From 593de06df956b7c272511a135901ef508a45e3e2 Mon Sep 17 00:00:00 2001
From: Karl Lessard <karl.lessard@gmail.com>
Date: Wed, 29 Jul 2020 09:54:11 -0400
Subject: [PATCH 01/11] Implements both Tensor and NdArray interfaces from same
 instance

---
 .../annotations/org/tensorflow/op/Ops.java    |   2 +-
 .../main/java/org/tensorflow/DataType.java    |  25 ++-
 .../main/java/org/tensorflow/DataTypes.java   |   8 +-
 .../java/org/tensorflow/EagerOperation.java   |   8 +-
 .../src/main/java/org/tensorflow/Operand.java |   6 +-
 .../src/main/java/org/tensorflow/Output.java  |   4 +-
 .../src/main/java/org/tensorflow/Session.java |   2 +-
 .../src/main/java/org/tensorflow/Tensor.java  | 167 +++---------------
 .../src/main/java/org/tensorflow/Tensors.java |  70 ++++++++
 ...fer.java => ByteSequenceTensorBuffer.java} |  16 +-
 .../internal/buffer/TensorBuffers.java        |   4 +-
 .../buffer/TensorRawDataBufferFactory.java    |   4 +-
 .../java/org/tensorflow/op/core/Constant.java |   4 +-
 .../org/tensorflow/types/BooleanTensor.java   | 126 +++++++++++++
 .../java/org/tensorflow/types/ByteTensor.java | 125 +++++++++++++
 .../org/tensorflow/types/DoubleTensor.java    | 126 +++++++++++++
 .../org/tensorflow/types/FloatTensor.java     | 126 +++++++++++++
 .../java/org/tensorflow/types/IntTensor.java  | 126 +++++++++++++
 .../java/org/tensorflow/types/LongTensor.java | 126 +++++++++++++
 .../java/org/tensorflow/types/RawTensor.java  |  66 +++++++
 .../org/tensorflow/types/StringTensor.java    | 125 +++++++++++++
 .../java/org/tensorflow/types/TBfloat16.java  |  50 +++---
 .../main/java/org/tensorflow/types/TBool.java |  44 +++--
 .../java/org/tensorflow/types/TFloat16.java   |  50 +++---
 .../java/org/tensorflow/types/TFloat32.java   |  47 +++--
 .../java/org/tensorflow/types/TFloat64.java   |  47 +++--
 .../java/org/tensorflow/types/TInt32.java     |  50 +++---
 .../java/org/tensorflow/types/TInt64.java     |  50 +++---
 .../java/org/tensorflow/types/TString.java    |  79 ++++-----
 .../java/org/tensorflow/types/TUint8.java     |  48 +++--
 .../org/tensorflow/types/family/TFloat.java   |   5 +
 .../test/java/org/tensorflow/TensorTest.java  |   2 +-
 32 files changed, 1311 insertions(+), 427 deletions(-)
 create mode 100644 tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Tensors.java
 rename tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/buffer/{StringTensorBuffer.java => ByteSequenceTensorBuffer.java} (91%)
 create mode 100644 tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/BooleanTensor.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/ByteTensor.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/DoubleTensor.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/FloatTensor.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/IntTensor.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/LongTensor.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/RawTensor.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/StringTensor.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/family/TFloat.java

diff --git a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/Ops.java b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/Ops.java
index 8c7aa1c0408..40f77b75711 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/Ops.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/Ops.java
@@ -1872,7 +1872,7 @@ public Constant<TString> constant(Charset charset, Shape shape, DataBuffer<Strin
    * @throws IllegalArgumentException If the tensor datatype or shape is not compatible with the
    *      buffer
    */
-  public <T extends TType> Constant<T> constant(DataType<T> type, Shape shape,
+  public <T extends TType & Tensor> Constant<T> constant(DataType<T> type, Shape shape,
       ByteDataBuffer data) {
     return Constant.tensorOf(scope, type, shape, data);
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/DataType.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/DataType.java
index 7b76b6dd02e..fa0603c6961 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/DataType.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/DataType.java
@@ -32,7 +32,7 @@
 public final class DataType<T extends TType> {
 
   @FunctionalInterface
-  public interface TensorMapper<T> {
+  public interface TensorInstantiator<T> {
 
     /**
      * Maps tensor memory to a data structure for manipulating elements of this type.
@@ -50,12 +50,11 @@ public interface TensorMapper<T> {
    * @param name readable-name for this type
    * @param value must match the corresponding TF_* value in the TensorFlow C API.
    * @param byteSize size of an element of this type, in bytes, -1 if unknown
+   * @param tensorMapper method for instantiating tensor from a native reference
    * @param <T> a tensor type
-   * @param tensorMapper method for mapping tensor memory to elements of this type
    */
-  public static <T extends TType> DataType<T> create(
-      String name, int value, int byteSize, TensorMapper<T> tensorMapper) {
-    return new DataType<>(name, value, byteSize, tensorMapper);
+  public static <T extends TType> DataType<T> create(String name, int value, int byteSize, TensorInstantiator<T> instantiator) {
+    return new DataType<>(name, value, byteSize, instantiator);
   }
 
   /**
@@ -158,24 +157,24 @@ int nativeCode() {
   }
 
   /**
-   * Maps a tensor to a data structure for manipulating elements of this type.
+   * Instantiate a tensor of this datatype from the provided native handle
    *
-   * @param tensor tensor to map
-   * @return data structure of elements of this type
+   * @param handle tensor native handle
+   * @return a tensor of this datatype
    */
-  T map(Tensor<T> tensor) {
-    return tensorMapper.apply(tensor.nativeHandle(), tensor.shape());
+  T instantiateTensor(TF_Tensor handle, Shape shape) {
+    return tensorInstantiator.apply(handle, shape);
   }
 
   private final int nativeCode;
   private final int byteSize;
   private final String name;
-  private final TensorMapper<T> tensorMapper;
+  private final TensorInstantiator<T> tensorInstantiator;
 
-  private DataType(String name, int nativeCode, int byteSize, TensorMapper<T> tensorMapper) {
+  private DataType(String name, int nativeCode, int byteSize, TensorInstantiator<T> tensorInstantiator) {
     this.name = name;
     this.nativeCode = nativeCode;
     this.byteSize = byteSize;
-    this.tensorMapper = tensorMapper;
+    this.tensorInstantiator = tensorInstantiator;
   }
 }
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/DataTypes.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/DataTypes.java
index 77c0de0c83f..03d613475b5 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/DataTypes.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/DataTypes.java
@@ -43,8 +43,8 @@ final class DataTypes {
    * @return data type for this code
    * @throws IllegalArgumentException if the code matches no registered data type
    */
-  static DataType<?> fromNativeCode(int nativeCode) {
-    DataType<?> dataType = DATA_TYPE_REGISTRY.get(nativeCode);
+  static DataType<? extends Tensor> fromNativeCode(int nativeCode) {
+    DataType<? extends Tensor> dataType = DATA_TYPE_REGISTRY.get(nativeCode);
     if (dataType == null) {
       throw new IllegalArgumentException(
           "DataType " + nativeCode + " is not recognized in Java (version " + TensorFlow.version() + ")");
@@ -52,7 +52,7 @@ static DataType<?> fromNativeCode(int nativeCode) {
     return dataType;
   }
 
-  private static final Map<Integer, DataType<?>> DATA_TYPE_REGISTRY = new HashMap<>();
+  private static final Map<Integer, DataType<? extends Tensor>> DATA_TYPE_REGISTRY = new HashMap<>();
 
   static {
     register(TBool.DTYPE);
@@ -68,7 +68,7 @@ static DataType<?> fromNativeCode(int nativeCode) {
 
   // TODO (karllessard): Right now this method is private but we might want to expose it
   //      to allow user to register custom data types?
-  private static void register(DataType<?> dataType) {
+  private static void register(DataType<? extends Tensor> dataType) {
     DATA_TYPE_REGISTRY.put(dataType.nativeCode(), dataType);
     DATA_TYPE_REGISTRY.put(dataType.nativeCode() + 100, dataType);
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/EagerOperation.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/EagerOperation.java
index 012981ac59c..690d5d1a4cb 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/EagerOperation.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/EagerOperation.java
@@ -162,7 +162,13 @@ private static Tensor<?> resolveTensorHandle(TFE_TensorHandle handle, EagerSessi
       TF_Status status = TF_Status.newStatus();
       TF_Tensor tensor = TFE_TensorHandleResolve(handle, status).withDeallocator();
       status.throwExceptionIfNotOK();
-      return Tensor.fromHandle(tensor, session);
+      Tensor<?> t = Tensors.fromHandle(tensor);
+      session.attach(t.nativeHandle());
+      // Now that the tensor life is attached to the eager session scope, closing it will
+      // implicitly detach it from its previous internal scope
+      // FIXME TENSOR REFACT : is that right and ok?
+      t.close();
+      return t;
     }
   }
 
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Operand.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Operand.java
index fa21f32d4ce..070fc55cb0f 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Operand.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Operand.java
@@ -60,7 +60,7 @@ public interface Operand<T extends TType> extends Op {
    * @return the tensor
    * @throws IllegalStateException if this is an operand of a graph
    */
-  default Tensor<T> asTensor() {
+  default T asTensor() {
     return asOutput().tensor();
   }
 
@@ -72,8 +72,10 @@ default Tensor<T> asTensor() {
    *
    * @return the tensor data
    * @throws IllegalStateException if this is an operand of a graph
+   * @deprecated use {@link #asTensor()} instead
    */
+  @Deprecated
   default T data() {
-    return asOutput().tensor().data();
+    return asTensor();
   }
 }
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Output.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Output.java
index a873df8ff4c..c6499377ded 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Output.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Output.java
@@ -80,8 +80,8 @@ public <U extends TType> Output<U> expect(DataType<U> dt) {
    * @see EagerSession
    */
   @SuppressWarnings("unchecked")
-  public Tensor<T> tensor() {
-    return (Tensor<T>) operation.tensor(index);
+  public T tensor() {
+    return (T)operation.tensor(index);
   }
 
   @Override
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Session.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Session.java
index 4e82f3944b8..b486c0b5dd7 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Session.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Session.java
@@ -667,7 +667,7 @@ private static RunMetadata run(
 
       for (int i = 0; i < noutputs; ++i) {
         TF_Tensor h = outputValues.get(TF_Tensor.class, i).withDeallocator();
-        outputTensors.add(Tensor.fromHandle(h));
+        outputTensors.add(Tensors.fromHandle(h));
       }
       try {
         return runMetadata != null ? RunMetadata.parseFrom(runMetadata.dataAsByteBuffer()) : null;
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Tensor.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Tensor.java
index 6787713418f..8f130296aba 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Tensor.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Tensor.java
@@ -15,10 +15,7 @@
 
 package org.tensorflow;
 
-import static org.tensorflow.internal.c_api.global.tensorflow.TF_Dim;
-import static org.tensorflow.internal.c_api.global.tensorflow.TF_NumDims;
 import static org.tensorflow.internal.c_api.global.tensorflow.TF_TensorByteSize;
-import static org.tensorflow.internal.c_api.global.tensorflow.TF_TensorType;
 
 import java.util.function.Consumer;
 import org.bytedeco.javacpp.PointerScope;
@@ -44,7 +41,7 @@
  * }
  * }</pre>
  */
-public final class Tensor<T extends TType> implements AutoCloseable {
+public interface Tensor<T extends TType> extends AutoCloseable {
 
   /**
    * Allocates a tensor of a given datatype and shape.
@@ -68,7 +65,7 @@ public final class Tensor<T extends TType> implements AutoCloseable {
    * @return an allocated but uninitialized tensor
    * @throws IllegalStateException if tensor failed to be allocated
    */
-  public static <T extends TType> Tensor<T> of(DataType<T> dtype, Shape shape) {
+  static <T extends TType & Tensor> T of(DataType<T> dtype, Shape shape) {
     return of(dtype, shape, shape.size() * dtype.byteSize());
   }
 
@@ -91,20 +88,8 @@ public static <T extends TType> Tensor<T> of(DataType<T> dtype, Shape shape) {
    *                                  store the tensor data
    * @throws IllegalStateException if tensor failed to be allocated
    */
-  public static <T extends TType> Tensor<T> of(DataType<T> dtype, Shape shape, long size) {
-    // Minimum requirements for datatypes of variable length cannot be verified in a relevant way so
-    // we only validate them for fixed length datatypes
-    if (!dtype.isVariableLength() && shape.size() * dtype.byteSize() > size) {
-      throw new IllegalArgumentException("Tensor size is not large enough to contain all scalar values");
-    }
-    Tensor<T> t = new Tensor<>(dtype, shape);
-    TF_Tensor nativeHandle = allocate(t.dtype.nativeCode(), shape.asArray(), size);
-    try (PointerScope scope = new PointerScope()) {
-        scope.attach(nativeHandle);
-        t.tensorHandle = nativeHandle;
-        t.tensorScope = scope.extend();
-        return t;
-    }
+  static <T extends TType & Tensor> T of(DataType<T> dtype, Shape shape, long size) {
+    return Tensors.allocate(dtype, shape, size);
   }
 
   /**
@@ -131,8 +116,7 @@ public static <T extends TType> Tensor<T> of(DataType<T> dtype, Shape shape, lon
    * @return an allocated and initialized tensor
    * @throws IllegalStateException if tensor failed to be allocated
    */
-  public static <T extends TType> Tensor<T> of(DataType<T> dtype, Shape shape,
-      Consumer<T> dataInitializer) {
+  static <T extends TType & Tensor> T of(DataType<T> dtype, Shape shape, Consumer<T> dataInitializer) {
     return of(dtype, shape, shape.size() * dtype.byteSize(), dataInitializer);
   }
 
@@ -149,18 +133,17 @@ public static <T extends TType> Tensor<T> of(DataType<T> dtype, Shape shape,
    * @param dtype datatype of the tensor
    * @param shape shape of the tensor
    * @param size size, in bytes, of the tensor
-   * @param dataInitializer method receiving accessor to the allocated tensor data for initialization
+   * @param tensorInit method receiving accessor to the allocated tensor data for initialization
    * @return an allocated and initialized tensor
    * @see #of(DataType, Shape, long, Consumer)
    * @throws IllegalArgumentException if {@code size} is smaller than the minimum space required to
    *                                  store the tensor data
    * @throws IllegalStateException if tensor failed to be allocated
    */
-  public static <T extends TType> Tensor<T> of(DataType<T> dtype, Shape shape, long size,
-      Consumer<T> dataInitializer) {
-    Tensor<T> tensor = of(dtype, shape, size);
+  static <T extends TType & Tensor> T of(DataType<T> dtype, Shape shape, long size, Consumer<T> tensorInit) {
+    T tensor = of(dtype, shape, size);
     try {
-      dataInitializer.accept(tensor.data());
+      tensorInit.accept(tensor);
       return tensor;
     } catch (Throwable t) {
       tensor.close();
@@ -181,10 +164,10 @@ public static <T extends TType> Tensor<T> of(DataType<T> dtype, Shape shape, lon
    * @throws IllegalArgumentException if {@code rawData} is not large enough to contain the tensor data
    * @throws IllegalStateException if tensor failed to be allocated with the given parameters
    */
-  public static <T extends TType> Tensor<T> of(DataType<T> dtype, Shape shape, ByteDataBuffer rawData) {
-    Tensor<T> t = of(dtype, shape, rawData.size());
-    rawData.copyTo(TensorBuffers.toBytes(t.nativeHandle()), rawData.size());
-    return t;
+  static <T extends TType & Tensor> T of(DataType<T> dtype, Shape shape, ByteDataBuffer rawData) {
+    T tensor = of(dtype, shape, rawData.size());
+    rawData.copyTo(TensorBuffers.toBytes(tensor.nativeHandle()), rawData.size());
+    return tensor;
   }
 
   /**
@@ -198,12 +181,12 @@ public static <T extends TType> Tensor<T> of(DataType<T> dtype, Shape shape, Byt
    *     {@code U}.
    */
   @SuppressWarnings("unchecked")
-  public <U extends TType> Tensor<U> expect(DataType<U> dt) {
-    if (!dt.equals(this.dtype)) {
+  default <U extends TType & Tensor> U expect(DataType<U> dt) {
+    if (!dt.equals(dataType())) {
       throw new IllegalArgumentException(
-          "Cannot cast from tensor of " + dtype + " to tensor of " + dt);
+          "Cannot cast from tensor of " + dataType() + " to tensor of " + dt);
     }
-    return ((Tensor<U>) this);
+    return (U)this;
   }
 
   /**
@@ -215,22 +198,13 @@ public <U extends TType> Tensor<U> expect(DataType<U> dt) {
    * <p>The Tensor object is no longer usable after {@code close} returns.
    */
   @Override
-  public void close() {
-    tensorScope.close();
-  }
+  void close();
 
   /** Returns the {@link DataType} of elements stored in the Tensor. */
-  public DataType<T> dataType() {
-    return dtype;
-  }
+  DataType<T> dataType();
 
   /** Returns the size, in bytes, of the tensor data. */
-  public long numBytes() {
-    if (numBytes == null) {
-      numBytes = TF_TensorByteSize(tensorHandle);
-    }
-    return numBytes;
-  }
+  long numBytes();
 
   /**
    * Returns the <a href="https://www.tensorflow.org/resources/dims_types.html#shape">shape</a> of
@@ -238,9 +212,7 @@ public long numBytes() {
    *
    * @return shape of this tensor
    */
-  public Shape shape() {
-    return shape;
-  }
+  Shape shape();
 
   /**
    * Returns the data of this tensor.
@@ -274,13 +246,9 @@ public Shape shape() {
    * @throws IllegalStateException if the tensor has been closed
    * @see NdArray
    */
-  public T data() {
-    if (data == null) {
-      data = dtype.map(this);
-    } else {
-      nativeHandle(); // Checks that the tensor has not been released or will throw
-    }
-    return data;
+  default T data() {
+    nativeHandle(); // make sure native handle is still valid
+    return (T) this;
   }
 
   /**
@@ -293,95 +261,14 @@ public T data() {
    * @return the tensor raw data mapped to a read-only byte buffer
    * @throws IllegalStateException if the tensor has been closed
    */
-  public ByteDataBuffer rawData() {
-    return TensorBuffers.toBytes(nativeHandle(), true);
-  }
-
-  /** Returns a string describing the type and shape of the Tensor. */
-  @Override
-  public String toString() {
-    return String.format("%s tensor with shape %s", dtype.toString(), shape);
-  }
-
-  /**
-   * Create a Tensor object from a handle to the C TF_Tensor object.
-   *
-   * <p>Takes ownership of the handle.
-   */
-  static Tensor<?> fromHandle(TF_Tensor handle) {
-    Tensor<?> t = new Tensor<>(DataTypes.fromNativeCode(dtype(handle)), Shape.of(shape(handle)));
-    try (PointerScope scope = new PointerScope()) {
-        scope.attach(handle);
-        t.tensorHandle = handle;
-        t.tensorScope = scope.extend();
-    }
-    return t;
-  }
-
-  /**
-   * Create an eager Tensor object from a handle to the C TF_Tensor object.
-   *
-   * <p>Takes ownership of the handle.
-   */
-  static Tensor<?> fromHandle(TF_Tensor handle, EagerSession session) {
-    Tensor<?> t = fromHandle(handle);
-    session.attach(handle);
-    t.tensorScope.detach(handle);
-    return t;
-  }
+  ByteDataBuffer rawData();
 
   /**
    * @return native handle to this tensor
    * @throws IllegalStateException if tensor has been closed
    */
-  TF_Tensor nativeHandle() {
-    return requireHandle(tensorHandle);
-  }
-
-  private PointerScope tensorScope;
-  private TF_Tensor tensorHandle;
-
-  private static TF_Tensor requireHandle(TF_Tensor handle) {
-    if (handle == null || handle.isNull()) {
-      throw new IllegalStateException("close() was called on the Tensor");
-    }
-    return handle;
-  }
-
-  private static TF_Tensor allocate(int dtype, long[] shape, long byteSize) {
-    TF_Tensor t = TF_Tensor.allocateTensor(dtype, shape, byteSize);
-    if (t == null || t.isNull()) {
-      throw new IllegalStateException("unable to allocate memory for the Tensor");
-    }
-    return t;
-  }
-
-  private static int dtype(TF_Tensor handle) {
-    requireHandle(handle);
-    return TF_TensorType(handle);
-  }
-
-  private static long[] shape(TF_Tensor handle) {
-    requireHandle(handle);
-    int numDims = TF_NumDims(handle);
-    long[] dims = new long[numDims];
-    for (int i = 0; i < numDims; ++i) {
-      dims[i] = TF_Dim(handle, i);
-    }
-    return dims;
-  }
+  TF_Tensor nativeHandle();
+}
 
-  private final DataType<T> dtype;
-  private final Shape shape;
-  private T data = null;
-  private Long numBytes = null;
 
-  private Tensor(DataType<T> dtype, Shape shape) {
-    this.dtype = dtype;
-    this.shape = shape;
-  }
 
-  static {
-    TensorFlow.init();
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Tensors.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Tensors.java
new file mode 100644
index 00000000000..4ac2e92f714
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Tensors.java
@@ -0,0 +1,70 @@
+package org.tensorflow;
+
+import static org.tensorflow.internal.c_api.global.tensorflow.TF_Dim;
+import static org.tensorflow.internal.c_api.global.tensorflow.TF_NumDims;
+import static org.tensorflow.internal.c_api.global.tensorflow.TF_TensorType;
+
+import org.bytedeco.javacpp.PointerScope;
+import org.tensorflow.internal.c_api.TF_Tensor;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.types.family.TType;
+
+final class Tensors {
+
+  static <T extends TType> T allocate(DataType<T> dataType, Shape shape, long size) {
+    // Minimum requirements for datatypes of variable length cannot be verified in a relevant way so
+    // we only validate them for fixed length datatypes
+    if (!dataType.isVariableLength() && shape.size() * dataType.byteSize() > size) {
+      throw new IllegalArgumentException("Tensor size is not large enough to contain all scalar values");
+    }
+    TF_Tensor nativeHandle = allocate(dataType.nativeCode(), shape.asArray(), size);
+    try (PointerScope scope = new PointerScope()) {
+      scope.attach(nativeHandle);
+      return dataType.instantiateTensor(nativeHandle, shape);
+    }
+  }
+
+  /**
+   * Create a Tensor object from a handle to the C TF_Tensor object.
+   *
+   * <p>Takes ownership of the handle.
+   */
+  static Tensor<?> fromHandle(TF_Tensor handle) {
+    DataType<? extends Tensor> dataType = DataTypes.fromNativeCode(dtype(handle));
+    Shape shape = Shape.of(shape(handle));
+    try (PointerScope scope = new PointerScope()) {
+      scope.attach(handle);
+      return dataType.instantiateTensor(handle, shape);
+    }
+  }
+
+  private static TF_Tensor requireHandle(TF_Tensor handle) {
+    if (handle == null || handle.isNull()) {
+      throw new IllegalStateException("close() was called on the Tensor");
+    }
+    return handle;
+  }
+
+  private static TF_Tensor allocate(int dtype, long[] shape, long byteSize) {
+    TF_Tensor t = TF_Tensor.allocateTensor(dtype, shape, byteSize);
+    if (t == null || t.isNull()) {
+      throw new IllegalStateException("unable to allocate memory for the Tensor");
+    }
+    return t;
+  }
+
+  private static int dtype(TF_Tensor handle) {
+    requireHandle(handle);
+    return TF_TensorType(handle);
+  }
+
+  private static long[] shape(TF_Tensor handle) {
+    requireHandle(handle);
+    int numDims = TF_NumDims(handle);
+    long[] dims = new long[numDims];
+    for (int i = 0; i < numDims; ++i) {
+      dims[i] = TF_Dim(handle, i);
+    }
+    return dims;
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/buffer/StringTensorBuffer.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/buffer/ByteSequenceTensorBuffer.java
similarity index 91%
rename from tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/buffer/StringTensorBuffer.java
rename to tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/buffer/ByteSequenceTensorBuffer.java
index 83cdab33452..bc80b2a164c 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/buffer/StringTensorBuffer.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/buffer/ByteSequenceTensorBuffer.java
@@ -50,7 +50,7 @@
  * <p>After its data has been initialized, the buffer is read-only as it is not possible to change
  * safely a value without reinitializing the whole data.
  */
-public class StringTensorBuffer extends AbstractDataBuffer<byte[]> {
+public class ByteSequenceTensorBuffer extends AbstractDataBuffer<byte[]> {
 
   /**
    * Computes how many bytes are required to store the given data in a string buffer.
@@ -66,7 +66,7 @@ public static <T> long computeSize(NdArray<T> data, Function<T, byte[]> getBytes
     // reserve space to store length and data of each values
     for (NdArray<T> scalar : data.scalars()) {
       byte[] elementBytes = getBytes.apply(scalar.getObject());
-      size += elementBytes.length + StringTensorBuffer.varintLength(elementBytes.length);
+      size += elementBytes.length + ByteSequenceTensorBuffer.varintLength(elementBytes.length);
     }
     return size;
   }
@@ -129,8 +129,8 @@ public boolean isReadOnly() {
 
   @Override
   public DataBuffer<byte[]> copyTo(DataBuffer<byte[]> dst, long size) {
-    if (size == size() && dst instanceof StringTensorBuffer) {
-      StringTensorBuffer tensorDst = (StringTensorBuffer) dst;
+    if (size == size() && dst instanceof ByteSequenceTensorBuffer) {
+      ByteSequenceTensorBuffer tensorDst = (ByteSequenceTensorBuffer) dst;
       if (offsets.size() != size || data.size() != size) {
         throw new IllegalArgumentException(
             "Cannot copy string tensor data to another tensor of a different size");
@@ -145,20 +145,20 @@ public DataBuffer<byte[]> copyTo(DataBuffer<byte[]> dst, long size) {
 
   @Override
   public DataBuffer<byte[]> offset(long index) {
-    return new StringTensorBuffer(offsets.offset(index), data);
+    return new ByteSequenceTensorBuffer(offsets.offset(index), data);
   }
 
   @Override
   public DataBuffer<byte[]> narrow(long size) {
-    return new StringTensorBuffer(offsets.narrow(size), data);
+    return new ByteSequenceTensorBuffer(offsets.narrow(size), data);
   }
 
   @Override
   public DataBuffer<byte[]> slice(long index, long size) {
-    return new StringTensorBuffer(offsets.slice(index, size), data);
+    return new ByteSequenceTensorBuffer(offsets.slice(index, size), data);
   }
 
-  StringTensorBuffer(LongDataBuffer offsets, ByteDataBuffer data) {
+  ByteSequenceTensorBuffer(LongDataBuffer offsets, ByteDataBuffer data) {
     this.offsets = offsets;
     this.data = data;
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/buffer/TensorBuffers.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/buffer/TensorBuffers.java
index f29396dd321..415c5ca35ef 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/buffer/TensorBuffers.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/buffer/TensorBuffers.java
@@ -156,7 +156,7 @@ public static BooleanDataBuffer toBooleans(TF_Tensor nativeTensor) {
    * @param nativeTensor native reference to the tensor
    * @return a string buffer
    */
-  public static StringTensorBuffer toStrings(TF_Tensor nativeTensor, long numElements) {
+  public static ByteSequenceTensorBuffer toStrings(TF_Tensor nativeTensor, long numElements) {
     Pointer tensorMemory = tensorMemory(nativeTensor);
     if (TensorRawDataBufferFactory.canBeUsed()) {
       return TensorRawDataBufferFactory.mapTensorToStrings(tensorMemory, numElements);
@@ -173,7 +173,7 @@ public static StringTensorBuffer toStrings(TF_Tensor nativeTensor, long numEleme
     dataBuffer.position((int)numElements * Long.BYTES);
     ByteDataBuffer data = DataBuffers.of(dataBuffer.slice());
 
-    return new StringTensorBuffer(offsets, data);
+    return new ByteSequenceTensorBuffer(offsets, data);
   }
 
   private static Pointer tensorMemory(TF_Tensor nativeTensor) {
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/buffer/TensorRawDataBufferFactory.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/buffer/TensorRawDataBufferFactory.java
index 1cfb1c9ab9a..dbaf31f1dcc 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/buffer/TensorRawDataBufferFactory.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/buffer/TensorRawDataBufferFactory.java
@@ -57,13 +57,13 @@ static BooleanDataBuffer mapTensorToBooleans(Pointer tensorMemory) {
     return mapNativeBooleans(tensorMemory.address(), tensorMemory.capacity(), false);
   }
 
-  static StringTensorBuffer mapTensorToStrings(Pointer tensorMemory, long numElements) {
+  static ByteSequenceTensorBuffer mapTensorToStrings(Pointer tensorMemory, long numElements) {
     long offsetByteSize = numElements * Long.BYTES;
     LongDataBuffer offsets = mapNativeLongs(tensorMemory.address(), offsetByteSize, false);
     ByteDataBuffer data = mapNativeBytes(
         tensorMemory.address() + offsetByteSize,
         tensorMemory.capacity() - offsetByteSize,
         false);
-    return new StringTensorBuffer(offsets, data);
+    return new ByteSequenceTensorBuffer(offsets, data);
   }
 }
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/op/core/Constant.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/op/core/Constant.java
index 6c214cc6819..c583bd8db0a 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/op/core/Constant.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/op/core/Constant.java
@@ -1004,9 +1004,9 @@ public static Constant<TUint8> tensorOf(Scope scope, Shape shape, ByteDataBuffer
    *     buffer
    */
   @Endpoint
-  public static <T extends TType> Constant<T> tensorOf(Scope scope, DataType<T> type, Shape shape,
+  public static <T extends TType & Tensor> Constant<T> tensorOf(Scope scope, DataType<T> type, Shape shape,
       ByteDataBuffer data) {
-    try (Tensor<T> value = Tensor.of(type, shape, data)) {
+    try (T value = Tensor.of(type, shape, data)) {
       return create(scope, value);
     }
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/BooleanTensor.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/BooleanTensor.java
new file mode 100644
index 00000000000..627062a990f
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/BooleanTensor.java
@@ -0,0 +1,126 @@
+package org.tensorflow.types;
+
+import org.tensorflow.DataType;
+import org.tensorflow.Tensor;
+import org.tensorflow.internal.c_api.TF_Tensor;
+import org.tensorflow.ndarray.BooleanNdArray;
+import org.tensorflow.ndarray.NdArray;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.ndarray.buffer.BooleanDataBuffer;
+import org.tensorflow.ndarray.buffer.ByteDataBuffer;
+import org.tensorflow.ndarray.buffer.DataBuffer;
+import org.tensorflow.ndarray.impl.dense.BooleanDenseNdArray;
+import org.tensorflow.ndarray.index.Index;
+import org.tensorflow.types.family.TType;
+
+public interface BooleanTensor<T extends BooleanTensor & TType> extends BooleanNdArray, Tensor<T> {
+
+  @Override
+  T setBoolean(boolean value, long... coordinates);
+
+  @Override
+  T set(NdArray<Boolean> src, long... coordinates);
+
+  @Override
+  T setObject(Boolean value, long... coordinates);
+
+  @Override
+  T copyTo(NdArray<Boolean> dst);
+
+  @Override
+  T read(DataBuffer<Boolean> dst);
+
+  @Override
+  T read(BooleanDataBuffer dst);
+
+  @Override
+  T write(DataBuffer<Boolean> src);
+
+  @Override
+  T write(BooleanDataBuffer src);
+}
+
+class BooleanTensorImpl<T extends BooleanTensor & TType> extends BooleanDenseNdArray implements BooleanTensor<T> {
+
+  @Override
+  public TF_Tensor nativeHandle() {
+    return rawTensor.nativeHandle();
+  }
+
+  @Override
+  public DataType<T> dataType() {
+    return rawTensor.dataType();
+  }
+
+  @Override
+  public Shape shape() {
+    return rawTensor.shape();
+  }
+
+  @Override
+  public long numBytes() {
+    return rawTensor.numBytes();
+  }
+
+  @Override
+  public ByteDataBuffer rawData() {
+    return rawTensor.rawData();
+  }
+
+  @Override
+  public void close() {
+    rawTensor.close();
+  }
+
+  @Override
+  public String toString() {
+    return rawTensor.toString();
+  }
+
+  @Override
+  public T setBoolean(boolean value, long... coordinates) {
+    return (T)super.setBoolean(value, coordinates);
+  }
+
+  @Override
+  public T setObject(Boolean value, long... coordinates) {
+    return (T)super.setObject(value, coordinates);
+  }
+
+  @Override
+  public T set(NdArray<Boolean> src, long... coordinates) {
+    return (T)super.set(src, coordinates);
+  }
+
+  @Override
+  public T copyTo(NdArray<Boolean> dst) {
+    return (T)super.copyTo(dst);
+  }
+
+  @Override
+  public T read(DataBuffer<Boolean> dst) {
+    return (T)super.read(dst);
+  }
+
+  @Override
+  public T read(BooleanDataBuffer dst) {
+    return (T)super.read(dst);
+  }
+
+  @Override
+  public T write(DataBuffer<Boolean> src) {
+    return (T)super.write(src);
+  }
+
+  @Override
+  public T write(BooleanDataBuffer src) {
+    return (T)super.write(src);
+  }
+
+  BooleanTensorImpl(TF_Tensor nativeHandle, DataType<T> dataType, Shape shape, BooleanDataBuffer buffer) {
+    super(buffer, shape);
+    this.rawTensor = new RawTensor(nativeHandle, dataType, shape);
+  }
+
+  private final RawTensor rawTensor;
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/ByteTensor.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/ByteTensor.java
new file mode 100644
index 00000000000..48451e3781c
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/ByteTensor.java
@@ -0,0 +1,125 @@
+package org.tensorflow.types;
+
+import org.tensorflow.DataType;
+import org.tensorflow.Tensor;
+import org.tensorflow.internal.c_api.TF_Tensor;
+import org.tensorflow.ndarray.ByteNdArray;
+import org.tensorflow.ndarray.NdArray;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.ndarray.buffer.ByteDataBuffer;
+import org.tensorflow.ndarray.buffer.DataBuffer;
+import org.tensorflow.ndarray.impl.dense.ByteDenseNdArray;
+import org.tensorflow.ndarray.index.Index;
+import org.tensorflow.types.family.TType;
+
+public interface ByteTensor<T extends ByteTensor & TType> extends ByteNdArray, Tensor<T> {
+
+  @Override
+  T setByte(byte value, long... coordinates);
+
+  @Override
+  T set(NdArray<Byte> src, long... coordinates);
+
+  @Override
+  T setObject(Byte value, long... coordinates);
+
+  @Override
+  T copyTo(NdArray<Byte> dst);
+
+  @Override
+  T read(DataBuffer<Byte> dst);
+
+  @Override
+  T read(ByteDataBuffer dst);
+
+  @Override
+  T write(DataBuffer<Byte> src);
+
+  @Override
+  T write(ByteDataBuffer src);
+}
+
+class ByteTensorImpl<T extends ByteTensor & TType> extends ByteDenseNdArray implements ByteTensor<T> {
+
+  @Override
+  public TF_Tensor nativeHandle() {
+    return rawTensor.nativeHandle();
+  }
+
+  @Override
+  public DataType<T> dataType() {
+    return rawTensor.dataType();
+  }
+
+  @Override
+  public Shape shape() {
+    return rawTensor.shape();
+  }
+
+  @Override
+  public long numBytes() {
+    return rawTensor.numBytes();
+  }
+
+  @Override
+  public ByteDataBuffer rawData() {
+    return rawTensor.rawData();
+  }
+
+  @Override
+  public void close() {
+    rawTensor.close();
+  }
+
+  @Override
+  public String toString() {
+    return rawTensor.toString();
+  }
+
+  @Override
+  public T setByte(byte value, long... coordinates) {
+    return (T)super.setByte(value, coordinates);
+  }
+
+  @Override
+  public T setObject(Byte value, long... coordinates) {
+    return (T)super.setObject(value, coordinates);
+  }
+
+  @Override
+  public T set(NdArray<Byte> src, long... coordinates) {
+    return (T)super.set(src, coordinates);
+  }
+
+  @Override
+  public T copyTo(NdArray<Byte> dst) {
+    return (T)super.copyTo(dst);
+  }
+
+  @Override
+  public T read(DataBuffer<Byte> dst) {
+    return (T)super.read(dst);
+  }
+
+  @Override
+  public T read(ByteDataBuffer dst) {
+    return (T)super.read(dst);
+  }
+
+  @Override
+  public T write(DataBuffer<Byte> src) {
+    return (T)super.write(src);
+  }
+
+  @Override
+  public T write(ByteDataBuffer src) {
+    return (T)super.write(src);
+  }
+
+  ByteTensorImpl(TF_Tensor nativeHandle, DataType<T> dataType, Shape shape, ByteDataBuffer buffer) {
+    super(buffer, shape);
+    this.rawTensor = new RawTensor(nativeHandle, dataType, shape);
+  }
+
+  private final RawTensor rawTensor;
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/DoubleTensor.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/DoubleTensor.java
new file mode 100644
index 00000000000..83fb0fa1c0d
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/DoubleTensor.java
@@ -0,0 +1,126 @@
+package org.tensorflow.types;
+
+import org.tensorflow.DataType;
+import org.tensorflow.Tensor;
+import org.tensorflow.internal.c_api.TF_Tensor;
+import org.tensorflow.ndarray.DoubleNdArray;
+import org.tensorflow.ndarray.NdArray;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.ndarray.buffer.ByteDataBuffer;
+import org.tensorflow.ndarray.buffer.DataBuffer;
+import org.tensorflow.ndarray.buffer.DoubleDataBuffer;
+import org.tensorflow.ndarray.impl.dense.DoubleDenseNdArray;
+import org.tensorflow.ndarray.index.Index;
+import org.tensorflow.types.family.TType;
+
+public interface DoubleTensor<T extends DoubleTensor & TType> extends DoubleNdArray, Tensor<T> {
+
+  @Override
+  T setDouble(double value, long... coordinates);
+
+  @Override
+  T set(NdArray<Double> src, long... coordinates);
+
+  @Override
+  T setObject(Double value, long... coordinates);
+
+  @Override
+  T copyTo(NdArray<Double> dst);
+
+  @Override
+  T read(DataBuffer<Double> dst);
+
+  @Override
+  T read(DoubleDataBuffer dst);
+
+  @Override
+  T write(DataBuffer<Double> src);
+
+  @Override
+  T write(DoubleDataBuffer src);
+}
+
+class DoubleTensorImpl<T extends DoubleTensor & TType> extends DoubleDenseNdArray implements DoubleTensor<T> {
+
+  @Override
+  public TF_Tensor nativeHandle() {
+    return rawTensor.nativeHandle();
+  }
+
+  @Override
+  public DataType<T> dataType() {
+    return rawTensor.dataType();
+  }
+
+  @Override
+  public Shape shape() {
+    return rawTensor.shape();
+  }
+
+  @Override
+  public long numBytes() {
+    return rawTensor.numBytes();
+  }
+
+  @Override
+  public ByteDataBuffer rawData() {
+    return rawTensor.rawData();
+  }
+
+  @Override
+  public void close() {
+    rawTensor.close();
+  }
+
+  @Override
+  public String toString() {
+    return rawTensor.toString();
+  }
+
+  @Override
+  public T setDouble(double value, long... coordinates) {
+    return (T)super.setDouble(value, coordinates);
+  }
+
+  @Override
+  public T setObject(Double value, long... coordinates) {
+    return (T)super.setObject(value, coordinates);
+  }
+
+  @Override
+  public T set(NdArray<Double> src, long... coordinates) {
+    return (T)super.set(src, coordinates);
+  }
+
+  @Override
+  public T copyTo(NdArray<Double> dst) {
+    return (T)super.copyTo(dst);
+  }
+
+  @Override
+  public T read(DataBuffer<Double> dst) {
+    return (T)super.read(dst);
+  }
+
+  @Override
+  public T read(DoubleDataBuffer dst) {
+    return (T)super.read(dst);
+  }
+
+  @Override
+  public T write(DataBuffer<Double> src) {
+    return (T)super.write(src);
+  }
+
+  @Override
+  public T write(DoubleDataBuffer src) {
+    return (T)super.write(src);
+  }
+
+  DoubleTensorImpl(TF_Tensor nativeHandle, DataType<T> dataType, Shape shape, DoubleDataBuffer buffer) {
+    super(buffer, shape);
+    this.rawTensor = new RawTensor(nativeHandle, dataType, shape);
+  }
+
+  private final RawTensor rawTensor;
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/FloatTensor.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/FloatTensor.java
new file mode 100644
index 00000000000..f819d073a4d
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/FloatTensor.java
@@ -0,0 +1,126 @@
+package org.tensorflow.types;
+
+import org.tensorflow.DataType;
+import org.tensorflow.Tensor;
+import org.tensorflow.internal.c_api.TF_Tensor;
+import org.tensorflow.ndarray.FloatNdArray;
+import org.tensorflow.ndarray.NdArray;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.ndarray.buffer.ByteDataBuffer;
+import org.tensorflow.ndarray.buffer.DataBuffer;
+import org.tensorflow.ndarray.buffer.FloatDataBuffer;
+import org.tensorflow.ndarray.impl.dense.FloatDenseNdArray;
+import org.tensorflow.ndarray.index.Index;
+import org.tensorflow.types.family.TType;
+
+public interface FloatTensor<T extends FloatTensor & TType> extends FloatNdArray, Tensor<T> {
+
+  @Override
+  T setFloat(float value, long... coordinates);
+
+  @Override
+  T set(NdArray<Float> src, long... coordinates);
+
+  @Override
+  T setObject(Float value, long... coordinates);
+
+  @Override
+  T copyTo(NdArray<Float> dst);
+
+  @Override
+  T read(DataBuffer<Float> dst);
+
+  @Override
+  T read(FloatDataBuffer dst);
+
+  @Override
+  T write(DataBuffer<Float> src);
+
+  @Override
+  T write(FloatDataBuffer src);
+}
+
+class FloatTensorImpl<T extends FloatTensor & TType> extends FloatDenseNdArray implements FloatTensor<T> {
+
+  @Override
+  public TF_Tensor nativeHandle() {
+    return rawTensor.nativeHandle();
+  }
+
+  @Override
+  public DataType<T> dataType() {
+    return rawTensor.dataType();
+  }
+
+  @Override
+  public Shape shape() {
+    return rawTensor.shape();
+  }
+
+  @Override
+  public long numBytes() {
+    return rawTensor.numBytes();
+  }
+
+  @Override
+  public ByteDataBuffer rawData() {
+    return rawTensor.rawData();
+  }
+
+  @Override
+  public void close() {
+    rawTensor.close();
+  }
+
+  @Override
+  public String toString() {
+    return rawTensor.toString();
+  }
+
+  @Override
+  public T setFloat(float value, long... coordinates) {
+    return (T)super.setFloat(value, coordinates);
+  }
+
+  @Override
+  public T setObject(Float value, long... coordinates) {
+    return (T)super.setObject(value, coordinates);
+  }
+
+  @Override
+  public T set(NdArray<Float> src, long... coordinates) {
+    return (T)super.set(src, coordinates);
+  }
+
+  @Override
+  public T copyTo(NdArray<Float> dst) {
+    return (T)super.copyTo(dst);
+  }
+
+  @Override
+  public T read(DataBuffer<Float> dst) {
+    return (T)super.read(dst);
+  }
+
+  @Override
+  public T read(FloatDataBuffer dst) {
+    return (T)super.read(dst);
+  }
+
+  @Override
+  public T write(DataBuffer<Float> src) {
+    return (T)super.write(src);
+  }
+
+  @Override
+  public T write(FloatDataBuffer src) {
+    return (T)super.write(src);
+  }
+
+  FloatTensorImpl(TF_Tensor nativeHandle, DataType<T> dataType, Shape shape, FloatDataBuffer buffer) {
+    super(buffer, shape);
+    this.rawTensor = new RawTensor(nativeHandle, dataType, shape);
+  }
+
+  private final RawTensor rawTensor;
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/IntTensor.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/IntTensor.java
new file mode 100644
index 00000000000..041232e83d8
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/IntTensor.java
@@ -0,0 +1,126 @@
+package org.tensorflow.types;
+
+import org.tensorflow.DataType;
+import org.tensorflow.Tensor;
+import org.tensorflow.internal.c_api.TF_Tensor;
+import org.tensorflow.ndarray.IntNdArray;
+import org.tensorflow.ndarray.NdArray;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.ndarray.buffer.ByteDataBuffer;
+import org.tensorflow.ndarray.buffer.DataBuffer;
+import org.tensorflow.ndarray.buffer.IntDataBuffer;
+import org.tensorflow.ndarray.impl.dense.IntDenseNdArray;
+import org.tensorflow.ndarray.index.Index;
+import org.tensorflow.types.family.TType;
+
+public interface IntTensor<T extends IntTensor & TType> extends IntNdArray, Tensor<T> {
+
+  @Override
+  T setInt(int value, long... coordinates);
+
+  @Override
+  T set(NdArray<Integer> src, long... coordinates);
+
+  @Override
+  T setObject(Integer value, long... coordinates);
+
+  @Override
+  T copyTo(NdArray<Integer> dst);
+
+  @Override
+  T read(DataBuffer<Integer> dst);
+
+  @Override
+  T read(IntDataBuffer dst);
+
+  @Override
+  T write(DataBuffer<Integer> src);
+
+  @Override
+  T write(IntDataBuffer src);
+}
+
+class IntTensorImpl<T extends IntTensor & TType> extends IntDenseNdArray implements IntTensor<T> {
+
+  @Override
+  public TF_Tensor nativeHandle() {
+    return rawTensor.nativeHandle();
+  }
+
+  @Override
+  public DataType<T> dataType() {
+    return rawTensor.dataType();
+  }
+
+  @Override
+  public Shape shape() {
+    return rawTensor.shape();
+  }
+
+  @Override
+  public long numBytes() {
+    return rawTensor.numBytes();
+  }
+
+  @Override
+  public ByteDataBuffer rawData() {
+    return rawTensor.rawData();
+  }
+
+  @Override
+  public void close() {
+    rawTensor.close();
+  }
+
+  @Override
+  public String toString() {
+    return rawTensor.toString();
+  }
+
+  @Override
+  public T setInt(int value, long... coordinates) {
+    return (T)super.setInt(value, coordinates);
+  }
+
+  @Override
+  public T setObject(Integer value, long... coordinates) {
+    return (T)super.setObject(value, coordinates);
+  }
+
+  @Override
+  public T set(NdArray<Integer> src, long... coordinates) {
+    return (T)super.set(src, coordinates);
+  }
+
+  @Override
+  public T copyTo(NdArray<Integer> dst) {
+    return (T)super.copyTo(dst);
+  }
+
+  @Override
+  public T read(DataBuffer<Integer> dst) {
+    return (T)super.read(dst);
+  }
+
+  @Override
+  public T read(IntDataBuffer dst) {
+    return (T)super.read(dst);
+  }
+
+  @Override
+  public T write(DataBuffer<Integer> src) {
+    return (T)super.write(src);
+  }
+
+  @Override
+  public T write(IntDataBuffer src) {
+    return (T)super.write(src);
+  }
+
+  IntTensorImpl(TF_Tensor nativeHandle, DataType<T> dataType, Shape shape, IntDataBuffer buffer) {
+    super(buffer, shape);
+    this.rawTensor = new RawTensor(nativeHandle, dataType, shape);
+  }
+
+  private final RawTensor rawTensor;
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/LongTensor.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/LongTensor.java
new file mode 100644
index 00000000000..b8965888f34
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/LongTensor.java
@@ -0,0 +1,126 @@
+package org.tensorflow.types;
+
+import org.tensorflow.DataType;
+import org.tensorflow.Tensor;
+import org.tensorflow.internal.c_api.TF_Tensor;
+import org.tensorflow.ndarray.LongNdArray;
+import org.tensorflow.ndarray.NdArray;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.ndarray.buffer.ByteDataBuffer;
+import org.tensorflow.ndarray.buffer.DataBuffer;
+import org.tensorflow.ndarray.buffer.LongDataBuffer;
+import org.tensorflow.ndarray.impl.dense.LongDenseNdArray;
+import org.tensorflow.ndarray.index.Index;
+import org.tensorflow.types.family.TType;
+
+public interface LongTensor<T extends LongTensor & TType> extends LongNdArray, Tensor<T> {
+
+  @Override
+  T setLong(long value, long... coordinates);
+
+  @Override
+  T set(NdArray<Long> src, long... coordinates);
+
+  @Override
+  T setObject(Long value, long... coordinates);
+
+  @Override
+  T copyTo(NdArray<Long> dst);
+
+  @Override
+  T read(DataBuffer<Long> dst);
+
+  @Override
+  T read(LongDataBuffer dst);
+
+  @Override
+  T write(DataBuffer<Long> src);
+
+  @Override
+  T write(LongDataBuffer src);
+}
+
+class LongTensorImpl<T extends LongTensor & TType> extends LongDenseNdArray implements LongTensor<T> {
+
+  @Override
+  public TF_Tensor nativeHandle() {
+    return rawTensor.nativeHandle();
+  }
+
+  @Override
+  public DataType<T> dataType() {
+    return rawTensor.dataType();
+  }
+
+  @Override
+  public Shape shape() {
+    return rawTensor.shape();
+  }
+
+  @Override
+  public long numBytes() {
+    return rawTensor.numBytes();
+  }
+
+  @Override
+  public ByteDataBuffer rawData() {
+    return rawTensor.rawData();
+  }
+
+  @Override
+  public void close() {
+    rawTensor.close();
+  }
+
+  @Override
+  public String toString() {
+    return rawTensor.toString();
+  }
+
+  @Override
+  public T setLong(long value, long... coordinates) {
+    return (T)super.setLong(value, coordinates);
+  }
+
+  @Override
+  public T setObject(Long value, long... coordinates) {
+    return (T)super.setObject(value, coordinates);
+  }
+
+  @Override
+  public T set(NdArray<Long> src, long... coordinates) {
+    return (T)super.set(src, coordinates);
+  }
+
+  @Override
+  public T copyTo(NdArray<Long> dst) {
+    return (T)super.copyTo(dst);
+  }
+
+  @Override
+  public T read(DataBuffer<Long> dst) {
+    return (T)super.read(dst);
+  }
+
+  @Override
+  public T read(LongDataBuffer dst) {
+    return (T)super.read(dst);
+  }
+
+  @Override
+  public T write(DataBuffer<Long> src) {
+    return (T)super.write(src);
+  }
+
+  @Override
+  public T write(LongDataBuffer src) {
+    return (T)super.write(src);
+  }
+
+  LongTensorImpl(TF_Tensor nativeHandle, DataType<T> dataType, Shape shape, LongDataBuffer buffer) {
+    super(buffer, shape);
+    this.rawTensor = new RawTensor(nativeHandle, dataType, shape);
+  }
+
+  private final RawTensor rawTensor;
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/RawTensor.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/RawTensor.java
new file mode 100644
index 00000000000..872546c8ae0
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/RawTensor.java
@@ -0,0 +1,66 @@
+package org.tensorflow.types;
+
+import static org.tensorflow.internal.c_api.global.tensorflow.TF_TensorByteSize;
+
+import org.bytedeco.javacpp.PointerScope;
+import org.tensorflow.DataType;
+import org.tensorflow.Tensor;
+import org.tensorflow.internal.buffer.TensorBuffers;
+import org.tensorflow.internal.c_api.TF_Tensor;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.ndarray.buffer.ByteDataBuffer;
+import org.tensorflow.types.family.TType;
+
+class RawTensor<T extends TType> implements Tensor<T> {
+
+  @Override
+  public TF_Tensor nativeHandle() {
+    if (nativeHandle.isNull()) {
+      throw new IllegalStateException("Tensor has been already released");
+    }
+    return nativeHandle;
+  }
+
+  @Override
+  public DataType<T> dataType() {
+    return dtype;
+  }
+
+  @Override
+  public Shape shape() {
+    return shape;
+  }
+
+  @Override
+  public long numBytes() {
+    return TF_TensorByteSize(nativeHandle);
+  }
+
+  @Override
+  public ByteDataBuffer rawData() {
+    return TensorBuffers.toBytes(nativeHandle, true);
+  }
+
+  @Override
+  public void close() {
+    tensorScope.close();
+  }
+
+  /** Returns a string describing the type and shape of the Tensor. */
+  @Override
+  public String toString() {
+    return String.format("%s tensor with shape %s", dataType().toString(), shape());
+  }
+
+  private final PointerScope tensorScope = new PointerScope();
+  private final TF_Tensor nativeHandle;
+  private final DataType<T> dtype;
+  private final Shape shape;
+
+  RawTensor(TF_Tensor nativeHandle, DataType<T> dtype, Shape shape) {
+    this.nativeHandle = nativeHandle;
+    this.dtype = dtype;
+    this.shape = shape;
+    this.tensorScope.attach(nativeHandle);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/StringTensor.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/StringTensor.java
new file mode 100644
index 00000000000..ca03e589360
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/StringTensor.java
@@ -0,0 +1,125 @@
+package org.tensorflow.types;
+
+import org.tensorflow.DataType;
+import org.tensorflow.Tensor;
+import org.tensorflow.internal.buffer.ByteSequenceTensorBuffer;
+import org.tensorflow.internal.c_api.TF_Tensor;
+import org.tensorflow.ndarray.NdArray;
+import org.tensorflow.ndarray.NdArrays;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.ndarray.buffer.ByteDataBuffer;
+import org.tensorflow.ndarray.buffer.DataBuffer;
+import org.tensorflow.ndarray.buffer.layout.DataLayout;
+import org.tensorflow.ndarray.impl.dense.DenseNdArray;
+import org.tensorflow.ndarray.index.Index;
+import org.tensorflow.types.family.TType;
+
+public interface StringTensor<T extends StringTensor & TType> extends NdArray<String>, Tensor<T> {
+
+  /**
+   * @return the tensor data as a n-dimensional array of raw byte sequences.
+   */
+  NdArray<byte[]> asBytes();
+
+  @Override
+  T set(NdArray<String> src, long... coordinates);
+
+  @Override
+  T setObject(String value, long... coordinates);
+
+  @Override
+  T copyTo(NdArray<String> dst);
+
+  @Override
+  T read(DataBuffer<String> dst);
+
+  @Override
+  T write(DataBuffer<String> src);
+}
+
+class StringTensorImpl<T extends StringTensor & TType> extends DenseNdArray<String> implements StringTensor<T> {
+
+  @Override
+  public NdArray<byte[]> asBytes() {
+    return NdArrays.wrap(shape(), rawBuffer);
+  }
+
+  @Override
+  public TF_Tensor nativeHandle() {
+    return rawTensor.nativeHandle();
+  }
+
+  @Override
+  public DataType<T> dataType() {
+    return rawTensor.dataType();
+  }
+
+  @Override
+  public Shape shape() {
+    return rawTensor.shape();
+  }
+
+  @Override
+  public long numBytes() {
+    return rawTensor.numBytes();
+  }
+
+  @Override
+  public ByteDataBuffer rawData() {
+    return rawTensor.rawData();
+  }
+
+  @Override
+  public void close() {
+    rawTensor.close();
+  }
+
+  @Override
+  public String toString() {
+    return rawTensor.toString();
+  }
+
+  @Override
+  public T setObject(String value, long... coordinates) {
+    return (T)super.setObject(value, coordinates);
+  }
+
+  @Override
+  public T set(NdArray<String> src, long... coordinates) {
+    return (T)super.set(src, coordinates);
+  }
+
+  @Override
+  public T copyTo(NdArray<String> dst) {
+    return (T)super.copyTo(dst);
+  }
+
+  @Override
+  public T read(DataBuffer<String> dst) {
+    return (T)super.read(dst);
+  }
+
+  @Override
+  public T write(DataBuffer<String> src) {
+    return (T)super.write(src);
+  }
+
+  protected ByteSequenceTensorBuffer rawBuffer() {
+    return rawBuffer;
+  }
+
+  StringTensorImpl(
+      TF_Tensor nativeHandle,
+      DataType<T> dataType,
+      Shape shape,
+      DataLayout<DataBuffer<byte[]>, String> layout,
+      ByteSequenceTensorBuffer buffer
+  ) {
+    super(layout.applyTo(buffer), shape);
+    this.rawTensor = new RawTensor(nativeHandle, dataType, shape);
+    this.rawBuffer = buffer;
+  }
+
+  private final RawTensor rawTensor;
+  private final ByteSequenceTensorBuffer rawBuffer;
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TBfloat16.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TBfloat16.java
index 50f6ea49b06..7e87343a979 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TBfloat16.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TBfloat16.java
@@ -23,14 +23,12 @@
 import org.tensorflow.exceptions.TensorFlowException;
 import org.tensorflow.internal.buffer.TensorBuffers;
 import org.tensorflow.internal.c_api.TF_Tensor;
+import org.tensorflow.ndarray.NdArray;
 import org.tensorflow.ndarray.Shape;
+import org.tensorflow.ndarray.StdArrays;
 import org.tensorflow.ndarray.buffer.FloatDataBuffer;
 import org.tensorflow.ndarray.buffer.layout.DataLayouts;
-import org.tensorflow.ndarray.FloatNdArray;
-import org.tensorflow.ndarray.NdArray;
-import org.tensorflow.ndarray.StdArrays;
-import org.tensorflow.ndarray.impl.dense.FloatDenseNdArray;
-import org.tensorflow.types.family.TFloating;
+import org.tensorflow.types.family.TFloat;
 
 /**
  * Brain 16-bit float tensor type.
@@ -48,12 +46,12 @@
  * <p>Note that some CPUs support the bfloat16 format natively, which can result in faster
  * computation compared to {@link TFloat16} when GPUs are not used.
  */
-public interface TBfloat16 extends FloatNdArray, TFloating {
+public interface TBfloat16 extends FloatNdArray, TNumber {
   /** readable-name for the data type */
   static final String NAME = "BFLOAT16";
 
   /** Type metadata */
-  DataType<TBfloat16> DTYPE = DataType.create(NAME, 14, 2, TBfloat16Impl::mapTensor);
+  DataType<TBfloat16> DTYPE = DataType.create(NAME, 14, 2, TBfloat16Impl::new);
 
   /**
    * Allocates a new tensor for storing a single float value.
@@ -61,8 +59,8 @@ public interface TBfloat16 extends FloatNdArray, TFloating {
    * @param value float to store in the new tensor
    * @return the new tensor
    */
-  static Tensor<TBfloat16> scalarOf(float value) {
-    return Tensor.of(DTYPE, Shape.scalar(), data -> data.setFloat(value));
+  static TBfloat16 scalarOf(float value) {
+    return Tensor.of(DTYPE, Shape.scalar(), t -> t.setFloat(value));
   }
 
   /**
@@ -71,11 +69,11 @@ static Tensor<TBfloat16> scalarOf(float value) {
    * @param values floats to store in the new tensor
    * @return the new tensor
    */
-  static Tensor<TBfloat16> vectorOf(float... values) {
+  static TBfloat16 vectorOf(float... values) {
     if (values == null) {
       throw new IllegalArgumentException();
     }
-    return Tensor.of(DTYPE, Shape.of(values.length), data -> StdArrays.copyTo(values, data));
+    return Tensor.of(DTYPE, Shape.of(values.length), t -> StdArrays.copyTo(values, t));
   }
 
   /**
@@ -86,7 +84,7 @@ static Tensor<TBfloat16> vectorOf(float... values) {
    * @param src the source array giving the shape and data to the new tensor
    * @return the new tensor
    */
-  static Tensor<TBfloat16> tensorOf(NdArray<Float> src) {
+  static TBfloat16 tensorOf(NdArray<Float> src) {
     return Tensor.of(DTYPE, src.shape(), src::copyTo);
   }
 
@@ -96,7 +94,7 @@ static Tensor<TBfloat16> tensorOf(NdArray<Float> src) {
    * @param shape shape of the tensor to allocate
    * @return the new tensor
    */
-  static Tensor<TBfloat16> tensorOf(Shape shape) {
+  static TBfloat16 tensorOf(Shape shape) {
     return Tensor.of(DTYPE, shape);
   }
 
@@ -107,32 +105,34 @@ static Tensor<TBfloat16> tensorOf(Shape shape) {
    * @param data buffer of floats to initialize the tensor with
    * @return the new tensor
    */
-  static Tensor<TBfloat16> tensorOf(Shape shape, FloatDataBuffer data) {
-    return Tensor.of(DTYPE, shape, d -> d.write(data));
+  static TBfloat16 tensorOf(Shape shape, FloatDataBuffer data) {
+    return Tensor.of(DTYPE, shape, t -> t.write(data));
   }
 
   /**
    * Allocates a new tensor of the given shape and initialize its data.
    *
    * @param shape shape of the tensor to allocate
-   * @param dataInit tensor data initializer
+   * @param tensorInit tensor data initializer
    * @return the new tensor
    * @throws TensorFlowException if the tensor cannot be allocated or initialized
    */
-  static Tensor<TBfloat16> tensorOf(Shape shape, Consumer<TBfloat16> dataInit) {
-    return Tensor.of(DTYPE, shape, dataInit);
+  static TBfloat16 tensorOf(Shape shape, Consumer<TBfloat16> tensorInit) {
+    return Tensor.of(DTYPE, shape, tensorInit);
   }
 }
 
-/** Hidden implementation of a {@code TBfloat16} */
-class TBfloat16Impl extends FloatDenseNdArray implements TBfloat16 {
+/**
+ * Hidden implementation of a {@code TBfloat16}
+ */
+class TBfloat16Impl extends FloatTensorImpl<TBfloat16> implements TBfloat16 {
 
-  static TBfloat16 mapTensor(TF_Tensor nativeTensor, Shape shape) {
-    return new TBfloat16Impl(
-        DataLayouts.BFLOAT16.applyTo(TensorBuffers.toShorts(nativeTensor)), shape);
+  TBfloat16Impl(TF_Tensor nativeTensorHandle, Shape shape) {
+    super(nativeTensorHandle, DTYPE, shape, mapMemory(nativeTensorHandle));
   }
 
-  private TBfloat16Impl(FloatDataBuffer buffer, Shape shape) {
-    super(buffer, shape);
+  private static FloatDataBuffer mapMemory(TF_Tensor nativeTensorHandle) {
+    return DataLayouts.BFLOAT16.applyTo(TensorBuffers.toShorts(nativeTensorHandle));
   }
 }
+
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TBool.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TBool.java
index 3cc72101893..ec39f0a73b5 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TBool.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TBool.java
@@ -17,22 +17,19 @@
 
 package org.tensorflow.types;
 
+import java.util.function.Consumer;
 import org.tensorflow.DataType;
 import org.tensorflow.Tensor;
 import org.tensorflow.exceptions.TensorFlowException;
 import org.tensorflow.internal.buffer.TensorBuffers;
 import org.tensorflow.internal.c_api.TF_Tensor;
-import org.tensorflow.ndarray.BooleanNdArray;
 import org.tensorflow.ndarray.NdArray;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.ndarray.StdArrays;
 import org.tensorflow.ndarray.buffer.BooleanDataBuffer;
 import org.tensorflow.ndarray.buffer.layout.DataLayouts;
-import org.tensorflow.ndarray.impl.dense.BooleanDenseNdArray;
 import org.tensorflow.types.family.TType;
 
-import java.util.function.Consumer;
-
 /**
  * Boolean tensor type.
  *
@@ -40,12 +37,13 @@
  * explicit mapping between Java boolean values and byte buffers using the {@link DataLayouts#BOOL
  * BOOL} layout, which may impact I/O performances.
  */
-public interface TBool extends BooleanNdArray, TType {
+public interface TBool extends BooleanTensor<TBool>, TType {
+
   /** readable-name for the data type */
   static final String NAME = "BOOL";
 
   /** Type metadata */
-  DataType<TBool> DTYPE = DataType.create(NAME, 10, 1, TBoolImpl::mapTensor);
+  DataType<TBool> DTYPE = DataType.create(NAME, 10, 1, TBoolImpl::new);
 
   /**
    * Allocates a new tensor for storing a single boolean value.
@@ -53,8 +51,8 @@ public interface TBool extends BooleanNdArray, TType {
    * @param value boolean to store in the new tensor
    * @return the new tensor
    */
-  static Tensor<TBool> scalarOf(boolean value) {
-    return Tensor.of(DTYPE, Shape.scalar(), data -> data.setBoolean(value));
+  static TBool scalarOf(boolean value) {
+    return Tensor.of(DTYPE, Shape.scalar(), t -> t.setBoolean(value));
   }
 
   /**
@@ -63,11 +61,11 @@ static Tensor<TBool> scalarOf(boolean value) {
    * @param values booleans to store in the new tensor
    * @return the new tensor
    */
-  static Tensor<TBool> vectorOf(boolean... values) {
+  static TBool vectorOf(boolean... values) {
     if (values == null) {
       throw new IllegalArgumentException();
     }
-    return Tensor.of(DTYPE, Shape.of(values.length), data -> StdArrays.copyTo(values, data));
+    return Tensor.of(DTYPE, Shape.of(values.length), t -> StdArrays.copyTo(values, t));
   }
 
   /**
@@ -78,7 +76,7 @@ static Tensor<TBool> vectorOf(boolean... values) {
    * @param src the source array giving the shape and data to the new tensor
    * @return the new tensor
    */
-  static Tensor<TBool> tensorOf(NdArray<Boolean> src) {
+  static TBool tensorOf(NdArray<Boolean> src) {
     return Tensor.of(DTYPE, src.shape(), src::copyTo);
   }
 
@@ -88,7 +86,7 @@ static Tensor<TBool> tensorOf(NdArray<Boolean> src) {
    * @param shape shape of the tensor to allocate
    * @return the new tensor
    */
-  static Tensor<TBool> tensorOf(Shape shape) {
+  static TBool tensorOf(Shape shape) {
     return Tensor.of(DTYPE, shape);
   }
 
@@ -99,7 +97,7 @@ static Tensor<TBool> tensorOf(Shape shape) {
    * @param data buffer of booleans to initialize the tensor with
    * @return the new tensor
    */
-  static Tensor<TBool> tensorOf(Shape shape, BooleanDataBuffer data) {
+  static TBool tensorOf(Shape shape, BooleanDataBuffer data) {
     return Tensor.of(DTYPE, shape, d -> d.write(data));
   }
 
@@ -107,23 +105,21 @@ static Tensor<TBool> tensorOf(Shape shape, BooleanDataBuffer data) {
    * Allocates a new tensor of the given shape and initialize its data.
    *
    * @param shape shape of the tensor to allocate
-   * @param dataInit tensor data initializer
+   * @param tensorInit tensor data initializer
    * @return the new tensor
    * @throws TensorFlowException if the tensor cannot be allocated or initialized
    */
-  static Tensor<TBool> tensorOf(Shape shape, Consumer<TBool> dataInit) {
-    return Tensor.of(DTYPE, shape, dataInit);
+  static TBool tensorOf(Shape shape, Consumer<TBool> tensorInit) {
+    return Tensor.of(DTYPE, shape, tensorInit);
   }
 }
 
-/** Hidden implementation of a {@code TBool} */
-class TBoolImpl extends BooleanDenseNdArray implements TBool {
-
-  static TBool mapTensor(TF_Tensor nativeTensor, Shape shape) {
-    return new TBoolImpl(TensorBuffers.toBooleans(nativeTensor), shape);
-  }
+/**
+ * Hidden implementation of a {@code TBool}
+ */
+class TBoolImpl extends BooleanTensorImpl<TBool> implements TBool {
 
-  private TBoolImpl(BooleanDataBuffer buffer, Shape shape) {
-    super(buffer, shape);
+  TBoolImpl(TF_Tensor nativeTensor, Shape shape) {
+    super(nativeTensor, DTYPE, shape, TensorBuffers.toBooleans(nativeTensor));
   }
 }
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TFloat16.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TFloat16.java
index 0cd441a1ff1..87b083dc8f0 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TFloat16.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TFloat16.java
@@ -23,14 +23,12 @@
 import org.tensorflow.exceptions.TensorFlowException;
 import org.tensorflow.internal.buffer.TensorBuffers;
 import org.tensorflow.internal.c_api.TF_Tensor;
+import org.tensorflow.ndarray.NdArray;
 import org.tensorflow.ndarray.Shape;
+import org.tensorflow.ndarray.StdArrays;
 import org.tensorflow.ndarray.buffer.FloatDataBuffer;
 import org.tensorflow.ndarray.buffer.layout.DataLayouts;
-import org.tensorflow.ndarray.FloatNdArray;
-import org.tensorflow.ndarray.NdArray;
-import org.tensorflow.ndarray.StdArrays;
-import org.tensorflow.ndarray.impl.dense.FloatDenseNdArray;
-import org.tensorflow.types.family.TFloating;
+import org.tensorflow.types.family.TFloat;
 
 /**
  * IEEE-754 half-precision 16-bit float tensor type.
@@ -45,13 +43,13 @@
  * most CPUs do not support this format natively. For CPU computation on 16-bit floats, the {@link
  * TBfloat16} tensor type might be a better option.
  */
-public interface TFloat16 extends FloatNdArray, TFloating {
+public interface TFloat16 extends FloatTensor<TFloat16>, TFloat {
 
   /** readable-name for the data type */
   static final String NAME = "FLOAT16";
 
   /** Type metadata */
-  DataType<TFloat16> DTYPE = DataType.create(NAME, 19, 2, TFloat16Impl::mapTensor);
+  DataType<TFloat16> DTYPE = DataType.create(NAME, 19, 2, TFloat16Impl::new);
 
   /**
    * Allocates a new tensor for storing a single float value.
@@ -59,8 +57,8 @@ public interface TFloat16 extends FloatNdArray, TFloating {
    * @param value float to store in the new tensor
    * @return the new tensor
    */
-  static Tensor<TFloat16> scalarOf(float value) {
-    return Tensor.of(DTYPE, Shape.scalar(), data -> data.setFloat(value));
+  static TFloat16 scalarOf(float value) {
+    return Tensor.of(DTYPE, Shape.scalar(), t -> t.setFloat(value));
   }
 
   /**
@@ -69,11 +67,11 @@ static Tensor<TFloat16> scalarOf(float value) {
    * @param values floats to store in the new tensor
    * @return the new tensor
    */
-  static Tensor<TFloat16> vectorOf(float... values) {
+  static TFloat16 vectorOf(float... values) {
     if (values == null) {
       throw new IllegalArgumentException();
     }
-    return Tensor.of(DTYPE, Shape.of(values.length), data -> StdArrays.copyTo(values, data));
+    return Tensor.of(DTYPE, Shape.of(values.length), t -> StdArrays.copyTo(values, t));
   }
 
   /**
@@ -84,7 +82,7 @@ static Tensor<TFloat16> vectorOf(float... values) {
    * @param src the source array giving the shape and data to the new tensor
    * @return the new tensor
    */
-  static Tensor<TFloat16> tensorOf(NdArray<Float> src) {
+  static TFloat16 tensorOf(NdArray<Float> src) {
     return Tensor.of(DTYPE, src.shape(), src::copyTo);
   }
 
@@ -94,7 +92,7 @@ static Tensor<TFloat16> tensorOf(NdArray<Float> src) {
    * @param shape shape of the tensor to allocate
    * @return the new tensor
    */
-  static Tensor<TFloat16> tensorOf(Shape shape) {
+  static TFloat16 tensorOf(Shape shape) {
     return Tensor.of(DTYPE, shape);
   }
 
@@ -105,32 +103,34 @@ static Tensor<TFloat16> tensorOf(Shape shape) {
    * @param data buffer of floats to initialize the tensor with
    * @return the new tensor
    */
-  static Tensor<TFloat16> tensorOf(Shape shape, FloatDataBuffer data) {
-    return Tensor.of(DTYPE, shape, d -> d.write(data));
+  static TFloat16 tensorOf(Shape shape, FloatDataBuffer data) {
+    return Tensor.of(DTYPE, shape, t -> t.write(data));
   }
 
   /**
    * Allocates a new tensor of the given shape and initialize its data.
    *
    * @param shape shape of the tensor to allocate
-   * @param dataInit tensor data initializer
+   * @param tensorInit tensor data initializer
    * @return the new tensor
    * @throws TensorFlowException if the tensor cannot be allocated or initialized
    */
-  static Tensor<TFloat16> tensorOf(Shape shape, Consumer<TFloat16> dataInit) {
-    return Tensor.of(DTYPE, shape, dataInit);
+  static TFloat16 tensorOf(Shape shape, Consumer<TFloat16> tensorInit) {
+    return Tensor.of(DTYPE, shape, tensorInit);
   }
 }
 
-/** Hidden implementation of a {@code TFloat16} */
-class TFloat16Impl extends FloatDenseNdArray implements TFloat16 {
+/**
+ * Hidden implementation of a {@code TFloat16}
+ */
+class TFloat16Impl extends FloatTensorImpl<TFloat16> implements TFloat16 {
 
-  static TFloat16 mapTensor(TF_Tensor nativeTensor, Shape shape) {
-    return new TFloat16Impl(
-        DataLayouts.FLOAT16.applyTo(TensorBuffers.toShorts(nativeTensor)), shape);
+  TFloat16Impl(TF_Tensor nativeTensor, Shape shape) {
+    super(nativeTensor, DTYPE, shape, mapMemory(nativeTensor));
   }
 
-  private TFloat16Impl(FloatDataBuffer buffer, Shape shape) {
-    super(buffer, shape);
+  private static FloatDataBuffer mapMemory(TF_Tensor nativeTensor) {
+    return DataLayouts.FLOAT16.applyTo(TensorBuffers.toShorts(nativeTensor));
   }
 }
+
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TFloat32.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TFloat32.java
index 571ec118ddc..73071c31ae6 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TFloat32.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TFloat32.java
@@ -23,22 +23,20 @@
 import org.tensorflow.exceptions.TensorFlowException;
 import org.tensorflow.internal.buffer.TensorBuffers;
 import org.tensorflow.internal.c_api.TF_Tensor;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.ndarray.buffer.FloatDataBuffer;
-import org.tensorflow.ndarray.FloatNdArray;
 import org.tensorflow.ndarray.NdArray;
+import org.tensorflow.ndarray.Shape;
 import org.tensorflow.ndarray.StdArrays;
 import org.tensorflow.ndarray.impl.dense.FloatDenseNdArray;
-import org.tensorflow.types.family.TFloating;
+import org.tensorflow.types.family.TNumber;
 
 /** IEEE-754 single-precision 32-bit float tensor type. */
-public interface TFloat32 extends FloatNdArray, TFloating {
+public interface TFloat32 extends FloatNdArray, TNumber {
 
   /** readable-name for the data type */
   static final String NAME = "FLOAT";
 
   /** Type metadata */
-  DataType<TFloat32> DTYPE = DataType.create(NAME, 1, 4, TFloat32Impl::mapTensor);
+  DataType<TFloat32> DTYPE = DataType.create(NAME, 1, 4, TFloat32Impl::new);
 
   /**
    * Allocates a new tensor for storing a single float value.
@@ -46,8 +44,8 @@ public interface TFloat32 extends FloatNdArray, TFloating {
    * @param value float to store in the new tensor
    * @return the new tensor
    */
-  static Tensor<TFloat32> scalarOf(float value) {
-    return Tensor.of(DTYPE, Shape.scalar(), data -> data.setFloat(value));
+  static TFloat32 scalarOf(float value) {
+    return Tensor.of(DTYPE, Shape.scalar(), t -> t.setFloat(value));
   }
 
   /**
@@ -56,11 +54,11 @@ static Tensor<TFloat32> scalarOf(float value) {
    * @param values floats to store in the new tensor
    * @return the new tensor
    */
-  static Tensor<TFloat32> vectorOf(float... values) {
+  static TFloat32 vectorOf(float... values) {
     if (values == null) {
       throw new IllegalArgumentException();
     }
-    return Tensor.of(DTYPE, Shape.of(values.length), data -> StdArrays.copyTo(values, data));
+    return Tensor.of(DTYPE, Shape.of(values.length), t -> StdArrays.copyTo(values, t));
   }
 
   /**
@@ -71,7 +69,7 @@ static Tensor<TFloat32> vectorOf(float... values) {
    * @param src the source array giving the shape and data to the new tensor
    * @return the new tensor
    */
-  static Tensor<TFloat32> tensorOf(NdArray<Float> src) {
+  static TFloat32 tensorOf(NdArray<Float> src) {
     return Tensor.of(DTYPE, src.shape(), src::copyTo);
   }
 
@@ -81,7 +79,7 @@ static Tensor<TFloat32> tensorOf(NdArray<Float> src) {
    * @param shape shape of the tensor to allocate
    * @return the new tensor
    */
-  static Tensor<TFloat32> tensorOf(Shape shape) {
+  static TFloat32 tensorOf(Shape shape) {
     return Tensor.of(DTYPE, shape);
   }
 
@@ -92,31 +90,30 @@ static Tensor<TFloat32> tensorOf(Shape shape) {
    * @param data buffer of floats to initialize the tensor with
    * @return the new tensor
    */
-  static Tensor<TFloat32> tensorOf(Shape shape, FloatDataBuffer data) {
-    return Tensor.of(DTYPE, shape, d -> d.write(data));
+  static TFloat32 tensorOf(Shape shape, FloatDataBuffer data) {
+    return Tensor.of(DTYPE, shape, t -> t.write(data));
   }
 
   /**
    * Allocates a new tensor of the given shape and initialize its data.
    *
    * @param shape shape of the tensor to allocate
-   * @param dataInit tensor data initializer
+   * @param tensorInit tensor data initializer
    * @return the new tensor
    * @throws TensorFlowException if the tensor cannot be allocated or initialized
    */
-  static Tensor<TFloat32> tensorOf(Shape shape, Consumer<TFloat32> dataInit) {
-    return Tensor.of(DTYPE, shape, dataInit);
+  static TFloat32 tensorOf(Shape shape, Consumer<TFloat32> tensorInit) {
+    return Tensor.of(DTYPE, shape, tensorInit);
   }
 }
 
-/** Hidden implementation of a {@code TFloat32} */
-class TFloat32Impl extends FloatDenseNdArray implements TFloat32 {
-
-  static TFloat32 mapTensor(TF_Tensor nativeTensor, Shape shape) {
-    return new TFloat32Impl(TensorBuffers.toFloats(nativeTensor), shape);
-  }
+/**
+ * Hidden implementation of a {@code TFloat32}
+ */
+class TFloat32Impl extends FloatTensorImpl<TFloat32> implements TFloat32 {
 
-  private TFloat32Impl(FloatDataBuffer buffer, Shape shape) {
-    super(buffer, shape);
+  TFloat32Impl(TF_Tensor nativeTensor, Shape shape) {
+    super(nativeTensor, DTYPE, shape, TensorBuffers.toFloats(nativeTensor));
   }
 }
+
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TFloat64.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TFloat64.java
index 5d2744c4b3c..fa2af3f2e60 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TFloat64.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TFloat64.java
@@ -23,23 +23,20 @@
 import org.tensorflow.exceptions.TensorFlowException;
 import org.tensorflow.internal.buffer.TensorBuffers;
 import org.tensorflow.internal.c_api.TF_Tensor;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.ndarray.buffer.DoubleDataBuffer;
-import org.tensorflow.ndarray.DoubleNdArray;
 import org.tensorflow.ndarray.NdArray;
+import org.tensorflow.ndarray.Shape;
 import org.tensorflow.ndarray.StdArrays;
 import org.tensorflow.ndarray.impl.dense.DoubleDenseNdArray;
-import org.tensorflow.types.family.TFloating;
-
+import org.tensorflow.types.family.TNumber;
 
 /** IEEE-754 double-precision 64-bit float tensor type. */
-public interface TFloat64 extends DoubleNdArray, TFloating {
+public interface TFloat64 extends DoubleNdArray, TNumber {
 
   /** readable-name for the data type */
   static final String NAME = "DOUBLE";
 
   /** Type metadata */
-  DataType<TFloat64> DTYPE = DataType.create(NAME, 2, 8, TFloat64Impl::mapTensor);
+  DataType<TFloat64> DTYPE = DataType.create(NAME, 2, 8, TFloat64Impl::new);
 
   /**
    * Allocates a new tensor for storing a single double value.
@@ -47,8 +44,8 @@ public interface TFloat64 extends DoubleNdArray, TFloating {
    * @param value double to store in the new tensor
    * @return the new tensor
    */
-  static Tensor<TFloat64> scalarOf(double value) {
-    return Tensor.of(DTYPE, Shape.scalar(), data -> data.setDouble(value));
+  static TFloat64 scalarOf(double value) {
+    return Tensor.of(DTYPE, Shape.scalar(), t -> t.setDouble(value));
   }
 
   /**
@@ -57,11 +54,11 @@ static Tensor<TFloat64> scalarOf(double value) {
    * @param values doubles to store in the new tensor
    * @return the new tensor
    */
-  static Tensor<TFloat64> vectorOf(double... values) {
+  static TFloat64 vectorOf(double... values) {
     if (values == null) {
       throw new IllegalArgumentException();
     }
-    return Tensor.of(DTYPE, Shape.of(values.length), data -> StdArrays.copyTo(values, data));
+    return Tensor.of(DTYPE, Shape.of(values.length), t -> StdArrays.copyTo(values, t));
   }
 
   /**
@@ -72,7 +69,7 @@ static Tensor<TFloat64> vectorOf(double... values) {
    * @param src the source array giving the shape and data to the new tensor
    * @return the new tensor
    */
-  static Tensor<TFloat64> tensorOf(NdArray<Double> src) {
+  static TFloat64 tensorOf(NdArray<Double> src) {
     return Tensor.of(DTYPE, src.shape(), src::copyTo);
   }
 
@@ -82,7 +79,7 @@ static Tensor<TFloat64> tensorOf(NdArray<Double> src) {
    * @param shape shape of the tensor to allocate
    * @return the new tensor
    */
-  static Tensor<TFloat64> tensorOf(Shape shape) {
+  static TFloat64 tensorOf(Shape shape) {
     return Tensor.of(DTYPE, shape);
   }
 
@@ -93,31 +90,29 @@ static Tensor<TFloat64> tensorOf(Shape shape) {
    * @param data buffer of doubles to initialize the tensor with
    * @return the new tensor
    */
-  static Tensor<TFloat64> tensorOf(Shape shape, DoubleDataBuffer data) {
-    return Tensor.of(DTYPE, shape, d -> d.write(data));
+  static TFloat64 tensorOf(Shape shape, DoubleDataBuffer data) {
+    return Tensor.of(DTYPE, shape, t -> t.write(data));
   }
 
   /**
    * Allocates a new tensor of the given shape and initialize its data.
    *
    * @param shape shape of the tensor to allocate
-   * @param dataInit tensor data initializer
+   * @param tensorInit tensor data initializer
    * @return the new tensor
    * @throws TensorFlowException if the tensor cannot be allocated or initialized
    */
-  static Tensor<TFloat64> tensorOf(Shape shape, Consumer<TFloat64> dataInit) {
-    return Tensor.of(DTYPE, shape, dataInit);
+  static TFloat64 tensorOf(Shape shape, Consumer<TFloat64> tensorInit) {
+    return Tensor.of(DTYPE, shape, tensorInit);
   }
 }
 
-/** Hidden implementation of a {@code TFloat64} */
-class TFloat64Impl extends DoubleDenseNdArray implements TFloat64 {
-
-  static TFloat64 mapTensor(TF_Tensor nativeTensor, Shape shape) {
-    return new TFloat64Impl(TensorBuffers.toDoubles(nativeTensor), shape);
-  }
+/**
+ * Hidden implementation of a {@code TFloat64}
+ */
+class TFloat64Impl extends DoubleTensorImpl<TFloat64> implements TFloat64 {
 
-  private TFloat64Impl(DoubleDataBuffer buffer, Shape shape) {
-    super(buffer, shape);
+  TFloat64Impl(TF_Tensor nativeTensor, Shape shape) {
+    super(nativeTensor, DTYPE, shape, TensorBuffers.toDoubles(nativeTensor));
   }
 }
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TInt32.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TInt32.java
index 4a1139ddde2..d84cb16b9f0 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TInt32.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TInt32.java
@@ -22,22 +22,22 @@
 import org.tensorflow.Tensor;
 import org.tensorflow.internal.buffer.TensorBuffers;
 import org.tensorflow.internal.c_api.TF_Tensor;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.ndarray.buffer.IntDataBuffer;
-import org.tensorflow.ndarray.IntNdArray;
 import org.tensorflow.ndarray.NdArray;
+import org.tensorflow.ndarray.Shape;
 import org.tensorflow.ndarray.StdArrays;
-import org.tensorflow.ndarray.impl.dense.IntDenseNdArray;
+import org.tensorflow.ndarray.buffer.IntDataBuffer;
 import org.tensorflow.types.family.TNumber;
 
-/** 32-bit signed integer tensor type. */
-public interface TInt32 extends IntNdArray, TNumber {
+/**
+ * 32-bit signed integer tensor type.
+ */
+public interface TInt32 extends IntTensor<TInt32>, TNumber {
 
   /** readable-name for the data type */
   static final String NAME = "INT32";
 
   /** Type metadata */
-  DataType<TInt32> DTYPE = DataType.create(NAME, 3, 4, TInt32Impl::mapTensor);
+  DataType<TInt32> DTYPE = DataType.create(NAME, 3, 4, TInt32Impl::new);
 
   /**
    * Allocates a new tensor for storing a single int value.
@@ -45,8 +45,8 @@ public interface TInt32 extends IntNdArray, TNumber {
    * @param value int to store in the new tensor
    * @return the new tensor
    */
-  static Tensor<TInt32> scalarOf(int value) {
-    return Tensor.of(DTYPE, Shape.scalar(), data -> data.setInt(value));
+  static TInt32 scalarOf(int value) {
+    return Tensor.of(DTYPE, Shape.scalar(), t -> t.setInt(value));
   }
 
   /**
@@ -56,11 +56,11 @@ static Tensor<TInt32> scalarOf(int value) {
    * @return the new tensor
    * @throws IllegalArgumentException if no values are provided
    */
-  static Tensor<TInt32> vectorOf(int... values) {
+  static TInt32 vectorOf(int... values) {
     if (values == null) {
       throw new IllegalArgumentException();
     }
-    return Tensor.of(DTYPE, Shape.of(values.length), data -> StdArrays.copyTo(values, data));
+    return Tensor.of(DTYPE, Shape.of(values.length), t -> StdArrays.copyTo(values, t));
   }
 
   /**
@@ -71,7 +71,7 @@ static Tensor<TInt32> vectorOf(int... values) {
    * @param src the source array giving the shape and data to the new tensor
    * @return the new tensor
    */
-  static Tensor<TInt32> tensorOf(NdArray<Integer> src) {
+  static TInt32 tensorOf(NdArray<Integer> src) {
     return Tensor.of(DTYPE, src.shape(), src::copyTo);
   }
 
@@ -81,7 +81,7 @@ static Tensor<TInt32> tensorOf(NdArray<Integer> src) {
    * @param shape shape of the tensor to allocate
    * @return the new tensor
    */
-  static Tensor<TInt32> tensorOf(Shape shape) {
+  static TInt32 tensorOf(Shape shape) {
     return Tensor.of(DTYPE, shape);
   }
 
@@ -92,30 +92,28 @@ static Tensor<TInt32> tensorOf(Shape shape) {
    * @param data buffer of ints to initialize the tensor with
    * @return the new tensor
    */
-  static Tensor<TInt32> tensorOf(Shape shape, IntDataBuffer data) {
-    return Tensor.of(DTYPE, shape, d -> d.write(data));
+  static TInt32 tensorOf(Shape shape, IntDataBuffer data) {
+    return Tensor.of(DTYPE, shape, t -> t.write(data));
   }
 
   /**
    * Allocates a new tensor of the given shape and initialize its data.
    *
    * @param shape shape of the tensor to allocate
-   * @param dataInit tensor data initializer
+   * @param tensorInit tensor data initializer
    * @return the new tensor
    */
-  static Tensor<TInt32> tensorOf(Shape shape, Consumer<TInt32> dataInit) {
-    return Tensor.of(DTYPE, shape, dataInit);
+  static TInt32 tensorOf(Shape shape, Consumer<TInt32> tensorInit) {
+    return Tensor.of(DTYPE, shape, tensorInit);
   }
 }
 
-/** Hidden implementation of a {@code TInt32} */
-class TInt32Impl extends IntDenseNdArray implements TInt32 {
-
-  static TInt32 mapTensor(TF_Tensor nativeTensor, Shape shape) {
-    return new TInt32Impl(TensorBuffers.toInts(nativeTensor), shape);
-  }
+/**
+ * Hidden implementation of a {@code TInt32}
+ */
+class TInt32Impl extends IntTensorImpl<TInt32> implements TInt32 {
 
-  private TInt32Impl(IntDataBuffer buffer, Shape shape) {
-    super(buffer, shape);
+  TInt32Impl(TF_Tensor nativeTensor, Shape shape) {
+    super(nativeTensor, DTYPE, shape, TensorBuffers.toInts(nativeTensor));
   }
 }
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TInt64.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TInt64.java
index 04fd4fd7799..5c15c379c09 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TInt64.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TInt64.java
@@ -23,22 +23,22 @@
 import org.tensorflow.exceptions.TensorFlowException;
 import org.tensorflow.internal.buffer.TensorBuffers;
 import org.tensorflow.internal.c_api.TF_Tensor;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.ndarray.buffer.LongDataBuffer;
-import org.tensorflow.ndarray.LongNdArray;
 import org.tensorflow.ndarray.NdArray;
+import org.tensorflow.ndarray.Shape;
 import org.tensorflow.ndarray.StdArrays;
-import org.tensorflow.ndarray.impl.dense.LongDenseNdArray;
+import org.tensorflow.ndarray.buffer.LongDataBuffer;
 import org.tensorflow.types.family.TNumber;
 
-/** 64-bit signed integer tensor type. */
-public interface TInt64 extends LongNdArray, TNumber {
+/**
+ * 64-bit signed integer tensor type.
+ */
+public interface TInt64 extends LongTensor<TInt64>, TNumber {
 
   /** readable-name for the data type */
   static final String NAME = "INT64";
 
   /** Type metadata */
-  DataType<TInt64> DTYPE = DataType.create(NAME, 9, 8, TInt64Impl::mapTensor);
+  DataType<TInt64> DTYPE = DataType.create(NAME, 9, 8, TInt64Impl::new);
 
   /**
    * Allocates a new tensor for storing a single long value.
@@ -46,8 +46,8 @@ public interface TInt64 extends LongNdArray, TNumber {
    * @param value long to store in the new tensor
    * @return the new tensor
    */
-  static Tensor<TInt64> scalarOf(long value) {
-    return Tensor.of(DTYPE, Shape.scalar(), data -> data.setLong(value));
+  static TInt64 scalarOf(long value) {
+    return Tensor.of(DTYPE, Shape.scalar(), t -> t.setLong(value));
   }
 
   /**
@@ -56,11 +56,11 @@ static Tensor<TInt64> scalarOf(long value) {
    * @param values longs to store in the new tensor
    * @return the new tensor
    */
-  static Tensor<TInt64> vectorOf(long... values) {
+  static TInt64 vectorOf(long... values) {
     if (values == null) {
       throw new IllegalArgumentException();
     }
-    return Tensor.of(DTYPE, Shape.of(values.length), data -> StdArrays.copyTo(values, data));
+    return Tensor.of(DTYPE, Shape.of(values.length), t -> StdArrays.copyTo(values, t));
   }
 
   /**
@@ -71,7 +71,7 @@ static Tensor<TInt64> vectorOf(long... values) {
    * @param src the source array giving the shape and data to the new tensor
    * @return the new tensor
    */
-  static Tensor<TInt64> tensorOf(NdArray<Long> src) {
+  static TInt64 tensorOf(NdArray<Long> src) {
     return Tensor.of(DTYPE, src.shape(), src::copyTo);
   }
 
@@ -81,7 +81,7 @@ static Tensor<TInt64> tensorOf(NdArray<Long> src) {
    * @param shape shape of the tensor to allocate
    * @return the new tensor
    */
-  static Tensor<TInt64> tensorOf(Shape shape) {
+  static TInt64 tensorOf(Shape shape) {
     return Tensor.of(DTYPE, shape);
   }
 
@@ -92,31 +92,29 @@ static Tensor<TInt64> tensorOf(Shape shape) {
    * @param data buffer of longs to initialize the tensor with
    * @return the new tensor
    */
-  static Tensor<TInt64> tensorOf(Shape shape, LongDataBuffer data) {
-    return Tensor.of(DTYPE, shape, d -> d.write(data));
+  static TInt64 tensorOf(Shape shape, LongDataBuffer data) {
+    return Tensor.of(DTYPE, shape, t -> t.write(data));
   }
 
   /**
    * Allocates a new tensor of the given shape and initialize its data.
    *
    * @param shape shape of the tensor to allocate
-   * @param dataInit tensor data initializer
+   * @param tensorInit tensor data initializer
    * @return the new tensor
    * @throws TensorFlowException if the tensor cannot be allocated or initialized
    */
-  static Tensor<TInt64> tensorOf(Shape shape, Consumer<TInt64> dataInit) {
-    return Tensor.of(DTYPE, shape, dataInit);
+  static TInt64 tensorOf(Shape shape, Consumer<TInt64> tensorInit) {
+    return Tensor.of(DTYPE, shape, tensorInit);
   }
 }
 
-/** Hidden implementation of a {@code TInt64} */
-class TInt64Impl extends LongDenseNdArray implements TInt64 {
-
-  static TInt64 mapTensor(TF_Tensor nativeTensor, Shape shape) {
-    return new TInt64Impl(TensorBuffers.toLongs(nativeTensor), shape);
-  }
+/**
+ * Hidden implementation of a {@code TInt64}
+ */
+class TInt64Impl extends LongTensorImpl<TInt64> implements TInt64 {
 
-  private TInt64Impl(LongDataBuffer buffer, Shape shape) {
-    super(buffer, shape);
+  TInt64Impl(TF_Tensor nativeTensor, Shape shape) {
+    super(nativeTensor, DTYPE, shape, TensorBuffers.toLongs(nativeTensor));
   }
 }
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TString.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TString.java
index 57a121edcf1..9cfcc0b0bbf 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TString.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TString.java
@@ -17,9 +17,12 @@
 
 package org.tensorflow.types;
 
+import java.nio.charset.Charset;
+import java.nio.charset.StandardCharsets;
+import java.util.function.Function;
 import org.tensorflow.DataType;
 import org.tensorflow.Tensor;
-import org.tensorflow.internal.buffer.StringTensorBuffer;
+import org.tensorflow.internal.buffer.ByteSequenceTensorBuffer;
 import org.tensorflow.internal.buffer.TensorBuffers;
 import org.tensorflow.internal.c_api.TF_Tensor;
 import org.tensorflow.ndarray.NdArray;
@@ -28,13 +31,8 @@
 import org.tensorflow.ndarray.buffer.DataBuffer;
 import org.tensorflow.ndarray.buffer.layout.DataLayout;
 import org.tensorflow.ndarray.buffer.layout.DataLayouts;
-import org.tensorflow.ndarray.impl.dense.DenseNdArray;
 import org.tensorflow.types.family.TType;
 
-import java.nio.charset.Charset;
-import java.nio.charset.StandardCharsets;
-import java.util.function.Function;
-
 /**
  * String type.
  *
@@ -44,13 +42,15 @@
  * its values initially, so TensorFlow can compute and allocate the right amount of memory. Then the
  * data in the tensor is initialized once and cannot be modified afterwards.
  */
-public interface TString extends NdArray<String>, TType {
+public interface TString extends StringTensor<TString>, TType {
 
   /** readable-name for the data type */
   static final String NAME = "STRING";
 
-  /** Type metadata */
-  DataType<TString> DTYPE = DataType.create(NAME, 7, -1, TStringImpl::mapTensor);
+  /**
+   * Type metadata
+   */
+  DataType<TString> DTYPE = DataType.create(NAME, 7, -1, TStringImpl::new);
 
   /**
    * Allocates a new tensor for storing a string scalar.
@@ -60,7 +60,7 @@ public interface TString extends NdArray<String>, TType {
    * @param value scalar value to store in the new tensor
    * @return the new tensor
    */
-  static Tensor<TString> scalarOf(String value) {
+  static TString scalarOf(String value) {
     return tensorOf(NdArrays.scalarOfObject(value));
   }
 
@@ -72,7 +72,7 @@ static Tensor<TString> scalarOf(String value) {
    * @param values values to store in the new tensor
    * @return the new tensor
    */
-  static Tensor<TString> vectorOf(String... values) {
+  static TString vectorOf(String... values) {
     if (values == null) {
       throw new IllegalArgumentException();
     }
@@ -88,7 +88,7 @@ static Tensor<TString> vectorOf(String... values) {
    * @param src the source array giving the shape and data to the new tensor
    * @return the new tensor
    */
-  static Tensor<TString> tensorOf(NdArray<String> src) {
+  static TString tensorOf(NdArray<String> src) {
     return tensorOf(StandardCharsets.UTF_8, src);
   }
 
@@ -113,7 +113,7 @@ static Tensor<TString> tensorOf(NdArray<String> src) {
    * @param src the source array giving the shape and data to the new tensor
    * @return the new tensor
    */
-  static Tensor<TString> tensorOf(Charset charset, NdArray<String> src) {
+  static TString tensorOf(Charset charset, NdArray<String> src) {
     return TStringImpl.createTensor(src, s -> s.getBytes(charset));
   }
 
@@ -127,7 +127,7 @@ static Tensor<TString> tensorOf(Charset charset, NdArray<String> src) {
    * @param data buffer of strings to initialize the tensor with
    * @return the new tensor
    */
-  static Tensor<TString> tensorOf(Shape shape, DataBuffer<String> data) {
+  static TString tensorOf(Shape shape, DataBuffer<String> data) {
     return tensorOf(NdArrays.wrap(shape, data));
   }
 
@@ -154,7 +154,7 @@ static Tensor<TString> tensorOf(Shape shape, DataBuffer<String> data) {
    * @param data buffer of strings to initialize the tensor with
    * @return the new tensor
    */
-  static Tensor<TString> tensorOf(Charset charset, Shape shape, DataBuffer<String> data) {
+  static TString tensorOf(Charset charset, Shape shape, DataBuffer<String> data) {
     return tensorOf(charset, NdArrays.wrap(shape, data));
   }
 
@@ -173,7 +173,7 @@ static Tensor<TString> tensorOf(Charset charset, Shape shape, DataBuffer<String>
    * @param src the source array giving the shape and data to the new tensor
    * @return the new tensor
    */
-  static Tensor<TString> tensorOfBytes(NdArray<byte[]> src) {
+  static TString tensorOfBytes(NdArray<byte[]> src) {
     return TStringImpl.createTensor(src, Function.identity());
   }
 
@@ -193,7 +193,7 @@ static Tensor<TString> tensorOfBytes(NdArray<byte[]> src) {
    * @param data the source array giving the shape and data to the new tensor
    * @return the new tensor
    */
-  static Tensor<TString> tensorOfBytes(Shape shape, DataBuffer<byte[]> data) {
+  static TString tensorOfBytes(Shape shape, DataBuffer<byte[]> data) {
     return tensorOfBytes(NdArrays.wrap(shape, data));
   }
 
@@ -214,46 +214,33 @@ static Tensor<TString> tensorOfBytes(Shape shape, DataBuffer<byte[]> data) {
    * @return string tensor data using this charset
    */
   TString using(Charset charset);
-
-  /** @return the tensor data as a n-dimensional array of raw byte sequences. */
-  NdArray<byte[]> asBytes();
 }
 
-/** Hidden implementation of a {@code TString} */
-class TStringImpl extends DenseNdArray<String> implements TString {
+/**
+ * Hidden implementation of a {@code TString}
+ */
+class TStringImpl extends StringTensorImpl<TString> implements TString {
 
   @Override
   public TString using(Charset charset) {
-    return new TStringImpl(tensorBuffer, DataLayouts.ofStrings(charset), shape());
-  }
-
-  @Override
-  public NdArray<byte[]> asBytes() {
-    return NdArrays.wrap(shape(), tensorBuffer);
+    return new TStringImpl(nativeHandle(), shape(), DataLayouts.ofStrings(charset), rawBuffer());
   }
 
-  static <T> Tensor<TString> createTensor(NdArray<T> src, Function<T, byte[]> getBytes) {
-    long size = StringTensorBuffer.computeSize(src, getBytes);
-    return Tensor.of(
-        TString.DTYPE,
-        src.shape(),
-        size,
-        data -> ((TStringImpl) data).tensorBuffer.init(src, getBytes));
+  static <T> TString createTensor(NdArray<T> src, Function<T, byte[]> getBytes) {
+    long size = ByteSequenceTensorBuffer.computeSize(src, getBytes);
+    return Tensor.of(TString.DTYPE, src.shape(), size, t ->
+        ((TStringImpl)t).rawBuffer().init(src, getBytes)
+    );
   }
 
-  static TString mapTensor(TF_Tensor nativeTensor, Shape shape) {
-    StringTensorBuffer buffer = TensorBuffers.toStrings(nativeTensor, shape.size());
-    return new TStringImpl(buffer, UTF_8_LAYOUT, shape);
+  TStringImpl(TF_Tensor nativeTensor, Shape shape) {
+    this(nativeTensor, shape, UTF_8_LAYOUT, TensorBuffers.toStrings(nativeTensor, shape.size()));
   }
 
-  private static DataLayout<DataBuffer<byte[]>, String> UTF_8_LAYOUT =
-      DataLayouts.ofStrings(StandardCharsets.UTF_8);
+  private static final DataLayout<DataBuffer<byte[]>, String> UTF_8_LAYOUT = DataLayouts.ofStrings(StandardCharsets.UTF_8);
 
-  private final StringTensorBuffer tensorBuffer;
-
-  private TStringImpl(
-      StringTensorBuffer buffer, DataLayout<DataBuffer<byte[]>, String> layout, Shape shape) {
-    super(layout.applyTo(buffer), shape);
-    tensorBuffer = buffer;
+  private TStringImpl(TF_Tensor nativeTensor, Shape shape, DataLayout<DataBuffer<byte[]>, String> layout, ByteSequenceTensorBuffer rawBuffer) {
+    super(nativeTensor, TString.DTYPE, shape, layout, rawBuffer);
   }
 }
+
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TUint8.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TUint8.java
index 365f41196fb..c00063585e0 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TUint8.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TUint8.java
@@ -23,22 +23,22 @@
 import org.tensorflow.exceptions.TensorFlowException;
 import org.tensorflow.internal.buffer.TensorBuffers;
 import org.tensorflow.internal.c_api.TF_Tensor;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.ndarray.buffer.ByteDataBuffer;
-import org.tensorflow.ndarray.ByteNdArray;
 import org.tensorflow.ndarray.NdArray;
+import org.tensorflow.ndarray.Shape;
 import org.tensorflow.ndarray.StdArrays;
-import org.tensorflow.ndarray.impl.dense.ByteDenseNdArray;
+import org.tensorflow.ndarray.buffer.ByteDataBuffer;
 import org.tensorflow.types.family.TNumber;
 
-/** 8-bit unsigned integer tensor type. */
-public interface TUint8 extends ByteNdArray, TNumber {
+/**
+ * 8-bit unsigned integer tensor type.
+ */
+public interface TUint8 extends ByteTensor<TUint8>, TNumber {
 
   /** readable-name for the data type */
   static final String NAME = "UINT8";
 
   /** Type metadata */
-  DataType<TUint8> DTYPE = DataType.create(NAME, 4, 1, TUint8Impl::mapTensor);
+  DataType<TUint8> DTYPE = DataType.create(NAME, 4, 1, TUint8Impl::new);
 
   /**
    * Allocates a new tensor for storing a single byte value.
@@ -46,8 +46,8 @@ public interface TUint8 extends ByteNdArray, TNumber {
    * @param value byte to store in the new tensor
    * @return the new tensor
    */
-  static Tensor<TUint8> scalarOf(byte value) {
-    return Tensor.of(DTYPE, Shape.scalar(), data -> data.setByte(value));
+  static TUint8 scalarOf(byte value) {
+    return Tensor.of(DTYPE, Shape.scalar(), t -> t.setByte(value));
   }
 
   /**
@@ -56,11 +56,11 @@ static Tensor<TUint8> scalarOf(byte value) {
    * @param values bytes to store in the new tensor
    * @return the new tensor
    */
-  static Tensor<TUint8> vectorOf(byte... values) {
+  static TUint8 vectorOf(byte... values) {
     if (values == null) {
       throw new IllegalArgumentException();
     }
-    return Tensor.of(DTYPE, Shape.of(values.length), data -> StdArrays.copyTo(values, data));
+    return Tensor.of(DTYPE, Shape.of(values.length), t -> StdArrays.copyTo(values, t));
   }
 
   /**
@@ -71,7 +71,7 @@ static Tensor<TUint8> vectorOf(byte... values) {
    * @param src the source array giving the shape and data to the new tensor
    * @return the new tensor
    */
-  static Tensor<TUint8> tensorOf(NdArray<Byte> src) {
+  static TUint8 tensorOf(NdArray<Byte> src) {
     return Tensor.of(DTYPE, src.shape(), src::copyTo);
   }
 
@@ -81,7 +81,7 @@ static Tensor<TUint8> tensorOf(NdArray<Byte> src) {
    * @param shape shape of the tensor to allocate
    * @return the new tensor
    */
-  static Tensor<TUint8> tensorOf(Shape shape) {
+  static TUint8 tensorOf(Shape shape) {
     return Tensor.of(DTYPE, shape);
   }
 
@@ -92,7 +92,7 @@ static Tensor<TUint8> tensorOf(Shape shape) {
    * @param data buffer of bytes to initialize the tensor with
    * @return the new tensor
    */
-  static Tensor<TUint8> tensorOf(Shape shape, ByteDataBuffer data) {
+  static TUint8 tensorOf(Shape shape, ByteDataBuffer data) {
     return Tensor.of(DTYPE, shape, d -> d.write(data));
   }
 
@@ -100,23 +100,21 @@ static Tensor<TUint8> tensorOf(Shape shape, ByteDataBuffer data) {
    * Allocates a new tensor of the given shape and initialize its data.
    *
    * @param shape shape of the tensor to allocate
-   * @param dataInit tensor data initializer
+   * @param tensorInit tensor data initializer
    * @return the new tensor
    * @throws TensorFlowException if the tensor cannot be allocated or initialized
    */
-  static Tensor<TUint8> tensorOf(Shape shape, Consumer<TUint8> dataInit) {
-    return Tensor.of(DTYPE, shape, dataInit);
+  static TUint8 tensorOf(Shape shape, Consumer<TUint8> tensorInit) {
+    return Tensor.of(DTYPE, shape, tensorInit);
   }
 }
 
-/** Hidden implementation of a {@code TUint8} */
-class TUint8Impl extends ByteDenseNdArray implements TUint8 {
-
-  static TUint8 mapTensor(TF_Tensor nativeTensor, Shape shape) {
-    return new TUint8Impl(TensorBuffers.toBytes(nativeTensor), shape);
-  }
+/**
+ * Hidden implementation of a {@code TUint8}
+ */
+class TUint8Impl extends ByteTensorImpl<TUint8> implements TUint8 {
 
-  private TUint8Impl(ByteDataBuffer buffer, Shape shape) {
-    super(buffer, shape);
+  TUint8Impl(TF_Tensor nativeTensor, Shape shape) {
+    super(nativeTensor, DTYPE, shape, TensorBuffers.toBytes(nativeTensor));
   }
 }
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/family/TFloat.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/family/TFloat.java
new file mode 100644
index 00000000000..297efc1f560
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/family/TFloat.java
@@ -0,0 +1,5 @@
+package org.tensorflow.types.family;
+
+public interface TFloat extends TNumber {
+
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/TensorTest.java b/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/TensorTest.java
index 01ef11efedd..6c60480dfda 100644
--- a/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/TensorTest.java
+++ b/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/TensorTest.java
@@ -504,7 +504,7 @@ public void fromHandle() {
     // close() on both Tensors.
     final FloatNdArray matrix = StdArrays.ndCopyOf(new float[][]{{1, 2, 3}, {4, 5, 6}});
     try (Tensor<TFloat32> src = TFloat32.tensorOf(matrix)) {
-      Tensor<TFloat32> cpy = Tensor.fromHandle(src.nativeHandle()).expect(TFloat32.DTYPE);
+      Tensor<TFloat32> cpy = Tensors.fromHandle(src.nativeHandle()).expect(TFloat32.DTYPE);
       assertEquals(src.dataType(), cpy.dataType());
       assertEquals(src.shape().numDimensions(), cpy.shape().numDimensions());
       assertEquals(src.shape(), cpy.shape());

From 25dcd110af1ad32e5bf10071e21dda47f1d9a0fb Mon Sep 17 00:00:00 2001
From: Karl Lessard <karl.lessard@gmail.com>
Date: Thu, 30 Jul 2020 12:58:04 -0400
Subject: [PATCH 02/11] Extends  from

---
 .../src/main/java/org/tensorflow/types/family/TFloat.java    | 5 -----
 1 file changed, 5 deletions(-)
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/family/TFloat.java

diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/family/TFloat.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/family/TFloat.java
deleted file mode 100644
index 297efc1f560..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/family/TFloat.java
+++ /dev/null
@@ -1,5 +0,0 @@
-package org.tensorflow.types.family;
-
-public interface TFloat extends TNumber {
-
-}

From 69adad5eedf402d9f4b7db01cc0cf312e34a0308 Mon Sep 17 00:00:00 2001
From: Karl Lessard <karl.lessard@gmail.com>
Date: Sun, 2 Aug 2020 09:37:24 -0400
Subject: [PATCH 03/11] Convert TType generic parameters to Tensor

---
 .../src/bazel/op_generator/op_generator.cc    |   9 +-
 .../src/bazel/op_generator/op_specs.cc        |   4 +-
 .../src/bazel/op_generator/source_writer.cc   |  11 +-
 .../org/tensorflow/op/BitwiseOps.java         |  13 +-
 .../org/tensorflow/op/DtypesOps.java          |  10 +-
 .../org/tensorflow/op/ImageOps.java           |  65 +-
 .../annotations/org/tensorflow/op/IoOps.java  |  23 +-
 .../org/tensorflow/op/LinalgOps.java          |  98 +--
 .../org/tensorflow/op/MathOps.java            | 289 ++++----
 .../annotations/org/tensorflow/op/NnOps.java  | 347 ++++------
 .../annotations/org/tensorflow/op/Ops.java    | 618 ++++++++----------
 .../org/tensorflow/op/QuantizationOps.java    |  23 +-
 .../org/tensorflow/op/RandomOps.java          |  57 +-
 .../org/tensorflow/op/ShapeOps.java           |  48 +-
 .../org/tensorflow/op/SignalOps.java          |  32 +-
 .../org/tensorflow/op/SparseOps.java          | 224 +++----
 .../org/tensorflow/op/StringsOps.java         |  11 +-
 .../org/tensorflow/op/SummaryOps.java         |  13 +-
 .../org/tensorflow/op/TrainOps.java           | 127 ++--
 .../annotations/org/tensorflow/op/XlaOps.java |  43 +-
 .../org/tensorflow/op/bitwise/BitwiseAnd.java |   6 +-
 .../org/tensorflow/op/bitwise/BitwiseOr.java  |   6 +-
 .../org/tensorflow/op/bitwise/BitwiseXor.java |   6 +-
 .../org/tensorflow/op/bitwise/Invert.java     |   6 +-
 .../org/tensorflow/op/bitwise/LeftShift.java  |   6 +-
 .../org/tensorflow/op/bitwise/RightShift.java |   6 +-
 .../tensorflow/op/collective/AllReduce.java   |   6 +-
 .../op/collective/BroadcastRecv.java          |   6 +-
 .../op/collective/BroadcastSend.java          |   6 +-
 .../gen/java/org/tensorflow/op/core/All.java  |   4 +-
 .../gen/java/org/tensorflow/op/core/Any.java  |   4 +-
 .../java/org/tensorflow/op/core/Assign.java   |   6 +-
 .../org/tensorflow/op/core/AssignAdd.java     |   6 +-
 .../op/core/AssignAddVariableOp.java          |   4 +-
 .../org/tensorflow/op/core/AssignSub.java     |   6 +-
 .../op/core/AssignSubVariableOp.java          |   4 +-
 .../tensorflow/op/core/AssignVariableOp.java  |   4 +-
 .../tensorflow/op/core/BarrierInsertMany.java |   4 +-
 .../org/tensorflow/op/core/BatchToSpace.java  |   6 +-
 .../tensorflow/op/core/BatchToSpaceNd.java    |   6 +-
 .../java/org/tensorflow/op/core/Bitcast.java  |   6 +-
 .../op/core/BroadcastDynamicShape.java        |   6 +-
 .../op/core/BroadcastGradientArgs.java        |   6 +-
 .../org/tensorflow/op/core/BroadcastTo.java   |   6 +-
 .../org/tensorflow/op/core/Bucketize.java     |   4 +-
 .../org/tensorflow/op/core/ClipByValue.java   |   6 +-
 .../tensorflow/op/core/CollectiveGather.java  |   6 +-
 .../java/org/tensorflow/op/core/Concat.java   |   6 +-
 .../gen/java/org/tensorflow/op/core/Copy.java |   6 +-
 .../java/org/tensorflow/op/core/CopyHost.java |   6 +-
 .../org/tensorflow/op/core/CountUpTo.java     |   6 +-
 .../java/org/tensorflow/op/core/DeepCopy.java |   6 +-
 .../op/core/DestroyTemporaryVariable.java     |   6 +-
 .../tensorflow/op/core/DummyMemoryCache.java  |   8 +-
 .../tensorflow/op/core/DynamicPartition.java  |   6 +-
 .../org/tensorflow/op/core/DynamicStitch.java |   6 +-
 .../org/tensorflow/op/core/EditDistance.java  |   4 +-
 .../java/org/tensorflow/op/core/Empty.java    |   6 +-
 .../tensorflow/op/core/EmptyTensorList.java   |  10 +-
 .../org/tensorflow/op/core/EnsureShape.java   |   6 +-
 .../java/org/tensorflow/op/core/Enter.java    |   6 +-
 .../gen/java/org/tensorflow/op/core/Exit.java |   6 +-
 .../org/tensorflow/op/core/ExpandDims.java    |   6 +-
 .../op/core/ExtractVolumePatches.java         |   6 +-
 .../gen/java/org/tensorflow/op/core/Fill.java |   6 +-
 .../org/tensorflow/op/core/Fingerprint.java   |   4 +-
 .../java/org/tensorflow/op/core/Gather.java   |   6 +-
 .../java/org/tensorflow/op/core/GatherNd.java |   6 +-
 .../tensorflow/op/core/GetSessionHandle.java  |  10 +-
 .../tensorflow/op/core/GetSessionTensor.java  |   6 +-
 .../tensorflow/op/core/GuaranteeConst.java    |   6 +-
 .../org/tensorflow/op/core/HashTable.java     |  10 +-
 .../op/core/HistogramFixedWidth.java          |   8 +-
 .../java/org/tensorflow/op/core/Identity.java |   6 +-
 .../org/tensorflow/op/core/IdentityN.java     |   6 +-
 .../tensorflow/op/core/ImmutableConst.java    |   6 +-
 .../tensorflow/op/core/InitializeTable.java   |   4 +-
 .../org/tensorflow/op/core/InplaceAdd.java    |   6 +-
 .../org/tensorflow/op/core/InplaceSub.java    |   6 +-
 .../org/tensorflow/op/core/InplaceUpdate.java |   6 +-
 .../op/core/IsVariableInitialized.java        |   4 +-
 .../java/org/tensorflow/op/core/LinSpace.java |  10 +-
 .../tensorflow/op/core/LookupTableExport.java |   6 +-
 .../tensorflow/op/core/LookupTableFind.java   |   6 +-
 .../tensorflow/op/core/LookupTableImport.java |   4 +-
 .../tensorflow/op/core/LookupTableInsert.java |   4 +-
 .../tensorflow/op/core/LookupTableRemove.java |   4 +-
 .../org/tensorflow/op/core/LowerBound.java    |   8 +-
 .../java/org/tensorflow/op/core/MapPeek.java  |   6 +-
 .../org/tensorflow/op/core/MapUnstage.java    |   6 +-
 .../gen/java/org/tensorflow/op/core/Max.java  |   6 +-
 .../java/org/tensorflow/op/core/Merge.java    |   6 +-
 .../gen/java/org/tensorflow/op/core/Min.java  |   6 +-
 .../org/tensorflow/op/core/MirrorPad.java     |   6 +-
 .../org/tensorflow/op/core/MirrorPadGrad.java |   6 +-
 .../tensorflow/op/core/MlirPassthroughOp.java |   6 +-
 .../op/core/MutableDenseHashTable.java        |  10 +-
 .../tensorflow/op/core/MutableHashTable.java  |  10 +-
 .../op/core/MutableHashTableOfTensors.java    |  10 +-
 .../java/org/tensorflow/op/core/Mutex.java    |   8 +-
 .../org/tensorflow/op/core/MutexLock.java     |   8 +-
 .../org/tensorflow/op/core/NcclAllReduce.java |   6 +-
 .../org/tensorflow/op/core/NcclBroadcast.java |   6 +-
 .../org/tensorflow/op/core/NcclReduce.java    |   6 +-
 .../org/tensorflow/op/core/NextIteration.java |   6 +-
 .../java/org/tensorflow/op/core/OneHot.java   |   6 +-
 .../java/org/tensorflow/op/core/OnesLike.java |   6 +-
 .../tensorflow/op/core/OrderedMapPeek.java    |   6 +-
 .../tensorflow/op/core/OrderedMapUnstage.java |   6 +-
 .../gen/java/org/tensorflow/op/core/Pad.java  |   6 +-
 .../tensorflow/op/core/ParallelConcat.java    |   6 +-
 .../op/core/ParallelDynamicStitch.java        |   6 +-
 .../org/tensorflow/op/core/Placeholder.java   |   6 +-
 .../op/core/PlaceholderWithDefault.java       |   6 +-
 .../gen/java/org/tensorflow/op/core/Prod.java |   6 +-
 .../tensorflow/op/core/QuantizedReshape.java  |   6 +-
 .../java/org/tensorflow/op/core/Range.java    |   6 +-
 .../gen/java/org/tensorflow/op/core/Rank.java |   4 +-
 .../tensorflow/op/core/ReadVariableOp.java    |   6 +-
 .../gen/java/org/tensorflow/op/core/Recv.java |   6 +-
 .../org/tensorflow/op/core/ReduceAll.java     |   4 +-
 .../org/tensorflow/op/core/ReduceAny.java     |   4 +-
 .../org/tensorflow/op/core/ReduceMax.java     |   6 +-
 .../org/tensorflow/op/core/ReduceMin.java     |   6 +-
 .../org/tensorflow/op/core/ReduceProd.java    |   6 +-
 .../org/tensorflow/op/core/ReduceSum.java     |   6 +-
 .../java/org/tensorflow/op/core/RefEnter.java |   6 +-
 .../java/org/tensorflow/op/core/RefExit.java  |   6 +-
 .../org/tensorflow/op/core/RefIdentity.java   |   6 +-
 .../java/org/tensorflow/op/core/RefMerge.java |   6 +-
 .../tensorflow/op/core/RefNextIteration.java  |   6 +-
 .../org/tensorflow/op/core/RefSelect.java     |   6 +-
 .../org/tensorflow/op/core/RefSwitch.java     |   6 +-
 .../op/core/RemoteFusedGraphExecute.java      |   6 +-
 .../java/org/tensorflow/op/core/Reshape.java  |   6 +-
 .../tensorflow/op/core/ResourceCountUpTo.java |   6 +-
 .../tensorflow/op/core/ResourceGather.java    |   6 +-
 .../tensorflow/op/core/ResourceGatherNd.java  |   6 +-
 .../op/core/ResourceScatterAdd.java           |   4 +-
 .../op/core/ResourceScatterDiv.java           |   4 +-
 .../op/core/ResourceScatterMax.java           |   4 +-
 .../op/core/ResourceScatterMin.java           |   4 +-
 .../op/core/ResourceScatterMul.java           |   4 +-
 .../op/core/ResourceScatterNdAdd.java         |   4 +-
 .../op/core/ResourceScatterNdSub.java         |   4 +-
 .../op/core/ResourceScatterNdUpdate.java      |   4 +-
 .../op/core/ResourceScatterSub.java           |   4 +-
 .../op/core/ResourceScatterUpdate.java        |   4 +-
 .../op/core/ResourceStridedSliceAssign.java   |   4 +-
 .../java/org/tensorflow/op/core/Reverse.java  |   6 +-
 .../tensorflow/op/core/ReverseSequence.java   |   6 +-
 .../gen/java/org/tensorflow/op/core/Roll.java |   6 +-
 .../org/tensorflow/op/core/ScatterAdd.java    |   6 +-
 .../org/tensorflow/op/core/ScatterDiv.java    |   6 +-
 .../org/tensorflow/op/core/ScatterMax.java    |   6 +-
 .../org/tensorflow/op/core/ScatterMin.java    |   6 +-
 .../org/tensorflow/op/core/ScatterMul.java    |   6 +-
 .../org/tensorflow/op/core/ScatterNd.java     |   6 +-
 .../org/tensorflow/op/core/ScatterNdAdd.java  |   6 +-
 .../op/core/ScatterNdNonAliasingAdd.java      |   6 +-
 .../org/tensorflow/op/core/ScatterNdSub.java  |   6 +-
 .../tensorflow/op/core/ScatterNdUpdate.java   |   6 +-
 .../org/tensorflow/op/core/ScatterSub.java    |   6 +-
 .../org/tensorflow/op/core/ScatterUpdate.java |   6 +-
 .../java/org/tensorflow/op/core/Select.java   |   6 +-
 .../gen/java/org/tensorflow/op/core/Send.java |   4 +-
 .../org/tensorflow/op/core/SetDiff1d.java     |   8 +-
 .../java/org/tensorflow/op/core/SetSize.java  |   4 +-
 .../java/org/tensorflow/op/core/Shape.java    |   8 +-
 .../java/org/tensorflow/op/core/ShapeN.java   |   8 +-
 .../gen/java/org/tensorflow/op/core/Size.java |   8 +-
 .../java/org/tensorflow/op/core/Slice.java    |   6 +-
 .../java/org/tensorflow/op/core/Snapshot.java |   6 +-
 .../tensorflow/op/core/SpaceToBatchNd.java    |   6 +-
 .../java/org/tensorflow/op/core/Split.java    |   6 +-
 .../java/org/tensorflow/op/core/SplitV.java   |   6 +-
 .../java/org/tensorflow/op/core/Squeeze.java  |   6 +-
 .../java/org/tensorflow/op/core/Stack.java    |   6 +-
 .../org/tensorflow/op/core/StagePeek.java     |   6 +-
 .../org/tensorflow/op/core/StopGradient.java  |   6 +-
 .../org/tensorflow/op/core/StridedSlice.java  |   6 +-
 .../op/core/StridedSliceAssign.java           |   6 +-
 .../tensorflow/op/core/StridedSliceGrad.java  |   6 +-
 .../gen/java/org/tensorflow/op/core/Sum.java  |   6 +-
 .../org/tensorflow/op/core/SwitchCond.java    |   6 +-
 .../tensorflow/op/core/TemporaryVariable.java |   6 +-
 .../org/tensorflow/op/core/TensorArray.java   |   4 +-
 .../tensorflow/op/core/TensorArrayConcat.java |   6 +-
 .../tensorflow/op/core/TensorArrayGather.java |   6 +-
 .../tensorflow/op/core/TensorArrayPack.java   |   6 +-
 .../tensorflow/op/core/TensorArrayRead.java   |   6 +-
 .../op/core/TensorArrayScatter.java           |   4 +-
 .../tensorflow/op/core/TensorArraySplit.java  |   4 +-
 .../tensorflow/op/core/TensorArrayUnpack.java |   4 +-
 .../tensorflow/op/core/TensorArrayWrite.java  |   4 +-
 .../TensorForestTreeResourceHandleOp.java     |   8 +-
 .../tensorflow/op/core/TensorListConcat.java  |   6 +-
 .../op/core/TensorListConcatLists.java        |  10 +-
 .../op/core/TensorListElementShape.java       |   6 +-
 .../op/core/TensorListFromTensor.java         |  10 +-
 .../tensorflow/op/core/TensorListGather.java  |   6 +-
 .../tensorflow/op/core/TensorListGetItem.java |   6 +-
 .../tensorflow/op/core/TensorListPopBack.java |   6 +-
 .../op/core/TensorListPushBack.java           |  10 +-
 .../op/core/TensorListPushBackBatch.java      |  10 +-
 .../tensorflow/op/core/TensorListReserve.java |  10 +-
 .../tensorflow/op/core/TensorListResize.java  |   8 +-
 .../tensorflow/op/core/TensorListScatter.java |  10 +-
 .../TensorListScatterIntoExistingList.java    |  10 +-
 .../tensorflow/op/core/TensorListSetItem.java |  10 +-
 .../tensorflow/op/core/TensorListSplit.java   |  10 +-
 .../tensorflow/op/core/TensorListStack.java   |   6 +-
 .../op/core/TensorScatterNdAdd.java           |   6 +-
 .../op/core/TensorScatterNdSub.java           |   6 +-
 .../op/core/TensorScatterNdUpdate.java        |   6 +-
 .../op/core/TensorStridedSliceUpdate.java     |   6 +-
 .../gen/java/org/tensorflow/op/core/Tile.java |   6 +-
 .../java/org/tensorflow/op/core/Unbatch.java  |   6 +-
 .../org/tensorflow/op/core/UnbatchGrad.java   |   6 +-
 .../java/org/tensorflow/op/core/Unique.java   |   8 +-
 .../tensorflow/op/core/UniqueWithCounts.java  |   8 +-
 .../org/tensorflow/op/core/UnravelIndex.java  |   6 +-
 .../java/org/tensorflow/op/core/Unstack.java  |   6 +-
 .../java/org/tensorflow/op/core/Unstage.java  |   6 +-
 .../org/tensorflow/op/core/UpperBound.java    |   8 +-
 .../org/tensorflow/op/core/VarHandleOp.java   |  10 +-
 .../java/org/tensorflow/op/core/Variable.java |   6 +-
 .../org/tensorflow/op/core/VariableShape.java |   6 +-
 .../java/org/tensorflow/op/core/Where.java    |   4 +-
 .../org/tensorflow/op/core/ZerosLike.java     |   6 +-
 .../tensorflow/op/data/AssertNextDataset.java |   8 +-
 .../tensorflow/op/data/AutoShardDataset.java  |   8 +-
 .../org/tensorflow/op/data/BatchDataset.java  |   8 +-
 .../op/data/BytesProducedStatsDataset.java    |   8 +-
 .../org/tensorflow/op/data/CSVDataset.java    |   8 +-
 .../org/tensorflow/op/data/CacheDataset.java  |   8 +-
 .../tensorflow/op/data/CacheDatasetV2.java    |   8 +-
 .../op/data/ChooseFastestDataset.java         |   8 +-
 .../op/data/ConcatenateDataset.java           |   8 +-
 .../tensorflow/op/data/DatasetFromGraph.java  |   8 +-
 .../op/data/DatasetToSingleElement.java       |   6 +-
 .../op/data/DenseToSparseBatchDataset.java    |   8 +-
 .../op/data/DirectedInterleaveDataset.java    |   8 +-
 .../op/data/FilterByLastComponentDataset.java |   8 +-
 .../op/data/FixedLengthRecordDataset.java     |   8 +-
 .../op/data/IgnoreErrorsDataset.java          |   8 +-
 .../java/org/tensorflow/op/data/Iterator.java |   8 +-
 .../op/data/IteratorFromStringHandle.java     |   8 +-
 .../tensorflow/op/data/IteratorGetNext.java   |   6 +-
 .../op/data/IteratorGetNextAsOptional.java    |   8 +-
 .../op/data/IteratorGetNextSync.java          |   6 +-
 .../org/tensorflow/op/data/LMDBDataset.java   |   8 +-
 .../op/data/LatencyStatsDataset.java          |   8 +-
 .../org/tensorflow/op/data/LeakyReluGrad.java |   6 +-
 .../op/data/MatchingFilesDataset.java         |   8 +-
 .../op/data/MaxIntraOpParallelismDataset.java |   8 +-
 .../org/tensorflow/op/data/ModelDataset.java  |   8 +-
 .../op/data/MultiDeviceIterator.java          |   8 +-
 .../MultiDeviceIteratorFromStringHandle.java  |   8 +-
 .../MultiDeviceIteratorGetNextFromShard.java  |   6 +-
 .../op/data/NonSerializableDataset.java       |   8 +-
 .../tensorflow/op/data/OptimizeDataset.java   |   8 +-
 .../tensorflow/op/data/OptionalFromValue.java |   8 +-
 .../tensorflow/op/data/OptionalGetValue.java  |   6 +-
 .../org/tensorflow/op/data/OptionalNone.java  |   8 +-
 .../op/data/PaddedBatchDataset.java           |   8 +-
 .../tensorflow/op/data/PrefetchDataset.java   |   8 +-
 .../op/data/PrivateThreadPoolDataset.java     |   8 +-
 .../org/tensorflow/op/data/RandomDataset.java |   8 +-
 .../org/tensorflow/op/data/RangeDataset.java  |   8 +-
 .../tensorflow/op/data/RebatchDataset.java    |   8 +-
 .../org/tensorflow/op/data/RepeatDataset.java |   8 +-
 .../tensorflow/op/data/SamplingDataset.java   |   8 +-
 .../tensorflow/op/data/SerializeIterator.java |   8 +-
 .../op/data/SetStatsAggregatorDataset.java    |   8 +-
 .../org/tensorflow/op/data/ShardDataset.java  |   8 +-
 .../op/data/ShuffleAndRepeatDataset.java      |   8 +-
 .../tensorflow/op/data/ShuffleDataset.java    |   8 +-
 .../org/tensorflow/op/data/SkipDataset.java   |   8 +-
 .../org/tensorflow/op/data/SleepDataset.java  |   8 +-
 .../op/data/SlidingWindowDataset.java         |   8 +-
 .../tensorflow/op/data/SnapshotDataset.java   |   8 +-
 .../op/data/SparseTensorSliceDataset.java     |  10 +-
 .../org/tensorflow/op/data/SqlDataset.java    |   8 +-
 .../op/data/StatsAggregatorHandle.java        |   8 +-
 .../org/tensorflow/op/data/TakeDataset.java   |   8 +-
 .../org/tensorflow/op/data/TensorDataset.java |   8 +-
 .../op/data/TensorSliceDataset.java           |   8 +-
 .../tensorflow/op/data/TextLineDataset.java   |   8 +-
 .../tensorflow/op/data/TfRecordDataset.java   |   8 +-
 .../tensorflow/op/data/ThreadPoolDataset.java |   8 +-
 .../tensorflow/op/data/ThreadPoolHandle.java  |   8 +-
 .../tensorflow/op/data/UnbatchDataset.java    |   8 +-
 .../org/tensorflow/op/data/UniqueDataset.java |   8 +-
 .../op/data/UnwrapDatasetVariant.java         |   8 +-
 .../org/tensorflow/op/data/WindowDataset.java |   8 +-
 .../op/data/WrapDatasetVariant.java           |   8 +-
 .../org/tensorflow/op/data/ZipDataset.java    |   8 +-
 .../AssertCardinalityDataset.java             |   8 +-
 .../data/experimental/AssertNextDataset.java  |   8 +-
 .../data/experimental/AutoShardDataset.java   |   8 +-
 .../BytesProducedStatsDataset.java            |   8 +-
 .../op/data/experimental/CSVDataset.java      |   8 +-
 .../experimental/ChooseFastestDataset.java    |   8 +-
 .../DenseToSparseBatchDataset.java            |   8 +-
 .../DirectedInterleaveDataset.java            |   8 +-
 .../experimental/IgnoreErrorsDataset.java     |   8 +-
 .../experimental/LatencyStatsDataset.java     |   8 +-
 .../op/data/experimental/LmdbDataset.java     |   8 +-
 .../experimental/MatchingFilesDataset.java    |   8 +-
 .../MaxIntraOpParallelismDataset.java         |   8 +-
 .../experimental/NonSerializableDataset.java  |   8 +-
 .../experimental/ParseExampleDataset.java     |   8 +-
 .../PrivateThreadPoolDataset.java             |   8 +-
 .../op/data/experimental/RandomDataset.java   |   8 +-
 .../op/data/experimental/RebatchDataset.java  |   8 +-
 .../SetStatsAggregatorDataset.java            |   8 +-
 .../op/data/experimental/SleepDataset.java    |   8 +-
 .../experimental/SlidingWindowDataset.java    |   8 +-
 .../op/data/experimental/SqlDataset.java      |   8 +-
 .../experimental/StatsAggregatorHandle.java   |   8 +-
 .../data/experimental/ThreadPoolDataset.java  |   8 +-
 .../data/experimental/ThreadPoolHandle.java   |   8 +-
 .../op/data/experimental/UnbatchDataset.java  |   8 +-
 .../op/data/experimental/UniqueDataset.java   |   8 +-
 .../op/debugging/CheckNumerics.java           |   6 +-
 .../op/debugging/DebugGradientIdentity.java   |   6 +-
 .../debugging/DebugGradientRefIdentity.java   |   6 +-
 .../op/debugging/DebugIdentity.java           |   6 +-
 .../op/debugging/DebugNanCount.java           |   4 +-
 .../op/debugging/DebugNumericsSummary.java    |   8 +-
 .../org/tensorflow/op/dtypes/AsString.java    |   4 +-
 .../java/org/tensorflow/op/dtypes/Cast.java   |   6 +-
 .../org/tensorflow/op/dtypes/Complex.java     |   6 +-
 .../java/org/tensorflow/op/dtypes/ToBool.java |   4 +-
 .../BoostedTreesEnsembleResourceHandleOp.java |   8 +-
 ...edTreesQuantileStreamResourceHandleOp.java |   8 +-
 .../tensorflow/op/image/AdjustContrast.java   |   6 +-
 .../org/tensorflow/op/image/AdjustHue.java    |   6 +-
 .../tensorflow/op/image/AdjustSaturation.java |   6 +-
 .../tensorflow/op/image/CropAndResize.java    |   4 +-
 .../op/image/CropAndResizeGradBoxes.java      |   4 +-
 .../op/image/CropAndResizeGradImage.java      |   6 +-
 .../org/tensorflow/op/image/DecodePng.java    |   6 +-
 .../op/image/DrawBoundingBoxes.java           |   6 +-
 .../org/tensorflow/op/image/EncodePng.java    |   4 +-
 .../op/image/ExtractImagePatches.java         |   6 +-
 .../tensorflow/op/image/ExtractJpegShape.java |   6 +-
 .../org/tensorflow/op/image/HsvToRgb.java     |   6 +-
 .../op/image/ImageProjectiveTransformV2.java  |   6 +-
 .../op/image/NonMaxSuppression.java           |   6 +-
 .../op/image/QuantizedResizeBilinear.java     |   6 +-
 .../org/tensorflow/op/image/RandomCrop.java   |   6 +-
 .../org/tensorflow/op/image/ResizeArea.java   |   4 +-
 .../tensorflow/op/image/ResizeBicubic.java    |   4 +-
 .../op/image/ResizeBicubicGrad.java           |   6 +-
 .../tensorflow/op/image/ResizeBilinear.java   |   4 +-
 .../op/image/ResizeBilinearGrad.java          |   6 +-
 .../op/image/ResizeNearestNeighbor.java       |   6 +-
 .../op/image/ResizeNearestNeighborGrad.java   |   6 +-
 .../org/tensorflow/op/image/RgbToHsv.java     |   6 +-
 .../op/image/SampleDistortedBoundingBox.java  |   6 +-
 .../op/image/ScaleAndTranslate.java           |   4 +-
 .../op/image/ScaleAndTranslateGrad.java       |   6 +-
 .../java/org/tensorflow/op/io/DecodeCsv.java  |   6 +-
 .../org/tensorflow/op/io/DecodePaddedRaw.java |   6 +-
 .../java/org/tensorflow/op/io/DecodeRaw.java  |   6 +-
 .../op/io/DeserializeManySparse.java          |   6 +-
 .../java/org/tensorflow/op/io/FifoQueue.java  |   8 +-
 .../op/io/FixedLengthRecordReader.java        |   8 +-
 .../org/tensorflow/op/io/IdentityReader.java  |   8 +-
 .../tensorflow/op/io/PaddingFifoQueue.java    |   8 +-
 .../org/tensorflow/op/io/ParseTensor.java     |   6 +-
 .../org/tensorflow/op/io/PriorityQueue.java   |   8 +-
 .../org/tensorflow/op/io/QueueDequeue.java    |   6 +-
 .../tensorflow/op/io/QueueDequeueMany.java    |   6 +-
 .../tensorflow/op/io/QueueDequeueUpTo.java    |   6 +-
 .../tensorflow/op/io/RandomShuffleQueue.java  |   8 +-
 .../tensorflow/op/io/SerializeManySparse.java |   8 +-
 .../org/tensorflow/op/io/SerializeSparse.java |   8 +-
 .../org/tensorflow/op/io/SerializeTensor.java |   4 +-
 .../org/tensorflow/op/io/TextLineReader.java  |   8 +-
 .../org/tensorflow/op/io/TfRecordReader.java  |   8 +-
 .../org/tensorflow/op/io/WholeFileReader.java |   8 +-
 .../org/tensorflow/op/linalg/BandPart.java    |   6 +-
 .../tensorflow/op/linalg/BatchCholesky.java   |   6 +-
 .../op/linalg/BatchCholeskyGrad.java          |   6 +-
 .../op/linalg/BatchMatrixBandPart.java        |   6 +-
 .../op/linalg/BatchMatrixDeterminant.java     |   6 +-
 .../tensorflow/op/linalg/BatchMatrixDiag.java |   6 +-
 .../op/linalg/BatchMatrixDiagPart.java        |   6 +-
 .../op/linalg/BatchMatrixInverse.java         |   6 +-
 .../op/linalg/BatchMatrixSetDiag.java         |   6 +-
 .../op/linalg/BatchMatrixSolve.java           |   6 +-
 .../op/linalg/BatchMatrixSolveLs.java         |   6 +-
 .../op/linalg/BatchMatrixTriangularSolve.java |   6 +-
 .../op/linalg/BatchSelfAdjointEig.java        |   6 +-
 .../org/tensorflow/op/linalg/BatchSvd.java    |   6 +-
 .../org/tensorflow/op/linalg/Cholesky.java    |   6 +-
 .../tensorflow/op/linalg/CholeskyGrad.java    |   6 +-
 .../op/linalg/ConjugateTranspose.java         |   6 +-
 .../java/org/tensorflow/op/linalg/Cross.java  |   6 +-
 .../java/org/tensorflow/op/linalg/Det.java    |   6 +-
 .../java/org/tensorflow/op/linalg/Eig.java    |   6 +-
 .../java/org/tensorflow/op/linalg/Einsum.java |   6 +-
 .../tensorflow/op/linalg/EuclideanNorm.java   |   6 +-
 .../java/org/tensorflow/op/linalg/Inv.java    |   6 +-
 .../op/linalg/LogMatrixDeterminant.java       |   6 +-
 .../gen/java/org/tensorflow/op/linalg/Lu.java |   8 +-
 .../java/org/tensorflow/op/linalg/MatMul.java |   6 +-
 .../org/tensorflow/op/linalg/MatrixDiag.java  |   6 +-
 .../tensorflow/op/linalg/MatrixDiagPart.java  |   6 +-
 .../op/linalg/MatrixDiagPartV3.java           |   6 +-
 .../tensorflow/op/linalg/MatrixDiagV3.java    |   6 +-
 .../tensorflow/op/linalg/MatrixLogarithm.java |   6 +-
 .../tensorflow/op/linalg/MatrixSetDiag.java   |   6 +-
 .../tensorflow/op/linalg/MatrixSolveLs.java   |   6 +-
 .../gen/java/org/tensorflow/op/linalg/Qr.java |   6 +-
 .../tensorflow/op/linalg/QuantizedMatMul.java |   6 +-
 .../op/linalg/QuantizedMatMulWithBias.java    |   6 +-
 .../QuantizedMatMulWithBiasAndRelu.java       |   6 +-
 ...zedMatMulWithBiasAndReluAndRequantize.java |   6 +-
 .../tensorflow/op/linalg/SelfAdjointEig.java  |   6 +-
 .../java/org/tensorflow/op/linalg/Solve.java  |   6 +-
 .../java/org/tensorflow/op/linalg/Sqrtm.java  |   6 +-
 .../java/org/tensorflow/op/linalg/Svd.java    |   6 +-
 .../org/tensorflow/op/linalg/TensorDiag.java  |   6 +-
 .../tensorflow/op/linalg/TensorDiagPart.java  |   6 +-
 .../org/tensorflow/op/linalg/Transpose.java   |   6 +-
 .../tensorflow/op/linalg/TriangularSolve.java |   6 +-
 .../op/linalg/TridiagonalMatMul.java          |   6 +-
 .../op/linalg/TridiagonalSolve.java           |   6 +-
 .../sparse/CSRSparseMatrixComponents.java     |   6 +-
 .../linalg/sparse/CSRSparseMatrixToDense.java |   6 +-
 .../sparse/CSRSparseMatrixToSparseTensor.java |   6 +-
 .../linalg/sparse/DenseToCSRSparseMatrix.java |  10 +-
 .../op/linalg/sparse/SparseMatrixAdd.java     |  10 +-
 .../op/linalg/sparse/SparseMatrixMatMul.java  |   6 +-
 .../op/linalg/sparse/SparseMatrixMul.java     |  10 +-
 .../op/linalg/sparse/SparseMatrixSoftmax.java |  10 +-
 .../sparse/SparseMatrixSoftmaxGrad.java       |  10 +-
 .../sparse/SparseMatrixSparseCholesky.java    |  10 +-
 .../sparse/SparseMatrixSparseMatMul.java      |  10 +-
 .../linalg/sparse/SparseMatrixTranspose.java  |  10 +-
 .../op/linalg/sparse/SparseMatrixZeros.java   |  10 +-
 .../sparse/SparseTensorToCSRSparseMatrix.java |  10 +-
 .../gen/java/org/tensorflow/op/math/Abs.java  |   6 +-
 .../org/tensorflow/op/math/AccumulateN.java   |   6 +-
 .../gen/java/org/tensorflow/op/math/Acos.java |   6 +-
 .../java/org/tensorflow/op/math/Acosh.java    |   6 +-
 .../gen/java/org/tensorflow/op/math/Add.java  |   6 +-
 .../gen/java/org/tensorflow/op/math/AddN.java |   6 +-
 .../java/org/tensorflow/op/math/Angle.java    |   8 +-
 .../tensorflow/op/math/ApproximateEqual.java  |   4 +-
 .../java/org/tensorflow/op/math/ArgMax.java   |   8 +-
 .../java/org/tensorflow/op/math/ArgMin.java   |   8 +-
 .../gen/java/org/tensorflow/op/math/Asin.java |   6 +-
 .../java/org/tensorflow/op/math/Asinh.java    |   6 +-
 .../gen/java/org/tensorflow/op/math/Atan.java |   6 +-
 .../java/org/tensorflow/op/math/Atan2.java    |   6 +-
 .../java/org/tensorflow/op/math/Atanh.java    |   6 +-
 .../org/tensorflow/op/math/BesselI0e.java     |  17 +-
 .../org/tensorflow/op/math/BesselI1e.java     |  17 +-
 .../java/org/tensorflow/op/math/Betainc.java  |   6 +-
 .../java/org/tensorflow/op/math/Bincount.java |   6 +-
 .../gen/java/org/tensorflow/op/math/Ceil.java |   6 +-
 .../tensorflow/op/math/CompareAndBitpack.java |   4 +-
 .../org/tensorflow/op/math/ComplexAbs.java    |   8 +-
 .../gen/java/org/tensorflow/op/math/Conj.java |   6 +-
 .../gen/java/org/tensorflow/op/math/Cos.java  |   6 +-
 .../gen/java/org/tensorflow/op/math/Cosh.java |   6 +-
 .../java/org/tensorflow/op/math/Cumprod.java  |   6 +-
 .../java/org/tensorflow/op/math/Cumsum.java   |   6 +-
 .../op/math/CumulativeLogsumexp.java          |   6 +-
 .../java/org/tensorflow/op/math/Digamma.java  |   6 +-
 .../gen/java/org/tensorflow/op/math/Div.java  |   6 +-
 .../java/org/tensorflow/op/math/DivNoNan.java |   6 +-
 .../java/org/tensorflow/op/math/Equal.java    |   4 +-
 .../gen/java/org/tensorflow/op/math/Erf.java  |   6 +-
 .../gen/java/org/tensorflow/op/math/Erfc.java |   6 +-
 .../gen/java/org/tensorflow/op/math/Exp.java  |   6 +-
 .../java/org/tensorflow/op/math/Expm1.java    |   6 +-
 .../java/org/tensorflow/op/math/Floor.java    |   6 +-
 .../java/org/tensorflow/op/math/FloorDiv.java |   6 +-
 .../java/org/tensorflow/op/math/FloorMod.java |   6 +-
 .../java/org/tensorflow/op/math/Greater.java  |   4 +-
 .../org/tensorflow/op/math/GreaterEqual.java  |   4 +-
 .../java/org/tensorflow/op/math/Igamma.java   |   6 +-
 .../org/tensorflow/op/math/IgammaGradA.java   |   6 +-
 .../java/org/tensorflow/op/math/Igammac.java  |   6 +-
 .../gen/java/org/tensorflow/op/math/Imag.java |   8 +-
 .../tensorflow/op/math/InvertPermutation.java |   6 +-
 .../java/org/tensorflow/op/math/IsFinite.java |   4 +-
 .../java/org/tensorflow/op/math/IsInf.java    |   4 +-
 .../java/org/tensorflow/op/math/IsNan.java    |   4 +-
 .../gen/java/org/tensorflow/op/math/Less.java |   4 +-
 .../org/tensorflow/op/math/LessEqual.java     |   4 +-
 .../java/org/tensorflow/op/math/Lgamma.java   |   6 +-
 .../gen/java/org/tensorflow/op/math/Log.java  |   6 +-
 .../java/org/tensorflow/op/math/Log1p.java    |   6 +-
 .../java/org/tensorflow/op/math/Maximum.java  |   6 +-
 .../gen/java/org/tensorflow/op/math/Mean.java |   6 +-
 .../java/org/tensorflow/op/math/Minimum.java  |   6 +-
 .../gen/java/org/tensorflow/op/math/Mod.java  |   6 +-
 .../gen/java/org/tensorflow/op/math/Mul.java  |   6 +-
 .../java/org/tensorflow/op/math/MulNoNan.java |   6 +-
 .../java/org/tensorflow/op/math/Ndtri.java    |   6 +-
 .../gen/java/org/tensorflow/op/math/Neg.java  |   6 +-
 .../org/tensorflow/op/math/NextAfter.java     |   6 +-
 .../java/org/tensorflow/op/math/NotEqual.java |   4 +-
 .../org/tensorflow/op/math/Polygamma.java     |   6 +-
 .../tensorflow/op/math/PopulationCount.java   |   4 +-
 .../gen/java/org/tensorflow/op/math/Pow.java  |   6 +-
 .../org/tensorflow/op/math/QuantizedAdd.java  |   6 +-
 .../org/tensorflow/op/math/QuantizedMul.java  |   6 +-
 .../gen/java/org/tensorflow/op/math/Real.java |   8 +-
 .../java/org/tensorflow/op/math/RealDiv.java  |   6 +-
 .../org/tensorflow/op/math/Reciprocal.java    |   6 +-
 .../tensorflow/op/math/ReciprocalGrad.java    |   6 +-
 .../math/RequantizationRangePerChannel.java   |   4 +-
 .../op/math/RequantizePerChannel.java         |   6 +-
 .../gen/java/org/tensorflow/op/math/Rint.java |   6 +-
 .../java/org/tensorflow/op/math/Round.java    |   6 +-
 .../java/org/tensorflow/op/math/Rsqrt.java    |   6 +-
 .../org/tensorflow/op/math/RsqrtGrad.java     |   6 +-
 .../org/tensorflow/op/math/SegmentMax.java    |   6 +-
 .../org/tensorflow/op/math/SegmentMean.java   |   6 +-
 .../org/tensorflow/op/math/SegmentMin.java    |   6 +-
 .../org/tensorflow/op/math/SegmentProd.java   |   6 +-
 .../org/tensorflow/op/math/SegmentSum.java    |   6 +-
 .../java/org/tensorflow/op/math/Sigmoid.java  |   6 +-
 .../org/tensorflow/op/math/SigmoidGrad.java   |   6 +-
 .../gen/java/org/tensorflow/op/math/Sign.java |   6 +-
 .../gen/java/org/tensorflow/op/math/Sin.java  |   6 +-
 .../gen/java/org/tensorflow/op/math/Sinh.java |   6 +-
 .../org/tensorflow/op/math/SobolSample.java   |   6 +-
 .../java/org/tensorflow/op/math/Softplus.java |   6 +-
 .../org/tensorflow/op/math/SoftplusGrad.java  |   6 +-
 .../gen/java/org/tensorflow/op/math/Sqrt.java |   6 +-
 .../java/org/tensorflow/op/math/SqrtGrad.java |   6 +-
 .../java/org/tensorflow/op/math/Square.java   |   6 +-
 .../tensorflow/op/math/SquaredDifference.java |   6 +-
 .../gen/java/org/tensorflow/op/math/Sub.java  |   6 +-
 .../gen/java/org/tensorflow/op/math/Tan.java  |   6 +-
 .../gen/java/org/tensorflow/op/math/Tanh.java |   6 +-
 .../java/org/tensorflow/op/math/TanhGrad.java |   6 +-
 .../org/tensorflow/op/math/TruncateDiv.java   |   6 +-
 .../org/tensorflow/op/math/TruncateMod.java   |   6 +-
 .../op/math/UnsortedSegmentMax.java           |   6 +-
 .../op/math/UnsortedSegmentMin.java           |   6 +-
 .../op/math/UnsortedSegmentProd.java          |   6 +-
 .../op/math/UnsortedSegmentSum.java           |   6 +-
 .../java/org/tensorflow/op/math/Xdivy.java    |   6 +-
 .../java/org/tensorflow/op/math/Xlog1py.java  |   6 +-
 .../java/org/tensorflow/op/math/Xlogy.java    |   6 +-
 .../gen/java/org/tensorflow/op/math/Zeta.java |   6 +-
 .../java/org/tensorflow/op/math/erfinv.java   |   6 +-
 .../org/tensorflow/op/math/special/Dawsn.java |   6 +-
 .../tensorflow/op/math/special/Expint.java    |   6 +-
 .../op/math/special/FresnelCos.java           |   6 +-
 .../op/math/special/FresnelSin.java           |   6 +-
 .../tensorflow/op/math/special/Spence.java    |   6 +-
 .../java/org/tensorflow/op/nn/AvgPool.java    |   6 +-
 .../java/org/tensorflow/op/nn/AvgPool3d.java  |   6 +-
 .../org/tensorflow/op/nn/AvgPool3dGrad.java   |   6 +-
 .../org/tensorflow/op/nn/AvgPoolGrad.java     |   6 +-
 .../nn/BatchNormWithGlobalNormalization.java  |   6 +-
 .../BatchNormWithGlobalNormalizationGrad.java |   6 +-
 .../java/org/tensorflow/op/nn/BiasAdd.java    |   6 +-
 .../org/tensorflow/op/nn/BiasAddGrad.java     |   6 +-
 .../java/org/tensorflow/op/nn/BlockLSTM.java  |   6 +-
 .../org/tensorflow/op/nn/BlockLSTMGrad.java   |   6 +-
 .../gen/java/org/tensorflow/op/nn/Conv2d.java |   6 +-
 .../op/nn/Conv2dBackpropFilter.java           |   6 +-
 .../tensorflow/op/nn/Conv2dBackpropInput.java |   6 +-
 .../gen/java/org/tensorflow/op/nn/Conv3d.java |   6 +-
 .../op/nn/Conv3dBackpropFilter.java           |   6 +-
 .../tensorflow/op/nn/Conv3dBackpropInput.java |   6 +-
 .../op/nn/CtcBeamSearchDecoder.java           |   6 +-
 .../tensorflow/op/nn/CtcGreedyDecoder.java    |   6 +-
 .../java/org/tensorflow/op/nn/CtcLoss.java    |   6 +-
 .../java/org/tensorflow/op/nn/CudnnRNN.java   |   6 +-
 .../tensorflow/op/nn/CudnnRNNBackprop.java    |   6 +-
 .../op/nn/CudnnRNNCanonicalToParams.java      |   6 +-
 .../op/nn/CudnnRNNParamsToCanonical.java      |   6 +-
 .../tensorflow/op/nn/CudnnRnnParamsSize.java  |   6 +-
 .../tensorflow/op/nn/DataFormatDimMap.java    |   6 +-
 .../op/nn/DataFormatVecPermute.java           |   6 +-
 .../org/tensorflow/op/nn/DepthToSpace.java    |   6 +-
 .../op/nn/DepthwiseConv2dNative.java          |   6 +-
 .../DepthwiseConv2dNativeBackpropFilter.java  |   6 +-
 .../DepthwiseConv2dNativeBackpropInput.java   |   6 +-
 .../java/org/tensorflow/op/nn/Dilation2d.java |   6 +-
 .../op/nn/Dilation2dBackpropFilter.java       |   6 +-
 .../op/nn/Dilation2dBackpropInput.java        |   6 +-
 .../gen/java/org/tensorflow/op/nn/Elu.java    |   6 +-
 .../java/org/tensorflow/op/nn/EluGrad.java    |   6 +-
 .../tensorflow/op/nn/FractionalAvgPool.java   |   6 +-
 .../op/nn/FractionalAvgPoolGrad.java          |   6 +-
 .../tensorflow/op/nn/FractionalMaxPool.java   |   6 +-
 .../op/nn/FractionalMaxPoolGrad.java          |   6 +-
 .../org/tensorflow/op/nn/FusedBatchNorm.java  |   6 +-
 .../tensorflow/op/nn/FusedBatchNormGrad.java  |   6 +-
 .../org/tensorflow/op/nn/FusedPadConv2d.java  |   6 +-
 .../op/nn/FusedResizeAndPadConv2d.java        |   6 +-
 .../org/tensorflow/op/nn/GRUBlockCell.java    |   6 +-
 .../tensorflow/op/nn/GRUBlockCellGrad.java    |   6 +-
 .../gen/java/org/tensorflow/op/nn/InTopK.java |   4 +-
 .../java/org/tensorflow/op/nn/InvGrad.java    |   6 +-
 .../gen/java/org/tensorflow/op/nn/L2Loss.java |   6 +-
 .../org/tensorflow/op/nn/LSTMBlockCell.java   |   6 +-
 .../tensorflow/op/nn/LSTMBlockCellGrad.java   |   6 +-
 .../java/org/tensorflow/op/nn/LeakyRelu.java  |   6 +-
 .../op/nn/LocalResponseNormalization.java     |   6 +-
 .../op/nn/LocalResponseNormalizationGrad.java |   6 +-
 .../java/org/tensorflow/op/nn/LogSoftmax.java |   6 +-
 .../java/org/tensorflow/op/nn/MaxPool.java    |   6 +-
 .../java/org/tensorflow/op/nn/MaxPool3d.java  |   6 +-
 .../org/tensorflow/op/nn/MaxPool3dGrad.java   |   6 +-
 .../tensorflow/op/nn/MaxPool3dGradGrad.java   |   6 +-
 .../org/tensorflow/op/nn/MaxPoolGrad.java     |   6 +-
 .../org/tensorflow/op/nn/MaxPoolGradGrad.java |   6 +-
 .../op/nn/MaxPoolGradGradWithArgmax.java      |   6 +-
 .../op/nn/MaxPoolGradWithArgmax.java          |   6 +-
 .../tensorflow/op/nn/MaxPoolWithArgmax.java   |   8 +-
 .../java/org/tensorflow/op/nn/NthElement.java |   6 +-
 .../tensorflow/op/nn/QuantizedAvgPool.java    |   6 +-
 ...tizedBatchNormWithGlobalNormalization.java |   6 +-
 .../tensorflow/op/nn/QuantizedBiasAdd.java    |   6 +-
 .../op/nn/QuantizedConv2DAndRelu.java         |   6 +-
 .../QuantizedConv2DAndReluAndRequantize.java  |   6 +-
 .../op/nn/QuantizedConv2DAndRequantize.java   |   6 +-
 .../op/nn/QuantizedConv2DPerChannel.java      |   6 +-
 .../op/nn/QuantizedConv2DWithBias.java        |   6 +-
 .../op/nn/QuantizedConv2DWithBiasAndRelu.java |   6 +-
 ...zedConv2DWithBiasAndReluAndRequantize.java |   6 +-
 .../QuantizedConv2DWithBiasAndRequantize.java |   6 +-
 ...WithBiasSignedSumAndReluAndRequantize.java |   6 +-
 .../nn/QuantizedConv2DWithBiasSumAndRelu.java |   6 +-
 ...Conv2DWithBiasSumAndReluAndRequantize.java |   6 +-
 .../org/tensorflow/op/nn/QuantizedConv2d.java |   6 +-
 .../op/nn/QuantizedDepthwiseConv2D.java       |   6 +-
 .../nn/QuantizedDepthwiseConv2DWithBias.java  |   6 +-
 ...antizedDepthwiseConv2DWithBiasAndRelu.java |   6 +-
 ...iseConv2DWithBiasAndReluAndRequantize.java |   6 +-
 .../op/nn/QuantizedInstanceNorm.java          |   6 +-
 .../tensorflow/op/nn/QuantizedMaxPool.java    |   6 +-
 .../org/tensorflow/op/nn/QuantizedRelu.java   |   6 +-
 .../org/tensorflow/op/nn/QuantizedRelu6.java  |   6 +-
 .../org/tensorflow/op/nn/QuantizedReluX.java  |   6 +-
 .../gen/java/org/tensorflow/op/nn/Relu.java   |   6 +-
 .../gen/java/org/tensorflow/op/nn/Relu6.java  |   6 +-
 .../java/org/tensorflow/op/nn/Relu6Grad.java  |   6 +-
 .../java/org/tensorflow/op/nn/ReluGrad.java   |   6 +-
 .../gen/java/org/tensorflow/op/nn/Selu.java   |   6 +-
 .../java/org/tensorflow/op/nn/SeluGrad.java   |   6 +-
 .../java/org/tensorflow/op/nn/Softmax.java    |   6 +-
 .../java/org/tensorflow/op/nn/Softsign.java   |   6 +-
 .../org/tensorflow/op/nn/SoftsignGrad.java    |   6 +-
 .../org/tensorflow/op/nn/SpaceToBatch.java    |   6 +-
 .../org/tensorflow/op/nn/SpaceToDepth.java    |   6 +-
 .../gen/java/org/tensorflow/op/nn/TopK.java   |   6 +-
 .../nn/raw/SoftmaxCrossEntropyWithLogits.java |  13 +-
 .../SparseSoftmaxCrossEntropyWithLogits.java  |  13 +-
 .../op/quantization/Dequantize.java           |   8 +-
 .../tensorflow/op/quantization/Quantize.java  |   6 +-
 .../quantization/QuantizeAndDequantize.java   |   6 +-
 .../QuantizeDownAndShrinkRange.java           |   6 +-
 .../op/quantization/QuantizedConcat.java      |   6 +-
 .../QuantizedMatMulWithBiasAndDequantize.java |   6 +-
 .../QuantizedMatMulWithBiasAndRequantize.java |   6 +-
 .../op/quantization/RequantizationRange.java  |   4 +-
 .../op/quantization/Requantize.java           |   6 +-
 .../tensorflow/op/ragged/RaggedGather.java    |   6 +-
 .../org/tensorflow/op/ragged/RaggedRange.java |   8 +-
 .../op/ragged/RaggedTensorFromVariant.java    |   8 +-
 .../op/ragged/RaggedTensorToSparse.java       |   6 +-
 .../op/ragged/RaggedTensorToTensor.java       |   6 +-
 .../op/ragged/RaggedTensorToVariant.java      |  10 +-
 .../org/tensorflow/op/random/Multinomial.java |   8 +-
 .../op/random/NonDeterministicInts.java       |   8 +-
 .../random/ParameterizedTruncatedNormal.java  |   6 +-
 .../org/tensorflow/op/random/RandomGamma.java |   6 +-
 .../tensorflow/op/random/RandomGammaGrad.java |   6 +-
 .../tensorflow/op/random/RandomPoisson.java   |   8 +-
 .../tensorflow/op/random/RandomShuffle.java   |   6 +-
 .../op/random/RandomStandardNormal.java       |   6 +-
 .../tensorflow/op/random/RandomUniform.java   |   6 +-
 .../op/random/RandomUniformInt.java           |   6 +-
 .../op/random/StatefulRandomBinomial.java     |   8 +-
 .../op/random/StatefulStandardNormal.java     |   8 +-
 .../op/random/StatefulTruncatedNormal.java    |   8 +-
 .../tensorflow/op/random/StatefulUniform.java |   8 +-
 .../op/random/StatefulUniformFullInt.java     |   6 +-
 .../op/random/StatefulUniformInt.java         |   6 +-
 .../op/random/StatelessMultinomial.java       |   8 +-
 .../op/random/StatelessRandomBinomial.java    |   8 +-
 .../op/random/StatelessRandomGamma.java       |   6 +-
 .../op/random/StatelessRandomNormal.java      |   8 +-
 .../op/random/StatelessRandomPoisson.java     |   6 +-
 .../op/random/StatelessRandomUniform.java     |   8 +-
 .../random/StatelessRandomUniformFullInt.java |   6 +-
 .../op/random/StatelessRandomUniformInt.java  |   6 +-
 .../op/random/StatelessTruncatedNormal.java   |   8 +-
 .../tensorflow/op/random/TruncatedNormal.java |   6 +-
 .../org/tensorflow/op/signal/BatchFft.java    |   8 +-
 .../org/tensorflow/op/signal/BatchFft2d.java  |   8 +-
 .../org/tensorflow/op/signal/BatchFft3d.java  |   8 +-
 .../org/tensorflow/op/signal/BatchIfft.java   |   8 +-
 .../org/tensorflow/op/signal/BatchIfft2d.java |   8 +-
 .../org/tensorflow/op/signal/BatchIfft3d.java |   8 +-
 .../java/org/tensorflow/op/signal/Fft.java    |   6 +-
 .../java/org/tensorflow/op/signal/Fft2d.java  |   6 +-
 .../java/org/tensorflow/op/signal/Fft3d.java  |   6 +-
 .../java/org/tensorflow/op/signal/Ifft.java   |   6 +-
 .../java/org/tensorflow/op/signal/Ifft2d.java |   6 +-
 .../java/org/tensorflow/op/signal/Ifft3d.java |   6 +-
 .../java/org/tensorflow/op/signal/Irfft.java  |   8 +-
 .../org/tensorflow/op/signal/Irfft2d.java     |   8 +-
 .../org/tensorflow/op/signal/Irfft3d.java     |   8 +-
 .../java/org/tensorflow/op/signal/Rfft.java   |   6 +-
 .../java/org/tensorflow/op/signal/Rfft2d.java |   6 +-
 .../java/org/tensorflow/op/signal/Rfft3d.java |   6 +-
 .../op/sparse/AddManySparseToTensorsMap.java  |   4 +-
 .../op/sparse/AddSparseToTensorsMap.java      |   4 +-
 .../op/sparse/DenseToDenseSetOperation.java   |   6 +-
 .../op/sparse/DenseToSparseSetOperation.java  |   6 +-
 .../op/sparse/DeserializeSparse.java          |   6 +-
 .../SparseAccumulatorApplyGradient.java       |   4 +-
 .../sparse/SparseAccumulatorTakeGradient.java |   6 +-
 .../org/tensorflow/op/sparse/SparseAdd.java   |   6 +-
 .../tensorflow/op/sparse/SparseAddGrad.java   |   6 +-
 .../tensorflow/op/sparse/SparseConcat.java    |   6 +-
 .../sparse/SparseConditionalAccumulator.java  |   4 +-
 .../org/tensorflow/op/sparse/SparseCross.java |  35 +-
 .../op/sparse/SparseDenseCwiseAdd.java        |   6 +-
 .../op/sparse/SparseDenseCwiseDiv.java        |   6 +-
 .../op/sparse/SparseDenseCwiseMul.java        |   6 +-
 .../op/sparse/SparseFillEmptyRows.java        |   6 +-
 .../op/sparse/SparseFillEmptyRowsGrad.java    |   6 +-
 .../tensorflow/op/sparse/SparseMatMul.java    |   4 +-
 .../tensorflow/op/sparse/SparseReduceMax.java |   6 +-
 .../op/sparse/SparseReduceMaxSparse.java      |   6 +-
 .../tensorflow/op/sparse/SparseReduceSum.java |   6 +-
 .../op/sparse/SparseReduceSumSparse.java      |   6 +-
 .../tensorflow/op/sparse/SparseReorder.java   |   6 +-
 .../op/sparse/SparseSegmentMean.java          |  10 +-
 .../op/sparse/SparseSegmentMeanGrad.java      |   9 +-
 .../SparseSegmentMeanWithNumSegments.java     |  12 +-
 .../op/sparse/SparseSegmentSqrtN.java         |  10 +-
 .../op/sparse/SparseSegmentSqrtNGrad.java     |   9 +-
 .../SparseSegmentSqrtNWithNumSegments.java    |  12 +-
 .../op/sparse/SparseSegmentSum.java           |  10 +-
 .../SparseSegmentSumWithNumSegments.java      |  12 +-
 .../org/tensorflow/op/sparse/SparseSlice.java |   6 +-
 .../tensorflow/op/sparse/SparseSliceGrad.java |   6 +-
 .../tensorflow/op/sparse/SparseSoftmax.java   |   6 +-
 .../op/sparse/SparseSparseMaximum.java        |   6 +-
 .../op/sparse/SparseSparseMinimum.java        |   6 +-
 .../org/tensorflow/op/sparse/SparseSplit.java |   6 +-
 .../op/sparse/SparseTensorDenseAdd.java       |   6 +-
 .../op/sparse/SparseTensorDenseMatMul.java    |   6 +-
 .../tensorflow/op/sparse/SparseToDense.java   |   6 +-
 .../op/sparse/SparseToSparseSetOperation.java |   6 +-
 .../sparse/TakeManySparseFromTensorsMap.java  |   6 +-
 .../tensorflow/op/strings/StringNGrams.java   |   6 +-
 .../org/tensorflow/op/strings/Substr.java     |   4 +-
 .../org/tensorflow/op/strings/ToNumber.java   |   6 +-
 .../tensorflow/op/strings/UnicodeDecode.java  |   6 +-
 .../op/strings/UnicodeDecodeWithOffsets.java  |   6 +-
 .../tensorflow/op/strings/UnicodeEncode.java  |   4 +-
 .../op/strings/UnsortedSegmentJoin.java       |   4 +-
 .../op/summary/HistogramSummary.java          |   4 +-
 .../tensorflow/op/summary/ImageSummary.java   |   3 +-
 .../tensorflow/op/summary/ScalarSummary.java  |   4 +-
 .../tensorflow/op/summary/SummaryWriter.java  |   8 +-
 .../tensorflow/op/summary/TensorSummary.java  |   4 +-
 .../op/summary/WriteHistogramSummary.java     |   4 +-
 .../op/summary/WriteImageSummary.java         |   4 +-
 .../op/summary/WriteScalarSummary.java        |   4 +-
 .../tensorflow/op/summary/WriteSummary.java   |   4 +-
 .../java/org/tensorflow/op/tpu/AllToAll.java  |   6 +-
 .../tensorflow/op/tpu/CollectivePermute.java  |   6 +-
 .../tensorflow/op/tpu/CrossReplicaSum.java    |   6 +-
 .../tpu/EnqueueTPUEmbeddingSparseBatch.java   |   4 +-
 .../EnqueueTPUEmbeddingSparseTensorBatch.java |   4 +-
 .../org/tensorflow/op/tpu/InfeedDequeue.java  |   6 +-
 .../tensorflow/op/tpu/InfeedDequeueTuple.java |   6 +-
 .../org/tensorflow/op/tpu/InfeedEnqueue.java  |   4 +-
 .../org/tensorflow/op/tpu/OutfeedDequeue.java |   6 +-
 .../op/tpu/OutfeedDequeueTuple.java           |   6 +-
 .../org/tensorflow/op/tpu/OutfeedEnqueue.java |   4 +-
 .../org/tensorflow/op/tpu/Prelinearize.java   |  10 +-
 .../tensorflow/op/tpu/PrelinearizeTuple.java  |   8 +-
 .../tensorflow/op/tpu/TPUReplicatedInput.java |   6 +-
 .../op/tpu/TPUReplicatedOutput.java           |   6 +-
 .../op/train/AccumulatorApplyGradient.java    |   4 +-
 .../op/train/AccumulatorTakeGradient.java     |   6 +-
 .../org/tensorflow/op/train/ApplyAdaMax.java  |   6 +-
 .../tensorflow/op/train/ApplyAdadelta.java    |   6 +-
 .../org/tensorflow/op/train/ApplyAdagrad.java |   6 +-
 .../tensorflow/op/train/ApplyAdagradDa.java   |   6 +-
 .../tensorflow/op/train/ApplyAdagradV2.java   |   6 +-
 .../org/tensorflow/op/train/ApplyAdam.java    |   6 +-
 .../org/tensorflow/op/train/ApplyAddSign.java |   6 +-
 .../op/train/ApplyCenteredRmsProp.java        |   6 +-
 .../org/tensorflow/op/train/ApplyFtrl.java    |   6 +-
 .../op/train/ApplyGradientDescent.java        |   6 +-
 .../tensorflow/op/train/ApplyMomentum.java    |   6 +-
 .../tensorflow/op/train/ApplyPowerSign.java   |   6 +-
 .../op/train/ApplyProximalAdagrad.java        |   6 +-
 .../train/ApplyProximalGradientDescent.java   |   6 +-
 .../org/tensorflow/op/train/ApplyRmsProp.java |   6 +-
 .../org/tensorflow/op/train/BatchMatMul.java  |   6 +-
 .../op/train/ConditionalAccumulator.java      |   4 +-
 .../tensorflow/op/train/PreventGradient.java  |   6 +-
 .../ResourceAccumulatorApplyGradient.java     |   4 +-
 .../ResourceAccumulatorTakeGradient.java      |   6 +-
 .../op/train/ResourceApplyAdaMax.java         |   4 +-
 .../op/train/ResourceApplyAdadelta.java       |   4 +-
 .../op/train/ResourceApplyAdagrad.java        |   4 +-
 .../op/train/ResourceApplyAdagradDa.java      |   4 +-
 .../op/train/ResourceApplyAdam.java           |   4 +-
 .../train/ResourceApplyAdamWithAmsgrad.java   |   4 +-
 .../op/train/ResourceApplyAddSign.java        |   4 +-
 .../train/ResourceApplyCenteredRmsProp.java   |   4 +-
 .../op/train/ResourceApplyFtrl.java           |   4 +-
 .../train/ResourceApplyGradientDescent.java   |   4 +-
 .../op/train/ResourceApplyKerasMomentum.java  |   4 +-
 .../op/train/ResourceApplyMomentum.java       |   4 +-
 .../op/train/ResourceApplyPowerSign.java      |   4 +-
 .../train/ResourceApplyProximalAdagrad.java   |   4 +-
 .../ResourceApplyProximalGradientDescent.java |   4 +-
 .../op/train/ResourceApplyRmsProp.java        |   4 +-
 .../train/ResourceConditionalAccumulator.java |  10 +-
 .../op/train/ResourceSparseApplyAdadelta.java |   4 +-
 .../op/train/ResourceSparseApplyAdagrad.java  |   4 +-
 .../train/ResourceSparseApplyAdagradDa.java   |   4 +-
 .../train/ResourceSparseApplyAdagradV2.java   |   4 +-
 .../ResourceSparseApplyCenteredRmsProp.java   |   4 +-
 .../op/train/ResourceSparseApplyFtrl.java     |   4 +-
 .../ResourceSparseApplyKerasMomentum.java     |   4 +-
 .../op/train/ResourceSparseApplyMomentum.java |   4 +-
 .../ResourceSparseApplyProximalAdagrad.java   |   4 +-
 ...rceSparseApplyProximalGradientDescent.java |   4 +-
 .../op/train/ResourceSparseApplyRmsProp.java  |   4 +-
 .../java/org/tensorflow/op/train/Restore.java |   6 +-
 .../org/tensorflow/op/train/RestoreSlice.java |   6 +-
 .../op/train/SparseApplyAdadelta.java         |   6 +-
 .../op/train/SparseApplyAdagrad.java          |   6 +-
 .../op/train/SparseApplyAdagradDa.java        |   6 +-
 .../op/train/SparseApplyCenteredRmsProp.java  |   6 +-
 .../tensorflow/op/train/SparseApplyFtrl.java  |   6 +-
 .../op/train/SparseApplyMomentum.java         |   6 +-
 .../op/train/SparseApplyProximalAdagrad.java  |   6 +-
 .../SparseApplyProximalGradientDescent.java   |   6 +-
 .../op/train/SparseApplyRmsProp.java          |   6 +-
 .../org/tensorflow/op/train/TileGrad.java     |   6 +-
 .../tensorflow/op/xla/BroadcastHelper.java    |   6 +-
 .../org/tensorflow/op/xla/ClusterOutput.java  |   6 +-
 .../gen/java/org/tensorflow/op/xla/Conv.java  |   6 +-
 .../gen/java/org/tensorflow/op/xla/Dot.java   |   6 +-
 .../org/tensorflow/op/xla/DynamicSlice.java   |   6 +-
 .../tensorflow/op/xla/DynamicUpdateSlice.java |   6 +-
 .../java/org/tensorflow/op/xla/Einsum.java    |   6 +-
 .../java/org/tensorflow/op/xla/Gather.java    |   6 +-
 .../org/tensorflow/op/xla/KeyValueSort.java   |   6 +-
 .../gen/java/org/tensorflow/op/xla/Pad.java   |   6 +-
 .../gen/java/org/tensorflow/op/xla/Recv.java  |   6 +-
 .../org/tensorflow/op/xla/SelfAdjointEig.java |   6 +-
 .../gen/java/org/tensorflow/op/xla/Send.java  |   4 +-
 .../java/org/tensorflow/op/xla/Sharding.java  |   6 +-
 .../gen/java/org/tensorflow/op/xla/Sort.java  |   6 +-
 .../gen/java/org/tensorflow/op/xla/Svd.java   |   6 +-
 .../org/tensorflow/AbstractOperation.java     |   3 +-
 .../main/java/org/tensorflow/DataType.java    | 103 +--
 .../src/main/java/org/tensorflow/Operand.java |   2 +-
 .../main/java/org/tensorflow/Operation.java   |   4 +-
 .../src/main/java/org/tensorflow/Output.java  |   5 +-
 .../src/main/java/org/tensorflow/Tensor.java  |  28 +-
 .../src/main/java/org/tensorflow/Tensors.java |   2 +-
 .../internal/tensor/BooleanTensorImpl.java    |  96 +++
 .../internal/tensor/ByteTensorImpl.java       |  95 +++
 .../internal/tensor/DoubleTensorImpl.java     |  96 +++
 .../internal/tensor/FloatTensorImpl.java      |  96 +++
 .../internal/tensor/IntTensorImpl.java        |  96 +++
 .../internal/tensor/LongTensorImpl.java       |  96 +++
 .../{types => internal/tensor}/RawTensor.java |  20 +-
 .../tensor/StringTensorImpl.java}             |  58 +-
 .../buffer/ByteSequenceTensorBuffer.java      |   2 +-
 .../{ => tensor}/buffer/TensorBuffers.java    |   2 +-
 .../buffer/TensorRawDataBufferFactory.java    |   2 +-
 .../java/org/tensorflow/op/core/Constant.java | 164 ++---
 .../org/tensorflow/op/core/Gradients.java     |   5 +-
 .../java/org/tensorflow/op/core/Helpers.java  |   4 +-
 .../java/org/tensorflow/op/core/Shapes.java   |  40 +-
 .../java/org/tensorflow/op/core/Zeros.java    |   6 +-
 .../org/tensorflow/tensor/BooleanTensor.java  |  40 ++
 .../org/tensorflow/tensor/ByteTensor.java     |  39 ++
 .../org/tensorflow/tensor/DoubleTensor.java   |  40 ++
 .../org/tensorflow/tensor/FloatTensor.java    |  40 ++
 .../java/org/tensorflow/tensor/IntTensor.java |  40 ++
 .../org/tensorflow/tensor/LongTensor.java     |  40 ++
 .../org/tensorflow/tensor/StringTensor.java   |  29 +
 .../org/tensorflow/types/BooleanTensor.java   | 126 ----
 .../java/org/tensorflow/types/ByteTensor.java | 125 ----
 .../org/tensorflow/types/DoubleTensor.java    | 126 ----
 .../org/tensorflow/types/FloatTensor.java     | 126 ----
 .../java/org/tensorflow/types/IntTensor.java  | 126 ----
 .../java/org/tensorflow/types/LongTensor.java | 126 ----
 .../java/org/tensorflow/types/TBfloat16.java  |  11 +-
 .../main/java/org/tensorflow/types/TBool.java |   8 +-
 .../java/org/tensorflow/types/TFloat16.java   |  10 +-
 .../java/org/tensorflow/types/TFloat32.java   |  16 +-
 .../java/org/tensorflow/types/TFloat64.java   |  16 +-
 .../java/org/tensorflow/types/TInt32.java     |   8 +-
 .../java/org/tensorflow/types/TInt64.java     |   8 +-
 .../java/org/tensorflow/types/TString.java    |  10 +-
 .../java/org/tensorflow/types/TUint8.java     |   8 +-
 .../tensorflow/types/family/package-info.java |   3 -
 .../tensorflow/EagerOperationBuilderTest.java |   2 +-
 .../org/tensorflow/EagerOperationTest.java    |   4 +-
 .../tensorflow/GraphOperationBuilderTest.java |  10 +-
 .../test/java/org/tensorflow/GraphTest.java   |  36 +-
 .../test/java/org/tensorflow/SessionTest.java |  36 +-
 .../test/java/org/tensorflow/TensorTest.java  | 142 ++--
 .../java/org/tensorflow/op/ScopeTest.java     |  21 +-
 .../op/core/GeneratedOperationsTest.java      |  12 +-
 .../org/tensorflow/op/core/GradientsTest.java |  14 +-
 .../org/tensorflow/op/core/ShapesTest.java    | 104 +--
 .../org/tensorflow/op/core/ZerosTest.java     |  28 +-
 .../types/NumericTypesTestBase.java           |  40 +-
 .../org/tensorflow/types/TBfloat16Test.java   |   2 +-
 .../org/tensorflow/types/TFloat16Test.java    |   2 +-
 .../org/tensorflow/types/TFloat32Test.java    |   2 +-
 .../org/tensorflow/types/TFloat64Test.java    |   2 +-
 .../java/org/tensorflow/types/TInt32Test.java |   2 +-
 .../java/org/tensorflow/types/TInt64Test.java |   2 +-
 .../org/tensorflow/types/TStringTest.java     |  41 +-
 .../java/org/tensorflow/types/TUint8Test.java |   2 +-
 940 files changed, 4925 insertions(+), 5212 deletions(-)
 create mode 100644 tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/BooleanTensorImpl.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/ByteTensorImpl.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/DoubleTensorImpl.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/FloatTensorImpl.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/IntTensorImpl.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/LongTensorImpl.java
 rename tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/{types => internal/tensor}/RawTensor.java (75%)
 rename tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/{types/StringTensor.java => internal/tensor/StringTensorImpl.java} (55%)
 rename tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/{ => tensor}/buffer/ByteSequenceTensorBuffer.java (99%)
 rename tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/{ => tensor}/buffer/TensorBuffers.java (99%)
 rename tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/{ => tensor}/buffer/TensorRawDataBufferFactory.java (98%)
 create mode 100644 tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/tensor/BooleanTensor.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/tensor/ByteTensor.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/tensor/DoubleTensor.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/tensor/FloatTensor.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/tensor/IntTensor.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/tensor/LongTensor.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/tensor/StringTensor.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/BooleanTensor.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/ByteTensor.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/DoubleTensor.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/FloatTensor.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/IntTensor.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/LongTensor.java

diff --git a/tensorflow-core/tensorflow-core-api/src/bazel/op_generator/op_generator.cc b/tensorflow-core/tensorflow-core-api/src/bazel/op_generator/op_generator.cc
index 843f3bdb247..b58b3d3a59f 100644
--- a/tensorflow-core/tensorflow-core-api/src/bazel/op_generator/op_generator.cc
+++ b/tensorflow-core/tensorflow-core-api/src/bazel/op_generator/op_generator.cc
@@ -348,7 +348,7 @@ void RenderInterfaceImpl(const OpSpec& op, RenderMode mode,
   if (mode == OPERAND) {
     bool cast2obj = output.type().wildcard();
     Type return_type = Type::Class("Output", "org.tensorflow")
-        .add_parameter(cast2obj ? Type::Class("TType", "org.tensorflow.types.family") : output.type());
+        .add_parameter(cast2obj ? Type::Interface("Tensor", "org.tensorflow") : output.type());
     Method as_output = Method::Create("asOutput", return_type)
         .add_annotation(Annotation::Create("Override"));
     if (cast2obj) {
@@ -366,7 +366,7 @@ void RenderInterfaceImpl(const OpSpec& op, RenderMode mode,
   } else if (mode == LIST_OPERAND) {
     Type operand = Type::Interface("Operand", "org.tensorflow");
     if (output.type().wildcard()) {
-      operand.add_parameter(Type::Class("TType", "org.tensorflow.types.family"));
+      operand.add_parameter(Type::Interface("Tensor", "org.tensorflow"));
     } else {
       operand.add_parameter(output.type());
     }
@@ -430,10 +430,9 @@ void GenerateOp(const OpSpec& op, const EndpointSpec& endpoint,
   RenderMode mode = DEFAULT;
   if (op.outputs().size() == 1) {
     const ArgumentSpec& output = op.outputs().front();
-    Type operand_type(output.type().wildcard() ? Type::Class("TType", "org.tensorflow.types.family")
+    Type operand_type(output.type().wildcard() ? Type::Interface("Tensor", "org.tensorflow")
                                                : output.type());
-    Type operand_inf(Type::Interface("Operand", "org.tensorflow")
-                         .add_parameter(operand_type));
+    Type operand_inf(Type::Interface("Operand", "org.tensorflow").add_parameter(operand_type));
     if (output.iterable()) {
       mode = LIST_OPERAND;
       op_class.add_supertype(Type::IterableOf(operand_inf));
diff --git a/tensorflow-core/tensorflow-core-api/src/bazel/op_generator/op_specs.cc b/tensorflow-core/tensorflow-core-api/src/bazel/op_generator/op_specs.cc
index c9e0525edb7..3c981bf8816 100644
--- a/tensorflow-core/tensorflow-core-api/src/bazel/op_generator/op_specs.cc
+++ b/tensorflow-core/tensorflow-core-api/src/bazel/op_generator/op_specs.cc
@@ -87,7 +87,7 @@ class TypeResolver {
       next_generic_letter_ = 'A';
     }
     return Type::Generic(string(1, generic_letter))
-        .add_supertype(Type::Class("TType", "org.tensorflow.types.family"));
+        .add_supertype(Type::Interface("Tensor", "org.tensorflow"));
   }
 };
 
@@ -158,7 +158,7 @@ std::pair<Type, Type> TypeResolver::TypesOf(const OpDef_AttrDef& attr_def,
   } else if (attr_type == "type") {
     Type type = *iterable_out ? Type::Wildcard() : NextGeneric();
     if (IsRealNumbers(attr_def.allowed_values())) {
-      type.add_supertype(Type::Class("TNumber", "org.tensorflow.types.family"));
+      type.add_supertype(Type::Interface("TNumber", "org.tensorflow.types.family"));
     }
     types = MakeTypePair(type, Type::Enum("DataType", "org.tensorflow"));
 
diff --git a/tensorflow-core/tensorflow-core-api/src/bazel/op_generator/source_writer.cc b/tensorflow-core/tensorflow-core-api/src/bazel/op_generator/source_writer.cc
index 8598b1d945d..9c36405c44c 100644
--- a/tensorflow-core/tensorflow-core-api/src/bazel/op_generator/source_writer.cc
+++ b/tensorflow-core/tensorflow-core-api/src/bazel/op_generator/source_writer.cc
@@ -321,8 +321,15 @@ SourceWriter& SourceWriter::WriteGenerics(
       Append(", ");
     }
     Append(pt->name());
-    if (!pt->supertypes().empty()) {
-      Append(" extends ").AppendType(pt->supertypes().front());
+    bool first_bound = true;
+    for (const Type& bound : pt->supertypes()) {
+      if (first_bound) {
+        Append(" extends ");
+        first_bound = false;
+      } else {
+        Append(" & ");
+      }
+      AppendType(bound);
     }
     first = false;
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/BitwiseOps.java b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/BitwiseOps.java
index 8ac6d565e51..9cc4506b77a 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/BitwiseOps.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/BitwiseOps.java
@@ -18,6 +18,7 @@
 package org.tensorflow.op;
 
 import org.tensorflow.Operand;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.bitwise.BitwiseAnd;
 import org.tensorflow.op.bitwise.BitwiseOr;
 import org.tensorflow.op.bitwise.BitwiseXor;
@@ -65,7 +66,7 @@ public final class BitwiseOps {
    * @param y
    * @return a new instance of BitwiseAnd
    */
-  public <T extends TNumber> BitwiseAnd<T> bitwiseAnd(Operand<T> x, Operand<T> y) {
+  public <T extends Tensor & TNumber> BitwiseAnd<T> bitwiseAnd(Operand<T> x, Operand<T> y) {
     return BitwiseAnd.create(scope, x, y);
   }
 
@@ -96,7 +97,7 @@ public <T extends TNumber> BitwiseAnd<T> bitwiseAnd(Operand<T> x, Operand<T> y)
    * @param y
    * @return a new instance of BitwiseOr
    */
-  public <T extends TNumber> BitwiseOr<T> bitwiseOr(Operand<T> x, Operand<T> y) {
+  public <T extends Tensor & TNumber> BitwiseOr<T> bitwiseOr(Operand<T> x, Operand<T> y) {
     return BitwiseOr.create(scope, x, y);
   }
 
@@ -127,7 +128,7 @@ public <T extends TNumber> BitwiseOr<T> bitwiseOr(Operand<T> x, Operand<T> y) {
    * @param y
    * @return a new instance of BitwiseXor
    */
-  public <T extends TNumber> BitwiseXor<T> bitwiseXor(Operand<T> x, Operand<T> y) {
+  public <T extends Tensor & TNumber> BitwiseXor<T> bitwiseXor(Operand<T> x, Operand<T> y) {
     return BitwiseXor.create(scope, x, y);
   }
 
@@ -178,7 +179,7 @@ public <T extends TNumber> BitwiseXor<T> bitwiseXor(Operand<T> x, Operand<T> y)
    * @param x
    * @return a new instance of Invert
    */
-  public <T extends TNumber> Invert<T> invert(Operand<T> x) {
+  public <T extends Tensor & TNumber> Invert<T> invert(Operand<T> x) {
     return Invert.create(scope, x);
   }
 
@@ -220,7 +221,7 @@ public <T extends TNumber> Invert<T> invert(Operand<T> x) {
    * @param y
    * @return a new instance of LeftShift
    */
-  public <T extends TNumber> LeftShift<T> leftShift(Operand<T> x, Operand<T> y) {
+  public <T extends Tensor & TNumber> LeftShift<T> leftShift(Operand<T> x, Operand<T> y) {
     return LeftShift.create(scope, x, y);
   }
 
@@ -265,7 +266,7 @@ public <T extends TNumber> LeftShift<T> leftShift(Operand<T> x, Operand<T> y) {
    * @param y
    * @return a new instance of RightShift
    */
-  public <T extends TNumber> RightShift<T> rightShift(Operand<T> x, Operand<T> y) {
+  public <T extends Tensor & TNumber> RightShift<T> rightShift(Operand<T> x, Operand<T> y) {
     return RightShift.create(scope, x, y);
   }
 }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/DtypesOps.java b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/DtypesOps.java
index 16d571a6428..92f4466f181 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/DtypesOps.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/DtypesOps.java
@@ -19,11 +19,11 @@
 
 import org.tensorflow.DataType;
 import org.tensorflow.Operand;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.dtypes.AsString;
 import org.tensorflow.op.dtypes.Cast;
 import org.tensorflow.op.dtypes.Complex;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * An API for building {@code dtypes} operations as {@link Op Op}s
@@ -57,7 +57,7 @@ public final class DtypesOps {
    * @param options carries optional attributes values
    * @return a new instance of AsString
    */
-  public <T extends TType> AsString asString(Operand<T> input, AsString.Options... options) {
+  public <T extends Tensor> AsString asString(Operand<T> input, AsString.Options... options) {
     return AsString.create(scope, input, options);
   }
 
@@ -70,7 +70,7 @@ public <T extends TType> AsString asString(Operand<T> input, AsString.Options...
    * @param options carries optional attributes values
    * @return a new instance of Cast
    */
-  public <U extends TType, T extends TType> Cast<U> cast(Operand<T> x, DataType<U> DstT,
+  public <U extends Tensor, T extends Tensor> Cast<U> cast(Operand<T> x, DataType<U> DstT,
       Cast.Options... options) {
     return Cast.create(scope, x, DstT, options);
   }
@@ -98,8 +98,8 @@ public <U extends TType, T extends TType> Cast<U> cast(Operand<T> x, DataType<U>
    * @param Tout
    * @return a new instance of Complex
    */
-  public <U extends TType, T extends TNumber> Complex<U> complex(Operand<T> real, Operand<T> imag,
-      DataType<U> Tout) {
+  public <U extends Tensor, T extends Tensor & TNumber> Complex<U> complex(Operand<T> real,
+      Operand<T> imag, DataType<U> Tout) {
     return Complex.create(scope, real, imag, Tout);
   }
 }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/ImageOps.java b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/ImageOps.java
index eea2fc4b8f1..0bcb966a2c6 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/ImageOps.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/ImageOps.java
@@ -20,6 +20,7 @@
 import java.util.List;
 import org.tensorflow.DataType;
 import org.tensorflow.Operand;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.image.AdjustContrast;
 import org.tensorflow.op.image.AdjustHue;
 import org.tensorflow.op.image.AdjustSaturation;
@@ -56,7 +57,6 @@
 import org.tensorflow.types.TString;
 import org.tensorflow.types.TUint8;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * An API for building {@code image} operations as {@link Op Op}s
@@ -88,7 +88,7 @@ public final class ImageOps {
    * @param contrastFactor A float multiplier for adjusting contrast.
    * @return a new instance of AdjustContrast
    */
-  public <T extends TNumber> AdjustContrast<T> adjustContrast(Operand<T> images,
+  public <T extends Tensor & TNumber> AdjustContrast<T> adjustContrast(Operand<T> images,
       Operand<TFloat32> contrastFactor) {
     return AdjustContrast.create(scope, images, contrastFactor);
   }
@@ -108,7 +108,8 @@ public <T extends TNumber> AdjustContrast<T> adjustContrast(Operand<T> images,
    * @param delta A float delta to add to the hue.
    * @return a new instance of AdjustHue
    */
-  public <T extends TNumber> AdjustHue<T> adjustHue(Operand<T> images, Operand<TFloat32> delta) {
+  public <T extends Tensor & TNumber> AdjustHue<T> adjustHue(Operand<T> images,
+      Operand<TFloat32> delta) {
     return AdjustHue.create(scope, images, delta);
   }
 
@@ -127,7 +128,7 @@ public <T extends TNumber> AdjustHue<T> adjustHue(Operand<T> images, Operand<TFl
    * @param scale A float scale to add to the saturation.
    * @return a new instance of AdjustSaturation
    */
-  public <T extends TNumber> AdjustSaturation<T> adjustSaturation(Operand<T> images,
+  public <T extends Tensor & TNumber> AdjustSaturation<T> adjustSaturation(Operand<T> images,
       Operand<TFloat32> scale) {
     return AdjustSaturation.create(scope, images, scale);
   }
@@ -211,8 +212,9 @@ public CombinedNonMaxSuppression combinedNonMaxSuppression(Operand<TFloat32> box
    * @param options carries optional attributes values
    * @return a new instance of CropAndResize
    */
-  public <T extends TNumber> CropAndResize cropAndResize(Operand<T> image, Operand<TFloat32> boxes,
-      Operand<TInt32> boxInd, Operand<TInt32> cropSize, CropAndResize.Options... options) {
+  public <T extends Tensor & TNumber> CropAndResize cropAndResize(Operand<T> image,
+      Operand<TFloat32> boxes, Operand<TInt32> boxInd, Operand<TInt32> cropSize,
+      CropAndResize.Options... options) {
     return CropAndResize.create(scope, image, boxes, boxInd, cropSize, options);
   }
 
@@ -237,8 +239,8 @@ public <T extends TNumber> CropAndResize cropAndResize(Operand<T> image, Operand
    * @param options carries optional attributes values
    * @return a new instance of CropAndResizeGradBoxes
    */
-  public <T extends TNumber> CropAndResizeGradBoxes cropAndResizeGradBoxes(Operand<TFloat32> grads,
-      Operand<T> image, Operand<TFloat32> boxes, Operand<TInt32> boxInd,
+  public <T extends Tensor & TNumber> CropAndResizeGradBoxes cropAndResizeGradBoxes(
+      Operand<TFloat32> grads, Operand<T> image, Operand<TFloat32> boxes, Operand<TInt32> boxInd,
       CropAndResizeGradBoxes.Options... options) {
     return CropAndResizeGradBoxes.create(scope, grads, image, boxes, boxInd, options);
   }
@@ -267,7 +269,7 @@ public <T extends TNumber> CropAndResizeGradBoxes cropAndResizeGradBoxes(Operand
    * @param options carries optional attributes values
    * @return a new instance of CropAndResizeGradImage
    */
-  public <T extends TNumber> CropAndResizeGradImage<T> cropAndResizeGradImage(
+  public <T extends Tensor & TNumber> CropAndResizeGradImage<T> cropAndResizeGradImage(
       Operand<TFloat32> grads, Operand<TFloat32> boxes, Operand<TInt32> boxInd,
       Operand<TInt32> imageSize, DataType<T> T, CropAndResizeGradImage.Options... options) {
     return CropAndResizeGradImage.create(scope, grads, boxes, boxInd, imageSize, T, options);
@@ -460,8 +462,8 @@ public DecodePng<TUint8> decodePng(Operand<TString> contents, DecodePng.Options.
    * @param options carries optional attributes values
    * @return a new instance of DecodePng
    */
-  public <T extends TNumber> DecodePng<T> decodePng(Operand<TString> contents, DataType<T> dtype,
-      DecodePng.Options... options) {
+  public <T extends Tensor & TNumber> DecodePng<T> decodePng(Operand<TString> contents,
+      DataType<T> dtype, DecodePng.Options... options) {
     return DecodePng.create(scope, contents, dtype, options);
   }
 
@@ -487,7 +489,7 @@ public <T extends TNumber> DecodePng<T> decodePng(Operand<TString> contents, Dat
    * @param colors 2-D. A list of RGBA colors to cycle through for the boxes.
    * @return a new instance of DrawBoundingBoxes
    */
-  public <T extends TNumber> DrawBoundingBoxes<T> drawBoundingBoxes(Operand<T> images,
+  public <T extends Tensor & TNumber> DrawBoundingBoxes<T> drawBoundingBoxes(Operand<T> images,
       Operand<TFloat32> boxes, Operand<TFloat32> colors) {
     return DrawBoundingBoxes.create(scope, images, boxes, colors);
   }
@@ -571,7 +573,8 @@ public EncodeJpegVariableQuality encodeJpegVariableQuality(Operand<TUint8> image
    * @param options carries optional attributes values
    * @return a new instance of EncodePng
    */
-  public <T extends TNumber> EncodePng encodePng(Operand<T> image, EncodePng.Options... options) {
+  public <T extends Tensor & TNumber> EncodePng encodePng(Operand<T> image,
+      EncodePng.Options... options) {
     return EncodePng.create(scope, image, options);
   }
 
@@ -592,7 +595,7 @@ public <T extends TNumber> EncodePng encodePng(Operand<T> image, EncodePng.Optio
    * @param padding The type of padding algorithm to use.
    * @return a new instance of ExtractImagePatches
    */
-  public <T extends TType> ExtractImagePatches<T> extractImagePatches(Operand<T> images,
+  public <T extends Tensor> ExtractImagePatches<T> extractImagePatches(Operand<T> images,
       List<Long> ksizes, List<Long> strides, List<Long> rates, String padding) {
     return ExtractImagePatches.create(scope, images, ksizes, strides, rates, padding);
   }
@@ -621,8 +624,8 @@ public ExtractJpegShape<TInt32> extractJpegShape(Operand<TString> contents) {
    *  Defaults to int32.
    * @return a new instance of ExtractJpegShape
    */
-  public <T extends TNumber> ExtractJpegShape<T> extractJpegShape(Operand<TString> contents,
-      DataType<T> outputType) {
+  public <T extends Tensor & TNumber> ExtractJpegShape<T> extractJpegShape(
+      Operand<TString> contents, DataType<T> outputType) {
     return ExtractJpegShape.create(scope, contents, outputType);
   }
 
@@ -639,7 +642,7 @@ public <T extends TNumber> ExtractJpegShape<T> extractJpegShape(Operand<TString>
    * @param images 1-D or higher rank. HSV data to convert. Last dimension must be size 3.
    * @return a new instance of HsvToRgb
    */
-  public <T extends TNumber> HsvToRgb<T> hsvToRgb(Operand<T> images) {
+  public <T extends Tensor & TNumber> HsvToRgb<T> hsvToRgb(Operand<T> images) {
     return HsvToRgb.create(scope, images);
   }
 
@@ -685,7 +688,7 @@ public <T extends TNumber> HsvToRgb<T> hsvToRgb(Operand<T> images) {
    * @param options carries optional attributes values
    * @return a new instance of NonMaxSuppression
    */
-  public <T extends TNumber> NonMaxSuppression<T> nonMaxSuppression(Operand<T> boxes,
+  public <T extends Tensor & TNumber> NonMaxSuppression<T> nonMaxSuppression(Operand<T> boxes,
       Operand<T> scores, Operand<TInt32> maxOutputSize, Operand<T> iouThreshold,
       Operand<T> scoreThreshold, Operand<T> softNmsSigma, NonMaxSuppression.Options... options) {
     return NonMaxSuppression.create(scope, boxes, scores, maxOutputSize, iouThreshold, scoreThreshold, softNmsSigma, options);
@@ -741,7 +744,7 @@ public NonMaxSuppressionWithOverlaps nonMaxSuppressionWithOverlaps(Operand<TFloa
    * @param options carries optional attributes values
    * @return a new instance of QuantizedResizeBilinear
    */
-  public <T extends TType> QuantizedResizeBilinear<T> quantizedResizeBilinear(Operand<T> images,
+  public <T extends Tensor> QuantizedResizeBilinear<T> quantizedResizeBilinear(Operand<T> images,
       Operand<TInt32> size, Operand<TFloat32> min, Operand<TFloat32> max,
       QuantizedResizeBilinear.Options... options) {
     return QuantizedResizeBilinear.create(scope, images, size, min, max, options);
@@ -763,8 +766,8 @@ public <T extends TType> QuantizedResizeBilinear<T> quantizedResizeBilinear(Oper
    * @param options carries optional attributes values
    * @return a new instance of RandomCrop
    */
-  public <T extends TNumber> RandomCrop<T> randomCrop(Operand<T> image, Operand<TInt64> size,
-      RandomCrop.Options... options) {
+  public <T extends Tensor & TNumber> RandomCrop<T> randomCrop(Operand<T> image,
+      Operand<TInt64> size, RandomCrop.Options... options) {
     return RandomCrop.create(scope, image, size, options);
   }
 
@@ -789,7 +792,7 @@ public <T extends TNumber> RandomCrop<T> randomCrop(Operand<T> image, Operand<TI
    * @param options carries optional attributes values
    * @return a new instance of ResizeArea
    */
-  public <T extends TNumber> ResizeArea resizeArea(Operand<T> images, Operand<TInt32> size,
+  public <T extends Tensor & TNumber> ResizeArea resizeArea(Operand<T> images, Operand<TInt32> size,
       ResizeArea.Options... options) {
     return ResizeArea.create(scope, images, size, options);
   }
@@ -805,8 +808,8 @@ public <T extends TNumber> ResizeArea resizeArea(Operand<T> images, Operand<TInt
    * @param options carries optional attributes values
    * @return a new instance of ResizeBicubic
    */
-  public <T extends TNumber> ResizeBicubic resizeBicubic(Operand<T> images, Operand<TInt32> size,
-      ResizeBicubic.Options... options) {
+  public <T extends Tensor & TNumber> ResizeBicubic resizeBicubic(Operand<T> images,
+      Operand<TInt32> size, ResizeBicubic.Options... options) {
     return ResizeBicubic.create(scope, images, size, options);
   }
 
@@ -821,8 +824,8 @@ public <T extends TNumber> ResizeBicubic resizeBicubic(Operand<T> images, Operan
    * @param options carries optional attributes values
    * @return a new instance of ResizeBilinear
    */
-  public <T extends TNumber> ResizeBilinear resizeBilinear(Operand<T> images, Operand<TInt32> size,
-      ResizeBilinear.Options... options) {
+  public <T extends Tensor & TNumber> ResizeBilinear resizeBilinear(Operand<T> images,
+      Operand<TInt32> size, ResizeBilinear.Options... options) {
     return ResizeBilinear.create(scope, images, size, options);
   }
 
@@ -836,8 +839,8 @@ public <T extends TNumber> ResizeBilinear resizeBilinear(Operand<T> images, Oper
    * @param options carries optional attributes values
    * @return a new instance of ResizeNearestNeighbor
    */
-  public <T extends TNumber> ResizeNearestNeighbor<T> resizeNearestNeighbor(Operand<T> images,
-      Operand<TInt32> size, ResizeNearestNeighbor.Options... options) {
+  public <T extends Tensor & TNumber> ResizeNearestNeighbor<T> resizeNearestNeighbor(
+      Operand<T> images, Operand<TInt32> size, ResizeNearestNeighbor.Options... options) {
     return ResizeNearestNeighbor.create(scope, images, size, options);
   }
 
@@ -867,7 +870,7 @@ public <T extends TNumber> ResizeNearestNeighbor<T> resizeNearestNeighbor(Operan
    * @param images 1-D or higher rank. RGB data to convert. Last dimension must be size 3.
    * @return a new instance of RgbToHsv
    */
-  public <T extends TNumber> RgbToHsv<T> rgbToHsv(Operand<T> images) {
+  public <T extends Tensor & TNumber> RgbToHsv<T> rgbToHsv(Operand<T> images) {
     return RgbToHsv.create(scope, images);
   }
 
@@ -922,7 +925,7 @@ public <T extends TNumber> RgbToHsv<T> rgbToHsv(Operand<T> images) {
    * @param options carries optional attributes values
    * @return a new instance of SampleDistortedBoundingBox
    */
-  public <T extends TNumber> SampleDistortedBoundingBox<T> sampleDistortedBoundingBox(
+  public <T extends Tensor & TNumber> SampleDistortedBoundingBox<T> sampleDistortedBoundingBox(
       Operand<T> imageSize, Operand<TFloat32> boundingBoxes, Operand<TFloat32> minObjectCovered,
       SampleDistortedBoundingBox.Options... options) {
     return SampleDistortedBoundingBox.create(scope, imageSize, boundingBoxes, minObjectCovered, options);
@@ -937,7 +940,7 @@ public <T extends TNumber> SampleDistortedBoundingBox<T> sampleDistortedBounding
    * @param options carries optional attributes values
    * @return a new instance of ScaleAndTranslate
    */
-  public <T extends TNumber> ScaleAndTranslate scaleAndTranslate(Operand<T> images,
+  public <T extends Tensor & TNumber> ScaleAndTranslate scaleAndTranslate(Operand<T> images,
       Operand<TInt32> size, Operand<TFloat32> scale, Operand<TFloat32> translation,
       ScaleAndTranslate.Options... options) {
     return ScaleAndTranslate.create(scope, images, size, scale, translation, options);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/IoOps.java b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/IoOps.java
index adc656dc5af..3f164450780 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/IoOps.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/IoOps.java
@@ -20,6 +20,7 @@
 import java.util.List;
 import org.tensorflow.DataType;
 import org.tensorflow.Operand;
+import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.io.DecodeBase64;
 import org.tensorflow.op.io.DecodeCompressed;
@@ -72,7 +73,6 @@
 import org.tensorflow.types.TInt64;
 import org.tensorflow.types.TString;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * An API for building {@code io} operations as {@link Op Op}s
@@ -167,8 +167,9 @@ public DecodeJsonExample decodeJsonExample(Operand<TString> jsonExamples) {
    * @param options carries optional attributes values
    * @return a new instance of DecodePaddedRaw
    */
-  public <T extends TNumber> DecodePaddedRaw<T> decodePaddedRaw(Operand<TString> inputBytes,
-      Operand<TInt32> fixedLength, DataType<T> outType, DecodePaddedRaw.Options... options) {
+  public <T extends Tensor & TNumber> DecodePaddedRaw<T> decodePaddedRaw(
+      Operand<TString> inputBytes, Operand<TInt32> fixedLength, DataType<T> outType,
+      DecodePaddedRaw.Options... options) {
     return DecodePaddedRaw.create(scope, inputBytes, fixedLength, outType, options);
   }
 
@@ -181,7 +182,7 @@ public <T extends TNumber> DecodePaddedRaw<T> decodePaddedRaw(Operand<TString> i
    * @param options carries optional attributes values
    * @return a new instance of DecodeRaw
    */
-  public <T extends TType> DecodeRaw<T> decodeRaw(Operand<TString> bytes, DataType<T> outType,
+  public <T extends Tensor> DecodeRaw<T> decodeRaw(Operand<TString> bytes, DataType<T> outType,
       DecodeRaw.Options... options) {
     return DecodeRaw.create(scope, bytes, outType, options);
   }
@@ -237,7 +238,7 @@ public <T extends TType> DecodeRaw<T> decodeRaw(Operand<TString> bytes, DataType
    * @param dtype The `dtype` of the serialized `SparseTensor` objects.
    * @return a new instance of DeserializeManySparse
    */
-  public <T extends TType> DeserializeManySparse<T> deserializeManySparse(
+  public <T extends Tensor> DeserializeManySparse<T> deserializeManySparse(
       Operand<TString> serializedSparse, DataType<T> dtype) {
     return DeserializeManySparse.create(scope, serializedSparse, dtype);
   }
@@ -565,7 +566,7 @@ public ParseSingleSequenceExample parseSingleSequenceExample(Operand<TString> se
    *  type of the serialized tensor and no implicit conversion will take place.
    * @return a new instance of ParseTensor
    */
-  public <T extends TType> ParseTensor<T> parseTensor(Operand<TString> serialized,
+  public <T extends Tensor> ParseTensor<T> parseTensor(Operand<TString> serialized,
       DataType<T> outType) {
     return ParseTensor.create(scope, serialized, outType);
   }
@@ -888,7 +889,7 @@ public ReaderSerializeState readerSerializeState(Operand<?> readerHandle) {
    * @param sparseShape 1-D.  The `shape` of the minibatch `SparseTensor`.
    * @return a new instance of SerializeManySparse
    */
-  public <T extends TType> SerializeManySparse<TString> serializeManySparse(
+  public <T extends Tensor> SerializeManySparse<TString> serializeManySparse(
       Operand<TInt64> sparseIndices, Operand<T> sparseValues, Operand<TInt64> sparseShape) {
     return SerializeManySparse.create(scope, sparseIndices, sparseValues, sparseShape);
   }
@@ -912,7 +913,7 @@ public <T extends TType> SerializeManySparse<TString> serializeManySparse(
    *  (default) and `variant`.
    * @return a new instance of SerializeManySparse
    */
-  public <U extends TType, T extends TType> SerializeManySparse<U> serializeManySparse(
+  public <U extends Tensor, T extends Tensor> SerializeManySparse<U> serializeManySparse(
       Operand<TInt64> sparseIndices, Operand<T> sparseValues, Operand<TInt64> sparseShape,
       DataType<U> outType) {
     return SerializeManySparse.create(scope, sparseIndices, sparseValues, sparseShape, outType);
@@ -927,7 +928,7 @@ public <U extends TType, T extends TType> SerializeManySparse<U> serializeManySp
    * @param sparseShape 1-D.  The `shape` of the `SparseTensor`.
    * @return a new instance of SerializeSparse
    */
-  public <T extends TType> SerializeSparse<TString> serializeSparse(Operand<TInt64> sparseIndices,
+  public <T extends Tensor> SerializeSparse<TString> serializeSparse(Operand<TInt64> sparseIndices,
       Operand<T> sparseValues, Operand<TInt64> sparseShape) {
     return SerializeSparse.create(scope, sparseIndices, sparseValues, sparseShape);
   }
@@ -943,7 +944,7 @@ public <T extends TType> SerializeSparse<TString> serializeSparse(Operand<TInt64
    *  (default) and `variant`.
    * @return a new instance of SerializeSparse
    */
-  public <U extends TType, T extends TType> SerializeSparse<U> serializeSparse(
+  public <U extends Tensor, T extends Tensor> SerializeSparse<U> serializeSparse(
       Operand<TInt64> sparseIndices, Operand<T> sparseValues, Operand<TInt64> sparseShape,
       DataType<U> outType) {
     return SerializeSparse.create(scope, sparseIndices, sparseValues, sparseShape, outType);
@@ -955,7 +956,7 @@ public <U extends TType, T extends TType> SerializeSparse<U> serializeSparse(
    * @param tensor A Tensor of type `T`.
    * @return a new instance of SerializeTensor
    */
-  public <T extends TType> SerializeTensor serializeTensor(Operand<T> tensor) {
+  public <T extends Tensor> SerializeTensor serializeTensor(Operand<T> tensor) {
     return SerializeTensor.create(scope, tensor);
   }
 
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/LinalgOps.java b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/LinalgOps.java
index b2242f1068e..7002747a317 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/LinalgOps.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/LinalgOps.java
@@ -19,6 +19,7 @@
 
 import org.tensorflow.DataType;
 import org.tensorflow.Operand;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.linalg.BandPart;
 import org.tensorflow.op.linalg.BatchCholesky;
 import org.tensorflow.op.linalg.BatchCholeskyGrad;
@@ -68,7 +69,6 @@
 import org.tensorflow.types.TInt64;
 import org.tensorflow.types.TString;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * An API for building {@code linalg} operations as {@link Op Op}s
@@ -128,7 +128,7 @@ public final class LinalgOps {
    *  entire upper triangle.
    * @return a new instance of BandPart
    */
-  public <T extends TType, U extends TNumber> BandPart<T> bandPart(Operand<T> input,
+  public <T extends Tensor, U extends Tensor & TNumber> BandPart<T> bandPart(Operand<T> input,
       Operand<U> numLower, Operand<U> numUpper) {
     return BandPart.create(scope, input, numLower, numUpper);
   }
@@ -139,7 +139,7 @@ public <T extends TType, U extends TNumber> BandPart<T> bandPart(Operand<T> inpu
    * @param input
    * @return a new instance of BatchCholesky
    */
-  public <T extends TNumber> BatchCholesky<T> batchCholesky(Operand<T> input) {
+  public <T extends Tensor & TNumber> BatchCholesky<T> batchCholesky(Operand<T> input) {
     return BatchCholesky.create(scope, input);
   }
 
@@ -150,7 +150,8 @@ public <T extends TNumber> BatchCholesky<T> batchCholesky(Operand<T> input) {
    * @param grad
    * @return a new instance of BatchCholeskyGrad
    */
-  public <T extends TNumber> BatchCholeskyGrad<T> batchCholeskyGrad(Operand<T> l, Operand<T> grad) {
+  public <T extends Tensor & TNumber> BatchCholeskyGrad<T> batchCholeskyGrad(Operand<T> l,
+      Operand<T> grad) {
     return BatchCholeskyGrad.create(scope, l, grad);
   }
 
@@ -162,7 +163,7 @@ public <T extends TNumber> BatchCholeskyGrad<T> batchCholeskyGrad(Operand<T> l,
    * @param numUpper
    * @return a new instance of BatchMatrixBandPart
    */
-  public <T extends TType> BatchMatrixBandPart<T> batchMatrixBandPart(Operand<T> input,
+  public <T extends Tensor> BatchMatrixBandPart<T> batchMatrixBandPart(Operand<T> input,
       Operand<TInt64> numLower, Operand<TInt64> numUpper) {
     return BatchMatrixBandPart.create(scope, input, numLower, numUpper);
   }
@@ -173,7 +174,7 @@ public <T extends TType> BatchMatrixBandPart<T> batchMatrixBandPart(Operand<T> i
    * @param input
    * @return a new instance of BatchMatrixDeterminant
    */
-  public <T extends TType> BatchMatrixDeterminant<T> batchMatrixDeterminant(Operand<T> input) {
+  public <T extends Tensor> BatchMatrixDeterminant<T> batchMatrixDeterminant(Operand<T> input) {
     return BatchMatrixDeterminant.create(scope, input);
   }
 
@@ -183,7 +184,7 @@ public <T extends TType> BatchMatrixDeterminant<T> batchMatrixDeterminant(Operan
    * @param diagonal
    * @return a new instance of BatchMatrixDiag
    */
-  public <T extends TType> BatchMatrixDiag<T> batchMatrixDiag(Operand<T> diagonal) {
+  public <T extends Tensor> BatchMatrixDiag<T> batchMatrixDiag(Operand<T> diagonal) {
     return BatchMatrixDiag.create(scope, diagonal);
   }
 
@@ -193,7 +194,7 @@ public <T extends TType> BatchMatrixDiag<T> batchMatrixDiag(Operand<T> diagonal)
    * @param input
    * @return a new instance of BatchMatrixDiagPart
    */
-  public <T extends TType> BatchMatrixDiagPart<T> batchMatrixDiagPart(Operand<T> input) {
+  public <T extends Tensor> BatchMatrixDiagPart<T> batchMatrixDiagPart(Operand<T> input) {
     return BatchMatrixDiagPart.create(scope, input);
   }
 
@@ -204,7 +205,7 @@ public <T extends TType> BatchMatrixDiagPart<T> batchMatrixDiagPart(Operand<T> i
    * @param options carries optional attributes values
    * @return a new instance of BatchMatrixInverse
    */
-  public <T extends TNumber> BatchMatrixInverse<T> batchMatrixInverse(Operand<T> input,
+  public <T extends Tensor & TNumber> BatchMatrixInverse<T> batchMatrixInverse(Operand<T> input,
       BatchMatrixInverse.Options... options) {
     return BatchMatrixInverse.create(scope, input, options);
   }
@@ -216,7 +217,7 @@ public <T extends TNumber> BatchMatrixInverse<T> batchMatrixInverse(Operand<T> i
    * @param diagonal
    * @return a new instance of BatchMatrixSetDiag
    */
-  public <T extends TType> BatchMatrixSetDiag<T> batchMatrixSetDiag(Operand<T> input,
+  public <T extends Tensor> BatchMatrixSetDiag<T> batchMatrixSetDiag(Operand<T> input,
       Operand<T> diagonal) {
     return BatchMatrixSetDiag.create(scope, input, diagonal);
   }
@@ -229,8 +230,8 @@ public <T extends TType> BatchMatrixSetDiag<T> batchMatrixSetDiag(Operand<T> inp
    * @param options carries optional attributes values
    * @return a new instance of BatchMatrixSolve
    */
-  public <T extends TNumber> BatchMatrixSolve<T> batchMatrixSolve(Operand<T> matrix, Operand<T> rhs,
-      BatchMatrixSolve.Options... options) {
+  public <T extends Tensor & TNumber> BatchMatrixSolve<T> batchMatrixSolve(Operand<T> matrix,
+      Operand<T> rhs, BatchMatrixSolve.Options... options) {
     return BatchMatrixSolve.create(scope, matrix, rhs, options);
   }
 
@@ -243,7 +244,7 @@ public <T extends TNumber> BatchMatrixSolve<T> batchMatrixSolve(Operand<T> matri
    * @param options carries optional attributes values
    * @return a new instance of BatchMatrixSolveLs
    */
-  public <T extends TNumber> BatchMatrixSolveLs<T> batchMatrixSolveLs(Operand<T> matrix,
+  public <T extends Tensor & TNumber> BatchMatrixSolveLs<T> batchMatrixSolveLs(Operand<T> matrix,
       Operand<T> rhs, Operand<TFloat64> l2Regularizer, BatchMatrixSolveLs.Options... options) {
     return BatchMatrixSolveLs.create(scope, matrix, rhs, l2Regularizer, options);
   }
@@ -256,7 +257,7 @@ public <T extends TNumber> BatchMatrixSolveLs<T> batchMatrixSolveLs(Operand<T> m
    * @param options carries optional attributes values
    * @return a new instance of BatchMatrixTriangularSolve
    */
-  public <T extends TNumber> BatchMatrixTriangularSolve<T> batchMatrixTriangularSolve(
+  public <T extends Tensor & TNumber> BatchMatrixTriangularSolve<T> batchMatrixTriangularSolve(
       Operand<T> matrix, Operand<T> rhs, BatchMatrixTriangularSolve.Options... options) {
     return BatchMatrixTriangularSolve.create(scope, matrix, rhs, options);
   }
@@ -268,7 +269,7 @@ public <T extends TNumber> BatchMatrixTriangularSolve<T> batchMatrixTriangularSo
    * @param options carries optional attributes values
    * @return a new instance of BatchSelfAdjointEig
    */
-  public <T extends TNumber> BatchSelfAdjointEig<T> batchSelfAdjointEig(Operand<T> input,
+  public <T extends Tensor & TNumber> BatchSelfAdjointEig<T> batchSelfAdjointEig(Operand<T> input,
       BatchSelfAdjointEig.Options... options) {
     return BatchSelfAdjointEig.create(scope, input, options);
   }
@@ -280,7 +281,7 @@ public <T extends TNumber> BatchSelfAdjointEig<T> batchSelfAdjointEig(Operand<T>
    * @param options carries optional attributes values
    * @return a new instance of BatchSvd
    */
-  public <T extends TType> BatchSvd<T> batchSvd(Operand<T> input, BatchSvd.Options... options) {
+  public <T extends Tensor> BatchSvd<T> batchSvd(Operand<T> input, BatchSvd.Options... options) {
     return BatchSvd.create(scope, input, options);
   }
 
@@ -305,7 +306,7 @@ public <T extends TType> BatchSvd<T> batchSvd(Operand<T> input, BatchSvd.Options
    * @param input Shape is `[..., M, M]`.
    * @return a new instance of Cholesky
    */
-  public <T extends TType> Cholesky<T> cholesky(Operand<T> input) {
+  public <T extends Tensor> Cholesky<T> cholesky(Operand<T> input) {
     return Cholesky.create(scope, input);
   }
 
@@ -324,7 +325,7 @@ public <T extends TType> Cholesky<T> cholesky(Operand<T> input) {
    *  this tensor.
    * @return a new instance of CholeskyGrad
    */
-  public <T extends TNumber> CholeskyGrad<T> choleskyGrad(Operand<T> l, Operand<T> grad) {
+  public <T extends Tensor & TNumber> CholeskyGrad<T> choleskyGrad(Operand<T> l, Operand<T> grad) {
     return CholeskyGrad.create(scope, l, grad);
   }
 
@@ -340,8 +341,8 @@ public <T extends TNumber> CholeskyGrad<T> choleskyGrad(Operand<T> l, Operand<T>
    * @param perm
    * @return a new instance of ConjugateTranspose
    */
-  public <T extends TType, U extends TNumber> ConjugateTranspose<T> conjugateTranspose(Operand<T> x,
-      Operand<U> perm) {
+  public <T extends Tensor, U extends Tensor & TNumber> ConjugateTranspose<T> conjugateTranspose(
+      Operand<T> x, Operand<U> perm) {
     return ConjugateTranspose.create(scope, x, perm);
   }
 
@@ -357,7 +358,7 @@ public <T extends TType, U extends TNumber> ConjugateTranspose<T> conjugateTrans
    * @param b Another tensor, of same type and shape as `a`.
    * @return a new instance of Cross
    */
-  public <T extends TNumber> Cross<T> cross(Operand<T> a, Operand<T> b) {
+  public <T extends Tensor & TNumber> Cross<T> cross(Operand<T> a, Operand<T> b) {
     return Cross.create(scope, a, b);
   }
 
@@ -372,7 +373,7 @@ public <T extends TNumber> Cross<T> cross(Operand<T> a, Operand<T> b) {
    * @param input Shape is `[..., M, M]`.
    * @return a new instance of Det
    */
-  public <T extends TType> Det<T> det(Operand<T> input) {
+  public <T extends Tensor> Det<T> det(Operand<T> input) {
     return Det.create(scope, input);
   }
 
@@ -396,7 +397,7 @@ public <T extends TType> Det<T> det(Operand<T> input) {
    * @param options carries optional attributes values
    * @return a new instance of Eig
    */
-  public <U extends TType, T extends TType> Eig<U> eig(Operand<T> input, DataType<U> Tout,
+  public <U extends Tensor, T extends Tensor> Eig<U> eig(Operand<T> input, DataType<U> Tout,
       Eig.Options... options) {
     return Eig.create(scope, input, Tout, options);
   }
@@ -484,7 +485,7 @@ public <U extends TType, T extends TType> Eig<U> eig(Operand<T> input, DataType<
    * @param equation String describing the Einstein Summation operation; in the format of np.einsum.
    * @return a new instance of Einsum
    */
-  public <T extends TType> Einsum<T> einsum(Iterable<Operand<T>> inputs, String equation) {
+  public <T extends Tensor> Einsum<T> einsum(Iterable<Operand<T>> inputs, String equation) {
     return Einsum.create(scope, inputs, equation);
   }
 
@@ -503,8 +504,8 @@ public <T extends TType> Einsum<T> einsum(Iterable<Operand<T>> inputs, String eq
    * @param options carries optional attributes values
    * @return a new instance of EuclideanNorm
    */
-  public <T extends TType, U extends TNumber> EuclideanNorm<T> euclideanNorm(Operand<T> input,
-      Operand<U> axis, EuclideanNorm.Options... options) {
+  public <T extends Tensor, U extends Tensor & TNumber> EuclideanNorm<T> euclideanNorm(
+      Operand<T> input, Operand<U> axis, EuclideanNorm.Options... options) {
     return EuclideanNorm.create(scope, input, axis, options);
   }
 
@@ -528,7 +529,7 @@ public <T extends TType, U extends TNumber> EuclideanNorm<T> euclideanNorm(Opera
    * @param options carries optional attributes values
    * @return a new instance of Inv
    */
-  public <T extends TType> Inv<T> inv(Operand<T> input, Inv.Options... options) {
+  public <T extends Tensor> Inv<T> inv(Operand<T> input, Inv.Options... options) {
     return Inv.create(scope, input, options);
   }
 
@@ -618,7 +619,7 @@ public LoadAndRemapMatrix loadAndRemapMatrix(Operand<TString> ckptPath,
    * @param input Shape is `[N, M, M]`.
    * @return a new instance of LogMatrixDeterminant
    */
-  public <T extends TType> LogMatrixDeterminant<T> logMatrixDeterminant(Operand<T> input) {
+  public <T extends Tensor> LogMatrixDeterminant<T> logMatrixDeterminant(Operand<T> input) {
     return LogMatrixDeterminant.create(scope, input);
   }
 
@@ -649,7 +650,7 @@ public <T extends TType> LogMatrixDeterminant<T> logMatrixDeterminant(Operand<T>
    *  size `[M, M]`.
    * @return a new instance of Lu
    */
-  public <T extends TType> Lu<T, TInt32> lu(Operand<T> input) {
+  public <T extends Tensor> Lu<T, TInt32> lu(Operand<T> input) {
     return Lu.create(scope, input);
   }
 
@@ -681,7 +682,7 @@ public <T extends TType> Lu<T, TInt32> lu(Operand<T> input) {
    * @param outputIdxType
    * @return a new instance of Lu
    */
-  public <T extends TType, U extends TNumber> Lu<T, U> lu(Operand<T> input,
+  public <T extends Tensor, U extends Tensor & TNumber> Lu<T, U> lu(Operand<T> input,
       DataType<U> outputIdxType) {
     return Lu.create(scope, input, outputIdxType);
   }
@@ -703,7 +704,8 @@ public <T extends TType, U extends TNumber> Lu<T, U> lu(Operand<T> input,
    * @param options carries optional attributes values
    * @return a new instance of MatMul
    */
-  public <T extends TType> MatMul<T> matMul(Operand<T> a, Operand<T> b, MatMul.Options... options) {
+  public <T extends Tensor> MatMul<T> matMul(Operand<T> a, Operand<T> b,
+      MatMul.Options... options) {
     return MatMul.create(scope, a, b, options);
   }
 
@@ -810,7 +812,7 @@ public <T extends TType> MatMul<T> matMul(Operand<T> a, Operand<T> b, MatMul.Opt
    *  Default is 0.
    * @return a new instance of MatrixDiag
    */
-  public <T extends TType> MatrixDiag<T> matrixDiag(Operand<T> diagonal, Operand<TInt32> k,
+  public <T extends Tensor> MatrixDiag<T> matrixDiag(Operand<T> diagonal, Operand<TInt32> k,
       Operand<TInt32> numRows, Operand<TInt32> numCols, Operand<T> paddingValue) {
     return MatrixDiag.create(scope, diagonal, k, numRows, numCols, paddingValue);
   }
@@ -894,7 +896,7 @@ public <T extends TType> MatrixDiag<T> matrixDiag(Operand<T> diagonal, Operand<T
    *  Default is 0.
    * @return a new instance of MatrixDiagPart
    */
-  public <T extends TType> MatrixDiagPart<T> matrixDiagPart(Operand<T> input, Operand<TInt32> k,
+  public <T extends Tensor> MatrixDiagPart<T> matrixDiagPart(Operand<T> input, Operand<TInt32> k,
       Operand<T> paddingValue) {
     return MatrixDiagPart.create(scope, input, k, paddingValue);
   }
@@ -1010,8 +1012,8 @@ public <T extends TType> MatrixDiagPart<T> matrixDiagPart(Operand<T> input, Oper
    * @param options carries optional attributes values
    * @return a new instance of MatrixDiagPartV3
    */
-  public <T extends TType> MatrixDiagPartV3<T> matrixDiagPartV3(Operand<T> input, Operand<TInt32> k,
-      Operand<T> paddingValue, MatrixDiagPartV3.Options... options) {
+  public <T extends Tensor> MatrixDiagPartV3<T> matrixDiagPartV3(Operand<T> input,
+      Operand<TInt32> k, Operand<T> paddingValue, MatrixDiagPartV3.Options... options) {
     return MatrixDiagPartV3.create(scope, input, k, paddingValue, options);
   }
 
@@ -1148,7 +1150,7 @@ public <T extends TType> MatrixDiagPartV3<T> matrixDiagPartV3(Operand<T> input,
    * @param options carries optional attributes values
    * @return a new instance of MatrixDiagV3
    */
-  public <T extends TType> MatrixDiagV3<T> matrixDiagV3(Operand<T> diagonal, Operand<TInt32> k,
+  public <T extends Tensor> MatrixDiagV3<T> matrixDiagV3(Operand<T> diagonal, Operand<TInt32> k,
       Operand<TInt32> numRows, Operand<TInt32> numCols, Operand<T> paddingValue,
       MatrixDiagV3.Options... options) {
     return MatrixDiagV3.create(scope, diagonal, k, numRows, numCols, paddingValue, options);
@@ -1269,7 +1271,7 @@ public <T extends TType> MatrixDiagV3<T> matrixDiagV3(Operand<T> diagonal, Opera
    * @param options carries optional attributes values
    * @return a new instance of MatrixSetDiag
    */
-  public <T extends TType> MatrixSetDiag<T> matrixSetDiag(Operand<T> input, Operand<T> diagonal,
+  public <T extends Tensor> MatrixSetDiag<T> matrixSetDiag(Operand<T> input, Operand<T> diagonal,
       Operand<TInt32> k, MatrixSetDiag.Options... options) {
     return MatrixSetDiag.create(scope, input, diagonal, k, options);
   }
@@ -1322,7 +1324,7 @@ public <T extends TType> MatrixSetDiag<T> matrixSetDiag(Operand<T> input, Operan
    * @param options carries optional attributes values
    * @return a new instance of MatrixSolveLs
    */
-  public <T extends TType> MatrixSolveLs<T> matrixSolveLs(Operand<T> matrix, Operand<T> rhs,
+  public <T extends Tensor> MatrixSolveLs<T> matrixSolveLs(Operand<T> matrix, Operand<T> rhs,
       Operand<TFloat64> l2Regularizer, MatrixSolveLs.Options... options) {
     return MatrixSolveLs.create(scope, matrix, rhs, l2Regularizer, options);
   }
@@ -1346,7 +1348,7 @@ public <T extends TType> MatrixSolveLs<T> matrixSolveLs(Operand<T> matrix, Opera
    * @param options carries optional attributes values
    * @return a new instance of Qr
    */
-  public <T extends TType> Qr<T> qr(Operand<T> input, Qr.Options... options) {
+  public <T extends Tensor> Qr<T> qr(Operand<T> input, Qr.Options... options) {
     return Qr.create(scope, input, options);
   }
 
@@ -1371,7 +1373,7 @@ public <T extends TType> Qr<T> qr(Operand<T> input, Qr.Options... options) {
    * @param options carries optional attributes values
    * @return a new instance of QuantizedMatMul
    */
-  public <V extends TType, T extends TType, U extends TType, W extends TType> QuantizedMatMul<V> quantizedMatMul(
+  public <V extends Tensor, T extends Tensor, U extends Tensor, W extends Tensor> QuantizedMatMul<V> quantizedMatMul(
       Operand<T> a, Operand<U> b, Operand<TFloat32> minA, Operand<TFloat32> maxA,
       Operand<TFloat32> minB, Operand<TFloat32> maxB, DataType<V> Toutput, DataType<W> Tactivation,
       QuantizedMatMul.Options... options) {
@@ -1397,7 +1399,7 @@ public <V extends TType, T extends TType, U extends TType, W extends TType> Quan
    * @param options carries optional attributes values
    * @return a new instance of SelfAdjointEig
    */
-  public <T extends TType> SelfAdjointEig<T> selfAdjointEig(Operand<T> input,
+  public <T extends Tensor> SelfAdjointEig<T> selfAdjointEig(Operand<T> input,
       SelfAdjointEig.Options... options) {
     return SelfAdjointEig.create(scope, input, options);
   }
@@ -1418,7 +1420,7 @@ public <T extends TType> SelfAdjointEig<T> selfAdjointEig(Operand<T> input,
    * @param options carries optional attributes values
    * @return a new instance of Solve
    */
-  public <T extends TType> Solve<T> solve(Operand<T> matrix, Operand<T> rhs,
+  public <T extends Tensor> Solve<T> solve(Operand<T> matrix, Operand<T> rhs,
       Solve.Options... options) {
     return Solve.create(scope, matrix, rhs, options);
   }
@@ -1446,7 +1448,7 @@ public <T extends TType> Solve<T> solve(Operand<T> matrix, Operand<T> rhs,
    * @param input Shape is `[..., M, M]`.
    * @return a new instance of Sqrtm
    */
-  public <T extends TType> Sqrtm<T> sqrtm(Operand<T> input) {
+  public <T extends Tensor> Sqrtm<T> sqrtm(Operand<T> input) {
     return Sqrtm.create(scope, input);
   }
 
@@ -1470,7 +1472,7 @@ public <T extends TType> Sqrtm<T> sqrtm(Operand<T> input) {
    * @param options carries optional attributes values
    * @return a new instance of Svd
    */
-  public <T extends TType> Svd<T> svd(Operand<T> input, Svd.Options... options) {
+  public <T extends Tensor> Svd<T> svd(Operand<T> input, Svd.Options... options) {
     return Svd.create(scope, input, options);
   }
 
@@ -1498,7 +1500,7 @@ public <T extends TType> Svd<T> svd(Operand<T> input, Svd.Options... options) {
    * @param diagonal Rank k tensor where k is at most 1.
    * @return a new instance of TensorDiag
    */
-  public <T extends TType> TensorDiag<T> tensorDiag(Operand<T> diagonal) {
+  public <T extends Tensor> TensorDiag<T> tensorDiag(Operand<T> diagonal) {
     return TensorDiag.create(scope, diagonal);
   }
 
@@ -1527,7 +1529,7 @@ public <T extends TType> TensorDiag<T> tensorDiag(Operand<T> diagonal) {
    * @param input Rank k tensor where k is even and not zero.
    * @return a new instance of TensorDiagPart
    */
-  public <T extends TType> TensorDiagPart<T> tensorDiagPart(Operand<T> input) {
+  public <T extends Tensor> TensorDiagPart<T> tensorDiagPart(Operand<T> input) {
     return TensorDiagPart.create(scope, input);
   }
 
@@ -1542,7 +1544,7 @@ public <T extends TType> TensorDiagPart<T> tensorDiagPart(Operand<T> input) {
    * @param perm
    * @return a new instance of Transpose
    */
-  public <T extends TType, U extends TNumber> Transpose<T> transpose(Operand<T> x,
+  public <T extends Tensor, U extends Tensor & TNumber> Transpose<T> transpose(Operand<T> x,
       Operand<U> perm) {
     return Transpose.create(scope, x, perm);
   }
@@ -1602,7 +1604,7 @@ public <T extends TType, U extends TNumber> Transpose<T> transpose(Operand<T> x,
    * @param options carries optional attributes values
    * @return a new instance of TriangularSolve
    */
-  public <T extends TType> TriangularSolve<T> triangularSolve(Operand<T> matrix, Operand<T> rhs,
+  public <T extends Tensor> TriangularSolve<T> triangularSolve(Operand<T> matrix, Operand<T> rhs,
       TriangularSolve.Options... options) {
     return TriangularSolve.create(scope, matrix, rhs, options);
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/MathOps.java b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/MathOps.java
index 1f08502ca44..3c983272e6d 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/MathOps.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/MathOps.java
@@ -19,6 +19,7 @@
 
 import org.tensorflow.DataType;
 import org.tensorflow.Operand;
+import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.math.Abs;
 import org.tensorflow.op.math.AccumulateN;
@@ -35,6 +36,8 @@
 import org.tensorflow.op.math.Atan;
 import org.tensorflow.op.math.Atan2;
 import org.tensorflow.op.math.Atanh;
+import org.tensorflow.op.math.BesselI0e;
+import org.tensorflow.op.math.BesselI1e;
 import org.tensorflow.op.math.Betainc;
 import org.tensorflow.op.math.Bincount;
 import org.tensorflow.op.math.Ceil;
@@ -45,7 +48,6 @@
 import org.tensorflow.op.math.Cosh;
 import org.tensorflow.op.math.Cumprod;
 import org.tensorflow.op.math.Cumsum;
-import org.tensorflow.op.math.DenseBincount;
 import org.tensorflow.op.math.Digamma;
 import org.tensorflow.op.math.Div;
 import org.tensorflow.op.math.DivNoNan;
@@ -128,7 +130,6 @@
 import org.tensorflow.types.TInt32;
 import org.tensorflow.types.TInt64;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * An API for building {@code math} operations as {@link Op Op}s
@@ -153,7 +154,7 @@ public final class MathOps {
    * @param x
    * @return a new instance of Abs
    */
-  public <T extends TNumber> Abs<T> abs(Operand<T> x) {
+  public <T extends Tensor & TNumber> Abs<T> abs(Operand<T> x) {
     return Abs.create(scope, x);
   }
 
@@ -174,7 +175,7 @@ public <T extends TNumber> Abs<T> abs(Operand<T> x) {
    * @param shape Shape of elements of `inputs`.
    * @return a new instance of AccumulateN
    */
-  public <T extends TType> AccumulateN<T> accumulateN(Iterable<Operand<T>> inputs, Shape shape) {
+  public <T extends Tensor> AccumulateN<T> accumulateN(Iterable<Operand<T>> inputs, Shape shape) {
     return AccumulateN.create(scope, inputs, shape);
   }
 
@@ -185,7 +186,7 @@ public <T extends TType> AccumulateN<T> accumulateN(Iterable<Operand<T>> inputs,
    * @param x
    * @return a new instance of Acos
    */
-  public <T extends TType> Acos<T> acos(Operand<T> x) {
+  public <T extends Tensor> Acos<T> acos(Operand<T> x) {
     return Acos.create(scope, x);
   }
 
@@ -203,7 +204,7 @@ public <T extends TType> Acos<T> acos(Operand<T> x) {
    * @param x
    * @return a new instance of Acosh
    */
-  public <T extends TType> Acosh<T> acosh(Operand<T> x) {
+  public <T extends Tensor> Acosh<T> acosh(Operand<T> x) {
     return Acosh.create(scope, x);
   }
 
@@ -218,7 +219,7 @@ public <T extends TType> Acosh<T> acosh(Operand<T> x) {
    * @param y
    * @return a new instance of Add
    */
-  public <T extends TType> Add<T> add(Operand<T> x, Operand<T> y) {
+  public <T extends Tensor> Add<T> add(Operand<T> x, Operand<T> y) {
     return Add.create(scope, x, y);
   }
 
@@ -236,7 +237,7 @@ public <T extends TType> Add<T> add(Operand<T> x, Operand<T> y) {
    * @param inputs
    * @return a new instance of AddN
    */
-  public <T extends TType> AddN<T> addN(Iterable<Operand<T>> inputs) {
+  public <T extends Tensor> AddN<T> addN(Iterable<Operand<T>> inputs) {
     return AddN.create(scope, inputs);
   }
 
@@ -262,7 +263,7 @@ public <T extends TType> AddN<T> addN(Iterable<Operand<T>> inputs) {
    * @param input
    * @return a new instance of Angle
    */
-  public <T extends TType> Angle<TFloat32> angle(Operand<T> input) {
+  public <T extends Tensor> Angle<TFloat32> angle(Operand<T> input) {
     return Angle.create(scope, input);
   }
 
@@ -289,7 +290,8 @@ public <T extends TType> Angle<TFloat32> angle(Operand<T> input) {
    * @param Tout
    * @return a new instance of Angle
    */
-  public <U extends TNumber, T extends TType> Angle<U> angle(Operand<T> input, DataType<U> Tout) {
+  public <U extends Tensor & TNumber, T extends Tensor> Angle<U> angle(Operand<T> input,
+      DataType<U> Tout) {
     return Angle.create(scope, input, Tout);
   }
 
@@ -301,7 +303,7 @@ public <U extends TNumber, T extends TType> Angle<U> angle(Operand<T> input, Dat
    * @param options carries optional attributes values
    * @return a new instance of ApproximateEqual
    */
-  public <T extends TType> ApproximateEqual approximateEqual(Operand<T> x, Operand<T> y,
+  public <T extends Tensor> ApproximateEqual approximateEqual(Operand<T> x, Operand<T> y,
       ApproximateEqual.Options... options) {
     return ApproximateEqual.create(scope, x, y, options);
   }
@@ -328,7 +330,7 @@ public <T extends TType> ApproximateEqual approximateEqual(Operand<T> x, Operand
    *  use dimension = 0.
    * @return a new instance of ArgMax
    */
-  public <T extends TType, U extends TNumber> ArgMax<TInt64> argMax(Operand<T> input,
+  public <T extends Tensor, U extends Tensor & TNumber> ArgMax<TInt64> argMax(Operand<T> input,
       Operand<U> dimension) {
     return ArgMax.create(scope, input, dimension);
   }
@@ -356,8 +358,8 @@ public <T extends TType, U extends TNumber> ArgMax<TInt64> argMax(Operand<T> inp
    * @param outputType
    * @return a new instance of ArgMax
    */
-  public <V extends TNumber, T extends TType, U extends TNumber> ArgMax<V> argMax(Operand<T> input,
-      Operand<U> dimension, DataType<V> outputType) {
+  public <V extends Tensor & TNumber, T extends Tensor, U extends Tensor & TNumber> ArgMax<V> argMax(
+      Operand<T> input, Operand<U> dimension, DataType<V> outputType) {
     return ArgMax.create(scope, input, dimension, outputType);
   }
 
@@ -383,7 +385,7 @@ public <V extends TNumber, T extends TType, U extends TNumber> ArgMax<V> argMax(
    *  use dimension = 0.
    * @return a new instance of ArgMin
    */
-  public <T extends TType, U extends TNumber> ArgMin<TInt64> argMin(Operand<T> input,
+  public <T extends Tensor, U extends Tensor & TNumber> ArgMin<TInt64> argMin(Operand<T> input,
       Operand<U> dimension) {
     return ArgMin.create(scope, input, dimension);
   }
@@ -411,8 +413,8 @@ public <T extends TType, U extends TNumber> ArgMin<TInt64> argMin(Operand<T> inp
    * @param outputType
    * @return a new instance of ArgMin
    */
-  public <V extends TNumber, T extends TType, U extends TNumber> ArgMin<V> argMin(Operand<T> input,
-      Operand<U> dimension, DataType<V> outputType) {
+  public <V extends Tensor & TNumber, T extends Tensor, U extends Tensor & TNumber> ArgMin<V> argMin(
+      Operand<T> input, Operand<U> dimension, DataType<V> outputType) {
     return ArgMin.create(scope, input, dimension, outputType);
   }
 
@@ -438,7 +440,7 @@ public <V extends TNumber, T extends TType, U extends TNumber> ArgMin<V> argMin(
    * @param x
    * @return a new instance of Asin
    */
-  public <T extends TType> Asin<T> asin(Operand<T> x) {
+  public <T extends Tensor> Asin<T> asin(Operand<T> x) {
     return Asin.create(scope, x);
   }
 
@@ -458,7 +460,7 @@ public <T extends TType> Asin<T> asin(Operand<T> x) {
    * @param x
    * @return a new instance of Asinh
    */
-  public <T extends TType> Asinh<T> asinh(Operand<T> x) {
+  public <T extends Tensor> Asinh<T> asinh(Operand<T> x) {
     return Asinh.create(scope, x);
   }
 
@@ -484,7 +486,7 @@ public <T extends TType> Asinh<T> asinh(Operand<T> x) {
    * @param x
    * @return a new instance of Atan
    */
-  public <T extends TType> Atan<T> atan(Operand<T> x) {
+  public <T extends Tensor> Atan<T> atan(Operand<T> x) {
     return Atan.create(scope, x);
   }
 
@@ -502,7 +504,7 @@ public <T extends TType> Atan<T> atan(Operand<T> x) {
    * @param x
    * @return a new instance of Atan2
    */
-  public <T extends TNumber> Atan2<T> atan2(Operand<T> y, Operand<T> x) {
+  public <T extends Tensor & TNumber> Atan2<T> atan2(Operand<T> y, Operand<T> x) {
     return Atan2.create(scope, y, x);
   }
 
@@ -524,10 +526,42 @@ public <T extends TNumber> Atan2<T> atan2(Operand<T> y, Operand<T> x) {
    * @param x
    * @return a new instance of Atanh
    */
-  public <T extends TType> Atanh<T> atanh(Operand<T> x) {
+  public <T extends Tensor> Atanh<T> atanh(Operand<T> x) {
     return Atanh.create(scope, x);
   }
 
+  /**
+   * Computes the Bessel i0e function of `x` element-wise.
+   *  <p>
+   *  Exponentially scaled modified Bessel function of order 0 defined as
+   *  `bessel_i0e(x) = exp(-abs(x)) bessel_i0(x)`.
+   *  <p>
+   *  This function is faster and numerically stabler than `bessel_i0(x)`.
+   *
+   * @param <T> data type for {@code y()} output
+   * @param x
+   * @return a new instance of BesselI0e
+   */
+  public <T extends Tensor & TNumber> BesselI0e<T> besselI0e(Operand<T> x) {
+    return BesselI0e.create(scope, x);
+  }
+
+  /**
+   * Computes the Bessel i1e function of `x` element-wise.
+   *  <p>
+   *  Exponentially scaled modified Bessel function of order 0 defined as
+   *  `bessel_i1e(x) = exp(-abs(x)) bessel_i1(x)`.
+   *  <p>
+   *  This function is faster and numerically stabler than `bessel_i1(x)`.
+   *
+   * @param <T> data type for {@code y()} output
+   * @param x
+   * @return a new instance of BesselI1e
+   */
+  public <T extends Tensor & TNumber> BesselI1e<T> besselI1e(Operand<T> x) {
+    return BesselI1e.create(scope, x);
+  }
+
   /**
    * Compute the regularized incomplete beta integral \\(I_x(a, b)\\).
    *  <p>
@@ -548,7 +582,7 @@ public <T extends TType> Atanh<T> atanh(Operand<T> x) {
    * @param x
    * @return a new instance of Betainc
    */
-  public <T extends TNumber> Betainc<T> betainc(Operand<T> a, Operand<T> b, Operand<T> x) {
+  public <T extends Tensor & TNumber> Betainc<T> betainc(Operand<T> a, Operand<T> b, Operand<T> x) {
     return Betainc.create(scope, a, b, x);
   }
 
@@ -571,8 +605,8 @@ public <T extends TNumber> Betainc<T> betainc(Operand<T> a, Operand<T> b, Operan
    *  equal to 1.
    * @return a new instance of Bincount
    */
-  public <T extends TNumber> Bincount<T> bincount(Operand<TInt32> arr, Operand<TInt32> size,
-      Operand<T> weights) {
+  public <T extends Tensor & TNumber> Bincount<T> bincount(Operand<TInt32> arr,
+      Operand<TInt32> size, Operand<T> weights) {
     return Bincount.create(scope, arr, size, weights);
   }
 
@@ -583,7 +617,7 @@ public <T extends TNumber> Bincount<T> bincount(Operand<TInt32> arr, Operand<TIn
    * @param x
    * @return a new instance of Ceil
    */
-  public <T extends TNumber> Ceil<T> ceil(Operand<T> x) {
+  public <T extends Tensor & TNumber> Ceil<T> ceil(Operand<T> x) {
     return Ceil.create(scope, x);
   }
 
@@ -616,7 +650,7 @@ public <T extends TNumber> Ceil<T> ceil(Operand<T> x) {
    * @param threshold Threshold to compare against.
    * @return a new instance of CompareAndBitpack
    */
-  public <T extends TType> CompareAndBitpack compareAndBitpack(Operand<T> input,
+  public <T extends Tensor> CompareAndBitpack compareAndBitpack(Operand<T> input,
       Operand<T> threshold) {
     return CompareAndBitpack.create(scope, input, threshold);
   }
@@ -633,7 +667,7 @@ public <T extends TType> CompareAndBitpack compareAndBitpack(Operand<T> input,
    * @param x
    * @return a new instance of ComplexAbs
    */
-  public <T extends TType> ComplexAbs<TFloat32> complexAbs(Operand<T> x) {
+  public <T extends Tensor> ComplexAbs<TFloat32> complexAbs(Operand<T> x) {
     return ComplexAbs.create(scope, x);
   }
 
@@ -650,7 +684,7 @@ public <T extends TType> ComplexAbs<TFloat32> complexAbs(Operand<T> x) {
    * @param Tout
    * @return a new instance of ComplexAbs
    */
-  public <U extends TNumber, T extends TType> ComplexAbs<U> complexAbs(Operand<T> x,
+  public <U extends Tensor & TNumber, T extends Tensor> ComplexAbs<U> complexAbs(Operand<T> x,
       DataType<U> Tout) {
     return ComplexAbs.create(scope, x, Tout);
   }
@@ -675,7 +709,7 @@ public <U extends TNumber, T extends TType> ComplexAbs<U> complexAbs(Operand<T>
    * @param input
    * @return a new instance of Conj
    */
-  public <T extends TType> Conj<T> conj(Operand<T> input) {
+  public <T extends Tensor> Conj<T> conj(Operand<T> input) {
     return Conj.create(scope, input);
   }
 
@@ -696,7 +730,7 @@ public <T extends TType> Conj<T> conj(Operand<T> input) {
    * @param x
    * @return a new instance of Cos
    */
-  public <T extends TType> Cos<T> cos(Operand<T> x) {
+  public <T extends Tensor> Cos<T> cos(Operand<T> x) {
     return Cos.create(scope, x);
   }
 
@@ -716,7 +750,7 @@ public <T extends TType> Cos<T> cos(Operand<T> x) {
    * @param x
    * @return a new instance of Cosh
    */
-  public <T extends TType> Cosh<T> cosh(Operand<T> x) {
+  public <T extends Tensor> Cosh<T> cosh(Operand<T> x) {
     return Cosh.create(scope, x);
   }
 
@@ -754,8 +788,8 @@ public <T extends TType> Cosh<T> cosh(Operand<T> x) {
    * @param options carries optional attributes values
    * @return a new instance of Cumprod
    */
-  public <T extends TType, U extends TNumber> Cumprod<T> cumprod(Operand<T> x, Operand<U> axis,
-      Cumprod.Options... options) {
+  public <T extends Tensor, U extends Tensor & TNumber> Cumprod<T> cumprod(Operand<T> x,
+      Operand<U> axis, Cumprod.Options... options) {
     return Cumprod.create(scope, x, axis, options);
   }
 
@@ -793,36 +827,11 @@ public <T extends TType, U extends TNumber> Cumprod<T> cumprod(Operand<T> x, Ope
    * @param options carries optional attributes values
    * @return a new instance of Cumsum
    */
-  public <T extends TType, U extends TNumber> Cumsum<T> cumsum(Operand<T> x, Operand<U> axis,
-      Cumsum.Options... options) {
+  public <T extends Tensor, U extends Tensor & TNumber> Cumsum<T> cumsum(Operand<T> x,
+      Operand<U> axis, Cumsum.Options... options) {
     return Cumsum.create(scope, x, axis, options);
   }
 
-  /**
-   * Counts the number of occurrences of each value in an integer array.
-   *  <p>
-   *  Outputs a vector with length `size` and the same dtype as `weights`. If
-   *  `weights` are empty, then index `i` stores the number of times the value `i` is
-   *  counted in `arr`. If `weights` are non-empty, then index `i` stores the sum of
-   *  the value in `weights` at each index where the corresponding value in `arr` is
-   *  `i`.
-   *  <p>
-   *  Values in `arr` outside of the range [0, size) are ignored.
-   *
-   * @param <U> data type for {@code output()} output
-   * @param input 1D or 2D int `Tensor`.
-   * @param size non-negative int scalar `Tensor`.
-   * @param weights is an int32, int64, float32, or float64 `Tensor` with the same
-   *  shape as `arr`, or a length-0 `Tensor`, in which case it acts as all weights
-   *  equal to 1.
-   * @param options carries optional attributes values
-   * @return a new instance of DenseBincount
-   */
-  public <U extends TNumber, T extends TNumber> DenseBincount<U> denseBincount(Operand<T> input,
-      Operand<T> size, Operand<U> weights, DenseBincount.Options... options) {
-    return DenseBincount.create(scope, input, size, weights, options);
-  }
-
   /**
    * Computes Psi, the derivative of Lgamma (the log of the absolute value of
    *  <p>
@@ -832,7 +841,7 @@ public <U extends TNumber, T extends TNumber> DenseBincount<U> denseBincount(Ope
    * @param x
    * @return a new instance of Digamma
    */
-  public <T extends TNumber> Digamma<T> digamma(Operand<T> x) {
+  public <T extends Tensor & TNumber> Digamma<T> digamma(Operand<T> x) {
     return Digamma.create(scope, x);
   }
 
@@ -847,7 +856,7 @@ public <T extends TNumber> Digamma<T> digamma(Operand<T> x) {
    * @param y
    * @return a new instance of Div
    */
-  public <T extends TType> Div<T> div(Operand<T> x, Operand<T> y) {
+  public <T extends Tensor> Div<T> div(Operand<T> x, Operand<T> y) {
     return Div.create(scope, x, y);
   }
 
@@ -863,7 +872,7 @@ public <T extends TType> Div<T> div(Operand<T> x, Operand<T> y) {
    * @param y
    * @return a new instance of DivNoNan
    */
-  public <T extends TType> DivNoNan<T> divNoNan(Operand<T> x, Operand<T> y) {
+  public <T extends Tensor> DivNoNan<T> divNoNan(Operand<T> x, Operand<T> y) {
     return DivNoNan.create(scope, x, y);
   }
 
@@ -887,7 +896,7 @@ public <T extends TType> DivNoNan<T> divNoNan(Operand<T> x, Operand<T> y) {
    * @param options carries optional attributes values
    * @return a new instance of Equal
    */
-  public <T extends TType> Equal equal(Operand<T> x, Operand<T> y, Equal.Options... options) {
+  public <T extends Tensor> Equal equal(Operand<T> x, Operand<T> y, Equal.Options... options) {
     return Equal.create(scope, x, y, options);
   }
 
@@ -898,7 +907,7 @@ public <T extends TType> Equal equal(Operand<T> x, Operand<T> y, Equal.Options..
    * @param x
    * @return a new instance of Erf
    */
-  public <T extends TNumber> Erf<T> erf(Operand<T> x) {
+  public <T extends Tensor & TNumber> Erf<T> erf(Operand<T> x) {
     return Erf.create(scope, x);
   }
 
@@ -909,7 +918,7 @@ public <T extends TNumber> Erf<T> erf(Operand<T> x) {
    * @param x
    * @return a new instance of Erfc
    */
-  public <T extends TNumber> Erfc<T> erfc(Operand<T> x) {
+  public <T extends Tensor & TNumber> Erfc<T> erfc(Operand<T> x) {
     return Erfc.create(scope, x);
   }
 
@@ -919,7 +928,7 @@ public <T extends TNumber> Erfc<T> erfc(Operand<T> x) {
    * @param x
    * @return a new instance of erfinv
    */
-  public <T extends TNumber> erfinv<T> erfinv(Operand<T> x) {
+  public <T extends Tensor & TNumber> erfinv<T> erfinv(Operand<T> x) {
     return erfinv.create(scope, x);
   }
 
@@ -955,7 +964,7 @@ public <T extends TNumber> erfinv<T> erfinv(Operand<T> x) {
    * @param x
    * @return a new instance of Exp
    */
-  public <T extends TType> Exp<T> exp(Operand<T> x) {
+  public <T extends Tensor> Exp<T> exp(Operand<T> x) {
     return Exp.create(scope, x);
   }
 
@@ -980,7 +989,7 @@ public <T extends TType> Exp<T> exp(Operand<T> x) {
    * @param x
    * @return a new instance of Expm1
    */
-  public <T extends TType> Expm1<T> expm1(Operand<T> x) {
+  public <T extends Tensor> Expm1<T> expm1(Operand<T> x) {
     return Expm1.create(scope, x);
   }
 
@@ -1000,7 +1009,7 @@ public Fact fact() {
    * @param x
    * @return a new instance of Floor
    */
-  public <T extends TNumber> Floor<T> floor(Operand<T> x) {
+  public <T extends Tensor & TNumber> Floor<T> floor(Operand<T> x) {
     return Floor.create(scope, x);
   }
 
@@ -1015,7 +1024,7 @@ public <T extends TNumber> Floor<T> floor(Operand<T> x) {
    * @param y
    * @return a new instance of FloorDiv
    */
-  public <T extends TType> FloorDiv<T> floorDiv(Operand<T> x, Operand<T> y) {
+  public <T extends Tensor> FloorDiv<T> floorDiv(Operand<T> x, Operand<T> y) {
     return FloorDiv.create(scope, x, y);
   }
 
@@ -1033,7 +1042,7 @@ public <T extends TType> FloorDiv<T> floorDiv(Operand<T> x, Operand<T> y) {
    * @param y
    * @return a new instance of FloorMod
    */
-  public <T extends TNumber> FloorMod<T> floorMod(Operand<T> x, Operand<T> y) {
+  public <T extends Tensor & TNumber> FloorMod<T> floorMod(Operand<T> x, Operand<T> y) {
     return FloorMod.create(scope, x, y);
   }
 
@@ -1058,7 +1067,7 @@ public <T extends TNumber> FloorMod<T> floorMod(Operand<T> x, Operand<T> y) {
    * @param y
    * @return a new instance of Greater
    */
-  public <T extends TNumber> Greater greater(Operand<T> x, Operand<T> y) {
+  public <T extends Tensor & TNumber> Greater greater(Operand<T> x, Operand<T> y) {
     return Greater.create(scope, x, y);
   }
 
@@ -1083,7 +1092,7 @@ public <T extends TNumber> Greater greater(Operand<T> x, Operand<T> y) {
    * @param y
    * @return a new instance of GreaterEqual
    */
-  public <T extends TNumber> GreaterEqual greaterEqual(Operand<T> x, Operand<T> y) {
+  public <T extends Tensor & TNumber> GreaterEqual greaterEqual(Operand<T> x, Operand<T> y) {
     return GreaterEqual.create(scope, x, y);
   }
 
@@ -1108,7 +1117,7 @@ public <T extends TNumber> GreaterEqual greaterEqual(Operand<T> x, Operand<T> y)
    * @param x
    * @return a new instance of Igamma
    */
-  public <T extends TNumber> Igamma<T> igamma(Operand<T> a, Operand<T> x) {
+  public <T extends Tensor & TNumber> Igamma<T> igamma(Operand<T> a, Operand<T> x) {
     return Igamma.create(scope, a, x);
   }
 
@@ -1133,7 +1142,7 @@ public <T extends TNumber> Igamma<T> igamma(Operand<T> a, Operand<T> x) {
    * @param x
    * @return a new instance of Igammac
    */
-  public <T extends TNumber> Igammac<T> igammac(Operand<T> a, Operand<T> x) {
+  public <T extends Tensor & TNumber> Igammac<T> igammac(Operand<T> a, Operand<T> x) {
     return Igammac.create(scope, a, x);
   }
 
@@ -1155,7 +1164,7 @@ public <T extends TNumber> Igammac<T> igammac(Operand<T> a, Operand<T> x) {
    * @param input
    * @return a new instance of Imag
    */
-  public <T extends TType> Imag<TFloat32> imag(Operand<T> input) {
+  public <T extends Tensor> Imag<TFloat32> imag(Operand<T> input) {
     return Imag.create(scope, input);
   }
 
@@ -1178,7 +1187,8 @@ public <T extends TType> Imag<TFloat32> imag(Operand<T> input) {
    * @param Tout
    * @return a new instance of Imag
    */
-  public <U extends TNumber, T extends TType> Imag<U> imag(Operand<T> input, DataType<U> Tout) {
+  public <U extends Tensor & TNumber, T extends Tensor> Imag<U> imag(Operand<T> input,
+      DataType<U> Tout) {
     return Imag.create(scope, input, Tout);
   }
 
@@ -1204,7 +1214,7 @@ public <U extends TNumber, T extends TType> Imag<U> imag(Operand<T> input, DataT
    * @param x 1-D.
    * @return a new instance of InvertPermutation
    */
-  public <T extends TNumber> InvertPermutation<T> invertPermutation(Operand<T> x) {
+  public <T extends Tensor & TNumber> InvertPermutation<T> invertPermutation(Operand<T> x) {
     return InvertPermutation.create(scope, x);
   }
 
@@ -1222,7 +1232,7 @@ public <T extends TNumber> InvertPermutation<T> invertPermutation(Operand<T> x)
    * @param x
    * @return a new instance of IsFinite
    */
-  public <T extends TNumber> IsFinite isFinite(Operand<T> x) {
+  public <T extends Tensor & TNumber> IsFinite isFinite(Operand<T> x) {
     return IsFinite.create(scope, x);
   }
 
@@ -1240,7 +1250,7 @@ public <T extends TNumber> IsFinite isFinite(Operand<T> x) {
    * @param x
    * @return a new instance of IsInf
    */
-  public <T extends TNumber> IsInf isInf(Operand<T> x) {
+  public <T extends Tensor & TNumber> IsInf isInf(Operand<T> x) {
     return IsInf.create(scope, x);
   }
 
@@ -1258,7 +1268,7 @@ public <T extends TNumber> IsInf isInf(Operand<T> x) {
    * @param x
    * @return a new instance of IsNan
    */
-  public <T extends TNumber> IsNan isNan(Operand<T> x) {
+  public <T extends Tensor & TNumber> IsNan isNan(Operand<T> x) {
     return IsNan.create(scope, x);
   }
 
@@ -1283,7 +1293,7 @@ public <T extends TNumber> IsNan isNan(Operand<T> x) {
    * @param y
    * @return a new instance of Less
    */
-  public <T extends TNumber> Less less(Operand<T> x, Operand<T> y) {
+  public <T extends Tensor & TNumber> Less less(Operand<T> x, Operand<T> y) {
     return Less.create(scope, x, y);
   }
 
@@ -1308,7 +1318,7 @@ public <T extends TNumber> Less less(Operand<T> x, Operand<T> y) {
    * @param y
    * @return a new instance of LessEqual
    */
-  public <T extends TNumber> LessEqual lessEqual(Operand<T> x, Operand<T> y) {
+  public <T extends Tensor & TNumber> LessEqual lessEqual(Operand<T> x, Operand<T> y) {
     return LessEqual.create(scope, x, y);
   }
 
@@ -1328,7 +1338,7 @@ public <T extends TNumber> LessEqual lessEqual(Operand<T> x, Operand<T> y) {
    * @param x
    * @return a new instance of Lgamma
    */
-  public <T extends TNumber> Lgamma<T> lgamma(Operand<T> x) {
+  public <T extends Tensor & TNumber> Lgamma<T> lgamma(Operand<T> x) {
     return Lgamma.create(scope, x);
   }
 
@@ -1347,7 +1357,7 @@ public <T extends TNumber> Lgamma<T> lgamma(Operand<T> x) {
    * @param x
    * @return a new instance of Log
    */
-  public <T extends TType> Log<T> log(Operand<T> x) {
+  public <T extends Tensor> Log<T> log(Operand<T> x) {
     return Log.create(scope, x);
   }
 
@@ -1366,7 +1376,7 @@ public <T extends TType> Log<T> log(Operand<T> x) {
    * @param x
    * @return a new instance of Log1p
    */
-  public <T extends TType> Log1p<T> log1p(Operand<T> x) {
+  public <T extends Tensor> Log1p<T> log1p(Operand<T> x) {
     return Log1p.create(scope, x);
   }
 
@@ -1419,7 +1429,7 @@ public LogicalOr logicalOr(Operand<TBool> x, Operand<TBool> y) {
    * @param y
    * @return a new instance of Maximum
    */
-  public <T extends TNumber> Maximum<T> maximum(Operand<T> x, Operand<T> y) {
+  public <T extends Tensor & TNumber> Maximum<T> maximum(Operand<T> x, Operand<T> y) {
     return Maximum.create(scope, x, y);
   }
 
@@ -1438,8 +1448,8 @@ public <T extends TNumber> Maximum<T> maximum(Operand<T> x, Operand<T> y) {
    * @param options carries optional attributes values
    * @return a new instance of Mean
    */
-  public <T extends TType, U extends TNumber> Mean<T> mean(Operand<T> input, Operand<U> axis,
-      Mean.Options... options) {
+  public <T extends Tensor, U extends Tensor & TNumber> Mean<T> mean(Operand<T> input,
+      Operand<U> axis, Mean.Options... options) {
     return Mean.create(scope, input, axis, options);
   }
 
@@ -1454,7 +1464,7 @@ public <T extends TType, U extends TNumber> Mean<T> mean(Operand<T> input, Opera
    * @param y
    * @return a new instance of Minimum
    */
-  public <T extends TNumber> Minimum<T> minimum(Operand<T> x, Operand<T> y) {
+  public <T extends Tensor & TNumber> Minimum<T> minimum(Operand<T> x, Operand<T> y) {
     return Minimum.create(scope, x, y);
   }
 
@@ -1472,7 +1482,7 @@ public <T extends TNumber> Minimum<T> minimum(Operand<T> x, Operand<T> y) {
    * @param y
    * @return a new instance of Mod
    */
-  public <T extends TNumber> Mod<T> mod(Operand<T> x, Operand<T> y) {
+  public <T extends Tensor & TNumber> Mod<T> mod(Operand<T> x, Operand<T> y) {
     return Mod.create(scope, x, y);
   }
 
@@ -1487,7 +1497,7 @@ public <T extends TNumber> Mod<T> mod(Operand<T> x, Operand<T> y) {
    * @param y
    * @return a new instance of Mul
    */
-  public <T extends TType> Mul<T> mul(Operand<T> x, Operand<T> y) {
+  public <T extends Tensor> Mul<T> mul(Operand<T> x, Operand<T> y) {
     return Mul.create(scope, x, y);
   }
 
@@ -1502,7 +1512,7 @@ public <T extends TType> Mul<T> mul(Operand<T> x, Operand<T> y) {
    * @param y
    * @return a new instance of MulNoNan
    */
-  public <T extends TType> MulNoNan<T> mulNoNan(Operand<T> x, Operand<T> y) {
+  public <T extends Tensor> MulNoNan<T> mulNoNan(Operand<T> x, Operand<T> y) {
     return MulNoNan.create(scope, x, y);
   }
 
@@ -1512,7 +1522,7 @@ public <T extends TType> MulNoNan<T> mulNoNan(Operand<T> x, Operand<T> y) {
    * @param x
    * @return a new instance of Ndtri
    */
-  public <T extends TNumber> Ndtri<T> ndtri(Operand<T> x) {
+  public <T extends Tensor & TNumber> Ndtri<T> ndtri(Operand<T> x) {
     return Ndtri.create(scope, x);
   }
 
@@ -1525,7 +1535,7 @@ public <T extends TNumber> Ndtri<T> ndtri(Operand<T> x) {
    * @param x
    * @return a new instance of Neg
    */
-  public <T extends TType> Neg<T> neg(Operand<T> x) {
+  public <T extends Tensor> Neg<T> neg(Operand<T> x) {
     return Neg.create(scope, x);
   }
 
@@ -1544,7 +1554,7 @@ public <T extends TType> Neg<T> neg(Operand<T> x) {
    * @param x2
    * @return a new instance of NextAfter
    */
-  public <T extends TNumber> NextAfter<T> nextAfter(Operand<T> x1, Operand<T> x2) {
+  public <T extends Tensor & TNumber> NextAfter<T> nextAfter(Operand<T> x1, Operand<T> x2) {
     return NextAfter.create(scope, x1, x2);
   }
 
@@ -1559,7 +1569,7 @@ public <T extends TNumber> NextAfter<T> nextAfter(Operand<T> x1, Operand<T> x2)
    * @param options carries optional attributes values
    * @return a new instance of NotEqual
    */
-  public <T extends TType> NotEqual notEqual(Operand<T> x, Operand<T> y,
+  public <T extends Tensor> NotEqual notEqual(Operand<T> x, Operand<T> y,
       NotEqual.Options... options) {
     return NotEqual.create(scope, x, y, options);
   }
@@ -1579,7 +1589,7 @@ public <T extends TType> NotEqual notEqual(Operand<T> x, Operand<T> y,
    * @param x
    * @return a new instance of Polygamma
    */
-  public <T extends TNumber> Polygamma<T> polygamma(Operand<T> a, Operand<T> x) {
+  public <T extends Tensor & TNumber> Polygamma<T> polygamma(Operand<T> a, Operand<T> x) {
     return Polygamma.create(scope, a, x);
   }
 
@@ -1596,7 +1606,7 @@ public <T extends TNumber> Polygamma<T> polygamma(Operand<T> a, Operand<T> x) {
    * @param x
    * @return a new instance of PopulationCount
    */
-  public <T extends TNumber> PopulationCount populationCount(Operand<T> x) {
+  public <T extends Tensor & TNumber> PopulationCount populationCount(Operand<T> x) {
     return PopulationCount.create(scope, x);
   }
 
@@ -1616,7 +1626,7 @@ public <T extends TNumber> PopulationCount populationCount(Operand<T> x) {
    * @param y
    * @return a new instance of Pow
    */
-  public <T extends TType> Pow<T> pow(Operand<T> x, Operand<T> y) {
+  public <T extends Tensor> Pow<T> pow(Operand<T> x, Operand<T> y) {
     return Pow.create(scope, x, y);
   }
 
@@ -1633,7 +1643,7 @@ public <T extends TType> Pow<T> pow(Operand<T> x, Operand<T> y) {
    * @param Toutput
    * @return a new instance of QuantizedAdd
    */
-  public <V extends TType, T extends TType, U extends TType> QuantizedAdd<V> quantizedAdd(
+  public <V extends Tensor, T extends Tensor, U extends Tensor> QuantizedAdd<V> quantizedAdd(
       Operand<T> x, Operand<U> y, Operand<TFloat32> minX, Operand<TFloat32> maxX,
       Operand<TFloat32> minY, Operand<TFloat32> maxY, DataType<V> Toutput) {
     return QuantizedAdd.create(scope, x, y, minX, maxX, minY, maxY, Toutput);
@@ -1652,7 +1662,7 @@ public <V extends TType, T extends TType, U extends TType> QuantizedAdd<V> quant
    * @param Toutput
    * @return a new instance of QuantizedMul
    */
-  public <V extends TType, T extends TType, U extends TType> QuantizedMul<V> quantizedMul(
+  public <V extends Tensor, T extends Tensor, U extends Tensor> QuantizedMul<V> quantizedMul(
       Operand<T> x, Operand<U> y, Operand<TFloat32> minX, Operand<TFloat32> maxX,
       Operand<TFloat32> minY, Operand<TFloat32> maxY, DataType<V> Toutput) {
     return QuantizedMul.create(scope, x, y, minX, maxX, minY, maxY, Toutput);
@@ -1676,7 +1686,7 @@ public <V extends TType, T extends TType, U extends TType> QuantizedMul<V> quant
    * @param input
    * @return a new instance of Real
    */
-  public <T extends TType> Real<TFloat32> real(Operand<T> input) {
+  public <T extends Tensor> Real<TFloat32> real(Operand<T> input) {
     return Real.create(scope, input);
   }
 
@@ -1699,7 +1709,8 @@ public <T extends TType> Real<TFloat32> real(Operand<T> input) {
    * @param Tout
    * @return a new instance of Real
    */
-  public <U extends TNumber, T extends TType> Real<U> real(Operand<T> input, DataType<U> Tout) {
+  public <U extends Tensor & TNumber, T extends Tensor> Real<U> real(Operand<T> input,
+      DataType<U> Tout) {
     return Real.create(scope, input, Tout);
   }
 
@@ -1716,7 +1727,7 @@ public <U extends TNumber, T extends TType> Real<U> real(Operand<T> input, DataT
    * @param y
    * @return a new instance of RealDiv
    */
-  public <T extends TType> RealDiv<T> realDiv(Operand<T> x, Operand<T> y) {
+  public <T extends Tensor> RealDiv<T> realDiv(Operand<T> x, Operand<T> y) {
     return RealDiv.create(scope, x, y);
   }
 
@@ -1729,7 +1740,7 @@ public <T extends TType> RealDiv<T> realDiv(Operand<T> x, Operand<T> y) {
    * @param x
    * @return a new instance of Reciprocal
    */
-  public <T extends TType> Reciprocal<T> reciprocal(Operand<T> x) {
+  public <T extends Tensor> Reciprocal<T> reciprocal(Operand<T> x) {
     return Reciprocal.create(scope, x);
   }
 
@@ -1749,7 +1760,7 @@ public <T extends TType> Reciprocal<T> reciprocal(Operand<T> x) {
    * @param x
    * @return a new instance of Rint
    */
-  public <T extends TNumber> Rint<T> rint(Operand<T> x) {
+  public <T extends Tensor & TNumber> Rint<T> rint(Operand<T> x) {
     return Rint.create(scope, x);
   }
 
@@ -1763,7 +1774,7 @@ public <T extends TNumber> Rint<T> rint(Operand<T> x) {
    * @param x
    * @return a new instance of Round
    */
-  public <T extends TType> Round<T> round(Operand<T> x) {
+  public <T extends Tensor> Round<T> round(Operand<T> x) {
     return Round.create(scope, x);
   }
 
@@ -1776,7 +1787,7 @@ public <T extends TType> Round<T> round(Operand<T> x) {
    * @param x
    * @return a new instance of Rsqrt
    */
-  public <T extends TType> Rsqrt<T> rsqrt(Operand<T> x) {
+  public <T extends Tensor> Rsqrt<T> rsqrt(Operand<T> x) {
     return Rsqrt.create(scope, x);
   }
 
@@ -1811,8 +1822,8 @@ public <T extends TType> Rsqrt<T> rsqrt(Operand<T> x) {
    *  first dimension.  Values should be sorted and can be repeated.
    * @return a new instance of SegmentMax
    */
-  public <T extends TNumber, U extends TNumber> SegmentMax<T> segmentMax(Operand<T> data,
-      Operand<U> segmentIds) {
+  public <T extends Tensor & TNumber, U extends Tensor & TNumber> SegmentMax<T> segmentMax(
+      Operand<T> data, Operand<U> segmentIds) {
     return SegmentMax.create(scope, data, segmentIds);
   }
 
@@ -1848,7 +1859,7 @@ public <T extends TNumber, U extends TNumber> SegmentMax<T> segmentMax(Operand<T
    *  first dimension.  Values should be sorted and can be repeated.
    * @return a new instance of SegmentMean
    */
-  public <T extends TType, U extends TNumber> SegmentMean<T> segmentMean(Operand<T> data,
+  public <T extends Tensor, U extends Tensor & TNumber> SegmentMean<T> segmentMean(Operand<T> data,
       Operand<U> segmentIds) {
     return SegmentMean.create(scope, data, segmentIds);
   }
@@ -1884,8 +1895,8 @@ public <T extends TType, U extends TNumber> SegmentMean<T> segmentMean(Operand<T
    *  first dimension.  Values should be sorted and can be repeated.
    * @return a new instance of SegmentMin
    */
-  public <T extends TNumber, U extends TNumber> SegmentMin<T> segmentMin(Operand<T> data,
-      Operand<U> segmentIds) {
+  public <T extends Tensor & TNumber, U extends Tensor & TNumber> SegmentMin<T> segmentMin(
+      Operand<T> data, Operand<U> segmentIds) {
     return SegmentMin.create(scope, data, segmentIds);
   }
 
@@ -1920,7 +1931,7 @@ public <T extends TNumber, U extends TNumber> SegmentMin<T> segmentMin(Operand<T
    *  first dimension.  Values should be sorted and can be repeated.
    * @return a new instance of SegmentProd
    */
-  public <T extends TType, U extends TNumber> SegmentProd<T> segmentProd(Operand<T> data,
+  public <T extends Tensor, U extends Tensor & TNumber> SegmentProd<T> segmentProd(Operand<T> data,
       Operand<U> segmentIds) {
     return SegmentProd.create(scope, data, segmentIds);
   }
@@ -1956,7 +1967,7 @@ public <T extends TType, U extends TNumber> SegmentProd<T> segmentProd(Operand<T
    *  first dimension.  Values should be sorted and can be repeated.
    * @return a new instance of SegmentSum
    */
-  public <T extends TType, U extends TNumber> SegmentSum<T> segmentSum(Operand<T> data,
+  public <T extends Tensor, U extends Tensor & TNumber> SegmentSum<T> segmentSum(Operand<T> data,
       Operand<U> segmentIds) {
     return SegmentSum.create(scope, data, segmentIds);
   }
@@ -1970,7 +1981,7 @@ public <T extends TType, U extends TNumber> SegmentSum<T> segmentSum(Operand<T>
    * @param x
    * @return a new instance of Sigmoid
    */
-  public <T extends TType> Sigmoid<T> sigmoid(Operand<T> x) {
+  public <T extends Tensor> Sigmoid<T> sigmoid(Operand<T> x) {
     return Sigmoid.create(scope, x);
   }
 
@@ -1989,7 +2000,7 @@ public <T extends TType> Sigmoid<T> sigmoid(Operand<T> x) {
    * @param x
    * @return a new instance of Sign
    */
-  public <T extends TType> Sign<T> sign(Operand<T> x) {
+  public <T extends Tensor> Sign<T> sign(Operand<T> x) {
     return Sign.create(scope, x);
   }
 
@@ -2009,7 +2020,7 @@ public <T extends TType> Sign<T> sign(Operand<T> x) {
    * @param x
    * @return a new instance of Sin
    */
-  public <T extends TType> Sin<T> sin(Operand<T> x) {
+  public <T extends Tensor> Sin<T> sin(Operand<T> x) {
     return Sin.create(scope, x);
   }
 
@@ -2029,7 +2040,7 @@ public <T extends TType> Sin<T> sin(Operand<T> x) {
    * @param x
    * @return a new instance of Sinh
    */
-  public <T extends TType> Sinh<T> sinh(Operand<T> x) {
+  public <T extends Tensor> Sinh<T> sinh(Operand<T> x) {
     return Sinh.create(scope, x);
   }
 
@@ -2040,7 +2051,7 @@ public <T extends TType> Sinh<T> sinh(Operand<T> x) {
    * @param features
    * @return a new instance of Softplus
    */
-  public <T extends TNumber> Softplus<T> softplus(Operand<T> features) {
+  public <T extends Tensor & TNumber> Softplus<T> softplus(Operand<T> features) {
     return Softplus.create(scope, features);
   }
 
@@ -2053,7 +2064,7 @@ public <T extends TNumber> Softplus<T> softplus(Operand<T> features) {
    * @param x
    * @return a new instance of Sqrt
    */
-  public <T extends TType> Sqrt<T> sqrt(Operand<T> x) {
+  public <T extends Tensor> Sqrt<T> sqrt(Operand<T> x) {
     return Sqrt.create(scope, x);
   }
 
@@ -2066,7 +2077,7 @@ public <T extends TType> Sqrt<T> sqrt(Operand<T> x) {
    * @param x
    * @return a new instance of Square
    */
-  public <T extends TType> Square<T> square(Operand<T> x) {
+  public <T extends Tensor> Square<T> square(Operand<T> x) {
     return Square.create(scope, x);
   }
 
@@ -2081,7 +2092,7 @@ public <T extends TType> Square<T> square(Operand<T> x) {
    * @param y
    * @return a new instance of SquaredDifference
    */
-  public <T extends TType> SquaredDifference<T> squaredDifference(Operand<T> x, Operand<T> y) {
+  public <T extends Tensor> SquaredDifference<T> squaredDifference(Operand<T> x, Operand<T> y) {
     return SquaredDifference.create(scope, x, y);
   }
 
@@ -2096,7 +2107,7 @@ public <T extends TType> SquaredDifference<T> squaredDifference(Operand<T> x, Op
    * @param y
    * @return a new instance of Sub
    */
-  public <T extends TType> Sub<T> sub(Operand<T> x, Operand<T> y) {
+  public <T extends Tensor> Sub<T> sub(Operand<T> x, Operand<T> y) {
     return Sub.create(scope, x, y);
   }
 
@@ -2117,7 +2128,7 @@ public <T extends TType> Sub<T> sub(Operand<T> x, Operand<T> y) {
    * @param x
    * @return a new instance of Tan
    */
-  public <T extends TType> Tan<T> tan(Operand<T> x) {
+  public <T extends Tensor> Tan<T> tan(Operand<T> x) {
     return Tan.create(scope, x);
   }
 
@@ -2137,7 +2148,7 @@ public <T extends TType> Tan<T> tan(Operand<T> x) {
    * @param x
    * @return a new instance of Tanh
    */
-  public <T extends TType> Tanh<T> tanh(Operand<T> x) {
+  public <T extends Tensor> Tanh<T> tanh(Operand<T> x) {
     return Tanh.create(scope, x);
   }
 
@@ -2157,7 +2168,7 @@ public <T extends TType> Tanh<T> tanh(Operand<T> x) {
    * @param y
    * @return a new instance of TruncateDiv
    */
-  public <T extends TType> TruncateDiv<T> truncateDiv(Operand<T> x, Operand<T> y) {
+  public <T extends Tensor> TruncateDiv<T> truncateDiv(Operand<T> x, Operand<T> y) {
     return TruncateDiv.create(scope, x, y);
   }
 
@@ -2175,7 +2186,7 @@ public <T extends TType> TruncateDiv<T> truncateDiv(Operand<T> x, Operand<T> y)
    * @param y
    * @return a new instance of TruncateMod
    */
-  public <T extends TNumber> TruncateMod<T> truncateMod(Operand<T> x, Operand<T> y) {
+  public <T extends Tensor & TNumber> TruncateMod<T> truncateMod(Operand<T> x, Operand<T> y) {
     return TruncateMod.create(scope, x, y);
   }
 
@@ -2218,7 +2229,7 @@ public <T extends TNumber> TruncateMod<T> truncateMod(Operand<T> x, Operand<T> y
    * @param numSegments
    * @return a new instance of UnsortedSegmentMax
    */
-  public <T extends TNumber, U extends TNumber, V extends TNumber> UnsortedSegmentMax<T> unsortedSegmentMax(
+  public <T extends Tensor & TNumber, U extends Tensor & TNumber, V extends Tensor & TNumber> UnsortedSegmentMax<T> unsortedSegmentMax(
       Operand<T> data, Operand<U> segmentIds, Operand<V> numSegments) {
     return UnsortedSegmentMax.create(scope, data, segmentIds, numSegments);
   }
@@ -2257,7 +2268,7 @@ public <T extends TNumber, U extends TNumber, V extends TNumber> UnsortedSegment
    * @param numSegments
    * @return a new instance of UnsortedSegmentMin
    */
-  public <T extends TNumber, U extends TNumber, V extends TNumber> UnsortedSegmentMin<T> unsortedSegmentMin(
+  public <T extends Tensor & TNumber, U extends Tensor & TNumber, V extends Tensor & TNumber> UnsortedSegmentMin<T> unsortedSegmentMin(
       Operand<T> data, Operand<U> segmentIds, Operand<V> numSegments) {
     return UnsortedSegmentMin.create(scope, data, segmentIds, numSegments);
   }
@@ -2295,7 +2306,7 @@ public <T extends TNumber, U extends TNumber, V extends TNumber> UnsortedSegment
    * @param numSegments
    * @return a new instance of UnsortedSegmentProd
    */
-  public <T extends TType, U extends TNumber, V extends TNumber> UnsortedSegmentProd<T> unsortedSegmentProd(
+  public <T extends Tensor, U extends Tensor & TNumber, V extends Tensor & TNumber> UnsortedSegmentProd<T> unsortedSegmentProd(
       Operand<T> data, Operand<U> segmentIds, Operand<V> numSegments) {
     return UnsortedSegmentProd.create(scope, data, segmentIds, numSegments);
   }
@@ -2335,7 +2346,7 @@ public <T extends TType, U extends TNumber, V extends TNumber> UnsortedSegmentPr
    * @param numSegments
    * @return a new instance of UnsortedSegmentSum
    */
-  public <T extends TType, U extends TNumber, V extends TNumber> UnsortedSegmentSum<T> unsortedSegmentSum(
+  public <T extends Tensor, U extends Tensor & TNumber, V extends Tensor & TNumber> UnsortedSegmentSum<T> unsortedSegmentSum(
       Operand<T> data, Operand<U> segmentIds, Operand<V> numSegments) {
     return UnsortedSegmentSum.create(scope, data, segmentIds, numSegments);
   }
@@ -2348,7 +2359,7 @@ public <T extends TType, U extends TNumber, V extends TNumber> UnsortedSegmentSu
    * @param y
    * @return a new instance of Xdivy
    */
-  public <T extends TType> Xdivy<T> xdivy(Operand<T> x, Operand<T> y) {
+  public <T extends Tensor> Xdivy<T> xdivy(Operand<T> x, Operand<T> y) {
     return Xdivy.create(scope, x, y);
   }
 
@@ -2360,7 +2371,7 @@ public <T extends TType> Xdivy<T> xdivy(Operand<T> x, Operand<T> y) {
    * @param y
    * @return a new instance of Xlog1py
    */
-  public <T extends TType> Xlog1py<T> xlog1py(Operand<T> x, Operand<T> y) {
+  public <T extends Tensor> Xlog1py<T> xlog1py(Operand<T> x, Operand<T> y) {
     return Xlog1py.create(scope, x, y);
   }
 
@@ -2372,7 +2383,7 @@ public <T extends TType> Xlog1py<T> xlog1py(Operand<T> x, Operand<T> y) {
    * @param y
    * @return a new instance of Xlogy
    */
-  public <T extends TType> Xlogy<T> xlogy(Operand<T> x, Operand<T> y) {
+  public <T extends Tensor> Xlogy<T> xlogy(Operand<T> x, Operand<T> y) {
     return Xlogy.create(scope, x, y);
   }
 
@@ -2388,7 +2399,7 @@ public <T extends TType> Xlogy<T> xlogy(Operand<T> x, Operand<T> y) {
    * @param q
    * @return a new instance of Zeta
    */
-  public <T extends TNumber> Zeta<T> zeta(Operand<T> x, Operand<T> q) {
+  public <T extends Tensor & TNumber> Zeta<T> zeta(Operand<T> x, Operand<T> q) {
     return Zeta.create(scope, x, q);
   }
 }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/NnOps.java b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/NnOps.java
index 81a24514a08..9eb29ba9835 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/NnOps.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/NnOps.java
@@ -20,6 +20,7 @@
 import java.util.List;
 import org.tensorflow.DataType;
 import org.tensorflow.Operand;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.nn.AvgPool;
 import org.tensorflow.op.nn.AvgPool3d;
 import org.tensorflow.op.nn.AvgPool3dGrad;
@@ -84,7 +85,6 @@
 import org.tensorflow.op.nn.Relu;
 import org.tensorflow.op.nn.Relu6;
 import org.tensorflow.op.nn.Selu;
-import org.tensorflow.op.nn.SigmoidCrossEntropyWithLogits;
 import org.tensorflow.op.nn.Softmax;
 import org.tensorflow.op.nn.SoftmaxCrossEntropyWithLogits;
 import org.tensorflow.op.nn.Softsign;
@@ -96,7 +96,6 @@
 import org.tensorflow.types.TInt32;
 import org.tensorflow.types.TInt64;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * An API for building {@code nn} operations as {@link Op Op}s
@@ -104,13 +103,10 @@
  * @see {@link Ops}
  */
 public final class NnOps {
-  public final NnRawOps raw;
-
   private final Scope scope;
 
   NnOps(Scope scope) {
     this.scope = scope;
-    raw = new NnRawOps(scope);
   }
 
   /**
@@ -127,7 +123,7 @@ public final class NnOps {
    * @param options carries optional attributes values
    * @return a new instance of AvgPool
    */
-  public <T extends TNumber> AvgPool<T> avgPool(Operand<T> value, List<Long> ksize,
+  public <T extends Tensor & TNumber> AvgPool<T> avgPool(Operand<T> value, List<Long> ksize,
       List<Long> strides, String padding, AvgPool.Options... options) {
     return AvgPool.create(scope, value, ksize, strides, padding, options);
   }
@@ -145,7 +141,7 @@ public <T extends TNumber> AvgPool<T> avgPool(Operand<T> value, List<Long> ksize
    * @param options carries optional attributes values
    * @return a new instance of AvgPool3d
    */
-  public <T extends TNumber> AvgPool3d<T> avgPool3d(Operand<T> input, List<Long> ksize,
+  public <T extends Tensor & TNumber> AvgPool3d<T> avgPool3d(Operand<T> input, List<Long> ksize,
       List<Long> strides, String padding, AvgPool3d.Options... options) {
     return AvgPool3d.create(scope, input, ksize, strides, padding, options);
   }
@@ -164,7 +160,7 @@ public <T extends TNumber> AvgPool3d<T> avgPool3d(Operand<T> input, List<Long> k
    * @param options carries optional attributes values
    * @return a new instance of AvgPool3dGrad
    */
-  public <T extends TNumber> AvgPool3dGrad<T> avgPool3dGrad(Operand<TInt32> origInputShape,
+  public <T extends Tensor & TNumber> AvgPool3dGrad<T> avgPool3dGrad(Operand<TInt32> origInputShape,
       Operand<T> grad, List<Long> ksize, List<Long> strides, String padding,
       AvgPool3dGrad.Options... options) {
     return AvgPool3dGrad.create(scope, origInputShape, grad, ksize, strides, padding, options);
@@ -193,7 +189,7 @@ public <T extends TNumber> AvgPool3dGrad<T> avgPool3dGrad(Operand<TInt32> origIn
    *  needs to be multiplied with gamma.
    * @return a new instance of BatchNormWithGlobalNormalization
    */
-  public <T extends TType> BatchNormWithGlobalNormalization<T> batchNormWithGlobalNormalization(
+  public <T extends Tensor> BatchNormWithGlobalNormalization<T> batchNormWithGlobalNormalization(
       Operand<T> t, Operand<T> m, Operand<T> v, Operand<T> beta, Operand<T> gamma,
       Float varianceEpsilon, Boolean scaleAfterNormalization) {
     return BatchNormWithGlobalNormalization.create(scope, t, m, v, beta, gamma, varianceEpsilon, scaleAfterNormalization);
@@ -221,7 +217,7 @@ public <T extends TType> BatchNormWithGlobalNormalization<T> batchNormWithGlobal
    *  needs to be multiplied with gamma.
    * @return a new instance of BatchNormWithGlobalNormalizationGrad
    */
-  public <T extends TType> BatchNormWithGlobalNormalizationGrad<T> batchNormWithGlobalNormalizationGrad(
+  public <T extends Tensor> BatchNormWithGlobalNormalizationGrad<T> batchNormWithGlobalNormalizationGrad(
       Operand<T> t, Operand<T> m, Operand<T> v, Operand<T> gamma, Operand<T> backprop,
       Float varianceEpsilon, Boolean scaleAfterNormalization) {
     return BatchNormWithGlobalNormalizationGrad.create(scope, t, m, v, gamma, backprop, varianceEpsilon, scaleAfterNormalization);
@@ -239,7 +235,7 @@ public <T extends TType> BatchNormWithGlobalNormalizationGrad<T> batchNormWithGl
    * @param options carries optional attributes values
    * @return a new instance of BiasAdd
    */
-  public <T extends TType> BiasAdd<T> biasAdd(Operand<T> value, Operand<T> bias,
+  public <T extends Tensor> BiasAdd<T> biasAdd(Operand<T> value, Operand<T> bias,
       BiasAdd.Options... options) {
     return BiasAdd.create(scope, value, bias, options);
   }
@@ -256,7 +252,7 @@ public <T extends TType> BiasAdd<T> biasAdd(Operand<T> value, Operand<T> bias,
    * @param options carries optional attributes values
    * @return a new instance of BiasAddGrad
    */
-  public <T extends TType> BiasAddGrad<T> biasAddGrad(Operand<T> outBackprop,
+  public <T extends Tensor> BiasAddGrad<T> biasAddGrad(Operand<T> outBackprop,
       BiasAddGrad.Options... options) {
     return BiasAddGrad.create(scope, outBackprop, options);
   }
@@ -317,7 +313,7 @@ public ComputeAccidentalHits computeAccidentalHits(Operand<TInt64> trueClasses,
    * @param options carries optional attributes values
    * @return a new instance of Conv2d
    */
-  public <T extends TNumber> Conv2d<T> conv2d(Operand<T> input, Operand<T> filter,
+  public <T extends Tensor & TNumber> Conv2d<T> conv2d(Operand<T> input, Operand<T> filter,
       List<Long> strides, String padding, Conv2d.Options... options) {
     return Conv2d.create(scope, input, filter, strides, padding, options);
   }
@@ -339,7 +335,7 @@ public <T extends TNumber> Conv2d<T> conv2d(Operand<T> input, Operand<T> filter,
    * @param options carries optional attributes values
    * @return a new instance of Conv2dBackpropFilter
    */
-  public <T extends TNumber> Conv2dBackpropFilter<T> conv2dBackpropFilter(Operand<T> input,
+  public <T extends Tensor & TNumber> Conv2dBackpropFilter<T> conv2dBackpropFilter(Operand<T> input,
       Operand<TInt32> filterSizes, Operand<T> outBackprop, List<Long> strides, String padding,
       Conv2dBackpropFilter.Options... options) {
     return Conv2dBackpropFilter.create(scope, input, filterSizes, outBackprop, strides, padding, options);
@@ -362,9 +358,9 @@ public <T extends TNumber> Conv2dBackpropFilter<T> conv2dBackpropFilter(Operand<
    * @param options carries optional attributes values
    * @return a new instance of Conv2dBackpropInput
    */
-  public <T extends TNumber> Conv2dBackpropInput<T> conv2dBackpropInput(Operand<TInt32> inputSizes,
-      Operand<T> filter, Operand<T> outBackprop, List<Long> strides, String padding,
-      Conv2dBackpropInput.Options... options) {
+  public <T extends Tensor & TNumber> Conv2dBackpropInput<T> conv2dBackpropInput(
+      Operand<TInt32> inputSizes, Operand<T> filter, Operand<T> outBackprop, List<Long> strides,
+      String padding, Conv2dBackpropInput.Options... options) {
     return Conv2dBackpropInput.create(scope, inputSizes, filter, outBackprop, strides, padding, options);
   }
 
@@ -387,7 +383,7 @@ public <T extends TNumber> Conv2dBackpropInput<T> conv2dBackpropInput(Operand<TI
    * @param options carries optional attributes values
    * @return a new instance of Conv3d
    */
-  public <T extends TNumber> Conv3d<T> conv3d(Operand<T> input, Operand<T> filter,
+  public <T extends Tensor & TNumber> Conv3d<T> conv3d(Operand<T> input, Operand<T> filter,
       List<Long> strides, String padding, Conv3d.Options... options) {
     return Conv3d.create(scope, input, filter, strides, padding, options);
   }
@@ -409,7 +405,7 @@ public <T extends TNumber> Conv3d<T> conv3d(Operand<T> input, Operand<T> filter,
    * @param options carries optional attributes values
    * @return a new instance of Conv3dBackpropFilter
    */
-  public <T extends TNumber> Conv3dBackpropFilter<T> conv3dBackpropFilter(Operand<T> input,
+  public <T extends Tensor & TNumber> Conv3dBackpropFilter<T> conv3dBackpropFilter(Operand<T> input,
       Operand<TInt32> filterSizes, Operand<T> outBackprop, List<Long> strides, String padding,
       Conv3dBackpropFilter.Options... options) {
     return Conv3dBackpropFilter.create(scope, input, filterSizes, outBackprop, strides, padding, options);
@@ -432,7 +428,7 @@ public <T extends TNumber> Conv3dBackpropFilter<T> conv3dBackpropFilter(Operand<
    * @param options carries optional attributes values
    * @return a new instance of Conv3dBackpropInput
    */
-  public <U extends TNumber, T extends TNumber> Conv3dBackpropInput<U> conv3dBackpropInput(
+  public <U extends Tensor & TNumber, T extends Tensor & TNumber> Conv3dBackpropInput<U> conv3dBackpropInput(
       Operand<T> inputSizes, Operand<U> filter, Operand<U> outBackprop, List<Long> strides,
       String padding, Conv3dBackpropInput.Options... options) {
     return Conv3dBackpropInput.create(scope, inputSizes, filter, outBackprop, strides, padding, options);
@@ -455,8 +451,8 @@ public <U extends TNumber, T extends TNumber> Conv3dBackpropInput<U> conv3dBackp
    * @param options carries optional attributes values
    * @return a new instance of CtcBeamSearchDecoder
    */
-  public <T extends TNumber> CtcBeamSearchDecoder<T> ctcBeamSearchDecoder(Operand<T> inputs,
-      Operand<TInt32> sequenceLength, Long beamWidth, Long topPaths,
+  public <T extends Tensor & TNumber> CtcBeamSearchDecoder<T> ctcBeamSearchDecoder(
+      Operand<T> inputs, Operand<TInt32> sequenceLength, Long beamWidth, Long topPaths,
       CtcBeamSearchDecoder.Options... options) {
     return CtcBeamSearchDecoder.create(scope, inputs, sequenceLength, beamWidth, topPaths, options);
   }
@@ -480,7 +476,7 @@ public <T extends TNumber> CtcBeamSearchDecoder<T> ctcBeamSearchDecoder(Operand<
    * @param options carries optional attributes values
    * @return a new instance of CtcGreedyDecoder
    */
-  public <T extends TNumber> CtcGreedyDecoder<T> ctcGreedyDecoder(Operand<T> inputs,
+  public <T extends Tensor & TNumber> CtcGreedyDecoder<T> ctcGreedyDecoder(Operand<T> inputs,
       Operand<TInt32> sequenceLength, CtcGreedyDecoder.Options... options) {
     return CtcGreedyDecoder.create(scope, inputs, sequenceLength, options);
   }
@@ -501,8 +497,9 @@ public <T extends TNumber> CtcGreedyDecoder<T> ctcGreedyDecoder(Operand<T> input
    * @param options carries optional attributes values
    * @return a new instance of CtcLoss
    */
-  public <T extends TNumber> CtcLoss<T> ctcLoss(Operand<T> inputs, Operand<TInt64> labelsIndices,
-      Operand<TInt32> labelsValues, Operand<TInt32> sequenceLength, CtcLoss.Options... options) {
+  public <T extends Tensor & TNumber> CtcLoss<T> ctcLoss(Operand<T> inputs,
+      Operand<TInt64> labelsIndices, Operand<TInt32> labelsValues, Operand<TInt32> sequenceLength,
+      CtcLoss.Options... options) {
     return CtcLoss.create(scope, inputs, labelsIndices, labelsValues, sequenceLength, options);
   }
 
@@ -549,7 +546,7 @@ public <T extends TNumber> CtcLoss<T> ctcLoss(Operand<T> inputs, Operand<TInt64>
    * @param options carries optional attributes values
    * @return a new instance of CudnnRNNCanonicalToParams
    */
-  public <T extends TNumber> CudnnRNNCanonicalToParams<T> cudnnRNNCanonicalToParams(
+  public <T extends Tensor & TNumber> CudnnRNNCanonicalToParams<T> cudnnRNNCanonicalToParams(
       Operand<TInt32> numLayers, Operand<TInt32> numUnits, Operand<TInt32> inputSize,
       Iterable<Operand<T>> weights, Iterable<Operand<T>> biases,
       CudnnRNNCanonicalToParams.Options... options) {
@@ -600,7 +597,7 @@ public <T extends TNumber> CudnnRNNCanonicalToParams<T> cudnnRNNCanonicalToParam
    * @param options carries optional attributes values
    * @return a new instance of CudnnRNNParamsToCanonical
    */
-  public <T extends TNumber> CudnnRNNParamsToCanonical<T> cudnnRNNParamsToCanonical(
+  public <T extends Tensor & TNumber> CudnnRNNParamsToCanonical<T> cudnnRNNParamsToCanonical(
       Operand<TInt32> numLayers, Operand<TInt32> numUnits, Operand<TInt32> inputSize,
       Operand<T> params, Long numParamsWeights, Long numParamsBiases,
       CudnnRNNParamsToCanonical.Options... options) {
@@ -641,7 +638,7 @@ public <T extends TNumber> CudnnRNNParamsToCanonical<T> cudnnRNNParamsToCanonica
    * @param options carries optional attributes values
    * @return a new instance of CudnnRnnParamsSize
    */
-  public <U extends TNumber, T extends TNumber> CudnnRnnParamsSize<U> cudnnRnnParamsSize(
+  public <U extends Tensor & TNumber, T extends Tensor & TNumber> CudnnRnnParamsSize<U> cudnnRnnParamsSize(
       Operand<TInt32> numLayers, Operand<TInt32> numUnits, Operand<TInt32> inputSize, DataType<T> T,
       DataType<U> S, CudnnRnnParamsSize.Options... options) {
     return CudnnRnnParamsSize.create(scope, numLayers, numUnits, inputSize, T, S, options);
@@ -658,7 +655,7 @@ public <U extends TNumber, T extends TNumber> CudnnRnnParamsSize<U> cudnnRnnPara
    * @param options carries optional attributes values
    * @return a new instance of DataFormatDimMap
    */
-  public <T extends TNumber> DataFormatDimMap<T> dataFormatDimMap(Operand<T> x,
+  public <T extends Tensor & TNumber> DataFormatDimMap<T> dataFormatDimMap(Operand<T> x,
       DataFormatDimMap.Options... options) {
     return DataFormatDimMap.create(scope, x, options);
   }
@@ -673,7 +670,7 @@ public <T extends TNumber> DataFormatDimMap<T> dataFormatDimMap(Operand<T> x,
    * @param options carries optional attributes values
    * @return a new instance of DataFormatVecPermute
    */
-  public <T extends TNumber> DataFormatVecPermute<T> dataFormatVecPermute(Operand<T> x,
+  public <T extends Tensor & TNumber> DataFormatVecPermute<T> dataFormatVecPermute(Operand<T> x,
       DataFormatVecPermute.Options... options) {
     return DataFormatVecPermute.create(scope, x, options);
   }
@@ -766,7 +763,7 @@ public <T extends TNumber> DataFormatVecPermute<T> dataFormatVecPermute(Operand<
    * @param options carries optional attributes values
    * @return a new instance of DepthToSpace
    */
-  public <T extends TType> DepthToSpace<T> depthToSpace(Operand<T> input, Long blockSize,
+  public <T extends Tensor> DepthToSpace<T> depthToSpace(Operand<T> input, Long blockSize,
       DepthToSpace.Options... options) {
     return DepthToSpace.create(scope, input, blockSize, options);
   }
@@ -800,8 +797,8 @@ public <T extends TType> DepthToSpace<T> depthToSpace(Operand<T> input, Long blo
    * @param options carries optional attributes values
    * @return a new instance of DepthwiseConv2dNative
    */
-  public <T extends TNumber> DepthwiseConv2dNative<T> depthwiseConv2dNative(Operand<T> input,
-      Operand<T> filter, List<Long> strides, String padding,
+  public <T extends Tensor & TNumber> DepthwiseConv2dNative<T> depthwiseConv2dNative(
+      Operand<T> input, Operand<T> filter, List<Long> strides, String padding,
       DepthwiseConv2dNative.Options... options) {
     return DepthwiseConv2dNative.create(scope, input, filter, strides, padding, options);
   }
@@ -826,7 +823,7 @@ public <T extends TNumber> DepthwiseConv2dNative<T> depthwiseConv2dNative(Operan
    * @param options carries optional attributes values
    * @return a new instance of DepthwiseConv2dNativeBackpropFilter
    */
-  public <T extends TNumber> DepthwiseConv2dNativeBackpropFilter<T> depthwiseConv2dNativeBackpropFilter(
+  public <T extends Tensor & TNumber> DepthwiseConv2dNativeBackpropFilter<T> depthwiseConv2dNativeBackpropFilter(
       Operand<T> input, Operand<TInt32> filterSizes, Operand<T> outBackprop, List<Long> strides,
       String padding, DepthwiseConv2dNativeBackpropFilter.Options... options) {
     return DepthwiseConv2dNativeBackpropFilter.create(scope, input, filterSizes, outBackprop, strides, padding, options);
@@ -851,7 +848,7 @@ public <T extends TNumber> DepthwiseConv2dNativeBackpropFilter<T> depthwiseConv2
    * @param options carries optional attributes values
    * @return a new instance of DepthwiseConv2dNativeBackpropInput
    */
-  public <T extends TNumber> DepthwiseConv2dNativeBackpropInput<T> depthwiseConv2dNativeBackpropInput(
+  public <T extends Tensor & TNumber> DepthwiseConv2dNativeBackpropInput<T> depthwiseConv2dNativeBackpropInput(
       Operand<TInt32> inputSizes, Operand<T> filter, Operand<T> outBackprop, List<Long> strides,
       String padding, DepthwiseConv2dNativeBackpropInput.Options... options) {
     return DepthwiseConv2dNativeBackpropInput.create(scope, inputSizes, filter, outBackprop, strides, padding, options);
@@ -894,7 +891,7 @@ public <T extends TNumber> DepthwiseConv2dNativeBackpropInput<T> depthwiseConv2d
    * @param padding The type of padding algorithm to use.
    * @return a new instance of Dilation2d
    */
-  public <T extends TNumber> Dilation2d<T> dilation2d(Operand<T> input, Operand<T> filter,
+  public <T extends Tensor & TNumber> Dilation2d<T> dilation2d(Operand<T> input, Operand<T> filter,
       List<Long> strides, List<Long> rates, String padding) {
     return Dilation2d.create(scope, input, filter, strides, rates, padding);
   }
@@ -913,9 +910,9 @@ public <T extends TNumber> Dilation2d<T> dilation2d(Operand<T> input, Operand<T>
    * @param padding The type of padding algorithm to use.
    * @return a new instance of Dilation2dBackpropFilter
    */
-  public <T extends TNumber> Dilation2dBackpropFilter<T> dilation2dBackpropFilter(Operand<T> input,
-      Operand<T> filter, Operand<T> outBackprop, List<Long> strides, List<Long> rates,
-      String padding) {
+  public <T extends Tensor & TNumber> Dilation2dBackpropFilter<T> dilation2dBackpropFilter(
+      Operand<T> input, Operand<T> filter, Operand<T> outBackprop, List<Long> strides,
+      List<Long> rates, String padding) {
     return Dilation2dBackpropFilter.create(scope, input, filter, outBackprop, strides, rates, padding);
   }
 
@@ -933,9 +930,9 @@ public <T extends TNumber> Dilation2dBackpropFilter<T> dilation2dBackpropFilter(
    * @param padding The type of padding algorithm to use.
    * @return a new instance of Dilation2dBackpropInput
    */
-  public <T extends TNumber> Dilation2dBackpropInput<T> dilation2dBackpropInput(Operand<T> input,
-      Operand<T> filter, Operand<T> outBackprop, List<Long> strides, List<Long> rates,
-      String padding) {
+  public <T extends Tensor & TNumber> Dilation2dBackpropInput<T> dilation2dBackpropInput(
+      Operand<T> input, Operand<T> filter, Operand<T> outBackprop, List<Long> strides,
+      List<Long> rates, String padding) {
     return Dilation2dBackpropInput.create(scope, input, filter, outBackprop, strides, rates, padding);
   }
 
@@ -949,7 +946,7 @@ public <T extends TNumber> Dilation2dBackpropInput<T> dilation2dBackpropInput(Op
    * @param features
    * @return a new instance of Elu
    */
-  public <T extends TNumber> Elu<T> elu(Operand<T> features) {
+  public <T extends Tensor & TNumber> Elu<T> elu(Operand<T> features) {
     return Elu.create(scope, features);
   }
 
@@ -1007,7 +1004,7 @@ public FixedUnigramCandidateSampler fixedUnigramCandidateSampler(Operand<TInt64>
    * @param options carries optional attributes values
    * @return a new instance of FractionalAvgPool
    */
-  public <T extends TNumber> FractionalAvgPool<T> fractionalAvgPool(Operand<T> value,
+  public <T extends Tensor & TNumber> FractionalAvgPool<T> fractionalAvgPool(Operand<T> value,
       List<Float> poolingRatio, FractionalAvgPool.Options... options) {
     return FractionalAvgPool.create(scope, value, poolingRatio, options);
   }
@@ -1055,7 +1052,7 @@ public <T extends TNumber> FractionalAvgPool<T> fractionalAvgPool(Operand<T> val
    * @param options carries optional attributes values
    * @return a new instance of FractionalMaxPool
    */
-  public <T extends TNumber> FractionalMaxPool<T> fractionalMaxPool(Operand<T> value,
+  public <T extends Tensor & TNumber> FractionalMaxPool<T> fractionalMaxPool(Operand<T> value,
       List<Float> poolingRatio, FractionalMaxPool.Options... options) {
     return FractionalMaxPool.create(scope, value, poolingRatio, options);
   }
@@ -1078,8 +1075,8 @@ public <T extends TNumber> FractionalMaxPool<T> fractionalMaxPool(Operand<T> val
    * @param options carries optional attributes values
    * @return a new instance of FusedBatchNorm
    */
-  public <T extends TNumber, U extends TNumber> FusedBatchNorm<T, U> fusedBatchNorm(Operand<T> x,
-      Operand<U> scale, Operand<U> offset, Operand<U> mean, Operand<U> variance,
+  public <T extends Tensor & TNumber, U extends Tensor & TNumber> FusedBatchNorm<T, U> fusedBatchNorm(
+      Operand<T> x, Operand<U> scale, Operand<U> offset, Operand<U> mean, Operand<U> variance,
       FusedBatchNorm.Options... options) {
     return FusedBatchNorm.create(scope, x, scale, offset, mean, variance, options);
   }
@@ -1110,7 +1107,7 @@ public <T extends TNumber, U extends TNumber> FusedBatchNorm<T, U> fusedBatchNor
    * @param options carries optional attributes values
    * @return a new instance of FusedBatchNormGrad
    */
-  public <T extends TNumber, U extends TNumber> FusedBatchNormGrad<T, U> fusedBatchNormGrad(
+  public <T extends Tensor & TNumber, U extends Tensor & TNumber> FusedBatchNormGrad<T, U> fusedBatchNormGrad(
       Operand<T> yBackprop, Operand<T> x, Operand<TFloat32> scale, Operand<U> reserveSpace1,
       Operand<U> reserveSpace2, Operand<U> reserveSpace3, FusedBatchNormGrad.Options... options) {
     return FusedBatchNormGrad.create(scope, yBackprop, x, scale, reserveSpace1, reserveSpace2, reserveSpace3, options);
@@ -1143,7 +1140,7 @@ public <T extends TNumber, U extends TNumber> FusedBatchNormGrad<T, U> fusedBatc
    * @param padding The type of padding algorithm to use.
    * @return a new instance of FusedPadConv2d
    */
-  public <T extends TNumber> FusedPadConv2d<T> fusedPadConv2d(Operand<T> input,
+  public <T extends Tensor & TNumber> FusedPadConv2d<T> fusedPadConv2d(Operand<T> input,
       Operand<TInt32> paddings, Operand<T> filter, String mode, List<Long> strides,
       String padding) {
     return FusedPadConv2d.create(scope, input, paddings, filter, mode, strides, padding);
@@ -1178,9 +1175,9 @@ public <T extends TNumber> FusedPadConv2d<T> fusedPadConv2d(Operand<T> input,
    * @param options carries optional attributes values
    * @return a new instance of FusedResizeAndPadConv2d
    */
-  public <T extends TNumber> FusedResizeAndPadConv2d<T> fusedResizeAndPadConv2d(Operand<T> input,
-      Operand<TInt32> size, Operand<TInt32> paddings, Operand<T> filter, String mode,
-      List<Long> strides, String padding, FusedResizeAndPadConv2d.Options... options) {
+  public <T extends Tensor & TNumber> FusedResizeAndPadConv2d<T> fusedResizeAndPadConv2d(
+      Operand<T> input, Operand<TInt32> size, Operand<TInt32> paddings, Operand<T> filter,
+      String mode, List<Long> strides, String padding, FusedResizeAndPadConv2d.Options... options) {
     return FusedResizeAndPadConv2d.create(scope, input, size, paddings, filter, mode, strides, padding, options);
   }
 
@@ -1207,8 +1204,8 @@ public <T extends TNumber> FusedResizeAndPadConv2d<T> fusedResizeAndPadConv2d(Op
    * @param k Number of top elements to look at for computing precision.
    * @return a new instance of InTopK
    */
-  public <T extends TNumber> InTopK inTopK(Operand<TFloat32> predictions, Operand<T> targets,
-      Operand<T> k) {
+  public <T extends Tensor & TNumber> InTopK inTopK(Operand<TFloat32> predictions,
+      Operand<T> targets, Operand<T> k) {
     return InTopK.create(scope, predictions, targets, k);
   }
 
@@ -1223,7 +1220,7 @@ public <T extends TNumber> InTopK inTopK(Operand<TFloat32> predictions, Operand<
    * @param t Typically 2-D, but may have any dimensions.
    * @return a new instance of L2Loss
    */
-  public <T extends TNumber> L2Loss<T> l2Loss(Operand<T> t) {
+  public <T extends Tensor & TNumber> L2Loss<T> l2Loss(Operand<T> t) {
     return L2Loss.create(scope, t);
   }
 
@@ -1290,7 +1287,7 @@ public LearnedUnigramCandidateSampler learnedUnigramCandidateSampler(Operand<TIn
    * @param options carries optional attributes values
    * @return a new instance of LocalResponseNormalization
    */
-  public <T extends TNumber> LocalResponseNormalization<T> localResponseNormalization(
+  public <T extends Tensor & TNumber> LocalResponseNormalization<T> localResponseNormalization(
       Operand<T> input, LocalResponseNormalization.Options... options) {
     return LocalResponseNormalization.create(scope, input, options);
   }
@@ -1306,7 +1303,7 @@ public <T extends TNumber> LocalResponseNormalization<T> localResponseNormalizat
    * @param logits 2-D with shape `[batch_size, num_classes]`.
    * @return a new instance of LogSoftmax
    */
-  public <T extends TNumber> LogSoftmax<T> logSoftmax(Operand<T> logits) {
+  public <T extends Tensor & TNumber> LogSoftmax<T> logSoftmax(Operand<T> logits) {
     return LogSoftmax.create(scope, logits);
   }
 
@@ -1322,7 +1319,7 @@ public <T extends TNumber> LogSoftmax<T> logSoftmax(Operand<T> logits) {
    * @param options carries optional attributes values
    * @return a new instance of MaxPool
    */
-  public <T extends TType> MaxPool<T> maxPool(Operand<T> input, Operand<TInt32> ksize,
+  public <T extends Tensor> MaxPool<T> maxPool(Operand<T> input, Operand<TInt32> ksize,
       Operand<TInt32> strides, String padding, MaxPool.Options... options) {
     return MaxPool.create(scope, input, ksize, strides, padding, options);
   }
@@ -1340,13 +1337,13 @@ public <T extends TType> MaxPool<T> maxPool(Operand<T> input, Operand<TInt32> ks
    * @param options carries optional attributes values
    * @return a new instance of MaxPool3d
    */
-  public <T extends TNumber> MaxPool3d<T> maxPool3d(Operand<T> input, List<Long> ksize,
+  public <T extends Tensor & TNumber> MaxPool3d<T> maxPool3d(Operand<T> input, List<Long> ksize,
       List<Long> strides, String padding, MaxPool3d.Options... options) {
     return MaxPool3d.create(scope, input, ksize, strides, padding, options);
   }
 
   /**
-   * Computes gradients of 3D max pooling function.
+   * Computes gradients of max pooling function.
    *
    * @param <U> data type for {@code output()} output
    * @param origInput The original input tensor.
@@ -1360,9 +1357,9 @@ public <T extends TNumber> MaxPool3d<T> maxPool3d(Operand<T> input, List<Long> k
    * @param options carries optional attributes values
    * @return a new instance of MaxPool3dGrad
    */
-  public <U extends TNumber, T extends TNumber> MaxPool3dGrad<U> maxPool3dGrad(Operand<T> origInput,
-      Operand<T> origOutput, Operand<U> grad, List<Long> ksize, List<Long> strides, String padding,
-      MaxPool3dGrad.Options... options) {
+  public <U extends Tensor & TNumber, T extends Tensor & TNumber> MaxPool3dGrad<U> maxPool3dGrad(
+      Operand<T> origInput, Operand<T> origOutput, Operand<U> grad, List<Long> ksize,
+      List<Long> strides, String padding, MaxPool3dGrad.Options... options) {
     return MaxPool3dGrad.create(scope, origInput, origOutput, grad, ksize, strides, padding, options);
   }
 
@@ -1381,7 +1378,7 @@ public <U extends TNumber, T extends TNumber> MaxPool3dGrad<U> maxPool3dGrad(Ope
    * @param options carries optional attributes values
    * @return a new instance of MaxPool3dGradGrad
    */
-  public <T extends TNumber> MaxPool3dGradGrad<T> maxPool3dGradGrad(Operand<T> origInput,
+  public <T extends Tensor & TNumber> MaxPool3dGradGrad<T> maxPool3dGradGrad(Operand<T> origInput,
       Operand<T> origOutput, Operand<T> grad, List<Long> ksize, List<Long> strides, String padding,
       MaxPool3dGradGrad.Options... options) {
     return MaxPool3dGradGrad.create(scope, origInput, origOutput, grad, ksize, strides, padding, options);
@@ -1401,9 +1398,9 @@ public <T extends TNumber> MaxPool3dGradGrad<T> maxPool3dGradGrad(Operand<T> ori
    * @param options carries optional attributes values
    * @return a new instance of MaxPoolGrad
    */
-  public <T extends TNumber> MaxPoolGrad<T> maxPoolGrad(Operand<T> origInput, Operand<T> origOutput,
-      Operand<T> grad, Operand<TInt32> ksize, Operand<TInt32> strides, String padding,
-      MaxPoolGrad.Options... options) {
+  public <T extends Tensor & TNumber> MaxPoolGrad<T> maxPoolGrad(Operand<T> origInput,
+      Operand<T> origOutput, Operand<T> grad, Operand<TInt32> ksize, Operand<TInt32> strides,
+      String padding, MaxPoolGrad.Options... options) {
     return MaxPoolGrad.create(scope, origInput, origOutput, grad, ksize, strides, padding, options);
   }
 
@@ -1421,7 +1418,7 @@ public <T extends TNumber> MaxPoolGrad<T> maxPoolGrad(Operand<T> origInput, Oper
    * @param options carries optional attributes values
    * @return a new instance of MaxPoolGradGrad
    */
-  public <T extends TNumber> MaxPoolGradGrad<T> maxPoolGradGrad(Operand<T> origInput,
+  public <T extends Tensor & TNumber> MaxPoolGradGrad<T> maxPoolGradGrad(Operand<T> origInput,
       Operand<T> origOutput, Operand<T> grad, Operand<TInt32> ksize, Operand<TInt32> strides,
       String padding, MaxPoolGradGrad.Options... options) {
     return MaxPoolGradGrad.create(scope, origInput, origOutput, grad, ksize, strides, padding, options);
@@ -1442,7 +1439,7 @@ public <T extends TNumber> MaxPoolGradGrad<T> maxPoolGradGrad(Operand<T> origInp
    * @param options carries optional attributes values
    * @return a new instance of MaxPoolGradGradWithArgmax
    */
-  public <T extends TNumber, U extends TNumber> MaxPoolGradGradWithArgmax<T> maxPoolGradGradWithArgmax(
+  public <T extends Tensor & TNumber, U extends Tensor & TNumber> MaxPoolGradGradWithArgmax<T> maxPoolGradGradWithArgmax(
       Operand<T> input, Operand<T> grad, Operand<U> argmax, List<Long> ksize, List<Long> strides,
       String padding, MaxPoolGradGradWithArgmax.Options... options) {
     return MaxPoolGradGradWithArgmax.create(scope, input, grad, argmax, ksize, strides, padding, options);
@@ -1471,8 +1468,9 @@ public <T extends TNumber, U extends TNumber> MaxPoolGradGradWithArgmax<T> maxPo
    * @param options carries optional attributes values
    * @return a new instance of MaxPoolWithArgmax
    */
-  public <T extends TNumber> MaxPoolWithArgmax<T, TInt64> maxPoolWithArgmax(Operand<T> input,
-      List<Long> ksize, List<Long> strides, String padding, MaxPoolWithArgmax.Options... options) {
+  public <T extends Tensor & TNumber> MaxPoolWithArgmax<T, TInt64> maxPoolWithArgmax(
+      Operand<T> input, List<Long> ksize, List<Long> strides, String padding,
+      MaxPoolWithArgmax.Options... options) {
     return MaxPoolWithArgmax.create(scope, input, ksize, strides, padding, options);
   }
 
@@ -1500,7 +1498,7 @@ public <T extends TNumber> MaxPoolWithArgmax<T, TInt64> maxPoolWithArgmax(Operan
    * @param options carries optional attributes values
    * @return a new instance of MaxPoolWithArgmax
    */
-  public <T extends TNumber, U extends TNumber> MaxPoolWithArgmax<T, U> maxPoolWithArgmax(
+  public <T extends Tensor & TNumber, U extends Tensor & TNumber> MaxPoolWithArgmax<T, U> maxPoolWithArgmax(
       Operand<T> input, List<Long> ksize, List<Long> strides, DataType<U> Targmax, String padding,
       MaxPoolWithArgmax.Options... options) {
     return MaxPoolWithArgmax.create(scope, input, ksize, strides, Targmax, padding, options);
@@ -1524,7 +1522,7 @@ public <T extends TNumber, U extends TNumber> MaxPoolWithArgmax<T, U> maxPoolWit
    * @param options carries optional attributes values
    * @return a new instance of NthElement
    */
-  public <T extends TNumber> NthElement<T> nthElement(Operand<T> input, Operand<TInt32> n,
+  public <T extends Tensor & TNumber> NthElement<T> nthElement(Operand<T> input, Operand<TInt32> n,
       NthElement.Options... options) {
     return NthElement.create(scope, input, n, options);
   }
@@ -1543,7 +1541,7 @@ public <T extends TNumber> NthElement<T> nthElement(Operand<T> input, Operand<TI
    * @param padding The type of padding algorithm to use.
    * @return a new instance of QuantizedAvgPool
    */
-  public <T extends TType> QuantizedAvgPool<T> quantizedAvgPool(Operand<T> input,
+  public <T extends Tensor> QuantizedAvgPool<T> quantizedAvgPool(Operand<T> input,
       Operand<TFloat32> minInput, Operand<TFloat32> maxInput, List<Long> ksize, List<Long> strides,
       String padding) {
     return QuantizedAvgPool.create(scope, input, minInput, maxInput, ksize, strides, padding);
@@ -1584,7 +1582,7 @@ public <T extends TType> QuantizedAvgPool<T> quantizedAvgPool(Operand<T> input,
    *  needs to be multiplied with gamma.
    * @return a new instance of QuantizedBatchNormWithGlobalNormalization
    */
-  public <U extends TType, T extends TType> QuantizedBatchNormWithGlobalNormalization<U> quantizedBatchNormWithGlobalNormalization(
+  public <U extends Tensor, T extends Tensor> QuantizedBatchNormWithGlobalNormalization<U> quantizedBatchNormWithGlobalNormalization(
       Operand<T> t, Operand<TFloat32> tMin, Operand<TFloat32> tMax, Operand<T> m,
       Operand<TFloat32> mMin, Operand<TFloat32> mMax, Operand<T> v, Operand<TFloat32> vMin,
       Operand<TFloat32> vMax, Operand<T> beta, Operand<TFloat32> betaMin, Operand<TFloat32> betaMax,
@@ -1608,7 +1606,7 @@ public <U extends TType, T extends TType> QuantizedBatchNormWithGlobalNormalizat
    * @param outType
    * @return a new instance of QuantizedBiasAdd
    */
-  public <V extends TType, T extends TType, U extends TType> QuantizedBiasAdd<V> quantizedBiasAdd(
+  public <V extends Tensor, T extends Tensor, U extends Tensor> QuantizedBiasAdd<V> quantizedBiasAdd(
       Operand<T> input, Operand<U> bias, Operand<TFloat32> minInput, Operand<TFloat32> maxInput,
       Operand<TFloat32> minBias, Operand<TFloat32> maxBias, DataType<V> outType) {
     return QuantizedBiasAdd.create(scope, input, bias, minInput, maxInput, minBias, maxBias, outType);
@@ -1636,7 +1634,7 @@ public <V extends TType, T extends TType, U extends TType> QuantizedBiasAdd<V> q
    * @param options carries optional attributes values
    * @return a new instance of QuantizedConv2d
    */
-  public <V extends TType, T extends TType, U extends TType> QuantizedConv2d<V> quantizedConv2d(
+  public <V extends Tensor, T extends Tensor, U extends Tensor> QuantizedConv2d<V> quantizedConv2d(
       Operand<T> input, Operand<U> filter, Operand<TFloat32> minInput, Operand<TFloat32> maxInput,
       Operand<TFloat32> minFilter, Operand<TFloat32> maxFilter, DataType<V> outType,
       List<Long> strides, String padding, QuantizedConv2d.Options... options) {
@@ -1653,7 +1651,7 @@ public <V extends TType, T extends TType, U extends TType> QuantizedConv2d<V> qu
    * @param options carries optional attributes values
    * @return a new instance of QuantizedInstanceNorm
    */
-  public <T extends TType> QuantizedInstanceNorm<T> quantizedInstanceNorm(Operand<T> x,
+  public <T extends Tensor> QuantizedInstanceNorm<T> quantizedInstanceNorm(Operand<T> x,
       Operand<TFloat32> xMin, Operand<TFloat32> xMax, QuantizedInstanceNorm.Options... options) {
     return QuantizedInstanceNorm.create(scope, x, xMin, xMax, options);
   }
@@ -1672,7 +1670,7 @@ public <T extends TType> QuantizedInstanceNorm<T> quantizedInstanceNorm(Operand<
    * @param padding The type of padding algorithm to use.
    * @return a new instance of QuantizedMaxPool
    */
-  public <T extends TType> QuantizedMaxPool<T> quantizedMaxPool(Operand<T> input,
+  public <T extends Tensor> QuantizedMaxPool<T> quantizedMaxPool(Operand<T> input,
       Operand<TFloat32> minInput, Operand<TFloat32> maxInput, List<Long> ksize, List<Long> strides,
       String padding) {
     return QuantizedMaxPool.create(scope, input, minInput, maxInput, ksize, strides, padding);
@@ -1688,7 +1686,7 @@ public <T extends TType> QuantizedMaxPool<T> quantizedMaxPool(Operand<T> input,
    * @param outType
    * @return a new instance of QuantizedRelu
    */
-  public <U extends TType, T extends TType> QuantizedRelu<U> quantizedRelu(Operand<T> features,
+  public <U extends Tensor, T extends Tensor> QuantizedRelu<U> quantizedRelu(Operand<T> features,
       Operand<TFloat32> minFeatures, Operand<TFloat32> maxFeatures, DataType<U> outType) {
     return QuantizedRelu.create(scope, features, minFeatures, maxFeatures, outType);
   }
@@ -1703,7 +1701,7 @@ public <U extends TType, T extends TType> QuantizedRelu<U> quantizedRelu(Operand
    * @param outType
    * @return a new instance of QuantizedRelu6
    */
-  public <U extends TType, T extends TType> QuantizedRelu6<U> quantizedRelu6(Operand<T> features,
+  public <U extends Tensor, T extends Tensor> QuantizedRelu6<U> quantizedRelu6(Operand<T> features,
       Operand<TFloat32> minFeatures, Operand<TFloat32> maxFeatures, DataType<U> outType) {
     return QuantizedRelu6.create(scope, features, minFeatures, maxFeatures, outType);
   }
@@ -1719,7 +1717,7 @@ public <U extends TType, T extends TType> QuantizedRelu6<U> quantizedRelu6(Opera
    * @param outType
    * @return a new instance of QuantizedReluX
    */
-  public <U extends TType, T extends TType> QuantizedReluX<U> quantizedReluX(Operand<T> features,
+  public <U extends Tensor, T extends Tensor> QuantizedReluX<U> quantizedReluX(Operand<T> features,
       Operand<TFloat32> maxValue, Operand<TFloat32> minFeatures, Operand<TFloat32> maxFeatures,
       DataType<U> outType) {
     return QuantizedReluX.create(scope, features, maxValue, minFeatures, maxFeatures, outType);
@@ -1737,7 +1735,7 @@ public <U extends TType, T extends TType> QuantizedReluX<U> quantizedReluX(Opera
    * @param features
    * @return a new instance of Relu
    */
-  public <T extends TType> Relu<T> relu(Operand<T> features) {
+  public <T extends Tensor> Relu<T> relu(Operand<T> features) {
     return Relu.create(scope, features);
   }
 
@@ -1748,7 +1746,7 @@ public <T extends TType> Relu<T> relu(Operand<T> features) {
    * @param features
    * @return a new instance of Relu6
    */
-  public <T extends TNumber> Relu6<T> relu6(Operand<T> features) {
+  public <T extends Tensor & TNumber> Relu6<T> relu6(Operand<T> features) {
     return Relu6.create(scope, features);
   }
 
@@ -1767,60 +1765,10 @@ public <T extends TNumber> Relu6<T> relu6(Operand<T> features) {
    * @param features
    * @return a new instance of Selu
    */
-  public <T extends TNumber> Selu<T> selu(Operand<T> features) {
+  public <T extends Tensor & TNumber> Selu<T> selu(Operand<T> features) {
     return Selu.create(scope, features);
   }
 
-  /**
-   * Computes sigmoid cross entropy given <code>logits</code>.
-   *
-   *  <p>Measures the probability error in discrete classification tasks in which each class is
-   *  independent and not mutually exclusive. For instance, one could perform multilabel
-   *  classification where a picture can contain both an elephant and a dog at the same time.
-   *
-   *  <p>For brevity, let <code>x = logits</code>, <code>z = labels</code>. The logistic loss in
-   *  pseudo-code is
-   *
-   *  <pre>
-   *  z * -log(sigmoid(x)) + (1 - z) * -log(1 - sigmoid(x))
-   *   = z * -log(1 / (1 + exp(-x))) + (1 - z) * -log(exp(-x) / (1 + exp(-x)))
-   *   = z * log(1 + exp(-x)) + (1 - z) * (-log(exp(-x)) + log(1 + exp(-x)))
-   *   = z * log(1 + exp(-x)) + (1 - z) * (x + log(1 + exp(-x))
-   *   = (1 - z) * x + log(1 + exp(-x))
-   *   = x - x * z + log(1 + exp(-x))
-   *  </pre>
-   *
-   *  <p>For <code>x < 0</code>, to avoid overflow in <code>exp(-x)</code>, we reformulate the above
-   *
-   *  <pre>
-   *  x - x * z + log(1 + exp(-x))
-   *   = log(exp(x)) - x * z + log(1 + exp(-x))
-   *   = - x * z + log(1 + exp(x))
-   *  </pre>
-   *
-   *  <p>Hence, to ensure stability and avoid overflow, the implementation uses this equivalent
-   *  formulation
-   *
-   *  <pre>
-   *    max(x, 0) - x * z + log(1 + exp(-abs(x)))
-   *  </pre>
-   *
-   *  <p></ode>logits</code> and <code>labels</code> must have the same type and shape.
-   *
-   *  <p>
-   *
-   * @param scope The TensorFlow scope
-   * @param labels the labels
-   * @param logits the logits of type float32 or float64
-   * @param <T> the type of labels and logits
-   * @return the component-wise logistic losses.
-   * @throws IllegalArgumentException if logits' and labels' do not have the same shape
-   */
-  public <T extends TNumber> Operand<T> sigmoidCrossEntropyWithLogits(Operand<T> labels,
-      Operand<T> logits) {
-    return SigmoidCrossEntropyWithLogits.sigmoidCrossEntropyWithLogits(scope, labels, logits);
-  }
-
   /**
    * Computes softmax activations.
    *  <p>
@@ -1832,59 +1780,25 @@ public <T extends TNumber> Operand<T> sigmoidCrossEntropyWithLogits(Operand<T> l
    * @param logits 2-D with shape `[batch_size, num_classes]`.
    * @return a new instance of Softmax
    */
-  public <T extends TNumber> Softmax<T> softmax(Operand<T> logits) {
+  public <T extends Tensor & TNumber> Softmax<T> softmax(Operand<T> logits) {
     return Softmax.create(scope, logits);
   }
 
   /**
-   * Computes softmax cross entropy between <code>logits</code> and <code>labels</code>.
-   *
-   *  <p>Measures the probability error in discrete classification tasks in which the classes are
-   *  mutually exclusive (each entry is in exactly one class). For example, each CIFAR-10 image is
-   *  labeled with one and only one label: an image can be a dog or a truck, but not both.
-   *
-   *  <p><b>NOTE:</b>
-   *
-   *  <p>While the classes are mutually exclusive, their probabilities need not be. All that is
-   *  required is that each row of <code>labels</code> is a valid probability distribution. If they
-   *  are not, the computation of the gradient will be incorrect.
-   *
-   *  <p>If using exclusive <code>labels</code> (wherein one and only one class is true at a time),
-   *  see {@link org.tensorflow.op.NnOps#sparseSoftmaxCrossEntropyWithLogits}
-   *
-   *  <p>Usage:
-   *
-   *  <pre>
-   *    Operand&lt;TFloat32&gt; logits =
-   *        tf.constant(new float[][] {{4.0F, 2.0F, 1.0F}, {0.0F, 5.0F, 1.0F}} );
-   *    Operand&lt;TFloat32&gt; labels =
-   *        tf.constant(new float[][] {{1.0F, 0.0F, 0.0F}, {0.0F, 0.8F, 0.2F}} );
-   *    Operand&lt;TFloat32&gt; output =
-   *        tf.nn.softmaxCrossEntropyWithLogits(labels, logits, -1);
-   *    // output Shape = [2]
-   *    // dataType = FLOAT (1)
-   *    // values { 0.169846, 0.824745 }
-   *  </pre>
-   *
-   *  <p>Backpropagation will happen into both <code>logits</code> and <code>labels</code>. To
-   *  disallow backpropagation into <code>labels</code>, pass label tensors through <code>
-   *  tf.stopGradient</code> before feeding it to this function.
+   * Computes softmax cross entropy cost and gradients to backpropagate.
+   *  <p>
+   *  Inputs are the logits, not probabilities.
    *
-   * @param scope current scope
-   * @param labels Each vector along the class dimension should hold a valid probability
-   *      distribution e.g. for the case in which labels are of shape <code>[batch_size, num_classes]
-   *      </code>, each row of <code>labels[i]</code> must be a valid probability distribution.
-   * @param logits Per-label activations, typically a linear output. These activation energies are
-   *      interpreted as unnormalized log probabilities.
-   * @param axis The class dimension. -1 is the last dimension.
-   * @param <T> the number type of the operands
-   * @return the softmax cross entropy loss. Its type is the same as <code>logits</code> and its
-   *      shape is the same as <code>labels</code> except that it does not have the last dimension of
-   *      <code>labels</code>.
+   * @param <T> data type for {@code loss()} output
+   * @param features batch_size x num_classes matrix
+   * @param labels batch_size x num_classes matrix
+   *  The caller must ensure that each batch of labels represents a valid
+   *  probability distribution.
+   * @return a new instance of SoftmaxCrossEntropyWithLogits
    */
-  public <T extends TNumber, U extends TNumber> Operand<T> softmaxCrossEntropyWithLogits(
-      Operand<U> labels, Operand<T> logits, int axis) {
-    return SoftmaxCrossEntropyWithLogits.softmaxCrossEntropyWithLogits(scope, labels, logits, axis);
+  public <T extends Tensor & TNumber> SoftmaxCrossEntropyWithLogits<T> softmaxCrossEntropyWithLogits(
+      Operand<T> features, Operand<T> labels) {
+    return SoftmaxCrossEntropyWithLogits.create(scope, features, labels);
   }
 
   /**
@@ -1894,7 +1808,7 @@ public <T extends TNumber, U extends TNumber> Operand<T> softmaxCrossEntropyWith
    * @param features
    * @return a new instance of Softsign
    */
-  public <T extends TNumber> Softsign<T> softsign(Operand<T> features) {
+  public <T extends Tensor & TNumber> Softsign<T> softsign(Operand<T> features) {
     return Softsign.create(scope, features);
   }
 
@@ -1983,8 +1897,8 @@ public <T extends TNumber> Softsign<T> softsign(Operand<T> features) {
    * @param blockSize
    * @return a new instance of SpaceToBatch
    */
-  public <T extends TType, U extends TNumber> SpaceToBatch<T> spaceToBatch(Operand<T> input,
-      Operand<U> paddings, Long blockSize) {
+  public <T extends Tensor, U extends Tensor & TNumber> SpaceToBatch<T> spaceToBatch(
+      Operand<T> input, Operand<U> paddings, Long blockSize) {
     return SpaceToBatch.create(scope, input, paddings, blockSize);
   }
 
@@ -2070,57 +1984,30 @@ public <T extends TType, U extends TNumber> SpaceToBatch<T> spaceToBatch(Operand
    * @param options carries optional attributes values
    * @return a new instance of SpaceToDepth
    */
-  public <T extends TType> SpaceToDepth<T> spaceToDepth(Operand<T> input, Long blockSize,
+  public <T extends Tensor> SpaceToDepth<T> spaceToDepth(Operand<T> input, Long blockSize,
       SpaceToDepth.Options... options) {
     return SpaceToDepth.create(scope, input, blockSize, options);
   }
 
   /**
-   * Computes sparse softmax cross entropy between <code>logits</code> and <code>labels</code>.
-   *
-   *  <p>Measures the probability error in discrete classification tasks in which the classes are
-   *  mutually exclusive (each entry is in exactly one class). For example, each CIFAR-10 image is
-   *  labeled with one and only one label: an image can be a dog or a truck, but not both.
-   *
-   *  <p><b>NOTE:</b>
-   *
-   *  <p>For this operation, the probability of a given label is considered exclusive. That is, soft
-   *  classes are not allowed, and the <code>labels</code> vector must provide a single specific
-   *  index for the true class for each row of <code>logits</code> (each minibatch entry). For soft
-   *  softmax classification with a probability distribution for each entry, {@link
-   *  org.tensorflow.op.NnOps#softmaxCrossEntropyWithLogits}.
-   *
-   *  <p><b>WARNING:</b>
-   *
-   *  <p>This op expects unscaled logits, since it performs a <code>softmax</code> on <code>logits
-   *  </code> internally for efficiency. Do not call this op with the output of <code>softmax</code>,
-   *  as it will produce incorrect results.
-   *
-   *  <p>A common use case is to have logits of shape <code>[batchSize, numClasses]</code> and have
-   *  labels of shape <code>[batchSize]</code>, but higher dimensions are supported, in which case
-   *  the <code>dim</code>-th dimension is assumed to be of size <code>numClasses</code>. <code>
-   *  logits</code> must have the <cod>dataType</cod> of <code>TFloat16</code>, <code>TFloat32</code>
-   *  , or <code>TFloat64</code>, and <code>labels</code> must have the dtype of <code>TInt32</code>
-   *  or <code>TInt64</code>.
+   * Computes softmax cross entropy cost and gradients to backpropagate.
+   *  <p>
+   *  Unlike `SoftmaxCrossEntropyWithLogits`, this operation does not accept
+   *  a matrix of label probabilities, but rather a single label per row
+   *  of features.  This label is considered to have probability 1.0 for the
+   *  given row.
+   *  <p>
+   *  Inputs are the logits, not probabilities.
    *
-   * @param scope current scope
-   * @param labels <code>Tensor</code> of shape <code>[d_0, d_1, ..., d_{r-1}]</code> (where <code>r
-   *      </code> is rank of <code>labels</code> and result) and the dataType is <code>TInt32</code>
-   *      or <code>TInt64</code>. Each entry in <code>labels</code> must be an index in <code>[0,
-   *      numClasses)</code>. Other values will raise an exception when this op is run on CPU, and
-   *      return <code>NaN</code> for corresponding loss and gradient rows on GPU.
-   * @param logits Per-label activations (typically a linear output) of shape <code>[d_0, d_1, ...,
-   *      d_{r-1}, numClasses]</code> and dataType of <code>TFloat16</code>, <code>TFloat32</code>,
-   *      or <code>TFloat64</code>. These activation energies are interpreted as unnormalized log
-   *      probabilities.
-   * @return A <code>Tensor</code> of the same shape as <code>labels</code> and of the same type as
-   *      <code>logits</code> with the softmax cross entropy loss.
-   * @throws IllegalArgumentException If logits are scalars (need to have rank >= 1) or if the rank
-   *      of the labels is not equal to the rank of the logits minus one.
+   * @param <T> data type for {@code loss()} output
+   * @param features batch_size x num_classes matrix
+   * @param labels batch_size vector with values in [0, num_classes).
+   *  This is the label for the given minibatch entry.
+   * @return a new instance of SparseSoftmaxCrossEntropyWithLogits
    */
-  public <T extends TNumber, U extends TNumber> Operand sparseSoftmaxCrossEntropyWithLogits(
-      Operand<T> labels, Operand<U> logits) {
-    return SparseSoftmaxCrossEntropyWithLogits.sparseSoftmaxCrossEntropyWithLogits(scope, labels, logits);
+  public <T extends Tensor & TNumber, U extends Tensor & TNumber> SparseSoftmaxCrossEntropyWithLogits<T> sparseSoftmaxCrossEntropyWithLogits(
+      Operand<T> features, Operand<U> labels) {
+    return SparseSoftmaxCrossEntropyWithLogits.create(scope, features, labels);
   }
 
   /**
@@ -2144,7 +2031,7 @@ public <T extends TNumber, U extends TNumber> Operand sparseSoftmaxCrossEntropyW
    * @param options carries optional attributes values
    * @return a new instance of TopK
    */
-  public <T extends TNumber> TopK<T> topK(Operand<T> input, Operand<TInt32> k,
+  public <T extends Tensor & TNumber> TopK<T> topK(Operand<T> input, Operand<TInt32> k,
       TopK.Options... options) {
     return TopK.create(scope, input, k, options);
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/Ops.java b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/Ops.java
index 40f77b75711..fa1b27e62bd 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/Ops.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/Ops.java
@@ -101,6 +101,7 @@
 import org.tensorflow.op.core.InplaceSub;
 import org.tensorflow.op.core.InplaceUpdate;
 import org.tensorflow.op.core.IsVariableInitialized;
+import org.tensorflow.op.core.LinSpace;
 import org.tensorflow.op.core.LookupTableExport;
 import org.tensorflow.op.core.LookupTableFind;
 import org.tensorflow.op.core.LookupTableImport;
@@ -166,8 +167,6 @@
 import org.tensorflow.op.core.ResourceScatterMin;
 import org.tensorflow.op.core.ResourceScatterMul;
 import org.tensorflow.op.core.ResourceScatterNdAdd;
-import org.tensorflow.op.core.ResourceScatterNdMax;
-import org.tensorflow.op.core.ResourceScatterNdMin;
 import org.tensorflow.op.core.ResourceScatterNdSub;
 import org.tensorflow.op.core.ResourceScatterNdUpdate;
 import org.tensorflow.op.core.ResourceScatterSub;
@@ -243,11 +242,7 @@
 import org.tensorflow.op.core.TensorListSetItem;
 import org.tensorflow.op.core.TensorListSplit;
 import org.tensorflow.op.core.TensorListStack;
-import org.tensorflow.op.core.TensorScatterMax;
-import org.tensorflow.op.core.TensorScatterMin;
 import org.tensorflow.op.core.TensorScatterNdAdd;
-import org.tensorflow.op.core.TensorScatterNdMax;
-import org.tensorflow.op.core.TensorScatterNdMin;
 import org.tensorflow.op.core.TensorScatterNdSub;
 import org.tensorflow.op.core.TensorScatterNdUpdate;
 import org.tensorflow.op.core.TensorStridedSliceUpdate;
@@ -266,8 +261,6 @@
 import org.tensorflow.op.core.Variable;
 import org.tensorflow.op.core.VariableShape;
 import org.tensorflow.op.core.Where;
-import org.tensorflow.op.core.XlaSpmdFullToShardShape;
-import org.tensorflow.op.core.XlaSpmdShardToFullShape;
 import org.tensorflow.op.core.Zeros;
 import org.tensorflow.op.core.ZerosLike;
 import org.tensorflow.types.TBool;
@@ -319,8 +312,6 @@ public final class Ops {
 
   public final ImageOps image;
 
-  public final RaggedOps ragged;
-
   public final DataOps data;
 
   public final ShapeOps shape;
@@ -358,7 +349,6 @@ private Ops(Scope scope) {
     nn = new NnOps(scope);
     summary = new SummaryOps(scope);
     image = new ImageOps(scope);
-    ragged = new RaggedOps(scope);
     data = new DataOps(scope);
     shape = new ShapeOps(scope);
     io = new IoOps(scope);
@@ -405,7 +395,7 @@ public Abort abort(Abort.Options... options) {
    * @param options carries optional attributes values
    * @return a new instance of All
    */
-  public <T extends TNumber> All all(Operand<TBool> input, Operand<T> axis,
+  public <T extends Tensor & TNumber> All all(Operand<TBool> input, Operand<T> axis,
       All.Options... options) {
     return All.create(scope, input, axis, options);
   }
@@ -424,7 +414,7 @@ public <T extends TNumber> All all(Operand<TBool> input, Operand<T> axis,
    * @param options carries optional attributes values
    * @return a new instance of Any
    */
-  public <T extends TNumber> Any any(Operand<TBool> input, Operand<T> axis,
+  public <T extends Tensor & TNumber> Any any(Operand<TBool> input, Operand<T> axis,
       Any.Options... options) {
     return Any.create(scope, input, axis, options);
   }
@@ -547,7 +537,7 @@ public AssertThat assertThat(Operand<TBool> condition, Iterable<Operand<?>> data
    * @param options carries optional attributes values
    * @return a new instance of Assign
    */
-  public <T extends TType> Assign<T> assign(Operand<T> ref, Operand<T> value,
+  public <T extends Tensor> Assign<T> assign(Operand<T> ref, Operand<T> value,
       Assign.Options... options) {
     return Assign.create(scope, ref, value, options);
   }
@@ -564,7 +554,7 @@ public <T extends TType> Assign<T> assign(Operand<T> ref, Operand<T> value,
    * @param options carries optional attributes values
    * @return a new instance of AssignAdd
    */
-  public <T extends TType> AssignAdd<T> assignAdd(Operand<T> ref, Operand<T> value,
+  public <T extends Tensor> AssignAdd<T> assignAdd(Operand<T> ref, Operand<T> value,
       AssignAdd.Options... options) {
     return AssignAdd.create(scope, ref, value, options);
   }
@@ -579,7 +569,7 @@ public <T extends TType> AssignAdd<T> assignAdd(Operand<T> ref, Operand<T> value
    * @param value the value by which the variable will be incremented.
    * @return a new instance of AssignAddVariableOp
    */
-  public <T extends TType> AssignAddVariableOp assignAddVariableOp(Operand<?> resource,
+  public <T extends Tensor> AssignAddVariableOp assignAddVariableOp(Operand<?> resource,
       Operand<T> value) {
     return AssignAddVariableOp.create(scope, resource, value);
   }
@@ -596,7 +586,7 @@ public <T extends TType> AssignAddVariableOp assignAddVariableOp(Operand<?> reso
    * @param options carries optional attributes values
    * @return a new instance of AssignSub
    */
-  public <T extends TType> AssignSub<T> assignSub(Operand<T> ref, Operand<T> value,
+  public <T extends Tensor> AssignSub<T> assignSub(Operand<T> ref, Operand<T> value,
       AssignSub.Options... options) {
     return AssignSub.create(scope, ref, value, options);
   }
@@ -611,7 +601,7 @@ public <T extends TType> AssignSub<T> assignSub(Operand<T> ref, Operand<T> value
    * @param value the value by which the variable will be incremented.
    * @return a new instance of AssignSubVariableOp
    */
-  public <T extends TType> AssignSubVariableOp assignSubVariableOp(Operand<?> resource,
+  public <T extends Tensor> AssignSubVariableOp assignSubVariableOp(Operand<?> resource,
       Operand<T> value) {
     return AssignSubVariableOp.create(scope, resource, value);
   }
@@ -626,7 +616,7 @@ public <T extends TType> AssignSubVariableOp assignSubVariableOp(Operand<?> reso
    * @param value the value to set the new tensor to use.
    * @return a new instance of AssignVariableOp
    */
-  public <T extends TType> AssignVariableOp assignVariableOp(Operand<?> resource,
+  public <T extends Tensor> AssignVariableOp assignVariableOp(Operand<?> resource,
       Operand<T> value) {
     return AssignVariableOp.create(scope, resource, value);
   }
@@ -694,7 +684,7 @@ public BarrierIncompleteSize barrierIncompleteSize(Operand<TString> handle) {
    * @param componentIndex The component of the barrier elements that is being assigned.
    * @return a new instance of BarrierInsertMany
    */
-  public <T extends TType> BarrierInsertMany barrierInsertMany(Operand<TString> handle,
+  public <T extends Tensor> BarrierInsertMany barrierInsertMany(Operand<TString> handle,
       Operand<TString> keys, Operand<T> values, Long componentIndex) {
     return BarrierInsertMany.create(scope, handle, keys, values, componentIndex);
   }
@@ -809,8 +799,8 @@ public Batch batch(Iterable<Operand<?>> inTensors, Long numBatchThreads, Long ma
    * @param blockSize
    * @return a new instance of BatchToSpace
    */
-  public <T extends TType, U extends TNumber> BatchToSpace<T> batchToSpace(Operand<T> input,
-      Operand<U> crops, Long blockSize) {
+  public <T extends Tensor, U extends Tensor & TNumber> BatchToSpace<T> batchToSpace(
+      Operand<T> input, Operand<U> crops, Long blockSize) {
     return BatchToSpace.create(scope, input, crops, blockSize);
   }
 
@@ -924,7 +914,7 @@ public <T extends TType, U extends TNumber> BatchToSpace<T> batchToSpace(Operand
    *  }</pre>
    * @return a new instance of BatchToSpaceNd
    */
-  public <T extends TType, U extends TNumber, V extends TNumber> BatchToSpaceNd<T> batchToSpaceNd(
+  public <T extends Tensor, U extends Tensor & TNumber, V extends Tensor & TNumber> BatchToSpaceNd<T> batchToSpaceNd(
       Operand<T> input, Operand<U> blockShape, Operand<V> crops) {
     return BatchToSpaceNd.create(scope, input, blockShape, crops);
   }
@@ -988,7 +978,8 @@ public <T extends TType, U extends TNumber, V extends TNumber> BatchToSpaceNd<T>
    * @param type
    * @return a new instance of Bitcast
    */
-  public <U extends TType, T extends TType> Bitcast<U> bitcast(Operand<T> input, DataType<U> type) {
+  public <U extends Tensor, T extends Tensor> Bitcast<U> bitcast(Operand<T> input,
+      DataType<U> type) {
     return Bitcast.create(scope, input, type);
   }
 
@@ -1003,7 +994,7 @@ public <U extends TType, T extends TType> Bitcast<U> bitcast(Operand<T> input, D
    * @param s1
    * @return a new instance of BroadcastDynamicShape
    */
-  public <T extends TNumber> BroadcastDynamicShape<T> broadcastDynamicShape(Operand<T> s0,
+  public <T extends Tensor & TNumber> BroadcastDynamicShape<T> broadcastDynamicShape(Operand<T> s0,
       Operand<T> s1) {
     return BroadcastDynamicShape.create(scope, s0, s1);
   }
@@ -1029,22 +1020,13 @@ public <T extends TNumber> BroadcastDynamicShape<T> broadcastDynamicShape(Operan
    *  <p>
    *  In the above example, the input Tensor with the shape of `[1, 3]`
    *  is broadcasted to output Tensor with shape of `[3, 3]`.
-   *  <p>
-   *  When doing broadcasted operations such as multiplying a tensor
-   *  by a scalar, broadcasting (usually) confers some time or space
-   *  benefit, as the broadcasted tensor is never materialized.
-   *  <p>
-   *  However, `broadcast_to` does not carry with it any such benefits.
-   *  The newly-created tensor takes the full memory of the broadcasted
-   *  shape. (In a graph context, `broadcast_to` might be fused to
-   *  subsequent operation and then be optimized away, however.)
    *
    * @param <T> data type for {@code output()} output
    * @param input A Tensor to broadcast.
    * @param shape An 1-D `int` Tensor. The shape of the desired output.
    * @return a new instance of BroadcastTo
    */
-  public <T extends TType, U extends TNumber> BroadcastTo<T> broadcastTo(Operand<T> input,
+  public <T extends Tensor, U extends Tensor & TNumber> BroadcastTo<T> broadcastTo(Operand<T> input,
       Operand<U> shape) {
     return BroadcastTo.create(scope, input, shape);
   }
@@ -1067,7 +1049,8 @@ public <T extends TType, U extends TNumber> BroadcastTo<T> broadcastTo(Operand<T
    * @param boundaries A sorted list of floats gives the boundary of the buckets.
    * @return a new instance of Bucketize
    */
-  public <T extends TNumber> Bucketize bucketize(Operand<T> input, List<Float> boundaries) {
+  public <T extends Tensor & TNumber> Bucketize bucketize(Operand<T> input,
+      List<Float> boundaries) {
     return Bucketize.create(scope, input, boundaries);
   }
 
@@ -1087,7 +1070,7 @@ public <T extends TNumber> Bucketize bucketize(Operand<T> input, List<Float> bou
    *  as `t`. The maximum value to clip by.
    * @return a new instance of ClipByValue
    */
-  public <T extends TType> ClipByValue<T> clipByValue(Operand<T> t, Operand<T> clipValueMin,
+  public <T extends Tensor> ClipByValue<T> clipByValue(Operand<T> t, Operand<T> clipValueMin,
       Operand<T> clipValueMax) {
     return ClipByValue.create(scope, t, clipValueMin, clipValueMax);
   }
@@ -1102,8 +1085,8 @@ public <T extends TType> ClipByValue<T> clipByValue(Operand<T> t, Operand<T> cli
    *  range [-rank(values), rank(values)).
    * @return a new instance of Concat
    */
-  public <T extends TType, U extends TNumber> Concat<T> concat(Iterable<Operand<T>> values,
-      Operand<U> axis) {
+  public <T extends Tensor, U extends Tensor & TNumber> Concat<T> concat(
+      Iterable<Operand<T>> values, Operand<U> axis) {
     return Concat.create(scope, values, axis);
   }
 
@@ -1713,7 +1696,7 @@ public Constant<TInt64> constant(Shape shape) {
    * @param tensor a Tensor holding the constant value
    * @return a constant of the same data type as `tensor`
    */
-  public <T extends TType> Constant<T> constant(Tensor<T> tensor) {
+  public <T extends Tensor & TType> Constant<T> constant(T tensor) {
     return Constant.create(scope, tensor);
   }
 
@@ -1872,7 +1855,7 @@ public Constant<TString> constant(Charset charset, Shape shape, DataBuffer<Strin
    * @throws IllegalArgumentException If the tensor datatype or shape is not compatible with the
    *      buffer
    */
-  public <T extends TType & Tensor> Constant<T> constant(DataType<T> type, Shape shape,
+  public <T extends Tensor & TType> Constant<T> constant(DataType<T> type, Shape shape,
       ByteDataBuffer data) {
     return Constant.tensorOf(scope, type, shape, data);
   }
@@ -1915,7 +1898,7 @@ public ControlTrigger controlTrigger() {
    *  'OutOfRange' error.
    * @return a new instance of CountUpTo
    */
-  public <T extends TNumber> CountUpTo<T> countUpTo(Operand<T> ref, Long limit) {
+  public <T extends Tensor & TNumber> CountUpTo<T> countUpTo(Operand<T> ref, Long limit) {
     return CountUpTo.create(scope, ref, limit);
   }
 
@@ -1926,7 +1909,7 @@ public <T extends TNumber> CountUpTo<T> countUpTo(Operand<T> ref, Long limit) {
    * @param x The source tensor of type `T`.
    * @return a new instance of DeepCopy
    */
-  public <T extends TType> DeepCopy<T> deepCopy(Operand<T> x) {
+  public <T extends Tensor> DeepCopy<T> deepCopy(Operand<T> x) {
     return DeepCopy.create(scope, x);
   }
 
@@ -1972,7 +1955,7 @@ public DestroyResourceOp destroyResourceOp(Operand<?> resource,
    *  'TemporaryVariable' op.
    * @return a new instance of DestroyTemporaryVariable
    */
-  public <T extends TType> DestroyTemporaryVariable<T> destroyTemporaryVariable(Operand<T> ref,
+  public <T extends Tensor> DestroyTemporaryVariable<T> destroyTemporaryVariable(Operand<T> ref,
       String varName) {
     return DestroyTemporaryVariable.create(scope, ref, varName);
   }
@@ -2020,7 +2003,7 @@ public <T extends TType> DestroyTemporaryVariable<T> destroyTemporaryVariable(Op
    * @param numPartitions The number of partitions to output.
    * @return a new instance of DynamicPartition
    */
-  public <T extends TType> DynamicPartition<T> dynamicPartition(Operand<T> data,
+  public <T extends Tensor> DynamicPartition<T> dynamicPartition(Operand<T> data,
       Operand<TInt32> partitions, Long numPartitions) {
     return DynamicPartition.create(scope, data, partitions, numPartitions);
   }
@@ -2088,7 +2071,7 @@ public <T extends TType> DynamicPartition<T> dynamicPartition(Operand<T> data,
    * @param data
    * @return a new instance of DynamicStitch
    */
-  public <T extends TType> DynamicStitch<T> dynamicStitch(Iterable<Operand<TInt32>> indices,
+  public <T extends Tensor> DynamicStitch<T> dynamicStitch(Iterable<Operand<TInt32>> indices,
       Iterable<Operand<T>> data) {
     return DynamicStitch.create(scope, indices, data);
   }
@@ -2117,7 +2100,7 @@ public <T extends TType> DynamicStitch<T> dynamicStitch(Iterable<Operand<TInt32>
    * @param options carries optional attributes values
    * @return a new instance of EditDistance
    */
-  public <T extends TType> EditDistance editDistance(Operand<TInt64> hypothesisIndices,
+  public <T extends Tensor> EditDistance editDistance(Operand<TInt64> hypothesisIndices,
       Operand<T> hypothesisValues, Operand<TInt64> hypothesisShape, Operand<TInt64> truthIndices,
       Operand<T> truthValues, Operand<TInt64> truthShape, EditDistance.Options... options) {
     return EditDistance.create(scope, hypothesisIndices, hypothesisValues, hypothesisShape, truthIndices, truthValues, truthShape, options);
@@ -2134,7 +2117,7 @@ public <T extends TType> EditDistance editDistance(Operand<TInt64> hypothesisInd
    * @param options carries optional attributes values
    * @return a new instance of Empty
    */
-  public <T extends TType> Empty<T> empty(Operand<TInt32> shape, DataType<T> dtype,
+  public <T extends Tensor> Empty<T> empty(Operand<TInt32> shape, DataType<T> dtype,
       Empty.Options... options) {
     return Empty.create(scope, shape, dtype, options);
   }
@@ -2154,7 +2137,7 @@ public <T extends TType> Empty<T> empty(Operand<TInt32> shape, DataType<T> dtype
    * @param elementDtype
    * @return a new instance of EmptyTensorList
    */
-  public <T extends TNumber, U extends TType> EmptyTensorList emptyTensorList(
+  public <T extends Tensor & TNumber, U extends Tensor> EmptyTensorList emptyTensorList(
       Operand<T> elementShape, Operand<TInt32> maxNumElements, DataType<U> elementDtype) {
     return EmptyTensorList.create(scope, elementShape, maxNumElements, elementDtype);
   }
@@ -2170,7 +2153,7 @@ public <T extends TNumber, U extends TType> EmptyTensorList emptyTensorList(
    * @param shape The expected (possibly partially specified) shape of the input tensor.
    * @return a new instance of EnsureShape
    */
-  public <T extends TType> EnsureShape<T> ensureShape(Operand<T> input, Shape shape) {
+  public <T extends Tensor> EnsureShape<T> ensureShape(Operand<T> input, Shape shape) {
     return EnsureShape.create(scope, input, shape);
   }
 
@@ -2213,7 +2196,7 @@ public <T extends TType> EnsureShape<T> ensureShape(Operand<T> input, Shape shap
    *  `[-rank(input) - 1, rank(input)]`.
    * @return a new instance of ExpandDims
    */
-  public <T extends TType, U extends TNumber> ExpandDims<T> expandDims(Operand<T> input,
+  public <T extends Tensor, U extends Tensor & TNumber> ExpandDims<T> expandDims(Operand<T> input,
       Operand<U> axis) {
     return ExpandDims.create(scope, input, axis);
   }
@@ -2235,7 +2218,7 @@ public <T extends TType, U extends TNumber> ExpandDims<T> expandDims(Operand<T>
    *  }</pre>
    * @return a new instance of ExtractVolumePatches
    */
-  public <T extends TNumber> ExtractVolumePatches<T> extractVolumePatches(Operand<T> input,
+  public <T extends Tensor & TNumber> ExtractVolumePatches<T> extractVolumePatches(Operand<T> input,
       List<Long> ksizes, List<Long> strides, String padding) {
     return ExtractVolumePatches.create(scope, input, ksizes, strides, padding);
   }
@@ -2274,7 +2257,8 @@ public <T extends TNumber> ExtractVolumePatches<T> extractVolumePatches(Operand<
    * @end_compatibility
    * @return a new instance of Fill
    */
-  public <U extends TType, T extends TNumber> Fill<U> fill(Operand<T> dims, Operand<U> value) {
+  public <U extends Tensor, T extends Tensor & TNumber> Fill<U> fill(Operand<T> dims,
+      Operand<U> value) {
     return Fill.create(scope, dims, value);
   }
 
@@ -2315,7 +2299,7 @@ public <U extends TType, T extends TNumber> Fill<U> fill(Operand<T> dims, Operan
    *  `farmhash::fingerprint64`.
    * @return a new instance of Fingerprint
    */
-  public <T extends TType> Fingerprint fingerprint(Operand<T> data, Operand<TString> method) {
+  public <T extends Tensor> Fingerprint fingerprint(Operand<T> data, Operand<TString> method) {
     return Fingerprint.create(scope, data, method);
   }
 
@@ -2323,8 +2307,8 @@ public <T extends TType> Fingerprint fingerprint(Operand<T> data, Operand<TStrin
    * Gather slices from `params` axis `axis` according to `indices`.
    *  <p>
    *  `indices` must be an integer tensor of any dimension (usually 0-D or 1-D).
-   *  Produces an output tensor with shape `params.shape[:axis] +
-   *  indices.shape[batch_dims:] + params.shape[axis + 1:]` where:
+   *  Produces an output tensor with shape `params.shape[:axis] + indices.shape +
+   *  params.shape[axis + 1:]` where:
    *  <pre>{@code
    *      # Scalar indices (output is rank(params) - 1).
    *      output[a_0, ..., a_n, b_0, ..., b_n] =
@@ -2357,8 +2341,8 @@ public <T extends TType> Fingerprint fingerprint(Operand<T> data, Operand<TStrin
    * @param options carries optional attributes values
    * @return a new instance of Gather
    */
-  public <T extends TType, U extends TNumber, V extends TNumber> Gather<T> gather(Operand<T> params,
-      Operand<U> indices, Operand<V> axis, Gather.Options... options) {
+  public <T extends Tensor, U extends Tensor & TNumber, V extends Tensor & TNumber> Gather<T> gather(
+      Operand<T> params, Operand<U> indices, Operand<V> axis, Gather.Options... options) {
     return Gather.create(scope, params, indices, axis, options);
   }
 
@@ -2463,7 +2447,7 @@ public <T extends TType, U extends TNumber, V extends TNumber> Gather<T> gather(
    * @param indices Index tensor.
    * @return a new instance of GatherNd
    */
-  public <T extends TType, U extends TNumber> GatherNd<T> gatherNd(Operand<T> params,
+  public <T extends Tensor, U extends Tensor & TNumber> GatherNd<T> gatherNd(Operand<T> params,
       Operand<U> indices) {
     return GatherNd.create(scope, params, indices);
   }
@@ -2474,7 +2458,7 @@ public <T extends TType, U extends TNumber> GatherNd<T> gatherNd(Operand<T> para
    * @param value The tensor to be stored.
    * @return a new instance of GetSessionHandle
    */
-  public <T extends TType> GetSessionHandle getSessionHandle(Operand<T> value) {
+  public <T extends Tensor> GetSessionHandle getSessionHandle(Operand<T> value) {
     return GetSessionHandle.create(scope, value);
   }
 
@@ -2486,7 +2470,7 @@ public <T extends TType> GetSessionHandle getSessionHandle(Operand<T> value) {
    * @param dtype The type of the output value.
    * @return a new instance of GetSessionTensor
    */
-  public <T extends TType> GetSessionTensor<T> getSessionTensor(Operand<TString> handle,
+  public <T extends Tensor> GetSessionTensor<T> getSessionTensor(Operand<TString> handle,
       DataType<T> dtype) {
     return GetSessionTensor.create(scope, handle, dtype);
   }
@@ -2551,7 +2535,7 @@ public Gradients gradients(Operand<?> y, Iterable<? extends Operand<?>> x,
    * @param input
    * @return a new instance of GuaranteeConst
    */
-  public <T extends TType> GuaranteeConst<T> guaranteeConst(Operand<T> input) {
+  public <T extends Tensor> GuaranteeConst<T> guaranteeConst(Operand<T> input) {
     return GuaranteeConst.create(scope, input);
   }
 
@@ -2567,7 +2551,7 @@ public <T extends TType> GuaranteeConst<T> guaranteeConst(Operand<T> input) {
    * @param options carries optional attributes values
    * @return a new instance of HashTable
    */
-  public <T extends TType, U extends TType> HashTable hashTable(DataType<T> keyDtype,
+  public <T extends Tensor, U extends Tensor> HashTable hashTable(DataType<T> keyDtype,
       DataType<U> valueDtype, HashTable.Options... options) {
     return HashTable.create(scope, keyDtype, valueDtype, options);
   }
@@ -2598,8 +2582,8 @@ public <T extends TType, U extends TType> HashTable hashTable(DataType<T> keyDty
    * @param nbins Scalar `int32 Tensor`.  Number of histogram bins.
    * @return a new instance of HistogramFixedWidth
    */
-  public <T extends TNumber> HistogramFixedWidth<TInt32> histogramFixedWidth(Operand<T> values,
-      Operand<T> valueRange, Operand<TInt32> nbins) {
+  public <T extends Tensor & TNumber> HistogramFixedWidth<TInt32> histogramFixedWidth(
+      Operand<T> values, Operand<T> valueRange, Operand<TInt32> nbins) {
     return HistogramFixedWidth.create(scope, values, valueRange, nbins);
   }
 
@@ -2630,7 +2614,7 @@ public <T extends TNumber> HistogramFixedWidth<TInt32> histogramFixedWidth(Opera
    * @param dtype
    * @return a new instance of HistogramFixedWidth
    */
-  public <U extends TNumber, T extends TNumber> HistogramFixedWidth<U> histogramFixedWidth(
+  public <U extends Tensor & TNumber, T extends Tensor & TNumber> HistogramFixedWidth<U> histogramFixedWidth(
       Operand<T> values, Operand<T> valueRange, Operand<TInt32> nbins, DataType<U> dtype) {
     return HistogramFixedWidth.create(scope, values, valueRange, nbins, dtype);
   }
@@ -2642,7 +2626,7 @@ public <U extends TNumber, T extends TNumber> HistogramFixedWidth<U> histogramFi
    * @param input
    * @return a new instance of Identity
    */
-  public <T extends TType> Identity<T> identity(Operand<T> input) {
+  public <T extends Tensor> Identity<T> identity(Operand<T> input) {
     return Identity.create(scope, input);
   }
 
@@ -2681,7 +2665,7 @@ public IdentityN identityN(Iterable<Operand<?>> input) {
    *  NewReadOnlyMemoryRegionFromFile in tensorflow::Env.
    * @return a new instance of ImmutableConst
    */
-  public <T extends TType> ImmutableConst<T> immutableConst(DataType<T> dtype, Shape shape,
+  public <T extends Tensor> ImmutableConst<T> immutableConst(DataType<T> dtype, Shape shape,
       String memoryRegionName) {
     return ImmutableConst.create(scope, dtype, shape, memoryRegionName);
   }
@@ -2771,8 +2755,8 @@ public void initAdd(Op initializer) {
    * @param values Values of type Tval.
    * @return a new instance of InitializeTable
    */
-  public <T extends TType, U extends TType> InitializeTable initializeTable(Operand<?> tableHandle,
-      Operand<T> keys, Operand<U> values) {
+  public <T extends Tensor, U extends Tensor> InitializeTable initializeTable(
+      Operand<?> tableHandle, Operand<T> keys, Operand<U> values) {
     return InitializeTable.create(scope, tableHandle, keys, values);
   }
 
@@ -2815,7 +2799,8 @@ public InitializeTableFromTextFile initializeTableFromTextFile(Operand<?> tableH
    * @param v A `Tensor` of type T. Same dimension sizes as x except the first dimension, which must be the same as i's size.
    * @return a new instance of InplaceAdd
    */
-  public <T extends TType> InplaceAdd<T> inplaceAdd(Operand<T> x, Operand<TInt32> i, Operand<T> v) {
+  public <T extends Tensor> InplaceAdd<T> inplaceAdd(Operand<T> x, Operand<TInt32> i,
+      Operand<T> v) {
     return InplaceAdd.create(scope, x, i, v);
   }
 
@@ -2830,17 +2815,15 @@ public <T extends TType> InplaceAdd<T> inplaceAdd(Operand<T> x, Operand<TInt32>
    * @param v A `Tensor` of type T. Same dimension sizes as x except the first dimension, which must be the same as i's size.
    * @return a new instance of InplaceSub
    */
-  public <T extends TType> InplaceSub<T> inplaceSub(Operand<T> x, Operand<TInt32> i, Operand<T> v) {
+  public <T extends Tensor> InplaceSub<T> inplaceSub(Operand<T> x, Operand<TInt32> i,
+      Operand<T> v) {
     return InplaceSub.create(scope, x, i, v);
   }
 
   /**
-   * Updates specified rows 'i' with values 'v'.
+   *     Updates specified rows with values in `v`.
    *  <p>
-   *  Computes `x[i, :] = v; return x`.
-   *  <p>
-   *  Originally this function is mutative however for compilation we make this
-   *  operation create / operate on a copy of `x`.
+   *      Computes `x[i, :] = v; return x`.
    *
    * @param <T> data type for {@code y()} output
    * @param x A tensor of type `T`.
@@ -2848,7 +2831,7 @@ public <T extends TType> InplaceSub<T> inplaceSub(Operand<T> x, Operand<TInt32>
    * @param v A `Tensor` of type T. Same dimension sizes as x except the first dimension, which must be the same as i's size.
    * @return a new instance of InplaceUpdate
    */
-  public <T extends TType> InplaceUpdate<T> inplaceUpdate(Operand<T> x, Operand<TInt32> i,
+  public <T extends Tensor> InplaceUpdate<T> inplaceUpdate(Operand<T> x, Operand<TInt32> i,
       Operand<T> v) {
     return InplaceUpdate.create(scope, x, i, v);
   }
@@ -2861,10 +2844,33 @@ public <T extends TType> InplaceUpdate<T> inplaceUpdate(Operand<T> x, Operand<TI
    * @param ref Should be from a `Variable` node. May be uninitialized.
    * @return a new instance of IsVariableInitialized
    */
-  public <T extends TType> IsVariableInitialized isVariableInitialized(Operand<T> ref) {
+  public <T extends Tensor> IsVariableInitialized isVariableInitialized(Operand<T> ref) {
     return IsVariableInitialized.create(scope, ref);
   }
 
+  /**
+   * Generates values in an interval.
+   *  <p>
+   *  A sequence of `num` evenly-spaced values are generated beginning at `start`.
+   *  If `num > 1`, the values in the sequence increase by `stop - start / num - 1`,
+   *  so that the last one is exactly `stop`.
+   *  <p>
+   *  For example:
+   *  <pre>{@code
+   *  tf.linspace(10.0, 12.0, 3, name="linspace") => [ 10.0  11.0  12.0]
+   *  }</pre>
+   *
+   * @param <T> data type for {@code output()} output
+   * @param start 0-D tensor. First entry in the range.
+   * @param stop 0-D tensor. Last entry in the range.
+   * @param num 0-D tensor. Number of values to generate.
+   * @return a new instance of LinSpace
+   */
+  public <T extends Tensor & TNumber, U extends Tensor & TNumber> LinSpace<T> linSpace(
+      Operand<T> start, Operand<T> stop, Operand<U> num) {
+    return LinSpace.create(scope, start, stop, num);
+  }
+
   /**
    * Outputs all keys and values in the table.
    *
@@ -2875,7 +2881,7 @@ public <T extends TType> IsVariableInitialized isVariableInitialized(Operand<T>
    * @param Tvalues
    * @return a new instance of LookupTableExport
    */
-  public <T extends TType, U extends TType> LookupTableExport<T, U> lookupTableExport(
+  public <T extends Tensor, U extends Tensor> LookupTableExport<T, U> lookupTableExport(
       Operand<?> tableHandle, DataType<T> Tkeys, DataType<U> Tvalues) {
     return LookupTableExport.create(scope, tableHandle, Tkeys, Tvalues);
   }
@@ -2895,7 +2901,7 @@ public <T extends TType, U extends TType> LookupTableExport<T, U> lookupTableExp
    * @param defaultValue
    * @return a new instance of LookupTableFind
    */
-  public <U extends TType, T extends TType> LookupTableFind<U> lookupTableFind(
+  public <U extends Tensor, T extends Tensor> LookupTableFind<U> lookupTableFind(
       Operand<?> tableHandle, Operand<T> keys, Operand<U> defaultValue) {
     return LookupTableFind.create(scope, tableHandle, keys, defaultValue);
   }
@@ -2911,7 +2917,7 @@ public <U extends TType, T extends TType> LookupTableFind<U> lookupTableFind(
    * @param values Values to associate with keys.
    * @return a new instance of LookupTableImport
    */
-  public <T extends TType, U extends TType> LookupTableImport lookupTableImport(
+  public <T extends Tensor, U extends Tensor> LookupTableImport lookupTableImport(
       Operand<?> tableHandle, Operand<T> keys, Operand<U> values) {
     return LookupTableImport.create(scope, tableHandle, keys, values);
   }
@@ -2927,7 +2933,7 @@ public <T extends TType, U extends TType> LookupTableImport lookupTableImport(
    * @param values Values to associate with keys.
    * @return a new instance of LookupTableInsert
    */
-  public <T extends TType, U extends TType> LookupTableInsert lookupTableInsert(
+  public <T extends Tensor, U extends Tensor> LookupTableInsert lookupTableInsert(
       Operand<?> tableHandle, Operand<T> keys, Operand<U> values) {
     return LookupTableInsert.create(scope, tableHandle, keys, values);
   }
@@ -3070,8 +3076,8 @@ public MapUnstageNoKey mapUnstageNoKey(Operand<TInt32> indices, List<DataType<?>
    * @param options carries optional attributes values
    * @return a new instance of Max
    */
-  public <T extends TType, U extends TNumber> Max<T> max(Operand<T> input, Operand<U> axis,
-      Max.Options... options) {
+  public <T extends Tensor, U extends Tensor & TNumber> Max<T> max(Operand<T> input,
+      Operand<U> axis, Max.Options... options) {
     return Max.create(scope, input, axis, options);
   }
 
@@ -3088,7 +3094,7 @@ public <T extends TType, U extends TNumber> Max<T> max(Operand<T> input, Operand
    * @param inputs The input tensors, exactly one of which will become available.
    * @return a new instance of Merge
    */
-  public <T extends TType> Merge<T> merge(Iterable<Operand<T>> inputs) {
+  public <T extends Tensor> Merge<T> merge(Iterable<Operand<T>> inputs) {
     return Merge.create(scope, inputs);
   }
 
@@ -3107,8 +3113,8 @@ public <T extends TType> Merge<T> merge(Iterable<Operand<T>> inputs) {
    * @param options carries optional attributes values
    * @return a new instance of Min
    */
-  public <T extends TType, U extends TNumber> Min<T> min(Operand<T> input, Operand<U> axis,
-      Min.Options... options) {
+  public <T extends Tensor, U extends Tensor & TNumber> Min<T> min(Operand<T> input,
+      Operand<U> axis, Min.Options... options) {
     return Min.create(scope, input, axis, options);
   }
 
@@ -3151,7 +3157,7 @@ public <T extends TType, U extends TNumber> Min<T> min(Operand<T> input, Operand
    *  it is `[1, 2, 3, 3, 2]` in symmetric mode.
    * @return a new instance of MirrorPad
    */
-  public <T extends TType, U extends TNumber> MirrorPad<T> mirrorPad(Operand<T> input,
+  public <T extends Tensor, U extends Tensor & TNumber> MirrorPad<T> mirrorPad(Operand<T> input,
       Operand<U> paddings, String mode) {
     return MirrorPad.create(scope, input, paddings, mode);
   }
@@ -3182,7 +3188,7 @@ public <T extends TType, U extends TNumber> MirrorPad<T> mirrorPad(Operand<T> in
    *  '''
    *
    * @tf.function def foo(x, y):
-   *    return mlir_passthrough_op([x, y], mlir_module, Toutputs=[tf.float32])
+   *    return = mlir_passthrough_op([x, y], mlir_module, Toutputs=[tf.float32])
    *
    *  graph_def = foo.get_concrete_function(tf.TensorSpec([10], tf.float32), tf.TensorSpec([10], tf.float32)).graph.as_graph_def()
    *  }</pre>
@@ -3213,7 +3219,7 @@ public MlirPassthroughOp mlirPassthroughOp(Iterable<Operand<?>> inputs, String m
    * @param options carries optional attributes values
    * @return a new instance of MutableDenseHashTable
    */
-  public <T extends TType, U extends TType> MutableDenseHashTable mutableDenseHashTable(
+  public <T extends Tensor, U extends Tensor> MutableDenseHashTable mutableDenseHashTable(
       Operand<T> emptyKey, Operand<T> deletedKey, DataType<U> valueDtype,
       MutableDenseHashTable.Options... options) {
     return MutableDenseHashTable.create(scope, emptyKey, deletedKey, valueDtype, options);
@@ -3231,8 +3237,8 @@ public <T extends TType, U extends TType> MutableDenseHashTable mutableDenseHash
    * @param options carries optional attributes values
    * @return a new instance of MutableHashTable
    */
-  public <T extends TType, U extends TType> MutableHashTable mutableHashTable(DataType<T> keyDtype,
-      DataType<U> valueDtype, MutableHashTable.Options... options) {
+  public <T extends Tensor, U extends Tensor> MutableHashTable mutableHashTable(
+      DataType<T> keyDtype, DataType<U> valueDtype, MutableHashTable.Options... options) {
     return MutableHashTable.create(scope, keyDtype, valueDtype, options);
   }
 
@@ -3248,7 +3254,7 @@ public <T extends TType, U extends TType> MutableHashTable mutableHashTable(Data
    * @param options carries optional attributes values
    * @return a new instance of MutableHashTableOfTensors
    */
-  public <T extends TType, U extends TType> MutableHashTableOfTensors mutableHashTableOfTensors(
+  public <T extends Tensor, U extends Tensor> MutableHashTableOfTensors mutableHashTableOfTensors(
       DataType<T> keyDtype, DataType<U> valueDtype, MutableHashTableOfTensors.Options... options) {
     return MutableHashTableOfTensors.create(scope, keyDtype, valueDtype, options);
   }
@@ -3316,7 +3322,7 @@ public MutexLock mutexLock(Operand<?> mutex) {
    * @param data The tensor to be made available to the next iteration.
    * @return a new instance of NextIteration
    */
-  public <T extends TType> NextIteration<T> nextIteration(Operand<T> data) {
+  public <T extends Tensor> NextIteration<T> nextIteration(Operand<T> data) {
     return NextIteration.create(scope, data);
   }
 
@@ -3420,7 +3426,7 @@ public NoOp noOp() {
    * @param options carries optional attributes values
    * @return a new instance of OneHot
    */
-  public <U extends TType, T extends TNumber> OneHot<U> oneHot(Operand<T> indices,
+  public <U extends Tensor, T extends Tensor & TNumber> OneHot<U> oneHot(Operand<T> indices,
       Operand<TInt32> depth, Operand<U> onValue, Operand<U> offValue, OneHot.Options... options) {
     return OneHot.create(scope, indices, depth, onValue, offValue, options);
   }
@@ -3432,7 +3438,7 @@ public <U extends TType, T extends TNumber> OneHot<U> oneHot(Operand<T> indices,
    * @param x a tensor of type T.
    * @return a new instance of OnesLike
    */
-  public <T extends TType> OnesLike<T> onesLike(Operand<T> x) {
+  public <T extends Tensor> OnesLike<T> onesLike(Operand<T> x) {
     return OnesLike.create(scope, x);
   }
 
@@ -3574,8 +3580,8 @@ public OrderedMapUnstageNoKey orderedMapUnstageNoKey(Operand<TInt32> indices,
    * @param constantValues
    * @return a new instance of Pad
    */
-  public <T extends TType, U extends TNumber> Pad<T> pad(Operand<T> input, Operand<U> paddings,
-      Operand<T> constantValues) {
+  public <T extends Tensor, U extends Tensor & TNumber> Pad<T> pad(Operand<T> input,
+      Operand<U> paddings, Operand<T> constantValues) {
     return Pad.create(scope, input, paddings, constantValues);
   }
 
@@ -3604,7 +3610,7 @@ public <T extends TType, U extends TNumber> Pad<T> pad(Operand<T> input, Operand
    *  but with the number of input values in the first dimension.
    * @return a new instance of ParallelConcat
    */
-  public <T extends TType> ParallelConcat<T> parallelConcat(Iterable<Operand<T>> values,
+  public <T extends Tensor> ParallelConcat<T> parallelConcat(Iterable<Operand<T>> values,
       Shape shape) {
     return ParallelConcat.create(scope, values, shape);
   }
@@ -3671,7 +3677,7 @@ public <T extends TType> ParallelConcat<T> parallelConcat(Iterable<Operand<T>> v
    * @param data
    * @return a new instance of ParallelDynamicStitch
    */
-  public <T extends TType> ParallelDynamicStitch<T> parallelDynamicStitch(
+  public <T extends Tensor> ParallelDynamicStitch<T> parallelDynamicStitch(
       Iterable<Operand<TInt32>> indices, Iterable<Operand<T>> data) {
     return ParallelDynamicStitch.create(scope, indices, data);
   }
@@ -3688,7 +3694,7 @@ public <T extends TType> ParallelDynamicStitch<T> parallelDynamicStitch(
    * @param options carries optional attributes values
    * @return a new instance of Placeholder
    */
-  public <T extends TType> Placeholder<T> placeholder(DataType<T> dtype,
+  public <T extends Tensor> Placeholder<T> placeholder(DataType<T> dtype,
       Placeholder.Options... options) {
     return Placeholder.create(scope, dtype, options);
   }
@@ -3701,7 +3707,7 @@ public <T extends TType> Placeholder<T> placeholder(DataType<T> dtype,
    * @param shape The (possibly partial) shape of the tensor.
    * @return a new instance of PlaceholderWithDefault
    */
-  public <T extends TType> PlaceholderWithDefault<T> placeholderWithDefault(Operand<T> input,
+  public <T extends Tensor> PlaceholderWithDefault<T> placeholderWithDefault(Operand<T> input,
       Shape shape) {
     return PlaceholderWithDefault.create(scope, input, shape);
   }
@@ -3734,8 +3740,8 @@ public Print print(Operand<TString> input, Print.Options... options) {
    * @param options carries optional attributes values
    * @return a new instance of Prod
    */
-  public <T extends TType, U extends TNumber> Prod<T> prod(Operand<T> input, Operand<U> axis,
-      Prod.Options... options) {
+  public <T extends Tensor, U extends Tensor & TNumber> Prod<T> prod(Operand<T> input,
+      Operand<U> axis, Prod.Options... options) {
     return Prod.create(scope, input, axis, options);
   }
 
@@ -3751,7 +3757,7 @@ public <T extends TType, U extends TNumber> Prod<T> prod(Operand<T> input, Opera
    * @param inputMax The maximum value of the input.
    * @return a new instance of QuantizedReshape
    */
-  public <T extends TType, U extends TNumber> QuantizedReshape<T> quantizedReshape(
+  public <T extends Tensor, U extends Tensor & TNumber> QuantizedReshape<T> quantizedReshape(
       Operand<T> tensor, Operand<U> shape, Operand<TFloat32> inputMin, Operand<TFloat32> inputMax) {
     return QuantizedReshape.create(scope, tensor, shape, inputMin, inputMax);
   }
@@ -3776,7 +3782,8 @@ public <T extends TType, U extends TNumber> QuantizedReshape<T> quantizedReshape
    * @param delta 0-D (scalar). Optional. Default is 1. Number that increments `start`.
    * @return a new instance of Range
    */
-  public <T extends TNumber> Range<T> range(Operand<T> start, Operand<T> limit, Operand<T> delta) {
+  public <T extends Tensor & TNumber> Range<T> range(Operand<T> start, Operand<T> limit,
+      Operand<T> delta) {
     return Range.create(scope, start, limit, delta);
   }
 
@@ -3798,7 +3805,7 @@ public <T extends TNumber> Range<T> range(Operand<T> start, Operand<T> limit, Op
    * @param input
    * @return a new instance of Rank
    */
-  public <T extends TType> Rank rank(Operand<T> input) {
+  public <T extends Tensor> Rank rank(Operand<T> input) {
     return Rank.create(scope, input);
   }
 
@@ -3817,7 +3824,7 @@ public <T extends TType> Rank rank(Operand<T> input) {
    * @param dtype the dtype of the value.
    * @return a new instance of ReadVariableOp
    */
-  public <T extends TType> ReadVariableOp<T> readVariableOp(Operand<?> resource,
+  public <T extends Tensor> ReadVariableOp<T> readVariableOp(Operand<?> resource,
       DataType<T> dtype) {
     return ReadVariableOp.create(scope, resource, dtype);
   }
@@ -3836,7 +3843,7 @@ public <T extends TType> ReadVariableOp<T> readVariableOp(Operand<?> resource,
    * @param options carries optional attributes values
    * @return a new instance of ReduceAll
    */
-  public <T extends TNumber> ReduceAll reduceAll(Operand<TBool> input, Operand<T> axis,
+  public <T extends Tensor & TNumber> ReduceAll reduceAll(Operand<TBool> input, Operand<T> axis,
       ReduceAll.Options... options) {
     return ReduceAll.create(scope, input, axis, options);
   }
@@ -3855,7 +3862,7 @@ public <T extends TNumber> ReduceAll reduceAll(Operand<TBool> input, Operand<T>
    * @param options carries optional attributes values
    * @return a new instance of ReduceAny
    */
-  public <T extends TNumber> ReduceAny reduceAny(Operand<TBool> input, Operand<T> axis,
+  public <T extends Tensor & TNumber> ReduceAny reduceAny(Operand<TBool> input, Operand<T> axis,
       ReduceAny.Options... options) {
     return ReduceAny.create(scope, input, axis, options);
   }
@@ -3875,7 +3882,7 @@ public <T extends TNumber> ReduceAny reduceAny(Operand<TBool> input, Operand<T>
    * @param options carries optional attributes values
    * @return a new instance of ReduceMax
    */
-  public <T extends TType, U extends TNumber> ReduceMax<T> reduceMax(Operand<T> input,
+  public <T extends Tensor, U extends Tensor & TNumber> ReduceMax<T> reduceMax(Operand<T> input,
       Operand<U> axis, ReduceMax.Options... options) {
     return ReduceMax.create(scope, input, axis, options);
   }
@@ -3895,7 +3902,7 @@ public <T extends TType, U extends TNumber> ReduceMax<T> reduceMax(Operand<T> in
    * @param options carries optional attributes values
    * @return a new instance of ReduceMin
    */
-  public <T extends TType, U extends TNumber> ReduceMin<T> reduceMin(Operand<T> input,
+  public <T extends Tensor, U extends Tensor & TNumber> ReduceMin<T> reduceMin(Operand<T> input,
       Operand<U> axis, ReduceMin.Options... options) {
     return ReduceMin.create(scope, input, axis, options);
   }
@@ -3915,7 +3922,7 @@ public <T extends TType, U extends TNumber> ReduceMin<T> reduceMin(Operand<T> in
    * @param options carries optional attributes values
    * @return a new instance of ReduceProd
    */
-  public <T extends TType, U extends TNumber> ReduceProd<T> reduceProd(Operand<T> input,
+  public <T extends Tensor, U extends Tensor & TNumber> ReduceProd<T> reduceProd(Operand<T> input,
       Operand<U> axis, ReduceProd.Options... options) {
     return ReduceProd.create(scope, input, axis, options);
   }
@@ -3935,7 +3942,7 @@ public <T extends TType, U extends TNumber> ReduceProd<T> reduceProd(Operand<T>
    * @param options carries optional attributes values
    * @return a new instance of ReduceSum
    */
-  public <T extends TType, U extends TNumber> ReduceSum<T> reduceSum(Operand<T> input,
+  public <T extends Tensor, U extends Tensor & TNumber> ReduceSum<T> reduceSum(Operand<T> input,
       Operand<U> axis, ReduceSum.Options... options) {
     return ReduceSum.create(scope, input, axis, options);
   }
@@ -3947,7 +3954,7 @@ public <T extends TType, U extends TNumber> ReduceSum<T> reduceSum(Operand<T> in
    * @param data The tensor to be made available to the next iteration.
    * @return a new instance of RefNextIteration
    */
-  public <T extends TType> RefNextIteration<T> refNextIteration(Operand<T> data) {
+  public <T extends Tensor> RefNextIteration<T> refNextIteration(Operand<T> data) {
     return RefNextIteration.create(scope, data);
   }
 
@@ -3959,7 +3966,7 @@ public <T extends TType> RefNextIteration<T> refNextIteration(Operand<T> data) {
    * @param inputs A list of ref tensors, one of which will be forwarded to `output`.
    * @return a new instance of RefSelect
    */
-  public <T extends TType> RefSelect<T> refSelect(Operand<TInt32> index,
+  public <T extends Tensor> RefSelect<T> refSelect(Operand<TInt32> index,
       Iterable<Operand<T>> inputs) {
     return RefSelect.create(scope, index, inputs);
   }
@@ -3977,7 +3984,7 @@ public <T extends TType> RefSelect<T> refSelect(Operand<TInt32> index,
    * @param pred A scalar that specifies which output port will receive data.
    * @return a new instance of RefSwitch
    */
-  public <T extends TType> RefSwitch<T> refSwitch(Operand<T> data, Operand<TBool> pred) {
+  public <T extends Tensor> RefSwitch<T> refSwitch(Operand<T> data, Operand<TBool> pred) {
     return RefSwitch.create(scope, data, pred);
   }
 
@@ -4070,7 +4077,7 @@ public RemoteFusedGraphExecute remoteFusedGraphExecute(Iterable<Operand<?>> inpu
    * @param shape Defines the shape of the output tensor.
    * @return a new instance of Reshape
    */
-  public <T extends TType, U extends TNumber> Reshape<T> reshape(Operand<T> tensor,
+  public <T extends Tensor, U extends Tensor & TNumber> Reshape<T> reshape(Operand<T> tensor,
       Operand<U> shape) {
     return Reshape.create(scope, tensor, shape);
   }
@@ -4085,8 +4092,8 @@ public <T extends TType, U extends TNumber> Reshape<T> reshape(Operand<T> tensor
    * @param T
    * @return a new instance of ResourceCountUpTo
    */
-  public <T extends TNumber> ResourceCountUpTo<T> resourceCountUpTo(Operand<?> resource, Long limit,
-      DataType<T> T) {
+  public <T extends Tensor & TNumber> ResourceCountUpTo<T> resourceCountUpTo(Operand<?> resource,
+      Long limit, DataType<T> T) {
     return ResourceCountUpTo.create(scope, resource, limit, T);
   }
 
@@ -4113,8 +4120,9 @@ public <T extends TNumber> ResourceCountUpTo<T> resourceCountUpTo(Operand<?> res
    * @param options carries optional attributes values
    * @return a new instance of ResourceGather
    */
-  public <U extends TType, T extends TNumber> ResourceGather<U> resourceGather(Operand<?> resource,
-      Operand<T> indices, DataType<U> dtype, ResourceGather.Options... options) {
+  public <U extends Tensor, T extends Tensor & TNumber> ResourceGather<U> resourceGather(
+      Operand<?> resource, Operand<T> indices, DataType<U> dtype,
+      ResourceGather.Options... options) {
     return ResourceGather.create(scope, resource, indices, dtype, options);
   }
 
@@ -4126,7 +4134,7 @@ public <U extends TType, T extends TNumber> ResourceGather<U> resourceGather(Ope
    * @param dtype
    * @return a new instance of ResourceGatherNd
    */
-  public <U extends TType, T extends TNumber> ResourceGatherNd<U> resourceGatherNd(
+  public <U extends Tensor, T extends Tensor & TNumber> ResourceGatherNd<U> resourceGatherNd(
       Operand<?> resource, Operand<T> indices, DataType<U> dtype) {
     return ResourceGatherNd.create(scope, resource, indices, dtype);
   }
@@ -4159,7 +4167,7 @@ public <U extends TType, T extends TNumber> ResourceGatherNd<U> resourceGatherNd
    * @param updates A tensor of updated values to add to `ref`.
    * @return a new instance of ResourceScatterAdd
    */
-  public <T extends TNumber, U extends TType> ResourceScatterAdd resourceScatterAdd(
+  public <T extends Tensor & TNumber, U extends Tensor> ResourceScatterAdd resourceScatterAdd(
       Operand<?> resource, Operand<T> indices, Operand<U> updates) {
     return ResourceScatterAdd.create(scope, resource, indices, updates);
   }
@@ -4192,7 +4200,7 @@ public <T extends TNumber, U extends TType> ResourceScatterAdd resourceScatterAd
    * @param updates A tensor of updated values to add to `ref`.
    * @return a new instance of ResourceScatterDiv
    */
-  public <T extends TNumber, U extends TType> ResourceScatterDiv resourceScatterDiv(
+  public <T extends Tensor & TNumber, U extends Tensor> ResourceScatterDiv resourceScatterDiv(
       Operand<?> resource, Operand<T> indices, Operand<U> updates) {
     return ResourceScatterDiv.create(scope, resource, indices, updates);
   }
@@ -4225,7 +4233,7 @@ public <T extends TNumber, U extends TType> ResourceScatterDiv resourceScatterDi
    * @param updates A tensor of updated values to add to `ref`.
    * @return a new instance of ResourceScatterMax
    */
-  public <T extends TNumber, U extends TType> ResourceScatterMax resourceScatterMax(
+  public <T extends Tensor & TNumber, U extends Tensor> ResourceScatterMax resourceScatterMax(
       Operand<?> resource, Operand<T> indices, Operand<U> updates) {
     return ResourceScatterMax.create(scope, resource, indices, updates);
   }
@@ -4258,7 +4266,7 @@ public <T extends TNumber, U extends TType> ResourceScatterMax resourceScatterMa
    * @param updates A tensor of updated values to add to `ref`.
    * @return a new instance of ResourceScatterMin
    */
-  public <T extends TNumber, U extends TType> ResourceScatterMin resourceScatterMin(
+  public <T extends Tensor & TNumber, U extends Tensor> ResourceScatterMin resourceScatterMin(
       Operand<?> resource, Operand<T> indices, Operand<U> updates) {
     return ResourceScatterMin.create(scope, resource, indices, updates);
   }
@@ -4291,7 +4299,7 @@ public <T extends TNumber, U extends TType> ResourceScatterMin resourceScatterMi
    * @param updates A tensor of updated values to add to `ref`.
    * @return a new instance of ResourceScatterMul
    */
-  public <T extends TNumber, U extends TType> ResourceScatterMul resourceScatterMul(
+  public <T extends Tensor & TNumber, U extends Tensor> ResourceScatterMul resourceScatterMul(
       Operand<?> resource, Operand<T> indices, Operand<U> updates) {
     return ResourceScatterMul.create(scope, resource, indices, updates);
   }
@@ -4337,44 +4345,12 @@ public <T extends TNumber, U extends TType> ResourceScatterMul resourceScatterMu
    * @param options carries optional attributes values
    * @return a new instance of ResourceScatterNdAdd
    */
-  public <T extends TNumber, U extends TType> ResourceScatterNdAdd resourceScatterNdAdd(
+  public <T extends Tensor & TNumber, U extends Tensor> ResourceScatterNdAdd resourceScatterNdAdd(
       Operand<?> ref, Operand<T> indices, Operand<U> updates,
       ResourceScatterNdAdd.Options... options) {
     return ResourceScatterNdAdd.create(scope, ref, indices, updates, options);
   }
 
-  /**
-   *
-   * @param ref A resource handle. Must be from a VarHandleOp.
-   * @param indices A Tensor. Must be one of the following types: int32, int64.
-   *  A tensor of indices into ref.
-   * @param updates A Tensor. Must have the same type as ref. A tensor of
-   *  values whose element wise max is taken with ref
-   * @param options carries optional attributes values
-   * @return a new instance of ResourceScatterNdMax
-   */
-  public <T extends TNumber, U extends TType> ResourceScatterNdMax resourceScatterNdMax(
-      Operand<?> ref, Operand<T> indices, Operand<U> updates,
-      ResourceScatterNdMax.Options... options) {
-    return ResourceScatterNdMax.create(scope, ref, indices, updates, options);
-  }
-
-  /**
-   *
-   * @param ref A resource handle. Must be from a VarHandleOp.
-   * @param indices A Tensor. Must be one of the following types: int32, int64.
-   *  A tensor of indices into ref.
-   * @param updates A Tensor. Must have the same type as ref. A tensor of
-   *  values whose element wise min is taken with ref.
-   * @param options carries optional attributes values
-   * @return a new instance of ResourceScatterNdMin
-   */
-  public <T extends TNumber, U extends TType> ResourceScatterNdMin resourceScatterNdMin(
-      Operand<?> ref, Operand<T> indices, Operand<U> updates,
-      ResourceScatterNdMin.Options... options) {
-    return ResourceScatterNdMin.create(scope, ref, indices, updates, options);
-  }
-
   /**
    * Applies sparse subtraction to individual values or slices in a Variable.
    *  <p>
@@ -4416,7 +4392,7 @@ public <T extends TNumber, U extends TType> ResourceScatterNdMin resourceScatter
    * @param options carries optional attributes values
    * @return a new instance of ResourceScatterNdSub
    */
-  public <T extends TNumber, U extends TType> ResourceScatterNdSub resourceScatterNdSub(
+  public <T extends Tensor & TNumber, U extends Tensor> ResourceScatterNdSub resourceScatterNdSub(
       Operand<?> ref, Operand<T> indices, Operand<U> updates,
       ResourceScatterNdSub.Options... options) {
     return ResourceScatterNdSub.create(scope, ref, indices, updates, options);
@@ -4465,7 +4441,7 @@ public <T extends TNumber, U extends TType> ResourceScatterNdSub resourceScatter
    * @param options carries optional attributes values
    * @return a new instance of ResourceScatterNdUpdate
    */
-  public <T extends TNumber, U extends TType> ResourceScatterNdUpdate resourceScatterNdUpdate(
+  public <T extends Tensor & TNumber, U extends Tensor> ResourceScatterNdUpdate resourceScatterNdUpdate(
       Operand<?> ref, Operand<T> indices, Operand<U> updates,
       ResourceScatterNdUpdate.Options... options) {
     return ResourceScatterNdUpdate.create(scope, ref, indices, updates, options);
@@ -4499,7 +4475,7 @@ public <T extends TNumber, U extends TType> ResourceScatterNdUpdate resourceScat
    * @param updates A tensor of updated values to add to `ref`.
    * @return a new instance of ResourceScatterSub
    */
-  public <T extends TNumber, U extends TType> ResourceScatterSub resourceScatterSub(
+  public <T extends Tensor & TNumber, U extends Tensor> ResourceScatterSub resourceScatterSub(
       Operand<?> resource, Operand<T> indices, Operand<U> updates) {
     return ResourceScatterSub.create(scope, resource, indices, updates);
   }
@@ -4523,7 +4499,7 @@ public <T extends TNumber, U extends TType> ResourceScatterSub resourceScatterSu
    * @param updates A tensor of updated values to add to `ref`.
    * @return a new instance of ResourceScatterUpdate
    */
-  public <T extends TNumber, U extends TType> ResourceScatterUpdate resourceScatterUpdate(
+  public <T extends Tensor & TNumber, U extends Tensor> ResourceScatterUpdate resourceScatterUpdate(
       Operand<?> resource, Operand<T> indices, Operand<U> updates) {
     return ResourceScatterUpdate.create(scope, resource, indices, updates);
   }
@@ -4546,7 +4522,7 @@ public <T extends TNumber, U extends TType> ResourceScatterUpdate resourceScatte
    * @param options carries optional attributes values
    * @return a new instance of ResourceStridedSliceAssign
    */
-  public <T extends TNumber, U extends TType> ResourceStridedSliceAssign resourceStridedSliceAssign(
+  public <T extends Tensor & TNumber, U extends Tensor> ResourceStridedSliceAssign resourceStridedSliceAssign(
       Operand<?> ref, Operand<T> begin, Operand<T> end, Operand<T> strides, Operand<U> value,
       ResourceStridedSliceAssign.Options... options) {
     return ResourceStridedSliceAssign.create(scope, ref, begin, end, strides, value, options);
@@ -4607,7 +4583,7 @@ public <T extends TNumber, U extends TType> ResourceStridedSliceAssign resourceS
    *  `[-rank(tensor), rank(tensor))`.
    * @return a new instance of Reverse
    */
-  public <T extends TType, U extends TNumber> Reverse<T> reverse(Operand<T> tensor,
+  public <T extends Tensor, U extends Tensor & TNumber> Reverse<T> reverse(Operand<T> tensor,
       Operand<U> axis) {
     return Reverse.create(scope, tensor, axis);
   }
@@ -4675,8 +4651,8 @@ public <T extends TType, U extends TNumber> Reverse<T> reverse(Operand<T> tensor
    * @param options carries optional attributes values
    * @return a new instance of ReverseSequence
    */
-  public <T extends TType, U extends TNumber> ReverseSequence<T> reverseSequence(Operand<T> input,
-      Operand<U> seqLengths, Long seqDim, ReverseSequence.Options... options) {
+  public <T extends Tensor, U extends Tensor & TNumber> ReverseSequence<T> reverseSequence(
+      Operand<T> input, Operand<U> seqLengths, Long seqDim, ReverseSequence.Options... options) {
     return ReverseSequence.create(scope, input, seqLengths, seqDim, options);
   }
 
@@ -4715,8 +4691,8 @@ public <T extends TType, U extends TNumber> ReverseSequence<T> reverseSequence(O
    *  axis.
    * @return a new instance of Roll
    */
-  public <T extends TType, U extends TNumber, V extends TNumber> Roll<T> roll(Operand<T> input,
-      Operand<U> shift, Operand<V> axis) {
+  public <T extends Tensor, U extends Tensor & TNumber, V extends Tensor & TNumber> Roll<T> roll(
+      Operand<T> input, Operand<U> shift, Operand<V> axis) {
     return Roll.create(scope, input, shift, axis);
   }
 
@@ -4817,7 +4793,7 @@ public Rpc rpc(Operand<TString> address, Operand<TString> method, Operand<TStrin
    * @param options carries optional attributes values
    * @return a new instance of ScatterAdd
    */
-  public <T extends TType, U extends TNumber> ScatterAdd<T> scatterAdd(Operand<T> ref,
+  public <T extends Tensor, U extends Tensor & TNumber> ScatterAdd<T> scatterAdd(Operand<T> ref,
       Operand<U> indices, Operand<T> updates, ScatterAdd.Options... options) {
     return ScatterAdd.create(scope, ref, indices, updates, options);
   }
@@ -4851,7 +4827,7 @@ public <T extends TType, U extends TNumber> ScatterAdd<T> scatterAdd(Operand<T>
    * @param options carries optional attributes values
    * @return a new instance of ScatterDiv
    */
-  public <T extends TType, U extends TNumber> ScatterDiv<T> scatterDiv(Operand<T> ref,
+  public <T extends Tensor, U extends Tensor & TNumber> ScatterDiv<T> scatterDiv(Operand<T> ref,
       Operand<U> indices, Operand<T> updates, ScatterDiv.Options... options) {
     return ScatterDiv.create(scope, ref, indices, updates, options);
   }
@@ -4889,8 +4865,8 @@ public <T extends TType, U extends TNumber> ScatterDiv<T> scatterDiv(Operand<T>
    * @param options carries optional attributes values
    * @return a new instance of ScatterMax
    */
-  public <T extends TNumber, U extends TNumber> ScatterMax<T> scatterMax(Operand<T> ref,
-      Operand<U> indices, Operand<T> updates, ScatterMax.Options... options) {
+  public <T extends Tensor & TNumber, U extends Tensor & TNumber> ScatterMax<T> scatterMax(
+      Operand<T> ref, Operand<U> indices, Operand<T> updates, ScatterMax.Options... options) {
     return ScatterMax.create(scope, ref, indices, updates, options);
   }
 
@@ -4927,8 +4903,8 @@ public <T extends TNumber, U extends TNumber> ScatterMax<T> scatterMax(Operand<T
    * @param options carries optional attributes values
    * @return a new instance of ScatterMin
    */
-  public <T extends TNumber, U extends TNumber> ScatterMin<T> scatterMin(Operand<T> ref,
-      Operand<U> indices, Operand<T> updates, ScatterMin.Options... options) {
+  public <T extends Tensor & TNumber, U extends Tensor & TNumber> ScatterMin<T> scatterMin(
+      Operand<T> ref, Operand<U> indices, Operand<T> updates, ScatterMin.Options... options) {
     return ScatterMin.create(scope, ref, indices, updates, options);
   }
 
@@ -4961,7 +4937,7 @@ public <T extends TNumber, U extends TNumber> ScatterMin<T> scatterMin(Operand<T
    * @param options carries optional attributes values
    * @return a new instance of ScatterMul
    */
-  public <T extends TType, U extends TNumber> ScatterMul<T> scatterMul(Operand<T> ref,
+  public <T extends Tensor, U extends Tensor & TNumber> ScatterMul<T> scatterMul(Operand<T> ref,
       Operand<U> indices, Operand<T> updates, ScatterMul.Options... options) {
     return ScatterMul.create(scope, ref, indices, updates, options);
   }
@@ -5052,7 +5028,7 @@ public <T extends TType, U extends TNumber> ScatterMul<T> scatterMul(Operand<T>
    * @param shape 1-D. The shape of the resulting tensor.
    * @return a new instance of ScatterNd
    */
-  public <U extends TType, T extends TNumber> ScatterNd<U> scatterNd(Operand<T> indices,
+  public <U extends Tensor, T extends Tensor & TNumber> ScatterNd<U> scatterNd(Operand<T> indices,
       Operand<U> updates, Operand<T> shape) {
     return ScatterNd.create(scope, indices, updates, shape);
   }
@@ -5099,7 +5075,7 @@ public <U extends TType, T extends TNumber> ScatterNd<U> scatterNd(Operand<T> in
    * @param options carries optional attributes values
    * @return a new instance of ScatterNdAdd
    */
-  public <T extends TType, U extends TNumber> ScatterNdAdd<T> scatterNdAdd(Operand<T> ref,
+  public <T extends Tensor, U extends Tensor & TNumber> ScatterNdAdd<T> scatterNdAdd(Operand<T> ref,
       Operand<U> indices, Operand<T> updates, ScatterNdAdd.Options... options) {
     return ScatterNdAdd.create(scope, ref, indices, updates, options);
   }
@@ -5149,7 +5125,7 @@ public <T extends TType, U extends TNumber> ScatterNdAdd<T> scatterNdAdd(Operand
    *  to add to `input`.
    * @return a new instance of ScatterNdNonAliasingAdd
    */
-  public <T extends TType, U extends TNumber> ScatterNdNonAliasingAdd<T> scatterNdNonAliasingAdd(
+  public <T extends Tensor, U extends Tensor & TNumber> ScatterNdNonAliasingAdd<T> scatterNdNonAliasingAdd(
       Operand<T> input, Operand<U> indices, Operand<T> updates) {
     return ScatterNdNonAliasingAdd.create(scope, input, indices, updates);
   }
@@ -5198,7 +5174,7 @@ public <T extends TType, U extends TNumber> ScatterNdNonAliasingAdd<T> scatterNd
    * @param options carries optional attributes values
    * @return a new instance of ScatterNdSub
    */
-  public <T extends TType, U extends TNumber> ScatterNdSub<T> scatterNdSub(Operand<T> ref,
+  public <T extends Tensor, U extends Tensor & TNumber> ScatterNdSub<T> scatterNdSub(Operand<T> ref,
       Operand<U> indices, Operand<T> updates, ScatterNdSub.Options... options) {
     return ScatterNdSub.create(scope, ref, indices, updates, options);
   }
@@ -5249,8 +5225,8 @@ public <T extends TType, U extends TNumber> ScatterNdSub<T> scatterNdSub(Operand
    * @param options carries optional attributes values
    * @return a new instance of ScatterNdUpdate
    */
-  public <T extends TType, U extends TNumber> ScatterNdUpdate<T> scatterNdUpdate(Operand<T> ref,
-      Operand<U> indices, Operand<T> updates, ScatterNdUpdate.Options... options) {
+  public <T extends Tensor, U extends Tensor & TNumber> ScatterNdUpdate<T> scatterNdUpdate(
+      Operand<T> ref, Operand<U> indices, Operand<T> updates, ScatterNdUpdate.Options... options) {
     return ScatterNdUpdate.create(scope, ref, indices, updates, options);
   }
 
@@ -5286,7 +5262,7 @@ public <T extends TType, U extends TNumber> ScatterNdUpdate<T> scatterNdUpdate(O
    * @param options carries optional attributes values
    * @return a new instance of ScatterSub
    */
-  public <T extends TType, U extends TNumber> ScatterSub<T> scatterSub(Operand<T> ref,
+  public <T extends Tensor, U extends Tensor & TNumber> ScatterSub<T> scatterSub(Operand<T> ref,
       Operand<U> indices, Operand<T> updates, ScatterSub.Options... options) {
     return ScatterSub.create(scope, ref, indices, updates, options);
   }
@@ -5327,8 +5303,8 @@ public <T extends TType, U extends TNumber> ScatterSub<T> scatterSub(Operand<T>
    * @param options carries optional attributes values
    * @return a new instance of ScatterUpdate
    */
-  public <T extends TType, U extends TNumber> ScatterUpdate<T> scatterUpdate(Operand<T> ref,
-      Operand<U> indices, Operand<T> updates, ScatterUpdate.Options... options) {
+  public <T extends Tensor, U extends Tensor & TNumber> ScatterUpdate<T> scatterUpdate(
+      Operand<T> ref, Operand<U> indices, Operand<T> updates, ScatterUpdate.Options... options) {
     return ScatterUpdate.create(scope, ref, indices, updates, options);
   }
 
@@ -5340,7 +5316,7 @@ public <T extends TType, U extends TNumber> ScatterUpdate<T> scatterUpdate(Opera
    * @param e
    * @return a new instance of Select
    */
-  public <T extends TType> Select<T> select(Operand<TBool> condition, Operand<T> t, Operand<T> e) {
+  public <T extends Tensor> Select<T> select(Operand<TBool> condition, Operand<T> t, Operand<T> e) {
     return Select.create(scope, condition, t, e);
   }
 
@@ -5372,7 +5348,7 @@ public <T extends TType> Select<T> select(Operand<TBool> condition, Operand<T> t
    * @param y 1-D. Values to remove.
    * @return a new instance of SetDiff1d
    */
-  public <T extends TType> SetDiff1d<T, TInt32> setDiff1d(Operand<T> x, Operand<T> y) {
+  public <T extends Tensor> SetDiff1d<T, TInt32> setDiff1d(Operand<T> x, Operand<T> y) {
     return SetDiff1d.create(scope, x, y);
   }
 
@@ -5405,8 +5381,8 @@ public <T extends TType> SetDiff1d<T, TInt32> setDiff1d(Operand<T> x, Operand<T>
    * @param outIdx
    * @return a new instance of SetDiff1d
    */
-  public <T extends TType, U extends TNumber> SetDiff1d<T, U> setDiff1d(Operand<T> x, Operand<T> y,
-      DataType<U> outIdx) {
+  public <T extends Tensor, U extends Tensor & TNumber> SetDiff1d<T, U> setDiff1d(Operand<T> x,
+      Operand<T> y, DataType<U> outIdx) {
     return SetDiff1d.create(scope, x, y, outIdx);
   }
 
@@ -5426,7 +5402,7 @@ public <T extends TType, U extends TNumber> SetDiff1d<T, U> setDiff1d(Operand<T>
    * @param options carries optional attributes values
    * @return a new instance of SetSize
    */
-  public <T extends TType> SetSize setSize(Operand<TInt64> setIndices, Operand<T> setValues,
+  public <T extends Tensor> SetSize setSize(Operand<TInt64> setIndices, Operand<T> setValues,
       Operand<TInt64> setShape, SetSize.Options... options) {
     return SetSize.create(scope, setIndices, setValues, setShape, options);
   }
@@ -5446,7 +5422,7 @@ public <T extends TType> SetSize setSize(Operand<TInt64> setIndices, Operand<T>
    * @param input
    * @return a new instance of Shape
    */
-  public <T extends TType> org.tensorflow.op.core.Shape<TInt32> shape(Operand<T> input) {
+  public <T extends Tensor> org.tensorflow.op.core.Shape<TInt32> shape(Operand<T> input) {
     return org.tensorflow.op.core.Shape.create(scope, input);
   }
 
@@ -5466,7 +5442,7 @@ public <T extends TType> org.tensorflow.op.core.Shape<TInt32> shape(Operand<T> i
    * @param outType
    * @return a new instance of Shape
    */
-  public <U extends TNumber, T extends TType> org.tensorflow.op.core.Shape<U> shape(
+  public <U extends Tensor & TNumber, T extends Tensor> org.tensorflow.op.core.Shape<U> shape(
       Operand<T> input, DataType<U> outType) {
     return org.tensorflow.op.core.Shape.create(scope, input, outType);
   }
@@ -5480,7 +5456,7 @@ public <U extends TNumber, T extends TType> org.tensorflow.op.core.Shape<U> shap
    * @param input
    * @return a new instance of ShapeN
    */
-  public <T extends TType> ShapeN<TInt32> shapeN(Iterable<Operand<T>> input) {
+  public <T extends Tensor> ShapeN<TInt32> shapeN(Iterable<Operand<T>> input) {
     return ShapeN.create(scope, input);
   }
 
@@ -5494,7 +5470,7 @@ public <T extends TType> ShapeN<TInt32> shapeN(Iterable<Operand<T>> input) {
    * @param outType
    * @return a new instance of ShapeN
    */
-  public <U extends TNumber, T extends TType> ShapeN<U> shapeN(Iterable<Operand<T>> input,
+  public <U extends Tensor & TNumber, T extends Tensor> ShapeN<U> shapeN(Iterable<Operand<T>> input,
       DataType<U> outType) {
     return ShapeN.create(scope, input, outType);
   }
@@ -5515,7 +5491,7 @@ public <U extends TNumber, T extends TType> ShapeN<U> shapeN(Iterable<Operand<T>
    * @param input
    * @return a new instance of Size
    */
-  public <T extends TType> Size<TInt32> size(Operand<T> input) {
+  public <T extends Tensor> Size<TInt32> size(Operand<T> input) {
     return Size.create(scope, input);
   }
 
@@ -5536,7 +5512,8 @@ public <T extends TType> Size<TInt32> size(Operand<T> input) {
    * @param outType
    * @return a new instance of Size
    */
-  public <U extends TNumber, T extends TType> Size<U> size(Operand<T> input, DataType<U> outType) {
+  public <U extends Tensor & TNumber, T extends Tensor> Size<U> size(Operand<T> input,
+      DataType<U> outType) {
     return Size.create(scope, input, outType);
   }
 
@@ -5572,8 +5549,8 @@ public Skipgram skipgram(String filename, Long batchSize, Skipgram.Options... op
    *  size[i] = input.dim_size(i) - begin[i]).
    * @return a new instance of Slice
    */
-  public <T extends TType, U extends TNumber> Slice<T> slice(Operand<T> input, Operand<U> begin,
-      Operand<U> size) {
+  public <T extends Tensor, U extends Tensor & TNumber> Slice<T> slice(Operand<T> input,
+      Operand<U> begin, Operand<U> size) {
     return Slice.create(scope, input, begin, size);
   }
 
@@ -5584,7 +5561,7 @@ public <T extends TType, U extends TNumber> Slice<T> slice(Operand<T> input, Ope
    * @param input
    * @return a new instance of Snapshot
    */
-  public <T extends TType> Snapshot<T> snapshot(Operand<T> input) {
+  public <T extends Tensor> Snapshot<T> snapshot(Operand<T> input) {
     return Snapshot.create(scope, input);
   }
 
@@ -5698,7 +5675,7 @@ public <T extends TType> Snapshot<T> snapshot(Operand<T> input) {
    *  regular convolution.
    * @return a new instance of SpaceToBatchNd
    */
-  public <T extends TType, U extends TNumber, V extends TNumber> SpaceToBatchNd<T> spaceToBatchNd(
+  public <T extends Tensor, U extends Tensor & TNumber, V extends Tensor & TNumber> SpaceToBatchNd<T> spaceToBatchNd(
       Operand<T> input, Operand<U> blockShape, Operand<V> paddings) {
     return SpaceToBatchNd.create(scope, input, blockShape, paddings);
   }
@@ -5714,7 +5691,7 @@ public <T extends TType, U extends TNumber, V extends TNumber> SpaceToBatchNd<T>
    *  `value.shape[split_dim]`.
    * @return a new instance of Split
    */
-  public <T extends TType> Split<T> split(Operand<TInt32> axis, Operand<T> value, Long numSplit) {
+  public <T extends Tensor> Split<T> split(Operand<TInt32> axis, Operand<T> value, Long numSplit) {
     return Split.create(scope, axis, value, numSplit);
   }
 
@@ -5731,7 +5708,7 @@ public <T extends TType> Split<T> split(Operand<TInt32> axis, Operand<T> value,
    * @param numSplit
    * @return a new instance of SplitV
    */
-  public <T extends TType, U extends TNumber> SplitV<T> splitV(Operand<T> value,
+  public <T extends Tensor, U extends Tensor & TNumber> SplitV<T> splitV(Operand<T> value,
       Operand<U> sizeSplits, Operand<TInt32> axis, Long numSplit) {
     return SplitV.create(scope, value, sizeSplits, axis, numSplit);
   }
@@ -5760,7 +5737,7 @@ public <T extends TType, U extends TNumber> SplitV<T> splitV(Operand<T> value,
    * @param options carries optional attributes values
    * @return a new instance of Squeeze
    */
-  public <T extends TType> Squeeze<T> squeeze(Operand<T> input, Squeeze.Options... options) {
+  public <T extends Tensor> Squeeze<T> squeeze(Operand<T> input, Squeeze.Options... options) {
     return Squeeze.create(scope, input, options);
   }
 
@@ -5790,7 +5767,7 @@ public <T extends TType> Squeeze<T> squeeze(Operand<T> input, Squeeze.Options...
    * @param options carries optional attributes values
    * @return a new instance of Stack
    */
-  public <T extends TType> Stack<T> stack(Iterable<Operand<T>> values, Stack.Options... options) {
+  public <T extends Tensor> Stack<T> stack(Iterable<Operand<T>> values, Stack.Options... options) {
     return Stack.create(scope, values, options);
   }
 
@@ -5880,7 +5857,7 @@ public StageSize stageSize(List<DataType<?>> dtypes, StageSize.Options... option
    * @param input
    * @return a new instance of StopGradient
    */
-  public <T extends TType> StopGradient<T> stopGradient(Operand<T> input) {
+  public <T extends Tensor> StopGradient<T> stopGradient(Operand<T> input) {
     return StopGradient.create(scope, input);
   }
 
@@ -5934,11 +5911,11 @@ public <T extends TType> StopGradient<T> stopGradient(Operand<T> input) {
    *  begin = [1, 2, x, x, 0, x] # x denotes don't care (usually 0)
    *  end = [2, 4, x, x, -3, x]
    *  strides = [1, 1, x, x, -1, 1]
-   *  begin_mask = 1<<4 | 1<<5 = 48
+   *  begin_mask = 1<<4 | 1 << 5 = 48
    *  end_mask = 1<<5 = 32
    *  ellipsis_mask = 1<<3 = 8
-   *  new_axis_mask = 1<<2 = 4
-   *  shrink_axis_mask = 1<<0 = 1
+   *  new_axis_mask = 1<<2 4
+   *  shrink_axis_mask = 1<<0
    *  }</pre>
    *  In this case if `foo.shape` is (5, 5, 5, 5, 5, 5) the final shape of
    *  the slice becomes (2, 1, 5, 5, 2, 5).
@@ -5991,8 +5968,9 @@ public <T extends TType> StopGradient<T> stopGradient(Operand<T> input) {
    * @param options carries optional attributes values
    * @return a new instance of StridedSlice
    */
-  public <T extends TType, U extends TNumber> StridedSlice<T> stridedSlice(Operand<T> input,
-      Operand<U> begin, Operand<U> end, Operand<U> strides, StridedSlice.Options... options) {
+  public <T extends Tensor, U extends Tensor & TNumber> StridedSlice<T> stridedSlice(
+      Operand<T> input, Operand<U> begin, Operand<U> end, Operand<U> strides,
+      StridedSlice.Options... options) {
     return StridedSlice.create(scope, input, begin, end, strides, options);
   }
 
@@ -6015,7 +5993,7 @@ public <T extends TType, U extends TNumber> StridedSlice<T> stridedSlice(Operand
    * @param options carries optional attributes values
    * @return a new instance of StridedSliceAssign
    */
-  public <T extends TType, U extends TNumber> StridedSliceAssign<T> stridedSliceAssign(
+  public <T extends Tensor, U extends Tensor & TNumber> StridedSliceAssign<T> stridedSliceAssign(
       Operand<T> ref, Operand<U> begin, Operand<U> end, Operand<U> strides, Operand<T> value,
       StridedSliceAssign.Options... options) {
     return StridedSliceAssign.create(scope, ref, begin, end, strides, value, options);
@@ -6042,8 +6020,8 @@ public <T extends TType, U extends TNumber> StridedSliceAssign<T> stridedSliceAs
    * @param options carries optional attributes values
    * @return a new instance of StridedSliceGrad
    */
-  public <U extends TType, T extends TNumber> StridedSliceGrad<U> stridedSliceGrad(Operand<T> shape,
-      Operand<T> begin, Operand<T> end, Operand<T> strides, Operand<U> dy,
+  public <U extends Tensor, T extends Tensor & TNumber> StridedSliceGrad<U> stridedSliceGrad(
+      Operand<T> shape, Operand<T> begin, Operand<T> end, Operand<T> strides, Operand<U> dy,
       StridedSliceGrad.Options... options) {
     return StridedSliceGrad.create(scope, shape, begin, end, strides, dy, options);
   }
@@ -6063,8 +6041,8 @@ public <U extends TType, T extends TNumber> StridedSliceGrad<U> stridedSliceGrad
    * @param options carries optional attributes values
    * @return a new instance of Sum
    */
-  public <T extends TType, U extends TNumber> Sum<T> sum(Operand<T> input, Operand<U> axis,
-      Sum.Options... options) {
+  public <T extends Tensor, U extends Tensor & TNumber> Sum<T> sum(Operand<T> input,
+      Operand<U> axis, Sum.Options... options) {
     return Sum.create(scope, input, axis, options);
   }
 
@@ -6081,7 +6059,7 @@ public <T extends TType, U extends TNumber> Sum<T> sum(Operand<T> input, Operand
    * @param pred A scalar that specifies which output port will receive data.
    * @return a new instance of SwitchCond
    */
-  public <T extends TType> SwitchCond<T> switchCond(Operand<T> data, Operand<TBool> pred) {
+  public <T extends Tensor> SwitchCond<T> switchCond(Operand<T> data, Operand<TBool> pred) {
     return SwitchCond.create(scope, data, pred);
   }
 
@@ -6109,7 +6087,7 @@ public <T extends TType> SwitchCond<T> switchCond(Operand<T> data, Operand<TBool
    * @param options carries optional attributes values
    * @return a new instance of TemporaryVariable
    */
-  public <T extends TType> TemporaryVariable<T> temporaryVariable(Shape shape, DataType<T> dtype,
+  public <T extends Tensor> TemporaryVariable<T> temporaryVariable(Shape shape, DataType<T> dtype,
       TemporaryVariable.Options... options) {
     return TemporaryVariable.create(scope, shape, dtype, options);
   }
@@ -6124,7 +6102,7 @@ public <T extends TType> TemporaryVariable<T> temporaryVariable(Shape shape, Dat
    * @param options carries optional attributes values
    * @return a new instance of TensorArray
    */
-  public <T extends TType> TensorArray tensorArray(Operand<TInt32> size, DataType<T> dtype,
+  public <T extends Tensor> TensorArray tensorArray(Operand<TInt32> size, DataType<T> dtype,
       TensorArray.Options... options) {
     return TensorArray.create(scope, size, dtype, options);
   }
@@ -6163,7 +6141,7 @@ public TensorArrayClose tensorArrayClose(Operand<?> handle) {
    * @param options carries optional attributes values
    * @return a new instance of TensorArrayConcat
    */
-  public <T extends TType> TensorArrayConcat<T> tensorArrayConcat(Operand<?> handle,
+  public <T extends Tensor> TensorArrayConcat<T> tensorArrayConcat(Operand<?> handle,
       Operand<TFloat32> flowIn, DataType<T> dtype, TensorArrayConcat.Options... options) {
     return TensorArrayConcat.create(scope, handle, flowIn, dtype, options);
   }
@@ -6181,7 +6159,7 @@ public <T extends TType> TensorArrayConcat<T> tensorArrayConcat(Operand<?> handl
    * @param options carries optional attributes values
    * @return a new instance of TensorArrayGather
    */
-  public <T extends TType> TensorArrayGather<T> tensorArrayGather(Operand<?> handle,
+  public <T extends Tensor> TensorArrayGather<T> tensorArrayGather(Operand<?> handle,
       Operand<TInt32> indices, Operand<TFloat32> flowIn, DataType<T> dtype,
       TensorArrayGather.Options... options) {
     return TensorArrayGather.create(scope, handle, indices, flowIn, dtype, options);
@@ -6269,7 +6247,7 @@ public TensorArrayGradWithShape tensorArrayGradWithShape(Operand<?> handle,
    * @param options carries optional attributes values
    * @return a new instance of TensorArrayPack
    */
-  public <T extends TType> TensorArrayPack<T> tensorArrayPack(Operand<TString> handle,
+  public <T extends Tensor> TensorArrayPack<T> tensorArrayPack(Operand<TString> handle,
       Operand<TFloat32> flowIn, DataType<T> dtype, TensorArrayPack.Options... options) {
     return TensorArrayPack.create(scope, handle, flowIn, dtype, options);
   }
@@ -6284,7 +6262,7 @@ public <T extends TType> TensorArrayPack<T> tensorArrayPack(Operand<TString> han
    * @param dtype The type of the elem that is returned.
    * @return a new instance of TensorArrayRead
    */
-  public <T extends TType> TensorArrayRead<T> tensorArrayRead(Operand<?> handle,
+  public <T extends Tensor> TensorArrayRead<T> tensorArrayRead(Operand<?> handle,
       Operand<TInt32> index, Operand<TFloat32> flowIn, DataType<T> dtype) {
     return TensorArrayRead.create(scope, handle, index, flowIn, dtype);
   }
@@ -6300,7 +6278,7 @@ public <T extends TType> TensorArrayRead<T> tensorArrayRead(Operand<?> handle,
    * @param flowIn A float scalar that enforces proper chaining of operations.
    * @return a new instance of TensorArrayScatter
    */
-  public <T extends TType> TensorArrayScatter tensorArrayScatter(Operand<?> handle,
+  public <T extends Tensor> TensorArrayScatter tensorArrayScatter(Operand<?> handle,
       Operand<TInt32> indices, Operand<T> value, Operand<TFloat32> flowIn) {
     return TensorArrayScatter.create(scope, handle, indices, value, flowIn);
   }
@@ -6347,7 +6325,7 @@ public TensorArraySize tensorArraySize(Operand<?> handle, Operand<TFloat32> flow
    * @param flowIn A float scalar that enforces proper chaining of operations.
    * @return a new instance of TensorArraySplit
    */
-  public <T extends TType> TensorArraySplit tensorArraySplit(Operand<?> handle, Operand<T> value,
+  public <T extends Tensor> TensorArraySplit tensorArraySplit(Operand<?> handle, Operand<T> value,
       Operand<TInt64> lengths, Operand<TFloat32> flowIn) {
     return TensorArraySplit.create(scope, handle, value, lengths, flowIn);
   }
@@ -6359,7 +6337,7 @@ public <T extends TType> TensorArraySplit tensorArraySplit(Operand<?> handle, Op
    * @param flowIn
    * @return a new instance of TensorArrayUnpack
    */
-  public <T extends TType> TensorArrayUnpack tensorArrayUnpack(Operand<TString> handle,
+  public <T extends Tensor> TensorArrayUnpack tensorArrayUnpack(Operand<TString> handle,
       Operand<T> value, Operand<TFloat32> flowIn) {
     return TensorArrayUnpack.create(scope, handle, value, flowIn);
   }
@@ -6373,7 +6351,7 @@ public <T extends TType> TensorArrayUnpack tensorArrayUnpack(Operand<TString> ha
    * @param flowIn A float scalar that enforces proper chaining of operations.
    * @return a new instance of TensorArrayWrite
    */
-  public <T extends TType> TensorArrayWrite tensorArrayWrite(Operand<?> handle,
+  public <T extends Tensor> TensorArrayWrite tensorArrayWrite(Operand<?> handle,
       Operand<TInt32> index, Operand<T> value, Operand<TFloat32> flowIn) {
     return TensorArrayWrite.create(scope, handle, index, value, flowIn);
   }
@@ -6400,7 +6378,7 @@ public <T extends TType> TensorArrayWrite tensorArrayWrite(Operand<?> handle,
    * @param elementDtype
    * @return a new instance of TensorListConcat
    */
-  public <U extends TType, T extends TNumber> TensorListConcat<U> tensorListConcat(
+  public <U extends Tensor, T extends Tensor & TNumber> TensorListConcat<U> tensorListConcat(
       Operand<?> inputHandle, Operand<T> elementShape, Operand<TInt64> leadingDims,
       DataType<U> elementDtype) {
     return TensorListConcat.create(scope, inputHandle, elementShape, leadingDims, elementDtype);
@@ -6413,7 +6391,7 @@ public <U extends TType, T extends TNumber> TensorListConcat<U> tensorListConcat
    * @param elementDtype
    * @return a new instance of TensorListConcatLists
    */
-  public <T extends TType> TensorListConcatLists tensorListConcatLists(Operand<?> inputA,
+  public <T extends Tensor> TensorListConcatLists tensorListConcatLists(Operand<?> inputA,
       Operand<?> inputB, DataType<T> elementDtype) {
     return TensorListConcatLists.create(scope, inputA, inputB, elementDtype);
   }
@@ -6429,7 +6407,7 @@ public <T extends TType> TensorListConcatLists tensorListConcatLists(Operand<?>
    * @param shapeType
    * @return a new instance of TensorListElementShape
    */
-  public <T extends TNumber> TensorListElementShape<T> tensorListElementShape(
+  public <T extends Tensor & TNumber> TensorListElementShape<T> tensorListElementShape(
       Operand<?> inputHandle, DataType<T> shapeType) {
     return TensorListElementShape.create(scope, inputHandle, shapeType);
   }
@@ -6446,7 +6424,7 @@ public <T extends TNumber> TensorListElementShape<T> tensorListElementShape(
    * @param elementShape
    * @return a new instance of TensorListFromTensor
    */
-  public <T extends TType, U extends TNumber> TensorListFromTensor tensorListFromTensor(
+  public <T extends Tensor, U extends Tensor & TNumber> TensorListFromTensor tensorListFromTensor(
       Operand<T> tensor, Operand<U> elementShape) {
     return TensorListFromTensor.create(scope, tensor, elementShape);
   }
@@ -6468,7 +6446,7 @@ public <T extends TType, U extends TNumber> TensorListFromTensor tensorListFromT
    * @param elementDtype
    * @return a new instance of TensorListGather
    */
-  public <T extends TType> TensorListGather<T> tensorListGather(Operand<?> inputHandle,
+  public <T extends Tensor> TensorListGather<T> tensorListGather(Operand<?> inputHandle,
       Operand<TInt32> indices, Operand<TInt32> elementShape, DataType<T> elementDtype) {
     return TensorListGather.create(scope, inputHandle, indices, elementShape, elementDtype);
   }
@@ -6482,7 +6460,7 @@ public <T extends TType> TensorListGather<T> tensorListGather(Operand<?> inputHa
    * @param elementDtype
    * @return a new instance of TensorListGetItem
    */
-  public <T extends TType> TensorListGetItem<T> tensorListGetItem(Operand<?> inputHandle,
+  public <T extends Tensor> TensorListGetItem<T> tensorListGetItem(Operand<?> inputHandle,
       Operand<TInt32> index, Operand<TInt32> elementShape, DataType<T> elementDtype) {
     return TensorListGetItem.create(scope, inputHandle, index, elementShape, elementDtype);
   }
@@ -6516,7 +6494,7 @@ public TensorListLength tensorListLength(Operand<?> inputHandle) {
    * @param elementDtype
    * @return a new instance of TensorListPopBack
    */
-  public <T extends TType> TensorListPopBack<T> tensorListPopBack(Operand<?> inputHandle,
+  public <T extends Tensor> TensorListPopBack<T> tensorListPopBack(Operand<?> inputHandle,
       Operand<TInt32> elementShape, DataType<T> elementDtype) {
     return TensorListPopBack.create(scope, inputHandle, elementShape, elementDtype);
   }
@@ -6534,7 +6512,7 @@ public <T extends TType> TensorListPopBack<T> tensorListPopBack(Operand<?> input
    * @param tensor
    * @return a new instance of TensorListPushBack
    */
-  public <T extends TType> TensorListPushBack tensorListPushBack(Operand<?> inputHandle,
+  public <T extends Tensor> TensorListPushBack tensorListPushBack(Operand<?> inputHandle,
       Operand<T> tensor) {
     return TensorListPushBack.create(scope, inputHandle, tensor);
   }
@@ -6545,7 +6523,7 @@ public <T extends TType> TensorListPushBack tensorListPushBack(Operand<?> inputH
    * @param tensor
    * @return a new instance of TensorListPushBackBatch
    */
-  public <T extends TType> TensorListPushBackBatch tensorListPushBackBatch(Operand<?> inputHandles,
+  public <T extends Tensor> TensorListPushBackBatch tensorListPushBackBatch(Operand<?> inputHandles,
       Operand<T> tensor) {
     return TensorListPushBackBatch.create(scope, inputHandles, tensor);
   }
@@ -6563,7 +6541,7 @@ public <T extends TType> TensorListPushBackBatch tensorListPushBackBatch(Operand
    * @param elementDtype
    * @return a new instance of TensorListReserve
    */
-  public <T extends TNumber, U extends TType> TensorListReserve tensorListReserve(
+  public <T extends Tensor & TNumber, U extends Tensor> TensorListReserve tensorListReserve(
       Operand<T> elementShape, Operand<TInt32> numElements, DataType<U> elementDtype) {
     return TensorListReserve.create(scope, elementShape, numElements, elementDtype);
   }
@@ -6604,8 +6582,9 @@ public TensorListResize tensorListResize(Operand<?> inputHandle, Operand<TInt32>
    * @param numElements
    * @return a new instance of TensorListScatter
    */
-  public <T extends TType, U extends TNumber> TensorListScatter tensorListScatter(Operand<T> tensor,
-      Operand<TInt32> indices, Operand<U> elementShape, Operand<TInt32> numElements) {
+  public <T extends Tensor, U extends Tensor & TNumber> TensorListScatter tensorListScatter(
+      Operand<T> tensor, Operand<TInt32> indices, Operand<U> elementShape,
+      Operand<TInt32> numElements) {
     return TensorListScatter.create(scope, tensor, indices, elementShape, numElements);
   }
 
@@ -6625,7 +6604,7 @@ public <T extends TType, U extends TNumber> TensorListScatter tensorListScatter(
    * @param indices
    * @return a new instance of TensorListScatterIntoExistingList
    */
-  public <T extends TType> TensorListScatterIntoExistingList tensorListScatterIntoExistingList(
+  public <T extends Tensor> TensorListScatterIntoExistingList tensorListScatterIntoExistingList(
       Operand<?> inputHandle, Operand<T> tensor, Operand<TInt32> indices) {
     return TensorListScatterIntoExistingList.create(scope, inputHandle, tensor, indices);
   }
@@ -6637,7 +6616,7 @@ public <T extends TType> TensorListScatterIntoExistingList tensorListScatterInto
    * @param item
    * @return a new instance of TensorListSetItem
    */
-  public <T extends TType> TensorListSetItem tensorListSetItem(Operand<?> inputHandle,
+  public <T extends Tensor> TensorListSetItem tensorListSetItem(Operand<?> inputHandle,
       Operand<TInt32> index, Operand<T> item) {
     return TensorListSetItem.create(scope, inputHandle, index, item);
   }
@@ -6658,8 +6637,8 @@ public <T extends TType> TensorListSetItem tensorListSetItem(Operand<?> inputHan
    * @param lengths
    * @return a new instance of TensorListSplit
    */
-  public <T extends TType, U extends TNumber> TensorListSplit tensorListSplit(Operand<T> tensor,
-      Operand<U> elementShape, Operand<TInt64> lengths) {
+  public <T extends Tensor, U extends Tensor & TNumber> TensorListSplit tensorListSplit(
+      Operand<T> tensor, Operand<U> elementShape, Operand<TInt64> lengths) {
     return TensorListSplit.create(scope, tensor, elementShape, lengths);
   }
 
@@ -6679,37 +6658,11 @@ public <T extends TType, U extends TNumber> TensorListSplit tensorListSplit(Oper
    * @param options carries optional attributes values
    * @return a new instance of TensorListStack
    */
-  public <T extends TType> TensorListStack<T> tensorListStack(Operand<?> inputHandle,
+  public <T extends Tensor> TensorListStack<T> tensorListStack(Operand<?> inputHandle,
       Operand<TInt32> elementShape, DataType<T> elementDtype, TensorListStack.Options... options) {
     return TensorListStack.create(scope, inputHandle, elementShape, elementDtype, options);
   }
 
-  /**
-   *
-   * @param <T> data type for {@code output()} output
-   * @param tensor Tensor to update.
-   * @param indices Index tensor.
-   * @param updates Updates to scatter into output.
-   * @return a new instance of TensorScatterMax
-   */
-  public <T extends TType, U extends TNumber> TensorScatterMax<T> tensorScatterMax(
-      Operand<T> tensor, Operand<U> indices, Operand<T> updates) {
-    return TensorScatterMax.create(scope, tensor, indices, updates);
-  }
-
-  /**
-   *
-   * @param <T> data type for {@code output()} output
-   * @param tensor Tensor to update.
-   * @param indices Index tensor.
-   * @param updates Updates to scatter into output.
-   * @return a new instance of TensorScatterMin
-   */
-  public <T extends TType, U extends TNumber> TensorScatterMin<T> tensorScatterMin(
-      Operand<T> tensor, Operand<U> indices, Operand<T> updates) {
-    return TensorScatterMin.create(scope, tensor, indices, updates);
-  }
-
   /**
    * Adds sparse `updates` to an existing tensor according to `indices`.
    *  <p>
@@ -6720,17 +6673,16 @@ public <T extends TType, U extends TNumber> TensorScatterMin<T> tensorScatterMin
    *  for the existing tensor cannot be re-used, a copy is made and updated.
    *  <p>
    *  `indices` is an integer tensor containing indices into a new tensor of shape
-   *  `tensor.shape`.  The last dimension of `indices` can be at most the rank of
-   *  `tensor.shape`:
+   *  `shape`.  The last dimension of `indices` can be at most the rank of `shape`:
    *  <p>
-   *      indices.shape[-1] <= tensor.shape.rank
+   *      indices.shape[-1] <= shape.rank
    *  <p>
    *  The last dimension of `indices` corresponds to indices into elements
-   *  (if `indices.shape[-1] = tensor.shape.rank`) or slices
-   *  (if `indices.shape[-1] < tensor.shape.rank`) along dimension
-   *  `indices.shape[-1]` of `tensor.shape`.  `updates` is a tensor with shape
+   *  (if `indices.shape[-1] = shape.rank`) or slices
+   *  (if `indices.shape[-1] < shape.rank`) along dimension `indices.shape[-1]` of
+   *  `shape`.  `updates` is a tensor with shape
    *  <p>
-   *      indices.shape[:-1] + tensor.shape[indices.shape[-1]:]
+   *      indices.shape[:-1] + shape[indices.shape[-1]:]
    *  <p>
    *  The simplest form of tensor_scatter_add is to add individual elements to a
    *  tensor by index. For example, say we want to add 4 elements in a rank-1
@@ -6779,37 +6731,11 @@ public <T extends TType, U extends TNumber> TensorScatterMin<T> tensorScatterMin
    * @param updates Updates to scatter into output.
    * @return a new instance of TensorScatterNdAdd
    */
-  public <T extends TType, U extends TNumber> TensorScatterNdAdd<T> tensorScatterNdAdd(
+  public <T extends Tensor, U extends Tensor & TNumber> TensorScatterNdAdd<T> tensorScatterNdAdd(
       Operand<T> tensor, Operand<U> indices, Operand<T> updates) {
     return TensorScatterNdAdd.create(scope, tensor, indices, updates);
   }
 
-  /**
-   *
-   * @param <T> data type for {@code output()} output
-   * @param tensor Tensor to update.
-   * @param indices Index tensor.
-   * @param updates Updates to scatter into output.
-   * @return a new instance of TensorScatterNdMax
-   */
-  public <T extends TType, U extends TNumber> TensorScatterNdMax<T> tensorScatterNdMax(
-      Operand<T> tensor, Operand<U> indices, Operand<T> updates) {
-    return TensorScatterNdMax.create(scope, tensor, indices, updates);
-  }
-
-  /**
-   *
-   * @param <T> data type for {@code output()} output
-   * @param tensor Tensor to update.
-   * @param indices Index tensor.
-   * @param updates Updates to scatter into output.
-   * @return a new instance of TensorScatterNdMin
-   */
-  public <T extends TType, U extends TNumber> TensorScatterNdMin<T> tensorScatterNdMin(
-      Operand<T> tensor, Operand<U> indices, Operand<T> updates) {
-    return TensorScatterNdMin.create(scope, tensor, indices, updates);
-  }
-
   /**
    * Subtracts sparse `updates` from an existing tensor according to `indices`.
    *  <p>
@@ -6878,7 +6804,7 @@ public <T extends TType, U extends TNumber> TensorScatterNdMin<T> tensorScatterN
    * @param updates Updates to scatter into output.
    * @return a new instance of TensorScatterNdSub
    */
-  public <T extends TType, U extends TNumber> TensorScatterNdSub<T> tensorScatterNdSub(
+  public <T extends Tensor, U extends Tensor & TNumber> TensorScatterNdSub<T> tensorScatterNdSub(
       Operand<T> tensor, Operand<U> indices, Operand<T> updates) {
     return TensorScatterNdSub.create(scope, tensor, indices, updates);
   }
@@ -6966,7 +6892,7 @@ public <T extends TType, U extends TNumber> TensorScatterNdSub<T> tensorScatterN
    * @param updates Updates to scatter into output.
    * @return a new instance of TensorScatterNdUpdate
    */
-  public <T extends TType, U extends TNumber> TensorScatterNdUpdate<T> tensorScatterNdUpdate(
+  public <T extends Tensor, U extends Tensor & TNumber> TensorScatterNdUpdate<T> tensorScatterNdUpdate(
       Operand<T> tensor, Operand<U> indices, Operand<T> updates) {
     return TensorScatterNdUpdate.create(scope, tensor, indices, updates);
   }
@@ -6990,7 +6916,7 @@ public <T extends TType, U extends TNumber> TensorScatterNdUpdate<T> tensorScatt
    * @param options carries optional attributes values
    * @return a new instance of TensorStridedSliceUpdate
    */
-  public <T extends TType, U extends TNumber> TensorStridedSliceUpdate<T> tensorStridedSliceUpdate(
+  public <T extends Tensor, U extends Tensor & TNumber> TensorStridedSliceUpdate<T> tensorStridedSliceUpdate(
       Operand<T> input, Operand<U> begin, Operand<U> end, Operand<U> strides, Operand<T> value,
       TensorStridedSliceUpdate.Options... options) {
     return TensorStridedSliceUpdate.create(scope, input, begin, end, strides, value, options);
@@ -7031,7 +6957,8 @@ public <T extends TType, U extends TNumber> TensorStridedSliceUpdate<T> tensorSt
    * @param multiples 1-D. Length must be the same as the number of dimensions in `input`
    * @return a new instance of Tile
    */
-  public <T extends TType, U extends TNumber> Tile<T> tile(Operand<T> input, Operand<U> multiples) {
+  public <T extends Tensor, U extends Tensor & TNumber> Tile<T> tile(Operand<T> input,
+      Operand<U> multiples) {
     return Tile.create(scope, input, multiples);
   }
 
@@ -7145,7 +7072,7 @@ public TryRpc tryRpc(Operand<TString> address, Operand<TString> method, Operand<
    * @param options carries optional attributes values
    * @return a new instance of Unbatch
    */
-  public <T extends TType> Unbatch<T> unbatch(Operand<T> batchedTensor, Operand<TInt64> batchIndex,
+  public <T extends Tensor> Unbatch<T> unbatch(Operand<T> batchedTensor, Operand<TInt64> batchIndex,
       Operand<TInt64> id, Long timeoutMicros, Unbatch.Options... options) {
     return Unbatch.create(scope, batchedTensor, batchIndex, id, timeoutMicros, options);
   }
@@ -7176,7 +7103,7 @@ public <T extends TType> Unbatch<T> unbatch(Operand<T> batchedTensor, Operand<TI
    * @param options carries optional attributes values
    * @return a new instance of UnbatchGrad
    */
-  public <T extends TType> UnbatchGrad<T> unbatchGrad(Operand<T> originalInput,
+  public <T extends Tensor> UnbatchGrad<T> unbatchGrad(Operand<T> originalInput,
       Operand<TInt64> batchIndex, Operand<T> grad, Operand<TInt64> id,
       UnbatchGrad.Options... options) {
     return UnbatchGrad.create(scope, originalInput, batchIndex, grad, id, options);
@@ -7231,7 +7158,7 @@ public <T extends TType> UnbatchGrad<T> unbatchGrad(Operand<T> originalInput,
    *  find the unique elements.
    * @return a new instance of Unique
    */
-  public <T extends TType, U extends TNumber> Unique<T, TInt32> unique(Operand<T> x,
+  public <T extends Tensor, U extends Tensor & TNumber> Unique<T, TInt32> unique(Operand<T> x,
       Operand<U> axis) {
     return Unique.create(scope, x, axis);
   }
@@ -7286,8 +7213,8 @@ public <T extends TType, U extends TNumber> Unique<T, TInt32> unique(Operand<T>
    * @param outIdx
    * @return a new instance of Unique
    */
-  public <T extends TType, V extends TNumber, U extends TNumber> Unique<T, V> unique(Operand<T> x,
-      Operand<U> axis, DataType<V> outIdx) {
+  public <T extends Tensor, V extends Tensor & TNumber, U extends Tensor & TNumber> Unique<T, V> unique(
+      Operand<T> x, Operand<U> axis, DataType<V> outIdx) {
     return Unique.create(scope, x, axis, outIdx);
   }
 
@@ -7344,7 +7271,7 @@ public <T extends TType, V extends TNumber, U extends TNumber> Unique<T, V> uniq
    *  find the unique elements.
    * @return a new instance of UniqueWithCounts
    */
-  public <T extends TType, U extends TNumber> UniqueWithCounts<T, TInt32> uniqueWithCounts(
+  public <T extends Tensor, U extends Tensor & TNumber> UniqueWithCounts<T, TInt32> uniqueWithCounts(
       Operand<T> x, Operand<U> axis) {
     return UniqueWithCounts.create(scope, x, axis);
   }
@@ -7403,7 +7330,7 @@ public <T extends TType, U extends TNumber> UniqueWithCounts<T, TInt32> uniqueWi
    * @param outIdx
    * @return a new instance of UniqueWithCounts
    */
-  public <T extends TType, V extends TNumber, U extends TNumber> UniqueWithCounts<T, V> uniqueWithCounts(
+  public <T extends Tensor, V extends Tensor & TNumber, U extends Tensor & TNumber> UniqueWithCounts<T, V> uniqueWithCounts(
       Operand<T> x, Operand<U> axis, DataType<V> outIdx) {
     return UniqueWithCounts.create(scope, x, axis, outIdx);
   }
@@ -7436,7 +7363,8 @@ public <T extends TType, V extends TNumber, U extends TNumber> UniqueWithCounts<
    *  indices.
    * @return a new instance of UnravelIndex
    */
-  public <T extends TNumber> UnravelIndex<T> unravelIndex(Operand<T> indices, Operand<T> dims) {
+  public <T extends Tensor & TNumber> UnravelIndex<T> unravelIndex(Operand<T> indices,
+      Operand<T> dims) {
     return UnravelIndex.create(scope, indices, dims);
   }
 
@@ -7462,7 +7390,7 @@ public <T extends TNumber> UnravelIndex<T> unravelIndex(Operand<T> indices, Oper
    * @param options carries optional attributes values
    * @return a new instance of Unstack
    */
-  public <T extends TType> Unstack<T> unstack(Operand<T> value, Long num,
+  public <T extends Tensor> Unstack<T> unstack(Operand<T> value, Long num,
       Unstack.Options... options) {
     return Unstack.create(scope, value, num, options);
   }
@@ -7490,7 +7418,7 @@ public Unstage unstage(List<DataType<?>> dtypes, Unstage.Options... options) {
    * @param options carries optional attributes values
    * @return a new instance of VarHandleOp
    */
-  public <T extends TType> VarHandleOp varHandleOp(DataType<T> dtype, Shape shape,
+  public <T extends Tensor> VarHandleOp varHandleOp(DataType<T> dtype, Shape shape,
       VarHandleOp.Options... options) {
     return VarHandleOp.create(scope, dtype, shape, options);
   }
@@ -7516,7 +7444,7 @@ public VarIsInitializedOp varIsInitializedOp(Operand<?> resource) {
    * @param options carries optional attributes values
    * @return a new instance of Variable
    */
-  public <T extends TType> Variable<T> variable(Operand<T> init, Variable.Options... options) {
+  public <T extends Tensor> Variable<T> variable(Operand<T> init, Variable.Options... options) {
     return Helpers.createVariableWithInit(scope, init, options);
   }
 
@@ -7533,7 +7461,7 @@ public <T extends TType> Variable<T> variable(Operand<T> init, Variable.Options.
    * @param options carries optional attributes values
    * @return a new instance of Variable
    */
-  public <T extends TType> Variable<T> variable(Shape shape, DataType<T> dtype,
+  public <T extends Tensor> Variable<T> variable(Shape shape, DataType<T> dtype,
       Variable.Options... options) {
     return Variable.create(scope, shape, dtype, options);
   }
@@ -7573,7 +7501,8 @@ public VariableShape<TInt32> variableShape(Operand<?> input) {
    * @param outType
    * @return a new instance of VariableShape
    */
-  public <T extends TNumber> VariableShape<T> variableShape(Operand<?> input, DataType<T> outType) {
+  public <T extends Tensor & TNumber> VariableShape<T> variableShape(Operand<?> input,
+      DataType<T> outType) {
     return VariableShape.create(scope, input, outType);
   }
 
@@ -7642,46 +7571,10 @@ public <T extends TNumber> VariableShape<T> variableShape(Operand<?> input, Data
    * @param condition
    * @return a new instance of Where
    */
-  public <T extends TType> Where where(Operand<T> condition) {
+  public <T extends Tensor> Where where(Operand<T> condition) {
     return Where.create(scope, condition);
   }
 
-  /**
-   * An op used by XLA SPMD partitioner to switch from automatic partitioning to
-   *  <p>
-   *  manual partitioning. It annotates the input (full-shape, to be automatically
-   *  partitioned) with the same sharding used by manual partitioning, and outputs a
-   *  shard-shaped tensor to be consumed by later manually-partitioned ops. If the
-   *  shape is not evenly partitionable, the padding region will be masked with 0s.
-   *
-   * @param <T> data type for {@code output()} output
-   * @param input
-   * @param manualSharding
-   * @return a new instance of XlaSpmdFullToShardShape
-   */
-  public <T extends TType> XlaSpmdFullToShardShape<T> xlaSpmdFullToShardShape(Operand<T> input,
-      String manualSharding) {
-    return XlaSpmdFullToShardShape.create(scope, input, manualSharding);
-  }
-
-  /**
-   * An op used by XLA SPMD partitioner to switch from manual partitioning to
-   *  <p>
-   *  automatic partitioning. It converts the shard-shaped, manually partitioned input
-   *  into full-shaped tensor to be partitioned automatically with the same sharding
-   *  used by manual partitioning.
-   *
-   * @param <T> data type for {@code output()} output
-   * @param input
-   * @param manualSharding
-   * @param fullShape
-   * @return a new instance of XlaSpmdShardToFullShape
-   */
-  public <T extends TType> XlaSpmdShardToFullShape<T> xlaSpmdShardToFullShape(Operand<T> input,
-      String manualSharding, Shape fullShape) {
-    return XlaSpmdShardToFullShape.create(scope, input, manualSharding, fullShape);
-  }
-
   /**
    * Creates a zeroed tensor given its type and shape.
    *
@@ -7691,7 +7584,8 @@ public <T extends TType> XlaSpmdShardToFullShape<T> xlaSpmdShardToFullShape(Oper
    * @return a constant tensor initialized with zeros
    * @throws IllegalArgumentException if the tensor type or shape cannot be initialized with zeros.
    */
-  public <T extends TType, U extends TNumber> Zeros<T> zeros(Operand<U> dims, DataType<T> type) {
+  public <T extends Tensor, U extends Tensor & TNumber> Zeros<T> zeros(Operand<U> dims,
+      DataType<T> type) {
     return Zeros.create(scope, dims, type);
   }
 
@@ -7702,7 +7596,7 @@ public <T extends TType, U extends TNumber> Zeros<T> zeros(Operand<U> dims, Data
    * @param x a tensor of type T.
    * @return a new instance of ZerosLike
    */
-  public <T extends TType> ZerosLike<T> zerosLike(Operand<T> x) {
+  public <T extends Tensor> ZerosLike<T> zerosLike(Operand<T> x) {
     return ZerosLike.create(scope, x);
   }
 
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/QuantizationOps.java b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/QuantizationOps.java
index aec0d667c65..be319493f11 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/QuantizationOps.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/QuantizationOps.java
@@ -19,6 +19,7 @@
 
 import org.tensorflow.DataType;
 import org.tensorflow.Operand;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.quantization.Dequantize;
 import org.tensorflow.op.quantization.FakeQuantWithMinMaxArgs;
 import org.tensorflow.op.quantization.FakeQuantWithMinMaxArgsGradient;
@@ -35,7 +36,6 @@
 import org.tensorflow.types.TFloat32;
 import org.tensorflow.types.TInt32;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * An API for building {@code quantization} operations as {@link Op Op}s
@@ -107,7 +107,7 @@ public final class QuantizationOps {
    * @param options carries optional attributes values
    * @return a new instance of Dequantize
    */
-  public <T extends TType> Dequantize<TFloat32> dequantize(Operand<T> input,
+  public <T extends Tensor> Dequantize<TFloat32> dequantize(Operand<T> input,
       Operand<TFloat32> minRange, Operand<TFloat32> maxRange, Dequantize.Options... options) {
     return Dequantize.create(scope, input, minRange, maxRange, options);
   }
@@ -172,7 +172,7 @@ public <T extends TType> Dequantize<TFloat32> dequantize(Operand<T> input,
    * @param options carries optional attributes values
    * @return a new instance of Dequantize
    */
-  public <U extends TNumber, T extends TType> Dequantize<U> dequantize(Operand<T> input,
+  public <U extends Tensor & TNumber, T extends Tensor> Dequantize<U> dequantize(Operand<T> input,
       Operand<TFloat32> minRange, Operand<TFloat32> maxRange, DataType<U> dtype,
       Dequantize.Options... options) {
     return Dequantize.create(scope, input, minRange, maxRange, dtype, options);
@@ -503,8 +503,9 @@ public FakeQuantWithMinMaxVarsPerChannelGradient fakeQuantWithMinMaxVarsPerChann
    * @param options carries optional attributes values
    * @return a new instance of Quantize
    */
-  public <T extends TType> Quantize<T> quantize(Operand<TFloat32> input, Operand<TFloat32> minRange,
-      Operand<TFloat32> maxRange, DataType<T> T, Quantize.Options... options) {
+  public <T extends Tensor> Quantize<T> quantize(Operand<TFloat32> input,
+      Operand<TFloat32> minRange, Operand<TFloat32> maxRange, DataType<T> T,
+      Quantize.Options... options) {
     return Quantize.create(scope, input, minRange, maxRange, T, options);
   }
 
@@ -522,8 +523,8 @@ public <T extends TType> Quantize<T> quantize(Operand<TFloat32> input, Operand<T
    * @param options carries optional attributes values
    * @return a new instance of QuantizeAndDequantize
    */
-  public <T extends TNumber> QuantizeAndDequantize<T> quantizeAndDequantize(Operand<T> input,
-      Operand<T> inputMin, Operand<T> inputMax, Operand<TInt32> numBits,
+  public <T extends Tensor & TNumber> QuantizeAndDequantize<T> quantizeAndDequantize(
+      Operand<T> input, Operand<T> inputMin, Operand<T> inputMax, Operand<TInt32> numBits,
       QuantizeAndDequantize.Options... options) {
     return QuantizeAndDequantize.create(scope, input, inputMin, inputMax, numBits, options);
   }
@@ -561,7 +562,7 @@ public <T extends TNumber> QuantizeAndDequantize<T> quantizeAndDequantize(Operan
    * @param outType The type of the output. Should be a lower bit depth than Tinput.
    * @return a new instance of QuantizeDownAndShrinkRange
    */
-  public <U extends TType, T extends TType> QuantizeDownAndShrinkRange<U> quantizeDownAndShrinkRange(
+  public <U extends Tensor, T extends Tensor> QuantizeDownAndShrinkRange<U> quantizeDownAndShrinkRange(
       Operand<T> input, Operand<TFloat32> inputMin, Operand<TFloat32> inputMax,
       DataType<U> outType) {
     return QuantizeDownAndShrinkRange.create(scope, input, inputMin, inputMax, outType);
@@ -579,7 +580,7 @@ public <U extends TType, T extends TType> QuantizeDownAndShrinkRange<U> quantize
    * @param inputMaxes The maximum scalar values for each of the input tensors.
    * @return a new instance of QuantizedConcat
    */
-  public <T extends TType> QuantizedConcat<T> quantizedConcat(Operand<TInt32> concatDim,
+  public <T extends Tensor> QuantizedConcat<T> quantizedConcat(Operand<TInt32> concatDim,
       Iterable<Operand<T>> values, Iterable<Operand<TFloat32>> inputMins,
       Iterable<Operand<TFloat32>> inputMaxes) {
     return QuantizedConcat.create(scope, concatDim, values, inputMins, inputMaxes);
@@ -598,7 +599,7 @@ public <T extends TType> QuantizedConcat<T> quantizedConcat(Operand<TInt32> conc
    * @param inputMax The float value that the maximum quantized input value represents.
    * @return a new instance of RequantizationRange
    */
-  public <T extends TType> RequantizationRange requantizationRange(Operand<T> input,
+  public <T extends Tensor> RequantizationRange requantizationRange(Operand<T> input,
       Operand<TFloat32> inputMin, Operand<TFloat32> inputMax) {
     return RequantizationRange.create(scope, input, inputMin, inputMax);
   }
@@ -623,7 +624,7 @@ public <T extends TType> RequantizationRange requantizationRange(Operand<T> inpu
    * @param outType The type of the output. Should be a lower bit depth than Tinput.
    * @return a new instance of Requantize
    */
-  public <U extends TType, T extends TType> Requantize<U> requantize(Operand<T> input,
+  public <U extends Tensor, T extends Tensor> Requantize<U> requantize(Operand<T> input,
       Operand<TFloat32> inputMin, Operand<TFloat32> inputMax, Operand<TFloat32> requestedOutputMin,
       Operand<TFloat32> requestedOutputMax, DataType<U> outType) {
     return Requantize.create(scope, input, inputMin, inputMax, requestedOutputMin, requestedOutputMax, outType);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/RandomOps.java b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/RandomOps.java
index 071e77c7a70..cb9410ebb2a 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/RandomOps.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/RandomOps.java
@@ -19,6 +19,7 @@
 
 import org.tensorflow.DataType;
 import org.tensorflow.Operand;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.random.AllCandidateSampler;
 import org.tensorflow.op.random.LogUniformCandidateSampler;
 import org.tensorflow.op.random.Multinomial;
@@ -42,7 +43,6 @@
 import org.tensorflow.types.TInt32;
 import org.tensorflow.types.TInt64;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * An API for building {@code random} operations as {@link Op Op}s
@@ -124,7 +124,7 @@ public LogUniformCandidateSampler logUniformCandidateSampler(Operand<TInt64> tru
    * @param options carries optional attributes values
    * @return a new instance of Multinomial
    */
-  public <T extends TNumber> Multinomial<TInt64> multinomial(Operand<T> logits,
+  public <T extends Tensor & TNumber> Multinomial<TInt64> multinomial(Operand<T> logits,
       Operand<TInt32> numSamples, Multinomial.Options... options) {
     return Multinomial.create(scope, logits, numSamples, options);
   }
@@ -140,8 +140,9 @@ public <T extends TNumber> Multinomial<TInt64> multinomial(Operand<T> logits,
    * @param options carries optional attributes values
    * @return a new instance of Multinomial
    */
-  public <U extends TNumber, T extends TNumber> Multinomial<U> multinomial(Operand<T> logits,
-      Operand<TInt32> numSamples, DataType<U> outputDtype, Multinomial.Options... options) {
+  public <U extends Tensor & TNumber, T extends Tensor & TNumber> Multinomial<U> multinomial(
+      Operand<T> logits, Operand<TInt32> numSamples, DataType<U> outputDtype,
+      Multinomial.Options... options) {
     return Multinomial.create(scope, logits, numSamples, outputDtype, options);
   }
 
@@ -161,7 +162,7 @@ public <U extends TNumber, T extends TNumber> Multinomial<U> multinomial(Operand
    * @param options carries optional attributes values
    * @return a new instance of ParameterizedTruncatedNormal
    */
-  public <U extends TNumber, T extends TNumber> ParameterizedTruncatedNormal<U> parameterizedTruncatedNormal(
+  public <U extends Tensor & TNumber, T extends Tensor & TNumber> ParameterizedTruncatedNormal<U> parameterizedTruncatedNormal(
       Operand<T> shape, Operand<U> means, Operand<U> stdevs, Operand<U> minvals, Operand<U> maxvals,
       ParameterizedTruncatedNormal.Options... options) {
     return ParameterizedTruncatedNormal.create(scope, shape, means, stdevs, minvals, maxvals, options);
@@ -182,8 +183,8 @@ public <U extends TNumber, T extends TNumber> ParameterizedTruncatedNormal<U> pa
    * @param options carries optional attributes values
    * @return a new instance of RandomGamma
    */
-  public <U extends TNumber, T extends TNumber> RandomGamma<U> randomGamma(Operand<T> shape,
-      Operand<U> alpha, RandomGamma.Options... options) {
+  public <U extends Tensor & TNumber, T extends Tensor & TNumber> RandomGamma<U> randomGamma(
+      Operand<T> shape, Operand<U> alpha, RandomGamma.Options... options) {
     return RandomGamma.create(scope, shape, alpha, options);
   }
 
@@ -208,7 +209,7 @@ public <U extends TNumber, T extends TNumber> RandomGamma<U> randomGamma(Operand
    * @param options carries optional attributes values
    * @return a new instance of RandomPoisson
    */
-  public <T extends TNumber, U extends TNumber> RandomPoisson<TInt64> randomPoisson(
+  public <T extends Tensor & TNumber, U extends Tensor & TNumber> RandomPoisson<TInt64> randomPoisson(
       Operand<T> shape, Operand<U> rate, RandomPoisson.Options... options) {
     return RandomPoisson.create(scope, shape, rate, options);
   }
@@ -235,7 +236,7 @@ public <T extends TNumber, U extends TNumber> RandomPoisson<TInt64> randomPoisso
    * @param options carries optional attributes values
    * @return a new instance of RandomPoisson
    */
-  public <V extends TNumber, T extends TNumber, U extends TNumber> RandomPoisson<V> randomPoisson(
+  public <V extends Tensor & TNumber, T extends Tensor & TNumber, U extends Tensor & TNumber> RandomPoisson<V> randomPoisson(
       Operand<T> shape, Operand<U> rate, DataType<V> dtype, RandomPoisson.Options... options) {
     return RandomPoisson.create(scope, shape, rate, dtype, options);
   }
@@ -257,7 +258,7 @@ public <V extends TNumber, T extends TNumber, U extends TNumber> RandomPoisson<V
    * @param options carries optional attributes values
    * @return a new instance of RandomShuffle
    */
-  public <T extends TType> RandomShuffle<T> randomShuffle(Operand<T> value,
+  public <T extends Tensor> RandomShuffle<T> randomShuffle(Operand<T> value,
       RandomShuffle.Options... options) {
     return RandomShuffle.create(scope, value, options);
   }
@@ -273,7 +274,7 @@ public <T extends TType> RandomShuffle<T> randomShuffle(Operand<T> value,
    * @param options carries optional attributes values
    * @return a new instance of RandomStandardNormal
    */
-  public <U extends TNumber, T extends TNumber> RandomStandardNormal<U> randomStandardNormal(
+  public <U extends Tensor & TNumber, T extends Tensor & TNumber> RandomStandardNormal<U> randomStandardNormal(
       Operand<T> shape, DataType<U> dtype, RandomStandardNormal.Options... options) {
     return RandomStandardNormal.create(scope, shape, dtype, options);
   }
@@ -290,8 +291,8 @@ public <U extends TNumber, T extends TNumber> RandomStandardNormal<U> randomStan
    * @param options carries optional attributes values
    * @return a new instance of RandomUniform
    */
-  public <U extends TNumber, T extends TNumber> RandomUniform<U> randomUniform(Operand<T> shape,
-      DataType<U> dtype, RandomUniform.Options... options) {
+  public <U extends Tensor & TNumber, T extends Tensor & TNumber> RandomUniform<U> randomUniform(
+      Operand<T> shape, DataType<U> dtype, RandomUniform.Options... options) {
     return RandomUniform.create(scope, shape, dtype, options);
   }
 
@@ -313,7 +314,7 @@ public <U extends TNumber, T extends TNumber> RandomUniform<U> randomUniform(Ope
    * @param options carries optional attributes values
    * @return a new instance of RandomUniformInt
    */
-  public <U extends TNumber, T extends TNumber> RandomUniformInt<U> randomUniformInt(
+  public <U extends Tensor & TNumber, T extends Tensor & TNumber> RandomUniformInt<U> randomUniformInt(
       Operand<T> shape, Operand<U> minval, Operand<U> maxval, RandomUniformInt.Options... options) {
     return RandomUniformInt.create(scope, shape, minval, maxval, options);
   }
@@ -339,7 +340,7 @@ public RecordInput recordInput(String filePattern, RecordInput.Options... option
    * @param probs
    * @return a new instance of StatefulRandomBinomial
    */
-  public <T extends TNumber, U extends TNumber> StatefulRandomBinomial<TInt64> statefulRandomBinomial(
+  public <T extends Tensor & TNumber, U extends Tensor & TNumber> StatefulRandomBinomial<TInt64> statefulRandomBinomial(
       Operand<?> resource, Operand<TInt64> algorithm, Operand<T> shape, Operand<U> counts,
       Operand<U> probs) {
     return StatefulRandomBinomial.create(scope, resource, algorithm, shape, counts, probs);
@@ -356,7 +357,7 @@ public <T extends TNumber, U extends TNumber> StatefulRandomBinomial<TInt64> sta
    * @param dtype
    * @return a new instance of StatefulRandomBinomial
    */
-  public <V extends TNumber, T extends TNumber, U extends TNumber> StatefulRandomBinomial<V> statefulRandomBinomial(
+  public <V extends Tensor & TNumber, T extends Tensor & TNumber, U extends Tensor & TNumber> StatefulRandomBinomial<V> statefulRandomBinomial(
       Operand<?> resource, Operand<TInt64> algorithm, Operand<T> shape, Operand<U> counts,
       Operand<U> probs, DataType<V> dtype) {
     return StatefulRandomBinomial.create(scope, resource, algorithm, shape, counts, probs, dtype);
@@ -373,7 +374,7 @@ public <V extends TNumber, T extends TNumber, U extends TNumber> StatefulRandomB
    * @param shape The shape of the output tensor.
    * @return a new instance of StatefulStandardNormal
    */
-  public <T extends TType> StatefulStandardNormal<TFloat32> statefulStandardNormal(
+  public <T extends Tensor> StatefulStandardNormal<TFloat32> statefulStandardNormal(
       Operand<?> resource, Operand<TInt64> algorithm, Operand<T> shape) {
     return StatefulStandardNormal.create(scope, resource, algorithm, shape);
   }
@@ -390,7 +391,7 @@ public <T extends TType> StatefulStandardNormal<TFloat32> statefulStandardNormal
    * @param dtype The type of the output.
    * @return a new instance of StatefulStandardNormal
    */
-  public <U extends TType, T extends TType> StatefulStandardNormal<U> statefulStandardNormal(
+  public <U extends Tensor, T extends Tensor> StatefulStandardNormal<U> statefulStandardNormal(
       Operand<?> resource, Operand<TInt64> algorithm, Operand<T> shape, DataType<U> dtype) {
     return StatefulStandardNormal.create(scope, resource, algorithm, shape, dtype);
   }
@@ -405,7 +406,7 @@ public <U extends TType, T extends TType> StatefulStandardNormal<U> statefulStan
    * @param seed 2 seeds (shape [2]).
    * @return a new instance of StatelessMultinomial
    */
-  public <T extends TNumber, U extends TNumber> StatelessMultinomial<TInt64> statelessMultinomial(
+  public <T extends Tensor & TNumber, U extends Tensor & TNumber> StatelessMultinomial<TInt64> statelessMultinomial(
       Operand<T> logits, Operand<TInt32> numSamples, Operand<U> seed) {
     return StatelessMultinomial.create(scope, logits, numSamples, seed);
   }
@@ -421,7 +422,7 @@ public <T extends TNumber, U extends TNumber> StatelessMultinomial<TInt64> state
    * @param outputDtype
    * @return a new instance of StatelessMultinomial
    */
-  public <V extends TNumber, T extends TNumber, U extends TNumber> StatelessMultinomial<V> statelessMultinomial(
+  public <V extends Tensor & TNumber, T extends Tensor & TNumber, U extends Tensor & TNumber> StatelessMultinomial<V> statelessMultinomial(
       Operand<T> logits, Operand<TInt32> numSamples, Operand<U> seed, DataType<V> outputDtype) {
     return StatelessMultinomial.create(scope, logits, numSamples, seed, outputDtype);
   }
@@ -438,7 +439,7 @@ public <V extends TNumber, T extends TNumber, U extends TNumber> StatelessMultin
    * @param seed 2 seeds (shape [2]).
    * @return a new instance of StatelessRandomNormal
    */
-  public <T extends TNumber, U extends TNumber> StatelessRandomNormal<TFloat32> statelessRandomNormal(
+  public <T extends Tensor & TNumber, U extends Tensor & TNumber> StatelessRandomNormal<TFloat32> statelessRandomNormal(
       Operand<T> shape, Operand<U> seed) {
     return StatelessRandomNormal.create(scope, shape, seed);
   }
@@ -456,7 +457,7 @@ public <T extends TNumber, U extends TNumber> StatelessRandomNormal<TFloat32> st
    * @param dtype The type of the output.
    * @return a new instance of StatelessRandomNormal
    */
-  public <V extends TNumber, T extends TNumber, U extends TNumber> StatelessRandomNormal<V> statelessRandomNormal(
+  public <V extends Tensor & TNumber, T extends Tensor & TNumber, U extends Tensor & TNumber> StatelessRandomNormal<V> statelessRandomNormal(
       Operand<T> shape, Operand<U> seed, DataType<V> dtype) {
     return StatelessRandomNormal.create(scope, shape, seed, dtype);
   }
@@ -474,7 +475,7 @@ public <V extends TNumber, T extends TNumber, U extends TNumber> StatelessRandom
    * @param seed 2 seeds (shape [2]).
    * @return a new instance of StatelessRandomUniform
    */
-  public <T extends TNumber, U extends TNumber> StatelessRandomUniform<TFloat32> statelessRandomUniform(
+  public <T extends Tensor & TNumber, U extends Tensor & TNumber> StatelessRandomUniform<TFloat32> statelessRandomUniform(
       Operand<T> shape, Operand<U> seed) {
     return StatelessRandomUniform.create(scope, shape, seed);
   }
@@ -493,7 +494,7 @@ public <T extends TNumber, U extends TNumber> StatelessRandomUniform<TFloat32> s
    * @param dtype The type of the output.
    * @return a new instance of StatelessRandomUniform
    */
-  public <V extends TNumber, T extends TNumber, U extends TNumber> StatelessRandomUniform<V> statelessRandomUniform(
+  public <V extends Tensor & TNumber, T extends Tensor & TNumber, U extends Tensor & TNumber> StatelessRandomUniform<V> statelessRandomUniform(
       Operand<T> shape, Operand<U> seed, DataType<V> dtype) {
     return StatelessRandomUniform.create(scope, shape, seed, dtype);
   }
@@ -512,7 +513,7 @@ public <V extends TNumber, T extends TNumber, U extends TNumber> StatelessRandom
    * @param seed 2 seeds (shape [2]).
    * @return a new instance of StatelessTruncatedNormal
    */
-  public <T extends TNumber, U extends TNumber> StatelessTruncatedNormal<TFloat32> statelessTruncatedNormal(
+  public <T extends Tensor & TNumber, U extends Tensor & TNumber> StatelessTruncatedNormal<TFloat32> statelessTruncatedNormal(
       Operand<T> shape, Operand<U> seed) {
     return StatelessTruncatedNormal.create(scope, shape, seed);
   }
@@ -532,7 +533,7 @@ public <T extends TNumber, U extends TNumber> StatelessTruncatedNormal<TFloat32>
    * @param dtype The type of the output.
    * @return a new instance of StatelessTruncatedNormal
    */
-  public <V extends TNumber, T extends TNumber, U extends TNumber> StatelessTruncatedNormal<V> statelessTruncatedNormal(
+  public <V extends Tensor & TNumber, T extends Tensor & TNumber, U extends Tensor & TNumber> StatelessTruncatedNormal<V> statelessTruncatedNormal(
       Operand<T> shape, Operand<U> seed, DataType<V> dtype) {
     return StatelessTruncatedNormal.create(scope, shape, seed, dtype);
   }
@@ -550,8 +551,8 @@ public <V extends TNumber, T extends TNumber, U extends TNumber> StatelessTrunca
    * @param options carries optional attributes values
    * @return a new instance of TruncatedNormal
    */
-  public <U extends TNumber, T extends TNumber> TruncatedNormal<U> truncatedNormal(Operand<T> shape,
-      DataType<U> dtype, TruncatedNormal.Options... options) {
+  public <U extends Tensor & TNumber, T extends Tensor & TNumber> TruncatedNormal<U> truncatedNormal(
+      Operand<T> shape, DataType<U> dtype, TruncatedNormal.Options... options) {
     return TruncatedNormal.create(scope, shape, dtype, options);
   }
 
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/ShapeOps.java b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/ShapeOps.java
index 81c692571f1..077dd664c61 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/ShapeOps.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/ShapeOps.java
@@ -19,12 +19,12 @@
 
 import org.tensorflow.DataType;
 import org.tensorflow.Operand;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.core.Shape;
 import org.tensorflow.op.core.Shapes;
 import org.tensorflow.types.TInt32;
 import org.tensorflow.types.TInt64;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * An API for building {@code shape} operations as {@link Op Op}s
@@ -78,7 +78,8 @@ public Operand<TInt32> append(Shape<TInt32> shape, int lastDimension) {
    *      operand representing a shape, followed by the dimensions of an operand representing a shape
    *      to append
    */
-  public <T extends TNumber> Operand<T> append(Operand<T> shape, Operand<T> shapeToAppend) {
+  public <T extends Tensor & TNumber> Operand<T> append(Operand<T> shape,
+      Operand<T> shapeToAppend) {
     return Shapes.append(scope, shape, shapeToAppend);
   }
 
@@ -90,7 +91,7 @@ public <T extends TNumber> Operand<T> append(Operand<T> shape, Operand<T> shapeT
    * @param operand the operand to flatten
    * @return the reshaped operand
    */
-  public <T extends TType> Operand<T> flatten(Operand<T> operand) {
+  public <T extends Tensor> Operand<T> flatten(Operand<T> operand) {
     return Shapes.flatten(scope, operand);
   }
 
@@ -115,7 +116,7 @@ public Operand<TInt32> flatten(Shape<TInt32> shape) {
    * @param dType the shape datatype
    * @return the reshaped operand
    */
-  public <T extends TType, U extends TNumber> Operand<T> flatten(Operand<T> operand,
+  public <T extends Tensor, U extends Tensor & TNumber> Operand<T> flatten(Operand<T> operand,
       DataType<U> dType) {
     return Shapes.flatten(scope, operand, dType);
   }
@@ -129,7 +130,7 @@ public <T extends TType, U extends TNumber> Operand<T> flatten(Operand<T> operan
    * @param dType the shape datatype
    * @return the flattened shape
    */
-  public <U extends TNumber> Operand<U> flatten(Shape<U> shape, DataType<U> dType) {
+  public <U extends Tensor & TNumber> Operand<U> flatten(Shape<U> shape, DataType<U> dType) {
     return Shapes.flatten(scope, shape, dType);
   }
 
@@ -153,7 +154,7 @@ public Operand<TInt32> head(Shape<TInt32> shape) {
    * @param <U> the shape datatype.
    * @return a 1-dimensional Operand containing the Shape's first dimension
    */
-  public <U extends TNumber> Operand<U> head(Shape<U> shape, DataType<U> dType) {
+  public <U extends Tensor & TNumber> Operand<U> head(Shape<U> shape, DataType<U> dType) {
     return Shapes.head(scope, shape, dType);
   }
 
@@ -177,7 +178,7 @@ public Operand<TInt32> numDimensions(Shape<TInt32> shape) {
    * @param dType the shape datatype
    * @return the number of dimensions
    */
-  public <U extends TNumber> Operand<U> numDimensions(Shape<U> shape, DataType<U> dType) {
+  public <U extends Tensor & TNumber> Operand<U> numDimensions(Shape<U> shape, DataType<U> dType) {
     return Shapes.numDimensions(scope, shape, dType);
   }
 
@@ -221,7 +222,8 @@ public Operand<TInt32> prepend(Shape<TInt32> shape, int firstDimension) {
    *      operand representing the shape to prepend, followed by the dimensions of an operand
    *      representing the shape
    */
-  public <T extends TNumber> Operand<T> prepend(Operand<T> shape, Operand<T> shapeToPrepend) {
+  public <T extends Tensor & TNumber> Operand<T> prepend(Operand<T> shape,
+      Operand<T> shapeToPrepend) {
     return Shapes.prepend(scope, shape, shapeToPrepend);
   }
 
@@ -234,7 +236,7 @@ public <T extends TNumber> Operand<T> prepend(Operand<T> shape, Operand<T> shape
    * @param axis the axis
    * @return the reshaped operand
    */
-  public <T extends TType> Operand<T> reduceDims(Operand<T> operand, Operand<TInt32> axis) {
+  public <T extends Tensor> Operand<T> reduceDims(Operand<T> operand, Operand<TInt32> axis) {
     return Shapes.reduceDims(scope, operand, axis);
   }
 
@@ -261,7 +263,7 @@ public Operand<TInt32> reduceDims(Shape<TInt32> shape, Operand<TInt32> axis) {
    * @param dType the shape datatype
    * @return the reshaped operand
    */
-  public <T extends TType, U extends TNumber> Operand<T> reduceDims(Operand<T> operand,
+  public <T extends Tensor, U extends Tensor & TNumber> Operand<T> reduceDims(Operand<T> operand,
       Operand<U> axis, DataType<U> dType) {
     return Shapes.reduceDims(scope, operand, axis, dType);
   }
@@ -276,7 +278,7 @@ public <T extends TType, U extends TNumber> Operand<T> reduceDims(Operand<T> ope
    * @param dType the shape datatype
    * @return the reduced shape
    */
-  public <U extends TNumber> Operand<U> reduceDims(Shape<U> shape, Operand<U> axis,
+  public <U extends Tensor & TNumber> Operand<U> reduceDims(Shape<U> shape, Operand<U> axis,
       DataType<U> dType) {
     return Shapes.reduceDims(scope, shape, axis, dType);
   }
@@ -300,7 +302,7 @@ public Operand<TInt32> size(Shape<TInt32> shape) {
    * @param dim the dimension
    * @return the size of the specified dimension
    */
-  public <T extends TType> Operand<TInt32> size(Operand<T> input, Operand<TInt32> dim) {
+  public <T extends Tensor> Operand<TInt32> size(Operand<T> input, Operand<TInt32> dim) {
     return Shapes.size(scope, input, dim);
   }
 
@@ -313,7 +315,7 @@ public <T extends TType> Operand<TInt32> size(Operand<T> input, Operand<TInt32>
    * @param dType the shape datatype
    * @return the size
    */
-  public <U extends TNumber> Operand<U> size(Shape<U> shape, DataType<U> dType) {
+  public <U extends Tensor & TNumber> Operand<U> size(Shape<U> shape, DataType<U> dType) {
     return Shapes.size(scope, shape, dType);
   }
 
@@ -339,8 +341,8 @@ public Operand<TInt32> size(Shape<TInt32> shape, Operand<TInt32> dim) {
    * @param dType the shape datatype
    * @return the size of the specified dimension
    */
-  public <T extends TType, U extends TNumber> Operand<U> size(Operand<T> input, Operand<U> dim,
-      DataType<U> dType) {
+  public <T extends Tensor, U extends Tensor & TNumber> Operand<U> size(Operand<T> input,
+      Operand<U> dim, DataType<U> dType) {
     return Shapes.size(scope, input, dim, dType);
   }
 
@@ -354,7 +356,8 @@ public <T extends TType, U extends TNumber> Operand<U> size(Operand<T> input, Op
    * @param dType the shape datatype
    * @return the size of the specified dimension
    */
-  public <U extends TNumber> Operand<U> size(Shape<U> shape, Operand<U> dim, DataType<U> dType) {
+  public <U extends Tensor & TNumber> Operand<U> size(Shape<U> shape, Operand<U> dim,
+      DataType<U> dType) {
     return Shapes.size(scope, shape, dim, dType);
   }
 
@@ -378,7 +381,7 @@ public Operand<TInt32> squeeze(Shape<TInt32> shape) {
    * @param dType the shape datatype.
    * @return the squeezed shape
    */
-  public <U extends TNumber> Operand<U> squeeze(Shape<U> shape, DataType<U> dType) {
+  public <U extends Tensor & TNumber> Operand<U> squeeze(Shape<U> shape, DataType<U> dType) {
     return Shapes.squeeze(scope, shape, dType);
   }
 
@@ -406,7 +409,7 @@ public Operand<TInt32> tail(Shape<TInt32> shape) {
    * @return a 1-dimensional Operand that contains the dimension matching the last dimension of the
    *      Shape
    */
-  public <U extends TNumber> Operand<U> tail(Shape<U> shape, DataType<U> dType) {
+  public <U extends Tensor & TNumber> Operand<U> tail(Shape<U> shape, DataType<U> dType) {
     return Shapes.tail(scope, shape, dType);
   }
 
@@ -436,7 +439,8 @@ public Operand<TInt32> take(Shape<TInt32> shape, Operand<TInt32> n) {
    * @return a 1-dimensional operand with the dimensions matching * the first n dimensions of the
    *      shape
    */
-  public <U extends TNumber> Operand<U> take(Shape<U> shape, Operand<U> n, DataType<U> dType) {
+  public <U extends Tensor & TNumber> Operand<U> take(Shape<U> shape, Operand<U> n,
+      DataType<U> dType) {
     return Shapes.take(scope, shape, n, dType);
   }
 
@@ -450,7 +454,8 @@ public <U extends TNumber> Operand<U> take(Shape<U> shape, Operand<U> n, DataTyp
    * @return a 1-dimensional operand containing the dimensions matching the last n dimensions of the
    *      shape
    */
-  public <U extends TNumber> Operand<TInt32> takeLast(Shape<TInt32> shape, Operand<TInt32> n) {
+  public <U extends Tensor & TNumber> Operand<TInt32> takeLast(Shape<TInt32> shape,
+      Operand<TInt32> n) {
     return Shapes.takeLast(scope, shape, n);
   }
 
@@ -466,7 +471,8 @@ public <U extends TNumber> Operand<TInt32> takeLast(Shape<TInt32> shape, Operand
    * @return a 1-dimensional operand containing the dimensions matching the last n dimensions of the
    *      shape
    */
-  public <U extends TNumber> Operand<U> takeLast(Shape<U> shape, Operand<U> n, DataType<U> dType) {
+  public <U extends Tensor & TNumber> Operand<U> takeLast(Shape<U> shape, Operand<U> n,
+      DataType<U> dType) {
     return Shapes.takeLast(scope, shape, n, dType);
   }
 }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/SignalOps.java b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/SignalOps.java
index f4ec7bdb48d..c15b82120f0 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/SignalOps.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/SignalOps.java
@@ -19,6 +19,7 @@
 
 import org.tensorflow.DataType;
 import org.tensorflow.Operand;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.signal.BatchFft;
 import org.tensorflow.op.signal.BatchFft2d;
 import org.tensorflow.op.signal.BatchFft3d;
@@ -40,7 +41,6 @@
 import org.tensorflow.types.TFloat32;
 import org.tensorflow.types.TInt32;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * An API for building {@code signal} operations as {@link Op Op}s
@@ -118,7 +118,7 @@ public BatchIfft3d batchIfft3d(Operand<?> input) {
    * @param input A complex tensor.
    * @return a new instance of Fft
    */
-  public <T extends TType> Fft<T> fft(Operand<T> input) {
+  public <T extends Tensor> Fft<T> fft(Operand<T> input) {
     return Fft.create(scope, input);
   }
 
@@ -132,7 +132,7 @@ public <T extends TType> Fft<T> fft(Operand<T> input) {
    * @param input A complex tensor.
    * @return a new instance of Fft2d
    */
-  public <T extends TType> Fft2d<T> fft2d(Operand<T> input) {
+  public <T extends Tensor> Fft2d<T> fft2d(Operand<T> input) {
     return Fft2d.create(scope, input);
   }
 
@@ -146,7 +146,7 @@ public <T extends TType> Fft2d<T> fft2d(Operand<T> input) {
    * @param input A complex tensor.
    * @return a new instance of Fft3d
    */
-  public <T extends TType> Fft3d<T> fft3d(Operand<T> input) {
+  public <T extends Tensor> Fft3d<T> fft3d(Operand<T> input) {
     return Fft3d.create(scope, input);
   }
 
@@ -160,7 +160,7 @@ public <T extends TType> Fft3d<T> fft3d(Operand<T> input) {
    * @param input A complex tensor.
    * @return a new instance of Ifft
    */
-  public <T extends TType> Ifft<T> ifft(Operand<T> input) {
+  public <T extends Tensor> Ifft<T> ifft(Operand<T> input) {
     return Ifft.create(scope, input);
   }
 
@@ -174,7 +174,7 @@ public <T extends TType> Ifft<T> ifft(Operand<T> input) {
    * @param input A complex tensor.
    * @return a new instance of Ifft2d
    */
-  public <T extends TType> Ifft2d<T> ifft2d(Operand<T> input) {
+  public <T extends Tensor> Ifft2d<T> ifft2d(Operand<T> input) {
     return Ifft2d.create(scope, input);
   }
 
@@ -188,7 +188,7 @@ public <T extends TType> Ifft2d<T> ifft2d(Operand<T> input) {
    * @param input A complex tensor.
    * @return a new instance of Ifft3d
    */
-  public <T extends TType> Ifft3d<T> ifft3d(Operand<T> input) {
+  public <T extends Tensor> Ifft3d<T> ifft3d(Operand<T> input) {
     return Ifft3d.create(scope, input);
   }
 
@@ -214,7 +214,7 @@ public <T extends TType> Ifft3d<T> ifft3d(Operand<T> input) {
    * @param fftLength An int32 tensor of shape [1]. The FFT length.
    * @return a new instance of Irfft
    */
-  public <T extends TType> Irfft<TFloat32> irfft(Operand<T> input, Operand<TInt32> fftLength) {
+  public <T extends Tensor> Irfft<TFloat32> irfft(Operand<T> input, Operand<TInt32> fftLength) {
     return Irfft.create(scope, input, fftLength);
   }
 
@@ -241,7 +241,7 @@ public <T extends TType> Irfft<TFloat32> irfft(Operand<T> input, Operand<TInt32>
    * @param Treal
    * @return a new instance of Irfft
    */
-  public <U extends TNumber, T extends TType> Irfft<U> irfft(Operand<T> input,
+  public <U extends Tensor & TNumber, T extends Tensor> Irfft<U> irfft(Operand<T> input,
       Operand<TInt32> fftLength, DataType<U> Treal) {
     return Irfft.create(scope, input, fftLength, Treal);
   }
@@ -269,7 +269,7 @@ public <U extends TNumber, T extends TType> Irfft<U> irfft(Operand<T> input,
    * @param fftLength An int32 tensor of shape [2]. The FFT length for each dimension.
    * @return a new instance of Irfft2d
    */
-  public <T extends TType> Irfft2d<TFloat32> irfft2d(Operand<T> input, Operand<TInt32> fftLength) {
+  public <T extends Tensor> Irfft2d<TFloat32> irfft2d(Operand<T> input, Operand<TInt32> fftLength) {
     return Irfft2d.create(scope, input, fftLength);
   }
 
@@ -297,7 +297,7 @@ public <T extends TType> Irfft2d<TFloat32> irfft2d(Operand<T> input, Operand<TIn
    * @param Treal
    * @return a new instance of Irfft2d
    */
-  public <U extends TNumber, T extends TType> Irfft2d<U> irfft2d(Operand<T> input,
+  public <U extends Tensor & TNumber, T extends Tensor> Irfft2d<U> irfft2d(Operand<T> input,
       Operand<TInt32> fftLength, DataType<U> Treal) {
     return Irfft2d.create(scope, input, fftLength, Treal);
   }
@@ -325,7 +325,7 @@ public <U extends TNumber, T extends TType> Irfft2d<U> irfft2d(Operand<T> input,
    * @param fftLength An int32 tensor of shape [3]. The FFT length for each dimension.
    * @return a new instance of Irfft3d
    */
-  public <T extends TType> Irfft3d<TFloat32> irfft3d(Operand<T> input, Operand<TInt32> fftLength) {
+  public <T extends Tensor> Irfft3d<TFloat32> irfft3d(Operand<T> input, Operand<TInt32> fftLength) {
     return Irfft3d.create(scope, input, fftLength);
   }
 
@@ -353,7 +353,7 @@ public <T extends TType> Irfft3d<TFloat32> irfft3d(Operand<T> input, Operand<TIn
    * @param Treal
    * @return a new instance of Irfft3d
    */
-  public <U extends TNumber, T extends TType> Irfft3d<U> irfft3d(Operand<T> input,
+  public <U extends Tensor & TNumber, T extends Tensor> Irfft3d<U> irfft3d(Operand<T> input,
       Operand<TInt32> fftLength, DataType<U> Treal) {
     return Irfft3d.create(scope, input, fftLength, Treal);
   }
@@ -378,7 +378,7 @@ public <U extends TNumber, T extends TType> Irfft3d<U> irfft3d(Operand<T> input,
    * @param Tcomplex
    * @return a new instance of Rfft
    */
-  public <U extends TType, T extends TNumber> Rfft<U> rfft(Operand<T> input,
+  public <U extends Tensor, T extends Tensor & TNumber> Rfft<U> rfft(Operand<T> input,
       Operand<TInt32> fftLength, DataType<U> Tcomplex) {
     return Rfft.create(scope, input, fftLength, Tcomplex);
   }
@@ -404,7 +404,7 @@ public <U extends TType, T extends TNumber> Rfft<U> rfft(Operand<T> input,
    * @param Tcomplex
    * @return a new instance of Rfft2d
    */
-  public <U extends TType, T extends TNumber> Rfft2d<U> rfft2d(Operand<T> input,
+  public <U extends Tensor, T extends Tensor & TNumber> Rfft2d<U> rfft2d(Operand<T> input,
       Operand<TInt32> fftLength, DataType<U> Tcomplex) {
     return Rfft2d.create(scope, input, fftLength, Tcomplex);
   }
@@ -430,7 +430,7 @@ public <U extends TType, T extends TNumber> Rfft2d<U> rfft2d(Operand<T> input,
    * @param Tcomplex
    * @return a new instance of Rfft3d
    */
-  public <U extends TType, T extends TNumber> Rfft3d<U> rfft3d(Operand<T> input,
+  public <U extends Tensor, T extends Tensor & TNumber> Rfft3d<U> rfft3d(Operand<T> input,
       Operand<TInt32> fftLength, DataType<U> Tcomplex) {
     return Rfft3d.create(scope, input, fftLength, Tcomplex);
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/SparseOps.java b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/SparseOps.java
index 42cdf9569d9..d8ba9f45ef2 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/SparseOps.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/SparseOps.java
@@ -19,6 +19,7 @@
 
 import org.tensorflow.DataType;
 import org.tensorflow.Operand;
+import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.sparse.AddManySparseToTensorsMap;
 import org.tensorflow.op.sparse.AddSparseToTensorsMap;
@@ -29,11 +30,9 @@
 import org.tensorflow.op.sparse.SparseAccumulatorTakeGradient;
 import org.tensorflow.op.sparse.SparseAdd;
 import org.tensorflow.op.sparse.SparseAddGrad;
-import org.tensorflow.op.sparse.SparseBincount;
 import org.tensorflow.op.sparse.SparseConcat;
 import org.tensorflow.op.sparse.SparseConditionalAccumulator;
 import org.tensorflow.op.sparse.SparseCross;
-import org.tensorflow.op.sparse.SparseCrossHashed;
 import org.tensorflow.op.sparse.SparseDenseCwiseAdd;
 import org.tensorflow.op.sparse.SparseDenseCwiseDiv;
 import org.tensorflow.op.sparse.SparseDenseCwiseMul;
@@ -65,12 +64,10 @@
 import org.tensorflow.op.sparse.SparseToDense;
 import org.tensorflow.op.sparse.SparseToSparseSetOperation;
 import org.tensorflow.op.sparse.TakeManySparseFromTensorsMap;
-import org.tensorflow.types.TBool;
 import org.tensorflow.types.TInt32;
 import org.tensorflow.types.TInt64;
 import org.tensorflow.types.TString;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * An API for building {@code sparse} operations as {@link Op Op}s
@@ -117,7 +114,7 @@ public final class SparseOps {
    * @param options carries optional attributes values
    * @return a new instance of AddManySparseToTensorsMap
    */
-  public <T extends TType> AddManySparseToTensorsMap addManySparseToTensorsMap(
+  public <T extends Tensor> AddManySparseToTensorsMap addManySparseToTensorsMap(
       Operand<TInt64> sparseIndices, Operand<T> sparseValues, Operand<TInt64> sparseShape,
       AddManySparseToTensorsMap.Options... options) {
     return AddManySparseToTensorsMap.create(scope, sparseIndices, sparseValues, sparseShape, options);
@@ -147,7 +144,7 @@ public <T extends TType> AddManySparseToTensorsMap addManySparseToTensorsMap(
    * @param options carries optional attributes values
    * @return a new instance of AddSparseToTensorsMap
    */
-  public <T extends TType> AddSparseToTensorsMap addSparseToTensorsMap(
+  public <T extends Tensor> AddSparseToTensorsMap addSparseToTensorsMap(
       Operand<TInt64> sparseIndices, Operand<T> sparseValues, Operand<TInt64> sparseShape,
       AddSparseToTensorsMap.Options... options) {
     return AddSparseToTensorsMap.create(scope, sparseIndices, sparseValues, sparseShape, options);
@@ -173,7 +170,7 @@ public <T extends TType> AddSparseToTensorsMap addSparseToTensorsMap(
    * @param options carries optional attributes values
    * @return a new instance of DenseToDenseSetOperation
    */
-  public <T extends TType> DenseToDenseSetOperation<T> denseToDenseSetOperation(Operand<T> set1,
+  public <T extends Tensor> DenseToDenseSetOperation<T> denseToDenseSetOperation(Operand<T> set1,
       Operand<T> set2, String setOperation, DenseToDenseSetOperation.Options... options) {
     return DenseToDenseSetOperation.create(scope, set1, set2, setOperation, options);
   }
@@ -211,7 +208,7 @@ public <T extends TType> DenseToDenseSetOperation<T> denseToDenseSetOperation(Op
    * @param options carries optional attributes values
    * @return a new instance of DenseToSparseSetOperation
    */
-  public <T extends TType> DenseToSparseSetOperation<T> denseToSparseSetOperation(Operand<T> set1,
+  public <T extends Tensor> DenseToSparseSetOperation<T> denseToSparseSetOperation(Operand<T> set1,
       Operand<TInt64> set2Indices, Operand<T> set2Values, Operand<TInt64> set2Shape,
       String setOperation, DenseToSparseSetOperation.Options... options) {
     return DenseToSparseSetOperation.create(scope, set1, set2Indices, set2Values, set2Shape, setOperation, options);
@@ -268,7 +265,7 @@ public <T extends TType> DenseToSparseSetOperation<T> denseToSparseSetOperation(
    * @param dtype The `dtype` of the serialized `SparseTensor` objects.
    * @return a new instance of DeserializeSparse
    */
-  public <U extends TType, T extends TType> DeserializeSparse<U> deserializeSparse(
+  public <U extends Tensor, T extends Tensor> DeserializeSparse<U> deserializeSparse(
       Operand<T> serializedSparse, DataType<U> dtype) {
     return DeserializeSparse.create(scope, serializedSparse, dtype);
   }
@@ -291,7 +288,7 @@ public <U extends TType, T extends TType> DeserializeSparse<U> deserializeSparse
    *  case the input is ignored during validation.
    * @return a new instance of SparseAccumulatorApplyGradient
    */
-  public <T extends TType> SparseAccumulatorApplyGradient sparseAccumulatorApplyGradient(
+  public <T extends Tensor> SparseAccumulatorApplyGradient sparseAccumulatorApplyGradient(
       Operand<TString> handle, Operand<TInt64> localStep, Operand<TInt64> gradientIndices,
       Operand<T> gradientValues, Operand<TInt64> gradientShape, Boolean hasKnownShape) {
     return SparseAccumulatorApplyGradient.create(scope, handle, localStep, gradientIndices, gradientValues, gradientShape, hasKnownShape);
@@ -314,7 +311,7 @@ public <T extends TType> SparseAccumulatorApplyGradient sparseAccumulatorApplyGr
    *  of the accumulator.
    * @return a new instance of SparseAccumulatorTakeGradient
    */
-  public <T extends TType> SparseAccumulatorTakeGradient<T> sparseAccumulatorTakeGradient(
+  public <T extends Tensor> SparseAccumulatorTakeGradient<T> sparseAccumulatorTakeGradient(
       Operand<TString> handle, Operand<TInt32> numRequired, DataType<T> dtype) {
     return SparseAccumulatorTakeGradient.create(scope, handle, numRequired, dtype);
   }
@@ -347,9 +344,9 @@ public <T extends TType> SparseAccumulatorTakeGradient<T> sparseAccumulatorTakeG
    *  pair takes space.
    * @return a new instance of SparseAdd
    */
-  public <T extends TType, U extends TNumber> SparseAdd<T> sparseAdd(Operand<TInt64> aIndices,
-      Operand<T> aValues, Operand<TInt64> aShape, Operand<TInt64> bIndices, Operand<T> bValues,
-      Operand<TInt64> bShape, Operand<U> thresh) {
+  public <T extends Tensor, U extends Tensor & TNumber> SparseAdd<T> sparseAdd(
+      Operand<TInt64> aIndices, Operand<T> aValues, Operand<TInt64> aShape,
+      Operand<TInt64> bIndices, Operand<T> bValues, Operand<TInt64> bShape, Operand<U> thresh) {
     return SparseAdd.create(scope, aIndices, aValues, aShape, bIndices, bValues, bShape, thresh);
   }
 
@@ -370,39 +367,11 @@ public <T extends TType, U extends TNumber> SparseAdd<T> sparseAdd(Operand<TInt6
    *  `[nnz(sum), ndims]`.
    * @return a new instance of SparseAddGrad
    */
-  public <T extends TType> SparseAddGrad<T> sparseAddGrad(Operand<T> backpropValGrad,
+  public <T extends Tensor> SparseAddGrad<T> sparseAddGrad(Operand<T> backpropValGrad,
       Operand<TInt64> aIndices, Operand<TInt64> bIndices, Operand<TInt64> sumIndices) {
     return SparseAddGrad.create(scope, backpropValGrad, aIndices, bIndices, sumIndices);
   }
 
-  /**
-   * Counts the number of occurrences of each value in an integer array.
-   *  <p>
-   *  Outputs a vector with length `size` and the same dtype as `weights`. If
-   *  `weights` are empty, then index `i` stores the number of times the value `i` is
-   *  counted in `arr`. If `weights` are non-empty, then index `i` stores the sum of
-   *  the value in `weights` at each index where the corresponding value in `arr` is
-   *  `i`.
-   *  <p>
-   *  Values in `arr` outside of the range [0, size) are ignored.
-   *
-   * @param <U> data type for {@code output()} output
-   * @param indices 2D int64 `Tensor`.
-   * @param values 1D int `Tensor`.
-   * @param denseShape 1D int64 `Tensor`.
-   * @param size non-negative int scalar `Tensor`.
-   * @param weights is an int32, int64, float32, or float64 `Tensor` with the same
-   *  shape as `input`, or a length-0 `Tensor`, in which case it acts as all weights
-   *  equal to 1.
-   * @param options carries optional attributes values
-   * @return a new instance of SparseBincount
-   */
-  public <U extends TNumber, T extends TNumber> SparseBincount<U> sparseBincount(
-      Operand<TInt64> indices, Operand<T> values, Operand<TInt64> denseShape, Operand<T> size,
-      Operand<U> weights, SparseBincount.Options... options) {
-    return SparseBincount.create(scope, indices, values, denseShape, size, weights, options);
-  }
-
   /**
    * Concatenates a list of `SparseTensor` along the specified dimension.
    *  <p>
@@ -456,7 +425,7 @@ public <U extends TNumber, T extends TNumber> SparseBincount<U> sparseBincount(
    *  where rank is the number of dimensions in each input `SparseTensor`.
    * @return a new instance of SparseConcat
    */
-  public <T extends TType> SparseConcat<T> sparseConcat(Iterable<Operand<TInt64>> indices,
+  public <T extends Tensor> SparseConcat<T> sparseConcat(Iterable<Operand<TInt64>> indices,
       Iterable<Operand<T>> values, Iterable<Operand<TInt64>> shapes, Long concatDim) {
     return SparseConcat.create(scope, indices, values, shapes, concatDim);
   }
@@ -476,7 +445,7 @@ public <T extends TType> SparseConcat<T> sparseConcat(Iterable<Operand<TInt64>>
    * @param options carries optional attributes values
    * @return a new instance of SparseConditionalAccumulator
    */
-  public <T extends TType> SparseConditionalAccumulator sparseConditionalAccumulator(
+  public <T extends Tensor> SparseConditionalAccumulator sparseConditionalAccumulator(
       DataType<T> dtype, Shape shape, SparseConditionalAccumulator.Options... options) {
     return SparseConditionalAccumulator.create(scope, dtype, shape, options);
   }
@@ -521,73 +490,26 @@ public <T extends TType> SparseConditionalAccumulator sparseConditionalAccumulat
    *                  Fingerprint64("g"), FingerprintCat64(
    *                      Fingerprint64("e"), Fingerprint64("c")))
    *
+   * @param <T> data type for {@code outputValues()} output
    * @param indices 2-D.  Indices of each input `SparseTensor`.
    * @param values 1-D.   values of each `SparseTensor`.
    * @param shapes 1-D.   Shapes of each `SparseTensor`.
    * @param denseInputs 2-D.    Columns represented by dense `Tensor`.
-   * @param sep string used when joining a list of string inputs, can be used as separator later.
-   * @return a new instance of SparseCross
-   */
-  public SparseCross sparseCross(Iterable<Operand<TInt64>> indices, Iterable<Operand<?>> values,
-      Iterable<Operand<TInt64>> shapes, Iterable<Operand<?>> denseInputs, Operand<TString> sep) {
-    return SparseCross.create(scope, indices, values, shapes, denseInputs, sep);
-  }
-
-  /**
-   * Generates sparse cross from a list of sparse and dense tensors.
-   *  <p>
-   *  The op takes two lists, one of 2D `SparseTensor` and one of 2D `Tensor`, each
-   *  representing features of one feature column. It outputs a 2D `SparseTensor` with
-   *  the batchwise crosses of these features.
-   *  <p>
-   *  For example, if the inputs are
-   *  <p>
-   *      inputs[0]: SparseTensor with shape = [2, 2]
-   *      [0, 0]: "a"
-   *      [1, 0]: "b"
-   *      [1, 1]: "c"
-   *  <p>
-   *      inputs[1]: SparseTensor with shape = [2, 1]
-   *      [0, 0]: "d"
-   *      [1, 0]: "e"
-   *  <p>
-   *      inputs[2]: Tensor [["f"], ["g"]]
-   *  <p>
-   *  then the output will be
-   *  <p>
-   *      shape = [2, 2]
-   *      [0, 0]: "a_X_d_X_f"
-   *      [1, 0]: "b_X_e_X_g"
-   *      [1, 1]: "c_X_e_X_g"
-   *  <p>
-   *  if hashed_output=true then the output will be
-   *  <p>
-   *      shape = [2, 2]
-   *      [0, 0]: FingerprintCat64(
-   *                  Fingerprint64("f"), FingerprintCat64(
-   *                      Fingerprint64("d"), Fingerprint64("a")))
-   *      [1, 0]: FingerprintCat64(
-   *                  Fingerprint64("g"), FingerprintCat64(
-   *                      Fingerprint64("e"), Fingerprint64("b")))
-   *      [1, 1]: FingerprintCat64(
-   *                  Fingerprint64("g"), FingerprintCat64(
-   *                      Fingerprint64("e"), Fingerprint64("c")))
-   *
-   * @param indices 2-D.  Indices of each input `SparseTensor`.
-   * @param values 1-D.   values of each `SparseTensor`.
-   * @param shapes 1-D.   Shapes of each `SparseTensor`.
-   * @param denseInputs 2-D.    Columns represented by dense `Tensor`.
+   * @param hashedOutput If true, returns the hash of the cross instead of the string.
+   *  This will allow us avoiding string manipulations.
    * @param numBuckets It is used if hashed_output is true.
    *  output = hashed_value%num_buckets if num_buckets > 0 else hashed_value.
-   * @param strongHash boolean, if true, siphash with salt will be used instead of farmhash.
-   * @param salt Specify the salt that will be used by the siphash function.
-   * @return a new instance of SparseCrossHashed
+   * @param hashKey Specify the hash_key that will be used by the `FingerprintCat64`
+   *  function to combine the crosses fingerprints.
+   * @param outType
+   * @param internalType
+   * @return a new instance of SparseCross
    */
-  public SparseCrossHashed sparseCrossHashed(Iterable<Operand<TInt64>> indices,
-      Iterable<Operand<?>> values, Iterable<Operand<TInt64>> shapes,
-      Iterable<Operand<?>> denseInputs, Operand<TInt64> numBuckets, Operand<TBool> strongHash,
-      Operand<TInt64> salt) {
-    return SparseCrossHashed.create(scope, indices, values, shapes, denseInputs, numBuckets, strongHash, salt);
+  public <T extends Tensor, U extends Tensor> SparseCross<T> sparseCross(
+      Iterable<Operand<TInt64>> indices, Iterable<Operand<?>> values,
+      Iterable<Operand<TInt64>> shapes, Iterable<Operand<?>> denseInputs, Boolean hashedOutput,
+      Long numBuckets, Long hashKey, DataType<T> outType, DataType<U> internalType) {
+    return SparseCross.create(scope, indices, values, shapes, denseInputs, hashedOutput, numBuckets, hashKey, outType, internalType);
   }
 
   /**
@@ -610,7 +532,7 @@ public SparseCrossHashed sparseCrossHashed(Iterable<Operand<TInt64>> indices,
    * @param dense `R`-D.  The dense Tensor operand.
    * @return a new instance of SparseDenseCwiseAdd
    */
-  public <T extends TType> SparseDenseCwiseAdd<T> sparseDenseCwiseAdd(Operand<TInt64> spIndices,
+  public <T extends Tensor> SparseDenseCwiseAdd<T> sparseDenseCwiseAdd(Operand<TInt64> spIndices,
       Operand<T> spValues, Operand<TInt64> spShape, Operand<T> dense) {
     return SparseDenseCwiseAdd.create(scope, spIndices, spValues, spShape, dense);
   }
@@ -629,7 +551,7 @@ public <T extends TType> SparseDenseCwiseAdd<T> sparseDenseCwiseAdd(Operand<TInt
    * @param dense `R`-D.  The dense Tensor operand.
    * @return a new instance of SparseDenseCwiseDiv
    */
-  public <T extends TType> SparseDenseCwiseDiv<T> sparseDenseCwiseDiv(Operand<TInt64> spIndices,
+  public <T extends Tensor> SparseDenseCwiseDiv<T> sparseDenseCwiseDiv(Operand<TInt64> spIndices,
       Operand<T> spValues, Operand<TInt64> spShape, Operand<T> dense) {
     return SparseDenseCwiseDiv.create(scope, spIndices, spValues, spShape, dense);
   }
@@ -652,7 +574,7 @@ public <T extends TType> SparseDenseCwiseDiv<T> sparseDenseCwiseDiv(Operand<TInt
    * @param dense `R`-D.  The dense Tensor operand.
    * @return a new instance of SparseDenseCwiseMul
    */
-  public <T extends TType> SparseDenseCwiseMul<T> sparseDenseCwiseMul(Operand<TInt64> spIndices,
+  public <T extends Tensor> SparseDenseCwiseMul<T> sparseDenseCwiseMul(Operand<TInt64> spIndices,
       Operand<T> spValues, Operand<TInt64> spShape, Operand<T> dense) {
     return SparseDenseCwiseMul.create(scope, spIndices, spValues, spShape, dense);
   }
@@ -706,7 +628,7 @@ public <T extends TType> SparseDenseCwiseMul<T> sparseDenseCwiseMul(Operand<TInt
    *  output indices: 2-D. the indices of the filled sparse tensor.
    * @return a new instance of SparseFillEmptyRows
    */
-  public <T extends TType> SparseFillEmptyRows<T> sparseFillEmptyRows(Operand<TInt64> indices,
+  public <T extends Tensor> SparseFillEmptyRows<T> sparseFillEmptyRows(Operand<TInt64> indices,
       Operand<T> values, Operand<TInt64> denseShape, Operand<T> defaultValue) {
     return SparseFillEmptyRows.create(scope, indices, values, denseShape, defaultValue);
   }
@@ -728,7 +650,7 @@ public <T extends TType> SparseFillEmptyRows<T> sparseFillEmptyRows(Operand<TInt
    * @param gradValues 1-D.  The gradients from backprop.
    * @return a new instance of SparseFillEmptyRowsGrad
    */
-  public <T extends TType> SparseFillEmptyRowsGrad<T> sparseFillEmptyRowsGrad(
+  public <T extends Tensor> SparseFillEmptyRowsGrad<T> sparseFillEmptyRowsGrad(
       Operand<TInt64> reverseIndexMap, Operand<T> gradValues) {
     return SparseFillEmptyRowsGrad.create(scope, reverseIndexMap, gradValues);
   }
@@ -751,8 +673,8 @@ public <T extends TType> SparseFillEmptyRowsGrad<T> sparseFillEmptyRowsGrad(
    * @param options carries optional attributes values
    * @return a new instance of SparseMatMul
    */
-  public <T extends TNumber, U extends TNumber> SparseMatMul sparseMatMul(Operand<T> a,
-      Operand<U> b, SparseMatMul.Options... options) {
+  public <T extends Tensor & TNumber, U extends Tensor & TNumber> SparseMatMul sparseMatMul(
+      Operand<T> a, Operand<U> b, SparseMatMul.Options... options) {
     return SparseMatMul.create(scope, a, b, options);
   }
 
@@ -781,9 +703,9 @@ public <T extends TNumber, U extends TNumber> SparseMatMul sparseMatMul(Operand<
    * @param options carries optional attributes values
    * @return a new instance of SparseReduceMax
    */
-  public <T extends TNumber> SparseReduceMax<T> sparseReduceMax(Operand<TInt64> inputIndices,
-      Operand<T> inputValues, Operand<TInt64> inputShape, Operand<TInt32> reductionAxes,
-      SparseReduceMax.Options... options) {
+  public <T extends Tensor & TNumber> SparseReduceMax<T> sparseReduceMax(
+      Operand<TInt64> inputIndices, Operand<T> inputValues, Operand<TInt64> inputShape,
+      Operand<TInt32> reductionAxes, SparseReduceMax.Options... options) {
     return SparseReduceMax.create(scope, inputIndices, inputValues, inputShape, reductionAxes, options);
   }
 
@@ -812,7 +734,7 @@ public <T extends TNumber> SparseReduceMax<T> sparseReduceMax(Operand<TInt64> in
    * @param options carries optional attributes values
    * @return a new instance of SparseReduceMaxSparse
    */
-  public <T extends TNumber> SparseReduceMaxSparse<T> sparseReduceMaxSparse(
+  public <T extends Tensor & TNumber> SparseReduceMaxSparse<T> sparseReduceMaxSparse(
       Operand<TInt64> inputIndices, Operand<T> inputValues, Operand<TInt64> inputShape,
       Operand<TInt32> reductionAxes, SparseReduceMaxSparse.Options... options) {
     return SparseReduceMaxSparse.create(scope, inputIndices, inputValues, inputShape, reductionAxes, options);
@@ -843,7 +765,7 @@ public <T extends TNumber> SparseReduceMaxSparse<T> sparseReduceMaxSparse(
    * @param options carries optional attributes values
    * @return a new instance of SparseReduceSum
    */
-  public <T extends TType> SparseReduceSum<T> sparseReduceSum(Operand<TInt64> inputIndices,
+  public <T extends Tensor> SparseReduceSum<T> sparseReduceSum(Operand<TInt64> inputIndices,
       Operand<T> inputValues, Operand<TInt64> inputShape, Operand<TInt32> reductionAxes,
       SparseReduceSum.Options... options) {
     return SparseReduceSum.create(scope, inputIndices, inputValues, inputShape, reductionAxes, options);
@@ -874,7 +796,7 @@ public <T extends TType> SparseReduceSum<T> sparseReduceSum(Operand<TInt64> inpu
    * @param options carries optional attributes values
    * @return a new instance of SparseReduceSumSparse
    */
-  public <T extends TType> SparseReduceSumSparse<T> sparseReduceSumSparse(
+  public <T extends Tensor> SparseReduceSumSparse<T> sparseReduceSumSparse(
       Operand<TInt64> inputIndices, Operand<T> inputValues, Operand<TInt64> inputShape,
       Operand<TInt32> reductionAxes, SparseReduceSumSparse.Options... options) {
     return SparseReduceSumSparse.create(scope, inputIndices, inputValues, inputShape, reductionAxes, options);
@@ -899,7 +821,7 @@ public <T extends TType> SparseReduceSumSparse<T> sparseReduceSumSparse(
    * @param inputShape 1-D.  Shape of the input SparseTensor.
    * @return a new instance of SparseReorder
    */
-  public <T extends TType> SparseReorder<T> sparseReorder(Operand<TInt64> inputIndices,
+  public <T extends Tensor> SparseReorder<T> sparseReorder(Operand<TInt64> inputIndices,
       Operand<T> inputValues, Operand<TInt64> inputShape) {
     return SparseReorder.create(scope, inputIndices, inputValues, inputShape);
   }
@@ -948,8 +870,8 @@ public SparseReshape sparseReshape(Operand<TInt64> inputIndices, Operand<TInt64>
    * @param segmentIds A 1-D tensor. Values should be sorted and can be repeated.
    * @return a new instance of SparseSegmentMean
    */
-  public <T extends TNumber, U extends TNumber, V extends TNumber> SparseSegmentMean<T> sparseSegmentMean(
-      Operand<T> data, Operand<U> indices, Operand<V> segmentIds) {
+  public <T extends Tensor & TNumber, U extends Tensor & TNumber> SparseSegmentMean<T> sparseSegmentMean(
+      Operand<T> data, Operand<U> indices, Operand<TInt32> segmentIds) {
     return SparseSegmentMean.create(scope, data, indices, segmentIds);
   }
 
@@ -966,8 +888,8 @@ public <T extends TNumber, U extends TNumber, V extends TNumber> SparseSegmentMe
    * @param outputDim0 dimension 0 of "data" passed to SparseSegmentMean op.
    * @return a new instance of SparseSegmentMeanGrad
    */
-  public <T extends TNumber, U extends TNumber, V extends TNumber> SparseSegmentMeanGrad<T> sparseSegmentMeanGrad(
-      Operand<T> grad, Operand<U> indices, Operand<V> segmentIds, Operand<TInt32> outputDim0) {
+  public <T extends Tensor & TNumber, U extends Tensor & TNumber> SparseSegmentMeanGrad<T> sparseSegmentMeanGrad(
+      Operand<T> grad, Operand<U> indices, Operand<TInt32> segmentIds, Operand<TInt32> outputDim0) {
     return SparseSegmentMeanGrad.create(scope, grad, indices, segmentIds, outputDim0);
   }
 
@@ -975,7 +897,7 @@ public <T extends TNumber, U extends TNumber, V extends TNumber> SparseSegmentMe
    * Computes the mean along sparse segments of a tensor.
    *  <p>
    *  Like `SparseSegmentMean`, but allows missing ids in `segment_ids`. If an id is
-   *  missing, the `output` tensor at that position will be zeroed.
+   *  misisng, the `output` tensor at that position will be zeroed.
    *  <p>
    *  Read
    *  [the section on segmentation](https://tensorflow.org/api_docs/python/tf/math#Segmentation)
@@ -988,8 +910,8 @@ public <T extends TNumber, U extends TNumber, V extends TNumber> SparseSegmentMe
    * @param numSegments Should equal the number of distinct segment IDs.
    * @return a new instance of SparseSegmentMeanWithNumSegments
    */
-  public <T extends TNumber, U extends TNumber, V extends TNumber, W extends TNumber> SparseSegmentMeanWithNumSegments<T> sparseSegmentMeanWithNumSegments(
-      Operand<T> data, Operand<U> indices, Operand<V> segmentIds, Operand<W> numSegments) {
+  public <T extends Tensor & TNumber, U extends Tensor & TNumber, V extends Tensor & TNumber> SparseSegmentMeanWithNumSegments<T> sparseSegmentMeanWithNumSegments(
+      Operand<T> data, Operand<U> indices, Operand<TInt32> segmentIds, Operand<V> numSegments) {
     return SparseSegmentMeanWithNumSegments.create(scope, data, indices, segmentIds, numSegments);
   }
 
@@ -1006,8 +928,8 @@ public <T extends TNumber, U extends TNumber, V extends TNumber, W extends TNumb
    * @param segmentIds A 1-D tensor. Values should be sorted and can be repeated.
    * @return a new instance of SparseSegmentSqrtN
    */
-  public <T extends TNumber, U extends TNumber, V extends TNumber> SparseSegmentSqrtN<T> sparseSegmentSqrtN(
-      Operand<T> data, Operand<U> indices, Operand<V> segmentIds) {
+  public <T extends Tensor & TNumber, U extends Tensor & TNumber> SparseSegmentSqrtN<T> sparseSegmentSqrtN(
+      Operand<T> data, Operand<U> indices, Operand<TInt32> segmentIds) {
     return SparseSegmentSqrtN.create(scope, data, indices, segmentIds);
   }
 
@@ -1024,8 +946,8 @@ public <T extends TNumber, U extends TNumber, V extends TNumber> SparseSegmentSq
    * @param outputDim0 dimension 0 of "data" passed to SparseSegmentSqrtN op.
    * @return a new instance of SparseSegmentSqrtNGrad
    */
-  public <T extends TNumber, U extends TNumber, V extends TNumber> SparseSegmentSqrtNGrad<T> sparseSegmentSqrtNGrad(
-      Operand<T> grad, Operand<U> indices, Operand<V> segmentIds, Operand<TInt32> outputDim0) {
+  public <T extends Tensor & TNumber, U extends Tensor & TNumber> SparseSegmentSqrtNGrad<T> sparseSegmentSqrtNGrad(
+      Operand<T> grad, Operand<U> indices, Operand<TInt32> segmentIds, Operand<TInt32> outputDim0) {
     return SparseSegmentSqrtNGrad.create(scope, grad, indices, segmentIds, outputDim0);
   }
 
@@ -1035,7 +957,7 @@ public <T extends TNumber, U extends TNumber, V extends TNumber> SparseSegmentSq
    *  N is the size of the segment being reduced.
    *  <p>
    *  Like `SparseSegmentSqrtN`, but allows missing ids in `segment_ids`. If an id is
-   *  missing, the `output` tensor at that position will be zeroed.
+   *  misisng, the `output` tensor at that position will be zeroed.
    *  <p>
    *  Read
    *  [the section on segmentation](https://tensorflow.org/api_docs/python/tf/math#Segmentation)
@@ -1048,8 +970,8 @@ public <T extends TNumber, U extends TNumber, V extends TNumber> SparseSegmentSq
    * @param numSegments Should equal the number of distinct segment IDs.
    * @return a new instance of SparseSegmentSqrtNWithNumSegments
    */
-  public <T extends TNumber, U extends TNumber, V extends TNumber, W extends TNumber> SparseSegmentSqrtNWithNumSegments<T> sparseSegmentSqrtNWithNumSegments(
-      Operand<T> data, Operand<U> indices, Operand<V> segmentIds, Operand<W> numSegments) {
+  public <T extends Tensor & TNumber, U extends Tensor & TNumber, V extends Tensor & TNumber> SparseSegmentSqrtNWithNumSegments<T> sparseSegmentSqrtNWithNumSegments(
+      Operand<T> data, Operand<U> indices, Operand<TInt32> segmentIds, Operand<V> numSegments) {
     return SparseSegmentSqrtNWithNumSegments.create(scope, data, indices, segmentIds, numSegments);
   }
 
@@ -1091,8 +1013,8 @@ public <T extends TNumber, U extends TNumber, V extends TNumber, W extends TNumb
    * @param segmentIds A 1-D tensor. Values should be sorted and can be repeated.
    * @return a new instance of SparseSegmentSum
    */
-  public <T extends TNumber, U extends TNumber, V extends TNumber> SparseSegmentSum<T> sparseSegmentSum(
-      Operand<T> data, Operand<U> indices, Operand<V> segmentIds) {
+  public <T extends Tensor & TNumber, U extends Tensor & TNumber> SparseSegmentSum<T> sparseSegmentSum(
+      Operand<T> data, Operand<U> indices, Operand<TInt32> segmentIds) {
     return SparseSegmentSum.create(scope, data, indices, segmentIds);
   }
 
@@ -1100,7 +1022,7 @@ public <T extends TNumber, U extends TNumber, V extends TNumber> SparseSegmentSu
    * Computes the sum along sparse segments of a tensor.
    *  <p>
    *  Like `SparseSegmentSum`, but allows missing ids in `segment_ids`. If an id is
-   *  missing, the `output` tensor at that position will be zeroed.
+   *  misisng, the `output` tensor at that position will be zeroed.
    *  <p>
    *  Read
    *  [the section on segmentation](https://tensorflow.org/api_docs/python/tf/sparse#Segmentation)
@@ -1133,8 +1055,8 @@ public <T extends TNumber, U extends TNumber, V extends TNumber> SparseSegmentSu
    * @param numSegments Should equal the number of distinct segment IDs.
    * @return a new instance of SparseSegmentSumWithNumSegments
    */
-  public <T extends TNumber, U extends TNumber, V extends TNumber, W extends TNumber> SparseSegmentSumWithNumSegments<T> sparseSegmentSumWithNumSegments(
-      Operand<T> data, Operand<U> indices, Operand<V> segmentIds, Operand<W> numSegments) {
+  public <T extends Tensor & TNumber, U extends Tensor & TNumber, V extends Tensor & TNumber> SparseSegmentSumWithNumSegments<T> sparseSegmentSumWithNumSegments(
+      Operand<T> data, Operand<U> indices, Operand<TInt32> segmentIds, Operand<V> numSegments) {
     return SparseSegmentSumWithNumSegments.create(scope, data, indices, segmentIds, numSegments);
   }
 
@@ -1167,7 +1089,7 @@ public <T extends TNumber, U extends TNumber, V extends TNumber, W extends TNumb
    *  sparse tensors.
    * @return a new instance of SparseSlice
    */
-  public <T extends TType> SparseSlice<T> sparseSlice(Operand<TInt64> indices, Operand<T> values,
+  public <T extends Tensor> SparseSlice<T> sparseSlice(Operand<TInt64> indices, Operand<T> values,
       Operand<TInt64> shape, Operand<TInt64> start, Operand<TInt64> size) {
     return SparseSlice.create(scope, indices, values, shape, start, size);
   }
@@ -1187,7 +1109,7 @@ public <T extends TType> SparseSlice<T> sparseSlice(Operand<TInt64> indices, Ope
    * @param outputIndices 2-D.  The `indices` of the sliced `SparseTensor`.
    * @return a new instance of SparseSliceGrad
    */
-  public <T extends TType> SparseSliceGrad<T> sparseSliceGrad(Operand<T> backpropValGrad,
+  public <T extends Tensor> SparseSliceGrad<T> sparseSliceGrad(Operand<T> backpropValGrad,
       Operand<TInt64> inputIndices, Operand<TInt64> inputStart, Operand<TInt64> outputIndices) {
     return SparseSliceGrad.create(scope, backpropValGrad, inputIndices, inputStart, outputIndices);
   }
@@ -1218,7 +1140,7 @@ public <T extends TType> SparseSliceGrad<T> sparseSliceGrad(Operand<T> backpropV
    * @param spShape 1-D.  Shape of the input SparseTensor.
    * @return a new instance of SparseSoftmax
    */
-  public <T extends TNumber> SparseSoftmax<T> sparseSoftmax(Operand<TInt64> spIndices,
+  public <T extends Tensor & TNumber> SparseSoftmax<T> sparseSoftmax(Operand<TInt64> spIndices,
       Operand<T> spValues, Operand<TInt64> spShape) {
     return SparseSoftmax.create(scope, spIndices, spValues, spShape);
   }
@@ -1238,9 +1160,9 @@ public <T extends TNumber> SparseSoftmax<T> sparseSoftmax(Operand<TInt64> spIndi
    * @param bShape counterpart to `a_shape` for the other operand; the two shapes must be equal.
    * @return a new instance of SparseSparseMaximum
    */
-  public <T extends TNumber> SparseSparseMaximum<T> sparseSparseMaximum(Operand<TInt64> aIndices,
-      Operand<T> aValues, Operand<TInt64> aShape, Operand<TInt64> bIndices, Operand<T> bValues,
-      Operand<TInt64> bShape) {
+  public <T extends Tensor & TNumber> SparseSparseMaximum<T> sparseSparseMaximum(
+      Operand<TInt64> aIndices, Operand<T> aValues, Operand<TInt64> aShape,
+      Operand<TInt64> bIndices, Operand<T> bValues, Operand<TInt64> bShape) {
     return SparseSparseMaximum.create(scope, aIndices, aValues, aShape, bIndices, bValues, bShape);
   }
 
@@ -1259,7 +1181,7 @@ public <T extends TNumber> SparseSparseMaximum<T> sparseSparseMaximum(Operand<TI
    * @param bShape counterpart to `a_shape` for the other operand; the two shapes must be equal.
    * @return a new instance of SparseSparseMinimum
    */
-  public <T extends TType> SparseSparseMinimum<T> sparseSparseMinimum(Operand<TInt64> aIndices,
+  public <T extends Tensor> SparseSparseMinimum<T> sparseSparseMinimum(Operand<TInt64> aIndices,
       Operand<T> aValues, Operand<TInt64> aShape, Operand<TInt64> bIndices, Operand<T> bValues,
       Operand<TInt64> bShape) {
     return SparseSparseMinimum.create(scope, aIndices, aValues, aShape, bIndices, bValues, bShape);
@@ -1297,7 +1219,7 @@ public <T extends TType> SparseSparseMinimum<T> sparseSparseMinimum(Operand<TInt
    * @param numSplit The number of ways to split.
    * @return a new instance of SparseSplit
    */
-  public <T extends TType> SparseSplit<T> sparseSplit(Operand<TInt64> splitDim,
+  public <T extends Tensor> SparseSplit<T> sparseSplit(Operand<TInt64> splitDim,
       Operand<TInt64> indices, Operand<T> values, Operand<TInt64> shape, Long numSplit) {
     return SparseSplit.create(scope, splitDim, indices, values, shape, numSplit);
   }
@@ -1314,7 +1236,7 @@ public <T extends TType> SparseSplit<T> sparseSplit(Operand<TInt64> splitDim,
    * @param b `ndims`-D Tensor.  With shape `a_shape`.
    * @return a new instance of SparseTensorDenseAdd
    */
-  public <U extends TType, T extends TNumber> SparseTensorDenseAdd<U> sparseTensorDenseAdd(
+  public <U extends Tensor, T extends Tensor & TNumber> SparseTensorDenseAdd<U> sparseTensorDenseAdd(
       Operand<T> aIndices, Operand<U> aValues, Operand<T> aShape, Operand<U> b) {
     return SparseTensorDenseAdd.create(scope, aIndices, aValues, aShape, b);
   }
@@ -1340,7 +1262,7 @@ public <U extends TType, T extends TNumber> SparseTensorDenseAdd<U> sparseTensor
    * @param options carries optional attributes values
    * @return a new instance of SparseTensorDenseMatMul
    */
-  public <U extends TType, T extends TNumber> SparseTensorDenseMatMul<U> sparseTensorDenseMatMul(
+  public <U extends Tensor, T extends Tensor & TNumber> SparseTensorDenseMatMul<U> sparseTensorDenseMatMul(
       Operand<T> aIndices, Operand<U> aValues, Operand<TInt64> aShape, Operand<U> b,
       SparseTensorDenseMatMul.Options... options) {
     return SparseTensorDenseMatMul.create(scope, aIndices, aValues, aShape, b, options);
@@ -1378,7 +1300,7 @@ public <U extends TType, T extends TNumber> SparseTensorDenseMatMul<U> sparseTen
    * @param options carries optional attributes values
    * @return a new instance of SparseToDense
    */
-  public <U extends TType, T extends TNumber> SparseToDense<U> sparseToDense(
+  public <U extends Tensor, T extends Tensor & TNumber> SparseToDense<U> sparseToDense(
       Operand<T> sparseIndices, Operand<T> outputShape, Operand<U> sparseValues,
       Operand<U> defaultValue, SparseToDense.Options... options) {
     return SparseToDense.create(scope, sparseIndices, outputShape, sparseValues, defaultValue, options);
@@ -1430,7 +1352,7 @@ public <U extends TType, T extends TNumber> SparseToDense<U> sparseToDense(
    * @param options carries optional attributes values
    * @return a new instance of SparseToSparseSetOperation
    */
-  public <T extends TType> SparseToSparseSetOperation<T> sparseToSparseSetOperation(
+  public <T extends Tensor> SparseToSparseSetOperation<T> sparseToSparseSetOperation(
       Operand<TInt64> set1Indices, Operand<T> set1Values, Operand<TInt64> set1Shape,
       Operand<TInt64> set2Indices, Operand<T> set2Values, Operand<TInt64> set2Shape,
       String setOperation, SparseToSparseSetOperation.Options... options) {
@@ -1492,7 +1414,7 @@ public <T extends TType> SparseToSparseSetOperation<T> sparseToSparseSetOperatio
    * @param options carries optional attributes values
    * @return a new instance of TakeManySparseFromTensorsMap
    */
-  public <T extends TType> TakeManySparseFromTensorsMap<T> takeManySparseFromTensorsMap(
+  public <T extends Tensor> TakeManySparseFromTensorsMap<T> takeManySparseFromTensorsMap(
       Operand<TInt64> sparseHandles, DataType<T> dtype,
       TakeManySparseFromTensorsMap.Options... options) {
     return TakeManySparseFromTensorsMap.create(scope, sparseHandles, dtype, options);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/StringsOps.java b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/StringsOps.java
index f6491843332..5e2be5647c5 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/StringsOps.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/StringsOps.java
@@ -20,6 +20,7 @@
 import java.util.List;
 import org.tensorflow.DataType;
 import org.tensorflow.Operand;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.strings.Join;
 import org.tensorflow.op.strings.Lower;
 import org.tensorflow.op.strings.ReduceJoin;
@@ -230,7 +231,7 @@ public StringLength stringLength(Operand<TString> input, StringLength.Options...
    * @param preserveShortSequences
    * @return a new instance of StringNGrams
    */
-  public <T extends TNumber> StringNGrams<T> stringNGrams(Operand<TString> data,
+  public <T extends Tensor & TNumber> StringNGrams<T> stringNGrams(Operand<TString> data,
       Operand<T> dataSplits, String separator, List<Long> ngramWidths, String leftPad,
       String rightPad, Long padWidth, Boolean preserveShortSequences) {
     return StringNGrams.create(scope, data, dataSplits, separator, ngramWidths, leftPad, rightPad, padWidth, preserveShortSequences);
@@ -367,8 +368,8 @@ public Strip strip(Operand<TString> input) {
    * @param options carries optional attributes values
    * @return a new instance of Substr
    */
-  public <T extends TNumber> Substr substr(Operand<TString> input, Operand<T> pos, Operand<T> len,
-      Substr.Options... options) {
+  public <T extends Tensor & TNumber> Substr substr(Operand<TString> input, Operand<T> pos,
+      Operand<T> len, Substr.Options... options) {
     return Substr.create(scope, input, pos, len, options);
   }
 
@@ -483,7 +484,7 @@ public ToNumber<TFloat32> toNumber(Operand<TString> stringTensor) {
    * @param outType The numeric type to interpret each string in `string_tensor` as.
    * @return a new instance of ToNumber
    */
-  public <T extends TNumber> ToNumber<T> toNumber(Operand<TString> stringTensor,
+  public <T extends Tensor & TNumber> ToNumber<T> toNumber(Operand<TString> stringTensor,
       DataType<T> outType) {
     return ToNumber.create(scope, stringTensor, outType);
   }
@@ -596,7 +597,7 @@ public UnicodeTranscode unicodeTranscode(Operand<TString> input, String inputEnc
    * @param options carries optional attributes values
    * @return a new instance of UnsortedSegmentJoin
    */
-  public <T extends TNumber, U extends TNumber> UnsortedSegmentJoin unsortedSegmentJoin(
+  public <T extends Tensor & TNumber, U extends Tensor & TNumber> UnsortedSegmentJoin unsortedSegmentJoin(
       Operand<TString> inputs, Operand<T> segmentIds, Operand<U> numSegments,
       UnsortedSegmentJoin.Options... options) {
     return UnsortedSegmentJoin.create(scope, inputs, segmentIds, numSegments, options);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/SummaryOps.java b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/SummaryOps.java
index 6143335b8fd..ce8b445610d 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/SummaryOps.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/SummaryOps.java
@@ -18,6 +18,7 @@
 package org.tensorflow.op;
 
 import org.tensorflow.Operand;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.summary.AudioSummary;
 import org.tensorflow.op.summary.HistogramSummary;
 import org.tensorflow.op.summary.ImageSummary;
@@ -27,7 +28,6 @@
 import org.tensorflow.types.TFloat32;
 import org.tensorflow.types.TString;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * An API for building {@code summary} operations as {@link Op Op}s
@@ -83,7 +83,7 @@ public AudioSummary audioSummary(Operand<TString> tag, Operand<TFloat32> tensor,
    * @param values Any shape. Values to use to build the histogram.
    * @return a new instance of HistogramSummary
    */
-  public <T extends TNumber> HistogramSummary histogramSummary(Operand<TString> tag,
+  public <T extends Tensor & TNumber> HistogramSummary histogramSummary(Operand<TString> tag,
       Operand<T> values) {
     return HistogramSummary.create(scope, tag, values);
   }
@@ -144,8 +144,8 @@ public <T extends TNumber> HistogramSummary histogramSummary(Operand<TString> ta
    * @param options carries optional attributes values
    * @return a new instance of ImageSummary
    */
-  public <T extends TNumber> ImageSummary imageSummary(Operand<TString> tag, Operand<T> tensor,
-      ImageSummary.Options... options) {
+  public <T extends Tensor & TNumber> ImageSummary imageSummary(Operand<TString> tag,
+      Operand<T> tensor, ImageSummary.Options... options) {
     return ImageSummary.create(scope, tag, tensor, options);
   }
 
@@ -178,7 +178,8 @@ public MergeSummary mergeSummary(Iterable<Operand<TString>> inputs) {
    * @param values Same shape as `tags.  Values for the summary.
    * @return a new instance of ScalarSummary
    */
-  public <T extends TNumber> ScalarSummary scalarSummary(Operand<TString> tags, Operand<T> values) {
+  public <T extends Tensor & TNumber> ScalarSummary scalarSummary(Operand<TString> tags,
+      Operand<T> values) {
     return ScalarSummary.create(scope, tags, values);
   }
 
@@ -191,7 +192,7 @@ public <T extends TNumber> ScalarSummary scalarSummary(Operand<TString> tags, Op
    *  data.
    * @return a new instance of TensorSummary
    */
-  public <T extends TType> TensorSummary tensorSummary(Operand<TString> tag, Operand<T> tensor,
+  public <T extends Tensor> TensorSummary tensorSummary(Operand<TString> tag, Operand<T> tensor,
       Operand<TString> serializedSummaryMetadata) {
     return TensorSummary.create(scope, tag, tensor, serializedSummaryMetadata);
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/TrainOps.java b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/TrainOps.java
index 2c5d8752136..e971180cd5d 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/TrainOps.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/TrainOps.java
@@ -20,6 +20,7 @@
 import java.util.List;
 import org.tensorflow.DataType;
 import org.tensorflow.Operand;
+import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.train.AccumulatorApplyGradient;
 import org.tensorflow.op.train.AccumulatorNumAccumulated;
@@ -88,7 +89,6 @@
 import org.tensorflow.types.TInt64;
 import org.tensorflow.types.TString;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * An API for building {@code train} operations as {@link Op Op}s
@@ -112,7 +112,7 @@ public final class TrainOps {
    * @param gradient A tensor of the gradient to be accumulated.
    * @return a new instance of AccumulatorApplyGradient
    */
-  public <T extends TType> AccumulatorApplyGradient accumulatorApplyGradient(
+  public <T extends Tensor> AccumulatorApplyGradient accumulatorApplyGradient(
       Operand<TString> handle, Operand<TInt64> localStep, Operand<T> gradient) {
     return AccumulatorApplyGradient.create(scope, handle, localStep, gradient);
   }
@@ -158,7 +158,7 @@ public AccumulatorSetGlobalStep accumulatorSetGlobalStep(Operand<TString> handle
    *  of the accumulator.
    * @return a new instance of AccumulatorTakeGradient
    */
-  public <T extends TType> AccumulatorTakeGradient<T> accumulatorTakeGradient(
+  public <T extends Tensor> AccumulatorTakeGradient<T> accumulatorTakeGradient(
       Operand<TString> handle, Operand<TInt32> numRequired, DataType<T> dtype) {
     return AccumulatorTakeGradient.create(scope, handle, numRequired, dtype);
   }
@@ -182,7 +182,7 @@ public <T extends TType> AccumulatorTakeGradient<T> accumulatorTakeGradient(
    * @param options carries optional attributes values
    * @return a new instance of ApplyAdadelta
    */
-  public <T extends TType> ApplyAdadelta<T> applyAdadelta(Operand<T> var, Operand<T> accum,
+  public <T extends Tensor> ApplyAdadelta<T> applyAdadelta(Operand<T> var, Operand<T> accum,
       Operand<T> accumUpdate, Operand<T> lr, Operand<T> rho, Operand<T> epsilon, Operand<T> grad,
       ApplyAdadelta.Options... options) {
     return ApplyAdadelta.create(scope, var, accum, accumUpdate, lr, rho, epsilon, grad, options);
@@ -202,7 +202,7 @@ public <T extends TType> ApplyAdadelta<T> applyAdadelta(Operand<T> var, Operand<
    * @param options carries optional attributes values
    * @return a new instance of ApplyAdagrad
    */
-  public <T extends TType> ApplyAdagrad<T> applyAdagrad(Operand<T> var, Operand<T> accum,
+  public <T extends Tensor> ApplyAdagrad<T> applyAdagrad(Operand<T> var, Operand<T> accum,
       Operand<T> lr, Operand<T> grad, ApplyAdagrad.Options... options) {
     return ApplyAdagrad.create(scope, var, accum, lr, grad, options);
   }
@@ -222,7 +222,7 @@ public <T extends TType> ApplyAdagrad<T> applyAdagrad(Operand<T> var, Operand<T>
    * @param options carries optional attributes values
    * @return a new instance of ApplyAdagradDa
    */
-  public <T extends TType> ApplyAdagradDa<T> applyAdagradDa(Operand<T> var,
+  public <T extends Tensor> ApplyAdagradDa<T> applyAdagradDa(Operand<T> var,
       Operand<T> gradientAccumulator, Operand<T> gradientSquaredAccumulator, Operand<T> grad,
       Operand<T> lr, Operand<T> l1, Operand<T> l2, Operand<TInt64> globalStep,
       ApplyAdagradDa.Options... options) {
@@ -251,7 +251,7 @@ public <T extends TType> ApplyAdagradDa<T> applyAdagradDa(Operand<T> var,
    * @param options carries optional attributes values
    * @return a new instance of ApplyAdam
    */
-  public <T extends TType> ApplyAdam<T> applyAdam(Operand<T> var, Operand<T> m, Operand<T> v,
+  public <T extends Tensor> ApplyAdam<T> applyAdam(Operand<T> var, Operand<T> m, Operand<T> v,
       Operand<T> beta1Power, Operand<T> beta2Power, Operand<T> lr, Operand<T> beta1,
       Operand<T> beta2, Operand<T> epsilon, Operand<T> grad, ApplyAdam.Options... options) {
     return ApplyAdam.create(scope, var, m, v, beta1Power, beta2Power, lr, beta1, beta2, epsilon, grad, options);
@@ -275,8 +275,8 @@ public <T extends TType> ApplyAdam<T> applyAdam(Operand<T> var, Operand<T> m, Op
    * @param options carries optional attributes values
    * @return a new instance of ApplyAddSign
    */
-  public <T extends TType> ApplyAddSign<T> applyAddSign(Operand<T> var, Operand<T> m, Operand<T> lr,
-      Operand<T> alpha, Operand<T> signDecay, Operand<T> beta, Operand<T> grad,
+  public <T extends Tensor> ApplyAddSign<T> applyAddSign(Operand<T> var, Operand<T> m,
+      Operand<T> lr, Operand<T> alpha, Operand<T> signDecay, Operand<T> beta, Operand<T> grad,
       ApplyAddSign.Options... options) {
     return ApplyAddSign.create(scope, var, m, lr, alpha, signDecay, beta, grad, options);
   }
@@ -316,7 +316,7 @@ public <T extends TType> ApplyAddSign<T> applyAddSign(Operand<T> var, Operand<T>
    * @param options carries optional attributes values
    * @return a new instance of ApplyCenteredRmsProp
    */
-  public <T extends TType> ApplyCenteredRmsProp<T> applyCenteredRmsProp(Operand<T> var,
+  public <T extends Tensor> ApplyCenteredRmsProp<T> applyCenteredRmsProp(Operand<T> var,
       Operand<T> mg, Operand<T> ms, Operand<T> mom, Operand<T> lr, Operand<T> rho,
       Operand<T> momentum, Operand<T> epsilon, Operand<T> grad,
       ApplyCenteredRmsProp.Options... options) {
@@ -347,7 +347,7 @@ public <T extends TType> ApplyCenteredRmsProp<T> applyCenteredRmsProp(Operand<T>
    * @param options carries optional attributes values
    * @return a new instance of ApplyFtrl
    */
-  public <T extends TType> ApplyFtrl<T> applyFtrl(Operand<T> var, Operand<T> accum,
+  public <T extends Tensor> ApplyFtrl<T> applyFtrl(Operand<T> var, Operand<T> accum,
       Operand<T> linear, Operand<T> grad, Operand<T> lr, Operand<T> l1, Operand<T> l2,
       Operand<T> l2Shrinkage, Operand<T> lrPower, ApplyFtrl.Options... options) {
     return ApplyFtrl.create(scope, var, accum, linear, grad, lr, l1, l2, l2Shrinkage, lrPower, options);
@@ -363,7 +363,7 @@ public <T extends TType> ApplyFtrl<T> applyFtrl(Operand<T> var, Operand<T> accum
    * @param options carries optional attributes values
    * @return a new instance of ApplyGradientDescent
    */
-  public <T extends TType> ApplyGradientDescent<T> applyGradientDescent(Operand<T> var,
+  public <T extends Tensor> ApplyGradientDescent<T> applyGradientDescent(Operand<T> var,
       Operand<T> alpha, Operand<T> delta, ApplyGradientDescent.Options... options) {
     return ApplyGradientDescent.create(scope, var, alpha, delta, options);
   }
@@ -385,7 +385,7 @@ public <T extends TType> ApplyGradientDescent<T> applyGradientDescent(Operand<T>
    * @param options carries optional attributes values
    * @return a new instance of ApplyMomentum
    */
-  public <T extends TType> ApplyMomentum<T> applyMomentum(Operand<T> var, Operand<T> accum,
+  public <T extends Tensor> ApplyMomentum<T> applyMomentum(Operand<T> var, Operand<T> accum,
       Operand<T> lr, Operand<T> grad, Operand<T> momentum, ApplyMomentum.Options... options) {
     return ApplyMomentum.create(scope, var, accum, lr, grad, momentum, options);
   }
@@ -408,7 +408,7 @@ public <T extends TType> ApplyMomentum<T> applyMomentum(Operand<T> var, Operand<
    * @param options carries optional attributes values
    * @return a new instance of ApplyPowerSign
    */
-  public <T extends TType> ApplyPowerSign<T> applyPowerSign(Operand<T> var, Operand<T> m,
+  public <T extends Tensor> ApplyPowerSign<T> applyPowerSign(Operand<T> var, Operand<T> m,
       Operand<T> lr, Operand<T> logbase, Operand<T> signDecay, Operand<T> beta, Operand<T> grad,
       ApplyPowerSign.Options... options) {
     return ApplyPowerSign.create(scope, var, m, lr, logbase, signDecay, beta, grad, options);
@@ -431,7 +431,7 @@ public <T extends TType> ApplyPowerSign<T> applyPowerSign(Operand<T> var, Operan
    * @param options carries optional attributes values
    * @return a new instance of ApplyProximalAdagrad
    */
-  public <T extends TType> ApplyProximalAdagrad<T> applyProximalAdagrad(Operand<T> var,
+  public <T extends Tensor> ApplyProximalAdagrad<T> applyProximalAdagrad(Operand<T> var,
       Operand<T> accum, Operand<T> lr, Operand<T> l1, Operand<T> l2, Operand<T> grad,
       ApplyProximalAdagrad.Options... options) {
     return ApplyProximalAdagrad.create(scope, var, accum, lr, l1, l2, grad, options);
@@ -452,7 +452,7 @@ public <T extends TType> ApplyProximalAdagrad<T> applyProximalAdagrad(Operand<T>
    * @param options carries optional attributes values
    * @return a new instance of ApplyProximalGradientDescent
    */
-  public <T extends TType> ApplyProximalGradientDescent<T> applyProximalGradientDescent(
+  public <T extends Tensor> ApplyProximalGradientDescent<T> applyProximalGradientDescent(
       Operand<T> var, Operand<T> alpha, Operand<T> l1, Operand<T> l2, Operand<T> delta,
       ApplyProximalGradientDescent.Options... options) {
     return ApplyProximalGradientDescent.create(scope, var, alpha, l1, l2, delta, options);
@@ -484,7 +484,7 @@ public <T extends TType> ApplyProximalGradientDescent<T> applyProximalGradientDe
    * @param options carries optional attributes values
    * @return a new instance of ApplyRmsProp
    */
-  public <T extends TType> ApplyRmsProp<T> applyRmsProp(Operand<T> var, Operand<T> ms,
+  public <T extends Tensor> ApplyRmsProp<T> applyRmsProp(Operand<T> var, Operand<T> ms,
       Operand<T> mom, Operand<T> lr, Operand<T> rho, Operand<T> momentum, Operand<T> epsilon,
       Operand<T> grad, ApplyRmsProp.Options... options) {
     return ApplyRmsProp.create(scope, var, ms, mom, lr, rho, momentum, epsilon, grad, options);
@@ -522,7 +522,7 @@ public <T extends TType> ApplyRmsProp<T> applyRmsProp(Operand<T> var, Operand<T>
    * @param options carries optional attributes values
    * @return a new instance of BatchMatMul
    */
-  public <T extends TType> BatchMatMul<T> batchMatMul(Operand<T> x, Operand<T> y,
+  public <T extends Tensor> BatchMatMul<T> batchMatMul(Operand<T> x, Operand<T> y,
       BatchMatMul.Options... options) {
     return BatchMatMul.create(scope, x, y, options);
   }
@@ -542,7 +542,7 @@ public <T extends TType> BatchMatMul<T> batchMatMul(Operand<T> x, Operand<T> y,
    * @param options carries optional attributes values
    * @return a new instance of ConditionalAccumulator
    */
-  public <T extends TType> ConditionalAccumulator conditionalAccumulator(DataType<T> dtype,
+  public <T extends Tensor> ConditionalAccumulator conditionalAccumulator(DataType<T> dtype,
       Shape shape, ConditionalAccumulator.Options... options) {
     return ConditionalAccumulator.create(scope, dtype, shape, options);
   }
@@ -649,7 +649,7 @@ public NegTrain negTrain(Operand<TFloat32> wIn, Operand<TFloat32> wOut, Operand<
    * @param options carries optional attributes values
    * @return a new instance of PreventGradient
    */
-  public <T extends TType> PreventGradient<T> preventGradient(Operand<T> input,
+  public <T extends Tensor> PreventGradient<T> preventGradient(Operand<T> input,
       PreventGradient.Options... options) {
     return PreventGradient.create(scope, input, options);
   }
@@ -672,7 +672,7 @@ public <T extends TType> PreventGradient<T> preventGradient(Operand<T> input,
    * @param options carries optional attributes values
    * @return a new instance of ResourceApplyAdadelta
    */
-  public <T extends TType> ResourceApplyAdadelta resourceApplyAdadelta(Operand<?> var,
+  public <T extends Tensor> ResourceApplyAdadelta resourceApplyAdadelta(Operand<?> var,
       Operand<?> accum, Operand<?> accumUpdate, Operand<T> lr, Operand<T> rho, Operand<T> epsilon,
       Operand<T> grad, ResourceApplyAdadelta.Options... options) {
     return ResourceApplyAdadelta.create(scope, var, accum, accumUpdate, lr, rho, epsilon, grad, options);
@@ -692,7 +692,7 @@ public <T extends TType> ResourceApplyAdadelta resourceApplyAdadelta(Operand<?>
    * @param options carries optional attributes values
    * @return a new instance of ResourceApplyAdagradDa
    */
-  public <T extends TType> ResourceApplyAdagradDa resourceApplyAdagradDa(Operand<?> var,
+  public <T extends Tensor> ResourceApplyAdagradDa resourceApplyAdagradDa(Operand<?> var,
       Operand<?> gradientAccumulator, Operand<?> gradientSquaredAccumulator, Operand<T> grad,
       Operand<T> lr, Operand<T> l1, Operand<T> l2, Operand<TInt64> globalStep,
       ResourceApplyAdagradDa.Options... options) {
@@ -720,7 +720,7 @@ public <T extends TType> ResourceApplyAdagradDa resourceApplyAdagradDa(Operand<?
    * @param options carries optional attributes values
    * @return a new instance of ResourceApplyAdam
    */
-  public <T extends TType> ResourceApplyAdam resourceApplyAdam(Operand<?> var, Operand<?> m,
+  public <T extends Tensor> ResourceApplyAdam resourceApplyAdam(Operand<?> var, Operand<?> m,
       Operand<?> v, Operand<T> beta1Power, Operand<T> beta2Power, Operand<T> lr, Operand<T> beta1,
       Operand<T> beta2, Operand<T> epsilon, Operand<T> grad, ResourceApplyAdam.Options... options) {
     return ResourceApplyAdam.create(scope, var, m, v, beta1Power, beta2Power, lr, beta1, beta2, epsilon, grad, options);
@@ -749,10 +749,10 @@ public <T extends TType> ResourceApplyAdam resourceApplyAdam(Operand<?> var, Ope
    * @param options carries optional attributes values
    * @return a new instance of ResourceApplyAdamWithAmsgrad
    */
-  public <T extends TType> ResourceApplyAdamWithAmsgrad resourceApplyAdamWithAmsgrad(Operand<?> var,
-      Operand<?> m, Operand<?> v, Operand<?> vhat, Operand<T> beta1Power, Operand<T> beta2Power,
-      Operand<T> lr, Operand<T> beta1, Operand<T> beta2, Operand<T> epsilon, Operand<T> grad,
-      ResourceApplyAdamWithAmsgrad.Options... options) {
+  public <T extends Tensor> ResourceApplyAdamWithAmsgrad resourceApplyAdamWithAmsgrad(
+      Operand<?> var, Operand<?> m, Operand<?> v, Operand<?> vhat, Operand<T> beta1Power,
+      Operand<T> beta2Power, Operand<T> lr, Operand<T> beta1, Operand<T> beta2, Operand<T> epsilon,
+      Operand<T> grad, ResourceApplyAdamWithAmsgrad.Options... options) {
     return ResourceApplyAdamWithAmsgrad.create(scope, var, m, v, vhat, beta1Power, beta2Power, lr, beta1, beta2, epsilon, grad, options);
   }
 
@@ -773,7 +773,7 @@ public <T extends TType> ResourceApplyAdamWithAmsgrad resourceApplyAdamWithAmsgr
    * @param options carries optional attributes values
    * @return a new instance of ResourceApplyAddSign
    */
-  public <T extends TType> ResourceApplyAddSign resourceApplyAddSign(Operand<?> var, Operand<?> m,
+  public <T extends Tensor> ResourceApplyAddSign resourceApplyAddSign(Operand<?> var, Operand<?> m,
       Operand<T> lr, Operand<T> alpha, Operand<T> signDecay, Operand<T> beta, Operand<T> grad,
       ResourceApplyAddSign.Options... options) {
     return ResourceApplyAddSign.create(scope, var, m, lr, alpha, signDecay, beta, grad, options);
@@ -813,8 +813,8 @@ public <T extends TType> ResourceApplyAddSign resourceApplyAddSign(Operand<?> va
    * @param options carries optional attributes values
    * @return a new instance of ResourceApplyCenteredRmsProp
    */
-  public <T extends TType> ResourceApplyCenteredRmsProp resourceApplyCenteredRmsProp(Operand<?> var,
-      Operand<?> mg, Operand<?> ms, Operand<?> mom, Operand<T> lr, Operand<T> rho,
+  public <T extends Tensor> ResourceApplyCenteredRmsProp resourceApplyCenteredRmsProp(
+      Operand<?> var, Operand<?> mg, Operand<?> ms, Operand<?> mom, Operand<T> lr, Operand<T> rho,
       Operand<T> momentum, Operand<T> epsilon, Operand<T> grad,
       ResourceApplyCenteredRmsProp.Options... options) {
     return ResourceApplyCenteredRmsProp.create(scope, var, mg, ms, mom, lr, rho, momentum, epsilon, grad, options);
@@ -843,7 +843,7 @@ public <T extends TType> ResourceApplyCenteredRmsProp resourceApplyCenteredRmsPr
    * @param options carries optional attributes values
    * @return a new instance of ResourceApplyFtrl
    */
-  public <T extends TType> ResourceApplyFtrl resourceApplyFtrl(Operand<?> var, Operand<?> accum,
+  public <T extends Tensor> ResourceApplyFtrl resourceApplyFtrl(Operand<?> var, Operand<?> accum,
       Operand<?> linear, Operand<T> grad, Operand<T> lr, Operand<T> l1, Operand<T> l2,
       Operand<T> l2Shrinkage, Operand<T> lrPower, ResourceApplyFtrl.Options... options) {
     return ResourceApplyFtrl.create(scope, var, accum, linear, grad, lr, l1, l2, l2Shrinkage, lrPower, options);
@@ -858,8 +858,9 @@ public <T extends TType> ResourceApplyFtrl resourceApplyFtrl(Operand<?> var, Ope
    * @param options carries optional attributes values
    * @return a new instance of ResourceApplyGradientDescent
    */
-  public <T extends TType> ResourceApplyGradientDescent resourceApplyGradientDescent(Operand<?> var,
-      Operand<T> alpha, Operand<T> delta, ResourceApplyGradientDescent.Options... options) {
+  public <T extends Tensor> ResourceApplyGradientDescent resourceApplyGradientDescent(
+      Operand<?> var, Operand<T> alpha, Operand<T> delta,
+      ResourceApplyGradientDescent.Options... options) {
     return ResourceApplyGradientDescent.create(scope, var, alpha, delta, options);
   }
 
@@ -879,7 +880,7 @@ public <T extends TType> ResourceApplyGradientDescent resourceApplyGradientDesce
    * @param options carries optional attributes values
    * @return a new instance of ResourceApplyKerasMomentum
    */
-  public <T extends TType> ResourceApplyKerasMomentum resourceApplyKerasMomentum(Operand<?> var,
+  public <T extends Tensor> ResourceApplyKerasMomentum resourceApplyKerasMomentum(Operand<?> var,
       Operand<?> accum, Operand<T> lr, Operand<T> grad, Operand<T> momentum,
       ResourceApplyKerasMomentum.Options... options) {
     return ResourceApplyKerasMomentum.create(scope, var, accum, lr, grad, momentum, options);
@@ -901,7 +902,7 @@ public <T extends TType> ResourceApplyKerasMomentum resourceApplyKerasMomentum(O
    * @param options carries optional attributes values
    * @return a new instance of ResourceApplyMomentum
    */
-  public <T extends TType> ResourceApplyMomentum resourceApplyMomentum(Operand<?> var,
+  public <T extends Tensor> ResourceApplyMomentum resourceApplyMomentum(Operand<?> var,
       Operand<?> accum, Operand<T> lr, Operand<T> grad, Operand<T> momentum,
       ResourceApplyMomentum.Options... options) {
     return ResourceApplyMomentum.create(scope, var, accum, lr, grad, momentum, options);
@@ -924,7 +925,7 @@ public <T extends TType> ResourceApplyMomentum resourceApplyMomentum(Operand<?>
    * @param options carries optional attributes values
    * @return a new instance of ResourceApplyPowerSign
    */
-  public <T extends TType> ResourceApplyPowerSign resourceApplyPowerSign(Operand<?> var,
+  public <T extends Tensor> ResourceApplyPowerSign resourceApplyPowerSign(Operand<?> var,
       Operand<?> m, Operand<T> lr, Operand<T> logbase, Operand<T> signDecay, Operand<T> beta,
       Operand<T> grad, ResourceApplyPowerSign.Options... options) {
     return ResourceApplyPowerSign.create(scope, var, m, lr, logbase, signDecay, beta, grad, options);
@@ -946,9 +947,9 @@ public <T extends TType> ResourceApplyPowerSign resourceApplyPowerSign(Operand<?
    * @param options carries optional attributes values
    * @return a new instance of ResourceApplyProximalAdagrad
    */
-  public <T extends TType> ResourceApplyProximalAdagrad resourceApplyProximalAdagrad(Operand<?> var,
-      Operand<?> accum, Operand<T> lr, Operand<T> l1, Operand<T> l2, Operand<T> grad,
-      ResourceApplyProximalAdagrad.Options... options) {
+  public <T extends Tensor> ResourceApplyProximalAdagrad resourceApplyProximalAdagrad(
+      Operand<?> var, Operand<?> accum, Operand<T> lr, Operand<T> l1, Operand<T> l2,
+      Operand<T> grad, ResourceApplyProximalAdagrad.Options... options) {
     return ResourceApplyProximalAdagrad.create(scope, var, accum, lr, l1, l2, grad, options);
   }
 
@@ -966,7 +967,7 @@ public <T extends TType> ResourceApplyProximalAdagrad resourceApplyProximalAdagr
    * @param options carries optional attributes values
    * @return a new instance of ResourceApplyProximalGradientDescent
    */
-  public <T extends TType> ResourceApplyProximalGradientDescent resourceApplyProximalGradientDescent(
+  public <T extends Tensor> ResourceApplyProximalGradientDescent resourceApplyProximalGradientDescent(
       Operand<?> var, Operand<T> alpha, Operand<T> l1, Operand<T> l2, Operand<T> delta,
       ResourceApplyProximalGradientDescent.Options... options) {
     return ResourceApplyProximalGradientDescent.create(scope, var, alpha, l1, l2, delta, options);
@@ -997,7 +998,7 @@ public <T extends TType> ResourceApplyProximalGradientDescent resourceApplyProxi
    * @param options carries optional attributes values
    * @return a new instance of ResourceApplyRmsProp
    */
-  public <T extends TType> ResourceApplyRmsProp resourceApplyRmsProp(Operand<?> var, Operand<?> ms,
+  public <T extends Tensor> ResourceApplyRmsProp resourceApplyRmsProp(Operand<?> var, Operand<?> ms,
       Operand<?> mom, Operand<T> lr, Operand<T> rho, Operand<T> momentum, Operand<T> epsilon,
       Operand<T> grad, ResourceApplyRmsProp.Options... options) {
     return ResourceApplyRmsProp.create(scope, var, ms, mom, lr, rho, momentum, epsilon, grad, options);
@@ -1017,7 +1018,7 @@ public <T extends TType> ResourceApplyRmsProp resourceApplyRmsProp(Operand<?> va
    * @param options carries optional attributes values
    * @return a new instance of ResourceSparseApplyAdadelta
    */
-  public <T extends TType, U extends TNumber> ResourceSparseApplyAdadelta resourceSparseApplyAdadelta(
+  public <T extends Tensor, U extends Tensor & TNumber> ResourceSparseApplyAdadelta resourceSparseApplyAdadelta(
       Operand<?> var, Operand<?> accum, Operand<?> accumUpdate, Operand<T> lr, Operand<T> rho,
       Operand<T> epsilon, Operand<T> grad, Operand<U> indices,
       ResourceSparseApplyAdadelta.Options... options) {
@@ -1039,7 +1040,7 @@ public <T extends TType, U extends TNumber> ResourceSparseApplyAdadelta resource
    * @param options carries optional attributes values
    * @return a new instance of ResourceSparseApplyAdagrad
    */
-  public <T extends TType, U extends TNumber> ResourceSparseApplyAdagrad resourceSparseApplyAdagrad(
+  public <T extends Tensor, U extends Tensor & TNumber> ResourceSparseApplyAdagrad resourceSparseApplyAdagrad(
       Operand<?> var, Operand<?> accum, Operand<T> lr, Operand<T> grad, Operand<U> indices,
       ResourceSparseApplyAdagrad.Options... options) {
     return ResourceSparseApplyAdagrad.create(scope, var, accum, lr, grad, indices, options);
@@ -1060,7 +1061,7 @@ public <T extends TType, U extends TNumber> ResourceSparseApplyAdagrad resourceS
    * @param options carries optional attributes values
    * @return a new instance of ResourceSparseApplyAdagradDa
    */
-  public <T extends TType, U extends TNumber> ResourceSparseApplyAdagradDa resourceSparseApplyAdagradDa(
+  public <T extends Tensor, U extends Tensor & TNumber> ResourceSparseApplyAdagradDa resourceSparseApplyAdagradDa(
       Operand<?> var, Operand<?> gradientAccumulator, Operand<?> gradientSquaredAccumulator,
       Operand<T> grad, Operand<U> indices, Operand<T> lr, Operand<T> l1, Operand<T> l2,
       Operand<TInt64> globalStep, ResourceSparseApplyAdagradDa.Options... options) {
@@ -1100,7 +1101,7 @@ public <T extends TType, U extends TNumber> ResourceSparseApplyAdagradDa resourc
    * @param options carries optional attributes values
    * @return a new instance of ResourceSparseApplyCenteredRmsProp
    */
-  public <T extends TType, U extends TNumber> ResourceSparseApplyCenteredRmsProp resourceSparseApplyCenteredRmsProp(
+  public <T extends Tensor, U extends Tensor & TNumber> ResourceSparseApplyCenteredRmsProp resourceSparseApplyCenteredRmsProp(
       Operand<?> var, Operand<?> mg, Operand<?> ms, Operand<?> mom, Operand<T> lr, Operand<T> rho,
       Operand<T> momentum, Operand<T> epsilon, Operand<T> grad, Operand<U> indices,
       ResourceSparseApplyCenteredRmsProp.Options... options) {
@@ -1132,7 +1133,7 @@ public <T extends TType, U extends TNumber> ResourceSparseApplyCenteredRmsProp r
    * @param options carries optional attributes values
    * @return a new instance of ResourceSparseApplyFtrl
    */
-  public <T extends TType, U extends TNumber> ResourceSparseApplyFtrl resourceSparseApplyFtrl(
+  public <T extends Tensor, U extends Tensor & TNumber> ResourceSparseApplyFtrl resourceSparseApplyFtrl(
       Operand<?> var, Operand<?> accum, Operand<?> linear, Operand<T> grad, Operand<U> indices,
       Operand<T> lr, Operand<T> l1, Operand<T> l2, Operand<T> l2Shrinkage, Operand<T> lrPower,
       ResourceSparseApplyFtrl.Options... options) {
@@ -1158,7 +1159,7 @@ public <T extends TType, U extends TNumber> ResourceSparseApplyFtrl resourceSpar
    * @param options carries optional attributes values
    * @return a new instance of ResourceSparseApplyKerasMomentum
    */
-  public <T extends TType, U extends TNumber> ResourceSparseApplyKerasMomentum resourceSparseApplyKerasMomentum(
+  public <T extends Tensor, U extends Tensor & TNumber> ResourceSparseApplyKerasMomentum resourceSparseApplyKerasMomentum(
       Operand<?> var, Operand<?> accum, Operand<T> lr, Operand<T> grad, Operand<U> indices,
       Operand<T> momentum, ResourceSparseApplyKerasMomentum.Options... options) {
     return ResourceSparseApplyKerasMomentum.create(scope, var, accum, lr, grad, indices, momentum, options);
@@ -1183,7 +1184,7 @@ public <T extends TType, U extends TNumber> ResourceSparseApplyKerasMomentum res
    * @param options carries optional attributes values
    * @return a new instance of ResourceSparseApplyMomentum
    */
-  public <T extends TType, U extends TNumber> ResourceSparseApplyMomentum resourceSparseApplyMomentum(
+  public <T extends Tensor, U extends Tensor & TNumber> ResourceSparseApplyMomentum resourceSparseApplyMomentum(
       Operand<?> var, Operand<?> accum, Operand<T> lr, Operand<T> grad, Operand<U> indices,
       Operand<T> momentum, ResourceSparseApplyMomentum.Options... options) {
     return ResourceSparseApplyMomentum.create(scope, var, accum, lr, grad, indices, momentum, options);
@@ -1208,7 +1209,7 @@ public <T extends TType, U extends TNumber> ResourceSparseApplyMomentum resource
    * @param options carries optional attributes values
    * @return a new instance of ResourceSparseApplyProximalAdagrad
    */
-  public <T extends TType, U extends TNumber> ResourceSparseApplyProximalAdagrad resourceSparseApplyProximalAdagrad(
+  public <T extends Tensor, U extends Tensor & TNumber> ResourceSparseApplyProximalAdagrad resourceSparseApplyProximalAdagrad(
       Operand<?> var, Operand<?> accum, Operand<T> lr, Operand<T> l1, Operand<T> l2,
       Operand<T> grad, Operand<U> indices, ResourceSparseApplyProximalAdagrad.Options... options) {
     return ResourceSparseApplyProximalAdagrad.create(scope, var, accum, lr, l1, l2, grad, indices, options);
@@ -1230,7 +1231,7 @@ public <T extends TType, U extends TNumber> ResourceSparseApplyProximalAdagrad r
    * @param options carries optional attributes values
    * @return a new instance of ResourceSparseApplyProximalGradientDescent
    */
-  public <T extends TType, U extends TNumber> ResourceSparseApplyProximalGradientDescent resourceSparseApplyProximalGradientDescent(
+  public <T extends Tensor, U extends Tensor & TNumber> ResourceSparseApplyProximalGradientDescent resourceSparseApplyProximalGradientDescent(
       Operand<?> var, Operand<T> alpha, Operand<T> l1, Operand<T> l2, Operand<T> grad,
       Operand<U> indices, ResourceSparseApplyProximalGradientDescent.Options... options) {
     return ResourceSparseApplyProximalGradientDescent.create(scope, var, alpha, l1, l2, grad, indices, options);
@@ -1262,7 +1263,7 @@ public <T extends TType, U extends TNumber> ResourceSparseApplyProximalGradientD
    * @param options carries optional attributes values
    * @return a new instance of ResourceSparseApplyRmsProp
    */
-  public <T extends TType, U extends TNumber> ResourceSparseApplyRmsProp resourceSparseApplyRmsProp(
+  public <T extends Tensor, U extends Tensor & TNumber> ResourceSparseApplyRmsProp resourceSparseApplyRmsProp(
       Operand<?> var, Operand<?> ms, Operand<?> mom, Operand<T> lr, Operand<T> rho,
       Operand<T> momentum, Operand<T> epsilon, Operand<T> grad, Operand<U> indices,
       ResourceSparseApplyRmsProp.Options... options) {
@@ -1320,7 +1321,7 @@ public Restore restore(Operand<TString> prefix, Operand<TString> tensorNames,
    * @param options carries optional attributes values
    * @return a new instance of RestoreSlice
    */
-  public <T extends TType> RestoreSlice<T> restoreSlice(Operand<TString> filePattern,
+  public <T extends Tensor> RestoreSlice<T> restoreSlice(Operand<TString> filePattern,
       Operand<TString> tensorName, Operand<TString> shapeAndSlice, DataType<T> dt,
       RestoreSlice.Options... options) {
     return RestoreSlice.create(scope, filePattern, tensorName, shapeAndSlice, dt, options);
@@ -1430,7 +1431,7 @@ public SdcaShrinkL1 sdcaShrinkL1(Iterable<Operand<TFloat32>> weights, Float l1,
    * @param options carries optional attributes values
    * @return a new instance of SparseApplyAdadelta
    */
-  public <T extends TType, U extends TNumber> SparseApplyAdadelta<T> sparseApplyAdadelta(
+  public <T extends Tensor, U extends Tensor & TNumber> SparseApplyAdadelta<T> sparseApplyAdadelta(
       Operand<T> var, Operand<T> accum, Operand<T> accumUpdate, Operand<T> lr, Operand<T> rho,
       Operand<T> epsilon, Operand<T> grad, Operand<U> indices,
       SparseApplyAdadelta.Options... options) {
@@ -1453,7 +1454,7 @@ public <T extends TType, U extends TNumber> SparseApplyAdadelta<T> sparseApplyAd
    * @param options carries optional attributes values
    * @return a new instance of SparseApplyAdagradDa
    */
-  public <T extends TType, U extends TNumber> SparseApplyAdagradDa<T> sparseApplyAdagradDa(
+  public <T extends Tensor, U extends Tensor & TNumber> SparseApplyAdagradDa<T> sparseApplyAdagradDa(
       Operand<T> var, Operand<T> gradientAccumulator, Operand<T> gradientSquaredAccumulator,
       Operand<T> grad, Operand<U> indices, Operand<T> lr, Operand<T> l1, Operand<T> l2,
       Operand<TInt64> globalStep, SparseApplyAdagradDa.Options... options) {
@@ -1494,7 +1495,7 @@ public <T extends TType, U extends TNumber> SparseApplyAdagradDa<T> sparseApplyA
    * @param options carries optional attributes values
    * @return a new instance of SparseApplyCenteredRmsProp
    */
-  public <T extends TType, U extends TNumber> SparseApplyCenteredRmsProp<T> sparseApplyCenteredRmsProp(
+  public <T extends Tensor, U extends Tensor & TNumber> SparseApplyCenteredRmsProp<T> sparseApplyCenteredRmsProp(
       Operand<T> var, Operand<T> mg, Operand<T> ms, Operand<T> mom, Operand<T> lr, Operand<T> rho,
       Operand<T> momentum, Operand<T> epsilon, Operand<T> grad, Operand<U> indices,
       SparseApplyCenteredRmsProp.Options... options) {
@@ -1527,9 +1528,9 @@ public <T extends TType, U extends TNumber> SparseApplyCenteredRmsProp<T> sparse
    * @param options carries optional attributes values
    * @return a new instance of SparseApplyFtrl
    */
-  public <T extends TType, U extends TNumber> SparseApplyFtrl<T> sparseApplyFtrl(Operand<T> var,
-      Operand<T> accum, Operand<T> linear, Operand<T> grad, Operand<U> indices, Operand<T> lr,
-      Operand<T> l1, Operand<T> l2, Operand<T> l2Shrinkage, Operand<T> lrPower,
+  public <T extends Tensor, U extends Tensor & TNumber> SparseApplyFtrl<T> sparseApplyFtrl(
+      Operand<T> var, Operand<T> accum, Operand<T> linear, Operand<T> grad, Operand<U> indices,
+      Operand<T> lr, Operand<T> l1, Operand<T> l2, Operand<T> l2Shrinkage, Operand<T> lrPower,
       SparseApplyFtrl.Options... options) {
     return SparseApplyFtrl.create(scope, var, accum, linear, grad, indices, lr, l1, l2, l2Shrinkage, lrPower, options);
   }
@@ -1554,7 +1555,7 @@ public <T extends TType, U extends TNumber> SparseApplyFtrl<T> sparseApplyFtrl(O
    * @param options carries optional attributes values
    * @return a new instance of SparseApplyMomentum
    */
-  public <T extends TType, U extends TNumber> SparseApplyMomentum<T> sparseApplyMomentum(
+  public <T extends Tensor, U extends Tensor & TNumber> SparseApplyMomentum<T> sparseApplyMomentum(
       Operand<T> var, Operand<T> accum, Operand<T> lr, Operand<T> grad, Operand<U> indices,
       Operand<T> momentum, SparseApplyMomentum.Options... options) {
     return SparseApplyMomentum.create(scope, var, accum, lr, grad, indices, momentum, options);
@@ -1580,7 +1581,7 @@ public <T extends TType, U extends TNumber> SparseApplyMomentum<T> sparseApplyMo
    * @param options carries optional attributes values
    * @return a new instance of SparseApplyProximalAdagrad
    */
-  public <T extends TType, U extends TNumber> SparseApplyProximalAdagrad<T> sparseApplyProximalAdagrad(
+  public <T extends Tensor, U extends Tensor & TNumber> SparseApplyProximalAdagrad<T> sparseApplyProximalAdagrad(
       Operand<T> var, Operand<T> accum, Operand<T> lr, Operand<T> l1, Operand<T> l2,
       Operand<T> grad, Operand<U> indices, SparseApplyProximalAdagrad.Options... options) {
     return SparseApplyProximalAdagrad.create(scope, var, accum, lr, l1, l2, grad, indices, options);
@@ -1603,7 +1604,7 @@ public <T extends TType, U extends TNumber> SparseApplyProximalAdagrad<T> sparse
    * @param options carries optional attributes values
    * @return a new instance of SparseApplyProximalGradientDescent
    */
-  public <T extends TType, U extends TNumber> SparseApplyProximalGradientDescent<T> sparseApplyProximalGradientDescent(
+  public <T extends Tensor, U extends Tensor & TNumber> SparseApplyProximalGradientDescent<T> sparseApplyProximalGradientDescent(
       Operand<T> var, Operand<T> alpha, Operand<T> l1, Operand<T> l2, Operand<T> grad,
       Operand<U> indices, SparseApplyProximalGradientDescent.Options... options) {
     return SparseApplyProximalGradientDescent.create(scope, var, alpha, l1, l2, grad, indices, options);
@@ -1636,7 +1637,7 @@ public <T extends TType, U extends TNumber> SparseApplyProximalGradientDescent<T
    * @param options carries optional attributes values
    * @return a new instance of SparseApplyRmsProp
    */
-  public <T extends TType, U extends TNumber> SparseApplyRmsProp<T> sparseApplyRmsProp(
+  public <T extends Tensor, U extends Tensor & TNumber> SparseApplyRmsProp<T> sparseApplyRmsProp(
       Operand<T> var, Operand<T> ms, Operand<T> mom, Operand<T> lr, Operand<T> rho,
       Operand<T> momentum, Operand<T> epsilon, Operand<T> grad, Operand<U> indices,
       SparseApplyRmsProp.Options... options) {
@@ -1655,7 +1656,7 @@ public <T extends TType, U extends TNumber> SparseApplyRmsProp<T> sparseApplyRms
    * @param multiples
    * @return a new instance of TileGrad
    */
-  public <T extends TType> TileGrad<T> tileGrad(Operand<T> input, Operand<TInt32> multiples) {
+  public <T extends Tensor> TileGrad<T> tileGrad(Operand<T> input, Operand<TInt32> multiples) {
     return TileGrad.create(scope, input, multiples);
   }
 }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/XlaOps.java b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/XlaOps.java
index 535972d4883..01d4ffa1ba7 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/XlaOps.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/XlaOps.java
@@ -19,6 +19,7 @@
 
 import org.tensorflow.DataType;
 import org.tensorflow.Operand;
+import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.xla.BroadcastHelper;
 import org.tensorflow.op.xla.ClusterOutput;
@@ -39,7 +40,6 @@
 import org.tensorflow.op.xla.Sort;
 import org.tensorflow.op.xla.Svd;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * An API for building {@code xla} operations as {@link Op Op}s
@@ -66,8 +66,8 @@ public final class XlaOps {
    * @param broadcastDims an XLA-style broadcast dimension specification
    * @return a new instance of BroadcastHelper
    */
-  public <T extends TType, U extends TNumber> BroadcastHelper<T> broadcastHelper(Operand<T> lhs,
-      Operand<T> rhs, Operand<U> broadcastDims) {
+  public <T extends Tensor, U extends Tensor & TNumber> BroadcastHelper<T> broadcastHelper(
+      Operand<T> lhs, Operand<T> rhs, Operand<U> broadcastDims) {
     return BroadcastHelper.create(scope, lhs, rhs, broadcastDims);
   }
 
@@ -78,7 +78,7 @@ public <T extends TType, U extends TNumber> BroadcastHelper<T> broadcastHelper(O
    * @param input
    * @return a new instance of ClusterOutput
    */
-  public <T extends TType> ClusterOutput<T> clusterOutput(Operand<T> input) {
+  public <T extends Tensor> ClusterOutput<T> clusterOutput(Operand<T> input) {
     return ClusterOutput.create(scope, input);
   }
 
@@ -100,7 +100,7 @@ public <T extends TType> ClusterOutput<T> clusterOutput(Operand<T> input) {
    * @param precisionConfig a serialized xla::PrecisionConfig proto.
    * @return a new instance of Conv
    */
-  public <T extends TType, U extends TNumber> Conv<T> conv(Operand<T> lhs, Operand<T> rhs,
+  public <T extends Tensor, U extends Tensor & TNumber> Conv<T> conv(Operand<T> lhs, Operand<T> rhs,
       Operand<U> windowStrides, Operand<U> padding, Operand<U> lhsDilation, Operand<U> rhsDilation,
       Operand<U> featureGroupCount, String dimensionNumbers, String precisionConfig) {
     return Conv.create(scope, lhs, rhs, windowStrides, padding, lhsDilation, rhsDilation, featureGroupCount, dimensionNumbers, precisionConfig);
@@ -137,7 +137,7 @@ public Dequantize dequantize(Operand<?> input, Float minRange, Float maxRange, S
    * @param precisionConfig a serialized xla::PrecisionConfig proto.
    * @return a new instance of Dot
    */
-  public <T extends TType> Dot<T> dot(Operand<T> lhs, Operand<T> rhs, String dimensionNumbers,
+  public <T extends Tensor> Dot<T> dot(Operand<T> lhs, Operand<T> rhs, String dimensionNumbers,
       String precisionConfig) {
     return Dot.create(scope, lhs, rhs, dimensionNumbers, precisionConfig);
   }
@@ -163,8 +163,8 @@ public <T extends TType> Dot<T> dot(Operand<T> lhs, Operand<T> rhs, String dimen
    * @param sizeIndices
    * @return a new instance of DynamicSlice
    */
-  public <T extends TType, U extends TNumber> DynamicSlice<T> dynamicSlice(Operand<T> input,
-      Operand<U> startIndices, Operand<U> sizeIndices) {
+  public <T extends Tensor, U extends Tensor & TNumber> DynamicSlice<T> dynamicSlice(
+      Operand<T> input, Operand<U> startIndices, Operand<U> sizeIndices) {
     return DynamicSlice.create(scope, input, startIndices, sizeIndices);
   }
 
@@ -188,7 +188,7 @@ public <T extends TType, U extends TNumber> DynamicSlice<T> dynamicSlice(Operand
    *  `input`.
    * @return a new instance of DynamicUpdateSlice
    */
-  public <T extends TType, U extends TNumber> DynamicUpdateSlice<T> dynamicUpdateSlice(
+  public <T extends Tensor, U extends Tensor & TNumber> DynamicUpdateSlice<T> dynamicUpdateSlice(
       Operand<T> input, Operand<T> update, Operand<U> indices) {
     return DynamicUpdateSlice.create(scope, input, update, indices);
   }
@@ -205,7 +205,7 @@ public <T extends TType, U extends TNumber> DynamicUpdateSlice<T> dynamicUpdateS
    * @param equation
    * @return a new instance of Einsum
    */
-  public <T extends TType> Einsum<T> einsum(Operand<T> a, Operand<T> b, String equation) {
+  public <T extends Tensor> Einsum<T> einsum(Operand<T> a, Operand<T> b, String equation) {
     return Einsum.create(scope, a, b, equation);
   }
 
@@ -222,7 +222,7 @@ public <T extends TType> Einsum<T> einsum(Operand<T> a, Operand<T> b, String equ
    * @param indicesAreSorted Boolean indicating if the indices are sorted.
    * @return a new instance of Gather
    */
-  public <T extends TType, U extends TNumber> Gather<T> gather(Operand<T> operand,
+  public <T extends Tensor, U extends Tensor & TNumber> Gather<T> gather(Operand<T> operand,
       Operand<U> startIndices, Operand<U> sliceSizes, String dimensionNumbers,
       Boolean indicesAreSorted) {
     return Gather.create(scope, operand, startIndices, sliceSizes, dimensionNumbers, indicesAreSorted);
@@ -242,8 +242,8 @@ public <T extends TType, U extends TNumber> Gather<T> gather(Operand<T> operand,
    * @param values A `Tensor` of type V.
    * @return a new instance of KeyValueSort
    */
-  public <T extends TNumber, U extends TType> KeyValueSort<T, U> keyValueSort(Operand<T> keys,
-      Operand<U> values) {
+  public <T extends Tensor & TNumber, U extends Tensor> KeyValueSort<T, U> keyValueSort(
+      Operand<T> keys, Operand<U> values) {
     return KeyValueSort.create(scope, keys, values);
   }
 
@@ -261,8 +261,9 @@ public <T extends TNumber, U extends TType> KeyValueSort<T, U> keyValueSort(Oper
    * @param paddingInterior the padding to apply between each input element.
    * @return a new instance of Pad
    */
-  public <T extends TType, U extends TNumber> Pad<T> pad(Operand<T> input, Operand<T> paddingValue,
-      Operand<U> paddingLow, Operand<U> paddingHigh, Operand<U> paddingInterior) {
+  public <T extends Tensor, U extends Tensor & TNumber> Pad<T> pad(Operand<T> input,
+      Operand<T> paddingValue, Operand<U> paddingLow, Operand<U> paddingHigh,
+      Operand<U> paddingInterior) {
     return Pad.create(scope, input, paddingValue, paddingLow, paddingHigh, paddingInterior);
   }
 
@@ -278,7 +279,7 @@ public <T extends TType, U extends TNumber> Pad<T> pad(Operand<T> input, Operand
    * @param shape The shape of the tensor.
    * @return a new instance of Recv
    */
-  public <T extends TType> Recv<T> recv(DataType<T> dtype, String tensorName, Shape shape) {
+  public <T extends Tensor> Recv<T> recv(DataType<T> dtype, String tensorName, Shape shape) {
     return Recv.create(scope, dtype, tensorName, shape);
   }
 
@@ -311,7 +312,7 @@ public ReplicaId replicaId() {
    * @param epsilon the tolerance ratio.
    * @return a new instance of SelfAdjointEig
    */
-  public <T extends TType> SelfAdjointEig<T> selfAdjointEig(Operand<T> a, Boolean lower,
+  public <T extends Tensor> SelfAdjointEig<T> selfAdjointEig(Operand<T> a, Boolean lower,
       Long maxIter, Float epsilon) {
     return SelfAdjointEig.create(scope, a, lower, maxIter, epsilon);
   }
@@ -326,7 +327,7 @@ public <T extends TType> SelfAdjointEig<T> selfAdjointEig(Operand<T> a, Boolean
    * @param tensorName A string key that identifies the channel.
    * @return a new instance of Send
    */
-  public <T extends TType> Send send(Operand<T> tensor, String tensorName) {
+  public <T extends Tensor> Send send(Operand<T> tensor, String tensorName) {
     return Send.create(scope, tensor, tensorName);
   }
 
@@ -337,7 +338,7 @@ public <T extends TType> Send send(Operand<T> tensor, String tensorName) {
    * @param input
    * @return a new instance of Sharding
    */
-  public <T extends TType> Sharding<T> sharding(Operand<T> input) {
+  public <T extends Tensor> Sharding<T> sharding(Operand<T> input) {
     return Sharding.create(scope, input);
   }
 
@@ -353,7 +354,7 @@ public <T extends TType> Sharding<T> sharding(Operand<T> input) {
    * @param input A `Tensor` of type T.
    * @return a new instance of Sort
    */
-  public <T extends TType> Sort<T> sort(Operand<T> input) {
+  public <T extends Tensor> Sort<T> sort(Operand<T> input) {
     return Sort.create(scope, input);
   }
 
@@ -375,7 +376,7 @@ public <T extends TType> Sort<T> sort(Operand<T> input) {
    * @param precisionConfig a serialized xla::PrecisionConfig proto.
    * @return a new instance of Svd
    */
-  public <T extends TType> Svd<T> svd(Operand<T> a, Long maxIter, Float epsilon,
+  public <T extends Tensor> Svd<T> svd(Operand<T> a, Long maxIter, Float epsilon,
       String precisionConfig) {
     return Svd.create(scope, a, maxIter, epsilon, precisionConfig);
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/bitwise/BitwiseAnd.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/bitwise/BitwiseAnd.java
index e199ff2201f..b4e37b6015e 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/bitwise/BitwiseAnd.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/bitwise/BitwiseAnd.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Elementwise computes the bitwise AND of `x` and `y`.
@@ -54,7 +54,7 @@
  * @param <T> data type for {@code z()} output
  */
 @Operator(group = "bitwise")
-public final class BitwiseAnd<T extends TNumber> extends RawOp implements Operand<T> {
+public final class BitwiseAnd<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new BitwiseAnd operation.
@@ -65,7 +65,7 @@ public final class BitwiseAnd<T extends TNumber> extends RawOp implements Operan
    * @return a new instance of BitwiseAnd
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> BitwiseAnd<T> create(Scope scope, Operand<T> x, Operand<T> y) {
+  public static <T extends Tensor & TNumber> BitwiseAnd<T> create(Scope scope, Operand<T> x, Operand<T> y) {
     OperationBuilder opBuilder = scope.env().opBuilder("BitwiseAnd", scope.makeOpName("BitwiseAnd"));
     opBuilder.addInput(x.asOutput());
     opBuilder.addInput(y.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/bitwise/BitwiseOr.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/bitwise/BitwiseOr.java
index 264c2bc340b..8440f0a46b1 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/bitwise/BitwiseOr.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/bitwise/BitwiseOr.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Elementwise computes the bitwise OR of `x` and `y`.
@@ -54,7 +54,7 @@
  * @param <T> data type for {@code z()} output
  */
 @Operator(group = "bitwise")
-public final class BitwiseOr<T extends TNumber> extends RawOp implements Operand<T> {
+public final class BitwiseOr<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new BitwiseOr operation.
@@ -65,7 +65,7 @@ public final class BitwiseOr<T extends TNumber> extends RawOp implements Operand
    * @return a new instance of BitwiseOr
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> BitwiseOr<T> create(Scope scope, Operand<T> x, Operand<T> y) {
+  public static <T extends Tensor & TNumber> BitwiseOr<T> create(Scope scope, Operand<T> x, Operand<T> y) {
     OperationBuilder opBuilder = scope.env().opBuilder("BitwiseOr", scope.makeOpName("BitwiseOr"));
     opBuilder.addInput(x.asOutput());
     opBuilder.addInput(y.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/bitwise/BitwiseXor.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/bitwise/BitwiseXor.java
index 1d8f668c175..c852846bc82 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/bitwise/BitwiseXor.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/bitwise/BitwiseXor.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Elementwise computes the bitwise XOR of `x` and `y`.
@@ -54,7 +54,7 @@
  * @param <T> data type for {@code z()} output
  */
 @Operator(group = "bitwise")
-public final class BitwiseXor<T extends TNumber> extends RawOp implements Operand<T> {
+public final class BitwiseXor<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new BitwiseXor operation.
@@ -65,7 +65,7 @@ public final class BitwiseXor<T extends TNumber> extends RawOp implements Operan
    * @return a new instance of BitwiseXor
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> BitwiseXor<T> create(Scope scope, Operand<T> x, Operand<T> y) {
+  public static <T extends Tensor & TNumber> BitwiseXor<T> create(Scope scope, Operand<T> x, Operand<T> y) {
     OperationBuilder opBuilder = scope.env().opBuilder("BitwiseXor", scope.makeOpName("BitwiseXor"));
     opBuilder.addInput(x.asOutput());
     opBuilder.addInput(y.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/bitwise/Invert.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/bitwise/Invert.java
index 9f8bdfd56d8..8cf61d86110 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/bitwise/Invert.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/bitwise/Invert.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Invert (flip) each bit of supported types; for example, type `uint8` value 01010101 becomes 10101010.
@@ -75,7 +75,7 @@
  * @param <T> data type for {@code y()} output
  */
 @Operator(group = "bitwise")
-public final class Invert<T extends TNumber> extends RawOp implements Operand<T> {
+public final class Invert<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Invert operation.
@@ -85,7 +85,7 @@ public final class Invert<T extends TNumber> extends RawOp implements Operand<T>
    * @return a new instance of Invert
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> Invert<T> create(Scope scope, Operand<T> x) {
+  public static <T extends Tensor & TNumber> Invert<T> create(Scope scope, Operand<T> x) {
     OperationBuilder opBuilder = scope.env().opBuilder("Invert", scope.makeOpName("Invert"));
     opBuilder.addInput(x.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/bitwise/LeftShift.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/bitwise/LeftShift.java
index f7a47534d81..1fa02ba6a51 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/bitwise/LeftShift.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/bitwise/LeftShift.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Elementwise computes the bitwise left-shift of `x` and `y`.
@@ -65,7 +65,7 @@
  * @param <T> data type for {@code z()} output
  */
 @Operator(group = "bitwise")
-public final class LeftShift<T extends TNumber> extends RawOp implements Operand<T> {
+public final class LeftShift<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new LeftShift operation.
@@ -76,7 +76,7 @@ public final class LeftShift<T extends TNumber> extends RawOp implements Operand
    * @return a new instance of LeftShift
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> LeftShift<T> create(Scope scope, Operand<T> x, Operand<T> y) {
+  public static <T extends Tensor & TNumber> LeftShift<T> create(Scope scope, Operand<T> x, Operand<T> y) {
     OperationBuilder opBuilder = scope.env().opBuilder("LeftShift", scope.makeOpName("LeftShift"));
     opBuilder.addInput(x.asOutput());
     opBuilder.addInput(y.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/bitwise/RightShift.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/bitwise/RightShift.java
index 99c5fe5766e..2bd25ff8ad8 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/bitwise/RightShift.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/bitwise/RightShift.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Elementwise computes the bitwise right-shift of `x` and `y`.
@@ -68,7 +68,7 @@
  * @param <T> data type for {@code z()} output
  */
 @Operator(group = "bitwise")
-public final class RightShift<T extends TNumber> extends RawOp implements Operand<T> {
+public final class RightShift<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new RightShift operation.
@@ -79,7 +79,7 @@ public final class RightShift<T extends TNumber> extends RawOp implements Operan
    * @return a new instance of RightShift
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> RightShift<T> create(Scope scope, Operand<T> x, Operand<T> y) {
+  public static <T extends Tensor & TNumber> RightShift<T> create(Scope scope, Operand<T> x, Operand<T> y) {
     OperationBuilder opBuilder = scope.env().opBuilder("RightShift", scope.makeOpName("RightShift"));
     opBuilder.addInput(x.asOutput());
     opBuilder.addInput(y.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/collective/AllReduce.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/collective/AllReduce.java
index d58bc1357f4..146533646cd 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/collective/AllReduce.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/collective/AllReduce.java
@@ -22,19 +22,19 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Mutually reduces multiple tensors of identical type and shape.
  * 
  * @param <T> data type for {@code output()} output
  */
-public final class AllReduce<T extends TNumber> extends RawOp implements Operand<T> {
+public final class AllReduce<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.collective.AllReduce}
@@ -88,7 +88,7 @@ private Options() {
    * @return a new instance of AllReduce
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> AllReduce<T> create(Scope scope, Operand<T> input, Long groupSize, Long groupKey, Long instanceKey, String mergeOp, String finalOp, List<Long> subdivOffsets, Options... options) {
+  public static <T extends Tensor & TNumber> AllReduce<T> create(Scope scope, Operand<T> input, Long groupSize, Long groupKey, Long instanceKey, String mergeOp, String finalOp, List<Long> subdivOffsets, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("CollectiveReduce", scope.makeOpName("AllReduce"));
     opBuilder.addInput(input.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/collective/BroadcastRecv.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/collective/BroadcastRecv.java
index ab938852275..cdd41fd3b3e 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/collective/BroadcastRecv.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/collective/BroadcastRecv.java
@@ -22,19 +22,19 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Receives a tensor value broadcast from another device.
  * 
  * @param <T> data type for {@code output()} output
  */
-public final class BroadcastRecv<T extends TType> extends RawOp implements Operand<T> {
+public final class BroadcastRecv<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.collective.BroadcastRecv}
@@ -77,7 +77,7 @@ private Options() {
    * @return a new instance of BroadcastRecv
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> BroadcastRecv<T> create(Scope scope, DataType<T> T, Long groupSize, Long groupKey, Long instanceKey, Shape shape, Options... options) {
+  public static <T extends Tensor> BroadcastRecv<T> create(Scope scope, DataType<T> T, Long groupSize, Long groupKey, Long instanceKey, Shape shape, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("CollectiveBcastRecv", scope.makeOpName("BroadcastRecv"));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("T", T);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/collective/BroadcastSend.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/collective/BroadcastSend.java
index f6b3d3ff307..5ebff67f14b 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/collective/BroadcastSend.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/collective/BroadcastSend.java
@@ -21,19 +21,19 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Broadcasts a tensor value to one or more other devices.
  * 
  * @param <T> data type for {@code output()} output
  */
-public final class BroadcastSend<T extends TType> extends RawOp implements Operand<T> {
+public final class BroadcastSend<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.collective.BroadcastSend}
@@ -76,7 +76,7 @@ private Options() {
    * @return a new instance of BroadcastSend
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> BroadcastSend<T> create(Scope scope, Operand<T> input, Long groupSize, Long groupKey, Long instanceKey, Shape shape, Options... options) {
+  public static <T extends Tensor> BroadcastSend<T> create(Scope scope, Operand<T> input, Long groupSize, Long groupKey, Long instanceKey, Shape shape, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("CollectiveBcastSend", scope.makeOpName("BroadcastSend"));
     opBuilder.addInput(input.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/All.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/All.java
index 909427d1a57..2e9bf958757 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/All.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/All.java
@@ -21,13 +21,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TBool;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes the "logical and" of elements across dimensions of a tensor.
@@ -70,7 +70,7 @@ private Options() {
    * @return a new instance of All
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> All create(Scope scope, Operand<TBool> input, Operand<T> axis, Options... options) {
+  public static <T extends Tensor & TNumber> All create(Scope scope, Operand<TBool> input, Operand<T> axis, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("All", scope.makeOpName("All"));
     opBuilder.addInput(input.asOutput());
     opBuilder.addInput(axis.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Any.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Any.java
index 0316e5e1a94..51cc483fb1d 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Any.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Any.java
@@ -21,13 +21,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TBool;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes the "logical or" of elements across dimensions of a tensor.
@@ -70,7 +70,7 @@ private Options() {
    * @return a new instance of Any
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> Any create(Scope scope, Operand<TBool> input, Operand<T> axis, Options... options) {
+  public static <T extends Tensor & TNumber> Any create(Scope scope, Operand<TBool> input, Operand<T> axis, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("Any", scope.makeOpName("Any"));
     opBuilder.addInput(input.asOutput());
     opBuilder.addInput(axis.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Assign.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Assign.java
index 27f66d3da23..021c6880596 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Assign.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Assign.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Update 'ref' by assigning 'value' to it.
@@ -36,7 +36,7 @@
  * @param <T> data type for {@code outputRef()} output
  */
 @Operator
-public final class Assign<T extends TType> extends RawOp implements Operand<T> {
+public final class Assign<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.core.Assign}
@@ -79,7 +79,7 @@ private Options() {
    * @return a new instance of Assign
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> Assign<T> create(Scope scope, Operand<T> ref, Operand<T> value, Options... options) {
+  public static <T extends Tensor> Assign<T> create(Scope scope, Operand<T> ref, Operand<T> value, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("Assign", scope.makeOpName("Assign"));
     opBuilder.addInput(ref.asOutput());
     opBuilder.addInput(value.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/AssignAdd.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/AssignAdd.java
index 770842d2764..c64a1924c42 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/AssignAdd.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/AssignAdd.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Update 'ref' by adding 'value' to it.
@@ -36,7 +36,7 @@
  * @param <T> data type for {@code outputRef()} output
  */
 @Operator
-public final class AssignAdd<T extends TType> extends RawOp implements Operand<T> {
+public final class AssignAdd<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.core.AssignAdd}
@@ -68,7 +68,7 @@ private Options() {
    * @return a new instance of AssignAdd
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> AssignAdd<T> create(Scope scope, Operand<T> ref, Operand<T> value, Options... options) {
+  public static <T extends Tensor> AssignAdd<T> create(Scope scope, Operand<T> ref, Operand<T> value, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("AssignAdd", scope.makeOpName("AssignAdd"));
     opBuilder.addInput(ref.asOutput());
     opBuilder.addInput(value.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/AssignAddVariableOp.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/AssignAddVariableOp.java
index 53edc808882..36c43524b12 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/AssignAddVariableOp.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/AssignAddVariableOp.java
@@ -20,11 +20,11 @@
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Adds a value to the current value of a variable.
@@ -44,7 +44,7 @@ public final class AssignAddVariableOp extends RawOp {
    * @return a new instance of AssignAddVariableOp
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> AssignAddVariableOp create(Scope scope, Operand<?> resource, Operand<T> value) {
+  public static <T extends Tensor> AssignAddVariableOp create(Scope scope, Operand<?> resource, Operand<T> value) {
     OperationBuilder opBuilder = scope.env().opBuilder("AssignAddVariableOp", scope.makeOpName("AssignAddVariableOp"));
     opBuilder.addInput(resource.asOutput());
     opBuilder.addInput(value.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/AssignSub.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/AssignSub.java
index 37841e0dadc..494f1dac6bb 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/AssignSub.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/AssignSub.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Update 'ref' by subtracting 'value' from it.
@@ -36,7 +36,7 @@
  * @param <T> data type for {@code outputRef()} output
  */
 @Operator
-public final class AssignSub<T extends TType> extends RawOp implements Operand<T> {
+public final class AssignSub<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.core.AssignSub}
@@ -68,7 +68,7 @@ private Options() {
    * @return a new instance of AssignSub
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> AssignSub<T> create(Scope scope, Operand<T> ref, Operand<T> value, Options... options) {
+  public static <T extends Tensor> AssignSub<T> create(Scope scope, Operand<T> ref, Operand<T> value, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("AssignSub", scope.makeOpName("AssignSub"));
     opBuilder.addInput(ref.asOutput());
     opBuilder.addInput(value.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/AssignSubVariableOp.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/AssignSubVariableOp.java
index 372a71b2168..c43611ebbba 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/AssignSubVariableOp.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/AssignSubVariableOp.java
@@ -20,11 +20,11 @@
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Subtracts a value from the current value of a variable.
@@ -44,7 +44,7 @@ public final class AssignSubVariableOp extends RawOp {
    * @return a new instance of AssignSubVariableOp
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> AssignSubVariableOp create(Scope scope, Operand<?> resource, Operand<T> value) {
+  public static <T extends Tensor> AssignSubVariableOp create(Scope scope, Operand<?> resource, Operand<T> value) {
     OperationBuilder opBuilder = scope.env().opBuilder("AssignSubVariableOp", scope.makeOpName("AssignSubVariableOp"));
     opBuilder.addInput(resource.asOutput());
     opBuilder.addInput(value.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/AssignVariableOp.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/AssignVariableOp.java
index ac08d62f9a8..3ffb2154021 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/AssignVariableOp.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/AssignVariableOp.java
@@ -20,11 +20,11 @@
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Assigns a new value to a variable.
@@ -44,7 +44,7 @@ public final class AssignVariableOp extends RawOp {
    * @return a new instance of AssignVariableOp
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> AssignVariableOp create(Scope scope, Operand<?> resource, Operand<T> value) {
+  public static <T extends Tensor> AssignVariableOp create(Scope scope, Operand<?> resource, Operand<T> value) {
     OperationBuilder opBuilder = scope.env().opBuilder("AssignVariableOp", scope.makeOpName("AssignVariableOp"));
     opBuilder.addInput(resource.asOutput());
     opBuilder.addInput(value.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BarrierInsertMany.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BarrierInsertMany.java
index b652c11a35c..272e6e1afe4 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BarrierInsertMany.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BarrierInsertMany.java
@@ -20,12 +20,12 @@
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TString;
-import org.tensorflow.types.family.TType;
 
 /**
  * For each key, assigns the respective value to the specified component.
@@ -50,7 +50,7 @@ public final class BarrierInsertMany extends RawOp {
    * @return a new instance of BarrierInsertMany
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> BarrierInsertMany create(Scope scope, Operand<TString> handle, Operand<TString> keys, Operand<T> values, Long componentIndex) {
+  public static <T extends Tensor> BarrierInsertMany create(Scope scope, Operand<TString> handle, Operand<TString> keys, Operand<T> values, Long componentIndex) {
     OperationBuilder opBuilder = scope.env().opBuilder("BarrierInsertMany", scope.makeOpName("BarrierInsertMany"));
     opBuilder.addInput(handle.asOutput());
     opBuilder.addInput(keys.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BatchToSpace.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BatchToSpace.java
index b20c60ee4ac..1d7356e3b95 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BatchToSpace.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BatchToSpace.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * BatchToSpace for 4-D tensors of type T.
@@ -42,7 +42,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator
-public final class BatchToSpace<T extends TType> extends RawOp implements Operand<T> {
+public final class BatchToSpace<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new BatchToSpace operation.
@@ -61,7 +61,7 @@ public final class BatchToSpace<T extends TType> extends RawOp implements Operan
    * @return a new instance of BatchToSpace
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> BatchToSpace<T> create(Scope scope, Operand<T> input, Operand<U> crops, Long blockSize) {
+  public static <T extends Tensor, U extends Tensor & TNumber> BatchToSpace<T> create(Scope scope, Operand<T> input, Operand<U> crops, Long blockSize) {
     OperationBuilder opBuilder = scope.env().opBuilder("BatchToSpace", scope.makeOpName("BatchToSpace"));
     opBuilder.addInput(input.asOutput());
     opBuilder.addInput(crops.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BatchToSpaceNd.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BatchToSpaceNd.java
index ef6f31d0f89..4396bd919c8 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BatchToSpaceNd.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BatchToSpaceNd.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * BatchToSpace for N-D tensors of type T.
@@ -41,7 +41,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator
-public final class BatchToSpaceNd<T extends TType> extends RawOp implements Operand<T> {
+public final class BatchToSpaceNd<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new BatchToSpaceNd operation.
@@ -148,7 +148,7 @@ public final class BatchToSpaceNd<T extends TType> extends RawOp implements Oper
    * @return a new instance of BatchToSpaceNd
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber, V extends TNumber> BatchToSpaceNd<T> create(Scope scope, Operand<T> input, Operand<U> blockShape, Operand<V> crops) {
+  public static <T extends Tensor, U extends Tensor & TNumber, V extends Tensor & TNumber> BatchToSpaceNd<T> create(Scope scope, Operand<T> input, Operand<U> blockShape, Operand<V> crops) {
     OperationBuilder opBuilder = scope.env().opBuilder("BatchToSpaceND", scope.makeOpName("BatchToSpaceNd"));
     opBuilder.addInput(input.asOutput());
     opBuilder.addInput(blockShape.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Bitcast.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Bitcast.java
index b01c8598ae6..da3b8569073 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Bitcast.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Bitcast.java
@@ -22,11 +22,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Bitcasts a tensor from one type to another without copying data.
@@ -85,7 +85,7 @@
  * @param <U> data type for {@code output()} output
  */
 @Operator
-public final class Bitcast<U extends TType> extends RawOp implements Operand<U> {
+public final class Bitcast<U extends Tensor> extends RawOp implements Operand<U> {
   
   /**
    * Factory method to create a class wrapping a new Bitcast operation.
@@ -96,7 +96,7 @@ public final class Bitcast<U extends TType> extends RawOp implements Operand<U>
    * @return a new instance of Bitcast
    */
   @Endpoint(describeByClass = true)
-  public static <U extends TType, T extends TType> Bitcast<U> create(Scope scope, Operand<T> input, DataType<U> type) {
+  public static <U extends Tensor, T extends Tensor> Bitcast<U> create(Scope scope, Operand<T> input, DataType<U> type) {
     OperationBuilder opBuilder = scope.env().opBuilder("Bitcast", scope.makeOpName("Bitcast"));
     opBuilder.addInput(input.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BroadcastDynamicShape.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BroadcastDynamicShape.java
index 3027e8234a4..d0e6dd8601e 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BroadcastDynamicShape.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BroadcastDynamicShape.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Return the shape of s0 op s1 with broadcast.
@@ -37,7 +37,7 @@
  * @param <T> data type for {@code r0()} output
  */
 @Operator
-public final class BroadcastDynamicShape<T extends TNumber> extends RawOp implements Operand<T> {
+public final class BroadcastDynamicShape<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new BroadcastDynamicShape operation.
@@ -48,7 +48,7 @@ public final class BroadcastDynamicShape<T extends TNumber> extends RawOp implem
    * @return a new instance of BroadcastDynamicShape
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> BroadcastDynamicShape<T> create(Scope scope, Operand<T> s0, Operand<T> s1) {
+  public static <T extends Tensor & TNumber> BroadcastDynamicShape<T> create(Scope scope, Operand<T> s0, Operand<T> s1) {
     OperationBuilder opBuilder = scope.env().opBuilder("BroadcastArgs", scope.makeOpName("BroadcastDynamicShape"));
     opBuilder.addInput(s0.asOutput());
     opBuilder.addInput(s1.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BroadcastGradientArgs.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BroadcastGradientArgs.java
index 2d95c71086e..e604ac4f87c 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BroadcastGradientArgs.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BroadcastGradientArgs.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Return the reduction indices for computing gradients of s0 op s1 with broadcast.
@@ -35,7 +35,7 @@
  * 
  * @param <T> data type for {@code r0()} output
  */
-public final class BroadcastGradientArgs<T extends TNumber> extends RawOp {
+public final class BroadcastGradientArgs<T extends Tensor & TNumber> extends RawOp {
   
   /**
    * Factory method to create a class wrapping a new BroadcastGradientArgs operation.
@@ -46,7 +46,7 @@ public final class BroadcastGradientArgs<T extends TNumber> extends RawOp {
    * @return a new instance of BroadcastGradientArgs
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> BroadcastGradientArgs<T> create(Scope scope, Operand<T> s0, Operand<T> s1) {
+  public static <T extends Tensor & TNumber> BroadcastGradientArgs<T> create(Scope scope, Operand<T> s0, Operand<T> s1) {
     OperationBuilder opBuilder = scope.env().opBuilder("BroadcastGradientArgs", scope.makeOpName("BroadcastGradientArgs"));
     opBuilder.addInput(s0.asOutput());
     opBuilder.addInput(s1.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BroadcastTo.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BroadcastTo.java
index e5ad5469501..31e6a0d4216 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BroadcastTo.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BroadcastTo.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Broadcast an array for a compatible shape.
@@ -62,7 +62,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator
-public final class BroadcastTo<T extends TType> extends RawOp implements Operand<T> {
+public final class BroadcastTo<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new BroadcastTo operation.
@@ -73,7 +73,7 @@ public final class BroadcastTo<T extends TType> extends RawOp implements Operand
    * @return a new instance of BroadcastTo
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> BroadcastTo<T> create(Scope scope, Operand<T> input, Operand<U> shape) {
+  public static <T extends Tensor, U extends Tensor & TNumber> BroadcastTo<T> create(Scope scope, Operand<T> input, Operand<U> shape) {
     OperationBuilder opBuilder = scope.env().opBuilder("BroadcastTo", scope.makeOpName("BroadcastTo"));
     opBuilder.addInput(input.asOutput());
     opBuilder.addInput(shape.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Bucketize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Bucketize.java
index 84e9b454d0d..8c69988030d 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Bucketize.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Bucketize.java
@@ -22,13 +22,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Bucketizes 'input' based on 'boundaries'.
@@ -56,7 +56,7 @@ public final class Bucketize extends RawOp implements Operand<TInt32> {
    * @return a new instance of Bucketize
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> Bucketize create(Scope scope, Operand<T> input, List<Float> boundaries) {
+  public static <T extends Tensor & TNumber> Bucketize create(Scope scope, Operand<T> input, List<Float> boundaries) {
     OperationBuilder opBuilder = scope.env().opBuilder("Bucketize", scope.makeOpName("Bucketize"));
     opBuilder.addInput(input.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ClipByValue.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ClipByValue.java
index 8a5b99d9389..43bff94ad09 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ClipByValue.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ClipByValue.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Clips tensor values to a specified min and max.
@@ -38,7 +38,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator
-public final class ClipByValue<T extends TType> extends RawOp implements Operand<T> {
+public final class ClipByValue<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new ClipByValue operation.
@@ -52,7 +52,7 @@ public final class ClipByValue<T extends TType> extends RawOp implements Operand
    * @return a new instance of ClipByValue
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> ClipByValue<T> create(Scope scope, Operand<T> t, Operand<T> clipValueMin, Operand<T> clipValueMax) {
+  public static <T extends Tensor> ClipByValue<T> create(Scope scope, Operand<T> t, Operand<T> clipValueMin, Operand<T> clipValueMax) {
     OperationBuilder opBuilder = scope.env().opBuilder("ClipByValue", scope.makeOpName("ClipByValue"));
     opBuilder.addInput(t.asOutput());
     opBuilder.addInput(clipValueMin.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/CollectiveGather.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/CollectiveGather.java
index bd9de3ebb33..6683a0822a7 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/CollectiveGather.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/CollectiveGather.java
@@ -21,20 +21,20 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Mutually accumulates multiple tensors of identical type and shape.
  * 
  * @param <T> data type for {@code output()} output
  */
-public final class CollectiveGather<T extends TNumber> extends RawOp implements Operand<T> {
+public final class CollectiveGather<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.core.CollectiveGather}
@@ -77,7 +77,7 @@ private Options() {
    * @return a new instance of CollectiveGather
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> CollectiveGather<T> create(Scope scope, Operand<T> input, Long groupSize, Long groupKey, Long instanceKey, Shape shape, Options... options) {
+  public static <T extends Tensor & TNumber> CollectiveGather<T> create(Scope scope, Operand<T> input, Long groupSize, Long groupKey, Long instanceKey, Shape shape, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("CollectiveGather", scope.makeOpName("CollectiveGather"));
     opBuilder.addInput(input.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Concat.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Concat.java
index 9782b3d6ec9..fa2898efb10 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Concat.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Concat.java
@@ -21,13 +21,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.Operands;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Concatenates tensors along one dimension.
@@ -35,7 +35,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator
-public final class Concat<T extends TType> extends RawOp implements Operand<T> {
+public final class Concat<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Concat operation.
@@ -48,7 +48,7 @@ public final class Concat<T extends TType> extends RawOp implements Operand<T> {
    * @return a new instance of Concat
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> Concat<T> create(Scope scope, Iterable<Operand<T>> values, Operand<U> axis) {
+  public static <T extends Tensor, U extends Tensor & TNumber> Concat<T> create(Scope scope, Iterable<Operand<T>> values, Operand<U> axis) {
     OperationBuilder opBuilder = scope.env().opBuilder("ConcatV2", scope.makeOpName("Concat"));
     opBuilder.addInputList(Operands.asOutputs(values));
     opBuilder.addInput(axis.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Copy.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Copy.java
index fb9cbbb122a..b114662e162 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Copy.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Copy.java
@@ -22,11 +22,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Copy a tensor from CPU-to-CPU or GPU-to-GPU.
@@ -42,7 +42,7 @@
  * 
  * @param <T> data type for {@code output()} output
  */
-public final class Copy<T extends TType> extends RawOp implements Operand<T> {
+public final class Copy<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.core.Copy}
@@ -85,7 +85,7 @@ private Options() {
    * @return a new instance of Copy
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> Copy<T> create(Scope scope, Operand<T> input, Options... options) {
+  public static <T extends Tensor> Copy<T> create(Scope scope, Operand<T> input, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("Copy", scope.makeOpName("Copy"));
     opBuilder.addInput(input.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/CopyHost.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/CopyHost.java
index 566258c0430..057b07cdb76 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/CopyHost.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/CopyHost.java
@@ -22,11 +22,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Copy a tensor to host.
@@ -40,7 +40,7 @@
  * 
  * @param <T> data type for {@code output()} output
  */
-public final class CopyHost<T extends TType> extends RawOp implements Operand<T> {
+public final class CopyHost<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.core.CopyHost}
@@ -83,7 +83,7 @@ private Options() {
    * @return a new instance of CopyHost
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> CopyHost<T> create(Scope scope, Operand<T> input, Options... options) {
+  public static <T extends Tensor> CopyHost<T> create(Scope scope, Operand<T> input, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("CopyHost", scope.makeOpName("CopyHost"));
     opBuilder.addInput(input.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/CountUpTo.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/CountUpTo.java
index 2884783695d..9a36603b148 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/CountUpTo.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/CountUpTo.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Increments 'ref' until it reaches 'limit'.
@@ -34,7 +34,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator
-public final class CountUpTo<T extends TNumber> extends RawOp implements Operand<T> {
+public final class CountUpTo<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new CountUpTo operation.
@@ -46,7 +46,7 @@ public final class CountUpTo<T extends TNumber> extends RawOp implements Operand
    * @return a new instance of CountUpTo
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> CountUpTo<T> create(Scope scope, Operand<T> ref, Long limit) {
+  public static <T extends Tensor & TNumber> CountUpTo<T> create(Scope scope, Operand<T> ref, Long limit) {
     OperationBuilder opBuilder = scope.env().opBuilder("CountUpTo", scope.makeOpName("CountUpTo"));
     opBuilder.addInput(ref.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/DeepCopy.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/DeepCopy.java
index da22489acc9..f7f8f85d83a 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/DeepCopy.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/DeepCopy.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Makes a copy of `x`.
@@ -33,7 +33,7 @@
  * @param <T> data type for {@code y()} output
  */
 @Operator
-public final class DeepCopy<T extends TType> extends RawOp implements Operand<T> {
+public final class DeepCopy<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new DeepCopy operation.
@@ -43,7 +43,7 @@ public final class DeepCopy<T extends TType> extends RawOp implements Operand<T>
    * @return a new instance of DeepCopy
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> DeepCopy<T> create(Scope scope, Operand<T> x) {
+  public static <T extends Tensor> DeepCopy<T> create(Scope scope, Operand<T> x) {
     OperationBuilder opBuilder = scope.env().opBuilder("DeepCopy", scope.makeOpName("DeepCopy"));
     opBuilder.addInput(x.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/DestroyTemporaryVariable.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/DestroyTemporaryVariable.java
index 7941a7a5870..161430a22e2 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/DestroyTemporaryVariable.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/DestroyTemporaryVariable.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Destroys the temporary variable and returns its final value.
@@ -41,7 +41,7 @@
  * @param <T> data type for {@code value()} output
  */
 @Operator
-public final class DestroyTemporaryVariable<T extends TType> extends RawOp implements Operand<T> {
+public final class DestroyTemporaryVariable<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new DestroyTemporaryVariable operation.
@@ -53,7 +53,7 @@ public final class DestroyTemporaryVariable<T extends TType> extends RawOp imple
    * @return a new instance of DestroyTemporaryVariable
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> DestroyTemporaryVariable<T> create(Scope scope, Operand<T> ref, String varName) {
+  public static <T extends Tensor> DestroyTemporaryVariable<T> create(Scope scope, Operand<T> ref, String varName) {
     OperationBuilder opBuilder = scope.env().opBuilder("DestroyTemporaryVariable", scope.makeOpName("DestroyTemporaryVariable"));
     opBuilder.addInput(ref.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/DummyMemoryCache.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/DummyMemoryCache.java
index 56f16d48a9b..e37c2c7f3ae 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/DummyMemoryCache.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/DummyMemoryCache.java
@@ -21,15 +21,15 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  */
-public final class DummyMemoryCache extends RawOp implements Operand<TType> {
+public final class DummyMemoryCache extends RawOp implements Operand<Tensor> {
   
   /**
    * Factory method to create a class wrapping a new DummyMemoryCache operation.
@@ -52,8 +52,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) handle;
+  public Output<Tensor> asOutput() {
+    return (Output<Tensor>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/DynamicPartition.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/DynamicPartition.java
index d9da09b5d34..73daaf7646e 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/DynamicPartition.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/DynamicPartition.java
@@ -24,12 +24,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
-import org.tensorflow.types.family.TType;
 
 /**
  * Partitions `data` into `num_partitions` tensors using indices from `partitions`.
@@ -71,7 +71,7 @@
  * @param <T> data type for {@code outputs()} output
  */
 @Operator
-public final class DynamicPartition<T extends TType> extends RawOp implements Iterable<Operand<T>> {
+public final class DynamicPartition<T extends Tensor> extends RawOp implements Iterable<Operand<T>> {
   
   /**
    * Factory method to create a class wrapping a new DynamicPartition operation.
@@ -83,7 +83,7 @@ public final class DynamicPartition<T extends TType> extends RawOp implements It
    * @return a new instance of DynamicPartition
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> DynamicPartition<T> create(Scope scope, Operand<T> data, Operand<TInt32> partitions, Long numPartitions) {
+  public static <T extends Tensor> DynamicPartition<T> create(Scope scope, Operand<T> data, Operand<TInt32> partitions, Long numPartitions) {
     OperationBuilder opBuilder = scope.env().opBuilder("DynamicPartition", scope.makeOpName("DynamicPartition"));
     opBuilder.addInput(data.asOutput());
     opBuilder.addInput(partitions.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/DynamicStitch.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/DynamicStitch.java
index b7dd72f8a57..590d9a2e3e1 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/DynamicStitch.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/DynamicStitch.java
@@ -21,13 +21,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.Operands;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
-import org.tensorflow.types.family.TType;
 
 /**
  * Interleave the values from the `data` tensors into a single tensor.
@@ -90,7 +90,7 @@
  * @param <T> data type for {@code merged()} output
  */
 @Operator
-public final class DynamicStitch<T extends TType> extends RawOp implements Operand<T> {
+public final class DynamicStitch<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new DynamicStitch operation.
@@ -101,7 +101,7 @@ public final class DynamicStitch<T extends TType> extends RawOp implements Opera
    * @return a new instance of DynamicStitch
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> DynamicStitch<T> create(Scope scope, Iterable<Operand<TInt32>> indices, Iterable<Operand<T>> data) {
+  public static <T extends Tensor> DynamicStitch<T> create(Scope scope, Iterable<Operand<TInt32>> indices, Iterable<Operand<T>> data) {
     OperationBuilder opBuilder = scope.env().opBuilder("DynamicStitch", scope.makeOpName("DynamicStitch"));
     opBuilder.addInputList(Operands.asOutputs(indices));
     opBuilder.addInputList(Operands.asOutputs(data));
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/EditDistance.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/EditDistance.java
index 29423a3c548..4220271bbf7 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/EditDistance.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/EditDistance.java
@@ -21,13 +21,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TFloat32;
 import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes the (possibly normalized) Levenshtein Edit Distance.
@@ -82,7 +82,7 @@ private Options() {
    * @return a new instance of EditDistance
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> EditDistance create(Scope scope, Operand<TInt64> hypothesisIndices, Operand<T> hypothesisValues, Operand<TInt64> hypothesisShape, Operand<TInt64> truthIndices, Operand<T> truthValues, Operand<TInt64> truthShape, Options... options) {
+  public static <T extends Tensor> EditDistance create(Scope scope, Operand<TInt64> hypothesisIndices, Operand<T> hypothesisValues, Operand<TInt64> hypothesisShape, Operand<TInt64> truthIndices, Operand<T> truthValues, Operand<TInt64> truthShape, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("EditDistance", scope.makeOpName("EditDistance"));
     opBuilder.addInput(hypothesisIndices.asOutput());
     opBuilder.addInput(hypothesisValues.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Empty.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Empty.java
index c8305349d37..6419d47510c 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Empty.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Empty.java
@@ -22,12 +22,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
-import org.tensorflow.types.family.TType;
 
 /**
  * Creates a tensor with the given shape.
@@ -37,7 +37,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator
-public final class Empty<T extends TType> extends RawOp implements Operand<T> {
+public final class Empty<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.core.Empty}
@@ -68,7 +68,7 @@ private Options() {
    * @return a new instance of Empty
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> Empty<T> create(Scope scope, Operand<TInt32> shape, DataType<T> dtype, Options... options) {
+  public static <T extends Tensor> Empty<T> create(Scope scope, Operand<TInt32> shape, DataType<T> dtype, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("Empty", scope.makeOpName("Empty"));
     opBuilder.addInput(shape.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/EmptyTensorList.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/EmptyTensorList.java
index 619fb90657f..545beb63c10 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/EmptyTensorList.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/EmptyTensorList.java
@@ -22,13 +22,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Creates and returns an empty tensor list.
@@ -41,7 +41,7 @@
  * element_shape: a shape compatible with that of elements in the list.
  */
 @Operator
-public final class EmptyTensorList extends RawOp implements Operand<TType> {
+public final class EmptyTensorList extends RawOp implements Operand<Tensor> {
   
   /**
    * Factory method to create a class wrapping a new EmptyTensorList operation.
@@ -53,7 +53,7 @@ public final class EmptyTensorList extends RawOp implements Operand<TType> {
    * @return a new instance of EmptyTensorList
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber, U extends TType> EmptyTensorList create(Scope scope, Operand<T> elementShape, Operand<TInt32> maxNumElements, DataType<U> elementDtype) {
+  public static <T extends Tensor & TNumber, U extends Tensor> EmptyTensorList create(Scope scope, Operand<T> elementShape, Operand<TInt32> maxNumElements, DataType<U> elementDtype) {
     OperationBuilder opBuilder = scope.env().opBuilder("EmptyTensorList", scope.makeOpName("EmptyTensorList"));
     opBuilder.addInput(elementShape.asOutput());
     opBuilder.addInput(maxNumElements.asOutput());
@@ -70,8 +70,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) handle;
+  public Output<Tensor> asOutput() {
+    return (Output<Tensor>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/EnsureShape.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/EnsureShape.java
index 8ea86427b7c..3ace0a8822a 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/EnsureShape.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/EnsureShape.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Ensures that the tensor's shape matches the expected shape.
@@ -37,7 +37,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator
-public final class EnsureShape<T extends TType> extends RawOp implements Operand<T> {
+public final class EnsureShape<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new EnsureShape operation.
@@ -48,7 +48,7 @@ public final class EnsureShape<T extends TType> extends RawOp implements Operand
    * @return a new instance of EnsureShape
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> EnsureShape<T> create(Scope scope, Operand<T> input, Shape shape) {
+  public static <T extends Tensor> EnsureShape<T> create(Scope scope, Operand<T> input, Shape shape) {
     OperationBuilder opBuilder = scope.env().opBuilder("EnsureShape", scope.makeOpName("EnsureShape"));
     opBuilder.addInput(input.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Enter.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Enter.java
index fa19763defb..a7ad6608dcb 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Enter.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Enter.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Creates or finds a child frame, and makes `data` available to the child frame.
@@ -38,7 +38,7 @@
  * 
  * @param <T> data type for {@code output()} output
  */
-public final class Enter<T extends TType> extends RawOp implements Operand<T> {
+public final class Enter<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.core.Enter}
@@ -78,7 +78,7 @@ private Options() {
    * @return a new instance of Enter
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> Enter<T> create(Scope scope, Operand<T> data, String frameName, Options... options) {
+  public static <T extends Tensor> Enter<T> create(Scope scope, Operand<T> data, String frameName, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("Enter", scope.makeOpName("Enter"));
     opBuilder.addInput(data.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Exit.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Exit.java
index 4e42bb0bb0e..9c57a566f31 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Exit.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Exit.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Exits the current frame to its parent frame.
@@ -34,7 +34,7 @@
  * 
  * @param <T> data type for {@code output()} output
  */
-public final class Exit<T extends TType> extends RawOp implements Operand<T> {
+public final class Exit<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Exit operation.
@@ -44,7 +44,7 @@ public final class Exit<T extends TType> extends RawOp implements Operand<T> {
    * @return a new instance of Exit
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> Exit<T> create(Scope scope, Operand<T> data) {
+  public static <T extends Tensor> Exit<T> create(Scope scope, Operand<T> data) {
     OperationBuilder opBuilder = scope.env().opBuilder("Exit", scope.makeOpName("Exit"));
     opBuilder.addInput(data.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ExpandDims.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ExpandDims.java
index eb0e977e163..2307b7cd5e1 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ExpandDims.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ExpandDims.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Inserts a dimension of 1 into a tensor's shape.
@@ -63,7 +63,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator
-public final class ExpandDims<T extends TType> extends RawOp implements Operand<T> {
+public final class ExpandDims<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new ExpandDims operation.
@@ -76,7 +76,7 @@ public final class ExpandDims<T extends TType> extends RawOp implements Operand<
    * @return a new instance of ExpandDims
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> ExpandDims<T> create(Scope scope, Operand<T> input, Operand<U> axis) {
+  public static <T extends Tensor, U extends Tensor & TNumber> ExpandDims<T> create(Scope scope, Operand<T> input, Operand<U> axis) {
     OperationBuilder opBuilder = scope.env().opBuilder("ExpandDims", scope.makeOpName("ExpandDims"));
     opBuilder.addInput(input.asOutput());
     opBuilder.addInput(axis.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ExtractVolumePatches.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ExtractVolumePatches.java
index 0b9bcd78e20..a1c5c6fe6d9 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ExtractVolumePatches.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ExtractVolumePatches.java
@@ -22,12 +22,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Extract `patches` from `input` and put them in the "depth" output dimension. 3D extension of `extract_image_patches`.
@@ -35,7 +35,7 @@
  * @param <T> data type for {@code patches()} output
  */
 @Operator
-public final class ExtractVolumePatches<T extends TNumber> extends RawOp implements Operand<T> {
+public final class ExtractVolumePatches<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new ExtractVolumePatches operation.
@@ -56,7 +56,7 @@ public final class ExtractVolumePatches<T extends TNumber> extends RawOp impleme
    * @return a new instance of ExtractVolumePatches
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> ExtractVolumePatches<T> create(Scope scope, Operand<T> input, List<Long> ksizes, List<Long> strides, String padding) {
+  public static <T extends Tensor & TNumber> ExtractVolumePatches<T> create(Scope scope, Operand<T> input, List<Long> ksizes, List<Long> strides, String padding) {
     OperationBuilder opBuilder = scope.env().opBuilder("ExtractVolumePatches", scope.makeOpName("ExtractVolumePatches"));
     opBuilder.addInput(input.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Fill.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Fill.java
index 79b827ea2aa..5a5eb9c69be 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Fill.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Fill.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Creates a tensor filled with a scalar value.
@@ -57,7 +57,7 @@
  * @param <U> data type for {@code output()} output
  */
 @Operator
-public final class Fill<U extends TType> extends RawOp implements Operand<U> {
+public final class Fill<U extends Tensor> extends RawOp implements Operand<U> {
   
   /**
    * Factory method to create a class wrapping a new Fill operation.
@@ -72,7 +72,7 @@ public final class Fill<U extends TType> extends RawOp implements Operand<U> {
    * @return a new instance of Fill
    */
   @Endpoint(describeByClass = true)
-  public static <U extends TType, T extends TNumber> Fill<U> create(Scope scope, Operand<T> dims, Operand<U> value) {
+  public static <U extends Tensor, T extends Tensor & TNumber> Fill<U> create(Scope scope, Operand<T> dims, Operand<U> value) {
     OperationBuilder opBuilder = scope.env().opBuilder("Fill", scope.makeOpName("Fill"));
     opBuilder.addInput(dims.asOutput());
     opBuilder.addInput(value.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Fingerprint.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Fingerprint.java
index 0ea88f2b1f1..90b4a868644 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Fingerprint.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Fingerprint.java
@@ -21,13 +21,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TString;
 import org.tensorflow.types.TUint8;
-import org.tensorflow.types.family.TType;
 
 /**
  * Generates fingerprint values.
@@ -74,7 +74,7 @@ public final class Fingerprint extends RawOp implements Operand<TUint8> {
    * @return a new instance of Fingerprint
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> Fingerprint create(Scope scope, Operand<T> data, Operand<TString> method) {
+  public static <T extends Tensor> Fingerprint create(Scope scope, Operand<T> data, Operand<TString> method) {
     OperationBuilder opBuilder = scope.env().opBuilder("Fingerprint", scope.makeOpName("Fingerprint"));
     opBuilder.addInput(data.asOutput());
     opBuilder.addInput(method.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Gather.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Gather.java
index a9bb2f72e69..33d82ad6a83 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Gather.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Gather.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Gather slices from `params` axis `axis` according to `indices`.
@@ -60,7 +60,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator
-public final class Gather<T extends TType> extends RawOp implements Operand<T> {
+public final class Gather<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.core.Gather}
@@ -94,7 +94,7 @@ private Options() {
    * @return a new instance of Gather
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber, V extends TNumber> Gather<T> create(Scope scope, Operand<T> params, Operand<U> indices, Operand<V> axis, Options... options) {
+  public static <T extends Tensor, U extends Tensor & TNumber, V extends Tensor & TNumber> Gather<T> create(Scope scope, Operand<T> params, Operand<U> indices, Operand<V> axis, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("GatherV2", scope.makeOpName("Gather"));
     opBuilder.addInput(params.asOutput());
     opBuilder.addInput(indices.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/GatherNd.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/GatherNd.java
index e932794001e..dc0e1e1a5e6 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/GatherNd.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/GatherNd.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Gather slices from `params` into a Tensor with shape specified by `indices`.
@@ -127,7 +127,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator
-public final class GatherNd<T extends TType> extends RawOp implements Operand<T> {
+public final class GatherNd<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new GatherNd operation.
@@ -138,7 +138,7 @@ public final class GatherNd<T extends TType> extends RawOp implements Operand<T>
    * @return a new instance of GatherNd
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> GatherNd<T> create(Scope scope, Operand<T> params, Operand<U> indices) {
+  public static <T extends Tensor, U extends Tensor & TNumber> GatherNd<T> create(Scope scope, Operand<T> params, Operand<U> indices) {
     OperationBuilder opBuilder = scope.env().opBuilder("GatherNd", scope.makeOpName("GatherNd"));
     opBuilder.addInput(params.asOutput());
     opBuilder.addInput(indices.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/GetSessionHandle.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/GetSessionHandle.java
index 93a31f1cba1..3eb38f77d30 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/GetSessionHandle.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/GetSessionHandle.java
@@ -21,17 +21,17 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Store the input tensor in the state of the current session.
  */
 @Operator
-public final class GetSessionHandle extends RawOp implements Operand<TType> {
+public final class GetSessionHandle extends RawOp implements Operand<Tensor> {
   
   /**
    * Factory method to create a class wrapping a new GetSessionHandle operation.
@@ -41,7 +41,7 @@ public final class GetSessionHandle extends RawOp implements Operand<TType> {
    * @return a new instance of GetSessionHandle
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> GetSessionHandle create(Scope scope, Operand<T> value) {
+  public static <T extends Tensor> GetSessionHandle create(Scope scope, Operand<T> value) {
     OperationBuilder opBuilder = scope.env().opBuilder("GetSessionHandleV2", scope.makeOpName("GetSessionHandle"));
     opBuilder.addInput(value.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
@@ -58,8 +58,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) handle;
+  public Output<Tensor> asOutput() {
+    return (Output<Tensor>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/GetSessionTensor.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/GetSessionTensor.java
index 93ba5af508c..50e0f66cff2 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/GetSessionTensor.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/GetSessionTensor.java
@@ -22,12 +22,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TString;
-import org.tensorflow.types.family.TType;
 
 /**
  * Get the value of the tensor specified by its handle.
@@ -35,7 +35,7 @@
  * @param <T> data type for {@code value()} output
  */
 @Operator
-public final class GetSessionTensor<T extends TType> extends RawOp implements Operand<T> {
+public final class GetSessionTensor<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new GetSessionTensor operation.
@@ -46,7 +46,7 @@ public final class GetSessionTensor<T extends TType> extends RawOp implements Op
    * @return a new instance of GetSessionTensor
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> GetSessionTensor<T> create(Scope scope, Operand<TString> handle, DataType<T> dtype) {
+  public static <T extends Tensor> GetSessionTensor<T> create(Scope scope, Operand<TString> handle, DataType<T> dtype) {
     OperationBuilder opBuilder = scope.env().opBuilder("GetSessionTensor", scope.makeOpName("GetSessionTensor"));
     opBuilder.addInput(handle.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/GuaranteeConst.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/GuaranteeConst.java
index df0a48ad28f..0c213bec82d 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/GuaranteeConst.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/GuaranteeConst.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Gives a guarantee to the TF runtime that the input tensor is a constant.
@@ -40,7 +40,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator
-public final class GuaranteeConst<T extends TType> extends RawOp implements Operand<T> {
+public final class GuaranteeConst<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new GuaranteeConst operation.
@@ -50,7 +50,7 @@ public final class GuaranteeConst<T extends TType> extends RawOp implements Oper
    * @return a new instance of GuaranteeConst
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> GuaranteeConst<T> create(Scope scope, Operand<T> input) {
+  public static <T extends Tensor> GuaranteeConst<T> create(Scope scope, Operand<T> input) {
     OperationBuilder opBuilder = scope.env().opBuilder("GuaranteeConst", scope.makeOpName("GuaranteeConst"));
     opBuilder.addInput(input.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/HashTable.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/HashTable.java
index 87d9cab4c3f..d02e8d37a01 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/HashTable.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/HashTable.java
@@ -22,11 +22,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Creates a non-initialized hash table.
@@ -36,7 +36,7 @@
  * table will be immutable.
  */
 @Operator
-public final class HashTable extends RawOp implements Operand<TType> {
+public final class HashTable extends RawOp implements Operand<Tensor> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.core.HashTable}
@@ -88,7 +88,7 @@ private Options() {
    * @return a new instance of HashTable
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TType> HashTable create(Scope scope, DataType<T> keyDtype, DataType<U> valueDtype, Options... options) {
+  public static <T extends Tensor, U extends Tensor> HashTable create(Scope scope, DataType<T> keyDtype, DataType<U> valueDtype, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("HashTableV2", scope.makeOpName("HashTable"));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("key_dtype", keyDtype);
@@ -142,8 +142,8 @@ public Output<?> tableHandle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) tableHandle;
+  public Output<Tensor> asOutput() {
+    return (Output<Tensor>) tableHandle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/HistogramFixedWidth.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/HistogramFixedWidth.java
index da1c1a7b713..efd20c1b3e9 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/HistogramFixedWidth.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/HistogramFixedWidth.java
@@ -22,13 +22,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Return histogram of values.
@@ -52,7 +52,7 @@
  * @param <U> data type for {@code out()} output
  */
 @Operator
-public final class HistogramFixedWidth<U extends TNumber> extends RawOp implements Operand<U> {
+public final class HistogramFixedWidth<U extends Tensor & TNumber> extends RawOp implements Operand<U> {
   
   /**
    * Factory method to create a class wrapping a new HistogramFixedWidth operation.
@@ -67,7 +67,7 @@ public final class HistogramFixedWidth<U extends TNumber> extends RawOp implemen
    * @return a new instance of HistogramFixedWidth
    */
   @Endpoint(describeByClass = true)
-  public static <U extends TNumber, T extends TNumber> HistogramFixedWidth<U> create(Scope scope, Operand<T> values, Operand<T> valueRange, Operand<TInt32> nbins, DataType<U> dtype) {
+  public static <U extends Tensor & TNumber, T extends Tensor & TNumber> HistogramFixedWidth<U> create(Scope scope, Operand<T> values, Operand<T> valueRange, Operand<TInt32> nbins, DataType<U> dtype) {
     OperationBuilder opBuilder = scope.env().opBuilder("HistogramFixedWidth", scope.makeOpName("HistogramFixedWidth"));
     opBuilder.addInput(values.asOutput());
     opBuilder.addInput(valueRange.asOutput());
@@ -89,7 +89,7 @@ public static <U extends TNumber, T extends TNumber> HistogramFixedWidth<U> crea
    * @return a new instance of HistogramFixedWidth
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> HistogramFixedWidth<TInt32> create(Scope scope, Operand<T> values, Operand<T> valueRange, Operand<TInt32> nbins) {
+  public static <T extends Tensor & TNumber> HistogramFixedWidth<TInt32> create(Scope scope, Operand<T> values, Operand<T> valueRange, Operand<TInt32> nbins) {
     return create(scope, values, valueRange, nbins, TInt32.DTYPE);
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Identity.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Identity.java
index 1cafffdf5c9..8db7b6ea287 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Identity.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Identity.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Return a tensor with the same shape and contents as the input tensor or value.
@@ -33,7 +33,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator
-public final class Identity<T extends TType> extends RawOp implements Operand<T> {
+public final class Identity<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Identity operation.
@@ -43,7 +43,7 @@ public final class Identity<T extends TType> extends RawOp implements Operand<T>
    * @return a new instance of Identity
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> Identity<T> create(Scope scope, Operand<T> input) {
+  public static <T extends Tensor> Identity<T> create(Scope scope, Operand<T> input) {
     OperationBuilder opBuilder = scope.env().opBuilder("Identity", scope.makeOpName("Identity"));
     opBuilder.addInput(input.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/IdentityN.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/IdentityN.java
index e000005eabf..dff191e5a23 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/IdentityN.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/IdentityN.java
@@ -24,12 +24,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.Operands;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Returns a list of tensors with the same shapes and contents as the input
@@ -51,7 +51,7 @@
  * 
  */
 @Operator
-public final class IdentityN extends RawOp implements Iterable<Operand<TType>> {
+public final class IdentityN extends RawOp implements Iterable<Operand<Tensor>> {
   
   /**
    * Factory method to create a class wrapping a new IdentityN operation.
@@ -76,7 +76,7 @@ public List<Output<?>> output() {
   
   @Override
   @SuppressWarnings({"rawtypes", "unchecked"})
-  public Iterator<Operand<TType>> iterator() {
+  public Iterator<Operand<Tensor>> iterator() {
     return (Iterator) output.iterator();
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ImmutableConst.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ImmutableConst.java
index ecbc3154498..86183c6ee44 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ImmutableConst.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ImmutableConst.java
@@ -22,12 +22,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Returns immutable tensor from memory region.
@@ -37,7 +37,7 @@
  * @param <T> data type for {@code tensor()} output
  */
 @Operator
-public final class ImmutableConst<T extends TType> extends RawOp implements Operand<T> {
+public final class ImmutableConst<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new ImmutableConst operation.
@@ -50,7 +50,7 @@ public final class ImmutableConst<T extends TType> extends RawOp implements Oper
    * @return a new instance of ImmutableConst
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> ImmutableConst<T> create(Scope scope, DataType<T> dtype, Shape shape, String memoryRegionName) {
+  public static <T extends Tensor> ImmutableConst<T> create(Scope scope, DataType<T> dtype, Shape shape, String memoryRegionName) {
     OperationBuilder opBuilder = scope.env().opBuilder("ImmutableConst", scope.makeOpName("ImmutableConst"));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("dtype", dtype);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/InitializeTable.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/InitializeTable.java
index 48662ed420d..50370b6905b 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/InitializeTable.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/InitializeTable.java
@@ -20,11 +20,11 @@
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Table initializer that takes two tensors for keys and values respectively.
@@ -42,7 +42,7 @@ public final class InitializeTable extends RawOp {
    * @return a new instance of InitializeTable
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TType> InitializeTable create(Scope scope, Operand<?> tableHandle, Operand<T> keys, Operand<U> values) {
+  public static <T extends Tensor, U extends Tensor> InitializeTable create(Scope scope, Operand<?> tableHandle, Operand<T> keys, Operand<U> values) {
     OperationBuilder opBuilder = scope.env().opBuilder("InitializeTableV2", scope.makeOpName("InitializeTable"));
     opBuilder.addInput(tableHandle.asOutput());
     opBuilder.addInput(keys.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/InplaceAdd.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/InplaceAdd.java
index 8de34bc569f..eeb7e331b79 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/InplaceAdd.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/InplaceAdd.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
-import org.tensorflow.types.family.TType;
 
 /**
  *     Adds v into specified rows of x.
@@ -36,7 +36,7 @@
  * @param <T> data type for {@code y()} output
  */
 @Operator
-public final class InplaceAdd<T extends TType> extends RawOp implements Operand<T> {
+public final class InplaceAdd<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new InplaceAdd operation.
@@ -48,7 +48,7 @@ public final class InplaceAdd<T extends TType> extends RawOp implements Operand<
    * @return a new instance of InplaceAdd
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> InplaceAdd<T> create(Scope scope, Operand<T> x, Operand<TInt32> i, Operand<T> v) {
+  public static <T extends Tensor> InplaceAdd<T> create(Scope scope, Operand<T> x, Operand<TInt32> i, Operand<T> v) {
     OperationBuilder opBuilder = scope.env().opBuilder("InplaceAdd", scope.makeOpName("InplaceAdd"));
     opBuilder.addInput(x.asOutput());
     opBuilder.addInput(i.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/InplaceSub.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/InplaceSub.java
index 999f793c869..55733c9b0ea 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/InplaceSub.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/InplaceSub.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
-import org.tensorflow.types.family.TType;
 
 /**
  *     Subtracts `v` into specified rows of `x`.
@@ -36,7 +36,7 @@
  * @param <T> data type for {@code y()} output
  */
 @Operator
-public final class InplaceSub<T extends TType> extends RawOp implements Operand<T> {
+public final class InplaceSub<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new InplaceSub operation.
@@ -48,7 +48,7 @@ public final class InplaceSub<T extends TType> extends RawOp implements Operand<
    * @return a new instance of InplaceSub
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> InplaceSub<T> create(Scope scope, Operand<T> x, Operand<TInt32> i, Operand<T> v) {
+  public static <T extends Tensor> InplaceSub<T> create(Scope scope, Operand<T> x, Operand<TInt32> i, Operand<T> v) {
     OperationBuilder opBuilder = scope.env().opBuilder("InplaceSub", scope.makeOpName("InplaceSub"));
     opBuilder.addInput(x.asOutput());
     opBuilder.addInput(i.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/InplaceUpdate.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/InplaceUpdate.java
index f7d2f79e418..dfbac30fa8a 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/InplaceUpdate.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/InplaceUpdate.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
-import org.tensorflow.types.family.TType;
 
 /**
  * Updates specified rows 'i' with values 'v'.
@@ -39,7 +39,7 @@
  * @param <T> data type for {@code y()} output
  */
 @Operator
-public final class InplaceUpdate<T extends TType> extends RawOp implements Operand<T> {
+public final class InplaceUpdate<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new InplaceUpdate operation.
@@ -51,7 +51,7 @@ public final class InplaceUpdate<T extends TType> extends RawOp implements Opera
    * @return a new instance of InplaceUpdate
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> InplaceUpdate<T> create(Scope scope, Operand<T> x, Operand<TInt32> i, Operand<T> v) {
+  public static <T extends Tensor> InplaceUpdate<T> create(Scope scope, Operand<T> x, Operand<TInt32> i, Operand<T> v) {
     OperationBuilder opBuilder = scope.env().opBuilder("InplaceUpdate", scope.makeOpName("InplaceUpdate"));
     opBuilder.addInput(x.asOutput());
     opBuilder.addInput(i.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/IsVariableInitialized.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/IsVariableInitialized.java
index e9dc32ec1d7..54ccb54b831 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/IsVariableInitialized.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/IsVariableInitialized.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TBool;
-import org.tensorflow.types.family.TType;
 
 /**
  * Checks whether a tensor has been initialized.
@@ -44,7 +44,7 @@ public final class IsVariableInitialized extends RawOp implements Operand<TBool>
    * @return a new instance of IsVariableInitialized
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> IsVariableInitialized create(Scope scope, Operand<T> ref) {
+  public static <T extends Tensor> IsVariableInitialized create(Scope scope, Operand<T> ref) {
     OperationBuilder opBuilder = scope.env().opBuilder("IsVariableInitialized", scope.makeOpName("IsVariableInitialized"));
     opBuilder.addInput(ref.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LinSpace.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LinSpace.java
index 04f3959820a..3162c4eec68 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LinSpace.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LinSpace.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Generates values in an interval.
@@ -43,7 +43,8 @@
  * 
  * @param <T> data type for {@code output()} output
  */
-public final class LinSpace<T extends TNumber> extends RawOp implements Operand<T> {
+@Operator
+public final class LinSpace<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new LinSpace operation.
@@ -55,7 +56,7 @@ public final class LinSpace<T extends TNumber> extends RawOp implements Operand<
    * @return a new instance of LinSpace
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber, U extends TNumber> LinSpace<T> create(Scope scope, Operand<T> start, Operand<T> stop, Operand<U> num) {
+  public static <T extends Tensor & TNumber, U extends Tensor & TNumber> LinSpace<T> create(Scope scope, Operand<T> start, Operand<T> stop, Operand<U> num) {
     OperationBuilder opBuilder = scope.env().opBuilder("LinSpace", scope.makeOpName("LinSpace"));
     opBuilder.addInput(start.asOutput());
     opBuilder.addInput(stop.asOutput());
@@ -76,9 +77,6 @@ public Output<T> asOutput() {
     return output;
   }
   
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "LinSpace";
-  
   private Output<T> output;
   
   private LinSpace(Operation operation) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LookupTableExport.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LookupTableExport.java
index 6bebbb35895..a106f123343 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LookupTableExport.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LookupTableExport.java
@@ -22,11 +22,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Outputs all keys and values in the table.
@@ -35,7 +35,7 @@
  * @param <U> data type for {@code values()} output
  */
 @Operator
-public final class LookupTableExport<T extends TType, U extends TType> extends RawOp {
+public final class LookupTableExport<T extends Tensor, U extends Tensor> extends RawOp {
   
   /**
    * Factory method to create a class wrapping a new LookupTableExport operation.
@@ -47,7 +47,7 @@ public final class LookupTableExport<T extends TType, U extends TType> extends R
    * @return a new instance of LookupTableExport
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TType> LookupTableExport<T, U> create(Scope scope, Operand<?> tableHandle, DataType<T> Tkeys, DataType<U> Tvalues) {
+  public static <T extends Tensor, U extends Tensor> LookupTableExport<T, U> create(Scope scope, Operand<?> tableHandle, DataType<T> Tkeys, DataType<U> Tvalues) {
     OperationBuilder opBuilder = scope.env().opBuilder("LookupTableExportV2", scope.makeOpName("LookupTableExport"));
     opBuilder.addInput(tableHandle.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LookupTableFind.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LookupTableFind.java
index bfc8e92e16f..beba84cda66 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LookupTableFind.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LookupTableFind.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Looks up keys in a table, outputs the corresponding values.
@@ -39,7 +39,7 @@
  * @param <U> data type for {@code values()} output
  */
 @Operator
-public final class LookupTableFind<U extends TType> extends RawOp implements Operand<U> {
+public final class LookupTableFind<U extends Tensor> extends RawOp implements Operand<U> {
   
   /**
    * Factory method to create a class wrapping a new LookupTableFind operation.
@@ -51,7 +51,7 @@ public final class LookupTableFind<U extends TType> extends RawOp implements Ope
    * @return a new instance of LookupTableFind
    */
   @Endpoint(describeByClass = true)
-  public static <U extends TType, T extends TType> LookupTableFind<U> create(Scope scope, Operand<?> tableHandle, Operand<T> keys, Operand<U> defaultValue) {
+  public static <U extends Tensor, T extends Tensor> LookupTableFind<U> create(Scope scope, Operand<?> tableHandle, Operand<T> keys, Operand<U> defaultValue) {
     OperationBuilder opBuilder = scope.env().opBuilder("LookupTableFindV2", scope.makeOpName("LookupTableFind"));
     opBuilder.addInput(tableHandle.asOutput());
     opBuilder.addInput(keys.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LookupTableImport.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LookupTableImport.java
index 9884a40e3cb..c5b9222bbf9 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LookupTableImport.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LookupTableImport.java
@@ -20,11 +20,11 @@
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Replaces the contents of the table with the specified keys and values.
@@ -45,7 +45,7 @@ public final class LookupTableImport extends RawOp {
    * @return a new instance of LookupTableImport
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TType> LookupTableImport create(Scope scope, Operand<?> tableHandle, Operand<T> keys, Operand<U> values) {
+  public static <T extends Tensor, U extends Tensor> LookupTableImport create(Scope scope, Operand<?> tableHandle, Operand<T> keys, Operand<U> values) {
     OperationBuilder opBuilder = scope.env().opBuilder("LookupTableImportV2", scope.makeOpName("LookupTableImport"));
     opBuilder.addInput(tableHandle.asOutput());
     opBuilder.addInput(keys.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LookupTableInsert.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LookupTableInsert.java
index 0f09ae25d1b..3055286604b 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LookupTableInsert.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LookupTableInsert.java
@@ -20,11 +20,11 @@
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Updates the table to associates keys with values.
@@ -45,7 +45,7 @@ public final class LookupTableInsert extends RawOp {
    * @return a new instance of LookupTableInsert
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TType> LookupTableInsert create(Scope scope, Operand<?> tableHandle, Operand<T> keys, Operand<U> values) {
+  public static <T extends Tensor, U extends Tensor> LookupTableInsert create(Scope scope, Operand<?> tableHandle, Operand<T> keys, Operand<U> values) {
     OperationBuilder opBuilder = scope.env().opBuilder("LookupTableInsertV2", scope.makeOpName("LookupTableInsert"));
     opBuilder.addInput(tableHandle.asOutput());
     opBuilder.addInput(keys.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LookupTableRemove.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LookupTableRemove.java
index 41463ad7539..54da587f516 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LookupTableRemove.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LookupTableRemove.java
@@ -20,11 +20,11 @@
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Removes keys and its associated values from a table.
@@ -43,7 +43,7 @@ public final class LookupTableRemove extends RawOp {
    * @return a new instance of LookupTableRemove
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> LookupTableRemove create(Scope scope, Operand<?> tableHandle, Operand<T> keys) {
+  public static <T extends Tensor> LookupTableRemove create(Scope scope, Operand<?> tableHandle, Operand<T> keys) {
     OperationBuilder opBuilder = scope.env().opBuilder("LookupTableRemoveV2", scope.makeOpName("LookupTableRemove"));
     opBuilder.addInput(tableHandle.asOutput());
     opBuilder.addInput(keys.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LowerBound.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LowerBound.java
index 6f0f5158cca..89387991fba 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LowerBound.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LowerBound.java
@@ -22,13 +22,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Applies lower_bound(sorted_search_values, values) along each row.
@@ -53,7 +53,7 @@
  * 
  * @param <U> data type for {@code output()} output
  */
-public final class LowerBound<U extends TNumber> extends RawOp implements Operand<U> {
+public final class LowerBound<U extends Tensor & TNumber> extends RawOp implements Operand<U> {
   
   /**
    * Factory method to create a class wrapping a new LowerBound operation.
@@ -66,7 +66,7 @@ public final class LowerBound<U extends TNumber> extends RawOp implements Operan
    * @return a new instance of LowerBound
    */
   @Endpoint(describeByClass = true)
-  public static <U extends TNumber, T extends TType> LowerBound<U> create(Scope scope, Operand<T> sortedInputs, Operand<T> values, DataType<U> outType) {
+  public static <U extends Tensor & TNumber, T extends Tensor> LowerBound<U> create(Scope scope, Operand<T> sortedInputs, Operand<T> values, DataType<U> outType) {
     OperationBuilder opBuilder = scope.env().opBuilder("LowerBound", scope.makeOpName("LowerBound"));
     opBuilder.addInput(sortedInputs.asOutput());
     opBuilder.addInput(values.asOutput());
@@ -85,7 +85,7 @@ public static <U extends TNumber, T extends TType> LowerBound<U> create(Scope sc
    * @return a new instance of LowerBound
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> LowerBound<TInt32> create(Scope scope, Operand<T> sortedInputs, Operand<T> values) {
+  public static <T extends Tensor> LowerBound<TInt32> create(Scope scope, Operand<T> sortedInputs, Operand<T> values) {
     return create(scope, sortedInputs, values, TInt32.DTYPE);
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MapPeek.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MapPeek.java
index 1925ca680ea..61541aa0306 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MapPeek.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MapPeek.java
@@ -25,13 +25,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
 import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
 
 /**
  * Op peeks at the values at the specified key.  If the
@@ -40,7 +40,7 @@
  * this op will block until it does.
  */
 @Operator
-public final class MapPeek extends RawOp implements Iterable<Operand<TType>> {
+public final class MapPeek extends RawOp implements Iterable<Operand<Tensor>> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.core.MapPeek}
@@ -164,7 +164,7 @@ public List<Output<?>> values() {
   
   @Override
   @SuppressWarnings({"rawtypes", "unchecked"})
-  public Iterator<Operand<TType>> iterator() {
+  public Iterator<Operand<Tensor>> iterator() {
     return (Iterator) values.iterator();
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MapUnstage.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MapUnstage.java
index 849f6f3ef6a..9f83b8b3346 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MapUnstage.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MapUnstage.java
@@ -25,13 +25,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
 import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
 
 /**
  * Op removes and returns the values associated with the key
@@ -40,7 +40,7 @@
  * does not contain this key, the op will block until it does.
  */
 @Operator
-public final class MapUnstage extends RawOp implements Iterable<Operand<TType>> {
+public final class MapUnstage extends RawOp implements Iterable<Operand<Tensor>> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.core.MapUnstage}
@@ -164,7 +164,7 @@ public List<Output<?>> values() {
   
   @Override
   @SuppressWarnings({"rawtypes", "unchecked"})
-  public Iterator<Operand<TType>> iterator() {
+  public Iterator<Operand<Tensor>> iterator() {
     return (Iterator) values.iterator();
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Max.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Max.java
index 94bcd0f544c..e26de0bcdc7 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Max.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Max.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes the maximum of elements across dimensions of a tensor.
@@ -39,7 +39,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator
-public final class Max<T extends TType> extends RawOp implements Operand<T> {
+public final class Max<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.core.Max}
@@ -71,7 +71,7 @@ private Options() {
    * @return a new instance of Max
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> Max<T> create(Scope scope, Operand<T> input, Operand<U> axis, Options... options) {
+  public static <T extends Tensor, U extends Tensor & TNumber> Max<T> create(Scope scope, Operand<T> input, Operand<U> axis, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("Max", scope.makeOpName("Max"));
     opBuilder.addInput(input.asOutput());
     opBuilder.addInput(axis.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Merge.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Merge.java
index 41eca5fc33f..6d70f5cbe93 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Merge.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Merge.java
@@ -21,13 +21,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.Operands;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
-import org.tensorflow.types.family.TType;
 
 /**
  * Forwards the value of an available tensor from `inputs` to `output`.
@@ -41,7 +41,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator
-public final class Merge<T extends TType> extends RawOp {
+public final class Merge<T extends Tensor> extends RawOp {
   
   /**
    * Factory method to create a class wrapping a new Merge operation.
@@ -51,7 +51,7 @@ public final class Merge<T extends TType> extends RawOp {
    * @return a new instance of Merge
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> Merge<T> create(Scope scope, Iterable<Operand<T>> inputs) {
+  public static <T extends Tensor> Merge<T> create(Scope scope, Iterable<Operand<T>> inputs) {
     OperationBuilder opBuilder = scope.env().opBuilder("Merge", scope.makeOpName("Merge"));
     opBuilder.addInputList(Operands.asOutputs(inputs));
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Min.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Min.java
index ed8ea366e7a..0052104ab5c 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Min.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Min.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes the minimum of elements across dimensions of a tensor.
@@ -39,7 +39,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator
-public final class Min<T extends TType> extends RawOp implements Operand<T> {
+public final class Min<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.core.Min}
@@ -71,7 +71,7 @@ private Options() {
    * @return a new instance of Min
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> Min<T> create(Scope scope, Operand<T> input, Operand<U> axis, Options... options) {
+  public static <T extends Tensor, U extends Tensor & TNumber> Min<T> create(Scope scope, Operand<T> input, Operand<U> axis, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("Min", scope.makeOpName("Min"));
     opBuilder.addInput(input.asOutput());
     opBuilder.addInput(axis.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MirrorPad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MirrorPad.java
index 6c8292a671a..f54b533b0b2 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MirrorPad.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MirrorPad.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Pads a tensor with mirrored values.
@@ -60,7 +60,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator
-public final class MirrorPad<T extends TType> extends RawOp implements Operand<T> {
+public final class MirrorPad<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new MirrorPad operation.
@@ -77,7 +77,7 @@ public final class MirrorPad<T extends TType> extends RawOp implements Operand<T
    * @return a new instance of MirrorPad
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> MirrorPad<T> create(Scope scope, Operand<T> input, Operand<U> paddings, String mode) {
+  public static <T extends Tensor, U extends Tensor & TNumber> MirrorPad<T> create(Scope scope, Operand<T> input, Operand<U> paddings, String mode) {
     OperationBuilder opBuilder = scope.env().opBuilder("MirrorPad", scope.makeOpName("MirrorPad"));
     opBuilder.addInput(input.asOutput());
     opBuilder.addInput(paddings.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MirrorPadGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MirrorPadGrad.java
index 7b1727c3fb5..1285a8b183d 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MirrorPadGrad.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MirrorPadGrad.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Gradient op for `MirrorPad` op. This op folds a mirror-padded tensor.
@@ -52,7 +52,7 @@
  * 
  * @param <T> data type for {@code output()} output
  */
-public final class MirrorPadGrad<T extends TType> extends RawOp implements Operand<T> {
+public final class MirrorPadGrad<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new MirrorPadGrad operation.
@@ -65,7 +65,7 @@ public final class MirrorPadGrad<T extends TType> extends RawOp implements Opera
    * @return a new instance of MirrorPadGrad
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> MirrorPadGrad<T> create(Scope scope, Operand<T> input, Operand<U> paddings, String mode) {
+  public static <T extends Tensor, U extends Tensor & TNumber> MirrorPadGrad<T> create(Scope scope, Operand<T> input, Operand<U> paddings, String mode) {
     OperationBuilder opBuilder = scope.env().opBuilder("MirrorPadGrad", scope.makeOpName("MirrorPadGrad"));
     opBuilder.addInput(input.asOutput());
     opBuilder.addInput(paddings.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MlirPassthroughOp.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MlirPassthroughOp.java
index cc278fdae8f..3898a019645 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MlirPassthroughOp.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MlirPassthroughOp.java
@@ -25,12 +25,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.Operands;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Wraps an arbitrary MLIR computation expressed as a module with a main() function.
@@ -66,7 +66,7 @@
  * 
  */
 @Operator
-public final class MlirPassthroughOp extends RawOp implements Iterable<Operand<TType>> {
+public final class MlirPassthroughOp extends RawOp implements Iterable<Operand<Tensor>> {
   
   /**
    * Factory method to create a class wrapping a new MlirPassthroughOp operation.
@@ -99,7 +99,7 @@ public List<Output<?>> outputs() {
   
   @Override
   @SuppressWarnings({"rawtypes", "unchecked"})
-  public Iterator<Operand<TType>> iterator() {
+  public Iterator<Operand<Tensor>> iterator() {
     return (Iterator) outputs.iterator();
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MutableDenseHashTable.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MutableDenseHashTable.java
index 4ff3c1f3ea6..1c21813f1aa 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MutableDenseHashTable.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MutableDenseHashTable.java
@@ -22,12 +22,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Creates an empty hash table that uses tensors as the backing store.
@@ -40,7 +40,7 @@
  * the insert operations. It does not support the initialization operation.
  */
 @Operator
-public final class MutableDenseHashTable extends RawOp implements Operand<TType> {
+public final class MutableDenseHashTable extends RawOp implements Operand<Tensor> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.core.MutableDenseHashTable}
@@ -122,7 +122,7 @@ private Options() {
    * @return a new instance of MutableDenseHashTable
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TType> MutableDenseHashTable create(Scope scope, Operand<T> emptyKey, Operand<T> deletedKey, DataType<U> valueDtype, Options... options) {
+  public static <T extends Tensor, U extends Tensor> MutableDenseHashTable create(Scope scope, Operand<T> emptyKey, Operand<T> deletedKey, DataType<U> valueDtype, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("MutableDenseHashTableV2", scope.makeOpName("MutableDenseHashTable"));
     opBuilder.addInput(emptyKey.asOutput());
     opBuilder.addInput(deletedKey.asOutput());
@@ -208,8 +208,8 @@ public Output<?> tableHandle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) tableHandle;
+  public Output<Tensor> asOutput() {
+    return (Output<Tensor>) tableHandle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MutableHashTable.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MutableHashTable.java
index a11789f9f34..b8ff6808125 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MutableHashTable.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MutableHashTable.java
@@ -22,11 +22,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Creates an empty hash table.
@@ -36,7 +36,7 @@
  * the insert operations. It does not support the initialization operation.
  */
 @Operator
-public final class MutableHashTable extends RawOp implements Operand<TType> {
+public final class MutableHashTable extends RawOp implements Operand<Tensor> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.core.MutableHashTable}
@@ -88,7 +88,7 @@ private Options() {
    * @return a new instance of MutableHashTable
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TType> MutableHashTable create(Scope scope, DataType<T> keyDtype, DataType<U> valueDtype, Options... options) {
+  public static <T extends Tensor, U extends Tensor> MutableHashTable create(Scope scope, DataType<T> keyDtype, DataType<U> valueDtype, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("MutableHashTableV2", scope.makeOpName("MutableHashTable"));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("key_dtype", keyDtype);
@@ -142,8 +142,8 @@ public Output<?> tableHandle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) tableHandle;
+  public Output<Tensor> asOutput() {
+    return (Output<Tensor>) tableHandle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MutableHashTableOfTensors.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MutableHashTableOfTensors.java
index 975f040ae3b..b444f3ffc68 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MutableHashTableOfTensors.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MutableHashTableOfTensors.java
@@ -22,12 +22,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Creates an empty hash table.
@@ -37,7 +37,7 @@
  * the insert operations. It does not support the initialization operation.
  */
 @Operator
-public final class MutableHashTableOfTensors extends RawOp implements Operand<TType> {
+public final class MutableHashTableOfTensors extends RawOp implements Operand<Tensor> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.core.MutableHashTableOfTensors}
@@ -97,7 +97,7 @@ private Options() {
    * @return a new instance of MutableHashTableOfTensors
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TType> MutableHashTableOfTensors create(Scope scope, DataType<T> keyDtype, DataType<U> valueDtype, Options... options) {
+  public static <T extends Tensor, U extends Tensor> MutableHashTableOfTensors create(Scope scope, DataType<T> keyDtype, DataType<U> valueDtype, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("MutableHashTableOfTensorsV2", scope.makeOpName("MutableHashTableOfTensors"));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("key_dtype", keyDtype);
@@ -160,8 +160,8 @@ public Output<?> tableHandle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) tableHandle;
+  public Output<Tensor> asOutput() {
+    return (Output<Tensor>) tableHandle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Mutex.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Mutex.java
index 1af2b439372..20325657299 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Mutex.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Mutex.java
@@ -21,17 +21,17 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Creates a Mutex resource that can be locked by `MutexLock`.
  */
 @Operator
-public final class Mutex extends RawOp implements Operand<TType> {
+public final class Mutex extends RawOp implements Operand<Tensor> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.core.Mutex}
@@ -112,8 +112,8 @@ public Output<?> resource() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) resource;
+  public Output<Tensor> asOutput() {
+    return (Output<Tensor>) resource;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MutexLock.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MutexLock.java
index de28b8f5025..ca11c7c739f 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MutexLock.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MutexLock.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Locks a mutex resource.  The output is the lock.  So long as the lock tensor
@@ -67,7 +67,7 @@
  * wish to ensure the usage is exclusive.
  */
 @Operator
-public final class MutexLock extends RawOp implements Operand<TType> {
+public final class MutexLock extends RawOp implements Operand<Tensor> {
   
   /**
    * Factory method to create a class wrapping a new MutexLock operation.
@@ -95,8 +95,8 @@ public Output<?> mutexLock() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) mutexLock;
+  public Output<Tensor> asOutput() {
+    return (Output<Tensor>) mutexLock;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/NcclAllReduce.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/NcclAllReduce.java
index a44611b5a41..fd293049078 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/NcclAllReduce.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/NcclAllReduce.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Outputs a tensor containing the reduction across all input tensors.
@@ -46,7 +46,7 @@
  * 
  * @param <T> data type for {@code output()} output
  */
-public final class NcclAllReduce<T extends TNumber> extends RawOp implements Operand<T> {
+public final class NcclAllReduce<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new NcclAllReduce operation.
@@ -59,7 +59,7 @@ public final class NcclAllReduce<T extends TNumber> extends RawOp implements Ope
    * @return a new instance of NcclAllReduce
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> NcclAllReduce<T> create(Scope scope, Operand<T> input, String reduction, Long numDevices, String sharedName) {
+  public static <T extends Tensor & TNumber> NcclAllReduce<T> create(Scope scope, Operand<T> input, String reduction, Long numDevices, String sharedName) {
     OperationBuilder opBuilder = scope.env().opBuilder("NcclAllReduce", scope.makeOpName("NcclAllReduce"));
     opBuilder.addInput(input.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/NcclBroadcast.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/NcclBroadcast.java
index da0832e437a..0d321927ac7 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/NcclBroadcast.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/NcclBroadcast.java
@@ -21,13 +21,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Sends `input` to all devices that are connected to the output.
@@ -44,7 +44,7 @@
  * 
  * @param <T> data type for {@code output()} output
  */
-public final class NcclBroadcast<T extends TNumber> extends RawOp implements Operand<T> {
+public final class NcclBroadcast<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new NcclBroadcast operation.
@@ -55,7 +55,7 @@ public final class NcclBroadcast<T extends TNumber> extends RawOp implements Ope
    * @return a new instance of NcclBroadcast
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> NcclBroadcast<T> create(Scope scope, Operand<T> input, Shape shape) {
+  public static <T extends Tensor & TNumber> NcclBroadcast<T> create(Scope scope, Operand<T> input, Shape shape) {
     OperationBuilder opBuilder = scope.env().opBuilder("NcclBroadcast", scope.makeOpName("NcclBroadcast"));
     opBuilder.addInput(input.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/NcclReduce.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/NcclReduce.java
index ced1473e60a..d7ff17f2243 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/NcclReduce.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/NcclReduce.java
@@ -21,13 +21,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.Operands;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Reduces `input` from `num_devices` using `reduction` to a single device.
@@ -43,7 +43,7 @@
  * 
  * @param <T> data type for {@code output()} output
  */
-public final class NcclReduce<T extends TNumber> extends RawOp implements Operand<T> {
+public final class NcclReduce<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new NcclReduce operation.
@@ -54,7 +54,7 @@ public final class NcclReduce<T extends TNumber> extends RawOp implements Operan
    * @return a new instance of NcclReduce
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> NcclReduce<T> create(Scope scope, Iterable<Operand<T>> input, String reduction) {
+  public static <T extends Tensor & TNumber> NcclReduce<T> create(Scope scope, Iterable<Operand<T>> input, String reduction) {
     OperationBuilder opBuilder = scope.env().opBuilder("NcclReduce", scope.makeOpName("NcclReduce"));
     opBuilder.addInputList(Operands.asOutputs(input));
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/NextIteration.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/NextIteration.java
index 69b6be6a810..ab2fb154b18 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/NextIteration.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/NextIteration.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Makes its input available to the next iteration.
@@ -33,7 +33,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator
-public final class NextIteration<T extends TType> extends RawOp implements Operand<T> {
+public final class NextIteration<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new NextIteration operation.
@@ -43,7 +43,7 @@ public final class NextIteration<T extends TType> extends RawOp implements Opera
    * @return a new instance of NextIteration
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> NextIteration<T> create(Scope scope, Operand<T> data) {
+  public static <T extends Tensor> NextIteration<T> create(Scope scope, Operand<T> data) {
     OperationBuilder opBuilder = scope.env().opBuilder("NextIteration", scope.makeOpName("NextIteration"));
     opBuilder.addInput(data.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OneHot.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OneHot.java
index 442b9ae9108..ba8a1ff3290 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OneHot.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OneHot.java
@@ -21,13 +21,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Returns a one-hot tensor.
@@ -116,7 +116,7 @@
  * @param <U> data type for {@code output()} output
  */
 @Operator
-public final class OneHot<U extends TType> extends RawOp implements Operand<U> {
+public final class OneHot<U extends Tensor> extends RawOp implements Operand<U> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.core.OneHot}
@@ -149,7 +149,7 @@ private Options() {
    * @return a new instance of OneHot
    */
   @Endpoint(describeByClass = true)
-  public static <U extends TType, T extends TNumber> OneHot<U> create(Scope scope, Operand<T> indices, Operand<TInt32> depth, Operand<U> onValue, Operand<U> offValue, Options... options) {
+  public static <U extends Tensor, T extends Tensor & TNumber> OneHot<U> create(Scope scope, Operand<T> indices, Operand<TInt32> depth, Operand<U> onValue, Operand<U> offValue, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("OneHot", scope.makeOpName("OneHot"));
     opBuilder.addInput(indices.asOutput());
     opBuilder.addInput(depth.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OnesLike.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OnesLike.java
index 986d0201f26..d565be08723 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OnesLike.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OnesLike.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Returns a tensor of ones with the same shape and type as x.
@@ -33,7 +33,7 @@
  * @param <T> data type for {@code y()} output
  */
 @Operator
-public final class OnesLike<T extends TType> extends RawOp implements Operand<T> {
+public final class OnesLike<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new OnesLike operation.
@@ -43,7 +43,7 @@ public final class OnesLike<T extends TType> extends RawOp implements Operand<T>
    * @return a new instance of OnesLike
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> OnesLike<T> create(Scope scope, Operand<T> x) {
+  public static <T extends Tensor> OnesLike<T> create(Scope scope, Operand<T> x) {
     OperationBuilder opBuilder = scope.env().opBuilder("OnesLike", scope.makeOpName("OnesLike"));
     opBuilder.addInput(x.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OrderedMapPeek.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OrderedMapPeek.java
index 21c6adcc039..80cd39960c0 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OrderedMapPeek.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OrderedMapPeek.java
@@ -25,13 +25,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
 import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
 
 /**
  * Op peeks at the values at the specified key.  If the
@@ -41,7 +41,7 @@
  * performance.
  */
 @Operator
-public final class OrderedMapPeek extends RawOp implements Iterable<Operand<TType>> {
+public final class OrderedMapPeek extends RawOp implements Iterable<Operand<Tensor>> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.core.OrderedMapPeek}
@@ -165,7 +165,7 @@ public List<Output<?>> values() {
   
   @Override
   @SuppressWarnings({"rawtypes", "unchecked"})
-  public Iterator<Operand<TType>> iterator() {
+  public Iterator<Operand<Tensor>> iterator() {
     return (Iterator) values.iterator();
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OrderedMapUnstage.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OrderedMapUnstage.java
index e2460e42dd8..d5e8769b433 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OrderedMapUnstage.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OrderedMapUnstage.java
@@ -25,13 +25,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
 import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
 
 /**
  * Op removes and returns the values associated with the key
@@ -40,7 +40,7 @@
  * does not contain this key, the op will block until it does.
  */
 @Operator
-public final class OrderedMapUnstage extends RawOp implements Iterable<Operand<TType>> {
+public final class OrderedMapUnstage extends RawOp implements Iterable<Operand<Tensor>> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.core.OrderedMapUnstage}
@@ -164,7 +164,7 @@ public List<Output<?>> values() {
   
   @Override
   @SuppressWarnings({"rawtypes", "unchecked"})
-  public Iterator<Operand<TType>> iterator() {
+  public Iterator<Operand<Tensor>> iterator() {
     return (Iterator) values.iterator();
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Pad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Pad.java
index 205e6ab0a62..27915f9670f 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Pad.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Pad.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Pads a tensor.
@@ -59,7 +59,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator
-public final class Pad<T extends TType> extends RawOp implements Operand<T> {
+public final class Pad<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Pad operation.
@@ -71,7 +71,7 @@ public final class Pad<T extends TType> extends RawOp implements Operand<T> {
    * @return a new instance of Pad
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> Pad<T> create(Scope scope, Operand<T> input, Operand<U> paddings, Operand<T> constantValues) {
+  public static <T extends Tensor, U extends Tensor & TNumber> Pad<T> create(Scope scope, Operand<T> input, Operand<U> paddings, Operand<T> constantValues) {
     OperationBuilder opBuilder = scope.env().opBuilder("PadV2", scope.makeOpName("Pad"));
     opBuilder.addInput(input.asOutput());
     opBuilder.addInput(paddings.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ParallelConcat.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ParallelConcat.java
index e373612bb68..e3675988da6 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ParallelConcat.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ParallelConcat.java
@@ -21,13 +21,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.Operands;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Concatenates a list of `N` tensors along the first dimension.
@@ -50,7 +50,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator
-public final class ParallelConcat<T extends TType> extends RawOp implements Operand<T> {
+public final class ParallelConcat<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new ParallelConcat operation.
@@ -63,7 +63,7 @@ public final class ParallelConcat<T extends TType> extends RawOp implements Oper
    * @return a new instance of ParallelConcat
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> ParallelConcat<T> create(Scope scope, Iterable<Operand<T>> values, Shape shape) {
+  public static <T extends Tensor> ParallelConcat<T> create(Scope scope, Iterable<Operand<T>> values, Shape shape) {
     OperationBuilder opBuilder = scope.env().opBuilder("ParallelConcat", scope.makeOpName("ParallelConcat"));
     opBuilder.addInputList(Operands.asOutputs(values));
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ParallelDynamicStitch.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ParallelDynamicStitch.java
index 2770fa01591..50cf7aaf7b9 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ParallelDynamicStitch.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ParallelDynamicStitch.java
@@ -21,13 +21,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.Operands;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
-import org.tensorflow.types.family.TType;
 
 /**
  * Interleave the values from the `data` tensors into a single tensor.
@@ -89,7 +89,7 @@
  * @param <T> data type for {@code merged()} output
  */
 @Operator
-public final class ParallelDynamicStitch<T extends TType> extends RawOp implements Operand<T> {
+public final class ParallelDynamicStitch<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new ParallelDynamicStitch operation.
@@ -100,7 +100,7 @@ public final class ParallelDynamicStitch<T extends TType> extends RawOp implemen
    * @return a new instance of ParallelDynamicStitch
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> ParallelDynamicStitch<T> create(Scope scope, Iterable<Operand<TInt32>> indices, Iterable<Operand<T>> data) {
+  public static <T extends Tensor> ParallelDynamicStitch<T> create(Scope scope, Iterable<Operand<TInt32>> indices, Iterable<Operand<T>> data) {
     OperationBuilder opBuilder = scope.env().opBuilder("ParallelDynamicStitch", scope.makeOpName("ParallelDynamicStitch"));
     opBuilder.addInputList(Operands.asOutputs(indices));
     opBuilder.addInputList(Operands.asOutputs(data));
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Placeholder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Placeholder.java
index caef9fc0783..dc96e6e9b36 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Placeholder.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Placeholder.java
@@ -22,12 +22,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * A placeholder op for a value that will be fed into the computation.
@@ -39,7 +39,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator
-public final class Placeholder<T extends TType> extends RawOp implements Operand<T> {
+public final class Placeholder<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.core.Placeholder}
@@ -70,7 +70,7 @@ private Options() {
    * @return a new instance of Placeholder
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> Placeholder<T> create(Scope scope, DataType<T> dtype, Options... options) {
+  public static <T extends Tensor> Placeholder<T> create(Scope scope, DataType<T> dtype, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("Placeholder", scope.makeOpName("Placeholder"));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("dtype", dtype);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/PlaceholderWithDefault.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/PlaceholderWithDefault.java
index 59a9ca223ab..0fb4696509f 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/PlaceholderWithDefault.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/PlaceholderWithDefault.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * A placeholder op that passes through `input` when its output is not fed.
@@ -34,7 +34,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator
-public final class PlaceholderWithDefault<T extends TType> extends RawOp implements Operand<T> {
+public final class PlaceholderWithDefault<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new PlaceholderWithDefault operation.
@@ -45,7 +45,7 @@ public final class PlaceholderWithDefault<T extends TType> extends RawOp impleme
    * @return a new instance of PlaceholderWithDefault
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> PlaceholderWithDefault<T> create(Scope scope, Operand<T> input, Shape shape) {
+  public static <T extends Tensor> PlaceholderWithDefault<T> create(Scope scope, Operand<T> input, Shape shape) {
     OperationBuilder opBuilder = scope.env().opBuilder("PlaceholderWithDefault", scope.makeOpName("PlaceholderWithDefault"));
     opBuilder.addInput(input.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Prod.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Prod.java
index 6bf1b7cd37d..f56bceb7edd 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Prod.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Prod.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes the product of elements across dimensions of a tensor.
@@ -39,7 +39,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator
-public final class Prod<T extends TType> extends RawOp implements Operand<T> {
+public final class Prod<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.core.Prod}
@@ -71,7 +71,7 @@ private Options() {
    * @return a new instance of Prod
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> Prod<T> create(Scope scope, Operand<T> input, Operand<U> axis, Options... options) {
+  public static <T extends Tensor, U extends Tensor & TNumber> Prod<T> create(Scope scope, Operand<T> input, Operand<U> axis, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("Prod", scope.makeOpName("Prod"));
     opBuilder.addInput(input.asOutput());
     opBuilder.addInput(axis.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/QuantizedReshape.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/QuantizedReshape.java
index e644855bbc9..fa59a2151c6 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/QuantizedReshape.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/QuantizedReshape.java
@@ -21,13 +21,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TFloat32;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Reshapes a quantized tensor as per the Reshape op.
@@ -37,7 +37,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator
-public final class QuantizedReshape<T extends TType> extends RawOp {
+public final class QuantizedReshape<T extends Tensor> extends RawOp {
   
   /**
    * Factory method to create a class wrapping a new QuantizedReshape operation.
@@ -50,7 +50,7 @@ public final class QuantizedReshape<T extends TType> extends RawOp {
    * @return a new instance of QuantizedReshape
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> QuantizedReshape<T> create(Scope scope, Operand<T> tensor, Operand<U> shape, Operand<TFloat32> inputMin, Operand<TFloat32> inputMax) {
+  public static <T extends Tensor, U extends Tensor & TNumber> QuantizedReshape<T> create(Scope scope, Operand<T> tensor, Operand<U> shape, Operand<TFloat32> inputMin, Operand<TFloat32> inputMax) {
     OperationBuilder opBuilder = scope.env().opBuilder("QuantizedReshape", scope.makeOpName("QuantizedReshape"));
     opBuilder.addInput(tensor.asOutput());
     opBuilder.addInput(shape.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Range.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Range.java
index 967a490382e..04d0235dedd 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Range.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Range.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Creates a sequence of numbers.
@@ -46,7 +46,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator
-public final class Range<T extends TNumber> extends RawOp implements Operand<T> {
+public final class Range<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Range operation.
@@ -58,7 +58,7 @@ public final class Range<T extends TNumber> extends RawOp implements Operand<T>
    * @return a new instance of Range
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> Range<T> create(Scope scope, Operand<T> start, Operand<T> limit, Operand<T> delta) {
+  public static <T extends Tensor & TNumber> Range<T> create(Scope scope, Operand<T> start, Operand<T> limit, Operand<T> delta) {
     OperationBuilder opBuilder = scope.env().opBuilder("Range", scope.makeOpName("Range"));
     opBuilder.addInput(start.asOutput());
     opBuilder.addInput(limit.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Rank.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Rank.java
index 071551d6492..8a5fafed6aa 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Rank.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Rank.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
-import org.tensorflow.types.family.TType;
 
 /**
  * Returns the rank of a tensor.
@@ -54,7 +54,7 @@ public final class Rank extends RawOp implements Operand<TInt32> {
    * @return a new instance of Rank
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> Rank create(Scope scope, Operand<T> input) {
+  public static <T extends Tensor> Rank create(Scope scope, Operand<T> input) {
     OperationBuilder opBuilder = scope.env().opBuilder("Rank", scope.makeOpName("Rank"));
     opBuilder.addInput(input.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ReadVariableOp.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ReadVariableOp.java
index aa2d1d2b9b2..198303a6b79 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ReadVariableOp.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ReadVariableOp.java
@@ -22,11 +22,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Reads the value of a variable.
@@ -41,7 +41,7 @@
  * @param <T> data type for {@code value()} output
  */
 @Operator
-public final class ReadVariableOp<T extends TType> extends RawOp implements Operand<T> {
+public final class ReadVariableOp<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new ReadVariableOp operation.
@@ -52,7 +52,7 @@ public final class ReadVariableOp<T extends TType> extends RawOp implements Oper
    * @return a new instance of ReadVariableOp
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> ReadVariableOp<T> create(Scope scope, Operand<?> resource, DataType<T> dtype) {
+  public static <T extends Tensor> ReadVariableOp<T> create(Scope scope, Operand<?> resource, DataType<T> dtype) {
     OperationBuilder opBuilder = scope.env().opBuilder("ReadVariableOp", scope.makeOpName("ReadVariableOp"));
     opBuilder.addInput(resource.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Recv.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Recv.java
index db6511c1583..20f3835f6e1 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Recv.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Recv.java
@@ -22,18 +22,18 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Receives the named tensor from send_device on recv_device.
  * 
  * @param <T> data type for {@code tensor()} output
  */
-public final class Recv<T extends TType> extends RawOp implements Operand<T> {
+public final class Recv<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.core.Recv}
@@ -70,7 +70,7 @@ private Options() {
    * @return a new instance of Recv
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> Recv<T> create(Scope scope, DataType<T> tensorType, String tensorName, String sendDevice, Long sendDeviceIncarnation, String recvDevice, Options... options) {
+  public static <T extends Tensor> Recv<T> create(Scope scope, DataType<T> tensorType, String tensorName, String sendDevice, Long sendDeviceIncarnation, String recvDevice, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("Recv", scope.makeOpName("Recv"));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("tensor_type", tensorType);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ReduceAll.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ReduceAll.java
index 9a5ad026ac8..8b975f18afa 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ReduceAll.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ReduceAll.java
@@ -21,13 +21,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TBool;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes the "logical and" of elements across dimensions of a tensor.
@@ -70,7 +70,7 @@ private Options() {
    * @return a new instance of ReduceAll
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> ReduceAll create(Scope scope, Operand<TBool> input, Operand<T> axis, Options... options) {
+  public static <T extends Tensor & TNumber> ReduceAll create(Scope scope, Operand<TBool> input, Operand<T> axis, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("All", scope.makeOpName("ReduceAll"));
     opBuilder.addInput(input.asOutput());
     opBuilder.addInput(axis.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ReduceAny.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ReduceAny.java
index de479629f97..28bda505e65 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ReduceAny.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ReduceAny.java
@@ -21,13 +21,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TBool;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes the "logical or" of elements across dimensions of a tensor.
@@ -70,7 +70,7 @@ private Options() {
    * @return a new instance of ReduceAny
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> ReduceAny create(Scope scope, Operand<TBool> input, Operand<T> axis, Options... options) {
+  public static <T extends Tensor & TNumber> ReduceAny create(Scope scope, Operand<TBool> input, Operand<T> axis, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("Any", scope.makeOpName("ReduceAny"));
     opBuilder.addInput(input.asOutput());
     opBuilder.addInput(axis.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ReduceMax.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ReduceMax.java
index 997481542b8..c153b785f4c 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ReduceMax.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ReduceMax.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes the maximum of elements across dimensions of a tensor.
@@ -39,7 +39,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator
-public final class ReduceMax<T extends TType> extends RawOp implements Operand<T> {
+public final class ReduceMax<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.core.ReduceMax}
@@ -71,7 +71,7 @@ private Options() {
    * @return a new instance of ReduceMax
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> ReduceMax<T> create(Scope scope, Operand<T> input, Operand<U> axis, Options... options) {
+  public static <T extends Tensor, U extends Tensor & TNumber> ReduceMax<T> create(Scope scope, Operand<T> input, Operand<U> axis, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("Max", scope.makeOpName("ReduceMax"));
     opBuilder.addInput(input.asOutput());
     opBuilder.addInput(axis.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ReduceMin.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ReduceMin.java
index 60d64c3a58c..8f676a509f1 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ReduceMin.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ReduceMin.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes the minimum of elements across dimensions of a tensor.
@@ -39,7 +39,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator
-public final class ReduceMin<T extends TType> extends RawOp implements Operand<T> {
+public final class ReduceMin<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.core.ReduceMin}
@@ -71,7 +71,7 @@ private Options() {
    * @return a new instance of ReduceMin
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> ReduceMin<T> create(Scope scope, Operand<T> input, Operand<U> axis, Options... options) {
+  public static <T extends Tensor, U extends Tensor & TNumber> ReduceMin<T> create(Scope scope, Operand<T> input, Operand<U> axis, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("Min", scope.makeOpName("ReduceMin"));
     opBuilder.addInput(input.asOutput());
     opBuilder.addInput(axis.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ReduceProd.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ReduceProd.java
index 7c9872758b1..cc352b66784 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ReduceProd.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ReduceProd.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes the product of elements across dimensions of a tensor.
@@ -39,7 +39,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator
-public final class ReduceProd<T extends TType> extends RawOp implements Operand<T> {
+public final class ReduceProd<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.core.ReduceProd}
@@ -71,7 +71,7 @@ private Options() {
    * @return a new instance of ReduceProd
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> ReduceProd<T> create(Scope scope, Operand<T> input, Operand<U> axis, Options... options) {
+  public static <T extends Tensor, U extends Tensor & TNumber> ReduceProd<T> create(Scope scope, Operand<T> input, Operand<U> axis, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("Prod", scope.makeOpName("ReduceProd"));
     opBuilder.addInput(input.asOutput());
     opBuilder.addInput(axis.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ReduceSum.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ReduceSum.java
index f87e5b22e15..d34cd8e448f 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ReduceSum.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ReduceSum.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes the sum of elements across dimensions of a tensor.
@@ -39,7 +39,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator
-public final class ReduceSum<T extends TType> extends RawOp implements Operand<T> {
+public final class ReduceSum<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.core.ReduceSum}
@@ -71,7 +71,7 @@ private Options() {
    * @return a new instance of ReduceSum
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> ReduceSum<T> create(Scope scope, Operand<T> input, Operand<U> axis, Options... options) {
+  public static <T extends Tensor, U extends Tensor & TNumber> ReduceSum<T> create(Scope scope, Operand<T> input, Operand<U> axis, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("Sum", scope.makeOpName("ReduceSum"));
     opBuilder.addInput(input.asOutput());
     opBuilder.addInput(axis.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/RefEnter.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/RefEnter.java
index 8292dcbb379..d6cfef23cab 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/RefEnter.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/RefEnter.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Creates or finds a child frame, and makes `data` available to the child frame.
@@ -37,7 +37,7 @@
  * 
  * @param <T> data type for {@code output()} output
  */
-public final class RefEnter<T extends TType> extends RawOp implements Operand<T> {
+public final class RefEnter<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.core.RefEnter}
@@ -77,7 +77,7 @@ private Options() {
    * @return a new instance of RefEnter
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> RefEnter<T> create(Scope scope, Operand<T> data, String frameName, Options... options) {
+  public static <T extends Tensor> RefEnter<T> create(Scope scope, Operand<T> data, String frameName, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("RefEnter", scope.makeOpName("RefEnter"));
     opBuilder.addInput(data.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/RefExit.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/RefExit.java
index d8896a59211..9cfeac60763 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/RefExit.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/RefExit.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Exits the current frame to its parent frame.
@@ -34,7 +34,7 @@
  * 
  * @param <T> data type for {@code output()} output
  */
-public final class RefExit<T extends TType> extends RawOp implements Operand<T> {
+public final class RefExit<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new RefExit operation.
@@ -44,7 +44,7 @@ public final class RefExit<T extends TType> extends RawOp implements Operand<T>
    * @return a new instance of RefExit
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> RefExit<T> create(Scope scope, Operand<T> data) {
+  public static <T extends Tensor> RefExit<T> create(Scope scope, Operand<T> data) {
     OperationBuilder opBuilder = scope.env().opBuilder("RefExit", scope.makeOpName("RefExit"));
     opBuilder.addInput(data.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/RefIdentity.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/RefIdentity.java
index 9deff66c3a9..cd757f96b48 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/RefIdentity.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/RefIdentity.java
@@ -21,18 +21,18 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Return the same ref tensor as the input ref tensor.
  * 
  * @param <T> data type for {@code output()} output
  */
-public final class RefIdentity<T extends TType> extends RawOp implements Operand<T> {
+public final class RefIdentity<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new RefIdentity operation.
@@ -42,7 +42,7 @@ public final class RefIdentity<T extends TType> extends RawOp implements Operand
    * @return a new instance of RefIdentity
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> RefIdentity<T> create(Scope scope, Operand<T> input) {
+  public static <T extends Tensor> RefIdentity<T> create(Scope scope, Operand<T> input) {
     OperationBuilder opBuilder = scope.env().opBuilder("RefIdentity", scope.makeOpName("RefIdentity"));
     opBuilder.addInput(input.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/RefMerge.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/RefMerge.java
index 990cfaaadb6..a8d2580fb7d 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/RefMerge.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/RefMerge.java
@@ -21,13 +21,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.Operands;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
-import org.tensorflow.types.family.TType;
 
 /**
  * Forwards the value of an available tensor from `inputs` to `output`.
@@ -40,7 +40,7 @@
  * 
  * @param <T> data type for {@code output()} output
  */
-public final class RefMerge<T extends TType> extends RawOp {
+public final class RefMerge<T extends Tensor> extends RawOp {
   
   /**
    * Factory method to create a class wrapping a new RefMerge operation.
@@ -50,7 +50,7 @@ public final class RefMerge<T extends TType> extends RawOp {
    * @return a new instance of RefMerge
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> RefMerge<T> create(Scope scope, Iterable<Operand<T>> inputs) {
+  public static <T extends Tensor> RefMerge<T> create(Scope scope, Iterable<Operand<T>> inputs) {
     OperationBuilder opBuilder = scope.env().opBuilder("RefMerge", scope.makeOpName("RefMerge"));
     opBuilder.addInputList(Operands.asOutputs(inputs));
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/RefNextIteration.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/RefNextIteration.java
index c151662820e..ad367e48118 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/RefNextIteration.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/RefNextIteration.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Makes its input available to the next iteration.
@@ -33,7 +33,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator
-public final class RefNextIteration<T extends TType> extends RawOp implements Operand<T> {
+public final class RefNextIteration<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new RefNextIteration operation.
@@ -43,7 +43,7 @@ public final class RefNextIteration<T extends TType> extends RawOp implements Op
    * @return a new instance of RefNextIteration
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> RefNextIteration<T> create(Scope scope, Operand<T> data) {
+  public static <T extends Tensor> RefNextIteration<T> create(Scope scope, Operand<T> data) {
     OperationBuilder opBuilder = scope.env().opBuilder("RefNextIteration", scope.makeOpName("RefNextIteration"));
     opBuilder.addInput(data.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/RefSelect.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/RefSelect.java
index 8be3ae40e9e..690a3e785d7 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/RefSelect.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/RefSelect.java
@@ -21,13 +21,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.Operands;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
-import org.tensorflow.types.family.TType;
 
 /**
  * Forwards the `index`th element of `inputs` to `output`.
@@ -35,7 +35,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator
-public final class RefSelect<T extends TType> extends RawOp implements Operand<T> {
+public final class RefSelect<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new RefSelect operation.
@@ -46,7 +46,7 @@ public final class RefSelect<T extends TType> extends RawOp implements Operand<T
    * @return a new instance of RefSelect
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> RefSelect<T> create(Scope scope, Operand<TInt32> index, Iterable<Operand<T>> inputs) {
+  public static <T extends Tensor> RefSelect<T> create(Scope scope, Operand<TInt32> index, Iterable<Operand<T>> inputs) {
     OperationBuilder opBuilder = scope.env().opBuilder("RefSelect", scope.makeOpName("RefSelect"));
     opBuilder.addInput(index.asOutput());
     opBuilder.addInputList(Operands.asOutputs(inputs));
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/RefSwitch.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/RefSwitch.java
index f0128f668fc..8233a076458 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/RefSwitch.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/RefSwitch.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TBool;
-import org.tensorflow.types.family.TType;
 
 /**
  * Forwards the ref tensor `data` to the output port determined by `pred`.
@@ -39,7 +39,7 @@
  * @param <T> data type for {@code outputFalse()} output
  */
 @Operator
-public final class RefSwitch<T extends TType> extends RawOp {
+public final class RefSwitch<T extends Tensor> extends RawOp {
   
   /**
    * Factory method to create a class wrapping a new RefSwitch operation.
@@ -50,7 +50,7 @@ public final class RefSwitch<T extends TType> extends RawOp {
    * @return a new instance of RefSwitch
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> RefSwitch<T> create(Scope scope, Operand<T> data, Operand<TBool> pred) {
+  public static <T extends Tensor> RefSwitch<T> create(Scope scope, Operand<T> data, Operand<TBool> pred) {
     OperationBuilder opBuilder = scope.env().opBuilder("RefSwitch", scope.makeOpName("RefSwitch"));
     opBuilder.addInput(data.asOutput());
     opBuilder.addInput(pred.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/RemoteFusedGraphExecute.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/RemoteFusedGraphExecute.java
index bd76549c976..6b46ba055ab 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/RemoteFusedGraphExecute.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/RemoteFusedGraphExecute.java
@@ -25,12 +25,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.Operands;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Execute a sub graph on a remote processor.
@@ -44,7 +44,7 @@
  * will be passed to consumer nodes as outputs of this node.
  */
 @Operator
-public final class RemoteFusedGraphExecute extends RawOp implements Iterable<Operand<TType>> {
+public final class RemoteFusedGraphExecute extends RawOp implements Iterable<Operand<Tensor>> {
   
   /**
    * Factory method to create a class wrapping a new RemoteFusedGraphExecute operation.
@@ -79,7 +79,7 @@ public List<Output<?>> outputs() {
   
   @Override
   @SuppressWarnings({"rawtypes", "unchecked"})
-  public Iterator<Operand<TType>> iterator() {
+  public Iterator<Operand<Tensor>> iterator() {
     return (Iterator) outputs.iterator();
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Reshape.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Reshape.java
index 2829a332b0e..4f012475d4a 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Reshape.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Reshape.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Reshapes a tensor.
@@ -94,7 +94,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator
-public final class Reshape<T extends TType> extends RawOp implements Operand<T> {
+public final class Reshape<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Reshape operation.
@@ -105,7 +105,7 @@ public final class Reshape<T extends TType> extends RawOp implements Operand<T>
    * @return a new instance of Reshape
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> Reshape<T> create(Scope scope, Operand<T> tensor, Operand<U> shape) {
+  public static <T extends Tensor, U extends Tensor & TNumber> Reshape<T> create(Scope scope, Operand<T> tensor, Operand<U> shape) {
     OperationBuilder opBuilder = scope.env().opBuilder("Reshape", scope.makeOpName("Reshape"));
     opBuilder.addInput(tensor.asOutput());
     opBuilder.addInput(shape.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceCountUpTo.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceCountUpTo.java
index 19f630dd014..5278d579d3e 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceCountUpTo.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceCountUpTo.java
@@ -22,12 +22,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Increments variable pointed to by 'resource' until it reaches 'limit'.
@@ -35,7 +35,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator
-public final class ResourceCountUpTo<T extends TNumber> extends RawOp implements Operand<T> {
+public final class ResourceCountUpTo<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new ResourceCountUpTo operation.
@@ -48,7 +48,7 @@ public final class ResourceCountUpTo<T extends TNumber> extends RawOp implements
    * @return a new instance of ResourceCountUpTo
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> ResourceCountUpTo<T> create(Scope scope, Operand<?> resource, Long limit, DataType<T> T) {
+  public static <T extends Tensor & TNumber> ResourceCountUpTo<T> create(Scope scope, Operand<?> resource, Long limit, DataType<T> T) {
     OperationBuilder opBuilder = scope.env().opBuilder("ResourceCountUpTo", scope.makeOpName("ResourceCountUpTo"));
     opBuilder.addInput(resource.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceGather.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceGather.java
index 38aa1fbd407..35974fdd095 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceGather.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceGather.java
@@ -22,12 +22,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Gather slices from the variable pointed to by `resource` according to `indices`.
@@ -49,7 +49,7 @@
  * @param <U> data type for {@code output()} output
  */
 @Operator
-public final class ResourceGather<U extends TType> extends RawOp implements Operand<U> {
+public final class ResourceGather<U extends Tensor> extends RawOp implements Operand<U> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.core.ResourceGather}
@@ -90,7 +90,7 @@ private Options() {
    * @return a new instance of ResourceGather
    */
   @Endpoint(describeByClass = true)
-  public static <U extends TType, T extends TNumber> ResourceGather<U> create(Scope scope, Operand<?> resource, Operand<T> indices, DataType<U> dtype, Options... options) {
+  public static <U extends Tensor, T extends Tensor & TNumber> ResourceGather<U> create(Scope scope, Operand<?> resource, Operand<T> indices, DataType<U> dtype, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ResourceGather", scope.makeOpName("ResourceGather"));
     opBuilder.addInput(resource.asOutput());
     opBuilder.addInput(indices.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceGatherNd.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceGatherNd.java
index 85e422179f7..d43d682e05a 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceGatherNd.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceGatherNd.java
@@ -22,18 +22,18 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * @param <U> data type for {@code output()} output
  */
 @Operator
-public final class ResourceGatherNd<U extends TType> extends RawOp implements Operand<U> {
+public final class ResourceGatherNd<U extends Tensor> extends RawOp implements Operand<U> {
   
   /**
    * Factory method to create a class wrapping a new ResourceGatherNd operation.
@@ -45,7 +45,7 @@ public final class ResourceGatherNd<U extends TType> extends RawOp implements Op
    * @return a new instance of ResourceGatherNd
    */
   @Endpoint(describeByClass = true)
-  public static <U extends TType, T extends TNumber> ResourceGatherNd<U> create(Scope scope, Operand<?> resource, Operand<T> indices, DataType<U> dtype) {
+  public static <U extends Tensor, T extends Tensor & TNumber> ResourceGatherNd<U> create(Scope scope, Operand<?> resource, Operand<T> indices, DataType<U> dtype) {
     OperationBuilder opBuilder = scope.env().opBuilder("ResourceGatherNd", scope.makeOpName("ResourceGatherNd"));
     opBuilder.addInput(resource.asOutput());
     opBuilder.addInput(indices.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterAdd.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterAdd.java
index 0966dd5fcc4..61f9428edef 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterAdd.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterAdd.java
@@ -20,12 +20,12 @@
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Adds sparse updates to the variable referenced by `resource`.
@@ -63,7 +63,7 @@ public final class ResourceScatterAdd extends RawOp {
    * @return a new instance of ResourceScatterAdd
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber, U extends TType> ResourceScatterAdd create(Scope scope, Operand<?> resource, Operand<T> indices, Operand<U> updates) {
+  public static <T extends Tensor & TNumber, U extends Tensor> ResourceScatterAdd create(Scope scope, Operand<?> resource, Operand<T> indices, Operand<U> updates) {
     OperationBuilder opBuilder = scope.env().opBuilder("ResourceScatterAdd", scope.makeOpName("ResourceScatterAdd"));
     opBuilder.addInput(resource.asOutput());
     opBuilder.addInput(indices.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterDiv.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterDiv.java
index 9560bddf284..3482a3886d3 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterDiv.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterDiv.java
@@ -20,12 +20,12 @@
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Divides sparse updates into the variable referenced by `resource`.
@@ -63,7 +63,7 @@ public final class ResourceScatterDiv extends RawOp {
    * @return a new instance of ResourceScatterDiv
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber, U extends TType> ResourceScatterDiv create(Scope scope, Operand<?> resource, Operand<T> indices, Operand<U> updates) {
+  public static <T extends Tensor & TNumber, U extends Tensor> ResourceScatterDiv create(Scope scope, Operand<?> resource, Operand<T> indices, Operand<U> updates) {
     OperationBuilder opBuilder = scope.env().opBuilder("ResourceScatterDiv", scope.makeOpName("ResourceScatterDiv"));
     opBuilder.addInput(resource.asOutput());
     opBuilder.addInput(indices.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterMax.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterMax.java
index ce952ee19ba..891d761acf8 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterMax.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterMax.java
@@ -20,12 +20,12 @@
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Reduces sparse updates into the variable referenced by `resource` using the `max` operation.
@@ -63,7 +63,7 @@ public final class ResourceScatterMax extends RawOp {
    * @return a new instance of ResourceScatterMax
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber, U extends TType> ResourceScatterMax create(Scope scope, Operand<?> resource, Operand<T> indices, Operand<U> updates) {
+  public static <T extends Tensor & TNumber, U extends Tensor> ResourceScatterMax create(Scope scope, Operand<?> resource, Operand<T> indices, Operand<U> updates) {
     OperationBuilder opBuilder = scope.env().opBuilder("ResourceScatterMax", scope.makeOpName("ResourceScatterMax"));
     opBuilder.addInput(resource.asOutput());
     opBuilder.addInput(indices.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterMin.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterMin.java
index 51ec6b7637e..2d083c02f2a 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterMin.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterMin.java
@@ -20,12 +20,12 @@
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Reduces sparse updates into the variable referenced by `resource` using the `min` operation.
@@ -63,7 +63,7 @@ public final class ResourceScatterMin extends RawOp {
    * @return a new instance of ResourceScatterMin
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber, U extends TType> ResourceScatterMin create(Scope scope, Operand<?> resource, Operand<T> indices, Operand<U> updates) {
+  public static <T extends Tensor & TNumber, U extends Tensor> ResourceScatterMin create(Scope scope, Operand<?> resource, Operand<T> indices, Operand<U> updates) {
     OperationBuilder opBuilder = scope.env().opBuilder("ResourceScatterMin", scope.makeOpName("ResourceScatterMin"));
     opBuilder.addInput(resource.asOutput());
     opBuilder.addInput(indices.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterMul.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterMul.java
index 2d5f71e006d..7acf531602c 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterMul.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterMul.java
@@ -20,12 +20,12 @@
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Multiplies sparse updates into the variable referenced by `resource`.
@@ -63,7 +63,7 @@ public final class ResourceScatterMul extends RawOp {
    * @return a new instance of ResourceScatterMul
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber, U extends TType> ResourceScatterMul create(Scope scope, Operand<?> resource, Operand<T> indices, Operand<U> updates) {
+  public static <T extends Tensor & TNumber, U extends Tensor> ResourceScatterMul create(Scope scope, Operand<?> resource, Operand<T> indices, Operand<U> updates) {
     OperationBuilder opBuilder = scope.env().opBuilder("ResourceScatterMul", scope.makeOpName("ResourceScatterMul"));
     opBuilder.addInput(resource.asOutput());
     opBuilder.addInput(indices.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterNdAdd.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterNdAdd.java
index 11e45c33098..a5a82746ec4 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterNdAdd.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterNdAdd.java
@@ -20,12 +20,12 @@
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Applies sparse addition to individual values or slices in a Variable.
@@ -97,7 +97,7 @@ private Options() {
    * @return a new instance of ResourceScatterNdAdd
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber, U extends TType> ResourceScatterNdAdd create(Scope scope, Operand<?> ref, Operand<T> indices, Operand<U> updates, Options... options) {
+  public static <T extends Tensor & TNumber, U extends Tensor> ResourceScatterNdAdd create(Scope scope, Operand<?> ref, Operand<T> indices, Operand<U> updates, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ResourceScatterNdAdd", scope.makeOpName("ResourceScatterNdAdd"));
     opBuilder.addInput(ref.asOutput());
     opBuilder.addInput(indices.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterNdSub.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterNdSub.java
index 267099b7cfc..129f51fd2ae 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterNdSub.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterNdSub.java
@@ -20,12 +20,12 @@
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Applies sparse subtraction to individual values or slices in a Variable.
@@ -97,7 +97,7 @@ private Options() {
    * @return a new instance of ResourceScatterNdSub
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber, U extends TType> ResourceScatterNdSub create(Scope scope, Operand<?> ref, Operand<T> indices, Operand<U> updates, Options... options) {
+  public static <T extends Tensor & TNumber, U extends Tensor> ResourceScatterNdSub create(Scope scope, Operand<?> ref, Operand<T> indices, Operand<U> updates, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ResourceScatterNdSub", scope.makeOpName("ResourceScatterNdSub"));
     opBuilder.addInput(ref.asOutput());
     opBuilder.addInput(indices.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterNdUpdate.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterNdUpdate.java
index 4a1e875bc97..d32d9c48717 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterNdUpdate.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterNdUpdate.java
@@ -20,12 +20,12 @@
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Applies sparse `updates` to individual values or slices within a given
@@ -99,7 +99,7 @@ private Options() {
    * @return a new instance of ResourceScatterNdUpdate
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber, U extends TType> ResourceScatterNdUpdate create(Scope scope, Operand<?> ref, Operand<T> indices, Operand<U> updates, Options... options) {
+  public static <T extends Tensor & TNumber, U extends Tensor> ResourceScatterNdUpdate create(Scope scope, Operand<?> ref, Operand<T> indices, Operand<U> updates, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ResourceScatterNdUpdate", scope.makeOpName("ResourceScatterNdUpdate"));
     opBuilder.addInput(ref.asOutput());
     opBuilder.addInput(indices.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterSub.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterSub.java
index 7b772fab997..89d66b01f58 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterSub.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterSub.java
@@ -20,12 +20,12 @@
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Subtracts sparse updates from the variable referenced by `resource`.
@@ -63,7 +63,7 @@ public final class ResourceScatterSub extends RawOp {
    * @return a new instance of ResourceScatterSub
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber, U extends TType> ResourceScatterSub create(Scope scope, Operand<?> resource, Operand<T> indices, Operand<U> updates) {
+  public static <T extends Tensor & TNumber, U extends Tensor> ResourceScatterSub create(Scope scope, Operand<?> resource, Operand<T> indices, Operand<U> updates) {
     OperationBuilder opBuilder = scope.env().opBuilder("ResourceScatterSub", scope.makeOpName("ResourceScatterSub"));
     opBuilder.addInput(resource.asOutput());
     opBuilder.addInput(indices.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterUpdate.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterUpdate.java
index 067ddf5f205..1f97ce5f21d 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterUpdate.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterUpdate.java
@@ -20,12 +20,12 @@
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Assigns sparse updates to the variable referenced by `resource`.
@@ -54,7 +54,7 @@ public final class ResourceScatterUpdate extends RawOp {
    * @return a new instance of ResourceScatterUpdate
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber, U extends TType> ResourceScatterUpdate create(Scope scope, Operand<?> resource, Operand<T> indices, Operand<U> updates) {
+  public static <T extends Tensor & TNumber, U extends Tensor> ResourceScatterUpdate create(Scope scope, Operand<?> resource, Operand<T> indices, Operand<U> updates) {
     OperationBuilder opBuilder = scope.env().opBuilder("ResourceScatterUpdate", scope.makeOpName("ResourceScatterUpdate"));
     opBuilder.addInput(resource.asOutput());
     opBuilder.addInput(indices.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceStridedSliceAssign.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceStridedSliceAssign.java
index 4deb4c55f64..e8a999a0e4c 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceStridedSliceAssign.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceStridedSliceAssign.java
@@ -20,12 +20,12 @@
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Assign `value` to the sliced l-value reference of `ref`.
@@ -108,7 +108,7 @@ private Options() {
    * @return a new instance of ResourceStridedSliceAssign
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber, U extends TType> ResourceStridedSliceAssign create(Scope scope, Operand<?> ref, Operand<T> begin, Operand<T> end, Operand<T> strides, Operand<U> value, Options... options) {
+  public static <T extends Tensor & TNumber, U extends Tensor> ResourceStridedSliceAssign create(Scope scope, Operand<?> ref, Operand<T> begin, Operand<T> end, Operand<T> strides, Operand<U> value, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ResourceStridedSliceAssign", scope.makeOpName("ResourceStridedSliceAssign"));
     opBuilder.addInput(ref.asOutput());
     opBuilder.addInput(begin.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Reverse.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Reverse.java
index 340336ee9f9..0334a7bce4c 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Reverse.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Reverse.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Reverses specific dimensions of a tensor.
@@ -81,7 +81,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator
-public final class Reverse<T extends TType> extends RawOp implements Operand<T> {
+public final class Reverse<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Reverse operation.
@@ -93,7 +93,7 @@ public final class Reverse<T extends TType> extends RawOp implements Operand<T>
    * @return a new instance of Reverse
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> Reverse<T> create(Scope scope, Operand<T> tensor, Operand<U> axis) {
+  public static <T extends Tensor, U extends Tensor & TNumber> Reverse<T> create(Scope scope, Operand<T> tensor, Operand<U> axis) {
     OperationBuilder opBuilder = scope.env().opBuilder("ReverseV2", scope.makeOpName("Reverse"));
     opBuilder.addInput(tensor.asOutput());
     opBuilder.addInput(axis.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ReverseSequence.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ReverseSequence.java
index 46fd2b54aa9..bc3e1b44bd2 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ReverseSequence.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ReverseSequence.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Reverses variable length slices.
@@ -87,7 +87,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator
-public final class ReverseSequence<T extends TType> extends RawOp implements Operand<T> {
+public final class ReverseSequence<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.core.ReverseSequence}
@@ -120,7 +120,7 @@ private Options() {
    * @return a new instance of ReverseSequence
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> ReverseSequence<T> create(Scope scope, Operand<T> input, Operand<U> seqLengths, Long seqDim, Options... options) {
+  public static <T extends Tensor, U extends Tensor & TNumber> ReverseSequence<T> create(Scope scope, Operand<T> input, Operand<U> seqLengths, Long seqDim, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ReverseSequence", scope.makeOpName("ReverseSequence"));
     opBuilder.addInput(input.asOutput());
     opBuilder.addInput(seqLengths.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Roll.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Roll.java
index 761be92e533..3b2cd8a7413 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Roll.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Roll.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Rolls the elements of a tensor along an axis.
@@ -55,7 +55,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator
-public final class Roll<T extends TType> extends RawOp implements Operand<T> {
+public final class Roll<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Roll operation.
@@ -73,7 +73,7 @@ public final class Roll<T extends TType> extends RawOp implements Operand<T> {
    * @return a new instance of Roll
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber, V extends TNumber> Roll<T> create(Scope scope, Operand<T> input, Operand<U> shift, Operand<V> axis) {
+  public static <T extends Tensor, U extends Tensor & TNumber, V extends Tensor & TNumber> Roll<T> create(Scope scope, Operand<T> input, Operand<U> shift, Operand<V> axis) {
     OperationBuilder opBuilder = scope.env().opBuilder("Roll", scope.makeOpName("Roll"));
     opBuilder.addInput(input.asOutput());
     opBuilder.addInput(shift.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterAdd.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterAdd.java
index c91e799a643..88def7cc922 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterAdd.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterAdd.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Adds sparse updates to a variable reference.
@@ -57,7 +57,7 @@
  * @param <T> data type for {@code outputRef()} output
  */
 @Operator
-public final class ScatterAdd<T extends TType> extends RawOp implements Operand<T> {
+public final class ScatterAdd<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.core.ScatterAdd}
@@ -90,7 +90,7 @@ private Options() {
    * @return a new instance of ScatterAdd
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> ScatterAdd<T> create(Scope scope, Operand<T> ref, Operand<U> indices, Operand<T> updates, Options... options) {
+  public static <T extends Tensor, U extends Tensor & TNumber> ScatterAdd<T> create(Scope scope, Operand<T> ref, Operand<U> indices, Operand<T> updates, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ScatterAdd", scope.makeOpName("ScatterAdd"));
     opBuilder.addInput(ref.asOutput());
     opBuilder.addInput(indices.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterDiv.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterDiv.java
index 12a27c67960..1f56778e276 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterDiv.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterDiv.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Divides a variable reference by sparse updates.
@@ -53,7 +53,7 @@
  * @param <T> data type for {@code outputRef()} output
  */
 @Operator
-public final class ScatterDiv<T extends TType> extends RawOp implements Operand<T> {
+public final class ScatterDiv<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.core.ScatterDiv}
@@ -86,7 +86,7 @@ private Options() {
    * @return a new instance of ScatterDiv
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> ScatterDiv<T> create(Scope scope, Operand<T> ref, Operand<U> indices, Operand<T> updates, Options... options) {
+  public static <T extends Tensor, U extends Tensor & TNumber> ScatterDiv<T> create(Scope scope, Operand<T> ref, Operand<U> indices, Operand<T> updates, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ScatterDiv", scope.makeOpName("ScatterDiv"));
     opBuilder.addInput(ref.asOutput());
     opBuilder.addInput(indices.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterMax.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterMax.java
index 20b84f016ea..441317eb24b 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterMax.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterMax.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Reduces sparse updates into a variable reference using the `max` operation.
@@ -57,7 +57,7 @@
  * @param <T> data type for {@code outputRef()} output
  */
 @Operator
-public final class ScatterMax<T extends TNumber> extends RawOp implements Operand<T> {
+public final class ScatterMax<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.core.ScatterMax}
@@ -90,7 +90,7 @@ private Options() {
    * @return a new instance of ScatterMax
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber, U extends TNumber> ScatterMax<T> create(Scope scope, Operand<T> ref, Operand<U> indices, Operand<T> updates, Options... options) {
+  public static <T extends Tensor & TNumber, U extends Tensor & TNumber> ScatterMax<T> create(Scope scope, Operand<T> ref, Operand<U> indices, Operand<T> updates, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ScatterMax", scope.makeOpName("ScatterMax"));
     opBuilder.addInput(ref.asOutput());
     opBuilder.addInput(indices.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterMin.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterMin.java
index 5010da14eaa..1b2e7cf358d 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterMin.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterMin.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Reduces sparse updates into a variable reference using the `min` operation.
@@ -57,7 +57,7 @@
  * @param <T> data type for {@code outputRef()} output
  */
 @Operator
-public final class ScatterMin<T extends TNumber> extends RawOp implements Operand<T> {
+public final class ScatterMin<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.core.ScatterMin}
@@ -90,7 +90,7 @@ private Options() {
    * @return a new instance of ScatterMin
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber, U extends TNumber> ScatterMin<T> create(Scope scope, Operand<T> ref, Operand<U> indices, Operand<T> updates, Options... options) {
+  public static <T extends Tensor & TNumber, U extends Tensor & TNumber> ScatterMin<T> create(Scope scope, Operand<T> ref, Operand<U> indices, Operand<T> updates, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ScatterMin", scope.makeOpName("ScatterMin"));
     opBuilder.addInput(ref.asOutput());
     opBuilder.addInput(indices.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterMul.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterMul.java
index ba0892e1a04..5ad16de15fb 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterMul.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterMul.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Multiplies sparse updates into a variable reference.
@@ -53,7 +53,7 @@
  * @param <T> data type for {@code outputRef()} output
  */
 @Operator
-public final class ScatterMul<T extends TType> extends RawOp implements Operand<T> {
+public final class ScatterMul<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.core.ScatterMul}
@@ -86,7 +86,7 @@ private Options() {
    * @return a new instance of ScatterMul
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> ScatterMul<T> create(Scope scope, Operand<T> ref, Operand<U> indices, Operand<T> updates, Options... options) {
+  public static <T extends Tensor, U extends Tensor & TNumber> ScatterMul<T> create(Scope scope, Operand<T> ref, Operand<U> indices, Operand<T> updates, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ScatterMul", scope.makeOpName("ScatterMul"));
     opBuilder.addInput(ref.asOutput());
     opBuilder.addInput(indices.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterNd.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterNd.java
index 86acb2935db..24098cc16bb 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterNd.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterNd.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Scatter `updates` into a new tensor according to `indices`.
@@ -111,7 +111,7 @@
  * @param <U> data type for {@code output()} output
  */
 @Operator
-public final class ScatterNd<U extends TType> extends RawOp implements Operand<U> {
+public final class ScatterNd<U extends Tensor> extends RawOp implements Operand<U> {
   
   /**
    * Factory method to create a class wrapping a new ScatterNd operation.
@@ -123,7 +123,7 @@ public final class ScatterNd<U extends TType> extends RawOp implements Operand<U
    * @return a new instance of ScatterNd
    */
   @Endpoint(describeByClass = true)
-  public static <U extends TType, T extends TNumber> ScatterNd<U> create(Scope scope, Operand<T> indices, Operand<U> updates, Operand<T> shape) {
+  public static <U extends Tensor, T extends Tensor & TNumber> ScatterNd<U> create(Scope scope, Operand<T> indices, Operand<U> updates, Operand<T> shape) {
     OperationBuilder opBuilder = scope.env().opBuilder("ScatterNd", scope.makeOpName("ScatterNd"));
     opBuilder.addInput(indices.asOutput());
     opBuilder.addInput(updates.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterNdAdd.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterNdAdd.java
index 09192e601e9..5d1ede126d6 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterNdAdd.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterNdAdd.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Applies sparse addition to individual values or slices in a Variable.
@@ -64,7 +64,7 @@
  * @param <T> data type for {@code outputRef()} output
  */
 @Operator
-public final class ScatterNdAdd<T extends TType> extends RawOp implements Operand<T> {
+public final class ScatterNdAdd<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.core.ScatterNdAdd}
@@ -100,7 +100,7 @@ private Options() {
    * @return a new instance of ScatterNdAdd
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> ScatterNdAdd<T> create(Scope scope, Operand<T> ref, Operand<U> indices, Operand<T> updates, Options... options) {
+  public static <T extends Tensor, U extends Tensor & TNumber> ScatterNdAdd<T> create(Scope scope, Operand<T> ref, Operand<U> indices, Operand<T> updates, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ScatterNdAdd", scope.makeOpName("ScatterNdAdd"));
     opBuilder.addInput(ref.asOutput());
     opBuilder.addInput(indices.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterNdNonAliasingAdd.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterNdNonAliasingAdd.java
index 98ce3cceba4..c2066854bee 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterNdNonAliasingAdd.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterNdNonAliasingAdd.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Applies sparse addition to `input` using individual values or slices
@@ -68,7 +68,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator
-public final class ScatterNdNonAliasingAdd<T extends TType> extends RawOp implements Operand<T> {
+public final class ScatterNdNonAliasingAdd<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new ScatterNdNonAliasingAdd operation.
@@ -82,7 +82,7 @@ public final class ScatterNdNonAliasingAdd<T extends TType> extends RawOp implem
    * @return a new instance of ScatterNdNonAliasingAdd
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> ScatterNdNonAliasingAdd<T> create(Scope scope, Operand<T> input, Operand<U> indices, Operand<T> updates) {
+  public static <T extends Tensor, U extends Tensor & TNumber> ScatterNdNonAliasingAdd<T> create(Scope scope, Operand<T> input, Operand<U> indices, Operand<T> updates) {
     OperationBuilder opBuilder = scope.env().opBuilder("ScatterNdNonAliasingAdd", scope.makeOpName("ScatterNdNonAliasingAdd"));
     opBuilder.addInput(input.asOutput());
     opBuilder.addInput(indices.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterNdSub.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterNdSub.java
index 39327cd553d..d6679bc2633 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterNdSub.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterNdSub.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Applies sparse subtraction to individual values or slices in a Variable.
@@ -66,7 +66,7 @@
  * @param <T> data type for {@code outputRef()} output
  */
 @Operator
-public final class ScatterNdSub<T extends TType> extends RawOp implements Operand<T> {
+public final class ScatterNdSub<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.core.ScatterNdSub}
@@ -102,7 +102,7 @@ private Options() {
    * @return a new instance of ScatterNdSub
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> ScatterNdSub<T> create(Scope scope, Operand<T> ref, Operand<U> indices, Operand<T> updates, Options... options) {
+  public static <T extends Tensor, U extends Tensor & TNumber> ScatterNdSub<T> create(Scope scope, Operand<T> ref, Operand<U> indices, Operand<T> updates, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ScatterNdSub", scope.makeOpName("ScatterNdSub"));
     opBuilder.addInput(ref.asOutput());
     opBuilder.addInput(indices.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterNdUpdate.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterNdUpdate.java
index 87e5949c07c..d9399495fbd 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterNdUpdate.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterNdUpdate.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Applies sparse `updates` to individual values or slices within a given
@@ -68,7 +68,7 @@
  * @param <T> data type for {@code outputRef()} output
  */
 @Operator
-public final class ScatterNdUpdate<T extends TType> extends RawOp implements Operand<T> {
+public final class ScatterNdUpdate<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.core.ScatterNdUpdate}
@@ -104,7 +104,7 @@ private Options() {
    * @return a new instance of ScatterNdUpdate
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> ScatterNdUpdate<T> create(Scope scope, Operand<T> ref, Operand<U> indices, Operand<T> updates, Options... options) {
+  public static <T extends Tensor, U extends Tensor & TNumber> ScatterNdUpdate<T> create(Scope scope, Operand<T> ref, Operand<U> indices, Operand<T> updates, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ScatterNdUpdate", scope.makeOpName("ScatterNdUpdate"));
     opBuilder.addInput(ref.asOutput());
     opBuilder.addInput(indices.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterSub.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterSub.java
index bad245da96e..df43d11f910 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterSub.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterSub.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Subtracts sparse updates to a variable reference.
@@ -56,7 +56,7 @@
  * @param <T> data type for {@code outputRef()} output
  */
 @Operator
-public final class ScatterSub<T extends TType> extends RawOp implements Operand<T> {
+public final class ScatterSub<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.core.ScatterSub}
@@ -89,7 +89,7 @@ private Options() {
    * @return a new instance of ScatterSub
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> ScatterSub<T> create(Scope scope, Operand<T> ref, Operand<U> indices, Operand<T> updates, Options... options) {
+  public static <T extends Tensor, U extends Tensor & TNumber> ScatterSub<T> create(Scope scope, Operand<T> ref, Operand<U> indices, Operand<T> updates, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ScatterSub", scope.makeOpName("ScatterSub"));
     opBuilder.addInput(ref.asOutput());
     opBuilder.addInput(indices.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterUpdate.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterUpdate.java
index 9790e2fc934..e71848cd77e 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterUpdate.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterUpdate.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Applies sparse updates to a variable reference.
@@ -60,7 +60,7 @@
  * @param <T> data type for {@code outputRef()} output
  */
 @Operator
-public final class ScatterUpdate<T extends TType> extends RawOp implements Operand<T> {
+public final class ScatterUpdate<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.core.ScatterUpdate}
@@ -93,7 +93,7 @@ private Options() {
    * @return a new instance of ScatterUpdate
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> ScatterUpdate<T> create(Scope scope, Operand<T> ref, Operand<U> indices, Operand<T> updates, Options... options) {
+  public static <T extends Tensor, U extends Tensor & TNumber> ScatterUpdate<T> create(Scope scope, Operand<T> ref, Operand<U> indices, Operand<T> updates, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ScatterUpdate", scope.makeOpName("ScatterUpdate"));
     opBuilder.addInput(ref.asOutput());
     opBuilder.addInput(indices.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Select.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Select.java
index 0aaa338eb1a..8adc04b790f 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Select.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Select.java
@@ -21,18 +21,18 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TBool;
-import org.tensorflow.types.family.TType;
 
 /**
  * @param <T> data type for {@code output()} output
  */
 @Operator
-public final class Select<T extends TType> extends RawOp implements Operand<T> {
+public final class Select<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Select operation.
@@ -44,7 +44,7 @@ public final class Select<T extends TType> extends RawOp implements Operand<T> {
    * @return a new instance of Select
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> Select<T> create(Scope scope, Operand<TBool> condition, Operand<T> t, Operand<T> e) {
+  public static <T extends Tensor> Select<T> create(Scope scope, Operand<TBool> condition, Operand<T> t, Operand<T> e) {
     OperationBuilder opBuilder = scope.env().opBuilder("SelectV2", scope.makeOpName("Select"));
     opBuilder.addInput(condition.asOutput());
     opBuilder.addInput(t.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Send.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Send.java
index d679b85319a..010fc778e9c 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Send.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Send.java
@@ -20,11 +20,11 @@
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Sends the named tensor from send_device to recv_device.
@@ -66,7 +66,7 @@ private Options() {
    * @return a new instance of Send
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> Send create(Scope scope, Operand<T> tensor, String tensorName, String sendDevice, Long sendDeviceIncarnation, String recvDevice, Options... options) {
+  public static <T extends Tensor> Send create(Scope scope, Operand<T> tensor, String tensorName, String sendDevice, Long sendDeviceIncarnation, String recvDevice, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("Send", scope.makeOpName("Send"));
     opBuilder.addInput(tensor.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/SetDiff1d.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/SetDiff1d.java
index c7e458931e5..082db21ea5f 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/SetDiff1d.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/SetDiff1d.java
@@ -22,13 +22,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes the difference between two lists of numbers or strings.
@@ -57,7 +57,7 @@
  * @param <U> data type for {@code idx()} output
  */
 @Operator
-public final class SetDiff1d<T extends TType, U extends TNumber> extends RawOp {
+public final class SetDiff1d<T extends Tensor, U extends Tensor & TNumber> extends RawOp {
   
   /**
    * Factory method to create a class wrapping a new SetDiff1d operation.
@@ -69,7 +69,7 @@ public final class SetDiff1d<T extends TType, U extends TNumber> extends RawOp {
    * @return a new instance of SetDiff1d
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> SetDiff1d<T, U> create(Scope scope, Operand<T> x, Operand<T> y, DataType<U> outIdx) {
+  public static <T extends Tensor, U extends Tensor & TNumber> SetDiff1d<T, U> create(Scope scope, Operand<T> x, Operand<T> y, DataType<U> outIdx) {
     OperationBuilder opBuilder = scope.env().opBuilder("ListDiff", scope.makeOpName("SetDiff1d"));
     opBuilder.addInput(x.asOutput());
     opBuilder.addInput(y.asOutput());
@@ -87,7 +87,7 @@ public static <T extends TType, U extends TNumber> SetDiff1d<T, U> create(Scope
    * @return a new instance of SetDiff1d
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> SetDiff1d<T, TInt32> create(Scope scope, Operand<T> x, Operand<T> y) {
+  public static <T extends Tensor> SetDiff1d<T, TInt32> create(Scope scope, Operand<T> x, Operand<T> y) {
     return create(scope, x, y, TInt32.DTYPE);
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/SetSize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/SetSize.java
index 2fcdf728542..26239a49412 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/SetSize.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/SetSize.java
@@ -21,13 +21,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
 import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
 
 /**
  * Number of unique elements along last dimension of input `set`.
@@ -72,7 +72,7 @@ private Options() {
    * @return a new instance of SetSize
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> SetSize create(Scope scope, Operand<TInt64> setIndices, Operand<T> setValues, Operand<TInt64> setShape, Options... options) {
+  public static <T extends Tensor> SetSize create(Scope scope, Operand<TInt64> setIndices, Operand<T> setValues, Operand<TInt64> setShape, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("SetSize", scope.makeOpName("SetSize"));
     opBuilder.addInput(setIndices.asOutput());
     opBuilder.addInput(setValues.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Shape.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Shape.java
index 5d613401fc5..c007fd32549 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Shape.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Shape.java
@@ -22,13 +22,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Returns the shape of a tensor.
@@ -45,7 +45,7 @@
  * @param <U> data type for {@code output()} output
  */
 @Operator
-public final class Shape<U extends TNumber> extends RawOp implements Operand<U> {
+public final class Shape<U extends Tensor & TNumber> extends RawOp implements Operand<U> {
   
   /**
    * Factory method to create a class wrapping a new Shape operation.
@@ -56,7 +56,7 @@ public final class Shape<U extends TNumber> extends RawOp implements Operand<U>
    * @return a new instance of Shape
    */
   @Endpoint(describeByClass = true)
-  public static <U extends TNumber, T extends TType> Shape<U> create(Scope scope, Operand<T> input, DataType<U> outType) {
+  public static <U extends Tensor & TNumber, T extends Tensor> Shape<U> create(Scope scope, Operand<T> input, DataType<U> outType) {
     OperationBuilder opBuilder = scope.env().opBuilder("Shape", scope.makeOpName("Shape"));
     opBuilder.addInput(input.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
@@ -72,7 +72,7 @@ public static <U extends TNumber, T extends TType> Shape<U> create(Scope scope,
    * @return a new instance of Shape
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> Shape<TInt32> create(Scope scope, Operand<T> input) {
+  public static <T extends Tensor> Shape<TInt32> create(Scope scope, Operand<T> input) {
     return create(scope, input, TInt32.DTYPE);
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ShapeN.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ShapeN.java
index 42340a8a83d..46fdba4223f 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ShapeN.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ShapeN.java
@@ -25,6 +25,7 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.Operands;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
@@ -32,7 +33,6 @@
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Returns shape of tensors.
@@ -42,7 +42,7 @@
  * @param <U> data type for {@code output()} output
  */
 @Operator
-public final class ShapeN<U extends TNumber> extends RawOp implements Iterable<Operand<U>> {
+public final class ShapeN<U extends Tensor & TNumber> extends RawOp implements Iterable<Operand<U>> {
   
   /**
    * Factory method to create a class wrapping a new ShapeN operation.
@@ -53,7 +53,7 @@ public final class ShapeN<U extends TNumber> extends RawOp implements Iterable<O
    * @return a new instance of ShapeN
    */
   @Endpoint(describeByClass = true)
-  public static <U extends TNumber, T extends TType> ShapeN<U> create(Scope scope, Iterable<Operand<T>> input, DataType<U> outType) {
+  public static <U extends Tensor & TNumber, T extends Tensor> ShapeN<U> create(Scope scope, Iterable<Operand<T>> input, DataType<U> outType) {
     OperationBuilder opBuilder = scope.env().opBuilder("ShapeN", scope.makeOpName("ShapeN"));
     opBuilder.addInputList(Operands.asOutputs(input));
     opBuilder = scope.applyControlDependencies(opBuilder);
@@ -69,7 +69,7 @@ public static <U extends TNumber, T extends TType> ShapeN<U> create(Scope scope,
    * @return a new instance of ShapeN
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> ShapeN<TInt32> create(Scope scope, Iterable<Operand<T>> input) {
+  public static <T extends Tensor> ShapeN<TInt32> create(Scope scope, Iterable<Operand<T>> input) {
     return create(scope, input, TInt32.DTYPE);
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Size.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Size.java
index 4aaccae0f30..c03e08df8fa 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Size.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Size.java
@@ -22,13 +22,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Returns the size of a tensor.
@@ -46,7 +46,7 @@
  * @param <U> data type for {@code output()} output
  */
 @Operator
-public final class Size<U extends TNumber> extends RawOp implements Operand<U> {
+public final class Size<U extends Tensor & TNumber> extends RawOp implements Operand<U> {
   
   /**
    * Factory method to create a class wrapping a new Size operation.
@@ -57,7 +57,7 @@ public final class Size<U extends TNumber> extends RawOp implements Operand<U> {
    * @return a new instance of Size
    */
   @Endpoint(describeByClass = true)
-  public static <U extends TNumber, T extends TType> Size<U> create(Scope scope, Operand<T> input, DataType<U> outType) {
+  public static <U extends Tensor & TNumber, T extends Tensor> Size<U> create(Scope scope, Operand<T> input, DataType<U> outType) {
     OperationBuilder opBuilder = scope.env().opBuilder("Size", scope.makeOpName("Size"));
     opBuilder.addInput(input.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
@@ -73,7 +73,7 @@ public static <U extends TNumber, T extends TType> Size<U> create(Scope scope, O
    * @return a new instance of Size
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> Size<TInt32> create(Scope scope, Operand<T> input) {
+  public static <T extends Tensor> Size<TInt32> create(Scope scope, Operand<T> input) {
     return create(scope, input, TInt32.DTYPE);
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Slice.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Slice.java
index 31edde00b81..1d5e2684f74 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Slice.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Slice.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Return a slice from 'input'.
@@ -41,7 +41,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator
-public final class Slice<T extends TType> extends RawOp implements Operand<T> {
+public final class Slice<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Slice operation.
@@ -57,7 +57,7 @@ public final class Slice<T extends TType> extends RawOp implements Operand<T> {
    * @return a new instance of Slice
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> Slice<T> create(Scope scope, Operand<T> input, Operand<U> begin, Operand<U> size) {
+  public static <T extends Tensor, U extends Tensor & TNumber> Slice<T> create(Scope scope, Operand<T> input, Operand<U> begin, Operand<U> size) {
     OperationBuilder opBuilder = scope.env().opBuilder("Slice", scope.makeOpName("Slice"));
     opBuilder.addInput(input.asOutput());
     opBuilder.addInput(begin.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Snapshot.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Snapshot.java
index 7209bccaaee..02328cbdc2d 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Snapshot.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Snapshot.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Returns a copy of the input tensor.
@@ -33,7 +33,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator
-public final class Snapshot<T extends TType> extends RawOp implements Operand<T> {
+public final class Snapshot<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Snapshot operation.
@@ -43,7 +43,7 @@ public final class Snapshot<T extends TType> extends RawOp implements Operand<T>
    * @return a new instance of Snapshot
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> Snapshot<T> create(Scope scope, Operand<T> input) {
+  public static <T extends Tensor> Snapshot<T> create(Scope scope, Operand<T> input) {
     OperationBuilder opBuilder = scope.env().opBuilder("Snapshot", scope.makeOpName("Snapshot"));
     opBuilder.addInput(input.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/SpaceToBatchNd.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/SpaceToBatchNd.java
index 150e18256e5..e1ebd3064a2 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/SpaceToBatchNd.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/SpaceToBatchNd.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * SpaceToBatch for N-D tensors of type T.
@@ -43,7 +43,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator
-public final class SpaceToBatchNd<T extends TType> extends RawOp implements Operand<T> {
+public final class SpaceToBatchNd<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new SpaceToBatchNd operation.
@@ -147,7 +147,7 @@ public final class SpaceToBatchNd<T extends TType> extends RawOp implements Oper
    * @return a new instance of SpaceToBatchNd
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber, V extends TNumber> SpaceToBatchNd<T> create(Scope scope, Operand<T> input, Operand<U> blockShape, Operand<V> paddings) {
+  public static <T extends Tensor, U extends Tensor & TNumber, V extends Tensor & TNumber> SpaceToBatchNd<T> create(Scope scope, Operand<T> input, Operand<U> blockShape, Operand<V> paddings) {
     OperationBuilder opBuilder = scope.env().opBuilder("SpaceToBatchND", scope.makeOpName("SpaceToBatchNd"));
     opBuilder.addInput(input.asOutput());
     opBuilder.addInput(blockShape.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Split.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Split.java
index fffd73f8ef5..97deffaa27d 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Split.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Split.java
@@ -24,12 +24,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
-import org.tensorflow.types.family.TType;
 
 /**
  * Splits a tensor into `num_split` tensors along one dimension.
@@ -37,7 +37,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator
-public final class Split<T extends TType> extends RawOp implements Iterable<Operand<T>> {
+public final class Split<T extends Tensor> extends RawOp implements Iterable<Operand<T>> {
   
   /**
    * Factory method to create a class wrapping a new Split operation.
@@ -51,7 +51,7 @@ public final class Split<T extends TType> extends RawOp implements Iterable<Oper
    * @return a new instance of Split
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> Split<T> create(Scope scope, Operand<TInt32> axis, Operand<T> value, Long numSplit) {
+  public static <T extends Tensor> Split<T> create(Scope scope, Operand<TInt32> axis, Operand<T> value, Long numSplit) {
     OperationBuilder opBuilder = scope.env().opBuilder("Split", scope.makeOpName("Split"));
     opBuilder.addInput(axis.asOutput());
     opBuilder.addInput(value.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/SplitV.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/SplitV.java
index 02b740c4d03..87538a25177 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/SplitV.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/SplitV.java
@@ -24,13 +24,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Splits a tensor into `num_split` tensors along one dimension.
@@ -38,7 +38,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator
-public final class SplitV<T extends TType> extends RawOp implements Iterable<Operand<T>> {
+public final class SplitV<T extends Tensor> extends RawOp implements Iterable<Operand<T>> {
   
   /**
    * Factory method to create a class wrapping a new SplitV operation.
@@ -54,7 +54,7 @@ public final class SplitV<T extends TType> extends RawOp implements Iterable<Ope
    * @return a new instance of SplitV
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> SplitV<T> create(Scope scope, Operand<T> value, Operand<U> sizeSplits, Operand<TInt32> axis, Long numSplit) {
+  public static <T extends Tensor, U extends Tensor & TNumber> SplitV<T> create(Scope scope, Operand<T> value, Operand<U> sizeSplits, Operand<TInt32> axis, Long numSplit) {
     OperationBuilder opBuilder = scope.env().opBuilder("SplitV", scope.makeOpName("SplitV"));
     opBuilder.addInput(value.asOutput());
     opBuilder.addInput(sizeSplits.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Squeeze.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Squeeze.java
index 3f962adc789..a229bc7ab23 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Squeeze.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Squeeze.java
@@ -22,11 +22,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Removes dimensions of size 1 from the shape of a tensor.
@@ -51,7 +51,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator
-public final class Squeeze<T extends TType> extends RawOp implements Operand<T> {
+public final class Squeeze<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.core.Squeeze}
@@ -83,7 +83,7 @@ private Options() {
    * @return a new instance of Squeeze
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> Squeeze<T> create(Scope scope, Operand<T> input, Options... options) {
+  public static <T extends Tensor> Squeeze<T> create(Scope scope, Operand<T> input, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("Squeeze", scope.makeOpName("Squeeze"));
     opBuilder.addInput(input.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Stack.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Stack.java
index f70f50b0d2b..7e22893da34 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Stack.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Stack.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.Operands;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Packs a list of `N` rank-`R` tensors into one rank-`(R+1)` tensor.
@@ -52,7 +52,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator
-public final class Stack<T extends TType> extends RawOp implements Operand<T> {
+public final class Stack<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.core.Stack}
@@ -83,7 +83,7 @@ private Options() {
    * @return a new instance of Stack
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> Stack<T> create(Scope scope, Iterable<Operand<T>> values, Options... options) {
+  public static <T extends Tensor> Stack<T> create(Scope scope, Iterable<Operand<T>> values, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("Pack", scope.makeOpName("Stack"));
     opBuilder.addInputList(Operands.asOutputs(values));
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/StagePeek.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/StagePeek.java
index 2126de722df..dab5018a122 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/StagePeek.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/StagePeek.java
@@ -25,12 +25,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
-import org.tensorflow.types.family.TType;
 
 /**
  * Op peeks at the values at the specified index.  If the
@@ -40,7 +40,7 @@
  * performance.
  */
 @Operator
-public final class StagePeek extends RawOp implements Iterable<Operand<TType>> {
+public final class StagePeek extends RawOp implements Iterable<Operand<Tensor>> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.core.StagePeek}
@@ -162,7 +162,7 @@ public List<Output<?>> values() {
   
   @Override
   @SuppressWarnings({"rawtypes", "unchecked"})
-  public Iterator<Operand<TType>> iterator() {
+  public Iterator<Operand<Tensor>> iterator() {
     return (Iterator) values.iterator();
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/StopGradient.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/StopGradient.java
index b3c7d144ca5..a5852cf0a97 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/StopGradient.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/StopGradient.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Stops gradient computation.
@@ -58,7 +58,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator
-public final class StopGradient<T extends TType> extends RawOp implements Operand<T> {
+public final class StopGradient<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new StopGradient operation.
@@ -68,7 +68,7 @@ public final class StopGradient<T extends TType> extends RawOp implements Operan
    * @return a new instance of StopGradient
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> StopGradient<T> create(Scope scope, Operand<T> input) {
+  public static <T extends Tensor> StopGradient<T> create(Scope scope, Operand<T> input) {
     OperationBuilder opBuilder = scope.env().opBuilder("StopGradient", scope.makeOpName("StopGradient"));
     opBuilder.addInput(input.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/StridedSlice.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/StridedSlice.java
index 8067f96728d..c2f38d29e4f 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/StridedSlice.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/StridedSlice.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Return a strided slice from `input`.
@@ -121,7 +121,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator
-public final class StridedSlice<T extends TType> extends RawOp implements Operand<T> {
+public final class StridedSlice<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.core.StridedSlice}
@@ -214,7 +214,7 @@ private Options() {
    * @return a new instance of StridedSlice
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> StridedSlice<T> create(Scope scope, Operand<T> input, Operand<U> begin, Operand<U> end, Operand<U> strides, Options... options) {
+  public static <T extends Tensor, U extends Tensor & TNumber> StridedSlice<T> create(Scope scope, Operand<T> input, Operand<U> begin, Operand<U> end, Operand<U> strides, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("StridedSlice", scope.makeOpName("StridedSlice"));
     opBuilder.addInput(input.asOutput());
     opBuilder.addInput(begin.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/StridedSliceAssign.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/StridedSliceAssign.java
index b6836c65fca..e7dec58902d 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/StridedSliceAssign.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/StridedSliceAssign.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Assign `value` to the sliced l-value reference of `ref`.
@@ -41,7 +41,7 @@
  * @param <T> data type for {@code outputRef()} output
  */
 @Operator
-public final class StridedSliceAssign<T extends TType> extends RawOp implements Operand<T> {
+public final class StridedSliceAssign<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.core.StridedSliceAssign}
@@ -111,7 +111,7 @@ private Options() {
    * @return a new instance of StridedSliceAssign
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> StridedSliceAssign<T> create(Scope scope, Operand<T> ref, Operand<U> begin, Operand<U> end, Operand<U> strides, Operand<T> value, Options... options) {
+  public static <T extends Tensor, U extends Tensor & TNumber> StridedSliceAssign<T> create(Scope scope, Operand<T> ref, Operand<U> begin, Operand<U> end, Operand<U> strides, Operand<T> value, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("StridedSliceAssign", scope.makeOpName("StridedSliceAssign"));
     opBuilder.addInput(ref.asOutput());
     opBuilder.addInput(begin.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/StridedSliceGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/StridedSliceGrad.java
index 241d6d567e9..9aa3f5b348b 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/StridedSliceGrad.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/StridedSliceGrad.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Returns the gradient of `StridedSlice`.
@@ -43,7 +43,7 @@
  * @param <U> data type for {@code output()} output
  */
 @Operator
-public final class StridedSliceGrad<U extends TType> extends RawOp implements Operand<U> {
+public final class StridedSliceGrad<U extends Tensor> extends RawOp implements Operand<U> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.core.StridedSliceGrad}
@@ -113,7 +113,7 @@ private Options() {
    * @return a new instance of StridedSliceGrad
    */
   @Endpoint(describeByClass = true)
-  public static <U extends TType, T extends TNumber> StridedSliceGrad<U> create(Scope scope, Operand<T> shape, Operand<T> begin, Operand<T> end, Operand<T> strides, Operand<U> dy, Options... options) {
+  public static <U extends Tensor, T extends Tensor & TNumber> StridedSliceGrad<U> create(Scope scope, Operand<T> shape, Operand<T> begin, Operand<T> end, Operand<T> strides, Operand<U> dy, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("StridedSliceGrad", scope.makeOpName("StridedSliceGrad"));
     opBuilder.addInput(shape.asOutput());
     opBuilder.addInput(begin.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Sum.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Sum.java
index c153cdf4760..af88da5baa2 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Sum.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Sum.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes the sum of elements across dimensions of a tensor.
@@ -39,7 +39,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator
-public final class Sum<T extends TType> extends RawOp implements Operand<T> {
+public final class Sum<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.core.Sum}
@@ -71,7 +71,7 @@ private Options() {
    * @return a new instance of Sum
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> Sum<T> create(Scope scope, Operand<T> input, Operand<U> axis, Options... options) {
+  public static <T extends Tensor, U extends Tensor & TNumber> Sum<T> create(Scope scope, Operand<T> input, Operand<U> axis, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("Sum", scope.makeOpName("Sum"));
     opBuilder.addInput(input.asOutput());
     opBuilder.addInput(axis.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/SwitchCond.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/SwitchCond.java
index 191eb6292d2..2b2b67d7f9c 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/SwitchCond.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/SwitchCond.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TBool;
-import org.tensorflow.types.family.TType;
 
 /**
  * Forwards `data` to the output port determined by `pred`.
@@ -39,7 +39,7 @@
  * @param <T> data type for {@code outputFalse()} output
  */
 @Operator
-public final class SwitchCond<T extends TType> extends RawOp {
+public final class SwitchCond<T extends Tensor> extends RawOp {
   
   /**
    * Factory method to create a class wrapping a new SwitchCond operation.
@@ -50,7 +50,7 @@ public final class SwitchCond<T extends TType> extends RawOp {
    * @return a new instance of SwitchCond
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> SwitchCond<T> create(Scope scope, Operand<T> data, Operand<TBool> pred) {
+  public static <T extends Tensor> SwitchCond<T> create(Scope scope, Operand<T> data, Operand<TBool> pred) {
     OperationBuilder opBuilder = scope.env().opBuilder("Switch", scope.makeOpName("SwitchCond"));
     opBuilder.addInput(data.asOutput());
     opBuilder.addInput(pred.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TemporaryVariable.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TemporaryVariable.java
index 43d9247fe21..18c5fc112e3 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TemporaryVariable.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TemporaryVariable.java
@@ -22,12 +22,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Returns a tensor that may be mutated, but only persists within a single step.
@@ -50,7 +50,7 @@
  * @param <T> data type for {@code ref()} output
  */
 @Operator
-public final class TemporaryVariable<T extends TType> extends RawOp implements Operand<T> {
+public final class TemporaryVariable<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.core.TemporaryVariable}
@@ -82,7 +82,7 @@ private Options() {
    * @return a new instance of TemporaryVariable
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> TemporaryVariable<T> create(Scope scope, Shape shape, DataType<T> dtype, Options... options) {
+  public static <T extends Tensor> TemporaryVariable<T> create(Scope scope, Shape shape, DataType<T> dtype, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("TemporaryVariable", scope.makeOpName("TemporaryVariable"));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("shape", shape);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArray.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArray.java
index f34dc414484..e484ce440cb 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArray.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArray.java
@@ -22,6 +22,7 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
@@ -29,7 +30,6 @@
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TFloat32;
 import org.tensorflow.types.TInt32;
-import org.tensorflow.types.family.TType;
 
 /**
  * An array of Tensors of given size.
@@ -116,7 +116,7 @@ private Options() {
    * @return a new instance of TensorArray
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> TensorArray create(Scope scope, Operand<TInt32> size, DataType<T> dtype, Options... options) {
+  public static <T extends Tensor> TensorArray create(Scope scope, Operand<TInt32> size, DataType<T> dtype, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("TensorArrayV3", scope.makeOpName("TensorArray"));
     opBuilder.addInput(size.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayConcat.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayConcat.java
index e25b7a2f958..9fa20bacc2f 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayConcat.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayConcat.java
@@ -22,6 +22,7 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
@@ -29,7 +30,6 @@
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TFloat32;
 import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
 
 /**
  * Concat the elements from the TensorArray into value `value`.
@@ -48,7 +48,7 @@
  * @param <T> data type for {@code value()} output
  */
 @Operator
-public final class TensorArrayConcat<T extends TType> extends RawOp {
+public final class TensorArrayConcat<T extends Tensor> extends RawOp {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.core.TensorArrayConcat}
@@ -83,7 +83,7 @@ private Options() {
    * @return a new instance of TensorArrayConcat
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> TensorArrayConcat<T> create(Scope scope, Operand<?> handle, Operand<TFloat32> flowIn, DataType<T> dtype, Options... options) {
+  public static <T extends Tensor> TensorArrayConcat<T> create(Scope scope, Operand<?> handle, Operand<TFloat32> flowIn, DataType<T> dtype, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("TensorArrayConcatV3", scope.makeOpName("TensorArrayConcat"));
     opBuilder.addInput(handle.asOutput());
     opBuilder.addInput(flowIn.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayGather.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayGather.java
index 13c6dce3122..5f5475866a9 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayGather.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayGather.java
@@ -22,6 +22,7 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
@@ -29,7 +30,6 @@
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TFloat32;
 import org.tensorflow.types.TInt32;
-import org.tensorflow.types.family.TType;
 
 /**
  * Gather specific elements from the TensorArray into output `value`.
@@ -39,7 +39,7 @@
  * @param <T> data type for {@code value()} output
  */
 @Operator
-public final class TensorArrayGather<T extends TType> extends RawOp implements Operand<T> {
+public final class TensorArrayGather<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.core.TensorArrayGather}
@@ -74,7 +74,7 @@ private Options() {
    * @return a new instance of TensorArrayGather
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> TensorArrayGather<T> create(Scope scope, Operand<?> handle, Operand<TInt32> indices, Operand<TFloat32> flowIn, DataType<T> dtype, Options... options) {
+  public static <T extends Tensor> TensorArrayGather<T> create(Scope scope, Operand<?> handle, Operand<TInt32> indices, Operand<TFloat32> flowIn, DataType<T> dtype, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("TensorArrayGatherV3", scope.makeOpName("TensorArrayGather"));
     opBuilder.addInput(handle.asOutput());
     opBuilder.addInput(indices.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayPack.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayPack.java
index 4b4be4ac089..aaccdfdc913 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayPack.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayPack.java
@@ -22,6 +22,7 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
@@ -29,13 +30,12 @@
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TFloat32;
 import org.tensorflow.types.TString;
-import org.tensorflow.types.family.TType;
 
 /**
  * @param <T> data type for {@code value()} output
  */
 @Operator
-public final class TensorArrayPack<T extends TType> extends RawOp implements Operand<T> {
+public final class TensorArrayPack<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.core.TensorArrayPack}
@@ -67,7 +67,7 @@ private Options() {
    * @return a new instance of TensorArrayPack
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> TensorArrayPack<T> create(Scope scope, Operand<TString> handle, Operand<TFloat32> flowIn, DataType<T> dtype, Options... options) {
+  public static <T extends Tensor> TensorArrayPack<T> create(Scope scope, Operand<TString> handle, Operand<TFloat32> flowIn, DataType<T> dtype, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("TensorArrayPack", scope.makeOpName("TensorArrayPack"));
     opBuilder.addInput(handle.asOutput());
     opBuilder.addInput(flowIn.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayRead.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayRead.java
index 55cf48bb020..44216cbab93 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayRead.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayRead.java
@@ -22,13 +22,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TFloat32;
 import org.tensorflow.types.TInt32;
-import org.tensorflow.types.family.TType;
 
 /**
  * Read an element from the TensorArray into output `value`.
@@ -36,7 +36,7 @@
  * @param <T> data type for {@code value()} output
  */
 @Operator
-public final class TensorArrayRead<T extends TType> extends RawOp implements Operand<T> {
+public final class TensorArrayRead<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new TensorArrayRead operation.
@@ -49,7 +49,7 @@ public final class TensorArrayRead<T extends TType> extends RawOp implements Ope
    * @return a new instance of TensorArrayRead
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> TensorArrayRead<T> create(Scope scope, Operand<?> handle, Operand<TInt32> index, Operand<TFloat32> flowIn, DataType<T> dtype) {
+  public static <T extends Tensor> TensorArrayRead<T> create(Scope scope, Operand<?> handle, Operand<TInt32> index, Operand<TFloat32> flowIn, DataType<T> dtype) {
     OperationBuilder opBuilder = scope.env().opBuilder("TensorArrayReadV3", scope.makeOpName("TensorArrayRead"));
     opBuilder.addInput(handle.asOutput());
     opBuilder.addInput(index.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayScatter.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayScatter.java
index 7b3eb238026..b0dac8b4452 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayScatter.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayScatter.java
@@ -21,13 +21,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TFloat32;
 import org.tensorflow.types.TInt32;
-import org.tensorflow.types.family.TType;
 
 /**
  * Scatter the data from the input value into specific TensorArray elements.
@@ -48,7 +48,7 @@ public final class TensorArrayScatter extends RawOp implements Operand<TFloat32>
    * @return a new instance of TensorArrayScatter
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> TensorArrayScatter create(Scope scope, Operand<?> handle, Operand<TInt32> indices, Operand<T> value, Operand<TFloat32> flowIn) {
+  public static <T extends Tensor> TensorArrayScatter create(Scope scope, Operand<?> handle, Operand<TInt32> indices, Operand<T> value, Operand<TFloat32> flowIn) {
     OperationBuilder opBuilder = scope.env().opBuilder("TensorArrayScatterV3", scope.makeOpName("TensorArrayScatter"));
     opBuilder.addInput(handle.asOutput());
     opBuilder.addInput(indices.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArraySplit.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArraySplit.java
index d6aedc34a4a..8b11e653a18 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArraySplit.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArraySplit.java
@@ -21,13 +21,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TFloat32;
 import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
 
 /**
  * Split the data from the input value into TensorArray elements.
@@ -69,7 +69,7 @@ public final class TensorArraySplit extends RawOp implements Operand<TFloat32> {
    * @return a new instance of TensorArraySplit
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> TensorArraySplit create(Scope scope, Operand<?> handle, Operand<T> value, Operand<TInt64> lengths, Operand<TFloat32> flowIn) {
+  public static <T extends Tensor> TensorArraySplit create(Scope scope, Operand<?> handle, Operand<T> value, Operand<TInt64> lengths, Operand<TFloat32> flowIn) {
     OperationBuilder opBuilder = scope.env().opBuilder("TensorArraySplitV3", scope.makeOpName("TensorArraySplit"));
     opBuilder.addInput(handle.asOutput());
     opBuilder.addInput(value.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayUnpack.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayUnpack.java
index 9d2d92c4b85..06509356f54 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayUnpack.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayUnpack.java
@@ -21,13 +21,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TFloat32;
 import org.tensorflow.types.TString;
-import org.tensorflow.types.family.TType;
 
 /**
  */
@@ -44,7 +44,7 @@ public final class TensorArrayUnpack extends RawOp implements Operand<TFloat32>
    * @return a new instance of TensorArrayUnpack
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> TensorArrayUnpack create(Scope scope, Operand<TString> handle, Operand<T> value, Operand<TFloat32> flowIn) {
+  public static <T extends Tensor> TensorArrayUnpack create(Scope scope, Operand<TString> handle, Operand<T> value, Operand<TFloat32> flowIn) {
     OperationBuilder opBuilder = scope.env().opBuilder("TensorArrayUnpack", scope.makeOpName("TensorArrayUnpack"));
     opBuilder.addInput(handle.asOutput());
     opBuilder.addInput(value.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayWrite.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayWrite.java
index a83a07a458b..cacb13634e0 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayWrite.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayWrite.java
@@ -21,13 +21,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TFloat32;
 import org.tensorflow.types.TInt32;
-import org.tensorflow.types.family.TType;
 
 /**
  * Push an element onto the tensor_array.
@@ -46,7 +46,7 @@ public final class TensorArrayWrite extends RawOp implements Operand<TFloat32> {
    * @return a new instance of TensorArrayWrite
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> TensorArrayWrite create(Scope scope, Operand<?> handle, Operand<TInt32> index, Operand<T> value, Operand<TFloat32> flowIn) {
+  public static <T extends Tensor> TensorArrayWrite create(Scope scope, Operand<?> handle, Operand<TInt32> index, Operand<T> value, Operand<TFloat32> flowIn) {
     OperationBuilder opBuilder = scope.env().opBuilder("TensorArrayWriteV3", scope.makeOpName("TensorArrayWrite"));
     opBuilder.addInput(handle.asOutput());
     opBuilder.addInput(index.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorForestTreeResourceHandleOp.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorForestTreeResourceHandleOp.java
index c72860e1ea7..f739dcb573b 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorForestTreeResourceHandleOp.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorForestTreeResourceHandleOp.java
@@ -21,16 +21,16 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Creates a handle to a TensorForestTreeResource
  */
-public final class TensorForestTreeResourceHandleOp extends RawOp implements Operand<TType> {
+public final class TensorForestTreeResourceHandleOp extends RawOp implements Operand<Tensor> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.core.TensorForestTreeResourceHandleOp}
@@ -106,8 +106,8 @@ public Output<?> resource() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) resource;
+  public Output<Tensor> asOutput() {
+    return (Output<Tensor>) resource;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListConcat.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListConcat.java
index 2f1771797e0..398b39a0746 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListConcat.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListConcat.java
@@ -22,13 +22,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Concats all tensors in the list along the 0th dimension.
@@ -49,7 +49,7 @@
  * @param <U> data type for {@code tensor()} output
  */
 @Operator
-public final class TensorListConcat<U extends TType> extends RawOp {
+public final class TensorListConcat<U extends Tensor> extends RawOp {
   
   /**
    * Factory method to create a class wrapping a new TensorListConcat operation.
@@ -62,7 +62,7 @@ public final class TensorListConcat<U extends TType> extends RawOp {
    * @return a new instance of TensorListConcat
    */
   @Endpoint(describeByClass = true)
-  public static <U extends TType, T extends TNumber> TensorListConcat<U> create(Scope scope, Operand<?> inputHandle, Operand<T> elementShape, Operand<TInt64> leadingDims, DataType<U> elementDtype) {
+  public static <U extends Tensor, T extends Tensor & TNumber> TensorListConcat<U> create(Scope scope, Operand<?> inputHandle, Operand<T> elementShape, Operand<TInt64> leadingDims, DataType<U> elementDtype) {
     OperationBuilder opBuilder = scope.env().opBuilder("TensorListConcatV2", scope.makeOpName("TensorListConcat"));
     opBuilder.addInput(inputHandle.asOutput());
     opBuilder.addInput(elementShape.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListConcatLists.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListConcatLists.java
index fdca8e2d6cd..f6347befb41 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListConcatLists.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListConcatLists.java
@@ -22,16 +22,16 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  */
 @Operator
-public final class TensorListConcatLists extends RawOp implements Operand<TType> {
+public final class TensorListConcatLists extends RawOp implements Operand<Tensor> {
   
   /**
    * Factory method to create a class wrapping a new TensorListConcatLists operation.
@@ -43,7 +43,7 @@ public final class TensorListConcatLists extends RawOp implements Operand<TType>
    * @return a new instance of TensorListConcatLists
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> TensorListConcatLists create(Scope scope, Operand<?> inputA, Operand<?> inputB, DataType<T> elementDtype) {
+  public static <T extends Tensor> TensorListConcatLists create(Scope scope, Operand<?> inputA, Operand<?> inputB, DataType<T> elementDtype) {
     OperationBuilder opBuilder = scope.env().opBuilder("TensorListConcatLists", scope.makeOpName("TensorListConcatLists"));
     opBuilder.addInput(inputA.asOutput());
     opBuilder.addInput(inputB.asOutput());
@@ -60,8 +60,8 @@ public Output<?> output() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) output;
+  public Output<Tensor> asOutput() {
+    return (Output<Tensor>) output;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListElementShape.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListElementShape.java
index 2c3e3a5b90c..1efb2ee1ac9 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListElementShape.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListElementShape.java
@@ -22,12 +22,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * The shape of the elements of the given list, as a tensor.
@@ -38,7 +38,7 @@
  * @param <T> data type for {@code elementShape()} output
  */
 @Operator
-public final class TensorListElementShape<T extends TNumber> extends RawOp implements Operand<T> {
+public final class TensorListElementShape<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new TensorListElementShape operation.
@@ -49,7 +49,7 @@ public final class TensorListElementShape<T extends TNumber> extends RawOp imple
    * @return a new instance of TensorListElementShape
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> TensorListElementShape<T> create(Scope scope, Operand<?> inputHandle, DataType<T> shapeType) {
+  public static <T extends Tensor & TNumber> TensorListElementShape<T> create(Scope scope, Operand<?> inputHandle, DataType<T> shapeType) {
     OperationBuilder opBuilder = scope.env().opBuilder("TensorListElementShape", scope.makeOpName("TensorListElementShape"));
     opBuilder.addInput(inputHandle.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListFromTensor.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListFromTensor.java
index 97a45046767..28829431aa8 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListFromTensor.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListFromTensor.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Creates a TensorList which, when stacked, has the value of `tensor`.
@@ -37,7 +37,7 @@
  * output_handle: The list.
  */
 @Operator
-public final class TensorListFromTensor extends RawOp implements Operand<TType> {
+public final class TensorListFromTensor extends RawOp implements Operand<Tensor> {
   
   /**
    * Factory method to create a class wrapping a new TensorListFromTensor operation.
@@ -48,7 +48,7 @@ public final class TensorListFromTensor extends RawOp implements Operand<TType>
    * @return a new instance of TensorListFromTensor
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> TensorListFromTensor create(Scope scope, Operand<T> tensor, Operand<U> elementShape) {
+  public static <T extends Tensor, U extends Tensor & TNumber> TensorListFromTensor create(Scope scope, Operand<T> tensor, Operand<U> elementShape) {
     OperationBuilder opBuilder = scope.env().opBuilder("TensorListFromTensor", scope.makeOpName("TensorListFromTensor"));
     opBuilder.addInput(tensor.asOutput());
     opBuilder.addInput(elementShape.asOutput());
@@ -64,8 +64,8 @@ public Output<?> outputHandle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) outputHandle;
+  public Output<Tensor> asOutput() {
+    return (Output<Tensor>) outputHandle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListGather.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListGather.java
index dbb565a29a6..856600ec5b8 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListGather.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListGather.java
@@ -22,12 +22,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
-import org.tensorflow.types.family.TType;
 
 /**
  * Creates a Tensor by indexing into the TensorList.
@@ -42,7 +42,7 @@
  * @param <T> data type for {@code values()} output
  */
 @Operator
-public final class TensorListGather<T extends TType> extends RawOp implements Operand<T> {
+public final class TensorListGather<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new TensorListGather operation.
@@ -55,7 +55,7 @@ public final class TensorListGather<T extends TType> extends RawOp implements Op
    * @return a new instance of TensorListGather
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> TensorListGather<T> create(Scope scope, Operand<?> inputHandle, Operand<TInt32> indices, Operand<TInt32> elementShape, DataType<T> elementDtype) {
+  public static <T extends Tensor> TensorListGather<T> create(Scope scope, Operand<?> inputHandle, Operand<TInt32> indices, Operand<TInt32> elementShape, DataType<T> elementDtype) {
     OperationBuilder opBuilder = scope.env().opBuilder("TensorListGather", scope.makeOpName("TensorListGather"));
     opBuilder.addInput(inputHandle.asOutput());
     opBuilder.addInput(indices.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListGetItem.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListGetItem.java
index 9af925e1425..07e2c30e56b 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListGetItem.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListGetItem.java
@@ -22,18 +22,18 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
-import org.tensorflow.types.family.TType;
 
 /**
  * @param <T> data type for {@code item()} output
  */
 @Operator
-public final class TensorListGetItem<T extends TType> extends RawOp implements Operand<T> {
+public final class TensorListGetItem<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new TensorListGetItem operation.
@@ -46,7 +46,7 @@ public final class TensorListGetItem<T extends TType> extends RawOp implements O
    * @return a new instance of TensorListGetItem
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> TensorListGetItem<T> create(Scope scope, Operand<?> inputHandle, Operand<TInt32> index, Operand<TInt32> elementShape, DataType<T> elementDtype) {
+  public static <T extends Tensor> TensorListGetItem<T> create(Scope scope, Operand<?> inputHandle, Operand<TInt32> index, Operand<TInt32> elementShape, DataType<T> elementDtype) {
     OperationBuilder opBuilder = scope.env().opBuilder("TensorListGetItem", scope.makeOpName("TensorListGetItem"));
     opBuilder.addInput(inputHandle.asOutput());
     opBuilder.addInput(index.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListPopBack.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListPopBack.java
index 96fd3433e07..428de964aeb 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListPopBack.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListPopBack.java
@@ -22,12 +22,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
-import org.tensorflow.types.family.TType;
 
 /**
  * Returns the last element of the input list as well as a list with all but that element.
@@ -42,7 +42,7 @@
  * @param <T> data type for {@code tensor()} output
  */
 @Operator
-public final class TensorListPopBack<T extends TType> extends RawOp {
+public final class TensorListPopBack<T extends Tensor> extends RawOp {
   
   /**
    * Factory method to create a class wrapping a new TensorListPopBack operation.
@@ -54,7 +54,7 @@ public final class TensorListPopBack<T extends TType> extends RawOp {
    * @return a new instance of TensorListPopBack
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> TensorListPopBack<T> create(Scope scope, Operand<?> inputHandle, Operand<TInt32> elementShape, DataType<T> elementDtype) {
+  public static <T extends Tensor> TensorListPopBack<T> create(Scope scope, Operand<?> inputHandle, Operand<TInt32> elementShape, DataType<T> elementDtype) {
     OperationBuilder opBuilder = scope.env().opBuilder("TensorListPopBack", scope.makeOpName("TensorListPopBack"));
     opBuilder.addInput(inputHandle.asOutput());
     opBuilder.addInput(elementShape.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListPushBack.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListPushBack.java
index 3bde4981422..99049e35893 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListPushBack.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListPushBack.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Returns a list which has the passed-in `Tensor` as last element and the other elements of the given list in `input_handle`.
@@ -37,7 +37,7 @@
  * element_shape: a shape compatible with that of elements in the list.
  */
 @Operator
-public final class TensorListPushBack extends RawOp implements Operand<TType> {
+public final class TensorListPushBack extends RawOp implements Operand<Tensor> {
   
   /**
    * Factory method to create a class wrapping a new TensorListPushBack operation.
@@ -48,7 +48,7 @@ public final class TensorListPushBack extends RawOp implements Operand<TType> {
    * @return a new instance of TensorListPushBack
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> TensorListPushBack create(Scope scope, Operand<?> inputHandle, Operand<T> tensor) {
+  public static <T extends Tensor> TensorListPushBack create(Scope scope, Operand<?> inputHandle, Operand<T> tensor) {
     OperationBuilder opBuilder = scope.env().opBuilder("TensorListPushBack", scope.makeOpName("TensorListPushBack"));
     opBuilder.addInput(inputHandle.asOutput());
     opBuilder.addInput(tensor.asOutput());
@@ -64,8 +64,8 @@ public Output<?> outputHandle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) outputHandle;
+  public Output<Tensor> asOutput() {
+    return (Output<Tensor>) outputHandle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListPushBackBatch.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListPushBackBatch.java
index de772f74891..991c3372083 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListPushBackBatch.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListPushBackBatch.java
@@ -21,16 +21,16 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  */
 @Operator
-public final class TensorListPushBackBatch extends RawOp implements Operand<TType> {
+public final class TensorListPushBackBatch extends RawOp implements Operand<Tensor> {
   
   /**
    * Factory method to create a class wrapping a new TensorListPushBackBatch operation.
@@ -41,7 +41,7 @@ public final class TensorListPushBackBatch extends RawOp implements Operand<TTyp
    * @return a new instance of TensorListPushBackBatch
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> TensorListPushBackBatch create(Scope scope, Operand<?> inputHandles, Operand<T> tensor) {
+  public static <T extends Tensor> TensorListPushBackBatch create(Scope scope, Operand<?> inputHandles, Operand<T> tensor) {
     OperationBuilder opBuilder = scope.env().opBuilder("TensorListPushBackBatch", scope.makeOpName("TensorListPushBackBatch"));
     opBuilder.addInput(inputHandles.asOutput());
     opBuilder.addInput(tensor.asOutput());
@@ -57,8 +57,8 @@ public Output<?> outputHandles() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) outputHandles;
+  public Output<Tensor> asOutput() {
+    return (Output<Tensor>) outputHandles;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListReserve.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListReserve.java
index 3f798e0f998..4b92a52ec2e 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListReserve.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListReserve.java
@@ -22,13 +22,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * List of the given size with empty elements.
@@ -39,7 +39,7 @@
  * element_dtype: the desired type of elements in the list.
  */
 @Operator
-public final class TensorListReserve extends RawOp implements Operand<TType> {
+public final class TensorListReserve extends RawOp implements Operand<Tensor> {
   
   /**
    * Factory method to create a class wrapping a new TensorListReserve operation.
@@ -51,7 +51,7 @@ public final class TensorListReserve extends RawOp implements Operand<TType> {
    * @return a new instance of TensorListReserve
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber, U extends TType> TensorListReserve create(Scope scope, Operand<T> elementShape, Operand<TInt32> numElements, DataType<U> elementDtype) {
+  public static <T extends Tensor & TNumber, U extends Tensor> TensorListReserve create(Scope scope, Operand<T> elementShape, Operand<TInt32> numElements, DataType<U> elementDtype) {
     OperationBuilder opBuilder = scope.env().opBuilder("TensorListReserve", scope.makeOpName("TensorListReserve"));
     opBuilder.addInput(elementShape.asOutput());
     opBuilder.addInput(numElements.asOutput());
@@ -68,8 +68,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) handle;
+  public Output<Tensor> asOutput() {
+    return (Output<Tensor>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListResize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListResize.java
index 4b70c31f277..a4cada56849 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListResize.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListResize.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
-import org.tensorflow.types.family.TType;
 
 /**
  * Resizes the list.
@@ -37,7 +37,7 @@
  * 
  */
 @Operator
-public final class TensorListResize extends RawOp implements Operand<TType> {
+public final class TensorListResize extends RawOp implements Operand<Tensor> {
   
   /**
    * Factory method to create a class wrapping a new TensorListResize operation.
@@ -64,8 +64,8 @@ public Output<?> outputHandle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) outputHandle;
+  public Output<Tensor> asOutput() {
+    return (Output<Tensor>) outputHandle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListScatter.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListScatter.java
index 700e3e715a7..c85c306a4a1 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListScatter.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListScatter.java
@@ -21,13 +21,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Creates a TensorList by indexing into a Tensor.
@@ -45,7 +45,7 @@
  * output_handle: The TensorList.
  */
 @Operator
-public final class TensorListScatter extends RawOp implements Operand<TType> {
+public final class TensorListScatter extends RawOp implements Operand<Tensor> {
   
   /**
    * Factory method to create a class wrapping a new TensorListScatter operation.
@@ -58,7 +58,7 @@ public final class TensorListScatter extends RawOp implements Operand<TType> {
    * @return a new instance of TensorListScatter
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> TensorListScatter create(Scope scope, Operand<T> tensor, Operand<TInt32> indices, Operand<U> elementShape, Operand<TInt32> numElements) {
+  public static <T extends Tensor, U extends Tensor & TNumber> TensorListScatter create(Scope scope, Operand<T> tensor, Operand<TInt32> indices, Operand<U> elementShape, Operand<TInt32> numElements) {
     OperationBuilder opBuilder = scope.env().opBuilder("TensorListScatterV2", scope.makeOpName("TensorListScatter"));
     opBuilder.addInput(tensor.asOutput());
     opBuilder.addInput(indices.asOutput());
@@ -76,8 +76,8 @@ public Output<?> outputHandle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) outputHandle;
+  public Output<Tensor> asOutput() {
+    return (Output<Tensor>) outputHandle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListScatterIntoExistingList.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListScatterIntoExistingList.java
index 4b91b1bc0b3..d85976c616b 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListScatterIntoExistingList.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListScatterIntoExistingList.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
-import org.tensorflow.types.family.TType;
 
 /**
  * Scatters tensor at indices in an input list.
@@ -40,7 +40,7 @@
  * output_handle: The TensorList.
  */
 @Operator
-public final class TensorListScatterIntoExistingList extends RawOp implements Operand<TType> {
+public final class TensorListScatterIntoExistingList extends RawOp implements Operand<Tensor> {
   
   /**
    * Factory method to create a class wrapping a new TensorListScatterIntoExistingList operation.
@@ -52,7 +52,7 @@ public final class TensorListScatterIntoExistingList extends RawOp implements Op
    * @return a new instance of TensorListScatterIntoExistingList
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> TensorListScatterIntoExistingList create(Scope scope, Operand<?> inputHandle, Operand<T> tensor, Operand<TInt32> indices) {
+  public static <T extends Tensor> TensorListScatterIntoExistingList create(Scope scope, Operand<?> inputHandle, Operand<T> tensor, Operand<TInt32> indices) {
     OperationBuilder opBuilder = scope.env().opBuilder("TensorListScatterIntoExistingList", scope.makeOpName("TensorListScatterIntoExistingList"));
     opBuilder.addInput(inputHandle.asOutput());
     opBuilder.addInput(tensor.asOutput());
@@ -69,8 +69,8 @@ public Output<?> outputHandle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) outputHandle;
+  public Output<Tensor> asOutput() {
+    return (Output<Tensor>) outputHandle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListSetItem.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListSetItem.java
index 358397a927e..801ad711a56 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListSetItem.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListSetItem.java
@@ -21,17 +21,17 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
-import org.tensorflow.types.family.TType;
 
 /**
  */
 @Operator
-public final class TensorListSetItem extends RawOp implements Operand<TType> {
+public final class TensorListSetItem extends RawOp implements Operand<Tensor> {
   
   /**
    * Factory method to create a class wrapping a new TensorListSetItem operation.
@@ -43,7 +43,7 @@ public final class TensorListSetItem extends RawOp implements Operand<TType> {
    * @return a new instance of TensorListSetItem
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> TensorListSetItem create(Scope scope, Operand<?> inputHandle, Operand<TInt32> index, Operand<T> item) {
+  public static <T extends Tensor> TensorListSetItem create(Scope scope, Operand<?> inputHandle, Operand<TInt32> index, Operand<T> item) {
     OperationBuilder opBuilder = scope.env().opBuilder("TensorListSetItem", scope.makeOpName("TensorListSetItem"));
     opBuilder.addInput(inputHandle.asOutput());
     opBuilder.addInput(index.asOutput());
@@ -60,8 +60,8 @@ public Output<?> outputHandle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) outputHandle;
+  public Output<Tensor> asOutput() {
+    return (Output<Tensor>) outputHandle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListSplit.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListSplit.java
index e93635d7713..f689050481c 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListSplit.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListSplit.java
@@ -21,13 +21,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Splits a tensor into a list.
@@ -41,7 +41,7 @@
  * output_handle: The list.
  */
 @Operator
-public final class TensorListSplit extends RawOp implements Operand<TType> {
+public final class TensorListSplit extends RawOp implements Operand<Tensor> {
   
   /**
    * Factory method to create a class wrapping a new TensorListSplit operation.
@@ -53,7 +53,7 @@ public final class TensorListSplit extends RawOp implements Operand<TType> {
    * @return a new instance of TensorListSplit
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> TensorListSplit create(Scope scope, Operand<T> tensor, Operand<U> elementShape, Operand<TInt64> lengths) {
+  public static <T extends Tensor, U extends Tensor & TNumber> TensorListSplit create(Scope scope, Operand<T> tensor, Operand<U> elementShape, Operand<TInt64> lengths) {
     OperationBuilder opBuilder = scope.env().opBuilder("TensorListSplit", scope.makeOpName("TensorListSplit"));
     opBuilder.addInput(tensor.asOutput());
     opBuilder.addInput(elementShape.asOutput());
@@ -70,8 +70,8 @@ public Output<?> outputHandle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) outputHandle;
+  public Output<Tensor> asOutput() {
+    return (Output<Tensor>) outputHandle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListStack.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListStack.java
index c0ecf388b3d..aeeed49c5bb 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListStack.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListStack.java
@@ -22,12 +22,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
-import org.tensorflow.types.family.TType;
 
 /**
  * Stacks all tensors in the list.
@@ -42,7 +42,7 @@
  * @param <T> data type for {@code tensor()} output
  */
 @Operator
-public final class TensorListStack<T extends TType> extends RawOp implements Operand<T> {
+public final class TensorListStack<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.core.TensorListStack}
@@ -74,7 +74,7 @@ private Options() {
    * @return a new instance of TensorListStack
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> TensorListStack<T> create(Scope scope, Operand<?> inputHandle, Operand<TInt32> elementShape, DataType<T> elementDtype, Options... options) {
+  public static <T extends Tensor> TensorListStack<T> create(Scope scope, Operand<?> inputHandle, Operand<TInt32> elementShape, DataType<T> elementDtype, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("TensorListStack", scope.makeOpName("TensorListStack"));
     opBuilder.addInput(inputHandle.asOutput());
     opBuilder.addInput(elementShape.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorScatterNdAdd.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorScatterNdAdd.java
index b5d32055bd9..9983210cb93 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorScatterNdAdd.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorScatterNdAdd.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Adds sparse `updates` to an existing tensor according to `indices`.
@@ -94,7 +94,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator
-public final class TensorScatterNdAdd<T extends TType> extends RawOp implements Operand<T> {
+public final class TensorScatterNdAdd<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new TensorScatterNdAdd operation.
@@ -106,7 +106,7 @@ public final class TensorScatterNdAdd<T extends TType> extends RawOp implements
    * @return a new instance of TensorScatterNdAdd
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> TensorScatterNdAdd<T> create(Scope scope, Operand<T> tensor, Operand<U> indices, Operand<T> updates) {
+  public static <T extends Tensor, U extends Tensor & TNumber> TensorScatterNdAdd<T> create(Scope scope, Operand<T> tensor, Operand<U> indices, Operand<T> updates) {
     OperationBuilder opBuilder = scope.env().opBuilder("TensorScatterAdd", scope.makeOpName("TensorScatterNdAdd"));
     opBuilder.addInput(tensor.asOutput());
     opBuilder.addInput(indices.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorScatterNdSub.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorScatterNdSub.java
index 14737a1fd95..c05e46b4050 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorScatterNdSub.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorScatterNdSub.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Subtracts sparse `updates` from an existing tensor according to `indices`.
@@ -93,7 +93,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator
-public final class TensorScatterNdSub<T extends TType> extends RawOp implements Operand<T> {
+public final class TensorScatterNdSub<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new TensorScatterNdSub operation.
@@ -105,7 +105,7 @@ public final class TensorScatterNdSub<T extends TType> extends RawOp implements
    * @return a new instance of TensorScatterNdSub
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> TensorScatterNdSub<T> create(Scope scope, Operand<T> tensor, Operand<U> indices, Operand<T> updates) {
+  public static <T extends Tensor, U extends Tensor & TNumber> TensorScatterNdSub<T> create(Scope scope, Operand<T> tensor, Operand<U> indices, Operand<T> updates) {
     OperationBuilder opBuilder = scope.env().opBuilder("TensorScatterSub", scope.makeOpName("TensorScatterNdSub"));
     opBuilder.addInput(tensor.asOutput());
     opBuilder.addInput(indices.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorScatterNdUpdate.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorScatterNdUpdate.java
index 7dff39f733f..c0147d8a796 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorScatterNdUpdate.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorScatterNdUpdate.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Scatter `updates` into an existing tensor according to `indices`.
@@ -108,7 +108,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator
-public final class TensorScatterNdUpdate<T extends TType> extends RawOp implements Operand<T> {
+public final class TensorScatterNdUpdate<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new TensorScatterNdUpdate operation.
@@ -120,7 +120,7 @@ public final class TensorScatterNdUpdate<T extends TType> extends RawOp implemen
    * @return a new instance of TensorScatterNdUpdate
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> TensorScatterNdUpdate<T> create(Scope scope, Operand<T> tensor, Operand<U> indices, Operand<T> updates) {
+  public static <T extends Tensor, U extends Tensor & TNumber> TensorScatterNdUpdate<T> create(Scope scope, Operand<T> tensor, Operand<U> indices, Operand<T> updates) {
     OperationBuilder opBuilder = scope.env().opBuilder("TensorScatterUpdate", scope.makeOpName("TensorScatterNdUpdate"));
     opBuilder.addInput(tensor.asOutput());
     opBuilder.addInput(indices.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorStridedSliceUpdate.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorStridedSliceUpdate.java
index c8c5e36e3c6..932013a0a05 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorStridedSliceUpdate.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorStridedSliceUpdate.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Assign `value` to the sliced l-value reference of `input`.
@@ -41,7 +41,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator
-public final class TensorStridedSliceUpdate<T extends TType> extends RawOp implements Operand<T> {
+public final class TensorStridedSliceUpdate<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.core.TensorStridedSliceUpdate}
@@ -111,7 +111,7 @@ private Options() {
    * @return a new instance of TensorStridedSliceUpdate
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> TensorStridedSliceUpdate<T> create(Scope scope, Operand<T> input, Operand<U> begin, Operand<U> end, Operand<U> strides, Operand<T> value, Options... options) {
+  public static <T extends Tensor, U extends Tensor & TNumber> TensorStridedSliceUpdate<T> create(Scope scope, Operand<T> input, Operand<U> begin, Operand<U> end, Operand<U> strides, Operand<T> value, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("TensorStridedSliceUpdate", scope.makeOpName("TensorStridedSliceUpdate"));
     opBuilder.addInput(input.asOutput());
     opBuilder.addInput(begin.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Tile.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Tile.java
index d28588cada5..b6943bbd8b3 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Tile.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Tile.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Constructs a tensor by tiling a given tensor.
@@ -61,7 +61,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator
-public final class Tile<T extends TType> extends RawOp implements Operand<T> {
+public final class Tile<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Tile operation.
@@ -72,7 +72,7 @@ public final class Tile<T extends TType> extends RawOp implements Operand<T> {
    * @return a new instance of Tile
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> Tile<T> create(Scope scope, Operand<T> input, Operand<U> multiples) {
+  public static <T extends Tensor, U extends Tensor & TNumber> Tile<T> create(Scope scope, Operand<T> input, Operand<U> multiples) {
     OperationBuilder opBuilder = scope.env().opBuilder("Tile", scope.makeOpName("Tile"));
     opBuilder.addInput(input.asOutput());
     opBuilder.addInput(multiples.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Unbatch.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Unbatch.java
index 84d2063ae8c..25fcf101630 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Unbatch.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Unbatch.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
 
 /**
  * Reverses the operation of Batch for a single output Tensor.
@@ -53,7 +53,7 @@
  * @param <T> data type for {@code unbatchedTensor()} output
  */
 @Operator
-public final class Unbatch<T extends TType> extends RawOp implements Operand<T> {
+public final class Unbatch<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.core.Unbatch}
@@ -95,7 +95,7 @@ private Options() {
    * @return a new instance of Unbatch
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> Unbatch<T> create(Scope scope, Operand<T> batchedTensor, Operand<TInt64> batchIndex, Operand<TInt64> id, Long timeoutMicros, Options... options) {
+  public static <T extends Tensor> Unbatch<T> create(Scope scope, Operand<T> batchedTensor, Operand<TInt64> batchIndex, Operand<TInt64> id, Long timeoutMicros, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("Unbatch", scope.makeOpName("Unbatch"));
     opBuilder.addInput(batchedTensor.asOutput());
     opBuilder.addInput(batchIndex.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/UnbatchGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/UnbatchGrad.java
index 10062f204fc..332b3b6967c 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/UnbatchGrad.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/UnbatchGrad.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
 
 /**
  * Gradient of Unbatch.
@@ -49,7 +49,7 @@
  * @param <T> data type for {@code batchedGrad()} output
  */
 @Operator
-public final class UnbatchGrad<T extends TType> extends RawOp implements Operand<T> {
+public final class UnbatchGrad<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.core.UnbatchGrad}
@@ -91,7 +91,7 @@ private Options() {
    * @return a new instance of UnbatchGrad
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> UnbatchGrad<T> create(Scope scope, Operand<T> originalInput, Operand<TInt64> batchIndex, Operand<T> grad, Operand<TInt64> id, Options... options) {
+  public static <T extends Tensor> UnbatchGrad<T> create(Scope scope, Operand<T> originalInput, Operand<TInt64> batchIndex, Operand<T> grad, Operand<TInt64> id, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("UnbatchGrad", scope.makeOpName("UnbatchGrad"));
     opBuilder.addInput(originalInput.asOutput());
     opBuilder.addInput(batchIndex.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Unique.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Unique.java
index f3b8b04aead..bb5f21ccacb 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Unique.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Unique.java
@@ -22,13 +22,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Finds unique elements along an axis of a tensor.
@@ -77,7 +77,7 @@
  * @param <V> data type for {@code idx()} output
  */
 @Operator
-public final class Unique<T extends TType, V extends TNumber> extends RawOp {
+public final class Unique<T extends Tensor, V extends Tensor & TNumber> extends RawOp {
   
   /**
    * Factory method to create a class wrapping a new Unique operation.
@@ -90,7 +90,7 @@ public final class Unique<T extends TType, V extends TNumber> extends RawOp {
    * @return a new instance of Unique
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType, V extends TNumber, U extends TNumber> Unique<T, V> create(Scope scope, Operand<T> x, Operand<U> axis, DataType<V> outIdx) {
+  public static <T extends Tensor, V extends Tensor & TNumber, U extends Tensor & TNumber> Unique<T, V> create(Scope scope, Operand<T> x, Operand<U> axis, DataType<V> outIdx) {
     OperationBuilder opBuilder = scope.env().opBuilder("UniqueV2", scope.makeOpName("Unique"));
     opBuilder.addInput(x.asOutput());
     opBuilder.addInput(axis.asOutput());
@@ -109,7 +109,7 @@ public static <T extends TType, V extends TNumber, U extends TNumber> Unique<T,
    * @return a new instance of Unique
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> Unique<T, TInt32> create(Scope scope, Operand<T> x, Operand<U> axis) {
+  public static <T extends Tensor, U extends Tensor & TNumber> Unique<T, TInt32> create(Scope scope, Operand<T> x, Operand<U> axis) {
     return create(scope, x, axis, TInt32.DTYPE);
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/UniqueWithCounts.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/UniqueWithCounts.java
index 732d4432333..bfdc83e0edf 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/UniqueWithCounts.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/UniqueWithCounts.java
@@ -22,13 +22,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Finds unique elements along an axis of a tensor.
@@ -81,7 +81,7 @@
  * @param <V> data type for {@code idx()} output
  */
 @Operator
-public final class UniqueWithCounts<T extends TType, V extends TNumber> extends RawOp {
+public final class UniqueWithCounts<T extends Tensor, V extends Tensor & TNumber> extends RawOp {
   
   /**
    * Factory method to create a class wrapping a new UniqueWithCounts operation.
@@ -94,7 +94,7 @@ public final class UniqueWithCounts<T extends TType, V extends TNumber> extends
    * @return a new instance of UniqueWithCounts
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType, V extends TNumber, U extends TNumber> UniqueWithCounts<T, V> create(Scope scope, Operand<T> x, Operand<U> axis, DataType<V> outIdx) {
+  public static <T extends Tensor, V extends Tensor & TNumber, U extends Tensor & TNumber> UniqueWithCounts<T, V> create(Scope scope, Operand<T> x, Operand<U> axis, DataType<V> outIdx) {
     OperationBuilder opBuilder = scope.env().opBuilder("UniqueWithCountsV2", scope.makeOpName("UniqueWithCounts"));
     opBuilder.addInput(x.asOutput());
     opBuilder.addInput(axis.asOutput());
@@ -113,7 +113,7 @@ public static <T extends TType, V extends TNumber, U extends TNumber> UniqueWith
    * @return a new instance of UniqueWithCounts
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> UniqueWithCounts<T, TInt32> create(Scope scope, Operand<T> x, Operand<U> axis) {
+  public static <T extends Tensor, U extends Tensor & TNumber> UniqueWithCounts<T, TInt32> create(Scope scope, Operand<T> x, Operand<U> axis) {
     return create(scope, x, axis, TInt32.DTYPE);
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/UnravelIndex.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/UnravelIndex.java
index ecc073e37a4..1f3c5c5c42d 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/UnravelIndex.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/UnravelIndex.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Converts an array of flat indices into a tuple of coordinate arrays.
@@ -53,7 +53,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator
-public final class UnravelIndex<T extends TNumber> extends RawOp implements Operand<T> {
+public final class UnravelIndex<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new UnravelIndex operation.
@@ -66,7 +66,7 @@ public final class UnravelIndex<T extends TNumber> extends RawOp implements Oper
    * @return a new instance of UnravelIndex
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> UnravelIndex<T> create(Scope scope, Operand<T> indices, Operand<T> dims) {
+  public static <T extends Tensor & TNumber> UnravelIndex<T> create(Scope scope, Operand<T> indices, Operand<T> dims) {
     OperationBuilder opBuilder = scope.env().opBuilder("UnravelIndex", scope.makeOpName("UnravelIndex"));
     opBuilder.addInput(indices.asOutput());
     opBuilder.addInput(dims.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Unstack.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Unstack.java
index a48f57289ae..4a1f2757be3 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Unstack.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Unstack.java
@@ -24,11 +24,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Unpacks a given dimension of a rank-`R` tensor into `num` rank-`(R-1)` tensors.
@@ -49,7 +49,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator
-public final class Unstack<T extends TType> extends RawOp implements Iterable<Operand<T>> {
+public final class Unstack<T extends Tensor> extends RawOp implements Iterable<Operand<T>> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.core.Unstack}
@@ -81,7 +81,7 @@ private Options() {
    * @return a new instance of Unstack
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> Unstack<T> create(Scope scope, Operand<T> value, Long num, Options... options) {
+  public static <T extends Tensor> Unstack<T> create(Scope scope, Operand<T> value, Long num, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("Unpack", scope.makeOpName("Unstack"));
     opBuilder.addInput(value.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Unstage.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Unstage.java
index 1878ce6fb88..4f683ed6775 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Unstage.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Unstage.java
@@ -25,11 +25,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Op is similar to a lightweight Dequeue.
@@ -38,7 +38,7 @@
  * capabilities and options.  This Op is optimized for performance.
  */
 @Operator
-public final class Unstage extends RawOp implements Iterable<Operand<TType>> {
+public final class Unstage extends RawOp implements Iterable<Operand<Tensor>> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.core.Unstage}
@@ -158,7 +158,7 @@ public List<Output<?>> values() {
   
   @Override
   @SuppressWarnings({"rawtypes", "unchecked"})
-  public Iterator<Operand<TType>> iterator() {
+  public Iterator<Operand<Tensor>> iterator() {
     return (Iterator) values.iterator();
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/UpperBound.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/UpperBound.java
index c46fbcba4de..d0338f5b1d9 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/UpperBound.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/UpperBound.java
@@ -22,13 +22,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Applies upper_bound(sorted_search_values, values) along each row.
@@ -53,7 +53,7 @@
  * 
  * @param <U> data type for {@code output()} output
  */
-public final class UpperBound<U extends TNumber> extends RawOp implements Operand<U> {
+public final class UpperBound<U extends Tensor & TNumber> extends RawOp implements Operand<U> {
   
   /**
    * Factory method to create a class wrapping a new UpperBound operation.
@@ -66,7 +66,7 @@ public final class UpperBound<U extends TNumber> extends RawOp implements Operan
    * @return a new instance of UpperBound
    */
   @Endpoint(describeByClass = true)
-  public static <U extends TNumber, T extends TType> UpperBound<U> create(Scope scope, Operand<T> sortedInputs, Operand<T> values, DataType<U> outType) {
+  public static <U extends Tensor & TNumber, T extends Tensor> UpperBound<U> create(Scope scope, Operand<T> sortedInputs, Operand<T> values, DataType<U> outType) {
     OperationBuilder opBuilder = scope.env().opBuilder("UpperBound", scope.makeOpName("UpperBound"));
     opBuilder.addInput(sortedInputs.asOutput());
     opBuilder.addInput(values.asOutput());
@@ -85,7 +85,7 @@ public static <U extends TNumber, T extends TType> UpperBound<U> create(Scope sc
    * @return a new instance of UpperBound
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> UpperBound<TInt32> create(Scope scope, Operand<T> sortedInputs, Operand<T> values) {
+  public static <T extends Tensor> UpperBound<TInt32> create(Scope scope, Operand<T> sortedInputs, Operand<T> values) {
     return create(scope, sortedInputs, values, TInt32.DTYPE);
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/VarHandleOp.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/VarHandleOp.java
index 92618c33d37..2cbd51dda7f 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/VarHandleOp.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/VarHandleOp.java
@@ -23,18 +23,18 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Creates a handle to a Variable resource.
  */
 @Operator
-public final class VarHandleOp extends RawOp implements Operand<TType> {
+public final class VarHandleOp extends RawOp implements Operand<Tensor> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.core.VarHandleOp}
@@ -85,7 +85,7 @@ private Options() {
    * @return a new instance of VarHandleOp
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> VarHandleOp create(Scope scope, DataType<T> dtype, Shape shape, Options... options) {
+  public static <T extends Tensor> VarHandleOp create(Scope scope, DataType<T> dtype, Shape shape, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("VarHandleOp", scope.makeOpName("VarHandleOp"));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("dtype", dtype);
@@ -140,8 +140,8 @@ public Output<?> resource() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) resource;
+  public Output<Tensor> asOutput() {
+    return (Output<Tensor>) resource;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Variable.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Variable.java
index 7353dfee688..94c577c329d 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Variable.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Variable.java
@@ -22,12 +22,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Holds state in the form of a tensor that persists across steps.
@@ -39,7 +39,7 @@
  * @param <T> data type for {@code ref()} output
  */
 @Operator
-public final class Variable<T extends TType> extends RawOp implements Operand<T> {
+public final class Variable<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.core.Variable}
@@ -81,7 +81,7 @@ private Options() {
    * @return a new instance of Variable
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> Variable<T> create(Scope scope, Shape shape, DataType<T> dtype, Options... options) {
+  public static <T extends Tensor> Variable<T> create(Scope scope, Shape shape, DataType<T> dtype, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("VariableV2", scope.makeOpName("Variable"));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("shape", shape);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/VariableShape.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/VariableShape.java
index 6573cb8de11..94ee6717f3d 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/VariableShape.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/VariableShape.java
@@ -22,13 +22,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Returns the shape of the variable pointed to by `resource`.
@@ -45,7 +45,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator
-public final class VariableShape<T extends TNumber> extends RawOp implements Operand<T> {
+public final class VariableShape<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new VariableShape operation.
@@ -56,7 +56,7 @@ public final class VariableShape<T extends TNumber> extends RawOp implements Ope
    * @return a new instance of VariableShape
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> VariableShape<T> create(Scope scope, Operand<?> input, DataType<T> outType) {
+  public static <T extends Tensor & TNumber> VariableShape<T> create(Scope scope, Operand<?> input, DataType<T> outType) {
     OperationBuilder opBuilder = scope.env().opBuilder("VariableShape", scope.makeOpName("VariableShape"));
     opBuilder.addInput(input.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Where.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Where.java
index a1af7f9282a..8419104bfbf 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Where.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Where.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
 
 /**
  * Returns locations of nonzero / true values in a tensor.
@@ -102,7 +102,7 @@ public final class Where extends RawOp implements Operand<TInt64> {
    * @return a new instance of Where
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> Where create(Scope scope, Operand<T> condition) {
+  public static <T extends Tensor> Where create(Scope scope, Operand<T> condition) {
     OperationBuilder opBuilder = scope.env().opBuilder("Where", scope.makeOpName("Where"));
     opBuilder.addInput(condition.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ZerosLike.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ZerosLike.java
index 5b9352ef6bd..818cde42d9c 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ZerosLike.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ZerosLike.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Returns a tensor of zeros with the same shape and type as x.
@@ -33,7 +33,7 @@
  * @param <T> data type for {@code y()} output
  */
 @Operator
-public final class ZerosLike<T extends TType> extends RawOp implements Operand<T> {
+public final class ZerosLike<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new ZerosLike operation.
@@ -43,7 +43,7 @@ public final class ZerosLike<T extends TType> extends RawOp implements Operand<T
    * @return a new instance of ZerosLike
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> ZerosLike<T> create(Scope scope, Operand<T> x) {
+  public static <T extends Tensor> ZerosLike<T> create(Scope scope, Operand<T> x) {
     OperationBuilder opBuilder = scope.env().opBuilder("ZerosLike", scope.makeOpName("ZerosLike"));
     opBuilder.addInput(x.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/AssertNextDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/AssertNextDataset.java
index c2825e29b93..5ffd7510dbc 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/AssertNextDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/AssertNextDataset.java
@@ -23,13 +23,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TString;
-import org.tensorflow.types.family.TType;
 
 /**
  * A transformation that asserts which transformations happen next.
@@ -43,7 +43,7 @@
  * means that the check happens <i>after</i> any static optimizations are applied
  * to the dataset graph.
  */
-public final class AssertNextDataset extends RawOp implements Operand<TType> {
+public final class AssertNextDataset extends RawOp implements Operand<Tensor> {
   
   /**
    * Factory method to create a class wrapping a new AssertNextDataset operation.
@@ -84,8 +84,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) handle;
+  public Output<Tensor> asOutput() {
+    return (Output<Tensor>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/AutoShardDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/AutoShardDataset.java
index aca90031261..54ccd69f19c 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/AutoShardDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/AutoShardDataset.java
@@ -23,13 +23,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
 
 /**
  * Creates a dataset that shards the input dataset.
@@ -42,7 +42,7 @@
  * This dataset will throw a NotFound error if we cannot shard the dataset
  * automatically.
  */
-public final class AutoShardDataset extends RawOp implements Operand<TType> {
+public final class AutoShardDataset extends RawOp implements Operand<Tensor> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.data.AutoShardDataset}
@@ -117,8 +117,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) handle;
+  public Output<Tensor> asOutput() {
+    return (Output<Tensor>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/BatchDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/BatchDataset.java
index b0fd6ef0c0c..3663c9d7bc4 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/BatchDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/BatchDataset.java
@@ -23,6 +23,7 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
@@ -30,13 +31,12 @@
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TBool;
 import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
 
 /**
  * Creates a dataset that batches `batch_size` elements from `input_dataset`.
  */
 @Operator(group = "data")
-public final class BatchDataset extends RawOp implements Operand<TType> {
+public final class BatchDataset extends RawOp implements Operand<Tensor> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.data.BatchDataset}
@@ -112,8 +112,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) handle;
+  public Output<Tensor> asOutput() {
+    return (Output<Tensor>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/BytesProducedStatsDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/BytesProducedStatsDataset.java
index a3b95d92909..9597c4063c2 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/BytesProducedStatsDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/BytesProducedStatsDataset.java
@@ -23,18 +23,18 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TString;
-import org.tensorflow.types.family.TType;
 
 /**
  * Records the bytes size of each element of `input_dataset` in a StatsAggregator.
  */
-public final class BytesProducedStatsDataset extends RawOp implements Operand<TType> {
+public final class BytesProducedStatsDataset extends RawOp implements Operand<Tensor> {
   
   /**
    * Factory method to create a class wrapping a new BytesProducedStatsDataset operation.
@@ -73,8 +73,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) handle;
+  public Output<Tensor> asOutput() {
+    return (Output<Tensor>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/CSVDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/CSVDataset.java
index f6c8f72232d..2781151f259 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/CSVDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/CSVDataset.java
@@ -22,6 +22,7 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.Operands;
 import org.tensorflow.op.RawOp;
@@ -31,12 +32,11 @@
 import org.tensorflow.types.TBool;
 import org.tensorflow.types.TInt64;
 import org.tensorflow.types.TString;
-import org.tensorflow.types.family.TType;
 
 /**
  */
 @Operator(group = "data")
-public final class CSVDataset extends RawOp implements Operand<TType> {
+public final class CSVDataset extends RawOp implements Operand<Tensor> {
   
   /**
    * Factory method to create a class wrapping a new CSVDataset operation.
@@ -83,8 +83,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) handle;
+  public Output<Tensor> asOutput() {
+    return (Output<Tensor>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/CacheDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/CacheDataset.java
index 07384fb6b3e..b15005af198 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/CacheDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/CacheDataset.java
@@ -23,13 +23,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TString;
-import org.tensorflow.types.family.TType;
 
 /**
  * Creates a dataset that caches elements from `input_dataset`.
@@ -39,7 +39,7 @@
  * (e.g. cannot be opened, contains tensors of the wrong shape / size), an error
  * will the returned when used.
  */
-public final class CacheDataset extends RawOp implements Operand<TType> {
+public final class CacheDataset extends RawOp implements Operand<Tensor> {
   
   /**
    * Factory method to create a class wrapping a new CacheDataset operation.
@@ -79,8 +79,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) handle;
+  public Output<Tensor> asOutput() {
+    return (Output<Tensor>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/CacheDatasetV2.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/CacheDatasetV2.java
index ba370108e04..5b111851102 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/CacheDatasetV2.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/CacheDatasetV2.java
@@ -23,17 +23,17 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TString;
-import org.tensorflow.types.family.TType;
 
 /**
  */
-public final class CacheDatasetV2 extends RawOp implements Operand<TType> {
+public final class CacheDatasetV2 extends RawOp implements Operand<Tensor> {
   
   /**
    * Factory method to create a class wrapping a new CacheDatasetV2 operation.
@@ -74,8 +74,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) handle;
+  public Output<Tensor> asOutput() {
+    return (Output<Tensor>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ChooseFastestDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ChooseFastestDataset.java
index a8d9ce445fe..e7726e4e5a9 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ChooseFastestDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ChooseFastestDataset.java
@@ -23,17 +23,17 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.Operands;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  */
-public final class ChooseFastestDataset extends RawOp implements Operand<TType> {
+public final class ChooseFastestDataset extends RawOp implements Operand<Tensor> {
   
   /**
    * Factory method to create a class wrapping a new ChooseFastestDataset operation.
@@ -72,8 +72,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) handle;
+  public Output<Tensor> asOutput() {
+    return (Output<Tensor>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ConcatenateDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ConcatenateDataset.java
index 19cf0b4a706..031296a8468 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ConcatenateDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ConcatenateDataset.java
@@ -23,18 +23,18 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Creates a dataset that concatenates `input_dataset` with `another_dataset`.
  */
 @Operator(group = "data")
-public final class ConcatenateDataset extends RawOp implements Operand<TType> {
+public final class ConcatenateDataset extends RawOp implements Operand<Tensor> {
   
   /**
    * Factory method to create a class wrapping a new ConcatenateDataset operation.
@@ -73,8 +73,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) handle;
+  public Output<Tensor> asOutput() {
+    return (Output<Tensor>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DatasetFromGraph.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DatasetFromGraph.java
index 6223baa7251..7d205ad6bf6 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DatasetFromGraph.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DatasetFromGraph.java
@@ -21,19 +21,19 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TString;
-import org.tensorflow.types.family.TType;
 
 /**
  * Creates a dataset from the given `graph_def`.
  * <p>
  * Creates a dataset from the provided `graph_def`.
  */
-public final class DatasetFromGraph extends RawOp implements Operand<TType> {
+public final class DatasetFromGraph extends RawOp implements Operand<Tensor> {
   
   /**
    * Factory method to create a class wrapping a new DatasetFromGraph operation.
@@ -59,8 +59,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) handle;
+  public Output<Tensor> asOutput() {
+    return (Output<Tensor>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DatasetToSingleElement.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DatasetToSingleElement.java
index 91fd5032439..ad7531dbfb2 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DatasetToSingleElement.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DatasetToSingleElement.java
@@ -25,17 +25,17 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Outputs the single element from the given dataset.
  */
-public final class DatasetToSingleElement extends RawOp implements Iterable<Operand<TType>> {
+public final class DatasetToSingleElement extends RawOp implements Iterable<Operand<Tensor>> {
   
   /**
    * Factory method to create a class wrapping a new DatasetToSingleElement operation.
@@ -73,7 +73,7 @@ public List<Output<?>> components() {
   
   @Override
   @SuppressWarnings({"rawtypes", "unchecked"})
-  public Iterator<Operand<TType>> iterator() {
+  public Iterator<Operand<Tensor>> iterator() {
     return (Iterator) components.iterator();
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DenseToSparseBatchDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DenseToSparseBatchDataset.java
index 32bd135c325..ccc03fe1983 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DenseToSparseBatchDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DenseToSparseBatchDataset.java
@@ -23,18 +23,18 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
 
 /**
  * Creates a dataset that batches input elements into a SparseTensor.
  */
-public final class DenseToSparseBatchDataset extends RawOp implements Operand<TType> {
+public final class DenseToSparseBatchDataset extends RawOp implements Operand<Tensor> {
   
   /**
    * Factory method to create a class wrapping a new DenseToSparseBatchDataset operation.
@@ -78,8 +78,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) handle;
+  public Output<Tensor> asOutput() {
+    return (Output<Tensor>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DirectedInterleaveDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DirectedInterleaveDataset.java
index 6883e2218d9..4218e1919f5 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DirectedInterleaveDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DirectedInterleaveDataset.java
@@ -23,18 +23,18 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.Operands;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * A substitute for `InterleaveDataset` on a fixed list of `N` datasets.
  */
-public final class DirectedInterleaveDataset extends RawOp implements Operand<TType> {
+public final class DirectedInterleaveDataset extends RawOp implements Operand<Tensor> {
   
   /**
    * Factory method to create a class wrapping a new DirectedInterleaveDataset operation.
@@ -75,8 +75,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) handle;
+  public Output<Tensor> asOutput() {
+    return (Output<Tensor>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/FilterByLastComponentDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/FilterByLastComponentDataset.java
index d6c2b576d7e..5e193d710a3 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/FilterByLastComponentDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/FilterByLastComponentDataset.java
@@ -23,17 +23,17 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Creates a dataset containing elements of first component of `input_dataset` having true in the last component.
  */
-public final class FilterByLastComponentDataset extends RawOp implements Operand<TType> {
+public final class FilterByLastComponentDataset extends RawOp implements Operand<Tensor> {
   
   /**
    * Factory method to create a class wrapping a new FilterByLastComponentDataset operation.
@@ -70,8 +70,8 @@ public Output<?> output() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) output;
+  public Output<Tensor> asOutput() {
+    return (Output<Tensor>) output;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/FixedLengthRecordDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/FixedLengthRecordDataset.java
index d1bfa77177b..105e1b6d3e4 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/FixedLengthRecordDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/FixedLengthRecordDataset.java
@@ -21,17 +21,17 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
 import org.tensorflow.types.TString;
-import org.tensorflow.types.family.TType;
 
 /**
  */
-public final class FixedLengthRecordDataset extends RawOp implements Operand<TType> {
+public final class FixedLengthRecordDataset extends RawOp implements Operand<Tensor> {
   
   /**
    * Factory method to create a class wrapping a new FixedLengthRecordDataset operation.
@@ -66,8 +66,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) handle;
+  public Output<Tensor> asOutput() {
+    return (Output<Tensor>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/IgnoreErrorsDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/IgnoreErrorsDataset.java
index 7ffca90d150..e6c6079b1a6 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/IgnoreErrorsDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/IgnoreErrorsDataset.java
@@ -23,17 +23,17 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Creates a dataset that contains the elements of `input_dataset` ignoring errors.
  */
-public final class IgnoreErrorsDataset extends RawOp implements Operand<TType> {
+public final class IgnoreErrorsDataset extends RawOp implements Operand<Tensor> {
   
   /**
    * Factory method to create a class wrapping a new IgnoreErrorsDataset operation.
@@ -70,8 +70,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) handle;
+  public Output<Tensor> asOutput() {
+    return (Output<Tensor>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/Iterator.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/Iterator.java
index 45030001714..2136bfd13e0 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/Iterator.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/Iterator.java
@@ -23,17 +23,17 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  */
 @Operator(group = "data")
-public final class Iterator extends RawOp implements Operand<TType> {
+public final class Iterator extends RawOp implements Operand<Tensor> {
   
   /**
    * Factory method to create a class wrapping a new Iterator operation.
@@ -72,8 +72,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) handle;
+  public Output<Tensor> asOutput() {
+    return (Output<Tensor>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/IteratorFromStringHandle.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/IteratorFromStringHandle.java
index 8264419635a..33b1143fa9e 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/IteratorFromStringHandle.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/IteratorFromStringHandle.java
@@ -23,17 +23,17 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TString;
-import org.tensorflow.types.family.TType;
 
 /**
  */
-public final class IteratorFromStringHandle extends RawOp implements Operand<TType> {
+public final class IteratorFromStringHandle extends RawOp implements Operand<Tensor> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.data.IteratorFromStringHandle}
@@ -102,8 +102,8 @@ public Output<?> resourceHandle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) resourceHandle;
+  public Output<Tensor> asOutput() {
+    return (Output<Tensor>) resourceHandle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/IteratorGetNext.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/IteratorGetNext.java
index b7be406e7bd..56efe443f96 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/IteratorGetNext.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/IteratorGetNext.java
@@ -25,18 +25,18 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Gets the next output from the given iterator .
  */
 @Operator(group = "data")
-public final class IteratorGetNext extends RawOp implements Iterable<Operand<TType>> {
+public final class IteratorGetNext extends RawOp implements Iterable<Operand<Tensor>> {
   
   /**
    * Factory method to create a class wrapping a new IteratorGetNext operation.
@@ -73,7 +73,7 @@ public List<Output<?>> components() {
   
   @Override
   @SuppressWarnings({"rawtypes", "unchecked"})
-  public Iterator<Operand<TType>> iterator() {
+  public Iterator<Operand<Tensor>> iterator() {
     return (Iterator) components.iterator();
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/IteratorGetNextAsOptional.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/IteratorGetNextAsOptional.java
index 682b7ba6f35..bd4deb65680 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/IteratorGetNextAsOptional.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/IteratorGetNextAsOptional.java
@@ -23,18 +23,18 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Gets the next output from the given iterator as an Optional variant.
  */
 @Operator(group = "data")
-public final class IteratorGetNextAsOptional extends RawOp implements Operand<TType> {
+public final class IteratorGetNextAsOptional extends RawOp implements Operand<Tensor> {
   
   /**
    * Factory method to create a class wrapping a new IteratorGetNextAsOptional operation.
@@ -71,8 +71,8 @@ public Output<?> optional() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) optional;
+  public Output<Tensor> asOutput() {
+    return (Output<Tensor>) optional;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/IteratorGetNextSync.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/IteratorGetNextSync.java
index 3cb4f072307..e718c779928 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/IteratorGetNextSync.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/IteratorGetNextSync.java
@@ -25,12 +25,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Gets the next output from the given iterator.
@@ -41,7 +41,7 @@
  * operations (e.g. in eager mode).
  */
 @Operator(group = "data")
-public final class IteratorGetNextSync extends RawOp implements Iterable<Operand<TType>> {
+public final class IteratorGetNextSync extends RawOp implements Iterable<Operand<Tensor>> {
   
   /**
    * Factory method to create a class wrapping a new IteratorGetNextSync operation.
@@ -78,7 +78,7 @@ public List<Output<?>> components() {
   
   @Override
   @SuppressWarnings({"rawtypes", "unchecked"})
-  public Iterator<Operand<TType>> iterator() {
+  public Iterator<Operand<Tensor>> iterator() {
     return (Iterator) components.iterator();
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/LMDBDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/LMDBDataset.java
index d0057d60323..fc17d2406ef 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/LMDBDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/LMDBDataset.java
@@ -23,13 +23,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TString;
-import org.tensorflow.types.family.TType;
 
 /**
  * Creates a dataset that emits the key-value pairs in one or more LMDB files.
@@ -45,7 +45,7 @@
  * LMDB uses different file formats on big- and little-endian machines.
  * `data.LMDBDataset` can only read files in the format of the host machine.
  */
-public final class LMDBDataset extends RawOp implements Operand<TType> {
+public final class LMDBDataset extends RawOp implements Operand<Tensor> {
   
   /**
    * Factory method to create a class wrapping a new LMDBDataset operation.
@@ -83,8 +83,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) handle;
+  public Output<Tensor> asOutput() {
+    return (Output<Tensor>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/LatencyStatsDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/LatencyStatsDataset.java
index 731b70867a7..0ed5f19a738 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/LatencyStatsDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/LatencyStatsDataset.java
@@ -23,18 +23,18 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TString;
-import org.tensorflow.types.family.TType;
 
 /**
  * Records the latency of producing `input_dataset` elements in a StatsAggregator.
  */
-public final class LatencyStatsDataset extends RawOp implements Operand<TType> {
+public final class LatencyStatsDataset extends RawOp implements Operand<Tensor> {
   
   /**
    * Factory method to create a class wrapping a new LatencyStatsDataset operation.
@@ -73,8 +73,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) handle;
+  public Output<Tensor> asOutput() {
+    return (Output<Tensor>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/LeakyReluGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/LeakyReluGrad.java
index 3fb21bd4a6e..ef7f5d3365f 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/LeakyReluGrad.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/LeakyReluGrad.java
@@ -21,19 +21,19 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes rectified linear gradients for a LeakyRelu operation.
  * 
  * @param <T> data type for {@code backprops()} output
  */
-public final class LeakyReluGrad<T extends TNumber> extends RawOp implements Operand<T> {
+public final class LeakyReluGrad<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.data.LeakyReluGrad}
@@ -65,7 +65,7 @@ private Options() {
    * @return a new instance of LeakyReluGrad
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> LeakyReluGrad<T> create(Scope scope, Operand<T> gradients, Operand<T> features, Options... options) {
+  public static <T extends Tensor & TNumber> LeakyReluGrad<T> create(Scope scope, Operand<T> gradients, Operand<T> features, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("LeakyReluGrad", scope.makeOpName("LeakyReluGrad"));
     opBuilder.addInput(gradients.asOutput());
     opBuilder.addInput(features.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MatchingFilesDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MatchingFilesDataset.java
index 2c798a8d807..fc88bc9b785 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MatchingFilesDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MatchingFilesDataset.java
@@ -21,16 +21,16 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TString;
-import org.tensorflow.types.family.TType;
 
 /**
  */
-public final class MatchingFilesDataset extends RawOp implements Operand<TType> {
+public final class MatchingFilesDataset extends RawOp implements Operand<Tensor> {
   
   /**
    * Factory method to create a class wrapping a new MatchingFilesDataset operation.
@@ -55,8 +55,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) handle;
+  public Output<Tensor> asOutput() {
+    return (Output<Tensor>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MaxIntraOpParallelismDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MaxIntraOpParallelismDataset.java
index d7c67924b55..423c401478a 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MaxIntraOpParallelismDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MaxIntraOpParallelismDataset.java
@@ -23,18 +23,18 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
 
 /**
  * Creates a dataset that overrides the maximum intra-op parallelism.
  */
-public final class MaxIntraOpParallelismDataset extends RawOp implements Operand<TType> {
+public final class MaxIntraOpParallelismDataset extends RawOp implements Operand<Tensor> {
   
   /**
    * Factory method to create a class wrapping a new MaxIntraOpParallelismDataset operation.
@@ -73,8 +73,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) handle;
+  public Output<Tensor> asOutput() {
+    return (Output<Tensor>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ModelDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ModelDataset.java
index 53b6f10204d..6f74e7e300c 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ModelDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ModelDataset.java
@@ -23,19 +23,19 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Identity transformation that models performance.
  * <p>
  * Identity transformation that models performance.
  */
-public final class ModelDataset extends RawOp implements Operand<TType> {
+public final class ModelDataset extends RawOp implements Operand<Tensor> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.data.ModelDataset}
@@ -125,8 +125,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) handle;
+  public Output<Tensor> asOutput() {
+    return (Output<Tensor>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MultiDeviceIterator.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MultiDeviceIterator.java
index a3dfeeb5665..e8d7a99e312 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MultiDeviceIterator.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MultiDeviceIterator.java
@@ -23,17 +23,17 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Creates a MultiDeviceIterator resource.
  */
-public final class MultiDeviceIterator extends RawOp implements Operand<TType> {
+public final class MultiDeviceIterator extends RawOp implements Operand<Tensor> {
   
   /**
    * Factory method to create a class wrapping a new MultiDeviceIterator operation.
@@ -81,8 +81,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) handle;
+  public Output<Tensor> asOutput() {
+    return (Output<Tensor>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MultiDeviceIteratorFromStringHandle.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MultiDeviceIteratorFromStringHandle.java
index a9b0b8ec07e..d3846fe1d7d 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MultiDeviceIteratorFromStringHandle.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MultiDeviceIteratorFromStringHandle.java
@@ -23,18 +23,18 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TString;
-import org.tensorflow.types.family.TType;
 
 /**
  * Generates a MultiDeviceIterator resource from its provided string handle.
  */
-public final class MultiDeviceIteratorFromStringHandle extends RawOp implements Operand<TType> {
+public final class MultiDeviceIteratorFromStringHandle extends RawOp implements Operand<Tensor> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.data.MultiDeviceIteratorFromStringHandle}
@@ -104,8 +104,8 @@ public Output<?> multiDeviceIterator() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) multiDeviceIterator;
+  public Output<Tensor> asOutput() {
+    return (Output<Tensor>) multiDeviceIterator;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MultiDeviceIteratorGetNextFromShard.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MultiDeviceIteratorGetNextFromShard.java
index 4781bc2d56e..c11cbdee0c1 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MultiDeviceIteratorGetNextFromShard.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MultiDeviceIteratorGetNextFromShard.java
@@ -25,6 +25,7 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
@@ -32,12 +33,11 @@
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
 import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
 
 /**
  * Gets next element for the provided shard number.
  */
-public final class MultiDeviceIteratorGetNextFromShard extends RawOp implements Iterable<Operand<TType>> {
+public final class MultiDeviceIteratorGetNextFromShard extends RawOp implements Iterable<Operand<Tensor>> {
   
   /**
    * Factory method to create a class wrapping a new MultiDeviceIteratorGetNextFromShard operation.
@@ -79,7 +79,7 @@ public List<Output<?>> components() {
   
   @Override
   @SuppressWarnings({"rawtypes", "unchecked"})
-  public Iterator<Operand<TType>> iterator() {
+  public Iterator<Operand<Tensor>> iterator() {
     return (Iterator) components.iterator();
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/NonSerializableDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/NonSerializableDataset.java
index 34043a14f9d..0f521de748e 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/NonSerializableDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/NonSerializableDataset.java
@@ -23,16 +23,16 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  */
-public final class NonSerializableDataset extends RawOp implements Operand<TType> {
+public final class NonSerializableDataset extends RawOp implements Operand<Tensor> {
   
   /**
    * Factory method to create a class wrapping a new NonSerializableDataset operation.
@@ -69,8 +69,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) handle;
+  public Output<Tensor> asOutput() {
+    return (Output<Tensor>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/OptimizeDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/OptimizeDataset.java
index 73e27b3ffc9..4a3b5902752 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/OptimizeDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/OptimizeDataset.java
@@ -23,20 +23,20 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TString;
-import org.tensorflow.types.family.TType;
 
 /**
  * Creates a dataset by applying optimizations to `input_dataset`.
  * <p>
  * Creates a dataset by applying optimizations to `input_dataset`.
  */
-public final class OptimizeDataset extends RawOp implements Operand<TType> {
+public final class OptimizeDataset extends RawOp implements Operand<Tensor> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.data.OptimizeDataset}
@@ -113,8 +113,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) handle;
+  public Output<Tensor> asOutput() {
+    return (Output<Tensor>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/OptionalFromValue.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/OptionalFromValue.java
index a8fc1d4eaea..5f69afca477 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/OptionalFromValue.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/OptionalFromValue.java
@@ -21,18 +21,18 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.Operands;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Constructs an Optional variant from a tuple of tensors.
  */
 @Operator(group = "data")
-public final class OptionalFromValue extends RawOp implements Operand<TType> {
+public final class OptionalFromValue extends RawOp implements Operand<Tensor> {
   
   /**
    * Factory method to create a class wrapping a new OptionalFromValue operation.
@@ -57,8 +57,8 @@ public Output<?> optional() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) optional;
+  public Output<Tensor> asOutput() {
+    return (Output<Tensor>) optional;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/OptionalGetValue.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/OptionalGetValue.java
index 8d3be67da03..af5e5ea0a70 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/OptionalGetValue.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/OptionalGetValue.java
@@ -25,18 +25,18 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Returns the value stored in an Optional variant or raises an error if none exists.
  */
 @Operator(group = "data")
-public final class OptionalGetValue extends RawOp implements Iterable<Operand<TType>> {
+public final class OptionalGetValue extends RawOp implements Iterable<Operand<Tensor>> {
   
   /**
    * Factory method to create a class wrapping a new OptionalGetValue operation.
@@ -73,7 +73,7 @@ public List<Output<?>> components() {
   
   @Override
   @SuppressWarnings({"rawtypes", "unchecked"})
-  public Iterator<Operand<TType>> iterator() {
+  public Iterator<Operand<Tensor>> iterator() {
     return (Iterator) components.iterator();
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/OptionalNone.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/OptionalNone.java
index fa3deabf28f..5652d9c0daf 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/OptionalNone.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/OptionalNone.java
@@ -21,17 +21,17 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Creates an Optional variant with no value.
  */
 @Operator(group = "data")
-public final class OptionalNone extends RawOp implements Operand<TType> {
+public final class OptionalNone extends RawOp implements Operand<Tensor> {
   
   /**
    * Factory method to create a class wrapping a new OptionalNone operation.
@@ -54,8 +54,8 @@ public Output<?> optional() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) optional;
+  public Output<Tensor> asOutput() {
+    return (Output<Tensor>) optional;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/PaddedBatchDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/PaddedBatchDataset.java
index 0685e5e9e74..4a7e5f06f6a 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/PaddedBatchDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/PaddedBatchDataset.java
@@ -22,6 +22,7 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.Operands;
 import org.tensorflow.op.RawOp;
@@ -30,12 +31,11 @@
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TBool;
 import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
 
 /**
  * Creates a dataset that batches and pads `batch_size` elements from the input.
  */
-public final class PaddedBatchDataset extends RawOp implements Operand<TType> {
+public final class PaddedBatchDataset extends RawOp implements Operand<Tensor> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.data.PaddedBatchDataset}
@@ -114,8 +114,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) handle;
+  public Output<Tensor> asOutput() {
+    return (Output<Tensor>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/PrefetchDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/PrefetchDataset.java
index 78995ad6d20..b0418e2722e 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/PrefetchDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/PrefetchDataset.java
@@ -23,18 +23,18 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
 
 /**
  * Creates a dataset that asynchronously prefetches elements from `input_dataset`.
  */
-public final class PrefetchDataset extends RawOp implements Operand<TType> {
+public final class PrefetchDataset extends RawOp implements Operand<Tensor> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.data.PrefetchDataset}
@@ -127,8 +127,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) handle;
+  public Output<Tensor> asOutput() {
+    return (Output<Tensor>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/PrivateThreadPoolDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/PrivateThreadPoolDataset.java
index b06618ce07a..5bc9fc4447e 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/PrivateThreadPoolDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/PrivateThreadPoolDataset.java
@@ -23,18 +23,18 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
 
 /**
  * Creates a dataset that uses a custom thread pool to compute `input_dataset`.
  */
-public final class PrivateThreadPoolDataset extends RawOp implements Operand<TType> {
+public final class PrivateThreadPoolDataset extends RawOp implements Operand<Tensor> {
   
   /**
    * Factory method to create a class wrapping a new PrivateThreadPoolDataset operation.
@@ -73,8 +73,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) handle;
+  public Output<Tensor> asOutput() {
+    return (Output<Tensor>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/RandomDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/RandomDataset.java
index 2b944e789dc..b70f81436d0 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/RandomDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/RandomDataset.java
@@ -23,13 +23,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
 
 /**
  * Creates a Dataset that returns pseudorandom numbers.
@@ -45,7 +45,7 @@
  * performed is determined by the `experimental_optimization.hoist_random_uniform`
  * option of `tf.data.Options`.
  */
-public final class RandomDataset extends RawOp implements Operand<TType> {
+public final class RandomDataset extends RawOp implements Operand<Tensor> {
   
   /**
    * Factory method to create a class wrapping a new RandomDataset operation.
@@ -86,8 +86,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) handle;
+  public Output<Tensor> asOutput() {
+    return (Output<Tensor>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/RangeDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/RangeDataset.java
index bc83590e03e..6db86ed9aaa 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/RangeDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/RangeDataset.java
@@ -23,19 +23,19 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
 
 /**
  * Creates a dataset with a range of values. Corresponds to python's xrange.
  */
 @Operator(group = "data")
-public final class RangeDataset extends RawOp implements Operand<TType> {
+public final class RangeDataset extends RawOp implements Operand<Tensor> {
   
   /**
    * Factory method to create a class wrapping a new RangeDataset operation.
@@ -76,8 +76,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) handle;
+  public Output<Tensor> asOutput() {
+    return (Output<Tensor>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/RebatchDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/RebatchDataset.java
index b5a21c7b6b8..a7ea93cc7a0 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/RebatchDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/RebatchDataset.java
@@ -23,13 +23,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
 
 /**
  * Creates a dataset that changes the batch size.
@@ -37,7 +37,7 @@
  * Creates a dataset that changes the batch size of the dataset to current batch
  * size // num_workers.
  */
-public final class RebatchDataset extends RawOp implements Operand<TType> {
+public final class RebatchDataset extends RawOp implements Operand<Tensor> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.data.RebatchDataset}
@@ -112,8 +112,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) handle;
+  public Output<Tensor> asOutput() {
+    return (Output<Tensor>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/RepeatDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/RepeatDataset.java
index b832025853c..33c501e2670 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/RepeatDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/RepeatDataset.java
@@ -23,19 +23,19 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
 
 /**
  * Creates a dataset that emits the outputs of `input_dataset` `count` times.
  */
 @Operator(group = "data")
-public final class RepeatDataset extends RawOp implements Operand<TType> {
+public final class RepeatDataset extends RawOp implements Operand<Tensor> {
   
   /**
    * Factory method to create a class wrapping a new RepeatDataset operation.
@@ -75,8 +75,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) handle;
+  public Output<Tensor> asOutput() {
+    return (Output<Tensor>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SamplingDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SamplingDataset.java
index 7876fe27587..59b597eb32f 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SamplingDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SamplingDataset.java
@@ -23,6 +23,7 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
@@ -30,7 +31,6 @@
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TFloat32;
 import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
 
 /**
  * Creates a dataset that takes a Bernoulli sample of the contents of another dataset.
@@ -41,7 +41,7 @@
  * `experimental_optimization.filter_with_random_uniform_fusion` option of
  * `tf.data.Options`.
  */
-public final class SamplingDataset extends RawOp implements Operand<TType> {
+public final class SamplingDataset extends RawOp implements Operand<Tensor> {
   
   /**
    * Factory method to create a class wrapping a new SamplingDataset operation.
@@ -85,8 +85,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) handle;
+  public Output<Tensor> asOutput() {
+    return (Output<Tensor>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SerializeIterator.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SerializeIterator.java
index f051994e6f5..1bc43bad658 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SerializeIterator.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SerializeIterator.java
@@ -21,17 +21,17 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Converts the given `resource_handle` representing an iterator to a variant tensor.
  */
 @Operator(group = "data")
-public final class SerializeIterator extends RawOp implements Operand<TType> {
+public final class SerializeIterator extends RawOp implements Operand<Tensor> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.data.SerializeIterator}
@@ -92,8 +92,8 @@ public Output<?> serialized() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) serialized;
+  public Output<Tensor> asOutput() {
+    return (Output<Tensor>) serialized;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SetStatsAggregatorDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SetStatsAggregatorDataset.java
index 5181c0b1f71..961776fb2b1 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SetStatsAggregatorDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SetStatsAggregatorDataset.java
@@ -23,17 +23,17 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TString;
-import org.tensorflow.types.family.TType;
 
 /**
  */
-public final class SetStatsAggregatorDataset extends RawOp implements Operand<TType> {
+public final class SetStatsAggregatorDataset extends RawOp implements Operand<Tensor> {
   
   /**
    * Factory method to create a class wrapping a new SetStatsAggregatorDataset operation.
@@ -76,8 +76,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) handle;
+  public Output<Tensor> asOutput() {
+    return (Output<Tensor>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ShardDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ShardDataset.java
index 62e18f3fadf..4c53f39cbc2 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ShardDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ShardDataset.java
@@ -23,18 +23,18 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
 
 /**
  * Creates a `Dataset` that includes only 1/`num_shards` of this dataset.
  */
-public final class ShardDataset extends RawOp implements Operand<TType> {
+public final class ShardDataset extends RawOp implements Operand<Tensor> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.data.ShardDataset}
@@ -109,8 +109,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) handle;
+  public Output<Tensor> asOutput() {
+    return (Output<Tensor>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ShuffleAndRepeatDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ShuffleAndRepeatDataset.java
index c5703e8e85c..9f2e233bbf8 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ShuffleAndRepeatDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ShuffleAndRepeatDataset.java
@@ -23,17 +23,17 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
 
 /**
  */
-public final class ShuffleAndRepeatDataset extends RawOp implements Operand<TType> {
+public final class ShuffleAndRepeatDataset extends RawOp implements Operand<Tensor> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.data.ShuffleAndRepeatDataset}
@@ -114,8 +114,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) handle;
+  public Output<Tensor> asOutput() {
+    return (Output<Tensor>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ShuffleDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ShuffleDataset.java
index 3dd522e319c..01ce2654dc6 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ShuffleDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ShuffleDataset.java
@@ -23,17 +23,17 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
 
 /**
  */
-public final class ShuffleDataset extends RawOp implements Operand<TType> {
+public final class ShuffleDataset extends RawOp implements Operand<Tensor> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.data.ShuffleDataset}
@@ -112,8 +112,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) handle;
+  public Output<Tensor> asOutput() {
+    return (Output<Tensor>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SkipDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SkipDataset.java
index 4e409b5a4ea..1ab6e64f1d2 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SkipDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SkipDataset.java
@@ -23,19 +23,19 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
 
 /**
  * Creates a dataset that skips `count` elements from the `input_dataset`.
  */
 @Operator(group = "data")
-public final class SkipDataset extends RawOp implements Operand<TType> {
+public final class SkipDataset extends RawOp implements Operand<Tensor> {
   
   /**
    * Factory method to create a class wrapping a new SkipDataset operation.
@@ -75,8 +75,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) handle;
+  public Output<Tensor> asOutput() {
+    return (Output<Tensor>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SleepDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SleepDataset.java
index 67605099106..614548839da 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SleepDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SleepDataset.java
@@ -23,17 +23,17 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
 
 /**
  */
-public final class SleepDataset extends RawOp implements Operand<TType> {
+public final class SleepDataset extends RawOp implements Operand<Tensor> {
   
   /**
    * Factory method to create a class wrapping a new SleepDataset operation.
@@ -72,8 +72,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) handle;
+  public Output<Tensor> asOutput() {
+    return (Output<Tensor>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SlidingWindowDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SlidingWindowDataset.java
index 89f440efd77..0dd1856499d 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SlidingWindowDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SlidingWindowDataset.java
@@ -23,18 +23,18 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
 
 /**
  * Creates a dataset that passes a sliding window over `input_dataset`.
  */
-public final class SlidingWindowDataset extends RawOp implements Operand<TType> {
+public final class SlidingWindowDataset extends RawOp implements Operand<Tensor> {
   
   /**
    * Factory method to create a class wrapping a new SlidingWindowDataset operation.
@@ -80,8 +80,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) handle;
+  public Output<Tensor> asOutput() {
+    return (Output<Tensor>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SnapshotDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SnapshotDataset.java
index f6bc66e8297..a5626396908 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SnapshotDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SnapshotDataset.java
@@ -23,13 +23,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TString;
-import org.tensorflow.types.family.TType;
 
 /**
  * Creates a dataset that will write to / read from a snapshot.
@@ -39,7 +39,7 @@
  * If not, it will run the preprocessing pipeline as usual, and write out a
  * snapshot of the data processed for future use.
  */
-public final class SnapshotDataset extends RawOp implements Operand<TType> {
+public final class SnapshotDataset extends RawOp implements Operand<Tensor> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.data.SnapshotDataset}
@@ -359,8 +359,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) handle;
+  public Output<Tensor> asOutput() {
+    return (Output<Tensor>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SparseTensorSliceDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SparseTensorSliceDataset.java
index fa5b61137ab..04de4360549 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SparseTensorSliceDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SparseTensorSliceDataset.java
@@ -21,17 +21,17 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
 
 /**
  * Creates a dataset that splits a SparseTensor into elements row-wise.
  */
-public final class SparseTensorSliceDataset extends RawOp implements Operand<TType> {
+public final class SparseTensorSliceDataset extends RawOp implements Operand<Tensor> {
   
   /**
    * Factory method to create a class wrapping a new SparseTensorSliceDataset operation.
@@ -43,7 +43,7 @@ public final class SparseTensorSliceDataset extends RawOp implements Operand<TTy
    * @return a new instance of SparseTensorSliceDataset
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> SparseTensorSliceDataset create(Scope scope, Operand<TInt64> indices, Operand<T> values, Operand<TInt64> denseShape) {
+  public static <T extends Tensor> SparseTensorSliceDataset create(Scope scope, Operand<TInt64> indices, Operand<T> values, Operand<TInt64> denseShape) {
     OperationBuilder opBuilder = scope.env().opBuilder("SparseTensorSliceDataset", scope.makeOpName("SparseTensorSliceDataset"));
     opBuilder.addInput(indices.asOutput());
     opBuilder.addInput(values.asOutput());
@@ -60,8 +60,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) handle;
+  public Output<Tensor> asOutput() {
+    return (Output<Tensor>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SqlDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SqlDataset.java
index 59a0a47cd4d..94bd2a8763c 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SqlDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SqlDataset.java
@@ -23,18 +23,18 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TString;
-import org.tensorflow.types.family.TType;
 
 /**
  * Creates a dataset that executes a SQL query and emits rows of the result set.
  */
-public final class SqlDataset extends RawOp implements Operand<TType> {
+public final class SqlDataset extends RawOp implements Operand<Tensor> {
   
   /**
    * Factory method to create a class wrapping a new SqlDataset operation.
@@ -75,8 +75,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) handle;
+  public Output<Tensor> asOutput() {
+    return (Output<Tensor>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/StatsAggregatorHandle.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/StatsAggregatorHandle.java
index f510b437c7d..0ca63d5f345 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/StatsAggregatorHandle.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/StatsAggregatorHandle.java
@@ -21,16 +21,16 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Creates a statistics manager resource.
  */
-public final class StatsAggregatorHandle extends RawOp implements Operand<TType> {
+public final class StatsAggregatorHandle extends RawOp implements Operand<Tensor> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.data.StatsAggregatorHandle}
@@ -106,8 +106,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) handle;
+  public Output<Tensor> asOutput() {
+    return (Output<Tensor>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/TakeDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/TakeDataset.java
index 3229f055924..a2a74c6e377 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/TakeDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/TakeDataset.java
@@ -23,19 +23,19 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
 
 /**
  * Creates a dataset that contains `count` elements from the `input_dataset`.
  */
 @Operator(group = "data")
-public final class TakeDataset extends RawOp implements Operand<TType> {
+public final class TakeDataset extends RawOp implements Operand<Tensor> {
   
   /**
    * Factory method to create a class wrapping a new TakeDataset operation.
@@ -76,8 +76,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) handle;
+  public Output<Tensor> asOutput() {
+    return (Output<Tensor>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/TensorDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/TensorDataset.java
index 974f9562859..03f508352d2 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/TensorDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/TensorDataset.java
@@ -22,18 +22,18 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.Operands;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Creates a dataset that emits `components` as a tuple of tensors once.
  */
-public final class TensorDataset extends RawOp implements Operand<TType> {
+public final class TensorDataset extends RawOp implements Operand<Tensor> {
   
   /**
    * Factory method to create a class wrapping a new TensorDataset operation.
@@ -64,8 +64,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) handle;
+  public Output<Tensor> asOutput() {
+    return (Output<Tensor>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/TensorSliceDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/TensorSliceDataset.java
index 0d0ca40de32..3593b6d0e4c 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/TensorSliceDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/TensorSliceDataset.java
@@ -22,19 +22,19 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.Operands;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Creates a dataset that emits each dim-0 slice of `components` once.
  */
 @Operator(group = "data")
-public final class TensorSliceDataset extends RawOp implements Operand<TType> {
+public final class TensorSliceDataset extends RawOp implements Operand<Tensor> {
   
   /**
    * Factory method to create a class wrapping a new TensorSliceDataset operation.
@@ -65,8 +65,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) handle;
+  public Output<Tensor> asOutput() {
+    return (Output<Tensor>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/TextLineDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/TextLineDataset.java
index a80851bb2ed..e66b001eb72 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/TextLineDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/TextLineDataset.java
@@ -21,19 +21,19 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
 import org.tensorflow.types.TString;
-import org.tensorflow.types.family.TType;
 
 /**
  * Creates a dataset that emits the lines of one or more text files.
  */
 @Operator(group = "data")
-public final class TextLineDataset extends RawOp implements Operand<TType> {
+public final class TextLineDataset extends RawOp implements Operand<Tensor> {
   
   /**
    * Factory method to create a class wrapping a new TextLineDataset operation.
@@ -64,8 +64,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) handle;
+  public Output<Tensor> asOutput() {
+    return (Output<Tensor>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/TfRecordDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/TfRecordDataset.java
index 4795e623f82..04043e1d38d 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/TfRecordDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/TfRecordDataset.java
@@ -21,19 +21,19 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
 import org.tensorflow.types.TString;
-import org.tensorflow.types.family.TType;
 
 /**
  * Creates a dataset that emits the records from one or more TFRecord files.
  */
 @Operator(group = "data")
-public final class TfRecordDataset extends RawOp implements Operand<TType> {
+public final class TfRecordDataset extends RawOp implements Operand<Tensor> {
   
   /**
    * Factory method to create a class wrapping a new TfRecordDataset operation.
@@ -65,8 +65,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) handle;
+  public Output<Tensor> asOutput() {
+    return (Output<Tensor>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ThreadPoolDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ThreadPoolDataset.java
index 0b1ab263ffa..2e5893a3274 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ThreadPoolDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ThreadPoolDataset.java
@@ -23,17 +23,17 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Creates a dataset that uses a custom thread pool to compute `input_dataset`.
  */
-public final class ThreadPoolDataset extends RawOp implements Operand<TType> {
+public final class ThreadPoolDataset extends RawOp implements Operand<Tensor> {
   
   /**
    * Factory method to create a class wrapping a new ThreadPoolDataset operation.
@@ -72,8 +72,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) handle;
+  public Output<Tensor> asOutput() {
+    return (Output<Tensor>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ThreadPoolHandle.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ThreadPoolHandle.java
index 6d9434cdff3..10847d750c4 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ThreadPoolHandle.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ThreadPoolHandle.java
@@ -21,16 +21,16 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Creates a dataset that uses a custom thread pool to compute `input_dataset`.
  */
-public final class ThreadPoolHandle extends RawOp implements Operand<TType> {
+public final class ThreadPoolHandle extends RawOp implements Operand<Tensor> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.data.ThreadPoolHandle}
@@ -135,8 +135,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) handle;
+  public Output<Tensor> asOutput() {
+    return (Output<Tensor>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/UnbatchDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/UnbatchDataset.java
index 1bca9b693cc..001d8d7aab0 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/UnbatchDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/UnbatchDataset.java
@@ -23,17 +23,17 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * A dataset that splits the elements of its input into multiple elements.
  */
-public final class UnbatchDataset extends RawOp implements Operand<TType> {
+public final class UnbatchDataset extends RawOp implements Operand<Tensor> {
   
   /**
    * Factory method to create a class wrapping a new UnbatchDataset operation.
@@ -70,8 +70,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) handle;
+  public Output<Tensor> asOutput() {
+    return (Output<Tensor>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/UniqueDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/UniqueDataset.java
index 817519ef6f8..26597f1d562 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/UniqueDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/UniqueDataset.java
@@ -23,17 +23,17 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Creates a dataset that contains the unique elements of `input_dataset`.
  */
-public final class UniqueDataset extends RawOp implements Operand<TType> {
+public final class UniqueDataset extends RawOp implements Operand<Tensor> {
   
   /**
    * Factory method to create a class wrapping a new UniqueDataset operation.
@@ -70,8 +70,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) handle;
+  public Output<Tensor> asOutput() {
+    return (Output<Tensor>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/UnwrapDatasetVariant.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/UnwrapDatasetVariant.java
index d5125db92c8..b3189ce18b9 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/UnwrapDatasetVariant.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/UnwrapDatasetVariant.java
@@ -21,15 +21,15 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  */
-public final class UnwrapDatasetVariant extends RawOp implements Operand<TType> {
+public final class UnwrapDatasetVariant extends RawOp implements Operand<Tensor> {
   
   /**
    * Factory method to create a class wrapping a new UnwrapDatasetVariant operation.
@@ -54,8 +54,8 @@ public Output<?> outputHandle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) outputHandle;
+  public Output<Tensor> asOutput() {
+    return (Output<Tensor>) outputHandle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/WindowDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/WindowDataset.java
index d5edcc917d8..4f48b788992 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/WindowDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/WindowDataset.java
@@ -23,6 +23,7 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
@@ -30,7 +31,6 @@
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TBool;
 import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
 
 /**
  *   Combines (nests of) input elements into a dataset of (nests of) windows.
@@ -76,7 +76,7 @@
  *   - `tf.data.Dataset.from_tensor_slices({"a": range(4)}).window(2)`
  *     produces `{{"a": {0, 1}}, {"a": {2, 3}}}`
  */
-public final class WindowDataset extends RawOp implements Operand<TType> {
+public final class WindowDataset extends RawOp implements Operand<Tensor> {
   
   /**
    * Factory method to create a class wrapping a new WindowDataset operation.
@@ -127,8 +127,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) handle;
+  public Output<Tensor> asOutput() {
+    return (Output<Tensor>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/WrapDatasetVariant.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/WrapDatasetVariant.java
index bcea6df7895..a7af207692b 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/WrapDatasetVariant.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/WrapDatasetVariant.java
@@ -21,15 +21,15 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  */
-public final class WrapDatasetVariant extends RawOp implements Operand<TType> {
+public final class WrapDatasetVariant extends RawOp implements Operand<Tensor> {
   
   /**
    * Factory method to create a class wrapping a new WrapDatasetVariant operation.
@@ -54,8 +54,8 @@ public Output<?> outputHandle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) outputHandle;
+  public Output<Tensor> asOutput() {
+    return (Output<Tensor>) outputHandle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ZipDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ZipDataset.java
index d4f54946f5f..36d9a1eb79b 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ZipDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ZipDataset.java
@@ -23,13 +23,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.Operands;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Creates a dataset that zips together `input_datasets`.
@@ -41,7 +41,7 @@
  * dataset, and no error will be raised if input datasets have different sizes.
  */
 @Operator(group = "data")
-public final class ZipDataset extends RawOp implements Operand<TType> {
+public final class ZipDataset extends RawOp implements Operand<Tensor> {
   
   /**
    * Factory method to create a class wrapping a new ZipDataset operation.
@@ -78,8 +78,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) handle;
+  public Output<Tensor> asOutput() {
+    return (Output<Tensor>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/AssertCardinalityDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/AssertCardinalityDataset.java
index 96c02bf5fc1..cc904d6035c 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/AssertCardinalityDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/AssertCardinalityDataset.java
@@ -23,17 +23,17 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
 
 /**
  */
-public final class AssertCardinalityDataset extends RawOp implements Operand<TType> {
+public final class AssertCardinalityDataset extends RawOp implements Operand<Tensor> {
   
   /**
    * Factory method to create a class wrapping a new AssertCardinalityDataset operation.
@@ -72,8 +72,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) handle;
+  public Output<Tensor> asOutput() {
+    return (Output<Tensor>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/AssertNextDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/AssertNextDataset.java
index cd0c5300df6..ac27b6276ea 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/AssertNextDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/AssertNextDataset.java
@@ -23,17 +23,17 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TString;
-import org.tensorflow.types.family.TType;
 
 /**
  */
-public final class AssertNextDataset extends RawOp implements Operand<TType> {
+public final class AssertNextDataset extends RawOp implements Operand<Tensor> {
   
   /**
    * Factory method to create a class wrapping a new AssertNextDataset operation.
@@ -72,8 +72,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) handle;
+  public Output<Tensor> asOutput() {
+    return (Output<Tensor>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/AutoShardDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/AutoShardDataset.java
index 3c1eb053091..b95e156bfb9 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/AutoShardDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/AutoShardDataset.java
@@ -23,13 +23,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
 
 /**
  * Creates a dataset that shards the input dataset.
@@ -42,7 +42,7 @@
  * This dataset will throw a NotFound error if we cannot shard the dataset
  * automatically.
  */
-public final class AutoShardDataset extends RawOp implements Operand<TType> {
+public final class AutoShardDataset extends RawOp implements Operand<Tensor> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.data.experimental.AutoShardDataset}
@@ -117,8 +117,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) handle;
+  public Output<Tensor> asOutput() {
+    return (Output<Tensor>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/BytesProducedStatsDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/BytesProducedStatsDataset.java
index 7fe1290b130..baf4210eb4e 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/BytesProducedStatsDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/BytesProducedStatsDataset.java
@@ -23,18 +23,18 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TString;
-import org.tensorflow.types.family.TType;
 
 /**
  * Records the bytes size of each element of `input_dataset` in a StatsAggregator.
  */
-public final class BytesProducedStatsDataset extends RawOp implements Operand<TType> {
+public final class BytesProducedStatsDataset extends RawOp implements Operand<Tensor> {
   
   /**
    * Factory method to create a class wrapping a new BytesProducedStatsDataset operation.
@@ -73,8 +73,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) handle;
+  public Output<Tensor> asOutput() {
+    return (Output<Tensor>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/CSVDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/CSVDataset.java
index d0498612cde..584080913d9 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/CSVDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/CSVDataset.java
@@ -22,6 +22,7 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.Operands;
 import org.tensorflow.op.RawOp;
@@ -31,11 +32,10 @@
 import org.tensorflow.types.TBool;
 import org.tensorflow.types.TInt64;
 import org.tensorflow.types.TString;
-import org.tensorflow.types.family.TType;
 
 /**
  */
-public final class CSVDataset extends RawOp implements Operand<TType> {
+public final class CSVDataset extends RawOp implements Operand<Tensor> {
   
   /**
    * Factory method to create a class wrapping a new CSVDataset operation.
@@ -82,8 +82,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) handle;
+  public Output<Tensor> asOutput() {
+    return (Output<Tensor>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/ChooseFastestDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/ChooseFastestDataset.java
index f731d948490..f2ca0bcaa95 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/ChooseFastestDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/ChooseFastestDataset.java
@@ -23,17 +23,17 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.Operands;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  */
-public final class ChooseFastestDataset extends RawOp implements Operand<TType> {
+public final class ChooseFastestDataset extends RawOp implements Operand<Tensor> {
   
   /**
    * Factory method to create a class wrapping a new ChooseFastestDataset operation.
@@ -72,8 +72,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) handle;
+  public Output<Tensor> asOutput() {
+    return (Output<Tensor>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/DenseToSparseBatchDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/DenseToSparseBatchDataset.java
index d464f422836..a079ea085db 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/DenseToSparseBatchDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/DenseToSparseBatchDataset.java
@@ -23,18 +23,18 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
 
 /**
  * Creates a dataset that batches input elements into a SparseTensor.
  */
-public final class DenseToSparseBatchDataset extends RawOp implements Operand<TType> {
+public final class DenseToSparseBatchDataset extends RawOp implements Operand<Tensor> {
   
   /**
    * Factory method to create a class wrapping a new DenseToSparseBatchDataset operation.
@@ -78,8 +78,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) handle;
+  public Output<Tensor> asOutput() {
+    return (Output<Tensor>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/DirectedInterleaveDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/DirectedInterleaveDataset.java
index 63a06a16201..3886abd836b 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/DirectedInterleaveDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/DirectedInterleaveDataset.java
@@ -23,18 +23,18 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.Operands;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * A substitute for `InterleaveDataset` on a fixed list of `N` datasets.
  */
-public final class DirectedInterleaveDataset extends RawOp implements Operand<TType> {
+public final class DirectedInterleaveDataset extends RawOp implements Operand<Tensor> {
   
   /**
    * Factory method to create a class wrapping a new DirectedInterleaveDataset operation.
@@ -75,8 +75,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) handle;
+  public Output<Tensor> asOutput() {
+    return (Output<Tensor>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/IgnoreErrorsDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/IgnoreErrorsDataset.java
index 22d734831da..5d592784bc2 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/IgnoreErrorsDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/IgnoreErrorsDataset.java
@@ -23,17 +23,17 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Creates a dataset that contains the elements of `input_dataset` ignoring errors.
  */
-public final class IgnoreErrorsDataset extends RawOp implements Operand<TType> {
+public final class IgnoreErrorsDataset extends RawOp implements Operand<Tensor> {
   
   /**
    * Factory method to create a class wrapping a new IgnoreErrorsDataset operation.
@@ -70,8 +70,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) handle;
+  public Output<Tensor> asOutput() {
+    return (Output<Tensor>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/LatencyStatsDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/LatencyStatsDataset.java
index ad901295c28..86ee0610c2a 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/LatencyStatsDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/LatencyStatsDataset.java
@@ -23,18 +23,18 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TString;
-import org.tensorflow.types.family.TType;
 
 /**
  * Records the latency of producing `input_dataset` elements in a StatsAggregator.
  */
-public final class LatencyStatsDataset extends RawOp implements Operand<TType> {
+public final class LatencyStatsDataset extends RawOp implements Operand<Tensor> {
   
   /**
    * Factory method to create a class wrapping a new LatencyStatsDataset operation.
@@ -73,8 +73,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) handle;
+  public Output<Tensor> asOutput() {
+    return (Output<Tensor>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/LmdbDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/LmdbDataset.java
index e6c489833af..9434c22b7bd 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/LmdbDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/LmdbDataset.java
@@ -23,17 +23,17 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TString;
-import org.tensorflow.types.family.TType;
 
 /**
  */
-public final class LmdbDataset extends RawOp implements Operand<TType> {
+public final class LmdbDataset extends RawOp implements Operand<Tensor> {
   
   /**
    * Factory method to create a class wrapping a new LmdbDataset operation.
@@ -70,8 +70,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) handle;
+  public Output<Tensor> asOutput() {
+    return (Output<Tensor>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/MatchingFilesDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/MatchingFilesDataset.java
index e2e29feeba4..f207ec77f3c 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/MatchingFilesDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/MatchingFilesDataset.java
@@ -21,16 +21,16 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TString;
-import org.tensorflow.types.family.TType;
 
 /**
  */
-public final class MatchingFilesDataset extends RawOp implements Operand<TType> {
+public final class MatchingFilesDataset extends RawOp implements Operand<Tensor> {
   
   /**
    * Factory method to create a class wrapping a new MatchingFilesDataset operation.
@@ -55,8 +55,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) handle;
+  public Output<Tensor> asOutput() {
+    return (Output<Tensor>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/MaxIntraOpParallelismDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/MaxIntraOpParallelismDataset.java
index fb69729fab0..d4cf2dcd57e 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/MaxIntraOpParallelismDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/MaxIntraOpParallelismDataset.java
@@ -23,18 +23,18 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
 
 /**
  * Creates a dataset that overrides the maximum intra-op parallelism.
  */
-public final class MaxIntraOpParallelismDataset extends RawOp implements Operand<TType> {
+public final class MaxIntraOpParallelismDataset extends RawOp implements Operand<Tensor> {
   
   /**
    * Factory method to create a class wrapping a new MaxIntraOpParallelismDataset operation.
@@ -73,8 +73,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) handle;
+  public Output<Tensor> asOutput() {
+    return (Output<Tensor>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/NonSerializableDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/NonSerializableDataset.java
index 9f07f37f804..015b710c2c5 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/NonSerializableDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/NonSerializableDataset.java
@@ -23,16 +23,16 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  */
-public final class NonSerializableDataset extends RawOp implements Operand<TType> {
+public final class NonSerializableDataset extends RawOp implements Operand<Tensor> {
   
   /**
    * Factory method to create a class wrapping a new NonSerializableDataset operation.
@@ -69,8 +69,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) handle;
+  public Output<Tensor> asOutput() {
+    return (Output<Tensor>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/ParseExampleDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/ParseExampleDataset.java
index 4c24cf99558..bcf05b640fe 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/ParseExampleDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/ParseExampleDataset.java
@@ -23,6 +23,7 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.Operands;
 import org.tensorflow.op.RawOp;
@@ -31,12 +32,11 @@
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Transforms `input_dataset` containing `Example` protos as vectors of DT_STRING into a dataset of `Tensor` or `SparseTensor` objects representing the parsed features.
  */
-public final class ParseExampleDataset extends RawOp implements Operand<TType> {
+public final class ParseExampleDataset extends RawOp implements Operand<Tensor> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.data.experimental.ParseExampleDataset}
@@ -191,8 +191,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) handle;
+  public Output<Tensor> asOutput() {
+    return (Output<Tensor>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/PrivateThreadPoolDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/PrivateThreadPoolDataset.java
index a52fd41140e..c1b9d9ced27 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/PrivateThreadPoolDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/PrivateThreadPoolDataset.java
@@ -23,18 +23,18 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
 
 /**
  * Creates a dataset that uses a custom thread pool to compute `input_dataset`.
  */
-public final class PrivateThreadPoolDataset extends RawOp implements Operand<TType> {
+public final class PrivateThreadPoolDataset extends RawOp implements Operand<Tensor> {
   
   /**
    * Factory method to create a class wrapping a new PrivateThreadPoolDataset operation.
@@ -73,8 +73,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) handle;
+  public Output<Tensor> asOutput() {
+    return (Output<Tensor>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/RandomDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/RandomDataset.java
index 8f9109ab3d5..772e89379b5 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/RandomDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/RandomDataset.java
@@ -23,18 +23,18 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
 
 /**
  * Creates a Dataset that returns pseudorandom numbers.
  */
-public final class RandomDataset extends RawOp implements Operand<TType> {
+public final class RandomDataset extends RawOp implements Operand<Tensor> {
   
   /**
    * Factory method to create a class wrapping a new RandomDataset operation.
@@ -75,8 +75,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) handle;
+  public Output<Tensor> asOutput() {
+    return (Output<Tensor>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/RebatchDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/RebatchDataset.java
index 6111166128e..215b1cc22eb 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/RebatchDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/RebatchDataset.java
@@ -23,13 +23,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
 
 /**
  * Creates a dataset that changes the batch size.
@@ -37,7 +37,7 @@
  * Creates a dataset that changes the batch size of the dataset to current batch
  * size // num_replicas.
  */
-public final class RebatchDataset extends RawOp implements Operand<TType> {
+public final class RebatchDataset extends RawOp implements Operand<Tensor> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.data.experimental.RebatchDataset}
@@ -112,8 +112,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) handle;
+  public Output<Tensor> asOutput() {
+    return (Output<Tensor>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/SetStatsAggregatorDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/SetStatsAggregatorDataset.java
index e2535dcecd7..dcbc74627d0 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/SetStatsAggregatorDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/SetStatsAggregatorDataset.java
@@ -23,17 +23,17 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TString;
-import org.tensorflow.types.family.TType;
 
 /**
  */
-public final class SetStatsAggregatorDataset extends RawOp implements Operand<TType> {
+public final class SetStatsAggregatorDataset extends RawOp implements Operand<Tensor> {
   
   /**
    * Factory method to create a class wrapping a new SetStatsAggregatorDataset operation.
@@ -76,8 +76,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) handle;
+  public Output<Tensor> asOutput() {
+    return (Output<Tensor>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/SleepDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/SleepDataset.java
index 97ddeb96999..394dfbf8d4b 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/SleepDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/SleepDataset.java
@@ -23,17 +23,17 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
 
 /**
  */
-public final class SleepDataset extends RawOp implements Operand<TType> {
+public final class SleepDataset extends RawOp implements Operand<Tensor> {
   
   /**
    * Factory method to create a class wrapping a new SleepDataset operation.
@@ -72,8 +72,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) handle;
+  public Output<Tensor> asOutput() {
+    return (Output<Tensor>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/SlidingWindowDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/SlidingWindowDataset.java
index b8dbded85e9..93ec0151c10 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/SlidingWindowDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/SlidingWindowDataset.java
@@ -23,18 +23,18 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
 
 /**
  * Creates a dataset that passes a sliding window over `input_dataset`.
  */
-public final class SlidingWindowDataset extends RawOp implements Operand<TType> {
+public final class SlidingWindowDataset extends RawOp implements Operand<Tensor> {
   
   /**
    * Factory method to create a class wrapping a new SlidingWindowDataset operation.
@@ -80,8 +80,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) handle;
+  public Output<Tensor> asOutput() {
+    return (Output<Tensor>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/SqlDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/SqlDataset.java
index a68f46a5c08..466191b26ba 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/SqlDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/SqlDataset.java
@@ -23,18 +23,18 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TString;
-import org.tensorflow.types.family.TType;
 
 /**
  * Creates a dataset that executes a SQL query and emits rows of the result set.
  */
-public final class SqlDataset extends RawOp implements Operand<TType> {
+public final class SqlDataset extends RawOp implements Operand<Tensor> {
   
   /**
    * Factory method to create a class wrapping a new SqlDataset operation.
@@ -75,8 +75,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) handle;
+  public Output<Tensor> asOutput() {
+    return (Output<Tensor>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/StatsAggregatorHandle.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/StatsAggregatorHandle.java
index ec2e57ae4e7..bc8c6ddcd6e 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/StatsAggregatorHandle.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/StatsAggregatorHandle.java
@@ -21,15 +21,15 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  */
-public final class StatsAggregatorHandle extends RawOp implements Operand<TType> {
+public final class StatsAggregatorHandle extends RawOp implements Operand<Tensor> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.data.experimental.StatsAggregatorHandle}
@@ -105,8 +105,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) handle;
+  public Output<Tensor> asOutput() {
+    return (Output<Tensor>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/ThreadPoolDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/ThreadPoolDataset.java
index e3fdc2b92cf..19522616699 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/ThreadPoolDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/ThreadPoolDataset.java
@@ -23,17 +23,17 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Creates a dataset that uses a custom thread pool to compute `input_dataset`.
  */
-public final class ThreadPoolDataset extends RawOp implements Operand<TType> {
+public final class ThreadPoolDataset extends RawOp implements Operand<Tensor> {
   
   /**
    * Factory method to create a class wrapping a new ThreadPoolDataset operation.
@@ -72,8 +72,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) handle;
+  public Output<Tensor> asOutput() {
+    return (Output<Tensor>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/ThreadPoolHandle.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/ThreadPoolHandle.java
index 15adc678211..1d43658188d 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/ThreadPoolHandle.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/ThreadPoolHandle.java
@@ -21,16 +21,16 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Creates a dataset that uses a custom thread pool to compute `input_dataset`.
  */
-public final class ThreadPoolHandle extends RawOp implements Operand<TType> {
+public final class ThreadPoolHandle extends RawOp implements Operand<Tensor> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.data.experimental.ThreadPoolHandle}
@@ -135,8 +135,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) handle;
+  public Output<Tensor> asOutput() {
+    return (Output<Tensor>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/UnbatchDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/UnbatchDataset.java
index dfa611faa0e..0d702b518ce 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/UnbatchDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/UnbatchDataset.java
@@ -23,17 +23,17 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * A dataset that splits the elements of its input into multiple elements.
  */
-public final class UnbatchDataset extends RawOp implements Operand<TType> {
+public final class UnbatchDataset extends RawOp implements Operand<Tensor> {
   
   /**
    * Factory method to create a class wrapping a new UnbatchDataset operation.
@@ -70,8 +70,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) handle;
+  public Output<Tensor> asOutput() {
+    return (Output<Tensor>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/UniqueDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/UniqueDataset.java
index eb213348e72..ea635f72d8f 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/UniqueDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/UniqueDataset.java
@@ -23,17 +23,17 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Creates a dataset that contains the unique elements of `input_dataset`.
  */
-public final class UniqueDataset extends RawOp implements Operand<TType> {
+public final class UniqueDataset extends RawOp implements Operand<Tensor> {
   
   /**
    * Factory method to create a class wrapping a new UniqueDataset operation.
@@ -70,8 +70,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) handle;
+  public Output<Tensor> asOutput() {
+    return (Output<Tensor>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/debugging/CheckNumerics.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/debugging/CheckNumerics.java
index e26417c561d..c14f3fd23d3 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/debugging/CheckNumerics.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/debugging/CheckNumerics.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Checks a tensor for NaN, -Inf and +Inf values.
@@ -38,7 +38,7 @@
  * 
  * @param <T> data type for {@code output()} output
  */
-public final class CheckNumerics<T extends TNumber> extends RawOp implements Operand<T> {
+public final class CheckNumerics<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new CheckNumerics operation.
@@ -49,7 +49,7 @@ public final class CheckNumerics<T extends TNumber> extends RawOp implements Ope
    * @return a new instance of CheckNumerics
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> CheckNumerics<T> create(Scope scope, Operand<T> tensor, String message) {
+  public static <T extends Tensor & TNumber> CheckNumerics<T> create(Scope scope, Operand<T> tensor, String message) {
     OperationBuilder opBuilder = scope.env().opBuilder("CheckNumericsV2", scope.makeOpName("CheckNumerics"));
     opBuilder.addInput(tensor.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/debugging/DebugGradientIdentity.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/debugging/DebugGradientIdentity.java
index 4214a50ab41..10b63b63587 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/debugging/DebugGradientIdentity.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/debugging/DebugGradientIdentity.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Identity op for gradient debugging.
@@ -36,7 +36,7 @@
  * 
  * @param <T> data type for {@code output()} output
  */
-public final class DebugGradientIdentity<T extends TType> extends RawOp implements Operand<T> {
+public final class DebugGradientIdentity<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new DebugGradientIdentity operation.
@@ -46,7 +46,7 @@ public final class DebugGradientIdentity<T extends TType> extends RawOp implemen
    * @return a new instance of DebugGradientIdentity
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> DebugGradientIdentity<T> create(Scope scope, Operand<T> input) {
+  public static <T extends Tensor> DebugGradientIdentity<T> create(Scope scope, Operand<T> input) {
     OperationBuilder opBuilder = scope.env().opBuilder("DebugGradientIdentity", scope.makeOpName("DebugGradientIdentity"));
     opBuilder.addInput(input.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/debugging/DebugGradientRefIdentity.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/debugging/DebugGradientRefIdentity.java
index fb4a1d4cfef..220ce202ab2 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/debugging/DebugGradientRefIdentity.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/debugging/DebugGradientRefIdentity.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Identity op for gradient debugging.
@@ -36,7 +36,7 @@
  * 
  * @param <T> data type for {@code output()} output
  */
-public final class DebugGradientRefIdentity<T extends TType> extends RawOp implements Operand<T> {
+public final class DebugGradientRefIdentity<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new DebugGradientRefIdentity operation.
@@ -46,7 +46,7 @@ public final class DebugGradientRefIdentity<T extends TType> extends RawOp imple
    * @return a new instance of DebugGradientRefIdentity
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> DebugGradientRefIdentity<T> create(Scope scope, Operand<T> input) {
+  public static <T extends Tensor> DebugGradientRefIdentity<T> create(Scope scope, Operand<T> input) {
     OperationBuilder opBuilder = scope.env().opBuilder("DebugGradientRefIdentity", scope.makeOpName("DebugGradientRefIdentity"));
     opBuilder.addInput(input.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/debugging/DebugIdentity.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/debugging/DebugIdentity.java
index 14f719c90f8..936d17bb297 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/debugging/DebugIdentity.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/debugging/DebugIdentity.java
@@ -22,11 +22,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Debug Identity V2 Op.
@@ -43,7 +43,7 @@
  * 
  * @param <T> data type for {@code output()} output
  */
-public final class DebugIdentity<T extends TType> extends RawOp implements Operand<T> {
+public final class DebugIdentity<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.debugging.DebugIdentity}
@@ -130,7 +130,7 @@ private Options() {
    * @return a new instance of DebugIdentity
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> DebugIdentity<T> create(Scope scope, Operand<T> input, Options... options) {
+  public static <T extends Tensor> DebugIdentity<T> create(Scope scope, Operand<T> input, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("DebugIdentityV2", scope.makeOpName("DebugIdentity"));
     opBuilder.addInput(input.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/debugging/DebugNanCount.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/debugging/DebugNanCount.java
index 30bc6931743..ff889ac64a0 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/debugging/DebugNanCount.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/debugging/DebugNanCount.java
@@ -22,12 +22,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
 
 /**
  * Debug NaN Value Counter Op.
@@ -97,7 +97,7 @@ private Options() {
    * @return a new instance of DebugNanCount
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> DebugNanCount create(Scope scope, Operand<T> input, Options... options) {
+  public static <T extends Tensor> DebugNanCount create(Scope scope, Operand<T> input, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("DebugNanCount", scope.makeOpName("DebugNanCount"));
     opBuilder.addInput(input.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/debugging/DebugNumericsSummary.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/debugging/DebugNumericsSummary.java
index 85c20b1eef0..1afff31e562 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/debugging/DebugNumericsSummary.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/debugging/DebugNumericsSummary.java
@@ -22,13 +22,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TFloat32;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Debug Numeric Summary V2 Op.
@@ -39,7 +39,7 @@
  * 
  * @param <U> data type for {@code output()} output
  */
-public final class DebugNumericsSummary<U extends TNumber> extends RawOp implements Operand<U> {
+public final class DebugNumericsSummary<U extends Tensor & TNumber> extends RawOp implements Operand<U> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.debugging.DebugNumericsSummary}
@@ -132,7 +132,7 @@ private Options() {
    * @return a new instance of DebugNumericsSummary
    */
   @Endpoint(describeByClass = true)
-  public static <U extends TNumber, T extends TType> DebugNumericsSummary<U> create(Scope scope, Operand<T> input, DataType<U> outputDtype, Options... options) {
+  public static <U extends Tensor & TNumber, T extends Tensor> DebugNumericsSummary<U> create(Scope scope, Operand<T> input, DataType<U> outputDtype, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("DebugNumericSummaryV2", scope.makeOpName("DebugNumericsSummary"));
     opBuilder.addInput(input.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
@@ -159,7 +159,7 @@ public static <U extends TNumber, T extends TType> DebugNumericsSummary<U> creat
    * @return a new instance of DebugNumericsSummary
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> DebugNumericsSummary<TFloat32> create(Scope scope, Operand<T> input, Options... options) {
+  public static <T extends Tensor> DebugNumericsSummary<TFloat32> create(Scope scope, Operand<T> input, Options... options) {
     return create(scope, input, TFloat32.DTYPE, options);
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/dtypes/AsString.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/dtypes/AsString.java
index 9d95a0ee6a0..3661a58f6a4 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/dtypes/AsString.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/dtypes/AsString.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TString;
-import org.tensorflow.types.family.TType;
 
 /**
  * Converts each entry in the given tensor to strings.
@@ -116,7 +116,7 @@ private Options() {
    * @return a new instance of AsString
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> AsString create(Scope scope, Operand<T> input, Options... options) {
+  public static <T extends Tensor> AsString create(Scope scope, Operand<T> input, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("AsString", scope.makeOpName("AsString"));
     opBuilder.addInput(input.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/dtypes/Cast.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/dtypes/Cast.java
index 49699f07c8c..d805dcb9783 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/dtypes/Cast.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/dtypes/Cast.java
@@ -22,11 +22,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Cast x of type SrcT to y of DstT.
@@ -34,7 +34,7 @@
  * @param <U> data type for {@code y()} output
  */
 @Operator(group = "dtypes")
-public final class Cast<U extends TType> extends RawOp implements Operand<U> {
+public final class Cast<U extends Tensor> extends RawOp implements Operand<U> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.dtypes.Cast}
@@ -65,7 +65,7 @@ private Options() {
    * @return a new instance of Cast
    */
   @Endpoint(describeByClass = true)
-  public static <U extends TType, T extends TType> Cast<U> create(Scope scope, Operand<T> x, DataType<U> DstT, Options... options) {
+  public static <U extends Tensor, T extends Tensor> Cast<U> create(Scope scope, Operand<T> x, DataType<U> DstT, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("Cast", scope.makeOpName("Cast"));
     opBuilder.addInput(x.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/dtypes/Complex.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/dtypes/Complex.java
index 5e8918785a0..fcf562a48dc 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/dtypes/Complex.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/dtypes/Complex.java
@@ -22,12 +22,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Converts two real numbers to a complex number.
@@ -50,7 +50,7 @@
  * @param <U> data type for {@code out()} output
  */
 @Operator(group = "dtypes")
-public final class Complex<U extends TType> extends RawOp implements Operand<U> {
+public final class Complex<U extends Tensor> extends RawOp implements Operand<U> {
   
   /**
    * Factory method to create a class wrapping a new Complex operation.
@@ -62,7 +62,7 @@ public final class Complex<U extends TType> extends RawOp implements Operand<U>
    * @return a new instance of Complex
    */
   @Endpoint(describeByClass = true)
-  public static <U extends TType, T extends TNumber> Complex<U> create(Scope scope, Operand<T> real, Operand<T> imag, DataType<U> Tout) {
+  public static <U extends Tensor, T extends Tensor & TNumber> Complex<U> create(Scope scope, Operand<T> real, Operand<T> imag, DataType<U> Tout) {
     OperationBuilder opBuilder = scope.env().opBuilder("Complex", scope.makeOpName("Complex"));
     opBuilder.addInput(real.asOutput());
     opBuilder.addInput(imag.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/dtypes/ToBool.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/dtypes/ToBool.java
index 72b3d46d00c..05987505418 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/dtypes/ToBool.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/dtypes/ToBool.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TBool;
-import org.tensorflow.types.family.TType;
 
 /**
  * Converts a tensor to a scalar predicate.
@@ -54,7 +54,7 @@ public final class ToBool extends RawOp implements Operand<TBool> {
    * @return a new instance of ToBool
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> ToBool create(Scope scope, Operand<T> input) {
+  public static <T extends Tensor> ToBool create(Scope scope, Operand<T> input) {
     OperationBuilder opBuilder = scope.env().opBuilder("ToBool", scope.makeOpName("ToBool"));
     opBuilder.addInput(input.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesEnsembleResourceHandleOp.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesEnsembleResourceHandleOp.java
index 13522dd2fd0..de542108127 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesEnsembleResourceHandleOp.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesEnsembleResourceHandleOp.java
@@ -21,16 +21,16 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Creates a handle to a BoostedTreesEnsembleResource
  */
-public final class BoostedTreesEnsembleResourceHandleOp extends RawOp implements Operand<TType> {
+public final class BoostedTreesEnsembleResourceHandleOp extends RawOp implements Operand<Tensor> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.estimator.BoostedTreesEnsembleResourceHandleOp}
@@ -106,8 +106,8 @@ public Output<?> resource() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) resource;
+  public Output<Tensor> asOutput() {
+    return (Output<Tensor>) resource;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesQuantileStreamResourceHandleOp.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesQuantileStreamResourceHandleOp.java
index db925d767f9..ebf3285a95f 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesQuantileStreamResourceHandleOp.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesQuantileStreamResourceHandleOp.java
@@ -21,16 +21,16 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Creates a handle to a BoostedTreesQuantileStreamResource.
  */
-public final class BoostedTreesQuantileStreamResourceHandleOp extends RawOp implements Operand<TType> {
+public final class BoostedTreesQuantileStreamResourceHandleOp extends RawOp implements Operand<Tensor> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.estimator.BoostedTreesQuantileStreamResourceHandleOp}
@@ -106,8 +106,8 @@ public Output<?> resource() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) resource;
+  public Output<Tensor> asOutput() {
+    return (Output<Tensor>) resource;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/AdjustContrast.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/AdjustContrast.java
index ab5b3f32e61..5bf4fa92401 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/AdjustContrast.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/AdjustContrast.java
@@ -21,13 +21,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TFloat32;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Adjust the contrast of one or more images.
@@ -45,7 +45,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "image")
-public final class AdjustContrast<T extends TNumber> extends RawOp implements Operand<T> {
+public final class AdjustContrast<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new AdjustContrast operation.
@@ -56,7 +56,7 @@ public final class AdjustContrast<T extends TNumber> extends RawOp implements Op
    * @return a new instance of AdjustContrast
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> AdjustContrast<T> create(Scope scope, Operand<T> images, Operand<TFloat32> contrastFactor) {
+  public static <T extends Tensor & TNumber> AdjustContrast<T> create(Scope scope, Operand<T> images, Operand<TFloat32> contrastFactor) {
     OperationBuilder opBuilder = scope.env().opBuilder("AdjustContrastv2", scope.makeOpName("AdjustContrast"));
     opBuilder.addInput(images.asOutput());
     opBuilder.addInput(contrastFactor.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/AdjustHue.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/AdjustHue.java
index 2fe15aa50f1..da6eb729a76 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/AdjustHue.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/AdjustHue.java
@@ -21,13 +21,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TFloat32;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Adjust the hue of one or more images.
@@ -42,7 +42,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "image")
-public final class AdjustHue<T extends TNumber> extends RawOp implements Operand<T> {
+public final class AdjustHue<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new AdjustHue operation.
@@ -53,7 +53,7 @@ public final class AdjustHue<T extends TNumber> extends RawOp implements Operand
    * @return a new instance of AdjustHue
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> AdjustHue<T> create(Scope scope, Operand<T> images, Operand<TFloat32> delta) {
+  public static <T extends Tensor & TNumber> AdjustHue<T> create(Scope scope, Operand<T> images, Operand<TFloat32> delta) {
     OperationBuilder opBuilder = scope.env().opBuilder("AdjustHue", scope.makeOpName("AdjustHue"));
     opBuilder.addInput(images.asOutput());
     opBuilder.addInput(delta.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/AdjustSaturation.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/AdjustSaturation.java
index 03270949652..ffe643d8434 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/AdjustSaturation.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/AdjustSaturation.java
@@ -21,13 +21,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TFloat32;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Adjust the saturation of one or more images.
@@ -42,7 +42,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "image")
-public final class AdjustSaturation<T extends TNumber> extends RawOp implements Operand<T> {
+public final class AdjustSaturation<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new AdjustSaturation operation.
@@ -53,7 +53,7 @@ public final class AdjustSaturation<T extends TNumber> extends RawOp implements
    * @return a new instance of AdjustSaturation
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> AdjustSaturation<T> create(Scope scope, Operand<T> images, Operand<TFloat32> scale) {
+  public static <T extends Tensor & TNumber> AdjustSaturation<T> create(Scope scope, Operand<T> images, Operand<TFloat32> scale) {
     OperationBuilder opBuilder = scope.env().opBuilder("AdjustSaturation", scope.makeOpName("AdjustSaturation"));
     opBuilder.addInput(images.asOutput());
     opBuilder.addInput(scale.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/CropAndResize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/CropAndResize.java
index f0048c1014b..5c30c811267 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/CropAndResize.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/CropAndResize.java
@@ -21,6 +21,7 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -28,7 +29,6 @@
 import org.tensorflow.types.TFloat32;
 import org.tensorflow.types.TInt32;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Extracts crops from the input image tensor and resizes them.
@@ -108,7 +108,7 @@ private Options() {
    * @return a new instance of CropAndResize
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> CropAndResize create(Scope scope, Operand<T> image, Operand<TFloat32> boxes, Operand<TInt32> boxInd, Operand<TInt32> cropSize, Options... options) {
+  public static <T extends Tensor & TNumber> CropAndResize create(Scope scope, Operand<T> image, Operand<TFloat32> boxes, Operand<TInt32> boxInd, Operand<TInt32> cropSize, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("CropAndResize", scope.makeOpName("CropAndResize"));
     opBuilder.addInput(image.asOutput());
     opBuilder.addInput(boxes.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/CropAndResizeGradBoxes.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/CropAndResizeGradBoxes.java
index 67263c1e576..adfb2f39082 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/CropAndResizeGradBoxes.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/CropAndResizeGradBoxes.java
@@ -21,6 +21,7 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -28,7 +29,6 @@
 import org.tensorflow.types.TFloat32;
 import org.tensorflow.types.TInt32;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes the gradient of the crop_and_resize op wrt the input boxes tensor.
@@ -79,7 +79,7 @@ private Options() {
    * @return a new instance of CropAndResizeGradBoxes
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> CropAndResizeGradBoxes create(Scope scope, Operand<TFloat32> grads, Operand<T> image, Operand<TFloat32> boxes, Operand<TInt32> boxInd, Options... options) {
+  public static <T extends Tensor & TNumber> CropAndResizeGradBoxes create(Scope scope, Operand<TFloat32> grads, Operand<T> image, Operand<TFloat32> boxes, Operand<TInt32> boxInd, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("CropAndResizeGradBoxes", scope.makeOpName("CropAndResizeGradBoxes"));
     opBuilder.addInput(grads.asOutput());
     opBuilder.addInput(image.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/CropAndResizeGradImage.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/CropAndResizeGradImage.java
index 6a6415f879d..79d9f6a5fc0 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/CropAndResizeGradImage.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/CropAndResizeGradImage.java
@@ -22,6 +22,7 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -29,7 +30,6 @@
 import org.tensorflow.types.TFloat32;
 import org.tensorflow.types.TInt32;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes the gradient of the crop_and_resize op wrt the input image tensor.
@@ -37,7 +37,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "image")
-public final class CropAndResizeGradImage<T extends TNumber> extends RawOp implements Operand<T> {
+public final class CropAndResizeGradImage<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.image.CropAndResizeGradImage}
@@ -84,7 +84,7 @@ private Options() {
    * @return a new instance of CropAndResizeGradImage
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> CropAndResizeGradImage<T> create(Scope scope, Operand<TFloat32> grads, Operand<TFloat32> boxes, Operand<TInt32> boxInd, Operand<TInt32> imageSize, DataType<T> T, Options... options) {
+  public static <T extends Tensor & TNumber> CropAndResizeGradImage<T> create(Scope scope, Operand<TFloat32> grads, Operand<TFloat32> boxes, Operand<TInt32> boxInd, Operand<TInt32> imageSize, DataType<T> T, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("CropAndResizeGradImage", scope.makeOpName("CropAndResizeGradImage"));
     opBuilder.addInput(grads.asOutput());
     opBuilder.addInput(boxes.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/DecodePng.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/DecodePng.java
index 1d97bc54df7..2d8a35af092 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/DecodePng.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/DecodePng.java
@@ -22,6 +22,7 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -29,7 +30,6 @@
 import org.tensorflow.types.TString;
 import org.tensorflow.types.TUint8;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Decode a PNG-encoded image to a uint8 or uint16 tensor.
@@ -61,7 +61,7 @@
  * @param <T> data type for {@code image()} output
  */
 @Operator(group = "image")
-public final class DecodePng<T extends TNumber> extends RawOp implements Operand<T> {
+public final class DecodePng<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.image.DecodePng}
@@ -92,7 +92,7 @@ private Options() {
    * @return a new instance of DecodePng
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> DecodePng<T> create(Scope scope, Operand<TString> contents, DataType<T> dtype, Options... options) {
+  public static <T extends Tensor & TNumber> DecodePng<T> create(Scope scope, Operand<TString> contents, DataType<T> dtype, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("DecodePng", scope.makeOpName("DecodePng"));
     opBuilder.addInput(contents.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/DrawBoundingBoxes.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/DrawBoundingBoxes.java
index ba674376d07..e1e6b6d4760 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/DrawBoundingBoxes.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/DrawBoundingBoxes.java
@@ -21,13 +21,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TFloat32;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Draw bounding boxes on a batch of images.
@@ -47,7 +47,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "image")
-public final class DrawBoundingBoxes<T extends TNumber> extends RawOp implements Operand<T> {
+public final class DrawBoundingBoxes<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new DrawBoundingBoxes operation.
@@ -60,7 +60,7 @@ public final class DrawBoundingBoxes<T extends TNumber> extends RawOp implements
    * @return a new instance of DrawBoundingBoxes
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> DrawBoundingBoxes<T> create(Scope scope, Operand<T> images, Operand<TFloat32> boxes, Operand<TFloat32> colors) {
+  public static <T extends Tensor & TNumber> DrawBoundingBoxes<T> create(Scope scope, Operand<T> images, Operand<TFloat32> boxes, Operand<TFloat32> colors) {
     OperationBuilder opBuilder = scope.env().opBuilder("DrawBoundingBoxesV2", scope.makeOpName("DrawBoundingBoxes"));
     opBuilder.addInput(images.asOutput());
     opBuilder.addInput(boxes.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/EncodePng.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/EncodePng.java
index cf5b5a0ebc5..6a75551dc4c 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/EncodePng.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/EncodePng.java
@@ -21,13 +21,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TString;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * PNG-encode an image.
@@ -83,7 +83,7 @@ private Options() {
    * @return a new instance of EncodePng
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> EncodePng create(Scope scope, Operand<T> image, Options... options) {
+  public static <T extends Tensor & TNumber> EncodePng create(Scope scope, Operand<T> image, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("EncodePng", scope.makeOpName("EncodePng"));
     opBuilder.addInput(image.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ExtractImagePatches.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ExtractImagePatches.java
index d02d422a005..8a342524c91 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ExtractImagePatches.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ExtractImagePatches.java
@@ -22,11 +22,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Extract `patches` from `images` and put them in the "depth" output dimension.
@@ -34,7 +34,7 @@
  * @param <T> data type for {@code patches()} output
  */
 @Operator(group = "image")
-public final class ExtractImagePatches<T extends TType> extends RawOp implements Operand<T> {
+public final class ExtractImagePatches<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new ExtractImagePatches operation.
@@ -54,7 +54,7 @@ public final class ExtractImagePatches<T extends TType> extends RawOp implements
    * @return a new instance of ExtractImagePatches
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> ExtractImagePatches<T> create(Scope scope, Operand<T> images, List<Long> ksizes, List<Long> strides, List<Long> rates, String padding) {
+  public static <T extends Tensor> ExtractImagePatches<T> create(Scope scope, Operand<T> images, List<Long> ksizes, List<Long> strides, List<Long> rates, String padding) {
     OperationBuilder opBuilder = scope.env().opBuilder("ExtractImagePatches", scope.makeOpName("ExtractImagePatches"));
     opBuilder.addInput(images.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ExtractJpegShape.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ExtractJpegShape.java
index dfa8c4ea7b3..53258a1723f 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ExtractJpegShape.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ExtractJpegShape.java
@@ -22,6 +22,7 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -29,7 +30,6 @@
 import org.tensorflow.types.TInt32;
 import org.tensorflow.types.TString;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Extract the shape information of a JPEG-encoded image.
@@ -39,7 +39,7 @@
  * @param <T> data type for {@code imageShape()} output
  */
 @Operator(group = "image")
-public final class ExtractJpegShape<T extends TNumber> extends RawOp implements Operand<T> {
+public final class ExtractJpegShape<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new ExtractJpegShape operation.
@@ -51,7 +51,7 @@ public final class ExtractJpegShape<T extends TNumber> extends RawOp implements
    * @return a new instance of ExtractJpegShape
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> ExtractJpegShape<T> create(Scope scope, Operand<TString> contents, DataType<T> outputType) {
+  public static <T extends Tensor & TNumber> ExtractJpegShape<T> create(Scope scope, Operand<TString> contents, DataType<T> outputType) {
     OperationBuilder opBuilder = scope.env().opBuilder("ExtractJpegShape", scope.makeOpName("ExtractJpegShape"));
     opBuilder.addInput(contents.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/HsvToRgb.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/HsvToRgb.java
index fb752a83518..06102fdee52 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/HsvToRgb.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/HsvToRgb.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Convert one or more images from HSV to RGB.
@@ -40,7 +40,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "image")
-public final class HsvToRgb<T extends TNumber> extends RawOp implements Operand<T> {
+public final class HsvToRgb<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new HsvToRgb operation.
@@ -50,7 +50,7 @@ public final class HsvToRgb<T extends TNumber> extends RawOp implements Operand<
    * @return a new instance of HsvToRgb
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> HsvToRgb<T> create(Scope scope, Operand<T> images) {
+  public static <T extends Tensor & TNumber> HsvToRgb<T> create(Scope scope, Operand<T> images) {
     OperationBuilder opBuilder = scope.env().opBuilder("HSVToRGB", scope.makeOpName("HsvToRgb"));
     opBuilder.addInput(images.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ImageProjectiveTransformV2.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ImageProjectiveTransformV2.java
index 3363d6a9804..15155c535c4 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ImageProjectiveTransformV2.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ImageProjectiveTransformV2.java
@@ -21,6 +21,7 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -28,7 +29,6 @@
 import org.tensorflow.types.TFloat32;
 import org.tensorflow.types.TInt32;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Applies the given transform to each of the images.
@@ -41,7 +41,7 @@
  * 
  * @param <T> data type for {@code transformedImages()} output
  */
-public final class ImageProjectiveTransformV2<T extends TNumber> extends RawOp implements Operand<T> {
+public final class ImageProjectiveTransformV2<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.image.ImageProjectiveTransformV2}
@@ -76,7 +76,7 @@ private Options() {
    * @return a new instance of ImageProjectiveTransformV2
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> ImageProjectiveTransformV2<T> create(Scope scope, Operand<T> images, Operand<TFloat32> transforms, Operand<TInt32> outputShape, String interpolation, Options... options) {
+  public static <T extends Tensor & TNumber> ImageProjectiveTransformV2<T> create(Scope scope, Operand<T> images, Operand<TFloat32> transforms, Operand<TInt32> outputShape, String interpolation, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ImageProjectiveTransformV2", scope.makeOpName("ImageProjectiveTransformV2"));
     opBuilder.addInput(images.asOutput());
     opBuilder.addInput(transforms.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/NonMaxSuppression.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/NonMaxSuppression.java
index 0745299e34f..77639f7bb60 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/NonMaxSuppression.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/NonMaxSuppression.java
@@ -21,13 +21,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Greedily selects a subset of bounding boxes in descending order of score,
@@ -58,7 +58,7 @@
  * @param <T> data type for {@code selectedScores()} output
  */
 @Operator(group = "image")
-public final class NonMaxSuppression<T extends TNumber> extends RawOp {
+public final class NonMaxSuppression<T extends Tensor & TNumber> extends RawOp {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.image.NonMaxSuppression}
@@ -100,7 +100,7 @@ private Options() {
    * @return a new instance of NonMaxSuppression
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> NonMaxSuppression<T> create(Scope scope, Operand<T> boxes, Operand<T> scores, Operand<TInt32> maxOutputSize, Operand<T> iouThreshold, Operand<T> scoreThreshold, Operand<T> softNmsSigma, Options... options) {
+  public static <T extends Tensor & TNumber> NonMaxSuppression<T> create(Scope scope, Operand<T> boxes, Operand<T> scores, Operand<TInt32> maxOutputSize, Operand<T> iouThreshold, Operand<T> scoreThreshold, Operand<T> softNmsSigma, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("NonMaxSuppressionV5", scope.makeOpName("NonMaxSuppression"));
     opBuilder.addInput(boxes.asOutput());
     opBuilder.addInput(scores.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/QuantizedResizeBilinear.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/QuantizedResizeBilinear.java
index 7eb6d8a3a92..9da9e33b053 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/QuantizedResizeBilinear.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/QuantizedResizeBilinear.java
@@ -21,13 +21,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TFloat32;
 import org.tensorflow.types.TInt32;
-import org.tensorflow.types.family.TType;
 
 /**
  * Resize quantized `images` to `size` using quantized bilinear interpolation.
@@ -37,7 +37,7 @@
  * @param <T> data type for {@code resizedImages()} output
  */
 @Operator(group = "image")
-public final class QuantizedResizeBilinear<T extends TType> extends RawOp {
+public final class QuantizedResizeBilinear<T extends Tensor> extends RawOp {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.image.QuantizedResizeBilinear}
@@ -81,7 +81,7 @@ private Options() {
    * @return a new instance of QuantizedResizeBilinear
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> QuantizedResizeBilinear<T> create(Scope scope, Operand<T> images, Operand<TInt32> size, Operand<TFloat32> min, Operand<TFloat32> max, Options... options) {
+  public static <T extends Tensor> QuantizedResizeBilinear<T> create(Scope scope, Operand<T> images, Operand<TInt32> size, Operand<TFloat32> min, Operand<TFloat32> max, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("QuantizedResizeBilinear", scope.makeOpName("QuantizedResizeBilinear"));
     opBuilder.addInput(images.asOutput());
     opBuilder.addInput(size.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/RandomCrop.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/RandomCrop.java
index 47d03d63187..8c5dabc23ed 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/RandomCrop.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/RandomCrop.java
@@ -21,13 +21,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Randomly crop `image`.
@@ -42,7 +42,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "image")
-public final class RandomCrop<T extends TNumber> extends RawOp implements Operand<T> {
+public final class RandomCrop<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.image.RandomCrop}
@@ -84,7 +84,7 @@ private Options() {
    * @return a new instance of RandomCrop
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> RandomCrop<T> create(Scope scope, Operand<T> image, Operand<TInt64> size, Options... options) {
+  public static <T extends Tensor & TNumber> RandomCrop<T> create(Scope scope, Operand<T> image, Operand<TInt64> size, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("RandomCrop", scope.makeOpName("RandomCrop"));
     opBuilder.addInput(image.asOutput());
     opBuilder.addInput(size.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ResizeArea.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ResizeArea.java
index 60d1e473ef0..da1a126950f 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ResizeArea.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ResizeArea.java
@@ -21,6 +21,7 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -28,7 +29,6 @@
 import org.tensorflow.types.TFloat32;
 import org.tensorflow.types.TInt32;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Resize `images` to `size` using area interpolation.
@@ -79,7 +79,7 @@ private Options() {
    * @return a new instance of ResizeArea
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> ResizeArea create(Scope scope, Operand<T> images, Operand<TInt32> size, Options... options) {
+  public static <T extends Tensor & TNumber> ResizeArea create(Scope scope, Operand<T> images, Operand<TInt32> size, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ResizeArea", scope.makeOpName("ResizeArea"));
     opBuilder.addInput(images.asOutput());
     opBuilder.addInput(size.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ResizeBicubic.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ResizeBicubic.java
index 2519eaa6ea9..750b4de7cb4 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ResizeBicubic.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ResizeBicubic.java
@@ -21,6 +21,7 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -28,7 +29,6 @@
 import org.tensorflow.types.TFloat32;
 import org.tensorflow.types.TInt32;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Resize `images` to `size` using bicubic interpolation.
@@ -78,7 +78,7 @@ private Options() {
    * @return a new instance of ResizeBicubic
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> ResizeBicubic create(Scope scope, Operand<T> images, Operand<TInt32> size, Options... options) {
+  public static <T extends Tensor & TNumber> ResizeBicubic create(Scope scope, Operand<T> images, Operand<TInt32> size, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ResizeBicubic", scope.makeOpName("ResizeBicubic"));
     opBuilder.addInput(images.asOutput());
     opBuilder.addInput(size.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ResizeBicubicGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ResizeBicubicGrad.java
index 96296cdf33f..0fb80f7f2c8 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ResizeBicubicGrad.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ResizeBicubicGrad.java
@@ -21,20 +21,20 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TFloat32;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes the gradient of bicubic interpolation.
  * 
  * @param <T> data type for {@code output()} output
  */
-public final class ResizeBicubicGrad<T extends TNumber> extends RawOp implements Operand<T> {
+public final class ResizeBicubicGrad<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.image.ResizeBicubicGrad}
@@ -76,7 +76,7 @@ private Options() {
    * @return a new instance of ResizeBicubicGrad
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> ResizeBicubicGrad<T> create(Scope scope, Operand<TFloat32> grads, Operand<T> originalImage, Options... options) {
+  public static <T extends Tensor & TNumber> ResizeBicubicGrad<T> create(Scope scope, Operand<TFloat32> grads, Operand<T> originalImage, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ResizeBicubicGrad", scope.makeOpName("ResizeBicubicGrad"));
     opBuilder.addInput(grads.asOutput());
     opBuilder.addInput(originalImage.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ResizeBilinear.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ResizeBilinear.java
index d0c28fed013..65d21cd2dcf 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ResizeBilinear.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ResizeBilinear.java
@@ -21,6 +21,7 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -28,7 +29,6 @@
 import org.tensorflow.types.TFloat32;
 import org.tensorflow.types.TInt32;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Resize `images` to `size` using bilinear interpolation.
@@ -78,7 +78,7 @@ private Options() {
    * @return a new instance of ResizeBilinear
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> ResizeBilinear create(Scope scope, Operand<T> images, Operand<TInt32> size, Options... options) {
+  public static <T extends Tensor & TNumber> ResizeBilinear create(Scope scope, Operand<T> images, Operand<TInt32> size, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ResizeBilinear", scope.makeOpName("ResizeBilinear"));
     opBuilder.addInput(images.asOutput());
     opBuilder.addInput(size.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ResizeBilinearGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ResizeBilinearGrad.java
index b8aef51ba30..cd643e7252b 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ResizeBilinearGrad.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ResizeBilinearGrad.java
@@ -21,20 +21,20 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TFloat32;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes the gradient of bilinear interpolation.
  * 
  * @param <T> data type for {@code output()} output
  */
-public final class ResizeBilinearGrad<T extends TNumber> extends RawOp implements Operand<T> {
+public final class ResizeBilinearGrad<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.image.ResizeBilinearGrad}
@@ -76,7 +76,7 @@ private Options() {
    * @return a new instance of ResizeBilinearGrad
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> ResizeBilinearGrad<T> create(Scope scope, Operand<TFloat32> grads, Operand<T> originalImage, Options... options) {
+  public static <T extends Tensor & TNumber> ResizeBilinearGrad<T> create(Scope scope, Operand<TFloat32> grads, Operand<T> originalImage, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ResizeBilinearGrad", scope.makeOpName("ResizeBilinearGrad"));
     opBuilder.addInput(grads.asOutput());
     opBuilder.addInput(originalImage.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ResizeNearestNeighbor.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ResizeNearestNeighbor.java
index dd7865841de..5ee41454f58 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ResizeNearestNeighbor.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ResizeNearestNeighbor.java
@@ -21,13 +21,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Resize `images` to `size` using nearest neighbor interpolation.
@@ -35,7 +35,7 @@
  * @param <T> data type for {@code resizedImages()} output
  */
 @Operator(group = "image")
-public final class ResizeNearestNeighbor<T extends TNumber> extends RawOp implements Operand<T> {
+public final class ResizeNearestNeighbor<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.image.ResizeNearestNeighbor}
@@ -77,7 +77,7 @@ private Options() {
    * @return a new instance of ResizeNearestNeighbor
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> ResizeNearestNeighbor<T> create(Scope scope, Operand<T> images, Operand<TInt32> size, Options... options) {
+  public static <T extends Tensor & TNumber> ResizeNearestNeighbor<T> create(Scope scope, Operand<T> images, Operand<TInt32> size, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ResizeNearestNeighbor", scope.makeOpName("ResizeNearestNeighbor"));
     opBuilder.addInput(images.asOutput());
     opBuilder.addInput(size.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ResizeNearestNeighborGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ResizeNearestNeighborGrad.java
index 4897c0f9dc6..dafa83b74c6 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ResizeNearestNeighborGrad.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ResizeNearestNeighborGrad.java
@@ -21,20 +21,20 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes the gradient of nearest neighbor interpolation.
  * 
  * @param <T> data type for {@code output()} output
  */
-public final class ResizeNearestNeighborGrad<T extends TNumber> extends RawOp implements Operand<T> {
+public final class ResizeNearestNeighborGrad<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.image.ResizeNearestNeighborGrad}
@@ -76,7 +76,7 @@ private Options() {
    * @return a new instance of ResizeNearestNeighborGrad
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> ResizeNearestNeighborGrad<T> create(Scope scope, Operand<T> grads, Operand<TInt32> size, Options... options) {
+  public static <T extends Tensor & TNumber> ResizeNearestNeighborGrad<T> create(Scope scope, Operand<T> grads, Operand<TInt32> size, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ResizeNearestNeighborGrad", scope.makeOpName("ResizeNearestNeighborGrad"));
     opBuilder.addInput(grads.asOutput());
     opBuilder.addInput(size.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/RgbToHsv.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/RgbToHsv.java
index acee7a6061e..87d08b14b08 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/RgbToHsv.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/RgbToHsv.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Converts one or more images from RGB to HSV.
@@ -54,7 +54,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "image")
-public final class RgbToHsv<T extends TNumber> extends RawOp implements Operand<T> {
+public final class RgbToHsv<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new RgbToHsv operation.
@@ -64,7 +64,7 @@ public final class RgbToHsv<T extends TNumber> extends RawOp implements Operand<
    * @return a new instance of RgbToHsv
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> RgbToHsv<T> create(Scope scope, Operand<T> images) {
+  public static <T extends Tensor & TNumber> RgbToHsv<T> create(Scope scope, Operand<T> images) {
     OperationBuilder opBuilder = scope.env().opBuilder("RGBToHSV", scope.makeOpName("RgbToHsv"));
     opBuilder.addInput(images.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/SampleDistortedBoundingBox.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/SampleDistortedBoundingBox.java
index 7f83b2ee84c..e9c391b177e 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/SampleDistortedBoundingBox.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/SampleDistortedBoundingBox.java
@@ -22,13 +22,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TFloat32;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Generate a single randomly distorted bounding box for an image.
@@ -73,7 +73,7 @@
  * @param <T> data type for {@code begin()} output
  */
 @Operator(group = "image")
-public final class SampleDistortedBoundingBox<T extends TNumber> extends RawOp {
+public final class SampleDistortedBoundingBox<T extends Tensor & TNumber> extends RawOp {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.image.SampleDistortedBoundingBox}
@@ -162,7 +162,7 @@ private Options() {
    * @return a new instance of SampleDistortedBoundingBox
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> SampleDistortedBoundingBox<T> create(Scope scope, Operand<T> imageSize, Operand<TFloat32> boundingBoxes, Operand<TFloat32> minObjectCovered, Options... options) {
+  public static <T extends Tensor & TNumber> SampleDistortedBoundingBox<T> create(Scope scope, Operand<T> imageSize, Operand<TFloat32> boundingBoxes, Operand<TFloat32> minObjectCovered, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("SampleDistortedBoundingBoxV2", scope.makeOpName("SampleDistortedBoundingBox"));
     opBuilder.addInput(imageSize.asOutput());
     opBuilder.addInput(boundingBoxes.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ScaleAndTranslate.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ScaleAndTranslate.java
index 26bf544be2e..e0c7f3bc22a 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ScaleAndTranslate.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ScaleAndTranslate.java
@@ -21,6 +21,7 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -28,7 +29,6 @@
 import org.tensorflow.types.TFloat32;
 import org.tensorflow.types.TInt32;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  */
@@ -75,7 +75,7 @@ private Options() {
    * @return a new instance of ScaleAndTranslate
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> ScaleAndTranslate create(Scope scope, Operand<T> images, Operand<TInt32> size, Operand<TFloat32> scale, Operand<TFloat32> translation, Options... options) {
+  public static <T extends Tensor & TNumber> ScaleAndTranslate create(Scope scope, Operand<T> images, Operand<TInt32> size, Operand<TFloat32> scale, Operand<TFloat32> translation, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ScaleAndTranslate", scope.makeOpName("ScaleAndTranslate"));
     opBuilder.addInput(images.asOutput());
     opBuilder.addInput(size.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ScaleAndTranslateGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ScaleAndTranslateGrad.java
index 4b8d9b0fc4b..b02e33bbfe0 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ScaleAndTranslateGrad.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ScaleAndTranslateGrad.java
@@ -21,18 +21,18 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TFloat32;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * @param <T> data type for {@code output()} output
  */
-public final class ScaleAndTranslateGrad<T extends TNumber> extends RawOp implements Operand<T> {
+public final class ScaleAndTranslateGrad<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.image.ScaleAndTranslateGrad}
@@ -74,7 +74,7 @@ private Options() {
    * @return a new instance of ScaleAndTranslateGrad
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> ScaleAndTranslateGrad<T> create(Scope scope, Operand<T> grads, Operand<T> originalImage, Operand<TFloat32> scale, Operand<TFloat32> translation, Options... options) {
+  public static <T extends Tensor & TNumber> ScaleAndTranslateGrad<T> create(Scope scope, Operand<T> grads, Operand<T> originalImage, Operand<TFloat32> scale, Operand<TFloat32> translation, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ScaleAndTranslateGrad", scope.makeOpName("ScaleAndTranslateGrad"));
     opBuilder.addInput(grads.asOutput());
     opBuilder.addInput(originalImage.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/DecodeCsv.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/DecodeCsv.java
index 6b417cb8717..3eb38c17efe 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/DecodeCsv.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/DecodeCsv.java
@@ -24,13 +24,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.Operands;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TString;
-import org.tensorflow.types.family.TType;
 
 /**
  * Convert CSV records to tensors. Each column maps to one tensor.
@@ -40,7 +40,7 @@
  * Note that we allow leading and trailing spaces with int or float field.
  */
 @Operator(group = "io")
-public final class DecodeCsv extends RawOp implements Iterable<Operand<TType>> {
+public final class DecodeCsv extends RawOp implements Iterable<Operand<Tensor>> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.io.DecodeCsv}
@@ -170,7 +170,7 @@ public List<Output<?>> output() {
   
   @Override
   @SuppressWarnings({"rawtypes", "unchecked"})
-  public Iterator<Operand<TType>> iterator() {
+  public Iterator<Operand<Tensor>> iterator() {
     return (Iterator) output.iterator();
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/DecodePaddedRaw.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/DecodePaddedRaw.java
index 820b0077ba5..ed4845e4216 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/DecodePaddedRaw.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/DecodePaddedRaw.java
@@ -22,6 +22,7 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -29,7 +30,6 @@
 import org.tensorflow.types.TInt32;
 import org.tensorflow.types.TString;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Reinterpret the bytes of a string as a vector of numbers.
@@ -37,7 +37,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "io")
-public final class DecodePaddedRaw<T extends TNumber> extends RawOp implements Operand<T> {
+public final class DecodePaddedRaw<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.io.DecodePaddedRaw}
@@ -71,7 +71,7 @@ private Options() {
    * @return a new instance of DecodePaddedRaw
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> DecodePaddedRaw<T> create(Scope scope, Operand<TString> inputBytes, Operand<TInt32> fixedLength, DataType<T> outType, Options... options) {
+  public static <T extends Tensor & TNumber> DecodePaddedRaw<T> create(Scope scope, Operand<TString> inputBytes, Operand<TInt32> fixedLength, DataType<T> outType, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("DecodePaddedRaw", scope.makeOpName("DecodePaddedRaw"));
     opBuilder.addInput(inputBytes.asOutput());
     opBuilder.addInput(fixedLength.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/DecodeRaw.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/DecodeRaw.java
index dfdabfdb466..922d7dbfb44 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/DecodeRaw.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/DecodeRaw.java
@@ -22,12 +22,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TString;
-import org.tensorflow.types.family.TType;
 
 /**
  * Reinterpret the bytes of a string as a vector of numbers.
@@ -35,7 +35,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "io")
-public final class DecodeRaw<T extends TType> extends RawOp implements Operand<T> {
+public final class DecodeRaw<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.io.DecodeRaw}
@@ -68,7 +68,7 @@ private Options() {
    * @return a new instance of DecodeRaw
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> DecodeRaw<T> create(Scope scope, Operand<TString> bytes, DataType<T> outType, Options... options) {
+  public static <T extends Tensor> DecodeRaw<T> create(Scope scope, Operand<TString> bytes, DataType<T> outType, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("DecodeRaw", scope.makeOpName("DecodeRaw"));
     opBuilder.addInput(bytes.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/DeserializeManySparse.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/DeserializeManySparse.java
index 8582681c073..c969ec7da77 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/DeserializeManySparse.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/DeserializeManySparse.java
@@ -22,13 +22,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
 import org.tensorflow.types.TString;
-import org.tensorflow.types.family.TType;
 
 /**
  * Deserialize and concatenate `SparseTensors` from a serialized minibatch.
@@ -78,7 +78,7 @@
  * @param <T> data type for {@code sparseValues()} output
  */
 @Operator(group = "io")
-public final class DeserializeManySparse<T extends TType> extends RawOp {
+public final class DeserializeManySparse<T extends Tensor> extends RawOp {
   
   /**
    * Factory method to create a class wrapping a new DeserializeManySparse operation.
@@ -90,7 +90,7 @@ public final class DeserializeManySparse<T extends TType> extends RawOp {
    * @return a new instance of DeserializeManySparse
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> DeserializeManySparse<T> create(Scope scope, Operand<TString> serializedSparse, DataType<T> dtype) {
+  public static <T extends Tensor> DeserializeManySparse<T> create(Scope scope, Operand<TString> serializedSparse, DataType<T> dtype) {
     OperationBuilder opBuilder = scope.env().opBuilder("DeserializeManySparse", scope.makeOpName("DeserializeManySparse"));
     opBuilder.addInput(serializedSparse.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/FifoQueue.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/FifoQueue.java
index 842985b49e9..1e897d50624 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/FifoQueue.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/FifoQueue.java
@@ -23,18 +23,18 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * A queue that produces elements in first-in first-out order.
  */
 @Operator(group = "io")
-public final class FifoQueue extends RawOp implements Operand<TType> {
+public final class FifoQueue extends RawOp implements Operand<Tensor> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.io.FifoQueue}
@@ -171,8 +171,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) handle;
+  public Output<Tensor> asOutput() {
+    return (Output<Tensor>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/FixedLengthRecordReader.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/FixedLengthRecordReader.java
index 3678b8190b9..ba1a213dcd2 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/FixedLengthRecordReader.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/FixedLengthRecordReader.java
@@ -21,17 +21,17 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * A Reader that outputs fixed-length records from a file.
  */
 @Operator(group = "io")
-public final class FixedLengthRecordReader extends RawOp implements Operand<TType> {
+public final class FixedLengthRecordReader extends RawOp implements Operand<Tensor> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.io.FixedLengthRecordReader}
@@ -194,8 +194,8 @@ public Output<?> readerHandle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) readerHandle;
+  public Output<Tensor> asOutput() {
+    return (Output<Tensor>) readerHandle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/IdentityReader.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/IdentityReader.java
index 99ae027a82e..f6af2da4b47 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/IdentityReader.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/IdentityReader.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * A Reader that outputs the queued work as both the key and value.
@@ -34,7 +34,7 @@
  * work string and output (work, work).
  */
 @Operator(group = "io")
-public final class IdentityReader extends RawOp implements Operand<TType> {
+public final class IdentityReader extends RawOp implements Operand<Tensor> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.io.IdentityReader}
@@ -115,8 +115,8 @@ public Output<?> readerHandle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) readerHandle;
+  public Output<Tensor> asOutput() {
+    return (Output<Tensor>) readerHandle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/PaddingFifoQueue.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/PaddingFifoQueue.java
index b72d8188038..712be3c7660 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/PaddingFifoQueue.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/PaddingFifoQueue.java
@@ -23,12 +23,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * A queue that produces elements in first-in first-out order.
@@ -38,7 +38,7 @@
  * size of any given element in the minibatch.  See below for details.
  */
 @Operator(group = "io")
-public final class PaddingFifoQueue extends RawOp implements Operand<TType> {
+public final class PaddingFifoQueue extends RawOp implements Operand<Tensor> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.io.PaddingFifoQueue}
@@ -183,8 +183,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) handle;
+  public Output<Tensor> asOutput() {
+    return (Output<Tensor>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ParseTensor.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ParseTensor.java
index bebdc1b4419..7d47f9380f6 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ParseTensor.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ParseTensor.java
@@ -22,12 +22,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TString;
-import org.tensorflow.types.family.TType;
 
 /**
  * Transforms a serialized tensorflow.TensorProto proto into a Tensor.
@@ -35,7 +35,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "io")
-public final class ParseTensor<T extends TType> extends RawOp implements Operand<T> {
+public final class ParseTensor<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new ParseTensor operation.
@@ -47,7 +47,7 @@ public final class ParseTensor<T extends TType> extends RawOp implements Operand
    * @return a new instance of ParseTensor
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> ParseTensor<T> create(Scope scope, Operand<TString> serialized, DataType<T> outType) {
+  public static <T extends Tensor> ParseTensor<T> create(Scope scope, Operand<TString> serialized, DataType<T> outType) {
     OperationBuilder opBuilder = scope.env().opBuilder("ParseTensor", scope.makeOpName("ParseTensor"));
     opBuilder.addInput(serialized.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/PriorityQueue.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/PriorityQueue.java
index 569d4ae7eb3..71d5aee9591 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/PriorityQueue.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/PriorityQueue.java
@@ -23,12 +23,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * A queue that produces elements sorted by the first component value.
@@ -40,7 +40,7 @@
  * entry in their input (resp. output) lists.
  */
 @Operator(group = "io")
-public final class PriorityQueue extends RawOp implements Operand<TType> {
+public final class PriorityQueue extends RawOp implements Operand<Tensor> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.io.PriorityQueue}
@@ -157,8 +157,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) handle;
+  public Output<Tensor> asOutput() {
+    return (Output<Tensor>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/QueueDequeue.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/QueueDequeue.java
index b74fb3ad3e8..52a1f40d890 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/QueueDequeue.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/QueueDequeue.java
@@ -25,11 +25,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Dequeues a tuple of one or more tensors from the given queue.
@@ -42,7 +42,7 @@
  * has been dequeued (or 'timeout_ms' elapses, if specified).
  */
 @Operator(group = "io")
-public final class QueueDequeue extends RawOp implements Iterable<Operand<TType>> {
+public final class QueueDequeue extends RawOp implements Iterable<Operand<Tensor>> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.io.QueueDequeue}
@@ -112,7 +112,7 @@ public List<Output<?>> components() {
   
   @Override
   @SuppressWarnings({"rawtypes", "unchecked"})
-  public Iterator<Operand<TType>> iterator() {
+  public Iterator<Operand<Tensor>> iterator() {
     return (Iterator) components.iterator();
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/QueueDequeueMany.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/QueueDequeueMany.java
index 2906fe6543c..770baa857d3 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/QueueDequeueMany.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/QueueDequeueMany.java
@@ -25,12 +25,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
-import org.tensorflow.types.family.TType;
 
 /**
  * Dequeues `n` tuples of one or more tensors from the given queue.
@@ -50,7 +50,7 @@
  * have been dequeued (or 'timeout_ms' elapses, if specified).
  */
 @Operator(group = "io")
-public final class QueueDequeueMany extends RawOp implements Iterable<Operand<TType>> {
+public final class QueueDequeueMany extends RawOp implements Iterable<Operand<Tensor>> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.io.QueueDequeueMany}
@@ -122,7 +122,7 @@ public List<Output<?>> components() {
   
   @Override
   @SuppressWarnings({"rawtypes", "unchecked"})
-  public Iterator<Operand<TType>> iterator() {
+  public Iterator<Operand<Tensor>> iterator() {
     return (Iterator) components.iterator();
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/QueueDequeueUpTo.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/QueueDequeueUpTo.java
index 51af788104c..7df58e083fd 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/QueueDequeueUpTo.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/QueueDequeueUpTo.java
@@ -25,12 +25,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
-import org.tensorflow.types.family.TType;
 
 /**
  * Dequeues `n` tuples of one or more tensors from the given queue.
@@ -54,7 +54,7 @@
  * component of the dequeued tuple.
  */
 @Operator(group = "io")
-public final class QueueDequeueUpTo extends RawOp implements Iterable<Operand<TType>> {
+public final class QueueDequeueUpTo extends RawOp implements Iterable<Operand<Tensor>> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.io.QueueDequeueUpTo}
@@ -126,7 +126,7 @@ public List<Output<?>> components() {
   
   @Override
   @SuppressWarnings({"rawtypes", "unchecked"})
-  public Iterator<Operand<TType>> iterator() {
+  public Iterator<Operand<Tensor>> iterator() {
     return (Iterator) components.iterator();
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/RandomShuffleQueue.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/RandomShuffleQueue.java
index cdde1debc95..cf81915b80a 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/RandomShuffleQueue.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/RandomShuffleQueue.java
@@ -23,18 +23,18 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * A queue that randomizes the order of elements.
  */
 @Operator(group = "io")
-public final class RandomShuffleQueue extends RawOp implements Operand<TType> {
+public final class RandomShuffleQueue extends RawOp implements Operand<Tensor> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.io.RandomShuffleQueue}
@@ -234,8 +234,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) handle;
+  public Output<Tensor> asOutput() {
+    return (Output<Tensor>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/SerializeManySparse.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/SerializeManySparse.java
index 0afbdcd6edd..afab861c445 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/SerializeManySparse.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/SerializeManySparse.java
@@ -22,13 +22,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
 import org.tensorflow.types.TString;
-import org.tensorflow.types.family.TType;
 
 /**
  * Serialize an `N`-minibatch `SparseTensor` into an `[N, 3]` `Tensor` object.
@@ -44,7 +44,7 @@
  * @param <U> data type for {@code serializedSparse()} output
  */
 @Operator(group = "io")
-public final class SerializeManySparse<U extends TType> extends RawOp implements Operand<U> {
+public final class SerializeManySparse<U extends Tensor> extends RawOp implements Operand<U> {
   
   /**
    * Factory method to create a class wrapping a new SerializeManySparse operation.
@@ -58,7 +58,7 @@ public final class SerializeManySparse<U extends TType> extends RawOp implements
    * @return a new instance of SerializeManySparse
    */
   @Endpoint(describeByClass = true)
-  public static <U extends TType, T extends TType> SerializeManySparse<U> create(Scope scope, Operand<TInt64> sparseIndices, Operand<T> sparseValues, Operand<TInt64> sparseShape, DataType<U> outType) {
+  public static <U extends Tensor, T extends Tensor> SerializeManySparse<U> create(Scope scope, Operand<TInt64> sparseIndices, Operand<T> sparseValues, Operand<TInt64> sparseShape, DataType<U> outType) {
     OperationBuilder opBuilder = scope.env().opBuilder("SerializeManySparse", scope.makeOpName("SerializeManySparse"));
     opBuilder.addInput(sparseIndices.asOutput());
     opBuilder.addInput(sparseValues.asOutput());
@@ -78,7 +78,7 @@ public static <U extends TType, T extends TType> SerializeManySparse<U> create(S
    * @return a new instance of SerializeManySparse
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> SerializeManySparse<TString> create(Scope scope, Operand<TInt64> sparseIndices, Operand<T> sparseValues, Operand<TInt64> sparseShape) {
+  public static <T extends Tensor> SerializeManySparse<TString> create(Scope scope, Operand<TInt64> sparseIndices, Operand<T> sparseValues, Operand<TInt64> sparseShape) {
     return create(scope, sparseIndices, sparseValues, sparseShape, TString.DTYPE);
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/SerializeSparse.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/SerializeSparse.java
index 2d8fc7838e4..744d573981d 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/SerializeSparse.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/SerializeSparse.java
@@ -22,13 +22,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
 import org.tensorflow.types.TString;
-import org.tensorflow.types.family.TType;
 
 /**
  * Serialize a `SparseTensor` into a `[3]` `Tensor` object.
@@ -36,7 +36,7 @@
  * @param <U> data type for {@code serializedSparse()} output
  */
 @Operator(group = "io")
-public final class SerializeSparse<U extends TType> extends RawOp implements Operand<U> {
+public final class SerializeSparse<U extends Tensor> extends RawOp implements Operand<U> {
   
   /**
    * Factory method to create a class wrapping a new SerializeSparse operation.
@@ -50,7 +50,7 @@ public final class SerializeSparse<U extends TType> extends RawOp implements Ope
    * @return a new instance of SerializeSparse
    */
   @Endpoint(describeByClass = true)
-  public static <U extends TType, T extends TType> SerializeSparse<U> create(Scope scope, Operand<TInt64> sparseIndices, Operand<T> sparseValues, Operand<TInt64> sparseShape, DataType<U> outType) {
+  public static <U extends Tensor, T extends Tensor> SerializeSparse<U> create(Scope scope, Operand<TInt64> sparseIndices, Operand<T> sparseValues, Operand<TInt64> sparseShape, DataType<U> outType) {
     OperationBuilder opBuilder = scope.env().opBuilder("SerializeSparse", scope.makeOpName("SerializeSparse"));
     opBuilder.addInput(sparseIndices.asOutput());
     opBuilder.addInput(sparseValues.asOutput());
@@ -70,7 +70,7 @@ public static <U extends TType, T extends TType> SerializeSparse<U> create(Scope
    * @return a new instance of SerializeSparse
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> SerializeSparse<TString> create(Scope scope, Operand<TInt64> sparseIndices, Operand<T> sparseValues, Operand<TInt64> sparseShape) {
+  public static <T extends Tensor> SerializeSparse<TString> create(Scope scope, Operand<TInt64> sparseIndices, Operand<T> sparseValues, Operand<TInt64> sparseShape) {
     return create(scope, sparseIndices, sparseValues, sparseShape, TString.DTYPE);
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/SerializeTensor.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/SerializeTensor.java
index d68320b85e3..cf5be111b9b 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/SerializeTensor.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/SerializeTensor.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TString;
-import org.tensorflow.types.family.TType;
 
 /**
  * Transforms a Tensor into a serialized TensorProto proto.
@@ -42,7 +42,7 @@ public final class SerializeTensor extends RawOp implements Operand<TString> {
    * @return a new instance of SerializeTensor
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> SerializeTensor create(Scope scope, Operand<T> tensor) {
+  public static <T extends Tensor> SerializeTensor create(Scope scope, Operand<T> tensor) {
     OperationBuilder opBuilder = scope.env().opBuilder("SerializeTensor", scope.makeOpName("SerializeTensor"));
     opBuilder.addInput(tensor.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/TextLineReader.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/TextLineReader.java
index 90bf079d2cf..d52738f708a 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/TextLineReader.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/TextLineReader.java
@@ -21,17 +21,17 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * A Reader that outputs the lines of a file delimited by '\n'.
  */
 @Operator(group = "io")
-public final class TextLineReader extends RawOp implements Operand<TType> {
+public final class TextLineReader extends RawOp implements Operand<Tensor> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.io.TextLineReader}
@@ -131,8 +131,8 @@ public Output<?> readerHandle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) readerHandle;
+  public Output<Tensor> asOutput() {
+    return (Output<Tensor>) readerHandle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/TfRecordReader.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/TfRecordReader.java
index a12145e7972..0adaeafa879 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/TfRecordReader.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/TfRecordReader.java
@@ -21,17 +21,17 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * A Reader that outputs the records from a TensorFlow Records file.
  */
 @Operator(group = "io")
-public final class TfRecordReader extends RawOp implements Operand<TType> {
+public final class TfRecordReader extends RawOp implements Operand<Tensor> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.io.TfRecordReader}
@@ -131,8 +131,8 @@ public Output<?> readerHandle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) readerHandle;
+  public Output<Tensor> asOutput() {
+    return (Output<Tensor>) readerHandle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/WholeFileReader.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/WholeFileReader.java
index 8cab4f9b09e..83480ec07b0 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/WholeFileReader.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/WholeFileReader.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * A Reader that outputs the entire contents of a file as a value.
@@ -34,7 +34,7 @@
  * be a filename (key) and the contents of that file (value).
  */
 @Operator(group = "io")
-public final class WholeFileReader extends RawOp implements Operand<TType> {
+public final class WholeFileReader extends RawOp implements Operand<Tensor> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.io.WholeFileReader}
@@ -115,8 +115,8 @@ public Output<?> readerHandle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) readerHandle;
+  public Output<Tensor> asOutput() {
+    return (Output<Tensor>) readerHandle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BandPart.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BandPart.java
index 11dbb5662a9..b6b5f90336a 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BandPart.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BandPart.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Copy a tensor setting everything outside a central band in each innermost matrix to zero.
@@ -70,7 +70,7 @@
  * @param <T> data type for {@code band()} output
  */
 @Operator(group = "linalg")
-public final class BandPart<T extends TType> extends RawOp implements Operand<T> {
+public final class BandPart<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new BandPart operation.
@@ -84,7 +84,7 @@ public final class BandPart<T extends TType> extends RawOp implements Operand<T>
    * @return a new instance of BandPart
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> BandPart<T> create(Scope scope, Operand<T> input, Operand<U> numLower, Operand<U> numUpper) {
+  public static <T extends Tensor, U extends Tensor & TNumber> BandPart<T> create(Scope scope, Operand<T> input, Operand<U> numLower, Operand<U> numUpper) {
     OperationBuilder opBuilder = scope.env().opBuilder("MatrixBandPart", scope.makeOpName("BandPart"));
     opBuilder.addInput(input.asOutput());
     opBuilder.addInput(numLower.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchCholesky.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchCholesky.java
index 733139f4359..08eeabdaa69 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchCholesky.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchCholesky.java
@@ -21,18 +21,18 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "linalg")
-public final class BatchCholesky<T extends TNumber> extends RawOp implements Operand<T> {
+public final class BatchCholesky<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new BatchCholesky operation.
@@ -42,7 +42,7 @@ public final class BatchCholesky<T extends TNumber> extends RawOp implements Ope
    * @return a new instance of BatchCholesky
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> BatchCholesky<T> create(Scope scope, Operand<T> input) {
+  public static <T extends Tensor & TNumber> BatchCholesky<T> create(Scope scope, Operand<T> input) {
     OperationBuilder opBuilder = scope.env().opBuilder("BatchCholesky", scope.makeOpName("BatchCholesky"));
     opBuilder.addInput(input.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchCholeskyGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchCholeskyGrad.java
index d56f45e8abf..47184b8de81 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchCholeskyGrad.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchCholeskyGrad.java
@@ -21,18 +21,18 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "linalg")
-public final class BatchCholeskyGrad<T extends TNumber> extends RawOp implements Operand<T> {
+public final class BatchCholeskyGrad<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new BatchCholeskyGrad operation.
@@ -43,7 +43,7 @@ public final class BatchCholeskyGrad<T extends TNumber> extends RawOp implements
    * @return a new instance of BatchCholeskyGrad
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> BatchCholeskyGrad<T> create(Scope scope, Operand<T> l, Operand<T> grad) {
+  public static <T extends Tensor & TNumber> BatchCholeskyGrad<T> create(Scope scope, Operand<T> l, Operand<T> grad) {
     OperationBuilder opBuilder = scope.env().opBuilder("BatchCholeskyGrad", scope.makeOpName("BatchCholeskyGrad"));
     opBuilder.addInput(l.asOutput());
     opBuilder.addInput(grad.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixBandPart.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixBandPart.java
index 8c990a8a71c..f857dc7588d 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixBandPart.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixBandPart.java
@@ -21,18 +21,18 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
 
 /**
  * @param <T> data type for {@code band()} output
  */
 @Operator(group = "linalg")
-public final class BatchMatrixBandPart<T extends TType> extends RawOp implements Operand<T> {
+public final class BatchMatrixBandPart<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new BatchMatrixBandPart operation.
@@ -44,7 +44,7 @@ public final class BatchMatrixBandPart<T extends TType> extends RawOp implements
    * @return a new instance of BatchMatrixBandPart
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> BatchMatrixBandPart<T> create(Scope scope, Operand<T> input, Operand<TInt64> numLower, Operand<TInt64> numUpper) {
+  public static <T extends Tensor> BatchMatrixBandPart<T> create(Scope scope, Operand<T> input, Operand<TInt64> numLower, Operand<TInt64> numUpper) {
     OperationBuilder opBuilder = scope.env().opBuilder("BatchMatrixBandPart", scope.makeOpName("BatchMatrixBandPart"));
     opBuilder.addInput(input.asOutput());
     opBuilder.addInput(numLower.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixDeterminant.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixDeterminant.java
index f513e9f0057..7b28f87b1e0 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixDeterminant.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixDeterminant.java
@@ -21,17 +21,17 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "linalg")
-public final class BatchMatrixDeterminant<T extends TType> extends RawOp implements Operand<T> {
+public final class BatchMatrixDeterminant<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new BatchMatrixDeterminant operation.
@@ -41,7 +41,7 @@ public final class BatchMatrixDeterminant<T extends TType> extends RawOp impleme
    * @return a new instance of BatchMatrixDeterminant
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> BatchMatrixDeterminant<T> create(Scope scope, Operand<T> input) {
+  public static <T extends Tensor> BatchMatrixDeterminant<T> create(Scope scope, Operand<T> input) {
     OperationBuilder opBuilder = scope.env().opBuilder("BatchMatrixDeterminant", scope.makeOpName("BatchMatrixDeterminant"));
     opBuilder.addInput(input.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixDiag.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixDiag.java
index b3f9071f9f6..5bd6f2770d7 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixDiag.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixDiag.java
@@ -21,17 +21,17 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "linalg")
-public final class BatchMatrixDiag<T extends TType> extends RawOp implements Operand<T> {
+public final class BatchMatrixDiag<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new BatchMatrixDiag operation.
@@ -41,7 +41,7 @@ public final class BatchMatrixDiag<T extends TType> extends RawOp implements Ope
    * @return a new instance of BatchMatrixDiag
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> BatchMatrixDiag<T> create(Scope scope, Operand<T> diagonal) {
+  public static <T extends Tensor> BatchMatrixDiag<T> create(Scope scope, Operand<T> diagonal) {
     OperationBuilder opBuilder = scope.env().opBuilder("BatchMatrixDiag", scope.makeOpName("BatchMatrixDiag"));
     opBuilder.addInput(diagonal.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixDiagPart.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixDiagPart.java
index 47f78ed623d..c174055130c 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixDiagPart.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixDiagPart.java
@@ -21,17 +21,17 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * @param <T> data type for {@code diagonal()} output
  */
 @Operator(group = "linalg")
-public final class BatchMatrixDiagPart<T extends TType> extends RawOp implements Operand<T> {
+public final class BatchMatrixDiagPart<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new BatchMatrixDiagPart operation.
@@ -41,7 +41,7 @@ public final class BatchMatrixDiagPart<T extends TType> extends RawOp implements
    * @return a new instance of BatchMatrixDiagPart
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> BatchMatrixDiagPart<T> create(Scope scope, Operand<T> input) {
+  public static <T extends Tensor> BatchMatrixDiagPart<T> create(Scope scope, Operand<T> input) {
     OperationBuilder opBuilder = scope.env().opBuilder("BatchMatrixDiagPart", scope.makeOpName("BatchMatrixDiagPart"));
     opBuilder.addInput(input.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixInverse.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixInverse.java
index 36bbee414e9..8f596e62777 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixInverse.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixInverse.java
@@ -21,18 +21,18 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "linalg")
-public final class BatchMatrixInverse<T extends TNumber> extends RawOp implements Operand<T> {
+public final class BatchMatrixInverse<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.linalg.BatchMatrixInverse}
@@ -62,7 +62,7 @@ private Options() {
    * @return a new instance of BatchMatrixInverse
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> BatchMatrixInverse<T> create(Scope scope, Operand<T> input, Options... options) {
+  public static <T extends Tensor & TNumber> BatchMatrixInverse<T> create(Scope scope, Operand<T> input, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("BatchMatrixInverse", scope.makeOpName("BatchMatrixInverse"));
     opBuilder.addInput(input.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixSetDiag.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixSetDiag.java
index d83ee0194e7..c02c4eaf541 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixSetDiag.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixSetDiag.java
@@ -21,17 +21,17 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "linalg")
-public final class BatchMatrixSetDiag<T extends TType> extends RawOp implements Operand<T> {
+public final class BatchMatrixSetDiag<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new BatchMatrixSetDiag operation.
@@ -42,7 +42,7 @@ public final class BatchMatrixSetDiag<T extends TType> extends RawOp implements
    * @return a new instance of BatchMatrixSetDiag
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> BatchMatrixSetDiag<T> create(Scope scope, Operand<T> input, Operand<T> diagonal) {
+  public static <T extends Tensor> BatchMatrixSetDiag<T> create(Scope scope, Operand<T> input, Operand<T> diagonal) {
     OperationBuilder opBuilder = scope.env().opBuilder("BatchMatrixSetDiag", scope.makeOpName("BatchMatrixSetDiag"));
     opBuilder.addInput(input.asOutput());
     opBuilder.addInput(diagonal.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixSolve.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixSolve.java
index 677b8aa0e7b..7def4a32599 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixSolve.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixSolve.java
@@ -21,18 +21,18 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "linalg")
-public final class BatchMatrixSolve<T extends TNumber> extends RawOp implements Operand<T> {
+public final class BatchMatrixSolve<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.linalg.BatchMatrixSolve}
@@ -63,7 +63,7 @@ private Options() {
    * @return a new instance of BatchMatrixSolve
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> BatchMatrixSolve<T> create(Scope scope, Operand<T> matrix, Operand<T> rhs, Options... options) {
+  public static <T extends Tensor & TNumber> BatchMatrixSolve<T> create(Scope scope, Operand<T> matrix, Operand<T> rhs, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("BatchMatrixSolve", scope.makeOpName("BatchMatrixSolve"));
     opBuilder.addInput(matrix.asOutput());
     opBuilder.addInput(rhs.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixSolveLs.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixSolveLs.java
index ae4a4366b3b..a25215e20f4 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixSolveLs.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixSolveLs.java
@@ -21,19 +21,19 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TFloat64;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "linalg")
-public final class BatchMatrixSolveLs<T extends TNumber> extends RawOp implements Operand<T> {
+public final class BatchMatrixSolveLs<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.linalg.BatchMatrixSolveLs}
@@ -65,7 +65,7 @@ private Options() {
    * @return a new instance of BatchMatrixSolveLs
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> BatchMatrixSolveLs<T> create(Scope scope, Operand<T> matrix, Operand<T> rhs, Operand<TFloat64> l2Regularizer, Options... options) {
+  public static <T extends Tensor & TNumber> BatchMatrixSolveLs<T> create(Scope scope, Operand<T> matrix, Operand<T> rhs, Operand<TFloat64> l2Regularizer, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("BatchMatrixSolveLs", scope.makeOpName("BatchMatrixSolveLs"));
     opBuilder.addInput(matrix.asOutput());
     opBuilder.addInput(rhs.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixTriangularSolve.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixTriangularSolve.java
index cb1baa772fd..2c8b9e2ad8e 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixTriangularSolve.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixTriangularSolve.java
@@ -21,18 +21,18 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "linalg")
-public final class BatchMatrixTriangularSolve<T extends TNumber> extends RawOp implements Operand<T> {
+public final class BatchMatrixTriangularSolve<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.linalg.BatchMatrixTriangularSolve}
@@ -72,7 +72,7 @@ private Options() {
    * @return a new instance of BatchMatrixTriangularSolve
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> BatchMatrixTriangularSolve<T> create(Scope scope, Operand<T> matrix, Operand<T> rhs, Options... options) {
+  public static <T extends Tensor & TNumber> BatchMatrixTriangularSolve<T> create(Scope scope, Operand<T> matrix, Operand<T> rhs, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("BatchMatrixTriangularSolve", scope.makeOpName("BatchMatrixTriangularSolve"));
     opBuilder.addInput(matrix.asOutput());
     opBuilder.addInput(rhs.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchSelfAdjointEig.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchSelfAdjointEig.java
index e589f5c49fe..2aa8d1f456b 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchSelfAdjointEig.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchSelfAdjointEig.java
@@ -21,18 +21,18 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * @param <T> data type for {@code e()} output
  */
 @Operator(group = "linalg")
-public final class BatchSelfAdjointEig<T extends TNumber> extends RawOp {
+public final class BatchSelfAdjointEig<T extends Tensor & TNumber> extends RawOp {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.linalg.BatchSelfAdjointEig}
@@ -62,7 +62,7 @@ private Options() {
    * @return a new instance of BatchSelfAdjointEig
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> BatchSelfAdjointEig<T> create(Scope scope, Operand<T> input, Options... options) {
+  public static <T extends Tensor & TNumber> BatchSelfAdjointEig<T> create(Scope scope, Operand<T> input, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("BatchSelfAdjointEigV2", scope.makeOpName("BatchSelfAdjointEig"));
     opBuilder.addInput(input.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchSvd.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchSvd.java
index 21db016713a..0b9c7cc82fb 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchSvd.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchSvd.java
@@ -21,17 +21,17 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * @param <T> data type for {@code s()} output
  */
 @Operator(group = "linalg")
-public final class BatchSvd<T extends TType> extends RawOp {
+public final class BatchSvd<T extends Tensor> extends RawOp {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.linalg.BatchSvd}
@@ -70,7 +70,7 @@ private Options() {
    * @return a new instance of BatchSvd
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> BatchSvd<T> create(Scope scope, Operand<T> input, Options... options) {
+  public static <T extends Tensor> BatchSvd<T> create(Scope scope, Operand<T> input, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("BatchSvd", scope.makeOpName("BatchSvd"));
     opBuilder.addInput(input.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Cholesky.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Cholesky.java
index 3acf951d5a7..fa9ee8d4ae5 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Cholesky.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Cholesky.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes the Cholesky decomposition of one or more square matrices.
@@ -47,7 +47,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "linalg")
-public final class Cholesky<T extends TType> extends RawOp implements Operand<T> {
+public final class Cholesky<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Cholesky operation.
@@ -57,7 +57,7 @@ public final class Cholesky<T extends TType> extends RawOp implements Operand<T>
    * @return a new instance of Cholesky
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> Cholesky<T> create(Scope scope, Operand<T> input) {
+  public static <T extends Tensor> Cholesky<T> create(Scope scope, Operand<T> input) {
     OperationBuilder opBuilder = scope.env().opBuilder("Cholesky", scope.makeOpName("Cholesky"));
     opBuilder.addInput(input.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/CholeskyGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/CholeskyGrad.java
index b4171907853..89d2868825c 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/CholeskyGrad.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/CholeskyGrad.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes the reverse mode backpropagated gradient of the Cholesky algorithm.
@@ -37,7 +37,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "linalg")
-public final class CholeskyGrad<T extends TNumber> extends RawOp implements Operand<T> {
+public final class CholeskyGrad<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new CholeskyGrad operation.
@@ -52,7 +52,7 @@ public final class CholeskyGrad<T extends TNumber> extends RawOp implements Oper
    * @return a new instance of CholeskyGrad
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> CholeskyGrad<T> create(Scope scope, Operand<T> l, Operand<T> grad) {
+  public static <T extends Tensor & TNumber> CholeskyGrad<T> create(Scope scope, Operand<T> l, Operand<T> grad) {
     OperationBuilder opBuilder = scope.env().opBuilder("CholeskyGrad", scope.makeOpName("CholeskyGrad"));
     opBuilder.addInput(l.asOutput());
     opBuilder.addInput(grad.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/ConjugateTranspose.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/ConjugateTranspose.java
index 2da45552f35..702176bac05 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/ConjugateTranspose.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/ConjugateTranspose.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Shuffle dimensions of x according to a permutation and conjugate the result.
@@ -38,7 +38,7 @@
  * @param <T> data type for {@code y()} output
  */
 @Operator(group = "linalg")
-public final class ConjugateTranspose<T extends TType> extends RawOp implements Operand<T> {
+public final class ConjugateTranspose<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new ConjugateTranspose operation.
@@ -49,7 +49,7 @@ public final class ConjugateTranspose<T extends TType> extends RawOp implements
    * @return a new instance of ConjugateTranspose
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> ConjugateTranspose<T> create(Scope scope, Operand<T> x, Operand<U> perm) {
+  public static <T extends Tensor, U extends Tensor & TNumber> ConjugateTranspose<T> create(Scope scope, Operand<T> x, Operand<U> perm) {
     OperationBuilder opBuilder = scope.env().opBuilder("ConjugateTranspose", scope.makeOpName("ConjugateTranspose"));
     opBuilder.addInput(x.asOutput());
     opBuilder.addInput(perm.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Cross.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Cross.java
index 0afc8a2bf60..2b2c99d8d6b 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Cross.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Cross.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Compute the pairwise cross product.
@@ -38,7 +38,7 @@
  * @param <T> data type for {@code product()} output
  */
 @Operator(group = "linalg")
-public final class Cross<T extends TNumber> extends RawOp implements Operand<T> {
+public final class Cross<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Cross operation.
@@ -49,7 +49,7 @@ public final class Cross<T extends TNumber> extends RawOp implements Operand<T>
    * @return a new instance of Cross
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> Cross<T> create(Scope scope, Operand<T> a, Operand<T> b) {
+  public static <T extends Tensor & TNumber> Cross<T> create(Scope scope, Operand<T> a, Operand<T> b) {
     OperationBuilder opBuilder = scope.env().opBuilder("Cross", scope.makeOpName("Cross"));
     opBuilder.addInput(a.asOutput());
     opBuilder.addInput(b.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Det.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Det.java
index 7134ef6ea44..e81f186064c 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Det.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Det.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes the determinant of one or more square matrices.
@@ -37,7 +37,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "linalg")
-public final class Det<T extends TType> extends RawOp implements Operand<T> {
+public final class Det<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Det operation.
@@ -47,7 +47,7 @@ public final class Det<T extends TType> extends RawOp implements Operand<T> {
    * @return a new instance of Det
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> Det<T> create(Scope scope, Operand<T> input) {
+  public static <T extends Tensor> Det<T> create(Scope scope, Operand<T> input) {
     OperationBuilder opBuilder = scope.env().opBuilder("MatrixDeterminant", scope.makeOpName("Det"));
     opBuilder.addInput(input.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Eig.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Eig.java
index 68c33487129..ebce453d78d 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Eig.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Eig.java
@@ -22,11 +22,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes the eigen decomposition of one or more square matrices.
@@ -46,7 +46,7 @@
  * @param <U> data type for {@code e()} output
  */
 @Operator(group = "linalg")
-public final class Eig<U extends TType> extends RawOp {
+public final class Eig<U extends Tensor> extends RawOp {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.linalg.Eig}
@@ -78,7 +78,7 @@ private Options() {
    * @return a new instance of Eig
    */
   @Endpoint(describeByClass = true)
-  public static <U extends TType, T extends TType> Eig<U> create(Scope scope, Operand<T> input, DataType<U> Tout, Options... options) {
+  public static <U extends Tensor, T extends Tensor> Eig<U> create(Scope scope, Operand<T> input, DataType<U> Tout, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("Eig", scope.makeOpName("Eig"));
     opBuilder.addInput(input.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Einsum.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Einsum.java
index 6817270edd1..f1ad82b2bcc 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Einsum.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Einsum.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.Operands;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Tensor contraction according to Einstein summation convention.
@@ -111,7 +111,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "linalg")
-public final class Einsum<T extends TType> extends RawOp implements Operand<T> {
+public final class Einsum<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Einsum operation.
@@ -122,7 +122,7 @@ public final class Einsum<T extends TType> extends RawOp implements Operand<T> {
    * @return a new instance of Einsum
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> Einsum<T> create(Scope scope, Iterable<Operand<T>> inputs, String equation) {
+  public static <T extends Tensor> Einsum<T> create(Scope scope, Iterable<Operand<T>> inputs, String equation) {
     OperationBuilder opBuilder = scope.env().opBuilder("Einsum", scope.makeOpName("Einsum"));
     opBuilder.addInputList(Operands.asOutputs(inputs));
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/EuclideanNorm.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/EuclideanNorm.java
index f8ce40f62bf..8eb67637d16 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/EuclideanNorm.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/EuclideanNorm.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes the euclidean norm of elements across dimensions of a tensor.
@@ -39,7 +39,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "linalg")
-public final class EuclideanNorm<T extends TType> extends RawOp implements Operand<T> {
+public final class EuclideanNorm<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.linalg.EuclideanNorm}
@@ -71,7 +71,7 @@ private Options() {
    * @return a new instance of EuclideanNorm
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> EuclideanNorm<T> create(Scope scope, Operand<T> input, Operand<U> axis, Options... options) {
+  public static <T extends Tensor, U extends Tensor & TNumber> EuclideanNorm<T> create(Scope scope, Operand<T> input, Operand<U> axis, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("EuclideanNorm", scope.makeOpName("EuclideanNorm"));
     opBuilder.addInput(input.asOutput());
     opBuilder.addInput(axis.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Inv.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Inv.java
index e482957bdac..d5cdb76b5e7 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Inv.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Inv.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes the inverse of one or more square invertible matrices or their
@@ -45,7 +45,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "linalg")
-public final class Inv<T extends TType> extends RawOp implements Operand<T> {
+public final class Inv<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.linalg.Inv}
@@ -75,7 +75,7 @@ private Options() {
    * @return a new instance of Inv
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> Inv<T> create(Scope scope, Operand<T> input, Options... options) {
+  public static <T extends Tensor> Inv<T> create(Scope scope, Operand<T> input, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("MatrixInverse", scope.makeOpName("Inv"));
     opBuilder.addInput(input.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/LogMatrixDeterminant.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/LogMatrixDeterminant.java
index 33781d561b4..ece21956d3d 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/LogMatrixDeterminant.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/LogMatrixDeterminant.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes the sign and the log of the absolute value of the determinant of
@@ -43,7 +43,7 @@
  * @param <T> data type for {@code sign()} output
  */
 @Operator(group = "linalg")
-public final class LogMatrixDeterminant<T extends TType> extends RawOp {
+public final class LogMatrixDeterminant<T extends Tensor> extends RawOp {
   
   /**
    * Factory method to create a class wrapping a new LogMatrixDeterminant operation.
@@ -53,7 +53,7 @@ public final class LogMatrixDeterminant<T extends TType> extends RawOp {
    * @return a new instance of LogMatrixDeterminant
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> LogMatrixDeterminant<T> create(Scope scope, Operand<T> input) {
+  public static <T extends Tensor> LogMatrixDeterminant<T> create(Scope scope, Operand<T> input) {
     OperationBuilder opBuilder = scope.env().opBuilder("LogMatrixDeterminant", scope.makeOpName("LogMatrixDeterminant"));
     opBuilder.addInput(input.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Lu.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Lu.java
index ce37bbcbea2..fa937ab6c24 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Lu.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Lu.java
@@ -22,13 +22,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes the LU decomposition of one or more square matrices.
@@ -55,7 +55,7 @@
  * @param <U> data type for {@code p()} output
  */
 @Operator(group = "linalg")
-public final class Lu<T extends TType, U extends TNumber> extends RawOp {
+public final class Lu<T extends Tensor, U extends Tensor & TNumber> extends RawOp {
   
   /**
    * Factory method to create a class wrapping a new Lu operation.
@@ -67,7 +67,7 @@ public final class Lu<T extends TType, U extends TNumber> extends RawOp {
    * @return a new instance of Lu
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> Lu<T, U> create(Scope scope, Operand<T> input, DataType<U> outputIdxType) {
+  public static <T extends Tensor, U extends Tensor & TNumber> Lu<T, U> create(Scope scope, Operand<T> input, DataType<U> outputIdxType) {
     OperationBuilder opBuilder = scope.env().opBuilder("Lu", scope.makeOpName("Lu"));
     opBuilder.addInput(input.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
@@ -84,7 +84,7 @@ public static <T extends TType, U extends TNumber> Lu<T, U> create(Scope scope,
    * @return a new instance of Lu
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> Lu<T, TInt32> create(Scope scope, Operand<T> input) {
+  public static <T extends Tensor> Lu<T, TInt32> create(Scope scope, Operand<T> input) {
     return create(scope, input, TInt32.DTYPE);
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/MatMul.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/MatMul.java
index 8b227e3910c..54443149d3b 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/MatMul.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/MatMul.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Multiply the matrix "a" by the matrix "b".
@@ -41,7 +41,7 @@
  * @param <T> data type for {@code product()} output
  */
 @Operator(group = "linalg")
-public final class MatMul<T extends TType> extends RawOp implements Operand<T> {
+public final class MatMul<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.linalg.MatMul}
@@ -81,7 +81,7 @@ private Options() {
    * @return a new instance of MatMul
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> MatMul<T> create(Scope scope, Operand<T> a, Operand<T> b, Options... options) {
+  public static <T extends Tensor> MatMul<T> create(Scope scope, Operand<T> a, Operand<T> b, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("MatMul", scope.makeOpName("MatMul"));
     opBuilder.addInput(a.asOutput());
     opBuilder.addInput(b.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/MatrixDiag.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/MatrixDiag.java
index b187ecc0930..0baa4506cc6 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/MatrixDiag.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/MatrixDiag.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
-import org.tensorflow.types.family.TType;
 
 /**
  * Returns a batched diagonal tensor with given batched diagonal values.
@@ -119,7 +119,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "linalg")
-public final class MatrixDiag<T extends TType> extends RawOp implements Operand<T> {
+public final class MatrixDiag<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new MatrixDiag operation.
@@ -141,7 +141,7 @@ public final class MatrixDiag<T extends TType> extends RawOp implements Operand<
    * @return a new instance of MatrixDiag
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> MatrixDiag<T> create(Scope scope, Operand<T> diagonal, Operand<TInt32> k, Operand<TInt32> numRows, Operand<TInt32> numCols, Operand<T> paddingValue) {
+  public static <T extends Tensor> MatrixDiag<T> create(Scope scope, Operand<T> diagonal, Operand<TInt32> k, Operand<TInt32> numRows, Operand<TInt32> numCols, Operand<T> paddingValue) {
     OperationBuilder opBuilder = scope.env().opBuilder("MatrixDiagV2", scope.makeOpName("MatrixDiag"));
     opBuilder.addInput(diagonal.asOutput());
     opBuilder.addInput(k.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/MatrixDiagPart.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/MatrixDiagPart.java
index c4dee3f7d76..73ade6ac51c 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/MatrixDiagPart.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/MatrixDiagPart.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
-import org.tensorflow.types.family.TType;
 
 /**
  * Returns the batched diagonal part of a batched tensor.
@@ -101,7 +101,7 @@
  * @param <T> data type for {@code diagonal()} output
  */
 @Operator(group = "linalg")
-public final class MatrixDiagPart<T extends TType> extends RawOp implements Operand<T> {
+public final class MatrixDiagPart<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new MatrixDiagPart operation.
@@ -117,7 +117,7 @@ public final class MatrixDiagPart<T extends TType> extends RawOp implements Oper
    * @return a new instance of MatrixDiagPart
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> MatrixDiagPart<T> create(Scope scope, Operand<T> input, Operand<TInt32> k, Operand<T> paddingValue) {
+  public static <T extends Tensor> MatrixDiagPart<T> create(Scope scope, Operand<T> input, Operand<TInt32> k, Operand<T> paddingValue) {
     OperationBuilder opBuilder = scope.env().opBuilder("MatrixDiagPartV2", scope.makeOpName("MatrixDiagPart"));
     opBuilder.addInput(input.asOutput());
     opBuilder.addInput(k.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/MatrixDiagPartV3.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/MatrixDiagPartV3.java
index 5803f9251f5..5ca61bd9c17 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/MatrixDiagPartV3.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/MatrixDiagPartV3.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
-import org.tensorflow.types.family.TType;
 
 /**
  * Returns the batched diagonal part of a batched tensor.
@@ -132,7 +132,7 @@
  * @param <T> data type for {@code diagonal()} output
  */
 @Operator(group = "linalg")
-public final class MatrixDiagPartV3<T extends TType> extends RawOp implements Operand<T> {
+public final class MatrixDiagPartV3<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.linalg.MatrixDiagPartV3}
@@ -174,7 +174,7 @@ private Options() {
    * @return a new instance of MatrixDiagPartV3
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> MatrixDiagPartV3<T> create(Scope scope, Operand<T> input, Operand<TInt32> k, Operand<T> paddingValue, Options... options) {
+  public static <T extends Tensor> MatrixDiagPartV3<T> create(Scope scope, Operand<T> input, Operand<TInt32> k, Operand<T> paddingValue, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("MatrixDiagPartV3", scope.makeOpName("MatrixDiagPartV3"));
     opBuilder.addInput(input.asOutput());
     opBuilder.addInput(k.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/MatrixDiagV3.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/MatrixDiagV3.java
index 10b57ecc19a..5c8a7c72023 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/MatrixDiagV3.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/MatrixDiagV3.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
-import org.tensorflow.types.family.TType;
 
 /**
  * Returns a batched diagonal tensor with given batched diagonal values.
@@ -148,7 +148,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "linalg")
-public final class MatrixDiagV3<T extends TType> extends RawOp implements Operand<T> {
+public final class MatrixDiagV3<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.linalg.MatrixDiagV3}
@@ -196,7 +196,7 @@ private Options() {
    * @return a new instance of MatrixDiagV3
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> MatrixDiagV3<T> create(Scope scope, Operand<T> diagonal, Operand<TInt32> k, Operand<TInt32> numRows, Operand<TInt32> numCols, Operand<T> paddingValue, Options... options) {
+  public static <T extends Tensor> MatrixDiagV3<T> create(Scope scope, Operand<T> diagonal, Operand<TInt32> k, Operand<TInt32> numRows, Operand<TInt32> numCols, Operand<T> paddingValue, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("MatrixDiagV3", scope.makeOpName("MatrixDiagV3"));
     opBuilder.addInput(diagonal.asOutput());
     opBuilder.addInput(k.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/MatrixLogarithm.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/MatrixLogarithm.java
index 71d3a2c9575..d098fbe0d4d 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/MatrixLogarithm.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/MatrixLogarithm.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes the matrix logarithm of one or more square matrices:
@@ -48,7 +48,7 @@
  * 
  * @param <T> data type for {@code output()} output
  */
-public final class MatrixLogarithm<T extends TType> extends RawOp implements Operand<T> {
+public final class MatrixLogarithm<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new MatrixLogarithm operation.
@@ -58,7 +58,7 @@ public final class MatrixLogarithm<T extends TType> extends RawOp implements Ope
    * @return a new instance of MatrixLogarithm
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> MatrixLogarithm<T> create(Scope scope, Operand<T> input) {
+  public static <T extends Tensor> MatrixLogarithm<T> create(Scope scope, Operand<T> input) {
     OperationBuilder opBuilder = scope.env().opBuilder("MatrixLogarithm", scope.makeOpName("MatrixLogarithm"));
     opBuilder.addInput(input.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/MatrixSetDiag.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/MatrixSetDiag.java
index 80367a3670b..c7ae957b0a5 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/MatrixSetDiag.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/MatrixSetDiag.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
-import org.tensorflow.types.family.TType;
 
 /**
  * Returns a batched matrix tensor with new batched diagonal values.
@@ -136,7 +136,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "linalg")
-public final class MatrixSetDiag<T extends TType> extends RawOp implements Operand<T> {
+public final class MatrixSetDiag<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.linalg.MatrixSetDiag}
@@ -178,7 +178,7 @@ private Options() {
    * @return a new instance of MatrixSetDiag
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> MatrixSetDiag<T> create(Scope scope, Operand<T> input, Operand<T> diagonal, Operand<TInt32> k, Options... options) {
+  public static <T extends Tensor> MatrixSetDiag<T> create(Scope scope, Operand<T> input, Operand<T> diagonal, Operand<TInt32> k, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("MatrixSetDiagV3", scope.makeOpName("MatrixSetDiag"));
     opBuilder.addInput(input.asOutput());
     opBuilder.addInput(diagonal.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/MatrixSolveLs.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/MatrixSolveLs.java
index 633145ce24e..0f8ce97fdec 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/MatrixSolveLs.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/MatrixSolveLs.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TFloat64;
-import org.tensorflow.types.family.TType;
 
 /**
  * Solves one or more linear least-squares problems.
@@ -69,7 +69,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "linalg")
-public final class MatrixSolveLs<T extends TType> extends RawOp implements Operand<T> {
+public final class MatrixSolveLs<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.linalg.MatrixSolveLs}
@@ -105,7 +105,7 @@ private Options() {
    * @return a new instance of MatrixSolveLs
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> MatrixSolveLs<T> create(Scope scope, Operand<T> matrix, Operand<T> rhs, Operand<TFloat64> l2Regularizer, Options... options) {
+  public static <T extends Tensor> MatrixSolveLs<T> create(Scope scope, Operand<T> matrix, Operand<T> rhs, Operand<TFloat64> l2Regularizer, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("MatrixSolveLs", scope.makeOpName("MatrixSolveLs"));
     opBuilder.addInput(matrix.asOutput());
     opBuilder.addInput(rhs.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Qr.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Qr.java
index cf58f41507f..c8fbd3511d4 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Qr.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Qr.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes the QR decompositions of one or more matrices.
@@ -44,7 +44,7 @@
  * @param <T> data type for {@code q()} output
  */
 @Operator(group = "linalg")
-public final class Qr<T extends TType> extends RawOp {
+public final class Qr<T extends Tensor> extends RawOp {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.linalg.Qr}
@@ -76,7 +76,7 @@ private Options() {
    * @return a new instance of Qr
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> Qr<T> create(Scope scope, Operand<T> input, Options... options) {
+  public static <T extends Tensor> Qr<T> create(Scope scope, Operand<T> input, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("Qr", scope.makeOpName("Qr"));
     opBuilder.addInput(input.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/QuantizedMatMul.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/QuantizedMatMul.java
index 445d27e1591..38a0f336092 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/QuantizedMatMul.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/QuantizedMatMul.java
@@ -22,12 +22,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.family.TType;
 
 /**
  * Perform a quantized matrix multiplication of  `a` by the matrix `b`.
@@ -40,7 +40,7 @@
  * @param <V> data type for {@code out()} output
  */
 @Operator(group = "linalg")
-public final class QuantizedMatMul<V extends TType> extends RawOp {
+public final class QuantizedMatMul<V extends Tensor> extends RawOp {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.linalg.QuantizedMatMul}
@@ -87,7 +87,7 @@ private Options() {
    * @return a new instance of QuantizedMatMul
    */
   @Endpoint(describeByClass = true)
-  public static <V extends TType, T extends TType, U extends TType, W extends TType> QuantizedMatMul<V> create(Scope scope, Operand<T> a, Operand<U> b, Operand<TFloat32> minA, Operand<TFloat32> maxA, Operand<TFloat32> minB, Operand<TFloat32> maxB, DataType<V> Toutput, DataType<W> Tactivation, Options... options) {
+  public static <V extends Tensor, T extends Tensor, U extends Tensor, W extends Tensor> QuantizedMatMul<V> create(Scope scope, Operand<T> a, Operand<U> b, Operand<TFloat32> minA, Operand<TFloat32> maxA, Operand<TFloat32> minB, Operand<TFloat32> maxB, DataType<V> Toutput, DataType<W> Tactivation, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("QuantizedMatMul", scope.makeOpName("QuantizedMatMul"));
     opBuilder.addInput(a.asOutput());
     opBuilder.addInput(b.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/QuantizedMatMulWithBias.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/QuantizedMatMulWithBias.java
index 52ba1e9576b..9a333aa8cf2 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/QuantizedMatMulWithBias.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/QuantizedMatMulWithBias.java
@@ -22,12 +22,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.family.TType;
 
 /**
  * Performs a quantized matrix multiplication of `a` by the matrix `b` with bias
@@ -41,7 +41,7 @@
  * 
  * @param <W> data type for {@code out()} output
  */
-public final class QuantizedMatMulWithBias<W extends TType> extends RawOp {
+public final class QuantizedMatMulWithBias<W extends Tensor> extends RawOp {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.linalg.QuantizedMatMulWithBias}
@@ -97,7 +97,7 @@ private Options() {
    * @return a new instance of QuantizedMatMulWithBias
    */
   @Endpoint(describeByClass = true)
-  public static <W extends TType, T extends TType, U extends TType, V extends TType> QuantizedMatMulWithBias<W> create(Scope scope, Operand<T> a, Operand<U> b, Operand<V> bias, Operand<TFloat32> minA, Operand<TFloat32> maxA, Operand<TFloat32> minB, Operand<TFloat32> maxB, DataType<W> Toutput, Options... options) {
+  public static <W extends Tensor, T extends Tensor, U extends Tensor, V extends Tensor> QuantizedMatMulWithBias<W> create(Scope scope, Operand<T> a, Operand<U> b, Operand<V> bias, Operand<TFloat32> minA, Operand<TFloat32> maxA, Operand<TFloat32> minB, Operand<TFloat32> maxB, DataType<W> Toutput, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("QuantizedMatMulWithBias", scope.makeOpName("QuantizedMatMulWithBias"));
     opBuilder.addInput(a.asOutput());
     opBuilder.addInput(b.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/QuantizedMatMulWithBiasAndRelu.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/QuantizedMatMulWithBiasAndRelu.java
index 4d7ba46f49e..354fd26a18a 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/QuantizedMatMulWithBiasAndRelu.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/QuantizedMatMulWithBiasAndRelu.java
@@ -22,12 +22,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.family.TType;
 
 /**
  * Perform a quantized matrix multiplication of  `a` by the matrix `b` with bias
@@ -42,7 +42,7 @@
  * 
  * @param <V> data type for {@code out()} output
  */
-public final class QuantizedMatMulWithBiasAndRelu<V extends TType> extends RawOp {
+public final class QuantizedMatMulWithBiasAndRelu<V extends Tensor> extends RawOp {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.linalg.QuantizedMatMulWithBiasAndRelu}
@@ -98,7 +98,7 @@ private Options() {
    * @return a new instance of QuantizedMatMulWithBiasAndRelu
    */
   @Endpoint(describeByClass = true)
-  public static <V extends TType, T extends TType, U extends TType> QuantizedMatMulWithBiasAndRelu<V> create(Scope scope, Operand<T> a, Operand<U> b, Operand<TFloat32> bias, Operand<TFloat32> minA, Operand<TFloat32> maxA, Operand<TFloat32> minB, Operand<TFloat32> maxB, DataType<V> Toutput, Options... options) {
+  public static <V extends Tensor, T extends Tensor, U extends Tensor> QuantizedMatMulWithBiasAndRelu<V> create(Scope scope, Operand<T> a, Operand<U> b, Operand<TFloat32> bias, Operand<TFloat32> minA, Operand<TFloat32> maxA, Operand<TFloat32> minB, Operand<TFloat32> maxB, DataType<V> Toutput, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("QuantizedMatMulWithBiasAndRelu", scope.makeOpName("QuantizedMatMulWithBiasAndRelu"));
     opBuilder.addInput(a.asOutput());
     opBuilder.addInput(b.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/QuantizedMatMulWithBiasAndReluAndRequantize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/QuantizedMatMulWithBiasAndReluAndRequantize.java
index e5639594805..a5548b7f368 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/QuantizedMatMulWithBiasAndReluAndRequantize.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/QuantizedMatMulWithBiasAndReluAndRequantize.java
@@ -22,12 +22,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.family.TType;
 
 /**
  * Perform a quantized matrix multiplication of  `a` by the matrix `b` with bias
@@ -43,7 +43,7 @@
  * 
  * @param <W> data type for {@code out()} output
  */
-public final class QuantizedMatMulWithBiasAndReluAndRequantize<W extends TType> extends RawOp {
+public final class QuantizedMatMulWithBiasAndReluAndRequantize<W extends Tensor> extends RawOp {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.linalg.QuantizedMatMulWithBiasAndReluAndRequantize}
@@ -101,7 +101,7 @@ private Options() {
    * @return a new instance of QuantizedMatMulWithBiasAndReluAndRequantize
    */
   @Endpoint(describeByClass = true)
-  public static <W extends TType, T extends TType, U extends TType, V extends TType> QuantizedMatMulWithBiasAndReluAndRequantize<W> create(Scope scope, Operand<T> a, Operand<U> b, Operand<V> bias, Operand<TFloat32> minA, Operand<TFloat32> maxA, Operand<TFloat32> minB, Operand<TFloat32> maxB, Operand<TFloat32> minFreezedOutput, Operand<TFloat32> maxFreezedOutput, DataType<W> Toutput, Options... options) {
+  public static <W extends Tensor, T extends Tensor, U extends Tensor, V extends Tensor> QuantizedMatMulWithBiasAndReluAndRequantize<W> create(Scope scope, Operand<T> a, Operand<U> b, Operand<V> bias, Operand<TFloat32> minA, Operand<TFloat32> maxA, Operand<TFloat32> minB, Operand<TFloat32> maxB, Operand<TFloat32> minFreezedOutput, Operand<TFloat32> maxFreezedOutput, DataType<W> Toutput, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("QuantizedMatMulWithBiasAndReluAndRequantize", scope.makeOpName("QuantizedMatMulWithBiasAndReluAndRequantize"));
     opBuilder.addInput(a.asOutput());
     opBuilder.addInput(b.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/SelfAdjointEig.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/SelfAdjointEig.java
index 60bbafe6e70..af4898246ee 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/SelfAdjointEig.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/SelfAdjointEig.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes the eigen decomposition of one or more square self-adjoint matrices.
@@ -45,7 +45,7 @@
  * @param <T> data type for {@code e()} output
  */
 @Operator(group = "linalg")
-public final class SelfAdjointEig<T extends TType> extends RawOp {
+public final class SelfAdjointEig<T extends Tensor> extends RawOp {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.linalg.SelfAdjointEig}
@@ -76,7 +76,7 @@ private Options() {
    * @return a new instance of SelfAdjointEig
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> SelfAdjointEig<T> create(Scope scope, Operand<T> input, Options... options) {
+  public static <T extends Tensor> SelfAdjointEig<T> create(Scope scope, Operand<T> input, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("SelfAdjointEigV2", scope.makeOpName("SelfAdjointEig"));
     opBuilder.addInput(input.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Solve.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Solve.java
index 62f1beb11c8..1966cfb65f1 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Solve.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Solve.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Solves systems of linear equations.
@@ -40,7 +40,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "linalg")
-public final class Solve<T extends TType> extends RawOp implements Operand<T> {
+public final class Solve<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.linalg.Solve}
@@ -72,7 +72,7 @@ private Options() {
    * @return a new instance of Solve
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> Solve<T> create(Scope scope, Operand<T> matrix, Operand<T> rhs, Options... options) {
+  public static <T extends Tensor> Solve<T> create(Scope scope, Operand<T> matrix, Operand<T> rhs, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("MatrixSolve", scope.makeOpName("Solve"));
     opBuilder.addInput(matrix.asOutput());
     opBuilder.addInput(rhs.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Sqrtm.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Sqrtm.java
index 35c3b88605c..36096b8c83c 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Sqrtm.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Sqrtm.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes the matrix square root of one or more square matrices:
@@ -49,7 +49,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "linalg")
-public final class Sqrtm<T extends TType> extends RawOp implements Operand<T> {
+public final class Sqrtm<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Sqrtm operation.
@@ -59,7 +59,7 @@ public final class Sqrtm<T extends TType> extends RawOp implements Operand<T> {
    * @return a new instance of Sqrtm
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> Sqrtm<T> create(Scope scope, Operand<T> input) {
+  public static <T extends Tensor> Sqrtm<T> create(Scope scope, Operand<T> input) {
     OperationBuilder opBuilder = scope.env().opBuilder("MatrixSquareRoot", scope.makeOpName("Sqrtm"));
     opBuilder.addInput(input.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Svd.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Svd.java
index 4ff72f0e139..60d85b7d76e 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Svd.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Svd.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes the singular value decompositions of one or more matrices.
@@ -45,7 +45,7 @@
  * @param <T> data type for {@code s()} output
  */
 @Operator(group = "linalg")
-public final class Svd<T extends TType> extends RawOp {
+public final class Svd<T extends Tensor> extends RawOp {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.linalg.Svd}
@@ -89,7 +89,7 @@ private Options() {
    * @return a new instance of Svd
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> Svd<T> create(Scope scope, Operand<T> input, Options... options) {
+  public static <T extends Tensor> Svd<T> create(Scope scope, Operand<T> input, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("Svd", scope.makeOpName("Svd"));
     opBuilder.addInput(input.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/TensorDiag.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/TensorDiag.java
index 126a8c5cde7..0bf02584809 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/TensorDiag.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/TensorDiag.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Returns a diagonal tensor with a given diagonal values.
@@ -51,7 +51,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "linalg")
-public final class TensorDiag<T extends TType> extends RawOp implements Operand<T> {
+public final class TensorDiag<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new TensorDiag operation.
@@ -61,7 +61,7 @@ public final class TensorDiag<T extends TType> extends RawOp implements Operand<
    * @return a new instance of TensorDiag
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> TensorDiag<T> create(Scope scope, Operand<T> diagonal) {
+  public static <T extends Tensor> TensorDiag<T> create(Scope scope, Operand<T> diagonal) {
     OperationBuilder opBuilder = scope.env().opBuilder("Diag", scope.makeOpName("TensorDiag"));
     opBuilder.addInput(diagonal.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/TensorDiagPart.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/TensorDiagPart.java
index e999ca96a4b..e218c245953 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/TensorDiagPart.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/TensorDiagPart.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Returns the diagonal part of the tensor.
@@ -52,7 +52,7 @@
  * @param <T> data type for {@code diagonal()} output
  */
 @Operator(group = "linalg")
-public final class TensorDiagPart<T extends TType> extends RawOp implements Operand<T> {
+public final class TensorDiagPart<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new TensorDiagPart operation.
@@ -62,7 +62,7 @@ public final class TensorDiagPart<T extends TType> extends RawOp implements Oper
    * @return a new instance of TensorDiagPart
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> TensorDiagPart<T> create(Scope scope, Operand<T> input) {
+  public static <T extends Tensor> TensorDiagPart<T> create(Scope scope, Operand<T> input) {
     OperationBuilder opBuilder = scope.env().opBuilder("DiagPart", scope.makeOpName("TensorDiagPart"));
     opBuilder.addInput(input.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Transpose.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Transpose.java
index 1df98fabe64..a736582d262 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Transpose.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Transpose.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Shuffle dimensions of x according to a permutation.
@@ -37,7 +37,7 @@
  * @param <T> data type for {@code y()} output
  */
 @Operator(group = "linalg")
-public final class Transpose<T extends TType> extends RawOp implements Operand<T> {
+public final class Transpose<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Transpose operation.
@@ -48,7 +48,7 @@ public final class Transpose<T extends TType> extends RawOp implements Operand<T
    * @return a new instance of Transpose
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> Transpose<T> create(Scope scope, Operand<T> x, Operand<U> perm) {
+  public static <T extends Tensor, U extends Tensor & TNumber> Transpose<T> create(Scope scope, Operand<T> x, Operand<U> perm) {
     OperationBuilder opBuilder = scope.env().opBuilder("Transpose", scope.makeOpName("Transpose"));
     opBuilder.addInput(x.asOutput());
     opBuilder.addInput(perm.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/TriangularSolve.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/TriangularSolve.java
index da86d99b6c6..209f3ef14be 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/TriangularSolve.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/TriangularSolve.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Solves systems of linear equations with upper or lower triangular matrices by backsubstitution.
@@ -80,7 +80,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "linalg")
-public final class TriangularSolve<T extends TType> extends RawOp implements Operand<T> {
+public final class TriangularSolve<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.linalg.TriangularSolve}
@@ -126,7 +126,7 @@ private Options() {
    * @return a new instance of TriangularSolve
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> TriangularSolve<T> create(Scope scope, Operand<T> matrix, Operand<T> rhs, Options... options) {
+  public static <T extends Tensor> TriangularSolve<T> create(Scope scope, Operand<T> matrix, Operand<T> rhs, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("MatrixTriangularSolve", scope.makeOpName("TriangularSolve"));
     opBuilder.addInput(matrix.asOutput());
     opBuilder.addInput(rhs.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/TridiagonalMatMul.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/TridiagonalMatMul.java
index 2ff1c18fc60..c2ea63fe6db 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/TridiagonalMatMul.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/TridiagonalMatMul.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Calculate product with tridiagonal matrix.
@@ -34,7 +34,7 @@
  * 
  * @param <T> data type for {@code output()} output
  */
-public final class TridiagonalMatMul<T extends TType> extends RawOp implements Operand<T> {
+public final class TridiagonalMatMul<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new TridiagonalMatMul operation.
@@ -51,7 +51,7 @@ public final class TridiagonalMatMul<T extends TType> extends RawOp implements O
    * @return a new instance of TridiagonalMatMul
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> TridiagonalMatMul<T> create(Scope scope, Operand<T> superdiag, Operand<T> maindiag, Operand<T> subdiag, Operand<T> rhs) {
+  public static <T extends Tensor> TridiagonalMatMul<T> create(Scope scope, Operand<T> superdiag, Operand<T> maindiag, Operand<T> subdiag, Operand<T> rhs) {
     OperationBuilder opBuilder = scope.env().opBuilder("TridiagonalMatMul", scope.makeOpName("TridiagonalMatMul"));
     opBuilder.addInput(superdiag.asOutput());
     opBuilder.addInput(maindiag.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/TridiagonalSolve.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/TridiagonalSolve.java
index 84a65b02b55..8ac8df38394 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/TridiagonalSolve.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/TridiagonalSolve.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Solves tridiagonal systems of equations.
@@ -40,7 +40,7 @@
  * 
  * @param <T> data type for {@code output()} output
  */
-public final class TridiagonalSolve<T extends TType> extends RawOp implements Operand<T> {
+public final class TridiagonalSolve<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.linalg.TridiagonalSolve}
@@ -76,7 +76,7 @@ private Options() {
    * @return a new instance of TridiagonalSolve
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> TridiagonalSolve<T> create(Scope scope, Operand<T> diagonals, Operand<T> rhs, Options... options) {
+  public static <T extends Tensor> TridiagonalSolve<T> create(Scope scope, Operand<T> diagonals, Operand<T> rhs, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("TridiagonalSolve", scope.makeOpName("TridiagonalSolve"));
     opBuilder.addInput(diagonals.asOutput());
     opBuilder.addInput(rhs.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/CSRSparseMatrixComponents.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/CSRSparseMatrixComponents.java
index 332382cec08..987e05915bb 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/CSRSparseMatrixComponents.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/CSRSparseMatrixComponents.java
@@ -22,12 +22,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
-import org.tensorflow.types.family.TType;
 
 /**
  * Reads out the CSR components at batch `index`.
@@ -37,7 +37,7 @@
  * 
  * @param <T> data type for {@code values()} output
  */
-public final class CSRSparseMatrixComponents<T extends TType> extends RawOp {
+public final class CSRSparseMatrixComponents<T extends Tensor> extends RawOp {
   
   /**
    * Factory method to create a class wrapping a new CSRSparseMatrixComponents operation.
@@ -49,7 +49,7 @@ public final class CSRSparseMatrixComponents<T extends TType> extends RawOp {
    * @return a new instance of CSRSparseMatrixComponents
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> CSRSparseMatrixComponents<T> create(Scope scope, Operand<?> csrSparseMatrix, Operand<TInt32> index, DataType<T> type) {
+  public static <T extends Tensor> CSRSparseMatrixComponents<T> create(Scope scope, Operand<?> csrSparseMatrix, Operand<TInt32> index, DataType<T> type) {
     OperationBuilder opBuilder = scope.env().opBuilder("CSRSparseMatrixComponents", scope.makeOpName("CSRSparseMatrixComponents"));
     opBuilder.addInput(csrSparseMatrix.asOutput());
     opBuilder.addInput(index.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/CSRSparseMatrixToDense.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/CSRSparseMatrixToDense.java
index 0d8ceba40d3..c0242c74690 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/CSRSparseMatrixToDense.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/CSRSparseMatrixToDense.java
@@ -22,18 +22,18 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Convert a (possibly batched) CSRSparseMatrix to dense.
  * 
  * @param <T> data type for {@code denseOutput()} output
  */
-public final class CSRSparseMatrixToDense<T extends TType> extends RawOp implements Operand<T> {
+public final class CSRSparseMatrixToDense<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new CSRSparseMatrixToDense operation.
@@ -44,7 +44,7 @@ public final class CSRSparseMatrixToDense<T extends TType> extends RawOp impleme
    * @return a new instance of CSRSparseMatrixToDense
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> CSRSparseMatrixToDense<T> create(Scope scope, Operand<?> sparseInput, DataType<T> type) {
+  public static <T extends Tensor> CSRSparseMatrixToDense<T> create(Scope scope, Operand<?> sparseInput, DataType<T> type) {
     OperationBuilder opBuilder = scope.env().opBuilder("CSRSparseMatrixToDense", scope.makeOpName("CSRSparseMatrixToDense"));
     opBuilder.addInput(sparseInput.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/CSRSparseMatrixToSparseTensor.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/CSRSparseMatrixToSparseTensor.java
index 0b5a2d41df8..7a5b1c46ba7 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/CSRSparseMatrixToSparseTensor.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/CSRSparseMatrixToSparseTensor.java
@@ -22,19 +22,19 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
 
 /**
  * Converts a (possibly batched) CSRSparesMatrix to a SparseTensor.
  * 
  * @param <T> data type for {@code values()} output
  */
-public final class CSRSparseMatrixToSparseTensor<T extends TType> extends RawOp {
+public final class CSRSparseMatrixToSparseTensor<T extends Tensor> extends RawOp {
   
   /**
    * Factory method to create a class wrapping a new CSRSparseMatrixToSparseTensor operation.
@@ -45,7 +45,7 @@ public final class CSRSparseMatrixToSparseTensor<T extends TType> extends RawOp
    * @return a new instance of CSRSparseMatrixToSparseTensor
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> CSRSparseMatrixToSparseTensor<T> create(Scope scope, Operand<?> sparseMatrix, DataType<T> type) {
+  public static <T extends Tensor> CSRSparseMatrixToSparseTensor<T> create(Scope scope, Operand<?> sparseMatrix, DataType<T> type) {
     OperationBuilder opBuilder = scope.env().opBuilder("CSRSparseMatrixToSparseTensor", scope.makeOpName("CSRSparseMatrixToSparseTensor"));
     opBuilder.addInput(sparseMatrix.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/DenseToCSRSparseMatrix.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/DenseToCSRSparseMatrix.java
index 0033c9a9760..97102309d63 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/DenseToCSRSparseMatrix.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/DenseToCSRSparseMatrix.java
@@ -21,17 +21,17 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
 
 /**
  * Converts a dense tensor to a (possibly batched) CSRSparseMatrix.
  */
-public final class DenseToCSRSparseMatrix extends RawOp implements Operand<TType> {
+public final class DenseToCSRSparseMatrix extends RawOp implements Operand<Tensor> {
   
   /**
    * Factory method to create a class wrapping a new DenseToCSRSparseMatrix operation.
@@ -42,7 +42,7 @@ public final class DenseToCSRSparseMatrix extends RawOp implements Operand<TType
    * @return a new instance of DenseToCSRSparseMatrix
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> DenseToCSRSparseMatrix create(Scope scope, Operand<T> denseInput, Operand<TInt64> indices) {
+  public static <T extends Tensor> DenseToCSRSparseMatrix create(Scope scope, Operand<T> denseInput, Operand<TInt64> indices) {
     OperationBuilder opBuilder = scope.env().opBuilder("DenseToCSRSparseMatrix", scope.makeOpName("DenseToCSRSparseMatrix"));
     opBuilder.addInput(denseInput.asOutput());
     opBuilder.addInput(indices.asOutput());
@@ -59,8 +59,8 @@ public Output<?> sparseOutput() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) sparseOutput;
+  public Output<Tensor> asOutput() {
+    return (Output<Tensor>) sparseOutput;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixAdd.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixAdd.java
index d8e676743a1..3e011c703ec 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixAdd.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixAdd.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Sparse addition of two CSR matrices, C = alpha * A + beta * B.
@@ -33,7 +33,7 @@
  * The gradients of SparseMatrixAdd outputs with respect to alpha and beta are not
  * currently defined (TensorFlow will return zeros for these entries).
  */
-public final class SparseMatrixAdd extends RawOp implements Operand<TType> {
+public final class SparseMatrixAdd extends RawOp implements Operand<Tensor> {
   
   /**
    * Factory method to create a class wrapping a new SparseMatrixAdd operation.
@@ -46,7 +46,7 @@ public final class SparseMatrixAdd extends RawOp implements Operand<TType> {
    * @return a new instance of SparseMatrixAdd
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> SparseMatrixAdd create(Scope scope, Operand<?> a, Operand<?> b, Operand<T> alpha, Operand<T> beta) {
+  public static <T extends Tensor> SparseMatrixAdd create(Scope scope, Operand<?> a, Operand<?> b, Operand<T> alpha, Operand<T> beta) {
     OperationBuilder opBuilder = scope.env().opBuilder("SparseMatrixAdd", scope.makeOpName("SparseMatrixAdd"));
     opBuilder.addInput(a.asOutput());
     opBuilder.addInput(b.asOutput());
@@ -65,8 +65,8 @@ public Output<?> c() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) c;
+  public Output<Tensor> asOutput() {
+    return (Output<Tensor>) c;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixMatMul.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixMatMul.java
index 2ee48d29d79..ebe3a8bc5c1 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixMatMul.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixMatMul.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Matrix-multiplies a sparse matrix with a dense matrix.
@@ -57,7 +57,7 @@
  * 
  * @param <T> data type for {@code output()} output
  */
-public final class SparseMatrixMatMul<T extends TType> extends RawOp implements Operand<T> {
+public final class SparseMatrixMatMul<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.linalg.sparse.SparseMatrixMatMul}
@@ -133,7 +133,7 @@ private Options() {
    * @return a new instance of SparseMatrixMatMul
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> SparseMatrixMatMul<T> create(Scope scope, Operand<?> a, Operand<T> b, Options... options) {
+  public static <T extends Tensor> SparseMatrixMatMul<T> create(Scope scope, Operand<?> a, Operand<T> b, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("SparseMatrixMatMul", scope.makeOpName("SparseMatrixMatMul"));
     opBuilder.addInput(a.asOutput());
     opBuilder.addInput(b.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixMul.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixMul.java
index ad1571580f5..35589e32682 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixMul.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixMul.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Element-wise multiplication of a sparse matrix with a dense tensor.
@@ -39,7 +39,7 @@
  * <b>NOTE</b> even if `b` is zero, the sparsity structure of the output does not
  * change.
  */
-public final class SparseMatrixMul extends RawOp implements Operand<TType> {
+public final class SparseMatrixMul extends RawOp implements Operand<Tensor> {
   
   /**
    * Factory method to create a class wrapping a new SparseMatrixMul operation.
@@ -50,7 +50,7 @@ public final class SparseMatrixMul extends RawOp implements Operand<TType> {
    * @return a new instance of SparseMatrixMul
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> SparseMatrixMul create(Scope scope, Operand<?> a, Operand<T> b) {
+  public static <T extends Tensor> SparseMatrixMul create(Scope scope, Operand<?> a, Operand<T> b) {
     OperationBuilder opBuilder = scope.env().opBuilder("SparseMatrixMul", scope.makeOpName("SparseMatrixMul"));
     opBuilder.addInput(a.asOutput());
     opBuilder.addInput(b.asOutput());
@@ -67,8 +67,8 @@ public Output<?> output() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) output;
+  public Output<Tensor> asOutput() {
+    return (Output<Tensor>) output;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixSoftmax.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixSoftmax.java
index 01cec0f84da..9feb5bad1c9 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixSoftmax.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixSoftmax.java
@@ -22,12 +22,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Calculates the softmax of a CSRSparseMatrix.
@@ -38,7 +38,7 @@
  * the output has the same sparsity structure as the input (though missing values
  * in the output may now be treated as having probability zero).
  */
-public final class SparseMatrixSoftmax extends RawOp implements Operand<TType> {
+public final class SparseMatrixSoftmax extends RawOp implements Operand<Tensor> {
   
   /**
    * Factory method to create a class wrapping a new SparseMatrixSoftmax operation.
@@ -49,7 +49,7 @@ public final class SparseMatrixSoftmax extends RawOp implements Operand<TType> {
    * @return a new instance of SparseMatrixSoftmax
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> SparseMatrixSoftmax create(Scope scope, Operand<?> logits, DataType<T> type) {
+  public static <T extends Tensor & TNumber> SparseMatrixSoftmax create(Scope scope, Operand<?> logits, DataType<T> type) {
     OperationBuilder opBuilder = scope.env().opBuilder("SparseMatrixSoftmax", scope.makeOpName("SparseMatrixSoftmax"));
     opBuilder.addInput(logits.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
@@ -66,8 +66,8 @@ public Output<?> softmax() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) softmax;
+  public Output<Tensor> asOutput() {
+    return (Output<Tensor>) softmax;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixSoftmaxGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixSoftmaxGrad.java
index 8ac96d72177..508e4be550e 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixSoftmaxGrad.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixSoftmaxGrad.java
@@ -22,17 +22,17 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Calculates the gradient of the SparseMatrixSoftmax op.
  */
-public final class SparseMatrixSoftmaxGrad extends RawOp implements Operand<TType> {
+public final class SparseMatrixSoftmaxGrad extends RawOp implements Operand<Tensor> {
   
   /**
    * Factory method to create a class wrapping a new SparseMatrixSoftmaxGrad operation.
@@ -44,7 +44,7 @@ public final class SparseMatrixSoftmaxGrad extends RawOp implements Operand<TTyp
    * @return a new instance of SparseMatrixSoftmaxGrad
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> SparseMatrixSoftmaxGrad create(Scope scope, Operand<?> softmax, Operand<?> gradSoftmax, DataType<T> type) {
+  public static <T extends Tensor & TNumber> SparseMatrixSoftmaxGrad create(Scope scope, Operand<?> softmax, Operand<?> gradSoftmax, DataType<T> type) {
     OperationBuilder opBuilder = scope.env().opBuilder("SparseMatrixSoftmaxGrad", scope.makeOpName("SparseMatrixSoftmaxGrad"));
     opBuilder.addInput(softmax.asOutput());
     opBuilder.addInput(gradSoftmax.asOutput());
@@ -62,8 +62,8 @@ public Output<?> gradient() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) gradient;
+  public Output<Tensor> asOutput() {
+    return (Output<Tensor>) gradient;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixSparseCholesky.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixSparseCholesky.java
index 82f97e209d9..79f31716eb3 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixSparseCholesky.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixSparseCholesky.java
@@ -22,12 +22,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes the sparse Cholesky decomposition of `input`.
@@ -103,7 +103,7 @@
  * permutation: A `Tensor`.
  * type: The type of `input`.
  */
-public final class SparseMatrixSparseCholesky extends RawOp implements Operand<TType> {
+public final class SparseMatrixSparseCholesky extends RawOp implements Operand<Tensor> {
   
   /**
    * Factory method to create a class wrapping a new SparseMatrixSparseCholesky operation.
@@ -115,7 +115,7 @@ public final class SparseMatrixSparseCholesky extends RawOp implements Operand<T
    * @return a new instance of SparseMatrixSparseCholesky
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> SparseMatrixSparseCholesky create(Scope scope, Operand<?> input, Operand<TInt32> permutation, DataType<T> type) {
+  public static <T extends Tensor> SparseMatrixSparseCholesky create(Scope scope, Operand<?> input, Operand<TInt32> permutation, DataType<T> type) {
     OperationBuilder opBuilder = scope.env().opBuilder("SparseMatrixSparseCholesky", scope.makeOpName("SparseMatrixSparseCholesky"));
     opBuilder.addInput(input.asOutput());
     opBuilder.addInput(permutation.asOutput());
@@ -133,8 +133,8 @@ public Output<?> output() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) output;
+  public Output<Tensor> asOutput() {
+    return (Output<Tensor>) output;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixSparseMatMul.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixSparseMatMul.java
index 373ab3a01e2..a46d9a5a3e4 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixSparseMatMul.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixSparseMatMul.java
@@ -22,11 +22,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Sparse-matrix-multiplies two CSR matrices `a` and `b`.
@@ -104,7 +104,7 @@
  * adjoint_a: If True, `a` adjointed before multiplication.
  * adjoint_b: If True, `b` adjointed before multiplication.
  */
-public final class SparseMatrixSparseMatMul extends RawOp implements Operand<TType> {
+public final class SparseMatrixSparseMatMul extends RawOp implements Operand<Tensor> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.linalg.sparse.SparseMatrixSparseMatMul}
@@ -163,7 +163,7 @@ private Options() {
    * @return a new instance of SparseMatrixSparseMatMul
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> SparseMatrixSparseMatMul create(Scope scope, Operand<?> a, Operand<?> b, DataType<T> type, Options... options) {
+  public static <T extends Tensor> SparseMatrixSparseMatMul create(Scope scope, Operand<?> a, Operand<?> b, DataType<T> type, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("SparseMatrixSparseMatMul", scope.makeOpName("SparseMatrixSparseMatMul"));
     opBuilder.addInput(a.asOutput());
     opBuilder.addInput(b.asOutput());
@@ -225,8 +225,8 @@ public Output<?> c() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) c;
+  public Output<Tensor> asOutput() {
+    return (Output<Tensor>) c;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixTranspose.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixTranspose.java
index 90e5ba4294f..6a2260099c8 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixTranspose.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixTranspose.java
@@ -22,11 +22,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Transposes the inner (matrix) dimensions of a CSRSparseMatrix.
@@ -34,7 +34,7 @@
  * Transposes the inner (matrix) dimensions of a SparseMatrix and optionally
  * conjugates its values.
  */
-public final class SparseMatrixTranspose extends RawOp implements Operand<TType> {
+public final class SparseMatrixTranspose extends RawOp implements Operand<Tensor> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.linalg.sparse.SparseMatrixTranspose}
@@ -65,7 +65,7 @@ private Options() {
    * @return a new instance of SparseMatrixTranspose
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> SparseMatrixTranspose create(Scope scope, Operand<?> input, DataType<T> type, Options... options) {
+  public static <T extends Tensor> SparseMatrixTranspose create(Scope scope, Operand<?> input, DataType<T> type, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("SparseMatrixTranspose", scope.makeOpName("SparseMatrixTranspose"));
     opBuilder.addInput(input.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
@@ -96,8 +96,8 @@ public Output<?> output() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) output;
+  public Output<Tensor> asOutput() {
+    return (Output<Tensor>) output;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixZeros.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixZeros.java
index 3c5bef9daed..d1c70f3529c 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixZeros.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixZeros.java
@@ -22,17 +22,17 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
 
 /**
  * Creates an all-zeros CSRSparseMatrix with shape `dense_shape`.
  */
-public final class SparseMatrixZeros extends RawOp implements Operand<TType> {
+public final class SparseMatrixZeros extends RawOp implements Operand<Tensor> {
   
   /**
    * Factory method to create a class wrapping a new SparseMatrixZeros operation.
@@ -43,7 +43,7 @@ public final class SparseMatrixZeros extends RawOp implements Operand<TType> {
    * @return a new instance of SparseMatrixZeros
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> SparseMatrixZeros create(Scope scope, Operand<TInt64> denseShape, DataType<T> type) {
+  public static <T extends Tensor> SparseMatrixZeros create(Scope scope, Operand<TInt64> denseShape, DataType<T> type) {
     OperationBuilder opBuilder = scope.env().opBuilder("SparseMatrixZeros", scope.makeOpName("SparseMatrixZeros"));
     opBuilder.addInput(denseShape.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
@@ -60,8 +60,8 @@ public Output<?> sparseMatrix() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) sparseMatrix;
+  public Output<Tensor> asOutput() {
+    return (Output<Tensor>) sparseMatrix;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseTensorToCSRSparseMatrix.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseTensorToCSRSparseMatrix.java
index afda1e8b5f1..7f1f8397f07 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseTensorToCSRSparseMatrix.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseTensorToCSRSparseMatrix.java
@@ -21,17 +21,17 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
 
 /**
  * Converts a SparseTensor to a (possibly batched) CSRSparseMatrix.
  */
-public final class SparseTensorToCSRSparseMatrix extends RawOp implements Operand<TType> {
+public final class SparseTensorToCSRSparseMatrix extends RawOp implements Operand<Tensor> {
   
   /**
    * Factory method to create a class wrapping a new SparseTensorToCSRSparseMatrix operation.
@@ -43,7 +43,7 @@ public final class SparseTensorToCSRSparseMatrix extends RawOp implements Operan
    * @return a new instance of SparseTensorToCSRSparseMatrix
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> SparseTensorToCSRSparseMatrix create(Scope scope, Operand<TInt64> indices, Operand<T> values, Operand<TInt64> denseShape) {
+  public static <T extends Tensor> SparseTensorToCSRSparseMatrix create(Scope scope, Operand<TInt64> indices, Operand<T> values, Operand<TInt64> denseShape) {
     OperationBuilder opBuilder = scope.env().opBuilder("SparseTensorToCSRSparseMatrix", scope.makeOpName("SparseTensorToCSRSparseMatrix"));
     opBuilder.addInput(indices.asOutput());
     opBuilder.addInput(values.asOutput());
@@ -61,8 +61,8 @@ public Output<?> sparseMatrix() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) sparseMatrix;
+  public Output<Tensor> asOutput() {
+    return (Output<Tensor>) sparseMatrix;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Abs.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Abs.java
index 7aa760a535b..061abeeb85e 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Abs.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Abs.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes the absolute value of a tensor.
@@ -38,7 +38,7 @@
  * @param <T> data type for {@code y()} output
  */
 @Operator(group = "math")
-public final class Abs<T extends TNumber> extends RawOp implements Operand<T> {
+public final class Abs<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Abs operation.
@@ -48,7 +48,7 @@ public final class Abs<T extends TNumber> extends RawOp implements Operand<T> {
    * @return a new instance of Abs
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> Abs<T> create(Scope scope, Operand<T> x) {
+  public static <T extends Tensor & TNumber> Abs<T> create(Scope scope, Operand<T> x) {
     OperationBuilder opBuilder = scope.env().opBuilder("Abs", scope.makeOpName("Abs"));
     opBuilder.addInput(x.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/AccumulateN.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/AccumulateN.java
index dc15d9f9d81..41ef83e8a42 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/AccumulateN.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/AccumulateN.java
@@ -21,13 +21,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.Operands;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Returns the element-wise sum of a list of tensors.
@@ -44,7 +44,7 @@
  * @param <T> data type for {@code sum()} output
  */
 @Operator(group = "math")
-public final class AccumulateN<T extends TType> extends RawOp implements Operand<T> {
+public final class AccumulateN<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new AccumulateN operation.
@@ -55,7 +55,7 @@ public final class AccumulateN<T extends TType> extends RawOp implements Operand
    * @return a new instance of AccumulateN
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> AccumulateN<T> create(Scope scope, Iterable<Operand<T>> inputs, Shape shape) {
+  public static <T extends Tensor> AccumulateN<T> create(Scope scope, Iterable<Operand<T>> inputs, Shape shape) {
     OperationBuilder opBuilder = scope.env().opBuilder("AccumulateNV2", scope.makeOpName("AccumulateN"));
     opBuilder.addInputList(Operands.asOutputs(inputs));
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Acos.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Acos.java
index 43ba8e54334..ab784548c80 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Acos.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Acos.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes acos of x element-wise.
@@ -33,7 +33,7 @@
  * @param <T> data type for {@code y()} output
  */
 @Operator(group = "math")
-public final class Acos<T extends TType> extends RawOp implements Operand<T> {
+public final class Acos<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Acos operation.
@@ -43,7 +43,7 @@ public final class Acos<T extends TType> extends RawOp implements Operand<T> {
    * @return a new instance of Acos
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> Acos<T> create(Scope scope, Operand<T> x) {
+  public static <T extends Tensor> Acos<T> create(Scope scope, Operand<T> x) {
     OperationBuilder opBuilder = scope.env().opBuilder("Acos", scope.makeOpName("Acos"));
     opBuilder.addInput(x.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Acosh.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Acosh.java
index 0d1aaf95237..9b432abae7f 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Acosh.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Acosh.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes inverse hyperbolic cosine of x element-wise.
@@ -41,7 +41,7 @@
  * @param <T> data type for {@code y()} output
  */
 @Operator(group = "math")
-public final class Acosh<T extends TType> extends RawOp implements Operand<T> {
+public final class Acosh<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Acosh operation.
@@ -51,7 +51,7 @@ public final class Acosh<T extends TType> extends RawOp implements Operand<T> {
    * @return a new instance of Acosh
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> Acosh<T> create(Scope scope, Operand<T> x) {
+  public static <T extends Tensor> Acosh<T> create(Scope scope, Operand<T> x) {
     OperationBuilder opBuilder = scope.env().opBuilder("Acosh", scope.makeOpName("Acosh"));
     opBuilder.addInput(x.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Add.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Add.java
index 86d7c82ae38..33501435d2e 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Add.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Add.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Returns x + y element-wise.
@@ -36,7 +36,7 @@
  * @param <T> data type for {@code z()} output
  */
 @Operator(group = "math")
-public final class Add<T extends TType> extends RawOp implements Operand<T> {
+public final class Add<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Add operation.
@@ -47,7 +47,7 @@ public final class Add<T extends TType> extends RawOp implements Operand<T> {
    * @return a new instance of Add
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> Add<T> create(Scope scope, Operand<T> x, Operand<T> y) {
+  public static <T extends Tensor> Add<T> create(Scope scope, Operand<T> x, Operand<T> y) {
     OperationBuilder opBuilder = scope.env().opBuilder("Add", scope.makeOpName("Add"));
     opBuilder.addInput(x.asOutput());
     opBuilder.addInput(y.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/AddN.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/AddN.java
index 16a5e22ce16..6004bcc15cf 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/AddN.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/AddN.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.Operands;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Add all input tensors element wise.
@@ -42,7 +42,7 @@
  * @param <T> data type for {@code sum()} output
  */
 @Operator(group = "math")
-public final class AddN<T extends TType> extends RawOp implements Operand<T> {
+public final class AddN<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new AddN operation.
@@ -52,7 +52,7 @@ public final class AddN<T extends TType> extends RawOp implements Operand<T> {
    * @return a new instance of AddN
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> AddN<T> create(Scope scope, Iterable<Operand<T>> inputs) {
+  public static <T extends Tensor> AddN<T> create(Scope scope, Iterable<Operand<T>> inputs) {
     OperationBuilder opBuilder = scope.env().opBuilder("AddN", scope.makeOpName("AddN"));
     opBuilder.addInputList(Operands.asOutputs(inputs));
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Angle.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Angle.java
index ce996785568..dfc0413f4d4 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Angle.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Angle.java
@@ -22,13 +22,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TFloat32;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Returns the argument of a complex number.
@@ -52,7 +52,7 @@
  * @param <U> data type for {@code output()} output
  */
 @Operator(group = "math")
-public final class Angle<U extends TNumber> extends RawOp implements Operand<U> {
+public final class Angle<U extends Tensor & TNumber> extends RawOp implements Operand<U> {
   
   /**
    * Factory method to create a class wrapping a new Angle operation.
@@ -63,7 +63,7 @@ public final class Angle<U extends TNumber> extends RawOp implements Operand<U>
    * @return a new instance of Angle
    */
   @Endpoint(describeByClass = true)
-  public static <U extends TNumber, T extends TType> Angle<U> create(Scope scope, Operand<T> input, DataType<U> Tout) {
+  public static <U extends Tensor & TNumber, T extends Tensor> Angle<U> create(Scope scope, Operand<T> input, DataType<U> Tout) {
     OperationBuilder opBuilder = scope.env().opBuilder("Angle", scope.makeOpName("Angle"));
     opBuilder.addInput(input.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
@@ -79,7 +79,7 @@ public static <U extends TNumber, T extends TType> Angle<U> create(Scope scope,
    * @return a new instance of Angle
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> Angle<TFloat32> create(Scope scope, Operand<T> input) {
+  public static <T extends Tensor> Angle<TFloat32> create(Scope scope, Operand<T> input) {
     return create(scope, input, TFloat32.DTYPE);
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/ApproximateEqual.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/ApproximateEqual.java
index 245c42f269d..49c3062640b 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/ApproximateEqual.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/ApproximateEqual.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TBool;
-import org.tensorflow.types.family.TType;
 
 /**
  * Returns the truth value of abs(x-y) < tolerance element-wise.
@@ -63,7 +63,7 @@ private Options() {
    * @return a new instance of ApproximateEqual
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> ApproximateEqual create(Scope scope, Operand<T> x, Operand<T> y, Options... options) {
+  public static <T extends Tensor> ApproximateEqual create(Scope scope, Operand<T> x, Operand<T> y, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ApproximateEqual", scope.makeOpName("ApproximateEqual"));
     opBuilder.addInput(x.asOutput());
     opBuilder.addInput(y.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/ArgMax.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/ArgMax.java
index b45c1268cb2..145bad3d4b7 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/ArgMax.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/ArgMax.java
@@ -22,13 +22,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Returns the index with the largest value across dimensions of a tensor.
@@ -49,7 +49,7 @@
  * @param <V> data type for {@code output()} output
  */
 @Operator(group = "math")
-public final class ArgMax<V extends TNumber> extends RawOp implements Operand<V> {
+public final class ArgMax<V extends Tensor & TNumber> extends RawOp implements Operand<V> {
   
   /**
    * Factory method to create a class wrapping a new ArgMax operation.
@@ -63,7 +63,7 @@ public final class ArgMax<V extends TNumber> extends RawOp implements Operand<V>
    * @return a new instance of ArgMax
    */
   @Endpoint(describeByClass = true)
-  public static <V extends TNumber, T extends TType, U extends TNumber> ArgMax<V> create(Scope scope, Operand<T> input, Operand<U> dimension, DataType<V> outputType) {
+  public static <V extends Tensor & TNumber, T extends Tensor, U extends Tensor & TNumber> ArgMax<V> create(Scope scope, Operand<T> input, Operand<U> dimension, DataType<V> outputType) {
     OperationBuilder opBuilder = scope.env().opBuilder("ArgMax", scope.makeOpName("ArgMax"));
     opBuilder.addInput(input.asOutput());
     opBuilder.addInput(dimension.asOutput());
@@ -83,7 +83,7 @@ public static <V extends TNumber, T extends TType, U extends TNumber> ArgMax<V>
    * @return a new instance of ArgMax
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> ArgMax<TInt64> create(Scope scope, Operand<T> input, Operand<U> dimension) {
+  public static <T extends Tensor, U extends Tensor & TNumber> ArgMax<TInt64> create(Scope scope, Operand<T> input, Operand<U> dimension) {
     return create(scope, input, dimension, TInt64.DTYPE);
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/ArgMin.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/ArgMin.java
index 5a49adecd22..1392faa0ff3 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/ArgMin.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/ArgMin.java
@@ -22,13 +22,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Returns the index with the smallest value across dimensions of a tensor.
@@ -49,7 +49,7 @@
  * @param <V> data type for {@code output()} output
  */
 @Operator(group = "math")
-public final class ArgMin<V extends TNumber> extends RawOp implements Operand<V> {
+public final class ArgMin<V extends Tensor & TNumber> extends RawOp implements Operand<V> {
   
   /**
    * Factory method to create a class wrapping a new ArgMin operation.
@@ -63,7 +63,7 @@ public final class ArgMin<V extends TNumber> extends RawOp implements Operand<V>
    * @return a new instance of ArgMin
    */
   @Endpoint(describeByClass = true)
-  public static <V extends TNumber, T extends TType, U extends TNumber> ArgMin<V> create(Scope scope, Operand<T> input, Operand<U> dimension, DataType<V> outputType) {
+  public static <V extends Tensor & TNumber, T extends Tensor, U extends Tensor & TNumber> ArgMin<V> create(Scope scope, Operand<T> input, Operand<U> dimension, DataType<V> outputType) {
     OperationBuilder opBuilder = scope.env().opBuilder("ArgMin", scope.makeOpName("ArgMin"));
     opBuilder.addInput(input.asOutput());
     opBuilder.addInput(dimension.asOutput());
@@ -83,7 +83,7 @@ public static <V extends TNumber, T extends TType, U extends TNumber> ArgMin<V>
    * @return a new instance of ArgMin
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> ArgMin<TInt64> create(Scope scope, Operand<T> input, Operand<U> dimension) {
+  public static <T extends Tensor, U extends Tensor & TNumber> ArgMin<TInt64> create(Scope scope, Operand<T> input, Operand<U> dimension) {
     return create(scope, input, dimension, TInt64.DTYPE);
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Asin.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Asin.java
index 941515b013e..43f0fe31313 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Asin.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Asin.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes the trignometric inverse sine of x element-wise.
@@ -49,7 +49,7 @@
  * @param <T> data type for {@code y()} output
  */
 @Operator(group = "math")
-public final class Asin<T extends TType> extends RawOp implements Operand<T> {
+public final class Asin<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Asin operation.
@@ -59,7 +59,7 @@ public final class Asin<T extends TType> extends RawOp implements Operand<T> {
    * @return a new instance of Asin
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> Asin<T> create(Scope scope, Operand<T> x) {
+  public static <T extends Tensor> Asin<T> create(Scope scope, Operand<T> x) {
     OperationBuilder opBuilder = scope.env().opBuilder("Asin", scope.makeOpName("Asin"));
     opBuilder.addInput(x.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Asinh.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Asinh.java
index c2451727573..a306e04d9b2 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Asinh.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Asinh.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes inverse hyperbolic sine of x element-wise.
@@ -43,7 +43,7 @@
  * @param <T> data type for {@code y()} output
  */
 @Operator(group = "math")
-public final class Asinh<T extends TType> extends RawOp implements Operand<T> {
+public final class Asinh<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Asinh operation.
@@ -53,7 +53,7 @@ public final class Asinh<T extends TType> extends RawOp implements Operand<T> {
    * @return a new instance of Asinh
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> Asinh<T> create(Scope scope, Operand<T> x) {
+  public static <T extends Tensor> Asinh<T> create(Scope scope, Operand<T> x) {
     OperationBuilder opBuilder = scope.env().opBuilder("Asinh", scope.makeOpName("Asinh"));
     opBuilder.addInput(x.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Atan.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Atan.java
index a3574d549bc..52c91362608 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Atan.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Atan.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes the trignometric inverse tangent of x element-wise.
@@ -49,7 +49,7 @@
  * @param <T> data type for {@code y()} output
  */
 @Operator(group = "math")
-public final class Atan<T extends TType> extends RawOp implements Operand<T> {
+public final class Atan<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Atan operation.
@@ -59,7 +59,7 @@ public final class Atan<T extends TType> extends RawOp implements Operand<T> {
    * @return a new instance of Atan
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> Atan<T> create(Scope scope, Operand<T> x) {
+  public static <T extends Tensor> Atan<T> create(Scope scope, Operand<T> x) {
     OperationBuilder opBuilder = scope.env().opBuilder("Atan", scope.makeOpName("Atan"));
     opBuilder.addInput(x.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Atan2.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Atan2.java
index d49bc324cb3..6b96355ca7c 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Atan2.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Atan2.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes arctangent of `y/x` element-wise, respecting signs of the arguments.
@@ -40,7 +40,7 @@
  * @param <T> data type for {@code z()} output
  */
 @Operator(group = "math")
-public final class Atan2<T extends TNumber> extends RawOp implements Operand<T> {
+public final class Atan2<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Atan2 operation.
@@ -51,7 +51,7 @@ public final class Atan2<T extends TNumber> extends RawOp implements Operand<T>
    * @return a new instance of Atan2
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> Atan2<T> create(Scope scope, Operand<T> y, Operand<T> x) {
+  public static <T extends Tensor & TNumber> Atan2<T> create(Scope scope, Operand<T> y, Operand<T> x) {
     OperationBuilder opBuilder = scope.env().opBuilder("Atan2", scope.makeOpName("Atan2"));
     opBuilder.addInput(y.asOutput());
     opBuilder.addInput(x.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Atanh.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Atanh.java
index 6c53c8828c0..222afb87f9f 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Atanh.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Atanh.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes inverse hyperbolic tangent of x element-wise.
@@ -45,7 +45,7 @@
  * @param <T> data type for {@code y()} output
  */
 @Operator(group = "math")
-public final class Atanh<T extends TType> extends RawOp implements Operand<T> {
+public final class Atanh<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Atanh operation.
@@ -55,7 +55,7 @@ public final class Atanh<T extends TType> extends RawOp implements Operand<T> {
    * @return a new instance of Atanh
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> Atanh<T> create(Scope scope, Operand<T> x) {
+  public static <T extends Tensor> Atanh<T> create(Scope scope, Operand<T> x) {
     OperationBuilder opBuilder = scope.env().opBuilder("Atanh", scope.makeOpName("Atanh"));
     opBuilder.addInput(x.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/BesselI0e.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/BesselI0e.java
index 816933ceb12..267c801fe6b 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/BesselI0e.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/BesselI0e.java
@@ -21,17 +21,25 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
+ * Computes the Bessel i0e function of `x` element-wise.
+ * <p>
+ * Exponentially scaled modified Bessel function of order 0 defined as
+ * `bessel_i0e(x) = exp(-abs(x)) bessel_i0(x)`.
+ * <p>
+ * This function is faster and numerically stabler than `bessel_i0(x)`.
+ * 
  * @param <T> data type for {@code y()} output
  */
-public final class BesselI0e<T extends TNumber> extends RawOp implements Operand<T> {
+@Operator(group = "math")
+public final class BesselI0e<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new BesselI0e operation.
@@ -41,7 +49,7 @@ public final class BesselI0e<T extends TNumber> extends RawOp implements Operand
    * @return a new instance of BesselI0e
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> BesselI0e<T> create(Scope scope, Operand<T> x) {
+  public static <T extends Tensor & TNumber> BesselI0e<T> create(Scope scope, Operand<T> x) {
     OperationBuilder opBuilder = scope.env().opBuilder("BesselI0e", scope.makeOpName("BesselI0e"));
     opBuilder.addInput(x.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
@@ -59,9 +67,6 @@ public Output<T> asOutput() {
     return y;
   }
   
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "BesselI0e";
-  
   private Output<T> y;
   
   private BesselI0e(Operation operation) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/BesselI1e.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/BesselI1e.java
index 3529c4a0bed..ca7571067b0 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/BesselI1e.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/BesselI1e.java
@@ -21,17 +21,25 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
+ * Computes the Bessel i1e function of `x` element-wise.
+ * <p>
+ * Exponentially scaled modified Bessel function of order 0 defined as
+ * `bessel_i1e(x) = exp(-abs(x)) bessel_i1(x)`.
+ * <p>
+ * This function is faster and numerically stabler than `bessel_i1(x)`.
+ * 
  * @param <T> data type for {@code y()} output
  */
-public final class BesselI1e<T extends TNumber> extends RawOp implements Operand<T> {
+@Operator(group = "math")
+public final class BesselI1e<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new BesselI1e operation.
@@ -41,7 +49,7 @@ public final class BesselI1e<T extends TNumber> extends RawOp implements Operand
    * @return a new instance of BesselI1e
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> BesselI1e<T> create(Scope scope, Operand<T> x) {
+  public static <T extends Tensor & TNumber> BesselI1e<T> create(Scope scope, Operand<T> x) {
     OperationBuilder opBuilder = scope.env().opBuilder("BesselI1e", scope.makeOpName("BesselI1e"));
     opBuilder.addInput(x.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
@@ -59,9 +67,6 @@ public Output<T> asOutput() {
     return y;
   }
   
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "BesselI1e";
-  
   private Output<T> y;
   
   private BesselI1e(Operation operation) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Betainc.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Betainc.java
index 53ecd2ed396..b252886b1c4 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Betainc.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Betainc.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Compute the regularized incomplete beta integral \\(I_x(a, b)\\).
@@ -45,7 +45,7 @@
  * @param <T> data type for {@code z()} output
  */
 @Operator(group = "math")
-public final class Betainc<T extends TNumber> extends RawOp implements Operand<T> {
+public final class Betainc<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Betainc operation.
@@ -57,7 +57,7 @@ public final class Betainc<T extends TNumber> extends RawOp implements Operand<T
    * @return a new instance of Betainc
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> Betainc<T> create(Scope scope, Operand<T> a, Operand<T> b, Operand<T> x) {
+  public static <T extends Tensor & TNumber> Betainc<T> create(Scope scope, Operand<T> a, Operand<T> b, Operand<T> x) {
     OperationBuilder opBuilder = scope.env().opBuilder("Betainc", scope.makeOpName("Betainc"));
     opBuilder.addInput(a.asOutput());
     opBuilder.addInput(b.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Bincount.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Bincount.java
index 4217584c4fb..dde12f54d47 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Bincount.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Bincount.java
@@ -21,13 +21,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Counts the number of occurrences of each value in an integer array.
@@ -43,7 +43,7 @@
  * @param <T> data type for {@code bins()} output
  */
 @Operator(group = "math")
-public final class Bincount<T extends TNumber> extends RawOp implements Operand<T> {
+public final class Bincount<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Bincount operation.
@@ -57,7 +57,7 @@ public final class Bincount<T extends TNumber> extends RawOp implements Operand<
    * @return a new instance of Bincount
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> Bincount<T> create(Scope scope, Operand<TInt32> arr, Operand<TInt32> size, Operand<T> weights) {
+  public static <T extends Tensor & TNumber> Bincount<T> create(Scope scope, Operand<TInt32> arr, Operand<TInt32> size, Operand<T> weights) {
     OperationBuilder opBuilder = scope.env().opBuilder("Bincount", scope.makeOpName("Bincount"));
     opBuilder.addInput(arr.asOutput());
     opBuilder.addInput(size.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Ceil.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Ceil.java
index c97e5b5ae5d..2d2d097c036 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Ceil.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Ceil.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Returns element-wise smallest integer not less than x.
@@ -34,7 +34,7 @@
  * @param <T> data type for {@code y()} output
  */
 @Operator(group = "math")
-public final class Ceil<T extends TNumber> extends RawOp implements Operand<T> {
+public final class Ceil<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Ceil operation.
@@ -44,7 +44,7 @@ public final class Ceil<T extends TNumber> extends RawOp implements Operand<T> {
    * @return a new instance of Ceil
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> Ceil<T> create(Scope scope, Operand<T> x) {
+  public static <T extends Tensor & TNumber> Ceil<T> create(Scope scope, Operand<T> x) {
     OperationBuilder opBuilder = scope.env().opBuilder("Ceil", scope.makeOpName("Ceil"));
     opBuilder.addInput(x.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/CompareAndBitpack.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/CompareAndBitpack.java
index 2538cc59248..84cf87274a2 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/CompareAndBitpack.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/CompareAndBitpack.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TUint8;
-import org.tensorflow.types.family.TType;
 
 /**
  * Compare values of `input` to `threshold` and pack resulting bits into a `uint8`.
@@ -65,7 +65,7 @@ public final class CompareAndBitpack extends RawOp implements Operand<TUint8> {
    * @return a new instance of CompareAndBitpack
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> CompareAndBitpack create(Scope scope, Operand<T> input, Operand<T> threshold) {
+  public static <T extends Tensor> CompareAndBitpack create(Scope scope, Operand<T> input, Operand<T> threshold) {
     OperationBuilder opBuilder = scope.env().opBuilder("CompareAndBitpack", scope.makeOpName("CompareAndBitpack"));
     opBuilder.addInput(input.asOutput());
     opBuilder.addInput(threshold.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/ComplexAbs.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/ComplexAbs.java
index 317744e519a..d1c7f2c3d25 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/ComplexAbs.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/ComplexAbs.java
@@ -22,13 +22,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TFloat32;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes the complex absolute value of a tensor.
@@ -41,7 +41,7 @@
  * @param <U> data type for {@code y()} output
  */
 @Operator(group = "math")
-public final class ComplexAbs<U extends TNumber> extends RawOp implements Operand<U> {
+public final class ComplexAbs<U extends Tensor & TNumber> extends RawOp implements Operand<U> {
   
   /**
    * Factory method to create a class wrapping a new ComplexAbs operation.
@@ -52,7 +52,7 @@ public final class ComplexAbs<U extends TNumber> extends RawOp implements Operan
    * @return a new instance of ComplexAbs
    */
   @Endpoint(describeByClass = true)
-  public static <U extends TNumber, T extends TType> ComplexAbs<U> create(Scope scope, Operand<T> x, DataType<U> Tout) {
+  public static <U extends Tensor & TNumber, T extends Tensor> ComplexAbs<U> create(Scope scope, Operand<T> x, DataType<U> Tout) {
     OperationBuilder opBuilder = scope.env().opBuilder("ComplexAbs", scope.makeOpName("ComplexAbs"));
     opBuilder.addInput(x.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
@@ -68,7 +68,7 @@ public static <U extends TNumber, T extends TType> ComplexAbs<U> create(Scope sc
    * @return a new instance of ComplexAbs
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> ComplexAbs<TFloat32> create(Scope scope, Operand<T> x) {
+  public static <T extends Tensor> ComplexAbs<TFloat32> create(Scope scope, Operand<T> x) {
     return create(scope, x, TFloat32.DTYPE);
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Conj.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Conj.java
index b565fe9d322..570f4bc5f29 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Conj.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Conj.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Returns the complex conjugate of a complex number.
@@ -47,7 +47,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "math")
-public final class Conj<T extends TType> extends RawOp implements Operand<T> {
+public final class Conj<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Conj operation.
@@ -57,7 +57,7 @@ public final class Conj<T extends TType> extends RawOp implements Operand<T> {
    * @return a new instance of Conj
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> Conj<T> create(Scope scope, Operand<T> input) {
+  public static <T extends Tensor> Conj<T> create(Scope scope, Operand<T> input) {
     OperationBuilder opBuilder = scope.env().opBuilder("Conj", scope.makeOpName("Conj"));
     opBuilder.addInput(input.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Cos.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Cos.java
index 29f81972b17..4bd6ef72d81 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Cos.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Cos.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes cos of x element-wise.
@@ -44,7 +44,7 @@
  * @param <T> data type for {@code y()} output
  */
 @Operator(group = "math")
-public final class Cos<T extends TType> extends RawOp implements Operand<T> {
+public final class Cos<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Cos operation.
@@ -54,7 +54,7 @@ public final class Cos<T extends TType> extends RawOp implements Operand<T> {
    * @return a new instance of Cos
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> Cos<T> create(Scope scope, Operand<T> x) {
+  public static <T extends Tensor> Cos<T> create(Scope scope, Operand<T> x) {
     OperationBuilder opBuilder = scope.env().opBuilder("Cos", scope.makeOpName("Cos"));
     opBuilder.addInput(x.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Cosh.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Cosh.java
index f65e3e952ea..8dbde201831 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Cosh.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Cosh.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes hyperbolic cosine of x element-wise.
@@ -43,7 +43,7 @@
  * @param <T> data type for {@code y()} output
  */
 @Operator(group = "math")
-public final class Cosh<T extends TType> extends RawOp implements Operand<T> {
+public final class Cosh<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Cosh operation.
@@ -53,7 +53,7 @@ public final class Cosh<T extends TType> extends RawOp implements Operand<T> {
    * @return a new instance of Cosh
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> Cosh<T> create(Scope scope, Operand<T> x) {
+  public static <T extends Tensor> Cosh<T> create(Scope scope, Operand<T> x) {
     OperationBuilder opBuilder = scope.env().opBuilder("Cosh", scope.makeOpName("Cosh"));
     opBuilder.addInput(x.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Cumprod.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Cumprod.java
index 52154403d72..f7eb392a1a4 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Cumprod.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Cumprod.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Compute the cumulative product of the tensor `x` along `axis`.
@@ -57,7 +57,7 @@
  * @param <T> data type for {@code out()} output
  */
 @Operator(group = "math")
-public final class Cumprod<T extends TType> extends RawOp implements Operand<T> {
+public final class Cumprod<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.math.Cumprod}
@@ -100,7 +100,7 @@ private Options() {
    * @return a new instance of Cumprod
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> Cumprod<T> create(Scope scope, Operand<T> x, Operand<U> axis, Options... options) {
+  public static <T extends Tensor, U extends Tensor & TNumber> Cumprod<T> create(Scope scope, Operand<T> x, Operand<U> axis, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("Cumprod", scope.makeOpName("Cumprod"));
     opBuilder.addInput(x.asOutput());
     opBuilder.addInput(axis.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Cumsum.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Cumsum.java
index 6895cbed39f..a2570547464 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Cumsum.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Cumsum.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Compute the cumulative sum of the tensor `x` along `axis`.
@@ -57,7 +57,7 @@
  * @param <T> data type for {@code out()} output
  */
 @Operator(group = "math")
-public final class Cumsum<T extends TType> extends RawOp implements Operand<T> {
+public final class Cumsum<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.math.Cumsum}
@@ -100,7 +100,7 @@ private Options() {
    * @return a new instance of Cumsum
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> Cumsum<T> create(Scope scope, Operand<T> x, Operand<U> axis, Options... options) {
+  public static <T extends Tensor, U extends Tensor & TNumber> Cumsum<T> create(Scope scope, Operand<T> x, Operand<U> axis, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("Cumsum", scope.makeOpName("Cumsum"));
     opBuilder.addInput(x.asOutput());
     opBuilder.addInput(axis.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/CumulativeLogsumexp.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/CumulativeLogsumexp.java
index 3c8e6f89870..2f741650242 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/CumulativeLogsumexp.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/CumulativeLogsumexp.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Compute the cumulative product of the tensor `x` along `axis`.
@@ -51,7 +51,7 @@
  * 
  * @param <T> data type for {@code out()} output
  */
-public final class CumulativeLogsumexp<T extends TNumber> extends RawOp implements Operand<T> {
+public final class CumulativeLogsumexp<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.math.CumulativeLogsumexp}
@@ -92,7 +92,7 @@ private Options() {
    * @return a new instance of CumulativeLogsumexp
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber, U extends TNumber> CumulativeLogsumexp<T> create(Scope scope, Operand<T> x, Operand<U> axis, Options... options) {
+  public static <T extends Tensor & TNumber, U extends Tensor & TNumber> CumulativeLogsumexp<T> create(Scope scope, Operand<T> x, Operand<U> axis, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("CumulativeLogsumexp", scope.makeOpName("CumulativeLogsumexp"));
     opBuilder.addInput(x.asOutput());
     opBuilder.addInput(axis.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Digamma.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Digamma.java
index 2fd494f637c..dde2fb8eb97 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Digamma.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Digamma.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes Psi, the derivative of Lgamma (the log of the absolute value of
@@ -36,7 +36,7 @@
  * @param <T> data type for {@code y()} output
  */
 @Operator(group = "math")
-public final class Digamma<T extends TNumber> extends RawOp implements Operand<T> {
+public final class Digamma<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Digamma operation.
@@ -46,7 +46,7 @@ public final class Digamma<T extends TNumber> extends RawOp implements Operand<T
    * @return a new instance of Digamma
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> Digamma<T> create(Scope scope, Operand<T> x) {
+  public static <T extends Tensor & TNumber> Digamma<T> create(Scope scope, Operand<T> x) {
     OperationBuilder opBuilder = scope.env().opBuilder("Digamma", scope.makeOpName("Digamma"));
     opBuilder.addInput(x.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Div.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Div.java
index 0cd80db04df..6a9962e93d4 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Div.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Div.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Returns x / y element-wise.
@@ -36,7 +36,7 @@
  * @param <T> data type for {@code z()} output
  */
 @Operator(group = "math")
-public final class Div<T extends TType> extends RawOp implements Operand<T> {
+public final class Div<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Div operation.
@@ -47,7 +47,7 @@ public final class Div<T extends TType> extends RawOp implements Operand<T> {
    * @return a new instance of Div
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> Div<T> create(Scope scope, Operand<T> x, Operand<T> y) {
+  public static <T extends Tensor> Div<T> create(Scope scope, Operand<T> x, Operand<T> y) {
     OperationBuilder opBuilder = scope.env().opBuilder("Div", scope.makeOpName("Div"));
     opBuilder.addInput(x.asOutput());
     opBuilder.addInput(y.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/DivNoNan.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/DivNoNan.java
index 7a1ea66a50d..913a118cdb1 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/DivNoNan.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/DivNoNan.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Returns 0 if the denominator is zero.
@@ -37,7 +37,7 @@
  * @param <T> data type for {@code z()} output
  */
 @Operator(group = "math")
-public final class DivNoNan<T extends TType> extends RawOp implements Operand<T> {
+public final class DivNoNan<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new DivNoNan operation.
@@ -48,7 +48,7 @@ public final class DivNoNan<T extends TType> extends RawOp implements Operand<T>
    * @return a new instance of DivNoNan
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> DivNoNan<T> create(Scope scope, Operand<T> x, Operand<T> y) {
+  public static <T extends Tensor> DivNoNan<T> create(Scope scope, Operand<T> x, Operand<T> y) {
     OperationBuilder opBuilder = scope.env().opBuilder("DivNoNan", scope.makeOpName("DivNoNan"));
     opBuilder.addInput(x.asOutput());
     opBuilder.addInput(y.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Equal.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Equal.java
index d67b38e9ca6..81e0e7ce04c 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Equal.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Equal.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TBool;
-import org.tensorflow.types.family.TType;
 
 /**
  * Returns the truth value of (x == y) element-wise.
@@ -76,7 +76,7 @@ private Options() {
    * @return a new instance of Equal
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> Equal create(Scope scope, Operand<T> x, Operand<T> y, Options... options) {
+  public static <T extends Tensor> Equal create(Scope scope, Operand<T> x, Operand<T> y, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("Equal", scope.makeOpName("Equal"));
     opBuilder.addInput(x.asOutput());
     opBuilder.addInput(y.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Erf.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Erf.java
index 9f86126bff4..9ef35983233 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Erf.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Erf.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes the Gauss error function of `x` element-wise.
@@ -34,7 +34,7 @@
  * @param <T> data type for {@code y()} output
  */
 @Operator(group = "math")
-public final class Erf<T extends TNumber> extends RawOp implements Operand<T> {
+public final class Erf<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Erf operation.
@@ -44,7 +44,7 @@ public final class Erf<T extends TNumber> extends RawOp implements Operand<T> {
    * @return a new instance of Erf
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> Erf<T> create(Scope scope, Operand<T> x) {
+  public static <T extends Tensor & TNumber> Erf<T> create(Scope scope, Operand<T> x) {
     OperationBuilder opBuilder = scope.env().opBuilder("Erf", scope.makeOpName("Erf"));
     opBuilder.addInput(x.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Erfc.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Erfc.java
index b94fecf9ede..9910632e7ff 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Erfc.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Erfc.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes the complementary error function of `x` element-wise.
@@ -34,7 +34,7 @@
  * @param <T> data type for {@code y()} output
  */
 @Operator(group = "math")
-public final class Erfc<T extends TNumber> extends RawOp implements Operand<T> {
+public final class Erfc<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Erfc operation.
@@ -44,7 +44,7 @@ public final class Erfc<T extends TNumber> extends RawOp implements Operand<T> {
    * @return a new instance of Erfc
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> Erfc<T> create(Scope scope, Operand<T> x) {
+  public static <T extends Tensor & TNumber> Erfc<T> create(Scope scope, Operand<T> x) {
     OperationBuilder opBuilder = scope.env().opBuilder("Erfc", scope.makeOpName("Erfc"));
     opBuilder.addInput(x.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Exp.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Exp.java
index cee48319f26..409e0d4124b 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Exp.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Exp.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes exponential of x element-wise.  \\(y = e^x\\).
@@ -59,7 +59,7 @@
  * @param <T> data type for {@code y()} output
  */
 @Operator(group = "math")
-public final class Exp<T extends TType> extends RawOp implements Operand<T> {
+public final class Exp<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Exp operation.
@@ -69,7 +69,7 @@ public final class Exp<T extends TType> extends RawOp implements Operand<T> {
    * @return a new instance of Exp
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> Exp<T> create(Scope scope, Operand<T> x) {
+  public static <T extends Tensor> Exp<T> create(Scope scope, Operand<T> x) {
     OperationBuilder opBuilder = scope.env().opBuilder("Exp", scope.makeOpName("Exp"));
     opBuilder.addInput(x.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Expm1.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Expm1.java
index 9d76f16b774..fe81e5afcf4 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Expm1.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Expm1.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes `exp(x) - 1` element-wise.
@@ -48,7 +48,7 @@
  * @param <T> data type for {@code y()} output
  */
 @Operator(group = "math")
-public final class Expm1<T extends TType> extends RawOp implements Operand<T> {
+public final class Expm1<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Expm1 operation.
@@ -58,7 +58,7 @@ public final class Expm1<T extends TType> extends RawOp implements Operand<T> {
    * @return a new instance of Expm1
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> Expm1<T> create(Scope scope, Operand<T> x) {
+  public static <T extends Tensor> Expm1<T> create(Scope scope, Operand<T> x) {
     OperationBuilder opBuilder = scope.env().opBuilder("Expm1", scope.makeOpName("Expm1"));
     opBuilder.addInput(x.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Floor.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Floor.java
index ac8d7a8ffba..924833d7f22 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Floor.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Floor.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Returns element-wise largest integer not greater than x.
@@ -34,7 +34,7 @@
  * @param <T> data type for {@code y()} output
  */
 @Operator(group = "math")
-public final class Floor<T extends TNumber> extends RawOp implements Operand<T> {
+public final class Floor<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Floor operation.
@@ -44,7 +44,7 @@ public final class Floor<T extends TNumber> extends RawOp implements Operand<T>
    * @return a new instance of Floor
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> Floor<T> create(Scope scope, Operand<T> x) {
+  public static <T extends Tensor & TNumber> Floor<T> create(Scope scope, Operand<T> x) {
     OperationBuilder opBuilder = scope.env().opBuilder("Floor", scope.makeOpName("Floor"));
     opBuilder.addInput(x.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/FloorDiv.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/FloorDiv.java
index 366b88bc6d9..cb7148d2e7b 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/FloorDiv.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/FloorDiv.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Returns x // y element-wise.
@@ -36,7 +36,7 @@
  * @param <T> data type for {@code z()} output
  */
 @Operator(group = "math")
-public final class FloorDiv<T extends TType> extends RawOp implements Operand<T> {
+public final class FloorDiv<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new FloorDiv operation.
@@ -47,7 +47,7 @@ public final class FloorDiv<T extends TType> extends RawOp implements Operand<T>
    * @return a new instance of FloorDiv
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> FloorDiv<T> create(Scope scope, Operand<T> x, Operand<T> y) {
+  public static <T extends Tensor> FloorDiv<T> create(Scope scope, Operand<T> x, Operand<T> y) {
     OperationBuilder opBuilder = scope.env().opBuilder("FloorDiv", scope.makeOpName("FloorDiv"));
     opBuilder.addInput(x.asOutput());
     opBuilder.addInput(y.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/FloorMod.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/FloorMod.java
index 715628b57a3..a5fcf94ac58 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/FloorMod.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/FloorMod.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Returns element-wise remainder of division. When `x < 0` xor `y < 0` is
@@ -40,7 +40,7 @@
  * @param <T> data type for {@code z()} output
  */
 @Operator(group = "math")
-public final class FloorMod<T extends TNumber> extends RawOp implements Operand<T> {
+public final class FloorMod<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new FloorMod operation.
@@ -51,7 +51,7 @@ public final class FloorMod<T extends TNumber> extends RawOp implements Operand<
    * @return a new instance of FloorMod
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> FloorMod<T> create(Scope scope, Operand<T> x, Operand<T> y) {
+  public static <T extends Tensor & TNumber> FloorMod<T> create(Scope scope, Operand<T> x, Operand<T> y) {
     OperationBuilder opBuilder = scope.env().opBuilder("FloorMod", scope.makeOpName("FloorMod"));
     opBuilder.addInput(x.asOutput());
     opBuilder.addInput(y.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Greater.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Greater.java
index e8ef3811fd1..7d1df45c855 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Greater.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Greater.java
@@ -21,13 +21,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TBool;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Returns the truth value of (x > y) element-wise.
@@ -59,7 +59,7 @@ public final class Greater extends RawOp implements Operand<TBool> {
    * @return a new instance of Greater
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> Greater create(Scope scope, Operand<T> x, Operand<T> y) {
+  public static <T extends Tensor & TNumber> Greater create(Scope scope, Operand<T> x, Operand<T> y) {
     OperationBuilder opBuilder = scope.env().opBuilder("Greater", scope.makeOpName("Greater"));
     opBuilder.addInput(x.asOutput());
     opBuilder.addInput(y.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/GreaterEqual.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/GreaterEqual.java
index 11a83743031..b2b0cebbd12 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/GreaterEqual.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/GreaterEqual.java
@@ -21,13 +21,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TBool;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Returns the truth value of (x >= y) element-wise.
@@ -59,7 +59,7 @@ public final class GreaterEqual extends RawOp implements Operand<TBool> {
    * @return a new instance of GreaterEqual
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> GreaterEqual create(Scope scope, Operand<T> x, Operand<T> y) {
+  public static <T extends Tensor & TNumber> GreaterEqual create(Scope scope, Operand<T> x, Operand<T> y) {
     OperationBuilder opBuilder = scope.env().opBuilder("GreaterEqual", scope.makeOpName("GreaterEqual"));
     opBuilder.addInput(x.asOutput());
     opBuilder.addInput(y.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Igamma.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Igamma.java
index a4687ef3ab1..a0427886360 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Igamma.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Igamma.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Compute the lower regularized incomplete Gamma function `P(a, x)`.
@@ -47,7 +47,7 @@
  * @param <T> data type for {@code z()} output
  */
 @Operator(group = "math")
-public final class Igamma<T extends TNumber> extends RawOp implements Operand<T> {
+public final class Igamma<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Igamma operation.
@@ -58,7 +58,7 @@ public final class Igamma<T extends TNumber> extends RawOp implements Operand<T>
    * @return a new instance of Igamma
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> Igamma<T> create(Scope scope, Operand<T> a, Operand<T> x) {
+  public static <T extends Tensor & TNumber> Igamma<T> create(Scope scope, Operand<T> a, Operand<T> x) {
     OperationBuilder opBuilder = scope.env().opBuilder("Igamma", scope.makeOpName("Igamma"));
     opBuilder.addInput(a.asOutput());
     opBuilder.addInput(x.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/IgammaGradA.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/IgammaGradA.java
index e57b8d886e1..4938de73273 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/IgammaGradA.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/IgammaGradA.java
@@ -21,19 +21,19 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes the gradient of `igamma(a, x)` wrt `a`.
  * 
  * @param <T> data type for {@code z()} output
  */
-public final class IgammaGradA<T extends TNumber> extends RawOp implements Operand<T> {
+public final class IgammaGradA<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new IgammaGradA operation.
@@ -44,7 +44,7 @@ public final class IgammaGradA<T extends TNumber> extends RawOp implements Opera
    * @return a new instance of IgammaGradA
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> IgammaGradA<T> create(Scope scope, Operand<T> a, Operand<T> x) {
+  public static <T extends Tensor & TNumber> IgammaGradA<T> create(Scope scope, Operand<T> a, Operand<T> x) {
     OperationBuilder opBuilder = scope.env().opBuilder("IgammaGradA", scope.makeOpName("IgammaGradA"));
     opBuilder.addInput(a.asOutput());
     opBuilder.addInput(x.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Igammac.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Igammac.java
index 1e2cb0cf0f7..6a0053183c0 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Igammac.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Igammac.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Compute the upper regularized incomplete Gamma function `Q(a, x)`.
@@ -47,7 +47,7 @@
  * @param <T> data type for {@code z()} output
  */
 @Operator(group = "math")
-public final class Igammac<T extends TNumber> extends RawOp implements Operand<T> {
+public final class Igammac<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Igammac operation.
@@ -58,7 +58,7 @@ public final class Igammac<T extends TNumber> extends RawOp implements Operand<T
    * @return a new instance of Igammac
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> Igammac<T> create(Scope scope, Operand<T> a, Operand<T> x) {
+  public static <T extends Tensor & TNumber> Igammac<T> create(Scope scope, Operand<T> a, Operand<T> x) {
     OperationBuilder opBuilder = scope.env().opBuilder("Igammac", scope.makeOpName("Igammac"));
     opBuilder.addInput(a.asOutput());
     opBuilder.addInput(x.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Imag.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Imag.java
index 48a8a7107a3..b706b5fea5c 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Imag.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Imag.java
@@ -22,13 +22,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TFloat32;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Returns the imaginary part of a complex number.
@@ -48,7 +48,7 @@
  * @param <U> data type for {@code output()} output
  */
 @Operator(group = "math")
-public final class Imag<U extends TNumber> extends RawOp implements Operand<U> {
+public final class Imag<U extends Tensor & TNumber> extends RawOp implements Operand<U> {
   
   /**
    * Factory method to create a class wrapping a new Imag operation.
@@ -59,7 +59,7 @@ public final class Imag<U extends TNumber> extends RawOp implements Operand<U> {
    * @return a new instance of Imag
    */
   @Endpoint(describeByClass = true)
-  public static <U extends TNumber, T extends TType> Imag<U> create(Scope scope, Operand<T> input, DataType<U> Tout) {
+  public static <U extends Tensor & TNumber, T extends Tensor> Imag<U> create(Scope scope, Operand<T> input, DataType<U> Tout) {
     OperationBuilder opBuilder = scope.env().opBuilder("Imag", scope.makeOpName("Imag"));
     opBuilder.addInput(input.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
@@ -75,7 +75,7 @@ public static <U extends TNumber, T extends TType> Imag<U> create(Scope scope, O
    * @return a new instance of Imag
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> Imag<TFloat32> create(Scope scope, Operand<T> input) {
+  public static <T extends Tensor> Imag<TFloat32> create(Scope scope, Operand<T> input) {
     return create(scope, input, TFloat32.DTYPE);
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/InvertPermutation.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/InvertPermutation.java
index de9e9c46932..2cd61865d17 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/InvertPermutation.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/InvertPermutation.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes the inverse permutation of a tensor.
@@ -50,7 +50,7 @@
  * @param <T> data type for {@code y()} output
  */
 @Operator(group = "math")
-public final class InvertPermutation<T extends TNumber> extends RawOp implements Operand<T> {
+public final class InvertPermutation<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new InvertPermutation operation.
@@ -60,7 +60,7 @@ public final class InvertPermutation<T extends TNumber> extends RawOp implements
    * @return a new instance of InvertPermutation
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> InvertPermutation<T> create(Scope scope, Operand<T> x) {
+  public static <T extends Tensor & TNumber> InvertPermutation<T> create(Scope scope, Operand<T> x) {
     OperationBuilder opBuilder = scope.env().opBuilder("InvertPermutation", scope.makeOpName("InvertPermutation"));
     opBuilder.addInput(x.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/IsFinite.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/IsFinite.java
index ffe9bb99f10..e97e7be7219 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/IsFinite.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/IsFinite.java
@@ -21,13 +21,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TBool;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Returns which elements of x are finite.
@@ -54,7 +54,7 @@ public final class IsFinite extends RawOp implements Operand<TBool> {
    * @return a new instance of IsFinite
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> IsFinite create(Scope scope, Operand<T> x) {
+  public static <T extends Tensor & TNumber> IsFinite create(Scope scope, Operand<T> x) {
     OperationBuilder opBuilder = scope.env().opBuilder("IsFinite", scope.makeOpName("IsFinite"));
     opBuilder.addInput(x.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/IsInf.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/IsInf.java
index 3d36e9fbfe9..15c7219c6c0 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/IsInf.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/IsInf.java
@@ -21,13 +21,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TBool;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Returns which elements of x are Inf.
@@ -54,7 +54,7 @@ public final class IsInf extends RawOp implements Operand<TBool> {
    * @return a new instance of IsInf
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> IsInf create(Scope scope, Operand<T> x) {
+  public static <T extends Tensor & TNumber> IsInf create(Scope scope, Operand<T> x) {
     OperationBuilder opBuilder = scope.env().opBuilder("IsInf", scope.makeOpName("IsInf"));
     opBuilder.addInput(x.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/IsNan.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/IsNan.java
index b58205a005c..bfcecce7968 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/IsNan.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/IsNan.java
@@ -21,13 +21,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TBool;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Returns which elements of x are NaN.
@@ -54,7 +54,7 @@ public final class IsNan extends RawOp implements Operand<TBool> {
    * @return a new instance of IsNan
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> IsNan create(Scope scope, Operand<T> x) {
+  public static <T extends Tensor & TNumber> IsNan create(Scope scope, Operand<T> x) {
     OperationBuilder opBuilder = scope.env().opBuilder("IsNan", scope.makeOpName("IsNan"));
     opBuilder.addInput(x.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Less.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Less.java
index 3d796296f01..8908b2e210d 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Less.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Less.java
@@ -21,13 +21,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TBool;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Returns the truth value of (x < y) element-wise.
@@ -59,7 +59,7 @@ public final class Less extends RawOp implements Operand<TBool> {
    * @return a new instance of Less
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> Less create(Scope scope, Operand<T> x, Operand<T> y) {
+  public static <T extends Tensor & TNumber> Less create(Scope scope, Operand<T> x, Operand<T> y) {
     OperationBuilder opBuilder = scope.env().opBuilder("Less", scope.makeOpName("Less"));
     opBuilder.addInput(x.asOutput());
     opBuilder.addInput(y.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/LessEqual.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/LessEqual.java
index 618cd7a9866..e4463ed29cd 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/LessEqual.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/LessEqual.java
@@ -21,13 +21,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TBool;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Returns the truth value of (x <= y) element-wise.
@@ -59,7 +59,7 @@ public final class LessEqual extends RawOp implements Operand<TBool> {
    * @return a new instance of LessEqual
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> LessEqual create(Scope scope, Operand<T> x, Operand<T> y) {
+  public static <T extends Tensor & TNumber> LessEqual create(Scope scope, Operand<T> x, Operand<T> y) {
     OperationBuilder opBuilder = scope.env().opBuilder("LessEqual", scope.makeOpName("LessEqual"));
     opBuilder.addInput(x.asOutput());
     opBuilder.addInput(y.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Lgamma.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Lgamma.java
index 9bc18b1b3b8..d1284ca2b3f 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Lgamma.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Lgamma.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes the log of the absolute value of `Gamma(x)` element-wise.
@@ -44,7 +44,7 @@
  * @param <T> data type for {@code y()} output
  */
 @Operator(group = "math")
-public final class Lgamma<T extends TNumber> extends RawOp implements Operand<T> {
+public final class Lgamma<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Lgamma operation.
@@ -54,7 +54,7 @@ public final class Lgamma<T extends TNumber> extends RawOp implements Operand<T>
    * @return a new instance of Lgamma
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> Lgamma<T> create(Scope scope, Operand<T> x) {
+  public static <T extends Tensor & TNumber> Lgamma<T> create(Scope scope, Operand<T> x) {
     OperationBuilder opBuilder = scope.env().opBuilder("Lgamma", scope.makeOpName("Lgamma"));
     opBuilder.addInput(x.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Log.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Log.java
index 9db418d1002..017d46cfec9 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Log.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Log.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes natural logarithm of x element-wise.
@@ -42,7 +42,7 @@
  * @param <T> data type for {@code y()} output
  */
 @Operator(group = "math")
-public final class Log<T extends TType> extends RawOp implements Operand<T> {
+public final class Log<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Log operation.
@@ -52,7 +52,7 @@ public final class Log<T extends TType> extends RawOp implements Operand<T> {
    * @return a new instance of Log
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> Log<T> create(Scope scope, Operand<T> x) {
+  public static <T extends Tensor> Log<T> create(Scope scope, Operand<T> x) {
     OperationBuilder opBuilder = scope.env().opBuilder("Log", scope.makeOpName("Log"));
     opBuilder.addInput(x.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Log1p.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Log1p.java
index 3cc2656ea9b..1fb34539fa2 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Log1p.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Log1p.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes natural logarithm of (1 + x) element-wise.
@@ -42,7 +42,7 @@
  * @param <T> data type for {@code y()} output
  */
 @Operator(group = "math")
-public final class Log1p<T extends TType> extends RawOp implements Operand<T> {
+public final class Log1p<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Log1p operation.
@@ -52,7 +52,7 @@ public final class Log1p<T extends TType> extends RawOp implements Operand<T> {
    * @return a new instance of Log1p
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> Log1p<T> create(Scope scope, Operand<T> x) {
+  public static <T extends Tensor> Log1p<T> create(Scope scope, Operand<T> x) {
     OperationBuilder opBuilder = scope.env().opBuilder("Log1p", scope.makeOpName("Log1p"));
     opBuilder.addInput(x.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Maximum.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Maximum.java
index 9f72150c5d3..0cc2abb4da6 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Maximum.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Maximum.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Returns the max of x and y (i.e. x > y ? x : y) element-wise.
@@ -37,7 +37,7 @@
  * @param <T> data type for {@code z()} output
  */
 @Operator(group = "math")
-public final class Maximum<T extends TNumber> extends RawOp implements Operand<T> {
+public final class Maximum<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Maximum operation.
@@ -48,7 +48,7 @@ public final class Maximum<T extends TNumber> extends RawOp implements Operand<T
    * @return a new instance of Maximum
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> Maximum<T> create(Scope scope, Operand<T> x, Operand<T> y) {
+  public static <T extends Tensor & TNumber> Maximum<T> create(Scope scope, Operand<T> x, Operand<T> y) {
     OperationBuilder opBuilder = scope.env().opBuilder("Maximum", scope.makeOpName("Maximum"));
     opBuilder.addInput(x.asOutput());
     opBuilder.addInput(y.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Mean.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Mean.java
index 7859b7ce10f..bf3641fb730 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Mean.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Mean.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes the mean of elements across dimensions of a tensor.
@@ -39,7 +39,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "math")
-public final class Mean<T extends TType> extends RawOp implements Operand<T> {
+public final class Mean<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.math.Mean}
@@ -71,7 +71,7 @@ private Options() {
    * @return a new instance of Mean
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> Mean<T> create(Scope scope, Operand<T> input, Operand<U> axis, Options... options) {
+  public static <T extends Tensor, U extends Tensor & TNumber> Mean<T> create(Scope scope, Operand<T> input, Operand<U> axis, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("Mean", scope.makeOpName("Mean"));
     opBuilder.addInput(input.asOutput());
     opBuilder.addInput(axis.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Minimum.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Minimum.java
index e11b6e484fc..8818d37fa92 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Minimum.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Minimum.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Returns the min of x and y (i.e. x < y ? x : y) element-wise.
@@ -37,7 +37,7 @@
  * @param <T> data type for {@code z()} output
  */
 @Operator(group = "math")
-public final class Minimum<T extends TNumber> extends RawOp implements Operand<T> {
+public final class Minimum<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Minimum operation.
@@ -48,7 +48,7 @@ public final class Minimum<T extends TNumber> extends RawOp implements Operand<T
    * @return a new instance of Minimum
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> Minimum<T> create(Scope scope, Operand<T> x, Operand<T> y) {
+  public static <T extends Tensor & TNumber> Minimum<T> create(Scope scope, Operand<T> x, Operand<T> y) {
     OperationBuilder opBuilder = scope.env().opBuilder("Minimum", scope.makeOpName("Minimum"));
     opBuilder.addInput(x.asOutput());
     opBuilder.addInput(y.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Mod.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Mod.java
index 7ba98b81b39..e2ea89b0d51 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Mod.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Mod.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Returns element-wise remainder of division. This emulates C semantics in that
@@ -40,7 +40,7 @@
  * @param <T> data type for {@code z()} output
  */
 @Operator(group = "math")
-public final class Mod<T extends TNumber> extends RawOp implements Operand<T> {
+public final class Mod<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Mod operation.
@@ -51,7 +51,7 @@ public final class Mod<T extends TNumber> extends RawOp implements Operand<T> {
    * @return a new instance of Mod
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> Mod<T> create(Scope scope, Operand<T> x, Operand<T> y) {
+  public static <T extends Tensor & TNumber> Mod<T> create(Scope scope, Operand<T> x, Operand<T> y) {
     OperationBuilder opBuilder = scope.env().opBuilder("Mod", scope.makeOpName("Mod"));
     opBuilder.addInput(x.asOutput());
     opBuilder.addInput(y.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Mul.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Mul.java
index 20d5f471ba8..9a71a3eec7a 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Mul.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Mul.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Returns x * y element-wise.
@@ -36,7 +36,7 @@
  * @param <T> data type for {@code z()} output
  */
 @Operator(group = "math")
-public final class Mul<T extends TType> extends RawOp implements Operand<T> {
+public final class Mul<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Mul operation.
@@ -47,7 +47,7 @@ public final class Mul<T extends TType> extends RawOp implements Operand<T> {
    * @return a new instance of Mul
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> Mul<T> create(Scope scope, Operand<T> x, Operand<T> y) {
+  public static <T extends Tensor> Mul<T> create(Scope scope, Operand<T> x, Operand<T> y) {
     OperationBuilder opBuilder = scope.env().opBuilder("Mul", scope.makeOpName("Mul"));
     opBuilder.addInput(x.asOutput());
     opBuilder.addInput(y.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/MulNoNan.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/MulNoNan.java
index 312a3025870..eda21a38bf0 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/MulNoNan.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/MulNoNan.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Returns x * y element-wise. Returns zero if y is zero, even if x if infinite or NaN.
@@ -36,7 +36,7 @@
  * @param <T> data type for {@code z()} output
  */
 @Operator(group = "math")
-public final class MulNoNan<T extends TType> extends RawOp implements Operand<T> {
+public final class MulNoNan<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new MulNoNan operation.
@@ -47,7 +47,7 @@ public final class MulNoNan<T extends TType> extends RawOp implements Operand<T>
    * @return a new instance of MulNoNan
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> MulNoNan<T> create(Scope scope, Operand<T> x, Operand<T> y) {
+  public static <T extends Tensor> MulNoNan<T> create(Scope scope, Operand<T> x, Operand<T> y) {
     OperationBuilder opBuilder = scope.env().opBuilder("MulNoNan", scope.makeOpName("MulNoNan"));
     opBuilder.addInput(x.asOutput());
     opBuilder.addInput(y.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Ndtri.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Ndtri.java
index 55dcf0c434d..8d9248f6012 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Ndtri.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Ndtri.java
@@ -21,18 +21,18 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * @param <T> data type for {@code y()} output
  */
 @Operator(group = "math")
-public final class Ndtri<T extends TNumber> extends RawOp implements Operand<T> {
+public final class Ndtri<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Ndtri operation.
@@ -42,7 +42,7 @@ public final class Ndtri<T extends TNumber> extends RawOp implements Operand<T>
    * @return a new instance of Ndtri
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> Ndtri<T> create(Scope scope, Operand<T> x) {
+  public static <T extends Tensor & TNumber> Ndtri<T> create(Scope scope, Operand<T> x) {
     OperationBuilder opBuilder = scope.env().opBuilder("Ndtri", scope.makeOpName("Ndtri"));
     opBuilder.addInput(x.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Neg.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Neg.java
index 53079692e32..341809a7260 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Neg.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Neg.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes numerical negative value element-wise.
@@ -35,7 +35,7 @@
  * @param <T> data type for {@code y()} output
  */
 @Operator(group = "math")
-public final class Neg<T extends TType> extends RawOp implements Operand<T> {
+public final class Neg<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Neg operation.
@@ -45,7 +45,7 @@ public final class Neg<T extends TType> extends RawOp implements Operand<T> {
    * @return a new instance of Neg
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> Neg<T> create(Scope scope, Operand<T> x) {
+  public static <T extends Tensor> Neg<T> create(Scope scope, Operand<T> x) {
     OperationBuilder opBuilder = scope.env().opBuilder("Neg", scope.makeOpName("Neg"));
     opBuilder.addInput(x.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/NextAfter.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/NextAfter.java
index 8fa53306eed..3d12fe37f42 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/NextAfter.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/NextAfter.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Returns the next representable value of `x1` in the direction of `x2`, element-wise.
@@ -42,7 +42,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "math")
-public final class NextAfter<T extends TNumber> extends RawOp implements Operand<T> {
+public final class NextAfter<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new NextAfter operation.
@@ -53,7 +53,7 @@ public final class NextAfter<T extends TNumber> extends RawOp implements Operand
    * @return a new instance of NextAfter
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> NextAfter<T> create(Scope scope, Operand<T> x1, Operand<T> x2) {
+  public static <T extends Tensor & TNumber> NextAfter<T> create(Scope scope, Operand<T> x1, Operand<T> x2) {
     OperationBuilder opBuilder = scope.env().opBuilder("NextAfter", scope.makeOpName("NextAfter"));
     opBuilder.addInput(x1.asOutput());
     opBuilder.addInput(x2.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/NotEqual.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/NotEqual.java
index 977688031c5..7beddd8f94f 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/NotEqual.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/NotEqual.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TBool;
-import org.tensorflow.types.family.TType;
 
 /**
  * Returns the truth value of (x != y) element-wise.
@@ -66,7 +66,7 @@ private Options() {
    * @return a new instance of NotEqual
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> NotEqual create(Scope scope, Operand<T> x, Operand<T> y, Options... options) {
+  public static <T extends Tensor> NotEqual create(Scope scope, Operand<T> x, Operand<T> y, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("NotEqual", scope.makeOpName("NotEqual"));
     opBuilder.addInput(x.asOutput());
     opBuilder.addInput(y.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Polygamma.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Polygamma.java
index d0021efc2ac..f8ac90aa7d4 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Polygamma.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Polygamma.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Compute the polygamma function \\(\psi^{(n)}(x)\\).
@@ -41,7 +41,7 @@
  * @param <T> data type for {@code z()} output
  */
 @Operator(group = "math")
-public final class Polygamma<T extends TNumber> extends RawOp implements Operand<T> {
+public final class Polygamma<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Polygamma operation.
@@ -52,7 +52,7 @@ public final class Polygamma<T extends TNumber> extends RawOp implements Operand
    * @return a new instance of Polygamma
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> Polygamma<T> create(Scope scope, Operand<T> a, Operand<T> x) {
+  public static <T extends Tensor & TNumber> Polygamma<T> create(Scope scope, Operand<T> a, Operand<T> x) {
     OperationBuilder opBuilder = scope.env().opBuilder("Polygamma", scope.makeOpName("Polygamma"));
     opBuilder.addInput(a.asOutput());
     opBuilder.addInput(x.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/PopulationCount.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/PopulationCount.java
index bc721897426..1a7e9cd65c6 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/PopulationCount.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/PopulationCount.java
@@ -21,13 +21,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TUint8;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes element-wise population count (a.k.a. popcount, bitsum, bitcount).
@@ -50,7 +50,7 @@ public final class PopulationCount extends RawOp implements Operand<TUint8> {
    * @return a new instance of PopulationCount
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> PopulationCount create(Scope scope, Operand<T> x) {
+  public static <T extends Tensor & TNumber> PopulationCount create(Scope scope, Operand<T> x) {
     OperationBuilder opBuilder = scope.env().opBuilder("PopulationCount", scope.makeOpName("PopulationCount"));
     opBuilder.addInput(x.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Pow.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Pow.java
index 27f26e431fe..b6195fa9e33 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Pow.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Pow.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes the power of one value to another.
@@ -42,7 +42,7 @@
  * @param <T> data type for {@code z()} output
  */
 @Operator(group = "math")
-public final class Pow<T extends TType> extends RawOp implements Operand<T> {
+public final class Pow<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Pow operation.
@@ -53,7 +53,7 @@ public final class Pow<T extends TType> extends RawOp implements Operand<T> {
    * @return a new instance of Pow
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> Pow<T> create(Scope scope, Operand<T> x, Operand<T> y) {
+  public static <T extends Tensor> Pow<T> create(Scope scope, Operand<T> x, Operand<T> y) {
     OperationBuilder opBuilder = scope.env().opBuilder("Pow", scope.makeOpName("Pow"));
     opBuilder.addInput(x.asOutput());
     opBuilder.addInput(y.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/QuantizedAdd.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/QuantizedAdd.java
index ccd6d2e5f98..a75ed47822b 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/QuantizedAdd.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/QuantizedAdd.java
@@ -22,12 +22,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.family.TType;
 
 /**
  * Returns x + y element-wise, working on quantized buffers.
@@ -35,7 +35,7 @@
  * @param <V> data type for {@code z()} output
  */
 @Operator(group = "math")
-public final class QuantizedAdd<V extends TType> extends RawOp {
+public final class QuantizedAdd<V extends Tensor> extends RawOp {
   
   /**
    * Factory method to create a class wrapping a new QuantizedAdd operation.
@@ -51,7 +51,7 @@ public final class QuantizedAdd<V extends TType> extends RawOp {
    * @return a new instance of QuantizedAdd
    */
   @Endpoint(describeByClass = true)
-  public static <V extends TType, T extends TType, U extends TType> QuantizedAdd<V> create(Scope scope, Operand<T> x, Operand<U> y, Operand<TFloat32> minX, Operand<TFloat32> maxX, Operand<TFloat32> minY, Operand<TFloat32> maxY, DataType<V> Toutput) {
+  public static <V extends Tensor, T extends Tensor, U extends Tensor> QuantizedAdd<V> create(Scope scope, Operand<T> x, Operand<U> y, Operand<TFloat32> minX, Operand<TFloat32> maxX, Operand<TFloat32> minY, Operand<TFloat32> maxY, DataType<V> Toutput) {
     OperationBuilder opBuilder = scope.env().opBuilder("QuantizedAdd", scope.makeOpName("QuantizedAdd"));
     opBuilder.addInput(x.asOutput());
     opBuilder.addInput(y.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/QuantizedMul.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/QuantizedMul.java
index e16dc423a63..be28c564b8c 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/QuantizedMul.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/QuantizedMul.java
@@ -22,12 +22,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.family.TType;
 
 /**
  * Returns x * y element-wise, working on quantized buffers.
@@ -35,7 +35,7 @@
  * @param <V> data type for {@code z()} output
  */
 @Operator(group = "math")
-public final class QuantizedMul<V extends TType> extends RawOp {
+public final class QuantizedMul<V extends Tensor> extends RawOp {
   
   /**
    * Factory method to create a class wrapping a new QuantizedMul operation.
@@ -51,7 +51,7 @@ public final class QuantizedMul<V extends TType> extends RawOp {
    * @return a new instance of QuantizedMul
    */
   @Endpoint(describeByClass = true)
-  public static <V extends TType, T extends TType, U extends TType> QuantizedMul<V> create(Scope scope, Operand<T> x, Operand<U> y, Operand<TFloat32> minX, Operand<TFloat32> maxX, Operand<TFloat32> minY, Operand<TFloat32> maxY, DataType<V> Toutput) {
+  public static <V extends Tensor, T extends Tensor, U extends Tensor> QuantizedMul<V> create(Scope scope, Operand<T> x, Operand<U> y, Operand<TFloat32> minX, Operand<TFloat32> maxX, Operand<TFloat32> minY, Operand<TFloat32> maxY, DataType<V> Toutput) {
     OperationBuilder opBuilder = scope.env().opBuilder("QuantizedMul", scope.makeOpName("QuantizedMul"));
     opBuilder.addInput(x.asOutput());
     opBuilder.addInput(y.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Real.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Real.java
index e113597afd8..94feb46b882 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Real.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Real.java
@@ -22,13 +22,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TFloat32;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Returns the real part of a complex number.
@@ -48,7 +48,7 @@
  * @param <U> data type for {@code output()} output
  */
 @Operator(group = "math")
-public final class Real<U extends TNumber> extends RawOp implements Operand<U> {
+public final class Real<U extends Tensor & TNumber> extends RawOp implements Operand<U> {
   
   /**
    * Factory method to create a class wrapping a new Real operation.
@@ -59,7 +59,7 @@ public final class Real<U extends TNumber> extends RawOp implements Operand<U> {
    * @return a new instance of Real
    */
   @Endpoint(describeByClass = true)
-  public static <U extends TNumber, T extends TType> Real<U> create(Scope scope, Operand<T> input, DataType<U> Tout) {
+  public static <U extends Tensor & TNumber, T extends Tensor> Real<U> create(Scope scope, Operand<T> input, DataType<U> Tout) {
     OperationBuilder opBuilder = scope.env().opBuilder("Real", scope.makeOpName("Real"));
     opBuilder.addInput(input.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
@@ -75,7 +75,7 @@ public static <U extends TNumber, T extends TType> Real<U> create(Scope scope, O
    * @return a new instance of Real
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> Real<TFloat32> create(Scope scope, Operand<T> input) {
+  public static <T extends Tensor> Real<TFloat32> create(Scope scope, Operand<T> input) {
     return create(scope, input, TFloat32.DTYPE);
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/RealDiv.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/RealDiv.java
index ec322c696db..e86d43a9c8d 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/RealDiv.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/RealDiv.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Returns x / y element-wise for real types.
@@ -38,7 +38,7 @@
  * @param <T> data type for {@code z()} output
  */
 @Operator(group = "math")
-public final class RealDiv<T extends TType> extends RawOp implements Operand<T> {
+public final class RealDiv<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new RealDiv operation.
@@ -49,7 +49,7 @@ public final class RealDiv<T extends TType> extends RawOp implements Operand<T>
    * @return a new instance of RealDiv
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> RealDiv<T> create(Scope scope, Operand<T> x, Operand<T> y) {
+  public static <T extends Tensor> RealDiv<T> create(Scope scope, Operand<T> x, Operand<T> y) {
     OperationBuilder opBuilder = scope.env().opBuilder("RealDiv", scope.makeOpName("RealDiv"));
     opBuilder.addInput(x.asOutput());
     opBuilder.addInput(y.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Reciprocal.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Reciprocal.java
index a0b53da3b7c..81ddad0f029 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Reciprocal.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Reciprocal.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes the reciprocal of x element-wise.
@@ -35,7 +35,7 @@
  * @param <T> data type for {@code y()} output
  */
 @Operator(group = "math")
-public final class Reciprocal<T extends TType> extends RawOp implements Operand<T> {
+public final class Reciprocal<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Reciprocal operation.
@@ -45,7 +45,7 @@ public final class Reciprocal<T extends TType> extends RawOp implements Operand<
    * @return a new instance of Reciprocal
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> Reciprocal<T> create(Scope scope, Operand<T> x) {
+  public static <T extends Tensor> Reciprocal<T> create(Scope scope, Operand<T> x) {
     OperationBuilder opBuilder = scope.env().opBuilder("Reciprocal", scope.makeOpName("Reciprocal"));
     opBuilder.addInput(x.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/ReciprocalGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/ReciprocalGrad.java
index dee41220ba9..09d4b5bc0a3 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/ReciprocalGrad.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/ReciprocalGrad.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes the gradient for the inverse of `x` wrt its input.
@@ -35,7 +35,7 @@
  * 
  * @param <T> data type for {@code z()} output
  */
-public final class ReciprocalGrad<T extends TType> extends RawOp implements Operand<T> {
+public final class ReciprocalGrad<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new ReciprocalGrad operation.
@@ -46,7 +46,7 @@ public final class ReciprocalGrad<T extends TType> extends RawOp implements Oper
    * @return a new instance of ReciprocalGrad
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> ReciprocalGrad<T> create(Scope scope, Operand<T> y, Operand<T> dy) {
+  public static <T extends Tensor> ReciprocalGrad<T> create(Scope scope, Operand<T> y, Operand<T> dy) {
     OperationBuilder opBuilder = scope.env().opBuilder("ReciprocalGrad", scope.makeOpName("ReciprocalGrad"));
     opBuilder.addInput(y.asOutput());
     opBuilder.addInput(dy.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/RequantizationRangePerChannel.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/RequantizationRangePerChannel.java
index feda79c70b2..428045a91c1 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/RequantizationRangePerChannel.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/RequantizationRangePerChannel.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes requantization range per channel.
@@ -45,7 +45,7 @@ public final class RequantizationRangePerChannel extends RawOp {
    * @return a new instance of RequantizationRangePerChannel
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> RequantizationRangePerChannel create(Scope scope, Operand<T> input, Operand<TFloat32> inputMin, Operand<TFloat32> inputMax, Float clipValueMax) {
+  public static <T extends Tensor> RequantizationRangePerChannel create(Scope scope, Operand<T> input, Operand<TFloat32> inputMin, Operand<TFloat32> inputMax, Float clipValueMax) {
     OperationBuilder opBuilder = scope.env().opBuilder("RequantizationRangePerChannel", scope.makeOpName("RequantizationRangePerChannel"));
     opBuilder.addInput(input.asOutput());
     opBuilder.addInput(inputMin.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/RequantizePerChannel.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/RequantizePerChannel.java
index 53a6201151e..f86e7d45a37 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/RequantizePerChannel.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/RequantizePerChannel.java
@@ -22,19 +22,19 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.family.TType;
 
 /**
  * Requantizes input with min and max values known per channel.
  * 
  * @param <U> data type for {@code output()} output
  */
-public final class RequantizePerChannel<U extends TType> extends RawOp {
+public final class RequantizePerChannel<U extends Tensor> extends RawOp {
   
   /**
    * Factory method to create a class wrapping a new RequantizePerChannel operation.
@@ -49,7 +49,7 @@ public final class RequantizePerChannel<U extends TType> extends RawOp {
    * @return a new instance of RequantizePerChannel
    */
   @Endpoint(describeByClass = true)
-  public static <U extends TType, T extends TType> RequantizePerChannel<U> create(Scope scope, Operand<T> input, Operand<TFloat32> inputMin, Operand<TFloat32> inputMax, Operand<TFloat32> requestedOutputMin, Operand<TFloat32> requestedOutputMax, DataType<U> outType) {
+  public static <U extends Tensor, T extends Tensor> RequantizePerChannel<U> create(Scope scope, Operand<T> input, Operand<TFloat32> inputMin, Operand<TFloat32> inputMax, Operand<TFloat32> requestedOutputMin, Operand<TFloat32> requestedOutputMax, DataType<U> outType) {
     OperationBuilder opBuilder = scope.env().opBuilder("RequantizePerChannel", scope.makeOpName("RequantizePerChannel"));
     opBuilder.addInput(input.asOutput());
     opBuilder.addInput(inputMin.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Rint.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Rint.java
index b7b152248a8..1aaa11b178d 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Rint.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Rint.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Returns element-wise integer closest to x.
@@ -44,7 +44,7 @@
  * @param <T> data type for {@code y()} output
  */
 @Operator(group = "math")
-public final class Rint<T extends TNumber> extends RawOp implements Operand<T> {
+public final class Rint<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Rint operation.
@@ -54,7 +54,7 @@ public final class Rint<T extends TNumber> extends RawOp implements Operand<T> {
    * @return a new instance of Rint
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> Rint<T> create(Scope scope, Operand<T> x) {
+  public static <T extends Tensor & TNumber> Rint<T> create(Scope scope, Operand<T> x) {
     OperationBuilder opBuilder = scope.env().opBuilder("Rint", scope.makeOpName("Rint"));
     opBuilder.addInput(x.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Round.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Round.java
index 7f892d28ed0..b43bba763f0 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Round.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Round.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Rounds the values of a tensor to the nearest integer, element-wise.
@@ -36,7 +36,7 @@
  * @param <T> data type for {@code y()} output
  */
 @Operator(group = "math")
-public final class Round<T extends TType> extends RawOp implements Operand<T> {
+public final class Round<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Round operation.
@@ -46,7 +46,7 @@ public final class Round<T extends TType> extends RawOp implements Operand<T> {
    * @return a new instance of Round
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> Round<T> create(Scope scope, Operand<T> x) {
+  public static <T extends Tensor> Round<T> create(Scope scope, Operand<T> x) {
     OperationBuilder opBuilder = scope.env().opBuilder("Round", scope.makeOpName("Round"));
     opBuilder.addInput(x.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Rsqrt.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Rsqrt.java
index de04c3f1204..ebe110213c4 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Rsqrt.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Rsqrt.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes reciprocal of square root of x element-wise.
@@ -35,7 +35,7 @@
  * @param <T> data type for {@code y()} output
  */
 @Operator(group = "math")
-public final class Rsqrt<T extends TType> extends RawOp implements Operand<T> {
+public final class Rsqrt<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Rsqrt operation.
@@ -45,7 +45,7 @@ public final class Rsqrt<T extends TType> extends RawOp implements Operand<T> {
    * @return a new instance of Rsqrt
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> Rsqrt<T> create(Scope scope, Operand<T> x) {
+  public static <T extends Tensor> Rsqrt<T> create(Scope scope, Operand<T> x) {
     OperationBuilder opBuilder = scope.env().opBuilder("Rsqrt", scope.makeOpName("Rsqrt"));
     opBuilder.addInput(x.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/RsqrtGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/RsqrtGrad.java
index ee0eec068f3..58f228a5c24 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/RsqrtGrad.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/RsqrtGrad.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes the gradient for the rsqrt of `x` wrt its input.
@@ -35,7 +35,7 @@
  * 
  * @param <T> data type for {@code z()} output
  */
-public final class RsqrtGrad<T extends TType> extends RawOp implements Operand<T> {
+public final class RsqrtGrad<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new RsqrtGrad operation.
@@ -46,7 +46,7 @@ public final class RsqrtGrad<T extends TType> extends RawOp implements Operand<T
    * @return a new instance of RsqrtGrad
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> RsqrtGrad<T> create(Scope scope, Operand<T> y, Operand<T> dy) {
+  public static <T extends Tensor> RsqrtGrad<T> create(Scope scope, Operand<T> y, Operand<T> dy) {
     OperationBuilder opBuilder = scope.env().opBuilder("RsqrtGrad", scope.makeOpName("RsqrtGrad"));
     opBuilder.addInput(y.asOutput());
     opBuilder.addInput(dy.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SegmentMax.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SegmentMax.java
index fdd40054420..80af885b628 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SegmentMax.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SegmentMax.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes the maximum along segments of a tensor.
@@ -57,7 +57,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "math")
-public final class SegmentMax<T extends TNumber> extends RawOp implements Operand<T> {
+public final class SegmentMax<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new SegmentMax operation.
@@ -69,7 +69,7 @@ public final class SegmentMax<T extends TNumber> extends RawOp implements Operan
    * @return a new instance of SegmentMax
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber, U extends TNumber> SegmentMax<T> create(Scope scope, Operand<T> data, Operand<U> segmentIds) {
+  public static <T extends Tensor & TNumber, U extends Tensor & TNumber> SegmentMax<T> create(Scope scope, Operand<T> data, Operand<U> segmentIds) {
     OperationBuilder opBuilder = scope.env().opBuilder("SegmentMax", scope.makeOpName("SegmentMax"));
     opBuilder.addInput(data.asOutput());
     opBuilder.addInput(segmentIds.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SegmentMean.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SegmentMean.java
index 2fe16ce7e12..46359a58ac9 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SegmentMean.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SegmentMean.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes the mean along segments of a tensor.
@@ -58,7 +58,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "math")
-public final class SegmentMean<T extends TType> extends RawOp implements Operand<T> {
+public final class SegmentMean<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new SegmentMean operation.
@@ -70,7 +70,7 @@ public final class SegmentMean<T extends TType> extends RawOp implements Operand
    * @return a new instance of SegmentMean
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> SegmentMean<T> create(Scope scope, Operand<T> data, Operand<U> segmentIds) {
+  public static <T extends Tensor, U extends Tensor & TNumber> SegmentMean<T> create(Scope scope, Operand<T> data, Operand<U> segmentIds) {
     OperationBuilder opBuilder = scope.env().opBuilder("SegmentMean", scope.makeOpName("SegmentMean"));
     opBuilder.addInput(data.asOutput());
     opBuilder.addInput(segmentIds.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SegmentMin.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SegmentMin.java
index b7ab590e976..05ea7605bc8 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SegmentMin.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SegmentMin.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes the minimum along segments of a tensor.
@@ -57,7 +57,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "math")
-public final class SegmentMin<T extends TNumber> extends RawOp implements Operand<T> {
+public final class SegmentMin<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new SegmentMin operation.
@@ -69,7 +69,7 @@ public final class SegmentMin<T extends TNumber> extends RawOp implements Operan
    * @return a new instance of SegmentMin
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber, U extends TNumber> SegmentMin<T> create(Scope scope, Operand<T> data, Operand<U> segmentIds) {
+  public static <T extends Tensor & TNumber, U extends Tensor & TNumber> SegmentMin<T> create(Scope scope, Operand<T> data, Operand<U> segmentIds) {
     OperationBuilder opBuilder = scope.env().opBuilder("SegmentMin", scope.makeOpName("SegmentMin"));
     opBuilder.addInput(data.asOutput());
     opBuilder.addInput(segmentIds.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SegmentProd.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SegmentProd.java
index 0f1179072e5..0fa63ed7bb4 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SegmentProd.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SegmentProd.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes the product along segments of a tensor.
@@ -57,7 +57,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "math")
-public final class SegmentProd<T extends TType> extends RawOp implements Operand<T> {
+public final class SegmentProd<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new SegmentProd operation.
@@ -69,7 +69,7 @@ public final class SegmentProd<T extends TType> extends RawOp implements Operand
    * @return a new instance of SegmentProd
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> SegmentProd<T> create(Scope scope, Operand<T> data, Operand<U> segmentIds) {
+  public static <T extends Tensor, U extends Tensor & TNumber> SegmentProd<T> create(Scope scope, Operand<T> data, Operand<U> segmentIds) {
     OperationBuilder opBuilder = scope.env().opBuilder("SegmentProd", scope.makeOpName("SegmentProd"));
     opBuilder.addInput(data.asOutput());
     opBuilder.addInput(segmentIds.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SegmentSum.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SegmentSum.java
index fceec1bfe43..fe5818cd263 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SegmentSum.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SegmentSum.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes the sum along segments of a tensor.
@@ -57,7 +57,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "math")
-public final class SegmentSum<T extends TType> extends RawOp implements Operand<T> {
+public final class SegmentSum<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new SegmentSum operation.
@@ -69,7 +69,7 @@ public final class SegmentSum<T extends TType> extends RawOp implements Operand<
    * @return a new instance of SegmentSum
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> SegmentSum<T> create(Scope scope, Operand<T> data, Operand<U> segmentIds) {
+  public static <T extends Tensor, U extends Tensor & TNumber> SegmentSum<T> create(Scope scope, Operand<T> data, Operand<U> segmentIds) {
     OperationBuilder opBuilder = scope.env().opBuilder("SegmentSum", scope.makeOpName("SegmentSum"));
     opBuilder.addInput(data.asOutput());
     opBuilder.addInput(segmentIds.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Sigmoid.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Sigmoid.java
index a54e4ae51ea..5ae3e66c143 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Sigmoid.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Sigmoid.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes sigmoid of `x` element-wise.
@@ -35,7 +35,7 @@
  * @param <T> data type for {@code y()} output
  */
 @Operator(group = "math")
-public final class Sigmoid<T extends TType> extends RawOp implements Operand<T> {
+public final class Sigmoid<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Sigmoid operation.
@@ -45,7 +45,7 @@ public final class Sigmoid<T extends TType> extends RawOp implements Operand<T>
    * @return a new instance of Sigmoid
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> Sigmoid<T> create(Scope scope, Operand<T> x) {
+  public static <T extends Tensor> Sigmoid<T> create(Scope scope, Operand<T> x) {
     OperationBuilder opBuilder = scope.env().opBuilder("Sigmoid", scope.makeOpName("Sigmoid"));
     opBuilder.addInput(x.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SigmoidGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SigmoidGrad.java
index 125397bfd65..b0a39e7d20f 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SigmoidGrad.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SigmoidGrad.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes the gradient of the sigmoid of `x` wrt its input.
@@ -35,7 +35,7 @@
  * 
  * @param <T> data type for {@code z()} output
  */
-public final class SigmoidGrad<T extends TType> extends RawOp implements Operand<T> {
+public final class SigmoidGrad<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new SigmoidGrad operation.
@@ -46,7 +46,7 @@ public final class SigmoidGrad<T extends TType> extends RawOp implements Operand
    * @return a new instance of SigmoidGrad
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> SigmoidGrad<T> create(Scope scope, Operand<T> y, Operand<T> dy) {
+  public static <T extends Tensor> SigmoidGrad<T> create(Scope scope, Operand<T> y, Operand<T> dy) {
     OperationBuilder opBuilder = scope.env().opBuilder("SigmoidGrad", scope.makeOpName("SigmoidGrad"));
     opBuilder.addInput(y.asOutput());
     opBuilder.addInput(dy.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Sign.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Sign.java
index c9d8609872e..90bcacd79e1 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Sign.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Sign.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Returns an element-wise indication of the sign of a number.
@@ -41,7 +41,7 @@
  * @param <T> data type for {@code y()} output
  */
 @Operator(group = "math")
-public final class Sign<T extends TType> extends RawOp implements Operand<T> {
+public final class Sign<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Sign operation.
@@ -51,7 +51,7 @@ public final class Sign<T extends TType> extends RawOp implements Operand<T> {
    * @return a new instance of Sign
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> Sign<T> create(Scope scope, Operand<T> x) {
+  public static <T extends Tensor> Sign<T> create(Scope scope, Operand<T> x) {
     OperationBuilder opBuilder = scope.env().opBuilder("Sign", scope.makeOpName("Sign"));
     opBuilder.addInput(x.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Sin.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Sin.java
index a17ea9d288e..f7d6e535e3c 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Sin.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Sin.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes sine of x element-wise.
@@ -43,7 +43,7 @@
  * @param <T> data type for {@code y()} output
  */
 @Operator(group = "math")
-public final class Sin<T extends TType> extends RawOp implements Operand<T> {
+public final class Sin<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Sin operation.
@@ -53,7 +53,7 @@ public final class Sin<T extends TType> extends RawOp implements Operand<T> {
    * @return a new instance of Sin
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> Sin<T> create(Scope scope, Operand<T> x) {
+  public static <T extends Tensor> Sin<T> create(Scope scope, Operand<T> x) {
     OperationBuilder opBuilder = scope.env().opBuilder("Sin", scope.makeOpName("Sin"));
     opBuilder.addInput(x.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Sinh.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Sinh.java
index 613d983eef5..435d05838c9 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Sinh.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Sinh.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes hyperbolic sine of x element-wise.
@@ -43,7 +43,7 @@
  * @param <T> data type for {@code y()} output
  */
 @Operator(group = "math")
-public final class Sinh<T extends TType> extends RawOp implements Operand<T> {
+public final class Sinh<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Sinh operation.
@@ -53,7 +53,7 @@ public final class Sinh<T extends TType> extends RawOp implements Operand<T> {
    * @return a new instance of Sinh
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> Sinh<T> create(Scope scope, Operand<T> x) {
+  public static <T extends Tensor> Sinh<T> create(Scope scope, Operand<T> x) {
     OperationBuilder opBuilder = scope.env().opBuilder("Sinh", scope.makeOpName("Sinh"));
     opBuilder.addInput(x.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SobolSample.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SobolSample.java
index 131cb7ed792..b12f674ea18 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SobolSample.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SobolSample.java
@@ -22,6 +22,7 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -29,7 +30,6 @@
 import org.tensorflow.types.TFloat32;
 import org.tensorflow.types.TInt32;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Generates points from the Sobol sequence.
@@ -39,7 +39,7 @@
  * 
  * @param <T> data type for {@code samples()} output
  */
-public final class SobolSample<T extends TNumber> extends RawOp implements Operand<T> {
+public final class SobolSample<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new SobolSample operation.
@@ -54,7 +54,7 @@ public final class SobolSample<T extends TNumber> extends RawOp implements Opera
    * @return a new instance of SobolSample
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> SobolSample<T> create(Scope scope, Operand<TInt32> dim, Operand<TInt32> numResults, Operand<TInt32> skip, DataType<T> dtype) {
+  public static <T extends Tensor & TNumber> SobolSample<T> create(Scope scope, Operand<TInt32> dim, Operand<TInt32> numResults, Operand<TInt32> skip, DataType<T> dtype) {
     OperationBuilder opBuilder = scope.env().opBuilder("SobolSample", scope.makeOpName("SobolSample"));
     opBuilder.addInput(dim.asOutput());
     opBuilder.addInput(numResults.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Softplus.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Softplus.java
index 79adcf30d76..34a5370db29 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Softplus.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Softplus.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes softplus: `log(exp(features) + 1)`.
@@ -34,7 +34,7 @@
  * @param <T> data type for {@code activations()} output
  */
 @Operator(group = "math")
-public final class Softplus<T extends TNumber> extends RawOp implements Operand<T> {
+public final class Softplus<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Softplus operation.
@@ -44,7 +44,7 @@ public final class Softplus<T extends TNumber> extends RawOp implements Operand<
    * @return a new instance of Softplus
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> Softplus<T> create(Scope scope, Operand<T> features) {
+  public static <T extends Tensor & TNumber> Softplus<T> create(Scope scope, Operand<T> features) {
     OperationBuilder opBuilder = scope.env().opBuilder("Softplus", scope.makeOpName("Softplus"));
     opBuilder.addInput(features.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SoftplusGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SoftplusGrad.java
index 90174757bc3..05463dcc468 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SoftplusGrad.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SoftplusGrad.java
@@ -21,19 +21,19 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes softplus gradients for a softplus operation.
  * 
  * @param <T> data type for {@code backprops()} output
  */
-public final class SoftplusGrad<T extends TNumber> extends RawOp implements Operand<T> {
+public final class SoftplusGrad<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new SoftplusGrad operation.
@@ -44,7 +44,7 @@ public final class SoftplusGrad<T extends TNumber> extends RawOp implements Oper
    * @return a new instance of SoftplusGrad
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> SoftplusGrad<T> create(Scope scope, Operand<T> gradients, Operand<T> features) {
+  public static <T extends Tensor & TNumber> SoftplusGrad<T> create(Scope scope, Operand<T> gradients, Operand<T> features) {
     OperationBuilder opBuilder = scope.env().opBuilder("SoftplusGrad", scope.makeOpName("SoftplusGrad"));
     opBuilder.addInput(gradients.asOutput());
     opBuilder.addInput(features.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Sqrt.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Sqrt.java
index de62810d7ce..d9d843713f4 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Sqrt.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Sqrt.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes square root of x element-wise.
@@ -35,7 +35,7 @@
  * @param <T> data type for {@code y()} output
  */
 @Operator(group = "math")
-public final class Sqrt<T extends TType> extends RawOp implements Operand<T> {
+public final class Sqrt<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Sqrt operation.
@@ -45,7 +45,7 @@ public final class Sqrt<T extends TType> extends RawOp implements Operand<T> {
    * @return a new instance of Sqrt
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> Sqrt<T> create(Scope scope, Operand<T> x) {
+  public static <T extends Tensor> Sqrt<T> create(Scope scope, Operand<T> x) {
     OperationBuilder opBuilder = scope.env().opBuilder("Sqrt", scope.makeOpName("Sqrt"));
     opBuilder.addInput(x.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SqrtGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SqrtGrad.java
index 2c4b301a60b..3039153a9b1 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SqrtGrad.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SqrtGrad.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes the gradient for the sqrt of `x` wrt its input.
@@ -35,7 +35,7 @@
  * 
  * @param <T> data type for {@code z()} output
  */
-public final class SqrtGrad<T extends TType> extends RawOp implements Operand<T> {
+public final class SqrtGrad<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new SqrtGrad operation.
@@ -46,7 +46,7 @@ public final class SqrtGrad<T extends TType> extends RawOp implements Operand<T>
    * @return a new instance of SqrtGrad
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> SqrtGrad<T> create(Scope scope, Operand<T> y, Operand<T> dy) {
+  public static <T extends Tensor> SqrtGrad<T> create(Scope scope, Operand<T> y, Operand<T> dy) {
     OperationBuilder opBuilder = scope.env().opBuilder("SqrtGrad", scope.makeOpName("SqrtGrad"));
     opBuilder.addInput(y.asOutput());
     opBuilder.addInput(dy.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Square.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Square.java
index 01ccd098cc8..e970833974b 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Square.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Square.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes square of x element-wise.
@@ -35,7 +35,7 @@
  * @param <T> data type for {@code y()} output
  */
 @Operator(group = "math")
-public final class Square<T extends TType> extends RawOp implements Operand<T> {
+public final class Square<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Square operation.
@@ -45,7 +45,7 @@ public final class Square<T extends TType> extends RawOp implements Operand<T> {
    * @return a new instance of Square
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> Square<T> create(Scope scope, Operand<T> x) {
+  public static <T extends Tensor> Square<T> create(Scope scope, Operand<T> x) {
     OperationBuilder opBuilder = scope.env().opBuilder("Square", scope.makeOpName("Square"));
     opBuilder.addInput(x.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SquaredDifference.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SquaredDifference.java
index bed2fe9bb77..3968a668e23 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SquaredDifference.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SquaredDifference.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Returns (x - y)(x - y) element-wise.
@@ -36,7 +36,7 @@
  * @param <T> data type for {@code z()} output
  */
 @Operator(group = "math")
-public final class SquaredDifference<T extends TType> extends RawOp implements Operand<T> {
+public final class SquaredDifference<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new SquaredDifference operation.
@@ -47,7 +47,7 @@ public final class SquaredDifference<T extends TType> extends RawOp implements O
    * @return a new instance of SquaredDifference
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> SquaredDifference<T> create(Scope scope, Operand<T> x, Operand<T> y) {
+  public static <T extends Tensor> SquaredDifference<T> create(Scope scope, Operand<T> x, Operand<T> y) {
     OperationBuilder opBuilder = scope.env().opBuilder("SquaredDifference", scope.makeOpName("SquaredDifference"));
     opBuilder.addInput(x.asOutput());
     opBuilder.addInput(y.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Sub.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Sub.java
index 3578a962622..c5d9132bb7b 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Sub.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Sub.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Returns x - y element-wise.
@@ -36,7 +36,7 @@
  * @param <T> data type for {@code z()} output
  */
 @Operator(group = "math")
-public final class Sub<T extends TType> extends RawOp implements Operand<T> {
+public final class Sub<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Sub operation.
@@ -47,7 +47,7 @@ public final class Sub<T extends TType> extends RawOp implements Operand<T> {
    * @return a new instance of Sub
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> Sub<T> create(Scope scope, Operand<T> x, Operand<T> y) {
+  public static <T extends Tensor> Sub<T> create(Scope scope, Operand<T> x, Operand<T> y) {
     OperationBuilder opBuilder = scope.env().opBuilder("Sub", scope.makeOpName("Sub"));
     opBuilder.addInput(x.asOutput());
     opBuilder.addInput(y.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Tan.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Tan.java
index f09d909e8cb..a0418328c0b 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Tan.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Tan.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes tan of x element-wise.
@@ -44,7 +44,7 @@
  * @param <T> data type for {@code y()} output
  */
 @Operator(group = "math")
-public final class Tan<T extends TType> extends RawOp implements Operand<T> {
+public final class Tan<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Tan operation.
@@ -54,7 +54,7 @@ public final class Tan<T extends TType> extends RawOp implements Operand<T> {
    * @return a new instance of Tan
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> Tan<T> create(Scope scope, Operand<T> x) {
+  public static <T extends Tensor> Tan<T> create(Scope scope, Operand<T> x) {
     OperationBuilder opBuilder = scope.env().opBuilder("Tan", scope.makeOpName("Tan"));
     opBuilder.addInput(x.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Tanh.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Tanh.java
index 04d9d1e092c..d8aed91518c 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Tanh.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Tanh.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes hyperbolic tangent of `x` element-wise.
@@ -43,7 +43,7 @@
  * @param <T> data type for {@code y()} output
  */
 @Operator(group = "math")
-public final class Tanh<T extends TType> extends RawOp implements Operand<T> {
+public final class Tanh<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Tanh operation.
@@ -53,7 +53,7 @@ public final class Tanh<T extends TType> extends RawOp implements Operand<T> {
    * @return a new instance of Tanh
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> Tanh<T> create(Scope scope, Operand<T> x) {
+  public static <T extends Tensor> Tanh<T> create(Scope scope, Operand<T> x) {
     OperationBuilder opBuilder = scope.env().opBuilder("Tanh", scope.makeOpName("Tanh"));
     opBuilder.addInput(x.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/TanhGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/TanhGrad.java
index 273c5a1f76d..5ab68795017 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/TanhGrad.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/TanhGrad.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes the gradient for the tanh of `x` wrt its input.
@@ -35,7 +35,7 @@
  * 
  * @param <T> data type for {@code z()} output
  */
-public final class TanhGrad<T extends TType> extends RawOp implements Operand<T> {
+public final class TanhGrad<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new TanhGrad operation.
@@ -46,7 +46,7 @@ public final class TanhGrad<T extends TType> extends RawOp implements Operand<T>
    * @return a new instance of TanhGrad
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> TanhGrad<T> create(Scope scope, Operand<T> y, Operand<T> dy) {
+  public static <T extends Tensor> TanhGrad<T> create(Scope scope, Operand<T> y, Operand<T> dy) {
     OperationBuilder opBuilder = scope.env().opBuilder("TanhGrad", scope.makeOpName("TanhGrad"));
     opBuilder.addInput(y.asOutput());
     opBuilder.addInput(dy.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/TruncateDiv.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/TruncateDiv.java
index 5cd4b52fbed..3463015f92e 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/TruncateDiv.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/TruncateDiv.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Returns x / y element-wise for integer types.
@@ -41,7 +41,7 @@
  * @param <T> data type for {@code z()} output
  */
 @Operator(group = "math")
-public final class TruncateDiv<T extends TType> extends RawOp implements Operand<T> {
+public final class TruncateDiv<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new TruncateDiv operation.
@@ -52,7 +52,7 @@ public final class TruncateDiv<T extends TType> extends RawOp implements Operand
    * @return a new instance of TruncateDiv
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> TruncateDiv<T> create(Scope scope, Operand<T> x, Operand<T> y) {
+  public static <T extends Tensor> TruncateDiv<T> create(Scope scope, Operand<T> x, Operand<T> y) {
     OperationBuilder opBuilder = scope.env().opBuilder("TruncateDiv", scope.makeOpName("TruncateDiv"));
     opBuilder.addInput(x.asOutput());
     opBuilder.addInput(y.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/TruncateMod.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/TruncateMod.java
index 48c92574eb4..12bce627fa6 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/TruncateMod.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/TruncateMod.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Returns element-wise remainder of division. This emulates C semantics in that
@@ -40,7 +40,7 @@
  * @param <T> data type for {@code z()} output
  */
 @Operator(group = "math")
-public final class TruncateMod<T extends TNumber> extends RawOp implements Operand<T> {
+public final class TruncateMod<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new TruncateMod operation.
@@ -51,7 +51,7 @@ public final class TruncateMod<T extends TNumber> extends RawOp implements Opera
    * @return a new instance of TruncateMod
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> TruncateMod<T> create(Scope scope, Operand<T> x, Operand<T> y) {
+  public static <T extends Tensor & TNumber> TruncateMod<T> create(Scope scope, Operand<T> x, Operand<T> y) {
     OperationBuilder opBuilder = scope.env().opBuilder("TruncateMod", scope.makeOpName("TruncateMod"));
     opBuilder.addInput(x.asOutput());
     opBuilder.addInput(y.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/UnsortedSegmentMax.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/UnsortedSegmentMax.java
index 582113e3320..da8d6a1a2a1 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/UnsortedSegmentMax.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/UnsortedSegmentMax.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes the maximum along segments of a tensor.
@@ -65,7 +65,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "math")
-public final class UnsortedSegmentMax<T extends TNumber> extends RawOp implements Operand<T> {
+public final class UnsortedSegmentMax<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new UnsortedSegmentMax operation.
@@ -77,7 +77,7 @@ public final class UnsortedSegmentMax<T extends TNumber> extends RawOp implement
    * @return a new instance of UnsortedSegmentMax
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber, U extends TNumber, V extends TNumber> UnsortedSegmentMax<T> create(Scope scope, Operand<T> data, Operand<U> segmentIds, Operand<V> numSegments) {
+  public static <T extends Tensor & TNumber, U extends Tensor & TNumber, V extends Tensor & TNumber> UnsortedSegmentMax<T> create(Scope scope, Operand<T> data, Operand<U> segmentIds, Operand<V> numSegments) {
     OperationBuilder opBuilder = scope.env().opBuilder("UnsortedSegmentMax", scope.makeOpName("UnsortedSegmentMax"));
     opBuilder.addInput(data.asOutput());
     opBuilder.addInput(segmentIds.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/UnsortedSegmentMin.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/UnsortedSegmentMin.java
index 397ac5706a0..5fcc1caa4fc 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/UnsortedSegmentMin.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/UnsortedSegmentMin.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes the minimum along segments of a tensor.
@@ -59,7 +59,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "math")
-public final class UnsortedSegmentMin<T extends TNumber> extends RawOp implements Operand<T> {
+public final class UnsortedSegmentMin<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new UnsortedSegmentMin operation.
@@ -71,7 +71,7 @@ public final class UnsortedSegmentMin<T extends TNumber> extends RawOp implement
    * @return a new instance of UnsortedSegmentMin
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber, U extends TNumber, V extends TNumber> UnsortedSegmentMin<T> create(Scope scope, Operand<T> data, Operand<U> segmentIds, Operand<V> numSegments) {
+  public static <T extends Tensor & TNumber, U extends Tensor & TNumber, V extends Tensor & TNumber> UnsortedSegmentMin<T> create(Scope scope, Operand<T> data, Operand<U> segmentIds, Operand<V> numSegments) {
     OperationBuilder opBuilder = scope.env().opBuilder("UnsortedSegmentMin", scope.makeOpName("UnsortedSegmentMin"));
     opBuilder.addInput(data.asOutput());
     opBuilder.addInput(segmentIds.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/UnsortedSegmentProd.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/UnsortedSegmentProd.java
index c65fc484e4c..e3eb374edc0 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/UnsortedSegmentProd.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/UnsortedSegmentProd.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes the product along segments of a tensor.
@@ -58,7 +58,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "math")
-public final class UnsortedSegmentProd<T extends TType> extends RawOp implements Operand<T> {
+public final class UnsortedSegmentProd<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new UnsortedSegmentProd operation.
@@ -70,7 +70,7 @@ public final class UnsortedSegmentProd<T extends TType> extends RawOp implements
    * @return a new instance of UnsortedSegmentProd
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber, V extends TNumber> UnsortedSegmentProd<T> create(Scope scope, Operand<T> data, Operand<U> segmentIds, Operand<V> numSegments) {
+  public static <T extends Tensor, U extends Tensor & TNumber, V extends Tensor & TNumber> UnsortedSegmentProd<T> create(Scope scope, Operand<T> data, Operand<U> segmentIds, Operand<V> numSegments) {
     OperationBuilder opBuilder = scope.env().opBuilder("UnsortedSegmentProd", scope.makeOpName("UnsortedSegmentProd"));
     opBuilder.addInput(data.asOutput());
     opBuilder.addInput(segmentIds.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/UnsortedSegmentSum.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/UnsortedSegmentSum.java
index aaaf5b7bd23..957b9409413 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/UnsortedSegmentSum.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/UnsortedSegmentSum.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes the sum along segments of a tensor.
@@ -61,7 +61,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "math")
-public final class UnsortedSegmentSum<T extends TType> extends RawOp implements Operand<T> {
+public final class UnsortedSegmentSum<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new UnsortedSegmentSum operation.
@@ -73,7 +73,7 @@ public final class UnsortedSegmentSum<T extends TType> extends RawOp implements
    * @return a new instance of UnsortedSegmentSum
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber, V extends TNumber> UnsortedSegmentSum<T> create(Scope scope, Operand<T> data, Operand<U> segmentIds, Operand<V> numSegments) {
+  public static <T extends Tensor, U extends Tensor & TNumber, V extends Tensor & TNumber> UnsortedSegmentSum<T> create(Scope scope, Operand<T> data, Operand<U> segmentIds, Operand<V> numSegments) {
     OperationBuilder opBuilder = scope.env().opBuilder("UnsortedSegmentSum", scope.makeOpName("UnsortedSegmentSum"));
     opBuilder.addInput(data.asOutput());
     opBuilder.addInput(segmentIds.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Xdivy.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Xdivy.java
index 1a5e57836fc..403cdd08849 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Xdivy.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Xdivy.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Returns 0 if x == 0, and x / y otherwise, elementwise.
@@ -33,7 +33,7 @@
  * @param <T> data type for {@code z()} output
  */
 @Operator(group = "math")
-public final class Xdivy<T extends TType> extends RawOp implements Operand<T> {
+public final class Xdivy<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Xdivy operation.
@@ -44,7 +44,7 @@ public final class Xdivy<T extends TType> extends RawOp implements Operand<T> {
    * @return a new instance of Xdivy
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> Xdivy<T> create(Scope scope, Operand<T> x, Operand<T> y) {
+  public static <T extends Tensor> Xdivy<T> create(Scope scope, Operand<T> x, Operand<T> y) {
     OperationBuilder opBuilder = scope.env().opBuilder("Xdivy", scope.makeOpName("Xdivy"));
     opBuilder.addInput(x.asOutput());
     opBuilder.addInput(y.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Xlog1py.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Xlog1py.java
index a15f29dab04..922d17a5913 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Xlog1py.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Xlog1py.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Returns 0 if x == 0, and x * log1p(y) otherwise, elementwise.
@@ -33,7 +33,7 @@
  * @param <T> data type for {@code z()} output
  */
 @Operator(group = "math")
-public final class Xlog1py<T extends TType> extends RawOp implements Operand<T> {
+public final class Xlog1py<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Xlog1py operation.
@@ -44,7 +44,7 @@ public final class Xlog1py<T extends TType> extends RawOp implements Operand<T>
    * @return a new instance of Xlog1py
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> Xlog1py<T> create(Scope scope, Operand<T> x, Operand<T> y) {
+  public static <T extends Tensor> Xlog1py<T> create(Scope scope, Operand<T> x, Operand<T> y) {
     OperationBuilder opBuilder = scope.env().opBuilder("Xlog1py", scope.makeOpName("Xlog1py"));
     opBuilder.addInput(x.asOutput());
     opBuilder.addInput(y.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Xlogy.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Xlogy.java
index a140c8aa6fe..c3faff9fc5f 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Xlogy.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Xlogy.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Returns 0 if x == 0, and x * log(y) otherwise, elementwise.
@@ -33,7 +33,7 @@
  * @param <T> data type for {@code z()} output
  */
 @Operator(group = "math")
-public final class Xlogy<T extends TType> extends RawOp implements Operand<T> {
+public final class Xlogy<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Xlogy operation.
@@ -44,7 +44,7 @@ public final class Xlogy<T extends TType> extends RawOp implements Operand<T> {
    * @return a new instance of Xlogy
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> Xlogy<T> create(Scope scope, Operand<T> x, Operand<T> y) {
+  public static <T extends Tensor> Xlogy<T> create(Scope scope, Operand<T> x, Operand<T> y) {
     OperationBuilder opBuilder = scope.env().opBuilder("Xlogy", scope.makeOpName("Xlogy"));
     opBuilder.addInput(x.asOutput());
     opBuilder.addInput(y.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Zeta.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Zeta.java
index 0f3cee188fc..73d33a70736 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Zeta.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Zeta.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Compute the Hurwitz zeta function \\(\zeta(x, q)\\).
@@ -38,7 +38,7 @@
  * @param <T> data type for {@code z()} output
  */
 @Operator(group = "math")
-public final class Zeta<T extends TNumber> extends RawOp implements Operand<T> {
+public final class Zeta<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Zeta operation.
@@ -49,7 +49,7 @@ public final class Zeta<T extends TNumber> extends RawOp implements Operand<T> {
    * @return a new instance of Zeta
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> Zeta<T> create(Scope scope, Operand<T> x, Operand<T> q) {
+  public static <T extends Tensor & TNumber> Zeta<T> create(Scope scope, Operand<T> x, Operand<T> q) {
     OperationBuilder opBuilder = scope.env().opBuilder("Zeta", scope.makeOpName("Zeta"));
     opBuilder.addInput(x.asOutput());
     opBuilder.addInput(q.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/erfinv.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/erfinv.java
index de2a4482f65..22fba9db393 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/erfinv.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/erfinv.java
@@ -21,18 +21,18 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * @param <T> data type for {@code y()} output
  */
 @Operator(group = "math")
-public final class erfinv<T extends TNumber> extends RawOp implements Operand<T> {
+public final class erfinv<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new erfinv operation.
@@ -42,7 +42,7 @@ public final class erfinv<T extends TNumber> extends RawOp implements Operand<T>
    * @return a new instance of erfinv
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> erfinv<T> create(Scope scope, Operand<T> x) {
+  public static <T extends Tensor & TNumber> erfinv<T> create(Scope scope, Operand<T> x) {
     OperationBuilder opBuilder = scope.env().opBuilder("Erfinv", scope.makeOpName("erfinv"));
     opBuilder.addInput(x.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/Dawsn.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/Dawsn.java
index 74388434149..6af6cea9003 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/Dawsn.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/Dawsn.java
@@ -21,17 +21,17 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * @param <T> data type for {@code y()} output
  */
-public final class Dawsn<T extends TNumber> extends RawOp implements Operand<T> {
+public final class Dawsn<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Dawsn operation.
@@ -41,7 +41,7 @@ public final class Dawsn<T extends TNumber> extends RawOp implements Operand<T>
    * @return a new instance of Dawsn
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> Dawsn<T> create(Scope scope, Operand<T> x) {
+  public static <T extends Tensor & TNumber> Dawsn<T> create(Scope scope, Operand<T> x) {
     OperationBuilder opBuilder = scope.env().opBuilder("Dawsn", scope.makeOpName("Dawsn"));
     opBuilder.addInput(x.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/Expint.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/Expint.java
index b36c55fdeb6..c2ca8f678bb 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/Expint.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/Expint.java
@@ -21,17 +21,17 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * @param <T> data type for {@code y()} output
  */
-public final class Expint<T extends TNumber> extends RawOp implements Operand<T> {
+public final class Expint<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Expint operation.
@@ -41,7 +41,7 @@ public final class Expint<T extends TNumber> extends RawOp implements Operand<T>
    * @return a new instance of Expint
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> Expint<T> create(Scope scope, Operand<T> x) {
+  public static <T extends Tensor & TNumber> Expint<T> create(Scope scope, Operand<T> x) {
     OperationBuilder opBuilder = scope.env().opBuilder("Expint", scope.makeOpName("Expint"));
     opBuilder.addInput(x.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/FresnelCos.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/FresnelCos.java
index bb9a9f47e78..0dcc2c2ef1a 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/FresnelCos.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/FresnelCos.java
@@ -21,17 +21,17 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * @param <T> data type for {@code y()} output
  */
-public final class FresnelCos<T extends TNumber> extends RawOp implements Operand<T> {
+public final class FresnelCos<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new FresnelCos operation.
@@ -41,7 +41,7 @@ public final class FresnelCos<T extends TNumber> extends RawOp implements Operan
    * @return a new instance of FresnelCos
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> FresnelCos<T> create(Scope scope, Operand<T> x) {
+  public static <T extends Tensor & TNumber> FresnelCos<T> create(Scope scope, Operand<T> x) {
     OperationBuilder opBuilder = scope.env().opBuilder("FresnelCos", scope.makeOpName("FresnelCos"));
     opBuilder.addInput(x.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/FresnelSin.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/FresnelSin.java
index 36681c87678..4fb9a95d462 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/FresnelSin.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/FresnelSin.java
@@ -21,17 +21,17 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * @param <T> data type for {@code y()} output
  */
-public final class FresnelSin<T extends TNumber> extends RawOp implements Operand<T> {
+public final class FresnelSin<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new FresnelSin operation.
@@ -41,7 +41,7 @@ public final class FresnelSin<T extends TNumber> extends RawOp implements Operan
    * @return a new instance of FresnelSin
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> FresnelSin<T> create(Scope scope, Operand<T> x) {
+  public static <T extends Tensor & TNumber> FresnelSin<T> create(Scope scope, Operand<T> x) {
     OperationBuilder opBuilder = scope.env().opBuilder("FresnelSin", scope.makeOpName("FresnelSin"));
     opBuilder.addInput(x.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/Spence.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/Spence.java
index ed613a28b1a..f7f1931e62d 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/Spence.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/Spence.java
@@ -21,17 +21,17 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * @param <T> data type for {@code y()} output
  */
-public final class Spence<T extends TNumber> extends RawOp implements Operand<T> {
+public final class Spence<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Spence operation.
@@ -41,7 +41,7 @@ public final class Spence<T extends TNumber> extends RawOp implements Operand<T>
    * @return a new instance of Spence
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> Spence<T> create(Scope scope, Operand<T> x) {
+  public static <T extends Tensor & TNumber> Spence<T> create(Scope scope, Operand<T> x) {
     OperationBuilder opBuilder = scope.env().opBuilder("Spence", scope.makeOpName("Spence"));
     opBuilder.addInput(x.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/AvgPool.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/AvgPool.java
index 527d1a49713..a45e80966c0 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/AvgPool.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/AvgPool.java
@@ -22,12 +22,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Performs average pooling on the input.
@@ -38,7 +38,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "nn")
-public final class AvgPool<T extends TNumber> extends RawOp implements Operand<T> {
+public final class AvgPool<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.nn.AvgPool}
@@ -75,7 +75,7 @@ private Options() {
    * @return a new instance of AvgPool
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> AvgPool<T> create(Scope scope, Operand<T> value, List<Long> ksize, List<Long> strides, String padding, Options... options) {
+  public static <T extends Tensor & TNumber> AvgPool<T> create(Scope scope, Operand<T> value, List<Long> ksize, List<Long> strides, String padding, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("AvgPool", scope.makeOpName("AvgPool"));
     opBuilder.addInput(value.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/AvgPool3d.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/AvgPool3d.java
index 87467c8a982..8c17badbb97 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/AvgPool3d.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/AvgPool3d.java
@@ -22,12 +22,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Performs 3D average pooling on the input.
@@ -35,7 +35,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "nn")
-public final class AvgPool3d<T extends TNumber> extends RawOp implements Operand<T> {
+public final class AvgPool3d<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.nn.AvgPool3d}
@@ -74,7 +74,7 @@ private Options() {
    * @return a new instance of AvgPool3d
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> AvgPool3d<T> create(Scope scope, Operand<T> input, List<Long> ksize, List<Long> strides, String padding, Options... options) {
+  public static <T extends Tensor & TNumber> AvgPool3d<T> create(Scope scope, Operand<T> input, List<Long> ksize, List<Long> strides, String padding, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("AvgPool3D", scope.makeOpName("AvgPool3d"));
     opBuilder.addInput(input.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/AvgPool3dGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/AvgPool3dGrad.java
index 3e34c87f9b6..5c75244a47e 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/AvgPool3dGrad.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/AvgPool3dGrad.java
@@ -22,13 +22,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes gradients of average pooling function.
@@ -36,7 +36,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "nn")
-public final class AvgPool3dGrad<T extends TNumber> extends RawOp implements Operand<T> {
+public final class AvgPool3dGrad<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.nn.AvgPool3dGrad}
@@ -76,7 +76,7 @@ private Options() {
    * @return a new instance of AvgPool3dGrad
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> AvgPool3dGrad<T> create(Scope scope, Operand<TInt32> origInputShape, Operand<T> grad, List<Long> ksize, List<Long> strides, String padding, Options... options) {
+  public static <T extends Tensor & TNumber> AvgPool3dGrad<T> create(Scope scope, Operand<TInt32> origInputShape, Operand<T> grad, List<Long> ksize, List<Long> strides, String padding, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("AvgPool3DGrad", scope.makeOpName("AvgPool3dGrad"));
     opBuilder.addInput(origInputShape.asOutput());
     opBuilder.addInput(grad.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/AvgPoolGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/AvgPoolGrad.java
index 7ec252eaac0..0229a5d08e7 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/AvgPoolGrad.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/AvgPoolGrad.java
@@ -22,20 +22,20 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes gradients of the average pooling function.
  * 
  * @param <T> data type for {@code output()} output
  */
-public final class AvgPoolGrad<T extends TNumber> extends RawOp implements Operand<T> {
+public final class AvgPoolGrad<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.nn.AvgPoolGrad}
@@ -74,7 +74,7 @@ private Options() {
    * @return a new instance of AvgPoolGrad
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> AvgPoolGrad<T> create(Scope scope, Operand<TInt32> origInputShape, Operand<T> grad, List<Long> ksize, List<Long> strides, String padding, Options... options) {
+  public static <T extends Tensor & TNumber> AvgPoolGrad<T> create(Scope scope, Operand<TInt32> origInputShape, Operand<T> grad, List<Long> ksize, List<Long> strides, String padding, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("AvgPoolGrad", scope.makeOpName("AvgPoolGrad"));
     opBuilder.addInput(origInputShape.asOutput());
     opBuilder.addInput(grad.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/BatchNormWithGlobalNormalization.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/BatchNormWithGlobalNormalization.java
index 6c30a52cf2e..964e00dab01 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/BatchNormWithGlobalNormalization.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/BatchNormWithGlobalNormalization.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Batch normalization.
@@ -35,7 +35,7 @@
  * @param <T> data type for {@code result()} output
  */
 @Operator(group = "nn")
-public final class BatchNormWithGlobalNormalization<T extends TType> extends RawOp implements Operand<T> {
+public final class BatchNormWithGlobalNormalization<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new BatchNormWithGlobalNormalization operation.
@@ -59,7 +59,7 @@ public final class BatchNormWithGlobalNormalization<T extends TType> extends Raw
    * @return a new instance of BatchNormWithGlobalNormalization
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> BatchNormWithGlobalNormalization<T> create(Scope scope, Operand<T> t, Operand<T> m, Operand<T> v, Operand<T> beta, Operand<T> gamma, Float varianceEpsilon, Boolean scaleAfterNormalization) {
+  public static <T extends Tensor> BatchNormWithGlobalNormalization<T> create(Scope scope, Operand<T> t, Operand<T> m, Operand<T> v, Operand<T> beta, Operand<T> gamma, Float varianceEpsilon, Boolean scaleAfterNormalization) {
     OperationBuilder opBuilder = scope.env().opBuilder("BatchNormWithGlobalNormalization", scope.makeOpName("BatchNormWithGlobalNormalization"));
     opBuilder.addInput(t.asOutput());
     opBuilder.addInput(m.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/BatchNormWithGlobalNormalizationGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/BatchNormWithGlobalNormalizationGrad.java
index be07a5eea25..150542bc939 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/BatchNormWithGlobalNormalizationGrad.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/BatchNormWithGlobalNormalizationGrad.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Gradients for batch normalization.
@@ -35,7 +35,7 @@
  * @param <T> data type for {@code dx()} output
  */
 @Operator(group = "nn")
-public final class BatchNormWithGlobalNormalizationGrad<T extends TType> extends RawOp {
+public final class BatchNormWithGlobalNormalizationGrad<T extends Tensor> extends RawOp {
   
   /**
    * Factory method to create a class wrapping a new BatchNormWithGlobalNormalizationGrad operation.
@@ -58,7 +58,7 @@ public final class BatchNormWithGlobalNormalizationGrad<T extends TType> extends
    * @return a new instance of BatchNormWithGlobalNormalizationGrad
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> BatchNormWithGlobalNormalizationGrad<T> create(Scope scope, Operand<T> t, Operand<T> m, Operand<T> v, Operand<T> gamma, Operand<T> backprop, Float varianceEpsilon, Boolean scaleAfterNormalization) {
+  public static <T extends Tensor> BatchNormWithGlobalNormalizationGrad<T> create(Scope scope, Operand<T> t, Operand<T> m, Operand<T> v, Operand<T> gamma, Operand<T> backprop, Float varianceEpsilon, Boolean scaleAfterNormalization) {
     OperationBuilder opBuilder = scope.env().opBuilder("BatchNormWithGlobalNormalizationGrad", scope.makeOpName("BatchNormWithGlobalNormalizationGrad"));
     opBuilder.addInput(t.asOutput());
     opBuilder.addInput(m.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/BiasAdd.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/BiasAdd.java
index 462eaa6edc7..418365edffc 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/BiasAdd.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/BiasAdd.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Adds `bias` to `value`.
@@ -36,7 +36,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "nn")
-public final class BiasAdd<T extends TType> extends RawOp implements Operand<T> {
+public final class BiasAdd<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.nn.BiasAdd}
@@ -73,7 +73,7 @@ private Options() {
    * @return a new instance of BiasAdd
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> BiasAdd<T> create(Scope scope, Operand<T> value, Operand<T> bias, Options... options) {
+  public static <T extends Tensor> BiasAdd<T> create(Scope scope, Operand<T> value, Operand<T> bias, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("BiasAdd", scope.makeOpName("BiasAdd"));
     opBuilder.addInput(value.asOutput());
     opBuilder.addInput(bias.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/BiasAddGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/BiasAddGrad.java
index df90cf9a2f3..0ec94390b18 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/BiasAddGrad.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/BiasAddGrad.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * The backward operation for "BiasAdd" on the "bias" tensor.
@@ -37,7 +37,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "nn")
-public final class BiasAddGrad<T extends TType> extends RawOp implements Operand<T> {
+public final class BiasAddGrad<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.nn.BiasAddGrad}
@@ -73,7 +73,7 @@ private Options() {
    * @return a new instance of BiasAddGrad
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> BiasAddGrad<T> create(Scope scope, Operand<T> outBackprop, Options... options) {
+  public static <T extends Tensor> BiasAddGrad<T> create(Scope scope, Operand<T> outBackprop, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("BiasAddGrad", scope.makeOpName("BiasAddGrad"));
     opBuilder.addInput(outBackprop.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/BlockLSTM.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/BlockLSTM.java
index 79de4f2f88c..9437b6d7d0f 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/BlockLSTM.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/BlockLSTM.java
@@ -21,13 +21,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes the LSTM cell forward propagation for all the time steps.
@@ -56,7 +56,7 @@
  * 
  * @param <T> data type for {@code i()} output
  */
-public final class BlockLSTM<T extends TNumber> extends RawOp {
+public final class BlockLSTM<T extends Tensor & TNumber> extends RawOp {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.nn.BlockLSTM}
@@ -104,7 +104,7 @@ private Options() {
    * @return a new instance of BlockLSTM
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> BlockLSTM<T> create(Scope scope, Operand<TInt64> seqLenMax, Operand<T> x, Operand<T> csPrev, Operand<T> hPrev, Operand<T> w, Operand<T> wci, Operand<T> wcf, Operand<T> wco, Operand<T> b, Options... options) {
+  public static <T extends Tensor & TNumber> BlockLSTM<T> create(Scope scope, Operand<TInt64> seqLenMax, Operand<T> x, Operand<T> csPrev, Operand<T> hPrev, Operand<T> w, Operand<T> wci, Operand<T> wcf, Operand<T> wco, Operand<T> b, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("BlockLSTMV2", scope.makeOpName("BlockLSTM"));
     opBuilder.addInput(seqLenMax.asOutput());
     opBuilder.addInput(x.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/BlockLSTMGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/BlockLSTMGrad.java
index 4e49f23cb6e..56a10702e5d 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/BlockLSTMGrad.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/BlockLSTMGrad.java
@@ -21,13 +21,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes the LSTM cell backward propagation for the entire time sequence.
@@ -36,7 +36,7 @@
  * 
  * @param <T> data type for {@code xGrad()} output
  */
-public final class BlockLSTMGrad<T extends TNumber> extends RawOp {
+public final class BlockLSTMGrad<T extends Tensor & TNumber> extends RawOp {
   
   /**
    * Factory method to create a class wrapping a new BlockLSTMGrad operation.
@@ -65,7 +65,7 @@ public final class BlockLSTMGrad<T extends TNumber> extends RawOp {
    * @return a new instance of BlockLSTMGrad
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> BlockLSTMGrad<T> create(Scope scope, Operand<TInt64> seqLenMax, Operand<T> x, Operand<T> csPrev, Operand<T> hPrev, Operand<T> w, Operand<T> wci, Operand<T> wcf, Operand<T> wco, Operand<T> b, Operand<T> i, Operand<T> cs, Operand<T> f, Operand<T> o, Operand<T> ci, Operand<T> co, Operand<T> h, Operand<T> csGrad, Operand<T> hGrad, Boolean usePeephole) {
+  public static <T extends Tensor & TNumber> BlockLSTMGrad<T> create(Scope scope, Operand<TInt64> seqLenMax, Operand<T> x, Operand<T> csPrev, Operand<T> hPrev, Operand<T> w, Operand<T> wci, Operand<T> wcf, Operand<T> wco, Operand<T> b, Operand<T> i, Operand<T> cs, Operand<T> f, Operand<T> o, Operand<T> ci, Operand<T> co, Operand<T> h, Operand<T> csGrad, Operand<T> hGrad, Boolean usePeephole) {
     OperationBuilder opBuilder = scope.env().opBuilder("BlockLSTMGradV2", scope.makeOpName("BlockLSTMGrad"));
     opBuilder.addInput(seqLenMax.asOutput());
     opBuilder.addInput(x.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Conv2d.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Conv2d.java
index 79f2022d807..e194c491408 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Conv2d.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Conv2d.java
@@ -22,12 +22,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes a 2-D convolution given 4-D `input` and `filter` tensors.
@@ -57,7 +57,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "nn")
-public final class Conv2d<T extends TNumber> extends RawOp implements Operand<T> {
+public final class Conv2d<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.nn.Conv2d}
@@ -132,7 +132,7 @@ private Options() {
    * @return a new instance of Conv2d
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> Conv2d<T> create(Scope scope, Operand<T> input, Operand<T> filter, List<Long> strides, String padding, Options... options) {
+  public static <T extends Tensor & TNumber> Conv2d<T> create(Scope scope, Operand<T> input, Operand<T> filter, List<Long> strides, String padding, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("Conv2D", scope.makeOpName("Conv2d"));
     opBuilder.addInput(input.asOutput());
     opBuilder.addInput(filter.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Conv2dBackpropFilter.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Conv2dBackpropFilter.java
index 9c145dc3d60..5e11f5bbcb1 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Conv2dBackpropFilter.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Conv2dBackpropFilter.java
@@ -22,13 +22,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes the gradients of convolution with respect to the filter.
@@ -36,7 +36,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "nn")
-public final class Conv2dBackpropFilter<T extends TNumber> extends RawOp implements Operand<T> {
+public final class Conv2dBackpropFilter<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.nn.Conv2dBackpropFilter}
@@ -113,7 +113,7 @@ private Options() {
    * @return a new instance of Conv2dBackpropFilter
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> Conv2dBackpropFilter<T> create(Scope scope, Operand<T> input, Operand<TInt32> filterSizes, Operand<T> outBackprop, List<Long> strides, String padding, Options... options) {
+  public static <T extends Tensor & TNumber> Conv2dBackpropFilter<T> create(Scope scope, Operand<T> input, Operand<TInt32> filterSizes, Operand<T> outBackprop, List<Long> strides, String padding, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("Conv2DBackpropFilter", scope.makeOpName("Conv2dBackpropFilter"));
     opBuilder.addInput(input.asOutput());
     opBuilder.addInput(filterSizes.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Conv2dBackpropInput.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Conv2dBackpropInput.java
index 492c4f2aebb..23fd6a45aed 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Conv2dBackpropInput.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Conv2dBackpropInput.java
@@ -22,13 +22,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes the gradients of convolution with respect to the input.
@@ -36,7 +36,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "nn")
-public final class Conv2dBackpropInput<T extends TNumber> extends RawOp implements Operand<T> {
+public final class Conv2dBackpropInput<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.nn.Conv2dBackpropInput}
@@ -113,7 +113,7 @@ private Options() {
    * @return a new instance of Conv2dBackpropInput
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> Conv2dBackpropInput<T> create(Scope scope, Operand<TInt32> inputSizes, Operand<T> filter, Operand<T> outBackprop, List<Long> strides, String padding, Options... options) {
+  public static <T extends Tensor & TNumber> Conv2dBackpropInput<T> create(Scope scope, Operand<TInt32> inputSizes, Operand<T> filter, Operand<T> outBackprop, List<Long> strides, String padding, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("Conv2DBackpropInput", scope.makeOpName("Conv2dBackpropInput"));
     opBuilder.addInput(inputSizes.asOutput());
     opBuilder.addInput(filter.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Conv3d.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Conv3d.java
index 94ae4ffd2b5..49817b63e2c 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Conv3d.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Conv3d.java
@@ -22,12 +22,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes a 3-D convolution given 5-D `input` and `filter` tensors.
@@ -41,7 +41,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "nn")
-public final class Conv3d<T extends TNumber> extends RawOp implements Operand<T> {
+public final class Conv3d<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.nn.Conv3d}
@@ -93,7 +93,7 @@ private Options() {
    * @return a new instance of Conv3d
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> Conv3d<T> create(Scope scope, Operand<T> input, Operand<T> filter, List<Long> strides, String padding, Options... options) {
+  public static <T extends Tensor & TNumber> Conv3d<T> create(Scope scope, Operand<T> input, Operand<T> filter, List<Long> strides, String padding, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("Conv3D", scope.makeOpName("Conv3d"));
     opBuilder.addInput(input.asOutput());
     opBuilder.addInput(filter.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Conv3dBackpropFilter.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Conv3dBackpropFilter.java
index 0d73e491717..7a6bc2549f4 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Conv3dBackpropFilter.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Conv3dBackpropFilter.java
@@ -22,13 +22,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes the gradients of 3-D convolution with respect to the filter.
@@ -36,7 +36,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "nn")
-public final class Conv3dBackpropFilter<T extends TNumber> extends RawOp implements Operand<T> {
+public final class Conv3dBackpropFilter<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.nn.Conv3dBackpropFilter}
@@ -92,7 +92,7 @@ private Options() {
    * @return a new instance of Conv3dBackpropFilter
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> Conv3dBackpropFilter<T> create(Scope scope, Operand<T> input, Operand<TInt32> filterSizes, Operand<T> outBackprop, List<Long> strides, String padding, Options... options) {
+  public static <T extends Tensor & TNumber> Conv3dBackpropFilter<T> create(Scope scope, Operand<T> input, Operand<TInt32> filterSizes, Operand<T> outBackprop, List<Long> strides, String padding, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("Conv3DBackpropFilterV2", scope.makeOpName("Conv3dBackpropFilter"));
     opBuilder.addInput(input.asOutput());
     opBuilder.addInput(filterSizes.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Conv3dBackpropInput.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Conv3dBackpropInput.java
index 8b153890811..478d4841a0c 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Conv3dBackpropInput.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Conv3dBackpropInput.java
@@ -22,12 +22,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes the gradients of 3-D convolution with respect to the input.
@@ -35,7 +35,7 @@
  * @param <U> data type for {@code output()} output
  */
 @Operator(group = "nn")
-public final class Conv3dBackpropInput<U extends TNumber> extends RawOp implements Operand<U> {
+public final class Conv3dBackpropInput<U extends Tensor & TNumber> extends RawOp implements Operand<U> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.nn.Conv3dBackpropInput}
@@ -91,7 +91,7 @@ private Options() {
    * @return a new instance of Conv3dBackpropInput
    */
   @Endpoint(describeByClass = true)
-  public static <U extends TNumber, T extends TNumber> Conv3dBackpropInput<U> create(Scope scope, Operand<T> inputSizes, Operand<U> filter, Operand<U> outBackprop, List<Long> strides, String padding, Options... options) {
+  public static <U extends Tensor & TNumber, T extends Tensor & TNumber> Conv3dBackpropInput<U> create(Scope scope, Operand<T> inputSizes, Operand<U> filter, Operand<U> outBackprop, List<Long> strides, String padding, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("Conv3DBackpropInputV2", scope.makeOpName("Conv3dBackpropInput"));
     opBuilder.addInput(inputSizes.asOutput());
     opBuilder.addInput(filter.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CtcBeamSearchDecoder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CtcBeamSearchDecoder.java
index 96f179641e7..d2f84a70d26 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CtcBeamSearchDecoder.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CtcBeamSearchDecoder.java
@@ -23,6 +23,7 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -30,7 +31,6 @@
 import org.tensorflow.types.TInt32;
 import org.tensorflow.types.TInt64;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Performs beam search decoding on the logits given in input.
@@ -44,7 +44,7 @@
  * @param <T> data type for {@code logProbability()} output
  */
 @Operator(group = "nn")
-public final class CtcBeamSearchDecoder<T extends TNumber> extends RawOp {
+public final class CtcBeamSearchDecoder<T extends Tensor & TNumber> extends RawOp {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.nn.CtcBeamSearchDecoder}
@@ -77,7 +77,7 @@ private Options() {
    * @return a new instance of CtcBeamSearchDecoder
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> CtcBeamSearchDecoder<T> create(Scope scope, Operand<T> inputs, Operand<TInt32> sequenceLength, Long beamWidth, Long topPaths, Options... options) {
+  public static <T extends Tensor & TNumber> CtcBeamSearchDecoder<T> create(Scope scope, Operand<T> inputs, Operand<TInt32> sequenceLength, Long beamWidth, Long topPaths, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("CTCBeamSearchDecoder", scope.makeOpName("CtcBeamSearchDecoder"));
     opBuilder.addInput(inputs.asOutput());
     opBuilder.addInput(sequenceLength.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CtcGreedyDecoder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CtcGreedyDecoder.java
index b8b35b8ceaa..93c505c22c7 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CtcGreedyDecoder.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CtcGreedyDecoder.java
@@ -21,6 +21,7 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -28,7 +29,6 @@
 import org.tensorflow.types.TInt32;
 import org.tensorflow.types.TInt64;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Performs greedy decoding on the logits given in inputs.
@@ -46,7 +46,7 @@
  * @param <T> data type for {@code logProbability()} output
  */
 @Operator(group = "nn")
-public final class CtcGreedyDecoder<T extends TNumber> extends RawOp {
+public final class CtcGreedyDecoder<T extends Tensor & TNumber> extends RawOp {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.nn.CtcGreedyDecoder}
@@ -77,7 +77,7 @@ private Options() {
    * @return a new instance of CtcGreedyDecoder
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> CtcGreedyDecoder<T> create(Scope scope, Operand<T> inputs, Operand<TInt32> sequenceLength, Options... options) {
+  public static <T extends Tensor & TNumber> CtcGreedyDecoder<T> create(Scope scope, Operand<T> inputs, Operand<TInt32> sequenceLength, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("CTCGreedyDecoder", scope.makeOpName("CtcGreedyDecoder"));
     opBuilder.addInput(inputs.asOutput());
     opBuilder.addInput(sequenceLength.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CtcLoss.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CtcLoss.java
index ceb60a4baf7..e464a6da181 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CtcLoss.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CtcLoss.java
@@ -21,6 +21,7 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -28,7 +29,6 @@
 import org.tensorflow.types.TInt32;
 import org.tensorflow.types.TInt64;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Calculates the CTC Loss (log probability) for each batch entry.  Also calculates
@@ -39,7 +39,7 @@
  * @param <T> data type for {@code loss()} output
  */
 @Operator(group = "nn")
-public final class CtcLoss<T extends TNumber> extends RawOp {
+public final class CtcLoss<T extends Tensor & TNumber> extends RawOp {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.nn.CtcLoss}
@@ -97,7 +97,7 @@ private Options() {
    * @return a new instance of CtcLoss
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> CtcLoss<T> create(Scope scope, Operand<T> inputs, Operand<TInt64> labelsIndices, Operand<TInt32> labelsValues, Operand<TInt32> sequenceLength, Options... options) {
+  public static <T extends Tensor & TNumber> CtcLoss<T> create(Scope scope, Operand<T> inputs, Operand<TInt64> labelsIndices, Operand<TInt32> labelsValues, Operand<TInt32> sequenceLength, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("CTCLoss", scope.makeOpName("CtcLoss"));
     opBuilder.addInput(inputs.asOutput());
     opBuilder.addInput(labelsIndices.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CudnnRNN.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CudnnRNN.java
index 7c179d7e578..1c482ad4f14 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CudnnRNN.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CudnnRNN.java
@@ -21,13 +21,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * A RNN backed by cuDNN.
@@ -72,7 +72,7 @@
  * 
  * @param <T> data type for {@code output()} output
  */
-public final class CudnnRNN<T extends TNumber> extends RawOp {
+public final class CudnnRNN<T extends Tensor & TNumber> extends RawOp {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.nn.CudnnRNN}
@@ -178,7 +178,7 @@ private Options() {
    * @return a new instance of CudnnRNN
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> CudnnRNN<T> create(Scope scope, Operand<T> input, Operand<T> inputH, Operand<T> inputC, Operand<T> params, Operand<TInt32> sequenceLengths, Options... options) {
+  public static <T extends Tensor & TNumber> CudnnRNN<T> create(Scope scope, Operand<T> input, Operand<T> inputH, Operand<T> inputC, Operand<T> params, Operand<TInt32> sequenceLengths, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("CudnnRNNV3", scope.makeOpName("CudnnRNN"));
     opBuilder.addInput(input.asOutput());
     opBuilder.addInput(inputH.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CudnnRNNBackprop.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CudnnRNNBackprop.java
index 3719c186716..0dd6347a4ab 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CudnnRNNBackprop.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CudnnRNNBackprop.java
@@ -21,13 +21,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Backprop step of CudnnRNNV3.
@@ -82,7 +82,7 @@
  * 
  * @param <T> data type for {@code inputBackprop()} output
  */
-public final class CudnnRNNBackprop<T extends TNumber> extends RawOp {
+public final class CudnnRNNBackprop<T extends Tensor & TNumber> extends RawOp {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.nn.CudnnRNNBackprop}
@@ -187,7 +187,7 @@ private Options() {
    * @return a new instance of CudnnRNNBackprop
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> CudnnRNNBackprop<T> create(Scope scope, Operand<T> input, Operand<T> inputH, Operand<T> inputC, Operand<T> params, Operand<TInt32> sequenceLengths, Operand<T> output, Operand<T> outputH, Operand<T> outputC, Operand<T> outputBackprop, Operand<T> outputHBackprop, Operand<T> outputCBackprop, Operand<T> reserveSpace, Operand<?> hostReserved, Options... options) {
+  public static <T extends Tensor & TNumber> CudnnRNNBackprop<T> create(Scope scope, Operand<T> input, Operand<T> inputH, Operand<T> inputC, Operand<T> params, Operand<TInt32> sequenceLengths, Operand<T> output, Operand<T> outputH, Operand<T> outputC, Operand<T> outputBackprop, Operand<T> outputHBackprop, Operand<T> outputCBackprop, Operand<T> reserveSpace, Operand<?> hostReserved, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("CudnnRNNBackpropV3", scope.makeOpName("CudnnRNNBackprop"));
     opBuilder.addInput(input.asOutput());
     opBuilder.addInput(inputH.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CudnnRNNCanonicalToParams.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CudnnRNNCanonicalToParams.java
index 155cfd2c0a4..efc3c0ed8db 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CudnnRNNCanonicalToParams.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CudnnRNNCanonicalToParams.java
@@ -21,6 +21,7 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.Operands;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
@@ -28,7 +29,6 @@
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Converts CudnnRNN params from canonical form to usable form. It supports the projection in LSTM.
@@ -67,7 +67,7 @@
  * @param <T> data type for {@code params()} output
  */
 @Operator(group = "nn")
-public final class CudnnRNNCanonicalToParams<T extends TNumber> extends RawOp implements Operand<T> {
+public final class CudnnRNNCanonicalToParams<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.nn.CudnnRNNCanonicalToParams}
@@ -155,7 +155,7 @@ private Options() {
    * @return a new instance of CudnnRNNCanonicalToParams
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> CudnnRNNCanonicalToParams<T> create(Scope scope, Operand<TInt32> numLayers, Operand<TInt32> numUnits, Operand<TInt32> inputSize, Iterable<Operand<T>> weights, Iterable<Operand<T>> biases, Options... options) {
+  public static <T extends Tensor & TNumber> CudnnRNNCanonicalToParams<T> create(Scope scope, Operand<TInt32> numLayers, Operand<TInt32> numUnits, Operand<TInt32> inputSize, Iterable<Operand<T>> weights, Iterable<Operand<T>> biases, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("CudnnRNNCanonicalToParamsV2", scope.makeOpName("CudnnRNNCanonicalToParams"));
     opBuilder.addInput(numLayers.asOutput());
     opBuilder.addInput(numUnits.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CudnnRNNParamsToCanonical.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CudnnRNNParamsToCanonical.java
index ea575f2d7a1..b6b09158eb6 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CudnnRNNParamsToCanonical.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CudnnRNNParamsToCanonical.java
@@ -23,13 +23,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Retrieves CudnnRNN params in canonical form. It supports the projection in LSTM.
@@ -68,7 +68,7 @@
  * @param <T> data type for {@code weights()} output
  */
 @Operator(group = "nn")
-public final class CudnnRNNParamsToCanonical<T extends TNumber> extends RawOp {
+public final class CudnnRNNParamsToCanonical<T extends Tensor & TNumber> extends RawOp {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.nn.CudnnRNNParamsToCanonical}
@@ -157,7 +157,7 @@ private Options() {
    * @return a new instance of CudnnRNNParamsToCanonical
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> CudnnRNNParamsToCanonical<T> create(Scope scope, Operand<TInt32> numLayers, Operand<TInt32> numUnits, Operand<TInt32> inputSize, Operand<T> params, Long numParamsWeights, Long numParamsBiases, Options... options) {
+  public static <T extends Tensor & TNumber> CudnnRNNParamsToCanonical<T> create(Scope scope, Operand<TInt32> numLayers, Operand<TInt32> numUnits, Operand<TInt32> inputSize, Operand<T> params, Long numParamsWeights, Long numParamsBiases, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("CudnnRNNParamsToCanonicalV2", scope.makeOpName("CudnnRNNParamsToCanonical"));
     opBuilder.addInput(numLayers.asOutput());
     opBuilder.addInput(numUnits.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CudnnRnnParamsSize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CudnnRnnParamsSize.java
index 94421d0aa6e..6758e30cb0d 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CudnnRnnParamsSize.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CudnnRnnParamsSize.java
@@ -22,13 +22,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes size of weights that can be used by a Cudnn RNN model.
@@ -58,7 +58,7 @@
  * @param <U> data type for {@code paramsSize()} output
  */
 @Operator(group = "nn")
-public final class CudnnRnnParamsSize<U extends TNumber> extends RawOp implements Operand<U> {
+public final class CudnnRnnParamsSize<U extends Tensor & TNumber> extends RawOp implements Operand<U> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.nn.CudnnRnnParamsSize}
@@ -146,7 +146,7 @@ private Options() {
    * @return a new instance of CudnnRnnParamsSize
    */
   @Endpoint(describeByClass = true)
-  public static <U extends TNumber, T extends TNumber> CudnnRnnParamsSize<U> create(Scope scope, Operand<TInt32> numLayers, Operand<TInt32> numUnits, Operand<TInt32> inputSize, DataType<T> T, DataType<U> S, Options... options) {
+  public static <U extends Tensor & TNumber, T extends Tensor & TNumber> CudnnRnnParamsSize<U> create(Scope scope, Operand<TInt32> numLayers, Operand<TInt32> numUnits, Operand<TInt32> inputSize, DataType<T> T, DataType<U> S, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("CudnnRNNParamsSize", scope.makeOpName("CudnnRnnParamsSize"));
     opBuilder.addInput(numLayers.asOutput());
     opBuilder.addInput(numUnits.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/DataFormatDimMap.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/DataFormatDimMap.java
index 6b1ff40761b..8e107e01b4e 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/DataFormatDimMap.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/DataFormatDimMap.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Returns the dimension index in the destination data format given the one in
@@ -36,7 +36,7 @@
  * @param <T> data type for {@code y()} output
  */
 @Operator(group = "nn")
-public final class DataFormatDimMap<T extends TNumber> extends RawOp implements Operand<T> {
+public final class DataFormatDimMap<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.nn.DataFormatDimMap}
@@ -76,7 +76,7 @@ private Options() {
    * @return a new instance of DataFormatDimMap
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> DataFormatDimMap<T> create(Scope scope, Operand<T> x, Options... options) {
+  public static <T extends Tensor & TNumber> DataFormatDimMap<T> create(Scope scope, Operand<T> x, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("DataFormatDimMap", scope.makeOpName("DataFormatDimMap"));
     opBuilder.addInput(x.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/DataFormatVecPermute.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/DataFormatVecPermute.java
index ba218c1923f..daf50364005 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/DataFormatVecPermute.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/DataFormatVecPermute.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Returns the permuted vector/tensor in the destination data format given the
@@ -36,7 +36,7 @@
  * @param <T> data type for {@code y()} output
  */
 @Operator(group = "nn")
-public final class DataFormatVecPermute<T extends TNumber> extends RawOp implements Operand<T> {
+public final class DataFormatVecPermute<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.nn.DataFormatVecPermute}
@@ -75,7 +75,7 @@ private Options() {
    * @return a new instance of DataFormatVecPermute
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> DataFormatVecPermute<T> create(Scope scope, Operand<T> x, Options... options) {
+  public static <T extends Tensor & TNumber> DataFormatVecPermute<T> create(Scope scope, Operand<T> x, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("DataFormatVecPermute", scope.makeOpName("DataFormatVecPermute"));
     opBuilder.addInput(x.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/DepthToSpace.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/DepthToSpace.java
index 6456eff1f95..1f84ac1f2bb 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/DepthToSpace.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/DepthToSpace.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * DepthToSpace for tensors of type T.
@@ -113,7 +113,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "nn")
-public final class DepthToSpace<T extends TType> extends RawOp implements Operand<T> {
+public final class DepthToSpace<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.nn.DepthToSpace}
@@ -144,7 +144,7 @@ private Options() {
    * @return a new instance of DepthToSpace
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> DepthToSpace<T> create(Scope scope, Operand<T> input, Long blockSize, Options... options) {
+  public static <T extends Tensor> DepthToSpace<T> create(Scope scope, Operand<T> input, Long blockSize, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("DepthToSpace", scope.makeOpName("DepthToSpace"));
     opBuilder.addInput(input.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/DepthwiseConv2dNative.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/DepthwiseConv2dNative.java
index 6549d8b9dfc..cfb054de485 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/DepthwiseConv2dNative.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/DepthwiseConv2dNative.java
@@ -22,12 +22,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes a 2-D depthwise convolution given 4-D `input` and `filter` tensors.
@@ -52,7 +52,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "nn")
-public final class DepthwiseConv2dNative<T extends TNumber> extends RawOp implements Operand<T> {
+public final class DepthwiseConv2dNative<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.nn.DepthwiseConv2dNative}
@@ -112,7 +112,7 @@ private Options() {
    * @return a new instance of DepthwiseConv2dNative
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> DepthwiseConv2dNative<T> create(Scope scope, Operand<T> input, Operand<T> filter, List<Long> strides, String padding, Options... options) {
+  public static <T extends Tensor & TNumber> DepthwiseConv2dNative<T> create(Scope scope, Operand<T> input, Operand<T> filter, List<Long> strides, String padding, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("DepthwiseConv2dNative", scope.makeOpName("DepthwiseConv2dNative"));
     opBuilder.addInput(input.asOutput());
     opBuilder.addInput(filter.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/DepthwiseConv2dNativeBackpropFilter.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/DepthwiseConv2dNativeBackpropFilter.java
index fadfdacc823..a9f82a9d2f4 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/DepthwiseConv2dNativeBackpropFilter.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/DepthwiseConv2dNativeBackpropFilter.java
@@ -22,13 +22,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes the gradients of depthwise convolution with respect to the filter.
@@ -36,7 +36,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "nn")
-public final class DepthwiseConv2dNativeBackpropFilter<T extends TNumber> extends RawOp implements Operand<T> {
+public final class DepthwiseConv2dNativeBackpropFilter<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.nn.DepthwiseConv2dNativeBackpropFilter}
@@ -104,7 +104,7 @@ private Options() {
    * @return a new instance of DepthwiseConv2dNativeBackpropFilter
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> DepthwiseConv2dNativeBackpropFilter<T> create(Scope scope, Operand<T> input, Operand<TInt32> filterSizes, Operand<T> outBackprop, List<Long> strides, String padding, Options... options) {
+  public static <T extends Tensor & TNumber> DepthwiseConv2dNativeBackpropFilter<T> create(Scope scope, Operand<T> input, Operand<TInt32> filterSizes, Operand<T> outBackprop, List<Long> strides, String padding, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("DepthwiseConv2dNativeBackpropFilter", scope.makeOpName("DepthwiseConv2dNativeBackpropFilter"));
     opBuilder.addInput(input.asOutput());
     opBuilder.addInput(filterSizes.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/DepthwiseConv2dNativeBackpropInput.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/DepthwiseConv2dNativeBackpropInput.java
index d2e3e733a01..51ff8495ff3 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/DepthwiseConv2dNativeBackpropInput.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/DepthwiseConv2dNativeBackpropInput.java
@@ -22,13 +22,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes the gradients of depthwise convolution with respect to the input.
@@ -36,7 +36,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "nn")
-public final class DepthwiseConv2dNativeBackpropInput<T extends TNumber> extends RawOp implements Operand<T> {
+public final class DepthwiseConv2dNativeBackpropInput<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.nn.DepthwiseConv2dNativeBackpropInput}
@@ -103,7 +103,7 @@ private Options() {
    * @return a new instance of DepthwiseConv2dNativeBackpropInput
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> DepthwiseConv2dNativeBackpropInput<T> create(Scope scope, Operand<TInt32> inputSizes, Operand<T> filter, Operand<T> outBackprop, List<Long> strides, String padding, Options... options) {
+  public static <T extends Tensor & TNumber> DepthwiseConv2dNativeBackpropInput<T> create(Scope scope, Operand<TInt32> inputSizes, Operand<T> filter, Operand<T> outBackprop, List<Long> strides, String padding, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("DepthwiseConv2dNativeBackpropInput", scope.makeOpName("DepthwiseConv2dNativeBackpropInput"));
     opBuilder.addInput(inputSizes.asOutput());
     opBuilder.addInput(filter.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Dilation2d.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Dilation2d.java
index c7135b20361..908feaf06a0 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Dilation2d.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Dilation2d.java
@@ -22,12 +22,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes the grayscale dilation of 4-D `input` and 3-D `filter` tensors.
@@ -59,7 +59,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "nn")
-public final class Dilation2d<T extends TNumber> extends RawOp implements Operand<T> {
+public final class Dilation2d<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Dilation2d operation.
@@ -75,7 +75,7 @@ public final class Dilation2d<T extends TNumber> extends RawOp implements Operan
    * @return a new instance of Dilation2d
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> Dilation2d<T> create(Scope scope, Operand<T> input, Operand<T> filter, List<Long> strides, List<Long> rates, String padding) {
+  public static <T extends Tensor & TNumber> Dilation2d<T> create(Scope scope, Operand<T> input, Operand<T> filter, List<Long> strides, List<Long> rates, String padding) {
     OperationBuilder opBuilder = scope.env().opBuilder("Dilation2D", scope.makeOpName("Dilation2d"));
     opBuilder.addInput(input.asOutput());
     opBuilder.addInput(filter.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Dilation2dBackpropFilter.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Dilation2dBackpropFilter.java
index 9254ff8c285..9f44fbee1f8 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Dilation2dBackpropFilter.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Dilation2dBackpropFilter.java
@@ -22,12 +22,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes the gradient of morphological 2-D dilation with respect to the filter.
@@ -35,7 +35,7 @@
  * @param <T> data type for {@code filterBackprop()} output
  */
 @Operator(group = "nn")
-public final class Dilation2dBackpropFilter<T extends TNumber> extends RawOp implements Operand<T> {
+public final class Dilation2dBackpropFilter<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Dilation2dBackpropFilter operation.
@@ -52,7 +52,7 @@ public final class Dilation2dBackpropFilter<T extends TNumber> extends RawOp imp
    * @return a new instance of Dilation2dBackpropFilter
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> Dilation2dBackpropFilter<T> create(Scope scope, Operand<T> input, Operand<T> filter, Operand<T> outBackprop, List<Long> strides, List<Long> rates, String padding) {
+  public static <T extends Tensor & TNumber> Dilation2dBackpropFilter<T> create(Scope scope, Operand<T> input, Operand<T> filter, Operand<T> outBackprop, List<Long> strides, List<Long> rates, String padding) {
     OperationBuilder opBuilder = scope.env().opBuilder("Dilation2DBackpropFilter", scope.makeOpName("Dilation2dBackpropFilter"));
     opBuilder.addInput(input.asOutput());
     opBuilder.addInput(filter.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Dilation2dBackpropInput.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Dilation2dBackpropInput.java
index 525e06182c5..bb71ec8c62c 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Dilation2dBackpropInput.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Dilation2dBackpropInput.java
@@ -22,12 +22,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes the gradient of morphological 2-D dilation with respect to the input.
@@ -35,7 +35,7 @@
  * @param <T> data type for {@code inBackprop()} output
  */
 @Operator(group = "nn")
-public final class Dilation2dBackpropInput<T extends TNumber> extends RawOp implements Operand<T> {
+public final class Dilation2dBackpropInput<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Dilation2dBackpropInput operation.
@@ -52,7 +52,7 @@ public final class Dilation2dBackpropInput<T extends TNumber> extends RawOp impl
    * @return a new instance of Dilation2dBackpropInput
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> Dilation2dBackpropInput<T> create(Scope scope, Operand<T> input, Operand<T> filter, Operand<T> outBackprop, List<Long> strides, List<Long> rates, String padding) {
+  public static <T extends Tensor & TNumber> Dilation2dBackpropInput<T> create(Scope scope, Operand<T> input, Operand<T> filter, Operand<T> outBackprop, List<Long> strides, List<Long> rates, String padding) {
     OperationBuilder opBuilder = scope.env().opBuilder("Dilation2DBackpropInput", scope.makeOpName("Dilation2dBackpropInput"));
     opBuilder.addInput(input.asOutput());
     opBuilder.addInput(filter.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Elu.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Elu.java
index daeef97895b..47259f7e42b 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Elu.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Elu.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes exponential linear: `exp(features) - 1` if < 0, `features` otherwise.
@@ -37,7 +37,7 @@
  * @param <T> data type for {@code activations()} output
  */
 @Operator(group = "nn")
-public final class Elu<T extends TNumber> extends RawOp implements Operand<T> {
+public final class Elu<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Elu operation.
@@ -47,7 +47,7 @@ public final class Elu<T extends TNumber> extends RawOp implements Operand<T> {
    * @return a new instance of Elu
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> Elu<T> create(Scope scope, Operand<T> features) {
+  public static <T extends Tensor & TNumber> Elu<T> create(Scope scope, Operand<T> features) {
     OperationBuilder opBuilder = scope.env().opBuilder("Elu", scope.makeOpName("Elu"));
     opBuilder.addInput(features.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/EluGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/EluGrad.java
index 664475879a0..3c058eae14e 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/EluGrad.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/EluGrad.java
@@ -21,19 +21,19 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes gradients for the exponential linear (Elu) operation.
  * 
  * @param <T> data type for {@code backprops()} output
  */
-public final class EluGrad<T extends TNumber> extends RawOp implements Operand<T> {
+public final class EluGrad<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new EluGrad operation.
@@ -44,7 +44,7 @@ public final class EluGrad<T extends TNumber> extends RawOp implements Operand<T
    * @return a new instance of EluGrad
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> EluGrad<T> create(Scope scope, Operand<T> gradients, Operand<T> outputs) {
+  public static <T extends Tensor & TNumber> EluGrad<T> create(Scope scope, Operand<T> gradients, Operand<T> outputs) {
     OperationBuilder opBuilder = scope.env().opBuilder("EluGrad", scope.makeOpName("EluGrad"));
     opBuilder.addInput(gradients.asOutput());
     opBuilder.addInput(outputs.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/FractionalAvgPool.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/FractionalAvgPool.java
index 101ea21ec8c..2dab6ebd063 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/FractionalAvgPool.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/FractionalAvgPool.java
@@ -22,13 +22,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Performs fractional average pooling on the input.
@@ -41,7 +41,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "nn")
-public final class FractionalAvgPool<T extends TNumber> extends RawOp {
+public final class FractionalAvgPool<T extends Tensor & TNumber> extends RawOp {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.nn.FractionalAvgPool}
@@ -128,7 +128,7 @@ private Options() {
    * @return a new instance of FractionalAvgPool
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> FractionalAvgPool<T> create(Scope scope, Operand<T> value, List<Float> poolingRatio, Options... options) {
+  public static <T extends Tensor & TNumber> FractionalAvgPool<T> create(Scope scope, Operand<T> value, List<Float> poolingRatio, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("FractionalAvgPool", scope.makeOpName("FractionalAvgPool"));
     opBuilder.addInput(value.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/FractionalAvgPoolGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/FractionalAvgPoolGrad.java
index 03ae0136311..5b12f66407e 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/FractionalAvgPoolGrad.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/FractionalAvgPoolGrad.java
@@ -21,13 +21,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes gradient of the FractionalAvgPool function.
@@ -40,7 +40,7 @@
  * 
  * @param <T> data type for {@code output()} output
  */
-public final class FractionalAvgPoolGrad<T extends TNumber> extends RawOp implements Operand<T> {
+public final class FractionalAvgPoolGrad<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.nn.FractionalAvgPoolGrad}
@@ -84,7 +84,7 @@ private Options() {
    * @return a new instance of FractionalAvgPoolGrad
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> FractionalAvgPoolGrad<T> create(Scope scope, Operand<TInt64> origInputTensorShape, Operand<T> outBackprop, Operand<TInt64> rowPoolingSequence, Operand<TInt64> colPoolingSequence, Options... options) {
+  public static <T extends Tensor & TNumber> FractionalAvgPoolGrad<T> create(Scope scope, Operand<TInt64> origInputTensorShape, Operand<T> outBackprop, Operand<TInt64> rowPoolingSequence, Operand<TInt64> colPoolingSequence, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("FractionalAvgPoolGrad", scope.makeOpName("FractionalAvgPoolGrad"));
     opBuilder.addInput(origInputTensorShape.asOutput());
     opBuilder.addInput(outBackprop.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/FractionalMaxPool.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/FractionalMaxPool.java
index a621e037740..2519445ae87 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/FractionalMaxPool.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/FractionalMaxPool.java
@@ -22,13 +22,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Performs fractional max pooling on the input.
@@ -65,7 +65,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "nn")
-public final class FractionalMaxPool<T extends TNumber> extends RawOp {
+public final class FractionalMaxPool<T extends Tensor & TNumber> extends RawOp {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.nn.FractionalMaxPool}
@@ -152,7 +152,7 @@ private Options() {
    * @return a new instance of FractionalMaxPool
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> FractionalMaxPool<T> create(Scope scope, Operand<T> value, List<Float> poolingRatio, Options... options) {
+  public static <T extends Tensor & TNumber> FractionalMaxPool<T> create(Scope scope, Operand<T> value, List<Float> poolingRatio, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("FractionalMaxPool", scope.makeOpName("FractionalMaxPool"));
     opBuilder.addInput(value.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/FractionalMaxPoolGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/FractionalMaxPoolGrad.java
index 1b2bcc62dbf..ff3c0159747 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/FractionalMaxPoolGrad.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/FractionalMaxPoolGrad.java
@@ -21,20 +21,20 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes gradient of the FractionalMaxPool function.
  * 
  * @param <T> data type for {@code output()} output
  */
-public final class FractionalMaxPoolGrad<T extends TNumber> extends RawOp implements Operand<T> {
+public final class FractionalMaxPoolGrad<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.nn.FractionalMaxPoolGrad}
@@ -79,7 +79,7 @@ private Options() {
    * @return a new instance of FractionalMaxPoolGrad
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> FractionalMaxPoolGrad<T> create(Scope scope, Operand<T> origInput, Operand<T> origOutput, Operand<T> outBackprop, Operand<TInt64> rowPoolingSequence, Operand<TInt64> colPoolingSequence, Options... options) {
+  public static <T extends Tensor & TNumber> FractionalMaxPoolGrad<T> create(Scope scope, Operand<T> origInput, Operand<T> origOutput, Operand<T> outBackprop, Operand<TInt64> rowPoolingSequence, Operand<TInt64> colPoolingSequence, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("FractionalMaxPoolGrad", scope.makeOpName("FractionalMaxPoolGrad"));
     opBuilder.addInput(origInput.asOutput());
     opBuilder.addInput(origOutput.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/FusedBatchNorm.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/FusedBatchNorm.java
index 04e8d71e1e5..9d7ff535c82 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/FusedBatchNorm.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/FusedBatchNorm.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Batch normalization.
@@ -38,7 +38,7 @@
  * @param <U> data type for {@code batchMean()} output
  */
 @Operator(group = "nn")
-public final class FusedBatchNorm<T extends TNumber, U extends TNumber> extends RawOp {
+public final class FusedBatchNorm<T extends Tensor & TNumber, U extends Tensor & TNumber> extends RawOp {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.nn.FusedBatchNorm}
@@ -102,7 +102,7 @@ private Options() {
    * @return a new instance of FusedBatchNorm
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber, U extends TNumber> FusedBatchNorm<T, U> create(Scope scope, Operand<T> x, Operand<U> scale, Operand<U> offset, Operand<U> mean, Operand<U> variance, Options... options) {
+  public static <T extends Tensor & TNumber, U extends Tensor & TNumber> FusedBatchNorm<T, U> create(Scope scope, Operand<T> x, Operand<U> scale, Operand<U> offset, Operand<U> mean, Operand<U> variance, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("FusedBatchNormV3", scope.makeOpName("FusedBatchNorm"));
     opBuilder.addInput(x.asOutput());
     opBuilder.addInput(scale.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/FusedBatchNormGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/FusedBatchNormGrad.java
index 4d2ecc74a4f..9059d890c06 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/FusedBatchNormGrad.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/FusedBatchNormGrad.java
@@ -21,13 +21,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TFloat32;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Gradient for batch normalization.
@@ -39,7 +39,7 @@
  * @param <U> data type for {@code scaleBackprop()} output
  */
 @Operator(group = "nn")
-public final class FusedBatchNormGrad<T extends TNumber, U extends TNumber> extends RawOp {
+public final class FusedBatchNormGrad<T extends Tensor & TNumber, U extends Tensor & TNumber> extends RawOp {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.nn.FusedBatchNormGrad}
@@ -103,7 +103,7 @@ private Options() {
    * @return a new instance of FusedBatchNormGrad
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber, U extends TNumber> FusedBatchNormGrad<T, U> create(Scope scope, Operand<T> yBackprop, Operand<T> x, Operand<TFloat32> scale, Operand<U> reserveSpace1, Operand<U> reserveSpace2, Operand<U> reserveSpace3, Options... options) {
+  public static <T extends Tensor & TNumber, U extends Tensor & TNumber> FusedBatchNormGrad<T, U> create(Scope scope, Operand<T> yBackprop, Operand<T> x, Operand<TFloat32> scale, Operand<U> reserveSpace1, Operand<U> reserveSpace2, Operand<U> reserveSpace3, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("FusedBatchNormGradV3", scope.makeOpName("FusedBatchNormGrad"));
     opBuilder.addInput(yBackprop.asOutput());
     opBuilder.addInput(x.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/FusedPadConv2d.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/FusedPadConv2d.java
index 2d3de921f0b..0d2d7367762 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/FusedPadConv2d.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/FusedPadConv2d.java
@@ -22,13 +22,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Performs a padding as a preprocess during a convolution.
@@ -48,7 +48,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "nn")
-public final class FusedPadConv2d<T extends TNumber> extends RawOp implements Operand<T> {
+public final class FusedPadConv2d<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new FusedPadConv2d operation.
@@ -66,7 +66,7 @@ public final class FusedPadConv2d<T extends TNumber> extends RawOp implements Op
    * @return a new instance of FusedPadConv2d
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> FusedPadConv2d<T> create(Scope scope, Operand<T> input, Operand<TInt32> paddings, Operand<T> filter, String mode, List<Long> strides, String padding) {
+  public static <T extends Tensor & TNumber> FusedPadConv2d<T> create(Scope scope, Operand<T> input, Operand<TInt32> paddings, Operand<T> filter, String mode, List<Long> strides, String padding) {
     OperationBuilder opBuilder = scope.env().opBuilder("FusedPadConv2D", scope.makeOpName("FusedPadConv2d"));
     opBuilder.addInput(input.asOutput());
     opBuilder.addInput(paddings.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/FusedResizeAndPadConv2d.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/FusedResizeAndPadConv2d.java
index 02f82242c7a..cd4dbc00464 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/FusedResizeAndPadConv2d.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/FusedResizeAndPadConv2d.java
@@ -22,13 +22,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Performs a resize and padding as a preprocess during a convolution.
@@ -47,7 +47,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "nn")
-public final class FusedResizeAndPadConv2d<T extends TNumber> extends RawOp implements Operand<T> {
+public final class FusedResizeAndPadConv2d<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.nn.FusedResizeAndPadConv2d}
@@ -88,7 +88,7 @@ private Options() {
    * @return a new instance of FusedResizeAndPadConv2d
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> FusedResizeAndPadConv2d<T> create(Scope scope, Operand<T> input, Operand<TInt32> size, Operand<TInt32> paddings, Operand<T> filter, String mode, List<Long> strides, String padding, Options... options) {
+  public static <T extends Tensor & TNumber> FusedResizeAndPadConv2d<T> create(Scope scope, Operand<T> input, Operand<TInt32> size, Operand<TInt32> paddings, Operand<T> filter, String mode, List<Long> strides, String padding, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("FusedResizeAndPadConv2D", scope.makeOpName("FusedResizeAndPadConv2d"));
     opBuilder.addInput(input.asOutput());
     opBuilder.addInput(size.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/GRUBlockCell.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/GRUBlockCell.java
index 446f43cfb0c..481916e5afe 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/GRUBlockCell.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/GRUBlockCell.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes the GRU cell forward propagation for 1 time step.
@@ -78,7 +78,7 @@
  * 
  * @param <T> data type for {@code r()} output
  */
-public final class GRUBlockCell<T extends TNumber> extends RawOp {
+public final class GRUBlockCell<T extends Tensor & TNumber> extends RawOp {
   
   /**
    * Factory method to create a class wrapping a new GRUBlockCell operation.
@@ -93,7 +93,7 @@ public final class GRUBlockCell<T extends TNumber> extends RawOp {
    * @return a new instance of GRUBlockCell
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> GRUBlockCell<T> create(Scope scope, Operand<T> x, Operand<T> hPrev, Operand<T> wRu, Operand<T> wC, Operand<T> bRu, Operand<T> bC) {
+  public static <T extends Tensor & TNumber> GRUBlockCell<T> create(Scope scope, Operand<T> x, Operand<T> hPrev, Operand<T> wRu, Operand<T> wC, Operand<T> bRu, Operand<T> bC) {
     OperationBuilder opBuilder = scope.env().opBuilder("GRUBlockCell", scope.makeOpName("GRUBlockCell"));
     opBuilder.addInput(x.asOutput());
     opBuilder.addInput(hPrev.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/GRUBlockCellGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/GRUBlockCellGrad.java
index 4a6b72b8e4c..274cdc6772b 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/GRUBlockCellGrad.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/GRUBlockCellGrad.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes the GRU cell back-propagation for 1 time step.
@@ -114,7 +114,7 @@
  * 
  * @param <T> data type for {@code dX()} output
  */
-public final class GRUBlockCellGrad<T extends TNumber> extends RawOp {
+public final class GRUBlockCellGrad<T extends Tensor & TNumber> extends RawOp {
   
   /**
    * Factory method to create a class wrapping a new GRUBlockCellGrad operation.
@@ -133,7 +133,7 @@ public final class GRUBlockCellGrad<T extends TNumber> extends RawOp {
    * @return a new instance of GRUBlockCellGrad
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> GRUBlockCellGrad<T> create(Scope scope, Operand<T> x, Operand<T> hPrev, Operand<T> wRu, Operand<T> wC, Operand<T> bRu, Operand<T> bC, Operand<T> r, Operand<T> u, Operand<T> c, Operand<T> dH) {
+  public static <T extends Tensor & TNumber> GRUBlockCellGrad<T> create(Scope scope, Operand<T> x, Operand<T> hPrev, Operand<T> wRu, Operand<T> wC, Operand<T> bRu, Operand<T> bC, Operand<T> r, Operand<T> u, Operand<T> c, Operand<T> dH) {
     OperationBuilder opBuilder = scope.env().opBuilder("GRUBlockCellGrad", scope.makeOpName("GRUBlockCellGrad"));
     opBuilder.addInput(x.asOutput());
     opBuilder.addInput(hPrev.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/InTopK.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/InTopK.java
index d38388a3fd5..869315541c9 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/InTopK.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/InTopK.java
@@ -21,6 +21,7 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -28,7 +29,6 @@
 import org.tensorflow.types.TBool;
 import org.tensorflow.types.TFloat32;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Says whether the targets are in the top `K` predictions.
@@ -61,7 +61,7 @@ public final class InTopK extends RawOp implements Operand<TBool> {
    * @return a new instance of InTopK
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> InTopK create(Scope scope, Operand<TFloat32> predictions, Operand<T> targets, Operand<T> k) {
+  public static <T extends Tensor & TNumber> InTopK create(Scope scope, Operand<TFloat32> predictions, Operand<T> targets, Operand<T> k) {
     OperationBuilder opBuilder = scope.env().opBuilder("InTopKV2", scope.makeOpName("InTopK"));
     opBuilder.addInput(predictions.asOutput());
     opBuilder.addInput(targets.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/InvGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/InvGrad.java
index 161d8771b83..c582161f771 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/InvGrad.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/InvGrad.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes the gradient for the inverse of `x` wrt its input.
@@ -35,7 +35,7 @@
  * 
  * @param <T> data type for {@code z()} output
  */
-public final class InvGrad<T extends TType> extends RawOp implements Operand<T> {
+public final class InvGrad<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new InvGrad operation.
@@ -46,7 +46,7 @@ public final class InvGrad<T extends TType> extends RawOp implements Operand<T>
    * @return a new instance of InvGrad
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> InvGrad<T> create(Scope scope, Operand<T> y, Operand<T> dy) {
+  public static <T extends Tensor> InvGrad<T> create(Scope scope, Operand<T> y, Operand<T> dy) {
     OperationBuilder opBuilder = scope.env().opBuilder("InvGrad", scope.makeOpName("InvGrad"));
     opBuilder.addInput(y.asOutput());
     opBuilder.addInput(dy.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/L2Loss.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/L2Loss.java
index b38d9e99d96..02f1771815c 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/L2Loss.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/L2Loss.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * L2 Loss.
@@ -38,7 +38,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "nn")
-public final class L2Loss<T extends TNumber> extends RawOp implements Operand<T> {
+public final class L2Loss<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new L2Loss operation.
@@ -48,7 +48,7 @@ public final class L2Loss<T extends TNumber> extends RawOp implements Operand<T>
    * @return a new instance of L2Loss
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> L2Loss<T> create(Scope scope, Operand<T> t) {
+  public static <T extends Tensor & TNumber> L2Loss<T> create(Scope scope, Operand<T> t) {
     OperationBuilder opBuilder = scope.env().opBuilder("L2Loss", scope.makeOpName("L2Loss"));
     opBuilder.addInput(t.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/LSTMBlockCell.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/LSTMBlockCell.java
index ab5d1b8aee2..ac4cc8ecffc 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/LSTMBlockCell.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/LSTMBlockCell.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes the LSTM cell forward propagation for 1 time step.
@@ -58,7 +58,7 @@
  * 
  * @param <T> data type for {@code i()} output
  */
-public final class LSTMBlockCell<T extends TNumber> extends RawOp {
+public final class LSTMBlockCell<T extends Tensor & TNumber> extends RawOp {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.nn.LSTMBlockCell}
@@ -113,7 +113,7 @@ private Options() {
    * @return a new instance of LSTMBlockCell
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> LSTMBlockCell<T> create(Scope scope, Operand<T> x, Operand<T> csPrev, Operand<T> hPrev, Operand<T> w, Operand<T> wci, Operand<T> wcf, Operand<T> wco, Operand<T> b, Options... options) {
+  public static <T extends Tensor & TNumber> LSTMBlockCell<T> create(Scope scope, Operand<T> x, Operand<T> csPrev, Operand<T> hPrev, Operand<T> w, Operand<T> wci, Operand<T> wcf, Operand<T> wco, Operand<T> b, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("LSTMBlockCell", scope.makeOpName("LSTMBlockCell"));
     opBuilder.addInput(x.asOutput());
     opBuilder.addInput(csPrev.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/LSTMBlockCellGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/LSTMBlockCellGrad.java
index 3635ec433d6..20541ebe8e8 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/LSTMBlockCellGrad.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/LSTMBlockCellGrad.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes the LSTM cell backward propagation for 1 timestep.
@@ -35,7 +35,7 @@
  * 
  * @param <T> data type for {@code csPrevGrad()} output
  */
-public final class LSTMBlockCellGrad<T extends TNumber> extends RawOp {
+public final class LSTMBlockCellGrad<T extends Tensor & TNumber> extends RawOp {
   
   /**
    * Factory method to create a class wrapping a new LSTMBlockCellGrad operation.
@@ -61,7 +61,7 @@ public final class LSTMBlockCellGrad<T extends TNumber> extends RawOp {
    * @return a new instance of LSTMBlockCellGrad
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> LSTMBlockCellGrad<T> create(Scope scope, Operand<T> x, Operand<T> csPrev, Operand<T> hPrev, Operand<T> w, Operand<T> wci, Operand<T> wcf, Operand<T> wco, Operand<T> b, Operand<T> i, Operand<T> cs, Operand<T> f, Operand<T> o, Operand<T> ci, Operand<T> co, Operand<T> csGrad, Operand<T> hGrad, Boolean usePeephole) {
+  public static <T extends Tensor & TNumber> LSTMBlockCellGrad<T> create(Scope scope, Operand<T> x, Operand<T> csPrev, Operand<T> hPrev, Operand<T> w, Operand<T> wci, Operand<T> wcf, Operand<T> wco, Operand<T> b, Operand<T> i, Operand<T> cs, Operand<T> f, Operand<T> o, Operand<T> ci, Operand<T> co, Operand<T> csGrad, Operand<T> hGrad, Boolean usePeephole) {
     OperationBuilder opBuilder = scope.env().opBuilder("LSTMBlockCellGrad", scope.makeOpName("LSTMBlockCellGrad"));
     opBuilder.addInput(x.asOutput());
     opBuilder.addInput(csPrev.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/LeakyRelu.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/LeakyRelu.java
index 8ca3f540cad..c64340ab6b4 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/LeakyRelu.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/LeakyRelu.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes rectified linear: `max(features, features * alpha)`.
@@ -34,7 +34,7 @@
  * @param <T> data type for {@code activations()} output
  */
 @Operator(group = "nn")
-public final class LeakyRelu<T extends TNumber> extends RawOp implements Operand<T> {
+public final class LeakyRelu<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.nn.LeakyRelu}
@@ -64,7 +64,7 @@ private Options() {
    * @return a new instance of LeakyRelu
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> LeakyRelu<T> create(Scope scope, Operand<T> features, Options... options) {
+  public static <T extends Tensor & TNumber> LeakyRelu<T> create(Scope scope, Operand<T> features, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("LeakyRelu", scope.makeOpName("LeakyRelu"));
     opBuilder.addInput(features.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/LocalResponseNormalization.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/LocalResponseNormalization.java
index a560c9cfb58..d04ddea39ce 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/LocalResponseNormalization.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/LocalResponseNormalization.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Local Response Normalization.
@@ -46,7 +46,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "nn")
-public final class LocalResponseNormalization<T extends TNumber> extends RawOp implements Operand<T> {
+public final class LocalResponseNormalization<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.nn.LocalResponseNormalization}
@@ -103,7 +103,7 @@ private Options() {
    * @return a new instance of LocalResponseNormalization
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> LocalResponseNormalization<T> create(Scope scope, Operand<T> input, Options... options) {
+  public static <T extends Tensor & TNumber> LocalResponseNormalization<T> create(Scope scope, Operand<T> input, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("LRN", scope.makeOpName("LocalResponseNormalization"));
     opBuilder.addInput(input.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/LocalResponseNormalizationGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/LocalResponseNormalizationGrad.java
index 08f10c12f26..fd9fd870829 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/LocalResponseNormalizationGrad.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/LocalResponseNormalizationGrad.java
@@ -21,19 +21,19 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Gradients for Local Response Normalization.
  * 
  * @param <T> data type for {@code output()} output
  */
-public final class LocalResponseNormalizationGrad<T extends TNumber> extends RawOp implements Operand<T> {
+public final class LocalResponseNormalizationGrad<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.nn.LocalResponseNormalizationGrad}
@@ -92,7 +92,7 @@ private Options() {
    * @return a new instance of LocalResponseNormalizationGrad
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> LocalResponseNormalizationGrad<T> create(Scope scope, Operand<T> inputGrads, Operand<T> inputImage, Operand<T> outputImage, Options... options) {
+  public static <T extends Tensor & TNumber> LocalResponseNormalizationGrad<T> create(Scope scope, Operand<T> inputGrads, Operand<T> inputImage, Operand<T> outputImage, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("LRNGrad", scope.makeOpName("LocalResponseNormalizationGrad"));
     opBuilder.addInput(inputGrads.asOutput());
     opBuilder.addInput(inputImage.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/LogSoftmax.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/LogSoftmax.java
index bd1bff467a1..80d8a2346bc 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/LogSoftmax.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/LogSoftmax.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes log softmax activations.
@@ -38,7 +38,7 @@
  * @param <T> data type for {@code logsoftmax()} output
  */
 @Operator(group = "nn")
-public final class LogSoftmax<T extends TNumber> extends RawOp implements Operand<T> {
+public final class LogSoftmax<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new LogSoftmax operation.
@@ -48,7 +48,7 @@ public final class LogSoftmax<T extends TNumber> extends RawOp implements Operan
    * @return a new instance of LogSoftmax
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> LogSoftmax<T> create(Scope scope, Operand<T> logits) {
+  public static <T extends Tensor & TNumber> LogSoftmax<T> create(Scope scope, Operand<T> logits) {
     OperationBuilder opBuilder = scope.env().opBuilder("LogSoftmax", scope.makeOpName("LogSoftmax"));
     opBuilder.addInput(logits.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPool.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPool.java
index f7b75951e31..90e952e747e 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPool.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPool.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
-import org.tensorflow.types.family.TType;
 
 /**
  * Performs max pooling on the input.
@@ -34,7 +34,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "nn")
-public final class MaxPool<T extends TType> extends RawOp implements Operand<T> {
+public final class MaxPool<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.nn.MaxPool}
@@ -72,7 +72,7 @@ private Options() {
    * @return a new instance of MaxPool
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> MaxPool<T> create(Scope scope, Operand<T> input, Operand<TInt32> ksize, Operand<TInt32> strides, String padding, Options... options) {
+  public static <T extends Tensor> MaxPool<T> create(Scope scope, Operand<T> input, Operand<TInt32> ksize, Operand<TInt32> strides, String padding, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("MaxPoolV2", scope.makeOpName("MaxPool"));
     opBuilder.addInput(input.asOutput());
     opBuilder.addInput(ksize.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPool3d.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPool3d.java
index 61bf43372cc..b6ac014641f 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPool3d.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPool3d.java
@@ -22,12 +22,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Performs 3D max pooling on the input.
@@ -35,7 +35,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "nn")
-public final class MaxPool3d<T extends TNumber> extends RawOp implements Operand<T> {
+public final class MaxPool3d<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.nn.MaxPool3d}
@@ -74,7 +74,7 @@ private Options() {
    * @return a new instance of MaxPool3d
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> MaxPool3d<T> create(Scope scope, Operand<T> input, List<Long> ksize, List<Long> strides, String padding, Options... options) {
+  public static <T extends Tensor & TNumber> MaxPool3d<T> create(Scope scope, Operand<T> input, List<Long> ksize, List<Long> strides, String padding, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("MaxPool3D", scope.makeOpName("MaxPool3d"));
     opBuilder.addInput(input.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPool3dGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPool3dGrad.java
index 2a4fe5e61f9..44d2cb4b276 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPool3dGrad.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPool3dGrad.java
@@ -22,12 +22,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes gradients of 3D max pooling function.
@@ -35,7 +35,7 @@
  * @param <U> data type for {@code output()} output
  */
 @Operator(group = "nn")
-public final class MaxPool3dGrad<U extends TNumber> extends RawOp implements Operand<U> {
+public final class MaxPool3dGrad<U extends Tensor & TNumber> extends RawOp implements Operand<U> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.nn.MaxPool3dGrad}
@@ -76,7 +76,7 @@ private Options() {
    * @return a new instance of MaxPool3dGrad
    */
   @Endpoint(describeByClass = true)
-  public static <U extends TNumber, T extends TNumber> MaxPool3dGrad<U> create(Scope scope, Operand<T> origInput, Operand<T> origOutput, Operand<U> grad, List<Long> ksize, List<Long> strides, String padding, Options... options) {
+  public static <U extends Tensor & TNumber, T extends Tensor & TNumber> MaxPool3dGrad<U> create(Scope scope, Operand<T> origInput, Operand<T> origOutput, Operand<U> grad, List<Long> ksize, List<Long> strides, String padding, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("MaxPool3DGrad", scope.makeOpName("MaxPool3dGrad"));
     opBuilder.addInput(origInput.asOutput());
     opBuilder.addInput(origOutput.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPool3dGradGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPool3dGradGrad.java
index 4243e1e5143..c17d23daa27 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPool3dGradGrad.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPool3dGradGrad.java
@@ -22,12 +22,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes second-order gradients of the maxpooling function.
@@ -35,7 +35,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "nn")
-public final class MaxPool3dGradGrad<T extends TNumber> extends RawOp implements Operand<T> {
+public final class MaxPool3dGradGrad<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.nn.MaxPool3dGradGrad}
@@ -76,7 +76,7 @@ private Options() {
    * @return a new instance of MaxPool3dGradGrad
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> MaxPool3dGradGrad<T> create(Scope scope, Operand<T> origInput, Operand<T> origOutput, Operand<T> grad, List<Long> ksize, List<Long> strides, String padding, Options... options) {
+  public static <T extends Tensor & TNumber> MaxPool3dGradGrad<T> create(Scope scope, Operand<T> origInput, Operand<T> origOutput, Operand<T> grad, List<Long> ksize, List<Long> strides, String padding, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("MaxPool3DGradGrad", scope.makeOpName("MaxPool3dGradGrad"));
     opBuilder.addInput(origInput.asOutput());
     opBuilder.addInput(origOutput.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPoolGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPoolGrad.java
index 6f5cbd2ce64..beea4fcdee5 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPoolGrad.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPoolGrad.java
@@ -21,13 +21,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes gradients of the maxpooling function.
@@ -35,7 +35,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "nn")
-public final class MaxPoolGrad<T extends TNumber> extends RawOp implements Operand<T> {
+public final class MaxPoolGrad<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.nn.MaxPoolGrad}
@@ -75,7 +75,7 @@ private Options() {
    * @return a new instance of MaxPoolGrad
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> MaxPoolGrad<T> create(Scope scope, Operand<T> origInput, Operand<T> origOutput, Operand<T> grad, Operand<TInt32> ksize, Operand<TInt32> strides, String padding, Options... options) {
+  public static <T extends Tensor & TNumber> MaxPoolGrad<T> create(Scope scope, Operand<T> origInput, Operand<T> origOutput, Operand<T> grad, Operand<TInt32> ksize, Operand<TInt32> strides, String padding, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("MaxPoolGradV2", scope.makeOpName("MaxPoolGrad"));
     opBuilder.addInput(origInput.asOutput());
     opBuilder.addInput(origOutput.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPoolGradGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPoolGradGrad.java
index 4f9255e9286..93a2b142125 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPoolGradGrad.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPoolGradGrad.java
@@ -21,13 +21,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes second-order gradients of the maxpooling function.
@@ -35,7 +35,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "nn")
-public final class MaxPoolGradGrad<T extends TNumber> extends RawOp implements Operand<T> {
+public final class MaxPoolGradGrad<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.nn.MaxPoolGradGrad}
@@ -75,7 +75,7 @@ private Options() {
    * @return a new instance of MaxPoolGradGrad
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> MaxPoolGradGrad<T> create(Scope scope, Operand<T> origInput, Operand<T> origOutput, Operand<T> grad, Operand<TInt32> ksize, Operand<TInt32> strides, String padding, Options... options) {
+  public static <T extends Tensor & TNumber> MaxPoolGradGrad<T> create(Scope scope, Operand<T> origInput, Operand<T> origOutput, Operand<T> grad, Operand<TInt32> ksize, Operand<TInt32> strides, String padding, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("MaxPoolGradGradV2", scope.makeOpName("MaxPoolGradGrad"));
     opBuilder.addInput(origInput.asOutput());
     opBuilder.addInput(origOutput.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPoolGradGradWithArgmax.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPoolGradGradWithArgmax.java
index 6258024daa9..d0a032189c4 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPoolGradGradWithArgmax.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPoolGradGradWithArgmax.java
@@ -22,12 +22,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes second-order gradients of the maxpooling function.
@@ -35,7 +35,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "nn")
-public final class MaxPoolGradGradWithArgmax<T extends TNumber> extends RawOp implements Operand<T> {
+public final class MaxPoolGradGradWithArgmax<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.nn.MaxPoolGradGradWithArgmax}
@@ -72,7 +72,7 @@ private Options() {
    * @return a new instance of MaxPoolGradGradWithArgmax
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber, U extends TNumber> MaxPoolGradGradWithArgmax<T> create(Scope scope, Operand<T> input, Operand<T> grad, Operand<U> argmax, List<Long> ksize, List<Long> strides, String padding, Options... options) {
+  public static <T extends Tensor & TNumber, U extends Tensor & TNumber> MaxPoolGradGradWithArgmax<T> create(Scope scope, Operand<T> input, Operand<T> grad, Operand<U> argmax, List<Long> ksize, List<Long> strides, String padding, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("MaxPoolGradGradWithArgmax", scope.makeOpName("MaxPoolGradGradWithArgmax"));
     opBuilder.addInput(input.asOutput());
     opBuilder.addInput(grad.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPoolGradWithArgmax.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPoolGradWithArgmax.java
index 329c88b4d33..0235d171d74 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPoolGradWithArgmax.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPoolGradWithArgmax.java
@@ -22,19 +22,19 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes gradients of the maxpooling function.
  * 
  * @param <T> data type for {@code output()} output
  */
-public final class MaxPoolGradWithArgmax<T extends TNumber> extends RawOp implements Operand<T> {
+public final class MaxPoolGradWithArgmax<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.nn.MaxPoolGradWithArgmax}
@@ -71,7 +71,7 @@ private Options() {
    * @return a new instance of MaxPoolGradWithArgmax
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber, U extends TNumber> MaxPoolGradWithArgmax<T> create(Scope scope, Operand<T> input, Operand<T> grad, Operand<U> argmax, List<Long> ksize, List<Long> strides, String padding, Options... options) {
+  public static <T extends Tensor & TNumber, U extends Tensor & TNumber> MaxPoolGradWithArgmax<T> create(Scope scope, Operand<T> input, Operand<T> grad, Operand<U> argmax, List<Long> ksize, List<Long> strides, String padding, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("MaxPoolGradWithArgmax", scope.makeOpName("MaxPoolGradWithArgmax"));
     opBuilder.addInput(input.asOutput());
     opBuilder.addInput(grad.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPoolWithArgmax.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPoolWithArgmax.java
index 63f92de1fea..901fbf451df 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPoolWithArgmax.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPoolWithArgmax.java
@@ -23,13 +23,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Performs max pooling on the input and outputs both max values and indices.
@@ -48,7 +48,7 @@
  * @param <U> data type for {@code argmax()} output
  */
 @Operator(group = "nn")
-public final class MaxPoolWithArgmax<T extends TNumber, U extends TNumber> extends RawOp {
+public final class MaxPoolWithArgmax<T extends Tensor & TNumber, U extends Tensor & TNumber> extends RawOp {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.nn.MaxPoolWithArgmax}
@@ -83,7 +83,7 @@ private Options() {
    * @return a new instance of MaxPoolWithArgmax
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber, U extends TNumber> MaxPoolWithArgmax<T, U> create(Scope scope, Operand<T> input, List<Long> ksize, List<Long> strides, DataType<U> Targmax, String padding, Options... options) {
+  public static <T extends Tensor & TNumber, U extends Tensor & TNumber> MaxPoolWithArgmax<T, U> create(Scope scope, Operand<T> input, List<Long> ksize, List<Long> strides, DataType<U> Targmax, String padding, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("MaxPoolWithArgmax", scope.makeOpName("MaxPoolWithArgmax"));
     opBuilder.addInput(input.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
@@ -122,7 +122,7 @@ public static <T extends TNumber, U extends TNumber> MaxPoolWithArgmax<T, U> cre
    * @return a new instance of MaxPoolWithArgmax
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> MaxPoolWithArgmax<T, TInt64> create(Scope scope, Operand<T> input, List<Long> ksize, List<Long> strides, String padding, Options... options) {
+  public static <T extends Tensor & TNumber> MaxPoolWithArgmax<T, TInt64> create(Scope scope, Operand<T> input, List<Long> ksize, List<Long> strides, String padding, Options... options) {
     return create(scope, input, ksize, strides, TInt64.DTYPE, padding, options);
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/NthElement.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/NthElement.java
index 340cd4b98f7..8decbf0e7ae 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/NthElement.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/NthElement.java
@@ -21,13 +21,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Finds values of the `n`-th order statistic for the last dimension.
@@ -43,7 +43,7 @@
  * @param <T> data type for {@code values()} output
  */
 @Operator(group = "nn")
-public final class NthElement<T extends TNumber> extends RawOp implements Operand<T> {
+public final class NthElement<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.nn.NthElement}
@@ -76,7 +76,7 @@ private Options() {
    * @return a new instance of NthElement
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> NthElement<T> create(Scope scope, Operand<T> input, Operand<TInt32> n, Options... options) {
+  public static <T extends Tensor & TNumber> NthElement<T> create(Scope scope, Operand<T> input, Operand<TInt32> n, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("NthElement", scope.makeOpName("NthElement"));
     opBuilder.addInput(input.asOutput());
     opBuilder.addInput(n.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedAvgPool.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedAvgPool.java
index 0da9f10b1ef..b521985edda 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedAvgPool.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedAvgPool.java
@@ -22,12 +22,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.family.TType;
 
 /**
  * Produces the average pool of the input tensor for quantized types.
@@ -35,7 +35,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "nn")
-public final class QuantizedAvgPool<T extends TType> extends RawOp {
+public final class QuantizedAvgPool<T extends Tensor> extends RawOp {
   
   /**
    * Factory method to create a class wrapping a new QuantizedAvgPool operation.
@@ -52,7 +52,7 @@ public final class QuantizedAvgPool<T extends TType> extends RawOp {
    * @return a new instance of QuantizedAvgPool
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> QuantizedAvgPool<T> create(Scope scope, Operand<T> input, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, List<Long> ksize, List<Long> strides, String padding) {
+  public static <T extends Tensor> QuantizedAvgPool<T> create(Scope scope, Operand<T> input, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, List<Long> ksize, List<Long> strides, String padding) {
     OperationBuilder opBuilder = scope.env().opBuilder("QuantizedAvgPool", scope.makeOpName("QuantizedAvgPool"));
     opBuilder.addInput(input.asOutput());
     opBuilder.addInput(minInput.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedBatchNormWithGlobalNormalization.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedBatchNormWithGlobalNormalization.java
index 41cc16017fc..f2c9999482d 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedBatchNormWithGlobalNormalization.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedBatchNormWithGlobalNormalization.java
@@ -22,12 +22,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.family.TType;
 
 /**
  * Quantized Batch normalization.
@@ -38,7 +38,7 @@
  * @param <U> data type for {@code result()} output
  */
 @Operator(group = "nn")
-public final class QuantizedBatchNormWithGlobalNormalization<U extends TType> extends RawOp {
+public final class QuantizedBatchNormWithGlobalNormalization<U extends Tensor> extends RawOp {
   
   /**
    * Factory method to create a class wrapping a new QuantizedBatchNormWithGlobalNormalization operation.
@@ -73,7 +73,7 @@ public final class QuantizedBatchNormWithGlobalNormalization<U extends TType> ex
    * @return a new instance of QuantizedBatchNormWithGlobalNormalization
    */
   @Endpoint(describeByClass = true)
-  public static <U extends TType, T extends TType> QuantizedBatchNormWithGlobalNormalization<U> create(Scope scope, Operand<T> t, Operand<TFloat32> tMin, Operand<TFloat32> tMax, Operand<T> m, Operand<TFloat32> mMin, Operand<TFloat32> mMax, Operand<T> v, Operand<TFloat32> vMin, Operand<TFloat32> vMax, Operand<T> beta, Operand<TFloat32> betaMin, Operand<TFloat32> betaMax, Operand<T> gamma, Operand<TFloat32> gammaMin, Operand<TFloat32> gammaMax, DataType<U> outType, Float varianceEpsilon, Boolean scaleAfterNormalization) {
+  public static <U extends Tensor, T extends Tensor> QuantizedBatchNormWithGlobalNormalization<U> create(Scope scope, Operand<T> t, Operand<TFloat32> tMin, Operand<TFloat32> tMax, Operand<T> m, Operand<TFloat32> mMin, Operand<TFloat32> mMax, Operand<T> v, Operand<TFloat32> vMin, Operand<TFloat32> vMax, Operand<T> beta, Operand<TFloat32> betaMin, Operand<TFloat32> betaMax, Operand<T> gamma, Operand<TFloat32> gammaMin, Operand<TFloat32> gammaMax, DataType<U> outType, Float varianceEpsilon, Boolean scaleAfterNormalization) {
     OperationBuilder opBuilder = scope.env().opBuilder("QuantizedBatchNormWithGlobalNormalization", scope.makeOpName("QuantizedBatchNormWithGlobalNormalization"));
     opBuilder.addInput(t.asOutput());
     opBuilder.addInput(tMin.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedBiasAdd.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedBiasAdd.java
index 62a8002ad49..f8abac79370 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedBiasAdd.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedBiasAdd.java
@@ -22,12 +22,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.family.TType;
 
 /**
  * Adds Tensor 'bias' to Tensor 'input' for Quantized types.
@@ -37,7 +37,7 @@
  * @param <V> data type for {@code output()} output
  */
 @Operator(group = "nn")
-public final class QuantizedBiasAdd<V extends TType> extends RawOp {
+public final class QuantizedBiasAdd<V extends Tensor> extends RawOp {
   
   /**
    * Factory method to create a class wrapping a new QuantizedBiasAdd operation.
@@ -53,7 +53,7 @@ public final class QuantizedBiasAdd<V extends TType> extends RawOp {
    * @return a new instance of QuantizedBiasAdd
    */
   @Endpoint(describeByClass = true)
-  public static <V extends TType, T extends TType, U extends TType> QuantizedBiasAdd<V> create(Scope scope, Operand<T> input, Operand<U> bias, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, Operand<TFloat32> minBias, Operand<TFloat32> maxBias, DataType<V> outType) {
+  public static <V extends Tensor, T extends Tensor, U extends Tensor> QuantizedBiasAdd<V> create(Scope scope, Operand<T> input, Operand<U> bias, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, Operand<TFloat32> minBias, Operand<TFloat32> maxBias, DataType<V> outType) {
     OperationBuilder opBuilder = scope.env().opBuilder("QuantizedBiasAdd", scope.makeOpName("QuantizedBiasAdd"));
     opBuilder.addInput(input.asOutput());
     opBuilder.addInput(bias.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DAndRelu.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DAndRelu.java
index 961a4245155..73ecbe49450 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DAndRelu.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DAndRelu.java
@@ -23,17 +23,17 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.family.TType;
 
 /**
  * @param <V> data type for {@code output()} output
  */
-public final class QuantizedConv2DAndRelu<V extends TType> extends RawOp {
+public final class QuantizedConv2DAndRelu<V extends Tensor> extends RawOp {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.nn.QuantizedConv2DAndRelu}
@@ -80,7 +80,7 @@ private Options() {
    * @return a new instance of QuantizedConv2DAndRelu
    */
   @Endpoint(describeByClass = true)
-  public static <V extends TType, T extends TType, U extends TType> QuantizedConv2DAndRelu<V> create(Scope scope, Operand<T> input, Operand<U> filter, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, Operand<TFloat32> minFilter, Operand<TFloat32> maxFilter, DataType<V> outType, List<Long> strides, String padding, Options... options) {
+  public static <V extends Tensor, T extends Tensor, U extends Tensor> QuantizedConv2DAndRelu<V> create(Scope scope, Operand<T> input, Operand<U> filter, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, Operand<TFloat32> minFilter, Operand<TFloat32> maxFilter, DataType<V> outType, List<Long> strides, String padding, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("QuantizedConv2DAndRelu", scope.makeOpName("QuantizedConv2DAndRelu"));
     opBuilder.addInput(input.asOutput());
     opBuilder.addInput(filter.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DAndReluAndRequantize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DAndReluAndRequantize.java
index 8b231edc09b..6d935c91097 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DAndReluAndRequantize.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DAndReluAndRequantize.java
@@ -23,17 +23,17 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.family.TType;
 
 /**
  * @param <V> data type for {@code output()} output
  */
-public final class QuantizedConv2DAndReluAndRequantize<V extends TType> extends RawOp {
+public final class QuantizedConv2DAndReluAndRequantize<V extends Tensor> extends RawOp {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.nn.QuantizedConv2DAndReluAndRequantize}
@@ -82,7 +82,7 @@ private Options() {
    * @return a new instance of QuantizedConv2DAndReluAndRequantize
    */
   @Endpoint(describeByClass = true)
-  public static <V extends TType, T extends TType, U extends TType> QuantizedConv2DAndReluAndRequantize<V> create(Scope scope, Operand<T> input, Operand<U> filter, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, Operand<TFloat32> minFilter, Operand<TFloat32> maxFilter, Operand<TFloat32> minFreezedOutput, Operand<TFloat32> maxFreezedOutput, DataType<V> outType, List<Long> strides, String padding, Options... options) {
+  public static <V extends Tensor, T extends Tensor, U extends Tensor> QuantizedConv2DAndReluAndRequantize<V> create(Scope scope, Operand<T> input, Operand<U> filter, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, Operand<TFloat32> minFilter, Operand<TFloat32> maxFilter, Operand<TFloat32> minFreezedOutput, Operand<TFloat32> maxFreezedOutput, DataType<V> outType, List<Long> strides, String padding, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("QuantizedConv2DAndReluAndRequantize", scope.makeOpName("QuantizedConv2DAndReluAndRequantize"));
     opBuilder.addInput(input.asOutput());
     opBuilder.addInput(filter.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DAndRequantize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DAndRequantize.java
index 8c9f1247d97..5661c43e321 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DAndRequantize.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DAndRequantize.java
@@ -23,17 +23,17 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.family.TType;
 
 /**
  * @param <V> data type for {@code output()} output
  */
-public final class QuantizedConv2DAndRequantize<V extends TType> extends RawOp {
+public final class QuantizedConv2DAndRequantize<V extends Tensor> extends RawOp {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.nn.QuantizedConv2DAndRequantize}
@@ -82,7 +82,7 @@ private Options() {
    * @return a new instance of QuantizedConv2DAndRequantize
    */
   @Endpoint(describeByClass = true)
-  public static <V extends TType, T extends TType, U extends TType> QuantizedConv2DAndRequantize<V> create(Scope scope, Operand<T> input, Operand<U> filter, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, Operand<TFloat32> minFilter, Operand<TFloat32> maxFilter, Operand<TFloat32> minFreezedOutput, Operand<TFloat32> maxFreezedOutput, DataType<V> outType, List<Long> strides, String padding, Options... options) {
+  public static <V extends Tensor, T extends Tensor, U extends Tensor> QuantizedConv2DAndRequantize<V> create(Scope scope, Operand<T> input, Operand<U> filter, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, Operand<TFloat32> minFilter, Operand<TFloat32> maxFilter, Operand<TFloat32> minFreezedOutput, Operand<TFloat32> maxFreezedOutput, DataType<V> outType, List<Long> strides, String padding, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("QuantizedConv2DAndRequantize", scope.makeOpName("QuantizedConv2DAndRequantize"));
     opBuilder.addInput(input.asOutput());
     opBuilder.addInput(filter.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DPerChannel.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DPerChannel.java
index efc0132e2a8..112e1f640a1 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DPerChannel.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DPerChannel.java
@@ -23,19 +23,19 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes QuantizedConv2D per channel.
  * 
  * @param <V> data type for {@code output()} output
  */
-public final class QuantizedConv2DPerChannel<V extends TType> extends RawOp {
+public final class QuantizedConv2DPerChannel<V extends Tensor> extends RawOp {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.nn.QuantizedConv2DPerChannel}
@@ -73,7 +73,7 @@ private Options() {
    * @return a new instance of QuantizedConv2DPerChannel
    */
   @Endpoint(describeByClass = true)
-  public static <V extends TType, T extends TType, U extends TType> QuantizedConv2DPerChannel<V> create(Scope scope, Operand<T> input, Operand<U> filter, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, Operand<TFloat32> minFilter, Operand<TFloat32> maxFilter, DataType<V> outType, List<Long> strides, String padding, Options... options) {
+  public static <V extends Tensor, T extends Tensor, U extends Tensor> QuantizedConv2DPerChannel<V> create(Scope scope, Operand<T> input, Operand<U> filter, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, Operand<TFloat32> minFilter, Operand<TFloat32> maxFilter, DataType<V> outType, List<Long> strides, String padding, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("QuantizedConv2DPerChannel", scope.makeOpName("QuantizedConv2DPerChannel"));
     opBuilder.addInput(input.asOutput());
     opBuilder.addInput(filter.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DWithBias.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DWithBias.java
index bc78b5b4ec1..94649797e81 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DWithBias.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DWithBias.java
@@ -23,17 +23,17 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.family.TType;
 
 /**
  * @param <V> data type for {@code output()} output
  */
-public final class QuantizedConv2DWithBias<V extends TType> extends RawOp {
+public final class QuantizedConv2DWithBias<V extends Tensor> extends RawOp {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.nn.QuantizedConv2DWithBias}
@@ -81,7 +81,7 @@ private Options() {
    * @return a new instance of QuantizedConv2DWithBias
    */
   @Endpoint(describeByClass = true)
-  public static <V extends TType, T extends TType, U extends TType> QuantizedConv2DWithBias<V> create(Scope scope, Operand<T> input, Operand<U> filter, Operand<TFloat32> bias, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, Operand<TFloat32> minFilter, Operand<TFloat32> maxFilter, DataType<V> outType, List<Long> strides, String padding, Options... options) {
+  public static <V extends Tensor, T extends Tensor, U extends Tensor> QuantizedConv2DWithBias<V> create(Scope scope, Operand<T> input, Operand<U> filter, Operand<TFloat32> bias, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, Operand<TFloat32> minFilter, Operand<TFloat32> maxFilter, DataType<V> outType, List<Long> strides, String padding, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("QuantizedConv2DWithBias", scope.makeOpName("QuantizedConv2DWithBias"));
     opBuilder.addInput(input.asOutput());
     opBuilder.addInput(filter.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DWithBiasAndRelu.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DWithBiasAndRelu.java
index 719150ee139..d14eed83c3f 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DWithBiasAndRelu.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DWithBiasAndRelu.java
@@ -23,17 +23,17 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.family.TType;
 
 /**
  * @param <V> data type for {@code output()} output
  */
-public final class QuantizedConv2DWithBiasAndRelu<V extends TType> extends RawOp {
+public final class QuantizedConv2DWithBiasAndRelu<V extends Tensor> extends RawOp {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.nn.QuantizedConv2DWithBiasAndRelu}
@@ -81,7 +81,7 @@ private Options() {
    * @return a new instance of QuantizedConv2DWithBiasAndRelu
    */
   @Endpoint(describeByClass = true)
-  public static <V extends TType, T extends TType, U extends TType> QuantizedConv2DWithBiasAndRelu<V> create(Scope scope, Operand<T> input, Operand<U> filter, Operand<TFloat32> bias, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, Operand<TFloat32> minFilter, Operand<TFloat32> maxFilter, DataType<V> outType, List<Long> strides, String padding, Options... options) {
+  public static <V extends Tensor, T extends Tensor, U extends Tensor> QuantizedConv2DWithBiasAndRelu<V> create(Scope scope, Operand<T> input, Operand<U> filter, Operand<TFloat32> bias, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, Operand<TFloat32> minFilter, Operand<TFloat32> maxFilter, DataType<V> outType, List<Long> strides, String padding, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("QuantizedConv2DWithBiasAndRelu", scope.makeOpName("QuantizedConv2DWithBiasAndRelu"));
     opBuilder.addInput(input.asOutput());
     opBuilder.addInput(filter.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DWithBiasAndReluAndRequantize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DWithBiasAndReluAndRequantize.java
index a61cab41d5e..293c9595d47 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DWithBiasAndReluAndRequantize.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DWithBiasAndReluAndRequantize.java
@@ -23,17 +23,17 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.family.TType;
 
 /**
  * @param <W> data type for {@code output()} output
  */
-public final class QuantizedConv2DWithBiasAndReluAndRequantize<W extends TType> extends RawOp {
+public final class QuantizedConv2DWithBiasAndReluAndRequantize<W extends Tensor> extends RawOp {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.nn.QuantizedConv2DWithBiasAndReluAndRequantize}
@@ -83,7 +83,7 @@ private Options() {
    * @return a new instance of QuantizedConv2DWithBiasAndReluAndRequantize
    */
   @Endpoint(describeByClass = true)
-  public static <W extends TType, T extends TType, U extends TType, V extends TType> QuantizedConv2DWithBiasAndReluAndRequantize<W> create(Scope scope, Operand<T> input, Operand<U> filter, Operand<V> bias, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, Operand<TFloat32> minFilter, Operand<TFloat32> maxFilter, Operand<TFloat32> minFreezedOutput, Operand<TFloat32> maxFreezedOutput, DataType<W> outType, List<Long> strides, String padding, Options... options) {
+  public static <W extends Tensor, T extends Tensor, U extends Tensor, V extends Tensor> QuantizedConv2DWithBiasAndReluAndRequantize<W> create(Scope scope, Operand<T> input, Operand<U> filter, Operand<V> bias, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, Operand<TFloat32> minFilter, Operand<TFloat32> maxFilter, Operand<TFloat32> minFreezedOutput, Operand<TFloat32> maxFreezedOutput, DataType<W> outType, List<Long> strides, String padding, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("QuantizedConv2DWithBiasAndReluAndRequantize", scope.makeOpName("QuantizedConv2DWithBiasAndReluAndRequantize"));
     opBuilder.addInput(input.asOutput());
     opBuilder.addInput(filter.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DWithBiasAndRequantize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DWithBiasAndRequantize.java
index af89fd48962..68ecc13322a 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DWithBiasAndRequantize.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DWithBiasAndRequantize.java
@@ -23,17 +23,17 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.family.TType;
 
 /**
  * @param <W> data type for {@code output()} output
  */
-public final class QuantizedConv2DWithBiasAndRequantize<W extends TType> extends RawOp {
+public final class QuantizedConv2DWithBiasAndRequantize<W extends Tensor> extends RawOp {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.nn.QuantizedConv2DWithBiasAndRequantize}
@@ -83,7 +83,7 @@ private Options() {
    * @return a new instance of QuantizedConv2DWithBiasAndRequantize
    */
   @Endpoint(describeByClass = true)
-  public static <W extends TType, T extends TType, U extends TType, V extends TType> QuantizedConv2DWithBiasAndRequantize<W> create(Scope scope, Operand<T> input, Operand<U> filter, Operand<V> bias, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, Operand<TFloat32> minFilter, Operand<TFloat32> maxFilter, Operand<TFloat32> minFreezedOutput, Operand<TFloat32> maxFreezedOutput, DataType<W> outType, List<Long> strides, String padding, Options... options) {
+  public static <W extends Tensor, T extends Tensor, U extends Tensor, V extends Tensor> QuantizedConv2DWithBiasAndRequantize<W> create(Scope scope, Operand<T> input, Operand<U> filter, Operand<V> bias, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, Operand<TFloat32> minFilter, Operand<TFloat32> maxFilter, Operand<TFloat32> minFreezedOutput, Operand<TFloat32> maxFreezedOutput, DataType<W> outType, List<Long> strides, String padding, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("QuantizedConv2DWithBiasAndRequantize", scope.makeOpName("QuantizedConv2DWithBiasAndRequantize"));
     opBuilder.addInput(input.asOutput());
     opBuilder.addInput(filter.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DWithBiasSignedSumAndReluAndRequantize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DWithBiasSignedSumAndReluAndRequantize.java
index cfa7770cf25..f549d05e064 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DWithBiasSignedSumAndReluAndRequantize.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DWithBiasSignedSumAndReluAndRequantize.java
@@ -23,17 +23,17 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.family.TType;
 
 /**
  * @param <X> data type for {@code output()} output
  */
-public final class QuantizedConv2DWithBiasSignedSumAndReluAndRequantize<X extends TType> extends RawOp {
+public final class QuantizedConv2DWithBiasSignedSumAndReluAndRequantize<X extends Tensor> extends RawOp {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.nn.QuantizedConv2DWithBiasSignedSumAndReluAndRequantize}
@@ -86,7 +86,7 @@ private Options() {
    * @return a new instance of QuantizedConv2DWithBiasSignedSumAndReluAndRequantize
    */
   @Endpoint(describeByClass = true)
-  public static <X extends TType, T extends TType, U extends TType, V extends TType, W extends TType> QuantizedConv2DWithBiasSignedSumAndReluAndRequantize<X> create(Scope scope, Operand<T> input, Operand<U> filter, Operand<V> bias, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, Operand<TFloat32> minFilter, Operand<TFloat32> maxFilter, Operand<TFloat32> minFreezedOutput, Operand<TFloat32> maxFreezedOutput, Operand<W> summand, Operand<TFloat32> minSummand, Operand<TFloat32> maxSummand, DataType<X> outType, List<Long> strides, String padding, Options... options) {
+  public static <X extends Tensor, T extends Tensor, U extends Tensor, V extends Tensor, W extends Tensor> QuantizedConv2DWithBiasSignedSumAndReluAndRequantize<X> create(Scope scope, Operand<T> input, Operand<U> filter, Operand<V> bias, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, Operand<TFloat32> minFilter, Operand<TFloat32> maxFilter, Operand<TFloat32> minFreezedOutput, Operand<TFloat32> maxFreezedOutput, Operand<W> summand, Operand<TFloat32> minSummand, Operand<TFloat32> maxSummand, DataType<X> outType, List<Long> strides, String padding, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("QuantizedConv2DWithBiasSignedSumAndReluAndRequantize", scope.makeOpName("QuantizedConv2DWithBiasSignedSumAndReluAndRequantize"));
     opBuilder.addInput(input.asOutput());
     opBuilder.addInput(filter.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DWithBiasSumAndRelu.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DWithBiasSumAndRelu.java
index 1096fa12578..ebb5799a648 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DWithBiasSumAndRelu.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DWithBiasSumAndRelu.java
@@ -23,17 +23,17 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.family.TType;
 
 /**
  * @param <V> data type for {@code output()} output
  */
-public final class QuantizedConv2DWithBiasSumAndRelu<V extends TType> extends RawOp {
+public final class QuantizedConv2DWithBiasSumAndRelu<V extends Tensor> extends RawOp {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.nn.QuantizedConv2DWithBiasSumAndRelu}
@@ -82,7 +82,7 @@ private Options() {
    * @return a new instance of QuantizedConv2DWithBiasSumAndRelu
    */
   @Endpoint(describeByClass = true)
-  public static <V extends TType, T extends TType, U extends TType> QuantizedConv2DWithBiasSumAndRelu<V> create(Scope scope, Operand<T> input, Operand<U> filter, Operand<TFloat32> bias, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, Operand<TFloat32> minFilter, Operand<TFloat32> maxFilter, Operand<TFloat32> summand, DataType<V> outType, List<Long> strides, String padding, Options... options) {
+  public static <V extends Tensor, T extends Tensor, U extends Tensor> QuantizedConv2DWithBiasSumAndRelu<V> create(Scope scope, Operand<T> input, Operand<U> filter, Operand<TFloat32> bias, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, Operand<TFloat32> minFilter, Operand<TFloat32> maxFilter, Operand<TFloat32> summand, DataType<V> outType, List<Long> strides, String padding, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("QuantizedConv2DWithBiasSumAndRelu", scope.makeOpName("QuantizedConv2DWithBiasSumAndRelu"));
     opBuilder.addInput(input.asOutput());
     opBuilder.addInput(filter.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DWithBiasSumAndReluAndRequantize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DWithBiasSumAndReluAndRequantize.java
index 55adb0e016d..cbcad086f19 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DWithBiasSumAndReluAndRequantize.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DWithBiasSumAndReluAndRequantize.java
@@ -23,17 +23,17 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.family.TType;
 
 /**
  * @param <X> data type for {@code output()} output
  */
-public final class QuantizedConv2DWithBiasSumAndReluAndRequantize<X extends TType> extends RawOp {
+public final class QuantizedConv2DWithBiasSumAndReluAndRequantize<X extends Tensor> extends RawOp {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.nn.QuantizedConv2DWithBiasSumAndReluAndRequantize}
@@ -86,7 +86,7 @@ private Options() {
    * @return a new instance of QuantizedConv2DWithBiasSumAndReluAndRequantize
    */
   @Endpoint(describeByClass = true)
-  public static <X extends TType, T extends TType, U extends TType, V extends TType, W extends TType> QuantizedConv2DWithBiasSumAndReluAndRequantize<X> create(Scope scope, Operand<T> input, Operand<U> filter, Operand<V> bias, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, Operand<TFloat32> minFilter, Operand<TFloat32> maxFilter, Operand<TFloat32> minFreezedOutput, Operand<TFloat32> maxFreezedOutput, Operand<W> summand, Operand<TFloat32> minSummand, Operand<TFloat32> maxSummand, DataType<X> outType, List<Long> strides, String padding, Options... options) {
+  public static <X extends Tensor, T extends Tensor, U extends Tensor, V extends Tensor, W extends Tensor> QuantizedConv2DWithBiasSumAndReluAndRequantize<X> create(Scope scope, Operand<T> input, Operand<U> filter, Operand<V> bias, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, Operand<TFloat32> minFilter, Operand<TFloat32> maxFilter, Operand<TFloat32> minFreezedOutput, Operand<TFloat32> maxFreezedOutput, Operand<W> summand, Operand<TFloat32> minSummand, Operand<TFloat32> maxSummand, DataType<X> outType, List<Long> strides, String padding, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("QuantizedConv2DWithBiasSumAndReluAndRequantize", scope.makeOpName("QuantizedConv2DWithBiasSumAndReluAndRequantize"));
     opBuilder.addInput(input.asOutput());
     opBuilder.addInput(filter.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2d.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2d.java
index 39913922bdb..a981e3d9f28 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2d.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2d.java
@@ -23,12 +23,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes a 2D convolution given quantized 4D input and filter tensors.
@@ -41,7 +41,7 @@
  * @param <V> data type for {@code output()} output
  */
 @Operator(group = "nn")
-public final class QuantizedConv2d<V extends TType> extends RawOp {
+public final class QuantizedConv2d<V extends Tensor> extends RawOp {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.nn.QuantizedConv2d}
@@ -84,7 +84,7 @@ private Options() {
    * @return a new instance of QuantizedConv2d
    */
   @Endpoint(describeByClass = true)
-  public static <V extends TType, T extends TType, U extends TType> QuantizedConv2d<V> create(Scope scope, Operand<T> input, Operand<U> filter, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, Operand<TFloat32> minFilter, Operand<TFloat32> maxFilter, DataType<V> outType, List<Long> strides, String padding, Options... options) {
+  public static <V extends Tensor, T extends Tensor, U extends Tensor> QuantizedConv2d<V> create(Scope scope, Operand<T> input, Operand<U> filter, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, Operand<TFloat32> minFilter, Operand<TFloat32> maxFilter, DataType<V> outType, List<Long> strides, String padding, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("QuantizedConv2D", scope.makeOpName("QuantizedConv2d"));
     opBuilder.addInput(input.asOutput());
     opBuilder.addInput(filter.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedDepthwiseConv2D.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedDepthwiseConv2D.java
index f64d9a7efbb..e4511f72a84 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedDepthwiseConv2D.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedDepthwiseConv2D.java
@@ -23,19 +23,19 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes quantized depthwise Conv2D.
  * 
  * @param <V> data type for {@code output()} output
  */
-public final class QuantizedDepthwiseConv2D<V extends TType> extends RawOp {
+public final class QuantizedDepthwiseConv2D<V extends Tensor> extends RawOp {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.nn.QuantizedDepthwiseConv2D}
@@ -73,7 +73,7 @@ private Options() {
    * @return a new instance of QuantizedDepthwiseConv2D
    */
   @Endpoint(describeByClass = true)
-  public static <V extends TType, T extends TType, U extends TType> QuantizedDepthwiseConv2D<V> create(Scope scope, Operand<T> input, Operand<U> filter, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, Operand<TFloat32> minFilter, Operand<TFloat32> maxFilter, DataType<V> outType, List<Long> strides, String padding, Options... options) {
+  public static <V extends Tensor, T extends Tensor, U extends Tensor> QuantizedDepthwiseConv2D<V> create(Scope scope, Operand<T> input, Operand<U> filter, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, Operand<TFloat32> minFilter, Operand<TFloat32> maxFilter, DataType<V> outType, List<Long> strides, String padding, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("QuantizedDepthwiseConv2D", scope.makeOpName("QuantizedDepthwiseConv2D"));
     opBuilder.addInput(input.asOutput());
     opBuilder.addInput(filter.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedDepthwiseConv2DWithBias.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedDepthwiseConv2DWithBias.java
index be5e2bb8657..b70e505a990 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedDepthwiseConv2DWithBias.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedDepthwiseConv2DWithBias.java
@@ -23,19 +23,19 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes quantized depthwise Conv2D with Bias.
  * 
  * @param <V> data type for {@code output()} output
  */
-public final class QuantizedDepthwiseConv2DWithBias<V extends TType> extends RawOp {
+public final class QuantizedDepthwiseConv2DWithBias<V extends Tensor> extends RawOp {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.nn.QuantizedDepthwiseConv2DWithBias}
@@ -74,7 +74,7 @@ private Options() {
    * @return a new instance of QuantizedDepthwiseConv2DWithBias
    */
   @Endpoint(describeByClass = true)
-  public static <V extends TType, T extends TType, U extends TType> QuantizedDepthwiseConv2DWithBias<V> create(Scope scope, Operand<T> input, Operand<U> filter, Operand<TFloat32> bias, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, Operand<TFloat32> minFilter, Operand<TFloat32> maxFilter, DataType<V> outType, List<Long> strides, String padding, Options... options) {
+  public static <V extends Tensor, T extends Tensor, U extends Tensor> QuantizedDepthwiseConv2DWithBias<V> create(Scope scope, Operand<T> input, Operand<U> filter, Operand<TFloat32> bias, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, Operand<TFloat32> minFilter, Operand<TFloat32> maxFilter, DataType<V> outType, List<Long> strides, String padding, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("QuantizedDepthwiseConv2DWithBias", scope.makeOpName("QuantizedDepthwiseConv2DWithBias"));
     opBuilder.addInput(input.asOutput());
     opBuilder.addInput(filter.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedDepthwiseConv2DWithBiasAndRelu.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedDepthwiseConv2DWithBiasAndRelu.java
index 8abd12b865f..46d2fd032a5 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedDepthwiseConv2DWithBiasAndRelu.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedDepthwiseConv2DWithBiasAndRelu.java
@@ -23,19 +23,19 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes quantized depthwise Conv2D with Bias and Relu.
  * 
  * @param <V> data type for {@code output()} output
  */
-public final class QuantizedDepthwiseConv2DWithBiasAndRelu<V extends TType> extends RawOp {
+public final class QuantizedDepthwiseConv2DWithBiasAndRelu<V extends Tensor> extends RawOp {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.nn.QuantizedDepthwiseConv2DWithBiasAndRelu}
@@ -83,7 +83,7 @@ private Options() {
    * @return a new instance of QuantizedDepthwiseConv2DWithBiasAndRelu
    */
   @Endpoint(describeByClass = true)
-  public static <V extends TType, T extends TType, U extends TType> QuantizedDepthwiseConv2DWithBiasAndRelu<V> create(Scope scope, Operand<T> input, Operand<U> filter, Operand<TFloat32> bias, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, Operand<TFloat32> minFilter, Operand<TFloat32> maxFilter, DataType<V> outType, List<Long> strides, String padding, Options... options) {
+  public static <V extends Tensor, T extends Tensor, U extends Tensor> QuantizedDepthwiseConv2DWithBiasAndRelu<V> create(Scope scope, Operand<T> input, Operand<U> filter, Operand<TFloat32> bias, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, Operand<TFloat32> minFilter, Operand<TFloat32> maxFilter, DataType<V> outType, List<Long> strides, String padding, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("QuantizedDepthwiseConv2DWithBiasAndRelu", scope.makeOpName("QuantizedDepthwiseConv2DWithBiasAndRelu"));
     opBuilder.addInput(input.asOutput());
     opBuilder.addInput(filter.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedDepthwiseConv2DWithBiasAndReluAndRequantize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedDepthwiseConv2DWithBiasAndReluAndRequantize.java
index 78c8048266e..c2638715575 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedDepthwiseConv2DWithBiasAndReluAndRequantize.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedDepthwiseConv2DWithBiasAndReluAndRequantize.java
@@ -23,19 +23,19 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes quantized depthwise Conv2D with Bias, Relu and Requantize.
  * 
  * @param <W> data type for {@code output()} output
  */
-public final class QuantizedDepthwiseConv2DWithBiasAndReluAndRequantize<W extends TType> extends RawOp {
+public final class QuantizedDepthwiseConv2DWithBiasAndReluAndRequantize<W extends Tensor> extends RawOp {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.nn.QuantizedDepthwiseConv2DWithBiasAndReluAndRequantize}
@@ -85,7 +85,7 @@ private Options() {
    * @return a new instance of QuantizedDepthwiseConv2DWithBiasAndReluAndRequantize
    */
   @Endpoint(describeByClass = true)
-  public static <W extends TType, T extends TType, U extends TType, V extends TType> QuantizedDepthwiseConv2DWithBiasAndReluAndRequantize<W> create(Scope scope, Operand<T> input, Operand<U> filter, Operand<V> bias, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, Operand<TFloat32> minFilter, Operand<TFloat32> maxFilter, Operand<TFloat32> minFreezedOutput, Operand<TFloat32> maxFreezedOutput, DataType<W> outType, List<Long> strides, String padding, Options... options) {
+  public static <W extends Tensor, T extends Tensor, U extends Tensor, V extends Tensor> QuantizedDepthwiseConv2DWithBiasAndReluAndRequantize<W> create(Scope scope, Operand<T> input, Operand<U> filter, Operand<V> bias, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, Operand<TFloat32> minFilter, Operand<TFloat32> maxFilter, Operand<TFloat32> minFreezedOutput, Operand<TFloat32> maxFreezedOutput, DataType<W> outType, List<Long> strides, String padding, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("QuantizedDepthwiseConv2DWithBiasAndReluAndRequantize", scope.makeOpName("QuantizedDepthwiseConv2DWithBiasAndReluAndRequantize"));
     opBuilder.addInput(input.asOutput());
     opBuilder.addInput(filter.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedInstanceNorm.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedInstanceNorm.java
index 839338a1458..87dac703ac9 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedInstanceNorm.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedInstanceNorm.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.family.TType;
 
 /**
  * Quantized Instance normalization.
@@ -34,7 +34,7 @@
  * @param <T> data type for {@code y()} output
  */
 @Operator(group = "nn")
-public final class QuantizedInstanceNorm<T extends TType> extends RawOp {
+public final class QuantizedInstanceNorm<T extends Tensor> extends RawOp {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.nn.QuantizedInstanceNorm}
@@ -104,7 +104,7 @@ private Options() {
    * @return a new instance of QuantizedInstanceNorm
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> QuantizedInstanceNorm<T> create(Scope scope, Operand<T> x, Operand<TFloat32> xMin, Operand<TFloat32> xMax, Options... options) {
+  public static <T extends Tensor> QuantizedInstanceNorm<T> create(Scope scope, Operand<T> x, Operand<TFloat32> xMin, Operand<TFloat32> xMax, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("QuantizedInstanceNorm", scope.makeOpName("QuantizedInstanceNorm"));
     opBuilder.addInput(x.asOutput());
     opBuilder.addInput(xMin.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedMaxPool.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedMaxPool.java
index 2cec96e411c..206df4fc20c 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedMaxPool.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedMaxPool.java
@@ -22,12 +22,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.family.TType;
 
 /**
  * Produces the max pool of the input tensor for quantized types.
@@ -35,7 +35,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "nn")
-public final class QuantizedMaxPool<T extends TType> extends RawOp {
+public final class QuantizedMaxPool<T extends Tensor> extends RawOp {
   
   /**
    * Factory method to create a class wrapping a new QuantizedMaxPool operation.
@@ -52,7 +52,7 @@ public final class QuantizedMaxPool<T extends TType> extends RawOp {
    * @return a new instance of QuantizedMaxPool
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> QuantizedMaxPool<T> create(Scope scope, Operand<T> input, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, List<Long> ksize, List<Long> strides, String padding) {
+  public static <T extends Tensor> QuantizedMaxPool<T> create(Scope scope, Operand<T> input, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, List<Long> ksize, List<Long> strides, String padding) {
     OperationBuilder opBuilder = scope.env().opBuilder("QuantizedMaxPool", scope.makeOpName("QuantizedMaxPool"));
     opBuilder.addInput(input.asOutput());
     opBuilder.addInput(minInput.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedRelu.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedRelu.java
index 308e14e9512..3064343acaa 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedRelu.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedRelu.java
@@ -22,12 +22,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes Quantized Rectified Linear: `max(features, 0)`
@@ -35,7 +35,7 @@
  * @param <U> data type for {@code activations()} output
  */
 @Operator(group = "nn")
-public final class QuantizedRelu<U extends TType> extends RawOp {
+public final class QuantizedRelu<U extends Tensor> extends RawOp {
   
   /**
    * Factory method to create a class wrapping a new QuantizedRelu operation.
@@ -48,7 +48,7 @@ public final class QuantizedRelu<U extends TType> extends RawOp {
    * @return a new instance of QuantizedRelu
    */
   @Endpoint(describeByClass = true)
-  public static <U extends TType, T extends TType> QuantizedRelu<U> create(Scope scope, Operand<T> features, Operand<TFloat32> minFeatures, Operand<TFloat32> maxFeatures, DataType<U> outType) {
+  public static <U extends Tensor, T extends Tensor> QuantizedRelu<U> create(Scope scope, Operand<T> features, Operand<TFloat32> minFeatures, Operand<TFloat32> maxFeatures, DataType<U> outType) {
     OperationBuilder opBuilder = scope.env().opBuilder("QuantizedRelu", scope.makeOpName("QuantizedRelu"));
     opBuilder.addInput(features.asOutput());
     opBuilder.addInput(minFeatures.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedRelu6.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedRelu6.java
index 46f04fd5722..60b45b90ba5 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedRelu6.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedRelu6.java
@@ -22,12 +22,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes Quantized Rectified Linear 6: `min(max(features, 0), 6)`
@@ -35,7 +35,7 @@
  * @param <U> data type for {@code activations()} output
  */
 @Operator(group = "nn")
-public final class QuantizedRelu6<U extends TType> extends RawOp {
+public final class QuantizedRelu6<U extends Tensor> extends RawOp {
   
   /**
    * Factory method to create a class wrapping a new QuantizedRelu6 operation.
@@ -48,7 +48,7 @@ public final class QuantizedRelu6<U extends TType> extends RawOp {
    * @return a new instance of QuantizedRelu6
    */
   @Endpoint(describeByClass = true)
-  public static <U extends TType, T extends TType> QuantizedRelu6<U> create(Scope scope, Operand<T> features, Operand<TFloat32> minFeatures, Operand<TFloat32> maxFeatures, DataType<U> outType) {
+  public static <U extends Tensor, T extends Tensor> QuantizedRelu6<U> create(Scope scope, Operand<T> features, Operand<TFloat32> minFeatures, Operand<TFloat32> maxFeatures, DataType<U> outType) {
     OperationBuilder opBuilder = scope.env().opBuilder("QuantizedRelu6", scope.makeOpName("QuantizedRelu6"));
     opBuilder.addInput(features.asOutput());
     opBuilder.addInput(minFeatures.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedReluX.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedReluX.java
index e47a6a5c043..17604d822d2 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedReluX.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedReluX.java
@@ -22,12 +22,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes Quantized Rectified Linear X: `min(max(features, 0), max_value)`
@@ -35,7 +35,7 @@
  * @param <U> data type for {@code activations()} output
  */
 @Operator(group = "nn")
-public final class QuantizedReluX<U extends TType> extends RawOp {
+public final class QuantizedReluX<U extends Tensor> extends RawOp {
   
   /**
    * Factory method to create a class wrapping a new QuantizedReluX operation.
@@ -49,7 +49,7 @@ public final class QuantizedReluX<U extends TType> extends RawOp {
    * @return a new instance of QuantizedReluX
    */
   @Endpoint(describeByClass = true)
-  public static <U extends TType, T extends TType> QuantizedReluX<U> create(Scope scope, Operand<T> features, Operand<TFloat32> maxValue, Operand<TFloat32> minFeatures, Operand<TFloat32> maxFeatures, DataType<U> outType) {
+  public static <U extends Tensor, T extends Tensor> QuantizedReluX<U> create(Scope scope, Operand<T> features, Operand<TFloat32> maxValue, Operand<TFloat32> minFeatures, Operand<TFloat32> maxFeatures, DataType<U> outType) {
     OperationBuilder opBuilder = scope.env().opBuilder("QuantizedReluX", scope.makeOpName("QuantizedReluX"));
     opBuilder.addInput(features.asOutput());
     opBuilder.addInput(maxValue.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Relu.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Relu.java
index c6a515d4c75..4ddf49d6677 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Relu.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Relu.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes rectified linear: `max(features, 0)`.
@@ -38,7 +38,7 @@
  * @param <T> data type for {@code activations()} output
  */
 @Operator(group = "nn")
-public final class Relu<T extends TType> extends RawOp implements Operand<T> {
+public final class Relu<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Relu operation.
@@ -48,7 +48,7 @@ public final class Relu<T extends TType> extends RawOp implements Operand<T> {
    * @return a new instance of Relu
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> Relu<T> create(Scope scope, Operand<T> features) {
+  public static <T extends Tensor> Relu<T> create(Scope scope, Operand<T> features) {
     OperationBuilder opBuilder = scope.env().opBuilder("Relu", scope.makeOpName("Relu"));
     opBuilder.addInput(features.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Relu6.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Relu6.java
index e4e674b7e35..cfd822bf138 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Relu6.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Relu6.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes rectified linear 6: `min(max(features, 0), 6)`.
@@ -34,7 +34,7 @@
  * @param <T> data type for {@code activations()} output
  */
 @Operator(group = "nn")
-public final class Relu6<T extends TNumber> extends RawOp implements Operand<T> {
+public final class Relu6<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Relu6 operation.
@@ -44,7 +44,7 @@ public final class Relu6<T extends TNumber> extends RawOp implements Operand<T>
    * @return a new instance of Relu6
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> Relu6<T> create(Scope scope, Operand<T> features) {
+  public static <T extends Tensor & TNumber> Relu6<T> create(Scope scope, Operand<T> features) {
     OperationBuilder opBuilder = scope.env().opBuilder("Relu6", scope.makeOpName("Relu6"));
     opBuilder.addInput(features.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Relu6Grad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Relu6Grad.java
index cf4edc9debf..3555a2fa244 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Relu6Grad.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Relu6Grad.java
@@ -21,19 +21,19 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes rectified linear 6 gradients for a Relu6 operation.
  * 
  * @param <T> data type for {@code backprops()} output
  */
-public final class Relu6Grad<T extends TNumber> extends RawOp implements Operand<T> {
+public final class Relu6Grad<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Relu6Grad operation.
@@ -45,7 +45,7 @@ public final class Relu6Grad<T extends TNumber> extends RawOp implements Operand
    * @return a new instance of Relu6Grad
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> Relu6Grad<T> create(Scope scope, Operand<T> gradients, Operand<T> features) {
+  public static <T extends Tensor & TNumber> Relu6Grad<T> create(Scope scope, Operand<T> gradients, Operand<T> features) {
     OperationBuilder opBuilder = scope.env().opBuilder("Relu6Grad", scope.makeOpName("Relu6Grad"));
     opBuilder.addInput(gradients.asOutput());
     opBuilder.addInput(features.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/ReluGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/ReluGrad.java
index d1c7cf7d44c..790d26427c6 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/ReluGrad.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/ReluGrad.java
@@ -21,19 +21,19 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes rectified linear gradients for a Relu operation.
  * 
  * @param <T> data type for {@code backprops()} output
  */
-public final class ReluGrad<T extends TNumber> extends RawOp implements Operand<T> {
+public final class ReluGrad<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new ReluGrad operation.
@@ -45,7 +45,7 @@ public final class ReluGrad<T extends TNumber> extends RawOp implements Operand<
    * @return a new instance of ReluGrad
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> ReluGrad<T> create(Scope scope, Operand<T> gradients, Operand<T> features) {
+  public static <T extends Tensor & TNumber> ReluGrad<T> create(Scope scope, Operand<T> gradients, Operand<T> features) {
     OperationBuilder opBuilder = scope.env().opBuilder("ReluGrad", scope.makeOpName("ReluGrad"));
     opBuilder.addInput(gradients.asOutput());
     opBuilder.addInput(features.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Selu.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Selu.java
index 2ce8484b299..802312a4f32 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Selu.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Selu.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes scaled exponential linear: `scale * alpha * (exp(features) - 1)`
@@ -42,7 +42,7 @@
  * @param <T> data type for {@code activations()} output
  */
 @Operator(group = "nn")
-public final class Selu<T extends TNumber> extends RawOp implements Operand<T> {
+public final class Selu<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Selu operation.
@@ -52,7 +52,7 @@ public final class Selu<T extends TNumber> extends RawOp implements Operand<T> {
    * @return a new instance of Selu
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> Selu<T> create(Scope scope, Operand<T> features) {
+  public static <T extends Tensor & TNumber> Selu<T> create(Scope scope, Operand<T> features) {
     OperationBuilder opBuilder = scope.env().opBuilder("Selu", scope.makeOpName("Selu"));
     opBuilder.addInput(features.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/SeluGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/SeluGrad.java
index 514ad00b38d..b6f5abc4e95 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/SeluGrad.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/SeluGrad.java
@@ -21,19 +21,19 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes gradients for the scaled exponential linear (Selu) operation.
  * 
  * @param <T> data type for {@code backprops()} output
  */
-public final class SeluGrad<T extends TNumber> extends RawOp implements Operand<T> {
+public final class SeluGrad<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new SeluGrad operation.
@@ -44,7 +44,7 @@ public final class SeluGrad<T extends TNumber> extends RawOp implements Operand<
    * @return a new instance of SeluGrad
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> SeluGrad<T> create(Scope scope, Operand<T> gradients, Operand<T> outputs) {
+  public static <T extends Tensor & TNumber> SeluGrad<T> create(Scope scope, Operand<T> gradients, Operand<T> outputs) {
     OperationBuilder opBuilder = scope.env().opBuilder("SeluGrad", scope.makeOpName("SeluGrad"));
     opBuilder.addInput(gradients.asOutput());
     opBuilder.addInput(outputs.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Softmax.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Softmax.java
index d9971b6667d..a3ed77b8e83 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Softmax.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Softmax.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes softmax activations.
@@ -38,7 +38,7 @@
  * @param <T> data type for {@code softmax()} output
  */
 @Operator(group = "nn")
-public final class Softmax<T extends TNumber> extends RawOp implements Operand<T> {
+public final class Softmax<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Softmax operation.
@@ -48,7 +48,7 @@ public final class Softmax<T extends TNumber> extends RawOp implements Operand<T
    * @return a new instance of Softmax
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> Softmax<T> create(Scope scope, Operand<T> logits) {
+  public static <T extends Tensor & TNumber> Softmax<T> create(Scope scope, Operand<T> logits) {
     OperationBuilder opBuilder = scope.env().opBuilder("Softmax", scope.makeOpName("Softmax"));
     opBuilder.addInput(logits.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Softsign.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Softsign.java
index e4276c679a3..8ea369d0b27 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Softsign.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Softsign.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes softsign: `features / (abs(features) + 1)`.
@@ -34,7 +34,7 @@
  * @param <T> data type for {@code activations()} output
  */
 @Operator(group = "nn")
-public final class Softsign<T extends TNumber> extends RawOp implements Operand<T> {
+public final class Softsign<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Softsign operation.
@@ -44,7 +44,7 @@ public final class Softsign<T extends TNumber> extends RawOp implements Operand<
    * @return a new instance of Softsign
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> Softsign<T> create(Scope scope, Operand<T> features) {
+  public static <T extends Tensor & TNumber> Softsign<T> create(Scope scope, Operand<T> features) {
     OperationBuilder opBuilder = scope.env().opBuilder("Softsign", scope.makeOpName("Softsign"));
     opBuilder.addInput(features.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/SoftsignGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/SoftsignGrad.java
index 220763e57a2..f8b55b3ecd9 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/SoftsignGrad.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/SoftsignGrad.java
@@ -21,19 +21,19 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes softsign gradients for a softsign operation.
  * 
  * @param <T> data type for {@code backprops()} output
  */
-public final class SoftsignGrad<T extends TNumber> extends RawOp implements Operand<T> {
+public final class SoftsignGrad<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new SoftsignGrad operation.
@@ -44,7 +44,7 @@ public final class SoftsignGrad<T extends TNumber> extends RawOp implements Oper
    * @return a new instance of SoftsignGrad
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> SoftsignGrad<T> create(Scope scope, Operand<T> gradients, Operand<T> features) {
+  public static <T extends Tensor & TNumber> SoftsignGrad<T> create(Scope scope, Operand<T> gradients, Operand<T> features) {
     OperationBuilder opBuilder = scope.env().opBuilder("SoftsignGrad", scope.makeOpName("SoftsignGrad"));
     opBuilder.addInput(gradients.asOutput());
     opBuilder.addInput(features.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/SpaceToBatch.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/SpaceToBatch.java
index 9b1796c99aa..93d0d1e3b21 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/SpaceToBatch.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/SpaceToBatch.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * SpaceToBatch for 4-D tensors of type T.
@@ -42,7 +42,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "nn")
-public final class SpaceToBatch<T extends TType> extends RawOp implements Operand<T> {
+public final class SpaceToBatch<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new SpaceToBatch operation.
@@ -122,7 +122,7 @@ public final class SpaceToBatch<T extends TType> extends RawOp implements Operan
    * @return a new instance of SpaceToBatch
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> SpaceToBatch<T> create(Scope scope, Operand<T> input, Operand<U> paddings, Long blockSize) {
+  public static <T extends Tensor, U extends Tensor & TNumber> SpaceToBatch<T> create(Scope scope, Operand<T> input, Operand<U> paddings, Long blockSize) {
     OperationBuilder opBuilder = scope.env().opBuilder("SpaceToBatch", scope.makeOpName("SpaceToBatch"));
     opBuilder.addInput(input.asOutput());
     opBuilder.addInput(paddings.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/SpaceToDepth.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/SpaceToDepth.java
index 0e9068c9b31..448902a1048 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/SpaceToDepth.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/SpaceToDepth.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * SpaceToDepth for tensors of type T.
@@ -107,7 +107,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "nn")
-public final class SpaceToDepth<T extends TType> extends RawOp implements Operand<T> {
+public final class SpaceToDepth<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.nn.SpaceToDepth}
@@ -138,7 +138,7 @@ private Options() {
    * @return a new instance of SpaceToDepth
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> SpaceToDepth<T> create(Scope scope, Operand<T> input, Long blockSize, Options... options) {
+  public static <T extends Tensor> SpaceToDepth<T> create(Scope scope, Operand<T> input, Long blockSize, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("SpaceToDepth", scope.makeOpName("SpaceToDepth"));
     opBuilder.addInput(input.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/TopK.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/TopK.java
index acd4ba679f7..5f90a2a5868 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/TopK.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/TopK.java
@@ -21,13 +21,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Finds values and indices of the `k` largest elements for the last dimension.
@@ -46,7 +46,7 @@
  * @param <T> data type for {@code values()} output
  */
 @Operator(group = "nn")
-public final class TopK<T extends TNumber> extends RawOp {
+public final class TopK<T extends Tensor & TNumber> extends RawOp {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.nn.TopK}
@@ -79,7 +79,7 @@ private Options() {
    * @return a new instance of TopK
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> TopK<T> create(Scope scope, Operand<T> input, Operand<TInt32> k, Options... options) {
+  public static <T extends Tensor & TNumber> TopK<T> create(Scope scope, Operand<T> input, Operand<TInt32> k, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("TopKV2", scope.makeOpName("TopK"));
     opBuilder.addInput(input.asOutput());
     opBuilder.addInput(k.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/raw/SoftmaxCrossEntropyWithLogits.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/raw/SoftmaxCrossEntropyWithLogits.java
index 4c23683d9ef..61a8bb03593 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/raw/SoftmaxCrossEntropyWithLogits.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/raw/SoftmaxCrossEntropyWithLogits.java
@@ -15,18 +15,18 @@
 
 // This class has been generated, DO NOT EDIT!
 
-package org.tensorflow.op.nn.raw;
+package org.tensorflow.op.nn;
 
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes softmax cross entropy cost and gradients to backpropagate.
@@ -35,8 +35,8 @@
  * 
  * @param <T> data type for {@code loss()} output
  */
-@Operator(group = "nn.raw")
-public final class SoftmaxCrossEntropyWithLogits<T extends TNumber> extends RawOp {
+@Operator(group = "nn")
+public final class SoftmaxCrossEntropyWithLogits<T extends Tensor & TNumber> extends RawOp {
   
   /**
    * Factory method to create a class wrapping a new SoftmaxCrossEntropyWithLogits operation.
@@ -49,7 +49,7 @@ public final class SoftmaxCrossEntropyWithLogits<T extends TNumber> extends RawO
    * @return a new instance of SoftmaxCrossEntropyWithLogits
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> SoftmaxCrossEntropyWithLogits<T> create(Scope scope, Operand<T> features, Operand<T> labels) {
+  public static <T extends Tensor & TNumber> SoftmaxCrossEntropyWithLogits<T> create(Scope scope, Operand<T> features, Operand<T> labels) {
     OperationBuilder opBuilder = scope.env().opBuilder("SoftmaxCrossEntropyWithLogits", scope.makeOpName("SoftmaxCrossEntropyWithLogits"));
     opBuilder.addInput(features.asOutput());
     opBuilder.addInput(labels.asOutput());
@@ -71,9 +71,6 @@ public Output<T> backprop() {
     return backprop;
   }
   
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "SoftmaxCrossEntropyWithLogits";
-  
   private Output<T> loss;
   private Output<T> backprop;
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/raw/SparseSoftmaxCrossEntropyWithLogits.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/raw/SparseSoftmaxCrossEntropyWithLogits.java
index e7cb45231de..9ddb57134df 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/raw/SparseSoftmaxCrossEntropyWithLogits.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/raw/SparseSoftmaxCrossEntropyWithLogits.java
@@ -15,18 +15,18 @@
 
 // This class has been generated, DO NOT EDIT!
 
-package org.tensorflow.op.nn.raw;
+package org.tensorflow.op.nn;
 
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes softmax cross entropy cost and gradients to backpropagate.
@@ -40,8 +40,8 @@
  * 
  * @param <T> data type for {@code loss()} output
  */
-@Operator(group = "nn.raw")
-public final class SparseSoftmaxCrossEntropyWithLogits<T extends TNumber> extends RawOp {
+@Operator(group = "nn")
+public final class SparseSoftmaxCrossEntropyWithLogits<T extends Tensor & TNumber> extends RawOp {
   
   /**
    * Factory method to create a class wrapping a new SparseSoftmaxCrossEntropyWithLogits operation.
@@ -53,7 +53,7 @@ public final class SparseSoftmaxCrossEntropyWithLogits<T extends TNumber> extend
    * @return a new instance of SparseSoftmaxCrossEntropyWithLogits
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber, U extends TNumber> SparseSoftmaxCrossEntropyWithLogits<T> create(Scope scope, Operand<T> features, Operand<U> labels) {
+  public static <T extends Tensor & TNumber, U extends Tensor & TNumber> SparseSoftmaxCrossEntropyWithLogits<T> create(Scope scope, Operand<T> features, Operand<U> labels) {
     OperationBuilder opBuilder = scope.env().opBuilder("SparseSoftmaxCrossEntropyWithLogits", scope.makeOpName("SparseSoftmaxCrossEntropyWithLogits"));
     opBuilder.addInput(features.asOutput());
     opBuilder.addInput(labels.asOutput());
@@ -75,9 +75,6 @@ public Output<T> backprop() {
     return backprop;
   }
   
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "SparseSoftmaxCrossEntropyWithLogits";
-  
   private Output<T> loss;
   private Output<T> backprop;
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/Dequantize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/Dequantize.java
index 760dd9fe913..feeed35e27d 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/Dequantize.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/Dequantize.java
@@ -22,13 +22,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TFloat32;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Dequantize the 'input' tensor into a float or bfloat16 Tensor.
@@ -85,7 +85,7 @@
  * @param <U> data type for {@code output()} output
  */
 @Operator(group = "quantization")
-public final class Dequantize<U extends TNumber> extends RawOp implements Operand<U> {
+public final class Dequantize<U extends Tensor & TNumber> extends RawOp implements Operand<U> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.quantization.Dequantize}
@@ -137,7 +137,7 @@ private Options() {
    * @return a new instance of Dequantize
    */
   @Endpoint(describeByClass = true)
-  public static <U extends TNumber, T extends TType> Dequantize<U> create(Scope scope, Operand<T> input, Operand<TFloat32> minRange, Operand<TFloat32> maxRange, DataType<U> dtype, Options... options) {
+  public static <U extends Tensor & TNumber, T extends Tensor> Dequantize<U> create(Scope scope, Operand<T> input, Operand<TFloat32> minRange, Operand<TFloat32> maxRange, DataType<U> dtype, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("Dequantize", scope.makeOpName("Dequantize"));
     opBuilder.addInput(input.asOutput());
     opBuilder.addInput(minRange.asOutput());
@@ -171,7 +171,7 @@ public static <U extends TNumber, T extends TType> Dequantize<U> create(Scope sc
    * @return a new instance of Dequantize
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> Dequantize<TFloat32> create(Scope scope, Operand<T> input, Operand<TFloat32> minRange, Operand<TFloat32> maxRange, Options... options) {
+  public static <T extends Tensor> Dequantize<TFloat32> create(Scope scope, Operand<T> input, Operand<TFloat32> minRange, Operand<TFloat32> maxRange, Options... options) {
     return create(scope, input, minRange, maxRange, TFloat32.DTYPE, options);
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/Quantize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/Quantize.java
index 37f58cfa512..9e3d5cf80d2 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/Quantize.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/Quantize.java
@@ -22,12 +22,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.family.TType;
 
 /**
  * Quantize the 'input' tensor of type float to 'output' tensor of type 'T'.
@@ -145,7 +145,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "quantization")
-public final class Quantize<T extends TType> extends RawOp {
+public final class Quantize<T extends Tensor> extends RawOp {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.quantization.Quantize}
@@ -220,7 +220,7 @@ private Options() {
    * @return a new instance of Quantize
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> Quantize<T> create(Scope scope, Operand<TFloat32> input, Operand<TFloat32> minRange, Operand<TFloat32> maxRange, DataType<T> T, Options... options) {
+  public static <T extends Tensor> Quantize<T> create(Scope scope, Operand<TFloat32> input, Operand<TFloat32> minRange, Operand<TFloat32> maxRange, DataType<T> T, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("QuantizeV2", scope.makeOpName("Quantize"));
     opBuilder.addInput(input.asOutput());
     opBuilder.addInput(minRange.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/QuantizeAndDequantize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/QuantizeAndDequantize.java
index fd75b330e41..81d3aeb3d34 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/QuantizeAndDequantize.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/QuantizeAndDequantize.java
@@ -21,13 +21,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Quantizes then dequantizes a tensor.
@@ -38,7 +38,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "quantization")
-public final class QuantizeAndDequantize<T extends TNumber> extends RawOp implements Operand<T> {
+public final class QuantizeAndDequantize<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.quantization.QuantizeAndDequantize}
@@ -98,7 +98,7 @@ private Options() {
    * @return a new instance of QuantizeAndDequantize
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> QuantizeAndDequantize<T> create(Scope scope, Operand<T> input, Operand<T> inputMin, Operand<T> inputMax, Operand<TInt32> numBits, Options... options) {
+  public static <T extends Tensor & TNumber> QuantizeAndDequantize<T> create(Scope scope, Operand<T> input, Operand<T> inputMin, Operand<T> inputMax, Operand<TInt32> numBits, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("QuantizeAndDequantizeV3", scope.makeOpName("QuantizeAndDequantize"));
     opBuilder.addInput(input.asOutput());
     opBuilder.addInput(inputMin.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/QuantizeDownAndShrinkRange.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/QuantizeDownAndShrinkRange.java
index 362375b40e2..39839df983b 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/QuantizeDownAndShrinkRange.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/QuantizeDownAndShrinkRange.java
@@ -22,12 +22,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.family.TType;
 
 /**
  * Convert the quantized 'input' tensor into a lower-precision 'output', using the
@@ -58,7 +58,7 @@
  * @param <U> data type for {@code output()} output
  */
 @Operator(group = "quantization")
-public final class QuantizeDownAndShrinkRange<U extends TType> extends RawOp {
+public final class QuantizeDownAndShrinkRange<U extends Tensor> extends RawOp {
   
   /**
    * Factory method to create a class wrapping a new QuantizeDownAndShrinkRange operation.
@@ -71,7 +71,7 @@ public final class QuantizeDownAndShrinkRange<U extends TType> extends RawOp {
    * @return a new instance of QuantizeDownAndShrinkRange
    */
   @Endpoint(describeByClass = true)
-  public static <U extends TType, T extends TType> QuantizeDownAndShrinkRange<U> create(Scope scope, Operand<T> input, Operand<TFloat32> inputMin, Operand<TFloat32> inputMax, DataType<U> outType) {
+  public static <U extends Tensor, T extends Tensor> QuantizeDownAndShrinkRange<U> create(Scope scope, Operand<T> input, Operand<TFloat32> inputMin, Operand<TFloat32> inputMax, DataType<U> outType) {
     OperationBuilder opBuilder = scope.env().opBuilder("QuantizeDownAndShrinkRange", scope.makeOpName("QuantizeDownAndShrinkRange"));
     opBuilder.addInput(input.asOutput());
     opBuilder.addInput(inputMin.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/QuantizedConcat.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/QuantizedConcat.java
index 43839d072de..acb4e43c5ef 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/QuantizedConcat.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/QuantizedConcat.java
@@ -21,6 +21,7 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.Operands;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
@@ -28,7 +29,6 @@
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TFloat32;
 import org.tensorflow.types.TInt32;
-import org.tensorflow.types.family.TType;
 
 /**
  * Concatenates quantized tensors along one dimension.
@@ -36,7 +36,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "quantization")
-public final class QuantizedConcat<T extends TType> extends RawOp {
+public final class QuantizedConcat<T extends Tensor> extends RawOp {
   
   /**
    * Factory method to create a class wrapping a new QuantizedConcat operation.
@@ -51,7 +51,7 @@ public final class QuantizedConcat<T extends TType> extends RawOp {
    * @return a new instance of QuantizedConcat
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> QuantizedConcat<T> create(Scope scope, Operand<TInt32> concatDim, Iterable<Operand<T>> values, Iterable<Operand<TFloat32>> inputMins, Iterable<Operand<TFloat32>> inputMaxes) {
+  public static <T extends Tensor> QuantizedConcat<T> create(Scope scope, Operand<TInt32> concatDim, Iterable<Operand<T>> values, Iterable<Operand<TFloat32>> inputMins, Iterable<Operand<TFloat32>> inputMaxes) {
     OperationBuilder opBuilder = scope.env().opBuilder("QuantizedConcat", scope.makeOpName("QuantizedConcat"));
     opBuilder.addInput(concatDim.asOutput());
     opBuilder.addInputList(Operands.asOutputs(values));
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/QuantizedMatMulWithBiasAndDequantize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/QuantizedMatMulWithBiasAndDequantize.java
index baa20635c51..d6083ce1d21 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/QuantizedMatMulWithBiasAndDequantize.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/QuantizedMatMulWithBiasAndDequantize.java
@@ -22,18 +22,18 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TFloat32;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * @param <W> data type for {@code out()} output
  */
-public final class QuantizedMatMulWithBiasAndDequantize<W extends TNumber> extends RawOp implements Operand<W> {
+public final class QuantizedMatMulWithBiasAndDequantize<W extends Tensor & TNumber> extends RawOp implements Operand<W> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.quantization.QuantizedMatMulWithBiasAndDequantize}
@@ -90,7 +90,7 @@ private Options() {
    * @return a new instance of QuantizedMatMulWithBiasAndDequantize
    */
   @Endpoint(describeByClass = true)
-  public static <W extends TNumber, T extends TType, U extends TType, V extends TType> QuantizedMatMulWithBiasAndDequantize<W> create(Scope scope, Operand<T> a, Operand<U> b, Operand<V> bias, Operand<TFloat32> minA, Operand<TFloat32> maxA, Operand<TFloat32> minB, Operand<TFloat32> maxB, Operand<TFloat32> minFreezedOutput, Operand<TFloat32> maxFreezedOutput, DataType<W> Toutput, Options... options) {
+  public static <W extends Tensor & TNumber, T extends Tensor, U extends Tensor, V extends Tensor> QuantizedMatMulWithBiasAndDequantize<W> create(Scope scope, Operand<T> a, Operand<U> b, Operand<V> bias, Operand<TFloat32> minA, Operand<TFloat32> maxA, Operand<TFloat32> minB, Operand<TFloat32> maxB, Operand<TFloat32> minFreezedOutput, Operand<TFloat32> maxFreezedOutput, DataType<W> Toutput, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("QuantizedMatMulWithBiasAndDequantize", scope.makeOpName("QuantizedMatMulWithBiasAndDequantize"));
     opBuilder.addInput(a.asOutput());
     opBuilder.addInput(b.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/QuantizedMatMulWithBiasAndRequantize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/QuantizedMatMulWithBiasAndRequantize.java
index 950221b1b94..7c0bf738249 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/QuantizedMatMulWithBiasAndRequantize.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/QuantizedMatMulWithBiasAndRequantize.java
@@ -22,17 +22,17 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.family.TType;
 
 /**
  * @param <W> data type for {@code out()} output
  */
-public final class QuantizedMatMulWithBiasAndRequantize<W extends TType> extends RawOp {
+public final class QuantizedMatMulWithBiasAndRequantize<W extends Tensor> extends RawOp {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.quantization.QuantizedMatMulWithBiasAndRequantize}
@@ -89,7 +89,7 @@ private Options() {
    * @return a new instance of QuantizedMatMulWithBiasAndRequantize
    */
   @Endpoint(describeByClass = true)
-  public static <W extends TType, T extends TType, U extends TType, V extends TType> QuantizedMatMulWithBiasAndRequantize<W> create(Scope scope, Operand<T> a, Operand<U> b, Operand<V> bias, Operand<TFloat32> minA, Operand<TFloat32> maxA, Operand<TFloat32> minB, Operand<TFloat32> maxB, Operand<TFloat32> minFreezedOutput, Operand<TFloat32> maxFreezedOutput, DataType<W> Toutput, Options... options) {
+  public static <W extends Tensor, T extends Tensor, U extends Tensor, V extends Tensor> QuantizedMatMulWithBiasAndRequantize<W> create(Scope scope, Operand<T> a, Operand<U> b, Operand<V> bias, Operand<TFloat32> minA, Operand<TFloat32> maxA, Operand<TFloat32> minB, Operand<TFloat32> maxB, Operand<TFloat32> minFreezedOutput, Operand<TFloat32> maxFreezedOutput, DataType<W> Toutput, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("QuantizedMatMulWithBiasAndRequantize", scope.makeOpName("QuantizedMatMulWithBiasAndRequantize"));
     opBuilder.addInput(a.asOutput());
     opBuilder.addInput(b.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/RequantizationRange.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/RequantizationRange.java
index 86ab731bb50..c2cf9076a8d 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/RequantizationRange.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/RequantizationRange.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes a range that covers the actual values present in a quantized tensor.
@@ -49,7 +49,7 @@ public final class RequantizationRange extends RawOp {
    * @return a new instance of RequantizationRange
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> RequantizationRange create(Scope scope, Operand<T> input, Operand<TFloat32> inputMin, Operand<TFloat32> inputMax) {
+  public static <T extends Tensor> RequantizationRange create(Scope scope, Operand<T> input, Operand<TFloat32> inputMin, Operand<TFloat32> inputMax) {
     OperationBuilder opBuilder = scope.env().opBuilder("RequantizationRange", scope.makeOpName("RequantizationRange"));
     opBuilder.addInput(input.asOutput());
     opBuilder.addInput(inputMin.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/Requantize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/Requantize.java
index 5df8ca0b622..700ade2cf9b 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/Requantize.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/Requantize.java
@@ -22,12 +22,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.family.TType;
 
 /**
  * Converts the quantized `input` tensor into a lower-precision `output`.
@@ -43,7 +43,7 @@
  * @param <U> data type for {@code output()} output
  */
 @Operator(group = "quantization")
-public final class Requantize<U extends TType> extends RawOp {
+public final class Requantize<U extends Tensor> extends RawOp {
   
   /**
    * Factory method to create a class wrapping a new Requantize operation.
@@ -58,7 +58,7 @@ public final class Requantize<U extends TType> extends RawOp {
    * @return a new instance of Requantize
    */
   @Endpoint(describeByClass = true)
-  public static <U extends TType, T extends TType> Requantize<U> create(Scope scope, Operand<T> input, Operand<TFloat32> inputMin, Operand<TFloat32> inputMax, Operand<TFloat32> requestedOutputMin, Operand<TFloat32> requestedOutputMax, DataType<U> outType) {
+  public static <U extends Tensor, T extends Tensor> Requantize<U> create(Scope scope, Operand<T> input, Operand<TFloat32> inputMin, Operand<TFloat32> inputMax, Operand<TFloat32> requestedOutputMin, Operand<TFloat32> requestedOutputMax, DataType<U> outType) {
     OperationBuilder opBuilder = scope.env().opBuilder("Requantize", scope.makeOpName("Requantize"));
     opBuilder.addInput(input.asOutput());
     opBuilder.addInput(inputMin.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/ragged/RaggedGather.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/ragged/RaggedGather.java
index 7cc7de6d3db..d7d1d94d854 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/ragged/RaggedGather.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/ragged/RaggedGather.java
@@ -23,13 +23,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.Operands;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Gather ragged slices from `params` axis `0` according to `indices`.
@@ -65,7 +65,7 @@
  * @param <T> data type for {@code outputNestedSplits()} output
  * @param <U> data type for {@code outputDenseValues()} output
  */
-public final class RaggedGather<T extends TNumber, U extends TType> extends RawOp {
+public final class RaggedGather<T extends Tensor & TNumber, U extends Tensor> extends RawOp {
   
   /**
    * Factory method to create a class wrapping a new RaggedGather operation.
@@ -84,7 +84,7 @@ public final class RaggedGather<T extends TNumber, U extends TType> extends RawO
    * @return a new instance of RaggedGather
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber, U extends TType, V extends TNumber> RaggedGather<T, U> create(Scope scope, Iterable<Operand<T>> paramsNestedSplits, Operand<U> paramsDenseValues, Operand<V> indices, Long OUTPUTRAGGEDRANK) {
+  public static <T extends Tensor & TNumber, U extends Tensor, V extends Tensor & TNumber> RaggedGather<T, U> create(Scope scope, Iterable<Operand<T>> paramsNestedSplits, Operand<U> paramsDenseValues, Operand<V> indices, Long OUTPUTRAGGEDRANK) {
     OperationBuilder opBuilder = scope.env().opBuilder("RaggedGather", scope.makeOpName("RaggedGather"));
     opBuilder.addInputList(Operands.asOutputs(paramsNestedSplits));
     opBuilder.addInput(paramsDenseValues.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/ragged/RaggedRange.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/ragged/RaggedRange.java
index 9d8f5594fd9..c8c28878b19 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/ragged/RaggedRange.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/ragged/RaggedRange.java
@@ -22,13 +22,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Returns a `RaggedTensor` containing the specified sequences of numbers.
@@ -51,7 +51,7 @@
  * @param <U> data type for {@code rtNestedSplits()} output
  * @param <T> data type for {@code rtDenseValues()} output
  */
-public final class RaggedRange<U extends TNumber, T extends TNumber> extends RawOp {
+public final class RaggedRange<U extends Tensor & TNumber, T extends Tensor & TNumber> extends RawOp {
   
   /**
    * Factory method to create a class wrapping a new RaggedRange operation.
@@ -64,7 +64,7 @@ public final class RaggedRange<U extends TNumber, T extends TNumber> extends Raw
    * @return a new instance of RaggedRange
    */
   @Endpoint(describeByClass = true)
-  public static <U extends TNumber, T extends TNumber> RaggedRange<U, T> create(Scope scope, Operand<T> starts, Operand<T> limits, Operand<T> deltas, DataType<U> Tsplits) {
+  public static <U extends Tensor & TNumber, T extends Tensor & TNumber> RaggedRange<U, T> create(Scope scope, Operand<T> starts, Operand<T> limits, Operand<T> deltas, DataType<U> Tsplits) {
     OperationBuilder opBuilder = scope.env().opBuilder("RaggedRange", scope.makeOpName("RaggedRange"));
     opBuilder.addInput(starts.asOutput());
     opBuilder.addInput(limits.asOutput());
@@ -84,7 +84,7 @@ public static <U extends TNumber, T extends TNumber> RaggedRange<U, T> create(Sc
    * @return a new instance of RaggedRange
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> RaggedRange<TInt64, T> create(Scope scope, Operand<T> starts, Operand<T> limits, Operand<T> deltas) {
+  public static <T extends Tensor & TNumber> RaggedRange<TInt64, T> create(Scope scope, Operand<T> starts, Operand<T> limits, Operand<T> deltas) {
     return create(scope, starts, limits, deltas, TInt64.DTYPE);
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/ragged/RaggedTensorFromVariant.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/ragged/RaggedTensorFromVariant.java
index c9dcfe54bda..36d159044e6 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/ragged/RaggedTensorFromVariant.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/ragged/RaggedTensorFromVariant.java
@@ -24,13 +24,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Decodes a `variant` Tensor into a `RaggedTensor`.
@@ -51,7 +51,7 @@
  * @param <U> data type for {@code outputNestedSplits()} output
  * @param <T> data type for {@code outputDenseValues()} output
  */
-public final class RaggedTensorFromVariant<U extends TNumber, T extends TType> extends RawOp {
+public final class RaggedTensorFromVariant<U extends Tensor & TNumber, T extends Tensor> extends RawOp {
   
   /**
    * Factory method to create a class wrapping a new RaggedTensorFromVariant operation.
@@ -67,7 +67,7 @@ public final class RaggedTensorFromVariant<U extends TNumber, T extends TType> e
    * @return a new instance of RaggedTensorFromVariant
    */
   @Endpoint(describeByClass = true)
-  public static <U extends TNumber, T extends TType> RaggedTensorFromVariant<U, T> create(Scope scope, Operand<?> encodedRagged, Long inputRaggedRank, Long outputRaggedRank, DataType<T> Tvalues, DataType<U> Tsplits) {
+  public static <U extends Tensor & TNumber, T extends Tensor> RaggedTensorFromVariant<U, T> create(Scope scope, Operand<?> encodedRagged, Long inputRaggedRank, Long outputRaggedRank, DataType<T> Tvalues, DataType<U> Tsplits) {
     OperationBuilder opBuilder = scope.env().opBuilder("RaggedTensorFromVariant", scope.makeOpName("RaggedTensorFromVariant"));
     opBuilder.addInput(encodedRagged.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
@@ -91,7 +91,7 @@ public static <U extends TNumber, T extends TType> RaggedTensorFromVariant<U, T>
    * @return a new instance of RaggedTensorFromVariant
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> RaggedTensorFromVariant<TInt64, T> create(Scope scope, Operand<?> encodedRagged, Long inputRaggedRank, Long outputRaggedRank, DataType<T> Tvalues) {
+  public static <T extends Tensor> RaggedTensorFromVariant<TInt64, T> create(Scope scope, Operand<?> encodedRagged, Long inputRaggedRank, Long outputRaggedRank, DataType<T> Tvalues) {
     return create(scope, encodedRagged, inputRaggedRank, outputRaggedRank, Tvalues, TInt64.DTYPE);
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/ragged/RaggedTensorToSparse.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/ragged/RaggedTensorToSparse.java
index 7906048788d..33db48e4972 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/ragged/RaggedTensorToSparse.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/ragged/RaggedTensorToSparse.java
@@ -21,6 +21,7 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.Operands;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
@@ -28,7 +29,6 @@
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Converts a `RaggedTensor` into a `SparseTensor` with the same values.
@@ -39,7 +39,7 @@
  * 
  * @param <U> data type for {@code sparseValues()} output
  */
-public final class RaggedTensorToSparse<U extends TType> extends RawOp {
+public final class RaggedTensorToSparse<U extends Tensor> extends RawOp {
   
   /**
    * Factory method to create a class wrapping a new RaggedTensorToSparse operation.
@@ -50,7 +50,7 @@ public final class RaggedTensorToSparse<U extends TType> extends RawOp {
    * @return a new instance of RaggedTensorToSparse
    */
   @Endpoint(describeByClass = true)
-  public static <U extends TType, T extends TNumber> RaggedTensorToSparse<U> create(Scope scope, Iterable<Operand<T>> rtNestedSplits, Operand<U> rtDenseValues) {
+  public static <U extends Tensor, T extends Tensor & TNumber> RaggedTensorToSparse<U> create(Scope scope, Iterable<Operand<T>> rtNestedSplits, Operand<U> rtDenseValues) {
     OperationBuilder opBuilder = scope.env().opBuilder("RaggedTensorToSparse", scope.makeOpName("RaggedTensorToSparse"));
     opBuilder.addInputList(Operands.asOutputs(rtNestedSplits));
     opBuilder.addInput(rtDenseValues.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/ragged/RaggedTensorToTensor.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/ragged/RaggedTensorToTensor.java
index e469e6fc020..7b29dd7e156 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/ragged/RaggedTensorToTensor.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/ragged/RaggedTensorToTensor.java
@@ -22,13 +22,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.Operands;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Create a dense tensor from a ragged tensor, possibly altering its shape.
@@ -58,7 +58,7 @@
  * 
  * @param <U> data type for {@code result()} output
  */
-public final class RaggedTensorToTensor<U extends TType> extends RawOp implements Operand<U> {
+public final class RaggedTensorToTensor<U extends Tensor> extends RawOp implements Operand<U> {
   
   /**
    * Factory method to create a class wrapping a new RaggedTensorToTensor operation.
@@ -106,7 +106,7 @@ public final class RaggedTensorToTensor<U extends TType> extends RawOp implement
    * @return a new instance of RaggedTensorToTensor
    */
   @Endpoint(describeByClass = true)
-  public static <U extends TType, T extends TNumber, V extends TNumber> RaggedTensorToTensor<U> create(Scope scope, Operand<T> shape, Operand<U> values, Operand<U> defaultValue, Iterable<Operand<V>> rowPartitionTensors, List<String> rowPartitionTypes) {
+  public static <U extends Tensor, T extends Tensor & TNumber, V extends Tensor & TNumber> RaggedTensorToTensor<U> create(Scope scope, Operand<T> shape, Operand<U> values, Operand<U> defaultValue, Iterable<Operand<V>> rowPartitionTensors, List<String> rowPartitionTypes) {
     OperationBuilder opBuilder = scope.env().opBuilder("RaggedTensorToTensor", scope.makeOpName("RaggedTensorToTensor"));
     opBuilder.addInput(shape.asOutput());
     opBuilder.addInput(values.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/ragged/RaggedTensorToVariant.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/ragged/RaggedTensorToVariant.java
index 6c22a0c4529..761f25691b6 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/ragged/RaggedTensorToVariant.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/ragged/RaggedTensorToVariant.java
@@ -21,13 +21,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.Operands;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Encodes a `RaggedTensor` into a `variant` Tensor.
@@ -45,7 +45,7 @@
  * corresponding decoding logic.
  * 
  */
-public final class RaggedTensorToVariant extends RawOp implements Operand<TType> {
+public final class RaggedTensorToVariant extends RawOp implements Operand<Tensor> {
   
   /**
    * Factory method to create a class wrapping a new RaggedTensorToVariant operation.
@@ -58,7 +58,7 @@ public final class RaggedTensorToVariant extends RawOp implements Operand<TType>
    * @return a new instance of RaggedTensorToVariant
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber, U extends TType> RaggedTensorToVariant create(Scope scope, Iterable<Operand<T>> rtNestedSplits, Operand<U> rtDenseValues, Boolean batchedInput) {
+  public static <T extends Tensor & TNumber, U extends Tensor> RaggedTensorToVariant create(Scope scope, Iterable<Operand<T>> rtNestedSplits, Operand<U> rtDenseValues, Boolean batchedInput) {
     OperationBuilder opBuilder = scope.env().opBuilder("RaggedTensorToVariant", scope.makeOpName("RaggedTensorToVariant"));
     opBuilder.addInputList(Operands.asOutputs(rtNestedSplits));
     opBuilder.addInput(rtDenseValues.asOutput());
@@ -76,8 +76,8 @@ public Output<?> encodedRagged() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) encodedRagged;
+  public Output<Tensor> asOutput() {
+    return (Output<Tensor>) encodedRagged;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/Multinomial.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/Multinomial.java
index 58cc57d6c52..1d2d115ad87 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/Multinomial.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/Multinomial.java
@@ -22,6 +22,7 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -29,7 +30,6 @@
 import org.tensorflow.types.TInt32;
 import org.tensorflow.types.TInt64;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Draws samples from a multinomial distribution.
@@ -37,7 +37,7 @@
  * @param <U> data type for {@code output()} output
  */
 @Operator(group = "random")
-public final class Multinomial<U extends TNumber> extends RawOp implements Operand<U> {
+public final class Multinomial<U extends Tensor & TNumber> extends RawOp implements Operand<U> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.random.Multinomial}
@@ -80,7 +80,7 @@ private Options() {
    * @return a new instance of Multinomial
    */
   @Endpoint(describeByClass = true)
-  public static <U extends TNumber, T extends TNumber> Multinomial<U> create(Scope scope, Operand<T> logits, Operand<TInt32> numSamples, DataType<U> outputDtype, Options... options) {
+  public static <U extends Tensor & TNumber, T extends Tensor & TNumber> Multinomial<U> create(Scope scope, Operand<T> logits, Operand<TInt32> numSamples, DataType<U> outputDtype, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("Multinomial", scope.makeOpName("Multinomial"));
     opBuilder.addInput(logits.asOutput());
     opBuilder.addInput(numSamples.asOutput());
@@ -110,7 +110,7 @@ public static <U extends TNumber, T extends TNumber> Multinomial<U> create(Scope
    * @return a new instance of Multinomial
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> Multinomial<TInt64> create(Scope scope, Operand<T> logits, Operand<TInt32> numSamples, Options... options) {
+  public static <T extends Tensor & TNumber> Multinomial<TInt64> create(Scope scope, Operand<T> logits, Operand<TInt32> numSamples, Options... options) {
     return create(scope, logits, numSamples, TInt64.DTYPE, options);
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/NonDeterministicInts.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/NonDeterministicInts.java
index 246974eaf6f..48ad4545107 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/NonDeterministicInts.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/NonDeterministicInts.java
@@ -22,12 +22,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
 
 /**
  * Non-deterministically generates some integers.
@@ -36,7 +36,7 @@
  * 
  * @param <U> data type for {@code output()} output
  */
-public final class NonDeterministicInts<U extends TType> extends RawOp implements Operand<U> {
+public final class NonDeterministicInts<U extends Tensor> extends RawOp implements Operand<U> {
   
   /**
    * Factory method to create a class wrapping a new NonDeterministicInts operation.
@@ -47,7 +47,7 @@ public final class NonDeterministicInts<U extends TType> extends RawOp implement
    * @return a new instance of NonDeterministicInts
    */
   @Endpoint(describeByClass = true)
-  public static <U extends TType, T extends TType> NonDeterministicInts<U> create(Scope scope, Operand<T> shape, DataType<U> dtype) {
+  public static <U extends Tensor, T extends Tensor> NonDeterministicInts<U> create(Scope scope, Operand<T> shape, DataType<U> dtype) {
     OperationBuilder opBuilder = scope.env().opBuilder("NonDeterministicInts", scope.makeOpName("NonDeterministicInts"));
     opBuilder.addInput(shape.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
@@ -63,7 +63,7 @@ public static <U extends TType, T extends TType> NonDeterministicInts<U> create(
    * @return a new instance of NonDeterministicInts
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> NonDeterministicInts<TInt64> create(Scope scope, Operand<T> shape) {
+  public static <T extends Tensor> NonDeterministicInts<TInt64> create(Scope scope, Operand<T> shape) {
     return create(scope, shape, TInt64.DTYPE);
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/ParameterizedTruncatedNormal.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/ParameterizedTruncatedNormal.java
index 4be50b9cde0..0a8871772a9 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/ParameterizedTruncatedNormal.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/ParameterizedTruncatedNormal.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Outputs random values from a normal distribution. The parameters may each be a
@@ -37,7 +37,7 @@
  * @param <U> data type for {@code output()} output
  */
 @Operator(group = "random")
-public final class ParameterizedTruncatedNormal<U extends TNumber> extends RawOp implements Operand<U> {
+public final class ParameterizedTruncatedNormal<U extends Tensor & TNumber> extends RawOp implements Operand<U> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.random.ParameterizedTruncatedNormal}
@@ -83,7 +83,7 @@ private Options() {
    * @return a new instance of ParameterizedTruncatedNormal
    */
   @Endpoint(describeByClass = true)
-  public static <U extends TNumber, T extends TNumber> ParameterizedTruncatedNormal<U> create(Scope scope, Operand<T> shape, Operand<U> means, Operand<U> stdevs, Operand<U> minvals, Operand<U> maxvals, Options... options) {
+  public static <U extends Tensor & TNumber, T extends Tensor & TNumber> ParameterizedTruncatedNormal<U> create(Scope scope, Operand<T> shape, Operand<U> means, Operand<U> stdevs, Operand<U> minvals, Operand<U> maxvals, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ParameterizedTruncatedNormal", scope.makeOpName("ParameterizedTruncatedNormal"));
     opBuilder.addInput(shape.asOutput());
     opBuilder.addInput(means.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RandomGamma.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RandomGamma.java
index 13963e09ecb..47aaf11c1a4 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RandomGamma.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RandomGamma.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Outputs random values from the Gamma distribution(s) described by alpha.
@@ -38,7 +38,7 @@
  * @param <U> data type for {@code output()} output
  */
 @Operator(group = "random")
-public final class RandomGamma<U extends TNumber> extends RawOp implements Operand<U> {
+public final class RandomGamma<U extends Tensor & TNumber> extends RawOp implements Operand<U> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.random.RandomGamma}
@@ -82,7 +82,7 @@ private Options() {
    * @return a new instance of RandomGamma
    */
   @Endpoint(describeByClass = true)
-  public static <U extends TNumber, T extends TNumber> RandomGamma<U> create(Scope scope, Operand<T> shape, Operand<U> alpha, Options... options) {
+  public static <U extends Tensor & TNumber, T extends Tensor & TNumber> RandomGamma<U> create(Scope scope, Operand<T> shape, Operand<U> alpha, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("RandomGamma", scope.makeOpName("RandomGamma"));
     opBuilder.addInput(shape.asOutput());
     opBuilder.addInput(alpha.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RandomGammaGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RandomGammaGrad.java
index ce3798cef3f..d66bc32b835 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RandomGammaGrad.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RandomGammaGrad.java
@@ -21,19 +21,19 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes the derivative of a Gamma random sample w.r.t. `alpha`.
  * 
  * @param <T> data type for {@code output()} output
  */
-public final class RandomGammaGrad<T extends TNumber> extends RawOp implements Operand<T> {
+public final class RandomGammaGrad<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new RandomGammaGrad operation.
@@ -44,7 +44,7 @@ public final class RandomGammaGrad<T extends TNumber> extends RawOp implements O
    * @return a new instance of RandomGammaGrad
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> RandomGammaGrad<T> create(Scope scope, Operand<T> alpha, Operand<T> sample) {
+  public static <T extends Tensor & TNumber> RandomGammaGrad<T> create(Scope scope, Operand<T> alpha, Operand<T> sample) {
     OperationBuilder opBuilder = scope.env().opBuilder("RandomGammaGrad", scope.makeOpName("RandomGammaGrad"));
     opBuilder.addInput(alpha.asOutput());
     opBuilder.addInput(sample.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RandomPoisson.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RandomPoisson.java
index d4b516343ac..c112d5a0def 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RandomPoisson.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RandomPoisson.java
@@ -22,13 +22,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Outputs random values from the Poisson distribution(s) described by rate.
@@ -46,7 +46,7 @@
  * @param <V> data type for {@code output()} output
  */
 @Operator(group = "random")
-public final class RandomPoisson<V extends TNumber> extends RawOp implements Operand<V> {
+public final class RandomPoisson<V extends Tensor & TNumber> extends RawOp implements Operand<V> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.random.RandomPoisson}
@@ -91,7 +91,7 @@ private Options() {
    * @return a new instance of RandomPoisson
    */
   @Endpoint(describeByClass = true)
-  public static <V extends TNumber, T extends TNumber, U extends TNumber> RandomPoisson<V> create(Scope scope, Operand<T> shape, Operand<U> rate, DataType<V> dtype, Options... options) {
+  public static <V extends Tensor & TNumber, T extends Tensor & TNumber, U extends Tensor & TNumber> RandomPoisson<V> create(Scope scope, Operand<T> shape, Operand<U> rate, DataType<V> dtype, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("RandomPoissonV2", scope.makeOpName("RandomPoisson"));
     opBuilder.addInput(shape.asOutput());
     opBuilder.addInput(rate.asOutput());
@@ -122,7 +122,7 @@ public static <V extends TNumber, T extends TNumber, U extends TNumber> RandomPo
    * @return a new instance of RandomPoisson
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber, U extends TNumber> RandomPoisson<TInt64> create(Scope scope, Operand<T> shape, Operand<U> rate, Options... options) {
+  public static <T extends Tensor & TNumber, U extends Tensor & TNumber> RandomPoisson<TInt64> create(Scope scope, Operand<T> shape, Operand<U> rate, Options... options) {
     return create(scope, shape, rate, TInt64.DTYPE, options);
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RandomShuffle.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RandomShuffle.java
index b08c31031ec..afe6ca061a4 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RandomShuffle.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RandomShuffle.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Randomly shuffles a tensor along its first dimension.
@@ -43,7 +43,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "random")
-public final class RandomShuffle<T extends TType> extends RawOp implements Operand<T> {
+public final class RandomShuffle<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.random.RandomShuffle}
@@ -84,7 +84,7 @@ private Options() {
    * @return a new instance of RandomShuffle
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> RandomShuffle<T> create(Scope scope, Operand<T> value, Options... options) {
+  public static <T extends Tensor> RandomShuffle<T> create(Scope scope, Operand<T> value, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("RandomShuffle", scope.makeOpName("RandomShuffle"));
     opBuilder.addInput(value.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RandomStandardNormal.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RandomStandardNormal.java
index bdd971cc19d..041a1ac0adb 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RandomStandardNormal.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RandomStandardNormal.java
@@ -22,12 +22,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Outputs random values from a normal distribution.
@@ -37,7 +37,7 @@
  * @param <U> data type for {@code output()} output
  */
 @Operator(group = "random")
-public final class RandomStandardNormal<U extends TNumber> extends RawOp implements Operand<U> {
+public final class RandomStandardNormal<U extends Tensor & TNumber> extends RawOp implements Operand<U> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.random.RandomStandardNormal}
@@ -79,7 +79,7 @@ private Options() {
    * @return a new instance of RandomStandardNormal
    */
   @Endpoint(describeByClass = true)
-  public static <U extends TNumber, T extends TNumber> RandomStandardNormal<U> create(Scope scope, Operand<T> shape, DataType<U> dtype, Options... options) {
+  public static <U extends Tensor & TNumber, T extends Tensor & TNumber> RandomStandardNormal<U> create(Scope scope, Operand<T> shape, DataType<U> dtype, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("RandomStandardNormal", scope.makeOpName("RandomStandardNormal"));
     opBuilder.addInput(shape.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RandomUniform.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RandomUniform.java
index 5e42c2d9691..d82f0cda1bf 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RandomUniform.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RandomUniform.java
@@ -22,12 +22,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Outputs random values from a uniform distribution.
@@ -38,7 +38,7 @@
  * @param <U> data type for {@code output()} output
  */
 @Operator(group = "random")
-public final class RandomUniform<U extends TNumber> extends RawOp implements Operand<U> {
+public final class RandomUniform<U extends Tensor & TNumber> extends RawOp implements Operand<U> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.random.RandomUniform}
@@ -80,7 +80,7 @@ private Options() {
    * @return a new instance of RandomUniform
    */
   @Endpoint(describeByClass = true)
-  public static <U extends TNumber, T extends TNumber> RandomUniform<U> create(Scope scope, Operand<T> shape, DataType<U> dtype, Options... options) {
+  public static <U extends Tensor & TNumber, T extends Tensor & TNumber> RandomUniform<U> create(Scope scope, Operand<T> shape, DataType<U> dtype, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("RandomUniform", scope.makeOpName("RandomUniform"));
     opBuilder.addInput(shape.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RandomUniformInt.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RandomUniformInt.java
index 5232135ac1c..893b03fdfde 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RandomUniformInt.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RandomUniformInt.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Outputs random integers from a uniform distribution.
@@ -42,7 +42,7 @@
  * @param <U> data type for {@code output()} output
  */
 @Operator(group = "random")
-public final class RandomUniformInt<U extends TNumber> extends RawOp implements Operand<U> {
+public final class RandomUniformInt<U extends Tensor & TNumber> extends RawOp implements Operand<U> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.random.RandomUniformInt}
@@ -85,7 +85,7 @@ private Options() {
    * @return a new instance of RandomUniformInt
    */
   @Endpoint(describeByClass = true)
-  public static <U extends TNumber, T extends TNumber> RandomUniformInt<U> create(Scope scope, Operand<T> shape, Operand<U> minval, Operand<U> maxval, Options... options) {
+  public static <U extends Tensor & TNumber, T extends Tensor & TNumber> RandomUniformInt<U> create(Scope scope, Operand<T> shape, Operand<U> minval, Operand<U> maxval, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("RandomUniformInt", scope.makeOpName("RandomUniformInt"));
     opBuilder.addInput(shape.asOutput());
     opBuilder.addInput(minval.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatefulRandomBinomial.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatefulRandomBinomial.java
index b3c2dfce166..4a5539b1bb7 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatefulRandomBinomial.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatefulRandomBinomial.java
@@ -22,19 +22,19 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * @param <V> data type for {@code output()} output
  */
 @Operator(group = "random")
-public final class StatefulRandomBinomial<V extends TNumber> extends RawOp implements Operand<V> {
+public final class StatefulRandomBinomial<V extends Tensor & TNumber> extends RawOp implements Operand<V> {
   
   /**
    * Factory method to create a class wrapping a new StatefulRandomBinomial operation.
@@ -49,7 +49,7 @@ public final class StatefulRandomBinomial<V extends TNumber> extends RawOp imple
    * @return a new instance of StatefulRandomBinomial
    */
   @Endpoint(describeByClass = true)
-  public static <V extends TNumber, T extends TNumber, U extends TNumber> StatefulRandomBinomial<V> create(Scope scope, Operand<?> resource, Operand<TInt64> algorithm, Operand<T> shape, Operand<U> counts, Operand<U> probs, DataType<V> dtype) {
+  public static <V extends Tensor & TNumber, T extends Tensor & TNumber, U extends Tensor & TNumber> StatefulRandomBinomial<V> create(Scope scope, Operand<?> resource, Operand<TInt64> algorithm, Operand<T> shape, Operand<U> counts, Operand<U> probs, DataType<V> dtype) {
     OperationBuilder opBuilder = scope.env().opBuilder("StatefulRandomBinomial", scope.makeOpName("StatefulRandomBinomial"));
     opBuilder.addInput(resource.asOutput());
     opBuilder.addInput(algorithm.asOutput());
@@ -73,7 +73,7 @@ public static <V extends TNumber, T extends TNumber, U extends TNumber> Stateful
    * @return a new instance of StatefulRandomBinomial
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber, U extends TNumber> StatefulRandomBinomial<TInt64> create(Scope scope, Operand<?> resource, Operand<TInt64> algorithm, Operand<T> shape, Operand<U> counts, Operand<U> probs) {
+  public static <T extends Tensor & TNumber, U extends Tensor & TNumber> StatefulRandomBinomial<TInt64> create(Scope scope, Operand<?> resource, Operand<TInt64> algorithm, Operand<T> shape, Operand<U> counts, Operand<U> probs) {
     return create(scope, resource, algorithm, shape, counts, probs, TInt64.DTYPE);
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatefulStandardNormal.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatefulStandardNormal.java
index 12aabfa2a73..2db4f99804a 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatefulStandardNormal.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatefulStandardNormal.java
@@ -22,13 +22,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TFloat32;
 import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
 
 /**
  * Outputs random values from a normal distribution.
@@ -38,7 +38,7 @@
  * @param <U> data type for {@code output()} output
  */
 @Operator(group = "random")
-public final class StatefulStandardNormal<U extends TType> extends RawOp implements Operand<U> {
+public final class StatefulStandardNormal<U extends Tensor> extends RawOp implements Operand<U> {
   
   /**
    * Factory method to create a class wrapping a new StatefulStandardNormal operation.
@@ -51,7 +51,7 @@ public final class StatefulStandardNormal<U extends TType> extends RawOp impleme
    * @return a new instance of StatefulStandardNormal
    */
   @Endpoint(describeByClass = true)
-  public static <U extends TType, T extends TType> StatefulStandardNormal<U> create(Scope scope, Operand<?> resource, Operand<TInt64> algorithm, Operand<T> shape, DataType<U> dtype) {
+  public static <U extends Tensor, T extends Tensor> StatefulStandardNormal<U> create(Scope scope, Operand<?> resource, Operand<TInt64> algorithm, Operand<T> shape, DataType<U> dtype) {
     OperationBuilder opBuilder = scope.env().opBuilder("StatefulStandardNormalV2", scope.makeOpName("StatefulStandardNormal"));
     opBuilder.addInput(resource.asOutput());
     opBuilder.addInput(algorithm.asOutput());
@@ -71,7 +71,7 @@ public static <U extends TType, T extends TType> StatefulStandardNormal<U> creat
    * @return a new instance of StatefulStandardNormal
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> StatefulStandardNormal<TFloat32> create(Scope scope, Operand<?> resource, Operand<TInt64> algorithm, Operand<T> shape) {
+  public static <T extends Tensor> StatefulStandardNormal<TFloat32> create(Scope scope, Operand<?> resource, Operand<TInt64> algorithm, Operand<T> shape) {
     return create(scope, resource, algorithm, shape, TFloat32.DTYPE);
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatefulTruncatedNormal.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatefulTruncatedNormal.java
index 86904de711f..334d9de6b1a 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatefulTruncatedNormal.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatefulTruncatedNormal.java
@@ -22,13 +22,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TFloat32;
 import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
 
 /**
  * Outputs random values from a truncated normal distribution.
@@ -39,7 +39,7 @@
  * 
  * @param <U> data type for {@code output()} output
  */
-public final class StatefulTruncatedNormal<U extends TType> extends RawOp implements Operand<U> {
+public final class StatefulTruncatedNormal<U extends Tensor> extends RawOp implements Operand<U> {
   
   /**
    * Factory method to create a class wrapping a new StatefulTruncatedNormal operation.
@@ -52,7 +52,7 @@ public final class StatefulTruncatedNormal<U extends TType> extends RawOp implem
    * @return a new instance of StatefulTruncatedNormal
    */
   @Endpoint(describeByClass = true)
-  public static <U extends TType, T extends TType> StatefulTruncatedNormal<U> create(Scope scope, Operand<?> resource, Operand<TInt64> algorithm, Operand<T> shape, DataType<U> dtype) {
+  public static <U extends Tensor, T extends Tensor> StatefulTruncatedNormal<U> create(Scope scope, Operand<?> resource, Operand<TInt64> algorithm, Operand<T> shape, DataType<U> dtype) {
     OperationBuilder opBuilder = scope.env().opBuilder("StatefulTruncatedNormal", scope.makeOpName("StatefulTruncatedNormal"));
     opBuilder.addInput(resource.asOutput());
     opBuilder.addInput(algorithm.asOutput());
@@ -72,7 +72,7 @@ public static <U extends TType, T extends TType> StatefulTruncatedNormal<U> crea
    * @return a new instance of StatefulTruncatedNormal
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> StatefulTruncatedNormal<TFloat32> create(Scope scope, Operand<?> resource, Operand<TInt64> algorithm, Operand<T> shape) {
+  public static <T extends Tensor> StatefulTruncatedNormal<TFloat32> create(Scope scope, Operand<?> resource, Operand<TInt64> algorithm, Operand<T> shape) {
     return create(scope, resource, algorithm, shape, TFloat32.DTYPE);
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatefulUniform.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatefulUniform.java
index dd9a2c10af0..6ca09b116a3 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatefulUniform.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatefulUniform.java
@@ -22,13 +22,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TFloat32;
 import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
 
 /**
  * Outputs random values from a uniform distribution.
@@ -38,7 +38,7 @@
  * 
  * @param <U> data type for {@code output()} output
  */
-public final class StatefulUniform<U extends TType> extends RawOp implements Operand<U> {
+public final class StatefulUniform<U extends Tensor> extends RawOp implements Operand<U> {
   
   /**
    * Factory method to create a class wrapping a new StatefulUniform operation.
@@ -51,7 +51,7 @@ public final class StatefulUniform<U extends TType> extends RawOp implements Ope
    * @return a new instance of StatefulUniform
    */
   @Endpoint(describeByClass = true)
-  public static <U extends TType, T extends TType> StatefulUniform<U> create(Scope scope, Operand<?> resource, Operand<TInt64> algorithm, Operand<T> shape, DataType<U> dtype) {
+  public static <U extends Tensor, T extends Tensor> StatefulUniform<U> create(Scope scope, Operand<?> resource, Operand<TInt64> algorithm, Operand<T> shape, DataType<U> dtype) {
     OperationBuilder opBuilder = scope.env().opBuilder("StatefulUniform", scope.makeOpName("StatefulUniform"));
     opBuilder.addInput(resource.asOutput());
     opBuilder.addInput(algorithm.asOutput());
@@ -71,7 +71,7 @@ public static <U extends TType, T extends TType> StatefulUniform<U> create(Scope
    * @return a new instance of StatefulUniform
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> StatefulUniform<TFloat32> create(Scope scope, Operand<?> resource, Operand<TInt64> algorithm, Operand<T> shape) {
+  public static <T extends Tensor> StatefulUniform<TFloat32> create(Scope scope, Operand<?> resource, Operand<TInt64> algorithm, Operand<T> shape) {
     return create(scope, resource, algorithm, shape, TFloat32.DTYPE);
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatefulUniformFullInt.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatefulUniformFullInt.java
index 0bf991bd72e..3e3c4b5f0c5 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatefulUniformFullInt.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatefulUniformFullInt.java
@@ -22,12 +22,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
 
 /**
  * Outputs random integers from a uniform distribution.
@@ -36,7 +36,7 @@
  * 
  * @param <U> data type for {@code output()} output
  */
-public final class StatefulUniformFullInt<U extends TType> extends RawOp implements Operand<U> {
+public final class StatefulUniformFullInt<U extends Tensor> extends RawOp implements Operand<U> {
   
   /**
    * Factory method to create a class wrapping a new StatefulUniformFullInt operation.
@@ -49,7 +49,7 @@ public final class StatefulUniformFullInt<U extends TType> extends RawOp impleme
    * @return a new instance of StatefulUniformFullInt
    */
   @Endpoint(describeByClass = true)
-  public static <U extends TType, T extends TType> StatefulUniformFullInt<U> create(Scope scope, Operand<?> resource, Operand<TInt64> algorithm, Operand<T> shape, DataType<U> dtype) {
+  public static <U extends Tensor, T extends Tensor> StatefulUniformFullInt<U> create(Scope scope, Operand<?> resource, Operand<TInt64> algorithm, Operand<T> shape, DataType<U> dtype) {
     OperationBuilder opBuilder = scope.env().opBuilder("StatefulUniformFullInt", scope.makeOpName("StatefulUniformFullInt"));
     opBuilder.addInput(resource.asOutput());
     opBuilder.addInput(algorithm.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatefulUniformInt.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatefulUniformInt.java
index 74337edc44d..56fa0646b35 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatefulUniformInt.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatefulUniformInt.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
 
 /**
  * Outputs random integers from a uniform distribution.
@@ -41,7 +41,7 @@
  * 
  * @param <U> data type for {@code output()} output
  */
-public final class StatefulUniformInt<U extends TType> extends RawOp implements Operand<U> {
+public final class StatefulUniformInt<U extends Tensor> extends RawOp implements Operand<U> {
   
   /**
    * Factory method to create a class wrapping a new StatefulUniformInt operation.
@@ -55,7 +55,7 @@ public final class StatefulUniformInt<U extends TType> extends RawOp implements
    * @return a new instance of StatefulUniformInt
    */
   @Endpoint(describeByClass = true)
-  public static <U extends TType, T extends TType> StatefulUniformInt<U> create(Scope scope, Operand<?> resource, Operand<TInt64> algorithm, Operand<T> shape, Operand<U> minval, Operand<U> maxval) {
+  public static <U extends Tensor, T extends Tensor> StatefulUniformInt<U> create(Scope scope, Operand<?> resource, Operand<TInt64> algorithm, Operand<T> shape, Operand<U> minval, Operand<U> maxval) {
     OperationBuilder opBuilder = scope.env().opBuilder("StatefulUniformInt", scope.makeOpName("StatefulUniformInt"));
     opBuilder.addInput(resource.asOutput());
     opBuilder.addInput(algorithm.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessMultinomial.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessMultinomial.java
index 1dbe0e5a7d0..b9a23f53cfd 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessMultinomial.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessMultinomial.java
@@ -22,6 +22,7 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -29,7 +30,6 @@
 import org.tensorflow.types.TInt32;
 import org.tensorflow.types.TInt64;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Draws samples from a multinomial distribution.
@@ -37,7 +37,7 @@
  * @param <V> data type for {@code output()} output
  */
 @Operator(group = "random")
-public final class StatelessMultinomial<V extends TNumber> extends RawOp implements Operand<V> {
+public final class StatelessMultinomial<V extends Tensor & TNumber> extends RawOp implements Operand<V> {
   
   /**
    * Factory method to create a class wrapping a new StatelessMultinomial operation.
@@ -51,7 +51,7 @@ public final class StatelessMultinomial<V extends TNumber> extends RawOp impleme
    * @return a new instance of StatelessMultinomial
    */
   @Endpoint(describeByClass = true)
-  public static <V extends TNumber, T extends TNumber, U extends TNumber> StatelessMultinomial<V> create(Scope scope, Operand<T> logits, Operand<TInt32> numSamples, Operand<U> seed, DataType<V> outputDtype) {
+  public static <V extends Tensor & TNumber, T extends Tensor & TNumber, U extends Tensor & TNumber> StatelessMultinomial<V> create(Scope scope, Operand<T> logits, Operand<TInt32> numSamples, Operand<U> seed, DataType<V> outputDtype) {
     OperationBuilder opBuilder = scope.env().opBuilder("StatelessMultinomial", scope.makeOpName("StatelessMultinomial"));
     opBuilder.addInput(logits.asOutput());
     opBuilder.addInput(numSamples.asOutput());
@@ -72,7 +72,7 @@ public static <V extends TNumber, T extends TNumber, U extends TNumber> Stateles
    * @return a new instance of StatelessMultinomial
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber, U extends TNumber> StatelessMultinomial<TInt64> create(Scope scope, Operand<T> logits, Operand<TInt32> numSamples, Operand<U> seed) {
+  public static <T extends Tensor & TNumber, U extends Tensor & TNumber> StatelessMultinomial<TInt64> create(Scope scope, Operand<T> logits, Operand<TInt32> numSamples, Operand<U> seed) {
     return create(scope, logits, numSamples, seed, TInt64.DTYPE);
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessRandomBinomial.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessRandomBinomial.java
index 03543495413..cfbf62891ba 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessRandomBinomial.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessRandomBinomial.java
@@ -22,13 +22,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Outputs deterministic pseudorandom random numbers from a binomial distribution.
@@ -39,7 +39,7 @@
  * 
  * @param <W> data type for {@code output()} output
  */
-public final class StatelessRandomBinomial<W extends TNumber> extends RawOp implements Operand<W> {
+public final class StatelessRandomBinomial<W extends Tensor & TNumber> extends RawOp implements Operand<W> {
   
   /**
    * Factory method to create a class wrapping a new StatelessRandomBinomial operation.
@@ -55,7 +55,7 @@ public final class StatelessRandomBinomial<W extends TNumber> extends RawOp impl
    * @return a new instance of StatelessRandomBinomial
    */
   @Endpoint(describeByClass = true)
-  public static <W extends TNumber, T extends TNumber, U extends TNumber, V extends TNumber> StatelessRandomBinomial<W> create(Scope scope, Operand<T> shape, Operand<U> seed, Operand<V> counts, Operand<V> probs, DataType<W> dtype) {
+  public static <W extends Tensor & TNumber, T extends Tensor & TNumber, U extends Tensor & TNumber, V extends Tensor & TNumber> StatelessRandomBinomial<W> create(Scope scope, Operand<T> shape, Operand<U> seed, Operand<V> counts, Operand<V> probs, DataType<W> dtype) {
     OperationBuilder opBuilder = scope.env().opBuilder("StatelessRandomBinomial", scope.makeOpName("StatelessRandomBinomial"));
     opBuilder.addInput(shape.asOutput());
     opBuilder.addInput(seed.asOutput());
@@ -79,7 +79,7 @@ public static <W extends TNumber, T extends TNumber, U extends TNumber, V extend
    * @return a new instance of StatelessRandomBinomial
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber, U extends TNumber, V extends TNumber> StatelessRandomBinomial<TInt64> create(Scope scope, Operand<T> shape, Operand<U> seed, Operand<V> counts, Operand<V> probs) {
+  public static <T extends Tensor & TNumber, U extends Tensor & TNumber, V extends Tensor & TNumber> StatelessRandomBinomial<TInt64> create(Scope scope, Operand<T> shape, Operand<U> seed, Operand<V> counts, Operand<V> probs) {
     return create(scope, shape, seed, counts, probs, TInt64.DTYPE);
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessRandomGamma.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessRandomGamma.java
index 2c4eef75ff7..0aabc88dd01 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessRandomGamma.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessRandomGamma.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Outputs deterministic pseudorandom random numbers from a gamma distribution.
@@ -37,7 +37,7 @@
  * 
  * @param <V> data type for {@code output()} output
  */
-public final class StatelessRandomGamma<V extends TNumber> extends RawOp implements Operand<V> {
+public final class StatelessRandomGamma<V extends Tensor & TNumber> extends RawOp implements Operand<V> {
   
   /**
    * Factory method to create a class wrapping a new StatelessRandomGamma operation.
@@ -50,7 +50,7 @@ public final class StatelessRandomGamma<V extends TNumber> extends RawOp impleme
    * @return a new instance of StatelessRandomGamma
    */
   @Endpoint(describeByClass = true)
-  public static <V extends TNumber, T extends TNumber, U extends TNumber> StatelessRandomGamma<V> create(Scope scope, Operand<T> shape, Operand<U> seed, Operand<V> alpha) {
+  public static <V extends Tensor & TNumber, T extends Tensor & TNumber, U extends Tensor & TNumber> StatelessRandomGamma<V> create(Scope scope, Operand<T> shape, Operand<U> seed, Operand<V> alpha) {
     OperationBuilder opBuilder = scope.env().opBuilder("StatelessRandomGammaV2", scope.makeOpName("StatelessRandomGamma"));
     opBuilder.addInput(shape.asOutput());
     opBuilder.addInput(seed.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessRandomNormal.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessRandomNormal.java
index 07c5298cca0..ef2f6a1b70e 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessRandomNormal.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessRandomNormal.java
@@ -22,13 +22,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TFloat32;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Outputs deterministic pseudorandom values from a normal distribution.
@@ -40,7 +40,7 @@
  * @param <V> data type for {@code output()} output
  */
 @Operator(group = "random")
-public final class StatelessRandomNormal<V extends TNumber> extends RawOp implements Operand<V> {
+public final class StatelessRandomNormal<V extends Tensor & TNumber> extends RawOp implements Operand<V> {
   
   /**
    * Factory method to create a class wrapping a new StatelessRandomNormal operation.
@@ -52,7 +52,7 @@ public final class StatelessRandomNormal<V extends TNumber> extends RawOp implem
    * @return a new instance of StatelessRandomNormal
    */
   @Endpoint(describeByClass = true)
-  public static <V extends TNumber, T extends TNumber, U extends TNumber> StatelessRandomNormal<V> create(Scope scope, Operand<T> shape, Operand<U> seed, DataType<V> dtype) {
+  public static <V extends Tensor & TNumber, T extends Tensor & TNumber, U extends Tensor & TNumber> StatelessRandomNormal<V> create(Scope scope, Operand<T> shape, Operand<U> seed, DataType<V> dtype) {
     OperationBuilder opBuilder = scope.env().opBuilder("StatelessRandomNormal", scope.makeOpName("StatelessRandomNormal"));
     opBuilder.addInput(shape.asOutput());
     opBuilder.addInput(seed.asOutput());
@@ -70,7 +70,7 @@ public static <V extends TNumber, T extends TNumber, U extends TNumber> Stateles
    * @return a new instance of StatelessRandomNormal
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber, U extends TNumber> StatelessRandomNormal<TFloat32> create(Scope scope, Operand<T> shape, Operand<U> seed) {
+  public static <T extends Tensor & TNumber, U extends Tensor & TNumber> StatelessRandomNormal<TFloat32> create(Scope scope, Operand<T> shape, Operand<U> seed) {
     return create(scope, shape, seed, TFloat32.DTYPE);
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessRandomPoisson.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessRandomPoisson.java
index e71c70e2c1f..d4249b20c59 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessRandomPoisson.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessRandomPoisson.java
@@ -22,12 +22,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Outputs deterministic pseudorandom random numbers from a Poisson distribution.
@@ -38,7 +38,7 @@
  * 
  * @param <W> data type for {@code output()} output
  */
-public final class StatelessRandomPoisson<W extends TNumber> extends RawOp implements Operand<W> {
+public final class StatelessRandomPoisson<W extends Tensor & TNumber> extends RawOp implements Operand<W> {
   
   /**
    * Factory method to create a class wrapping a new StatelessRandomPoisson operation.
@@ -52,7 +52,7 @@ public final class StatelessRandomPoisson<W extends TNumber> extends RawOp imple
    * @return a new instance of StatelessRandomPoisson
    */
   @Endpoint(describeByClass = true)
-  public static <W extends TNumber, T extends TNumber, U extends TNumber, V extends TNumber> StatelessRandomPoisson<W> create(Scope scope, Operand<T> shape, Operand<U> seed, Operand<V> lam, DataType<W> dtype) {
+  public static <W extends Tensor & TNumber, T extends Tensor & TNumber, U extends Tensor & TNumber, V extends Tensor & TNumber> StatelessRandomPoisson<W> create(Scope scope, Operand<T> shape, Operand<U> seed, Operand<V> lam, DataType<W> dtype) {
     OperationBuilder opBuilder = scope.env().opBuilder("StatelessRandomPoisson", scope.makeOpName("StatelessRandomPoisson"));
     opBuilder.addInput(shape.asOutput());
     opBuilder.addInput(seed.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessRandomUniform.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessRandomUniform.java
index 9eb6edc67e5..dafea214cda 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessRandomUniform.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessRandomUniform.java
@@ -22,13 +22,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TFloat32;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Outputs deterministic pseudorandom random values from a uniform distribution.
@@ -41,7 +41,7 @@
  * @param <V> data type for {@code output()} output
  */
 @Operator(group = "random")
-public final class StatelessRandomUniform<V extends TNumber> extends RawOp implements Operand<V> {
+public final class StatelessRandomUniform<V extends Tensor & TNumber> extends RawOp implements Operand<V> {
   
   /**
    * Factory method to create a class wrapping a new StatelessRandomUniform operation.
@@ -53,7 +53,7 @@ public final class StatelessRandomUniform<V extends TNumber> extends RawOp imple
    * @return a new instance of StatelessRandomUniform
    */
   @Endpoint(describeByClass = true)
-  public static <V extends TNumber, T extends TNumber, U extends TNumber> StatelessRandomUniform<V> create(Scope scope, Operand<T> shape, Operand<U> seed, DataType<V> dtype) {
+  public static <V extends Tensor & TNumber, T extends Tensor & TNumber, U extends Tensor & TNumber> StatelessRandomUniform<V> create(Scope scope, Operand<T> shape, Operand<U> seed, DataType<V> dtype) {
     OperationBuilder opBuilder = scope.env().opBuilder("StatelessRandomUniform", scope.makeOpName("StatelessRandomUniform"));
     opBuilder.addInput(shape.asOutput());
     opBuilder.addInput(seed.asOutput());
@@ -71,7 +71,7 @@ public static <V extends TNumber, T extends TNumber, U extends TNumber> Stateles
    * @return a new instance of StatelessRandomUniform
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber, U extends TNumber> StatelessRandomUniform<TFloat32> create(Scope scope, Operand<T> shape, Operand<U> seed) {
+  public static <T extends Tensor & TNumber, U extends Tensor & TNumber> StatelessRandomUniform<TFloat32> create(Scope scope, Operand<T> shape, Operand<U> seed) {
     return create(scope, shape, seed, TFloat32.DTYPE);
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessRandomUniformFullInt.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessRandomUniformFullInt.java
index 291cce74d19..eff49a245ff 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessRandomUniformFullInt.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessRandomUniformFullInt.java
@@ -22,12 +22,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Outputs deterministic pseudorandom random integers from a uniform distribution.
@@ -38,7 +38,7 @@
  * 
  * @param <V> data type for {@code output()} output
  */
-public final class StatelessRandomUniformFullInt<V extends TNumber> extends RawOp implements Operand<V> {
+public final class StatelessRandomUniformFullInt<V extends Tensor & TNumber> extends RawOp implements Operand<V> {
   
   /**
    * Factory method to create a class wrapping a new StatelessRandomUniformFullInt operation.
@@ -50,7 +50,7 @@ public final class StatelessRandomUniformFullInt<V extends TNumber> extends RawO
    * @return a new instance of StatelessRandomUniformFullInt
    */
   @Endpoint(describeByClass = true)
-  public static <V extends TNumber, T extends TNumber, U extends TNumber> StatelessRandomUniformFullInt<V> create(Scope scope, Operand<T> shape, Operand<U> seed, DataType<V> dtype) {
+  public static <V extends Tensor & TNumber, T extends Tensor & TNumber, U extends Tensor & TNumber> StatelessRandomUniformFullInt<V> create(Scope scope, Operand<T> shape, Operand<U> seed, DataType<V> dtype) {
     OperationBuilder opBuilder = scope.env().opBuilder("StatelessRandomUniformFullInt", scope.makeOpName("StatelessRandomUniformFullInt"));
     opBuilder.addInput(shape.asOutput());
     opBuilder.addInput(seed.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessRandomUniformInt.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessRandomUniformInt.java
index 4695718a186..f916718ebe9 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessRandomUniformInt.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessRandomUniformInt.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Outputs deterministic pseudorandom random integers from a uniform distribution.
@@ -37,7 +37,7 @@
  * 
  * @param <V> data type for {@code output()} output
  */
-public final class StatelessRandomUniformInt<V extends TNumber> extends RawOp implements Operand<V> {
+public final class StatelessRandomUniformInt<V extends Tensor & TNumber> extends RawOp implements Operand<V> {
   
   /**
    * Factory method to create a class wrapping a new StatelessRandomUniformInt operation.
@@ -50,7 +50,7 @@ public final class StatelessRandomUniformInt<V extends TNumber> extends RawOp im
    * @return a new instance of StatelessRandomUniformInt
    */
   @Endpoint(describeByClass = true)
-  public static <V extends TNumber, T extends TNumber, U extends TNumber> StatelessRandomUniformInt<V> create(Scope scope, Operand<T> shape, Operand<U> seed, Operand<V> minval, Operand<V> maxval) {
+  public static <V extends Tensor & TNumber, T extends Tensor & TNumber, U extends Tensor & TNumber> StatelessRandomUniformInt<V> create(Scope scope, Operand<T> shape, Operand<U> seed, Operand<V> minval, Operand<V> maxval) {
     OperationBuilder opBuilder = scope.env().opBuilder("StatelessRandomUniformInt", scope.makeOpName("StatelessRandomUniformInt"));
     opBuilder.addInput(shape.asOutput());
     opBuilder.addInput(seed.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessTruncatedNormal.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessTruncatedNormal.java
index 3d4761fc4df..d9e8bb3ce05 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessTruncatedNormal.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessTruncatedNormal.java
@@ -22,13 +22,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TFloat32;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Outputs deterministic pseudorandom values from a truncated normal distribution.
@@ -42,7 +42,7 @@
  * @param <V> data type for {@code output()} output
  */
 @Operator(group = "random")
-public final class StatelessTruncatedNormal<V extends TNumber> extends RawOp implements Operand<V> {
+public final class StatelessTruncatedNormal<V extends Tensor & TNumber> extends RawOp implements Operand<V> {
   
   /**
    * Factory method to create a class wrapping a new StatelessTruncatedNormal operation.
@@ -54,7 +54,7 @@ public final class StatelessTruncatedNormal<V extends TNumber> extends RawOp imp
    * @return a new instance of StatelessTruncatedNormal
    */
   @Endpoint(describeByClass = true)
-  public static <V extends TNumber, T extends TNumber, U extends TNumber> StatelessTruncatedNormal<V> create(Scope scope, Operand<T> shape, Operand<U> seed, DataType<V> dtype) {
+  public static <V extends Tensor & TNumber, T extends Tensor & TNumber, U extends Tensor & TNumber> StatelessTruncatedNormal<V> create(Scope scope, Operand<T> shape, Operand<U> seed, DataType<V> dtype) {
     OperationBuilder opBuilder = scope.env().opBuilder("StatelessTruncatedNormal", scope.makeOpName("StatelessTruncatedNormal"));
     opBuilder.addInput(shape.asOutput());
     opBuilder.addInput(seed.asOutput());
@@ -72,7 +72,7 @@ public static <V extends TNumber, T extends TNumber, U extends TNumber> Stateles
    * @return a new instance of StatelessTruncatedNormal
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber, U extends TNumber> StatelessTruncatedNormal<TFloat32> create(Scope scope, Operand<T> shape, Operand<U> seed) {
+  public static <T extends Tensor & TNumber, U extends Tensor & TNumber> StatelessTruncatedNormal<TFloat32> create(Scope scope, Operand<T> shape, Operand<U> seed) {
     return create(scope, shape, seed, TFloat32.DTYPE);
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/TruncatedNormal.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/TruncatedNormal.java
index 7b88d720386..1004c261237 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/TruncatedNormal.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/TruncatedNormal.java
@@ -22,12 +22,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Outputs random values from a truncated normal distribution.
@@ -39,7 +39,7 @@
  * @param <U> data type for {@code output()} output
  */
 @Operator(group = "random")
-public final class TruncatedNormal<U extends TNumber> extends RawOp implements Operand<U> {
+public final class TruncatedNormal<U extends Tensor & TNumber> extends RawOp implements Operand<U> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.random.TruncatedNormal}
@@ -81,7 +81,7 @@ private Options() {
    * @return a new instance of TruncatedNormal
    */
   @Endpoint(describeByClass = true)
-  public static <U extends TNumber, T extends TNumber> TruncatedNormal<U> create(Scope scope, Operand<T> shape, DataType<U> dtype, Options... options) {
+  public static <U extends Tensor & TNumber, T extends Tensor & TNumber> TruncatedNormal<U> create(Scope scope, Operand<T> shape, DataType<U> dtype, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("TruncatedNormal", scope.makeOpName("TruncatedNormal"));
     opBuilder.addInput(shape.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/BatchFft.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/BatchFft.java
index 9eaacf4b26a..80b665f059d 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/BatchFft.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/BatchFft.java
@@ -21,16 +21,16 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  */
 @Operator(group = "signal")
-public final class BatchFft extends RawOp implements Operand<TType> {
+public final class BatchFft extends RawOp implements Operand<Tensor> {
   
   /**
    * Factory method to create a class wrapping a new BatchFft operation.
@@ -55,8 +55,8 @@ public Output<?> output() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) output;
+  public Output<Tensor> asOutput() {
+    return (Output<Tensor>) output;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/BatchFft2d.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/BatchFft2d.java
index 551d7c3f139..e8e48f76cd9 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/BatchFft2d.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/BatchFft2d.java
@@ -21,16 +21,16 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  */
 @Operator(group = "signal")
-public final class BatchFft2d extends RawOp implements Operand<TType> {
+public final class BatchFft2d extends RawOp implements Operand<Tensor> {
   
   /**
    * Factory method to create a class wrapping a new BatchFft2d operation.
@@ -55,8 +55,8 @@ public Output<?> output() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) output;
+  public Output<Tensor> asOutput() {
+    return (Output<Tensor>) output;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/BatchFft3d.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/BatchFft3d.java
index beeb97ee414..538689e7ea7 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/BatchFft3d.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/BatchFft3d.java
@@ -21,16 +21,16 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  */
 @Operator(group = "signal")
-public final class BatchFft3d extends RawOp implements Operand<TType> {
+public final class BatchFft3d extends RawOp implements Operand<Tensor> {
   
   /**
    * Factory method to create a class wrapping a new BatchFft3d operation.
@@ -55,8 +55,8 @@ public Output<?> output() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) output;
+  public Output<Tensor> asOutput() {
+    return (Output<Tensor>) output;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/BatchIfft.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/BatchIfft.java
index 262b02d2bc5..0f7529241fd 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/BatchIfft.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/BatchIfft.java
@@ -21,16 +21,16 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  */
 @Operator(group = "signal")
-public final class BatchIfft extends RawOp implements Operand<TType> {
+public final class BatchIfft extends RawOp implements Operand<Tensor> {
   
   /**
    * Factory method to create a class wrapping a new BatchIfft operation.
@@ -55,8 +55,8 @@ public Output<?> output() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) output;
+  public Output<Tensor> asOutput() {
+    return (Output<Tensor>) output;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/BatchIfft2d.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/BatchIfft2d.java
index 2145f201419..127dae4d1cd 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/BatchIfft2d.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/BatchIfft2d.java
@@ -21,16 +21,16 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  */
 @Operator(group = "signal")
-public final class BatchIfft2d extends RawOp implements Operand<TType> {
+public final class BatchIfft2d extends RawOp implements Operand<Tensor> {
   
   /**
    * Factory method to create a class wrapping a new BatchIfft2d operation.
@@ -55,8 +55,8 @@ public Output<?> output() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) output;
+  public Output<Tensor> asOutput() {
+    return (Output<Tensor>) output;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/BatchIfft3d.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/BatchIfft3d.java
index 6a271795308..3876316fd85 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/BatchIfft3d.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/BatchIfft3d.java
@@ -21,16 +21,16 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  */
 @Operator(group = "signal")
-public final class BatchIfft3d extends RawOp implements Operand<TType> {
+public final class BatchIfft3d extends RawOp implements Operand<Tensor> {
   
   /**
    * Factory method to create a class wrapping a new BatchIfft3d operation.
@@ -55,8 +55,8 @@ public Output<?> output() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) output;
+  public Output<Tensor> asOutput() {
+    return (Output<Tensor>) output;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Fft.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Fft.java
index 632500fad5e..6599e4b2c5e 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Fft.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Fft.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Fast Fourier transform.
@@ -36,7 +36,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "signal")
-public final class Fft<T extends TType> extends RawOp implements Operand<T> {
+public final class Fft<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Fft operation.
@@ -46,7 +46,7 @@ public final class Fft<T extends TType> extends RawOp implements Operand<T> {
    * @return a new instance of Fft
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> Fft<T> create(Scope scope, Operand<T> input) {
+  public static <T extends Tensor> Fft<T> create(Scope scope, Operand<T> input) {
     OperationBuilder opBuilder = scope.env().opBuilder("FFT", scope.makeOpName("Fft"));
     opBuilder.addInput(input.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Fft2d.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Fft2d.java
index 201b2e3c8af..d605fd8c944 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Fft2d.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Fft2d.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * 2D fast Fourier transform.
@@ -36,7 +36,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "signal")
-public final class Fft2d<T extends TType> extends RawOp implements Operand<T> {
+public final class Fft2d<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Fft2d operation.
@@ -46,7 +46,7 @@ public final class Fft2d<T extends TType> extends RawOp implements Operand<T> {
    * @return a new instance of Fft2d
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> Fft2d<T> create(Scope scope, Operand<T> input) {
+  public static <T extends Tensor> Fft2d<T> create(Scope scope, Operand<T> input) {
     OperationBuilder opBuilder = scope.env().opBuilder("FFT2D", scope.makeOpName("Fft2d"));
     opBuilder.addInput(input.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Fft3d.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Fft3d.java
index 7e840fbfb62..819d584bcd9 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Fft3d.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Fft3d.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * 3D fast Fourier transform.
@@ -36,7 +36,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "signal")
-public final class Fft3d<T extends TType> extends RawOp implements Operand<T> {
+public final class Fft3d<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Fft3d operation.
@@ -46,7 +46,7 @@ public final class Fft3d<T extends TType> extends RawOp implements Operand<T> {
    * @return a new instance of Fft3d
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> Fft3d<T> create(Scope scope, Operand<T> input) {
+  public static <T extends Tensor> Fft3d<T> create(Scope scope, Operand<T> input) {
     OperationBuilder opBuilder = scope.env().opBuilder("FFT3D", scope.makeOpName("Fft3d"));
     opBuilder.addInput(input.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Ifft.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Ifft.java
index c020956beec..62af7dc0af4 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Ifft.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Ifft.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Inverse fast Fourier transform.
@@ -36,7 +36,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "signal")
-public final class Ifft<T extends TType> extends RawOp implements Operand<T> {
+public final class Ifft<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Ifft operation.
@@ -46,7 +46,7 @@ public final class Ifft<T extends TType> extends RawOp implements Operand<T> {
    * @return a new instance of Ifft
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> Ifft<T> create(Scope scope, Operand<T> input) {
+  public static <T extends Tensor> Ifft<T> create(Scope scope, Operand<T> input) {
     OperationBuilder opBuilder = scope.env().opBuilder("IFFT", scope.makeOpName("Ifft"));
     opBuilder.addInput(input.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Ifft2d.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Ifft2d.java
index c8cac107ff4..03394267780 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Ifft2d.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Ifft2d.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Inverse 2D fast Fourier transform.
@@ -36,7 +36,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "signal")
-public final class Ifft2d<T extends TType> extends RawOp implements Operand<T> {
+public final class Ifft2d<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Ifft2d operation.
@@ -46,7 +46,7 @@ public final class Ifft2d<T extends TType> extends RawOp implements Operand<T> {
    * @return a new instance of Ifft2d
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> Ifft2d<T> create(Scope scope, Operand<T> input) {
+  public static <T extends Tensor> Ifft2d<T> create(Scope scope, Operand<T> input) {
     OperationBuilder opBuilder = scope.env().opBuilder("IFFT2D", scope.makeOpName("Ifft2d"));
     opBuilder.addInput(input.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Ifft3d.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Ifft3d.java
index ce7f87a9aa2..1c2eb56e45e 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Ifft3d.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Ifft3d.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Inverse 3D fast Fourier transform.
@@ -36,7 +36,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "signal")
-public final class Ifft3d<T extends TType> extends RawOp implements Operand<T> {
+public final class Ifft3d<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Ifft3d operation.
@@ -46,7 +46,7 @@ public final class Ifft3d<T extends TType> extends RawOp implements Operand<T> {
    * @return a new instance of Ifft3d
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> Ifft3d<T> create(Scope scope, Operand<T> input) {
+  public static <T extends Tensor> Ifft3d<T> create(Scope scope, Operand<T> input) {
     OperationBuilder opBuilder = scope.env().opBuilder("IFFT3D", scope.makeOpName("Ifft3d"));
     opBuilder.addInput(input.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Irfft.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Irfft.java
index 80d3bb85291..fb48c602bf7 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Irfft.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Irfft.java
@@ -22,6 +22,7 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -29,7 +30,6 @@
 import org.tensorflow.types.TFloat32;
 import org.tensorflow.types.TInt32;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Inverse real-valued fast Fourier transform.
@@ -51,7 +51,7 @@
  * @param <U> data type for {@code output()} output
  */
 @Operator(group = "signal")
-public final class Irfft<U extends TNumber> extends RawOp implements Operand<U> {
+public final class Irfft<U extends Tensor & TNumber> extends RawOp implements Operand<U> {
   
   /**
    * Factory method to create a class wrapping a new Irfft operation.
@@ -63,7 +63,7 @@ public final class Irfft<U extends TNumber> extends RawOp implements Operand<U>
    * @return a new instance of Irfft
    */
   @Endpoint(describeByClass = true)
-  public static <U extends TNumber, T extends TType> Irfft<U> create(Scope scope, Operand<T> input, Operand<TInt32> fftLength, DataType<U> Treal) {
+  public static <U extends Tensor & TNumber, T extends Tensor> Irfft<U> create(Scope scope, Operand<T> input, Operand<TInt32> fftLength, DataType<U> Treal) {
     OperationBuilder opBuilder = scope.env().opBuilder("IRFFT", scope.makeOpName("Irfft"));
     opBuilder.addInput(input.asOutput());
     opBuilder.addInput(fftLength.asOutput());
@@ -81,7 +81,7 @@ public static <U extends TNumber, T extends TType> Irfft<U> create(Scope scope,
    * @return a new instance of Irfft
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> Irfft<TFloat32> create(Scope scope, Operand<T> input, Operand<TInt32> fftLength) {
+  public static <T extends Tensor> Irfft<TFloat32> create(Scope scope, Operand<T> input, Operand<TInt32> fftLength) {
     return create(scope, input, fftLength, TFloat32.DTYPE);
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Irfft2d.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Irfft2d.java
index 8acf23a4f23..2b525b692ad 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Irfft2d.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Irfft2d.java
@@ -22,6 +22,7 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -29,7 +30,6 @@
 import org.tensorflow.types.TFloat32;
 import org.tensorflow.types.TInt32;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Inverse 2D real-valued fast Fourier transform.
@@ -52,7 +52,7 @@
  * @param <U> data type for {@code output()} output
  */
 @Operator(group = "signal")
-public final class Irfft2d<U extends TNumber> extends RawOp implements Operand<U> {
+public final class Irfft2d<U extends Tensor & TNumber> extends RawOp implements Operand<U> {
   
   /**
    * Factory method to create a class wrapping a new Irfft2d operation.
@@ -64,7 +64,7 @@ public final class Irfft2d<U extends TNumber> extends RawOp implements Operand<U
    * @return a new instance of Irfft2d
    */
   @Endpoint(describeByClass = true)
-  public static <U extends TNumber, T extends TType> Irfft2d<U> create(Scope scope, Operand<T> input, Operand<TInt32> fftLength, DataType<U> Treal) {
+  public static <U extends Tensor & TNumber, T extends Tensor> Irfft2d<U> create(Scope scope, Operand<T> input, Operand<TInt32> fftLength, DataType<U> Treal) {
     OperationBuilder opBuilder = scope.env().opBuilder("IRFFT2D", scope.makeOpName("Irfft2d"));
     opBuilder.addInput(input.asOutput());
     opBuilder.addInput(fftLength.asOutput());
@@ -82,7 +82,7 @@ public static <U extends TNumber, T extends TType> Irfft2d<U> create(Scope scope
    * @return a new instance of Irfft2d
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> Irfft2d<TFloat32> create(Scope scope, Operand<T> input, Operand<TInt32> fftLength) {
+  public static <T extends Tensor> Irfft2d<TFloat32> create(Scope scope, Operand<T> input, Operand<TInt32> fftLength) {
     return create(scope, input, fftLength, TFloat32.DTYPE);
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Irfft3d.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Irfft3d.java
index c7b9efabfd2..a36fc387339 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Irfft3d.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Irfft3d.java
@@ -22,6 +22,7 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -29,7 +30,6 @@
 import org.tensorflow.types.TFloat32;
 import org.tensorflow.types.TInt32;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Inverse 3D real-valued fast Fourier transform.
@@ -52,7 +52,7 @@
  * @param <U> data type for {@code output()} output
  */
 @Operator(group = "signal")
-public final class Irfft3d<U extends TNumber> extends RawOp implements Operand<U> {
+public final class Irfft3d<U extends Tensor & TNumber> extends RawOp implements Operand<U> {
   
   /**
    * Factory method to create a class wrapping a new Irfft3d operation.
@@ -64,7 +64,7 @@ public final class Irfft3d<U extends TNumber> extends RawOp implements Operand<U
    * @return a new instance of Irfft3d
    */
   @Endpoint(describeByClass = true)
-  public static <U extends TNumber, T extends TType> Irfft3d<U> create(Scope scope, Operand<T> input, Operand<TInt32> fftLength, DataType<U> Treal) {
+  public static <U extends Tensor & TNumber, T extends Tensor> Irfft3d<U> create(Scope scope, Operand<T> input, Operand<TInt32> fftLength, DataType<U> Treal) {
     OperationBuilder opBuilder = scope.env().opBuilder("IRFFT3D", scope.makeOpName("Irfft3d"));
     opBuilder.addInput(input.asOutput());
     opBuilder.addInput(fftLength.asOutput());
@@ -82,7 +82,7 @@ public static <U extends TNumber, T extends TType> Irfft3d<U> create(Scope scope
    * @return a new instance of Irfft3d
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> Irfft3d<TFloat32> create(Scope scope, Operand<T> input, Operand<TInt32> fftLength) {
+  public static <T extends Tensor> Irfft3d<TFloat32> create(Scope scope, Operand<T> input, Operand<TInt32> fftLength) {
     return create(scope, input, fftLength, TFloat32.DTYPE);
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Rfft.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Rfft.java
index 9764bcbf0f2..c70d871ac9a 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Rfft.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Rfft.java
@@ -22,13 +22,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Real-valued fast Fourier transform.
@@ -47,7 +47,7 @@
  * @param <U> data type for {@code output()} output
  */
 @Operator(group = "signal")
-public final class Rfft<U extends TType> extends RawOp implements Operand<U> {
+public final class Rfft<U extends Tensor> extends RawOp implements Operand<U> {
   
   /**
    * Factory method to create a class wrapping a new Rfft operation.
@@ -59,7 +59,7 @@ public final class Rfft<U extends TType> extends RawOp implements Operand<U> {
    * @return a new instance of Rfft
    */
   @Endpoint(describeByClass = true)
-  public static <U extends TType, T extends TNumber> Rfft<U> create(Scope scope, Operand<T> input, Operand<TInt32> fftLength, DataType<U> Tcomplex) {
+  public static <U extends Tensor, T extends Tensor & TNumber> Rfft<U> create(Scope scope, Operand<T> input, Operand<TInt32> fftLength, DataType<U> Tcomplex) {
     OperationBuilder opBuilder = scope.env().opBuilder("RFFT", scope.makeOpName("Rfft"));
     opBuilder.addInput(input.asOutput());
     opBuilder.addInput(fftLength.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Rfft2d.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Rfft2d.java
index 91187dced7b..32fae79f26c 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Rfft2d.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Rfft2d.java
@@ -22,13 +22,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * 2D real-valued fast Fourier transform.
@@ -48,7 +48,7 @@
  * @param <U> data type for {@code output()} output
  */
 @Operator(group = "signal")
-public final class Rfft2d<U extends TType> extends RawOp implements Operand<U> {
+public final class Rfft2d<U extends Tensor> extends RawOp implements Operand<U> {
   
   /**
    * Factory method to create a class wrapping a new Rfft2d operation.
@@ -60,7 +60,7 @@ public final class Rfft2d<U extends TType> extends RawOp implements Operand<U> {
    * @return a new instance of Rfft2d
    */
   @Endpoint(describeByClass = true)
-  public static <U extends TType, T extends TNumber> Rfft2d<U> create(Scope scope, Operand<T> input, Operand<TInt32> fftLength, DataType<U> Tcomplex) {
+  public static <U extends Tensor, T extends Tensor & TNumber> Rfft2d<U> create(Scope scope, Operand<T> input, Operand<TInt32> fftLength, DataType<U> Tcomplex) {
     OperationBuilder opBuilder = scope.env().opBuilder("RFFT2D", scope.makeOpName("Rfft2d"));
     opBuilder.addInput(input.asOutput());
     opBuilder.addInput(fftLength.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Rfft3d.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Rfft3d.java
index 1eb113e9cf1..5524ec596c4 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Rfft3d.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Rfft3d.java
@@ -22,13 +22,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * 3D real-valued fast Fourier transform.
@@ -48,7 +48,7 @@
  * @param <U> data type for {@code output()} output
  */
 @Operator(group = "signal")
-public final class Rfft3d<U extends TType> extends RawOp implements Operand<U> {
+public final class Rfft3d<U extends Tensor> extends RawOp implements Operand<U> {
   
   /**
    * Factory method to create a class wrapping a new Rfft3d operation.
@@ -60,7 +60,7 @@ public final class Rfft3d<U extends TType> extends RawOp implements Operand<U> {
    * @return a new instance of Rfft3d
    */
   @Endpoint(describeByClass = true)
-  public static <U extends TType, T extends TNumber> Rfft3d<U> create(Scope scope, Operand<T> input, Operand<TInt32> fftLength, DataType<U> Tcomplex) {
+  public static <U extends Tensor, T extends Tensor & TNumber> Rfft3d<U> create(Scope scope, Operand<T> input, Operand<TInt32> fftLength, DataType<U> Tcomplex) {
     OperationBuilder opBuilder = scope.env().opBuilder("RFFT3D", scope.makeOpName("Rfft3d"));
     opBuilder.addInput(input.asOutput());
     opBuilder.addInput(fftLength.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/AddManySparseToTensorsMap.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/AddManySparseToTensorsMap.java
index ec66167f7e6..6a49325a74b 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/AddManySparseToTensorsMap.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/AddManySparseToTensorsMap.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
 
 /**
  * Add an `N`-minibatch `SparseTensor` to a `SparseTensorsMap`, return `N` handles.
@@ -98,7 +98,7 @@ private Options() {
    * @return a new instance of AddManySparseToTensorsMap
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> AddManySparseToTensorsMap create(Scope scope, Operand<TInt64> sparseIndices, Operand<T> sparseValues, Operand<TInt64> sparseShape, Options... options) {
+  public static <T extends Tensor> AddManySparseToTensorsMap create(Scope scope, Operand<TInt64> sparseIndices, Operand<T> sparseValues, Operand<TInt64> sparseShape, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("AddManySparseToTensorsMap", scope.makeOpName("AddManySparseToTensorsMap"));
     opBuilder.addInput(sparseIndices.asOutput());
     opBuilder.addInput(sparseValues.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/AddSparseToTensorsMap.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/AddSparseToTensorsMap.java
index 153e326aa9d..b8b14b04d91 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/AddSparseToTensorsMap.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/AddSparseToTensorsMap.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
 
 /**
  * Add a `SparseTensor` to a `SparseTensorsMap` return its handle.
@@ -89,7 +89,7 @@ private Options() {
    * @return a new instance of AddSparseToTensorsMap
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> AddSparseToTensorsMap create(Scope scope, Operand<TInt64> sparseIndices, Operand<T> sparseValues, Operand<TInt64> sparseShape, Options... options) {
+  public static <T extends Tensor> AddSparseToTensorsMap create(Scope scope, Operand<TInt64> sparseIndices, Operand<T> sparseValues, Operand<TInt64> sparseShape, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("AddSparseToTensorsMap", scope.makeOpName("AddSparseToTensorsMap"));
     opBuilder.addInput(sparseIndices.asOutput());
     opBuilder.addInput(sparseValues.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/DenseToDenseSetOperation.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/DenseToDenseSetOperation.java
index 15e62be4fa8..d8b50867170 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/DenseToDenseSetOperation.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/DenseToDenseSetOperation.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
 
 /**
  * Applies set operation along last dimension of 2 `Tensor` inputs.
@@ -42,7 +42,7 @@
  * @param <T> data type for {@code resultValues()} output
  */
 @Operator(group = "sparse")
-public final class DenseToDenseSetOperation<T extends TType> extends RawOp {
+public final class DenseToDenseSetOperation<T extends Tensor> extends RawOp {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.sparse.DenseToDenseSetOperation}
@@ -76,7 +76,7 @@ private Options() {
    * @return a new instance of DenseToDenseSetOperation
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> DenseToDenseSetOperation<T> create(Scope scope, Operand<T> set1, Operand<T> set2, String setOperation, Options... options) {
+  public static <T extends Tensor> DenseToDenseSetOperation<T> create(Scope scope, Operand<T> set1, Operand<T> set2, String setOperation, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("DenseToDenseSetOperation", scope.makeOpName("DenseToDenseSetOperation"));
     opBuilder.addInput(set1.asOutput());
     opBuilder.addInput(set2.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/DenseToSparseSetOperation.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/DenseToSparseSetOperation.java
index 107f32524ea..bc83ee8e6b5 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/DenseToSparseSetOperation.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/DenseToSparseSetOperation.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
 
 /**
  * Applies set operation along last dimension of `Tensor` and `SparseTensor`.
@@ -50,7 +50,7 @@
  * @param <T> data type for {@code resultValues()} output
  */
 @Operator(group = "sparse")
-public final class DenseToSparseSetOperation<T extends TType> extends RawOp {
+public final class DenseToSparseSetOperation<T extends Tensor> extends RawOp {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.sparse.DenseToSparseSetOperation}
@@ -89,7 +89,7 @@ private Options() {
    * @return a new instance of DenseToSparseSetOperation
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> DenseToSparseSetOperation<T> create(Scope scope, Operand<T> set1, Operand<TInt64> set2Indices, Operand<T> set2Values, Operand<TInt64> set2Shape, String setOperation, Options... options) {
+  public static <T extends Tensor> DenseToSparseSetOperation<T> create(Scope scope, Operand<T> set1, Operand<TInt64> set2Indices, Operand<T> set2Values, Operand<TInt64> set2Shape, String setOperation, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("DenseToSparseSetOperation", scope.makeOpName("DenseToSparseSetOperation"));
     opBuilder.addInput(set1.asOutput());
     opBuilder.addInput(set2Indices.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/DeserializeSparse.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/DeserializeSparse.java
index a1a6bb2fa49..33c0f45fb6a 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/DeserializeSparse.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/DeserializeSparse.java
@@ -22,12 +22,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
 
 /**
  * Deserialize `SparseTensor` objects.
@@ -77,7 +77,7 @@
  * @param <U> data type for {@code sparseValues()} output
  */
 @Operator(group = "sparse")
-public final class DeserializeSparse<U extends TType> extends RawOp {
+public final class DeserializeSparse<U extends Tensor> extends RawOp {
   
   /**
    * Factory method to create a class wrapping a new DeserializeSparse operation.
@@ -89,7 +89,7 @@ public final class DeserializeSparse<U extends TType> extends RawOp {
    * @return a new instance of DeserializeSparse
    */
   @Endpoint(describeByClass = true)
-  public static <U extends TType, T extends TType> DeserializeSparse<U> create(Scope scope, Operand<T> serializedSparse, DataType<U> dtype) {
+  public static <U extends Tensor, T extends Tensor> DeserializeSparse<U> create(Scope scope, Operand<T> serializedSparse, DataType<U> dtype) {
     OperationBuilder opBuilder = scope.env().opBuilder("DeserializeSparse", scope.makeOpName("DeserializeSparse"));
     opBuilder.addInput(serializedSparse.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseAccumulatorApplyGradient.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseAccumulatorApplyGradient.java
index 328fe0c49ea..6cedd5f287f 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseAccumulatorApplyGradient.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseAccumulatorApplyGradient.java
@@ -20,13 +20,13 @@
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
 import org.tensorflow.types.TString;
-import org.tensorflow.types.family.TType;
 
 /**
  * Applies a sparse gradient to a given accumulator.
@@ -54,7 +54,7 @@ public final class SparseAccumulatorApplyGradient extends RawOp {
    * @return a new instance of SparseAccumulatorApplyGradient
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> SparseAccumulatorApplyGradient create(Scope scope, Operand<TString> handle, Operand<TInt64> localStep, Operand<TInt64> gradientIndices, Operand<T> gradientValues, Operand<TInt64> gradientShape, Boolean hasKnownShape) {
+  public static <T extends Tensor> SparseAccumulatorApplyGradient create(Scope scope, Operand<TString> handle, Operand<TInt64> localStep, Operand<TInt64> gradientIndices, Operand<T> gradientValues, Operand<TInt64> gradientShape, Boolean hasKnownShape) {
     OperationBuilder opBuilder = scope.env().opBuilder("SparseAccumulatorApplyGradient", scope.makeOpName("SparseAccumulatorApplyGradient"));
     opBuilder.addInput(handle.asOutput());
     opBuilder.addInput(localStep.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseAccumulatorTakeGradient.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseAccumulatorTakeGradient.java
index 87cbd112e57..d226209221a 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseAccumulatorTakeGradient.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseAccumulatorTakeGradient.java
@@ -22,6 +22,7 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -29,7 +30,6 @@
 import org.tensorflow.types.TInt32;
 import org.tensorflow.types.TInt64;
 import org.tensorflow.types.TString;
-import org.tensorflow.types.family.TType;
 
 /**
  * Extracts the average sparse gradient in a SparseConditionalAccumulator.
@@ -44,7 +44,7 @@
  * @param <T> data type for {@code values()} output
  */
 @Operator(group = "sparse")
-public final class SparseAccumulatorTakeGradient<T extends TType> extends RawOp {
+public final class SparseAccumulatorTakeGradient<T extends Tensor> extends RawOp {
   
   /**
    * Factory method to create a class wrapping a new SparseAccumulatorTakeGradient operation.
@@ -57,7 +57,7 @@ public final class SparseAccumulatorTakeGradient<T extends TType> extends RawOp
    * @return a new instance of SparseAccumulatorTakeGradient
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> SparseAccumulatorTakeGradient<T> create(Scope scope, Operand<TString> handle, Operand<TInt32> numRequired, DataType<T> dtype) {
+  public static <T extends Tensor> SparseAccumulatorTakeGradient<T> create(Scope scope, Operand<TString> handle, Operand<TInt32> numRequired, DataType<T> dtype) {
     OperationBuilder opBuilder = scope.env().opBuilder("SparseAccumulatorTakeGradient", scope.makeOpName("SparseAccumulatorTakeGradient"));
     opBuilder.addInput(handle.asOutput());
     opBuilder.addInput(numRequired.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseAdd.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseAdd.java
index bc8beb287bb..5882e508deb 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseAdd.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseAdd.java
@@ -21,13 +21,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Adds two `SparseTensor` objects to produce another `SparseTensor`.
@@ -49,7 +49,7 @@
  * @param <T> data type for {@code sumValues()} output
  */
 @Operator(group = "sparse")
-public final class SparseAdd<T extends TType> extends RawOp {
+public final class SparseAdd<T extends Tensor> extends RawOp {
   
   /**
    * Factory method to create a class wrapping a new SparseAdd operation.
@@ -66,7 +66,7 @@ public final class SparseAdd<T extends TType> extends RawOp {
    * @return a new instance of SparseAdd
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> SparseAdd<T> create(Scope scope, Operand<TInt64> aIndices, Operand<T> aValues, Operand<TInt64> aShape, Operand<TInt64> bIndices, Operand<T> bValues, Operand<TInt64> bShape, Operand<U> thresh) {
+  public static <T extends Tensor, U extends Tensor & TNumber> SparseAdd<T> create(Scope scope, Operand<TInt64> aIndices, Operand<T> aValues, Operand<TInt64> aShape, Operand<TInt64> bIndices, Operand<T> bValues, Operand<TInt64> bShape, Operand<U> thresh) {
     OperationBuilder opBuilder = scope.env().opBuilder("SparseAdd", scope.makeOpName("SparseAdd"));
     opBuilder.addInput(aIndices.asOutput());
     opBuilder.addInput(aValues.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseAddGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseAddGrad.java
index c3fdf2952a9..b13a206294d 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseAddGrad.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseAddGrad.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
 
 /**
  * The gradient operator for the SparseAdd op.
@@ -39,7 +39,7 @@
  * @param <T> data type for {@code aValGrad()} output
  */
 @Operator(group = "sparse")
-public final class SparseAddGrad<T extends TType> extends RawOp {
+public final class SparseAddGrad<T extends Tensor> extends RawOp {
   
   /**
    * Factory method to create a class wrapping a new SparseAddGrad operation.
@@ -54,7 +54,7 @@ public final class SparseAddGrad<T extends TType> extends RawOp {
    * @return a new instance of SparseAddGrad
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> SparseAddGrad<T> create(Scope scope, Operand<T> backpropValGrad, Operand<TInt64> aIndices, Operand<TInt64> bIndices, Operand<TInt64> sumIndices) {
+  public static <T extends Tensor> SparseAddGrad<T> create(Scope scope, Operand<T> backpropValGrad, Operand<TInt64> aIndices, Operand<TInt64> bIndices, Operand<TInt64> sumIndices) {
     OperationBuilder opBuilder = scope.env().opBuilder("SparseAddGrad", scope.makeOpName("SparseAddGrad"));
     opBuilder.addInput(backpropValGrad.asOutput());
     opBuilder.addInput(aIndices.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseConcat.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseConcat.java
index c6866997e30..b61b2a0a834 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseConcat.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseConcat.java
@@ -21,13 +21,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.Operands;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
 
 /**
  * Concatenates a list of `SparseTensor` along the specified dimension.
@@ -77,7 +77,7 @@
  * @param <T> data type for {@code outputValues()} output
  */
 @Operator(group = "sparse")
-public final class SparseConcat<T extends TType> extends RawOp {
+public final class SparseConcat<T extends Tensor> extends RawOp {
   
   /**
    * Factory method to create a class wrapping a new SparseConcat operation.
@@ -91,7 +91,7 @@ public final class SparseConcat<T extends TType> extends RawOp {
    * @return a new instance of SparseConcat
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> SparseConcat<T> create(Scope scope, Iterable<Operand<TInt64>> indices, Iterable<Operand<T>> values, Iterable<Operand<TInt64>> shapes, Long concatDim) {
+  public static <T extends Tensor> SparseConcat<T> create(Scope scope, Iterable<Operand<TInt64>> indices, Iterable<Operand<T>> values, Iterable<Operand<TInt64>> shapes, Long concatDim) {
     OperationBuilder opBuilder = scope.env().opBuilder("SparseConcat", scope.makeOpName("SparseConcat"));
     opBuilder.addInputList(Operands.asOutputs(indices));
     opBuilder.addInputList(Operands.asOutputs(values));
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseConditionalAccumulator.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseConditionalAccumulator.java
index b38fc0a6c46..4a7a5ed2ce2 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseConditionalAccumulator.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseConditionalAccumulator.java
@@ -22,13 +22,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TString;
-import org.tensorflow.types.family.TType;
 
 /**
  * A conditional accumulator for aggregating sparse gradients.
@@ -92,7 +92,7 @@ private Options() {
    * @return a new instance of SparseConditionalAccumulator
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> SparseConditionalAccumulator create(Scope scope, DataType<T> dtype, Shape shape, Options... options) {
+  public static <T extends Tensor> SparseConditionalAccumulator create(Scope scope, DataType<T> dtype, Shape shape, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("SparseConditionalAccumulator", scope.makeOpName("SparseConditionalAccumulator"));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("dtype", dtype);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseCross.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseCross.java
index 1cd471349c2..dfc030fc809 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseCross.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseCross.java
@@ -17,17 +17,18 @@
 
 package org.tensorflow.op.sparse;
 
+import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.Operands;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
-import org.tensorflow.types.TString;
 
 /**
  * Generates sparse cross from a list of sparse and dense tensors.
@@ -68,9 +69,11 @@
  *     [1, 1]: FingerprintCat64(
  *                 Fingerprint64("g"), FingerprintCat64(
  *                     Fingerprint64("e"), Fingerprint64("c")))
+ * 
+ * @param <T> data type for {@code outputValues()} output
  */
 @Operator(group = "sparse")
-public final class SparseCross extends RawOp {
+public final class SparseCross<T extends Tensor> extends RawOp {
   
   /**
    * Factory method to create a class wrapping a new SparseCross operation.
@@ -80,19 +83,30 @@ public final class SparseCross extends RawOp {
    * @param values 1-D.   values of each `SparseTensor`.
    * @param shapes 1-D.   Shapes of each `SparseTensor`.
    * @param denseInputs 2-D.    Columns represented by dense `Tensor`.
-   * @param sep string used when joining a list of string inputs, can be used as separator later.
+   * @param hashedOutput If true, returns the hash of the cross instead of the string.
+   * This will allow us avoiding string manipulations.
+   * @param numBuckets It is used if hashed_output is true.
+   * output = hashed_value%num_buckets if num_buckets > 0 else hashed_value.
+   * @param hashKey Specify the hash_key that will be used by the `FingerprintCat64`
+   * function to combine the crosses fingerprints.
+   * @param outType 
+   * @param internalType 
    * @return a new instance of SparseCross
    */
   @Endpoint(describeByClass = true)
-  public static SparseCross create(Scope scope, Iterable<Operand<TInt64>> indices, Iterable<Operand<?>> values, Iterable<Operand<TInt64>> shapes, Iterable<Operand<?>> denseInputs, Operand<TString> sep) {
-    OperationBuilder opBuilder = scope.env().opBuilder("SparseCrossV2", scope.makeOpName("SparseCross"));
+  public static <T extends Tensor, U extends Tensor> SparseCross<T> create(Scope scope, Iterable<Operand<TInt64>> indices, Iterable<Operand<?>> values, Iterable<Operand<TInt64>> shapes, Iterable<Operand<?>> denseInputs, Boolean hashedOutput, Long numBuckets, Long hashKey, DataType<T> outType, DataType<U> internalType) {
+    OperationBuilder opBuilder = scope.env().opBuilder("SparseCross", scope.makeOpName("SparseCross"));
     opBuilder.addInputList(Operands.asOutputs(indices));
     opBuilder.addInputList(Operands.asOutputs(values));
     opBuilder.addInputList(Operands.asOutputs(shapes));
     opBuilder.addInputList(Operands.asOutputs(denseInputs));
-    opBuilder.addInput(sep.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
-    return new SparseCross(opBuilder.build());
+    opBuilder.setAttr("hashed_output", hashedOutput);
+    opBuilder.setAttr("num_buckets", numBuckets);
+    opBuilder.setAttr("hash_key", hashKey);
+    opBuilder.setAttr("out_type", outType);
+    opBuilder.setAttr("internal_type", internalType);
+    return new SparseCross<T>(opBuilder.build());
   }
   
   /**
@@ -106,7 +120,7 @@ public Output<TInt64> outputIndices() {
    * 1-D.  Non-empty values of the concatenated or hashed
    * `SparseTensor`.
    */
-  public Output<TString> outputValues() {
+  public Output<T> outputValues() {
     return outputValues;
   }
   
@@ -117,11 +131,8 @@ public Output<TInt64> outputShape() {
     return outputShape;
   }
   
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "SparseCrossV2";
-  
   private Output<TInt64> outputIndices;
-  private Output<TString> outputValues;
+  private Output<T> outputValues;
   private Output<TInt64> outputShape;
   
   private SparseCross(Operation operation) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseDenseCwiseAdd.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseDenseCwiseAdd.java
index 40c5fc1446b..c07d4c50e22 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseDenseCwiseAdd.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseDenseCwiseAdd.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
 
 /**
  * Adds up a SparseTensor and a dense Tensor, using these special rules:
@@ -43,7 +43,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "sparse")
-public final class SparseDenseCwiseAdd<T extends TType> extends RawOp implements Operand<T> {
+public final class SparseDenseCwiseAdd<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new SparseDenseCwiseAdd operation.
@@ -57,7 +57,7 @@ public final class SparseDenseCwiseAdd<T extends TType> extends RawOp implements
    * @return a new instance of SparseDenseCwiseAdd
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> SparseDenseCwiseAdd<T> create(Scope scope, Operand<TInt64> spIndices, Operand<T> spValues, Operand<TInt64> spShape, Operand<T> dense) {
+  public static <T extends Tensor> SparseDenseCwiseAdd<T> create(Scope scope, Operand<TInt64> spIndices, Operand<T> spValues, Operand<TInt64> spShape, Operand<T> dense) {
     OperationBuilder opBuilder = scope.env().opBuilder("SparseDenseCwiseAdd", scope.makeOpName("SparseDenseCwiseAdd"));
     opBuilder.addInput(spIndices.asOutput());
     opBuilder.addInput(spValues.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseDenseCwiseDiv.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseDenseCwiseDiv.java
index 1d88f78e482..b2ac0fe6a0c 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseDenseCwiseDiv.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseDenseCwiseDiv.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
 
 /**
  * Component-wise divides a SparseTensor by a dense Tensor.
@@ -37,7 +37,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "sparse")
-public final class SparseDenseCwiseDiv<T extends TType> extends RawOp implements Operand<T> {
+public final class SparseDenseCwiseDiv<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new SparseDenseCwiseDiv operation.
@@ -51,7 +51,7 @@ public final class SparseDenseCwiseDiv<T extends TType> extends RawOp implements
    * @return a new instance of SparseDenseCwiseDiv
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> SparseDenseCwiseDiv<T> create(Scope scope, Operand<TInt64> spIndices, Operand<T> spValues, Operand<TInt64> spShape, Operand<T> dense) {
+  public static <T extends Tensor> SparseDenseCwiseDiv<T> create(Scope scope, Operand<TInt64> spIndices, Operand<T> spValues, Operand<TInt64> spShape, Operand<T> dense) {
     OperationBuilder opBuilder = scope.env().opBuilder("SparseDenseCwiseDiv", scope.makeOpName("SparseDenseCwiseDiv"));
     opBuilder.addInput(spIndices.asOutput());
     opBuilder.addInput(spValues.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseDenseCwiseMul.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseDenseCwiseMul.java
index b0144601eb3..cc36ff3cf9f 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseDenseCwiseMul.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseDenseCwiseMul.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
 
 /**
  * Component-wise multiplies a SparseTensor by a dense Tensor.
@@ -41,7 +41,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "sparse")
-public final class SparseDenseCwiseMul<T extends TType> extends RawOp implements Operand<T> {
+public final class SparseDenseCwiseMul<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new SparseDenseCwiseMul operation.
@@ -55,7 +55,7 @@ public final class SparseDenseCwiseMul<T extends TType> extends RawOp implements
    * @return a new instance of SparseDenseCwiseMul
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> SparseDenseCwiseMul<T> create(Scope scope, Operand<TInt64> spIndices, Operand<T> spValues, Operand<TInt64> spShape, Operand<T> dense) {
+  public static <T extends Tensor> SparseDenseCwiseMul<T> create(Scope scope, Operand<TInt64> spIndices, Operand<T> spValues, Operand<TInt64> spShape, Operand<T> dense) {
     OperationBuilder opBuilder = scope.env().opBuilder("SparseDenseCwiseMul", scope.makeOpName("SparseDenseCwiseMul"));
     opBuilder.addInput(spIndices.asOutput());
     opBuilder.addInput(spValues.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseFillEmptyRows.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseFillEmptyRows.java
index 6c8066061fc..6a05b2ddd9f 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseFillEmptyRows.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseFillEmptyRows.java
@@ -21,13 +21,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TBool;
 import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
 
 /**
  * Fills empty rows in the input 2-D `SparseTensor` with a default value.
@@ -72,7 +72,7 @@
  * @param <T> data type for {@code outputValues()} output
  */
 @Operator(group = "sparse")
-public final class SparseFillEmptyRows<T extends TType> extends RawOp {
+public final class SparseFillEmptyRows<T extends Tensor> extends RawOp {
   
   /**
    * Factory method to create a class wrapping a new SparseFillEmptyRows operation.
@@ -87,7 +87,7 @@ public final class SparseFillEmptyRows<T extends TType> extends RawOp {
    * @return a new instance of SparseFillEmptyRows
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> SparseFillEmptyRows<T> create(Scope scope, Operand<TInt64> indices, Operand<T> values, Operand<TInt64> denseShape, Operand<T> defaultValue) {
+  public static <T extends Tensor> SparseFillEmptyRows<T> create(Scope scope, Operand<TInt64> indices, Operand<T> values, Operand<TInt64> denseShape, Operand<T> defaultValue) {
     OperationBuilder opBuilder = scope.env().opBuilder("SparseFillEmptyRows", scope.makeOpName("SparseFillEmptyRows"));
     opBuilder.addInput(indices.asOutput());
     opBuilder.addInput(values.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseFillEmptyRowsGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseFillEmptyRowsGrad.java
index 3942e8bc8b3..a8688193f67 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseFillEmptyRowsGrad.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseFillEmptyRowsGrad.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
 
 /**
  * The gradient of SparseFillEmptyRows.
@@ -43,7 +43,7 @@
  * @param <T> data type for {@code dValues()} output
  */
 @Operator(group = "sparse")
-public final class SparseFillEmptyRowsGrad<T extends TType> extends RawOp {
+public final class SparseFillEmptyRowsGrad<T extends Tensor> extends RawOp {
   
   /**
    * Factory method to create a class wrapping a new SparseFillEmptyRowsGrad operation.
@@ -54,7 +54,7 @@ public final class SparseFillEmptyRowsGrad<T extends TType> extends RawOp {
    * @return a new instance of SparseFillEmptyRowsGrad
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> SparseFillEmptyRowsGrad<T> create(Scope scope, Operand<TInt64> reverseIndexMap, Operand<T> gradValues) {
+  public static <T extends Tensor> SparseFillEmptyRowsGrad<T> create(Scope scope, Operand<TInt64> reverseIndexMap, Operand<T> gradValues) {
     OperationBuilder opBuilder = scope.env().opBuilder("SparseFillEmptyRowsGrad", scope.makeOpName("SparseFillEmptyRowsGrad"));
     opBuilder.addInput(reverseIndexMap.asOutput());
     opBuilder.addInput(gradValues.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseMatMul.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseMatMul.java
index 295063d560a..03ebba44774 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseMatMul.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseMatMul.java
@@ -21,13 +21,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TFloat32;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Multiply matrix "a" by matrix "b".
@@ -101,7 +101,7 @@ private Options() {
    * @return a new instance of SparseMatMul
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber, U extends TNumber> SparseMatMul create(Scope scope, Operand<T> a, Operand<U> b, Options... options) {
+  public static <T extends Tensor & TNumber, U extends Tensor & TNumber> SparseMatMul create(Scope scope, Operand<T> a, Operand<U> b, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("SparseMatMul", scope.makeOpName("SparseMatMul"));
     opBuilder.addInput(a.asOutput());
     opBuilder.addInput(b.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseReduceMax.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseReduceMax.java
index 5288f7b5b88..f21237e0352 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseReduceMax.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseReduceMax.java
@@ -21,6 +21,7 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -28,7 +29,6 @@
 import org.tensorflow.types.TInt32;
 import org.tensorflow.types.TInt64;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes the max of elements across dimensions of a SparseTensor.
@@ -49,7 +49,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "sparse")
-public final class SparseReduceMax<T extends TNumber> extends RawOp implements Operand<T> {
+public final class SparseReduceMax<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.sparse.SparseReduceMax}
@@ -83,7 +83,7 @@ private Options() {
    * @return a new instance of SparseReduceMax
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> SparseReduceMax<T> create(Scope scope, Operand<TInt64> inputIndices, Operand<T> inputValues, Operand<TInt64> inputShape, Operand<TInt32> reductionAxes, Options... options) {
+  public static <T extends Tensor & TNumber> SparseReduceMax<T> create(Scope scope, Operand<TInt64> inputIndices, Operand<T> inputValues, Operand<TInt64> inputShape, Operand<TInt32> reductionAxes, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("SparseReduceMax", scope.makeOpName("SparseReduceMax"));
     opBuilder.addInput(inputIndices.asOutput());
     opBuilder.addInput(inputValues.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseReduceMaxSparse.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseReduceMaxSparse.java
index 93ce28bb66a..a9d196426e9 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseReduceMaxSparse.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseReduceMaxSparse.java
@@ -21,6 +21,7 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -28,7 +29,6 @@
 import org.tensorflow.types.TInt32;
 import org.tensorflow.types.TInt64;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes the max of elements across dimensions of a SparseTensor.
@@ -49,7 +49,7 @@
  * @param <T> data type for {@code outputValues()} output
  */
 @Operator(group = "sparse")
-public final class SparseReduceMaxSparse<T extends TNumber> extends RawOp {
+public final class SparseReduceMaxSparse<T extends Tensor & TNumber> extends RawOp {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.sparse.SparseReduceMaxSparse}
@@ -83,7 +83,7 @@ private Options() {
    * @return a new instance of SparseReduceMaxSparse
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> SparseReduceMaxSparse<T> create(Scope scope, Operand<TInt64> inputIndices, Operand<T> inputValues, Operand<TInt64> inputShape, Operand<TInt32> reductionAxes, Options... options) {
+  public static <T extends Tensor & TNumber> SparseReduceMaxSparse<T> create(Scope scope, Operand<TInt64> inputIndices, Operand<T> inputValues, Operand<TInt64> inputShape, Operand<TInt32> reductionAxes, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("SparseReduceMaxSparse", scope.makeOpName("SparseReduceMaxSparse"));
     opBuilder.addInput(inputIndices.asOutput());
     opBuilder.addInput(inputValues.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseReduceSum.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseReduceSum.java
index 138bb1aab5b..daf9d4cca0e 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseReduceSum.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseReduceSum.java
@@ -21,13 +21,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
 import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes the sum of elements across dimensions of a SparseTensor.
@@ -48,7 +48,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "sparse")
-public final class SparseReduceSum<T extends TType> extends RawOp implements Operand<T> {
+public final class SparseReduceSum<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.sparse.SparseReduceSum}
@@ -82,7 +82,7 @@ private Options() {
    * @return a new instance of SparseReduceSum
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> SparseReduceSum<T> create(Scope scope, Operand<TInt64> inputIndices, Operand<T> inputValues, Operand<TInt64> inputShape, Operand<TInt32> reductionAxes, Options... options) {
+  public static <T extends Tensor> SparseReduceSum<T> create(Scope scope, Operand<TInt64> inputIndices, Operand<T> inputValues, Operand<TInt64> inputShape, Operand<TInt32> reductionAxes, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("SparseReduceSum", scope.makeOpName("SparseReduceSum"));
     opBuilder.addInput(inputIndices.asOutput());
     opBuilder.addInput(inputValues.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseReduceSumSparse.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseReduceSumSparse.java
index e27a1a98861..4f4c839c694 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseReduceSumSparse.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseReduceSumSparse.java
@@ -21,13 +21,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
 import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes the sum of elements across dimensions of a SparseTensor.
@@ -48,7 +48,7 @@
  * @param <T> data type for {@code outputValues()} output
  */
 @Operator(group = "sparse")
-public final class SparseReduceSumSparse<T extends TType> extends RawOp {
+public final class SparseReduceSumSparse<T extends Tensor> extends RawOp {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.sparse.SparseReduceSumSparse}
@@ -82,7 +82,7 @@ private Options() {
    * @return a new instance of SparseReduceSumSparse
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> SparseReduceSumSparse<T> create(Scope scope, Operand<TInt64> inputIndices, Operand<T> inputValues, Operand<TInt64> inputShape, Operand<TInt32> reductionAxes, Options... options) {
+  public static <T extends Tensor> SparseReduceSumSparse<T> create(Scope scope, Operand<TInt64> inputIndices, Operand<T> inputValues, Operand<TInt64> inputShape, Operand<TInt32> reductionAxes, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("SparseReduceSumSparse", scope.makeOpName("SparseReduceSumSparse"));
     opBuilder.addInput(inputIndices.asOutput());
     opBuilder.addInput(inputValues.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseReorder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseReorder.java
index 16713d1f767..aacd2fab8e3 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseReorder.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseReorder.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
 
 /**
  * Reorders a SparseTensor into the canonical, row-major ordering.
@@ -43,7 +43,7 @@
  * @param <T> data type for {@code outputValues()} output
  */
 @Operator(group = "sparse")
-public final class SparseReorder<T extends TType> extends RawOp {
+public final class SparseReorder<T extends Tensor> extends RawOp {
   
   /**
    * Factory method to create a class wrapping a new SparseReorder operation.
@@ -56,7 +56,7 @@ public final class SparseReorder<T extends TType> extends RawOp {
    * @return a new instance of SparseReorder
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> SparseReorder<T> create(Scope scope, Operand<TInt64> inputIndices, Operand<T> inputValues, Operand<TInt64> inputShape) {
+  public static <T extends Tensor> SparseReorder<T> create(Scope scope, Operand<TInt64> inputIndices, Operand<T> inputValues, Operand<TInt64> inputShape) {
     OperationBuilder opBuilder = scope.env().opBuilder("SparseReorder", scope.makeOpName("SparseReorder"));
     opBuilder.addInput(inputIndices.asOutput());
     opBuilder.addInput(inputValues.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentMean.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentMean.java
index 14dd3df8c2f..276076198c0 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentMean.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentMean.java
@@ -21,12 +21,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes the mean along sparse segments of a tensor.
@@ -39,7 +40,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "sparse")
-public final class SparseSegmentMean<T extends TNumber> extends RawOp implements Operand<T> {
+public final class SparseSegmentMean<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new SparseSegmentMean operation.
@@ -51,7 +52,7 @@ public final class SparseSegmentMean<T extends TNumber> extends RawOp implements
    * @return a new instance of SparseSegmentMean
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber, U extends TNumber, V extends TNumber> SparseSegmentMean<T> create(Scope scope, Operand<T> data, Operand<U> indices, Operand<V> segmentIds) {
+  public static <T extends Tensor & TNumber, U extends Tensor & TNumber> SparseSegmentMean<T> create(Scope scope, Operand<T> data, Operand<U> indices, Operand<TInt32> segmentIds) {
     OperationBuilder opBuilder = scope.env().opBuilder("SparseSegmentMean", scope.makeOpName("SparseSegmentMean"));
     opBuilder.addInput(data.asOutput());
     opBuilder.addInput(indices.asOutput());
@@ -73,9 +74,6 @@ public Output<T> asOutput() {
     return output;
   }
   
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "SparseSegmentMean";
-  
   private Output<T> output;
   
   private SparseSegmentMean(Operation operation) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentMeanGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentMeanGrad.java
index b5ae14a0f03..5abe339711b 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentMeanGrad.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentMeanGrad.java
@@ -21,13 +21,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes gradients for SparseSegmentMean.
@@ -38,7 +38,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "sparse")
-public final class SparseSegmentMeanGrad<T extends TNumber> extends RawOp implements Operand<T> {
+public final class SparseSegmentMeanGrad<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new SparseSegmentMeanGrad operation.
@@ -51,7 +51,7 @@ public final class SparseSegmentMeanGrad<T extends TNumber> extends RawOp implem
    * @return a new instance of SparseSegmentMeanGrad
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber, U extends TNumber, V extends TNumber> SparseSegmentMeanGrad<T> create(Scope scope, Operand<T> grad, Operand<U> indices, Operand<V> segmentIds, Operand<TInt32> outputDim0) {
+  public static <T extends Tensor & TNumber, U extends Tensor & TNumber> SparseSegmentMeanGrad<T> create(Scope scope, Operand<T> grad, Operand<U> indices, Operand<TInt32> segmentIds, Operand<TInt32> outputDim0) {
     OperationBuilder opBuilder = scope.env().opBuilder("SparseSegmentMeanGrad", scope.makeOpName("SparseSegmentMeanGrad"));
     opBuilder.addInput(grad.asOutput());
     opBuilder.addInput(indices.asOutput());
@@ -72,9 +72,6 @@ public Output<T> asOutput() {
     return output;
   }
   
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "SparseSegmentMeanGrad";
-  
   private Output<T> output;
   
   private SparseSegmentMeanGrad(Operation operation) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentMeanWithNumSegments.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentMeanWithNumSegments.java
index 214d4983c81..2e32109e991 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentMeanWithNumSegments.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentMeanWithNumSegments.java
@@ -21,18 +21,19 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes the mean along sparse segments of a tensor.
  * <p>
  * Like `SparseSegmentMean`, but allows missing ids in `segment_ids`. If an id is
- * missing, the `output` tensor at that position will be zeroed.
+ * misisng, the `output` tensor at that position will be zeroed.
  * <p>
  * Read
  * [the section on segmentation](https://tensorflow.org/api_docs/python/tf/math#Segmentation)
@@ -41,7 +42,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "sparse")
-public final class SparseSegmentMeanWithNumSegments<T extends TNumber> extends RawOp implements Operand<T> {
+public final class SparseSegmentMeanWithNumSegments<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new SparseSegmentMeanWithNumSegments operation.
@@ -54,7 +55,7 @@ public final class SparseSegmentMeanWithNumSegments<T extends TNumber> extends R
    * @return a new instance of SparseSegmentMeanWithNumSegments
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber, U extends TNumber, V extends TNumber, W extends TNumber> SparseSegmentMeanWithNumSegments<T> create(Scope scope, Operand<T> data, Operand<U> indices, Operand<V> segmentIds, Operand<W> numSegments) {
+  public static <T extends Tensor & TNumber, U extends Tensor & TNumber, V extends Tensor & TNumber> SparseSegmentMeanWithNumSegments<T> create(Scope scope, Operand<T> data, Operand<U> indices, Operand<TInt32> segmentIds, Operand<V> numSegments) {
     OperationBuilder opBuilder = scope.env().opBuilder("SparseSegmentMeanWithNumSegments", scope.makeOpName("SparseSegmentMeanWithNumSegments"));
     opBuilder.addInput(data.asOutput());
     opBuilder.addInput(indices.asOutput());
@@ -77,9 +78,6 @@ public Output<T> asOutput() {
     return output;
   }
   
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "SparseSegmentMeanWithNumSegments";
-  
   private Output<T> output;
   
   private SparseSegmentMeanWithNumSegments(Operation operation) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentSqrtN.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentSqrtN.java
index c1889eae169..ce500c7a976 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentSqrtN.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentSqrtN.java
@@ -21,12 +21,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes the sum along sparse segments of a tensor divided by the sqrt of N.
@@ -39,7 +40,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "sparse")
-public final class SparseSegmentSqrtN<T extends TNumber> extends RawOp implements Operand<T> {
+public final class SparseSegmentSqrtN<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new SparseSegmentSqrtN operation.
@@ -51,7 +52,7 @@ public final class SparseSegmentSqrtN<T extends TNumber> extends RawOp implement
    * @return a new instance of SparseSegmentSqrtN
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber, U extends TNumber, V extends TNumber> SparseSegmentSqrtN<T> create(Scope scope, Operand<T> data, Operand<U> indices, Operand<V> segmentIds) {
+  public static <T extends Tensor & TNumber, U extends Tensor & TNumber> SparseSegmentSqrtN<T> create(Scope scope, Operand<T> data, Operand<U> indices, Operand<TInt32> segmentIds) {
     OperationBuilder opBuilder = scope.env().opBuilder("SparseSegmentSqrtN", scope.makeOpName("SparseSegmentSqrtN"));
     opBuilder.addInput(data.asOutput());
     opBuilder.addInput(indices.asOutput());
@@ -73,9 +74,6 @@ public Output<T> asOutput() {
     return output;
   }
   
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "SparseSegmentSqrtN";
-  
   private Output<T> output;
   
   private SparseSegmentSqrtN(Operation operation) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentSqrtNGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentSqrtNGrad.java
index 3366dbac0c8..706e8cd9d15 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentSqrtNGrad.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentSqrtNGrad.java
@@ -21,13 +21,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes gradients for SparseSegmentSqrtN.
@@ -38,7 +38,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "sparse")
-public final class SparseSegmentSqrtNGrad<T extends TNumber> extends RawOp implements Operand<T> {
+public final class SparseSegmentSqrtNGrad<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new SparseSegmentSqrtNGrad operation.
@@ -51,7 +51,7 @@ public final class SparseSegmentSqrtNGrad<T extends TNumber> extends RawOp imple
    * @return a new instance of SparseSegmentSqrtNGrad
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber, U extends TNumber, V extends TNumber> SparseSegmentSqrtNGrad<T> create(Scope scope, Operand<T> grad, Operand<U> indices, Operand<V> segmentIds, Operand<TInt32> outputDim0) {
+  public static <T extends Tensor & TNumber, U extends Tensor & TNumber> SparseSegmentSqrtNGrad<T> create(Scope scope, Operand<T> grad, Operand<U> indices, Operand<TInt32> segmentIds, Operand<TInt32> outputDim0) {
     OperationBuilder opBuilder = scope.env().opBuilder("SparseSegmentSqrtNGrad", scope.makeOpName("SparseSegmentSqrtNGrad"));
     opBuilder.addInput(grad.asOutput());
     opBuilder.addInput(indices.asOutput());
@@ -72,9 +72,6 @@ public Output<T> asOutput() {
     return output;
   }
   
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "SparseSegmentSqrtNGrad";
-  
   private Output<T> output;
   
   private SparseSegmentSqrtNGrad(Operation operation) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentSqrtNWithNumSegments.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentSqrtNWithNumSegments.java
index 41000f65cdf..0d6d1081a7d 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentSqrtNWithNumSegments.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentSqrtNWithNumSegments.java
@@ -21,12 +21,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes the sum along sparse segments of a tensor divided by the sqrt of N.
@@ -34,7 +35,7 @@
  * N is the size of the segment being reduced.
  * <p>
  * Like `SparseSegmentSqrtN`, but allows missing ids in `segment_ids`. If an id is
- * missing, the `output` tensor at that position will be zeroed.
+ * misisng, the `output` tensor at that position will be zeroed.
  * <p>
  * Read
  * [the section on segmentation](https://tensorflow.org/api_docs/python/tf/math#Segmentation)
@@ -43,7 +44,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "sparse")
-public final class SparseSegmentSqrtNWithNumSegments<T extends TNumber> extends RawOp implements Operand<T> {
+public final class SparseSegmentSqrtNWithNumSegments<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new SparseSegmentSqrtNWithNumSegments operation.
@@ -56,7 +57,7 @@ public final class SparseSegmentSqrtNWithNumSegments<T extends TNumber> extends
    * @return a new instance of SparseSegmentSqrtNWithNumSegments
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber, U extends TNumber, V extends TNumber, W extends TNumber> SparseSegmentSqrtNWithNumSegments<T> create(Scope scope, Operand<T> data, Operand<U> indices, Operand<V> segmentIds, Operand<W> numSegments) {
+  public static <T extends Tensor & TNumber, U extends Tensor & TNumber, V extends Tensor & TNumber> SparseSegmentSqrtNWithNumSegments<T> create(Scope scope, Operand<T> data, Operand<U> indices, Operand<TInt32> segmentIds, Operand<V> numSegments) {
     OperationBuilder opBuilder = scope.env().opBuilder("SparseSegmentSqrtNWithNumSegments", scope.makeOpName("SparseSegmentSqrtNWithNumSegments"));
     opBuilder.addInput(data.asOutput());
     opBuilder.addInput(indices.asOutput());
@@ -79,9 +80,6 @@ public Output<T> asOutput() {
     return output;
   }
   
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "SparseSegmentSqrtNWithNumSegments";
-  
   private Output<T> output;
   
   private SparseSegmentSqrtNWithNumSegments(Operation operation) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentSum.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentSum.java
index 991653b2c02..13bcf882abe 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentSum.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentSum.java
@@ -21,12 +21,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes the sum along sparse segments of a tensor.
@@ -64,7 +65,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "sparse")
-public final class SparseSegmentSum<T extends TNumber> extends RawOp implements Operand<T> {
+public final class SparseSegmentSum<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new SparseSegmentSum operation.
@@ -76,7 +77,7 @@ public final class SparseSegmentSum<T extends TNumber> extends RawOp implements
    * @return a new instance of SparseSegmentSum
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber, U extends TNumber, V extends TNumber> SparseSegmentSum<T> create(Scope scope, Operand<T> data, Operand<U> indices, Operand<V> segmentIds) {
+  public static <T extends Tensor & TNumber, U extends Tensor & TNumber> SparseSegmentSum<T> create(Scope scope, Operand<T> data, Operand<U> indices, Operand<TInt32> segmentIds) {
     OperationBuilder opBuilder = scope.env().opBuilder("SparseSegmentSum", scope.makeOpName("SparseSegmentSum"));
     opBuilder.addInput(data.asOutput());
     opBuilder.addInput(indices.asOutput());
@@ -98,9 +99,6 @@ public Output<T> asOutput() {
     return output;
   }
   
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "SparseSegmentSum";
-  
   private Output<T> output;
   
   private SparseSegmentSum(Operation operation) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentSumWithNumSegments.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentSumWithNumSegments.java
index f5414151cbe..598d563c82b 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentSumWithNumSegments.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentSumWithNumSegments.java
@@ -21,18 +21,19 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes the sum along sparse segments of a tensor.
  * <p>
  * Like `SparseSegmentSum`, but allows missing ids in `segment_ids`. If an id is
- * missing, the `output` tensor at that position will be zeroed.
+ * misisng, the `output` tensor at that position will be zeroed.
  * <p>
  * Read
  * [the section on segmentation](https://tensorflow.org/api_docs/python/tf/sparse#Segmentation)
@@ -62,7 +63,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "sparse")
-public final class SparseSegmentSumWithNumSegments<T extends TNumber> extends RawOp implements Operand<T> {
+public final class SparseSegmentSumWithNumSegments<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new SparseSegmentSumWithNumSegments operation.
@@ -75,7 +76,7 @@ public final class SparseSegmentSumWithNumSegments<T extends TNumber> extends Ra
    * @return a new instance of SparseSegmentSumWithNumSegments
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber, U extends TNumber, V extends TNumber, W extends TNumber> SparseSegmentSumWithNumSegments<T> create(Scope scope, Operand<T> data, Operand<U> indices, Operand<V> segmentIds, Operand<W> numSegments) {
+  public static <T extends Tensor & TNumber, U extends Tensor & TNumber, V extends Tensor & TNumber> SparseSegmentSumWithNumSegments<T> create(Scope scope, Operand<T> data, Operand<U> indices, Operand<TInt32> segmentIds, Operand<V> numSegments) {
     OperationBuilder opBuilder = scope.env().opBuilder("SparseSegmentSumWithNumSegments", scope.makeOpName("SparseSegmentSumWithNumSegments"));
     opBuilder.addInput(data.asOutput());
     opBuilder.addInput(indices.asOutput());
@@ -98,9 +99,6 @@ public Output<T> asOutput() {
     return output;
   }
   
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "SparseSegmentSumWithNumSegments";
-  
   private Output<T> output;
   
   private SparseSegmentSumWithNumSegments(Operation operation) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSlice.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSlice.java
index 4ed30d8e2eb..ce9c123ab64 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSlice.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSlice.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
 
 /**
  * Slice a `SparseTensor` based on the `start` and `size`.
@@ -50,7 +50,7 @@
  * @param <T> data type for {@code outputValues()} output
  */
 @Operator(group = "sparse")
-public final class SparseSlice<T extends TType> extends RawOp {
+public final class SparseSlice<T extends Tensor> extends RawOp {
   
   /**
    * Factory method to create a class wrapping a new SparseSlice operation.
@@ -66,7 +66,7 @@ public final class SparseSlice<T extends TType> extends RawOp {
    * @return a new instance of SparseSlice
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> SparseSlice<T> create(Scope scope, Operand<TInt64> indices, Operand<T> values, Operand<TInt64> shape, Operand<TInt64> start, Operand<TInt64> size) {
+  public static <T extends Tensor> SparseSlice<T> create(Scope scope, Operand<TInt64> indices, Operand<T> values, Operand<TInt64> shape, Operand<TInt64> start, Operand<TInt64> size) {
     OperationBuilder opBuilder = scope.env().opBuilder("SparseSlice", scope.makeOpName("SparseSlice"));
     opBuilder.addInput(indices.asOutput());
     opBuilder.addInput(values.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSliceGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSliceGrad.java
index 480631eccbc..a5860715cbe 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSliceGrad.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSliceGrad.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
 
 /**
  * The gradient operator for the SparseSlice op.
@@ -38,7 +38,7 @@
  * @param <T> data type for {@code valGrad()} output
  */
 @Operator(group = "sparse")
-public final class SparseSliceGrad<T extends TType> extends RawOp implements Operand<T> {
+public final class SparseSliceGrad<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new SparseSliceGrad operation.
@@ -52,7 +52,7 @@ public final class SparseSliceGrad<T extends TType> extends RawOp implements Ope
    * @return a new instance of SparseSliceGrad
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> SparseSliceGrad<T> create(Scope scope, Operand<T> backpropValGrad, Operand<TInt64> inputIndices, Operand<TInt64> inputStart, Operand<TInt64> outputIndices) {
+  public static <T extends Tensor> SparseSliceGrad<T> create(Scope scope, Operand<T> backpropValGrad, Operand<TInt64> inputIndices, Operand<TInt64> inputStart, Operand<TInt64> outputIndices) {
     OperationBuilder opBuilder = scope.env().opBuilder("SparseSliceGrad", scope.makeOpName("SparseSliceGrad"));
     opBuilder.addInput(backpropValGrad.asOutput());
     opBuilder.addInput(inputIndices.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSoftmax.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSoftmax.java
index 0906213115f..24e6a3553e9 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSoftmax.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSoftmax.java
@@ -21,13 +21,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Applies softmax to a batched N-D `SparseTensor`.
@@ -51,7 +51,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "sparse")
-public final class SparseSoftmax<T extends TNumber> extends RawOp implements Operand<T> {
+public final class SparseSoftmax<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new SparseSoftmax operation.
@@ -64,7 +64,7 @@ public final class SparseSoftmax<T extends TNumber> extends RawOp implements Ope
    * @return a new instance of SparseSoftmax
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> SparseSoftmax<T> create(Scope scope, Operand<TInt64> spIndices, Operand<T> spValues, Operand<TInt64> spShape) {
+  public static <T extends Tensor & TNumber> SparseSoftmax<T> create(Scope scope, Operand<TInt64> spIndices, Operand<T> spValues, Operand<TInt64> spShape) {
     OperationBuilder opBuilder = scope.env().opBuilder("SparseSoftmax", scope.makeOpName("SparseSoftmax"));
     opBuilder.addInput(spIndices.asOutput());
     opBuilder.addInput(spValues.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSparseMaximum.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSparseMaximum.java
index 3cbf1a6ef7d..a993f2e3e7a 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSparseMaximum.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSparseMaximum.java
@@ -21,13 +21,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Returns the element-wise max of two SparseTensors.
@@ -37,7 +37,7 @@
  * @param <T> data type for {@code outputValues()} output
  */
 @Operator(group = "sparse")
-public final class SparseSparseMaximum<T extends TNumber> extends RawOp {
+public final class SparseSparseMaximum<T extends Tensor & TNumber> extends RawOp {
   
   /**
    * Factory method to create a class wrapping a new SparseSparseMaximum operation.
@@ -53,7 +53,7 @@ public final class SparseSparseMaximum<T extends TNumber> extends RawOp {
    * @return a new instance of SparseSparseMaximum
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> SparseSparseMaximum<T> create(Scope scope, Operand<TInt64> aIndices, Operand<T> aValues, Operand<TInt64> aShape, Operand<TInt64> bIndices, Operand<T> bValues, Operand<TInt64> bShape) {
+  public static <T extends Tensor & TNumber> SparseSparseMaximum<T> create(Scope scope, Operand<TInt64> aIndices, Operand<T> aValues, Operand<TInt64> aShape, Operand<TInt64> bIndices, Operand<T> bValues, Operand<TInt64> bShape) {
     OperationBuilder opBuilder = scope.env().opBuilder("SparseSparseMaximum", scope.makeOpName("SparseSparseMaximum"));
     opBuilder.addInput(aIndices.asOutput());
     opBuilder.addInput(aValues.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSparseMinimum.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSparseMinimum.java
index b2f6aedbccb..6a0d20c559b 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSparseMinimum.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSparseMinimum.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
 
 /**
  * Returns the element-wise min of two SparseTensors.
@@ -36,7 +36,7 @@
  * @param <T> data type for {@code outputValues()} output
  */
 @Operator(group = "sparse")
-public final class SparseSparseMinimum<T extends TType> extends RawOp {
+public final class SparseSparseMinimum<T extends Tensor> extends RawOp {
   
   /**
    * Factory method to create a class wrapping a new SparseSparseMinimum operation.
@@ -52,7 +52,7 @@ public final class SparseSparseMinimum<T extends TType> extends RawOp {
    * @return a new instance of SparseSparseMinimum
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> SparseSparseMinimum<T> create(Scope scope, Operand<TInt64> aIndices, Operand<T> aValues, Operand<TInt64> aShape, Operand<TInt64> bIndices, Operand<T> bValues, Operand<TInt64> bShape) {
+  public static <T extends Tensor> SparseSparseMinimum<T> create(Scope scope, Operand<TInt64> aIndices, Operand<T> aValues, Operand<TInt64> aShape, Operand<TInt64> bIndices, Operand<T> bValues, Operand<TInt64> bShape) {
     OperationBuilder opBuilder = scope.env().opBuilder("SparseSparseMinimum", scope.makeOpName("SparseSparseMinimum"));
     opBuilder.addInput(aIndices.asOutput());
     opBuilder.addInput(aValues.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSplit.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSplit.java
index e93f6ebe40e..4970cca6cb2 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSplit.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSplit.java
@@ -23,12 +23,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
 
 /**
  * Split a `SparseTensor` into `num_split` tensors along one dimension.
@@ -54,7 +54,7 @@
  * @param <T> data type for {@code outputValues()} output
  */
 @Operator(group = "sparse")
-public final class SparseSplit<T extends TType> extends RawOp {
+public final class SparseSplit<T extends Tensor> extends RawOp {
   
   /**
    * Factory method to create a class wrapping a new SparseSplit operation.
@@ -71,7 +71,7 @@ public final class SparseSplit<T extends TType> extends RawOp {
    * @return a new instance of SparseSplit
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> SparseSplit<T> create(Scope scope, Operand<TInt64> splitDim, Operand<TInt64> indices, Operand<T> values, Operand<TInt64> shape, Long numSplit) {
+  public static <T extends Tensor> SparseSplit<T> create(Scope scope, Operand<TInt64> splitDim, Operand<TInt64> indices, Operand<T> values, Operand<TInt64> shape, Long numSplit) {
     OperationBuilder opBuilder = scope.env().opBuilder("SparseSplit", scope.makeOpName("SparseSplit"));
     opBuilder.addInput(splitDim.asOutput());
     opBuilder.addInput(indices.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseTensorDenseAdd.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseTensorDenseAdd.java
index 9959d57d16b..f1d0dc84d2b 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseTensorDenseAdd.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseTensorDenseAdd.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Adds up a `SparseTensor` and a dense `Tensor`, producing a dense `Tensor`.
@@ -36,7 +36,7 @@
  * @param <U> data type for {@code output()} output
  */
 @Operator(group = "sparse")
-public final class SparseTensorDenseAdd<U extends TType> extends RawOp implements Operand<U> {
+public final class SparseTensorDenseAdd<U extends Tensor> extends RawOp implements Operand<U> {
   
   /**
    * Factory method to create a class wrapping a new SparseTensorDenseAdd operation.
@@ -49,7 +49,7 @@ public final class SparseTensorDenseAdd<U extends TType> extends RawOp implement
    * @return a new instance of SparseTensorDenseAdd
    */
   @Endpoint(describeByClass = true)
-  public static <U extends TType, T extends TNumber> SparseTensorDenseAdd<U> create(Scope scope, Operand<T> aIndices, Operand<U> aValues, Operand<T> aShape, Operand<U> b) {
+  public static <U extends Tensor, T extends Tensor & TNumber> SparseTensorDenseAdd<U> create(Scope scope, Operand<T> aIndices, Operand<U> aValues, Operand<T> aShape, Operand<U> b) {
     OperationBuilder opBuilder = scope.env().opBuilder("SparseTensorDenseAdd", scope.makeOpName("SparseTensorDenseAdd"));
     opBuilder.addInput(aIndices.asOutput());
     opBuilder.addInput(aValues.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseTensorDenseMatMul.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseTensorDenseMatMul.java
index 63f3d3116c0..ecfb1c780a4 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseTensorDenseMatMul.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseTensorDenseMatMul.java
@@ -21,13 +21,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Multiply SparseTensor (of rank 2) "A" by dense matrix "B".
@@ -45,7 +45,7 @@
  * @param <U> data type for {@code product()} output
  */
 @Operator(group = "sparse")
-public final class SparseTensorDenseMatMul<U extends TType> extends RawOp implements Operand<U> {
+public final class SparseTensorDenseMatMul<U extends Tensor> extends RawOp implements Operand<U> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.sparse.SparseTensorDenseMatMul}
@@ -89,7 +89,7 @@ private Options() {
    * @return a new instance of SparseTensorDenseMatMul
    */
   @Endpoint(describeByClass = true)
-  public static <U extends TType, T extends TNumber> SparseTensorDenseMatMul<U> create(Scope scope, Operand<T> aIndices, Operand<U> aValues, Operand<TInt64> aShape, Operand<U> b, Options... options) {
+  public static <U extends Tensor, T extends Tensor & TNumber> SparseTensorDenseMatMul<U> create(Scope scope, Operand<T> aIndices, Operand<U> aValues, Operand<TInt64> aShape, Operand<U> b, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("SparseTensorDenseMatMul", scope.makeOpName("SparseTensorDenseMatMul"));
     opBuilder.addInput(aIndices.asOutput());
     opBuilder.addInput(aValues.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseToDense.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseToDense.java
index 0f86af1327d..ff006e0cf5a 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseToDense.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseToDense.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Converts a sparse representation into a dense tensor.
@@ -52,7 +52,7 @@
  * @param <U> data type for {@code dense()} output
  */
 @Operator(group = "sparse")
-public final class SparseToDense<U extends TType> extends RawOp implements Operand<U> {
+public final class SparseToDense<U extends Tensor> extends RawOp implements Operand<U> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.sparse.SparseToDense}
@@ -89,7 +89,7 @@ private Options() {
    * @return a new instance of SparseToDense
    */
   @Endpoint(describeByClass = true)
-  public static <U extends TType, T extends TNumber> SparseToDense<U> create(Scope scope, Operand<T> sparseIndices, Operand<T> outputShape, Operand<U> sparseValues, Operand<U> defaultValue, Options... options) {
+  public static <U extends Tensor, T extends Tensor & TNumber> SparseToDense<U> create(Scope scope, Operand<T> sparseIndices, Operand<T> outputShape, Operand<U> sparseValues, Operand<U> defaultValue, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("SparseToDense", scope.makeOpName("SparseToDense"));
     opBuilder.addInput(sparseIndices.asOutput());
     opBuilder.addInput(outputShape.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseToSparseSetOperation.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseToSparseSetOperation.java
index 31cbe67a914..c0284a23632 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseToSparseSetOperation.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseToSparseSetOperation.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
 
 /**
  * Applies set operation along last dimension of 2 `SparseTensor` inputs.
@@ -58,7 +58,7 @@
  * @param <T> data type for {@code resultValues()} output
  */
 @Operator(group = "sparse")
-public final class SparseToSparseSetOperation<T extends TType> extends RawOp {
+public final class SparseToSparseSetOperation<T extends Tensor> extends RawOp {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.sparse.SparseToSparseSetOperation}
@@ -102,7 +102,7 @@ private Options() {
    * @return a new instance of SparseToSparseSetOperation
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> SparseToSparseSetOperation<T> create(Scope scope, Operand<TInt64> set1Indices, Operand<T> set1Values, Operand<TInt64> set1Shape, Operand<TInt64> set2Indices, Operand<T> set2Values, Operand<TInt64> set2Shape, String setOperation, Options... options) {
+  public static <T extends Tensor> SparseToSparseSetOperation<T> create(Scope scope, Operand<TInt64> set1Indices, Operand<T> set1Values, Operand<TInt64> set1Shape, Operand<TInt64> set2Indices, Operand<T> set2Values, Operand<TInt64> set2Shape, String setOperation, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("SparseToSparseSetOperation", scope.makeOpName("SparseToSparseSetOperation"));
     opBuilder.addInput(set1Indices.asOutput());
     opBuilder.addInput(set1Values.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/TakeManySparseFromTensorsMap.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/TakeManySparseFromTensorsMap.java
index 7fd5242cb20..0909bf70863 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/TakeManySparseFromTensorsMap.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/TakeManySparseFromTensorsMap.java
@@ -22,12 +22,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
 
 /**
  * Read `SparseTensors` from a `SparseTensorsMap` and concatenate them.
@@ -80,7 +80,7 @@
  * @param <T> data type for {@code sparseValues()} output
  */
 @Operator(group = "sparse")
-public final class TakeManySparseFromTensorsMap<T extends TType> extends RawOp {
+public final class TakeManySparseFromTensorsMap<T extends Tensor> extends RawOp {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.sparse.TakeManySparseFromTensorsMap}
@@ -124,7 +124,7 @@ private Options() {
    * @return a new instance of TakeManySparseFromTensorsMap
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> TakeManySparseFromTensorsMap<T> create(Scope scope, Operand<TInt64> sparseHandles, DataType<T> dtype, Options... options) {
+  public static <T extends Tensor> TakeManySparseFromTensorsMap<T> create(Scope scope, Operand<TInt64> sparseHandles, DataType<T> dtype, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("TakeManySparseFromTensorsMap", scope.makeOpName("TakeManySparseFromTensorsMap"));
     opBuilder.addInput(sparseHandles.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/StringNGrams.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/StringNGrams.java
index 6d68c11fe1b..b420ae031f0 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/StringNGrams.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/StringNGrams.java
@@ -22,13 +22,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TString;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Creates ngrams from ragged string data.
@@ -40,7 +40,7 @@
  * @param <T> data type for {@code ngramsSplits()} output
  */
 @Operator(group = "strings")
-public final class StringNGrams<T extends TNumber> extends RawOp {
+public final class StringNGrams<T extends Tensor & TNumber> extends RawOp {
   
   /**
    * Factory method to create a class wrapping a new StringNGrams operation.
@@ -63,7 +63,7 @@ public final class StringNGrams<T extends TNumber> extends RawOp {
    * @return a new instance of StringNGrams
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> StringNGrams<T> create(Scope scope, Operand<TString> data, Operand<T> dataSplits, String separator, List<Long> ngramWidths, String leftPad, String rightPad, Long padWidth, Boolean preserveShortSequences) {
+  public static <T extends Tensor & TNumber> StringNGrams<T> create(Scope scope, Operand<TString> data, Operand<T> dataSplits, String separator, List<Long> ngramWidths, String leftPad, String rightPad, Long padWidth, Boolean preserveShortSequences) {
     OperationBuilder opBuilder = scope.env().opBuilder("StringNGrams", scope.makeOpName("StringNGrams"));
     opBuilder.addInput(data.asOutput());
     opBuilder.addInput(dataSplits.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/Substr.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/Substr.java
index f1eb5a05485..abb39bdd67a 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/Substr.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/Substr.java
@@ -21,13 +21,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TString;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Return substrings from `Tensor` of strings.
@@ -145,7 +145,7 @@ private Options() {
    * @return a new instance of Substr
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> Substr create(Scope scope, Operand<TString> input, Operand<T> pos, Operand<T> len, Options... options) {
+  public static <T extends Tensor & TNumber> Substr create(Scope scope, Operand<TString> input, Operand<T> pos, Operand<T> len, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("Substr", scope.makeOpName("Substr"));
     opBuilder.addInput(input.asOutput());
     opBuilder.addInput(pos.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/ToNumber.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/ToNumber.java
index e8e5e4039b9..8a6ae49eec3 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/ToNumber.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/ToNumber.java
@@ -22,6 +22,7 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -29,7 +30,6 @@
 import org.tensorflow.types.TFloat32;
 import org.tensorflow.types.TString;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Converts each string in the input Tensor to the specified numeric type.
@@ -47,7 +47,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "strings")
-public final class ToNumber<T extends TNumber> extends RawOp implements Operand<T> {
+public final class ToNumber<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new ToNumber operation.
@@ -58,7 +58,7 @@ public final class ToNumber<T extends TNumber> extends RawOp implements Operand<
    * @return a new instance of ToNumber
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> ToNumber<T> create(Scope scope, Operand<TString> stringTensor, DataType<T> outType) {
+  public static <T extends Tensor & TNumber> ToNumber<T> create(Scope scope, Operand<TString> stringTensor, DataType<T> outType) {
     OperationBuilder opBuilder = scope.env().opBuilder("StringToNumber", scope.makeOpName("ToNumber"));
     opBuilder.addInput(stringTensor.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/UnicodeDecode.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/UnicodeDecode.java
index a39b08b1f5a..78e046da645 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/UnicodeDecode.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/UnicodeDecode.java
@@ -22,6 +22,7 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -30,7 +31,6 @@
 import org.tensorflow.types.TInt64;
 import org.tensorflow.types.TString;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Decodes each string in `input` into a sequence of Unicode code points.
@@ -54,7 +54,7 @@
  * 
  * @param <T> data type for {@code rowSplits()} output
  */
-public final class UnicodeDecode<T extends TNumber> extends RawOp {
+public final class UnicodeDecode<T extends Tensor & TNumber> extends RawOp {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.strings.UnicodeDecode}
@@ -116,7 +116,7 @@ private Options() {
    * @return a new instance of UnicodeDecode
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> UnicodeDecode<T> create(Scope scope, Operand<TString> input, String inputEncoding, DataType<T> Tsplits, Options... options) {
+  public static <T extends Tensor & TNumber> UnicodeDecode<T> create(Scope scope, Operand<TString> input, String inputEncoding, DataType<T> Tsplits, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("UnicodeDecode", scope.makeOpName("UnicodeDecode"));
     opBuilder.addInput(input.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/UnicodeDecodeWithOffsets.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/UnicodeDecodeWithOffsets.java
index ce6977b63ef..8900040c0c3 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/UnicodeDecodeWithOffsets.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/UnicodeDecodeWithOffsets.java
@@ -22,6 +22,7 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -30,7 +31,6 @@
 import org.tensorflow.types.TInt64;
 import org.tensorflow.types.TString;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Decodes each string in `input` into a sequence of Unicode code points.
@@ -60,7 +60,7 @@
  * 
  * @param <T> data type for {@code rowSplits()} output
  */
-public final class UnicodeDecodeWithOffsets<T extends TNumber> extends RawOp {
+public final class UnicodeDecodeWithOffsets<T extends Tensor & TNumber> extends RawOp {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.strings.UnicodeDecodeWithOffsets}
@@ -122,7 +122,7 @@ private Options() {
    * @return a new instance of UnicodeDecodeWithOffsets
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> UnicodeDecodeWithOffsets<T> create(Scope scope, Operand<TString> input, String inputEncoding, DataType<T> Tsplits, Options... options) {
+  public static <T extends Tensor & TNumber> UnicodeDecodeWithOffsets<T> create(Scope scope, Operand<TString> input, String inputEncoding, DataType<T> Tsplits, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("UnicodeDecodeWithOffsets", scope.makeOpName("UnicodeDecodeWithOffsets"));
     opBuilder.addInput(input.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/UnicodeEncode.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/UnicodeEncode.java
index ab83ef9dcda..f6e29c08385 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/UnicodeEncode.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/UnicodeEncode.java
@@ -21,6 +21,7 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -28,7 +29,6 @@
 import org.tensorflow.types.TInt32;
 import org.tensorflow.types.TString;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Encode a tensor of ints into unicode strings.
@@ -102,7 +102,7 @@ private Options() {
    * @return a new instance of UnicodeEncode
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> UnicodeEncode create(Scope scope, Operand<TInt32> inputValues, Operand<T> inputSplits, String outputEncoding, Options... options) {
+  public static <T extends Tensor & TNumber> UnicodeEncode create(Scope scope, Operand<TInt32> inputValues, Operand<T> inputSplits, String outputEncoding, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("UnicodeEncode", scope.makeOpName("UnicodeEncode"));
     opBuilder.addInput(inputValues.asOutput());
     opBuilder.addInput(inputSplits.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/UnsortedSegmentJoin.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/UnsortedSegmentJoin.java
index df1bbc68b68..7ada46e3e7f 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/UnsortedSegmentJoin.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/UnsortedSegmentJoin.java
@@ -21,13 +21,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TString;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Joins the elements of `inputs` based on `segment_ids`.
@@ -93,7 +93,7 @@ private Options() {
    * @return a new instance of UnsortedSegmentJoin
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber, U extends TNumber> UnsortedSegmentJoin create(Scope scope, Operand<TString> inputs, Operand<T> segmentIds, Operand<U> numSegments, Options... options) {
+  public static <T extends Tensor & TNumber, U extends Tensor & TNumber> UnsortedSegmentJoin create(Scope scope, Operand<TString> inputs, Operand<T> segmentIds, Operand<U> numSegments, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("UnsortedSegmentJoin", scope.makeOpName("UnsortedSegmentJoin"));
     opBuilder.addInput(inputs.asOutput());
     opBuilder.addInput(segmentIds.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/HistogramSummary.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/HistogramSummary.java
index 669cbf5c2fb..f7ca3602773 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/HistogramSummary.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/HistogramSummary.java
@@ -21,13 +21,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TString;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Outputs a `Summary` protocol buffer with a histogram.
@@ -50,7 +50,7 @@ public final class HistogramSummary extends RawOp implements Operand<TString> {
    * @return a new instance of HistogramSummary
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> HistogramSummary create(Scope scope, Operand<TString> tag, Operand<T> values) {
+  public static <T extends Tensor & TNumber> HistogramSummary create(Scope scope, Operand<TString> tag, Operand<T> values) {
     OperationBuilder opBuilder = scope.env().opBuilder("HistogramSummary", scope.makeOpName("HistogramSummary"));
     opBuilder.addInput(tag.asOutput());
     opBuilder.addInput(values.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/ImageSummary.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/ImageSummary.java
index 5b0f9f8859f..ac4702a0f6d 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/ImageSummary.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/ImageSummary.java
@@ -28,7 +28,6 @@
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TString;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Outputs a `Summary` protocol buffer with images.
@@ -122,7 +121,7 @@ private Options() {
    * @return a new instance of ImageSummary
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> ImageSummary create(Scope scope, Operand<TString> tag, Operand<T> tensor, Options... options) {
+  public static <T extends Tensor & TNumber> ImageSummary create(Scope scope, Operand<TString> tag, Operand<T> tensor, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ImageSummary", scope.makeOpName("ImageSummary"));
     opBuilder.addInput(tag.asOutput());
     opBuilder.addInput(tensor.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/ScalarSummary.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/ScalarSummary.java
index 416251aa6b1..f2cb1135396 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/ScalarSummary.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/ScalarSummary.java
@@ -21,13 +21,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TString;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Outputs a `Summary` protocol buffer with scalar values.
@@ -47,7 +47,7 @@ public final class ScalarSummary extends RawOp implements Operand<TString> {
    * @return a new instance of ScalarSummary
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> ScalarSummary create(Scope scope, Operand<TString> tags, Operand<T> values) {
+  public static <T extends Tensor & TNumber> ScalarSummary create(Scope scope, Operand<TString> tags, Operand<T> values) {
     OperationBuilder opBuilder = scope.env().opBuilder("ScalarSummary", scope.makeOpName("ScalarSummary"));
     opBuilder.addInput(tags.asOutput());
     opBuilder.addInput(values.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/SummaryWriter.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/SummaryWriter.java
index 131cb6e4a9d..3aff9146666 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/SummaryWriter.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/SummaryWriter.java
@@ -21,15 +21,15 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  */
-public final class SummaryWriter extends RawOp implements Operand<TType> {
+public final class SummaryWriter extends RawOp implements Operand<Tensor> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.summary.SummaryWriter}
@@ -105,8 +105,8 @@ public Output<?> writer() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) writer;
+  public Output<Tensor> asOutput() {
+    return (Output<Tensor>) writer;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/TensorSummary.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/TensorSummary.java
index 85e0f8d533b..386b626778c 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/TensorSummary.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/TensorSummary.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TString;
-import org.tensorflow.types.family.TType;
 
 /**
  * Outputs a `Summary` protocol buffer with a tensor and per-plugin data.
@@ -45,7 +45,7 @@ public final class TensorSummary extends RawOp implements Operand<TString> {
    * @return a new instance of TensorSummary
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> TensorSummary create(Scope scope, Operand<TString> tag, Operand<T> tensor, Operand<TString> serializedSummaryMetadata) {
+  public static <T extends Tensor> TensorSummary create(Scope scope, Operand<TString> tag, Operand<T> tensor, Operand<TString> serializedSummaryMetadata) {
     OperationBuilder opBuilder = scope.env().opBuilder("TensorSummaryV2", scope.makeOpName("TensorSummary"));
     opBuilder.addInput(tag.asOutput());
     opBuilder.addInput(tensor.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/WriteHistogramSummary.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/WriteHistogramSummary.java
index 2069cefafa4..59841f8cc7e 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/WriteHistogramSummary.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/WriteHistogramSummary.java
@@ -20,6 +20,7 @@
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -27,7 +28,6 @@
 import org.tensorflow.types.TInt64;
 import org.tensorflow.types.TString;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  */
@@ -44,7 +44,7 @@ public final class WriteHistogramSummary extends RawOp {
    * @return a new instance of WriteHistogramSummary
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> WriteHistogramSummary create(Scope scope, Operand<?> writer, Operand<TInt64> step, Operand<TString> tag, Operand<T> values) {
+  public static <T extends Tensor & TNumber> WriteHistogramSummary create(Scope scope, Operand<?> writer, Operand<TInt64> step, Operand<TString> tag, Operand<T> values) {
     OperationBuilder opBuilder = scope.env().opBuilder("WriteHistogramSummary", scope.makeOpName("WriteHistogramSummary"));
     opBuilder.addInput(writer.asOutput());
     opBuilder.addInput(step.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/WriteImageSummary.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/WriteImageSummary.java
index 757ddf59a1c..e66bafad757 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/WriteImageSummary.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/WriteImageSummary.java
@@ -20,6 +20,7 @@
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -28,7 +29,6 @@
 import org.tensorflow.types.TString;
 import org.tensorflow.types.TUint8;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  */
@@ -66,7 +66,7 @@ private Options() {
    * @return a new instance of WriteImageSummary
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> WriteImageSummary create(Scope scope, Operand<?> writer, Operand<TInt64> step, Operand<TString> tag, Operand<T> tensor, Operand<TUint8> badColor, Options... options) {
+  public static <T extends Tensor & TNumber> WriteImageSummary create(Scope scope, Operand<?> writer, Operand<TInt64> step, Operand<TString> tag, Operand<T> tensor, Operand<TUint8> badColor, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("WriteImageSummary", scope.makeOpName("WriteImageSummary"));
     opBuilder.addInput(writer.asOutput());
     opBuilder.addInput(step.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/WriteScalarSummary.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/WriteScalarSummary.java
index f173651001a..63e61b54584 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/WriteScalarSummary.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/WriteScalarSummary.java
@@ -20,6 +20,7 @@
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -27,7 +28,6 @@
 import org.tensorflow.types.TInt64;
 import org.tensorflow.types.TString;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  */
@@ -44,7 +44,7 @@ public final class WriteScalarSummary extends RawOp {
    * @return a new instance of WriteScalarSummary
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> WriteScalarSummary create(Scope scope, Operand<?> writer, Operand<TInt64> step, Operand<TString> tag, Operand<T> value) {
+  public static <T extends Tensor & TNumber> WriteScalarSummary create(Scope scope, Operand<?> writer, Operand<TInt64> step, Operand<TString> tag, Operand<T> value) {
     OperationBuilder opBuilder = scope.env().opBuilder("WriteScalarSummary", scope.makeOpName("WriteScalarSummary"));
     opBuilder.addInput(writer.asOutput());
     opBuilder.addInput(step.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/WriteSummary.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/WriteSummary.java
index 5404e593f27..eb431459f27 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/WriteSummary.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/WriteSummary.java
@@ -20,13 +20,13 @@
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
 import org.tensorflow.types.TString;
-import org.tensorflow.types.family.TType;
 
 /**
  */
@@ -44,7 +44,7 @@ public final class WriteSummary extends RawOp {
    * @return a new instance of WriteSummary
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> WriteSummary create(Scope scope, Operand<?> writer, Operand<TInt64> step, Operand<T> tensor, Operand<TString> tag, Operand<TString> summaryMetadata) {
+  public static <T extends Tensor> WriteSummary create(Scope scope, Operand<?> writer, Operand<TInt64> step, Operand<T> tensor, Operand<TString> tag, Operand<TString> summaryMetadata) {
     OperationBuilder opBuilder = scope.env().opBuilder("WriteSummary", scope.makeOpName("WriteSummary"));
     opBuilder.addInput(writer.asOutput());
     opBuilder.addInput(step.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/AllToAll.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/AllToAll.java
index 9bbac83ac2c..ca0c638cd10 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/AllToAll.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/AllToAll.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
-import org.tensorflow.types.family.TType;
 
 /**
  * An Op to exchange data across TPU replicas.
@@ -50,7 +50,7 @@
  * 
  * @param <T> data type for {@code output()} output
  */
-public final class AllToAll<T extends TType> extends RawOp implements Operand<T> {
+public final class AllToAll<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new AllToAll operation.
@@ -67,7 +67,7 @@ public final class AllToAll<T extends TType> extends RawOp implements Operand<T>
    * @return a new instance of AllToAll
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> AllToAll<T> create(Scope scope, Operand<T> input, Operand<TInt32> groupAssignment, Long concatDimension, Long splitDimension, Long splitCount) {
+  public static <T extends Tensor> AllToAll<T> create(Scope scope, Operand<T> input, Operand<TInt32> groupAssignment, Long concatDimension, Long splitDimension, Long splitCount) {
     OperationBuilder opBuilder = scope.env().opBuilder("AllToAll", scope.makeOpName("AllToAll"));
     opBuilder.addInput(input.asOutput());
     opBuilder.addInput(groupAssignment.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/CollectivePermute.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/CollectivePermute.java
index bafa9a893f9..6382cbfab4e 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/CollectivePermute.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/CollectivePermute.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
-import org.tensorflow.types.family.TType;
 
 /**
  * An Op to permute tensors across replicated TPU instances.
@@ -39,7 +39,7 @@
  * 
  * @param <T> data type for {@code output()} output
  */
-public final class CollectivePermute<T extends TType> extends RawOp implements Operand<T> {
+public final class CollectivePermute<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new CollectivePermute operation.
@@ -51,7 +51,7 @@ public final class CollectivePermute<T extends TType> extends RawOp implements O
    * @return a new instance of CollectivePermute
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> CollectivePermute<T> create(Scope scope, Operand<T> input, Operand<TInt32> sourceTargetPairs) {
+  public static <T extends Tensor> CollectivePermute<T> create(Scope scope, Operand<T> input, Operand<TInt32> sourceTargetPairs) {
     OperationBuilder opBuilder = scope.env().opBuilder("CollectivePermute", scope.makeOpName("CollectivePermute"));
     opBuilder.addInput(input.asOutput());
     opBuilder.addInput(sourceTargetPairs.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/CrossReplicaSum.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/CrossReplicaSum.java
index 79dc79410a1..7a1ec781270 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/CrossReplicaSum.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/CrossReplicaSum.java
@@ -21,13 +21,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * An Op to sum inputs across replicated TPU instances.
@@ -41,7 +41,7 @@
  * 
  * @param <T> data type for {@code output()} output
  */
-public final class CrossReplicaSum<T extends TNumber> extends RawOp implements Operand<T> {
+public final class CrossReplicaSum<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new CrossReplicaSum operation.
@@ -54,7 +54,7 @@ public final class CrossReplicaSum<T extends TNumber> extends RawOp implements O
    * @return a new instance of CrossReplicaSum
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> CrossReplicaSum<T> create(Scope scope, Operand<T> input, Operand<TInt32> groupAssignment) {
+  public static <T extends Tensor & TNumber> CrossReplicaSum<T> create(Scope scope, Operand<T> input, Operand<TInt32> groupAssignment) {
     OperationBuilder opBuilder = scope.env().opBuilder("CrossReplicaSum", scope.makeOpName("CrossReplicaSum"));
     opBuilder.addInput(input.asOutput());
     opBuilder.addInput(groupAssignment.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/EnqueueTPUEmbeddingSparseBatch.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/EnqueueTPUEmbeddingSparseBatch.java
index 2cb7dfb674b..8e7f74431c2 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/EnqueueTPUEmbeddingSparseBatch.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/EnqueueTPUEmbeddingSparseBatch.java
@@ -21,6 +21,7 @@
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.Operands;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
@@ -28,7 +29,6 @@
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TString;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * An op that enqueues TPUEmbedding input indices from a SparseTensor.
@@ -99,7 +99,7 @@ private Options() {
    * @return a new instance of EnqueueTPUEmbeddingSparseBatch
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber, U extends TNumber, V extends TNumber> EnqueueTPUEmbeddingSparseBatch create(Scope scope, Iterable<Operand<T>> sampleIndices, Iterable<Operand<U>> embeddingIndices, Iterable<Operand<V>> aggregationWeights, Operand<TString> modeOverride, Options... options) {
+  public static <T extends Tensor & TNumber, U extends Tensor & TNumber, V extends Tensor & TNumber> EnqueueTPUEmbeddingSparseBatch create(Scope scope, Iterable<Operand<T>> sampleIndices, Iterable<Operand<U>> embeddingIndices, Iterable<Operand<V>> aggregationWeights, Operand<TString> modeOverride, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("EnqueueTPUEmbeddingSparseBatch", scope.makeOpName("EnqueueTPUEmbeddingSparseBatch"));
     opBuilder.addInputList(Operands.asOutputs(sampleIndices));
     opBuilder.addInputList(Operands.asOutputs(embeddingIndices));
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/EnqueueTPUEmbeddingSparseTensorBatch.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/EnqueueTPUEmbeddingSparseTensorBatch.java
index 3d93c6a0f71..b9b713fc110 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/EnqueueTPUEmbeddingSparseTensorBatch.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/EnqueueTPUEmbeddingSparseTensorBatch.java
@@ -21,6 +21,7 @@
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.Operands;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
@@ -28,7 +29,6 @@
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TString;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Eases the porting of code that uses tf.nn.embedding_lookup_sparse().
@@ -112,7 +112,7 @@ private Options() {
    * @return a new instance of EnqueueTPUEmbeddingSparseTensorBatch
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber, U extends TNumber, V extends TNumber> EnqueueTPUEmbeddingSparseTensorBatch create(Scope scope, Iterable<Operand<T>> sampleIndices, Iterable<Operand<U>> embeddingIndices, Iterable<Operand<V>> aggregationWeights, Operand<TString> modeOverride, List<Long> tableIds, Options... options) {
+  public static <T extends Tensor & TNumber, U extends Tensor & TNumber, V extends Tensor & TNumber> EnqueueTPUEmbeddingSparseTensorBatch create(Scope scope, Iterable<Operand<T>> sampleIndices, Iterable<Operand<U>> embeddingIndices, Iterable<Operand<V>> aggregationWeights, Operand<TString> modeOverride, List<Long> tableIds, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("EnqueueTPUEmbeddingSparseTensorBatch", scope.makeOpName("EnqueueTPUEmbeddingSparseTensorBatch"));
     opBuilder.addInputList(Operands.asOutputs(sampleIndices));
     opBuilder.addInputList(Operands.asOutputs(embeddingIndices));
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/InfeedDequeue.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/InfeedDequeue.java
index ad4d5f52fa8..e7008a92e99 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/InfeedDequeue.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/InfeedDequeue.java
@@ -22,19 +22,19 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * A placeholder op for a value that will be fed into the computation.
  * 
  * @param <T> data type for {@code output()} output
  */
-public final class InfeedDequeue<T extends TType> extends RawOp implements Operand<T> {
+public final class InfeedDequeue<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new InfeedDequeue operation.
@@ -45,7 +45,7 @@ public final class InfeedDequeue<T extends TType> extends RawOp implements Opera
    * @return a new instance of InfeedDequeue
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> InfeedDequeue<T> create(Scope scope, DataType<T> dtype, Shape shape) {
+  public static <T extends Tensor> InfeedDequeue<T> create(Scope scope, DataType<T> dtype, Shape shape) {
     OperationBuilder opBuilder = scope.env().opBuilder("InfeedDequeue", scope.makeOpName("InfeedDequeue"));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("dtype", dtype);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/InfeedDequeueTuple.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/InfeedDequeueTuple.java
index f471da5b5d7..2a2d5f7ed8c 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/InfeedDequeueTuple.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/InfeedDequeueTuple.java
@@ -25,17 +25,17 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Fetches multiple values from infeed as an XLA tuple.
  */
-public final class InfeedDequeueTuple extends RawOp implements Iterable<Operand<TType>> {
+public final class InfeedDequeueTuple extends RawOp implements Iterable<Operand<Tensor>> {
   
   /**
    * Factory method to create a class wrapping a new InfeedDequeueTuple operation.
@@ -71,7 +71,7 @@ public List<Output<?>> outputs() {
   
   @Override
   @SuppressWarnings({"rawtypes", "unchecked"})
-  public Iterator<Operand<TType>> iterator() {
+  public Iterator<Operand<Tensor>> iterator() {
     return (Iterator) outputs.iterator();
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/InfeedEnqueue.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/InfeedEnqueue.java
index 391d51a9ab0..4e816591eea 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/InfeedEnqueue.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/InfeedEnqueue.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
+import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * An op which feeds a single Tensor value into the computation.
@@ -83,7 +83,7 @@ private Options() {
    * @return a new instance of InfeedEnqueue
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> InfeedEnqueue create(Scope scope, Operand<T> input, Options... options) {
+  public static <T extends Tensor> InfeedEnqueue create(Scope scope, Operand<T> input, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("InfeedEnqueue", scope.makeOpName("InfeedEnqueue"));
     opBuilder.addInput(input.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/OutfeedDequeue.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/OutfeedDequeue.java
index 3be811cd039..645ff872328 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/OutfeedDequeue.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/OutfeedDequeue.java
@@ -22,12 +22,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Retrieves a single tensor from the computation outfeed.
@@ -36,7 +36,7 @@
  * 
  * @param <T> data type for {@code output()} output
  */
-public final class OutfeedDequeue<T extends TType> extends RawOp implements Operand<T> {
+public final class OutfeedDequeue<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.tpu.OutfeedDequeue}
@@ -69,7 +69,7 @@ private Options() {
    * @return a new instance of OutfeedDequeue
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> OutfeedDequeue<T> create(Scope scope, DataType<T> dtype, Shape shape, Options... options) {
+  public static <T extends Tensor> OutfeedDequeue<T> create(Scope scope, DataType<T> dtype, Shape shape, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("OutfeedDequeue", scope.makeOpName("OutfeedDequeue"));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("dtype", dtype);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/OutfeedDequeueTuple.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/OutfeedDequeueTuple.java
index 6b9110232d8..fe6bfa90ea4 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/OutfeedDequeueTuple.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/OutfeedDequeueTuple.java
@@ -25,12 +25,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Retrieve multiple values from the computation outfeed.
@@ -38,7 +38,7 @@
  * This operation will block indefinitely until data is available. Output `i`
  * corresponds to XLA tuple element `i`.
  */
-public final class OutfeedDequeueTuple extends RawOp implements Iterable<Operand<TType>> {
+public final class OutfeedDequeueTuple extends RawOp implements Iterable<Operand<Tensor>> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.tpu.OutfeedDequeueTuple}
@@ -112,7 +112,7 @@ public List<Output<?>> outputs() {
   
   @Override
   @SuppressWarnings({"rawtypes", "unchecked"})
-  public Iterator<Operand<TType>> iterator() {
+  public Iterator<Operand<Tensor>> iterator() {
     return (Iterator) outputs.iterator();
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/OutfeedEnqueue.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/OutfeedEnqueue.java
index 5b5f059a81e..4192f8b8dbb 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/OutfeedEnqueue.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/OutfeedEnqueue.java
@@ -20,11 +20,11 @@
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Enqueue a Tensor on the computation outfeed.
@@ -39,7 +39,7 @@ public final class OutfeedEnqueue extends RawOp {
    * @return a new instance of OutfeedEnqueue
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> OutfeedEnqueue create(Scope scope, Operand<T> input) {
+  public static <T extends Tensor> OutfeedEnqueue create(Scope scope, Operand<T> input) {
     OperationBuilder opBuilder = scope.env().opBuilder("OutfeedEnqueue", scope.makeOpName("OutfeedEnqueue"));
     opBuilder.addInput(input.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/Prelinearize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/Prelinearize.java
index c5c9573fb75..27bb0405fa1 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/Prelinearize.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/Prelinearize.java
@@ -22,17 +22,17 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * An op which linearizes one Tensor value to an opaque variant tensor.
  */
-public final class Prelinearize extends RawOp implements Operand<TType> {
+public final class Prelinearize extends RawOp implements Operand<Tensor> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.tpu.Prelinearize}
@@ -73,7 +73,7 @@ private Options() {
    * @return a new instance of Prelinearize
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> Prelinearize create(Scope scope, Operand<T> input, Options... options) {
+  public static <T extends Tensor> Prelinearize create(Scope scope, Operand<T> input, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("Prelinearize", scope.makeOpName("Prelinearize"));
     opBuilder.addInput(input.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
@@ -118,8 +118,8 @@ public Output<?> output() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) output;
+  public Output<Tensor> asOutput() {
+    return (Output<Tensor>) output;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/PrelinearizeTuple.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/PrelinearizeTuple.java
index 6ff29f7b63a..0b7b48778b9 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/PrelinearizeTuple.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/PrelinearizeTuple.java
@@ -22,18 +22,18 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.Operands;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * An op which linearizes multiple Tensor values to an opaque variant tensor.
  */
-public final class PrelinearizeTuple extends RawOp implements Operand<TType> {
+public final class PrelinearizeTuple extends RawOp implements Operand<Tensor> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.tpu.PrelinearizeTuple}
@@ -108,8 +108,8 @@ public Output<?> output() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) output;
+  public Output<Tensor> asOutput() {
+    return (Output<Tensor>) output;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/TPUReplicatedInput.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/TPUReplicatedInput.java
index 51fe43aea06..58959270d9b 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/TPUReplicatedInput.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/TPUReplicatedInput.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.Operands;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Connects N inputs to an N-way replicated TPU computation.
@@ -45,7 +45,7 @@
  * 
  * @param <T> data type for {@code output()} output
  */
-public final class TPUReplicatedInput<T extends TType> extends RawOp implements Operand<T> {
+public final class TPUReplicatedInput<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.tpu.TPUReplicatedInput}
@@ -93,7 +93,7 @@ private Options() {
    * @return a new instance of TPUReplicatedInput
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> TPUReplicatedInput<T> create(Scope scope, Iterable<Operand<T>> inputs, Options... options) {
+  public static <T extends Tensor> TPUReplicatedInput<T> create(Scope scope, Iterable<Operand<T>> inputs, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("TPUReplicatedInput", scope.makeOpName("TPUReplicatedInput"));
     opBuilder.addInputList(Operands.asOutputs(inputs));
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/TPUReplicatedOutput.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/TPUReplicatedOutput.java
index 24c9418b0c3..36c52cbb6b4 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/TPUReplicatedOutput.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/TPUReplicatedOutput.java
@@ -24,11 +24,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Connects N outputs from an N-way replicated TPU computation.
@@ -45,7 +45,7 @@
  * 
  * @param <T> data type for {@code outputs()} output
  */
-public final class TPUReplicatedOutput<T extends TType> extends RawOp implements Iterable<Operand<T>> {
+public final class TPUReplicatedOutput<T extends Tensor> extends RawOp implements Iterable<Operand<T>> {
   
   /**
    * Factory method to create a class wrapping a new TPUReplicatedOutput operation.
@@ -56,7 +56,7 @@ public final class TPUReplicatedOutput<T extends TType> extends RawOp implements
    * @return a new instance of TPUReplicatedOutput
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> TPUReplicatedOutput<T> create(Scope scope, Operand<T> input, Long numReplicas) {
+  public static <T extends Tensor> TPUReplicatedOutput<T> create(Scope scope, Operand<T> input, Long numReplicas) {
     OperationBuilder opBuilder = scope.env().opBuilder("TPUReplicatedOutput", scope.makeOpName("TPUReplicatedOutput"));
     opBuilder.addInput(input.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/AccumulatorApplyGradient.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/AccumulatorApplyGradient.java
index 504acba4c96..eedb748560b 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/AccumulatorApplyGradient.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/AccumulatorApplyGradient.java
@@ -20,13 +20,13 @@
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
 import org.tensorflow.types.TString;
-import org.tensorflow.types.family.TType;
 
 /**
  * Applies a gradient to a given accumulator.
@@ -46,7 +46,7 @@ public final class AccumulatorApplyGradient extends RawOp {
    * @return a new instance of AccumulatorApplyGradient
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> AccumulatorApplyGradient create(Scope scope, Operand<TString> handle, Operand<TInt64> localStep, Operand<T> gradient) {
+  public static <T extends Tensor> AccumulatorApplyGradient create(Scope scope, Operand<TString> handle, Operand<TInt64> localStep, Operand<T> gradient) {
     OperationBuilder opBuilder = scope.env().opBuilder("AccumulatorApplyGradient", scope.makeOpName("AccumulatorApplyGradient"));
     opBuilder.addInput(handle.asOutput());
     opBuilder.addInput(localStep.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/AccumulatorTakeGradient.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/AccumulatorTakeGradient.java
index 51de5485845..95b394d3c45 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/AccumulatorTakeGradient.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/AccumulatorTakeGradient.java
@@ -22,13 +22,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
 import org.tensorflow.types.TString;
-import org.tensorflow.types.family.TType;
 
 /**
  * Extracts the average gradient in the given ConditionalAccumulator.
@@ -42,7 +42,7 @@
  * @param <T> data type for {@code average()} output
  */
 @Operator(group = "train")
-public final class AccumulatorTakeGradient<T extends TType> extends RawOp implements Operand<T> {
+public final class AccumulatorTakeGradient<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new AccumulatorTakeGradient operation.
@@ -55,7 +55,7 @@ public final class AccumulatorTakeGradient<T extends TType> extends RawOp implem
    * @return a new instance of AccumulatorTakeGradient
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> AccumulatorTakeGradient<T> create(Scope scope, Operand<TString> handle, Operand<TInt32> numRequired, DataType<T> dtype) {
+  public static <T extends Tensor> AccumulatorTakeGradient<T> create(Scope scope, Operand<TString> handle, Operand<TInt32> numRequired, DataType<T> dtype) {
     OperationBuilder opBuilder = scope.env().opBuilder("AccumulatorTakeGradient", scope.makeOpName("AccumulatorTakeGradient"));
     opBuilder.addInput(handle.asOutput());
     opBuilder.addInput(numRequired.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyAdaMax.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyAdaMax.java
index 1b24ac1fe26..62de9bcf1fc 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyAdaMax.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyAdaMax.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Update '*var' according to the AdaMax algorithm.
@@ -36,7 +36,7 @@
  * 
  * @param <T> data type for {@code out()} output
  */
-public final class ApplyAdaMax<T extends TType> extends RawOp implements Operand<T> {
+public final class ApplyAdaMax<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.train.ApplyAdaMax}
@@ -76,7 +76,7 @@ private Options() {
    * @return a new instance of ApplyAdaMax
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> ApplyAdaMax<T> create(Scope scope, Operand<T> var, Operand<T> m, Operand<T> v, Operand<T> beta1Power, Operand<T> lr, Operand<T> beta1, Operand<T> beta2, Operand<T> epsilon, Operand<T> grad, Options... options) {
+  public static <T extends Tensor> ApplyAdaMax<T> create(Scope scope, Operand<T> var, Operand<T> m, Operand<T> v, Operand<T> beta1Power, Operand<T> lr, Operand<T> beta1, Operand<T> beta2, Operand<T> epsilon, Operand<T> grad, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ApplyAdaMax", scope.makeOpName("ApplyAdaMax"));
     opBuilder.addInput(var.asOutput());
     opBuilder.addInput(m.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyAdadelta.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyAdadelta.java
index 7e68618da7d..0a75fc0d386 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyAdadelta.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyAdadelta.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Update '*var' according to the adadelta scheme.
@@ -38,7 +38,7 @@
  * @param <T> data type for {@code out()} output
  */
 @Operator(group = "train")
-public final class ApplyAdadelta<T extends TType> extends RawOp implements Operand<T> {
+public final class ApplyAdadelta<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.train.ApplyAdadelta}
@@ -75,7 +75,7 @@ private Options() {
    * @return a new instance of ApplyAdadelta
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> ApplyAdadelta<T> create(Scope scope, Operand<T> var, Operand<T> accum, Operand<T> accumUpdate, Operand<T> lr, Operand<T> rho, Operand<T> epsilon, Operand<T> grad, Options... options) {
+  public static <T extends Tensor> ApplyAdadelta<T> create(Scope scope, Operand<T> var, Operand<T> accum, Operand<T> accumUpdate, Operand<T> lr, Operand<T> rho, Operand<T> epsilon, Operand<T> grad, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ApplyAdadelta", scope.makeOpName("ApplyAdadelta"));
     opBuilder.addInput(var.asOutput());
     opBuilder.addInput(accum.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyAdagrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyAdagrad.java
index c8311ac541a..0cf6eed8c09 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyAdagrad.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyAdagrad.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Update '*var' according to the adagrad scheme.
@@ -36,7 +36,7 @@
  * @param <T> data type for {@code out()} output
  */
 @Operator(group = "train")
-public final class ApplyAdagrad<T extends TType> extends RawOp implements Operand<T> {
+public final class ApplyAdagrad<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.train.ApplyAdagrad}
@@ -80,7 +80,7 @@ private Options() {
    * @return a new instance of ApplyAdagrad
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> ApplyAdagrad<T> create(Scope scope, Operand<T> var, Operand<T> accum, Operand<T> lr, Operand<T> grad, Options... options) {
+  public static <T extends Tensor> ApplyAdagrad<T> create(Scope scope, Operand<T> var, Operand<T> accum, Operand<T> lr, Operand<T> grad, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ApplyAdagrad", scope.makeOpName("ApplyAdagrad"));
     opBuilder.addInput(var.asOutput());
     opBuilder.addInput(accum.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyAdagradDa.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyAdagradDa.java
index 3bb9f9cae4b..7b09995fa1c 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyAdagradDa.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyAdagradDa.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
 
 /**
  * Update '*var' according to the proximal adagrad scheme.
@@ -34,7 +34,7 @@
  * @param <T> data type for {@code out()} output
  */
 @Operator(group = "train")
-public final class ApplyAdagradDa<T extends TType> extends RawOp implements Operand<T> {
+public final class ApplyAdagradDa<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.train.ApplyAdagradDa}
@@ -72,7 +72,7 @@ private Options() {
    * @return a new instance of ApplyAdagradDa
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> ApplyAdagradDa<T> create(Scope scope, Operand<T> var, Operand<T> gradientAccumulator, Operand<T> gradientSquaredAccumulator, Operand<T> grad, Operand<T> lr, Operand<T> l1, Operand<T> l2, Operand<TInt64> globalStep, Options... options) {
+  public static <T extends Tensor> ApplyAdagradDa<T> create(Scope scope, Operand<T> var, Operand<T> gradientAccumulator, Operand<T> gradientSquaredAccumulator, Operand<T> grad, Operand<T> lr, Operand<T> l1, Operand<T> l2, Operand<TInt64> globalStep, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ApplyAdagradDA", scope.makeOpName("ApplyAdagradDa"));
     opBuilder.addInput(var.asOutput());
     opBuilder.addInput(gradientAccumulator.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyAdagradV2.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyAdagradV2.java
index da399e31eda..6b50f9ffed3 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyAdagradV2.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyAdagradV2.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Update '*var' according to the adagrad scheme.
@@ -35,7 +35,7 @@
  * 
  * @param <T> data type for {@code out()} output
  */
-public final class ApplyAdagradV2<T extends TType> extends RawOp implements Operand<T> {
+public final class ApplyAdagradV2<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.train.ApplyAdagradV2}
@@ -80,7 +80,7 @@ private Options() {
    * @return a new instance of ApplyAdagradV2
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> ApplyAdagradV2<T> create(Scope scope, Operand<T> var, Operand<T> accum, Operand<T> lr, Operand<T> epsilon, Operand<T> grad, Options... options) {
+  public static <T extends Tensor> ApplyAdagradV2<T> create(Scope scope, Operand<T> var, Operand<T> accum, Operand<T> lr, Operand<T> epsilon, Operand<T> grad, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ApplyAdagradV2", scope.makeOpName("ApplyAdagradV2"));
     opBuilder.addInput(var.asOutput());
     opBuilder.addInput(accum.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyAdam.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyAdam.java
index 31d98fa299b..b662b1f0203 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyAdam.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyAdam.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Update '*var' according to the Adam algorithm.
@@ -38,7 +38,7 @@
  * @param <T> data type for {@code out()} output
  */
 @Operator(group = "train")
-public final class ApplyAdam<T extends TType> extends RawOp implements Operand<T> {
+public final class ApplyAdam<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.train.ApplyAdam}
@@ -88,7 +88,7 @@ private Options() {
    * @return a new instance of ApplyAdam
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> ApplyAdam<T> create(Scope scope, Operand<T> var, Operand<T> m, Operand<T> v, Operand<T> beta1Power, Operand<T> beta2Power, Operand<T> lr, Operand<T> beta1, Operand<T> beta2, Operand<T> epsilon, Operand<T> grad, Options... options) {
+  public static <T extends Tensor> ApplyAdam<T> create(Scope scope, Operand<T> var, Operand<T> m, Operand<T> v, Operand<T> beta1Power, Operand<T> beta2Power, Operand<T> lr, Operand<T> beta1, Operand<T> beta2, Operand<T> epsilon, Operand<T> grad, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ApplyAdam", scope.makeOpName("ApplyAdam"));
     opBuilder.addInput(var.asOutput());
     opBuilder.addInput(m.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyAddSign.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyAddSign.java
index 1fe2cd756af..970172dcf14 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyAddSign.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyAddSign.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Update '*var' according to the AddSign update.
@@ -37,7 +37,7 @@
  * @param <T> data type for {@code out()} output
  */
 @Operator(group = "train")
-public final class ApplyAddSign<T extends TType> extends RawOp implements Operand<T> {
+public final class ApplyAddSign<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.train.ApplyAddSign}
@@ -75,7 +75,7 @@ private Options() {
    * @return a new instance of ApplyAddSign
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> ApplyAddSign<T> create(Scope scope, Operand<T> var, Operand<T> m, Operand<T> lr, Operand<T> alpha, Operand<T> signDecay, Operand<T> beta, Operand<T> grad, Options... options) {
+  public static <T extends Tensor> ApplyAddSign<T> create(Scope scope, Operand<T> var, Operand<T> m, Operand<T> lr, Operand<T> alpha, Operand<T> signDecay, Operand<T> beta, Operand<T> grad, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ApplyAddSign", scope.makeOpName("ApplyAddSign"));
     opBuilder.addInput(var.asOutput());
     opBuilder.addInput(m.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyCenteredRmsProp.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyCenteredRmsProp.java
index 22aeeed0d91..97a43dedd51 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyCenteredRmsProp.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyCenteredRmsProp.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Update '*var' according to the centered RMSProp algorithm.
@@ -52,7 +52,7 @@
  * @param <T> data type for {@code out()} output
  */
 @Operator(group = "train")
-public final class ApplyCenteredRmsProp<T extends TType> extends RawOp implements Operand<T> {
+public final class ApplyCenteredRmsProp<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.train.ApplyCenteredRmsProp}
@@ -92,7 +92,7 @@ private Options() {
    * @return a new instance of ApplyCenteredRmsProp
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> ApplyCenteredRmsProp<T> create(Scope scope, Operand<T> var, Operand<T> mg, Operand<T> ms, Operand<T> mom, Operand<T> lr, Operand<T> rho, Operand<T> momentum, Operand<T> epsilon, Operand<T> grad, Options... options) {
+  public static <T extends Tensor> ApplyCenteredRmsProp<T> create(Scope scope, Operand<T> var, Operand<T> mg, Operand<T> ms, Operand<T> mom, Operand<T> lr, Operand<T> rho, Operand<T> momentum, Operand<T> epsilon, Operand<T> grad, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ApplyCenteredRMSProp", scope.makeOpName("ApplyCenteredRmsProp"));
     opBuilder.addInput(var.asOutput());
     opBuilder.addInput(mg.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyFtrl.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyFtrl.java
index f85cf2769c0..368fa40cb87 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyFtrl.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyFtrl.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Update '*var' according to the Ftrl-proximal scheme.
@@ -41,7 +41,7 @@
  * @param <T> data type for {@code out()} output
  */
 @Operator(group = "train")
-public final class ApplyFtrl<T extends TType> extends RawOp implements Operand<T> {
+public final class ApplyFtrl<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.train.ApplyFtrl}
@@ -90,7 +90,7 @@ private Options() {
    * @return a new instance of ApplyFtrl
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> ApplyFtrl<T> create(Scope scope, Operand<T> var, Operand<T> accum, Operand<T> linear, Operand<T> grad, Operand<T> lr, Operand<T> l1, Operand<T> l2, Operand<T> l2Shrinkage, Operand<T> lrPower, Options... options) {
+  public static <T extends Tensor> ApplyFtrl<T> create(Scope scope, Operand<T> var, Operand<T> accum, Operand<T> linear, Operand<T> grad, Operand<T> lr, Operand<T> l1, Operand<T> l2, Operand<T> l2Shrinkage, Operand<T> lrPower, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ApplyFtrlV2", scope.makeOpName("ApplyFtrl"));
     opBuilder.addInput(var.asOutput());
     opBuilder.addInput(accum.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyGradientDescent.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyGradientDescent.java
index 729312860fd..4cfb6878c25 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyGradientDescent.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyGradientDescent.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Update '*var' by subtracting 'alpha' * 'delta' from it.
@@ -33,7 +33,7 @@
  * @param <T> data type for {@code out()} output
  */
 @Operator(group = "train")
-public final class ApplyGradientDescent<T extends TType> extends RawOp implements Operand<T> {
+public final class ApplyGradientDescent<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.train.ApplyGradientDescent}
@@ -66,7 +66,7 @@ private Options() {
    * @return a new instance of ApplyGradientDescent
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> ApplyGradientDescent<T> create(Scope scope, Operand<T> var, Operand<T> alpha, Operand<T> delta, Options... options) {
+  public static <T extends Tensor> ApplyGradientDescent<T> create(Scope scope, Operand<T> var, Operand<T> alpha, Operand<T> delta, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ApplyGradientDescent", scope.makeOpName("ApplyGradientDescent"));
     opBuilder.addInput(var.asOutput());
     opBuilder.addInput(alpha.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyMomentum.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyMomentum.java
index d7885da2e32..24e221691e1 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyMomentum.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyMomentum.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Update '*var' according to the momentum scheme.
@@ -38,7 +38,7 @@
  * @param <T> data type for {@code out()} output
  */
 @Operator(group = "train")
-public final class ApplyMomentum<T extends TType> extends RawOp implements Operand<T> {
+public final class ApplyMomentum<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.train.ApplyMomentum}
@@ -85,7 +85,7 @@ private Options() {
    * @return a new instance of ApplyMomentum
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> ApplyMomentum<T> create(Scope scope, Operand<T> var, Operand<T> accum, Operand<T> lr, Operand<T> grad, Operand<T> momentum, Options... options) {
+  public static <T extends Tensor> ApplyMomentum<T> create(Scope scope, Operand<T> var, Operand<T> accum, Operand<T> lr, Operand<T> grad, Operand<T> momentum, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ApplyMomentum", scope.makeOpName("ApplyMomentum"));
     opBuilder.addInput(var.asOutput());
     opBuilder.addInput(accum.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyPowerSign.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyPowerSign.java
index d508e939714..a1a3d8000bc 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyPowerSign.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyPowerSign.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Update '*var' according to the AddSign update.
@@ -37,7 +37,7 @@
  * @param <T> data type for {@code out()} output
  */
 @Operator(group = "train")
-public final class ApplyPowerSign<T extends TType> extends RawOp implements Operand<T> {
+public final class ApplyPowerSign<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.train.ApplyPowerSign}
@@ -75,7 +75,7 @@ private Options() {
    * @return a new instance of ApplyPowerSign
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> ApplyPowerSign<T> create(Scope scope, Operand<T> var, Operand<T> m, Operand<T> lr, Operand<T> logbase, Operand<T> signDecay, Operand<T> beta, Operand<T> grad, Options... options) {
+  public static <T extends Tensor> ApplyPowerSign<T> create(Scope scope, Operand<T> var, Operand<T> m, Operand<T> lr, Operand<T> logbase, Operand<T> signDecay, Operand<T> beta, Operand<T> grad, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ApplyPowerSign", scope.makeOpName("ApplyPowerSign"));
     opBuilder.addInput(var.asOutput());
     opBuilder.addInput(m.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyProximalAdagrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyProximalAdagrad.java
index bd710eb08ca..afec66ccab6 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyProximalAdagrad.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyProximalAdagrad.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Update '*var' and '*accum' according to FOBOS with Adagrad learning rate.
@@ -37,7 +37,7 @@
  * @param <T> data type for {@code out()} output
  */
 @Operator(group = "train")
-public final class ApplyProximalAdagrad<T extends TType> extends RawOp implements Operand<T> {
+public final class ApplyProximalAdagrad<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.train.ApplyProximalAdagrad}
@@ -73,7 +73,7 @@ private Options() {
    * @return a new instance of ApplyProximalAdagrad
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> ApplyProximalAdagrad<T> create(Scope scope, Operand<T> var, Operand<T> accum, Operand<T> lr, Operand<T> l1, Operand<T> l2, Operand<T> grad, Options... options) {
+  public static <T extends Tensor> ApplyProximalAdagrad<T> create(Scope scope, Operand<T> var, Operand<T> accum, Operand<T> lr, Operand<T> l1, Operand<T> l2, Operand<T> grad, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ApplyProximalAdagrad", scope.makeOpName("ApplyProximalAdagrad"));
     opBuilder.addInput(var.asOutput());
     opBuilder.addInput(accum.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyProximalGradientDescent.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyProximalGradientDescent.java
index eb5e9a96dc9..b3040099f58 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyProximalGradientDescent.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyProximalGradientDescent.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Update '*var' as FOBOS algorithm with fixed learning rate.
@@ -36,7 +36,7 @@
  * @param <T> data type for {@code out()} output
  */
 @Operator(group = "train")
-public final class ApplyProximalGradientDescent<T extends TType> extends RawOp implements Operand<T> {
+public final class ApplyProximalGradientDescent<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.train.ApplyProximalGradientDescent}
@@ -71,7 +71,7 @@ private Options() {
    * @return a new instance of ApplyProximalGradientDescent
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> ApplyProximalGradientDescent<T> create(Scope scope, Operand<T> var, Operand<T> alpha, Operand<T> l1, Operand<T> l2, Operand<T> delta, Options... options) {
+  public static <T extends Tensor> ApplyProximalGradientDescent<T> create(Scope scope, Operand<T> var, Operand<T> alpha, Operand<T> l1, Operand<T> l2, Operand<T> delta, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ApplyProximalGradientDescent", scope.makeOpName("ApplyProximalGradientDescent"));
     opBuilder.addInput(var.asOutput());
     opBuilder.addInput(alpha.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyRmsProp.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyRmsProp.java
index 388fc062c5d..2b1bd6bb0b1 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyRmsProp.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyRmsProp.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Update '*var' according to the RMSProp algorithm.
@@ -44,7 +44,7 @@
  * @param <T> data type for {@code out()} output
  */
 @Operator(group = "train")
-public final class ApplyRmsProp<T extends TType> extends RawOp implements Operand<T> {
+public final class ApplyRmsProp<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.train.ApplyRmsProp}
@@ -83,7 +83,7 @@ private Options() {
    * @return a new instance of ApplyRmsProp
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> ApplyRmsProp<T> create(Scope scope, Operand<T> var, Operand<T> ms, Operand<T> mom, Operand<T> lr, Operand<T> rho, Operand<T> momentum, Operand<T> epsilon, Operand<T> grad, Options... options) {
+  public static <T extends Tensor> ApplyRmsProp<T> create(Scope scope, Operand<T> var, Operand<T> ms, Operand<T> mom, Operand<T> lr, Operand<T> rho, Operand<T> momentum, Operand<T> epsilon, Operand<T> grad, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ApplyRMSProp", scope.makeOpName("ApplyRmsProp"));
     opBuilder.addInput(var.asOutput());
     opBuilder.addInput(ms.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/BatchMatMul.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/BatchMatMul.java
index 3b6da5b0811..beac69ae46d 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/BatchMatMul.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/BatchMatMul.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Multiplies slices of two tensors in batches.
@@ -57,7 +57,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "train")
-public final class BatchMatMul<T extends TType> extends RawOp implements Operand<T> {
+public final class BatchMatMul<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.train.BatchMatMul}
@@ -97,7 +97,7 @@ private Options() {
    * @return a new instance of BatchMatMul
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> BatchMatMul<T> create(Scope scope, Operand<T> x, Operand<T> y, Options... options) {
+  public static <T extends Tensor> BatchMatMul<T> create(Scope scope, Operand<T> x, Operand<T> y, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("BatchMatMulV2", scope.makeOpName("BatchMatMul"));
     opBuilder.addInput(x.asOutput());
     opBuilder.addInput(y.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ConditionalAccumulator.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ConditionalAccumulator.java
index f35e9f5001f..4cad6d75ff2 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ConditionalAccumulator.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ConditionalAccumulator.java
@@ -22,13 +22,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TString;
-import org.tensorflow.types.family.TType;
 
 /**
  * A conditional accumulator for aggregating gradients.
@@ -92,7 +92,7 @@ private Options() {
    * @return a new instance of ConditionalAccumulator
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> ConditionalAccumulator create(Scope scope, DataType<T> dtype, Shape shape, Options... options) {
+  public static <T extends Tensor> ConditionalAccumulator create(Scope scope, DataType<T> dtype, Shape shape, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ConditionalAccumulator", scope.makeOpName("ConditionalAccumulator"));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("dtype", dtype);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/PreventGradient.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/PreventGradient.java
index 03d342477a9..8911db15871 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/PreventGradient.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/PreventGradient.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * An identity op that triggers an error if a gradient is requested.
@@ -41,7 +41,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "train")
-public final class PreventGradient<T extends TType> extends RawOp implements Operand<T> {
+public final class PreventGradient<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.train.PreventGradient}
@@ -72,7 +72,7 @@ private Options() {
    * @return a new instance of PreventGradient
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> PreventGradient<T> create(Scope scope, Operand<T> input, Options... options) {
+  public static <T extends Tensor> PreventGradient<T> create(Scope scope, Operand<T> input, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("PreventGradient", scope.makeOpName("PreventGradient"));
     opBuilder.addInput(input.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceAccumulatorApplyGradient.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceAccumulatorApplyGradient.java
index 560c3400c3b..7631a6f32b4 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceAccumulatorApplyGradient.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceAccumulatorApplyGradient.java
@@ -20,12 +20,12 @@
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
 
 /**
  * Applies a gradient to a given accumulator.
@@ -44,7 +44,7 @@ public final class ResourceAccumulatorApplyGradient extends RawOp {
    * @return a new instance of ResourceAccumulatorApplyGradient
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> ResourceAccumulatorApplyGradient create(Scope scope, Operand<?> handle, Operand<TInt64> localStep, Operand<T> gradient) {
+  public static <T extends Tensor> ResourceAccumulatorApplyGradient create(Scope scope, Operand<?> handle, Operand<TInt64> localStep, Operand<T> gradient) {
     OperationBuilder opBuilder = scope.env().opBuilder("ResourceAccumulatorApplyGradient", scope.makeOpName("ResourceAccumulatorApplyGradient"));
     opBuilder.addInput(handle.asOutput());
     opBuilder.addInput(localStep.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceAccumulatorTakeGradient.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceAccumulatorTakeGradient.java
index c27fb1db524..2f31831fff8 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceAccumulatorTakeGradient.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceAccumulatorTakeGradient.java
@@ -22,12 +22,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
-import org.tensorflow.types.family.TType;
 
 /**
  * Extracts the average gradient in the given ConditionalAccumulator.
@@ -40,7 +40,7 @@
  * 
  * @param <T> data type for {@code average()} output
  */
-public final class ResourceAccumulatorTakeGradient<T extends TType> extends RawOp implements Operand<T> {
+public final class ResourceAccumulatorTakeGradient<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new ResourceAccumulatorTakeGradient operation.
@@ -53,7 +53,7 @@ public final class ResourceAccumulatorTakeGradient<T extends TType> extends RawO
    * @return a new instance of ResourceAccumulatorTakeGradient
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> ResourceAccumulatorTakeGradient<T> create(Scope scope, Operand<?> handle, Operand<TInt32> numRequired, DataType<T> dtype) {
+  public static <T extends Tensor> ResourceAccumulatorTakeGradient<T> create(Scope scope, Operand<?> handle, Operand<TInt32> numRequired, DataType<T> dtype) {
     OperationBuilder opBuilder = scope.env().opBuilder("ResourceAccumulatorTakeGradient", scope.makeOpName("ResourceAccumulatorTakeGradient"));
     opBuilder.addInput(handle.asOutput());
     opBuilder.addInput(numRequired.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyAdaMax.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyAdaMax.java
index 169da75fecd..eb4720db721 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyAdaMax.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyAdaMax.java
@@ -20,11 +20,11 @@
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Update '*var' according to the AdaMax algorithm.
@@ -73,7 +73,7 @@ private Options() {
    * @return a new instance of ResourceApplyAdaMax
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> ResourceApplyAdaMax create(Scope scope, Operand<?> var, Operand<?> m, Operand<?> v, Operand<T> beta1Power, Operand<T> lr, Operand<T> beta1, Operand<T> beta2, Operand<T> epsilon, Operand<T> grad, Options... options) {
+  public static <T extends Tensor> ResourceApplyAdaMax create(Scope scope, Operand<?> var, Operand<?> m, Operand<?> v, Operand<T> beta1Power, Operand<T> lr, Operand<T> beta1, Operand<T> beta2, Operand<T> epsilon, Operand<T> grad, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ResourceApplyAdaMax", scope.makeOpName("ResourceApplyAdaMax"));
     opBuilder.addInput(var.asOutput());
     opBuilder.addInput(m.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyAdadelta.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyAdadelta.java
index 4323a39de45..19687ce7090 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyAdadelta.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyAdadelta.java
@@ -20,11 +20,11 @@
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Update '*var' according to the adadelta scheme.
@@ -72,7 +72,7 @@ private Options() {
    * @return a new instance of ResourceApplyAdadelta
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> ResourceApplyAdadelta create(Scope scope, Operand<?> var, Operand<?> accum, Operand<?> accumUpdate, Operand<T> lr, Operand<T> rho, Operand<T> epsilon, Operand<T> grad, Options... options) {
+  public static <T extends Tensor> ResourceApplyAdadelta create(Scope scope, Operand<?> var, Operand<?> accum, Operand<?> accumUpdate, Operand<T> lr, Operand<T> rho, Operand<T> epsilon, Operand<T> grad, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ResourceApplyAdadelta", scope.makeOpName("ResourceApplyAdadelta"));
     opBuilder.addInput(var.asOutput());
     opBuilder.addInput(accum.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyAdagrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyAdagrad.java
index b60ad1fc0e0..2ab492152d6 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyAdagrad.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyAdagrad.java
@@ -20,11 +20,11 @@
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Update '*var' according to the adagrad scheme.
@@ -77,7 +77,7 @@ private Options() {
    * @return a new instance of ResourceApplyAdagrad
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> ResourceApplyAdagrad create(Scope scope, Operand<?> var, Operand<?> accum, Operand<T> lr, Operand<T> epsilon, Operand<T> grad, Options... options) {
+  public static <T extends Tensor> ResourceApplyAdagrad create(Scope scope, Operand<?> var, Operand<?> accum, Operand<T> lr, Operand<T> epsilon, Operand<T> grad, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ResourceApplyAdagradV2", scope.makeOpName("ResourceApplyAdagrad"));
     opBuilder.addInput(var.asOutput());
     opBuilder.addInput(accum.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyAdagradDa.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyAdagradDa.java
index 7c6f06634ef..a2ea0b0b917 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyAdagradDa.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyAdagradDa.java
@@ -20,12 +20,12 @@
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
 
 /**
  * Update '*var' according to the proximal adagrad scheme.
@@ -69,7 +69,7 @@ private Options() {
    * @return a new instance of ResourceApplyAdagradDa
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> ResourceApplyAdagradDa create(Scope scope, Operand<?> var, Operand<?> gradientAccumulator, Operand<?> gradientSquaredAccumulator, Operand<T> grad, Operand<T> lr, Operand<T> l1, Operand<T> l2, Operand<TInt64> globalStep, Options... options) {
+  public static <T extends Tensor> ResourceApplyAdagradDa create(Scope scope, Operand<?> var, Operand<?> gradientAccumulator, Operand<?> gradientSquaredAccumulator, Operand<T> grad, Operand<T> lr, Operand<T> l1, Operand<T> l2, Operand<TInt64> globalStep, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ResourceApplyAdagradDA", scope.makeOpName("ResourceApplyAdagradDa"));
     opBuilder.addInput(var.asOutput());
     opBuilder.addInput(gradientAccumulator.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyAdam.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyAdam.java
index 4a1aea5d355..7acf59a7c16 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyAdam.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyAdam.java
@@ -20,11 +20,11 @@
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Update '*var' according to the Adam algorithm.
@@ -85,7 +85,7 @@ private Options() {
    * @return a new instance of ResourceApplyAdam
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> ResourceApplyAdam create(Scope scope, Operand<?> var, Operand<?> m, Operand<?> v, Operand<T> beta1Power, Operand<T> beta2Power, Operand<T> lr, Operand<T> beta1, Operand<T> beta2, Operand<T> epsilon, Operand<T> grad, Options... options) {
+  public static <T extends Tensor> ResourceApplyAdam create(Scope scope, Operand<?> var, Operand<?> m, Operand<?> v, Operand<T> beta1Power, Operand<T> beta2Power, Operand<T> lr, Operand<T> beta1, Operand<T> beta2, Operand<T> epsilon, Operand<T> grad, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ResourceApplyAdam", scope.makeOpName("ResourceApplyAdam"));
     opBuilder.addInput(var.asOutput());
     opBuilder.addInput(m.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyAdamWithAmsgrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyAdamWithAmsgrad.java
index a436bc7fdd2..6d1806cd6c8 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyAdamWithAmsgrad.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyAdamWithAmsgrad.java
@@ -20,11 +20,11 @@
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Update '*var' according to the Adam algorithm.
@@ -78,7 +78,7 @@ private Options() {
    * @return a new instance of ResourceApplyAdamWithAmsgrad
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> ResourceApplyAdamWithAmsgrad create(Scope scope, Operand<?> var, Operand<?> m, Operand<?> v, Operand<?> vhat, Operand<T> beta1Power, Operand<T> beta2Power, Operand<T> lr, Operand<T> beta1, Operand<T> beta2, Operand<T> epsilon, Operand<T> grad, Options... options) {
+  public static <T extends Tensor> ResourceApplyAdamWithAmsgrad create(Scope scope, Operand<?> var, Operand<?> m, Operand<?> v, Operand<?> vhat, Operand<T> beta1Power, Operand<T> beta2Power, Operand<T> lr, Operand<T> beta1, Operand<T> beta2, Operand<T> epsilon, Operand<T> grad, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ResourceApplyAdamWithAmsgrad", scope.makeOpName("ResourceApplyAdamWithAmsgrad"));
     opBuilder.addInput(var.asOutput());
     opBuilder.addInput(m.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyAddSign.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyAddSign.java
index 85c9c587979..f8670ff14c5 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyAddSign.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyAddSign.java
@@ -20,11 +20,11 @@
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Update '*var' according to the AddSign update.
@@ -72,7 +72,7 @@ private Options() {
    * @return a new instance of ResourceApplyAddSign
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> ResourceApplyAddSign create(Scope scope, Operand<?> var, Operand<?> m, Operand<T> lr, Operand<T> alpha, Operand<T> signDecay, Operand<T> beta, Operand<T> grad, Options... options) {
+  public static <T extends Tensor> ResourceApplyAddSign create(Scope scope, Operand<?> var, Operand<?> m, Operand<T> lr, Operand<T> alpha, Operand<T> signDecay, Operand<T> beta, Operand<T> grad, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ResourceApplyAddSign", scope.makeOpName("ResourceApplyAddSign"));
     opBuilder.addInput(var.asOutput());
     opBuilder.addInput(m.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyCenteredRmsProp.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyCenteredRmsProp.java
index 6fc3a8a02ff..e1499e57edd 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyCenteredRmsProp.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyCenteredRmsProp.java
@@ -20,11 +20,11 @@
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Update '*var' according to the centered RMSProp algorithm.
@@ -89,7 +89,7 @@ private Options() {
    * @return a new instance of ResourceApplyCenteredRmsProp
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> ResourceApplyCenteredRmsProp create(Scope scope, Operand<?> var, Operand<?> mg, Operand<?> ms, Operand<?> mom, Operand<T> lr, Operand<T> rho, Operand<T> momentum, Operand<T> epsilon, Operand<T> grad, Options... options) {
+  public static <T extends Tensor> ResourceApplyCenteredRmsProp create(Scope scope, Operand<?> var, Operand<?> mg, Operand<?> ms, Operand<?> mom, Operand<T> lr, Operand<T> rho, Operand<T> momentum, Operand<T> epsilon, Operand<T> grad, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ResourceApplyCenteredRMSProp", scope.makeOpName("ResourceApplyCenteredRmsProp"));
     opBuilder.addInput(var.asOutput());
     opBuilder.addInput(mg.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyFtrl.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyFtrl.java
index e69b6b99959..acb8879bc29 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyFtrl.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyFtrl.java
@@ -20,11 +20,11 @@
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Update '*var' according to the Ftrl-proximal scheme.
@@ -87,7 +87,7 @@ private Options() {
    * @return a new instance of ResourceApplyFtrl
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> ResourceApplyFtrl create(Scope scope, Operand<?> var, Operand<?> accum, Operand<?> linear, Operand<T> grad, Operand<T> lr, Operand<T> l1, Operand<T> l2, Operand<T> l2Shrinkage, Operand<T> lrPower, Options... options) {
+  public static <T extends Tensor> ResourceApplyFtrl create(Scope scope, Operand<?> var, Operand<?> accum, Operand<?> linear, Operand<T> grad, Operand<T> lr, Operand<T> l1, Operand<T> l2, Operand<T> l2Shrinkage, Operand<T> lrPower, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ResourceApplyFtrlV2", scope.makeOpName("ResourceApplyFtrl"));
     opBuilder.addInput(var.asOutput());
     opBuilder.addInput(accum.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyGradientDescent.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyGradientDescent.java
index f33c6b9ca87..701d8691eff 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyGradientDescent.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyGradientDescent.java
@@ -20,11 +20,11 @@
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Update '*var' by subtracting 'alpha' * 'delta' from it.
@@ -63,7 +63,7 @@ private Options() {
    * @return a new instance of ResourceApplyGradientDescent
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> ResourceApplyGradientDescent create(Scope scope, Operand<?> var, Operand<T> alpha, Operand<T> delta, Options... options) {
+  public static <T extends Tensor> ResourceApplyGradientDescent create(Scope scope, Operand<?> var, Operand<T> alpha, Operand<T> delta, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ResourceApplyGradientDescent", scope.makeOpName("ResourceApplyGradientDescent"));
     opBuilder.addInput(var.asOutput());
     opBuilder.addInput(alpha.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyKerasMomentum.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyKerasMomentum.java
index 3922439dcad..bb00f90ece5 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyKerasMomentum.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyKerasMomentum.java
@@ -20,11 +20,11 @@
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Update '*var' according to the momentum scheme.
@@ -82,7 +82,7 @@ private Options() {
    * @return a new instance of ResourceApplyKerasMomentum
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> ResourceApplyKerasMomentum create(Scope scope, Operand<?> var, Operand<?> accum, Operand<T> lr, Operand<T> grad, Operand<T> momentum, Options... options) {
+  public static <T extends Tensor> ResourceApplyKerasMomentum create(Scope scope, Operand<?> var, Operand<?> accum, Operand<T> lr, Operand<T> grad, Operand<T> momentum, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ResourceApplyKerasMomentum", scope.makeOpName("ResourceApplyKerasMomentum"));
     opBuilder.addInput(var.asOutput());
     opBuilder.addInput(accum.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyMomentum.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyMomentum.java
index c554c8a939d..4ab1f15ca92 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyMomentum.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyMomentum.java
@@ -20,11 +20,11 @@
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Update '*var' according to the momentum scheme.
@@ -82,7 +82,7 @@ private Options() {
    * @return a new instance of ResourceApplyMomentum
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> ResourceApplyMomentum create(Scope scope, Operand<?> var, Operand<?> accum, Operand<T> lr, Operand<T> grad, Operand<T> momentum, Options... options) {
+  public static <T extends Tensor> ResourceApplyMomentum create(Scope scope, Operand<?> var, Operand<?> accum, Operand<T> lr, Operand<T> grad, Operand<T> momentum, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ResourceApplyMomentum", scope.makeOpName("ResourceApplyMomentum"));
     opBuilder.addInput(var.asOutput());
     opBuilder.addInput(accum.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyPowerSign.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyPowerSign.java
index 662c2253264..a7444fe0fa5 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyPowerSign.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyPowerSign.java
@@ -20,11 +20,11 @@
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Update '*var' according to the AddSign update.
@@ -72,7 +72,7 @@ private Options() {
    * @return a new instance of ResourceApplyPowerSign
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> ResourceApplyPowerSign create(Scope scope, Operand<?> var, Operand<?> m, Operand<T> lr, Operand<T> logbase, Operand<T> signDecay, Operand<T> beta, Operand<T> grad, Options... options) {
+  public static <T extends Tensor> ResourceApplyPowerSign create(Scope scope, Operand<?> var, Operand<?> m, Operand<T> lr, Operand<T> logbase, Operand<T> signDecay, Operand<T> beta, Operand<T> grad, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ResourceApplyPowerSign", scope.makeOpName("ResourceApplyPowerSign"));
     opBuilder.addInput(var.asOutput());
     opBuilder.addInput(m.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyProximalAdagrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyProximalAdagrad.java
index 8036d891e33..53e885b6324 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyProximalAdagrad.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyProximalAdagrad.java
@@ -20,11 +20,11 @@
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Update '*var' and '*accum' according to FOBOS with Adagrad learning rate.
@@ -70,7 +70,7 @@ private Options() {
    * @return a new instance of ResourceApplyProximalAdagrad
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> ResourceApplyProximalAdagrad create(Scope scope, Operand<?> var, Operand<?> accum, Operand<T> lr, Operand<T> l1, Operand<T> l2, Operand<T> grad, Options... options) {
+  public static <T extends Tensor> ResourceApplyProximalAdagrad create(Scope scope, Operand<?> var, Operand<?> accum, Operand<T> lr, Operand<T> l1, Operand<T> l2, Operand<T> grad, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ResourceApplyProximalAdagrad", scope.makeOpName("ResourceApplyProximalAdagrad"));
     opBuilder.addInput(var.asOutput());
     opBuilder.addInput(accum.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyProximalGradientDescent.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyProximalGradientDescent.java
index 3b217c88c67..9df791f64ce 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyProximalGradientDescent.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyProximalGradientDescent.java
@@ -20,11 +20,11 @@
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Update '*var' as FOBOS algorithm with fixed learning rate.
@@ -68,7 +68,7 @@ private Options() {
    * @return a new instance of ResourceApplyProximalGradientDescent
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> ResourceApplyProximalGradientDescent create(Scope scope, Operand<?> var, Operand<T> alpha, Operand<T> l1, Operand<T> l2, Operand<T> delta, Options... options) {
+  public static <T extends Tensor> ResourceApplyProximalGradientDescent create(Scope scope, Operand<?> var, Operand<T> alpha, Operand<T> l1, Operand<T> l2, Operand<T> delta, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ResourceApplyProximalGradientDescent", scope.makeOpName("ResourceApplyProximalGradientDescent"));
     opBuilder.addInput(var.asOutput());
     opBuilder.addInput(alpha.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyRmsProp.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyRmsProp.java
index ae42295c1f7..302afd570d1 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyRmsProp.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyRmsProp.java
@@ -20,11 +20,11 @@
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Update '*var' according to the RMSProp algorithm.
@@ -80,7 +80,7 @@ private Options() {
    * @return a new instance of ResourceApplyRmsProp
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> ResourceApplyRmsProp create(Scope scope, Operand<?> var, Operand<?> ms, Operand<?> mom, Operand<T> lr, Operand<T> rho, Operand<T> momentum, Operand<T> epsilon, Operand<T> grad, Options... options) {
+  public static <T extends Tensor> ResourceApplyRmsProp create(Scope scope, Operand<?> var, Operand<?> ms, Operand<?> mom, Operand<T> lr, Operand<T> rho, Operand<T> momentum, Operand<T> epsilon, Operand<T> grad, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ResourceApplyRMSProp", scope.makeOpName("ResourceApplyRmsProp"));
     opBuilder.addInput(var.asOutput());
     opBuilder.addInput(ms.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceConditionalAccumulator.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceConditionalAccumulator.java
index 2935d3ca6bc..d03834634ed 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceConditionalAccumulator.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceConditionalAccumulator.java
@@ -22,12 +22,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * A conditional accumulator for aggregating gradients.
@@ -41,7 +41,7 @@
  * This is a resource version of ConditionalAccumulator that will work in TF2.0
  * with tf.cond version 2.
  */
-public final class ResourceConditionalAccumulator extends RawOp implements Operand<TType> {
+public final class ResourceConditionalAccumulator extends RawOp implements Operand<Tensor> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.train.ResourceConditionalAccumulator}
@@ -92,7 +92,7 @@ private Options() {
    * @return a new instance of ResourceConditionalAccumulator
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> ResourceConditionalAccumulator create(Scope scope, DataType<T> dtype, Shape shape, Options... options) {
+  public static <T extends Tensor> ResourceConditionalAccumulator create(Scope scope, DataType<T> dtype, Shape shape, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ResourceConditionalAccumulator", scope.makeOpName("ResourceConditionalAccumulator"));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("dtype", dtype);
@@ -145,8 +145,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) handle;
+  public Output<Tensor> asOutput() {
+    return (Output<Tensor>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyAdadelta.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyAdadelta.java
index baea98fc1f7..78056617597 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyAdadelta.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyAdadelta.java
@@ -20,12 +20,12 @@
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * var: Should be from a Variable().
@@ -69,7 +69,7 @@ private Options() {
    * @return a new instance of ResourceSparseApplyAdadelta
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> ResourceSparseApplyAdadelta create(Scope scope, Operand<?> var, Operand<?> accum, Operand<?> accumUpdate, Operand<T> lr, Operand<T> rho, Operand<T> epsilon, Operand<T> grad, Operand<U> indices, Options... options) {
+  public static <T extends Tensor, U extends Tensor & TNumber> ResourceSparseApplyAdadelta create(Scope scope, Operand<?> var, Operand<?> accum, Operand<?> accumUpdate, Operand<T> lr, Operand<T> rho, Operand<T> epsilon, Operand<T> grad, Operand<U> indices, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ResourceSparseApplyAdadelta", scope.makeOpName("ResourceSparseApplyAdadelta"));
     opBuilder.addInput(var.asOutput());
     opBuilder.addInput(accum.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyAdagrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyAdagrad.java
index f7816e78d0c..b04a01c5411 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyAdagrad.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyAdagrad.java
@@ -20,12 +20,12 @@
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Update relevant entries in '*var' and '*accum' according to the adagrad scheme.
@@ -80,7 +80,7 @@ private Options() {
    * @return a new instance of ResourceSparseApplyAdagrad
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> ResourceSparseApplyAdagrad create(Scope scope, Operand<?> var, Operand<?> accum, Operand<T> lr, Operand<T> grad, Operand<U> indices, Options... options) {
+  public static <T extends Tensor, U extends Tensor & TNumber> ResourceSparseApplyAdagrad create(Scope scope, Operand<?> var, Operand<?> accum, Operand<T> lr, Operand<T> grad, Operand<U> indices, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ResourceSparseApplyAdagrad", scope.makeOpName("ResourceSparseApplyAdagrad"));
     opBuilder.addInput(var.asOutput());
     opBuilder.addInput(accum.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyAdagradDa.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyAdagradDa.java
index 417eca86a80..64450eadd93 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyAdagradDa.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyAdagradDa.java
@@ -20,13 +20,13 @@
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Update entries in '*var' and '*accum' according to the proximal adagrad scheme.
@@ -71,7 +71,7 @@ private Options() {
    * @return a new instance of ResourceSparseApplyAdagradDa
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> ResourceSparseApplyAdagradDa create(Scope scope, Operand<?> var, Operand<?> gradientAccumulator, Operand<?> gradientSquaredAccumulator, Operand<T> grad, Operand<U> indices, Operand<T> lr, Operand<T> l1, Operand<T> l2, Operand<TInt64> globalStep, Options... options) {
+  public static <T extends Tensor, U extends Tensor & TNumber> ResourceSparseApplyAdagradDa create(Scope scope, Operand<?> var, Operand<?> gradientAccumulator, Operand<?> gradientSquaredAccumulator, Operand<T> grad, Operand<U> indices, Operand<T> lr, Operand<T> l1, Operand<T> l2, Operand<TInt64> globalStep, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ResourceSparseApplyAdagradDA", scope.makeOpName("ResourceSparseApplyAdagradDa"));
     opBuilder.addInput(var.asOutput());
     opBuilder.addInput(gradientAccumulator.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyAdagradV2.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyAdagradV2.java
index f60d192c368..8ea0ceb8390 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyAdagradV2.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyAdagradV2.java
@@ -20,12 +20,12 @@
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Update relevant entries in '*var' and '*accum' according to the adagrad scheme.
@@ -80,7 +80,7 @@ private Options() {
    * @return a new instance of ResourceSparseApplyAdagradV2
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> ResourceSparseApplyAdagradV2 create(Scope scope, Operand<?> var, Operand<?> accum, Operand<T> lr, Operand<T> epsilon, Operand<T> grad, Operand<U> indices, Options... options) {
+  public static <T extends Tensor, U extends Tensor & TNumber> ResourceSparseApplyAdagradV2 create(Scope scope, Operand<?> var, Operand<?> accum, Operand<T> lr, Operand<T> epsilon, Operand<T> grad, Operand<U> indices, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ResourceSparseApplyAdagradV2", scope.makeOpName("ResourceSparseApplyAdagradV2"));
     opBuilder.addInput(var.asOutput());
     opBuilder.addInput(accum.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyCenteredRmsProp.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyCenteredRmsProp.java
index d6806c36abf..a71f6524b61 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyCenteredRmsProp.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyCenteredRmsProp.java
@@ -20,12 +20,12 @@
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Update '*var' according to the centered RMSProp algorithm.
@@ -89,7 +89,7 @@ private Options() {
    * @return a new instance of ResourceSparseApplyCenteredRmsProp
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> ResourceSparseApplyCenteredRmsProp create(Scope scope, Operand<?> var, Operand<?> mg, Operand<?> ms, Operand<?> mom, Operand<T> lr, Operand<T> rho, Operand<T> momentum, Operand<T> epsilon, Operand<T> grad, Operand<U> indices, Options... options) {
+  public static <T extends Tensor, U extends Tensor & TNumber> ResourceSparseApplyCenteredRmsProp create(Scope scope, Operand<?> var, Operand<?> mg, Operand<?> ms, Operand<?> mom, Operand<T> lr, Operand<T> rho, Operand<T> momentum, Operand<T> epsilon, Operand<T> grad, Operand<U> indices, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ResourceSparseApplyCenteredRMSProp", scope.makeOpName("ResourceSparseApplyCenteredRmsProp"));
     opBuilder.addInput(var.asOutput());
     opBuilder.addInput(mg.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyFtrl.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyFtrl.java
index a13382272c8..c6b83eb9091 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyFtrl.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyFtrl.java
@@ -20,12 +20,12 @@
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Update relevant entries in '*var' according to the Ftrl-proximal scheme.
@@ -90,7 +90,7 @@ private Options() {
    * @return a new instance of ResourceSparseApplyFtrl
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> ResourceSparseApplyFtrl create(Scope scope, Operand<?> var, Operand<?> accum, Operand<?> linear, Operand<T> grad, Operand<U> indices, Operand<T> lr, Operand<T> l1, Operand<T> l2, Operand<T> l2Shrinkage, Operand<T> lrPower, Options... options) {
+  public static <T extends Tensor, U extends Tensor & TNumber> ResourceSparseApplyFtrl create(Scope scope, Operand<?> var, Operand<?> accum, Operand<?> linear, Operand<T> grad, Operand<U> indices, Operand<T> lr, Operand<T> l1, Operand<T> l2, Operand<T> l2Shrinkage, Operand<T> lrPower, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ResourceSparseApplyFtrlV2", scope.makeOpName("ResourceSparseApplyFtrl"));
     opBuilder.addInput(var.asOutput());
     opBuilder.addInput(accum.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyKerasMomentum.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyKerasMomentum.java
index b385403f989..648217b3d8b 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyKerasMomentum.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyKerasMomentum.java
@@ -20,12 +20,12 @@
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Update relevant entries in '*var' and '*accum' according to the momentum scheme.
@@ -86,7 +86,7 @@ private Options() {
    * @return a new instance of ResourceSparseApplyKerasMomentum
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> ResourceSparseApplyKerasMomentum create(Scope scope, Operand<?> var, Operand<?> accum, Operand<T> lr, Operand<T> grad, Operand<U> indices, Operand<T> momentum, Options... options) {
+  public static <T extends Tensor, U extends Tensor & TNumber> ResourceSparseApplyKerasMomentum create(Scope scope, Operand<?> var, Operand<?> accum, Operand<T> lr, Operand<T> grad, Operand<U> indices, Operand<T> momentum, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ResourceSparseApplyKerasMomentum", scope.makeOpName("ResourceSparseApplyKerasMomentum"));
     opBuilder.addInput(var.asOutput());
     opBuilder.addInput(accum.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyMomentum.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyMomentum.java
index bc303bfbbf0..0b217881abf 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyMomentum.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyMomentum.java
@@ -20,12 +20,12 @@
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Update relevant entries in '*var' and '*accum' according to the momentum scheme.
@@ -86,7 +86,7 @@ private Options() {
    * @return a new instance of ResourceSparseApplyMomentum
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> ResourceSparseApplyMomentum create(Scope scope, Operand<?> var, Operand<?> accum, Operand<T> lr, Operand<T> grad, Operand<U> indices, Operand<T> momentum, Options... options) {
+  public static <T extends Tensor, U extends Tensor & TNumber> ResourceSparseApplyMomentum create(Scope scope, Operand<?> var, Operand<?> accum, Operand<T> lr, Operand<T> grad, Operand<U> indices, Operand<T> momentum, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ResourceSparseApplyMomentum", scope.makeOpName("ResourceSparseApplyMomentum"));
     opBuilder.addInput(var.asOutput());
     opBuilder.addInput(accum.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyProximalAdagrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyProximalAdagrad.java
index 678601d6aea..efbc4f836c2 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyProximalAdagrad.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyProximalAdagrad.java
@@ -20,12 +20,12 @@
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Sparse update entries in '*var' and '*accum' according to FOBOS algorithm.
@@ -74,7 +74,7 @@ private Options() {
    * @return a new instance of ResourceSparseApplyProximalAdagrad
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> ResourceSparseApplyProximalAdagrad create(Scope scope, Operand<?> var, Operand<?> accum, Operand<T> lr, Operand<T> l1, Operand<T> l2, Operand<T> grad, Operand<U> indices, Options... options) {
+  public static <T extends Tensor, U extends Tensor & TNumber> ResourceSparseApplyProximalAdagrad create(Scope scope, Operand<?> var, Operand<?> accum, Operand<T> lr, Operand<T> l1, Operand<T> l2, Operand<T> grad, Operand<U> indices, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ResourceSparseApplyProximalAdagrad", scope.makeOpName("ResourceSparseApplyProximalAdagrad"));
     opBuilder.addInput(var.asOutput());
     opBuilder.addInput(accum.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyProximalGradientDescent.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyProximalGradientDescent.java
index 11ad213524c..77194576cfe 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyProximalGradientDescent.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyProximalGradientDescent.java
@@ -20,12 +20,12 @@
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Sparse update '*var' as FOBOS algorithm with fixed learning rate.
@@ -71,7 +71,7 @@ private Options() {
    * @return a new instance of ResourceSparseApplyProximalGradientDescent
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> ResourceSparseApplyProximalGradientDescent create(Scope scope, Operand<?> var, Operand<T> alpha, Operand<T> l1, Operand<T> l2, Operand<T> grad, Operand<U> indices, Options... options) {
+  public static <T extends Tensor, U extends Tensor & TNumber> ResourceSparseApplyProximalGradientDescent create(Scope scope, Operand<?> var, Operand<T> alpha, Operand<T> l1, Operand<T> l2, Operand<T> grad, Operand<U> indices, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ResourceSparseApplyProximalGradientDescent", scope.makeOpName("ResourceSparseApplyProximalGradientDescent"));
     opBuilder.addInput(var.asOutput());
     opBuilder.addInput(alpha.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyRmsProp.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyRmsProp.java
index 8c519504f89..76a1623708b 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyRmsProp.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyRmsProp.java
@@ -20,12 +20,12 @@
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Update '*var' according to the RMSProp algorithm.
@@ -82,7 +82,7 @@ private Options() {
    * @return a new instance of ResourceSparseApplyRmsProp
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> ResourceSparseApplyRmsProp create(Scope scope, Operand<?> var, Operand<?> ms, Operand<?> mom, Operand<T> lr, Operand<T> rho, Operand<T> momentum, Operand<T> epsilon, Operand<T> grad, Operand<U> indices, Options... options) {
+  public static <T extends Tensor, U extends Tensor & TNumber> ResourceSparseApplyRmsProp create(Scope scope, Operand<?> var, Operand<?> ms, Operand<?> mom, Operand<T> lr, Operand<T> rho, Operand<T> momentum, Operand<T> epsilon, Operand<T> grad, Operand<U> indices, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ResourceSparseApplyRMSProp", scope.makeOpName("ResourceSparseApplyRmsProp"));
     opBuilder.addInput(var.asOutput());
     opBuilder.addInput(ms.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/Restore.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/Restore.java
index 5a2466f501d..29e595675f8 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/Restore.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/Restore.java
@@ -25,12 +25,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TString;
-import org.tensorflow.types.family.TType;
 
 /**
  * Restores tensors from a V2 checkpoint.
@@ -50,7 +50,7 @@
  * Callers must ensure all the named tensors are indeed stored in the checkpoint.
  */
 @Operator(group = "train")
-public final class Restore extends RawOp implements Iterable<Operand<TType>> {
+public final class Restore extends RawOp implements Iterable<Operand<Tensor>> {
   
   /**
    * Factory method to create a class wrapping a new Restore operation.
@@ -89,7 +89,7 @@ public List<Output<?>> tensors() {
   
   @Override
   @SuppressWarnings({"rawtypes", "unchecked"})
-  public Iterator<Operand<TType>> iterator() {
+  public Iterator<Operand<Tensor>> iterator() {
     return (Iterator) tensors.iterator();
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/RestoreSlice.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/RestoreSlice.java
index 631a127a44f..7a9e8382ca1 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/RestoreSlice.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/RestoreSlice.java
@@ -22,12 +22,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TString;
-import org.tensorflow.types.family.TType;
 
 /**
  * Restores a tensor from checkpoint files.
@@ -42,7 +42,7 @@
  * @param <T> data type for {@code tensor()} output
  */
 @Operator(group = "train")
-public final class RestoreSlice<T extends TType> extends RawOp implements Operand<T> {
+public final class RestoreSlice<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.train.RestoreSlice}
@@ -79,7 +79,7 @@ private Options() {
    * @return a new instance of RestoreSlice
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> RestoreSlice<T> create(Scope scope, Operand<TString> filePattern, Operand<TString> tensorName, Operand<TString> shapeAndSlice, DataType<T> dt, Options... options) {
+  public static <T extends Tensor> RestoreSlice<T> create(Scope scope, Operand<TString> filePattern, Operand<TString> tensorName, Operand<TString> shapeAndSlice, DataType<T> dt, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("RestoreSlice", scope.makeOpName("RestoreSlice"));
     opBuilder.addInput(filePattern.asOutput());
     opBuilder.addInput(tensorName.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyAdadelta.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyAdadelta.java
index 190449c8ca7..4cf10cb7fcd 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyAdadelta.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyAdadelta.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * var: Should be from a Variable().
@@ -34,7 +34,7 @@
  * @param <T> data type for {@code out()} output
  */
 @Operator(group = "train")
-public final class SparseApplyAdadelta<T extends TType> extends RawOp implements Operand<T> {
+public final class SparseApplyAdadelta<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.train.SparseApplyAdadelta}
@@ -72,7 +72,7 @@ private Options() {
    * @return a new instance of SparseApplyAdadelta
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> SparseApplyAdadelta<T> create(Scope scope, Operand<T> var, Operand<T> accum, Operand<T> accumUpdate, Operand<T> lr, Operand<T> rho, Operand<T> epsilon, Operand<T> grad, Operand<U> indices, Options... options) {
+  public static <T extends Tensor, U extends Tensor & TNumber> SparseApplyAdadelta<T> create(Scope scope, Operand<T> var, Operand<T> accum, Operand<T> accumUpdate, Operand<T> lr, Operand<T> rho, Operand<T> epsilon, Operand<T> grad, Operand<U> indices, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("SparseApplyAdadelta", scope.makeOpName("SparseApplyAdadelta"));
     opBuilder.addInput(var.asOutput());
     opBuilder.addInput(accum.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyAdagrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyAdagrad.java
index 71776f4fd26..8745fd9f4ae 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyAdagrad.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyAdagrad.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Update relevant entries in '*var' and '*accum' according to the adagrad scheme.
@@ -37,7 +37,7 @@
  * 
  * @param <T> data type for {@code out()} output
  */
-public final class SparseApplyAdagrad<T extends TType> extends RawOp implements Operand<T> {
+public final class SparseApplyAdagrad<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.train.SparseApplyAdagrad}
@@ -83,7 +83,7 @@ private Options() {
    * @return a new instance of SparseApplyAdagrad
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> SparseApplyAdagrad<T> create(Scope scope, Operand<T> var, Operand<T> accum, Operand<T> lr, Operand<T> epsilon, Operand<T> grad, Operand<U> indices, Options... options) {
+  public static <T extends Tensor, U extends Tensor & TNumber> SparseApplyAdagrad<T> create(Scope scope, Operand<T> var, Operand<T> accum, Operand<T> lr, Operand<T> epsilon, Operand<T> grad, Operand<U> indices, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("SparseApplyAdagradV2", scope.makeOpName("SparseApplyAdagrad"));
     opBuilder.addInput(var.asOutput());
     opBuilder.addInput(accum.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyAdagradDa.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyAdagradDa.java
index 7f13dd080dd..e2952193b28 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyAdagradDa.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyAdagradDa.java
@@ -21,13 +21,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Update entries in '*var' and '*accum' according to the proximal adagrad scheme.
@@ -35,7 +35,7 @@
  * @param <T> data type for {@code out()} output
  */
 @Operator(group = "train")
-public final class SparseApplyAdagradDa<T extends TType> extends RawOp implements Operand<T> {
+public final class SparseApplyAdagradDa<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.train.SparseApplyAdagradDa}
@@ -74,7 +74,7 @@ private Options() {
    * @return a new instance of SparseApplyAdagradDa
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> SparseApplyAdagradDa<T> create(Scope scope, Operand<T> var, Operand<T> gradientAccumulator, Operand<T> gradientSquaredAccumulator, Operand<T> grad, Operand<U> indices, Operand<T> lr, Operand<T> l1, Operand<T> l2, Operand<TInt64> globalStep, Options... options) {
+  public static <T extends Tensor, U extends Tensor & TNumber> SparseApplyAdagradDa<T> create(Scope scope, Operand<T> var, Operand<T> gradientAccumulator, Operand<T> gradientSquaredAccumulator, Operand<T> grad, Operand<U> indices, Operand<T> lr, Operand<T> l1, Operand<T> l2, Operand<TInt64> globalStep, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("SparseApplyAdagradDA", scope.makeOpName("SparseApplyAdagradDa"));
     opBuilder.addInput(var.asOutput());
     opBuilder.addInput(gradientAccumulator.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyCenteredRmsProp.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyCenteredRmsProp.java
index b73275139d1..ca7cab533b3 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyCenteredRmsProp.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyCenteredRmsProp.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Update '*var' according to the centered RMSProp algorithm.
@@ -51,7 +51,7 @@
  * @param <T> data type for {@code out()} output
  */
 @Operator(group = "train")
-public final class SparseApplyCenteredRmsProp<T extends TType> extends RawOp implements Operand<T> {
+public final class SparseApplyCenteredRmsProp<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.train.SparseApplyCenteredRmsProp}
@@ -92,7 +92,7 @@ private Options() {
    * @return a new instance of SparseApplyCenteredRmsProp
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> SparseApplyCenteredRmsProp<T> create(Scope scope, Operand<T> var, Operand<T> mg, Operand<T> ms, Operand<T> mom, Operand<T> lr, Operand<T> rho, Operand<T> momentum, Operand<T> epsilon, Operand<T> grad, Operand<U> indices, Options... options) {
+  public static <T extends Tensor, U extends Tensor & TNumber> SparseApplyCenteredRmsProp<T> create(Scope scope, Operand<T> var, Operand<T> mg, Operand<T> ms, Operand<T> mom, Operand<T> lr, Operand<T> rho, Operand<T> momentum, Operand<T> epsilon, Operand<T> grad, Operand<U> indices, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("SparseApplyCenteredRMSProp", scope.makeOpName("SparseApplyCenteredRmsProp"));
     opBuilder.addInput(var.asOutput());
     opBuilder.addInput(mg.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyFtrl.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyFtrl.java
index 2eb27c33f45..44967501a85 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyFtrl.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyFtrl.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Update relevant entries in '*var' according to the Ftrl-proximal scheme.
@@ -43,7 +43,7 @@
  * @param <T> data type for {@code out()} output
  */
 @Operator(group = "train")
-public final class SparseApplyFtrl<T extends TType> extends RawOp implements Operand<T> {
+public final class SparseApplyFtrl<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.train.SparseApplyFtrl}
@@ -93,7 +93,7 @@ private Options() {
    * @return a new instance of SparseApplyFtrl
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> SparseApplyFtrl<T> create(Scope scope, Operand<T> var, Operand<T> accum, Operand<T> linear, Operand<T> grad, Operand<U> indices, Operand<T> lr, Operand<T> l1, Operand<T> l2, Operand<T> l2Shrinkage, Operand<T> lrPower, Options... options) {
+  public static <T extends Tensor, U extends Tensor & TNumber> SparseApplyFtrl<T> create(Scope scope, Operand<T> var, Operand<T> accum, Operand<T> linear, Operand<T> grad, Operand<U> indices, Operand<T> lr, Operand<T> l1, Operand<T> l2, Operand<T> l2Shrinkage, Operand<T> lrPower, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("SparseApplyFtrlV2", scope.makeOpName("SparseApplyFtrl"));
     opBuilder.addInput(var.asOutput());
     opBuilder.addInput(accum.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyMomentum.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyMomentum.java
index 5e8bf197092..65de971b0ee 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyMomentum.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyMomentum.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Update relevant entries in '*var' and '*accum' according to the momentum scheme.
@@ -41,7 +41,7 @@
  * @param <T> data type for {@code out()} output
  */
 @Operator(group = "train")
-public final class SparseApplyMomentum<T extends TType> extends RawOp implements Operand<T> {
+public final class SparseApplyMomentum<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.train.SparseApplyMomentum}
@@ -89,7 +89,7 @@ private Options() {
    * @return a new instance of SparseApplyMomentum
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> SparseApplyMomentum<T> create(Scope scope, Operand<T> var, Operand<T> accum, Operand<T> lr, Operand<T> grad, Operand<U> indices, Operand<T> momentum, Options... options) {
+  public static <T extends Tensor, U extends Tensor & TNumber> SparseApplyMomentum<T> create(Scope scope, Operand<T> var, Operand<T> accum, Operand<T> lr, Operand<T> grad, Operand<U> indices, Operand<T> momentum, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("SparseApplyMomentum", scope.makeOpName("SparseApplyMomentum"));
     opBuilder.addInput(var.asOutput());
     opBuilder.addInput(accum.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyProximalAdagrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyProximalAdagrad.java
index 19ede0d5f71..b2f420a9922 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyProximalAdagrad.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyProximalAdagrad.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Sparse update entries in '*var' and '*accum' according to FOBOS algorithm.
@@ -40,7 +40,7 @@
  * @param <T> data type for {@code out()} output
  */
 @Operator(group = "train")
-public final class SparseApplyProximalAdagrad<T extends TType> extends RawOp implements Operand<T> {
+public final class SparseApplyProximalAdagrad<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.train.SparseApplyProximalAdagrad}
@@ -77,7 +77,7 @@ private Options() {
    * @return a new instance of SparseApplyProximalAdagrad
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> SparseApplyProximalAdagrad<T> create(Scope scope, Operand<T> var, Operand<T> accum, Operand<T> lr, Operand<T> l1, Operand<T> l2, Operand<T> grad, Operand<U> indices, Options... options) {
+  public static <T extends Tensor, U extends Tensor & TNumber> SparseApplyProximalAdagrad<T> create(Scope scope, Operand<T> var, Operand<T> accum, Operand<T> lr, Operand<T> l1, Operand<T> l2, Operand<T> grad, Operand<U> indices, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("SparseApplyProximalAdagrad", scope.makeOpName("SparseApplyProximalAdagrad"));
     opBuilder.addInput(var.asOutput());
     opBuilder.addInput(accum.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyProximalGradientDescent.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyProximalGradientDescent.java
index 14735403f02..8e4a87f1b6f 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyProximalGradientDescent.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyProximalGradientDescent.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Sparse update '*var' as FOBOS algorithm with fixed learning rate.
@@ -38,7 +38,7 @@
  * @param <T> data type for {@code out()} output
  */
 @Operator(group = "train")
-public final class SparseApplyProximalGradientDescent<T extends TType> extends RawOp implements Operand<T> {
+public final class SparseApplyProximalGradientDescent<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.train.SparseApplyProximalGradientDescent}
@@ -74,7 +74,7 @@ private Options() {
    * @return a new instance of SparseApplyProximalGradientDescent
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> SparseApplyProximalGradientDescent<T> create(Scope scope, Operand<T> var, Operand<T> alpha, Operand<T> l1, Operand<T> l2, Operand<T> grad, Operand<U> indices, Options... options) {
+  public static <T extends Tensor, U extends Tensor & TNumber> SparseApplyProximalGradientDescent<T> create(Scope scope, Operand<T> var, Operand<T> alpha, Operand<T> l1, Operand<T> l2, Operand<T> grad, Operand<U> indices, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("SparseApplyProximalGradientDescent", scope.makeOpName("SparseApplyProximalGradientDescent"));
     opBuilder.addInput(var.asOutput());
     opBuilder.addInput(alpha.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyRmsProp.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyRmsProp.java
index c0b39b65e5e..3528d012eb1 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyRmsProp.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyRmsProp.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Update '*var' according to the RMSProp algorithm.
@@ -45,7 +45,7 @@
  * @param <T> data type for {@code out()} output
  */
 @Operator(group = "train")
-public final class SparseApplyRmsProp<T extends TType> extends RawOp implements Operand<T> {
+public final class SparseApplyRmsProp<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.train.SparseApplyRmsProp}
@@ -85,7 +85,7 @@ private Options() {
    * @return a new instance of SparseApplyRmsProp
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> SparseApplyRmsProp<T> create(Scope scope, Operand<T> var, Operand<T> ms, Operand<T> mom, Operand<T> lr, Operand<T> rho, Operand<T> momentum, Operand<T> epsilon, Operand<T> grad, Operand<U> indices, Options... options) {
+  public static <T extends Tensor, U extends Tensor & TNumber> SparseApplyRmsProp<T> create(Scope scope, Operand<T> var, Operand<T> ms, Operand<T> mom, Operand<T> lr, Operand<T> rho, Operand<T> momentum, Operand<T> epsilon, Operand<T> grad, Operand<U> indices, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("SparseApplyRMSProp", scope.makeOpName("SparseApplyRmsProp"));
     opBuilder.addInput(var.asOutput());
     opBuilder.addInput(ms.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/TileGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/TileGrad.java
index 1bcfde7b183..46a7eac1e27 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/TileGrad.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/TileGrad.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
-import org.tensorflow.types.family.TType;
 
 /**
  * Returns the gradient of `Tile`.
@@ -38,7 +38,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "train")
-public final class TileGrad<T extends TType> extends RawOp implements Operand<T> {
+public final class TileGrad<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new TileGrad operation.
@@ -49,7 +49,7 @@ public final class TileGrad<T extends TType> extends RawOp implements Operand<T>
    * @return a new instance of TileGrad
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> TileGrad<T> create(Scope scope, Operand<T> input, Operand<TInt32> multiples) {
+  public static <T extends Tensor> TileGrad<T> create(Scope scope, Operand<T> input, Operand<TInt32> multiples) {
     OperationBuilder opBuilder = scope.env().opBuilder("TileGrad", scope.makeOpName("TileGrad"));
     opBuilder.addInput(input.asOutput());
     opBuilder.addInput(multiples.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/BroadcastHelper.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/BroadcastHelper.java
index e06919d04bd..770f551d5ae 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/BroadcastHelper.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/BroadcastHelper.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Helper operator for performing XLA-style broadcasts
@@ -38,7 +38,7 @@
  * @param <T> data type for {@code lhsOutput()} output
  */
 @Operator(group = "xla")
-public final class BroadcastHelper<T extends TType> extends RawOp {
+public final class BroadcastHelper<T extends Tensor> extends RawOp {
   
   /**
    * Factory method to create a class wrapping a new BroadcastHelper operation.
@@ -50,7 +50,7 @@ public final class BroadcastHelper<T extends TType> extends RawOp {
    * @return a new instance of BroadcastHelper
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> BroadcastHelper<T> create(Scope scope, Operand<T> lhs, Operand<T> rhs, Operand<U> broadcastDims) {
+  public static <T extends Tensor, U extends Tensor & TNumber> BroadcastHelper<T> create(Scope scope, Operand<T> lhs, Operand<T> rhs, Operand<U> broadcastDims) {
     OperationBuilder opBuilder = scope.env().opBuilder("XlaBroadcastHelper", scope.makeOpName("BroadcastHelper"));
     opBuilder.addInput(lhs.asOutput());
     opBuilder.addInput(rhs.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/ClusterOutput.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/ClusterOutput.java
index 717b65c931e..1a9b7fa18f4 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/ClusterOutput.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/ClusterOutput.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Operator that connects the output of an XLA computation to other consumer graph nodes.
@@ -33,7 +33,7 @@
  * @param <T> data type for {@code outputs()} output
  */
 @Operator(group = "xla")
-public final class ClusterOutput<T extends TType> extends RawOp implements Operand<T> {
+public final class ClusterOutput<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new ClusterOutput operation.
@@ -43,7 +43,7 @@ public final class ClusterOutput<T extends TType> extends RawOp implements Opera
    * @return a new instance of ClusterOutput
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> ClusterOutput<T> create(Scope scope, Operand<T> input) {
+  public static <T extends Tensor> ClusterOutput<T> create(Scope scope, Operand<T> input) {
     OperationBuilder opBuilder = scope.env().opBuilder("XlaClusterOutput", scope.makeOpName("ClusterOutput"));
     opBuilder.addInput(input.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Conv.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Conv.java
index 05cec5b6d51..e2d53e58b2f 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Conv.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Conv.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Wraps the XLA ConvGeneralDilated operator, documented at
@@ -37,7 +37,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "xla")
-public final class Conv<T extends TType> extends RawOp implements Operand<T> {
+public final class Conv<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Conv operation.
@@ -55,7 +55,7 @@ public final class Conv<T extends TType> extends RawOp implements Operand<T> {
    * @return a new instance of Conv
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> Conv<T> create(Scope scope, Operand<T> lhs, Operand<T> rhs, Operand<U> windowStrides, Operand<U> padding, Operand<U> lhsDilation, Operand<U> rhsDilation, Operand<U> featureGroupCount, String dimensionNumbers, String precisionConfig) {
+  public static <T extends Tensor, U extends Tensor & TNumber> Conv<T> create(Scope scope, Operand<T> lhs, Operand<T> rhs, Operand<U> windowStrides, Operand<U> padding, Operand<U> lhsDilation, Operand<U> rhsDilation, Operand<U> featureGroupCount, String dimensionNumbers, String precisionConfig) {
     OperationBuilder opBuilder = scope.env().opBuilder("XlaConv", scope.makeOpName("Conv"));
     opBuilder.addInput(lhs.asOutput());
     opBuilder.addInput(rhs.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Dot.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Dot.java
index 9cb83d061cb..e4cb77489a6 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Dot.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Dot.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Wraps the XLA DotGeneral operator, documented at
@@ -36,7 +36,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "xla")
-public final class Dot<T extends TType> extends RawOp implements Operand<T> {
+public final class Dot<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Dot operation.
@@ -49,7 +49,7 @@ public final class Dot<T extends TType> extends RawOp implements Operand<T> {
    * @return a new instance of Dot
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> Dot<T> create(Scope scope, Operand<T> lhs, Operand<T> rhs, String dimensionNumbers, String precisionConfig) {
+  public static <T extends Tensor> Dot<T> create(Scope scope, Operand<T> lhs, Operand<T> rhs, String dimensionNumbers, String precisionConfig) {
     OperationBuilder opBuilder = scope.env().opBuilder("XlaDot", scope.makeOpName("Dot"));
     opBuilder.addInput(lhs.asOutput());
     opBuilder.addInput(rhs.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/DynamicSlice.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/DynamicSlice.java
index b9e2bebf9b8..e3ea1be1b5e 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/DynamicSlice.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/DynamicSlice.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Wraps the XLA DynamicSlice operator, documented at
@@ -43,7 +43,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "xla")
-public final class DynamicSlice<T extends TType> extends RawOp implements Operand<T> {
+public final class DynamicSlice<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new DynamicSlice operation.
@@ -58,7 +58,7 @@ public final class DynamicSlice<T extends TType> extends RawOp implements Operan
    * @return a new instance of DynamicSlice
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> DynamicSlice<T> create(Scope scope, Operand<T> input, Operand<U> startIndices, Operand<U> sizeIndices) {
+  public static <T extends Tensor, U extends Tensor & TNumber> DynamicSlice<T> create(Scope scope, Operand<T> input, Operand<U> startIndices, Operand<U> sizeIndices) {
     OperationBuilder opBuilder = scope.env().opBuilder("XlaDynamicSlice", scope.makeOpName("DynamicSlice"));
     opBuilder.addInput(input.asOutput());
     opBuilder.addInput(startIndices.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/DynamicUpdateSlice.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/DynamicUpdateSlice.java
index dccbdd10da8..7782c2d9587 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/DynamicUpdateSlice.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/DynamicUpdateSlice.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Wraps the XLA DynamicUpdateSlice operator, documented at
@@ -44,7 +44,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "xla")
-public final class DynamicUpdateSlice<T extends TType> extends RawOp implements Operand<T> {
+public final class DynamicUpdateSlice<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new DynamicUpdateSlice operation.
@@ -57,7 +57,7 @@ public final class DynamicUpdateSlice<T extends TType> extends RawOp implements
    * @return a new instance of DynamicUpdateSlice
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> DynamicUpdateSlice<T> create(Scope scope, Operand<T> input, Operand<T> update, Operand<U> indices) {
+  public static <T extends Tensor, U extends Tensor & TNumber> DynamicUpdateSlice<T> create(Scope scope, Operand<T> input, Operand<T> update, Operand<U> indices) {
     OperationBuilder opBuilder = scope.env().opBuilder("XlaDynamicUpdateSlice", scope.makeOpName("DynamicUpdateSlice"));
     opBuilder.addInput(input.asOutput());
     opBuilder.addInput(update.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Einsum.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Einsum.java
index 451fb27bc49..d83b2a7b487 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Einsum.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Einsum.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * An op which supports basic einsum op with 2 inputs and 1 output.
@@ -36,7 +36,7 @@
  * @param <T> data type for {@code product()} output
  */
 @Operator(group = "xla")
-public final class Einsum<T extends TType> extends RawOp implements Operand<T> {
+public final class Einsum<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Einsum operation.
@@ -48,7 +48,7 @@ public final class Einsum<T extends TType> extends RawOp implements Operand<T> {
    * @return a new instance of Einsum
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> Einsum<T> create(Scope scope, Operand<T> a, Operand<T> b, String equation) {
+  public static <T extends Tensor> Einsum<T> create(Scope scope, Operand<T> a, Operand<T> b, String equation) {
     OperationBuilder opBuilder = scope.env().opBuilder("XlaEinsum", scope.makeOpName("Einsum"));
     opBuilder.addInput(a.asOutput());
     opBuilder.addInput(b.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Gather.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Gather.java
index 9daa1808620..f1fd72ba2a5 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Gather.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Gather.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Wraps the XLA Gather operator documented at
@@ -36,7 +36,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "xla")
-public final class Gather<T extends TType> extends RawOp implements Operand<T> {
+public final class Gather<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Gather operation.
@@ -50,7 +50,7 @@ public final class Gather<T extends TType> extends RawOp implements Operand<T> {
    * @return a new instance of Gather
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> Gather<T> create(Scope scope, Operand<T> operand, Operand<U> startIndices, Operand<U> sliceSizes, String dimensionNumbers, Boolean indicesAreSorted) {
+  public static <T extends Tensor, U extends Tensor & TNumber> Gather<T> create(Scope scope, Operand<T> operand, Operand<U> startIndices, Operand<U> sliceSizes, String dimensionNumbers, Boolean indicesAreSorted) {
     OperationBuilder opBuilder = scope.env().opBuilder("XlaGather", scope.makeOpName("Gather"));
     opBuilder.addInput(operand.asOutput());
     opBuilder.addInput(startIndices.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/KeyValueSort.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/KeyValueSort.java
index d1cf154bd8a..5f60de49d10 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/KeyValueSort.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/KeyValueSort.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Wraps the XLA Sort operator, documented at
@@ -40,7 +40,7 @@
  * @param <U> data type for {@code sortedValues()} output
  */
 @Operator(group = "xla")
-public final class KeyValueSort<T extends TNumber, U extends TType> extends RawOp {
+public final class KeyValueSort<T extends Tensor & TNumber, U extends Tensor> extends RawOp {
   
   /**
    * Factory method to create a class wrapping a new KeyValueSort operation.
@@ -51,7 +51,7 @@ public final class KeyValueSort<T extends TNumber, U extends TType> extends RawO
    * @return a new instance of KeyValueSort
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber, U extends TType> KeyValueSort<T, U> create(Scope scope, Operand<T> keys, Operand<U> values) {
+  public static <T extends Tensor & TNumber, U extends Tensor> KeyValueSort<T, U> create(Scope scope, Operand<T> keys, Operand<U> values) {
     OperationBuilder opBuilder = scope.env().opBuilder("XlaKeyValueSort", scope.makeOpName("KeyValueSort"));
     opBuilder.addInput(keys.asOutput());
     opBuilder.addInput(values.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Pad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Pad.java
index c56631c48f2..b2b7b511256 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Pad.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Pad.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Wraps the XLA Pad operator, documented at
@@ -37,7 +37,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "xla")
-public final class Pad<T extends TType> extends RawOp implements Operand<T> {
+public final class Pad<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Pad operation.
@@ -51,7 +51,7 @@ public final class Pad<T extends TType> extends RawOp implements Operand<T> {
    * @return a new instance of Pad
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> Pad<T> create(Scope scope, Operand<T> input, Operand<T> paddingValue, Operand<U> paddingLow, Operand<U> paddingHigh, Operand<U> paddingInterior) {
+  public static <T extends Tensor, U extends Tensor & TNumber> Pad<T> create(Scope scope, Operand<T> input, Operand<T> paddingValue, Operand<U> paddingLow, Operand<U> paddingHigh, Operand<U> paddingInterior) {
     OperationBuilder opBuilder = scope.env().opBuilder("XlaPad", scope.makeOpName("Pad"));
     opBuilder.addInput(input.asOutput());
     opBuilder.addInput(paddingValue.asOutput());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Recv.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Recv.java
index fe26ee27587..b53d6f36c55 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Recv.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Recv.java
@@ -22,12 +22,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Receives the named tensor from another XLA computation. Wraps the XLA Recv
@@ -38,7 +38,7 @@
  * @param <T> data type for {@code tensor()} output
  */
 @Operator(group = "xla")
-public final class Recv<T extends TType> extends RawOp implements Operand<T> {
+public final class Recv<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Recv operation.
@@ -50,7 +50,7 @@ public final class Recv<T extends TType> extends RawOp implements Operand<T> {
    * @return a new instance of Recv
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> Recv<T> create(Scope scope, DataType<T> dtype, String tensorName, Shape shape) {
+  public static <T extends Tensor> Recv<T> create(Scope scope, DataType<T> dtype, String tensorName, Shape shape) {
     OperationBuilder opBuilder = scope.env().opBuilder("XlaRecv", scope.makeOpName("Recv"));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("dtype", dtype);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/SelfAdjointEig.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/SelfAdjointEig.java
index 9ce4468fdfe..b36fc427d4f 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/SelfAdjointEig.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/SelfAdjointEig.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes the eigen decomposition of a batch of self-adjoint matrices
@@ -39,7 +39,7 @@
  * @param <T> data type for {@code w()} output
  */
 @Operator(group = "xla")
-public final class SelfAdjointEig<T extends TType> extends RawOp {
+public final class SelfAdjointEig<T extends Tensor> extends RawOp {
   
   /**
    * Factory method to create a class wrapping a new SelfAdjointEig operation.
@@ -56,7 +56,7 @@ public final class SelfAdjointEig<T extends TType> extends RawOp {
    * @return a new instance of SelfAdjointEig
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> SelfAdjointEig<T> create(Scope scope, Operand<T> a, Boolean lower, Long maxIter, Float epsilon) {
+  public static <T extends Tensor> SelfAdjointEig<T> create(Scope scope, Operand<T> a, Boolean lower, Long maxIter, Float epsilon) {
     OperationBuilder opBuilder = scope.env().opBuilder("XlaSelfAdjointEig", scope.makeOpName("SelfAdjointEig"));
     opBuilder.addInput(a.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Send.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Send.java
index b18a86458ca..7e8287d8766 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Send.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Send.java
@@ -20,11 +20,11 @@
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Sends the named tensor to another XLA computation. Wraps the XLA Send operator
@@ -44,7 +44,7 @@ public final class Send extends RawOp {
    * @return a new instance of Send
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> Send create(Scope scope, Operand<T> tensor, String tensorName) {
+  public static <T extends Tensor> Send create(Scope scope, Operand<T> tensor, String tensorName) {
     OperationBuilder opBuilder = scope.env().opBuilder("XlaSend", scope.makeOpName("Send"));
     opBuilder.addInput(tensor.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Sharding.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Sharding.java
index dbc255aa607..cb483010a40 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Sharding.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Sharding.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * An op which shards the input based on the given sharding attribute.
@@ -33,7 +33,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "xla")
-public final class Sharding<T extends TType> extends RawOp implements Operand<T> {
+public final class Sharding<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Sharding operation.
@@ -43,7 +43,7 @@ public final class Sharding<T extends TType> extends RawOp implements Operand<T>
    * @return a new instance of Sharding
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> Sharding<T> create(Scope scope, Operand<T> input) {
+  public static <T extends Tensor> Sharding<T> create(Scope scope, Operand<T> input) {
     OperationBuilder opBuilder = scope.env().opBuilder("XlaSharding", scope.makeOpName("Sharding"));
     opBuilder.addInput(input.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Sort.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Sort.java
index d7f9aae6449..11b15492709 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Sort.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Sort.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Wraps the XLA Sort operator, documented at
@@ -38,7 +38,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "xla")
-public final class Sort<T extends TType> extends RawOp implements Operand<T> {
+public final class Sort<T extends Tensor> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Sort operation.
@@ -48,7 +48,7 @@ public final class Sort<T extends TType> extends RawOp implements Operand<T> {
    * @return a new instance of Sort
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> Sort<T> create(Scope scope, Operand<T> input) {
+  public static <T extends Tensor> Sort<T> create(Scope scope, Operand<T> input) {
     OperationBuilder opBuilder = scope.env().opBuilder("XlaSort", scope.makeOpName("Sort"));
     opBuilder.addInput(input.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Svd.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Svd.java
index cd0a67f33b4..b5dc57f6531 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Svd.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Svd.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Computes the eigen decomposition of a batch of self-adjoint matrices
@@ -38,7 +38,7 @@
  * @param <T> data type for {@code s()} output
  */
 @Operator(group = "xla")
-public final class Svd<T extends TType> extends RawOp {
+public final class Svd<T extends Tensor> extends RawOp {
   
   /**
    * Factory method to create a class wrapping a new Svd operation.
@@ -54,7 +54,7 @@ public final class Svd<T extends TType> extends RawOp {
    * @return a new instance of Svd
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> Svd<T> create(Scope scope, Operand<T> a, Long maxIter, Float epsilon, String precisionConfig) {
+  public static <T extends Tensor> Svd<T> create(Scope scope, Operand<T> a, Long maxIter, Float epsilon, String precisionConfig) {
     OperationBuilder opBuilder = scope.env().opBuilder("XlaSvd", scope.makeOpName("Svd"));
     opBuilder.addInput(a.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/AbstractOperation.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/AbstractOperation.java
index 96da6bc5ff4..71a9bc45bcf 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/AbstractOperation.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/AbstractOperation.java
@@ -17,7 +17,6 @@
 
 import org.bytedeco.javacpp.Pointer;
 import org.tensorflow.ndarray.Shape;
-import org.tensorflow.types.family.TType;
 
 /**
  * Base class for {@link Operation} implementations.
@@ -37,7 +36,7 @@ public Output<?>[] outputList(int idx, int length) {
   }
 
   @Override
-  public <T extends TType> Output<T> output(int idx) {
+  public <T extends Tensor> Output<T> output(int idx) {
     return new Output<>(this, idx);
   }
 
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/DataType.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/DataType.java
index fa0603c6961..974a04f3333 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/DataType.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/DataType.java
@@ -17,19 +17,9 @@
 
 import org.tensorflow.internal.c_api.TF_Tensor;
 import org.tensorflow.ndarray.Shape;
-import org.tensorflow.types.TBfloat16;
-import org.tensorflow.types.TBool;
-import org.tensorflow.types.TFloat16;
-import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.TFloat64;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.TUint8;
-import org.tensorflow.types.TString;
-import org.tensorflow.types.family.TType;
 
 /** Represents a type of elements in a {@link Tensor} */
-public final class DataType<T extends TType> {
+public final class DataType<T extends Tensor> {
 
   @FunctionalInterface
   public interface TensorInstantiator<T> {
@@ -51,97 +41,28 @@ public interface TensorInstantiator<T> {
    * @param value must match the corresponding TF_* value in the TensorFlow C API.
    * @param byteSize size of an element of this type, in bytes, -1 if unknown
    * @param tensorMapper method for instantiating tensor from a native reference
-   * @param <T> a tensor type
    */
-  public static <T extends TType> DataType<T> create(String name, int value, int byteSize, TensorInstantiator<T> instantiator) {
+  public static <T extends Tensor> DataType<T> create(String name, int value, int byteSize, TensorInstantiator<T> instantiator) {
     return new DataType<>(name, value, byteSize, instantiator);
   }
 
   /**
-   * Gets the DataType associated with the readable-name for the type
-   * <p>The name must match exactly the name used to create the desired DataType</p>
-   *
-   * @param name readable-name for the type
-   * @return the DataType
-   * @throws java.lang.IllegalArgumentException if the name is not a valid data type name
-   * @throws java.lang.NullPointerException if name is null
+   * Returns the size of an element of this type, in bytes, or -1 if element size is variable.
    */
-  public static DataType<? extends TType> of(String name) {
-    switch (name) {
-      case TBfloat16.NAME:
-        return TBfloat16.DTYPE;
-      case TFloat16.NAME:
-        return TFloat16.DTYPE;
-      case TFloat32.NAME:
-        return TFloat32.DTYPE;
-      case TFloat64.NAME:
-        return TFloat64.DTYPE;
-      case TUint8.NAME:
-        return TUint8.DTYPE;
-      case TInt32.NAME:
-        return TInt32.DTYPE;
-      case TInt64.NAME:
-        return TInt64.DTYPE;
-      case TBool.NAME:
-        return TBool.DTYPE;
-      case TString.NAME:
-        return TString.DTYPE;
-      default:
-        throw new IllegalArgumentException(String.format("%s is an unknown DataType", name));
-    }
-  }
-
-  /** Returns true if this data type represents a floating point type */
-  public boolean isFloating() {
-    switch (this.name()) {
-      case TBfloat16.NAME:
-      case TFloat16.NAME:
-      case TFloat32.NAME:
-      case TFloat64.NAME:
-        return true;
-      default:
-        return false;
-    }
-  }
-
-  /** Returns true if this data type represents an integer type */
-  public boolean isInteger() {
-    switch (this.name()) {
-      case TInt32.NAME:
-      case TInt64.NAME:
-      case TUint8.NAME:
-        return true;
-      default:
-        return false;
-    }
-  }
-
-  /** Returns true if this data type represents a numeric type */
-  public boolean isNumeric() {
-    return isFloating() || isInteger();
-  }
-
-  /** Returns true if this data type represents a boolean type */
-  public boolean isBoolean() {
-    return this.name().equals(TBool.NAME);
-  }
-
-  /** Returns true if this data type represents a string type */
-  public boolean isString() {
-    return this.name().equals(TString.NAME);
-  }
-
-  /** Returns the size of an element of this type, in bytes, or -1 if element size is variable. */
   public int byteSize() {
     return byteSize;
   }
 
-  /** Returns true if this datatype has elements of variable length */
+  /**
+   * Returns true if this datatype has elements of variable length
+   */
   public boolean isVariableLength() {
     return byteSize == -1;
   }
 
-  /** Returns a readable name for this type */
+  /**
+   * Returns a readable name for this type
+   */
   public String name() {
     return name;
   }
@@ -151,8 +72,10 @@ public String toString() {
     return name + " (" + nativeCode + ")";
   }
 
-  /** Returns the numeric code for this datatype, as recognized by the native library (C API) */
-  int nativeCode() {
+  /**
+   * Returns the numeric code for this datatype, as recognized by the native library (C API)
+   */
+  public int nativeCode() {
     return nativeCode;
   }
 
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Operand.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Operand.java
index 070fc55cb0f..78228d1a5ec 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Operand.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Operand.java
@@ -39,7 +39,7 @@
  * tf.concat(split, tf.constant(0));
  * }</pre>
  */
-public interface Operand<T extends TType> extends Op {
+public interface Operand<T extends Tensor> extends Op {
 
   /**
    * Returns the symbolic handle of the tensor.
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Operation.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Operation.java
index 1cc175da161..dd7903e07a6 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Operation.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Operation.java
@@ -15,8 +15,6 @@
 
 package org.tensorflow;
 
-import org.tensorflow.types.family.TType;
-
 /**
  * Performs computation on Tensors.
  *
@@ -70,7 +68,7 @@ public interface Operation {
    * @param <T> The expected element type of the tensors produced by this output.
    * @param idx The index of the output among the outputs produced by this operation.
    */
-  <T extends TType> Output<T> output(int idx);
+  <T extends Tensor> Output<T> output(int idx);
 
   /**
    * Returns the size of the given inputs list of Tensors for this operation.
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Output.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Output.java
index c6499377ded..944ed507ee1 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Output.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Output.java
@@ -18,7 +18,6 @@
 import java.util.Objects;
 import org.bytedeco.javacpp.Pointer;
 import org.tensorflow.ndarray.Shape;
-import org.tensorflow.types.family.TType;
 
 /**
  * A symbolic handle to a tensor produced by an {@link Operation}.
@@ -29,7 +28,7 @@
  * <p>By implementing the {@link Operand} interface, instances of this class also act as operands to
  * {@link org.tensorflow.op.Op Op} instances.
  */
-public final class Output<T extends TType> implements Operand<T> {
+public final class Output<T extends Tensor> implements Operand<T> {
 
   /** Returns the index into the outputs of the Operation. */
   public int index() {
@@ -56,7 +55,7 @@ public DataType<T> dataType() {
    *     {@code U}.
    */
   @SuppressWarnings("unchecked")
-  public <U extends TType> Output<U> expect(DataType<U> dt) {
+  public <U extends Tensor> Output<U> expect(DataType<U> dt) {
     if (!dt.equals(this.dataType())) {
       throw new IllegalArgumentException(
           "Cannot cast from output of " + this.dataType() + " to output of " + dt);
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Tensor.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Tensor.java
index 8f130296aba..64ba11d2e14 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Tensor.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Tensor.java
@@ -15,16 +15,12 @@
 
 package org.tensorflow;
 
-import static org.tensorflow.internal.c_api.global.tensorflow.TF_TensorByteSize;
-
 import java.util.function.Consumer;
-import org.bytedeco.javacpp.PointerScope;
-import org.tensorflow.internal.buffer.TensorBuffers;
+import org.tensorflow.internal.tensor.buffer.TensorBuffers;
 import org.tensorflow.internal.c_api.TF_Tensor;
 import org.tensorflow.ndarray.NdArray;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.ndarray.buffer.ByteDataBuffer;
-import org.tensorflow.types.family.TType;
 
 /**
  * A statically typed multi-dimensional array whose elements are of a type described by T.
@@ -41,7 +37,7 @@
  * }
  * }</pre>
  */
-public interface Tensor<T extends TType> extends AutoCloseable {
+public interface Tensor<T> extends NdArray<T>, AutoCloseable {
 
   /**
    * Allocates a tensor of a given datatype and shape.
@@ -65,7 +61,7 @@ public interface Tensor<T extends TType> extends AutoCloseable {
    * @return an allocated but uninitialized tensor
    * @throws IllegalStateException if tensor failed to be allocated
    */
-  static <T extends TType & Tensor> T of(DataType<T> dtype, Shape shape) {
+  static <T extends Tensor> T of(DataType<T> dtype, Shape shape) {
     return of(dtype, shape, shape.size() * dtype.byteSize());
   }
 
@@ -88,7 +84,7 @@ static <T extends TType & Tensor> T of(DataType<T> dtype, Shape shape) {
    *                                  store the tensor data
    * @throws IllegalStateException if tensor failed to be allocated
    */
-  static <T extends TType & Tensor> T of(DataType<T> dtype, Shape shape, long size) {
+  static <T extends Tensor> T of(DataType<T> dtype, Shape shape, long size) {
     return Tensors.allocate(dtype, shape, size);
   }
 
@@ -116,7 +112,7 @@ static <T extends TType & Tensor> T of(DataType<T> dtype, Shape shape, long size
    * @return an allocated and initialized tensor
    * @throws IllegalStateException if tensor failed to be allocated
    */
-  static <T extends TType & Tensor> T of(DataType<T> dtype, Shape shape, Consumer<T> dataInitializer) {
+  static <T extends Tensor> T of(DataType<T> dtype, Shape shape, Consumer<T> dataInitializer) {
     return of(dtype, shape, shape.size() * dtype.byteSize(), dataInitializer);
   }
 
@@ -140,7 +136,7 @@ static <T extends TType & Tensor> T of(DataType<T> dtype, Shape shape, Consumer<
    *                                  store the tensor data
    * @throws IllegalStateException if tensor failed to be allocated
    */
-  static <T extends TType & Tensor> T of(DataType<T> dtype, Shape shape, long size, Consumer<T> tensorInit) {
+  static <T extends Tensor> T of(DataType<T> dtype, Shape shape, long size, Consumer<T> tensorInit) {
     T tensor = of(dtype, shape, size);
     try {
       tensorInit.accept(tensor);
@@ -164,7 +160,7 @@ static <T extends TType & Tensor> T of(DataType<T> dtype, Shape shape, long size
    * @throws IllegalArgumentException if {@code rawData} is not large enough to contain the tensor data
    * @throws IllegalStateException if tensor failed to be allocated with the given parameters
    */
-  static <T extends TType & Tensor> T of(DataType<T> dtype, Shape shape, ByteDataBuffer rawData) {
+  static <T extends Tensor> T of(DataType<T> dtype, Shape shape, ByteDataBuffer rawData) {
     T tensor = of(dtype, shape, rawData.size());
     rawData.copyTo(TensorBuffers.toBytes(tensor.nativeHandle()), rawData.size());
     return tensor;
@@ -181,7 +177,7 @@ static <T extends TType & Tensor> T of(DataType<T> dtype, Shape shape, ByteDataB
    *     {@code U}.
    */
   @SuppressWarnings("unchecked")
-  default <U extends TType & Tensor> U expect(DataType<U> dt) {
+  default <U extends Tensor> U expect(DataType<U> dt) {
     if (!dt.equals(dataType())) {
       throw new IllegalArgumentException(
           "Cannot cast from tensor of " + dataType() + " to tensor of " + dt);
@@ -201,7 +197,7 @@ default <U extends TType & Tensor> U expect(DataType<U> dt) {
   void close();
 
   /** Returns the {@link DataType} of elements stored in the Tensor. */
-  DataType<T> dataType();
+  DataType<?> dataType();
 
   /** Returns the size, in bytes, of the tensor data. */
   long numBytes();
@@ -245,10 +241,12 @@ default <U extends TType & Tensor> U expect(DataType<U> dt) {
    * @return the tensor data mapped to an n-dimensional space
    * @throws IllegalStateException if the tensor has been closed
    * @see NdArray
+   * @deprecated since tensor implements {@code NdArray}
    */
-  default T data() {
+  @Deprecated
+  default Tensor<T> data() {
     nativeHandle(); // make sure native handle is still valid
-    return (T) this;
+    return this;
   }
 
   /**
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Tensors.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Tensors.java
index 4ac2e92f714..b04a0f297b9 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Tensors.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Tensors.java
@@ -11,7 +11,7 @@
 
 final class Tensors {
 
-  static <T extends TType> T allocate(DataType<T> dataType, Shape shape, long size) {
+  static <T extends Tensor> T allocate(DataType<T> dataType, Shape shape, long size) {
     // Minimum requirements for datatypes of variable length cannot be verified in a relevant way so
     // we only validate them for fixed length datatypes
     if (!dataType.isVariableLength() && shape.size() * dataType.byteSize() > size) {
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/BooleanTensorImpl.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/BooleanTensorImpl.java
new file mode 100644
index 00000000000..09318072a7f
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/BooleanTensorImpl.java
@@ -0,0 +1,96 @@
+package org.tensorflow.internal.tensor;
+
+import org.tensorflow.DataType;
+import org.tensorflow.internal.c_api.TF_Tensor;
+import org.tensorflow.ndarray.NdArray;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.ndarray.buffer.BooleanDataBuffer;
+import org.tensorflow.ndarray.buffer.ByteDataBuffer;
+import org.tensorflow.ndarray.buffer.DataBuffer;
+import org.tensorflow.ndarray.impl.dense.BooleanDenseNdArray;
+import org.tensorflow.tensor.BooleanTensor;
+
+public class BooleanTensorImpl extends BooleanDenseNdArray implements BooleanTensor {
+
+  @Override
+  public TF_Tensor nativeHandle() {
+    return rawTensor.nativeHandle();
+  }
+
+  @Override
+  public DataType<?> dataType() {
+    return rawTensor.dataType();
+  }
+
+  @Override
+  public Shape shape() {
+    return rawTensor.shape();
+  }
+
+  @Override
+  public long numBytes() {
+    return rawTensor.numBytes();
+  }
+
+  @Override
+  public ByteDataBuffer rawData() {
+    return rawTensor.rawData();
+  }
+
+  @Override
+  public void close() {
+    rawTensor.close();
+  }
+
+  @Override
+  public String toString() {
+    return rawTensor.toString();
+  }
+
+  @Override
+  public BooleanTensor setBoolean(boolean value, long... coordinates) {
+    return (BooleanTensor)super.setBoolean(value, coordinates);
+  }
+
+  @Override
+  public BooleanTensor setObject(Boolean value, long... coordinates) {
+    return (BooleanTensor)super.setObject(value, coordinates);
+  }
+
+  @Override
+  public BooleanTensor set(NdArray<Boolean> src, long... coordinates) {
+    return (BooleanTensor)super.set(src, coordinates);
+  }
+
+  @Override
+  public BooleanTensor copyTo(NdArray<Boolean> dst) {
+    return (BooleanTensor)super.copyTo(dst);
+  }
+
+  @Override
+  public BooleanTensor read(DataBuffer<Boolean> dst) {
+    return (BooleanTensor)super.read(dst);
+  }
+
+  @Override
+  public BooleanTensor read(BooleanDataBuffer dst) {
+    return (BooleanTensor)super.read(dst);
+  }
+
+  @Override
+  public BooleanTensor write(DataBuffer<Boolean> src) {
+    return (BooleanTensor)super.write(src);
+  }
+
+  @Override
+  public BooleanTensor write(BooleanDataBuffer src) {
+    return (BooleanTensor)super.write(src);
+  }
+
+  public BooleanTensorImpl(TF_Tensor nativeHandle, DataType<?> dataType, Shape shape, BooleanDataBuffer buffer) {
+    super(buffer, shape);
+    this.rawTensor = new RawTensor(nativeHandle, dataType, shape);
+  }
+
+  private final RawTensor rawTensor;
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/ByteTensorImpl.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/ByteTensorImpl.java
new file mode 100644
index 00000000000..7f4f4742ff6
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/ByteTensorImpl.java
@@ -0,0 +1,95 @@
+package org.tensorflow.internal.tensor;
+
+import org.tensorflow.DataType;
+import org.tensorflow.internal.c_api.TF_Tensor;
+import org.tensorflow.ndarray.NdArray;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.ndarray.buffer.ByteDataBuffer;
+import org.tensorflow.ndarray.buffer.DataBuffer;
+import org.tensorflow.ndarray.impl.dense.ByteDenseNdArray;
+import org.tensorflow.tensor.ByteTensor;
+
+public class ByteTensorImpl extends ByteDenseNdArray implements ByteTensor {
+
+  @Override
+  public TF_Tensor nativeHandle() {
+    return rawTensor.nativeHandle();
+  }
+
+  @Override
+  public DataType<?> dataType() {
+    return rawTensor.dataType();
+  }
+
+  @Override
+  public Shape shape() {
+    return rawTensor.shape();
+  }
+
+  @Override
+  public long numBytes() {
+    return rawTensor.numBytes();
+  }
+
+  @Override
+  public ByteDataBuffer rawData() {
+    return rawTensor.rawData();
+  }
+
+  @Override
+  public void close() {
+    rawTensor.close();
+  }
+
+  @Override
+  public String toString() {
+    return rawTensor.toString();
+  }
+
+  @Override
+  public ByteTensor setByte(byte value, long... coordinates) {
+    return (ByteTensor)super.setByte(value, coordinates);
+  }
+
+  @Override
+  public ByteTensor setObject(Byte value, long... coordinates) {
+    return (ByteTensor)super.setObject(value, coordinates);
+  }
+
+  @Override
+  public ByteTensor set(NdArray<Byte> src, long... coordinates) {
+    return (ByteTensor)super.set(src, coordinates);
+  }
+
+  @Override
+  public ByteTensor copyTo(NdArray<Byte> dst) {
+    return (ByteTensor)super.copyTo(dst);
+  }
+
+  @Override
+  public ByteTensor read(DataBuffer<Byte> dst) {
+    return (ByteTensor)super.read(dst);
+  }
+
+  @Override
+  public ByteTensor read(ByteDataBuffer dst) {
+    return (ByteTensor)super.read(dst);
+  }
+
+  @Override
+  public ByteTensor write(DataBuffer<Byte> src) {
+    return (ByteTensor)super.write(src);
+  }
+
+  @Override
+  public ByteTensor write(ByteDataBuffer src) {
+    return (ByteTensor)super.write(src);
+  }
+
+  public ByteTensorImpl(TF_Tensor nativeHandle, DataType<?> dataType, Shape shape, ByteDataBuffer buffer) {
+    super(buffer, shape);
+    this.rawTensor = new RawTensor(nativeHandle, dataType, shape);
+  }
+
+  private final RawTensor rawTensor;
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/DoubleTensorImpl.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/DoubleTensorImpl.java
new file mode 100644
index 00000000000..f4da880c976
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/DoubleTensorImpl.java
@@ -0,0 +1,96 @@
+package org.tensorflow.internal.tensor;
+
+import org.tensorflow.DataType;
+import org.tensorflow.internal.c_api.TF_Tensor;
+import org.tensorflow.ndarray.NdArray;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.ndarray.buffer.ByteDataBuffer;
+import org.tensorflow.ndarray.buffer.DataBuffer;
+import org.tensorflow.ndarray.buffer.DoubleDataBuffer;
+import org.tensorflow.ndarray.impl.dense.DoubleDenseNdArray;
+import org.tensorflow.tensor.DoubleTensor;
+
+public class DoubleTensorImpl extends DoubleDenseNdArray implements DoubleTensor {
+
+  @Override
+  public TF_Tensor nativeHandle() {
+    return rawTensor.nativeHandle();
+  }
+
+  @Override
+  public DataType<?> dataType() {
+    return rawTensor.dataType();
+  }
+
+  @Override
+  public Shape shape() {
+    return rawTensor.shape();
+  }
+
+  @Override
+  public long numBytes() {
+    return rawTensor.numBytes();
+  }
+
+  @Override
+  public ByteDataBuffer rawData() {
+    return rawTensor.rawData();
+  }
+
+  @Override
+  public void close() {
+    rawTensor.close();
+  }
+
+  @Override
+  public String toString() {
+    return rawTensor.toString();
+  }
+
+  @Override
+  public DoubleTensor setDouble(double value, long... coordinates) {
+    return (DoubleTensor)super.setDouble(value, coordinates);
+  }
+
+  @Override
+  public DoubleTensor setObject(Double value, long... coordinates) {
+    return (DoubleTensor)super.setObject(value, coordinates);
+  }
+
+  @Override
+  public DoubleTensor set(NdArray<Double> src, long... coordinates) {
+    return (DoubleTensor)super.set(src, coordinates);
+  }
+
+  @Override
+  public DoubleTensor copyTo(NdArray<Double> dst) {
+    return (DoubleTensor)super.copyTo(dst);
+  }
+
+  @Override
+  public DoubleTensor read(DataBuffer<Double> dst) {
+    return (DoubleTensor)super.read(dst);
+  }
+
+  @Override
+  public DoubleTensor read(DoubleDataBuffer dst) {
+    return (DoubleTensor)super.read(dst);
+  }
+
+  @Override
+  public DoubleTensor write(DataBuffer<Double> src) {
+    return (DoubleTensor)super.write(src);
+  }
+
+  @Override
+  public DoubleTensor write(DoubleDataBuffer src) {
+    return (DoubleTensor)super.write(src);
+  }
+
+  public DoubleTensorImpl(TF_Tensor nativeHandle, DataType<?> dataType, Shape shape, DoubleDataBuffer buffer) {
+    super(buffer, shape);
+    this.rawTensor = new RawTensor(nativeHandle, dataType, shape);
+  }
+
+  private final RawTensor rawTensor;
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/FloatTensorImpl.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/FloatTensorImpl.java
new file mode 100644
index 00000000000..7aa4e824123
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/FloatTensorImpl.java
@@ -0,0 +1,96 @@
+package org.tensorflow.internal.tensor;
+
+import org.tensorflow.DataType;
+import org.tensorflow.internal.c_api.TF_Tensor;
+import org.tensorflow.ndarray.NdArray;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.ndarray.buffer.ByteDataBuffer;
+import org.tensorflow.ndarray.buffer.DataBuffer;
+import org.tensorflow.ndarray.buffer.FloatDataBuffer;
+import org.tensorflow.ndarray.impl.dense.FloatDenseNdArray;
+import org.tensorflow.tensor.FloatTensor;
+
+public class FloatTensorImpl extends FloatDenseNdArray implements FloatTensor {
+
+  @Override
+  public TF_Tensor nativeHandle() {
+    return rawTensor.nativeHandle();
+  }
+
+  @Override
+  public DataType<?> dataType() {
+    return rawTensor.dataType();
+  }
+
+  @Override
+  public Shape shape() {
+    return rawTensor.shape();
+  }
+
+  @Override
+  public long numBytes() {
+    return rawTensor.numBytes();
+  }
+
+  @Override
+  public ByteDataBuffer rawData() {
+    return rawTensor.rawData();
+  }
+
+  @Override
+  public void close() {
+    rawTensor.close();
+  }
+
+  @Override
+  public String toString() {
+    return rawTensor.toString();
+  }
+
+  @Override
+  public FloatTensor setFloat(float value, long... coordinates) {
+    return (FloatTensor)super.setFloat(value, coordinates);
+  }
+
+  @Override
+  public FloatTensor setObject(Float value, long... coordinates) {
+    return (FloatTensor)super.setObject(value, coordinates);
+  }
+
+  @Override
+  public FloatTensor set(NdArray<Float> src, long... coordinates) {
+    return (FloatTensor)super.set(src, coordinates);
+  }
+
+  @Override
+  public FloatTensor copyTo(NdArray<Float> dst) {
+    return (FloatTensor)super.copyTo(dst);
+  }
+
+  @Override
+  public FloatTensor read(DataBuffer<Float> dst) {
+    return (FloatTensor)super.read(dst);
+  }
+
+  @Override
+  public FloatTensor read(FloatDataBuffer dst) {
+    return (FloatTensor)super.read(dst);
+  }
+
+  @Override
+  public FloatTensor write(DataBuffer<Float> src) {
+    return (FloatTensor)super.write(src);
+  }
+
+  @Override
+  public FloatTensor write(FloatDataBuffer src) {
+    return (FloatTensor)super.write(src);
+  }
+
+  public FloatTensorImpl(TF_Tensor nativeHandle, DataType<?> dataType, Shape shape, FloatDataBuffer buffer) {
+    super(buffer, shape);
+    this.rawTensor = new RawTensor(nativeHandle, dataType, shape);
+  }
+
+  private final RawTensor rawTensor;
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/IntTensorImpl.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/IntTensorImpl.java
new file mode 100644
index 00000000000..6febdda8c93
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/IntTensorImpl.java
@@ -0,0 +1,96 @@
+package org.tensorflow.internal.tensor;
+
+import org.tensorflow.DataType;
+import org.tensorflow.internal.c_api.TF_Tensor;
+import org.tensorflow.ndarray.NdArray;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.ndarray.buffer.ByteDataBuffer;
+import org.tensorflow.ndarray.buffer.DataBuffer;
+import org.tensorflow.ndarray.buffer.IntDataBuffer;
+import org.tensorflow.ndarray.impl.dense.IntDenseNdArray;
+import org.tensorflow.tensor.IntTensor;
+
+public class IntTensorImpl extends IntDenseNdArray implements IntTensor {
+
+  @Override
+  public TF_Tensor nativeHandle() {
+    return rawTensor.nativeHandle();
+  }
+
+  @Override
+  public DataType<?> dataType() {
+    return rawTensor.dataType();
+  }
+
+  @Override
+  public Shape shape() {
+    return rawTensor.shape();
+  }
+
+  @Override
+  public long numBytes() {
+    return rawTensor.numBytes();
+  }
+
+  @Override
+  public ByteDataBuffer rawData() {
+    return rawTensor.rawData();
+  }
+
+  @Override
+  public void close() {
+    rawTensor.close();
+  }
+
+  @Override
+  public String toString() {
+    return rawTensor.toString();
+  }
+
+  @Override
+  public IntTensor setInt(int value, long... coordinates) {
+    return (IntTensor)super.setInt(value, coordinates);
+  }
+
+  @Override
+  public IntTensor setObject(Integer value, long... coordinates) {
+    return (IntTensor)super.setObject(value, coordinates);
+  }
+
+  @Override
+  public IntTensor set(NdArray<Integer> src, long... coordinates) {
+    return (IntTensor)super.set(src, coordinates);
+  }
+
+  @Override
+  public IntTensor copyTo(NdArray<Integer> dst) {
+    return (IntTensor)super.copyTo(dst);
+  }
+
+  @Override
+  public IntTensor read(DataBuffer<Integer> dst) {
+    return (IntTensor)super.read(dst);
+  }
+
+  @Override
+  public IntTensor read(IntDataBuffer dst) {
+    return (IntTensor)super.read(dst);
+  }
+
+  @Override
+  public IntTensor write(DataBuffer<Integer> src) {
+    return (IntTensor)super.write(src);
+  }
+
+  @Override
+  public IntTensor write(IntDataBuffer src) {
+    return (IntTensor)super.write(src);
+  }
+
+  public IntTensorImpl(TF_Tensor nativeHandle, DataType<?> dataType, Shape shape, IntDataBuffer buffer) {
+    super(buffer, shape);
+    this.rawTensor = new RawTensor(nativeHandle, dataType, shape);
+  }
+
+  private final RawTensor rawTensor;
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/LongTensorImpl.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/LongTensorImpl.java
new file mode 100644
index 00000000000..7dc68cf58ce
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/LongTensorImpl.java
@@ -0,0 +1,96 @@
+package org.tensorflow.internal.tensor;
+
+import org.tensorflow.DataType;
+import org.tensorflow.internal.c_api.TF_Tensor;
+import org.tensorflow.ndarray.NdArray;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.ndarray.buffer.ByteDataBuffer;
+import org.tensorflow.ndarray.buffer.DataBuffer;
+import org.tensorflow.ndarray.buffer.LongDataBuffer;
+import org.tensorflow.ndarray.impl.dense.LongDenseNdArray;
+import org.tensorflow.tensor.LongTensor;
+
+public class LongTensorImpl extends LongDenseNdArray implements LongTensor {
+
+  @Override
+  public TF_Tensor nativeHandle() {
+    return rawTensor.nativeHandle();
+  }
+
+  @Override
+  public DataType<?> dataType() {
+    return rawTensor.dataType();
+  }
+
+  @Override
+  public Shape shape() {
+    return rawTensor.shape();
+  }
+
+  @Override
+  public long numBytes() {
+    return rawTensor.numBytes();
+  }
+
+  @Override
+  public ByteDataBuffer rawData() {
+    return rawTensor.rawData();
+  }
+
+  @Override
+  public void close() {
+    rawTensor.close();
+  }
+
+  @Override
+  public String toString() {
+    return rawTensor.toString();
+  }
+
+  @Override
+  public LongTensor setLong(long value, long... coordinates) {
+    return (LongTensor)super.setLong(value, coordinates);
+  }
+
+  @Override
+  public LongTensor setObject(Long value, long... coordinates) {
+    return (LongTensor)super.setObject(value, coordinates);
+  }
+
+  @Override
+  public LongTensor set(NdArray<Long> src, long... coordinates) {
+    return (LongTensor)super.set(src, coordinates);
+  }
+
+  @Override
+  public LongTensor copyTo(NdArray<Long> dst) {
+    return (LongTensor)super.copyTo(dst);
+  }
+
+  @Override
+  public LongTensor read(DataBuffer<Long> dst) {
+    return (LongTensor)super.read(dst);
+  }
+
+  @Override
+  public LongTensor read(LongDataBuffer dst) {
+    return (LongTensor)super.read(dst);
+  }
+
+  @Override
+  public LongTensor write(DataBuffer<Long> src) {
+    return (LongTensor)super.write(src);
+  }
+
+  @Override
+  public LongTensor write(LongDataBuffer src) {
+    return (LongTensor)super.write(src);
+  }
+
+  public LongTensorImpl(TF_Tensor nativeHandle, DataType<?> dataType, Shape shape, LongDataBuffer buffer) {
+    super(buffer, shape);
+    this.rawTensor = new RawTensor(nativeHandle, dataType, shape);
+  }
+
+  private final RawTensor rawTensor;
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/RawTensor.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/RawTensor.java
similarity index 75%
rename from tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/RawTensor.java
rename to tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/RawTensor.java
index 872546c8ae0..98b9720b727 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/RawTensor.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/RawTensor.java
@@ -1,19 +1,16 @@
-package org.tensorflow.types;
+package org.tensorflow.internal.tensor;
 
 import static org.tensorflow.internal.c_api.global.tensorflow.TF_TensorByteSize;
 
 import org.bytedeco.javacpp.PointerScope;
 import org.tensorflow.DataType;
-import org.tensorflow.Tensor;
-import org.tensorflow.internal.buffer.TensorBuffers;
+import org.tensorflow.internal.tensor.buffer.TensorBuffers;
 import org.tensorflow.internal.c_api.TF_Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.ndarray.buffer.ByteDataBuffer;
-import org.tensorflow.types.family.TType;
 
-class RawTensor<T extends TType> implements Tensor<T> {
+class RawTensor<T> {
 
-  @Override
   public TF_Tensor nativeHandle() {
     if (nativeHandle.isNull()) {
       throw new IllegalStateException("Tensor has been already released");
@@ -21,27 +18,22 @@ public TF_Tensor nativeHandle() {
     return nativeHandle;
   }
 
-  @Override
-  public DataType<T> dataType() {
+  public DataType<?> dataType() {
     return dtype;
   }
 
-  @Override
   public Shape shape() {
     return shape;
   }
 
-  @Override
   public long numBytes() {
     return TF_TensorByteSize(nativeHandle);
   }
 
-  @Override
   public ByteDataBuffer rawData() {
     return TensorBuffers.toBytes(nativeHandle, true);
   }
 
-  @Override
   public void close() {
     tensorScope.close();
   }
@@ -54,10 +46,10 @@ public String toString() {
 
   private final PointerScope tensorScope = new PointerScope();
   private final TF_Tensor nativeHandle;
-  private final DataType<T> dtype;
+  private final DataType<?> dtype;
   private final Shape shape;
 
-  RawTensor(TF_Tensor nativeHandle, DataType<T> dtype, Shape shape) {
+  RawTensor(TF_Tensor nativeHandle, DataType<?> dtype, Shape shape) {
     this.nativeHandle = nativeHandle;
     this.dtype = dtype;
     this.shape = shape;
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/StringTensor.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/StringTensorImpl.java
similarity index 55%
rename from tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/StringTensor.java
rename to tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/StringTensorImpl.java
index ca03e589360..3ee3966a22f 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/StringTensor.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/StringTensorImpl.java
@@ -1,8 +1,8 @@
-package org.tensorflow.types;
+package org.tensorflow.internal.tensor;
 
 import org.tensorflow.DataType;
 import org.tensorflow.Tensor;
-import org.tensorflow.internal.buffer.ByteSequenceTensorBuffer;
+import org.tensorflow.internal.tensor.buffer.ByteSequenceTensorBuffer;
 import org.tensorflow.internal.c_api.TF_Tensor;
 import org.tensorflow.ndarray.NdArray;
 import org.tensorflow.ndarray.NdArrays;
@@ -11,33 +11,9 @@
 import org.tensorflow.ndarray.buffer.DataBuffer;
 import org.tensorflow.ndarray.buffer.layout.DataLayout;
 import org.tensorflow.ndarray.impl.dense.DenseNdArray;
-import org.tensorflow.ndarray.index.Index;
-import org.tensorflow.types.family.TType;
+import org.tensorflow.tensor.StringTensor;
 
-public interface StringTensor<T extends StringTensor & TType> extends NdArray<String>, Tensor<T> {
-
-  /**
-   * @return the tensor data as a n-dimensional array of raw byte sequences.
-   */
-  NdArray<byte[]> asBytes();
-
-  @Override
-  T set(NdArray<String> src, long... coordinates);
-
-  @Override
-  T setObject(String value, long... coordinates);
-
-  @Override
-  T copyTo(NdArray<String> dst);
-
-  @Override
-  T read(DataBuffer<String> dst);
-
-  @Override
-  T write(DataBuffer<String> src);
-}
-
-class StringTensorImpl<T extends StringTensor & TType> extends DenseNdArray<String> implements StringTensor<T> {
+public class StringTensorImpl extends DenseNdArray<String> implements StringTensor {
 
   @Override
   public NdArray<byte[]> asBytes() {
@@ -50,7 +26,7 @@ public TF_Tensor nativeHandle() {
   }
 
   @Override
-  public DataType<T> dataType() {
+  public DataType<Tensor<String>> dataType() {
     return rawTensor.dataType();
   }
 
@@ -80,37 +56,37 @@ public String toString() {
   }
 
   @Override
-  public T setObject(String value, long... coordinates) {
-    return (T)super.setObject(value, coordinates);
+  public Tensor<String> setObject(String value, long... coordinates) {
+    return (Tensor<String>)super.setObject(value, coordinates);
   }
 
   @Override
-  public T set(NdArray<String> src, long... coordinates) {
-    return (T)super.set(src, coordinates);
+  public Tensor<String> set(NdArray<String> src, long... coordinates) {
+    return (Tensor<String>)super.set(src, coordinates);
   }
 
   @Override
-  public T copyTo(NdArray<String> dst) {
-    return (T)super.copyTo(dst);
+  public Tensor<String> copyTo(NdArray<String> dst) {
+    return (Tensor<String>)super.copyTo(dst);
   }
 
   @Override
-  public T read(DataBuffer<String> dst) {
-    return (T)super.read(dst);
+  public Tensor<String> read(DataBuffer<String> dst) {
+    return (Tensor<String>)super.read(dst);
   }
 
   @Override
-  public T write(DataBuffer<String> src) {
-    return (T)super.write(src);
+  public Tensor<String> write(DataBuffer<String> src) {
+    return (Tensor<String>)super.write(src);
   }
 
   protected ByteSequenceTensorBuffer rawBuffer() {
     return rawBuffer;
   }
 
-  StringTensorImpl(
+  public StringTensorImpl(
       TF_Tensor nativeHandle,
-      DataType<T> dataType,
+      DataType<?> dataType,
       Shape shape,
       DataLayout<DataBuffer<byte[]>, String> layout,
       ByteSequenceTensorBuffer buffer
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/buffer/ByteSequenceTensorBuffer.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/buffer/ByteSequenceTensorBuffer.java
similarity index 99%
rename from tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/buffer/ByteSequenceTensorBuffer.java
rename to tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/buffer/ByteSequenceTensorBuffer.java
index bc80b2a164c..1ab7cb55bc8 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/buffer/ByteSequenceTensorBuffer.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/buffer/ByteSequenceTensorBuffer.java
@@ -15,7 +15,7 @@
  *  =======================================================================
  */
 
-package org.tensorflow.internal.buffer;
+package org.tensorflow.internal.tensor.buffer;
 
 import java.nio.ReadOnlyBufferException;
 import java.util.function.Function;
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/buffer/TensorBuffers.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/buffer/TensorBuffers.java
similarity index 99%
rename from tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/buffer/TensorBuffers.java
rename to tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/buffer/TensorBuffers.java
index 415c5ca35ef..f06de6631c1 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/buffer/TensorBuffers.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/buffer/TensorBuffers.java
@@ -15,7 +15,7 @@
  *  =======================================================================
  */
 
-package org.tensorflow.internal.buffer;
+package org.tensorflow.internal.tensor.buffer;
 
 import static org.tensorflow.internal.c_api.global.tensorflow.TF_TensorByteSize;
 import static org.tensorflow.internal.c_api.global.tensorflow.TF_TensorData;
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/buffer/TensorRawDataBufferFactory.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/buffer/TensorRawDataBufferFactory.java
similarity index 98%
rename from tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/buffer/TensorRawDataBufferFactory.java
rename to tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/buffer/TensorRawDataBufferFactory.java
index dbaf31f1dcc..3a35edf0485 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/buffer/TensorRawDataBufferFactory.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/buffer/TensorRawDataBufferFactory.java
@@ -15,7 +15,7 @@
  *  =======================================================================
  */
 
-package org.tensorflow.internal.buffer;
+package org.tensorflow.internal.tensor.buffer;
 
 import org.bytedeco.javacpp.Pointer;
 import org.tensorflow.ndarray.buffer.BooleanDataBuffer;
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/op/core/Constant.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/op/core/Constant.java
index c583bd8db0a..2fc69dd9188 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/op/core/Constant.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/op/core/Constant.java
@@ -21,18 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.Output;
 import org.tensorflow.Tensor;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.ndarray.buffer.BooleanDataBuffer;
-import org.tensorflow.ndarray.buffer.ByteDataBuffer;
-import org.tensorflow.ndarray.buffer.DataBuffer;
-import org.tensorflow.ndarray.buffer.DoubleDataBuffer;
-import org.tensorflow.ndarray.buffer.FloatDataBuffer;
-import org.tensorflow.ndarray.buffer.IntDataBuffer;
-import org.tensorflow.ndarray.buffer.LongDataBuffer;
 import org.tensorflow.ndarray.BooleanNdArray;
 import org.tensorflow.ndarray.ByteNdArray;
 import org.tensorflow.ndarray.DoubleNdArray;
@@ -41,7 +29,19 @@
 import org.tensorflow.ndarray.LongNdArray;
 import org.tensorflow.ndarray.NdArray;
 import org.tensorflow.ndarray.NdArrays;
+import org.tensorflow.ndarray.Shape;
 import org.tensorflow.ndarray.StdArrays;
+import org.tensorflow.ndarray.buffer.BooleanDataBuffer;
+import org.tensorflow.ndarray.buffer.ByteDataBuffer;
+import org.tensorflow.ndarray.buffer.DataBuffer;
+import org.tensorflow.ndarray.buffer.DoubleDataBuffer;
+import org.tensorflow.ndarray.buffer.FloatDataBuffer;
+import org.tensorflow.ndarray.buffer.IntDataBuffer;
+import org.tensorflow.ndarray.buffer.LongDataBuffer;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TBool;
 import org.tensorflow.types.TFloat32;
 import org.tensorflow.types.TFloat64;
@@ -66,7 +66,7 @@
  * }</pre>
  */
 @Operator
-public final class Constant<T extends TType> extends RawOp implements Operand<T> {
+public final class Constant<T extends Tensor> extends RawOp implements Operand<T> {
 
   /**
    * Creates a constant containing a single {@code int} element.
@@ -77,7 +77,7 @@ public final class Constant<T extends TType> extends RawOp implements Operand<T>
    */
   @Endpoint
   public static Constant<TInt32> scalarOf(Scope scope, int data) {
-    try (Tensor<TInt32> value = TInt32.scalarOf(data)) {
+    try (TInt32 value = TInt32.scalarOf(data)) {
       return create(scope, value);
     }
   }
@@ -92,7 +92,7 @@ public static Constant<TInt32> scalarOf(Scope scope, int data) {
    */
   @Endpoint
   public static Constant<TInt32> vectorOf(Scope scope, int[] data) {
-    try (Tensor<TInt32> value = TInt32.vectorOf(data)) {
+    try (TInt32 value = TInt32.vectorOf(data)) {
       return create(scope, value);
     }
   }
@@ -122,7 +122,7 @@ public static Constant<TInt32> arrayOf(Scope scope, int... data) {
    */
   @Endpoint
   public static Constant<TInt32> tensorOf(Scope scope, int[][] data) {
-    try (Tensor<TInt32> value = TInt32.tensorOf(StdArrays.shapeOf(data), t -> StdArrays.copyTo(data,
+    try (TInt32 value = TInt32.tensorOf(StdArrays.shapeOf(data), t -> StdArrays.copyTo(data,
         t))) {
       return create(scope, value);
     }
@@ -138,7 +138,7 @@ public static Constant<TInt32> tensorOf(Scope scope, int[][] data) {
    */
   @Endpoint
   public static Constant<TInt32> tensorOf(Scope scope, int[][][] data) {
-    try (Tensor<TInt32> value = TInt32.tensorOf(StdArrays.shapeOf(data), t -> StdArrays.copyTo(data,
+    try (TInt32 value = TInt32.tensorOf(StdArrays.shapeOf(data), t -> StdArrays.copyTo(data,
         t))) {
       return create(scope, value);
     }
@@ -154,7 +154,7 @@ public static Constant<TInt32> tensorOf(Scope scope, int[][][] data) {
    */
   @Endpoint
   public static Constant<TInt32> tensorOf(Scope scope, int[][][][] data) {
-    try (Tensor<TInt32> value = TInt32.tensorOf(StdArrays.shapeOf(data), t -> StdArrays.copyTo(data,
+    try (TInt32 value = TInt32.tensorOf(StdArrays.shapeOf(data), t -> StdArrays.copyTo(data,
         t))) {
       return create(scope, value);
     }
@@ -170,7 +170,7 @@ public static Constant<TInt32> tensorOf(Scope scope, int[][][][] data) {
    */
   @Endpoint
   public static Constant<TInt32> tensorOf(Scope scope, int[][][][][] data) {
-    try (Tensor<TInt32> value = TInt32.tensorOf(StdArrays.shapeOf(data), t -> StdArrays.copyTo(data,
+    try (TInt32 value = TInt32.tensorOf(StdArrays.shapeOf(data), t -> StdArrays.copyTo(data,
         t))) {
       return create(scope, value);
     }
@@ -186,7 +186,7 @@ public static Constant<TInt32> tensorOf(Scope scope, int[][][][][] data) {
    */
   @Endpoint
   public static Constant<TInt32> tensorOf(Scope scope, int[][][][][][] data) {
-    try (Tensor<TInt32> value = TInt32.tensorOf(StdArrays.shapeOf(data), t -> StdArrays.copyTo(data,
+    try (TInt32 value = TInt32.tensorOf(StdArrays.shapeOf(data), t -> StdArrays.copyTo(data,
         t))) {
       return create(scope, value);
     }
@@ -201,7 +201,7 @@ public static Constant<TInt32> tensorOf(Scope scope, int[][][][][][] data) {
    */
   @Endpoint
   public static Constant<TInt32> tensorOf(Scope scope, IntNdArray data) {
-    try (Tensor<TInt32> value = TInt32.tensorOf(data)) {
+    try (TInt32 value = TInt32.tensorOf(data)) {
       return create(scope, value);
     }
   }
@@ -217,7 +217,7 @@ public static Constant<TInt32> tensorOf(Scope scope, IntNdArray data) {
    */
   @Endpoint
   public static Constant<TInt32> tensorOf(Scope scope, Shape shape, IntDataBuffer data) {
-    try (Tensor<TInt32> value = TInt32.tensorOf(shape, data)) {
+    try (TInt32 value = TInt32.tensorOf(shape, data)) {
       return create(scope, value);
     }
   }
@@ -231,7 +231,7 @@ public static Constant<TInt32> tensorOf(Scope scope, Shape shape, IntDataBuffer
    */
   @Endpoint
   public static Constant<TFloat32> scalarOf(Scope scope, float data) {
-    try (Tensor<TFloat32> value = TFloat32.scalarOf(data)) {
+    try (TFloat32 value = TFloat32.scalarOf(data)) {
       return create(scope, value);
     }
   }
@@ -246,7 +246,7 @@ public static Constant<TFloat32> scalarOf(Scope scope, float data) {
    */
   @Endpoint
   public static Constant<TFloat32> vectorOf(Scope scope, float[] data) {
-    try (Tensor<TFloat32> value = TFloat32.vectorOf(data)) {
+    try (TFloat32 value = TFloat32.vectorOf(data)) {
       return create(scope, value);
     }
   }
@@ -276,7 +276,7 @@ public static Constant<TFloat32> arrayOf(Scope scope, float... data) {
    */
   @Endpoint
   public static Constant<TFloat32> tensorOf(Scope scope, float[][] data) {
-    try (Tensor<TFloat32> value = TFloat32.tensorOf(StdArrays.shapeOf(data), t -> StdArrays.copyTo(
+    try (TFloat32 value = TFloat32.tensorOf(StdArrays.shapeOf(data), t -> StdArrays.copyTo(
         data, t))) {
       return create(scope, value);
     }
@@ -292,7 +292,7 @@ public static Constant<TFloat32> tensorOf(Scope scope, float[][] data) {
    */
   @Endpoint
   public static Constant<TFloat32> tensorOf(Scope scope, float[][][] data) {
-    try (Tensor<TFloat32> value = TFloat32.tensorOf(StdArrays.shapeOf(data), t -> StdArrays.copyTo(
+    try (TFloat32 value = TFloat32.tensorOf(StdArrays.shapeOf(data), t -> StdArrays.copyTo(
         data, t))) {
       return create(scope, value);
     }
@@ -308,7 +308,7 @@ public static Constant<TFloat32> tensorOf(Scope scope, float[][][] data) {
    */
   @Endpoint
   public static Constant<TFloat32> tensorOf(Scope scope, float[][][][] data) {
-    try (Tensor<TFloat32> value = TFloat32.tensorOf(StdArrays.shapeOf(data), t -> StdArrays.copyTo(
+    try (TFloat32 value = TFloat32.tensorOf(StdArrays.shapeOf(data), t -> StdArrays.copyTo(
         data, t))) {
       return create(scope, value);
     }
@@ -324,7 +324,7 @@ public static Constant<TFloat32> tensorOf(Scope scope, float[][][][] data) {
    */
   @Endpoint
   public static Constant<TFloat32> tensorOf(Scope scope, float[][][][][] data) {
-    try (Tensor<TFloat32> value = TFloat32.tensorOf(StdArrays.shapeOf(data), t -> StdArrays.copyTo(
+    try (TFloat32 value = TFloat32.tensorOf(StdArrays.shapeOf(data), t -> StdArrays.copyTo(
         data, t))) {
       return create(scope, value);
     }
@@ -340,7 +340,7 @@ public static Constant<TFloat32> tensorOf(Scope scope, float[][][][][] data) {
    */
   @Endpoint
   public static Constant<TFloat32> tensorOf(Scope scope, float[][][][][][] data) {
-    try (Tensor<TFloat32> value = TFloat32.tensorOf(StdArrays.shapeOf(data), t -> StdArrays.copyTo(
+    try (TFloat32 value = TFloat32.tensorOf(StdArrays.shapeOf(data), t -> StdArrays.copyTo(
         data, t))) {
       return create(scope, value);
     }
@@ -355,7 +355,7 @@ public static Constant<TFloat32> tensorOf(Scope scope, float[][][][][][] data) {
    */
   @Endpoint
   public static Constant<TFloat32> tensorOf(Scope scope, FloatNdArray data) {
-    try (Tensor<TFloat32> value = TFloat32.tensorOf(data)) {
+    try (TFloat32 value = TFloat32.tensorOf(data)) {
       return create(scope, value);
     }
   }
@@ -371,7 +371,7 @@ public static Constant<TFloat32> tensorOf(Scope scope, FloatNdArray data) {
    */
   @Endpoint
   public static Constant<TFloat32> tensorOf(Scope scope, Shape shape, FloatDataBuffer data) {
-    try (Tensor<TFloat32> value = TFloat32.tensorOf(shape, data)) {
+    try (TFloat32 value = TFloat32.tensorOf(shape, data)) {
       return create(scope, value);
     }
   }
@@ -385,7 +385,7 @@ public static Constant<TFloat32> tensorOf(Scope scope, Shape shape, FloatDataBuf
    */
   @Endpoint
   public static Constant<TFloat64> scalarOf(Scope scope, double data) {
-    try (Tensor<TFloat64> value = TFloat64.scalarOf(data)) {
+    try (TFloat64 value = TFloat64.scalarOf(data)) {
       return create(scope, value);
     }
   }
@@ -400,7 +400,7 @@ public static Constant<TFloat64> scalarOf(Scope scope, double data) {
    */
   @Endpoint
   public static Constant<TFloat64> vectorOf(Scope scope, double[] data) {
-    try (Tensor<TFloat64> value = TFloat64.vectorOf(data)) {
+    try (TFloat64 value = TFloat64.vectorOf(data)) {
       return create(scope, value);
     }
   }
@@ -430,7 +430,7 @@ public static Constant<TFloat64> arrayOf(Scope scope, double... data) {
    */
   @Endpoint
   public static Constant<TFloat64> tensorOf(Scope scope, double[][] data) {
-    try (Tensor<TFloat64> value = TFloat64.tensorOf(StdArrays.shapeOf(data), t -> StdArrays.copyTo(
+    try (TFloat64 value = TFloat64.tensorOf(StdArrays.shapeOf(data), t -> StdArrays.copyTo(
         data, t))) {
       return create(scope, value);
     }
@@ -446,7 +446,7 @@ public static Constant<TFloat64> tensorOf(Scope scope, double[][] data) {
    */
   @Endpoint
   public static Constant<TFloat64> tensorOf(Scope scope, double[][][] data) {
-    try (Tensor<TFloat64> value = TFloat64.tensorOf(StdArrays.shapeOf(data), t -> StdArrays.copyTo(
+    try (TFloat64 value = TFloat64.tensorOf(StdArrays.shapeOf(data), t -> StdArrays.copyTo(
         data, t))) {
       return create(scope, value);
     }
@@ -462,7 +462,7 @@ public static Constant<TFloat64> tensorOf(Scope scope, double[][][] data) {
    */
   @Endpoint
   public static Constant<TFloat64> tensorOf(Scope scope, double[][][][] data) {
-    try (Tensor<TFloat64> value = TFloat64.tensorOf(StdArrays.shapeOf(data), t -> StdArrays.copyTo(
+    try (TFloat64 value = TFloat64.tensorOf(StdArrays.shapeOf(data), t -> StdArrays.copyTo(
         data, t))) {
       return create(scope, value);
     }
@@ -478,7 +478,7 @@ public static Constant<TFloat64> tensorOf(Scope scope, double[][][][] data) {
    */
   @Endpoint
   public static Constant<TFloat64> tensorOf(Scope scope, double[][][][][] data) {
-    try (Tensor<TFloat64> value = TFloat64.tensorOf(StdArrays.shapeOf(data), t -> StdArrays.copyTo(
+    try (TFloat64 value = TFloat64.tensorOf(StdArrays.shapeOf(data), t -> StdArrays.copyTo(
         data, t))) {
       return create(scope, value);
     }
@@ -494,7 +494,7 @@ public static Constant<TFloat64> tensorOf(Scope scope, double[][][][][] data) {
    */
   @Endpoint
   public static Constant<TFloat64> tensorOf(Scope scope, double[][][][][][] data) {
-    try (Tensor<TFloat64> value = TFloat64.tensorOf(StdArrays.shapeOf(data), t -> StdArrays.copyTo(
+    try (TFloat64 value = TFloat64.tensorOf(StdArrays.shapeOf(data), t -> StdArrays.copyTo(
         data, t))) {
       return create(scope, value);
     }
@@ -509,7 +509,7 @@ public static Constant<TFloat64> tensorOf(Scope scope, double[][][][][][] data)
    */
   @Endpoint
   public static Constant<TFloat64> tensorOf(Scope scope, DoubleNdArray data) {
-    try (Tensor<TFloat64> value = TFloat64.tensorOf(data)) {
+    try (TFloat64 value = TFloat64.tensorOf(data)) {
       return create(scope, value);
     }
   }
@@ -525,7 +525,7 @@ public static Constant<TFloat64> tensorOf(Scope scope, DoubleNdArray data) {
    */
   @Endpoint
   public static Constant<TFloat64> tensorOf(Scope scope, Shape shape, DoubleDataBuffer data) {
-    try (Tensor<TFloat64> value = TFloat64.tensorOf(shape, data)) {
+    try (TFloat64 value = TFloat64.tensorOf(shape, data)) {
       return create(scope, value);
     }
   }
@@ -539,7 +539,7 @@ public static Constant<TFloat64> tensorOf(Scope scope, Shape shape, DoubleDataBu
    */
   @Endpoint
   public static Constant<TInt64> scalarOf(Scope scope, long data) {
-    try (Tensor<TInt64> value = TInt64.scalarOf(data)) {
+    try (TInt64 value = TInt64.scalarOf(data)) {
       return create(scope, value);
     }
   }
@@ -554,7 +554,7 @@ public static Constant<TInt64> scalarOf(Scope scope, long data) {
    */
   @Endpoint
   public static Constant<TInt64> vectorOf(Scope scope, long[] data) {
-    try (Tensor<TInt64> value = TInt64.vectorOf(data)) {
+    try (TInt64 value = TInt64.vectorOf(data)) {
       return create(scope, value);
     }
   }
@@ -569,7 +569,7 @@ public static Constant<TInt64> vectorOf(Scope scope, long[] data) {
    */
   @Endpoint
   public static Constant<TInt64> tensorOf(Scope scope, long[][] data) {
-    try (Tensor<TInt64> value = TInt64.tensorOf(StdArrays.shapeOf(data), t -> StdArrays.copyTo(data,
+    try (TInt64 value = TInt64.tensorOf(StdArrays.shapeOf(data), t -> StdArrays.copyTo(data,
         t))) {
       return create(scope, value);
     }
@@ -600,7 +600,7 @@ public static Constant<TInt64> arrayOf(Scope scope, long... data) {
    */
   @Endpoint
   public static Constant<TInt64> tensorOf(Scope scope, long[][][] data) {
-    try (Tensor<TInt64> value = TInt64.tensorOf(StdArrays.shapeOf(data), t -> StdArrays.copyTo(data,
+    try (TInt64 value = TInt64.tensorOf(StdArrays.shapeOf(data), t -> StdArrays.copyTo(data,
         t))) {
       return create(scope, value);
     }
@@ -616,7 +616,7 @@ public static Constant<TInt64> tensorOf(Scope scope, long[][][] data) {
    */
   @Endpoint
   public static Constant<TInt64> tensorOf(Scope scope, long[][][][] data) {
-    try (Tensor<TInt64> value = TInt64.tensorOf(StdArrays.shapeOf(data), t -> StdArrays.copyTo(data,
+    try (TInt64 value = TInt64.tensorOf(StdArrays.shapeOf(data), t -> StdArrays.copyTo(data,
         t))) {
       return create(scope, value);
     }
@@ -632,7 +632,7 @@ public static Constant<TInt64> tensorOf(Scope scope, long[][][][] data) {
    */
   @Endpoint
   public static Constant<TInt64> tensorOf(Scope scope, long[][][][][] data) {
-    try (Tensor<TInt64> value = TInt64.tensorOf(StdArrays.shapeOf(data), t -> StdArrays.copyTo(data,
+    try (TInt64 value = TInt64.tensorOf(StdArrays.shapeOf(data), t -> StdArrays.copyTo(data,
         t))) {
       return create(scope, value);
     }
@@ -648,7 +648,7 @@ public static Constant<TInt64> tensorOf(Scope scope, long[][][][][] data) {
    */
   @Endpoint
   public static Constant<TInt64> tensorOf(Scope scope, long[][][][][][] data) {
-    try (Tensor<TInt64> value = TInt64.tensorOf(StdArrays.shapeOf(data), t -> StdArrays.copyTo(data,
+    try (TInt64 value = TInt64.tensorOf(StdArrays.shapeOf(data), t -> StdArrays.copyTo(data,
         t))) {
       return create(scope, value);
     }
@@ -663,7 +663,7 @@ public static Constant<TInt64> tensorOf(Scope scope, long[][][][][][] data) {
    */
   @Endpoint
   public static Constant<TInt64> tensorOf(Scope scope, LongNdArray data) {
-    try (Tensor<TInt64> value = TInt64.tensorOf(data)) {
+    try (TInt64 value = TInt64.tensorOf(data)) {
       return create(scope, value);
     }
   }
@@ -679,7 +679,7 @@ public static Constant<TInt64> tensorOf(Scope scope, LongNdArray data) {
    */
   @Endpoint
   public static Constant<TInt64> tensorOf(Scope scope, Shape shape, LongDataBuffer data) {
-    try (Tensor<TInt64> value = TInt64.tensorOf(shape, data)) {
+    try (TInt64 value = TInt64.tensorOf(shape, data)) {
       return create(scope, value);
     }
   }
@@ -693,7 +693,7 @@ public static Constant<TInt64> tensorOf(Scope scope, Shape shape, LongDataBuffer
    */
   @Endpoint
   public static Constant<TBool> scalarOf(Scope scope, boolean data) {
-    try (Tensor<TBool> value = TBool.scalarOf(data)) {
+    try (TBool value = TBool.scalarOf(data)) {
       return create(scope, value);
     }
   }
@@ -708,7 +708,7 @@ public static Constant<TBool> scalarOf(Scope scope, boolean data) {
    */
   @Endpoint
   public static Constant<TBool> vectorOf(Scope scope, boolean[] data) {
-    try (Tensor<TBool> value = TBool.vectorOf(data)) {
+    try (TBool value = TBool.vectorOf(data)) {
       return create(scope, value);
     }
   }
@@ -738,7 +738,7 @@ public static Constant<TBool> arrayOf(Scope scope, boolean... data) {
    */
   @Endpoint
   public static Constant<TBool> tensorOf(Scope scope, boolean[][] data) {
-    try (Tensor<TBool> value = TBool.tensorOf(StdArrays.shapeOf(data), t -> StdArrays.copyTo(data,
+    try (TBool value = TBool.tensorOf(StdArrays.shapeOf(data), t -> StdArrays.copyTo(data,
         t))) {
       return create(scope, value);
     }
@@ -754,7 +754,7 @@ public static Constant<TBool> tensorOf(Scope scope, boolean[][] data) {
    */
   @Endpoint
   public static Constant<TBool> tensorOf(Scope scope, boolean[][][] data) {
-    try (Tensor<TBool> value = TBool.tensorOf(StdArrays.shapeOf(data), t -> StdArrays.copyTo(data,
+    try (TBool value = TBool.tensorOf(StdArrays.shapeOf(data), t -> StdArrays.copyTo(data,
         t))) {
       return create(scope, value);
     }
@@ -770,7 +770,7 @@ public static Constant<TBool> tensorOf(Scope scope, boolean[][][] data) {
    */
   @Endpoint
   public static Constant<TBool> tensorOf(Scope scope, boolean[][][][] data) {
-    try (Tensor<TBool> value = TBool.tensorOf(StdArrays.shapeOf(data), t -> StdArrays.copyTo(data,
+    try (TBool value = TBool.tensorOf(StdArrays.shapeOf(data), t -> StdArrays.copyTo(data,
         t))) {
       return create(scope, value);
     }
@@ -786,7 +786,7 @@ public static Constant<TBool> tensorOf(Scope scope, boolean[][][][] data) {
    */
   @Endpoint
   public static Constant<TBool> tensorOf(Scope scope, boolean[][][][][] data) {
-    try (Tensor<TBool> value = TBool.tensorOf(StdArrays.shapeOf(data), t -> StdArrays.copyTo(data,
+    try (TBool value = TBool.tensorOf(StdArrays.shapeOf(data), t -> StdArrays.copyTo(data,
         t))) {
       return create(scope, value);
     }
@@ -802,7 +802,7 @@ public static Constant<TBool> tensorOf(Scope scope, boolean[][][][][] data) {
    */
   @Endpoint
   public static Constant<TBool> tensorOf(Scope scope, boolean[][][][][][] data) {
-    try (Tensor<TBool> value = TBool.tensorOf(StdArrays.shapeOf(data), t -> StdArrays.copyTo(data,
+    try (TBool value = TBool.tensorOf(StdArrays.shapeOf(data), t -> StdArrays.copyTo(data,
         t))) {
       return create(scope, value);
     }
@@ -817,7 +817,7 @@ public static Constant<TBool> tensorOf(Scope scope, boolean[][][][][][] data) {
    */
   @Endpoint
   public static Constant<TBool> tensorOf(Scope scope, BooleanNdArray data) {
-    try (Tensor<TBool> value = TBool.tensorOf(data)) {
+    try (TBool value = TBool.tensorOf(data)) {
       return create(scope, value);
     }
   }
@@ -833,7 +833,7 @@ public static Constant<TBool> tensorOf(Scope scope, BooleanNdArray data) {
    */
   @Endpoint
   public static Constant<TBool> tensorOf(Scope scope, Shape shape, BooleanDataBuffer data) {
-    try (Tensor<TBool> value = TBool.tensorOf(shape, data)) {
+    try (TBool value = TBool.tensorOf(shape, data)) {
       return create(scope, value);
     }
   }
@@ -847,7 +847,7 @@ public static Constant<TBool> tensorOf(Scope scope, Shape shape, BooleanDataBuff
    */
   @Endpoint
   public static Constant<TUint8> scalarOf(Scope scope, byte data) {
-    try (Tensor<TUint8> value = TUint8.scalarOf(data)) {
+    try (TUint8 value = TUint8.scalarOf(data)) {
       return create(scope, value);
     }
   }
@@ -862,7 +862,7 @@ public static Constant<TUint8> scalarOf(Scope scope, byte data) {
    */
   @Endpoint
   public static Constant<TUint8> vectorOf(Scope scope, byte[] data) {
-    try (Tensor<TUint8> value = TUint8.vectorOf(data)) {
+    try (TUint8 value = TUint8.vectorOf(data)) {
       return create(scope, value);
     }
   }
@@ -892,7 +892,7 @@ public static Constant<TUint8> arrayOf(Scope scope, byte... data) {
    */
   @Endpoint
   public static Constant<TUint8> tensorOf(Scope scope, byte[][] data) {
-    try (Tensor<TUint8> value = TUint8.tensorOf(StdArrays.shapeOf(data), d -> StdArrays.copyTo(data,
+    try (TUint8 value = TUint8.tensorOf(StdArrays.shapeOf(data), d -> StdArrays.copyTo(data,
         d))) {
       return create(scope, value);
     }
@@ -908,7 +908,7 @@ public static Constant<TUint8> tensorOf(Scope scope, byte[][] data) {
    */
   @Endpoint
   public static Constant<TUint8> tensorOf(Scope scope, byte[][][] data) {
-    try (Tensor<TUint8> value = TUint8.tensorOf(StdArrays.shapeOf(data), d -> StdArrays.copyTo(data,
+    try (TUint8 value = TUint8.tensorOf(StdArrays.shapeOf(data), d -> StdArrays.copyTo(data,
         d))) {
       return create(scope, value);
     }
@@ -924,7 +924,7 @@ public static Constant<TUint8> tensorOf(Scope scope, byte[][][] data) {
    */
   @Endpoint
   public static Constant<TUint8> tensorOf(Scope scope, byte[][][][] data) {
-    try (Tensor<TUint8> value = TUint8.tensorOf(StdArrays.shapeOf(data), d -> StdArrays.copyTo(data,
+    try (TUint8 value = TUint8.tensorOf(StdArrays.shapeOf(data), d -> StdArrays.copyTo(data,
         d))) {
       return create(scope, value);
     }
@@ -940,7 +940,7 @@ public static Constant<TUint8> tensorOf(Scope scope, byte[][][][] data) {
    */
   @Endpoint
   public static Constant<TUint8> tensorOf(Scope scope, byte[][][][][] data) {
-    try (Tensor<TUint8> value = TUint8.tensorOf(StdArrays.shapeOf(data), d -> StdArrays.copyTo(data,
+    try (TUint8 value = TUint8.tensorOf(StdArrays.shapeOf(data), d -> StdArrays.copyTo(data,
         d))) {
       return create(scope, value);
     }
@@ -956,7 +956,7 @@ public static Constant<TUint8> tensorOf(Scope scope, byte[][][][][] data) {
    */
   @Endpoint
   public static Constant<TUint8> tensorOf(Scope scope, byte[][][][][][] data) {
-    try (Tensor<TUint8> value = TUint8.tensorOf(StdArrays.shapeOf(data), d -> StdArrays.copyTo(data,
+    try (TUint8 value = TUint8.tensorOf(StdArrays.shapeOf(data), d -> StdArrays.copyTo(data,
         d))) {
       return create(scope, value);
     }
@@ -971,7 +971,7 @@ public static Constant<TUint8> tensorOf(Scope scope, byte[][][][][][] data) {
    */
   @Endpoint
   public static Constant<TUint8> tensorOf(Scope scope, ByteNdArray data) {
-    try (Tensor<TUint8> value = TUint8.tensorOf(data)) {
+    try (TUint8 value = TUint8.tensorOf(data)) {
       return create(scope, value);
     }
   }
@@ -987,7 +987,7 @@ public static Constant<TUint8> tensorOf(Scope scope, ByteNdArray data) {
    */
   @Endpoint
   public static Constant<TUint8> tensorOf(Scope scope, Shape shape, ByteDataBuffer data) {
-    try (Tensor<TUint8> value = TUint8.tensorOf(shape, data)) {
+    try (TUint8 value = TUint8.tensorOf(shape, data)) {
       return create(scope, value);
     }
   }
@@ -1004,7 +1004,7 @@ public static Constant<TUint8> tensorOf(Scope scope, Shape shape, ByteDataBuffer
    *     buffer
    */
   @Endpoint
-  public static <T extends TType & Tensor> Constant<T> tensorOf(Scope scope, DataType<T> type, Shape shape,
+  public static <T extends Tensor & TType> Constant<T> tensorOf(Scope scope, DataType<T> type, Shape shape,
       ByteDataBuffer data) {
     try (T value = Tensor.of(type, shape, data)) {
       return create(scope, value);
@@ -1020,7 +1020,7 @@ public static <T extends TType & Tensor> Constant<T> tensorOf(Scope scope, DataT
    */
   @Endpoint
   public static Constant<TString> scalarOf(Scope scope, String data) {
-    try (Tensor<TString> value = TString.scalarOf(data)) {
+    try (TString value = TString.scalarOf(data)) {
       return create(scope, value);
     }
   }
@@ -1035,7 +1035,7 @@ public static Constant<TString> scalarOf(Scope scope, String data) {
    */
   @Endpoint
   public static Constant<TString> scalarOf(Scope scope, Charset charset, String data) {
-    try (Tensor<TString> value = TString.tensorOf(charset, NdArrays.scalarOfObject(data))) {
+    try (TString value = TString.tensorOf(charset, NdArrays.scalarOfObject(data))) {
       return create(scope, value);
     }
   }
@@ -1049,7 +1049,7 @@ public static Constant<TString> scalarOf(Scope scope, Charset charset, String da
    */
   public static Constant<TString> vectorOf(Scope scope, String[] data) {
     NdArray<String> src = NdArrays.vectorOfObjects(data);
-    try (Tensor<TString> value = TString.tensorOf(src)) {
+    try (TString value = TString.tensorOf(src)) {
       return create(scope, value);
     }
   }
@@ -1065,7 +1065,7 @@ public static Constant<TString> vectorOf(Scope scope, String[] data) {
    */
   @Endpoint
   public static Constant<TString> vectorOf(Scope scope, Charset charset, String[] data) {
-    try (Tensor<TString> value = TString.tensorOf(charset, NdArrays.vectorOfObjects(data))) {
+    try (TString value = TString.tensorOf(charset, NdArrays.vectorOfObjects(data))) {
       return Constant.create(scope, value);
     }
   }
@@ -1112,7 +1112,7 @@ public static Constant<TString> arrayOf(Scope scope, Charset charset, String...
   public static Constant<TString> tensorOf(Scope scope, String[][] data) {
     NdArray<String> src = NdArrays.ofObjects(String.class, StdArrays.shapeOf(data));
     StdArrays.copyTo(data, src);
-    try (Tensor<TString> value = TString.tensorOf(src)) {
+    try (TString value = TString.tensorOf(src)) {
       return create(scope, value);
     }
   }
@@ -1127,7 +1127,7 @@ public static Constant<TString> tensorOf(Scope scope, String[][] data) {
   public static Constant<TString> tensorOf(Scope scope, String[][][] data) {
     NdArray<String> src = NdArrays.ofObjects(String.class, StdArrays.shapeOf(data));
     StdArrays.copyTo(data, src);
-    try (Tensor<TString> value = TString.tensorOf(src)) {
+    try (TString value = TString.tensorOf(src)) {
       return create(scope, value);
     }
   }
@@ -1142,7 +1142,7 @@ public static Constant<TString> tensorOf(Scope scope, String[][][] data) {
   public static Constant<TString> tensorOf(Scope scope, String[][][][] data) {
     NdArray<String> src = NdArrays.ofObjects(String.class, StdArrays.shapeOf(data));
     StdArrays.copyTo(data, src);
-    try (Tensor<TString> value = TString.tensorOf(src)) {
+    try (TString value = TString.tensorOf(src)) {
       return create(scope, value);
     }
   }
@@ -1157,7 +1157,7 @@ public static Constant<TString> tensorOf(Scope scope, String[][][][] data) {
   public static Constant<TString> tensorOf(Scope scope, String[][][][][] data) {
     NdArray<String> src = NdArrays.ofObjects(String.class, StdArrays.shapeOf(data));
     StdArrays.copyTo(data, src);
-    try (Tensor<TString> value = TString.tensorOf(src)) {
+    try (TString value = TString.tensorOf(src)) {
       return create(scope, value);
     }
   }
@@ -1172,7 +1172,7 @@ public static Constant<TString> tensorOf(Scope scope, String[][][][][] data) {
   public static Constant<TString> tensorOf(Scope scope, String[][][][][][] data) {
     NdArray<String> src = NdArrays.ofObjects(String.class, StdArrays.shapeOf(data));
     StdArrays.copyTo(data, src);
-    try (Tensor<TString> value = TString.tensorOf(src)) {
+    try (TString value = TString.tensorOf(src)) {
       return create(scope, value);
     }
   }
@@ -1187,7 +1187,7 @@ public static Constant<TString> tensorOf(Scope scope, String[][][][][][] data) {
    */
   @Endpoint
   public static Constant<TString> tensorOf(Scope scope, NdArray<String> data) {
-    try (Tensor<TString> value = TString.tensorOf(data)) {
+    try (TString value = TString.tensorOf(data)) {
       return create(scope, value);
     }
   }
@@ -1203,7 +1203,7 @@ public static Constant<TString> tensorOf(Scope scope, NdArray<String> data) {
    */
   @Endpoint
   public static Constant<TString> tensorOf(Scope scope, Charset charset, NdArray<String> data) {
-    try (Tensor<TString> value = TString.tensorOf(charset, data)) {
+    try (TString value = TString.tensorOf(charset, data)) {
       return create(scope, value);
     }
   }
@@ -1220,7 +1220,7 @@ public static Constant<TString> tensorOf(Scope scope, Charset charset, NdArray<S
    */
   @Endpoint
   public static Constant<TString> tensorOf(Scope scope, Shape shape, DataBuffer<String> data) {
-    try (Tensor<TString> value = TString.tensorOf(shape, data)) {
+    try (TString value = TString.tensorOf(shape, data)) {
       return create(scope, value);
     }
   }
@@ -1238,7 +1238,7 @@ public static Constant<TString> tensorOf(Scope scope, Shape shape, DataBuffer<St
   @Endpoint
   public static Constant<TString> tensorOf(Scope scope, Charset charset, Shape shape,
       DataBuffer<String> data) {
-    try (Tensor<TString> value = TString.tensorOf(charset, shape, data)) {
+    try (TString value = TString.tensorOf(charset, shape, data)) {
       return create(scope, value);
     }
   }
@@ -1264,7 +1264,7 @@ public static Constant<TInt64> tensorOf(Scope scope, Shape shape) {
    * @return a constant of the same data type as `tensor`
    */
   @Endpoint
-  public static <T extends TType> Constant<T> create(Scope scope, Tensor<T> tensor) {
+  public static <T extends Tensor & TType> Constant<T> create(Scope scope, T tensor) {
     return new Constant<>(
         scope
             .env()
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/op/core/Gradients.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/op/core/Gradients.java
index 2827276c32c..f5192697628 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/op/core/Gradients.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/op/core/Gradients.java
@@ -22,12 +22,11 @@
 import org.tensorflow.Graph;
 import org.tensorflow.Operand;
 import org.tensorflow.Output;
-import org.tensorflow.op.Op;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.Operands;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Adds operations to compute the partial derivatives of sum of {@code y}s w.r.t {@code x}s,
@@ -158,7 +157,7 @@ public List<Output<?>> dy() {
    * @param index The index of the output among the gradients added by this operation
    */
   @SuppressWarnings("unchecked")
-  public <T extends TType> Output<T> dy(int index) {
+  public <T extends Tensor> Output<T> dy(int index) {
     return (Output<T>) dy.get(index);
   }
 
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/op/core/Helpers.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/op/core/Helpers.java
index f9ce837fe60..123f7eb826b 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/op/core/Helpers.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/op/core/Helpers.java
@@ -17,10 +17,10 @@
 
 import org.tensorflow.Operand;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
 
 /**
  * Container class for core methods which add or perform several operations
@@ -45,7 +45,7 @@ private Helpers() {}
      * @return a new instance of Variable
      */
     @Endpoint(name = "variable")
-    public static <T extends TType> Variable<T> createVariableWithInit(Scope scope, Operand<T> init, Variable.Options... options) {
+    public static <T extends Tensor> Variable<T> createVariableWithInit(Scope scope, Operand<T> init, Variable.Options... options) {
         Output<T> initOutput = init.asOutput();
         Variable<T> newVar = Variable.create(scope,initOutput.shape(), initOutput.dataType(), options);
         Assign<T> assignOp = Assign.create(scope, newVar, init);
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/op/core/Shapes.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/op/core/Shapes.java
index 613cb729341..14bc0604ce6 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/op/core/Shapes.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/op/core/Shapes.java
@@ -18,6 +18,7 @@
 
 import org.tensorflow.DataType;
 import org.tensorflow.Operand;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
@@ -31,7 +32,6 @@
 import org.tensorflow.types.TInt32;
 import org.tensorflow.types.TInt64;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * An operator providing methods on org.tensorflow.op.core.Shape tensors and 1d operands that
@@ -67,7 +67,7 @@ public abstract class Shapes {
    * @return the reshaped operand
    */
   @Endpoint(name = "flatten")
-  public static <T extends TType> Operand<T> flatten(Scope scope, Operand<T> operand) {
+  public static <T extends Tensor> Operand<T> flatten(Scope scope, Operand<T> operand) {
     return flatten(scope, operand, TInt32.DTYPE);
   }
 
@@ -82,7 +82,7 @@ public static <T extends TType> Operand<T> flatten(Scope scope, Operand<T> opera
    * @return the reshaped operand
    */
   @Endpoint(name = "flatten")
-  public static <T extends TType, U extends TNumber> Operand<T> flatten(
+  public static <T extends Tensor, U extends Tensor & TNumber> Operand<T> flatten(
       Scope scope, Operand<T> operand, DataType<U> dType) {
     Operand<U> flatShape = flatten(scope, Shape.create(scope, operand, dType), dType);
     return Reshape.create(scope, operand, flatShape);
@@ -110,7 +110,7 @@ public static Operand<TInt32> flatten(Scope scope, Shape<TInt32> shape) {
    * @return the flattened shape
    */
   @Endpoint(name = "flatten")
-  public static <U extends TNumber> Operand<U> flatten(
+  public static <U extends Tensor & TNumber> Operand<U> flatten(
       Scope scope, Shape<U> shape, DataType<U> dType) {
     return ExpandDims.create(
         scope,
@@ -140,7 +140,7 @@ public static Operand<TInt32> size(Scope scope, Shape<TInt32> shape) {
    * @return the size
    */
   @Endpoint(name = "size")
-  public static <U extends TNumber> Operand<U> size(
+  public static <U extends Tensor & TNumber> Operand<U> size(
       Scope scope, Shape<U> shape, DataType<U> dType) {
     Slice<U> dims =
         Slice.create(
@@ -178,7 +178,7 @@ public static Operand<TInt32> size(Scope scope, Shape<TInt32> shape, Operand<TIn
    * @return the size of the specified dimension
    */
   @Endpoint(name = "size")
-  public static <U extends TNumber> Operand<U> size(
+  public static <U extends Tensor & TNumber> Operand<U> size(
       Scope scope, Shape<U> shape, Operand<U> dim, DataType<U> dType) {
     return Slice.create(
         scope,
@@ -199,7 +199,7 @@ public static <U extends TNumber> Operand<U> size(
    * @return the size of the specified dimension
    */
   @Endpoint(name = "size")
-  public static <T extends TType> Operand<TInt32> size(
+  public static <T extends Tensor> Operand<TInt32> size(
       Scope scope, Operand<T> input, Operand<TInt32> dim) {
     return size(scope, input, dim, TInt32.DTYPE);
   }
@@ -215,7 +215,7 @@ public static <T extends TType> Operand<TInt32> size(
    * @return the size of the specified dimension
    */
   @Endpoint(name = "size")
-  public static <T extends TType, U extends TNumber> Operand<U> size(
+  public static <T extends Tensor, U extends Tensor & TNumber> Operand<U> size(
       Scope scope, Operand<T> input, Operand<U> dim, DataType<U> dType) {
     return size(scope, Shape.create(scope, input, dType), dim, dType);
   }
@@ -242,7 +242,7 @@ public static Operand<TInt32> numDimensions(Scope scope, Shape<TInt32> shape) {
    * @return the number of dimensions
    */
   @Endpoint(name = "numDimensions")
-  public static <U extends TNumber> Operand<U> numDimensions(
+  public static <U extends Tensor & TNumber> Operand<U> numDimensions(
       Scope scope, Shape<U> shape, DataType<U> dType) {
     return Size.create(scope, shape, dType);
   }
@@ -257,7 +257,7 @@ public static <U extends TNumber> Operand<U> numDimensions(
    * @return the reshaped operand
    */
   @Endpoint(name = "reduceDims")
-  public static <T extends TType> Operand<T> reduceDims(
+  public static <T extends Tensor> Operand<T> reduceDims(
       Scope scope, Operand<T> operand, Operand<TInt32> axis) {
     return reduceDims(scope, operand, axis, TInt32.DTYPE);
   }
@@ -274,7 +274,7 @@ public static <T extends TType> Operand<T> reduceDims(
    * @return the reshaped operand
    */
   @Endpoint(name = "reduceDims")
-  public static <T extends TType, U extends TNumber> Operand<T> reduceDims(
+  public static <T extends Tensor, U extends Tensor & TNumber> Operand<T> reduceDims(
       Scope scope, Operand<T> operand, Operand<U> axis, DataType<U> dType) {
     Shape<U> newShape = Shape.create(scope, operand, dType);
     return Reshape.create(scope, operand, reduceDims(scope, newShape, axis, dType));
@@ -304,7 +304,7 @@ public static Operand<TInt32> reduceDims(Scope scope, Shape<TInt32> shape, Opera
    * @return the reduced shape
    */
   @Endpoint(name = "reduceDims")
-  public static <U extends TNumber> Operand<U> reduceDims(
+  public static <U extends Tensor & TNumber> Operand<U> reduceDims(
       Scope scope, Shape<U> shape, Operand<U> axis, DataType<U> dType) {
     Size<U> rank = Size.create(scope, shape, dType);
     axis = FloorMod.create(scope, axis, rank);
@@ -356,7 +356,7 @@ public static Operand<TInt32> squeeze(Scope scope, Shape<TInt32> shape) {
    * @return the squeezed shape
    */
   @Endpoint(name = "squeeze")
-  public static <U extends TNumber> Operand<U> squeeze(
+  public static <U extends Tensor & TNumber> Operand<U> squeeze(
       Scope scope, Shape<U> shape, DataType<U> dType) {
     Operand<TBool> mask =
         NotEqual.create(scope, shape, Cast.create(scope, OnesLike.create(scope, shape), dType));
@@ -386,7 +386,7 @@ public static Operand<TInt32> head(Scope scope, Shape<TInt32> shape) {
    * @return a 1-dimensional Operand containing the Shape's first dimension
    */
   @Endpoint(name = "head")
-  public static <U extends TNumber> Operand<U> head(
+  public static <U extends Tensor & TNumber> Operand<U> head(
       Scope scope, Shape<U> shape, DataType<U> dType) {
     return take(scope, shape, Cast.create(scope, Constant.scalarOf(scope, 1), dType), dType);
   }
@@ -419,7 +419,7 @@ public static Operand<TInt32> take(Scope scope, Shape<TInt32> shape, Operand<TIn
    *     shape
    */
   @Endpoint(name = "take")
-  public static <U extends TNumber> Operand<U> take(
+  public static <U extends Tensor & TNumber> Operand<U> take(
       Scope scope, Shape<U> shape, Operand<U> n, DataType<U> dType) {
     return Slice.create(
         scope,
@@ -454,7 +454,7 @@ public static Operand<TInt32> tail(Scope scope, Shape<TInt32> shape) {
    *     Shape
    */
   @Endpoint(name = "tail")
-  public static <U extends TNumber> Operand<U> tail(
+  public static <U extends Tensor & TNumber> Operand<U> tail(
       Scope scope, Shape<U> shape, DataType<U> dType) {
     return takeLast(scope, shape, Cast.create(scope, Constant.scalarOf(scope, 1), dType), dType);
   }
@@ -470,7 +470,7 @@ public static <U extends TNumber> Operand<U> tail(
    *     shape
    */
   @Endpoint(name = "takeLast")
-  public static <U extends TNumber> Operand<TInt32> takeLast(
+  public static <U extends Tensor & TNumber> Operand<TInt32> takeLast(
       Scope scope, Shape<TInt32> shape, Operand<TInt32> n) {
     return takeLast(scope, shape, n, TInt32.DTYPE);
   }
@@ -488,7 +488,7 @@ public static <U extends TNumber> Operand<TInt32> takeLast(
    *     shape
    */
   @Endpoint(name = "takeLast")
-  public static <U extends TNumber> Operand<U> takeLast(
+  public static <U extends Tensor & TNumber> Operand<U> takeLast(
       Scope scope, Shape<U> shape, Operand<U> n, DataType<U> dType) {
 
     Size<U> rank = Size.create(scope, shape, dType);
@@ -548,7 +548,7 @@ public static Operand<TInt64> prepend(Scope scope, Shape<TInt64> shape, long fir
    *     representing the shape
    */
   @Endpoint(name = "prepend")
-  public static <T extends TNumber> Operand<T> prepend(
+  public static <T extends Tensor & TNumber> Operand<T> prepend(
       Scope scope, Operand<T> shape, Operand<T> shapeToPrepend) {
 
     return Concat.create(scope, Arrays.asList(shapeToPrepend, shape), Constant.scalarOf(scope, 0));
@@ -600,7 +600,7 @@ public static Operand<TInt64> append(Scope scope, Shape<TInt64> shape, long last
    *     to append
    */
   @Endpoint(name = "append")
-  public static <T extends TNumber> Operand<T> append(
+  public static <T extends Tensor & TNumber> Operand<T> append(
       Scope scope, Operand<T> shape, Operand<T> shapeToAppend) {
     return Concat.create(scope, Arrays.asList(shape, shapeToAppend), Constant.scalarOf(scope, 0));
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/op/core/Zeros.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/op/core/Zeros.java
index 4aad417b117..389be13e59d 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/op/core/Zeros.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/op/core/Zeros.java
@@ -18,6 +18,7 @@
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.Output;
+import org.tensorflow.Tensor;
 import org.tensorflow.op.Op;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -25,7 +26,6 @@
 import org.tensorflow.op.dtypes.Cast;
 import org.tensorflow.types.TString;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * An operator creating a constant initialized with zeros of the shape given by `dims`.
@@ -38,7 +38,7 @@
  * @param <T> constant type
  */
 @Operator
-public final class Zeros<T extends TType> implements Op, Operand<T> {
+public final class Zeros<T extends Tensor> implements Op, Operand<T> {
 
   /**
    * Creates a zeroed tensor given its type and shape.
@@ -51,7 +51,7 @@ public final class Zeros<T extends TType> implements Op, Operand<T> {
    */
   @Endpoint
   @SuppressWarnings("unchecked")
-  public static <T extends TType, U extends TNumber> Zeros<T> create(Scope scope, Operand<U> dims, DataType<T> type) {
+  public static <T extends Tensor, U extends Tensor & TNumber> Zeros<T> create(Scope scope, Operand<U> dims, DataType<T> type) {
     Scope zerosScope = scope.withSubScope("Zeros");
     Operand<T> zero;
     if (type == TString.DTYPE) {
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/tensor/BooleanTensor.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/tensor/BooleanTensor.java
new file mode 100644
index 00000000000..ac270e7ad0e
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/tensor/BooleanTensor.java
@@ -0,0 +1,40 @@
+package org.tensorflow.tensor;
+
+import org.tensorflow.DataType;
+import org.tensorflow.Tensor;
+import org.tensorflow.internal.c_api.TF_Tensor;
+import org.tensorflow.ndarray.BooleanNdArray;
+import org.tensorflow.ndarray.NdArray;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.ndarray.buffer.BooleanDataBuffer;
+import org.tensorflow.ndarray.buffer.ByteDataBuffer;
+import org.tensorflow.ndarray.buffer.DataBuffer;
+import org.tensorflow.ndarray.impl.dense.BooleanDenseNdArray;
+
+public interface BooleanTensor extends BooleanNdArray, Tensor<Boolean> {
+
+  @Override
+  BooleanTensor setBoolean(boolean value, long... coordinates);
+
+  @Override
+  BooleanTensor set(NdArray<Boolean> src, long... coordinates);
+
+  @Override
+  BooleanTensor setObject(Boolean value, long... coordinates);
+
+  @Override
+  BooleanTensor copyTo(NdArray<Boolean> dst);
+
+  @Override
+  BooleanTensor read(DataBuffer<Boolean> dst);
+
+  @Override
+  BooleanTensor read(BooleanDataBuffer dst);
+
+  @Override
+  BooleanTensor write(DataBuffer<Boolean> src);
+
+  @Override
+  BooleanTensor write(BooleanDataBuffer src);
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/tensor/ByteTensor.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/tensor/ByteTensor.java
new file mode 100644
index 00000000000..ba18867765b
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/tensor/ByteTensor.java
@@ -0,0 +1,39 @@
+package org.tensorflow.tensor;
+
+import org.tensorflow.DataType;
+import org.tensorflow.Tensor;
+import org.tensorflow.internal.c_api.TF_Tensor;
+import org.tensorflow.ndarray.ByteNdArray;
+import org.tensorflow.ndarray.NdArray;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.ndarray.buffer.ByteDataBuffer;
+import org.tensorflow.ndarray.buffer.DataBuffer;
+import org.tensorflow.ndarray.impl.dense.ByteDenseNdArray;
+
+public interface ByteTensor extends ByteNdArray, Tensor<Byte> {
+
+  @Override
+  ByteTensor setByte(byte value, long... coordinates);
+
+  @Override
+  ByteTensor set(NdArray<Byte> src, long... coordinates);
+
+  @Override
+  ByteTensor setObject(Byte value, long... coordinates);
+
+  @Override
+  ByteTensor copyTo(NdArray<Byte> dst);
+
+  @Override
+  ByteTensor read(DataBuffer<Byte> dst);
+
+  @Override
+  ByteTensor read(ByteDataBuffer dst);
+
+  @Override
+  ByteTensor write(DataBuffer<Byte> src);
+
+  @Override
+  ByteTensor write(ByteDataBuffer src);
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/tensor/DoubleTensor.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/tensor/DoubleTensor.java
new file mode 100644
index 00000000000..4f8b1b5b0c2
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/tensor/DoubleTensor.java
@@ -0,0 +1,40 @@
+package org.tensorflow.tensor;
+
+import org.tensorflow.DataType;
+import org.tensorflow.Tensor;
+import org.tensorflow.internal.c_api.TF_Tensor;
+import org.tensorflow.ndarray.DoubleNdArray;
+import org.tensorflow.ndarray.NdArray;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.ndarray.buffer.ByteDataBuffer;
+import org.tensorflow.ndarray.buffer.DataBuffer;
+import org.tensorflow.ndarray.buffer.DoubleDataBuffer;
+import org.tensorflow.ndarray.impl.dense.DoubleDenseNdArray;
+
+public interface DoubleTensor extends DoubleNdArray, Tensor<Double> {
+
+  @Override
+  DoubleTensor setDouble(double value, long... coordinates);
+
+  @Override
+  DoubleTensor set(NdArray<Double> src, long... coordinates);
+
+  @Override
+  DoubleTensor setObject(Double value, long... coordinates);
+
+  @Override
+  DoubleTensor copyTo(NdArray<Double> dst);
+
+  @Override
+  DoubleTensor read(DataBuffer<Double> dst);
+
+  @Override
+  DoubleTensor read(DoubleDataBuffer dst);
+
+  @Override
+  DoubleTensor write(DataBuffer<Double> src);
+
+  @Override
+  DoubleTensor write(DoubleDataBuffer src);
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/tensor/FloatTensor.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/tensor/FloatTensor.java
new file mode 100644
index 00000000000..c06e9c5f0d9
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/tensor/FloatTensor.java
@@ -0,0 +1,40 @@
+package org.tensorflow.tensor;
+
+import org.tensorflow.DataType;
+import org.tensorflow.Tensor;
+import org.tensorflow.internal.c_api.TF_Tensor;
+import org.tensorflow.ndarray.FloatNdArray;
+import org.tensorflow.ndarray.NdArray;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.ndarray.buffer.ByteDataBuffer;
+import org.tensorflow.ndarray.buffer.DataBuffer;
+import org.tensorflow.ndarray.buffer.FloatDataBuffer;
+import org.tensorflow.ndarray.impl.dense.FloatDenseNdArray;
+
+public interface FloatTensor extends FloatNdArray, Tensor<Float> {
+
+  @Override
+  FloatTensor setFloat(float value, long... coordinates);
+
+  @Override
+  FloatTensor set(NdArray<Float> src, long... coordinates);
+
+  @Override
+  FloatTensor setObject(Float value, long... coordinates);
+
+  @Override
+  FloatTensor copyTo(NdArray<Float> dst);
+
+  @Override
+  FloatTensor read(DataBuffer<Float> dst);
+
+  @Override
+  FloatTensor read(FloatDataBuffer dst);
+
+  @Override
+  FloatTensor write(DataBuffer<Float> src);
+
+  @Override
+  FloatTensor write(FloatDataBuffer src);
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/tensor/IntTensor.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/tensor/IntTensor.java
new file mode 100644
index 00000000000..52d3930a844
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/tensor/IntTensor.java
@@ -0,0 +1,40 @@
+package org.tensorflow.tensor;
+
+import org.tensorflow.DataType;
+import org.tensorflow.Tensor;
+import org.tensorflow.internal.c_api.TF_Tensor;
+import org.tensorflow.ndarray.IntNdArray;
+import org.tensorflow.ndarray.NdArray;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.ndarray.buffer.ByteDataBuffer;
+import org.tensorflow.ndarray.buffer.DataBuffer;
+import org.tensorflow.ndarray.buffer.IntDataBuffer;
+import org.tensorflow.ndarray.impl.dense.IntDenseNdArray;
+
+public interface IntTensor extends IntNdArray, Tensor<Integer> {
+
+  @Override
+  IntTensor setInt(int value, long... coordinates);
+
+  @Override
+  IntTensor set(NdArray<Integer> src, long... coordinates);
+
+  @Override
+  IntTensor setObject(Integer value, long... coordinates);
+
+  @Override
+  IntTensor copyTo(NdArray<Integer> dst);
+
+  @Override
+  IntTensor read(DataBuffer<Integer> dst);
+
+  @Override
+  IntTensor read(IntDataBuffer dst);
+
+  @Override
+  IntTensor write(DataBuffer<Integer> src);
+
+  @Override
+  IntTensor write(IntDataBuffer src);
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/tensor/LongTensor.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/tensor/LongTensor.java
new file mode 100644
index 00000000000..1a29bb278d6
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/tensor/LongTensor.java
@@ -0,0 +1,40 @@
+package org.tensorflow.tensor;
+
+import org.tensorflow.DataType;
+import org.tensorflow.Tensor;
+import org.tensorflow.internal.c_api.TF_Tensor;
+import org.tensorflow.ndarray.LongNdArray;
+import org.tensorflow.ndarray.NdArray;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.ndarray.buffer.ByteDataBuffer;
+import org.tensorflow.ndarray.buffer.DataBuffer;
+import org.tensorflow.ndarray.buffer.LongDataBuffer;
+import org.tensorflow.ndarray.impl.dense.LongDenseNdArray;
+
+public interface LongTensor extends LongNdArray, Tensor<Long> {
+
+  @Override
+  LongTensor setLong(long value, long... coordinates);
+
+  @Override
+  LongTensor set(NdArray<Long> src, long... coordinates);
+
+  @Override
+  LongTensor setObject(Long value, long... coordinates);
+
+  @Override
+  LongTensor copyTo(NdArray<Long> dst);
+
+  @Override
+  LongTensor read(DataBuffer<Long> dst);
+
+  @Override
+  LongTensor read(LongDataBuffer dst);
+
+  @Override
+  LongTensor write(DataBuffer<Long> src);
+
+  @Override
+  LongTensor write(LongDataBuffer src);
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/tensor/StringTensor.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/tensor/StringTensor.java
new file mode 100644
index 00000000000..78a9c2d43a0
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/tensor/StringTensor.java
@@ -0,0 +1,29 @@
+package org.tensorflow.tensor;
+
+import org.tensorflow.Tensor;
+import org.tensorflow.ndarray.NdArray;
+import org.tensorflow.ndarray.buffer.DataBuffer;
+
+public interface StringTensor extends Tensor<String> {
+
+  /**
+   * @return the tensor data as a n-dimensional array of raw byte sequences.
+   */
+  NdArray<byte[]> asBytes();
+
+  @Override
+  Tensor<String> set(NdArray<String> src, long... coordinates);
+
+  @Override
+  Tensor<String> setObject(String value, long... coordinates);
+
+  @Override
+  Tensor<String> copyTo(NdArray<String> dst);
+
+  @Override
+  Tensor<String> read(DataBuffer<String> dst);
+
+  @Override
+  Tensor<String> write(DataBuffer<String> src);
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/BooleanTensor.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/BooleanTensor.java
deleted file mode 100644
index 627062a990f..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/BooleanTensor.java
+++ /dev/null
@@ -1,126 +0,0 @@
-package org.tensorflow.types;
-
-import org.tensorflow.DataType;
-import org.tensorflow.Tensor;
-import org.tensorflow.internal.c_api.TF_Tensor;
-import org.tensorflow.ndarray.BooleanNdArray;
-import org.tensorflow.ndarray.NdArray;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.ndarray.buffer.BooleanDataBuffer;
-import org.tensorflow.ndarray.buffer.ByteDataBuffer;
-import org.tensorflow.ndarray.buffer.DataBuffer;
-import org.tensorflow.ndarray.impl.dense.BooleanDenseNdArray;
-import org.tensorflow.ndarray.index.Index;
-import org.tensorflow.types.family.TType;
-
-public interface BooleanTensor<T extends BooleanTensor & TType> extends BooleanNdArray, Tensor<T> {
-
-  @Override
-  T setBoolean(boolean value, long... coordinates);
-
-  @Override
-  T set(NdArray<Boolean> src, long... coordinates);
-
-  @Override
-  T setObject(Boolean value, long... coordinates);
-
-  @Override
-  T copyTo(NdArray<Boolean> dst);
-
-  @Override
-  T read(DataBuffer<Boolean> dst);
-
-  @Override
-  T read(BooleanDataBuffer dst);
-
-  @Override
-  T write(DataBuffer<Boolean> src);
-
-  @Override
-  T write(BooleanDataBuffer src);
-}
-
-class BooleanTensorImpl<T extends BooleanTensor & TType> extends BooleanDenseNdArray implements BooleanTensor<T> {
-
-  @Override
-  public TF_Tensor nativeHandle() {
-    return rawTensor.nativeHandle();
-  }
-
-  @Override
-  public DataType<T> dataType() {
-    return rawTensor.dataType();
-  }
-
-  @Override
-  public Shape shape() {
-    return rawTensor.shape();
-  }
-
-  @Override
-  public long numBytes() {
-    return rawTensor.numBytes();
-  }
-
-  @Override
-  public ByteDataBuffer rawData() {
-    return rawTensor.rawData();
-  }
-
-  @Override
-  public void close() {
-    rawTensor.close();
-  }
-
-  @Override
-  public String toString() {
-    return rawTensor.toString();
-  }
-
-  @Override
-  public T setBoolean(boolean value, long... coordinates) {
-    return (T)super.setBoolean(value, coordinates);
-  }
-
-  @Override
-  public T setObject(Boolean value, long... coordinates) {
-    return (T)super.setObject(value, coordinates);
-  }
-
-  @Override
-  public T set(NdArray<Boolean> src, long... coordinates) {
-    return (T)super.set(src, coordinates);
-  }
-
-  @Override
-  public T copyTo(NdArray<Boolean> dst) {
-    return (T)super.copyTo(dst);
-  }
-
-  @Override
-  public T read(DataBuffer<Boolean> dst) {
-    return (T)super.read(dst);
-  }
-
-  @Override
-  public T read(BooleanDataBuffer dst) {
-    return (T)super.read(dst);
-  }
-
-  @Override
-  public T write(DataBuffer<Boolean> src) {
-    return (T)super.write(src);
-  }
-
-  @Override
-  public T write(BooleanDataBuffer src) {
-    return (T)super.write(src);
-  }
-
-  BooleanTensorImpl(TF_Tensor nativeHandle, DataType<T> dataType, Shape shape, BooleanDataBuffer buffer) {
-    super(buffer, shape);
-    this.rawTensor = new RawTensor(nativeHandle, dataType, shape);
-  }
-
-  private final RawTensor rawTensor;
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/ByteTensor.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/ByteTensor.java
deleted file mode 100644
index 48451e3781c..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/ByteTensor.java
+++ /dev/null
@@ -1,125 +0,0 @@
-package org.tensorflow.types;
-
-import org.tensorflow.DataType;
-import org.tensorflow.Tensor;
-import org.tensorflow.internal.c_api.TF_Tensor;
-import org.tensorflow.ndarray.ByteNdArray;
-import org.tensorflow.ndarray.NdArray;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.ndarray.buffer.ByteDataBuffer;
-import org.tensorflow.ndarray.buffer.DataBuffer;
-import org.tensorflow.ndarray.impl.dense.ByteDenseNdArray;
-import org.tensorflow.ndarray.index.Index;
-import org.tensorflow.types.family.TType;
-
-public interface ByteTensor<T extends ByteTensor & TType> extends ByteNdArray, Tensor<T> {
-
-  @Override
-  T setByte(byte value, long... coordinates);
-
-  @Override
-  T set(NdArray<Byte> src, long... coordinates);
-
-  @Override
-  T setObject(Byte value, long... coordinates);
-
-  @Override
-  T copyTo(NdArray<Byte> dst);
-
-  @Override
-  T read(DataBuffer<Byte> dst);
-
-  @Override
-  T read(ByteDataBuffer dst);
-
-  @Override
-  T write(DataBuffer<Byte> src);
-
-  @Override
-  T write(ByteDataBuffer src);
-}
-
-class ByteTensorImpl<T extends ByteTensor & TType> extends ByteDenseNdArray implements ByteTensor<T> {
-
-  @Override
-  public TF_Tensor nativeHandle() {
-    return rawTensor.nativeHandle();
-  }
-
-  @Override
-  public DataType<T> dataType() {
-    return rawTensor.dataType();
-  }
-
-  @Override
-  public Shape shape() {
-    return rawTensor.shape();
-  }
-
-  @Override
-  public long numBytes() {
-    return rawTensor.numBytes();
-  }
-
-  @Override
-  public ByteDataBuffer rawData() {
-    return rawTensor.rawData();
-  }
-
-  @Override
-  public void close() {
-    rawTensor.close();
-  }
-
-  @Override
-  public String toString() {
-    return rawTensor.toString();
-  }
-
-  @Override
-  public T setByte(byte value, long... coordinates) {
-    return (T)super.setByte(value, coordinates);
-  }
-
-  @Override
-  public T setObject(Byte value, long... coordinates) {
-    return (T)super.setObject(value, coordinates);
-  }
-
-  @Override
-  public T set(NdArray<Byte> src, long... coordinates) {
-    return (T)super.set(src, coordinates);
-  }
-
-  @Override
-  public T copyTo(NdArray<Byte> dst) {
-    return (T)super.copyTo(dst);
-  }
-
-  @Override
-  public T read(DataBuffer<Byte> dst) {
-    return (T)super.read(dst);
-  }
-
-  @Override
-  public T read(ByteDataBuffer dst) {
-    return (T)super.read(dst);
-  }
-
-  @Override
-  public T write(DataBuffer<Byte> src) {
-    return (T)super.write(src);
-  }
-
-  @Override
-  public T write(ByteDataBuffer src) {
-    return (T)super.write(src);
-  }
-
-  ByteTensorImpl(TF_Tensor nativeHandle, DataType<T> dataType, Shape shape, ByteDataBuffer buffer) {
-    super(buffer, shape);
-    this.rawTensor = new RawTensor(nativeHandle, dataType, shape);
-  }
-
-  private final RawTensor rawTensor;
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/DoubleTensor.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/DoubleTensor.java
deleted file mode 100644
index 83fb0fa1c0d..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/DoubleTensor.java
+++ /dev/null
@@ -1,126 +0,0 @@
-package org.tensorflow.types;
-
-import org.tensorflow.DataType;
-import org.tensorflow.Tensor;
-import org.tensorflow.internal.c_api.TF_Tensor;
-import org.tensorflow.ndarray.DoubleNdArray;
-import org.tensorflow.ndarray.NdArray;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.ndarray.buffer.ByteDataBuffer;
-import org.tensorflow.ndarray.buffer.DataBuffer;
-import org.tensorflow.ndarray.buffer.DoubleDataBuffer;
-import org.tensorflow.ndarray.impl.dense.DoubleDenseNdArray;
-import org.tensorflow.ndarray.index.Index;
-import org.tensorflow.types.family.TType;
-
-public interface DoubleTensor<T extends DoubleTensor & TType> extends DoubleNdArray, Tensor<T> {
-
-  @Override
-  T setDouble(double value, long... coordinates);
-
-  @Override
-  T set(NdArray<Double> src, long... coordinates);
-
-  @Override
-  T setObject(Double value, long... coordinates);
-
-  @Override
-  T copyTo(NdArray<Double> dst);
-
-  @Override
-  T read(DataBuffer<Double> dst);
-
-  @Override
-  T read(DoubleDataBuffer dst);
-
-  @Override
-  T write(DataBuffer<Double> src);
-
-  @Override
-  T write(DoubleDataBuffer src);
-}
-
-class DoubleTensorImpl<T extends DoubleTensor & TType> extends DoubleDenseNdArray implements DoubleTensor<T> {
-
-  @Override
-  public TF_Tensor nativeHandle() {
-    return rawTensor.nativeHandle();
-  }
-
-  @Override
-  public DataType<T> dataType() {
-    return rawTensor.dataType();
-  }
-
-  @Override
-  public Shape shape() {
-    return rawTensor.shape();
-  }
-
-  @Override
-  public long numBytes() {
-    return rawTensor.numBytes();
-  }
-
-  @Override
-  public ByteDataBuffer rawData() {
-    return rawTensor.rawData();
-  }
-
-  @Override
-  public void close() {
-    rawTensor.close();
-  }
-
-  @Override
-  public String toString() {
-    return rawTensor.toString();
-  }
-
-  @Override
-  public T setDouble(double value, long... coordinates) {
-    return (T)super.setDouble(value, coordinates);
-  }
-
-  @Override
-  public T setObject(Double value, long... coordinates) {
-    return (T)super.setObject(value, coordinates);
-  }
-
-  @Override
-  public T set(NdArray<Double> src, long... coordinates) {
-    return (T)super.set(src, coordinates);
-  }
-
-  @Override
-  public T copyTo(NdArray<Double> dst) {
-    return (T)super.copyTo(dst);
-  }
-
-  @Override
-  public T read(DataBuffer<Double> dst) {
-    return (T)super.read(dst);
-  }
-
-  @Override
-  public T read(DoubleDataBuffer dst) {
-    return (T)super.read(dst);
-  }
-
-  @Override
-  public T write(DataBuffer<Double> src) {
-    return (T)super.write(src);
-  }
-
-  @Override
-  public T write(DoubleDataBuffer src) {
-    return (T)super.write(src);
-  }
-
-  DoubleTensorImpl(TF_Tensor nativeHandle, DataType<T> dataType, Shape shape, DoubleDataBuffer buffer) {
-    super(buffer, shape);
-    this.rawTensor = new RawTensor(nativeHandle, dataType, shape);
-  }
-
-  private final RawTensor rawTensor;
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/FloatTensor.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/FloatTensor.java
deleted file mode 100644
index f819d073a4d..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/FloatTensor.java
+++ /dev/null
@@ -1,126 +0,0 @@
-package org.tensorflow.types;
-
-import org.tensorflow.DataType;
-import org.tensorflow.Tensor;
-import org.tensorflow.internal.c_api.TF_Tensor;
-import org.tensorflow.ndarray.FloatNdArray;
-import org.tensorflow.ndarray.NdArray;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.ndarray.buffer.ByteDataBuffer;
-import org.tensorflow.ndarray.buffer.DataBuffer;
-import org.tensorflow.ndarray.buffer.FloatDataBuffer;
-import org.tensorflow.ndarray.impl.dense.FloatDenseNdArray;
-import org.tensorflow.ndarray.index.Index;
-import org.tensorflow.types.family.TType;
-
-public interface FloatTensor<T extends FloatTensor & TType> extends FloatNdArray, Tensor<T> {
-
-  @Override
-  T setFloat(float value, long... coordinates);
-
-  @Override
-  T set(NdArray<Float> src, long... coordinates);
-
-  @Override
-  T setObject(Float value, long... coordinates);
-
-  @Override
-  T copyTo(NdArray<Float> dst);
-
-  @Override
-  T read(DataBuffer<Float> dst);
-
-  @Override
-  T read(FloatDataBuffer dst);
-
-  @Override
-  T write(DataBuffer<Float> src);
-
-  @Override
-  T write(FloatDataBuffer src);
-}
-
-class FloatTensorImpl<T extends FloatTensor & TType> extends FloatDenseNdArray implements FloatTensor<T> {
-
-  @Override
-  public TF_Tensor nativeHandle() {
-    return rawTensor.nativeHandle();
-  }
-
-  @Override
-  public DataType<T> dataType() {
-    return rawTensor.dataType();
-  }
-
-  @Override
-  public Shape shape() {
-    return rawTensor.shape();
-  }
-
-  @Override
-  public long numBytes() {
-    return rawTensor.numBytes();
-  }
-
-  @Override
-  public ByteDataBuffer rawData() {
-    return rawTensor.rawData();
-  }
-
-  @Override
-  public void close() {
-    rawTensor.close();
-  }
-
-  @Override
-  public String toString() {
-    return rawTensor.toString();
-  }
-
-  @Override
-  public T setFloat(float value, long... coordinates) {
-    return (T)super.setFloat(value, coordinates);
-  }
-
-  @Override
-  public T setObject(Float value, long... coordinates) {
-    return (T)super.setObject(value, coordinates);
-  }
-
-  @Override
-  public T set(NdArray<Float> src, long... coordinates) {
-    return (T)super.set(src, coordinates);
-  }
-
-  @Override
-  public T copyTo(NdArray<Float> dst) {
-    return (T)super.copyTo(dst);
-  }
-
-  @Override
-  public T read(DataBuffer<Float> dst) {
-    return (T)super.read(dst);
-  }
-
-  @Override
-  public T read(FloatDataBuffer dst) {
-    return (T)super.read(dst);
-  }
-
-  @Override
-  public T write(DataBuffer<Float> src) {
-    return (T)super.write(src);
-  }
-
-  @Override
-  public T write(FloatDataBuffer src) {
-    return (T)super.write(src);
-  }
-
-  FloatTensorImpl(TF_Tensor nativeHandle, DataType<T> dataType, Shape shape, FloatDataBuffer buffer) {
-    super(buffer, shape);
-    this.rawTensor = new RawTensor(nativeHandle, dataType, shape);
-  }
-
-  private final RawTensor rawTensor;
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/IntTensor.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/IntTensor.java
deleted file mode 100644
index 041232e83d8..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/IntTensor.java
+++ /dev/null
@@ -1,126 +0,0 @@
-package org.tensorflow.types;
-
-import org.tensorflow.DataType;
-import org.tensorflow.Tensor;
-import org.tensorflow.internal.c_api.TF_Tensor;
-import org.tensorflow.ndarray.IntNdArray;
-import org.tensorflow.ndarray.NdArray;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.ndarray.buffer.ByteDataBuffer;
-import org.tensorflow.ndarray.buffer.DataBuffer;
-import org.tensorflow.ndarray.buffer.IntDataBuffer;
-import org.tensorflow.ndarray.impl.dense.IntDenseNdArray;
-import org.tensorflow.ndarray.index.Index;
-import org.tensorflow.types.family.TType;
-
-public interface IntTensor<T extends IntTensor & TType> extends IntNdArray, Tensor<T> {
-
-  @Override
-  T setInt(int value, long... coordinates);
-
-  @Override
-  T set(NdArray<Integer> src, long... coordinates);
-
-  @Override
-  T setObject(Integer value, long... coordinates);
-
-  @Override
-  T copyTo(NdArray<Integer> dst);
-
-  @Override
-  T read(DataBuffer<Integer> dst);
-
-  @Override
-  T read(IntDataBuffer dst);
-
-  @Override
-  T write(DataBuffer<Integer> src);
-
-  @Override
-  T write(IntDataBuffer src);
-}
-
-class IntTensorImpl<T extends IntTensor & TType> extends IntDenseNdArray implements IntTensor<T> {
-
-  @Override
-  public TF_Tensor nativeHandle() {
-    return rawTensor.nativeHandle();
-  }
-
-  @Override
-  public DataType<T> dataType() {
-    return rawTensor.dataType();
-  }
-
-  @Override
-  public Shape shape() {
-    return rawTensor.shape();
-  }
-
-  @Override
-  public long numBytes() {
-    return rawTensor.numBytes();
-  }
-
-  @Override
-  public ByteDataBuffer rawData() {
-    return rawTensor.rawData();
-  }
-
-  @Override
-  public void close() {
-    rawTensor.close();
-  }
-
-  @Override
-  public String toString() {
-    return rawTensor.toString();
-  }
-
-  @Override
-  public T setInt(int value, long... coordinates) {
-    return (T)super.setInt(value, coordinates);
-  }
-
-  @Override
-  public T setObject(Integer value, long... coordinates) {
-    return (T)super.setObject(value, coordinates);
-  }
-
-  @Override
-  public T set(NdArray<Integer> src, long... coordinates) {
-    return (T)super.set(src, coordinates);
-  }
-
-  @Override
-  public T copyTo(NdArray<Integer> dst) {
-    return (T)super.copyTo(dst);
-  }
-
-  @Override
-  public T read(DataBuffer<Integer> dst) {
-    return (T)super.read(dst);
-  }
-
-  @Override
-  public T read(IntDataBuffer dst) {
-    return (T)super.read(dst);
-  }
-
-  @Override
-  public T write(DataBuffer<Integer> src) {
-    return (T)super.write(src);
-  }
-
-  @Override
-  public T write(IntDataBuffer src) {
-    return (T)super.write(src);
-  }
-
-  IntTensorImpl(TF_Tensor nativeHandle, DataType<T> dataType, Shape shape, IntDataBuffer buffer) {
-    super(buffer, shape);
-    this.rawTensor = new RawTensor(nativeHandle, dataType, shape);
-  }
-
-  private final RawTensor rawTensor;
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/LongTensor.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/LongTensor.java
deleted file mode 100644
index b8965888f34..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/LongTensor.java
+++ /dev/null
@@ -1,126 +0,0 @@
-package org.tensorflow.types;
-
-import org.tensorflow.DataType;
-import org.tensorflow.Tensor;
-import org.tensorflow.internal.c_api.TF_Tensor;
-import org.tensorflow.ndarray.LongNdArray;
-import org.tensorflow.ndarray.NdArray;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.ndarray.buffer.ByteDataBuffer;
-import org.tensorflow.ndarray.buffer.DataBuffer;
-import org.tensorflow.ndarray.buffer.LongDataBuffer;
-import org.tensorflow.ndarray.impl.dense.LongDenseNdArray;
-import org.tensorflow.ndarray.index.Index;
-import org.tensorflow.types.family.TType;
-
-public interface LongTensor<T extends LongTensor & TType> extends LongNdArray, Tensor<T> {
-
-  @Override
-  T setLong(long value, long... coordinates);
-
-  @Override
-  T set(NdArray<Long> src, long... coordinates);
-
-  @Override
-  T setObject(Long value, long... coordinates);
-
-  @Override
-  T copyTo(NdArray<Long> dst);
-
-  @Override
-  T read(DataBuffer<Long> dst);
-
-  @Override
-  T read(LongDataBuffer dst);
-
-  @Override
-  T write(DataBuffer<Long> src);
-
-  @Override
-  T write(LongDataBuffer src);
-}
-
-class LongTensorImpl<T extends LongTensor & TType> extends LongDenseNdArray implements LongTensor<T> {
-
-  @Override
-  public TF_Tensor nativeHandle() {
-    return rawTensor.nativeHandle();
-  }
-
-  @Override
-  public DataType<T> dataType() {
-    return rawTensor.dataType();
-  }
-
-  @Override
-  public Shape shape() {
-    return rawTensor.shape();
-  }
-
-  @Override
-  public long numBytes() {
-    return rawTensor.numBytes();
-  }
-
-  @Override
-  public ByteDataBuffer rawData() {
-    return rawTensor.rawData();
-  }
-
-  @Override
-  public void close() {
-    rawTensor.close();
-  }
-
-  @Override
-  public String toString() {
-    return rawTensor.toString();
-  }
-
-  @Override
-  public T setLong(long value, long... coordinates) {
-    return (T)super.setLong(value, coordinates);
-  }
-
-  @Override
-  public T setObject(Long value, long... coordinates) {
-    return (T)super.setObject(value, coordinates);
-  }
-
-  @Override
-  public T set(NdArray<Long> src, long... coordinates) {
-    return (T)super.set(src, coordinates);
-  }
-
-  @Override
-  public T copyTo(NdArray<Long> dst) {
-    return (T)super.copyTo(dst);
-  }
-
-  @Override
-  public T read(DataBuffer<Long> dst) {
-    return (T)super.read(dst);
-  }
-
-  @Override
-  public T read(LongDataBuffer dst) {
-    return (T)super.read(dst);
-  }
-
-  @Override
-  public T write(DataBuffer<Long> src) {
-    return (T)super.write(src);
-  }
-
-  @Override
-  public T write(LongDataBuffer src) {
-    return (T)super.write(src);
-  }
-
-  LongTensorImpl(TF_Tensor nativeHandle, DataType<T> dataType, Shape shape, LongDataBuffer buffer) {
-    super(buffer, shape);
-    this.rawTensor = new RawTensor(nativeHandle, dataType, shape);
-  }
-
-  private final RawTensor rawTensor;
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TBfloat16.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TBfloat16.java
index 7e87343a979..efd69dc1db1 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TBfloat16.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TBfloat16.java
@@ -21,14 +21,16 @@
 import org.tensorflow.DataType;
 import org.tensorflow.Tensor;
 import org.tensorflow.exceptions.TensorFlowException;
-import org.tensorflow.internal.buffer.TensorBuffers;
+import org.tensorflow.internal.tensor.buffer.TensorBuffers;
 import org.tensorflow.internal.c_api.TF_Tensor;
+import org.tensorflow.internal.tensor.FloatTensorImpl;
 import org.tensorflow.ndarray.NdArray;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.ndarray.StdArrays;
 import org.tensorflow.ndarray.buffer.FloatDataBuffer;
 import org.tensorflow.ndarray.buffer.layout.DataLayouts;
-import org.tensorflow.types.family.TFloat;
+import org.tensorflow.tensor.FloatTensor;
+import org.tensorflow.types.family.TFloating;
 
 /**
  * Brain 16-bit float tensor type.
@@ -46,7 +48,8 @@
  * <p>Note that some CPUs support the bfloat16 format natively, which can result in faster
  * computation compared to {@link TFloat16} when GPUs are not used.
  */
-public interface TBfloat16 extends FloatNdArray, TNumber {
+public interface TBfloat16 extends FloatTensor, TFloating {
+
   /** readable-name for the data type */
   static final String NAME = "BFLOAT16";
 
@@ -125,7 +128,7 @@ static TBfloat16 tensorOf(Shape shape, Consumer<TBfloat16> tensorInit) {
 /**
  * Hidden implementation of a {@code TBfloat16}
  */
-class TBfloat16Impl extends FloatTensorImpl<TBfloat16> implements TBfloat16 {
+class TBfloat16Impl extends FloatTensorImpl implements TBfloat16 {
 
   TBfloat16Impl(TF_Tensor nativeTensorHandle, Shape shape) {
     super(nativeTensorHandle, DTYPE, shape, mapMemory(nativeTensorHandle));
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TBool.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TBool.java
index ec39f0a73b5..ad016a9db18 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TBool.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TBool.java
@@ -21,13 +21,15 @@
 import org.tensorflow.DataType;
 import org.tensorflow.Tensor;
 import org.tensorflow.exceptions.TensorFlowException;
-import org.tensorflow.internal.buffer.TensorBuffers;
+import org.tensorflow.internal.tensor.buffer.TensorBuffers;
 import org.tensorflow.internal.c_api.TF_Tensor;
+import org.tensorflow.internal.tensor.BooleanTensorImpl;
 import org.tensorflow.ndarray.NdArray;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.ndarray.StdArrays;
 import org.tensorflow.ndarray.buffer.BooleanDataBuffer;
 import org.tensorflow.ndarray.buffer.layout.DataLayouts;
+import org.tensorflow.tensor.BooleanTensor;
 import org.tensorflow.types.family.TType;
 
 /**
@@ -37,7 +39,7 @@
  * explicit mapping between Java boolean values and byte buffers using the {@link DataLayouts#BOOL
  * BOOL} layout, which may impact I/O performances.
  */
-public interface TBool extends BooleanTensor<TBool>, TType {
+public interface TBool extends BooleanTensor, TType {
 
   /** readable-name for the data type */
   static final String NAME = "BOOL";
@@ -117,7 +119,7 @@ static TBool tensorOf(Shape shape, Consumer<TBool> tensorInit) {
 /**
  * Hidden implementation of a {@code TBool}
  */
-class TBoolImpl extends BooleanTensorImpl<TBool> implements TBool {
+class TBoolImpl extends BooleanTensorImpl implements TBool {
 
   TBoolImpl(TF_Tensor nativeTensor, Shape shape) {
     super(nativeTensor, DTYPE, shape, TensorBuffers.toBooleans(nativeTensor));
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TFloat16.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TFloat16.java
index 87b083dc8f0..db86071514c 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TFloat16.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TFloat16.java
@@ -21,14 +21,16 @@
 import org.tensorflow.DataType;
 import org.tensorflow.Tensor;
 import org.tensorflow.exceptions.TensorFlowException;
-import org.tensorflow.internal.buffer.TensorBuffers;
+import org.tensorflow.internal.tensor.buffer.TensorBuffers;
 import org.tensorflow.internal.c_api.TF_Tensor;
+import org.tensorflow.internal.tensor.FloatTensorImpl;
 import org.tensorflow.ndarray.NdArray;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.ndarray.StdArrays;
 import org.tensorflow.ndarray.buffer.FloatDataBuffer;
 import org.tensorflow.ndarray.buffer.layout.DataLayouts;
-import org.tensorflow.types.family.TFloat;
+import org.tensorflow.tensor.FloatTensor;
+import org.tensorflow.types.family.TFloating;
 
 /**
  * IEEE-754 half-precision 16-bit float tensor type.
@@ -43,7 +45,7 @@
  * most CPUs do not support this format natively. For CPU computation on 16-bit floats, the {@link
  * TBfloat16} tensor type might be a better option.
  */
-public interface TFloat16 extends FloatTensor<TFloat16>, TFloat {
+public interface TFloat16 extends FloatTensor, TFloating {
 
   /** readable-name for the data type */
   static final String NAME = "FLOAT16";
@@ -123,7 +125,7 @@ static TFloat16 tensorOf(Shape shape, Consumer<TFloat16> tensorInit) {
 /**
  * Hidden implementation of a {@code TFloat16}
  */
-class TFloat16Impl extends FloatTensorImpl<TFloat16> implements TFloat16 {
+class TFloat16Impl extends FloatTensorImpl implements TFloat16 {
 
   TFloat16Impl(TF_Tensor nativeTensor, Shape shape) {
     super(nativeTensor, DTYPE, shape, mapMemory(nativeTensor));
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TFloat32.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TFloat32.java
index 73071c31ae6..96149a32e0e 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TFloat32.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TFloat32.java
@@ -21,16 +21,20 @@
 import org.tensorflow.DataType;
 import org.tensorflow.Tensor;
 import org.tensorflow.exceptions.TensorFlowException;
-import org.tensorflow.internal.buffer.TensorBuffers;
+import org.tensorflow.internal.tensor.buffer.TensorBuffers;
 import org.tensorflow.internal.c_api.TF_Tensor;
+import org.tensorflow.internal.tensor.FloatTensorImpl;
 import org.tensorflow.ndarray.NdArray;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.ndarray.StdArrays;
-import org.tensorflow.ndarray.impl.dense.FloatDenseNdArray;
-import org.tensorflow.types.family.TNumber;
+import org.tensorflow.ndarray.buffer.FloatDataBuffer;
+import org.tensorflow.tensor.FloatTensor;
+import org.tensorflow.types.family.TFloating;
 
-/** IEEE-754 single-precision 32-bit float tensor type. */
-public interface TFloat32 extends FloatNdArray, TNumber {
+/**
+ * IEEE-754 single-precision 32-bit float tensor type.
+ */
+public interface TFloat32 extends FloatTensor, TFloating {
 
   /** readable-name for the data type */
   static final String NAME = "FLOAT";
@@ -110,7 +114,7 @@ static TFloat32 tensorOf(Shape shape, Consumer<TFloat32> tensorInit) {
 /**
  * Hidden implementation of a {@code TFloat32}
  */
-class TFloat32Impl extends FloatTensorImpl<TFloat32> implements TFloat32 {
+class TFloat32Impl extends FloatTensorImpl implements TFloat32 {
 
   TFloat32Impl(TF_Tensor nativeTensor, Shape shape) {
     super(nativeTensor, DTYPE, shape, TensorBuffers.toFloats(nativeTensor));
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TFloat64.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TFloat64.java
index fa2af3f2e60..db455552950 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TFloat64.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TFloat64.java
@@ -21,16 +21,20 @@
 import org.tensorflow.DataType;
 import org.tensorflow.Tensor;
 import org.tensorflow.exceptions.TensorFlowException;
-import org.tensorflow.internal.buffer.TensorBuffers;
+import org.tensorflow.internal.tensor.buffer.TensorBuffers;
 import org.tensorflow.internal.c_api.TF_Tensor;
+import org.tensorflow.internal.tensor.DoubleTensorImpl;
 import org.tensorflow.ndarray.NdArray;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.ndarray.StdArrays;
-import org.tensorflow.ndarray.impl.dense.DoubleDenseNdArray;
-import org.tensorflow.types.family.TNumber;
+import org.tensorflow.ndarray.buffer.DoubleDataBuffer;
+import org.tensorflow.tensor.DoubleTensor;
+import org.tensorflow.types.family.TFloating;
 
-/** IEEE-754 double-precision 64-bit float tensor type. */
-public interface TFloat64 extends DoubleNdArray, TNumber {
+/**
+ * IEEE-754 double-precision 64-bit float tensor type.
+ */
+public interface TFloat64 extends DoubleTensor, TFloating {
 
   /** readable-name for the data type */
   static final String NAME = "DOUBLE";
@@ -110,7 +114,7 @@ static TFloat64 tensorOf(Shape shape, Consumer<TFloat64> tensorInit) {
 /**
  * Hidden implementation of a {@code TFloat64}
  */
-class TFloat64Impl extends DoubleTensorImpl<TFloat64> implements TFloat64 {
+class TFloat64Impl extends DoubleTensorImpl implements TFloat64 {
 
   TFloat64Impl(TF_Tensor nativeTensor, Shape shape) {
     super(nativeTensor, DTYPE, shape, TensorBuffers.toDoubles(nativeTensor));
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TInt32.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TInt32.java
index d84cb16b9f0..6218ac06398 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TInt32.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TInt32.java
@@ -20,18 +20,20 @@
 import java.util.function.Consumer;
 import org.tensorflow.DataType;
 import org.tensorflow.Tensor;
-import org.tensorflow.internal.buffer.TensorBuffers;
+import org.tensorflow.internal.tensor.buffer.TensorBuffers;
 import org.tensorflow.internal.c_api.TF_Tensor;
+import org.tensorflow.internal.tensor.IntTensorImpl;
 import org.tensorflow.ndarray.NdArray;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.ndarray.StdArrays;
 import org.tensorflow.ndarray.buffer.IntDataBuffer;
+import org.tensorflow.tensor.IntTensor;
 import org.tensorflow.types.family.TNumber;
 
 /**
  * 32-bit signed integer tensor type.
  */
-public interface TInt32 extends IntTensor<TInt32>, TNumber {
+public interface TInt32 extends IntTensor, TNumber {
 
   /** readable-name for the data type */
   static final String NAME = "INT32";
@@ -111,7 +113,7 @@ static TInt32 tensorOf(Shape shape, Consumer<TInt32> tensorInit) {
 /**
  * Hidden implementation of a {@code TInt32}
  */
-class TInt32Impl extends IntTensorImpl<TInt32> implements TInt32 {
+class TInt32Impl extends IntTensorImpl implements TInt32 {
 
   TInt32Impl(TF_Tensor nativeTensor, Shape shape) {
     super(nativeTensor, DTYPE, shape, TensorBuffers.toInts(nativeTensor));
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TInt64.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TInt64.java
index 5c15c379c09..1a55a19458f 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TInt64.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TInt64.java
@@ -21,18 +21,20 @@
 import org.tensorflow.DataType;
 import org.tensorflow.Tensor;
 import org.tensorflow.exceptions.TensorFlowException;
-import org.tensorflow.internal.buffer.TensorBuffers;
+import org.tensorflow.internal.tensor.buffer.TensorBuffers;
 import org.tensorflow.internal.c_api.TF_Tensor;
+import org.tensorflow.internal.tensor.LongTensorImpl;
 import org.tensorflow.ndarray.NdArray;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.ndarray.StdArrays;
 import org.tensorflow.ndarray.buffer.LongDataBuffer;
+import org.tensorflow.tensor.LongTensor;
 import org.tensorflow.types.family.TNumber;
 
 /**
  * 64-bit signed integer tensor type.
  */
-public interface TInt64 extends LongTensor<TInt64>, TNumber {
+public interface TInt64 extends LongTensor, TNumber {
 
   /** readable-name for the data type */
   static final String NAME = "INT64";
@@ -112,7 +114,7 @@ static TInt64 tensorOf(Shape shape, Consumer<TInt64> tensorInit) {
 /**
  * Hidden implementation of a {@code TInt64}
  */
-class TInt64Impl extends LongTensorImpl<TInt64> implements TInt64 {
+class TInt64Impl extends LongTensorImpl implements TInt64 {
 
   TInt64Impl(TF_Tensor nativeTensor, Shape shape) {
     super(nativeTensor, DTYPE, shape, TensorBuffers.toLongs(nativeTensor));
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TString.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TString.java
index 9cfcc0b0bbf..901857f24a8 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TString.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TString.java
@@ -22,15 +22,17 @@
 import java.util.function.Function;
 import org.tensorflow.DataType;
 import org.tensorflow.Tensor;
-import org.tensorflow.internal.buffer.ByteSequenceTensorBuffer;
-import org.tensorflow.internal.buffer.TensorBuffers;
+import org.tensorflow.internal.tensor.buffer.ByteSequenceTensorBuffer;
+import org.tensorflow.internal.tensor.buffer.TensorBuffers;
 import org.tensorflow.internal.c_api.TF_Tensor;
+import org.tensorflow.internal.tensor.StringTensorImpl;
 import org.tensorflow.ndarray.NdArray;
 import org.tensorflow.ndarray.NdArrays;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.ndarray.buffer.DataBuffer;
 import org.tensorflow.ndarray.buffer.layout.DataLayout;
 import org.tensorflow.ndarray.buffer.layout.DataLayouts;
+import org.tensorflow.tensor.StringTensor;
 import org.tensorflow.types.family.TType;
 
 /**
@@ -42,7 +44,7 @@
  * its values initially, so TensorFlow can compute and allocate the right amount of memory. Then the
  * data in the tensor is initialized once and cannot be modified afterwards.
  */
-public interface TString extends StringTensor<TString>, TType {
+public interface TString extends StringTensor, TType {
 
   /** readable-name for the data type */
   static final String NAME = "STRING";
@@ -219,7 +221,7 @@ static TString tensorOfBytes(Shape shape, DataBuffer<byte[]> data) {
 /**
  * Hidden implementation of a {@code TString}
  */
-class TStringImpl extends StringTensorImpl<TString> implements TString {
+class TStringImpl extends StringTensorImpl implements TString {
 
   @Override
   public TString using(Charset charset) {
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TUint8.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TUint8.java
index c00063585e0..5907cfbfc12 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TUint8.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TUint8.java
@@ -21,18 +21,20 @@
 import org.tensorflow.DataType;
 import org.tensorflow.Tensor;
 import org.tensorflow.exceptions.TensorFlowException;
-import org.tensorflow.internal.buffer.TensorBuffers;
+import org.tensorflow.internal.tensor.buffer.TensorBuffers;
 import org.tensorflow.internal.c_api.TF_Tensor;
+import org.tensorflow.internal.tensor.ByteTensorImpl;
 import org.tensorflow.ndarray.NdArray;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.ndarray.StdArrays;
 import org.tensorflow.ndarray.buffer.ByteDataBuffer;
+import org.tensorflow.tensor.ByteTensor;
 import org.tensorflow.types.family.TNumber;
 
 /**
  * 8-bit unsigned integer tensor type.
  */
-public interface TUint8 extends ByteTensor<TUint8>, TNumber {
+public interface TUint8 extends ByteTensor, TNumber {
 
   /** readable-name for the data type */
   static final String NAME = "UINT8";
@@ -112,7 +114,7 @@ static TUint8 tensorOf(Shape shape, Consumer<TUint8> tensorInit) {
 /**
  * Hidden implementation of a {@code TUint8}
  */
-class TUint8Impl extends ByteTensorImpl<TUint8> implements TUint8 {
+class TUint8Impl extends ByteTensorImpl implements TUint8 {
 
   TUint8Impl(TF_Tensor nativeTensor, Shape shape) {
     super(nativeTensor, DTYPE, shape, TensorBuffers.toBytes(nativeTensor));
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/family/package-info.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/family/package-info.java
index fcf1f554133..79e9e93fea3 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/family/package-info.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/family/package-info.java
@@ -21,8 +21,5 @@
  * <p>Some operations enforces that only operands of a type from a given family can be passed
  * in argument. For example, if an operation only allows numeric operands, such operands must be
  * bound to the {@link org.tensorflow.types.family.TNumber TNumber} interface.
- *
- * <p>All tensor types is bound to {@link org.tensorflow.types.family.TType TType}, which lays at
- * the root of the family hierarchy.
  */
 package org.tensorflow.types.family;
diff --git a/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/EagerOperationBuilderTest.java b/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/EagerOperationBuilderTest.java
index 6802ead9592..8fb3f22cd8d 100644
--- a/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/EagerOperationBuilderTest.java
+++ b/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/EagerOperationBuilderTest.java
@@ -93,7 +93,7 @@ public void setAttrs() {
     try (EagerSession session = EagerSession.create()) {
       Ops tf = Ops.create(session);
       // dtype, tensor attributes.
-      try (Tensor<TInt32> t = TInt32.scalarOf(1)) {
+      try (TInt32 t = TInt32.scalarOf(1)) {
         opBuilder(session, "Const", "DataTypeAndTensor")
             .setAttr("dtype", TInt32.DTYPE)
             .setAttr("value", t)
diff --git a/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/EagerOperationTest.java b/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/EagerOperationTest.java
index 09d2214cc6a..6f069506f5f 100644
--- a/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/EagerOperationTest.java
+++ b/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/EagerOperationTest.java
@@ -46,7 +46,7 @@ public void failToCreateIfSessionIsClosed() {
   @Test
   public void outputDataTypeAndShape() {
     try (EagerSession session = EagerSession.create();
-         Tensor<TInt32> t = TInt32.tensorOf(Shape.of(2, 3))) {
+         TInt32 t = TInt32.tensorOf(Shape.of(2, 3))) {
       EagerOperation op =
           opBuilder(session, "Const", "OutputAttrs")
               .setAttr("dtype", TInt32.DTYPE)
@@ -67,7 +67,7 @@ public void outputTensor() {
               .addInput(tf.constant(2).asOutput())
               .addInput(tf.constant(4).asOutput())
               .build();
-      assertEquals(6, add.tensor(0).expect(TInt32.DTYPE).data().getInt());
+      assertEquals(6, add.tensor(0).expect(TInt32.DTYPE).getInt());
 
       // Validate that we retrieve the right shape and datatype from the tensor
       // that has been resolved
diff --git a/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/GraphOperationBuilderTest.java b/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/GraphOperationBuilderTest.java
index 35bfa808238..7573a25ac13 100644
--- a/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/GraphOperationBuilderTest.java
+++ b/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/GraphOperationBuilderTest.java
@@ -50,7 +50,7 @@ public void failWhenMixingOperationsOnDifferentGraphs() {
   @Test
   public void failOnUseAfterBuild() {
     try (Graph g = new Graph();
-        Tensor<TInt32> t = TInt32.scalarOf(1)) {
+        TInt32 t = TInt32.scalarOf(1)) {
       OperationBuilder b =
           g.opBuilder("Const", "Const").setAttr("dtype", t.dataType()).setAttr("value", t);
       b.build();
@@ -66,7 +66,7 @@ public void failOnUseAfterBuild() {
   public void failOnUseAfterGraphClose() {
     OperationBuilder b = null;
     try (Graph g = new Graph();
-        Tensor<TInt32> t = TInt32.scalarOf(1)) {
+        TInt32 t = TInt32.scalarOf(1)) {
       b = g.opBuilder("Const", "Const").setAttr("dtype", t.dataType()).setAttr("value", t);
     }
     try {
@@ -88,7 +88,7 @@ public void setAttr() {
     try (Graph g = new Graph()) {
       Ops tf = Ops.create(g);
       // dtype, tensor attributes.
-      try (Tensor<TInt32> t = TInt32.scalarOf(1)) {
+      try (TInt32 t = TInt32.scalarOf(1)) {
         g.opBuilder("Const", "DataTypeAndTensor")
             .setAttr("dtype", TInt32.DTYPE)
             .setAttr("value", t)
@@ -169,8 +169,8 @@ public void setAttrShapeList() {
   public void addControlInput() {
     try (Graph g = new Graph();
         Session s = new Session(g);
-        Tensor<TBool> yes = TBool.scalarOf(true);
-        Tensor<TBool> no = TBool.scalarOf(false)) {
+        TBool yes = TBool.scalarOf(true);
+        TBool no = TBool.scalarOf(false)) {
       Ops tf = Ops.create(g);
       Output<TBool> placeholder = tf.placeholder(TBool.DTYPE).asOutput();
       GraphOperation check =
diff --git a/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/GraphTest.java b/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/GraphTest.java
index de376015e3f..efa6c09f2a5 100644
--- a/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/GraphTest.java
+++ b/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/GraphTest.java
@@ -157,8 +157,8 @@ public void addGradientsToGraph() {
       assertEquals(TFloat32.DTYPE, grads1[0].dataType());
       assertEquals(TFloat32.DTYPE, grads1[1].dataType());
       
-      try (Tensor<TFloat32> c1 = TFloat32.scalarOf(3.0f);
-          Tensor<TFloat32> c2 = TFloat32.scalarOf(2.0f);
+      try (TFloat32 c1 = TFloat32.scalarOf(3.0f);
+          TFloat32 c2 = TFloat32.scalarOf(2.0f);
           AutoCloseableList<Tensor<?>> outputs = new AutoCloseableList<>(
               s.runner()
                   .feed(x1, c1)
@@ -169,9 +169,9 @@ public void addGradientsToGraph() {
                   .run())) {
      
         assertEquals(3, outputs.size());
-        assertEquals(108.0f, outputs.get(0).expect(TFloat32.DTYPE).data().getFloat(), 0.0f);
-        assertEquals(6.0f, outputs.get(1).expect(TFloat32.DTYPE).data().getFloat(), 0.0f);
-        assertEquals(1.0f, outputs.get(2).expect(TFloat32.DTYPE).data().getFloat(), 0.0f);
+        assertEquals(108.0f, outputs.get(0).expect(TFloat32.DTYPE).getFloat(), 0.0f);
+        assertEquals(6.0f, outputs.get(1).expect(TFloat32.DTYPE).getFloat(), 0.0f);
+        assertEquals(1.0f, outputs.get(2).expect(TFloat32.DTYPE).getFloat(), 0.0f);
       }
     }
   }
@@ -191,14 +191,14 @@ public void addGradientSumsToGraph() {
       assertEquals(1, grad.length);
       assertEquals(TFloat32.DTYPE, grad[0].dataType());
 
-      try (Tensor<TFloat32> c = TFloat32.scalarOf(3.0f);
-          Tensor<TFloat32> output = s.runner()
+      try (TFloat32 c = TFloat32.scalarOf(3.0f);
+          TFloat32 output = s.runner()
               .feed(x, c)
               .fetch(grad[0])
               .run()
               .get(0)
               .expect(TFloat32.DTYPE)) {
-        assertEquals(114.0f, output.data().getFloat(), 0.0f);
+        assertEquals(114.0f, output.getFloat(), 0.0f);
       }
     }
   }
@@ -223,14 +223,14 @@ public void addGradientsWithInitialValuesToGraph() {
       assertEquals(1, grad1.length);
       assertEquals(TFloat32.DTYPE, grad1[0].dataType());
 
-      try (Tensor<TFloat32> c = TFloat32.scalarOf(3.0f);
-          Tensor<TFloat32> output = s.runner()
+      try (TFloat32 c = TFloat32.scalarOf(3.0f);
+          TFloat32 output = s.runner()
               .feed(x, c)
               .fetch(grad1[0])
               .run()
               .get(0)
               .expect(TFloat32.DTYPE)) {
-        assertEquals(108.0f, output.data().getFloat(), 0.0f);
+        assertEquals(108.0f, output.getFloat(), 0.0f);
       }
     }
   }
@@ -284,14 +284,14 @@ public void buildWhileLoopSingleInput() {
           },
           "test_loop");
 
-      try (Tensor<TInt32> c = TInt32.scalarOf(2);
-          Tensor<TInt32> output = s.runner()
+      try (TInt32 c = TInt32.scalarOf(2);
+          TInt32 output = s.runner()
               .feed(input, c)
               .fetch(loopOutputs[0])
               .run()
               .get(0)
               .expect(TInt32.DTYPE)) {
-        assertEquals(16, output.data().getInt()); // ((2^2)^2)
+        assertEquals(16, output.getInt()); // ((2^2)^2)
       }
     }
   }
@@ -320,8 +320,8 @@ public void buildWhileLoopMultipleInputs() {
           },
           "test_loop");
 
-      try (Tensor<TInt32> c1 = TInt32.scalarOf(2);
-          Tensor<TInt32> c2 = TInt32.scalarOf(5);
+      try (TInt32 c1 = TInt32.scalarOf(2);
+          TInt32 c2 = TInt32.scalarOf(5);
           AutoCloseableList<Tensor<?>> outputs =
               new AutoCloseableList<>(
                   s.runner()
@@ -331,8 +331,8 @@ public void buildWhileLoopMultipleInputs() {
                       .fetch(loopOutputs[1])
                       .run())) {
         assertEquals(2, outputs.size());
-        assertEquals(16, outputs.get(0).expect(TInt32.DTYPE).data().getInt()); // ((2^2)^2)
-        assertEquals(625, outputs.get(1).expect(TInt32.DTYPE).data().getInt()); // ((5^2)^2)
+        assertEquals(16, outputs.get(0).expect(TInt32.DTYPE).getInt()); // ((2^2)^2)
+        assertEquals(625, outputs.get(1).expect(TInt32.DTYPE).getInt()); // ((5^2)^2)
       }
     }
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/SessionTest.java b/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/SessionTest.java
index fa41af32a29..9e8191434bc 100644
--- a/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/SessionTest.java
+++ b/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/SessionTest.java
@@ -48,11 +48,11 @@ public void runUsingOperationNames() {
         Session s = new Session(g)) {
       Ops tf = Ops.create(g);
       transpose_A_times_X(tf, new int[][] {{2}, {3}});
-      try (Tensor<TInt32> x = TInt32.tensorOf(StdArrays.ndCopyOf(new int[][] {{5}, {7}}));
+      try (TInt32 x = TInt32.tensorOf(StdArrays.ndCopyOf(new int[][] {{5}, {7}}));
           AutoCloseableList<Tensor<?>> outputs =
               new AutoCloseableList<>(s.runner().feed("X", x).fetch("Y").run())) {
         assertEquals(1, outputs.size());
-        assertEquals(31, outputs.get(0).expect(TInt32.DTYPE).data().getInt(0, 0));
+        assertEquals(31, outputs.get(0).expect(TInt32.DTYPE).getInt(0, 0));
       }
     }
   }
@@ -65,11 +65,11 @@ public void runUsingOperationHandles() {
       transpose_A_times_X(tf, new int[][] {{2}, {3}});
       Output<TInt32> feed = g.operation("X").output(0);
       Output<TInt32> fetch = g.operation("Y").output(0);
-      try (Tensor<TInt32> x = TInt32.tensorOf(StdArrays.ndCopyOf(new int[][] {{5}, {7}}));
+      try (TInt32 x = TInt32.tensorOf(StdArrays.ndCopyOf(new int[][] {{5}, {7}}));
           AutoCloseableList<Tensor<?>> outputs =
               new AutoCloseableList<>(s.runner().feed(feed, x).fetch(fetch).run())) {
         assertEquals(1, outputs.size());
-        assertEquals(31, outputs.get(0).expect(TInt32.DTYPE).data().getInt(0, 0));
+        assertEquals(31, outputs.get(0).expect(TInt32.DTYPE).getInt(0, 0));
       }
     }
   }
@@ -83,14 +83,14 @@ public void runUsingColonSeparatedNames() {
       tf.math.add(split.output().get(0), split.output().get(1));
 
       // Fetch using colon separated names.
-      try (Tensor<TInt32> fetched =
+      try (TInt32 fetched =
           s.runner().fetch("Split:1").run().get(0).expect(TInt32.DTYPE)) {
-        assertEquals(3, fetched.data().getInt(0));
-        assertEquals(4, fetched.data().getInt(1));
+        assertEquals(3, fetched.getInt(0));
+        assertEquals(4, fetched.getInt(1));
       }
       // Feed using colon separated names.
-      try (Tensor<TInt32> fed = TInt32.vectorOf(4, 3, 2, 1);
-          Tensor<TInt32> fetched =
+      try (TInt32 fed = TInt32.vectorOf(4, 3, 2, 1);
+          TInt32 fetched =
               s.runner()
                   .feed("Split:0", fed)
                   .feed("Split:1", fed)
@@ -98,7 +98,7 @@ public void runUsingColonSeparatedNames() {
                   .run()
                   .get(0)
                   .expect(TInt32.DTYPE)) {
-        assertEquals(NdArrays.vectorOf(8, 6, 4, 2), fetched.data());
+        assertEquals(NdArrays.vectorOf(8, 6, 4, 2), fetched);
       }
     }
   }
@@ -109,7 +109,7 @@ public void runWithMetadata() {
         Session s = new Session(g)) {
       Ops tf = Ops.create(g);
       transpose_A_times_X(tf, new int[][] {{2}, {3}});
-      try (Tensor<TInt32> x = TInt32.tensorOf(StdArrays.ndCopyOf(new int[][] {{5}, {7}}))) {
+      try (TInt32 x = TInt32.tensorOf(StdArrays.ndCopyOf(new int[][] {{5}, {7}}))) {
         Session.Run result =
             s.runner()
                 .feed("X", x)
@@ -119,7 +119,7 @@ public void runWithMetadata() {
         // Sanity check on outputs.
         AutoCloseableList<Tensor<?>> outputs = new AutoCloseableList<>(result.outputs);
         assertEquals(1, outputs.size());
-        assertEquals(31, outputs.get(0).expect(TInt32.DTYPE).data().getInt(0, 0));
+        assertEquals(31, outputs.get(0).expect(TInt32.DTYPE).getInt(0, 0));
         // Sanity check on metadata
         assertNotNull(result.metadata);
         assertTrue(result.metadata.hasStepStats(), result.metadata.toString());
@@ -138,8 +138,8 @@ public void runMultipleOutputs() {
       AutoCloseableList<Tensor<?>> outputs =
           new AutoCloseableList<>(s.runner().fetch("c2").fetch("c1").run());
       assertEquals(2, outputs.size());
-      assertEquals(31415, outputs.get(0).expect(TInt32.DTYPE).data().getInt());
-      assertEquals(2718, outputs.get(1).expect(TInt32.DTYPE).data().getInt());
+      assertEquals(31415, outputs.get(0).expect(TInt32.DTYPE).getInt());
+      assertEquals(2718, outputs.get(1).expect(TInt32.DTYPE).getInt());
       outputs.close();
     }
   }
@@ -177,8 +177,8 @@ public void runInit() {
       try (Session s = new Session(g)) {
         s.run(tf.init());
 
-        try (Tensor<TInt32> t = s.runner().fetch(add).run().get(0).expect(TInt32.DTYPE)) {
-          assertEquals(30, t.data().getInt());
+        try (TInt32 t = s.runner().fetch(add).run().get(0).expect(TInt32.DTYPE)) {
+          assertEquals(30, t.getInt());
         }
       }
     }
@@ -198,8 +198,8 @@ public void runInitByName() {
       try (Session s = new Session(g)) {
         s.run("init_test");
 
-        try (Tensor<TInt32> t = s.runner().fetch(add).run().get(0).expect(TInt32.DTYPE)) {
-          assertEquals(30, t.data().getInt());
+        try (TInt32 t = s.runner().fetch(add).run().get(0).expect(TInt32.DTYPE)) {
+          assertEquals(30, t.getInt());
         }
         try {
           s.run("wrong_name");
diff --git a/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/TensorTest.java b/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/TensorTest.java
index 6c60480dfda..162f1ac1c95 100644
--- a/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/TensorTest.java
+++ b/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/TensorTest.java
@@ -64,22 +64,22 @@ public void createWithRawData() {
     String strings = "test";
     Shape strings_shape = Shape.scalar();
     byte[] strings_; // raw TF_STRING
-    try (Tensor<TString> t = TString.tensorOf(NdArrays.scalarOfObject(strings))) {
+    try (TString t = TString.tensorOf(NdArrays.scalarOfObject(strings))) {
       strings_ = new byte[(int)t.numBytes()];
       t.rawData().read(strings_);
     }
 
     // validate creating a tensor using a raw data byte buffers
     {
-      try (Tensor<TBool> t = Tensor.of(TBool.DTYPE, bools_shape, DataBuffers.of(bools_))) {
+      try (TBool t = Tensor.of(TBool.DTYPE, bools_shape, DataBuffers.of(bools_))) {
         boolean[] actual = new boolean[bools_.length];
-        t.data().read(DataBuffers.of(actual));
+        t.read(DataBuffers.of(actual));
         assertArrayEquals(bools, actual);
       }
 
       // note: the buffer is expected to contain raw TF_STRING (as per C API)
-      try (Tensor<TString> t = Tensor.of(TString.DTYPE, strings_shape, DataBuffers.of(strings_))) {
-        assertEquals(strings, t.data().getObject());
+      try (TString t = Tensor.of(TString.DTYPE, strings_shape, DataBuffers.of(strings_))) {
+        assertEquals(strings, t.getObject());
       }
     }
 
@@ -87,15 +87,15 @@ public void createWithRawData() {
     {
       DoubleBuffer buf = ByteBuffer.allocateDirect(8 * doubles.length).order(ByteOrder.nativeOrder())
           .asDoubleBuffer().put(doubles);
-      try (Tensor<TFloat64> t = TFloat64.tensorOf(doubles_shape, d -> d.write(DataBuffers.of(buf)))) {
+      try (TFloat64 t = TFloat64.tensorOf(doubles_shape, d -> d.write(DataBuffers.of(buf)))) {
         double[] actual = new double[doubles.length];
-        t.data().read(DataBuffers.of(actual));
+        t.read(DataBuffers.of(actual));
         assertArrayEquals(doubles, actual, EPSILON);
       }
     }
 
     // validate shape checking
-    try (Tensor<TBool> t = Tensor.of(TBool.DTYPE, Shape.of(bools_.length * 2), DataBuffers.of(bools_))) {
+    try (TBool t = Tensor.of(TBool.DTYPE, Shape.of(bools_.length * 2), DataBuffers.of(bools_))) {
       fail("should have failed on incompatible buffer");
     } catch (IllegalArgumentException e) {
       // expected
@@ -111,9 +111,9 @@ public void createFromBufferWithNativeByteOrder() {
             .asDoubleBuffer()
             .put(doubles);
     flipBuffer(buf);
-    try (Tensor<TFloat64> t = TFloat64.tensorOf(Shape.of(4), DataBuffers.of(buf))) {
+    try (TFloat64 t = TFloat64.tensorOf(Shape.of(4), DataBuffers.of(buf))) {
       double[] actual = new double[doubles.length];
-      t.data().read(DataBuffers.of(actual));
+      t.read(DataBuffers.of(actual));
       assertArrayEquals(doubles, actual, EPSILON);
     }
   }
@@ -130,9 +130,9 @@ public void createFromBufferWithNonNativeByteOrder() {
             .asDoubleBuffer()
             .put(doubles);
     flipBuffer(buf);
-    try (Tensor<TFloat64> t = TFloat64.tensorOf(Shape.of(4), DataBuffers.of(buf))) {
+    try (TFloat64 t = TFloat64.tensorOf(Shape.of(4), DataBuffers.of(buf))) {
       double[] actual = new double[doubles.length];
-      t.data().read(DataBuffers.of(actual));
+      t.read(DataBuffers.of(actual));
       assertArrayEquals(doubles, actual, EPSILON);
     }
   }
@@ -147,24 +147,24 @@ public void createWithTypedBuffer() {
     // validate creating a tensor using a typed buffer
     {
       Shape shape = Shape.of(4);
-      try (Tensor<TFloat64> t = TFloat64.tensorOf(shape, DataBuffers.of(doubles))) {
+      try (TFloat64 t = TFloat64.tensorOf(shape, DataBuffers.of(doubles))) {
         DoubleBuffer actual = DoubleBuffer.allocate(doubles.capacity());
-        t.data().read(DataBuffers.of(actual));
+        t.read(DataBuffers.of(actual));
         assertEquals(doubles, actual);
       }
-      try (Tensor<TFloat32> t = TFloat32.tensorOf(shape, DataBuffers.of(floats))) {
+      try (TFloat32 t = TFloat32.tensorOf(shape, DataBuffers.of(floats))) {
         FloatBuffer actual = FloatBuffer.allocate(floats.capacity());
-        t.data().read(DataBuffers.of(actual));
+        t.read(DataBuffers.of(actual));
         assertEquals(floats, actual);
       }
-      try (Tensor<TInt32> t = TInt32.tensorOf(shape, DataBuffers.of(ints))) {
+      try (TInt32 t = TInt32.tensorOf(shape, DataBuffers.of(ints))) {
         IntBuffer actual = IntBuffer.allocate(ints.capacity());
-        t.data().read(DataBuffers.of(actual));
+        t.read(DataBuffers.of(actual));
         assertEquals(ints, actual);
       }
-      try (Tensor<TInt64> t = TInt64.tensorOf(shape, DataBuffers.of(longs))) {
+      try (TInt64 t = TInt64.tensorOf(shape, DataBuffers.of(longs))) {
         LongBuffer actual = LongBuffer.allocate(longs.capacity());
-        t.data().read(DataBuffers.of(actual));
+        t.read(DataBuffers.of(actual));
         assertEquals(longs, actual);
       }
     }
@@ -172,22 +172,22 @@ public void createWithTypedBuffer() {
     // validate shape-checking
     {
       Shape shape = Shape.of(5);
-      try (Tensor<TFloat64> t = TFloat64.tensorOf(shape, DataBuffers.of(doubles))) {
+      try (TFloat64 t = TFloat64.tensorOf(shape, DataBuffers.of(doubles))) {
         fail("should have failed on incompatible buffer");
       } catch (BufferUnderflowException e) {
         // expected
       }
-      try (Tensor<TFloat32> t = TFloat32.tensorOf(shape, DataBuffers.of(floats))) {
+      try (TFloat32 t = TFloat32.tensorOf(shape, DataBuffers.of(floats))) {
         fail("should have failed on incompatible buffer");
       } catch (BufferUnderflowException e) {
         // expected
       }
-      try (Tensor<TInt32> t = TInt32.tensorOf(shape, DataBuffers.of(ints))) {
+      try (TInt32 t = TInt32.tensorOf(shape, DataBuffers.of(ints))) {
         fail("should have failed on incompatible buffer");
       } catch (BufferUnderflowException e) {
         // expected
       }
-      try (Tensor<TInt64> t = TInt64.tensorOf(shape, DataBuffers.of(longs))) {
+      try (TInt64 t = TInt64.tensorOf(shape, DataBuffers.of(longs))) {
         fail("should have failed on incompatible buffer");
       } catch (BufferUnderflowException e) {
         // expected
@@ -203,11 +203,11 @@ public void readFromRawData() {
     long[] longs = {1L, 2L, 3L};
     boolean[] bools = {true, false, true};
 
-    try (Tensor<TInt32> tints = TInt32.vectorOf(ints);
-        Tensor<TFloat32> tfloats = TFloat32.vectorOf(floats);
-        Tensor<TFloat64> tdoubles = TFloat64.vectorOf(doubles);
-        Tensor<TInt64> tlongs = TInt64.vectorOf(longs);
-        Tensor<TBool> tbools = TBool.vectorOf(bools)) {
+    try (TInt32 tints = TInt32.vectorOf(ints);
+        TFloat32 tfloats = TFloat32.vectorOf(floats);
+        TFloat64 tdoubles = TFloat64.vectorOf(doubles);
+        TInt64 tlongs = TInt64.vectorOf(longs);
+        TBool tbools = TBool.vectorOf(bools)) {
 
       // validate that any datatype is readable with ByteBuffer (content, position)
       {
@@ -266,79 +266,79 @@ public void readFromRawData() {
 
   @Test
   public void scalars() {
-    try (Tensor<TFloat32> t = TFloat32.scalarOf(2.718f)) {
+    try (TFloat32 t = TFloat32.scalarOf(2.718f)) {
       assertEquals(TFloat32.DTYPE, t.dataType());
       assertEquals(0, t.shape().numDimensions());
-      assertEquals(2.718f, t.data().getFloat(), EPSILON_F);
+      assertEquals(2.718f, t.getFloat(), EPSILON_F);
     }
 
-    try (Tensor<TFloat64> t = TFloat64.scalarOf(3.1415)) {
+    try (TFloat64 t = TFloat64.scalarOf(3.1415)) {
       assertEquals(TFloat64.DTYPE, t.dataType());
       assertEquals(0, t.shape().numDimensions());
-      assertEquals(3.1415, t.data().getDouble(), EPSILON);
+      assertEquals(3.1415, t.getDouble(), EPSILON);
     }
 
-    try (Tensor<TInt32> t = TInt32.scalarOf(-33)) {
+    try (TInt32 t = TInt32.scalarOf(-33)) {
       assertEquals(TInt32.DTYPE, t.dataType());
       assertEquals(0, t.shape().numDimensions());
-      assertEquals(-33, t.data().getInt());
+      assertEquals(-33, t.getInt());
     }
 
-    try (Tensor<TInt64> t = TInt64.scalarOf(8589934592L)) {
+    try (TInt64 t = TInt64.scalarOf(8589934592L)) {
       assertEquals(TInt64.DTYPE, t.dataType());
       assertEquals(0, t.shape().numDimensions());
-      assertEquals(8589934592L, t.data().getLong());
+      assertEquals(8589934592L, t.getLong());
     }
 
-    try (Tensor<TBool> t = TBool.scalarOf(true)) {
+    try (TBool t = TBool.scalarOf(true)) {
       assertEquals(TBool.DTYPE, t.dataType());
       assertEquals(0, t.shape().numDimensions());
-      assertTrue(t.data().getBoolean());
+      assertTrue(t.getBoolean());
     }
 
-    try (Tensor<TString> t = TString.scalarOf("sombrero")) {
+    try (TString t = TString.scalarOf("sombrero")) {
       assertEquals(TString.DTYPE, t.dataType());
       assertEquals(0, t.shape().numDimensions());
-      assertEquals("sombrero", t.data().getObject());
+      assertEquals("sombrero", t.getObject());
     }
 
     final byte[] bytes = {1, 2, 3, 4};
-    try (Tensor<TString> t = TString.tensorOfBytes(NdArrays.scalarOfObject(bytes))) {
+    try (TString t = TString.tensorOfBytes(NdArrays.scalarOfObject(bytes))) {
       assertEquals(TString.DTYPE, t.dataType());
       assertEquals(0, t.shape().numDimensions());
-      assertArrayEquals(bytes, t.data().asBytes().getObject());
+      assertArrayEquals(bytes, t.asBytes().getObject());
     }
   }
 
   @Test
   public void nDimensional() {
     DoubleNdArray vector = StdArrays.ndCopyOf(new double[]{1.414, 2.718, 3.1415});
-    try (Tensor<TFloat64> t = TFloat64.tensorOf(vector)) {
+    try (TFloat64 t = TFloat64.tensorOf(vector)) {
       assertEquals(TFloat64.DTYPE, t.dataType());
       assertEquals(1, t.shape().numDimensions());
       assertEquals(3, t.shape().size(0));
-      assertEquals(vector, t.data());
+      assertEquals(vector, t);
     }
 
     IntNdArray matrix = StdArrays.ndCopyOf(new int[][]{{1, 2, 3}, {4, 5, 6}});
-    try (Tensor<TInt32> t = TInt32.tensorOf(matrix)) {
+    try (TInt32 t = TInt32.tensorOf(matrix)) {
       assertEquals(TInt32.DTYPE, t.dataType());
       assertEquals(2, t.shape().numDimensions());
       assertEquals(2, t.shape().size(0));
       assertEquals(3, t.shape().size(1));
-      assertEquals(matrix, t.data());
+      assertEquals(matrix, t);
     }
 
     LongNdArray threeD = StdArrays.ndCopyOf(new long[][][]{
       {{1}, {3}, {5}, {7}, {9}}, {{2}, {4}, {6}, {8}, {0}},
     });
-    try (Tensor<TInt64> t = TInt64.tensorOf(threeD)) {
+    try (TInt64 t = TInt64.tensorOf(threeD)) {
       assertEquals(TInt64.DTYPE, t.dataType());
       assertEquals(3, t.shape().numDimensions());
       assertEquals(2, t.shape().size(0));
       assertEquals(5, t.shape().size(1));
       assertEquals(1, t.shape().size(2));
-      assertEquals(threeD, t.data());
+      assertEquals(threeD, t);
     }
 
     BooleanNdArray fourD = StdArrays.ndCopyOf(new boolean[][][][]{
@@ -346,14 +346,14 @@ public void nDimensional() {
       {{{false, false, true, true}, {false, true, false, false}}},
       {{{false, true, false, true}, {false, true, true, false}}},
     });
-    try (Tensor<TBool> t = TBool.tensorOf(fourD)) {
+    try (TBool t = TBool.tensorOf(fourD)) {
       assertEquals(TBool.DTYPE, t.dataType());
       assertEquals(4, t.shape().numDimensions());
       assertEquals(3, t.shape().size(0));
       assertEquals(1, t.shape().size(1));
       assertEquals(2, t.shape().size(2));
       assertEquals(4, t.shape().size(3));
-      assertEquals(fourD, t.data());
+      assertEquals(fourD, t);
     }
   }
 
@@ -365,36 +365,36 @@ public void testNDimensionalStringTensor() {
         matrix.setObject(String.format("(%d, %d) = %d", i, j, i << j), i, j);
       }
     }
-    try (Tensor<TString> t = TString.tensorOf(matrix)) {
+    try (TString t = TString.tensorOf(matrix)) {
       assertEquals(TString.DTYPE, t.dataType());
       assertEquals(2, t.shape().numDimensions());
       assertEquals(4, t.shape().size(0));
       assertEquals(3, t.shape().size(1));
-      assertEquals(matrix, t.data());
+      assertEquals(matrix, t);
     }
 
     NdArray<byte[]> byteMatrix = NdArrays.ofObjects(byte[].class, matrix.shape());
     matrix.scalars().forEachIndexed((i, s) -> byteMatrix.setObject(s.getObject().getBytes(UTF_8), i));
-    try (Tensor<TString> t = TString.tensorOfBytes(byteMatrix)) {
+    try (TString t = TString.tensorOfBytes(byteMatrix)) {
       assertEquals(TString.DTYPE, t.dataType());
       assertEquals(2, t.shape().numDimensions());
       assertEquals(4, t.shape().size(0));
       assertEquals(3, t.shape().size(1));
-      assertEquals(byteMatrix, t.data().asBytes());
-      assertEquals(matrix, t.data());
+      assertEquals(byteMatrix, t.asBytes());
+      assertEquals(matrix, t);
     }
   }
 
   @Test
   public void testUint8TensorFromArray() {
     byte[] vector = new byte[] {1, 2, 3, 4};
-    try (Tensor<TUint8> t = TUint8.vectorOf(vector)) {
+    try (TUint8 t = TUint8.vectorOf(vector)) {
       assertEquals(TUint8.DTYPE, t.dataType());
       assertEquals(1, t.shape().numDimensions());
       assertEquals(4, t.shape().size(0));
 
       byte[] got = new byte[4];
-      t.data().read(DataBuffers.of(got));
+      t.read(DataBuffers.of(got));
       assertArrayEquals(vector, got);
     }
   }
@@ -402,13 +402,13 @@ public void testUint8TensorFromArray() {
   @Test
   public void testCreateFromArrayOfBoxed() {
     Integer[] vector = new Integer[] {1, 2, 3, 4};
-    try (Tensor<TInt32> t = TInt32.tensorOf(Shape.of(4), d -> d.write(DataBuffers.ofObjects(vector)))) {
+    try (TInt32 t = TInt32.tensorOf(Shape.of(4), d -> d.write(DataBuffers.ofObjects(vector)))) {
       assertEquals(TInt32.DTYPE, t.dataType());
       assertEquals(1, t.shape().numDimensions());
       assertEquals(4, t.shape().size(0));
 
       Integer[] got = new Integer[4];
-      t.data().read(DataBuffers.ofObjects(got));
+      t.read(DataBuffers.ofObjects(got));
       assertArrayEquals(vector, got);
     }
   }
@@ -421,7 +421,7 @@ public void failCreateOnMismatchedDimensions() {
         invalid[x][y] = new int[x + y + 1];
       }
     }
-    try (Tensor<TInt32> t = TInt32.tensorOf(StdArrays.ndCopyOf(invalid))) {
+    try (TInt32 t = TInt32.tensorOf(StdArrays.ndCopyOf(invalid))) {
       fail("Tensor.create() should fail because of differing sizes in the 3rd dimension");
     } catch (IllegalArgumentException e) {
       // The expected exception.
@@ -433,11 +433,11 @@ public void tensorWithZeroDimension() {
     // Note: Historically, TF Java failed on purpose when trying to allocate a tensor with a shape
     // that has one or more dimensions set to 0 elements. But Python API allows it, so we should do
     // the same.
-    try (Tensor<TInt32> t = TInt32.tensorOf(Shape.of(3, 0, 1))) {
+    try (TInt32 t = TInt32.tensorOf(Shape.of(3, 0, 1))) {
       assertEquals(0, t.numBytes());
       assertEquals(0, t.shape().size());
     }
-    try (Tensor<TInt32> t = TInt32.tensorOf(StdArrays.ndCopyOf(new int[3][0][1]))) {
+    try (TInt32 t = TInt32.tensorOf(StdArrays.ndCopyOf(new int[3][0][1]))) {
       assertEquals(0, t.numBytes());
       assertEquals(0, t.shape().size());
     }
@@ -445,10 +445,10 @@ public void tensorWithZeroDimension() {
 
   @Test
   public void allocateTensorWithSize() {
-    try (Tensor<TInt32> t = Tensor.of(TInt32.DTYPE, Shape.of(2, 2, 2), 8 * TInt32.DTYPE.byteSize())) {
+    try (TInt32 t = Tensor.of(TInt32.DTYPE, Shape.of(2, 2, 2), 8 * TInt32.DTYPE.byteSize())) {
       // ok
     }
-    try (Tensor<TInt32> t = Tensor.of(TInt32.DTYPE, Shape.of(2, 2, 2), 9 * TInt32.DTYPE.byteSize())) {
+    try (TInt32 t = Tensor.of(TInt32.DTYPE, Shape.of(2, 2, 2), 9 * TInt32.DTYPE.byteSize())) {
       // ok (size requested is larger that minimum space required)
     }
     try {
@@ -473,12 +473,12 @@ public void useAfterClose() {
 
   @Test
   public void eagerTensorIsReleasedAfterSessionIsClosed() {
-    Tensor<TInt32> sum;
+    TInt32 sum;
     try (EagerSession session = EagerSession.create()) {
       Ops tf = Ops.create(session);
       sum = tf.math.add(tf.constant(10), tf.constant(20)).asTensor();
       sum.nativeHandle(); // does not throw
-      assertEquals(30, sum.data().getInt());
+      assertEquals(30, sum.getInt());
     }
     try {
       sum.nativeHandle();
@@ -487,7 +487,7 @@ public void eagerTensorIsReleasedAfterSessionIsClosed() {
       // as expected
     }
     try {
-      sum.data().getInt();
+      sum.getInt();
       fail("Tensor data should not be accessible after tensor is closed");
     } catch (IllegalStateException e) {
       // as expected
@@ -503,12 +503,12 @@ public void fromHandle() {
     // An exception is made for this test, where the pitfalls of this is avoided by not calling
     // close() on both Tensors.
     final FloatNdArray matrix = StdArrays.ndCopyOf(new float[][]{{1, 2, 3}, {4, 5, 6}});
-    try (Tensor<TFloat32> src = TFloat32.tensorOf(matrix)) {
-      Tensor<TFloat32> cpy = Tensors.fromHandle(src.nativeHandle()).expect(TFloat32.DTYPE);
+    try (TFloat32 src = TFloat32.tensorOf(matrix)) {
+      TFloat32 cpy = Tensors.fromHandle(src.nativeHandle()).expect(TFloat32.DTYPE);
       assertEquals(src.dataType(), cpy.dataType());
       assertEquals(src.shape().numDimensions(), cpy.shape().numDimensions());
       assertEquals(src.shape(), cpy.shape());
-      assertEquals(matrix, cpy.data());
+      assertEquals(matrix, cpy);
     }
   }
 
diff --git a/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/op/ScopeTest.java b/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/op/ScopeTest.java
index bbebfd5f454..288c7d7b18b 100644
--- a/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/op/ScopeTest.java
+++ b/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/op/ScopeTest.java
@@ -25,7 +25,6 @@
 import org.tensorflow.Session;
 import org.tensorflow.Tensor;
 import org.tensorflow.types.TInt32;
-import org.tensorflow.types.family.TType;
 
 /** Unit tests for {@link org.tensorflow.op.Scope}. */
 public class ScopeTest {
@@ -169,16 +168,16 @@ public void composite() {
       // assertNotNull(g.operation("variance/zero"));
 
       // Verify correct results as well.
-      Tensor<TInt32> result =
+      TInt32 result =
           sess.runner().fetch(var1.output()).run().get(0).expect(TInt32.DTYPE);
-      assertEquals(21704, result.data().getInt());
+      assertEquals(21704, result.getInt());
       result = sess.runner().fetch(var2.output()).run().get(0).expect(TInt32.DTYPE);
-      assertEquals(21704, result.data().getInt());
+      assertEquals(21704, result.getInt());
     }
   }
 
   // "handwritten" sample operator classes
-  private static final class Const<T extends TType> {
+  private static final class Const<T extends Tensor> {
     private final Output<T> output;
 
     static Const<TInt32> create(Scope s, int v) {
@@ -189,7 +188,7 @@ static Const<TInt32> create(Scope s, int[] v) {
       return create(s, TInt32.vectorOf(v));
     }
 
-    static <T extends TType> Const<T> create(Scope s, Tensor<T> value) {
+    static <T extends Tensor> Const<T> create(Scope s, T value) {
       return new Const<>(
           s.env()
               .opBuilder("Const", s.makeOpName("Const"))
@@ -208,10 +207,10 @@ Output<T> output() {
     }
   }
 
-  private static final class Mean<T extends TType> {
+  private static final class Mean<T extends Tensor> {
     private final Output<T> output;
 
-    static <T extends TType> Mean<T> create(Scope s, Output<T> input, Output<T> reductionIndices) {
+    static <T extends Tensor> Mean<T> create(Scope s, Output<T> input, Output<T> reductionIndices) {
       return new Mean<>(
           s.env()
               .opBuilder("Mean", s.makeOpName("Mean"))
@@ -230,10 +229,10 @@ Output<T> output() {
     }
   }
 
-  private static final class SquaredDifference<T extends TType> {
+  private static final class SquaredDifference<T extends Tensor> {
     private final Output<T> output;
 
-    static <T extends TType> SquaredDifference<T> create(Scope s, Output<T> x, Output<T> y) {
+    static <T extends Tensor> SquaredDifference<T> create(Scope s, Output<T> x, Output<T> y) {
       return new SquaredDifference<>(
           s.env()
               .opBuilder("SquaredDifference", s.makeOpName("SquaredDifference"))
@@ -252,7 +251,7 @@ Output<T> output() {
     }
   }
 
-  private static final class Variance<T extends TType> {
+  private static final class Variance<T extends Tensor> {
     private final Output<T> output;
 
     static Variance<TInt32> create(Scope base, Output<TInt32> x) {
diff --git a/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/op/core/GeneratedOperationsTest.java b/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/op/core/GeneratedOperationsTest.java
index a337bd73098..8d0c4764a91 100644
--- a/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/op/core/GeneratedOperationsTest.java
+++ b/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/op/core/GeneratedOperationsTest.java
@@ -36,8 +36,8 @@ public void tensorInputTensorOutput() {
         Session sess = new Session(g)) {
       Ops ops = Ops.create(g);
       Operand<TInt32> x = ops.math.add(ops.constant(1), ops.constant(2));
-      try (Tensor<TInt32> result = sess.runner().fetch(x).run().get(0).expect(TInt32.DTYPE)) {
-        assertEquals(3, result.data().getInt());
+      try (TInt32 result = sess.runner().fetch(x).run().get(0).expect(TInt32.DTYPE)) {
+        assertEquals(3, result.getInt());
       }
     }
   }
@@ -52,8 +52,8 @@ public void testListInputTensorOutput() {
       inputs.add(ops.constant(2));
       inputs.add(ops.constant(3));
       Operand<TInt32> x = ops.math.addN(inputs);
-      try (Tensor<TInt32> result = sess.runner().fetch(x).run().get(0).expect(TInt32.DTYPE)) {
-        assertEquals(6, result.data().getInt());
+      try (TInt32 result = sess.runner().fetch(x).run().get(0).expect(TInt32.DTYPE)) {
+        assertEquals(6, result.getInt());
       }
     }
   }
@@ -77,8 +77,8 @@ public void testControlDependencies() {
       Operand<TInt32> x =
           ops.withControlDependencies(controls).math.add(variable, ops.constant(0));
       sess.runner().addTarget(initVariable).run();
-      try (Tensor<TInt32> result = sess.runner().fetch(x).run().get(0).expect(TInt32.DTYPE)) {
-        assertEquals(3, result.data().getInt());
+      try (TInt32 result = sess.runner().fetch(x).run().get(0).expect(TInt32.DTYPE)) {
+        assertEquals(3, result.getInt());
       }
     }
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/op/core/GradientsTest.java b/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/op/core/GradientsTest.java
index fe1503d415f..3303904f9e6 100644
--- a/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/op/core/GradientsTest.java
+++ b/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/op/core/GradientsTest.java
@@ -47,13 +47,13 @@ public void createGradients() {
       assertNotNull(grads.dy());
       assertEquals(2, grads.dy().size());
 
-      try (Tensor<TFloat32> c = TFloat32.scalarOf(3.0f);
+      try (TFloat32 c = TFloat32.scalarOf(3.0f);
           AutoCloseableList<Tensor<?>> outputs =
               new AutoCloseableList<>(
                   sess.runner().feed(x, c).fetch(grads.dy(0)).fetch(grads.dy(1)).run())) {
 
-        assertEquals(108.0f, outputs.get(0).expect(TFloat32.DTYPE).data().getFloat(), 0.0f);
-        assertEquals(18.0f, outputs.get(1).expect(TFloat32.DTYPE).data().getFloat(), 0.0f);
+        assertEquals(108.0f, outputs.get(0).expect(TFloat32.DTYPE).getFloat(), 0.0f);
+        assertEquals(18.0f, outputs.get(1).expect(TFloat32.DTYPE).getFloat(), 0.0f);
       }
     }
   }
@@ -74,11 +74,11 @@ public void createGradientsWithSum() {
       assertNotNull(grads.dy());
       assertEquals(1, grads.dy().size());
 
-      try (Tensor<TFloat32> c = TFloat32.scalarOf(3.0f);
+      try (TFloat32 c = TFloat32.scalarOf(3.0f);
           AutoCloseableList<Tensor<?>> outputs =
               new AutoCloseableList<>(sess.runner().feed(x, c).fetch(grads.dy(0)).run())) {
 
-        assertEquals(114.0f, outputs.get(0).expect(TFloat32.DTYPE).data().getFloat(), 0.0f);
+        assertEquals(114.0f, outputs.get(0).expect(TFloat32.DTYPE).getFloat(), 0.0f);
       }
     }
   }
@@ -100,12 +100,12 @@ public void createGradientsWithInitialValues() {
       assertNotNull(grads1.dy());
       assertEquals(1, grads1.dy().size());
 
-      try (Tensor<TFloat32> c = TFloat32.scalarOf(3.0f);
+      try (TFloat32 c = TFloat32.scalarOf(3.0f);
           AutoCloseableList<Tensor<?>> outputs =
               new AutoCloseableList<>(
                   sess.runner().feed(x, c).fetch(grads1.dy(0)).run())) {
 
-        assertEquals(108.0f, outputs.get(0).expect(TFloat32.DTYPE).data().getFloat(), 0.0f);
+        assertEquals(108.0f, outputs.get(0).expect(TFloat32.DTYPE).getFloat(), 0.0f);
       }
     }
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/op/core/ShapesTest.java b/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/op/core/ShapesTest.java
index d5eb7412ea3..583b3d14a0f 100644
--- a/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/op/core/ShapesTest.java
+++ b/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/op/core/ShapesTest.java
@@ -47,15 +47,15 @@ public void testFlatten_Operand() {
       Shape<TInt64> tfshape = Shape.create(scope, actual, TInt64.DTYPE);
 
       AtomicInteger index = new AtomicInteger();
-      try (Tensor<TInt64> result1 =
+      try (TInt64 result1 =
               session.runner().fetch(tfshape.asOutput()).run().get(0).expect(TInt64.DTYPE);
-          Tensor<TInt64> result2 =
+          TInt64 result2 =
               session.runner().fetch(expResult.asOutput()).run().get(0).expect(TInt64.DTYPE)) {
         result1
-            .data()
+
             .scalars()
             .forEach(
-                s -> assertEquals(result2.data().getLong(index.getAndIncrement()), s.getLong()));
+                s -> assertEquals(result2.getLong(index.getAndIncrement()), s.getLong()));
       }
     }
   }
@@ -75,12 +75,12 @@ public void testFlatten_Shape() {
       AtomicInteger index = new AtomicInteger();
       flattened
           .asOutput()
-          .data()
+          .asTensor()
           .scalars()
           .forEach(
               s ->
                   assertEquals(
-                      expShape.asOutput().data().getLong(index.getAndIncrement()), s.getLong()));
+                      expShape.asTensor().getLong(index.getAndIncrement()), s.getLong()));
     }
   }
 
@@ -97,9 +97,9 @@ public void testSize_Shape() {
       Operand<TInt64> size = Shapes.size(scope, tfshape, TInt64.DTYPE);
 
       AtomicInteger index = new AtomicInteger();
-      try (Tensor<TInt64> result1 =
+      try (TInt64 result1 =
           session.runner().fetch(size.asOutput()).run().get(0).expect(TInt64.DTYPE)) {
-        result1.data().scalars().forEach(s -> assertEquals(8, s.getLong()));
+        result1.scalars().forEach(s -> assertEquals(8, s.getLong()));
       }
     }
   }
@@ -116,21 +116,21 @@ public void testSize_Shape_Operand() {
       Shape<TInt32> tfshape = Shape.create(scope, actual);
 
       Operand<TInt32> size = Shapes.size(scope, tfshape, Constant.scalarOf(scope, 0));
-      try (Tensor<TInt32> result =
+      try (TInt32 result =
           session.runner().fetch(size.asOutput()).run().get(0).expect(TInt32.DTYPE)) {
-        result.data().scalars().forEach(s -> assertEquals(4, s.getInt()));
+        result.scalars().forEach(s -> assertEquals(4, s.getInt()));
       }
 
       size = Shapes.size(scope, tfshape, Constant.scalarOf(scope, 1));
-      try (Tensor<TInt32> result =
+      try (TInt32 result =
           session.runner().fetch(size.asOutput()).run().get(0).expect(TInt32.DTYPE)) {
-        result.data().scalars().forEach(s -> assertEquals(2, s.getInt()));
+        result.scalars().forEach(s -> assertEquals(2, s.getInt()));
       }
 
       size = Shapes.size(scope, tfshape, Constant.scalarOf(scope, 2));
-      try (Tensor<TInt32> result =
+      try (TInt32 result =
           session.runner().fetch(size.asOutput()).run().get(0).expect(TInt32.DTYPE)) {
-        result.data().scalars().forEach(s -> assertEquals(1, s.getInt()));
+        result.scalars().forEach(s -> assertEquals(1, s.getInt()));
       }
     }
   }
@@ -146,21 +146,21 @@ public void testSize_Operand_Operand() {
           Reshape.create(scope, operand, Constant.vectorOf(scope, new long[] {4, 2, 1}));
 
       Operand<TInt32> size = Shapes.size(scope, actual, Constant.scalarOf(scope, 0));
-      try (Tensor<TInt32> result =
+      try (TInt32 result =
           session.runner().fetch(size.asOutput()).run().get(0).expect(TInt32.DTYPE)) {
-        result.data().scalars().forEach(s -> assertEquals(4, s.getInt()));
+        result.scalars().forEach(s -> assertEquals(4, s.getInt()));
       }
 
       size = Shapes.size(scope, actual, Constant.scalarOf(scope, 1));
-      try (Tensor<TInt32> result =
+      try (TInt32 result =
           session.runner().fetch(size.asOutput()).run().get(0).expect(TInt32.DTYPE)) {
-        result.data().scalars().forEach(s -> assertEquals(2, s.getInt()));
+        result.scalars().forEach(s -> assertEquals(2, s.getInt()));
       }
 
       size = Shapes.size(scope, actual, Constant.scalarOf(scope, 2));
-      try (Tensor<TInt32> result =
+      try (TInt32 result =
           session.runner().fetch(size.asOutput()).run().get(0).expect(TInt32.DTYPE)) {
-        result.data().scalars().forEach(s -> assertEquals(1, s.getInt()));
+        result.scalars().forEach(s -> assertEquals(1, s.getInt()));
       }
     }
   }
@@ -177,9 +177,9 @@ public void testNumDimensions() {
       Shape<TInt32> tfshape = Shape.create(scope, actual);
 
       Operand<TInt32> nDims = Shapes.numDimensions(scope, tfshape);
-      try (Tensor<TInt32> result =
+      try (TInt32 result =
           session.runner().fetch(nDims.asOutput()).run().get(0).expect(TInt32.DTYPE)) {
-        result.data().scalars().forEach(s -> assertEquals(3, s.getInt()));
+        result.scalars().forEach(s -> assertEquals(3, s.getInt()));
       }
     }
   }
@@ -199,7 +199,7 @@ public void testReduceDims_Operand_Operand() {
       AtomicInteger index = new AtomicInteger();
       int[] expected = {8};
       reducedShape
-          .data()
+          .asTensor()
           .scalars()
           .forEach(
               s -> {
@@ -224,7 +224,7 @@ public void testReduceDims_Shape_Operand() {
       AtomicInteger index = new AtomicInteger();
       int[] expected1 = {8};
       reducedShape
-          .data()
+          .asTensor()
           .scalars()
           .forEach(
               s -> {
@@ -237,7 +237,7 @@ public void testReduceDims_Shape_Operand() {
       index.set(0);
       int[] expected2 = {2, 4};
       reducedShape
-          .data()
+          .asTensor()
           .scalars()
           .forEach(
               s -> {
@@ -250,7 +250,7 @@ public void testReduceDims_Shape_Operand() {
       index.set(0);
       int[] expected3 = {2, 2, 2};
       reducedShape
-          .data()
+          .asTensor()
           .scalars()
           .forEach(
               s -> {
@@ -274,10 +274,10 @@ public void testSqueeze() {
       Operand<TInt32> squeezed = Shapes.squeeze(scope, tfshape);
       AtomicInteger index = new AtomicInteger();
       int[] expected = {4, 2};
-      try (Tensor<TInt32> result =
+      try (TInt32 result =
           session.runner().fetch(squeezed.asOutput()).run().get(0).expect(TInt32.DTYPE)) {
         result
-            .data()
+
             .scalars()
             .forEach(
                 s -> {
@@ -301,10 +301,10 @@ public void testHead() {
       Operand<TInt32> head = Shapes.head(scope, tfshape);
       AtomicInteger index = new AtomicInteger();
       int[] expected = {4};
-      try (Tensor<TInt32> result =
+      try (TInt32 result =
           session.runner().fetch(head.asOutput()).run().get(0).expect(TInt32.DTYPE)) {
         result
-            .data()
+
             .scalars()
             .forEach(
                 s -> {
@@ -328,10 +328,10 @@ public void testTake() {
       Operand<TInt32> take = Shapes.take(scope, tfshape, Constant.scalarOf(scope, 2));
       AtomicInteger index = new AtomicInteger();
       int[] expected = {4, 1};
-      try (Tensor<TInt32> result =
+      try (TInt32 result =
           session.runner().fetch(take.asOutput()).run().get(0).expect(TInt32.DTYPE)) {
         result
-            .data()
+
             .scalars()
             .forEach(
                 s -> {
@@ -355,10 +355,10 @@ public void testTail() {
       Operand<TInt32> tail = Shapes.tail(scope, tfshape);
       AtomicInteger index = new AtomicInteger();
       int[] expected = {1};
-      try (Tensor<TInt32> result =
+      try (TInt32 result =
           session.runner().fetch(tail.asOutput()).run().get(0).expect(TInt32.DTYPE)) {
         result
-            .data()
+
             .scalars()
             .forEach(
                 s -> {
@@ -382,10 +382,10 @@ public void testTakeLast() {
       Operand<TInt32> takeLast = Shapes.takeLast(scope, tfshape, Constant.scalarOf(scope, 3));
       AtomicInteger index = new AtomicInteger();
       int[] expected = {1, 2, 1};
-      try (Tensor<TInt32> result =
+      try (TInt32 result =
           session.runner().fetch(takeLast.asOutput()).run().get(0).expect(TInt32.DTYPE)) {
         result
-            .data()
+
             .scalars()
             .forEach(
                 s -> {
@@ -408,10 +408,10 @@ public void testPrependInt() {
       Operand<TInt32> prepend = Shapes.prepend(scope, tfshape, 3);
       AtomicInteger index = new AtomicInteger();
       int[] expected = {3, 4, 2};
-      try (Tensor<TInt32> result =
+      try (TInt32 result =
           session.runner().fetch(prepend.asOutput()).run().get(0).expect(TInt32.DTYPE)) {
         result
-            .data()
+
             .scalars()
             .forEach(
                 s -> {
@@ -434,10 +434,10 @@ public void testPrependLong() {
       Operand<TInt64> prepend = Shapes.prepend(scope, tfshape, 1L);
       AtomicInteger index = new AtomicInteger();
       long[] expected = {1, 4, 2};
-      try (Tensor<TInt64> result =
+      try (TInt64 result =
           session.runner().fetch(prepend.asOutput()).run().get(0).expect(TInt64.DTYPE)) {
         result
-            .data()
+
             .scalars()
             .forEach(
                 s -> {
@@ -465,10 +465,10 @@ public void testPrependShapeTInt32() {
       Operand<TInt32> prepend = Shapes.prepend(scope, tfshape1, tfshape2);
       AtomicInteger index = new AtomicInteger();
       int[] expected = {2, 4, 4, 2};
-      try (Tensor<TInt32> result =
+      try (TInt32 result =
           session.runner().fetch(prepend.asOutput()).run().get(0).expect(TInt32.DTYPE)) {
         result
-            .data()
+
             .scalars()
             .forEach(
                 s -> {
@@ -496,10 +496,10 @@ public void testPrependShapeTInt64() {
       Operand<TInt64> prepend = Shapes.prepend(scope, tfshape1, tfshape2);
       AtomicInteger index = new AtomicInteger();
       long[] expected = {2, 4, 4, 2};
-      try (Tensor<TInt64> result =
+      try (TInt64 result =
           session.runner().fetch(prepend.asOutput()).run().get(0).expect(TInt64.DTYPE)) {
         result
-            .data()
+
             .scalars()
             .forEach(
                 s -> {
@@ -522,10 +522,10 @@ public void testAppendLong() {
       Operand<TInt64> append = Shapes.append(scope, tfshape, 2L);
       AtomicInteger index = new AtomicInteger();
       long[] expected = {4L, 2L, 2L};
-      try (Tensor<TInt64> result =
+      try (TInt64 result =
           session.runner().fetch(append.asOutput()).run().get(0).expect(TInt64.DTYPE)) {
         result
-            .data()
+
             .scalars()
             .forEach(
                 s -> {
@@ -548,10 +548,10 @@ public void testAppendInt() {
       Operand<TInt32> append = Shapes.append(scope, tfshape, 2);
       AtomicInteger index = new AtomicInteger();
       int[] expected = {4, 2, 2};
-      try (Tensor<TInt32> result =
+      try (TInt32 result =
           session.runner().fetch(append.asOutput()).run().get(0).expect(TInt32.DTYPE)) {
         result
-            .data()
+
             .scalars()
             .forEach(
                 s -> {
@@ -579,10 +579,10 @@ public void testAppendShapeTInt32() {
       Operand<TInt32> append = Shapes.append(scope, tfshape1, tfshape2);
       AtomicInteger index = new AtomicInteger();
       int[] expected = {4, 2, 2, 4};
-      try (Tensor<TInt32> result =
+      try (TInt32 result =
           session.runner().fetch(append.asOutput()).run().get(0).expect(TInt32.DTYPE)) {
         result
-            .data()
+
             .scalars()
             .forEach(
                 s -> {
@@ -610,10 +610,10 @@ public void testAppendShapeTInt64() {
       Operand<TInt64> append = Shapes.append(scope, tfshape1, tfshape2);
       AtomicInteger index = new AtomicInteger();
       long[] expected = {4, 2, 2, 4};
-      try (Tensor<TInt64> result =
+      try (TInt64 result =
           session.runner().fetch(append.asOutput()).run().get(0).expect(TInt64.DTYPE)) {
         result
-            .data()
+
             .scalars()
             .forEach(
                 s -> {
diff --git a/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/op/core/ZerosTest.java b/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/op/core/ZerosTest.java
index 9600f8b38fc..55dcef5ae3c 100644
--- a/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/op/core/ZerosTest.java
+++ b/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/op/core/ZerosTest.java
@@ -42,8 +42,8 @@ public void createIntZeros() {
       Scope scope = new Scope(g);
       long[] shape = {2, 2};
       Zeros<TInt32> op = Zeros.create(scope, Constant.vectorOf(scope, shape), TInt32.DTYPE);
-      try (Tensor<TInt32> result = sess.runner().fetch(op).run().get(0).expect(TInt32.DTYPE)) {
-        result.data().scalars().forEach(s -> assertEquals(0, s.getInt()));
+      try (TInt32 result = sess.runner().fetch(op).run().get(0).expect(TInt32.DTYPE)) {
+        result.scalars().forEach(s -> assertEquals(0, s.getInt()));
       }
     }
   }
@@ -55,8 +55,8 @@ public void createFloatZeros() {
       Scope scope = new Scope(g);
       long[] shape = {2, 2};
       Zeros<TFloat32> op = Zeros.create(scope, Constant.vectorOf(scope, shape), TFloat32.DTYPE);
-      try (Tensor<TFloat32> result = sess.runner().fetch(op.asOutput()).run().get(0).expect(TFloat32.DTYPE)) {
-        result.data().scalars().forEach(s -> assertEquals(0.0f, s.getFloat(), 0));
+      try (TFloat32 result = sess.runner().fetch(op.asOutput()).run().get(0).expect(TFloat32.DTYPE)) {
+        result.scalars().forEach(s -> assertEquals(0.0f, s.getFloat(), 0));
       }
     }
   }
@@ -68,8 +68,8 @@ public void createDoubleZeros() {
       Scope scope = new Scope(g);
       long[] shape = {2, 2};
       Zeros<TFloat64> op = Zeros.create(scope, Constant.vectorOf(scope, shape), TFloat64.DTYPE);
-      try (Tensor<TFloat64> result = sess.runner().fetch(op.asOutput()).run().get(0).expect(TFloat64.DTYPE)) {
-        result.data().scalars().forEach(s -> assertEquals(0.0f, s.getDouble(), 0));
+      try (TFloat64 result = sess.runner().fetch(op.asOutput()).run().get(0).expect(TFloat64.DTYPE)) {
+        result.scalars().forEach(s -> assertEquals(0.0f, s.getDouble(), 0));
       }
     }
   }
@@ -81,8 +81,8 @@ public void createLongZeros() {
       Scope scope = new Scope(g);
       long[] shape = {2, 2};
       Zeros<TInt64> op = Zeros.create(scope, Constant.vectorOf(scope, shape), TInt64.DTYPE);
-      try (Tensor<TInt64> result = sess.runner().fetch(op.asOutput()).run().get(0).expect(TInt64.DTYPE)) {
-        result.data().scalars().forEach(s -> assertEquals(0L, s.getLong()));
+      try (TInt64 result = sess.runner().fetch(op.asOutput()).run().get(0).expect(TInt64.DTYPE)) {
+        result.scalars().forEach(s -> assertEquals(0L, s.getLong()));
       }
     }
   }
@@ -94,8 +94,8 @@ public void createBooleanZeros() {
       Scope scope = new Scope(g);
       long[] shape = {2, 2};
       Zeros<TBool> op = Zeros.create(scope, Constant.vectorOf(scope, shape), TBool.DTYPE);
-      try (Tensor<TBool> result = sess.runner().fetch(op.asOutput()).run().get(0).expect(TBool.DTYPE)) {
-        result.data().scalars().forEach(s -> assertFalse(s.getBoolean()));
+      try (TBool result = sess.runner().fetch(op.asOutput()).run().get(0).expect(TBool.DTYPE)) {
+        result.scalars().forEach(s -> assertFalse(s.getBoolean()));
       }
    }
   }
@@ -107,8 +107,8 @@ public void createUint8Zeros() {
       Scope scope = new Scope(g);
       long[] shape = {2, 2};
       Zeros<TUint8> op = Zeros.create(scope, Constant.vectorOf(scope, shape), TUint8.DTYPE);
-      try (Tensor<TUint8> result = sess.runner().fetch(op.asOutput()).run().get(0).expect(TUint8.DTYPE)) {
-        result.data().scalars().forEach(s -> assertEquals(0, s.getByte()));
+      try (TUint8 result = sess.runner().fetch(op.asOutput()).run().get(0).expect(TUint8.DTYPE)) {
+        result.scalars().forEach(s -> assertEquals(0, s.getByte()));
       }
     }
   }
@@ -120,8 +120,8 @@ public void createStringZeros() {
       Scope scope = new Scope(g);
       long[] shape = {2, 2};
       Zeros<TString> op = Zeros.create(scope, Constant.vectorOf(scope, shape), TString.DTYPE);
-      try (Tensor<TString> result = sess.runner().fetch(op.asOutput()).run().get(0).expect(TString.DTYPE)) {
-        result.data().scalars().forEach(s -> assertTrue(s.getObject().isEmpty()));
+      try (TString result = sess.runner().fetch(op.asOutput()).run().get(0).expect(TString.DTYPE)) {
+        result.scalars().forEach(s -> assertTrue(s.getObject().isEmpty()));
       }
     }
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/types/NumericTypesTestBase.java b/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/types/NumericTypesTestBase.java
index 87b24b0da2a..de4dc09035f 100644
--- a/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/types/NumericTypesTestBase.java
+++ b/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/types/NumericTypesTestBase.java
@@ -27,36 +27,35 @@
 import org.tensorflow.op.math.Sub;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.ndarray.IntNdArray;
-import org.tensorflow.ndarray.NdArray;
 import org.tensorflow.ndarray.NdArrays;
 import org.tensorflow.ndarray.index.Indices;
-import org.tensorflow.types.family.TNumber;
+import org.tensorflow.tensor.IntTensor;
+import org.tensorflow.types.family.TType;
 
-abstract class NumericTypesTestBase<T extends TNumber & NdArray<U>, U> {
+abstract class NumericTypesTestBase<T extends Tensor<U> & TType, U> {
 
   @Test
   public void initializeTensorsWithZeros() {
     // Allocate a tensor of 32-bits integer of the shape (2, 3, 2)
-    Tensor<T> tensor = allocateTensor(Shape.of(2, 3, 2));
-    NdArray<U> tensorData = tensor.data();
+    T tensor = allocateTensor(Shape.of(2, 3, 2));
 
-    assertEquals(3, tensorData.rank());
-    assertEquals(12, tensorData.size());
+    assertEquals(3, tensor.rank());
+    assertEquals(12, tensor.size());
 
     try (EagerSession session = EagerSession.create()) {
       Ops tf = Ops.create(session);
 
       // Initialize tensor memory with zeros and take a snapshot
-      tensorData.scalars().forEach(scalar -> scalar.setObject(valueOf(0)));
+      tensor.scalars().forEach(scalar -> scalar.setObject(valueOf(0)));
       Constant<T> x = tf.constant(tensor);
 
       // Initialize the same tensor memory with ones and take a snapshot
-      tensorData.scalars().forEach(scalar -> scalar.setObject(valueOf(1)));
+      tensor.scalars().forEach(scalar -> scalar.setObject(valueOf(1)));
       Constant<T> y = tf.constant(tensor);
 
       // Subtract y from x and validate the result
       Sub<T> sub = tf.math.sub(x, y);
-      sub.data().scalars().forEach(scalar ->
+      sub.asTensor().scalars().forEach(scalar ->
           assertEquals(valueOf(-1), scalar.getObject())
       );
     }
@@ -67,22 +66,21 @@ public void genericTest() {
     IntNdArray heapData = NdArrays.vectorOf(0, 1, 2, 3);
 
     // Creates a 2x2 matrix
-    try (Tensor<TInt32> tensor = TInt32.tensorOf(Shape.of(2, 2))) {
-      IntNdArray tensorData = tensor.data();
+    try (IntTensor tensor = TInt32.tensorOf(Shape.of(2, 2))) {
 
       // Copy first 2 values of the vector to the first row of the matrix
-      tensorData.set(heapData.slice(Indices.range(0, 2)), 0);
+      tensor.set(heapData.slice(Indices.range(0, 2)), 0);
 
       // Copy values at an odd position in the vector as the second row of the matrix
-      tensorData.set(heapData.slice(Indices.odd()), 1);
+      tensor.set(heapData.slice(Indices.odd()), 1);
 
-      assertEquals(0, tensorData.getInt(0, 0));
-      assertEquals(1, tensorData.getInt(0, 1));
-      assertEquals(1, tensorData.getInt(1, 0));
-      assertEquals(3, tensorData.getInt(1, 1));
+      assertEquals(0, tensor.getInt(0, 0));
+      assertEquals(1, tensor.getInt(0, 1));
+      assertEquals(1, tensor.getInt(1, 0));
+      assertEquals(3, tensor.getInt(1, 1));
 
       // Read rows of the tensor in reverse order
-      IntNdArray reversedTensorData = tensorData.slice(Indices.all(), Indices.flip());
+      IntNdArray reversedTensorData = tensor.slice(Indices.all(), Indices.flip());
 
       assertEquals(1, reversedTensorData.getInt(0, 0));
       assertEquals(0, reversedTensorData.getInt(0, 1));
@@ -97,14 +95,14 @@ public void genericTest() {
         IntNdArray result = tf.math.pow(x, x).data();
 
         // Validate result by computing the same operation in Java
-        tensorData.scalars().forEachIndexed((coords, s) ->
+        tensor.scalars().forEachIndexed((coords, s) ->
           assertEquals(Math.pow(s.getInt(), s.getInt()), result.getInt(coords), 1e-7f)
         );
       }
     }
   }
 
-  abstract Tensor<T> allocateTensor(Shape shape);
+  abstract T allocateTensor(Shape shape);
 
   abstract U valueOf(Integer value);
 }
diff --git a/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/types/TBfloat16Test.java b/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/types/TBfloat16Test.java
index 8681e805e3d..1ef6a24fc8b 100644
--- a/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/types/TBfloat16Test.java
+++ b/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/types/TBfloat16Test.java
@@ -23,7 +23,7 @@
 public class TBfloat16Test extends NumericTypesTestBase<TBfloat16, Float> {
 
   @Override
-  Tensor<TBfloat16> allocateTensor(Shape shape) {
+  TBfloat16 allocateTensor(Shape shape) {
     return TBfloat16.tensorOf(shape);
   }
 
diff --git a/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/types/TFloat16Test.java b/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/types/TFloat16Test.java
index b72fe6fc01c..b0b2fa908f0 100644
--- a/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/types/TFloat16Test.java
+++ b/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/types/TFloat16Test.java
@@ -23,7 +23,7 @@
 public class TFloat16Test extends NumericTypesTestBase<TFloat16, Float> {
 
   @Override
-  Tensor<TFloat16> allocateTensor(Shape shape) {
+  TFloat16 allocateTensor(Shape shape) {
     return TFloat16.tensorOf(shape);
   }
 
diff --git a/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/types/TFloat32Test.java b/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/types/TFloat32Test.java
index c4b1f6023f3..21ef1586d12 100644
--- a/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/types/TFloat32Test.java
+++ b/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/types/TFloat32Test.java
@@ -23,7 +23,7 @@
 public class TFloat32Test extends NumericTypesTestBase<TFloat32, Float> {
 
   @Override
-  Tensor<TFloat32> allocateTensor(Shape shape) {
+  TFloat32 allocateTensor(Shape shape) {
     return TFloat32.tensorOf(shape);
   }
 
diff --git a/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/types/TFloat64Test.java b/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/types/TFloat64Test.java
index 0e9c8947d0f..cb9de0bea19 100644
--- a/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/types/TFloat64Test.java
+++ b/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/types/TFloat64Test.java
@@ -23,7 +23,7 @@
 public class TFloat64Test extends NumericTypesTestBase<TFloat64, Double> {
 
   @Override
-  Tensor<TFloat64> allocateTensor(Shape shape) {
+  TFloat64 allocateTensor(Shape shape) {
     return TFloat64.tensorOf(shape);
   }
 
diff --git a/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/types/TInt32Test.java b/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/types/TInt32Test.java
index c52394bf210..8388c2834b2 100644
--- a/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/types/TInt32Test.java
+++ b/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/types/TInt32Test.java
@@ -23,7 +23,7 @@
 public class TInt32Test extends NumericTypesTestBase<TInt32, Integer> {
 
   @Override
-  Tensor<TInt32> allocateTensor(Shape shape) {
+  TInt32 allocateTensor(Shape shape) {
     return TInt32.tensorOf(shape);
   }
 
diff --git a/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/types/TInt64Test.java b/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/types/TInt64Test.java
index 261ac546fc5..55e4c7ead87 100644
--- a/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/types/TInt64Test.java
+++ b/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/types/TInt64Test.java
@@ -23,7 +23,7 @@
 public class TInt64Test extends NumericTypesTestBase<TInt64, Long> {
 
   @Override
-  Tensor<TInt64> allocateTensor(Shape shape) {
+  TInt64 allocateTensor(Shape shape) {
     return TInt64.tensorOf(shape);
   }
 
diff --git a/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/types/TStringTest.java b/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/types/TStringTest.java
index a4700aa652f..6d2b293be8b 100644
--- a/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/types/TStringTest.java
+++ b/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/types/TStringTest.java
@@ -32,13 +32,10 @@ public class TStringTest {
 
   @Test
   public void createScalar() {
-    Tensor<TString> tensor = TString.scalarOf("Pretty vacant");
+    TString tensor = TString.scalarOf("Pretty vacant");
     assertNotNull(tensor);
-
-    TString data = tensor.data();
-    assertNotNull(data);
-    assertEquals(Shape.scalar(), data.shape());
-    assertEquals("Pretty vacant", data.getObject());
+    assertEquals(Shape.scalar(), tensor.shape());
+    assertEquals("Pretty vacant", tensor.getObject());
   }
 
     @Test
@@ -55,14 +52,11 @@ public void createrScalarLongerThan127() {
 
     @Test
   public void createVector() {
-    Tensor<TString> tensor = TString.vectorOf("Pretty", "vacant");
+    TString tensor = TString.vectorOf("Pretty", "vacant");
     assertNotNull(tensor);
-
-    TString data = tensor.data();
-    assertNotNull(data);
-    assertEquals(Shape.of(2), data.shape());
-    assertEquals("Pretty", data.getObject(0));
-    assertEquals("vacant", data.getObject(1));
+    assertEquals(Shape.of(2), tensor.shape());
+    assertEquals("Pretty", tensor.getObject(0));
+    assertEquals("vacant", tensor.getObject(1));
   }
 
   @Test
@@ -73,30 +67,27 @@ public void createCopy() {
         .setObject("New", 1, 0)
         .setObject("York", 1, 1);
 
-    Tensor<TString> tensor = TString.tensorOf(strings);
+    TString tensor = TString.tensorOf(strings);
     assertNotNull(tensor);
-
-    TString data = tensor.data();
-    assertNotNull(data);
     strings.scalars().forEachIndexed((idx, s) ->
-        assertEquals(s.getObject(), data.getObject(idx))
+        assertEquals(s.getObject(), tensor.getObject(idx))
     );
   }
 
   @Test
   public void defaultCharsetIsUtf8() {
-    Tensor<TString> tensor = TString.tensorOf(NdArrays.scalarOfObject(BABY_CHICK));
-    byte[] bytes = tensor.data().asBytes().getObject();
+    TString tensor = TString.tensorOf(NdArrays.scalarOfObject(BABY_CHICK));
+    byte[] bytes = tensor.asBytes().getObject();
     assertArrayEquals(new byte[] { (byte)0xF0, (byte)0x9F, (byte)0x90, (byte)0xA5 }, bytes);
     assertEquals(BABY_CHICK, tensor.data().getObject());
   }
 
   @Test
   public void usingDifferentCharset() {
-    Tensor<TString> tensor = TString.tensorOf(StandardCharsets.UTF_16LE, NdArrays.scalarOfObject(BABY_CHICK));
-    byte[] bytes = tensor.data().asBytes().getObject();
+    TString tensor = TString.tensorOf(StandardCharsets.UTF_16LE, NdArrays.scalarOfObject(BABY_CHICK));
+    byte[] bytes = tensor.asBytes().getObject();
     assertArrayEquals(new byte[] { (byte)0x3D, (byte)0xD8, (byte)0x25, (byte)0xDC }, bytes);
-    assertEquals(BABY_CHICK, tensor.data().using(StandardCharsets.UTF_16LE).getObject());
+    assertEquals(BABY_CHICK, tensor.using(StandardCharsets.UTF_16LE).getObject());
   }
 
   @Test
@@ -106,11 +97,11 @@ public void initializingTensorWithRawBytes() {
     for (int i = 0; i < strings.length; ++i) {
       bytes.setObject(strings[i].getBytes(), i);
     }
-    Tensor<TString> tensor = TString.tensorOfBytes(bytes);
+    TString tensor = TString.tensorOfBytes(bytes);
     assertNotNull(tensor);
     assertEquals(bytes.shape(), tensor.shape());
 
-    NdArray<byte[]> tensorBytes = tensor.data().asBytes();
+    NdArray<byte[]> tensorBytes = tensor.asBytes();
     for (int i = 0; i < strings.length; ++i) {
       assertArrayEquals(bytes.getObject(i), tensorBytes.getObject(i));
     }
diff --git a/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/types/TUint8Test.java b/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/types/TUint8Test.java
index cc83087e018..eb2ddd5bd38 100644
--- a/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/types/TUint8Test.java
+++ b/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/types/TUint8Test.java
@@ -23,7 +23,7 @@
 public class TUint8Test extends NumericTypesTestBase<TUint8, Byte> {
 
   @Override
-  Tensor<TUint8> allocateTensor(Shape shape) {
+  TUint8 allocateTensor(Shape shape) {
     return TUint8.tensorOf(shape);
   }
 

From f92adb3b8992d5106349f5ef26e0dd5602edddeb Mon Sep 17 00:00:00 2001
From: Karl Lessard <karl.lessard@gmail.com>
Date: Tue, 4 Aug 2020 10:01:18 -0400
Subject: [PATCH 04/11] Delegate NdArray calls on tensors

---
 .../java/org/tensorflow/ndarray/NdArrays.java |   8 +-
 .../src/bazel/op_generator/op_generator.cc    |  11 +-
 .../src/bazel/op_generator/op_specs.cc        |  25 +-
 .../org/tensorflow/op/BitwiseOps.java         |  13 +-
 .../org/tensorflow/op/DtypesOps.java          |  10 +-
 .../org/tensorflow/op/ImageOps.java           |  65 ++-
 .../annotations/org/tensorflow/op/IoOps.java  |  23 +-
 .../org/tensorflow/op/LinalgOps.java          |  98 ++--
 .../org/tensorflow/op/MathOps.java            | 233 +++++----
 .../annotations/org/tensorflow/op/NnOps.java  | 178 ++++---
 .../annotations/org/tensorflow/op/Ops.java    | 447 +++++++++---------
 .../org/tensorflow/op/QuantizationOps.java    |  23 +-
 .../org/tensorflow/op/RandomOps.java          |  57 ++-
 .../org/tensorflow/op/ShapeOps.java           |  48 +-
 .../org/tensorflow/op/SignalOps.java          |  32 +-
 .../org/tensorflow/op/SparseOps.java          | 100 ++--
 .../org/tensorflow/op/StringsOps.java         |  11 +-
 .../org/tensorflow/op/SummaryOps.java         |  13 +-
 .../org/tensorflow/op/TrainOps.java           | 127 +++--
 .../annotations/org/tensorflow/op/XlaOps.java |  43 +-
 .../tensorflow/op/audio/AudioSpectrogram.java |   4 +-
 .../org/tensorflow/op/audio/DecodeWav.java    |   2 +-
 .../org/tensorflow/op/audio/EncodeWav.java    |   6 +-
 .../java/org/tensorflow/op/audio/Mfcc.java    |   6 +-
 .../org/tensorflow/op/bitwise/BitwiseAnd.java |  11 +-
 .../org/tensorflow/op/bitwise/BitwiseOr.java  |  11 +-
 .../org/tensorflow/op/bitwise/BitwiseXor.java |  11 +-
 .../org/tensorflow/op/bitwise/Invert.java     |   9 +-
 .../org/tensorflow/op/bitwise/LeftShift.java  |  11 +-
 .../org/tensorflow/op/bitwise/RightShift.java |  11 +-
 .../op/cluster/KMC2ChainInitialization.java   |   6 +-
 .../cluster/KmeansPlusPlusInitialization.java |  10 +-
 .../tensorflow/op/collective/AllReduce.java   |   9 +-
 .../op/collective/BroadcastRecv.java          |   8 +-
 .../op/collective/BroadcastSend.java          |  10 +-
 .../gen/java/org/tensorflow/op/core/All.java  |   9 +-
 .../gen/java/org/tensorflow/op/core/Any.java  |   9 +-
 .../org/tensorflow/op/core/AssertThat.java    |   4 +-
 .../java/org/tensorflow/op/core/Assign.java   |  12 +-
 .../org/tensorflow/op/core/AssignAdd.java     |  12 +-
 .../op/core/AssignAddVariableOp.java          |   8 +-
 .../org/tensorflow/op/core/AssignSub.java     |  12 +-
 .../op/core/AssignSubVariableOp.java          |   8 +-
 .../tensorflow/op/core/AssignVariableOp.java  |   8 +-
 .../java/org/tensorflow/op/core/Barrier.java  |   2 +-
 .../org/tensorflow/op/core/BarrierClose.java  |   2 +-
 .../op/core/BarrierIncompleteSize.java        |   4 +-
 .../tensorflow/op/core/BarrierInsertMany.java |  10 +-
 .../tensorflow/op/core/BarrierReadySize.java  |   4 +-
 .../tensorflow/op/core/BarrierTakeMany.java   |   4 +-
 .../java/org/tensorflow/op/core/Batch.java    |   2 +-
 .../org/tensorflow/op/core/BatchToSpace.java  |  12 +-
 .../tensorflow/op/core/BatchToSpaceNd.java    |  14 +-
 .../java/org/tensorflow/op/core/Bitcast.java  |  10 +-
 .../op/core/BroadcastDynamicShape.java        |  11 +-
 .../op/core/BroadcastGradientArgs.java        |   9 +-
 .../org/tensorflow/op/core/BroadcastTo.java   |  12 +-
 .../org/tensorflow/op/core/Bucketize.java     |   7 +-
 .../org/tensorflow/op/core/ClipByValue.java   |  14 +-
 .../tensorflow/op/core/CollectiveGather.java  |   9 +-
 .../java/org/tensorflow/op/core/Concat.java   |  12 +-
 .../tensorflow/op/core/ConsumeMutexLock.java  |   2 +-
 .../gen/java/org/tensorflow/op/core/Copy.java |  10 +-
 .../java/org/tensorflow/op/core/CopyHost.java |  10 +-
 .../org/tensorflow/op/core/CountUpTo.java     |   9 +-
 .../org/tensorflow/op/core/DecodeProto.java   |   2 +-
 .../java/org/tensorflow/op/core/DeepCopy.java |  10 +-
 .../op/core/DeleteSessionTensor.java          |   2 +-
 .../tensorflow/op/core/DestroyResourceOp.java |   2 +-
 .../op/core/DestroyTemporaryVariable.java     |  10 +-
 .../tensorflow/op/core/DummyMemoryCache.java  |   8 +-
 .../tensorflow/op/core/DynamicPartition.java  |  10 +-
 .../org/tensorflow/op/core/DynamicStitch.java |  12 +-
 .../org/tensorflow/op/core/EditDistance.java  |  18 +-
 .../java/org/tensorflow/op/core/Empty.java    |  10 +-
 .../tensorflow/op/core/EmptyTensorList.java   |  14 +-
 .../org/tensorflow/op/core/EncodeProto.java   |   6 +-
 .../org/tensorflow/op/core/EnsureShape.java   |  10 +-
 .../java/org/tensorflow/op/core/Enter.java    |  10 +-
 .../gen/java/org/tensorflow/op/core/Exit.java |  10 +-
 .../org/tensorflow/op/core/ExpandDims.java    |  12 +-
 .../op/core/ExtractVolumePatches.java         |   9 +-
 .../gen/java/org/tensorflow/op/core/Fill.java |  12 +-
 .../org/tensorflow/op/core/Fingerprint.java   |  10 +-
 .../java/org/tensorflow/op/core/Gather.java   |  14 +-
 .../java/org/tensorflow/op/core/GatherNd.java |  12 +-
 .../tensorflow/op/core/GetSessionHandle.java  |  12 +-
 .../tensorflow/op/core/GetSessionTensor.java  |  10 +-
 .../tensorflow/op/core/GuaranteeConst.java    |  10 +-
 .../org/tensorflow/op/core/HashTable.java     |  10 +-
 .../op/core/HistogramFixedWidth.java          |  15 +-
 .../java/org/tensorflow/op/core/Identity.java |  10 +-
 .../org/tensorflow/op/core/IdentityN.java     |   8 +-
 .../tensorflow/op/core/ImmutableConst.java    |   8 +-
 .../tensorflow/op/core/InitializeTable.java   |  10 +-
 .../op/core/InitializeTableFromTextFile.java  |   4 +-
 .../org/tensorflow/op/core/InplaceAdd.java    |  14 +-
 .../org/tensorflow/op/core/InplaceSub.java    |  14 +-
 .../org/tensorflow/op/core/InplaceUpdate.java |  14 +-
 .../op/core/IsVariableInitialized.java        |   8 +-
 .../java/org/tensorflow/op/core/LinSpace.java |  13 +-
 .../tensorflow/op/core/LookupTableExport.java |   8 +-
 .../tensorflow/op/core/LookupTableFind.java   |  14 +-
 .../tensorflow/op/core/LookupTableImport.java |  10 +-
 .../tensorflow/op/core/LookupTableInsert.java |  10 +-
 .../tensorflow/op/core/LookupTableRemove.java |   8 +-
 .../tensorflow/op/core/LookupTableSize.java   |   4 +-
 .../java/org/tensorflow/op/core/LoopCond.java |   4 +-
 .../org/tensorflow/op/core/LowerBound.java    |  14 +-
 .../tensorflow/op/core/MapIncompleteSize.java |   2 +-
 .../java/org/tensorflow/op/core/MapPeek.java  |  10 +-
 .../java/org/tensorflow/op/core/MapSize.java  |   2 +-
 .../java/org/tensorflow/op/core/MapStage.java |   6 +-
 .../org/tensorflow/op/core/MapUnstage.java    |  10 +-
 .../tensorflow/op/core/MapUnstageNoKey.java   |   2 +-
 .../gen/java/org/tensorflow/op/core/Max.java  |  12 +-
 .../java/org/tensorflow/op/core/Merge.java    |   8 +-
 .../gen/java/org/tensorflow/op/core/Min.java  |  12 +-
 .../org/tensorflow/op/core/MirrorPad.java     |  12 +-
 .../org/tensorflow/op/core/MirrorPadGrad.java |  12 +-
 .../tensorflow/op/core/MlirPassthroughOp.java |   8 +-
 .../op/core/MutableDenseHashTable.java        |  14 +-
 .../tensorflow/op/core/MutableHashTable.java  |  10 +-
 .../op/core/MutableHashTableOfTensors.java    |  10 +-
 .../java/org/tensorflow/op/core/Mutex.java    |   8 +-
 .../org/tensorflow/op/core/MutexLock.java     |  10 +-
 .../org/tensorflow/op/core/NcclAllReduce.java |   9 +-
 .../org/tensorflow/op/core/NcclBroadcast.java |   9 +-
 .../org/tensorflow/op/core/NcclReduce.java    |   9 +-
 .../org/tensorflow/op/core/NextIteration.java |  10 +-
 .../java/org/tensorflow/op/core/OneHot.java   |  16 +-
 .../java/org/tensorflow/op/core/OnesLike.java |  10 +-
 .../op/core/OrderedMapIncompleteSize.java     |   2 +-
 .../tensorflow/op/core/OrderedMapPeek.java    |  10 +-
 .../tensorflow/op/core/OrderedMapSize.java    |   2 +-
 .../tensorflow/op/core/OrderedMapStage.java   |   6 +-
 .../tensorflow/op/core/OrderedMapUnstage.java |  10 +-
 .../op/core/OrderedMapUnstageNoKey.java       |   2 +-
 .../gen/java/org/tensorflow/op/core/Pad.java  |  14 +-
 .../tensorflow/op/core/ParallelConcat.java    |  10 +-
 .../op/core/ParallelDynamicStitch.java        |  12 +-
 .../org/tensorflow/op/core/Placeholder.java   |   8 +-
 .../op/core/PlaceholderWithDefault.java       |  10 +-
 .../java/org/tensorflow/op/core/Print.java    |   2 +-
 .../gen/java/org/tensorflow/op/core/Prod.java |  12 +-
 .../tensorflow/op/core/QuantizedReshape.java  |  14 +-
 .../java/org/tensorflow/op/core/Range.java    |  13 +-
 .../gen/java/org/tensorflow/op/core/Rank.java |   8 +-
 .../tensorflow/op/core/ReadVariableOp.java    |  10 +-
 .../gen/java/org/tensorflow/op/core/Recv.java |   8 +-
 .../org/tensorflow/op/core/ReduceAll.java     |   9 +-
 .../org/tensorflow/op/core/ReduceAny.java     |   9 +-
 .../org/tensorflow/op/core/ReduceMax.java     |  12 +-
 .../org/tensorflow/op/core/ReduceMin.java     |  12 +-
 .../org/tensorflow/op/core/ReduceProd.java    |  12 +-
 .../org/tensorflow/op/core/ReduceSum.java     |  12 +-
 .../java/org/tensorflow/op/core/RefEnter.java |  10 +-
 .../java/org/tensorflow/op/core/RefExit.java  |  10 +-
 .../org/tensorflow/op/core/RefIdentity.java   |  10 +-
 .../java/org/tensorflow/op/core/RefMerge.java |   8 +-
 .../tensorflow/op/core/RefNextIteration.java  |  10 +-
 .../org/tensorflow/op/core/RefSelect.java     |  12 +-
 .../org/tensorflow/op/core/RefSwitch.java     |  10 +-
 .../op/core/RemoteFusedGraphExecute.java      |   8 +-
 .../java/org/tensorflow/op/core/Reshape.java  |  12 +-
 .../tensorflow/op/core/ResourceCountUpTo.java |   9 +-
 .../tensorflow/op/core/ResourceGather.java    |  12 +-
 .../tensorflow/op/core/ResourceGatherNd.java  |  12 +-
 .../op/core/ResourceScatterAdd.java           |  10 +-
 .../op/core/ResourceScatterDiv.java           |  10 +-
 .../op/core/ResourceScatterMax.java           |  10 +-
 .../op/core/ResourceScatterMin.java           |  10 +-
 .../op/core/ResourceScatterMul.java           |  10 +-
 .../op/core/ResourceScatterNdAdd.java         |  10 +-
 .../op/core/ResourceScatterNdSub.java         |  10 +-
 .../op/core/ResourceScatterNdUpdate.java      |  10 +-
 .../op/core/ResourceScatterSub.java           |  10 +-
 .../op/core/ResourceScatterUpdate.java        |  10 +-
 .../op/core/ResourceStridedSliceAssign.java   |  14 +-
 .../java/org/tensorflow/op/core/Reverse.java  |  12 +-
 .../tensorflow/op/core/ReverseSequence.java   |  12 +-
 .../gen/java/org/tensorflow/op/core/Roll.java |  14 +-
 .../gen/java/org/tensorflow/op/core/Rpc.java  |   8 +-
 .../org/tensorflow/op/core/ScatterAdd.java    |  14 +-
 .../org/tensorflow/op/core/ScatterDiv.java    |  14 +-
 .../org/tensorflow/op/core/ScatterMax.java    |  13 +-
 .../org/tensorflow/op/core/ScatterMin.java    |  13 +-
 .../org/tensorflow/op/core/ScatterMul.java    |  14 +-
 .../org/tensorflow/op/core/ScatterNd.java     |  14 +-
 .../org/tensorflow/op/core/ScatterNdAdd.java  |  14 +-
 .../op/core/ScatterNdNonAliasingAdd.java      |  14 +-
 .../org/tensorflow/op/core/ScatterNdSub.java  |  14 +-
 .../tensorflow/op/core/ScatterNdUpdate.java   |  14 +-
 .../org/tensorflow/op/core/ScatterSub.java    |  14 +-
 .../org/tensorflow/op/core/ScatterUpdate.java |  14 +-
 .../java/org/tensorflow/op/core/Select.java   |  14 +-
 .../gen/java/org/tensorflow/op/core/Send.java |   6 +-
 .../org/tensorflow/op/core/SetDiff1d.java     |  12 +-
 .../java/org/tensorflow/op/core/SetSize.java  |  12 +-
 .../java/org/tensorflow/op/core/Shape.java    |  12 +-
 .../java/org/tensorflow/op/core/ShapeN.java   |  10 +-
 .../gen/java/org/tensorflow/op/core/Size.java |  12 +-
 .../java/org/tensorflow/op/core/Slice.java    |  14 +-
 .../java/org/tensorflow/op/core/Snapshot.java |  10 +-
 .../tensorflow/op/core/SpaceToBatchNd.java    |  14 +-
 .../java/org/tensorflow/op/core/Split.java    |  10 +-
 .../java/org/tensorflow/op/core/SplitV.java   |  12 +-
 .../java/org/tensorflow/op/core/Squeeze.java  |  10 +-
 .../java/org/tensorflow/op/core/Stack.java    |  10 +-
 .../java/org/tensorflow/op/core/Stage.java    |   2 +-
 .../org/tensorflow/op/core/StagePeek.java     |   8 +-
 .../org/tensorflow/op/core/StageSize.java     |   2 +-
 .../org/tensorflow/op/core/StopGradient.java  |  10 +-
 .../org/tensorflow/op/core/StridedSlice.java  |  16 +-
 .../op/core/StridedSliceAssign.java           |  18 +-
 .../tensorflow/op/core/StridedSliceGrad.java  |  18 +-
 .../gen/java/org/tensorflow/op/core/Sum.java  |  12 +-
 .../org/tensorflow/op/core/SwitchCond.java    |  10 +-
 .../tensorflow/op/core/TemporaryVariable.java |   8 +-
 .../org/tensorflow/op/core/TensorArray.java   |   6 +-
 .../tensorflow/op/core/TensorArrayClose.java  |   2 +-
 .../tensorflow/op/core/TensorArrayConcat.java |  10 +-
 .../tensorflow/op/core/TensorArrayGather.java |  14 +-
 .../tensorflow/op/core/TensorArrayGrad.java   |   4 +-
 .../op/core/TensorArrayGradWithShape.java     |   6 +-
 .../tensorflow/op/core/TensorArrayPack.java   |  12 +-
 .../tensorflow/op/core/TensorArrayRead.java   |  14 +-
 .../op/core/TensorArrayScatter.java           |  14 +-
 .../tensorflow/op/core/TensorArraySize.java   |   6 +-
 .../tensorflow/op/core/TensorArraySplit.java  |  14 +-
 .../tensorflow/op/core/TensorArrayUnpack.java |  12 +-
 .../tensorflow/op/core/TensorArrayWrite.java  |  14 +-
 .../core/TensorForestCreateTreeVariable.java  |   4 +-
 .../op/core/TensorForestTreeDeserialize.java  |   4 +-
 .../core/TensorForestTreeIsInitializedOp.java |   4 +-
 .../op/core/TensorForestTreePredict.java      |   6 +-
 .../TensorForestTreeResourceHandleOp.java     |   8 +-
 .../op/core/TensorForestTreeSerialize.java    |   4 +-
 .../op/core/TensorForestTreeSize.java         |   4 +-
 .../tensorflow/op/core/TensorListConcat.java  |  12 +-
 .../op/core/TensorListConcatLists.java        |  14 +-
 .../op/core/TensorListElementShape.java       |   9 +-
 .../op/core/TensorListFromTensor.java         |  14 +-
 .../tensorflow/op/core/TensorListGather.java  |  14 +-
 .../tensorflow/op/core/TensorListGetItem.java |  14 +-
 .../tensorflow/op/core/TensorListLength.java  |   4 +-
 .../tensorflow/op/core/TensorListPopBack.java |  10 +-
 .../op/core/TensorListPushBack.java           |  14 +-
 .../op/core/TensorListPushBackBatch.java      |  14 +-
 .../tensorflow/op/core/TensorListReserve.java |  14 +-
 .../tensorflow/op/core/TensorListResize.java  |  12 +-
 .../tensorflow/op/core/TensorListScatter.java |  18 +-
 .../TensorListScatterIntoExistingList.java    |  16 +-
 .../tensorflow/op/core/TensorListSetItem.java |  16 +-
 .../tensorflow/op/core/TensorListSplit.java   |  16 +-
 .../tensorflow/op/core/TensorListStack.java   |  12 +-
 .../op/core/TensorScatterNdAdd.java           |  14 +-
 .../op/core/TensorScatterNdSub.java           |  14 +-
 .../op/core/TensorScatterNdUpdate.java        |  14 +-
 .../op/core/TensorStridedSliceUpdate.java     |  18 +-
 .../gen/java/org/tensorflow/op/core/Tile.java |  12 +-
 .../org/tensorflow/op/core/Timestamp.java     |   2 +-
 .../java/org/tensorflow/op/core/TryRpc.java   |   6 +-
 .../java/org/tensorflow/op/core/Unbatch.java  |  14 +-
 .../org/tensorflow/op/core/UnbatchGrad.java   |  16 +-
 .../java/org/tensorflow/op/core/Unique.java   |  12 +-
 .../tensorflow/op/core/UniqueWithCounts.java  |  12 +-
 .../org/tensorflow/op/core/UnravelIndex.java  |  11 +-
 .../java/org/tensorflow/op/core/Unstack.java  |   8 +-
 .../java/org/tensorflow/op/core/Unstage.java  |   6 +-
 .../org/tensorflow/op/core/UpperBound.java    |  14 +-
 .../org/tensorflow/op/core/VarHandleOp.java   |  10 +-
 .../op/core/VarIsInitializedOp.java           |   4 +-
 .../java/org/tensorflow/op/core/Variable.java |   8 +-
 .../org/tensorflow/op/core/VariableShape.java |   9 +-
 .../java/org/tensorflow/op/core/Where.java    |   8 +-
 .../org/tensorflow/op/core/ZerosLike.java     |  10 +-
 .../tensorflow/op/data/AssertNextDataset.java |  12 +-
 .../tensorflow/op/data/AutoShardDataset.java  |  14 +-
 .../org/tensorflow/op/data/BatchDataset.java  |  14 +-
 .../op/data/BytesProducedStatsDataset.java    |  12 +-
 .../org/tensorflow/op/data/CSVDataset.java    |  26 +-
 .../org/tensorflow/op/data/CacheDataset.java  |  12 +-
 .../tensorflow/op/data/CacheDatasetV2.java    |  14 +-
 .../op/data/ChooseFastestDataset.java         |  10 +-
 .../op/data/ConcatenateDataset.java           |  12 +-
 .../op/data/DatasetCardinality.java           |   4 +-
 .../tensorflow/op/data/DatasetFromGraph.java  |  10 +-
 .../tensorflow/op/data/DatasetToGraph.java    |   4 +-
 .../op/data/DatasetToSingleElement.java       |   8 +-
 .../tensorflow/op/data/DatasetToTfRecord.java |   6 +-
 .../tensorflow/op/data/DeleteIterator.java    |   4 +-
 .../tensorflow/op/data/DeleteMemoryCache.java |   4 +-
 .../op/data/DeleteMultiDeviceIterator.java    |   6 +-
 .../op/data/DenseToSparseBatchDataset.java    |  14 +-
 .../op/data/DeserializeIterator.java          |   4 +-
 .../op/data/DirectedInterleaveDataset.java    |  12 +-
 .../op/data/FilterByLastComponentDataset.java |  10 +-
 .../op/data/FixedLengthRecordDataset.java     |  20 +-
 .../op/data/IgnoreErrorsDataset.java          |  10 +-
 .../java/org/tensorflow/op/data/Iterator.java |   8 +-
 .../op/data/IteratorFromStringHandle.java     |  10 +-
 .../tensorflow/op/data/IteratorGetDevice.java |   4 +-
 .../tensorflow/op/data/IteratorGetNext.java   |   8 +-
 .../op/data/IteratorGetNextAsOptional.java    |  10 +-
 .../op/data/IteratorGetNextSync.java          |   8 +-
 .../op/data/IteratorToStringHandle.java       |   4 +-
 .../org/tensorflow/op/data/LMDBDataset.java   |  10 +-
 .../op/data/LatencyStatsDataset.java          |  12 +-
 .../org/tensorflow/op/data/LeakyReluGrad.java |  11 +-
 .../org/tensorflow/op/data/MakeIterator.java  |   4 +-
 .../op/data/MatchingFilesDataset.java         |  10 +-
 .../op/data/MaxIntraOpParallelismDataset.java |  12 +-
 .../org/tensorflow/op/data/ModelDataset.java  |  10 +-
 .../op/data/MultiDeviceIterator.java          |   8 +-
 .../MultiDeviceIteratorFromStringHandle.java  |  10 +-
 .../MultiDeviceIteratorGetNextFromShard.java  |  12 +-
 .../op/data/MultiDeviceIteratorInit.java      |   8 +-
 .../MultiDeviceIteratorToStringHandle.java    |   4 +-
 .../op/data/NonSerializableDataset.java       |  10 +-
 .../tensorflow/op/data/OptimizeDataset.java   |  12 +-
 .../tensorflow/op/data/OptionalFromValue.java |  10 +-
 .../tensorflow/op/data/OptionalGetValue.java  |   8 +-
 .../tensorflow/op/data/OptionalHasValue.java  |   4 +-
 .../org/tensorflow/op/data/OptionalNone.java  |   8 +-
 .../op/data/PaddedBatchDataset.java           |  18 +-
 .../tensorflow/op/data/PrefetchDataset.java   |  12 +-
 .../op/data/PrivateThreadPoolDataset.java     |  12 +-
 .../org/tensorflow/op/data/RandomDataset.java |  12 +-
 .../org/tensorflow/op/data/RangeDataset.java  |  14 +-
 .../tensorflow/op/data/RebatchDataset.java    |  12 +-
 .../org/tensorflow/op/data/RepeatDataset.java |  12 +-
 .../tensorflow/op/data/SamplingDataset.java   |  16 +-
 .../tensorflow/op/data/SerializeIterator.java |  10 +-
 .../op/data/SetStatsAggregatorDataset.java    |  16 +-
 .../org/tensorflow/op/data/ShardDataset.java  |  14 +-
 .../op/data/ShuffleAndRepeatDataset.java      |  77 +--
 .../tensorflow/op/data/ShuffleDataset.java    |  59 +--
 .../org/tensorflow/op/data/SkipDataset.java   |  12 +-
 .../org/tensorflow/op/data/SleepDataset.java  |  12 +-
 .../op/data/SlidingWindowDataset.java         |  16 +-
 .../tensorflow/op/data/SnapshotDataset.java   |  12 +-
 .../op/data/SparseTensorSliceDataset.java     |  16 +-
 .../org/tensorflow/op/data/SqlDataset.java    |  14 +-
 .../op/data/StatsAggregatorHandle.java        |   8 +-
 .../org/tensorflow/op/data/TakeDataset.java   |  12 +-
 .../org/tensorflow/op/data/TensorDataset.java |  10 +-
 .../op/data/TensorSliceDataset.java           |  10 +-
 .../tensorflow/op/data/TextLineDataset.java   |  14 +-
 .../tensorflow/op/data/TfRecordDataset.java   |  14 +-
 .../tensorflow/op/data/ThreadPoolDataset.java |  12 +-
 .../tensorflow/op/data/ThreadPoolHandle.java  |   8 +-
 .../tensorflow/op/data/UnbatchDataset.java    |  10 +-
 .../org/tensorflow/op/data/UniqueDataset.java |  10 +-
 .../op/data/UnwrapDatasetVariant.java         |  10 +-
 .../org/tensorflow/op/data/WindowDataset.java |  18 +-
 .../op/data/WrapDatasetVariant.java           |  10 +-
 .../org/tensorflow/op/data/ZipDataset.java    |  10 +-
 .../AssertCardinalityDataset.java             |  12 +-
 .../data/experimental/AssertNextDataset.java  |  12 +-
 .../data/experimental/AutoShardDataset.java   |  14 +-
 .../BytesProducedStatsDataset.java            |  12 +-
 .../op/data/experimental/CSVDataset.java      |  26 +-
 .../experimental/ChooseFastestDataset.java    |  10 +-
 .../data/experimental/DatasetCardinality.java |   4 +-
 .../data/experimental/DatasetToTFRecord.java  |   6 +-
 .../DenseToSparseBatchDataset.java            |  14 +-
 .../DirectedInterleaveDataset.java            |  12 +-
 .../experimental/IgnoreErrorsDataset.java     |  10 +-
 .../data/experimental/IteratorGetDevice.java  |   4 +-
 .../experimental/LatencyStatsDataset.java     |  12 +-
 .../op/data/experimental/LmdbDataset.java     |  10 +-
 .../experimental/MatchingFilesDataset.java    |  10 +-
 .../MaxIntraOpParallelismDataset.java         |  12 +-
 .../experimental/NonSerializableDataset.java  |  10 +-
 .../experimental/ParseExampleDataset.java     |  15 +-
 .../PrivateThreadPoolDataset.java             |  12 +-
 .../op/data/experimental/RandomDataset.java   |  12 +-
 .../op/data/experimental/RebatchDataset.java  |  12 +-
 .../SetStatsAggregatorDataset.java            |  16 +-
 .../op/data/experimental/SleepDataset.java    |  12 +-
 .../experimental/SlidingWindowDataset.java    |  16 +-
 .../op/data/experimental/SqlDataset.java      |  14 +-
 .../experimental/StatsAggregatorHandle.java   |   8 +-
 .../StatsAggregatorSetSummaryWriter.java      |   4 +-
 .../experimental/StatsAggregatorSummary.java  |   4 +-
 .../data/experimental/ThreadPoolDataset.java  |  12 +-
 .../data/experimental/ThreadPoolHandle.java   |   8 +-
 .../op/data/experimental/UnbatchDataset.java  |  10 +-
 .../op/data/experimental/UniqueDataset.java   |  10 +-
 .../op/debugging/CheckNumerics.java           |   9 +-
 .../op/debugging/DebugGradientIdentity.java   |  10 +-
 .../debugging/DebugGradientRefIdentity.java   |  10 +-
 .../op/debugging/DebugIdentity.java           |  10 +-
 .../op/debugging/DebugNanCount.java           |   8 +-
 .../op/debugging/DebugNumericsSummary.java    |  12 +-
 .../org/tensorflow/op/dtypes/AsString.java    |   8 +-
 .../java/org/tensorflow/op/dtypes/Cast.java   |  10 +-
 .../org/tensorflow/op/dtypes/Complex.java     |  12 +-
 .../java/org/tensorflow/op/dtypes/ToBool.java |   8 +-
 .../estimator/BoostedTreesAggregateStats.java |  10 +-
 .../op/estimator/BoostedTreesBucketize.java   |   4 +-
 ...BoostedTreesCalculateBestFeatureSplit.java |  12 +-
 ...ostedTreesCalculateBestFeatureSplitV2.java |  16 +-
 ...stedTreesCalculateBestGainsPerFeature.java |  12 +-
 .../op/estimator/BoostedTreesCenterBias.java  |  12 +-
 .../estimator/BoostedTreesCreateEnsemble.java |   6 +-
 ...stedTreesCreateQuantileStreamResource.java |   6 +-
 .../BoostedTreesDeserializeEnsemble.java      |   6 +-
 .../BoostedTreesEnsembleResourceHandleOp.java |   8 +-
 .../BoostedTreesExampleDebugOutputs.java      |   6 +-
 .../BoostedTreesFlushQuantileSummaries.java   |   2 +-
 .../BoostedTreesGetEnsembleStates.java        |   2 +-
 .../BoostedTreesMakeQuantileSummaries.java    |   6 +-
 .../BoostedTreesMakeStatsSummary.java         |  10 +-
 .../op/estimator/BoostedTreesPredict.java     |   6 +-
 ...eesQuantileStreamResourceAddSummaries.java |   4 +-
 ...reesQuantileStreamResourceDeserialize.java |   4 +-
 ...ostedTreesQuantileStreamResourceFlush.java |   4 +-
 ...tileStreamResourceGetBucketBoundaries.java |   2 +-
 ...edTreesQuantileStreamResourceHandleOp.java |   8 +-
 .../BoostedTreesSerializeEnsemble.java        |   2 +-
 .../BoostedTreesSparseAggregateStats.java     |  12 +-
 ...dTreesSparseCalculateBestFeatureSplit.java |  16 +-
 .../BoostedTreesTrainingPredict.java          |   8 +-
 .../estimator/BoostedTreesUpdateEnsemble.java |  18 +-
 .../BoostedTreesUpdateEnsembleV2.java         |  24 +-
 .../IsBoostedTreesEnsembleInitialized.java    |   4 +-
 ...reesQuantileStreamResourceInitialized.java |   4 +-
 .../tensorflow/op/image/AdjustContrast.java   |  11 +-
 .../org/tensorflow/op/image/AdjustHue.java    |  11 +-
 .../tensorflow/op/image/AdjustSaturation.java |  11 +-
 .../op/image/CombinedNonMaxSuppression.java   |  12 +-
 .../tensorflow/op/image/CropAndResize.java    |  13 +-
 .../op/image/CropAndResizeGradBoxes.java      |  13 +-
 .../op/image/CropAndResizeGradImage.java      |  15 +-
 .../op/image/DecodeAndCropJpeg.java           |   6 +-
 .../org/tensorflow/op/image/DecodeBmp.java    |   4 +-
 .../org/tensorflow/op/image/DecodeGif.java    |   4 +-
 .../org/tensorflow/op/image/DecodeJpeg.java   |   4 +-
 .../org/tensorflow/op/image/DecodePng.java    |   9 +-
 .../op/image/DrawBoundingBoxes.java           |  13 +-
 .../org/tensorflow/op/image/EncodeJpeg.java   |   4 +-
 .../op/image/EncodeJpegVariableQuality.java   |   6 +-
 .../org/tensorflow/op/image/EncodePng.java    |   7 +-
 .../tensorflow/op/image/ExtractGlimpse.java   |  14 +-
 .../op/image/ExtractImagePatches.java         |  10 +-
 .../tensorflow/op/image/ExtractJpegShape.java |   9 +-
 .../image/GenerateBoundingBoxProposals.java   |  14 +-
 .../org/tensorflow/op/image/HsvToRgb.java     |   9 +-
 .../op/image/ImageProjectiveTransformV2.java  |  13 +-
 .../tensorflow/op/image/NearestNeighbors.java |   6 +-
 .../op/image/NonMaxSuppression.java           |  17 +-
 .../image/NonMaxSuppressionWithOverlaps.java  |  12 +-
 .../op/image/QuantizedResizeBilinear.java     |  14 +-
 .../org/tensorflow/op/image/RandomCrop.java   |  11 +-
 .../org/tensorflow/op/image/ResizeArea.java   |   9 +-
 .../tensorflow/op/image/ResizeBicubic.java    |   9 +-
 .../op/image/ResizeBicubicGrad.java           |  11 +-
 .../tensorflow/op/image/ResizeBilinear.java   |   9 +-
 .../op/image/ResizeBilinearGrad.java          |  11 +-
 .../op/image/ResizeNearestNeighbor.java       |  11 +-
 .../op/image/ResizeNearestNeighborGrad.java   |  11 +-
 .../org/tensorflow/op/image/RgbToHsv.java     |   9 +-
 .../op/image/SampleDistortedBoundingBox.java  |  11 +-
 .../op/image/ScaleAndTranslate.java           |  13 +-
 .../op/image/ScaleAndTranslateGrad.java       |  15 +-
 .../org/tensorflow/op/io/DecodeBase64.java    |   4 +-
 .../tensorflow/op/io/DecodeCompressed.java    |   4 +-
 .../java/org/tensorflow/op/io/DecodeCsv.java  |  10 +-
 .../tensorflow/op/io/DecodeJsonExample.java   |   4 +-
 .../org/tensorflow/op/io/DecodePaddedRaw.java |  11 +-
 .../java/org/tensorflow/op/io/DecodeRaw.java  |  10 +-
 .../op/io/DeserializeManySparse.java          |   8 +-
 .../org/tensorflow/op/io/EncodeBase64.java    |   4 +-
 .../java/org/tensorflow/op/io/FifoQueue.java  |   8 +-
 .../op/io/FixedLengthRecordReader.java        |   8 +-
 .../org/tensorflow/op/io/IdentityReader.java  |   8 +-
 .../java/org/tensorflow/op/io/LmdbReader.java |   2 +-
 .../org/tensorflow/op/io/MatchingFiles.java   |   4 +-
 .../tensorflow/op/io/PaddingFifoQueue.java    |   8 +-
 .../org/tensorflow/op/io/ParseExample.java    |  13 +-
 .../op/io/ParseSequenceExample.java           |  21 +-
 .../tensorflow/op/io/ParseSingleExample.java  |   4 +-
 .../op/io/ParseSingleSequenceExample.java     |  16 +-
 .../org/tensorflow/op/io/ParseTensor.java     |  10 +-
 .../org/tensorflow/op/io/PriorityQueue.java   |   8 +-
 .../java/org/tensorflow/op/io/QueueClose.java |   2 +-
 .../org/tensorflow/op/io/QueueDequeue.java    |   8 +-
 .../tensorflow/op/io/QueueDequeueMany.java    |  10 +-
 .../tensorflow/op/io/QueueDequeueUpTo.java    |  10 +-
 .../org/tensorflow/op/io/QueueEnqueue.java    |   4 +-
 .../tensorflow/op/io/QueueEnqueueMany.java    |   4 +-
 .../org/tensorflow/op/io/QueueIsClosed.java   |   4 +-
 .../java/org/tensorflow/op/io/QueueSize.java  |   4 +-
 .../tensorflow/op/io/RandomShuffleQueue.java  |   8 +-
 .../java/org/tensorflow/op/io/ReadFile.java   |   4 +-
 .../op/io/ReaderNumRecordsProduced.java       |   4 +-
 .../op/io/ReaderNumWorkUnitsCompleted.java    |   4 +-
 .../java/org/tensorflow/op/io/ReaderRead.java |   4 +-
 .../org/tensorflow/op/io/ReaderReadUpTo.java  |   6 +-
 .../org/tensorflow/op/io/ReaderReset.java     |   2 +-
 .../tensorflow/op/io/ReaderRestoreState.java  |   4 +-
 .../op/io/ReaderSerializeState.java           |   4 +-
 .../tensorflow/op/io/SerializeManySparse.java |  16 +-
 .../org/tensorflow/op/io/SerializeSparse.java |  16 +-
 .../org/tensorflow/op/io/SerializeTensor.java |   8 +-
 .../org/tensorflow/op/io/ShardedFilename.java |   8 +-
 .../org/tensorflow/op/io/ShardedFilespec.java |   6 +-
 .../org/tensorflow/op/io/TextLineReader.java  |   8 +-
 .../org/tensorflow/op/io/TfRecordReader.java  |   8 +-
 .../org/tensorflow/op/io/WholeFileReader.java |   8 +-
 .../java/org/tensorflow/op/io/WriteFile.java  |   4 +-
 .../org/tensorflow/op/linalg/BandPart.java    |  14 +-
 .../tensorflow/op/linalg/BatchCholesky.java   |   9 +-
 .../op/linalg/BatchCholeskyGrad.java          |  11 +-
 .../op/linalg/BatchMatrixBandPart.java        |  14 +-
 .../op/linalg/BatchMatrixDeterminant.java     |  10 +-
 .../tensorflow/op/linalg/BatchMatrixDiag.java |  10 +-
 .../op/linalg/BatchMatrixDiagPart.java        |  10 +-
 .../op/linalg/BatchMatrixInverse.java         |   9 +-
 .../op/linalg/BatchMatrixSetDiag.java         |  12 +-
 .../op/linalg/BatchMatrixSolve.java           |  11 +-
 .../op/linalg/BatchMatrixSolveLs.java         |  13 +-
 .../op/linalg/BatchMatrixTriangularSolve.java |  11 +-
 .../op/linalg/BatchSelfAdjointEig.java        |   7 +-
 .../org/tensorflow/op/linalg/BatchSvd.java    |   8 +-
 .../org/tensorflow/op/linalg/Cholesky.java    |  10 +-
 .../tensorflow/op/linalg/CholeskyGrad.java    |  11 +-
 .../op/linalg/ConjugateTranspose.java         |  12 +-
 .../java/org/tensorflow/op/linalg/Cross.java  |  11 +-
 .../java/org/tensorflow/op/linalg/Det.java    |  10 +-
 .../java/org/tensorflow/op/linalg/Eig.java    |   8 +-
 .../java/org/tensorflow/op/linalg/Einsum.java |  10 +-
 .../tensorflow/op/linalg/EuclideanNorm.java   |  12 +-
 .../java/org/tensorflow/op/linalg/Inv.java    |  10 +-
 .../op/linalg/LoadAndRemapMatrix.java         |  12 +-
 .../op/linalg/LogMatrixDeterminant.java       |   8 +-
 .../gen/java/org/tensorflow/op/linalg/Lu.java |  10 +-
 .../java/org/tensorflow/op/linalg/MatMul.java |  12 +-
 .../org/tensorflow/op/linalg/MatrixDiag.java  |  18 +-
 .../tensorflow/op/linalg/MatrixDiagPart.java  |  14 +-
 .../op/linalg/MatrixDiagPartV3.java           |  14 +-
 .../tensorflow/op/linalg/MatrixDiagV3.java    |  18 +-
 .../tensorflow/op/linalg/MatrixLogarithm.java |  10 +-
 .../tensorflow/op/linalg/MatrixSetDiag.java   |  14 +-
 .../tensorflow/op/linalg/MatrixSolveLs.java   |  14 +-
 .../gen/java/org/tensorflow/op/linalg/Qr.java |   8 +-
 .../tensorflow/op/linalg/QuantizedMatMul.java |  18 +-
 .../op/linalg/QuantizedMatMulWithBias.java    |  20 +-
 .../QuantizedMatMulWithBiasAndRelu.java       |  20 +-
 ...zedMatMulWithBiasAndReluAndRequantize.java |  24 +-
 .../tensorflow/op/linalg/SelfAdjointEig.java  |   8 +-
 .../java/org/tensorflow/op/linalg/Solve.java  |  12 +-
 .../java/org/tensorflow/op/linalg/Sqrtm.java  |  10 +-
 .../java/org/tensorflow/op/linalg/Svd.java    |   8 +-
 .../org/tensorflow/op/linalg/TensorDiag.java  |  10 +-
 .../tensorflow/op/linalg/TensorDiagPart.java  |  10 +-
 .../org/tensorflow/op/linalg/Transpose.java   |  12 +-
 .../tensorflow/op/linalg/TriangularSolve.java |  12 +-
 .../op/linalg/TridiagonalMatMul.java          |  16 +-
 .../op/linalg/TridiagonalSolve.java           |  12 +-
 .../sparse/CSRSparseMatrixComponents.java     |  10 +-
 .../linalg/sparse/CSRSparseMatrixToDense.java |  10 +-
 .../sparse/CSRSparseMatrixToSparseTensor.java |   8 +-
 .../linalg/sparse/DenseToCSRSparseMatrix.java |  14 +-
 .../op/linalg/sparse/SparseMatrixAdd.java     |  18 +-
 .../op/linalg/sparse/SparseMatrixMatMul.java  |  12 +-
 .../op/linalg/sparse/SparseMatrixMul.java     |  14 +-
 .../op/linalg/sparse/SparseMatrixNNZ.java     |   4 +-
 .../sparse/SparseMatrixOrderingAMD.java       |   4 +-
 .../op/linalg/sparse/SparseMatrixSoftmax.java |  12 +-
 .../sparse/SparseMatrixSoftmaxGrad.java       |  14 +-
 .../sparse/SparseMatrixSparseCholesky.java    |  14 +-
 .../sparse/SparseMatrixSparseMatMul.java      |  14 +-
 .../linalg/sparse/SparseMatrixTranspose.java  |  12 +-
 .../op/linalg/sparse/SparseMatrixZeros.java   |  12 +-
 .../sparse/SparseTensorToCSRSparseMatrix.java |  16 +-
 .../gen/java/org/tensorflow/op/math/Abs.java  |   9 +-
 .../org/tensorflow/op/math/AccumulateN.java   |  10 +-
 .../gen/java/org/tensorflow/op/math/Acos.java |  10 +-
 .../java/org/tensorflow/op/math/Acosh.java    |  10 +-
 .../gen/java/org/tensorflow/op/math/Add.java  |  12 +-
 .../gen/java/org/tensorflow/op/math/AddN.java |  10 +-
 .../java/org/tensorflow/op/math/Angle.java    |  12 +-
 .../tensorflow/op/math/ApproximateEqual.java  |  10 +-
 .../java/org/tensorflow/op/math/ArgMax.java   |  14 +-
 .../java/org/tensorflow/op/math/ArgMin.java   |  14 +-
 .../gen/java/org/tensorflow/op/math/Asin.java |  10 +-
 .../java/org/tensorflow/op/math/Asinh.java    |  10 +-
 .../gen/java/org/tensorflow/op/math/Atan.java |  10 +-
 .../java/org/tensorflow/op/math/Atan2.java    |  11 +-
 .../java/org/tensorflow/op/math/Atanh.java    |  10 +-
 .../org/tensorflow/op/math/BesselI0e.java     |   9 +-
 .../org/tensorflow/op/math/BesselI1e.java     |   9 +-
 .../java/org/tensorflow/op/math/Betainc.java  |  13 +-
 .../java/org/tensorflow/op/math/Bincount.java |  13 +-
 .../gen/java/org/tensorflow/op/math/Ceil.java |   9 +-
 .../tensorflow/op/math/CompareAndBitpack.java |  10 +-
 .../org/tensorflow/op/math/ComplexAbs.java    |  12 +-
 .../gen/java/org/tensorflow/op/math/Conj.java |  10 +-
 .../gen/java/org/tensorflow/op/math/Cos.java  |  10 +-
 .../gen/java/org/tensorflow/op/math/Cosh.java |  10 +-
 .../java/org/tensorflow/op/math/Cumprod.java  |  12 +-
 .../java/org/tensorflow/op/math/Cumsum.java   |  12 +-
 .../op/math/CumulativeLogsumexp.java          |  11 +-
 .../java/org/tensorflow/op/math/Digamma.java  |   9 +-
 .../gen/java/org/tensorflow/op/math/Div.java  |  12 +-
 .../java/org/tensorflow/op/math/DivNoNan.java |  12 +-
 .../java/org/tensorflow/op/math/Equal.java    |  10 +-
 .../gen/java/org/tensorflow/op/math/Erf.java  |   9 +-
 .../gen/java/org/tensorflow/op/math/Erfc.java |   9 +-
 .../gen/java/org/tensorflow/op/math/Exp.java  |  10 +-
 .../java/org/tensorflow/op/math/Expm1.java    |  10 +-
 .../gen/java/org/tensorflow/op/math/Fact.java |   2 +-
 .../java/org/tensorflow/op/math/Floor.java    |   9 +-
 .../java/org/tensorflow/op/math/FloorDiv.java |  12 +-
 .../java/org/tensorflow/op/math/FloorMod.java |  11 +-
 .../java/org/tensorflow/op/math/Greater.java  |   9 +-
 .../org/tensorflow/op/math/GreaterEqual.java  |   9 +-
 .../java/org/tensorflow/op/math/Igamma.java   |  11 +-
 .../org/tensorflow/op/math/IgammaGradA.java   |  11 +-
 .../java/org/tensorflow/op/math/Igammac.java  |  11 +-
 .../gen/java/org/tensorflow/op/math/Imag.java |  12 +-
 .../tensorflow/op/math/InvertPermutation.java |   9 +-
 .../java/org/tensorflow/op/math/IsFinite.java |   7 +-
 .../java/org/tensorflow/op/math/IsInf.java    |   7 +-
 .../java/org/tensorflow/op/math/IsNan.java    |   7 +-
 .../gen/java/org/tensorflow/op/math/Less.java |   9 +-
 .../org/tensorflow/op/math/LessEqual.java     |   9 +-
 .../java/org/tensorflow/op/math/Lgamma.java   |   9 +-
 .../gen/java/org/tensorflow/op/math/Log.java  |  10 +-
 .../java/org/tensorflow/op/math/Log1p.java    |  10 +-
 .../org/tensorflow/op/math/LogicalAnd.java    |   6 +-
 .../org/tensorflow/op/math/LogicalNot.java    |   4 +-
 .../org/tensorflow/op/math/LogicalOr.java     |   6 +-
 .../java/org/tensorflow/op/math/Maximum.java  |  11 +-
 .../gen/java/org/tensorflow/op/math/Mean.java |  12 +-
 .../java/org/tensorflow/op/math/Minimum.java  |  11 +-
 .../gen/java/org/tensorflow/op/math/Mod.java  |  11 +-
 .../gen/java/org/tensorflow/op/math/Mul.java  |  12 +-
 .../java/org/tensorflow/op/math/MulNoNan.java |  12 +-
 .../java/org/tensorflow/op/math/Ndtri.java    |   9 +-
 .../gen/java/org/tensorflow/op/math/Neg.java  |  10 +-
 .../org/tensorflow/op/math/NextAfter.java     |  11 +-
 .../java/org/tensorflow/op/math/NotEqual.java |  10 +-
 .../org/tensorflow/op/math/Polygamma.java     |  11 +-
 .../tensorflow/op/math/PopulationCount.java   |   7 +-
 .../gen/java/org/tensorflow/op/math/Pow.java  |  12 +-
 .../org/tensorflow/op/math/QuantizedAdd.java  |  18 +-
 .../org/tensorflow/op/math/QuantizedMul.java  |  18 +-
 .../gen/java/org/tensorflow/op/math/Real.java |  12 +-
 .../java/org/tensorflow/op/math/RealDiv.java  |  12 +-
 .../org/tensorflow/op/math/Reciprocal.java    |  10 +-
 .../tensorflow/op/math/ReciprocalGrad.java    |  12 +-
 .../math/RequantizationRangePerChannel.java   |  10 +-
 .../op/math/RequantizePerChannel.java         |  16 +-
 .../gen/java/org/tensorflow/op/math/Rint.java |   9 +-
 .../java/org/tensorflow/op/math/Round.java    |  10 +-
 .../java/org/tensorflow/op/math/Rsqrt.java    |  10 +-
 .../org/tensorflow/op/math/RsqrtGrad.java     |  12 +-
 .../org/tensorflow/op/math/SegmentMax.java    |  11 +-
 .../org/tensorflow/op/math/SegmentMean.java   |  12 +-
 .../org/tensorflow/op/math/SegmentMin.java    |  11 +-
 .../org/tensorflow/op/math/SegmentProd.java   |  12 +-
 .../org/tensorflow/op/math/SegmentSum.java    |  12 +-
 .../java/org/tensorflow/op/math/Sigmoid.java  |  10 +-
 .../org/tensorflow/op/math/SigmoidGrad.java   |  12 +-
 .../gen/java/org/tensorflow/op/math/Sign.java |  10 +-
 .../gen/java/org/tensorflow/op/math/Sin.java  |  10 +-
 .../gen/java/org/tensorflow/op/math/Sinh.java |  10 +-
 .../org/tensorflow/op/math/SobolSample.java   |  13 +-
 .../java/org/tensorflow/op/math/Softplus.java |   9 +-
 .../org/tensorflow/op/math/SoftplusGrad.java  |  11 +-
 .../gen/java/org/tensorflow/op/math/Sqrt.java |  10 +-
 .../java/org/tensorflow/op/math/SqrtGrad.java |  12 +-
 .../java/org/tensorflow/op/math/Square.java   |  10 +-
 .../tensorflow/op/math/SquaredDifference.java |  12 +-
 .../gen/java/org/tensorflow/op/math/Sub.java  |  12 +-
 .../gen/java/org/tensorflow/op/math/Tan.java  |  10 +-
 .../gen/java/org/tensorflow/op/math/Tanh.java |  10 +-
 .../java/org/tensorflow/op/math/TanhGrad.java |  12 +-
 .../org/tensorflow/op/math/TruncateDiv.java   |  12 +-
 .../org/tensorflow/op/math/TruncateMod.java   |  11 +-
 .../op/math/UnsortedSegmentMax.java           |  13 +-
 .../op/math/UnsortedSegmentMin.java           |  13 +-
 .../op/math/UnsortedSegmentProd.java          |  14 +-
 .../op/math/UnsortedSegmentSum.java           |  14 +-
 .../java/org/tensorflow/op/math/Xdivy.java    |  12 +-
 .../java/org/tensorflow/op/math/Xlog1py.java  |  12 +-
 .../java/org/tensorflow/op/math/Xlogy.java    |  12 +-
 .../gen/java/org/tensorflow/op/math/Zeta.java |  11 +-
 .../java/org/tensorflow/op/math/erfinv.java   |   9 +-
 .../org/tensorflow/op/math/special/Dawsn.java |   9 +-
 .../tensorflow/op/math/special/Expint.java    |   9 +-
 .../op/math/special/FresnelCos.java           |   9 +-
 .../op/math/special/FresnelSin.java           |   9 +-
 .../tensorflow/op/math/special/Spence.java    |   9 +-
 .../java/org/tensorflow/op/nn/AvgPool.java    |   9 +-
 .../java/org/tensorflow/op/nn/AvgPool3d.java  |   9 +-
 .../org/tensorflow/op/nn/AvgPool3dGrad.java   |  11 +-
 .../org/tensorflow/op/nn/AvgPoolGrad.java     |  11 +-
 .../nn/BatchNormWithGlobalNormalization.java  |  18 +-
 .../BatchNormWithGlobalNormalizationGrad.java |  16 +-
 .../java/org/tensorflow/op/nn/BiasAdd.java    |  12 +-
 .../org/tensorflow/op/nn/BiasAddGrad.java     |  10 +-
 .../java/org/tensorflow/op/nn/BlockLSTM.java  |  23 +-
 .../org/tensorflow/op/nn/BlockLSTMGrad.java   |  41 +-
 .../java/org/tensorflow/op/nn/CTCLossV2.java  |   8 +-
 .../op/nn/ComputeAccidentalHits.java          |   4 +-
 .../gen/java/org/tensorflow/op/nn/Conv2d.java |  11 +-
 .../op/nn/Conv2dBackpropFilter.java           |  13 +-
 .../tensorflow/op/nn/Conv2dBackpropInput.java |  13 +-
 .../gen/java/org/tensorflow/op/nn/Conv3d.java |  11 +-
 .../op/nn/Conv3dBackpropFilter.java           |  13 +-
 .../tensorflow/op/nn/Conv3dBackpropInput.java |  13 +-
 .../op/nn/CtcBeamSearchDecoder.java           |   9 +-
 .../tensorflow/op/nn/CtcGreedyDecoder.java    |   9 +-
 .../java/org/tensorflow/op/nn/CtcLoss.java    |  13 +-
 .../java/org/tensorflow/op/nn/CudnnRNN.java   |  15 +-
 .../tensorflow/op/nn/CudnnRNNBackprop.java    |  31 +-
 .../op/nn/CudnnRNNCanonicalToParams.java      |  17 +-
 .../op/nn/CudnnRNNParamsToCanonical.java      |  13 +-
 .../tensorflow/op/nn/CudnnRnnParamsSize.java  |  13 +-
 .../tensorflow/op/nn/DataFormatDimMap.java    |   9 +-
 .../op/nn/DataFormatVecPermute.java           |   9 +-
 .../org/tensorflow/op/nn/DepthToSpace.java    |  10 +-
 .../op/nn/DepthwiseConv2dNative.java          |  11 +-
 .../DepthwiseConv2dNativeBackpropFilter.java  |  13 +-
 .../DepthwiseConv2dNativeBackpropInput.java   |  13 +-
 .../java/org/tensorflow/op/nn/Dilation2d.java |  11 +-
 .../op/nn/Dilation2dBackpropFilter.java       |  13 +-
 .../op/nn/Dilation2dBackpropInput.java        |  13 +-
 .../gen/java/org/tensorflow/op/nn/Elu.java    |   9 +-
 .../java/org/tensorflow/op/nn/EluGrad.java    |  11 +-
 .../op/nn/FixedUnigramCandidateSampler.java   |   2 +-
 .../tensorflow/op/nn/FractionalAvgPool.java   |   7 +-
 .../op/nn/FractionalAvgPoolGrad.java          |  15 +-
 .../tensorflow/op/nn/FractionalMaxPool.java   |   7 +-
 .../op/nn/FractionalMaxPoolGrad.java          |  17 +-
 .../org/tensorflow/op/nn/FusedBatchNorm.java  |  15 +-
 .../tensorflow/op/nn/FusedBatchNormGrad.java  |  17 +-
 .../org/tensorflow/op/nn/FusedPadConv2d.java  |  13 +-
 .../op/nn/FusedResizeAndPadConv2d.java        |  15 +-
 .../org/tensorflow/op/nn/GRUBlockCell.java    |  17 +-
 .../tensorflow/op/nn/GRUBlockCellGrad.java    |  25 +-
 .../gen/java/org/tensorflow/op/nn/InTopK.java |  11 +-
 .../java/org/tensorflow/op/nn/InvGrad.java    |  12 +-
 .../gen/java/org/tensorflow/op/nn/L2Loss.java |   9 +-
 .../org/tensorflow/op/nn/LSTMBlockCell.java   |  21 +-
 .../tensorflow/op/nn/LSTMBlockCellGrad.java   |  37 +-
 .../java/org/tensorflow/op/nn/LeakyRelu.java  |   9 +-
 .../op/nn/LearnedUnigramCandidateSampler.java |   2 +-
 .../op/nn/LocalResponseNormalization.java     |   9 +-
 .../op/nn/LocalResponseNormalizationGrad.java |  13 +-
 .../java/org/tensorflow/op/nn/LogSoftmax.java |   9 +-
 .../java/org/tensorflow/op/nn/MaxPool.java    |  14 +-
 .../java/org/tensorflow/op/nn/MaxPool3d.java  |   9 +-
 .../org/tensorflow/op/nn/MaxPool3dGrad.java   |  13 +-
 .../tensorflow/op/nn/MaxPool3dGradGrad.java   |  13 +-
 .../org/tensorflow/op/nn/MaxPoolGrad.java     |  17 +-
 .../org/tensorflow/op/nn/MaxPoolGradGrad.java |  17 +-
 .../op/nn/MaxPoolGradGradWithArgmax.java      |  13 +-
 .../op/nn/MaxPoolGradWithArgmax.java          |  13 +-
 .../tensorflow/op/nn/MaxPoolWithArgmax.java   |   9 +-
 .../java/org/tensorflow/op/nn/NthElement.java |  11 +-
 .../tensorflow/op/nn/QuantizedAvgPool.java    |  12 +-
 ...tizedBatchNormWithGlobalNormalization.java |  36 +-
 .../tensorflow/op/nn/QuantizedBiasAdd.java    |  18 +-
 .../op/nn/QuantizedConv2DAndRelu.java         |  18 +-
 .../QuantizedConv2DAndReluAndRequantize.java  |  22 +-
 .../op/nn/QuantizedConv2DAndRequantize.java   |  22 +-
 .../op/nn/QuantizedConv2DPerChannel.java      |  18 +-
 .../op/nn/QuantizedConv2DWithBias.java        |  20 +-
 .../op/nn/QuantizedConv2DWithBiasAndRelu.java |  20 +-
 ...zedConv2DWithBiasAndReluAndRequantize.java |  24 +-
 .../QuantizedConv2DWithBiasAndRequantize.java |  24 +-
 ...WithBiasSignedSumAndReluAndRequantize.java |  30 +-
 .../nn/QuantizedConv2DWithBiasSumAndRelu.java |  22 +-
 ...Conv2DWithBiasSumAndReluAndRequantize.java |  30 +-
 .../org/tensorflow/op/nn/QuantizedConv2d.java |  18 +-
 .../op/nn/QuantizedDepthwiseConv2D.java       |  18 +-
 .../nn/QuantizedDepthwiseConv2DWithBias.java  |  20 +-
 ...antizedDepthwiseConv2DWithBiasAndRelu.java |  20 +-
 ...iseConv2DWithBiasAndReluAndRequantize.java |  24 +-
 .../op/nn/QuantizedInstanceNorm.java          |  12 +-
 .../tensorflow/op/nn/QuantizedMaxPool.java    |  12 +-
 .../org/tensorflow/op/nn/QuantizedRelu.java   |  12 +-
 .../org/tensorflow/op/nn/QuantizedRelu6.java  |  12 +-
 .../org/tensorflow/op/nn/QuantizedReluX.java  |  14 +-
 .../gen/java/org/tensorflow/op/nn/Relu.java   |  10 +-
 .../gen/java/org/tensorflow/op/nn/Relu6.java  |   9 +-
 .../java/org/tensorflow/op/nn/Relu6Grad.java  |  11 +-
 .../java/org/tensorflow/op/nn/ReluGrad.java   |  11 +-
 .../gen/java/org/tensorflow/op/nn/Selu.java   |   9 +-
 .../java/org/tensorflow/op/nn/SeluGrad.java   |  11 +-
 .../java/org/tensorflow/op/nn/Softmax.java    |   9 +-
 .../java/org/tensorflow/op/nn/Softsign.java   |   9 +-
 .../org/tensorflow/op/nn/SoftsignGrad.java    |  11 +-
 .../org/tensorflow/op/nn/SpaceToBatch.java    |  12 +-
 .../org/tensorflow/op/nn/SpaceToDepth.java    |  10 +-
 .../gen/java/org/tensorflow/op/nn/TopK.java   |   9 +-
 .../nn/raw/SoftmaxCrossEntropyWithLogits.java |   9 +-
 .../SparseSoftmaxCrossEntropyWithLogits.java  |   9 +-
 .../op/quantization/Dequantize.java           |  16 +-
 .../quantization/FakeQuantWithMinMaxArgs.java |   4 +-
 .../FakeQuantWithMinMaxArgsGradient.java      |   6 +-
 .../quantization/FakeQuantWithMinMaxVars.java |   8 +-
 .../FakeQuantWithMinMaxVarsGradient.java      |   8 +-
 .../FakeQuantWithMinMaxVarsPerChannel.java    |   8 +-
 ...QuantWithMinMaxVarsPerChannelGradient.java |   8 +-
 .../tensorflow/op/quantization/Quantize.java  |  12 +-
 .../quantization/QuantizeAndDequantize.java   |  15 +-
 .../QuantizeDownAndShrinkRange.java           |  12 +-
 .../op/quantization/QuantizedConcat.java      |  14 +-
 .../QuantizedMatMulWithBiasAndDequantize.java |  26 +-
 .../QuantizedMatMulWithBiasAndRequantize.java |  24 +-
 .../op/quantization/RequantizationRange.java  |  10 +-
 .../op/quantization/Requantize.java           |  16 +-
 .../tensorflow/op/ragged/RaggedGather.java    |  12 +-
 .../org/tensorflow/op/ragged/RaggedRange.java |  13 +-
 .../op/ragged/RaggedTensorFromVariant.java    |  10 +-
 .../op/ragged/RaggedTensorToSparse.java       |  10 +-
 .../op/ragged/RaggedTensorToTensor.java       |  16 +-
 .../op/ragged/RaggedTensorToVariant.java      |  14 +-
 .../op/random/AllCandidateSampler.java        |   2 +-
 .../random/AnonymousRandomSeedGenerator.java  |   4 +-
 .../op/random/DeleteRandomSeedGenerator.java  |   4 +-
 .../op/random/LogUniformCandidateSampler.java |   2 +-
 .../org/tensorflow/op/random/Multinomial.java |  13 +-
 .../op/random/NonDeterministicInts.java       |  12 +-
 .../random/ParameterizedTruncatedNormal.java  |  17 +-
 .../org/tensorflow/op/random/RandomGamma.java |  11 +-
 .../tensorflow/op/random/RandomGammaGrad.java |  11 +-
 .../tensorflow/op/random/RandomPoisson.java   |  13 +-
 .../tensorflow/op/random/RandomShuffle.java   |  10 +-
 .../op/random/RandomStandardNormal.java       |   9 +-
 .../tensorflow/op/random/RandomUniform.java   |   9 +-
 .../op/random/RandomUniformInt.java           |  13 +-
 .../org/tensorflow/op/random/RecordInput.java |   2 +-
 .../org/tensorflow/op/random/RngSkip.java     |   6 +-
 .../op/random/StatefulRandomBinomial.java     |  19 +-
 .../op/random/StatefulStandardNormal.java     |  16 +-
 .../op/random/StatefulTruncatedNormal.java    |  16 +-
 .../tensorflow/op/random/StatefulUniform.java |  16 +-
 .../op/random/StatefulUniformFullInt.java     |  14 +-
 .../op/random/StatefulUniformInt.java         |  18 +-
 .../op/random/StatelessMultinomial.java       |  15 +-
 .../op/random/StatelessRandomBinomial.java    |  17 +-
 .../op/random/StatelessRandomGamma.java       |  13 +-
 .../op/random/StatelessRandomNormal.java      |  13 +-
 .../op/random/StatelessRandomPoisson.java     |  13 +-
 .../op/random/StatelessRandomUniform.java     |  13 +-
 .../random/StatelessRandomUniformFullInt.java |  11 +-
 .../op/random/StatelessRandomUniformInt.java  |  15 +-
 .../op/random/StatelessTruncatedNormal.java   |  13 +-
 .../tensorflow/op/random/TruncatedNormal.java |   9 +-
 .../op/random/UniformCandidateSampler.java    |   2 +-
 .../org/tensorflow/op/signal/BatchFft.java    |  10 +-
 .../org/tensorflow/op/signal/BatchFft2d.java  |  10 +-
 .../org/tensorflow/op/signal/BatchFft3d.java  |  10 +-
 .../org/tensorflow/op/signal/BatchIfft.java   |  10 +-
 .../org/tensorflow/op/signal/BatchIfft2d.java |  10 +-
 .../org/tensorflow/op/signal/BatchIfft3d.java |  10 +-
 .../java/org/tensorflow/op/signal/Fft.java    |  10 +-
 .../java/org/tensorflow/op/signal/Fft2d.java  |  10 +-
 .../java/org/tensorflow/op/signal/Fft3d.java  |  10 +-
 .../java/org/tensorflow/op/signal/Ifft.java   |  10 +-
 .../java/org/tensorflow/op/signal/Ifft2d.java |  10 +-
 .../java/org/tensorflow/op/signal/Ifft3d.java |  10 +-
 .../java/org/tensorflow/op/signal/Irfft.java  |  14 +-
 .../org/tensorflow/op/signal/Irfft2d.java     |  14 +-
 .../org/tensorflow/op/signal/Irfft3d.java     |  14 +-
 .../java/org/tensorflow/op/signal/Rfft.java   |  12 +-
 .../java/org/tensorflow/op/signal/Rfft2d.java |  12 +-
 .../java/org/tensorflow/op/signal/Rfft3d.java |  12 +-
 .../op/sparse/AddManySparseToTensorsMap.java  |  12 +-
 .../op/sparse/AddSparseToTensorsMap.java      |  12 +-
 .../op/sparse/DenseToDenseSetOperation.java   |  10 +-
 .../op/sparse/DenseToSparseSetOperation.java  |  14 +-
 .../op/sparse/DeserializeSparse.java          |   8 +-
 .../SparseAccumulatorApplyGradient.java       |  14 +-
 .../sparse/SparseAccumulatorTakeGradient.java |  10 +-
 .../org/tensorflow/op/sparse/SparseAdd.java   |  20 +-
 .../tensorflow/op/sparse/SparseAddGrad.java   |  14 +-
 .../tensorflow/op/sparse/SparseConcat.java    |  12 +-
 .../sparse/SparseConditionalAccumulator.java  |   6 +-
 .../org/tensorflow/op/sparse/SparseCross.java |  14 +-
 .../op/sparse/SparseDenseCwiseAdd.java        |  16 +-
 .../op/sparse/SparseDenseCwiseDiv.java        |  16 +-
 .../op/sparse/SparseDenseCwiseMul.java        |  16 +-
 .../op/sparse/SparseFillEmptyRows.java        |  14 +-
 .../op/sparse/SparseFillEmptyRowsGrad.java    |  10 +-
 .../tensorflow/op/sparse/SparseMatMul.java    |   9 +-
 .../tensorflow/op/sparse/SparseReduceMax.java |  15 +-
 .../op/sparse/SparseReduceMaxSparse.java      |  13 +-
 .../tensorflow/op/sparse/SparseReduceSum.java |  16 +-
 .../op/sparse/SparseReduceSumSparse.java      |  14 +-
 .../tensorflow/op/sparse/SparseReorder.java   |  12 +-
 .../tensorflow/op/sparse/SparseReshape.java   |   6 +-
 .../op/sparse/SparseSegmentMean.java          |  13 +-
 .../op/sparse/SparseSegmentMeanGrad.java      |  15 +-
 .../SparseSegmentMeanWithNumSegments.java     |  15 +-
 .../op/sparse/SparseSegmentSqrtN.java         |  13 +-
 .../op/sparse/SparseSegmentSqrtNGrad.java     |  15 +-
 .../SparseSegmentSqrtNWithNumSegments.java    |  15 +-
 .../op/sparse/SparseSegmentSum.java           |  13 +-
 .../SparseSegmentSumWithNumSegments.java      |  15 +-
 .../org/tensorflow/op/sparse/SparseSlice.java |  16 +-
 .../tensorflow/op/sparse/SparseSliceGrad.java |  16 +-
 .../tensorflow/op/sparse/SparseSoftmax.java   |  13 +-
 .../op/sparse/SparseSparseMaximum.java        |  17 +-
 .../op/sparse/SparseSparseMinimum.java        |  18 +-
 .../org/tensorflow/op/sparse/SparseSplit.java |  14 +-
 .../op/sparse/SparseTensorDenseAdd.java       |  16 +-
 .../op/sparse/SparseTensorDenseMatMul.java    |  16 +-
 .../tensorflow/op/sparse/SparseToDense.java   |  16 +-
 .../op/sparse/SparseToSparseSetOperation.java |  18 +-
 .../sparse/TakeManySparseFromTensorsMap.java  |   8 +-
 .../java/org/tensorflow/op/strings/Join.java  |   4 +-
 .../java/org/tensorflow/op/strings/Lower.java |   4 +-
 .../org/tensorflow/op/strings/ReduceJoin.java |   6 +-
 .../tensorflow/op/strings/RegexFullMatch.java |   6 +-
 .../tensorflow/op/strings/RegexReplace.java   |   8 +-
 .../op/strings/StaticRegexFullMatch.java      |   4 +-
 .../op/strings/StaticRegexReplace.java        |   4 +-
 .../tensorflow/op/strings/StringFormat.java   |   4 +-
 .../tensorflow/op/strings/StringLength.java   |   4 +-
 .../tensorflow/op/strings/StringNGrams.java   |   9 +-
 .../tensorflow/op/strings/StringSplit.java    |   4 +-
 .../java/org/tensorflow/op/strings/Strip.java |   4 +-
 .../org/tensorflow/op/strings/Substr.java     |  11 +-
 .../tensorflow/op/strings/ToHashBucket.java   |   4 +-
 .../op/strings/ToHashBucketFast.java          |   4 +-
 .../op/strings/ToHashBucketStrong.java        |   4 +-
 .../org/tensorflow/op/strings/ToNumber.java   |   9 +-
 .../tensorflow/op/strings/UnicodeDecode.java  |   7 +-
 .../op/strings/UnicodeDecodeWithOffsets.java  |   7 +-
 .../tensorflow/op/strings/UnicodeEncode.java  |   9 +-
 .../tensorflow/op/strings/UnicodeScript.java  |   4 +-
 .../op/strings/UnicodeTranscode.java          |   4 +-
 .../op/strings/UnsortedSegmentJoin.java       |  11 +-
 .../java/org/tensorflow/op/strings/Upper.java |   4 +-
 .../tensorflow/op/summary/AudioSummary.java   |   8 +-
 .../op/summary/CloseSummaryWriter.java        |   2 +-
 .../op/summary/CreateSummaryDbWriter.java     |  10 +-
 .../op/summary/CreateSummaryFileWriter.java   |  10 +-
 .../op/summary/FlushSummaryWriter.java        |   2 +-
 .../op/summary/HistogramSummary.java          |   9 +-
 .../tensorflow/op/summary/ImageSummary.java   |   8 +-
 .../tensorflow/op/summary/ImportEvent.java    |   4 +-
 .../tensorflow/op/summary/MergeSummary.java   |   4 +-
 .../tensorflow/op/summary/ScalarSummary.java  |   9 +-
 .../op/summary/StatsAggregatorSummary.java    |   4 +-
 .../tensorflow/op/summary/SummaryWriter.java  |   8 +-
 .../tensorflow/op/summary/TensorSummary.java  |  12 +-
 .../op/summary/WriteAudioSummary.java         |  10 +-
 .../op/summary/WriteGraphSummary.java         |   6 +-
 .../op/summary/WriteHistogramSummary.java     |  11 +-
 .../op/summary/WriteImageSummary.java         |  13 +-
 .../op/summary/WriteRawProtoSummary.java      |   6 +-
 .../op/summary/WriteScalarSummary.java        |  11 +-
 .../tensorflow/op/summary/WriteSummary.java   |  14 +-
 .../java/org/tensorflow/op/tpu/AllToAll.java  |  12 +-
 .../tensorflow/op/tpu/CollectivePermute.java  |  12 +-
 .../op/tpu/ConfigureDistributedTPU.java       |   2 +-
 .../tensorflow/op/tpu/CrossReplicaSum.java    |  11 +-
 .../tpu/EnqueueTPUEmbeddingIntegerBatch.java  |   4 +-
 .../tpu/EnqueueTPUEmbeddingSparseBatch.java   |  11 +-
 .../EnqueueTPUEmbeddingSparseTensorBatch.java |  11 +-
 .../org/tensorflow/op/tpu/InfeedDequeue.java  |   8 +-
 .../tensorflow/op/tpu/InfeedDequeueTuple.java |   6 +-
 .../org/tensorflow/op/tpu/InfeedEnqueue.java  |   6 +-
 .../tpu/InfeedEnqueuePrelinearizedBuffer.java |   2 +-
 .../tensorflow/op/tpu/InfeedEnqueueTuple.java |   2 +-
 .../tpu/LoadTPUEmbeddingADAMParameters.java   |   6 +-
 ...EmbeddingADAMParametersGradAccumDebug.java |   8 +-
 .../LoadTPUEmbeddingAdadeltaParameters.java   |   6 +-
 ...ddingAdadeltaParametersGradAccumDebug.java |   8 +-
 .../LoadTPUEmbeddingAdagradParameters.java    |   4 +-
 ...eddingAdagradParametersGradAccumDebug.java |   6 +-
 ...TPUEmbeddingCenteredRMSPropParameters.java |   8 +-
 .../tpu/LoadTPUEmbeddingFTRLParameters.java   |   6 +-
 ...EmbeddingFTRLParametersGradAccumDebug.java |   8 +-
 ...TPUEmbeddingMDLAdagradLightParameters.java |   8 +-
 .../LoadTPUEmbeddingMomentumParameters.java   |   4 +-
 ...ddingMomentumParametersGradAccumDebug.java |   6 +-
 ...TPUEmbeddingProximalAdagradParameters.java |   4 +-
 ...oximalAdagradParametersGradAccumDebug.java |   6 +-
 .../LoadTPUEmbeddingRMSPropParameters.java    |   6 +-
 ...eddingRMSPropParametersGradAccumDebug.java |   8 +-
 ...ngStochasticGradientDescentParameters.java |   2 +-
 .../org/tensorflow/op/tpu/OutfeedDequeue.java |   8 +-
 .../op/tpu/OutfeedDequeueTuple.java           |   6 +-
 .../org/tensorflow/op/tpu/OutfeedEnqueue.java |   6 +-
 .../op/tpu/OutfeedEnqueueTuple.java           |   2 +-
 .../org/tensorflow/op/tpu/Prelinearize.java   |  12 +-
 .../tensorflow/op/tpu/PrelinearizeTuple.java  |  10 +-
 ...ngStochasticGradientDescentParameters.java |   2 +-
 .../op/tpu/SendTPUEmbeddingGradients.java     |   4 +-
 .../op/tpu/TPUCompilationResult.java          |   2 +-
 .../op/tpu/TPUEmbeddingActivations.java       |   6 +-
 .../tensorflow/op/tpu/TPUOrdinalSelector.java |   2 +-
 .../tensorflow/op/tpu/TPUReplicatedInput.java |  10 +-
 .../op/tpu/TPUReplicatedOutput.java           |   8 +-
 .../tensorflow/op/tpu/WorkerHeartbeat.java    |   4 +-
 .../op/train/AccumulatorApplyGradient.java    |  10 +-
 .../op/train/AccumulatorNumAccumulated.java   |   4 +-
 .../op/train/AccumulatorSetGlobalStep.java    |   4 +-
 .../op/train/AccumulatorTakeGradient.java     |  12 +-
 .../org/tensorflow/op/train/ApplyAdaMax.java  |  26 +-
 .../tensorflow/op/train/ApplyAdadelta.java    |  22 +-
 .../org/tensorflow/op/train/ApplyAdagrad.java |  16 +-
 .../tensorflow/op/train/ApplyAdagradDa.java   |  24 +-
 .../tensorflow/op/train/ApplyAdagradV2.java   |  18 +-
 .../org/tensorflow/op/train/ApplyAdam.java    |  28 +-
 .../org/tensorflow/op/train/ApplyAddSign.java |  22 +-
 .../op/train/ApplyCenteredRmsProp.java        |  26 +-
 .../org/tensorflow/op/train/ApplyFtrl.java    |  26 +-
 .../op/train/ApplyGradientDescent.java        |  14 +-
 .../tensorflow/op/train/ApplyMomentum.java    |  18 +-
 .../tensorflow/op/train/ApplyPowerSign.java   |  22 +-
 .../op/train/ApplyProximalAdagrad.java        |  20 +-
 .../train/ApplyProximalGradientDescent.java   |  18 +-
 .../org/tensorflow/op/train/ApplyRmsProp.java |  24 +-
 .../org/tensorflow/op/train/BatchMatMul.java  |  12 +-
 .../op/train/ConditionalAccumulator.java      |   6 +-
 .../op/train/GenerateVocabRemapping.java      |   4 +-
 .../op/train/MergeV2Checkpoints.java          |   4 +-
 .../org/tensorflow/op/train/NegTrain.java     |  10 +-
 .../tensorflow/op/train/PreventGradient.java  |  10 +-
 .../ResourceAccumulatorApplyGradient.java     |  10 +-
 .../ResourceAccumulatorNumAccumulated.java    |   4 +-
 .../ResourceAccumulatorSetGlobalStep.java     |   4 +-
 .../ResourceAccumulatorTakeGradient.java      |  12 +-
 .../op/train/ResourceApplyAdaMax.java         |  22 +-
 .../op/train/ResourceApplyAdadelta.java       |  18 +-
 .../op/train/ResourceApplyAdagrad.java        |  14 +-
 .../op/train/ResourceApplyAdagradDa.java      |  20 +-
 .../op/train/ResourceApplyAdam.java           |  24 +-
 .../train/ResourceApplyAdamWithAmsgrad.java   |  26 +-
 .../op/train/ResourceApplyAddSign.java        |  18 +-
 .../train/ResourceApplyCenteredRmsProp.java   |  22 +-
 .../op/train/ResourceApplyFtrl.java           |  22 +-
 .../train/ResourceApplyGradientDescent.java   |  10 +-
 .../op/train/ResourceApplyKerasMomentum.java  |  14 +-
 .../op/train/ResourceApplyMomentum.java       |  14 +-
 .../op/train/ResourceApplyPowerSign.java      |  18 +-
 .../train/ResourceApplyProximalAdagrad.java   |  16 +-
 .../ResourceApplyProximalGradientDescent.java |  14 +-
 .../op/train/ResourceApplyRmsProp.java        |  20 +-
 .../train/ResourceConditionalAccumulator.java |  10 +-
 .../op/train/ResourceSparseApplyAdadelta.java |  20 +-
 .../op/train/ResourceSparseApplyAdagrad.java  |  14 +-
 .../train/ResourceSparseApplyAdagradDa.java   |  22 +-
 .../train/ResourceSparseApplyAdagradV2.java   |  16 +-
 .../ResourceSparseApplyCenteredRmsProp.java   |  24 +-
 .../op/train/ResourceSparseApplyFtrl.java     |  24 +-
 .../ResourceSparseApplyKerasMomentum.java     |  16 +-
 .../op/train/ResourceSparseApplyMomentum.java |  16 +-
 .../ResourceSparseApplyProximalAdagrad.java   |  18 +-
 ...rceSparseApplyProximalGradientDescent.java |  16 +-
 .../op/train/ResourceSparseApplyRmsProp.java  |  22 +-
 .../java/org/tensorflow/op/train/Restore.java |  12 +-
 .../org/tensorflow/op/train/RestoreSlice.java |  14 +-
 .../java/org/tensorflow/op/train/Save.java    |   8 +-
 .../org/tensorflow/op/train/SaveSlices.java   |   8 +-
 .../org/tensorflow/op/train/SdcaFprint.java   |   4 +-
 .../tensorflow/op/train/SdcaOptimizer.java    |  20 +-
 .../org/tensorflow/op/train/SdcaShrinkL1.java |   2 +-
 .../op/train/SparseApplyAdadelta.java         |  24 +-
 .../op/train/SparseApplyAdagrad.java          |  20 +-
 .../op/train/SparseApplyAdagradDa.java        |  26 +-
 .../op/train/SparseApplyCenteredRmsProp.java  |  28 +-
 .../tensorflow/op/train/SparseApplyFtrl.java  |  28 +-
 .../op/train/SparseApplyMomentum.java         |  20 +-
 .../op/train/SparseApplyProximalAdagrad.java  |  22 +-
 .../SparseApplyProximalGradientDescent.java   |  20 +-
 .../op/train/SparseApplyRmsProp.java          |  26 +-
 .../org/tensorflow/op/train/TileGrad.java     |  12 +-
 .../tensorflow/op/xla/BroadcastHelper.java    |  12 +-
 .../org/tensorflow/op/xla/ClusterOutput.java  |  10 +-
 .../gen/java/org/tensorflow/op/xla/Conv.java  |  22 +-
 .../org/tensorflow/op/xla/Dequantize.java     |   4 +-
 .../gen/java/org/tensorflow/op/xla/Dot.java   |  12 +-
 .../org/tensorflow/op/xla/DynamicSlice.java   |  14 +-
 .../tensorflow/op/xla/DynamicUpdateSlice.java |  14 +-
 .../java/org/tensorflow/op/xla/Einsum.java    |  12 +-
 .../java/org/tensorflow/op/xla/Gather.java    |  14 +-
 .../org/tensorflow/op/xla/KeyValueSort.java   |  10 +-
 .../gen/java/org/tensorflow/op/xla/Pad.java   |  18 +-
 .../gen/java/org/tensorflow/op/xla/Recv.java  |   8 +-
 .../java/org/tensorflow/op/xla/ReplicaId.java |   2 +-
 .../org/tensorflow/op/xla/SelfAdjointEig.java |   8 +-
 .../gen/java/org/tensorflow/op/xla/Send.java  |   6 +-
 .../java/org/tensorflow/op/xla/Sharding.java  |  10 +-
 .../gen/java/org/tensorflow/op/xla/Sort.java  |  10 +-
 .../gen/java/org/tensorflow/op/xla/Svd.java   |   8 +-
 .../org/tensorflow/AbstractOperation.java     |   3 +-
 .../java/org/tensorflow/AbstractTensor.java   | 105 ++++
 .../main/java/org/tensorflow/DataType.java    |   5 +-
 .../java/org/tensorflow/EagerOperation.java   |  12 +-
 .../org/tensorflow/EagerOperationBuilder.java |   2 +-
 .../org/tensorflow/GraphOperationBuilder.java |   4 +-
 .../src/main/java/org/tensorflow/Operand.java |  35 +-
 .../main/java/org/tensorflow/Operation.java   |   4 +-
 .../src/main/java/org/tensorflow/Output.java  |   8 +-
 .../src/main/java/org/tensorflow/Session.java |   2 +-
 .../src/main/java/org/tensorflow/Tensor.java  | 209 +-------
 .../src/main/java/org/tensorflow/Tensors.java | 141 +++++-
 .../internal/tensor/BooleanTensorImpl.java    | 115 +++--
 .../internal/tensor/ByteTensorImpl.java       | 114 +++--
 .../internal/tensor/DoubleTensorImpl.java     | 113 +++--
 .../internal/tensor/FloatTensorImpl.java      | 112 +++--
 .../internal/tensor/IntTensorImpl.java        | 113 +++--
 .../internal/tensor/LongTensorImpl.java       | 113 +++--
 .../tensorflow/internal/tensor/RawTensor.java |  58 ---
 .../internal/tensor/StringTensorImpl.java     | 115 +++--
 .../main/java/org/tensorflow/op/Operands.java |   9 +-
 .../java/org/tensorflow/op/core/Constant.java |  14 +-
 .../org/tensorflow/op/core/Gradients.java     |  13 +-
 .../java/org/tensorflow/op/core/Helpers.java  |   4 +-
 .../java/org/tensorflow/op/core/Shapes.java   |  44 +-
 .../java/org/tensorflow/op/core/Zeros.java    |   8 +-
 .../org/tensorflow/tensor/BooleanTensor.java  |  11 +-
 .../org/tensorflow/tensor/ByteTensor.java     |  10 +-
 .../org/tensorflow/tensor/DoubleTensor.java   |  11 +-
 .../org/tensorflow/tensor/FloatTensor.java    |  11 +-
 .../java/org/tensorflow/tensor/IntTensor.java |  11 +-
 .../org/tensorflow/tensor/LongTensor.java     |  11 +-
 .../org/tensorflow/tensor/StringTensor.java   |  16 +-
 .../java/org/tensorflow/types/TBfloat16.java  |  18 +-
 .../main/java/org/tensorflow/types/TBool.java |  18 +-
 .../java/org/tensorflow/types/TFloat16.java   |  18 +-
 .../java/org/tensorflow/types/TFloat32.java   |  18 +-
 .../java/org/tensorflow/types/TFloat64.java   |  18 +-
 .../java/org/tensorflow/types/TInt32.java     |  18 +-
 .../java/org/tensorflow/types/TInt64.java     |  18 +-
 .../java/org/tensorflow/types/TString.java    |  12 +-
 .../java/org/tensorflow/types/TUint8.java     |  18 +-
 .../org/tensorflow/types/family/TNumber.java  |   2 +-
 .../org/tensorflow/types/family/TType.java    |  38 +-
 .../test/java/org/tensorflow/TensorTest.java  |  36 +-
 .../java/org/tensorflow/op/OperandsTest.java  |   2 +-
 .../java/org/tensorflow/op/ScopeTest.java     |  15 +-
 .../org/tensorflow/op/core/ConstantTest.java  |  25 +-
 .../types/NumericTypesTestBase.java           |  70 +--
 .../org/tensorflow/types/TBfloat16Test.java   |   7 +
 .../org/tensorflow/types/TFloat16Test.java    |   7 +
 .../org/tensorflow/types/TFloat32Test.java    |   7 +
 .../org/tensorflow/types/TFloat64Test.java    |   7 +
 .../java/org/tensorflow/types/TInt32Test.java |  16 +
 .../java/org/tensorflow/types/TInt64Test.java |   7 +
 .../org/tensorflow/types/TStringTest.java     |   2 +-
 .../java/org/tensorflow/types/TUint8Test.java |   7 +
 .../framework/data/DatasetIterator.java       |   2 +-
 .../framework/data/BatchDatasetTest.java      |  51 +-
 .../framework/data/DatasetIteratorTest.java   |  18 +-
 .../framework/data/MapDatasetTest.java        |  20 +-
 .../framework/data/SkipDatasetTest.java       |   9 +-
 .../framework/data/TakeDatasetTest.java       |   9 +-
 1161 files changed, 7800 insertions(+), 7760 deletions(-)
 create mode 100644 tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/AbstractTensor.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/RawTensor.java

diff --git a/ndarray/src/main/java/org/tensorflow/ndarray/NdArrays.java b/ndarray/src/main/java/org/tensorflow/ndarray/NdArrays.java
index 21b33402e98..8ad55cae7ed 100644
--- a/ndarray/src/main/java/org/tensorflow/ndarray/NdArrays.java
+++ b/ndarray/src/main/java/org/tensorflow/ndarray/NdArrays.java
@@ -65,7 +65,7 @@ public static ByteNdArray vectorOf(byte... values) {
     if (values == null) {
       throw new IllegalArgumentException("Values cannot be null");
     }
-    return wrap(DataBuffers.of(values, false, false), Shape.of(values.length));
+    return wrap(Shape.of(values.length), DataBuffers.of(values, false, false));
   }
 
   /**
@@ -81,19 +81,19 @@ public static ByteNdArray ofBytes(Shape shape) {
     if (shape == null) {
       throw new IllegalArgumentException("Shape cannot be null");
     }
-    return wrap(DataBuffers.ofBytes(shape.size()), shape);
+    return wrap(shape, DataBuffers.ofBytes(shape.size()));
   }
 
   /**
    * Wraps a buffer in a byte N-dimensional array of a given shape.
    *
-   * @param buffer buffer to wrap
    * @param shape shape of the array
+   * @param buffer buffer to wrap
    * @return new byte N-dimensional array
    * @throws IllegalArgumentException if shape is null, has unknown dimensions or has size bigger
    *                                  in the buffer size
    */
-  public static ByteNdArray wrap(ByteDataBuffer buffer, Shape shape) {
+  public static ByteNdArray wrap(Shape shape, ByteDataBuffer buffer) {
     return ByteDenseNdArray.create(buffer, shape);
   }
 
diff --git a/tensorflow-core/tensorflow-core-api/src/bazel/op_generator/op_generator.cc b/tensorflow-core/tensorflow-core-api/src/bazel/op_generator/op_generator.cc
index b58b3d3a59f..667ce463d26 100644
--- a/tensorflow-core/tensorflow-core-api/src/bazel/op_generator/op_generator.cc
+++ b/tensorflow-core/tensorflow-core-api/src/bazel/op_generator/op_generator.cc
@@ -237,12 +237,12 @@ void RenderFactoryMethods(const OpSpec& op, const Type& op_class,
   writer->EndLine();
   for (const ArgumentSpec& input : op.inputs()) {
     if (input.iterable()) {
-      writer->Append("opBuilder.addInputList(Operands.asOutputs(" +
+      writer->Append("opBuilder.addInputList(Operands.asOutputs(scope, " +
                      input.var().name() + "));");
       writer->EndLine();
     } else {
       writer->Append("opBuilder.addInput(" + input.var().name() +
-                     ".asOutput());");
+                     ".asOutput(scope));");
       writer->EndLine();
     }
   }
@@ -348,8 +348,9 @@ void RenderInterfaceImpl(const OpSpec& op, RenderMode mode,
   if (mode == OPERAND) {
     bool cast2obj = output.type().wildcard();
     Type return_type = Type::Class("Output", "org.tensorflow")
-        .add_parameter(cast2obj ? Type::Interface("Tensor", "org.tensorflow") : output.type());
+        .add_parameter(cast2obj ? Type::Interface("TType", "org.tensorflow.types.family") : output.type());
     Method as_output = Method::Create("asOutput", return_type)
+        .add_argument(Variable::Create("scope", Type::Class("Scope", "org.tensorflow.op")))
         .add_annotation(Annotation::Create("Override"));
     if (cast2obj) {
       as_output.add_annotation(
@@ -366,7 +367,7 @@ void RenderInterfaceImpl(const OpSpec& op, RenderMode mode,
   } else if (mode == LIST_OPERAND) {
     Type operand = Type::Interface("Operand", "org.tensorflow");
     if (output.type().wildcard()) {
-      operand.add_parameter(Type::Interface("Tensor", "org.tensorflow"));
+      operand.add_parameter(Type::Interface("TType", "org.tensorflow.types.family"));
     } else {
       operand.add_parameter(output.type());
     }
@@ -430,7 +431,7 @@ void GenerateOp(const OpSpec& op, const EndpointSpec& endpoint,
   RenderMode mode = DEFAULT;
   if (op.outputs().size() == 1) {
     const ArgumentSpec& output = op.outputs().front();
-    Type operand_type(output.type().wildcard() ? Type::Interface("Tensor", "org.tensorflow")
+    Type operand_type(output.type().wildcard() ? Type::Interface("TType", "org.tensorflow.types.family")
                                                : output.type());
     Type operand_inf(Type::Interface("Operand", "org.tensorflow").add_parameter(operand_type));
     if (output.iterable()) {
diff --git a/tensorflow-core/tensorflow-core-api/src/bazel/op_generator/op_specs.cc b/tensorflow-core/tensorflow-core-api/src/bazel/op_generator/op_specs.cc
index 3c981bf8816..7906613569a 100644
--- a/tensorflow-core/tensorflow-core-api/src/bazel/op_generator/op_specs.cc
+++ b/tensorflow-core/tensorflow-core-api/src/bazel/op_generator/op_specs.cc
@@ -65,6 +65,13 @@ class TypeResolver {
   std::pair<Type, Type> TypesOf(const OpDef_AttrDef& attr_def,
                                 bool* iterable_out);
 
+  // Returns the highest type family this attribute is part of
+  //
+  // For example, if the attribute is of type 'bool', the base 'TType' family
+  // returned. But if it represents a number, like a float or an integer,
+  // then 'TNumber' (which supersedes 'TType') is returned.
+  Type FamilyOf(const OpDef_AttrDef& attr_def);
+
   // Returns true if the type of this attribute has already been resolved
   bool IsAttributeVisited(const string& attr_name) {
     return visited_attrs_.find(attr_name) != visited_attrs_.cend();
@@ -81,13 +88,12 @@ class TypeResolver {
   std::pair<Type, Type> MakeTypePair(const Type& type) {
     return std::make_pair(type, type);
   }
-  Type NextGeneric() {
+  Type NextGeneric(const Type& typeFamily) {
     char generic_letter = next_generic_letter_++;
     if (next_generic_letter_ > 'Z') {
       next_generic_letter_ = 'A';
     }
-    return Type::Generic(string(1, generic_letter))
-        .add_supertype(Type::Interface("Tensor", "org.tensorflow"));
+    return Type::Generic(string(1, generic_letter)).add_supertype(typeFamily);
   }
 };
 
@@ -156,10 +162,7 @@ std::pair<Type, Type> TypeResolver::TypesOf(const OpDef_AttrDef& attr_def,
                              .add_parameter(Type::Wildcard()));
 
   } else if (attr_type == "type") {
-    Type type = *iterable_out ? Type::Wildcard() : NextGeneric();
-    if (IsRealNumbers(attr_def.allowed_values())) {
-      type.add_supertype(Type::Interface("TNumber", "org.tensorflow.types.family"));
-    }
+    Type type = *iterable_out ? Type::Wildcard() : NextGeneric(FamilyOf(attr_def));
     types = MakeTypePair(type, Type::Enum("DataType", "org.tensorflow"));
 
   } else {
@@ -170,6 +173,14 @@ std::pair<Type, Type> TypeResolver::TypesOf(const OpDef_AttrDef& attr_def,
   return types;
 }
 
+Type TypeResolver::FamilyOf(const OpDef_AttrDef& attr_def) {
+  // TODO (karlllessard): add more type families
+  if (IsRealNumbers(attr_def.allowed_values())) {
+    return Type::Interface("TNumber", "org.tensorflow.types.family");
+  }
+  return Type::Interface("TType", "org.tensorflow.types.family");
+}
+
 string SnakeToCamelCase(const string& str, bool upper = false) {
   string result;
   bool cap = upper;
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/BitwiseOps.java b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/BitwiseOps.java
index 9cc4506b77a..8ac6d565e51 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/BitwiseOps.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/BitwiseOps.java
@@ -18,7 +18,6 @@
 package org.tensorflow.op;
 
 import org.tensorflow.Operand;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.bitwise.BitwiseAnd;
 import org.tensorflow.op.bitwise.BitwiseOr;
 import org.tensorflow.op.bitwise.BitwiseXor;
@@ -66,7 +65,7 @@ public final class BitwiseOps {
    * @param y
    * @return a new instance of BitwiseAnd
    */
-  public <T extends Tensor & TNumber> BitwiseAnd<T> bitwiseAnd(Operand<T> x, Operand<T> y) {
+  public <T extends TNumber> BitwiseAnd<T> bitwiseAnd(Operand<T> x, Operand<T> y) {
     return BitwiseAnd.create(scope, x, y);
   }
 
@@ -97,7 +96,7 @@ public <T extends Tensor & TNumber> BitwiseAnd<T> bitwiseAnd(Operand<T> x, Opera
    * @param y
    * @return a new instance of BitwiseOr
    */
-  public <T extends Tensor & TNumber> BitwiseOr<T> bitwiseOr(Operand<T> x, Operand<T> y) {
+  public <T extends TNumber> BitwiseOr<T> bitwiseOr(Operand<T> x, Operand<T> y) {
     return BitwiseOr.create(scope, x, y);
   }
 
@@ -128,7 +127,7 @@ public <T extends Tensor & TNumber> BitwiseOr<T> bitwiseOr(Operand<T> x, Operand
    * @param y
    * @return a new instance of BitwiseXor
    */
-  public <T extends Tensor & TNumber> BitwiseXor<T> bitwiseXor(Operand<T> x, Operand<T> y) {
+  public <T extends TNumber> BitwiseXor<T> bitwiseXor(Operand<T> x, Operand<T> y) {
     return BitwiseXor.create(scope, x, y);
   }
 
@@ -179,7 +178,7 @@ public <T extends Tensor & TNumber> BitwiseXor<T> bitwiseXor(Operand<T> x, Opera
    * @param x
    * @return a new instance of Invert
    */
-  public <T extends Tensor & TNumber> Invert<T> invert(Operand<T> x) {
+  public <T extends TNumber> Invert<T> invert(Operand<T> x) {
     return Invert.create(scope, x);
   }
 
@@ -221,7 +220,7 @@ public <T extends Tensor & TNumber> Invert<T> invert(Operand<T> x) {
    * @param y
    * @return a new instance of LeftShift
    */
-  public <T extends Tensor & TNumber> LeftShift<T> leftShift(Operand<T> x, Operand<T> y) {
+  public <T extends TNumber> LeftShift<T> leftShift(Operand<T> x, Operand<T> y) {
     return LeftShift.create(scope, x, y);
   }
 
@@ -266,7 +265,7 @@ public <T extends Tensor & TNumber> LeftShift<T> leftShift(Operand<T> x, Operand
    * @param y
    * @return a new instance of RightShift
    */
-  public <T extends Tensor & TNumber> RightShift<T> rightShift(Operand<T> x, Operand<T> y) {
+  public <T extends TNumber> RightShift<T> rightShift(Operand<T> x, Operand<T> y) {
     return RightShift.create(scope, x, y);
   }
 }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/DtypesOps.java b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/DtypesOps.java
index 92f4466f181..16d571a6428 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/DtypesOps.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/DtypesOps.java
@@ -19,11 +19,11 @@
 
 import org.tensorflow.DataType;
 import org.tensorflow.Operand;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.dtypes.AsString;
 import org.tensorflow.op.dtypes.Cast;
 import org.tensorflow.op.dtypes.Complex;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * An API for building {@code dtypes} operations as {@link Op Op}s
@@ -57,7 +57,7 @@ public final class DtypesOps {
    * @param options carries optional attributes values
    * @return a new instance of AsString
    */
-  public <T extends Tensor> AsString asString(Operand<T> input, AsString.Options... options) {
+  public <T extends TType> AsString asString(Operand<T> input, AsString.Options... options) {
     return AsString.create(scope, input, options);
   }
 
@@ -70,7 +70,7 @@ public <T extends Tensor> AsString asString(Operand<T> input, AsString.Options..
    * @param options carries optional attributes values
    * @return a new instance of Cast
    */
-  public <U extends Tensor, T extends Tensor> Cast<U> cast(Operand<T> x, DataType<U> DstT,
+  public <U extends TType, T extends TType> Cast<U> cast(Operand<T> x, DataType<U> DstT,
       Cast.Options... options) {
     return Cast.create(scope, x, DstT, options);
   }
@@ -98,8 +98,8 @@ public <U extends Tensor, T extends Tensor> Cast<U> cast(Operand<T> x, DataType<
    * @param Tout
    * @return a new instance of Complex
    */
-  public <U extends Tensor, T extends Tensor & TNumber> Complex<U> complex(Operand<T> real,
-      Operand<T> imag, DataType<U> Tout) {
+  public <U extends TType, T extends TNumber> Complex<U> complex(Operand<T> real, Operand<T> imag,
+      DataType<U> Tout) {
     return Complex.create(scope, real, imag, Tout);
   }
 }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/ImageOps.java b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/ImageOps.java
index 0bcb966a2c6..eea2fc4b8f1 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/ImageOps.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/ImageOps.java
@@ -20,7 +20,6 @@
 import java.util.List;
 import org.tensorflow.DataType;
 import org.tensorflow.Operand;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.image.AdjustContrast;
 import org.tensorflow.op.image.AdjustHue;
 import org.tensorflow.op.image.AdjustSaturation;
@@ -57,6 +56,7 @@
 import org.tensorflow.types.TString;
 import org.tensorflow.types.TUint8;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * An API for building {@code image} operations as {@link Op Op}s
@@ -88,7 +88,7 @@ public final class ImageOps {
    * @param contrastFactor A float multiplier for adjusting contrast.
    * @return a new instance of AdjustContrast
    */
-  public <T extends Tensor & TNumber> AdjustContrast<T> adjustContrast(Operand<T> images,
+  public <T extends TNumber> AdjustContrast<T> adjustContrast(Operand<T> images,
       Operand<TFloat32> contrastFactor) {
     return AdjustContrast.create(scope, images, contrastFactor);
   }
@@ -108,8 +108,7 @@ public <T extends Tensor & TNumber> AdjustContrast<T> adjustContrast(Operand<T>
    * @param delta A float delta to add to the hue.
    * @return a new instance of AdjustHue
    */
-  public <T extends Tensor & TNumber> AdjustHue<T> adjustHue(Operand<T> images,
-      Operand<TFloat32> delta) {
+  public <T extends TNumber> AdjustHue<T> adjustHue(Operand<T> images, Operand<TFloat32> delta) {
     return AdjustHue.create(scope, images, delta);
   }
 
@@ -128,7 +127,7 @@ public <T extends Tensor & TNumber> AdjustHue<T> adjustHue(Operand<T> images,
    * @param scale A float scale to add to the saturation.
    * @return a new instance of AdjustSaturation
    */
-  public <T extends Tensor & TNumber> AdjustSaturation<T> adjustSaturation(Operand<T> images,
+  public <T extends TNumber> AdjustSaturation<T> adjustSaturation(Operand<T> images,
       Operand<TFloat32> scale) {
     return AdjustSaturation.create(scope, images, scale);
   }
@@ -212,9 +211,8 @@ public CombinedNonMaxSuppression combinedNonMaxSuppression(Operand<TFloat32> box
    * @param options carries optional attributes values
    * @return a new instance of CropAndResize
    */
-  public <T extends Tensor & TNumber> CropAndResize cropAndResize(Operand<T> image,
-      Operand<TFloat32> boxes, Operand<TInt32> boxInd, Operand<TInt32> cropSize,
-      CropAndResize.Options... options) {
+  public <T extends TNumber> CropAndResize cropAndResize(Operand<T> image, Operand<TFloat32> boxes,
+      Operand<TInt32> boxInd, Operand<TInt32> cropSize, CropAndResize.Options... options) {
     return CropAndResize.create(scope, image, boxes, boxInd, cropSize, options);
   }
 
@@ -239,8 +237,8 @@ public <T extends Tensor & TNumber> CropAndResize cropAndResize(Operand<T> image
    * @param options carries optional attributes values
    * @return a new instance of CropAndResizeGradBoxes
    */
-  public <T extends Tensor & TNumber> CropAndResizeGradBoxes cropAndResizeGradBoxes(
-      Operand<TFloat32> grads, Operand<T> image, Operand<TFloat32> boxes, Operand<TInt32> boxInd,
+  public <T extends TNumber> CropAndResizeGradBoxes cropAndResizeGradBoxes(Operand<TFloat32> grads,
+      Operand<T> image, Operand<TFloat32> boxes, Operand<TInt32> boxInd,
       CropAndResizeGradBoxes.Options... options) {
     return CropAndResizeGradBoxes.create(scope, grads, image, boxes, boxInd, options);
   }
@@ -269,7 +267,7 @@ public <T extends Tensor & TNumber> CropAndResizeGradBoxes cropAndResizeGradBoxe
    * @param options carries optional attributes values
    * @return a new instance of CropAndResizeGradImage
    */
-  public <T extends Tensor & TNumber> CropAndResizeGradImage<T> cropAndResizeGradImage(
+  public <T extends TNumber> CropAndResizeGradImage<T> cropAndResizeGradImage(
       Operand<TFloat32> grads, Operand<TFloat32> boxes, Operand<TInt32> boxInd,
       Operand<TInt32> imageSize, DataType<T> T, CropAndResizeGradImage.Options... options) {
     return CropAndResizeGradImage.create(scope, grads, boxes, boxInd, imageSize, T, options);
@@ -462,8 +460,8 @@ public DecodePng<TUint8> decodePng(Operand<TString> contents, DecodePng.Options.
    * @param options carries optional attributes values
    * @return a new instance of DecodePng
    */
-  public <T extends Tensor & TNumber> DecodePng<T> decodePng(Operand<TString> contents,
-      DataType<T> dtype, DecodePng.Options... options) {
+  public <T extends TNumber> DecodePng<T> decodePng(Operand<TString> contents, DataType<T> dtype,
+      DecodePng.Options... options) {
     return DecodePng.create(scope, contents, dtype, options);
   }
 
@@ -489,7 +487,7 @@ public <T extends Tensor & TNumber> DecodePng<T> decodePng(Operand<TString> cont
    * @param colors 2-D. A list of RGBA colors to cycle through for the boxes.
    * @return a new instance of DrawBoundingBoxes
    */
-  public <T extends Tensor & TNumber> DrawBoundingBoxes<T> drawBoundingBoxes(Operand<T> images,
+  public <T extends TNumber> DrawBoundingBoxes<T> drawBoundingBoxes(Operand<T> images,
       Operand<TFloat32> boxes, Operand<TFloat32> colors) {
     return DrawBoundingBoxes.create(scope, images, boxes, colors);
   }
@@ -573,8 +571,7 @@ public EncodeJpegVariableQuality encodeJpegVariableQuality(Operand<TUint8> image
    * @param options carries optional attributes values
    * @return a new instance of EncodePng
    */
-  public <T extends Tensor & TNumber> EncodePng encodePng(Operand<T> image,
-      EncodePng.Options... options) {
+  public <T extends TNumber> EncodePng encodePng(Operand<T> image, EncodePng.Options... options) {
     return EncodePng.create(scope, image, options);
   }
 
@@ -595,7 +592,7 @@ public <T extends Tensor & TNumber> EncodePng encodePng(Operand<T> image,
    * @param padding The type of padding algorithm to use.
    * @return a new instance of ExtractImagePatches
    */
-  public <T extends Tensor> ExtractImagePatches<T> extractImagePatches(Operand<T> images,
+  public <T extends TType> ExtractImagePatches<T> extractImagePatches(Operand<T> images,
       List<Long> ksizes, List<Long> strides, List<Long> rates, String padding) {
     return ExtractImagePatches.create(scope, images, ksizes, strides, rates, padding);
   }
@@ -624,8 +621,8 @@ public ExtractJpegShape<TInt32> extractJpegShape(Operand<TString> contents) {
    *  Defaults to int32.
    * @return a new instance of ExtractJpegShape
    */
-  public <T extends Tensor & TNumber> ExtractJpegShape<T> extractJpegShape(
-      Operand<TString> contents, DataType<T> outputType) {
+  public <T extends TNumber> ExtractJpegShape<T> extractJpegShape(Operand<TString> contents,
+      DataType<T> outputType) {
     return ExtractJpegShape.create(scope, contents, outputType);
   }
 
@@ -642,7 +639,7 @@ public <T extends Tensor & TNumber> ExtractJpegShape<T> extractJpegShape(
    * @param images 1-D or higher rank. HSV data to convert. Last dimension must be size 3.
    * @return a new instance of HsvToRgb
    */
-  public <T extends Tensor & TNumber> HsvToRgb<T> hsvToRgb(Operand<T> images) {
+  public <T extends TNumber> HsvToRgb<T> hsvToRgb(Operand<T> images) {
     return HsvToRgb.create(scope, images);
   }
 
@@ -688,7 +685,7 @@ public <T extends Tensor & TNumber> HsvToRgb<T> hsvToRgb(Operand<T> images) {
    * @param options carries optional attributes values
    * @return a new instance of NonMaxSuppression
    */
-  public <T extends Tensor & TNumber> NonMaxSuppression<T> nonMaxSuppression(Operand<T> boxes,
+  public <T extends TNumber> NonMaxSuppression<T> nonMaxSuppression(Operand<T> boxes,
       Operand<T> scores, Operand<TInt32> maxOutputSize, Operand<T> iouThreshold,
       Operand<T> scoreThreshold, Operand<T> softNmsSigma, NonMaxSuppression.Options... options) {
     return NonMaxSuppression.create(scope, boxes, scores, maxOutputSize, iouThreshold, scoreThreshold, softNmsSigma, options);
@@ -744,7 +741,7 @@ public NonMaxSuppressionWithOverlaps nonMaxSuppressionWithOverlaps(Operand<TFloa
    * @param options carries optional attributes values
    * @return a new instance of QuantizedResizeBilinear
    */
-  public <T extends Tensor> QuantizedResizeBilinear<T> quantizedResizeBilinear(Operand<T> images,
+  public <T extends TType> QuantizedResizeBilinear<T> quantizedResizeBilinear(Operand<T> images,
       Operand<TInt32> size, Operand<TFloat32> min, Operand<TFloat32> max,
       QuantizedResizeBilinear.Options... options) {
     return QuantizedResizeBilinear.create(scope, images, size, min, max, options);
@@ -766,8 +763,8 @@ public <T extends Tensor> QuantizedResizeBilinear<T> quantizedResizeBilinear(Ope
    * @param options carries optional attributes values
    * @return a new instance of RandomCrop
    */
-  public <T extends Tensor & TNumber> RandomCrop<T> randomCrop(Operand<T> image,
-      Operand<TInt64> size, RandomCrop.Options... options) {
+  public <T extends TNumber> RandomCrop<T> randomCrop(Operand<T> image, Operand<TInt64> size,
+      RandomCrop.Options... options) {
     return RandomCrop.create(scope, image, size, options);
   }
 
@@ -792,7 +789,7 @@ public <T extends Tensor & TNumber> RandomCrop<T> randomCrop(Operand<T> image,
    * @param options carries optional attributes values
    * @return a new instance of ResizeArea
    */
-  public <T extends Tensor & TNumber> ResizeArea resizeArea(Operand<T> images, Operand<TInt32> size,
+  public <T extends TNumber> ResizeArea resizeArea(Operand<T> images, Operand<TInt32> size,
       ResizeArea.Options... options) {
     return ResizeArea.create(scope, images, size, options);
   }
@@ -808,8 +805,8 @@ public <T extends Tensor & TNumber> ResizeArea resizeArea(Operand<T> images, Ope
    * @param options carries optional attributes values
    * @return a new instance of ResizeBicubic
    */
-  public <T extends Tensor & TNumber> ResizeBicubic resizeBicubic(Operand<T> images,
-      Operand<TInt32> size, ResizeBicubic.Options... options) {
+  public <T extends TNumber> ResizeBicubic resizeBicubic(Operand<T> images, Operand<TInt32> size,
+      ResizeBicubic.Options... options) {
     return ResizeBicubic.create(scope, images, size, options);
   }
 
@@ -824,8 +821,8 @@ public <T extends Tensor & TNumber> ResizeBicubic resizeBicubic(Operand<T> image
    * @param options carries optional attributes values
    * @return a new instance of ResizeBilinear
    */
-  public <T extends Tensor & TNumber> ResizeBilinear resizeBilinear(Operand<T> images,
-      Operand<TInt32> size, ResizeBilinear.Options... options) {
+  public <T extends TNumber> ResizeBilinear resizeBilinear(Operand<T> images, Operand<TInt32> size,
+      ResizeBilinear.Options... options) {
     return ResizeBilinear.create(scope, images, size, options);
   }
 
@@ -839,8 +836,8 @@ public <T extends Tensor & TNumber> ResizeBilinear resizeBilinear(Operand<T> ima
    * @param options carries optional attributes values
    * @return a new instance of ResizeNearestNeighbor
    */
-  public <T extends Tensor & TNumber> ResizeNearestNeighbor<T> resizeNearestNeighbor(
-      Operand<T> images, Operand<TInt32> size, ResizeNearestNeighbor.Options... options) {
+  public <T extends TNumber> ResizeNearestNeighbor<T> resizeNearestNeighbor(Operand<T> images,
+      Operand<TInt32> size, ResizeNearestNeighbor.Options... options) {
     return ResizeNearestNeighbor.create(scope, images, size, options);
   }
 
@@ -870,7 +867,7 @@ public <T extends Tensor & TNumber> ResizeNearestNeighbor<T> resizeNearestNeighb
    * @param images 1-D or higher rank. RGB data to convert. Last dimension must be size 3.
    * @return a new instance of RgbToHsv
    */
-  public <T extends Tensor & TNumber> RgbToHsv<T> rgbToHsv(Operand<T> images) {
+  public <T extends TNumber> RgbToHsv<T> rgbToHsv(Operand<T> images) {
     return RgbToHsv.create(scope, images);
   }
 
@@ -925,7 +922,7 @@ public <T extends Tensor & TNumber> RgbToHsv<T> rgbToHsv(Operand<T> images) {
    * @param options carries optional attributes values
    * @return a new instance of SampleDistortedBoundingBox
    */
-  public <T extends Tensor & TNumber> SampleDistortedBoundingBox<T> sampleDistortedBoundingBox(
+  public <T extends TNumber> SampleDistortedBoundingBox<T> sampleDistortedBoundingBox(
       Operand<T> imageSize, Operand<TFloat32> boundingBoxes, Operand<TFloat32> minObjectCovered,
       SampleDistortedBoundingBox.Options... options) {
     return SampleDistortedBoundingBox.create(scope, imageSize, boundingBoxes, minObjectCovered, options);
@@ -940,7 +937,7 @@ public <T extends Tensor & TNumber> SampleDistortedBoundingBox<T> sampleDistorte
    * @param options carries optional attributes values
    * @return a new instance of ScaleAndTranslate
    */
-  public <T extends Tensor & TNumber> ScaleAndTranslate scaleAndTranslate(Operand<T> images,
+  public <T extends TNumber> ScaleAndTranslate scaleAndTranslate(Operand<T> images,
       Operand<TInt32> size, Operand<TFloat32> scale, Operand<TFloat32> translation,
       ScaleAndTranslate.Options... options) {
     return ScaleAndTranslate.create(scope, images, size, scale, translation, options);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/IoOps.java b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/IoOps.java
index 3f164450780..adc656dc5af 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/IoOps.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/IoOps.java
@@ -20,7 +20,6 @@
 import java.util.List;
 import org.tensorflow.DataType;
 import org.tensorflow.Operand;
-import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.io.DecodeBase64;
 import org.tensorflow.op.io.DecodeCompressed;
@@ -73,6 +72,7 @@
 import org.tensorflow.types.TInt64;
 import org.tensorflow.types.TString;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * An API for building {@code io} operations as {@link Op Op}s
@@ -167,9 +167,8 @@ public DecodeJsonExample decodeJsonExample(Operand<TString> jsonExamples) {
    * @param options carries optional attributes values
    * @return a new instance of DecodePaddedRaw
    */
-  public <T extends Tensor & TNumber> DecodePaddedRaw<T> decodePaddedRaw(
-      Operand<TString> inputBytes, Operand<TInt32> fixedLength, DataType<T> outType,
-      DecodePaddedRaw.Options... options) {
+  public <T extends TNumber> DecodePaddedRaw<T> decodePaddedRaw(Operand<TString> inputBytes,
+      Operand<TInt32> fixedLength, DataType<T> outType, DecodePaddedRaw.Options... options) {
     return DecodePaddedRaw.create(scope, inputBytes, fixedLength, outType, options);
   }
 
@@ -182,7 +181,7 @@ public <T extends Tensor & TNumber> DecodePaddedRaw<T> decodePaddedRaw(
    * @param options carries optional attributes values
    * @return a new instance of DecodeRaw
    */
-  public <T extends Tensor> DecodeRaw<T> decodeRaw(Operand<TString> bytes, DataType<T> outType,
+  public <T extends TType> DecodeRaw<T> decodeRaw(Operand<TString> bytes, DataType<T> outType,
       DecodeRaw.Options... options) {
     return DecodeRaw.create(scope, bytes, outType, options);
   }
@@ -238,7 +237,7 @@ public <T extends Tensor> DecodeRaw<T> decodeRaw(Operand<TString> bytes, DataTyp
    * @param dtype The `dtype` of the serialized `SparseTensor` objects.
    * @return a new instance of DeserializeManySparse
    */
-  public <T extends Tensor> DeserializeManySparse<T> deserializeManySparse(
+  public <T extends TType> DeserializeManySparse<T> deserializeManySparse(
       Operand<TString> serializedSparse, DataType<T> dtype) {
     return DeserializeManySparse.create(scope, serializedSparse, dtype);
   }
@@ -566,7 +565,7 @@ public ParseSingleSequenceExample parseSingleSequenceExample(Operand<TString> se
    *  type of the serialized tensor and no implicit conversion will take place.
    * @return a new instance of ParseTensor
    */
-  public <T extends Tensor> ParseTensor<T> parseTensor(Operand<TString> serialized,
+  public <T extends TType> ParseTensor<T> parseTensor(Operand<TString> serialized,
       DataType<T> outType) {
     return ParseTensor.create(scope, serialized, outType);
   }
@@ -889,7 +888,7 @@ public ReaderSerializeState readerSerializeState(Operand<?> readerHandle) {
    * @param sparseShape 1-D.  The `shape` of the minibatch `SparseTensor`.
    * @return a new instance of SerializeManySparse
    */
-  public <T extends Tensor> SerializeManySparse<TString> serializeManySparse(
+  public <T extends TType> SerializeManySparse<TString> serializeManySparse(
       Operand<TInt64> sparseIndices, Operand<T> sparseValues, Operand<TInt64> sparseShape) {
     return SerializeManySparse.create(scope, sparseIndices, sparseValues, sparseShape);
   }
@@ -913,7 +912,7 @@ public <T extends Tensor> SerializeManySparse<TString> serializeManySparse(
    *  (default) and `variant`.
    * @return a new instance of SerializeManySparse
    */
-  public <U extends Tensor, T extends Tensor> SerializeManySparse<U> serializeManySparse(
+  public <U extends TType, T extends TType> SerializeManySparse<U> serializeManySparse(
       Operand<TInt64> sparseIndices, Operand<T> sparseValues, Operand<TInt64> sparseShape,
       DataType<U> outType) {
     return SerializeManySparse.create(scope, sparseIndices, sparseValues, sparseShape, outType);
@@ -928,7 +927,7 @@ public <U extends Tensor, T extends Tensor> SerializeManySparse<U> serializeMany
    * @param sparseShape 1-D.  The `shape` of the `SparseTensor`.
    * @return a new instance of SerializeSparse
    */
-  public <T extends Tensor> SerializeSparse<TString> serializeSparse(Operand<TInt64> sparseIndices,
+  public <T extends TType> SerializeSparse<TString> serializeSparse(Operand<TInt64> sparseIndices,
       Operand<T> sparseValues, Operand<TInt64> sparseShape) {
     return SerializeSparse.create(scope, sparseIndices, sparseValues, sparseShape);
   }
@@ -944,7 +943,7 @@ public <T extends Tensor> SerializeSparse<TString> serializeSparse(Operand<TInt6
    *  (default) and `variant`.
    * @return a new instance of SerializeSparse
    */
-  public <U extends Tensor, T extends Tensor> SerializeSparse<U> serializeSparse(
+  public <U extends TType, T extends TType> SerializeSparse<U> serializeSparse(
       Operand<TInt64> sparseIndices, Operand<T> sparseValues, Operand<TInt64> sparseShape,
       DataType<U> outType) {
     return SerializeSparse.create(scope, sparseIndices, sparseValues, sparseShape, outType);
@@ -956,7 +955,7 @@ public <U extends Tensor, T extends Tensor> SerializeSparse<U> serializeSparse(
    * @param tensor A Tensor of type `T`.
    * @return a new instance of SerializeTensor
    */
-  public <T extends Tensor> SerializeTensor serializeTensor(Operand<T> tensor) {
+  public <T extends TType> SerializeTensor serializeTensor(Operand<T> tensor) {
     return SerializeTensor.create(scope, tensor);
   }
 
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/LinalgOps.java b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/LinalgOps.java
index 7002747a317..b2242f1068e 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/LinalgOps.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/LinalgOps.java
@@ -19,7 +19,6 @@
 
 import org.tensorflow.DataType;
 import org.tensorflow.Operand;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.linalg.BandPart;
 import org.tensorflow.op.linalg.BatchCholesky;
 import org.tensorflow.op.linalg.BatchCholeskyGrad;
@@ -69,6 +68,7 @@
 import org.tensorflow.types.TInt64;
 import org.tensorflow.types.TString;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * An API for building {@code linalg} operations as {@link Op Op}s
@@ -128,7 +128,7 @@ public final class LinalgOps {
    *  entire upper triangle.
    * @return a new instance of BandPart
    */
-  public <T extends Tensor, U extends Tensor & TNumber> BandPart<T> bandPart(Operand<T> input,
+  public <T extends TType, U extends TNumber> BandPart<T> bandPart(Operand<T> input,
       Operand<U> numLower, Operand<U> numUpper) {
     return BandPart.create(scope, input, numLower, numUpper);
   }
@@ -139,7 +139,7 @@ public <T extends Tensor, U extends Tensor & TNumber> BandPart<T> bandPart(Opera
    * @param input
    * @return a new instance of BatchCholesky
    */
-  public <T extends Tensor & TNumber> BatchCholesky<T> batchCholesky(Operand<T> input) {
+  public <T extends TNumber> BatchCholesky<T> batchCholesky(Operand<T> input) {
     return BatchCholesky.create(scope, input);
   }
 
@@ -150,8 +150,7 @@ public <T extends Tensor & TNumber> BatchCholesky<T> batchCholesky(Operand<T> in
    * @param grad
    * @return a new instance of BatchCholeskyGrad
    */
-  public <T extends Tensor & TNumber> BatchCholeskyGrad<T> batchCholeskyGrad(Operand<T> l,
-      Operand<T> grad) {
+  public <T extends TNumber> BatchCholeskyGrad<T> batchCholeskyGrad(Operand<T> l, Operand<T> grad) {
     return BatchCholeskyGrad.create(scope, l, grad);
   }
 
@@ -163,7 +162,7 @@ public <T extends Tensor & TNumber> BatchCholeskyGrad<T> batchCholeskyGrad(Opera
    * @param numUpper
    * @return a new instance of BatchMatrixBandPart
    */
-  public <T extends Tensor> BatchMatrixBandPart<T> batchMatrixBandPart(Operand<T> input,
+  public <T extends TType> BatchMatrixBandPart<T> batchMatrixBandPart(Operand<T> input,
       Operand<TInt64> numLower, Operand<TInt64> numUpper) {
     return BatchMatrixBandPart.create(scope, input, numLower, numUpper);
   }
@@ -174,7 +173,7 @@ public <T extends Tensor> BatchMatrixBandPart<T> batchMatrixBandPart(Operand<T>
    * @param input
    * @return a new instance of BatchMatrixDeterminant
    */
-  public <T extends Tensor> BatchMatrixDeterminant<T> batchMatrixDeterminant(Operand<T> input) {
+  public <T extends TType> BatchMatrixDeterminant<T> batchMatrixDeterminant(Operand<T> input) {
     return BatchMatrixDeterminant.create(scope, input);
   }
 
@@ -184,7 +183,7 @@ public <T extends Tensor> BatchMatrixDeterminant<T> batchMatrixDeterminant(Opera
    * @param diagonal
    * @return a new instance of BatchMatrixDiag
    */
-  public <T extends Tensor> BatchMatrixDiag<T> batchMatrixDiag(Operand<T> diagonal) {
+  public <T extends TType> BatchMatrixDiag<T> batchMatrixDiag(Operand<T> diagonal) {
     return BatchMatrixDiag.create(scope, diagonal);
   }
 
@@ -194,7 +193,7 @@ public <T extends Tensor> BatchMatrixDiag<T> batchMatrixDiag(Operand<T> diagonal
    * @param input
    * @return a new instance of BatchMatrixDiagPart
    */
-  public <T extends Tensor> BatchMatrixDiagPart<T> batchMatrixDiagPart(Operand<T> input) {
+  public <T extends TType> BatchMatrixDiagPart<T> batchMatrixDiagPart(Operand<T> input) {
     return BatchMatrixDiagPart.create(scope, input);
   }
 
@@ -205,7 +204,7 @@ public <T extends Tensor> BatchMatrixDiagPart<T> batchMatrixDiagPart(Operand<T>
    * @param options carries optional attributes values
    * @return a new instance of BatchMatrixInverse
    */
-  public <T extends Tensor & TNumber> BatchMatrixInverse<T> batchMatrixInverse(Operand<T> input,
+  public <T extends TNumber> BatchMatrixInverse<T> batchMatrixInverse(Operand<T> input,
       BatchMatrixInverse.Options... options) {
     return BatchMatrixInverse.create(scope, input, options);
   }
@@ -217,7 +216,7 @@ public <T extends Tensor & TNumber> BatchMatrixInverse<T> batchMatrixInverse(Ope
    * @param diagonal
    * @return a new instance of BatchMatrixSetDiag
    */
-  public <T extends Tensor> BatchMatrixSetDiag<T> batchMatrixSetDiag(Operand<T> input,
+  public <T extends TType> BatchMatrixSetDiag<T> batchMatrixSetDiag(Operand<T> input,
       Operand<T> diagonal) {
     return BatchMatrixSetDiag.create(scope, input, diagonal);
   }
@@ -230,8 +229,8 @@ public <T extends Tensor> BatchMatrixSetDiag<T> batchMatrixSetDiag(Operand<T> in
    * @param options carries optional attributes values
    * @return a new instance of BatchMatrixSolve
    */
-  public <T extends Tensor & TNumber> BatchMatrixSolve<T> batchMatrixSolve(Operand<T> matrix,
-      Operand<T> rhs, BatchMatrixSolve.Options... options) {
+  public <T extends TNumber> BatchMatrixSolve<T> batchMatrixSolve(Operand<T> matrix, Operand<T> rhs,
+      BatchMatrixSolve.Options... options) {
     return BatchMatrixSolve.create(scope, matrix, rhs, options);
   }
 
@@ -244,7 +243,7 @@ public <T extends Tensor & TNumber> BatchMatrixSolve<T> batchMatrixSolve(Operand
    * @param options carries optional attributes values
    * @return a new instance of BatchMatrixSolveLs
    */
-  public <T extends Tensor & TNumber> BatchMatrixSolveLs<T> batchMatrixSolveLs(Operand<T> matrix,
+  public <T extends TNumber> BatchMatrixSolveLs<T> batchMatrixSolveLs(Operand<T> matrix,
       Operand<T> rhs, Operand<TFloat64> l2Regularizer, BatchMatrixSolveLs.Options... options) {
     return BatchMatrixSolveLs.create(scope, matrix, rhs, l2Regularizer, options);
   }
@@ -257,7 +256,7 @@ public <T extends Tensor & TNumber> BatchMatrixSolveLs<T> batchMatrixSolveLs(Ope
    * @param options carries optional attributes values
    * @return a new instance of BatchMatrixTriangularSolve
    */
-  public <T extends Tensor & TNumber> BatchMatrixTriangularSolve<T> batchMatrixTriangularSolve(
+  public <T extends TNumber> BatchMatrixTriangularSolve<T> batchMatrixTriangularSolve(
       Operand<T> matrix, Operand<T> rhs, BatchMatrixTriangularSolve.Options... options) {
     return BatchMatrixTriangularSolve.create(scope, matrix, rhs, options);
   }
@@ -269,7 +268,7 @@ public <T extends Tensor & TNumber> BatchMatrixTriangularSolve<T> batchMatrixTri
    * @param options carries optional attributes values
    * @return a new instance of BatchSelfAdjointEig
    */
-  public <T extends Tensor & TNumber> BatchSelfAdjointEig<T> batchSelfAdjointEig(Operand<T> input,
+  public <T extends TNumber> BatchSelfAdjointEig<T> batchSelfAdjointEig(Operand<T> input,
       BatchSelfAdjointEig.Options... options) {
     return BatchSelfAdjointEig.create(scope, input, options);
   }
@@ -281,7 +280,7 @@ public <T extends Tensor & TNumber> BatchSelfAdjointEig<T> batchSelfAdjointEig(O
    * @param options carries optional attributes values
    * @return a new instance of BatchSvd
    */
-  public <T extends Tensor> BatchSvd<T> batchSvd(Operand<T> input, BatchSvd.Options... options) {
+  public <T extends TType> BatchSvd<T> batchSvd(Operand<T> input, BatchSvd.Options... options) {
     return BatchSvd.create(scope, input, options);
   }
 
@@ -306,7 +305,7 @@ public <T extends Tensor> BatchSvd<T> batchSvd(Operand<T> input, BatchSvd.Option
    * @param input Shape is `[..., M, M]`.
    * @return a new instance of Cholesky
    */
-  public <T extends Tensor> Cholesky<T> cholesky(Operand<T> input) {
+  public <T extends TType> Cholesky<T> cholesky(Operand<T> input) {
     return Cholesky.create(scope, input);
   }
 
@@ -325,7 +324,7 @@ public <T extends Tensor> Cholesky<T> cholesky(Operand<T> input) {
    *  this tensor.
    * @return a new instance of CholeskyGrad
    */
-  public <T extends Tensor & TNumber> CholeskyGrad<T> choleskyGrad(Operand<T> l, Operand<T> grad) {
+  public <T extends TNumber> CholeskyGrad<T> choleskyGrad(Operand<T> l, Operand<T> grad) {
     return CholeskyGrad.create(scope, l, grad);
   }
 
@@ -341,8 +340,8 @@ public <T extends Tensor & TNumber> CholeskyGrad<T> choleskyGrad(Operand<T> l, O
    * @param perm
    * @return a new instance of ConjugateTranspose
    */
-  public <T extends Tensor, U extends Tensor & TNumber> ConjugateTranspose<T> conjugateTranspose(
-      Operand<T> x, Operand<U> perm) {
+  public <T extends TType, U extends TNumber> ConjugateTranspose<T> conjugateTranspose(Operand<T> x,
+      Operand<U> perm) {
     return ConjugateTranspose.create(scope, x, perm);
   }
 
@@ -358,7 +357,7 @@ public <T extends Tensor, U extends Tensor & TNumber> ConjugateTranspose<T> conj
    * @param b Another tensor, of same type and shape as `a`.
    * @return a new instance of Cross
    */
-  public <T extends Tensor & TNumber> Cross<T> cross(Operand<T> a, Operand<T> b) {
+  public <T extends TNumber> Cross<T> cross(Operand<T> a, Operand<T> b) {
     return Cross.create(scope, a, b);
   }
 
@@ -373,7 +372,7 @@ public <T extends Tensor & TNumber> Cross<T> cross(Operand<T> a, Operand<T> b) {
    * @param input Shape is `[..., M, M]`.
    * @return a new instance of Det
    */
-  public <T extends Tensor> Det<T> det(Operand<T> input) {
+  public <T extends TType> Det<T> det(Operand<T> input) {
     return Det.create(scope, input);
   }
 
@@ -397,7 +396,7 @@ public <T extends Tensor> Det<T> det(Operand<T> input) {
    * @param options carries optional attributes values
    * @return a new instance of Eig
    */
-  public <U extends Tensor, T extends Tensor> Eig<U> eig(Operand<T> input, DataType<U> Tout,
+  public <U extends TType, T extends TType> Eig<U> eig(Operand<T> input, DataType<U> Tout,
       Eig.Options... options) {
     return Eig.create(scope, input, Tout, options);
   }
@@ -485,7 +484,7 @@ public <U extends Tensor, T extends Tensor> Eig<U> eig(Operand<T> input, DataTyp
    * @param equation String describing the Einstein Summation operation; in the format of np.einsum.
    * @return a new instance of Einsum
    */
-  public <T extends Tensor> Einsum<T> einsum(Iterable<Operand<T>> inputs, String equation) {
+  public <T extends TType> Einsum<T> einsum(Iterable<Operand<T>> inputs, String equation) {
     return Einsum.create(scope, inputs, equation);
   }
 
@@ -504,8 +503,8 @@ public <T extends Tensor> Einsum<T> einsum(Iterable<Operand<T>> inputs, String e
    * @param options carries optional attributes values
    * @return a new instance of EuclideanNorm
    */
-  public <T extends Tensor, U extends Tensor & TNumber> EuclideanNorm<T> euclideanNorm(
-      Operand<T> input, Operand<U> axis, EuclideanNorm.Options... options) {
+  public <T extends TType, U extends TNumber> EuclideanNorm<T> euclideanNorm(Operand<T> input,
+      Operand<U> axis, EuclideanNorm.Options... options) {
     return EuclideanNorm.create(scope, input, axis, options);
   }
 
@@ -529,7 +528,7 @@ public <T extends Tensor, U extends Tensor & TNumber> EuclideanNorm<T> euclidean
    * @param options carries optional attributes values
    * @return a new instance of Inv
    */
-  public <T extends Tensor> Inv<T> inv(Operand<T> input, Inv.Options... options) {
+  public <T extends TType> Inv<T> inv(Operand<T> input, Inv.Options... options) {
     return Inv.create(scope, input, options);
   }
 
@@ -619,7 +618,7 @@ public LoadAndRemapMatrix loadAndRemapMatrix(Operand<TString> ckptPath,
    * @param input Shape is `[N, M, M]`.
    * @return a new instance of LogMatrixDeterminant
    */
-  public <T extends Tensor> LogMatrixDeterminant<T> logMatrixDeterminant(Operand<T> input) {
+  public <T extends TType> LogMatrixDeterminant<T> logMatrixDeterminant(Operand<T> input) {
     return LogMatrixDeterminant.create(scope, input);
   }
 
@@ -650,7 +649,7 @@ public <T extends Tensor> LogMatrixDeterminant<T> logMatrixDeterminant(Operand<T
    *  size `[M, M]`.
    * @return a new instance of Lu
    */
-  public <T extends Tensor> Lu<T, TInt32> lu(Operand<T> input) {
+  public <T extends TType> Lu<T, TInt32> lu(Operand<T> input) {
     return Lu.create(scope, input);
   }
 
@@ -682,7 +681,7 @@ public <T extends Tensor> Lu<T, TInt32> lu(Operand<T> input) {
    * @param outputIdxType
    * @return a new instance of Lu
    */
-  public <T extends Tensor, U extends Tensor & TNumber> Lu<T, U> lu(Operand<T> input,
+  public <T extends TType, U extends TNumber> Lu<T, U> lu(Operand<T> input,
       DataType<U> outputIdxType) {
     return Lu.create(scope, input, outputIdxType);
   }
@@ -704,8 +703,7 @@ public <T extends Tensor, U extends Tensor & TNumber> Lu<T, U> lu(Operand<T> inp
    * @param options carries optional attributes values
    * @return a new instance of MatMul
    */
-  public <T extends Tensor> MatMul<T> matMul(Operand<T> a, Operand<T> b,
-      MatMul.Options... options) {
+  public <T extends TType> MatMul<T> matMul(Operand<T> a, Operand<T> b, MatMul.Options... options) {
     return MatMul.create(scope, a, b, options);
   }
 
@@ -812,7 +810,7 @@ public <T extends Tensor> MatMul<T> matMul(Operand<T> a, Operand<T> b,
    *  Default is 0.
    * @return a new instance of MatrixDiag
    */
-  public <T extends Tensor> MatrixDiag<T> matrixDiag(Operand<T> diagonal, Operand<TInt32> k,
+  public <T extends TType> MatrixDiag<T> matrixDiag(Operand<T> diagonal, Operand<TInt32> k,
       Operand<TInt32> numRows, Operand<TInt32> numCols, Operand<T> paddingValue) {
     return MatrixDiag.create(scope, diagonal, k, numRows, numCols, paddingValue);
   }
@@ -896,7 +894,7 @@ public <T extends Tensor> MatrixDiag<T> matrixDiag(Operand<T> diagonal, Operand<
    *  Default is 0.
    * @return a new instance of MatrixDiagPart
    */
-  public <T extends Tensor> MatrixDiagPart<T> matrixDiagPart(Operand<T> input, Operand<TInt32> k,
+  public <T extends TType> MatrixDiagPart<T> matrixDiagPart(Operand<T> input, Operand<TInt32> k,
       Operand<T> paddingValue) {
     return MatrixDiagPart.create(scope, input, k, paddingValue);
   }
@@ -1012,8 +1010,8 @@ public <T extends Tensor> MatrixDiagPart<T> matrixDiagPart(Operand<T> input, Ope
    * @param options carries optional attributes values
    * @return a new instance of MatrixDiagPartV3
    */
-  public <T extends Tensor> MatrixDiagPartV3<T> matrixDiagPartV3(Operand<T> input,
-      Operand<TInt32> k, Operand<T> paddingValue, MatrixDiagPartV3.Options... options) {
+  public <T extends TType> MatrixDiagPartV3<T> matrixDiagPartV3(Operand<T> input, Operand<TInt32> k,
+      Operand<T> paddingValue, MatrixDiagPartV3.Options... options) {
     return MatrixDiagPartV3.create(scope, input, k, paddingValue, options);
   }
 
@@ -1150,7 +1148,7 @@ public <T extends Tensor> MatrixDiagPartV3<T> matrixDiagPartV3(Operand<T> input,
    * @param options carries optional attributes values
    * @return a new instance of MatrixDiagV3
    */
-  public <T extends Tensor> MatrixDiagV3<T> matrixDiagV3(Operand<T> diagonal, Operand<TInt32> k,
+  public <T extends TType> MatrixDiagV3<T> matrixDiagV3(Operand<T> diagonal, Operand<TInt32> k,
       Operand<TInt32> numRows, Operand<TInt32> numCols, Operand<T> paddingValue,
       MatrixDiagV3.Options... options) {
     return MatrixDiagV3.create(scope, diagonal, k, numRows, numCols, paddingValue, options);
@@ -1271,7 +1269,7 @@ public <T extends Tensor> MatrixDiagV3<T> matrixDiagV3(Operand<T> diagonal, Oper
    * @param options carries optional attributes values
    * @return a new instance of MatrixSetDiag
    */
-  public <T extends Tensor> MatrixSetDiag<T> matrixSetDiag(Operand<T> input, Operand<T> diagonal,
+  public <T extends TType> MatrixSetDiag<T> matrixSetDiag(Operand<T> input, Operand<T> diagonal,
       Operand<TInt32> k, MatrixSetDiag.Options... options) {
     return MatrixSetDiag.create(scope, input, diagonal, k, options);
   }
@@ -1324,7 +1322,7 @@ public <T extends Tensor> MatrixSetDiag<T> matrixSetDiag(Operand<T> input, Opera
    * @param options carries optional attributes values
    * @return a new instance of MatrixSolveLs
    */
-  public <T extends Tensor> MatrixSolveLs<T> matrixSolveLs(Operand<T> matrix, Operand<T> rhs,
+  public <T extends TType> MatrixSolveLs<T> matrixSolveLs(Operand<T> matrix, Operand<T> rhs,
       Operand<TFloat64> l2Regularizer, MatrixSolveLs.Options... options) {
     return MatrixSolveLs.create(scope, matrix, rhs, l2Regularizer, options);
   }
@@ -1348,7 +1346,7 @@ public <T extends Tensor> MatrixSolveLs<T> matrixSolveLs(Operand<T> matrix, Oper
    * @param options carries optional attributes values
    * @return a new instance of Qr
    */
-  public <T extends Tensor> Qr<T> qr(Operand<T> input, Qr.Options... options) {
+  public <T extends TType> Qr<T> qr(Operand<T> input, Qr.Options... options) {
     return Qr.create(scope, input, options);
   }
 
@@ -1373,7 +1371,7 @@ public <T extends Tensor> Qr<T> qr(Operand<T> input, Qr.Options... options) {
    * @param options carries optional attributes values
    * @return a new instance of QuantizedMatMul
    */
-  public <V extends Tensor, T extends Tensor, U extends Tensor, W extends Tensor> QuantizedMatMul<V> quantizedMatMul(
+  public <V extends TType, T extends TType, U extends TType, W extends TType> QuantizedMatMul<V> quantizedMatMul(
       Operand<T> a, Operand<U> b, Operand<TFloat32> minA, Operand<TFloat32> maxA,
       Operand<TFloat32> minB, Operand<TFloat32> maxB, DataType<V> Toutput, DataType<W> Tactivation,
       QuantizedMatMul.Options... options) {
@@ -1399,7 +1397,7 @@ public <V extends Tensor, T extends Tensor, U extends Tensor, W extends Tensor>
    * @param options carries optional attributes values
    * @return a new instance of SelfAdjointEig
    */
-  public <T extends Tensor> SelfAdjointEig<T> selfAdjointEig(Operand<T> input,
+  public <T extends TType> SelfAdjointEig<T> selfAdjointEig(Operand<T> input,
       SelfAdjointEig.Options... options) {
     return SelfAdjointEig.create(scope, input, options);
   }
@@ -1420,7 +1418,7 @@ public <T extends Tensor> SelfAdjointEig<T> selfAdjointEig(Operand<T> input,
    * @param options carries optional attributes values
    * @return a new instance of Solve
    */
-  public <T extends Tensor> Solve<T> solve(Operand<T> matrix, Operand<T> rhs,
+  public <T extends TType> Solve<T> solve(Operand<T> matrix, Operand<T> rhs,
       Solve.Options... options) {
     return Solve.create(scope, matrix, rhs, options);
   }
@@ -1448,7 +1446,7 @@ public <T extends Tensor> Solve<T> solve(Operand<T> matrix, Operand<T> rhs,
    * @param input Shape is `[..., M, M]`.
    * @return a new instance of Sqrtm
    */
-  public <T extends Tensor> Sqrtm<T> sqrtm(Operand<T> input) {
+  public <T extends TType> Sqrtm<T> sqrtm(Operand<T> input) {
     return Sqrtm.create(scope, input);
   }
 
@@ -1472,7 +1470,7 @@ public <T extends Tensor> Sqrtm<T> sqrtm(Operand<T> input) {
    * @param options carries optional attributes values
    * @return a new instance of Svd
    */
-  public <T extends Tensor> Svd<T> svd(Operand<T> input, Svd.Options... options) {
+  public <T extends TType> Svd<T> svd(Operand<T> input, Svd.Options... options) {
     return Svd.create(scope, input, options);
   }
 
@@ -1500,7 +1498,7 @@ public <T extends Tensor> Svd<T> svd(Operand<T> input, Svd.Options... options) {
    * @param diagonal Rank k tensor where k is at most 1.
    * @return a new instance of TensorDiag
    */
-  public <T extends Tensor> TensorDiag<T> tensorDiag(Operand<T> diagonal) {
+  public <T extends TType> TensorDiag<T> tensorDiag(Operand<T> diagonal) {
     return TensorDiag.create(scope, diagonal);
   }
 
@@ -1529,7 +1527,7 @@ public <T extends Tensor> TensorDiag<T> tensorDiag(Operand<T> diagonal) {
    * @param input Rank k tensor where k is even and not zero.
    * @return a new instance of TensorDiagPart
    */
-  public <T extends Tensor> TensorDiagPart<T> tensorDiagPart(Operand<T> input) {
+  public <T extends TType> TensorDiagPart<T> tensorDiagPart(Operand<T> input) {
     return TensorDiagPart.create(scope, input);
   }
 
@@ -1544,7 +1542,7 @@ public <T extends Tensor> TensorDiagPart<T> tensorDiagPart(Operand<T> input) {
    * @param perm
    * @return a new instance of Transpose
    */
-  public <T extends Tensor, U extends Tensor & TNumber> Transpose<T> transpose(Operand<T> x,
+  public <T extends TType, U extends TNumber> Transpose<T> transpose(Operand<T> x,
       Operand<U> perm) {
     return Transpose.create(scope, x, perm);
   }
@@ -1604,7 +1602,7 @@ public <T extends Tensor, U extends Tensor & TNumber> Transpose<T> transpose(Ope
    * @param options carries optional attributes values
    * @return a new instance of TriangularSolve
    */
-  public <T extends Tensor> TriangularSolve<T> triangularSolve(Operand<T> matrix, Operand<T> rhs,
+  public <T extends TType> TriangularSolve<T> triangularSolve(Operand<T> matrix, Operand<T> rhs,
       TriangularSolve.Options... options) {
     return TriangularSolve.create(scope, matrix, rhs, options);
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/MathOps.java b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/MathOps.java
index 3c983272e6d..8da2c9ee3cd 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/MathOps.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/MathOps.java
@@ -19,7 +19,6 @@
 
 import org.tensorflow.DataType;
 import org.tensorflow.Operand;
-import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.math.Abs;
 import org.tensorflow.op.math.AccumulateN;
@@ -130,6 +129,7 @@
 import org.tensorflow.types.TInt32;
 import org.tensorflow.types.TInt64;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * An API for building {@code math} operations as {@link Op Op}s
@@ -154,7 +154,7 @@ public final class MathOps {
    * @param x
    * @return a new instance of Abs
    */
-  public <T extends Tensor & TNumber> Abs<T> abs(Operand<T> x) {
+  public <T extends TNumber> Abs<T> abs(Operand<T> x) {
     return Abs.create(scope, x);
   }
 
@@ -175,7 +175,7 @@ public <T extends Tensor & TNumber> Abs<T> abs(Operand<T> x) {
    * @param shape Shape of elements of `inputs`.
    * @return a new instance of AccumulateN
    */
-  public <T extends Tensor> AccumulateN<T> accumulateN(Iterable<Operand<T>> inputs, Shape shape) {
+  public <T extends TType> AccumulateN<T> accumulateN(Iterable<Operand<T>> inputs, Shape shape) {
     return AccumulateN.create(scope, inputs, shape);
   }
 
@@ -186,7 +186,7 @@ public <T extends Tensor> AccumulateN<T> accumulateN(Iterable<Operand<T>> inputs
    * @param x
    * @return a new instance of Acos
    */
-  public <T extends Tensor> Acos<T> acos(Operand<T> x) {
+  public <T extends TType> Acos<T> acos(Operand<T> x) {
     return Acos.create(scope, x);
   }
 
@@ -204,7 +204,7 @@ public <T extends Tensor> Acos<T> acos(Operand<T> x) {
    * @param x
    * @return a new instance of Acosh
    */
-  public <T extends Tensor> Acosh<T> acosh(Operand<T> x) {
+  public <T extends TType> Acosh<T> acosh(Operand<T> x) {
     return Acosh.create(scope, x);
   }
 
@@ -219,7 +219,7 @@ public <T extends Tensor> Acosh<T> acosh(Operand<T> x) {
    * @param y
    * @return a new instance of Add
    */
-  public <T extends Tensor> Add<T> add(Operand<T> x, Operand<T> y) {
+  public <T extends TType> Add<T> add(Operand<T> x, Operand<T> y) {
     return Add.create(scope, x, y);
   }
 
@@ -237,7 +237,7 @@ public <T extends Tensor> Add<T> add(Operand<T> x, Operand<T> y) {
    * @param inputs
    * @return a new instance of AddN
    */
-  public <T extends Tensor> AddN<T> addN(Iterable<Operand<T>> inputs) {
+  public <T extends TType> AddN<T> addN(Iterable<Operand<T>> inputs) {
     return AddN.create(scope, inputs);
   }
 
@@ -263,7 +263,7 @@ public <T extends Tensor> AddN<T> addN(Iterable<Operand<T>> inputs) {
    * @param input
    * @return a new instance of Angle
    */
-  public <T extends Tensor> Angle<TFloat32> angle(Operand<T> input) {
+  public <T extends TType> Angle<TFloat32> angle(Operand<T> input) {
     return Angle.create(scope, input);
   }
 
@@ -290,8 +290,7 @@ public <T extends Tensor> Angle<TFloat32> angle(Operand<T> input) {
    * @param Tout
    * @return a new instance of Angle
    */
-  public <U extends Tensor & TNumber, T extends Tensor> Angle<U> angle(Operand<T> input,
-      DataType<U> Tout) {
+  public <U extends TNumber, T extends TType> Angle<U> angle(Operand<T> input, DataType<U> Tout) {
     return Angle.create(scope, input, Tout);
   }
 
@@ -303,7 +302,7 @@ public <U extends Tensor & TNumber, T extends Tensor> Angle<U> angle(Operand<T>
    * @param options carries optional attributes values
    * @return a new instance of ApproximateEqual
    */
-  public <T extends Tensor> ApproximateEqual approximateEqual(Operand<T> x, Operand<T> y,
+  public <T extends TType> ApproximateEqual approximateEqual(Operand<T> x, Operand<T> y,
       ApproximateEqual.Options... options) {
     return ApproximateEqual.create(scope, x, y, options);
   }
@@ -330,7 +329,7 @@ public <T extends Tensor> ApproximateEqual approximateEqual(Operand<T> x, Operan
    *  use dimension = 0.
    * @return a new instance of ArgMax
    */
-  public <T extends Tensor, U extends Tensor & TNumber> ArgMax<TInt64> argMax(Operand<T> input,
+  public <T extends TType, U extends TNumber> ArgMax<TInt64> argMax(Operand<T> input,
       Operand<U> dimension) {
     return ArgMax.create(scope, input, dimension);
   }
@@ -358,8 +357,8 @@ public <T extends Tensor, U extends Tensor & TNumber> ArgMax<TInt64> argMax(Oper
    * @param outputType
    * @return a new instance of ArgMax
    */
-  public <V extends Tensor & TNumber, T extends Tensor, U extends Tensor & TNumber> ArgMax<V> argMax(
-      Operand<T> input, Operand<U> dimension, DataType<V> outputType) {
+  public <V extends TNumber, T extends TType, U extends TNumber> ArgMax<V> argMax(Operand<T> input,
+      Operand<U> dimension, DataType<V> outputType) {
     return ArgMax.create(scope, input, dimension, outputType);
   }
 
@@ -385,7 +384,7 @@ public <V extends Tensor & TNumber, T extends Tensor, U extends Tensor & TNumber
    *  use dimension = 0.
    * @return a new instance of ArgMin
    */
-  public <T extends Tensor, U extends Tensor & TNumber> ArgMin<TInt64> argMin(Operand<T> input,
+  public <T extends TType, U extends TNumber> ArgMin<TInt64> argMin(Operand<T> input,
       Operand<U> dimension) {
     return ArgMin.create(scope, input, dimension);
   }
@@ -413,8 +412,8 @@ public <T extends Tensor, U extends Tensor & TNumber> ArgMin<TInt64> argMin(Oper
    * @param outputType
    * @return a new instance of ArgMin
    */
-  public <V extends Tensor & TNumber, T extends Tensor, U extends Tensor & TNumber> ArgMin<V> argMin(
-      Operand<T> input, Operand<U> dimension, DataType<V> outputType) {
+  public <V extends TNumber, T extends TType, U extends TNumber> ArgMin<V> argMin(Operand<T> input,
+      Operand<U> dimension, DataType<V> outputType) {
     return ArgMin.create(scope, input, dimension, outputType);
   }
 
@@ -440,7 +439,7 @@ public <V extends Tensor & TNumber, T extends Tensor, U extends Tensor & TNumber
    * @param x
    * @return a new instance of Asin
    */
-  public <T extends Tensor> Asin<T> asin(Operand<T> x) {
+  public <T extends TType> Asin<T> asin(Operand<T> x) {
     return Asin.create(scope, x);
   }
 
@@ -460,7 +459,7 @@ public <T extends Tensor> Asin<T> asin(Operand<T> x) {
    * @param x
    * @return a new instance of Asinh
    */
-  public <T extends Tensor> Asinh<T> asinh(Operand<T> x) {
+  public <T extends TType> Asinh<T> asinh(Operand<T> x) {
     return Asinh.create(scope, x);
   }
 
@@ -486,7 +485,7 @@ public <T extends Tensor> Asinh<T> asinh(Operand<T> x) {
    * @param x
    * @return a new instance of Atan
    */
-  public <T extends Tensor> Atan<T> atan(Operand<T> x) {
+  public <T extends TType> Atan<T> atan(Operand<T> x) {
     return Atan.create(scope, x);
   }
 
@@ -504,7 +503,7 @@ public <T extends Tensor> Atan<T> atan(Operand<T> x) {
    * @param x
    * @return a new instance of Atan2
    */
-  public <T extends Tensor & TNumber> Atan2<T> atan2(Operand<T> y, Operand<T> x) {
+  public <T extends TNumber> Atan2<T> atan2(Operand<T> y, Operand<T> x) {
     return Atan2.create(scope, y, x);
   }
 
@@ -526,7 +525,7 @@ public <T extends Tensor & TNumber> Atan2<T> atan2(Operand<T> y, Operand<T> x) {
    * @param x
    * @return a new instance of Atanh
    */
-  public <T extends Tensor> Atanh<T> atanh(Operand<T> x) {
+  public <T extends TType> Atanh<T> atanh(Operand<T> x) {
     return Atanh.create(scope, x);
   }
 
@@ -542,7 +541,7 @@ public <T extends Tensor> Atanh<T> atanh(Operand<T> x) {
    * @param x
    * @return a new instance of BesselI0e
    */
-  public <T extends Tensor & TNumber> BesselI0e<T> besselI0e(Operand<T> x) {
+  public <T extends TNumber> BesselI0e<T> besselI0e(Operand<T> x) {
     return BesselI0e.create(scope, x);
   }
 
@@ -558,7 +557,7 @@ public <T extends Tensor & TNumber> BesselI0e<T> besselI0e(Operand<T> x) {
    * @param x
    * @return a new instance of BesselI1e
    */
-  public <T extends Tensor & TNumber> BesselI1e<T> besselI1e(Operand<T> x) {
+  public <T extends TNumber> BesselI1e<T> besselI1e(Operand<T> x) {
     return BesselI1e.create(scope, x);
   }
 
@@ -582,7 +581,7 @@ public <T extends Tensor & TNumber> BesselI1e<T> besselI1e(Operand<T> x) {
    * @param x
    * @return a new instance of Betainc
    */
-  public <T extends Tensor & TNumber> Betainc<T> betainc(Operand<T> a, Operand<T> b, Operand<T> x) {
+  public <T extends TNumber> Betainc<T> betainc(Operand<T> a, Operand<T> b, Operand<T> x) {
     return Betainc.create(scope, a, b, x);
   }
 
@@ -605,8 +604,8 @@ public <T extends Tensor & TNumber> Betainc<T> betainc(Operand<T> a, Operand<T>
    *  equal to 1.
    * @return a new instance of Bincount
    */
-  public <T extends Tensor & TNumber> Bincount<T> bincount(Operand<TInt32> arr,
-      Operand<TInt32> size, Operand<T> weights) {
+  public <T extends TNumber> Bincount<T> bincount(Operand<TInt32> arr, Operand<TInt32> size,
+      Operand<T> weights) {
     return Bincount.create(scope, arr, size, weights);
   }
 
@@ -617,7 +616,7 @@ public <T extends Tensor & TNumber> Bincount<T> bincount(Operand<TInt32> arr,
    * @param x
    * @return a new instance of Ceil
    */
-  public <T extends Tensor & TNumber> Ceil<T> ceil(Operand<T> x) {
+  public <T extends TNumber> Ceil<T> ceil(Operand<T> x) {
     return Ceil.create(scope, x);
   }
 
@@ -650,7 +649,7 @@ public <T extends Tensor & TNumber> Ceil<T> ceil(Operand<T> x) {
    * @param threshold Threshold to compare against.
    * @return a new instance of CompareAndBitpack
    */
-  public <T extends Tensor> CompareAndBitpack compareAndBitpack(Operand<T> input,
+  public <T extends TType> CompareAndBitpack compareAndBitpack(Operand<T> input,
       Operand<T> threshold) {
     return CompareAndBitpack.create(scope, input, threshold);
   }
@@ -667,7 +666,7 @@ public <T extends Tensor> CompareAndBitpack compareAndBitpack(Operand<T> input,
    * @param x
    * @return a new instance of ComplexAbs
    */
-  public <T extends Tensor> ComplexAbs<TFloat32> complexAbs(Operand<T> x) {
+  public <T extends TType> ComplexAbs<TFloat32> complexAbs(Operand<T> x) {
     return ComplexAbs.create(scope, x);
   }
 
@@ -684,7 +683,7 @@ public <T extends Tensor> ComplexAbs<TFloat32> complexAbs(Operand<T> x) {
    * @param Tout
    * @return a new instance of ComplexAbs
    */
-  public <U extends Tensor & TNumber, T extends Tensor> ComplexAbs<U> complexAbs(Operand<T> x,
+  public <U extends TNumber, T extends TType> ComplexAbs<U> complexAbs(Operand<T> x,
       DataType<U> Tout) {
     return ComplexAbs.create(scope, x, Tout);
   }
@@ -709,7 +708,7 @@ public <U extends Tensor & TNumber, T extends Tensor> ComplexAbs<U> complexAbs(O
    * @param input
    * @return a new instance of Conj
    */
-  public <T extends Tensor> Conj<T> conj(Operand<T> input) {
+  public <T extends TType> Conj<T> conj(Operand<T> input) {
     return Conj.create(scope, input);
   }
 
@@ -730,7 +729,7 @@ public <T extends Tensor> Conj<T> conj(Operand<T> input) {
    * @param x
    * @return a new instance of Cos
    */
-  public <T extends Tensor> Cos<T> cos(Operand<T> x) {
+  public <T extends TType> Cos<T> cos(Operand<T> x) {
     return Cos.create(scope, x);
   }
 
@@ -750,7 +749,7 @@ public <T extends Tensor> Cos<T> cos(Operand<T> x) {
    * @param x
    * @return a new instance of Cosh
    */
-  public <T extends Tensor> Cosh<T> cosh(Operand<T> x) {
+  public <T extends TType> Cosh<T> cosh(Operand<T> x) {
     return Cosh.create(scope, x);
   }
 
@@ -788,8 +787,8 @@ public <T extends Tensor> Cosh<T> cosh(Operand<T> x) {
    * @param options carries optional attributes values
    * @return a new instance of Cumprod
    */
-  public <T extends Tensor, U extends Tensor & TNumber> Cumprod<T> cumprod(Operand<T> x,
-      Operand<U> axis, Cumprod.Options... options) {
+  public <T extends TType, U extends TNumber> Cumprod<T> cumprod(Operand<T> x, Operand<U> axis,
+      Cumprod.Options... options) {
     return Cumprod.create(scope, x, axis, options);
   }
 
@@ -827,8 +826,8 @@ public <T extends Tensor, U extends Tensor & TNumber> Cumprod<T> cumprod(Operand
    * @param options carries optional attributes values
    * @return a new instance of Cumsum
    */
-  public <T extends Tensor, U extends Tensor & TNumber> Cumsum<T> cumsum(Operand<T> x,
-      Operand<U> axis, Cumsum.Options... options) {
+  public <T extends TType, U extends TNumber> Cumsum<T> cumsum(Operand<T> x, Operand<U> axis,
+      Cumsum.Options... options) {
     return Cumsum.create(scope, x, axis, options);
   }
 
@@ -841,7 +840,7 @@ public <T extends Tensor, U extends Tensor & TNumber> Cumsum<T> cumsum(Operand<T
    * @param x
    * @return a new instance of Digamma
    */
-  public <T extends Tensor & TNumber> Digamma<T> digamma(Operand<T> x) {
+  public <T extends TNumber> Digamma<T> digamma(Operand<T> x) {
     return Digamma.create(scope, x);
   }
 
@@ -856,7 +855,7 @@ public <T extends Tensor & TNumber> Digamma<T> digamma(Operand<T> x) {
    * @param y
    * @return a new instance of Div
    */
-  public <T extends Tensor> Div<T> div(Operand<T> x, Operand<T> y) {
+  public <T extends TType> Div<T> div(Operand<T> x, Operand<T> y) {
     return Div.create(scope, x, y);
   }
 
@@ -872,7 +871,7 @@ public <T extends Tensor> Div<T> div(Operand<T> x, Operand<T> y) {
    * @param y
    * @return a new instance of DivNoNan
    */
-  public <T extends Tensor> DivNoNan<T> divNoNan(Operand<T> x, Operand<T> y) {
+  public <T extends TType> DivNoNan<T> divNoNan(Operand<T> x, Operand<T> y) {
     return DivNoNan.create(scope, x, y);
   }
 
@@ -896,7 +895,7 @@ public <T extends Tensor> DivNoNan<T> divNoNan(Operand<T> x, Operand<T> y) {
    * @param options carries optional attributes values
    * @return a new instance of Equal
    */
-  public <T extends Tensor> Equal equal(Operand<T> x, Operand<T> y, Equal.Options... options) {
+  public <T extends TType> Equal equal(Operand<T> x, Operand<T> y, Equal.Options... options) {
     return Equal.create(scope, x, y, options);
   }
 
@@ -907,7 +906,7 @@ public <T extends Tensor> Equal equal(Operand<T> x, Operand<T> y, Equal.Options.
    * @param x
    * @return a new instance of Erf
    */
-  public <T extends Tensor & TNumber> Erf<T> erf(Operand<T> x) {
+  public <T extends TNumber> Erf<T> erf(Operand<T> x) {
     return Erf.create(scope, x);
   }
 
@@ -918,7 +917,7 @@ public <T extends Tensor & TNumber> Erf<T> erf(Operand<T> x) {
    * @param x
    * @return a new instance of Erfc
    */
-  public <T extends Tensor & TNumber> Erfc<T> erfc(Operand<T> x) {
+  public <T extends TNumber> Erfc<T> erfc(Operand<T> x) {
     return Erfc.create(scope, x);
   }
 
@@ -928,7 +927,7 @@ public <T extends Tensor & TNumber> Erfc<T> erfc(Operand<T> x) {
    * @param x
    * @return a new instance of erfinv
    */
-  public <T extends Tensor & TNumber> erfinv<T> erfinv(Operand<T> x) {
+  public <T extends TNumber> erfinv<T> erfinv(Operand<T> x) {
     return erfinv.create(scope, x);
   }
 
@@ -964,7 +963,7 @@ public <T extends Tensor & TNumber> erfinv<T> erfinv(Operand<T> x) {
    * @param x
    * @return a new instance of Exp
    */
-  public <T extends Tensor> Exp<T> exp(Operand<T> x) {
+  public <T extends TType> Exp<T> exp(Operand<T> x) {
     return Exp.create(scope, x);
   }
 
@@ -989,7 +988,7 @@ public <T extends Tensor> Exp<T> exp(Operand<T> x) {
    * @param x
    * @return a new instance of Expm1
    */
-  public <T extends Tensor> Expm1<T> expm1(Operand<T> x) {
+  public <T extends TType> Expm1<T> expm1(Operand<T> x) {
     return Expm1.create(scope, x);
   }
 
@@ -1009,7 +1008,7 @@ public Fact fact() {
    * @param x
    * @return a new instance of Floor
    */
-  public <T extends Tensor & TNumber> Floor<T> floor(Operand<T> x) {
+  public <T extends TNumber> Floor<T> floor(Operand<T> x) {
     return Floor.create(scope, x);
   }
 
@@ -1024,7 +1023,7 @@ public <T extends Tensor & TNumber> Floor<T> floor(Operand<T> x) {
    * @param y
    * @return a new instance of FloorDiv
    */
-  public <T extends Tensor> FloorDiv<T> floorDiv(Operand<T> x, Operand<T> y) {
+  public <T extends TType> FloorDiv<T> floorDiv(Operand<T> x, Operand<T> y) {
     return FloorDiv.create(scope, x, y);
   }
 
@@ -1042,7 +1041,7 @@ public <T extends Tensor> FloorDiv<T> floorDiv(Operand<T> x, Operand<T> y) {
    * @param y
    * @return a new instance of FloorMod
    */
-  public <T extends Tensor & TNumber> FloorMod<T> floorMod(Operand<T> x, Operand<T> y) {
+  public <T extends TNumber> FloorMod<T> floorMod(Operand<T> x, Operand<T> y) {
     return FloorMod.create(scope, x, y);
   }
 
@@ -1067,7 +1066,7 @@ public <T extends Tensor & TNumber> FloorMod<T> floorMod(Operand<T> x, Operand<T
    * @param y
    * @return a new instance of Greater
    */
-  public <T extends Tensor & TNumber> Greater greater(Operand<T> x, Operand<T> y) {
+  public <T extends TNumber> Greater greater(Operand<T> x, Operand<T> y) {
     return Greater.create(scope, x, y);
   }
 
@@ -1092,7 +1091,7 @@ public <T extends Tensor & TNumber> Greater greater(Operand<T> x, Operand<T> y)
    * @param y
    * @return a new instance of GreaterEqual
    */
-  public <T extends Tensor & TNumber> GreaterEqual greaterEqual(Operand<T> x, Operand<T> y) {
+  public <T extends TNumber> GreaterEqual greaterEqual(Operand<T> x, Operand<T> y) {
     return GreaterEqual.create(scope, x, y);
   }
 
@@ -1117,7 +1116,7 @@ public <T extends Tensor & TNumber> GreaterEqual greaterEqual(Operand<T> x, Oper
    * @param x
    * @return a new instance of Igamma
    */
-  public <T extends Tensor & TNumber> Igamma<T> igamma(Operand<T> a, Operand<T> x) {
+  public <T extends TNumber> Igamma<T> igamma(Operand<T> a, Operand<T> x) {
     return Igamma.create(scope, a, x);
   }
 
@@ -1142,7 +1141,7 @@ public <T extends Tensor & TNumber> Igamma<T> igamma(Operand<T> a, Operand<T> x)
    * @param x
    * @return a new instance of Igammac
    */
-  public <T extends Tensor & TNumber> Igammac<T> igammac(Operand<T> a, Operand<T> x) {
+  public <T extends TNumber> Igammac<T> igammac(Operand<T> a, Operand<T> x) {
     return Igammac.create(scope, a, x);
   }
 
@@ -1164,7 +1163,7 @@ public <T extends Tensor & TNumber> Igammac<T> igammac(Operand<T> a, Operand<T>
    * @param input
    * @return a new instance of Imag
    */
-  public <T extends Tensor> Imag<TFloat32> imag(Operand<T> input) {
+  public <T extends TType> Imag<TFloat32> imag(Operand<T> input) {
     return Imag.create(scope, input);
   }
 
@@ -1187,8 +1186,7 @@ public <T extends Tensor> Imag<TFloat32> imag(Operand<T> input) {
    * @param Tout
    * @return a new instance of Imag
    */
-  public <U extends Tensor & TNumber, T extends Tensor> Imag<U> imag(Operand<T> input,
-      DataType<U> Tout) {
+  public <U extends TNumber, T extends TType> Imag<U> imag(Operand<T> input, DataType<U> Tout) {
     return Imag.create(scope, input, Tout);
   }
 
@@ -1214,7 +1212,7 @@ public <U extends Tensor & TNumber, T extends Tensor> Imag<U> imag(Operand<T> in
    * @param x 1-D.
    * @return a new instance of InvertPermutation
    */
-  public <T extends Tensor & TNumber> InvertPermutation<T> invertPermutation(Operand<T> x) {
+  public <T extends TNumber> InvertPermutation<T> invertPermutation(Operand<T> x) {
     return InvertPermutation.create(scope, x);
   }
 
@@ -1232,7 +1230,7 @@ public <T extends Tensor & TNumber> InvertPermutation<T> invertPermutation(Opera
    * @param x
    * @return a new instance of IsFinite
    */
-  public <T extends Tensor & TNumber> IsFinite isFinite(Operand<T> x) {
+  public <T extends TNumber> IsFinite isFinite(Operand<T> x) {
     return IsFinite.create(scope, x);
   }
 
@@ -1250,7 +1248,7 @@ public <T extends Tensor & TNumber> IsFinite isFinite(Operand<T> x) {
    * @param x
    * @return a new instance of IsInf
    */
-  public <T extends Tensor & TNumber> IsInf isInf(Operand<T> x) {
+  public <T extends TNumber> IsInf isInf(Operand<T> x) {
     return IsInf.create(scope, x);
   }
 
@@ -1268,7 +1266,7 @@ public <T extends Tensor & TNumber> IsInf isInf(Operand<T> x) {
    * @param x
    * @return a new instance of IsNan
    */
-  public <T extends Tensor & TNumber> IsNan isNan(Operand<T> x) {
+  public <T extends TNumber> IsNan isNan(Operand<T> x) {
     return IsNan.create(scope, x);
   }
 
@@ -1293,7 +1291,7 @@ public <T extends Tensor & TNumber> IsNan isNan(Operand<T> x) {
    * @param y
    * @return a new instance of Less
    */
-  public <T extends Tensor & TNumber> Less less(Operand<T> x, Operand<T> y) {
+  public <T extends TNumber> Less less(Operand<T> x, Operand<T> y) {
     return Less.create(scope, x, y);
   }
 
@@ -1318,7 +1316,7 @@ public <T extends Tensor & TNumber> Less less(Operand<T> x, Operand<T> y) {
    * @param y
    * @return a new instance of LessEqual
    */
-  public <T extends Tensor & TNumber> LessEqual lessEqual(Operand<T> x, Operand<T> y) {
+  public <T extends TNumber> LessEqual lessEqual(Operand<T> x, Operand<T> y) {
     return LessEqual.create(scope, x, y);
   }
 
@@ -1338,7 +1336,7 @@ public <T extends Tensor & TNumber> LessEqual lessEqual(Operand<T> x, Operand<T>
    * @param x
    * @return a new instance of Lgamma
    */
-  public <T extends Tensor & TNumber> Lgamma<T> lgamma(Operand<T> x) {
+  public <T extends TNumber> Lgamma<T> lgamma(Operand<T> x) {
     return Lgamma.create(scope, x);
   }
 
@@ -1357,7 +1355,7 @@ public <T extends Tensor & TNumber> Lgamma<T> lgamma(Operand<T> x) {
    * @param x
    * @return a new instance of Log
    */
-  public <T extends Tensor> Log<T> log(Operand<T> x) {
+  public <T extends TType> Log<T> log(Operand<T> x) {
     return Log.create(scope, x);
   }
 
@@ -1376,7 +1374,7 @@ public <T extends Tensor> Log<T> log(Operand<T> x) {
    * @param x
    * @return a new instance of Log1p
    */
-  public <T extends Tensor> Log1p<T> log1p(Operand<T> x) {
+  public <T extends TType> Log1p<T> log1p(Operand<T> x) {
     return Log1p.create(scope, x);
   }
 
@@ -1429,7 +1427,7 @@ public LogicalOr logicalOr(Operand<TBool> x, Operand<TBool> y) {
    * @param y
    * @return a new instance of Maximum
    */
-  public <T extends Tensor & TNumber> Maximum<T> maximum(Operand<T> x, Operand<T> y) {
+  public <T extends TNumber> Maximum<T> maximum(Operand<T> x, Operand<T> y) {
     return Maximum.create(scope, x, y);
   }
 
@@ -1448,8 +1446,8 @@ public <T extends Tensor & TNumber> Maximum<T> maximum(Operand<T> x, Operand<T>
    * @param options carries optional attributes values
    * @return a new instance of Mean
    */
-  public <T extends Tensor, U extends Tensor & TNumber> Mean<T> mean(Operand<T> input,
-      Operand<U> axis, Mean.Options... options) {
+  public <T extends TType, U extends TNumber> Mean<T> mean(Operand<T> input, Operand<U> axis,
+      Mean.Options... options) {
     return Mean.create(scope, input, axis, options);
   }
 
@@ -1464,7 +1462,7 @@ public <T extends Tensor, U extends Tensor & TNumber> Mean<T> mean(Operand<T> in
    * @param y
    * @return a new instance of Minimum
    */
-  public <T extends Tensor & TNumber> Minimum<T> minimum(Operand<T> x, Operand<T> y) {
+  public <T extends TNumber> Minimum<T> minimum(Operand<T> x, Operand<T> y) {
     return Minimum.create(scope, x, y);
   }
 
@@ -1482,7 +1480,7 @@ public <T extends Tensor & TNumber> Minimum<T> minimum(Operand<T> x, Operand<T>
    * @param y
    * @return a new instance of Mod
    */
-  public <T extends Tensor & TNumber> Mod<T> mod(Operand<T> x, Operand<T> y) {
+  public <T extends TNumber> Mod<T> mod(Operand<T> x, Operand<T> y) {
     return Mod.create(scope, x, y);
   }
 
@@ -1497,7 +1495,7 @@ public <T extends Tensor & TNumber> Mod<T> mod(Operand<T> x, Operand<T> y) {
    * @param y
    * @return a new instance of Mul
    */
-  public <T extends Tensor> Mul<T> mul(Operand<T> x, Operand<T> y) {
+  public <T extends TType> Mul<T> mul(Operand<T> x, Operand<T> y) {
     return Mul.create(scope, x, y);
   }
 
@@ -1512,7 +1510,7 @@ public <T extends Tensor> Mul<T> mul(Operand<T> x, Operand<T> y) {
    * @param y
    * @return a new instance of MulNoNan
    */
-  public <T extends Tensor> MulNoNan<T> mulNoNan(Operand<T> x, Operand<T> y) {
+  public <T extends TType> MulNoNan<T> mulNoNan(Operand<T> x, Operand<T> y) {
     return MulNoNan.create(scope, x, y);
   }
 
@@ -1522,7 +1520,7 @@ public <T extends Tensor> MulNoNan<T> mulNoNan(Operand<T> x, Operand<T> y) {
    * @param x
    * @return a new instance of Ndtri
    */
-  public <T extends Tensor & TNumber> Ndtri<T> ndtri(Operand<T> x) {
+  public <T extends TNumber> Ndtri<T> ndtri(Operand<T> x) {
     return Ndtri.create(scope, x);
   }
 
@@ -1535,7 +1533,7 @@ public <T extends Tensor & TNumber> Ndtri<T> ndtri(Operand<T> x) {
    * @param x
    * @return a new instance of Neg
    */
-  public <T extends Tensor> Neg<T> neg(Operand<T> x) {
+  public <T extends TType> Neg<T> neg(Operand<T> x) {
     return Neg.create(scope, x);
   }
 
@@ -1554,7 +1552,7 @@ public <T extends Tensor> Neg<T> neg(Operand<T> x) {
    * @param x2
    * @return a new instance of NextAfter
    */
-  public <T extends Tensor & TNumber> NextAfter<T> nextAfter(Operand<T> x1, Operand<T> x2) {
+  public <T extends TNumber> NextAfter<T> nextAfter(Operand<T> x1, Operand<T> x2) {
     return NextAfter.create(scope, x1, x2);
   }
 
@@ -1569,7 +1567,7 @@ public <T extends Tensor & TNumber> NextAfter<T> nextAfter(Operand<T> x1, Operan
    * @param options carries optional attributes values
    * @return a new instance of NotEqual
    */
-  public <T extends Tensor> NotEqual notEqual(Operand<T> x, Operand<T> y,
+  public <T extends TType> NotEqual notEqual(Operand<T> x, Operand<T> y,
       NotEqual.Options... options) {
     return NotEqual.create(scope, x, y, options);
   }
@@ -1589,7 +1587,7 @@ public <T extends Tensor> NotEqual notEqual(Operand<T> x, Operand<T> y,
    * @param x
    * @return a new instance of Polygamma
    */
-  public <T extends Tensor & TNumber> Polygamma<T> polygamma(Operand<T> a, Operand<T> x) {
+  public <T extends TNumber> Polygamma<T> polygamma(Operand<T> a, Operand<T> x) {
     return Polygamma.create(scope, a, x);
   }
 
@@ -1606,7 +1604,7 @@ public <T extends Tensor & TNumber> Polygamma<T> polygamma(Operand<T> a, Operand
    * @param x
    * @return a new instance of PopulationCount
    */
-  public <T extends Tensor & TNumber> PopulationCount populationCount(Operand<T> x) {
+  public <T extends TNumber> PopulationCount populationCount(Operand<T> x) {
     return PopulationCount.create(scope, x);
   }
 
@@ -1626,7 +1624,7 @@ public <T extends Tensor & TNumber> PopulationCount populationCount(Operand<T> x
    * @param y
    * @return a new instance of Pow
    */
-  public <T extends Tensor> Pow<T> pow(Operand<T> x, Operand<T> y) {
+  public <T extends TType> Pow<T> pow(Operand<T> x, Operand<T> y) {
     return Pow.create(scope, x, y);
   }
 
@@ -1643,7 +1641,7 @@ public <T extends Tensor> Pow<T> pow(Operand<T> x, Operand<T> y) {
    * @param Toutput
    * @return a new instance of QuantizedAdd
    */
-  public <V extends Tensor, T extends Tensor, U extends Tensor> QuantizedAdd<V> quantizedAdd(
+  public <V extends TType, T extends TType, U extends TType> QuantizedAdd<V> quantizedAdd(
       Operand<T> x, Operand<U> y, Operand<TFloat32> minX, Operand<TFloat32> maxX,
       Operand<TFloat32> minY, Operand<TFloat32> maxY, DataType<V> Toutput) {
     return QuantizedAdd.create(scope, x, y, minX, maxX, minY, maxY, Toutput);
@@ -1662,7 +1660,7 @@ public <V extends Tensor, T extends Tensor, U extends Tensor> QuantizedAdd<V> qu
    * @param Toutput
    * @return a new instance of QuantizedMul
    */
-  public <V extends Tensor, T extends Tensor, U extends Tensor> QuantizedMul<V> quantizedMul(
+  public <V extends TType, T extends TType, U extends TType> QuantizedMul<V> quantizedMul(
       Operand<T> x, Operand<U> y, Operand<TFloat32> minX, Operand<TFloat32> maxX,
       Operand<TFloat32> minY, Operand<TFloat32> maxY, DataType<V> Toutput) {
     return QuantizedMul.create(scope, x, y, minX, maxX, minY, maxY, Toutput);
@@ -1686,7 +1684,7 @@ public <V extends Tensor, T extends Tensor, U extends Tensor> QuantizedMul<V> qu
    * @param input
    * @return a new instance of Real
    */
-  public <T extends Tensor> Real<TFloat32> real(Operand<T> input) {
+  public <T extends TType> Real<TFloat32> real(Operand<T> input) {
     return Real.create(scope, input);
   }
 
@@ -1709,8 +1707,7 @@ public <T extends Tensor> Real<TFloat32> real(Operand<T> input) {
    * @param Tout
    * @return a new instance of Real
    */
-  public <U extends Tensor & TNumber, T extends Tensor> Real<U> real(Operand<T> input,
-      DataType<U> Tout) {
+  public <U extends TNumber, T extends TType> Real<U> real(Operand<T> input, DataType<U> Tout) {
     return Real.create(scope, input, Tout);
   }
 
@@ -1727,7 +1724,7 @@ public <U extends Tensor & TNumber, T extends Tensor> Real<U> real(Operand<T> in
    * @param y
    * @return a new instance of RealDiv
    */
-  public <T extends Tensor> RealDiv<T> realDiv(Operand<T> x, Operand<T> y) {
+  public <T extends TType> RealDiv<T> realDiv(Operand<T> x, Operand<T> y) {
     return RealDiv.create(scope, x, y);
   }
 
@@ -1740,7 +1737,7 @@ public <T extends Tensor> RealDiv<T> realDiv(Operand<T> x, Operand<T> y) {
    * @param x
    * @return a new instance of Reciprocal
    */
-  public <T extends Tensor> Reciprocal<T> reciprocal(Operand<T> x) {
+  public <T extends TType> Reciprocal<T> reciprocal(Operand<T> x) {
     return Reciprocal.create(scope, x);
   }
 
@@ -1760,7 +1757,7 @@ public <T extends Tensor> Reciprocal<T> reciprocal(Operand<T> x) {
    * @param x
    * @return a new instance of Rint
    */
-  public <T extends Tensor & TNumber> Rint<T> rint(Operand<T> x) {
+  public <T extends TNumber> Rint<T> rint(Operand<T> x) {
     return Rint.create(scope, x);
   }
 
@@ -1774,7 +1771,7 @@ public <T extends Tensor & TNumber> Rint<T> rint(Operand<T> x) {
    * @param x
    * @return a new instance of Round
    */
-  public <T extends Tensor> Round<T> round(Operand<T> x) {
+  public <T extends TType> Round<T> round(Operand<T> x) {
     return Round.create(scope, x);
   }
 
@@ -1787,7 +1784,7 @@ public <T extends Tensor> Round<T> round(Operand<T> x) {
    * @param x
    * @return a new instance of Rsqrt
    */
-  public <T extends Tensor> Rsqrt<T> rsqrt(Operand<T> x) {
+  public <T extends TType> Rsqrt<T> rsqrt(Operand<T> x) {
     return Rsqrt.create(scope, x);
   }
 
@@ -1822,8 +1819,8 @@ public <T extends Tensor> Rsqrt<T> rsqrt(Operand<T> x) {
    *  first dimension.  Values should be sorted and can be repeated.
    * @return a new instance of SegmentMax
    */
-  public <T extends Tensor & TNumber, U extends Tensor & TNumber> SegmentMax<T> segmentMax(
-      Operand<T> data, Operand<U> segmentIds) {
+  public <T extends TNumber, U extends TNumber> SegmentMax<T> segmentMax(Operand<T> data,
+      Operand<U> segmentIds) {
     return SegmentMax.create(scope, data, segmentIds);
   }
 
@@ -1859,7 +1856,7 @@ public <T extends Tensor & TNumber, U extends Tensor & TNumber> SegmentMax<T> se
    *  first dimension.  Values should be sorted and can be repeated.
    * @return a new instance of SegmentMean
    */
-  public <T extends Tensor, U extends Tensor & TNumber> SegmentMean<T> segmentMean(Operand<T> data,
+  public <T extends TType, U extends TNumber> SegmentMean<T> segmentMean(Operand<T> data,
       Operand<U> segmentIds) {
     return SegmentMean.create(scope, data, segmentIds);
   }
@@ -1895,8 +1892,8 @@ public <T extends Tensor, U extends Tensor & TNumber> SegmentMean<T> segmentMean
    *  first dimension.  Values should be sorted and can be repeated.
    * @return a new instance of SegmentMin
    */
-  public <T extends Tensor & TNumber, U extends Tensor & TNumber> SegmentMin<T> segmentMin(
-      Operand<T> data, Operand<U> segmentIds) {
+  public <T extends TNumber, U extends TNumber> SegmentMin<T> segmentMin(Operand<T> data,
+      Operand<U> segmentIds) {
     return SegmentMin.create(scope, data, segmentIds);
   }
 
@@ -1931,7 +1928,7 @@ public <T extends Tensor & TNumber, U extends Tensor & TNumber> SegmentMin<T> se
    *  first dimension.  Values should be sorted and can be repeated.
    * @return a new instance of SegmentProd
    */
-  public <T extends Tensor, U extends Tensor & TNumber> SegmentProd<T> segmentProd(Operand<T> data,
+  public <T extends TType, U extends TNumber> SegmentProd<T> segmentProd(Operand<T> data,
       Operand<U> segmentIds) {
     return SegmentProd.create(scope, data, segmentIds);
   }
@@ -1967,7 +1964,7 @@ public <T extends Tensor, U extends Tensor & TNumber> SegmentProd<T> segmentProd
    *  first dimension.  Values should be sorted and can be repeated.
    * @return a new instance of SegmentSum
    */
-  public <T extends Tensor, U extends Tensor & TNumber> SegmentSum<T> segmentSum(Operand<T> data,
+  public <T extends TType, U extends TNumber> SegmentSum<T> segmentSum(Operand<T> data,
       Operand<U> segmentIds) {
     return SegmentSum.create(scope, data, segmentIds);
   }
@@ -1981,7 +1978,7 @@ public <T extends Tensor, U extends Tensor & TNumber> SegmentSum<T> segmentSum(O
    * @param x
    * @return a new instance of Sigmoid
    */
-  public <T extends Tensor> Sigmoid<T> sigmoid(Operand<T> x) {
+  public <T extends TType> Sigmoid<T> sigmoid(Operand<T> x) {
     return Sigmoid.create(scope, x);
   }
 
@@ -2000,7 +1997,7 @@ public <T extends Tensor> Sigmoid<T> sigmoid(Operand<T> x) {
    * @param x
    * @return a new instance of Sign
    */
-  public <T extends Tensor> Sign<T> sign(Operand<T> x) {
+  public <T extends TType> Sign<T> sign(Operand<T> x) {
     return Sign.create(scope, x);
   }
 
@@ -2020,7 +2017,7 @@ public <T extends Tensor> Sign<T> sign(Operand<T> x) {
    * @param x
    * @return a new instance of Sin
    */
-  public <T extends Tensor> Sin<T> sin(Operand<T> x) {
+  public <T extends TType> Sin<T> sin(Operand<T> x) {
     return Sin.create(scope, x);
   }
 
@@ -2040,7 +2037,7 @@ public <T extends Tensor> Sin<T> sin(Operand<T> x) {
    * @param x
    * @return a new instance of Sinh
    */
-  public <T extends Tensor> Sinh<T> sinh(Operand<T> x) {
+  public <T extends TType> Sinh<T> sinh(Operand<T> x) {
     return Sinh.create(scope, x);
   }
 
@@ -2051,7 +2048,7 @@ public <T extends Tensor> Sinh<T> sinh(Operand<T> x) {
    * @param features
    * @return a new instance of Softplus
    */
-  public <T extends Tensor & TNumber> Softplus<T> softplus(Operand<T> features) {
+  public <T extends TNumber> Softplus<T> softplus(Operand<T> features) {
     return Softplus.create(scope, features);
   }
 
@@ -2064,7 +2061,7 @@ public <T extends Tensor & TNumber> Softplus<T> softplus(Operand<T> features) {
    * @param x
    * @return a new instance of Sqrt
    */
-  public <T extends Tensor> Sqrt<T> sqrt(Operand<T> x) {
+  public <T extends TType> Sqrt<T> sqrt(Operand<T> x) {
     return Sqrt.create(scope, x);
   }
 
@@ -2077,7 +2074,7 @@ public <T extends Tensor> Sqrt<T> sqrt(Operand<T> x) {
    * @param x
    * @return a new instance of Square
    */
-  public <T extends Tensor> Square<T> square(Operand<T> x) {
+  public <T extends TType> Square<T> square(Operand<T> x) {
     return Square.create(scope, x);
   }
 
@@ -2092,7 +2089,7 @@ public <T extends Tensor> Square<T> square(Operand<T> x) {
    * @param y
    * @return a new instance of SquaredDifference
    */
-  public <T extends Tensor> SquaredDifference<T> squaredDifference(Operand<T> x, Operand<T> y) {
+  public <T extends TType> SquaredDifference<T> squaredDifference(Operand<T> x, Operand<T> y) {
     return SquaredDifference.create(scope, x, y);
   }
 
@@ -2107,7 +2104,7 @@ public <T extends Tensor> SquaredDifference<T> squaredDifference(Operand<T> x, O
    * @param y
    * @return a new instance of Sub
    */
-  public <T extends Tensor> Sub<T> sub(Operand<T> x, Operand<T> y) {
+  public <T extends TType> Sub<T> sub(Operand<T> x, Operand<T> y) {
     return Sub.create(scope, x, y);
   }
 
@@ -2128,7 +2125,7 @@ public <T extends Tensor> Sub<T> sub(Operand<T> x, Operand<T> y) {
    * @param x
    * @return a new instance of Tan
    */
-  public <T extends Tensor> Tan<T> tan(Operand<T> x) {
+  public <T extends TType> Tan<T> tan(Operand<T> x) {
     return Tan.create(scope, x);
   }
 
@@ -2148,7 +2145,7 @@ public <T extends Tensor> Tan<T> tan(Operand<T> x) {
    * @param x
    * @return a new instance of Tanh
    */
-  public <T extends Tensor> Tanh<T> tanh(Operand<T> x) {
+  public <T extends TType> Tanh<T> tanh(Operand<T> x) {
     return Tanh.create(scope, x);
   }
 
@@ -2168,7 +2165,7 @@ public <T extends Tensor> Tanh<T> tanh(Operand<T> x) {
    * @param y
    * @return a new instance of TruncateDiv
    */
-  public <T extends Tensor> TruncateDiv<T> truncateDiv(Operand<T> x, Operand<T> y) {
+  public <T extends TType> TruncateDiv<T> truncateDiv(Operand<T> x, Operand<T> y) {
     return TruncateDiv.create(scope, x, y);
   }
 
@@ -2186,7 +2183,7 @@ public <T extends Tensor> TruncateDiv<T> truncateDiv(Operand<T> x, Operand<T> y)
    * @param y
    * @return a new instance of TruncateMod
    */
-  public <T extends Tensor & TNumber> TruncateMod<T> truncateMod(Operand<T> x, Operand<T> y) {
+  public <T extends TNumber> TruncateMod<T> truncateMod(Operand<T> x, Operand<T> y) {
     return TruncateMod.create(scope, x, y);
   }
 
@@ -2229,7 +2226,7 @@ public <T extends Tensor & TNumber> TruncateMod<T> truncateMod(Operand<T> x, Ope
    * @param numSegments
    * @return a new instance of UnsortedSegmentMax
    */
-  public <T extends Tensor & TNumber, U extends Tensor & TNumber, V extends Tensor & TNumber> UnsortedSegmentMax<T> unsortedSegmentMax(
+  public <T extends TNumber, U extends TNumber, V extends TNumber> UnsortedSegmentMax<T> unsortedSegmentMax(
       Operand<T> data, Operand<U> segmentIds, Operand<V> numSegments) {
     return UnsortedSegmentMax.create(scope, data, segmentIds, numSegments);
   }
@@ -2268,7 +2265,7 @@ public <T extends Tensor & TNumber, U extends Tensor & TNumber, V extends Tensor
    * @param numSegments
    * @return a new instance of UnsortedSegmentMin
    */
-  public <T extends Tensor & TNumber, U extends Tensor & TNumber, V extends Tensor & TNumber> UnsortedSegmentMin<T> unsortedSegmentMin(
+  public <T extends TNumber, U extends TNumber, V extends TNumber> UnsortedSegmentMin<T> unsortedSegmentMin(
       Operand<T> data, Operand<U> segmentIds, Operand<V> numSegments) {
     return UnsortedSegmentMin.create(scope, data, segmentIds, numSegments);
   }
@@ -2306,7 +2303,7 @@ public <T extends Tensor & TNumber, U extends Tensor & TNumber, V extends Tensor
    * @param numSegments
    * @return a new instance of UnsortedSegmentProd
    */
-  public <T extends Tensor, U extends Tensor & TNumber, V extends Tensor & TNumber> UnsortedSegmentProd<T> unsortedSegmentProd(
+  public <T extends TType, U extends TNumber, V extends TNumber> UnsortedSegmentProd<T> unsortedSegmentProd(
       Operand<T> data, Operand<U> segmentIds, Operand<V> numSegments) {
     return UnsortedSegmentProd.create(scope, data, segmentIds, numSegments);
   }
@@ -2346,7 +2343,7 @@ public <T extends Tensor, U extends Tensor & TNumber, V extends Tensor & TNumber
    * @param numSegments
    * @return a new instance of UnsortedSegmentSum
    */
-  public <T extends Tensor, U extends Tensor & TNumber, V extends Tensor & TNumber> UnsortedSegmentSum<T> unsortedSegmentSum(
+  public <T extends TType, U extends TNumber, V extends TNumber> UnsortedSegmentSum<T> unsortedSegmentSum(
       Operand<T> data, Operand<U> segmentIds, Operand<V> numSegments) {
     return UnsortedSegmentSum.create(scope, data, segmentIds, numSegments);
   }
@@ -2359,7 +2356,7 @@ public <T extends Tensor, U extends Tensor & TNumber, V extends Tensor & TNumber
    * @param y
    * @return a new instance of Xdivy
    */
-  public <T extends Tensor> Xdivy<T> xdivy(Operand<T> x, Operand<T> y) {
+  public <T extends TType> Xdivy<T> xdivy(Operand<T> x, Operand<T> y) {
     return Xdivy.create(scope, x, y);
   }
 
@@ -2371,7 +2368,7 @@ public <T extends Tensor> Xdivy<T> xdivy(Operand<T> x, Operand<T> y) {
    * @param y
    * @return a new instance of Xlog1py
    */
-  public <T extends Tensor> Xlog1py<T> xlog1py(Operand<T> x, Operand<T> y) {
+  public <T extends TType> Xlog1py<T> xlog1py(Operand<T> x, Operand<T> y) {
     return Xlog1py.create(scope, x, y);
   }
 
@@ -2383,7 +2380,7 @@ public <T extends Tensor> Xlog1py<T> xlog1py(Operand<T> x, Operand<T> y) {
    * @param y
    * @return a new instance of Xlogy
    */
-  public <T extends Tensor> Xlogy<T> xlogy(Operand<T> x, Operand<T> y) {
+  public <T extends TType> Xlogy<T> xlogy(Operand<T> x, Operand<T> y) {
     return Xlogy.create(scope, x, y);
   }
 
@@ -2399,7 +2396,7 @@ public <T extends Tensor> Xlogy<T> xlogy(Operand<T> x, Operand<T> y) {
    * @param q
    * @return a new instance of Zeta
    */
-  public <T extends Tensor & TNumber> Zeta<T> zeta(Operand<T> x, Operand<T> q) {
+  public <T extends TNumber> Zeta<T> zeta(Operand<T> x, Operand<T> q) {
     return Zeta.create(scope, x, q);
   }
 }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/NnOps.java b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/NnOps.java
index 9eb29ba9835..fb57feee8c4 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/NnOps.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/NnOps.java
@@ -20,7 +20,6 @@
 import java.util.List;
 import org.tensorflow.DataType;
 import org.tensorflow.Operand;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.nn.AvgPool;
 import org.tensorflow.op.nn.AvgPool3d;
 import org.tensorflow.op.nn.AvgPool3dGrad;
@@ -96,6 +95,7 @@
 import org.tensorflow.types.TInt32;
 import org.tensorflow.types.TInt64;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * An API for building {@code nn} operations as {@link Op Op}s
@@ -123,7 +123,7 @@ public final class NnOps {
    * @param options carries optional attributes values
    * @return a new instance of AvgPool
    */
-  public <T extends Tensor & TNumber> AvgPool<T> avgPool(Operand<T> value, List<Long> ksize,
+  public <T extends TNumber> AvgPool<T> avgPool(Operand<T> value, List<Long> ksize,
       List<Long> strides, String padding, AvgPool.Options... options) {
     return AvgPool.create(scope, value, ksize, strides, padding, options);
   }
@@ -141,7 +141,7 @@ public <T extends Tensor & TNumber> AvgPool<T> avgPool(Operand<T> value, List<Lo
    * @param options carries optional attributes values
    * @return a new instance of AvgPool3d
    */
-  public <T extends Tensor & TNumber> AvgPool3d<T> avgPool3d(Operand<T> input, List<Long> ksize,
+  public <T extends TNumber> AvgPool3d<T> avgPool3d(Operand<T> input, List<Long> ksize,
       List<Long> strides, String padding, AvgPool3d.Options... options) {
     return AvgPool3d.create(scope, input, ksize, strides, padding, options);
   }
@@ -160,7 +160,7 @@ public <T extends Tensor & TNumber> AvgPool3d<T> avgPool3d(Operand<T> input, Lis
    * @param options carries optional attributes values
    * @return a new instance of AvgPool3dGrad
    */
-  public <T extends Tensor & TNumber> AvgPool3dGrad<T> avgPool3dGrad(Operand<TInt32> origInputShape,
+  public <T extends TNumber> AvgPool3dGrad<T> avgPool3dGrad(Operand<TInt32> origInputShape,
       Operand<T> grad, List<Long> ksize, List<Long> strides, String padding,
       AvgPool3dGrad.Options... options) {
     return AvgPool3dGrad.create(scope, origInputShape, grad, ksize, strides, padding, options);
@@ -189,7 +189,7 @@ public <T extends Tensor & TNumber> AvgPool3dGrad<T> avgPool3dGrad(Operand<TInt3
    *  needs to be multiplied with gamma.
    * @return a new instance of BatchNormWithGlobalNormalization
    */
-  public <T extends Tensor> BatchNormWithGlobalNormalization<T> batchNormWithGlobalNormalization(
+  public <T extends TType> BatchNormWithGlobalNormalization<T> batchNormWithGlobalNormalization(
       Operand<T> t, Operand<T> m, Operand<T> v, Operand<T> beta, Operand<T> gamma,
       Float varianceEpsilon, Boolean scaleAfterNormalization) {
     return BatchNormWithGlobalNormalization.create(scope, t, m, v, beta, gamma, varianceEpsilon, scaleAfterNormalization);
@@ -217,7 +217,7 @@ public <T extends Tensor> BatchNormWithGlobalNormalization<T> batchNormWithGloba
    *  needs to be multiplied with gamma.
    * @return a new instance of BatchNormWithGlobalNormalizationGrad
    */
-  public <T extends Tensor> BatchNormWithGlobalNormalizationGrad<T> batchNormWithGlobalNormalizationGrad(
+  public <T extends TType> BatchNormWithGlobalNormalizationGrad<T> batchNormWithGlobalNormalizationGrad(
       Operand<T> t, Operand<T> m, Operand<T> v, Operand<T> gamma, Operand<T> backprop,
       Float varianceEpsilon, Boolean scaleAfterNormalization) {
     return BatchNormWithGlobalNormalizationGrad.create(scope, t, m, v, gamma, backprop, varianceEpsilon, scaleAfterNormalization);
@@ -235,7 +235,7 @@ public <T extends Tensor> BatchNormWithGlobalNormalizationGrad<T> batchNormWithG
    * @param options carries optional attributes values
    * @return a new instance of BiasAdd
    */
-  public <T extends Tensor> BiasAdd<T> biasAdd(Operand<T> value, Operand<T> bias,
+  public <T extends TType> BiasAdd<T> biasAdd(Operand<T> value, Operand<T> bias,
       BiasAdd.Options... options) {
     return BiasAdd.create(scope, value, bias, options);
   }
@@ -252,7 +252,7 @@ public <T extends Tensor> BiasAdd<T> biasAdd(Operand<T> value, Operand<T> bias,
    * @param options carries optional attributes values
    * @return a new instance of BiasAddGrad
    */
-  public <T extends Tensor> BiasAddGrad<T> biasAddGrad(Operand<T> outBackprop,
+  public <T extends TType> BiasAddGrad<T> biasAddGrad(Operand<T> outBackprop,
       BiasAddGrad.Options... options) {
     return BiasAddGrad.create(scope, outBackprop, options);
   }
@@ -313,7 +313,7 @@ public ComputeAccidentalHits computeAccidentalHits(Operand<TInt64> trueClasses,
    * @param options carries optional attributes values
    * @return a new instance of Conv2d
    */
-  public <T extends Tensor & TNumber> Conv2d<T> conv2d(Operand<T> input, Operand<T> filter,
+  public <T extends TNumber> Conv2d<T> conv2d(Operand<T> input, Operand<T> filter,
       List<Long> strides, String padding, Conv2d.Options... options) {
     return Conv2d.create(scope, input, filter, strides, padding, options);
   }
@@ -335,7 +335,7 @@ public <T extends Tensor & TNumber> Conv2d<T> conv2d(Operand<T> input, Operand<T
    * @param options carries optional attributes values
    * @return a new instance of Conv2dBackpropFilter
    */
-  public <T extends Tensor & TNumber> Conv2dBackpropFilter<T> conv2dBackpropFilter(Operand<T> input,
+  public <T extends TNumber> Conv2dBackpropFilter<T> conv2dBackpropFilter(Operand<T> input,
       Operand<TInt32> filterSizes, Operand<T> outBackprop, List<Long> strides, String padding,
       Conv2dBackpropFilter.Options... options) {
     return Conv2dBackpropFilter.create(scope, input, filterSizes, outBackprop, strides, padding, options);
@@ -358,9 +358,9 @@ public <T extends Tensor & TNumber> Conv2dBackpropFilter<T> conv2dBackpropFilter
    * @param options carries optional attributes values
    * @return a new instance of Conv2dBackpropInput
    */
-  public <T extends Tensor & TNumber> Conv2dBackpropInput<T> conv2dBackpropInput(
-      Operand<TInt32> inputSizes, Operand<T> filter, Operand<T> outBackprop, List<Long> strides,
-      String padding, Conv2dBackpropInput.Options... options) {
+  public <T extends TNumber> Conv2dBackpropInput<T> conv2dBackpropInput(Operand<TInt32> inputSizes,
+      Operand<T> filter, Operand<T> outBackprop, List<Long> strides, String padding,
+      Conv2dBackpropInput.Options... options) {
     return Conv2dBackpropInput.create(scope, inputSizes, filter, outBackprop, strides, padding, options);
   }
 
@@ -383,7 +383,7 @@ public <T extends Tensor & TNumber> Conv2dBackpropInput<T> conv2dBackpropInput(
    * @param options carries optional attributes values
    * @return a new instance of Conv3d
    */
-  public <T extends Tensor & TNumber> Conv3d<T> conv3d(Operand<T> input, Operand<T> filter,
+  public <T extends TNumber> Conv3d<T> conv3d(Operand<T> input, Operand<T> filter,
       List<Long> strides, String padding, Conv3d.Options... options) {
     return Conv3d.create(scope, input, filter, strides, padding, options);
   }
@@ -405,7 +405,7 @@ public <T extends Tensor & TNumber> Conv3d<T> conv3d(Operand<T> input, Operand<T
    * @param options carries optional attributes values
    * @return a new instance of Conv3dBackpropFilter
    */
-  public <T extends Tensor & TNumber> Conv3dBackpropFilter<T> conv3dBackpropFilter(Operand<T> input,
+  public <T extends TNumber> Conv3dBackpropFilter<T> conv3dBackpropFilter(Operand<T> input,
       Operand<TInt32> filterSizes, Operand<T> outBackprop, List<Long> strides, String padding,
       Conv3dBackpropFilter.Options... options) {
     return Conv3dBackpropFilter.create(scope, input, filterSizes, outBackprop, strides, padding, options);
@@ -428,7 +428,7 @@ public <T extends Tensor & TNumber> Conv3dBackpropFilter<T> conv3dBackpropFilter
    * @param options carries optional attributes values
    * @return a new instance of Conv3dBackpropInput
    */
-  public <U extends Tensor & TNumber, T extends Tensor & TNumber> Conv3dBackpropInput<U> conv3dBackpropInput(
+  public <U extends TNumber, T extends TNumber> Conv3dBackpropInput<U> conv3dBackpropInput(
       Operand<T> inputSizes, Operand<U> filter, Operand<U> outBackprop, List<Long> strides,
       String padding, Conv3dBackpropInput.Options... options) {
     return Conv3dBackpropInput.create(scope, inputSizes, filter, outBackprop, strides, padding, options);
@@ -451,8 +451,8 @@ public <U extends Tensor & TNumber, T extends Tensor & TNumber> Conv3dBackpropIn
    * @param options carries optional attributes values
    * @return a new instance of CtcBeamSearchDecoder
    */
-  public <T extends Tensor & TNumber> CtcBeamSearchDecoder<T> ctcBeamSearchDecoder(
-      Operand<T> inputs, Operand<TInt32> sequenceLength, Long beamWidth, Long topPaths,
+  public <T extends TNumber> CtcBeamSearchDecoder<T> ctcBeamSearchDecoder(Operand<T> inputs,
+      Operand<TInt32> sequenceLength, Long beamWidth, Long topPaths,
       CtcBeamSearchDecoder.Options... options) {
     return CtcBeamSearchDecoder.create(scope, inputs, sequenceLength, beamWidth, topPaths, options);
   }
@@ -476,7 +476,7 @@ public <T extends Tensor & TNumber> CtcBeamSearchDecoder<T> ctcBeamSearchDecoder
    * @param options carries optional attributes values
    * @return a new instance of CtcGreedyDecoder
    */
-  public <T extends Tensor & TNumber> CtcGreedyDecoder<T> ctcGreedyDecoder(Operand<T> inputs,
+  public <T extends TNumber> CtcGreedyDecoder<T> ctcGreedyDecoder(Operand<T> inputs,
       Operand<TInt32> sequenceLength, CtcGreedyDecoder.Options... options) {
     return CtcGreedyDecoder.create(scope, inputs, sequenceLength, options);
   }
@@ -497,9 +497,8 @@ public <T extends Tensor & TNumber> CtcGreedyDecoder<T> ctcGreedyDecoder(Operand
    * @param options carries optional attributes values
    * @return a new instance of CtcLoss
    */
-  public <T extends Tensor & TNumber> CtcLoss<T> ctcLoss(Operand<T> inputs,
-      Operand<TInt64> labelsIndices, Operand<TInt32> labelsValues, Operand<TInt32> sequenceLength,
-      CtcLoss.Options... options) {
+  public <T extends TNumber> CtcLoss<T> ctcLoss(Operand<T> inputs, Operand<TInt64> labelsIndices,
+      Operand<TInt32> labelsValues, Operand<TInt32> sequenceLength, CtcLoss.Options... options) {
     return CtcLoss.create(scope, inputs, labelsIndices, labelsValues, sequenceLength, options);
   }
 
@@ -546,7 +545,7 @@ public <T extends Tensor & TNumber> CtcLoss<T> ctcLoss(Operand<T> inputs,
    * @param options carries optional attributes values
    * @return a new instance of CudnnRNNCanonicalToParams
    */
-  public <T extends Tensor & TNumber> CudnnRNNCanonicalToParams<T> cudnnRNNCanonicalToParams(
+  public <T extends TNumber> CudnnRNNCanonicalToParams<T> cudnnRNNCanonicalToParams(
       Operand<TInt32> numLayers, Operand<TInt32> numUnits, Operand<TInt32> inputSize,
       Iterable<Operand<T>> weights, Iterable<Operand<T>> biases,
       CudnnRNNCanonicalToParams.Options... options) {
@@ -597,7 +596,7 @@ public <T extends Tensor & TNumber> CudnnRNNCanonicalToParams<T> cudnnRNNCanonic
    * @param options carries optional attributes values
    * @return a new instance of CudnnRNNParamsToCanonical
    */
-  public <T extends Tensor & TNumber> CudnnRNNParamsToCanonical<T> cudnnRNNParamsToCanonical(
+  public <T extends TNumber> CudnnRNNParamsToCanonical<T> cudnnRNNParamsToCanonical(
       Operand<TInt32> numLayers, Operand<TInt32> numUnits, Operand<TInt32> inputSize,
       Operand<T> params, Long numParamsWeights, Long numParamsBiases,
       CudnnRNNParamsToCanonical.Options... options) {
@@ -638,7 +637,7 @@ public <T extends Tensor & TNumber> CudnnRNNParamsToCanonical<T> cudnnRNNParamsT
    * @param options carries optional attributes values
    * @return a new instance of CudnnRnnParamsSize
    */
-  public <U extends Tensor & TNumber, T extends Tensor & TNumber> CudnnRnnParamsSize<U> cudnnRnnParamsSize(
+  public <U extends TNumber, T extends TNumber> CudnnRnnParamsSize<U> cudnnRnnParamsSize(
       Operand<TInt32> numLayers, Operand<TInt32> numUnits, Operand<TInt32> inputSize, DataType<T> T,
       DataType<U> S, CudnnRnnParamsSize.Options... options) {
     return CudnnRnnParamsSize.create(scope, numLayers, numUnits, inputSize, T, S, options);
@@ -655,7 +654,7 @@ public <U extends Tensor & TNumber, T extends Tensor & TNumber> CudnnRnnParamsSi
    * @param options carries optional attributes values
    * @return a new instance of DataFormatDimMap
    */
-  public <T extends Tensor & TNumber> DataFormatDimMap<T> dataFormatDimMap(Operand<T> x,
+  public <T extends TNumber> DataFormatDimMap<T> dataFormatDimMap(Operand<T> x,
       DataFormatDimMap.Options... options) {
     return DataFormatDimMap.create(scope, x, options);
   }
@@ -670,7 +669,7 @@ public <T extends Tensor & TNumber> DataFormatDimMap<T> dataFormatDimMap(Operand
    * @param options carries optional attributes values
    * @return a new instance of DataFormatVecPermute
    */
-  public <T extends Tensor & TNumber> DataFormatVecPermute<T> dataFormatVecPermute(Operand<T> x,
+  public <T extends TNumber> DataFormatVecPermute<T> dataFormatVecPermute(Operand<T> x,
       DataFormatVecPermute.Options... options) {
     return DataFormatVecPermute.create(scope, x, options);
   }
@@ -763,7 +762,7 @@ public <T extends Tensor & TNumber> DataFormatVecPermute<T> dataFormatVecPermute
    * @param options carries optional attributes values
    * @return a new instance of DepthToSpace
    */
-  public <T extends Tensor> DepthToSpace<T> depthToSpace(Operand<T> input, Long blockSize,
+  public <T extends TType> DepthToSpace<T> depthToSpace(Operand<T> input, Long blockSize,
       DepthToSpace.Options... options) {
     return DepthToSpace.create(scope, input, blockSize, options);
   }
@@ -797,8 +796,8 @@ public <T extends Tensor> DepthToSpace<T> depthToSpace(Operand<T> input, Long bl
    * @param options carries optional attributes values
    * @return a new instance of DepthwiseConv2dNative
    */
-  public <T extends Tensor & TNumber> DepthwiseConv2dNative<T> depthwiseConv2dNative(
-      Operand<T> input, Operand<T> filter, List<Long> strides, String padding,
+  public <T extends TNumber> DepthwiseConv2dNative<T> depthwiseConv2dNative(Operand<T> input,
+      Operand<T> filter, List<Long> strides, String padding,
       DepthwiseConv2dNative.Options... options) {
     return DepthwiseConv2dNative.create(scope, input, filter, strides, padding, options);
   }
@@ -823,7 +822,7 @@ public <T extends Tensor & TNumber> DepthwiseConv2dNative<T> depthwiseConv2dNati
    * @param options carries optional attributes values
    * @return a new instance of DepthwiseConv2dNativeBackpropFilter
    */
-  public <T extends Tensor & TNumber> DepthwiseConv2dNativeBackpropFilter<T> depthwiseConv2dNativeBackpropFilter(
+  public <T extends TNumber> DepthwiseConv2dNativeBackpropFilter<T> depthwiseConv2dNativeBackpropFilter(
       Operand<T> input, Operand<TInt32> filterSizes, Operand<T> outBackprop, List<Long> strides,
       String padding, DepthwiseConv2dNativeBackpropFilter.Options... options) {
     return DepthwiseConv2dNativeBackpropFilter.create(scope, input, filterSizes, outBackprop, strides, padding, options);
@@ -848,7 +847,7 @@ public <T extends Tensor & TNumber> DepthwiseConv2dNativeBackpropFilter<T> depth
    * @param options carries optional attributes values
    * @return a new instance of DepthwiseConv2dNativeBackpropInput
    */
-  public <T extends Tensor & TNumber> DepthwiseConv2dNativeBackpropInput<T> depthwiseConv2dNativeBackpropInput(
+  public <T extends TNumber> DepthwiseConv2dNativeBackpropInput<T> depthwiseConv2dNativeBackpropInput(
       Operand<TInt32> inputSizes, Operand<T> filter, Operand<T> outBackprop, List<Long> strides,
       String padding, DepthwiseConv2dNativeBackpropInput.Options... options) {
     return DepthwiseConv2dNativeBackpropInput.create(scope, inputSizes, filter, outBackprop, strides, padding, options);
@@ -891,7 +890,7 @@ public <T extends Tensor & TNumber> DepthwiseConv2dNativeBackpropInput<T> depthw
    * @param padding The type of padding algorithm to use.
    * @return a new instance of Dilation2d
    */
-  public <T extends Tensor & TNumber> Dilation2d<T> dilation2d(Operand<T> input, Operand<T> filter,
+  public <T extends TNumber> Dilation2d<T> dilation2d(Operand<T> input, Operand<T> filter,
       List<Long> strides, List<Long> rates, String padding) {
     return Dilation2d.create(scope, input, filter, strides, rates, padding);
   }
@@ -910,9 +909,9 @@ public <T extends Tensor & TNumber> Dilation2d<T> dilation2d(Operand<T> input, O
    * @param padding The type of padding algorithm to use.
    * @return a new instance of Dilation2dBackpropFilter
    */
-  public <T extends Tensor & TNumber> Dilation2dBackpropFilter<T> dilation2dBackpropFilter(
-      Operand<T> input, Operand<T> filter, Operand<T> outBackprop, List<Long> strides,
-      List<Long> rates, String padding) {
+  public <T extends TNumber> Dilation2dBackpropFilter<T> dilation2dBackpropFilter(Operand<T> input,
+      Operand<T> filter, Operand<T> outBackprop, List<Long> strides, List<Long> rates,
+      String padding) {
     return Dilation2dBackpropFilter.create(scope, input, filter, outBackprop, strides, rates, padding);
   }
 
@@ -930,9 +929,9 @@ public <T extends Tensor & TNumber> Dilation2dBackpropFilter<T> dilation2dBackpr
    * @param padding The type of padding algorithm to use.
    * @return a new instance of Dilation2dBackpropInput
    */
-  public <T extends Tensor & TNumber> Dilation2dBackpropInput<T> dilation2dBackpropInput(
-      Operand<T> input, Operand<T> filter, Operand<T> outBackprop, List<Long> strides,
-      List<Long> rates, String padding) {
+  public <T extends TNumber> Dilation2dBackpropInput<T> dilation2dBackpropInput(Operand<T> input,
+      Operand<T> filter, Operand<T> outBackprop, List<Long> strides, List<Long> rates,
+      String padding) {
     return Dilation2dBackpropInput.create(scope, input, filter, outBackprop, strides, rates, padding);
   }
 
@@ -946,7 +945,7 @@ public <T extends Tensor & TNumber> Dilation2dBackpropInput<T> dilation2dBackpro
    * @param features
    * @return a new instance of Elu
    */
-  public <T extends Tensor & TNumber> Elu<T> elu(Operand<T> features) {
+  public <T extends TNumber> Elu<T> elu(Operand<T> features) {
     return Elu.create(scope, features);
   }
 
@@ -1004,7 +1003,7 @@ public FixedUnigramCandidateSampler fixedUnigramCandidateSampler(Operand<TInt64>
    * @param options carries optional attributes values
    * @return a new instance of FractionalAvgPool
    */
-  public <T extends Tensor & TNumber> FractionalAvgPool<T> fractionalAvgPool(Operand<T> value,
+  public <T extends TNumber> FractionalAvgPool<T> fractionalAvgPool(Operand<T> value,
       List<Float> poolingRatio, FractionalAvgPool.Options... options) {
     return FractionalAvgPool.create(scope, value, poolingRatio, options);
   }
@@ -1052,7 +1051,7 @@ public <T extends Tensor & TNumber> FractionalAvgPool<T> fractionalAvgPool(Opera
    * @param options carries optional attributes values
    * @return a new instance of FractionalMaxPool
    */
-  public <T extends Tensor & TNumber> FractionalMaxPool<T> fractionalMaxPool(Operand<T> value,
+  public <T extends TNumber> FractionalMaxPool<T> fractionalMaxPool(Operand<T> value,
       List<Float> poolingRatio, FractionalMaxPool.Options... options) {
     return FractionalMaxPool.create(scope, value, poolingRatio, options);
   }
@@ -1075,8 +1074,8 @@ public <T extends Tensor & TNumber> FractionalMaxPool<T> fractionalMaxPool(Opera
    * @param options carries optional attributes values
    * @return a new instance of FusedBatchNorm
    */
-  public <T extends Tensor & TNumber, U extends Tensor & TNumber> FusedBatchNorm<T, U> fusedBatchNorm(
-      Operand<T> x, Operand<U> scale, Operand<U> offset, Operand<U> mean, Operand<U> variance,
+  public <T extends TNumber, U extends TNumber> FusedBatchNorm<T, U> fusedBatchNorm(Operand<T> x,
+      Operand<U> scale, Operand<U> offset, Operand<U> mean, Operand<U> variance,
       FusedBatchNorm.Options... options) {
     return FusedBatchNorm.create(scope, x, scale, offset, mean, variance, options);
   }
@@ -1107,7 +1106,7 @@ public <T extends Tensor & TNumber, U extends Tensor & TNumber> FusedBatchNorm<T
    * @param options carries optional attributes values
    * @return a new instance of FusedBatchNormGrad
    */
-  public <T extends Tensor & TNumber, U extends Tensor & TNumber> FusedBatchNormGrad<T, U> fusedBatchNormGrad(
+  public <T extends TNumber, U extends TNumber> FusedBatchNormGrad<T, U> fusedBatchNormGrad(
       Operand<T> yBackprop, Operand<T> x, Operand<TFloat32> scale, Operand<U> reserveSpace1,
       Operand<U> reserveSpace2, Operand<U> reserveSpace3, FusedBatchNormGrad.Options... options) {
     return FusedBatchNormGrad.create(scope, yBackprop, x, scale, reserveSpace1, reserveSpace2, reserveSpace3, options);
@@ -1140,7 +1139,7 @@ public <T extends Tensor & TNumber, U extends Tensor & TNumber> FusedBatchNormGr
    * @param padding The type of padding algorithm to use.
    * @return a new instance of FusedPadConv2d
    */
-  public <T extends Tensor & TNumber> FusedPadConv2d<T> fusedPadConv2d(Operand<T> input,
+  public <T extends TNumber> FusedPadConv2d<T> fusedPadConv2d(Operand<T> input,
       Operand<TInt32> paddings, Operand<T> filter, String mode, List<Long> strides,
       String padding) {
     return FusedPadConv2d.create(scope, input, paddings, filter, mode, strides, padding);
@@ -1175,9 +1174,9 @@ public <T extends Tensor & TNumber> FusedPadConv2d<T> fusedPadConv2d(Operand<T>
    * @param options carries optional attributes values
    * @return a new instance of FusedResizeAndPadConv2d
    */
-  public <T extends Tensor & TNumber> FusedResizeAndPadConv2d<T> fusedResizeAndPadConv2d(
-      Operand<T> input, Operand<TInt32> size, Operand<TInt32> paddings, Operand<T> filter,
-      String mode, List<Long> strides, String padding, FusedResizeAndPadConv2d.Options... options) {
+  public <T extends TNumber> FusedResizeAndPadConv2d<T> fusedResizeAndPadConv2d(Operand<T> input,
+      Operand<TInt32> size, Operand<TInt32> paddings, Operand<T> filter, String mode,
+      List<Long> strides, String padding, FusedResizeAndPadConv2d.Options... options) {
     return FusedResizeAndPadConv2d.create(scope, input, size, paddings, filter, mode, strides, padding, options);
   }
 
@@ -1204,8 +1203,8 @@ public <T extends Tensor & TNumber> FusedResizeAndPadConv2d<T> fusedResizeAndPad
    * @param k Number of top elements to look at for computing precision.
    * @return a new instance of InTopK
    */
-  public <T extends Tensor & TNumber> InTopK inTopK(Operand<TFloat32> predictions,
-      Operand<T> targets, Operand<T> k) {
+  public <T extends TNumber> InTopK inTopK(Operand<TFloat32> predictions, Operand<T> targets,
+      Operand<T> k) {
     return InTopK.create(scope, predictions, targets, k);
   }
 
@@ -1220,7 +1219,7 @@ public <T extends Tensor & TNumber> InTopK inTopK(Operand<TFloat32> predictions,
    * @param t Typically 2-D, but may have any dimensions.
    * @return a new instance of L2Loss
    */
-  public <T extends Tensor & TNumber> L2Loss<T> l2Loss(Operand<T> t) {
+  public <T extends TNumber> L2Loss<T> l2Loss(Operand<T> t) {
     return L2Loss.create(scope, t);
   }
 
@@ -1287,7 +1286,7 @@ public LearnedUnigramCandidateSampler learnedUnigramCandidateSampler(Operand<TIn
    * @param options carries optional attributes values
    * @return a new instance of LocalResponseNormalization
    */
-  public <T extends Tensor & TNumber> LocalResponseNormalization<T> localResponseNormalization(
+  public <T extends TNumber> LocalResponseNormalization<T> localResponseNormalization(
       Operand<T> input, LocalResponseNormalization.Options... options) {
     return LocalResponseNormalization.create(scope, input, options);
   }
@@ -1303,7 +1302,7 @@ public <T extends Tensor & TNumber> LocalResponseNormalization<T> localResponseN
    * @param logits 2-D with shape `[batch_size, num_classes]`.
    * @return a new instance of LogSoftmax
    */
-  public <T extends Tensor & TNumber> LogSoftmax<T> logSoftmax(Operand<T> logits) {
+  public <T extends TNumber> LogSoftmax<T> logSoftmax(Operand<T> logits) {
     return LogSoftmax.create(scope, logits);
   }
 
@@ -1319,7 +1318,7 @@ public <T extends Tensor & TNumber> LogSoftmax<T> logSoftmax(Operand<T> logits)
    * @param options carries optional attributes values
    * @return a new instance of MaxPool
    */
-  public <T extends Tensor> MaxPool<T> maxPool(Operand<T> input, Operand<TInt32> ksize,
+  public <T extends TType> MaxPool<T> maxPool(Operand<T> input, Operand<TInt32> ksize,
       Operand<TInt32> strides, String padding, MaxPool.Options... options) {
     return MaxPool.create(scope, input, ksize, strides, padding, options);
   }
@@ -1337,7 +1336,7 @@ public <T extends Tensor> MaxPool<T> maxPool(Operand<T> input, Operand<TInt32> k
    * @param options carries optional attributes values
    * @return a new instance of MaxPool3d
    */
-  public <T extends Tensor & TNumber> MaxPool3d<T> maxPool3d(Operand<T> input, List<Long> ksize,
+  public <T extends TNumber> MaxPool3d<T> maxPool3d(Operand<T> input, List<Long> ksize,
       List<Long> strides, String padding, MaxPool3d.Options... options) {
     return MaxPool3d.create(scope, input, ksize, strides, padding, options);
   }
@@ -1357,9 +1356,9 @@ public <T extends Tensor & TNumber> MaxPool3d<T> maxPool3d(Operand<T> input, Lis
    * @param options carries optional attributes values
    * @return a new instance of MaxPool3dGrad
    */
-  public <U extends Tensor & TNumber, T extends Tensor & TNumber> MaxPool3dGrad<U> maxPool3dGrad(
-      Operand<T> origInput, Operand<T> origOutput, Operand<U> grad, List<Long> ksize,
-      List<Long> strides, String padding, MaxPool3dGrad.Options... options) {
+  public <U extends TNumber, T extends TNumber> MaxPool3dGrad<U> maxPool3dGrad(Operand<T> origInput,
+      Operand<T> origOutput, Operand<U> grad, List<Long> ksize, List<Long> strides, String padding,
+      MaxPool3dGrad.Options... options) {
     return MaxPool3dGrad.create(scope, origInput, origOutput, grad, ksize, strides, padding, options);
   }
 
@@ -1378,7 +1377,7 @@ public <U extends Tensor & TNumber, T extends Tensor & TNumber> MaxPool3dGrad<U>
    * @param options carries optional attributes values
    * @return a new instance of MaxPool3dGradGrad
    */
-  public <T extends Tensor & TNumber> MaxPool3dGradGrad<T> maxPool3dGradGrad(Operand<T> origInput,
+  public <T extends TNumber> MaxPool3dGradGrad<T> maxPool3dGradGrad(Operand<T> origInput,
       Operand<T> origOutput, Operand<T> grad, List<Long> ksize, List<Long> strides, String padding,
       MaxPool3dGradGrad.Options... options) {
     return MaxPool3dGradGrad.create(scope, origInput, origOutput, grad, ksize, strides, padding, options);
@@ -1398,9 +1397,9 @@ public <T extends Tensor & TNumber> MaxPool3dGradGrad<T> maxPool3dGradGrad(Opera
    * @param options carries optional attributes values
    * @return a new instance of MaxPoolGrad
    */
-  public <T extends Tensor & TNumber> MaxPoolGrad<T> maxPoolGrad(Operand<T> origInput,
-      Operand<T> origOutput, Operand<T> grad, Operand<TInt32> ksize, Operand<TInt32> strides,
-      String padding, MaxPoolGrad.Options... options) {
+  public <T extends TNumber> MaxPoolGrad<T> maxPoolGrad(Operand<T> origInput, Operand<T> origOutput,
+      Operand<T> grad, Operand<TInt32> ksize, Operand<TInt32> strides, String padding,
+      MaxPoolGrad.Options... options) {
     return MaxPoolGrad.create(scope, origInput, origOutput, grad, ksize, strides, padding, options);
   }
 
@@ -1418,7 +1417,7 @@ public <T extends Tensor & TNumber> MaxPoolGrad<T> maxPoolGrad(Operand<T> origIn
    * @param options carries optional attributes values
    * @return a new instance of MaxPoolGradGrad
    */
-  public <T extends Tensor & TNumber> MaxPoolGradGrad<T> maxPoolGradGrad(Operand<T> origInput,
+  public <T extends TNumber> MaxPoolGradGrad<T> maxPoolGradGrad(Operand<T> origInput,
       Operand<T> origOutput, Operand<T> grad, Operand<TInt32> ksize, Operand<TInt32> strides,
       String padding, MaxPoolGradGrad.Options... options) {
     return MaxPoolGradGrad.create(scope, origInput, origOutput, grad, ksize, strides, padding, options);
@@ -1439,7 +1438,7 @@ public <T extends Tensor & TNumber> MaxPoolGradGrad<T> maxPoolGradGrad(Operand<T
    * @param options carries optional attributes values
    * @return a new instance of MaxPoolGradGradWithArgmax
    */
-  public <T extends Tensor & TNumber, U extends Tensor & TNumber> MaxPoolGradGradWithArgmax<T> maxPoolGradGradWithArgmax(
+  public <T extends TNumber, U extends TNumber> MaxPoolGradGradWithArgmax<T> maxPoolGradGradWithArgmax(
       Operand<T> input, Operand<T> grad, Operand<U> argmax, List<Long> ksize, List<Long> strides,
       String padding, MaxPoolGradGradWithArgmax.Options... options) {
     return MaxPoolGradGradWithArgmax.create(scope, input, grad, argmax, ksize, strides, padding, options);
@@ -1468,9 +1467,8 @@ public <T extends Tensor & TNumber, U extends Tensor & TNumber> MaxPoolGradGradW
    * @param options carries optional attributes values
    * @return a new instance of MaxPoolWithArgmax
    */
-  public <T extends Tensor & TNumber> MaxPoolWithArgmax<T, TInt64> maxPoolWithArgmax(
-      Operand<T> input, List<Long> ksize, List<Long> strides, String padding,
-      MaxPoolWithArgmax.Options... options) {
+  public <T extends TNumber> MaxPoolWithArgmax<T, TInt64> maxPoolWithArgmax(Operand<T> input,
+      List<Long> ksize, List<Long> strides, String padding, MaxPoolWithArgmax.Options... options) {
     return MaxPoolWithArgmax.create(scope, input, ksize, strides, padding, options);
   }
 
@@ -1498,7 +1496,7 @@ public <T extends Tensor & TNumber> MaxPoolWithArgmax<T, TInt64> maxPoolWithArgm
    * @param options carries optional attributes values
    * @return a new instance of MaxPoolWithArgmax
    */
-  public <T extends Tensor & TNumber, U extends Tensor & TNumber> MaxPoolWithArgmax<T, U> maxPoolWithArgmax(
+  public <T extends TNumber, U extends TNumber> MaxPoolWithArgmax<T, U> maxPoolWithArgmax(
       Operand<T> input, List<Long> ksize, List<Long> strides, DataType<U> Targmax, String padding,
       MaxPoolWithArgmax.Options... options) {
     return MaxPoolWithArgmax.create(scope, input, ksize, strides, Targmax, padding, options);
@@ -1522,7 +1520,7 @@ public <T extends Tensor & TNumber, U extends Tensor & TNumber> MaxPoolWithArgma
    * @param options carries optional attributes values
    * @return a new instance of NthElement
    */
-  public <T extends Tensor & TNumber> NthElement<T> nthElement(Operand<T> input, Operand<TInt32> n,
+  public <T extends TNumber> NthElement<T> nthElement(Operand<T> input, Operand<TInt32> n,
       NthElement.Options... options) {
     return NthElement.create(scope, input, n, options);
   }
@@ -1541,7 +1539,7 @@ public <T extends Tensor & TNumber> NthElement<T> nthElement(Operand<T> input, O
    * @param padding The type of padding algorithm to use.
    * @return a new instance of QuantizedAvgPool
    */
-  public <T extends Tensor> QuantizedAvgPool<T> quantizedAvgPool(Operand<T> input,
+  public <T extends TType> QuantizedAvgPool<T> quantizedAvgPool(Operand<T> input,
       Operand<TFloat32> minInput, Operand<TFloat32> maxInput, List<Long> ksize, List<Long> strides,
       String padding) {
     return QuantizedAvgPool.create(scope, input, minInput, maxInput, ksize, strides, padding);
@@ -1582,7 +1580,7 @@ public <T extends Tensor> QuantizedAvgPool<T> quantizedAvgPool(Operand<T> input,
    *  needs to be multiplied with gamma.
    * @return a new instance of QuantizedBatchNormWithGlobalNormalization
    */
-  public <U extends Tensor, T extends Tensor> QuantizedBatchNormWithGlobalNormalization<U> quantizedBatchNormWithGlobalNormalization(
+  public <U extends TType, T extends TType> QuantizedBatchNormWithGlobalNormalization<U> quantizedBatchNormWithGlobalNormalization(
       Operand<T> t, Operand<TFloat32> tMin, Operand<TFloat32> tMax, Operand<T> m,
       Operand<TFloat32> mMin, Operand<TFloat32> mMax, Operand<T> v, Operand<TFloat32> vMin,
       Operand<TFloat32> vMax, Operand<T> beta, Operand<TFloat32> betaMin, Operand<TFloat32> betaMax,
@@ -1606,7 +1604,7 @@ public <U extends Tensor, T extends Tensor> QuantizedBatchNormWithGlobalNormaliz
    * @param outType
    * @return a new instance of QuantizedBiasAdd
    */
-  public <V extends Tensor, T extends Tensor, U extends Tensor> QuantizedBiasAdd<V> quantizedBiasAdd(
+  public <V extends TType, T extends TType, U extends TType> QuantizedBiasAdd<V> quantizedBiasAdd(
       Operand<T> input, Operand<U> bias, Operand<TFloat32> minInput, Operand<TFloat32> maxInput,
       Operand<TFloat32> minBias, Operand<TFloat32> maxBias, DataType<V> outType) {
     return QuantizedBiasAdd.create(scope, input, bias, minInput, maxInput, minBias, maxBias, outType);
@@ -1634,7 +1632,7 @@ public <V extends Tensor, T extends Tensor, U extends Tensor> QuantizedBiasAdd<V
    * @param options carries optional attributes values
    * @return a new instance of QuantizedConv2d
    */
-  public <V extends Tensor, T extends Tensor, U extends Tensor> QuantizedConv2d<V> quantizedConv2d(
+  public <V extends TType, T extends TType, U extends TType> QuantizedConv2d<V> quantizedConv2d(
       Operand<T> input, Operand<U> filter, Operand<TFloat32> minInput, Operand<TFloat32> maxInput,
       Operand<TFloat32> minFilter, Operand<TFloat32> maxFilter, DataType<V> outType,
       List<Long> strides, String padding, QuantizedConv2d.Options... options) {
@@ -1651,7 +1649,7 @@ public <V extends Tensor, T extends Tensor, U extends Tensor> QuantizedConv2d<V>
    * @param options carries optional attributes values
    * @return a new instance of QuantizedInstanceNorm
    */
-  public <T extends Tensor> QuantizedInstanceNorm<T> quantizedInstanceNorm(Operand<T> x,
+  public <T extends TType> QuantizedInstanceNorm<T> quantizedInstanceNorm(Operand<T> x,
       Operand<TFloat32> xMin, Operand<TFloat32> xMax, QuantizedInstanceNorm.Options... options) {
     return QuantizedInstanceNorm.create(scope, x, xMin, xMax, options);
   }
@@ -1670,7 +1668,7 @@ public <T extends Tensor> QuantizedInstanceNorm<T> quantizedInstanceNorm(Operand
    * @param padding The type of padding algorithm to use.
    * @return a new instance of QuantizedMaxPool
    */
-  public <T extends Tensor> QuantizedMaxPool<T> quantizedMaxPool(Operand<T> input,
+  public <T extends TType> QuantizedMaxPool<T> quantizedMaxPool(Operand<T> input,
       Operand<TFloat32> minInput, Operand<TFloat32> maxInput, List<Long> ksize, List<Long> strides,
       String padding) {
     return QuantizedMaxPool.create(scope, input, minInput, maxInput, ksize, strides, padding);
@@ -1686,7 +1684,7 @@ public <T extends Tensor> QuantizedMaxPool<T> quantizedMaxPool(Operand<T> input,
    * @param outType
    * @return a new instance of QuantizedRelu
    */
-  public <U extends Tensor, T extends Tensor> QuantizedRelu<U> quantizedRelu(Operand<T> features,
+  public <U extends TType, T extends TType> QuantizedRelu<U> quantizedRelu(Operand<T> features,
       Operand<TFloat32> minFeatures, Operand<TFloat32> maxFeatures, DataType<U> outType) {
     return QuantizedRelu.create(scope, features, minFeatures, maxFeatures, outType);
   }
@@ -1701,7 +1699,7 @@ public <U extends Tensor, T extends Tensor> QuantizedRelu<U> quantizedRelu(Opera
    * @param outType
    * @return a new instance of QuantizedRelu6
    */
-  public <U extends Tensor, T extends Tensor> QuantizedRelu6<U> quantizedRelu6(Operand<T> features,
+  public <U extends TType, T extends TType> QuantizedRelu6<U> quantizedRelu6(Operand<T> features,
       Operand<TFloat32> minFeatures, Operand<TFloat32> maxFeatures, DataType<U> outType) {
     return QuantizedRelu6.create(scope, features, minFeatures, maxFeatures, outType);
   }
@@ -1717,7 +1715,7 @@ public <U extends Tensor, T extends Tensor> QuantizedRelu6<U> quantizedRelu6(Ope
    * @param outType
    * @return a new instance of QuantizedReluX
    */
-  public <U extends Tensor, T extends Tensor> QuantizedReluX<U> quantizedReluX(Operand<T> features,
+  public <U extends TType, T extends TType> QuantizedReluX<U> quantizedReluX(Operand<T> features,
       Operand<TFloat32> maxValue, Operand<TFloat32> minFeatures, Operand<TFloat32> maxFeatures,
       DataType<U> outType) {
     return QuantizedReluX.create(scope, features, maxValue, minFeatures, maxFeatures, outType);
@@ -1735,7 +1733,7 @@ public <U extends Tensor, T extends Tensor> QuantizedReluX<U> quantizedReluX(Ope
    * @param features
    * @return a new instance of Relu
    */
-  public <T extends Tensor> Relu<T> relu(Operand<T> features) {
+  public <T extends TType> Relu<T> relu(Operand<T> features) {
     return Relu.create(scope, features);
   }
 
@@ -1746,7 +1744,7 @@ public <T extends Tensor> Relu<T> relu(Operand<T> features) {
    * @param features
    * @return a new instance of Relu6
    */
-  public <T extends Tensor & TNumber> Relu6<T> relu6(Operand<T> features) {
+  public <T extends TNumber> Relu6<T> relu6(Operand<T> features) {
     return Relu6.create(scope, features);
   }
 
@@ -1765,7 +1763,7 @@ public <T extends Tensor & TNumber> Relu6<T> relu6(Operand<T> features) {
    * @param features
    * @return a new instance of Selu
    */
-  public <T extends Tensor & TNumber> Selu<T> selu(Operand<T> features) {
+  public <T extends TNumber> Selu<T> selu(Operand<T> features) {
     return Selu.create(scope, features);
   }
 
@@ -1780,7 +1778,7 @@ public <T extends Tensor & TNumber> Selu<T> selu(Operand<T> features) {
    * @param logits 2-D with shape `[batch_size, num_classes]`.
    * @return a new instance of Softmax
    */
-  public <T extends Tensor & TNumber> Softmax<T> softmax(Operand<T> logits) {
+  public <T extends TNumber> Softmax<T> softmax(Operand<T> logits) {
     return Softmax.create(scope, logits);
   }
 
@@ -1796,7 +1794,7 @@ public <T extends Tensor & TNumber> Softmax<T> softmax(Operand<T> logits) {
    *  probability distribution.
    * @return a new instance of SoftmaxCrossEntropyWithLogits
    */
-  public <T extends Tensor & TNumber> SoftmaxCrossEntropyWithLogits<T> softmaxCrossEntropyWithLogits(
+  public <T extends TNumber> SoftmaxCrossEntropyWithLogits<T> softmaxCrossEntropyWithLogits(
       Operand<T> features, Operand<T> labels) {
     return SoftmaxCrossEntropyWithLogits.create(scope, features, labels);
   }
@@ -1808,7 +1806,7 @@ public <T extends Tensor & TNumber> SoftmaxCrossEntropyWithLogits<T> softmaxCros
    * @param features
    * @return a new instance of Softsign
    */
-  public <T extends Tensor & TNumber> Softsign<T> softsign(Operand<T> features) {
+  public <T extends TNumber> Softsign<T> softsign(Operand<T> features) {
     return Softsign.create(scope, features);
   }
 
@@ -1897,8 +1895,8 @@ public <T extends Tensor & TNumber> Softsign<T> softsign(Operand<T> features) {
    * @param blockSize
    * @return a new instance of SpaceToBatch
    */
-  public <T extends Tensor, U extends Tensor & TNumber> SpaceToBatch<T> spaceToBatch(
-      Operand<T> input, Operand<U> paddings, Long blockSize) {
+  public <T extends TType, U extends TNumber> SpaceToBatch<T> spaceToBatch(Operand<T> input,
+      Operand<U> paddings, Long blockSize) {
     return SpaceToBatch.create(scope, input, paddings, blockSize);
   }
 
@@ -1984,7 +1982,7 @@ public <T extends Tensor, U extends Tensor & TNumber> SpaceToBatch<T> spaceToBat
    * @param options carries optional attributes values
    * @return a new instance of SpaceToDepth
    */
-  public <T extends Tensor> SpaceToDepth<T> spaceToDepth(Operand<T> input, Long blockSize,
+  public <T extends TType> SpaceToDepth<T> spaceToDepth(Operand<T> input, Long blockSize,
       SpaceToDepth.Options... options) {
     return SpaceToDepth.create(scope, input, blockSize, options);
   }
@@ -2005,7 +2003,7 @@ public <T extends Tensor> SpaceToDepth<T> spaceToDepth(Operand<T> input, Long bl
    *  This is the label for the given minibatch entry.
    * @return a new instance of SparseSoftmaxCrossEntropyWithLogits
    */
-  public <T extends Tensor & TNumber, U extends Tensor & TNumber> SparseSoftmaxCrossEntropyWithLogits<T> sparseSoftmaxCrossEntropyWithLogits(
+  public <T extends TNumber, U extends TNumber> SparseSoftmaxCrossEntropyWithLogits<T> sparseSoftmaxCrossEntropyWithLogits(
       Operand<T> features, Operand<U> labels) {
     return SparseSoftmaxCrossEntropyWithLogits.create(scope, features, labels);
   }
@@ -2031,7 +2029,7 @@ public <T extends Tensor & TNumber, U extends Tensor & TNumber> SparseSoftmaxCro
    * @param options carries optional attributes values
    * @return a new instance of TopK
    */
-  public <T extends Tensor & TNumber> TopK<T> topK(Operand<T> input, Operand<TInt32> k,
+  public <T extends TNumber> TopK<T> topK(Operand<T> input, Operand<TInt32> k,
       TopK.Options... options) {
     return TopK.create(scope, input, k, options);
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/Ops.java b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/Ops.java
index fa1b27e62bd..b17d3c81d31 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/Ops.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/Ops.java
@@ -23,7 +23,6 @@
 import org.tensorflow.EagerSession;
 import org.tensorflow.ExecutionEnvironment;
 import org.tensorflow.Operand;
-import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.BooleanNdArray;
 import org.tensorflow.ndarray.ByteNdArray;
 import org.tensorflow.ndarray.DoubleNdArray;
@@ -395,7 +394,7 @@ public Abort abort(Abort.Options... options) {
    * @param options carries optional attributes values
    * @return a new instance of All
    */
-  public <T extends Tensor & TNumber> All all(Operand<TBool> input, Operand<T> axis,
+  public <T extends TNumber> All all(Operand<TBool> input, Operand<T> axis,
       All.Options... options) {
     return All.create(scope, input, axis, options);
   }
@@ -414,7 +413,7 @@ public <T extends Tensor & TNumber> All all(Operand<TBool> input, Operand<T> axi
    * @param options carries optional attributes values
    * @return a new instance of Any
    */
-  public <T extends Tensor & TNumber> Any any(Operand<TBool> input, Operand<T> axis,
+  public <T extends TNumber> Any any(Operand<TBool> input, Operand<T> axis,
       Any.Options... options) {
     return Any.create(scope, input, axis, options);
   }
@@ -537,7 +536,7 @@ public AssertThat assertThat(Operand<TBool> condition, Iterable<Operand<?>> data
    * @param options carries optional attributes values
    * @return a new instance of Assign
    */
-  public <T extends Tensor> Assign<T> assign(Operand<T> ref, Operand<T> value,
+  public <T extends TType> Assign<T> assign(Operand<T> ref, Operand<T> value,
       Assign.Options... options) {
     return Assign.create(scope, ref, value, options);
   }
@@ -554,7 +553,7 @@ public <T extends Tensor> Assign<T> assign(Operand<T> ref, Operand<T> value,
    * @param options carries optional attributes values
    * @return a new instance of AssignAdd
    */
-  public <T extends Tensor> AssignAdd<T> assignAdd(Operand<T> ref, Operand<T> value,
+  public <T extends TType> AssignAdd<T> assignAdd(Operand<T> ref, Operand<T> value,
       AssignAdd.Options... options) {
     return AssignAdd.create(scope, ref, value, options);
   }
@@ -569,7 +568,7 @@ public <T extends Tensor> AssignAdd<T> assignAdd(Operand<T> ref, Operand<T> valu
    * @param value the value by which the variable will be incremented.
    * @return a new instance of AssignAddVariableOp
    */
-  public <T extends Tensor> AssignAddVariableOp assignAddVariableOp(Operand<?> resource,
+  public <T extends TType> AssignAddVariableOp assignAddVariableOp(Operand<?> resource,
       Operand<T> value) {
     return AssignAddVariableOp.create(scope, resource, value);
   }
@@ -586,7 +585,7 @@ public <T extends Tensor> AssignAddVariableOp assignAddVariableOp(Operand<?> res
    * @param options carries optional attributes values
    * @return a new instance of AssignSub
    */
-  public <T extends Tensor> AssignSub<T> assignSub(Operand<T> ref, Operand<T> value,
+  public <T extends TType> AssignSub<T> assignSub(Operand<T> ref, Operand<T> value,
       AssignSub.Options... options) {
     return AssignSub.create(scope, ref, value, options);
   }
@@ -601,7 +600,7 @@ public <T extends Tensor> AssignSub<T> assignSub(Operand<T> ref, Operand<T> valu
    * @param value the value by which the variable will be incremented.
    * @return a new instance of AssignSubVariableOp
    */
-  public <T extends Tensor> AssignSubVariableOp assignSubVariableOp(Operand<?> resource,
+  public <T extends TType> AssignSubVariableOp assignSubVariableOp(Operand<?> resource,
       Operand<T> value) {
     return AssignSubVariableOp.create(scope, resource, value);
   }
@@ -616,7 +615,7 @@ public <T extends Tensor> AssignSubVariableOp assignSubVariableOp(Operand<?> res
    * @param value the value to set the new tensor to use.
    * @return a new instance of AssignVariableOp
    */
-  public <T extends Tensor> AssignVariableOp assignVariableOp(Operand<?> resource,
+  public <T extends TType> AssignVariableOp assignVariableOp(Operand<?> resource,
       Operand<T> value) {
     return AssignVariableOp.create(scope, resource, value);
   }
@@ -684,7 +683,7 @@ public BarrierIncompleteSize barrierIncompleteSize(Operand<TString> handle) {
    * @param componentIndex The component of the barrier elements that is being assigned.
    * @return a new instance of BarrierInsertMany
    */
-  public <T extends Tensor> BarrierInsertMany barrierInsertMany(Operand<TString> handle,
+  public <T extends TType> BarrierInsertMany barrierInsertMany(Operand<TString> handle,
       Operand<TString> keys, Operand<T> values, Long componentIndex) {
     return BarrierInsertMany.create(scope, handle, keys, values, componentIndex);
   }
@@ -799,8 +798,8 @@ public Batch batch(Iterable<Operand<?>> inTensors, Long numBatchThreads, Long ma
    * @param blockSize
    * @return a new instance of BatchToSpace
    */
-  public <T extends Tensor, U extends Tensor & TNumber> BatchToSpace<T> batchToSpace(
-      Operand<T> input, Operand<U> crops, Long blockSize) {
+  public <T extends TType, U extends TNumber> BatchToSpace<T> batchToSpace(Operand<T> input,
+      Operand<U> crops, Long blockSize) {
     return BatchToSpace.create(scope, input, crops, blockSize);
   }
 
@@ -914,7 +913,7 @@ public <T extends Tensor, U extends Tensor & TNumber> BatchToSpace<T> batchToSpa
    *  }</pre>
    * @return a new instance of BatchToSpaceNd
    */
-  public <T extends Tensor, U extends Tensor & TNumber, V extends Tensor & TNumber> BatchToSpaceNd<T> batchToSpaceNd(
+  public <T extends TType, U extends TNumber, V extends TNumber> BatchToSpaceNd<T> batchToSpaceNd(
       Operand<T> input, Operand<U> blockShape, Operand<V> crops) {
     return BatchToSpaceNd.create(scope, input, blockShape, crops);
   }
@@ -978,8 +977,7 @@ public <T extends Tensor, U extends Tensor & TNumber, V extends Tensor & TNumber
    * @param type
    * @return a new instance of Bitcast
    */
-  public <U extends Tensor, T extends Tensor> Bitcast<U> bitcast(Operand<T> input,
-      DataType<U> type) {
+  public <U extends TType, T extends TType> Bitcast<U> bitcast(Operand<T> input, DataType<U> type) {
     return Bitcast.create(scope, input, type);
   }
 
@@ -994,7 +992,7 @@ public <U extends Tensor, T extends Tensor> Bitcast<U> bitcast(Operand<T> input,
    * @param s1
    * @return a new instance of BroadcastDynamicShape
    */
-  public <T extends Tensor & TNumber> BroadcastDynamicShape<T> broadcastDynamicShape(Operand<T> s0,
+  public <T extends TNumber> BroadcastDynamicShape<T> broadcastDynamicShape(Operand<T> s0,
       Operand<T> s1) {
     return BroadcastDynamicShape.create(scope, s0, s1);
   }
@@ -1026,7 +1024,7 @@ public <T extends Tensor & TNumber> BroadcastDynamicShape<T> broadcastDynamicSha
    * @param shape An 1-D `int` Tensor. The shape of the desired output.
    * @return a new instance of BroadcastTo
    */
-  public <T extends Tensor, U extends Tensor & TNumber> BroadcastTo<T> broadcastTo(Operand<T> input,
+  public <T extends TType, U extends TNumber> BroadcastTo<T> broadcastTo(Operand<T> input,
       Operand<U> shape) {
     return BroadcastTo.create(scope, input, shape);
   }
@@ -1049,11 +1047,21 @@ public <T extends Tensor, U extends Tensor & TNumber> BroadcastTo<T> broadcastTo
    * @param boundaries A sorted list of floats gives the boundary of the buckets.
    * @return a new instance of Bucketize
    */
-  public <T extends Tensor & TNumber> Bucketize bucketize(Operand<T> input,
-      List<Float> boundaries) {
+  public <T extends TNumber> Bucketize bucketize(Operand<T> input, List<Float> boundaries) {
     return Bucketize.create(scope, input, boundaries);
   }
 
+  /**
+   * Create a constant from a Tensor.
+   *
+   * @param scope is a scope used to add the underlying operation.
+   * @param tensor a Tensor holding the constant value
+   * @return a constant of the same data type as `tensor`
+   */
+  public <T extends TType> Constant<T> capture(T tensor) {
+    return Constant.create(scope, tensor);
+  }
+
   /**
    * Clips tensor values to a specified min and max.
    *  <p>
@@ -1070,7 +1078,7 @@ public <T extends Tensor & TNumber> Bucketize bucketize(Operand<T> input,
    *  as `t`. The maximum value to clip by.
    * @return a new instance of ClipByValue
    */
-  public <T extends Tensor> ClipByValue<T> clipByValue(Operand<T> t, Operand<T> clipValueMin,
+  public <T extends TType> ClipByValue<T> clipByValue(Operand<T> t, Operand<T> clipValueMin,
       Operand<T> clipValueMax) {
     return ClipByValue.create(scope, t, clipValueMin, clipValueMax);
   }
@@ -1085,8 +1093,8 @@ public <T extends Tensor> ClipByValue<T> clipByValue(Operand<T> t, Operand<T> cl
    *  range [-rank(values), rank(values)).
    * @return a new instance of Concat
    */
-  public <T extends Tensor, U extends Tensor & TNumber> Concat<T> concat(
-      Iterable<Operand<T>> values, Operand<U> axis) {
+  public <T extends TType, U extends TNumber> Concat<T> concat(Iterable<Operand<T>> values,
+      Operand<U> axis) {
     return Concat.create(scope, values, axis);
   }
 
@@ -1689,17 +1697,6 @@ public Constant<TInt64> constant(Shape shape) {
     return Constant.tensorOf(scope, shape);
   }
 
-  /**
-   * Create a constant from a Tensor.
-   *
-   * @param scope is a scope used to add the underlying operation.
-   * @param tensor a Tensor holding the constant value
-   * @return a constant of the same data type as `tensor`
-   */
-  public <T extends Tensor & TType> Constant<T> constant(T tensor) {
-    return Constant.create(scope, tensor);
-  }
-
   /**
    * Creates a constant of {@code String} elements, using the given charset.
    *
@@ -1855,7 +1852,7 @@ public Constant<TString> constant(Charset charset, Shape shape, DataBuffer<Strin
    * @throws IllegalArgumentException If the tensor datatype or shape is not compatible with the
    *      buffer
    */
-  public <T extends Tensor & TType> Constant<T> constant(DataType<T> type, Shape shape,
+  public <T extends TType> Constant<T> constant(DataType<T> type, Shape shape,
       ByteDataBuffer data) {
     return Constant.tensorOf(scope, type, shape, data);
   }
@@ -1898,7 +1895,7 @@ public ControlTrigger controlTrigger() {
    *  'OutOfRange' error.
    * @return a new instance of CountUpTo
    */
-  public <T extends Tensor & TNumber> CountUpTo<T> countUpTo(Operand<T> ref, Long limit) {
+  public <T extends TNumber> CountUpTo<T> countUpTo(Operand<T> ref, Long limit) {
     return CountUpTo.create(scope, ref, limit);
   }
 
@@ -1909,7 +1906,7 @@ public <T extends Tensor & TNumber> CountUpTo<T> countUpTo(Operand<T> ref, Long
    * @param x The source tensor of type `T`.
    * @return a new instance of DeepCopy
    */
-  public <T extends Tensor> DeepCopy<T> deepCopy(Operand<T> x) {
+  public <T extends TType> DeepCopy<T> deepCopy(Operand<T> x) {
     return DeepCopy.create(scope, x);
   }
 
@@ -1955,7 +1952,7 @@ public DestroyResourceOp destroyResourceOp(Operand<?> resource,
    *  'TemporaryVariable' op.
    * @return a new instance of DestroyTemporaryVariable
    */
-  public <T extends Tensor> DestroyTemporaryVariable<T> destroyTemporaryVariable(Operand<T> ref,
+  public <T extends TType> DestroyTemporaryVariable<T> destroyTemporaryVariable(Operand<T> ref,
       String varName) {
     return DestroyTemporaryVariable.create(scope, ref, varName);
   }
@@ -2003,7 +2000,7 @@ public <T extends Tensor> DestroyTemporaryVariable<T> destroyTemporaryVariable(O
    * @param numPartitions The number of partitions to output.
    * @return a new instance of DynamicPartition
    */
-  public <T extends Tensor> DynamicPartition<T> dynamicPartition(Operand<T> data,
+  public <T extends TType> DynamicPartition<T> dynamicPartition(Operand<T> data,
       Operand<TInt32> partitions, Long numPartitions) {
     return DynamicPartition.create(scope, data, partitions, numPartitions);
   }
@@ -2071,7 +2068,7 @@ public <T extends Tensor> DynamicPartition<T> dynamicPartition(Operand<T> data,
    * @param data
    * @return a new instance of DynamicStitch
    */
-  public <T extends Tensor> DynamicStitch<T> dynamicStitch(Iterable<Operand<TInt32>> indices,
+  public <T extends TType> DynamicStitch<T> dynamicStitch(Iterable<Operand<TInt32>> indices,
       Iterable<Operand<T>> data) {
     return DynamicStitch.create(scope, indices, data);
   }
@@ -2100,7 +2097,7 @@ public <T extends Tensor> DynamicStitch<T> dynamicStitch(Iterable<Operand<TInt32
    * @param options carries optional attributes values
    * @return a new instance of EditDistance
    */
-  public <T extends Tensor> EditDistance editDistance(Operand<TInt64> hypothesisIndices,
+  public <T extends TType> EditDistance editDistance(Operand<TInt64> hypothesisIndices,
       Operand<T> hypothesisValues, Operand<TInt64> hypothesisShape, Operand<TInt64> truthIndices,
       Operand<T> truthValues, Operand<TInt64> truthShape, EditDistance.Options... options) {
     return EditDistance.create(scope, hypothesisIndices, hypothesisValues, hypothesisShape, truthIndices, truthValues, truthShape, options);
@@ -2117,7 +2114,7 @@ public <T extends Tensor> EditDistance editDistance(Operand<TInt64> hypothesisIn
    * @param options carries optional attributes values
    * @return a new instance of Empty
    */
-  public <T extends Tensor> Empty<T> empty(Operand<TInt32> shape, DataType<T> dtype,
+  public <T extends TType> Empty<T> empty(Operand<TInt32> shape, DataType<T> dtype,
       Empty.Options... options) {
     return Empty.create(scope, shape, dtype, options);
   }
@@ -2137,7 +2134,7 @@ public <T extends Tensor> Empty<T> empty(Operand<TInt32> shape, DataType<T> dtyp
    * @param elementDtype
    * @return a new instance of EmptyTensorList
    */
-  public <T extends Tensor & TNumber, U extends Tensor> EmptyTensorList emptyTensorList(
+  public <T extends TNumber, U extends TType> EmptyTensorList emptyTensorList(
       Operand<T> elementShape, Operand<TInt32> maxNumElements, DataType<U> elementDtype) {
     return EmptyTensorList.create(scope, elementShape, maxNumElements, elementDtype);
   }
@@ -2153,7 +2150,7 @@ public <T extends Tensor & TNumber, U extends Tensor> EmptyTensorList emptyTenso
    * @param shape The expected (possibly partially specified) shape of the input tensor.
    * @return a new instance of EnsureShape
    */
-  public <T extends Tensor> EnsureShape<T> ensureShape(Operand<T> input, Shape shape) {
+  public <T extends TType> EnsureShape<T> ensureShape(Operand<T> input, Shape shape) {
     return EnsureShape.create(scope, input, shape);
   }
 
@@ -2196,7 +2193,7 @@ public <T extends Tensor> EnsureShape<T> ensureShape(Operand<T> input, Shape sha
    *  `[-rank(input) - 1, rank(input)]`.
    * @return a new instance of ExpandDims
    */
-  public <T extends Tensor, U extends Tensor & TNumber> ExpandDims<T> expandDims(Operand<T> input,
+  public <T extends TType, U extends TNumber> ExpandDims<T> expandDims(Operand<T> input,
       Operand<U> axis) {
     return ExpandDims.create(scope, input, axis);
   }
@@ -2218,7 +2215,7 @@ public <T extends Tensor, U extends Tensor & TNumber> ExpandDims<T> expandDims(O
    *  }</pre>
    * @return a new instance of ExtractVolumePatches
    */
-  public <T extends Tensor & TNumber> ExtractVolumePatches<T> extractVolumePatches(Operand<T> input,
+  public <T extends TNumber> ExtractVolumePatches<T> extractVolumePatches(Operand<T> input,
       List<Long> ksizes, List<Long> strides, String padding) {
     return ExtractVolumePatches.create(scope, input, ksizes, strides, padding);
   }
@@ -2257,8 +2254,7 @@ public <T extends Tensor & TNumber> ExtractVolumePatches<T> extractVolumePatches
    * @end_compatibility
    * @return a new instance of Fill
    */
-  public <U extends Tensor, T extends Tensor & TNumber> Fill<U> fill(Operand<T> dims,
-      Operand<U> value) {
+  public <U extends TType, T extends TNumber> Fill<U> fill(Operand<T> dims, Operand<U> value) {
     return Fill.create(scope, dims, value);
   }
 
@@ -2299,7 +2295,7 @@ public <U extends Tensor, T extends Tensor & TNumber> Fill<U> fill(Operand<T> di
    *  `farmhash::fingerprint64`.
    * @return a new instance of Fingerprint
    */
-  public <T extends Tensor> Fingerprint fingerprint(Operand<T> data, Operand<TString> method) {
+  public <T extends TType> Fingerprint fingerprint(Operand<T> data, Operand<TString> method) {
     return Fingerprint.create(scope, data, method);
   }
 
@@ -2341,8 +2337,8 @@ public <T extends Tensor> Fingerprint fingerprint(Operand<T> data, Operand<TStri
    * @param options carries optional attributes values
    * @return a new instance of Gather
    */
-  public <T extends Tensor, U extends Tensor & TNumber, V extends Tensor & TNumber> Gather<T> gather(
-      Operand<T> params, Operand<U> indices, Operand<V> axis, Gather.Options... options) {
+  public <T extends TType, U extends TNumber, V extends TNumber> Gather<T> gather(Operand<T> params,
+      Operand<U> indices, Operand<V> axis, Gather.Options... options) {
     return Gather.create(scope, params, indices, axis, options);
   }
 
@@ -2447,7 +2443,7 @@ public <T extends Tensor, U extends Tensor & TNumber, V extends Tensor & TNumber
    * @param indices Index tensor.
    * @return a new instance of GatherNd
    */
-  public <T extends Tensor, U extends Tensor & TNumber> GatherNd<T> gatherNd(Operand<T> params,
+  public <T extends TType, U extends TNumber> GatherNd<T> gatherNd(Operand<T> params,
       Operand<U> indices) {
     return GatherNd.create(scope, params, indices);
   }
@@ -2458,7 +2454,7 @@ public <T extends Tensor, U extends Tensor & TNumber> GatherNd<T> gatherNd(Opera
    * @param value The tensor to be stored.
    * @return a new instance of GetSessionHandle
    */
-  public <T extends Tensor> GetSessionHandle getSessionHandle(Operand<T> value) {
+  public <T extends TType> GetSessionHandle getSessionHandle(Operand<T> value) {
     return GetSessionHandle.create(scope, value);
   }
 
@@ -2470,7 +2466,7 @@ public <T extends Tensor> GetSessionHandle getSessionHandle(Operand<T> value) {
    * @param dtype The type of the output value.
    * @return a new instance of GetSessionTensor
    */
-  public <T extends Tensor> GetSessionTensor<T> getSessionTensor(Operand<TString> handle,
+  public <T extends TType> GetSessionTensor<T> getSessionTensor(Operand<TString> handle,
       DataType<T> dtype) {
     return GetSessionTensor.create(scope, handle, dtype);
   }
@@ -2535,7 +2531,7 @@ public Gradients gradients(Operand<?> y, Iterable<? extends Operand<?>> x,
    * @param input
    * @return a new instance of GuaranteeConst
    */
-  public <T extends Tensor> GuaranteeConst<T> guaranteeConst(Operand<T> input) {
+  public <T extends TType> GuaranteeConst<T> guaranteeConst(Operand<T> input) {
     return GuaranteeConst.create(scope, input);
   }
 
@@ -2551,7 +2547,7 @@ public <T extends Tensor> GuaranteeConst<T> guaranteeConst(Operand<T> input) {
    * @param options carries optional attributes values
    * @return a new instance of HashTable
    */
-  public <T extends Tensor, U extends Tensor> HashTable hashTable(DataType<T> keyDtype,
+  public <T extends TType, U extends TType> HashTable hashTable(DataType<T> keyDtype,
       DataType<U> valueDtype, HashTable.Options... options) {
     return HashTable.create(scope, keyDtype, valueDtype, options);
   }
@@ -2582,8 +2578,8 @@ public <T extends Tensor, U extends Tensor> HashTable hashTable(DataType<T> keyD
    * @param nbins Scalar `int32 Tensor`.  Number of histogram bins.
    * @return a new instance of HistogramFixedWidth
    */
-  public <T extends Tensor & TNumber> HistogramFixedWidth<TInt32> histogramFixedWidth(
-      Operand<T> values, Operand<T> valueRange, Operand<TInt32> nbins) {
+  public <T extends TNumber> HistogramFixedWidth<TInt32> histogramFixedWidth(Operand<T> values,
+      Operand<T> valueRange, Operand<TInt32> nbins) {
     return HistogramFixedWidth.create(scope, values, valueRange, nbins);
   }
 
@@ -2614,7 +2610,7 @@ public <T extends Tensor & TNumber> HistogramFixedWidth<TInt32> histogramFixedWi
    * @param dtype
    * @return a new instance of HistogramFixedWidth
    */
-  public <U extends Tensor & TNumber, T extends Tensor & TNumber> HistogramFixedWidth<U> histogramFixedWidth(
+  public <U extends TNumber, T extends TNumber> HistogramFixedWidth<U> histogramFixedWidth(
       Operand<T> values, Operand<T> valueRange, Operand<TInt32> nbins, DataType<U> dtype) {
     return HistogramFixedWidth.create(scope, values, valueRange, nbins, dtype);
   }
@@ -2626,7 +2622,7 @@ public <U extends Tensor & TNumber, T extends Tensor & TNumber> HistogramFixedWi
    * @param input
    * @return a new instance of Identity
    */
-  public <T extends Tensor> Identity<T> identity(Operand<T> input) {
+  public <T extends TType> Identity<T> identity(Operand<T> input) {
     return Identity.create(scope, input);
   }
 
@@ -2665,7 +2661,7 @@ public IdentityN identityN(Iterable<Operand<?>> input) {
    *  NewReadOnlyMemoryRegionFromFile in tensorflow::Env.
    * @return a new instance of ImmutableConst
    */
-  public <T extends Tensor> ImmutableConst<T> immutableConst(DataType<T> dtype, Shape shape,
+  public <T extends TType> ImmutableConst<T> immutableConst(DataType<T> dtype, Shape shape,
       String memoryRegionName) {
     return ImmutableConst.create(scope, dtype, shape, memoryRegionName);
   }
@@ -2755,8 +2751,8 @@ public void initAdd(Op initializer) {
    * @param values Values of type Tval.
    * @return a new instance of InitializeTable
    */
-  public <T extends Tensor, U extends Tensor> InitializeTable initializeTable(
-      Operand<?> tableHandle, Operand<T> keys, Operand<U> values) {
+  public <T extends TType, U extends TType> InitializeTable initializeTable(Operand<?> tableHandle,
+      Operand<T> keys, Operand<U> values) {
     return InitializeTable.create(scope, tableHandle, keys, values);
   }
 
@@ -2799,8 +2795,7 @@ public InitializeTableFromTextFile initializeTableFromTextFile(Operand<?> tableH
    * @param v A `Tensor` of type T. Same dimension sizes as x except the first dimension, which must be the same as i's size.
    * @return a new instance of InplaceAdd
    */
-  public <T extends Tensor> InplaceAdd<T> inplaceAdd(Operand<T> x, Operand<TInt32> i,
-      Operand<T> v) {
+  public <T extends TType> InplaceAdd<T> inplaceAdd(Operand<T> x, Operand<TInt32> i, Operand<T> v) {
     return InplaceAdd.create(scope, x, i, v);
   }
 
@@ -2815,8 +2810,7 @@ public <T extends Tensor> InplaceAdd<T> inplaceAdd(Operand<T> x, Operand<TInt32>
    * @param v A `Tensor` of type T. Same dimension sizes as x except the first dimension, which must be the same as i's size.
    * @return a new instance of InplaceSub
    */
-  public <T extends Tensor> InplaceSub<T> inplaceSub(Operand<T> x, Operand<TInt32> i,
-      Operand<T> v) {
+  public <T extends TType> InplaceSub<T> inplaceSub(Operand<T> x, Operand<TInt32> i, Operand<T> v) {
     return InplaceSub.create(scope, x, i, v);
   }
 
@@ -2831,7 +2825,7 @@ public <T extends Tensor> InplaceSub<T> inplaceSub(Operand<T> x, Operand<TInt32>
    * @param v A `Tensor` of type T. Same dimension sizes as x except the first dimension, which must be the same as i's size.
    * @return a new instance of InplaceUpdate
    */
-  public <T extends Tensor> InplaceUpdate<T> inplaceUpdate(Operand<T> x, Operand<TInt32> i,
+  public <T extends TType> InplaceUpdate<T> inplaceUpdate(Operand<T> x, Operand<TInt32> i,
       Operand<T> v) {
     return InplaceUpdate.create(scope, x, i, v);
   }
@@ -2844,7 +2838,7 @@ public <T extends Tensor> InplaceUpdate<T> inplaceUpdate(Operand<T> x, Operand<T
    * @param ref Should be from a `Variable` node. May be uninitialized.
    * @return a new instance of IsVariableInitialized
    */
-  public <T extends Tensor> IsVariableInitialized isVariableInitialized(Operand<T> ref) {
+  public <T extends TType> IsVariableInitialized isVariableInitialized(Operand<T> ref) {
     return IsVariableInitialized.create(scope, ref);
   }
 
@@ -2866,8 +2860,8 @@ public <T extends Tensor> IsVariableInitialized isVariableInitialized(Operand<T>
    * @param num 0-D tensor. Number of values to generate.
    * @return a new instance of LinSpace
    */
-  public <T extends Tensor & TNumber, U extends Tensor & TNumber> LinSpace<T> linSpace(
-      Operand<T> start, Operand<T> stop, Operand<U> num) {
+  public <T extends TNumber, U extends TNumber> LinSpace<T> linSpace(Operand<T> start,
+      Operand<T> stop, Operand<U> num) {
     return LinSpace.create(scope, start, stop, num);
   }
 
@@ -2881,7 +2875,7 @@ public <T extends Tensor & TNumber, U extends Tensor & TNumber> LinSpace<T> linS
    * @param Tvalues
    * @return a new instance of LookupTableExport
    */
-  public <T extends Tensor, U extends Tensor> LookupTableExport<T, U> lookupTableExport(
+  public <T extends TType, U extends TType> LookupTableExport<T, U> lookupTableExport(
       Operand<?> tableHandle, DataType<T> Tkeys, DataType<U> Tvalues) {
     return LookupTableExport.create(scope, tableHandle, Tkeys, Tvalues);
   }
@@ -2901,7 +2895,7 @@ public <T extends Tensor, U extends Tensor> LookupTableExport<T, U> lookupTableE
    * @param defaultValue
    * @return a new instance of LookupTableFind
    */
-  public <U extends Tensor, T extends Tensor> LookupTableFind<U> lookupTableFind(
+  public <U extends TType, T extends TType> LookupTableFind<U> lookupTableFind(
       Operand<?> tableHandle, Operand<T> keys, Operand<U> defaultValue) {
     return LookupTableFind.create(scope, tableHandle, keys, defaultValue);
   }
@@ -2917,7 +2911,7 @@ public <U extends Tensor, T extends Tensor> LookupTableFind<U> lookupTableFind(
    * @param values Values to associate with keys.
    * @return a new instance of LookupTableImport
    */
-  public <T extends Tensor, U extends Tensor> LookupTableImport lookupTableImport(
+  public <T extends TType, U extends TType> LookupTableImport lookupTableImport(
       Operand<?> tableHandle, Operand<T> keys, Operand<U> values) {
     return LookupTableImport.create(scope, tableHandle, keys, values);
   }
@@ -2933,7 +2927,7 @@ public <T extends Tensor, U extends Tensor> LookupTableImport lookupTableImport(
    * @param values Values to associate with keys.
    * @return a new instance of LookupTableInsert
    */
-  public <T extends Tensor, U extends Tensor> LookupTableInsert lookupTableInsert(
+  public <T extends TType, U extends TType> LookupTableInsert lookupTableInsert(
       Operand<?> tableHandle, Operand<T> keys, Operand<U> values) {
     return LookupTableInsert.create(scope, tableHandle, keys, values);
   }
@@ -3076,8 +3070,8 @@ public MapUnstageNoKey mapUnstageNoKey(Operand<TInt32> indices, List<DataType<?>
    * @param options carries optional attributes values
    * @return a new instance of Max
    */
-  public <T extends Tensor, U extends Tensor & TNumber> Max<T> max(Operand<T> input,
-      Operand<U> axis, Max.Options... options) {
+  public <T extends TType, U extends TNumber> Max<T> max(Operand<T> input, Operand<U> axis,
+      Max.Options... options) {
     return Max.create(scope, input, axis, options);
   }
 
@@ -3094,7 +3088,7 @@ public <T extends Tensor, U extends Tensor & TNumber> Max<T> max(Operand<T> inpu
    * @param inputs The input tensors, exactly one of which will become available.
    * @return a new instance of Merge
    */
-  public <T extends Tensor> Merge<T> merge(Iterable<Operand<T>> inputs) {
+  public <T extends TType> Merge<T> merge(Iterable<Operand<T>> inputs) {
     return Merge.create(scope, inputs);
   }
 
@@ -3113,8 +3107,8 @@ public <T extends Tensor> Merge<T> merge(Iterable<Operand<T>> inputs) {
    * @param options carries optional attributes values
    * @return a new instance of Min
    */
-  public <T extends Tensor, U extends Tensor & TNumber> Min<T> min(Operand<T> input,
-      Operand<U> axis, Min.Options... options) {
+  public <T extends TType, U extends TNumber> Min<T> min(Operand<T> input, Operand<U> axis,
+      Min.Options... options) {
     return Min.create(scope, input, axis, options);
   }
 
@@ -3157,7 +3151,7 @@ public <T extends Tensor, U extends Tensor & TNumber> Min<T> min(Operand<T> inpu
    *  it is `[1, 2, 3, 3, 2]` in symmetric mode.
    * @return a new instance of MirrorPad
    */
-  public <T extends Tensor, U extends Tensor & TNumber> MirrorPad<T> mirrorPad(Operand<T> input,
+  public <T extends TType, U extends TNumber> MirrorPad<T> mirrorPad(Operand<T> input,
       Operand<U> paddings, String mode) {
     return MirrorPad.create(scope, input, paddings, mode);
   }
@@ -3219,7 +3213,7 @@ public MlirPassthroughOp mlirPassthroughOp(Iterable<Operand<?>> inputs, String m
    * @param options carries optional attributes values
    * @return a new instance of MutableDenseHashTable
    */
-  public <T extends Tensor, U extends Tensor> MutableDenseHashTable mutableDenseHashTable(
+  public <T extends TType, U extends TType> MutableDenseHashTable mutableDenseHashTable(
       Operand<T> emptyKey, Operand<T> deletedKey, DataType<U> valueDtype,
       MutableDenseHashTable.Options... options) {
     return MutableDenseHashTable.create(scope, emptyKey, deletedKey, valueDtype, options);
@@ -3237,8 +3231,8 @@ public <T extends Tensor, U extends Tensor> MutableDenseHashTable mutableDenseHa
    * @param options carries optional attributes values
    * @return a new instance of MutableHashTable
    */
-  public <T extends Tensor, U extends Tensor> MutableHashTable mutableHashTable(
-      DataType<T> keyDtype, DataType<U> valueDtype, MutableHashTable.Options... options) {
+  public <T extends TType, U extends TType> MutableHashTable mutableHashTable(DataType<T> keyDtype,
+      DataType<U> valueDtype, MutableHashTable.Options... options) {
     return MutableHashTable.create(scope, keyDtype, valueDtype, options);
   }
 
@@ -3254,7 +3248,7 @@ public <T extends Tensor, U extends Tensor> MutableHashTable mutableHashTable(
    * @param options carries optional attributes values
    * @return a new instance of MutableHashTableOfTensors
    */
-  public <T extends Tensor, U extends Tensor> MutableHashTableOfTensors mutableHashTableOfTensors(
+  public <T extends TType, U extends TType> MutableHashTableOfTensors mutableHashTableOfTensors(
       DataType<T> keyDtype, DataType<U> valueDtype, MutableHashTableOfTensors.Options... options) {
     return MutableHashTableOfTensors.create(scope, keyDtype, valueDtype, options);
   }
@@ -3322,7 +3316,7 @@ public MutexLock mutexLock(Operand<?> mutex) {
    * @param data The tensor to be made available to the next iteration.
    * @return a new instance of NextIteration
    */
-  public <T extends Tensor> NextIteration<T> nextIteration(Operand<T> data) {
+  public <T extends TType> NextIteration<T> nextIteration(Operand<T> data) {
     return NextIteration.create(scope, data);
   }
 
@@ -3426,7 +3420,7 @@ public NoOp noOp() {
    * @param options carries optional attributes values
    * @return a new instance of OneHot
    */
-  public <U extends Tensor, T extends Tensor & TNumber> OneHot<U> oneHot(Operand<T> indices,
+  public <U extends TType, T extends TNumber> OneHot<U> oneHot(Operand<T> indices,
       Operand<TInt32> depth, Operand<U> onValue, Operand<U> offValue, OneHot.Options... options) {
     return OneHot.create(scope, indices, depth, onValue, offValue, options);
   }
@@ -3438,7 +3432,7 @@ public <U extends Tensor, T extends Tensor & TNumber> OneHot<U> oneHot(Operand<T
    * @param x a tensor of type T.
    * @return a new instance of OnesLike
    */
-  public <T extends Tensor> OnesLike<T> onesLike(Operand<T> x) {
+  public <T extends TType> OnesLike<T> onesLike(Operand<T> x) {
     return OnesLike.create(scope, x);
   }
 
@@ -3580,8 +3574,8 @@ public OrderedMapUnstageNoKey orderedMapUnstageNoKey(Operand<TInt32> indices,
    * @param constantValues
    * @return a new instance of Pad
    */
-  public <T extends Tensor, U extends Tensor & TNumber> Pad<T> pad(Operand<T> input,
-      Operand<U> paddings, Operand<T> constantValues) {
+  public <T extends TType, U extends TNumber> Pad<T> pad(Operand<T> input, Operand<U> paddings,
+      Operand<T> constantValues) {
     return Pad.create(scope, input, paddings, constantValues);
   }
 
@@ -3610,7 +3604,7 @@ public <T extends Tensor, U extends Tensor & TNumber> Pad<T> pad(Operand<T> inpu
    *  but with the number of input values in the first dimension.
    * @return a new instance of ParallelConcat
    */
-  public <T extends Tensor> ParallelConcat<T> parallelConcat(Iterable<Operand<T>> values,
+  public <T extends TType> ParallelConcat<T> parallelConcat(Iterable<Operand<T>> values,
       Shape shape) {
     return ParallelConcat.create(scope, values, shape);
   }
@@ -3677,7 +3671,7 @@ public <T extends Tensor> ParallelConcat<T> parallelConcat(Iterable<Operand<T>>
    * @param data
    * @return a new instance of ParallelDynamicStitch
    */
-  public <T extends Tensor> ParallelDynamicStitch<T> parallelDynamicStitch(
+  public <T extends TType> ParallelDynamicStitch<T> parallelDynamicStitch(
       Iterable<Operand<TInt32>> indices, Iterable<Operand<T>> data) {
     return ParallelDynamicStitch.create(scope, indices, data);
   }
@@ -3694,7 +3688,7 @@ public <T extends Tensor> ParallelDynamicStitch<T> parallelDynamicStitch(
    * @param options carries optional attributes values
    * @return a new instance of Placeholder
    */
-  public <T extends Tensor> Placeholder<T> placeholder(DataType<T> dtype,
+  public <T extends TType> Placeholder<T> placeholder(DataType<T> dtype,
       Placeholder.Options... options) {
     return Placeholder.create(scope, dtype, options);
   }
@@ -3707,7 +3701,7 @@ public <T extends Tensor> Placeholder<T> placeholder(DataType<T> dtype,
    * @param shape The (possibly partial) shape of the tensor.
    * @return a new instance of PlaceholderWithDefault
    */
-  public <T extends Tensor> PlaceholderWithDefault<T> placeholderWithDefault(Operand<T> input,
+  public <T extends TType> PlaceholderWithDefault<T> placeholderWithDefault(Operand<T> input,
       Shape shape) {
     return PlaceholderWithDefault.create(scope, input, shape);
   }
@@ -3740,8 +3734,8 @@ public Print print(Operand<TString> input, Print.Options... options) {
    * @param options carries optional attributes values
    * @return a new instance of Prod
    */
-  public <T extends Tensor, U extends Tensor & TNumber> Prod<T> prod(Operand<T> input,
-      Operand<U> axis, Prod.Options... options) {
+  public <T extends TType, U extends TNumber> Prod<T> prod(Operand<T> input, Operand<U> axis,
+      Prod.Options... options) {
     return Prod.create(scope, input, axis, options);
   }
 
@@ -3757,7 +3751,7 @@ public <T extends Tensor, U extends Tensor & TNumber> Prod<T> prod(Operand<T> in
    * @param inputMax The maximum value of the input.
    * @return a new instance of QuantizedReshape
    */
-  public <T extends Tensor, U extends Tensor & TNumber> QuantizedReshape<T> quantizedReshape(
+  public <T extends TType, U extends TNumber> QuantizedReshape<T> quantizedReshape(
       Operand<T> tensor, Operand<U> shape, Operand<TFloat32> inputMin, Operand<TFloat32> inputMax) {
     return QuantizedReshape.create(scope, tensor, shape, inputMin, inputMax);
   }
@@ -3782,8 +3776,7 @@ public <T extends Tensor, U extends Tensor & TNumber> QuantizedReshape<T> quanti
    * @param delta 0-D (scalar). Optional. Default is 1. Number that increments `start`.
    * @return a new instance of Range
    */
-  public <T extends Tensor & TNumber> Range<T> range(Operand<T> start, Operand<T> limit,
-      Operand<T> delta) {
+  public <T extends TNumber> Range<T> range(Operand<T> start, Operand<T> limit, Operand<T> delta) {
     return Range.create(scope, start, limit, delta);
   }
 
@@ -3805,7 +3798,7 @@ public <T extends Tensor & TNumber> Range<T> range(Operand<T> start, Operand<T>
    * @param input
    * @return a new instance of Rank
    */
-  public <T extends Tensor> Rank rank(Operand<T> input) {
+  public <T extends TType> Rank rank(Operand<T> input) {
     return Rank.create(scope, input);
   }
 
@@ -3824,7 +3817,7 @@ public <T extends Tensor> Rank rank(Operand<T> input) {
    * @param dtype the dtype of the value.
    * @return a new instance of ReadVariableOp
    */
-  public <T extends Tensor> ReadVariableOp<T> readVariableOp(Operand<?> resource,
+  public <T extends TType> ReadVariableOp<T> readVariableOp(Operand<?> resource,
       DataType<T> dtype) {
     return ReadVariableOp.create(scope, resource, dtype);
   }
@@ -3843,7 +3836,7 @@ public <T extends Tensor> ReadVariableOp<T> readVariableOp(Operand<?> resource,
    * @param options carries optional attributes values
    * @return a new instance of ReduceAll
    */
-  public <T extends Tensor & TNumber> ReduceAll reduceAll(Operand<TBool> input, Operand<T> axis,
+  public <T extends TNumber> ReduceAll reduceAll(Operand<TBool> input, Operand<T> axis,
       ReduceAll.Options... options) {
     return ReduceAll.create(scope, input, axis, options);
   }
@@ -3862,7 +3855,7 @@ public <T extends Tensor & TNumber> ReduceAll reduceAll(Operand<TBool> input, Op
    * @param options carries optional attributes values
    * @return a new instance of ReduceAny
    */
-  public <T extends Tensor & TNumber> ReduceAny reduceAny(Operand<TBool> input, Operand<T> axis,
+  public <T extends TNumber> ReduceAny reduceAny(Operand<TBool> input, Operand<T> axis,
       ReduceAny.Options... options) {
     return ReduceAny.create(scope, input, axis, options);
   }
@@ -3882,7 +3875,7 @@ public <T extends Tensor & TNumber> ReduceAny reduceAny(Operand<TBool> input, Op
    * @param options carries optional attributes values
    * @return a new instance of ReduceMax
    */
-  public <T extends Tensor, U extends Tensor & TNumber> ReduceMax<T> reduceMax(Operand<T> input,
+  public <T extends TType, U extends TNumber> ReduceMax<T> reduceMax(Operand<T> input,
       Operand<U> axis, ReduceMax.Options... options) {
     return ReduceMax.create(scope, input, axis, options);
   }
@@ -3902,7 +3895,7 @@ public <T extends Tensor, U extends Tensor & TNumber> ReduceMax<T> reduceMax(Ope
    * @param options carries optional attributes values
    * @return a new instance of ReduceMin
    */
-  public <T extends Tensor, U extends Tensor & TNumber> ReduceMin<T> reduceMin(Operand<T> input,
+  public <T extends TType, U extends TNumber> ReduceMin<T> reduceMin(Operand<T> input,
       Operand<U> axis, ReduceMin.Options... options) {
     return ReduceMin.create(scope, input, axis, options);
   }
@@ -3922,7 +3915,7 @@ public <T extends Tensor, U extends Tensor & TNumber> ReduceMin<T> reduceMin(Ope
    * @param options carries optional attributes values
    * @return a new instance of ReduceProd
    */
-  public <T extends Tensor, U extends Tensor & TNumber> ReduceProd<T> reduceProd(Operand<T> input,
+  public <T extends TType, U extends TNumber> ReduceProd<T> reduceProd(Operand<T> input,
       Operand<U> axis, ReduceProd.Options... options) {
     return ReduceProd.create(scope, input, axis, options);
   }
@@ -3942,7 +3935,7 @@ public <T extends Tensor, U extends Tensor & TNumber> ReduceProd<T> reduceProd(O
    * @param options carries optional attributes values
    * @return a new instance of ReduceSum
    */
-  public <T extends Tensor, U extends Tensor & TNumber> ReduceSum<T> reduceSum(Operand<T> input,
+  public <T extends TType, U extends TNumber> ReduceSum<T> reduceSum(Operand<T> input,
       Operand<U> axis, ReduceSum.Options... options) {
     return ReduceSum.create(scope, input, axis, options);
   }
@@ -3954,7 +3947,7 @@ public <T extends Tensor, U extends Tensor & TNumber> ReduceSum<T> reduceSum(Ope
    * @param data The tensor to be made available to the next iteration.
    * @return a new instance of RefNextIteration
    */
-  public <T extends Tensor> RefNextIteration<T> refNextIteration(Operand<T> data) {
+  public <T extends TType> RefNextIteration<T> refNextIteration(Operand<T> data) {
     return RefNextIteration.create(scope, data);
   }
 
@@ -3966,7 +3959,7 @@ public <T extends Tensor> RefNextIteration<T> refNextIteration(Operand<T> data)
    * @param inputs A list of ref tensors, one of which will be forwarded to `output`.
    * @return a new instance of RefSelect
    */
-  public <T extends Tensor> RefSelect<T> refSelect(Operand<TInt32> index,
+  public <T extends TType> RefSelect<T> refSelect(Operand<TInt32> index,
       Iterable<Operand<T>> inputs) {
     return RefSelect.create(scope, index, inputs);
   }
@@ -3984,7 +3977,7 @@ public <T extends Tensor> RefSelect<T> refSelect(Operand<TInt32> index,
    * @param pred A scalar that specifies which output port will receive data.
    * @return a new instance of RefSwitch
    */
-  public <T extends Tensor> RefSwitch<T> refSwitch(Operand<T> data, Operand<TBool> pred) {
+  public <T extends TType> RefSwitch<T> refSwitch(Operand<T> data, Operand<TBool> pred) {
     return RefSwitch.create(scope, data, pred);
   }
 
@@ -4077,7 +4070,7 @@ public RemoteFusedGraphExecute remoteFusedGraphExecute(Iterable<Operand<?>> inpu
    * @param shape Defines the shape of the output tensor.
    * @return a new instance of Reshape
    */
-  public <T extends Tensor, U extends Tensor & TNumber> Reshape<T> reshape(Operand<T> tensor,
+  public <T extends TType, U extends TNumber> Reshape<T> reshape(Operand<T> tensor,
       Operand<U> shape) {
     return Reshape.create(scope, tensor, shape);
   }
@@ -4092,8 +4085,8 @@ public <T extends Tensor, U extends Tensor & TNumber> Reshape<T> reshape(Operand
    * @param T
    * @return a new instance of ResourceCountUpTo
    */
-  public <T extends Tensor & TNumber> ResourceCountUpTo<T> resourceCountUpTo(Operand<?> resource,
-      Long limit, DataType<T> T) {
+  public <T extends TNumber> ResourceCountUpTo<T> resourceCountUpTo(Operand<?> resource, Long limit,
+      DataType<T> T) {
     return ResourceCountUpTo.create(scope, resource, limit, T);
   }
 
@@ -4120,9 +4113,8 @@ public <T extends Tensor & TNumber> ResourceCountUpTo<T> resourceCountUpTo(Opera
    * @param options carries optional attributes values
    * @return a new instance of ResourceGather
    */
-  public <U extends Tensor, T extends Tensor & TNumber> ResourceGather<U> resourceGather(
-      Operand<?> resource, Operand<T> indices, DataType<U> dtype,
-      ResourceGather.Options... options) {
+  public <U extends TType, T extends TNumber> ResourceGather<U> resourceGather(Operand<?> resource,
+      Operand<T> indices, DataType<U> dtype, ResourceGather.Options... options) {
     return ResourceGather.create(scope, resource, indices, dtype, options);
   }
 
@@ -4134,7 +4126,7 @@ public <U extends Tensor, T extends Tensor & TNumber> ResourceGather<U> resource
    * @param dtype
    * @return a new instance of ResourceGatherNd
    */
-  public <U extends Tensor, T extends Tensor & TNumber> ResourceGatherNd<U> resourceGatherNd(
+  public <U extends TType, T extends TNumber> ResourceGatherNd<U> resourceGatherNd(
       Operand<?> resource, Operand<T> indices, DataType<U> dtype) {
     return ResourceGatherNd.create(scope, resource, indices, dtype);
   }
@@ -4167,7 +4159,7 @@ public <U extends Tensor, T extends Tensor & TNumber> ResourceGatherNd<U> resour
    * @param updates A tensor of updated values to add to `ref`.
    * @return a new instance of ResourceScatterAdd
    */
-  public <T extends Tensor & TNumber, U extends Tensor> ResourceScatterAdd resourceScatterAdd(
+  public <T extends TNumber, U extends TType> ResourceScatterAdd resourceScatterAdd(
       Operand<?> resource, Operand<T> indices, Operand<U> updates) {
     return ResourceScatterAdd.create(scope, resource, indices, updates);
   }
@@ -4200,7 +4192,7 @@ public <T extends Tensor & TNumber, U extends Tensor> ResourceScatterAdd resourc
    * @param updates A tensor of updated values to add to `ref`.
    * @return a new instance of ResourceScatterDiv
    */
-  public <T extends Tensor & TNumber, U extends Tensor> ResourceScatterDiv resourceScatterDiv(
+  public <T extends TNumber, U extends TType> ResourceScatterDiv resourceScatterDiv(
       Operand<?> resource, Operand<T> indices, Operand<U> updates) {
     return ResourceScatterDiv.create(scope, resource, indices, updates);
   }
@@ -4233,7 +4225,7 @@ public <T extends Tensor & TNumber, U extends Tensor> ResourceScatterDiv resourc
    * @param updates A tensor of updated values to add to `ref`.
    * @return a new instance of ResourceScatterMax
    */
-  public <T extends Tensor & TNumber, U extends Tensor> ResourceScatterMax resourceScatterMax(
+  public <T extends TNumber, U extends TType> ResourceScatterMax resourceScatterMax(
       Operand<?> resource, Operand<T> indices, Operand<U> updates) {
     return ResourceScatterMax.create(scope, resource, indices, updates);
   }
@@ -4266,7 +4258,7 @@ public <T extends Tensor & TNumber, U extends Tensor> ResourceScatterMax resourc
    * @param updates A tensor of updated values to add to `ref`.
    * @return a new instance of ResourceScatterMin
    */
-  public <T extends Tensor & TNumber, U extends Tensor> ResourceScatterMin resourceScatterMin(
+  public <T extends TNumber, U extends TType> ResourceScatterMin resourceScatterMin(
       Operand<?> resource, Operand<T> indices, Operand<U> updates) {
     return ResourceScatterMin.create(scope, resource, indices, updates);
   }
@@ -4299,7 +4291,7 @@ public <T extends Tensor & TNumber, U extends Tensor> ResourceScatterMin resourc
    * @param updates A tensor of updated values to add to `ref`.
    * @return a new instance of ResourceScatterMul
    */
-  public <T extends Tensor & TNumber, U extends Tensor> ResourceScatterMul resourceScatterMul(
+  public <T extends TNumber, U extends TType> ResourceScatterMul resourceScatterMul(
       Operand<?> resource, Operand<T> indices, Operand<U> updates) {
     return ResourceScatterMul.create(scope, resource, indices, updates);
   }
@@ -4345,7 +4337,7 @@ public <T extends Tensor & TNumber, U extends Tensor> ResourceScatterMul resourc
    * @param options carries optional attributes values
    * @return a new instance of ResourceScatterNdAdd
    */
-  public <T extends Tensor & TNumber, U extends Tensor> ResourceScatterNdAdd resourceScatterNdAdd(
+  public <T extends TNumber, U extends TType> ResourceScatterNdAdd resourceScatterNdAdd(
       Operand<?> ref, Operand<T> indices, Operand<U> updates,
       ResourceScatterNdAdd.Options... options) {
     return ResourceScatterNdAdd.create(scope, ref, indices, updates, options);
@@ -4392,7 +4384,7 @@ public <T extends Tensor & TNumber, U extends Tensor> ResourceScatterNdAdd resou
    * @param options carries optional attributes values
    * @return a new instance of ResourceScatterNdSub
    */
-  public <T extends Tensor & TNumber, U extends Tensor> ResourceScatterNdSub resourceScatterNdSub(
+  public <T extends TNumber, U extends TType> ResourceScatterNdSub resourceScatterNdSub(
       Operand<?> ref, Operand<T> indices, Operand<U> updates,
       ResourceScatterNdSub.Options... options) {
     return ResourceScatterNdSub.create(scope, ref, indices, updates, options);
@@ -4441,7 +4433,7 @@ public <T extends Tensor & TNumber, U extends Tensor> ResourceScatterNdSub resou
    * @param options carries optional attributes values
    * @return a new instance of ResourceScatterNdUpdate
    */
-  public <T extends Tensor & TNumber, U extends Tensor> ResourceScatterNdUpdate resourceScatterNdUpdate(
+  public <T extends TNumber, U extends TType> ResourceScatterNdUpdate resourceScatterNdUpdate(
       Operand<?> ref, Operand<T> indices, Operand<U> updates,
       ResourceScatterNdUpdate.Options... options) {
     return ResourceScatterNdUpdate.create(scope, ref, indices, updates, options);
@@ -4475,7 +4467,7 @@ public <T extends Tensor & TNumber, U extends Tensor> ResourceScatterNdUpdate re
    * @param updates A tensor of updated values to add to `ref`.
    * @return a new instance of ResourceScatterSub
    */
-  public <T extends Tensor & TNumber, U extends Tensor> ResourceScatterSub resourceScatterSub(
+  public <T extends TNumber, U extends TType> ResourceScatterSub resourceScatterSub(
       Operand<?> resource, Operand<T> indices, Operand<U> updates) {
     return ResourceScatterSub.create(scope, resource, indices, updates);
   }
@@ -4499,7 +4491,7 @@ public <T extends Tensor & TNumber, U extends Tensor> ResourceScatterSub resourc
    * @param updates A tensor of updated values to add to `ref`.
    * @return a new instance of ResourceScatterUpdate
    */
-  public <T extends Tensor & TNumber, U extends Tensor> ResourceScatterUpdate resourceScatterUpdate(
+  public <T extends TNumber, U extends TType> ResourceScatterUpdate resourceScatterUpdate(
       Operand<?> resource, Operand<T> indices, Operand<U> updates) {
     return ResourceScatterUpdate.create(scope, resource, indices, updates);
   }
@@ -4522,7 +4514,7 @@ public <T extends Tensor & TNumber, U extends Tensor> ResourceScatterUpdate reso
    * @param options carries optional attributes values
    * @return a new instance of ResourceStridedSliceAssign
    */
-  public <T extends Tensor & TNumber, U extends Tensor> ResourceStridedSliceAssign resourceStridedSliceAssign(
+  public <T extends TNumber, U extends TType> ResourceStridedSliceAssign resourceStridedSliceAssign(
       Operand<?> ref, Operand<T> begin, Operand<T> end, Operand<T> strides, Operand<U> value,
       ResourceStridedSliceAssign.Options... options) {
     return ResourceStridedSliceAssign.create(scope, ref, begin, end, strides, value, options);
@@ -4583,7 +4575,7 @@ public <T extends Tensor & TNumber, U extends Tensor> ResourceStridedSliceAssign
    *  `[-rank(tensor), rank(tensor))`.
    * @return a new instance of Reverse
    */
-  public <T extends Tensor, U extends Tensor & TNumber> Reverse<T> reverse(Operand<T> tensor,
+  public <T extends TType, U extends TNumber> Reverse<T> reverse(Operand<T> tensor,
       Operand<U> axis) {
     return Reverse.create(scope, tensor, axis);
   }
@@ -4651,8 +4643,8 @@ public <T extends Tensor, U extends Tensor & TNumber> Reverse<T> reverse(Operand
    * @param options carries optional attributes values
    * @return a new instance of ReverseSequence
    */
-  public <T extends Tensor, U extends Tensor & TNumber> ReverseSequence<T> reverseSequence(
-      Operand<T> input, Operand<U> seqLengths, Long seqDim, ReverseSequence.Options... options) {
+  public <T extends TType, U extends TNumber> ReverseSequence<T> reverseSequence(Operand<T> input,
+      Operand<U> seqLengths, Long seqDim, ReverseSequence.Options... options) {
     return ReverseSequence.create(scope, input, seqLengths, seqDim, options);
   }
 
@@ -4691,8 +4683,8 @@ public <T extends Tensor, U extends Tensor & TNumber> ReverseSequence<T> reverse
    *  axis.
    * @return a new instance of Roll
    */
-  public <T extends Tensor, U extends Tensor & TNumber, V extends Tensor & TNumber> Roll<T> roll(
-      Operand<T> input, Operand<U> shift, Operand<V> axis) {
+  public <T extends TType, U extends TNumber, V extends TNumber> Roll<T> roll(Operand<T> input,
+      Operand<U> shift, Operand<V> axis) {
     return Roll.create(scope, input, shift, axis);
   }
 
@@ -4793,7 +4785,7 @@ public Rpc rpc(Operand<TString> address, Operand<TString> method, Operand<TStrin
    * @param options carries optional attributes values
    * @return a new instance of ScatterAdd
    */
-  public <T extends Tensor, U extends Tensor & TNumber> ScatterAdd<T> scatterAdd(Operand<T> ref,
+  public <T extends TType, U extends TNumber> ScatterAdd<T> scatterAdd(Operand<T> ref,
       Operand<U> indices, Operand<T> updates, ScatterAdd.Options... options) {
     return ScatterAdd.create(scope, ref, indices, updates, options);
   }
@@ -4827,7 +4819,7 @@ public <T extends Tensor, U extends Tensor & TNumber> ScatterAdd<T> scatterAdd(O
    * @param options carries optional attributes values
    * @return a new instance of ScatterDiv
    */
-  public <T extends Tensor, U extends Tensor & TNumber> ScatterDiv<T> scatterDiv(Operand<T> ref,
+  public <T extends TType, U extends TNumber> ScatterDiv<T> scatterDiv(Operand<T> ref,
       Operand<U> indices, Operand<T> updates, ScatterDiv.Options... options) {
     return ScatterDiv.create(scope, ref, indices, updates, options);
   }
@@ -4865,8 +4857,8 @@ public <T extends Tensor, U extends Tensor & TNumber> ScatterDiv<T> scatterDiv(O
    * @param options carries optional attributes values
    * @return a new instance of ScatterMax
    */
-  public <T extends Tensor & TNumber, U extends Tensor & TNumber> ScatterMax<T> scatterMax(
-      Operand<T> ref, Operand<U> indices, Operand<T> updates, ScatterMax.Options... options) {
+  public <T extends TNumber, U extends TNumber> ScatterMax<T> scatterMax(Operand<T> ref,
+      Operand<U> indices, Operand<T> updates, ScatterMax.Options... options) {
     return ScatterMax.create(scope, ref, indices, updates, options);
   }
 
@@ -4903,8 +4895,8 @@ public <T extends Tensor & TNumber, U extends Tensor & TNumber> ScatterMax<T> sc
    * @param options carries optional attributes values
    * @return a new instance of ScatterMin
    */
-  public <T extends Tensor & TNumber, U extends Tensor & TNumber> ScatterMin<T> scatterMin(
-      Operand<T> ref, Operand<U> indices, Operand<T> updates, ScatterMin.Options... options) {
+  public <T extends TNumber, U extends TNumber> ScatterMin<T> scatterMin(Operand<T> ref,
+      Operand<U> indices, Operand<T> updates, ScatterMin.Options... options) {
     return ScatterMin.create(scope, ref, indices, updates, options);
   }
 
@@ -4937,7 +4929,7 @@ public <T extends Tensor & TNumber, U extends Tensor & TNumber> ScatterMin<T> sc
    * @param options carries optional attributes values
    * @return a new instance of ScatterMul
    */
-  public <T extends Tensor, U extends Tensor & TNumber> ScatterMul<T> scatterMul(Operand<T> ref,
+  public <T extends TType, U extends TNumber> ScatterMul<T> scatterMul(Operand<T> ref,
       Operand<U> indices, Operand<T> updates, ScatterMul.Options... options) {
     return ScatterMul.create(scope, ref, indices, updates, options);
   }
@@ -5028,7 +5020,7 @@ public <T extends Tensor, U extends Tensor & TNumber> ScatterMul<T> scatterMul(O
    * @param shape 1-D. The shape of the resulting tensor.
    * @return a new instance of ScatterNd
    */
-  public <U extends Tensor, T extends Tensor & TNumber> ScatterNd<U> scatterNd(Operand<T> indices,
+  public <U extends TType, T extends TNumber> ScatterNd<U> scatterNd(Operand<T> indices,
       Operand<U> updates, Operand<T> shape) {
     return ScatterNd.create(scope, indices, updates, shape);
   }
@@ -5075,7 +5067,7 @@ public <U extends Tensor, T extends Tensor & TNumber> ScatterNd<U> scatterNd(Ope
    * @param options carries optional attributes values
    * @return a new instance of ScatterNdAdd
    */
-  public <T extends Tensor, U extends Tensor & TNumber> ScatterNdAdd<T> scatterNdAdd(Operand<T> ref,
+  public <T extends TType, U extends TNumber> ScatterNdAdd<T> scatterNdAdd(Operand<T> ref,
       Operand<U> indices, Operand<T> updates, ScatterNdAdd.Options... options) {
     return ScatterNdAdd.create(scope, ref, indices, updates, options);
   }
@@ -5125,7 +5117,7 @@ public <T extends Tensor, U extends Tensor & TNumber> ScatterNdAdd<T> scatterNdA
    *  to add to `input`.
    * @return a new instance of ScatterNdNonAliasingAdd
    */
-  public <T extends Tensor, U extends Tensor & TNumber> ScatterNdNonAliasingAdd<T> scatterNdNonAliasingAdd(
+  public <T extends TType, U extends TNumber> ScatterNdNonAliasingAdd<T> scatterNdNonAliasingAdd(
       Operand<T> input, Operand<U> indices, Operand<T> updates) {
     return ScatterNdNonAliasingAdd.create(scope, input, indices, updates);
   }
@@ -5174,7 +5166,7 @@ public <T extends Tensor, U extends Tensor & TNumber> ScatterNdNonAliasingAdd<T>
    * @param options carries optional attributes values
    * @return a new instance of ScatterNdSub
    */
-  public <T extends Tensor, U extends Tensor & TNumber> ScatterNdSub<T> scatterNdSub(Operand<T> ref,
+  public <T extends TType, U extends TNumber> ScatterNdSub<T> scatterNdSub(Operand<T> ref,
       Operand<U> indices, Operand<T> updates, ScatterNdSub.Options... options) {
     return ScatterNdSub.create(scope, ref, indices, updates, options);
   }
@@ -5225,8 +5217,8 @@ public <T extends Tensor, U extends Tensor & TNumber> ScatterNdSub<T> scatterNdS
    * @param options carries optional attributes values
    * @return a new instance of ScatterNdUpdate
    */
-  public <T extends Tensor, U extends Tensor & TNumber> ScatterNdUpdate<T> scatterNdUpdate(
-      Operand<T> ref, Operand<U> indices, Operand<T> updates, ScatterNdUpdate.Options... options) {
+  public <T extends TType, U extends TNumber> ScatterNdUpdate<T> scatterNdUpdate(Operand<T> ref,
+      Operand<U> indices, Operand<T> updates, ScatterNdUpdate.Options... options) {
     return ScatterNdUpdate.create(scope, ref, indices, updates, options);
   }
 
@@ -5262,7 +5254,7 @@ public <T extends Tensor, U extends Tensor & TNumber> ScatterNdUpdate<T> scatter
    * @param options carries optional attributes values
    * @return a new instance of ScatterSub
    */
-  public <T extends Tensor, U extends Tensor & TNumber> ScatterSub<T> scatterSub(Operand<T> ref,
+  public <T extends TType, U extends TNumber> ScatterSub<T> scatterSub(Operand<T> ref,
       Operand<U> indices, Operand<T> updates, ScatterSub.Options... options) {
     return ScatterSub.create(scope, ref, indices, updates, options);
   }
@@ -5303,8 +5295,8 @@ public <T extends Tensor, U extends Tensor & TNumber> ScatterSub<T> scatterSub(O
    * @param options carries optional attributes values
    * @return a new instance of ScatterUpdate
    */
-  public <T extends Tensor, U extends Tensor & TNumber> ScatterUpdate<T> scatterUpdate(
-      Operand<T> ref, Operand<U> indices, Operand<T> updates, ScatterUpdate.Options... options) {
+  public <T extends TType, U extends TNumber> ScatterUpdate<T> scatterUpdate(Operand<T> ref,
+      Operand<U> indices, Operand<T> updates, ScatterUpdate.Options... options) {
     return ScatterUpdate.create(scope, ref, indices, updates, options);
   }
 
@@ -5316,7 +5308,7 @@ public <T extends Tensor, U extends Tensor & TNumber> ScatterUpdate<T> scatterUp
    * @param e
    * @return a new instance of Select
    */
-  public <T extends Tensor> Select<T> select(Operand<TBool> condition, Operand<T> t, Operand<T> e) {
+  public <T extends TType> Select<T> select(Operand<TBool> condition, Operand<T> t, Operand<T> e) {
     return Select.create(scope, condition, t, e);
   }
 
@@ -5348,7 +5340,7 @@ public <T extends Tensor> Select<T> select(Operand<TBool> condition, Operand<T>
    * @param y 1-D. Values to remove.
    * @return a new instance of SetDiff1d
    */
-  public <T extends Tensor> SetDiff1d<T, TInt32> setDiff1d(Operand<T> x, Operand<T> y) {
+  public <T extends TType> SetDiff1d<T, TInt32> setDiff1d(Operand<T> x, Operand<T> y) {
     return SetDiff1d.create(scope, x, y);
   }
 
@@ -5381,8 +5373,8 @@ public <T extends Tensor> SetDiff1d<T, TInt32> setDiff1d(Operand<T> x, Operand<T
    * @param outIdx
    * @return a new instance of SetDiff1d
    */
-  public <T extends Tensor, U extends Tensor & TNumber> SetDiff1d<T, U> setDiff1d(Operand<T> x,
-      Operand<T> y, DataType<U> outIdx) {
+  public <T extends TType, U extends TNumber> SetDiff1d<T, U> setDiff1d(Operand<T> x, Operand<T> y,
+      DataType<U> outIdx) {
     return SetDiff1d.create(scope, x, y, outIdx);
   }
 
@@ -5402,7 +5394,7 @@ public <T extends Tensor, U extends Tensor & TNumber> SetDiff1d<T, U> setDiff1d(
    * @param options carries optional attributes values
    * @return a new instance of SetSize
    */
-  public <T extends Tensor> SetSize setSize(Operand<TInt64> setIndices, Operand<T> setValues,
+  public <T extends TType> SetSize setSize(Operand<TInt64> setIndices, Operand<T> setValues,
       Operand<TInt64> setShape, SetSize.Options... options) {
     return SetSize.create(scope, setIndices, setValues, setShape, options);
   }
@@ -5422,7 +5414,7 @@ public <T extends Tensor> SetSize setSize(Operand<TInt64> setIndices, Operand<T>
    * @param input
    * @return a new instance of Shape
    */
-  public <T extends Tensor> org.tensorflow.op.core.Shape<TInt32> shape(Operand<T> input) {
+  public <T extends TType> org.tensorflow.op.core.Shape<TInt32> shape(Operand<T> input) {
     return org.tensorflow.op.core.Shape.create(scope, input);
   }
 
@@ -5442,7 +5434,7 @@ public <T extends Tensor> org.tensorflow.op.core.Shape<TInt32> shape(Operand<T>
    * @param outType
    * @return a new instance of Shape
    */
-  public <U extends Tensor & TNumber, T extends Tensor> org.tensorflow.op.core.Shape<U> shape(
+  public <U extends TNumber, T extends TType> org.tensorflow.op.core.Shape<U> shape(
       Operand<T> input, DataType<U> outType) {
     return org.tensorflow.op.core.Shape.create(scope, input, outType);
   }
@@ -5456,7 +5448,7 @@ public <U extends Tensor & TNumber, T extends Tensor> org.tensorflow.op.core.Sha
    * @param input
    * @return a new instance of ShapeN
    */
-  public <T extends Tensor> ShapeN<TInt32> shapeN(Iterable<Operand<T>> input) {
+  public <T extends TType> ShapeN<TInt32> shapeN(Iterable<Operand<T>> input) {
     return ShapeN.create(scope, input);
   }
 
@@ -5470,7 +5462,7 @@ public <T extends Tensor> ShapeN<TInt32> shapeN(Iterable<Operand<T>> input) {
    * @param outType
    * @return a new instance of ShapeN
    */
-  public <U extends Tensor & TNumber, T extends Tensor> ShapeN<U> shapeN(Iterable<Operand<T>> input,
+  public <U extends TNumber, T extends TType> ShapeN<U> shapeN(Iterable<Operand<T>> input,
       DataType<U> outType) {
     return ShapeN.create(scope, input, outType);
   }
@@ -5491,7 +5483,7 @@ public <U extends Tensor & TNumber, T extends Tensor> ShapeN<U> shapeN(Iterable<
    * @param input
    * @return a new instance of Size
    */
-  public <T extends Tensor> Size<TInt32> size(Operand<T> input) {
+  public <T extends TType> Size<TInt32> size(Operand<T> input) {
     return Size.create(scope, input);
   }
 
@@ -5512,8 +5504,7 @@ public <T extends Tensor> Size<TInt32> size(Operand<T> input) {
    * @param outType
    * @return a new instance of Size
    */
-  public <U extends Tensor & TNumber, T extends Tensor> Size<U> size(Operand<T> input,
-      DataType<U> outType) {
+  public <U extends TNumber, T extends TType> Size<U> size(Operand<T> input, DataType<U> outType) {
     return Size.create(scope, input, outType);
   }
 
@@ -5549,8 +5540,8 @@ public Skipgram skipgram(String filename, Long batchSize, Skipgram.Options... op
    *  size[i] = input.dim_size(i) - begin[i]).
    * @return a new instance of Slice
    */
-  public <T extends Tensor, U extends Tensor & TNumber> Slice<T> slice(Operand<T> input,
-      Operand<U> begin, Operand<U> size) {
+  public <T extends TType, U extends TNumber> Slice<T> slice(Operand<T> input, Operand<U> begin,
+      Operand<U> size) {
     return Slice.create(scope, input, begin, size);
   }
 
@@ -5561,7 +5552,7 @@ public <T extends Tensor, U extends Tensor & TNumber> Slice<T> slice(Operand<T>
    * @param input
    * @return a new instance of Snapshot
    */
-  public <T extends Tensor> Snapshot<T> snapshot(Operand<T> input) {
+  public <T extends TType> Snapshot<T> snapshot(Operand<T> input) {
     return Snapshot.create(scope, input);
   }
 
@@ -5675,7 +5666,7 @@ public <T extends Tensor> Snapshot<T> snapshot(Operand<T> input) {
    *  regular convolution.
    * @return a new instance of SpaceToBatchNd
    */
-  public <T extends Tensor, U extends Tensor & TNumber, V extends Tensor & TNumber> SpaceToBatchNd<T> spaceToBatchNd(
+  public <T extends TType, U extends TNumber, V extends TNumber> SpaceToBatchNd<T> spaceToBatchNd(
       Operand<T> input, Operand<U> blockShape, Operand<V> paddings) {
     return SpaceToBatchNd.create(scope, input, blockShape, paddings);
   }
@@ -5691,7 +5682,7 @@ public <T extends Tensor, U extends Tensor & TNumber, V extends Tensor & TNumber
    *  `value.shape[split_dim]`.
    * @return a new instance of Split
    */
-  public <T extends Tensor> Split<T> split(Operand<TInt32> axis, Operand<T> value, Long numSplit) {
+  public <T extends TType> Split<T> split(Operand<TInt32> axis, Operand<T> value, Long numSplit) {
     return Split.create(scope, axis, value, numSplit);
   }
 
@@ -5708,7 +5699,7 @@ public <T extends Tensor> Split<T> split(Operand<TInt32> axis, Operand<T> value,
    * @param numSplit
    * @return a new instance of SplitV
    */
-  public <T extends Tensor, U extends Tensor & TNumber> SplitV<T> splitV(Operand<T> value,
+  public <T extends TType, U extends TNumber> SplitV<T> splitV(Operand<T> value,
       Operand<U> sizeSplits, Operand<TInt32> axis, Long numSplit) {
     return SplitV.create(scope, value, sizeSplits, axis, numSplit);
   }
@@ -5737,7 +5728,7 @@ public <T extends Tensor, U extends Tensor & TNumber> SplitV<T> splitV(Operand<T
    * @param options carries optional attributes values
    * @return a new instance of Squeeze
    */
-  public <T extends Tensor> Squeeze<T> squeeze(Operand<T> input, Squeeze.Options... options) {
+  public <T extends TType> Squeeze<T> squeeze(Operand<T> input, Squeeze.Options... options) {
     return Squeeze.create(scope, input, options);
   }
 
@@ -5767,7 +5758,7 @@ public <T extends Tensor> Squeeze<T> squeeze(Operand<T> input, Squeeze.Options..
    * @param options carries optional attributes values
    * @return a new instance of Stack
    */
-  public <T extends Tensor> Stack<T> stack(Iterable<Operand<T>> values, Stack.Options... options) {
+  public <T extends TType> Stack<T> stack(Iterable<Operand<T>> values, Stack.Options... options) {
     return Stack.create(scope, values, options);
   }
 
@@ -5857,7 +5848,7 @@ public StageSize stageSize(List<DataType<?>> dtypes, StageSize.Options... option
    * @param input
    * @return a new instance of StopGradient
    */
-  public <T extends Tensor> StopGradient<T> stopGradient(Operand<T> input) {
+  public <T extends TType> StopGradient<T> stopGradient(Operand<T> input) {
     return StopGradient.create(scope, input);
   }
 
@@ -5968,9 +5959,8 @@ public <T extends Tensor> StopGradient<T> stopGradient(Operand<T> input) {
    * @param options carries optional attributes values
    * @return a new instance of StridedSlice
    */
-  public <T extends Tensor, U extends Tensor & TNumber> StridedSlice<T> stridedSlice(
-      Operand<T> input, Operand<U> begin, Operand<U> end, Operand<U> strides,
-      StridedSlice.Options... options) {
+  public <T extends TType, U extends TNumber> StridedSlice<T> stridedSlice(Operand<T> input,
+      Operand<U> begin, Operand<U> end, Operand<U> strides, StridedSlice.Options... options) {
     return StridedSlice.create(scope, input, begin, end, strides, options);
   }
 
@@ -5993,7 +5983,7 @@ public <T extends Tensor, U extends Tensor & TNumber> StridedSlice<T> stridedSli
    * @param options carries optional attributes values
    * @return a new instance of StridedSliceAssign
    */
-  public <T extends Tensor, U extends Tensor & TNumber> StridedSliceAssign<T> stridedSliceAssign(
+  public <T extends TType, U extends TNumber> StridedSliceAssign<T> stridedSliceAssign(
       Operand<T> ref, Operand<U> begin, Operand<U> end, Operand<U> strides, Operand<T> value,
       StridedSliceAssign.Options... options) {
     return StridedSliceAssign.create(scope, ref, begin, end, strides, value, options);
@@ -6020,8 +6010,8 @@ public <T extends Tensor, U extends Tensor & TNumber> StridedSliceAssign<T> stri
    * @param options carries optional attributes values
    * @return a new instance of StridedSliceGrad
    */
-  public <U extends Tensor, T extends Tensor & TNumber> StridedSliceGrad<U> stridedSliceGrad(
-      Operand<T> shape, Operand<T> begin, Operand<T> end, Operand<T> strides, Operand<U> dy,
+  public <U extends TType, T extends TNumber> StridedSliceGrad<U> stridedSliceGrad(Operand<T> shape,
+      Operand<T> begin, Operand<T> end, Operand<T> strides, Operand<U> dy,
       StridedSliceGrad.Options... options) {
     return StridedSliceGrad.create(scope, shape, begin, end, strides, dy, options);
   }
@@ -6041,8 +6031,8 @@ public <U extends Tensor, T extends Tensor & TNumber> StridedSliceGrad<U> stride
    * @param options carries optional attributes values
    * @return a new instance of Sum
    */
-  public <T extends Tensor, U extends Tensor & TNumber> Sum<T> sum(Operand<T> input,
-      Operand<U> axis, Sum.Options... options) {
+  public <T extends TType, U extends TNumber> Sum<T> sum(Operand<T> input, Operand<U> axis,
+      Sum.Options... options) {
     return Sum.create(scope, input, axis, options);
   }
 
@@ -6059,7 +6049,7 @@ public <T extends Tensor, U extends Tensor & TNumber> Sum<T> sum(Operand<T> inpu
    * @param pred A scalar that specifies which output port will receive data.
    * @return a new instance of SwitchCond
    */
-  public <T extends Tensor> SwitchCond<T> switchCond(Operand<T> data, Operand<TBool> pred) {
+  public <T extends TType> SwitchCond<T> switchCond(Operand<T> data, Operand<TBool> pred) {
     return SwitchCond.create(scope, data, pred);
   }
 
@@ -6087,7 +6077,7 @@ public <T extends Tensor> SwitchCond<T> switchCond(Operand<T> data, Operand<TBoo
    * @param options carries optional attributes values
    * @return a new instance of TemporaryVariable
    */
-  public <T extends Tensor> TemporaryVariable<T> temporaryVariable(Shape shape, DataType<T> dtype,
+  public <T extends TType> TemporaryVariable<T> temporaryVariable(Shape shape, DataType<T> dtype,
       TemporaryVariable.Options... options) {
     return TemporaryVariable.create(scope, shape, dtype, options);
   }
@@ -6102,7 +6092,7 @@ public <T extends Tensor> TemporaryVariable<T> temporaryVariable(Shape shape, Da
    * @param options carries optional attributes values
    * @return a new instance of TensorArray
    */
-  public <T extends Tensor> TensorArray tensorArray(Operand<TInt32> size, DataType<T> dtype,
+  public <T extends TType> TensorArray tensorArray(Operand<TInt32> size, DataType<T> dtype,
       TensorArray.Options... options) {
     return TensorArray.create(scope, size, dtype, options);
   }
@@ -6141,7 +6131,7 @@ public TensorArrayClose tensorArrayClose(Operand<?> handle) {
    * @param options carries optional attributes values
    * @return a new instance of TensorArrayConcat
    */
-  public <T extends Tensor> TensorArrayConcat<T> tensorArrayConcat(Operand<?> handle,
+  public <T extends TType> TensorArrayConcat<T> tensorArrayConcat(Operand<?> handle,
       Operand<TFloat32> flowIn, DataType<T> dtype, TensorArrayConcat.Options... options) {
     return TensorArrayConcat.create(scope, handle, flowIn, dtype, options);
   }
@@ -6159,7 +6149,7 @@ public <T extends Tensor> TensorArrayConcat<T> tensorArrayConcat(Operand<?> hand
    * @param options carries optional attributes values
    * @return a new instance of TensorArrayGather
    */
-  public <T extends Tensor> TensorArrayGather<T> tensorArrayGather(Operand<?> handle,
+  public <T extends TType> TensorArrayGather<T> tensorArrayGather(Operand<?> handle,
       Operand<TInt32> indices, Operand<TFloat32> flowIn, DataType<T> dtype,
       TensorArrayGather.Options... options) {
     return TensorArrayGather.create(scope, handle, indices, flowIn, dtype, options);
@@ -6247,7 +6237,7 @@ public TensorArrayGradWithShape tensorArrayGradWithShape(Operand<?> handle,
    * @param options carries optional attributes values
    * @return a new instance of TensorArrayPack
    */
-  public <T extends Tensor> TensorArrayPack<T> tensorArrayPack(Operand<TString> handle,
+  public <T extends TType> TensorArrayPack<T> tensorArrayPack(Operand<TString> handle,
       Operand<TFloat32> flowIn, DataType<T> dtype, TensorArrayPack.Options... options) {
     return TensorArrayPack.create(scope, handle, flowIn, dtype, options);
   }
@@ -6262,7 +6252,7 @@ public <T extends Tensor> TensorArrayPack<T> tensorArrayPack(Operand<TString> ha
    * @param dtype The type of the elem that is returned.
    * @return a new instance of TensorArrayRead
    */
-  public <T extends Tensor> TensorArrayRead<T> tensorArrayRead(Operand<?> handle,
+  public <T extends TType> TensorArrayRead<T> tensorArrayRead(Operand<?> handle,
       Operand<TInt32> index, Operand<TFloat32> flowIn, DataType<T> dtype) {
     return TensorArrayRead.create(scope, handle, index, flowIn, dtype);
   }
@@ -6278,7 +6268,7 @@ public <T extends Tensor> TensorArrayRead<T> tensorArrayRead(Operand<?> handle,
    * @param flowIn A float scalar that enforces proper chaining of operations.
    * @return a new instance of TensorArrayScatter
    */
-  public <T extends Tensor> TensorArrayScatter tensorArrayScatter(Operand<?> handle,
+  public <T extends TType> TensorArrayScatter tensorArrayScatter(Operand<?> handle,
       Operand<TInt32> indices, Operand<T> value, Operand<TFloat32> flowIn) {
     return TensorArrayScatter.create(scope, handle, indices, value, flowIn);
   }
@@ -6325,7 +6315,7 @@ public TensorArraySize tensorArraySize(Operand<?> handle, Operand<TFloat32> flow
    * @param flowIn A float scalar that enforces proper chaining of operations.
    * @return a new instance of TensorArraySplit
    */
-  public <T extends Tensor> TensorArraySplit tensorArraySplit(Operand<?> handle, Operand<T> value,
+  public <T extends TType> TensorArraySplit tensorArraySplit(Operand<?> handle, Operand<T> value,
       Operand<TInt64> lengths, Operand<TFloat32> flowIn) {
     return TensorArraySplit.create(scope, handle, value, lengths, flowIn);
   }
@@ -6337,7 +6327,7 @@ public <T extends Tensor> TensorArraySplit tensorArraySplit(Operand<?> handle, O
    * @param flowIn
    * @return a new instance of TensorArrayUnpack
    */
-  public <T extends Tensor> TensorArrayUnpack tensorArrayUnpack(Operand<TString> handle,
+  public <T extends TType> TensorArrayUnpack tensorArrayUnpack(Operand<TString> handle,
       Operand<T> value, Operand<TFloat32> flowIn) {
     return TensorArrayUnpack.create(scope, handle, value, flowIn);
   }
@@ -6351,7 +6341,7 @@ public <T extends Tensor> TensorArrayUnpack tensorArrayUnpack(Operand<TString> h
    * @param flowIn A float scalar that enforces proper chaining of operations.
    * @return a new instance of TensorArrayWrite
    */
-  public <T extends Tensor> TensorArrayWrite tensorArrayWrite(Operand<?> handle,
+  public <T extends TType> TensorArrayWrite tensorArrayWrite(Operand<?> handle,
       Operand<TInt32> index, Operand<T> value, Operand<TFloat32> flowIn) {
     return TensorArrayWrite.create(scope, handle, index, value, flowIn);
   }
@@ -6378,7 +6368,7 @@ public <T extends Tensor> TensorArrayWrite tensorArrayWrite(Operand<?> handle,
    * @param elementDtype
    * @return a new instance of TensorListConcat
    */
-  public <U extends Tensor, T extends Tensor & TNumber> TensorListConcat<U> tensorListConcat(
+  public <U extends TType, T extends TNumber> TensorListConcat<U> tensorListConcat(
       Operand<?> inputHandle, Operand<T> elementShape, Operand<TInt64> leadingDims,
       DataType<U> elementDtype) {
     return TensorListConcat.create(scope, inputHandle, elementShape, leadingDims, elementDtype);
@@ -6391,7 +6381,7 @@ public <U extends Tensor, T extends Tensor & TNumber> TensorListConcat<U> tensor
    * @param elementDtype
    * @return a new instance of TensorListConcatLists
    */
-  public <T extends Tensor> TensorListConcatLists tensorListConcatLists(Operand<?> inputA,
+  public <T extends TType> TensorListConcatLists tensorListConcatLists(Operand<?> inputA,
       Operand<?> inputB, DataType<T> elementDtype) {
     return TensorListConcatLists.create(scope, inputA, inputB, elementDtype);
   }
@@ -6407,7 +6397,7 @@ public <T extends Tensor> TensorListConcatLists tensorListConcatLists(Operand<?>
    * @param shapeType
    * @return a new instance of TensorListElementShape
    */
-  public <T extends Tensor & TNumber> TensorListElementShape<T> tensorListElementShape(
+  public <T extends TNumber> TensorListElementShape<T> tensorListElementShape(
       Operand<?> inputHandle, DataType<T> shapeType) {
     return TensorListElementShape.create(scope, inputHandle, shapeType);
   }
@@ -6424,7 +6414,7 @@ public <T extends Tensor & TNumber> TensorListElementShape<T> tensorListElementS
    * @param elementShape
    * @return a new instance of TensorListFromTensor
    */
-  public <T extends Tensor, U extends Tensor & TNumber> TensorListFromTensor tensorListFromTensor(
+  public <T extends TType, U extends TNumber> TensorListFromTensor tensorListFromTensor(
       Operand<T> tensor, Operand<U> elementShape) {
     return TensorListFromTensor.create(scope, tensor, elementShape);
   }
@@ -6446,7 +6436,7 @@ public <T extends Tensor, U extends Tensor & TNumber> TensorListFromTensor tenso
    * @param elementDtype
    * @return a new instance of TensorListGather
    */
-  public <T extends Tensor> TensorListGather<T> tensorListGather(Operand<?> inputHandle,
+  public <T extends TType> TensorListGather<T> tensorListGather(Operand<?> inputHandle,
       Operand<TInt32> indices, Operand<TInt32> elementShape, DataType<T> elementDtype) {
     return TensorListGather.create(scope, inputHandle, indices, elementShape, elementDtype);
   }
@@ -6460,7 +6450,7 @@ public <T extends Tensor> TensorListGather<T> tensorListGather(Operand<?> inputH
    * @param elementDtype
    * @return a new instance of TensorListGetItem
    */
-  public <T extends Tensor> TensorListGetItem<T> tensorListGetItem(Operand<?> inputHandle,
+  public <T extends TType> TensorListGetItem<T> tensorListGetItem(Operand<?> inputHandle,
       Operand<TInt32> index, Operand<TInt32> elementShape, DataType<T> elementDtype) {
     return TensorListGetItem.create(scope, inputHandle, index, elementShape, elementDtype);
   }
@@ -6494,7 +6484,7 @@ public TensorListLength tensorListLength(Operand<?> inputHandle) {
    * @param elementDtype
    * @return a new instance of TensorListPopBack
    */
-  public <T extends Tensor> TensorListPopBack<T> tensorListPopBack(Operand<?> inputHandle,
+  public <T extends TType> TensorListPopBack<T> tensorListPopBack(Operand<?> inputHandle,
       Operand<TInt32> elementShape, DataType<T> elementDtype) {
     return TensorListPopBack.create(scope, inputHandle, elementShape, elementDtype);
   }
@@ -6512,7 +6502,7 @@ public <T extends Tensor> TensorListPopBack<T> tensorListPopBack(Operand<?> inpu
    * @param tensor
    * @return a new instance of TensorListPushBack
    */
-  public <T extends Tensor> TensorListPushBack tensorListPushBack(Operand<?> inputHandle,
+  public <T extends TType> TensorListPushBack tensorListPushBack(Operand<?> inputHandle,
       Operand<T> tensor) {
     return TensorListPushBack.create(scope, inputHandle, tensor);
   }
@@ -6523,7 +6513,7 @@ public <T extends Tensor> TensorListPushBack tensorListPushBack(Operand<?> input
    * @param tensor
    * @return a new instance of TensorListPushBackBatch
    */
-  public <T extends Tensor> TensorListPushBackBatch tensorListPushBackBatch(Operand<?> inputHandles,
+  public <T extends TType> TensorListPushBackBatch tensorListPushBackBatch(Operand<?> inputHandles,
       Operand<T> tensor) {
     return TensorListPushBackBatch.create(scope, inputHandles, tensor);
   }
@@ -6541,7 +6531,7 @@ public <T extends Tensor> TensorListPushBackBatch tensorListPushBackBatch(Operan
    * @param elementDtype
    * @return a new instance of TensorListReserve
    */
-  public <T extends Tensor & TNumber, U extends Tensor> TensorListReserve tensorListReserve(
+  public <T extends TNumber, U extends TType> TensorListReserve tensorListReserve(
       Operand<T> elementShape, Operand<TInt32> numElements, DataType<U> elementDtype) {
     return TensorListReserve.create(scope, elementShape, numElements, elementDtype);
   }
@@ -6582,9 +6572,8 @@ public TensorListResize tensorListResize(Operand<?> inputHandle, Operand<TInt32>
    * @param numElements
    * @return a new instance of TensorListScatter
    */
-  public <T extends Tensor, U extends Tensor & TNumber> TensorListScatter tensorListScatter(
-      Operand<T> tensor, Operand<TInt32> indices, Operand<U> elementShape,
-      Operand<TInt32> numElements) {
+  public <T extends TType, U extends TNumber> TensorListScatter tensorListScatter(Operand<T> tensor,
+      Operand<TInt32> indices, Operand<U> elementShape, Operand<TInt32> numElements) {
     return TensorListScatter.create(scope, tensor, indices, elementShape, numElements);
   }
 
@@ -6604,7 +6593,7 @@ public <T extends Tensor, U extends Tensor & TNumber> TensorListScatter tensorLi
    * @param indices
    * @return a new instance of TensorListScatterIntoExistingList
    */
-  public <T extends Tensor> TensorListScatterIntoExistingList tensorListScatterIntoExistingList(
+  public <T extends TType> TensorListScatterIntoExistingList tensorListScatterIntoExistingList(
       Operand<?> inputHandle, Operand<T> tensor, Operand<TInt32> indices) {
     return TensorListScatterIntoExistingList.create(scope, inputHandle, tensor, indices);
   }
@@ -6616,7 +6605,7 @@ public <T extends Tensor> TensorListScatterIntoExistingList tensorListScatterInt
    * @param item
    * @return a new instance of TensorListSetItem
    */
-  public <T extends Tensor> TensorListSetItem tensorListSetItem(Operand<?> inputHandle,
+  public <T extends TType> TensorListSetItem tensorListSetItem(Operand<?> inputHandle,
       Operand<TInt32> index, Operand<T> item) {
     return TensorListSetItem.create(scope, inputHandle, index, item);
   }
@@ -6637,8 +6626,8 @@ public <T extends Tensor> TensorListSetItem tensorListSetItem(Operand<?> inputHa
    * @param lengths
    * @return a new instance of TensorListSplit
    */
-  public <T extends Tensor, U extends Tensor & TNumber> TensorListSplit tensorListSplit(
-      Operand<T> tensor, Operand<U> elementShape, Operand<TInt64> lengths) {
+  public <T extends TType, U extends TNumber> TensorListSplit tensorListSplit(Operand<T> tensor,
+      Operand<U> elementShape, Operand<TInt64> lengths) {
     return TensorListSplit.create(scope, tensor, elementShape, lengths);
   }
 
@@ -6658,7 +6647,7 @@ public <T extends Tensor, U extends Tensor & TNumber> TensorListSplit tensorList
    * @param options carries optional attributes values
    * @return a new instance of TensorListStack
    */
-  public <T extends Tensor> TensorListStack<T> tensorListStack(Operand<?> inputHandle,
+  public <T extends TType> TensorListStack<T> tensorListStack(Operand<?> inputHandle,
       Operand<TInt32> elementShape, DataType<T> elementDtype, TensorListStack.Options... options) {
     return TensorListStack.create(scope, inputHandle, elementShape, elementDtype, options);
   }
@@ -6731,7 +6720,7 @@ public <T extends Tensor> TensorListStack<T> tensorListStack(Operand<?> inputHan
    * @param updates Updates to scatter into output.
    * @return a new instance of TensorScatterNdAdd
    */
-  public <T extends Tensor, U extends Tensor & TNumber> TensorScatterNdAdd<T> tensorScatterNdAdd(
+  public <T extends TType, U extends TNumber> TensorScatterNdAdd<T> tensorScatterNdAdd(
       Operand<T> tensor, Operand<U> indices, Operand<T> updates) {
     return TensorScatterNdAdd.create(scope, tensor, indices, updates);
   }
@@ -6804,7 +6793,7 @@ public <T extends Tensor, U extends Tensor & TNumber> TensorScatterNdAdd<T> tens
    * @param updates Updates to scatter into output.
    * @return a new instance of TensorScatterNdSub
    */
-  public <T extends Tensor, U extends Tensor & TNumber> TensorScatterNdSub<T> tensorScatterNdSub(
+  public <T extends TType, U extends TNumber> TensorScatterNdSub<T> tensorScatterNdSub(
       Operand<T> tensor, Operand<U> indices, Operand<T> updates) {
     return TensorScatterNdSub.create(scope, tensor, indices, updates);
   }
@@ -6892,7 +6881,7 @@ public <T extends Tensor, U extends Tensor & TNumber> TensorScatterNdSub<T> tens
    * @param updates Updates to scatter into output.
    * @return a new instance of TensorScatterNdUpdate
    */
-  public <T extends Tensor, U extends Tensor & TNumber> TensorScatterNdUpdate<T> tensorScatterNdUpdate(
+  public <T extends TType, U extends TNumber> TensorScatterNdUpdate<T> tensorScatterNdUpdate(
       Operand<T> tensor, Operand<U> indices, Operand<T> updates) {
     return TensorScatterNdUpdate.create(scope, tensor, indices, updates);
   }
@@ -6916,7 +6905,7 @@ public <T extends Tensor, U extends Tensor & TNumber> TensorScatterNdUpdate<T> t
    * @param options carries optional attributes values
    * @return a new instance of TensorStridedSliceUpdate
    */
-  public <T extends Tensor, U extends Tensor & TNumber> TensorStridedSliceUpdate<T> tensorStridedSliceUpdate(
+  public <T extends TType, U extends TNumber> TensorStridedSliceUpdate<T> tensorStridedSliceUpdate(
       Operand<T> input, Operand<U> begin, Operand<U> end, Operand<U> strides, Operand<T> value,
       TensorStridedSliceUpdate.Options... options) {
     return TensorStridedSliceUpdate.create(scope, input, begin, end, strides, value, options);
@@ -6957,8 +6946,7 @@ public <T extends Tensor, U extends Tensor & TNumber> TensorStridedSliceUpdate<T
    * @param multiples 1-D. Length must be the same as the number of dimensions in `input`
    * @return a new instance of Tile
    */
-  public <T extends Tensor, U extends Tensor & TNumber> Tile<T> tile(Operand<T> input,
-      Operand<U> multiples) {
+  public <T extends TType, U extends TNumber> Tile<T> tile(Operand<T> input, Operand<U> multiples) {
     return Tile.create(scope, input, multiples);
   }
 
@@ -7072,7 +7060,7 @@ public TryRpc tryRpc(Operand<TString> address, Operand<TString> method, Operand<
    * @param options carries optional attributes values
    * @return a new instance of Unbatch
    */
-  public <T extends Tensor> Unbatch<T> unbatch(Operand<T> batchedTensor, Operand<TInt64> batchIndex,
+  public <T extends TType> Unbatch<T> unbatch(Operand<T> batchedTensor, Operand<TInt64> batchIndex,
       Operand<TInt64> id, Long timeoutMicros, Unbatch.Options... options) {
     return Unbatch.create(scope, batchedTensor, batchIndex, id, timeoutMicros, options);
   }
@@ -7103,7 +7091,7 @@ public <T extends Tensor> Unbatch<T> unbatch(Operand<T> batchedTensor, Operand<T
    * @param options carries optional attributes values
    * @return a new instance of UnbatchGrad
    */
-  public <T extends Tensor> UnbatchGrad<T> unbatchGrad(Operand<T> originalInput,
+  public <T extends TType> UnbatchGrad<T> unbatchGrad(Operand<T> originalInput,
       Operand<TInt64> batchIndex, Operand<T> grad, Operand<TInt64> id,
       UnbatchGrad.Options... options) {
     return UnbatchGrad.create(scope, originalInput, batchIndex, grad, id, options);
@@ -7158,7 +7146,7 @@ public <T extends Tensor> UnbatchGrad<T> unbatchGrad(Operand<T> originalInput,
    *  find the unique elements.
    * @return a new instance of Unique
    */
-  public <T extends Tensor, U extends Tensor & TNumber> Unique<T, TInt32> unique(Operand<T> x,
+  public <T extends TType, U extends TNumber> Unique<T, TInt32> unique(Operand<T> x,
       Operand<U> axis) {
     return Unique.create(scope, x, axis);
   }
@@ -7213,8 +7201,8 @@ public <T extends Tensor, U extends Tensor & TNumber> Unique<T, TInt32> unique(O
    * @param outIdx
    * @return a new instance of Unique
    */
-  public <T extends Tensor, V extends Tensor & TNumber, U extends Tensor & TNumber> Unique<T, V> unique(
-      Operand<T> x, Operand<U> axis, DataType<V> outIdx) {
+  public <T extends TType, V extends TNumber, U extends TNumber> Unique<T, V> unique(Operand<T> x,
+      Operand<U> axis, DataType<V> outIdx) {
     return Unique.create(scope, x, axis, outIdx);
   }
 
@@ -7271,7 +7259,7 @@ public <T extends Tensor, V extends Tensor & TNumber, U extends Tensor & TNumber
    *  find the unique elements.
    * @return a new instance of UniqueWithCounts
    */
-  public <T extends Tensor, U extends Tensor & TNumber> UniqueWithCounts<T, TInt32> uniqueWithCounts(
+  public <T extends TType, U extends TNumber> UniqueWithCounts<T, TInt32> uniqueWithCounts(
       Operand<T> x, Operand<U> axis) {
     return UniqueWithCounts.create(scope, x, axis);
   }
@@ -7330,7 +7318,7 @@ public <T extends Tensor, U extends Tensor & TNumber> UniqueWithCounts<T, TInt32
    * @param outIdx
    * @return a new instance of UniqueWithCounts
    */
-  public <T extends Tensor, V extends Tensor & TNumber, U extends Tensor & TNumber> UniqueWithCounts<T, V> uniqueWithCounts(
+  public <T extends TType, V extends TNumber, U extends TNumber> UniqueWithCounts<T, V> uniqueWithCounts(
       Operand<T> x, Operand<U> axis, DataType<V> outIdx) {
     return UniqueWithCounts.create(scope, x, axis, outIdx);
   }
@@ -7363,8 +7351,7 @@ public <T extends Tensor, V extends Tensor & TNumber, U extends Tensor & TNumber
    *  indices.
    * @return a new instance of UnravelIndex
    */
-  public <T extends Tensor & TNumber> UnravelIndex<T> unravelIndex(Operand<T> indices,
-      Operand<T> dims) {
+  public <T extends TNumber> UnravelIndex<T> unravelIndex(Operand<T> indices, Operand<T> dims) {
     return UnravelIndex.create(scope, indices, dims);
   }
 
@@ -7390,7 +7377,7 @@ public <T extends Tensor & TNumber> UnravelIndex<T> unravelIndex(Operand<T> indi
    * @param options carries optional attributes values
    * @return a new instance of Unstack
    */
-  public <T extends Tensor> Unstack<T> unstack(Operand<T> value, Long num,
+  public <T extends TType> Unstack<T> unstack(Operand<T> value, Long num,
       Unstack.Options... options) {
     return Unstack.create(scope, value, num, options);
   }
@@ -7418,7 +7405,7 @@ public Unstage unstage(List<DataType<?>> dtypes, Unstage.Options... options) {
    * @param options carries optional attributes values
    * @return a new instance of VarHandleOp
    */
-  public <T extends Tensor> VarHandleOp varHandleOp(DataType<T> dtype, Shape shape,
+  public <T extends TType> VarHandleOp varHandleOp(DataType<T> dtype, Shape shape,
       VarHandleOp.Options... options) {
     return VarHandleOp.create(scope, dtype, shape, options);
   }
@@ -7444,7 +7431,7 @@ public VarIsInitializedOp varIsInitializedOp(Operand<?> resource) {
    * @param options carries optional attributes values
    * @return a new instance of Variable
    */
-  public <T extends Tensor> Variable<T> variable(Operand<T> init, Variable.Options... options) {
+  public <T extends TType> Variable<T> variable(Operand<T> init, Variable.Options... options) {
     return Helpers.createVariableWithInit(scope, init, options);
   }
 
@@ -7461,7 +7448,7 @@ public <T extends Tensor> Variable<T> variable(Operand<T> init, Variable.Options
    * @param options carries optional attributes values
    * @return a new instance of Variable
    */
-  public <T extends Tensor> Variable<T> variable(Shape shape, DataType<T> dtype,
+  public <T extends TType> Variable<T> variable(Shape shape, DataType<T> dtype,
       Variable.Options... options) {
     return Variable.create(scope, shape, dtype, options);
   }
@@ -7501,8 +7488,7 @@ public VariableShape<TInt32> variableShape(Operand<?> input) {
    * @param outType
    * @return a new instance of VariableShape
    */
-  public <T extends Tensor & TNumber> VariableShape<T> variableShape(Operand<?> input,
-      DataType<T> outType) {
+  public <T extends TNumber> VariableShape<T> variableShape(Operand<?> input, DataType<T> outType) {
     return VariableShape.create(scope, input, outType);
   }
 
@@ -7571,7 +7557,7 @@ public <T extends Tensor & TNumber> VariableShape<T> variableShape(Operand<?> in
    * @param condition
    * @return a new instance of Where
    */
-  public <T extends Tensor> Where where(Operand<T> condition) {
+  public <T extends TType> Where where(Operand<T> condition) {
     return Where.create(scope, condition);
   }
 
@@ -7584,8 +7570,7 @@ public <T extends Tensor> Where where(Operand<T> condition) {
    * @return a constant tensor initialized with zeros
    * @throws IllegalArgumentException if the tensor type or shape cannot be initialized with zeros.
    */
-  public <T extends Tensor, U extends Tensor & TNumber> Zeros<T> zeros(Operand<U> dims,
-      DataType<T> type) {
+  public <T extends TType, U extends TNumber> Zeros<T> zeros(Operand<U> dims, DataType<T> type) {
     return Zeros.create(scope, dims, type);
   }
 
@@ -7596,7 +7581,7 @@ public <T extends Tensor, U extends Tensor & TNumber> Zeros<T> zeros(Operand<U>
    * @param x a tensor of type T.
    * @return a new instance of ZerosLike
    */
-  public <T extends Tensor> ZerosLike<T> zerosLike(Operand<T> x) {
+  public <T extends TType> ZerosLike<T> zerosLike(Operand<T> x) {
     return ZerosLike.create(scope, x);
   }
 
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/QuantizationOps.java b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/QuantizationOps.java
index be319493f11..aec0d667c65 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/QuantizationOps.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/QuantizationOps.java
@@ -19,7 +19,6 @@
 
 import org.tensorflow.DataType;
 import org.tensorflow.Operand;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.quantization.Dequantize;
 import org.tensorflow.op.quantization.FakeQuantWithMinMaxArgs;
 import org.tensorflow.op.quantization.FakeQuantWithMinMaxArgsGradient;
@@ -36,6 +35,7 @@
 import org.tensorflow.types.TFloat32;
 import org.tensorflow.types.TInt32;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * An API for building {@code quantization} operations as {@link Op Op}s
@@ -107,7 +107,7 @@ public final class QuantizationOps {
    * @param options carries optional attributes values
    * @return a new instance of Dequantize
    */
-  public <T extends Tensor> Dequantize<TFloat32> dequantize(Operand<T> input,
+  public <T extends TType> Dequantize<TFloat32> dequantize(Operand<T> input,
       Operand<TFloat32> minRange, Operand<TFloat32> maxRange, Dequantize.Options... options) {
     return Dequantize.create(scope, input, minRange, maxRange, options);
   }
@@ -172,7 +172,7 @@ public <T extends Tensor> Dequantize<TFloat32> dequantize(Operand<T> input,
    * @param options carries optional attributes values
    * @return a new instance of Dequantize
    */
-  public <U extends Tensor & TNumber, T extends Tensor> Dequantize<U> dequantize(Operand<T> input,
+  public <U extends TNumber, T extends TType> Dequantize<U> dequantize(Operand<T> input,
       Operand<TFloat32> minRange, Operand<TFloat32> maxRange, DataType<U> dtype,
       Dequantize.Options... options) {
     return Dequantize.create(scope, input, minRange, maxRange, dtype, options);
@@ -503,9 +503,8 @@ public FakeQuantWithMinMaxVarsPerChannelGradient fakeQuantWithMinMaxVarsPerChann
    * @param options carries optional attributes values
    * @return a new instance of Quantize
    */
-  public <T extends Tensor> Quantize<T> quantize(Operand<TFloat32> input,
-      Operand<TFloat32> minRange, Operand<TFloat32> maxRange, DataType<T> T,
-      Quantize.Options... options) {
+  public <T extends TType> Quantize<T> quantize(Operand<TFloat32> input, Operand<TFloat32> minRange,
+      Operand<TFloat32> maxRange, DataType<T> T, Quantize.Options... options) {
     return Quantize.create(scope, input, minRange, maxRange, T, options);
   }
 
@@ -523,8 +522,8 @@ public <T extends Tensor> Quantize<T> quantize(Operand<TFloat32> input,
    * @param options carries optional attributes values
    * @return a new instance of QuantizeAndDequantize
    */
-  public <T extends Tensor & TNumber> QuantizeAndDequantize<T> quantizeAndDequantize(
-      Operand<T> input, Operand<T> inputMin, Operand<T> inputMax, Operand<TInt32> numBits,
+  public <T extends TNumber> QuantizeAndDequantize<T> quantizeAndDequantize(Operand<T> input,
+      Operand<T> inputMin, Operand<T> inputMax, Operand<TInt32> numBits,
       QuantizeAndDequantize.Options... options) {
     return QuantizeAndDequantize.create(scope, input, inputMin, inputMax, numBits, options);
   }
@@ -562,7 +561,7 @@ public <T extends Tensor & TNumber> QuantizeAndDequantize<T> quantizeAndDequanti
    * @param outType The type of the output. Should be a lower bit depth than Tinput.
    * @return a new instance of QuantizeDownAndShrinkRange
    */
-  public <U extends Tensor, T extends Tensor> QuantizeDownAndShrinkRange<U> quantizeDownAndShrinkRange(
+  public <U extends TType, T extends TType> QuantizeDownAndShrinkRange<U> quantizeDownAndShrinkRange(
       Operand<T> input, Operand<TFloat32> inputMin, Operand<TFloat32> inputMax,
       DataType<U> outType) {
     return QuantizeDownAndShrinkRange.create(scope, input, inputMin, inputMax, outType);
@@ -580,7 +579,7 @@ public <U extends Tensor, T extends Tensor> QuantizeDownAndShrinkRange<U> quanti
    * @param inputMaxes The maximum scalar values for each of the input tensors.
    * @return a new instance of QuantizedConcat
    */
-  public <T extends Tensor> QuantizedConcat<T> quantizedConcat(Operand<TInt32> concatDim,
+  public <T extends TType> QuantizedConcat<T> quantizedConcat(Operand<TInt32> concatDim,
       Iterable<Operand<T>> values, Iterable<Operand<TFloat32>> inputMins,
       Iterable<Operand<TFloat32>> inputMaxes) {
     return QuantizedConcat.create(scope, concatDim, values, inputMins, inputMaxes);
@@ -599,7 +598,7 @@ public <T extends Tensor> QuantizedConcat<T> quantizedConcat(Operand<TInt32> con
    * @param inputMax The float value that the maximum quantized input value represents.
    * @return a new instance of RequantizationRange
    */
-  public <T extends Tensor> RequantizationRange requantizationRange(Operand<T> input,
+  public <T extends TType> RequantizationRange requantizationRange(Operand<T> input,
       Operand<TFloat32> inputMin, Operand<TFloat32> inputMax) {
     return RequantizationRange.create(scope, input, inputMin, inputMax);
   }
@@ -624,7 +623,7 @@ public <T extends Tensor> RequantizationRange requantizationRange(Operand<T> inp
    * @param outType The type of the output. Should be a lower bit depth than Tinput.
    * @return a new instance of Requantize
    */
-  public <U extends Tensor, T extends Tensor> Requantize<U> requantize(Operand<T> input,
+  public <U extends TType, T extends TType> Requantize<U> requantize(Operand<T> input,
       Operand<TFloat32> inputMin, Operand<TFloat32> inputMax, Operand<TFloat32> requestedOutputMin,
       Operand<TFloat32> requestedOutputMax, DataType<U> outType) {
     return Requantize.create(scope, input, inputMin, inputMax, requestedOutputMin, requestedOutputMax, outType);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/RandomOps.java b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/RandomOps.java
index cb9410ebb2a..071e77c7a70 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/RandomOps.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/RandomOps.java
@@ -19,7 +19,6 @@
 
 import org.tensorflow.DataType;
 import org.tensorflow.Operand;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.random.AllCandidateSampler;
 import org.tensorflow.op.random.LogUniformCandidateSampler;
 import org.tensorflow.op.random.Multinomial;
@@ -43,6 +42,7 @@
 import org.tensorflow.types.TInt32;
 import org.tensorflow.types.TInt64;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * An API for building {@code random} operations as {@link Op Op}s
@@ -124,7 +124,7 @@ public LogUniformCandidateSampler logUniformCandidateSampler(Operand<TInt64> tru
    * @param options carries optional attributes values
    * @return a new instance of Multinomial
    */
-  public <T extends Tensor & TNumber> Multinomial<TInt64> multinomial(Operand<T> logits,
+  public <T extends TNumber> Multinomial<TInt64> multinomial(Operand<T> logits,
       Operand<TInt32> numSamples, Multinomial.Options... options) {
     return Multinomial.create(scope, logits, numSamples, options);
   }
@@ -140,9 +140,8 @@ public <T extends Tensor & TNumber> Multinomial<TInt64> multinomial(Operand<T> l
    * @param options carries optional attributes values
    * @return a new instance of Multinomial
    */
-  public <U extends Tensor & TNumber, T extends Tensor & TNumber> Multinomial<U> multinomial(
-      Operand<T> logits, Operand<TInt32> numSamples, DataType<U> outputDtype,
-      Multinomial.Options... options) {
+  public <U extends TNumber, T extends TNumber> Multinomial<U> multinomial(Operand<T> logits,
+      Operand<TInt32> numSamples, DataType<U> outputDtype, Multinomial.Options... options) {
     return Multinomial.create(scope, logits, numSamples, outputDtype, options);
   }
 
@@ -162,7 +161,7 @@ public <U extends Tensor & TNumber, T extends Tensor & TNumber> Multinomial<U> m
    * @param options carries optional attributes values
    * @return a new instance of ParameterizedTruncatedNormal
    */
-  public <U extends Tensor & TNumber, T extends Tensor & TNumber> ParameterizedTruncatedNormal<U> parameterizedTruncatedNormal(
+  public <U extends TNumber, T extends TNumber> ParameterizedTruncatedNormal<U> parameterizedTruncatedNormal(
       Operand<T> shape, Operand<U> means, Operand<U> stdevs, Operand<U> minvals, Operand<U> maxvals,
       ParameterizedTruncatedNormal.Options... options) {
     return ParameterizedTruncatedNormal.create(scope, shape, means, stdevs, minvals, maxvals, options);
@@ -183,8 +182,8 @@ public <U extends Tensor & TNumber, T extends Tensor & TNumber> ParameterizedTru
    * @param options carries optional attributes values
    * @return a new instance of RandomGamma
    */
-  public <U extends Tensor & TNumber, T extends Tensor & TNumber> RandomGamma<U> randomGamma(
-      Operand<T> shape, Operand<U> alpha, RandomGamma.Options... options) {
+  public <U extends TNumber, T extends TNumber> RandomGamma<U> randomGamma(Operand<T> shape,
+      Operand<U> alpha, RandomGamma.Options... options) {
     return RandomGamma.create(scope, shape, alpha, options);
   }
 
@@ -209,7 +208,7 @@ public <U extends Tensor & TNumber, T extends Tensor & TNumber> RandomGamma<U> r
    * @param options carries optional attributes values
    * @return a new instance of RandomPoisson
    */
-  public <T extends Tensor & TNumber, U extends Tensor & TNumber> RandomPoisson<TInt64> randomPoisson(
+  public <T extends TNumber, U extends TNumber> RandomPoisson<TInt64> randomPoisson(
       Operand<T> shape, Operand<U> rate, RandomPoisson.Options... options) {
     return RandomPoisson.create(scope, shape, rate, options);
   }
@@ -236,7 +235,7 @@ public <T extends Tensor & TNumber, U extends Tensor & TNumber> RandomPoisson<TI
    * @param options carries optional attributes values
    * @return a new instance of RandomPoisson
    */
-  public <V extends Tensor & TNumber, T extends Tensor & TNumber, U extends Tensor & TNumber> RandomPoisson<V> randomPoisson(
+  public <V extends TNumber, T extends TNumber, U extends TNumber> RandomPoisson<V> randomPoisson(
       Operand<T> shape, Operand<U> rate, DataType<V> dtype, RandomPoisson.Options... options) {
     return RandomPoisson.create(scope, shape, rate, dtype, options);
   }
@@ -258,7 +257,7 @@ public <V extends Tensor & TNumber, T extends Tensor & TNumber, U extends Tensor
    * @param options carries optional attributes values
    * @return a new instance of RandomShuffle
    */
-  public <T extends Tensor> RandomShuffle<T> randomShuffle(Operand<T> value,
+  public <T extends TType> RandomShuffle<T> randomShuffle(Operand<T> value,
       RandomShuffle.Options... options) {
     return RandomShuffle.create(scope, value, options);
   }
@@ -274,7 +273,7 @@ public <T extends Tensor> RandomShuffle<T> randomShuffle(Operand<T> value,
    * @param options carries optional attributes values
    * @return a new instance of RandomStandardNormal
    */
-  public <U extends Tensor & TNumber, T extends Tensor & TNumber> RandomStandardNormal<U> randomStandardNormal(
+  public <U extends TNumber, T extends TNumber> RandomStandardNormal<U> randomStandardNormal(
       Operand<T> shape, DataType<U> dtype, RandomStandardNormal.Options... options) {
     return RandomStandardNormal.create(scope, shape, dtype, options);
   }
@@ -291,8 +290,8 @@ public <U extends Tensor & TNumber, T extends Tensor & TNumber> RandomStandardNo
    * @param options carries optional attributes values
    * @return a new instance of RandomUniform
    */
-  public <U extends Tensor & TNumber, T extends Tensor & TNumber> RandomUniform<U> randomUniform(
-      Operand<T> shape, DataType<U> dtype, RandomUniform.Options... options) {
+  public <U extends TNumber, T extends TNumber> RandomUniform<U> randomUniform(Operand<T> shape,
+      DataType<U> dtype, RandomUniform.Options... options) {
     return RandomUniform.create(scope, shape, dtype, options);
   }
 
@@ -314,7 +313,7 @@ public <U extends Tensor & TNumber, T extends Tensor & TNumber> RandomUniform<U>
    * @param options carries optional attributes values
    * @return a new instance of RandomUniformInt
    */
-  public <U extends Tensor & TNumber, T extends Tensor & TNumber> RandomUniformInt<U> randomUniformInt(
+  public <U extends TNumber, T extends TNumber> RandomUniformInt<U> randomUniformInt(
       Operand<T> shape, Operand<U> minval, Operand<U> maxval, RandomUniformInt.Options... options) {
     return RandomUniformInt.create(scope, shape, minval, maxval, options);
   }
@@ -340,7 +339,7 @@ public RecordInput recordInput(String filePattern, RecordInput.Options... option
    * @param probs
    * @return a new instance of StatefulRandomBinomial
    */
-  public <T extends Tensor & TNumber, U extends Tensor & TNumber> StatefulRandomBinomial<TInt64> statefulRandomBinomial(
+  public <T extends TNumber, U extends TNumber> StatefulRandomBinomial<TInt64> statefulRandomBinomial(
       Operand<?> resource, Operand<TInt64> algorithm, Operand<T> shape, Operand<U> counts,
       Operand<U> probs) {
     return StatefulRandomBinomial.create(scope, resource, algorithm, shape, counts, probs);
@@ -357,7 +356,7 @@ public <T extends Tensor & TNumber, U extends Tensor & TNumber> StatefulRandomBi
    * @param dtype
    * @return a new instance of StatefulRandomBinomial
    */
-  public <V extends Tensor & TNumber, T extends Tensor & TNumber, U extends Tensor & TNumber> StatefulRandomBinomial<V> statefulRandomBinomial(
+  public <V extends TNumber, T extends TNumber, U extends TNumber> StatefulRandomBinomial<V> statefulRandomBinomial(
       Operand<?> resource, Operand<TInt64> algorithm, Operand<T> shape, Operand<U> counts,
       Operand<U> probs, DataType<V> dtype) {
     return StatefulRandomBinomial.create(scope, resource, algorithm, shape, counts, probs, dtype);
@@ -374,7 +373,7 @@ public <V extends Tensor & TNumber, T extends Tensor & TNumber, U extends Tensor
    * @param shape The shape of the output tensor.
    * @return a new instance of StatefulStandardNormal
    */
-  public <T extends Tensor> StatefulStandardNormal<TFloat32> statefulStandardNormal(
+  public <T extends TType> StatefulStandardNormal<TFloat32> statefulStandardNormal(
       Operand<?> resource, Operand<TInt64> algorithm, Operand<T> shape) {
     return StatefulStandardNormal.create(scope, resource, algorithm, shape);
   }
@@ -391,7 +390,7 @@ public <T extends Tensor> StatefulStandardNormal<TFloat32> statefulStandardNorma
    * @param dtype The type of the output.
    * @return a new instance of StatefulStandardNormal
    */
-  public <U extends Tensor, T extends Tensor> StatefulStandardNormal<U> statefulStandardNormal(
+  public <U extends TType, T extends TType> StatefulStandardNormal<U> statefulStandardNormal(
       Operand<?> resource, Operand<TInt64> algorithm, Operand<T> shape, DataType<U> dtype) {
     return StatefulStandardNormal.create(scope, resource, algorithm, shape, dtype);
   }
@@ -406,7 +405,7 @@ public <U extends Tensor, T extends Tensor> StatefulStandardNormal<U> statefulSt
    * @param seed 2 seeds (shape [2]).
    * @return a new instance of StatelessMultinomial
    */
-  public <T extends Tensor & TNumber, U extends Tensor & TNumber> StatelessMultinomial<TInt64> statelessMultinomial(
+  public <T extends TNumber, U extends TNumber> StatelessMultinomial<TInt64> statelessMultinomial(
       Operand<T> logits, Operand<TInt32> numSamples, Operand<U> seed) {
     return StatelessMultinomial.create(scope, logits, numSamples, seed);
   }
@@ -422,7 +421,7 @@ public <T extends Tensor & TNumber, U extends Tensor & TNumber> StatelessMultino
    * @param outputDtype
    * @return a new instance of StatelessMultinomial
    */
-  public <V extends Tensor & TNumber, T extends Tensor & TNumber, U extends Tensor & TNumber> StatelessMultinomial<V> statelessMultinomial(
+  public <V extends TNumber, T extends TNumber, U extends TNumber> StatelessMultinomial<V> statelessMultinomial(
       Operand<T> logits, Operand<TInt32> numSamples, Operand<U> seed, DataType<V> outputDtype) {
     return StatelessMultinomial.create(scope, logits, numSamples, seed, outputDtype);
   }
@@ -439,7 +438,7 @@ public <V extends Tensor & TNumber, T extends Tensor & TNumber, U extends Tensor
    * @param seed 2 seeds (shape [2]).
    * @return a new instance of StatelessRandomNormal
    */
-  public <T extends Tensor & TNumber, U extends Tensor & TNumber> StatelessRandomNormal<TFloat32> statelessRandomNormal(
+  public <T extends TNumber, U extends TNumber> StatelessRandomNormal<TFloat32> statelessRandomNormal(
       Operand<T> shape, Operand<U> seed) {
     return StatelessRandomNormal.create(scope, shape, seed);
   }
@@ -457,7 +456,7 @@ public <T extends Tensor & TNumber, U extends Tensor & TNumber> StatelessRandomN
    * @param dtype The type of the output.
    * @return a new instance of StatelessRandomNormal
    */
-  public <V extends Tensor & TNumber, T extends Tensor & TNumber, U extends Tensor & TNumber> StatelessRandomNormal<V> statelessRandomNormal(
+  public <V extends TNumber, T extends TNumber, U extends TNumber> StatelessRandomNormal<V> statelessRandomNormal(
       Operand<T> shape, Operand<U> seed, DataType<V> dtype) {
     return StatelessRandomNormal.create(scope, shape, seed, dtype);
   }
@@ -475,7 +474,7 @@ public <V extends Tensor & TNumber, T extends Tensor & TNumber, U extends Tensor
    * @param seed 2 seeds (shape [2]).
    * @return a new instance of StatelessRandomUniform
    */
-  public <T extends Tensor & TNumber, U extends Tensor & TNumber> StatelessRandomUniform<TFloat32> statelessRandomUniform(
+  public <T extends TNumber, U extends TNumber> StatelessRandomUniform<TFloat32> statelessRandomUniform(
       Operand<T> shape, Operand<U> seed) {
     return StatelessRandomUniform.create(scope, shape, seed);
   }
@@ -494,7 +493,7 @@ public <T extends Tensor & TNumber, U extends Tensor & TNumber> StatelessRandomU
    * @param dtype The type of the output.
    * @return a new instance of StatelessRandomUniform
    */
-  public <V extends Tensor & TNumber, T extends Tensor & TNumber, U extends Tensor & TNumber> StatelessRandomUniform<V> statelessRandomUniform(
+  public <V extends TNumber, T extends TNumber, U extends TNumber> StatelessRandomUniform<V> statelessRandomUniform(
       Operand<T> shape, Operand<U> seed, DataType<V> dtype) {
     return StatelessRandomUniform.create(scope, shape, seed, dtype);
   }
@@ -513,7 +512,7 @@ public <V extends Tensor & TNumber, T extends Tensor & TNumber, U extends Tensor
    * @param seed 2 seeds (shape [2]).
    * @return a new instance of StatelessTruncatedNormal
    */
-  public <T extends Tensor & TNumber, U extends Tensor & TNumber> StatelessTruncatedNormal<TFloat32> statelessTruncatedNormal(
+  public <T extends TNumber, U extends TNumber> StatelessTruncatedNormal<TFloat32> statelessTruncatedNormal(
       Operand<T> shape, Operand<U> seed) {
     return StatelessTruncatedNormal.create(scope, shape, seed);
   }
@@ -533,7 +532,7 @@ public <T extends Tensor & TNumber, U extends Tensor & TNumber> StatelessTruncat
    * @param dtype The type of the output.
    * @return a new instance of StatelessTruncatedNormal
    */
-  public <V extends Tensor & TNumber, T extends Tensor & TNumber, U extends Tensor & TNumber> StatelessTruncatedNormal<V> statelessTruncatedNormal(
+  public <V extends TNumber, T extends TNumber, U extends TNumber> StatelessTruncatedNormal<V> statelessTruncatedNormal(
       Operand<T> shape, Operand<U> seed, DataType<V> dtype) {
     return StatelessTruncatedNormal.create(scope, shape, seed, dtype);
   }
@@ -551,8 +550,8 @@ public <V extends Tensor & TNumber, T extends Tensor & TNumber, U extends Tensor
    * @param options carries optional attributes values
    * @return a new instance of TruncatedNormal
    */
-  public <U extends Tensor & TNumber, T extends Tensor & TNumber> TruncatedNormal<U> truncatedNormal(
-      Operand<T> shape, DataType<U> dtype, TruncatedNormal.Options... options) {
+  public <U extends TNumber, T extends TNumber> TruncatedNormal<U> truncatedNormal(Operand<T> shape,
+      DataType<U> dtype, TruncatedNormal.Options... options) {
     return TruncatedNormal.create(scope, shape, dtype, options);
   }
 
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/ShapeOps.java b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/ShapeOps.java
index 077dd664c61..81c692571f1 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/ShapeOps.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/ShapeOps.java
@@ -19,12 +19,12 @@
 
 import org.tensorflow.DataType;
 import org.tensorflow.Operand;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.core.Shape;
 import org.tensorflow.op.core.Shapes;
 import org.tensorflow.types.TInt32;
 import org.tensorflow.types.TInt64;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * An API for building {@code shape} operations as {@link Op Op}s
@@ -78,8 +78,7 @@ public Operand<TInt32> append(Shape<TInt32> shape, int lastDimension) {
    *      operand representing a shape, followed by the dimensions of an operand representing a shape
    *      to append
    */
-  public <T extends Tensor & TNumber> Operand<T> append(Operand<T> shape,
-      Operand<T> shapeToAppend) {
+  public <T extends TNumber> Operand<T> append(Operand<T> shape, Operand<T> shapeToAppend) {
     return Shapes.append(scope, shape, shapeToAppend);
   }
 
@@ -91,7 +90,7 @@ public <T extends Tensor & TNumber> Operand<T> append(Operand<T> shape,
    * @param operand the operand to flatten
    * @return the reshaped operand
    */
-  public <T extends Tensor> Operand<T> flatten(Operand<T> operand) {
+  public <T extends TType> Operand<T> flatten(Operand<T> operand) {
     return Shapes.flatten(scope, operand);
   }
 
@@ -116,7 +115,7 @@ public Operand<TInt32> flatten(Shape<TInt32> shape) {
    * @param dType the shape datatype
    * @return the reshaped operand
    */
-  public <T extends Tensor, U extends Tensor & TNumber> Operand<T> flatten(Operand<T> operand,
+  public <T extends TType, U extends TNumber> Operand<T> flatten(Operand<T> operand,
       DataType<U> dType) {
     return Shapes.flatten(scope, operand, dType);
   }
@@ -130,7 +129,7 @@ public <T extends Tensor, U extends Tensor & TNumber> Operand<T> flatten(Operand
    * @param dType the shape datatype
    * @return the flattened shape
    */
-  public <U extends Tensor & TNumber> Operand<U> flatten(Shape<U> shape, DataType<U> dType) {
+  public <U extends TNumber> Operand<U> flatten(Shape<U> shape, DataType<U> dType) {
     return Shapes.flatten(scope, shape, dType);
   }
 
@@ -154,7 +153,7 @@ public Operand<TInt32> head(Shape<TInt32> shape) {
    * @param <U> the shape datatype.
    * @return a 1-dimensional Operand containing the Shape's first dimension
    */
-  public <U extends Tensor & TNumber> Operand<U> head(Shape<U> shape, DataType<U> dType) {
+  public <U extends TNumber> Operand<U> head(Shape<U> shape, DataType<U> dType) {
     return Shapes.head(scope, shape, dType);
   }
 
@@ -178,7 +177,7 @@ public Operand<TInt32> numDimensions(Shape<TInt32> shape) {
    * @param dType the shape datatype
    * @return the number of dimensions
    */
-  public <U extends Tensor & TNumber> Operand<U> numDimensions(Shape<U> shape, DataType<U> dType) {
+  public <U extends TNumber> Operand<U> numDimensions(Shape<U> shape, DataType<U> dType) {
     return Shapes.numDimensions(scope, shape, dType);
   }
 
@@ -222,8 +221,7 @@ public Operand<TInt32> prepend(Shape<TInt32> shape, int firstDimension) {
    *      operand representing the shape to prepend, followed by the dimensions of an operand
    *      representing the shape
    */
-  public <T extends Tensor & TNumber> Operand<T> prepend(Operand<T> shape,
-      Operand<T> shapeToPrepend) {
+  public <T extends TNumber> Operand<T> prepend(Operand<T> shape, Operand<T> shapeToPrepend) {
     return Shapes.prepend(scope, shape, shapeToPrepend);
   }
 
@@ -236,7 +234,7 @@ public <T extends Tensor & TNumber> Operand<T> prepend(Operand<T> shape,
    * @param axis the axis
    * @return the reshaped operand
    */
-  public <T extends Tensor> Operand<T> reduceDims(Operand<T> operand, Operand<TInt32> axis) {
+  public <T extends TType> Operand<T> reduceDims(Operand<T> operand, Operand<TInt32> axis) {
     return Shapes.reduceDims(scope, operand, axis);
   }
 
@@ -263,7 +261,7 @@ public Operand<TInt32> reduceDims(Shape<TInt32> shape, Operand<TInt32> axis) {
    * @param dType the shape datatype
    * @return the reshaped operand
    */
-  public <T extends Tensor, U extends Tensor & TNumber> Operand<T> reduceDims(Operand<T> operand,
+  public <T extends TType, U extends TNumber> Operand<T> reduceDims(Operand<T> operand,
       Operand<U> axis, DataType<U> dType) {
     return Shapes.reduceDims(scope, operand, axis, dType);
   }
@@ -278,7 +276,7 @@ public <T extends Tensor, U extends Tensor & TNumber> Operand<T> reduceDims(Oper
    * @param dType the shape datatype
    * @return the reduced shape
    */
-  public <U extends Tensor & TNumber> Operand<U> reduceDims(Shape<U> shape, Operand<U> axis,
+  public <U extends TNumber> Operand<U> reduceDims(Shape<U> shape, Operand<U> axis,
       DataType<U> dType) {
     return Shapes.reduceDims(scope, shape, axis, dType);
   }
@@ -302,7 +300,7 @@ public Operand<TInt32> size(Shape<TInt32> shape) {
    * @param dim the dimension
    * @return the size of the specified dimension
    */
-  public <T extends Tensor> Operand<TInt32> size(Operand<T> input, Operand<TInt32> dim) {
+  public <T extends TType> Operand<TInt32> size(Operand<T> input, Operand<TInt32> dim) {
     return Shapes.size(scope, input, dim);
   }
 
@@ -315,7 +313,7 @@ public <T extends Tensor> Operand<TInt32> size(Operand<T> input, Operand<TInt32>
    * @param dType the shape datatype
    * @return the size
    */
-  public <U extends Tensor & TNumber> Operand<U> size(Shape<U> shape, DataType<U> dType) {
+  public <U extends TNumber> Operand<U> size(Shape<U> shape, DataType<U> dType) {
     return Shapes.size(scope, shape, dType);
   }
 
@@ -341,8 +339,8 @@ public Operand<TInt32> size(Shape<TInt32> shape, Operand<TInt32> dim) {
    * @param dType the shape datatype
    * @return the size of the specified dimension
    */
-  public <T extends Tensor, U extends Tensor & TNumber> Operand<U> size(Operand<T> input,
-      Operand<U> dim, DataType<U> dType) {
+  public <T extends TType, U extends TNumber> Operand<U> size(Operand<T> input, Operand<U> dim,
+      DataType<U> dType) {
     return Shapes.size(scope, input, dim, dType);
   }
 
@@ -356,8 +354,7 @@ public <T extends Tensor, U extends Tensor & TNumber> Operand<U> size(Operand<T>
    * @param dType the shape datatype
    * @return the size of the specified dimension
    */
-  public <U extends Tensor & TNumber> Operand<U> size(Shape<U> shape, Operand<U> dim,
-      DataType<U> dType) {
+  public <U extends TNumber> Operand<U> size(Shape<U> shape, Operand<U> dim, DataType<U> dType) {
     return Shapes.size(scope, shape, dim, dType);
   }
 
@@ -381,7 +378,7 @@ public Operand<TInt32> squeeze(Shape<TInt32> shape) {
    * @param dType the shape datatype.
    * @return the squeezed shape
    */
-  public <U extends Tensor & TNumber> Operand<U> squeeze(Shape<U> shape, DataType<U> dType) {
+  public <U extends TNumber> Operand<U> squeeze(Shape<U> shape, DataType<U> dType) {
     return Shapes.squeeze(scope, shape, dType);
   }
 
@@ -409,7 +406,7 @@ public Operand<TInt32> tail(Shape<TInt32> shape) {
    * @return a 1-dimensional Operand that contains the dimension matching the last dimension of the
    *      Shape
    */
-  public <U extends Tensor & TNumber> Operand<U> tail(Shape<U> shape, DataType<U> dType) {
+  public <U extends TNumber> Operand<U> tail(Shape<U> shape, DataType<U> dType) {
     return Shapes.tail(scope, shape, dType);
   }
 
@@ -439,8 +436,7 @@ public Operand<TInt32> take(Shape<TInt32> shape, Operand<TInt32> n) {
    * @return a 1-dimensional operand with the dimensions matching * the first n dimensions of the
    *      shape
    */
-  public <U extends Tensor & TNumber> Operand<U> take(Shape<U> shape, Operand<U> n,
-      DataType<U> dType) {
+  public <U extends TNumber> Operand<U> take(Shape<U> shape, Operand<U> n, DataType<U> dType) {
     return Shapes.take(scope, shape, n, dType);
   }
 
@@ -454,8 +450,7 @@ public <U extends Tensor & TNumber> Operand<U> take(Shape<U> shape, Operand<U> n
    * @return a 1-dimensional operand containing the dimensions matching the last n dimensions of the
    *      shape
    */
-  public <U extends Tensor & TNumber> Operand<TInt32> takeLast(Shape<TInt32> shape,
-      Operand<TInt32> n) {
+  public <U extends TNumber> Operand<TInt32> takeLast(Shape<TInt32> shape, Operand<TInt32> n) {
     return Shapes.takeLast(scope, shape, n);
   }
 
@@ -471,8 +466,7 @@ public <U extends Tensor & TNumber> Operand<TInt32> takeLast(Shape<TInt32> shape
    * @return a 1-dimensional operand containing the dimensions matching the last n dimensions of the
    *      shape
    */
-  public <U extends Tensor & TNumber> Operand<U> takeLast(Shape<U> shape, Operand<U> n,
-      DataType<U> dType) {
+  public <U extends TNumber> Operand<U> takeLast(Shape<U> shape, Operand<U> n, DataType<U> dType) {
     return Shapes.takeLast(scope, shape, n, dType);
   }
 }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/SignalOps.java b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/SignalOps.java
index c15b82120f0..f4ec7bdb48d 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/SignalOps.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/SignalOps.java
@@ -19,7 +19,6 @@
 
 import org.tensorflow.DataType;
 import org.tensorflow.Operand;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.signal.BatchFft;
 import org.tensorflow.op.signal.BatchFft2d;
 import org.tensorflow.op.signal.BatchFft3d;
@@ -41,6 +40,7 @@
 import org.tensorflow.types.TFloat32;
 import org.tensorflow.types.TInt32;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * An API for building {@code signal} operations as {@link Op Op}s
@@ -118,7 +118,7 @@ public BatchIfft3d batchIfft3d(Operand<?> input) {
    * @param input A complex tensor.
    * @return a new instance of Fft
    */
-  public <T extends Tensor> Fft<T> fft(Operand<T> input) {
+  public <T extends TType> Fft<T> fft(Operand<T> input) {
     return Fft.create(scope, input);
   }
 
@@ -132,7 +132,7 @@ public <T extends Tensor> Fft<T> fft(Operand<T> input) {
    * @param input A complex tensor.
    * @return a new instance of Fft2d
    */
-  public <T extends Tensor> Fft2d<T> fft2d(Operand<T> input) {
+  public <T extends TType> Fft2d<T> fft2d(Operand<T> input) {
     return Fft2d.create(scope, input);
   }
 
@@ -146,7 +146,7 @@ public <T extends Tensor> Fft2d<T> fft2d(Operand<T> input) {
    * @param input A complex tensor.
    * @return a new instance of Fft3d
    */
-  public <T extends Tensor> Fft3d<T> fft3d(Operand<T> input) {
+  public <T extends TType> Fft3d<T> fft3d(Operand<T> input) {
     return Fft3d.create(scope, input);
   }
 
@@ -160,7 +160,7 @@ public <T extends Tensor> Fft3d<T> fft3d(Operand<T> input) {
    * @param input A complex tensor.
    * @return a new instance of Ifft
    */
-  public <T extends Tensor> Ifft<T> ifft(Operand<T> input) {
+  public <T extends TType> Ifft<T> ifft(Operand<T> input) {
     return Ifft.create(scope, input);
   }
 
@@ -174,7 +174,7 @@ public <T extends Tensor> Ifft<T> ifft(Operand<T> input) {
    * @param input A complex tensor.
    * @return a new instance of Ifft2d
    */
-  public <T extends Tensor> Ifft2d<T> ifft2d(Operand<T> input) {
+  public <T extends TType> Ifft2d<T> ifft2d(Operand<T> input) {
     return Ifft2d.create(scope, input);
   }
 
@@ -188,7 +188,7 @@ public <T extends Tensor> Ifft2d<T> ifft2d(Operand<T> input) {
    * @param input A complex tensor.
    * @return a new instance of Ifft3d
    */
-  public <T extends Tensor> Ifft3d<T> ifft3d(Operand<T> input) {
+  public <T extends TType> Ifft3d<T> ifft3d(Operand<T> input) {
     return Ifft3d.create(scope, input);
   }
 
@@ -214,7 +214,7 @@ public <T extends Tensor> Ifft3d<T> ifft3d(Operand<T> input) {
    * @param fftLength An int32 tensor of shape [1]. The FFT length.
    * @return a new instance of Irfft
    */
-  public <T extends Tensor> Irfft<TFloat32> irfft(Operand<T> input, Operand<TInt32> fftLength) {
+  public <T extends TType> Irfft<TFloat32> irfft(Operand<T> input, Operand<TInt32> fftLength) {
     return Irfft.create(scope, input, fftLength);
   }
 
@@ -241,7 +241,7 @@ public <T extends Tensor> Irfft<TFloat32> irfft(Operand<T> input, Operand<TInt32
    * @param Treal
    * @return a new instance of Irfft
    */
-  public <U extends Tensor & TNumber, T extends Tensor> Irfft<U> irfft(Operand<T> input,
+  public <U extends TNumber, T extends TType> Irfft<U> irfft(Operand<T> input,
       Operand<TInt32> fftLength, DataType<U> Treal) {
     return Irfft.create(scope, input, fftLength, Treal);
   }
@@ -269,7 +269,7 @@ public <U extends Tensor & TNumber, T extends Tensor> Irfft<U> irfft(Operand<T>
    * @param fftLength An int32 tensor of shape [2]. The FFT length for each dimension.
    * @return a new instance of Irfft2d
    */
-  public <T extends Tensor> Irfft2d<TFloat32> irfft2d(Operand<T> input, Operand<TInt32> fftLength) {
+  public <T extends TType> Irfft2d<TFloat32> irfft2d(Operand<T> input, Operand<TInt32> fftLength) {
     return Irfft2d.create(scope, input, fftLength);
   }
 
@@ -297,7 +297,7 @@ public <T extends Tensor> Irfft2d<TFloat32> irfft2d(Operand<T> input, Operand<TI
    * @param Treal
    * @return a new instance of Irfft2d
    */
-  public <U extends Tensor & TNumber, T extends Tensor> Irfft2d<U> irfft2d(Operand<T> input,
+  public <U extends TNumber, T extends TType> Irfft2d<U> irfft2d(Operand<T> input,
       Operand<TInt32> fftLength, DataType<U> Treal) {
     return Irfft2d.create(scope, input, fftLength, Treal);
   }
@@ -325,7 +325,7 @@ public <U extends Tensor & TNumber, T extends Tensor> Irfft2d<U> irfft2d(Operand
    * @param fftLength An int32 tensor of shape [3]. The FFT length for each dimension.
    * @return a new instance of Irfft3d
    */
-  public <T extends Tensor> Irfft3d<TFloat32> irfft3d(Operand<T> input, Operand<TInt32> fftLength) {
+  public <T extends TType> Irfft3d<TFloat32> irfft3d(Operand<T> input, Operand<TInt32> fftLength) {
     return Irfft3d.create(scope, input, fftLength);
   }
 
@@ -353,7 +353,7 @@ public <T extends Tensor> Irfft3d<TFloat32> irfft3d(Operand<T> input, Operand<TI
    * @param Treal
    * @return a new instance of Irfft3d
    */
-  public <U extends Tensor & TNumber, T extends Tensor> Irfft3d<U> irfft3d(Operand<T> input,
+  public <U extends TNumber, T extends TType> Irfft3d<U> irfft3d(Operand<T> input,
       Operand<TInt32> fftLength, DataType<U> Treal) {
     return Irfft3d.create(scope, input, fftLength, Treal);
   }
@@ -378,7 +378,7 @@ public <U extends Tensor & TNumber, T extends Tensor> Irfft3d<U> irfft3d(Operand
    * @param Tcomplex
    * @return a new instance of Rfft
    */
-  public <U extends Tensor, T extends Tensor & TNumber> Rfft<U> rfft(Operand<T> input,
+  public <U extends TType, T extends TNumber> Rfft<U> rfft(Operand<T> input,
       Operand<TInt32> fftLength, DataType<U> Tcomplex) {
     return Rfft.create(scope, input, fftLength, Tcomplex);
   }
@@ -404,7 +404,7 @@ public <U extends Tensor, T extends Tensor & TNumber> Rfft<U> rfft(Operand<T> in
    * @param Tcomplex
    * @return a new instance of Rfft2d
    */
-  public <U extends Tensor, T extends Tensor & TNumber> Rfft2d<U> rfft2d(Operand<T> input,
+  public <U extends TType, T extends TNumber> Rfft2d<U> rfft2d(Operand<T> input,
       Operand<TInt32> fftLength, DataType<U> Tcomplex) {
     return Rfft2d.create(scope, input, fftLength, Tcomplex);
   }
@@ -430,7 +430,7 @@ public <U extends Tensor, T extends Tensor & TNumber> Rfft2d<U> rfft2d(Operand<T
    * @param Tcomplex
    * @return a new instance of Rfft3d
    */
-  public <U extends Tensor, T extends Tensor & TNumber> Rfft3d<U> rfft3d(Operand<T> input,
+  public <U extends TType, T extends TNumber> Rfft3d<U> rfft3d(Operand<T> input,
       Operand<TInt32> fftLength, DataType<U> Tcomplex) {
     return Rfft3d.create(scope, input, fftLength, Tcomplex);
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/SparseOps.java b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/SparseOps.java
index d8ba9f45ef2..6e26fec2275 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/SparseOps.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/SparseOps.java
@@ -19,7 +19,6 @@
 
 import org.tensorflow.DataType;
 import org.tensorflow.Operand;
-import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.sparse.AddManySparseToTensorsMap;
 import org.tensorflow.op.sparse.AddSparseToTensorsMap;
@@ -68,6 +67,7 @@
 import org.tensorflow.types.TInt64;
 import org.tensorflow.types.TString;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * An API for building {@code sparse} operations as {@link Op Op}s
@@ -114,7 +114,7 @@ public final class SparseOps {
    * @param options carries optional attributes values
    * @return a new instance of AddManySparseToTensorsMap
    */
-  public <T extends Tensor> AddManySparseToTensorsMap addManySparseToTensorsMap(
+  public <T extends TType> AddManySparseToTensorsMap addManySparseToTensorsMap(
       Operand<TInt64> sparseIndices, Operand<T> sparseValues, Operand<TInt64> sparseShape,
       AddManySparseToTensorsMap.Options... options) {
     return AddManySparseToTensorsMap.create(scope, sparseIndices, sparseValues, sparseShape, options);
@@ -144,7 +144,7 @@ public <T extends Tensor> AddManySparseToTensorsMap addManySparseToTensorsMap(
    * @param options carries optional attributes values
    * @return a new instance of AddSparseToTensorsMap
    */
-  public <T extends Tensor> AddSparseToTensorsMap addSparseToTensorsMap(
+  public <T extends TType> AddSparseToTensorsMap addSparseToTensorsMap(
       Operand<TInt64> sparseIndices, Operand<T> sparseValues, Operand<TInt64> sparseShape,
       AddSparseToTensorsMap.Options... options) {
     return AddSparseToTensorsMap.create(scope, sparseIndices, sparseValues, sparseShape, options);
@@ -170,7 +170,7 @@ public <T extends Tensor> AddSparseToTensorsMap addSparseToTensorsMap(
    * @param options carries optional attributes values
    * @return a new instance of DenseToDenseSetOperation
    */
-  public <T extends Tensor> DenseToDenseSetOperation<T> denseToDenseSetOperation(Operand<T> set1,
+  public <T extends TType> DenseToDenseSetOperation<T> denseToDenseSetOperation(Operand<T> set1,
       Operand<T> set2, String setOperation, DenseToDenseSetOperation.Options... options) {
     return DenseToDenseSetOperation.create(scope, set1, set2, setOperation, options);
   }
@@ -208,7 +208,7 @@ public <T extends Tensor> DenseToDenseSetOperation<T> denseToDenseSetOperation(O
    * @param options carries optional attributes values
    * @return a new instance of DenseToSparseSetOperation
    */
-  public <T extends Tensor> DenseToSparseSetOperation<T> denseToSparseSetOperation(Operand<T> set1,
+  public <T extends TType> DenseToSparseSetOperation<T> denseToSparseSetOperation(Operand<T> set1,
       Operand<TInt64> set2Indices, Operand<T> set2Values, Operand<TInt64> set2Shape,
       String setOperation, DenseToSparseSetOperation.Options... options) {
     return DenseToSparseSetOperation.create(scope, set1, set2Indices, set2Values, set2Shape, setOperation, options);
@@ -265,7 +265,7 @@ public <T extends Tensor> DenseToSparseSetOperation<T> denseToSparseSetOperation
    * @param dtype The `dtype` of the serialized `SparseTensor` objects.
    * @return a new instance of DeserializeSparse
    */
-  public <U extends Tensor, T extends Tensor> DeserializeSparse<U> deserializeSparse(
+  public <U extends TType, T extends TType> DeserializeSparse<U> deserializeSparse(
       Operand<T> serializedSparse, DataType<U> dtype) {
     return DeserializeSparse.create(scope, serializedSparse, dtype);
   }
@@ -288,7 +288,7 @@ public <U extends Tensor, T extends Tensor> DeserializeSparse<U> deserializeSpar
    *  case the input is ignored during validation.
    * @return a new instance of SparseAccumulatorApplyGradient
    */
-  public <T extends Tensor> SparseAccumulatorApplyGradient sparseAccumulatorApplyGradient(
+  public <T extends TType> SparseAccumulatorApplyGradient sparseAccumulatorApplyGradient(
       Operand<TString> handle, Operand<TInt64> localStep, Operand<TInt64> gradientIndices,
       Operand<T> gradientValues, Operand<TInt64> gradientShape, Boolean hasKnownShape) {
     return SparseAccumulatorApplyGradient.create(scope, handle, localStep, gradientIndices, gradientValues, gradientShape, hasKnownShape);
@@ -311,7 +311,7 @@ public <T extends Tensor> SparseAccumulatorApplyGradient sparseAccumulatorApplyG
    *  of the accumulator.
    * @return a new instance of SparseAccumulatorTakeGradient
    */
-  public <T extends Tensor> SparseAccumulatorTakeGradient<T> sparseAccumulatorTakeGradient(
+  public <T extends TType> SparseAccumulatorTakeGradient<T> sparseAccumulatorTakeGradient(
       Operand<TString> handle, Operand<TInt32> numRequired, DataType<T> dtype) {
     return SparseAccumulatorTakeGradient.create(scope, handle, numRequired, dtype);
   }
@@ -344,9 +344,9 @@ public <T extends Tensor> SparseAccumulatorTakeGradient<T> sparseAccumulatorTake
    *  pair takes space.
    * @return a new instance of SparseAdd
    */
-  public <T extends Tensor, U extends Tensor & TNumber> SparseAdd<T> sparseAdd(
-      Operand<TInt64> aIndices, Operand<T> aValues, Operand<TInt64> aShape,
-      Operand<TInt64> bIndices, Operand<T> bValues, Operand<TInt64> bShape, Operand<U> thresh) {
+  public <T extends TType, U extends TNumber> SparseAdd<T> sparseAdd(Operand<TInt64> aIndices,
+      Operand<T> aValues, Operand<TInt64> aShape, Operand<TInt64> bIndices, Operand<T> bValues,
+      Operand<TInt64> bShape, Operand<U> thresh) {
     return SparseAdd.create(scope, aIndices, aValues, aShape, bIndices, bValues, bShape, thresh);
   }
 
@@ -367,7 +367,7 @@ public <T extends Tensor, U extends Tensor & TNumber> SparseAdd<T> sparseAdd(
    *  `[nnz(sum), ndims]`.
    * @return a new instance of SparseAddGrad
    */
-  public <T extends Tensor> SparseAddGrad<T> sparseAddGrad(Operand<T> backpropValGrad,
+  public <T extends TType> SparseAddGrad<T> sparseAddGrad(Operand<T> backpropValGrad,
       Operand<TInt64> aIndices, Operand<TInt64> bIndices, Operand<TInt64> sumIndices) {
     return SparseAddGrad.create(scope, backpropValGrad, aIndices, bIndices, sumIndices);
   }
@@ -425,7 +425,7 @@ public <T extends Tensor> SparseAddGrad<T> sparseAddGrad(Operand<T> backpropValG
    *  where rank is the number of dimensions in each input `SparseTensor`.
    * @return a new instance of SparseConcat
    */
-  public <T extends Tensor> SparseConcat<T> sparseConcat(Iterable<Operand<TInt64>> indices,
+  public <T extends TType> SparseConcat<T> sparseConcat(Iterable<Operand<TInt64>> indices,
       Iterable<Operand<T>> values, Iterable<Operand<TInt64>> shapes, Long concatDim) {
     return SparseConcat.create(scope, indices, values, shapes, concatDim);
   }
@@ -445,7 +445,7 @@ public <T extends Tensor> SparseConcat<T> sparseConcat(Iterable<Operand<TInt64>>
    * @param options carries optional attributes values
    * @return a new instance of SparseConditionalAccumulator
    */
-  public <T extends Tensor> SparseConditionalAccumulator sparseConditionalAccumulator(
+  public <T extends TType> SparseConditionalAccumulator sparseConditionalAccumulator(
       DataType<T> dtype, Shape shape, SparseConditionalAccumulator.Options... options) {
     return SparseConditionalAccumulator.create(scope, dtype, shape, options);
   }
@@ -505,7 +505,7 @@ public <T extends Tensor> SparseConditionalAccumulator sparseConditionalAccumula
    * @param internalType
    * @return a new instance of SparseCross
    */
-  public <T extends Tensor, U extends Tensor> SparseCross<T> sparseCross(
+  public <T extends TType, U extends TType> SparseCross<T> sparseCross(
       Iterable<Operand<TInt64>> indices, Iterable<Operand<?>> values,
       Iterable<Operand<TInt64>> shapes, Iterable<Operand<?>> denseInputs, Boolean hashedOutput,
       Long numBuckets, Long hashKey, DataType<T> outType, DataType<U> internalType) {
@@ -532,7 +532,7 @@ public <T extends Tensor, U extends Tensor> SparseCross<T> sparseCross(
    * @param dense `R`-D.  The dense Tensor operand.
    * @return a new instance of SparseDenseCwiseAdd
    */
-  public <T extends Tensor> SparseDenseCwiseAdd<T> sparseDenseCwiseAdd(Operand<TInt64> spIndices,
+  public <T extends TType> SparseDenseCwiseAdd<T> sparseDenseCwiseAdd(Operand<TInt64> spIndices,
       Operand<T> spValues, Operand<TInt64> spShape, Operand<T> dense) {
     return SparseDenseCwiseAdd.create(scope, spIndices, spValues, spShape, dense);
   }
@@ -551,7 +551,7 @@ public <T extends Tensor> SparseDenseCwiseAdd<T> sparseDenseCwiseAdd(Operand<TIn
    * @param dense `R`-D.  The dense Tensor operand.
    * @return a new instance of SparseDenseCwiseDiv
    */
-  public <T extends Tensor> SparseDenseCwiseDiv<T> sparseDenseCwiseDiv(Operand<TInt64> spIndices,
+  public <T extends TType> SparseDenseCwiseDiv<T> sparseDenseCwiseDiv(Operand<TInt64> spIndices,
       Operand<T> spValues, Operand<TInt64> spShape, Operand<T> dense) {
     return SparseDenseCwiseDiv.create(scope, spIndices, spValues, spShape, dense);
   }
@@ -574,7 +574,7 @@ public <T extends Tensor> SparseDenseCwiseDiv<T> sparseDenseCwiseDiv(Operand<TIn
    * @param dense `R`-D.  The dense Tensor operand.
    * @return a new instance of SparseDenseCwiseMul
    */
-  public <T extends Tensor> SparseDenseCwiseMul<T> sparseDenseCwiseMul(Operand<TInt64> spIndices,
+  public <T extends TType> SparseDenseCwiseMul<T> sparseDenseCwiseMul(Operand<TInt64> spIndices,
       Operand<T> spValues, Operand<TInt64> spShape, Operand<T> dense) {
     return SparseDenseCwiseMul.create(scope, spIndices, spValues, spShape, dense);
   }
@@ -628,7 +628,7 @@ public <T extends Tensor> SparseDenseCwiseMul<T> sparseDenseCwiseMul(Operand<TIn
    *  output indices: 2-D. the indices of the filled sparse tensor.
    * @return a new instance of SparseFillEmptyRows
    */
-  public <T extends Tensor> SparseFillEmptyRows<T> sparseFillEmptyRows(Operand<TInt64> indices,
+  public <T extends TType> SparseFillEmptyRows<T> sparseFillEmptyRows(Operand<TInt64> indices,
       Operand<T> values, Operand<TInt64> denseShape, Operand<T> defaultValue) {
     return SparseFillEmptyRows.create(scope, indices, values, denseShape, defaultValue);
   }
@@ -650,7 +650,7 @@ public <T extends Tensor> SparseFillEmptyRows<T> sparseFillEmptyRows(Operand<TIn
    * @param gradValues 1-D.  The gradients from backprop.
    * @return a new instance of SparseFillEmptyRowsGrad
    */
-  public <T extends Tensor> SparseFillEmptyRowsGrad<T> sparseFillEmptyRowsGrad(
+  public <T extends TType> SparseFillEmptyRowsGrad<T> sparseFillEmptyRowsGrad(
       Operand<TInt64> reverseIndexMap, Operand<T> gradValues) {
     return SparseFillEmptyRowsGrad.create(scope, reverseIndexMap, gradValues);
   }
@@ -673,8 +673,8 @@ public <T extends Tensor> SparseFillEmptyRowsGrad<T> sparseFillEmptyRowsGrad(
    * @param options carries optional attributes values
    * @return a new instance of SparseMatMul
    */
-  public <T extends Tensor & TNumber, U extends Tensor & TNumber> SparseMatMul sparseMatMul(
-      Operand<T> a, Operand<U> b, SparseMatMul.Options... options) {
+  public <T extends TNumber, U extends TNumber> SparseMatMul sparseMatMul(Operand<T> a,
+      Operand<U> b, SparseMatMul.Options... options) {
     return SparseMatMul.create(scope, a, b, options);
   }
 
@@ -703,9 +703,9 @@ public <T extends Tensor & TNumber, U extends Tensor & TNumber> SparseMatMul spa
    * @param options carries optional attributes values
    * @return a new instance of SparseReduceMax
    */
-  public <T extends Tensor & TNumber> SparseReduceMax<T> sparseReduceMax(
-      Operand<TInt64> inputIndices, Operand<T> inputValues, Operand<TInt64> inputShape,
-      Operand<TInt32> reductionAxes, SparseReduceMax.Options... options) {
+  public <T extends TNumber> SparseReduceMax<T> sparseReduceMax(Operand<TInt64> inputIndices,
+      Operand<T> inputValues, Operand<TInt64> inputShape, Operand<TInt32> reductionAxes,
+      SparseReduceMax.Options... options) {
     return SparseReduceMax.create(scope, inputIndices, inputValues, inputShape, reductionAxes, options);
   }
 
@@ -734,7 +734,7 @@ public <T extends Tensor & TNumber> SparseReduceMax<T> sparseReduceMax(
    * @param options carries optional attributes values
    * @return a new instance of SparseReduceMaxSparse
    */
-  public <T extends Tensor & TNumber> SparseReduceMaxSparse<T> sparseReduceMaxSparse(
+  public <T extends TNumber> SparseReduceMaxSparse<T> sparseReduceMaxSparse(
       Operand<TInt64> inputIndices, Operand<T> inputValues, Operand<TInt64> inputShape,
       Operand<TInt32> reductionAxes, SparseReduceMaxSparse.Options... options) {
     return SparseReduceMaxSparse.create(scope, inputIndices, inputValues, inputShape, reductionAxes, options);
@@ -765,7 +765,7 @@ public <T extends Tensor & TNumber> SparseReduceMaxSparse<T> sparseReduceMaxSpar
    * @param options carries optional attributes values
    * @return a new instance of SparseReduceSum
    */
-  public <T extends Tensor> SparseReduceSum<T> sparseReduceSum(Operand<TInt64> inputIndices,
+  public <T extends TType> SparseReduceSum<T> sparseReduceSum(Operand<TInt64> inputIndices,
       Operand<T> inputValues, Operand<TInt64> inputShape, Operand<TInt32> reductionAxes,
       SparseReduceSum.Options... options) {
     return SparseReduceSum.create(scope, inputIndices, inputValues, inputShape, reductionAxes, options);
@@ -796,7 +796,7 @@ public <T extends Tensor> SparseReduceSum<T> sparseReduceSum(Operand<TInt64> inp
    * @param options carries optional attributes values
    * @return a new instance of SparseReduceSumSparse
    */
-  public <T extends Tensor> SparseReduceSumSparse<T> sparseReduceSumSparse(
+  public <T extends TType> SparseReduceSumSparse<T> sparseReduceSumSparse(
       Operand<TInt64> inputIndices, Operand<T> inputValues, Operand<TInt64> inputShape,
       Operand<TInt32> reductionAxes, SparseReduceSumSparse.Options... options) {
     return SparseReduceSumSparse.create(scope, inputIndices, inputValues, inputShape, reductionAxes, options);
@@ -821,7 +821,7 @@ public <T extends Tensor> SparseReduceSumSparse<T> sparseReduceSumSparse(
    * @param inputShape 1-D.  Shape of the input SparseTensor.
    * @return a new instance of SparseReorder
    */
-  public <T extends Tensor> SparseReorder<T> sparseReorder(Operand<TInt64> inputIndices,
+  public <T extends TType> SparseReorder<T> sparseReorder(Operand<TInt64> inputIndices,
       Operand<T> inputValues, Operand<TInt64> inputShape) {
     return SparseReorder.create(scope, inputIndices, inputValues, inputShape);
   }
@@ -870,7 +870,7 @@ public SparseReshape sparseReshape(Operand<TInt64> inputIndices, Operand<TInt64>
    * @param segmentIds A 1-D tensor. Values should be sorted and can be repeated.
    * @return a new instance of SparseSegmentMean
    */
-  public <T extends Tensor & TNumber, U extends Tensor & TNumber> SparseSegmentMean<T> sparseSegmentMean(
+  public <T extends TNumber, U extends TNumber> SparseSegmentMean<T> sparseSegmentMean(
       Operand<T> data, Operand<U> indices, Operand<TInt32> segmentIds) {
     return SparseSegmentMean.create(scope, data, indices, segmentIds);
   }
@@ -888,7 +888,7 @@ public <T extends Tensor & TNumber, U extends Tensor & TNumber> SparseSegmentMea
    * @param outputDim0 dimension 0 of "data" passed to SparseSegmentMean op.
    * @return a new instance of SparseSegmentMeanGrad
    */
-  public <T extends Tensor & TNumber, U extends Tensor & TNumber> SparseSegmentMeanGrad<T> sparseSegmentMeanGrad(
+  public <T extends TNumber, U extends TNumber> SparseSegmentMeanGrad<T> sparseSegmentMeanGrad(
       Operand<T> grad, Operand<U> indices, Operand<TInt32> segmentIds, Operand<TInt32> outputDim0) {
     return SparseSegmentMeanGrad.create(scope, grad, indices, segmentIds, outputDim0);
   }
@@ -910,7 +910,7 @@ public <T extends Tensor & TNumber, U extends Tensor & TNumber> SparseSegmentMea
    * @param numSegments Should equal the number of distinct segment IDs.
    * @return a new instance of SparseSegmentMeanWithNumSegments
    */
-  public <T extends Tensor & TNumber, U extends Tensor & TNumber, V extends Tensor & TNumber> SparseSegmentMeanWithNumSegments<T> sparseSegmentMeanWithNumSegments(
+  public <T extends TNumber, U extends TNumber, V extends TNumber> SparseSegmentMeanWithNumSegments<T> sparseSegmentMeanWithNumSegments(
       Operand<T> data, Operand<U> indices, Operand<TInt32> segmentIds, Operand<V> numSegments) {
     return SparseSegmentMeanWithNumSegments.create(scope, data, indices, segmentIds, numSegments);
   }
@@ -928,7 +928,7 @@ public <T extends Tensor & TNumber, U extends Tensor & TNumber, V extends Tensor
    * @param segmentIds A 1-D tensor. Values should be sorted and can be repeated.
    * @return a new instance of SparseSegmentSqrtN
    */
-  public <T extends Tensor & TNumber, U extends Tensor & TNumber> SparseSegmentSqrtN<T> sparseSegmentSqrtN(
+  public <T extends TNumber, U extends TNumber> SparseSegmentSqrtN<T> sparseSegmentSqrtN(
       Operand<T> data, Operand<U> indices, Operand<TInt32> segmentIds) {
     return SparseSegmentSqrtN.create(scope, data, indices, segmentIds);
   }
@@ -946,7 +946,7 @@ public <T extends Tensor & TNumber, U extends Tensor & TNumber> SparseSegmentSqr
    * @param outputDim0 dimension 0 of "data" passed to SparseSegmentSqrtN op.
    * @return a new instance of SparseSegmentSqrtNGrad
    */
-  public <T extends Tensor & TNumber, U extends Tensor & TNumber> SparseSegmentSqrtNGrad<T> sparseSegmentSqrtNGrad(
+  public <T extends TNumber, U extends TNumber> SparseSegmentSqrtNGrad<T> sparseSegmentSqrtNGrad(
       Operand<T> grad, Operand<U> indices, Operand<TInt32> segmentIds, Operand<TInt32> outputDim0) {
     return SparseSegmentSqrtNGrad.create(scope, grad, indices, segmentIds, outputDim0);
   }
@@ -970,7 +970,7 @@ public <T extends Tensor & TNumber, U extends Tensor & TNumber> SparseSegmentSqr
    * @param numSegments Should equal the number of distinct segment IDs.
    * @return a new instance of SparseSegmentSqrtNWithNumSegments
    */
-  public <T extends Tensor & TNumber, U extends Tensor & TNumber, V extends Tensor & TNumber> SparseSegmentSqrtNWithNumSegments<T> sparseSegmentSqrtNWithNumSegments(
+  public <T extends TNumber, U extends TNumber, V extends TNumber> SparseSegmentSqrtNWithNumSegments<T> sparseSegmentSqrtNWithNumSegments(
       Operand<T> data, Operand<U> indices, Operand<TInt32> segmentIds, Operand<V> numSegments) {
     return SparseSegmentSqrtNWithNumSegments.create(scope, data, indices, segmentIds, numSegments);
   }
@@ -1013,7 +1013,7 @@ public <T extends Tensor & TNumber, U extends Tensor & TNumber, V extends Tensor
    * @param segmentIds A 1-D tensor. Values should be sorted and can be repeated.
    * @return a new instance of SparseSegmentSum
    */
-  public <T extends Tensor & TNumber, U extends Tensor & TNumber> SparseSegmentSum<T> sparseSegmentSum(
+  public <T extends TNumber, U extends TNumber> SparseSegmentSum<T> sparseSegmentSum(
       Operand<T> data, Operand<U> indices, Operand<TInt32> segmentIds) {
     return SparseSegmentSum.create(scope, data, indices, segmentIds);
   }
@@ -1055,7 +1055,7 @@ public <T extends Tensor & TNumber, U extends Tensor & TNumber> SparseSegmentSum
    * @param numSegments Should equal the number of distinct segment IDs.
    * @return a new instance of SparseSegmentSumWithNumSegments
    */
-  public <T extends Tensor & TNumber, U extends Tensor & TNumber, V extends Tensor & TNumber> SparseSegmentSumWithNumSegments<T> sparseSegmentSumWithNumSegments(
+  public <T extends TNumber, U extends TNumber, V extends TNumber> SparseSegmentSumWithNumSegments<T> sparseSegmentSumWithNumSegments(
       Operand<T> data, Operand<U> indices, Operand<TInt32> segmentIds, Operand<V> numSegments) {
     return SparseSegmentSumWithNumSegments.create(scope, data, indices, segmentIds, numSegments);
   }
@@ -1089,7 +1089,7 @@ public <T extends Tensor & TNumber, U extends Tensor & TNumber, V extends Tensor
    *  sparse tensors.
    * @return a new instance of SparseSlice
    */
-  public <T extends Tensor> SparseSlice<T> sparseSlice(Operand<TInt64> indices, Operand<T> values,
+  public <T extends TType> SparseSlice<T> sparseSlice(Operand<TInt64> indices, Operand<T> values,
       Operand<TInt64> shape, Operand<TInt64> start, Operand<TInt64> size) {
     return SparseSlice.create(scope, indices, values, shape, start, size);
   }
@@ -1109,7 +1109,7 @@ public <T extends Tensor> SparseSlice<T> sparseSlice(Operand<TInt64> indices, Op
    * @param outputIndices 2-D.  The `indices` of the sliced `SparseTensor`.
    * @return a new instance of SparseSliceGrad
    */
-  public <T extends Tensor> SparseSliceGrad<T> sparseSliceGrad(Operand<T> backpropValGrad,
+  public <T extends TType> SparseSliceGrad<T> sparseSliceGrad(Operand<T> backpropValGrad,
       Operand<TInt64> inputIndices, Operand<TInt64> inputStart, Operand<TInt64> outputIndices) {
     return SparseSliceGrad.create(scope, backpropValGrad, inputIndices, inputStart, outputIndices);
   }
@@ -1140,7 +1140,7 @@ public <T extends Tensor> SparseSliceGrad<T> sparseSliceGrad(Operand<T> backprop
    * @param spShape 1-D.  Shape of the input SparseTensor.
    * @return a new instance of SparseSoftmax
    */
-  public <T extends Tensor & TNumber> SparseSoftmax<T> sparseSoftmax(Operand<TInt64> spIndices,
+  public <T extends TNumber> SparseSoftmax<T> sparseSoftmax(Operand<TInt64> spIndices,
       Operand<T> spValues, Operand<TInt64> spShape) {
     return SparseSoftmax.create(scope, spIndices, spValues, spShape);
   }
@@ -1160,9 +1160,9 @@ public <T extends Tensor & TNumber> SparseSoftmax<T> sparseSoftmax(Operand<TInt6
    * @param bShape counterpart to `a_shape` for the other operand; the two shapes must be equal.
    * @return a new instance of SparseSparseMaximum
    */
-  public <T extends Tensor & TNumber> SparseSparseMaximum<T> sparseSparseMaximum(
-      Operand<TInt64> aIndices, Operand<T> aValues, Operand<TInt64> aShape,
-      Operand<TInt64> bIndices, Operand<T> bValues, Operand<TInt64> bShape) {
+  public <T extends TNumber> SparseSparseMaximum<T> sparseSparseMaximum(Operand<TInt64> aIndices,
+      Operand<T> aValues, Operand<TInt64> aShape, Operand<TInt64> bIndices, Operand<T> bValues,
+      Operand<TInt64> bShape) {
     return SparseSparseMaximum.create(scope, aIndices, aValues, aShape, bIndices, bValues, bShape);
   }
 
@@ -1181,7 +1181,7 @@ public <T extends Tensor & TNumber> SparseSparseMaximum<T> sparseSparseMaximum(
    * @param bShape counterpart to `a_shape` for the other operand; the two shapes must be equal.
    * @return a new instance of SparseSparseMinimum
    */
-  public <T extends Tensor> SparseSparseMinimum<T> sparseSparseMinimum(Operand<TInt64> aIndices,
+  public <T extends TType> SparseSparseMinimum<T> sparseSparseMinimum(Operand<TInt64> aIndices,
       Operand<T> aValues, Operand<TInt64> aShape, Operand<TInt64> bIndices, Operand<T> bValues,
       Operand<TInt64> bShape) {
     return SparseSparseMinimum.create(scope, aIndices, aValues, aShape, bIndices, bValues, bShape);
@@ -1219,7 +1219,7 @@ public <T extends Tensor> SparseSparseMinimum<T> sparseSparseMinimum(Operand<TIn
    * @param numSplit The number of ways to split.
    * @return a new instance of SparseSplit
    */
-  public <T extends Tensor> SparseSplit<T> sparseSplit(Operand<TInt64> splitDim,
+  public <T extends TType> SparseSplit<T> sparseSplit(Operand<TInt64> splitDim,
       Operand<TInt64> indices, Operand<T> values, Operand<TInt64> shape, Long numSplit) {
     return SparseSplit.create(scope, splitDim, indices, values, shape, numSplit);
   }
@@ -1236,7 +1236,7 @@ public <T extends Tensor> SparseSplit<T> sparseSplit(Operand<TInt64> splitDim,
    * @param b `ndims`-D Tensor.  With shape `a_shape`.
    * @return a new instance of SparseTensorDenseAdd
    */
-  public <U extends Tensor, T extends Tensor & TNumber> SparseTensorDenseAdd<U> sparseTensorDenseAdd(
+  public <U extends TType, T extends TNumber> SparseTensorDenseAdd<U> sparseTensorDenseAdd(
       Operand<T> aIndices, Operand<U> aValues, Operand<T> aShape, Operand<U> b) {
     return SparseTensorDenseAdd.create(scope, aIndices, aValues, aShape, b);
   }
@@ -1262,7 +1262,7 @@ public <U extends Tensor, T extends Tensor & TNumber> SparseTensorDenseAdd<U> sp
    * @param options carries optional attributes values
    * @return a new instance of SparseTensorDenseMatMul
    */
-  public <U extends Tensor, T extends Tensor & TNumber> SparseTensorDenseMatMul<U> sparseTensorDenseMatMul(
+  public <U extends TType, T extends TNumber> SparseTensorDenseMatMul<U> sparseTensorDenseMatMul(
       Operand<T> aIndices, Operand<U> aValues, Operand<TInt64> aShape, Operand<U> b,
       SparseTensorDenseMatMul.Options... options) {
     return SparseTensorDenseMatMul.create(scope, aIndices, aValues, aShape, b, options);
@@ -1300,7 +1300,7 @@ public <U extends Tensor, T extends Tensor & TNumber> SparseTensorDenseMatMul<U>
    * @param options carries optional attributes values
    * @return a new instance of SparseToDense
    */
-  public <U extends Tensor, T extends Tensor & TNumber> SparseToDense<U> sparseToDense(
+  public <U extends TType, T extends TNumber> SparseToDense<U> sparseToDense(
       Operand<T> sparseIndices, Operand<T> outputShape, Operand<U> sparseValues,
       Operand<U> defaultValue, SparseToDense.Options... options) {
     return SparseToDense.create(scope, sparseIndices, outputShape, sparseValues, defaultValue, options);
@@ -1352,7 +1352,7 @@ public <U extends Tensor, T extends Tensor & TNumber> SparseToDense<U> sparseToD
    * @param options carries optional attributes values
    * @return a new instance of SparseToSparseSetOperation
    */
-  public <T extends Tensor> SparseToSparseSetOperation<T> sparseToSparseSetOperation(
+  public <T extends TType> SparseToSparseSetOperation<T> sparseToSparseSetOperation(
       Operand<TInt64> set1Indices, Operand<T> set1Values, Operand<TInt64> set1Shape,
       Operand<TInt64> set2Indices, Operand<T> set2Values, Operand<TInt64> set2Shape,
       String setOperation, SparseToSparseSetOperation.Options... options) {
@@ -1414,7 +1414,7 @@ public <T extends Tensor> SparseToSparseSetOperation<T> sparseToSparseSetOperati
    * @param options carries optional attributes values
    * @return a new instance of TakeManySparseFromTensorsMap
    */
-  public <T extends Tensor> TakeManySparseFromTensorsMap<T> takeManySparseFromTensorsMap(
+  public <T extends TType> TakeManySparseFromTensorsMap<T> takeManySparseFromTensorsMap(
       Operand<TInt64> sparseHandles, DataType<T> dtype,
       TakeManySparseFromTensorsMap.Options... options) {
     return TakeManySparseFromTensorsMap.create(scope, sparseHandles, dtype, options);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/StringsOps.java b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/StringsOps.java
index 5e2be5647c5..f6491843332 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/StringsOps.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/StringsOps.java
@@ -20,7 +20,6 @@
 import java.util.List;
 import org.tensorflow.DataType;
 import org.tensorflow.Operand;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.strings.Join;
 import org.tensorflow.op.strings.Lower;
 import org.tensorflow.op.strings.ReduceJoin;
@@ -231,7 +230,7 @@ public StringLength stringLength(Operand<TString> input, StringLength.Options...
    * @param preserveShortSequences
    * @return a new instance of StringNGrams
    */
-  public <T extends Tensor & TNumber> StringNGrams<T> stringNGrams(Operand<TString> data,
+  public <T extends TNumber> StringNGrams<T> stringNGrams(Operand<TString> data,
       Operand<T> dataSplits, String separator, List<Long> ngramWidths, String leftPad,
       String rightPad, Long padWidth, Boolean preserveShortSequences) {
     return StringNGrams.create(scope, data, dataSplits, separator, ngramWidths, leftPad, rightPad, padWidth, preserveShortSequences);
@@ -368,8 +367,8 @@ public Strip strip(Operand<TString> input) {
    * @param options carries optional attributes values
    * @return a new instance of Substr
    */
-  public <T extends Tensor & TNumber> Substr substr(Operand<TString> input, Operand<T> pos,
-      Operand<T> len, Substr.Options... options) {
+  public <T extends TNumber> Substr substr(Operand<TString> input, Operand<T> pos, Operand<T> len,
+      Substr.Options... options) {
     return Substr.create(scope, input, pos, len, options);
   }
 
@@ -484,7 +483,7 @@ public ToNumber<TFloat32> toNumber(Operand<TString> stringTensor) {
    * @param outType The numeric type to interpret each string in `string_tensor` as.
    * @return a new instance of ToNumber
    */
-  public <T extends Tensor & TNumber> ToNumber<T> toNumber(Operand<TString> stringTensor,
+  public <T extends TNumber> ToNumber<T> toNumber(Operand<TString> stringTensor,
       DataType<T> outType) {
     return ToNumber.create(scope, stringTensor, outType);
   }
@@ -597,7 +596,7 @@ public UnicodeTranscode unicodeTranscode(Operand<TString> input, String inputEnc
    * @param options carries optional attributes values
    * @return a new instance of UnsortedSegmentJoin
    */
-  public <T extends Tensor & TNumber, U extends Tensor & TNumber> UnsortedSegmentJoin unsortedSegmentJoin(
+  public <T extends TNumber, U extends TNumber> UnsortedSegmentJoin unsortedSegmentJoin(
       Operand<TString> inputs, Operand<T> segmentIds, Operand<U> numSegments,
       UnsortedSegmentJoin.Options... options) {
     return UnsortedSegmentJoin.create(scope, inputs, segmentIds, numSegments, options);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/SummaryOps.java b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/SummaryOps.java
index ce8b445610d..6143335b8fd 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/SummaryOps.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/SummaryOps.java
@@ -18,7 +18,6 @@
 package org.tensorflow.op;
 
 import org.tensorflow.Operand;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.summary.AudioSummary;
 import org.tensorflow.op.summary.HistogramSummary;
 import org.tensorflow.op.summary.ImageSummary;
@@ -28,6 +27,7 @@
 import org.tensorflow.types.TFloat32;
 import org.tensorflow.types.TString;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * An API for building {@code summary} operations as {@link Op Op}s
@@ -83,7 +83,7 @@ public AudioSummary audioSummary(Operand<TString> tag, Operand<TFloat32> tensor,
    * @param values Any shape. Values to use to build the histogram.
    * @return a new instance of HistogramSummary
    */
-  public <T extends Tensor & TNumber> HistogramSummary histogramSummary(Operand<TString> tag,
+  public <T extends TNumber> HistogramSummary histogramSummary(Operand<TString> tag,
       Operand<T> values) {
     return HistogramSummary.create(scope, tag, values);
   }
@@ -144,8 +144,8 @@ public <T extends Tensor & TNumber> HistogramSummary histogramSummary(Operand<TS
    * @param options carries optional attributes values
    * @return a new instance of ImageSummary
    */
-  public <T extends Tensor & TNumber> ImageSummary imageSummary(Operand<TString> tag,
-      Operand<T> tensor, ImageSummary.Options... options) {
+  public <T extends TNumber> ImageSummary imageSummary(Operand<TString> tag, Operand<T> tensor,
+      ImageSummary.Options... options) {
     return ImageSummary.create(scope, tag, tensor, options);
   }
 
@@ -178,8 +178,7 @@ public MergeSummary mergeSummary(Iterable<Operand<TString>> inputs) {
    * @param values Same shape as `tags.  Values for the summary.
    * @return a new instance of ScalarSummary
    */
-  public <T extends Tensor & TNumber> ScalarSummary scalarSummary(Operand<TString> tags,
-      Operand<T> values) {
+  public <T extends TNumber> ScalarSummary scalarSummary(Operand<TString> tags, Operand<T> values) {
     return ScalarSummary.create(scope, tags, values);
   }
 
@@ -192,7 +191,7 @@ public <T extends Tensor & TNumber> ScalarSummary scalarSummary(Operand<TString>
    *  data.
    * @return a new instance of TensorSummary
    */
-  public <T extends Tensor> TensorSummary tensorSummary(Operand<TString> tag, Operand<T> tensor,
+  public <T extends TType> TensorSummary tensorSummary(Operand<TString> tag, Operand<T> tensor,
       Operand<TString> serializedSummaryMetadata) {
     return TensorSummary.create(scope, tag, tensor, serializedSummaryMetadata);
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/TrainOps.java b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/TrainOps.java
index e971180cd5d..2c5d8752136 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/TrainOps.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/TrainOps.java
@@ -20,7 +20,6 @@
 import java.util.List;
 import org.tensorflow.DataType;
 import org.tensorflow.Operand;
-import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.train.AccumulatorApplyGradient;
 import org.tensorflow.op.train.AccumulatorNumAccumulated;
@@ -89,6 +88,7 @@
 import org.tensorflow.types.TInt64;
 import org.tensorflow.types.TString;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * An API for building {@code train} operations as {@link Op Op}s
@@ -112,7 +112,7 @@ public final class TrainOps {
    * @param gradient A tensor of the gradient to be accumulated.
    * @return a new instance of AccumulatorApplyGradient
    */
-  public <T extends Tensor> AccumulatorApplyGradient accumulatorApplyGradient(
+  public <T extends TType> AccumulatorApplyGradient accumulatorApplyGradient(
       Operand<TString> handle, Operand<TInt64> localStep, Operand<T> gradient) {
     return AccumulatorApplyGradient.create(scope, handle, localStep, gradient);
   }
@@ -158,7 +158,7 @@ public AccumulatorSetGlobalStep accumulatorSetGlobalStep(Operand<TString> handle
    *  of the accumulator.
    * @return a new instance of AccumulatorTakeGradient
    */
-  public <T extends Tensor> AccumulatorTakeGradient<T> accumulatorTakeGradient(
+  public <T extends TType> AccumulatorTakeGradient<T> accumulatorTakeGradient(
       Operand<TString> handle, Operand<TInt32> numRequired, DataType<T> dtype) {
     return AccumulatorTakeGradient.create(scope, handle, numRequired, dtype);
   }
@@ -182,7 +182,7 @@ public <T extends Tensor> AccumulatorTakeGradient<T> accumulatorTakeGradient(
    * @param options carries optional attributes values
    * @return a new instance of ApplyAdadelta
    */
-  public <T extends Tensor> ApplyAdadelta<T> applyAdadelta(Operand<T> var, Operand<T> accum,
+  public <T extends TType> ApplyAdadelta<T> applyAdadelta(Operand<T> var, Operand<T> accum,
       Operand<T> accumUpdate, Operand<T> lr, Operand<T> rho, Operand<T> epsilon, Operand<T> grad,
       ApplyAdadelta.Options... options) {
     return ApplyAdadelta.create(scope, var, accum, accumUpdate, lr, rho, epsilon, grad, options);
@@ -202,7 +202,7 @@ public <T extends Tensor> ApplyAdadelta<T> applyAdadelta(Operand<T> var, Operand
    * @param options carries optional attributes values
    * @return a new instance of ApplyAdagrad
    */
-  public <T extends Tensor> ApplyAdagrad<T> applyAdagrad(Operand<T> var, Operand<T> accum,
+  public <T extends TType> ApplyAdagrad<T> applyAdagrad(Operand<T> var, Operand<T> accum,
       Operand<T> lr, Operand<T> grad, ApplyAdagrad.Options... options) {
     return ApplyAdagrad.create(scope, var, accum, lr, grad, options);
   }
@@ -222,7 +222,7 @@ public <T extends Tensor> ApplyAdagrad<T> applyAdagrad(Operand<T> var, Operand<T
    * @param options carries optional attributes values
    * @return a new instance of ApplyAdagradDa
    */
-  public <T extends Tensor> ApplyAdagradDa<T> applyAdagradDa(Operand<T> var,
+  public <T extends TType> ApplyAdagradDa<T> applyAdagradDa(Operand<T> var,
       Operand<T> gradientAccumulator, Operand<T> gradientSquaredAccumulator, Operand<T> grad,
       Operand<T> lr, Operand<T> l1, Operand<T> l2, Operand<TInt64> globalStep,
       ApplyAdagradDa.Options... options) {
@@ -251,7 +251,7 @@ public <T extends Tensor> ApplyAdagradDa<T> applyAdagradDa(Operand<T> var,
    * @param options carries optional attributes values
    * @return a new instance of ApplyAdam
    */
-  public <T extends Tensor> ApplyAdam<T> applyAdam(Operand<T> var, Operand<T> m, Operand<T> v,
+  public <T extends TType> ApplyAdam<T> applyAdam(Operand<T> var, Operand<T> m, Operand<T> v,
       Operand<T> beta1Power, Operand<T> beta2Power, Operand<T> lr, Operand<T> beta1,
       Operand<T> beta2, Operand<T> epsilon, Operand<T> grad, ApplyAdam.Options... options) {
     return ApplyAdam.create(scope, var, m, v, beta1Power, beta2Power, lr, beta1, beta2, epsilon, grad, options);
@@ -275,8 +275,8 @@ public <T extends Tensor> ApplyAdam<T> applyAdam(Operand<T> var, Operand<T> m, O
    * @param options carries optional attributes values
    * @return a new instance of ApplyAddSign
    */
-  public <T extends Tensor> ApplyAddSign<T> applyAddSign(Operand<T> var, Operand<T> m,
-      Operand<T> lr, Operand<T> alpha, Operand<T> signDecay, Operand<T> beta, Operand<T> grad,
+  public <T extends TType> ApplyAddSign<T> applyAddSign(Operand<T> var, Operand<T> m, Operand<T> lr,
+      Operand<T> alpha, Operand<T> signDecay, Operand<T> beta, Operand<T> grad,
       ApplyAddSign.Options... options) {
     return ApplyAddSign.create(scope, var, m, lr, alpha, signDecay, beta, grad, options);
   }
@@ -316,7 +316,7 @@ public <T extends Tensor> ApplyAddSign<T> applyAddSign(Operand<T> var, Operand<T
    * @param options carries optional attributes values
    * @return a new instance of ApplyCenteredRmsProp
    */
-  public <T extends Tensor> ApplyCenteredRmsProp<T> applyCenteredRmsProp(Operand<T> var,
+  public <T extends TType> ApplyCenteredRmsProp<T> applyCenteredRmsProp(Operand<T> var,
       Operand<T> mg, Operand<T> ms, Operand<T> mom, Operand<T> lr, Operand<T> rho,
       Operand<T> momentum, Operand<T> epsilon, Operand<T> grad,
       ApplyCenteredRmsProp.Options... options) {
@@ -347,7 +347,7 @@ public <T extends Tensor> ApplyCenteredRmsProp<T> applyCenteredRmsProp(Operand<T
    * @param options carries optional attributes values
    * @return a new instance of ApplyFtrl
    */
-  public <T extends Tensor> ApplyFtrl<T> applyFtrl(Operand<T> var, Operand<T> accum,
+  public <T extends TType> ApplyFtrl<T> applyFtrl(Operand<T> var, Operand<T> accum,
       Operand<T> linear, Operand<T> grad, Operand<T> lr, Operand<T> l1, Operand<T> l2,
       Operand<T> l2Shrinkage, Operand<T> lrPower, ApplyFtrl.Options... options) {
     return ApplyFtrl.create(scope, var, accum, linear, grad, lr, l1, l2, l2Shrinkage, lrPower, options);
@@ -363,7 +363,7 @@ public <T extends Tensor> ApplyFtrl<T> applyFtrl(Operand<T> var, Operand<T> accu
    * @param options carries optional attributes values
    * @return a new instance of ApplyGradientDescent
    */
-  public <T extends Tensor> ApplyGradientDescent<T> applyGradientDescent(Operand<T> var,
+  public <T extends TType> ApplyGradientDescent<T> applyGradientDescent(Operand<T> var,
       Operand<T> alpha, Operand<T> delta, ApplyGradientDescent.Options... options) {
     return ApplyGradientDescent.create(scope, var, alpha, delta, options);
   }
@@ -385,7 +385,7 @@ public <T extends Tensor> ApplyGradientDescent<T> applyGradientDescent(Operand<T
    * @param options carries optional attributes values
    * @return a new instance of ApplyMomentum
    */
-  public <T extends Tensor> ApplyMomentum<T> applyMomentum(Operand<T> var, Operand<T> accum,
+  public <T extends TType> ApplyMomentum<T> applyMomentum(Operand<T> var, Operand<T> accum,
       Operand<T> lr, Operand<T> grad, Operand<T> momentum, ApplyMomentum.Options... options) {
     return ApplyMomentum.create(scope, var, accum, lr, grad, momentum, options);
   }
@@ -408,7 +408,7 @@ public <T extends Tensor> ApplyMomentum<T> applyMomentum(Operand<T> var, Operand
    * @param options carries optional attributes values
    * @return a new instance of ApplyPowerSign
    */
-  public <T extends Tensor> ApplyPowerSign<T> applyPowerSign(Operand<T> var, Operand<T> m,
+  public <T extends TType> ApplyPowerSign<T> applyPowerSign(Operand<T> var, Operand<T> m,
       Operand<T> lr, Operand<T> logbase, Operand<T> signDecay, Operand<T> beta, Operand<T> grad,
       ApplyPowerSign.Options... options) {
     return ApplyPowerSign.create(scope, var, m, lr, logbase, signDecay, beta, grad, options);
@@ -431,7 +431,7 @@ public <T extends Tensor> ApplyPowerSign<T> applyPowerSign(Operand<T> var, Opera
    * @param options carries optional attributes values
    * @return a new instance of ApplyProximalAdagrad
    */
-  public <T extends Tensor> ApplyProximalAdagrad<T> applyProximalAdagrad(Operand<T> var,
+  public <T extends TType> ApplyProximalAdagrad<T> applyProximalAdagrad(Operand<T> var,
       Operand<T> accum, Operand<T> lr, Operand<T> l1, Operand<T> l2, Operand<T> grad,
       ApplyProximalAdagrad.Options... options) {
     return ApplyProximalAdagrad.create(scope, var, accum, lr, l1, l2, grad, options);
@@ -452,7 +452,7 @@ public <T extends Tensor> ApplyProximalAdagrad<T> applyProximalAdagrad(Operand<T
    * @param options carries optional attributes values
    * @return a new instance of ApplyProximalGradientDescent
    */
-  public <T extends Tensor> ApplyProximalGradientDescent<T> applyProximalGradientDescent(
+  public <T extends TType> ApplyProximalGradientDescent<T> applyProximalGradientDescent(
       Operand<T> var, Operand<T> alpha, Operand<T> l1, Operand<T> l2, Operand<T> delta,
       ApplyProximalGradientDescent.Options... options) {
     return ApplyProximalGradientDescent.create(scope, var, alpha, l1, l2, delta, options);
@@ -484,7 +484,7 @@ public <T extends Tensor> ApplyProximalGradientDescent<T> applyProximalGradientD
    * @param options carries optional attributes values
    * @return a new instance of ApplyRmsProp
    */
-  public <T extends Tensor> ApplyRmsProp<T> applyRmsProp(Operand<T> var, Operand<T> ms,
+  public <T extends TType> ApplyRmsProp<T> applyRmsProp(Operand<T> var, Operand<T> ms,
       Operand<T> mom, Operand<T> lr, Operand<T> rho, Operand<T> momentum, Operand<T> epsilon,
       Operand<T> grad, ApplyRmsProp.Options... options) {
     return ApplyRmsProp.create(scope, var, ms, mom, lr, rho, momentum, epsilon, grad, options);
@@ -522,7 +522,7 @@ public <T extends Tensor> ApplyRmsProp<T> applyRmsProp(Operand<T> var, Operand<T
    * @param options carries optional attributes values
    * @return a new instance of BatchMatMul
    */
-  public <T extends Tensor> BatchMatMul<T> batchMatMul(Operand<T> x, Operand<T> y,
+  public <T extends TType> BatchMatMul<T> batchMatMul(Operand<T> x, Operand<T> y,
       BatchMatMul.Options... options) {
     return BatchMatMul.create(scope, x, y, options);
   }
@@ -542,7 +542,7 @@ public <T extends Tensor> BatchMatMul<T> batchMatMul(Operand<T> x, Operand<T> y,
    * @param options carries optional attributes values
    * @return a new instance of ConditionalAccumulator
    */
-  public <T extends Tensor> ConditionalAccumulator conditionalAccumulator(DataType<T> dtype,
+  public <T extends TType> ConditionalAccumulator conditionalAccumulator(DataType<T> dtype,
       Shape shape, ConditionalAccumulator.Options... options) {
     return ConditionalAccumulator.create(scope, dtype, shape, options);
   }
@@ -649,7 +649,7 @@ public NegTrain negTrain(Operand<TFloat32> wIn, Operand<TFloat32> wOut, Operand<
    * @param options carries optional attributes values
    * @return a new instance of PreventGradient
    */
-  public <T extends Tensor> PreventGradient<T> preventGradient(Operand<T> input,
+  public <T extends TType> PreventGradient<T> preventGradient(Operand<T> input,
       PreventGradient.Options... options) {
     return PreventGradient.create(scope, input, options);
   }
@@ -672,7 +672,7 @@ public <T extends Tensor> PreventGradient<T> preventGradient(Operand<T> input,
    * @param options carries optional attributes values
    * @return a new instance of ResourceApplyAdadelta
    */
-  public <T extends Tensor> ResourceApplyAdadelta resourceApplyAdadelta(Operand<?> var,
+  public <T extends TType> ResourceApplyAdadelta resourceApplyAdadelta(Operand<?> var,
       Operand<?> accum, Operand<?> accumUpdate, Operand<T> lr, Operand<T> rho, Operand<T> epsilon,
       Operand<T> grad, ResourceApplyAdadelta.Options... options) {
     return ResourceApplyAdadelta.create(scope, var, accum, accumUpdate, lr, rho, epsilon, grad, options);
@@ -692,7 +692,7 @@ public <T extends Tensor> ResourceApplyAdadelta resourceApplyAdadelta(Operand<?>
    * @param options carries optional attributes values
    * @return a new instance of ResourceApplyAdagradDa
    */
-  public <T extends Tensor> ResourceApplyAdagradDa resourceApplyAdagradDa(Operand<?> var,
+  public <T extends TType> ResourceApplyAdagradDa resourceApplyAdagradDa(Operand<?> var,
       Operand<?> gradientAccumulator, Operand<?> gradientSquaredAccumulator, Operand<T> grad,
       Operand<T> lr, Operand<T> l1, Operand<T> l2, Operand<TInt64> globalStep,
       ResourceApplyAdagradDa.Options... options) {
@@ -720,7 +720,7 @@ public <T extends Tensor> ResourceApplyAdagradDa resourceApplyAdagradDa(Operand<
    * @param options carries optional attributes values
    * @return a new instance of ResourceApplyAdam
    */
-  public <T extends Tensor> ResourceApplyAdam resourceApplyAdam(Operand<?> var, Operand<?> m,
+  public <T extends TType> ResourceApplyAdam resourceApplyAdam(Operand<?> var, Operand<?> m,
       Operand<?> v, Operand<T> beta1Power, Operand<T> beta2Power, Operand<T> lr, Operand<T> beta1,
       Operand<T> beta2, Operand<T> epsilon, Operand<T> grad, ResourceApplyAdam.Options... options) {
     return ResourceApplyAdam.create(scope, var, m, v, beta1Power, beta2Power, lr, beta1, beta2, epsilon, grad, options);
@@ -749,10 +749,10 @@ public <T extends Tensor> ResourceApplyAdam resourceApplyAdam(Operand<?> var, Op
    * @param options carries optional attributes values
    * @return a new instance of ResourceApplyAdamWithAmsgrad
    */
-  public <T extends Tensor> ResourceApplyAdamWithAmsgrad resourceApplyAdamWithAmsgrad(
-      Operand<?> var, Operand<?> m, Operand<?> v, Operand<?> vhat, Operand<T> beta1Power,
-      Operand<T> beta2Power, Operand<T> lr, Operand<T> beta1, Operand<T> beta2, Operand<T> epsilon,
-      Operand<T> grad, ResourceApplyAdamWithAmsgrad.Options... options) {
+  public <T extends TType> ResourceApplyAdamWithAmsgrad resourceApplyAdamWithAmsgrad(Operand<?> var,
+      Operand<?> m, Operand<?> v, Operand<?> vhat, Operand<T> beta1Power, Operand<T> beta2Power,
+      Operand<T> lr, Operand<T> beta1, Operand<T> beta2, Operand<T> epsilon, Operand<T> grad,
+      ResourceApplyAdamWithAmsgrad.Options... options) {
     return ResourceApplyAdamWithAmsgrad.create(scope, var, m, v, vhat, beta1Power, beta2Power, lr, beta1, beta2, epsilon, grad, options);
   }
 
@@ -773,7 +773,7 @@ public <T extends Tensor> ResourceApplyAdamWithAmsgrad resourceApplyAdamWithAmsg
    * @param options carries optional attributes values
    * @return a new instance of ResourceApplyAddSign
    */
-  public <T extends Tensor> ResourceApplyAddSign resourceApplyAddSign(Operand<?> var, Operand<?> m,
+  public <T extends TType> ResourceApplyAddSign resourceApplyAddSign(Operand<?> var, Operand<?> m,
       Operand<T> lr, Operand<T> alpha, Operand<T> signDecay, Operand<T> beta, Operand<T> grad,
       ResourceApplyAddSign.Options... options) {
     return ResourceApplyAddSign.create(scope, var, m, lr, alpha, signDecay, beta, grad, options);
@@ -813,8 +813,8 @@ public <T extends Tensor> ResourceApplyAddSign resourceApplyAddSign(Operand<?> v
    * @param options carries optional attributes values
    * @return a new instance of ResourceApplyCenteredRmsProp
    */
-  public <T extends Tensor> ResourceApplyCenteredRmsProp resourceApplyCenteredRmsProp(
-      Operand<?> var, Operand<?> mg, Operand<?> ms, Operand<?> mom, Operand<T> lr, Operand<T> rho,
+  public <T extends TType> ResourceApplyCenteredRmsProp resourceApplyCenteredRmsProp(Operand<?> var,
+      Operand<?> mg, Operand<?> ms, Operand<?> mom, Operand<T> lr, Operand<T> rho,
       Operand<T> momentum, Operand<T> epsilon, Operand<T> grad,
       ResourceApplyCenteredRmsProp.Options... options) {
     return ResourceApplyCenteredRmsProp.create(scope, var, mg, ms, mom, lr, rho, momentum, epsilon, grad, options);
@@ -843,7 +843,7 @@ public <T extends Tensor> ResourceApplyCenteredRmsProp resourceApplyCenteredRmsP
    * @param options carries optional attributes values
    * @return a new instance of ResourceApplyFtrl
    */
-  public <T extends Tensor> ResourceApplyFtrl resourceApplyFtrl(Operand<?> var, Operand<?> accum,
+  public <T extends TType> ResourceApplyFtrl resourceApplyFtrl(Operand<?> var, Operand<?> accum,
       Operand<?> linear, Operand<T> grad, Operand<T> lr, Operand<T> l1, Operand<T> l2,
       Operand<T> l2Shrinkage, Operand<T> lrPower, ResourceApplyFtrl.Options... options) {
     return ResourceApplyFtrl.create(scope, var, accum, linear, grad, lr, l1, l2, l2Shrinkage, lrPower, options);
@@ -858,9 +858,8 @@ public <T extends Tensor> ResourceApplyFtrl resourceApplyFtrl(Operand<?> var, Op
    * @param options carries optional attributes values
    * @return a new instance of ResourceApplyGradientDescent
    */
-  public <T extends Tensor> ResourceApplyGradientDescent resourceApplyGradientDescent(
-      Operand<?> var, Operand<T> alpha, Operand<T> delta,
-      ResourceApplyGradientDescent.Options... options) {
+  public <T extends TType> ResourceApplyGradientDescent resourceApplyGradientDescent(Operand<?> var,
+      Operand<T> alpha, Operand<T> delta, ResourceApplyGradientDescent.Options... options) {
     return ResourceApplyGradientDescent.create(scope, var, alpha, delta, options);
   }
 
@@ -880,7 +879,7 @@ public <T extends Tensor> ResourceApplyGradientDescent resourceApplyGradientDesc
    * @param options carries optional attributes values
    * @return a new instance of ResourceApplyKerasMomentum
    */
-  public <T extends Tensor> ResourceApplyKerasMomentum resourceApplyKerasMomentum(Operand<?> var,
+  public <T extends TType> ResourceApplyKerasMomentum resourceApplyKerasMomentum(Operand<?> var,
       Operand<?> accum, Operand<T> lr, Operand<T> grad, Operand<T> momentum,
       ResourceApplyKerasMomentum.Options... options) {
     return ResourceApplyKerasMomentum.create(scope, var, accum, lr, grad, momentum, options);
@@ -902,7 +901,7 @@ public <T extends Tensor> ResourceApplyKerasMomentum resourceApplyKerasMomentum(
    * @param options carries optional attributes values
    * @return a new instance of ResourceApplyMomentum
    */
-  public <T extends Tensor> ResourceApplyMomentum resourceApplyMomentum(Operand<?> var,
+  public <T extends TType> ResourceApplyMomentum resourceApplyMomentum(Operand<?> var,
       Operand<?> accum, Operand<T> lr, Operand<T> grad, Operand<T> momentum,
       ResourceApplyMomentum.Options... options) {
     return ResourceApplyMomentum.create(scope, var, accum, lr, grad, momentum, options);
@@ -925,7 +924,7 @@ public <T extends Tensor> ResourceApplyMomentum resourceApplyMomentum(Operand<?>
    * @param options carries optional attributes values
    * @return a new instance of ResourceApplyPowerSign
    */
-  public <T extends Tensor> ResourceApplyPowerSign resourceApplyPowerSign(Operand<?> var,
+  public <T extends TType> ResourceApplyPowerSign resourceApplyPowerSign(Operand<?> var,
       Operand<?> m, Operand<T> lr, Operand<T> logbase, Operand<T> signDecay, Operand<T> beta,
       Operand<T> grad, ResourceApplyPowerSign.Options... options) {
     return ResourceApplyPowerSign.create(scope, var, m, lr, logbase, signDecay, beta, grad, options);
@@ -947,9 +946,9 @@ public <T extends Tensor> ResourceApplyPowerSign resourceApplyPowerSign(Operand<
    * @param options carries optional attributes values
    * @return a new instance of ResourceApplyProximalAdagrad
    */
-  public <T extends Tensor> ResourceApplyProximalAdagrad resourceApplyProximalAdagrad(
-      Operand<?> var, Operand<?> accum, Operand<T> lr, Operand<T> l1, Operand<T> l2,
-      Operand<T> grad, ResourceApplyProximalAdagrad.Options... options) {
+  public <T extends TType> ResourceApplyProximalAdagrad resourceApplyProximalAdagrad(Operand<?> var,
+      Operand<?> accum, Operand<T> lr, Operand<T> l1, Operand<T> l2, Operand<T> grad,
+      ResourceApplyProximalAdagrad.Options... options) {
     return ResourceApplyProximalAdagrad.create(scope, var, accum, lr, l1, l2, grad, options);
   }
 
@@ -967,7 +966,7 @@ public <T extends Tensor> ResourceApplyProximalAdagrad resourceApplyProximalAdag
    * @param options carries optional attributes values
    * @return a new instance of ResourceApplyProximalGradientDescent
    */
-  public <T extends Tensor> ResourceApplyProximalGradientDescent resourceApplyProximalGradientDescent(
+  public <T extends TType> ResourceApplyProximalGradientDescent resourceApplyProximalGradientDescent(
       Operand<?> var, Operand<T> alpha, Operand<T> l1, Operand<T> l2, Operand<T> delta,
       ResourceApplyProximalGradientDescent.Options... options) {
     return ResourceApplyProximalGradientDescent.create(scope, var, alpha, l1, l2, delta, options);
@@ -998,7 +997,7 @@ public <T extends Tensor> ResourceApplyProximalGradientDescent resourceApplyProx
    * @param options carries optional attributes values
    * @return a new instance of ResourceApplyRmsProp
    */
-  public <T extends Tensor> ResourceApplyRmsProp resourceApplyRmsProp(Operand<?> var, Operand<?> ms,
+  public <T extends TType> ResourceApplyRmsProp resourceApplyRmsProp(Operand<?> var, Operand<?> ms,
       Operand<?> mom, Operand<T> lr, Operand<T> rho, Operand<T> momentum, Operand<T> epsilon,
       Operand<T> grad, ResourceApplyRmsProp.Options... options) {
     return ResourceApplyRmsProp.create(scope, var, ms, mom, lr, rho, momentum, epsilon, grad, options);
@@ -1018,7 +1017,7 @@ public <T extends Tensor> ResourceApplyRmsProp resourceApplyRmsProp(Operand<?> v
    * @param options carries optional attributes values
    * @return a new instance of ResourceSparseApplyAdadelta
    */
-  public <T extends Tensor, U extends Tensor & TNumber> ResourceSparseApplyAdadelta resourceSparseApplyAdadelta(
+  public <T extends TType, U extends TNumber> ResourceSparseApplyAdadelta resourceSparseApplyAdadelta(
       Operand<?> var, Operand<?> accum, Operand<?> accumUpdate, Operand<T> lr, Operand<T> rho,
       Operand<T> epsilon, Operand<T> grad, Operand<U> indices,
       ResourceSparseApplyAdadelta.Options... options) {
@@ -1040,7 +1039,7 @@ public <T extends Tensor, U extends Tensor & TNumber> ResourceSparseApplyAdadelt
    * @param options carries optional attributes values
    * @return a new instance of ResourceSparseApplyAdagrad
    */
-  public <T extends Tensor, U extends Tensor & TNumber> ResourceSparseApplyAdagrad resourceSparseApplyAdagrad(
+  public <T extends TType, U extends TNumber> ResourceSparseApplyAdagrad resourceSparseApplyAdagrad(
       Operand<?> var, Operand<?> accum, Operand<T> lr, Operand<T> grad, Operand<U> indices,
       ResourceSparseApplyAdagrad.Options... options) {
     return ResourceSparseApplyAdagrad.create(scope, var, accum, lr, grad, indices, options);
@@ -1061,7 +1060,7 @@ public <T extends Tensor, U extends Tensor & TNumber> ResourceSparseApplyAdagrad
    * @param options carries optional attributes values
    * @return a new instance of ResourceSparseApplyAdagradDa
    */
-  public <T extends Tensor, U extends Tensor & TNumber> ResourceSparseApplyAdagradDa resourceSparseApplyAdagradDa(
+  public <T extends TType, U extends TNumber> ResourceSparseApplyAdagradDa resourceSparseApplyAdagradDa(
       Operand<?> var, Operand<?> gradientAccumulator, Operand<?> gradientSquaredAccumulator,
       Operand<T> grad, Operand<U> indices, Operand<T> lr, Operand<T> l1, Operand<T> l2,
       Operand<TInt64> globalStep, ResourceSparseApplyAdagradDa.Options... options) {
@@ -1101,7 +1100,7 @@ public <T extends Tensor, U extends Tensor & TNumber> ResourceSparseApplyAdagrad
    * @param options carries optional attributes values
    * @return a new instance of ResourceSparseApplyCenteredRmsProp
    */
-  public <T extends Tensor, U extends Tensor & TNumber> ResourceSparseApplyCenteredRmsProp resourceSparseApplyCenteredRmsProp(
+  public <T extends TType, U extends TNumber> ResourceSparseApplyCenteredRmsProp resourceSparseApplyCenteredRmsProp(
       Operand<?> var, Operand<?> mg, Operand<?> ms, Operand<?> mom, Operand<T> lr, Operand<T> rho,
       Operand<T> momentum, Operand<T> epsilon, Operand<T> grad, Operand<U> indices,
       ResourceSparseApplyCenteredRmsProp.Options... options) {
@@ -1133,7 +1132,7 @@ public <T extends Tensor, U extends Tensor & TNumber> ResourceSparseApplyCentere
    * @param options carries optional attributes values
    * @return a new instance of ResourceSparseApplyFtrl
    */
-  public <T extends Tensor, U extends Tensor & TNumber> ResourceSparseApplyFtrl resourceSparseApplyFtrl(
+  public <T extends TType, U extends TNumber> ResourceSparseApplyFtrl resourceSparseApplyFtrl(
       Operand<?> var, Operand<?> accum, Operand<?> linear, Operand<T> grad, Operand<U> indices,
       Operand<T> lr, Operand<T> l1, Operand<T> l2, Operand<T> l2Shrinkage, Operand<T> lrPower,
       ResourceSparseApplyFtrl.Options... options) {
@@ -1159,7 +1158,7 @@ public <T extends Tensor, U extends Tensor & TNumber> ResourceSparseApplyFtrl re
    * @param options carries optional attributes values
    * @return a new instance of ResourceSparseApplyKerasMomentum
    */
-  public <T extends Tensor, U extends Tensor & TNumber> ResourceSparseApplyKerasMomentum resourceSparseApplyKerasMomentum(
+  public <T extends TType, U extends TNumber> ResourceSparseApplyKerasMomentum resourceSparseApplyKerasMomentum(
       Operand<?> var, Operand<?> accum, Operand<T> lr, Operand<T> grad, Operand<U> indices,
       Operand<T> momentum, ResourceSparseApplyKerasMomentum.Options... options) {
     return ResourceSparseApplyKerasMomentum.create(scope, var, accum, lr, grad, indices, momentum, options);
@@ -1184,7 +1183,7 @@ public <T extends Tensor, U extends Tensor & TNumber> ResourceSparseApplyKerasMo
    * @param options carries optional attributes values
    * @return a new instance of ResourceSparseApplyMomentum
    */
-  public <T extends Tensor, U extends Tensor & TNumber> ResourceSparseApplyMomentum resourceSparseApplyMomentum(
+  public <T extends TType, U extends TNumber> ResourceSparseApplyMomentum resourceSparseApplyMomentum(
       Operand<?> var, Operand<?> accum, Operand<T> lr, Operand<T> grad, Operand<U> indices,
       Operand<T> momentum, ResourceSparseApplyMomentum.Options... options) {
     return ResourceSparseApplyMomentum.create(scope, var, accum, lr, grad, indices, momentum, options);
@@ -1209,7 +1208,7 @@ public <T extends Tensor, U extends Tensor & TNumber> ResourceSparseApplyMomentu
    * @param options carries optional attributes values
    * @return a new instance of ResourceSparseApplyProximalAdagrad
    */
-  public <T extends Tensor, U extends Tensor & TNumber> ResourceSparseApplyProximalAdagrad resourceSparseApplyProximalAdagrad(
+  public <T extends TType, U extends TNumber> ResourceSparseApplyProximalAdagrad resourceSparseApplyProximalAdagrad(
       Operand<?> var, Operand<?> accum, Operand<T> lr, Operand<T> l1, Operand<T> l2,
       Operand<T> grad, Operand<U> indices, ResourceSparseApplyProximalAdagrad.Options... options) {
     return ResourceSparseApplyProximalAdagrad.create(scope, var, accum, lr, l1, l2, grad, indices, options);
@@ -1231,7 +1230,7 @@ public <T extends Tensor, U extends Tensor & TNumber> ResourceSparseApplyProxima
    * @param options carries optional attributes values
    * @return a new instance of ResourceSparseApplyProximalGradientDescent
    */
-  public <T extends Tensor, U extends Tensor & TNumber> ResourceSparseApplyProximalGradientDescent resourceSparseApplyProximalGradientDescent(
+  public <T extends TType, U extends TNumber> ResourceSparseApplyProximalGradientDescent resourceSparseApplyProximalGradientDescent(
       Operand<?> var, Operand<T> alpha, Operand<T> l1, Operand<T> l2, Operand<T> grad,
       Operand<U> indices, ResourceSparseApplyProximalGradientDescent.Options... options) {
     return ResourceSparseApplyProximalGradientDescent.create(scope, var, alpha, l1, l2, grad, indices, options);
@@ -1263,7 +1262,7 @@ public <T extends Tensor, U extends Tensor & TNumber> ResourceSparseApplyProxima
    * @param options carries optional attributes values
    * @return a new instance of ResourceSparseApplyRmsProp
    */
-  public <T extends Tensor, U extends Tensor & TNumber> ResourceSparseApplyRmsProp resourceSparseApplyRmsProp(
+  public <T extends TType, U extends TNumber> ResourceSparseApplyRmsProp resourceSparseApplyRmsProp(
       Operand<?> var, Operand<?> ms, Operand<?> mom, Operand<T> lr, Operand<T> rho,
       Operand<T> momentum, Operand<T> epsilon, Operand<T> grad, Operand<U> indices,
       ResourceSparseApplyRmsProp.Options... options) {
@@ -1321,7 +1320,7 @@ public Restore restore(Operand<TString> prefix, Operand<TString> tensorNames,
    * @param options carries optional attributes values
    * @return a new instance of RestoreSlice
    */
-  public <T extends Tensor> RestoreSlice<T> restoreSlice(Operand<TString> filePattern,
+  public <T extends TType> RestoreSlice<T> restoreSlice(Operand<TString> filePattern,
       Operand<TString> tensorName, Operand<TString> shapeAndSlice, DataType<T> dt,
       RestoreSlice.Options... options) {
     return RestoreSlice.create(scope, filePattern, tensorName, shapeAndSlice, dt, options);
@@ -1431,7 +1430,7 @@ public SdcaShrinkL1 sdcaShrinkL1(Iterable<Operand<TFloat32>> weights, Float l1,
    * @param options carries optional attributes values
    * @return a new instance of SparseApplyAdadelta
    */
-  public <T extends Tensor, U extends Tensor & TNumber> SparseApplyAdadelta<T> sparseApplyAdadelta(
+  public <T extends TType, U extends TNumber> SparseApplyAdadelta<T> sparseApplyAdadelta(
       Operand<T> var, Operand<T> accum, Operand<T> accumUpdate, Operand<T> lr, Operand<T> rho,
       Operand<T> epsilon, Operand<T> grad, Operand<U> indices,
       SparseApplyAdadelta.Options... options) {
@@ -1454,7 +1453,7 @@ public <T extends Tensor, U extends Tensor & TNumber> SparseApplyAdadelta<T> spa
    * @param options carries optional attributes values
    * @return a new instance of SparseApplyAdagradDa
    */
-  public <T extends Tensor, U extends Tensor & TNumber> SparseApplyAdagradDa<T> sparseApplyAdagradDa(
+  public <T extends TType, U extends TNumber> SparseApplyAdagradDa<T> sparseApplyAdagradDa(
       Operand<T> var, Operand<T> gradientAccumulator, Operand<T> gradientSquaredAccumulator,
       Operand<T> grad, Operand<U> indices, Operand<T> lr, Operand<T> l1, Operand<T> l2,
       Operand<TInt64> globalStep, SparseApplyAdagradDa.Options... options) {
@@ -1495,7 +1494,7 @@ public <T extends Tensor, U extends Tensor & TNumber> SparseApplyAdagradDa<T> sp
    * @param options carries optional attributes values
    * @return a new instance of SparseApplyCenteredRmsProp
    */
-  public <T extends Tensor, U extends Tensor & TNumber> SparseApplyCenteredRmsProp<T> sparseApplyCenteredRmsProp(
+  public <T extends TType, U extends TNumber> SparseApplyCenteredRmsProp<T> sparseApplyCenteredRmsProp(
       Operand<T> var, Operand<T> mg, Operand<T> ms, Operand<T> mom, Operand<T> lr, Operand<T> rho,
       Operand<T> momentum, Operand<T> epsilon, Operand<T> grad, Operand<U> indices,
       SparseApplyCenteredRmsProp.Options... options) {
@@ -1528,9 +1527,9 @@ public <T extends Tensor, U extends Tensor & TNumber> SparseApplyCenteredRmsProp
    * @param options carries optional attributes values
    * @return a new instance of SparseApplyFtrl
    */
-  public <T extends Tensor, U extends Tensor & TNumber> SparseApplyFtrl<T> sparseApplyFtrl(
-      Operand<T> var, Operand<T> accum, Operand<T> linear, Operand<T> grad, Operand<U> indices,
-      Operand<T> lr, Operand<T> l1, Operand<T> l2, Operand<T> l2Shrinkage, Operand<T> lrPower,
+  public <T extends TType, U extends TNumber> SparseApplyFtrl<T> sparseApplyFtrl(Operand<T> var,
+      Operand<T> accum, Operand<T> linear, Operand<T> grad, Operand<U> indices, Operand<T> lr,
+      Operand<T> l1, Operand<T> l2, Operand<T> l2Shrinkage, Operand<T> lrPower,
       SparseApplyFtrl.Options... options) {
     return SparseApplyFtrl.create(scope, var, accum, linear, grad, indices, lr, l1, l2, l2Shrinkage, lrPower, options);
   }
@@ -1555,7 +1554,7 @@ public <T extends Tensor, U extends Tensor & TNumber> SparseApplyFtrl<T> sparseA
    * @param options carries optional attributes values
    * @return a new instance of SparseApplyMomentum
    */
-  public <T extends Tensor, U extends Tensor & TNumber> SparseApplyMomentum<T> sparseApplyMomentum(
+  public <T extends TType, U extends TNumber> SparseApplyMomentum<T> sparseApplyMomentum(
       Operand<T> var, Operand<T> accum, Operand<T> lr, Operand<T> grad, Operand<U> indices,
       Operand<T> momentum, SparseApplyMomentum.Options... options) {
     return SparseApplyMomentum.create(scope, var, accum, lr, grad, indices, momentum, options);
@@ -1581,7 +1580,7 @@ public <T extends Tensor, U extends Tensor & TNumber> SparseApplyMomentum<T> spa
    * @param options carries optional attributes values
    * @return a new instance of SparseApplyProximalAdagrad
    */
-  public <T extends Tensor, U extends Tensor & TNumber> SparseApplyProximalAdagrad<T> sparseApplyProximalAdagrad(
+  public <T extends TType, U extends TNumber> SparseApplyProximalAdagrad<T> sparseApplyProximalAdagrad(
       Operand<T> var, Operand<T> accum, Operand<T> lr, Operand<T> l1, Operand<T> l2,
       Operand<T> grad, Operand<U> indices, SparseApplyProximalAdagrad.Options... options) {
     return SparseApplyProximalAdagrad.create(scope, var, accum, lr, l1, l2, grad, indices, options);
@@ -1604,7 +1603,7 @@ public <T extends Tensor, U extends Tensor & TNumber> SparseApplyProximalAdagrad
    * @param options carries optional attributes values
    * @return a new instance of SparseApplyProximalGradientDescent
    */
-  public <T extends Tensor, U extends Tensor & TNumber> SparseApplyProximalGradientDescent<T> sparseApplyProximalGradientDescent(
+  public <T extends TType, U extends TNumber> SparseApplyProximalGradientDescent<T> sparseApplyProximalGradientDescent(
       Operand<T> var, Operand<T> alpha, Operand<T> l1, Operand<T> l2, Operand<T> grad,
       Operand<U> indices, SparseApplyProximalGradientDescent.Options... options) {
     return SparseApplyProximalGradientDescent.create(scope, var, alpha, l1, l2, grad, indices, options);
@@ -1637,7 +1636,7 @@ public <T extends Tensor, U extends Tensor & TNumber> SparseApplyProximalGradien
    * @param options carries optional attributes values
    * @return a new instance of SparseApplyRmsProp
    */
-  public <T extends Tensor, U extends Tensor & TNumber> SparseApplyRmsProp<T> sparseApplyRmsProp(
+  public <T extends TType, U extends TNumber> SparseApplyRmsProp<T> sparseApplyRmsProp(
       Operand<T> var, Operand<T> ms, Operand<T> mom, Operand<T> lr, Operand<T> rho,
       Operand<T> momentum, Operand<T> epsilon, Operand<T> grad, Operand<U> indices,
       SparseApplyRmsProp.Options... options) {
@@ -1656,7 +1655,7 @@ public <T extends Tensor, U extends Tensor & TNumber> SparseApplyRmsProp<T> spar
    * @param multiples
    * @return a new instance of TileGrad
    */
-  public <T extends Tensor> TileGrad<T> tileGrad(Operand<T> input, Operand<TInt32> multiples) {
+  public <T extends TType> TileGrad<T> tileGrad(Operand<T> input, Operand<TInt32> multiples) {
     return TileGrad.create(scope, input, multiples);
   }
 }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/XlaOps.java b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/XlaOps.java
index 01d4ffa1ba7..535972d4883 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/XlaOps.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/XlaOps.java
@@ -19,7 +19,6 @@
 
 import org.tensorflow.DataType;
 import org.tensorflow.Operand;
-import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.xla.BroadcastHelper;
 import org.tensorflow.op.xla.ClusterOutput;
@@ -40,6 +39,7 @@
 import org.tensorflow.op.xla.Sort;
 import org.tensorflow.op.xla.Svd;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * An API for building {@code xla} operations as {@link Op Op}s
@@ -66,8 +66,8 @@ public final class XlaOps {
    * @param broadcastDims an XLA-style broadcast dimension specification
    * @return a new instance of BroadcastHelper
    */
-  public <T extends Tensor, U extends Tensor & TNumber> BroadcastHelper<T> broadcastHelper(
-      Operand<T> lhs, Operand<T> rhs, Operand<U> broadcastDims) {
+  public <T extends TType, U extends TNumber> BroadcastHelper<T> broadcastHelper(Operand<T> lhs,
+      Operand<T> rhs, Operand<U> broadcastDims) {
     return BroadcastHelper.create(scope, lhs, rhs, broadcastDims);
   }
 
@@ -78,7 +78,7 @@ public <T extends Tensor, U extends Tensor & TNumber> BroadcastHelper<T> broadca
    * @param input
    * @return a new instance of ClusterOutput
    */
-  public <T extends Tensor> ClusterOutput<T> clusterOutput(Operand<T> input) {
+  public <T extends TType> ClusterOutput<T> clusterOutput(Operand<T> input) {
     return ClusterOutput.create(scope, input);
   }
 
@@ -100,7 +100,7 @@ public <T extends Tensor> ClusterOutput<T> clusterOutput(Operand<T> input) {
    * @param precisionConfig a serialized xla::PrecisionConfig proto.
    * @return a new instance of Conv
    */
-  public <T extends Tensor, U extends Tensor & TNumber> Conv<T> conv(Operand<T> lhs, Operand<T> rhs,
+  public <T extends TType, U extends TNumber> Conv<T> conv(Operand<T> lhs, Operand<T> rhs,
       Operand<U> windowStrides, Operand<U> padding, Operand<U> lhsDilation, Operand<U> rhsDilation,
       Operand<U> featureGroupCount, String dimensionNumbers, String precisionConfig) {
     return Conv.create(scope, lhs, rhs, windowStrides, padding, lhsDilation, rhsDilation, featureGroupCount, dimensionNumbers, precisionConfig);
@@ -137,7 +137,7 @@ public Dequantize dequantize(Operand<?> input, Float minRange, Float maxRange, S
    * @param precisionConfig a serialized xla::PrecisionConfig proto.
    * @return a new instance of Dot
    */
-  public <T extends Tensor> Dot<T> dot(Operand<T> lhs, Operand<T> rhs, String dimensionNumbers,
+  public <T extends TType> Dot<T> dot(Operand<T> lhs, Operand<T> rhs, String dimensionNumbers,
       String precisionConfig) {
     return Dot.create(scope, lhs, rhs, dimensionNumbers, precisionConfig);
   }
@@ -163,8 +163,8 @@ public <T extends Tensor> Dot<T> dot(Operand<T> lhs, Operand<T> rhs, String dime
    * @param sizeIndices
    * @return a new instance of DynamicSlice
    */
-  public <T extends Tensor, U extends Tensor & TNumber> DynamicSlice<T> dynamicSlice(
-      Operand<T> input, Operand<U> startIndices, Operand<U> sizeIndices) {
+  public <T extends TType, U extends TNumber> DynamicSlice<T> dynamicSlice(Operand<T> input,
+      Operand<U> startIndices, Operand<U> sizeIndices) {
     return DynamicSlice.create(scope, input, startIndices, sizeIndices);
   }
 
@@ -188,7 +188,7 @@ public <T extends Tensor, U extends Tensor & TNumber> DynamicSlice<T> dynamicSli
    *  `input`.
    * @return a new instance of DynamicUpdateSlice
    */
-  public <T extends Tensor, U extends Tensor & TNumber> DynamicUpdateSlice<T> dynamicUpdateSlice(
+  public <T extends TType, U extends TNumber> DynamicUpdateSlice<T> dynamicUpdateSlice(
       Operand<T> input, Operand<T> update, Operand<U> indices) {
     return DynamicUpdateSlice.create(scope, input, update, indices);
   }
@@ -205,7 +205,7 @@ public <T extends Tensor, U extends Tensor & TNumber> DynamicUpdateSlice<T> dyna
    * @param equation
    * @return a new instance of Einsum
    */
-  public <T extends Tensor> Einsum<T> einsum(Operand<T> a, Operand<T> b, String equation) {
+  public <T extends TType> Einsum<T> einsum(Operand<T> a, Operand<T> b, String equation) {
     return Einsum.create(scope, a, b, equation);
   }
 
@@ -222,7 +222,7 @@ public <T extends Tensor> Einsum<T> einsum(Operand<T> a, Operand<T> b, String eq
    * @param indicesAreSorted Boolean indicating if the indices are sorted.
    * @return a new instance of Gather
    */
-  public <T extends Tensor, U extends Tensor & TNumber> Gather<T> gather(Operand<T> operand,
+  public <T extends TType, U extends TNumber> Gather<T> gather(Operand<T> operand,
       Operand<U> startIndices, Operand<U> sliceSizes, String dimensionNumbers,
       Boolean indicesAreSorted) {
     return Gather.create(scope, operand, startIndices, sliceSizes, dimensionNumbers, indicesAreSorted);
@@ -242,8 +242,8 @@ public <T extends Tensor, U extends Tensor & TNumber> Gather<T> gather(Operand<T
    * @param values A `Tensor` of type V.
    * @return a new instance of KeyValueSort
    */
-  public <T extends Tensor & TNumber, U extends Tensor> KeyValueSort<T, U> keyValueSort(
-      Operand<T> keys, Operand<U> values) {
+  public <T extends TNumber, U extends TType> KeyValueSort<T, U> keyValueSort(Operand<T> keys,
+      Operand<U> values) {
     return KeyValueSort.create(scope, keys, values);
   }
 
@@ -261,9 +261,8 @@ public <T extends Tensor & TNumber, U extends Tensor> KeyValueSort<T, U> keyValu
    * @param paddingInterior the padding to apply between each input element.
    * @return a new instance of Pad
    */
-  public <T extends Tensor, U extends Tensor & TNumber> Pad<T> pad(Operand<T> input,
-      Operand<T> paddingValue, Operand<U> paddingLow, Operand<U> paddingHigh,
-      Operand<U> paddingInterior) {
+  public <T extends TType, U extends TNumber> Pad<T> pad(Operand<T> input, Operand<T> paddingValue,
+      Operand<U> paddingLow, Operand<U> paddingHigh, Operand<U> paddingInterior) {
     return Pad.create(scope, input, paddingValue, paddingLow, paddingHigh, paddingInterior);
   }
 
@@ -279,7 +278,7 @@ public <T extends Tensor, U extends Tensor & TNumber> Pad<T> pad(Operand<T> inpu
    * @param shape The shape of the tensor.
    * @return a new instance of Recv
    */
-  public <T extends Tensor> Recv<T> recv(DataType<T> dtype, String tensorName, Shape shape) {
+  public <T extends TType> Recv<T> recv(DataType<T> dtype, String tensorName, Shape shape) {
     return Recv.create(scope, dtype, tensorName, shape);
   }
 
@@ -312,7 +311,7 @@ public ReplicaId replicaId() {
    * @param epsilon the tolerance ratio.
    * @return a new instance of SelfAdjointEig
    */
-  public <T extends Tensor> SelfAdjointEig<T> selfAdjointEig(Operand<T> a, Boolean lower,
+  public <T extends TType> SelfAdjointEig<T> selfAdjointEig(Operand<T> a, Boolean lower,
       Long maxIter, Float epsilon) {
     return SelfAdjointEig.create(scope, a, lower, maxIter, epsilon);
   }
@@ -327,7 +326,7 @@ public <T extends Tensor> SelfAdjointEig<T> selfAdjointEig(Operand<T> a, Boolean
    * @param tensorName A string key that identifies the channel.
    * @return a new instance of Send
    */
-  public <T extends Tensor> Send send(Operand<T> tensor, String tensorName) {
+  public <T extends TType> Send send(Operand<T> tensor, String tensorName) {
     return Send.create(scope, tensor, tensorName);
   }
 
@@ -338,7 +337,7 @@ public <T extends Tensor> Send send(Operand<T> tensor, String tensorName) {
    * @param input
    * @return a new instance of Sharding
    */
-  public <T extends Tensor> Sharding<T> sharding(Operand<T> input) {
+  public <T extends TType> Sharding<T> sharding(Operand<T> input) {
     return Sharding.create(scope, input);
   }
 
@@ -354,7 +353,7 @@ public <T extends Tensor> Sharding<T> sharding(Operand<T> input) {
    * @param input A `Tensor` of type T.
    * @return a new instance of Sort
    */
-  public <T extends Tensor> Sort<T> sort(Operand<T> input) {
+  public <T extends TType> Sort<T> sort(Operand<T> input) {
     return Sort.create(scope, input);
   }
 
@@ -376,7 +375,7 @@ public <T extends Tensor> Sort<T> sort(Operand<T> input) {
    * @param precisionConfig a serialized xla::PrecisionConfig proto.
    * @return a new instance of Svd
    */
-  public <T extends Tensor> Svd<T> svd(Operand<T> a, Long maxIter, Float epsilon,
+  public <T extends TType> Svd<T> svd(Operand<T> a, Long maxIter, Float epsilon,
       String precisionConfig) {
     return Svd.create(scope, a, maxIter, epsilon, precisionConfig);
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/audio/AudioSpectrogram.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/audio/AudioSpectrogram.java
index 28fdb7791bc..1e88316cdd1 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/audio/AudioSpectrogram.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/audio/AudioSpectrogram.java
@@ -93,7 +93,7 @@ private Options() {
   @Endpoint(describeByClass = true)
   public static AudioSpectrogram create(Scope scope, Operand<TFloat32> input, Long windowSize, Long stride, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("AudioSpectrogram", scope.makeOpName("AudioSpectrogram"));
-    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("window_size", windowSize);
     opBuilder.setAttr("stride", stride);
@@ -123,7 +123,7 @@ public Output<TFloat32> spectrogram() {
   }
   
   @Override
-  public Output<TFloat32> asOutput() {
+  public Output<TFloat32> asOutput(Scope scope) {
     return spectrogram;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/audio/DecodeWav.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/audio/DecodeWav.java
index ec80ffdaf7c..d4ed42012af 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/audio/DecodeWav.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/audio/DecodeWav.java
@@ -89,7 +89,7 @@ private Options() {
   @Endpoint(describeByClass = true)
   public static DecodeWav create(Scope scope, Operand<TString> contents, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("DecodeWav", scope.makeOpName("DecodeWav"));
-    opBuilder.addInput(contents.asOutput());
+    opBuilder.addInput(contents.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/audio/EncodeWav.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/audio/EncodeWav.java
index b1920318564..3a7fc674e51 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/audio/EncodeWav.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/audio/EncodeWav.java
@@ -54,8 +54,8 @@ public final class EncodeWav extends RawOp implements Operand<TString> {
   @Endpoint(describeByClass = true)
   public static EncodeWav create(Scope scope, Operand<TFloat32> audio, Operand<TInt32> sampleRate) {
     OperationBuilder opBuilder = scope.env().opBuilder("EncodeWav", scope.makeOpName("EncodeWav"));
-    opBuilder.addInput(audio.asOutput());
-    opBuilder.addInput(sampleRate.asOutput());
+    opBuilder.addInput(audio.asOutput(scope));
+    opBuilder.addInput(sampleRate.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new EncodeWav(opBuilder.build());
   }
@@ -68,7 +68,7 @@ public Output<TString> contents() {
   }
   
   @Override
-  public Output<TString> asOutput() {
+  public Output<TString> asOutput(Scope scope) {
     return contents;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/audio/Mfcc.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/audio/Mfcc.java
index 07ab4a13215..1bf6eaca34d 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/audio/Mfcc.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/audio/Mfcc.java
@@ -102,8 +102,8 @@ private Options() {
   @Endpoint(describeByClass = true)
   public static Mfcc create(Scope scope, Operand<TFloat32> spectrogram, Operand<TInt32> sampleRate, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("Mfcc", scope.makeOpName("Mfcc"));
-    opBuilder.addInput(spectrogram.asOutput());
-    opBuilder.addInput(sampleRate.asOutput());
+    opBuilder.addInput(spectrogram.asOutput(scope));
+    opBuilder.addInput(sampleRate.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -161,7 +161,7 @@ public Output<TFloat32> output() {
   }
   
   @Override
-  public Output<TFloat32> asOutput() {
+  public Output<TFloat32> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/bitwise/BitwiseAnd.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/bitwise/BitwiseAnd.java
index b4e37b6015e..99ef38701b5 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/bitwise/BitwiseAnd.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/bitwise/BitwiseAnd.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -54,7 +53,7 @@
  * @param <T> data type for {@code z()} output
  */
 @Operator(group = "bitwise")
-public final class BitwiseAnd<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class BitwiseAnd<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new BitwiseAnd operation.
@@ -65,10 +64,10 @@ public final class BitwiseAnd<T extends Tensor & TNumber> extends RawOp implemen
    * @return a new instance of BitwiseAnd
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> BitwiseAnd<T> create(Scope scope, Operand<T> x, Operand<T> y) {
+  public static <T extends TNumber> BitwiseAnd<T> create(Scope scope, Operand<T> x, Operand<T> y) {
     OperationBuilder opBuilder = scope.env().opBuilder("BitwiseAnd", scope.makeOpName("BitwiseAnd"));
-    opBuilder.addInput(x.asOutput());
-    opBuilder.addInput(y.asOutput());
+    opBuilder.addInput(x.asOutput(scope));
+    opBuilder.addInput(y.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new BitwiseAnd<T>(opBuilder.build());
   }
@@ -80,7 +79,7 @@ public Output<T> z() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return z;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/bitwise/BitwiseOr.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/bitwise/BitwiseOr.java
index 8440f0a46b1..5532fc43f65 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/bitwise/BitwiseOr.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/bitwise/BitwiseOr.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -54,7 +53,7 @@
  * @param <T> data type for {@code z()} output
  */
 @Operator(group = "bitwise")
-public final class BitwiseOr<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class BitwiseOr<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new BitwiseOr operation.
@@ -65,10 +64,10 @@ public final class BitwiseOr<T extends Tensor & TNumber> extends RawOp implement
    * @return a new instance of BitwiseOr
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> BitwiseOr<T> create(Scope scope, Operand<T> x, Operand<T> y) {
+  public static <T extends TNumber> BitwiseOr<T> create(Scope scope, Operand<T> x, Operand<T> y) {
     OperationBuilder opBuilder = scope.env().opBuilder("BitwiseOr", scope.makeOpName("BitwiseOr"));
-    opBuilder.addInput(x.asOutput());
-    opBuilder.addInput(y.asOutput());
+    opBuilder.addInput(x.asOutput(scope));
+    opBuilder.addInput(y.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new BitwiseOr<T>(opBuilder.build());
   }
@@ -80,7 +79,7 @@ public Output<T> z() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return z;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/bitwise/BitwiseXor.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/bitwise/BitwiseXor.java
index c852846bc82..624cd8a1df3 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/bitwise/BitwiseXor.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/bitwise/BitwiseXor.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -54,7 +53,7 @@
  * @param <T> data type for {@code z()} output
  */
 @Operator(group = "bitwise")
-public final class BitwiseXor<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class BitwiseXor<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new BitwiseXor operation.
@@ -65,10 +64,10 @@ public final class BitwiseXor<T extends Tensor & TNumber> extends RawOp implemen
    * @return a new instance of BitwiseXor
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> BitwiseXor<T> create(Scope scope, Operand<T> x, Operand<T> y) {
+  public static <T extends TNumber> BitwiseXor<T> create(Scope scope, Operand<T> x, Operand<T> y) {
     OperationBuilder opBuilder = scope.env().opBuilder("BitwiseXor", scope.makeOpName("BitwiseXor"));
-    opBuilder.addInput(x.asOutput());
-    opBuilder.addInput(y.asOutput());
+    opBuilder.addInput(x.asOutput(scope));
+    opBuilder.addInput(y.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new BitwiseXor<T>(opBuilder.build());
   }
@@ -80,7 +79,7 @@ public Output<T> z() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return z;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/bitwise/Invert.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/bitwise/Invert.java
index 8cf61d86110..c0f882d0791 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/bitwise/Invert.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/bitwise/Invert.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -75,7 +74,7 @@
  * @param <T> data type for {@code y()} output
  */
 @Operator(group = "bitwise")
-public final class Invert<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class Invert<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Invert operation.
@@ -85,9 +84,9 @@ public final class Invert<T extends Tensor & TNumber> extends RawOp implements O
    * @return a new instance of Invert
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> Invert<T> create(Scope scope, Operand<T> x) {
+  public static <T extends TNumber> Invert<T> create(Scope scope, Operand<T> x) {
     OperationBuilder opBuilder = scope.env().opBuilder("Invert", scope.makeOpName("Invert"));
-    opBuilder.addInput(x.asOutput());
+    opBuilder.addInput(x.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new Invert<T>(opBuilder.build());
   }
@@ -99,7 +98,7 @@ public Output<T> y() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return y;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/bitwise/LeftShift.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/bitwise/LeftShift.java
index 1fa02ba6a51..3de00d2b037 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/bitwise/LeftShift.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/bitwise/LeftShift.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -65,7 +64,7 @@
  * @param <T> data type for {@code z()} output
  */
 @Operator(group = "bitwise")
-public final class LeftShift<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class LeftShift<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new LeftShift operation.
@@ -76,10 +75,10 @@ public final class LeftShift<T extends Tensor & TNumber> extends RawOp implement
    * @return a new instance of LeftShift
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> LeftShift<T> create(Scope scope, Operand<T> x, Operand<T> y) {
+  public static <T extends TNumber> LeftShift<T> create(Scope scope, Operand<T> x, Operand<T> y) {
     OperationBuilder opBuilder = scope.env().opBuilder("LeftShift", scope.makeOpName("LeftShift"));
-    opBuilder.addInput(x.asOutput());
-    opBuilder.addInput(y.asOutput());
+    opBuilder.addInput(x.asOutput(scope));
+    opBuilder.addInput(y.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new LeftShift<T>(opBuilder.build());
   }
@@ -91,7 +90,7 @@ public Output<T> z() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return z;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/bitwise/RightShift.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/bitwise/RightShift.java
index 2bd25ff8ad8..00117180819 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/bitwise/RightShift.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/bitwise/RightShift.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -68,7 +67,7 @@
  * @param <T> data type for {@code z()} output
  */
 @Operator(group = "bitwise")
-public final class RightShift<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class RightShift<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new RightShift operation.
@@ -79,10 +78,10 @@ public final class RightShift<T extends Tensor & TNumber> extends RawOp implemen
    * @return a new instance of RightShift
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> RightShift<T> create(Scope scope, Operand<T> x, Operand<T> y) {
+  public static <T extends TNumber> RightShift<T> create(Scope scope, Operand<T> x, Operand<T> y) {
     OperationBuilder opBuilder = scope.env().opBuilder("RightShift", scope.makeOpName("RightShift"));
-    opBuilder.addInput(x.asOutput());
-    opBuilder.addInput(y.asOutput());
+    opBuilder.addInput(x.asOutput(scope));
+    opBuilder.addInput(y.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new RightShift<T>(opBuilder.build());
   }
@@ -94,7 +93,7 @@ public Output<T> z() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return z;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/cluster/KMC2ChainInitialization.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/cluster/KMC2ChainInitialization.java
index 272695dad5c..4b8a4723e1a 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/cluster/KMC2ChainInitialization.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/cluster/KMC2ChainInitialization.java
@@ -50,8 +50,8 @@ public final class KMC2ChainInitialization extends RawOp implements Operand<TInt
   @Endpoint(describeByClass = true)
   public static KMC2ChainInitialization create(Scope scope, Operand<TFloat32> distances, Operand<TInt64> seed) {
     OperationBuilder opBuilder = scope.env().opBuilder("KMC2ChainInitialization", scope.makeOpName("KMC2ChainInitialization"));
-    opBuilder.addInput(distances.asOutput());
-    opBuilder.addInput(seed.asOutput());
+    opBuilder.addInput(distances.asOutput(scope));
+    opBuilder.addInput(seed.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new KMC2ChainInitialization(opBuilder.build());
   }
@@ -64,7 +64,7 @@ public Output<TInt64> index() {
   }
   
   @Override
-  public Output<TInt64> asOutput() {
+  public Output<TInt64> asOutput(Scope scope) {
     return index;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/cluster/KmeansPlusPlusInitialization.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/cluster/KmeansPlusPlusInitialization.java
index 63d8ef01222..7f142468754 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/cluster/KmeansPlusPlusInitialization.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/cluster/KmeansPlusPlusInitialization.java
@@ -54,10 +54,10 @@ public final class KmeansPlusPlusInitialization extends RawOp implements Operand
   @Endpoint(describeByClass = true)
   public static KmeansPlusPlusInitialization create(Scope scope, Operand<TFloat32> points, Operand<TInt64> numToSample, Operand<TInt64> seed, Operand<TInt64> numRetriesPerSample) {
     OperationBuilder opBuilder = scope.env().opBuilder("KmeansPlusPlusInitialization", scope.makeOpName("KmeansPlusPlusInitialization"));
-    opBuilder.addInput(points.asOutput());
-    opBuilder.addInput(numToSample.asOutput());
-    opBuilder.addInput(seed.asOutput());
-    opBuilder.addInput(numRetriesPerSample.asOutput());
+    opBuilder.addInput(points.asOutput(scope));
+    opBuilder.addInput(numToSample.asOutput(scope));
+    opBuilder.addInput(seed.asOutput(scope));
+    opBuilder.addInput(numRetriesPerSample.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new KmeansPlusPlusInitialization(opBuilder.build());
   }
@@ -70,7 +70,7 @@ public Output<TFloat32> samples() {
   }
   
   @Override
-  public Output<TFloat32> asOutput() {
+  public Output<TFloat32> asOutput(Scope scope) {
     return samples;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/collective/AllReduce.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/collective/AllReduce.java
index 146533646cd..6c832043526 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/collective/AllReduce.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/collective/AllReduce.java
@@ -22,7 +22,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -34,7 +33,7 @@
  * 
  * @param <T> data type for {@code output()} output
  */
-public final class AllReduce<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class AllReduce<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.collective.AllReduce}
@@ -88,9 +87,9 @@ private Options() {
    * @return a new instance of AllReduce
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> AllReduce<T> create(Scope scope, Operand<T> input, Long groupSize, Long groupKey, Long instanceKey, String mergeOp, String finalOp, List<Long> subdivOffsets, Options... options) {
+  public static <T extends TNumber> AllReduce<T> create(Scope scope, Operand<T> input, Long groupSize, Long groupKey, Long instanceKey, String mergeOp, String finalOp, List<Long> subdivOffsets, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("CollectiveReduce", scope.makeOpName("AllReduce"));
-    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("group_size", groupSize);
     opBuilder.setAttr("group_key", groupKey);
@@ -150,7 +149,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/collective/BroadcastRecv.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/collective/BroadcastRecv.java
index cdd41fd3b3e..ba1adab34f3 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/collective/BroadcastRecv.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/collective/BroadcastRecv.java
@@ -22,19 +22,19 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Receives a tensor value broadcast from another device.
  * 
  * @param <T> data type for {@code output()} output
  */
-public final class BroadcastRecv<T extends Tensor> extends RawOp implements Operand<T> {
+public final class BroadcastRecv<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.collective.BroadcastRecv}
@@ -77,7 +77,7 @@ private Options() {
    * @return a new instance of BroadcastRecv
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> BroadcastRecv<T> create(Scope scope, DataType<T> T, Long groupSize, Long groupKey, Long instanceKey, Shape shape, Options... options) {
+  public static <T extends TType> BroadcastRecv<T> create(Scope scope, DataType<T> T, Long groupSize, Long groupKey, Long instanceKey, Shape shape, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("CollectiveBcastRecv", scope.makeOpName("BroadcastRecv"));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("T", T);
@@ -119,7 +119,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/collective/BroadcastSend.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/collective/BroadcastSend.java
index 5ebff67f14b..ad95ac7a58b 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/collective/BroadcastSend.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/collective/BroadcastSend.java
@@ -21,19 +21,19 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Broadcasts a tensor value to one or more other devices.
  * 
  * @param <T> data type for {@code output()} output
  */
-public final class BroadcastSend<T extends Tensor> extends RawOp implements Operand<T> {
+public final class BroadcastSend<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.collective.BroadcastSend}
@@ -76,9 +76,9 @@ private Options() {
    * @return a new instance of BroadcastSend
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> BroadcastSend<T> create(Scope scope, Operand<T> input, Long groupSize, Long groupKey, Long instanceKey, Shape shape, Options... options) {
+  public static <T extends TType> BroadcastSend<T> create(Scope scope, Operand<T> input, Long groupSize, Long groupKey, Long instanceKey, Shape shape, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("CollectiveBcastSend", scope.makeOpName("BroadcastSend"));
-    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("group_size", groupSize);
     opBuilder.setAttr("group_key", groupKey);
@@ -118,7 +118,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/All.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/All.java
index 2e9bf958757..ae82b78ad75 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/All.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/All.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -70,10 +69,10 @@ private Options() {
    * @return a new instance of All
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> All create(Scope scope, Operand<TBool> input, Operand<T> axis, Options... options) {
+  public static <T extends TNumber> All create(Scope scope, Operand<TBool> input, Operand<T> axis, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("All", scope.makeOpName("All"));
-    opBuilder.addInput(input.asOutput());
-    opBuilder.addInput(axis.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
+    opBuilder.addInput(axis.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -100,7 +99,7 @@ public Output<TBool> output() {
   }
   
   @Override
-  public Output<TBool> asOutput() {
+  public Output<TBool> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Any.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Any.java
index 51cc483fb1d..1be811c16f0 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Any.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Any.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -70,10 +69,10 @@ private Options() {
    * @return a new instance of Any
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> Any create(Scope scope, Operand<TBool> input, Operand<T> axis, Options... options) {
+  public static <T extends TNumber> Any create(Scope scope, Operand<TBool> input, Operand<T> axis, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("Any", scope.makeOpName("Any"));
-    opBuilder.addInput(input.asOutput());
-    opBuilder.addInput(axis.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
+    opBuilder.addInput(axis.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -100,7 +99,7 @@ public Output<TBool> output() {
   }
   
   @Override
-  public Output<TBool> asOutput() {
+  public Output<TBool> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/AssertThat.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/AssertThat.java
index dce70c04e5a..19bb1cda9d8 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/AssertThat.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/AssertThat.java
@@ -67,8 +67,8 @@ private Options() {
   @Endpoint(describeByClass = true)
   public static AssertThat create(Scope scope, Operand<TBool> condition, Iterable<Operand<?>> data, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("Assert", scope.makeOpName("AssertThat"));
-    opBuilder.addInput(condition.asOutput());
-    opBuilder.addInputList(Operands.asOutputs(data));
+    opBuilder.addInput(condition.asOutput(scope));
+    opBuilder.addInputList(Operands.asOutputs(scope, data));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Assign.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Assign.java
index 021c6880596..f0b5b6174c3 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Assign.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Assign.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Update 'ref' by assigning 'value' to it.
@@ -36,7 +36,7 @@
  * @param <T> data type for {@code outputRef()} output
  */
 @Operator
-public final class Assign<T extends Tensor> extends RawOp implements Operand<T> {
+public final class Assign<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.core.Assign}
@@ -79,10 +79,10 @@ private Options() {
    * @return a new instance of Assign
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> Assign<T> create(Scope scope, Operand<T> ref, Operand<T> value, Options... options) {
+  public static <T extends TType> Assign<T> create(Scope scope, Operand<T> ref, Operand<T> value, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("Assign", scope.makeOpName("Assign"));
-    opBuilder.addInput(ref.asOutput());
-    opBuilder.addInput(value.asOutput());
+    opBuilder.addInput(ref.asOutput(scope));
+    opBuilder.addInput(value.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -123,7 +123,7 @@ public Output<T> outputRef() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return outputRef;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/AssignAdd.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/AssignAdd.java
index c64a1924c42..8c9052a4672 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/AssignAdd.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/AssignAdd.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Update 'ref' by adding 'value' to it.
@@ -36,7 +36,7 @@
  * @param <T> data type for {@code outputRef()} output
  */
 @Operator
-public final class AssignAdd<T extends Tensor> extends RawOp implements Operand<T> {
+public final class AssignAdd<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.core.AssignAdd}
@@ -68,10 +68,10 @@ private Options() {
    * @return a new instance of AssignAdd
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> AssignAdd<T> create(Scope scope, Operand<T> ref, Operand<T> value, Options... options) {
+  public static <T extends TType> AssignAdd<T> create(Scope scope, Operand<T> ref, Operand<T> value, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("AssignAdd", scope.makeOpName("AssignAdd"));
-    opBuilder.addInput(ref.asOutput());
-    opBuilder.addInput(value.asOutput());
+    opBuilder.addInput(ref.asOutput(scope));
+    opBuilder.addInput(value.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -100,7 +100,7 @@ public Output<T> outputRef() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return outputRef;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/AssignAddVariableOp.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/AssignAddVariableOp.java
index 36c43524b12..641f8a435eb 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/AssignAddVariableOp.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/AssignAddVariableOp.java
@@ -20,11 +20,11 @@
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Adds a value to the current value of a variable.
@@ -44,10 +44,10 @@ public final class AssignAddVariableOp extends RawOp {
    * @return a new instance of AssignAddVariableOp
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> AssignAddVariableOp create(Scope scope, Operand<?> resource, Operand<T> value) {
+  public static <T extends TType> AssignAddVariableOp create(Scope scope, Operand<?> resource, Operand<T> value) {
     OperationBuilder opBuilder = scope.env().opBuilder("AssignAddVariableOp", scope.makeOpName("AssignAddVariableOp"));
-    opBuilder.addInput(resource.asOutput());
-    opBuilder.addInput(value.asOutput());
+    opBuilder.addInput(resource.asOutput(scope));
+    opBuilder.addInput(value.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new AssignAddVariableOp(opBuilder.build());
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/AssignSub.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/AssignSub.java
index 494f1dac6bb..0aec7a7afac 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/AssignSub.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/AssignSub.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Update 'ref' by subtracting 'value' from it.
@@ -36,7 +36,7 @@
  * @param <T> data type for {@code outputRef()} output
  */
 @Operator
-public final class AssignSub<T extends Tensor> extends RawOp implements Operand<T> {
+public final class AssignSub<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.core.AssignSub}
@@ -68,10 +68,10 @@ private Options() {
    * @return a new instance of AssignSub
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> AssignSub<T> create(Scope scope, Operand<T> ref, Operand<T> value, Options... options) {
+  public static <T extends TType> AssignSub<T> create(Scope scope, Operand<T> ref, Operand<T> value, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("AssignSub", scope.makeOpName("AssignSub"));
-    opBuilder.addInput(ref.asOutput());
-    opBuilder.addInput(value.asOutput());
+    opBuilder.addInput(ref.asOutput(scope));
+    opBuilder.addInput(value.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -100,7 +100,7 @@ public Output<T> outputRef() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return outputRef;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/AssignSubVariableOp.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/AssignSubVariableOp.java
index c43611ebbba..44ef8585450 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/AssignSubVariableOp.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/AssignSubVariableOp.java
@@ -20,11 +20,11 @@
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Subtracts a value from the current value of a variable.
@@ -44,10 +44,10 @@ public final class AssignSubVariableOp extends RawOp {
    * @return a new instance of AssignSubVariableOp
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> AssignSubVariableOp create(Scope scope, Operand<?> resource, Operand<T> value) {
+  public static <T extends TType> AssignSubVariableOp create(Scope scope, Operand<?> resource, Operand<T> value) {
     OperationBuilder opBuilder = scope.env().opBuilder("AssignSubVariableOp", scope.makeOpName("AssignSubVariableOp"));
-    opBuilder.addInput(resource.asOutput());
-    opBuilder.addInput(value.asOutput());
+    opBuilder.addInput(resource.asOutput(scope));
+    opBuilder.addInput(value.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new AssignSubVariableOp(opBuilder.build());
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/AssignVariableOp.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/AssignVariableOp.java
index 3ffb2154021..581f980f517 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/AssignVariableOp.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/AssignVariableOp.java
@@ -20,11 +20,11 @@
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Assigns a new value to a variable.
@@ -44,10 +44,10 @@ public final class AssignVariableOp extends RawOp {
    * @return a new instance of AssignVariableOp
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> AssignVariableOp create(Scope scope, Operand<?> resource, Operand<T> value) {
+  public static <T extends TType> AssignVariableOp create(Scope scope, Operand<?> resource, Operand<T> value) {
     OperationBuilder opBuilder = scope.env().opBuilder("AssignVariableOp", scope.makeOpName("AssignVariableOp"));
-    opBuilder.addInput(resource.asOutput());
-    opBuilder.addInput(value.asOutput());
+    opBuilder.addInput(resource.asOutput(scope));
+    opBuilder.addInput(value.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new AssignVariableOp(opBuilder.build());
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Barrier.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Barrier.java
index b9c5c84083f..4fcb8ef8f21 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Barrier.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Barrier.java
@@ -177,7 +177,7 @@ public Output<TString> handle() {
   }
   
   @Override
-  public Output<TString> asOutput() {
+  public Output<TString> asOutput(Scope scope) {
     return handle;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BarrierClose.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BarrierClose.java
index 514f4f50edf..e4ec2900d0a 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BarrierClose.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BarrierClose.java
@@ -71,7 +71,7 @@ private Options() {
   @Endpoint(describeByClass = true)
   public static BarrierClose create(Scope scope, Operand<TString> handle, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("BarrierClose", scope.makeOpName("BarrierClose"));
-    opBuilder.addInput(handle.asOutput());
+    opBuilder.addInput(handle.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BarrierIncompleteSize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BarrierIncompleteSize.java
index 72dbe1533d6..f4c404030d9 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BarrierIncompleteSize.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BarrierIncompleteSize.java
@@ -44,7 +44,7 @@ public final class BarrierIncompleteSize extends RawOp implements Operand<TInt32
   @Endpoint(describeByClass = true)
   public static BarrierIncompleteSize create(Scope scope, Operand<TString> handle) {
     OperationBuilder opBuilder = scope.env().opBuilder("BarrierIncompleteSize", scope.makeOpName("BarrierIncompleteSize"));
-    opBuilder.addInput(handle.asOutput());
+    opBuilder.addInput(handle.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new BarrierIncompleteSize(opBuilder.build());
   }
@@ -58,7 +58,7 @@ public Output<TInt32> size() {
   }
   
   @Override
-  public Output<TInt32> asOutput() {
+  public Output<TInt32> asOutput(Scope scope) {
     return size;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BarrierInsertMany.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BarrierInsertMany.java
index 272e6e1afe4..85ee4c294cf 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BarrierInsertMany.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BarrierInsertMany.java
@@ -20,12 +20,12 @@
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TString;
+import org.tensorflow.types.family.TType;
 
 /**
  * For each key, assigns the respective value to the specified component.
@@ -50,11 +50,11 @@ public final class BarrierInsertMany extends RawOp {
    * @return a new instance of BarrierInsertMany
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> BarrierInsertMany create(Scope scope, Operand<TString> handle, Operand<TString> keys, Operand<T> values, Long componentIndex) {
+  public static <T extends TType> BarrierInsertMany create(Scope scope, Operand<TString> handle, Operand<TString> keys, Operand<T> values, Long componentIndex) {
     OperationBuilder opBuilder = scope.env().opBuilder("BarrierInsertMany", scope.makeOpName("BarrierInsertMany"));
-    opBuilder.addInput(handle.asOutput());
-    opBuilder.addInput(keys.asOutput());
-    opBuilder.addInput(values.asOutput());
+    opBuilder.addInput(handle.asOutput(scope));
+    opBuilder.addInput(keys.asOutput(scope));
+    opBuilder.addInput(values.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("component_index", componentIndex);
     return new BarrierInsertMany(opBuilder.build());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BarrierReadySize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BarrierReadySize.java
index 1cb29f97e1d..0eda2364bca 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BarrierReadySize.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BarrierReadySize.java
@@ -44,7 +44,7 @@ public final class BarrierReadySize extends RawOp implements Operand<TInt32> {
   @Endpoint(describeByClass = true)
   public static BarrierReadySize create(Scope scope, Operand<TString> handle) {
     OperationBuilder opBuilder = scope.env().opBuilder("BarrierReadySize", scope.makeOpName("BarrierReadySize"));
-    opBuilder.addInput(handle.asOutput());
+    opBuilder.addInput(handle.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new BarrierReadySize(opBuilder.build());
   }
@@ -58,7 +58,7 @@ public Output<TInt32> size() {
   }
   
   @Override
-  public Output<TInt32> asOutput() {
+  public Output<TInt32> asOutput(Scope scope) {
     return size;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BarrierTakeMany.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BarrierTakeMany.java
index 6c391fab5fa..c5a22ae6027 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BarrierTakeMany.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BarrierTakeMany.java
@@ -100,8 +100,8 @@ private Options() {
   @Endpoint(describeByClass = true)
   public static BarrierTakeMany create(Scope scope, Operand<TString> handle, Operand<TInt32> numElements, List<DataType<?>> componentTypes, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("BarrierTakeMany", scope.makeOpName("BarrierTakeMany"));
-    opBuilder.addInput(handle.asOutput());
-    opBuilder.addInput(numElements.asOutput());
+    opBuilder.addInput(handle.asOutput(scope));
+    opBuilder.addInput(numElements.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     DataType[] componentTypesArray = new DataType[componentTypes.size()];
     for (int i = 0; i < componentTypesArray.length; ++i) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Batch.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Batch.java
index 21d7cf0a6c0..b8b418f978b 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Batch.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Batch.java
@@ -143,7 +143,7 @@ private Options() {
   @Endpoint(describeByClass = true)
   public static Batch create(Scope scope, Iterable<Operand<?>> inTensors, Long numBatchThreads, Long maxBatchSize, Long batchTimeoutMicros, Long gradTimeoutMicros, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("Batch", scope.makeOpName("Batch"));
-    opBuilder.addInputList(Operands.asOutputs(inTensors));
+    opBuilder.addInputList(Operands.asOutputs(scope, inTensors));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("num_batch_threads", numBatchThreads);
     opBuilder.setAttr("max_batch_size", maxBatchSize);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BatchToSpace.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BatchToSpace.java
index 1d7356e3b95..5bd9366d82d 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BatchToSpace.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BatchToSpace.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * BatchToSpace for 4-D tensors of type T.
@@ -42,7 +42,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator
-public final class BatchToSpace<T extends Tensor> extends RawOp implements Operand<T> {
+public final class BatchToSpace<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new BatchToSpace operation.
@@ -61,10 +61,10 @@ public final class BatchToSpace<T extends Tensor> extends RawOp implements Opera
    * @return a new instance of BatchToSpace
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor, U extends Tensor & TNumber> BatchToSpace<T> create(Scope scope, Operand<T> input, Operand<U> crops, Long blockSize) {
+  public static <T extends TType, U extends TNumber> BatchToSpace<T> create(Scope scope, Operand<T> input, Operand<U> crops, Long blockSize) {
     OperationBuilder opBuilder = scope.env().opBuilder("BatchToSpace", scope.makeOpName("BatchToSpace"));
-    opBuilder.addInput(input.asOutput());
-    opBuilder.addInput(crops.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
+    opBuilder.addInput(crops.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("block_size", blockSize);
     return new BatchToSpace<T>(opBuilder.build());
@@ -130,7 +130,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BatchToSpaceNd.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BatchToSpaceNd.java
index 4396bd919c8..ae991e8bf52 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BatchToSpaceNd.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BatchToSpaceNd.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * BatchToSpace for N-D tensors of type T.
@@ -41,7 +41,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator
-public final class BatchToSpaceNd<T extends Tensor> extends RawOp implements Operand<T> {
+public final class BatchToSpaceNd<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new BatchToSpaceNd operation.
@@ -148,11 +148,11 @@ public final class BatchToSpaceNd<T extends Tensor> extends RawOp implements Ope
    * @return a new instance of BatchToSpaceNd
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor, U extends Tensor & TNumber, V extends Tensor & TNumber> BatchToSpaceNd<T> create(Scope scope, Operand<T> input, Operand<U> blockShape, Operand<V> crops) {
+  public static <T extends TType, U extends TNumber, V extends TNumber> BatchToSpaceNd<T> create(Scope scope, Operand<T> input, Operand<U> blockShape, Operand<V> crops) {
     OperationBuilder opBuilder = scope.env().opBuilder("BatchToSpaceND", scope.makeOpName("BatchToSpaceNd"));
-    opBuilder.addInput(input.asOutput());
-    opBuilder.addInput(blockShape.asOutput());
-    opBuilder.addInput(crops.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
+    opBuilder.addInput(blockShape.asOutput(scope));
+    opBuilder.addInput(crops.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new BatchToSpaceNd<T>(opBuilder.build());
   }
@@ -164,7 +164,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Bitcast.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Bitcast.java
index da3b8569073..37391484e43 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Bitcast.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Bitcast.java
@@ -22,11 +22,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Bitcasts a tensor from one type to another without copying data.
@@ -85,7 +85,7 @@
  * @param <U> data type for {@code output()} output
  */
 @Operator
-public final class Bitcast<U extends Tensor> extends RawOp implements Operand<U> {
+public final class Bitcast<U extends TType> extends RawOp implements Operand<U> {
   
   /**
    * Factory method to create a class wrapping a new Bitcast operation.
@@ -96,9 +96,9 @@ public final class Bitcast<U extends Tensor> extends RawOp implements Operand<U>
    * @return a new instance of Bitcast
    */
   @Endpoint(describeByClass = true)
-  public static <U extends Tensor, T extends Tensor> Bitcast<U> create(Scope scope, Operand<T> input, DataType<U> type) {
+  public static <U extends TType, T extends TType> Bitcast<U> create(Scope scope, Operand<T> input, DataType<U> type) {
     OperationBuilder opBuilder = scope.env().opBuilder("Bitcast", scope.makeOpName("Bitcast"));
-    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("type", type);
     return new Bitcast<U>(opBuilder.build());
@@ -111,7 +111,7 @@ public Output<U> output() {
   }
   
   @Override
-  public Output<U> asOutput() {
+  public Output<U> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BroadcastDynamicShape.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BroadcastDynamicShape.java
index d0e6dd8601e..51e8295f2b3 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BroadcastDynamicShape.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BroadcastDynamicShape.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -37,7 +36,7 @@
  * @param <T> data type for {@code r0()} output
  */
 @Operator
-public final class BroadcastDynamicShape<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class BroadcastDynamicShape<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new BroadcastDynamicShape operation.
@@ -48,10 +47,10 @@ public final class BroadcastDynamicShape<T extends Tensor & TNumber> extends Raw
    * @return a new instance of BroadcastDynamicShape
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> BroadcastDynamicShape<T> create(Scope scope, Operand<T> s0, Operand<T> s1) {
+  public static <T extends TNumber> BroadcastDynamicShape<T> create(Scope scope, Operand<T> s0, Operand<T> s1) {
     OperationBuilder opBuilder = scope.env().opBuilder("BroadcastArgs", scope.makeOpName("BroadcastDynamicShape"));
-    opBuilder.addInput(s0.asOutput());
-    opBuilder.addInput(s1.asOutput());
+    opBuilder.addInput(s0.asOutput(scope));
+    opBuilder.addInput(s1.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new BroadcastDynamicShape<T>(opBuilder.build());
   }
@@ -63,7 +62,7 @@ public Output<T> r0() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return r0;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BroadcastGradientArgs.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BroadcastGradientArgs.java
index e604ac4f87c..d5bbded593e 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BroadcastGradientArgs.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BroadcastGradientArgs.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -35,7 +34,7 @@
  * 
  * @param <T> data type for {@code r0()} output
  */
-public final class BroadcastGradientArgs<T extends Tensor & TNumber> extends RawOp {
+public final class BroadcastGradientArgs<T extends TNumber> extends RawOp {
   
   /**
    * Factory method to create a class wrapping a new BroadcastGradientArgs operation.
@@ -46,10 +45,10 @@ public final class BroadcastGradientArgs<T extends Tensor & TNumber> extends Raw
    * @return a new instance of BroadcastGradientArgs
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> BroadcastGradientArgs<T> create(Scope scope, Operand<T> s0, Operand<T> s1) {
+  public static <T extends TNumber> BroadcastGradientArgs<T> create(Scope scope, Operand<T> s0, Operand<T> s1) {
     OperationBuilder opBuilder = scope.env().opBuilder("BroadcastGradientArgs", scope.makeOpName("BroadcastGradientArgs"));
-    opBuilder.addInput(s0.asOutput());
-    opBuilder.addInput(s1.asOutput());
+    opBuilder.addInput(s0.asOutput(scope));
+    opBuilder.addInput(s1.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new BroadcastGradientArgs<T>(opBuilder.build());
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BroadcastTo.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BroadcastTo.java
index 31e6a0d4216..b99bbb51cf5 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BroadcastTo.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BroadcastTo.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * Broadcast an array for a compatible shape.
@@ -62,7 +62,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator
-public final class BroadcastTo<T extends Tensor> extends RawOp implements Operand<T> {
+public final class BroadcastTo<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new BroadcastTo operation.
@@ -73,10 +73,10 @@ public final class BroadcastTo<T extends Tensor> extends RawOp implements Operan
    * @return a new instance of BroadcastTo
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor, U extends Tensor & TNumber> BroadcastTo<T> create(Scope scope, Operand<T> input, Operand<U> shape) {
+  public static <T extends TType, U extends TNumber> BroadcastTo<T> create(Scope scope, Operand<T> input, Operand<U> shape) {
     OperationBuilder opBuilder = scope.env().opBuilder("BroadcastTo", scope.makeOpName("BroadcastTo"));
-    opBuilder.addInput(input.asOutput());
-    opBuilder.addInput(shape.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
+    opBuilder.addInput(shape.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new BroadcastTo<T>(opBuilder.build());
   }
@@ -89,7 +89,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Bucketize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Bucketize.java
index 8c69988030d..ee6a3660c37 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Bucketize.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Bucketize.java
@@ -22,7 +22,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -56,9 +55,9 @@ public final class Bucketize extends RawOp implements Operand<TInt32> {
    * @return a new instance of Bucketize
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> Bucketize create(Scope scope, Operand<T> input, List<Float> boundaries) {
+  public static <T extends TNumber> Bucketize create(Scope scope, Operand<T> input, List<Float> boundaries) {
     OperationBuilder opBuilder = scope.env().opBuilder("Bucketize", scope.makeOpName("Bucketize"));
-    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     float[] boundariesArray = new float[boundaries.size()];
     for (int i = 0; i < boundariesArray.length; ++i) {
@@ -80,7 +79,7 @@ public Output<TInt32> output() {
   }
   
   @Override
-  public Output<TInt32> asOutput() {
+  public Output<TInt32> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ClipByValue.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ClipByValue.java
index 43bff94ad09..7d4859664ab 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ClipByValue.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ClipByValue.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Clips tensor values to a specified min and max.
@@ -38,7 +38,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator
-public final class ClipByValue<T extends Tensor> extends RawOp implements Operand<T> {
+public final class ClipByValue<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new ClipByValue operation.
@@ -52,11 +52,11 @@ public final class ClipByValue<T extends Tensor> extends RawOp implements Operan
    * @return a new instance of ClipByValue
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> ClipByValue<T> create(Scope scope, Operand<T> t, Operand<T> clipValueMin, Operand<T> clipValueMax) {
+  public static <T extends TType> ClipByValue<T> create(Scope scope, Operand<T> t, Operand<T> clipValueMin, Operand<T> clipValueMax) {
     OperationBuilder opBuilder = scope.env().opBuilder("ClipByValue", scope.makeOpName("ClipByValue"));
-    opBuilder.addInput(t.asOutput());
-    opBuilder.addInput(clipValueMin.asOutput());
-    opBuilder.addInput(clipValueMax.asOutput());
+    opBuilder.addInput(t.asOutput(scope));
+    opBuilder.addInput(clipValueMin.asOutput(scope));
+    opBuilder.addInput(clipValueMax.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new ClipByValue<T>(opBuilder.build());
   }
@@ -69,7 +69,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/CollectiveGather.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/CollectiveGather.java
index 6683a0822a7..2b504684e85 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/CollectiveGather.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/CollectiveGather.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
@@ -34,7 +33,7 @@
  * 
  * @param <T> data type for {@code output()} output
  */
-public final class CollectiveGather<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class CollectiveGather<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.core.CollectiveGather}
@@ -77,9 +76,9 @@ private Options() {
    * @return a new instance of CollectiveGather
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> CollectiveGather<T> create(Scope scope, Operand<T> input, Long groupSize, Long groupKey, Long instanceKey, Shape shape, Options... options) {
+  public static <T extends TNumber> CollectiveGather<T> create(Scope scope, Operand<T> input, Long groupSize, Long groupKey, Long instanceKey, Shape shape, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("CollectiveGather", scope.makeOpName("CollectiveGather"));
-    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("group_size", groupSize);
     opBuilder.setAttr("group_key", groupKey);
@@ -119,7 +118,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Concat.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Concat.java
index fa2898efb10..116cea2d4aa 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Concat.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Concat.java
@@ -21,13 +21,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.Operands;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * Concatenates tensors along one dimension.
@@ -35,7 +35,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator
-public final class Concat<T extends Tensor> extends RawOp implements Operand<T> {
+public final class Concat<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Concat operation.
@@ -48,10 +48,10 @@ public final class Concat<T extends Tensor> extends RawOp implements Operand<T>
    * @return a new instance of Concat
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor, U extends Tensor & TNumber> Concat<T> create(Scope scope, Iterable<Operand<T>> values, Operand<U> axis) {
+  public static <T extends TType, U extends TNumber> Concat<T> create(Scope scope, Iterable<Operand<T>> values, Operand<U> axis) {
     OperationBuilder opBuilder = scope.env().opBuilder("ConcatV2", scope.makeOpName("Concat"));
-    opBuilder.addInputList(Operands.asOutputs(values));
-    opBuilder.addInput(axis.asOutput());
+    opBuilder.addInputList(Operands.asOutputs(scope, values));
+    opBuilder.addInput(axis.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new Concat<T>(opBuilder.build());
   }
@@ -66,7 +66,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ConsumeMutexLock.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ConsumeMutexLock.java
index 094f6d5e4b0..2c88d65429f 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ConsumeMutexLock.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ConsumeMutexLock.java
@@ -49,7 +49,7 @@ public final class ConsumeMutexLock extends RawOp {
   @Endpoint(describeByClass = true)
   public static ConsumeMutexLock create(Scope scope, Operand<?> mutexLock) {
     OperationBuilder opBuilder = scope.env().opBuilder("ConsumeMutexLock", scope.makeOpName("ConsumeMutexLock"));
-    opBuilder.addInput(mutexLock.asOutput());
+    opBuilder.addInput(mutexLock.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new ConsumeMutexLock(opBuilder.build());
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Copy.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Copy.java
index b114662e162..13a00037a10 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Copy.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Copy.java
@@ -22,11 +22,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Copy a tensor from CPU-to-CPU or GPU-to-GPU.
@@ -42,7 +42,7 @@
  * 
  * @param <T> data type for {@code output()} output
  */
-public final class Copy<T extends Tensor> extends RawOp implements Operand<T> {
+public final class Copy<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.core.Copy}
@@ -85,9 +85,9 @@ private Options() {
    * @return a new instance of Copy
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> Copy<T> create(Scope scope, Operand<T> input, Options... options) {
+  public static <T extends TType> Copy<T> create(Scope scope, Operand<T> input, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("Copy", scope.makeOpName("Copy"));
-    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -131,7 +131,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/CopyHost.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/CopyHost.java
index 057b07cdb76..d18b72eaaba 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/CopyHost.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/CopyHost.java
@@ -22,11 +22,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Copy a tensor to host.
@@ -40,7 +40,7 @@
  * 
  * @param <T> data type for {@code output()} output
  */
-public final class CopyHost<T extends Tensor> extends RawOp implements Operand<T> {
+public final class CopyHost<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.core.CopyHost}
@@ -83,9 +83,9 @@ private Options() {
    * @return a new instance of CopyHost
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> CopyHost<T> create(Scope scope, Operand<T> input, Options... options) {
+  public static <T extends TType> CopyHost<T> create(Scope scope, Operand<T> input, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("CopyHost", scope.makeOpName("CopyHost"));
-    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -129,7 +129,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/CountUpTo.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/CountUpTo.java
index 9a36603b148..47aad076465 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/CountUpTo.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/CountUpTo.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -34,7 +33,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator
-public final class CountUpTo<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class CountUpTo<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new CountUpTo operation.
@@ -46,9 +45,9 @@ public final class CountUpTo<T extends Tensor & TNumber> extends RawOp implement
    * @return a new instance of CountUpTo
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> CountUpTo<T> create(Scope scope, Operand<T> ref, Long limit) {
+  public static <T extends TNumber> CountUpTo<T> create(Scope scope, Operand<T> ref, Long limit) {
     OperationBuilder opBuilder = scope.env().opBuilder("CountUpTo", scope.makeOpName("CountUpTo"));
-    opBuilder.addInput(ref.asOutput());
+    opBuilder.addInput(ref.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("limit", limit);
     return new CountUpTo<T>(opBuilder.build());
@@ -63,7 +62,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/DecodeProto.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/DecodeProto.java
index 28af55f8465..5d01ad3ac00 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/DecodeProto.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/DecodeProto.java
@@ -137,7 +137,7 @@ private Options() {
   @Endpoint(describeByClass = true)
   public static DecodeProto create(Scope scope, Operand<TString> bytes, String messageType, List<String> fieldNames, List<DataType<?>> outputTypes, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("DecodeProtoV2", scope.makeOpName("DecodeProto"));
-    opBuilder.addInput(bytes.asOutput());
+    opBuilder.addInput(bytes.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("message_type", messageType);
     String[] fieldNamesArray = new String[fieldNames.size()];
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/DeepCopy.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/DeepCopy.java
index f7f8f85d83a..e6bcb413ff8 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/DeepCopy.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/DeepCopy.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Makes a copy of `x`.
@@ -33,7 +33,7 @@
  * @param <T> data type for {@code y()} output
  */
 @Operator
-public final class DeepCopy<T extends Tensor> extends RawOp implements Operand<T> {
+public final class DeepCopy<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new DeepCopy operation.
@@ -43,9 +43,9 @@ public final class DeepCopy<T extends Tensor> extends RawOp implements Operand<T
    * @return a new instance of DeepCopy
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> DeepCopy<T> create(Scope scope, Operand<T> x) {
+  public static <T extends TType> DeepCopy<T> create(Scope scope, Operand<T> x) {
     OperationBuilder opBuilder = scope.env().opBuilder("DeepCopy", scope.makeOpName("DeepCopy"));
-    opBuilder.addInput(x.asOutput());
+    opBuilder.addInput(x.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new DeepCopy<T>(opBuilder.build());
   }
@@ -59,7 +59,7 @@ public Output<T> y() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return y;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/DeleteSessionTensor.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/DeleteSessionTensor.java
index 5f92cc26ca2..30f29c07e20 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/DeleteSessionTensor.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/DeleteSessionTensor.java
@@ -42,7 +42,7 @@ public final class DeleteSessionTensor extends RawOp {
   @Endpoint(describeByClass = true)
   public static DeleteSessionTensor create(Scope scope, Operand<TString> handle) {
     OperationBuilder opBuilder = scope.env().opBuilder("DeleteSessionTensor", scope.makeOpName("DeleteSessionTensor"));
-    opBuilder.addInput(handle.asOutput());
+    opBuilder.addInput(handle.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new DeleteSessionTensor(opBuilder.build());
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/DestroyResourceOp.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/DestroyResourceOp.java
index 8a427166874..b4dd2ba4acd 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/DestroyResourceOp.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/DestroyResourceOp.java
@@ -65,7 +65,7 @@ private Options() {
   @Endpoint(describeByClass = true)
   public static DestroyResourceOp create(Scope scope, Operand<?> resource, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("DestroyResourceOp", scope.makeOpName("DestroyResourceOp"));
-    opBuilder.addInput(resource.asOutput());
+    opBuilder.addInput(resource.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/DestroyTemporaryVariable.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/DestroyTemporaryVariable.java
index 161430a22e2..6dd40821a47 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/DestroyTemporaryVariable.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/DestroyTemporaryVariable.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Destroys the temporary variable and returns its final value.
@@ -41,7 +41,7 @@
  * @param <T> data type for {@code value()} output
  */
 @Operator
-public final class DestroyTemporaryVariable<T extends Tensor> extends RawOp implements Operand<T> {
+public final class DestroyTemporaryVariable<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new DestroyTemporaryVariable operation.
@@ -53,9 +53,9 @@ public final class DestroyTemporaryVariable<T extends Tensor> extends RawOp impl
    * @return a new instance of DestroyTemporaryVariable
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> DestroyTemporaryVariable<T> create(Scope scope, Operand<T> ref, String varName) {
+  public static <T extends TType> DestroyTemporaryVariable<T> create(Scope scope, Operand<T> ref, String varName) {
     OperationBuilder opBuilder = scope.env().opBuilder("DestroyTemporaryVariable", scope.makeOpName("DestroyTemporaryVariable"));
-    opBuilder.addInput(ref.asOutput());
+    opBuilder.addInput(ref.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("var_name", varName);
     return new DestroyTemporaryVariable<T>(opBuilder.build());
@@ -68,7 +68,7 @@ public Output<T> value() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return value;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/DummyMemoryCache.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/DummyMemoryCache.java
index e37c2c7f3ae..efc6061d293 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/DummyMemoryCache.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/DummyMemoryCache.java
@@ -21,15 +21,15 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  */
-public final class DummyMemoryCache extends RawOp implements Operand<Tensor> {
+public final class DummyMemoryCache extends RawOp implements Operand<TType> {
   
   /**
    * Factory method to create a class wrapping a new DummyMemoryCache operation.
@@ -52,8 +52,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<Tensor> asOutput() {
-    return (Output<Tensor>) handle;
+  public Output<TType> asOutput(Scope scope) {
+    return (Output<TType>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/DynamicPartition.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/DynamicPartition.java
index 73daaf7646e..3561ff823d5 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/DynamicPartition.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/DynamicPartition.java
@@ -24,12 +24,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TType;
 
 /**
  * Partitions `data` into `num_partitions` tensors using indices from `partitions`.
@@ -71,7 +71,7 @@
  * @param <T> data type for {@code outputs()} output
  */
 @Operator
-public final class DynamicPartition<T extends Tensor> extends RawOp implements Iterable<Operand<T>> {
+public final class DynamicPartition<T extends TType> extends RawOp implements Iterable<Operand<T>> {
   
   /**
    * Factory method to create a class wrapping a new DynamicPartition operation.
@@ -83,10 +83,10 @@ public final class DynamicPartition<T extends Tensor> extends RawOp implements I
    * @return a new instance of DynamicPartition
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> DynamicPartition<T> create(Scope scope, Operand<T> data, Operand<TInt32> partitions, Long numPartitions) {
+  public static <T extends TType> DynamicPartition<T> create(Scope scope, Operand<T> data, Operand<TInt32> partitions, Long numPartitions) {
     OperationBuilder opBuilder = scope.env().opBuilder("DynamicPartition", scope.makeOpName("DynamicPartition"));
-    opBuilder.addInput(data.asOutput());
-    opBuilder.addInput(partitions.asOutput());
+    opBuilder.addInput(data.asOutput(scope));
+    opBuilder.addInput(partitions.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("num_partitions", numPartitions);
     return new DynamicPartition<T>(opBuilder.build());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/DynamicStitch.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/DynamicStitch.java
index 590d9a2e3e1..c2f3f9e06c6 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/DynamicStitch.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/DynamicStitch.java
@@ -21,13 +21,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.Operands;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TType;
 
 /**
  * Interleave the values from the `data` tensors into a single tensor.
@@ -90,7 +90,7 @@
  * @param <T> data type for {@code merged()} output
  */
 @Operator
-public final class DynamicStitch<T extends Tensor> extends RawOp implements Operand<T> {
+public final class DynamicStitch<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new DynamicStitch operation.
@@ -101,10 +101,10 @@ public final class DynamicStitch<T extends Tensor> extends RawOp implements Oper
    * @return a new instance of DynamicStitch
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> DynamicStitch<T> create(Scope scope, Iterable<Operand<TInt32>> indices, Iterable<Operand<T>> data) {
+  public static <T extends TType> DynamicStitch<T> create(Scope scope, Iterable<Operand<TInt32>> indices, Iterable<Operand<T>> data) {
     OperationBuilder opBuilder = scope.env().opBuilder("DynamicStitch", scope.makeOpName("DynamicStitch"));
-    opBuilder.addInputList(Operands.asOutputs(indices));
-    opBuilder.addInputList(Operands.asOutputs(data));
+    opBuilder.addInputList(Operands.asOutputs(scope, indices));
+    opBuilder.addInputList(Operands.asOutputs(scope, data));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new DynamicStitch<T>(opBuilder.build());
   }
@@ -116,7 +116,7 @@ public Output<T> merged() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return merged;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/EditDistance.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/EditDistance.java
index 4220271bbf7..f914700e9f4 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/EditDistance.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/EditDistance.java
@@ -21,13 +21,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TFloat32;
 import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
 
 /**
  * Computes the (possibly normalized) Levenshtein Edit Distance.
@@ -82,14 +82,14 @@ private Options() {
    * @return a new instance of EditDistance
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> EditDistance create(Scope scope, Operand<TInt64> hypothesisIndices, Operand<T> hypothesisValues, Operand<TInt64> hypothesisShape, Operand<TInt64> truthIndices, Operand<T> truthValues, Operand<TInt64> truthShape, Options... options) {
+  public static <T extends TType> EditDistance create(Scope scope, Operand<TInt64> hypothesisIndices, Operand<T> hypothesisValues, Operand<TInt64> hypothesisShape, Operand<TInt64> truthIndices, Operand<T> truthValues, Operand<TInt64> truthShape, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("EditDistance", scope.makeOpName("EditDistance"));
-    opBuilder.addInput(hypothesisIndices.asOutput());
-    opBuilder.addInput(hypothesisValues.asOutput());
-    opBuilder.addInput(hypothesisShape.asOutput());
-    opBuilder.addInput(truthIndices.asOutput());
-    opBuilder.addInput(truthValues.asOutput());
-    opBuilder.addInput(truthShape.asOutput());
+    opBuilder.addInput(hypothesisIndices.asOutput(scope));
+    opBuilder.addInput(hypothesisValues.asOutput(scope));
+    opBuilder.addInput(hypothesisShape.asOutput(scope));
+    opBuilder.addInput(truthIndices.asOutput(scope));
+    opBuilder.addInput(truthValues.asOutput(scope));
+    opBuilder.addInput(truthShape.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -147,7 +147,7 @@ public Output<TFloat32> output() {
   }
   
   @Override
-  public Output<TFloat32> asOutput() {
+  public Output<TFloat32> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Empty.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Empty.java
index 6419d47510c..b78e44b0709 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Empty.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Empty.java
@@ -22,12 +22,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TType;
 
 /**
  * Creates a tensor with the given shape.
@@ -37,7 +37,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator
-public final class Empty<T extends Tensor> extends RawOp implements Operand<T> {
+public final class Empty<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.core.Empty}
@@ -68,9 +68,9 @@ private Options() {
    * @return a new instance of Empty
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> Empty<T> create(Scope scope, Operand<TInt32> shape, DataType<T> dtype, Options... options) {
+  public static <T extends TType> Empty<T> create(Scope scope, Operand<TInt32> shape, DataType<T> dtype, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("Empty", scope.makeOpName("Empty"));
-    opBuilder.addInput(shape.asOutput());
+    opBuilder.addInput(shape.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("dtype", dtype);
     if (options != null) {
@@ -98,7 +98,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/EmptyTensorList.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/EmptyTensorList.java
index 545beb63c10..68a46956ad6 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/EmptyTensorList.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/EmptyTensorList.java
@@ -22,13 +22,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * Creates and returns an empty tensor list.
@@ -41,7 +41,7 @@
  * element_shape: a shape compatible with that of elements in the list.
  */
 @Operator
-public final class EmptyTensorList extends RawOp implements Operand<Tensor> {
+public final class EmptyTensorList extends RawOp implements Operand<TType> {
   
   /**
    * Factory method to create a class wrapping a new EmptyTensorList operation.
@@ -53,10 +53,10 @@ public final class EmptyTensorList extends RawOp implements Operand<Tensor> {
    * @return a new instance of EmptyTensorList
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber, U extends Tensor> EmptyTensorList create(Scope scope, Operand<T> elementShape, Operand<TInt32> maxNumElements, DataType<U> elementDtype) {
+  public static <T extends TNumber, U extends TType> EmptyTensorList create(Scope scope, Operand<T> elementShape, Operand<TInt32> maxNumElements, DataType<U> elementDtype) {
     OperationBuilder opBuilder = scope.env().opBuilder("EmptyTensorList", scope.makeOpName("EmptyTensorList"));
-    opBuilder.addInput(elementShape.asOutput());
-    opBuilder.addInput(maxNumElements.asOutput());
+    opBuilder.addInput(elementShape.asOutput(scope));
+    opBuilder.addInput(maxNumElements.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("element_dtype", elementDtype);
     return new EmptyTensorList(opBuilder.build());
@@ -70,8 +70,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<Tensor> asOutput() {
-    return (Output<Tensor>) handle;
+  public Output<TType> asOutput(Scope scope) {
+    return (Output<TType>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/EncodeProto.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/EncodeProto.java
index 4d0396dcb9d..9e7a5076a37 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/EncodeProto.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/EncodeProto.java
@@ -109,8 +109,8 @@ private Options() {
   @Endpoint(describeByClass = true)
   public static EncodeProto create(Scope scope, Operand<TInt32> sizes, Iterable<Operand<?>> values, List<String> fieldNames, String messageType, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("EncodeProto", scope.makeOpName("EncodeProto"));
-    opBuilder.addInput(sizes.asOutput());
-    opBuilder.addInputList(Operands.asOutputs(values));
+    opBuilder.addInput(sizes.asOutput(scope));
+    opBuilder.addInputList(Operands.asOutputs(scope, values));
     opBuilder = scope.applyControlDependencies(opBuilder);
     String[] fieldNamesArray = new String[fieldNames.size()];
     for (int i = 0; i < fieldNamesArray.length; ++i) {
@@ -143,7 +143,7 @@ public Output<TString> bytes() {
   }
   
   @Override
-  public Output<TString> asOutput() {
+  public Output<TString> asOutput(Scope scope) {
     return bytes;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/EnsureShape.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/EnsureShape.java
index 3ace0a8822a..0babbe5c112 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/EnsureShape.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/EnsureShape.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Ensures that the tensor's shape matches the expected shape.
@@ -37,7 +37,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator
-public final class EnsureShape<T extends Tensor> extends RawOp implements Operand<T> {
+public final class EnsureShape<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new EnsureShape operation.
@@ -48,9 +48,9 @@ public final class EnsureShape<T extends Tensor> extends RawOp implements Operan
    * @return a new instance of EnsureShape
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> EnsureShape<T> create(Scope scope, Operand<T> input, Shape shape) {
+  public static <T extends TType> EnsureShape<T> create(Scope scope, Operand<T> input, Shape shape) {
     OperationBuilder opBuilder = scope.env().opBuilder("EnsureShape", scope.makeOpName("EnsureShape"));
-    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("shape", shape);
     return new EnsureShape<T>(opBuilder.build());
@@ -64,7 +64,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Enter.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Enter.java
index a7ad6608dcb..9fb174a1942 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Enter.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Enter.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Creates or finds a child frame, and makes `data` available to the child frame.
@@ -38,7 +38,7 @@
  * 
  * @param <T> data type for {@code output()} output
  */
-public final class Enter<T extends Tensor> extends RawOp implements Operand<T> {
+public final class Enter<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.core.Enter}
@@ -78,9 +78,9 @@ private Options() {
    * @return a new instance of Enter
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> Enter<T> create(Scope scope, Operand<T> data, String frameName, Options... options) {
+  public static <T extends TType> Enter<T> create(Scope scope, Operand<T> data, String frameName, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("Enter", scope.makeOpName("Enter"));
-    opBuilder.addInput(data.asOutput());
+    opBuilder.addInput(data.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("frame_name", frameName);
     if (options != null) {
@@ -118,7 +118,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Exit.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Exit.java
index 9c57a566f31..076464de2cc 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Exit.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Exit.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Exits the current frame to its parent frame.
@@ -34,7 +34,7 @@
  * 
  * @param <T> data type for {@code output()} output
  */
-public final class Exit<T extends Tensor> extends RawOp implements Operand<T> {
+public final class Exit<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Exit operation.
@@ -44,9 +44,9 @@ public final class Exit<T extends Tensor> extends RawOp implements Operand<T> {
    * @return a new instance of Exit
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> Exit<T> create(Scope scope, Operand<T> data) {
+  public static <T extends TType> Exit<T> create(Scope scope, Operand<T> data) {
     OperationBuilder opBuilder = scope.env().opBuilder("Exit", scope.makeOpName("Exit"));
-    opBuilder.addInput(data.asOutput());
+    opBuilder.addInput(data.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new Exit<T>(opBuilder.build());
   }
@@ -59,7 +59,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ExpandDims.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ExpandDims.java
index 2307b7cd5e1..a961f9abc77 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ExpandDims.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ExpandDims.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * Inserts a dimension of 1 into a tensor's shape.
@@ -63,7 +63,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator
-public final class ExpandDims<T extends Tensor> extends RawOp implements Operand<T> {
+public final class ExpandDims<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new ExpandDims operation.
@@ -76,10 +76,10 @@ public final class ExpandDims<T extends Tensor> extends RawOp implements Operand
    * @return a new instance of ExpandDims
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor, U extends Tensor & TNumber> ExpandDims<T> create(Scope scope, Operand<T> input, Operand<U> axis) {
+  public static <T extends TType, U extends TNumber> ExpandDims<T> create(Scope scope, Operand<T> input, Operand<U> axis) {
     OperationBuilder opBuilder = scope.env().opBuilder("ExpandDims", scope.makeOpName("ExpandDims"));
-    opBuilder.addInput(input.asOutput());
-    opBuilder.addInput(axis.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
+    opBuilder.addInput(axis.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new ExpandDims<T>(opBuilder.build());
   }
@@ -93,7 +93,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ExtractVolumePatches.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ExtractVolumePatches.java
index a1c5c6fe6d9..b68e98b22cd 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ExtractVolumePatches.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ExtractVolumePatches.java
@@ -22,7 +22,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -35,7 +34,7 @@
  * @param <T> data type for {@code patches()} output
  */
 @Operator
-public final class ExtractVolumePatches<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class ExtractVolumePatches<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new ExtractVolumePatches operation.
@@ -56,9 +55,9 @@ public final class ExtractVolumePatches<T extends Tensor & TNumber> extends RawO
    * @return a new instance of ExtractVolumePatches
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> ExtractVolumePatches<T> create(Scope scope, Operand<T> input, List<Long> ksizes, List<Long> strides, String padding) {
+  public static <T extends TNumber> ExtractVolumePatches<T> create(Scope scope, Operand<T> input, List<Long> ksizes, List<Long> strides, String padding) {
     OperationBuilder opBuilder = scope.env().opBuilder("ExtractVolumePatches", scope.makeOpName("ExtractVolumePatches"));
-    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     long[] ksizesArray = new long[ksizes.size()];
     for (int i = 0; i < ksizesArray.length; ++i) {
@@ -86,7 +85,7 @@ public Output<T> patches() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return patches;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Fill.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Fill.java
index 5a5eb9c69be..5f45e136e63 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Fill.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Fill.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * Creates a tensor filled with a scalar value.
@@ -57,7 +57,7 @@
  * @param <U> data type for {@code output()} output
  */
 @Operator
-public final class Fill<U extends Tensor> extends RawOp implements Operand<U> {
+public final class Fill<U extends TType> extends RawOp implements Operand<U> {
   
   /**
    * Factory method to create a class wrapping a new Fill operation.
@@ -72,10 +72,10 @@ public final class Fill<U extends Tensor> extends RawOp implements Operand<U> {
    * @return a new instance of Fill
    */
   @Endpoint(describeByClass = true)
-  public static <U extends Tensor, T extends Tensor & TNumber> Fill<U> create(Scope scope, Operand<T> dims, Operand<U> value) {
+  public static <U extends TType, T extends TNumber> Fill<U> create(Scope scope, Operand<T> dims, Operand<U> value) {
     OperationBuilder opBuilder = scope.env().opBuilder("Fill", scope.makeOpName("Fill"));
-    opBuilder.addInput(dims.asOutput());
-    opBuilder.addInput(value.asOutput());
+    opBuilder.addInput(dims.asOutput(scope));
+    opBuilder.addInput(value.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new Fill<U>(opBuilder.build());
   }
@@ -87,7 +87,7 @@ public Output<U> output() {
   }
   
   @Override
-  public Output<U> asOutput() {
+  public Output<U> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Fingerprint.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Fingerprint.java
index 90b4a868644..e4c2874e8c3 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Fingerprint.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Fingerprint.java
@@ -21,13 +21,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TString;
 import org.tensorflow.types.TUint8;
+import org.tensorflow.types.family.TType;
 
 /**
  * Generates fingerprint values.
@@ -74,10 +74,10 @@ public final class Fingerprint extends RawOp implements Operand<TUint8> {
    * @return a new instance of Fingerprint
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> Fingerprint create(Scope scope, Operand<T> data, Operand<TString> method) {
+  public static <T extends TType> Fingerprint create(Scope scope, Operand<T> data, Operand<TString> method) {
     OperationBuilder opBuilder = scope.env().opBuilder("Fingerprint", scope.makeOpName("Fingerprint"));
-    opBuilder.addInput(data.asOutput());
-    opBuilder.addInput(method.asOutput());
+    opBuilder.addInput(data.asOutput(scope));
+    opBuilder.addInput(method.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new Fingerprint(opBuilder.build());
   }
@@ -92,7 +92,7 @@ public Output<TUint8> fingerprint() {
   }
   
   @Override
-  public Output<TUint8> asOutput() {
+  public Output<TUint8> asOutput(Scope scope) {
     return fingerprint;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Gather.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Gather.java
index 33d82ad6a83..cf2f3f99ec4 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Gather.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Gather.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * Gather slices from `params` axis `axis` according to `indices`.
@@ -60,7 +60,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator
-public final class Gather<T extends Tensor> extends RawOp implements Operand<T> {
+public final class Gather<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.core.Gather}
@@ -94,11 +94,11 @@ private Options() {
    * @return a new instance of Gather
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor, U extends Tensor & TNumber, V extends Tensor & TNumber> Gather<T> create(Scope scope, Operand<T> params, Operand<U> indices, Operand<V> axis, Options... options) {
+  public static <T extends TType, U extends TNumber, V extends TNumber> Gather<T> create(Scope scope, Operand<T> params, Operand<U> indices, Operand<V> axis, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("GatherV2", scope.makeOpName("Gather"));
-    opBuilder.addInput(params.asOutput());
-    opBuilder.addInput(indices.asOutput());
-    opBuilder.addInput(axis.asOutput());
+    opBuilder.addInput(params.asOutput(scope));
+    opBuilder.addInput(indices.asOutput(scope));
+    opBuilder.addInput(axis.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -126,7 +126,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/GatherNd.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/GatherNd.java
index dc0e1e1a5e6..4a10f694208 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/GatherNd.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/GatherNd.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * Gather slices from `params` into a Tensor with shape specified by `indices`.
@@ -127,7 +127,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator
-public final class GatherNd<T extends Tensor> extends RawOp implements Operand<T> {
+public final class GatherNd<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new GatherNd operation.
@@ -138,10 +138,10 @@ public final class GatherNd<T extends Tensor> extends RawOp implements Operand<T
    * @return a new instance of GatherNd
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor, U extends Tensor & TNumber> GatherNd<T> create(Scope scope, Operand<T> params, Operand<U> indices) {
+  public static <T extends TType, U extends TNumber> GatherNd<T> create(Scope scope, Operand<T> params, Operand<U> indices) {
     OperationBuilder opBuilder = scope.env().opBuilder("GatherNd", scope.makeOpName("GatherNd"));
-    opBuilder.addInput(params.asOutput());
-    opBuilder.addInput(indices.asOutput());
+    opBuilder.addInput(params.asOutput(scope));
+    opBuilder.addInput(indices.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new GatherNd<T>(opBuilder.build());
   }
@@ -155,7 +155,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/GetSessionHandle.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/GetSessionHandle.java
index 3eb38f77d30..efde73b2a96 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/GetSessionHandle.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/GetSessionHandle.java
@@ -21,17 +21,17 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Store the input tensor in the state of the current session.
  */
 @Operator
-public final class GetSessionHandle extends RawOp implements Operand<Tensor> {
+public final class GetSessionHandle extends RawOp implements Operand<TType> {
   
   /**
    * Factory method to create a class wrapping a new GetSessionHandle operation.
@@ -41,9 +41,9 @@ public final class GetSessionHandle extends RawOp implements Operand<Tensor> {
    * @return a new instance of GetSessionHandle
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> GetSessionHandle create(Scope scope, Operand<T> value) {
+  public static <T extends TType> GetSessionHandle create(Scope scope, Operand<T> value) {
     OperationBuilder opBuilder = scope.env().opBuilder("GetSessionHandleV2", scope.makeOpName("GetSessionHandle"));
-    opBuilder.addInput(value.asOutput());
+    opBuilder.addInput(value.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new GetSessionHandle(opBuilder.build());
   }
@@ -58,8 +58,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<Tensor> asOutput() {
-    return (Output<Tensor>) handle;
+  public Output<TType> asOutput(Scope scope) {
+    return (Output<TType>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/GetSessionTensor.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/GetSessionTensor.java
index 50e0f66cff2..a9abc7821c8 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/GetSessionTensor.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/GetSessionTensor.java
@@ -22,12 +22,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TString;
+import org.tensorflow.types.family.TType;
 
 /**
  * Get the value of the tensor specified by its handle.
@@ -35,7 +35,7 @@
  * @param <T> data type for {@code value()} output
  */
 @Operator
-public final class GetSessionTensor<T extends Tensor> extends RawOp implements Operand<T> {
+public final class GetSessionTensor<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new GetSessionTensor operation.
@@ -46,9 +46,9 @@ public final class GetSessionTensor<T extends Tensor> extends RawOp implements O
    * @return a new instance of GetSessionTensor
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> GetSessionTensor<T> create(Scope scope, Operand<TString> handle, DataType<T> dtype) {
+  public static <T extends TType> GetSessionTensor<T> create(Scope scope, Operand<TString> handle, DataType<T> dtype) {
     OperationBuilder opBuilder = scope.env().opBuilder("GetSessionTensor", scope.makeOpName("GetSessionTensor"));
-    opBuilder.addInput(handle.asOutput());
+    opBuilder.addInput(handle.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("dtype", dtype);
     return new GetSessionTensor<T>(opBuilder.build());
@@ -62,7 +62,7 @@ public Output<T> value() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return value;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/GuaranteeConst.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/GuaranteeConst.java
index 0c213bec82d..e3f4468fa28 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/GuaranteeConst.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/GuaranteeConst.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Gives a guarantee to the TF runtime that the input tensor is a constant.
@@ -40,7 +40,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator
-public final class GuaranteeConst<T extends Tensor> extends RawOp implements Operand<T> {
+public final class GuaranteeConst<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new GuaranteeConst operation.
@@ -50,9 +50,9 @@ public final class GuaranteeConst<T extends Tensor> extends RawOp implements Ope
    * @return a new instance of GuaranteeConst
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> GuaranteeConst<T> create(Scope scope, Operand<T> input) {
+  public static <T extends TType> GuaranteeConst<T> create(Scope scope, Operand<T> input) {
     OperationBuilder opBuilder = scope.env().opBuilder("GuaranteeConst", scope.makeOpName("GuaranteeConst"));
-    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new GuaranteeConst<T>(opBuilder.build());
   }
@@ -64,7 +64,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/HashTable.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/HashTable.java
index d02e8d37a01..e191d086497 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/HashTable.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/HashTable.java
@@ -22,11 +22,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Creates a non-initialized hash table.
@@ -36,7 +36,7 @@
  * table will be immutable.
  */
 @Operator
-public final class HashTable extends RawOp implements Operand<Tensor> {
+public final class HashTable extends RawOp implements Operand<TType> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.core.HashTable}
@@ -88,7 +88,7 @@ private Options() {
    * @return a new instance of HashTable
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor, U extends Tensor> HashTable create(Scope scope, DataType<T> keyDtype, DataType<U> valueDtype, Options... options) {
+  public static <T extends TType, U extends TType> HashTable create(Scope scope, DataType<T> keyDtype, DataType<U> valueDtype, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("HashTableV2", scope.makeOpName("HashTable"));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("key_dtype", keyDtype);
@@ -142,8 +142,8 @@ public Output<?> tableHandle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<Tensor> asOutput() {
-    return (Output<Tensor>) tableHandle;
+  public Output<TType> asOutput(Scope scope) {
+    return (Output<TType>) tableHandle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/HistogramFixedWidth.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/HistogramFixedWidth.java
index efd20c1b3e9..e6f46c71057 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/HistogramFixedWidth.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/HistogramFixedWidth.java
@@ -22,7 +22,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -52,7 +51,7 @@
  * @param <U> data type for {@code out()} output
  */
 @Operator
-public final class HistogramFixedWidth<U extends Tensor & TNumber> extends RawOp implements Operand<U> {
+public final class HistogramFixedWidth<U extends TNumber> extends RawOp implements Operand<U> {
   
   /**
    * Factory method to create a class wrapping a new HistogramFixedWidth operation.
@@ -67,11 +66,11 @@ public final class HistogramFixedWidth<U extends Tensor & TNumber> extends RawOp
    * @return a new instance of HistogramFixedWidth
    */
   @Endpoint(describeByClass = true)
-  public static <U extends Tensor & TNumber, T extends Tensor & TNumber> HistogramFixedWidth<U> create(Scope scope, Operand<T> values, Operand<T> valueRange, Operand<TInt32> nbins, DataType<U> dtype) {
+  public static <U extends TNumber, T extends TNumber> HistogramFixedWidth<U> create(Scope scope, Operand<T> values, Operand<T> valueRange, Operand<TInt32> nbins, DataType<U> dtype) {
     OperationBuilder opBuilder = scope.env().opBuilder("HistogramFixedWidth", scope.makeOpName("HistogramFixedWidth"));
-    opBuilder.addInput(values.asOutput());
-    opBuilder.addInput(valueRange.asOutput());
-    opBuilder.addInput(nbins.asOutput());
+    opBuilder.addInput(values.asOutput(scope));
+    opBuilder.addInput(valueRange.asOutput(scope));
+    opBuilder.addInput(nbins.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("dtype", dtype);
     return new HistogramFixedWidth<U>(opBuilder.build());
@@ -89,7 +88,7 @@ public static <U extends Tensor & TNumber, T extends Tensor & TNumber> Histogram
    * @return a new instance of HistogramFixedWidth
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> HistogramFixedWidth<TInt32> create(Scope scope, Operand<T> values, Operand<T> valueRange, Operand<TInt32> nbins) {
+  public static <T extends TNumber> HistogramFixedWidth<TInt32> create(Scope scope, Operand<T> values, Operand<T> valueRange, Operand<TInt32> nbins) {
     return create(scope, values, valueRange, nbins, TInt32.DTYPE);
   }
   
@@ -101,7 +100,7 @@ public Output<U> out() {
   }
   
   @Override
-  public Output<U> asOutput() {
+  public Output<U> asOutput(Scope scope) {
     return out;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Identity.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Identity.java
index 8db7b6ea287..c0731947bbb 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Identity.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Identity.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Return a tensor with the same shape and contents as the input tensor or value.
@@ -33,7 +33,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator
-public final class Identity<T extends Tensor> extends RawOp implements Operand<T> {
+public final class Identity<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Identity operation.
@@ -43,9 +43,9 @@ public final class Identity<T extends Tensor> extends RawOp implements Operand<T
    * @return a new instance of Identity
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> Identity<T> create(Scope scope, Operand<T> input) {
+  public static <T extends TType> Identity<T> create(Scope scope, Operand<T> input) {
     OperationBuilder opBuilder = scope.env().opBuilder("Identity", scope.makeOpName("Identity"));
-    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new Identity<T>(opBuilder.build());
   }
@@ -57,7 +57,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/IdentityN.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/IdentityN.java
index dff191e5a23..43364cf6d91 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/IdentityN.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/IdentityN.java
@@ -24,12 +24,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.Operands;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Returns a list of tensors with the same shapes and contents as the input
@@ -51,7 +51,7 @@
  * 
  */
 @Operator
-public final class IdentityN extends RawOp implements Iterable<Operand<Tensor>> {
+public final class IdentityN extends RawOp implements Iterable<Operand<TType>> {
   
   /**
    * Factory method to create a class wrapping a new IdentityN operation.
@@ -63,7 +63,7 @@ public final class IdentityN extends RawOp implements Iterable<Operand<Tensor>>
   @Endpoint(describeByClass = true)
   public static IdentityN create(Scope scope, Iterable<Operand<?>> input) {
     OperationBuilder opBuilder = scope.env().opBuilder("IdentityN", scope.makeOpName("IdentityN"));
-    opBuilder.addInputList(Operands.asOutputs(input));
+    opBuilder.addInputList(Operands.asOutputs(scope, input));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new IdentityN(opBuilder.build());
   }
@@ -76,7 +76,7 @@ public List<Output<?>> output() {
   
   @Override
   @SuppressWarnings({"rawtypes", "unchecked"})
-  public Iterator<Operand<Tensor>> iterator() {
+  public Iterator<Operand<TType>> iterator() {
     return (Iterator) output.iterator();
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ImmutableConst.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ImmutableConst.java
index 86183c6ee44..a6976162ff0 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ImmutableConst.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ImmutableConst.java
@@ -22,12 +22,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Returns immutable tensor from memory region.
@@ -37,7 +37,7 @@
  * @param <T> data type for {@code tensor()} output
  */
 @Operator
-public final class ImmutableConst<T extends Tensor> extends RawOp implements Operand<T> {
+public final class ImmutableConst<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new ImmutableConst operation.
@@ -50,7 +50,7 @@ public final class ImmutableConst<T extends Tensor> extends RawOp implements Ope
    * @return a new instance of ImmutableConst
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> ImmutableConst<T> create(Scope scope, DataType<T> dtype, Shape shape, String memoryRegionName) {
+  public static <T extends TType> ImmutableConst<T> create(Scope scope, DataType<T> dtype, Shape shape, String memoryRegionName) {
     OperationBuilder opBuilder = scope.env().opBuilder("ImmutableConst", scope.makeOpName("ImmutableConst"));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("dtype", dtype);
@@ -66,7 +66,7 @@ public Output<T> tensor() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return tensor;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/InitializeTable.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/InitializeTable.java
index 50370b6905b..e6e09c248b0 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/InitializeTable.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/InitializeTable.java
@@ -20,11 +20,11 @@
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Table initializer that takes two tensors for keys and values respectively.
@@ -42,11 +42,11 @@ public final class InitializeTable extends RawOp {
    * @return a new instance of InitializeTable
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor, U extends Tensor> InitializeTable create(Scope scope, Operand<?> tableHandle, Operand<T> keys, Operand<U> values) {
+  public static <T extends TType, U extends TType> InitializeTable create(Scope scope, Operand<?> tableHandle, Operand<T> keys, Operand<U> values) {
     OperationBuilder opBuilder = scope.env().opBuilder("InitializeTableV2", scope.makeOpName("InitializeTable"));
-    opBuilder.addInput(tableHandle.asOutput());
-    opBuilder.addInput(keys.asOutput());
-    opBuilder.addInput(values.asOutput());
+    opBuilder.addInput(tableHandle.asOutput(scope));
+    opBuilder.addInput(keys.asOutput(scope));
+    opBuilder.addInput(values.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new InitializeTable(opBuilder.build());
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/InitializeTableFromTextFile.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/InitializeTableFromTextFile.java
index 2050c4d8628..f830a07e387 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/InitializeTableFromTextFile.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/InitializeTableFromTextFile.java
@@ -86,8 +86,8 @@ private Options() {
   @Endpoint(describeByClass = true)
   public static InitializeTableFromTextFile create(Scope scope, Operand<?> tableHandle, Operand<TString> filename, Long keyIndex, Long valueIndex, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("InitializeTableFromTextFileV2", scope.makeOpName("InitializeTableFromTextFile"));
-    opBuilder.addInput(tableHandle.asOutput());
-    opBuilder.addInput(filename.asOutput());
+    opBuilder.addInput(tableHandle.asOutput(scope));
+    opBuilder.addInput(filename.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("key_index", keyIndex);
     opBuilder.setAttr("value_index", valueIndex);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/InplaceAdd.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/InplaceAdd.java
index eeb7e331b79..0a84f9d71eb 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/InplaceAdd.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/InplaceAdd.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TType;
 
 /**
  *     Adds v into specified rows of x.
@@ -36,7 +36,7 @@
  * @param <T> data type for {@code y()} output
  */
 @Operator
-public final class InplaceAdd<T extends Tensor> extends RawOp implements Operand<T> {
+public final class InplaceAdd<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new InplaceAdd operation.
@@ -48,11 +48,11 @@ public final class InplaceAdd<T extends Tensor> extends RawOp implements Operand
    * @return a new instance of InplaceAdd
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> InplaceAdd<T> create(Scope scope, Operand<T> x, Operand<TInt32> i, Operand<T> v) {
+  public static <T extends TType> InplaceAdd<T> create(Scope scope, Operand<T> x, Operand<TInt32> i, Operand<T> v) {
     OperationBuilder opBuilder = scope.env().opBuilder("InplaceAdd", scope.makeOpName("InplaceAdd"));
-    opBuilder.addInput(x.asOutput());
-    opBuilder.addInput(i.asOutput());
-    opBuilder.addInput(v.asOutput());
+    opBuilder.addInput(x.asOutput(scope));
+    opBuilder.addInput(i.asOutput(scope));
+    opBuilder.addInput(v.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new InplaceAdd<T>(opBuilder.build());
   }
@@ -65,7 +65,7 @@ public Output<T> y() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return y;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/InplaceSub.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/InplaceSub.java
index 55733c9b0ea..ea18d37f3c0 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/InplaceSub.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/InplaceSub.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TType;
 
 /**
  *     Subtracts `v` into specified rows of `x`.
@@ -36,7 +36,7 @@
  * @param <T> data type for {@code y()} output
  */
 @Operator
-public final class InplaceSub<T extends Tensor> extends RawOp implements Operand<T> {
+public final class InplaceSub<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new InplaceSub operation.
@@ -48,11 +48,11 @@ public final class InplaceSub<T extends Tensor> extends RawOp implements Operand
    * @return a new instance of InplaceSub
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> InplaceSub<T> create(Scope scope, Operand<T> x, Operand<TInt32> i, Operand<T> v) {
+  public static <T extends TType> InplaceSub<T> create(Scope scope, Operand<T> x, Operand<TInt32> i, Operand<T> v) {
     OperationBuilder opBuilder = scope.env().opBuilder("InplaceSub", scope.makeOpName("InplaceSub"));
-    opBuilder.addInput(x.asOutput());
-    opBuilder.addInput(i.asOutput());
-    opBuilder.addInput(v.asOutput());
+    opBuilder.addInput(x.asOutput(scope));
+    opBuilder.addInput(i.asOutput(scope));
+    opBuilder.addInput(v.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new InplaceSub<T>(opBuilder.build());
   }
@@ -65,7 +65,7 @@ public Output<T> y() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return y;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/InplaceUpdate.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/InplaceUpdate.java
index dfbac30fa8a..691c8a39f1d 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/InplaceUpdate.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/InplaceUpdate.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TType;
 
 /**
  * Updates specified rows 'i' with values 'v'.
@@ -39,7 +39,7 @@
  * @param <T> data type for {@code y()} output
  */
 @Operator
-public final class InplaceUpdate<T extends Tensor> extends RawOp implements Operand<T> {
+public final class InplaceUpdate<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new InplaceUpdate operation.
@@ -51,11 +51,11 @@ public final class InplaceUpdate<T extends Tensor> extends RawOp implements Oper
    * @return a new instance of InplaceUpdate
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> InplaceUpdate<T> create(Scope scope, Operand<T> x, Operand<TInt32> i, Operand<T> v) {
+  public static <T extends TType> InplaceUpdate<T> create(Scope scope, Operand<T> x, Operand<TInt32> i, Operand<T> v) {
     OperationBuilder opBuilder = scope.env().opBuilder("InplaceUpdate", scope.makeOpName("InplaceUpdate"));
-    opBuilder.addInput(x.asOutput());
-    opBuilder.addInput(i.asOutput());
-    opBuilder.addInput(v.asOutput());
+    opBuilder.addInput(x.asOutput(scope));
+    opBuilder.addInput(i.asOutput(scope));
+    opBuilder.addInput(v.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new InplaceUpdate<T>(opBuilder.build());
   }
@@ -68,7 +68,7 @@ public Output<T> y() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return y;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/IsVariableInitialized.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/IsVariableInitialized.java
index 54ccb54b831..28aae03928c 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/IsVariableInitialized.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/IsVariableInitialized.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TBool;
+import org.tensorflow.types.family.TType;
 
 /**
  * Checks whether a tensor has been initialized.
@@ -44,9 +44,9 @@ public final class IsVariableInitialized extends RawOp implements Operand<TBool>
    * @return a new instance of IsVariableInitialized
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> IsVariableInitialized create(Scope scope, Operand<T> ref) {
+  public static <T extends TType> IsVariableInitialized create(Scope scope, Operand<T> ref) {
     OperationBuilder opBuilder = scope.env().opBuilder("IsVariableInitialized", scope.makeOpName("IsVariableInitialized"));
-    opBuilder.addInput(ref.asOutput());
+    opBuilder.addInput(ref.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new IsVariableInitialized(opBuilder.build());
   }
@@ -58,7 +58,7 @@ public Output<TBool> isInitialized() {
   }
   
   @Override
-  public Output<TBool> asOutput() {
+  public Output<TBool> asOutput(Scope scope) {
     return isInitialized;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LinSpace.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LinSpace.java
index 3162c4eec68..d2061ea6553 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LinSpace.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LinSpace.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -44,7 +43,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator
-public final class LinSpace<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class LinSpace<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new LinSpace operation.
@@ -56,11 +55,11 @@ public final class LinSpace<T extends Tensor & TNumber> extends RawOp implements
    * @return a new instance of LinSpace
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber, U extends Tensor & TNumber> LinSpace<T> create(Scope scope, Operand<T> start, Operand<T> stop, Operand<U> num) {
+  public static <T extends TNumber, U extends TNumber> LinSpace<T> create(Scope scope, Operand<T> start, Operand<T> stop, Operand<U> num) {
     OperationBuilder opBuilder = scope.env().opBuilder("LinSpace", scope.makeOpName("LinSpace"));
-    opBuilder.addInput(start.asOutput());
-    opBuilder.addInput(stop.asOutput());
-    opBuilder.addInput(num.asOutput());
+    opBuilder.addInput(start.asOutput(scope));
+    opBuilder.addInput(stop.asOutput(scope));
+    opBuilder.addInput(num.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new LinSpace<T>(opBuilder.build());
   }
@@ -73,7 +72,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LookupTableExport.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LookupTableExport.java
index a106f123343..6d59e0404f8 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LookupTableExport.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LookupTableExport.java
@@ -22,11 +22,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Outputs all keys and values in the table.
@@ -35,7 +35,7 @@
  * @param <U> data type for {@code values()} output
  */
 @Operator
-public final class LookupTableExport<T extends Tensor, U extends Tensor> extends RawOp {
+public final class LookupTableExport<T extends TType, U extends TType> extends RawOp {
   
   /**
    * Factory method to create a class wrapping a new LookupTableExport operation.
@@ -47,9 +47,9 @@ public final class LookupTableExport<T extends Tensor, U extends Tensor> extends
    * @return a new instance of LookupTableExport
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor, U extends Tensor> LookupTableExport<T, U> create(Scope scope, Operand<?> tableHandle, DataType<T> Tkeys, DataType<U> Tvalues) {
+  public static <T extends TType, U extends TType> LookupTableExport<T, U> create(Scope scope, Operand<?> tableHandle, DataType<T> Tkeys, DataType<U> Tvalues) {
     OperationBuilder opBuilder = scope.env().opBuilder("LookupTableExportV2", scope.makeOpName("LookupTableExport"));
-    opBuilder.addInput(tableHandle.asOutput());
+    opBuilder.addInput(tableHandle.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("Tkeys", Tkeys);
     opBuilder.setAttr("Tvalues", Tvalues);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LookupTableFind.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LookupTableFind.java
index beba84cda66..381907fabce 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LookupTableFind.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LookupTableFind.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Looks up keys in a table, outputs the corresponding values.
@@ -39,7 +39,7 @@
  * @param <U> data type for {@code values()} output
  */
 @Operator
-public final class LookupTableFind<U extends Tensor> extends RawOp implements Operand<U> {
+public final class LookupTableFind<U extends TType> extends RawOp implements Operand<U> {
   
   /**
    * Factory method to create a class wrapping a new LookupTableFind operation.
@@ -51,11 +51,11 @@ public final class LookupTableFind<U extends Tensor> extends RawOp implements Op
    * @return a new instance of LookupTableFind
    */
   @Endpoint(describeByClass = true)
-  public static <U extends Tensor, T extends Tensor> LookupTableFind<U> create(Scope scope, Operand<?> tableHandle, Operand<T> keys, Operand<U> defaultValue) {
+  public static <U extends TType, T extends TType> LookupTableFind<U> create(Scope scope, Operand<?> tableHandle, Operand<T> keys, Operand<U> defaultValue) {
     OperationBuilder opBuilder = scope.env().opBuilder("LookupTableFindV2", scope.makeOpName("LookupTableFind"));
-    opBuilder.addInput(tableHandle.asOutput());
-    opBuilder.addInput(keys.asOutput());
-    opBuilder.addInput(defaultValue.asOutput());
+    opBuilder.addInput(tableHandle.asOutput(scope));
+    opBuilder.addInput(keys.asOutput(scope));
+    opBuilder.addInput(defaultValue.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new LookupTableFind<U>(opBuilder.build());
   }
@@ -69,7 +69,7 @@ public Output<U> values() {
   }
   
   @Override
-  public Output<U> asOutput() {
+  public Output<U> asOutput(Scope scope) {
     return values;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LookupTableImport.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LookupTableImport.java
index c5b9222bbf9..4b5121b7911 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LookupTableImport.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LookupTableImport.java
@@ -20,11 +20,11 @@
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Replaces the contents of the table with the specified keys and values.
@@ -45,11 +45,11 @@ public final class LookupTableImport extends RawOp {
    * @return a new instance of LookupTableImport
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor, U extends Tensor> LookupTableImport create(Scope scope, Operand<?> tableHandle, Operand<T> keys, Operand<U> values) {
+  public static <T extends TType, U extends TType> LookupTableImport create(Scope scope, Operand<?> tableHandle, Operand<T> keys, Operand<U> values) {
     OperationBuilder opBuilder = scope.env().opBuilder("LookupTableImportV2", scope.makeOpName("LookupTableImport"));
-    opBuilder.addInput(tableHandle.asOutput());
-    opBuilder.addInput(keys.asOutput());
-    opBuilder.addInput(values.asOutput());
+    opBuilder.addInput(tableHandle.asOutput(scope));
+    opBuilder.addInput(keys.asOutput(scope));
+    opBuilder.addInput(values.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new LookupTableImport(opBuilder.build());
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LookupTableInsert.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LookupTableInsert.java
index 3055286604b..6276b435130 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LookupTableInsert.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LookupTableInsert.java
@@ -20,11 +20,11 @@
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Updates the table to associates keys with values.
@@ -45,11 +45,11 @@ public final class LookupTableInsert extends RawOp {
    * @return a new instance of LookupTableInsert
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor, U extends Tensor> LookupTableInsert create(Scope scope, Operand<?> tableHandle, Operand<T> keys, Operand<U> values) {
+  public static <T extends TType, U extends TType> LookupTableInsert create(Scope scope, Operand<?> tableHandle, Operand<T> keys, Operand<U> values) {
     OperationBuilder opBuilder = scope.env().opBuilder("LookupTableInsertV2", scope.makeOpName("LookupTableInsert"));
-    opBuilder.addInput(tableHandle.asOutput());
-    opBuilder.addInput(keys.asOutput());
-    opBuilder.addInput(values.asOutput());
+    opBuilder.addInput(tableHandle.asOutput(scope));
+    opBuilder.addInput(keys.asOutput(scope));
+    opBuilder.addInput(values.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new LookupTableInsert(opBuilder.build());
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LookupTableRemove.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LookupTableRemove.java
index 54da587f516..ee86784c34f 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LookupTableRemove.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LookupTableRemove.java
@@ -20,11 +20,11 @@
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Removes keys and its associated values from a table.
@@ -43,10 +43,10 @@ public final class LookupTableRemove extends RawOp {
    * @return a new instance of LookupTableRemove
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> LookupTableRemove create(Scope scope, Operand<?> tableHandle, Operand<T> keys) {
+  public static <T extends TType> LookupTableRemove create(Scope scope, Operand<?> tableHandle, Operand<T> keys) {
     OperationBuilder opBuilder = scope.env().opBuilder("LookupTableRemoveV2", scope.makeOpName("LookupTableRemove"));
-    opBuilder.addInput(tableHandle.asOutput());
-    opBuilder.addInput(keys.asOutput());
+    opBuilder.addInput(tableHandle.asOutput(scope));
+    opBuilder.addInput(keys.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new LookupTableRemove(opBuilder.build());
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LookupTableSize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LookupTableSize.java
index 10e84953496..a362eb3c54c 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LookupTableSize.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LookupTableSize.java
@@ -43,7 +43,7 @@ public final class LookupTableSize extends RawOp implements Operand<TInt64> {
   @Endpoint(describeByClass = true)
   public static LookupTableSize create(Scope scope, Operand<?> tableHandle) {
     OperationBuilder opBuilder = scope.env().opBuilder("LookupTableSizeV2", scope.makeOpName("LookupTableSize"));
-    opBuilder.addInput(tableHandle.asOutput());
+    opBuilder.addInput(tableHandle.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new LookupTableSize(opBuilder.build());
   }
@@ -56,7 +56,7 @@ public Output<TInt64> size() {
   }
   
   @Override
-  public Output<TInt64> asOutput() {
+  public Output<TInt64> asOutput(Scope scope) {
     return size;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LoopCond.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LoopCond.java
index 57b82598f84..59ac003a6f7 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LoopCond.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LoopCond.java
@@ -46,7 +46,7 @@ public final class LoopCond extends RawOp implements Operand<TBool> {
   @Endpoint(describeByClass = true)
   public static LoopCond create(Scope scope, Operand<TBool> input) {
     OperationBuilder opBuilder = scope.env().opBuilder("LoopCond", scope.makeOpName("LoopCond"));
-    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new LoopCond(opBuilder.build());
   }
@@ -59,7 +59,7 @@ public Output<TBool> output() {
   }
   
   @Override
-  public Output<TBool> asOutput() {
+  public Output<TBool> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LowerBound.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LowerBound.java
index 89387991fba..8d6341857be 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LowerBound.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LowerBound.java
@@ -22,13 +22,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * Applies lower_bound(sorted_search_values, values) along each row.
@@ -53,7 +53,7 @@
  * 
  * @param <U> data type for {@code output()} output
  */
-public final class LowerBound<U extends Tensor & TNumber> extends RawOp implements Operand<U> {
+public final class LowerBound<U extends TNumber> extends RawOp implements Operand<U> {
   
   /**
    * Factory method to create a class wrapping a new LowerBound operation.
@@ -66,10 +66,10 @@ public final class LowerBound<U extends Tensor & TNumber> extends RawOp implemen
    * @return a new instance of LowerBound
    */
   @Endpoint(describeByClass = true)
-  public static <U extends Tensor & TNumber, T extends Tensor> LowerBound<U> create(Scope scope, Operand<T> sortedInputs, Operand<T> values, DataType<U> outType) {
+  public static <U extends TNumber, T extends TType> LowerBound<U> create(Scope scope, Operand<T> sortedInputs, Operand<T> values, DataType<U> outType) {
     OperationBuilder opBuilder = scope.env().opBuilder("LowerBound", scope.makeOpName("LowerBound"));
-    opBuilder.addInput(sortedInputs.asOutput());
-    opBuilder.addInput(values.asOutput());
+    opBuilder.addInput(sortedInputs.asOutput(scope));
+    opBuilder.addInput(values.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("out_type", outType);
     return new LowerBound<U>(opBuilder.build());
@@ -85,7 +85,7 @@ public static <U extends Tensor & TNumber, T extends Tensor> LowerBound<U> creat
    * @return a new instance of LowerBound
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> LowerBound<TInt32> create(Scope scope, Operand<T> sortedInputs, Operand<T> values) {
+  public static <T extends TType> LowerBound<TInt32> create(Scope scope, Operand<T> sortedInputs, Operand<T> values) {
     return create(scope, sortedInputs, values, TInt32.DTYPE);
   }
   
@@ -99,7 +99,7 @@ public Output<U> output() {
   }
   
   @Override
-  public Output<U> asOutput() {
+  public Output<U> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MapIncompleteSize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MapIncompleteSize.java
index 19e9e87a08a..18cc4e059b1 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MapIncompleteSize.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MapIncompleteSize.java
@@ -152,7 +152,7 @@ public Output<TInt32> size() {
   }
   
   @Override
-  public Output<TInt32> asOutput() {
+  public Output<TInt32> asOutput(Scope scope) {
     return size;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MapPeek.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MapPeek.java
index 61541aa0306..7408c06644e 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MapPeek.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MapPeek.java
@@ -25,13 +25,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
 import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
 
 /**
  * Op peeks at the values at the specified key.  If the
@@ -40,7 +40,7 @@
  * this op will block until it does.
  */
 @Operator
-public final class MapPeek extends RawOp implements Iterable<Operand<Tensor>> {
+public final class MapPeek extends RawOp implements Iterable<Operand<TType>> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.core.MapPeek}
@@ -101,8 +101,8 @@ private Options() {
   @Endpoint(describeByClass = true)
   public static MapPeek create(Scope scope, Operand<TInt64> key, Operand<TInt32> indices, List<DataType<?>> dtypes, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("MapPeek", scope.makeOpName("MapPeek"));
-    opBuilder.addInput(key.asOutput());
-    opBuilder.addInput(indices.asOutput());
+    opBuilder.addInput(key.asOutput(scope));
+    opBuilder.addInput(indices.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     DataType[] dtypesArray = new DataType[dtypes.size()];
     for (int i = 0; i < dtypesArray.length; ++i) {
@@ -164,7 +164,7 @@ public List<Output<?>> values() {
   
   @Override
   @SuppressWarnings({"rawtypes", "unchecked"})
-  public Iterator<Operand<Tensor>> iterator() {
+  public Iterator<Operand<TType>> iterator() {
     return (Iterator) values.iterator();
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MapSize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MapSize.java
index 7f4eea906f5..b6e467af2d3 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MapSize.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MapSize.java
@@ -152,7 +152,7 @@ public Output<TInt32> size() {
   }
   
   @Override
-  public Output<TInt32> asOutput() {
+  public Output<TInt32> asOutput(Scope scope) {
     return size;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MapStage.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MapStage.java
index 9291b32d53b..ebbdd445b62 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MapStage.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MapStage.java
@@ -99,9 +99,9 @@ private Options() {
   @Endpoint(describeByClass = true)
   public static MapStage create(Scope scope, Operand<TInt64> key, Operand<TInt32> indices, Iterable<Operand<?>> values, List<DataType<?>> dtypes, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("MapStage", scope.makeOpName("MapStage"));
-    opBuilder.addInput(key.asOutput());
-    opBuilder.addInput(indices.asOutput());
-    opBuilder.addInputList(Operands.asOutputs(values));
+    opBuilder.addInput(key.asOutput(scope));
+    opBuilder.addInput(indices.asOutput(scope));
+    opBuilder.addInputList(Operands.asOutputs(scope, values));
     opBuilder = scope.applyControlDependencies(opBuilder);
     DataType[] dtypesArray = new DataType[dtypes.size()];
     for (int i = 0; i < dtypesArray.length; ++i) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MapUnstage.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MapUnstage.java
index 9f83b8b3346..0d8e9d854d7 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MapUnstage.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MapUnstage.java
@@ -25,13 +25,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
 import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
 
 /**
  * Op removes and returns the values associated with the key
@@ -40,7 +40,7 @@
  * does not contain this key, the op will block until it does.
  */
 @Operator
-public final class MapUnstage extends RawOp implements Iterable<Operand<Tensor>> {
+public final class MapUnstage extends RawOp implements Iterable<Operand<TType>> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.core.MapUnstage}
@@ -101,8 +101,8 @@ private Options() {
   @Endpoint(describeByClass = true)
   public static MapUnstage create(Scope scope, Operand<TInt64> key, Operand<TInt32> indices, List<DataType<?>> dtypes, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("MapUnstage", scope.makeOpName("MapUnstage"));
-    opBuilder.addInput(key.asOutput());
-    opBuilder.addInput(indices.asOutput());
+    opBuilder.addInput(key.asOutput(scope));
+    opBuilder.addInput(indices.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     DataType[] dtypesArray = new DataType[dtypes.size()];
     for (int i = 0; i < dtypesArray.length; ++i) {
@@ -164,7 +164,7 @@ public List<Output<?>> values() {
   
   @Override
   @SuppressWarnings({"rawtypes", "unchecked"})
-  public Iterator<Operand<Tensor>> iterator() {
+  public Iterator<Operand<TType>> iterator() {
     return (Iterator) values.iterator();
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MapUnstageNoKey.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MapUnstageNoKey.java
index 10a119ec6c4..58e49e3616f 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MapUnstageNoKey.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MapUnstageNoKey.java
@@ -98,7 +98,7 @@ private Options() {
   @Endpoint(describeByClass = true)
   public static MapUnstageNoKey create(Scope scope, Operand<TInt32> indices, List<DataType<?>> dtypes, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("MapUnstageNoKey", scope.makeOpName("MapUnstageNoKey"));
-    opBuilder.addInput(indices.asOutput());
+    opBuilder.addInput(indices.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     DataType[] dtypesArray = new DataType[dtypes.size()];
     for (int i = 0; i < dtypesArray.length; ++i) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Max.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Max.java
index e26de0bcdc7..84a862f4435 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Max.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Max.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * Computes the maximum of elements across dimensions of a tensor.
@@ -39,7 +39,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator
-public final class Max<T extends Tensor> extends RawOp implements Operand<T> {
+public final class Max<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.core.Max}
@@ -71,10 +71,10 @@ private Options() {
    * @return a new instance of Max
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor, U extends Tensor & TNumber> Max<T> create(Scope scope, Operand<T> input, Operand<U> axis, Options... options) {
+  public static <T extends TType, U extends TNumber> Max<T> create(Scope scope, Operand<T> input, Operand<U> axis, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("Max", scope.makeOpName("Max"));
-    opBuilder.addInput(input.asOutput());
-    opBuilder.addInput(axis.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
+    opBuilder.addInput(axis.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -101,7 +101,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Merge.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Merge.java
index 6d70f5cbe93..7b6247f50f9 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Merge.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Merge.java
@@ -21,13 +21,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.Operands;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TType;
 
 /**
  * Forwards the value of an available tensor from `inputs` to `output`.
@@ -41,7 +41,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator
-public final class Merge<T extends Tensor> extends RawOp {
+public final class Merge<T extends TType> extends RawOp {
   
   /**
    * Factory method to create a class wrapping a new Merge operation.
@@ -51,9 +51,9 @@ public final class Merge<T extends Tensor> extends RawOp {
    * @return a new instance of Merge
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> Merge<T> create(Scope scope, Iterable<Operand<T>> inputs) {
+  public static <T extends TType> Merge<T> create(Scope scope, Iterable<Operand<T>> inputs) {
     OperationBuilder opBuilder = scope.env().opBuilder("Merge", scope.makeOpName("Merge"));
-    opBuilder.addInputList(Operands.asOutputs(inputs));
+    opBuilder.addInputList(Operands.asOutputs(scope, inputs));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new Merge<T>(opBuilder.build());
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Min.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Min.java
index 0052104ab5c..a6772fd009c 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Min.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Min.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * Computes the minimum of elements across dimensions of a tensor.
@@ -39,7 +39,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator
-public final class Min<T extends Tensor> extends RawOp implements Operand<T> {
+public final class Min<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.core.Min}
@@ -71,10 +71,10 @@ private Options() {
    * @return a new instance of Min
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor, U extends Tensor & TNumber> Min<T> create(Scope scope, Operand<T> input, Operand<U> axis, Options... options) {
+  public static <T extends TType, U extends TNumber> Min<T> create(Scope scope, Operand<T> input, Operand<U> axis, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("Min", scope.makeOpName("Min"));
-    opBuilder.addInput(input.asOutput());
-    opBuilder.addInput(axis.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
+    opBuilder.addInput(axis.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -101,7 +101,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MirrorPad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MirrorPad.java
index f54b533b0b2..9a10a23b155 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MirrorPad.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MirrorPad.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * Pads a tensor with mirrored values.
@@ -60,7 +60,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator
-public final class MirrorPad<T extends Tensor> extends RawOp implements Operand<T> {
+public final class MirrorPad<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new MirrorPad operation.
@@ -77,10 +77,10 @@ public final class MirrorPad<T extends Tensor> extends RawOp implements Operand<
    * @return a new instance of MirrorPad
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor, U extends Tensor & TNumber> MirrorPad<T> create(Scope scope, Operand<T> input, Operand<U> paddings, String mode) {
+  public static <T extends TType, U extends TNumber> MirrorPad<T> create(Scope scope, Operand<T> input, Operand<U> paddings, String mode) {
     OperationBuilder opBuilder = scope.env().opBuilder("MirrorPad", scope.makeOpName("MirrorPad"));
-    opBuilder.addInput(input.asOutput());
-    opBuilder.addInput(paddings.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
+    opBuilder.addInput(paddings.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("mode", mode);
     return new MirrorPad<T>(opBuilder.build());
@@ -94,7 +94,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MirrorPadGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MirrorPadGrad.java
index 1285a8b183d..71a93147bd0 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MirrorPadGrad.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MirrorPadGrad.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * Gradient op for `MirrorPad` op. This op folds a mirror-padded tensor.
@@ -52,7 +52,7 @@
  * 
  * @param <T> data type for {@code output()} output
  */
-public final class MirrorPadGrad<T extends Tensor> extends RawOp implements Operand<T> {
+public final class MirrorPadGrad<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new MirrorPadGrad operation.
@@ -65,10 +65,10 @@ public final class MirrorPadGrad<T extends Tensor> extends RawOp implements Oper
    * @return a new instance of MirrorPadGrad
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor, U extends Tensor & TNumber> MirrorPadGrad<T> create(Scope scope, Operand<T> input, Operand<U> paddings, String mode) {
+  public static <T extends TType, U extends TNumber> MirrorPadGrad<T> create(Scope scope, Operand<T> input, Operand<U> paddings, String mode) {
     OperationBuilder opBuilder = scope.env().opBuilder("MirrorPadGrad", scope.makeOpName("MirrorPadGrad"));
-    opBuilder.addInput(input.asOutput());
-    opBuilder.addInput(paddings.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
+    opBuilder.addInput(paddings.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("mode", mode);
     return new MirrorPadGrad<T>(opBuilder.build());
@@ -82,7 +82,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MlirPassthroughOp.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MlirPassthroughOp.java
index 3898a019645..0dcd4a25d20 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MlirPassthroughOp.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MlirPassthroughOp.java
@@ -25,12 +25,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.Operands;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Wraps an arbitrary MLIR computation expressed as a module with a main() function.
@@ -66,7 +66,7 @@
  * 
  */
 @Operator
-public final class MlirPassthroughOp extends RawOp implements Iterable<Operand<Tensor>> {
+public final class MlirPassthroughOp extends RawOp implements Iterable<Operand<TType>> {
   
   /**
    * Factory method to create a class wrapping a new MlirPassthroughOp operation.
@@ -80,7 +80,7 @@ public final class MlirPassthroughOp extends RawOp implements Iterable<Operand<T
   @Endpoint(describeByClass = true)
   public static MlirPassthroughOp create(Scope scope, Iterable<Operand<?>> inputs, String mlirModule, List<DataType<?>> Toutputs) {
     OperationBuilder opBuilder = scope.env().opBuilder("MlirPassthroughOp", scope.makeOpName("MlirPassthroughOp"));
-    opBuilder.addInputList(Operands.asOutputs(inputs));
+    opBuilder.addInputList(Operands.asOutputs(scope, inputs));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("mlir_module", mlirModule);
     DataType[] ToutputsArray = new DataType[Toutputs.size()];
@@ -99,7 +99,7 @@ public List<Output<?>> outputs() {
   
   @Override
   @SuppressWarnings({"rawtypes", "unchecked"})
-  public Iterator<Operand<Tensor>> iterator() {
+  public Iterator<Operand<TType>> iterator() {
     return (Iterator) outputs.iterator();
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MutableDenseHashTable.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MutableDenseHashTable.java
index 1c21813f1aa..b9fa39dc7c1 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MutableDenseHashTable.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MutableDenseHashTable.java
@@ -22,12 +22,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Creates an empty hash table that uses tensors as the backing store.
@@ -40,7 +40,7 @@
  * the insert operations. It does not support the initialization operation.
  */
 @Operator
-public final class MutableDenseHashTable extends RawOp implements Operand<Tensor> {
+public final class MutableDenseHashTable extends RawOp implements Operand<TType> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.core.MutableDenseHashTable}
@@ -122,10 +122,10 @@ private Options() {
    * @return a new instance of MutableDenseHashTable
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor, U extends Tensor> MutableDenseHashTable create(Scope scope, Operand<T> emptyKey, Operand<T> deletedKey, DataType<U> valueDtype, Options... options) {
+  public static <T extends TType, U extends TType> MutableDenseHashTable create(Scope scope, Operand<T> emptyKey, Operand<T> deletedKey, DataType<U> valueDtype, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("MutableDenseHashTableV2", scope.makeOpName("MutableDenseHashTable"));
-    opBuilder.addInput(emptyKey.asOutput());
-    opBuilder.addInput(deletedKey.asOutput());
+    opBuilder.addInput(emptyKey.asOutput(scope));
+    opBuilder.addInput(deletedKey.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("value_dtype", valueDtype);
     if (options != null) {
@@ -208,8 +208,8 @@ public Output<?> tableHandle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<Tensor> asOutput() {
-    return (Output<Tensor>) tableHandle;
+  public Output<TType> asOutput(Scope scope) {
+    return (Output<TType>) tableHandle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MutableHashTable.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MutableHashTable.java
index b8ff6808125..b108d397980 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MutableHashTable.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MutableHashTable.java
@@ -22,11 +22,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Creates an empty hash table.
@@ -36,7 +36,7 @@
  * the insert operations. It does not support the initialization operation.
  */
 @Operator
-public final class MutableHashTable extends RawOp implements Operand<Tensor> {
+public final class MutableHashTable extends RawOp implements Operand<TType> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.core.MutableHashTable}
@@ -88,7 +88,7 @@ private Options() {
    * @return a new instance of MutableHashTable
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor, U extends Tensor> MutableHashTable create(Scope scope, DataType<T> keyDtype, DataType<U> valueDtype, Options... options) {
+  public static <T extends TType, U extends TType> MutableHashTable create(Scope scope, DataType<T> keyDtype, DataType<U> valueDtype, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("MutableHashTableV2", scope.makeOpName("MutableHashTable"));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("key_dtype", keyDtype);
@@ -142,8 +142,8 @@ public Output<?> tableHandle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<Tensor> asOutput() {
-    return (Output<Tensor>) tableHandle;
+  public Output<TType> asOutput(Scope scope) {
+    return (Output<TType>) tableHandle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MutableHashTableOfTensors.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MutableHashTableOfTensors.java
index b444f3ffc68..5ad501f5175 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MutableHashTableOfTensors.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MutableHashTableOfTensors.java
@@ -22,12 +22,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Creates an empty hash table.
@@ -37,7 +37,7 @@
  * the insert operations. It does not support the initialization operation.
  */
 @Operator
-public final class MutableHashTableOfTensors extends RawOp implements Operand<Tensor> {
+public final class MutableHashTableOfTensors extends RawOp implements Operand<TType> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.core.MutableHashTableOfTensors}
@@ -97,7 +97,7 @@ private Options() {
    * @return a new instance of MutableHashTableOfTensors
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor, U extends Tensor> MutableHashTableOfTensors create(Scope scope, DataType<T> keyDtype, DataType<U> valueDtype, Options... options) {
+  public static <T extends TType, U extends TType> MutableHashTableOfTensors create(Scope scope, DataType<T> keyDtype, DataType<U> valueDtype, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("MutableHashTableOfTensorsV2", scope.makeOpName("MutableHashTableOfTensors"));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("key_dtype", keyDtype);
@@ -160,8 +160,8 @@ public Output<?> tableHandle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<Tensor> asOutput() {
-    return (Output<Tensor>) tableHandle;
+  public Output<TType> asOutput(Scope scope) {
+    return (Output<TType>) tableHandle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Mutex.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Mutex.java
index 20325657299..62c1434e39b 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Mutex.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Mutex.java
@@ -21,17 +21,17 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Creates a Mutex resource that can be locked by `MutexLock`.
  */
 @Operator
-public final class Mutex extends RawOp implements Operand<Tensor> {
+public final class Mutex extends RawOp implements Operand<TType> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.core.Mutex}
@@ -112,8 +112,8 @@ public Output<?> resource() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<Tensor> asOutput() {
-    return (Output<Tensor>) resource;
+  public Output<TType> asOutput(Scope scope) {
+    return (Output<TType>) resource;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MutexLock.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MutexLock.java
index ca11c7c739f..ac2927b03c9 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MutexLock.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MutexLock.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Locks a mutex resource.  The output is the lock.  So long as the lock tensor
@@ -67,7 +67,7 @@
  * wish to ensure the usage is exclusive.
  */
 @Operator
-public final class MutexLock extends RawOp implements Operand<Tensor> {
+public final class MutexLock extends RawOp implements Operand<TType> {
   
   /**
    * Factory method to create a class wrapping a new MutexLock operation.
@@ -79,7 +79,7 @@ public final class MutexLock extends RawOp implements Operand<Tensor> {
   @Endpoint(describeByClass = true)
   public static MutexLock create(Scope scope, Operand<?> mutex) {
     OperationBuilder opBuilder = scope.env().opBuilder("MutexLock", scope.makeOpName("MutexLock"));
-    opBuilder.addInput(mutex.asOutput());
+    opBuilder.addInput(mutex.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new MutexLock(opBuilder.build());
   }
@@ -95,8 +95,8 @@ public Output<?> mutexLock() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<Tensor> asOutput() {
-    return (Output<Tensor>) mutexLock;
+  public Output<TType> asOutput(Scope scope) {
+    return (Output<TType>) mutexLock;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/NcclAllReduce.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/NcclAllReduce.java
index fd293049078..a13de0e82ab 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/NcclAllReduce.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/NcclAllReduce.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -46,7 +45,7 @@
  * 
  * @param <T> data type for {@code output()} output
  */
-public final class NcclAllReduce<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class NcclAllReduce<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new NcclAllReduce operation.
@@ -59,9 +58,9 @@ public final class NcclAllReduce<T extends Tensor & TNumber> extends RawOp imple
    * @return a new instance of NcclAllReduce
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> NcclAllReduce<T> create(Scope scope, Operand<T> input, String reduction, Long numDevices, String sharedName) {
+  public static <T extends TNumber> NcclAllReduce<T> create(Scope scope, Operand<T> input, String reduction, Long numDevices, String sharedName) {
     OperationBuilder opBuilder = scope.env().opBuilder("NcclAllReduce", scope.makeOpName("NcclAllReduce"));
-    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("reduction", reduction);
     opBuilder.setAttr("num_devices", numDevices);
@@ -76,7 +75,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/NcclBroadcast.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/NcclBroadcast.java
index 0d321927ac7..c4b727efa7e 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/NcclBroadcast.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/NcclBroadcast.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
@@ -44,7 +43,7 @@
  * 
  * @param <T> data type for {@code output()} output
  */
-public final class NcclBroadcast<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class NcclBroadcast<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new NcclBroadcast operation.
@@ -55,9 +54,9 @@ public final class NcclBroadcast<T extends Tensor & TNumber> extends RawOp imple
    * @return a new instance of NcclBroadcast
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> NcclBroadcast<T> create(Scope scope, Operand<T> input, Shape shape) {
+  public static <T extends TNumber> NcclBroadcast<T> create(Scope scope, Operand<T> input, Shape shape) {
     OperationBuilder opBuilder = scope.env().opBuilder("NcclBroadcast", scope.makeOpName("NcclBroadcast"));
-    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("shape", shape);
     return new NcclBroadcast<T>(opBuilder.build());
@@ -70,7 +69,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/NcclReduce.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/NcclReduce.java
index d7ff17f2243..fecac09710d 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/NcclReduce.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/NcclReduce.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.Operands;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
@@ -43,7 +42,7 @@
  * 
  * @param <T> data type for {@code output()} output
  */
-public final class NcclReduce<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class NcclReduce<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new NcclReduce operation.
@@ -54,9 +53,9 @@ public final class NcclReduce<T extends Tensor & TNumber> extends RawOp implemen
    * @return a new instance of NcclReduce
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> NcclReduce<T> create(Scope scope, Iterable<Operand<T>> input, String reduction) {
+  public static <T extends TNumber> NcclReduce<T> create(Scope scope, Iterable<Operand<T>> input, String reduction) {
     OperationBuilder opBuilder = scope.env().opBuilder("NcclReduce", scope.makeOpName("NcclReduce"));
-    opBuilder.addInputList(Operands.asOutputs(input));
+    opBuilder.addInputList(Operands.asOutputs(scope, input));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("reduction", reduction);
     return new NcclReduce<T>(opBuilder.build());
@@ -69,7 +68,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/NextIteration.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/NextIteration.java
index ab2fb154b18..d246bdfd764 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/NextIteration.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/NextIteration.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Makes its input available to the next iteration.
@@ -33,7 +33,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator
-public final class NextIteration<T extends Tensor> extends RawOp implements Operand<T> {
+public final class NextIteration<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new NextIteration operation.
@@ -43,9 +43,9 @@ public final class NextIteration<T extends Tensor> extends RawOp implements Oper
    * @return a new instance of NextIteration
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> NextIteration<T> create(Scope scope, Operand<T> data) {
+  public static <T extends TType> NextIteration<T> create(Scope scope, Operand<T> data) {
     OperationBuilder opBuilder = scope.env().opBuilder("NextIteration", scope.makeOpName("NextIteration"));
-    opBuilder.addInput(data.asOutput());
+    opBuilder.addInput(data.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new NextIteration<T>(opBuilder.build());
   }
@@ -58,7 +58,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OneHot.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OneHot.java
index ba8a1ff3290..7969d18bd4d 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OneHot.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OneHot.java
@@ -21,13 +21,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * Returns a one-hot tensor.
@@ -116,7 +116,7 @@
  * @param <U> data type for {@code output()} output
  */
 @Operator
-public final class OneHot<U extends Tensor> extends RawOp implements Operand<U> {
+public final class OneHot<U extends TType> extends RawOp implements Operand<U> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.core.OneHot}
@@ -149,12 +149,12 @@ private Options() {
    * @return a new instance of OneHot
    */
   @Endpoint(describeByClass = true)
-  public static <U extends Tensor, T extends Tensor & TNumber> OneHot<U> create(Scope scope, Operand<T> indices, Operand<TInt32> depth, Operand<U> onValue, Operand<U> offValue, Options... options) {
+  public static <U extends TType, T extends TNumber> OneHot<U> create(Scope scope, Operand<T> indices, Operand<TInt32> depth, Operand<U> onValue, Operand<U> offValue, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("OneHot", scope.makeOpName("OneHot"));
-    opBuilder.addInput(indices.asOutput());
-    opBuilder.addInput(depth.asOutput());
-    opBuilder.addInput(onValue.asOutput());
-    opBuilder.addInput(offValue.asOutput());
+    opBuilder.addInput(indices.asOutput(scope));
+    opBuilder.addInput(depth.asOutput(scope));
+    opBuilder.addInput(onValue.asOutput(scope));
+    opBuilder.addInput(offValue.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -181,7 +181,7 @@ public Output<U> output() {
   }
   
   @Override
-  public Output<U> asOutput() {
+  public Output<U> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OnesLike.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OnesLike.java
index d565be08723..53255e2fc85 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OnesLike.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OnesLike.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Returns a tensor of ones with the same shape and type as x.
@@ -33,7 +33,7 @@
  * @param <T> data type for {@code y()} output
  */
 @Operator
-public final class OnesLike<T extends Tensor> extends RawOp implements Operand<T> {
+public final class OnesLike<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new OnesLike operation.
@@ -43,9 +43,9 @@ public final class OnesLike<T extends Tensor> extends RawOp implements Operand<T
    * @return a new instance of OnesLike
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> OnesLike<T> create(Scope scope, Operand<T> x) {
+  public static <T extends TType> OnesLike<T> create(Scope scope, Operand<T> x) {
     OperationBuilder opBuilder = scope.env().opBuilder("OnesLike", scope.makeOpName("OnesLike"));
-    opBuilder.addInput(x.asOutput());
+    opBuilder.addInput(x.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new OnesLike<T>(opBuilder.build());
   }
@@ -58,7 +58,7 @@ public Output<T> y() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return y;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OrderedMapIncompleteSize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OrderedMapIncompleteSize.java
index 865810568db..b92e8cb4273 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OrderedMapIncompleteSize.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OrderedMapIncompleteSize.java
@@ -152,7 +152,7 @@ public Output<TInt32> size() {
   }
   
   @Override
-  public Output<TInt32> asOutput() {
+  public Output<TInt32> asOutput(Scope scope) {
     return size;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OrderedMapPeek.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OrderedMapPeek.java
index 80cd39960c0..45ae4630076 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OrderedMapPeek.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OrderedMapPeek.java
@@ -25,13 +25,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
 import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
 
 /**
  * Op peeks at the values at the specified key.  If the
@@ -41,7 +41,7 @@
  * performance.
  */
 @Operator
-public final class OrderedMapPeek extends RawOp implements Iterable<Operand<Tensor>> {
+public final class OrderedMapPeek extends RawOp implements Iterable<Operand<TType>> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.core.OrderedMapPeek}
@@ -102,8 +102,8 @@ private Options() {
   @Endpoint(describeByClass = true)
   public static OrderedMapPeek create(Scope scope, Operand<TInt64> key, Operand<TInt32> indices, List<DataType<?>> dtypes, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("OrderedMapPeek", scope.makeOpName("OrderedMapPeek"));
-    opBuilder.addInput(key.asOutput());
-    opBuilder.addInput(indices.asOutput());
+    opBuilder.addInput(key.asOutput(scope));
+    opBuilder.addInput(indices.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     DataType[] dtypesArray = new DataType[dtypes.size()];
     for (int i = 0; i < dtypesArray.length; ++i) {
@@ -165,7 +165,7 @@ public List<Output<?>> values() {
   
   @Override
   @SuppressWarnings({"rawtypes", "unchecked"})
-  public Iterator<Operand<Tensor>> iterator() {
+  public Iterator<Operand<TType>> iterator() {
     return (Iterator) values.iterator();
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OrderedMapSize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OrderedMapSize.java
index afdee7de1bd..d818185302d 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OrderedMapSize.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OrderedMapSize.java
@@ -152,7 +152,7 @@ public Output<TInt32> size() {
   }
   
   @Override
-  public Output<TInt32> asOutput() {
+  public Output<TInt32> asOutput(Scope scope) {
     return size;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OrderedMapStage.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OrderedMapStage.java
index 7e02973e3c6..523fabafa6e 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OrderedMapStage.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OrderedMapStage.java
@@ -101,9 +101,9 @@ private Options() {
   @Endpoint(describeByClass = true)
   public static OrderedMapStage create(Scope scope, Operand<TInt64> key, Operand<TInt32> indices, Iterable<Operand<?>> values, List<DataType<?>> dtypes, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("OrderedMapStage", scope.makeOpName("OrderedMapStage"));
-    opBuilder.addInput(key.asOutput());
-    opBuilder.addInput(indices.asOutput());
-    opBuilder.addInputList(Operands.asOutputs(values));
+    opBuilder.addInput(key.asOutput(scope));
+    opBuilder.addInput(indices.asOutput(scope));
+    opBuilder.addInputList(Operands.asOutputs(scope, values));
     opBuilder = scope.applyControlDependencies(opBuilder);
     DataType[] dtypesArray = new DataType[dtypes.size()];
     for (int i = 0; i < dtypesArray.length; ++i) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OrderedMapUnstage.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OrderedMapUnstage.java
index d5e8769b433..b9289984e5f 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OrderedMapUnstage.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OrderedMapUnstage.java
@@ -25,13 +25,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
 import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
 
 /**
  * Op removes and returns the values associated with the key
@@ -40,7 +40,7 @@
  * does not contain this key, the op will block until it does.
  */
 @Operator
-public final class OrderedMapUnstage extends RawOp implements Iterable<Operand<Tensor>> {
+public final class OrderedMapUnstage extends RawOp implements Iterable<Operand<TType>> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.core.OrderedMapUnstage}
@@ -101,8 +101,8 @@ private Options() {
   @Endpoint(describeByClass = true)
   public static OrderedMapUnstage create(Scope scope, Operand<TInt64> key, Operand<TInt32> indices, List<DataType<?>> dtypes, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("OrderedMapUnstage", scope.makeOpName("OrderedMapUnstage"));
-    opBuilder.addInput(key.asOutput());
-    opBuilder.addInput(indices.asOutput());
+    opBuilder.addInput(key.asOutput(scope));
+    opBuilder.addInput(indices.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     DataType[] dtypesArray = new DataType[dtypes.size()];
     for (int i = 0; i < dtypesArray.length; ++i) {
@@ -164,7 +164,7 @@ public List<Output<?>> values() {
   
   @Override
   @SuppressWarnings({"rawtypes", "unchecked"})
-  public Iterator<Operand<Tensor>> iterator() {
+  public Iterator<Operand<TType>> iterator() {
     return (Iterator) values.iterator();
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OrderedMapUnstageNoKey.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OrderedMapUnstageNoKey.java
index f20a23b9806..c401e84c1c3 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OrderedMapUnstageNoKey.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OrderedMapUnstageNoKey.java
@@ -98,7 +98,7 @@ private Options() {
   @Endpoint(describeByClass = true)
   public static OrderedMapUnstageNoKey create(Scope scope, Operand<TInt32> indices, List<DataType<?>> dtypes, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("OrderedMapUnstageNoKey", scope.makeOpName("OrderedMapUnstageNoKey"));
-    opBuilder.addInput(indices.asOutput());
+    opBuilder.addInput(indices.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     DataType[] dtypesArray = new DataType[dtypes.size()];
     for (int i = 0; i < dtypesArray.length; ++i) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Pad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Pad.java
index 27915f9670f..bcb29dfb937 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Pad.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Pad.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * Pads a tensor.
@@ -59,7 +59,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator
-public final class Pad<T extends Tensor> extends RawOp implements Operand<T> {
+public final class Pad<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Pad operation.
@@ -71,11 +71,11 @@ public final class Pad<T extends Tensor> extends RawOp implements Operand<T> {
    * @return a new instance of Pad
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor, U extends Tensor & TNumber> Pad<T> create(Scope scope, Operand<T> input, Operand<U> paddings, Operand<T> constantValues) {
+  public static <T extends TType, U extends TNumber> Pad<T> create(Scope scope, Operand<T> input, Operand<U> paddings, Operand<T> constantValues) {
     OperationBuilder opBuilder = scope.env().opBuilder("PadV2", scope.makeOpName("Pad"));
-    opBuilder.addInput(input.asOutput());
-    opBuilder.addInput(paddings.asOutput());
-    opBuilder.addInput(constantValues.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
+    opBuilder.addInput(paddings.asOutput(scope));
+    opBuilder.addInput(constantValues.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new Pad<T>(opBuilder.build());
   }
@@ -87,7 +87,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ParallelConcat.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ParallelConcat.java
index e3675988da6..fdc0b3a6763 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ParallelConcat.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ParallelConcat.java
@@ -21,13 +21,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.Operands;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Concatenates a list of `N` tensors along the first dimension.
@@ -50,7 +50,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator
-public final class ParallelConcat<T extends Tensor> extends RawOp implements Operand<T> {
+public final class ParallelConcat<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new ParallelConcat operation.
@@ -63,9 +63,9 @@ public final class ParallelConcat<T extends Tensor> extends RawOp implements Ope
    * @return a new instance of ParallelConcat
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> ParallelConcat<T> create(Scope scope, Iterable<Operand<T>> values, Shape shape) {
+  public static <T extends TType> ParallelConcat<T> create(Scope scope, Iterable<Operand<T>> values, Shape shape) {
     OperationBuilder opBuilder = scope.env().opBuilder("ParallelConcat", scope.makeOpName("ParallelConcat"));
-    opBuilder.addInputList(Operands.asOutputs(values));
+    opBuilder.addInputList(Operands.asOutputs(scope, values));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("shape", shape);
     return new ParallelConcat<T>(opBuilder.build());
@@ -79,7 +79,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ParallelDynamicStitch.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ParallelDynamicStitch.java
index 50cf7aaf7b9..f6a4e709687 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ParallelDynamicStitch.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ParallelDynamicStitch.java
@@ -21,13 +21,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.Operands;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TType;
 
 /**
  * Interleave the values from the `data` tensors into a single tensor.
@@ -89,7 +89,7 @@
  * @param <T> data type for {@code merged()} output
  */
 @Operator
-public final class ParallelDynamicStitch<T extends Tensor> extends RawOp implements Operand<T> {
+public final class ParallelDynamicStitch<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new ParallelDynamicStitch operation.
@@ -100,10 +100,10 @@ public final class ParallelDynamicStitch<T extends Tensor> extends RawOp impleme
    * @return a new instance of ParallelDynamicStitch
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> ParallelDynamicStitch<T> create(Scope scope, Iterable<Operand<TInt32>> indices, Iterable<Operand<T>> data) {
+  public static <T extends TType> ParallelDynamicStitch<T> create(Scope scope, Iterable<Operand<TInt32>> indices, Iterable<Operand<T>> data) {
     OperationBuilder opBuilder = scope.env().opBuilder("ParallelDynamicStitch", scope.makeOpName("ParallelDynamicStitch"));
-    opBuilder.addInputList(Operands.asOutputs(indices));
-    opBuilder.addInputList(Operands.asOutputs(data));
+    opBuilder.addInputList(Operands.asOutputs(scope, indices));
+    opBuilder.addInputList(Operands.asOutputs(scope, data));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new ParallelDynamicStitch<T>(opBuilder.build());
   }
@@ -115,7 +115,7 @@ public Output<T> merged() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return merged;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Placeholder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Placeholder.java
index dc96e6e9b36..b8bbafd09b6 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Placeholder.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Placeholder.java
@@ -22,12 +22,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * A placeholder op for a value that will be fed into the computation.
@@ -39,7 +39,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator
-public final class Placeholder<T extends Tensor> extends RawOp implements Operand<T> {
+public final class Placeholder<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.core.Placeholder}
@@ -70,7 +70,7 @@ private Options() {
    * @return a new instance of Placeholder
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> Placeholder<T> create(Scope scope, DataType<T> dtype, Options... options) {
+  public static <T extends TType> Placeholder<T> create(Scope scope, DataType<T> dtype, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("Placeholder", scope.makeOpName("Placeholder"));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("dtype", dtype);
@@ -100,7 +100,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/PlaceholderWithDefault.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/PlaceholderWithDefault.java
index 0fb4696509f..87d6691333a 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/PlaceholderWithDefault.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/PlaceholderWithDefault.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * A placeholder op that passes through `input` when its output is not fed.
@@ -34,7 +34,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator
-public final class PlaceholderWithDefault<T extends Tensor> extends RawOp implements Operand<T> {
+public final class PlaceholderWithDefault<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new PlaceholderWithDefault operation.
@@ -45,9 +45,9 @@ public final class PlaceholderWithDefault<T extends Tensor> extends RawOp implem
    * @return a new instance of PlaceholderWithDefault
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> PlaceholderWithDefault<T> create(Scope scope, Operand<T> input, Shape shape) {
+  public static <T extends TType> PlaceholderWithDefault<T> create(Scope scope, Operand<T> input, Shape shape) {
     OperationBuilder opBuilder = scope.env().opBuilder("PlaceholderWithDefault", scope.makeOpName("PlaceholderWithDefault"));
-    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("shape", shape);
     return new PlaceholderWithDefault<T>(opBuilder.build());
@@ -61,7 +61,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Print.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Print.java
index 52b933329a0..9406d259ee3 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Print.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Print.java
@@ -73,7 +73,7 @@ private Options() {
   @Endpoint(describeByClass = true)
   public static Print create(Scope scope, Operand<TString> input, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("PrintV2", scope.makeOpName("Print"));
-    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Prod.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Prod.java
index f56bceb7edd..16e401cb760 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Prod.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Prod.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * Computes the product of elements across dimensions of a tensor.
@@ -39,7 +39,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator
-public final class Prod<T extends Tensor> extends RawOp implements Operand<T> {
+public final class Prod<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.core.Prod}
@@ -71,10 +71,10 @@ private Options() {
    * @return a new instance of Prod
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor, U extends Tensor & TNumber> Prod<T> create(Scope scope, Operand<T> input, Operand<U> axis, Options... options) {
+  public static <T extends TType, U extends TNumber> Prod<T> create(Scope scope, Operand<T> input, Operand<U> axis, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("Prod", scope.makeOpName("Prod"));
-    opBuilder.addInput(input.asOutput());
-    opBuilder.addInput(axis.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
+    opBuilder.addInput(axis.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -101,7 +101,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/QuantizedReshape.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/QuantizedReshape.java
index fa59a2151c6..b2de966aa32 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/QuantizedReshape.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/QuantizedReshape.java
@@ -21,13 +21,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TFloat32;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * Reshapes a quantized tensor as per the Reshape op.
@@ -37,7 +37,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator
-public final class QuantizedReshape<T extends Tensor> extends RawOp {
+public final class QuantizedReshape<T extends TType> extends RawOp {
   
   /**
    * Factory method to create a class wrapping a new QuantizedReshape operation.
@@ -50,12 +50,12 @@ public final class QuantizedReshape<T extends Tensor> extends RawOp {
    * @return a new instance of QuantizedReshape
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor, U extends Tensor & TNumber> QuantizedReshape<T> create(Scope scope, Operand<T> tensor, Operand<U> shape, Operand<TFloat32> inputMin, Operand<TFloat32> inputMax) {
+  public static <T extends TType, U extends TNumber> QuantizedReshape<T> create(Scope scope, Operand<T> tensor, Operand<U> shape, Operand<TFloat32> inputMin, Operand<TFloat32> inputMax) {
     OperationBuilder opBuilder = scope.env().opBuilder("QuantizedReshape", scope.makeOpName("QuantizedReshape"));
-    opBuilder.addInput(tensor.asOutput());
-    opBuilder.addInput(shape.asOutput());
-    opBuilder.addInput(inputMin.asOutput());
-    opBuilder.addInput(inputMax.asOutput());
+    opBuilder.addInput(tensor.asOutput(scope));
+    opBuilder.addInput(shape.asOutput(scope));
+    opBuilder.addInput(inputMin.asOutput(scope));
+    opBuilder.addInput(inputMax.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new QuantizedReshape<T>(opBuilder.build());
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Range.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Range.java
index 04d0235dedd..528ea8017c5 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Range.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Range.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -46,7 +45,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator
-public final class Range<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class Range<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Range operation.
@@ -58,11 +57,11 @@ public final class Range<T extends Tensor & TNumber> extends RawOp implements Op
    * @return a new instance of Range
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> Range<T> create(Scope scope, Operand<T> start, Operand<T> limit, Operand<T> delta) {
+  public static <T extends TNumber> Range<T> create(Scope scope, Operand<T> start, Operand<T> limit, Operand<T> delta) {
     OperationBuilder opBuilder = scope.env().opBuilder("Range", scope.makeOpName("Range"));
-    opBuilder.addInput(start.asOutput());
-    opBuilder.addInput(limit.asOutput());
-    opBuilder.addInput(delta.asOutput());
+    opBuilder.addInput(start.asOutput(scope));
+    opBuilder.addInput(limit.asOutput(scope));
+    opBuilder.addInput(delta.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new Range<T>(opBuilder.build());
   }
@@ -75,7 +74,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Rank.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Rank.java
index 8a5fafed6aa..84fd0a9c755 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Rank.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Rank.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TType;
 
 /**
  * Returns the rank of a tensor.
@@ -54,9 +54,9 @@ public final class Rank extends RawOp implements Operand<TInt32> {
    * @return a new instance of Rank
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> Rank create(Scope scope, Operand<T> input) {
+  public static <T extends TType> Rank create(Scope scope, Operand<T> input) {
     OperationBuilder opBuilder = scope.env().opBuilder("Rank", scope.makeOpName("Rank"));
-    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new Rank(opBuilder.build());
   }
@@ -68,7 +68,7 @@ public Output<TInt32> output() {
   }
   
   @Override
-  public Output<TInt32> asOutput() {
+  public Output<TInt32> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ReadVariableOp.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ReadVariableOp.java
index 198303a6b79..2f23e6aed41 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ReadVariableOp.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ReadVariableOp.java
@@ -22,11 +22,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Reads the value of a variable.
@@ -41,7 +41,7 @@
  * @param <T> data type for {@code value()} output
  */
 @Operator
-public final class ReadVariableOp<T extends Tensor> extends RawOp implements Operand<T> {
+public final class ReadVariableOp<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new ReadVariableOp operation.
@@ -52,9 +52,9 @@ public final class ReadVariableOp<T extends Tensor> extends RawOp implements Ope
    * @return a new instance of ReadVariableOp
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> ReadVariableOp<T> create(Scope scope, Operand<?> resource, DataType<T> dtype) {
+  public static <T extends TType> ReadVariableOp<T> create(Scope scope, Operand<?> resource, DataType<T> dtype) {
     OperationBuilder opBuilder = scope.env().opBuilder("ReadVariableOp", scope.makeOpName("ReadVariableOp"));
-    opBuilder.addInput(resource.asOutput());
+    opBuilder.addInput(resource.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("dtype", dtype);
     return new ReadVariableOp<T>(opBuilder.build());
@@ -67,7 +67,7 @@ public Output<T> value() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return value;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Recv.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Recv.java
index 20f3835f6e1..1ebad338979 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Recv.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Recv.java
@@ -22,18 +22,18 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Receives the named tensor from send_device on recv_device.
  * 
  * @param <T> data type for {@code tensor()} output
  */
-public final class Recv<T extends Tensor> extends RawOp implements Operand<T> {
+public final class Recv<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.core.Recv}
@@ -70,7 +70,7 @@ private Options() {
    * @return a new instance of Recv
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> Recv<T> create(Scope scope, DataType<T> tensorType, String tensorName, String sendDevice, Long sendDeviceIncarnation, String recvDevice, Options... options) {
+  public static <T extends TType> Recv<T> create(Scope scope, DataType<T> tensorType, String tensorName, String sendDevice, Long sendDeviceIncarnation, String recvDevice, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("Recv", scope.makeOpName("Recv"));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("tensor_type", tensorType);
@@ -106,7 +106,7 @@ public Output<T> tensor() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return tensor;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ReduceAll.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ReduceAll.java
index 8b975f18afa..0bdb46ee38c 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ReduceAll.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ReduceAll.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -70,10 +69,10 @@ private Options() {
    * @return a new instance of ReduceAll
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> ReduceAll create(Scope scope, Operand<TBool> input, Operand<T> axis, Options... options) {
+  public static <T extends TNumber> ReduceAll create(Scope scope, Operand<TBool> input, Operand<T> axis, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("All", scope.makeOpName("ReduceAll"));
-    opBuilder.addInput(input.asOutput());
-    opBuilder.addInput(axis.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
+    opBuilder.addInput(axis.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -100,7 +99,7 @@ public Output<TBool> output() {
   }
   
   @Override
-  public Output<TBool> asOutput() {
+  public Output<TBool> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ReduceAny.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ReduceAny.java
index 28bda505e65..b54d2a5c18d 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ReduceAny.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ReduceAny.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -70,10 +69,10 @@ private Options() {
    * @return a new instance of ReduceAny
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> ReduceAny create(Scope scope, Operand<TBool> input, Operand<T> axis, Options... options) {
+  public static <T extends TNumber> ReduceAny create(Scope scope, Operand<TBool> input, Operand<T> axis, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("Any", scope.makeOpName("ReduceAny"));
-    opBuilder.addInput(input.asOutput());
-    opBuilder.addInput(axis.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
+    opBuilder.addInput(axis.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -100,7 +99,7 @@ public Output<TBool> output() {
   }
   
   @Override
-  public Output<TBool> asOutput() {
+  public Output<TBool> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ReduceMax.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ReduceMax.java
index c153b785f4c..355390df9c5 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ReduceMax.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ReduceMax.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * Computes the maximum of elements across dimensions of a tensor.
@@ -39,7 +39,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator
-public final class ReduceMax<T extends Tensor> extends RawOp implements Operand<T> {
+public final class ReduceMax<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.core.ReduceMax}
@@ -71,10 +71,10 @@ private Options() {
    * @return a new instance of ReduceMax
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor, U extends Tensor & TNumber> ReduceMax<T> create(Scope scope, Operand<T> input, Operand<U> axis, Options... options) {
+  public static <T extends TType, U extends TNumber> ReduceMax<T> create(Scope scope, Operand<T> input, Operand<U> axis, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("Max", scope.makeOpName("ReduceMax"));
-    opBuilder.addInput(input.asOutput());
-    opBuilder.addInput(axis.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
+    opBuilder.addInput(axis.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -101,7 +101,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ReduceMin.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ReduceMin.java
index 8f676a509f1..0e5b82c5267 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ReduceMin.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ReduceMin.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * Computes the minimum of elements across dimensions of a tensor.
@@ -39,7 +39,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator
-public final class ReduceMin<T extends Tensor> extends RawOp implements Operand<T> {
+public final class ReduceMin<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.core.ReduceMin}
@@ -71,10 +71,10 @@ private Options() {
    * @return a new instance of ReduceMin
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor, U extends Tensor & TNumber> ReduceMin<T> create(Scope scope, Operand<T> input, Operand<U> axis, Options... options) {
+  public static <T extends TType, U extends TNumber> ReduceMin<T> create(Scope scope, Operand<T> input, Operand<U> axis, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("Min", scope.makeOpName("ReduceMin"));
-    opBuilder.addInput(input.asOutput());
-    opBuilder.addInput(axis.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
+    opBuilder.addInput(axis.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -101,7 +101,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ReduceProd.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ReduceProd.java
index cc352b66784..779372c1079 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ReduceProd.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ReduceProd.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * Computes the product of elements across dimensions of a tensor.
@@ -39,7 +39,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator
-public final class ReduceProd<T extends Tensor> extends RawOp implements Operand<T> {
+public final class ReduceProd<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.core.ReduceProd}
@@ -71,10 +71,10 @@ private Options() {
    * @return a new instance of ReduceProd
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor, U extends Tensor & TNumber> ReduceProd<T> create(Scope scope, Operand<T> input, Operand<U> axis, Options... options) {
+  public static <T extends TType, U extends TNumber> ReduceProd<T> create(Scope scope, Operand<T> input, Operand<U> axis, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("Prod", scope.makeOpName("ReduceProd"));
-    opBuilder.addInput(input.asOutput());
-    opBuilder.addInput(axis.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
+    opBuilder.addInput(axis.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -101,7 +101,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ReduceSum.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ReduceSum.java
index d34cd8e448f..5cbce3b033c 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ReduceSum.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ReduceSum.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * Computes the sum of elements across dimensions of a tensor.
@@ -39,7 +39,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator
-public final class ReduceSum<T extends Tensor> extends RawOp implements Operand<T> {
+public final class ReduceSum<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.core.ReduceSum}
@@ -71,10 +71,10 @@ private Options() {
    * @return a new instance of ReduceSum
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor, U extends Tensor & TNumber> ReduceSum<T> create(Scope scope, Operand<T> input, Operand<U> axis, Options... options) {
+  public static <T extends TType, U extends TNumber> ReduceSum<T> create(Scope scope, Operand<T> input, Operand<U> axis, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("Sum", scope.makeOpName("ReduceSum"));
-    opBuilder.addInput(input.asOutput());
-    opBuilder.addInput(axis.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
+    opBuilder.addInput(axis.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -101,7 +101,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/RefEnter.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/RefEnter.java
index d6cfef23cab..f36104fc2dd 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/RefEnter.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/RefEnter.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Creates or finds a child frame, and makes `data` available to the child frame.
@@ -37,7 +37,7 @@
  * 
  * @param <T> data type for {@code output()} output
  */
-public final class RefEnter<T extends Tensor> extends RawOp implements Operand<T> {
+public final class RefEnter<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.core.RefEnter}
@@ -77,9 +77,9 @@ private Options() {
    * @return a new instance of RefEnter
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> RefEnter<T> create(Scope scope, Operand<T> data, String frameName, Options... options) {
+  public static <T extends TType> RefEnter<T> create(Scope scope, Operand<T> data, String frameName, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("RefEnter", scope.makeOpName("RefEnter"));
-    opBuilder.addInput(data.asOutput());
+    opBuilder.addInput(data.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("frame_name", frameName);
     if (options != null) {
@@ -117,7 +117,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/RefExit.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/RefExit.java
index 9cfeac60763..0438696ed02 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/RefExit.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/RefExit.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Exits the current frame to its parent frame.
@@ -34,7 +34,7 @@
  * 
  * @param <T> data type for {@code output()} output
  */
-public final class RefExit<T extends Tensor> extends RawOp implements Operand<T> {
+public final class RefExit<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new RefExit operation.
@@ -44,9 +44,9 @@ public final class RefExit<T extends Tensor> extends RawOp implements Operand<T>
    * @return a new instance of RefExit
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> RefExit<T> create(Scope scope, Operand<T> data) {
+  public static <T extends TType> RefExit<T> create(Scope scope, Operand<T> data) {
     OperationBuilder opBuilder = scope.env().opBuilder("RefExit", scope.makeOpName("RefExit"));
-    opBuilder.addInput(data.asOutput());
+    opBuilder.addInput(data.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new RefExit<T>(opBuilder.build());
   }
@@ -59,7 +59,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/RefIdentity.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/RefIdentity.java
index cd757f96b48..e3da108a249 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/RefIdentity.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/RefIdentity.java
@@ -21,18 +21,18 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Return the same ref tensor as the input ref tensor.
  * 
  * @param <T> data type for {@code output()} output
  */
-public final class RefIdentity<T extends Tensor> extends RawOp implements Operand<T> {
+public final class RefIdentity<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new RefIdentity operation.
@@ -42,9 +42,9 @@ public final class RefIdentity<T extends Tensor> extends RawOp implements Operan
    * @return a new instance of RefIdentity
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> RefIdentity<T> create(Scope scope, Operand<T> input) {
+  public static <T extends TType> RefIdentity<T> create(Scope scope, Operand<T> input) {
     OperationBuilder opBuilder = scope.env().opBuilder("RefIdentity", scope.makeOpName("RefIdentity"));
-    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new RefIdentity<T>(opBuilder.build());
   }
@@ -56,7 +56,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/RefMerge.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/RefMerge.java
index a8d2580fb7d..04ee46da33e 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/RefMerge.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/RefMerge.java
@@ -21,13 +21,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.Operands;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TType;
 
 /**
  * Forwards the value of an available tensor from `inputs` to `output`.
@@ -40,7 +40,7 @@
  * 
  * @param <T> data type for {@code output()} output
  */
-public final class RefMerge<T extends Tensor> extends RawOp {
+public final class RefMerge<T extends TType> extends RawOp {
   
   /**
    * Factory method to create a class wrapping a new RefMerge operation.
@@ -50,9 +50,9 @@ public final class RefMerge<T extends Tensor> extends RawOp {
    * @return a new instance of RefMerge
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> RefMerge<T> create(Scope scope, Iterable<Operand<T>> inputs) {
+  public static <T extends TType> RefMerge<T> create(Scope scope, Iterable<Operand<T>> inputs) {
     OperationBuilder opBuilder = scope.env().opBuilder("RefMerge", scope.makeOpName("RefMerge"));
-    opBuilder.addInputList(Operands.asOutputs(inputs));
+    opBuilder.addInputList(Operands.asOutputs(scope, inputs));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new RefMerge<T>(opBuilder.build());
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/RefNextIteration.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/RefNextIteration.java
index ad367e48118..935207485a4 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/RefNextIteration.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/RefNextIteration.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Makes its input available to the next iteration.
@@ -33,7 +33,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator
-public final class RefNextIteration<T extends Tensor> extends RawOp implements Operand<T> {
+public final class RefNextIteration<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new RefNextIteration operation.
@@ -43,9 +43,9 @@ public final class RefNextIteration<T extends Tensor> extends RawOp implements O
    * @return a new instance of RefNextIteration
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> RefNextIteration<T> create(Scope scope, Operand<T> data) {
+  public static <T extends TType> RefNextIteration<T> create(Scope scope, Operand<T> data) {
     OperationBuilder opBuilder = scope.env().opBuilder("RefNextIteration", scope.makeOpName("RefNextIteration"));
-    opBuilder.addInput(data.asOutput());
+    opBuilder.addInput(data.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new RefNextIteration<T>(opBuilder.build());
   }
@@ -58,7 +58,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/RefSelect.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/RefSelect.java
index 690a3e785d7..1678f30289f 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/RefSelect.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/RefSelect.java
@@ -21,13 +21,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.Operands;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TType;
 
 /**
  * Forwards the `index`th element of `inputs` to `output`.
@@ -35,7 +35,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator
-public final class RefSelect<T extends Tensor> extends RawOp implements Operand<T> {
+public final class RefSelect<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new RefSelect operation.
@@ -46,10 +46,10 @@ public final class RefSelect<T extends Tensor> extends RawOp implements Operand<
    * @return a new instance of RefSelect
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> RefSelect<T> create(Scope scope, Operand<TInt32> index, Iterable<Operand<T>> inputs) {
+  public static <T extends TType> RefSelect<T> create(Scope scope, Operand<TInt32> index, Iterable<Operand<T>> inputs) {
     OperationBuilder opBuilder = scope.env().opBuilder("RefSelect", scope.makeOpName("RefSelect"));
-    opBuilder.addInput(index.asOutput());
-    opBuilder.addInputList(Operands.asOutputs(inputs));
+    opBuilder.addInput(index.asOutput(scope));
+    opBuilder.addInputList(Operands.asOutputs(scope, inputs));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new RefSelect<T>(opBuilder.build());
   }
@@ -62,7 +62,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/RefSwitch.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/RefSwitch.java
index 8233a076458..be17ff3fb60 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/RefSwitch.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/RefSwitch.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TBool;
+import org.tensorflow.types.family.TType;
 
 /**
  * Forwards the ref tensor `data` to the output port determined by `pred`.
@@ -39,7 +39,7 @@
  * @param <T> data type for {@code outputFalse()} output
  */
 @Operator
-public final class RefSwitch<T extends Tensor> extends RawOp {
+public final class RefSwitch<T extends TType> extends RawOp {
   
   /**
    * Factory method to create a class wrapping a new RefSwitch operation.
@@ -50,10 +50,10 @@ public final class RefSwitch<T extends Tensor> extends RawOp {
    * @return a new instance of RefSwitch
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> RefSwitch<T> create(Scope scope, Operand<T> data, Operand<TBool> pred) {
+  public static <T extends TType> RefSwitch<T> create(Scope scope, Operand<T> data, Operand<TBool> pred) {
     OperationBuilder opBuilder = scope.env().opBuilder("RefSwitch", scope.makeOpName("RefSwitch"));
-    opBuilder.addInput(data.asOutput());
-    opBuilder.addInput(pred.asOutput());
+    opBuilder.addInput(data.asOutput(scope));
+    opBuilder.addInput(pred.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new RefSwitch<T>(opBuilder.build());
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/RemoteFusedGraphExecute.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/RemoteFusedGraphExecute.java
index 6b46ba055ab..5462c762ac0 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/RemoteFusedGraphExecute.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/RemoteFusedGraphExecute.java
@@ -25,12 +25,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.Operands;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Execute a sub graph on a remote processor.
@@ -44,7 +44,7 @@
  * will be passed to consumer nodes as outputs of this node.
  */
 @Operator
-public final class RemoteFusedGraphExecute extends RawOp implements Iterable<Operand<Tensor>> {
+public final class RemoteFusedGraphExecute extends RawOp implements Iterable<Operand<TType>> {
   
   /**
    * Factory method to create a class wrapping a new RemoteFusedGraphExecute operation.
@@ -59,7 +59,7 @@ public final class RemoteFusedGraphExecute extends RawOp implements Iterable<Ope
   @Endpoint(describeByClass = true)
   public static RemoteFusedGraphExecute create(Scope scope, Iterable<Operand<?>> inputs, List<DataType<?>> Toutputs, String serializedRemoteFusedGraphExecuteInfo) {
     OperationBuilder opBuilder = scope.env().opBuilder("RemoteFusedGraphExecute", scope.makeOpName("RemoteFusedGraphExecute"));
-    opBuilder.addInputList(Operands.asOutputs(inputs));
+    opBuilder.addInputList(Operands.asOutputs(scope, inputs));
     opBuilder = scope.applyControlDependencies(opBuilder);
     DataType[] ToutputsArray = new DataType[Toutputs.size()];
     for (int i = 0; i < ToutputsArray.length; ++i) {
@@ -79,7 +79,7 @@ public List<Output<?>> outputs() {
   
   @Override
   @SuppressWarnings({"rawtypes", "unchecked"})
-  public Iterator<Operand<Tensor>> iterator() {
+  public Iterator<Operand<TType>> iterator() {
     return (Iterator) outputs.iterator();
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Reshape.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Reshape.java
index 4f012475d4a..4e88d4d7f03 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Reshape.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Reshape.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * Reshapes a tensor.
@@ -94,7 +94,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator
-public final class Reshape<T extends Tensor> extends RawOp implements Operand<T> {
+public final class Reshape<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Reshape operation.
@@ -105,10 +105,10 @@ public final class Reshape<T extends Tensor> extends RawOp implements Operand<T>
    * @return a new instance of Reshape
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor, U extends Tensor & TNumber> Reshape<T> create(Scope scope, Operand<T> tensor, Operand<U> shape) {
+  public static <T extends TType, U extends TNumber> Reshape<T> create(Scope scope, Operand<T> tensor, Operand<U> shape) {
     OperationBuilder opBuilder = scope.env().opBuilder("Reshape", scope.makeOpName("Reshape"));
-    opBuilder.addInput(tensor.asOutput());
-    opBuilder.addInput(shape.asOutput());
+    opBuilder.addInput(tensor.asOutput(scope));
+    opBuilder.addInput(shape.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new Reshape<T>(opBuilder.build());
   }
@@ -120,7 +120,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceCountUpTo.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceCountUpTo.java
index 5278d579d3e..9070824ccc2 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceCountUpTo.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceCountUpTo.java
@@ -22,7 +22,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -35,7 +34,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator
-public final class ResourceCountUpTo<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class ResourceCountUpTo<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new ResourceCountUpTo operation.
@@ -48,9 +47,9 @@ public final class ResourceCountUpTo<T extends Tensor & TNumber> extends RawOp i
    * @return a new instance of ResourceCountUpTo
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> ResourceCountUpTo<T> create(Scope scope, Operand<?> resource, Long limit, DataType<T> T) {
+  public static <T extends TNumber> ResourceCountUpTo<T> create(Scope scope, Operand<?> resource, Long limit, DataType<T> T) {
     OperationBuilder opBuilder = scope.env().opBuilder("ResourceCountUpTo", scope.makeOpName("ResourceCountUpTo"));
-    opBuilder.addInput(resource.asOutput());
+    opBuilder.addInput(resource.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("limit", limit);
     opBuilder.setAttr("T", T);
@@ -66,7 +65,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceGather.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceGather.java
index 35974fdd095..f112b891109 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceGather.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceGather.java
@@ -22,12 +22,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * Gather slices from the variable pointed to by `resource` according to `indices`.
@@ -49,7 +49,7 @@
  * @param <U> data type for {@code output()} output
  */
 @Operator
-public final class ResourceGather<U extends Tensor> extends RawOp implements Operand<U> {
+public final class ResourceGather<U extends TType> extends RawOp implements Operand<U> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.core.ResourceGather}
@@ -90,10 +90,10 @@ private Options() {
    * @return a new instance of ResourceGather
    */
   @Endpoint(describeByClass = true)
-  public static <U extends Tensor, T extends Tensor & TNumber> ResourceGather<U> create(Scope scope, Operand<?> resource, Operand<T> indices, DataType<U> dtype, Options... options) {
+  public static <U extends TType, T extends TNumber> ResourceGather<U> create(Scope scope, Operand<?> resource, Operand<T> indices, DataType<U> dtype, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ResourceGather", scope.makeOpName("ResourceGather"));
-    opBuilder.addInput(resource.asOutput());
-    opBuilder.addInput(indices.asOutput());
+    opBuilder.addInput(resource.asOutput(scope));
+    opBuilder.addInput(indices.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("dtype", dtype);
     if (options != null) {
@@ -130,7 +130,7 @@ public Output<U> output() {
   }
   
   @Override
-  public Output<U> asOutput() {
+  public Output<U> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceGatherNd.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceGatherNd.java
index d43d682e05a..46120fa6249 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceGatherNd.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceGatherNd.java
@@ -22,18 +22,18 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * @param <U> data type for {@code output()} output
  */
 @Operator
-public final class ResourceGatherNd<U extends Tensor> extends RawOp implements Operand<U> {
+public final class ResourceGatherNd<U extends TType> extends RawOp implements Operand<U> {
   
   /**
    * Factory method to create a class wrapping a new ResourceGatherNd operation.
@@ -45,10 +45,10 @@ public final class ResourceGatherNd<U extends Tensor> extends RawOp implements O
    * @return a new instance of ResourceGatherNd
    */
   @Endpoint(describeByClass = true)
-  public static <U extends Tensor, T extends Tensor & TNumber> ResourceGatherNd<U> create(Scope scope, Operand<?> resource, Operand<T> indices, DataType<U> dtype) {
+  public static <U extends TType, T extends TNumber> ResourceGatherNd<U> create(Scope scope, Operand<?> resource, Operand<T> indices, DataType<U> dtype) {
     OperationBuilder opBuilder = scope.env().opBuilder("ResourceGatherNd", scope.makeOpName("ResourceGatherNd"));
-    opBuilder.addInput(resource.asOutput());
-    opBuilder.addInput(indices.asOutput());
+    opBuilder.addInput(resource.asOutput(scope));
+    opBuilder.addInput(indices.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("dtype", dtype);
     return new ResourceGatherNd<U>(opBuilder.build());
@@ -61,7 +61,7 @@ public Output<U> output() {
   }
   
   @Override
-  public Output<U> asOutput() {
+  public Output<U> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterAdd.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterAdd.java
index 61f9428edef..9091ceaf189 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterAdd.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterAdd.java
@@ -20,12 +20,12 @@
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * Adds sparse updates to the variable referenced by `resource`.
@@ -63,11 +63,11 @@ public final class ResourceScatterAdd extends RawOp {
    * @return a new instance of ResourceScatterAdd
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber, U extends Tensor> ResourceScatterAdd create(Scope scope, Operand<?> resource, Operand<T> indices, Operand<U> updates) {
+  public static <T extends TNumber, U extends TType> ResourceScatterAdd create(Scope scope, Operand<?> resource, Operand<T> indices, Operand<U> updates) {
     OperationBuilder opBuilder = scope.env().opBuilder("ResourceScatterAdd", scope.makeOpName("ResourceScatterAdd"));
-    opBuilder.addInput(resource.asOutput());
-    opBuilder.addInput(indices.asOutput());
-    opBuilder.addInput(updates.asOutput());
+    opBuilder.addInput(resource.asOutput(scope));
+    opBuilder.addInput(indices.asOutput(scope));
+    opBuilder.addInput(updates.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new ResourceScatterAdd(opBuilder.build());
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterDiv.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterDiv.java
index 3482a3886d3..3125cde19b0 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterDiv.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterDiv.java
@@ -20,12 +20,12 @@
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * Divides sparse updates into the variable referenced by `resource`.
@@ -63,11 +63,11 @@ public final class ResourceScatterDiv extends RawOp {
    * @return a new instance of ResourceScatterDiv
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber, U extends Tensor> ResourceScatterDiv create(Scope scope, Operand<?> resource, Operand<T> indices, Operand<U> updates) {
+  public static <T extends TNumber, U extends TType> ResourceScatterDiv create(Scope scope, Operand<?> resource, Operand<T> indices, Operand<U> updates) {
     OperationBuilder opBuilder = scope.env().opBuilder("ResourceScatterDiv", scope.makeOpName("ResourceScatterDiv"));
-    opBuilder.addInput(resource.asOutput());
-    opBuilder.addInput(indices.asOutput());
-    opBuilder.addInput(updates.asOutput());
+    opBuilder.addInput(resource.asOutput(scope));
+    opBuilder.addInput(indices.asOutput(scope));
+    opBuilder.addInput(updates.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new ResourceScatterDiv(opBuilder.build());
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterMax.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterMax.java
index 891d761acf8..f06e1862bfd 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterMax.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterMax.java
@@ -20,12 +20,12 @@
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * Reduces sparse updates into the variable referenced by `resource` using the `max` operation.
@@ -63,11 +63,11 @@ public final class ResourceScatterMax extends RawOp {
    * @return a new instance of ResourceScatterMax
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber, U extends Tensor> ResourceScatterMax create(Scope scope, Operand<?> resource, Operand<T> indices, Operand<U> updates) {
+  public static <T extends TNumber, U extends TType> ResourceScatterMax create(Scope scope, Operand<?> resource, Operand<T> indices, Operand<U> updates) {
     OperationBuilder opBuilder = scope.env().opBuilder("ResourceScatterMax", scope.makeOpName("ResourceScatterMax"));
-    opBuilder.addInput(resource.asOutput());
-    opBuilder.addInput(indices.asOutput());
-    opBuilder.addInput(updates.asOutput());
+    opBuilder.addInput(resource.asOutput(scope));
+    opBuilder.addInput(indices.asOutput(scope));
+    opBuilder.addInput(updates.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new ResourceScatterMax(opBuilder.build());
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterMin.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterMin.java
index 2d083c02f2a..b7f4ff1e51c 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterMin.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterMin.java
@@ -20,12 +20,12 @@
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * Reduces sparse updates into the variable referenced by `resource` using the `min` operation.
@@ -63,11 +63,11 @@ public final class ResourceScatterMin extends RawOp {
    * @return a new instance of ResourceScatterMin
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber, U extends Tensor> ResourceScatterMin create(Scope scope, Operand<?> resource, Operand<T> indices, Operand<U> updates) {
+  public static <T extends TNumber, U extends TType> ResourceScatterMin create(Scope scope, Operand<?> resource, Operand<T> indices, Operand<U> updates) {
     OperationBuilder opBuilder = scope.env().opBuilder("ResourceScatterMin", scope.makeOpName("ResourceScatterMin"));
-    opBuilder.addInput(resource.asOutput());
-    opBuilder.addInput(indices.asOutput());
-    opBuilder.addInput(updates.asOutput());
+    opBuilder.addInput(resource.asOutput(scope));
+    opBuilder.addInput(indices.asOutput(scope));
+    opBuilder.addInput(updates.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new ResourceScatterMin(opBuilder.build());
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterMul.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterMul.java
index 7acf531602c..181b4c372b8 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterMul.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterMul.java
@@ -20,12 +20,12 @@
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * Multiplies sparse updates into the variable referenced by `resource`.
@@ -63,11 +63,11 @@ public final class ResourceScatterMul extends RawOp {
    * @return a new instance of ResourceScatterMul
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber, U extends Tensor> ResourceScatterMul create(Scope scope, Operand<?> resource, Operand<T> indices, Operand<U> updates) {
+  public static <T extends TNumber, U extends TType> ResourceScatterMul create(Scope scope, Operand<?> resource, Operand<T> indices, Operand<U> updates) {
     OperationBuilder opBuilder = scope.env().opBuilder("ResourceScatterMul", scope.makeOpName("ResourceScatterMul"));
-    opBuilder.addInput(resource.asOutput());
-    opBuilder.addInput(indices.asOutput());
-    opBuilder.addInput(updates.asOutput());
+    opBuilder.addInput(resource.asOutput(scope));
+    opBuilder.addInput(indices.asOutput(scope));
+    opBuilder.addInput(updates.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new ResourceScatterMul(opBuilder.build());
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterNdAdd.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterNdAdd.java
index a5a82746ec4..83c9ddc52ed 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterNdAdd.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterNdAdd.java
@@ -20,12 +20,12 @@
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * Applies sparse addition to individual values or slices in a Variable.
@@ -97,11 +97,11 @@ private Options() {
    * @return a new instance of ResourceScatterNdAdd
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber, U extends Tensor> ResourceScatterNdAdd create(Scope scope, Operand<?> ref, Operand<T> indices, Operand<U> updates, Options... options) {
+  public static <T extends TNumber, U extends TType> ResourceScatterNdAdd create(Scope scope, Operand<?> ref, Operand<T> indices, Operand<U> updates, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ResourceScatterNdAdd", scope.makeOpName("ResourceScatterNdAdd"));
-    opBuilder.addInput(ref.asOutput());
-    opBuilder.addInput(indices.asOutput());
-    opBuilder.addInput(updates.asOutput());
+    opBuilder.addInput(ref.asOutput(scope));
+    opBuilder.addInput(indices.asOutput(scope));
+    opBuilder.addInput(updates.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterNdSub.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterNdSub.java
index 129f51fd2ae..4bf0dd018ce 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterNdSub.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterNdSub.java
@@ -20,12 +20,12 @@
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * Applies sparse subtraction to individual values or slices in a Variable.
@@ -97,11 +97,11 @@ private Options() {
    * @return a new instance of ResourceScatterNdSub
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber, U extends Tensor> ResourceScatterNdSub create(Scope scope, Operand<?> ref, Operand<T> indices, Operand<U> updates, Options... options) {
+  public static <T extends TNumber, U extends TType> ResourceScatterNdSub create(Scope scope, Operand<?> ref, Operand<T> indices, Operand<U> updates, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ResourceScatterNdSub", scope.makeOpName("ResourceScatterNdSub"));
-    opBuilder.addInput(ref.asOutput());
-    opBuilder.addInput(indices.asOutput());
-    opBuilder.addInput(updates.asOutput());
+    opBuilder.addInput(ref.asOutput(scope));
+    opBuilder.addInput(indices.asOutput(scope));
+    opBuilder.addInput(updates.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterNdUpdate.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterNdUpdate.java
index d32d9c48717..aec3c3dd214 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterNdUpdate.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterNdUpdate.java
@@ -20,12 +20,12 @@
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * Applies sparse `updates` to individual values or slices within a given
@@ -99,11 +99,11 @@ private Options() {
    * @return a new instance of ResourceScatterNdUpdate
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber, U extends Tensor> ResourceScatterNdUpdate create(Scope scope, Operand<?> ref, Operand<T> indices, Operand<U> updates, Options... options) {
+  public static <T extends TNumber, U extends TType> ResourceScatterNdUpdate create(Scope scope, Operand<?> ref, Operand<T> indices, Operand<U> updates, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ResourceScatterNdUpdate", scope.makeOpName("ResourceScatterNdUpdate"));
-    opBuilder.addInput(ref.asOutput());
-    opBuilder.addInput(indices.asOutput());
-    opBuilder.addInput(updates.asOutput());
+    opBuilder.addInput(ref.asOutput(scope));
+    opBuilder.addInput(indices.asOutput(scope));
+    opBuilder.addInput(updates.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterSub.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterSub.java
index 89d66b01f58..306cda55863 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterSub.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterSub.java
@@ -20,12 +20,12 @@
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * Subtracts sparse updates from the variable referenced by `resource`.
@@ -63,11 +63,11 @@ public final class ResourceScatterSub extends RawOp {
    * @return a new instance of ResourceScatterSub
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber, U extends Tensor> ResourceScatterSub create(Scope scope, Operand<?> resource, Operand<T> indices, Operand<U> updates) {
+  public static <T extends TNumber, U extends TType> ResourceScatterSub create(Scope scope, Operand<?> resource, Operand<T> indices, Operand<U> updates) {
     OperationBuilder opBuilder = scope.env().opBuilder("ResourceScatterSub", scope.makeOpName("ResourceScatterSub"));
-    opBuilder.addInput(resource.asOutput());
-    opBuilder.addInput(indices.asOutput());
-    opBuilder.addInput(updates.asOutput());
+    opBuilder.addInput(resource.asOutput(scope));
+    opBuilder.addInput(indices.asOutput(scope));
+    opBuilder.addInput(updates.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new ResourceScatterSub(opBuilder.build());
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterUpdate.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterUpdate.java
index 1f97ce5f21d..ad18af339c1 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterUpdate.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterUpdate.java
@@ -20,12 +20,12 @@
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * Assigns sparse updates to the variable referenced by `resource`.
@@ -54,11 +54,11 @@ public final class ResourceScatterUpdate extends RawOp {
    * @return a new instance of ResourceScatterUpdate
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber, U extends Tensor> ResourceScatterUpdate create(Scope scope, Operand<?> resource, Operand<T> indices, Operand<U> updates) {
+  public static <T extends TNumber, U extends TType> ResourceScatterUpdate create(Scope scope, Operand<?> resource, Operand<T> indices, Operand<U> updates) {
     OperationBuilder opBuilder = scope.env().opBuilder("ResourceScatterUpdate", scope.makeOpName("ResourceScatterUpdate"));
-    opBuilder.addInput(resource.asOutput());
-    opBuilder.addInput(indices.asOutput());
-    opBuilder.addInput(updates.asOutput());
+    opBuilder.addInput(resource.asOutput(scope));
+    opBuilder.addInput(indices.asOutput(scope));
+    opBuilder.addInput(updates.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new ResourceScatterUpdate(opBuilder.build());
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceStridedSliceAssign.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceStridedSliceAssign.java
index e8a999a0e4c..ce676c8c2b6 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceStridedSliceAssign.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceStridedSliceAssign.java
@@ -20,12 +20,12 @@
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * Assign `value` to the sliced l-value reference of `ref`.
@@ -108,13 +108,13 @@ private Options() {
    * @return a new instance of ResourceStridedSliceAssign
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber, U extends Tensor> ResourceStridedSliceAssign create(Scope scope, Operand<?> ref, Operand<T> begin, Operand<T> end, Operand<T> strides, Operand<U> value, Options... options) {
+  public static <T extends TNumber, U extends TType> ResourceStridedSliceAssign create(Scope scope, Operand<?> ref, Operand<T> begin, Operand<T> end, Operand<T> strides, Operand<U> value, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ResourceStridedSliceAssign", scope.makeOpName("ResourceStridedSliceAssign"));
-    opBuilder.addInput(ref.asOutput());
-    opBuilder.addInput(begin.asOutput());
-    opBuilder.addInput(end.asOutput());
-    opBuilder.addInput(strides.asOutput());
-    opBuilder.addInput(value.asOutput());
+    opBuilder.addInput(ref.asOutput(scope));
+    opBuilder.addInput(begin.asOutput(scope));
+    opBuilder.addInput(end.asOutput(scope));
+    opBuilder.addInput(strides.asOutput(scope));
+    opBuilder.addInput(value.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Reverse.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Reverse.java
index 0334a7bce4c..3db73f986f2 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Reverse.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Reverse.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * Reverses specific dimensions of a tensor.
@@ -81,7 +81,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator
-public final class Reverse<T extends Tensor> extends RawOp implements Operand<T> {
+public final class Reverse<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Reverse operation.
@@ -93,10 +93,10 @@ public final class Reverse<T extends Tensor> extends RawOp implements Operand<T>
    * @return a new instance of Reverse
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor, U extends Tensor & TNumber> Reverse<T> create(Scope scope, Operand<T> tensor, Operand<U> axis) {
+  public static <T extends TType, U extends TNumber> Reverse<T> create(Scope scope, Operand<T> tensor, Operand<U> axis) {
     OperationBuilder opBuilder = scope.env().opBuilder("ReverseV2", scope.makeOpName("Reverse"));
-    opBuilder.addInput(tensor.asOutput());
-    opBuilder.addInput(axis.asOutput());
+    opBuilder.addInput(tensor.asOutput(scope));
+    opBuilder.addInput(axis.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new Reverse<T>(opBuilder.build());
   }
@@ -109,7 +109,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ReverseSequence.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ReverseSequence.java
index bc3e1b44bd2..aed27ec03df 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ReverseSequence.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ReverseSequence.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * Reverses variable length slices.
@@ -87,7 +87,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator
-public final class ReverseSequence<T extends Tensor> extends RawOp implements Operand<T> {
+public final class ReverseSequence<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.core.ReverseSequence}
@@ -120,10 +120,10 @@ private Options() {
    * @return a new instance of ReverseSequence
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor, U extends Tensor & TNumber> ReverseSequence<T> create(Scope scope, Operand<T> input, Operand<U> seqLengths, Long seqDim, Options... options) {
+  public static <T extends TType, U extends TNumber> ReverseSequence<T> create(Scope scope, Operand<T> input, Operand<U> seqLengths, Long seqDim, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ReverseSequence", scope.makeOpName("ReverseSequence"));
-    opBuilder.addInput(input.asOutput());
-    opBuilder.addInput(seqLengths.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
+    opBuilder.addInput(seqLengths.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("seq_dim", seqDim);
     if (options != null) {
@@ -151,7 +151,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Roll.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Roll.java
index 3b2cd8a7413..e3baaa91241 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Roll.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Roll.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * Rolls the elements of a tensor along an axis.
@@ -55,7 +55,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator
-public final class Roll<T extends Tensor> extends RawOp implements Operand<T> {
+public final class Roll<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Roll operation.
@@ -73,11 +73,11 @@ public final class Roll<T extends Tensor> extends RawOp implements Operand<T> {
    * @return a new instance of Roll
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor, U extends Tensor & TNumber, V extends Tensor & TNumber> Roll<T> create(Scope scope, Operand<T> input, Operand<U> shift, Operand<V> axis) {
+  public static <T extends TType, U extends TNumber, V extends TNumber> Roll<T> create(Scope scope, Operand<T> input, Operand<U> shift, Operand<V> axis) {
     OperationBuilder opBuilder = scope.env().opBuilder("Roll", scope.makeOpName("Roll"));
-    opBuilder.addInput(input.asOutput());
-    opBuilder.addInput(shift.asOutput());
-    opBuilder.addInput(axis.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
+    opBuilder.addInput(shift.asOutput(scope));
+    opBuilder.addInput(axis.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new Roll<T>(opBuilder.build());
   }
@@ -92,7 +92,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Rpc.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Rpc.java
index ad3195933bc..ae82347d5d4 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Rpc.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Rpc.java
@@ -138,9 +138,9 @@ private Options() {
   @Endpoint(describeByClass = true)
   public static Rpc create(Scope scope, Operand<TString> address, Operand<TString> method, Operand<TString> request, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("Rpc", scope.makeOpName("Rpc"));
-    opBuilder.addInput(address.asOutput());
-    opBuilder.addInput(method.asOutput());
-    opBuilder.addInput(request.asOutput());
+    opBuilder.addInput(address.asOutput(scope));
+    opBuilder.addInput(method.asOutput(scope));
+    opBuilder.addInput(request.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -192,7 +192,7 @@ public Output<TString> response() {
   }
   
   @Override
-  public Output<TString> asOutput() {
+  public Output<TString> asOutput(Scope scope) {
     return response;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterAdd.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterAdd.java
index 88def7cc922..88e107f365c 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterAdd.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterAdd.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * Adds sparse updates to a variable reference.
@@ -57,7 +57,7 @@
  * @param <T> data type for {@code outputRef()} output
  */
 @Operator
-public final class ScatterAdd<T extends Tensor> extends RawOp implements Operand<T> {
+public final class ScatterAdd<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.core.ScatterAdd}
@@ -90,11 +90,11 @@ private Options() {
    * @return a new instance of ScatterAdd
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor, U extends Tensor & TNumber> ScatterAdd<T> create(Scope scope, Operand<T> ref, Operand<U> indices, Operand<T> updates, Options... options) {
+  public static <T extends TType, U extends TNumber> ScatterAdd<T> create(Scope scope, Operand<T> ref, Operand<U> indices, Operand<T> updates, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ScatterAdd", scope.makeOpName("ScatterAdd"));
-    opBuilder.addInput(ref.asOutput());
-    opBuilder.addInput(indices.asOutput());
-    opBuilder.addInput(updates.asOutput());
+    opBuilder.addInput(ref.asOutput(scope));
+    opBuilder.addInput(indices.asOutput(scope));
+    opBuilder.addInput(updates.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -123,7 +123,7 @@ public Output<T> outputRef() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return outputRef;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterDiv.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterDiv.java
index 1f56778e276..c9c1e2e0772 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterDiv.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterDiv.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * Divides a variable reference by sparse updates.
@@ -53,7 +53,7 @@
  * @param <T> data type for {@code outputRef()} output
  */
 @Operator
-public final class ScatterDiv<T extends Tensor> extends RawOp implements Operand<T> {
+public final class ScatterDiv<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.core.ScatterDiv}
@@ -86,11 +86,11 @@ private Options() {
    * @return a new instance of ScatterDiv
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor, U extends Tensor & TNumber> ScatterDiv<T> create(Scope scope, Operand<T> ref, Operand<U> indices, Operand<T> updates, Options... options) {
+  public static <T extends TType, U extends TNumber> ScatterDiv<T> create(Scope scope, Operand<T> ref, Operand<U> indices, Operand<T> updates, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ScatterDiv", scope.makeOpName("ScatterDiv"));
-    opBuilder.addInput(ref.asOutput());
-    opBuilder.addInput(indices.asOutput());
-    opBuilder.addInput(updates.asOutput());
+    opBuilder.addInput(ref.asOutput(scope));
+    opBuilder.addInput(indices.asOutput(scope));
+    opBuilder.addInput(updates.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -119,7 +119,7 @@ public Output<T> outputRef() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return outputRef;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterMax.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterMax.java
index 441317eb24b..75ef0b8455c 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterMax.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterMax.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -57,7 +56,7 @@
  * @param <T> data type for {@code outputRef()} output
  */
 @Operator
-public final class ScatterMax<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class ScatterMax<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.core.ScatterMax}
@@ -90,11 +89,11 @@ private Options() {
    * @return a new instance of ScatterMax
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber, U extends Tensor & TNumber> ScatterMax<T> create(Scope scope, Operand<T> ref, Operand<U> indices, Operand<T> updates, Options... options) {
+  public static <T extends TNumber, U extends TNumber> ScatterMax<T> create(Scope scope, Operand<T> ref, Operand<U> indices, Operand<T> updates, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ScatterMax", scope.makeOpName("ScatterMax"));
-    opBuilder.addInput(ref.asOutput());
-    opBuilder.addInput(indices.asOutput());
-    opBuilder.addInput(updates.asOutput());
+    opBuilder.addInput(ref.asOutput(scope));
+    opBuilder.addInput(indices.asOutput(scope));
+    opBuilder.addInput(updates.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -123,7 +122,7 @@ public Output<T> outputRef() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return outputRef;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterMin.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterMin.java
index 1b2e7cf358d..b14c0f3f84e 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterMin.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterMin.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -57,7 +56,7 @@
  * @param <T> data type for {@code outputRef()} output
  */
 @Operator
-public final class ScatterMin<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class ScatterMin<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.core.ScatterMin}
@@ -90,11 +89,11 @@ private Options() {
    * @return a new instance of ScatterMin
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber, U extends Tensor & TNumber> ScatterMin<T> create(Scope scope, Operand<T> ref, Operand<U> indices, Operand<T> updates, Options... options) {
+  public static <T extends TNumber, U extends TNumber> ScatterMin<T> create(Scope scope, Operand<T> ref, Operand<U> indices, Operand<T> updates, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ScatterMin", scope.makeOpName("ScatterMin"));
-    opBuilder.addInput(ref.asOutput());
-    opBuilder.addInput(indices.asOutput());
-    opBuilder.addInput(updates.asOutput());
+    opBuilder.addInput(ref.asOutput(scope));
+    opBuilder.addInput(indices.asOutput(scope));
+    opBuilder.addInput(updates.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -123,7 +122,7 @@ public Output<T> outputRef() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return outputRef;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterMul.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterMul.java
index 5ad16de15fb..c2497cb69ad 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterMul.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterMul.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * Multiplies sparse updates into a variable reference.
@@ -53,7 +53,7 @@
  * @param <T> data type for {@code outputRef()} output
  */
 @Operator
-public final class ScatterMul<T extends Tensor> extends RawOp implements Operand<T> {
+public final class ScatterMul<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.core.ScatterMul}
@@ -86,11 +86,11 @@ private Options() {
    * @return a new instance of ScatterMul
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor, U extends Tensor & TNumber> ScatterMul<T> create(Scope scope, Operand<T> ref, Operand<U> indices, Operand<T> updates, Options... options) {
+  public static <T extends TType, U extends TNumber> ScatterMul<T> create(Scope scope, Operand<T> ref, Operand<U> indices, Operand<T> updates, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ScatterMul", scope.makeOpName("ScatterMul"));
-    opBuilder.addInput(ref.asOutput());
-    opBuilder.addInput(indices.asOutput());
-    opBuilder.addInput(updates.asOutput());
+    opBuilder.addInput(ref.asOutput(scope));
+    opBuilder.addInput(indices.asOutput(scope));
+    opBuilder.addInput(updates.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -119,7 +119,7 @@ public Output<T> outputRef() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return outputRef;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterNd.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterNd.java
index 24098cc16bb..efba30f95a1 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterNd.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterNd.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * Scatter `updates` into a new tensor according to `indices`.
@@ -111,7 +111,7 @@
  * @param <U> data type for {@code output()} output
  */
 @Operator
-public final class ScatterNd<U extends Tensor> extends RawOp implements Operand<U> {
+public final class ScatterNd<U extends TType> extends RawOp implements Operand<U> {
   
   /**
    * Factory method to create a class wrapping a new ScatterNd operation.
@@ -123,11 +123,11 @@ public final class ScatterNd<U extends Tensor> extends RawOp implements Operand<
    * @return a new instance of ScatterNd
    */
   @Endpoint(describeByClass = true)
-  public static <U extends Tensor, T extends Tensor & TNumber> ScatterNd<U> create(Scope scope, Operand<T> indices, Operand<U> updates, Operand<T> shape) {
+  public static <U extends TType, T extends TNumber> ScatterNd<U> create(Scope scope, Operand<T> indices, Operand<U> updates, Operand<T> shape) {
     OperationBuilder opBuilder = scope.env().opBuilder("ScatterNd", scope.makeOpName("ScatterNd"));
-    opBuilder.addInput(indices.asOutput());
-    opBuilder.addInput(updates.asOutput());
-    opBuilder.addInput(shape.asOutput());
+    opBuilder.addInput(indices.asOutput(scope));
+    opBuilder.addInput(updates.asOutput(scope));
+    opBuilder.addInput(shape.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new ScatterNd<U>(opBuilder.build());
   }
@@ -141,7 +141,7 @@ public Output<U> output() {
   }
   
   @Override
-  public Output<U> asOutput() {
+  public Output<U> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterNdAdd.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterNdAdd.java
index 5d1ede126d6..92d40566bc2 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterNdAdd.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterNdAdd.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * Applies sparse addition to individual values or slices in a Variable.
@@ -64,7 +64,7 @@
  * @param <T> data type for {@code outputRef()} output
  */
 @Operator
-public final class ScatterNdAdd<T extends Tensor> extends RawOp implements Operand<T> {
+public final class ScatterNdAdd<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.core.ScatterNdAdd}
@@ -100,11 +100,11 @@ private Options() {
    * @return a new instance of ScatterNdAdd
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor, U extends Tensor & TNumber> ScatterNdAdd<T> create(Scope scope, Operand<T> ref, Operand<U> indices, Operand<T> updates, Options... options) {
+  public static <T extends TType, U extends TNumber> ScatterNdAdd<T> create(Scope scope, Operand<T> ref, Operand<U> indices, Operand<T> updates, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ScatterNdAdd", scope.makeOpName("ScatterNdAdd"));
-    opBuilder.addInput(ref.asOutput());
-    opBuilder.addInput(indices.asOutput());
-    opBuilder.addInput(updates.asOutput());
+    opBuilder.addInput(ref.asOutput(scope));
+    opBuilder.addInput(indices.asOutput(scope));
+    opBuilder.addInput(updates.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -134,7 +134,7 @@ public Output<T> outputRef() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return outputRef;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterNdNonAliasingAdd.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterNdNonAliasingAdd.java
index c2066854bee..a76b06dd47c 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterNdNonAliasingAdd.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterNdNonAliasingAdd.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * Applies sparse addition to `input` using individual values or slices
@@ -68,7 +68,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator
-public final class ScatterNdNonAliasingAdd<T extends Tensor> extends RawOp implements Operand<T> {
+public final class ScatterNdNonAliasingAdd<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new ScatterNdNonAliasingAdd operation.
@@ -82,11 +82,11 @@ public final class ScatterNdNonAliasingAdd<T extends Tensor> extends RawOp imple
    * @return a new instance of ScatterNdNonAliasingAdd
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor, U extends Tensor & TNumber> ScatterNdNonAliasingAdd<T> create(Scope scope, Operand<T> input, Operand<U> indices, Operand<T> updates) {
+  public static <T extends TType, U extends TNumber> ScatterNdNonAliasingAdd<T> create(Scope scope, Operand<T> input, Operand<U> indices, Operand<T> updates) {
     OperationBuilder opBuilder = scope.env().opBuilder("ScatterNdNonAliasingAdd", scope.makeOpName("ScatterNdNonAliasingAdd"));
-    opBuilder.addInput(input.asOutput());
-    opBuilder.addInput(indices.asOutput());
-    opBuilder.addInput(updates.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
+    opBuilder.addInput(indices.asOutput(scope));
+    opBuilder.addInput(updates.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new ScatterNdNonAliasingAdd<T>(opBuilder.build());
   }
@@ -100,7 +100,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterNdSub.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterNdSub.java
index d6679bc2633..402367ab5d1 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterNdSub.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterNdSub.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * Applies sparse subtraction to individual values or slices in a Variable.
@@ -66,7 +66,7 @@
  * @param <T> data type for {@code outputRef()} output
  */
 @Operator
-public final class ScatterNdSub<T extends Tensor> extends RawOp implements Operand<T> {
+public final class ScatterNdSub<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.core.ScatterNdSub}
@@ -102,11 +102,11 @@ private Options() {
    * @return a new instance of ScatterNdSub
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor, U extends Tensor & TNumber> ScatterNdSub<T> create(Scope scope, Operand<T> ref, Operand<U> indices, Operand<T> updates, Options... options) {
+  public static <T extends TType, U extends TNumber> ScatterNdSub<T> create(Scope scope, Operand<T> ref, Operand<U> indices, Operand<T> updates, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ScatterNdSub", scope.makeOpName("ScatterNdSub"));
-    opBuilder.addInput(ref.asOutput());
-    opBuilder.addInput(indices.asOutput());
-    opBuilder.addInput(updates.asOutput());
+    opBuilder.addInput(ref.asOutput(scope));
+    opBuilder.addInput(indices.asOutput(scope));
+    opBuilder.addInput(updates.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -136,7 +136,7 @@ public Output<T> outputRef() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return outputRef;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterNdUpdate.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterNdUpdate.java
index d9399495fbd..62e7d3461db 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterNdUpdate.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterNdUpdate.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * Applies sparse `updates` to individual values or slices within a given
@@ -68,7 +68,7 @@
  * @param <T> data type for {@code outputRef()} output
  */
 @Operator
-public final class ScatterNdUpdate<T extends Tensor> extends RawOp implements Operand<T> {
+public final class ScatterNdUpdate<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.core.ScatterNdUpdate}
@@ -104,11 +104,11 @@ private Options() {
    * @return a new instance of ScatterNdUpdate
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor, U extends Tensor & TNumber> ScatterNdUpdate<T> create(Scope scope, Operand<T> ref, Operand<U> indices, Operand<T> updates, Options... options) {
+  public static <T extends TType, U extends TNumber> ScatterNdUpdate<T> create(Scope scope, Operand<T> ref, Operand<U> indices, Operand<T> updates, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ScatterNdUpdate", scope.makeOpName("ScatterNdUpdate"));
-    opBuilder.addInput(ref.asOutput());
-    opBuilder.addInput(indices.asOutput());
-    opBuilder.addInput(updates.asOutput());
+    opBuilder.addInput(ref.asOutput(scope));
+    opBuilder.addInput(indices.asOutput(scope));
+    opBuilder.addInput(updates.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -138,7 +138,7 @@ public Output<T> outputRef() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return outputRef;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterSub.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterSub.java
index df43d11f910..ae5a81ffd50 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterSub.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterSub.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * Subtracts sparse updates to a variable reference.
@@ -56,7 +56,7 @@
  * @param <T> data type for {@code outputRef()} output
  */
 @Operator
-public final class ScatterSub<T extends Tensor> extends RawOp implements Operand<T> {
+public final class ScatterSub<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.core.ScatterSub}
@@ -89,11 +89,11 @@ private Options() {
    * @return a new instance of ScatterSub
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor, U extends Tensor & TNumber> ScatterSub<T> create(Scope scope, Operand<T> ref, Operand<U> indices, Operand<T> updates, Options... options) {
+  public static <T extends TType, U extends TNumber> ScatterSub<T> create(Scope scope, Operand<T> ref, Operand<U> indices, Operand<T> updates, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ScatterSub", scope.makeOpName("ScatterSub"));
-    opBuilder.addInput(ref.asOutput());
-    opBuilder.addInput(indices.asOutput());
-    opBuilder.addInput(updates.asOutput());
+    opBuilder.addInput(ref.asOutput(scope));
+    opBuilder.addInput(indices.asOutput(scope));
+    opBuilder.addInput(updates.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -122,7 +122,7 @@ public Output<T> outputRef() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return outputRef;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterUpdate.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterUpdate.java
index e71848cd77e..5768a097438 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterUpdate.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterUpdate.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * Applies sparse updates to a variable reference.
@@ -60,7 +60,7 @@
  * @param <T> data type for {@code outputRef()} output
  */
 @Operator
-public final class ScatterUpdate<T extends Tensor> extends RawOp implements Operand<T> {
+public final class ScatterUpdate<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.core.ScatterUpdate}
@@ -93,11 +93,11 @@ private Options() {
    * @return a new instance of ScatterUpdate
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor, U extends Tensor & TNumber> ScatterUpdate<T> create(Scope scope, Operand<T> ref, Operand<U> indices, Operand<T> updates, Options... options) {
+  public static <T extends TType, U extends TNumber> ScatterUpdate<T> create(Scope scope, Operand<T> ref, Operand<U> indices, Operand<T> updates, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ScatterUpdate", scope.makeOpName("ScatterUpdate"));
-    opBuilder.addInput(ref.asOutput());
-    opBuilder.addInput(indices.asOutput());
-    opBuilder.addInput(updates.asOutput());
+    opBuilder.addInput(ref.asOutput(scope));
+    opBuilder.addInput(indices.asOutput(scope));
+    opBuilder.addInput(updates.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -126,7 +126,7 @@ public Output<T> outputRef() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return outputRef;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Select.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Select.java
index 8adc04b790f..caa85e1ea44 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Select.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Select.java
@@ -21,18 +21,18 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TBool;
+import org.tensorflow.types.family.TType;
 
 /**
  * @param <T> data type for {@code output()} output
  */
 @Operator
-public final class Select<T extends Tensor> extends RawOp implements Operand<T> {
+public final class Select<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Select operation.
@@ -44,11 +44,11 @@ public final class Select<T extends Tensor> extends RawOp implements Operand<T>
    * @return a new instance of Select
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> Select<T> create(Scope scope, Operand<TBool> condition, Operand<T> t, Operand<T> e) {
+  public static <T extends TType> Select<T> create(Scope scope, Operand<TBool> condition, Operand<T> t, Operand<T> e) {
     OperationBuilder opBuilder = scope.env().opBuilder("SelectV2", scope.makeOpName("Select"));
-    opBuilder.addInput(condition.asOutput());
-    opBuilder.addInput(t.asOutput());
-    opBuilder.addInput(e.asOutput());
+    opBuilder.addInput(condition.asOutput(scope));
+    opBuilder.addInput(t.asOutput(scope));
+    opBuilder.addInput(e.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new Select<T>(opBuilder.build());
   }
@@ -60,7 +60,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Send.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Send.java
index 010fc778e9c..c3a5d086ad1 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Send.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Send.java
@@ -20,11 +20,11 @@
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Sends the named tensor from send_device to recv_device.
@@ -66,9 +66,9 @@ private Options() {
    * @return a new instance of Send
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> Send create(Scope scope, Operand<T> tensor, String tensorName, String sendDevice, Long sendDeviceIncarnation, String recvDevice, Options... options) {
+  public static <T extends TType> Send create(Scope scope, Operand<T> tensor, String tensorName, String sendDevice, Long sendDeviceIncarnation, String recvDevice, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("Send", scope.makeOpName("Send"));
-    opBuilder.addInput(tensor.asOutput());
+    opBuilder.addInput(tensor.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("tensor_name", tensorName);
     opBuilder.setAttr("send_device", sendDevice);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/SetDiff1d.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/SetDiff1d.java
index 082db21ea5f..c0130ee43c3 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/SetDiff1d.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/SetDiff1d.java
@@ -22,13 +22,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * Computes the difference between two lists of numbers or strings.
@@ -57,7 +57,7 @@
  * @param <U> data type for {@code idx()} output
  */
 @Operator
-public final class SetDiff1d<T extends Tensor, U extends Tensor & TNumber> extends RawOp {
+public final class SetDiff1d<T extends TType, U extends TNumber> extends RawOp {
   
   /**
    * Factory method to create a class wrapping a new SetDiff1d operation.
@@ -69,10 +69,10 @@ public final class SetDiff1d<T extends Tensor, U extends Tensor & TNumber> exten
    * @return a new instance of SetDiff1d
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor, U extends Tensor & TNumber> SetDiff1d<T, U> create(Scope scope, Operand<T> x, Operand<T> y, DataType<U> outIdx) {
+  public static <T extends TType, U extends TNumber> SetDiff1d<T, U> create(Scope scope, Operand<T> x, Operand<T> y, DataType<U> outIdx) {
     OperationBuilder opBuilder = scope.env().opBuilder("ListDiff", scope.makeOpName("SetDiff1d"));
-    opBuilder.addInput(x.asOutput());
-    opBuilder.addInput(y.asOutput());
+    opBuilder.addInput(x.asOutput(scope));
+    opBuilder.addInput(y.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("out_idx", outIdx);
     return new SetDiff1d<T, U>(opBuilder.build());
@@ -87,7 +87,7 @@ public static <T extends Tensor, U extends Tensor & TNumber> SetDiff1d<T, U> cre
    * @return a new instance of SetDiff1d
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> SetDiff1d<T, TInt32> create(Scope scope, Operand<T> x, Operand<T> y) {
+  public static <T extends TType> SetDiff1d<T, TInt32> create(Scope scope, Operand<T> x, Operand<T> y) {
     return create(scope, x, y, TInt32.DTYPE);
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/SetSize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/SetSize.java
index 26239a49412..e5f73a23881 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/SetSize.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/SetSize.java
@@ -21,13 +21,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
 import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
 
 /**
  * Number of unique elements along last dimension of input `set`.
@@ -72,11 +72,11 @@ private Options() {
    * @return a new instance of SetSize
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> SetSize create(Scope scope, Operand<TInt64> setIndices, Operand<T> setValues, Operand<TInt64> setShape, Options... options) {
+  public static <T extends TType> SetSize create(Scope scope, Operand<TInt64> setIndices, Operand<T> setValues, Operand<TInt64> setShape, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("SetSize", scope.makeOpName("SetSize"));
-    opBuilder.addInput(setIndices.asOutput());
-    opBuilder.addInput(setValues.asOutput());
-    opBuilder.addInput(setShape.asOutput());
+    opBuilder.addInput(setIndices.asOutput(scope));
+    opBuilder.addInput(setValues.asOutput(scope));
+    opBuilder.addInput(setShape.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -105,7 +105,7 @@ public Output<TInt32> size() {
   }
   
   @Override
-  public Output<TInt32> asOutput() {
+  public Output<TInt32> asOutput(Scope scope) {
     return size;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Shape.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Shape.java
index c007fd32549..3867a539be3 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Shape.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Shape.java
@@ -22,13 +22,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * Returns the shape of a tensor.
@@ -45,7 +45,7 @@
  * @param <U> data type for {@code output()} output
  */
 @Operator
-public final class Shape<U extends Tensor & TNumber> extends RawOp implements Operand<U> {
+public final class Shape<U extends TNumber> extends RawOp implements Operand<U> {
   
   /**
    * Factory method to create a class wrapping a new Shape operation.
@@ -56,9 +56,9 @@ public final class Shape<U extends Tensor & TNumber> extends RawOp implements Op
    * @return a new instance of Shape
    */
   @Endpoint(describeByClass = true)
-  public static <U extends Tensor & TNumber, T extends Tensor> Shape<U> create(Scope scope, Operand<T> input, DataType<U> outType) {
+  public static <U extends TNumber, T extends TType> Shape<U> create(Scope scope, Operand<T> input, DataType<U> outType) {
     OperationBuilder opBuilder = scope.env().opBuilder("Shape", scope.makeOpName("Shape"));
-    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("out_type", outType);
     return new Shape<U>(opBuilder.build());
@@ -72,7 +72,7 @@ public static <U extends Tensor & TNumber, T extends Tensor> Shape<U> create(Sco
    * @return a new instance of Shape
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> Shape<TInt32> create(Scope scope, Operand<T> input) {
+  public static <T extends TType> Shape<TInt32> create(Scope scope, Operand<T> input) {
     return create(scope, input, TInt32.DTYPE);
   }
   
@@ -83,7 +83,7 @@ public Output<U> output() {
   }
   
   @Override
-  public Output<U> asOutput() {
+  public Output<U> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ShapeN.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ShapeN.java
index 46fdba4223f..d390088de03 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ShapeN.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ShapeN.java
@@ -25,7 +25,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.Operands;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
@@ -33,6 +32,7 @@
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * Returns shape of tensors.
@@ -42,7 +42,7 @@
  * @param <U> data type for {@code output()} output
  */
 @Operator
-public final class ShapeN<U extends Tensor & TNumber> extends RawOp implements Iterable<Operand<U>> {
+public final class ShapeN<U extends TNumber> extends RawOp implements Iterable<Operand<U>> {
   
   /**
    * Factory method to create a class wrapping a new ShapeN operation.
@@ -53,9 +53,9 @@ public final class ShapeN<U extends Tensor & TNumber> extends RawOp implements I
    * @return a new instance of ShapeN
    */
   @Endpoint(describeByClass = true)
-  public static <U extends Tensor & TNumber, T extends Tensor> ShapeN<U> create(Scope scope, Iterable<Operand<T>> input, DataType<U> outType) {
+  public static <U extends TNumber, T extends TType> ShapeN<U> create(Scope scope, Iterable<Operand<T>> input, DataType<U> outType) {
     OperationBuilder opBuilder = scope.env().opBuilder("ShapeN", scope.makeOpName("ShapeN"));
-    opBuilder.addInputList(Operands.asOutputs(input));
+    opBuilder.addInputList(Operands.asOutputs(scope, input));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("out_type", outType);
     return new ShapeN<U>(opBuilder.build());
@@ -69,7 +69,7 @@ public static <U extends Tensor & TNumber, T extends Tensor> ShapeN<U> create(Sc
    * @return a new instance of ShapeN
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> ShapeN<TInt32> create(Scope scope, Iterable<Operand<T>> input) {
+  public static <T extends TType> ShapeN<TInt32> create(Scope scope, Iterable<Operand<T>> input) {
     return create(scope, input, TInt32.DTYPE);
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Size.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Size.java
index c03e08df8fa..18ccf934513 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Size.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Size.java
@@ -22,13 +22,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * Returns the size of a tensor.
@@ -46,7 +46,7 @@
  * @param <U> data type for {@code output()} output
  */
 @Operator
-public final class Size<U extends Tensor & TNumber> extends RawOp implements Operand<U> {
+public final class Size<U extends TNumber> extends RawOp implements Operand<U> {
   
   /**
    * Factory method to create a class wrapping a new Size operation.
@@ -57,9 +57,9 @@ public final class Size<U extends Tensor & TNumber> extends RawOp implements Ope
    * @return a new instance of Size
    */
   @Endpoint(describeByClass = true)
-  public static <U extends Tensor & TNumber, T extends Tensor> Size<U> create(Scope scope, Operand<T> input, DataType<U> outType) {
+  public static <U extends TNumber, T extends TType> Size<U> create(Scope scope, Operand<T> input, DataType<U> outType) {
     OperationBuilder opBuilder = scope.env().opBuilder("Size", scope.makeOpName("Size"));
-    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("out_type", outType);
     return new Size<U>(opBuilder.build());
@@ -73,7 +73,7 @@ public static <U extends Tensor & TNumber, T extends Tensor> Size<U> create(Scop
    * @return a new instance of Size
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> Size<TInt32> create(Scope scope, Operand<T> input) {
+  public static <T extends TType> Size<TInt32> create(Scope scope, Operand<T> input) {
     return create(scope, input, TInt32.DTYPE);
   }
   
@@ -84,7 +84,7 @@ public Output<U> output() {
   }
   
   @Override
-  public Output<U> asOutput() {
+  public Output<U> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Slice.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Slice.java
index 1d5e2684f74..5b93a65307e 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Slice.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Slice.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * Return a slice from 'input'.
@@ -41,7 +41,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator
-public final class Slice<T extends Tensor> extends RawOp implements Operand<T> {
+public final class Slice<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Slice operation.
@@ -57,11 +57,11 @@ public final class Slice<T extends Tensor> extends RawOp implements Operand<T> {
    * @return a new instance of Slice
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor, U extends Tensor & TNumber> Slice<T> create(Scope scope, Operand<T> input, Operand<U> begin, Operand<U> size) {
+  public static <T extends TType, U extends TNumber> Slice<T> create(Scope scope, Operand<T> input, Operand<U> begin, Operand<U> size) {
     OperationBuilder opBuilder = scope.env().opBuilder("Slice", scope.makeOpName("Slice"));
-    opBuilder.addInput(input.asOutput());
-    opBuilder.addInput(begin.asOutput());
-    opBuilder.addInput(size.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
+    opBuilder.addInput(begin.asOutput(scope));
+    opBuilder.addInput(size.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new Slice<T>(opBuilder.build());
   }
@@ -73,7 +73,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Snapshot.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Snapshot.java
index 02328cbdc2d..9e654bb65da 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Snapshot.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Snapshot.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Returns a copy of the input tensor.
@@ -33,7 +33,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator
-public final class Snapshot<T extends Tensor> extends RawOp implements Operand<T> {
+public final class Snapshot<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Snapshot operation.
@@ -43,9 +43,9 @@ public final class Snapshot<T extends Tensor> extends RawOp implements Operand<T
    * @return a new instance of Snapshot
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> Snapshot<T> create(Scope scope, Operand<T> input) {
+  public static <T extends TType> Snapshot<T> create(Scope scope, Operand<T> input) {
     OperationBuilder opBuilder = scope.env().opBuilder("Snapshot", scope.makeOpName("Snapshot"));
-    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new Snapshot<T>(opBuilder.build());
   }
@@ -57,7 +57,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/SpaceToBatchNd.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/SpaceToBatchNd.java
index e1ebd3064a2..11d64395abf 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/SpaceToBatchNd.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/SpaceToBatchNd.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * SpaceToBatch for N-D tensors of type T.
@@ -43,7 +43,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator
-public final class SpaceToBatchNd<T extends Tensor> extends RawOp implements Operand<T> {
+public final class SpaceToBatchNd<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new SpaceToBatchNd operation.
@@ -147,11 +147,11 @@ public final class SpaceToBatchNd<T extends Tensor> extends RawOp implements Ope
    * @return a new instance of SpaceToBatchNd
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor, U extends Tensor & TNumber, V extends Tensor & TNumber> SpaceToBatchNd<T> create(Scope scope, Operand<T> input, Operand<U> blockShape, Operand<V> paddings) {
+  public static <T extends TType, U extends TNumber, V extends TNumber> SpaceToBatchNd<T> create(Scope scope, Operand<T> input, Operand<U> blockShape, Operand<V> paddings) {
     OperationBuilder opBuilder = scope.env().opBuilder("SpaceToBatchND", scope.makeOpName("SpaceToBatchNd"));
-    opBuilder.addInput(input.asOutput());
-    opBuilder.addInput(blockShape.asOutput());
-    opBuilder.addInput(paddings.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
+    opBuilder.addInput(blockShape.asOutput(scope));
+    opBuilder.addInput(paddings.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new SpaceToBatchNd<T>(opBuilder.build());
   }
@@ -163,7 +163,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Split.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Split.java
index 97deffaa27d..55fbae5d9cd 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Split.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Split.java
@@ -24,12 +24,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TType;
 
 /**
  * Splits a tensor into `num_split` tensors along one dimension.
@@ -37,7 +37,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator
-public final class Split<T extends Tensor> extends RawOp implements Iterable<Operand<T>> {
+public final class Split<T extends TType> extends RawOp implements Iterable<Operand<T>> {
   
   /**
    * Factory method to create a class wrapping a new Split operation.
@@ -51,10 +51,10 @@ public final class Split<T extends Tensor> extends RawOp implements Iterable<Ope
    * @return a new instance of Split
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> Split<T> create(Scope scope, Operand<TInt32> axis, Operand<T> value, Long numSplit) {
+  public static <T extends TType> Split<T> create(Scope scope, Operand<TInt32> axis, Operand<T> value, Long numSplit) {
     OperationBuilder opBuilder = scope.env().opBuilder("Split", scope.makeOpName("Split"));
-    opBuilder.addInput(axis.asOutput());
-    opBuilder.addInput(value.asOutput());
+    opBuilder.addInput(axis.asOutput(scope));
+    opBuilder.addInput(value.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("num_split", numSplit);
     return new Split<T>(opBuilder.build());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/SplitV.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/SplitV.java
index 87538a25177..7972f7f523f 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/SplitV.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/SplitV.java
@@ -24,13 +24,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * Splits a tensor into `num_split` tensors along one dimension.
@@ -38,7 +38,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator
-public final class SplitV<T extends Tensor> extends RawOp implements Iterable<Operand<T>> {
+public final class SplitV<T extends TType> extends RawOp implements Iterable<Operand<T>> {
   
   /**
    * Factory method to create a class wrapping a new SplitV operation.
@@ -54,11 +54,11 @@ public final class SplitV<T extends Tensor> extends RawOp implements Iterable<Op
    * @return a new instance of SplitV
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor, U extends Tensor & TNumber> SplitV<T> create(Scope scope, Operand<T> value, Operand<U> sizeSplits, Operand<TInt32> axis, Long numSplit) {
+  public static <T extends TType, U extends TNumber> SplitV<T> create(Scope scope, Operand<T> value, Operand<U> sizeSplits, Operand<TInt32> axis, Long numSplit) {
     OperationBuilder opBuilder = scope.env().opBuilder("SplitV", scope.makeOpName("SplitV"));
-    opBuilder.addInput(value.asOutput());
-    opBuilder.addInput(sizeSplits.asOutput());
-    opBuilder.addInput(axis.asOutput());
+    opBuilder.addInput(value.asOutput(scope));
+    opBuilder.addInput(sizeSplits.asOutput(scope));
+    opBuilder.addInput(axis.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("num_split", numSplit);
     return new SplitV<T>(opBuilder.build());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Squeeze.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Squeeze.java
index a229bc7ab23..4eb9838a298 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Squeeze.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Squeeze.java
@@ -22,11 +22,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Removes dimensions of size 1 from the shape of a tensor.
@@ -51,7 +51,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator
-public final class Squeeze<T extends Tensor> extends RawOp implements Operand<T> {
+public final class Squeeze<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.core.Squeeze}
@@ -83,9 +83,9 @@ private Options() {
    * @return a new instance of Squeeze
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> Squeeze<T> create(Scope scope, Operand<T> input, Options... options) {
+  public static <T extends TType> Squeeze<T> create(Scope scope, Operand<T> input, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("Squeeze", scope.makeOpName("Squeeze"));
-    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -119,7 +119,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Stack.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Stack.java
index 7e22893da34..0b68d56ff7e 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Stack.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Stack.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.Operands;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Packs a list of `N` rank-`R` tensors into one rank-`(R+1)` tensor.
@@ -52,7 +52,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator
-public final class Stack<T extends Tensor> extends RawOp implements Operand<T> {
+public final class Stack<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.core.Stack}
@@ -83,9 +83,9 @@ private Options() {
    * @return a new instance of Stack
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> Stack<T> create(Scope scope, Iterable<Operand<T>> values, Options... options) {
+  public static <T extends TType> Stack<T> create(Scope scope, Iterable<Operand<T>> values, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("Pack", scope.makeOpName("Stack"));
-    opBuilder.addInputList(Operands.asOutputs(values));
+    opBuilder.addInputList(Operands.asOutputs(scope, values));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -113,7 +113,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Stage.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Stage.java
index 526462b02f4..ee809f3e79f 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Stage.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Stage.java
@@ -96,7 +96,7 @@ private Options() {
   @Endpoint(describeByClass = true)
   public static Stage create(Scope scope, Iterable<Operand<?>> values, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("Stage", scope.makeOpName("Stage"));
-    opBuilder.addInputList(Operands.asOutputs(values));
+    opBuilder.addInputList(Operands.asOutputs(scope, values));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/StagePeek.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/StagePeek.java
index dab5018a122..9256b9c9ef0 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/StagePeek.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/StagePeek.java
@@ -25,12 +25,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TType;
 
 /**
  * Op peeks at the values at the specified index.  If the
@@ -40,7 +40,7 @@
  * performance.
  */
 @Operator
-public final class StagePeek extends RawOp implements Iterable<Operand<Tensor>> {
+public final class StagePeek extends RawOp implements Iterable<Operand<TType>> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.core.StagePeek}
@@ -100,7 +100,7 @@ private Options() {
   @Endpoint(describeByClass = true)
   public static StagePeek create(Scope scope, Operand<TInt32> index, List<DataType<?>> dtypes, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("StagePeek", scope.makeOpName("StagePeek"));
-    opBuilder.addInput(index.asOutput());
+    opBuilder.addInput(index.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     DataType[] dtypesArray = new DataType[dtypes.size()];
     for (int i = 0; i < dtypesArray.length; ++i) {
@@ -162,7 +162,7 @@ public List<Output<?>> values() {
   
   @Override
   @SuppressWarnings({"rawtypes", "unchecked"})
-  public Iterator<Operand<Tensor>> iterator() {
+  public Iterator<Operand<TType>> iterator() {
     return (Iterator) values.iterator();
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/StageSize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/StageSize.java
index 94ef566e708..fb489970ed1 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/StageSize.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/StageSize.java
@@ -152,7 +152,7 @@ public Output<TInt32> size() {
   }
   
   @Override
-  public Output<TInt32> asOutput() {
+  public Output<TInt32> asOutput(Scope scope) {
     return size;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/StopGradient.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/StopGradient.java
index a5852cf0a97..f57784f8f1f 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/StopGradient.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/StopGradient.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Stops gradient computation.
@@ -58,7 +58,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator
-public final class StopGradient<T extends Tensor> extends RawOp implements Operand<T> {
+public final class StopGradient<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new StopGradient operation.
@@ -68,9 +68,9 @@ public final class StopGradient<T extends Tensor> extends RawOp implements Opera
    * @return a new instance of StopGradient
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> StopGradient<T> create(Scope scope, Operand<T> input) {
+  public static <T extends TType> StopGradient<T> create(Scope scope, Operand<T> input) {
     OperationBuilder opBuilder = scope.env().opBuilder("StopGradient", scope.makeOpName("StopGradient"));
-    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new StopGradient<T>(opBuilder.build());
   }
@@ -82,7 +82,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/StridedSlice.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/StridedSlice.java
index c2f38d29e4f..5fb0ee67946 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/StridedSlice.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/StridedSlice.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * Return a strided slice from `input`.
@@ -121,7 +121,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator
-public final class StridedSlice<T extends Tensor> extends RawOp implements Operand<T> {
+public final class StridedSlice<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.core.StridedSlice}
@@ -214,12 +214,12 @@ private Options() {
    * @return a new instance of StridedSlice
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor, U extends Tensor & TNumber> StridedSlice<T> create(Scope scope, Operand<T> input, Operand<U> begin, Operand<U> end, Operand<U> strides, Options... options) {
+  public static <T extends TType, U extends TNumber> StridedSlice<T> create(Scope scope, Operand<T> input, Operand<U> begin, Operand<U> end, Operand<U> strides, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("StridedSlice", scope.makeOpName("StridedSlice"));
-    opBuilder.addInput(input.asOutput());
-    opBuilder.addInput(begin.asOutput());
-    opBuilder.addInput(end.asOutput());
-    opBuilder.addInput(strides.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
+    opBuilder.addInput(begin.asOutput(scope));
+    opBuilder.addInput(end.asOutput(scope));
+    opBuilder.addInput(strides.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -300,7 +300,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/StridedSliceAssign.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/StridedSliceAssign.java
index e7dec58902d..0cb552ea895 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/StridedSliceAssign.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/StridedSliceAssign.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * Assign `value` to the sliced l-value reference of `ref`.
@@ -41,7 +41,7 @@
  * @param <T> data type for {@code outputRef()} output
  */
 @Operator
-public final class StridedSliceAssign<T extends Tensor> extends RawOp implements Operand<T> {
+public final class StridedSliceAssign<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.core.StridedSliceAssign}
@@ -111,13 +111,13 @@ private Options() {
    * @return a new instance of StridedSliceAssign
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor, U extends Tensor & TNumber> StridedSliceAssign<T> create(Scope scope, Operand<T> ref, Operand<U> begin, Operand<U> end, Operand<U> strides, Operand<T> value, Options... options) {
+  public static <T extends TType, U extends TNumber> StridedSliceAssign<T> create(Scope scope, Operand<T> ref, Operand<U> begin, Operand<U> end, Operand<U> strides, Operand<T> value, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("StridedSliceAssign", scope.makeOpName("StridedSliceAssign"));
-    opBuilder.addInput(ref.asOutput());
-    opBuilder.addInput(begin.asOutput());
-    opBuilder.addInput(end.asOutput());
-    opBuilder.addInput(strides.asOutput());
-    opBuilder.addInput(value.asOutput());
+    opBuilder.addInput(ref.asOutput(scope));
+    opBuilder.addInput(begin.asOutput(scope));
+    opBuilder.addInput(end.asOutput(scope));
+    opBuilder.addInput(strides.asOutput(scope));
+    opBuilder.addInput(value.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -183,7 +183,7 @@ public Output<T> outputRef() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return outputRef;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/StridedSliceGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/StridedSliceGrad.java
index 9aa3f5b348b..bbf0e022ca6 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/StridedSliceGrad.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/StridedSliceGrad.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * Returns the gradient of `StridedSlice`.
@@ -43,7 +43,7 @@
  * @param <U> data type for {@code output()} output
  */
 @Operator
-public final class StridedSliceGrad<U extends Tensor> extends RawOp implements Operand<U> {
+public final class StridedSliceGrad<U extends TType> extends RawOp implements Operand<U> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.core.StridedSliceGrad}
@@ -113,13 +113,13 @@ private Options() {
    * @return a new instance of StridedSliceGrad
    */
   @Endpoint(describeByClass = true)
-  public static <U extends Tensor, T extends Tensor & TNumber> StridedSliceGrad<U> create(Scope scope, Operand<T> shape, Operand<T> begin, Operand<T> end, Operand<T> strides, Operand<U> dy, Options... options) {
+  public static <U extends TType, T extends TNumber> StridedSliceGrad<U> create(Scope scope, Operand<T> shape, Operand<T> begin, Operand<T> end, Operand<T> strides, Operand<U> dy, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("StridedSliceGrad", scope.makeOpName("StridedSliceGrad"));
-    opBuilder.addInput(shape.asOutput());
-    opBuilder.addInput(begin.asOutput());
-    opBuilder.addInput(end.asOutput());
-    opBuilder.addInput(strides.asOutput());
-    opBuilder.addInput(dy.asOutput());
+    opBuilder.addInput(shape.asOutput(scope));
+    opBuilder.addInput(begin.asOutput(scope));
+    opBuilder.addInput(end.asOutput(scope));
+    opBuilder.addInput(strides.asOutput(scope));
+    opBuilder.addInput(dy.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -185,7 +185,7 @@ public Output<U> output() {
   }
   
   @Override
-  public Output<U> asOutput() {
+  public Output<U> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Sum.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Sum.java
index af88da5baa2..ae5eca98bc7 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Sum.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Sum.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * Computes the sum of elements across dimensions of a tensor.
@@ -39,7 +39,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator
-public final class Sum<T extends Tensor> extends RawOp implements Operand<T> {
+public final class Sum<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.core.Sum}
@@ -71,10 +71,10 @@ private Options() {
    * @return a new instance of Sum
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor, U extends Tensor & TNumber> Sum<T> create(Scope scope, Operand<T> input, Operand<U> axis, Options... options) {
+  public static <T extends TType, U extends TNumber> Sum<T> create(Scope scope, Operand<T> input, Operand<U> axis, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("Sum", scope.makeOpName("Sum"));
-    opBuilder.addInput(input.asOutput());
-    opBuilder.addInput(axis.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
+    opBuilder.addInput(axis.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -101,7 +101,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/SwitchCond.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/SwitchCond.java
index 2b2b67d7f9c..49bd0008e47 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/SwitchCond.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/SwitchCond.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TBool;
+import org.tensorflow.types.family.TType;
 
 /**
  * Forwards `data` to the output port determined by `pred`.
@@ -39,7 +39,7 @@
  * @param <T> data type for {@code outputFalse()} output
  */
 @Operator
-public final class SwitchCond<T extends Tensor> extends RawOp {
+public final class SwitchCond<T extends TType> extends RawOp {
   
   /**
    * Factory method to create a class wrapping a new SwitchCond operation.
@@ -50,10 +50,10 @@ public final class SwitchCond<T extends Tensor> extends RawOp {
    * @return a new instance of SwitchCond
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> SwitchCond<T> create(Scope scope, Operand<T> data, Operand<TBool> pred) {
+  public static <T extends TType> SwitchCond<T> create(Scope scope, Operand<T> data, Operand<TBool> pred) {
     OperationBuilder opBuilder = scope.env().opBuilder("Switch", scope.makeOpName("SwitchCond"));
-    opBuilder.addInput(data.asOutput());
-    opBuilder.addInput(pred.asOutput());
+    opBuilder.addInput(data.asOutput(scope));
+    opBuilder.addInput(pred.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new SwitchCond<T>(opBuilder.build());
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TemporaryVariable.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TemporaryVariable.java
index 18c5fc112e3..85213671649 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TemporaryVariable.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TemporaryVariable.java
@@ -22,12 +22,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Returns a tensor that may be mutated, but only persists within a single step.
@@ -50,7 +50,7 @@
  * @param <T> data type for {@code ref()} output
  */
 @Operator
-public final class TemporaryVariable<T extends Tensor> extends RawOp implements Operand<T> {
+public final class TemporaryVariable<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.core.TemporaryVariable}
@@ -82,7 +82,7 @@ private Options() {
    * @return a new instance of TemporaryVariable
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> TemporaryVariable<T> create(Scope scope, Shape shape, DataType<T> dtype, Options... options) {
+  public static <T extends TType> TemporaryVariable<T> create(Scope scope, Shape shape, DataType<T> dtype, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("TemporaryVariable", scope.makeOpName("TemporaryVariable"));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("shape", shape);
@@ -113,7 +113,7 @@ public Output<T> ref() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return ref;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArray.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArray.java
index e484ce440cb..41fea3a831a 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArray.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArray.java
@@ -22,7 +22,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
@@ -30,6 +29,7 @@
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TFloat32;
 import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TType;
 
 /**
  * An array of Tensors of given size.
@@ -116,9 +116,9 @@ private Options() {
    * @return a new instance of TensorArray
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> TensorArray create(Scope scope, Operand<TInt32> size, DataType<T> dtype, Options... options) {
+  public static <T extends TType> TensorArray create(Scope scope, Operand<TInt32> size, DataType<T> dtype, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("TensorArrayV3", scope.makeOpName("TensorArray"));
-    opBuilder.addInput(size.asOutput());
+    opBuilder.addInput(size.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("dtype", dtype);
     if (options != null) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayClose.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayClose.java
index 62180e8e5ff..047dda8e506 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayClose.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayClose.java
@@ -44,7 +44,7 @@ public final class TensorArrayClose extends RawOp {
   @Endpoint(describeByClass = true)
   public static TensorArrayClose create(Scope scope, Operand<?> handle) {
     OperationBuilder opBuilder = scope.env().opBuilder("TensorArrayCloseV3", scope.makeOpName("TensorArrayClose"));
-    opBuilder.addInput(handle.asOutput());
+    opBuilder.addInput(handle.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new TensorArrayClose(opBuilder.build());
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayConcat.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayConcat.java
index 9fa20bacc2f..76b519c6ff5 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayConcat.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayConcat.java
@@ -22,7 +22,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
@@ -30,6 +29,7 @@
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TFloat32;
 import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
 
 /**
  * Concat the elements from the TensorArray into value `value`.
@@ -48,7 +48,7 @@
  * @param <T> data type for {@code value()} output
  */
 @Operator
-public final class TensorArrayConcat<T extends Tensor> extends RawOp {
+public final class TensorArrayConcat<T extends TType> extends RawOp {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.core.TensorArrayConcat}
@@ -83,10 +83,10 @@ private Options() {
    * @return a new instance of TensorArrayConcat
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> TensorArrayConcat<T> create(Scope scope, Operand<?> handle, Operand<TFloat32> flowIn, DataType<T> dtype, Options... options) {
+  public static <T extends TType> TensorArrayConcat<T> create(Scope scope, Operand<?> handle, Operand<TFloat32> flowIn, DataType<T> dtype, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("TensorArrayConcatV3", scope.makeOpName("TensorArrayConcat"));
-    opBuilder.addInput(handle.asOutput());
-    opBuilder.addInput(flowIn.asOutput());
+    opBuilder.addInput(handle.asOutput(scope));
+    opBuilder.addInput(flowIn.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("dtype", dtype);
     if (options != null) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayGather.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayGather.java
index 5f5475866a9..4595687db01 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayGather.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayGather.java
@@ -22,7 +22,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
@@ -30,6 +29,7 @@
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TFloat32;
 import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TType;
 
 /**
  * Gather specific elements from the TensorArray into output `value`.
@@ -39,7 +39,7 @@
  * @param <T> data type for {@code value()} output
  */
 @Operator
-public final class TensorArrayGather<T extends Tensor> extends RawOp implements Operand<T> {
+public final class TensorArrayGather<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.core.TensorArrayGather}
@@ -74,11 +74,11 @@ private Options() {
    * @return a new instance of TensorArrayGather
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> TensorArrayGather<T> create(Scope scope, Operand<?> handle, Operand<TInt32> indices, Operand<TFloat32> flowIn, DataType<T> dtype, Options... options) {
+  public static <T extends TType> TensorArrayGather<T> create(Scope scope, Operand<?> handle, Operand<TInt32> indices, Operand<TFloat32> flowIn, DataType<T> dtype, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("TensorArrayGatherV3", scope.makeOpName("TensorArrayGather"));
-    opBuilder.addInput(handle.asOutput());
-    opBuilder.addInput(indices.asOutput());
-    opBuilder.addInput(flowIn.asOutput());
+    opBuilder.addInput(handle.asOutput(scope));
+    opBuilder.addInput(indices.asOutput(scope));
+    opBuilder.addInput(flowIn.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("dtype", dtype);
     if (options != null) {
@@ -109,7 +109,7 @@ public Output<T> value() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return value;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayGrad.java
index dca362c6934..961a09cb278 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayGrad.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayGrad.java
@@ -83,8 +83,8 @@ public final class TensorArrayGrad extends RawOp {
   @Endpoint(describeByClass = true)
   public static TensorArrayGrad create(Scope scope, Operand<?> handle, Operand<TFloat32> flowIn, String source) {
     OperationBuilder opBuilder = scope.env().opBuilder("TensorArrayGradV3", scope.makeOpName("TensorArrayGrad"));
-    opBuilder.addInput(handle.asOutput());
-    opBuilder.addInput(flowIn.asOutput());
+    opBuilder.addInput(handle.asOutput(scope));
+    opBuilder.addInput(flowIn.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("source", source);
     return new TensorArrayGrad(opBuilder.build());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayGradWithShape.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayGradWithShape.java
index 6128775dce6..6420095d6c5 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayGradWithShape.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayGradWithShape.java
@@ -55,9 +55,9 @@ public final class TensorArrayGradWithShape extends RawOp {
   @Endpoint(describeByClass = true)
   public static TensorArrayGradWithShape create(Scope scope, Operand<?> handle, Operand<TFloat32> flowIn, Operand<TInt32> shapeToPrepend, String source) {
     OperationBuilder opBuilder = scope.env().opBuilder("TensorArrayGradWithShape", scope.makeOpName("TensorArrayGradWithShape"));
-    opBuilder.addInput(handle.asOutput());
-    opBuilder.addInput(flowIn.asOutput());
-    opBuilder.addInput(shapeToPrepend.asOutput());
+    opBuilder.addInput(handle.asOutput(scope));
+    opBuilder.addInput(flowIn.asOutput(scope));
+    opBuilder.addInput(shapeToPrepend.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("source", source);
     return new TensorArrayGradWithShape(opBuilder.build());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayPack.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayPack.java
index aaccdfdc913..e2ad54ec5a3 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayPack.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayPack.java
@@ -22,7 +22,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
@@ -30,12 +29,13 @@
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TFloat32;
 import org.tensorflow.types.TString;
+import org.tensorflow.types.family.TType;
 
 /**
  * @param <T> data type for {@code value()} output
  */
 @Operator
-public final class TensorArrayPack<T extends Tensor> extends RawOp implements Operand<T> {
+public final class TensorArrayPack<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.core.TensorArrayPack}
@@ -67,10 +67,10 @@ private Options() {
    * @return a new instance of TensorArrayPack
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> TensorArrayPack<T> create(Scope scope, Operand<TString> handle, Operand<TFloat32> flowIn, DataType<T> dtype, Options... options) {
+  public static <T extends TType> TensorArrayPack<T> create(Scope scope, Operand<TString> handle, Operand<TFloat32> flowIn, DataType<T> dtype, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("TensorArrayPack", scope.makeOpName("TensorArrayPack"));
-    opBuilder.addInput(handle.asOutput());
-    opBuilder.addInput(flowIn.asOutput());
+    opBuilder.addInput(handle.asOutput(scope));
+    opBuilder.addInput(flowIn.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("dtype", dtype);
     if (options != null) {
@@ -97,7 +97,7 @@ public Output<T> value() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return value;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayRead.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayRead.java
index 44216cbab93..2634a6c0cf2 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayRead.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayRead.java
@@ -22,13 +22,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TFloat32;
 import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TType;
 
 /**
  * Read an element from the TensorArray into output `value`.
@@ -36,7 +36,7 @@
  * @param <T> data type for {@code value()} output
  */
 @Operator
-public final class TensorArrayRead<T extends Tensor> extends RawOp implements Operand<T> {
+public final class TensorArrayRead<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new TensorArrayRead operation.
@@ -49,11 +49,11 @@ public final class TensorArrayRead<T extends Tensor> extends RawOp implements Op
    * @return a new instance of TensorArrayRead
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> TensorArrayRead<T> create(Scope scope, Operand<?> handle, Operand<TInt32> index, Operand<TFloat32> flowIn, DataType<T> dtype) {
+  public static <T extends TType> TensorArrayRead<T> create(Scope scope, Operand<?> handle, Operand<TInt32> index, Operand<TFloat32> flowIn, DataType<T> dtype) {
     OperationBuilder opBuilder = scope.env().opBuilder("TensorArrayReadV3", scope.makeOpName("TensorArrayRead"));
-    opBuilder.addInput(handle.asOutput());
-    opBuilder.addInput(index.asOutput());
-    opBuilder.addInput(flowIn.asOutput());
+    opBuilder.addInput(handle.asOutput(scope));
+    opBuilder.addInput(index.asOutput(scope));
+    opBuilder.addInput(flowIn.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("dtype", dtype);
     return new TensorArrayRead<T>(opBuilder.build());
@@ -67,7 +67,7 @@ public Output<T> value() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return value;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayScatter.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayScatter.java
index b0dac8b4452..6053ce43fbe 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayScatter.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayScatter.java
@@ -21,13 +21,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TFloat32;
 import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TType;
 
 /**
  * Scatter the data from the input value into specific TensorArray elements.
@@ -48,12 +48,12 @@ public final class TensorArrayScatter extends RawOp implements Operand<TFloat32>
    * @return a new instance of TensorArrayScatter
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> TensorArrayScatter create(Scope scope, Operand<?> handle, Operand<TInt32> indices, Operand<T> value, Operand<TFloat32> flowIn) {
+  public static <T extends TType> TensorArrayScatter create(Scope scope, Operand<?> handle, Operand<TInt32> indices, Operand<T> value, Operand<TFloat32> flowIn) {
     OperationBuilder opBuilder = scope.env().opBuilder("TensorArrayScatterV3", scope.makeOpName("TensorArrayScatter"));
-    opBuilder.addInput(handle.asOutput());
-    opBuilder.addInput(indices.asOutput());
-    opBuilder.addInput(value.asOutput());
-    opBuilder.addInput(flowIn.asOutput());
+    opBuilder.addInput(handle.asOutput(scope));
+    opBuilder.addInput(indices.asOutput(scope));
+    opBuilder.addInput(value.asOutput(scope));
+    opBuilder.addInput(flowIn.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new TensorArrayScatter(opBuilder.build());
   }
@@ -66,7 +66,7 @@ public Output<TFloat32> flowOut() {
   }
   
   @Override
-  public Output<TFloat32> asOutput() {
+  public Output<TFloat32> asOutput(Scope scope) {
     return flowOut;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArraySize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArraySize.java
index 3e7987bc388..41883931ca6 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArraySize.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArraySize.java
@@ -45,8 +45,8 @@ public final class TensorArraySize extends RawOp implements Operand<TInt32> {
   @Endpoint(describeByClass = true)
   public static TensorArraySize create(Scope scope, Operand<?> handle, Operand<TFloat32> flowIn) {
     OperationBuilder opBuilder = scope.env().opBuilder("TensorArraySizeV3", scope.makeOpName("TensorArraySize"));
-    opBuilder.addInput(handle.asOutput());
-    opBuilder.addInput(flowIn.asOutput());
+    opBuilder.addInput(handle.asOutput(scope));
+    opBuilder.addInput(flowIn.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new TensorArraySize(opBuilder.build());
   }
@@ -59,7 +59,7 @@ public Output<TInt32> size() {
   }
   
   @Override
-  public Output<TInt32> asOutput() {
+  public Output<TInt32> asOutput(Scope scope) {
     return size;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArraySplit.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArraySplit.java
index 8b11e653a18..4b9bc963fac 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArraySplit.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArraySplit.java
@@ -21,13 +21,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TFloat32;
 import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
 
 /**
  * Split the data from the input value into TensorArray elements.
@@ -69,12 +69,12 @@ public final class TensorArraySplit extends RawOp implements Operand<TFloat32> {
    * @return a new instance of TensorArraySplit
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> TensorArraySplit create(Scope scope, Operand<?> handle, Operand<T> value, Operand<TInt64> lengths, Operand<TFloat32> flowIn) {
+  public static <T extends TType> TensorArraySplit create(Scope scope, Operand<?> handle, Operand<T> value, Operand<TInt64> lengths, Operand<TFloat32> flowIn) {
     OperationBuilder opBuilder = scope.env().opBuilder("TensorArraySplitV3", scope.makeOpName("TensorArraySplit"));
-    opBuilder.addInput(handle.asOutput());
-    opBuilder.addInput(value.asOutput());
-    opBuilder.addInput(lengths.asOutput());
-    opBuilder.addInput(flowIn.asOutput());
+    opBuilder.addInput(handle.asOutput(scope));
+    opBuilder.addInput(value.asOutput(scope));
+    opBuilder.addInput(lengths.asOutput(scope));
+    opBuilder.addInput(flowIn.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new TensorArraySplit(opBuilder.build());
   }
@@ -87,7 +87,7 @@ public Output<TFloat32> flowOut() {
   }
   
   @Override
-  public Output<TFloat32> asOutput() {
+  public Output<TFloat32> asOutput(Scope scope) {
     return flowOut;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayUnpack.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayUnpack.java
index 06509356f54..f179119693c 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayUnpack.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayUnpack.java
@@ -21,13 +21,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TFloat32;
 import org.tensorflow.types.TString;
+import org.tensorflow.types.family.TType;
 
 /**
  */
@@ -44,11 +44,11 @@ public final class TensorArrayUnpack extends RawOp implements Operand<TFloat32>
    * @return a new instance of TensorArrayUnpack
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> TensorArrayUnpack create(Scope scope, Operand<TString> handle, Operand<T> value, Operand<TFloat32> flowIn) {
+  public static <T extends TType> TensorArrayUnpack create(Scope scope, Operand<TString> handle, Operand<T> value, Operand<TFloat32> flowIn) {
     OperationBuilder opBuilder = scope.env().opBuilder("TensorArrayUnpack", scope.makeOpName("TensorArrayUnpack"));
-    opBuilder.addInput(handle.asOutput());
-    opBuilder.addInput(value.asOutput());
-    opBuilder.addInput(flowIn.asOutput());
+    opBuilder.addInput(handle.asOutput(scope));
+    opBuilder.addInput(value.asOutput(scope));
+    opBuilder.addInput(flowIn.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new TensorArrayUnpack(opBuilder.build());
   }
@@ -60,7 +60,7 @@ public Output<TFloat32> flowOut() {
   }
   
   @Override
-  public Output<TFloat32> asOutput() {
+  public Output<TFloat32> asOutput(Scope scope) {
     return flowOut;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayWrite.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayWrite.java
index cacb13634e0..1a37e6474a4 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayWrite.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayWrite.java
@@ -21,13 +21,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TFloat32;
 import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TType;
 
 /**
  * Push an element onto the tensor_array.
@@ -46,12 +46,12 @@ public final class TensorArrayWrite extends RawOp implements Operand<TFloat32> {
    * @return a new instance of TensorArrayWrite
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> TensorArrayWrite create(Scope scope, Operand<?> handle, Operand<TInt32> index, Operand<T> value, Operand<TFloat32> flowIn) {
+  public static <T extends TType> TensorArrayWrite create(Scope scope, Operand<?> handle, Operand<TInt32> index, Operand<T> value, Operand<TFloat32> flowIn) {
     OperationBuilder opBuilder = scope.env().opBuilder("TensorArrayWriteV3", scope.makeOpName("TensorArrayWrite"));
-    opBuilder.addInput(handle.asOutput());
-    opBuilder.addInput(index.asOutput());
-    opBuilder.addInput(value.asOutput());
-    opBuilder.addInput(flowIn.asOutput());
+    opBuilder.addInput(handle.asOutput(scope));
+    opBuilder.addInput(index.asOutput(scope));
+    opBuilder.addInput(value.asOutput(scope));
+    opBuilder.addInput(flowIn.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new TensorArrayWrite(opBuilder.build());
   }
@@ -64,7 +64,7 @@ public Output<TFloat32> flowOut() {
   }
   
   @Override
-  public Output<TFloat32> asOutput() {
+  public Output<TFloat32> asOutput(Scope scope) {
     return flowOut;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorForestCreateTreeVariable.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorForestCreateTreeVariable.java
index 5ca6ffa1cd0..3daad106a9e 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorForestCreateTreeVariable.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorForestCreateTreeVariable.java
@@ -42,8 +42,8 @@ public final class TensorForestCreateTreeVariable extends RawOp {
   @Endpoint(describeByClass = true)
   public static TensorForestCreateTreeVariable create(Scope scope, Operand<?> treeHandle, Operand<TString> treeConfig) {
     OperationBuilder opBuilder = scope.env().opBuilder("TensorForestCreateTreeVariable", scope.makeOpName("TensorForestCreateTreeVariable"));
-    opBuilder.addInput(treeHandle.asOutput());
-    opBuilder.addInput(treeConfig.asOutput());
+    opBuilder.addInput(treeHandle.asOutput(scope));
+    opBuilder.addInput(treeConfig.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new TensorForestCreateTreeVariable(opBuilder.build());
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorForestTreeDeserialize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorForestTreeDeserialize.java
index a5e1638035e..cdbbc60af61 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorForestTreeDeserialize.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorForestTreeDeserialize.java
@@ -42,8 +42,8 @@ public final class TensorForestTreeDeserialize extends RawOp {
   @Endpoint(describeByClass = true)
   public static TensorForestTreeDeserialize create(Scope scope, Operand<?> treeHandle, Operand<TString> treeConfig) {
     OperationBuilder opBuilder = scope.env().opBuilder("TensorForestTreeDeserialize", scope.makeOpName("TensorForestTreeDeserialize"));
-    opBuilder.addInput(treeHandle.asOutput());
-    opBuilder.addInput(treeConfig.asOutput());
+    opBuilder.addInput(treeHandle.asOutput(scope));
+    opBuilder.addInput(treeConfig.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new TensorForestTreeDeserialize(opBuilder.build());
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorForestTreeIsInitializedOp.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorForestTreeIsInitializedOp.java
index 4dfe64cdf19..f15882776a6 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorForestTreeIsInitializedOp.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorForestTreeIsInitializedOp.java
@@ -42,7 +42,7 @@ public final class TensorForestTreeIsInitializedOp extends RawOp implements Oper
   @Endpoint(describeByClass = true)
   public static TensorForestTreeIsInitializedOp create(Scope scope, Operand<?> treeHandle) {
     OperationBuilder opBuilder = scope.env().opBuilder("TensorForestTreeIsInitializedOp", scope.makeOpName("TensorForestTreeIsInitializedOp"));
-    opBuilder.addInput(treeHandle.asOutput());
+    opBuilder.addInput(treeHandle.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new TensorForestTreeIsInitializedOp(opBuilder.build());
   }
@@ -55,7 +55,7 @@ public Output<TBool> isInitialized() {
   }
   
   @Override
-  public Output<TBool> asOutput() {
+  public Output<TBool> asOutput(Scope scope) {
     return isInitialized;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorForestTreePredict.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorForestTreePredict.java
index 3962212c5b7..4b8268d0258 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorForestTreePredict.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorForestTreePredict.java
@@ -44,8 +44,8 @@ public final class TensorForestTreePredict extends RawOp implements Operand<TFlo
   @Endpoint(describeByClass = true)
   public static TensorForestTreePredict create(Scope scope, Operand<?> treeHandle, Operand<TFloat32> denseFeatures, Long logitsDimension) {
     OperationBuilder opBuilder = scope.env().opBuilder("TensorForestTreePredict", scope.makeOpName("TensorForestTreePredict"));
-    opBuilder.addInput(treeHandle.asOutput());
-    opBuilder.addInput(denseFeatures.asOutput());
+    opBuilder.addInput(treeHandle.asOutput(scope));
+    opBuilder.addInput(denseFeatures.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("logits_dimension", logitsDimension);
     return new TensorForestTreePredict(opBuilder.build());
@@ -59,7 +59,7 @@ public Output<TFloat32> logits() {
   }
   
   @Override
-  public Output<TFloat32> asOutput() {
+  public Output<TFloat32> asOutput(Scope scope) {
     return logits;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorForestTreeResourceHandleOp.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorForestTreeResourceHandleOp.java
index f739dcb573b..d2a7541b05c 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorForestTreeResourceHandleOp.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorForestTreeResourceHandleOp.java
@@ -21,16 +21,16 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Creates a handle to a TensorForestTreeResource
  */
-public final class TensorForestTreeResourceHandleOp extends RawOp implements Operand<Tensor> {
+public final class TensorForestTreeResourceHandleOp extends RawOp implements Operand<TType> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.core.TensorForestTreeResourceHandleOp}
@@ -106,8 +106,8 @@ public Output<?> resource() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<Tensor> asOutput() {
-    return (Output<Tensor>) resource;
+  public Output<TType> asOutput(Scope scope) {
+    return (Output<TType>) resource;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorForestTreeSerialize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorForestTreeSerialize.java
index d46e23a08ab..fbb182ff56d 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorForestTreeSerialize.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorForestTreeSerialize.java
@@ -42,7 +42,7 @@ public final class TensorForestTreeSerialize extends RawOp implements Operand<TS
   @Endpoint(describeByClass = true)
   public static TensorForestTreeSerialize create(Scope scope, Operand<?> treeHandle) {
     OperationBuilder opBuilder = scope.env().opBuilder("TensorForestTreeSerialize", scope.makeOpName("TensorForestTreeSerialize"));
-    opBuilder.addInput(treeHandle.asOutput());
+    opBuilder.addInput(treeHandle.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new TensorForestTreeSerialize(opBuilder.build());
   }
@@ -55,7 +55,7 @@ public Output<TString> treeConfig() {
   }
   
   @Override
-  public Output<TString> asOutput() {
+  public Output<TString> asOutput(Scope scope) {
     return treeConfig;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorForestTreeSize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorForestTreeSize.java
index 17cf008d470..ce4abb39160 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorForestTreeSize.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorForestTreeSize.java
@@ -42,7 +42,7 @@ public final class TensorForestTreeSize extends RawOp implements Operand<TInt32>
   @Endpoint(describeByClass = true)
   public static TensorForestTreeSize create(Scope scope, Operand<?> treeHandle) {
     OperationBuilder opBuilder = scope.env().opBuilder("TensorForestTreeSize", scope.makeOpName("TensorForestTreeSize"));
-    opBuilder.addInput(treeHandle.asOutput());
+    opBuilder.addInput(treeHandle.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new TensorForestTreeSize(opBuilder.build());
   }
@@ -55,7 +55,7 @@ public Output<TInt32> treeSize() {
   }
   
   @Override
-  public Output<TInt32> asOutput() {
+  public Output<TInt32> asOutput(Scope scope) {
     return treeSize;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListConcat.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListConcat.java
index 398b39a0746..da39e92abbc 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListConcat.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListConcat.java
@@ -22,13 +22,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * Concats all tensors in the list along the 0th dimension.
@@ -49,7 +49,7 @@
  * @param <U> data type for {@code tensor()} output
  */
 @Operator
-public final class TensorListConcat<U extends Tensor> extends RawOp {
+public final class TensorListConcat<U extends TType> extends RawOp {
   
   /**
    * Factory method to create a class wrapping a new TensorListConcat operation.
@@ -62,11 +62,11 @@ public final class TensorListConcat<U extends Tensor> extends RawOp {
    * @return a new instance of TensorListConcat
    */
   @Endpoint(describeByClass = true)
-  public static <U extends Tensor, T extends Tensor & TNumber> TensorListConcat<U> create(Scope scope, Operand<?> inputHandle, Operand<T> elementShape, Operand<TInt64> leadingDims, DataType<U> elementDtype) {
+  public static <U extends TType, T extends TNumber> TensorListConcat<U> create(Scope scope, Operand<?> inputHandle, Operand<T> elementShape, Operand<TInt64> leadingDims, DataType<U> elementDtype) {
     OperationBuilder opBuilder = scope.env().opBuilder("TensorListConcatV2", scope.makeOpName("TensorListConcat"));
-    opBuilder.addInput(inputHandle.asOutput());
-    opBuilder.addInput(elementShape.asOutput());
-    opBuilder.addInput(leadingDims.asOutput());
+    opBuilder.addInput(inputHandle.asOutput(scope));
+    opBuilder.addInput(elementShape.asOutput(scope));
+    opBuilder.addInput(leadingDims.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("element_dtype", elementDtype);
     return new TensorListConcat<U>(opBuilder.build());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListConcatLists.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListConcatLists.java
index f6347befb41..ae959e28808 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListConcatLists.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListConcatLists.java
@@ -22,16 +22,16 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  */
 @Operator
-public final class TensorListConcatLists extends RawOp implements Operand<Tensor> {
+public final class TensorListConcatLists extends RawOp implements Operand<TType> {
   
   /**
    * Factory method to create a class wrapping a new TensorListConcatLists operation.
@@ -43,10 +43,10 @@ public final class TensorListConcatLists extends RawOp implements Operand<Tensor
    * @return a new instance of TensorListConcatLists
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> TensorListConcatLists create(Scope scope, Operand<?> inputA, Operand<?> inputB, DataType<T> elementDtype) {
+  public static <T extends TType> TensorListConcatLists create(Scope scope, Operand<?> inputA, Operand<?> inputB, DataType<T> elementDtype) {
     OperationBuilder opBuilder = scope.env().opBuilder("TensorListConcatLists", scope.makeOpName("TensorListConcatLists"));
-    opBuilder.addInput(inputA.asOutput());
-    opBuilder.addInput(inputB.asOutput());
+    opBuilder.addInput(inputA.asOutput(scope));
+    opBuilder.addInput(inputB.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("element_dtype", elementDtype);
     return new TensorListConcatLists(opBuilder.build());
@@ -60,8 +60,8 @@ public Output<?> output() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<Tensor> asOutput() {
-    return (Output<Tensor>) output;
+  public Output<TType> asOutput(Scope scope) {
+    return (Output<TType>) output;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListElementShape.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListElementShape.java
index 1efb2ee1ac9..f9b95898d6d 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListElementShape.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListElementShape.java
@@ -22,7 +22,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -38,7 +37,7 @@
  * @param <T> data type for {@code elementShape()} output
  */
 @Operator
-public final class TensorListElementShape<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class TensorListElementShape<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new TensorListElementShape operation.
@@ -49,9 +48,9 @@ public final class TensorListElementShape<T extends Tensor & TNumber> extends Ra
    * @return a new instance of TensorListElementShape
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> TensorListElementShape<T> create(Scope scope, Operand<?> inputHandle, DataType<T> shapeType) {
+  public static <T extends TNumber> TensorListElementShape<T> create(Scope scope, Operand<?> inputHandle, DataType<T> shapeType) {
     OperationBuilder opBuilder = scope.env().opBuilder("TensorListElementShape", scope.makeOpName("TensorListElementShape"));
-    opBuilder.addInput(inputHandle.asOutput());
+    opBuilder.addInput(inputHandle.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("shape_type", shapeType);
     return new TensorListElementShape<T>(opBuilder.build());
@@ -64,7 +63,7 @@ public Output<T> elementShape() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return elementShape;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListFromTensor.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListFromTensor.java
index 28829431aa8..d95c1718653 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListFromTensor.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListFromTensor.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * Creates a TensorList which, when stacked, has the value of `tensor`.
@@ -37,7 +37,7 @@
  * output_handle: The list.
  */
 @Operator
-public final class TensorListFromTensor extends RawOp implements Operand<Tensor> {
+public final class TensorListFromTensor extends RawOp implements Operand<TType> {
   
   /**
    * Factory method to create a class wrapping a new TensorListFromTensor operation.
@@ -48,10 +48,10 @@ public final class TensorListFromTensor extends RawOp implements Operand<Tensor>
    * @return a new instance of TensorListFromTensor
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor, U extends Tensor & TNumber> TensorListFromTensor create(Scope scope, Operand<T> tensor, Operand<U> elementShape) {
+  public static <T extends TType, U extends TNumber> TensorListFromTensor create(Scope scope, Operand<T> tensor, Operand<U> elementShape) {
     OperationBuilder opBuilder = scope.env().opBuilder("TensorListFromTensor", scope.makeOpName("TensorListFromTensor"));
-    opBuilder.addInput(tensor.asOutput());
-    opBuilder.addInput(elementShape.asOutput());
+    opBuilder.addInput(tensor.asOutput(scope));
+    opBuilder.addInput(elementShape.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new TensorListFromTensor(opBuilder.build());
   }
@@ -64,8 +64,8 @@ public Output<?> outputHandle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<Tensor> asOutput() {
-    return (Output<Tensor>) outputHandle;
+  public Output<TType> asOutput(Scope scope) {
+    return (Output<TType>) outputHandle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListGather.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListGather.java
index 856600ec5b8..224fe935e5b 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListGather.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListGather.java
@@ -22,12 +22,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TType;
 
 /**
  * Creates a Tensor by indexing into the TensorList.
@@ -42,7 +42,7 @@
  * @param <T> data type for {@code values()} output
  */
 @Operator
-public final class TensorListGather<T extends Tensor> extends RawOp implements Operand<T> {
+public final class TensorListGather<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new TensorListGather operation.
@@ -55,11 +55,11 @@ public final class TensorListGather<T extends Tensor> extends RawOp implements O
    * @return a new instance of TensorListGather
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> TensorListGather<T> create(Scope scope, Operand<?> inputHandle, Operand<TInt32> indices, Operand<TInt32> elementShape, DataType<T> elementDtype) {
+  public static <T extends TType> TensorListGather<T> create(Scope scope, Operand<?> inputHandle, Operand<TInt32> indices, Operand<TInt32> elementShape, DataType<T> elementDtype) {
     OperationBuilder opBuilder = scope.env().opBuilder("TensorListGather", scope.makeOpName("TensorListGather"));
-    opBuilder.addInput(inputHandle.asOutput());
-    opBuilder.addInput(indices.asOutput());
-    opBuilder.addInput(elementShape.asOutput());
+    opBuilder.addInput(inputHandle.asOutput(scope));
+    opBuilder.addInput(indices.asOutput(scope));
+    opBuilder.addInput(elementShape.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("element_dtype", elementDtype);
     return new TensorListGather<T>(opBuilder.build());
@@ -72,7 +72,7 @@ public Output<T> values() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return values;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListGetItem.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListGetItem.java
index 07e2c30e56b..7f435a21a27 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListGetItem.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListGetItem.java
@@ -22,18 +22,18 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TType;
 
 /**
  * @param <T> data type for {@code item()} output
  */
 @Operator
-public final class TensorListGetItem<T extends Tensor> extends RawOp implements Operand<T> {
+public final class TensorListGetItem<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new TensorListGetItem operation.
@@ -46,11 +46,11 @@ public final class TensorListGetItem<T extends Tensor> extends RawOp implements
    * @return a new instance of TensorListGetItem
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> TensorListGetItem<T> create(Scope scope, Operand<?> inputHandle, Operand<TInt32> index, Operand<TInt32> elementShape, DataType<T> elementDtype) {
+  public static <T extends TType> TensorListGetItem<T> create(Scope scope, Operand<?> inputHandle, Operand<TInt32> index, Operand<TInt32> elementShape, DataType<T> elementDtype) {
     OperationBuilder opBuilder = scope.env().opBuilder("TensorListGetItem", scope.makeOpName("TensorListGetItem"));
-    opBuilder.addInput(inputHandle.asOutput());
-    opBuilder.addInput(index.asOutput());
-    opBuilder.addInput(elementShape.asOutput());
+    opBuilder.addInput(inputHandle.asOutput(scope));
+    opBuilder.addInput(index.asOutput(scope));
+    opBuilder.addInput(elementShape.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("element_dtype", elementDtype);
     return new TensorListGetItem<T>(opBuilder.build());
@@ -63,7 +63,7 @@ public Output<T> item() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return item;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListLength.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListLength.java
index d2dc93a6e06..2b4e4055630 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListLength.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListLength.java
@@ -46,7 +46,7 @@ public final class TensorListLength extends RawOp implements Operand<TInt32> {
   @Endpoint(describeByClass = true)
   public static TensorListLength create(Scope scope, Operand<?> inputHandle) {
     OperationBuilder opBuilder = scope.env().opBuilder("TensorListLength", scope.makeOpName("TensorListLength"));
-    opBuilder.addInput(inputHandle.asOutput());
+    opBuilder.addInput(inputHandle.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new TensorListLength(opBuilder.build());
   }
@@ -58,7 +58,7 @@ public Output<TInt32> length() {
   }
   
   @Override
-  public Output<TInt32> asOutput() {
+  public Output<TInt32> asOutput(Scope scope) {
     return length;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListPopBack.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListPopBack.java
index 428de964aeb..4489f7ae733 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListPopBack.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListPopBack.java
@@ -22,12 +22,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TType;
 
 /**
  * Returns the last element of the input list as well as a list with all but that element.
@@ -42,7 +42,7 @@
  * @param <T> data type for {@code tensor()} output
  */
 @Operator
-public final class TensorListPopBack<T extends Tensor> extends RawOp {
+public final class TensorListPopBack<T extends TType> extends RawOp {
   
   /**
    * Factory method to create a class wrapping a new TensorListPopBack operation.
@@ -54,10 +54,10 @@ public final class TensorListPopBack<T extends Tensor> extends RawOp {
    * @return a new instance of TensorListPopBack
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> TensorListPopBack<T> create(Scope scope, Operand<?> inputHandle, Operand<TInt32> elementShape, DataType<T> elementDtype) {
+  public static <T extends TType> TensorListPopBack<T> create(Scope scope, Operand<?> inputHandle, Operand<TInt32> elementShape, DataType<T> elementDtype) {
     OperationBuilder opBuilder = scope.env().opBuilder("TensorListPopBack", scope.makeOpName("TensorListPopBack"));
-    opBuilder.addInput(inputHandle.asOutput());
-    opBuilder.addInput(elementShape.asOutput());
+    opBuilder.addInput(inputHandle.asOutput(scope));
+    opBuilder.addInput(elementShape.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("element_dtype", elementDtype);
     return new TensorListPopBack<T>(opBuilder.build());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListPushBack.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListPushBack.java
index 99049e35893..ad8bac3d00a 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListPushBack.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListPushBack.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Returns a list which has the passed-in `Tensor` as last element and the other elements of the given list in `input_handle`.
@@ -37,7 +37,7 @@
  * element_shape: a shape compatible with that of elements in the list.
  */
 @Operator
-public final class TensorListPushBack extends RawOp implements Operand<Tensor> {
+public final class TensorListPushBack extends RawOp implements Operand<TType> {
   
   /**
    * Factory method to create a class wrapping a new TensorListPushBack operation.
@@ -48,10 +48,10 @@ public final class TensorListPushBack extends RawOp implements Operand<Tensor> {
    * @return a new instance of TensorListPushBack
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> TensorListPushBack create(Scope scope, Operand<?> inputHandle, Operand<T> tensor) {
+  public static <T extends TType> TensorListPushBack create(Scope scope, Operand<?> inputHandle, Operand<T> tensor) {
     OperationBuilder opBuilder = scope.env().opBuilder("TensorListPushBack", scope.makeOpName("TensorListPushBack"));
-    opBuilder.addInput(inputHandle.asOutput());
-    opBuilder.addInput(tensor.asOutput());
+    opBuilder.addInput(inputHandle.asOutput(scope));
+    opBuilder.addInput(tensor.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new TensorListPushBack(opBuilder.build());
   }
@@ -64,8 +64,8 @@ public Output<?> outputHandle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<Tensor> asOutput() {
-    return (Output<Tensor>) outputHandle;
+  public Output<TType> asOutput(Scope scope) {
+    return (Output<TType>) outputHandle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListPushBackBatch.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListPushBackBatch.java
index 991c3372083..74f3ca489fd 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListPushBackBatch.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListPushBackBatch.java
@@ -21,16 +21,16 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  */
 @Operator
-public final class TensorListPushBackBatch extends RawOp implements Operand<Tensor> {
+public final class TensorListPushBackBatch extends RawOp implements Operand<TType> {
   
   /**
    * Factory method to create a class wrapping a new TensorListPushBackBatch operation.
@@ -41,10 +41,10 @@ public final class TensorListPushBackBatch extends RawOp implements Operand<Tens
    * @return a new instance of TensorListPushBackBatch
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> TensorListPushBackBatch create(Scope scope, Operand<?> inputHandles, Operand<T> tensor) {
+  public static <T extends TType> TensorListPushBackBatch create(Scope scope, Operand<?> inputHandles, Operand<T> tensor) {
     OperationBuilder opBuilder = scope.env().opBuilder("TensorListPushBackBatch", scope.makeOpName("TensorListPushBackBatch"));
-    opBuilder.addInput(inputHandles.asOutput());
-    opBuilder.addInput(tensor.asOutput());
+    opBuilder.addInput(inputHandles.asOutput(scope));
+    opBuilder.addInput(tensor.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new TensorListPushBackBatch(opBuilder.build());
   }
@@ -57,8 +57,8 @@ public Output<?> outputHandles() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<Tensor> asOutput() {
-    return (Output<Tensor>) outputHandles;
+  public Output<TType> asOutput(Scope scope) {
+    return (Output<TType>) outputHandles;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListReserve.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListReserve.java
index 4b92a52ec2e..c5a2e63bff5 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListReserve.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListReserve.java
@@ -22,13 +22,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * List of the given size with empty elements.
@@ -39,7 +39,7 @@
  * element_dtype: the desired type of elements in the list.
  */
 @Operator
-public final class TensorListReserve extends RawOp implements Operand<Tensor> {
+public final class TensorListReserve extends RawOp implements Operand<TType> {
   
   /**
    * Factory method to create a class wrapping a new TensorListReserve operation.
@@ -51,10 +51,10 @@ public final class TensorListReserve extends RawOp implements Operand<Tensor> {
    * @return a new instance of TensorListReserve
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber, U extends Tensor> TensorListReserve create(Scope scope, Operand<T> elementShape, Operand<TInt32> numElements, DataType<U> elementDtype) {
+  public static <T extends TNumber, U extends TType> TensorListReserve create(Scope scope, Operand<T> elementShape, Operand<TInt32> numElements, DataType<U> elementDtype) {
     OperationBuilder opBuilder = scope.env().opBuilder("TensorListReserve", scope.makeOpName("TensorListReserve"));
-    opBuilder.addInput(elementShape.asOutput());
-    opBuilder.addInput(numElements.asOutput());
+    opBuilder.addInput(elementShape.asOutput(scope));
+    opBuilder.addInput(numElements.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("element_dtype", elementDtype);
     return new TensorListReserve(opBuilder.build());
@@ -68,8 +68,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<Tensor> asOutput() {
-    return (Output<Tensor>) handle;
+  public Output<TType> asOutput(Scope scope) {
+    return (Output<TType>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListResize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListResize.java
index a4cada56849..9bd592af106 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListResize.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListResize.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TType;
 
 /**
  * Resizes the list.
@@ -37,7 +37,7 @@
  * 
  */
 @Operator
-public final class TensorListResize extends RawOp implements Operand<Tensor> {
+public final class TensorListResize extends RawOp implements Operand<TType> {
   
   /**
    * Factory method to create a class wrapping a new TensorListResize operation.
@@ -50,8 +50,8 @@ public final class TensorListResize extends RawOp implements Operand<Tensor> {
   @Endpoint(describeByClass = true)
   public static TensorListResize create(Scope scope, Operand<?> inputHandle, Operand<TInt32> size) {
     OperationBuilder opBuilder = scope.env().opBuilder("TensorListResize", scope.makeOpName("TensorListResize"));
-    opBuilder.addInput(inputHandle.asOutput());
-    opBuilder.addInput(size.asOutput());
+    opBuilder.addInput(inputHandle.asOutput(scope));
+    opBuilder.addInput(size.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new TensorListResize(opBuilder.build());
   }
@@ -64,8 +64,8 @@ public Output<?> outputHandle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<Tensor> asOutput() {
-    return (Output<Tensor>) outputHandle;
+  public Output<TType> asOutput(Scope scope) {
+    return (Output<TType>) outputHandle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListScatter.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListScatter.java
index c85c306a4a1..173ccd4c7f9 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListScatter.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListScatter.java
@@ -21,13 +21,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * Creates a TensorList by indexing into a Tensor.
@@ -45,7 +45,7 @@
  * output_handle: The TensorList.
  */
 @Operator
-public final class TensorListScatter extends RawOp implements Operand<Tensor> {
+public final class TensorListScatter extends RawOp implements Operand<TType> {
   
   /**
    * Factory method to create a class wrapping a new TensorListScatter operation.
@@ -58,12 +58,12 @@ public final class TensorListScatter extends RawOp implements Operand<Tensor> {
    * @return a new instance of TensorListScatter
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor, U extends Tensor & TNumber> TensorListScatter create(Scope scope, Operand<T> tensor, Operand<TInt32> indices, Operand<U> elementShape, Operand<TInt32> numElements) {
+  public static <T extends TType, U extends TNumber> TensorListScatter create(Scope scope, Operand<T> tensor, Operand<TInt32> indices, Operand<U> elementShape, Operand<TInt32> numElements) {
     OperationBuilder opBuilder = scope.env().opBuilder("TensorListScatterV2", scope.makeOpName("TensorListScatter"));
-    opBuilder.addInput(tensor.asOutput());
-    opBuilder.addInput(indices.asOutput());
-    opBuilder.addInput(elementShape.asOutput());
-    opBuilder.addInput(numElements.asOutput());
+    opBuilder.addInput(tensor.asOutput(scope));
+    opBuilder.addInput(indices.asOutput(scope));
+    opBuilder.addInput(elementShape.asOutput(scope));
+    opBuilder.addInput(numElements.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new TensorListScatter(opBuilder.build());
   }
@@ -76,8 +76,8 @@ public Output<?> outputHandle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<Tensor> asOutput() {
-    return (Output<Tensor>) outputHandle;
+  public Output<TType> asOutput(Scope scope) {
+    return (Output<TType>) outputHandle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListScatterIntoExistingList.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListScatterIntoExistingList.java
index d85976c616b..7350939b6d7 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListScatterIntoExistingList.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListScatterIntoExistingList.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TType;
 
 /**
  * Scatters tensor at indices in an input list.
@@ -40,7 +40,7 @@
  * output_handle: The TensorList.
  */
 @Operator
-public final class TensorListScatterIntoExistingList extends RawOp implements Operand<Tensor> {
+public final class TensorListScatterIntoExistingList extends RawOp implements Operand<TType> {
   
   /**
    * Factory method to create a class wrapping a new TensorListScatterIntoExistingList operation.
@@ -52,11 +52,11 @@ public final class TensorListScatterIntoExistingList extends RawOp implements Op
    * @return a new instance of TensorListScatterIntoExistingList
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> TensorListScatterIntoExistingList create(Scope scope, Operand<?> inputHandle, Operand<T> tensor, Operand<TInt32> indices) {
+  public static <T extends TType> TensorListScatterIntoExistingList create(Scope scope, Operand<?> inputHandle, Operand<T> tensor, Operand<TInt32> indices) {
     OperationBuilder opBuilder = scope.env().opBuilder("TensorListScatterIntoExistingList", scope.makeOpName("TensorListScatterIntoExistingList"));
-    opBuilder.addInput(inputHandle.asOutput());
-    opBuilder.addInput(tensor.asOutput());
-    opBuilder.addInput(indices.asOutput());
+    opBuilder.addInput(inputHandle.asOutput(scope));
+    opBuilder.addInput(tensor.asOutput(scope));
+    opBuilder.addInput(indices.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new TensorListScatterIntoExistingList(opBuilder.build());
   }
@@ -69,8 +69,8 @@ public Output<?> outputHandle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<Tensor> asOutput() {
-    return (Output<Tensor>) outputHandle;
+  public Output<TType> asOutput(Scope scope) {
+    return (Output<TType>) outputHandle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListSetItem.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListSetItem.java
index 801ad711a56..8a3ceb52d01 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListSetItem.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListSetItem.java
@@ -21,17 +21,17 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TType;
 
 /**
  */
 @Operator
-public final class TensorListSetItem extends RawOp implements Operand<Tensor> {
+public final class TensorListSetItem extends RawOp implements Operand<TType> {
   
   /**
    * Factory method to create a class wrapping a new TensorListSetItem operation.
@@ -43,11 +43,11 @@ public final class TensorListSetItem extends RawOp implements Operand<Tensor> {
    * @return a new instance of TensorListSetItem
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> TensorListSetItem create(Scope scope, Operand<?> inputHandle, Operand<TInt32> index, Operand<T> item) {
+  public static <T extends TType> TensorListSetItem create(Scope scope, Operand<?> inputHandle, Operand<TInt32> index, Operand<T> item) {
     OperationBuilder opBuilder = scope.env().opBuilder("TensorListSetItem", scope.makeOpName("TensorListSetItem"));
-    opBuilder.addInput(inputHandle.asOutput());
-    opBuilder.addInput(index.asOutput());
-    opBuilder.addInput(item.asOutput());
+    opBuilder.addInput(inputHandle.asOutput(scope));
+    opBuilder.addInput(index.asOutput(scope));
+    opBuilder.addInput(item.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new TensorListSetItem(opBuilder.build());
   }
@@ -60,8 +60,8 @@ public Output<?> outputHandle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<Tensor> asOutput() {
-    return (Output<Tensor>) outputHandle;
+  public Output<TType> asOutput(Scope scope) {
+    return (Output<TType>) outputHandle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListSplit.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListSplit.java
index f689050481c..209eb066259 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListSplit.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListSplit.java
@@ -21,13 +21,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * Splits a tensor into a list.
@@ -41,7 +41,7 @@
  * output_handle: The list.
  */
 @Operator
-public final class TensorListSplit extends RawOp implements Operand<Tensor> {
+public final class TensorListSplit extends RawOp implements Operand<TType> {
   
   /**
    * Factory method to create a class wrapping a new TensorListSplit operation.
@@ -53,11 +53,11 @@ public final class TensorListSplit extends RawOp implements Operand<Tensor> {
    * @return a new instance of TensorListSplit
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor, U extends Tensor & TNumber> TensorListSplit create(Scope scope, Operand<T> tensor, Operand<U> elementShape, Operand<TInt64> lengths) {
+  public static <T extends TType, U extends TNumber> TensorListSplit create(Scope scope, Operand<T> tensor, Operand<U> elementShape, Operand<TInt64> lengths) {
     OperationBuilder opBuilder = scope.env().opBuilder("TensorListSplit", scope.makeOpName("TensorListSplit"));
-    opBuilder.addInput(tensor.asOutput());
-    opBuilder.addInput(elementShape.asOutput());
-    opBuilder.addInput(lengths.asOutput());
+    opBuilder.addInput(tensor.asOutput(scope));
+    opBuilder.addInput(elementShape.asOutput(scope));
+    opBuilder.addInput(lengths.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new TensorListSplit(opBuilder.build());
   }
@@ -70,8 +70,8 @@ public Output<?> outputHandle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<Tensor> asOutput() {
-    return (Output<Tensor>) outputHandle;
+  public Output<TType> asOutput(Scope scope) {
+    return (Output<TType>) outputHandle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListStack.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListStack.java
index aeeed49c5bb..7919721b524 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListStack.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListStack.java
@@ -22,12 +22,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TType;
 
 /**
  * Stacks all tensors in the list.
@@ -42,7 +42,7 @@
  * @param <T> data type for {@code tensor()} output
  */
 @Operator
-public final class TensorListStack<T extends Tensor> extends RawOp implements Operand<T> {
+public final class TensorListStack<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.core.TensorListStack}
@@ -74,10 +74,10 @@ private Options() {
    * @return a new instance of TensorListStack
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> TensorListStack<T> create(Scope scope, Operand<?> inputHandle, Operand<TInt32> elementShape, DataType<T> elementDtype, Options... options) {
+  public static <T extends TType> TensorListStack<T> create(Scope scope, Operand<?> inputHandle, Operand<TInt32> elementShape, DataType<T> elementDtype, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("TensorListStack", scope.makeOpName("TensorListStack"));
-    opBuilder.addInput(inputHandle.asOutput());
-    opBuilder.addInput(elementShape.asOutput());
+    opBuilder.addInput(inputHandle.asOutput(scope));
+    opBuilder.addInput(elementShape.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("element_dtype", elementDtype);
     if (options != null) {
@@ -104,7 +104,7 @@ public Output<T> tensor() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return tensor;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorScatterNdAdd.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorScatterNdAdd.java
index 9983210cb93..85647215fa2 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorScatterNdAdd.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorScatterNdAdd.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * Adds sparse `updates` to an existing tensor according to `indices`.
@@ -94,7 +94,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator
-public final class TensorScatterNdAdd<T extends Tensor> extends RawOp implements Operand<T> {
+public final class TensorScatterNdAdd<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new TensorScatterNdAdd operation.
@@ -106,11 +106,11 @@ public final class TensorScatterNdAdd<T extends Tensor> extends RawOp implements
    * @return a new instance of TensorScatterNdAdd
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor, U extends Tensor & TNumber> TensorScatterNdAdd<T> create(Scope scope, Operand<T> tensor, Operand<U> indices, Operand<T> updates) {
+  public static <T extends TType, U extends TNumber> TensorScatterNdAdd<T> create(Scope scope, Operand<T> tensor, Operand<U> indices, Operand<T> updates) {
     OperationBuilder opBuilder = scope.env().opBuilder("TensorScatterAdd", scope.makeOpName("TensorScatterNdAdd"));
-    opBuilder.addInput(tensor.asOutput());
-    opBuilder.addInput(indices.asOutput());
-    opBuilder.addInput(updates.asOutput());
+    opBuilder.addInput(tensor.asOutput(scope));
+    opBuilder.addInput(indices.asOutput(scope));
+    opBuilder.addInput(updates.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new TensorScatterNdAdd<T>(opBuilder.build());
   }
@@ -123,7 +123,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorScatterNdSub.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorScatterNdSub.java
index c05e46b4050..a97d99fc197 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorScatterNdSub.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorScatterNdSub.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * Subtracts sparse `updates` from an existing tensor according to `indices`.
@@ -93,7 +93,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator
-public final class TensorScatterNdSub<T extends Tensor> extends RawOp implements Operand<T> {
+public final class TensorScatterNdSub<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new TensorScatterNdSub operation.
@@ -105,11 +105,11 @@ public final class TensorScatterNdSub<T extends Tensor> extends RawOp implements
    * @return a new instance of TensorScatterNdSub
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor, U extends Tensor & TNumber> TensorScatterNdSub<T> create(Scope scope, Operand<T> tensor, Operand<U> indices, Operand<T> updates) {
+  public static <T extends TType, U extends TNumber> TensorScatterNdSub<T> create(Scope scope, Operand<T> tensor, Operand<U> indices, Operand<T> updates) {
     OperationBuilder opBuilder = scope.env().opBuilder("TensorScatterSub", scope.makeOpName("TensorScatterNdSub"));
-    opBuilder.addInput(tensor.asOutput());
-    opBuilder.addInput(indices.asOutput());
-    opBuilder.addInput(updates.asOutput());
+    opBuilder.addInput(tensor.asOutput(scope));
+    opBuilder.addInput(indices.asOutput(scope));
+    opBuilder.addInput(updates.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new TensorScatterNdSub<T>(opBuilder.build());
   }
@@ -122,7 +122,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorScatterNdUpdate.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorScatterNdUpdate.java
index c0147d8a796..0c4cf483e57 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorScatterNdUpdate.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorScatterNdUpdate.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * Scatter `updates` into an existing tensor according to `indices`.
@@ -108,7 +108,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator
-public final class TensorScatterNdUpdate<T extends Tensor> extends RawOp implements Operand<T> {
+public final class TensorScatterNdUpdate<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new TensorScatterNdUpdate operation.
@@ -120,11 +120,11 @@ public final class TensorScatterNdUpdate<T extends Tensor> extends RawOp impleme
    * @return a new instance of TensorScatterNdUpdate
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor, U extends Tensor & TNumber> TensorScatterNdUpdate<T> create(Scope scope, Operand<T> tensor, Operand<U> indices, Operand<T> updates) {
+  public static <T extends TType, U extends TNumber> TensorScatterNdUpdate<T> create(Scope scope, Operand<T> tensor, Operand<U> indices, Operand<T> updates) {
     OperationBuilder opBuilder = scope.env().opBuilder("TensorScatterUpdate", scope.makeOpName("TensorScatterNdUpdate"));
-    opBuilder.addInput(tensor.asOutput());
-    opBuilder.addInput(indices.asOutput());
-    opBuilder.addInput(updates.asOutput());
+    opBuilder.addInput(tensor.asOutput(scope));
+    opBuilder.addInput(indices.asOutput(scope));
+    opBuilder.addInput(updates.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new TensorScatterNdUpdate<T>(opBuilder.build());
   }
@@ -138,7 +138,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorStridedSliceUpdate.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorStridedSliceUpdate.java
index 932013a0a05..ff0777f607d 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorStridedSliceUpdate.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorStridedSliceUpdate.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * Assign `value` to the sliced l-value reference of `input`.
@@ -41,7 +41,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator
-public final class TensorStridedSliceUpdate<T extends Tensor> extends RawOp implements Operand<T> {
+public final class TensorStridedSliceUpdate<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.core.TensorStridedSliceUpdate}
@@ -111,13 +111,13 @@ private Options() {
    * @return a new instance of TensorStridedSliceUpdate
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor, U extends Tensor & TNumber> TensorStridedSliceUpdate<T> create(Scope scope, Operand<T> input, Operand<U> begin, Operand<U> end, Operand<U> strides, Operand<T> value, Options... options) {
+  public static <T extends TType, U extends TNumber> TensorStridedSliceUpdate<T> create(Scope scope, Operand<T> input, Operand<U> begin, Operand<U> end, Operand<U> strides, Operand<T> value, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("TensorStridedSliceUpdate", scope.makeOpName("TensorStridedSliceUpdate"));
-    opBuilder.addInput(input.asOutput());
-    opBuilder.addInput(begin.asOutput());
-    opBuilder.addInput(end.asOutput());
-    opBuilder.addInput(strides.asOutput());
-    opBuilder.addInput(value.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
+    opBuilder.addInput(begin.asOutput(scope));
+    opBuilder.addInput(end.asOutput(scope));
+    opBuilder.addInput(strides.asOutput(scope));
+    opBuilder.addInput(value.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -183,7 +183,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Tile.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Tile.java
index b6943bbd8b3..82749c7897e 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Tile.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Tile.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * Constructs a tensor by tiling a given tensor.
@@ -61,7 +61,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator
-public final class Tile<T extends Tensor> extends RawOp implements Operand<T> {
+public final class Tile<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Tile operation.
@@ -72,10 +72,10 @@ public final class Tile<T extends Tensor> extends RawOp implements Operand<T> {
    * @return a new instance of Tile
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor, U extends Tensor & TNumber> Tile<T> create(Scope scope, Operand<T> input, Operand<U> multiples) {
+  public static <T extends TType, U extends TNumber> Tile<T> create(Scope scope, Operand<T> input, Operand<U> multiples) {
     OperationBuilder opBuilder = scope.env().opBuilder("Tile", scope.makeOpName("Tile"));
-    opBuilder.addInput(input.asOutput());
-    opBuilder.addInput(multiples.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
+    opBuilder.addInput(multiples.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new Tile<T>(opBuilder.build());
   }
@@ -87,7 +87,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Timestamp.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Timestamp.java
index cbbf650a32f..f0830c6ffe2 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Timestamp.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Timestamp.java
@@ -58,7 +58,7 @@ public Output<TFloat64> ts() {
   }
   
   @Override
-  public Output<TFloat64> asOutput() {
+  public Output<TFloat64> asOutput(Scope scope) {
     return ts;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TryRpc.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TryRpc.java
index 3aa6c3a76ad..72e98564759 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TryRpc.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TryRpc.java
@@ -141,9 +141,9 @@ private Options() {
   @Endpoint(describeByClass = true)
   public static TryRpc create(Scope scope, Operand<TString> address, Operand<TString> method, Operand<TString> request, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("TryRpc", scope.makeOpName("TryRpc"));
-    opBuilder.addInput(address.asOutput());
-    opBuilder.addInput(method.asOutput());
-    opBuilder.addInput(request.asOutput());
+    opBuilder.addInput(address.asOutput(scope));
+    opBuilder.addInput(method.asOutput(scope));
+    opBuilder.addInput(request.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Unbatch.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Unbatch.java
index 25fcf101630..266df8c4af5 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Unbatch.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Unbatch.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
 
 /**
  * Reverses the operation of Batch for a single output Tensor.
@@ -53,7 +53,7 @@
  * @param <T> data type for {@code unbatchedTensor()} output
  */
 @Operator
-public final class Unbatch<T extends Tensor> extends RawOp implements Operand<T> {
+public final class Unbatch<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.core.Unbatch}
@@ -95,11 +95,11 @@ private Options() {
    * @return a new instance of Unbatch
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> Unbatch<T> create(Scope scope, Operand<T> batchedTensor, Operand<TInt64> batchIndex, Operand<TInt64> id, Long timeoutMicros, Options... options) {
+  public static <T extends TType> Unbatch<T> create(Scope scope, Operand<T> batchedTensor, Operand<TInt64> batchIndex, Operand<TInt64> id, Long timeoutMicros, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("Unbatch", scope.makeOpName("Unbatch"));
-    opBuilder.addInput(batchedTensor.asOutput());
-    opBuilder.addInput(batchIndex.asOutput());
-    opBuilder.addInput(id.asOutput());
+    opBuilder.addInput(batchedTensor.asOutput(scope));
+    opBuilder.addInput(batchIndex.asOutput(scope));
+    opBuilder.addInput(id.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("timeout_micros", timeoutMicros);
     if (options != null) {
@@ -136,7 +136,7 @@ public Output<T> unbatchedTensor() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return unbatchedTensor;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/UnbatchGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/UnbatchGrad.java
index 332b3b6967c..b821aaac605 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/UnbatchGrad.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/UnbatchGrad.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
 
 /**
  * Gradient of Unbatch.
@@ -49,7 +49,7 @@
  * @param <T> data type for {@code batchedGrad()} output
  */
 @Operator
-public final class UnbatchGrad<T extends Tensor> extends RawOp implements Operand<T> {
+public final class UnbatchGrad<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.core.UnbatchGrad}
@@ -91,12 +91,12 @@ private Options() {
    * @return a new instance of UnbatchGrad
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> UnbatchGrad<T> create(Scope scope, Operand<T> originalInput, Operand<TInt64> batchIndex, Operand<T> grad, Operand<TInt64> id, Options... options) {
+  public static <T extends TType> UnbatchGrad<T> create(Scope scope, Operand<T> originalInput, Operand<TInt64> batchIndex, Operand<T> grad, Operand<TInt64> id, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("UnbatchGrad", scope.makeOpName("UnbatchGrad"));
-    opBuilder.addInput(originalInput.asOutput());
-    opBuilder.addInput(batchIndex.asOutput());
-    opBuilder.addInput(grad.asOutput());
-    opBuilder.addInput(id.asOutput());
+    opBuilder.addInput(originalInput.asOutput(scope));
+    opBuilder.addInput(batchIndex.asOutput(scope));
+    opBuilder.addInput(grad.asOutput(scope));
+    opBuilder.addInput(id.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -132,7 +132,7 @@ public Output<T> batchedGrad() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return batchedGrad;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Unique.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Unique.java
index bb5f21ccacb..a3f64684cd7 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Unique.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Unique.java
@@ -22,13 +22,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * Finds unique elements along an axis of a tensor.
@@ -77,7 +77,7 @@
  * @param <V> data type for {@code idx()} output
  */
 @Operator
-public final class Unique<T extends Tensor, V extends Tensor & TNumber> extends RawOp {
+public final class Unique<T extends TType, V extends TNumber> extends RawOp {
   
   /**
    * Factory method to create a class wrapping a new Unique operation.
@@ -90,10 +90,10 @@ public final class Unique<T extends Tensor, V extends Tensor & TNumber> extends
    * @return a new instance of Unique
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor, V extends Tensor & TNumber, U extends Tensor & TNumber> Unique<T, V> create(Scope scope, Operand<T> x, Operand<U> axis, DataType<V> outIdx) {
+  public static <T extends TType, V extends TNumber, U extends TNumber> Unique<T, V> create(Scope scope, Operand<T> x, Operand<U> axis, DataType<V> outIdx) {
     OperationBuilder opBuilder = scope.env().opBuilder("UniqueV2", scope.makeOpName("Unique"));
-    opBuilder.addInput(x.asOutput());
-    opBuilder.addInput(axis.asOutput());
+    opBuilder.addInput(x.asOutput(scope));
+    opBuilder.addInput(axis.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("out_idx", outIdx);
     return new Unique<T, V>(opBuilder.build());
@@ -109,7 +109,7 @@ public static <T extends Tensor, V extends Tensor & TNumber, U extends Tensor &
    * @return a new instance of Unique
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor, U extends Tensor & TNumber> Unique<T, TInt32> create(Scope scope, Operand<T> x, Operand<U> axis) {
+  public static <T extends TType, U extends TNumber> Unique<T, TInt32> create(Scope scope, Operand<T> x, Operand<U> axis) {
     return create(scope, x, axis, TInt32.DTYPE);
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/UniqueWithCounts.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/UniqueWithCounts.java
index bfdc83e0edf..62804827da5 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/UniqueWithCounts.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/UniqueWithCounts.java
@@ -22,13 +22,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * Finds unique elements along an axis of a tensor.
@@ -81,7 +81,7 @@
  * @param <V> data type for {@code idx()} output
  */
 @Operator
-public final class UniqueWithCounts<T extends Tensor, V extends Tensor & TNumber> extends RawOp {
+public final class UniqueWithCounts<T extends TType, V extends TNumber> extends RawOp {
   
   /**
    * Factory method to create a class wrapping a new UniqueWithCounts operation.
@@ -94,10 +94,10 @@ public final class UniqueWithCounts<T extends Tensor, V extends Tensor & TNumber
    * @return a new instance of UniqueWithCounts
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor, V extends Tensor & TNumber, U extends Tensor & TNumber> UniqueWithCounts<T, V> create(Scope scope, Operand<T> x, Operand<U> axis, DataType<V> outIdx) {
+  public static <T extends TType, V extends TNumber, U extends TNumber> UniqueWithCounts<T, V> create(Scope scope, Operand<T> x, Operand<U> axis, DataType<V> outIdx) {
     OperationBuilder opBuilder = scope.env().opBuilder("UniqueWithCountsV2", scope.makeOpName("UniqueWithCounts"));
-    opBuilder.addInput(x.asOutput());
-    opBuilder.addInput(axis.asOutput());
+    opBuilder.addInput(x.asOutput(scope));
+    opBuilder.addInput(axis.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("out_idx", outIdx);
     return new UniqueWithCounts<T, V>(opBuilder.build());
@@ -113,7 +113,7 @@ public static <T extends Tensor, V extends Tensor & TNumber, U extends Tensor &
    * @return a new instance of UniqueWithCounts
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor, U extends Tensor & TNumber> UniqueWithCounts<T, TInt32> create(Scope scope, Operand<T> x, Operand<U> axis) {
+  public static <T extends TType, U extends TNumber> UniqueWithCounts<T, TInt32> create(Scope scope, Operand<T> x, Operand<U> axis) {
     return create(scope, x, axis, TInt32.DTYPE);
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/UnravelIndex.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/UnravelIndex.java
index 1f3c5c5c42d..7f7c8b50bf8 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/UnravelIndex.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/UnravelIndex.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -53,7 +52,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator
-public final class UnravelIndex<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class UnravelIndex<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new UnravelIndex operation.
@@ -66,10 +65,10 @@ public final class UnravelIndex<T extends Tensor & TNumber> extends RawOp implem
    * @return a new instance of UnravelIndex
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> UnravelIndex<T> create(Scope scope, Operand<T> indices, Operand<T> dims) {
+  public static <T extends TNumber> UnravelIndex<T> create(Scope scope, Operand<T> indices, Operand<T> dims) {
     OperationBuilder opBuilder = scope.env().opBuilder("UnravelIndex", scope.makeOpName("UnravelIndex"));
-    opBuilder.addInput(indices.asOutput());
-    opBuilder.addInput(dims.asOutput());
+    opBuilder.addInput(indices.asOutput(scope));
+    opBuilder.addInput(dims.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new UnravelIndex<T>(opBuilder.build());
   }
@@ -83,7 +82,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Unstack.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Unstack.java
index 4a1f2757be3..4902acb7ae3 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Unstack.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Unstack.java
@@ -24,11 +24,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Unpacks a given dimension of a rank-`R` tensor into `num` rank-`(R-1)` tensors.
@@ -49,7 +49,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator
-public final class Unstack<T extends Tensor> extends RawOp implements Iterable<Operand<T>> {
+public final class Unstack<T extends TType> extends RawOp implements Iterable<Operand<T>> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.core.Unstack}
@@ -81,9 +81,9 @@ private Options() {
    * @return a new instance of Unstack
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> Unstack<T> create(Scope scope, Operand<T> value, Long num, Options... options) {
+  public static <T extends TType> Unstack<T> create(Scope scope, Operand<T> value, Long num, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("Unpack", scope.makeOpName("Unstack"));
-    opBuilder.addInput(value.asOutput());
+    opBuilder.addInput(value.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("num", num);
     if (options != null) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Unstage.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Unstage.java
index 4f683ed6775..1878ce6fb88 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Unstage.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Unstage.java
@@ -25,11 +25,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Op is similar to a lightweight Dequeue.
@@ -38,7 +38,7 @@
  * capabilities and options.  This Op is optimized for performance.
  */
 @Operator
-public final class Unstage extends RawOp implements Iterable<Operand<Tensor>> {
+public final class Unstage extends RawOp implements Iterable<Operand<TType>> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.core.Unstage}
@@ -158,7 +158,7 @@ public List<Output<?>> values() {
   
   @Override
   @SuppressWarnings({"rawtypes", "unchecked"})
-  public Iterator<Operand<Tensor>> iterator() {
+  public Iterator<Operand<TType>> iterator() {
     return (Iterator) values.iterator();
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/UpperBound.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/UpperBound.java
index d0338f5b1d9..f7117601ec3 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/UpperBound.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/UpperBound.java
@@ -22,13 +22,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * Applies upper_bound(sorted_search_values, values) along each row.
@@ -53,7 +53,7 @@
  * 
  * @param <U> data type for {@code output()} output
  */
-public final class UpperBound<U extends Tensor & TNumber> extends RawOp implements Operand<U> {
+public final class UpperBound<U extends TNumber> extends RawOp implements Operand<U> {
   
   /**
    * Factory method to create a class wrapping a new UpperBound operation.
@@ -66,10 +66,10 @@ public final class UpperBound<U extends Tensor & TNumber> extends RawOp implemen
    * @return a new instance of UpperBound
    */
   @Endpoint(describeByClass = true)
-  public static <U extends Tensor & TNumber, T extends Tensor> UpperBound<U> create(Scope scope, Operand<T> sortedInputs, Operand<T> values, DataType<U> outType) {
+  public static <U extends TNumber, T extends TType> UpperBound<U> create(Scope scope, Operand<T> sortedInputs, Operand<T> values, DataType<U> outType) {
     OperationBuilder opBuilder = scope.env().opBuilder("UpperBound", scope.makeOpName("UpperBound"));
-    opBuilder.addInput(sortedInputs.asOutput());
-    opBuilder.addInput(values.asOutput());
+    opBuilder.addInput(sortedInputs.asOutput(scope));
+    opBuilder.addInput(values.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("out_type", outType);
     return new UpperBound<U>(opBuilder.build());
@@ -85,7 +85,7 @@ public static <U extends Tensor & TNumber, T extends Tensor> UpperBound<U> creat
    * @return a new instance of UpperBound
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> UpperBound<TInt32> create(Scope scope, Operand<T> sortedInputs, Operand<T> values) {
+  public static <T extends TType> UpperBound<TInt32> create(Scope scope, Operand<T> sortedInputs, Operand<T> values) {
     return create(scope, sortedInputs, values, TInt32.DTYPE);
   }
   
@@ -99,7 +99,7 @@ public Output<U> output() {
   }
   
   @Override
-  public Output<U> asOutput() {
+  public Output<U> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/VarHandleOp.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/VarHandleOp.java
index 2cbd51dda7f..2bd97143dcb 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/VarHandleOp.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/VarHandleOp.java
@@ -23,18 +23,18 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Creates a handle to a Variable resource.
  */
 @Operator
-public final class VarHandleOp extends RawOp implements Operand<Tensor> {
+public final class VarHandleOp extends RawOp implements Operand<TType> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.core.VarHandleOp}
@@ -85,7 +85,7 @@ private Options() {
    * @return a new instance of VarHandleOp
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> VarHandleOp create(Scope scope, DataType<T> dtype, Shape shape, Options... options) {
+  public static <T extends TType> VarHandleOp create(Scope scope, DataType<T> dtype, Shape shape, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("VarHandleOp", scope.makeOpName("VarHandleOp"));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("dtype", dtype);
@@ -140,8 +140,8 @@ public Output<?> resource() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<Tensor> asOutput() {
-    return (Output<Tensor>) resource;
+  public Output<TType> asOutput(Scope scope) {
+    return (Output<TType>) resource;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/VarIsInitializedOp.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/VarIsInitializedOp.java
index f2c3df0456d..6fa5ffc7162 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/VarIsInitializedOp.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/VarIsInitializedOp.java
@@ -43,7 +43,7 @@ public final class VarIsInitializedOp extends RawOp implements Operand<TBool> {
   @Endpoint(describeByClass = true)
   public static VarIsInitializedOp create(Scope scope, Operand<?> resource) {
     OperationBuilder opBuilder = scope.env().opBuilder("VarIsInitializedOp", scope.makeOpName("VarIsInitializedOp"));
-    opBuilder.addInput(resource.asOutput());
+    opBuilder.addInput(resource.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new VarIsInitializedOp(opBuilder.build());
   }
@@ -57,7 +57,7 @@ public Output<TBool> isInitialized() {
   }
   
   @Override
-  public Output<TBool> asOutput() {
+  public Output<TBool> asOutput(Scope scope) {
     return isInitialized;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Variable.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Variable.java
index 94c577c329d..1c0c9151156 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Variable.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Variable.java
@@ -22,12 +22,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Holds state in the form of a tensor that persists across steps.
@@ -39,7 +39,7 @@
  * @param <T> data type for {@code ref()} output
  */
 @Operator
-public final class Variable<T extends Tensor> extends RawOp implements Operand<T> {
+public final class Variable<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.core.Variable}
@@ -81,7 +81,7 @@ private Options() {
    * @return a new instance of Variable
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> Variable<T> create(Scope scope, Shape shape, DataType<T> dtype, Options... options) {
+  public static <T extends TType> Variable<T> create(Scope scope, Shape shape, DataType<T> dtype, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("VariableV2", scope.makeOpName("Variable"));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("shape", shape);
@@ -123,7 +123,7 @@ public Output<T> ref() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return ref;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/VariableShape.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/VariableShape.java
index 94ee6717f3d..97ae695bcc4 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/VariableShape.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/VariableShape.java
@@ -22,7 +22,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -45,7 +44,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator
-public final class VariableShape<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class VariableShape<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new VariableShape operation.
@@ -56,9 +55,9 @@ public final class VariableShape<T extends Tensor & TNumber> extends RawOp imple
    * @return a new instance of VariableShape
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> VariableShape<T> create(Scope scope, Operand<?> input, DataType<T> outType) {
+  public static <T extends TNumber> VariableShape<T> create(Scope scope, Operand<?> input, DataType<T> outType) {
     OperationBuilder opBuilder = scope.env().opBuilder("VariableShape", scope.makeOpName("VariableShape"));
-    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("out_type", outType);
     return new VariableShape<T>(opBuilder.build());
@@ -83,7 +82,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Where.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Where.java
index 8419104bfbf..791ba807489 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Where.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Where.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
 
 /**
  * Returns locations of nonzero / true values in a tensor.
@@ -102,9 +102,9 @@ public final class Where extends RawOp implements Operand<TInt64> {
    * @return a new instance of Where
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> Where create(Scope scope, Operand<T> condition) {
+  public static <T extends TType> Where create(Scope scope, Operand<T> condition) {
     OperationBuilder opBuilder = scope.env().opBuilder("Where", scope.makeOpName("Where"));
-    opBuilder.addInput(condition.asOutput());
+    opBuilder.addInput(condition.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new Where(opBuilder.build());
   }
@@ -116,7 +116,7 @@ public Output<TInt64> index() {
   }
   
   @Override
-  public Output<TInt64> asOutput() {
+  public Output<TInt64> asOutput(Scope scope) {
     return index;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ZerosLike.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ZerosLike.java
index 818cde42d9c..1d79777cc35 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ZerosLike.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ZerosLike.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Returns a tensor of zeros with the same shape and type as x.
@@ -33,7 +33,7 @@
  * @param <T> data type for {@code y()} output
  */
 @Operator
-public final class ZerosLike<T extends Tensor> extends RawOp implements Operand<T> {
+public final class ZerosLike<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new ZerosLike operation.
@@ -43,9 +43,9 @@ public final class ZerosLike<T extends Tensor> extends RawOp implements Operand<
    * @return a new instance of ZerosLike
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> ZerosLike<T> create(Scope scope, Operand<T> x) {
+  public static <T extends TType> ZerosLike<T> create(Scope scope, Operand<T> x) {
     OperationBuilder opBuilder = scope.env().opBuilder("ZerosLike", scope.makeOpName("ZerosLike"));
-    opBuilder.addInput(x.asOutput());
+    opBuilder.addInput(x.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new ZerosLike<T>(opBuilder.build());
   }
@@ -58,7 +58,7 @@ public Output<T> y() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return y;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/AssertNextDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/AssertNextDataset.java
index 5ffd7510dbc..ee9c8977d29 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/AssertNextDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/AssertNextDataset.java
@@ -23,13 +23,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TString;
+import org.tensorflow.types.family.TType;
 
 /**
  * A transformation that asserts which transformations happen next.
@@ -43,7 +43,7 @@
  * means that the check happens <i>after</i> any static optimizations are applied
  * to the dataset graph.
  */
-public final class AssertNextDataset extends RawOp implements Operand<Tensor> {
+public final class AssertNextDataset extends RawOp implements Operand<TType> {
   
   /**
    * Factory method to create a class wrapping a new AssertNextDataset operation.
@@ -60,8 +60,8 @@ public final class AssertNextDataset extends RawOp implements Operand<Tensor> {
   @Endpoint(describeByClass = true)
   public static AssertNextDataset create(Scope scope, Operand<?> inputDataset, Operand<TString> transformations, List<DataType<?>> outputTypes, List<Shape> outputShapes) {
     OperationBuilder opBuilder = scope.env().opBuilder("AssertNextDataset", scope.makeOpName("AssertNextDataset"));
-    opBuilder.addInput(inputDataset.asOutput());
-    opBuilder.addInput(transformations.asOutput());
+    opBuilder.addInput(inputDataset.asOutput(scope));
+    opBuilder.addInput(transformations.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     DataType[] outputTypesArray = new DataType[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
@@ -84,8 +84,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<Tensor> asOutput() {
-    return (Output<Tensor>) handle;
+  public Output<TType> asOutput(Scope scope) {
+    return (Output<TType>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/AutoShardDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/AutoShardDataset.java
index 54ccd69f19c..180ba4f5d71 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/AutoShardDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/AutoShardDataset.java
@@ -23,13 +23,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
 
 /**
  * Creates a dataset that shards the input dataset.
@@ -42,7 +42,7 @@
  * This dataset will throw a NotFound error if we cannot shard the dataset
  * automatically.
  */
-public final class AutoShardDataset extends RawOp implements Operand<Tensor> {
+public final class AutoShardDataset extends RawOp implements Operand<TType> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.data.AutoShardDataset}
@@ -78,9 +78,9 @@ private Options() {
   @Endpoint(describeByClass = true)
   public static AutoShardDataset create(Scope scope, Operand<?> inputDataset, Operand<TInt64> numWorkers, Operand<TInt64> index, List<DataType<?>> outputTypes, List<Shape> outputShapes, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("AutoShardDataset", scope.makeOpName("AutoShardDataset"));
-    opBuilder.addInput(inputDataset.asOutput());
-    opBuilder.addInput(numWorkers.asOutput());
-    opBuilder.addInput(index.asOutput());
+    opBuilder.addInput(inputDataset.asOutput(scope));
+    opBuilder.addInput(numWorkers.asOutput(scope));
+    opBuilder.addInput(index.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     DataType[] outputTypesArray = new DataType[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
@@ -117,8 +117,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<Tensor> asOutput() {
-    return (Output<Tensor>) handle;
+  public Output<TType> asOutput(Scope scope) {
+    return (Output<TType>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/BatchDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/BatchDataset.java
index 3663c9d7bc4..7a827ad43eb 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/BatchDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/BatchDataset.java
@@ -23,7 +23,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
@@ -31,12 +30,13 @@
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TBool;
 import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
 
 /**
  * Creates a dataset that batches `batch_size` elements from `input_dataset`.
  */
 @Operator(group = "data")
-public final class BatchDataset extends RawOp implements Operand<Tensor> {
+public final class BatchDataset extends RawOp implements Operand<TType> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.data.BatchDataset}
@@ -73,9 +73,9 @@ private Options() {
   @Endpoint(describeByClass = true)
   public static BatchDataset create(Scope scope, Operand<?> inputDataset, Operand<TInt64> batchSize, Operand<TBool> dropRemainder, List<DataType<?>> outputTypes, List<Shape> outputShapes, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("BatchDatasetV2", scope.makeOpName("BatchDataset"));
-    opBuilder.addInput(inputDataset.asOutput());
-    opBuilder.addInput(batchSize.asOutput());
-    opBuilder.addInput(dropRemainder.asOutput());
+    opBuilder.addInput(inputDataset.asOutput(scope));
+    opBuilder.addInput(batchSize.asOutput(scope));
+    opBuilder.addInput(dropRemainder.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     DataType[] outputTypesArray = new DataType[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
@@ -112,8 +112,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<Tensor> asOutput() {
-    return (Output<Tensor>) handle;
+  public Output<TType> asOutput(Scope scope) {
+    return (Output<TType>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/BytesProducedStatsDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/BytesProducedStatsDataset.java
index 9597c4063c2..7b73e98c293 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/BytesProducedStatsDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/BytesProducedStatsDataset.java
@@ -23,18 +23,18 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TString;
+import org.tensorflow.types.family.TType;
 
 /**
  * Records the bytes size of each element of `input_dataset` in a StatsAggregator.
  */
-public final class BytesProducedStatsDataset extends RawOp implements Operand<Tensor> {
+public final class BytesProducedStatsDataset extends RawOp implements Operand<TType> {
   
   /**
    * Factory method to create a class wrapping a new BytesProducedStatsDataset operation.
@@ -49,8 +49,8 @@ public final class BytesProducedStatsDataset extends RawOp implements Operand<Te
   @Endpoint(describeByClass = true)
   public static BytesProducedStatsDataset create(Scope scope, Operand<?> inputDataset, Operand<TString> tag, List<DataType<?>> outputTypes, List<Shape> outputShapes) {
     OperationBuilder opBuilder = scope.env().opBuilder("BytesProducedStatsDataset", scope.makeOpName("BytesProducedStatsDataset"));
-    opBuilder.addInput(inputDataset.asOutput());
-    opBuilder.addInput(tag.asOutput());
+    opBuilder.addInput(inputDataset.asOutput(scope));
+    opBuilder.addInput(tag.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     DataType[] outputTypesArray = new DataType[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
@@ -73,8 +73,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<Tensor> asOutput() {
-    return (Output<Tensor>) handle;
+  public Output<TType> asOutput(Scope scope) {
+    return (Output<TType>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/CSVDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/CSVDataset.java
index 2781151f259..1466a507bfc 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/CSVDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/CSVDataset.java
@@ -22,7 +22,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.Operands;
 import org.tensorflow.op.RawOp;
@@ -32,11 +31,12 @@
 import org.tensorflow.types.TBool;
 import org.tensorflow.types.TInt64;
 import org.tensorflow.types.TString;
+import org.tensorflow.types.family.TType;
 
 /**
  */
 @Operator(group = "data")
-public final class CSVDataset extends RawOp implements Operand<Tensor> {
+public final class CSVDataset extends RawOp implements Operand<TType> {
   
   /**
    * Factory method to create a class wrapping a new CSVDataset operation.
@@ -57,15 +57,15 @@ public final class CSVDataset extends RawOp implements Operand<Tensor> {
   @Endpoint(describeByClass = true)
   public static CSVDataset create(Scope scope, Operand<TString> filenames, Operand<TString> compressionType, Operand<TInt64> bufferSize, Operand<TBool> header, Operand<TString> fieldDelim, Operand<TBool> useQuoteDelim, Operand<TString> naValue, Operand<TInt64> selectCols, Iterable<Operand<?>> recordDefaults, List<Shape> outputShapes) {
     OperationBuilder opBuilder = scope.env().opBuilder("CSVDataset", scope.makeOpName("CSVDataset"));
-    opBuilder.addInput(filenames.asOutput());
-    opBuilder.addInput(compressionType.asOutput());
-    opBuilder.addInput(bufferSize.asOutput());
-    opBuilder.addInput(header.asOutput());
-    opBuilder.addInput(fieldDelim.asOutput());
-    opBuilder.addInput(useQuoteDelim.asOutput());
-    opBuilder.addInput(naValue.asOutput());
-    opBuilder.addInput(selectCols.asOutput());
-    opBuilder.addInputList(Operands.asOutputs(recordDefaults));
+    opBuilder.addInput(filenames.asOutput(scope));
+    opBuilder.addInput(compressionType.asOutput(scope));
+    opBuilder.addInput(bufferSize.asOutput(scope));
+    opBuilder.addInput(header.asOutput(scope));
+    opBuilder.addInput(fieldDelim.asOutput(scope));
+    opBuilder.addInput(useQuoteDelim.asOutput(scope));
+    opBuilder.addInput(naValue.asOutput(scope));
+    opBuilder.addInput(selectCols.asOutput(scope));
+    opBuilder.addInputList(Operands.asOutputs(scope, recordDefaults));
     opBuilder = scope.applyControlDependencies(opBuilder);
     Shape[] outputShapesArray = new Shape[outputShapes.size()];
     for (int i = 0; i < outputShapesArray.length; ++i) {
@@ -83,8 +83,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<Tensor> asOutput() {
-    return (Output<Tensor>) handle;
+  public Output<TType> asOutput(Scope scope) {
+    return (Output<TType>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/CacheDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/CacheDataset.java
index b15005af198..1dd8e604bc4 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/CacheDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/CacheDataset.java
@@ -23,13 +23,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TString;
+import org.tensorflow.types.family.TType;
 
 /**
  * Creates a dataset that caches elements from `input_dataset`.
@@ -39,7 +39,7 @@
  * (e.g. cannot be opened, contains tensors of the wrong shape / size), an error
  * will the returned when used.
  */
-public final class CacheDataset extends RawOp implements Operand<Tensor> {
+public final class CacheDataset extends RawOp implements Operand<TType> {
   
   /**
    * Factory method to create a class wrapping a new CacheDataset operation.
@@ -55,8 +55,8 @@ public final class CacheDataset extends RawOp implements Operand<Tensor> {
   @Endpoint(describeByClass = true)
   public static CacheDataset create(Scope scope, Operand<?> inputDataset, Operand<TString> filename, List<DataType<?>> outputTypes, List<Shape> outputShapes) {
     OperationBuilder opBuilder = scope.env().opBuilder("CacheDataset", scope.makeOpName("CacheDataset"));
-    opBuilder.addInput(inputDataset.asOutput());
-    opBuilder.addInput(filename.asOutput());
+    opBuilder.addInput(inputDataset.asOutput(scope));
+    opBuilder.addInput(filename.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     DataType[] outputTypesArray = new DataType[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
@@ -79,8 +79,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<Tensor> asOutput() {
-    return (Output<Tensor>) handle;
+  public Output<TType> asOutput(Scope scope) {
+    return (Output<TType>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/CacheDatasetV2.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/CacheDatasetV2.java
index 5b111851102..7dac3483939 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/CacheDatasetV2.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/CacheDatasetV2.java
@@ -23,17 +23,17 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TString;
+import org.tensorflow.types.family.TType;
 
 /**
  */
-public final class CacheDatasetV2 extends RawOp implements Operand<Tensor> {
+public final class CacheDatasetV2 extends RawOp implements Operand<TType> {
   
   /**
    * Factory method to create a class wrapping a new CacheDatasetV2 operation.
@@ -49,9 +49,9 @@ public final class CacheDatasetV2 extends RawOp implements Operand<Tensor> {
   @Endpoint(describeByClass = true)
   public static CacheDatasetV2 create(Scope scope, Operand<?> inputDataset, Operand<TString> filename, Operand<?> cache, List<DataType<?>> outputTypes, List<Shape> outputShapes) {
     OperationBuilder opBuilder = scope.env().opBuilder("CacheDatasetV2", scope.makeOpName("CacheDatasetV2"));
-    opBuilder.addInput(inputDataset.asOutput());
-    opBuilder.addInput(filename.asOutput());
-    opBuilder.addInput(cache.asOutput());
+    opBuilder.addInput(inputDataset.asOutput(scope));
+    opBuilder.addInput(filename.asOutput(scope));
+    opBuilder.addInput(cache.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     DataType[] outputTypesArray = new DataType[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
@@ -74,8 +74,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<Tensor> asOutput() {
-    return (Output<Tensor>) handle;
+  public Output<TType> asOutput(Scope scope) {
+    return (Output<TType>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ChooseFastestDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ChooseFastestDataset.java
index e7726e4e5a9..e1e2e718f21 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ChooseFastestDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ChooseFastestDataset.java
@@ -23,17 +23,17 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.Operands;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  */
-public final class ChooseFastestDataset extends RawOp implements Operand<Tensor> {
+public final class ChooseFastestDataset extends RawOp implements Operand<TType> {
   
   /**
    * Factory method to create a class wrapping a new ChooseFastestDataset operation.
@@ -48,7 +48,7 @@ public final class ChooseFastestDataset extends RawOp implements Operand<Tensor>
   @Endpoint(describeByClass = true)
   public static ChooseFastestDataset create(Scope scope, Iterable<Operand<?>> inputDatasets, Long numExperiments, List<DataType<?>> outputTypes, List<Shape> outputShapes) {
     OperationBuilder opBuilder = scope.env().opBuilder("ChooseFastestDataset", scope.makeOpName("ChooseFastestDataset"));
-    opBuilder.addInputList(Operands.asOutputs(inputDatasets));
+    opBuilder.addInputList(Operands.asOutputs(scope, inputDatasets));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("num_experiments", numExperiments);
     DataType[] outputTypesArray = new DataType[outputTypes.size()];
@@ -72,8 +72,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<Tensor> asOutput() {
-    return (Output<Tensor>) handle;
+  public Output<TType> asOutput(Scope scope) {
+    return (Output<TType>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ConcatenateDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ConcatenateDataset.java
index 031296a8468..7eb4581f462 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ConcatenateDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ConcatenateDataset.java
@@ -23,18 +23,18 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Creates a dataset that concatenates `input_dataset` with `another_dataset`.
  */
 @Operator(group = "data")
-public final class ConcatenateDataset extends RawOp implements Operand<Tensor> {
+public final class ConcatenateDataset extends RawOp implements Operand<TType> {
   
   /**
    * Factory method to create a class wrapping a new ConcatenateDataset operation.
@@ -49,8 +49,8 @@ public final class ConcatenateDataset extends RawOp implements Operand<Tensor> {
   @Endpoint(describeByClass = true)
   public static ConcatenateDataset create(Scope scope, Operand<?> inputDataset, Operand<?> anotherDataset, List<DataType<?>> outputTypes, List<Shape> outputShapes) {
     OperationBuilder opBuilder = scope.env().opBuilder("ConcatenateDataset", scope.makeOpName("ConcatenateDataset"));
-    opBuilder.addInput(inputDataset.asOutput());
-    opBuilder.addInput(anotherDataset.asOutput());
+    opBuilder.addInput(inputDataset.asOutput(scope));
+    opBuilder.addInput(anotherDataset.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     DataType[] outputTypesArray = new DataType[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
@@ -73,8 +73,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<Tensor> asOutput() {
-    return (Output<Tensor>) handle;
+  public Output<TType> asOutput(Scope scope) {
+    return (Output<TType>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DatasetCardinality.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DatasetCardinality.java
index d851c211420..71b02bb2977 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DatasetCardinality.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DatasetCardinality.java
@@ -44,7 +44,7 @@ public final class DatasetCardinality extends RawOp implements Operand<TInt64> {
   @Endpoint(describeByClass = true)
   public static DatasetCardinality create(Scope scope, Operand<?> inputDataset) {
     OperationBuilder opBuilder = scope.env().opBuilder("DatasetCardinality", scope.makeOpName("DatasetCardinality"));
-    opBuilder.addInput(inputDataset.asOutput());
+    opBuilder.addInput(inputDataset.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new DatasetCardinality(opBuilder.build());
   }
@@ -58,7 +58,7 @@ public Output<TInt64> cardinality() {
   }
   
   @Override
-  public Output<TInt64> asOutput() {
+  public Output<TInt64> asOutput(Scope scope) {
     return cardinality;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DatasetFromGraph.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DatasetFromGraph.java
index 7d205ad6bf6..feaa06769fc 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DatasetFromGraph.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DatasetFromGraph.java
@@ -21,19 +21,19 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TString;
+import org.tensorflow.types.family.TType;
 
 /**
  * Creates a dataset from the given `graph_def`.
  * <p>
  * Creates a dataset from the provided `graph_def`.
  */
-public final class DatasetFromGraph extends RawOp implements Operand<Tensor> {
+public final class DatasetFromGraph extends RawOp implements Operand<TType> {
   
   /**
    * Factory method to create a class wrapping a new DatasetFromGraph operation.
@@ -45,7 +45,7 @@ public final class DatasetFromGraph extends RawOp implements Operand<Tensor> {
   @Endpoint(describeByClass = true)
   public static DatasetFromGraph create(Scope scope, Operand<TString> graphDef) {
     OperationBuilder opBuilder = scope.env().opBuilder("DatasetFromGraph", scope.makeOpName("DatasetFromGraph"));
-    opBuilder.addInput(graphDef.asOutput());
+    opBuilder.addInput(graphDef.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new DatasetFromGraph(opBuilder.build());
   }
@@ -59,8 +59,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<Tensor> asOutput() {
-    return (Output<Tensor>) handle;
+  public Output<TType> asOutput(Scope scope) {
+    return (Output<TType>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DatasetToGraph.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DatasetToGraph.java
index afaab68330a..805cc39b424 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DatasetToGraph.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DatasetToGraph.java
@@ -73,7 +73,7 @@ private Options() {
   @Endpoint(describeByClass = true)
   public static DatasetToGraph create(Scope scope, Operand<?> inputDataset, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("DatasetToGraphV2", scope.makeOpName("DatasetToGraph"));
-    opBuilder.addInput(inputDataset.asOutput());
+    opBuilder.addInput(inputDataset.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -110,7 +110,7 @@ public Output<TString> graph() {
   }
   
   @Override
-  public Output<TString> asOutput() {
+  public Output<TString> asOutput(Scope scope) {
     return graph;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DatasetToSingleElement.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DatasetToSingleElement.java
index ad7531dbfb2..69c4a717eb4 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DatasetToSingleElement.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DatasetToSingleElement.java
@@ -25,17 +25,17 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Outputs the single element from the given dataset.
  */
-public final class DatasetToSingleElement extends RawOp implements Iterable<Operand<Tensor>> {
+public final class DatasetToSingleElement extends RawOp implements Iterable<Operand<TType>> {
   
   /**
    * Factory method to create a class wrapping a new DatasetToSingleElement operation.
@@ -49,7 +49,7 @@ public final class DatasetToSingleElement extends RawOp implements Iterable<Oper
   @Endpoint(describeByClass = true)
   public static DatasetToSingleElement create(Scope scope, Operand<?> dataset, List<DataType<?>> outputTypes, List<Shape> outputShapes) {
     OperationBuilder opBuilder = scope.env().opBuilder("DatasetToSingleElement", scope.makeOpName("DatasetToSingleElement"));
-    opBuilder.addInput(dataset.asOutput());
+    opBuilder.addInput(dataset.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     DataType[] outputTypesArray = new DataType[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
@@ -73,7 +73,7 @@ public List<Output<?>> components() {
   
   @Override
   @SuppressWarnings({"rawtypes", "unchecked"})
-  public Iterator<Operand<Tensor>> iterator() {
+  public Iterator<Operand<TType>> iterator() {
     return (Iterator) components.iterator();
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DatasetToTfRecord.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DatasetToTfRecord.java
index 114e11074dc..016eec90e86 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DatasetToTfRecord.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DatasetToTfRecord.java
@@ -44,9 +44,9 @@ public final class DatasetToTfRecord extends RawOp {
   @Endpoint(describeByClass = true)
   public static DatasetToTfRecord create(Scope scope, Operand<?> inputDataset, Operand<TString> filename, Operand<TString> compressionType) {
     OperationBuilder opBuilder = scope.env().opBuilder("DatasetToTFRecord", scope.makeOpName("DatasetToTfRecord"));
-    opBuilder.addInput(inputDataset.asOutput());
-    opBuilder.addInput(filename.asOutput());
-    opBuilder.addInput(compressionType.asOutput());
+    opBuilder.addInput(inputDataset.asOutput(scope));
+    opBuilder.addInput(filename.asOutput(scope));
+    opBuilder.addInput(compressionType.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new DatasetToTfRecord(opBuilder.build());
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DeleteIterator.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DeleteIterator.java
index 69f3af096bb..384f7f8369c 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DeleteIterator.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DeleteIterator.java
@@ -42,8 +42,8 @@ public final class DeleteIterator extends RawOp {
   @Endpoint(describeByClass = true)
   public static DeleteIterator create(Scope scope, Operand<?> handle, Operand<?> deleter) {
     OperationBuilder opBuilder = scope.env().opBuilder("DeleteIterator", scope.makeOpName("DeleteIterator"));
-    opBuilder.addInput(handle.asOutput());
-    opBuilder.addInput(deleter.asOutput());
+    opBuilder.addInput(handle.asOutput(scope));
+    opBuilder.addInput(deleter.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new DeleteIterator(opBuilder.build());
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DeleteMemoryCache.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DeleteMemoryCache.java
index 21c33030b66..c83f6878dc2 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DeleteMemoryCache.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DeleteMemoryCache.java
@@ -40,8 +40,8 @@ public final class DeleteMemoryCache extends RawOp {
   @Endpoint(describeByClass = true)
   public static DeleteMemoryCache create(Scope scope, Operand<?> handle, Operand<?> deleter) {
     OperationBuilder opBuilder = scope.env().opBuilder("DeleteMemoryCache", scope.makeOpName("DeleteMemoryCache"));
-    opBuilder.addInput(handle.asOutput());
-    opBuilder.addInput(deleter.asOutput());
+    opBuilder.addInput(handle.asOutput(scope));
+    opBuilder.addInput(deleter.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new DeleteMemoryCache(opBuilder.build());
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DeleteMultiDeviceIterator.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DeleteMultiDeviceIterator.java
index 966d6a7dbf1..de98ff6a78e 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DeleteMultiDeviceIterator.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DeleteMultiDeviceIterator.java
@@ -43,9 +43,9 @@ public final class DeleteMultiDeviceIterator extends RawOp {
   @Endpoint(describeByClass = true)
   public static DeleteMultiDeviceIterator create(Scope scope, Operand<?> multiDeviceIterator, Iterable<Operand<?>> iterators, Operand<?> deleter) {
     OperationBuilder opBuilder = scope.env().opBuilder("DeleteMultiDeviceIterator", scope.makeOpName("DeleteMultiDeviceIterator"));
-    opBuilder.addInput(multiDeviceIterator.asOutput());
-    opBuilder.addInputList(Operands.asOutputs(iterators));
-    opBuilder.addInput(deleter.asOutput());
+    opBuilder.addInput(multiDeviceIterator.asOutput(scope));
+    opBuilder.addInputList(Operands.asOutputs(scope, iterators));
+    opBuilder.addInput(deleter.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new DeleteMultiDeviceIterator(opBuilder.build());
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DenseToSparseBatchDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DenseToSparseBatchDataset.java
index ccc03fe1983..2f89979fa98 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DenseToSparseBatchDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DenseToSparseBatchDataset.java
@@ -23,18 +23,18 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
 
 /**
  * Creates a dataset that batches input elements into a SparseTensor.
  */
-public final class DenseToSparseBatchDataset extends RawOp implements Operand<Tensor> {
+public final class DenseToSparseBatchDataset extends RawOp implements Operand<TType> {
   
   /**
    * Factory method to create a class wrapping a new DenseToSparseBatchDataset operation.
@@ -53,9 +53,9 @@ public final class DenseToSparseBatchDataset extends RawOp implements Operand<Te
   @Endpoint(describeByClass = true)
   public static DenseToSparseBatchDataset create(Scope scope, Operand<?> inputDataset, Operand<TInt64> batchSize, Operand<TInt64> rowShape, List<DataType<?>> outputTypes, List<Shape> outputShapes) {
     OperationBuilder opBuilder = scope.env().opBuilder("DenseToSparseBatchDataset", scope.makeOpName("DenseToSparseBatchDataset"));
-    opBuilder.addInput(inputDataset.asOutput());
-    opBuilder.addInput(batchSize.asOutput());
-    opBuilder.addInput(rowShape.asOutput());
+    opBuilder.addInput(inputDataset.asOutput(scope));
+    opBuilder.addInput(batchSize.asOutput(scope));
+    opBuilder.addInput(rowShape.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     DataType[] outputTypesArray = new DataType[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
@@ -78,8 +78,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<Tensor> asOutput() {
-    return (Output<Tensor>) handle;
+  public Output<TType> asOutput(Scope scope) {
+    return (Output<TType>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DeserializeIterator.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DeserializeIterator.java
index 4f772fd5028..fb7873f6f42 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DeserializeIterator.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DeserializeIterator.java
@@ -43,8 +43,8 @@ public final class DeserializeIterator extends RawOp {
   @Endpoint(describeByClass = true)
   public static DeserializeIterator create(Scope scope, Operand<?> resourceHandle, Operand<?> serialized) {
     OperationBuilder opBuilder = scope.env().opBuilder("DeserializeIterator", scope.makeOpName("DeserializeIterator"));
-    opBuilder.addInput(resourceHandle.asOutput());
-    opBuilder.addInput(serialized.asOutput());
+    opBuilder.addInput(resourceHandle.asOutput(scope));
+    opBuilder.addInput(serialized.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new DeserializeIterator(opBuilder.build());
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DirectedInterleaveDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DirectedInterleaveDataset.java
index 4218e1919f5..9fff9e7e6e7 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DirectedInterleaveDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DirectedInterleaveDataset.java
@@ -23,18 +23,18 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.Operands;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * A substitute for `InterleaveDataset` on a fixed list of `N` datasets.
  */
-public final class DirectedInterleaveDataset extends RawOp implements Operand<Tensor> {
+public final class DirectedInterleaveDataset extends RawOp implements Operand<TType> {
   
   /**
    * Factory method to create a class wrapping a new DirectedInterleaveDataset operation.
@@ -51,8 +51,8 @@ public final class DirectedInterleaveDataset extends RawOp implements Operand<Te
   @Endpoint(describeByClass = true)
   public static DirectedInterleaveDataset create(Scope scope, Operand<?> selectorInputDataset, Iterable<Operand<?>> dataInputDatasets, List<DataType<?>> outputTypes, List<Shape> outputShapes) {
     OperationBuilder opBuilder = scope.env().opBuilder("DirectedInterleaveDataset", scope.makeOpName("DirectedInterleaveDataset"));
-    opBuilder.addInput(selectorInputDataset.asOutput());
-    opBuilder.addInputList(Operands.asOutputs(dataInputDatasets));
+    opBuilder.addInput(selectorInputDataset.asOutput(scope));
+    opBuilder.addInputList(Operands.asOutputs(scope, dataInputDatasets));
     opBuilder = scope.applyControlDependencies(opBuilder);
     DataType[] outputTypesArray = new DataType[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
@@ -75,8 +75,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<Tensor> asOutput() {
-    return (Output<Tensor>) handle;
+  public Output<TType> asOutput(Scope scope) {
+    return (Output<TType>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/FilterByLastComponentDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/FilterByLastComponentDataset.java
index 5e193d710a3..7d1421f46da 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/FilterByLastComponentDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/FilterByLastComponentDataset.java
@@ -23,17 +23,17 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Creates a dataset containing elements of first component of `input_dataset` having true in the last component.
  */
-public final class FilterByLastComponentDataset extends RawOp implements Operand<Tensor> {
+public final class FilterByLastComponentDataset extends RawOp implements Operand<TType> {
   
   /**
    * Factory method to create a class wrapping a new FilterByLastComponentDataset operation.
@@ -47,7 +47,7 @@ public final class FilterByLastComponentDataset extends RawOp implements Operand
   @Endpoint(describeByClass = true)
   public static FilterByLastComponentDataset create(Scope scope, Operand<?> inputDataset, List<DataType<?>> outputTypes, List<Shape> outputShapes) {
     OperationBuilder opBuilder = scope.env().opBuilder("FilterByLastComponentDataset", scope.makeOpName("FilterByLastComponentDataset"));
-    opBuilder.addInput(inputDataset.asOutput());
+    opBuilder.addInput(inputDataset.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     DataType[] outputTypesArray = new DataType[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
@@ -70,8 +70,8 @@ public Output<?> output() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<Tensor> asOutput() {
-    return (Output<Tensor>) output;
+  public Output<TType> asOutput(Scope scope) {
+    return (Output<TType>) output;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/FixedLengthRecordDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/FixedLengthRecordDataset.java
index 105e1b6d3e4..9dc927efb6f 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/FixedLengthRecordDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/FixedLengthRecordDataset.java
@@ -21,17 +21,17 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
 import org.tensorflow.types.TString;
+import org.tensorflow.types.family.TType;
 
 /**
  */
-public final class FixedLengthRecordDataset extends RawOp implements Operand<Tensor> {
+public final class FixedLengthRecordDataset extends RawOp implements Operand<TType> {
   
   /**
    * Factory method to create a class wrapping a new FixedLengthRecordDataset operation.
@@ -48,12 +48,12 @@ public final class FixedLengthRecordDataset extends RawOp implements Operand<Ten
   @Endpoint(describeByClass = true)
   public static FixedLengthRecordDataset create(Scope scope, Operand<TString> filenames, Operand<TInt64> headerBytes, Operand<TInt64> recordBytes, Operand<TInt64> footerBytes, Operand<TInt64> bufferSize, Operand<TString> compressionType) {
     OperationBuilder opBuilder = scope.env().opBuilder("FixedLengthRecordDatasetV2", scope.makeOpName("FixedLengthRecordDataset"));
-    opBuilder.addInput(filenames.asOutput());
-    opBuilder.addInput(headerBytes.asOutput());
-    opBuilder.addInput(recordBytes.asOutput());
-    opBuilder.addInput(footerBytes.asOutput());
-    opBuilder.addInput(bufferSize.asOutput());
-    opBuilder.addInput(compressionType.asOutput());
+    opBuilder.addInput(filenames.asOutput(scope));
+    opBuilder.addInput(headerBytes.asOutput(scope));
+    opBuilder.addInput(recordBytes.asOutput(scope));
+    opBuilder.addInput(footerBytes.asOutput(scope));
+    opBuilder.addInput(bufferSize.asOutput(scope));
+    opBuilder.addInput(compressionType.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new FixedLengthRecordDataset(opBuilder.build());
   }
@@ -66,8 +66,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<Tensor> asOutput() {
-    return (Output<Tensor>) handle;
+  public Output<TType> asOutput(Scope scope) {
+    return (Output<TType>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/IgnoreErrorsDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/IgnoreErrorsDataset.java
index e6c6079b1a6..b58693e2031 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/IgnoreErrorsDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/IgnoreErrorsDataset.java
@@ -23,17 +23,17 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Creates a dataset that contains the elements of `input_dataset` ignoring errors.
  */
-public final class IgnoreErrorsDataset extends RawOp implements Operand<Tensor> {
+public final class IgnoreErrorsDataset extends RawOp implements Operand<TType> {
   
   /**
    * Factory method to create a class wrapping a new IgnoreErrorsDataset operation.
@@ -47,7 +47,7 @@ public final class IgnoreErrorsDataset extends RawOp implements Operand<Tensor>
   @Endpoint(describeByClass = true)
   public static IgnoreErrorsDataset create(Scope scope, Operand<?> inputDataset, List<DataType<?>> outputTypes, List<Shape> outputShapes) {
     OperationBuilder opBuilder = scope.env().opBuilder("IgnoreErrorsDataset", scope.makeOpName("IgnoreErrorsDataset"));
-    opBuilder.addInput(inputDataset.asOutput());
+    opBuilder.addInput(inputDataset.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     DataType[] outputTypesArray = new DataType[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
@@ -70,8 +70,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<Tensor> asOutput() {
-    return (Output<Tensor>) handle;
+  public Output<TType> asOutput(Scope scope) {
+    return (Output<TType>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/Iterator.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/Iterator.java
index 2136bfd13e0..d3c03cd40b0 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/Iterator.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/Iterator.java
@@ -23,17 +23,17 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  */
 @Operator(group = "data")
-public final class Iterator extends RawOp implements Operand<Tensor> {
+public final class Iterator extends RawOp implements Operand<TType> {
   
   /**
    * Factory method to create a class wrapping a new Iterator operation.
@@ -72,8 +72,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<Tensor> asOutput() {
-    return (Output<Tensor>) handle;
+  public Output<TType> asOutput(Scope scope) {
+    return (Output<TType>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/IteratorFromStringHandle.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/IteratorFromStringHandle.java
index 33b1143fa9e..c06bcd607f1 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/IteratorFromStringHandle.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/IteratorFromStringHandle.java
@@ -23,17 +23,17 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TString;
+import org.tensorflow.types.family.TType;
 
 /**
  */
-public final class IteratorFromStringHandle extends RawOp implements Operand<Tensor> {
+public final class IteratorFromStringHandle extends RawOp implements Operand<TType> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.data.IteratorFromStringHandle}
@@ -66,7 +66,7 @@ private Options() {
   @Endpoint(describeByClass = true)
   public static IteratorFromStringHandle create(Scope scope, Operand<TString> stringHandle, List<DataType<?>> outputTypes, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("IteratorFromStringHandleV2", scope.makeOpName("IteratorFromStringHandle"));
-    opBuilder.addInput(stringHandle.asOutput());
+    opBuilder.addInput(stringHandle.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     DataType[] outputTypesArray = new DataType[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
@@ -102,8 +102,8 @@ public Output<?> resourceHandle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<Tensor> asOutput() {
-    return (Output<Tensor>) resourceHandle;
+  public Output<TType> asOutput(Scope scope) {
+    return (Output<TType>) resourceHandle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/IteratorGetDevice.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/IteratorGetDevice.java
index d2aee159583..d02fadac0f7 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/IteratorGetDevice.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/IteratorGetDevice.java
@@ -42,7 +42,7 @@ public final class IteratorGetDevice extends RawOp implements Operand<TString> {
   @Endpoint(describeByClass = true)
   public static IteratorGetDevice create(Scope scope, Operand<?> resource) {
     OperationBuilder opBuilder = scope.env().opBuilder("IteratorGetDevice", scope.makeOpName("IteratorGetDevice"));
-    opBuilder.addInput(resource.asOutput());
+    opBuilder.addInput(resource.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new IteratorGetDevice(opBuilder.build());
   }
@@ -54,7 +54,7 @@ public Output<TString> device() {
   }
   
   @Override
-  public Output<TString> asOutput() {
+  public Output<TString> asOutput(Scope scope) {
     return device;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/IteratorGetNext.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/IteratorGetNext.java
index 56efe443f96..97c55d78184 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/IteratorGetNext.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/IteratorGetNext.java
@@ -25,18 +25,18 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Gets the next output from the given iterator .
  */
 @Operator(group = "data")
-public final class IteratorGetNext extends RawOp implements Iterable<Operand<Tensor>> {
+public final class IteratorGetNext extends RawOp implements Iterable<Operand<TType>> {
   
   /**
    * Factory method to create a class wrapping a new IteratorGetNext operation.
@@ -50,7 +50,7 @@ public final class IteratorGetNext extends RawOp implements Iterable<Operand<Ten
   @Endpoint(describeByClass = true)
   public static IteratorGetNext create(Scope scope, Operand<?> iterator, List<DataType<?>> outputTypes, List<Shape> outputShapes) {
     OperationBuilder opBuilder = scope.env().opBuilder("IteratorGetNext", scope.makeOpName("IteratorGetNext"));
-    opBuilder.addInput(iterator.asOutput());
+    opBuilder.addInput(iterator.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     DataType[] outputTypesArray = new DataType[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
@@ -73,7 +73,7 @@ public List<Output<?>> components() {
   
   @Override
   @SuppressWarnings({"rawtypes", "unchecked"})
-  public Iterator<Operand<Tensor>> iterator() {
+  public Iterator<Operand<TType>> iterator() {
     return (Iterator) components.iterator();
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/IteratorGetNextAsOptional.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/IteratorGetNextAsOptional.java
index bd4deb65680..1243a53e435 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/IteratorGetNextAsOptional.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/IteratorGetNextAsOptional.java
@@ -23,18 +23,18 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Gets the next output from the given iterator as an Optional variant.
  */
 @Operator(group = "data")
-public final class IteratorGetNextAsOptional extends RawOp implements Operand<Tensor> {
+public final class IteratorGetNextAsOptional extends RawOp implements Operand<TType> {
   
   /**
    * Factory method to create a class wrapping a new IteratorGetNextAsOptional operation.
@@ -48,7 +48,7 @@ public final class IteratorGetNextAsOptional extends RawOp implements Operand<Te
   @Endpoint(describeByClass = true)
   public static IteratorGetNextAsOptional create(Scope scope, Operand<?> iterator, List<DataType<?>> outputTypes, List<Shape> outputShapes) {
     OperationBuilder opBuilder = scope.env().opBuilder("IteratorGetNextAsOptional", scope.makeOpName("IteratorGetNextAsOptional"));
-    opBuilder.addInput(iterator.asOutput());
+    opBuilder.addInput(iterator.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     DataType[] outputTypesArray = new DataType[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
@@ -71,8 +71,8 @@ public Output<?> optional() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<Tensor> asOutput() {
-    return (Output<Tensor>) optional;
+  public Output<TType> asOutput(Scope scope) {
+    return (Output<TType>) optional;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/IteratorGetNextSync.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/IteratorGetNextSync.java
index e718c779928..84670f3fc5b 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/IteratorGetNextSync.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/IteratorGetNextSync.java
@@ -25,12 +25,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Gets the next output from the given iterator.
@@ -41,7 +41,7 @@
  * operations (e.g. in eager mode).
  */
 @Operator(group = "data")
-public final class IteratorGetNextSync extends RawOp implements Iterable<Operand<Tensor>> {
+public final class IteratorGetNextSync extends RawOp implements Iterable<Operand<TType>> {
   
   /**
    * Factory method to create a class wrapping a new IteratorGetNextSync operation.
@@ -55,7 +55,7 @@ public final class IteratorGetNextSync extends RawOp implements Iterable<Operand
   @Endpoint(describeByClass = true)
   public static IteratorGetNextSync create(Scope scope, Operand<?> iterator, List<DataType<?>> outputTypes, List<Shape> outputShapes) {
     OperationBuilder opBuilder = scope.env().opBuilder("IteratorGetNextSync", scope.makeOpName("IteratorGetNextSync"));
-    opBuilder.addInput(iterator.asOutput());
+    opBuilder.addInput(iterator.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     DataType[] outputTypesArray = new DataType[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
@@ -78,7 +78,7 @@ public List<Output<?>> components() {
   
   @Override
   @SuppressWarnings({"rawtypes", "unchecked"})
-  public Iterator<Operand<Tensor>> iterator() {
+  public Iterator<Operand<TType>> iterator() {
     return (Iterator) components.iterator();
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/IteratorToStringHandle.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/IteratorToStringHandle.java
index d715d8b8f2d..b868405a984 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/IteratorToStringHandle.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/IteratorToStringHandle.java
@@ -43,7 +43,7 @@ public final class IteratorToStringHandle extends RawOp implements Operand<TStri
   @Endpoint(describeByClass = true)
   public static IteratorToStringHandle create(Scope scope, Operand<?> resourceHandle) {
     OperationBuilder opBuilder = scope.env().opBuilder("IteratorToStringHandle", scope.makeOpName("IteratorToStringHandle"));
-    opBuilder.addInput(resourceHandle.asOutput());
+    opBuilder.addInput(resourceHandle.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new IteratorToStringHandle(opBuilder.build());
   }
@@ -56,7 +56,7 @@ public Output<TString> stringHandle() {
   }
   
   @Override
-  public Output<TString> asOutput() {
+  public Output<TString> asOutput(Scope scope) {
     return stringHandle;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/LMDBDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/LMDBDataset.java
index fc17d2406ef..208785fad35 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/LMDBDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/LMDBDataset.java
@@ -23,13 +23,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TString;
+import org.tensorflow.types.family.TType;
 
 /**
  * Creates a dataset that emits the key-value pairs in one or more LMDB files.
@@ -45,7 +45,7 @@
  * LMDB uses different file formats on big- and little-endian machines.
  * `data.LMDBDataset` can only read files in the format of the host machine.
  */
-public final class LMDBDataset extends RawOp implements Operand<Tensor> {
+public final class LMDBDataset extends RawOp implements Operand<TType> {
   
   /**
    * Factory method to create a class wrapping a new LMDBDataset operation.
@@ -60,7 +60,7 @@ public final class LMDBDataset extends RawOp implements Operand<Tensor> {
   @Endpoint(describeByClass = true)
   public static LMDBDataset create(Scope scope, Operand<TString> filenames, List<DataType<?>> outputTypes, List<Shape> outputShapes) {
     OperationBuilder opBuilder = scope.env().opBuilder("LMDBDataset", scope.makeOpName("LMDBDataset"));
-    opBuilder.addInput(filenames.asOutput());
+    opBuilder.addInput(filenames.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     DataType[] outputTypesArray = new DataType[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
@@ -83,8 +83,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<Tensor> asOutput() {
-    return (Output<Tensor>) handle;
+  public Output<TType> asOutput(Scope scope) {
+    return (Output<TType>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/LatencyStatsDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/LatencyStatsDataset.java
index 0ed5f19a738..1a4af719e32 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/LatencyStatsDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/LatencyStatsDataset.java
@@ -23,18 +23,18 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TString;
+import org.tensorflow.types.family.TType;
 
 /**
  * Records the latency of producing `input_dataset` elements in a StatsAggregator.
  */
-public final class LatencyStatsDataset extends RawOp implements Operand<Tensor> {
+public final class LatencyStatsDataset extends RawOp implements Operand<TType> {
   
   /**
    * Factory method to create a class wrapping a new LatencyStatsDataset operation.
@@ -49,8 +49,8 @@ public final class LatencyStatsDataset extends RawOp implements Operand<Tensor>
   @Endpoint(describeByClass = true)
   public static LatencyStatsDataset create(Scope scope, Operand<?> inputDataset, Operand<TString> tag, List<DataType<?>> outputTypes, List<Shape> outputShapes) {
     OperationBuilder opBuilder = scope.env().opBuilder("LatencyStatsDataset", scope.makeOpName("LatencyStatsDataset"));
-    opBuilder.addInput(inputDataset.asOutput());
-    opBuilder.addInput(tag.asOutput());
+    opBuilder.addInput(inputDataset.asOutput(scope));
+    opBuilder.addInput(tag.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     DataType[] outputTypesArray = new DataType[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
@@ -73,8 +73,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<Tensor> asOutput() {
-    return (Output<Tensor>) handle;
+  public Output<TType> asOutput(Scope scope) {
+    return (Output<TType>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/LeakyReluGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/LeakyReluGrad.java
index ef7f5d3365f..46f9410dc26 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/LeakyReluGrad.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/LeakyReluGrad.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -33,7 +32,7 @@
  * 
  * @param <T> data type for {@code backprops()} output
  */
-public final class LeakyReluGrad<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class LeakyReluGrad<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.data.LeakyReluGrad}
@@ -65,10 +64,10 @@ private Options() {
    * @return a new instance of LeakyReluGrad
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> LeakyReluGrad<T> create(Scope scope, Operand<T> gradients, Operand<T> features, Options... options) {
+  public static <T extends TNumber> LeakyReluGrad<T> create(Scope scope, Operand<T> gradients, Operand<T> features, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("LeakyReluGrad", scope.makeOpName("LeakyReluGrad"));
-    opBuilder.addInput(gradients.asOutput());
-    opBuilder.addInput(features.asOutput());
+    opBuilder.addInput(gradients.asOutput(scope));
+    opBuilder.addInput(features.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -95,7 +94,7 @@ public Output<T> backprops() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return backprops;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MakeIterator.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MakeIterator.java
index 4aace25184e..663d2f9e869 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MakeIterator.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MakeIterator.java
@@ -45,8 +45,8 @@ public final class MakeIterator extends RawOp {
   @Endpoint(describeByClass = true)
   public static MakeIterator create(Scope scope, Operand<?> dataset, Operand<?> iterator) {
     OperationBuilder opBuilder = scope.env().opBuilder("MakeIterator", scope.makeOpName("MakeIterator"));
-    opBuilder.addInput(dataset.asOutput());
-    opBuilder.addInput(iterator.asOutput());
+    opBuilder.addInput(dataset.asOutput(scope));
+    opBuilder.addInput(iterator.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new MakeIterator(opBuilder.build());
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MatchingFilesDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MatchingFilesDataset.java
index fc88bc9b785..884c14f3ef0 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MatchingFilesDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MatchingFilesDataset.java
@@ -21,16 +21,16 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TString;
+import org.tensorflow.types.family.TType;
 
 /**
  */
-public final class MatchingFilesDataset extends RawOp implements Operand<Tensor> {
+public final class MatchingFilesDataset extends RawOp implements Operand<TType> {
   
   /**
    * Factory method to create a class wrapping a new MatchingFilesDataset operation.
@@ -42,7 +42,7 @@ public final class MatchingFilesDataset extends RawOp implements Operand<Tensor>
   @Endpoint(describeByClass = true)
   public static MatchingFilesDataset create(Scope scope, Operand<TString> patterns) {
     OperationBuilder opBuilder = scope.env().opBuilder("MatchingFilesDataset", scope.makeOpName("MatchingFilesDataset"));
-    opBuilder.addInput(patterns.asOutput());
+    opBuilder.addInput(patterns.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new MatchingFilesDataset(opBuilder.build());
   }
@@ -55,8 +55,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<Tensor> asOutput() {
-    return (Output<Tensor>) handle;
+  public Output<TType> asOutput(Scope scope) {
+    return (Output<TType>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MaxIntraOpParallelismDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MaxIntraOpParallelismDataset.java
index 423c401478a..4d365c94ed3 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MaxIntraOpParallelismDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MaxIntraOpParallelismDataset.java
@@ -23,18 +23,18 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
 
 /**
  * Creates a dataset that overrides the maximum intra-op parallelism.
  */
-public final class MaxIntraOpParallelismDataset extends RawOp implements Operand<Tensor> {
+public final class MaxIntraOpParallelismDataset extends RawOp implements Operand<TType> {
   
   /**
    * Factory method to create a class wrapping a new MaxIntraOpParallelismDataset operation.
@@ -49,8 +49,8 @@ public final class MaxIntraOpParallelismDataset extends RawOp implements Operand
   @Endpoint(describeByClass = true)
   public static MaxIntraOpParallelismDataset create(Scope scope, Operand<?> inputDataset, Operand<TInt64> maxIntraOpParallelism, List<DataType<?>> outputTypes, List<Shape> outputShapes) {
     OperationBuilder opBuilder = scope.env().opBuilder("MaxIntraOpParallelismDataset", scope.makeOpName("MaxIntraOpParallelismDataset"));
-    opBuilder.addInput(inputDataset.asOutput());
-    opBuilder.addInput(maxIntraOpParallelism.asOutput());
+    opBuilder.addInput(inputDataset.asOutput(scope));
+    opBuilder.addInput(maxIntraOpParallelism.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     DataType[] outputTypesArray = new DataType[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
@@ -73,8 +73,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<Tensor> asOutput() {
-    return (Output<Tensor>) handle;
+  public Output<TType> asOutput(Scope scope) {
+    return (Output<TType>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ModelDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ModelDataset.java
index 6f74e7e300c..9fed6fcaa30 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ModelDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ModelDataset.java
@@ -23,19 +23,19 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Identity transformation that models performance.
  * <p>
  * Identity transformation that models performance.
  */
-public final class ModelDataset extends RawOp implements Operand<Tensor> {
+public final class ModelDataset extends RawOp implements Operand<TType> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.data.ModelDataset}
@@ -78,7 +78,7 @@ private Options() {
   @Endpoint(describeByClass = true)
   public static ModelDataset create(Scope scope, Operand<?> inputDataset, List<DataType<?>> outputTypes, List<Shape> outputShapes, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ModelDataset", scope.makeOpName("ModelDataset"));
-    opBuilder.addInput(inputDataset.asOutput());
+    opBuilder.addInput(inputDataset.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     DataType[] outputTypesArray = new DataType[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
@@ -125,8 +125,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<Tensor> asOutput() {
-    return (Output<Tensor>) handle;
+  public Output<TType> asOutput(Scope scope) {
+    return (Output<TType>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MultiDeviceIterator.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MultiDeviceIterator.java
index e8d7a99e312..1ac9874ab6a 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MultiDeviceIterator.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MultiDeviceIterator.java
@@ -23,17 +23,17 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Creates a MultiDeviceIterator resource.
  */
-public final class MultiDeviceIterator extends RawOp implements Operand<Tensor> {
+public final class MultiDeviceIterator extends RawOp implements Operand<TType> {
   
   /**
    * Factory method to create a class wrapping a new MultiDeviceIterator operation.
@@ -81,8 +81,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<Tensor> asOutput() {
-    return (Output<Tensor>) handle;
+  public Output<TType> asOutput(Scope scope) {
+    return (Output<TType>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MultiDeviceIteratorFromStringHandle.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MultiDeviceIteratorFromStringHandle.java
index d3846fe1d7d..61723160f2b 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MultiDeviceIteratorFromStringHandle.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MultiDeviceIteratorFromStringHandle.java
@@ -23,18 +23,18 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TString;
+import org.tensorflow.types.family.TType;
 
 /**
  * Generates a MultiDeviceIterator resource from its provided string handle.
  */
-public final class MultiDeviceIteratorFromStringHandle extends RawOp implements Operand<Tensor> {
+public final class MultiDeviceIteratorFromStringHandle extends RawOp implements Operand<TType> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.data.MultiDeviceIteratorFromStringHandle}
@@ -67,7 +67,7 @@ private Options() {
   @Endpoint(describeByClass = true)
   public static MultiDeviceIteratorFromStringHandle create(Scope scope, Operand<TString> stringHandle, List<DataType<?>> outputTypes, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("MultiDeviceIteratorFromStringHandle", scope.makeOpName("MultiDeviceIteratorFromStringHandle"));
-    opBuilder.addInput(stringHandle.asOutput());
+    opBuilder.addInput(stringHandle.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     DataType[] outputTypesArray = new DataType[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
@@ -104,8 +104,8 @@ public Output<?> multiDeviceIterator() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<Tensor> asOutput() {
-    return (Output<Tensor>) multiDeviceIterator;
+  public Output<TType> asOutput(Scope scope) {
+    return (Output<TType>) multiDeviceIterator;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MultiDeviceIteratorGetNextFromShard.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MultiDeviceIteratorGetNextFromShard.java
index c11cbdee0c1..d08b76a7305 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MultiDeviceIteratorGetNextFromShard.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MultiDeviceIteratorGetNextFromShard.java
@@ -25,7 +25,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
@@ -33,11 +32,12 @@
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
 import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
 
 /**
  * Gets next element for the provided shard number.
  */
-public final class MultiDeviceIteratorGetNextFromShard extends RawOp implements Iterable<Operand<Tensor>> {
+public final class MultiDeviceIteratorGetNextFromShard extends RawOp implements Iterable<Operand<TType>> {
   
   /**
    * Factory method to create a class wrapping a new MultiDeviceIteratorGetNextFromShard operation.
@@ -53,9 +53,9 @@ public final class MultiDeviceIteratorGetNextFromShard extends RawOp implements
   @Endpoint(describeByClass = true)
   public static MultiDeviceIteratorGetNextFromShard create(Scope scope, Operand<?> multiDeviceIterator, Operand<TInt32> shardNum, Operand<TInt64> incarnationId, List<DataType<?>> outputTypes, List<Shape> outputShapes) {
     OperationBuilder opBuilder = scope.env().opBuilder("MultiDeviceIteratorGetNextFromShard", scope.makeOpName("MultiDeviceIteratorGetNextFromShard"));
-    opBuilder.addInput(multiDeviceIterator.asOutput());
-    opBuilder.addInput(shardNum.asOutput());
-    opBuilder.addInput(incarnationId.asOutput());
+    opBuilder.addInput(multiDeviceIterator.asOutput(scope));
+    opBuilder.addInput(shardNum.asOutput(scope));
+    opBuilder.addInput(incarnationId.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     DataType[] outputTypesArray = new DataType[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
@@ -79,7 +79,7 @@ public List<Output<?>> components() {
   
   @Override
   @SuppressWarnings({"rawtypes", "unchecked"})
-  public Iterator<Operand<Tensor>> iterator() {
+  public Iterator<Operand<TType>> iterator() {
     return (Iterator) components.iterator();
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MultiDeviceIteratorInit.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MultiDeviceIteratorInit.java
index a8e957c64b5..0b1a70d6a8f 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MultiDeviceIteratorInit.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MultiDeviceIteratorInit.java
@@ -44,9 +44,9 @@ public final class MultiDeviceIteratorInit extends RawOp implements Operand<TInt
   @Endpoint(describeByClass = true)
   public static MultiDeviceIteratorInit create(Scope scope, Operand<?> dataset, Operand<?> multiDeviceIterator, Operand<TInt64> maxBufferSize) {
     OperationBuilder opBuilder = scope.env().opBuilder("MultiDeviceIteratorInit", scope.makeOpName("MultiDeviceIteratorInit"));
-    opBuilder.addInput(dataset.asOutput());
-    opBuilder.addInput(multiDeviceIterator.asOutput());
-    opBuilder.addInput(maxBufferSize.asOutput());
+    opBuilder.addInput(dataset.asOutput(scope));
+    opBuilder.addInput(multiDeviceIterator.asOutput(scope));
+    opBuilder.addInput(maxBufferSize.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new MultiDeviceIteratorInit(opBuilder.build());
   }
@@ -60,7 +60,7 @@ public Output<TInt64> incarnationId() {
   }
   
   @Override
-  public Output<TInt64> asOutput() {
+  public Output<TInt64> asOutput(Scope scope) {
     return incarnationId;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MultiDeviceIteratorToStringHandle.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MultiDeviceIteratorToStringHandle.java
index 5697c7e7699..3efeb3052e9 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MultiDeviceIteratorToStringHandle.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MultiDeviceIteratorToStringHandle.java
@@ -42,7 +42,7 @@ public final class MultiDeviceIteratorToStringHandle extends RawOp implements Op
   @Endpoint(describeByClass = true)
   public static MultiDeviceIteratorToStringHandle create(Scope scope, Operand<?> multiDeviceIterator) {
     OperationBuilder opBuilder = scope.env().opBuilder("MultiDeviceIteratorToStringHandle", scope.makeOpName("MultiDeviceIteratorToStringHandle"));
-    opBuilder.addInput(multiDeviceIterator.asOutput());
+    opBuilder.addInput(multiDeviceIterator.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new MultiDeviceIteratorToStringHandle(opBuilder.build());
   }
@@ -55,7 +55,7 @@ public Output<TString> stringHandle() {
   }
   
   @Override
-  public Output<TString> asOutput() {
+  public Output<TString> asOutput(Scope scope) {
     return stringHandle;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/NonSerializableDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/NonSerializableDataset.java
index 0f521de748e..fb0c21e115e 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/NonSerializableDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/NonSerializableDataset.java
@@ -23,16 +23,16 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  */
-public final class NonSerializableDataset extends RawOp implements Operand<Tensor> {
+public final class NonSerializableDataset extends RawOp implements Operand<TType> {
   
   /**
    * Factory method to create a class wrapping a new NonSerializableDataset operation.
@@ -46,7 +46,7 @@ public final class NonSerializableDataset extends RawOp implements Operand<Tenso
   @Endpoint(describeByClass = true)
   public static NonSerializableDataset create(Scope scope, Operand<?> inputDataset, List<DataType<?>> outputTypes, List<Shape> outputShapes) {
     OperationBuilder opBuilder = scope.env().opBuilder("NonSerializableDataset", scope.makeOpName("NonSerializableDataset"));
-    opBuilder.addInput(inputDataset.asOutput());
+    opBuilder.addInput(inputDataset.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     DataType[] outputTypesArray = new DataType[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
@@ -69,8 +69,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<Tensor> asOutput() {
-    return (Output<Tensor>) handle;
+  public Output<TType> asOutput(Scope scope) {
+    return (Output<TType>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/OptimizeDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/OptimizeDataset.java
index 4a3b5902752..d82141b019b 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/OptimizeDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/OptimizeDataset.java
@@ -23,20 +23,20 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TString;
+import org.tensorflow.types.family.TType;
 
 /**
  * Creates a dataset by applying optimizations to `input_dataset`.
  * <p>
  * Creates a dataset by applying optimizations to `input_dataset`.
  */
-public final class OptimizeDataset extends RawOp implements Operand<Tensor> {
+public final class OptimizeDataset extends RawOp implements Operand<TType> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.data.OptimizeDataset}
@@ -71,8 +71,8 @@ private Options() {
   @Endpoint(describeByClass = true)
   public static OptimizeDataset create(Scope scope, Operand<?> inputDataset, Operand<TString> optimizations, List<DataType<?>> outputTypes, List<Shape> outputShapes, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("OptimizeDataset", scope.makeOpName("OptimizeDataset"));
-    opBuilder.addInput(inputDataset.asOutput());
-    opBuilder.addInput(optimizations.asOutput());
+    opBuilder.addInput(inputDataset.asOutput(scope));
+    opBuilder.addInput(optimizations.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     DataType[] outputTypesArray = new DataType[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
@@ -113,8 +113,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<Tensor> asOutput() {
-    return (Output<Tensor>) handle;
+  public Output<TType> asOutput(Scope scope) {
+    return (Output<TType>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/OptionalFromValue.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/OptionalFromValue.java
index 5f69afca477..df33632dc0b 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/OptionalFromValue.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/OptionalFromValue.java
@@ -21,18 +21,18 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.Operands;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Constructs an Optional variant from a tuple of tensors.
  */
 @Operator(group = "data")
-public final class OptionalFromValue extends RawOp implements Operand<Tensor> {
+public final class OptionalFromValue extends RawOp implements Operand<TType> {
   
   /**
    * Factory method to create a class wrapping a new OptionalFromValue operation.
@@ -44,7 +44,7 @@ public final class OptionalFromValue extends RawOp implements Operand<Tensor> {
   @Endpoint(describeByClass = true)
   public static OptionalFromValue create(Scope scope, Iterable<Operand<?>> components) {
     OperationBuilder opBuilder = scope.env().opBuilder("OptionalFromValue", scope.makeOpName("OptionalFromValue"));
-    opBuilder.addInputList(Operands.asOutputs(components));
+    opBuilder.addInputList(Operands.asOutputs(scope, components));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new OptionalFromValue(opBuilder.build());
   }
@@ -57,8 +57,8 @@ public Output<?> optional() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<Tensor> asOutput() {
-    return (Output<Tensor>) optional;
+  public Output<TType> asOutput(Scope scope) {
+    return (Output<TType>) optional;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/OptionalGetValue.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/OptionalGetValue.java
index af5e5ea0a70..a6ec01b632c 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/OptionalGetValue.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/OptionalGetValue.java
@@ -25,18 +25,18 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Returns the value stored in an Optional variant or raises an error if none exists.
  */
 @Operator(group = "data")
-public final class OptionalGetValue extends RawOp implements Iterable<Operand<Tensor>> {
+public final class OptionalGetValue extends RawOp implements Iterable<Operand<TType>> {
   
   /**
    * Factory method to create a class wrapping a new OptionalGetValue operation.
@@ -50,7 +50,7 @@ public final class OptionalGetValue extends RawOp implements Iterable<Operand<Te
   @Endpoint(describeByClass = true)
   public static OptionalGetValue create(Scope scope, Operand<?> optional, List<DataType<?>> outputTypes, List<Shape> outputShapes) {
     OperationBuilder opBuilder = scope.env().opBuilder("OptionalGetValue", scope.makeOpName("OptionalGetValue"));
-    opBuilder.addInput(optional.asOutput());
+    opBuilder.addInput(optional.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     DataType[] outputTypesArray = new DataType[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
@@ -73,7 +73,7 @@ public List<Output<?>> components() {
   
   @Override
   @SuppressWarnings({"rawtypes", "unchecked"})
-  public Iterator<Operand<Tensor>> iterator() {
+  public Iterator<Operand<TType>> iterator() {
     return (Iterator) components.iterator();
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/OptionalHasValue.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/OptionalHasValue.java
index af5900c38a9..d0fb8dabaef 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/OptionalHasValue.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/OptionalHasValue.java
@@ -43,7 +43,7 @@ public final class OptionalHasValue extends RawOp implements Operand<TBool> {
   @Endpoint(describeByClass = true)
   public static OptionalHasValue create(Scope scope, Operand<?> optional) {
     OperationBuilder opBuilder = scope.env().opBuilder("OptionalHasValue", scope.makeOpName("OptionalHasValue"));
-    opBuilder.addInput(optional.asOutput());
+    opBuilder.addInput(optional.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new OptionalHasValue(opBuilder.build());
   }
@@ -55,7 +55,7 @@ public Output<TBool> hasValue() {
   }
   
   @Override
-  public Output<TBool> asOutput() {
+  public Output<TBool> asOutput(Scope scope) {
     return hasValue;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/OptionalNone.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/OptionalNone.java
index 5652d9c0daf..be0d557c02e 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/OptionalNone.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/OptionalNone.java
@@ -21,17 +21,17 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Creates an Optional variant with no value.
  */
 @Operator(group = "data")
-public final class OptionalNone extends RawOp implements Operand<Tensor> {
+public final class OptionalNone extends RawOp implements Operand<TType> {
   
   /**
    * Factory method to create a class wrapping a new OptionalNone operation.
@@ -54,8 +54,8 @@ public Output<?> optional() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<Tensor> asOutput() {
-    return (Output<Tensor>) optional;
+  public Output<TType> asOutput(Scope scope) {
+    return (Output<TType>) optional;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/PaddedBatchDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/PaddedBatchDataset.java
index 4a7e5f06f6a..97e3f9b5808 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/PaddedBatchDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/PaddedBatchDataset.java
@@ -22,7 +22,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.Operands;
 import org.tensorflow.op.RawOp;
@@ -31,11 +30,12 @@
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TBool;
 import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
 
 /**
  * Creates a dataset that batches and pads `batch_size` elements from the input.
  */
-public final class PaddedBatchDataset extends RawOp implements Operand<Tensor> {
+public final class PaddedBatchDataset extends RawOp implements Operand<TType> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.data.PaddedBatchDataset}
@@ -78,11 +78,11 @@ private Options() {
   @Endpoint(describeByClass = true)
   public static PaddedBatchDataset create(Scope scope, Operand<?> inputDataset, Operand<TInt64> batchSize, Iterable<Operand<TInt64>> paddedShapes, Iterable<Operand<?>> paddingValues, Operand<TBool> dropRemainder, List<Shape> outputShapes, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("PaddedBatchDatasetV2", scope.makeOpName("PaddedBatchDataset"));
-    opBuilder.addInput(inputDataset.asOutput());
-    opBuilder.addInput(batchSize.asOutput());
-    opBuilder.addInputList(Operands.asOutputs(paddedShapes));
-    opBuilder.addInputList(Operands.asOutputs(paddingValues));
-    opBuilder.addInput(dropRemainder.asOutput());
+    opBuilder.addInput(inputDataset.asOutput(scope));
+    opBuilder.addInput(batchSize.asOutput(scope));
+    opBuilder.addInputList(Operands.asOutputs(scope, paddedShapes));
+    opBuilder.addInputList(Operands.asOutputs(scope, paddingValues));
+    opBuilder.addInput(dropRemainder.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     Shape[] outputShapesArray = new Shape[outputShapes.size()];
     for (int i = 0; i < outputShapesArray.length; ++i) {
@@ -114,8 +114,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<Tensor> asOutput() {
-    return (Output<Tensor>) handle;
+  public Output<TType> asOutput(Scope scope) {
+    return (Output<TType>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/PrefetchDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/PrefetchDataset.java
index b0418e2722e..fd8a0cc8260 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/PrefetchDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/PrefetchDataset.java
@@ -23,18 +23,18 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
 
 /**
  * Creates a dataset that asynchronously prefetches elements from `input_dataset`.
  */
-public final class PrefetchDataset extends RawOp implements Operand<Tensor> {
+public final class PrefetchDataset extends RawOp implements Operand<TType> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.data.PrefetchDataset}
@@ -79,8 +79,8 @@ private Options() {
   @Endpoint(describeByClass = true)
   public static PrefetchDataset create(Scope scope, Operand<?> inputDataset, Operand<TInt64> bufferSize, List<DataType<?>> outputTypes, List<Shape> outputShapes, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("PrefetchDataset", scope.makeOpName("PrefetchDataset"));
-    opBuilder.addInput(inputDataset.asOutput());
-    opBuilder.addInput(bufferSize.asOutput());
+    opBuilder.addInput(inputDataset.asOutput(scope));
+    opBuilder.addInput(bufferSize.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     DataType[] outputTypesArray = new DataType[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
@@ -127,8 +127,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<Tensor> asOutput() {
-    return (Output<Tensor>) handle;
+  public Output<TType> asOutput(Scope scope) {
+    return (Output<TType>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/PrivateThreadPoolDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/PrivateThreadPoolDataset.java
index 5bc9fc4447e..30a9fce5626 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/PrivateThreadPoolDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/PrivateThreadPoolDataset.java
@@ -23,18 +23,18 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
 
 /**
  * Creates a dataset that uses a custom thread pool to compute `input_dataset`.
  */
-public final class PrivateThreadPoolDataset extends RawOp implements Operand<Tensor> {
+public final class PrivateThreadPoolDataset extends RawOp implements Operand<TType> {
   
   /**
    * Factory method to create a class wrapping a new PrivateThreadPoolDataset operation.
@@ -49,8 +49,8 @@ public final class PrivateThreadPoolDataset extends RawOp implements Operand<Ten
   @Endpoint(describeByClass = true)
   public static PrivateThreadPoolDataset create(Scope scope, Operand<?> inputDataset, Operand<TInt64> numThreads, List<DataType<?>> outputTypes, List<Shape> outputShapes) {
     OperationBuilder opBuilder = scope.env().opBuilder("PrivateThreadPoolDataset", scope.makeOpName("PrivateThreadPoolDataset"));
-    opBuilder.addInput(inputDataset.asOutput());
-    opBuilder.addInput(numThreads.asOutput());
+    opBuilder.addInput(inputDataset.asOutput(scope));
+    opBuilder.addInput(numThreads.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     DataType[] outputTypesArray = new DataType[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
@@ -73,8 +73,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<Tensor> asOutput() {
-    return (Output<Tensor>) handle;
+  public Output<TType> asOutput(Scope scope) {
+    return (Output<TType>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/RandomDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/RandomDataset.java
index b70f81436d0..749a517522e 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/RandomDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/RandomDataset.java
@@ -23,13 +23,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
 
 /**
  * Creates a Dataset that returns pseudorandom numbers.
@@ -45,7 +45,7 @@
  * performed is determined by the `experimental_optimization.hoist_random_uniform`
  * option of `tf.data.Options`.
  */
-public final class RandomDataset extends RawOp implements Operand<Tensor> {
+public final class RandomDataset extends RawOp implements Operand<TType> {
   
   /**
    * Factory method to create a class wrapping a new RandomDataset operation.
@@ -62,8 +62,8 @@ public final class RandomDataset extends RawOp implements Operand<Tensor> {
   @Endpoint(describeByClass = true)
   public static RandomDataset create(Scope scope, Operand<TInt64> seed, Operand<TInt64> seed2, List<DataType<?>> outputTypes, List<Shape> outputShapes) {
     OperationBuilder opBuilder = scope.env().opBuilder("RandomDataset", scope.makeOpName("RandomDataset"));
-    opBuilder.addInput(seed.asOutput());
-    opBuilder.addInput(seed2.asOutput());
+    opBuilder.addInput(seed.asOutput(scope));
+    opBuilder.addInput(seed2.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     DataType[] outputTypesArray = new DataType[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
@@ -86,8 +86,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<Tensor> asOutput() {
-    return (Output<Tensor>) handle;
+  public Output<TType> asOutput(Scope scope) {
+    return (Output<TType>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/RangeDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/RangeDataset.java
index 6db86ed9aaa..6c45680f1a5 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/RangeDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/RangeDataset.java
@@ -23,19 +23,19 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
 
 /**
  * Creates a dataset with a range of values. Corresponds to python's xrange.
  */
 @Operator(group = "data")
-public final class RangeDataset extends RawOp implements Operand<Tensor> {
+public final class RangeDataset extends RawOp implements Operand<TType> {
   
   /**
    * Factory method to create a class wrapping a new RangeDataset operation.
@@ -51,9 +51,9 @@ public final class RangeDataset extends RawOp implements Operand<Tensor> {
   @Endpoint(describeByClass = true)
   public static RangeDataset create(Scope scope, Operand<TInt64> start, Operand<TInt64> stop, Operand<TInt64> step, List<DataType<?>> outputTypes, List<Shape> outputShapes) {
     OperationBuilder opBuilder = scope.env().opBuilder("RangeDataset", scope.makeOpName("RangeDataset"));
-    opBuilder.addInput(start.asOutput());
-    opBuilder.addInput(stop.asOutput());
-    opBuilder.addInput(step.asOutput());
+    opBuilder.addInput(start.asOutput(scope));
+    opBuilder.addInput(stop.asOutput(scope));
+    opBuilder.addInput(step.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     DataType[] outputTypesArray = new DataType[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
@@ -76,8 +76,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<Tensor> asOutput() {
-    return (Output<Tensor>) handle;
+  public Output<TType> asOutput(Scope scope) {
+    return (Output<TType>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/RebatchDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/RebatchDataset.java
index a7ea93cc7a0..2b741f37f7f 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/RebatchDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/RebatchDataset.java
@@ -23,13 +23,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
 
 /**
  * Creates a dataset that changes the batch size.
@@ -37,7 +37,7 @@
  * Creates a dataset that changes the batch size of the dataset to current batch
  * size // num_workers.
  */
-public final class RebatchDataset extends RawOp implements Operand<Tensor> {
+public final class RebatchDataset extends RawOp implements Operand<TType> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.data.RebatchDataset}
@@ -74,8 +74,8 @@ private Options() {
   @Endpoint(describeByClass = true)
   public static RebatchDataset create(Scope scope, Operand<?> inputDataset, Operand<TInt64> numReplicas, List<DataType<?>> outputTypes, List<Shape> outputShapes, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("RebatchDataset", scope.makeOpName("RebatchDataset"));
-    opBuilder.addInput(inputDataset.asOutput());
-    opBuilder.addInput(numReplicas.asOutput());
+    opBuilder.addInput(inputDataset.asOutput(scope));
+    opBuilder.addInput(numReplicas.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     DataType[] outputTypesArray = new DataType[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
@@ -112,8 +112,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<Tensor> asOutput() {
-    return (Output<Tensor>) handle;
+  public Output<TType> asOutput(Scope scope) {
+    return (Output<TType>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/RepeatDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/RepeatDataset.java
index 33c501e2670..e4e432f2a1b 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/RepeatDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/RepeatDataset.java
@@ -23,19 +23,19 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
 
 /**
  * Creates a dataset that emits the outputs of `input_dataset` `count` times.
  */
 @Operator(group = "data")
-public final class RepeatDataset extends RawOp implements Operand<Tensor> {
+public final class RepeatDataset extends RawOp implements Operand<TType> {
   
   /**
    * Factory method to create a class wrapping a new RepeatDataset operation.
@@ -51,8 +51,8 @@ public final class RepeatDataset extends RawOp implements Operand<Tensor> {
   @Endpoint(describeByClass = true)
   public static RepeatDataset create(Scope scope, Operand<?> inputDataset, Operand<TInt64> count, List<DataType<?>> outputTypes, List<Shape> outputShapes) {
     OperationBuilder opBuilder = scope.env().opBuilder("RepeatDataset", scope.makeOpName("RepeatDataset"));
-    opBuilder.addInput(inputDataset.asOutput());
-    opBuilder.addInput(count.asOutput());
+    opBuilder.addInput(inputDataset.asOutput(scope));
+    opBuilder.addInput(count.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     DataType[] outputTypesArray = new DataType[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
@@ -75,8 +75,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<Tensor> asOutput() {
-    return (Output<Tensor>) handle;
+  public Output<TType> asOutput(Scope scope) {
+    return (Output<TType>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SamplingDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SamplingDataset.java
index 59b597eb32f..d535f6855b4 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SamplingDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SamplingDataset.java
@@ -23,7 +23,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
@@ -31,6 +30,7 @@
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TFloat32;
 import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
 
 /**
  * Creates a dataset that takes a Bernoulli sample of the contents of another dataset.
@@ -41,7 +41,7 @@
  * `experimental_optimization.filter_with_random_uniform_fusion` option of
  * `tf.data.Options`.
  */
-public final class SamplingDataset extends RawOp implements Operand<Tensor> {
+public final class SamplingDataset extends RawOp implements Operand<TType> {
   
   /**
    * Factory method to create a class wrapping a new SamplingDataset operation.
@@ -59,10 +59,10 @@ public final class SamplingDataset extends RawOp implements Operand<Tensor> {
   @Endpoint(describeByClass = true)
   public static SamplingDataset create(Scope scope, Operand<?> inputDataset, Operand<TFloat32> rate, Operand<TInt64> seed, Operand<TInt64> seed2, List<DataType<?>> outputTypes, List<Shape> outputShapes) {
     OperationBuilder opBuilder = scope.env().opBuilder("SamplingDataset", scope.makeOpName("SamplingDataset"));
-    opBuilder.addInput(inputDataset.asOutput());
-    opBuilder.addInput(rate.asOutput());
-    opBuilder.addInput(seed.asOutput());
-    opBuilder.addInput(seed2.asOutput());
+    opBuilder.addInput(inputDataset.asOutput(scope));
+    opBuilder.addInput(rate.asOutput(scope));
+    opBuilder.addInput(seed.asOutput(scope));
+    opBuilder.addInput(seed2.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     DataType[] outputTypesArray = new DataType[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
@@ -85,8 +85,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<Tensor> asOutput() {
-    return (Output<Tensor>) handle;
+  public Output<TType> asOutput(Scope scope) {
+    return (Output<TType>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SerializeIterator.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SerializeIterator.java
index 1bc43bad658..e4b8a08231e 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SerializeIterator.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SerializeIterator.java
@@ -21,17 +21,17 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Converts the given `resource_handle` representing an iterator to a variant tensor.
  */
 @Operator(group = "data")
-public final class SerializeIterator extends RawOp implements Operand<Tensor> {
+public final class SerializeIterator extends RawOp implements Operand<TType> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.data.SerializeIterator}
@@ -63,7 +63,7 @@ private Options() {
   @Endpoint(describeByClass = true)
   public static SerializeIterator create(Scope scope, Operand<?> resourceHandle, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("SerializeIterator", scope.makeOpName("SerializeIterator"));
-    opBuilder.addInput(resourceHandle.asOutput());
+    opBuilder.addInput(resourceHandle.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -92,8 +92,8 @@ public Output<?> serialized() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<Tensor> asOutput() {
-    return (Output<Tensor>) serialized;
+  public Output<TType> asOutput(Scope scope) {
+    return (Output<TType>) serialized;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SetStatsAggregatorDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SetStatsAggregatorDataset.java
index 961776fb2b1..6c123952bdc 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SetStatsAggregatorDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SetStatsAggregatorDataset.java
@@ -23,17 +23,17 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TString;
+import org.tensorflow.types.family.TType;
 
 /**
  */
-public final class SetStatsAggregatorDataset extends RawOp implements Operand<Tensor> {
+public final class SetStatsAggregatorDataset extends RawOp implements Operand<TType> {
   
   /**
    * Factory method to create a class wrapping a new SetStatsAggregatorDataset operation.
@@ -50,10 +50,10 @@ public final class SetStatsAggregatorDataset extends RawOp implements Operand<Te
   @Endpoint(describeByClass = true)
   public static SetStatsAggregatorDataset create(Scope scope, Operand<?> inputDataset, Operand<?> statsAggregator, Operand<TString> tag, Operand<TString> counterPrefix, List<DataType<?>> outputTypes, List<Shape> outputShapes) {
     OperationBuilder opBuilder = scope.env().opBuilder("SetStatsAggregatorDataset", scope.makeOpName("SetStatsAggregatorDataset"));
-    opBuilder.addInput(inputDataset.asOutput());
-    opBuilder.addInput(statsAggregator.asOutput());
-    opBuilder.addInput(tag.asOutput());
-    opBuilder.addInput(counterPrefix.asOutput());
+    opBuilder.addInput(inputDataset.asOutput(scope));
+    opBuilder.addInput(statsAggregator.asOutput(scope));
+    opBuilder.addInput(tag.asOutput(scope));
+    opBuilder.addInput(counterPrefix.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     DataType[] outputTypesArray = new DataType[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
@@ -76,8 +76,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<Tensor> asOutput() {
-    return (Output<Tensor>) handle;
+  public Output<TType> asOutput(Scope scope) {
+    return (Output<TType>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ShardDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ShardDataset.java
index 4c53f39cbc2..7a77bd7eed7 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ShardDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ShardDataset.java
@@ -23,18 +23,18 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
 
 /**
  * Creates a `Dataset` that includes only 1/`num_shards` of this dataset.
  */
-public final class ShardDataset extends RawOp implements Operand<Tensor> {
+public final class ShardDataset extends RawOp implements Operand<TType> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.data.ShardDataset}
@@ -70,9 +70,9 @@ private Options() {
   @Endpoint(describeByClass = true)
   public static ShardDataset create(Scope scope, Operand<?> inputDataset, Operand<TInt64> numShards, Operand<TInt64> index, List<DataType<?>> outputTypes, List<Shape> outputShapes, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ShardDataset", scope.makeOpName("ShardDataset"));
-    opBuilder.addInput(inputDataset.asOutput());
-    opBuilder.addInput(numShards.asOutput());
-    opBuilder.addInput(index.asOutput());
+    opBuilder.addInput(inputDataset.asOutput(scope));
+    opBuilder.addInput(numShards.asOutput(scope));
+    opBuilder.addInput(index.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     DataType[] outputTypesArray = new DataType[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
@@ -109,8 +109,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<Tensor> asOutput() {
-    return (Output<Tensor>) handle;
+  public Output<TType> asOutput(Scope scope) {
+    return (Output<TType>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ShuffleAndRepeatDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ShuffleAndRepeatDataset.java
index 9f2e233bbf8..13a629bb3c8 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ShuffleAndRepeatDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ShuffleAndRepeatDataset.java
@@ -23,61 +23,47 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
 
 /**
+ * Creates a dataset that shuffles and repeats elements from `input_dataset`
+ * <p>
+ * pseudorandomly.
  */
-public final class ShuffleAndRepeatDataset extends RawOp implements Operand<Tensor> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.data.ShuffleAndRepeatDataset}
-   */
-  public static class Options {
-    
-    /**
-     * @param reshuffleEachIteration 
-     */
-    public Options reshuffleEachIteration(Boolean reshuffleEachIteration) {
-      this.reshuffleEachIteration = reshuffleEachIteration;
-      return this;
-    }
-    
-    private Boolean reshuffleEachIteration;
-    
-    private Options() {
-    }
-  }
+public final class ShuffleAndRepeatDataset extends RawOp implements Operand<TType> {
   
   /**
    * Factory method to create a class wrapping a new ShuffleAndRepeatDataset operation.
    * 
    * @param scope current scope
    * @param inputDataset 
-   * @param bufferSize 
-   * @param seed 
-   * @param seed2 
-   * @param count 
-   * @param seedGenerator 
+   * @param bufferSize The number of output elements to buffer in an iterator over
+   * this dataset. Compare with the `min_after_dequeue` attr when creating a
+   * `RandomShuffleQueue`.
+   * @param seed A scalar seed for the random number generator. If either `seed` or
+   * `seed2` is set to be non-zero, the random number generator is seeded
+   * by the given seed.  Otherwise, a random seed is used.
+   * @param seed2 A second scalar seed to avoid seed collision.
+   * @param count A scalar representing the number of times the underlying dataset
+   * should be repeated. The default is `-1`, which results in infinite repetition.
    * @param outputTypes 
    * @param outputShapes 
-   * @param options carries optional attributes values
    * @return a new instance of ShuffleAndRepeatDataset
    */
   @Endpoint(describeByClass = true)
-  public static ShuffleAndRepeatDataset create(Scope scope, Operand<?> inputDataset, Operand<TInt64> bufferSize, Operand<TInt64> seed, Operand<TInt64> seed2, Operand<TInt64> count, Operand<?> seedGenerator, List<DataType<?>> outputTypes, List<Shape> outputShapes, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ShuffleAndRepeatDatasetV2", scope.makeOpName("ShuffleAndRepeatDataset"));
-    opBuilder.addInput(inputDataset.asOutput());
-    opBuilder.addInput(bufferSize.asOutput());
-    opBuilder.addInput(seed.asOutput());
-    opBuilder.addInput(seed2.asOutput());
-    opBuilder.addInput(count.asOutput());
-    opBuilder.addInput(seedGenerator.asOutput());
+  public static ShuffleAndRepeatDataset create(Scope scope, Operand<?> inputDataset, Operand<TInt64> bufferSize, Operand<TInt64> seed, Operand<TInt64> seed2, Operand<TInt64> count, List<DataType<?>> outputTypes, List<Shape> outputShapes) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ShuffleAndRepeatDataset", scope.makeOpName("ShuffleAndRepeatDataset"));
+    opBuilder.addInput(inputDataset.asOutput(scope));
+    opBuilder.addInput(bufferSize.asOutput(scope));
+    opBuilder.addInput(seed.asOutput(scope));
+    opBuilder.addInput(seed2.asOutput(scope));
+    opBuilder.addInput(count.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     DataType[] outputTypesArray = new DataType[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
@@ -89,23 +75,9 @@ public static ShuffleAndRepeatDataset create(Scope scope, Operand<?> inputDatase
       outputShapesArray[i] = outputShapes.get(i);
     }
     opBuilder.setAttr("output_shapes", outputShapesArray);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.reshuffleEachIteration != null) {
-          opBuilder.setAttr("reshuffle_each_iteration", opts.reshuffleEachIteration);
-        }
-      }
-    }
     return new ShuffleAndRepeatDataset(opBuilder.build());
   }
   
-  /**
-   * @param reshuffleEachIteration 
-   */
-  public static Options reshuffleEachIteration(Boolean reshuffleEachIteration) {
-    return new Options().reshuffleEachIteration(reshuffleEachIteration);
-  }
-  
   /**
    */
   public Output<?> handle() {
@@ -114,13 +86,10 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<Tensor> asOutput() {
-    return (Output<Tensor>) handle;
+  public Output<TType> asOutput(Scope scope) {
+    return (Output<TType>) handle;
   }
   
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ShuffleAndRepeatDatasetV2";
-  
   private Output<?> handle;
   
   private ShuffleAndRepeatDataset(Operation operation) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ShuffleDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ShuffleDataset.java
index 01ce2654dc6..1401683e34f 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ShuffleDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ShuffleDataset.java
@@ -23,36 +23,17 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
 
 /**
  */
-public final class ShuffleDataset extends RawOp implements Operand<Tensor> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.data.ShuffleDataset}
-   */
-  public static class Options {
-    
-    /**
-     * @param reshuffleEachIteration 
-     */
-    public Options reshuffleEachIteration(Boolean reshuffleEachIteration) {
-      this.reshuffleEachIteration = reshuffleEachIteration;
-      return this;
-    }
-    
-    private Boolean reshuffleEachIteration;
-    
-    private Options() {
-    }
-  }
+public final class ShuffleDataset extends RawOp implements Operand<TType> {
   
   /**
    * Factory method to create a class wrapping a new ShuffleDataset operation.
@@ -60,22 +41,17 @@ private Options() {
    * @param scope current scope
    * @param inputDataset 
    * @param bufferSize 
-   * @param seed 
-   * @param seed2 
    * @param seedGenerator 
    * @param outputTypes 
    * @param outputShapes 
-   * @param options carries optional attributes values
    * @return a new instance of ShuffleDataset
    */
   @Endpoint(describeByClass = true)
-  public static ShuffleDataset create(Scope scope, Operand<?> inputDataset, Operand<TInt64> bufferSize, Operand<TInt64> seed, Operand<TInt64> seed2, Operand<?> seedGenerator, List<DataType<?>> outputTypes, List<Shape> outputShapes, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ShuffleDatasetV3", scope.makeOpName("ShuffleDataset"));
-    opBuilder.addInput(inputDataset.asOutput());
-    opBuilder.addInput(bufferSize.asOutput());
-    opBuilder.addInput(seed.asOutput());
-    opBuilder.addInput(seed2.asOutput());
-    opBuilder.addInput(seedGenerator.asOutput());
+  public static ShuffleDataset create(Scope scope, Operand<?> inputDataset, Operand<TInt64> bufferSize, Operand<?> seedGenerator, List<DataType<?>> outputTypes, List<Shape> outputShapes) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ShuffleDatasetV2", scope.makeOpName("ShuffleDataset"));
+    opBuilder.addInput(inputDataset.asOutput(scope));
+    opBuilder.addInput(bufferSize.asOutput(scope));
+    opBuilder.addInput(seedGenerator.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     DataType[] outputTypesArray = new DataType[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
@@ -87,23 +63,9 @@ public static ShuffleDataset create(Scope scope, Operand<?> inputDataset, Operan
       outputShapesArray[i] = outputShapes.get(i);
     }
     opBuilder.setAttr("output_shapes", outputShapesArray);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.reshuffleEachIteration != null) {
-          opBuilder.setAttr("reshuffle_each_iteration", opts.reshuffleEachIteration);
-        }
-      }
-    }
     return new ShuffleDataset(opBuilder.build());
   }
   
-  /**
-   * @param reshuffleEachIteration 
-   */
-  public static Options reshuffleEachIteration(Boolean reshuffleEachIteration) {
-    return new Options().reshuffleEachIteration(reshuffleEachIteration);
-  }
-  
   /**
    */
   public Output<?> handle() {
@@ -112,13 +74,10 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<Tensor> asOutput() {
-    return (Output<Tensor>) handle;
+  public Output<TType> asOutput(Scope scope) {
+    return (Output<TType>) handle;
   }
   
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ShuffleDatasetV3";
-  
   private Output<?> handle;
   
   private ShuffleDataset(Operation operation) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SkipDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SkipDataset.java
index 1ab6e64f1d2..266bc4a9ce3 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SkipDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SkipDataset.java
@@ -23,19 +23,19 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
 
 /**
  * Creates a dataset that skips `count` elements from the `input_dataset`.
  */
 @Operator(group = "data")
-public final class SkipDataset extends RawOp implements Operand<Tensor> {
+public final class SkipDataset extends RawOp implements Operand<TType> {
   
   /**
    * Factory method to create a class wrapping a new SkipDataset operation.
@@ -51,8 +51,8 @@ public final class SkipDataset extends RawOp implements Operand<Tensor> {
   @Endpoint(describeByClass = true)
   public static SkipDataset create(Scope scope, Operand<?> inputDataset, Operand<TInt64> count, List<DataType<?>> outputTypes, List<Shape> outputShapes) {
     OperationBuilder opBuilder = scope.env().opBuilder("SkipDataset", scope.makeOpName("SkipDataset"));
-    opBuilder.addInput(inputDataset.asOutput());
-    opBuilder.addInput(count.asOutput());
+    opBuilder.addInput(inputDataset.asOutput(scope));
+    opBuilder.addInput(count.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     DataType[] outputTypesArray = new DataType[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
@@ -75,8 +75,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<Tensor> asOutput() {
-    return (Output<Tensor>) handle;
+  public Output<TType> asOutput(Scope scope) {
+    return (Output<TType>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SleepDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SleepDataset.java
index 614548839da..965eece0409 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SleepDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SleepDataset.java
@@ -23,17 +23,17 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
 
 /**
  */
-public final class SleepDataset extends RawOp implements Operand<Tensor> {
+public final class SleepDataset extends RawOp implements Operand<TType> {
   
   /**
    * Factory method to create a class wrapping a new SleepDataset operation.
@@ -48,8 +48,8 @@ public final class SleepDataset extends RawOp implements Operand<Tensor> {
   @Endpoint(describeByClass = true)
   public static SleepDataset create(Scope scope, Operand<?> inputDataset, Operand<TInt64> sleepMicroseconds, List<DataType<?>> outputTypes, List<Shape> outputShapes) {
     OperationBuilder opBuilder = scope.env().opBuilder("SleepDataset", scope.makeOpName("SleepDataset"));
-    opBuilder.addInput(inputDataset.asOutput());
-    opBuilder.addInput(sleepMicroseconds.asOutput());
+    opBuilder.addInput(inputDataset.asOutput(scope));
+    opBuilder.addInput(sleepMicroseconds.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     DataType[] outputTypesArray = new DataType[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
@@ -72,8 +72,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<Tensor> asOutput() {
-    return (Output<Tensor>) handle;
+  public Output<TType> asOutput(Scope scope) {
+    return (Output<TType>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SlidingWindowDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SlidingWindowDataset.java
index 0dd1856499d..374166983b3 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SlidingWindowDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SlidingWindowDataset.java
@@ -23,18 +23,18 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
 
 /**
  * Creates a dataset that passes a sliding window over `input_dataset`.
  */
-public final class SlidingWindowDataset extends RawOp implements Operand<Tensor> {
+public final class SlidingWindowDataset extends RawOp implements Operand<TType> {
   
   /**
    * Factory method to create a class wrapping a new SlidingWindowDataset operation.
@@ -54,10 +54,10 @@ public final class SlidingWindowDataset extends RawOp implements Operand<Tensor>
   @Endpoint(describeByClass = true)
   public static SlidingWindowDataset create(Scope scope, Operand<?> inputDataset, Operand<TInt64> windowSize, Operand<TInt64> windowShift, Operand<TInt64> windowStride, List<DataType<?>> outputTypes, List<Shape> outputShapes) {
     OperationBuilder opBuilder = scope.env().opBuilder("SlidingWindowDataset", scope.makeOpName("SlidingWindowDataset"));
-    opBuilder.addInput(inputDataset.asOutput());
-    opBuilder.addInput(windowSize.asOutput());
-    opBuilder.addInput(windowShift.asOutput());
-    opBuilder.addInput(windowStride.asOutput());
+    opBuilder.addInput(inputDataset.asOutput(scope));
+    opBuilder.addInput(windowSize.asOutput(scope));
+    opBuilder.addInput(windowShift.asOutput(scope));
+    opBuilder.addInput(windowStride.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     DataType[] outputTypesArray = new DataType[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
@@ -80,8 +80,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<Tensor> asOutput() {
-    return (Output<Tensor>) handle;
+  public Output<TType> asOutput(Scope scope) {
+    return (Output<TType>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SnapshotDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SnapshotDataset.java
index a5626396908..89b1246381b 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SnapshotDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SnapshotDataset.java
@@ -23,13 +23,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TString;
+import org.tensorflow.types.family.TType;
 
 /**
  * Creates a dataset that will write to / read from a snapshot.
@@ -39,7 +39,7 @@
  * If not, it will run the preprocessing pipeline as usual, and write out a
  * snapshot of the data processed for future use.
  */
-public final class SnapshotDataset extends RawOp implements Operand<Tensor> {
+public final class SnapshotDataset extends RawOp implements Operand<TType> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.data.SnapshotDataset}
@@ -191,8 +191,8 @@ private Options() {
   @Endpoint(describeByClass = true)
   public static SnapshotDataset create(Scope scope, Operand<?> inputDataset, Operand<TString> path, List<DataType<?>> outputTypes, List<Shape> outputShapes, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("SnapshotDataset", scope.makeOpName("SnapshotDataset"));
-    opBuilder.addInput(inputDataset.asOutput());
-    opBuilder.addInput(path.asOutput());
+    opBuilder.addInput(inputDataset.asOutput(scope));
+    opBuilder.addInput(path.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     DataType[] outputTypesArray = new DataType[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
@@ -359,8 +359,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<Tensor> asOutput() {
-    return (Output<Tensor>) handle;
+  public Output<TType> asOutput(Scope scope) {
+    return (Output<TType>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SparseTensorSliceDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SparseTensorSliceDataset.java
index 04de4360549..23785d96b37 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SparseTensorSliceDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SparseTensorSliceDataset.java
@@ -21,17 +21,17 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
 
 /**
  * Creates a dataset that splits a SparseTensor into elements row-wise.
  */
-public final class SparseTensorSliceDataset extends RawOp implements Operand<Tensor> {
+public final class SparseTensorSliceDataset extends RawOp implements Operand<TType> {
   
   /**
    * Factory method to create a class wrapping a new SparseTensorSliceDataset operation.
@@ -43,11 +43,11 @@ public final class SparseTensorSliceDataset extends RawOp implements Operand<Ten
    * @return a new instance of SparseTensorSliceDataset
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> SparseTensorSliceDataset create(Scope scope, Operand<TInt64> indices, Operand<T> values, Operand<TInt64> denseShape) {
+  public static <T extends TType> SparseTensorSliceDataset create(Scope scope, Operand<TInt64> indices, Operand<T> values, Operand<TInt64> denseShape) {
     OperationBuilder opBuilder = scope.env().opBuilder("SparseTensorSliceDataset", scope.makeOpName("SparseTensorSliceDataset"));
-    opBuilder.addInput(indices.asOutput());
-    opBuilder.addInput(values.asOutput());
-    opBuilder.addInput(denseShape.asOutput());
+    opBuilder.addInput(indices.asOutput(scope));
+    opBuilder.addInput(values.asOutput(scope));
+    opBuilder.addInput(denseShape.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new SparseTensorSliceDataset(opBuilder.build());
   }
@@ -60,8 +60,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<Tensor> asOutput() {
-    return (Output<Tensor>) handle;
+  public Output<TType> asOutput(Scope scope) {
+    return (Output<TType>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SqlDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SqlDataset.java
index 94bd2a8763c..68088e5353a 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SqlDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SqlDataset.java
@@ -23,18 +23,18 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TString;
+import org.tensorflow.types.family.TType;
 
 /**
  * Creates a dataset that executes a SQL query and emits rows of the result set.
  */
-public final class SqlDataset extends RawOp implements Operand<Tensor> {
+public final class SqlDataset extends RawOp implements Operand<TType> {
   
   /**
    * Factory method to create a class wrapping a new SqlDataset operation.
@@ -50,9 +50,9 @@ public final class SqlDataset extends RawOp implements Operand<Tensor> {
   @Endpoint(describeByClass = true)
   public static SqlDataset create(Scope scope, Operand<TString> driverName, Operand<TString> dataSourceName, Operand<TString> query, List<DataType<?>> outputTypes, List<Shape> outputShapes) {
     OperationBuilder opBuilder = scope.env().opBuilder("SqlDataset", scope.makeOpName("SqlDataset"));
-    opBuilder.addInput(driverName.asOutput());
-    opBuilder.addInput(dataSourceName.asOutput());
-    opBuilder.addInput(query.asOutput());
+    opBuilder.addInput(driverName.asOutput(scope));
+    opBuilder.addInput(dataSourceName.asOutput(scope));
+    opBuilder.addInput(query.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     DataType[] outputTypesArray = new DataType[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
@@ -75,8 +75,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<Tensor> asOutput() {
-    return (Output<Tensor>) handle;
+  public Output<TType> asOutput(Scope scope) {
+    return (Output<TType>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/StatsAggregatorHandle.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/StatsAggregatorHandle.java
index 0ca63d5f345..f96d0930a23 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/StatsAggregatorHandle.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/StatsAggregatorHandle.java
@@ -21,16 +21,16 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Creates a statistics manager resource.
  */
-public final class StatsAggregatorHandle extends RawOp implements Operand<Tensor> {
+public final class StatsAggregatorHandle extends RawOp implements Operand<TType> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.data.StatsAggregatorHandle}
@@ -106,8 +106,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<Tensor> asOutput() {
-    return (Output<Tensor>) handle;
+  public Output<TType> asOutput(Scope scope) {
+    return (Output<TType>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/TakeDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/TakeDataset.java
index a2a74c6e377..25199e03482 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/TakeDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/TakeDataset.java
@@ -23,19 +23,19 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
 
 /**
  * Creates a dataset that contains `count` elements from the `input_dataset`.
  */
 @Operator(group = "data")
-public final class TakeDataset extends RawOp implements Operand<Tensor> {
+public final class TakeDataset extends RawOp implements Operand<TType> {
   
   /**
    * Factory method to create a class wrapping a new TakeDataset operation.
@@ -52,8 +52,8 @@ public final class TakeDataset extends RawOp implements Operand<Tensor> {
   @Endpoint(describeByClass = true)
   public static TakeDataset create(Scope scope, Operand<?> inputDataset, Operand<TInt64> count, List<DataType<?>> outputTypes, List<Shape> outputShapes) {
     OperationBuilder opBuilder = scope.env().opBuilder("TakeDataset", scope.makeOpName("TakeDataset"));
-    opBuilder.addInput(inputDataset.asOutput());
-    opBuilder.addInput(count.asOutput());
+    opBuilder.addInput(inputDataset.asOutput(scope));
+    opBuilder.addInput(count.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     DataType[] outputTypesArray = new DataType[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
@@ -76,8 +76,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<Tensor> asOutput() {
-    return (Output<Tensor>) handle;
+  public Output<TType> asOutput(Scope scope) {
+    return (Output<TType>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/TensorDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/TensorDataset.java
index 03f508352d2..943ee03785e 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/TensorDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/TensorDataset.java
@@ -22,18 +22,18 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.Operands;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Creates a dataset that emits `components` as a tuple of tensors once.
  */
-public final class TensorDataset extends RawOp implements Operand<Tensor> {
+public final class TensorDataset extends RawOp implements Operand<TType> {
   
   /**
    * Factory method to create a class wrapping a new TensorDataset operation.
@@ -46,7 +46,7 @@ public final class TensorDataset extends RawOp implements Operand<Tensor> {
   @Endpoint(describeByClass = true)
   public static TensorDataset create(Scope scope, Iterable<Operand<?>> components, List<Shape> outputShapes) {
     OperationBuilder opBuilder = scope.env().opBuilder("TensorDataset", scope.makeOpName("TensorDataset"));
-    opBuilder.addInputList(Operands.asOutputs(components));
+    opBuilder.addInputList(Operands.asOutputs(scope, components));
     opBuilder = scope.applyControlDependencies(opBuilder);
     Shape[] outputShapesArray = new Shape[outputShapes.size()];
     for (int i = 0; i < outputShapesArray.length; ++i) {
@@ -64,8 +64,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<Tensor> asOutput() {
-    return (Output<Tensor>) handle;
+  public Output<TType> asOutput(Scope scope) {
+    return (Output<TType>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/TensorSliceDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/TensorSliceDataset.java
index 3593b6d0e4c..038991f405b 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/TensorSliceDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/TensorSliceDataset.java
@@ -22,19 +22,19 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.Operands;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Creates a dataset that emits each dim-0 slice of `components` once.
  */
 @Operator(group = "data")
-public final class TensorSliceDataset extends RawOp implements Operand<Tensor> {
+public final class TensorSliceDataset extends RawOp implements Operand<TType> {
   
   /**
    * Factory method to create a class wrapping a new TensorSliceDataset operation.
@@ -47,7 +47,7 @@ public final class TensorSliceDataset extends RawOp implements Operand<Tensor> {
   @Endpoint(describeByClass = true)
   public static TensorSliceDataset create(Scope scope, Iterable<Operand<?>> components, List<Shape> outputShapes) {
     OperationBuilder opBuilder = scope.env().opBuilder("TensorSliceDataset", scope.makeOpName("TensorSliceDataset"));
-    opBuilder.addInputList(Operands.asOutputs(components));
+    opBuilder.addInputList(Operands.asOutputs(scope, components));
     opBuilder = scope.applyControlDependencies(opBuilder);
     Shape[] outputShapesArray = new Shape[outputShapes.size()];
     for (int i = 0; i < outputShapesArray.length; ++i) {
@@ -65,8 +65,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<Tensor> asOutput() {
-    return (Output<Tensor>) handle;
+  public Output<TType> asOutput(Scope scope) {
+    return (Output<TType>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/TextLineDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/TextLineDataset.java
index e66b001eb72..c5f0b5de678 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/TextLineDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/TextLineDataset.java
@@ -21,19 +21,19 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
 import org.tensorflow.types.TString;
+import org.tensorflow.types.family.TType;
 
 /**
  * Creates a dataset that emits the lines of one or more text files.
  */
 @Operator(group = "data")
-public final class TextLineDataset extends RawOp implements Operand<Tensor> {
+public final class TextLineDataset extends RawOp implements Operand<TType> {
   
   /**
    * Factory method to create a class wrapping a new TextLineDataset operation.
@@ -49,9 +49,9 @@ public final class TextLineDataset extends RawOp implements Operand<Tensor> {
   @Endpoint(describeByClass = true)
   public static TextLineDataset create(Scope scope, Operand<TString> filenames, Operand<TString> compressionType, Operand<TInt64> bufferSize) {
     OperationBuilder opBuilder = scope.env().opBuilder("TextLineDataset", scope.makeOpName("TextLineDataset"));
-    opBuilder.addInput(filenames.asOutput());
-    opBuilder.addInput(compressionType.asOutput());
-    opBuilder.addInput(bufferSize.asOutput());
+    opBuilder.addInput(filenames.asOutput(scope));
+    opBuilder.addInput(compressionType.asOutput(scope));
+    opBuilder.addInput(bufferSize.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new TextLineDataset(opBuilder.build());
   }
@@ -64,8 +64,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<Tensor> asOutput() {
-    return (Output<Tensor>) handle;
+  public Output<TType> asOutput(Scope scope) {
+    return (Output<TType>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/TfRecordDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/TfRecordDataset.java
index 04043e1d38d..d87da730c87 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/TfRecordDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/TfRecordDataset.java
@@ -21,19 +21,19 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
 import org.tensorflow.types.TString;
+import org.tensorflow.types.family.TType;
 
 /**
  * Creates a dataset that emits the records from one or more TFRecord files.
  */
 @Operator(group = "data")
-public final class TfRecordDataset extends RawOp implements Operand<Tensor> {
+public final class TfRecordDataset extends RawOp implements Operand<TType> {
   
   /**
    * Factory method to create a class wrapping a new TfRecordDataset operation.
@@ -50,9 +50,9 @@ public final class TfRecordDataset extends RawOp implements Operand<Tensor> {
   @Endpoint(describeByClass = true)
   public static TfRecordDataset create(Scope scope, Operand<TString> filenames, Operand<TString> compressionType, Operand<TInt64> bufferSize) {
     OperationBuilder opBuilder = scope.env().opBuilder("TFRecordDataset", scope.makeOpName("TfRecordDataset"));
-    opBuilder.addInput(filenames.asOutput());
-    opBuilder.addInput(compressionType.asOutput());
-    opBuilder.addInput(bufferSize.asOutput());
+    opBuilder.addInput(filenames.asOutput(scope));
+    opBuilder.addInput(compressionType.asOutput(scope));
+    opBuilder.addInput(bufferSize.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new TfRecordDataset(opBuilder.build());
   }
@@ -65,8 +65,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<Tensor> asOutput() {
-    return (Output<Tensor>) handle;
+  public Output<TType> asOutput(Scope scope) {
+    return (Output<TType>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ThreadPoolDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ThreadPoolDataset.java
index 2e5893a3274..001f6e6ebe1 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ThreadPoolDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ThreadPoolDataset.java
@@ -23,17 +23,17 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Creates a dataset that uses a custom thread pool to compute `input_dataset`.
  */
-public final class ThreadPoolDataset extends RawOp implements Operand<Tensor> {
+public final class ThreadPoolDataset extends RawOp implements Operand<TType> {
   
   /**
    * Factory method to create a class wrapping a new ThreadPoolDataset operation.
@@ -48,8 +48,8 @@ public final class ThreadPoolDataset extends RawOp implements Operand<Tensor> {
   @Endpoint(describeByClass = true)
   public static ThreadPoolDataset create(Scope scope, Operand<?> inputDataset, Operand<?> threadPool, List<DataType<?>> outputTypes, List<Shape> outputShapes) {
     OperationBuilder opBuilder = scope.env().opBuilder("ThreadPoolDataset", scope.makeOpName("ThreadPoolDataset"));
-    opBuilder.addInput(inputDataset.asOutput());
-    opBuilder.addInput(threadPool.asOutput());
+    opBuilder.addInput(inputDataset.asOutput(scope));
+    opBuilder.addInput(threadPool.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     DataType[] outputTypesArray = new DataType[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
@@ -72,8 +72,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<Tensor> asOutput() {
-    return (Output<Tensor>) handle;
+  public Output<TType> asOutput(Scope scope) {
+    return (Output<TType>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ThreadPoolHandle.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ThreadPoolHandle.java
index 10847d750c4..ef7ccb677a9 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ThreadPoolHandle.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ThreadPoolHandle.java
@@ -21,16 +21,16 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Creates a dataset that uses a custom thread pool to compute `input_dataset`.
  */
-public final class ThreadPoolHandle extends RawOp implements Operand<Tensor> {
+public final class ThreadPoolHandle extends RawOp implements Operand<TType> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.data.ThreadPoolHandle}
@@ -135,8 +135,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<Tensor> asOutput() {
-    return (Output<Tensor>) handle;
+  public Output<TType> asOutput(Scope scope) {
+    return (Output<TType>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/UnbatchDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/UnbatchDataset.java
index 001d8d7aab0..568a57e051f 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/UnbatchDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/UnbatchDataset.java
@@ -23,17 +23,17 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * A dataset that splits the elements of its input into multiple elements.
  */
-public final class UnbatchDataset extends RawOp implements Operand<Tensor> {
+public final class UnbatchDataset extends RawOp implements Operand<TType> {
   
   /**
    * Factory method to create a class wrapping a new UnbatchDataset operation.
@@ -47,7 +47,7 @@ public final class UnbatchDataset extends RawOp implements Operand<Tensor> {
   @Endpoint(describeByClass = true)
   public static UnbatchDataset create(Scope scope, Operand<?> inputDataset, List<DataType<?>> outputTypes, List<Shape> outputShapes) {
     OperationBuilder opBuilder = scope.env().opBuilder("UnbatchDataset", scope.makeOpName("UnbatchDataset"));
-    opBuilder.addInput(inputDataset.asOutput());
+    opBuilder.addInput(inputDataset.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     DataType[] outputTypesArray = new DataType[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
@@ -70,8 +70,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<Tensor> asOutput() {
-    return (Output<Tensor>) handle;
+  public Output<TType> asOutput(Scope scope) {
+    return (Output<TType>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/UniqueDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/UniqueDataset.java
index 26597f1d562..10cd6e1d5a3 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/UniqueDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/UniqueDataset.java
@@ -23,17 +23,17 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Creates a dataset that contains the unique elements of `input_dataset`.
  */
-public final class UniqueDataset extends RawOp implements Operand<Tensor> {
+public final class UniqueDataset extends RawOp implements Operand<TType> {
   
   /**
    * Factory method to create a class wrapping a new UniqueDataset operation.
@@ -47,7 +47,7 @@ public final class UniqueDataset extends RawOp implements Operand<Tensor> {
   @Endpoint(describeByClass = true)
   public static UniqueDataset create(Scope scope, Operand<?> inputDataset, List<DataType<?>> outputTypes, List<Shape> outputShapes) {
     OperationBuilder opBuilder = scope.env().opBuilder("UniqueDataset", scope.makeOpName("UniqueDataset"));
-    opBuilder.addInput(inputDataset.asOutput());
+    opBuilder.addInput(inputDataset.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     DataType[] outputTypesArray = new DataType[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
@@ -70,8 +70,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<Tensor> asOutput() {
-    return (Output<Tensor>) handle;
+  public Output<TType> asOutput(Scope scope) {
+    return (Output<TType>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/UnwrapDatasetVariant.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/UnwrapDatasetVariant.java
index b3189ce18b9..982d43e9168 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/UnwrapDatasetVariant.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/UnwrapDatasetVariant.java
@@ -21,15 +21,15 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  */
-public final class UnwrapDatasetVariant extends RawOp implements Operand<Tensor> {
+public final class UnwrapDatasetVariant extends RawOp implements Operand<TType> {
   
   /**
    * Factory method to create a class wrapping a new UnwrapDatasetVariant operation.
@@ -41,7 +41,7 @@ public final class UnwrapDatasetVariant extends RawOp implements Operand<Tensor>
   @Endpoint(describeByClass = true)
   public static UnwrapDatasetVariant create(Scope scope, Operand<?> inputHandle) {
     OperationBuilder opBuilder = scope.env().opBuilder("UnwrapDatasetVariant", scope.makeOpName("UnwrapDatasetVariant"));
-    opBuilder.addInput(inputHandle.asOutput());
+    opBuilder.addInput(inputHandle.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new UnwrapDatasetVariant(opBuilder.build());
   }
@@ -54,8 +54,8 @@ public Output<?> outputHandle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<Tensor> asOutput() {
-    return (Output<Tensor>) outputHandle;
+  public Output<TType> asOutput(Scope scope) {
+    return (Output<TType>) outputHandle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/WindowDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/WindowDataset.java
index 4f48b788992..df6088f1346 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/WindowDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/WindowDataset.java
@@ -23,7 +23,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
@@ -31,6 +30,7 @@
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TBool;
 import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
 
 /**
  *   Combines (nests of) input elements into a dataset of (nests of) windows.
@@ -76,7 +76,7 @@
  *   - `tf.data.Dataset.from_tensor_slices({"a": range(4)}).window(2)`
  *     produces `{{"a": {0, 1}}, {"a": {2, 3}}}`
  */
-public final class WindowDataset extends RawOp implements Operand<Tensor> {
+public final class WindowDataset extends RawOp implements Operand<TType> {
   
   /**
    * Factory method to create a class wrapping a new WindowDataset operation.
@@ -100,11 +100,11 @@ public final class WindowDataset extends RawOp implements Operand<Tensor> {
   @Endpoint(describeByClass = true)
   public static WindowDataset create(Scope scope, Operand<?> inputDataset, Operand<TInt64> size, Operand<TInt64> shift, Operand<TInt64> stride, Operand<TBool> dropRemainder, List<DataType<?>> outputTypes, List<Shape> outputShapes) {
     OperationBuilder opBuilder = scope.env().opBuilder("WindowDataset", scope.makeOpName("WindowDataset"));
-    opBuilder.addInput(inputDataset.asOutput());
-    opBuilder.addInput(size.asOutput());
-    opBuilder.addInput(shift.asOutput());
-    opBuilder.addInput(stride.asOutput());
-    opBuilder.addInput(dropRemainder.asOutput());
+    opBuilder.addInput(inputDataset.asOutput(scope));
+    opBuilder.addInput(size.asOutput(scope));
+    opBuilder.addInput(shift.asOutput(scope));
+    opBuilder.addInput(stride.asOutput(scope));
+    opBuilder.addInput(dropRemainder.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     DataType[] outputTypesArray = new DataType[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
@@ -127,8 +127,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<Tensor> asOutput() {
-    return (Output<Tensor>) handle;
+  public Output<TType> asOutput(Scope scope) {
+    return (Output<TType>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/WrapDatasetVariant.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/WrapDatasetVariant.java
index a7af207692b..8a80750fcf5 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/WrapDatasetVariant.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/WrapDatasetVariant.java
@@ -21,15 +21,15 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  */
-public final class WrapDatasetVariant extends RawOp implements Operand<Tensor> {
+public final class WrapDatasetVariant extends RawOp implements Operand<TType> {
   
   /**
    * Factory method to create a class wrapping a new WrapDatasetVariant operation.
@@ -41,7 +41,7 @@ public final class WrapDatasetVariant extends RawOp implements Operand<Tensor> {
   @Endpoint(describeByClass = true)
   public static WrapDatasetVariant create(Scope scope, Operand<?> inputHandle) {
     OperationBuilder opBuilder = scope.env().opBuilder("WrapDatasetVariant", scope.makeOpName("WrapDatasetVariant"));
-    opBuilder.addInput(inputHandle.asOutput());
+    opBuilder.addInput(inputHandle.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new WrapDatasetVariant(opBuilder.build());
   }
@@ -54,8 +54,8 @@ public Output<?> outputHandle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<Tensor> asOutput() {
-    return (Output<Tensor>) outputHandle;
+  public Output<TType> asOutput(Scope scope) {
+    return (Output<TType>) outputHandle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ZipDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ZipDataset.java
index 36d9a1eb79b..8cf4d8abc6a 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ZipDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ZipDataset.java
@@ -23,13 +23,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.Operands;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Creates a dataset that zips together `input_datasets`.
@@ -41,7 +41,7 @@
  * dataset, and no error will be raised if input datasets have different sizes.
  */
 @Operator(group = "data")
-public final class ZipDataset extends RawOp implements Operand<Tensor> {
+public final class ZipDataset extends RawOp implements Operand<TType> {
   
   /**
    * Factory method to create a class wrapping a new ZipDataset operation.
@@ -55,7 +55,7 @@ public final class ZipDataset extends RawOp implements Operand<Tensor> {
   @Endpoint(describeByClass = true)
   public static ZipDataset create(Scope scope, Iterable<Operand<?>> inputDatasets, List<DataType<?>> outputTypes, List<Shape> outputShapes) {
     OperationBuilder opBuilder = scope.env().opBuilder("ZipDataset", scope.makeOpName("ZipDataset"));
-    opBuilder.addInputList(Operands.asOutputs(inputDatasets));
+    opBuilder.addInputList(Operands.asOutputs(scope, inputDatasets));
     opBuilder = scope.applyControlDependencies(opBuilder);
     DataType[] outputTypesArray = new DataType[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
@@ -78,8 +78,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<Tensor> asOutput() {
-    return (Output<Tensor>) handle;
+  public Output<TType> asOutput(Scope scope) {
+    return (Output<TType>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/AssertCardinalityDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/AssertCardinalityDataset.java
index cc904d6035c..67ba6f37fba 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/AssertCardinalityDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/AssertCardinalityDataset.java
@@ -23,17 +23,17 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
 
 /**
  */
-public final class AssertCardinalityDataset extends RawOp implements Operand<Tensor> {
+public final class AssertCardinalityDataset extends RawOp implements Operand<TType> {
   
   /**
    * Factory method to create a class wrapping a new AssertCardinalityDataset operation.
@@ -48,8 +48,8 @@ public final class AssertCardinalityDataset extends RawOp implements Operand<Ten
   @Endpoint(describeByClass = true)
   public static AssertCardinalityDataset create(Scope scope, Operand<?> inputDataset, Operand<TInt64> cardinality, List<DataType<?>> outputTypes, List<Shape> outputShapes) {
     OperationBuilder opBuilder = scope.env().opBuilder("AssertCardinalityDataset", scope.makeOpName("AssertCardinalityDataset"));
-    opBuilder.addInput(inputDataset.asOutput());
-    opBuilder.addInput(cardinality.asOutput());
+    opBuilder.addInput(inputDataset.asOutput(scope));
+    opBuilder.addInput(cardinality.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     DataType[] outputTypesArray = new DataType[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
@@ -72,8 +72,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<Tensor> asOutput() {
-    return (Output<Tensor>) handle;
+  public Output<TType> asOutput(Scope scope) {
+    return (Output<TType>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/AssertNextDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/AssertNextDataset.java
index ac27b6276ea..45008edaf8e 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/AssertNextDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/AssertNextDataset.java
@@ -23,17 +23,17 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TString;
+import org.tensorflow.types.family.TType;
 
 /**
  */
-public final class AssertNextDataset extends RawOp implements Operand<Tensor> {
+public final class AssertNextDataset extends RawOp implements Operand<TType> {
   
   /**
    * Factory method to create a class wrapping a new AssertNextDataset operation.
@@ -48,8 +48,8 @@ public final class AssertNextDataset extends RawOp implements Operand<Tensor> {
   @Endpoint(describeByClass = true)
   public static AssertNextDataset create(Scope scope, Operand<?> inputDataset, Operand<TString> transformations, List<DataType<?>> outputTypes, List<Shape> outputShapes) {
     OperationBuilder opBuilder = scope.env().opBuilder("ExperimentalAssertNextDataset", scope.makeOpName("AssertNextDataset"));
-    opBuilder.addInput(inputDataset.asOutput());
-    opBuilder.addInput(transformations.asOutput());
+    opBuilder.addInput(inputDataset.asOutput(scope));
+    opBuilder.addInput(transformations.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     DataType[] outputTypesArray = new DataType[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
@@ -72,8 +72,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<Tensor> asOutput() {
-    return (Output<Tensor>) handle;
+  public Output<TType> asOutput(Scope scope) {
+    return (Output<TType>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/AutoShardDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/AutoShardDataset.java
index b95e156bfb9..2eab290b2ab 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/AutoShardDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/AutoShardDataset.java
@@ -23,13 +23,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
 
 /**
  * Creates a dataset that shards the input dataset.
@@ -42,7 +42,7 @@
  * This dataset will throw a NotFound error if we cannot shard the dataset
  * automatically.
  */
-public final class AutoShardDataset extends RawOp implements Operand<Tensor> {
+public final class AutoShardDataset extends RawOp implements Operand<TType> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.data.experimental.AutoShardDataset}
@@ -78,9 +78,9 @@ private Options() {
   @Endpoint(describeByClass = true)
   public static AutoShardDataset create(Scope scope, Operand<?> inputDataset, Operand<TInt64> numWorkers, Operand<TInt64> index, List<DataType<?>> outputTypes, List<Shape> outputShapes, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ExperimentalAutoShardDataset", scope.makeOpName("AutoShardDataset"));
-    opBuilder.addInput(inputDataset.asOutput());
-    opBuilder.addInput(numWorkers.asOutput());
-    opBuilder.addInput(index.asOutput());
+    opBuilder.addInput(inputDataset.asOutput(scope));
+    opBuilder.addInput(numWorkers.asOutput(scope));
+    opBuilder.addInput(index.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     DataType[] outputTypesArray = new DataType[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
@@ -117,8 +117,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<Tensor> asOutput() {
-    return (Output<Tensor>) handle;
+  public Output<TType> asOutput(Scope scope) {
+    return (Output<TType>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/BytesProducedStatsDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/BytesProducedStatsDataset.java
index baf4210eb4e..3619b6f73ed 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/BytesProducedStatsDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/BytesProducedStatsDataset.java
@@ -23,18 +23,18 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TString;
+import org.tensorflow.types.family.TType;
 
 /**
  * Records the bytes size of each element of `input_dataset` in a StatsAggregator.
  */
-public final class BytesProducedStatsDataset extends RawOp implements Operand<Tensor> {
+public final class BytesProducedStatsDataset extends RawOp implements Operand<TType> {
   
   /**
    * Factory method to create a class wrapping a new BytesProducedStatsDataset operation.
@@ -49,8 +49,8 @@ public final class BytesProducedStatsDataset extends RawOp implements Operand<Te
   @Endpoint(describeByClass = true)
   public static BytesProducedStatsDataset create(Scope scope, Operand<?> inputDataset, Operand<TString> tag, List<DataType<?>> outputTypes, List<Shape> outputShapes) {
     OperationBuilder opBuilder = scope.env().opBuilder("ExperimentalBytesProducedStatsDataset", scope.makeOpName("BytesProducedStatsDataset"));
-    opBuilder.addInput(inputDataset.asOutput());
-    opBuilder.addInput(tag.asOutput());
+    opBuilder.addInput(inputDataset.asOutput(scope));
+    opBuilder.addInput(tag.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     DataType[] outputTypesArray = new DataType[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
@@ -73,8 +73,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<Tensor> asOutput() {
-    return (Output<Tensor>) handle;
+  public Output<TType> asOutput(Scope scope) {
+    return (Output<TType>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/CSVDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/CSVDataset.java
index 584080913d9..1fa07608599 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/CSVDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/CSVDataset.java
@@ -22,7 +22,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.Operands;
 import org.tensorflow.op.RawOp;
@@ -32,10 +31,11 @@
 import org.tensorflow.types.TBool;
 import org.tensorflow.types.TInt64;
 import org.tensorflow.types.TString;
+import org.tensorflow.types.family.TType;
 
 /**
  */
-public final class CSVDataset extends RawOp implements Operand<Tensor> {
+public final class CSVDataset extends RawOp implements Operand<TType> {
   
   /**
    * Factory method to create a class wrapping a new CSVDataset operation.
@@ -56,15 +56,15 @@ public final class CSVDataset extends RawOp implements Operand<Tensor> {
   @Endpoint(describeByClass = true)
   public static CSVDataset create(Scope scope, Operand<TString> filenames, Operand<TString> compressionType, Operand<TInt64> bufferSize, Operand<TBool> header, Operand<TString> fieldDelim, Operand<TBool> useQuoteDelim, Operand<TString> naValue, Operand<TInt64> selectCols, Iterable<Operand<?>> recordDefaults, List<Shape> outputShapes) {
     OperationBuilder opBuilder = scope.env().opBuilder("ExperimentalCSVDataset", scope.makeOpName("CSVDataset"));
-    opBuilder.addInput(filenames.asOutput());
-    opBuilder.addInput(compressionType.asOutput());
-    opBuilder.addInput(bufferSize.asOutput());
-    opBuilder.addInput(header.asOutput());
-    opBuilder.addInput(fieldDelim.asOutput());
-    opBuilder.addInput(useQuoteDelim.asOutput());
-    opBuilder.addInput(naValue.asOutput());
-    opBuilder.addInput(selectCols.asOutput());
-    opBuilder.addInputList(Operands.asOutputs(recordDefaults));
+    opBuilder.addInput(filenames.asOutput(scope));
+    opBuilder.addInput(compressionType.asOutput(scope));
+    opBuilder.addInput(bufferSize.asOutput(scope));
+    opBuilder.addInput(header.asOutput(scope));
+    opBuilder.addInput(fieldDelim.asOutput(scope));
+    opBuilder.addInput(useQuoteDelim.asOutput(scope));
+    opBuilder.addInput(naValue.asOutput(scope));
+    opBuilder.addInput(selectCols.asOutput(scope));
+    opBuilder.addInputList(Operands.asOutputs(scope, recordDefaults));
     opBuilder = scope.applyControlDependencies(opBuilder);
     Shape[] outputShapesArray = new Shape[outputShapes.size()];
     for (int i = 0; i < outputShapesArray.length; ++i) {
@@ -82,8 +82,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<Tensor> asOutput() {
-    return (Output<Tensor>) handle;
+  public Output<TType> asOutput(Scope scope) {
+    return (Output<TType>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/ChooseFastestDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/ChooseFastestDataset.java
index f2ca0bcaa95..627a9e4784b 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/ChooseFastestDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/ChooseFastestDataset.java
@@ -23,17 +23,17 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.Operands;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  */
-public final class ChooseFastestDataset extends RawOp implements Operand<Tensor> {
+public final class ChooseFastestDataset extends RawOp implements Operand<TType> {
   
   /**
    * Factory method to create a class wrapping a new ChooseFastestDataset operation.
@@ -48,7 +48,7 @@ public final class ChooseFastestDataset extends RawOp implements Operand<Tensor>
   @Endpoint(describeByClass = true)
   public static ChooseFastestDataset create(Scope scope, Iterable<Operand<?>> inputDatasets, Long numExperiments, List<DataType<?>> outputTypes, List<Shape> outputShapes) {
     OperationBuilder opBuilder = scope.env().opBuilder("ExperimentalChooseFastestDataset", scope.makeOpName("ChooseFastestDataset"));
-    opBuilder.addInputList(Operands.asOutputs(inputDatasets));
+    opBuilder.addInputList(Operands.asOutputs(scope, inputDatasets));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("num_experiments", numExperiments);
     DataType[] outputTypesArray = new DataType[outputTypes.size()];
@@ -72,8 +72,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<Tensor> asOutput() {
-    return (Output<Tensor>) handle;
+  public Output<TType> asOutput(Scope scope) {
+    return (Output<TType>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/DatasetCardinality.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/DatasetCardinality.java
index 81ca505ca91..76aea531c3d 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/DatasetCardinality.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/DatasetCardinality.java
@@ -44,7 +44,7 @@ public final class DatasetCardinality extends RawOp implements Operand<TInt64> {
   @Endpoint(describeByClass = true)
   public static DatasetCardinality create(Scope scope, Operand<?> inputDataset) {
     OperationBuilder opBuilder = scope.env().opBuilder("ExperimentalDatasetCardinality", scope.makeOpName("DatasetCardinality"));
-    opBuilder.addInput(inputDataset.asOutput());
+    opBuilder.addInput(inputDataset.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new DatasetCardinality(opBuilder.build());
   }
@@ -58,7 +58,7 @@ public Output<TInt64> cardinality() {
   }
   
   @Override
-  public Output<TInt64> asOutput() {
+  public Output<TInt64> asOutput(Scope scope) {
     return cardinality;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/DatasetToTFRecord.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/DatasetToTFRecord.java
index 6e0a0b8f2dc..ab2fda4e241 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/DatasetToTFRecord.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/DatasetToTFRecord.java
@@ -44,9 +44,9 @@ public final class DatasetToTFRecord extends RawOp {
   @Endpoint(describeByClass = true)
   public static DatasetToTFRecord create(Scope scope, Operand<?> inputDataset, Operand<TString> filename, Operand<TString> compressionType) {
     OperationBuilder opBuilder = scope.env().opBuilder("ExperimentalDatasetToTFRecord", scope.makeOpName("DatasetToTFRecord"));
-    opBuilder.addInput(inputDataset.asOutput());
-    opBuilder.addInput(filename.asOutput());
-    opBuilder.addInput(compressionType.asOutput());
+    opBuilder.addInput(inputDataset.asOutput(scope));
+    opBuilder.addInput(filename.asOutput(scope));
+    opBuilder.addInput(compressionType.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new DatasetToTFRecord(opBuilder.build());
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/DenseToSparseBatchDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/DenseToSparseBatchDataset.java
index a079ea085db..a1cae77fa4e 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/DenseToSparseBatchDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/DenseToSparseBatchDataset.java
@@ -23,18 +23,18 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
 
 /**
  * Creates a dataset that batches input elements into a SparseTensor.
  */
-public final class DenseToSparseBatchDataset extends RawOp implements Operand<Tensor> {
+public final class DenseToSparseBatchDataset extends RawOp implements Operand<TType> {
   
   /**
    * Factory method to create a class wrapping a new DenseToSparseBatchDataset operation.
@@ -53,9 +53,9 @@ public final class DenseToSparseBatchDataset extends RawOp implements Operand<Te
   @Endpoint(describeByClass = true)
   public static DenseToSparseBatchDataset create(Scope scope, Operand<?> inputDataset, Operand<TInt64> batchSize, Operand<TInt64> rowShape, List<DataType<?>> outputTypes, List<Shape> outputShapes) {
     OperationBuilder opBuilder = scope.env().opBuilder("ExperimentalDenseToSparseBatchDataset", scope.makeOpName("DenseToSparseBatchDataset"));
-    opBuilder.addInput(inputDataset.asOutput());
-    opBuilder.addInput(batchSize.asOutput());
-    opBuilder.addInput(rowShape.asOutput());
+    opBuilder.addInput(inputDataset.asOutput(scope));
+    opBuilder.addInput(batchSize.asOutput(scope));
+    opBuilder.addInput(rowShape.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     DataType[] outputTypesArray = new DataType[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
@@ -78,8 +78,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<Tensor> asOutput() {
-    return (Output<Tensor>) handle;
+  public Output<TType> asOutput(Scope scope) {
+    return (Output<TType>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/DirectedInterleaveDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/DirectedInterleaveDataset.java
index 3886abd836b..9fda314f40c 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/DirectedInterleaveDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/DirectedInterleaveDataset.java
@@ -23,18 +23,18 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.Operands;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * A substitute for `InterleaveDataset` on a fixed list of `N` datasets.
  */
-public final class DirectedInterleaveDataset extends RawOp implements Operand<Tensor> {
+public final class DirectedInterleaveDataset extends RawOp implements Operand<TType> {
   
   /**
    * Factory method to create a class wrapping a new DirectedInterleaveDataset operation.
@@ -51,8 +51,8 @@ public final class DirectedInterleaveDataset extends RawOp implements Operand<Te
   @Endpoint(describeByClass = true)
   public static DirectedInterleaveDataset create(Scope scope, Operand<?> selectorInputDataset, Iterable<Operand<?>> dataInputDatasets, List<DataType<?>> outputTypes, List<Shape> outputShapes) {
     OperationBuilder opBuilder = scope.env().opBuilder("ExperimentalDirectedInterleaveDataset", scope.makeOpName("DirectedInterleaveDataset"));
-    opBuilder.addInput(selectorInputDataset.asOutput());
-    opBuilder.addInputList(Operands.asOutputs(dataInputDatasets));
+    opBuilder.addInput(selectorInputDataset.asOutput(scope));
+    opBuilder.addInputList(Operands.asOutputs(scope, dataInputDatasets));
     opBuilder = scope.applyControlDependencies(opBuilder);
     DataType[] outputTypesArray = new DataType[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
@@ -75,8 +75,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<Tensor> asOutput() {
-    return (Output<Tensor>) handle;
+  public Output<TType> asOutput(Scope scope) {
+    return (Output<TType>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/IgnoreErrorsDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/IgnoreErrorsDataset.java
index 5d592784bc2..a14afc598a5 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/IgnoreErrorsDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/IgnoreErrorsDataset.java
@@ -23,17 +23,17 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Creates a dataset that contains the elements of `input_dataset` ignoring errors.
  */
-public final class IgnoreErrorsDataset extends RawOp implements Operand<Tensor> {
+public final class IgnoreErrorsDataset extends RawOp implements Operand<TType> {
   
   /**
    * Factory method to create a class wrapping a new IgnoreErrorsDataset operation.
@@ -47,7 +47,7 @@ public final class IgnoreErrorsDataset extends RawOp implements Operand<Tensor>
   @Endpoint(describeByClass = true)
   public static IgnoreErrorsDataset create(Scope scope, Operand<?> inputDataset, List<DataType<?>> outputTypes, List<Shape> outputShapes) {
     OperationBuilder opBuilder = scope.env().opBuilder("ExperimentalIgnoreErrorsDataset", scope.makeOpName("IgnoreErrorsDataset"));
-    opBuilder.addInput(inputDataset.asOutput());
+    opBuilder.addInput(inputDataset.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     DataType[] outputTypesArray = new DataType[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
@@ -70,8 +70,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<Tensor> asOutput() {
-    return (Output<Tensor>) handle;
+  public Output<TType> asOutput(Scope scope) {
+    return (Output<TType>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/IteratorGetDevice.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/IteratorGetDevice.java
index 7602eca94fc..f83241f5450 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/IteratorGetDevice.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/IteratorGetDevice.java
@@ -42,7 +42,7 @@ public final class IteratorGetDevice extends RawOp implements Operand<TString> {
   @Endpoint(describeByClass = true)
   public static IteratorGetDevice create(Scope scope, Operand<?> resource) {
     OperationBuilder opBuilder = scope.env().opBuilder("ExperimentalIteratorGetDevice", scope.makeOpName("IteratorGetDevice"));
-    opBuilder.addInput(resource.asOutput());
+    opBuilder.addInput(resource.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new IteratorGetDevice(opBuilder.build());
   }
@@ -54,7 +54,7 @@ public Output<TString> device() {
   }
   
   @Override
-  public Output<TString> asOutput() {
+  public Output<TString> asOutput(Scope scope) {
     return device;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/LatencyStatsDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/LatencyStatsDataset.java
index 86ee0610c2a..8073d397acd 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/LatencyStatsDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/LatencyStatsDataset.java
@@ -23,18 +23,18 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TString;
+import org.tensorflow.types.family.TType;
 
 /**
  * Records the latency of producing `input_dataset` elements in a StatsAggregator.
  */
-public final class LatencyStatsDataset extends RawOp implements Operand<Tensor> {
+public final class LatencyStatsDataset extends RawOp implements Operand<TType> {
   
   /**
    * Factory method to create a class wrapping a new LatencyStatsDataset operation.
@@ -49,8 +49,8 @@ public final class LatencyStatsDataset extends RawOp implements Operand<Tensor>
   @Endpoint(describeByClass = true)
   public static LatencyStatsDataset create(Scope scope, Operand<?> inputDataset, Operand<TString> tag, List<DataType<?>> outputTypes, List<Shape> outputShapes) {
     OperationBuilder opBuilder = scope.env().opBuilder("ExperimentalLatencyStatsDataset", scope.makeOpName("LatencyStatsDataset"));
-    opBuilder.addInput(inputDataset.asOutput());
-    opBuilder.addInput(tag.asOutput());
+    opBuilder.addInput(inputDataset.asOutput(scope));
+    opBuilder.addInput(tag.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     DataType[] outputTypesArray = new DataType[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
@@ -73,8 +73,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<Tensor> asOutput() {
-    return (Output<Tensor>) handle;
+  public Output<TType> asOutput(Scope scope) {
+    return (Output<TType>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/LmdbDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/LmdbDataset.java
index 9434c22b7bd..4c04fab718a 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/LmdbDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/LmdbDataset.java
@@ -23,17 +23,17 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TString;
+import org.tensorflow.types.family.TType;
 
 /**
  */
-public final class LmdbDataset extends RawOp implements Operand<Tensor> {
+public final class LmdbDataset extends RawOp implements Operand<TType> {
   
   /**
    * Factory method to create a class wrapping a new LmdbDataset operation.
@@ -47,7 +47,7 @@ public final class LmdbDataset extends RawOp implements Operand<Tensor> {
   @Endpoint(describeByClass = true)
   public static LmdbDataset create(Scope scope, Operand<TString> filenames, List<DataType<?>> outputTypes, List<Shape> outputShapes) {
     OperationBuilder opBuilder = scope.env().opBuilder("ExperimentalLMDBDataset", scope.makeOpName("LmdbDataset"));
-    opBuilder.addInput(filenames.asOutput());
+    opBuilder.addInput(filenames.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     DataType[] outputTypesArray = new DataType[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
@@ -70,8 +70,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<Tensor> asOutput() {
-    return (Output<Tensor>) handle;
+  public Output<TType> asOutput(Scope scope) {
+    return (Output<TType>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/MatchingFilesDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/MatchingFilesDataset.java
index f207ec77f3c..9d4db8c1f39 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/MatchingFilesDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/MatchingFilesDataset.java
@@ -21,16 +21,16 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TString;
+import org.tensorflow.types.family.TType;
 
 /**
  */
-public final class MatchingFilesDataset extends RawOp implements Operand<Tensor> {
+public final class MatchingFilesDataset extends RawOp implements Operand<TType> {
   
   /**
    * Factory method to create a class wrapping a new MatchingFilesDataset operation.
@@ -42,7 +42,7 @@ public final class MatchingFilesDataset extends RawOp implements Operand<Tensor>
   @Endpoint(describeByClass = true)
   public static MatchingFilesDataset create(Scope scope, Operand<TString> patterns) {
     OperationBuilder opBuilder = scope.env().opBuilder("ExperimentalMatchingFilesDataset", scope.makeOpName("MatchingFilesDataset"));
-    opBuilder.addInput(patterns.asOutput());
+    opBuilder.addInput(patterns.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new MatchingFilesDataset(opBuilder.build());
   }
@@ -55,8 +55,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<Tensor> asOutput() {
-    return (Output<Tensor>) handle;
+  public Output<TType> asOutput(Scope scope) {
+    return (Output<TType>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/MaxIntraOpParallelismDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/MaxIntraOpParallelismDataset.java
index d4cf2dcd57e..16e5d1a9d51 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/MaxIntraOpParallelismDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/MaxIntraOpParallelismDataset.java
@@ -23,18 +23,18 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
 
 /**
  * Creates a dataset that overrides the maximum intra-op parallelism.
  */
-public final class MaxIntraOpParallelismDataset extends RawOp implements Operand<Tensor> {
+public final class MaxIntraOpParallelismDataset extends RawOp implements Operand<TType> {
   
   /**
    * Factory method to create a class wrapping a new MaxIntraOpParallelismDataset operation.
@@ -49,8 +49,8 @@ public final class MaxIntraOpParallelismDataset extends RawOp implements Operand
   @Endpoint(describeByClass = true)
   public static MaxIntraOpParallelismDataset create(Scope scope, Operand<?> inputDataset, Operand<TInt64> maxIntraOpParallelism, List<DataType<?>> outputTypes, List<Shape> outputShapes) {
     OperationBuilder opBuilder = scope.env().opBuilder("ExperimentalMaxIntraOpParallelismDataset", scope.makeOpName("MaxIntraOpParallelismDataset"));
-    opBuilder.addInput(inputDataset.asOutput());
-    opBuilder.addInput(maxIntraOpParallelism.asOutput());
+    opBuilder.addInput(inputDataset.asOutput(scope));
+    opBuilder.addInput(maxIntraOpParallelism.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     DataType[] outputTypesArray = new DataType[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
@@ -73,8 +73,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<Tensor> asOutput() {
-    return (Output<Tensor>) handle;
+  public Output<TType> asOutput(Scope scope) {
+    return (Output<TType>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/NonSerializableDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/NonSerializableDataset.java
index 015b710c2c5..4ab5e682300 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/NonSerializableDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/NonSerializableDataset.java
@@ -23,16 +23,16 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  */
-public final class NonSerializableDataset extends RawOp implements Operand<Tensor> {
+public final class NonSerializableDataset extends RawOp implements Operand<TType> {
   
   /**
    * Factory method to create a class wrapping a new NonSerializableDataset operation.
@@ -46,7 +46,7 @@ public final class NonSerializableDataset extends RawOp implements Operand<Tenso
   @Endpoint(describeByClass = true)
   public static NonSerializableDataset create(Scope scope, Operand<?> inputDataset, List<DataType<?>> outputTypes, List<Shape> outputShapes) {
     OperationBuilder opBuilder = scope.env().opBuilder("ExperimentalNonSerializableDataset", scope.makeOpName("NonSerializableDataset"));
-    opBuilder.addInput(inputDataset.asOutput());
+    opBuilder.addInput(inputDataset.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     DataType[] outputTypesArray = new DataType[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
@@ -69,8 +69,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<Tensor> asOutput() {
-    return (Output<Tensor>) handle;
+  public Output<TType> asOutput(Scope scope) {
+    return (Output<TType>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/ParseExampleDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/ParseExampleDataset.java
index bcf05b640fe..104939a7cf9 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/ParseExampleDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/ParseExampleDataset.java
@@ -23,7 +23,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.Operands;
 import org.tensorflow.op.RawOp;
@@ -31,12 +30,12 @@
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * Transforms `input_dataset` containing `Example` protos as vectors of DT_STRING into a dataset of `Tensor` or `SparseTensor` objects representing the parsed features.
  */
-public final class ParseExampleDataset extends RawOp implements Operand<Tensor> {
+public final class ParseExampleDataset extends RawOp implements Operand<TType> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.data.experimental.ParseExampleDataset}
@@ -104,9 +103,9 @@ private Options() {
   @Endpoint(describeByClass = true)
   public static ParseExampleDataset create(Scope scope, Operand<?> inputDataset, Operand<TInt64> numParallelCalls, Iterable<Operand<?>> denseDefaults, List<String> sparseKeys, List<String> denseKeys, List<DataType<?>> sparseTypes, List<Shape> denseShapes, List<DataType<?>> outputTypes, List<Shape> outputShapes, List<DataType<?>> raggedValueTypes, List<DataType<?>> raggedSplitTypes, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ParseExampleDatasetV2", scope.makeOpName("ParseExampleDataset"));
-    opBuilder.addInput(inputDataset.asOutput());
-    opBuilder.addInput(numParallelCalls.asOutput());
-    opBuilder.addInputList(Operands.asOutputs(denseDefaults));
+    opBuilder.addInput(inputDataset.asOutput(scope));
+    opBuilder.addInput(numParallelCalls.asOutput(scope));
+    opBuilder.addInputList(Operands.asOutputs(scope, denseDefaults));
     opBuilder = scope.applyControlDependencies(opBuilder);
     String[] sparseKeysArray = new String[sparseKeys.size()];
     for (int i = 0; i < sparseKeysArray.length; ++i) {
@@ -191,8 +190,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<Tensor> asOutput() {
-    return (Output<Tensor>) handle;
+  public Output<TType> asOutput(Scope scope) {
+    return (Output<TType>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/PrivateThreadPoolDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/PrivateThreadPoolDataset.java
index c1b9d9ced27..2f31ef615da 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/PrivateThreadPoolDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/PrivateThreadPoolDataset.java
@@ -23,18 +23,18 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
 
 /**
  * Creates a dataset that uses a custom thread pool to compute `input_dataset`.
  */
-public final class PrivateThreadPoolDataset extends RawOp implements Operand<Tensor> {
+public final class PrivateThreadPoolDataset extends RawOp implements Operand<TType> {
   
   /**
    * Factory method to create a class wrapping a new PrivateThreadPoolDataset operation.
@@ -49,8 +49,8 @@ public final class PrivateThreadPoolDataset extends RawOp implements Operand<Ten
   @Endpoint(describeByClass = true)
   public static PrivateThreadPoolDataset create(Scope scope, Operand<?> inputDataset, Operand<TInt64> numThreads, List<DataType<?>> outputTypes, List<Shape> outputShapes) {
     OperationBuilder opBuilder = scope.env().opBuilder("ExperimentalPrivateThreadPoolDataset", scope.makeOpName("PrivateThreadPoolDataset"));
-    opBuilder.addInput(inputDataset.asOutput());
-    opBuilder.addInput(numThreads.asOutput());
+    opBuilder.addInput(inputDataset.asOutput(scope));
+    opBuilder.addInput(numThreads.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     DataType[] outputTypesArray = new DataType[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
@@ -73,8 +73,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<Tensor> asOutput() {
-    return (Output<Tensor>) handle;
+  public Output<TType> asOutput(Scope scope) {
+    return (Output<TType>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/RandomDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/RandomDataset.java
index 772e89379b5..2ad127b91d6 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/RandomDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/RandomDataset.java
@@ -23,18 +23,18 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
 
 /**
  * Creates a Dataset that returns pseudorandom numbers.
  */
-public final class RandomDataset extends RawOp implements Operand<Tensor> {
+public final class RandomDataset extends RawOp implements Operand<TType> {
   
   /**
    * Factory method to create a class wrapping a new RandomDataset operation.
@@ -51,8 +51,8 @@ public final class RandomDataset extends RawOp implements Operand<Tensor> {
   @Endpoint(describeByClass = true)
   public static RandomDataset create(Scope scope, Operand<TInt64> seed, Operand<TInt64> seed2, List<DataType<?>> outputTypes, List<Shape> outputShapes) {
     OperationBuilder opBuilder = scope.env().opBuilder("ExperimentalRandomDataset", scope.makeOpName("RandomDataset"));
-    opBuilder.addInput(seed.asOutput());
-    opBuilder.addInput(seed2.asOutput());
+    opBuilder.addInput(seed.asOutput(scope));
+    opBuilder.addInput(seed2.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     DataType[] outputTypesArray = new DataType[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
@@ -75,8 +75,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<Tensor> asOutput() {
-    return (Output<Tensor>) handle;
+  public Output<TType> asOutput(Scope scope) {
+    return (Output<TType>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/RebatchDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/RebatchDataset.java
index 215b1cc22eb..9c0ee15dcce 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/RebatchDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/RebatchDataset.java
@@ -23,13 +23,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
 
 /**
  * Creates a dataset that changes the batch size.
@@ -37,7 +37,7 @@
  * Creates a dataset that changes the batch size of the dataset to current batch
  * size // num_replicas.
  */
-public final class RebatchDataset extends RawOp implements Operand<Tensor> {
+public final class RebatchDataset extends RawOp implements Operand<TType> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.data.experimental.RebatchDataset}
@@ -74,8 +74,8 @@ private Options() {
   @Endpoint(describeByClass = true)
   public static RebatchDataset create(Scope scope, Operand<?> inputDataset, Operand<TInt64> numReplicas, List<DataType<?>> outputTypes, List<Shape> outputShapes, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ExperimentalRebatchDataset", scope.makeOpName("RebatchDataset"));
-    opBuilder.addInput(inputDataset.asOutput());
-    opBuilder.addInput(numReplicas.asOutput());
+    opBuilder.addInput(inputDataset.asOutput(scope));
+    opBuilder.addInput(numReplicas.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     DataType[] outputTypesArray = new DataType[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
@@ -112,8 +112,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<Tensor> asOutput() {
-    return (Output<Tensor>) handle;
+  public Output<TType> asOutput(Scope scope) {
+    return (Output<TType>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/SetStatsAggregatorDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/SetStatsAggregatorDataset.java
index dcbc74627d0..9e80103945d 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/SetStatsAggregatorDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/SetStatsAggregatorDataset.java
@@ -23,17 +23,17 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TString;
+import org.tensorflow.types.family.TType;
 
 /**
  */
-public final class SetStatsAggregatorDataset extends RawOp implements Operand<Tensor> {
+public final class SetStatsAggregatorDataset extends RawOp implements Operand<TType> {
   
   /**
    * Factory method to create a class wrapping a new SetStatsAggregatorDataset operation.
@@ -50,10 +50,10 @@ public final class SetStatsAggregatorDataset extends RawOp implements Operand<Te
   @Endpoint(describeByClass = true)
   public static SetStatsAggregatorDataset create(Scope scope, Operand<?> inputDataset, Operand<?> statsAggregator, Operand<TString> tag, Operand<TString> counterPrefix, List<DataType<?>> outputTypes, List<Shape> outputShapes) {
     OperationBuilder opBuilder = scope.env().opBuilder("ExperimentalSetStatsAggregatorDataset", scope.makeOpName("SetStatsAggregatorDataset"));
-    opBuilder.addInput(inputDataset.asOutput());
-    opBuilder.addInput(statsAggregator.asOutput());
-    opBuilder.addInput(tag.asOutput());
-    opBuilder.addInput(counterPrefix.asOutput());
+    opBuilder.addInput(inputDataset.asOutput(scope));
+    opBuilder.addInput(statsAggregator.asOutput(scope));
+    opBuilder.addInput(tag.asOutput(scope));
+    opBuilder.addInput(counterPrefix.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     DataType[] outputTypesArray = new DataType[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
@@ -76,8 +76,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<Tensor> asOutput() {
-    return (Output<Tensor>) handle;
+  public Output<TType> asOutput(Scope scope) {
+    return (Output<TType>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/SleepDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/SleepDataset.java
index 394dfbf8d4b..bf8753e5210 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/SleepDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/SleepDataset.java
@@ -23,17 +23,17 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
 
 /**
  */
-public final class SleepDataset extends RawOp implements Operand<Tensor> {
+public final class SleepDataset extends RawOp implements Operand<TType> {
   
   /**
    * Factory method to create a class wrapping a new SleepDataset operation.
@@ -48,8 +48,8 @@ public final class SleepDataset extends RawOp implements Operand<Tensor> {
   @Endpoint(describeByClass = true)
   public static SleepDataset create(Scope scope, Operand<?> inputDataset, Operand<TInt64> sleepMicroseconds, List<DataType<?>> outputTypes, List<Shape> outputShapes) {
     OperationBuilder opBuilder = scope.env().opBuilder("ExperimentalSleepDataset", scope.makeOpName("SleepDataset"));
-    opBuilder.addInput(inputDataset.asOutput());
-    opBuilder.addInput(sleepMicroseconds.asOutput());
+    opBuilder.addInput(inputDataset.asOutput(scope));
+    opBuilder.addInput(sleepMicroseconds.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     DataType[] outputTypesArray = new DataType[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
@@ -72,8 +72,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<Tensor> asOutput() {
-    return (Output<Tensor>) handle;
+  public Output<TType> asOutput(Scope scope) {
+    return (Output<TType>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/SlidingWindowDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/SlidingWindowDataset.java
index 93ec0151c10..6c583c31979 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/SlidingWindowDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/SlidingWindowDataset.java
@@ -23,18 +23,18 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
 
 /**
  * Creates a dataset that passes a sliding window over `input_dataset`.
  */
-public final class SlidingWindowDataset extends RawOp implements Operand<Tensor> {
+public final class SlidingWindowDataset extends RawOp implements Operand<TType> {
   
   /**
    * Factory method to create a class wrapping a new SlidingWindowDataset operation.
@@ -54,10 +54,10 @@ public final class SlidingWindowDataset extends RawOp implements Operand<Tensor>
   @Endpoint(describeByClass = true)
   public static SlidingWindowDataset create(Scope scope, Operand<?> inputDataset, Operand<TInt64> windowSize, Operand<TInt64> windowShift, Operand<TInt64> windowStride, List<DataType<?>> outputTypes, List<Shape> outputShapes) {
     OperationBuilder opBuilder = scope.env().opBuilder("ExperimentalSlidingWindowDataset", scope.makeOpName("SlidingWindowDataset"));
-    opBuilder.addInput(inputDataset.asOutput());
-    opBuilder.addInput(windowSize.asOutput());
-    opBuilder.addInput(windowShift.asOutput());
-    opBuilder.addInput(windowStride.asOutput());
+    opBuilder.addInput(inputDataset.asOutput(scope));
+    opBuilder.addInput(windowSize.asOutput(scope));
+    opBuilder.addInput(windowShift.asOutput(scope));
+    opBuilder.addInput(windowStride.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     DataType[] outputTypesArray = new DataType[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
@@ -80,8 +80,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<Tensor> asOutput() {
-    return (Output<Tensor>) handle;
+  public Output<TType> asOutput(Scope scope) {
+    return (Output<TType>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/SqlDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/SqlDataset.java
index 466191b26ba..c0d2f90ed3b 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/SqlDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/SqlDataset.java
@@ -23,18 +23,18 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TString;
+import org.tensorflow.types.family.TType;
 
 /**
  * Creates a dataset that executes a SQL query and emits rows of the result set.
  */
-public final class SqlDataset extends RawOp implements Operand<Tensor> {
+public final class SqlDataset extends RawOp implements Operand<TType> {
   
   /**
    * Factory method to create a class wrapping a new SqlDataset operation.
@@ -50,9 +50,9 @@ public final class SqlDataset extends RawOp implements Operand<Tensor> {
   @Endpoint(describeByClass = true)
   public static SqlDataset create(Scope scope, Operand<TString> driverName, Operand<TString> dataSourceName, Operand<TString> query, List<DataType<?>> outputTypes, List<Shape> outputShapes) {
     OperationBuilder opBuilder = scope.env().opBuilder("ExperimentalSqlDataset", scope.makeOpName("SqlDataset"));
-    opBuilder.addInput(driverName.asOutput());
-    opBuilder.addInput(dataSourceName.asOutput());
-    opBuilder.addInput(query.asOutput());
+    opBuilder.addInput(driverName.asOutput(scope));
+    opBuilder.addInput(dataSourceName.asOutput(scope));
+    opBuilder.addInput(query.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     DataType[] outputTypesArray = new DataType[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
@@ -75,8 +75,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<Tensor> asOutput() {
-    return (Output<Tensor>) handle;
+  public Output<TType> asOutput(Scope scope) {
+    return (Output<TType>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/StatsAggregatorHandle.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/StatsAggregatorHandle.java
index bc8c6ddcd6e..c79ea93abb7 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/StatsAggregatorHandle.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/StatsAggregatorHandle.java
@@ -21,15 +21,15 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  */
-public final class StatsAggregatorHandle extends RawOp implements Operand<Tensor> {
+public final class StatsAggregatorHandle extends RawOp implements Operand<TType> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.data.experimental.StatsAggregatorHandle}
@@ -105,8 +105,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<Tensor> asOutput() {
-    return (Output<Tensor>) handle;
+  public Output<TType> asOutput(Scope scope) {
+    return (Output<TType>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/StatsAggregatorSetSummaryWriter.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/StatsAggregatorSetSummaryWriter.java
index 1af246d8313..52bf7e0be75 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/StatsAggregatorSetSummaryWriter.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/StatsAggregatorSetSummaryWriter.java
@@ -41,8 +41,8 @@ public final class StatsAggregatorSetSummaryWriter extends RawOp {
   @Endpoint(describeByClass = true)
   public static StatsAggregatorSetSummaryWriter create(Scope scope, Operand<?> statsAggregator, Operand<?> summary) {
     OperationBuilder opBuilder = scope.env().opBuilder("StatsAggregatorSetSummaryWriter", scope.makeOpName("StatsAggregatorSetSummaryWriter"));
-    opBuilder.addInput(statsAggregator.asOutput());
-    opBuilder.addInput(summary.asOutput());
+    opBuilder.addInput(statsAggregator.asOutput(scope));
+    opBuilder.addInput(summary.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new StatsAggregatorSetSummaryWriter(opBuilder.build());
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/StatsAggregatorSummary.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/StatsAggregatorSummary.java
index a24bfdbaada..ec12eaed6c7 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/StatsAggregatorSummary.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/StatsAggregatorSummary.java
@@ -42,7 +42,7 @@ public final class StatsAggregatorSummary extends RawOp implements Operand<TStri
   @Endpoint(describeByClass = true)
   public static StatsAggregatorSummary create(Scope scope, Operand<?> iterator) {
     OperationBuilder opBuilder = scope.env().opBuilder("ExperimentalStatsAggregatorSummary", scope.makeOpName("StatsAggregatorSummary"));
-    opBuilder.addInput(iterator.asOutput());
+    opBuilder.addInput(iterator.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new StatsAggregatorSummary(opBuilder.build());
   }
@@ -54,7 +54,7 @@ public Output<TString> summary() {
   }
   
   @Override
-  public Output<TString> asOutput() {
+  public Output<TString> asOutput(Scope scope) {
     return summary;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/ThreadPoolDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/ThreadPoolDataset.java
index 19522616699..1e3c110f8c5 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/ThreadPoolDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/ThreadPoolDataset.java
@@ -23,17 +23,17 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Creates a dataset that uses a custom thread pool to compute `input_dataset`.
  */
-public final class ThreadPoolDataset extends RawOp implements Operand<Tensor> {
+public final class ThreadPoolDataset extends RawOp implements Operand<TType> {
   
   /**
    * Factory method to create a class wrapping a new ThreadPoolDataset operation.
@@ -48,8 +48,8 @@ public final class ThreadPoolDataset extends RawOp implements Operand<Tensor> {
   @Endpoint(describeByClass = true)
   public static ThreadPoolDataset create(Scope scope, Operand<?> inputDataset, Operand<?> threadPool, List<DataType<?>> outputTypes, List<Shape> outputShapes) {
     OperationBuilder opBuilder = scope.env().opBuilder("ExperimentalThreadPoolDataset", scope.makeOpName("ThreadPoolDataset"));
-    opBuilder.addInput(inputDataset.asOutput());
-    opBuilder.addInput(threadPool.asOutput());
+    opBuilder.addInput(inputDataset.asOutput(scope));
+    opBuilder.addInput(threadPool.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     DataType[] outputTypesArray = new DataType[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
@@ -72,8 +72,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<Tensor> asOutput() {
-    return (Output<Tensor>) handle;
+  public Output<TType> asOutput(Scope scope) {
+    return (Output<TType>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/ThreadPoolHandle.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/ThreadPoolHandle.java
index 1d43658188d..534c614c51b 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/ThreadPoolHandle.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/ThreadPoolHandle.java
@@ -21,16 +21,16 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Creates a dataset that uses a custom thread pool to compute `input_dataset`.
  */
-public final class ThreadPoolHandle extends RawOp implements Operand<Tensor> {
+public final class ThreadPoolHandle extends RawOp implements Operand<TType> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.data.experimental.ThreadPoolHandle}
@@ -135,8 +135,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<Tensor> asOutput() {
-    return (Output<Tensor>) handle;
+  public Output<TType> asOutput(Scope scope) {
+    return (Output<TType>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/UnbatchDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/UnbatchDataset.java
index 0d702b518ce..6a34aeef823 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/UnbatchDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/UnbatchDataset.java
@@ -23,17 +23,17 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * A dataset that splits the elements of its input into multiple elements.
  */
-public final class UnbatchDataset extends RawOp implements Operand<Tensor> {
+public final class UnbatchDataset extends RawOp implements Operand<TType> {
   
   /**
    * Factory method to create a class wrapping a new UnbatchDataset operation.
@@ -47,7 +47,7 @@ public final class UnbatchDataset extends RawOp implements Operand<Tensor> {
   @Endpoint(describeByClass = true)
   public static UnbatchDataset create(Scope scope, Operand<?> inputDataset, List<DataType<?>> outputTypes, List<Shape> outputShapes) {
     OperationBuilder opBuilder = scope.env().opBuilder("ExperimentalUnbatchDataset", scope.makeOpName("UnbatchDataset"));
-    opBuilder.addInput(inputDataset.asOutput());
+    opBuilder.addInput(inputDataset.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     DataType[] outputTypesArray = new DataType[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
@@ -70,8 +70,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<Tensor> asOutput() {
-    return (Output<Tensor>) handle;
+  public Output<TType> asOutput(Scope scope) {
+    return (Output<TType>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/UniqueDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/UniqueDataset.java
index ea635f72d8f..88c2928b20e 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/UniqueDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/UniqueDataset.java
@@ -23,17 +23,17 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Creates a dataset that contains the unique elements of `input_dataset`.
  */
-public final class UniqueDataset extends RawOp implements Operand<Tensor> {
+public final class UniqueDataset extends RawOp implements Operand<TType> {
   
   /**
    * Factory method to create a class wrapping a new UniqueDataset operation.
@@ -47,7 +47,7 @@ public final class UniqueDataset extends RawOp implements Operand<Tensor> {
   @Endpoint(describeByClass = true)
   public static UniqueDataset create(Scope scope, Operand<?> inputDataset, List<DataType<?>> outputTypes, List<Shape> outputShapes) {
     OperationBuilder opBuilder = scope.env().opBuilder("ExperimentalUniqueDataset", scope.makeOpName("UniqueDataset"));
-    opBuilder.addInput(inputDataset.asOutput());
+    opBuilder.addInput(inputDataset.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     DataType[] outputTypesArray = new DataType[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
@@ -70,8 +70,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<Tensor> asOutput() {
-    return (Output<Tensor>) handle;
+  public Output<TType> asOutput(Scope scope) {
+    return (Output<TType>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/debugging/CheckNumerics.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/debugging/CheckNumerics.java
index c14f3fd23d3..d6285c89f54 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/debugging/CheckNumerics.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/debugging/CheckNumerics.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -38,7 +37,7 @@
  * 
  * @param <T> data type for {@code output()} output
  */
-public final class CheckNumerics<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class CheckNumerics<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new CheckNumerics operation.
@@ -49,9 +48,9 @@ public final class CheckNumerics<T extends Tensor & TNumber> extends RawOp imple
    * @return a new instance of CheckNumerics
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> CheckNumerics<T> create(Scope scope, Operand<T> tensor, String message) {
+  public static <T extends TNumber> CheckNumerics<T> create(Scope scope, Operand<T> tensor, String message) {
     OperationBuilder opBuilder = scope.env().opBuilder("CheckNumericsV2", scope.makeOpName("CheckNumerics"));
-    opBuilder.addInput(tensor.asOutput());
+    opBuilder.addInput(tensor.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("message", message);
     return new CheckNumerics<T>(opBuilder.build());
@@ -64,7 +63,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/debugging/DebugGradientIdentity.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/debugging/DebugGradientIdentity.java
index 10b63b63587..b851b34d426 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/debugging/DebugGradientIdentity.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/debugging/DebugGradientIdentity.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Identity op for gradient debugging.
@@ -36,7 +36,7 @@
  * 
  * @param <T> data type for {@code output()} output
  */
-public final class DebugGradientIdentity<T extends Tensor> extends RawOp implements Operand<T> {
+public final class DebugGradientIdentity<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new DebugGradientIdentity operation.
@@ -46,9 +46,9 @@ public final class DebugGradientIdentity<T extends Tensor> extends RawOp impleme
    * @return a new instance of DebugGradientIdentity
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> DebugGradientIdentity<T> create(Scope scope, Operand<T> input) {
+  public static <T extends TType> DebugGradientIdentity<T> create(Scope scope, Operand<T> input) {
     OperationBuilder opBuilder = scope.env().opBuilder("DebugGradientIdentity", scope.makeOpName("DebugGradientIdentity"));
-    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new DebugGradientIdentity<T>(opBuilder.build());
   }
@@ -60,7 +60,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/debugging/DebugGradientRefIdentity.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/debugging/DebugGradientRefIdentity.java
index 220ce202ab2..035ced72d8b 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/debugging/DebugGradientRefIdentity.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/debugging/DebugGradientRefIdentity.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Identity op for gradient debugging.
@@ -36,7 +36,7 @@
  * 
  * @param <T> data type for {@code output()} output
  */
-public final class DebugGradientRefIdentity<T extends Tensor> extends RawOp implements Operand<T> {
+public final class DebugGradientRefIdentity<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new DebugGradientRefIdentity operation.
@@ -46,9 +46,9 @@ public final class DebugGradientRefIdentity<T extends Tensor> extends RawOp impl
    * @return a new instance of DebugGradientRefIdentity
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> DebugGradientRefIdentity<T> create(Scope scope, Operand<T> input) {
+  public static <T extends TType> DebugGradientRefIdentity<T> create(Scope scope, Operand<T> input) {
     OperationBuilder opBuilder = scope.env().opBuilder("DebugGradientRefIdentity", scope.makeOpName("DebugGradientRefIdentity"));
-    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new DebugGradientRefIdentity<T>(opBuilder.build());
   }
@@ -60,7 +60,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/debugging/DebugIdentity.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/debugging/DebugIdentity.java
index 936d17bb297..4f075a98f6e 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/debugging/DebugIdentity.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/debugging/DebugIdentity.java
@@ -22,11 +22,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Debug Identity V2 Op.
@@ -43,7 +43,7 @@
  * 
  * @param <T> data type for {@code output()} output
  */
-public final class DebugIdentity<T extends Tensor> extends RawOp implements Operand<T> {
+public final class DebugIdentity<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.debugging.DebugIdentity}
@@ -130,9 +130,9 @@ private Options() {
    * @return a new instance of DebugIdentity
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> DebugIdentity<T> create(Scope scope, Operand<T> input, Options... options) {
+  public static <T extends TType> DebugIdentity<T> create(Scope scope, Operand<T> input, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("DebugIdentityV2", scope.makeOpName("DebugIdentity"));
-    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -225,7 +225,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/debugging/DebugNanCount.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/debugging/DebugNanCount.java
index ff889ac64a0..0c4666b1818 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/debugging/DebugNanCount.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/debugging/DebugNanCount.java
@@ -22,12 +22,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
 
 /**
  * Debug NaN Value Counter Op.
@@ -97,9 +97,9 @@ private Options() {
    * @return a new instance of DebugNanCount
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> DebugNanCount create(Scope scope, Operand<T> input, Options... options) {
+  public static <T extends TType> DebugNanCount create(Scope scope, Operand<T> input, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("DebugNanCount", scope.makeOpName("DebugNanCount"));
-    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -165,7 +165,7 @@ public Output<TInt64> output() {
   }
   
   @Override
-  public Output<TInt64> asOutput() {
+  public Output<TInt64> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/debugging/DebugNumericsSummary.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/debugging/DebugNumericsSummary.java
index 1afff31e562..e30b0390eb2 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/debugging/DebugNumericsSummary.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/debugging/DebugNumericsSummary.java
@@ -22,13 +22,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TFloat32;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * Debug Numeric Summary V2 Op.
@@ -39,7 +39,7 @@
  * 
  * @param <U> data type for {@code output()} output
  */
-public final class DebugNumericsSummary<U extends Tensor & TNumber> extends RawOp implements Operand<U> {
+public final class DebugNumericsSummary<U extends TNumber> extends RawOp implements Operand<U> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.debugging.DebugNumericsSummary}
@@ -132,9 +132,9 @@ private Options() {
    * @return a new instance of DebugNumericsSummary
    */
   @Endpoint(describeByClass = true)
-  public static <U extends Tensor & TNumber, T extends Tensor> DebugNumericsSummary<U> create(Scope scope, Operand<T> input, DataType<U> outputDtype, Options... options) {
+  public static <U extends TNumber, T extends TType> DebugNumericsSummary<U> create(Scope scope, Operand<T> input, DataType<U> outputDtype, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("DebugNumericSummaryV2", scope.makeOpName("DebugNumericsSummary"));
-    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("output_dtype", outputDtype);
     if (options != null) {
@@ -159,7 +159,7 @@ public static <U extends Tensor & TNumber, T extends Tensor> DebugNumericsSummar
    * @return a new instance of DebugNumericsSummary
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> DebugNumericsSummary<TFloat32> create(Scope scope, Operand<T> input, Options... options) {
+  public static <T extends TType> DebugNumericsSummary<TFloat32> create(Scope scope, Operand<T> input, Options... options) {
     return create(scope, input, TFloat32.DTYPE, options);
   }
   
@@ -237,7 +237,7 @@ public Output<U> output() {
   }
   
   @Override
-  public Output<U> asOutput() {
+  public Output<U> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/dtypes/AsString.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/dtypes/AsString.java
index 3661a58f6a4..17d4f15222b 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/dtypes/AsString.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/dtypes/AsString.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TString;
+import org.tensorflow.types.family.TType;
 
 /**
  * Converts each entry in the given tensor to strings.
@@ -116,9 +116,9 @@ private Options() {
    * @return a new instance of AsString
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> AsString create(Scope scope, Operand<T> input, Options... options) {
+  public static <T extends TType> AsString create(Scope scope, Operand<T> input, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("AsString", scope.makeOpName("AsString"));
-    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -189,7 +189,7 @@ public Output<TString> output() {
   }
   
   @Override
-  public Output<TString> asOutput() {
+  public Output<TString> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/dtypes/Cast.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/dtypes/Cast.java
index d805dcb9783..2f3400bfd3b 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/dtypes/Cast.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/dtypes/Cast.java
@@ -22,11 +22,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Cast x of type SrcT to y of DstT.
@@ -34,7 +34,7 @@
  * @param <U> data type for {@code y()} output
  */
 @Operator(group = "dtypes")
-public final class Cast<U extends Tensor> extends RawOp implements Operand<U> {
+public final class Cast<U extends TType> extends RawOp implements Operand<U> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.dtypes.Cast}
@@ -65,9 +65,9 @@ private Options() {
    * @return a new instance of Cast
    */
   @Endpoint(describeByClass = true)
-  public static <U extends Tensor, T extends Tensor> Cast<U> create(Scope scope, Operand<T> x, DataType<U> DstT, Options... options) {
+  public static <U extends TType, T extends TType> Cast<U> create(Scope scope, Operand<T> x, DataType<U> DstT, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("Cast", scope.makeOpName("Cast"));
-    opBuilder.addInput(x.asOutput());
+    opBuilder.addInput(x.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("DstT", DstT);
     if (options != null) {
@@ -94,7 +94,7 @@ public Output<U> y() {
   }
   
   @Override
-  public Output<U> asOutput() {
+  public Output<U> asOutput(Scope scope) {
     return y;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/dtypes/Complex.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/dtypes/Complex.java
index fcf562a48dc..a9711798846 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/dtypes/Complex.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/dtypes/Complex.java
@@ -22,12 +22,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * Converts two real numbers to a complex number.
@@ -50,7 +50,7 @@
  * @param <U> data type for {@code out()} output
  */
 @Operator(group = "dtypes")
-public final class Complex<U extends Tensor> extends RawOp implements Operand<U> {
+public final class Complex<U extends TType> extends RawOp implements Operand<U> {
   
   /**
    * Factory method to create a class wrapping a new Complex operation.
@@ -62,10 +62,10 @@ public final class Complex<U extends Tensor> extends RawOp implements Operand<U>
    * @return a new instance of Complex
    */
   @Endpoint(describeByClass = true)
-  public static <U extends Tensor, T extends Tensor & TNumber> Complex<U> create(Scope scope, Operand<T> real, Operand<T> imag, DataType<U> Tout) {
+  public static <U extends TType, T extends TNumber> Complex<U> create(Scope scope, Operand<T> real, Operand<T> imag, DataType<U> Tout) {
     OperationBuilder opBuilder = scope.env().opBuilder("Complex", scope.makeOpName("Complex"));
-    opBuilder.addInput(real.asOutput());
-    opBuilder.addInput(imag.asOutput());
+    opBuilder.addInput(real.asOutput(scope));
+    opBuilder.addInput(imag.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("Tout", Tout);
     return new Complex<U>(opBuilder.build());
@@ -78,7 +78,7 @@ public Output<U> out() {
   }
   
   @Override
-  public Output<U> asOutput() {
+  public Output<U> asOutput(Scope scope) {
     return out;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/dtypes/ToBool.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/dtypes/ToBool.java
index 05987505418..693100c20da 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/dtypes/ToBool.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/dtypes/ToBool.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TBool;
+import org.tensorflow.types.family.TType;
 
 /**
  * Converts a tensor to a scalar predicate.
@@ -54,9 +54,9 @@ public final class ToBool extends RawOp implements Operand<TBool> {
    * @return a new instance of ToBool
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> ToBool create(Scope scope, Operand<T> input) {
+  public static <T extends TType> ToBool create(Scope scope, Operand<T> input) {
     OperationBuilder opBuilder = scope.env().opBuilder("ToBool", scope.makeOpName("ToBool"));
-    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new ToBool(opBuilder.build());
   }
@@ -68,7 +68,7 @@ public Output<TBool> output() {
   }
   
   @Override
-  public Output<TBool> asOutput() {
+  public Output<TBool> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesAggregateStats.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesAggregateStats.java
index ac772ea4ae0..03ade5c79e1 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesAggregateStats.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesAggregateStats.java
@@ -50,10 +50,10 @@ public final class BoostedTreesAggregateStats extends RawOp implements Operand<T
   @Endpoint(describeByClass = true)
   public static BoostedTreesAggregateStats create(Scope scope, Operand<TInt32> nodeIds, Operand<TFloat32> gradients, Operand<TFloat32> hessians, Operand<TInt32> feature, Long maxSplits, Long numBuckets) {
     OperationBuilder opBuilder = scope.env().opBuilder("BoostedTreesAggregateStats", scope.makeOpName("BoostedTreesAggregateStats"));
-    opBuilder.addInput(nodeIds.asOutput());
-    opBuilder.addInput(gradients.asOutput());
-    opBuilder.addInput(hessians.asOutput());
-    opBuilder.addInput(feature.asOutput());
+    opBuilder.addInput(nodeIds.asOutput(scope));
+    opBuilder.addInput(gradients.asOutput(scope));
+    opBuilder.addInput(hessians.asOutput(scope));
+    opBuilder.addInput(feature.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("max_splits", maxSplits);
     opBuilder.setAttr("num_buckets", numBuckets);
@@ -69,7 +69,7 @@ public Output<TFloat32> statsSummary() {
   }
   
   @Override
-  public Output<TFloat32> asOutput() {
+  public Output<TFloat32> asOutput(Scope scope) {
     return statsSummary;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesBucketize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesBucketize.java
index 6d57706c19e..f13498bd012 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesBucketize.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesBucketize.java
@@ -52,8 +52,8 @@ public final class BoostedTreesBucketize extends RawOp implements Iterable<Opera
   @Endpoint(describeByClass = true)
   public static BoostedTreesBucketize create(Scope scope, Iterable<Operand<TFloat32>> floatValues, Iterable<Operand<TFloat32>> bucketBoundaries) {
     OperationBuilder opBuilder = scope.env().opBuilder("BoostedTreesBucketize", scope.makeOpName("BoostedTreesBucketize"));
-    opBuilder.addInputList(Operands.asOutputs(floatValues));
-    opBuilder.addInputList(Operands.asOutputs(bucketBoundaries));
+    opBuilder.addInputList(Operands.asOutputs(scope, floatValues));
+    opBuilder.addInputList(Operands.asOutputs(scope, bucketBoundaries));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new BoostedTreesBucketize(opBuilder.build());
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesCalculateBestFeatureSplit.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesCalculateBestFeatureSplit.java
index 41f1d3a6910..1e6fe616a4f 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesCalculateBestFeatureSplit.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesCalculateBestFeatureSplit.java
@@ -79,12 +79,12 @@ private Options() {
   @Endpoint(describeByClass = true)
   public static BoostedTreesCalculateBestFeatureSplit create(Scope scope, Operand<TInt32> nodeIdRange, Operand<TFloat32> statsSummary, Operand<TFloat32> l1, Operand<TFloat32> l2, Operand<TFloat32> treeComplexity, Operand<TFloat32> minNodeWeight, Long logitsDimension, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("BoostedTreesCalculateBestFeatureSplit", scope.makeOpName("BoostedTreesCalculateBestFeatureSplit"));
-    opBuilder.addInput(nodeIdRange.asOutput());
-    opBuilder.addInput(statsSummary.asOutput());
-    opBuilder.addInput(l1.asOutput());
-    opBuilder.addInput(l2.asOutput());
-    opBuilder.addInput(treeComplexity.asOutput());
-    opBuilder.addInput(minNodeWeight.asOutput());
+    opBuilder.addInput(nodeIdRange.asOutput(scope));
+    opBuilder.addInput(statsSummary.asOutput(scope));
+    opBuilder.addInput(l1.asOutput(scope));
+    opBuilder.addInput(l2.asOutput(scope));
+    opBuilder.addInput(treeComplexity.asOutput(scope));
+    opBuilder.addInput(minNodeWeight.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("logits_dimension", logitsDimension);
     if (options != null) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesCalculateBestFeatureSplitV2.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesCalculateBestFeatureSplitV2.java
index 94f4afa4386..85d6580d293 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesCalculateBestFeatureSplitV2.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesCalculateBestFeatureSplitV2.java
@@ -62,14 +62,14 @@ public final class BoostedTreesCalculateBestFeatureSplitV2 extends RawOp {
   @Endpoint(describeByClass = true)
   public static BoostedTreesCalculateBestFeatureSplitV2 create(Scope scope, Operand<TInt32> nodeIdRange, Iterable<Operand<TFloat32>> statsSummariesList, Operand<TString> splitTypes, Operand<TInt32> candidateFeatureIds, Operand<TFloat32> l1, Operand<TFloat32> l2, Operand<TFloat32> treeComplexity, Operand<TFloat32> minNodeWeight, Long logitsDimension) {
     OperationBuilder opBuilder = scope.env().opBuilder("BoostedTreesCalculateBestFeatureSplitV2", scope.makeOpName("BoostedTreesCalculateBestFeatureSplitV2"));
-    opBuilder.addInput(nodeIdRange.asOutput());
-    opBuilder.addInputList(Operands.asOutputs(statsSummariesList));
-    opBuilder.addInput(splitTypes.asOutput());
-    opBuilder.addInput(candidateFeatureIds.asOutput());
-    opBuilder.addInput(l1.asOutput());
-    opBuilder.addInput(l2.asOutput());
-    opBuilder.addInput(treeComplexity.asOutput());
-    opBuilder.addInput(minNodeWeight.asOutput());
+    opBuilder.addInput(nodeIdRange.asOutput(scope));
+    opBuilder.addInputList(Operands.asOutputs(scope, statsSummariesList));
+    opBuilder.addInput(splitTypes.asOutput(scope));
+    opBuilder.addInput(candidateFeatureIds.asOutput(scope));
+    opBuilder.addInput(l1.asOutput(scope));
+    opBuilder.addInput(l2.asOutput(scope));
+    opBuilder.addInput(treeComplexity.asOutput(scope));
+    opBuilder.addInput(minNodeWeight.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("logits_dimension", logitsDimension);
     return new BoostedTreesCalculateBestFeatureSplitV2(opBuilder.build());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesCalculateBestGainsPerFeature.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesCalculateBestGainsPerFeature.java
index daade3457ae..5b4efb8d789 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesCalculateBestGainsPerFeature.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesCalculateBestGainsPerFeature.java
@@ -61,12 +61,12 @@ public final class BoostedTreesCalculateBestGainsPerFeature extends RawOp {
   @Endpoint(describeByClass = true)
   public static BoostedTreesCalculateBestGainsPerFeature create(Scope scope, Operand<TInt32> nodeIdRange, Iterable<Operand<TFloat32>> statsSummaryList, Operand<TFloat32> l1, Operand<TFloat32> l2, Operand<TFloat32> treeComplexity, Operand<TFloat32> minNodeWeight, Long maxSplits) {
     OperationBuilder opBuilder = scope.env().opBuilder("BoostedTreesCalculateBestGainsPerFeature", scope.makeOpName("BoostedTreesCalculateBestGainsPerFeature"));
-    opBuilder.addInput(nodeIdRange.asOutput());
-    opBuilder.addInputList(Operands.asOutputs(statsSummaryList));
-    opBuilder.addInput(l1.asOutput());
-    opBuilder.addInput(l2.asOutput());
-    opBuilder.addInput(treeComplexity.asOutput());
-    opBuilder.addInput(minNodeWeight.asOutput());
+    opBuilder.addInput(nodeIdRange.asOutput(scope));
+    opBuilder.addInputList(Operands.asOutputs(scope, statsSummaryList));
+    opBuilder.addInput(l1.asOutput(scope));
+    opBuilder.addInput(l2.asOutput(scope));
+    opBuilder.addInput(treeComplexity.asOutput(scope));
+    opBuilder.addInput(minNodeWeight.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("max_splits", maxSplits);
     return new BoostedTreesCalculateBestGainsPerFeature(opBuilder.build());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesCenterBias.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesCenterBias.java
index 7359b6039e2..9bea51c509c 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesCenterBias.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesCenterBias.java
@@ -47,11 +47,11 @@ public final class BoostedTreesCenterBias extends RawOp implements Operand<TBool
   @Endpoint(describeByClass = true)
   public static BoostedTreesCenterBias create(Scope scope, Operand<?> treeEnsembleHandle, Operand<TFloat32> meanGradients, Operand<TFloat32> meanHessians, Operand<TFloat32> l1, Operand<TFloat32> l2) {
     OperationBuilder opBuilder = scope.env().opBuilder("BoostedTreesCenterBias", scope.makeOpName("BoostedTreesCenterBias"));
-    opBuilder.addInput(treeEnsembleHandle.asOutput());
-    opBuilder.addInput(meanGradients.asOutput());
-    opBuilder.addInput(meanHessians.asOutput());
-    opBuilder.addInput(l1.asOutput());
-    opBuilder.addInput(l2.asOutput());
+    opBuilder.addInput(treeEnsembleHandle.asOutput(scope));
+    opBuilder.addInput(meanGradients.asOutput(scope));
+    opBuilder.addInput(meanHessians.asOutput(scope));
+    opBuilder.addInput(l1.asOutput(scope));
+    opBuilder.addInput(l2.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new BoostedTreesCenterBias(opBuilder.build());
   }
@@ -64,7 +64,7 @@ public Output<TBool> continueCentering() {
   }
   
   @Override
-  public Output<TBool> asOutput() {
+  public Output<TBool> asOutput(Scope scope) {
     return continueCentering;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesCreateEnsemble.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesCreateEnsemble.java
index 8841988b36d..b7d2ff14a67 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesCreateEnsemble.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesCreateEnsemble.java
@@ -44,9 +44,9 @@ public final class BoostedTreesCreateEnsemble extends RawOp {
   @Endpoint(describeByClass = true)
   public static BoostedTreesCreateEnsemble create(Scope scope, Operand<?> treeEnsembleHandle, Operand<TInt64> stampToken, Operand<TString> treeEnsembleSerialized) {
     OperationBuilder opBuilder = scope.env().opBuilder("BoostedTreesCreateEnsemble", scope.makeOpName("BoostedTreesCreateEnsemble"));
-    opBuilder.addInput(treeEnsembleHandle.asOutput());
-    opBuilder.addInput(stampToken.asOutput());
-    opBuilder.addInput(treeEnsembleSerialized.asOutput());
+    opBuilder.addInput(treeEnsembleHandle.asOutput(scope));
+    opBuilder.addInput(stampToken.asOutput(scope));
+    opBuilder.addInput(treeEnsembleSerialized.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new BoostedTreesCreateEnsemble(opBuilder.build());
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesCreateQuantileStreamResource.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesCreateQuantileStreamResource.java
index 802a61ecb2a..698f8356d53 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesCreateQuantileStreamResource.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesCreateQuantileStreamResource.java
@@ -64,9 +64,9 @@ private Options() {
   @Endpoint(describeByClass = true)
   public static BoostedTreesCreateQuantileStreamResource create(Scope scope, Operand<?> quantileStreamResourceHandle, Operand<TFloat32> epsilon, Operand<TInt64> numStreams, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("BoostedTreesCreateQuantileStreamResource", scope.makeOpName("BoostedTreesCreateQuantileStreamResource"));
-    opBuilder.addInput(quantileStreamResourceHandle.asOutput());
-    opBuilder.addInput(epsilon.asOutput());
-    opBuilder.addInput(numStreams.asOutput());
+    opBuilder.addInput(quantileStreamResourceHandle.asOutput(scope));
+    opBuilder.addInput(epsilon.asOutput(scope));
+    opBuilder.addInput(numStreams.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesDeserializeEnsemble.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesDeserializeEnsemble.java
index 15371fb4df9..a9b02360384 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesDeserializeEnsemble.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesDeserializeEnsemble.java
@@ -46,9 +46,9 @@ public final class BoostedTreesDeserializeEnsemble extends RawOp {
   @Endpoint(describeByClass = true)
   public static BoostedTreesDeserializeEnsemble create(Scope scope, Operand<?> treeEnsembleHandle, Operand<TInt64> stampToken, Operand<TString> treeEnsembleSerialized) {
     OperationBuilder opBuilder = scope.env().opBuilder("BoostedTreesDeserializeEnsemble", scope.makeOpName("BoostedTreesDeserializeEnsemble"));
-    opBuilder.addInput(treeEnsembleHandle.asOutput());
-    opBuilder.addInput(stampToken.asOutput());
-    opBuilder.addInput(treeEnsembleSerialized.asOutput());
+    opBuilder.addInput(treeEnsembleHandle.asOutput(scope));
+    opBuilder.addInput(stampToken.asOutput(scope));
+    opBuilder.addInput(treeEnsembleSerialized.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new BoostedTreesDeserializeEnsemble(opBuilder.build());
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesEnsembleResourceHandleOp.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesEnsembleResourceHandleOp.java
index de542108127..e2067f08a5b 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesEnsembleResourceHandleOp.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesEnsembleResourceHandleOp.java
@@ -21,16 +21,16 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Creates a handle to a BoostedTreesEnsembleResource
  */
-public final class BoostedTreesEnsembleResourceHandleOp extends RawOp implements Operand<Tensor> {
+public final class BoostedTreesEnsembleResourceHandleOp extends RawOp implements Operand<TType> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.estimator.BoostedTreesEnsembleResourceHandleOp}
@@ -106,8 +106,8 @@ public Output<?> resource() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<Tensor> asOutput() {
-    return (Output<Tensor>) resource;
+  public Output<TType> asOutput(Scope scope) {
+    return (Output<TType>) resource;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesExampleDebugOutputs.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesExampleDebugOutputs.java
index 8618f401ce4..058430e3315 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesExampleDebugOutputs.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesExampleDebugOutputs.java
@@ -52,8 +52,8 @@ public final class BoostedTreesExampleDebugOutputs extends RawOp implements Oper
   @Endpoint(describeByClass = true)
   public static BoostedTreesExampleDebugOutputs create(Scope scope, Operand<?> treeEnsembleHandle, Iterable<Operand<TInt32>> bucketizedFeatures, Long logitsDimension) {
     OperationBuilder opBuilder = scope.env().opBuilder("BoostedTreesExampleDebugOutputs", scope.makeOpName("BoostedTreesExampleDebugOutputs"));
-    opBuilder.addInput(treeEnsembleHandle.asOutput());
-    opBuilder.addInputList(Operands.asOutputs(bucketizedFeatures));
+    opBuilder.addInput(treeEnsembleHandle.asOutput(scope));
+    opBuilder.addInputList(Operands.asOutputs(scope, bucketizedFeatures));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("logits_dimension", logitsDimension);
     return new BoostedTreesExampleDebugOutputs(opBuilder.build());
@@ -67,7 +67,7 @@ public Output<TString> examplesDebugOutputsSerialized() {
   }
   
   @Override
-  public Output<TString> asOutput() {
+  public Output<TString> asOutput(Scope scope) {
     return examplesDebugOutputsSerialized;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesFlushQuantileSummaries.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesFlushQuantileSummaries.java
index d22acc2ca1b..f3e6381a923 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesFlushQuantileSummaries.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesFlushQuantileSummaries.java
@@ -50,7 +50,7 @@ public final class BoostedTreesFlushQuantileSummaries extends RawOp implements I
   @Endpoint(describeByClass = true)
   public static BoostedTreesFlushQuantileSummaries create(Scope scope, Operand<?> quantileStreamResourceHandle, Long numFeatures) {
     OperationBuilder opBuilder = scope.env().opBuilder("BoostedTreesFlushQuantileSummaries", scope.makeOpName("BoostedTreesFlushQuantileSummaries"));
-    opBuilder.addInput(quantileStreamResourceHandle.asOutput());
+    opBuilder.addInput(quantileStreamResourceHandle.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("num_features", numFeatures);
     return new BoostedTreesFlushQuantileSummaries(opBuilder.build());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesGetEnsembleStates.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesGetEnsembleStates.java
index 36e8eb3c4be..deba0d48ee9 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesGetEnsembleStates.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesGetEnsembleStates.java
@@ -43,7 +43,7 @@ public final class BoostedTreesGetEnsembleStates extends RawOp {
   @Endpoint(describeByClass = true)
   public static BoostedTreesGetEnsembleStates create(Scope scope, Operand<?> treeEnsembleHandle) {
     OperationBuilder opBuilder = scope.env().opBuilder("BoostedTreesGetEnsembleStates", scope.makeOpName("BoostedTreesGetEnsembleStates"));
-    opBuilder.addInput(treeEnsembleHandle.asOutput());
+    opBuilder.addInput(treeEnsembleHandle.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new BoostedTreesGetEnsembleStates(opBuilder.build());
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesMakeQuantileSummaries.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesMakeQuantileSummaries.java
index cf4156bb2ce..cbf32069e02 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesMakeQuantileSummaries.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesMakeQuantileSummaries.java
@@ -51,9 +51,9 @@ public final class BoostedTreesMakeQuantileSummaries extends RawOp implements It
   @Endpoint(describeByClass = true)
   public static BoostedTreesMakeQuantileSummaries create(Scope scope, Iterable<Operand<TFloat32>> floatValues, Operand<TFloat32> exampleWeights, Operand<TFloat32> epsilon) {
     OperationBuilder opBuilder = scope.env().opBuilder("BoostedTreesMakeQuantileSummaries", scope.makeOpName("BoostedTreesMakeQuantileSummaries"));
-    opBuilder.addInputList(Operands.asOutputs(floatValues));
-    opBuilder.addInput(exampleWeights.asOutput());
-    opBuilder.addInput(epsilon.asOutput());
+    opBuilder.addInputList(Operands.asOutputs(scope, floatValues));
+    opBuilder.addInput(exampleWeights.asOutput(scope));
+    opBuilder.addInput(epsilon.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new BoostedTreesMakeQuantileSummaries(opBuilder.build());
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesMakeStatsSummary.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesMakeStatsSummary.java
index 626e05a6f54..6b73fd4903d 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesMakeStatsSummary.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesMakeStatsSummary.java
@@ -51,10 +51,10 @@ public final class BoostedTreesMakeStatsSummary extends RawOp implements Operand
   @Endpoint(describeByClass = true)
   public static BoostedTreesMakeStatsSummary create(Scope scope, Operand<TInt32> nodeIds, Operand<TFloat32> gradients, Operand<TFloat32> hessians, Iterable<Operand<TInt32>> bucketizedFeaturesList, Long maxSplits, Long numBuckets) {
     OperationBuilder opBuilder = scope.env().opBuilder("BoostedTreesMakeStatsSummary", scope.makeOpName("BoostedTreesMakeStatsSummary"));
-    opBuilder.addInput(nodeIds.asOutput());
-    opBuilder.addInput(gradients.asOutput());
-    opBuilder.addInput(hessians.asOutput());
-    opBuilder.addInputList(Operands.asOutputs(bucketizedFeaturesList));
+    opBuilder.addInput(nodeIds.asOutput(scope));
+    opBuilder.addInput(gradients.asOutput(scope));
+    opBuilder.addInput(hessians.asOutput(scope));
+    opBuilder.addInputList(Operands.asOutputs(scope, bucketizedFeaturesList));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("max_splits", maxSplits);
     opBuilder.setAttr("num_buckets", numBuckets);
@@ -69,7 +69,7 @@ public Output<TFloat32> statsSummary() {
   }
   
   @Override
-  public Output<TFloat32> asOutput() {
+  public Output<TFloat32> asOutput(Scope scope) {
     return statsSummary;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesPredict.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesPredict.java
index e9c7d4a2be1..f5bdd283ae9 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesPredict.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesPredict.java
@@ -51,8 +51,8 @@ public final class BoostedTreesPredict extends RawOp implements Operand<TFloat32
   @Endpoint(describeByClass = true)
   public static BoostedTreesPredict create(Scope scope, Operand<?> treeEnsembleHandle, Iterable<Operand<TInt32>> bucketizedFeatures, Long logitsDimension) {
     OperationBuilder opBuilder = scope.env().opBuilder("BoostedTreesPredict", scope.makeOpName("BoostedTreesPredict"));
-    opBuilder.addInput(treeEnsembleHandle.asOutput());
-    opBuilder.addInputList(Operands.asOutputs(bucketizedFeatures));
+    opBuilder.addInput(treeEnsembleHandle.asOutput(scope));
+    opBuilder.addInputList(Operands.asOutputs(scope, bucketizedFeatures));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("logits_dimension", logitsDimension);
     return new BoostedTreesPredict(opBuilder.build());
@@ -66,7 +66,7 @@ public Output<TFloat32> logits() {
   }
   
   @Override
-  public Output<TFloat32> asOutput() {
+  public Output<TFloat32> asOutput(Scope scope) {
     return logits;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesQuantileStreamResourceAddSummaries.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesQuantileStreamResourceAddSummaries.java
index 418ff3b2ff6..140bcc8f951 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesQuantileStreamResourceAddSummaries.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesQuantileStreamResourceAddSummaries.java
@@ -47,8 +47,8 @@ public final class BoostedTreesQuantileStreamResourceAddSummaries extends RawOp
   @Endpoint(describeByClass = true)
   public static BoostedTreesQuantileStreamResourceAddSummaries create(Scope scope, Operand<?> quantileStreamResourceHandle, Iterable<Operand<TFloat32>> summaries) {
     OperationBuilder opBuilder = scope.env().opBuilder("BoostedTreesQuantileStreamResourceAddSummaries", scope.makeOpName("BoostedTreesQuantileStreamResourceAddSummaries"));
-    opBuilder.addInput(quantileStreamResourceHandle.asOutput());
-    opBuilder.addInputList(Operands.asOutputs(summaries));
+    opBuilder.addInput(quantileStreamResourceHandle.asOutput(scope));
+    opBuilder.addInputList(Operands.asOutputs(scope, summaries));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new BoostedTreesQuantileStreamResourceAddSummaries(opBuilder.build());
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesQuantileStreamResourceDeserialize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesQuantileStreamResourceDeserialize.java
index 6efb58ed60c..11073dec531 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesQuantileStreamResourceDeserialize.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesQuantileStreamResourceDeserialize.java
@@ -45,8 +45,8 @@ public final class BoostedTreesQuantileStreamResourceDeserialize extends RawOp {
   @Endpoint(describeByClass = true)
   public static BoostedTreesQuantileStreamResourceDeserialize create(Scope scope, Operand<?> quantileStreamResourceHandle, Iterable<Operand<TFloat32>> bucketBoundaries) {
     OperationBuilder opBuilder = scope.env().opBuilder("BoostedTreesQuantileStreamResourceDeserialize", scope.makeOpName("BoostedTreesQuantileStreamResourceDeserialize"));
-    opBuilder.addInput(quantileStreamResourceHandle.asOutput());
-    opBuilder.addInputList(Operands.asOutputs(bucketBoundaries));
+    opBuilder.addInput(quantileStreamResourceHandle.asOutput(scope));
+    opBuilder.addInputList(Operands.asOutputs(scope, bucketBoundaries));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new BoostedTreesQuantileStreamResourceDeserialize(opBuilder.build());
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesQuantileStreamResourceFlush.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesQuantileStreamResourceFlush.java
index cc10434a582..64163d59d9a 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesQuantileStreamResourceFlush.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesQuantileStreamResourceFlush.java
@@ -69,8 +69,8 @@ private Options() {
   @Endpoint(describeByClass = true)
   public static BoostedTreesQuantileStreamResourceFlush create(Scope scope, Operand<?> quantileStreamResourceHandle, Operand<TInt64> numBuckets, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("BoostedTreesQuantileStreamResourceFlush", scope.makeOpName("BoostedTreesQuantileStreamResourceFlush"));
-    opBuilder.addInput(quantileStreamResourceHandle.asOutput());
-    opBuilder.addInput(numBuckets.asOutput());
+    opBuilder.addInput(quantileStreamResourceHandle.asOutput(scope));
+    opBuilder.addInput(numBuckets.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesQuantileStreamResourceGetBucketBoundaries.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesQuantileStreamResourceGetBucketBoundaries.java
index 5d421152254..ac6be6f10c5 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesQuantileStreamResourceGetBucketBoundaries.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesQuantileStreamResourceGetBucketBoundaries.java
@@ -49,7 +49,7 @@ public final class BoostedTreesQuantileStreamResourceGetBucketBoundaries extends
   @Endpoint(describeByClass = true)
   public static BoostedTreesQuantileStreamResourceGetBucketBoundaries create(Scope scope, Operand<?> quantileStreamResourceHandle, Long numFeatures) {
     OperationBuilder opBuilder = scope.env().opBuilder("BoostedTreesQuantileStreamResourceGetBucketBoundaries", scope.makeOpName("BoostedTreesQuantileStreamResourceGetBucketBoundaries"));
-    opBuilder.addInput(quantileStreamResourceHandle.asOutput());
+    opBuilder.addInput(quantileStreamResourceHandle.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("num_features", numFeatures);
     return new BoostedTreesQuantileStreamResourceGetBucketBoundaries(opBuilder.build());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesQuantileStreamResourceHandleOp.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesQuantileStreamResourceHandleOp.java
index ebf3285a95f..20ee0ddcb71 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesQuantileStreamResourceHandleOp.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesQuantileStreamResourceHandleOp.java
@@ -21,16 +21,16 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Creates a handle to a BoostedTreesQuantileStreamResource.
  */
-public final class BoostedTreesQuantileStreamResourceHandleOp extends RawOp implements Operand<Tensor> {
+public final class BoostedTreesQuantileStreamResourceHandleOp extends RawOp implements Operand<TType> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.estimator.BoostedTreesQuantileStreamResourceHandleOp}
@@ -106,8 +106,8 @@ public Output<?> resource() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<Tensor> asOutput() {
-    return (Output<Tensor>) resource;
+  public Output<TType> asOutput(Scope scope) {
+    return (Output<TType>) resource;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesSerializeEnsemble.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesSerializeEnsemble.java
index 287c9e9054a..4755654dee8 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesSerializeEnsemble.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesSerializeEnsemble.java
@@ -43,7 +43,7 @@ public final class BoostedTreesSerializeEnsemble extends RawOp {
   @Endpoint(describeByClass = true)
   public static BoostedTreesSerializeEnsemble create(Scope scope, Operand<?> treeEnsembleHandle) {
     OperationBuilder opBuilder = scope.env().opBuilder("BoostedTreesSerializeEnsemble", scope.makeOpName("BoostedTreesSerializeEnsemble"));
-    opBuilder.addInput(treeEnsembleHandle.asOutput());
+    opBuilder.addInput(treeEnsembleHandle.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new BoostedTreesSerializeEnsemble(opBuilder.build());
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesSparseAggregateStats.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesSparseAggregateStats.java
index 3923ac0e905..b8e0ba4c030 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesSparseAggregateStats.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesSparseAggregateStats.java
@@ -58,12 +58,12 @@ public final class BoostedTreesSparseAggregateStats extends RawOp {
   @Endpoint(describeByClass = true)
   public static BoostedTreesSparseAggregateStats create(Scope scope, Operand<TInt32> nodeIds, Operand<TFloat32> gradients, Operand<TFloat32> hessians, Operand<TInt32> featureIndices, Operand<TInt32> featureValues, Operand<TInt32> featureShape, Long maxSplits, Long numBuckets) {
     OperationBuilder opBuilder = scope.env().opBuilder("BoostedTreesSparseAggregateStats", scope.makeOpName("BoostedTreesSparseAggregateStats"));
-    opBuilder.addInput(nodeIds.asOutput());
-    opBuilder.addInput(gradients.asOutput());
-    opBuilder.addInput(hessians.asOutput());
-    opBuilder.addInput(featureIndices.asOutput());
-    opBuilder.addInput(featureValues.asOutput());
-    opBuilder.addInput(featureShape.asOutput());
+    opBuilder.addInput(nodeIds.asOutput(scope));
+    opBuilder.addInput(gradients.asOutput(scope));
+    opBuilder.addInput(hessians.asOutput(scope));
+    opBuilder.addInput(featureIndices.asOutput(scope));
+    opBuilder.addInput(featureValues.asOutput(scope));
+    opBuilder.addInput(featureShape.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("max_splits", maxSplits);
     opBuilder.setAttr("num_buckets", numBuckets);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesSparseCalculateBestFeatureSplit.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesSparseCalculateBestFeatureSplit.java
index b2f2dd1eebf..3d20166d9cf 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesSparseCalculateBestFeatureSplit.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesSparseCalculateBestFeatureSplit.java
@@ -81,14 +81,14 @@ private Options() {
   @Endpoint(describeByClass = true)
   public static BoostedTreesSparseCalculateBestFeatureSplit create(Scope scope, Operand<TInt32> nodeIdRange, Operand<TInt32> statsSummaryIndices, Operand<TFloat32> statsSummaryValues, Operand<TInt32> statsSummaryShape, Operand<TFloat32> l1, Operand<TFloat32> l2, Operand<TFloat32> treeComplexity, Operand<TFloat32> minNodeWeight, Long logitsDimension, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("BoostedTreesSparseCalculateBestFeatureSplit", scope.makeOpName("BoostedTreesSparseCalculateBestFeatureSplit"));
-    opBuilder.addInput(nodeIdRange.asOutput());
-    opBuilder.addInput(statsSummaryIndices.asOutput());
-    opBuilder.addInput(statsSummaryValues.asOutput());
-    opBuilder.addInput(statsSummaryShape.asOutput());
-    opBuilder.addInput(l1.asOutput());
-    opBuilder.addInput(l2.asOutput());
-    opBuilder.addInput(treeComplexity.asOutput());
-    opBuilder.addInput(minNodeWeight.asOutput());
+    opBuilder.addInput(nodeIdRange.asOutput(scope));
+    opBuilder.addInput(statsSummaryIndices.asOutput(scope));
+    opBuilder.addInput(statsSummaryValues.asOutput(scope));
+    opBuilder.addInput(statsSummaryShape.asOutput(scope));
+    opBuilder.addInput(l1.asOutput(scope));
+    opBuilder.addInput(l2.asOutput(scope));
+    opBuilder.addInput(treeComplexity.asOutput(scope));
+    opBuilder.addInput(minNodeWeight.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("logits_dimension", logitsDimension);
     if (options != null) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesTrainingPredict.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesTrainingPredict.java
index a81509950b9..560e2b20948 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesTrainingPredict.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesTrainingPredict.java
@@ -56,10 +56,10 @@ public final class BoostedTreesTrainingPredict extends RawOp {
   @Endpoint(describeByClass = true)
   public static BoostedTreesTrainingPredict create(Scope scope, Operand<?> treeEnsembleHandle, Operand<TInt32> cachedTreeIds, Operand<TInt32> cachedNodeIds, Iterable<Operand<TInt32>> bucketizedFeatures, Long logitsDimension) {
     OperationBuilder opBuilder = scope.env().opBuilder("BoostedTreesTrainingPredict", scope.makeOpName("BoostedTreesTrainingPredict"));
-    opBuilder.addInput(treeEnsembleHandle.asOutput());
-    opBuilder.addInput(cachedTreeIds.asOutput());
-    opBuilder.addInput(cachedNodeIds.asOutput());
-    opBuilder.addInputList(Operands.asOutputs(bucketizedFeatures));
+    opBuilder.addInput(treeEnsembleHandle.asOutput(scope));
+    opBuilder.addInput(cachedTreeIds.asOutput(scope));
+    opBuilder.addInput(cachedNodeIds.asOutput(scope));
+    opBuilder.addInputList(Operands.asOutputs(scope, bucketizedFeatures));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("logits_dimension", logitsDimension);
     return new BoostedTreesTrainingPredict(opBuilder.build());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesUpdateEnsemble.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesUpdateEnsemble.java
index e6ddcf3d2da..1e304c4319a 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesUpdateEnsemble.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesUpdateEnsemble.java
@@ -62,15 +62,15 @@ public final class BoostedTreesUpdateEnsemble extends RawOp {
   @Endpoint(describeByClass = true)
   public static BoostedTreesUpdateEnsemble create(Scope scope, Operand<?> treeEnsembleHandle, Operand<TInt32> featureIds, Iterable<Operand<TInt32>> nodeIds, Iterable<Operand<TFloat32>> gains, Iterable<Operand<TInt32>> thresholds, Iterable<Operand<TFloat32>> leftNodeContribs, Iterable<Operand<TFloat32>> rightNodeContribs, Operand<TInt32> maxDepth, Operand<TFloat32> learningRate, Long pruningMode) {
     OperationBuilder opBuilder = scope.env().opBuilder("BoostedTreesUpdateEnsemble", scope.makeOpName("BoostedTreesUpdateEnsemble"));
-    opBuilder.addInput(treeEnsembleHandle.asOutput());
-    opBuilder.addInput(featureIds.asOutput());
-    opBuilder.addInputList(Operands.asOutputs(nodeIds));
-    opBuilder.addInputList(Operands.asOutputs(gains));
-    opBuilder.addInputList(Operands.asOutputs(thresholds));
-    opBuilder.addInputList(Operands.asOutputs(leftNodeContribs));
-    opBuilder.addInputList(Operands.asOutputs(rightNodeContribs));
-    opBuilder.addInput(maxDepth.asOutput());
-    opBuilder.addInput(learningRate.asOutput());
+    opBuilder.addInput(treeEnsembleHandle.asOutput(scope));
+    opBuilder.addInput(featureIds.asOutput(scope));
+    opBuilder.addInputList(Operands.asOutputs(scope, nodeIds));
+    opBuilder.addInputList(Operands.asOutputs(scope, gains));
+    opBuilder.addInputList(Operands.asOutputs(scope, thresholds));
+    opBuilder.addInputList(Operands.asOutputs(scope, leftNodeContribs));
+    opBuilder.addInputList(Operands.asOutputs(scope, rightNodeContribs));
+    opBuilder.addInput(maxDepth.asOutput(scope));
+    opBuilder.addInput(learningRate.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("pruning_mode", pruningMode);
     return new BoostedTreesUpdateEnsemble(opBuilder.build());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesUpdateEnsembleV2.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesUpdateEnsembleV2.java
index ceaff116fd1..653cdb0a3ea 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesUpdateEnsembleV2.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesUpdateEnsembleV2.java
@@ -85,18 +85,18 @@ private Options() {
   @Endpoint(describeByClass = true)
   public static BoostedTreesUpdateEnsembleV2 create(Scope scope, Operand<?> treeEnsembleHandle, Iterable<Operand<TInt32>> featureIds, Iterable<Operand<TInt32>> dimensionIds, Iterable<Operand<TInt32>> nodeIds, Iterable<Operand<TFloat32>> gains, Iterable<Operand<TInt32>> thresholds, Iterable<Operand<TFloat32>> leftNodeContribs, Iterable<Operand<TFloat32>> rightNodeContribs, Iterable<Operand<TString>> splitTypes, Operand<TInt32> maxDepth, Operand<TFloat32> learningRate, Operand<TInt32> pruningMode, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("BoostedTreesUpdateEnsembleV2", scope.makeOpName("BoostedTreesUpdateEnsembleV2"));
-    opBuilder.addInput(treeEnsembleHandle.asOutput());
-    opBuilder.addInputList(Operands.asOutputs(featureIds));
-    opBuilder.addInputList(Operands.asOutputs(dimensionIds));
-    opBuilder.addInputList(Operands.asOutputs(nodeIds));
-    opBuilder.addInputList(Operands.asOutputs(gains));
-    opBuilder.addInputList(Operands.asOutputs(thresholds));
-    opBuilder.addInputList(Operands.asOutputs(leftNodeContribs));
-    opBuilder.addInputList(Operands.asOutputs(rightNodeContribs));
-    opBuilder.addInputList(Operands.asOutputs(splitTypes));
-    opBuilder.addInput(maxDepth.asOutput());
-    opBuilder.addInput(learningRate.asOutput());
-    opBuilder.addInput(pruningMode.asOutput());
+    opBuilder.addInput(treeEnsembleHandle.asOutput(scope));
+    opBuilder.addInputList(Operands.asOutputs(scope, featureIds));
+    opBuilder.addInputList(Operands.asOutputs(scope, dimensionIds));
+    opBuilder.addInputList(Operands.asOutputs(scope, nodeIds));
+    opBuilder.addInputList(Operands.asOutputs(scope, gains));
+    opBuilder.addInputList(Operands.asOutputs(scope, thresholds));
+    opBuilder.addInputList(Operands.asOutputs(scope, leftNodeContribs));
+    opBuilder.addInputList(Operands.asOutputs(scope, rightNodeContribs));
+    opBuilder.addInputList(Operands.asOutputs(scope, splitTypes));
+    opBuilder.addInput(maxDepth.asOutput(scope));
+    opBuilder.addInput(learningRate.asOutput(scope));
+    opBuilder.addInput(pruningMode.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/IsBoostedTreesEnsembleInitialized.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/IsBoostedTreesEnsembleInitialized.java
index 7bf185408df..acccce061ff 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/IsBoostedTreesEnsembleInitialized.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/IsBoostedTreesEnsembleInitialized.java
@@ -42,7 +42,7 @@ public final class IsBoostedTreesEnsembleInitialized extends RawOp implements Op
   @Endpoint(describeByClass = true)
   public static IsBoostedTreesEnsembleInitialized create(Scope scope, Operand<?> treeEnsembleHandle) {
     OperationBuilder opBuilder = scope.env().opBuilder("IsBoostedTreesEnsembleInitialized", scope.makeOpName("IsBoostedTreesEnsembleInitialized"));
-    opBuilder.addInput(treeEnsembleHandle.asOutput());
+    opBuilder.addInput(treeEnsembleHandle.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new IsBoostedTreesEnsembleInitialized(opBuilder.build());
   }
@@ -55,7 +55,7 @@ public Output<TBool> isInitialized() {
   }
   
   @Override
-  public Output<TBool> asOutput() {
+  public Output<TBool> asOutput(Scope scope) {
     return isInitialized;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/IsBoostedTreesQuantileStreamResourceInitialized.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/IsBoostedTreesQuantileStreamResourceInitialized.java
index 66ad4435dfd..e16086691a8 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/IsBoostedTreesQuantileStreamResourceInitialized.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/IsBoostedTreesQuantileStreamResourceInitialized.java
@@ -44,7 +44,7 @@ public final class IsBoostedTreesQuantileStreamResourceInitialized extends RawOp
   @Endpoint(describeByClass = true)
   public static IsBoostedTreesQuantileStreamResourceInitialized create(Scope scope, Operand<?> quantileStreamResourceHandle) {
     OperationBuilder opBuilder = scope.env().opBuilder("IsBoostedTreesQuantileStreamResourceInitialized", scope.makeOpName("IsBoostedTreesQuantileStreamResourceInitialized"));
-    opBuilder.addInput(quantileStreamResourceHandle.asOutput());
+    opBuilder.addInput(quantileStreamResourceHandle.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new IsBoostedTreesQuantileStreamResourceInitialized(opBuilder.build());
   }
@@ -57,7 +57,7 @@ public Output<TBool> isInitialized() {
   }
   
   @Override
-  public Output<TBool> asOutput() {
+  public Output<TBool> asOutput(Scope scope) {
     return isInitialized;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/AdjustContrast.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/AdjustContrast.java
index 5bf4fa92401..a5f824f5145 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/AdjustContrast.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/AdjustContrast.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -45,7 +44,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "image")
-public final class AdjustContrast<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class AdjustContrast<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new AdjustContrast operation.
@@ -56,10 +55,10 @@ public final class AdjustContrast<T extends Tensor & TNumber> extends RawOp impl
    * @return a new instance of AdjustContrast
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> AdjustContrast<T> create(Scope scope, Operand<T> images, Operand<TFloat32> contrastFactor) {
+  public static <T extends TNumber> AdjustContrast<T> create(Scope scope, Operand<T> images, Operand<TFloat32> contrastFactor) {
     OperationBuilder opBuilder = scope.env().opBuilder("AdjustContrastv2", scope.makeOpName("AdjustContrast"));
-    opBuilder.addInput(images.asOutput());
-    opBuilder.addInput(contrastFactor.asOutput());
+    opBuilder.addInput(images.asOutput(scope));
+    opBuilder.addInput(contrastFactor.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new AdjustContrast<T>(opBuilder.build());
   }
@@ -72,7 +71,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/AdjustHue.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/AdjustHue.java
index da6eb729a76..00d49018a7e 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/AdjustHue.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/AdjustHue.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -42,7 +41,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "image")
-public final class AdjustHue<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class AdjustHue<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new AdjustHue operation.
@@ -53,10 +52,10 @@ public final class AdjustHue<T extends Tensor & TNumber> extends RawOp implement
    * @return a new instance of AdjustHue
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> AdjustHue<T> create(Scope scope, Operand<T> images, Operand<TFloat32> delta) {
+  public static <T extends TNumber> AdjustHue<T> create(Scope scope, Operand<T> images, Operand<TFloat32> delta) {
     OperationBuilder opBuilder = scope.env().opBuilder("AdjustHue", scope.makeOpName("AdjustHue"));
-    opBuilder.addInput(images.asOutput());
-    opBuilder.addInput(delta.asOutput());
+    opBuilder.addInput(images.asOutput(scope));
+    opBuilder.addInput(delta.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new AdjustHue<T>(opBuilder.build());
   }
@@ -69,7 +68,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/AdjustSaturation.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/AdjustSaturation.java
index ffe643d8434..d0100ce4f8a 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/AdjustSaturation.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/AdjustSaturation.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -42,7 +41,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "image")
-public final class AdjustSaturation<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class AdjustSaturation<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new AdjustSaturation operation.
@@ -53,10 +52,10 @@ public final class AdjustSaturation<T extends Tensor & TNumber> extends RawOp im
    * @return a new instance of AdjustSaturation
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> AdjustSaturation<T> create(Scope scope, Operand<T> images, Operand<TFloat32> scale) {
+  public static <T extends TNumber> AdjustSaturation<T> create(Scope scope, Operand<T> images, Operand<TFloat32> scale) {
     OperationBuilder opBuilder = scope.env().opBuilder("AdjustSaturation", scope.makeOpName("AdjustSaturation"));
-    opBuilder.addInput(images.asOutput());
-    opBuilder.addInput(scale.asOutput());
+    opBuilder.addInput(images.asOutput(scope));
+    opBuilder.addInput(scale.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new AdjustSaturation<T>(opBuilder.build());
   }
@@ -69,7 +68,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/CombinedNonMaxSuppression.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/CombinedNonMaxSuppression.java
index 42bf54e7143..4868f373f58 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/CombinedNonMaxSuppression.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/CombinedNonMaxSuppression.java
@@ -104,12 +104,12 @@ private Options() {
   @Endpoint(describeByClass = true)
   public static CombinedNonMaxSuppression create(Scope scope, Operand<TFloat32> boxes, Operand<TFloat32> scores, Operand<TInt32> maxOutputSizePerClass, Operand<TInt32> maxTotalSize, Operand<TFloat32> iouThreshold, Operand<TFloat32> scoreThreshold, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("CombinedNonMaxSuppression", scope.makeOpName("CombinedNonMaxSuppression"));
-    opBuilder.addInput(boxes.asOutput());
-    opBuilder.addInput(scores.asOutput());
-    opBuilder.addInput(maxOutputSizePerClass.asOutput());
-    opBuilder.addInput(maxTotalSize.asOutput());
-    opBuilder.addInput(iouThreshold.asOutput());
-    opBuilder.addInput(scoreThreshold.asOutput());
+    opBuilder.addInput(boxes.asOutput(scope));
+    opBuilder.addInput(scores.asOutput(scope));
+    opBuilder.addInput(maxOutputSizePerClass.asOutput(scope));
+    opBuilder.addInput(maxTotalSize.asOutput(scope));
+    opBuilder.addInput(iouThreshold.asOutput(scope));
+    opBuilder.addInput(scoreThreshold.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/CropAndResize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/CropAndResize.java
index 5c30c811267..bae22372323 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/CropAndResize.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/CropAndResize.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -108,12 +107,12 @@ private Options() {
    * @return a new instance of CropAndResize
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> CropAndResize create(Scope scope, Operand<T> image, Operand<TFloat32> boxes, Operand<TInt32> boxInd, Operand<TInt32> cropSize, Options... options) {
+  public static <T extends TNumber> CropAndResize create(Scope scope, Operand<T> image, Operand<TFloat32> boxes, Operand<TInt32> boxInd, Operand<TInt32> cropSize, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("CropAndResize", scope.makeOpName("CropAndResize"));
-    opBuilder.addInput(image.asOutput());
-    opBuilder.addInput(boxes.asOutput());
-    opBuilder.addInput(boxInd.asOutput());
-    opBuilder.addInput(cropSize.asOutput());
+    opBuilder.addInput(image.asOutput(scope));
+    opBuilder.addInput(boxes.asOutput(scope));
+    opBuilder.addInput(boxInd.asOutput(scope));
+    opBuilder.addInput(cropSize.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -152,7 +151,7 @@ public Output<TFloat32> crops() {
   }
   
   @Override
-  public Output<TFloat32> asOutput() {
+  public Output<TFloat32> asOutput(Scope scope) {
     return crops;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/CropAndResizeGradBoxes.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/CropAndResizeGradBoxes.java
index adfb2f39082..3806d28fd2a 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/CropAndResizeGradBoxes.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/CropAndResizeGradBoxes.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -79,12 +78,12 @@ private Options() {
    * @return a new instance of CropAndResizeGradBoxes
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> CropAndResizeGradBoxes create(Scope scope, Operand<TFloat32> grads, Operand<T> image, Operand<TFloat32> boxes, Operand<TInt32> boxInd, Options... options) {
+  public static <T extends TNumber> CropAndResizeGradBoxes create(Scope scope, Operand<TFloat32> grads, Operand<T> image, Operand<TFloat32> boxes, Operand<TInt32> boxInd, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("CropAndResizeGradBoxes", scope.makeOpName("CropAndResizeGradBoxes"));
-    opBuilder.addInput(grads.asOutput());
-    opBuilder.addInput(image.asOutput());
-    opBuilder.addInput(boxes.asOutput());
-    opBuilder.addInput(boxInd.asOutput());
+    opBuilder.addInput(grads.asOutput(scope));
+    opBuilder.addInput(image.asOutput(scope));
+    opBuilder.addInput(boxes.asOutput(scope));
+    opBuilder.addInput(boxInd.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -112,7 +111,7 @@ public Output<TFloat32> output() {
   }
   
   @Override
-  public Output<TFloat32> asOutput() {
+  public Output<TFloat32> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/CropAndResizeGradImage.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/CropAndResizeGradImage.java
index 79d9f6a5fc0..e3a70effa94 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/CropAndResizeGradImage.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/CropAndResizeGradImage.java
@@ -22,7 +22,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -37,7 +36,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "image")
-public final class CropAndResizeGradImage<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class CropAndResizeGradImage<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.image.CropAndResizeGradImage}
@@ -84,12 +83,12 @@ private Options() {
    * @return a new instance of CropAndResizeGradImage
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> CropAndResizeGradImage<T> create(Scope scope, Operand<TFloat32> grads, Operand<TFloat32> boxes, Operand<TInt32> boxInd, Operand<TInt32> imageSize, DataType<T> T, Options... options) {
+  public static <T extends TNumber> CropAndResizeGradImage<T> create(Scope scope, Operand<TFloat32> grads, Operand<TFloat32> boxes, Operand<TInt32> boxInd, Operand<TInt32> imageSize, DataType<T> T, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("CropAndResizeGradImage", scope.makeOpName("CropAndResizeGradImage"));
-    opBuilder.addInput(grads.asOutput());
-    opBuilder.addInput(boxes.asOutput());
-    opBuilder.addInput(boxInd.asOutput());
-    opBuilder.addInput(imageSize.asOutput());
+    opBuilder.addInput(grads.asOutput(scope));
+    opBuilder.addInput(boxes.asOutput(scope));
+    opBuilder.addInput(boxInd.asOutput(scope));
+    opBuilder.addInput(imageSize.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("T", T);
     if (options != null) {
@@ -118,7 +117,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/DecodeAndCropJpeg.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/DecodeAndCropJpeg.java
index 0c2ac4b1471..dc8282ec77a 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/DecodeAndCropJpeg.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/DecodeAndCropJpeg.java
@@ -143,8 +143,8 @@ private Options() {
   @Endpoint(describeByClass = true)
   public static DecodeAndCropJpeg create(Scope scope, Operand<TString> contents, Operand<TInt32> cropWindow, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("DecodeAndCropJpeg", scope.makeOpName("DecodeAndCropJpeg"));
-    opBuilder.addInput(contents.asOutput());
-    opBuilder.addInput(cropWindow.asOutput());
+    opBuilder.addInput(contents.asOutput(scope));
+    opBuilder.addInput(cropWindow.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -228,7 +228,7 @@ public Output<TUint8> image() {
   }
   
   @Override
-  public Output<TUint8> asOutput() {
+  public Output<TUint8> asOutput(Scope scope) {
     return image;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/DecodeBmp.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/DecodeBmp.java
index 3801822db4e..7ffa278060a 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/DecodeBmp.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/DecodeBmp.java
@@ -78,7 +78,7 @@ private Options() {
   @Endpoint(describeByClass = true)
   public static DecodeBmp create(Scope scope, Operand<TString> contents, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("DecodeBmp", scope.makeOpName("DecodeBmp"));
-    opBuilder.addInput(contents.asOutput());
+    opBuilder.addInput(contents.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -105,7 +105,7 @@ public Output<TUint8> image() {
   }
   
   @Override
-  public Output<TUint8> asOutput() {
+  public Output<TUint8> asOutput(Scope scope) {
     return image;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/DecodeGif.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/DecodeGif.java
index 7c25b8c5b0c..6ec523b5a34 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/DecodeGif.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/DecodeGif.java
@@ -53,7 +53,7 @@ public final class DecodeGif extends RawOp implements Operand<TUint8> {
   @Endpoint(describeByClass = true)
   public static DecodeGif create(Scope scope, Operand<TString> contents) {
     OperationBuilder opBuilder = scope.env().opBuilder("DecodeGif", scope.makeOpName("DecodeGif"));
-    opBuilder.addInput(contents.asOutput());
+    opBuilder.addInput(contents.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new DecodeGif(opBuilder.build());
   }
@@ -66,7 +66,7 @@ public Output<TUint8> image() {
   }
   
   @Override
-  public Output<TUint8> asOutput() {
+  public Output<TUint8> asOutput(Scope scope) {
     return image;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/DecodeJpeg.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/DecodeJpeg.java
index e2b20460b2e..854b391f83b 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/DecodeJpeg.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/DecodeJpeg.java
@@ -141,7 +141,7 @@ private Options() {
   @Endpoint(describeByClass = true)
   public static DecodeJpeg create(Scope scope, Operand<TString> contents, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("DecodeJpeg", scope.makeOpName("DecodeJpeg"));
-    opBuilder.addInput(contents.asOutput());
+    opBuilder.addInput(contents.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -225,7 +225,7 @@ public Output<TUint8> image() {
   }
   
   @Override
-  public Output<TUint8> asOutput() {
+  public Output<TUint8> asOutput(Scope scope) {
     return image;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/DecodePng.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/DecodePng.java
index 2d8a35af092..0f0b995b868 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/DecodePng.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/DecodePng.java
@@ -22,7 +22,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -61,7 +60,7 @@
  * @param <T> data type for {@code image()} output
  */
 @Operator(group = "image")
-public final class DecodePng<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class DecodePng<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.image.DecodePng}
@@ -92,9 +91,9 @@ private Options() {
    * @return a new instance of DecodePng
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> DecodePng<T> create(Scope scope, Operand<TString> contents, DataType<T> dtype, Options... options) {
+  public static <T extends TNumber> DecodePng<T> create(Scope scope, Operand<TString> contents, DataType<T> dtype, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("DecodePng", scope.makeOpName("DecodePng"));
-    opBuilder.addInput(contents.asOutput());
+    opBuilder.addInput(contents.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("dtype", dtype);
     if (options != null) {
@@ -135,7 +134,7 @@ public Output<T> image() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return image;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/DrawBoundingBoxes.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/DrawBoundingBoxes.java
index e1e6b6d4760..ad0e84f294d 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/DrawBoundingBoxes.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/DrawBoundingBoxes.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -47,7 +46,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "image")
-public final class DrawBoundingBoxes<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class DrawBoundingBoxes<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new DrawBoundingBoxes operation.
@@ -60,11 +59,11 @@ public final class DrawBoundingBoxes<T extends Tensor & TNumber> extends RawOp i
    * @return a new instance of DrawBoundingBoxes
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> DrawBoundingBoxes<T> create(Scope scope, Operand<T> images, Operand<TFloat32> boxes, Operand<TFloat32> colors) {
+  public static <T extends TNumber> DrawBoundingBoxes<T> create(Scope scope, Operand<T> images, Operand<TFloat32> boxes, Operand<TFloat32> colors) {
     OperationBuilder opBuilder = scope.env().opBuilder("DrawBoundingBoxesV2", scope.makeOpName("DrawBoundingBoxes"));
-    opBuilder.addInput(images.asOutput());
-    opBuilder.addInput(boxes.asOutput());
-    opBuilder.addInput(colors.asOutput());
+    opBuilder.addInput(images.asOutput(scope));
+    opBuilder.addInput(boxes.asOutput(scope));
+    opBuilder.addInput(colors.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new DrawBoundingBoxes<T>(opBuilder.build());
   }
@@ -78,7 +77,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/EncodeJpeg.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/EncodeJpeg.java
index 920961e1115..6d504201441 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/EncodeJpeg.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/EncodeJpeg.java
@@ -163,7 +163,7 @@ private Options() {
   @Endpoint(describeByClass = true)
   public static EncodeJpeg create(Scope scope, Operand<TUint8> image, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("EncodeJpeg", scope.makeOpName("EncodeJpeg"));
-    opBuilder.addInput(image.asOutput());
+    opBuilder.addInput(image.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -271,7 +271,7 @@ public Output<TString> contents() {
   }
   
   @Override
-  public Output<TString> asOutput() {
+  public Output<TString> asOutput(Scope scope) {
     return contents;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/EncodeJpegVariableQuality.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/EncodeJpegVariableQuality.java
index c07fc341f00..84a39b998c1 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/EncodeJpegVariableQuality.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/EncodeJpegVariableQuality.java
@@ -50,8 +50,8 @@ public final class EncodeJpegVariableQuality extends RawOp implements Operand<TS
   @Endpoint(describeByClass = true)
   public static EncodeJpegVariableQuality create(Scope scope, Operand<TUint8> images, Operand<TInt32> quality) {
     OperationBuilder opBuilder = scope.env().opBuilder("EncodeJpegVariableQuality", scope.makeOpName("EncodeJpegVariableQuality"));
-    opBuilder.addInput(images.asOutput());
-    opBuilder.addInput(quality.asOutput());
+    opBuilder.addInput(images.asOutput(scope));
+    opBuilder.addInput(quality.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new EncodeJpegVariableQuality(opBuilder.build());
   }
@@ -64,7 +64,7 @@ public Output<TString> contents() {
   }
   
   @Override
-  public Output<TString> asOutput() {
+  public Output<TString> asOutput(Scope scope) {
     return contents;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/EncodePng.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/EncodePng.java
index 6a75551dc4c..a43391c0264 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/EncodePng.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/EncodePng.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -83,9 +82,9 @@ private Options() {
    * @return a new instance of EncodePng
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> EncodePng create(Scope scope, Operand<T> image, Options... options) {
+  public static <T extends TNumber> EncodePng create(Scope scope, Operand<T> image, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("EncodePng", scope.makeOpName("EncodePng"));
-    opBuilder.addInput(image.asOutput());
+    opBuilder.addInput(image.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -112,7 +111,7 @@ public Output<TString> contents() {
   }
   
   @Override
-  public Output<TString> asOutput() {
+  public Output<TString> asOutput(Scope scope) {
     return contents;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ExtractGlimpse.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ExtractGlimpse.java
index 05bc1d924b3..2f41f806d48 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ExtractGlimpse.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ExtractGlimpse.java
@@ -58,6 +58,7 @@
  * If the coordinates are not normalized they are interpreted as
  *   numbers of pixels.
  */
+@Operator(group = "image")
 public final class ExtractGlimpse extends RawOp implements Operand<TFloat32> {
   
   /**
@@ -127,10 +128,10 @@ private Options() {
    */
   @Endpoint(describeByClass = true)
   public static ExtractGlimpse create(Scope scope, Operand<TFloat32> input, Operand<TInt32> size, Operand<TFloat32> offsets, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ExtractGlimpseV2", scope.makeOpName("ExtractGlimpse"));
-    opBuilder.addInput(input.asOutput());
-    opBuilder.addInput(size.asOutput());
-    opBuilder.addInput(offsets.asOutput());
+    OperationBuilder opBuilder = scope.env().opBuilder("ExtractGlimpse", scope.makeOpName("ExtractGlimpse"));
+    opBuilder.addInput(input.asOutput(scope));
+    opBuilder.addInput(size.asOutput(scope));
+    opBuilder.addInput(offsets.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -194,13 +195,10 @@ public Output<TFloat32> glimpse() {
   }
   
   @Override
-  public Output<TFloat32> asOutput() {
+  public Output<TFloat32> asOutput(Scope scope) {
     return glimpse;
   }
   
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ExtractGlimpseV2";
-  
   private Output<TFloat32> glimpse;
   
   private ExtractGlimpse(Operation operation) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ExtractImagePatches.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ExtractImagePatches.java
index 8a342524c91..7e9f055783f 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ExtractImagePatches.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ExtractImagePatches.java
@@ -22,11 +22,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Extract `patches` from `images` and put them in the "depth" output dimension.
@@ -34,7 +34,7 @@
  * @param <T> data type for {@code patches()} output
  */
 @Operator(group = "image")
-public final class ExtractImagePatches<T extends Tensor> extends RawOp implements Operand<T> {
+public final class ExtractImagePatches<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new ExtractImagePatches operation.
@@ -54,9 +54,9 @@ public final class ExtractImagePatches<T extends Tensor> extends RawOp implement
    * @return a new instance of ExtractImagePatches
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> ExtractImagePatches<T> create(Scope scope, Operand<T> images, List<Long> ksizes, List<Long> strides, List<Long> rates, String padding) {
+  public static <T extends TType> ExtractImagePatches<T> create(Scope scope, Operand<T> images, List<Long> ksizes, List<Long> strides, List<Long> rates, String padding) {
     OperationBuilder opBuilder = scope.env().opBuilder("ExtractImagePatches", scope.makeOpName("ExtractImagePatches"));
-    opBuilder.addInput(images.asOutput());
+    opBuilder.addInput(images.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     long[] ksizesArray = new long[ksizes.size()];
     for (int i = 0; i < ksizesArray.length; ++i) {
@@ -88,7 +88,7 @@ public Output<T> patches() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return patches;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ExtractJpegShape.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ExtractJpegShape.java
index 53258a1723f..d5932d9fe45 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ExtractJpegShape.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ExtractJpegShape.java
@@ -22,7 +22,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -39,7 +38,7 @@
  * @param <T> data type for {@code imageShape()} output
  */
 @Operator(group = "image")
-public final class ExtractJpegShape<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class ExtractJpegShape<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new ExtractJpegShape operation.
@@ -51,9 +50,9 @@ public final class ExtractJpegShape<T extends Tensor & TNumber> extends RawOp im
    * @return a new instance of ExtractJpegShape
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> ExtractJpegShape<T> create(Scope scope, Operand<TString> contents, DataType<T> outputType) {
+  public static <T extends TNumber> ExtractJpegShape<T> create(Scope scope, Operand<TString> contents, DataType<T> outputType) {
     OperationBuilder opBuilder = scope.env().opBuilder("ExtractJpegShape", scope.makeOpName("ExtractJpegShape"));
-    opBuilder.addInput(contents.asOutput());
+    opBuilder.addInput(contents.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("output_type", outputType);
     return new ExtractJpegShape<T>(opBuilder.build());
@@ -79,7 +78,7 @@ public Output<T> imageShape() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return imageShape;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/GenerateBoundingBoxProposals.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/GenerateBoundingBoxProposals.java
index e287a419cdd..4b83bea8c54 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/GenerateBoundingBoxProposals.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/GenerateBoundingBoxProposals.java
@@ -82,13 +82,13 @@ private Options() {
   @Endpoint(describeByClass = true)
   public static GenerateBoundingBoxProposals create(Scope scope, Operand<TFloat32> scores, Operand<TFloat32> bboxDeltas, Operand<TFloat32> imageInfo, Operand<TFloat32> anchors, Operand<TFloat32> nmsThreshold, Operand<TInt32> preNmsTopn, Operand<TFloat32> minSize, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("GenerateBoundingBoxProposals", scope.makeOpName("GenerateBoundingBoxProposals"));
-    opBuilder.addInput(scores.asOutput());
-    opBuilder.addInput(bboxDeltas.asOutput());
-    opBuilder.addInput(imageInfo.asOutput());
-    opBuilder.addInput(anchors.asOutput());
-    opBuilder.addInput(nmsThreshold.asOutput());
-    opBuilder.addInput(preNmsTopn.asOutput());
-    opBuilder.addInput(minSize.asOutput());
+    opBuilder.addInput(scores.asOutput(scope));
+    opBuilder.addInput(bboxDeltas.asOutput(scope));
+    opBuilder.addInput(imageInfo.asOutput(scope));
+    opBuilder.addInput(anchors.asOutput(scope));
+    opBuilder.addInput(nmsThreshold.asOutput(scope));
+    opBuilder.addInput(preNmsTopn.asOutput(scope));
+    opBuilder.addInput(minSize.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/HsvToRgb.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/HsvToRgb.java
index 06102fdee52..ac6876c570d 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/HsvToRgb.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/HsvToRgb.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -40,7 +39,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "image")
-public final class HsvToRgb<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class HsvToRgb<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new HsvToRgb operation.
@@ -50,9 +49,9 @@ public final class HsvToRgb<T extends Tensor & TNumber> extends RawOp implements
    * @return a new instance of HsvToRgb
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> HsvToRgb<T> create(Scope scope, Operand<T> images) {
+  public static <T extends TNumber> HsvToRgb<T> create(Scope scope, Operand<T> images) {
     OperationBuilder opBuilder = scope.env().opBuilder("HSVToRGB", scope.makeOpName("HsvToRgb"));
-    opBuilder.addInput(images.asOutput());
+    opBuilder.addInput(images.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new HsvToRgb<T>(opBuilder.build());
   }
@@ -65,7 +64,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ImageProjectiveTransformV2.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ImageProjectiveTransformV2.java
index 15155c535c4..250e6659599 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ImageProjectiveTransformV2.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ImageProjectiveTransformV2.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -41,7 +40,7 @@
  * 
  * @param <T> data type for {@code transformedImages()} output
  */
-public final class ImageProjectiveTransformV2<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class ImageProjectiveTransformV2<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.image.ImageProjectiveTransformV2}
@@ -76,11 +75,11 @@ private Options() {
    * @return a new instance of ImageProjectiveTransformV2
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> ImageProjectiveTransformV2<T> create(Scope scope, Operand<T> images, Operand<TFloat32> transforms, Operand<TInt32> outputShape, String interpolation, Options... options) {
+  public static <T extends TNumber> ImageProjectiveTransformV2<T> create(Scope scope, Operand<T> images, Operand<TFloat32> transforms, Operand<TInt32> outputShape, String interpolation, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ImageProjectiveTransformV2", scope.makeOpName("ImageProjectiveTransformV2"));
-    opBuilder.addInput(images.asOutput());
-    opBuilder.addInput(transforms.asOutput());
-    opBuilder.addInput(outputShape.asOutput());
+    opBuilder.addInput(images.asOutput(scope));
+    opBuilder.addInput(transforms.asOutput(scope));
+    opBuilder.addInput(outputShape.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("interpolation", interpolation);
     if (options != null) {
@@ -109,7 +108,7 @@ public Output<T> transformedImages() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return transformedImages;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/NearestNeighbors.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/NearestNeighbors.java
index f3ad7cb890b..0dd8f6d45d7 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/NearestNeighbors.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/NearestNeighbors.java
@@ -50,9 +50,9 @@ public final class NearestNeighbors extends RawOp {
   @Endpoint(describeByClass = true)
   public static NearestNeighbors create(Scope scope, Operand<TFloat32> points, Operand<TFloat32> centers, Operand<TInt64> k) {
     OperationBuilder opBuilder = scope.env().opBuilder("NearestNeighbors", scope.makeOpName("NearestNeighbors"));
-    opBuilder.addInput(points.asOutput());
-    opBuilder.addInput(centers.asOutput());
-    opBuilder.addInput(k.asOutput());
+    opBuilder.addInput(points.asOutput(scope));
+    opBuilder.addInput(centers.asOutput(scope));
+    opBuilder.addInput(k.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new NearestNeighbors(opBuilder.build());
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/NonMaxSuppression.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/NonMaxSuppression.java
index 77639f7bb60..12f9714a150 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/NonMaxSuppression.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/NonMaxSuppression.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -58,7 +57,7 @@
  * @param <T> data type for {@code selectedScores()} output
  */
 @Operator(group = "image")
-public final class NonMaxSuppression<T extends Tensor & TNumber> extends RawOp {
+public final class NonMaxSuppression<T extends TNumber> extends RawOp {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.image.NonMaxSuppression}
@@ -100,14 +99,14 @@ private Options() {
    * @return a new instance of NonMaxSuppression
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> NonMaxSuppression<T> create(Scope scope, Operand<T> boxes, Operand<T> scores, Operand<TInt32> maxOutputSize, Operand<T> iouThreshold, Operand<T> scoreThreshold, Operand<T> softNmsSigma, Options... options) {
+  public static <T extends TNumber> NonMaxSuppression<T> create(Scope scope, Operand<T> boxes, Operand<T> scores, Operand<TInt32> maxOutputSize, Operand<T> iouThreshold, Operand<T> scoreThreshold, Operand<T> softNmsSigma, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("NonMaxSuppressionV5", scope.makeOpName("NonMaxSuppression"));
-    opBuilder.addInput(boxes.asOutput());
-    opBuilder.addInput(scores.asOutput());
-    opBuilder.addInput(maxOutputSize.asOutput());
-    opBuilder.addInput(iouThreshold.asOutput());
-    opBuilder.addInput(scoreThreshold.asOutput());
-    opBuilder.addInput(softNmsSigma.asOutput());
+    opBuilder.addInput(boxes.asOutput(scope));
+    opBuilder.addInput(scores.asOutput(scope));
+    opBuilder.addInput(maxOutputSize.asOutput(scope));
+    opBuilder.addInput(iouThreshold.asOutput(scope));
+    opBuilder.addInput(scoreThreshold.asOutput(scope));
+    opBuilder.addInput(softNmsSigma.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/NonMaxSuppressionWithOverlaps.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/NonMaxSuppressionWithOverlaps.java
index 073aee26ff9..952746a3108 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/NonMaxSuppressionWithOverlaps.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/NonMaxSuppressionWithOverlaps.java
@@ -68,11 +68,11 @@ public final class NonMaxSuppressionWithOverlaps extends RawOp implements Operan
   @Endpoint(describeByClass = true)
   public static NonMaxSuppressionWithOverlaps create(Scope scope, Operand<TFloat32> overlaps, Operand<TFloat32> scores, Operand<TInt32> maxOutputSize, Operand<TFloat32> overlapThreshold, Operand<TFloat32> scoreThreshold) {
     OperationBuilder opBuilder = scope.env().opBuilder("NonMaxSuppressionWithOverlaps", scope.makeOpName("NonMaxSuppressionWithOverlaps"));
-    opBuilder.addInput(overlaps.asOutput());
-    opBuilder.addInput(scores.asOutput());
-    opBuilder.addInput(maxOutputSize.asOutput());
-    opBuilder.addInput(overlapThreshold.asOutput());
-    opBuilder.addInput(scoreThreshold.asOutput());
+    opBuilder.addInput(overlaps.asOutput(scope));
+    opBuilder.addInput(scores.asOutput(scope));
+    opBuilder.addInput(maxOutputSize.asOutput(scope));
+    opBuilder.addInput(overlapThreshold.asOutput(scope));
+    opBuilder.addInput(scoreThreshold.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new NonMaxSuppressionWithOverlaps(opBuilder.build());
   }
@@ -86,7 +86,7 @@ public Output<TInt32> selectedIndices() {
   }
   
   @Override
-  public Output<TInt32> asOutput() {
+  public Output<TInt32> asOutput(Scope scope) {
     return selectedIndices;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/QuantizedResizeBilinear.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/QuantizedResizeBilinear.java
index 9da9e33b053..004caafdb5e 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/QuantizedResizeBilinear.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/QuantizedResizeBilinear.java
@@ -21,13 +21,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TFloat32;
 import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TType;
 
 /**
  * Resize quantized `images` to `size` using quantized bilinear interpolation.
@@ -37,7 +37,7 @@
  * @param <T> data type for {@code resizedImages()} output
  */
 @Operator(group = "image")
-public final class QuantizedResizeBilinear<T extends Tensor> extends RawOp {
+public final class QuantizedResizeBilinear<T extends TType> extends RawOp {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.image.QuantizedResizeBilinear}
@@ -81,12 +81,12 @@ private Options() {
    * @return a new instance of QuantizedResizeBilinear
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> QuantizedResizeBilinear<T> create(Scope scope, Operand<T> images, Operand<TInt32> size, Operand<TFloat32> min, Operand<TFloat32> max, Options... options) {
+  public static <T extends TType> QuantizedResizeBilinear<T> create(Scope scope, Operand<T> images, Operand<TInt32> size, Operand<TFloat32> min, Operand<TFloat32> max, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("QuantizedResizeBilinear", scope.makeOpName("QuantizedResizeBilinear"));
-    opBuilder.addInput(images.asOutput());
-    opBuilder.addInput(size.asOutput());
-    opBuilder.addInput(min.asOutput());
-    opBuilder.addInput(max.asOutput());
+    opBuilder.addInput(images.asOutput(scope));
+    opBuilder.addInput(size.asOutput(scope));
+    opBuilder.addInput(min.asOutput(scope));
+    opBuilder.addInput(max.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/RandomCrop.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/RandomCrop.java
index 8c5dabc23ed..1a5a2a4f47c 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/RandomCrop.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/RandomCrop.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -42,7 +41,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "image")
-public final class RandomCrop<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class RandomCrop<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.image.RandomCrop}
@@ -84,10 +83,10 @@ private Options() {
    * @return a new instance of RandomCrop
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> RandomCrop<T> create(Scope scope, Operand<T> image, Operand<TInt64> size, Options... options) {
+  public static <T extends TNumber> RandomCrop<T> create(Scope scope, Operand<T> image, Operand<TInt64> size, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("RandomCrop", scope.makeOpName("RandomCrop"));
-    opBuilder.addInput(image.asOutput());
-    opBuilder.addInput(size.asOutput());
+    opBuilder.addInput(image.asOutput(scope));
+    opBuilder.addInput(size.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -126,7 +125,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ResizeArea.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ResizeArea.java
index da1a126950f..e4252030c3a 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ResizeArea.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ResizeArea.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -79,10 +78,10 @@ private Options() {
    * @return a new instance of ResizeArea
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> ResizeArea create(Scope scope, Operand<T> images, Operand<TInt32> size, Options... options) {
+  public static <T extends TNumber> ResizeArea create(Scope scope, Operand<T> images, Operand<TInt32> size, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ResizeArea", scope.makeOpName("ResizeArea"));
-    opBuilder.addInput(images.asOutput());
-    opBuilder.addInput(size.asOutput());
+    opBuilder.addInput(images.asOutput(scope));
+    opBuilder.addInput(size.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -111,7 +110,7 @@ public Output<TFloat32> resizedImages() {
   }
   
   @Override
-  public Output<TFloat32> asOutput() {
+  public Output<TFloat32> asOutput(Scope scope) {
     return resizedImages;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ResizeBicubic.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ResizeBicubic.java
index 750b4de7cb4..caf2a49dba5 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ResizeBicubic.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ResizeBicubic.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -78,10 +77,10 @@ private Options() {
    * @return a new instance of ResizeBicubic
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> ResizeBicubic create(Scope scope, Operand<T> images, Operand<TInt32> size, Options... options) {
+  public static <T extends TNumber> ResizeBicubic create(Scope scope, Operand<T> images, Operand<TInt32> size, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ResizeBicubic", scope.makeOpName("ResizeBicubic"));
-    opBuilder.addInput(images.asOutput());
-    opBuilder.addInput(size.asOutput());
+    opBuilder.addInput(images.asOutput(scope));
+    opBuilder.addInput(size.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -120,7 +119,7 @@ public Output<TFloat32> resizedImages() {
   }
   
   @Override
-  public Output<TFloat32> asOutput() {
+  public Output<TFloat32> asOutput(Scope scope) {
     return resizedImages;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ResizeBicubicGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ResizeBicubicGrad.java
index 0fb80f7f2c8..623941bbc98 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ResizeBicubicGrad.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ResizeBicubicGrad.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -34,7 +33,7 @@
  * 
  * @param <T> data type for {@code output()} output
  */
-public final class ResizeBicubicGrad<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class ResizeBicubicGrad<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.image.ResizeBicubicGrad}
@@ -76,10 +75,10 @@ private Options() {
    * @return a new instance of ResizeBicubicGrad
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> ResizeBicubicGrad<T> create(Scope scope, Operand<TFloat32> grads, Operand<T> originalImage, Options... options) {
+  public static <T extends TNumber> ResizeBicubicGrad<T> create(Scope scope, Operand<TFloat32> grads, Operand<T> originalImage, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ResizeBicubicGrad", scope.makeOpName("ResizeBicubicGrad"));
-    opBuilder.addInput(grads.asOutput());
-    opBuilder.addInput(originalImage.asOutput());
+    opBuilder.addInput(grads.asOutput(scope));
+    opBuilder.addInput(originalImage.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -119,7 +118,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ResizeBilinear.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ResizeBilinear.java
index 65d21cd2dcf..c65c9c8d176 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ResizeBilinear.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ResizeBilinear.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -78,10 +77,10 @@ private Options() {
    * @return a new instance of ResizeBilinear
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> ResizeBilinear create(Scope scope, Operand<T> images, Operand<TInt32> size, Options... options) {
+  public static <T extends TNumber> ResizeBilinear create(Scope scope, Operand<T> images, Operand<TInt32> size, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ResizeBilinear", scope.makeOpName("ResizeBilinear"));
-    opBuilder.addInput(images.asOutput());
-    opBuilder.addInput(size.asOutput());
+    opBuilder.addInput(images.asOutput(scope));
+    opBuilder.addInput(size.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -120,7 +119,7 @@ public Output<TFloat32> resizedImages() {
   }
   
   @Override
-  public Output<TFloat32> asOutput() {
+  public Output<TFloat32> asOutput(Scope scope) {
     return resizedImages;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ResizeBilinearGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ResizeBilinearGrad.java
index cd643e7252b..bacb51b9775 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ResizeBilinearGrad.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ResizeBilinearGrad.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -34,7 +33,7 @@
  * 
  * @param <T> data type for {@code output()} output
  */
-public final class ResizeBilinearGrad<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class ResizeBilinearGrad<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.image.ResizeBilinearGrad}
@@ -76,10 +75,10 @@ private Options() {
    * @return a new instance of ResizeBilinearGrad
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> ResizeBilinearGrad<T> create(Scope scope, Operand<TFloat32> grads, Operand<T> originalImage, Options... options) {
+  public static <T extends TNumber> ResizeBilinearGrad<T> create(Scope scope, Operand<TFloat32> grads, Operand<T> originalImage, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ResizeBilinearGrad", scope.makeOpName("ResizeBilinearGrad"));
-    opBuilder.addInput(grads.asOutput());
-    opBuilder.addInput(originalImage.asOutput());
+    opBuilder.addInput(grads.asOutput(scope));
+    opBuilder.addInput(originalImage.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -119,7 +118,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ResizeNearestNeighbor.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ResizeNearestNeighbor.java
index 5ee41454f58..a7a8ceca11b 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ResizeNearestNeighbor.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ResizeNearestNeighbor.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -35,7 +34,7 @@
  * @param <T> data type for {@code resizedImages()} output
  */
 @Operator(group = "image")
-public final class ResizeNearestNeighbor<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class ResizeNearestNeighbor<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.image.ResizeNearestNeighbor}
@@ -77,10 +76,10 @@ private Options() {
    * @return a new instance of ResizeNearestNeighbor
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> ResizeNearestNeighbor<T> create(Scope scope, Operand<T> images, Operand<TInt32> size, Options... options) {
+  public static <T extends TNumber> ResizeNearestNeighbor<T> create(Scope scope, Operand<T> images, Operand<TInt32> size, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ResizeNearestNeighbor", scope.makeOpName("ResizeNearestNeighbor"));
-    opBuilder.addInput(images.asOutput());
-    opBuilder.addInput(size.asOutput());
+    opBuilder.addInput(images.asOutput(scope));
+    opBuilder.addInput(size.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -119,7 +118,7 @@ public Output<T> resizedImages() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return resizedImages;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ResizeNearestNeighborGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ResizeNearestNeighborGrad.java
index dafa83b74c6..fdee4f35ace 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ResizeNearestNeighborGrad.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ResizeNearestNeighborGrad.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -34,7 +33,7 @@
  * 
  * @param <T> data type for {@code output()} output
  */
-public final class ResizeNearestNeighborGrad<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class ResizeNearestNeighborGrad<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.image.ResizeNearestNeighborGrad}
@@ -76,10 +75,10 @@ private Options() {
    * @return a new instance of ResizeNearestNeighborGrad
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> ResizeNearestNeighborGrad<T> create(Scope scope, Operand<T> grads, Operand<TInt32> size, Options... options) {
+  public static <T extends TNumber> ResizeNearestNeighborGrad<T> create(Scope scope, Operand<T> grads, Operand<TInt32> size, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ResizeNearestNeighborGrad", scope.makeOpName("ResizeNearestNeighborGrad"));
-    opBuilder.addInput(grads.asOutput());
-    opBuilder.addInput(size.asOutput());
+    opBuilder.addInput(grads.asOutput(scope));
+    opBuilder.addInput(size.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -118,7 +117,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/RgbToHsv.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/RgbToHsv.java
index 87d08b14b08..9e86f495a7d 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/RgbToHsv.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/RgbToHsv.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -54,7 +53,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "image")
-public final class RgbToHsv<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class RgbToHsv<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new RgbToHsv operation.
@@ -64,9 +63,9 @@ public final class RgbToHsv<T extends Tensor & TNumber> extends RawOp implements
    * @return a new instance of RgbToHsv
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> RgbToHsv<T> create(Scope scope, Operand<T> images) {
+  public static <T extends TNumber> RgbToHsv<T> create(Scope scope, Operand<T> images) {
     OperationBuilder opBuilder = scope.env().opBuilder("RGBToHSV", scope.makeOpName("RgbToHsv"));
-    opBuilder.addInput(images.asOutput());
+    opBuilder.addInput(images.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new RgbToHsv<T>(opBuilder.build());
   }
@@ -79,7 +78,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/SampleDistortedBoundingBox.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/SampleDistortedBoundingBox.java
index e9c391b177e..fe4bc34d9bd 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/SampleDistortedBoundingBox.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/SampleDistortedBoundingBox.java
@@ -22,7 +22,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -73,7 +72,7 @@
  * @param <T> data type for {@code begin()} output
  */
 @Operator(group = "image")
-public final class SampleDistortedBoundingBox<T extends Tensor & TNumber> extends RawOp {
+public final class SampleDistortedBoundingBox<T extends TNumber> extends RawOp {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.image.SampleDistortedBoundingBox}
@@ -162,11 +161,11 @@ private Options() {
    * @return a new instance of SampleDistortedBoundingBox
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> SampleDistortedBoundingBox<T> create(Scope scope, Operand<T> imageSize, Operand<TFloat32> boundingBoxes, Operand<TFloat32> minObjectCovered, Options... options) {
+  public static <T extends TNumber> SampleDistortedBoundingBox<T> create(Scope scope, Operand<T> imageSize, Operand<TFloat32> boundingBoxes, Operand<TFloat32> minObjectCovered, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("SampleDistortedBoundingBoxV2", scope.makeOpName("SampleDistortedBoundingBox"));
-    opBuilder.addInput(imageSize.asOutput());
-    opBuilder.addInput(boundingBoxes.asOutput());
-    opBuilder.addInput(minObjectCovered.asOutput());
+    opBuilder.addInput(imageSize.asOutput(scope));
+    opBuilder.addInput(boundingBoxes.asOutput(scope));
+    opBuilder.addInput(minObjectCovered.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ScaleAndTranslate.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ScaleAndTranslate.java
index e0c7f3bc22a..12928534943 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ScaleAndTranslate.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ScaleAndTranslate.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -75,12 +74,12 @@ private Options() {
    * @return a new instance of ScaleAndTranslate
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> ScaleAndTranslate create(Scope scope, Operand<T> images, Operand<TInt32> size, Operand<TFloat32> scale, Operand<TFloat32> translation, Options... options) {
+  public static <T extends TNumber> ScaleAndTranslate create(Scope scope, Operand<T> images, Operand<TInt32> size, Operand<TFloat32> scale, Operand<TFloat32> translation, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ScaleAndTranslate", scope.makeOpName("ScaleAndTranslate"));
-    opBuilder.addInput(images.asOutput());
-    opBuilder.addInput(size.asOutput());
-    opBuilder.addInput(scale.asOutput());
-    opBuilder.addInput(translation.asOutput());
+    opBuilder.addInput(images.asOutput(scope));
+    opBuilder.addInput(size.asOutput(scope));
+    opBuilder.addInput(scale.asOutput(scope));
+    opBuilder.addInput(translation.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -116,7 +115,7 @@ public Output<TFloat32> resizedImages() {
   }
   
   @Override
-  public Output<TFloat32> asOutput() {
+  public Output<TFloat32> asOutput(Scope scope) {
     return resizedImages;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ScaleAndTranslateGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ScaleAndTranslateGrad.java
index b02e33bbfe0..2c0c858b38c 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ScaleAndTranslateGrad.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ScaleAndTranslateGrad.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -32,7 +31,7 @@
 /**
  * @param <T> data type for {@code output()} output
  */
-public final class ScaleAndTranslateGrad<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class ScaleAndTranslateGrad<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.image.ScaleAndTranslateGrad}
@@ -74,12 +73,12 @@ private Options() {
    * @return a new instance of ScaleAndTranslateGrad
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> ScaleAndTranslateGrad<T> create(Scope scope, Operand<T> grads, Operand<T> originalImage, Operand<TFloat32> scale, Operand<TFloat32> translation, Options... options) {
+  public static <T extends TNumber> ScaleAndTranslateGrad<T> create(Scope scope, Operand<T> grads, Operand<T> originalImage, Operand<TFloat32> scale, Operand<TFloat32> translation, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ScaleAndTranslateGrad", scope.makeOpName("ScaleAndTranslateGrad"));
-    opBuilder.addInput(grads.asOutput());
-    opBuilder.addInput(originalImage.asOutput());
-    opBuilder.addInput(scale.asOutput());
-    opBuilder.addInput(translation.asOutput());
+    opBuilder.addInput(grads.asOutput(scope));
+    opBuilder.addInput(originalImage.asOutput(scope));
+    opBuilder.addInput(scale.asOutput(scope));
+    opBuilder.addInput(translation.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -115,7 +114,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/DecodeBase64.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/DecodeBase64.java
index db026fd820c..4545ccd70c9 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/DecodeBase64.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/DecodeBase64.java
@@ -46,7 +46,7 @@ public final class DecodeBase64 extends RawOp implements Operand<TString> {
   @Endpoint(describeByClass = true)
   public static DecodeBase64 create(Scope scope, Operand<TString> input) {
     OperationBuilder opBuilder = scope.env().opBuilder("DecodeBase64", scope.makeOpName("DecodeBase64"));
-    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new DecodeBase64(opBuilder.build());
   }
@@ -59,7 +59,7 @@ public Output<TString> output() {
   }
   
   @Override
-  public Output<TString> asOutput() {
+  public Output<TString> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/DecodeCompressed.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/DecodeCompressed.java
index 745dba6d063..b314e3b31b5 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/DecodeCompressed.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/DecodeCompressed.java
@@ -71,7 +71,7 @@ private Options() {
   @Endpoint(describeByClass = true)
   public static DecodeCompressed create(Scope scope, Operand<TString> bytes, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("DecodeCompressed", scope.makeOpName("DecodeCompressed"));
-    opBuilder.addInput(bytes.asOutput());
+    opBuilder.addInput(bytes.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -100,7 +100,7 @@ public Output<TString> output() {
   }
   
   @Override
-  public Output<TString> asOutput() {
+  public Output<TString> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/DecodeCsv.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/DecodeCsv.java
index 3eb38c17efe..84fe9d1d814 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/DecodeCsv.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/DecodeCsv.java
@@ -24,13 +24,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.Operands;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TString;
+import org.tensorflow.types.family.TType;
 
 /**
  * Convert CSV records to tensors. Each column maps to one tensor.
@@ -40,7 +40,7 @@
  * Note that we allow leading and trailing spaces with int or float field.
  */
 @Operator(group = "io")
-public final class DecodeCsv extends RawOp implements Iterable<Operand<Tensor>> {
+public final class DecodeCsv extends RawOp implements Iterable<Operand<TType>> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.io.DecodeCsv}
@@ -105,8 +105,8 @@ private Options() {
   @Endpoint(describeByClass = true)
   public static DecodeCsv create(Scope scope, Operand<TString> records, Iterable<Operand<?>> recordDefaults, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("DecodeCSV", scope.makeOpName("DecodeCsv"));
-    opBuilder.addInput(records.asOutput());
-    opBuilder.addInputList(Operands.asOutputs(recordDefaults));
+    opBuilder.addInput(records.asOutput(scope));
+    opBuilder.addInputList(Operands.asOutputs(scope, recordDefaults));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -170,7 +170,7 @@ public List<Output<?>> output() {
   
   @Override
   @SuppressWarnings({"rawtypes", "unchecked"})
-  public Iterator<Operand<Tensor>> iterator() {
+  public Iterator<Operand<TType>> iterator() {
     return (Iterator) output.iterator();
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/DecodeJsonExample.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/DecodeJsonExample.java
index f3da58ca98d..c46049e8e9b 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/DecodeJsonExample.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/DecodeJsonExample.java
@@ -51,7 +51,7 @@ public final class DecodeJsonExample extends RawOp implements Operand<TString> {
   @Endpoint(describeByClass = true)
   public static DecodeJsonExample create(Scope scope, Operand<TString> jsonExamples) {
     OperationBuilder opBuilder = scope.env().opBuilder("DecodeJSONExample", scope.makeOpName("DecodeJsonExample"));
-    opBuilder.addInput(jsonExamples.asOutput());
+    opBuilder.addInput(jsonExamples.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new DecodeJsonExample(opBuilder.build());
   }
@@ -65,7 +65,7 @@ public Output<TString> binaryExamples() {
   }
   
   @Override
-  public Output<TString> asOutput() {
+  public Output<TString> asOutput(Scope scope) {
     return binaryExamples;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/DecodePaddedRaw.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/DecodePaddedRaw.java
index ed4845e4216..6eb973f6e01 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/DecodePaddedRaw.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/DecodePaddedRaw.java
@@ -22,7 +22,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -37,7 +36,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "io")
-public final class DecodePaddedRaw<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class DecodePaddedRaw<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.io.DecodePaddedRaw}
@@ -71,10 +70,10 @@ private Options() {
    * @return a new instance of DecodePaddedRaw
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> DecodePaddedRaw<T> create(Scope scope, Operand<TString> inputBytes, Operand<TInt32> fixedLength, DataType<T> outType, Options... options) {
+  public static <T extends TNumber> DecodePaddedRaw<T> create(Scope scope, Operand<TString> inputBytes, Operand<TInt32> fixedLength, DataType<T> outType, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("DecodePaddedRaw", scope.makeOpName("DecodePaddedRaw"));
-    opBuilder.addInput(inputBytes.asOutput());
-    opBuilder.addInput(fixedLength.asOutput());
+    opBuilder.addInput(inputBytes.asOutput(scope));
+    opBuilder.addInput(fixedLength.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("out_type", outType);
     if (options != null) {
@@ -105,7 +104,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/DecodeRaw.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/DecodeRaw.java
index 922d7dbfb44..ec062e780a3 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/DecodeRaw.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/DecodeRaw.java
@@ -22,12 +22,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TString;
+import org.tensorflow.types.family.TType;
 
 /**
  * Reinterpret the bytes of a string as a vector of numbers.
@@ -35,7 +35,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "io")
-public final class DecodeRaw<T extends Tensor> extends RawOp implements Operand<T> {
+public final class DecodeRaw<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.io.DecodeRaw}
@@ -68,9 +68,9 @@ private Options() {
    * @return a new instance of DecodeRaw
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> DecodeRaw<T> create(Scope scope, Operand<TString> bytes, DataType<T> outType, Options... options) {
+  public static <T extends TType> DecodeRaw<T> create(Scope scope, Operand<TString> bytes, DataType<T> outType, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("DecodeRaw", scope.makeOpName("DecodeRaw"));
-    opBuilder.addInput(bytes.asOutput());
+    opBuilder.addInput(bytes.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("out_type", outType);
     if (options != null) {
@@ -102,7 +102,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/DeserializeManySparse.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/DeserializeManySparse.java
index c969ec7da77..d31ab4de4d0 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/DeserializeManySparse.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/DeserializeManySparse.java
@@ -22,13 +22,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
 import org.tensorflow.types.TString;
+import org.tensorflow.types.family.TType;
 
 /**
  * Deserialize and concatenate `SparseTensors` from a serialized minibatch.
@@ -78,7 +78,7 @@
  * @param <T> data type for {@code sparseValues()} output
  */
 @Operator(group = "io")
-public final class DeserializeManySparse<T extends Tensor> extends RawOp {
+public final class DeserializeManySparse<T extends TType> extends RawOp {
   
   /**
    * Factory method to create a class wrapping a new DeserializeManySparse operation.
@@ -90,9 +90,9 @@ public final class DeserializeManySparse<T extends Tensor> extends RawOp {
    * @return a new instance of DeserializeManySparse
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> DeserializeManySparse<T> create(Scope scope, Operand<TString> serializedSparse, DataType<T> dtype) {
+  public static <T extends TType> DeserializeManySparse<T> create(Scope scope, Operand<TString> serializedSparse, DataType<T> dtype) {
     OperationBuilder opBuilder = scope.env().opBuilder("DeserializeManySparse", scope.makeOpName("DeserializeManySparse"));
-    opBuilder.addInput(serializedSparse.asOutput());
+    opBuilder.addInput(serializedSparse.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("dtype", dtype);
     return new DeserializeManySparse<T>(opBuilder.build());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/EncodeBase64.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/EncodeBase64.java
index ddb331b4a48..219ad0c422b 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/EncodeBase64.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/EncodeBase64.java
@@ -70,7 +70,7 @@ private Options() {
   @Endpoint(describeByClass = true)
   public static EncodeBase64 create(Scope scope, Operand<TString> input, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("EncodeBase64", scope.makeOpName("EncodeBase64"));
-    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -97,7 +97,7 @@ public Output<TString> output() {
   }
   
   @Override
-  public Output<TString> asOutput() {
+  public Output<TString> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/FifoQueue.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/FifoQueue.java
index 1e897d50624..e7509130d50 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/FifoQueue.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/FifoQueue.java
@@ -23,18 +23,18 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * A queue that produces elements in first-in first-out order.
  */
 @Operator(group = "io")
-public final class FifoQueue extends RawOp implements Operand<Tensor> {
+public final class FifoQueue extends RawOp implements Operand<TType> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.io.FifoQueue}
@@ -171,8 +171,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<Tensor> asOutput() {
-    return (Output<Tensor>) handle;
+  public Output<TType> asOutput(Scope scope) {
+    return (Output<TType>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/FixedLengthRecordReader.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/FixedLengthRecordReader.java
index ba1a213dcd2..2cc926ea765 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/FixedLengthRecordReader.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/FixedLengthRecordReader.java
@@ -21,17 +21,17 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * A Reader that outputs fixed-length records from a file.
  */
 @Operator(group = "io")
-public final class FixedLengthRecordReader extends RawOp implements Operand<Tensor> {
+public final class FixedLengthRecordReader extends RawOp implements Operand<TType> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.io.FixedLengthRecordReader}
@@ -194,8 +194,8 @@ public Output<?> readerHandle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<Tensor> asOutput() {
-    return (Output<Tensor>) readerHandle;
+  public Output<TType> asOutput(Scope scope) {
+    return (Output<TType>) readerHandle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/IdentityReader.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/IdentityReader.java
index f6af2da4b47..5649e2da49f 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/IdentityReader.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/IdentityReader.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * A Reader that outputs the queued work as both the key and value.
@@ -34,7 +34,7 @@
  * work string and output (work, work).
  */
 @Operator(group = "io")
-public final class IdentityReader extends RawOp implements Operand<Tensor> {
+public final class IdentityReader extends RawOp implements Operand<TType> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.io.IdentityReader}
@@ -115,8 +115,8 @@ public Output<?> readerHandle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<Tensor> asOutput() {
-    return (Output<Tensor>) readerHandle;
+  public Output<TType> asOutput(Scope scope) {
+    return (Output<TType>) readerHandle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/LmdbReader.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/LmdbReader.java
index b3b10b2809c..affdd871f44 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/LmdbReader.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/LmdbReader.java
@@ -111,7 +111,7 @@ public Output<TString> readerHandle() {
   }
   
   @Override
-  public Output<TString> asOutput() {
+  public Output<TString> asOutput(Scope scope) {
     return readerHandle;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/MatchingFiles.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/MatchingFiles.java
index ed2b33adea3..7441e1bd9eb 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/MatchingFiles.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/MatchingFiles.java
@@ -47,7 +47,7 @@ public final class MatchingFiles extends RawOp implements Operand<TString> {
   @Endpoint(describeByClass = true)
   public static MatchingFiles create(Scope scope, Operand<TString> pattern) {
     OperationBuilder opBuilder = scope.env().opBuilder("MatchingFiles", scope.makeOpName("MatchingFiles"));
-    opBuilder.addInput(pattern.asOutput());
+    opBuilder.addInput(pattern.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new MatchingFiles(opBuilder.build());
   }
@@ -60,7 +60,7 @@ public Output<TString> filenames() {
   }
   
   @Override
-  public Output<TString> asOutput() {
+  public Output<TString> asOutput(Scope scope) {
     return filenames;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/PaddingFifoQueue.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/PaddingFifoQueue.java
index 712be3c7660..5fb8d2c4134 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/PaddingFifoQueue.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/PaddingFifoQueue.java
@@ -23,12 +23,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * A queue that produces elements in first-in first-out order.
@@ -38,7 +38,7 @@
  * size of any given element in the minibatch.  See below for details.
  */
 @Operator(group = "io")
-public final class PaddingFifoQueue extends RawOp implements Operand<Tensor> {
+public final class PaddingFifoQueue extends RawOp implements Operand<TType> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.io.PaddingFifoQueue}
@@ -183,8 +183,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<Tensor> asOutput() {
-    return (Output<Tensor>) handle;
+  public Output<TType> asOutput(Scope scope) {
+    return (Output<TType>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ParseExample.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ParseExample.java
index bd8f0f4a603..c6c5be5046e 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ParseExample.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ParseExample.java
@@ -32,7 +32,6 @@
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
 import org.tensorflow.types.TString;
-import org.tensorflow.types.family.TNumber;
 
 /**
  * Transforms a vector of tf.Example protos (as strings) into typed tensors.
@@ -99,12 +98,12 @@ public final class ParseExample extends RawOp {
   @Endpoint(describeByClass = true)
   public static ParseExample create(Scope scope, Operand<TString> serialized, Operand<TString> names, Operand<TString> sparseKeys, Operand<TString> denseKeys, Operand<TString> raggedKeys, Iterable<Operand<?>> denseDefaults, Long numSparse, List<DataType<?>> sparseTypes, List<DataType<?>> raggedValueTypes, List<DataType<?>> raggedSplitTypes, List<Shape> denseShapes) {
     OperationBuilder opBuilder = scope.env().opBuilder("ParseExampleV2", scope.makeOpName("ParseExample"));
-    opBuilder.addInput(serialized.asOutput());
-    opBuilder.addInput(names.asOutput());
-    opBuilder.addInput(sparseKeys.asOutput());
-    opBuilder.addInput(denseKeys.asOutput());
-    opBuilder.addInput(raggedKeys.asOutput());
-    opBuilder.addInputList(Operands.asOutputs(denseDefaults));
+    opBuilder.addInput(serialized.asOutput(scope));
+    opBuilder.addInput(names.asOutput(scope));
+    opBuilder.addInput(sparseKeys.asOutput(scope));
+    opBuilder.addInput(denseKeys.asOutput(scope));
+    opBuilder.addInput(raggedKeys.asOutput(scope));
+    opBuilder.addInputList(Operands.asOutputs(scope, denseDefaults));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("num_sparse", numSparse);
     DataType[] sparseTypesArray = new DataType[sparseTypes.size()];
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ParseSequenceExample.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ParseSequenceExample.java
index 1c10b138c55..d1a60de633f 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ParseSequenceExample.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ParseSequenceExample.java
@@ -33,7 +33,6 @@
 import org.tensorflow.types.TBool;
 import org.tensorflow.types.TInt64;
 import org.tensorflow.types.TString;
-import org.tensorflow.types.family.TNumber;
 
 /**
  * Transforms a vector of tf.io.SequenceExample protos (as strings) into
@@ -155,16 +154,16 @@ private Options() {
   @Endpoint(describeByClass = true)
   public static ParseSequenceExample create(Scope scope, Operand<TString> serialized, Operand<TString> debugName, Operand<TString> contextSparseKeys, Operand<TString> contextDenseKeys, Operand<TString> contextRaggedKeys, Operand<TString> featureListSparseKeys, Operand<TString> featureListDenseKeys, Operand<TString> featureListRaggedKeys, Operand<TBool> featureListDenseMissingAssumedEmpty, Iterable<Operand<?>> contextDenseDefaults, List<DataType<?>> contextSparseTypes, List<DataType<?>> contextRaggedValueTypes, List<DataType<?>> contextRaggedSplitTypes, List<DataType<?>> featureListDenseTypes, List<DataType<?>> featureListSparseTypes, List<DataType<?>> featureListRaggedValueTypes, List<DataType<?>> featureListRaggedSplitTypes, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ParseSequenceExampleV2", scope.makeOpName("ParseSequenceExample"));
-    opBuilder.addInput(serialized.asOutput());
-    opBuilder.addInput(debugName.asOutput());
-    opBuilder.addInput(contextSparseKeys.asOutput());
-    opBuilder.addInput(contextDenseKeys.asOutput());
-    opBuilder.addInput(contextRaggedKeys.asOutput());
-    opBuilder.addInput(featureListSparseKeys.asOutput());
-    opBuilder.addInput(featureListDenseKeys.asOutput());
-    opBuilder.addInput(featureListRaggedKeys.asOutput());
-    opBuilder.addInput(featureListDenseMissingAssumedEmpty.asOutput());
-    opBuilder.addInputList(Operands.asOutputs(contextDenseDefaults));
+    opBuilder.addInput(serialized.asOutput(scope));
+    opBuilder.addInput(debugName.asOutput(scope));
+    opBuilder.addInput(contextSparseKeys.asOutput(scope));
+    opBuilder.addInput(contextDenseKeys.asOutput(scope));
+    opBuilder.addInput(contextRaggedKeys.asOutput(scope));
+    opBuilder.addInput(featureListSparseKeys.asOutput(scope));
+    opBuilder.addInput(featureListDenseKeys.asOutput(scope));
+    opBuilder.addInput(featureListRaggedKeys.asOutput(scope));
+    opBuilder.addInput(featureListDenseMissingAssumedEmpty.asOutput(scope));
+    opBuilder.addInputList(Operands.asOutputs(scope, contextDenseDefaults));
     opBuilder = scope.applyControlDependencies(opBuilder);
     DataType[] contextSparseTypesArray = new DataType[contextSparseTypes.size()];
     for (int i = 0; i < contextSparseTypesArray.length; ++i) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ParseSingleExample.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ParseSingleExample.java
index 2c09ba8ac4c..c05bd4f21ce 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ParseSingleExample.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ParseSingleExample.java
@@ -78,8 +78,8 @@ public final class ParseSingleExample extends RawOp {
   @Endpoint(describeByClass = true)
   public static ParseSingleExample create(Scope scope, Operand<TString> serialized, Iterable<Operand<?>> denseDefaults, Long numSparse, List<String> sparseKeys, List<String> denseKeys, List<DataType<?>> sparseTypes, List<Shape> denseShapes) {
     OperationBuilder opBuilder = scope.env().opBuilder("ParseSingleExample", scope.makeOpName("ParseSingleExample"));
-    opBuilder.addInput(serialized.asOutput());
-    opBuilder.addInputList(Operands.asOutputs(denseDefaults));
+    opBuilder.addInput(serialized.asOutput(scope));
+    opBuilder.addInputList(Operands.asOutputs(scope, denseDefaults));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("num_sparse", numSparse);
     String[] sparseKeysArray = new String[sparseKeys.size()];
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ParseSingleSequenceExample.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ParseSingleSequenceExample.java
index bb1e1eaa576..96a8bd9c413 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ParseSingleSequenceExample.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ParseSingleSequenceExample.java
@@ -124,14 +124,14 @@ private Options() {
   @Endpoint(describeByClass = true)
   public static ParseSingleSequenceExample create(Scope scope, Operand<TString> serialized, Operand<TString> featureListDenseMissingAssumedEmpty, Iterable<Operand<TString>> contextSparseKeys, Iterable<Operand<TString>> contextDenseKeys, Iterable<Operand<TString>> featureListSparseKeys, Iterable<Operand<TString>> featureListDenseKeys, Iterable<Operand<?>> contextDenseDefaults, Operand<TString> debugName, List<DataType<?>> contextSparseTypes, List<DataType<?>> featureListDenseTypes, List<DataType<?>> featureListSparseTypes, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ParseSingleSequenceExample", scope.makeOpName("ParseSingleSequenceExample"));
-    opBuilder.addInput(serialized.asOutput());
-    opBuilder.addInput(featureListDenseMissingAssumedEmpty.asOutput());
-    opBuilder.addInputList(Operands.asOutputs(contextSparseKeys));
-    opBuilder.addInputList(Operands.asOutputs(contextDenseKeys));
-    opBuilder.addInputList(Operands.asOutputs(featureListSparseKeys));
-    opBuilder.addInputList(Operands.asOutputs(featureListDenseKeys));
-    opBuilder.addInputList(Operands.asOutputs(contextDenseDefaults));
-    opBuilder.addInput(debugName.asOutput());
+    opBuilder.addInput(serialized.asOutput(scope));
+    opBuilder.addInput(featureListDenseMissingAssumedEmpty.asOutput(scope));
+    opBuilder.addInputList(Operands.asOutputs(scope, contextSparseKeys));
+    opBuilder.addInputList(Operands.asOutputs(scope, contextDenseKeys));
+    opBuilder.addInputList(Operands.asOutputs(scope, featureListSparseKeys));
+    opBuilder.addInputList(Operands.asOutputs(scope, featureListDenseKeys));
+    opBuilder.addInputList(Operands.asOutputs(scope, contextDenseDefaults));
+    opBuilder.addInput(debugName.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     DataType[] contextSparseTypesArray = new DataType[contextSparseTypes.size()];
     for (int i = 0; i < contextSparseTypesArray.length; ++i) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ParseTensor.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ParseTensor.java
index 7d47f9380f6..1fe2d9cebaf 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ParseTensor.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ParseTensor.java
@@ -22,12 +22,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TString;
+import org.tensorflow.types.family.TType;
 
 /**
  * Transforms a serialized tensorflow.TensorProto proto into a Tensor.
@@ -35,7 +35,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "io")
-public final class ParseTensor<T extends Tensor> extends RawOp implements Operand<T> {
+public final class ParseTensor<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new ParseTensor operation.
@@ -47,9 +47,9 @@ public final class ParseTensor<T extends Tensor> extends RawOp implements Operan
    * @return a new instance of ParseTensor
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> ParseTensor<T> create(Scope scope, Operand<TString> serialized, DataType<T> outType) {
+  public static <T extends TType> ParseTensor<T> create(Scope scope, Operand<TString> serialized, DataType<T> outType) {
     OperationBuilder opBuilder = scope.env().opBuilder("ParseTensor", scope.makeOpName("ParseTensor"));
-    opBuilder.addInput(serialized.asOutput());
+    opBuilder.addInput(serialized.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("out_type", outType);
     return new ParseTensor<T>(opBuilder.build());
@@ -63,7 +63,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/PriorityQueue.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/PriorityQueue.java
index 71d5aee9591..f3abb008e73 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/PriorityQueue.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/PriorityQueue.java
@@ -23,12 +23,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * A queue that produces elements sorted by the first component value.
@@ -40,7 +40,7 @@
  * entry in their input (resp. output) lists.
  */
 @Operator(group = "io")
-public final class PriorityQueue extends RawOp implements Operand<Tensor> {
+public final class PriorityQueue extends RawOp implements Operand<TType> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.io.PriorityQueue}
@@ -157,8 +157,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<Tensor> asOutput() {
-    return (Output<Tensor>) handle;
+  public Output<TType> asOutput(Scope scope) {
+    return (Output<TType>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/QueueClose.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/QueueClose.java
index ea7791f143e..dcbace2ecd3 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/QueueClose.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/QueueClose.java
@@ -68,7 +68,7 @@ private Options() {
   @Endpoint(describeByClass = true)
   public static QueueClose create(Scope scope, Operand<?> handle, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("QueueCloseV2", scope.makeOpName("QueueClose"));
-    opBuilder.addInput(handle.asOutput());
+    opBuilder.addInput(handle.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/QueueDequeue.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/QueueDequeue.java
index 52a1f40d890..2da189694d7 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/QueueDequeue.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/QueueDequeue.java
@@ -25,11 +25,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Dequeues a tuple of one or more tensors from the given queue.
@@ -42,7 +42,7 @@
  * has been dequeued (or 'timeout_ms' elapses, if specified).
  */
 @Operator(group = "io")
-public final class QueueDequeue extends RawOp implements Iterable<Operand<Tensor>> {
+public final class QueueDequeue extends RawOp implements Iterable<Operand<TType>> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.io.QueueDequeue}
@@ -77,7 +77,7 @@ private Options() {
   @Endpoint(describeByClass = true)
   public static QueueDequeue create(Scope scope, Operand<?> handle, List<DataType<?>> componentTypes, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("QueueDequeueV2", scope.makeOpName("QueueDequeue"));
-    opBuilder.addInput(handle.asOutput());
+    opBuilder.addInput(handle.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     DataType[] componentTypesArray = new DataType[componentTypes.size()];
     for (int i = 0; i < componentTypesArray.length; ++i) {
@@ -112,7 +112,7 @@ public List<Output<?>> components() {
   
   @Override
   @SuppressWarnings({"rawtypes", "unchecked"})
-  public Iterator<Operand<Tensor>> iterator() {
+  public Iterator<Operand<TType>> iterator() {
     return (Iterator) components.iterator();
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/QueueDequeueMany.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/QueueDequeueMany.java
index 770baa857d3..18d79dfa2de 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/QueueDequeueMany.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/QueueDequeueMany.java
@@ -25,12 +25,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TType;
 
 /**
  * Dequeues `n` tuples of one or more tensors from the given queue.
@@ -50,7 +50,7 @@
  * have been dequeued (or 'timeout_ms' elapses, if specified).
  */
 @Operator(group = "io")
-public final class QueueDequeueMany extends RawOp implements Iterable<Operand<Tensor>> {
+public final class QueueDequeueMany extends RawOp implements Iterable<Operand<TType>> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.io.QueueDequeueMany}
@@ -86,8 +86,8 @@ private Options() {
   @Endpoint(describeByClass = true)
   public static QueueDequeueMany create(Scope scope, Operand<?> handle, Operand<TInt32> n, List<DataType<?>> componentTypes, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("QueueDequeueManyV2", scope.makeOpName("QueueDequeueMany"));
-    opBuilder.addInput(handle.asOutput());
-    opBuilder.addInput(n.asOutput());
+    opBuilder.addInput(handle.asOutput(scope));
+    opBuilder.addInput(n.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     DataType[] componentTypesArray = new DataType[componentTypes.size()];
     for (int i = 0; i < componentTypesArray.length; ++i) {
@@ -122,7 +122,7 @@ public List<Output<?>> components() {
   
   @Override
   @SuppressWarnings({"rawtypes", "unchecked"})
-  public Iterator<Operand<Tensor>> iterator() {
+  public Iterator<Operand<TType>> iterator() {
     return (Iterator) components.iterator();
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/QueueDequeueUpTo.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/QueueDequeueUpTo.java
index 7df58e083fd..4358a202318 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/QueueDequeueUpTo.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/QueueDequeueUpTo.java
@@ -25,12 +25,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TType;
 
 /**
  * Dequeues `n` tuples of one or more tensors from the given queue.
@@ -54,7 +54,7 @@
  * component of the dequeued tuple.
  */
 @Operator(group = "io")
-public final class QueueDequeueUpTo extends RawOp implements Iterable<Operand<Tensor>> {
+public final class QueueDequeueUpTo extends RawOp implements Iterable<Operand<TType>> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.io.QueueDequeueUpTo}
@@ -90,8 +90,8 @@ private Options() {
   @Endpoint(describeByClass = true)
   public static QueueDequeueUpTo create(Scope scope, Operand<?> handle, Operand<TInt32> n, List<DataType<?>> componentTypes, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("QueueDequeueUpToV2", scope.makeOpName("QueueDequeueUpTo"));
-    opBuilder.addInput(handle.asOutput());
-    opBuilder.addInput(n.asOutput());
+    opBuilder.addInput(handle.asOutput(scope));
+    opBuilder.addInput(n.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     DataType[] componentTypesArray = new DataType[componentTypes.size()];
     for (int i = 0; i < componentTypesArray.length; ++i) {
@@ -126,7 +126,7 @@ public List<Output<?>> components() {
   
   @Override
   @SuppressWarnings({"rawtypes", "unchecked"})
-  public Iterator<Operand<Tensor>> iterator() {
+  public Iterator<Operand<TType>> iterator() {
     return (Iterator) components.iterator();
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/QueueEnqueue.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/QueueEnqueue.java
index a159b0cd17c..b37f9d4f848 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/QueueEnqueue.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/QueueEnqueue.java
@@ -71,8 +71,8 @@ private Options() {
   @Endpoint(describeByClass = true)
   public static QueueEnqueue create(Scope scope, Operand<?> handle, Iterable<Operand<?>> components, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("QueueEnqueueV2", scope.makeOpName("QueueEnqueue"));
-    opBuilder.addInput(handle.asOutput());
-    opBuilder.addInputList(Operands.asOutputs(components));
+    opBuilder.addInput(handle.asOutput(scope));
+    opBuilder.addInputList(Operands.asOutputs(scope, components));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/QueueEnqueueMany.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/QueueEnqueueMany.java
index b1f9cbd6807..b308e7aa7d2 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/QueueEnqueueMany.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/QueueEnqueueMany.java
@@ -76,8 +76,8 @@ private Options() {
   @Endpoint(describeByClass = true)
   public static QueueEnqueueMany create(Scope scope, Operand<?> handle, Iterable<Operand<?>> components, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("QueueEnqueueManyV2", scope.makeOpName("QueueEnqueueMany"));
-    opBuilder.addInput(handle.asOutput());
-    opBuilder.addInputList(Operands.asOutputs(components));
+    opBuilder.addInput(handle.asOutput(scope));
+    opBuilder.addInputList(Operands.asOutputs(scope, components));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/QueueIsClosed.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/QueueIsClosed.java
index 895965a3c6c..0cb6ac9997b 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/QueueIsClosed.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/QueueIsClosed.java
@@ -46,7 +46,7 @@ public final class QueueIsClosed extends RawOp implements Operand<TBool> {
   @Endpoint(describeByClass = true)
   public static QueueIsClosed create(Scope scope, Operand<?> handle) {
     OperationBuilder opBuilder = scope.env().opBuilder("QueueIsClosedV2", scope.makeOpName("QueueIsClosed"));
-    opBuilder.addInput(handle.asOutput());
+    opBuilder.addInput(handle.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new QueueIsClosed(opBuilder.build());
   }
@@ -58,7 +58,7 @@ public Output<TBool> isClosed() {
   }
   
   @Override
-  public Output<TBool> asOutput() {
+  public Output<TBool> asOutput(Scope scope) {
     return isClosed;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/QueueSize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/QueueSize.java
index cefc1b3fd11..cfb5c24dbaa 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/QueueSize.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/QueueSize.java
@@ -43,7 +43,7 @@ public final class QueueSize extends RawOp implements Operand<TInt32> {
   @Endpoint(describeByClass = true)
   public static QueueSize create(Scope scope, Operand<?> handle) {
     OperationBuilder opBuilder = scope.env().opBuilder("QueueSizeV2", scope.makeOpName("QueueSize"));
-    opBuilder.addInput(handle.asOutput());
+    opBuilder.addInput(handle.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new QueueSize(opBuilder.build());
   }
@@ -56,7 +56,7 @@ public Output<TInt32> size() {
   }
   
   @Override
-  public Output<TInt32> asOutput() {
+  public Output<TInt32> asOutput(Scope scope) {
     return size;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/RandomShuffleQueue.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/RandomShuffleQueue.java
index cf81915b80a..42bfd44f9ad 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/RandomShuffleQueue.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/RandomShuffleQueue.java
@@ -23,18 +23,18 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * A queue that randomizes the order of elements.
  */
 @Operator(group = "io")
-public final class RandomShuffleQueue extends RawOp implements Operand<Tensor> {
+public final class RandomShuffleQueue extends RawOp implements Operand<TType> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.io.RandomShuffleQueue}
@@ -234,8 +234,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<Tensor> asOutput() {
-    return (Output<Tensor>) handle;
+  public Output<TType> asOutput(Scope scope) {
+    return (Output<TType>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ReadFile.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ReadFile.java
index 7109022751a..0911125282c 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ReadFile.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ReadFile.java
@@ -43,7 +43,7 @@ public final class ReadFile extends RawOp implements Operand<TString> {
   @Endpoint(describeByClass = true)
   public static ReadFile create(Scope scope, Operand<TString> filename) {
     OperationBuilder opBuilder = scope.env().opBuilder("ReadFile", scope.makeOpName("ReadFile"));
-    opBuilder.addInput(filename.asOutput());
+    opBuilder.addInput(filename.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new ReadFile(opBuilder.build());
   }
@@ -55,7 +55,7 @@ public Output<TString> contents() {
   }
   
   @Override
-  public Output<TString> asOutput() {
+  public Output<TString> asOutput(Scope scope) {
     return contents;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ReaderNumRecordsProduced.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ReaderNumRecordsProduced.java
index 8e4cb9c5692..25c000c9a5e 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ReaderNumRecordsProduced.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ReaderNumRecordsProduced.java
@@ -46,7 +46,7 @@ public final class ReaderNumRecordsProduced extends RawOp implements Operand<TIn
   @Endpoint(describeByClass = true)
   public static ReaderNumRecordsProduced create(Scope scope, Operand<?> readerHandle) {
     OperationBuilder opBuilder = scope.env().opBuilder("ReaderNumRecordsProducedV2", scope.makeOpName("ReaderNumRecordsProduced"));
-    opBuilder.addInput(readerHandle.asOutput());
+    opBuilder.addInput(readerHandle.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new ReaderNumRecordsProduced(opBuilder.build());
   }
@@ -58,7 +58,7 @@ public Output<TInt64> recordsProduced() {
   }
   
   @Override
-  public Output<TInt64> asOutput() {
+  public Output<TInt64> asOutput(Scope scope) {
     return recordsProduced;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ReaderNumWorkUnitsCompleted.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ReaderNumWorkUnitsCompleted.java
index bbd6fc8075c..3f4b161768c 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ReaderNumWorkUnitsCompleted.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ReaderNumWorkUnitsCompleted.java
@@ -43,7 +43,7 @@ public final class ReaderNumWorkUnitsCompleted extends RawOp implements Operand<
   @Endpoint(describeByClass = true)
   public static ReaderNumWorkUnitsCompleted create(Scope scope, Operand<?> readerHandle) {
     OperationBuilder opBuilder = scope.env().opBuilder("ReaderNumWorkUnitsCompletedV2", scope.makeOpName("ReaderNumWorkUnitsCompleted"));
-    opBuilder.addInput(readerHandle.asOutput());
+    opBuilder.addInput(readerHandle.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new ReaderNumWorkUnitsCompleted(opBuilder.build());
   }
@@ -55,7 +55,7 @@ public Output<TInt64> unitsCompleted() {
   }
   
   @Override
-  public Output<TInt64> asOutput() {
+  public Output<TInt64> asOutput(Scope scope) {
     return unitsCompleted;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ReaderRead.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ReaderRead.java
index 9b7110b10ae..7c52b881b02 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ReaderRead.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ReaderRead.java
@@ -48,8 +48,8 @@ public final class ReaderRead extends RawOp {
   @Endpoint(describeByClass = true)
   public static ReaderRead create(Scope scope, Operand<?> readerHandle, Operand<?> queueHandle) {
     OperationBuilder opBuilder = scope.env().opBuilder("ReaderReadV2", scope.makeOpName("ReaderRead"));
-    opBuilder.addInput(readerHandle.asOutput());
-    opBuilder.addInput(queueHandle.asOutput());
+    opBuilder.addInput(readerHandle.asOutput(scope));
+    opBuilder.addInput(queueHandle.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new ReaderRead(opBuilder.build());
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ReaderReadUpTo.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ReaderReadUpTo.java
index e22d5b95576..168ebf8db6a 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ReaderReadUpTo.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ReaderReadUpTo.java
@@ -51,9 +51,9 @@ public final class ReaderReadUpTo extends RawOp {
   @Endpoint(describeByClass = true)
   public static ReaderReadUpTo create(Scope scope, Operand<?> readerHandle, Operand<?> queueHandle, Operand<TInt64> numRecords) {
     OperationBuilder opBuilder = scope.env().opBuilder("ReaderReadUpToV2", scope.makeOpName("ReaderReadUpTo"));
-    opBuilder.addInput(readerHandle.asOutput());
-    opBuilder.addInput(queueHandle.asOutput());
-    opBuilder.addInput(numRecords.asOutput());
+    opBuilder.addInput(readerHandle.asOutput(scope));
+    opBuilder.addInput(queueHandle.asOutput(scope));
+    opBuilder.addInput(numRecords.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new ReaderReadUpTo(opBuilder.build());
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ReaderReset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ReaderReset.java
index 243d4a72080..ffb6898ac39 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ReaderReset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ReaderReset.java
@@ -41,7 +41,7 @@ public final class ReaderReset extends RawOp {
   @Endpoint(describeByClass = true)
   public static ReaderReset create(Scope scope, Operand<?> readerHandle) {
     OperationBuilder opBuilder = scope.env().opBuilder("ReaderResetV2", scope.makeOpName("ReaderReset"));
-    opBuilder.addInput(readerHandle.asOutput());
+    opBuilder.addInput(readerHandle.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new ReaderReset(opBuilder.build());
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ReaderRestoreState.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ReaderRestoreState.java
index 431ba079ffc..8ecfd87479e 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ReaderRestoreState.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ReaderRestoreState.java
@@ -47,8 +47,8 @@ public final class ReaderRestoreState extends RawOp {
   @Endpoint(describeByClass = true)
   public static ReaderRestoreState create(Scope scope, Operand<?> readerHandle, Operand<TString> state) {
     OperationBuilder opBuilder = scope.env().opBuilder("ReaderRestoreStateV2", scope.makeOpName("ReaderRestoreState"));
-    opBuilder.addInput(readerHandle.asOutput());
-    opBuilder.addInput(state.asOutput());
+    opBuilder.addInput(readerHandle.asOutput(scope));
+    opBuilder.addInput(state.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new ReaderRestoreState(opBuilder.build());
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ReaderSerializeState.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ReaderSerializeState.java
index 72b927ed171..affefa95016 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ReaderSerializeState.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ReaderSerializeState.java
@@ -46,7 +46,7 @@ public final class ReaderSerializeState extends RawOp implements Operand<TString
   @Endpoint(describeByClass = true)
   public static ReaderSerializeState create(Scope scope, Operand<?> readerHandle) {
     OperationBuilder opBuilder = scope.env().opBuilder("ReaderSerializeStateV2", scope.makeOpName("ReaderSerializeState"));
-    opBuilder.addInput(readerHandle.asOutput());
+    opBuilder.addInput(readerHandle.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new ReaderSerializeState(opBuilder.build());
   }
@@ -58,7 +58,7 @@ public Output<TString> state() {
   }
   
   @Override
-  public Output<TString> asOutput() {
+  public Output<TString> asOutput(Scope scope) {
     return state;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/SerializeManySparse.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/SerializeManySparse.java
index afab861c445..a29e431d44c 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/SerializeManySparse.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/SerializeManySparse.java
@@ -22,13 +22,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
 import org.tensorflow.types.TString;
+import org.tensorflow.types.family.TType;
 
 /**
  * Serialize an `N`-minibatch `SparseTensor` into an `[N, 3]` `Tensor` object.
@@ -44,7 +44,7 @@
  * @param <U> data type for {@code serializedSparse()} output
  */
 @Operator(group = "io")
-public final class SerializeManySparse<U extends Tensor> extends RawOp implements Operand<U> {
+public final class SerializeManySparse<U extends TType> extends RawOp implements Operand<U> {
   
   /**
    * Factory method to create a class wrapping a new SerializeManySparse operation.
@@ -58,11 +58,11 @@ public final class SerializeManySparse<U extends Tensor> extends RawOp implement
    * @return a new instance of SerializeManySparse
    */
   @Endpoint(describeByClass = true)
-  public static <U extends Tensor, T extends Tensor> SerializeManySparse<U> create(Scope scope, Operand<TInt64> sparseIndices, Operand<T> sparseValues, Operand<TInt64> sparseShape, DataType<U> outType) {
+  public static <U extends TType, T extends TType> SerializeManySparse<U> create(Scope scope, Operand<TInt64> sparseIndices, Operand<T> sparseValues, Operand<TInt64> sparseShape, DataType<U> outType) {
     OperationBuilder opBuilder = scope.env().opBuilder("SerializeManySparse", scope.makeOpName("SerializeManySparse"));
-    opBuilder.addInput(sparseIndices.asOutput());
-    opBuilder.addInput(sparseValues.asOutput());
-    opBuilder.addInput(sparseShape.asOutput());
+    opBuilder.addInput(sparseIndices.asOutput(scope));
+    opBuilder.addInput(sparseValues.asOutput(scope));
+    opBuilder.addInput(sparseShape.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("out_type", outType);
     return new SerializeManySparse<U>(opBuilder.build());
@@ -78,7 +78,7 @@ public static <U extends Tensor, T extends Tensor> SerializeManySparse<U> create
    * @return a new instance of SerializeManySparse
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> SerializeManySparse<TString> create(Scope scope, Operand<TInt64> sparseIndices, Operand<T> sparseValues, Operand<TInt64> sparseShape) {
+  public static <T extends TType> SerializeManySparse<TString> create(Scope scope, Operand<TInt64> sparseIndices, Operand<T> sparseValues, Operand<TInt64> sparseShape) {
     return create(scope, sparseIndices, sparseValues, sparseShape, TString.DTYPE);
   }
   
@@ -89,7 +89,7 @@ public Output<U> serializedSparse() {
   }
   
   @Override
-  public Output<U> asOutput() {
+  public Output<U> asOutput(Scope scope) {
     return serializedSparse;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/SerializeSparse.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/SerializeSparse.java
index 744d573981d..c4bdb3bbb74 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/SerializeSparse.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/SerializeSparse.java
@@ -22,13 +22,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
 import org.tensorflow.types.TString;
+import org.tensorflow.types.family.TType;
 
 /**
  * Serialize a `SparseTensor` into a `[3]` `Tensor` object.
@@ -36,7 +36,7 @@
  * @param <U> data type for {@code serializedSparse()} output
  */
 @Operator(group = "io")
-public final class SerializeSparse<U extends Tensor> extends RawOp implements Operand<U> {
+public final class SerializeSparse<U extends TType> extends RawOp implements Operand<U> {
   
   /**
    * Factory method to create a class wrapping a new SerializeSparse operation.
@@ -50,11 +50,11 @@ public final class SerializeSparse<U extends Tensor> extends RawOp implements Op
    * @return a new instance of SerializeSparse
    */
   @Endpoint(describeByClass = true)
-  public static <U extends Tensor, T extends Tensor> SerializeSparse<U> create(Scope scope, Operand<TInt64> sparseIndices, Operand<T> sparseValues, Operand<TInt64> sparseShape, DataType<U> outType) {
+  public static <U extends TType, T extends TType> SerializeSparse<U> create(Scope scope, Operand<TInt64> sparseIndices, Operand<T> sparseValues, Operand<TInt64> sparseShape, DataType<U> outType) {
     OperationBuilder opBuilder = scope.env().opBuilder("SerializeSparse", scope.makeOpName("SerializeSparse"));
-    opBuilder.addInput(sparseIndices.asOutput());
-    opBuilder.addInput(sparseValues.asOutput());
-    opBuilder.addInput(sparseShape.asOutput());
+    opBuilder.addInput(sparseIndices.asOutput(scope));
+    opBuilder.addInput(sparseValues.asOutput(scope));
+    opBuilder.addInput(sparseShape.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("out_type", outType);
     return new SerializeSparse<U>(opBuilder.build());
@@ -70,7 +70,7 @@ public static <U extends Tensor, T extends Tensor> SerializeSparse<U> create(Sco
    * @return a new instance of SerializeSparse
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> SerializeSparse<TString> create(Scope scope, Operand<TInt64> sparseIndices, Operand<T> sparseValues, Operand<TInt64> sparseShape) {
+  public static <T extends TType> SerializeSparse<TString> create(Scope scope, Operand<TInt64> sparseIndices, Operand<T> sparseValues, Operand<TInt64> sparseShape) {
     return create(scope, sparseIndices, sparseValues, sparseShape, TString.DTYPE);
   }
   
@@ -81,7 +81,7 @@ public Output<U> serializedSparse() {
   }
   
   @Override
-  public Output<U> asOutput() {
+  public Output<U> asOutput(Scope scope) {
     return serializedSparse;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/SerializeTensor.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/SerializeTensor.java
index cf5be111b9b..cc3b71be5d9 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/SerializeTensor.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/SerializeTensor.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TString;
+import org.tensorflow.types.family.TType;
 
 /**
  * Transforms a Tensor into a serialized TensorProto proto.
@@ -42,9 +42,9 @@ public final class SerializeTensor extends RawOp implements Operand<TString> {
    * @return a new instance of SerializeTensor
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> SerializeTensor create(Scope scope, Operand<T> tensor) {
+  public static <T extends TType> SerializeTensor create(Scope scope, Operand<T> tensor) {
     OperationBuilder opBuilder = scope.env().opBuilder("SerializeTensor", scope.makeOpName("SerializeTensor"));
-    opBuilder.addInput(tensor.asOutput());
+    opBuilder.addInput(tensor.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new SerializeTensor(opBuilder.build());
   }
@@ -57,7 +57,7 @@ public Output<TString> serialized() {
   }
   
   @Override
-  public Output<TString> asOutput() {
+  public Output<TString> asOutput(Scope scope) {
     return serialized;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ShardedFilename.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ShardedFilename.java
index 88b422a5f6d..4ddc9959485 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ShardedFilename.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ShardedFilename.java
@@ -48,9 +48,9 @@ public final class ShardedFilename extends RawOp implements Operand<TString> {
   @Endpoint(describeByClass = true)
   public static ShardedFilename create(Scope scope, Operand<TString> basename, Operand<TInt32> shard, Operand<TInt32> numShards) {
     OperationBuilder opBuilder = scope.env().opBuilder("ShardedFilename", scope.makeOpName("ShardedFilename"));
-    opBuilder.addInput(basename.asOutput());
-    opBuilder.addInput(shard.asOutput());
-    opBuilder.addInput(numShards.asOutput());
+    opBuilder.addInput(basename.asOutput(scope));
+    opBuilder.addInput(shard.asOutput(scope));
+    opBuilder.addInput(numShards.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new ShardedFilename(opBuilder.build());
   }
@@ -62,7 +62,7 @@ public Output<TString> filename() {
   }
   
   @Override
-  public Output<TString> asOutput() {
+  public Output<TString> asOutput(Scope scope) {
     return filename;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ShardedFilespec.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ShardedFilespec.java
index f04c0cf1a58..b48c4d37d88 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ShardedFilespec.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ShardedFilespec.java
@@ -45,8 +45,8 @@ public final class ShardedFilespec extends RawOp implements Operand<TString> {
   @Endpoint(describeByClass = true)
   public static ShardedFilespec create(Scope scope, Operand<TString> basename, Operand<TInt32> numShards) {
     OperationBuilder opBuilder = scope.env().opBuilder("ShardedFilespec", scope.makeOpName("ShardedFilespec"));
-    opBuilder.addInput(basename.asOutput());
-    opBuilder.addInput(numShards.asOutput());
+    opBuilder.addInput(basename.asOutput(scope));
+    opBuilder.addInput(numShards.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new ShardedFilespec(opBuilder.build());
   }
@@ -58,7 +58,7 @@ public Output<TString> filename() {
   }
   
   @Override
-  public Output<TString> asOutput() {
+  public Output<TString> asOutput(Scope scope) {
     return filename;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/TextLineReader.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/TextLineReader.java
index d52738f708a..5db34187e96 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/TextLineReader.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/TextLineReader.java
@@ -21,17 +21,17 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * A Reader that outputs the lines of a file delimited by '\n'.
  */
 @Operator(group = "io")
-public final class TextLineReader extends RawOp implements Operand<Tensor> {
+public final class TextLineReader extends RawOp implements Operand<TType> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.io.TextLineReader}
@@ -131,8 +131,8 @@ public Output<?> readerHandle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<Tensor> asOutput() {
-    return (Output<Tensor>) readerHandle;
+  public Output<TType> asOutput(Scope scope) {
+    return (Output<TType>) readerHandle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/TfRecordReader.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/TfRecordReader.java
index 0adaeafa879..502ff7fe719 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/TfRecordReader.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/TfRecordReader.java
@@ -21,17 +21,17 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * A Reader that outputs the records from a TensorFlow Records file.
  */
 @Operator(group = "io")
-public final class TfRecordReader extends RawOp implements Operand<Tensor> {
+public final class TfRecordReader extends RawOp implements Operand<TType> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.io.TfRecordReader}
@@ -131,8 +131,8 @@ public Output<?> readerHandle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<Tensor> asOutput() {
-    return (Output<Tensor>) readerHandle;
+  public Output<TType> asOutput(Scope scope) {
+    return (Output<TType>) readerHandle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/WholeFileReader.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/WholeFileReader.java
index 83480ec07b0..7be00710bdb 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/WholeFileReader.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/WholeFileReader.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * A Reader that outputs the entire contents of a file as a value.
@@ -34,7 +34,7 @@
  * be a filename (key) and the contents of that file (value).
  */
 @Operator(group = "io")
-public final class WholeFileReader extends RawOp implements Operand<Tensor> {
+public final class WholeFileReader extends RawOp implements Operand<TType> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.io.WholeFileReader}
@@ -115,8 +115,8 @@ public Output<?> readerHandle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<Tensor> asOutput() {
-    return (Output<Tensor>) readerHandle;
+  public Output<TType> asOutput(Scope scope) {
+    return (Output<TType>) readerHandle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/WriteFile.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/WriteFile.java
index d1c9dd9b9c8..e43565c431a 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/WriteFile.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/WriteFile.java
@@ -45,8 +45,8 @@ public final class WriteFile extends RawOp {
   @Endpoint(describeByClass = true)
   public static WriteFile create(Scope scope, Operand<TString> filename, Operand<TString> contents) {
     OperationBuilder opBuilder = scope.env().opBuilder("WriteFile", scope.makeOpName("WriteFile"));
-    opBuilder.addInput(filename.asOutput());
-    opBuilder.addInput(contents.asOutput());
+    opBuilder.addInput(filename.asOutput(scope));
+    opBuilder.addInput(contents.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new WriteFile(opBuilder.build());
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BandPart.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BandPart.java
index b6b5f90336a..9cebc6f40d1 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BandPart.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BandPart.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * Copy a tensor setting everything outside a central band in each innermost matrix to zero.
@@ -70,7 +70,7 @@
  * @param <T> data type for {@code band()} output
  */
 @Operator(group = "linalg")
-public final class BandPart<T extends Tensor> extends RawOp implements Operand<T> {
+public final class BandPart<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new BandPart operation.
@@ -84,11 +84,11 @@ public final class BandPart<T extends Tensor> extends RawOp implements Operand<T
    * @return a new instance of BandPart
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor, U extends Tensor & TNumber> BandPart<T> create(Scope scope, Operand<T> input, Operand<U> numLower, Operand<U> numUpper) {
+  public static <T extends TType, U extends TNumber> BandPart<T> create(Scope scope, Operand<T> input, Operand<U> numLower, Operand<U> numUpper) {
     OperationBuilder opBuilder = scope.env().opBuilder("MatrixBandPart", scope.makeOpName("BandPart"));
-    opBuilder.addInput(input.asOutput());
-    opBuilder.addInput(numLower.asOutput());
-    opBuilder.addInput(numUpper.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
+    opBuilder.addInput(numLower.asOutput(scope));
+    opBuilder.addInput(numUpper.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new BandPart<T>(opBuilder.build());
   }
@@ -101,7 +101,7 @@ public Output<T> band() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return band;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchCholesky.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchCholesky.java
index 08eeabdaa69..088f7e118b0 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchCholesky.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchCholesky.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -32,7 +31,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "linalg")
-public final class BatchCholesky<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class BatchCholesky<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new BatchCholesky operation.
@@ -42,9 +41,9 @@ public final class BatchCholesky<T extends Tensor & TNumber> extends RawOp imple
    * @return a new instance of BatchCholesky
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> BatchCholesky<T> create(Scope scope, Operand<T> input) {
+  public static <T extends TNumber> BatchCholesky<T> create(Scope scope, Operand<T> input) {
     OperationBuilder opBuilder = scope.env().opBuilder("BatchCholesky", scope.makeOpName("BatchCholesky"));
-    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new BatchCholesky<T>(opBuilder.build());
   }
@@ -56,7 +55,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchCholeskyGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchCholeskyGrad.java
index 47184b8de81..608b2e9d4a5 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchCholeskyGrad.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchCholeskyGrad.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -32,7 +31,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "linalg")
-public final class BatchCholeskyGrad<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class BatchCholeskyGrad<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new BatchCholeskyGrad operation.
@@ -43,10 +42,10 @@ public final class BatchCholeskyGrad<T extends Tensor & TNumber> extends RawOp i
    * @return a new instance of BatchCholeskyGrad
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> BatchCholeskyGrad<T> create(Scope scope, Operand<T> l, Operand<T> grad) {
+  public static <T extends TNumber> BatchCholeskyGrad<T> create(Scope scope, Operand<T> l, Operand<T> grad) {
     OperationBuilder opBuilder = scope.env().opBuilder("BatchCholeskyGrad", scope.makeOpName("BatchCholeskyGrad"));
-    opBuilder.addInput(l.asOutput());
-    opBuilder.addInput(grad.asOutput());
+    opBuilder.addInput(l.asOutput(scope));
+    opBuilder.addInput(grad.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new BatchCholeskyGrad<T>(opBuilder.build());
   }
@@ -58,7 +57,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixBandPart.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixBandPart.java
index f857dc7588d..4ba6f583f74 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixBandPart.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixBandPart.java
@@ -21,18 +21,18 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
 
 /**
  * @param <T> data type for {@code band()} output
  */
 @Operator(group = "linalg")
-public final class BatchMatrixBandPart<T extends Tensor> extends RawOp implements Operand<T> {
+public final class BatchMatrixBandPart<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new BatchMatrixBandPart operation.
@@ -44,11 +44,11 @@ public final class BatchMatrixBandPart<T extends Tensor> extends RawOp implement
    * @return a new instance of BatchMatrixBandPart
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> BatchMatrixBandPart<T> create(Scope scope, Operand<T> input, Operand<TInt64> numLower, Operand<TInt64> numUpper) {
+  public static <T extends TType> BatchMatrixBandPart<T> create(Scope scope, Operand<T> input, Operand<TInt64> numLower, Operand<TInt64> numUpper) {
     OperationBuilder opBuilder = scope.env().opBuilder("BatchMatrixBandPart", scope.makeOpName("BatchMatrixBandPart"));
-    opBuilder.addInput(input.asOutput());
-    opBuilder.addInput(numLower.asOutput());
-    opBuilder.addInput(numUpper.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
+    opBuilder.addInput(numLower.asOutput(scope));
+    opBuilder.addInput(numUpper.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new BatchMatrixBandPart<T>(opBuilder.build());
   }
@@ -60,7 +60,7 @@ public Output<T> band() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return band;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixDeterminant.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixDeterminant.java
index 7b28f87b1e0..a29c53c36c4 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixDeterminant.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixDeterminant.java
@@ -21,17 +21,17 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "linalg")
-public final class BatchMatrixDeterminant<T extends Tensor> extends RawOp implements Operand<T> {
+public final class BatchMatrixDeterminant<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new BatchMatrixDeterminant operation.
@@ -41,9 +41,9 @@ public final class BatchMatrixDeterminant<T extends Tensor> extends RawOp implem
    * @return a new instance of BatchMatrixDeterminant
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> BatchMatrixDeterminant<T> create(Scope scope, Operand<T> input) {
+  public static <T extends TType> BatchMatrixDeterminant<T> create(Scope scope, Operand<T> input) {
     OperationBuilder opBuilder = scope.env().opBuilder("BatchMatrixDeterminant", scope.makeOpName("BatchMatrixDeterminant"));
-    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new BatchMatrixDeterminant<T>(opBuilder.build());
   }
@@ -55,7 +55,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixDiag.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixDiag.java
index 5bd6f2770d7..78d4860ca43 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixDiag.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixDiag.java
@@ -21,17 +21,17 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "linalg")
-public final class BatchMatrixDiag<T extends Tensor> extends RawOp implements Operand<T> {
+public final class BatchMatrixDiag<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new BatchMatrixDiag operation.
@@ -41,9 +41,9 @@ public final class BatchMatrixDiag<T extends Tensor> extends RawOp implements Op
    * @return a new instance of BatchMatrixDiag
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> BatchMatrixDiag<T> create(Scope scope, Operand<T> diagonal) {
+  public static <T extends TType> BatchMatrixDiag<T> create(Scope scope, Operand<T> diagonal) {
     OperationBuilder opBuilder = scope.env().opBuilder("BatchMatrixDiag", scope.makeOpName("BatchMatrixDiag"));
-    opBuilder.addInput(diagonal.asOutput());
+    opBuilder.addInput(diagonal.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new BatchMatrixDiag<T>(opBuilder.build());
   }
@@ -55,7 +55,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixDiagPart.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixDiagPart.java
index c174055130c..9ba0bacf370 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixDiagPart.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixDiagPart.java
@@ -21,17 +21,17 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * @param <T> data type for {@code diagonal()} output
  */
 @Operator(group = "linalg")
-public final class BatchMatrixDiagPart<T extends Tensor> extends RawOp implements Operand<T> {
+public final class BatchMatrixDiagPart<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new BatchMatrixDiagPart operation.
@@ -41,9 +41,9 @@ public final class BatchMatrixDiagPart<T extends Tensor> extends RawOp implement
    * @return a new instance of BatchMatrixDiagPart
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> BatchMatrixDiagPart<T> create(Scope scope, Operand<T> input) {
+  public static <T extends TType> BatchMatrixDiagPart<T> create(Scope scope, Operand<T> input) {
     OperationBuilder opBuilder = scope.env().opBuilder("BatchMatrixDiagPart", scope.makeOpName("BatchMatrixDiagPart"));
-    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new BatchMatrixDiagPart<T>(opBuilder.build());
   }
@@ -55,7 +55,7 @@ public Output<T> diagonal() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return diagonal;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixInverse.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixInverse.java
index 8f596e62777..1590430cfe6 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixInverse.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixInverse.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -32,7 +31,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "linalg")
-public final class BatchMatrixInverse<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class BatchMatrixInverse<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.linalg.BatchMatrixInverse}
@@ -62,9 +61,9 @@ private Options() {
    * @return a new instance of BatchMatrixInverse
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> BatchMatrixInverse<T> create(Scope scope, Operand<T> input, Options... options) {
+  public static <T extends TNumber> BatchMatrixInverse<T> create(Scope scope, Operand<T> input, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("BatchMatrixInverse", scope.makeOpName("BatchMatrixInverse"));
-    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -90,7 +89,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixSetDiag.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixSetDiag.java
index c02c4eaf541..b3a4ad61eb5 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixSetDiag.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixSetDiag.java
@@ -21,17 +21,17 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "linalg")
-public final class BatchMatrixSetDiag<T extends Tensor> extends RawOp implements Operand<T> {
+public final class BatchMatrixSetDiag<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new BatchMatrixSetDiag operation.
@@ -42,10 +42,10 @@ public final class BatchMatrixSetDiag<T extends Tensor> extends RawOp implements
    * @return a new instance of BatchMatrixSetDiag
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> BatchMatrixSetDiag<T> create(Scope scope, Operand<T> input, Operand<T> diagonal) {
+  public static <T extends TType> BatchMatrixSetDiag<T> create(Scope scope, Operand<T> input, Operand<T> diagonal) {
     OperationBuilder opBuilder = scope.env().opBuilder("BatchMatrixSetDiag", scope.makeOpName("BatchMatrixSetDiag"));
-    opBuilder.addInput(input.asOutput());
-    opBuilder.addInput(diagonal.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
+    opBuilder.addInput(diagonal.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new BatchMatrixSetDiag<T>(opBuilder.build());
   }
@@ -57,7 +57,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixSolve.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixSolve.java
index 7def4a32599..6ebfa446613 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixSolve.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixSolve.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -32,7 +31,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "linalg")
-public final class BatchMatrixSolve<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class BatchMatrixSolve<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.linalg.BatchMatrixSolve}
@@ -63,10 +62,10 @@ private Options() {
    * @return a new instance of BatchMatrixSolve
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> BatchMatrixSolve<T> create(Scope scope, Operand<T> matrix, Operand<T> rhs, Options... options) {
+  public static <T extends TNumber> BatchMatrixSolve<T> create(Scope scope, Operand<T> matrix, Operand<T> rhs, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("BatchMatrixSolve", scope.makeOpName("BatchMatrixSolve"));
-    opBuilder.addInput(matrix.asOutput());
-    opBuilder.addInput(rhs.asOutput());
+    opBuilder.addInput(matrix.asOutput(scope));
+    opBuilder.addInput(rhs.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -92,7 +91,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixSolveLs.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixSolveLs.java
index a25215e20f4..722c3a5cf09 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixSolveLs.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixSolveLs.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -33,7 +32,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "linalg")
-public final class BatchMatrixSolveLs<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class BatchMatrixSolveLs<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.linalg.BatchMatrixSolveLs}
@@ -65,11 +64,11 @@ private Options() {
    * @return a new instance of BatchMatrixSolveLs
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> BatchMatrixSolveLs<T> create(Scope scope, Operand<T> matrix, Operand<T> rhs, Operand<TFloat64> l2Regularizer, Options... options) {
+  public static <T extends TNumber> BatchMatrixSolveLs<T> create(Scope scope, Operand<T> matrix, Operand<T> rhs, Operand<TFloat64> l2Regularizer, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("BatchMatrixSolveLs", scope.makeOpName("BatchMatrixSolveLs"));
-    opBuilder.addInput(matrix.asOutput());
-    opBuilder.addInput(rhs.asOutput());
-    opBuilder.addInput(l2Regularizer.asOutput());
+    opBuilder.addInput(matrix.asOutput(scope));
+    opBuilder.addInput(rhs.asOutput(scope));
+    opBuilder.addInput(l2Regularizer.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -95,7 +94,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixTriangularSolve.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixTriangularSolve.java
index 2c8b9e2ad8e..0062969a197 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixTriangularSolve.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixTriangularSolve.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -32,7 +31,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "linalg")
-public final class BatchMatrixTriangularSolve<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class BatchMatrixTriangularSolve<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.linalg.BatchMatrixTriangularSolve}
@@ -72,10 +71,10 @@ private Options() {
    * @return a new instance of BatchMatrixTriangularSolve
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> BatchMatrixTriangularSolve<T> create(Scope scope, Operand<T> matrix, Operand<T> rhs, Options... options) {
+  public static <T extends TNumber> BatchMatrixTriangularSolve<T> create(Scope scope, Operand<T> matrix, Operand<T> rhs, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("BatchMatrixTriangularSolve", scope.makeOpName("BatchMatrixTriangularSolve"));
-    opBuilder.addInput(matrix.asOutput());
-    opBuilder.addInput(rhs.asOutput());
+    opBuilder.addInput(matrix.asOutput(scope));
+    opBuilder.addInput(rhs.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -111,7 +110,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchSelfAdjointEig.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchSelfAdjointEig.java
index 2aa8d1f456b..b9e5dd053c5 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchSelfAdjointEig.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchSelfAdjointEig.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -32,7 +31,7 @@
  * @param <T> data type for {@code e()} output
  */
 @Operator(group = "linalg")
-public final class BatchSelfAdjointEig<T extends Tensor & TNumber> extends RawOp {
+public final class BatchSelfAdjointEig<T extends TNumber> extends RawOp {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.linalg.BatchSelfAdjointEig}
@@ -62,9 +61,9 @@ private Options() {
    * @return a new instance of BatchSelfAdjointEig
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> BatchSelfAdjointEig<T> create(Scope scope, Operand<T> input, Options... options) {
+  public static <T extends TNumber> BatchSelfAdjointEig<T> create(Scope scope, Operand<T> input, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("BatchSelfAdjointEigV2", scope.makeOpName("BatchSelfAdjointEig"));
-    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchSvd.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchSvd.java
index 0b9c7cc82fb..76ed6341fce 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchSvd.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchSvd.java
@@ -21,17 +21,17 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * @param <T> data type for {@code s()} output
  */
 @Operator(group = "linalg")
-public final class BatchSvd<T extends Tensor> extends RawOp {
+public final class BatchSvd<T extends TType> extends RawOp {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.linalg.BatchSvd}
@@ -70,9 +70,9 @@ private Options() {
    * @return a new instance of BatchSvd
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> BatchSvd<T> create(Scope scope, Operand<T> input, Options... options) {
+  public static <T extends TType> BatchSvd<T> create(Scope scope, Operand<T> input, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("BatchSvd", scope.makeOpName("BatchSvd"));
-    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Cholesky.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Cholesky.java
index fa9ee8d4ae5..c38d8b21256 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Cholesky.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Cholesky.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Computes the Cholesky decomposition of one or more square matrices.
@@ -47,7 +47,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "linalg")
-public final class Cholesky<T extends Tensor> extends RawOp implements Operand<T> {
+public final class Cholesky<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Cholesky operation.
@@ -57,9 +57,9 @@ public final class Cholesky<T extends Tensor> extends RawOp implements Operand<T
    * @return a new instance of Cholesky
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> Cholesky<T> create(Scope scope, Operand<T> input) {
+  public static <T extends TType> Cholesky<T> create(Scope scope, Operand<T> input) {
     OperationBuilder opBuilder = scope.env().opBuilder("Cholesky", scope.makeOpName("Cholesky"));
-    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new Cholesky<T>(opBuilder.build());
   }
@@ -72,7 +72,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/CholeskyGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/CholeskyGrad.java
index 89d2868825c..e73fa93b7b2 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/CholeskyGrad.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/CholeskyGrad.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -37,7 +36,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "linalg")
-public final class CholeskyGrad<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class CholeskyGrad<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new CholeskyGrad operation.
@@ -52,10 +51,10 @@ public final class CholeskyGrad<T extends Tensor & TNumber> extends RawOp implem
    * @return a new instance of CholeskyGrad
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> CholeskyGrad<T> create(Scope scope, Operand<T> l, Operand<T> grad) {
+  public static <T extends TNumber> CholeskyGrad<T> create(Scope scope, Operand<T> l, Operand<T> grad) {
     OperationBuilder opBuilder = scope.env().opBuilder("CholeskyGrad", scope.makeOpName("CholeskyGrad"));
-    opBuilder.addInput(l.asOutput());
-    opBuilder.addInput(grad.asOutput());
+    opBuilder.addInput(l.asOutput(scope));
+    opBuilder.addInput(grad.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new CholeskyGrad<T>(opBuilder.build());
   }
@@ -68,7 +67,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/ConjugateTranspose.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/ConjugateTranspose.java
index 702176bac05..b6219b24a6a 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/ConjugateTranspose.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/ConjugateTranspose.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * Shuffle dimensions of x according to a permutation and conjugate the result.
@@ -38,7 +38,7 @@
  * @param <T> data type for {@code y()} output
  */
 @Operator(group = "linalg")
-public final class ConjugateTranspose<T extends Tensor> extends RawOp implements Operand<T> {
+public final class ConjugateTranspose<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new ConjugateTranspose operation.
@@ -49,10 +49,10 @@ public final class ConjugateTranspose<T extends Tensor> extends RawOp implements
    * @return a new instance of ConjugateTranspose
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor, U extends Tensor & TNumber> ConjugateTranspose<T> create(Scope scope, Operand<T> x, Operand<U> perm) {
+  public static <T extends TType, U extends TNumber> ConjugateTranspose<T> create(Scope scope, Operand<T> x, Operand<U> perm) {
     OperationBuilder opBuilder = scope.env().opBuilder("ConjugateTranspose", scope.makeOpName("ConjugateTranspose"));
-    opBuilder.addInput(x.asOutput());
-    opBuilder.addInput(perm.asOutput());
+    opBuilder.addInput(x.asOutput(scope));
+    opBuilder.addInput(perm.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new ConjugateTranspose<T>(opBuilder.build());
   }
@@ -64,7 +64,7 @@ public Output<T> y() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return y;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Cross.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Cross.java
index 2b2c99d8d6b..1ab0e9b3c0f 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Cross.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Cross.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -38,7 +37,7 @@
  * @param <T> data type for {@code product()} output
  */
 @Operator(group = "linalg")
-public final class Cross<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class Cross<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Cross operation.
@@ -49,10 +48,10 @@ public final class Cross<T extends Tensor & TNumber> extends RawOp implements Op
    * @return a new instance of Cross
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> Cross<T> create(Scope scope, Operand<T> a, Operand<T> b) {
+  public static <T extends TNumber> Cross<T> create(Scope scope, Operand<T> a, Operand<T> b) {
     OperationBuilder opBuilder = scope.env().opBuilder("Cross", scope.makeOpName("Cross"));
-    opBuilder.addInput(a.asOutput());
-    opBuilder.addInput(b.asOutput());
+    opBuilder.addInput(a.asOutput(scope));
+    opBuilder.addInput(b.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new Cross<T>(opBuilder.build());
   }
@@ -65,7 +64,7 @@ public Output<T> product() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return product;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Det.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Det.java
index e81f186064c..d4f7dfe0eab 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Det.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Det.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Computes the determinant of one or more square matrices.
@@ -37,7 +37,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "linalg")
-public final class Det<T extends Tensor> extends RawOp implements Operand<T> {
+public final class Det<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Det operation.
@@ -47,9 +47,9 @@ public final class Det<T extends Tensor> extends RawOp implements Operand<T> {
    * @return a new instance of Det
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> Det<T> create(Scope scope, Operand<T> input) {
+  public static <T extends TType> Det<T> create(Scope scope, Operand<T> input) {
     OperationBuilder opBuilder = scope.env().opBuilder("MatrixDeterminant", scope.makeOpName("Det"));
-    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new Det<T>(opBuilder.build());
   }
@@ -62,7 +62,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Eig.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Eig.java
index ebce453d78d..327f9be5396 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Eig.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Eig.java
@@ -22,11 +22,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Computes the eigen decomposition of one or more square matrices.
@@ -46,7 +46,7 @@
  * @param <U> data type for {@code e()} output
  */
 @Operator(group = "linalg")
-public final class Eig<U extends Tensor> extends RawOp {
+public final class Eig<U extends TType> extends RawOp {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.linalg.Eig}
@@ -78,9 +78,9 @@ private Options() {
    * @return a new instance of Eig
    */
   @Endpoint(describeByClass = true)
-  public static <U extends Tensor, T extends Tensor> Eig<U> create(Scope scope, Operand<T> input, DataType<U> Tout, Options... options) {
+  public static <U extends TType, T extends TType> Eig<U> create(Scope scope, Operand<T> input, DataType<U> Tout, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("Eig", scope.makeOpName("Eig"));
-    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("Tout", Tout);
     if (options != null) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Einsum.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Einsum.java
index f1ad82b2bcc..0e3091d05f2 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Einsum.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Einsum.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.Operands;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Tensor contraction according to Einstein summation convention.
@@ -111,7 +111,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "linalg")
-public final class Einsum<T extends Tensor> extends RawOp implements Operand<T> {
+public final class Einsum<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Einsum operation.
@@ -122,9 +122,9 @@ public final class Einsum<T extends Tensor> extends RawOp implements Operand<T>
    * @return a new instance of Einsum
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> Einsum<T> create(Scope scope, Iterable<Operand<T>> inputs, String equation) {
+  public static <T extends TType> Einsum<T> create(Scope scope, Iterable<Operand<T>> inputs, String equation) {
     OperationBuilder opBuilder = scope.env().opBuilder("Einsum", scope.makeOpName("Einsum"));
-    opBuilder.addInputList(Operands.asOutputs(inputs));
+    opBuilder.addInputList(Operands.asOutputs(scope, inputs));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("equation", equation);
     return new Einsum<T>(opBuilder.build());
@@ -138,7 +138,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/EuclideanNorm.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/EuclideanNorm.java
index 8eb67637d16..62dae10d44b 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/EuclideanNorm.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/EuclideanNorm.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * Computes the euclidean norm of elements across dimensions of a tensor.
@@ -39,7 +39,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "linalg")
-public final class EuclideanNorm<T extends Tensor> extends RawOp implements Operand<T> {
+public final class EuclideanNorm<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.linalg.EuclideanNorm}
@@ -71,10 +71,10 @@ private Options() {
    * @return a new instance of EuclideanNorm
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor, U extends Tensor & TNumber> EuclideanNorm<T> create(Scope scope, Operand<T> input, Operand<U> axis, Options... options) {
+  public static <T extends TType, U extends TNumber> EuclideanNorm<T> create(Scope scope, Operand<T> input, Operand<U> axis, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("EuclideanNorm", scope.makeOpName("EuclideanNorm"));
-    opBuilder.addInput(input.asOutput());
-    opBuilder.addInput(axis.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
+    opBuilder.addInput(axis.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -101,7 +101,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Inv.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Inv.java
index d5cdb76b5e7..dca78a8b23c 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Inv.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Inv.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Computes the inverse of one or more square invertible matrices or their
@@ -45,7 +45,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "linalg")
-public final class Inv<T extends Tensor> extends RawOp implements Operand<T> {
+public final class Inv<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.linalg.Inv}
@@ -75,9 +75,9 @@ private Options() {
    * @return a new instance of Inv
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> Inv<T> create(Scope scope, Operand<T> input, Options... options) {
+  public static <T extends TType> Inv<T> create(Scope scope, Operand<T> input, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("MatrixInverse", scope.makeOpName("Inv"));
-    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -108,7 +108,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/LoadAndRemapMatrix.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/LoadAndRemapMatrix.java
index 1010c503eba..31a4d445dc8 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/LoadAndRemapMatrix.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/LoadAndRemapMatrix.java
@@ -122,11 +122,11 @@ private Options() {
   @Endpoint(describeByClass = true)
   public static LoadAndRemapMatrix create(Scope scope, Operand<TString> ckptPath, Operand<TString> oldTensorName, Operand<TInt64> rowRemapping, Operand<TInt64> colRemapping, Operand<TFloat32> initializingValues, Long numRows, Long numCols, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("LoadAndRemapMatrix", scope.makeOpName("LoadAndRemapMatrix"));
-    opBuilder.addInput(ckptPath.asOutput());
-    opBuilder.addInput(oldTensorName.asOutput());
-    opBuilder.addInput(rowRemapping.asOutput());
-    opBuilder.addInput(colRemapping.asOutput());
-    opBuilder.addInput(initializingValues.asOutput());
+    opBuilder.addInput(ckptPath.asOutput(scope));
+    opBuilder.addInput(oldTensorName.asOutput(scope));
+    opBuilder.addInput(rowRemapping.asOutput(scope));
+    opBuilder.addInput(colRemapping.asOutput(scope));
+    opBuilder.addInput(initializingValues.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("num_rows", numRows);
     opBuilder.setAttr("num_cols", numCols);
@@ -158,7 +158,7 @@ public Output<TFloat32> outputMatrix() {
   }
   
   @Override
-  public Output<TFloat32> asOutput() {
+  public Output<TFloat32> asOutput(Scope scope) {
     return outputMatrix;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/LogMatrixDeterminant.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/LogMatrixDeterminant.java
index ece21956d3d..de1e4a08bb1 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/LogMatrixDeterminant.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/LogMatrixDeterminant.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Computes the sign and the log of the absolute value of the determinant of
@@ -43,7 +43,7 @@
  * @param <T> data type for {@code sign()} output
  */
 @Operator(group = "linalg")
-public final class LogMatrixDeterminant<T extends Tensor> extends RawOp {
+public final class LogMatrixDeterminant<T extends TType> extends RawOp {
   
   /**
    * Factory method to create a class wrapping a new LogMatrixDeterminant operation.
@@ -53,9 +53,9 @@ public final class LogMatrixDeterminant<T extends Tensor> extends RawOp {
    * @return a new instance of LogMatrixDeterminant
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> LogMatrixDeterminant<T> create(Scope scope, Operand<T> input) {
+  public static <T extends TType> LogMatrixDeterminant<T> create(Scope scope, Operand<T> input) {
     OperationBuilder opBuilder = scope.env().opBuilder("LogMatrixDeterminant", scope.makeOpName("LogMatrixDeterminant"));
-    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new LogMatrixDeterminant<T>(opBuilder.build());
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Lu.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Lu.java
index fa937ab6c24..dad690bff5e 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Lu.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Lu.java
@@ -22,13 +22,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * Computes the LU decomposition of one or more square matrices.
@@ -55,7 +55,7 @@
  * @param <U> data type for {@code p()} output
  */
 @Operator(group = "linalg")
-public final class Lu<T extends Tensor, U extends Tensor & TNumber> extends RawOp {
+public final class Lu<T extends TType, U extends TNumber> extends RawOp {
   
   /**
    * Factory method to create a class wrapping a new Lu operation.
@@ -67,9 +67,9 @@ public final class Lu<T extends Tensor, U extends Tensor & TNumber> extends RawO
    * @return a new instance of Lu
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor, U extends Tensor & TNumber> Lu<T, U> create(Scope scope, Operand<T> input, DataType<U> outputIdxType) {
+  public static <T extends TType, U extends TNumber> Lu<T, U> create(Scope scope, Operand<T> input, DataType<U> outputIdxType) {
     OperationBuilder opBuilder = scope.env().opBuilder("Lu", scope.makeOpName("Lu"));
-    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("output_idx_type", outputIdxType);
     return new Lu<T, U>(opBuilder.build());
@@ -84,7 +84,7 @@ public static <T extends Tensor, U extends Tensor & TNumber> Lu<T, U> create(Sco
    * @return a new instance of Lu
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> Lu<T, TInt32> create(Scope scope, Operand<T> input) {
+  public static <T extends TType> Lu<T, TInt32> create(Scope scope, Operand<T> input) {
     return create(scope, input, TInt32.DTYPE);
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/MatMul.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/MatMul.java
index 54443149d3b..700f03d4d39 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/MatMul.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/MatMul.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Multiply the matrix "a" by the matrix "b".
@@ -41,7 +41,7 @@
  * @param <T> data type for {@code product()} output
  */
 @Operator(group = "linalg")
-public final class MatMul<T extends Tensor> extends RawOp implements Operand<T> {
+public final class MatMul<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.linalg.MatMul}
@@ -81,10 +81,10 @@ private Options() {
    * @return a new instance of MatMul
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> MatMul<T> create(Scope scope, Operand<T> a, Operand<T> b, Options... options) {
+  public static <T extends TType> MatMul<T> create(Scope scope, Operand<T> a, Operand<T> b, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("MatMul", scope.makeOpName("MatMul"));
-    opBuilder.addInput(a.asOutput());
-    opBuilder.addInput(b.asOutput());
+    opBuilder.addInput(a.asOutput(scope));
+    opBuilder.addInput(b.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -120,7 +120,7 @@ public Output<T> product() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return product;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/MatrixDiag.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/MatrixDiag.java
index 0baa4506cc6..621564ab1be 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/MatrixDiag.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/MatrixDiag.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TType;
 
 /**
  * Returns a batched diagonal tensor with given batched diagonal values.
@@ -119,7 +119,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "linalg")
-public final class MatrixDiag<T extends Tensor> extends RawOp implements Operand<T> {
+public final class MatrixDiag<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new MatrixDiag operation.
@@ -141,13 +141,13 @@ public final class MatrixDiag<T extends Tensor> extends RawOp implements Operand
    * @return a new instance of MatrixDiag
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> MatrixDiag<T> create(Scope scope, Operand<T> diagonal, Operand<TInt32> k, Operand<TInt32> numRows, Operand<TInt32> numCols, Operand<T> paddingValue) {
+  public static <T extends TType> MatrixDiag<T> create(Scope scope, Operand<T> diagonal, Operand<TInt32> k, Operand<TInt32> numRows, Operand<TInt32> numCols, Operand<T> paddingValue) {
     OperationBuilder opBuilder = scope.env().opBuilder("MatrixDiagV2", scope.makeOpName("MatrixDiag"));
-    opBuilder.addInput(diagonal.asOutput());
-    opBuilder.addInput(k.asOutput());
-    opBuilder.addInput(numRows.asOutput());
-    opBuilder.addInput(numCols.asOutput());
-    opBuilder.addInput(paddingValue.asOutput());
+    opBuilder.addInput(diagonal.asOutput(scope));
+    opBuilder.addInput(k.asOutput(scope));
+    opBuilder.addInput(numRows.asOutput(scope));
+    opBuilder.addInput(numCols.asOutput(scope));
+    opBuilder.addInput(paddingValue.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new MatrixDiag<T>(opBuilder.build());
   }
@@ -160,7 +160,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/MatrixDiagPart.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/MatrixDiagPart.java
index 73ade6ac51c..a9c11cf82dc 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/MatrixDiagPart.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/MatrixDiagPart.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TType;
 
 /**
  * Returns the batched diagonal part of a batched tensor.
@@ -101,7 +101,7 @@
  * @param <T> data type for {@code diagonal()} output
  */
 @Operator(group = "linalg")
-public final class MatrixDiagPart<T extends Tensor> extends RawOp implements Operand<T> {
+public final class MatrixDiagPart<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new MatrixDiagPart operation.
@@ -117,11 +117,11 @@ public final class MatrixDiagPart<T extends Tensor> extends RawOp implements Ope
    * @return a new instance of MatrixDiagPart
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> MatrixDiagPart<T> create(Scope scope, Operand<T> input, Operand<TInt32> k, Operand<T> paddingValue) {
+  public static <T extends TType> MatrixDiagPart<T> create(Scope scope, Operand<T> input, Operand<TInt32> k, Operand<T> paddingValue) {
     OperationBuilder opBuilder = scope.env().opBuilder("MatrixDiagPartV2", scope.makeOpName("MatrixDiagPart"));
-    opBuilder.addInput(input.asOutput());
-    opBuilder.addInput(k.asOutput());
-    opBuilder.addInput(paddingValue.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
+    opBuilder.addInput(k.asOutput(scope));
+    opBuilder.addInput(paddingValue.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new MatrixDiagPart<T>(opBuilder.build());
   }
@@ -134,7 +134,7 @@ public Output<T> diagonal() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return diagonal;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/MatrixDiagPartV3.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/MatrixDiagPartV3.java
index 5ca61bd9c17..1d671ac2c66 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/MatrixDiagPartV3.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/MatrixDiagPartV3.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TType;
 
 /**
  * Returns the batched diagonal part of a batched tensor.
@@ -132,7 +132,7 @@
  * @param <T> data type for {@code diagonal()} output
  */
 @Operator(group = "linalg")
-public final class MatrixDiagPartV3<T extends Tensor> extends RawOp implements Operand<T> {
+public final class MatrixDiagPartV3<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.linalg.MatrixDiagPartV3}
@@ -174,11 +174,11 @@ private Options() {
    * @return a new instance of MatrixDiagPartV3
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> MatrixDiagPartV3<T> create(Scope scope, Operand<T> input, Operand<TInt32> k, Operand<T> paddingValue, Options... options) {
+  public static <T extends TType> MatrixDiagPartV3<T> create(Scope scope, Operand<T> input, Operand<TInt32> k, Operand<T> paddingValue, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("MatrixDiagPartV3", scope.makeOpName("MatrixDiagPartV3"));
-    opBuilder.addInput(input.asOutput());
-    opBuilder.addInput(k.asOutput());
-    opBuilder.addInput(paddingValue.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
+    opBuilder.addInput(k.asOutput(scope));
+    opBuilder.addInput(paddingValue.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -211,7 +211,7 @@ public Output<T> diagonal() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return diagonal;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/MatrixDiagV3.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/MatrixDiagV3.java
index 5c8a7c72023..204954aabbb 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/MatrixDiagV3.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/MatrixDiagV3.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TType;
 
 /**
  * Returns a batched diagonal tensor with given batched diagonal values.
@@ -148,7 +148,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "linalg")
-public final class MatrixDiagV3<T extends Tensor> extends RawOp implements Operand<T> {
+public final class MatrixDiagV3<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.linalg.MatrixDiagV3}
@@ -196,13 +196,13 @@ private Options() {
    * @return a new instance of MatrixDiagV3
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> MatrixDiagV3<T> create(Scope scope, Operand<T> diagonal, Operand<TInt32> k, Operand<TInt32> numRows, Operand<TInt32> numCols, Operand<T> paddingValue, Options... options) {
+  public static <T extends TType> MatrixDiagV3<T> create(Scope scope, Operand<T> diagonal, Operand<TInt32> k, Operand<TInt32> numRows, Operand<TInt32> numCols, Operand<T> paddingValue, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("MatrixDiagV3", scope.makeOpName("MatrixDiagV3"));
-    opBuilder.addInput(diagonal.asOutput());
-    opBuilder.addInput(k.asOutput());
-    opBuilder.addInput(numRows.asOutput());
-    opBuilder.addInput(numCols.asOutput());
-    opBuilder.addInput(paddingValue.asOutput());
+    opBuilder.addInput(diagonal.asOutput(scope));
+    opBuilder.addInput(k.asOutput(scope));
+    opBuilder.addInput(numRows.asOutput(scope));
+    opBuilder.addInput(numCols.asOutput(scope));
+    opBuilder.addInput(paddingValue.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -235,7 +235,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/MatrixLogarithm.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/MatrixLogarithm.java
index d098fbe0d4d..a43eabb415b 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/MatrixLogarithm.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/MatrixLogarithm.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Computes the matrix logarithm of one or more square matrices:
@@ -48,7 +48,7 @@
  * 
  * @param <T> data type for {@code output()} output
  */
-public final class MatrixLogarithm<T extends Tensor> extends RawOp implements Operand<T> {
+public final class MatrixLogarithm<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new MatrixLogarithm operation.
@@ -58,9 +58,9 @@ public final class MatrixLogarithm<T extends Tensor> extends RawOp implements Op
    * @return a new instance of MatrixLogarithm
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> MatrixLogarithm<T> create(Scope scope, Operand<T> input) {
+  public static <T extends TType> MatrixLogarithm<T> create(Scope scope, Operand<T> input) {
     OperationBuilder opBuilder = scope.env().opBuilder("MatrixLogarithm", scope.makeOpName("MatrixLogarithm"));
-    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new MatrixLogarithm<T>(opBuilder.build());
   }
@@ -77,7 +77,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/MatrixSetDiag.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/MatrixSetDiag.java
index c7ae957b0a5..d57dbb92fc8 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/MatrixSetDiag.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/MatrixSetDiag.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TType;
 
 /**
  * Returns a batched matrix tensor with new batched diagonal values.
@@ -136,7 +136,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "linalg")
-public final class MatrixSetDiag<T extends Tensor> extends RawOp implements Operand<T> {
+public final class MatrixSetDiag<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.linalg.MatrixSetDiag}
@@ -178,11 +178,11 @@ private Options() {
    * @return a new instance of MatrixSetDiag
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> MatrixSetDiag<T> create(Scope scope, Operand<T> input, Operand<T> diagonal, Operand<TInt32> k, Options... options) {
+  public static <T extends TType> MatrixSetDiag<T> create(Scope scope, Operand<T> input, Operand<T> diagonal, Operand<TInt32> k, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("MatrixSetDiagV3", scope.makeOpName("MatrixSetDiag"));
-    opBuilder.addInput(input.asOutput());
-    opBuilder.addInput(diagonal.asOutput());
-    opBuilder.addInput(k.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
+    opBuilder.addInput(diagonal.asOutput(scope));
+    opBuilder.addInput(k.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -215,7 +215,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/MatrixSolveLs.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/MatrixSolveLs.java
index 0f8ce97fdec..cea92eb0f45 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/MatrixSolveLs.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/MatrixSolveLs.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TFloat64;
+import org.tensorflow.types.family.TType;
 
 /**
  * Solves one or more linear least-squares problems.
@@ -69,7 +69,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "linalg")
-public final class MatrixSolveLs<T extends Tensor> extends RawOp implements Operand<T> {
+public final class MatrixSolveLs<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.linalg.MatrixSolveLs}
@@ -105,11 +105,11 @@ private Options() {
    * @return a new instance of MatrixSolveLs
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> MatrixSolveLs<T> create(Scope scope, Operand<T> matrix, Operand<T> rhs, Operand<TFloat64> l2Regularizer, Options... options) {
+  public static <T extends TType> MatrixSolveLs<T> create(Scope scope, Operand<T> matrix, Operand<T> rhs, Operand<TFloat64> l2Regularizer, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("MatrixSolveLs", scope.makeOpName("MatrixSolveLs"));
-    opBuilder.addInput(matrix.asOutput());
-    opBuilder.addInput(rhs.asOutput());
-    opBuilder.addInput(l2Regularizer.asOutput());
+    opBuilder.addInput(matrix.asOutput(scope));
+    opBuilder.addInput(rhs.asOutput(scope));
+    opBuilder.addInput(l2Regularizer.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -136,7 +136,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Qr.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Qr.java
index c8fbd3511d4..483c709d75c 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Qr.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Qr.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Computes the QR decompositions of one or more matrices.
@@ -44,7 +44,7 @@
  * @param <T> data type for {@code q()} output
  */
 @Operator(group = "linalg")
-public final class Qr<T extends Tensor> extends RawOp {
+public final class Qr<T extends TType> extends RawOp {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.linalg.Qr}
@@ -76,9 +76,9 @@ private Options() {
    * @return a new instance of Qr
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> Qr<T> create(Scope scope, Operand<T> input, Options... options) {
+  public static <T extends TType> Qr<T> create(Scope scope, Operand<T> input, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("Qr", scope.makeOpName("Qr"));
-    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/QuantizedMatMul.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/QuantizedMatMul.java
index 38a0f336092..eaa128d3d35 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/QuantizedMatMul.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/QuantizedMatMul.java
@@ -22,12 +22,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.family.TType;
 
 /**
  * Perform a quantized matrix multiplication of  `a` by the matrix `b`.
@@ -40,7 +40,7 @@
  * @param <V> data type for {@code out()} output
  */
 @Operator(group = "linalg")
-public final class QuantizedMatMul<V extends Tensor> extends RawOp {
+public final class QuantizedMatMul<V extends TType> extends RawOp {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.linalg.QuantizedMatMul}
@@ -87,14 +87,14 @@ private Options() {
    * @return a new instance of QuantizedMatMul
    */
   @Endpoint(describeByClass = true)
-  public static <V extends Tensor, T extends Tensor, U extends Tensor, W extends Tensor> QuantizedMatMul<V> create(Scope scope, Operand<T> a, Operand<U> b, Operand<TFloat32> minA, Operand<TFloat32> maxA, Operand<TFloat32> minB, Operand<TFloat32> maxB, DataType<V> Toutput, DataType<W> Tactivation, Options... options) {
+  public static <V extends TType, T extends TType, U extends TType, W extends TType> QuantizedMatMul<V> create(Scope scope, Operand<T> a, Operand<U> b, Operand<TFloat32> minA, Operand<TFloat32> maxA, Operand<TFloat32> minB, Operand<TFloat32> maxB, DataType<V> Toutput, DataType<W> Tactivation, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("QuantizedMatMul", scope.makeOpName("QuantizedMatMul"));
-    opBuilder.addInput(a.asOutput());
-    opBuilder.addInput(b.asOutput());
-    opBuilder.addInput(minA.asOutput());
-    opBuilder.addInput(maxA.asOutput());
-    opBuilder.addInput(minB.asOutput());
-    opBuilder.addInput(maxB.asOutput());
+    opBuilder.addInput(a.asOutput(scope));
+    opBuilder.addInput(b.asOutput(scope));
+    opBuilder.addInput(minA.asOutput(scope));
+    opBuilder.addInput(maxA.asOutput(scope));
+    opBuilder.addInput(minB.asOutput(scope));
+    opBuilder.addInput(maxB.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("Toutput", Toutput);
     opBuilder.setAttr("Tactivation", Tactivation);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/QuantizedMatMulWithBias.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/QuantizedMatMulWithBias.java
index 9a333aa8cf2..775046e4353 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/QuantizedMatMulWithBias.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/QuantizedMatMulWithBias.java
@@ -22,12 +22,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.family.TType;
 
 /**
  * Performs a quantized matrix multiplication of `a` by the matrix `b` with bias
@@ -41,7 +41,7 @@
  * 
  * @param <W> data type for {@code out()} output
  */
-public final class QuantizedMatMulWithBias<W extends Tensor> extends RawOp {
+public final class QuantizedMatMulWithBias<W extends TType> extends RawOp {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.linalg.QuantizedMatMulWithBias}
@@ -97,15 +97,15 @@ private Options() {
    * @return a new instance of QuantizedMatMulWithBias
    */
   @Endpoint(describeByClass = true)
-  public static <W extends Tensor, T extends Tensor, U extends Tensor, V extends Tensor> QuantizedMatMulWithBias<W> create(Scope scope, Operand<T> a, Operand<U> b, Operand<V> bias, Operand<TFloat32> minA, Operand<TFloat32> maxA, Operand<TFloat32> minB, Operand<TFloat32> maxB, DataType<W> Toutput, Options... options) {
+  public static <W extends TType, T extends TType, U extends TType, V extends TType> QuantizedMatMulWithBias<W> create(Scope scope, Operand<T> a, Operand<U> b, Operand<V> bias, Operand<TFloat32> minA, Operand<TFloat32> maxA, Operand<TFloat32> minB, Operand<TFloat32> maxB, DataType<W> Toutput, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("QuantizedMatMulWithBias", scope.makeOpName("QuantizedMatMulWithBias"));
-    opBuilder.addInput(a.asOutput());
-    opBuilder.addInput(b.asOutput());
-    opBuilder.addInput(bias.asOutput());
-    opBuilder.addInput(minA.asOutput());
-    opBuilder.addInput(maxA.asOutput());
-    opBuilder.addInput(minB.asOutput());
-    opBuilder.addInput(maxB.asOutput());
+    opBuilder.addInput(a.asOutput(scope));
+    opBuilder.addInput(b.asOutput(scope));
+    opBuilder.addInput(bias.asOutput(scope));
+    opBuilder.addInput(minA.asOutput(scope));
+    opBuilder.addInput(maxA.asOutput(scope));
+    opBuilder.addInput(minB.asOutput(scope));
+    opBuilder.addInput(maxB.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("Toutput", Toutput);
     if (options != null) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/QuantizedMatMulWithBiasAndRelu.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/QuantizedMatMulWithBiasAndRelu.java
index 354fd26a18a..4612d13dbb7 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/QuantizedMatMulWithBiasAndRelu.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/QuantizedMatMulWithBiasAndRelu.java
@@ -22,12 +22,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.family.TType;
 
 /**
  * Perform a quantized matrix multiplication of  `a` by the matrix `b` with bias
@@ -42,7 +42,7 @@
  * 
  * @param <V> data type for {@code out()} output
  */
-public final class QuantizedMatMulWithBiasAndRelu<V extends Tensor> extends RawOp {
+public final class QuantizedMatMulWithBiasAndRelu<V extends TType> extends RawOp {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.linalg.QuantizedMatMulWithBiasAndRelu}
@@ -98,15 +98,15 @@ private Options() {
    * @return a new instance of QuantizedMatMulWithBiasAndRelu
    */
   @Endpoint(describeByClass = true)
-  public static <V extends Tensor, T extends Tensor, U extends Tensor> QuantizedMatMulWithBiasAndRelu<V> create(Scope scope, Operand<T> a, Operand<U> b, Operand<TFloat32> bias, Operand<TFloat32> minA, Operand<TFloat32> maxA, Operand<TFloat32> minB, Operand<TFloat32> maxB, DataType<V> Toutput, Options... options) {
+  public static <V extends TType, T extends TType, U extends TType> QuantizedMatMulWithBiasAndRelu<V> create(Scope scope, Operand<T> a, Operand<U> b, Operand<TFloat32> bias, Operand<TFloat32> minA, Operand<TFloat32> maxA, Operand<TFloat32> minB, Operand<TFloat32> maxB, DataType<V> Toutput, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("QuantizedMatMulWithBiasAndRelu", scope.makeOpName("QuantizedMatMulWithBiasAndRelu"));
-    opBuilder.addInput(a.asOutput());
-    opBuilder.addInput(b.asOutput());
-    opBuilder.addInput(bias.asOutput());
-    opBuilder.addInput(minA.asOutput());
-    opBuilder.addInput(maxA.asOutput());
-    opBuilder.addInput(minB.asOutput());
-    opBuilder.addInput(maxB.asOutput());
+    opBuilder.addInput(a.asOutput(scope));
+    opBuilder.addInput(b.asOutput(scope));
+    opBuilder.addInput(bias.asOutput(scope));
+    opBuilder.addInput(minA.asOutput(scope));
+    opBuilder.addInput(maxA.asOutput(scope));
+    opBuilder.addInput(minB.asOutput(scope));
+    opBuilder.addInput(maxB.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("Toutput", Toutput);
     if (options != null) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/QuantizedMatMulWithBiasAndReluAndRequantize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/QuantizedMatMulWithBiasAndReluAndRequantize.java
index a5548b7f368..48cbdd88f21 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/QuantizedMatMulWithBiasAndReluAndRequantize.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/QuantizedMatMulWithBiasAndReluAndRequantize.java
@@ -22,12 +22,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.family.TType;
 
 /**
  * Perform a quantized matrix multiplication of  `a` by the matrix `b` with bias
@@ -43,7 +43,7 @@
  * 
  * @param <W> data type for {@code out()} output
  */
-public final class QuantizedMatMulWithBiasAndReluAndRequantize<W extends Tensor> extends RawOp {
+public final class QuantizedMatMulWithBiasAndReluAndRequantize<W extends TType> extends RawOp {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.linalg.QuantizedMatMulWithBiasAndReluAndRequantize}
@@ -101,17 +101,17 @@ private Options() {
    * @return a new instance of QuantizedMatMulWithBiasAndReluAndRequantize
    */
   @Endpoint(describeByClass = true)
-  public static <W extends Tensor, T extends Tensor, U extends Tensor, V extends Tensor> QuantizedMatMulWithBiasAndReluAndRequantize<W> create(Scope scope, Operand<T> a, Operand<U> b, Operand<V> bias, Operand<TFloat32> minA, Operand<TFloat32> maxA, Operand<TFloat32> minB, Operand<TFloat32> maxB, Operand<TFloat32> minFreezedOutput, Operand<TFloat32> maxFreezedOutput, DataType<W> Toutput, Options... options) {
+  public static <W extends TType, T extends TType, U extends TType, V extends TType> QuantizedMatMulWithBiasAndReluAndRequantize<W> create(Scope scope, Operand<T> a, Operand<U> b, Operand<V> bias, Operand<TFloat32> minA, Operand<TFloat32> maxA, Operand<TFloat32> minB, Operand<TFloat32> maxB, Operand<TFloat32> minFreezedOutput, Operand<TFloat32> maxFreezedOutput, DataType<W> Toutput, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("QuantizedMatMulWithBiasAndReluAndRequantize", scope.makeOpName("QuantizedMatMulWithBiasAndReluAndRequantize"));
-    opBuilder.addInput(a.asOutput());
-    opBuilder.addInput(b.asOutput());
-    opBuilder.addInput(bias.asOutput());
-    opBuilder.addInput(minA.asOutput());
-    opBuilder.addInput(maxA.asOutput());
-    opBuilder.addInput(minB.asOutput());
-    opBuilder.addInput(maxB.asOutput());
-    opBuilder.addInput(minFreezedOutput.asOutput());
-    opBuilder.addInput(maxFreezedOutput.asOutput());
+    opBuilder.addInput(a.asOutput(scope));
+    opBuilder.addInput(b.asOutput(scope));
+    opBuilder.addInput(bias.asOutput(scope));
+    opBuilder.addInput(minA.asOutput(scope));
+    opBuilder.addInput(maxA.asOutput(scope));
+    opBuilder.addInput(minB.asOutput(scope));
+    opBuilder.addInput(maxB.asOutput(scope));
+    opBuilder.addInput(minFreezedOutput.asOutput(scope));
+    opBuilder.addInput(maxFreezedOutput.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("Toutput", Toutput);
     if (options != null) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/SelfAdjointEig.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/SelfAdjointEig.java
index af4898246ee..954e94c10be 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/SelfAdjointEig.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/SelfAdjointEig.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Computes the eigen decomposition of one or more square self-adjoint matrices.
@@ -45,7 +45,7 @@
  * @param <T> data type for {@code e()} output
  */
 @Operator(group = "linalg")
-public final class SelfAdjointEig<T extends Tensor> extends RawOp {
+public final class SelfAdjointEig<T extends TType> extends RawOp {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.linalg.SelfAdjointEig}
@@ -76,9 +76,9 @@ private Options() {
    * @return a new instance of SelfAdjointEig
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> SelfAdjointEig<T> create(Scope scope, Operand<T> input, Options... options) {
+  public static <T extends TType> SelfAdjointEig<T> create(Scope scope, Operand<T> input, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("SelfAdjointEigV2", scope.makeOpName("SelfAdjointEig"));
-    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Solve.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Solve.java
index 1966cfb65f1..ab85e800b71 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Solve.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Solve.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Solves systems of linear equations.
@@ -40,7 +40,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "linalg")
-public final class Solve<T extends Tensor> extends RawOp implements Operand<T> {
+public final class Solve<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.linalg.Solve}
@@ -72,10 +72,10 @@ private Options() {
    * @return a new instance of Solve
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> Solve<T> create(Scope scope, Operand<T> matrix, Operand<T> rhs, Options... options) {
+  public static <T extends TType> Solve<T> create(Scope scope, Operand<T> matrix, Operand<T> rhs, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("MatrixSolve", scope.makeOpName("Solve"));
-    opBuilder.addInput(matrix.asOutput());
-    opBuilder.addInput(rhs.asOutput());
+    opBuilder.addInput(matrix.asOutput(scope));
+    opBuilder.addInput(rhs.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -103,7 +103,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Sqrtm.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Sqrtm.java
index 36096b8c83c..4334b4a1534 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Sqrtm.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Sqrtm.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Computes the matrix square root of one or more square matrices:
@@ -49,7 +49,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "linalg")
-public final class Sqrtm<T extends Tensor> extends RawOp implements Operand<T> {
+public final class Sqrtm<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Sqrtm operation.
@@ -59,9 +59,9 @@ public final class Sqrtm<T extends Tensor> extends RawOp implements Operand<T> {
    * @return a new instance of Sqrtm
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> Sqrtm<T> create(Scope scope, Operand<T> input) {
+  public static <T extends TType> Sqrtm<T> create(Scope scope, Operand<T> input) {
     OperationBuilder opBuilder = scope.env().opBuilder("MatrixSquareRoot", scope.makeOpName("Sqrtm"));
-    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new Sqrtm<T>(opBuilder.build());
   }
@@ -78,7 +78,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Svd.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Svd.java
index 60d85b7d76e..d020ab02358 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Svd.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Svd.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Computes the singular value decompositions of one or more matrices.
@@ -45,7 +45,7 @@
  * @param <T> data type for {@code s()} output
  */
 @Operator(group = "linalg")
-public final class Svd<T extends Tensor> extends RawOp {
+public final class Svd<T extends TType> extends RawOp {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.linalg.Svd}
@@ -89,9 +89,9 @@ private Options() {
    * @return a new instance of Svd
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> Svd<T> create(Scope scope, Operand<T> input, Options... options) {
+  public static <T extends TType> Svd<T> create(Scope scope, Operand<T> input, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("Svd", scope.makeOpName("Svd"));
-    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/TensorDiag.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/TensorDiag.java
index 0bf02584809..bcaac11e213 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/TensorDiag.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/TensorDiag.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Returns a diagonal tensor with a given diagonal values.
@@ -51,7 +51,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "linalg")
-public final class TensorDiag<T extends Tensor> extends RawOp implements Operand<T> {
+public final class TensorDiag<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new TensorDiag operation.
@@ -61,9 +61,9 @@ public final class TensorDiag<T extends Tensor> extends RawOp implements Operand
    * @return a new instance of TensorDiag
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> TensorDiag<T> create(Scope scope, Operand<T> diagonal) {
+  public static <T extends TType> TensorDiag<T> create(Scope scope, Operand<T> diagonal) {
     OperationBuilder opBuilder = scope.env().opBuilder("Diag", scope.makeOpName("TensorDiag"));
-    opBuilder.addInput(diagonal.asOutput());
+    opBuilder.addInput(diagonal.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new TensorDiag<T>(opBuilder.build());
   }
@@ -75,7 +75,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/TensorDiagPart.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/TensorDiagPart.java
index e218c245953..ef2a3b258c5 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/TensorDiagPart.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/TensorDiagPart.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Returns the diagonal part of the tensor.
@@ -52,7 +52,7 @@
  * @param <T> data type for {@code diagonal()} output
  */
 @Operator(group = "linalg")
-public final class TensorDiagPart<T extends Tensor> extends RawOp implements Operand<T> {
+public final class TensorDiagPart<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new TensorDiagPart operation.
@@ -62,9 +62,9 @@ public final class TensorDiagPart<T extends Tensor> extends RawOp implements Ope
    * @return a new instance of TensorDiagPart
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> TensorDiagPart<T> create(Scope scope, Operand<T> input) {
+  public static <T extends TType> TensorDiagPart<T> create(Scope scope, Operand<T> input) {
     OperationBuilder opBuilder = scope.env().opBuilder("DiagPart", scope.makeOpName("TensorDiagPart"));
-    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new TensorDiagPart<T>(opBuilder.build());
   }
@@ -77,7 +77,7 @@ public Output<T> diagonal() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return diagonal;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Transpose.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Transpose.java
index a736582d262..7662ff4e3d6 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Transpose.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Transpose.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * Shuffle dimensions of x according to a permutation.
@@ -37,7 +37,7 @@
  * @param <T> data type for {@code y()} output
  */
 @Operator(group = "linalg")
-public final class Transpose<T extends Tensor> extends RawOp implements Operand<T> {
+public final class Transpose<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Transpose operation.
@@ -48,10 +48,10 @@ public final class Transpose<T extends Tensor> extends RawOp implements Operand<
    * @return a new instance of Transpose
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor, U extends Tensor & TNumber> Transpose<T> create(Scope scope, Operand<T> x, Operand<U> perm) {
+  public static <T extends TType, U extends TNumber> Transpose<T> create(Scope scope, Operand<T> x, Operand<U> perm) {
     OperationBuilder opBuilder = scope.env().opBuilder("Transpose", scope.makeOpName("Transpose"));
-    opBuilder.addInput(x.asOutput());
-    opBuilder.addInput(perm.asOutput());
+    opBuilder.addInput(x.asOutput(scope));
+    opBuilder.addInput(perm.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new Transpose<T>(opBuilder.build());
   }
@@ -63,7 +63,7 @@ public Output<T> y() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return y;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/TriangularSolve.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/TriangularSolve.java
index 209f3ef14be..9e61c01a83a 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/TriangularSolve.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/TriangularSolve.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Solves systems of linear equations with upper or lower triangular matrices by backsubstitution.
@@ -80,7 +80,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "linalg")
-public final class TriangularSolve<T extends Tensor> extends RawOp implements Operand<T> {
+public final class TriangularSolve<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.linalg.TriangularSolve}
@@ -126,10 +126,10 @@ private Options() {
    * @return a new instance of TriangularSolve
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> TriangularSolve<T> create(Scope scope, Operand<T> matrix, Operand<T> rhs, Options... options) {
+  public static <T extends TType> TriangularSolve<T> create(Scope scope, Operand<T> matrix, Operand<T> rhs, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("MatrixTriangularSolve", scope.makeOpName("TriangularSolve"));
-    opBuilder.addInput(matrix.asOutput());
-    opBuilder.addInput(rhs.asOutput());
+    opBuilder.addInput(matrix.asOutput(scope));
+    opBuilder.addInput(rhs.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -172,7 +172,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/TridiagonalMatMul.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/TridiagonalMatMul.java
index c2ea63fe6db..c700a24f346 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/TridiagonalMatMul.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/TridiagonalMatMul.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Calculate product with tridiagonal matrix.
@@ -34,7 +34,7 @@
  * 
  * @param <T> data type for {@code output()} output
  */
-public final class TridiagonalMatMul<T extends Tensor> extends RawOp implements Operand<T> {
+public final class TridiagonalMatMul<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new TridiagonalMatMul operation.
@@ -51,12 +51,12 @@ public final class TridiagonalMatMul<T extends Tensor> extends RawOp implements
    * @return a new instance of TridiagonalMatMul
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> TridiagonalMatMul<T> create(Scope scope, Operand<T> superdiag, Operand<T> maindiag, Operand<T> subdiag, Operand<T> rhs) {
+  public static <T extends TType> TridiagonalMatMul<T> create(Scope scope, Operand<T> superdiag, Operand<T> maindiag, Operand<T> subdiag, Operand<T> rhs) {
     OperationBuilder opBuilder = scope.env().opBuilder("TridiagonalMatMul", scope.makeOpName("TridiagonalMatMul"));
-    opBuilder.addInput(superdiag.asOutput());
-    opBuilder.addInput(maindiag.asOutput());
-    opBuilder.addInput(subdiag.asOutput());
-    opBuilder.addInput(rhs.asOutput());
+    opBuilder.addInput(superdiag.asOutput(scope));
+    opBuilder.addInput(maindiag.asOutput(scope));
+    opBuilder.addInput(subdiag.asOutput(scope));
+    opBuilder.addInput(rhs.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new TridiagonalMatMul<T>(opBuilder.build());
   }
@@ -69,7 +69,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/TridiagonalSolve.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/TridiagonalSolve.java
index 8ac8df38394..6865f480ef2 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/TridiagonalSolve.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/TridiagonalSolve.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Solves tridiagonal systems of equations.
@@ -40,7 +40,7 @@
  * 
  * @param <T> data type for {@code output()} output
  */
-public final class TridiagonalSolve<T extends Tensor> extends RawOp implements Operand<T> {
+public final class TridiagonalSolve<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.linalg.TridiagonalSolve}
@@ -76,10 +76,10 @@ private Options() {
    * @return a new instance of TridiagonalSolve
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> TridiagonalSolve<T> create(Scope scope, Operand<T> diagonals, Operand<T> rhs, Options... options) {
+  public static <T extends TType> TridiagonalSolve<T> create(Scope scope, Operand<T> diagonals, Operand<T> rhs, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("TridiagonalSolve", scope.makeOpName("TridiagonalSolve"));
-    opBuilder.addInput(diagonals.asOutput());
-    opBuilder.addInput(rhs.asOutput());
+    opBuilder.addInput(diagonals.asOutput(scope));
+    opBuilder.addInput(rhs.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -107,7 +107,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/CSRSparseMatrixComponents.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/CSRSparseMatrixComponents.java
index 987e05915bb..bb2dbb735ff 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/CSRSparseMatrixComponents.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/CSRSparseMatrixComponents.java
@@ -22,12 +22,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TType;
 
 /**
  * Reads out the CSR components at batch `index`.
@@ -37,7 +37,7 @@
  * 
  * @param <T> data type for {@code values()} output
  */
-public final class CSRSparseMatrixComponents<T extends Tensor> extends RawOp {
+public final class CSRSparseMatrixComponents<T extends TType> extends RawOp {
   
   /**
    * Factory method to create a class wrapping a new CSRSparseMatrixComponents operation.
@@ -49,10 +49,10 @@ public final class CSRSparseMatrixComponents<T extends Tensor> extends RawOp {
    * @return a new instance of CSRSparseMatrixComponents
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> CSRSparseMatrixComponents<T> create(Scope scope, Operand<?> csrSparseMatrix, Operand<TInt32> index, DataType<T> type) {
+  public static <T extends TType> CSRSparseMatrixComponents<T> create(Scope scope, Operand<?> csrSparseMatrix, Operand<TInt32> index, DataType<T> type) {
     OperationBuilder opBuilder = scope.env().opBuilder("CSRSparseMatrixComponents", scope.makeOpName("CSRSparseMatrixComponents"));
-    opBuilder.addInput(csrSparseMatrix.asOutput());
-    opBuilder.addInput(index.asOutput());
+    opBuilder.addInput(csrSparseMatrix.asOutput(scope));
+    opBuilder.addInput(index.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("type", type);
     return new CSRSparseMatrixComponents<T>(opBuilder.build());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/CSRSparseMatrixToDense.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/CSRSparseMatrixToDense.java
index c0242c74690..3c5caaeeb1b 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/CSRSparseMatrixToDense.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/CSRSparseMatrixToDense.java
@@ -22,18 +22,18 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Convert a (possibly batched) CSRSparseMatrix to dense.
  * 
  * @param <T> data type for {@code denseOutput()} output
  */
-public final class CSRSparseMatrixToDense<T extends Tensor> extends RawOp implements Operand<T> {
+public final class CSRSparseMatrixToDense<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new CSRSparseMatrixToDense operation.
@@ -44,9 +44,9 @@ public final class CSRSparseMatrixToDense<T extends Tensor> extends RawOp implem
    * @return a new instance of CSRSparseMatrixToDense
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> CSRSparseMatrixToDense<T> create(Scope scope, Operand<?> sparseInput, DataType<T> type) {
+  public static <T extends TType> CSRSparseMatrixToDense<T> create(Scope scope, Operand<?> sparseInput, DataType<T> type) {
     OperationBuilder opBuilder = scope.env().opBuilder("CSRSparseMatrixToDense", scope.makeOpName("CSRSparseMatrixToDense"));
-    opBuilder.addInput(sparseInput.asOutput());
+    opBuilder.addInput(sparseInput.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("type", type);
     return new CSRSparseMatrixToDense<T>(opBuilder.build());
@@ -60,7 +60,7 @@ public Output<T> denseOutput() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return denseOutput;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/CSRSparseMatrixToSparseTensor.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/CSRSparseMatrixToSparseTensor.java
index 7a5b1c46ba7..365ef176926 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/CSRSparseMatrixToSparseTensor.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/CSRSparseMatrixToSparseTensor.java
@@ -22,19 +22,19 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
 
 /**
  * Converts a (possibly batched) CSRSparesMatrix to a SparseTensor.
  * 
  * @param <T> data type for {@code values()} output
  */
-public final class CSRSparseMatrixToSparseTensor<T extends Tensor> extends RawOp {
+public final class CSRSparseMatrixToSparseTensor<T extends TType> extends RawOp {
   
   /**
    * Factory method to create a class wrapping a new CSRSparseMatrixToSparseTensor operation.
@@ -45,9 +45,9 @@ public final class CSRSparseMatrixToSparseTensor<T extends Tensor> extends RawOp
    * @return a new instance of CSRSparseMatrixToSparseTensor
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> CSRSparseMatrixToSparseTensor<T> create(Scope scope, Operand<?> sparseMatrix, DataType<T> type) {
+  public static <T extends TType> CSRSparseMatrixToSparseTensor<T> create(Scope scope, Operand<?> sparseMatrix, DataType<T> type) {
     OperationBuilder opBuilder = scope.env().opBuilder("CSRSparseMatrixToSparseTensor", scope.makeOpName("CSRSparseMatrixToSparseTensor"));
-    opBuilder.addInput(sparseMatrix.asOutput());
+    opBuilder.addInput(sparseMatrix.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("type", type);
     return new CSRSparseMatrixToSparseTensor<T>(opBuilder.build());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/DenseToCSRSparseMatrix.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/DenseToCSRSparseMatrix.java
index 97102309d63..cdf128a932c 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/DenseToCSRSparseMatrix.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/DenseToCSRSparseMatrix.java
@@ -21,17 +21,17 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
 
 /**
  * Converts a dense tensor to a (possibly batched) CSRSparseMatrix.
  */
-public final class DenseToCSRSparseMatrix extends RawOp implements Operand<Tensor> {
+public final class DenseToCSRSparseMatrix extends RawOp implements Operand<TType> {
   
   /**
    * Factory method to create a class wrapping a new DenseToCSRSparseMatrix operation.
@@ -42,10 +42,10 @@ public final class DenseToCSRSparseMatrix extends RawOp implements Operand<Tenso
    * @return a new instance of DenseToCSRSparseMatrix
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> DenseToCSRSparseMatrix create(Scope scope, Operand<T> denseInput, Operand<TInt64> indices) {
+  public static <T extends TType> DenseToCSRSparseMatrix create(Scope scope, Operand<T> denseInput, Operand<TInt64> indices) {
     OperationBuilder opBuilder = scope.env().opBuilder("DenseToCSRSparseMatrix", scope.makeOpName("DenseToCSRSparseMatrix"));
-    opBuilder.addInput(denseInput.asOutput());
-    opBuilder.addInput(indices.asOutput());
+    opBuilder.addInput(denseInput.asOutput(scope));
+    opBuilder.addInput(indices.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new DenseToCSRSparseMatrix(opBuilder.build());
   }
@@ -59,8 +59,8 @@ public Output<?> sparseOutput() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<Tensor> asOutput() {
-    return (Output<Tensor>) sparseOutput;
+  public Output<TType> asOutput(Scope scope) {
+    return (Output<TType>) sparseOutput;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixAdd.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixAdd.java
index 3e011c703ec..4b50062963d 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixAdd.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixAdd.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Sparse addition of two CSR matrices, C = alpha * A + beta * B.
@@ -33,7 +33,7 @@
  * The gradients of SparseMatrixAdd outputs with respect to alpha and beta are not
  * currently defined (TensorFlow will return zeros for these entries).
  */
-public final class SparseMatrixAdd extends RawOp implements Operand<Tensor> {
+public final class SparseMatrixAdd extends RawOp implements Operand<TType> {
   
   /**
    * Factory method to create a class wrapping a new SparseMatrixAdd operation.
@@ -46,12 +46,12 @@ public final class SparseMatrixAdd extends RawOp implements Operand<Tensor> {
    * @return a new instance of SparseMatrixAdd
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> SparseMatrixAdd create(Scope scope, Operand<?> a, Operand<?> b, Operand<T> alpha, Operand<T> beta) {
+  public static <T extends TType> SparseMatrixAdd create(Scope scope, Operand<?> a, Operand<?> b, Operand<T> alpha, Operand<T> beta) {
     OperationBuilder opBuilder = scope.env().opBuilder("SparseMatrixAdd", scope.makeOpName("SparseMatrixAdd"));
-    opBuilder.addInput(a.asOutput());
-    opBuilder.addInput(b.asOutput());
-    opBuilder.addInput(alpha.asOutput());
-    opBuilder.addInput(beta.asOutput());
+    opBuilder.addInput(a.asOutput(scope));
+    opBuilder.addInput(b.asOutput(scope));
+    opBuilder.addInput(alpha.asOutput(scope));
+    opBuilder.addInput(beta.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new SparseMatrixAdd(opBuilder.build());
   }
@@ -65,8 +65,8 @@ public Output<?> c() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<Tensor> asOutput() {
-    return (Output<Tensor>) c;
+  public Output<TType> asOutput(Scope scope) {
+    return (Output<TType>) c;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixMatMul.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixMatMul.java
index ebe3a8bc5c1..1668438105b 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixMatMul.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixMatMul.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Matrix-multiplies a sparse matrix with a dense matrix.
@@ -57,7 +57,7 @@
  * 
  * @param <T> data type for {@code output()} output
  */
-public final class SparseMatrixMatMul<T extends Tensor> extends RawOp implements Operand<T> {
+public final class SparseMatrixMatMul<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.linalg.sparse.SparseMatrixMatMul}
@@ -133,10 +133,10 @@ private Options() {
    * @return a new instance of SparseMatrixMatMul
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> SparseMatrixMatMul<T> create(Scope scope, Operand<?> a, Operand<T> b, Options... options) {
+  public static <T extends TType> SparseMatrixMatMul<T> create(Scope scope, Operand<?> a, Operand<T> b, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("SparseMatrixMatMul", scope.makeOpName("SparseMatrixMatMul"));
-    opBuilder.addInput(a.asOutput());
-    opBuilder.addInput(b.asOutput());
+    opBuilder.addInput(a.asOutput(scope));
+    opBuilder.addInput(b.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -213,7 +213,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixMul.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixMul.java
index 35589e32682..d378dac7e35 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixMul.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixMul.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Element-wise multiplication of a sparse matrix with a dense tensor.
@@ -39,7 +39,7 @@
  * <b>NOTE</b> even if `b` is zero, the sparsity structure of the output does not
  * change.
  */
-public final class SparseMatrixMul extends RawOp implements Operand<Tensor> {
+public final class SparseMatrixMul extends RawOp implements Operand<TType> {
   
   /**
    * Factory method to create a class wrapping a new SparseMatrixMul operation.
@@ -50,10 +50,10 @@ public final class SparseMatrixMul extends RawOp implements Operand<Tensor> {
    * @return a new instance of SparseMatrixMul
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> SparseMatrixMul create(Scope scope, Operand<?> a, Operand<T> b) {
+  public static <T extends TType> SparseMatrixMul create(Scope scope, Operand<?> a, Operand<T> b) {
     OperationBuilder opBuilder = scope.env().opBuilder("SparseMatrixMul", scope.makeOpName("SparseMatrixMul"));
-    opBuilder.addInput(a.asOutput());
-    opBuilder.addInput(b.asOutput());
+    opBuilder.addInput(a.asOutput(scope));
+    opBuilder.addInput(b.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new SparseMatrixMul(opBuilder.build());
   }
@@ -67,8 +67,8 @@ public Output<?> output() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<Tensor> asOutput() {
-    return (Output<Tensor>) output;
+  public Output<TType> asOutput(Scope scope) {
+    return (Output<TType>) output;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixNNZ.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixNNZ.java
index e730d181699..b17c3549720 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixNNZ.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixNNZ.java
@@ -42,7 +42,7 @@ public final class SparseMatrixNNZ extends RawOp implements Operand<TInt32> {
   @Endpoint(describeByClass = true)
   public static SparseMatrixNNZ create(Scope scope, Operand<?> sparseMatrix) {
     OperationBuilder opBuilder = scope.env().opBuilder("SparseMatrixNNZ", scope.makeOpName("SparseMatrixNNZ"));
-    opBuilder.addInput(sparseMatrix.asOutput());
+    opBuilder.addInput(sparseMatrix.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new SparseMatrixNNZ(opBuilder.build());
   }
@@ -55,7 +55,7 @@ public Output<TInt32> nnz() {
   }
   
   @Override
-  public Output<TInt32> asOutput() {
+  public Output<TInt32> asOutput(Scope scope) {
     return nnz;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixOrderingAMD.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixOrderingAMD.java
index b833601b72e..ad8264db7e3 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixOrderingAMD.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixOrderingAMD.java
@@ -90,7 +90,7 @@ public final class SparseMatrixOrderingAMD extends RawOp implements Operand<TInt
   @Endpoint(describeByClass = true)
   public static SparseMatrixOrderingAMD create(Scope scope, Operand<?> input) {
     OperationBuilder opBuilder = scope.env().opBuilder("SparseMatrixOrderingAMD", scope.makeOpName("SparseMatrixOrderingAMD"));
-    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new SparseMatrixOrderingAMD(opBuilder.build());
   }
@@ -103,7 +103,7 @@ public Output<TInt32> output() {
   }
   
   @Override
-  public Output<TInt32> asOutput() {
+  public Output<TInt32> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixSoftmax.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixSoftmax.java
index 9feb5bad1c9..bf04b321e7c 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixSoftmax.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixSoftmax.java
@@ -22,12 +22,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * Calculates the softmax of a CSRSparseMatrix.
@@ -38,7 +38,7 @@
  * the output has the same sparsity structure as the input (though missing values
  * in the output may now be treated as having probability zero).
  */
-public final class SparseMatrixSoftmax extends RawOp implements Operand<Tensor> {
+public final class SparseMatrixSoftmax extends RawOp implements Operand<TType> {
   
   /**
    * Factory method to create a class wrapping a new SparseMatrixSoftmax operation.
@@ -49,9 +49,9 @@ public final class SparseMatrixSoftmax extends RawOp implements Operand<Tensor>
    * @return a new instance of SparseMatrixSoftmax
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> SparseMatrixSoftmax create(Scope scope, Operand<?> logits, DataType<T> type) {
+  public static <T extends TNumber> SparseMatrixSoftmax create(Scope scope, Operand<?> logits, DataType<T> type) {
     OperationBuilder opBuilder = scope.env().opBuilder("SparseMatrixSoftmax", scope.makeOpName("SparseMatrixSoftmax"));
-    opBuilder.addInput(logits.asOutput());
+    opBuilder.addInput(logits.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("type", type);
     return new SparseMatrixSoftmax(opBuilder.build());
@@ -66,8 +66,8 @@ public Output<?> softmax() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<Tensor> asOutput() {
-    return (Output<Tensor>) softmax;
+  public Output<TType> asOutput(Scope scope) {
+    return (Output<TType>) softmax;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixSoftmaxGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixSoftmaxGrad.java
index 508e4be550e..bdc65e0f1e7 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixSoftmaxGrad.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixSoftmaxGrad.java
@@ -22,17 +22,17 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * Calculates the gradient of the SparseMatrixSoftmax op.
  */
-public final class SparseMatrixSoftmaxGrad extends RawOp implements Operand<Tensor> {
+public final class SparseMatrixSoftmaxGrad extends RawOp implements Operand<TType> {
   
   /**
    * Factory method to create a class wrapping a new SparseMatrixSoftmaxGrad operation.
@@ -44,10 +44,10 @@ public final class SparseMatrixSoftmaxGrad extends RawOp implements Operand<Tens
    * @return a new instance of SparseMatrixSoftmaxGrad
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> SparseMatrixSoftmaxGrad create(Scope scope, Operand<?> softmax, Operand<?> gradSoftmax, DataType<T> type) {
+  public static <T extends TNumber> SparseMatrixSoftmaxGrad create(Scope scope, Operand<?> softmax, Operand<?> gradSoftmax, DataType<T> type) {
     OperationBuilder opBuilder = scope.env().opBuilder("SparseMatrixSoftmaxGrad", scope.makeOpName("SparseMatrixSoftmaxGrad"));
-    opBuilder.addInput(softmax.asOutput());
-    opBuilder.addInput(gradSoftmax.asOutput());
+    opBuilder.addInput(softmax.asOutput(scope));
+    opBuilder.addInput(gradSoftmax.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("type", type);
     return new SparseMatrixSoftmaxGrad(opBuilder.build());
@@ -62,8 +62,8 @@ public Output<?> gradient() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<Tensor> asOutput() {
-    return (Output<Tensor>) gradient;
+  public Output<TType> asOutput(Scope scope) {
+    return (Output<TType>) gradient;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixSparseCholesky.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixSparseCholesky.java
index 79f31716eb3..c92f753bdad 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixSparseCholesky.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixSparseCholesky.java
@@ -22,12 +22,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TType;
 
 /**
  * Computes the sparse Cholesky decomposition of `input`.
@@ -103,7 +103,7 @@
  * permutation: A `Tensor`.
  * type: The type of `input`.
  */
-public final class SparseMatrixSparseCholesky extends RawOp implements Operand<Tensor> {
+public final class SparseMatrixSparseCholesky extends RawOp implements Operand<TType> {
   
   /**
    * Factory method to create a class wrapping a new SparseMatrixSparseCholesky operation.
@@ -115,10 +115,10 @@ public final class SparseMatrixSparseCholesky extends RawOp implements Operand<T
    * @return a new instance of SparseMatrixSparseCholesky
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> SparseMatrixSparseCholesky create(Scope scope, Operand<?> input, Operand<TInt32> permutation, DataType<T> type) {
+  public static <T extends TType> SparseMatrixSparseCholesky create(Scope scope, Operand<?> input, Operand<TInt32> permutation, DataType<T> type) {
     OperationBuilder opBuilder = scope.env().opBuilder("SparseMatrixSparseCholesky", scope.makeOpName("SparseMatrixSparseCholesky"));
-    opBuilder.addInput(input.asOutput());
-    opBuilder.addInput(permutation.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
+    opBuilder.addInput(permutation.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("type", type);
     return new SparseMatrixSparseCholesky(opBuilder.build());
@@ -133,8 +133,8 @@ public Output<?> output() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<Tensor> asOutput() {
-    return (Output<Tensor>) output;
+  public Output<TType> asOutput(Scope scope) {
+    return (Output<TType>) output;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixSparseMatMul.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixSparseMatMul.java
index a46d9a5a3e4..dbd4c82fb98 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixSparseMatMul.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixSparseMatMul.java
@@ -22,11 +22,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Sparse-matrix-multiplies two CSR matrices `a` and `b`.
@@ -104,7 +104,7 @@
  * adjoint_a: If True, `a` adjointed before multiplication.
  * adjoint_b: If True, `b` adjointed before multiplication.
  */
-public final class SparseMatrixSparseMatMul extends RawOp implements Operand<Tensor> {
+public final class SparseMatrixSparseMatMul extends RawOp implements Operand<TType> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.linalg.sparse.SparseMatrixSparseMatMul}
@@ -163,10 +163,10 @@ private Options() {
    * @return a new instance of SparseMatrixSparseMatMul
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> SparseMatrixSparseMatMul create(Scope scope, Operand<?> a, Operand<?> b, DataType<T> type, Options... options) {
+  public static <T extends TType> SparseMatrixSparseMatMul create(Scope scope, Operand<?> a, Operand<?> b, DataType<T> type, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("SparseMatrixSparseMatMul", scope.makeOpName("SparseMatrixSparseMatMul"));
-    opBuilder.addInput(a.asOutput());
-    opBuilder.addInput(b.asOutput());
+    opBuilder.addInput(a.asOutput(scope));
+    opBuilder.addInput(b.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("type", type);
     if (options != null) {
@@ -225,8 +225,8 @@ public Output<?> c() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<Tensor> asOutput() {
-    return (Output<Tensor>) c;
+  public Output<TType> asOutput(Scope scope) {
+    return (Output<TType>) c;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixTranspose.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixTranspose.java
index 6a2260099c8..c858a37383a 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixTranspose.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixTranspose.java
@@ -22,11 +22,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Transposes the inner (matrix) dimensions of a CSRSparseMatrix.
@@ -34,7 +34,7 @@
  * Transposes the inner (matrix) dimensions of a SparseMatrix and optionally
  * conjugates its values.
  */
-public final class SparseMatrixTranspose extends RawOp implements Operand<Tensor> {
+public final class SparseMatrixTranspose extends RawOp implements Operand<TType> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.linalg.sparse.SparseMatrixTranspose}
@@ -65,9 +65,9 @@ private Options() {
    * @return a new instance of SparseMatrixTranspose
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> SparseMatrixTranspose create(Scope scope, Operand<?> input, DataType<T> type, Options... options) {
+  public static <T extends TType> SparseMatrixTranspose create(Scope scope, Operand<?> input, DataType<T> type, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("SparseMatrixTranspose", scope.makeOpName("SparseMatrixTranspose"));
-    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("type", type);
     if (options != null) {
@@ -96,8 +96,8 @@ public Output<?> output() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<Tensor> asOutput() {
-    return (Output<Tensor>) output;
+  public Output<TType> asOutput(Scope scope) {
+    return (Output<TType>) output;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixZeros.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixZeros.java
index d1c70f3529c..bdb0ac426d7 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixZeros.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixZeros.java
@@ -22,17 +22,17 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
 
 /**
  * Creates an all-zeros CSRSparseMatrix with shape `dense_shape`.
  */
-public final class SparseMatrixZeros extends RawOp implements Operand<Tensor> {
+public final class SparseMatrixZeros extends RawOp implements Operand<TType> {
   
   /**
    * Factory method to create a class wrapping a new SparseMatrixZeros operation.
@@ -43,9 +43,9 @@ public final class SparseMatrixZeros extends RawOp implements Operand<Tensor> {
    * @return a new instance of SparseMatrixZeros
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> SparseMatrixZeros create(Scope scope, Operand<TInt64> denseShape, DataType<T> type) {
+  public static <T extends TType> SparseMatrixZeros create(Scope scope, Operand<TInt64> denseShape, DataType<T> type) {
     OperationBuilder opBuilder = scope.env().opBuilder("SparseMatrixZeros", scope.makeOpName("SparseMatrixZeros"));
-    opBuilder.addInput(denseShape.asOutput());
+    opBuilder.addInput(denseShape.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("type", type);
     return new SparseMatrixZeros(opBuilder.build());
@@ -60,8 +60,8 @@ public Output<?> sparseMatrix() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<Tensor> asOutput() {
-    return (Output<Tensor>) sparseMatrix;
+  public Output<TType> asOutput(Scope scope) {
+    return (Output<TType>) sparseMatrix;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseTensorToCSRSparseMatrix.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseTensorToCSRSparseMatrix.java
index 7f1f8397f07..f3cf07dbb45 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseTensorToCSRSparseMatrix.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseTensorToCSRSparseMatrix.java
@@ -21,17 +21,17 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
 
 /**
  * Converts a SparseTensor to a (possibly batched) CSRSparseMatrix.
  */
-public final class SparseTensorToCSRSparseMatrix extends RawOp implements Operand<Tensor> {
+public final class SparseTensorToCSRSparseMatrix extends RawOp implements Operand<TType> {
   
   /**
    * Factory method to create a class wrapping a new SparseTensorToCSRSparseMatrix operation.
@@ -43,11 +43,11 @@ public final class SparseTensorToCSRSparseMatrix extends RawOp implements Operan
    * @return a new instance of SparseTensorToCSRSparseMatrix
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> SparseTensorToCSRSparseMatrix create(Scope scope, Operand<TInt64> indices, Operand<T> values, Operand<TInt64> denseShape) {
+  public static <T extends TType> SparseTensorToCSRSparseMatrix create(Scope scope, Operand<TInt64> indices, Operand<T> values, Operand<TInt64> denseShape) {
     OperationBuilder opBuilder = scope.env().opBuilder("SparseTensorToCSRSparseMatrix", scope.makeOpName("SparseTensorToCSRSparseMatrix"));
-    opBuilder.addInput(indices.asOutput());
-    opBuilder.addInput(values.asOutput());
-    opBuilder.addInput(denseShape.asOutput());
+    opBuilder.addInput(indices.asOutput(scope));
+    opBuilder.addInput(values.asOutput(scope));
+    opBuilder.addInput(denseShape.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new SparseTensorToCSRSparseMatrix(opBuilder.build());
   }
@@ -61,8 +61,8 @@ public Output<?> sparseMatrix() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<Tensor> asOutput() {
-    return (Output<Tensor>) sparseMatrix;
+  public Output<TType> asOutput(Scope scope) {
+    return (Output<TType>) sparseMatrix;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Abs.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Abs.java
index 061abeeb85e..3f52e09a1c5 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Abs.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Abs.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -38,7 +37,7 @@
  * @param <T> data type for {@code y()} output
  */
 @Operator(group = "math")
-public final class Abs<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class Abs<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Abs operation.
@@ -48,9 +47,9 @@ public final class Abs<T extends Tensor & TNumber> extends RawOp implements Oper
    * @return a new instance of Abs
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> Abs<T> create(Scope scope, Operand<T> x) {
+  public static <T extends TNumber> Abs<T> create(Scope scope, Operand<T> x) {
     OperationBuilder opBuilder = scope.env().opBuilder("Abs", scope.makeOpName("Abs"));
-    opBuilder.addInput(x.asOutput());
+    opBuilder.addInput(x.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new Abs<T>(opBuilder.build());
   }
@@ -62,7 +61,7 @@ public Output<T> y() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return y;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/AccumulateN.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/AccumulateN.java
index 41ef83e8a42..cdac6836f17 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/AccumulateN.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/AccumulateN.java
@@ -21,13 +21,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.Operands;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Returns the element-wise sum of a list of tensors.
@@ -44,7 +44,7 @@
  * @param <T> data type for {@code sum()} output
  */
 @Operator(group = "math")
-public final class AccumulateN<T extends Tensor> extends RawOp implements Operand<T> {
+public final class AccumulateN<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new AccumulateN operation.
@@ -55,9 +55,9 @@ public final class AccumulateN<T extends Tensor> extends RawOp implements Operan
    * @return a new instance of AccumulateN
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> AccumulateN<T> create(Scope scope, Iterable<Operand<T>> inputs, Shape shape) {
+  public static <T extends TType> AccumulateN<T> create(Scope scope, Iterable<Operand<T>> inputs, Shape shape) {
     OperationBuilder opBuilder = scope.env().opBuilder("AccumulateNV2", scope.makeOpName("AccumulateN"));
-    opBuilder.addInputList(Operands.asOutputs(inputs));
+    opBuilder.addInputList(Operands.asOutputs(scope, inputs));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("shape", shape);
     return new AccumulateN<T>(opBuilder.build());
@@ -70,7 +70,7 @@ public Output<T> sum() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return sum;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Acos.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Acos.java
index ab784548c80..e4a8200e5ea 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Acos.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Acos.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Computes acos of x element-wise.
@@ -33,7 +33,7 @@
  * @param <T> data type for {@code y()} output
  */
 @Operator(group = "math")
-public final class Acos<T extends Tensor> extends RawOp implements Operand<T> {
+public final class Acos<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Acos operation.
@@ -43,9 +43,9 @@ public final class Acos<T extends Tensor> extends RawOp implements Operand<T> {
    * @return a new instance of Acos
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> Acos<T> create(Scope scope, Operand<T> x) {
+  public static <T extends TType> Acos<T> create(Scope scope, Operand<T> x) {
     OperationBuilder opBuilder = scope.env().opBuilder("Acos", scope.makeOpName("Acos"));
-    opBuilder.addInput(x.asOutput());
+    opBuilder.addInput(x.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new Acos<T>(opBuilder.build());
   }
@@ -57,7 +57,7 @@ public Output<T> y() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return y;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Acosh.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Acosh.java
index 9b432abae7f..7f2bda57117 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Acosh.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Acosh.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Computes inverse hyperbolic cosine of x element-wise.
@@ -41,7 +41,7 @@
  * @param <T> data type for {@code y()} output
  */
 @Operator(group = "math")
-public final class Acosh<T extends Tensor> extends RawOp implements Operand<T> {
+public final class Acosh<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Acosh operation.
@@ -51,9 +51,9 @@ public final class Acosh<T extends Tensor> extends RawOp implements Operand<T> {
    * @return a new instance of Acosh
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> Acosh<T> create(Scope scope, Operand<T> x) {
+  public static <T extends TType> Acosh<T> create(Scope scope, Operand<T> x) {
     OperationBuilder opBuilder = scope.env().opBuilder("Acosh", scope.makeOpName("Acosh"));
-    opBuilder.addInput(x.asOutput());
+    opBuilder.addInput(x.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new Acosh<T>(opBuilder.build());
   }
@@ -65,7 +65,7 @@ public Output<T> y() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return y;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Add.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Add.java
index 33501435d2e..f2a374fb9e3 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Add.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Add.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Returns x + y element-wise.
@@ -36,7 +36,7 @@
  * @param <T> data type for {@code z()} output
  */
 @Operator(group = "math")
-public final class Add<T extends Tensor> extends RawOp implements Operand<T> {
+public final class Add<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Add operation.
@@ -47,10 +47,10 @@ public final class Add<T extends Tensor> extends RawOp implements Operand<T> {
    * @return a new instance of Add
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> Add<T> create(Scope scope, Operand<T> x, Operand<T> y) {
+  public static <T extends TType> Add<T> create(Scope scope, Operand<T> x, Operand<T> y) {
     OperationBuilder opBuilder = scope.env().opBuilder("Add", scope.makeOpName("Add"));
-    opBuilder.addInput(x.asOutput());
-    opBuilder.addInput(y.asOutput());
+    opBuilder.addInput(x.asOutput(scope));
+    opBuilder.addInput(y.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new Add<T>(opBuilder.build());
   }
@@ -62,7 +62,7 @@ public Output<T> z() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return z;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/AddN.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/AddN.java
index 6004bcc15cf..99815542073 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/AddN.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/AddN.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.Operands;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Add all input tensors element wise.
@@ -42,7 +42,7 @@
  * @param <T> data type for {@code sum()} output
  */
 @Operator(group = "math")
-public final class AddN<T extends Tensor> extends RawOp implements Operand<T> {
+public final class AddN<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new AddN operation.
@@ -52,9 +52,9 @@ public final class AddN<T extends Tensor> extends RawOp implements Operand<T> {
    * @return a new instance of AddN
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> AddN<T> create(Scope scope, Iterable<Operand<T>> inputs) {
+  public static <T extends TType> AddN<T> create(Scope scope, Iterable<Operand<T>> inputs) {
     OperationBuilder opBuilder = scope.env().opBuilder("AddN", scope.makeOpName("AddN"));
-    opBuilder.addInputList(Operands.asOutputs(inputs));
+    opBuilder.addInputList(Operands.asOutputs(scope, inputs));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new AddN<T>(opBuilder.build());
   }
@@ -66,7 +66,7 @@ public Output<T> sum() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return sum;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Angle.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Angle.java
index dfc0413f4d4..5c1d9b6919b 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Angle.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Angle.java
@@ -22,13 +22,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TFloat32;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * Returns the argument of a complex number.
@@ -52,7 +52,7 @@
  * @param <U> data type for {@code output()} output
  */
 @Operator(group = "math")
-public final class Angle<U extends Tensor & TNumber> extends RawOp implements Operand<U> {
+public final class Angle<U extends TNumber> extends RawOp implements Operand<U> {
   
   /**
    * Factory method to create a class wrapping a new Angle operation.
@@ -63,9 +63,9 @@ public final class Angle<U extends Tensor & TNumber> extends RawOp implements Op
    * @return a new instance of Angle
    */
   @Endpoint(describeByClass = true)
-  public static <U extends Tensor & TNumber, T extends Tensor> Angle<U> create(Scope scope, Operand<T> input, DataType<U> Tout) {
+  public static <U extends TNumber, T extends TType> Angle<U> create(Scope scope, Operand<T> input, DataType<U> Tout) {
     OperationBuilder opBuilder = scope.env().opBuilder("Angle", scope.makeOpName("Angle"));
-    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("Tout", Tout);
     return new Angle<U>(opBuilder.build());
@@ -79,7 +79,7 @@ public static <U extends Tensor & TNumber, T extends Tensor> Angle<U> create(Sco
    * @return a new instance of Angle
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> Angle<TFloat32> create(Scope scope, Operand<T> input) {
+  public static <T extends TType> Angle<TFloat32> create(Scope scope, Operand<T> input) {
     return create(scope, input, TFloat32.DTYPE);
   }
   
@@ -90,7 +90,7 @@ public Output<U> output() {
   }
   
   @Override
-  public Output<U> asOutput() {
+  public Output<U> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/ApproximateEqual.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/ApproximateEqual.java
index 49c3062640b..753f503f64b 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/ApproximateEqual.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/ApproximateEqual.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TBool;
+import org.tensorflow.types.family.TType;
 
 /**
  * Returns the truth value of abs(x-y) < tolerance element-wise.
@@ -63,10 +63,10 @@ private Options() {
    * @return a new instance of ApproximateEqual
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> ApproximateEqual create(Scope scope, Operand<T> x, Operand<T> y, Options... options) {
+  public static <T extends TType> ApproximateEqual create(Scope scope, Operand<T> x, Operand<T> y, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ApproximateEqual", scope.makeOpName("ApproximateEqual"));
-    opBuilder.addInput(x.asOutput());
-    opBuilder.addInput(y.asOutput());
+    opBuilder.addInput(x.asOutput(scope));
+    opBuilder.addInput(y.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -92,7 +92,7 @@ public Output<TBool> z() {
   }
   
   @Override
-  public Output<TBool> asOutput() {
+  public Output<TBool> asOutput(Scope scope) {
     return z;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/ArgMax.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/ArgMax.java
index 145bad3d4b7..955d47f8ba3 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/ArgMax.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/ArgMax.java
@@ -22,13 +22,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * Returns the index with the largest value across dimensions of a tensor.
@@ -49,7 +49,7 @@
  * @param <V> data type for {@code output()} output
  */
 @Operator(group = "math")
-public final class ArgMax<V extends Tensor & TNumber> extends RawOp implements Operand<V> {
+public final class ArgMax<V extends TNumber> extends RawOp implements Operand<V> {
   
   /**
    * Factory method to create a class wrapping a new ArgMax operation.
@@ -63,10 +63,10 @@ public final class ArgMax<V extends Tensor & TNumber> extends RawOp implements O
    * @return a new instance of ArgMax
    */
   @Endpoint(describeByClass = true)
-  public static <V extends Tensor & TNumber, T extends Tensor, U extends Tensor & TNumber> ArgMax<V> create(Scope scope, Operand<T> input, Operand<U> dimension, DataType<V> outputType) {
+  public static <V extends TNumber, T extends TType, U extends TNumber> ArgMax<V> create(Scope scope, Operand<T> input, Operand<U> dimension, DataType<V> outputType) {
     OperationBuilder opBuilder = scope.env().opBuilder("ArgMax", scope.makeOpName("ArgMax"));
-    opBuilder.addInput(input.asOutput());
-    opBuilder.addInput(dimension.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
+    opBuilder.addInput(dimension.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("output_type", outputType);
     return new ArgMax<V>(opBuilder.build());
@@ -83,7 +83,7 @@ public static <V extends Tensor & TNumber, T extends Tensor, U extends Tensor &
    * @return a new instance of ArgMax
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor, U extends Tensor & TNumber> ArgMax<TInt64> create(Scope scope, Operand<T> input, Operand<U> dimension) {
+  public static <T extends TType, U extends TNumber> ArgMax<TInt64> create(Scope scope, Operand<T> input, Operand<U> dimension) {
     return create(scope, input, dimension, TInt64.DTYPE);
   }
   
@@ -94,7 +94,7 @@ public Output<V> output() {
   }
   
   @Override
-  public Output<V> asOutput() {
+  public Output<V> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/ArgMin.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/ArgMin.java
index 1392faa0ff3..73281d654b1 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/ArgMin.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/ArgMin.java
@@ -22,13 +22,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * Returns the index with the smallest value across dimensions of a tensor.
@@ -49,7 +49,7 @@
  * @param <V> data type for {@code output()} output
  */
 @Operator(group = "math")
-public final class ArgMin<V extends Tensor & TNumber> extends RawOp implements Operand<V> {
+public final class ArgMin<V extends TNumber> extends RawOp implements Operand<V> {
   
   /**
    * Factory method to create a class wrapping a new ArgMin operation.
@@ -63,10 +63,10 @@ public final class ArgMin<V extends Tensor & TNumber> extends RawOp implements O
    * @return a new instance of ArgMin
    */
   @Endpoint(describeByClass = true)
-  public static <V extends Tensor & TNumber, T extends Tensor, U extends Tensor & TNumber> ArgMin<V> create(Scope scope, Operand<T> input, Operand<U> dimension, DataType<V> outputType) {
+  public static <V extends TNumber, T extends TType, U extends TNumber> ArgMin<V> create(Scope scope, Operand<T> input, Operand<U> dimension, DataType<V> outputType) {
     OperationBuilder opBuilder = scope.env().opBuilder("ArgMin", scope.makeOpName("ArgMin"));
-    opBuilder.addInput(input.asOutput());
-    opBuilder.addInput(dimension.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
+    opBuilder.addInput(dimension.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("output_type", outputType);
     return new ArgMin<V>(opBuilder.build());
@@ -83,7 +83,7 @@ public static <V extends Tensor & TNumber, T extends Tensor, U extends Tensor &
    * @return a new instance of ArgMin
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor, U extends Tensor & TNumber> ArgMin<TInt64> create(Scope scope, Operand<T> input, Operand<U> dimension) {
+  public static <T extends TType, U extends TNumber> ArgMin<TInt64> create(Scope scope, Operand<T> input, Operand<U> dimension) {
     return create(scope, input, dimension, TInt64.DTYPE);
   }
   
@@ -94,7 +94,7 @@ public Output<V> output() {
   }
   
   @Override
-  public Output<V> asOutput() {
+  public Output<V> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Asin.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Asin.java
index 43f0fe31313..48750003eb1 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Asin.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Asin.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Computes the trignometric inverse sine of x element-wise.
@@ -49,7 +49,7 @@
  * @param <T> data type for {@code y()} output
  */
 @Operator(group = "math")
-public final class Asin<T extends Tensor> extends RawOp implements Operand<T> {
+public final class Asin<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Asin operation.
@@ -59,9 +59,9 @@ public final class Asin<T extends Tensor> extends RawOp implements Operand<T> {
    * @return a new instance of Asin
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> Asin<T> create(Scope scope, Operand<T> x) {
+  public static <T extends TType> Asin<T> create(Scope scope, Operand<T> x) {
     OperationBuilder opBuilder = scope.env().opBuilder("Asin", scope.makeOpName("Asin"));
-    opBuilder.addInput(x.asOutput());
+    opBuilder.addInput(x.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new Asin<T>(opBuilder.build());
   }
@@ -73,7 +73,7 @@ public Output<T> y() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return y;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Asinh.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Asinh.java
index a306e04d9b2..ae067dffb7c 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Asinh.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Asinh.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Computes inverse hyperbolic sine of x element-wise.
@@ -43,7 +43,7 @@
  * @param <T> data type for {@code y()} output
  */
 @Operator(group = "math")
-public final class Asinh<T extends Tensor> extends RawOp implements Operand<T> {
+public final class Asinh<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Asinh operation.
@@ -53,9 +53,9 @@ public final class Asinh<T extends Tensor> extends RawOp implements Operand<T> {
    * @return a new instance of Asinh
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> Asinh<T> create(Scope scope, Operand<T> x) {
+  public static <T extends TType> Asinh<T> create(Scope scope, Operand<T> x) {
     OperationBuilder opBuilder = scope.env().opBuilder("Asinh", scope.makeOpName("Asinh"));
-    opBuilder.addInput(x.asOutput());
+    opBuilder.addInput(x.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new Asinh<T>(opBuilder.build());
   }
@@ -67,7 +67,7 @@ public Output<T> y() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return y;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Atan.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Atan.java
index 52c91362608..f78ec14e057 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Atan.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Atan.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Computes the trignometric inverse tangent of x element-wise.
@@ -49,7 +49,7 @@
  * @param <T> data type for {@code y()} output
  */
 @Operator(group = "math")
-public final class Atan<T extends Tensor> extends RawOp implements Operand<T> {
+public final class Atan<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Atan operation.
@@ -59,9 +59,9 @@ public final class Atan<T extends Tensor> extends RawOp implements Operand<T> {
    * @return a new instance of Atan
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> Atan<T> create(Scope scope, Operand<T> x) {
+  public static <T extends TType> Atan<T> create(Scope scope, Operand<T> x) {
     OperationBuilder opBuilder = scope.env().opBuilder("Atan", scope.makeOpName("Atan"));
-    opBuilder.addInput(x.asOutput());
+    opBuilder.addInput(x.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new Atan<T>(opBuilder.build());
   }
@@ -73,7 +73,7 @@ public Output<T> y() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return y;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Atan2.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Atan2.java
index 6b96355ca7c..afb5011d4ce 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Atan2.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Atan2.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -40,7 +39,7 @@
  * @param <T> data type for {@code z()} output
  */
 @Operator(group = "math")
-public final class Atan2<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class Atan2<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Atan2 operation.
@@ -51,10 +50,10 @@ public final class Atan2<T extends Tensor & TNumber> extends RawOp implements Op
    * @return a new instance of Atan2
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> Atan2<T> create(Scope scope, Operand<T> y, Operand<T> x) {
+  public static <T extends TNumber> Atan2<T> create(Scope scope, Operand<T> y, Operand<T> x) {
     OperationBuilder opBuilder = scope.env().opBuilder("Atan2", scope.makeOpName("Atan2"));
-    opBuilder.addInput(y.asOutput());
-    opBuilder.addInput(x.asOutput());
+    opBuilder.addInput(y.asOutput(scope));
+    opBuilder.addInput(x.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new Atan2<T>(opBuilder.build());
   }
@@ -66,7 +65,7 @@ public Output<T> z() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return z;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Atanh.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Atanh.java
index 222afb87f9f..c8f3fbd8878 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Atanh.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Atanh.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Computes inverse hyperbolic tangent of x element-wise.
@@ -45,7 +45,7 @@
  * @param <T> data type for {@code y()} output
  */
 @Operator(group = "math")
-public final class Atanh<T extends Tensor> extends RawOp implements Operand<T> {
+public final class Atanh<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Atanh operation.
@@ -55,9 +55,9 @@ public final class Atanh<T extends Tensor> extends RawOp implements Operand<T> {
    * @return a new instance of Atanh
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> Atanh<T> create(Scope scope, Operand<T> x) {
+  public static <T extends TType> Atanh<T> create(Scope scope, Operand<T> x) {
     OperationBuilder opBuilder = scope.env().opBuilder("Atanh", scope.makeOpName("Atanh"));
-    opBuilder.addInput(x.asOutput());
+    opBuilder.addInput(x.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new Atanh<T>(opBuilder.build());
   }
@@ -69,7 +69,7 @@ public Output<T> y() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return y;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/BesselI0e.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/BesselI0e.java
index 267c801fe6b..fec91e77944 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/BesselI0e.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/BesselI0e.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -39,7 +38,7 @@
  * @param <T> data type for {@code y()} output
  */
 @Operator(group = "math")
-public final class BesselI0e<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class BesselI0e<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new BesselI0e operation.
@@ -49,9 +48,9 @@ public final class BesselI0e<T extends Tensor & TNumber> extends RawOp implement
    * @return a new instance of BesselI0e
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> BesselI0e<T> create(Scope scope, Operand<T> x) {
+  public static <T extends TNumber> BesselI0e<T> create(Scope scope, Operand<T> x) {
     OperationBuilder opBuilder = scope.env().opBuilder("BesselI0e", scope.makeOpName("BesselI0e"));
-    opBuilder.addInput(x.asOutput());
+    opBuilder.addInput(x.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new BesselI0e<T>(opBuilder.build());
   }
@@ -63,7 +62,7 @@ public Output<T> y() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return y;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/BesselI1e.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/BesselI1e.java
index ca7571067b0..8d27f3b7bd1 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/BesselI1e.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/BesselI1e.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -39,7 +38,7 @@
  * @param <T> data type for {@code y()} output
  */
 @Operator(group = "math")
-public final class BesselI1e<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class BesselI1e<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new BesselI1e operation.
@@ -49,9 +48,9 @@ public final class BesselI1e<T extends Tensor & TNumber> extends RawOp implement
    * @return a new instance of BesselI1e
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> BesselI1e<T> create(Scope scope, Operand<T> x) {
+  public static <T extends TNumber> BesselI1e<T> create(Scope scope, Operand<T> x) {
     OperationBuilder opBuilder = scope.env().opBuilder("BesselI1e", scope.makeOpName("BesselI1e"));
-    opBuilder.addInput(x.asOutput());
+    opBuilder.addInput(x.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new BesselI1e<T>(opBuilder.build());
   }
@@ -63,7 +62,7 @@ public Output<T> y() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return y;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Betainc.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Betainc.java
index b252886b1c4..066d1504311 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Betainc.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Betainc.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -45,7 +44,7 @@
  * @param <T> data type for {@code z()} output
  */
 @Operator(group = "math")
-public final class Betainc<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class Betainc<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Betainc operation.
@@ -57,11 +56,11 @@ public final class Betainc<T extends Tensor & TNumber> extends RawOp implements
    * @return a new instance of Betainc
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> Betainc<T> create(Scope scope, Operand<T> a, Operand<T> b, Operand<T> x) {
+  public static <T extends TNumber> Betainc<T> create(Scope scope, Operand<T> a, Operand<T> b, Operand<T> x) {
     OperationBuilder opBuilder = scope.env().opBuilder("Betainc", scope.makeOpName("Betainc"));
-    opBuilder.addInput(a.asOutput());
-    opBuilder.addInput(b.asOutput());
-    opBuilder.addInput(x.asOutput());
+    opBuilder.addInput(a.asOutput(scope));
+    opBuilder.addInput(b.asOutput(scope));
+    opBuilder.addInput(x.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new Betainc<T>(opBuilder.build());
   }
@@ -73,7 +72,7 @@ public Output<T> z() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return z;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Bincount.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Bincount.java
index dde12f54d47..0e15d315b76 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Bincount.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Bincount.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -43,7 +42,7 @@
  * @param <T> data type for {@code bins()} output
  */
 @Operator(group = "math")
-public final class Bincount<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class Bincount<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Bincount operation.
@@ -57,11 +56,11 @@ public final class Bincount<T extends Tensor & TNumber> extends RawOp implements
    * @return a new instance of Bincount
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> Bincount<T> create(Scope scope, Operand<TInt32> arr, Operand<TInt32> size, Operand<T> weights) {
+  public static <T extends TNumber> Bincount<T> create(Scope scope, Operand<TInt32> arr, Operand<TInt32> size, Operand<T> weights) {
     OperationBuilder opBuilder = scope.env().opBuilder("Bincount", scope.makeOpName("Bincount"));
-    opBuilder.addInput(arr.asOutput());
-    opBuilder.addInput(size.asOutput());
-    opBuilder.addInput(weights.asOutput());
+    opBuilder.addInput(arr.asOutput(scope));
+    opBuilder.addInput(size.asOutput(scope));
+    opBuilder.addInput(weights.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new Bincount<T>(opBuilder.build());
   }
@@ -75,7 +74,7 @@ public Output<T> bins() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return bins;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Ceil.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Ceil.java
index 2d2d097c036..9f1081f9998 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Ceil.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Ceil.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -34,7 +33,7 @@
  * @param <T> data type for {@code y()} output
  */
 @Operator(group = "math")
-public final class Ceil<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class Ceil<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Ceil operation.
@@ -44,9 +43,9 @@ public final class Ceil<T extends Tensor & TNumber> extends RawOp implements Ope
    * @return a new instance of Ceil
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> Ceil<T> create(Scope scope, Operand<T> x) {
+  public static <T extends TNumber> Ceil<T> create(Scope scope, Operand<T> x) {
     OperationBuilder opBuilder = scope.env().opBuilder("Ceil", scope.makeOpName("Ceil"));
-    opBuilder.addInput(x.asOutput());
+    opBuilder.addInput(x.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new Ceil<T>(opBuilder.build());
   }
@@ -58,7 +57,7 @@ public Output<T> y() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return y;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/CompareAndBitpack.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/CompareAndBitpack.java
index 84cf87274a2..6ca6f9324aa 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/CompareAndBitpack.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/CompareAndBitpack.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TUint8;
+import org.tensorflow.types.family.TType;
 
 /**
  * Compare values of `input` to `threshold` and pack resulting bits into a `uint8`.
@@ -65,10 +65,10 @@ public final class CompareAndBitpack extends RawOp implements Operand<TUint8> {
    * @return a new instance of CompareAndBitpack
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> CompareAndBitpack create(Scope scope, Operand<T> input, Operand<T> threshold) {
+  public static <T extends TType> CompareAndBitpack create(Scope scope, Operand<T> input, Operand<T> threshold) {
     OperationBuilder opBuilder = scope.env().opBuilder("CompareAndBitpack", scope.makeOpName("CompareAndBitpack"));
-    opBuilder.addInput(input.asOutput());
-    opBuilder.addInput(threshold.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
+    opBuilder.addInput(threshold.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new CompareAndBitpack(opBuilder.build());
   }
@@ -81,7 +81,7 @@ public Output<TUint8> output() {
   }
   
   @Override
-  public Output<TUint8> asOutput() {
+  public Output<TUint8> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/ComplexAbs.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/ComplexAbs.java
index d1c7f2c3d25..a384396d001 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/ComplexAbs.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/ComplexAbs.java
@@ -22,13 +22,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TFloat32;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * Computes the complex absolute value of a tensor.
@@ -41,7 +41,7 @@
  * @param <U> data type for {@code y()} output
  */
 @Operator(group = "math")
-public final class ComplexAbs<U extends Tensor & TNumber> extends RawOp implements Operand<U> {
+public final class ComplexAbs<U extends TNumber> extends RawOp implements Operand<U> {
   
   /**
    * Factory method to create a class wrapping a new ComplexAbs operation.
@@ -52,9 +52,9 @@ public final class ComplexAbs<U extends Tensor & TNumber> extends RawOp implemen
    * @return a new instance of ComplexAbs
    */
   @Endpoint(describeByClass = true)
-  public static <U extends Tensor & TNumber, T extends Tensor> ComplexAbs<U> create(Scope scope, Operand<T> x, DataType<U> Tout) {
+  public static <U extends TNumber, T extends TType> ComplexAbs<U> create(Scope scope, Operand<T> x, DataType<U> Tout) {
     OperationBuilder opBuilder = scope.env().opBuilder("ComplexAbs", scope.makeOpName("ComplexAbs"));
-    opBuilder.addInput(x.asOutput());
+    opBuilder.addInput(x.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("Tout", Tout);
     return new ComplexAbs<U>(opBuilder.build());
@@ -68,7 +68,7 @@ public static <U extends Tensor & TNumber, T extends Tensor> ComplexAbs<U> creat
    * @return a new instance of ComplexAbs
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> ComplexAbs<TFloat32> create(Scope scope, Operand<T> x) {
+  public static <T extends TType> ComplexAbs<TFloat32> create(Scope scope, Operand<T> x) {
     return create(scope, x, TFloat32.DTYPE);
   }
   
@@ -79,7 +79,7 @@ public Output<U> y() {
   }
   
   @Override
-  public Output<U> asOutput() {
+  public Output<U> asOutput(Scope scope) {
     return y;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Conj.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Conj.java
index 570f4bc5f29..27ca2ff1e88 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Conj.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Conj.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Returns the complex conjugate of a complex number.
@@ -47,7 +47,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "math")
-public final class Conj<T extends Tensor> extends RawOp implements Operand<T> {
+public final class Conj<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Conj operation.
@@ -57,9 +57,9 @@ public final class Conj<T extends Tensor> extends RawOp implements Operand<T> {
    * @return a new instance of Conj
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> Conj<T> create(Scope scope, Operand<T> input) {
+  public static <T extends TType> Conj<T> create(Scope scope, Operand<T> input) {
     OperationBuilder opBuilder = scope.env().opBuilder("Conj", scope.makeOpName("Conj"));
-    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new Conj<T>(opBuilder.build());
   }
@@ -71,7 +71,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Cos.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Cos.java
index 4bd6ef72d81..f35c3eddb16 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Cos.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Cos.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Computes cos of x element-wise.
@@ -44,7 +44,7 @@
  * @param <T> data type for {@code y()} output
  */
 @Operator(group = "math")
-public final class Cos<T extends Tensor> extends RawOp implements Operand<T> {
+public final class Cos<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Cos operation.
@@ -54,9 +54,9 @@ public final class Cos<T extends Tensor> extends RawOp implements Operand<T> {
    * @return a new instance of Cos
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> Cos<T> create(Scope scope, Operand<T> x) {
+  public static <T extends TType> Cos<T> create(Scope scope, Operand<T> x) {
     OperationBuilder opBuilder = scope.env().opBuilder("Cos", scope.makeOpName("Cos"));
-    opBuilder.addInput(x.asOutput());
+    opBuilder.addInput(x.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new Cos<T>(opBuilder.build());
   }
@@ -68,7 +68,7 @@ public Output<T> y() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return y;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Cosh.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Cosh.java
index 8dbde201831..554123b820d 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Cosh.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Cosh.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Computes hyperbolic cosine of x element-wise.
@@ -43,7 +43,7 @@
  * @param <T> data type for {@code y()} output
  */
 @Operator(group = "math")
-public final class Cosh<T extends Tensor> extends RawOp implements Operand<T> {
+public final class Cosh<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Cosh operation.
@@ -53,9 +53,9 @@ public final class Cosh<T extends Tensor> extends RawOp implements Operand<T> {
    * @return a new instance of Cosh
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> Cosh<T> create(Scope scope, Operand<T> x) {
+  public static <T extends TType> Cosh<T> create(Scope scope, Operand<T> x) {
     OperationBuilder opBuilder = scope.env().opBuilder("Cosh", scope.makeOpName("Cosh"));
-    opBuilder.addInput(x.asOutput());
+    opBuilder.addInput(x.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new Cosh<T>(opBuilder.build());
   }
@@ -67,7 +67,7 @@ public Output<T> y() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return y;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Cumprod.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Cumprod.java
index f7eb392a1a4..a2563630107 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Cumprod.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Cumprod.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * Compute the cumulative product of the tensor `x` along `axis`.
@@ -57,7 +57,7 @@
  * @param <T> data type for {@code out()} output
  */
 @Operator(group = "math")
-public final class Cumprod<T extends Tensor> extends RawOp implements Operand<T> {
+public final class Cumprod<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.math.Cumprod}
@@ -100,10 +100,10 @@ private Options() {
    * @return a new instance of Cumprod
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor, U extends Tensor & TNumber> Cumprod<T> create(Scope scope, Operand<T> x, Operand<U> axis, Options... options) {
+  public static <T extends TType, U extends TNumber> Cumprod<T> create(Scope scope, Operand<T> x, Operand<U> axis, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("Cumprod", scope.makeOpName("Cumprod"));
-    opBuilder.addInput(x.asOutput());
-    opBuilder.addInput(axis.asOutput());
+    opBuilder.addInput(x.asOutput(scope));
+    opBuilder.addInput(axis.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -139,7 +139,7 @@ public Output<T> out() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return out;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Cumsum.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Cumsum.java
index a2570547464..47761661bbe 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Cumsum.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Cumsum.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * Compute the cumulative sum of the tensor `x` along `axis`.
@@ -57,7 +57,7 @@
  * @param <T> data type for {@code out()} output
  */
 @Operator(group = "math")
-public final class Cumsum<T extends Tensor> extends RawOp implements Operand<T> {
+public final class Cumsum<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.math.Cumsum}
@@ -100,10 +100,10 @@ private Options() {
    * @return a new instance of Cumsum
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor, U extends Tensor & TNumber> Cumsum<T> create(Scope scope, Operand<T> x, Operand<U> axis, Options... options) {
+  public static <T extends TType, U extends TNumber> Cumsum<T> create(Scope scope, Operand<T> x, Operand<U> axis, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("Cumsum", scope.makeOpName("Cumsum"));
-    opBuilder.addInput(x.asOutput());
-    opBuilder.addInput(axis.asOutput());
+    opBuilder.addInput(x.asOutput(scope));
+    opBuilder.addInput(axis.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -139,7 +139,7 @@ public Output<T> out() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return out;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/CumulativeLogsumexp.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/CumulativeLogsumexp.java
index 2f741650242..640d4cb05ca 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/CumulativeLogsumexp.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/CumulativeLogsumexp.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -51,7 +50,7 @@
  * 
  * @param <T> data type for {@code out()} output
  */
-public final class CumulativeLogsumexp<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class CumulativeLogsumexp<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.math.CumulativeLogsumexp}
@@ -92,10 +91,10 @@ private Options() {
    * @return a new instance of CumulativeLogsumexp
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber, U extends Tensor & TNumber> CumulativeLogsumexp<T> create(Scope scope, Operand<T> x, Operand<U> axis, Options... options) {
+  public static <T extends TNumber, U extends TNumber> CumulativeLogsumexp<T> create(Scope scope, Operand<T> x, Operand<U> axis, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("CumulativeLogsumexp", scope.makeOpName("CumulativeLogsumexp"));
-    opBuilder.addInput(x.asOutput());
-    opBuilder.addInput(axis.asOutput());
+    opBuilder.addInput(x.asOutput(scope));
+    opBuilder.addInput(axis.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -131,7 +130,7 @@ public Output<T> out() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return out;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Digamma.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Digamma.java
index dde2fb8eb97..d1d87138af7 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Digamma.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Digamma.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -36,7 +35,7 @@
  * @param <T> data type for {@code y()} output
  */
 @Operator(group = "math")
-public final class Digamma<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class Digamma<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Digamma operation.
@@ -46,9 +45,9 @@ public final class Digamma<T extends Tensor & TNumber> extends RawOp implements
    * @return a new instance of Digamma
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> Digamma<T> create(Scope scope, Operand<T> x) {
+  public static <T extends TNumber> Digamma<T> create(Scope scope, Operand<T> x) {
     OperationBuilder opBuilder = scope.env().opBuilder("Digamma", scope.makeOpName("Digamma"));
-    opBuilder.addInput(x.asOutput());
+    opBuilder.addInput(x.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new Digamma<T>(opBuilder.build());
   }
@@ -60,7 +59,7 @@ public Output<T> y() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return y;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Div.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Div.java
index 6a9962e93d4..f86f9909a49 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Div.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Div.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Returns x / y element-wise.
@@ -36,7 +36,7 @@
  * @param <T> data type for {@code z()} output
  */
 @Operator(group = "math")
-public final class Div<T extends Tensor> extends RawOp implements Operand<T> {
+public final class Div<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Div operation.
@@ -47,10 +47,10 @@ public final class Div<T extends Tensor> extends RawOp implements Operand<T> {
    * @return a new instance of Div
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> Div<T> create(Scope scope, Operand<T> x, Operand<T> y) {
+  public static <T extends TType> Div<T> create(Scope scope, Operand<T> x, Operand<T> y) {
     OperationBuilder opBuilder = scope.env().opBuilder("Div", scope.makeOpName("Div"));
-    opBuilder.addInput(x.asOutput());
-    opBuilder.addInput(y.asOutput());
+    opBuilder.addInput(x.asOutput(scope));
+    opBuilder.addInput(y.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new Div<T>(opBuilder.build());
   }
@@ -62,7 +62,7 @@ public Output<T> z() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return z;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/DivNoNan.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/DivNoNan.java
index 913a118cdb1..3f4607025fd 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/DivNoNan.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/DivNoNan.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Returns 0 if the denominator is zero.
@@ -37,7 +37,7 @@
  * @param <T> data type for {@code z()} output
  */
 @Operator(group = "math")
-public final class DivNoNan<T extends Tensor> extends RawOp implements Operand<T> {
+public final class DivNoNan<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new DivNoNan operation.
@@ -48,10 +48,10 @@ public final class DivNoNan<T extends Tensor> extends RawOp implements Operand<T
    * @return a new instance of DivNoNan
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> DivNoNan<T> create(Scope scope, Operand<T> x, Operand<T> y) {
+  public static <T extends TType> DivNoNan<T> create(Scope scope, Operand<T> x, Operand<T> y) {
     OperationBuilder opBuilder = scope.env().opBuilder("DivNoNan", scope.makeOpName("DivNoNan"));
-    opBuilder.addInput(x.asOutput());
-    opBuilder.addInput(y.asOutput());
+    opBuilder.addInput(x.asOutput(scope));
+    opBuilder.addInput(y.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new DivNoNan<T>(opBuilder.build());
   }
@@ -63,7 +63,7 @@ public Output<T> z() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return z;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Equal.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Equal.java
index 81e0e7ce04c..868bb3e0690 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Equal.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Equal.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TBool;
+import org.tensorflow.types.family.TType;
 
 /**
  * Returns the truth value of (x == y) element-wise.
@@ -76,10 +76,10 @@ private Options() {
    * @return a new instance of Equal
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> Equal create(Scope scope, Operand<T> x, Operand<T> y, Options... options) {
+  public static <T extends TType> Equal create(Scope scope, Operand<T> x, Operand<T> y, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("Equal", scope.makeOpName("Equal"));
-    opBuilder.addInput(x.asOutput());
-    opBuilder.addInput(y.asOutput());
+    opBuilder.addInput(x.asOutput(scope));
+    opBuilder.addInput(y.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -105,7 +105,7 @@ public Output<TBool> z() {
   }
   
   @Override
-  public Output<TBool> asOutput() {
+  public Output<TBool> asOutput(Scope scope) {
     return z;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Erf.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Erf.java
index 9ef35983233..f335ca9acf9 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Erf.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Erf.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -34,7 +33,7 @@
  * @param <T> data type for {@code y()} output
  */
 @Operator(group = "math")
-public final class Erf<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class Erf<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Erf operation.
@@ -44,9 +43,9 @@ public final class Erf<T extends Tensor & TNumber> extends RawOp implements Oper
    * @return a new instance of Erf
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> Erf<T> create(Scope scope, Operand<T> x) {
+  public static <T extends TNumber> Erf<T> create(Scope scope, Operand<T> x) {
     OperationBuilder opBuilder = scope.env().opBuilder("Erf", scope.makeOpName("Erf"));
-    opBuilder.addInput(x.asOutput());
+    opBuilder.addInput(x.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new Erf<T>(opBuilder.build());
   }
@@ -58,7 +57,7 @@ public Output<T> y() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return y;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Erfc.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Erfc.java
index 9910632e7ff..1eb9d1eb3d0 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Erfc.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Erfc.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -34,7 +33,7 @@
  * @param <T> data type for {@code y()} output
  */
 @Operator(group = "math")
-public final class Erfc<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class Erfc<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Erfc operation.
@@ -44,9 +43,9 @@ public final class Erfc<T extends Tensor & TNumber> extends RawOp implements Ope
    * @return a new instance of Erfc
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> Erfc<T> create(Scope scope, Operand<T> x) {
+  public static <T extends TNumber> Erfc<T> create(Scope scope, Operand<T> x) {
     OperationBuilder opBuilder = scope.env().opBuilder("Erfc", scope.makeOpName("Erfc"));
-    opBuilder.addInput(x.asOutput());
+    opBuilder.addInput(x.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new Erfc<T>(opBuilder.build());
   }
@@ -58,7 +57,7 @@ public Output<T> y() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return y;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Exp.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Exp.java
index 409e0d4124b..e37760b90f4 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Exp.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Exp.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Computes exponential of x element-wise.  \\(y = e^x\\).
@@ -59,7 +59,7 @@
  * @param <T> data type for {@code y()} output
  */
 @Operator(group = "math")
-public final class Exp<T extends Tensor> extends RawOp implements Operand<T> {
+public final class Exp<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Exp operation.
@@ -69,9 +69,9 @@ public final class Exp<T extends Tensor> extends RawOp implements Operand<T> {
    * @return a new instance of Exp
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> Exp<T> create(Scope scope, Operand<T> x) {
+  public static <T extends TType> Exp<T> create(Scope scope, Operand<T> x) {
     OperationBuilder opBuilder = scope.env().opBuilder("Exp", scope.makeOpName("Exp"));
-    opBuilder.addInput(x.asOutput());
+    opBuilder.addInput(x.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new Exp<T>(opBuilder.build());
   }
@@ -83,7 +83,7 @@ public Output<T> y() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return y;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Expm1.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Expm1.java
index fe81e5afcf4..ad951dc71aa 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Expm1.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Expm1.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Computes `exp(x) - 1` element-wise.
@@ -48,7 +48,7 @@
  * @param <T> data type for {@code y()} output
  */
 @Operator(group = "math")
-public final class Expm1<T extends Tensor> extends RawOp implements Operand<T> {
+public final class Expm1<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Expm1 operation.
@@ -58,9 +58,9 @@ public final class Expm1<T extends Tensor> extends RawOp implements Operand<T> {
    * @return a new instance of Expm1
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> Expm1<T> create(Scope scope, Operand<T> x) {
+  public static <T extends TType> Expm1<T> create(Scope scope, Operand<T> x) {
     OperationBuilder opBuilder = scope.env().opBuilder("Expm1", scope.makeOpName("Expm1"));
-    opBuilder.addInput(x.asOutput());
+    opBuilder.addInput(x.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new Expm1<T>(opBuilder.build());
   }
@@ -72,7 +72,7 @@ public Output<T> y() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return y;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Fact.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Fact.java
index 31d78966166..dd199eb7759 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Fact.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Fact.java
@@ -53,7 +53,7 @@ public Output<TString> fact() {
   }
   
   @Override
-  public Output<TString> asOutput() {
+  public Output<TString> asOutput(Scope scope) {
     return fact;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Floor.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Floor.java
index 924833d7f22..190c1c3f7a1 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Floor.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Floor.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -34,7 +33,7 @@
  * @param <T> data type for {@code y()} output
  */
 @Operator(group = "math")
-public final class Floor<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class Floor<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Floor operation.
@@ -44,9 +43,9 @@ public final class Floor<T extends Tensor & TNumber> extends RawOp implements Op
    * @return a new instance of Floor
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> Floor<T> create(Scope scope, Operand<T> x) {
+  public static <T extends TNumber> Floor<T> create(Scope scope, Operand<T> x) {
     OperationBuilder opBuilder = scope.env().opBuilder("Floor", scope.makeOpName("Floor"));
-    opBuilder.addInput(x.asOutput());
+    opBuilder.addInput(x.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new Floor<T>(opBuilder.build());
   }
@@ -58,7 +57,7 @@ public Output<T> y() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return y;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/FloorDiv.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/FloorDiv.java
index cb7148d2e7b..2d92c796459 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/FloorDiv.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/FloorDiv.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Returns x // y element-wise.
@@ -36,7 +36,7 @@
  * @param <T> data type for {@code z()} output
  */
 @Operator(group = "math")
-public final class FloorDiv<T extends Tensor> extends RawOp implements Operand<T> {
+public final class FloorDiv<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new FloorDiv operation.
@@ -47,10 +47,10 @@ public final class FloorDiv<T extends Tensor> extends RawOp implements Operand<T
    * @return a new instance of FloorDiv
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> FloorDiv<T> create(Scope scope, Operand<T> x, Operand<T> y) {
+  public static <T extends TType> FloorDiv<T> create(Scope scope, Operand<T> x, Operand<T> y) {
     OperationBuilder opBuilder = scope.env().opBuilder("FloorDiv", scope.makeOpName("FloorDiv"));
-    opBuilder.addInput(x.asOutput());
-    opBuilder.addInput(y.asOutput());
+    opBuilder.addInput(x.asOutput(scope));
+    opBuilder.addInput(y.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new FloorDiv<T>(opBuilder.build());
   }
@@ -62,7 +62,7 @@ public Output<T> z() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return z;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/FloorMod.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/FloorMod.java
index a5fcf94ac58..a8502331592 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/FloorMod.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/FloorMod.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -40,7 +39,7 @@
  * @param <T> data type for {@code z()} output
  */
 @Operator(group = "math")
-public final class FloorMod<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class FloorMod<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new FloorMod operation.
@@ -51,10 +50,10 @@ public final class FloorMod<T extends Tensor & TNumber> extends RawOp implements
    * @return a new instance of FloorMod
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> FloorMod<T> create(Scope scope, Operand<T> x, Operand<T> y) {
+  public static <T extends TNumber> FloorMod<T> create(Scope scope, Operand<T> x, Operand<T> y) {
     OperationBuilder opBuilder = scope.env().opBuilder("FloorMod", scope.makeOpName("FloorMod"));
-    opBuilder.addInput(x.asOutput());
-    opBuilder.addInput(y.asOutput());
+    opBuilder.addInput(x.asOutput(scope));
+    opBuilder.addInput(y.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new FloorMod<T>(opBuilder.build());
   }
@@ -66,7 +65,7 @@ public Output<T> z() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return z;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Greater.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Greater.java
index 7d1df45c855..5a5ab1ce6e8 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Greater.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Greater.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -59,10 +58,10 @@ public final class Greater extends RawOp implements Operand<TBool> {
    * @return a new instance of Greater
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> Greater create(Scope scope, Operand<T> x, Operand<T> y) {
+  public static <T extends TNumber> Greater create(Scope scope, Operand<T> x, Operand<T> y) {
     OperationBuilder opBuilder = scope.env().opBuilder("Greater", scope.makeOpName("Greater"));
-    opBuilder.addInput(x.asOutput());
-    opBuilder.addInput(y.asOutput());
+    opBuilder.addInput(x.asOutput(scope));
+    opBuilder.addInput(y.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new Greater(opBuilder.build());
   }
@@ -74,7 +73,7 @@ public Output<TBool> z() {
   }
   
   @Override
-  public Output<TBool> asOutput() {
+  public Output<TBool> asOutput(Scope scope) {
     return z;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/GreaterEqual.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/GreaterEqual.java
index b2b0cebbd12..88543131abb 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/GreaterEqual.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/GreaterEqual.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -59,10 +58,10 @@ public final class GreaterEqual extends RawOp implements Operand<TBool> {
    * @return a new instance of GreaterEqual
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> GreaterEqual create(Scope scope, Operand<T> x, Operand<T> y) {
+  public static <T extends TNumber> GreaterEqual create(Scope scope, Operand<T> x, Operand<T> y) {
     OperationBuilder opBuilder = scope.env().opBuilder("GreaterEqual", scope.makeOpName("GreaterEqual"));
-    opBuilder.addInput(x.asOutput());
-    opBuilder.addInput(y.asOutput());
+    opBuilder.addInput(x.asOutput(scope));
+    opBuilder.addInput(y.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new GreaterEqual(opBuilder.build());
   }
@@ -74,7 +73,7 @@ public Output<TBool> z() {
   }
   
   @Override
-  public Output<TBool> asOutput() {
+  public Output<TBool> asOutput(Scope scope) {
     return z;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Igamma.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Igamma.java
index a0427886360..141cc603bae 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Igamma.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Igamma.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -47,7 +46,7 @@
  * @param <T> data type for {@code z()} output
  */
 @Operator(group = "math")
-public final class Igamma<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class Igamma<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Igamma operation.
@@ -58,10 +57,10 @@ public final class Igamma<T extends Tensor & TNumber> extends RawOp implements O
    * @return a new instance of Igamma
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> Igamma<T> create(Scope scope, Operand<T> a, Operand<T> x) {
+  public static <T extends TNumber> Igamma<T> create(Scope scope, Operand<T> a, Operand<T> x) {
     OperationBuilder opBuilder = scope.env().opBuilder("Igamma", scope.makeOpName("Igamma"));
-    opBuilder.addInput(a.asOutput());
-    opBuilder.addInput(x.asOutput());
+    opBuilder.addInput(a.asOutput(scope));
+    opBuilder.addInput(x.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new Igamma<T>(opBuilder.build());
   }
@@ -73,7 +72,7 @@ public Output<T> z() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return z;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/IgammaGradA.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/IgammaGradA.java
index 4938de73273..5320b437133 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/IgammaGradA.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/IgammaGradA.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -33,7 +32,7 @@
  * 
  * @param <T> data type for {@code z()} output
  */
-public final class IgammaGradA<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class IgammaGradA<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new IgammaGradA operation.
@@ -44,10 +43,10 @@ public final class IgammaGradA<T extends Tensor & TNumber> extends RawOp impleme
    * @return a new instance of IgammaGradA
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> IgammaGradA<T> create(Scope scope, Operand<T> a, Operand<T> x) {
+  public static <T extends TNumber> IgammaGradA<T> create(Scope scope, Operand<T> a, Operand<T> x) {
     OperationBuilder opBuilder = scope.env().opBuilder("IgammaGradA", scope.makeOpName("IgammaGradA"));
-    opBuilder.addInput(a.asOutput());
-    opBuilder.addInput(x.asOutput());
+    opBuilder.addInput(a.asOutput(scope));
+    opBuilder.addInput(x.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new IgammaGradA<T>(opBuilder.build());
   }
@@ -59,7 +58,7 @@ public Output<T> z() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return z;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Igammac.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Igammac.java
index 6a0053183c0..597e944d1fb 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Igammac.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Igammac.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -47,7 +46,7 @@
  * @param <T> data type for {@code z()} output
  */
 @Operator(group = "math")
-public final class Igammac<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class Igammac<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Igammac operation.
@@ -58,10 +57,10 @@ public final class Igammac<T extends Tensor & TNumber> extends RawOp implements
    * @return a new instance of Igammac
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> Igammac<T> create(Scope scope, Operand<T> a, Operand<T> x) {
+  public static <T extends TNumber> Igammac<T> create(Scope scope, Operand<T> a, Operand<T> x) {
     OperationBuilder opBuilder = scope.env().opBuilder("Igammac", scope.makeOpName("Igammac"));
-    opBuilder.addInput(a.asOutput());
-    opBuilder.addInput(x.asOutput());
+    opBuilder.addInput(a.asOutput(scope));
+    opBuilder.addInput(x.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new Igammac<T>(opBuilder.build());
   }
@@ -73,7 +72,7 @@ public Output<T> z() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return z;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Imag.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Imag.java
index b706b5fea5c..3dd31bcc850 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Imag.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Imag.java
@@ -22,13 +22,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TFloat32;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * Returns the imaginary part of a complex number.
@@ -48,7 +48,7 @@
  * @param <U> data type for {@code output()} output
  */
 @Operator(group = "math")
-public final class Imag<U extends Tensor & TNumber> extends RawOp implements Operand<U> {
+public final class Imag<U extends TNumber> extends RawOp implements Operand<U> {
   
   /**
    * Factory method to create a class wrapping a new Imag operation.
@@ -59,9 +59,9 @@ public final class Imag<U extends Tensor & TNumber> extends RawOp implements Ope
    * @return a new instance of Imag
    */
   @Endpoint(describeByClass = true)
-  public static <U extends Tensor & TNumber, T extends Tensor> Imag<U> create(Scope scope, Operand<T> input, DataType<U> Tout) {
+  public static <U extends TNumber, T extends TType> Imag<U> create(Scope scope, Operand<T> input, DataType<U> Tout) {
     OperationBuilder opBuilder = scope.env().opBuilder("Imag", scope.makeOpName("Imag"));
-    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("Tout", Tout);
     return new Imag<U>(opBuilder.build());
@@ -75,7 +75,7 @@ public static <U extends Tensor & TNumber, T extends Tensor> Imag<U> create(Scop
    * @return a new instance of Imag
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> Imag<TFloat32> create(Scope scope, Operand<T> input) {
+  public static <T extends TType> Imag<TFloat32> create(Scope scope, Operand<T> input) {
     return create(scope, input, TFloat32.DTYPE);
   }
   
@@ -86,7 +86,7 @@ public Output<U> output() {
   }
   
   @Override
-  public Output<U> asOutput() {
+  public Output<U> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/InvertPermutation.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/InvertPermutation.java
index 2cd61865d17..e6f38ded46b 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/InvertPermutation.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/InvertPermutation.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -50,7 +49,7 @@
  * @param <T> data type for {@code y()} output
  */
 @Operator(group = "math")
-public final class InvertPermutation<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class InvertPermutation<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new InvertPermutation operation.
@@ -60,9 +59,9 @@ public final class InvertPermutation<T extends Tensor & TNumber> extends RawOp i
    * @return a new instance of InvertPermutation
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> InvertPermutation<T> create(Scope scope, Operand<T> x) {
+  public static <T extends TNumber> InvertPermutation<T> create(Scope scope, Operand<T> x) {
     OperationBuilder opBuilder = scope.env().opBuilder("InvertPermutation", scope.makeOpName("InvertPermutation"));
-    opBuilder.addInput(x.asOutput());
+    opBuilder.addInput(x.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new InvertPermutation<T>(opBuilder.build());
   }
@@ -75,7 +74,7 @@ public Output<T> y() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return y;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/IsFinite.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/IsFinite.java
index e97e7be7219..ec1566a22ae 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/IsFinite.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/IsFinite.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -54,9 +53,9 @@ public final class IsFinite extends RawOp implements Operand<TBool> {
    * @return a new instance of IsFinite
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> IsFinite create(Scope scope, Operand<T> x) {
+  public static <T extends TNumber> IsFinite create(Scope scope, Operand<T> x) {
     OperationBuilder opBuilder = scope.env().opBuilder("IsFinite", scope.makeOpName("IsFinite"));
-    opBuilder.addInput(x.asOutput());
+    opBuilder.addInput(x.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new IsFinite(opBuilder.build());
   }
@@ -68,7 +67,7 @@ public Output<TBool> y() {
   }
   
   @Override
-  public Output<TBool> asOutput() {
+  public Output<TBool> asOutput(Scope scope) {
     return y;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/IsInf.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/IsInf.java
index 15c7219c6c0..7b91b7e4563 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/IsInf.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/IsInf.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -54,9 +53,9 @@ public final class IsInf extends RawOp implements Operand<TBool> {
    * @return a new instance of IsInf
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> IsInf create(Scope scope, Operand<T> x) {
+  public static <T extends TNumber> IsInf create(Scope scope, Operand<T> x) {
     OperationBuilder opBuilder = scope.env().opBuilder("IsInf", scope.makeOpName("IsInf"));
-    opBuilder.addInput(x.asOutput());
+    opBuilder.addInput(x.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new IsInf(opBuilder.build());
   }
@@ -68,7 +67,7 @@ public Output<TBool> y() {
   }
   
   @Override
-  public Output<TBool> asOutput() {
+  public Output<TBool> asOutput(Scope scope) {
     return y;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/IsNan.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/IsNan.java
index bfcecce7968..c02865ab759 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/IsNan.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/IsNan.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -54,9 +53,9 @@ public final class IsNan extends RawOp implements Operand<TBool> {
    * @return a new instance of IsNan
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> IsNan create(Scope scope, Operand<T> x) {
+  public static <T extends TNumber> IsNan create(Scope scope, Operand<T> x) {
     OperationBuilder opBuilder = scope.env().opBuilder("IsNan", scope.makeOpName("IsNan"));
-    opBuilder.addInput(x.asOutput());
+    opBuilder.addInput(x.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new IsNan(opBuilder.build());
   }
@@ -68,7 +67,7 @@ public Output<TBool> y() {
   }
   
   @Override
-  public Output<TBool> asOutput() {
+  public Output<TBool> asOutput(Scope scope) {
     return y;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Less.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Less.java
index 8908b2e210d..7ce3046b5b7 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Less.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Less.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -59,10 +58,10 @@ public final class Less extends RawOp implements Operand<TBool> {
    * @return a new instance of Less
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> Less create(Scope scope, Operand<T> x, Operand<T> y) {
+  public static <T extends TNumber> Less create(Scope scope, Operand<T> x, Operand<T> y) {
     OperationBuilder opBuilder = scope.env().opBuilder("Less", scope.makeOpName("Less"));
-    opBuilder.addInput(x.asOutput());
-    opBuilder.addInput(y.asOutput());
+    opBuilder.addInput(x.asOutput(scope));
+    opBuilder.addInput(y.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new Less(opBuilder.build());
   }
@@ -74,7 +73,7 @@ public Output<TBool> z() {
   }
   
   @Override
-  public Output<TBool> asOutput() {
+  public Output<TBool> asOutput(Scope scope) {
     return z;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/LessEqual.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/LessEqual.java
index e4463ed29cd..0e36cbc0223 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/LessEqual.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/LessEqual.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -59,10 +58,10 @@ public final class LessEqual extends RawOp implements Operand<TBool> {
    * @return a new instance of LessEqual
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> LessEqual create(Scope scope, Operand<T> x, Operand<T> y) {
+  public static <T extends TNumber> LessEqual create(Scope scope, Operand<T> x, Operand<T> y) {
     OperationBuilder opBuilder = scope.env().opBuilder("LessEqual", scope.makeOpName("LessEqual"));
-    opBuilder.addInput(x.asOutput());
-    opBuilder.addInput(y.asOutput());
+    opBuilder.addInput(x.asOutput(scope));
+    opBuilder.addInput(y.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new LessEqual(opBuilder.build());
   }
@@ -74,7 +73,7 @@ public Output<TBool> z() {
   }
   
   @Override
-  public Output<TBool> asOutput() {
+  public Output<TBool> asOutput(Scope scope) {
     return z;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Lgamma.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Lgamma.java
index d1284ca2b3f..d37aced67e9 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Lgamma.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Lgamma.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -44,7 +43,7 @@
  * @param <T> data type for {@code y()} output
  */
 @Operator(group = "math")
-public final class Lgamma<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class Lgamma<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Lgamma operation.
@@ -54,9 +53,9 @@ public final class Lgamma<T extends Tensor & TNumber> extends RawOp implements O
    * @return a new instance of Lgamma
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> Lgamma<T> create(Scope scope, Operand<T> x) {
+  public static <T extends TNumber> Lgamma<T> create(Scope scope, Operand<T> x) {
     OperationBuilder opBuilder = scope.env().opBuilder("Lgamma", scope.makeOpName("Lgamma"));
-    opBuilder.addInput(x.asOutput());
+    opBuilder.addInput(x.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new Lgamma<T>(opBuilder.build());
   }
@@ -68,7 +67,7 @@ public Output<T> y() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return y;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Log.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Log.java
index 017d46cfec9..a4c7d04325d 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Log.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Log.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Computes natural logarithm of x element-wise.
@@ -42,7 +42,7 @@
  * @param <T> data type for {@code y()} output
  */
 @Operator(group = "math")
-public final class Log<T extends Tensor> extends RawOp implements Operand<T> {
+public final class Log<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Log operation.
@@ -52,9 +52,9 @@ public final class Log<T extends Tensor> extends RawOp implements Operand<T> {
    * @return a new instance of Log
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> Log<T> create(Scope scope, Operand<T> x) {
+  public static <T extends TType> Log<T> create(Scope scope, Operand<T> x) {
     OperationBuilder opBuilder = scope.env().opBuilder("Log", scope.makeOpName("Log"));
-    opBuilder.addInput(x.asOutput());
+    opBuilder.addInput(x.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new Log<T>(opBuilder.build());
   }
@@ -66,7 +66,7 @@ public Output<T> y() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return y;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Log1p.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Log1p.java
index 1fb34539fa2..3f1c418d3ed 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Log1p.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Log1p.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Computes natural logarithm of (1 + x) element-wise.
@@ -42,7 +42,7 @@
  * @param <T> data type for {@code y()} output
  */
 @Operator(group = "math")
-public final class Log1p<T extends Tensor> extends RawOp implements Operand<T> {
+public final class Log1p<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Log1p operation.
@@ -52,9 +52,9 @@ public final class Log1p<T extends Tensor> extends RawOp implements Operand<T> {
    * @return a new instance of Log1p
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> Log1p<T> create(Scope scope, Operand<T> x) {
+  public static <T extends TType> Log1p<T> create(Scope scope, Operand<T> x) {
     OperationBuilder opBuilder = scope.env().opBuilder("Log1p", scope.makeOpName("Log1p"));
-    opBuilder.addInput(x.asOutput());
+    opBuilder.addInput(x.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new Log1p<T>(opBuilder.build());
   }
@@ -66,7 +66,7 @@ public Output<T> y() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return y;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/LogicalAnd.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/LogicalAnd.java
index 89bb2837f46..a77dfbe0a73 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/LogicalAnd.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/LogicalAnd.java
@@ -47,8 +47,8 @@ public final class LogicalAnd extends RawOp implements Operand<TBool> {
   @Endpoint(describeByClass = true)
   public static LogicalAnd create(Scope scope, Operand<TBool> x, Operand<TBool> y) {
     OperationBuilder opBuilder = scope.env().opBuilder("LogicalAnd", scope.makeOpName("LogicalAnd"));
-    opBuilder.addInput(x.asOutput());
-    opBuilder.addInput(y.asOutput());
+    opBuilder.addInput(x.asOutput(scope));
+    opBuilder.addInput(y.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new LogicalAnd(opBuilder.build());
   }
@@ -60,7 +60,7 @@ public Output<TBool> z() {
   }
   
   @Override
-  public Output<TBool> asOutput() {
+  public Output<TBool> asOutput(Scope scope) {
     return z;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/LogicalNot.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/LogicalNot.java
index f59c85fd320..58376781559 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/LogicalNot.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/LogicalNot.java
@@ -43,7 +43,7 @@ public final class LogicalNot extends RawOp implements Operand<TBool> {
   @Endpoint(describeByClass = true)
   public static LogicalNot create(Scope scope, Operand<TBool> x) {
     OperationBuilder opBuilder = scope.env().opBuilder("LogicalNot", scope.makeOpName("LogicalNot"));
-    opBuilder.addInput(x.asOutput());
+    opBuilder.addInput(x.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new LogicalNot(opBuilder.build());
   }
@@ -56,7 +56,7 @@ public Output<TBool> y() {
   }
   
   @Override
-  public Output<TBool> asOutput() {
+  public Output<TBool> asOutput(Scope scope) {
     return y;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/LogicalOr.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/LogicalOr.java
index 604e3ea18f5..39ae7295a27 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/LogicalOr.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/LogicalOr.java
@@ -47,8 +47,8 @@ public final class LogicalOr extends RawOp implements Operand<TBool> {
   @Endpoint(describeByClass = true)
   public static LogicalOr create(Scope scope, Operand<TBool> x, Operand<TBool> y) {
     OperationBuilder opBuilder = scope.env().opBuilder("LogicalOr", scope.makeOpName("LogicalOr"));
-    opBuilder.addInput(x.asOutput());
-    opBuilder.addInput(y.asOutput());
+    opBuilder.addInput(x.asOutput(scope));
+    opBuilder.addInput(y.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new LogicalOr(opBuilder.build());
   }
@@ -60,7 +60,7 @@ public Output<TBool> z() {
   }
   
   @Override
-  public Output<TBool> asOutput() {
+  public Output<TBool> asOutput(Scope scope) {
     return z;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Maximum.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Maximum.java
index 0cc2abb4da6..d9ea1d45870 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Maximum.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Maximum.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -37,7 +36,7 @@
  * @param <T> data type for {@code z()} output
  */
 @Operator(group = "math")
-public final class Maximum<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class Maximum<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Maximum operation.
@@ -48,10 +47,10 @@ public final class Maximum<T extends Tensor & TNumber> extends RawOp implements
    * @return a new instance of Maximum
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> Maximum<T> create(Scope scope, Operand<T> x, Operand<T> y) {
+  public static <T extends TNumber> Maximum<T> create(Scope scope, Operand<T> x, Operand<T> y) {
     OperationBuilder opBuilder = scope.env().opBuilder("Maximum", scope.makeOpName("Maximum"));
-    opBuilder.addInput(x.asOutput());
-    opBuilder.addInput(y.asOutput());
+    opBuilder.addInput(x.asOutput(scope));
+    opBuilder.addInput(y.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new Maximum<T>(opBuilder.build());
   }
@@ -63,7 +62,7 @@ public Output<T> z() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return z;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Mean.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Mean.java
index bf3641fb730..4923b136110 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Mean.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Mean.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * Computes the mean of elements across dimensions of a tensor.
@@ -39,7 +39,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "math")
-public final class Mean<T extends Tensor> extends RawOp implements Operand<T> {
+public final class Mean<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.math.Mean}
@@ -71,10 +71,10 @@ private Options() {
    * @return a new instance of Mean
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor, U extends Tensor & TNumber> Mean<T> create(Scope scope, Operand<T> input, Operand<U> axis, Options... options) {
+  public static <T extends TType, U extends TNumber> Mean<T> create(Scope scope, Operand<T> input, Operand<U> axis, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("Mean", scope.makeOpName("Mean"));
-    opBuilder.addInput(input.asOutput());
-    opBuilder.addInput(axis.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
+    opBuilder.addInput(axis.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -101,7 +101,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Minimum.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Minimum.java
index 8818d37fa92..a9a7934a26a 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Minimum.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Minimum.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -37,7 +36,7 @@
  * @param <T> data type for {@code z()} output
  */
 @Operator(group = "math")
-public final class Minimum<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class Minimum<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Minimum operation.
@@ -48,10 +47,10 @@ public final class Minimum<T extends Tensor & TNumber> extends RawOp implements
    * @return a new instance of Minimum
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> Minimum<T> create(Scope scope, Operand<T> x, Operand<T> y) {
+  public static <T extends TNumber> Minimum<T> create(Scope scope, Operand<T> x, Operand<T> y) {
     OperationBuilder opBuilder = scope.env().opBuilder("Minimum", scope.makeOpName("Minimum"));
-    opBuilder.addInput(x.asOutput());
-    opBuilder.addInput(y.asOutput());
+    opBuilder.addInput(x.asOutput(scope));
+    opBuilder.addInput(y.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new Minimum<T>(opBuilder.build());
   }
@@ -63,7 +62,7 @@ public Output<T> z() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return z;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Mod.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Mod.java
index e2ea89b0d51..db3c2dbd24e 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Mod.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Mod.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -40,7 +39,7 @@
  * @param <T> data type for {@code z()} output
  */
 @Operator(group = "math")
-public final class Mod<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class Mod<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Mod operation.
@@ -51,10 +50,10 @@ public final class Mod<T extends Tensor & TNumber> extends RawOp implements Oper
    * @return a new instance of Mod
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> Mod<T> create(Scope scope, Operand<T> x, Operand<T> y) {
+  public static <T extends TNumber> Mod<T> create(Scope scope, Operand<T> x, Operand<T> y) {
     OperationBuilder opBuilder = scope.env().opBuilder("Mod", scope.makeOpName("Mod"));
-    opBuilder.addInput(x.asOutput());
-    opBuilder.addInput(y.asOutput());
+    opBuilder.addInput(x.asOutput(scope));
+    opBuilder.addInput(y.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new Mod<T>(opBuilder.build());
   }
@@ -66,7 +65,7 @@ public Output<T> z() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return z;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Mul.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Mul.java
index 9a71a3eec7a..778a19ceb4c 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Mul.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Mul.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Returns x * y element-wise.
@@ -36,7 +36,7 @@
  * @param <T> data type for {@code z()} output
  */
 @Operator(group = "math")
-public final class Mul<T extends Tensor> extends RawOp implements Operand<T> {
+public final class Mul<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Mul operation.
@@ -47,10 +47,10 @@ public final class Mul<T extends Tensor> extends RawOp implements Operand<T> {
    * @return a new instance of Mul
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> Mul<T> create(Scope scope, Operand<T> x, Operand<T> y) {
+  public static <T extends TType> Mul<T> create(Scope scope, Operand<T> x, Operand<T> y) {
     OperationBuilder opBuilder = scope.env().opBuilder("Mul", scope.makeOpName("Mul"));
-    opBuilder.addInput(x.asOutput());
-    opBuilder.addInput(y.asOutput());
+    opBuilder.addInput(x.asOutput(scope));
+    opBuilder.addInput(y.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new Mul<T>(opBuilder.build());
   }
@@ -62,7 +62,7 @@ public Output<T> z() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return z;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/MulNoNan.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/MulNoNan.java
index eda21a38bf0..4c73331b7ca 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/MulNoNan.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/MulNoNan.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Returns x * y element-wise. Returns zero if y is zero, even if x if infinite or NaN.
@@ -36,7 +36,7 @@
  * @param <T> data type for {@code z()} output
  */
 @Operator(group = "math")
-public final class MulNoNan<T extends Tensor> extends RawOp implements Operand<T> {
+public final class MulNoNan<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new MulNoNan operation.
@@ -47,10 +47,10 @@ public final class MulNoNan<T extends Tensor> extends RawOp implements Operand<T
    * @return a new instance of MulNoNan
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> MulNoNan<T> create(Scope scope, Operand<T> x, Operand<T> y) {
+  public static <T extends TType> MulNoNan<T> create(Scope scope, Operand<T> x, Operand<T> y) {
     OperationBuilder opBuilder = scope.env().opBuilder("MulNoNan", scope.makeOpName("MulNoNan"));
-    opBuilder.addInput(x.asOutput());
-    opBuilder.addInput(y.asOutput());
+    opBuilder.addInput(x.asOutput(scope));
+    opBuilder.addInput(y.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new MulNoNan<T>(opBuilder.build());
   }
@@ -62,7 +62,7 @@ public Output<T> z() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return z;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Ndtri.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Ndtri.java
index 8d9248f6012..72108ba8d51 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Ndtri.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Ndtri.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -32,7 +31,7 @@
  * @param <T> data type for {@code y()} output
  */
 @Operator(group = "math")
-public final class Ndtri<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class Ndtri<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Ndtri operation.
@@ -42,9 +41,9 @@ public final class Ndtri<T extends Tensor & TNumber> extends RawOp implements Op
    * @return a new instance of Ndtri
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> Ndtri<T> create(Scope scope, Operand<T> x) {
+  public static <T extends TNumber> Ndtri<T> create(Scope scope, Operand<T> x) {
     OperationBuilder opBuilder = scope.env().opBuilder("Ndtri", scope.makeOpName("Ndtri"));
-    opBuilder.addInput(x.asOutput());
+    opBuilder.addInput(x.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new Ndtri<T>(opBuilder.build());
   }
@@ -56,7 +55,7 @@ public Output<T> y() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return y;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Neg.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Neg.java
index 341809a7260..90615ef3cbe 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Neg.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Neg.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Computes numerical negative value element-wise.
@@ -35,7 +35,7 @@
  * @param <T> data type for {@code y()} output
  */
 @Operator(group = "math")
-public final class Neg<T extends Tensor> extends RawOp implements Operand<T> {
+public final class Neg<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Neg operation.
@@ -45,9 +45,9 @@ public final class Neg<T extends Tensor> extends RawOp implements Operand<T> {
    * @return a new instance of Neg
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> Neg<T> create(Scope scope, Operand<T> x) {
+  public static <T extends TType> Neg<T> create(Scope scope, Operand<T> x) {
     OperationBuilder opBuilder = scope.env().opBuilder("Neg", scope.makeOpName("Neg"));
-    opBuilder.addInput(x.asOutput());
+    opBuilder.addInput(x.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new Neg<T>(opBuilder.build());
   }
@@ -59,7 +59,7 @@ public Output<T> y() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return y;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/NextAfter.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/NextAfter.java
index 3d12fe37f42..11bb2c1699a 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/NextAfter.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/NextAfter.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -42,7 +41,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "math")
-public final class NextAfter<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class NextAfter<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new NextAfter operation.
@@ -53,10 +52,10 @@ public final class NextAfter<T extends Tensor & TNumber> extends RawOp implement
    * @return a new instance of NextAfter
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> NextAfter<T> create(Scope scope, Operand<T> x1, Operand<T> x2) {
+  public static <T extends TNumber> NextAfter<T> create(Scope scope, Operand<T> x1, Operand<T> x2) {
     OperationBuilder opBuilder = scope.env().opBuilder("NextAfter", scope.makeOpName("NextAfter"));
-    opBuilder.addInput(x1.asOutput());
-    opBuilder.addInput(x2.asOutput());
+    opBuilder.addInput(x1.asOutput(scope));
+    opBuilder.addInput(x2.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new NextAfter<T>(opBuilder.build());
   }
@@ -68,7 +67,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/NotEqual.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/NotEqual.java
index 7beddd8f94f..0ca58afe9b9 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/NotEqual.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/NotEqual.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TBool;
+import org.tensorflow.types.family.TType;
 
 /**
  * Returns the truth value of (x != y) element-wise.
@@ -66,10 +66,10 @@ private Options() {
    * @return a new instance of NotEqual
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> NotEqual create(Scope scope, Operand<T> x, Operand<T> y, Options... options) {
+  public static <T extends TType> NotEqual create(Scope scope, Operand<T> x, Operand<T> y, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("NotEqual", scope.makeOpName("NotEqual"));
-    opBuilder.addInput(x.asOutput());
-    opBuilder.addInput(y.asOutput());
+    opBuilder.addInput(x.asOutput(scope));
+    opBuilder.addInput(y.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -95,7 +95,7 @@ public Output<TBool> z() {
   }
   
   @Override
-  public Output<TBool> asOutput() {
+  public Output<TBool> asOutput(Scope scope) {
     return z;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Polygamma.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Polygamma.java
index f8ac90aa7d4..792f45e251b 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Polygamma.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Polygamma.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -41,7 +40,7 @@
  * @param <T> data type for {@code z()} output
  */
 @Operator(group = "math")
-public final class Polygamma<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class Polygamma<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Polygamma operation.
@@ -52,10 +51,10 @@ public final class Polygamma<T extends Tensor & TNumber> extends RawOp implement
    * @return a new instance of Polygamma
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> Polygamma<T> create(Scope scope, Operand<T> a, Operand<T> x) {
+  public static <T extends TNumber> Polygamma<T> create(Scope scope, Operand<T> a, Operand<T> x) {
     OperationBuilder opBuilder = scope.env().opBuilder("Polygamma", scope.makeOpName("Polygamma"));
-    opBuilder.addInput(a.asOutput());
-    opBuilder.addInput(x.asOutput());
+    opBuilder.addInput(a.asOutput(scope));
+    opBuilder.addInput(x.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new Polygamma<T>(opBuilder.build());
   }
@@ -67,7 +66,7 @@ public Output<T> z() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return z;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/PopulationCount.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/PopulationCount.java
index 1a7e9cd65c6..aa56a025b24 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/PopulationCount.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/PopulationCount.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -50,9 +49,9 @@ public final class PopulationCount extends RawOp implements Operand<TUint8> {
    * @return a new instance of PopulationCount
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> PopulationCount create(Scope scope, Operand<T> x) {
+  public static <T extends TNumber> PopulationCount create(Scope scope, Operand<T> x) {
     OperationBuilder opBuilder = scope.env().opBuilder("PopulationCount", scope.makeOpName("PopulationCount"));
-    opBuilder.addInput(x.asOutput());
+    opBuilder.addInput(x.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new PopulationCount(opBuilder.build());
   }
@@ -64,7 +63,7 @@ public Output<TUint8> y() {
   }
   
   @Override
-  public Output<TUint8> asOutput() {
+  public Output<TUint8> asOutput(Scope scope) {
     return y;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Pow.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Pow.java
index b6195fa9e33..d632c6b6bb4 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Pow.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Pow.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Computes the power of one value to another.
@@ -42,7 +42,7 @@
  * @param <T> data type for {@code z()} output
  */
 @Operator(group = "math")
-public final class Pow<T extends Tensor> extends RawOp implements Operand<T> {
+public final class Pow<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Pow operation.
@@ -53,10 +53,10 @@ public final class Pow<T extends Tensor> extends RawOp implements Operand<T> {
    * @return a new instance of Pow
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> Pow<T> create(Scope scope, Operand<T> x, Operand<T> y) {
+  public static <T extends TType> Pow<T> create(Scope scope, Operand<T> x, Operand<T> y) {
     OperationBuilder opBuilder = scope.env().opBuilder("Pow", scope.makeOpName("Pow"));
-    opBuilder.addInput(x.asOutput());
-    opBuilder.addInput(y.asOutput());
+    opBuilder.addInput(x.asOutput(scope));
+    opBuilder.addInput(y.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new Pow<T>(opBuilder.build());
   }
@@ -68,7 +68,7 @@ public Output<T> z() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return z;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/QuantizedAdd.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/QuantizedAdd.java
index a75ed47822b..a35716ae88c 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/QuantizedAdd.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/QuantizedAdd.java
@@ -22,12 +22,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.family.TType;
 
 /**
  * Returns x + y element-wise, working on quantized buffers.
@@ -35,7 +35,7 @@
  * @param <V> data type for {@code z()} output
  */
 @Operator(group = "math")
-public final class QuantizedAdd<V extends Tensor> extends RawOp {
+public final class QuantizedAdd<V extends TType> extends RawOp {
   
   /**
    * Factory method to create a class wrapping a new QuantizedAdd operation.
@@ -51,14 +51,14 @@ public final class QuantizedAdd<V extends Tensor> extends RawOp {
    * @return a new instance of QuantizedAdd
    */
   @Endpoint(describeByClass = true)
-  public static <V extends Tensor, T extends Tensor, U extends Tensor> QuantizedAdd<V> create(Scope scope, Operand<T> x, Operand<U> y, Operand<TFloat32> minX, Operand<TFloat32> maxX, Operand<TFloat32> minY, Operand<TFloat32> maxY, DataType<V> Toutput) {
+  public static <V extends TType, T extends TType, U extends TType> QuantizedAdd<V> create(Scope scope, Operand<T> x, Operand<U> y, Operand<TFloat32> minX, Operand<TFloat32> maxX, Operand<TFloat32> minY, Operand<TFloat32> maxY, DataType<V> Toutput) {
     OperationBuilder opBuilder = scope.env().opBuilder("QuantizedAdd", scope.makeOpName("QuantizedAdd"));
-    opBuilder.addInput(x.asOutput());
-    opBuilder.addInput(y.asOutput());
-    opBuilder.addInput(minX.asOutput());
-    opBuilder.addInput(maxX.asOutput());
-    opBuilder.addInput(minY.asOutput());
-    opBuilder.addInput(maxY.asOutput());
+    opBuilder.addInput(x.asOutput(scope));
+    opBuilder.addInput(y.asOutput(scope));
+    opBuilder.addInput(minX.asOutput(scope));
+    opBuilder.addInput(maxX.asOutput(scope));
+    opBuilder.addInput(minY.asOutput(scope));
+    opBuilder.addInput(maxY.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("Toutput", Toutput);
     return new QuantizedAdd<V>(opBuilder.build());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/QuantizedMul.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/QuantizedMul.java
index be28c564b8c..89dd9d4fbb0 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/QuantizedMul.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/QuantizedMul.java
@@ -22,12 +22,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.family.TType;
 
 /**
  * Returns x * y element-wise, working on quantized buffers.
@@ -35,7 +35,7 @@
  * @param <V> data type for {@code z()} output
  */
 @Operator(group = "math")
-public final class QuantizedMul<V extends Tensor> extends RawOp {
+public final class QuantizedMul<V extends TType> extends RawOp {
   
   /**
    * Factory method to create a class wrapping a new QuantizedMul operation.
@@ -51,14 +51,14 @@ public final class QuantizedMul<V extends Tensor> extends RawOp {
    * @return a new instance of QuantizedMul
    */
   @Endpoint(describeByClass = true)
-  public static <V extends Tensor, T extends Tensor, U extends Tensor> QuantizedMul<V> create(Scope scope, Operand<T> x, Operand<U> y, Operand<TFloat32> minX, Operand<TFloat32> maxX, Operand<TFloat32> minY, Operand<TFloat32> maxY, DataType<V> Toutput) {
+  public static <V extends TType, T extends TType, U extends TType> QuantizedMul<V> create(Scope scope, Operand<T> x, Operand<U> y, Operand<TFloat32> minX, Operand<TFloat32> maxX, Operand<TFloat32> minY, Operand<TFloat32> maxY, DataType<V> Toutput) {
     OperationBuilder opBuilder = scope.env().opBuilder("QuantizedMul", scope.makeOpName("QuantizedMul"));
-    opBuilder.addInput(x.asOutput());
-    opBuilder.addInput(y.asOutput());
-    opBuilder.addInput(minX.asOutput());
-    opBuilder.addInput(maxX.asOutput());
-    opBuilder.addInput(minY.asOutput());
-    opBuilder.addInput(maxY.asOutput());
+    opBuilder.addInput(x.asOutput(scope));
+    opBuilder.addInput(y.asOutput(scope));
+    opBuilder.addInput(minX.asOutput(scope));
+    opBuilder.addInput(maxX.asOutput(scope));
+    opBuilder.addInput(minY.asOutput(scope));
+    opBuilder.addInput(maxY.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("Toutput", Toutput);
     return new QuantizedMul<V>(opBuilder.build());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Real.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Real.java
index 94feb46b882..651f4c41b83 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Real.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Real.java
@@ -22,13 +22,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TFloat32;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * Returns the real part of a complex number.
@@ -48,7 +48,7 @@
  * @param <U> data type for {@code output()} output
  */
 @Operator(group = "math")
-public final class Real<U extends Tensor & TNumber> extends RawOp implements Operand<U> {
+public final class Real<U extends TNumber> extends RawOp implements Operand<U> {
   
   /**
    * Factory method to create a class wrapping a new Real operation.
@@ -59,9 +59,9 @@ public final class Real<U extends Tensor & TNumber> extends RawOp implements Ope
    * @return a new instance of Real
    */
   @Endpoint(describeByClass = true)
-  public static <U extends Tensor & TNumber, T extends Tensor> Real<U> create(Scope scope, Operand<T> input, DataType<U> Tout) {
+  public static <U extends TNumber, T extends TType> Real<U> create(Scope scope, Operand<T> input, DataType<U> Tout) {
     OperationBuilder opBuilder = scope.env().opBuilder("Real", scope.makeOpName("Real"));
-    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("Tout", Tout);
     return new Real<U>(opBuilder.build());
@@ -75,7 +75,7 @@ public static <U extends Tensor & TNumber, T extends Tensor> Real<U> create(Scop
    * @return a new instance of Real
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> Real<TFloat32> create(Scope scope, Operand<T> input) {
+  public static <T extends TType> Real<TFloat32> create(Scope scope, Operand<T> input) {
     return create(scope, input, TFloat32.DTYPE);
   }
   
@@ -86,7 +86,7 @@ public Output<U> output() {
   }
   
   @Override
-  public Output<U> asOutput() {
+  public Output<U> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/RealDiv.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/RealDiv.java
index e86d43a9c8d..472cbb80e1f 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/RealDiv.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/RealDiv.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Returns x / y element-wise for real types.
@@ -38,7 +38,7 @@
  * @param <T> data type for {@code z()} output
  */
 @Operator(group = "math")
-public final class RealDiv<T extends Tensor> extends RawOp implements Operand<T> {
+public final class RealDiv<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new RealDiv operation.
@@ -49,10 +49,10 @@ public final class RealDiv<T extends Tensor> extends RawOp implements Operand<T>
    * @return a new instance of RealDiv
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> RealDiv<T> create(Scope scope, Operand<T> x, Operand<T> y) {
+  public static <T extends TType> RealDiv<T> create(Scope scope, Operand<T> x, Operand<T> y) {
     OperationBuilder opBuilder = scope.env().opBuilder("RealDiv", scope.makeOpName("RealDiv"));
-    opBuilder.addInput(x.asOutput());
-    opBuilder.addInput(y.asOutput());
+    opBuilder.addInput(x.asOutput(scope));
+    opBuilder.addInput(y.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new RealDiv<T>(opBuilder.build());
   }
@@ -64,7 +64,7 @@ public Output<T> z() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return z;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Reciprocal.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Reciprocal.java
index 81ddad0f029..187fa436621 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Reciprocal.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Reciprocal.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Computes the reciprocal of x element-wise.
@@ -35,7 +35,7 @@
  * @param <T> data type for {@code y()} output
  */
 @Operator(group = "math")
-public final class Reciprocal<T extends Tensor> extends RawOp implements Operand<T> {
+public final class Reciprocal<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Reciprocal operation.
@@ -45,9 +45,9 @@ public final class Reciprocal<T extends Tensor> extends RawOp implements Operand
    * @return a new instance of Reciprocal
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> Reciprocal<T> create(Scope scope, Operand<T> x) {
+  public static <T extends TType> Reciprocal<T> create(Scope scope, Operand<T> x) {
     OperationBuilder opBuilder = scope.env().opBuilder("Reciprocal", scope.makeOpName("Reciprocal"));
-    opBuilder.addInput(x.asOutput());
+    opBuilder.addInput(x.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new Reciprocal<T>(opBuilder.build());
   }
@@ -59,7 +59,7 @@ public Output<T> y() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return y;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/ReciprocalGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/ReciprocalGrad.java
index 09d4b5bc0a3..8ee6127d7cc 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/ReciprocalGrad.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/ReciprocalGrad.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Computes the gradient for the inverse of `x` wrt its input.
@@ -35,7 +35,7 @@
  * 
  * @param <T> data type for {@code z()} output
  */
-public final class ReciprocalGrad<T extends Tensor> extends RawOp implements Operand<T> {
+public final class ReciprocalGrad<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new ReciprocalGrad operation.
@@ -46,10 +46,10 @@ public final class ReciprocalGrad<T extends Tensor> extends RawOp implements Ope
    * @return a new instance of ReciprocalGrad
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> ReciprocalGrad<T> create(Scope scope, Operand<T> y, Operand<T> dy) {
+  public static <T extends TType> ReciprocalGrad<T> create(Scope scope, Operand<T> y, Operand<T> dy) {
     OperationBuilder opBuilder = scope.env().opBuilder("ReciprocalGrad", scope.makeOpName("ReciprocalGrad"));
-    opBuilder.addInput(y.asOutput());
-    opBuilder.addInput(dy.asOutput());
+    opBuilder.addInput(y.asOutput(scope));
+    opBuilder.addInput(dy.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new ReciprocalGrad<T>(opBuilder.build());
   }
@@ -61,7 +61,7 @@ public Output<T> z() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return z;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/RequantizationRangePerChannel.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/RequantizationRangePerChannel.java
index 428045a91c1..5dfaaa2ceab 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/RequantizationRangePerChannel.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/RequantizationRangePerChannel.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.family.TType;
 
 /**
  * Computes requantization range per channel.
@@ -45,11 +45,11 @@ public final class RequantizationRangePerChannel extends RawOp {
    * @return a new instance of RequantizationRangePerChannel
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> RequantizationRangePerChannel create(Scope scope, Operand<T> input, Operand<TFloat32> inputMin, Operand<TFloat32> inputMax, Float clipValueMax) {
+  public static <T extends TType> RequantizationRangePerChannel create(Scope scope, Operand<T> input, Operand<TFloat32> inputMin, Operand<TFloat32> inputMax, Float clipValueMax) {
     OperationBuilder opBuilder = scope.env().opBuilder("RequantizationRangePerChannel", scope.makeOpName("RequantizationRangePerChannel"));
-    opBuilder.addInput(input.asOutput());
-    opBuilder.addInput(inputMin.asOutput());
-    opBuilder.addInput(inputMax.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
+    opBuilder.addInput(inputMin.asOutput(scope));
+    opBuilder.addInput(inputMax.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("clip_value_max", clipValueMax);
     return new RequantizationRangePerChannel(opBuilder.build());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/RequantizePerChannel.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/RequantizePerChannel.java
index f86e7d45a37..bc1d250245c 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/RequantizePerChannel.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/RequantizePerChannel.java
@@ -22,19 +22,19 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.family.TType;
 
 /**
  * Requantizes input with min and max values known per channel.
  * 
  * @param <U> data type for {@code output()} output
  */
-public final class RequantizePerChannel<U extends Tensor> extends RawOp {
+public final class RequantizePerChannel<U extends TType> extends RawOp {
   
   /**
    * Factory method to create a class wrapping a new RequantizePerChannel operation.
@@ -49,13 +49,13 @@ public final class RequantizePerChannel<U extends Tensor> extends RawOp {
    * @return a new instance of RequantizePerChannel
    */
   @Endpoint(describeByClass = true)
-  public static <U extends Tensor, T extends Tensor> RequantizePerChannel<U> create(Scope scope, Operand<T> input, Operand<TFloat32> inputMin, Operand<TFloat32> inputMax, Operand<TFloat32> requestedOutputMin, Operand<TFloat32> requestedOutputMax, DataType<U> outType) {
+  public static <U extends TType, T extends TType> RequantizePerChannel<U> create(Scope scope, Operand<T> input, Operand<TFloat32> inputMin, Operand<TFloat32> inputMax, Operand<TFloat32> requestedOutputMin, Operand<TFloat32> requestedOutputMax, DataType<U> outType) {
     OperationBuilder opBuilder = scope.env().opBuilder("RequantizePerChannel", scope.makeOpName("RequantizePerChannel"));
-    opBuilder.addInput(input.asOutput());
-    opBuilder.addInput(inputMin.asOutput());
-    opBuilder.addInput(inputMax.asOutput());
-    opBuilder.addInput(requestedOutputMin.asOutput());
-    opBuilder.addInput(requestedOutputMax.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
+    opBuilder.addInput(inputMin.asOutput(scope));
+    opBuilder.addInput(inputMax.asOutput(scope));
+    opBuilder.addInput(requestedOutputMin.asOutput(scope));
+    opBuilder.addInput(requestedOutputMax.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("out_type", outType);
     return new RequantizePerChannel<U>(opBuilder.build());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Rint.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Rint.java
index 1aaa11b178d..5ebd67d43ae 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Rint.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Rint.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -44,7 +43,7 @@
  * @param <T> data type for {@code y()} output
  */
 @Operator(group = "math")
-public final class Rint<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class Rint<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Rint operation.
@@ -54,9 +53,9 @@ public final class Rint<T extends Tensor & TNumber> extends RawOp implements Ope
    * @return a new instance of Rint
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> Rint<T> create(Scope scope, Operand<T> x) {
+  public static <T extends TNumber> Rint<T> create(Scope scope, Operand<T> x) {
     OperationBuilder opBuilder = scope.env().opBuilder("Rint", scope.makeOpName("Rint"));
-    opBuilder.addInput(x.asOutput());
+    opBuilder.addInput(x.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new Rint<T>(opBuilder.build());
   }
@@ -68,7 +67,7 @@ public Output<T> y() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return y;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Round.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Round.java
index b43bba763f0..ea8e3cfb878 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Round.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Round.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Rounds the values of a tensor to the nearest integer, element-wise.
@@ -36,7 +36,7 @@
  * @param <T> data type for {@code y()} output
  */
 @Operator(group = "math")
-public final class Round<T extends Tensor> extends RawOp implements Operand<T> {
+public final class Round<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Round operation.
@@ -46,9 +46,9 @@ public final class Round<T extends Tensor> extends RawOp implements Operand<T> {
    * @return a new instance of Round
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> Round<T> create(Scope scope, Operand<T> x) {
+  public static <T extends TType> Round<T> create(Scope scope, Operand<T> x) {
     OperationBuilder opBuilder = scope.env().opBuilder("Round", scope.makeOpName("Round"));
-    opBuilder.addInput(x.asOutput());
+    opBuilder.addInput(x.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new Round<T>(opBuilder.build());
   }
@@ -60,7 +60,7 @@ public Output<T> y() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return y;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Rsqrt.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Rsqrt.java
index ebe110213c4..198e2ac0f9e 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Rsqrt.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Rsqrt.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Computes reciprocal of square root of x element-wise.
@@ -35,7 +35,7 @@
  * @param <T> data type for {@code y()} output
  */
 @Operator(group = "math")
-public final class Rsqrt<T extends Tensor> extends RawOp implements Operand<T> {
+public final class Rsqrt<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Rsqrt operation.
@@ -45,9 +45,9 @@ public final class Rsqrt<T extends Tensor> extends RawOp implements Operand<T> {
    * @return a new instance of Rsqrt
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> Rsqrt<T> create(Scope scope, Operand<T> x) {
+  public static <T extends TType> Rsqrt<T> create(Scope scope, Operand<T> x) {
     OperationBuilder opBuilder = scope.env().opBuilder("Rsqrt", scope.makeOpName("Rsqrt"));
-    opBuilder.addInput(x.asOutput());
+    opBuilder.addInput(x.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new Rsqrt<T>(opBuilder.build());
   }
@@ -59,7 +59,7 @@ public Output<T> y() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return y;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/RsqrtGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/RsqrtGrad.java
index 58f228a5c24..c7fd2ca1502 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/RsqrtGrad.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/RsqrtGrad.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Computes the gradient for the rsqrt of `x` wrt its input.
@@ -35,7 +35,7 @@
  * 
  * @param <T> data type for {@code z()} output
  */
-public final class RsqrtGrad<T extends Tensor> extends RawOp implements Operand<T> {
+public final class RsqrtGrad<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new RsqrtGrad operation.
@@ -46,10 +46,10 @@ public final class RsqrtGrad<T extends Tensor> extends RawOp implements Operand<
    * @return a new instance of RsqrtGrad
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> RsqrtGrad<T> create(Scope scope, Operand<T> y, Operand<T> dy) {
+  public static <T extends TType> RsqrtGrad<T> create(Scope scope, Operand<T> y, Operand<T> dy) {
     OperationBuilder opBuilder = scope.env().opBuilder("RsqrtGrad", scope.makeOpName("RsqrtGrad"));
-    opBuilder.addInput(y.asOutput());
-    opBuilder.addInput(dy.asOutput());
+    opBuilder.addInput(y.asOutput(scope));
+    opBuilder.addInput(dy.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new RsqrtGrad<T>(opBuilder.build());
   }
@@ -61,7 +61,7 @@ public Output<T> z() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return z;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SegmentMax.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SegmentMax.java
index 80af885b628..ff02ffa3bbc 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SegmentMax.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SegmentMax.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -57,7 +56,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "math")
-public final class SegmentMax<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class SegmentMax<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new SegmentMax operation.
@@ -69,10 +68,10 @@ public final class SegmentMax<T extends Tensor & TNumber> extends RawOp implemen
    * @return a new instance of SegmentMax
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber, U extends Tensor & TNumber> SegmentMax<T> create(Scope scope, Operand<T> data, Operand<U> segmentIds) {
+  public static <T extends TNumber, U extends TNumber> SegmentMax<T> create(Scope scope, Operand<T> data, Operand<U> segmentIds) {
     OperationBuilder opBuilder = scope.env().opBuilder("SegmentMax", scope.makeOpName("SegmentMax"));
-    opBuilder.addInput(data.asOutput());
-    opBuilder.addInput(segmentIds.asOutput());
+    opBuilder.addInput(data.asOutput(scope));
+    opBuilder.addInput(segmentIds.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new SegmentMax<T>(opBuilder.build());
   }
@@ -86,7 +85,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SegmentMean.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SegmentMean.java
index 46359a58ac9..cdd1acc7fa3 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SegmentMean.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SegmentMean.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * Computes the mean along segments of a tensor.
@@ -58,7 +58,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "math")
-public final class SegmentMean<T extends Tensor> extends RawOp implements Operand<T> {
+public final class SegmentMean<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new SegmentMean operation.
@@ -70,10 +70,10 @@ public final class SegmentMean<T extends Tensor> extends RawOp implements Operan
    * @return a new instance of SegmentMean
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor, U extends Tensor & TNumber> SegmentMean<T> create(Scope scope, Operand<T> data, Operand<U> segmentIds) {
+  public static <T extends TType, U extends TNumber> SegmentMean<T> create(Scope scope, Operand<T> data, Operand<U> segmentIds) {
     OperationBuilder opBuilder = scope.env().opBuilder("SegmentMean", scope.makeOpName("SegmentMean"));
-    opBuilder.addInput(data.asOutput());
-    opBuilder.addInput(segmentIds.asOutput());
+    opBuilder.addInput(data.asOutput(scope));
+    opBuilder.addInput(segmentIds.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new SegmentMean<T>(opBuilder.build());
   }
@@ -87,7 +87,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SegmentMin.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SegmentMin.java
index 05ea7605bc8..e40683c8aa3 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SegmentMin.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SegmentMin.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -57,7 +56,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "math")
-public final class SegmentMin<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class SegmentMin<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new SegmentMin operation.
@@ -69,10 +68,10 @@ public final class SegmentMin<T extends Tensor & TNumber> extends RawOp implemen
    * @return a new instance of SegmentMin
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber, U extends Tensor & TNumber> SegmentMin<T> create(Scope scope, Operand<T> data, Operand<U> segmentIds) {
+  public static <T extends TNumber, U extends TNumber> SegmentMin<T> create(Scope scope, Operand<T> data, Operand<U> segmentIds) {
     OperationBuilder opBuilder = scope.env().opBuilder("SegmentMin", scope.makeOpName("SegmentMin"));
-    opBuilder.addInput(data.asOutput());
-    opBuilder.addInput(segmentIds.asOutput());
+    opBuilder.addInput(data.asOutput(scope));
+    opBuilder.addInput(segmentIds.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new SegmentMin<T>(opBuilder.build());
   }
@@ -86,7 +85,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SegmentProd.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SegmentProd.java
index 0fa63ed7bb4..b5757ecb249 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SegmentProd.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SegmentProd.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * Computes the product along segments of a tensor.
@@ -57,7 +57,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "math")
-public final class SegmentProd<T extends Tensor> extends RawOp implements Operand<T> {
+public final class SegmentProd<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new SegmentProd operation.
@@ -69,10 +69,10 @@ public final class SegmentProd<T extends Tensor> extends RawOp implements Operan
    * @return a new instance of SegmentProd
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor, U extends Tensor & TNumber> SegmentProd<T> create(Scope scope, Operand<T> data, Operand<U> segmentIds) {
+  public static <T extends TType, U extends TNumber> SegmentProd<T> create(Scope scope, Operand<T> data, Operand<U> segmentIds) {
     OperationBuilder opBuilder = scope.env().opBuilder("SegmentProd", scope.makeOpName("SegmentProd"));
-    opBuilder.addInput(data.asOutput());
-    opBuilder.addInput(segmentIds.asOutput());
+    opBuilder.addInput(data.asOutput(scope));
+    opBuilder.addInput(segmentIds.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new SegmentProd<T>(opBuilder.build());
   }
@@ -86,7 +86,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SegmentSum.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SegmentSum.java
index fe5818cd263..1bda10f8878 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SegmentSum.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SegmentSum.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * Computes the sum along segments of a tensor.
@@ -57,7 +57,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "math")
-public final class SegmentSum<T extends Tensor> extends RawOp implements Operand<T> {
+public final class SegmentSum<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new SegmentSum operation.
@@ -69,10 +69,10 @@ public final class SegmentSum<T extends Tensor> extends RawOp implements Operand
    * @return a new instance of SegmentSum
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor, U extends Tensor & TNumber> SegmentSum<T> create(Scope scope, Operand<T> data, Operand<U> segmentIds) {
+  public static <T extends TType, U extends TNumber> SegmentSum<T> create(Scope scope, Operand<T> data, Operand<U> segmentIds) {
     OperationBuilder opBuilder = scope.env().opBuilder("SegmentSum", scope.makeOpName("SegmentSum"));
-    opBuilder.addInput(data.asOutput());
-    opBuilder.addInput(segmentIds.asOutput());
+    opBuilder.addInput(data.asOutput(scope));
+    opBuilder.addInput(segmentIds.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new SegmentSum<T>(opBuilder.build());
   }
@@ -86,7 +86,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Sigmoid.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Sigmoid.java
index 5ae3e66c143..f8a06245fe5 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Sigmoid.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Sigmoid.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Computes sigmoid of `x` element-wise.
@@ -35,7 +35,7 @@
  * @param <T> data type for {@code y()} output
  */
 @Operator(group = "math")
-public final class Sigmoid<T extends Tensor> extends RawOp implements Operand<T> {
+public final class Sigmoid<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Sigmoid operation.
@@ -45,9 +45,9 @@ public final class Sigmoid<T extends Tensor> extends RawOp implements Operand<T>
    * @return a new instance of Sigmoid
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> Sigmoid<T> create(Scope scope, Operand<T> x) {
+  public static <T extends TType> Sigmoid<T> create(Scope scope, Operand<T> x) {
     OperationBuilder opBuilder = scope.env().opBuilder("Sigmoid", scope.makeOpName("Sigmoid"));
-    opBuilder.addInput(x.asOutput());
+    opBuilder.addInput(x.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new Sigmoid<T>(opBuilder.build());
   }
@@ -59,7 +59,7 @@ public Output<T> y() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return y;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SigmoidGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SigmoidGrad.java
index b0a39e7d20f..ea0613e0f1c 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SigmoidGrad.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SigmoidGrad.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Computes the gradient of the sigmoid of `x` wrt its input.
@@ -35,7 +35,7 @@
  * 
  * @param <T> data type for {@code z()} output
  */
-public final class SigmoidGrad<T extends Tensor> extends RawOp implements Operand<T> {
+public final class SigmoidGrad<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new SigmoidGrad operation.
@@ -46,10 +46,10 @@ public final class SigmoidGrad<T extends Tensor> extends RawOp implements Operan
    * @return a new instance of SigmoidGrad
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> SigmoidGrad<T> create(Scope scope, Operand<T> y, Operand<T> dy) {
+  public static <T extends TType> SigmoidGrad<T> create(Scope scope, Operand<T> y, Operand<T> dy) {
     OperationBuilder opBuilder = scope.env().opBuilder("SigmoidGrad", scope.makeOpName("SigmoidGrad"));
-    opBuilder.addInput(y.asOutput());
-    opBuilder.addInput(dy.asOutput());
+    opBuilder.addInput(y.asOutput(scope));
+    opBuilder.addInput(dy.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new SigmoidGrad<T>(opBuilder.build());
   }
@@ -61,7 +61,7 @@ public Output<T> z() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return z;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Sign.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Sign.java
index 90bcacd79e1..7144fe432a9 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Sign.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Sign.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Returns an element-wise indication of the sign of a number.
@@ -41,7 +41,7 @@
  * @param <T> data type for {@code y()} output
  */
 @Operator(group = "math")
-public final class Sign<T extends Tensor> extends RawOp implements Operand<T> {
+public final class Sign<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Sign operation.
@@ -51,9 +51,9 @@ public final class Sign<T extends Tensor> extends RawOp implements Operand<T> {
    * @return a new instance of Sign
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> Sign<T> create(Scope scope, Operand<T> x) {
+  public static <T extends TType> Sign<T> create(Scope scope, Operand<T> x) {
     OperationBuilder opBuilder = scope.env().opBuilder("Sign", scope.makeOpName("Sign"));
-    opBuilder.addInput(x.asOutput());
+    opBuilder.addInput(x.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new Sign<T>(opBuilder.build());
   }
@@ -65,7 +65,7 @@ public Output<T> y() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return y;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Sin.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Sin.java
index f7d6e535e3c..1a60f3d1df0 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Sin.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Sin.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Computes sine of x element-wise.
@@ -43,7 +43,7 @@
  * @param <T> data type for {@code y()} output
  */
 @Operator(group = "math")
-public final class Sin<T extends Tensor> extends RawOp implements Operand<T> {
+public final class Sin<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Sin operation.
@@ -53,9 +53,9 @@ public final class Sin<T extends Tensor> extends RawOp implements Operand<T> {
    * @return a new instance of Sin
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> Sin<T> create(Scope scope, Operand<T> x) {
+  public static <T extends TType> Sin<T> create(Scope scope, Operand<T> x) {
     OperationBuilder opBuilder = scope.env().opBuilder("Sin", scope.makeOpName("Sin"));
-    opBuilder.addInput(x.asOutput());
+    opBuilder.addInput(x.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new Sin<T>(opBuilder.build());
   }
@@ -67,7 +67,7 @@ public Output<T> y() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return y;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Sinh.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Sinh.java
index 435d05838c9..f82339505f3 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Sinh.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Sinh.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Computes hyperbolic sine of x element-wise.
@@ -43,7 +43,7 @@
  * @param <T> data type for {@code y()} output
  */
 @Operator(group = "math")
-public final class Sinh<T extends Tensor> extends RawOp implements Operand<T> {
+public final class Sinh<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Sinh operation.
@@ -53,9 +53,9 @@ public final class Sinh<T extends Tensor> extends RawOp implements Operand<T> {
    * @return a new instance of Sinh
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> Sinh<T> create(Scope scope, Operand<T> x) {
+  public static <T extends TType> Sinh<T> create(Scope scope, Operand<T> x) {
     OperationBuilder opBuilder = scope.env().opBuilder("Sinh", scope.makeOpName("Sinh"));
-    opBuilder.addInput(x.asOutput());
+    opBuilder.addInput(x.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new Sinh<T>(opBuilder.build());
   }
@@ -67,7 +67,7 @@ public Output<T> y() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return y;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SobolSample.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SobolSample.java
index b12f674ea18..bda81e6b304 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SobolSample.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SobolSample.java
@@ -22,7 +22,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -39,7 +38,7 @@
  * 
  * @param <T> data type for {@code samples()} output
  */
-public final class SobolSample<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class SobolSample<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new SobolSample operation.
@@ -54,11 +53,11 @@ public final class SobolSample<T extends Tensor & TNumber> extends RawOp impleme
    * @return a new instance of SobolSample
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> SobolSample<T> create(Scope scope, Operand<TInt32> dim, Operand<TInt32> numResults, Operand<TInt32> skip, DataType<T> dtype) {
+  public static <T extends TNumber> SobolSample<T> create(Scope scope, Operand<TInt32> dim, Operand<TInt32> numResults, Operand<TInt32> skip, DataType<T> dtype) {
     OperationBuilder opBuilder = scope.env().opBuilder("SobolSample", scope.makeOpName("SobolSample"));
-    opBuilder.addInput(dim.asOutput());
-    opBuilder.addInput(numResults.asOutput());
-    opBuilder.addInput(skip.asOutput());
+    opBuilder.addInput(dim.asOutput(scope));
+    opBuilder.addInput(numResults.asOutput(scope));
+    opBuilder.addInput(skip.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("dtype", dtype);
     return new SobolSample<T>(opBuilder.build());
@@ -88,7 +87,7 @@ public Output<T> samples() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return samples;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Softplus.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Softplus.java
index 34a5370db29..1a07ecfb747 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Softplus.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Softplus.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -34,7 +33,7 @@
  * @param <T> data type for {@code activations()} output
  */
 @Operator(group = "math")
-public final class Softplus<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class Softplus<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Softplus operation.
@@ -44,9 +43,9 @@ public final class Softplus<T extends Tensor & TNumber> extends RawOp implements
    * @return a new instance of Softplus
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> Softplus<T> create(Scope scope, Operand<T> features) {
+  public static <T extends TNumber> Softplus<T> create(Scope scope, Operand<T> features) {
     OperationBuilder opBuilder = scope.env().opBuilder("Softplus", scope.makeOpName("Softplus"));
-    opBuilder.addInput(features.asOutput());
+    opBuilder.addInput(features.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new Softplus<T>(opBuilder.build());
   }
@@ -58,7 +57,7 @@ public Output<T> activations() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return activations;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SoftplusGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SoftplusGrad.java
index 05463dcc468..9bc0aee229f 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SoftplusGrad.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SoftplusGrad.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -33,7 +32,7 @@
  * 
  * @param <T> data type for {@code backprops()} output
  */
-public final class SoftplusGrad<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class SoftplusGrad<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new SoftplusGrad operation.
@@ -44,10 +43,10 @@ public final class SoftplusGrad<T extends Tensor & TNumber> extends RawOp implem
    * @return a new instance of SoftplusGrad
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> SoftplusGrad<T> create(Scope scope, Operand<T> gradients, Operand<T> features) {
+  public static <T extends TNumber> SoftplusGrad<T> create(Scope scope, Operand<T> gradients, Operand<T> features) {
     OperationBuilder opBuilder = scope.env().opBuilder("SoftplusGrad", scope.makeOpName("SoftplusGrad"));
-    opBuilder.addInput(gradients.asOutput());
-    opBuilder.addInput(features.asOutput());
+    opBuilder.addInput(gradients.asOutput(scope));
+    opBuilder.addInput(features.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new SoftplusGrad<T>(opBuilder.build());
   }
@@ -60,7 +59,7 @@ public Output<T> backprops() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return backprops;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Sqrt.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Sqrt.java
index d9d843713f4..dad9c1d6cec 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Sqrt.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Sqrt.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Computes square root of x element-wise.
@@ -35,7 +35,7 @@
  * @param <T> data type for {@code y()} output
  */
 @Operator(group = "math")
-public final class Sqrt<T extends Tensor> extends RawOp implements Operand<T> {
+public final class Sqrt<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Sqrt operation.
@@ -45,9 +45,9 @@ public final class Sqrt<T extends Tensor> extends RawOp implements Operand<T> {
    * @return a new instance of Sqrt
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> Sqrt<T> create(Scope scope, Operand<T> x) {
+  public static <T extends TType> Sqrt<T> create(Scope scope, Operand<T> x) {
     OperationBuilder opBuilder = scope.env().opBuilder("Sqrt", scope.makeOpName("Sqrt"));
-    opBuilder.addInput(x.asOutput());
+    opBuilder.addInput(x.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new Sqrt<T>(opBuilder.build());
   }
@@ -59,7 +59,7 @@ public Output<T> y() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return y;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SqrtGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SqrtGrad.java
index 3039153a9b1..0206518124c 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SqrtGrad.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SqrtGrad.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Computes the gradient for the sqrt of `x` wrt its input.
@@ -35,7 +35,7 @@
  * 
  * @param <T> data type for {@code z()} output
  */
-public final class SqrtGrad<T extends Tensor> extends RawOp implements Operand<T> {
+public final class SqrtGrad<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new SqrtGrad operation.
@@ -46,10 +46,10 @@ public final class SqrtGrad<T extends Tensor> extends RawOp implements Operand<T
    * @return a new instance of SqrtGrad
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> SqrtGrad<T> create(Scope scope, Operand<T> y, Operand<T> dy) {
+  public static <T extends TType> SqrtGrad<T> create(Scope scope, Operand<T> y, Operand<T> dy) {
     OperationBuilder opBuilder = scope.env().opBuilder("SqrtGrad", scope.makeOpName("SqrtGrad"));
-    opBuilder.addInput(y.asOutput());
-    opBuilder.addInput(dy.asOutput());
+    opBuilder.addInput(y.asOutput(scope));
+    opBuilder.addInput(dy.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new SqrtGrad<T>(opBuilder.build());
   }
@@ -61,7 +61,7 @@ public Output<T> z() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return z;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Square.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Square.java
index e970833974b..896cf528ca3 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Square.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Square.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Computes square of x element-wise.
@@ -35,7 +35,7 @@
  * @param <T> data type for {@code y()} output
  */
 @Operator(group = "math")
-public final class Square<T extends Tensor> extends RawOp implements Operand<T> {
+public final class Square<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Square operation.
@@ -45,9 +45,9 @@ public final class Square<T extends Tensor> extends RawOp implements Operand<T>
    * @return a new instance of Square
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> Square<T> create(Scope scope, Operand<T> x) {
+  public static <T extends TType> Square<T> create(Scope scope, Operand<T> x) {
     OperationBuilder opBuilder = scope.env().opBuilder("Square", scope.makeOpName("Square"));
-    opBuilder.addInput(x.asOutput());
+    opBuilder.addInput(x.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new Square<T>(opBuilder.build());
   }
@@ -59,7 +59,7 @@ public Output<T> y() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return y;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SquaredDifference.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SquaredDifference.java
index 3968a668e23..b0cc11688d7 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SquaredDifference.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SquaredDifference.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Returns (x - y)(x - y) element-wise.
@@ -36,7 +36,7 @@
  * @param <T> data type for {@code z()} output
  */
 @Operator(group = "math")
-public final class SquaredDifference<T extends Tensor> extends RawOp implements Operand<T> {
+public final class SquaredDifference<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new SquaredDifference operation.
@@ -47,10 +47,10 @@ public final class SquaredDifference<T extends Tensor> extends RawOp implements
    * @return a new instance of SquaredDifference
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> SquaredDifference<T> create(Scope scope, Operand<T> x, Operand<T> y) {
+  public static <T extends TType> SquaredDifference<T> create(Scope scope, Operand<T> x, Operand<T> y) {
     OperationBuilder opBuilder = scope.env().opBuilder("SquaredDifference", scope.makeOpName("SquaredDifference"));
-    opBuilder.addInput(x.asOutput());
-    opBuilder.addInput(y.asOutput());
+    opBuilder.addInput(x.asOutput(scope));
+    opBuilder.addInput(y.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new SquaredDifference<T>(opBuilder.build());
   }
@@ -62,7 +62,7 @@ public Output<T> z() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return z;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Sub.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Sub.java
index c5d9132bb7b..cff65b6c38b 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Sub.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Sub.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Returns x - y element-wise.
@@ -36,7 +36,7 @@
  * @param <T> data type for {@code z()} output
  */
 @Operator(group = "math")
-public final class Sub<T extends Tensor> extends RawOp implements Operand<T> {
+public final class Sub<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Sub operation.
@@ -47,10 +47,10 @@ public final class Sub<T extends Tensor> extends RawOp implements Operand<T> {
    * @return a new instance of Sub
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> Sub<T> create(Scope scope, Operand<T> x, Operand<T> y) {
+  public static <T extends TType> Sub<T> create(Scope scope, Operand<T> x, Operand<T> y) {
     OperationBuilder opBuilder = scope.env().opBuilder("Sub", scope.makeOpName("Sub"));
-    opBuilder.addInput(x.asOutput());
-    opBuilder.addInput(y.asOutput());
+    opBuilder.addInput(x.asOutput(scope));
+    opBuilder.addInput(y.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new Sub<T>(opBuilder.build());
   }
@@ -62,7 +62,7 @@ public Output<T> z() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return z;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Tan.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Tan.java
index a0418328c0b..07e2f9ddf06 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Tan.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Tan.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Computes tan of x element-wise.
@@ -44,7 +44,7 @@
  * @param <T> data type for {@code y()} output
  */
 @Operator(group = "math")
-public final class Tan<T extends Tensor> extends RawOp implements Operand<T> {
+public final class Tan<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Tan operation.
@@ -54,9 +54,9 @@ public final class Tan<T extends Tensor> extends RawOp implements Operand<T> {
    * @return a new instance of Tan
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> Tan<T> create(Scope scope, Operand<T> x) {
+  public static <T extends TType> Tan<T> create(Scope scope, Operand<T> x) {
     OperationBuilder opBuilder = scope.env().opBuilder("Tan", scope.makeOpName("Tan"));
-    opBuilder.addInput(x.asOutput());
+    opBuilder.addInput(x.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new Tan<T>(opBuilder.build());
   }
@@ -68,7 +68,7 @@ public Output<T> y() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return y;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Tanh.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Tanh.java
index d8aed91518c..66ae2614338 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Tanh.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Tanh.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Computes hyperbolic tangent of `x` element-wise.
@@ -43,7 +43,7 @@
  * @param <T> data type for {@code y()} output
  */
 @Operator(group = "math")
-public final class Tanh<T extends Tensor> extends RawOp implements Operand<T> {
+public final class Tanh<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Tanh operation.
@@ -53,9 +53,9 @@ public final class Tanh<T extends Tensor> extends RawOp implements Operand<T> {
    * @return a new instance of Tanh
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> Tanh<T> create(Scope scope, Operand<T> x) {
+  public static <T extends TType> Tanh<T> create(Scope scope, Operand<T> x) {
     OperationBuilder opBuilder = scope.env().opBuilder("Tanh", scope.makeOpName("Tanh"));
-    opBuilder.addInput(x.asOutput());
+    opBuilder.addInput(x.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new Tanh<T>(opBuilder.build());
   }
@@ -67,7 +67,7 @@ public Output<T> y() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return y;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/TanhGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/TanhGrad.java
index 5ab68795017..4e8b54ddba8 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/TanhGrad.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/TanhGrad.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Computes the gradient for the tanh of `x` wrt its input.
@@ -35,7 +35,7 @@
  * 
  * @param <T> data type for {@code z()} output
  */
-public final class TanhGrad<T extends Tensor> extends RawOp implements Operand<T> {
+public final class TanhGrad<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new TanhGrad operation.
@@ -46,10 +46,10 @@ public final class TanhGrad<T extends Tensor> extends RawOp implements Operand<T
    * @return a new instance of TanhGrad
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> TanhGrad<T> create(Scope scope, Operand<T> y, Operand<T> dy) {
+  public static <T extends TType> TanhGrad<T> create(Scope scope, Operand<T> y, Operand<T> dy) {
     OperationBuilder opBuilder = scope.env().opBuilder("TanhGrad", scope.makeOpName("TanhGrad"));
-    opBuilder.addInput(y.asOutput());
-    opBuilder.addInput(dy.asOutput());
+    opBuilder.addInput(y.asOutput(scope));
+    opBuilder.addInput(dy.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new TanhGrad<T>(opBuilder.build());
   }
@@ -61,7 +61,7 @@ public Output<T> z() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return z;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/TruncateDiv.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/TruncateDiv.java
index 3463015f92e..8ee52866c66 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/TruncateDiv.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/TruncateDiv.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Returns x / y element-wise for integer types.
@@ -41,7 +41,7 @@
  * @param <T> data type for {@code z()} output
  */
 @Operator(group = "math")
-public final class TruncateDiv<T extends Tensor> extends RawOp implements Operand<T> {
+public final class TruncateDiv<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new TruncateDiv operation.
@@ -52,10 +52,10 @@ public final class TruncateDiv<T extends Tensor> extends RawOp implements Operan
    * @return a new instance of TruncateDiv
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> TruncateDiv<T> create(Scope scope, Operand<T> x, Operand<T> y) {
+  public static <T extends TType> TruncateDiv<T> create(Scope scope, Operand<T> x, Operand<T> y) {
     OperationBuilder opBuilder = scope.env().opBuilder("TruncateDiv", scope.makeOpName("TruncateDiv"));
-    opBuilder.addInput(x.asOutput());
-    opBuilder.addInput(y.asOutput());
+    opBuilder.addInput(x.asOutput(scope));
+    opBuilder.addInput(y.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new TruncateDiv<T>(opBuilder.build());
   }
@@ -67,7 +67,7 @@ public Output<T> z() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return z;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/TruncateMod.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/TruncateMod.java
index 12bce627fa6..8ff26efff5b 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/TruncateMod.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/TruncateMod.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -40,7 +39,7 @@
  * @param <T> data type for {@code z()} output
  */
 @Operator(group = "math")
-public final class TruncateMod<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class TruncateMod<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new TruncateMod operation.
@@ -51,10 +50,10 @@ public final class TruncateMod<T extends Tensor & TNumber> extends RawOp impleme
    * @return a new instance of TruncateMod
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> TruncateMod<T> create(Scope scope, Operand<T> x, Operand<T> y) {
+  public static <T extends TNumber> TruncateMod<T> create(Scope scope, Operand<T> x, Operand<T> y) {
     OperationBuilder opBuilder = scope.env().opBuilder("TruncateMod", scope.makeOpName("TruncateMod"));
-    opBuilder.addInput(x.asOutput());
-    opBuilder.addInput(y.asOutput());
+    opBuilder.addInput(x.asOutput(scope));
+    opBuilder.addInput(y.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new TruncateMod<T>(opBuilder.build());
   }
@@ -66,7 +65,7 @@ public Output<T> z() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return z;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/UnsortedSegmentMax.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/UnsortedSegmentMax.java
index da8d6a1a2a1..bdf89cd2ad8 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/UnsortedSegmentMax.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/UnsortedSegmentMax.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -65,7 +64,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "math")
-public final class UnsortedSegmentMax<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class UnsortedSegmentMax<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new UnsortedSegmentMax operation.
@@ -77,11 +76,11 @@ public final class UnsortedSegmentMax<T extends Tensor & TNumber> extends RawOp
    * @return a new instance of UnsortedSegmentMax
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber, U extends Tensor & TNumber, V extends Tensor & TNumber> UnsortedSegmentMax<T> create(Scope scope, Operand<T> data, Operand<U> segmentIds, Operand<V> numSegments) {
+  public static <T extends TNumber, U extends TNumber, V extends TNumber> UnsortedSegmentMax<T> create(Scope scope, Operand<T> data, Operand<U> segmentIds, Operand<V> numSegments) {
     OperationBuilder opBuilder = scope.env().opBuilder("UnsortedSegmentMax", scope.makeOpName("UnsortedSegmentMax"));
-    opBuilder.addInput(data.asOutput());
-    opBuilder.addInput(segmentIds.asOutput());
-    opBuilder.addInput(numSegments.asOutput());
+    opBuilder.addInput(data.asOutput(scope));
+    opBuilder.addInput(segmentIds.asOutput(scope));
+    opBuilder.addInput(numSegments.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new UnsortedSegmentMax<T>(opBuilder.build());
   }
@@ -96,7 +95,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/UnsortedSegmentMin.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/UnsortedSegmentMin.java
index 5fcc1caa4fc..5c237ccadac 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/UnsortedSegmentMin.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/UnsortedSegmentMin.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -59,7 +58,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "math")
-public final class UnsortedSegmentMin<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class UnsortedSegmentMin<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new UnsortedSegmentMin operation.
@@ -71,11 +70,11 @@ public final class UnsortedSegmentMin<T extends Tensor & TNumber> extends RawOp
    * @return a new instance of UnsortedSegmentMin
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber, U extends Tensor & TNumber, V extends Tensor & TNumber> UnsortedSegmentMin<T> create(Scope scope, Operand<T> data, Operand<U> segmentIds, Operand<V> numSegments) {
+  public static <T extends TNumber, U extends TNumber, V extends TNumber> UnsortedSegmentMin<T> create(Scope scope, Operand<T> data, Operand<U> segmentIds, Operand<V> numSegments) {
     OperationBuilder opBuilder = scope.env().opBuilder("UnsortedSegmentMin", scope.makeOpName("UnsortedSegmentMin"));
-    opBuilder.addInput(data.asOutput());
-    opBuilder.addInput(segmentIds.asOutput());
-    opBuilder.addInput(numSegments.asOutput());
+    opBuilder.addInput(data.asOutput(scope));
+    opBuilder.addInput(segmentIds.asOutput(scope));
+    opBuilder.addInput(numSegments.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new UnsortedSegmentMin<T>(opBuilder.build());
   }
@@ -90,7 +89,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/UnsortedSegmentProd.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/UnsortedSegmentProd.java
index e3eb374edc0..d2eefdc418e 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/UnsortedSegmentProd.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/UnsortedSegmentProd.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * Computes the product along segments of a tensor.
@@ -58,7 +58,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "math")
-public final class UnsortedSegmentProd<T extends Tensor> extends RawOp implements Operand<T> {
+public final class UnsortedSegmentProd<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new UnsortedSegmentProd operation.
@@ -70,11 +70,11 @@ public final class UnsortedSegmentProd<T extends Tensor> extends RawOp implement
    * @return a new instance of UnsortedSegmentProd
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor, U extends Tensor & TNumber, V extends Tensor & TNumber> UnsortedSegmentProd<T> create(Scope scope, Operand<T> data, Operand<U> segmentIds, Operand<V> numSegments) {
+  public static <T extends TType, U extends TNumber, V extends TNumber> UnsortedSegmentProd<T> create(Scope scope, Operand<T> data, Operand<U> segmentIds, Operand<V> numSegments) {
     OperationBuilder opBuilder = scope.env().opBuilder("UnsortedSegmentProd", scope.makeOpName("UnsortedSegmentProd"));
-    opBuilder.addInput(data.asOutput());
-    opBuilder.addInput(segmentIds.asOutput());
-    opBuilder.addInput(numSegments.asOutput());
+    opBuilder.addInput(data.asOutput(scope));
+    opBuilder.addInput(segmentIds.asOutput(scope));
+    opBuilder.addInput(numSegments.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new UnsortedSegmentProd<T>(opBuilder.build());
   }
@@ -89,7 +89,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/UnsortedSegmentSum.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/UnsortedSegmentSum.java
index 957b9409413..7108e10e216 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/UnsortedSegmentSum.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/UnsortedSegmentSum.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * Computes the sum along segments of a tensor.
@@ -61,7 +61,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "math")
-public final class UnsortedSegmentSum<T extends Tensor> extends RawOp implements Operand<T> {
+public final class UnsortedSegmentSum<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new UnsortedSegmentSum operation.
@@ -73,11 +73,11 @@ public final class UnsortedSegmentSum<T extends Tensor> extends RawOp implements
    * @return a new instance of UnsortedSegmentSum
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor, U extends Tensor & TNumber, V extends Tensor & TNumber> UnsortedSegmentSum<T> create(Scope scope, Operand<T> data, Operand<U> segmentIds, Operand<V> numSegments) {
+  public static <T extends TType, U extends TNumber, V extends TNumber> UnsortedSegmentSum<T> create(Scope scope, Operand<T> data, Operand<U> segmentIds, Operand<V> numSegments) {
     OperationBuilder opBuilder = scope.env().opBuilder("UnsortedSegmentSum", scope.makeOpName("UnsortedSegmentSum"));
-    opBuilder.addInput(data.asOutput());
-    opBuilder.addInput(segmentIds.asOutput());
-    opBuilder.addInput(numSegments.asOutput());
+    opBuilder.addInput(data.asOutput(scope));
+    opBuilder.addInput(segmentIds.asOutput(scope));
+    opBuilder.addInput(numSegments.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new UnsortedSegmentSum<T>(opBuilder.build());
   }
@@ -92,7 +92,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Xdivy.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Xdivy.java
index 403cdd08849..5c825f899d5 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Xdivy.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Xdivy.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Returns 0 if x == 0, and x / y otherwise, elementwise.
@@ -33,7 +33,7 @@
  * @param <T> data type for {@code z()} output
  */
 @Operator(group = "math")
-public final class Xdivy<T extends Tensor> extends RawOp implements Operand<T> {
+public final class Xdivy<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Xdivy operation.
@@ -44,10 +44,10 @@ public final class Xdivy<T extends Tensor> extends RawOp implements Operand<T> {
    * @return a new instance of Xdivy
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> Xdivy<T> create(Scope scope, Operand<T> x, Operand<T> y) {
+  public static <T extends TType> Xdivy<T> create(Scope scope, Operand<T> x, Operand<T> y) {
     OperationBuilder opBuilder = scope.env().opBuilder("Xdivy", scope.makeOpName("Xdivy"));
-    opBuilder.addInput(x.asOutput());
-    opBuilder.addInput(y.asOutput());
+    opBuilder.addInput(x.asOutput(scope));
+    opBuilder.addInput(y.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new Xdivy<T>(opBuilder.build());
   }
@@ -59,7 +59,7 @@ public Output<T> z() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return z;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Xlog1py.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Xlog1py.java
index 922d17a5913..c53e83bbc51 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Xlog1py.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Xlog1py.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Returns 0 if x == 0, and x * log1p(y) otherwise, elementwise.
@@ -33,7 +33,7 @@
  * @param <T> data type for {@code z()} output
  */
 @Operator(group = "math")
-public final class Xlog1py<T extends Tensor> extends RawOp implements Operand<T> {
+public final class Xlog1py<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Xlog1py operation.
@@ -44,10 +44,10 @@ public final class Xlog1py<T extends Tensor> extends RawOp implements Operand<T>
    * @return a new instance of Xlog1py
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> Xlog1py<T> create(Scope scope, Operand<T> x, Operand<T> y) {
+  public static <T extends TType> Xlog1py<T> create(Scope scope, Operand<T> x, Operand<T> y) {
     OperationBuilder opBuilder = scope.env().opBuilder("Xlog1py", scope.makeOpName("Xlog1py"));
-    opBuilder.addInput(x.asOutput());
-    opBuilder.addInput(y.asOutput());
+    opBuilder.addInput(x.asOutput(scope));
+    opBuilder.addInput(y.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new Xlog1py<T>(opBuilder.build());
   }
@@ -59,7 +59,7 @@ public Output<T> z() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return z;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Xlogy.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Xlogy.java
index c3faff9fc5f..8985e6cc7e7 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Xlogy.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Xlogy.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Returns 0 if x == 0, and x * log(y) otherwise, elementwise.
@@ -33,7 +33,7 @@
  * @param <T> data type for {@code z()} output
  */
 @Operator(group = "math")
-public final class Xlogy<T extends Tensor> extends RawOp implements Operand<T> {
+public final class Xlogy<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Xlogy operation.
@@ -44,10 +44,10 @@ public final class Xlogy<T extends Tensor> extends RawOp implements Operand<T> {
    * @return a new instance of Xlogy
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> Xlogy<T> create(Scope scope, Operand<T> x, Operand<T> y) {
+  public static <T extends TType> Xlogy<T> create(Scope scope, Operand<T> x, Operand<T> y) {
     OperationBuilder opBuilder = scope.env().opBuilder("Xlogy", scope.makeOpName("Xlogy"));
-    opBuilder.addInput(x.asOutput());
-    opBuilder.addInput(y.asOutput());
+    opBuilder.addInput(x.asOutput(scope));
+    opBuilder.addInput(y.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new Xlogy<T>(opBuilder.build());
   }
@@ -59,7 +59,7 @@ public Output<T> z() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return z;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Zeta.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Zeta.java
index 73d33a70736..47cdb43465a 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Zeta.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Zeta.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -38,7 +37,7 @@
  * @param <T> data type for {@code z()} output
  */
 @Operator(group = "math")
-public final class Zeta<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class Zeta<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Zeta operation.
@@ -49,10 +48,10 @@ public final class Zeta<T extends Tensor & TNumber> extends RawOp implements Ope
    * @return a new instance of Zeta
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> Zeta<T> create(Scope scope, Operand<T> x, Operand<T> q) {
+  public static <T extends TNumber> Zeta<T> create(Scope scope, Operand<T> x, Operand<T> q) {
     OperationBuilder opBuilder = scope.env().opBuilder("Zeta", scope.makeOpName("Zeta"));
-    opBuilder.addInput(x.asOutput());
-    opBuilder.addInput(q.asOutput());
+    opBuilder.addInput(x.asOutput(scope));
+    opBuilder.addInput(q.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new Zeta<T>(opBuilder.build());
   }
@@ -64,7 +63,7 @@ public Output<T> z() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return z;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/erfinv.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/erfinv.java
index 22fba9db393..837d55d977b 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/erfinv.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/erfinv.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -32,7 +31,7 @@
  * @param <T> data type for {@code y()} output
  */
 @Operator(group = "math")
-public final class erfinv<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class erfinv<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new erfinv operation.
@@ -42,9 +41,9 @@ public final class erfinv<T extends Tensor & TNumber> extends RawOp implements O
    * @return a new instance of erfinv
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> erfinv<T> create(Scope scope, Operand<T> x) {
+  public static <T extends TNumber> erfinv<T> create(Scope scope, Operand<T> x) {
     OperationBuilder opBuilder = scope.env().opBuilder("Erfinv", scope.makeOpName("erfinv"));
-    opBuilder.addInput(x.asOutput());
+    opBuilder.addInput(x.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new erfinv<T>(opBuilder.build());
   }
@@ -56,7 +55,7 @@ public Output<T> y() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return y;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/Dawsn.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/Dawsn.java
index 6af6cea9003..a57ede341f8 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/Dawsn.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/Dawsn.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -31,7 +30,7 @@
 /**
  * @param <T> data type for {@code y()} output
  */
-public final class Dawsn<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class Dawsn<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Dawsn operation.
@@ -41,9 +40,9 @@ public final class Dawsn<T extends Tensor & TNumber> extends RawOp implements Op
    * @return a new instance of Dawsn
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> Dawsn<T> create(Scope scope, Operand<T> x) {
+  public static <T extends TNumber> Dawsn<T> create(Scope scope, Operand<T> x) {
     OperationBuilder opBuilder = scope.env().opBuilder("Dawsn", scope.makeOpName("Dawsn"));
-    opBuilder.addInput(x.asOutput());
+    opBuilder.addInput(x.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new Dawsn<T>(opBuilder.build());
   }
@@ -55,7 +54,7 @@ public Output<T> y() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return y;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/Expint.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/Expint.java
index c2ca8f678bb..c3ba2e7f3ed 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/Expint.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/Expint.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -31,7 +30,7 @@
 /**
  * @param <T> data type for {@code y()} output
  */
-public final class Expint<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class Expint<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Expint operation.
@@ -41,9 +40,9 @@ public final class Expint<T extends Tensor & TNumber> extends RawOp implements O
    * @return a new instance of Expint
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> Expint<T> create(Scope scope, Operand<T> x) {
+  public static <T extends TNumber> Expint<T> create(Scope scope, Operand<T> x) {
     OperationBuilder opBuilder = scope.env().opBuilder("Expint", scope.makeOpName("Expint"));
-    opBuilder.addInput(x.asOutput());
+    opBuilder.addInput(x.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new Expint<T>(opBuilder.build());
   }
@@ -55,7 +54,7 @@ public Output<T> y() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return y;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/FresnelCos.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/FresnelCos.java
index 0dcc2c2ef1a..43e19ecdf72 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/FresnelCos.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/FresnelCos.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -31,7 +30,7 @@
 /**
  * @param <T> data type for {@code y()} output
  */
-public final class FresnelCos<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class FresnelCos<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new FresnelCos operation.
@@ -41,9 +40,9 @@ public final class FresnelCos<T extends Tensor & TNumber> extends RawOp implemen
    * @return a new instance of FresnelCos
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> FresnelCos<T> create(Scope scope, Operand<T> x) {
+  public static <T extends TNumber> FresnelCos<T> create(Scope scope, Operand<T> x) {
     OperationBuilder opBuilder = scope.env().opBuilder("FresnelCos", scope.makeOpName("FresnelCos"));
-    opBuilder.addInput(x.asOutput());
+    opBuilder.addInput(x.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new FresnelCos<T>(opBuilder.build());
   }
@@ -55,7 +54,7 @@ public Output<T> y() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return y;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/FresnelSin.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/FresnelSin.java
index 4fb9a95d462..dc35737b1b3 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/FresnelSin.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/FresnelSin.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -31,7 +30,7 @@
 /**
  * @param <T> data type for {@code y()} output
  */
-public final class FresnelSin<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class FresnelSin<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new FresnelSin operation.
@@ -41,9 +40,9 @@ public final class FresnelSin<T extends Tensor & TNumber> extends RawOp implemen
    * @return a new instance of FresnelSin
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> FresnelSin<T> create(Scope scope, Operand<T> x) {
+  public static <T extends TNumber> FresnelSin<T> create(Scope scope, Operand<T> x) {
     OperationBuilder opBuilder = scope.env().opBuilder("FresnelSin", scope.makeOpName("FresnelSin"));
-    opBuilder.addInput(x.asOutput());
+    opBuilder.addInput(x.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new FresnelSin<T>(opBuilder.build());
   }
@@ -55,7 +54,7 @@ public Output<T> y() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return y;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/Spence.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/Spence.java
index f7f1931e62d..1e820823960 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/Spence.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/Spence.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -31,7 +30,7 @@
 /**
  * @param <T> data type for {@code y()} output
  */
-public final class Spence<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class Spence<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Spence operation.
@@ -41,9 +40,9 @@ public final class Spence<T extends Tensor & TNumber> extends RawOp implements O
    * @return a new instance of Spence
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> Spence<T> create(Scope scope, Operand<T> x) {
+  public static <T extends TNumber> Spence<T> create(Scope scope, Operand<T> x) {
     OperationBuilder opBuilder = scope.env().opBuilder("Spence", scope.makeOpName("Spence"));
-    opBuilder.addInput(x.asOutput());
+    opBuilder.addInput(x.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new Spence<T>(opBuilder.build());
   }
@@ -55,7 +54,7 @@ public Output<T> y() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return y;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/AvgPool.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/AvgPool.java
index a45e80966c0..d3417d490b0 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/AvgPool.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/AvgPool.java
@@ -22,7 +22,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -38,7 +37,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "nn")
-public final class AvgPool<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class AvgPool<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.nn.AvgPool}
@@ -75,9 +74,9 @@ private Options() {
    * @return a new instance of AvgPool
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> AvgPool<T> create(Scope scope, Operand<T> value, List<Long> ksize, List<Long> strides, String padding, Options... options) {
+  public static <T extends TNumber> AvgPool<T> create(Scope scope, Operand<T> value, List<Long> ksize, List<Long> strides, String padding, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("AvgPool", scope.makeOpName("AvgPool"));
-    opBuilder.addInput(value.asOutput());
+    opBuilder.addInput(value.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     long[] ksizeArray = new long[ksize.size()];
     for (int i = 0; i < ksizeArray.length; ++i) {
@@ -119,7 +118,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/AvgPool3d.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/AvgPool3d.java
index 8c17badbb97..0b9fac0931d 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/AvgPool3d.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/AvgPool3d.java
@@ -22,7 +22,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -35,7 +34,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "nn")
-public final class AvgPool3d<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class AvgPool3d<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.nn.AvgPool3d}
@@ -74,9 +73,9 @@ private Options() {
    * @return a new instance of AvgPool3d
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> AvgPool3d<T> create(Scope scope, Operand<T> input, List<Long> ksize, List<Long> strides, String padding, Options... options) {
+  public static <T extends TNumber> AvgPool3d<T> create(Scope scope, Operand<T> input, List<Long> ksize, List<Long> strides, String padding, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("AvgPool3D", scope.makeOpName("AvgPool3d"));
-    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     long[] ksizeArray = new long[ksize.size()];
     for (int i = 0; i < ksizeArray.length; ++i) {
@@ -118,7 +117,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/AvgPool3dGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/AvgPool3dGrad.java
index 5c75244a47e..4465248b03a 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/AvgPool3dGrad.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/AvgPool3dGrad.java
@@ -22,7 +22,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -36,7 +35,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "nn")
-public final class AvgPool3dGrad<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class AvgPool3dGrad<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.nn.AvgPool3dGrad}
@@ -76,10 +75,10 @@ private Options() {
    * @return a new instance of AvgPool3dGrad
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> AvgPool3dGrad<T> create(Scope scope, Operand<TInt32> origInputShape, Operand<T> grad, List<Long> ksize, List<Long> strides, String padding, Options... options) {
+  public static <T extends TNumber> AvgPool3dGrad<T> create(Scope scope, Operand<TInt32> origInputShape, Operand<T> grad, List<Long> ksize, List<Long> strides, String padding, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("AvgPool3DGrad", scope.makeOpName("AvgPool3dGrad"));
-    opBuilder.addInput(origInputShape.asOutput());
-    opBuilder.addInput(grad.asOutput());
+    opBuilder.addInput(origInputShape.asOutput(scope));
+    opBuilder.addInput(grad.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     long[] ksizeArray = new long[ksize.size()];
     for (int i = 0; i < ksizeArray.length; ++i) {
@@ -121,7 +120,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/AvgPoolGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/AvgPoolGrad.java
index 0229a5d08e7..490bf1c6e89 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/AvgPoolGrad.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/AvgPoolGrad.java
@@ -22,7 +22,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -35,7 +34,7 @@
  * 
  * @param <T> data type for {@code output()} output
  */
-public final class AvgPoolGrad<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class AvgPoolGrad<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.nn.AvgPoolGrad}
@@ -74,10 +73,10 @@ private Options() {
    * @return a new instance of AvgPoolGrad
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> AvgPoolGrad<T> create(Scope scope, Operand<TInt32> origInputShape, Operand<T> grad, List<Long> ksize, List<Long> strides, String padding, Options... options) {
+  public static <T extends TNumber> AvgPoolGrad<T> create(Scope scope, Operand<TInt32> origInputShape, Operand<T> grad, List<Long> ksize, List<Long> strides, String padding, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("AvgPoolGrad", scope.makeOpName("AvgPoolGrad"));
-    opBuilder.addInput(origInputShape.asOutput());
-    opBuilder.addInput(grad.asOutput());
+    opBuilder.addInput(origInputShape.asOutput(scope));
+    opBuilder.addInput(grad.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     long[] ksizeArray = new long[ksize.size()];
     for (int i = 0; i < ksizeArray.length; ++i) {
@@ -119,7 +118,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/BatchNormWithGlobalNormalization.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/BatchNormWithGlobalNormalization.java
index 964e00dab01..7b09a43f339 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/BatchNormWithGlobalNormalization.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/BatchNormWithGlobalNormalization.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Batch normalization.
@@ -35,7 +35,7 @@
  * @param <T> data type for {@code result()} output
  */
 @Operator(group = "nn")
-public final class BatchNormWithGlobalNormalization<T extends Tensor> extends RawOp implements Operand<T> {
+public final class BatchNormWithGlobalNormalization<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new BatchNormWithGlobalNormalization operation.
@@ -59,13 +59,13 @@ public final class BatchNormWithGlobalNormalization<T extends Tensor> extends Ra
    * @return a new instance of BatchNormWithGlobalNormalization
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> BatchNormWithGlobalNormalization<T> create(Scope scope, Operand<T> t, Operand<T> m, Operand<T> v, Operand<T> beta, Operand<T> gamma, Float varianceEpsilon, Boolean scaleAfterNormalization) {
+  public static <T extends TType> BatchNormWithGlobalNormalization<T> create(Scope scope, Operand<T> t, Operand<T> m, Operand<T> v, Operand<T> beta, Operand<T> gamma, Float varianceEpsilon, Boolean scaleAfterNormalization) {
     OperationBuilder opBuilder = scope.env().opBuilder("BatchNormWithGlobalNormalization", scope.makeOpName("BatchNormWithGlobalNormalization"));
-    opBuilder.addInput(t.asOutput());
-    opBuilder.addInput(m.asOutput());
-    opBuilder.addInput(v.asOutput());
-    opBuilder.addInput(beta.asOutput());
-    opBuilder.addInput(gamma.asOutput());
+    opBuilder.addInput(t.asOutput(scope));
+    opBuilder.addInput(m.asOutput(scope));
+    opBuilder.addInput(v.asOutput(scope));
+    opBuilder.addInput(beta.asOutput(scope));
+    opBuilder.addInput(gamma.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("variance_epsilon", varianceEpsilon);
     opBuilder.setAttr("scale_after_normalization", scaleAfterNormalization);
@@ -79,7 +79,7 @@ public Output<T> result() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return result;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/BatchNormWithGlobalNormalizationGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/BatchNormWithGlobalNormalizationGrad.java
index 150542bc939..812058fff34 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/BatchNormWithGlobalNormalizationGrad.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/BatchNormWithGlobalNormalizationGrad.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Gradients for batch normalization.
@@ -35,7 +35,7 @@
  * @param <T> data type for {@code dx()} output
  */
 @Operator(group = "nn")
-public final class BatchNormWithGlobalNormalizationGrad<T extends Tensor> extends RawOp {
+public final class BatchNormWithGlobalNormalizationGrad<T extends TType> extends RawOp {
   
   /**
    * Factory method to create a class wrapping a new BatchNormWithGlobalNormalizationGrad operation.
@@ -58,13 +58,13 @@ public final class BatchNormWithGlobalNormalizationGrad<T extends Tensor> extend
    * @return a new instance of BatchNormWithGlobalNormalizationGrad
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> BatchNormWithGlobalNormalizationGrad<T> create(Scope scope, Operand<T> t, Operand<T> m, Operand<T> v, Operand<T> gamma, Operand<T> backprop, Float varianceEpsilon, Boolean scaleAfterNormalization) {
+  public static <T extends TType> BatchNormWithGlobalNormalizationGrad<T> create(Scope scope, Operand<T> t, Operand<T> m, Operand<T> v, Operand<T> gamma, Operand<T> backprop, Float varianceEpsilon, Boolean scaleAfterNormalization) {
     OperationBuilder opBuilder = scope.env().opBuilder("BatchNormWithGlobalNormalizationGrad", scope.makeOpName("BatchNormWithGlobalNormalizationGrad"));
-    opBuilder.addInput(t.asOutput());
-    opBuilder.addInput(m.asOutput());
-    opBuilder.addInput(v.asOutput());
-    opBuilder.addInput(gamma.asOutput());
-    opBuilder.addInput(backprop.asOutput());
+    opBuilder.addInput(t.asOutput(scope));
+    opBuilder.addInput(m.asOutput(scope));
+    opBuilder.addInput(v.asOutput(scope));
+    opBuilder.addInput(gamma.asOutput(scope));
+    opBuilder.addInput(backprop.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("variance_epsilon", varianceEpsilon);
     opBuilder.setAttr("scale_after_normalization", scaleAfterNormalization);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/BiasAdd.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/BiasAdd.java
index 418365edffc..d57cf9c67e1 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/BiasAdd.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/BiasAdd.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Adds `bias` to `value`.
@@ -36,7 +36,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "nn")
-public final class BiasAdd<T extends Tensor> extends RawOp implements Operand<T> {
+public final class BiasAdd<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.nn.BiasAdd}
@@ -73,10 +73,10 @@ private Options() {
    * @return a new instance of BiasAdd
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> BiasAdd<T> create(Scope scope, Operand<T> value, Operand<T> bias, Options... options) {
+  public static <T extends TType> BiasAdd<T> create(Scope scope, Operand<T> value, Operand<T> bias, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("BiasAdd", scope.makeOpName("BiasAdd"));
-    opBuilder.addInput(value.asOutput());
-    opBuilder.addInput(bias.asOutput());
+    opBuilder.addInput(value.asOutput(scope));
+    opBuilder.addInput(bias.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -109,7 +109,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/BiasAddGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/BiasAddGrad.java
index 0ec94390b18..ea4aadd373a 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/BiasAddGrad.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/BiasAddGrad.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * The backward operation for "BiasAdd" on the "bias" tensor.
@@ -37,7 +37,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "nn")
-public final class BiasAddGrad<T extends Tensor> extends RawOp implements Operand<T> {
+public final class BiasAddGrad<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.nn.BiasAddGrad}
@@ -73,9 +73,9 @@ private Options() {
    * @return a new instance of BiasAddGrad
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> BiasAddGrad<T> create(Scope scope, Operand<T> outBackprop, Options... options) {
+  public static <T extends TType> BiasAddGrad<T> create(Scope scope, Operand<T> outBackprop, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("BiasAddGrad", scope.makeOpName("BiasAddGrad"));
-    opBuilder.addInput(outBackprop.asOutput());
+    opBuilder.addInput(outBackprop.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -108,7 +108,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/BlockLSTM.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/BlockLSTM.java
index 9437b6d7d0f..88ec1a64daf 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/BlockLSTM.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/BlockLSTM.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -56,7 +55,7 @@
  * 
  * @param <T> data type for {@code i()} output
  */
-public final class BlockLSTM<T extends Tensor & TNumber> extends RawOp {
+public final class BlockLSTM<T extends TNumber> extends RawOp {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.nn.BlockLSTM}
@@ -104,17 +103,17 @@ private Options() {
    * @return a new instance of BlockLSTM
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> BlockLSTM<T> create(Scope scope, Operand<TInt64> seqLenMax, Operand<T> x, Operand<T> csPrev, Operand<T> hPrev, Operand<T> w, Operand<T> wci, Operand<T> wcf, Operand<T> wco, Operand<T> b, Options... options) {
+  public static <T extends TNumber> BlockLSTM<T> create(Scope scope, Operand<TInt64> seqLenMax, Operand<T> x, Operand<T> csPrev, Operand<T> hPrev, Operand<T> w, Operand<T> wci, Operand<T> wcf, Operand<T> wco, Operand<T> b, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("BlockLSTMV2", scope.makeOpName("BlockLSTM"));
-    opBuilder.addInput(seqLenMax.asOutput());
-    opBuilder.addInput(x.asOutput());
-    opBuilder.addInput(csPrev.asOutput());
-    opBuilder.addInput(hPrev.asOutput());
-    opBuilder.addInput(w.asOutput());
-    opBuilder.addInput(wci.asOutput());
-    opBuilder.addInput(wcf.asOutput());
-    opBuilder.addInput(wco.asOutput());
-    opBuilder.addInput(b.asOutput());
+    opBuilder.addInput(seqLenMax.asOutput(scope));
+    opBuilder.addInput(x.asOutput(scope));
+    opBuilder.addInput(csPrev.asOutput(scope));
+    opBuilder.addInput(hPrev.asOutput(scope));
+    opBuilder.addInput(w.asOutput(scope));
+    opBuilder.addInput(wci.asOutput(scope));
+    opBuilder.addInput(wcf.asOutput(scope));
+    opBuilder.addInput(wco.asOutput(scope));
+    opBuilder.addInput(b.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/BlockLSTMGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/BlockLSTMGrad.java
index 56a10702e5d..4f8907f3ee4 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/BlockLSTMGrad.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/BlockLSTMGrad.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -36,7 +35,7 @@
  * 
  * @param <T> data type for {@code xGrad()} output
  */
-public final class BlockLSTMGrad<T extends Tensor & TNumber> extends RawOp {
+public final class BlockLSTMGrad<T extends TNumber> extends RawOp {
   
   /**
    * Factory method to create a class wrapping a new BlockLSTMGrad operation.
@@ -65,26 +64,26 @@ public final class BlockLSTMGrad<T extends Tensor & TNumber> extends RawOp {
    * @return a new instance of BlockLSTMGrad
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> BlockLSTMGrad<T> create(Scope scope, Operand<TInt64> seqLenMax, Operand<T> x, Operand<T> csPrev, Operand<T> hPrev, Operand<T> w, Operand<T> wci, Operand<T> wcf, Operand<T> wco, Operand<T> b, Operand<T> i, Operand<T> cs, Operand<T> f, Operand<T> o, Operand<T> ci, Operand<T> co, Operand<T> h, Operand<T> csGrad, Operand<T> hGrad, Boolean usePeephole) {
+  public static <T extends TNumber> BlockLSTMGrad<T> create(Scope scope, Operand<TInt64> seqLenMax, Operand<T> x, Operand<T> csPrev, Operand<T> hPrev, Operand<T> w, Operand<T> wci, Operand<T> wcf, Operand<T> wco, Operand<T> b, Operand<T> i, Operand<T> cs, Operand<T> f, Operand<T> o, Operand<T> ci, Operand<T> co, Operand<T> h, Operand<T> csGrad, Operand<T> hGrad, Boolean usePeephole) {
     OperationBuilder opBuilder = scope.env().opBuilder("BlockLSTMGradV2", scope.makeOpName("BlockLSTMGrad"));
-    opBuilder.addInput(seqLenMax.asOutput());
-    opBuilder.addInput(x.asOutput());
-    opBuilder.addInput(csPrev.asOutput());
-    opBuilder.addInput(hPrev.asOutput());
-    opBuilder.addInput(w.asOutput());
-    opBuilder.addInput(wci.asOutput());
-    opBuilder.addInput(wcf.asOutput());
-    opBuilder.addInput(wco.asOutput());
-    opBuilder.addInput(b.asOutput());
-    opBuilder.addInput(i.asOutput());
-    opBuilder.addInput(cs.asOutput());
-    opBuilder.addInput(f.asOutput());
-    opBuilder.addInput(o.asOutput());
-    opBuilder.addInput(ci.asOutput());
-    opBuilder.addInput(co.asOutput());
-    opBuilder.addInput(h.asOutput());
-    opBuilder.addInput(csGrad.asOutput());
-    opBuilder.addInput(hGrad.asOutput());
+    opBuilder.addInput(seqLenMax.asOutput(scope));
+    opBuilder.addInput(x.asOutput(scope));
+    opBuilder.addInput(csPrev.asOutput(scope));
+    opBuilder.addInput(hPrev.asOutput(scope));
+    opBuilder.addInput(w.asOutput(scope));
+    opBuilder.addInput(wci.asOutput(scope));
+    opBuilder.addInput(wcf.asOutput(scope));
+    opBuilder.addInput(wco.asOutput(scope));
+    opBuilder.addInput(b.asOutput(scope));
+    opBuilder.addInput(i.asOutput(scope));
+    opBuilder.addInput(cs.asOutput(scope));
+    opBuilder.addInput(f.asOutput(scope));
+    opBuilder.addInput(o.asOutput(scope));
+    opBuilder.addInput(ci.asOutput(scope));
+    opBuilder.addInput(co.asOutput(scope));
+    opBuilder.addInput(h.asOutput(scope));
+    opBuilder.addInput(csGrad.asOutput(scope));
+    opBuilder.addInput(hGrad.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("use_peephole", usePeephole);
     return new BlockLSTMGrad<T>(opBuilder.build());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CTCLossV2.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CTCLossV2.java
index c60fef4b2a7..c79ab1414fd 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CTCLossV2.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CTCLossV2.java
@@ -96,10 +96,10 @@ private Options() {
   @Endpoint(describeByClass = true)
   public static CTCLossV2 create(Scope scope, Operand<TFloat32> inputs, Operand<TInt64> labelsIndices, Operand<TInt32> labelsValues, Operand<TInt32> sequenceLength, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("CTCLossV2", scope.makeOpName("CTCLossV2"));
-    opBuilder.addInput(inputs.asOutput());
-    opBuilder.addInput(labelsIndices.asOutput());
-    opBuilder.addInput(labelsValues.asOutput());
-    opBuilder.addInput(sequenceLength.asOutput());
+    opBuilder.addInput(inputs.asOutput(scope));
+    opBuilder.addInput(labelsIndices.asOutput(scope));
+    opBuilder.addInput(labelsValues.asOutput(scope));
+    opBuilder.addInput(sequenceLength.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/ComputeAccidentalHits.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/ComputeAccidentalHits.java
index f6daea0df2e..6329a45cb74 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/ComputeAccidentalHits.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/ComputeAccidentalHits.java
@@ -83,8 +83,8 @@ private Options() {
   @Endpoint(describeByClass = true)
   public static ComputeAccidentalHits create(Scope scope, Operand<TInt64> trueClasses, Operand<TInt64> sampledCandidates, Long numTrue, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ComputeAccidentalHits", scope.makeOpName("ComputeAccidentalHits"));
-    opBuilder.addInput(trueClasses.asOutput());
-    opBuilder.addInput(sampledCandidates.asOutput());
+    opBuilder.addInput(trueClasses.asOutput(scope));
+    opBuilder.addInput(sampledCandidates.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("num_true", numTrue);
     if (options != null) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Conv2d.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Conv2d.java
index e194c491408..af5f16acbb2 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Conv2d.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Conv2d.java
@@ -22,7 +22,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -57,7 +56,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "nn")
-public final class Conv2d<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class Conv2d<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.nn.Conv2d}
@@ -132,10 +131,10 @@ private Options() {
    * @return a new instance of Conv2d
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> Conv2d<T> create(Scope scope, Operand<T> input, Operand<T> filter, List<Long> strides, String padding, Options... options) {
+  public static <T extends TNumber> Conv2d<T> create(Scope scope, Operand<T> input, Operand<T> filter, List<Long> strides, String padding, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("Conv2D", scope.makeOpName("Conv2d"));
-    opBuilder.addInput(input.asOutput());
-    opBuilder.addInput(filter.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
+    opBuilder.addInput(filter.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     long[] stridesArray = new long[strides.size()];
     for (int i = 0; i < stridesArray.length; ++i) {
@@ -218,7 +217,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Conv2dBackpropFilter.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Conv2dBackpropFilter.java
index 5e11f5bbcb1..17a7b97f78e 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Conv2dBackpropFilter.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Conv2dBackpropFilter.java
@@ -22,7 +22,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -36,7 +35,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "nn")
-public final class Conv2dBackpropFilter<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class Conv2dBackpropFilter<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.nn.Conv2dBackpropFilter}
@@ -113,11 +112,11 @@ private Options() {
    * @return a new instance of Conv2dBackpropFilter
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> Conv2dBackpropFilter<T> create(Scope scope, Operand<T> input, Operand<TInt32> filterSizes, Operand<T> outBackprop, List<Long> strides, String padding, Options... options) {
+  public static <T extends TNumber> Conv2dBackpropFilter<T> create(Scope scope, Operand<T> input, Operand<TInt32> filterSizes, Operand<T> outBackprop, List<Long> strides, String padding, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("Conv2DBackpropFilter", scope.makeOpName("Conv2dBackpropFilter"));
-    opBuilder.addInput(input.asOutput());
-    opBuilder.addInput(filterSizes.asOutput());
-    opBuilder.addInput(outBackprop.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
+    opBuilder.addInput(filterSizes.asOutput(scope));
+    opBuilder.addInput(outBackprop.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     long[] stridesArray = new long[strides.size()];
     for (int i = 0; i < stridesArray.length; ++i) {
@@ -201,7 +200,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Conv2dBackpropInput.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Conv2dBackpropInput.java
index 23fd6a45aed..6e95b5906ff 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Conv2dBackpropInput.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Conv2dBackpropInput.java
@@ -22,7 +22,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -36,7 +35,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "nn")
-public final class Conv2dBackpropInput<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class Conv2dBackpropInput<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.nn.Conv2dBackpropInput}
@@ -113,11 +112,11 @@ private Options() {
    * @return a new instance of Conv2dBackpropInput
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> Conv2dBackpropInput<T> create(Scope scope, Operand<TInt32> inputSizes, Operand<T> filter, Operand<T> outBackprop, List<Long> strides, String padding, Options... options) {
+  public static <T extends TNumber> Conv2dBackpropInput<T> create(Scope scope, Operand<TInt32> inputSizes, Operand<T> filter, Operand<T> outBackprop, List<Long> strides, String padding, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("Conv2DBackpropInput", scope.makeOpName("Conv2dBackpropInput"));
-    opBuilder.addInput(inputSizes.asOutput());
-    opBuilder.addInput(filter.asOutput());
-    opBuilder.addInput(outBackprop.asOutput());
+    opBuilder.addInput(inputSizes.asOutput(scope));
+    opBuilder.addInput(filter.asOutput(scope));
+    opBuilder.addInput(outBackprop.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     long[] stridesArray = new long[strides.size()];
     for (int i = 0; i < stridesArray.length; ++i) {
@@ -200,7 +199,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Conv3d.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Conv3d.java
index 49817b63e2c..20e44b309f6 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Conv3d.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Conv3d.java
@@ -22,7 +22,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -41,7 +40,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "nn")
-public final class Conv3d<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class Conv3d<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.nn.Conv3d}
@@ -93,10 +92,10 @@ private Options() {
    * @return a new instance of Conv3d
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> Conv3d<T> create(Scope scope, Operand<T> input, Operand<T> filter, List<Long> strides, String padding, Options... options) {
+  public static <T extends TNumber> Conv3d<T> create(Scope scope, Operand<T> input, Operand<T> filter, List<Long> strides, String padding, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("Conv3D", scope.makeOpName("Conv3d"));
-    opBuilder.addInput(input.asOutput());
-    opBuilder.addInput(filter.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
+    opBuilder.addInput(filter.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     long[] stridesArray = new long[strides.size()];
     for (int i = 0; i < stridesArray.length; ++i) {
@@ -150,7 +149,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Conv3dBackpropFilter.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Conv3dBackpropFilter.java
index 7a6bc2549f4..94c4dbd8a01 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Conv3dBackpropFilter.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Conv3dBackpropFilter.java
@@ -22,7 +22,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -36,7 +35,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "nn")
-public final class Conv3dBackpropFilter<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class Conv3dBackpropFilter<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.nn.Conv3dBackpropFilter}
@@ -92,11 +91,11 @@ private Options() {
    * @return a new instance of Conv3dBackpropFilter
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> Conv3dBackpropFilter<T> create(Scope scope, Operand<T> input, Operand<TInt32> filterSizes, Operand<T> outBackprop, List<Long> strides, String padding, Options... options) {
+  public static <T extends TNumber> Conv3dBackpropFilter<T> create(Scope scope, Operand<T> input, Operand<TInt32> filterSizes, Operand<T> outBackprop, List<Long> strides, String padding, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("Conv3DBackpropFilterV2", scope.makeOpName("Conv3dBackpropFilter"));
-    opBuilder.addInput(input.asOutput());
-    opBuilder.addInput(filterSizes.asOutput());
-    opBuilder.addInput(outBackprop.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
+    opBuilder.addInput(filterSizes.asOutput(scope));
+    opBuilder.addInput(outBackprop.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     long[] stridesArray = new long[strides.size()];
     for (int i = 0; i < stridesArray.length; ++i) {
@@ -150,7 +149,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Conv3dBackpropInput.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Conv3dBackpropInput.java
index 478d4841a0c..102305387a7 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Conv3dBackpropInput.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Conv3dBackpropInput.java
@@ -22,7 +22,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -35,7 +34,7 @@
  * @param <U> data type for {@code output()} output
  */
 @Operator(group = "nn")
-public final class Conv3dBackpropInput<U extends Tensor & TNumber> extends RawOp implements Operand<U> {
+public final class Conv3dBackpropInput<U extends TNumber> extends RawOp implements Operand<U> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.nn.Conv3dBackpropInput}
@@ -91,11 +90,11 @@ private Options() {
    * @return a new instance of Conv3dBackpropInput
    */
   @Endpoint(describeByClass = true)
-  public static <U extends Tensor & TNumber, T extends Tensor & TNumber> Conv3dBackpropInput<U> create(Scope scope, Operand<T> inputSizes, Operand<U> filter, Operand<U> outBackprop, List<Long> strides, String padding, Options... options) {
+  public static <U extends TNumber, T extends TNumber> Conv3dBackpropInput<U> create(Scope scope, Operand<T> inputSizes, Operand<U> filter, Operand<U> outBackprop, List<Long> strides, String padding, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("Conv3DBackpropInputV2", scope.makeOpName("Conv3dBackpropInput"));
-    opBuilder.addInput(inputSizes.asOutput());
-    opBuilder.addInput(filter.asOutput());
-    opBuilder.addInput(outBackprop.asOutput());
+    opBuilder.addInput(inputSizes.asOutput(scope));
+    opBuilder.addInput(filter.asOutput(scope));
+    opBuilder.addInput(outBackprop.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     long[] stridesArray = new long[strides.size()];
     for (int i = 0; i < stridesArray.length; ++i) {
@@ -149,7 +148,7 @@ public Output<U> output() {
   }
   
   @Override
-  public Output<U> asOutput() {
+  public Output<U> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CtcBeamSearchDecoder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CtcBeamSearchDecoder.java
index d2f84a70d26..b9952e52777 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CtcBeamSearchDecoder.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CtcBeamSearchDecoder.java
@@ -23,7 +23,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -44,7 +43,7 @@
  * @param <T> data type for {@code logProbability()} output
  */
 @Operator(group = "nn")
-public final class CtcBeamSearchDecoder<T extends Tensor & TNumber> extends RawOp {
+public final class CtcBeamSearchDecoder<T extends TNumber> extends RawOp {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.nn.CtcBeamSearchDecoder}
@@ -77,10 +76,10 @@ private Options() {
    * @return a new instance of CtcBeamSearchDecoder
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> CtcBeamSearchDecoder<T> create(Scope scope, Operand<T> inputs, Operand<TInt32> sequenceLength, Long beamWidth, Long topPaths, Options... options) {
+  public static <T extends TNumber> CtcBeamSearchDecoder<T> create(Scope scope, Operand<T> inputs, Operand<TInt32> sequenceLength, Long beamWidth, Long topPaths, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("CTCBeamSearchDecoder", scope.makeOpName("CtcBeamSearchDecoder"));
-    opBuilder.addInput(inputs.asOutput());
-    opBuilder.addInput(sequenceLength.asOutput());
+    opBuilder.addInput(inputs.asOutput(scope));
+    opBuilder.addInput(sequenceLength.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("beam_width", beamWidth);
     opBuilder.setAttr("top_paths", topPaths);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CtcGreedyDecoder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CtcGreedyDecoder.java
index 93c505c22c7..8ba71cd466e 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CtcGreedyDecoder.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CtcGreedyDecoder.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -46,7 +45,7 @@
  * @param <T> data type for {@code logProbability()} output
  */
 @Operator(group = "nn")
-public final class CtcGreedyDecoder<T extends Tensor & TNumber> extends RawOp {
+public final class CtcGreedyDecoder<T extends TNumber> extends RawOp {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.nn.CtcGreedyDecoder}
@@ -77,10 +76,10 @@ private Options() {
    * @return a new instance of CtcGreedyDecoder
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> CtcGreedyDecoder<T> create(Scope scope, Operand<T> inputs, Operand<TInt32> sequenceLength, Options... options) {
+  public static <T extends TNumber> CtcGreedyDecoder<T> create(Scope scope, Operand<T> inputs, Operand<TInt32> sequenceLength, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("CTCGreedyDecoder", scope.makeOpName("CtcGreedyDecoder"));
-    opBuilder.addInput(inputs.asOutput());
-    opBuilder.addInput(sequenceLength.asOutput());
+    opBuilder.addInput(inputs.asOutput(scope));
+    opBuilder.addInput(sequenceLength.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CtcLoss.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CtcLoss.java
index e464a6da181..db4df7fe909 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CtcLoss.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CtcLoss.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -39,7 +38,7 @@
  * @param <T> data type for {@code loss()} output
  */
 @Operator(group = "nn")
-public final class CtcLoss<T extends Tensor & TNumber> extends RawOp {
+public final class CtcLoss<T extends TNumber> extends RawOp {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.nn.CtcLoss}
@@ -97,12 +96,12 @@ private Options() {
    * @return a new instance of CtcLoss
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> CtcLoss<T> create(Scope scope, Operand<T> inputs, Operand<TInt64> labelsIndices, Operand<TInt32> labelsValues, Operand<TInt32> sequenceLength, Options... options) {
+  public static <T extends TNumber> CtcLoss<T> create(Scope scope, Operand<T> inputs, Operand<TInt64> labelsIndices, Operand<TInt32> labelsValues, Operand<TInt32> sequenceLength, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("CTCLoss", scope.makeOpName("CtcLoss"));
-    opBuilder.addInput(inputs.asOutput());
-    opBuilder.addInput(labelsIndices.asOutput());
-    opBuilder.addInput(labelsValues.asOutput());
-    opBuilder.addInput(sequenceLength.asOutput());
+    opBuilder.addInput(inputs.asOutput(scope));
+    opBuilder.addInput(labelsIndices.asOutput(scope));
+    opBuilder.addInput(labelsValues.asOutput(scope));
+    opBuilder.addInput(sequenceLength.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CudnnRNN.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CudnnRNN.java
index 1c482ad4f14..a3fc18e8ec5 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CudnnRNN.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CudnnRNN.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -72,7 +71,7 @@
  * 
  * @param <T> data type for {@code output()} output
  */
-public final class CudnnRNN<T extends Tensor & TNumber> extends RawOp {
+public final class CudnnRNN<T extends TNumber> extends RawOp {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.nn.CudnnRNN}
@@ -178,13 +177,13 @@ private Options() {
    * @return a new instance of CudnnRNN
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> CudnnRNN<T> create(Scope scope, Operand<T> input, Operand<T> inputH, Operand<T> inputC, Operand<T> params, Operand<TInt32> sequenceLengths, Options... options) {
+  public static <T extends TNumber> CudnnRNN<T> create(Scope scope, Operand<T> input, Operand<T> inputH, Operand<T> inputC, Operand<T> params, Operand<TInt32> sequenceLengths, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("CudnnRNNV3", scope.makeOpName("CudnnRNN"));
-    opBuilder.addInput(input.asOutput());
-    opBuilder.addInput(inputH.asOutput());
-    opBuilder.addInput(inputC.asOutput());
-    opBuilder.addInput(params.asOutput());
-    opBuilder.addInput(sequenceLengths.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
+    opBuilder.addInput(inputH.asOutput(scope));
+    opBuilder.addInput(inputC.asOutput(scope));
+    opBuilder.addInput(params.asOutput(scope));
+    opBuilder.addInput(sequenceLengths.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CudnnRNNBackprop.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CudnnRNNBackprop.java
index 0dd6347a4ab..53c32de11cc 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CudnnRNNBackprop.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CudnnRNNBackprop.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -82,7 +81,7 @@
  * 
  * @param <T> data type for {@code inputBackprop()} output
  */
-public final class CudnnRNNBackprop<T extends Tensor & TNumber> extends RawOp {
+public final class CudnnRNNBackprop<T extends TNumber> extends RawOp {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.nn.CudnnRNNBackprop}
@@ -187,21 +186,21 @@ private Options() {
    * @return a new instance of CudnnRNNBackprop
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> CudnnRNNBackprop<T> create(Scope scope, Operand<T> input, Operand<T> inputH, Operand<T> inputC, Operand<T> params, Operand<TInt32> sequenceLengths, Operand<T> output, Operand<T> outputH, Operand<T> outputC, Operand<T> outputBackprop, Operand<T> outputHBackprop, Operand<T> outputCBackprop, Operand<T> reserveSpace, Operand<?> hostReserved, Options... options) {
+  public static <T extends TNumber> CudnnRNNBackprop<T> create(Scope scope, Operand<T> input, Operand<T> inputH, Operand<T> inputC, Operand<T> params, Operand<TInt32> sequenceLengths, Operand<T> output, Operand<T> outputH, Operand<T> outputC, Operand<T> outputBackprop, Operand<T> outputHBackprop, Operand<T> outputCBackprop, Operand<T> reserveSpace, Operand<?> hostReserved, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("CudnnRNNBackpropV3", scope.makeOpName("CudnnRNNBackprop"));
-    opBuilder.addInput(input.asOutput());
-    opBuilder.addInput(inputH.asOutput());
-    opBuilder.addInput(inputC.asOutput());
-    opBuilder.addInput(params.asOutput());
-    opBuilder.addInput(sequenceLengths.asOutput());
-    opBuilder.addInput(output.asOutput());
-    opBuilder.addInput(outputH.asOutput());
-    opBuilder.addInput(outputC.asOutput());
-    opBuilder.addInput(outputBackprop.asOutput());
-    opBuilder.addInput(outputHBackprop.asOutput());
-    opBuilder.addInput(outputCBackprop.asOutput());
-    opBuilder.addInput(reserveSpace.asOutput());
-    opBuilder.addInput(hostReserved.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
+    opBuilder.addInput(inputH.asOutput(scope));
+    opBuilder.addInput(inputC.asOutput(scope));
+    opBuilder.addInput(params.asOutput(scope));
+    opBuilder.addInput(sequenceLengths.asOutput(scope));
+    opBuilder.addInput(output.asOutput(scope));
+    opBuilder.addInput(outputH.asOutput(scope));
+    opBuilder.addInput(outputC.asOutput(scope));
+    opBuilder.addInput(outputBackprop.asOutput(scope));
+    opBuilder.addInput(outputHBackprop.asOutput(scope));
+    opBuilder.addInput(outputCBackprop.asOutput(scope));
+    opBuilder.addInput(reserveSpace.asOutput(scope));
+    opBuilder.addInput(hostReserved.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CudnnRNNCanonicalToParams.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CudnnRNNCanonicalToParams.java
index efc3c0ed8db..c92c6e8c8d0 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CudnnRNNCanonicalToParams.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CudnnRNNCanonicalToParams.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.Operands;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
@@ -67,7 +66,7 @@
  * @param <T> data type for {@code params()} output
  */
 @Operator(group = "nn")
-public final class CudnnRNNCanonicalToParams<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class CudnnRNNCanonicalToParams<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.nn.CudnnRNNCanonicalToParams}
@@ -155,13 +154,13 @@ private Options() {
    * @return a new instance of CudnnRNNCanonicalToParams
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> CudnnRNNCanonicalToParams<T> create(Scope scope, Operand<TInt32> numLayers, Operand<TInt32> numUnits, Operand<TInt32> inputSize, Iterable<Operand<T>> weights, Iterable<Operand<T>> biases, Options... options) {
+  public static <T extends TNumber> CudnnRNNCanonicalToParams<T> create(Scope scope, Operand<TInt32> numLayers, Operand<TInt32> numUnits, Operand<TInt32> inputSize, Iterable<Operand<T>> weights, Iterable<Operand<T>> biases, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("CudnnRNNCanonicalToParamsV2", scope.makeOpName("CudnnRNNCanonicalToParams"));
-    opBuilder.addInput(numLayers.asOutput());
-    opBuilder.addInput(numUnits.asOutput());
-    opBuilder.addInput(inputSize.asOutput());
-    opBuilder.addInputList(Operands.asOutputs(weights));
-    opBuilder.addInputList(Operands.asOutputs(biases));
+    opBuilder.addInput(numLayers.asOutput(scope));
+    opBuilder.addInput(numUnits.asOutput(scope));
+    opBuilder.addInput(inputSize.asOutput(scope));
+    opBuilder.addInputList(Operands.asOutputs(scope, weights));
+    opBuilder.addInputList(Operands.asOutputs(scope, biases));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -247,7 +246,7 @@ public Output<T> params() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return params;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CudnnRNNParamsToCanonical.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CudnnRNNParamsToCanonical.java
index b6b09158eb6..c76cef9ced0 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CudnnRNNParamsToCanonical.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CudnnRNNParamsToCanonical.java
@@ -23,7 +23,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -68,7 +67,7 @@
  * @param <T> data type for {@code weights()} output
  */
 @Operator(group = "nn")
-public final class CudnnRNNParamsToCanonical<T extends Tensor & TNumber> extends RawOp {
+public final class CudnnRNNParamsToCanonical<T extends TNumber> extends RawOp {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.nn.CudnnRNNParamsToCanonical}
@@ -157,12 +156,12 @@ private Options() {
    * @return a new instance of CudnnRNNParamsToCanonical
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> CudnnRNNParamsToCanonical<T> create(Scope scope, Operand<TInt32> numLayers, Operand<TInt32> numUnits, Operand<TInt32> inputSize, Operand<T> params, Long numParamsWeights, Long numParamsBiases, Options... options) {
+  public static <T extends TNumber> CudnnRNNParamsToCanonical<T> create(Scope scope, Operand<TInt32> numLayers, Operand<TInt32> numUnits, Operand<TInt32> inputSize, Operand<T> params, Long numParamsWeights, Long numParamsBiases, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("CudnnRNNParamsToCanonicalV2", scope.makeOpName("CudnnRNNParamsToCanonical"));
-    opBuilder.addInput(numLayers.asOutput());
-    opBuilder.addInput(numUnits.asOutput());
-    opBuilder.addInput(inputSize.asOutput());
-    opBuilder.addInput(params.asOutput());
+    opBuilder.addInput(numLayers.asOutput(scope));
+    opBuilder.addInput(numUnits.asOutput(scope));
+    opBuilder.addInput(inputSize.asOutput(scope));
+    opBuilder.addInput(params.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("num_params_weights", numParamsWeights);
     opBuilder.setAttr("num_params_biases", numParamsBiases);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CudnnRnnParamsSize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CudnnRnnParamsSize.java
index 6758e30cb0d..cdffe933e65 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CudnnRnnParamsSize.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CudnnRnnParamsSize.java
@@ -22,7 +22,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -58,7 +57,7 @@
  * @param <U> data type for {@code paramsSize()} output
  */
 @Operator(group = "nn")
-public final class CudnnRnnParamsSize<U extends Tensor & TNumber> extends RawOp implements Operand<U> {
+public final class CudnnRnnParamsSize<U extends TNumber> extends RawOp implements Operand<U> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.nn.CudnnRnnParamsSize}
@@ -146,11 +145,11 @@ private Options() {
    * @return a new instance of CudnnRnnParamsSize
    */
   @Endpoint(describeByClass = true)
-  public static <U extends Tensor & TNumber, T extends Tensor & TNumber> CudnnRnnParamsSize<U> create(Scope scope, Operand<TInt32> numLayers, Operand<TInt32> numUnits, Operand<TInt32> inputSize, DataType<T> T, DataType<U> S, Options... options) {
+  public static <U extends TNumber, T extends TNumber> CudnnRnnParamsSize<U> create(Scope scope, Operand<TInt32> numLayers, Operand<TInt32> numUnits, Operand<TInt32> inputSize, DataType<T> T, DataType<U> S, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("CudnnRNNParamsSize", scope.makeOpName("CudnnRnnParamsSize"));
-    opBuilder.addInput(numLayers.asOutput());
-    opBuilder.addInput(numUnits.asOutput());
-    opBuilder.addInput(inputSize.asOutput());
+    opBuilder.addInput(numLayers.asOutput(scope));
+    opBuilder.addInput(numUnits.asOutput(scope));
+    opBuilder.addInput(inputSize.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("T", T);
     opBuilder.setAttr("S", S);
@@ -238,7 +237,7 @@ public Output<U> paramsSize() {
   }
   
   @Override
-  public Output<U> asOutput() {
+  public Output<U> asOutput(Scope scope) {
     return paramsSize;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/DataFormatDimMap.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/DataFormatDimMap.java
index 8e107e01b4e..8090d201190 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/DataFormatDimMap.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/DataFormatDimMap.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -36,7 +35,7 @@
  * @param <T> data type for {@code y()} output
  */
 @Operator(group = "nn")
-public final class DataFormatDimMap<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class DataFormatDimMap<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.nn.DataFormatDimMap}
@@ -76,9 +75,9 @@ private Options() {
    * @return a new instance of DataFormatDimMap
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> DataFormatDimMap<T> create(Scope scope, Operand<T> x, Options... options) {
+  public static <T extends TNumber> DataFormatDimMap<T> create(Scope scope, Operand<T> x, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("DataFormatDimMap", scope.makeOpName("DataFormatDimMap"));
-    opBuilder.addInput(x.asOutput());
+    opBuilder.addInput(x.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -115,7 +114,7 @@ public Output<T> y() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return y;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/DataFormatVecPermute.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/DataFormatVecPermute.java
index daf50364005..f9b7f096ed3 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/DataFormatVecPermute.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/DataFormatVecPermute.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -36,7 +35,7 @@
  * @param <T> data type for {@code y()} output
  */
 @Operator(group = "nn")
-public final class DataFormatVecPermute<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class DataFormatVecPermute<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.nn.DataFormatVecPermute}
@@ -75,9 +74,9 @@ private Options() {
    * @return a new instance of DataFormatVecPermute
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> DataFormatVecPermute<T> create(Scope scope, Operand<T> x, Options... options) {
+  public static <T extends TNumber> DataFormatVecPermute<T> create(Scope scope, Operand<T> x, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("DataFormatVecPermute", scope.makeOpName("DataFormatVecPermute"));
-    opBuilder.addInput(x.asOutput());
+    opBuilder.addInput(x.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -114,7 +113,7 @@ public Output<T> y() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return y;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/DepthToSpace.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/DepthToSpace.java
index 1f84ac1f2bb..a35e6e9e474 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/DepthToSpace.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/DepthToSpace.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * DepthToSpace for tensors of type T.
@@ -113,7 +113,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "nn")
-public final class DepthToSpace<T extends Tensor> extends RawOp implements Operand<T> {
+public final class DepthToSpace<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.nn.DepthToSpace}
@@ -144,9 +144,9 @@ private Options() {
    * @return a new instance of DepthToSpace
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> DepthToSpace<T> create(Scope scope, Operand<T> input, Long blockSize, Options... options) {
+  public static <T extends TType> DepthToSpace<T> create(Scope scope, Operand<T> input, Long blockSize, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("DepthToSpace", scope.makeOpName("DepthToSpace"));
-    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("block_size", blockSize);
     if (options != null) {
@@ -173,7 +173,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/DepthwiseConv2dNative.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/DepthwiseConv2dNative.java
index cfb054de485..5e799acd0a5 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/DepthwiseConv2dNative.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/DepthwiseConv2dNative.java
@@ -22,7 +22,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -52,7 +51,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "nn")
-public final class DepthwiseConv2dNative<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class DepthwiseConv2dNative<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.nn.DepthwiseConv2dNative}
@@ -112,10 +111,10 @@ private Options() {
    * @return a new instance of DepthwiseConv2dNative
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> DepthwiseConv2dNative<T> create(Scope scope, Operand<T> input, Operand<T> filter, List<Long> strides, String padding, Options... options) {
+  public static <T extends TNumber> DepthwiseConv2dNative<T> create(Scope scope, Operand<T> input, Operand<T> filter, List<Long> strides, String padding, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("DepthwiseConv2dNative", scope.makeOpName("DepthwiseConv2dNative"));
-    opBuilder.addInput(input.asOutput());
-    opBuilder.addInput(filter.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
+    opBuilder.addInput(filter.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     long[] stridesArray = new long[strides.size()];
     for (int i = 0; i < stridesArray.length; ++i) {
@@ -183,7 +182,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/DepthwiseConv2dNativeBackpropFilter.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/DepthwiseConv2dNativeBackpropFilter.java
index a9f82a9d2f4..bee307ed16b 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/DepthwiseConv2dNativeBackpropFilter.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/DepthwiseConv2dNativeBackpropFilter.java
@@ -22,7 +22,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -36,7 +35,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "nn")
-public final class DepthwiseConv2dNativeBackpropFilter<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class DepthwiseConv2dNativeBackpropFilter<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.nn.DepthwiseConv2dNativeBackpropFilter}
@@ -104,11 +103,11 @@ private Options() {
    * @return a new instance of DepthwiseConv2dNativeBackpropFilter
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> DepthwiseConv2dNativeBackpropFilter<T> create(Scope scope, Operand<T> input, Operand<TInt32> filterSizes, Operand<T> outBackprop, List<Long> strides, String padding, Options... options) {
+  public static <T extends TNumber> DepthwiseConv2dNativeBackpropFilter<T> create(Scope scope, Operand<T> input, Operand<TInt32> filterSizes, Operand<T> outBackprop, List<Long> strides, String padding, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("DepthwiseConv2dNativeBackpropFilter", scope.makeOpName("DepthwiseConv2dNativeBackpropFilter"));
-    opBuilder.addInput(input.asOutput());
-    opBuilder.addInput(filterSizes.asOutput());
-    opBuilder.addInput(outBackprop.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
+    opBuilder.addInput(filterSizes.asOutput(scope));
+    opBuilder.addInput(outBackprop.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     long[] stridesArray = new long[strides.size()];
     for (int i = 0; i < stridesArray.length; ++i) {
@@ -179,7 +178,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/DepthwiseConv2dNativeBackpropInput.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/DepthwiseConv2dNativeBackpropInput.java
index 51ff8495ff3..99d8e2933f2 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/DepthwiseConv2dNativeBackpropInput.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/DepthwiseConv2dNativeBackpropInput.java
@@ -22,7 +22,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -36,7 +35,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "nn")
-public final class DepthwiseConv2dNativeBackpropInput<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class DepthwiseConv2dNativeBackpropInput<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.nn.DepthwiseConv2dNativeBackpropInput}
@@ -103,11 +102,11 @@ private Options() {
    * @return a new instance of DepthwiseConv2dNativeBackpropInput
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> DepthwiseConv2dNativeBackpropInput<T> create(Scope scope, Operand<TInt32> inputSizes, Operand<T> filter, Operand<T> outBackprop, List<Long> strides, String padding, Options... options) {
+  public static <T extends TNumber> DepthwiseConv2dNativeBackpropInput<T> create(Scope scope, Operand<TInt32> inputSizes, Operand<T> filter, Operand<T> outBackprop, List<Long> strides, String padding, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("DepthwiseConv2dNativeBackpropInput", scope.makeOpName("DepthwiseConv2dNativeBackpropInput"));
-    opBuilder.addInput(inputSizes.asOutput());
-    opBuilder.addInput(filter.asOutput());
-    opBuilder.addInput(outBackprop.asOutput());
+    opBuilder.addInput(inputSizes.asOutput(scope));
+    opBuilder.addInput(filter.asOutput(scope));
+    opBuilder.addInput(outBackprop.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     long[] stridesArray = new long[strides.size()];
     for (int i = 0; i < stridesArray.length; ++i) {
@@ -179,7 +178,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Dilation2d.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Dilation2d.java
index 908feaf06a0..e07c5ca6047 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Dilation2d.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Dilation2d.java
@@ -22,7 +22,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -59,7 +58,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "nn")
-public final class Dilation2d<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class Dilation2d<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Dilation2d operation.
@@ -75,10 +74,10 @@ public final class Dilation2d<T extends Tensor & TNumber> extends RawOp implemen
    * @return a new instance of Dilation2d
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> Dilation2d<T> create(Scope scope, Operand<T> input, Operand<T> filter, List<Long> strides, List<Long> rates, String padding) {
+  public static <T extends TNumber> Dilation2d<T> create(Scope scope, Operand<T> input, Operand<T> filter, List<Long> strides, List<Long> rates, String padding) {
     OperationBuilder opBuilder = scope.env().opBuilder("Dilation2D", scope.makeOpName("Dilation2d"));
-    opBuilder.addInput(input.asOutput());
-    opBuilder.addInput(filter.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
+    opBuilder.addInput(filter.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     long[] stridesArray = new long[strides.size()];
     for (int i = 0; i < stridesArray.length; ++i) {
@@ -102,7 +101,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Dilation2dBackpropFilter.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Dilation2dBackpropFilter.java
index 9f44fbee1f8..b9f59b07796 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Dilation2dBackpropFilter.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Dilation2dBackpropFilter.java
@@ -22,7 +22,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -35,7 +34,7 @@
  * @param <T> data type for {@code filterBackprop()} output
  */
 @Operator(group = "nn")
-public final class Dilation2dBackpropFilter<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class Dilation2dBackpropFilter<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Dilation2dBackpropFilter operation.
@@ -52,11 +51,11 @@ public final class Dilation2dBackpropFilter<T extends Tensor & TNumber> extends
    * @return a new instance of Dilation2dBackpropFilter
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> Dilation2dBackpropFilter<T> create(Scope scope, Operand<T> input, Operand<T> filter, Operand<T> outBackprop, List<Long> strides, List<Long> rates, String padding) {
+  public static <T extends TNumber> Dilation2dBackpropFilter<T> create(Scope scope, Operand<T> input, Operand<T> filter, Operand<T> outBackprop, List<Long> strides, List<Long> rates, String padding) {
     OperationBuilder opBuilder = scope.env().opBuilder("Dilation2DBackpropFilter", scope.makeOpName("Dilation2dBackpropFilter"));
-    opBuilder.addInput(input.asOutput());
-    opBuilder.addInput(filter.asOutput());
-    opBuilder.addInput(outBackprop.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
+    opBuilder.addInput(filter.asOutput(scope));
+    opBuilder.addInput(outBackprop.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     long[] stridesArray = new long[strides.size()];
     for (int i = 0; i < stridesArray.length; ++i) {
@@ -80,7 +79,7 @@ public Output<T> filterBackprop() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return filterBackprop;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Dilation2dBackpropInput.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Dilation2dBackpropInput.java
index bb71ec8c62c..7634c99cb6a 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Dilation2dBackpropInput.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Dilation2dBackpropInput.java
@@ -22,7 +22,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -35,7 +34,7 @@
  * @param <T> data type for {@code inBackprop()} output
  */
 @Operator(group = "nn")
-public final class Dilation2dBackpropInput<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class Dilation2dBackpropInput<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Dilation2dBackpropInput operation.
@@ -52,11 +51,11 @@ public final class Dilation2dBackpropInput<T extends Tensor & TNumber> extends R
    * @return a new instance of Dilation2dBackpropInput
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> Dilation2dBackpropInput<T> create(Scope scope, Operand<T> input, Operand<T> filter, Operand<T> outBackprop, List<Long> strides, List<Long> rates, String padding) {
+  public static <T extends TNumber> Dilation2dBackpropInput<T> create(Scope scope, Operand<T> input, Operand<T> filter, Operand<T> outBackprop, List<Long> strides, List<Long> rates, String padding) {
     OperationBuilder opBuilder = scope.env().opBuilder("Dilation2DBackpropInput", scope.makeOpName("Dilation2dBackpropInput"));
-    opBuilder.addInput(input.asOutput());
-    opBuilder.addInput(filter.asOutput());
-    opBuilder.addInput(outBackprop.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
+    opBuilder.addInput(filter.asOutput(scope));
+    opBuilder.addInput(outBackprop.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     long[] stridesArray = new long[strides.size()];
     for (int i = 0; i < stridesArray.length; ++i) {
@@ -80,7 +79,7 @@ public Output<T> inBackprop() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return inBackprop;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Elu.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Elu.java
index 47259f7e42b..6c855873ce4 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Elu.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Elu.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -37,7 +36,7 @@
  * @param <T> data type for {@code activations()} output
  */
 @Operator(group = "nn")
-public final class Elu<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class Elu<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Elu operation.
@@ -47,9 +46,9 @@ public final class Elu<T extends Tensor & TNumber> extends RawOp implements Oper
    * @return a new instance of Elu
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> Elu<T> create(Scope scope, Operand<T> features) {
+  public static <T extends TNumber> Elu<T> create(Scope scope, Operand<T> features) {
     OperationBuilder opBuilder = scope.env().opBuilder("Elu", scope.makeOpName("Elu"));
-    opBuilder.addInput(features.asOutput());
+    opBuilder.addInput(features.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new Elu<T>(opBuilder.build());
   }
@@ -61,7 +60,7 @@ public Output<T> activations() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return activations;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/EluGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/EluGrad.java
index 3c058eae14e..9776b7c2f23 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/EluGrad.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/EluGrad.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -33,7 +32,7 @@
  * 
  * @param <T> data type for {@code backprops()} output
  */
-public final class EluGrad<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class EluGrad<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new EluGrad operation.
@@ -44,10 +43,10 @@ public final class EluGrad<T extends Tensor & TNumber> extends RawOp implements
    * @return a new instance of EluGrad
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> EluGrad<T> create(Scope scope, Operand<T> gradients, Operand<T> outputs) {
+  public static <T extends TNumber> EluGrad<T> create(Scope scope, Operand<T> gradients, Operand<T> outputs) {
     OperationBuilder opBuilder = scope.env().opBuilder("EluGrad", scope.makeOpName("EluGrad"));
-    opBuilder.addInput(gradients.asOutput());
-    opBuilder.addInput(outputs.asOutput());
+    opBuilder.addInput(gradients.asOutput(scope));
+    opBuilder.addInput(outputs.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new EluGrad<T>(opBuilder.build());
   }
@@ -61,7 +60,7 @@ public Output<T> backprops() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return backprops;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/FixedUnigramCandidateSampler.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/FixedUnigramCandidateSampler.java
index 25fd11071f9..50255afbf49 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/FixedUnigramCandidateSampler.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/FixedUnigramCandidateSampler.java
@@ -169,7 +169,7 @@ private Options() {
   @Endpoint(describeByClass = true)
   public static FixedUnigramCandidateSampler create(Scope scope, Operand<TInt64> trueClasses, Long numTrue, Long numSampled, Boolean unique, Long rangeMax, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("FixedUnigramCandidateSampler", scope.makeOpName("FixedUnigramCandidateSampler"));
-    opBuilder.addInput(trueClasses.asOutput());
+    opBuilder.addInput(trueClasses.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("num_true", numTrue);
     opBuilder.setAttr("num_sampled", numSampled);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/FractionalAvgPool.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/FractionalAvgPool.java
index 2dab6ebd063..2c3888ab6a5 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/FractionalAvgPool.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/FractionalAvgPool.java
@@ -22,7 +22,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -41,7 +40,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "nn")
-public final class FractionalAvgPool<T extends Tensor & TNumber> extends RawOp {
+public final class FractionalAvgPool<T extends TNumber> extends RawOp {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.nn.FractionalAvgPool}
@@ -128,9 +127,9 @@ private Options() {
    * @return a new instance of FractionalAvgPool
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> FractionalAvgPool<T> create(Scope scope, Operand<T> value, List<Float> poolingRatio, Options... options) {
+  public static <T extends TNumber> FractionalAvgPool<T> create(Scope scope, Operand<T> value, List<Float> poolingRatio, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("FractionalAvgPool", scope.makeOpName("FractionalAvgPool"));
-    opBuilder.addInput(value.asOutput());
+    opBuilder.addInput(value.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     float[] poolingRatioArray = new float[poolingRatio.size()];
     for (int i = 0; i < poolingRatioArray.length; ++i) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/FractionalAvgPoolGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/FractionalAvgPoolGrad.java
index 5b12f66407e..0d052aee581 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/FractionalAvgPoolGrad.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/FractionalAvgPoolGrad.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -40,7 +39,7 @@
  * 
  * @param <T> data type for {@code output()} output
  */
-public final class FractionalAvgPoolGrad<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class FractionalAvgPoolGrad<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.nn.FractionalAvgPoolGrad}
@@ -84,12 +83,12 @@ private Options() {
    * @return a new instance of FractionalAvgPoolGrad
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> FractionalAvgPoolGrad<T> create(Scope scope, Operand<TInt64> origInputTensorShape, Operand<T> outBackprop, Operand<TInt64> rowPoolingSequence, Operand<TInt64> colPoolingSequence, Options... options) {
+  public static <T extends TNumber> FractionalAvgPoolGrad<T> create(Scope scope, Operand<TInt64> origInputTensorShape, Operand<T> outBackprop, Operand<TInt64> rowPoolingSequence, Operand<TInt64> colPoolingSequence, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("FractionalAvgPoolGrad", scope.makeOpName("FractionalAvgPoolGrad"));
-    opBuilder.addInput(origInputTensorShape.asOutput());
-    opBuilder.addInput(outBackprop.asOutput());
-    opBuilder.addInput(rowPoolingSequence.asOutput());
-    opBuilder.addInput(colPoolingSequence.asOutput());
+    opBuilder.addInput(origInputTensorShape.asOutput(scope));
+    opBuilder.addInput(outBackprop.asOutput(scope));
+    opBuilder.addInput(rowPoolingSequence.asOutput(scope));
+    opBuilder.addInput(colPoolingSequence.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -124,7 +123,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/FractionalMaxPool.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/FractionalMaxPool.java
index 2519445ae87..234ec684451 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/FractionalMaxPool.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/FractionalMaxPool.java
@@ -22,7 +22,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -65,7 +64,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "nn")
-public final class FractionalMaxPool<T extends Tensor & TNumber> extends RawOp {
+public final class FractionalMaxPool<T extends TNumber> extends RawOp {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.nn.FractionalMaxPool}
@@ -152,9 +151,9 @@ private Options() {
    * @return a new instance of FractionalMaxPool
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> FractionalMaxPool<T> create(Scope scope, Operand<T> value, List<Float> poolingRatio, Options... options) {
+  public static <T extends TNumber> FractionalMaxPool<T> create(Scope scope, Operand<T> value, List<Float> poolingRatio, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("FractionalMaxPool", scope.makeOpName("FractionalMaxPool"));
-    opBuilder.addInput(value.asOutput());
+    opBuilder.addInput(value.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     float[] poolingRatioArray = new float[poolingRatio.size()];
     for (int i = 0; i < poolingRatioArray.length; ++i) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/FractionalMaxPoolGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/FractionalMaxPoolGrad.java
index ff3c0159747..3a9be5c7cd2 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/FractionalMaxPoolGrad.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/FractionalMaxPoolGrad.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -34,7 +33,7 @@
  * 
  * @param <T> data type for {@code output()} output
  */
-public final class FractionalMaxPoolGrad<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class FractionalMaxPoolGrad<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.nn.FractionalMaxPoolGrad}
@@ -79,13 +78,13 @@ private Options() {
    * @return a new instance of FractionalMaxPoolGrad
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> FractionalMaxPoolGrad<T> create(Scope scope, Operand<T> origInput, Operand<T> origOutput, Operand<T> outBackprop, Operand<TInt64> rowPoolingSequence, Operand<TInt64> colPoolingSequence, Options... options) {
+  public static <T extends TNumber> FractionalMaxPoolGrad<T> create(Scope scope, Operand<T> origInput, Operand<T> origOutput, Operand<T> outBackprop, Operand<TInt64> rowPoolingSequence, Operand<TInt64> colPoolingSequence, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("FractionalMaxPoolGrad", scope.makeOpName("FractionalMaxPoolGrad"));
-    opBuilder.addInput(origInput.asOutput());
-    opBuilder.addInput(origOutput.asOutput());
-    opBuilder.addInput(outBackprop.asOutput());
-    opBuilder.addInput(rowPoolingSequence.asOutput());
-    opBuilder.addInput(colPoolingSequence.asOutput());
+    opBuilder.addInput(origInput.asOutput(scope));
+    opBuilder.addInput(origOutput.asOutput(scope));
+    opBuilder.addInput(outBackprop.asOutput(scope));
+    opBuilder.addInput(rowPoolingSequence.asOutput(scope));
+    opBuilder.addInput(colPoolingSequence.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -120,7 +119,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/FusedBatchNorm.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/FusedBatchNorm.java
index 9d7ff535c82..1e61f4e0060 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/FusedBatchNorm.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/FusedBatchNorm.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -38,7 +37,7 @@
  * @param <U> data type for {@code batchMean()} output
  */
 @Operator(group = "nn")
-public final class FusedBatchNorm<T extends Tensor & TNumber, U extends Tensor & TNumber> extends RawOp {
+public final class FusedBatchNorm<T extends TNumber, U extends TNumber> extends RawOp {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.nn.FusedBatchNorm}
@@ -102,13 +101,13 @@ private Options() {
    * @return a new instance of FusedBatchNorm
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber, U extends Tensor & TNumber> FusedBatchNorm<T, U> create(Scope scope, Operand<T> x, Operand<U> scale, Operand<U> offset, Operand<U> mean, Operand<U> variance, Options... options) {
+  public static <T extends TNumber, U extends TNumber> FusedBatchNorm<T, U> create(Scope scope, Operand<T> x, Operand<U> scale, Operand<U> offset, Operand<U> mean, Operand<U> variance, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("FusedBatchNormV3", scope.makeOpName("FusedBatchNorm"));
-    opBuilder.addInput(x.asOutput());
-    opBuilder.addInput(scale.asOutput());
-    opBuilder.addInput(offset.asOutput());
-    opBuilder.addInput(mean.asOutput());
-    opBuilder.addInput(variance.asOutput());
+    opBuilder.addInput(x.asOutput(scope));
+    opBuilder.addInput(scale.asOutput(scope));
+    opBuilder.addInput(offset.asOutput(scope));
+    opBuilder.addInput(mean.asOutput(scope));
+    opBuilder.addInput(variance.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/FusedBatchNormGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/FusedBatchNormGrad.java
index 9059d890c06..abeb9214bab 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/FusedBatchNormGrad.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/FusedBatchNormGrad.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -39,7 +38,7 @@
  * @param <U> data type for {@code scaleBackprop()} output
  */
 @Operator(group = "nn")
-public final class FusedBatchNormGrad<T extends Tensor & TNumber, U extends Tensor & TNumber> extends RawOp {
+public final class FusedBatchNormGrad<T extends TNumber, U extends TNumber> extends RawOp {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.nn.FusedBatchNormGrad}
@@ -103,14 +102,14 @@ private Options() {
    * @return a new instance of FusedBatchNormGrad
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber, U extends Tensor & TNumber> FusedBatchNormGrad<T, U> create(Scope scope, Operand<T> yBackprop, Operand<T> x, Operand<TFloat32> scale, Operand<U> reserveSpace1, Operand<U> reserveSpace2, Operand<U> reserveSpace3, Options... options) {
+  public static <T extends TNumber, U extends TNumber> FusedBatchNormGrad<T, U> create(Scope scope, Operand<T> yBackprop, Operand<T> x, Operand<TFloat32> scale, Operand<U> reserveSpace1, Operand<U> reserveSpace2, Operand<U> reserveSpace3, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("FusedBatchNormGradV3", scope.makeOpName("FusedBatchNormGrad"));
-    opBuilder.addInput(yBackprop.asOutput());
-    opBuilder.addInput(x.asOutput());
-    opBuilder.addInput(scale.asOutput());
-    opBuilder.addInput(reserveSpace1.asOutput());
-    opBuilder.addInput(reserveSpace2.asOutput());
-    opBuilder.addInput(reserveSpace3.asOutput());
+    opBuilder.addInput(yBackprop.asOutput(scope));
+    opBuilder.addInput(x.asOutput(scope));
+    opBuilder.addInput(scale.asOutput(scope));
+    opBuilder.addInput(reserveSpace1.asOutput(scope));
+    opBuilder.addInput(reserveSpace2.asOutput(scope));
+    opBuilder.addInput(reserveSpace3.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/FusedPadConv2d.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/FusedPadConv2d.java
index 0d2d7367762..2bd86f38e37 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/FusedPadConv2d.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/FusedPadConv2d.java
@@ -22,7 +22,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -48,7 +47,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "nn")
-public final class FusedPadConv2d<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class FusedPadConv2d<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new FusedPadConv2d operation.
@@ -66,11 +65,11 @@ public final class FusedPadConv2d<T extends Tensor & TNumber> extends RawOp impl
    * @return a new instance of FusedPadConv2d
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> FusedPadConv2d<T> create(Scope scope, Operand<T> input, Operand<TInt32> paddings, Operand<T> filter, String mode, List<Long> strides, String padding) {
+  public static <T extends TNumber> FusedPadConv2d<T> create(Scope scope, Operand<T> input, Operand<TInt32> paddings, Operand<T> filter, String mode, List<Long> strides, String padding) {
     OperationBuilder opBuilder = scope.env().opBuilder("FusedPadConv2D", scope.makeOpName("FusedPadConv2d"));
-    opBuilder.addInput(input.asOutput());
-    opBuilder.addInput(paddings.asOutput());
-    opBuilder.addInput(filter.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
+    opBuilder.addInput(paddings.asOutput(scope));
+    opBuilder.addInput(filter.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("mode", mode);
     long[] stridesArray = new long[strides.size()];
@@ -89,7 +88,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/FusedResizeAndPadConv2d.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/FusedResizeAndPadConv2d.java
index cd4dbc00464..e4f8fa54396 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/FusedResizeAndPadConv2d.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/FusedResizeAndPadConv2d.java
@@ -22,7 +22,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -47,7 +46,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "nn")
-public final class FusedResizeAndPadConv2d<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class FusedResizeAndPadConv2d<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.nn.FusedResizeAndPadConv2d}
@@ -88,12 +87,12 @@ private Options() {
    * @return a new instance of FusedResizeAndPadConv2d
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> FusedResizeAndPadConv2d<T> create(Scope scope, Operand<T> input, Operand<TInt32> size, Operand<TInt32> paddings, Operand<T> filter, String mode, List<Long> strides, String padding, Options... options) {
+  public static <T extends TNumber> FusedResizeAndPadConv2d<T> create(Scope scope, Operand<T> input, Operand<TInt32> size, Operand<TInt32> paddings, Operand<T> filter, String mode, List<Long> strides, String padding, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("FusedResizeAndPadConv2D", scope.makeOpName("FusedResizeAndPadConv2d"));
-    opBuilder.addInput(input.asOutput());
-    opBuilder.addInput(size.asOutput());
-    opBuilder.addInput(paddings.asOutput());
-    opBuilder.addInput(filter.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
+    opBuilder.addInput(size.asOutput(scope));
+    opBuilder.addInput(paddings.asOutput(scope));
+    opBuilder.addInput(filter.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("mode", mode);
     long[] stridesArray = new long[strides.size()];
@@ -127,7 +126,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/GRUBlockCell.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/GRUBlockCell.java
index 481916e5afe..c51e1db1bdd 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/GRUBlockCell.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/GRUBlockCell.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -78,7 +77,7 @@
  * 
  * @param <T> data type for {@code r()} output
  */
-public final class GRUBlockCell<T extends Tensor & TNumber> extends RawOp {
+public final class GRUBlockCell<T extends TNumber> extends RawOp {
   
   /**
    * Factory method to create a class wrapping a new GRUBlockCell operation.
@@ -93,14 +92,14 @@ public final class GRUBlockCell<T extends Tensor & TNumber> extends RawOp {
    * @return a new instance of GRUBlockCell
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> GRUBlockCell<T> create(Scope scope, Operand<T> x, Operand<T> hPrev, Operand<T> wRu, Operand<T> wC, Operand<T> bRu, Operand<T> bC) {
+  public static <T extends TNumber> GRUBlockCell<T> create(Scope scope, Operand<T> x, Operand<T> hPrev, Operand<T> wRu, Operand<T> wC, Operand<T> bRu, Operand<T> bC) {
     OperationBuilder opBuilder = scope.env().opBuilder("GRUBlockCell", scope.makeOpName("GRUBlockCell"));
-    opBuilder.addInput(x.asOutput());
-    opBuilder.addInput(hPrev.asOutput());
-    opBuilder.addInput(wRu.asOutput());
-    opBuilder.addInput(wC.asOutput());
-    opBuilder.addInput(bRu.asOutput());
-    opBuilder.addInput(bC.asOutput());
+    opBuilder.addInput(x.asOutput(scope));
+    opBuilder.addInput(hPrev.asOutput(scope));
+    opBuilder.addInput(wRu.asOutput(scope));
+    opBuilder.addInput(wC.asOutput(scope));
+    opBuilder.addInput(bRu.asOutput(scope));
+    opBuilder.addInput(bC.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new GRUBlockCell<T>(opBuilder.build());
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/GRUBlockCellGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/GRUBlockCellGrad.java
index 274cdc6772b..1e93416bcc6 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/GRUBlockCellGrad.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/GRUBlockCellGrad.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -114,7 +113,7 @@
  * 
  * @param <T> data type for {@code dX()} output
  */
-public final class GRUBlockCellGrad<T extends Tensor & TNumber> extends RawOp {
+public final class GRUBlockCellGrad<T extends TNumber> extends RawOp {
   
   /**
    * Factory method to create a class wrapping a new GRUBlockCellGrad operation.
@@ -133,18 +132,18 @@ public final class GRUBlockCellGrad<T extends Tensor & TNumber> extends RawOp {
    * @return a new instance of GRUBlockCellGrad
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> GRUBlockCellGrad<T> create(Scope scope, Operand<T> x, Operand<T> hPrev, Operand<T> wRu, Operand<T> wC, Operand<T> bRu, Operand<T> bC, Operand<T> r, Operand<T> u, Operand<T> c, Operand<T> dH) {
+  public static <T extends TNumber> GRUBlockCellGrad<T> create(Scope scope, Operand<T> x, Operand<T> hPrev, Operand<T> wRu, Operand<T> wC, Operand<T> bRu, Operand<T> bC, Operand<T> r, Operand<T> u, Operand<T> c, Operand<T> dH) {
     OperationBuilder opBuilder = scope.env().opBuilder("GRUBlockCellGrad", scope.makeOpName("GRUBlockCellGrad"));
-    opBuilder.addInput(x.asOutput());
-    opBuilder.addInput(hPrev.asOutput());
-    opBuilder.addInput(wRu.asOutput());
-    opBuilder.addInput(wC.asOutput());
-    opBuilder.addInput(bRu.asOutput());
-    opBuilder.addInput(bC.asOutput());
-    opBuilder.addInput(r.asOutput());
-    opBuilder.addInput(u.asOutput());
-    opBuilder.addInput(c.asOutput());
-    opBuilder.addInput(dH.asOutput());
+    opBuilder.addInput(x.asOutput(scope));
+    opBuilder.addInput(hPrev.asOutput(scope));
+    opBuilder.addInput(wRu.asOutput(scope));
+    opBuilder.addInput(wC.asOutput(scope));
+    opBuilder.addInput(bRu.asOutput(scope));
+    opBuilder.addInput(bC.asOutput(scope));
+    opBuilder.addInput(r.asOutput(scope));
+    opBuilder.addInput(u.asOutput(scope));
+    opBuilder.addInput(c.asOutput(scope));
+    opBuilder.addInput(dH.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new GRUBlockCellGrad<T>(opBuilder.build());
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/InTopK.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/InTopK.java
index 869315541c9..28a53b5b13f 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/InTopK.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/InTopK.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -61,11 +60,11 @@ public final class InTopK extends RawOp implements Operand<TBool> {
    * @return a new instance of InTopK
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> InTopK create(Scope scope, Operand<TFloat32> predictions, Operand<T> targets, Operand<T> k) {
+  public static <T extends TNumber> InTopK create(Scope scope, Operand<TFloat32> predictions, Operand<T> targets, Operand<T> k) {
     OperationBuilder opBuilder = scope.env().opBuilder("InTopKV2", scope.makeOpName("InTopK"));
-    opBuilder.addInput(predictions.asOutput());
-    opBuilder.addInput(targets.asOutput());
-    opBuilder.addInput(k.asOutput());
+    opBuilder.addInput(predictions.asOutput(scope));
+    opBuilder.addInput(targets.asOutput(scope));
+    opBuilder.addInput(k.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new InTopK(opBuilder.build());
   }
@@ -78,7 +77,7 @@ public Output<TBool> precision() {
   }
   
   @Override
-  public Output<TBool> asOutput() {
+  public Output<TBool> asOutput(Scope scope) {
     return precision;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/InvGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/InvGrad.java
index c582161f771..c1143025c38 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/InvGrad.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/InvGrad.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Computes the gradient for the inverse of `x` wrt its input.
@@ -35,7 +35,7 @@
  * 
  * @param <T> data type for {@code z()} output
  */
-public final class InvGrad<T extends Tensor> extends RawOp implements Operand<T> {
+public final class InvGrad<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new InvGrad operation.
@@ -46,10 +46,10 @@ public final class InvGrad<T extends Tensor> extends RawOp implements Operand<T>
    * @return a new instance of InvGrad
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> InvGrad<T> create(Scope scope, Operand<T> y, Operand<T> dy) {
+  public static <T extends TType> InvGrad<T> create(Scope scope, Operand<T> y, Operand<T> dy) {
     OperationBuilder opBuilder = scope.env().opBuilder("InvGrad", scope.makeOpName("InvGrad"));
-    opBuilder.addInput(y.asOutput());
-    opBuilder.addInput(dy.asOutput());
+    opBuilder.addInput(y.asOutput(scope));
+    opBuilder.addInput(dy.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new InvGrad<T>(opBuilder.build());
   }
@@ -61,7 +61,7 @@ public Output<T> z() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return z;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/L2Loss.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/L2Loss.java
index 02f1771815c..2c72e84e897 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/L2Loss.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/L2Loss.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -38,7 +37,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "nn")
-public final class L2Loss<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class L2Loss<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new L2Loss operation.
@@ -48,9 +47,9 @@ public final class L2Loss<T extends Tensor & TNumber> extends RawOp implements O
    * @return a new instance of L2Loss
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> L2Loss<T> create(Scope scope, Operand<T> t) {
+  public static <T extends TNumber> L2Loss<T> create(Scope scope, Operand<T> t) {
     OperationBuilder opBuilder = scope.env().opBuilder("L2Loss", scope.makeOpName("L2Loss"));
-    opBuilder.addInput(t.asOutput());
+    opBuilder.addInput(t.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new L2Loss<T>(opBuilder.build());
   }
@@ -63,7 +62,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/LSTMBlockCell.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/LSTMBlockCell.java
index ac4cc8ecffc..d733379eac1 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/LSTMBlockCell.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/LSTMBlockCell.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -58,7 +57,7 @@
  * 
  * @param <T> data type for {@code i()} output
  */
-public final class LSTMBlockCell<T extends Tensor & TNumber> extends RawOp {
+public final class LSTMBlockCell<T extends TNumber> extends RawOp {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.nn.LSTMBlockCell}
@@ -113,16 +112,16 @@ private Options() {
    * @return a new instance of LSTMBlockCell
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> LSTMBlockCell<T> create(Scope scope, Operand<T> x, Operand<T> csPrev, Operand<T> hPrev, Operand<T> w, Operand<T> wci, Operand<T> wcf, Operand<T> wco, Operand<T> b, Options... options) {
+  public static <T extends TNumber> LSTMBlockCell<T> create(Scope scope, Operand<T> x, Operand<T> csPrev, Operand<T> hPrev, Operand<T> w, Operand<T> wci, Operand<T> wcf, Operand<T> wco, Operand<T> b, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("LSTMBlockCell", scope.makeOpName("LSTMBlockCell"));
-    opBuilder.addInput(x.asOutput());
-    opBuilder.addInput(csPrev.asOutput());
-    opBuilder.addInput(hPrev.asOutput());
-    opBuilder.addInput(w.asOutput());
-    opBuilder.addInput(wci.asOutput());
-    opBuilder.addInput(wcf.asOutput());
-    opBuilder.addInput(wco.asOutput());
-    opBuilder.addInput(b.asOutput());
+    opBuilder.addInput(x.asOutput(scope));
+    opBuilder.addInput(csPrev.asOutput(scope));
+    opBuilder.addInput(hPrev.asOutput(scope));
+    opBuilder.addInput(w.asOutput(scope));
+    opBuilder.addInput(wci.asOutput(scope));
+    opBuilder.addInput(wcf.asOutput(scope));
+    opBuilder.addInput(wco.asOutput(scope));
+    opBuilder.addInput(b.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/LSTMBlockCellGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/LSTMBlockCellGrad.java
index 20541ebe8e8..e515f99e9e9 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/LSTMBlockCellGrad.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/LSTMBlockCellGrad.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -35,7 +34,7 @@
  * 
  * @param <T> data type for {@code csPrevGrad()} output
  */
-public final class LSTMBlockCellGrad<T extends Tensor & TNumber> extends RawOp {
+public final class LSTMBlockCellGrad<T extends TNumber> extends RawOp {
   
   /**
    * Factory method to create a class wrapping a new LSTMBlockCellGrad operation.
@@ -61,24 +60,24 @@ public final class LSTMBlockCellGrad<T extends Tensor & TNumber> extends RawOp {
    * @return a new instance of LSTMBlockCellGrad
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> LSTMBlockCellGrad<T> create(Scope scope, Operand<T> x, Operand<T> csPrev, Operand<T> hPrev, Operand<T> w, Operand<T> wci, Operand<T> wcf, Operand<T> wco, Operand<T> b, Operand<T> i, Operand<T> cs, Operand<T> f, Operand<T> o, Operand<T> ci, Operand<T> co, Operand<T> csGrad, Operand<T> hGrad, Boolean usePeephole) {
+  public static <T extends TNumber> LSTMBlockCellGrad<T> create(Scope scope, Operand<T> x, Operand<T> csPrev, Operand<T> hPrev, Operand<T> w, Operand<T> wci, Operand<T> wcf, Operand<T> wco, Operand<T> b, Operand<T> i, Operand<T> cs, Operand<T> f, Operand<T> o, Operand<T> ci, Operand<T> co, Operand<T> csGrad, Operand<T> hGrad, Boolean usePeephole) {
     OperationBuilder opBuilder = scope.env().opBuilder("LSTMBlockCellGrad", scope.makeOpName("LSTMBlockCellGrad"));
-    opBuilder.addInput(x.asOutput());
-    opBuilder.addInput(csPrev.asOutput());
-    opBuilder.addInput(hPrev.asOutput());
-    opBuilder.addInput(w.asOutput());
-    opBuilder.addInput(wci.asOutput());
-    opBuilder.addInput(wcf.asOutput());
-    opBuilder.addInput(wco.asOutput());
-    opBuilder.addInput(b.asOutput());
-    opBuilder.addInput(i.asOutput());
-    opBuilder.addInput(cs.asOutput());
-    opBuilder.addInput(f.asOutput());
-    opBuilder.addInput(o.asOutput());
-    opBuilder.addInput(ci.asOutput());
-    opBuilder.addInput(co.asOutput());
-    opBuilder.addInput(csGrad.asOutput());
-    opBuilder.addInput(hGrad.asOutput());
+    opBuilder.addInput(x.asOutput(scope));
+    opBuilder.addInput(csPrev.asOutput(scope));
+    opBuilder.addInput(hPrev.asOutput(scope));
+    opBuilder.addInput(w.asOutput(scope));
+    opBuilder.addInput(wci.asOutput(scope));
+    opBuilder.addInput(wcf.asOutput(scope));
+    opBuilder.addInput(wco.asOutput(scope));
+    opBuilder.addInput(b.asOutput(scope));
+    opBuilder.addInput(i.asOutput(scope));
+    opBuilder.addInput(cs.asOutput(scope));
+    opBuilder.addInput(f.asOutput(scope));
+    opBuilder.addInput(o.asOutput(scope));
+    opBuilder.addInput(ci.asOutput(scope));
+    opBuilder.addInput(co.asOutput(scope));
+    opBuilder.addInput(csGrad.asOutput(scope));
+    opBuilder.addInput(hGrad.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("use_peephole", usePeephole);
     return new LSTMBlockCellGrad<T>(opBuilder.build());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/LeakyRelu.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/LeakyRelu.java
index c64340ab6b4..8db6282d4c9 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/LeakyRelu.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/LeakyRelu.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -34,7 +33,7 @@
  * @param <T> data type for {@code activations()} output
  */
 @Operator(group = "nn")
-public final class LeakyRelu<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class LeakyRelu<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.nn.LeakyRelu}
@@ -64,9 +63,9 @@ private Options() {
    * @return a new instance of LeakyRelu
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> LeakyRelu<T> create(Scope scope, Operand<T> features, Options... options) {
+  public static <T extends TNumber> LeakyRelu<T> create(Scope scope, Operand<T> features, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("LeakyRelu", scope.makeOpName("LeakyRelu"));
-    opBuilder.addInput(features.asOutput());
+    opBuilder.addInput(features.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -92,7 +91,7 @@ public Output<T> activations() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return activations;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/LearnedUnigramCandidateSampler.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/LearnedUnigramCandidateSampler.java
index 3f850ce672c..668c1cd3b1e 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/LearnedUnigramCandidateSampler.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/LearnedUnigramCandidateSampler.java
@@ -92,7 +92,7 @@ private Options() {
   @Endpoint(describeByClass = true)
   public static LearnedUnigramCandidateSampler create(Scope scope, Operand<TInt64> trueClasses, Long numTrue, Long numSampled, Boolean unique, Long rangeMax, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("LearnedUnigramCandidateSampler", scope.makeOpName("LearnedUnigramCandidateSampler"));
-    opBuilder.addInput(trueClasses.asOutput());
+    opBuilder.addInput(trueClasses.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("num_true", numTrue);
     opBuilder.setAttr("num_sampled", numSampled);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/LocalResponseNormalization.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/LocalResponseNormalization.java
index d04ddea39ce..9347b69f656 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/LocalResponseNormalization.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/LocalResponseNormalization.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -46,7 +45,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "nn")
-public final class LocalResponseNormalization<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class LocalResponseNormalization<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.nn.LocalResponseNormalization}
@@ -103,9 +102,9 @@ private Options() {
    * @return a new instance of LocalResponseNormalization
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> LocalResponseNormalization<T> create(Scope scope, Operand<T> input, Options... options) {
+  public static <T extends TNumber> LocalResponseNormalization<T> create(Scope scope, Operand<T> input, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("LRN", scope.makeOpName("LocalResponseNormalization"));
-    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -161,7 +160,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/LocalResponseNormalizationGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/LocalResponseNormalizationGrad.java
index fd9fd870829..086ea52f53e 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/LocalResponseNormalizationGrad.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/LocalResponseNormalizationGrad.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -33,7 +32,7 @@
  * 
  * @param <T> data type for {@code output()} output
  */
-public final class LocalResponseNormalizationGrad<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class LocalResponseNormalizationGrad<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.nn.LocalResponseNormalizationGrad}
@@ -92,11 +91,11 @@ private Options() {
    * @return a new instance of LocalResponseNormalizationGrad
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> LocalResponseNormalizationGrad<T> create(Scope scope, Operand<T> inputGrads, Operand<T> inputImage, Operand<T> outputImage, Options... options) {
+  public static <T extends TNumber> LocalResponseNormalizationGrad<T> create(Scope scope, Operand<T> inputGrads, Operand<T> inputImage, Operand<T> outputImage, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("LRNGrad", scope.makeOpName("LocalResponseNormalizationGrad"));
-    opBuilder.addInput(inputGrads.asOutput());
-    opBuilder.addInput(inputImage.asOutput());
-    opBuilder.addInput(outputImage.asOutput());
+    opBuilder.addInput(inputGrads.asOutput(scope));
+    opBuilder.addInput(inputImage.asOutput(scope));
+    opBuilder.addInput(outputImage.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -153,7 +152,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/LogSoftmax.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/LogSoftmax.java
index 80d8a2346bc..dd035488742 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/LogSoftmax.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/LogSoftmax.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -38,7 +37,7 @@
  * @param <T> data type for {@code logsoftmax()} output
  */
 @Operator(group = "nn")
-public final class LogSoftmax<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class LogSoftmax<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new LogSoftmax operation.
@@ -48,9 +47,9 @@ public final class LogSoftmax<T extends Tensor & TNumber> extends RawOp implemen
    * @return a new instance of LogSoftmax
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> LogSoftmax<T> create(Scope scope, Operand<T> logits) {
+  public static <T extends TNumber> LogSoftmax<T> create(Scope scope, Operand<T> logits) {
     OperationBuilder opBuilder = scope.env().opBuilder("LogSoftmax", scope.makeOpName("LogSoftmax"));
-    opBuilder.addInput(logits.asOutput());
+    opBuilder.addInput(logits.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new LogSoftmax<T>(opBuilder.build());
   }
@@ -63,7 +62,7 @@ public Output<T> logsoftmax() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return logsoftmax;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPool.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPool.java
index 90e952e747e..c2e5977e1c7 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPool.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPool.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TType;
 
 /**
  * Performs max pooling on the input.
@@ -34,7 +34,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "nn")
-public final class MaxPool<T extends Tensor> extends RawOp implements Operand<T> {
+public final class MaxPool<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.nn.MaxPool}
@@ -72,11 +72,11 @@ private Options() {
    * @return a new instance of MaxPool
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> MaxPool<T> create(Scope scope, Operand<T> input, Operand<TInt32> ksize, Operand<TInt32> strides, String padding, Options... options) {
+  public static <T extends TType> MaxPool<T> create(Scope scope, Operand<T> input, Operand<TInt32> ksize, Operand<TInt32> strides, String padding, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("MaxPoolV2", scope.makeOpName("MaxPool"));
-    opBuilder.addInput(input.asOutput());
-    opBuilder.addInput(ksize.asOutput());
-    opBuilder.addInput(strides.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
+    opBuilder.addInput(ksize.asOutput(scope));
+    opBuilder.addInput(strides.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("padding", padding);
     if (options != null) {
@@ -108,7 +108,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPool3d.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPool3d.java
index b6ac014641f..72a8241fd93 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPool3d.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPool3d.java
@@ -22,7 +22,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -35,7 +34,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "nn")
-public final class MaxPool3d<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class MaxPool3d<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.nn.MaxPool3d}
@@ -74,9 +73,9 @@ private Options() {
    * @return a new instance of MaxPool3d
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> MaxPool3d<T> create(Scope scope, Operand<T> input, List<Long> ksize, List<Long> strides, String padding, Options... options) {
+  public static <T extends TNumber> MaxPool3d<T> create(Scope scope, Operand<T> input, List<Long> ksize, List<Long> strides, String padding, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("MaxPool3D", scope.makeOpName("MaxPool3d"));
-    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     long[] ksizeArray = new long[ksize.size()];
     for (int i = 0; i < ksizeArray.length; ++i) {
@@ -118,7 +117,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPool3dGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPool3dGrad.java
index 44d2cb4b276..3fd6f901792 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPool3dGrad.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPool3dGrad.java
@@ -22,7 +22,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -35,7 +34,7 @@
  * @param <U> data type for {@code output()} output
  */
 @Operator(group = "nn")
-public final class MaxPool3dGrad<U extends Tensor & TNumber> extends RawOp implements Operand<U> {
+public final class MaxPool3dGrad<U extends TNumber> extends RawOp implements Operand<U> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.nn.MaxPool3dGrad}
@@ -76,11 +75,11 @@ private Options() {
    * @return a new instance of MaxPool3dGrad
    */
   @Endpoint(describeByClass = true)
-  public static <U extends Tensor & TNumber, T extends Tensor & TNumber> MaxPool3dGrad<U> create(Scope scope, Operand<T> origInput, Operand<T> origOutput, Operand<U> grad, List<Long> ksize, List<Long> strides, String padding, Options... options) {
+  public static <U extends TNumber, T extends TNumber> MaxPool3dGrad<U> create(Scope scope, Operand<T> origInput, Operand<T> origOutput, Operand<U> grad, List<Long> ksize, List<Long> strides, String padding, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("MaxPool3DGrad", scope.makeOpName("MaxPool3dGrad"));
-    opBuilder.addInput(origInput.asOutput());
-    opBuilder.addInput(origOutput.asOutput());
-    opBuilder.addInput(grad.asOutput());
+    opBuilder.addInput(origInput.asOutput(scope));
+    opBuilder.addInput(origOutput.asOutput(scope));
+    opBuilder.addInput(grad.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     long[] ksizeArray = new long[ksize.size()];
     for (int i = 0; i < ksizeArray.length; ++i) {
@@ -121,7 +120,7 @@ public Output<U> output() {
   }
   
   @Override
-  public Output<U> asOutput() {
+  public Output<U> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPool3dGradGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPool3dGradGrad.java
index c17d23daa27..e9ccd76db0d 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPool3dGradGrad.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPool3dGradGrad.java
@@ -22,7 +22,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -35,7 +34,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "nn")
-public final class MaxPool3dGradGrad<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class MaxPool3dGradGrad<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.nn.MaxPool3dGradGrad}
@@ -76,11 +75,11 @@ private Options() {
    * @return a new instance of MaxPool3dGradGrad
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> MaxPool3dGradGrad<T> create(Scope scope, Operand<T> origInput, Operand<T> origOutput, Operand<T> grad, List<Long> ksize, List<Long> strides, String padding, Options... options) {
+  public static <T extends TNumber> MaxPool3dGradGrad<T> create(Scope scope, Operand<T> origInput, Operand<T> origOutput, Operand<T> grad, List<Long> ksize, List<Long> strides, String padding, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("MaxPool3DGradGrad", scope.makeOpName("MaxPool3dGradGrad"));
-    opBuilder.addInput(origInput.asOutput());
-    opBuilder.addInput(origOutput.asOutput());
-    opBuilder.addInput(grad.asOutput());
+    opBuilder.addInput(origInput.asOutput(scope));
+    opBuilder.addInput(origOutput.asOutput(scope));
+    opBuilder.addInput(grad.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     long[] ksizeArray = new long[ksize.size()];
     for (int i = 0; i < ksizeArray.length; ++i) {
@@ -122,7 +121,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPoolGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPoolGrad.java
index beea4fcdee5..c1951a034f9 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPoolGrad.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPoolGrad.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -35,7 +34,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "nn")
-public final class MaxPoolGrad<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class MaxPoolGrad<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.nn.MaxPoolGrad}
@@ -75,13 +74,13 @@ private Options() {
    * @return a new instance of MaxPoolGrad
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> MaxPoolGrad<T> create(Scope scope, Operand<T> origInput, Operand<T> origOutput, Operand<T> grad, Operand<TInt32> ksize, Operand<TInt32> strides, String padding, Options... options) {
+  public static <T extends TNumber> MaxPoolGrad<T> create(Scope scope, Operand<T> origInput, Operand<T> origOutput, Operand<T> grad, Operand<TInt32> ksize, Operand<TInt32> strides, String padding, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("MaxPoolGradV2", scope.makeOpName("MaxPoolGrad"));
-    opBuilder.addInput(origInput.asOutput());
-    opBuilder.addInput(origOutput.asOutput());
-    opBuilder.addInput(grad.asOutput());
-    opBuilder.addInput(ksize.asOutput());
-    opBuilder.addInput(strides.asOutput());
+    opBuilder.addInput(origInput.asOutput(scope));
+    opBuilder.addInput(origOutput.asOutput(scope));
+    opBuilder.addInput(grad.asOutput(scope));
+    opBuilder.addInput(ksize.asOutput(scope));
+    opBuilder.addInput(strides.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("padding", padding);
     if (options != null) {
@@ -113,7 +112,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPoolGradGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPoolGradGrad.java
index 93a2b142125..3810b6845a5 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPoolGradGrad.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPoolGradGrad.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -35,7 +34,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "nn")
-public final class MaxPoolGradGrad<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class MaxPoolGradGrad<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.nn.MaxPoolGradGrad}
@@ -75,13 +74,13 @@ private Options() {
    * @return a new instance of MaxPoolGradGrad
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> MaxPoolGradGrad<T> create(Scope scope, Operand<T> origInput, Operand<T> origOutput, Operand<T> grad, Operand<TInt32> ksize, Operand<TInt32> strides, String padding, Options... options) {
+  public static <T extends TNumber> MaxPoolGradGrad<T> create(Scope scope, Operand<T> origInput, Operand<T> origOutput, Operand<T> grad, Operand<TInt32> ksize, Operand<TInt32> strides, String padding, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("MaxPoolGradGradV2", scope.makeOpName("MaxPoolGradGrad"));
-    opBuilder.addInput(origInput.asOutput());
-    opBuilder.addInput(origOutput.asOutput());
-    opBuilder.addInput(grad.asOutput());
-    opBuilder.addInput(ksize.asOutput());
-    opBuilder.addInput(strides.asOutput());
+    opBuilder.addInput(origInput.asOutput(scope));
+    opBuilder.addInput(origOutput.asOutput(scope));
+    opBuilder.addInput(grad.asOutput(scope));
+    opBuilder.addInput(ksize.asOutput(scope));
+    opBuilder.addInput(strides.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("padding", padding);
     if (options != null) {
@@ -113,7 +112,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPoolGradGradWithArgmax.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPoolGradGradWithArgmax.java
index d0a032189c4..8f22d65be65 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPoolGradGradWithArgmax.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPoolGradGradWithArgmax.java
@@ -22,7 +22,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -35,7 +34,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "nn")
-public final class MaxPoolGradGradWithArgmax<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class MaxPoolGradGradWithArgmax<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.nn.MaxPoolGradGradWithArgmax}
@@ -72,11 +71,11 @@ private Options() {
    * @return a new instance of MaxPoolGradGradWithArgmax
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber, U extends Tensor & TNumber> MaxPoolGradGradWithArgmax<T> create(Scope scope, Operand<T> input, Operand<T> grad, Operand<U> argmax, List<Long> ksize, List<Long> strides, String padding, Options... options) {
+  public static <T extends TNumber, U extends TNumber> MaxPoolGradGradWithArgmax<T> create(Scope scope, Operand<T> input, Operand<T> grad, Operand<U> argmax, List<Long> ksize, List<Long> strides, String padding, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("MaxPoolGradGradWithArgmax", scope.makeOpName("MaxPoolGradGradWithArgmax"));
-    opBuilder.addInput(input.asOutput());
-    opBuilder.addInput(grad.asOutput());
-    opBuilder.addInput(argmax.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
+    opBuilder.addInput(grad.asOutput(scope));
+    opBuilder.addInput(argmax.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     long[] ksizeArray = new long[ksize.size()];
     for (int i = 0; i < ksizeArray.length; ++i) {
@@ -114,7 +113,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPoolGradWithArgmax.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPoolGradWithArgmax.java
index 0235d171d74..551b3ef0594 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPoolGradWithArgmax.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPoolGradWithArgmax.java
@@ -22,7 +22,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -34,7 +33,7 @@
  * 
  * @param <T> data type for {@code output()} output
  */
-public final class MaxPoolGradWithArgmax<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class MaxPoolGradWithArgmax<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.nn.MaxPoolGradWithArgmax}
@@ -71,11 +70,11 @@ private Options() {
    * @return a new instance of MaxPoolGradWithArgmax
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber, U extends Tensor & TNumber> MaxPoolGradWithArgmax<T> create(Scope scope, Operand<T> input, Operand<T> grad, Operand<U> argmax, List<Long> ksize, List<Long> strides, String padding, Options... options) {
+  public static <T extends TNumber, U extends TNumber> MaxPoolGradWithArgmax<T> create(Scope scope, Operand<T> input, Operand<T> grad, Operand<U> argmax, List<Long> ksize, List<Long> strides, String padding, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("MaxPoolGradWithArgmax", scope.makeOpName("MaxPoolGradWithArgmax"));
-    opBuilder.addInput(input.asOutput());
-    opBuilder.addInput(grad.asOutput());
-    opBuilder.addInput(argmax.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
+    opBuilder.addInput(grad.asOutput(scope));
+    opBuilder.addInput(argmax.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     long[] ksizeArray = new long[ksize.size()];
     for (int i = 0; i < ksizeArray.length; ++i) {
@@ -113,7 +112,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPoolWithArgmax.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPoolWithArgmax.java
index 901fbf451df..8e5fbba6a69 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPoolWithArgmax.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPoolWithArgmax.java
@@ -23,7 +23,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -48,7 +47,7 @@
  * @param <U> data type for {@code argmax()} output
  */
 @Operator(group = "nn")
-public final class MaxPoolWithArgmax<T extends Tensor & TNumber, U extends Tensor & TNumber> extends RawOp {
+public final class MaxPoolWithArgmax<T extends TNumber, U extends TNumber> extends RawOp {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.nn.MaxPoolWithArgmax}
@@ -83,9 +82,9 @@ private Options() {
    * @return a new instance of MaxPoolWithArgmax
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber, U extends Tensor & TNumber> MaxPoolWithArgmax<T, U> create(Scope scope, Operand<T> input, List<Long> ksize, List<Long> strides, DataType<U> Targmax, String padding, Options... options) {
+  public static <T extends TNumber, U extends TNumber> MaxPoolWithArgmax<T, U> create(Scope scope, Operand<T> input, List<Long> ksize, List<Long> strides, DataType<U> Targmax, String padding, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("MaxPoolWithArgmax", scope.makeOpName("MaxPoolWithArgmax"));
-    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     long[] ksizeArray = new long[ksize.size()];
     for (int i = 0; i < ksizeArray.length; ++i) {
@@ -122,7 +121,7 @@ public static <T extends Tensor & TNumber, U extends Tensor & TNumber> MaxPoolWi
    * @return a new instance of MaxPoolWithArgmax
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> MaxPoolWithArgmax<T, TInt64> create(Scope scope, Operand<T> input, List<Long> ksize, List<Long> strides, String padding, Options... options) {
+  public static <T extends TNumber> MaxPoolWithArgmax<T, TInt64> create(Scope scope, Operand<T> input, List<Long> ksize, List<Long> strides, String padding, Options... options) {
     return create(scope, input, ksize, strides, TInt64.DTYPE, padding, options);
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/NthElement.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/NthElement.java
index 8decbf0e7ae..9f55dce0be7 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/NthElement.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/NthElement.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -43,7 +42,7 @@
  * @param <T> data type for {@code values()} output
  */
 @Operator(group = "nn")
-public final class NthElement<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class NthElement<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.nn.NthElement}
@@ -76,10 +75,10 @@ private Options() {
    * @return a new instance of NthElement
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> NthElement<T> create(Scope scope, Operand<T> input, Operand<TInt32> n, Options... options) {
+  public static <T extends TNumber> NthElement<T> create(Scope scope, Operand<T> input, Operand<TInt32> n, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("NthElement", scope.makeOpName("NthElement"));
-    opBuilder.addInput(input.asOutput());
-    opBuilder.addInput(n.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
+    opBuilder.addInput(n.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -107,7 +106,7 @@ public Output<T> values() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return values;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedAvgPool.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedAvgPool.java
index b521985edda..f573d560413 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedAvgPool.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedAvgPool.java
@@ -22,12 +22,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.family.TType;
 
 /**
  * Produces the average pool of the input tensor for quantized types.
@@ -35,7 +35,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "nn")
-public final class QuantizedAvgPool<T extends Tensor> extends RawOp {
+public final class QuantizedAvgPool<T extends TType> extends RawOp {
   
   /**
    * Factory method to create a class wrapping a new QuantizedAvgPool operation.
@@ -52,11 +52,11 @@ public final class QuantizedAvgPool<T extends Tensor> extends RawOp {
    * @return a new instance of QuantizedAvgPool
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> QuantizedAvgPool<T> create(Scope scope, Operand<T> input, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, List<Long> ksize, List<Long> strides, String padding) {
+  public static <T extends TType> QuantizedAvgPool<T> create(Scope scope, Operand<T> input, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, List<Long> ksize, List<Long> strides, String padding) {
     OperationBuilder opBuilder = scope.env().opBuilder("QuantizedAvgPool", scope.makeOpName("QuantizedAvgPool"));
-    opBuilder.addInput(input.asOutput());
-    opBuilder.addInput(minInput.asOutput());
-    opBuilder.addInput(maxInput.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
+    opBuilder.addInput(minInput.asOutput(scope));
+    opBuilder.addInput(maxInput.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     long[] ksizeArray = new long[ksize.size()];
     for (int i = 0; i < ksizeArray.length; ++i) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedBatchNormWithGlobalNormalization.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedBatchNormWithGlobalNormalization.java
index f2c9999482d..9c9e29fe78f 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedBatchNormWithGlobalNormalization.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedBatchNormWithGlobalNormalization.java
@@ -22,12 +22,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.family.TType;
 
 /**
  * Quantized Batch normalization.
@@ -38,7 +38,7 @@
  * @param <U> data type for {@code result()} output
  */
 @Operator(group = "nn")
-public final class QuantizedBatchNormWithGlobalNormalization<U extends Tensor> extends RawOp {
+public final class QuantizedBatchNormWithGlobalNormalization<U extends TType> extends RawOp {
   
   /**
    * Factory method to create a class wrapping a new QuantizedBatchNormWithGlobalNormalization operation.
@@ -73,23 +73,23 @@ public final class QuantizedBatchNormWithGlobalNormalization<U extends Tensor> e
    * @return a new instance of QuantizedBatchNormWithGlobalNormalization
    */
   @Endpoint(describeByClass = true)
-  public static <U extends Tensor, T extends Tensor> QuantizedBatchNormWithGlobalNormalization<U> create(Scope scope, Operand<T> t, Operand<TFloat32> tMin, Operand<TFloat32> tMax, Operand<T> m, Operand<TFloat32> mMin, Operand<TFloat32> mMax, Operand<T> v, Operand<TFloat32> vMin, Operand<TFloat32> vMax, Operand<T> beta, Operand<TFloat32> betaMin, Operand<TFloat32> betaMax, Operand<T> gamma, Operand<TFloat32> gammaMin, Operand<TFloat32> gammaMax, DataType<U> outType, Float varianceEpsilon, Boolean scaleAfterNormalization) {
+  public static <U extends TType, T extends TType> QuantizedBatchNormWithGlobalNormalization<U> create(Scope scope, Operand<T> t, Operand<TFloat32> tMin, Operand<TFloat32> tMax, Operand<T> m, Operand<TFloat32> mMin, Operand<TFloat32> mMax, Operand<T> v, Operand<TFloat32> vMin, Operand<TFloat32> vMax, Operand<T> beta, Operand<TFloat32> betaMin, Operand<TFloat32> betaMax, Operand<T> gamma, Operand<TFloat32> gammaMin, Operand<TFloat32> gammaMax, DataType<U> outType, Float varianceEpsilon, Boolean scaleAfterNormalization) {
     OperationBuilder opBuilder = scope.env().opBuilder("QuantizedBatchNormWithGlobalNormalization", scope.makeOpName("QuantizedBatchNormWithGlobalNormalization"));
-    opBuilder.addInput(t.asOutput());
-    opBuilder.addInput(tMin.asOutput());
-    opBuilder.addInput(tMax.asOutput());
-    opBuilder.addInput(m.asOutput());
-    opBuilder.addInput(mMin.asOutput());
-    opBuilder.addInput(mMax.asOutput());
-    opBuilder.addInput(v.asOutput());
-    opBuilder.addInput(vMin.asOutput());
-    opBuilder.addInput(vMax.asOutput());
-    opBuilder.addInput(beta.asOutput());
-    opBuilder.addInput(betaMin.asOutput());
-    opBuilder.addInput(betaMax.asOutput());
-    opBuilder.addInput(gamma.asOutput());
-    opBuilder.addInput(gammaMin.asOutput());
-    opBuilder.addInput(gammaMax.asOutput());
+    opBuilder.addInput(t.asOutput(scope));
+    opBuilder.addInput(tMin.asOutput(scope));
+    opBuilder.addInput(tMax.asOutput(scope));
+    opBuilder.addInput(m.asOutput(scope));
+    opBuilder.addInput(mMin.asOutput(scope));
+    opBuilder.addInput(mMax.asOutput(scope));
+    opBuilder.addInput(v.asOutput(scope));
+    opBuilder.addInput(vMin.asOutput(scope));
+    opBuilder.addInput(vMax.asOutput(scope));
+    opBuilder.addInput(beta.asOutput(scope));
+    opBuilder.addInput(betaMin.asOutput(scope));
+    opBuilder.addInput(betaMax.asOutput(scope));
+    opBuilder.addInput(gamma.asOutput(scope));
+    opBuilder.addInput(gammaMin.asOutput(scope));
+    opBuilder.addInput(gammaMax.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("out_type", outType);
     opBuilder.setAttr("variance_epsilon", varianceEpsilon);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedBiasAdd.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedBiasAdd.java
index f8abac79370..a1e464190bb 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedBiasAdd.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedBiasAdd.java
@@ -22,12 +22,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.family.TType;
 
 /**
  * Adds Tensor 'bias' to Tensor 'input' for Quantized types.
@@ -37,7 +37,7 @@
  * @param <V> data type for {@code output()} output
  */
 @Operator(group = "nn")
-public final class QuantizedBiasAdd<V extends Tensor> extends RawOp {
+public final class QuantizedBiasAdd<V extends TType> extends RawOp {
   
   /**
    * Factory method to create a class wrapping a new QuantizedBiasAdd operation.
@@ -53,14 +53,14 @@ public final class QuantizedBiasAdd<V extends Tensor> extends RawOp {
    * @return a new instance of QuantizedBiasAdd
    */
   @Endpoint(describeByClass = true)
-  public static <V extends Tensor, T extends Tensor, U extends Tensor> QuantizedBiasAdd<V> create(Scope scope, Operand<T> input, Operand<U> bias, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, Operand<TFloat32> minBias, Operand<TFloat32> maxBias, DataType<V> outType) {
+  public static <V extends TType, T extends TType, U extends TType> QuantizedBiasAdd<V> create(Scope scope, Operand<T> input, Operand<U> bias, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, Operand<TFloat32> minBias, Operand<TFloat32> maxBias, DataType<V> outType) {
     OperationBuilder opBuilder = scope.env().opBuilder("QuantizedBiasAdd", scope.makeOpName("QuantizedBiasAdd"));
-    opBuilder.addInput(input.asOutput());
-    opBuilder.addInput(bias.asOutput());
-    opBuilder.addInput(minInput.asOutput());
-    opBuilder.addInput(maxInput.asOutput());
-    opBuilder.addInput(minBias.asOutput());
-    opBuilder.addInput(maxBias.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
+    opBuilder.addInput(bias.asOutput(scope));
+    opBuilder.addInput(minInput.asOutput(scope));
+    opBuilder.addInput(maxInput.asOutput(scope));
+    opBuilder.addInput(minBias.asOutput(scope));
+    opBuilder.addInput(maxBias.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("out_type", outType);
     return new QuantizedBiasAdd<V>(opBuilder.build());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DAndRelu.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DAndRelu.java
index 73ecbe49450..12f565c6a8a 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DAndRelu.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DAndRelu.java
@@ -23,17 +23,17 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.family.TType;
 
 /**
  * @param <V> data type for {@code output()} output
  */
-public final class QuantizedConv2DAndRelu<V extends Tensor> extends RawOp {
+public final class QuantizedConv2DAndRelu<V extends TType> extends RawOp {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.nn.QuantizedConv2DAndRelu}
@@ -80,14 +80,14 @@ private Options() {
    * @return a new instance of QuantizedConv2DAndRelu
    */
   @Endpoint(describeByClass = true)
-  public static <V extends Tensor, T extends Tensor, U extends Tensor> QuantizedConv2DAndRelu<V> create(Scope scope, Operand<T> input, Operand<U> filter, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, Operand<TFloat32> minFilter, Operand<TFloat32> maxFilter, DataType<V> outType, List<Long> strides, String padding, Options... options) {
+  public static <V extends TType, T extends TType, U extends TType> QuantizedConv2DAndRelu<V> create(Scope scope, Operand<T> input, Operand<U> filter, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, Operand<TFloat32> minFilter, Operand<TFloat32> maxFilter, DataType<V> outType, List<Long> strides, String padding, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("QuantizedConv2DAndRelu", scope.makeOpName("QuantizedConv2DAndRelu"));
-    opBuilder.addInput(input.asOutput());
-    opBuilder.addInput(filter.asOutput());
-    opBuilder.addInput(minInput.asOutput());
-    opBuilder.addInput(maxInput.asOutput());
-    opBuilder.addInput(minFilter.asOutput());
-    opBuilder.addInput(maxFilter.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
+    opBuilder.addInput(filter.asOutput(scope));
+    opBuilder.addInput(minInput.asOutput(scope));
+    opBuilder.addInput(maxInput.asOutput(scope));
+    opBuilder.addInput(minFilter.asOutput(scope));
+    opBuilder.addInput(maxFilter.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("out_type", outType);
     long[] stridesArray = new long[strides.size()];
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DAndReluAndRequantize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DAndReluAndRequantize.java
index 6d935c91097..a7fe326d647 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DAndReluAndRequantize.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DAndReluAndRequantize.java
@@ -23,17 +23,17 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.family.TType;
 
 /**
  * @param <V> data type for {@code output()} output
  */
-public final class QuantizedConv2DAndReluAndRequantize<V extends Tensor> extends RawOp {
+public final class QuantizedConv2DAndReluAndRequantize<V extends TType> extends RawOp {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.nn.QuantizedConv2DAndReluAndRequantize}
@@ -82,16 +82,16 @@ private Options() {
    * @return a new instance of QuantizedConv2DAndReluAndRequantize
    */
   @Endpoint(describeByClass = true)
-  public static <V extends Tensor, T extends Tensor, U extends Tensor> QuantizedConv2DAndReluAndRequantize<V> create(Scope scope, Operand<T> input, Operand<U> filter, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, Operand<TFloat32> minFilter, Operand<TFloat32> maxFilter, Operand<TFloat32> minFreezedOutput, Operand<TFloat32> maxFreezedOutput, DataType<V> outType, List<Long> strides, String padding, Options... options) {
+  public static <V extends TType, T extends TType, U extends TType> QuantizedConv2DAndReluAndRequantize<V> create(Scope scope, Operand<T> input, Operand<U> filter, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, Operand<TFloat32> minFilter, Operand<TFloat32> maxFilter, Operand<TFloat32> minFreezedOutput, Operand<TFloat32> maxFreezedOutput, DataType<V> outType, List<Long> strides, String padding, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("QuantizedConv2DAndReluAndRequantize", scope.makeOpName("QuantizedConv2DAndReluAndRequantize"));
-    opBuilder.addInput(input.asOutput());
-    opBuilder.addInput(filter.asOutput());
-    opBuilder.addInput(minInput.asOutput());
-    opBuilder.addInput(maxInput.asOutput());
-    opBuilder.addInput(minFilter.asOutput());
-    opBuilder.addInput(maxFilter.asOutput());
-    opBuilder.addInput(minFreezedOutput.asOutput());
-    opBuilder.addInput(maxFreezedOutput.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
+    opBuilder.addInput(filter.asOutput(scope));
+    opBuilder.addInput(minInput.asOutput(scope));
+    opBuilder.addInput(maxInput.asOutput(scope));
+    opBuilder.addInput(minFilter.asOutput(scope));
+    opBuilder.addInput(maxFilter.asOutput(scope));
+    opBuilder.addInput(minFreezedOutput.asOutput(scope));
+    opBuilder.addInput(maxFreezedOutput.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("out_type", outType);
     long[] stridesArray = new long[strides.size()];
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DAndRequantize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DAndRequantize.java
index 5661c43e321..6ec1e000d71 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DAndRequantize.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DAndRequantize.java
@@ -23,17 +23,17 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.family.TType;
 
 /**
  * @param <V> data type for {@code output()} output
  */
-public final class QuantizedConv2DAndRequantize<V extends Tensor> extends RawOp {
+public final class QuantizedConv2DAndRequantize<V extends TType> extends RawOp {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.nn.QuantizedConv2DAndRequantize}
@@ -82,16 +82,16 @@ private Options() {
    * @return a new instance of QuantizedConv2DAndRequantize
    */
   @Endpoint(describeByClass = true)
-  public static <V extends Tensor, T extends Tensor, U extends Tensor> QuantizedConv2DAndRequantize<V> create(Scope scope, Operand<T> input, Operand<U> filter, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, Operand<TFloat32> minFilter, Operand<TFloat32> maxFilter, Operand<TFloat32> minFreezedOutput, Operand<TFloat32> maxFreezedOutput, DataType<V> outType, List<Long> strides, String padding, Options... options) {
+  public static <V extends TType, T extends TType, U extends TType> QuantizedConv2DAndRequantize<V> create(Scope scope, Operand<T> input, Operand<U> filter, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, Operand<TFloat32> minFilter, Operand<TFloat32> maxFilter, Operand<TFloat32> minFreezedOutput, Operand<TFloat32> maxFreezedOutput, DataType<V> outType, List<Long> strides, String padding, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("QuantizedConv2DAndRequantize", scope.makeOpName("QuantizedConv2DAndRequantize"));
-    opBuilder.addInput(input.asOutput());
-    opBuilder.addInput(filter.asOutput());
-    opBuilder.addInput(minInput.asOutput());
-    opBuilder.addInput(maxInput.asOutput());
-    opBuilder.addInput(minFilter.asOutput());
-    opBuilder.addInput(maxFilter.asOutput());
-    opBuilder.addInput(minFreezedOutput.asOutput());
-    opBuilder.addInput(maxFreezedOutput.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
+    opBuilder.addInput(filter.asOutput(scope));
+    opBuilder.addInput(minInput.asOutput(scope));
+    opBuilder.addInput(maxInput.asOutput(scope));
+    opBuilder.addInput(minFilter.asOutput(scope));
+    opBuilder.addInput(maxFilter.asOutput(scope));
+    opBuilder.addInput(minFreezedOutput.asOutput(scope));
+    opBuilder.addInput(maxFreezedOutput.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("out_type", outType);
     long[] stridesArray = new long[strides.size()];
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DPerChannel.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DPerChannel.java
index 112e1f640a1..12a8159a1c4 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DPerChannel.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DPerChannel.java
@@ -23,19 +23,19 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.family.TType;
 
 /**
  * Computes QuantizedConv2D per channel.
  * 
  * @param <V> data type for {@code output()} output
  */
-public final class QuantizedConv2DPerChannel<V extends Tensor> extends RawOp {
+public final class QuantizedConv2DPerChannel<V extends TType> extends RawOp {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.nn.QuantizedConv2DPerChannel}
@@ -73,14 +73,14 @@ private Options() {
    * @return a new instance of QuantizedConv2DPerChannel
    */
   @Endpoint(describeByClass = true)
-  public static <V extends Tensor, T extends Tensor, U extends Tensor> QuantizedConv2DPerChannel<V> create(Scope scope, Operand<T> input, Operand<U> filter, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, Operand<TFloat32> minFilter, Operand<TFloat32> maxFilter, DataType<V> outType, List<Long> strides, String padding, Options... options) {
+  public static <V extends TType, T extends TType, U extends TType> QuantizedConv2DPerChannel<V> create(Scope scope, Operand<T> input, Operand<U> filter, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, Operand<TFloat32> minFilter, Operand<TFloat32> maxFilter, DataType<V> outType, List<Long> strides, String padding, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("QuantizedConv2DPerChannel", scope.makeOpName("QuantizedConv2DPerChannel"));
-    opBuilder.addInput(input.asOutput());
-    opBuilder.addInput(filter.asOutput());
-    opBuilder.addInput(minInput.asOutput());
-    opBuilder.addInput(maxInput.asOutput());
-    opBuilder.addInput(minFilter.asOutput());
-    opBuilder.addInput(maxFilter.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
+    opBuilder.addInput(filter.asOutput(scope));
+    opBuilder.addInput(minInput.asOutput(scope));
+    opBuilder.addInput(maxInput.asOutput(scope));
+    opBuilder.addInput(minFilter.asOutput(scope));
+    opBuilder.addInput(maxFilter.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("out_type", outType);
     long[] stridesArray = new long[strides.size()];
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DWithBias.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DWithBias.java
index 94649797e81..0e0e229f266 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DWithBias.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DWithBias.java
@@ -23,17 +23,17 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.family.TType;
 
 /**
  * @param <V> data type for {@code output()} output
  */
-public final class QuantizedConv2DWithBias<V extends Tensor> extends RawOp {
+public final class QuantizedConv2DWithBias<V extends TType> extends RawOp {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.nn.QuantizedConv2DWithBias}
@@ -81,15 +81,15 @@ private Options() {
    * @return a new instance of QuantizedConv2DWithBias
    */
   @Endpoint(describeByClass = true)
-  public static <V extends Tensor, T extends Tensor, U extends Tensor> QuantizedConv2DWithBias<V> create(Scope scope, Operand<T> input, Operand<U> filter, Operand<TFloat32> bias, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, Operand<TFloat32> minFilter, Operand<TFloat32> maxFilter, DataType<V> outType, List<Long> strides, String padding, Options... options) {
+  public static <V extends TType, T extends TType, U extends TType> QuantizedConv2DWithBias<V> create(Scope scope, Operand<T> input, Operand<U> filter, Operand<TFloat32> bias, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, Operand<TFloat32> minFilter, Operand<TFloat32> maxFilter, DataType<V> outType, List<Long> strides, String padding, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("QuantizedConv2DWithBias", scope.makeOpName("QuantizedConv2DWithBias"));
-    opBuilder.addInput(input.asOutput());
-    opBuilder.addInput(filter.asOutput());
-    opBuilder.addInput(bias.asOutput());
-    opBuilder.addInput(minInput.asOutput());
-    opBuilder.addInput(maxInput.asOutput());
-    opBuilder.addInput(minFilter.asOutput());
-    opBuilder.addInput(maxFilter.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
+    opBuilder.addInput(filter.asOutput(scope));
+    opBuilder.addInput(bias.asOutput(scope));
+    opBuilder.addInput(minInput.asOutput(scope));
+    opBuilder.addInput(maxInput.asOutput(scope));
+    opBuilder.addInput(minFilter.asOutput(scope));
+    opBuilder.addInput(maxFilter.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("out_type", outType);
     long[] stridesArray = new long[strides.size()];
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DWithBiasAndRelu.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DWithBiasAndRelu.java
index d14eed83c3f..fe01df2a4d8 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DWithBiasAndRelu.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DWithBiasAndRelu.java
@@ -23,17 +23,17 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.family.TType;
 
 /**
  * @param <V> data type for {@code output()} output
  */
-public final class QuantizedConv2DWithBiasAndRelu<V extends Tensor> extends RawOp {
+public final class QuantizedConv2DWithBiasAndRelu<V extends TType> extends RawOp {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.nn.QuantizedConv2DWithBiasAndRelu}
@@ -81,15 +81,15 @@ private Options() {
    * @return a new instance of QuantizedConv2DWithBiasAndRelu
    */
   @Endpoint(describeByClass = true)
-  public static <V extends Tensor, T extends Tensor, U extends Tensor> QuantizedConv2DWithBiasAndRelu<V> create(Scope scope, Operand<T> input, Operand<U> filter, Operand<TFloat32> bias, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, Operand<TFloat32> minFilter, Operand<TFloat32> maxFilter, DataType<V> outType, List<Long> strides, String padding, Options... options) {
+  public static <V extends TType, T extends TType, U extends TType> QuantizedConv2DWithBiasAndRelu<V> create(Scope scope, Operand<T> input, Operand<U> filter, Operand<TFloat32> bias, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, Operand<TFloat32> minFilter, Operand<TFloat32> maxFilter, DataType<V> outType, List<Long> strides, String padding, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("QuantizedConv2DWithBiasAndRelu", scope.makeOpName("QuantizedConv2DWithBiasAndRelu"));
-    opBuilder.addInput(input.asOutput());
-    opBuilder.addInput(filter.asOutput());
-    opBuilder.addInput(bias.asOutput());
-    opBuilder.addInput(minInput.asOutput());
-    opBuilder.addInput(maxInput.asOutput());
-    opBuilder.addInput(minFilter.asOutput());
-    opBuilder.addInput(maxFilter.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
+    opBuilder.addInput(filter.asOutput(scope));
+    opBuilder.addInput(bias.asOutput(scope));
+    opBuilder.addInput(minInput.asOutput(scope));
+    opBuilder.addInput(maxInput.asOutput(scope));
+    opBuilder.addInput(minFilter.asOutput(scope));
+    opBuilder.addInput(maxFilter.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("out_type", outType);
     long[] stridesArray = new long[strides.size()];
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DWithBiasAndReluAndRequantize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DWithBiasAndReluAndRequantize.java
index 293c9595d47..207ac36f953 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DWithBiasAndReluAndRequantize.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DWithBiasAndReluAndRequantize.java
@@ -23,17 +23,17 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.family.TType;
 
 /**
  * @param <W> data type for {@code output()} output
  */
-public final class QuantizedConv2DWithBiasAndReluAndRequantize<W extends Tensor> extends RawOp {
+public final class QuantizedConv2DWithBiasAndReluAndRequantize<W extends TType> extends RawOp {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.nn.QuantizedConv2DWithBiasAndReluAndRequantize}
@@ -83,17 +83,17 @@ private Options() {
    * @return a new instance of QuantizedConv2DWithBiasAndReluAndRequantize
    */
   @Endpoint(describeByClass = true)
-  public static <W extends Tensor, T extends Tensor, U extends Tensor, V extends Tensor> QuantizedConv2DWithBiasAndReluAndRequantize<W> create(Scope scope, Operand<T> input, Operand<U> filter, Operand<V> bias, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, Operand<TFloat32> minFilter, Operand<TFloat32> maxFilter, Operand<TFloat32> minFreezedOutput, Operand<TFloat32> maxFreezedOutput, DataType<W> outType, List<Long> strides, String padding, Options... options) {
+  public static <W extends TType, T extends TType, U extends TType, V extends TType> QuantizedConv2DWithBiasAndReluAndRequantize<W> create(Scope scope, Operand<T> input, Operand<U> filter, Operand<V> bias, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, Operand<TFloat32> minFilter, Operand<TFloat32> maxFilter, Operand<TFloat32> minFreezedOutput, Operand<TFloat32> maxFreezedOutput, DataType<W> outType, List<Long> strides, String padding, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("QuantizedConv2DWithBiasAndReluAndRequantize", scope.makeOpName("QuantizedConv2DWithBiasAndReluAndRequantize"));
-    opBuilder.addInput(input.asOutput());
-    opBuilder.addInput(filter.asOutput());
-    opBuilder.addInput(bias.asOutput());
-    opBuilder.addInput(minInput.asOutput());
-    opBuilder.addInput(maxInput.asOutput());
-    opBuilder.addInput(minFilter.asOutput());
-    opBuilder.addInput(maxFilter.asOutput());
-    opBuilder.addInput(minFreezedOutput.asOutput());
-    opBuilder.addInput(maxFreezedOutput.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
+    opBuilder.addInput(filter.asOutput(scope));
+    opBuilder.addInput(bias.asOutput(scope));
+    opBuilder.addInput(minInput.asOutput(scope));
+    opBuilder.addInput(maxInput.asOutput(scope));
+    opBuilder.addInput(minFilter.asOutput(scope));
+    opBuilder.addInput(maxFilter.asOutput(scope));
+    opBuilder.addInput(minFreezedOutput.asOutput(scope));
+    opBuilder.addInput(maxFreezedOutput.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("out_type", outType);
     long[] stridesArray = new long[strides.size()];
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DWithBiasAndRequantize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DWithBiasAndRequantize.java
index 68ecc13322a..5ff329393b7 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DWithBiasAndRequantize.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DWithBiasAndRequantize.java
@@ -23,17 +23,17 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.family.TType;
 
 /**
  * @param <W> data type for {@code output()} output
  */
-public final class QuantizedConv2DWithBiasAndRequantize<W extends Tensor> extends RawOp {
+public final class QuantizedConv2DWithBiasAndRequantize<W extends TType> extends RawOp {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.nn.QuantizedConv2DWithBiasAndRequantize}
@@ -83,17 +83,17 @@ private Options() {
    * @return a new instance of QuantizedConv2DWithBiasAndRequantize
    */
   @Endpoint(describeByClass = true)
-  public static <W extends Tensor, T extends Tensor, U extends Tensor, V extends Tensor> QuantizedConv2DWithBiasAndRequantize<W> create(Scope scope, Operand<T> input, Operand<U> filter, Operand<V> bias, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, Operand<TFloat32> minFilter, Operand<TFloat32> maxFilter, Operand<TFloat32> minFreezedOutput, Operand<TFloat32> maxFreezedOutput, DataType<W> outType, List<Long> strides, String padding, Options... options) {
+  public static <W extends TType, T extends TType, U extends TType, V extends TType> QuantizedConv2DWithBiasAndRequantize<W> create(Scope scope, Operand<T> input, Operand<U> filter, Operand<V> bias, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, Operand<TFloat32> minFilter, Operand<TFloat32> maxFilter, Operand<TFloat32> minFreezedOutput, Operand<TFloat32> maxFreezedOutput, DataType<W> outType, List<Long> strides, String padding, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("QuantizedConv2DWithBiasAndRequantize", scope.makeOpName("QuantizedConv2DWithBiasAndRequantize"));
-    opBuilder.addInput(input.asOutput());
-    opBuilder.addInput(filter.asOutput());
-    opBuilder.addInput(bias.asOutput());
-    opBuilder.addInput(minInput.asOutput());
-    opBuilder.addInput(maxInput.asOutput());
-    opBuilder.addInput(minFilter.asOutput());
-    opBuilder.addInput(maxFilter.asOutput());
-    opBuilder.addInput(minFreezedOutput.asOutput());
-    opBuilder.addInput(maxFreezedOutput.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
+    opBuilder.addInput(filter.asOutput(scope));
+    opBuilder.addInput(bias.asOutput(scope));
+    opBuilder.addInput(minInput.asOutput(scope));
+    opBuilder.addInput(maxInput.asOutput(scope));
+    opBuilder.addInput(minFilter.asOutput(scope));
+    opBuilder.addInput(maxFilter.asOutput(scope));
+    opBuilder.addInput(minFreezedOutput.asOutput(scope));
+    opBuilder.addInput(maxFreezedOutput.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("out_type", outType);
     long[] stridesArray = new long[strides.size()];
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DWithBiasSignedSumAndReluAndRequantize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DWithBiasSignedSumAndReluAndRequantize.java
index f549d05e064..13f86998835 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DWithBiasSignedSumAndReluAndRequantize.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DWithBiasSignedSumAndReluAndRequantize.java
@@ -23,17 +23,17 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.family.TType;
 
 /**
  * @param <X> data type for {@code output()} output
  */
-public final class QuantizedConv2DWithBiasSignedSumAndReluAndRequantize<X extends Tensor> extends RawOp {
+public final class QuantizedConv2DWithBiasSignedSumAndReluAndRequantize<X extends TType> extends RawOp {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.nn.QuantizedConv2DWithBiasSignedSumAndReluAndRequantize}
@@ -86,20 +86,20 @@ private Options() {
    * @return a new instance of QuantizedConv2DWithBiasSignedSumAndReluAndRequantize
    */
   @Endpoint(describeByClass = true)
-  public static <X extends Tensor, T extends Tensor, U extends Tensor, V extends Tensor, W extends Tensor> QuantizedConv2DWithBiasSignedSumAndReluAndRequantize<X> create(Scope scope, Operand<T> input, Operand<U> filter, Operand<V> bias, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, Operand<TFloat32> minFilter, Operand<TFloat32> maxFilter, Operand<TFloat32> minFreezedOutput, Operand<TFloat32> maxFreezedOutput, Operand<W> summand, Operand<TFloat32> minSummand, Operand<TFloat32> maxSummand, DataType<X> outType, List<Long> strides, String padding, Options... options) {
+  public static <X extends TType, T extends TType, U extends TType, V extends TType, W extends TType> QuantizedConv2DWithBiasSignedSumAndReluAndRequantize<X> create(Scope scope, Operand<T> input, Operand<U> filter, Operand<V> bias, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, Operand<TFloat32> minFilter, Operand<TFloat32> maxFilter, Operand<TFloat32> minFreezedOutput, Operand<TFloat32> maxFreezedOutput, Operand<W> summand, Operand<TFloat32> minSummand, Operand<TFloat32> maxSummand, DataType<X> outType, List<Long> strides, String padding, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("QuantizedConv2DWithBiasSignedSumAndReluAndRequantize", scope.makeOpName("QuantizedConv2DWithBiasSignedSumAndReluAndRequantize"));
-    opBuilder.addInput(input.asOutput());
-    opBuilder.addInput(filter.asOutput());
-    opBuilder.addInput(bias.asOutput());
-    opBuilder.addInput(minInput.asOutput());
-    opBuilder.addInput(maxInput.asOutput());
-    opBuilder.addInput(minFilter.asOutput());
-    opBuilder.addInput(maxFilter.asOutput());
-    opBuilder.addInput(minFreezedOutput.asOutput());
-    opBuilder.addInput(maxFreezedOutput.asOutput());
-    opBuilder.addInput(summand.asOutput());
-    opBuilder.addInput(minSummand.asOutput());
-    opBuilder.addInput(maxSummand.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
+    opBuilder.addInput(filter.asOutput(scope));
+    opBuilder.addInput(bias.asOutput(scope));
+    opBuilder.addInput(minInput.asOutput(scope));
+    opBuilder.addInput(maxInput.asOutput(scope));
+    opBuilder.addInput(minFilter.asOutput(scope));
+    opBuilder.addInput(maxFilter.asOutput(scope));
+    opBuilder.addInput(minFreezedOutput.asOutput(scope));
+    opBuilder.addInput(maxFreezedOutput.asOutput(scope));
+    opBuilder.addInput(summand.asOutput(scope));
+    opBuilder.addInput(minSummand.asOutput(scope));
+    opBuilder.addInput(maxSummand.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("out_type", outType);
     long[] stridesArray = new long[strides.size()];
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DWithBiasSumAndRelu.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DWithBiasSumAndRelu.java
index ebb5799a648..bed47e5ac80 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DWithBiasSumAndRelu.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DWithBiasSumAndRelu.java
@@ -23,17 +23,17 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.family.TType;
 
 /**
  * @param <V> data type for {@code output()} output
  */
-public final class QuantizedConv2DWithBiasSumAndRelu<V extends Tensor> extends RawOp {
+public final class QuantizedConv2DWithBiasSumAndRelu<V extends TType> extends RawOp {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.nn.QuantizedConv2DWithBiasSumAndRelu}
@@ -82,16 +82,16 @@ private Options() {
    * @return a new instance of QuantizedConv2DWithBiasSumAndRelu
    */
   @Endpoint(describeByClass = true)
-  public static <V extends Tensor, T extends Tensor, U extends Tensor> QuantizedConv2DWithBiasSumAndRelu<V> create(Scope scope, Operand<T> input, Operand<U> filter, Operand<TFloat32> bias, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, Operand<TFloat32> minFilter, Operand<TFloat32> maxFilter, Operand<TFloat32> summand, DataType<V> outType, List<Long> strides, String padding, Options... options) {
+  public static <V extends TType, T extends TType, U extends TType> QuantizedConv2DWithBiasSumAndRelu<V> create(Scope scope, Operand<T> input, Operand<U> filter, Operand<TFloat32> bias, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, Operand<TFloat32> minFilter, Operand<TFloat32> maxFilter, Operand<TFloat32> summand, DataType<V> outType, List<Long> strides, String padding, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("QuantizedConv2DWithBiasSumAndRelu", scope.makeOpName("QuantizedConv2DWithBiasSumAndRelu"));
-    opBuilder.addInput(input.asOutput());
-    opBuilder.addInput(filter.asOutput());
-    opBuilder.addInput(bias.asOutput());
-    opBuilder.addInput(minInput.asOutput());
-    opBuilder.addInput(maxInput.asOutput());
-    opBuilder.addInput(minFilter.asOutput());
-    opBuilder.addInput(maxFilter.asOutput());
-    opBuilder.addInput(summand.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
+    opBuilder.addInput(filter.asOutput(scope));
+    opBuilder.addInput(bias.asOutput(scope));
+    opBuilder.addInput(minInput.asOutput(scope));
+    opBuilder.addInput(maxInput.asOutput(scope));
+    opBuilder.addInput(minFilter.asOutput(scope));
+    opBuilder.addInput(maxFilter.asOutput(scope));
+    opBuilder.addInput(summand.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("out_type", outType);
     long[] stridesArray = new long[strides.size()];
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DWithBiasSumAndReluAndRequantize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DWithBiasSumAndReluAndRequantize.java
index cbcad086f19..72279500957 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DWithBiasSumAndReluAndRequantize.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DWithBiasSumAndReluAndRequantize.java
@@ -23,17 +23,17 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.family.TType;
 
 /**
  * @param <X> data type for {@code output()} output
  */
-public final class QuantizedConv2DWithBiasSumAndReluAndRequantize<X extends Tensor> extends RawOp {
+public final class QuantizedConv2DWithBiasSumAndReluAndRequantize<X extends TType> extends RawOp {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.nn.QuantizedConv2DWithBiasSumAndReluAndRequantize}
@@ -86,20 +86,20 @@ private Options() {
    * @return a new instance of QuantizedConv2DWithBiasSumAndReluAndRequantize
    */
   @Endpoint(describeByClass = true)
-  public static <X extends Tensor, T extends Tensor, U extends Tensor, V extends Tensor, W extends Tensor> QuantizedConv2DWithBiasSumAndReluAndRequantize<X> create(Scope scope, Operand<T> input, Operand<U> filter, Operand<V> bias, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, Operand<TFloat32> minFilter, Operand<TFloat32> maxFilter, Operand<TFloat32> minFreezedOutput, Operand<TFloat32> maxFreezedOutput, Operand<W> summand, Operand<TFloat32> minSummand, Operand<TFloat32> maxSummand, DataType<X> outType, List<Long> strides, String padding, Options... options) {
+  public static <X extends TType, T extends TType, U extends TType, V extends TType, W extends TType> QuantizedConv2DWithBiasSumAndReluAndRequantize<X> create(Scope scope, Operand<T> input, Operand<U> filter, Operand<V> bias, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, Operand<TFloat32> minFilter, Operand<TFloat32> maxFilter, Operand<TFloat32> minFreezedOutput, Operand<TFloat32> maxFreezedOutput, Operand<W> summand, Operand<TFloat32> minSummand, Operand<TFloat32> maxSummand, DataType<X> outType, List<Long> strides, String padding, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("QuantizedConv2DWithBiasSumAndReluAndRequantize", scope.makeOpName("QuantizedConv2DWithBiasSumAndReluAndRequantize"));
-    opBuilder.addInput(input.asOutput());
-    opBuilder.addInput(filter.asOutput());
-    opBuilder.addInput(bias.asOutput());
-    opBuilder.addInput(minInput.asOutput());
-    opBuilder.addInput(maxInput.asOutput());
-    opBuilder.addInput(minFilter.asOutput());
-    opBuilder.addInput(maxFilter.asOutput());
-    opBuilder.addInput(minFreezedOutput.asOutput());
-    opBuilder.addInput(maxFreezedOutput.asOutput());
-    opBuilder.addInput(summand.asOutput());
-    opBuilder.addInput(minSummand.asOutput());
-    opBuilder.addInput(maxSummand.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
+    opBuilder.addInput(filter.asOutput(scope));
+    opBuilder.addInput(bias.asOutput(scope));
+    opBuilder.addInput(minInput.asOutput(scope));
+    opBuilder.addInput(maxInput.asOutput(scope));
+    opBuilder.addInput(minFilter.asOutput(scope));
+    opBuilder.addInput(maxFilter.asOutput(scope));
+    opBuilder.addInput(minFreezedOutput.asOutput(scope));
+    opBuilder.addInput(maxFreezedOutput.asOutput(scope));
+    opBuilder.addInput(summand.asOutput(scope));
+    opBuilder.addInput(minSummand.asOutput(scope));
+    opBuilder.addInput(maxSummand.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("out_type", outType);
     long[] stridesArray = new long[strides.size()];
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2d.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2d.java
index a981e3d9f28..b4ad6e2485d 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2d.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2d.java
@@ -23,12 +23,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.family.TType;
 
 /**
  * Computes a 2D convolution given quantized 4D input and filter tensors.
@@ -41,7 +41,7 @@
  * @param <V> data type for {@code output()} output
  */
 @Operator(group = "nn")
-public final class QuantizedConv2d<V extends Tensor> extends RawOp {
+public final class QuantizedConv2d<V extends TType> extends RawOp {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.nn.QuantizedConv2d}
@@ -84,14 +84,14 @@ private Options() {
    * @return a new instance of QuantizedConv2d
    */
   @Endpoint(describeByClass = true)
-  public static <V extends Tensor, T extends Tensor, U extends Tensor> QuantizedConv2d<V> create(Scope scope, Operand<T> input, Operand<U> filter, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, Operand<TFloat32> minFilter, Operand<TFloat32> maxFilter, DataType<V> outType, List<Long> strides, String padding, Options... options) {
+  public static <V extends TType, T extends TType, U extends TType> QuantizedConv2d<V> create(Scope scope, Operand<T> input, Operand<U> filter, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, Operand<TFloat32> minFilter, Operand<TFloat32> maxFilter, DataType<V> outType, List<Long> strides, String padding, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("QuantizedConv2D", scope.makeOpName("QuantizedConv2d"));
-    opBuilder.addInput(input.asOutput());
-    opBuilder.addInput(filter.asOutput());
-    opBuilder.addInput(minInput.asOutput());
-    opBuilder.addInput(maxInput.asOutput());
-    opBuilder.addInput(minFilter.asOutput());
-    opBuilder.addInput(maxFilter.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
+    opBuilder.addInput(filter.asOutput(scope));
+    opBuilder.addInput(minInput.asOutput(scope));
+    opBuilder.addInput(maxInput.asOutput(scope));
+    opBuilder.addInput(minFilter.asOutput(scope));
+    opBuilder.addInput(maxFilter.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("out_type", outType);
     long[] stridesArray = new long[strides.size()];
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedDepthwiseConv2D.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedDepthwiseConv2D.java
index e4511f72a84..715fba4d7af 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedDepthwiseConv2D.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedDepthwiseConv2D.java
@@ -23,19 +23,19 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.family.TType;
 
 /**
  * Computes quantized depthwise Conv2D.
  * 
  * @param <V> data type for {@code output()} output
  */
-public final class QuantizedDepthwiseConv2D<V extends Tensor> extends RawOp {
+public final class QuantizedDepthwiseConv2D<V extends TType> extends RawOp {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.nn.QuantizedDepthwiseConv2D}
@@ -73,14 +73,14 @@ private Options() {
    * @return a new instance of QuantizedDepthwiseConv2D
    */
   @Endpoint(describeByClass = true)
-  public static <V extends Tensor, T extends Tensor, U extends Tensor> QuantizedDepthwiseConv2D<V> create(Scope scope, Operand<T> input, Operand<U> filter, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, Operand<TFloat32> minFilter, Operand<TFloat32> maxFilter, DataType<V> outType, List<Long> strides, String padding, Options... options) {
+  public static <V extends TType, T extends TType, U extends TType> QuantizedDepthwiseConv2D<V> create(Scope scope, Operand<T> input, Operand<U> filter, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, Operand<TFloat32> minFilter, Operand<TFloat32> maxFilter, DataType<V> outType, List<Long> strides, String padding, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("QuantizedDepthwiseConv2D", scope.makeOpName("QuantizedDepthwiseConv2D"));
-    opBuilder.addInput(input.asOutput());
-    opBuilder.addInput(filter.asOutput());
-    opBuilder.addInput(minInput.asOutput());
-    opBuilder.addInput(maxInput.asOutput());
-    opBuilder.addInput(minFilter.asOutput());
-    opBuilder.addInput(maxFilter.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
+    opBuilder.addInput(filter.asOutput(scope));
+    opBuilder.addInput(minInput.asOutput(scope));
+    opBuilder.addInput(maxInput.asOutput(scope));
+    opBuilder.addInput(minFilter.asOutput(scope));
+    opBuilder.addInput(maxFilter.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("out_type", outType);
     long[] stridesArray = new long[strides.size()];
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedDepthwiseConv2DWithBias.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedDepthwiseConv2DWithBias.java
index b70e505a990..7e5e703b462 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedDepthwiseConv2DWithBias.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedDepthwiseConv2DWithBias.java
@@ -23,19 +23,19 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.family.TType;
 
 /**
  * Computes quantized depthwise Conv2D with Bias.
  * 
  * @param <V> data type for {@code output()} output
  */
-public final class QuantizedDepthwiseConv2DWithBias<V extends Tensor> extends RawOp {
+public final class QuantizedDepthwiseConv2DWithBias<V extends TType> extends RawOp {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.nn.QuantizedDepthwiseConv2DWithBias}
@@ -74,15 +74,15 @@ private Options() {
    * @return a new instance of QuantizedDepthwiseConv2DWithBias
    */
   @Endpoint(describeByClass = true)
-  public static <V extends Tensor, T extends Tensor, U extends Tensor> QuantizedDepthwiseConv2DWithBias<V> create(Scope scope, Operand<T> input, Operand<U> filter, Operand<TFloat32> bias, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, Operand<TFloat32> minFilter, Operand<TFloat32> maxFilter, DataType<V> outType, List<Long> strides, String padding, Options... options) {
+  public static <V extends TType, T extends TType, U extends TType> QuantizedDepthwiseConv2DWithBias<V> create(Scope scope, Operand<T> input, Operand<U> filter, Operand<TFloat32> bias, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, Operand<TFloat32> minFilter, Operand<TFloat32> maxFilter, DataType<V> outType, List<Long> strides, String padding, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("QuantizedDepthwiseConv2DWithBias", scope.makeOpName("QuantizedDepthwiseConv2DWithBias"));
-    opBuilder.addInput(input.asOutput());
-    opBuilder.addInput(filter.asOutput());
-    opBuilder.addInput(bias.asOutput());
-    opBuilder.addInput(minInput.asOutput());
-    opBuilder.addInput(maxInput.asOutput());
-    opBuilder.addInput(minFilter.asOutput());
-    opBuilder.addInput(maxFilter.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
+    opBuilder.addInput(filter.asOutput(scope));
+    opBuilder.addInput(bias.asOutput(scope));
+    opBuilder.addInput(minInput.asOutput(scope));
+    opBuilder.addInput(maxInput.asOutput(scope));
+    opBuilder.addInput(minFilter.asOutput(scope));
+    opBuilder.addInput(maxFilter.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("out_type", outType);
     long[] stridesArray = new long[strides.size()];
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedDepthwiseConv2DWithBiasAndRelu.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedDepthwiseConv2DWithBiasAndRelu.java
index 46d2fd032a5..bdf1762476f 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedDepthwiseConv2DWithBiasAndRelu.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedDepthwiseConv2DWithBiasAndRelu.java
@@ -23,19 +23,19 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.family.TType;
 
 /**
  * Computes quantized depthwise Conv2D with Bias and Relu.
  * 
  * @param <V> data type for {@code output()} output
  */
-public final class QuantizedDepthwiseConv2DWithBiasAndRelu<V extends Tensor> extends RawOp {
+public final class QuantizedDepthwiseConv2DWithBiasAndRelu<V extends TType> extends RawOp {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.nn.QuantizedDepthwiseConv2DWithBiasAndRelu}
@@ -83,15 +83,15 @@ private Options() {
    * @return a new instance of QuantizedDepthwiseConv2DWithBiasAndRelu
    */
   @Endpoint(describeByClass = true)
-  public static <V extends Tensor, T extends Tensor, U extends Tensor> QuantizedDepthwiseConv2DWithBiasAndRelu<V> create(Scope scope, Operand<T> input, Operand<U> filter, Operand<TFloat32> bias, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, Operand<TFloat32> minFilter, Operand<TFloat32> maxFilter, DataType<V> outType, List<Long> strides, String padding, Options... options) {
+  public static <V extends TType, T extends TType, U extends TType> QuantizedDepthwiseConv2DWithBiasAndRelu<V> create(Scope scope, Operand<T> input, Operand<U> filter, Operand<TFloat32> bias, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, Operand<TFloat32> minFilter, Operand<TFloat32> maxFilter, DataType<V> outType, List<Long> strides, String padding, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("QuantizedDepthwiseConv2DWithBiasAndRelu", scope.makeOpName("QuantizedDepthwiseConv2DWithBiasAndRelu"));
-    opBuilder.addInput(input.asOutput());
-    opBuilder.addInput(filter.asOutput());
-    opBuilder.addInput(bias.asOutput());
-    opBuilder.addInput(minInput.asOutput());
-    opBuilder.addInput(maxInput.asOutput());
-    opBuilder.addInput(minFilter.asOutput());
-    opBuilder.addInput(maxFilter.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
+    opBuilder.addInput(filter.asOutput(scope));
+    opBuilder.addInput(bias.asOutput(scope));
+    opBuilder.addInput(minInput.asOutput(scope));
+    opBuilder.addInput(maxInput.asOutput(scope));
+    opBuilder.addInput(minFilter.asOutput(scope));
+    opBuilder.addInput(maxFilter.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("out_type", outType);
     long[] stridesArray = new long[strides.size()];
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedDepthwiseConv2DWithBiasAndReluAndRequantize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedDepthwiseConv2DWithBiasAndReluAndRequantize.java
index c2638715575..11193af5bfe 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedDepthwiseConv2DWithBiasAndReluAndRequantize.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedDepthwiseConv2DWithBiasAndReluAndRequantize.java
@@ -23,19 +23,19 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.family.TType;
 
 /**
  * Computes quantized depthwise Conv2D with Bias, Relu and Requantize.
  * 
  * @param <W> data type for {@code output()} output
  */
-public final class QuantizedDepthwiseConv2DWithBiasAndReluAndRequantize<W extends Tensor> extends RawOp {
+public final class QuantizedDepthwiseConv2DWithBiasAndReluAndRequantize<W extends TType> extends RawOp {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.nn.QuantizedDepthwiseConv2DWithBiasAndReluAndRequantize}
@@ -85,17 +85,17 @@ private Options() {
    * @return a new instance of QuantizedDepthwiseConv2DWithBiasAndReluAndRequantize
    */
   @Endpoint(describeByClass = true)
-  public static <W extends Tensor, T extends Tensor, U extends Tensor, V extends Tensor> QuantizedDepthwiseConv2DWithBiasAndReluAndRequantize<W> create(Scope scope, Operand<T> input, Operand<U> filter, Operand<V> bias, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, Operand<TFloat32> minFilter, Operand<TFloat32> maxFilter, Operand<TFloat32> minFreezedOutput, Operand<TFloat32> maxFreezedOutput, DataType<W> outType, List<Long> strides, String padding, Options... options) {
+  public static <W extends TType, T extends TType, U extends TType, V extends TType> QuantizedDepthwiseConv2DWithBiasAndReluAndRequantize<W> create(Scope scope, Operand<T> input, Operand<U> filter, Operand<V> bias, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, Operand<TFloat32> minFilter, Operand<TFloat32> maxFilter, Operand<TFloat32> minFreezedOutput, Operand<TFloat32> maxFreezedOutput, DataType<W> outType, List<Long> strides, String padding, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("QuantizedDepthwiseConv2DWithBiasAndReluAndRequantize", scope.makeOpName("QuantizedDepthwiseConv2DWithBiasAndReluAndRequantize"));
-    opBuilder.addInput(input.asOutput());
-    opBuilder.addInput(filter.asOutput());
-    opBuilder.addInput(bias.asOutput());
-    opBuilder.addInput(minInput.asOutput());
-    opBuilder.addInput(maxInput.asOutput());
-    opBuilder.addInput(minFilter.asOutput());
-    opBuilder.addInput(maxFilter.asOutput());
-    opBuilder.addInput(minFreezedOutput.asOutput());
-    opBuilder.addInput(maxFreezedOutput.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
+    opBuilder.addInput(filter.asOutput(scope));
+    opBuilder.addInput(bias.asOutput(scope));
+    opBuilder.addInput(minInput.asOutput(scope));
+    opBuilder.addInput(maxInput.asOutput(scope));
+    opBuilder.addInput(minFilter.asOutput(scope));
+    opBuilder.addInput(maxFilter.asOutput(scope));
+    opBuilder.addInput(minFreezedOutput.asOutput(scope));
+    opBuilder.addInput(maxFreezedOutput.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("out_type", outType);
     long[] stridesArray = new long[strides.size()];
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedInstanceNorm.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedInstanceNorm.java
index 87dac703ac9..5c12e5a6b71 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedInstanceNorm.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedInstanceNorm.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.family.TType;
 
 /**
  * Quantized Instance normalization.
@@ -34,7 +34,7 @@
  * @param <T> data type for {@code y()} output
  */
 @Operator(group = "nn")
-public final class QuantizedInstanceNorm<T extends Tensor> extends RawOp {
+public final class QuantizedInstanceNorm<T extends TType> extends RawOp {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.nn.QuantizedInstanceNorm}
@@ -104,11 +104,11 @@ private Options() {
    * @return a new instance of QuantizedInstanceNorm
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> QuantizedInstanceNorm<T> create(Scope scope, Operand<T> x, Operand<TFloat32> xMin, Operand<TFloat32> xMax, Options... options) {
+  public static <T extends TType> QuantizedInstanceNorm<T> create(Scope scope, Operand<T> x, Operand<TFloat32> xMin, Operand<TFloat32> xMax, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("QuantizedInstanceNorm", scope.makeOpName("QuantizedInstanceNorm"));
-    opBuilder.addInput(x.asOutput());
-    opBuilder.addInput(xMin.asOutput());
-    opBuilder.addInput(xMax.asOutput());
+    opBuilder.addInput(x.asOutput(scope));
+    opBuilder.addInput(xMin.asOutput(scope));
+    opBuilder.addInput(xMax.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedMaxPool.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedMaxPool.java
index 206df4fc20c..aaea42bd906 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedMaxPool.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedMaxPool.java
@@ -22,12 +22,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.family.TType;
 
 /**
  * Produces the max pool of the input tensor for quantized types.
@@ -35,7 +35,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "nn")
-public final class QuantizedMaxPool<T extends Tensor> extends RawOp {
+public final class QuantizedMaxPool<T extends TType> extends RawOp {
   
   /**
    * Factory method to create a class wrapping a new QuantizedMaxPool operation.
@@ -52,11 +52,11 @@ public final class QuantizedMaxPool<T extends Tensor> extends RawOp {
    * @return a new instance of QuantizedMaxPool
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> QuantizedMaxPool<T> create(Scope scope, Operand<T> input, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, List<Long> ksize, List<Long> strides, String padding) {
+  public static <T extends TType> QuantizedMaxPool<T> create(Scope scope, Operand<T> input, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, List<Long> ksize, List<Long> strides, String padding) {
     OperationBuilder opBuilder = scope.env().opBuilder("QuantizedMaxPool", scope.makeOpName("QuantizedMaxPool"));
-    opBuilder.addInput(input.asOutput());
-    opBuilder.addInput(minInput.asOutput());
-    opBuilder.addInput(maxInput.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
+    opBuilder.addInput(minInput.asOutput(scope));
+    opBuilder.addInput(maxInput.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     long[] ksizeArray = new long[ksize.size()];
     for (int i = 0; i < ksizeArray.length; ++i) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedRelu.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedRelu.java
index 3064343acaa..69e5f6e5a46 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedRelu.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedRelu.java
@@ -22,12 +22,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.family.TType;
 
 /**
  * Computes Quantized Rectified Linear: `max(features, 0)`
@@ -35,7 +35,7 @@
  * @param <U> data type for {@code activations()} output
  */
 @Operator(group = "nn")
-public final class QuantizedRelu<U extends Tensor> extends RawOp {
+public final class QuantizedRelu<U extends TType> extends RawOp {
   
   /**
    * Factory method to create a class wrapping a new QuantizedRelu operation.
@@ -48,11 +48,11 @@ public final class QuantizedRelu<U extends Tensor> extends RawOp {
    * @return a new instance of QuantizedRelu
    */
   @Endpoint(describeByClass = true)
-  public static <U extends Tensor, T extends Tensor> QuantizedRelu<U> create(Scope scope, Operand<T> features, Operand<TFloat32> minFeatures, Operand<TFloat32> maxFeatures, DataType<U> outType) {
+  public static <U extends TType, T extends TType> QuantizedRelu<U> create(Scope scope, Operand<T> features, Operand<TFloat32> minFeatures, Operand<TFloat32> maxFeatures, DataType<U> outType) {
     OperationBuilder opBuilder = scope.env().opBuilder("QuantizedRelu", scope.makeOpName("QuantizedRelu"));
-    opBuilder.addInput(features.asOutput());
-    opBuilder.addInput(minFeatures.asOutput());
-    opBuilder.addInput(maxFeatures.asOutput());
+    opBuilder.addInput(features.asOutput(scope));
+    opBuilder.addInput(minFeatures.asOutput(scope));
+    opBuilder.addInput(maxFeatures.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("out_type", outType);
     return new QuantizedRelu<U>(opBuilder.build());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedRelu6.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedRelu6.java
index 60b45b90ba5..365fbf97003 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedRelu6.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedRelu6.java
@@ -22,12 +22,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.family.TType;
 
 /**
  * Computes Quantized Rectified Linear 6: `min(max(features, 0), 6)`
@@ -35,7 +35,7 @@
  * @param <U> data type for {@code activations()} output
  */
 @Operator(group = "nn")
-public final class QuantizedRelu6<U extends Tensor> extends RawOp {
+public final class QuantizedRelu6<U extends TType> extends RawOp {
   
   /**
    * Factory method to create a class wrapping a new QuantizedRelu6 operation.
@@ -48,11 +48,11 @@ public final class QuantizedRelu6<U extends Tensor> extends RawOp {
    * @return a new instance of QuantizedRelu6
    */
   @Endpoint(describeByClass = true)
-  public static <U extends Tensor, T extends Tensor> QuantizedRelu6<U> create(Scope scope, Operand<T> features, Operand<TFloat32> minFeatures, Operand<TFloat32> maxFeatures, DataType<U> outType) {
+  public static <U extends TType, T extends TType> QuantizedRelu6<U> create(Scope scope, Operand<T> features, Operand<TFloat32> minFeatures, Operand<TFloat32> maxFeatures, DataType<U> outType) {
     OperationBuilder opBuilder = scope.env().opBuilder("QuantizedRelu6", scope.makeOpName("QuantizedRelu6"));
-    opBuilder.addInput(features.asOutput());
-    opBuilder.addInput(minFeatures.asOutput());
-    opBuilder.addInput(maxFeatures.asOutput());
+    opBuilder.addInput(features.asOutput(scope));
+    opBuilder.addInput(minFeatures.asOutput(scope));
+    opBuilder.addInput(maxFeatures.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("out_type", outType);
     return new QuantizedRelu6<U>(opBuilder.build());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedReluX.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedReluX.java
index 17604d822d2..73826f85087 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedReluX.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedReluX.java
@@ -22,12 +22,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.family.TType;
 
 /**
  * Computes Quantized Rectified Linear X: `min(max(features, 0), max_value)`
@@ -35,7 +35,7 @@
  * @param <U> data type for {@code activations()} output
  */
 @Operator(group = "nn")
-public final class QuantizedReluX<U extends Tensor> extends RawOp {
+public final class QuantizedReluX<U extends TType> extends RawOp {
   
   /**
    * Factory method to create a class wrapping a new QuantizedReluX operation.
@@ -49,12 +49,12 @@ public final class QuantizedReluX<U extends Tensor> extends RawOp {
    * @return a new instance of QuantizedReluX
    */
   @Endpoint(describeByClass = true)
-  public static <U extends Tensor, T extends Tensor> QuantizedReluX<U> create(Scope scope, Operand<T> features, Operand<TFloat32> maxValue, Operand<TFloat32> minFeatures, Operand<TFloat32> maxFeatures, DataType<U> outType) {
+  public static <U extends TType, T extends TType> QuantizedReluX<U> create(Scope scope, Operand<T> features, Operand<TFloat32> maxValue, Operand<TFloat32> minFeatures, Operand<TFloat32> maxFeatures, DataType<U> outType) {
     OperationBuilder opBuilder = scope.env().opBuilder("QuantizedReluX", scope.makeOpName("QuantizedReluX"));
-    opBuilder.addInput(features.asOutput());
-    opBuilder.addInput(maxValue.asOutput());
-    opBuilder.addInput(minFeatures.asOutput());
-    opBuilder.addInput(maxFeatures.asOutput());
+    opBuilder.addInput(features.asOutput(scope));
+    opBuilder.addInput(maxValue.asOutput(scope));
+    opBuilder.addInput(minFeatures.asOutput(scope));
+    opBuilder.addInput(maxFeatures.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("out_type", outType);
     return new QuantizedReluX<U>(opBuilder.build());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Relu.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Relu.java
index 4ddf49d6677..3482ad1f7e4 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Relu.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Relu.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Computes rectified linear: `max(features, 0)`.
@@ -38,7 +38,7 @@
  * @param <T> data type for {@code activations()} output
  */
 @Operator(group = "nn")
-public final class Relu<T extends Tensor> extends RawOp implements Operand<T> {
+public final class Relu<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Relu operation.
@@ -48,9 +48,9 @@ public final class Relu<T extends Tensor> extends RawOp implements Operand<T> {
    * @return a new instance of Relu
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> Relu<T> create(Scope scope, Operand<T> features) {
+  public static <T extends TType> Relu<T> create(Scope scope, Operand<T> features) {
     OperationBuilder opBuilder = scope.env().opBuilder("Relu", scope.makeOpName("Relu"));
-    opBuilder.addInput(features.asOutput());
+    opBuilder.addInput(features.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new Relu<T>(opBuilder.build());
   }
@@ -62,7 +62,7 @@ public Output<T> activations() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return activations;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Relu6.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Relu6.java
index cfd822bf138..f497fec9130 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Relu6.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Relu6.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -34,7 +33,7 @@
  * @param <T> data type for {@code activations()} output
  */
 @Operator(group = "nn")
-public final class Relu6<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class Relu6<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Relu6 operation.
@@ -44,9 +43,9 @@ public final class Relu6<T extends Tensor & TNumber> extends RawOp implements Op
    * @return a new instance of Relu6
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> Relu6<T> create(Scope scope, Operand<T> features) {
+  public static <T extends TNumber> Relu6<T> create(Scope scope, Operand<T> features) {
     OperationBuilder opBuilder = scope.env().opBuilder("Relu6", scope.makeOpName("Relu6"));
-    opBuilder.addInput(features.asOutput());
+    opBuilder.addInput(features.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new Relu6<T>(opBuilder.build());
   }
@@ -58,7 +57,7 @@ public Output<T> activations() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return activations;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Relu6Grad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Relu6Grad.java
index 3555a2fa244..77fa55e828a 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Relu6Grad.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Relu6Grad.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -33,7 +32,7 @@
  * 
  * @param <T> data type for {@code backprops()} output
  */
-public final class Relu6Grad<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class Relu6Grad<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Relu6Grad operation.
@@ -45,10 +44,10 @@ public final class Relu6Grad<T extends Tensor & TNumber> extends RawOp implement
    * @return a new instance of Relu6Grad
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> Relu6Grad<T> create(Scope scope, Operand<T> gradients, Operand<T> features) {
+  public static <T extends TNumber> Relu6Grad<T> create(Scope scope, Operand<T> gradients, Operand<T> features) {
     OperationBuilder opBuilder = scope.env().opBuilder("Relu6Grad", scope.makeOpName("Relu6Grad"));
-    opBuilder.addInput(gradients.asOutput());
-    opBuilder.addInput(features.asOutput());
+    opBuilder.addInput(gradients.asOutput(scope));
+    opBuilder.addInput(features.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new Relu6Grad<T>(opBuilder.build());
   }
@@ -62,7 +61,7 @@ public Output<T> backprops() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return backprops;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/ReluGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/ReluGrad.java
index 790d26427c6..936a7a8df6b 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/ReluGrad.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/ReluGrad.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -33,7 +32,7 @@
  * 
  * @param <T> data type for {@code backprops()} output
  */
-public final class ReluGrad<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class ReluGrad<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new ReluGrad operation.
@@ -45,10 +44,10 @@ public final class ReluGrad<T extends Tensor & TNumber> extends RawOp implements
    * @return a new instance of ReluGrad
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> ReluGrad<T> create(Scope scope, Operand<T> gradients, Operand<T> features) {
+  public static <T extends TNumber> ReluGrad<T> create(Scope scope, Operand<T> gradients, Operand<T> features) {
     OperationBuilder opBuilder = scope.env().opBuilder("ReluGrad", scope.makeOpName("ReluGrad"));
-    opBuilder.addInput(gradients.asOutput());
-    opBuilder.addInput(features.asOutput());
+    opBuilder.addInput(gradients.asOutput(scope));
+    opBuilder.addInput(features.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new ReluGrad<T>(opBuilder.build());
   }
@@ -61,7 +60,7 @@ public Output<T> backprops() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return backprops;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Selu.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Selu.java
index 802312a4f32..9a1cc72c687 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Selu.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Selu.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -42,7 +41,7 @@
  * @param <T> data type for {@code activations()} output
  */
 @Operator(group = "nn")
-public final class Selu<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class Selu<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Selu operation.
@@ -52,9 +51,9 @@ public final class Selu<T extends Tensor & TNumber> extends RawOp implements Ope
    * @return a new instance of Selu
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> Selu<T> create(Scope scope, Operand<T> features) {
+  public static <T extends TNumber> Selu<T> create(Scope scope, Operand<T> features) {
     OperationBuilder opBuilder = scope.env().opBuilder("Selu", scope.makeOpName("Selu"));
-    opBuilder.addInput(features.asOutput());
+    opBuilder.addInput(features.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new Selu<T>(opBuilder.build());
   }
@@ -66,7 +65,7 @@ public Output<T> activations() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return activations;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/SeluGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/SeluGrad.java
index b6f5abc4e95..5acf2b14f93 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/SeluGrad.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/SeluGrad.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -33,7 +32,7 @@
  * 
  * @param <T> data type for {@code backprops()} output
  */
-public final class SeluGrad<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class SeluGrad<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new SeluGrad operation.
@@ -44,10 +43,10 @@ public final class SeluGrad<T extends Tensor & TNumber> extends RawOp implements
    * @return a new instance of SeluGrad
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> SeluGrad<T> create(Scope scope, Operand<T> gradients, Operand<T> outputs) {
+  public static <T extends TNumber> SeluGrad<T> create(Scope scope, Operand<T> gradients, Operand<T> outputs) {
     OperationBuilder opBuilder = scope.env().opBuilder("SeluGrad", scope.makeOpName("SeluGrad"));
-    opBuilder.addInput(gradients.asOutput());
-    opBuilder.addInput(outputs.asOutput());
+    opBuilder.addInput(gradients.asOutput(scope));
+    opBuilder.addInput(outputs.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new SeluGrad<T>(opBuilder.build());
   }
@@ -61,7 +60,7 @@ public Output<T> backprops() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return backprops;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Softmax.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Softmax.java
index a3ed77b8e83..3cef45fd7a7 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Softmax.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Softmax.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -38,7 +37,7 @@
  * @param <T> data type for {@code softmax()} output
  */
 @Operator(group = "nn")
-public final class Softmax<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class Softmax<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Softmax operation.
@@ -48,9 +47,9 @@ public final class Softmax<T extends Tensor & TNumber> extends RawOp implements
    * @return a new instance of Softmax
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> Softmax<T> create(Scope scope, Operand<T> logits) {
+  public static <T extends TNumber> Softmax<T> create(Scope scope, Operand<T> logits) {
     OperationBuilder opBuilder = scope.env().opBuilder("Softmax", scope.makeOpName("Softmax"));
-    opBuilder.addInput(logits.asOutput());
+    opBuilder.addInput(logits.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new Softmax<T>(opBuilder.build());
   }
@@ -63,7 +62,7 @@ public Output<T> softmax() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return softmax;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Softsign.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Softsign.java
index 8ea369d0b27..b3494d281a7 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Softsign.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Softsign.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -34,7 +33,7 @@
  * @param <T> data type for {@code activations()} output
  */
 @Operator(group = "nn")
-public final class Softsign<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class Softsign<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Softsign operation.
@@ -44,9 +43,9 @@ public final class Softsign<T extends Tensor & TNumber> extends RawOp implements
    * @return a new instance of Softsign
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> Softsign<T> create(Scope scope, Operand<T> features) {
+  public static <T extends TNumber> Softsign<T> create(Scope scope, Operand<T> features) {
     OperationBuilder opBuilder = scope.env().opBuilder("Softsign", scope.makeOpName("Softsign"));
-    opBuilder.addInput(features.asOutput());
+    opBuilder.addInput(features.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new Softsign<T>(opBuilder.build());
   }
@@ -58,7 +57,7 @@ public Output<T> activations() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return activations;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/SoftsignGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/SoftsignGrad.java
index f8b55b3ecd9..70f6a9077fb 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/SoftsignGrad.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/SoftsignGrad.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -33,7 +32,7 @@
  * 
  * @param <T> data type for {@code backprops()} output
  */
-public final class SoftsignGrad<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class SoftsignGrad<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new SoftsignGrad operation.
@@ -44,10 +43,10 @@ public final class SoftsignGrad<T extends Tensor & TNumber> extends RawOp implem
    * @return a new instance of SoftsignGrad
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> SoftsignGrad<T> create(Scope scope, Operand<T> gradients, Operand<T> features) {
+  public static <T extends TNumber> SoftsignGrad<T> create(Scope scope, Operand<T> gradients, Operand<T> features) {
     OperationBuilder opBuilder = scope.env().opBuilder("SoftsignGrad", scope.makeOpName("SoftsignGrad"));
-    opBuilder.addInput(gradients.asOutput());
-    opBuilder.addInput(features.asOutput());
+    opBuilder.addInput(gradients.asOutput(scope));
+    opBuilder.addInput(features.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new SoftsignGrad<T>(opBuilder.build());
   }
@@ -60,7 +59,7 @@ public Output<T> backprops() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return backprops;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/SpaceToBatch.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/SpaceToBatch.java
index 93d0d1e3b21..1c51c37b53b 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/SpaceToBatch.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/SpaceToBatch.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * SpaceToBatch for 4-D tensors of type T.
@@ -42,7 +42,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "nn")
-public final class SpaceToBatch<T extends Tensor> extends RawOp implements Operand<T> {
+public final class SpaceToBatch<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new SpaceToBatch operation.
@@ -122,10 +122,10 @@ public final class SpaceToBatch<T extends Tensor> extends RawOp implements Opera
    * @return a new instance of SpaceToBatch
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor, U extends Tensor & TNumber> SpaceToBatch<T> create(Scope scope, Operand<T> input, Operand<U> paddings, Long blockSize) {
+  public static <T extends TType, U extends TNumber> SpaceToBatch<T> create(Scope scope, Operand<T> input, Operand<U> paddings, Long blockSize) {
     OperationBuilder opBuilder = scope.env().opBuilder("SpaceToBatch", scope.makeOpName("SpaceToBatch"));
-    opBuilder.addInput(input.asOutput());
-    opBuilder.addInput(paddings.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
+    opBuilder.addInput(paddings.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("block_size", blockSize);
     return new SpaceToBatch<T>(opBuilder.build());
@@ -138,7 +138,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/SpaceToDepth.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/SpaceToDepth.java
index 448902a1048..404c97f889e 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/SpaceToDepth.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/SpaceToDepth.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * SpaceToDepth for tensors of type T.
@@ -107,7 +107,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "nn")
-public final class SpaceToDepth<T extends Tensor> extends RawOp implements Operand<T> {
+public final class SpaceToDepth<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.nn.SpaceToDepth}
@@ -138,9 +138,9 @@ private Options() {
    * @return a new instance of SpaceToDepth
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> SpaceToDepth<T> create(Scope scope, Operand<T> input, Long blockSize, Options... options) {
+  public static <T extends TType> SpaceToDepth<T> create(Scope scope, Operand<T> input, Long blockSize, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("SpaceToDepth", scope.makeOpName("SpaceToDepth"));
-    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("block_size", blockSize);
     if (options != null) {
@@ -167,7 +167,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/TopK.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/TopK.java
index 5f90a2a5868..cd87a1374bb 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/TopK.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/TopK.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -46,7 +45,7 @@
  * @param <T> data type for {@code values()} output
  */
 @Operator(group = "nn")
-public final class TopK<T extends Tensor & TNumber> extends RawOp {
+public final class TopK<T extends TNumber> extends RawOp {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.nn.TopK}
@@ -79,10 +78,10 @@ private Options() {
    * @return a new instance of TopK
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> TopK<T> create(Scope scope, Operand<T> input, Operand<TInt32> k, Options... options) {
+  public static <T extends TNumber> TopK<T> create(Scope scope, Operand<T> input, Operand<TInt32> k, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("TopKV2", scope.makeOpName("TopK"));
-    opBuilder.addInput(input.asOutput());
-    opBuilder.addInput(k.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
+    opBuilder.addInput(k.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/raw/SoftmaxCrossEntropyWithLogits.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/raw/SoftmaxCrossEntropyWithLogits.java
index 61a8bb03593..32e55c4f4ac 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/raw/SoftmaxCrossEntropyWithLogits.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/raw/SoftmaxCrossEntropyWithLogits.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -36,7 +35,7 @@
  * @param <T> data type for {@code loss()} output
  */
 @Operator(group = "nn")
-public final class SoftmaxCrossEntropyWithLogits<T extends Tensor & TNumber> extends RawOp {
+public final class SoftmaxCrossEntropyWithLogits<T extends TNumber> extends RawOp {
   
   /**
    * Factory method to create a class wrapping a new SoftmaxCrossEntropyWithLogits operation.
@@ -49,10 +48,10 @@ public final class SoftmaxCrossEntropyWithLogits<T extends Tensor & TNumber> ext
    * @return a new instance of SoftmaxCrossEntropyWithLogits
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> SoftmaxCrossEntropyWithLogits<T> create(Scope scope, Operand<T> features, Operand<T> labels) {
+  public static <T extends TNumber> SoftmaxCrossEntropyWithLogits<T> create(Scope scope, Operand<T> features, Operand<T> labels) {
     OperationBuilder opBuilder = scope.env().opBuilder("SoftmaxCrossEntropyWithLogits", scope.makeOpName("SoftmaxCrossEntropyWithLogits"));
-    opBuilder.addInput(features.asOutput());
-    opBuilder.addInput(labels.asOutput());
+    opBuilder.addInput(features.asOutput(scope));
+    opBuilder.addInput(labels.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new SoftmaxCrossEntropyWithLogits<T>(opBuilder.build());
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/raw/SparseSoftmaxCrossEntropyWithLogits.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/raw/SparseSoftmaxCrossEntropyWithLogits.java
index 9ddb57134df..0c62793e97e 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/raw/SparseSoftmaxCrossEntropyWithLogits.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/raw/SparseSoftmaxCrossEntropyWithLogits.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -41,7 +40,7 @@
  * @param <T> data type for {@code loss()} output
  */
 @Operator(group = "nn")
-public final class SparseSoftmaxCrossEntropyWithLogits<T extends Tensor & TNumber> extends RawOp {
+public final class SparseSoftmaxCrossEntropyWithLogits<T extends TNumber> extends RawOp {
   
   /**
    * Factory method to create a class wrapping a new SparseSoftmaxCrossEntropyWithLogits operation.
@@ -53,10 +52,10 @@ public final class SparseSoftmaxCrossEntropyWithLogits<T extends Tensor & TNumbe
    * @return a new instance of SparseSoftmaxCrossEntropyWithLogits
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber, U extends Tensor & TNumber> SparseSoftmaxCrossEntropyWithLogits<T> create(Scope scope, Operand<T> features, Operand<U> labels) {
+  public static <T extends TNumber, U extends TNumber> SparseSoftmaxCrossEntropyWithLogits<T> create(Scope scope, Operand<T> features, Operand<U> labels) {
     OperationBuilder opBuilder = scope.env().opBuilder("SparseSoftmaxCrossEntropyWithLogits", scope.makeOpName("SparseSoftmaxCrossEntropyWithLogits"));
-    opBuilder.addInput(features.asOutput());
-    opBuilder.addInput(labels.asOutput());
+    opBuilder.addInput(features.asOutput(scope));
+    opBuilder.addInput(labels.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new SparseSoftmaxCrossEntropyWithLogits<T>(opBuilder.build());
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/Dequantize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/Dequantize.java
index feeed35e27d..b0bb05757cb 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/Dequantize.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/Dequantize.java
@@ -22,13 +22,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TFloat32;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * Dequantize the 'input' tensor into a float or bfloat16 Tensor.
@@ -85,7 +85,7 @@
  * @param <U> data type for {@code output()} output
  */
 @Operator(group = "quantization")
-public final class Dequantize<U extends Tensor & TNumber> extends RawOp implements Operand<U> {
+public final class Dequantize<U extends TNumber> extends RawOp implements Operand<U> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.quantization.Dequantize}
@@ -137,11 +137,11 @@ private Options() {
    * @return a new instance of Dequantize
    */
   @Endpoint(describeByClass = true)
-  public static <U extends Tensor & TNumber, T extends Tensor> Dequantize<U> create(Scope scope, Operand<T> input, Operand<TFloat32> minRange, Operand<TFloat32> maxRange, DataType<U> dtype, Options... options) {
+  public static <U extends TNumber, T extends TType> Dequantize<U> create(Scope scope, Operand<T> input, Operand<TFloat32> minRange, Operand<TFloat32> maxRange, DataType<U> dtype, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("Dequantize", scope.makeOpName("Dequantize"));
-    opBuilder.addInput(input.asOutput());
-    opBuilder.addInput(minRange.asOutput());
-    opBuilder.addInput(maxRange.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
+    opBuilder.addInput(minRange.asOutput(scope));
+    opBuilder.addInput(maxRange.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("dtype", dtype);
     if (options != null) {
@@ -171,7 +171,7 @@ public static <U extends Tensor & TNumber, T extends Tensor> Dequantize<U> creat
    * @return a new instance of Dequantize
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> Dequantize<TFloat32> create(Scope scope, Operand<T> input, Operand<TFloat32> minRange, Operand<TFloat32> maxRange, Options... options) {
+  public static <T extends TType> Dequantize<TFloat32> create(Scope scope, Operand<T> input, Operand<TFloat32> minRange, Operand<TFloat32> maxRange, Options... options) {
     return create(scope, input, minRange, maxRange, TFloat32.DTYPE, options);
   }
   
@@ -203,7 +203,7 @@ public Output<U> output() {
   }
   
   @Override
-  public Output<U> asOutput() {
+  public Output<U> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/FakeQuantWithMinMaxArgs.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/FakeQuantWithMinMaxArgs.java
index 986097339a9..3cbac1bdf23 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/FakeQuantWithMinMaxArgs.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/FakeQuantWithMinMaxArgs.java
@@ -123,7 +123,7 @@ private Options() {
   @Endpoint(describeByClass = true)
   public static FakeQuantWithMinMaxArgs create(Scope scope, Operand<TFloat32> inputs, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("FakeQuantWithMinMaxArgs", scope.makeOpName("FakeQuantWithMinMaxArgs"));
-    opBuilder.addInput(inputs.asOutput());
+    opBuilder.addInput(inputs.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -179,7 +179,7 @@ public Output<TFloat32> outputs() {
   }
   
   @Override
-  public Output<TFloat32> asOutput() {
+  public Output<TFloat32> asOutput(Scope scope) {
     return outputs;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/FakeQuantWithMinMaxArgsGradient.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/FakeQuantWithMinMaxArgsGradient.java
index 62057eaefed..031d0d71a69 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/FakeQuantWithMinMaxArgsGradient.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/FakeQuantWithMinMaxArgsGradient.java
@@ -91,8 +91,8 @@ private Options() {
   @Endpoint(describeByClass = true)
   public static FakeQuantWithMinMaxArgsGradient create(Scope scope, Operand<TFloat32> gradients, Operand<TFloat32> inputs, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("FakeQuantWithMinMaxArgsGradient", scope.makeOpName("FakeQuantWithMinMaxArgsGradient"));
-    opBuilder.addInput(gradients.asOutput());
-    opBuilder.addInput(inputs.asOutput());
+    opBuilder.addInput(gradients.asOutput(scope));
+    opBuilder.addInput(inputs.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -150,7 +150,7 @@ public Output<TFloat32> backprops() {
   }
   
   @Override
-  public Output<TFloat32> asOutput() {
+  public Output<TFloat32> asOutput(Scope scope) {
     return backprops;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/FakeQuantWithMinMaxVars.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/FakeQuantWithMinMaxVars.java
index b2c1ae3880f..53bdcc5dc0f 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/FakeQuantWithMinMaxVars.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/FakeQuantWithMinMaxVars.java
@@ -111,9 +111,9 @@ private Options() {
   @Endpoint(describeByClass = true)
   public static FakeQuantWithMinMaxVars create(Scope scope, Operand<TFloat32> inputs, Operand<TFloat32> min, Operand<TFloat32> max, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("FakeQuantWithMinMaxVars", scope.makeOpName("FakeQuantWithMinMaxVars"));
-    opBuilder.addInput(inputs.asOutput());
-    opBuilder.addInput(min.asOutput());
-    opBuilder.addInput(max.asOutput());
+    opBuilder.addInput(inputs.asOutput(scope));
+    opBuilder.addInput(min.asOutput(scope));
+    opBuilder.addInput(max.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -149,7 +149,7 @@ public Output<TFloat32> outputs() {
   }
   
   @Override
-  public Output<TFloat32> asOutput() {
+  public Output<TFloat32> asOutput(Scope scope) {
     return outputs;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/FakeQuantWithMinMaxVarsGradient.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/FakeQuantWithMinMaxVarsGradient.java
index e9fcd339a75..f487a5bb806 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/FakeQuantWithMinMaxVarsGradient.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/FakeQuantWithMinMaxVarsGradient.java
@@ -76,10 +76,10 @@ private Options() {
   @Endpoint(describeByClass = true)
   public static FakeQuantWithMinMaxVarsGradient create(Scope scope, Operand<TFloat32> gradients, Operand<TFloat32> inputs, Operand<TFloat32> min, Operand<TFloat32> max, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("FakeQuantWithMinMaxVarsGradient", scope.makeOpName("FakeQuantWithMinMaxVarsGradient"));
-    opBuilder.addInput(gradients.asOutput());
-    opBuilder.addInput(inputs.asOutput());
-    opBuilder.addInput(min.asOutput());
-    opBuilder.addInput(max.asOutput());
+    opBuilder.addInput(gradients.asOutput(scope));
+    opBuilder.addInput(inputs.asOutput(scope));
+    opBuilder.addInput(min.asOutput(scope));
+    opBuilder.addInput(max.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/FakeQuantWithMinMaxVarsPerChannel.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/FakeQuantWithMinMaxVarsPerChannel.java
index ce5649f6c0b..62a5fbe4554 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/FakeQuantWithMinMaxVarsPerChannel.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/FakeQuantWithMinMaxVarsPerChannel.java
@@ -112,9 +112,9 @@ private Options() {
   @Endpoint(describeByClass = true)
   public static FakeQuantWithMinMaxVarsPerChannel create(Scope scope, Operand<TFloat32> inputs, Operand<TFloat32> min, Operand<TFloat32> max, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("FakeQuantWithMinMaxVarsPerChannel", scope.makeOpName("FakeQuantWithMinMaxVarsPerChannel"));
-    opBuilder.addInput(inputs.asOutput());
-    opBuilder.addInput(min.asOutput());
-    opBuilder.addInput(max.asOutput());
+    opBuilder.addInput(inputs.asOutput(scope));
+    opBuilder.addInput(min.asOutput(scope));
+    opBuilder.addInput(max.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -150,7 +150,7 @@ public Output<TFloat32> outputs() {
   }
   
   @Override
-  public Output<TFloat32> asOutput() {
+  public Output<TFloat32> asOutput(Scope scope) {
     return outputs;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/FakeQuantWithMinMaxVarsPerChannelGradient.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/FakeQuantWithMinMaxVarsPerChannelGradient.java
index 4fba9b56115..02fd3200c68 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/FakeQuantWithMinMaxVarsPerChannelGradient.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/FakeQuantWithMinMaxVarsPerChannelGradient.java
@@ -78,10 +78,10 @@ private Options() {
   @Endpoint(describeByClass = true)
   public static FakeQuantWithMinMaxVarsPerChannelGradient create(Scope scope, Operand<TFloat32> gradients, Operand<TFloat32> inputs, Operand<TFloat32> min, Operand<TFloat32> max, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("FakeQuantWithMinMaxVarsPerChannelGradient", scope.makeOpName("FakeQuantWithMinMaxVarsPerChannelGradient"));
-    opBuilder.addInput(gradients.asOutput());
-    opBuilder.addInput(inputs.asOutput());
-    opBuilder.addInput(min.asOutput());
-    opBuilder.addInput(max.asOutput());
+    opBuilder.addInput(gradients.asOutput(scope));
+    opBuilder.addInput(inputs.asOutput(scope));
+    opBuilder.addInput(min.asOutput(scope));
+    opBuilder.addInput(max.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/Quantize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/Quantize.java
index 9e3d5cf80d2..5e9e0e5ca83 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/Quantize.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/Quantize.java
@@ -22,12 +22,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.family.TType;
 
 /**
  * Quantize the 'input' tensor of type float to 'output' tensor of type 'T'.
@@ -145,7 +145,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "quantization")
-public final class Quantize<T extends Tensor> extends RawOp {
+public final class Quantize<T extends TType> extends RawOp {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.quantization.Quantize}
@@ -220,11 +220,11 @@ private Options() {
    * @return a new instance of Quantize
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> Quantize<T> create(Scope scope, Operand<TFloat32> input, Operand<TFloat32> minRange, Operand<TFloat32> maxRange, DataType<T> T, Options... options) {
+  public static <T extends TType> Quantize<T> create(Scope scope, Operand<TFloat32> input, Operand<TFloat32> minRange, Operand<TFloat32> maxRange, DataType<T> T, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("QuantizeV2", scope.makeOpName("Quantize"));
-    opBuilder.addInput(input.asOutput());
-    opBuilder.addInput(minRange.asOutput());
-    opBuilder.addInput(maxRange.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
+    opBuilder.addInput(minRange.asOutput(scope));
+    opBuilder.addInput(maxRange.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("T", T);
     if (options != null) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/QuantizeAndDequantize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/QuantizeAndDequantize.java
index 81d3aeb3d34..aefb818c77d 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/QuantizeAndDequantize.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/QuantizeAndDequantize.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -38,7 +37,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "quantization")
-public final class QuantizeAndDequantize<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class QuantizeAndDequantize<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.quantization.QuantizeAndDequantize}
@@ -98,12 +97,12 @@ private Options() {
    * @return a new instance of QuantizeAndDequantize
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> QuantizeAndDequantize<T> create(Scope scope, Operand<T> input, Operand<T> inputMin, Operand<T> inputMax, Operand<TInt32> numBits, Options... options) {
+  public static <T extends TNumber> QuantizeAndDequantize<T> create(Scope scope, Operand<T> input, Operand<T> inputMin, Operand<T> inputMax, Operand<TInt32> numBits, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("QuantizeAndDequantizeV3", scope.makeOpName("QuantizeAndDequantize"));
-    opBuilder.addInput(input.asOutput());
-    opBuilder.addInput(inputMin.asOutput());
-    opBuilder.addInput(inputMax.asOutput());
-    opBuilder.addInput(numBits.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
+    opBuilder.addInput(inputMin.asOutput(scope));
+    opBuilder.addInput(inputMax.asOutput(scope));
+    opBuilder.addInput(numBits.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -159,7 +158,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/QuantizeDownAndShrinkRange.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/QuantizeDownAndShrinkRange.java
index 39839df983b..849191286c3 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/QuantizeDownAndShrinkRange.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/QuantizeDownAndShrinkRange.java
@@ -22,12 +22,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.family.TType;
 
 /**
  * Convert the quantized 'input' tensor into a lower-precision 'output', using the
@@ -58,7 +58,7 @@
  * @param <U> data type for {@code output()} output
  */
 @Operator(group = "quantization")
-public final class QuantizeDownAndShrinkRange<U extends Tensor> extends RawOp {
+public final class QuantizeDownAndShrinkRange<U extends TType> extends RawOp {
   
   /**
    * Factory method to create a class wrapping a new QuantizeDownAndShrinkRange operation.
@@ -71,11 +71,11 @@ public final class QuantizeDownAndShrinkRange<U extends Tensor> extends RawOp {
    * @return a new instance of QuantizeDownAndShrinkRange
    */
   @Endpoint(describeByClass = true)
-  public static <U extends Tensor, T extends Tensor> QuantizeDownAndShrinkRange<U> create(Scope scope, Operand<T> input, Operand<TFloat32> inputMin, Operand<TFloat32> inputMax, DataType<U> outType) {
+  public static <U extends TType, T extends TType> QuantizeDownAndShrinkRange<U> create(Scope scope, Operand<T> input, Operand<TFloat32> inputMin, Operand<TFloat32> inputMax, DataType<U> outType) {
     OperationBuilder opBuilder = scope.env().opBuilder("QuantizeDownAndShrinkRange", scope.makeOpName("QuantizeDownAndShrinkRange"));
-    opBuilder.addInput(input.asOutput());
-    opBuilder.addInput(inputMin.asOutput());
-    opBuilder.addInput(inputMax.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
+    opBuilder.addInput(inputMin.asOutput(scope));
+    opBuilder.addInput(inputMax.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("out_type", outType);
     return new QuantizeDownAndShrinkRange<U>(opBuilder.build());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/QuantizedConcat.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/QuantizedConcat.java
index acb4e43c5ef..714e49fa347 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/QuantizedConcat.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/QuantizedConcat.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.Operands;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
@@ -29,6 +28,7 @@
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TFloat32;
 import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TType;
 
 /**
  * Concatenates quantized tensors along one dimension.
@@ -36,7 +36,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "quantization")
-public final class QuantizedConcat<T extends Tensor> extends RawOp {
+public final class QuantizedConcat<T extends TType> extends RawOp {
   
   /**
    * Factory method to create a class wrapping a new QuantizedConcat operation.
@@ -51,12 +51,12 @@ public final class QuantizedConcat<T extends Tensor> extends RawOp {
    * @return a new instance of QuantizedConcat
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> QuantizedConcat<T> create(Scope scope, Operand<TInt32> concatDim, Iterable<Operand<T>> values, Iterable<Operand<TFloat32>> inputMins, Iterable<Operand<TFloat32>> inputMaxes) {
+  public static <T extends TType> QuantizedConcat<T> create(Scope scope, Operand<TInt32> concatDim, Iterable<Operand<T>> values, Iterable<Operand<TFloat32>> inputMins, Iterable<Operand<TFloat32>> inputMaxes) {
     OperationBuilder opBuilder = scope.env().opBuilder("QuantizedConcat", scope.makeOpName("QuantizedConcat"));
-    opBuilder.addInput(concatDim.asOutput());
-    opBuilder.addInputList(Operands.asOutputs(values));
-    opBuilder.addInputList(Operands.asOutputs(inputMins));
-    opBuilder.addInputList(Operands.asOutputs(inputMaxes));
+    opBuilder.addInput(concatDim.asOutput(scope));
+    opBuilder.addInputList(Operands.asOutputs(scope, values));
+    opBuilder.addInputList(Operands.asOutputs(scope, inputMins));
+    opBuilder.addInputList(Operands.asOutputs(scope, inputMaxes));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new QuantizedConcat<T>(opBuilder.build());
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/QuantizedMatMulWithBiasAndDequantize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/QuantizedMatMulWithBiasAndDequantize.java
index d6083ce1d21..ef25c6d0b00 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/QuantizedMatMulWithBiasAndDequantize.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/QuantizedMatMulWithBiasAndDequantize.java
@@ -22,18 +22,18 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TFloat32;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * @param <W> data type for {@code out()} output
  */
-public final class QuantizedMatMulWithBiasAndDequantize<W extends Tensor & TNumber> extends RawOp implements Operand<W> {
+public final class QuantizedMatMulWithBiasAndDequantize<W extends TNumber> extends RawOp implements Operand<W> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.quantization.QuantizedMatMulWithBiasAndDequantize}
@@ -90,17 +90,17 @@ private Options() {
    * @return a new instance of QuantizedMatMulWithBiasAndDequantize
    */
   @Endpoint(describeByClass = true)
-  public static <W extends Tensor & TNumber, T extends Tensor, U extends Tensor, V extends Tensor> QuantizedMatMulWithBiasAndDequantize<W> create(Scope scope, Operand<T> a, Operand<U> b, Operand<V> bias, Operand<TFloat32> minA, Operand<TFloat32> maxA, Operand<TFloat32> minB, Operand<TFloat32> maxB, Operand<TFloat32> minFreezedOutput, Operand<TFloat32> maxFreezedOutput, DataType<W> Toutput, Options... options) {
+  public static <W extends TNumber, T extends TType, U extends TType, V extends TType> QuantizedMatMulWithBiasAndDequantize<W> create(Scope scope, Operand<T> a, Operand<U> b, Operand<V> bias, Operand<TFloat32> minA, Operand<TFloat32> maxA, Operand<TFloat32> minB, Operand<TFloat32> maxB, Operand<TFloat32> minFreezedOutput, Operand<TFloat32> maxFreezedOutput, DataType<W> Toutput, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("QuantizedMatMulWithBiasAndDequantize", scope.makeOpName("QuantizedMatMulWithBiasAndDequantize"));
-    opBuilder.addInput(a.asOutput());
-    opBuilder.addInput(b.asOutput());
-    opBuilder.addInput(bias.asOutput());
-    opBuilder.addInput(minA.asOutput());
-    opBuilder.addInput(maxA.asOutput());
-    opBuilder.addInput(minB.asOutput());
-    opBuilder.addInput(maxB.asOutput());
-    opBuilder.addInput(minFreezedOutput.asOutput());
-    opBuilder.addInput(maxFreezedOutput.asOutput());
+    opBuilder.addInput(a.asOutput(scope));
+    opBuilder.addInput(b.asOutput(scope));
+    opBuilder.addInput(bias.asOutput(scope));
+    opBuilder.addInput(minA.asOutput(scope));
+    opBuilder.addInput(maxA.asOutput(scope));
+    opBuilder.addInput(minB.asOutput(scope));
+    opBuilder.addInput(maxB.asOutput(scope));
+    opBuilder.addInput(minFreezedOutput.asOutput(scope));
+    opBuilder.addInput(maxFreezedOutput.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("Toutput", Toutput);
     if (options != null) {
@@ -147,7 +147,7 @@ public Output<W> out() {
   }
   
   @Override
-  public Output<W> asOutput() {
+  public Output<W> asOutput(Scope scope) {
     return out;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/QuantizedMatMulWithBiasAndRequantize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/QuantizedMatMulWithBiasAndRequantize.java
index 7c0bf738249..54ef67fd237 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/QuantizedMatMulWithBiasAndRequantize.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/QuantizedMatMulWithBiasAndRequantize.java
@@ -22,17 +22,17 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.family.TType;
 
 /**
  * @param <W> data type for {@code out()} output
  */
-public final class QuantizedMatMulWithBiasAndRequantize<W extends Tensor> extends RawOp {
+public final class QuantizedMatMulWithBiasAndRequantize<W extends TType> extends RawOp {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.quantization.QuantizedMatMulWithBiasAndRequantize}
@@ -89,17 +89,17 @@ private Options() {
    * @return a new instance of QuantizedMatMulWithBiasAndRequantize
    */
   @Endpoint(describeByClass = true)
-  public static <W extends Tensor, T extends Tensor, U extends Tensor, V extends Tensor> QuantizedMatMulWithBiasAndRequantize<W> create(Scope scope, Operand<T> a, Operand<U> b, Operand<V> bias, Operand<TFloat32> minA, Operand<TFloat32> maxA, Operand<TFloat32> minB, Operand<TFloat32> maxB, Operand<TFloat32> minFreezedOutput, Operand<TFloat32> maxFreezedOutput, DataType<W> Toutput, Options... options) {
+  public static <W extends TType, T extends TType, U extends TType, V extends TType> QuantizedMatMulWithBiasAndRequantize<W> create(Scope scope, Operand<T> a, Operand<U> b, Operand<V> bias, Operand<TFloat32> minA, Operand<TFloat32> maxA, Operand<TFloat32> minB, Operand<TFloat32> maxB, Operand<TFloat32> minFreezedOutput, Operand<TFloat32> maxFreezedOutput, DataType<W> Toutput, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("QuantizedMatMulWithBiasAndRequantize", scope.makeOpName("QuantizedMatMulWithBiasAndRequantize"));
-    opBuilder.addInput(a.asOutput());
-    opBuilder.addInput(b.asOutput());
-    opBuilder.addInput(bias.asOutput());
-    opBuilder.addInput(minA.asOutput());
-    opBuilder.addInput(maxA.asOutput());
-    opBuilder.addInput(minB.asOutput());
-    opBuilder.addInput(maxB.asOutput());
-    opBuilder.addInput(minFreezedOutput.asOutput());
-    opBuilder.addInput(maxFreezedOutput.asOutput());
+    opBuilder.addInput(a.asOutput(scope));
+    opBuilder.addInput(b.asOutput(scope));
+    opBuilder.addInput(bias.asOutput(scope));
+    opBuilder.addInput(minA.asOutput(scope));
+    opBuilder.addInput(maxA.asOutput(scope));
+    opBuilder.addInput(minB.asOutput(scope));
+    opBuilder.addInput(maxB.asOutput(scope));
+    opBuilder.addInput(minFreezedOutput.asOutput(scope));
+    opBuilder.addInput(maxFreezedOutput.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("Toutput", Toutput);
     if (options != null) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/RequantizationRange.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/RequantizationRange.java
index c2cf9076a8d..2a57d9b435d 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/RequantizationRange.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/RequantizationRange.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.family.TType;
 
 /**
  * Computes a range that covers the actual values present in a quantized tensor.
@@ -49,11 +49,11 @@ public final class RequantizationRange extends RawOp {
    * @return a new instance of RequantizationRange
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> RequantizationRange create(Scope scope, Operand<T> input, Operand<TFloat32> inputMin, Operand<TFloat32> inputMax) {
+  public static <T extends TType> RequantizationRange create(Scope scope, Operand<T> input, Operand<TFloat32> inputMin, Operand<TFloat32> inputMax) {
     OperationBuilder opBuilder = scope.env().opBuilder("RequantizationRange", scope.makeOpName("RequantizationRange"));
-    opBuilder.addInput(input.asOutput());
-    opBuilder.addInput(inputMin.asOutput());
-    opBuilder.addInput(inputMax.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
+    opBuilder.addInput(inputMin.asOutput(scope));
+    opBuilder.addInput(inputMax.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new RequantizationRange(opBuilder.build());
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/Requantize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/Requantize.java
index 700ade2cf9b..f0a8541a854 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/Requantize.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/Requantize.java
@@ -22,12 +22,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.family.TType;
 
 /**
  * Converts the quantized `input` tensor into a lower-precision `output`.
@@ -43,7 +43,7 @@
  * @param <U> data type for {@code output()} output
  */
 @Operator(group = "quantization")
-public final class Requantize<U extends Tensor> extends RawOp {
+public final class Requantize<U extends TType> extends RawOp {
   
   /**
    * Factory method to create a class wrapping a new Requantize operation.
@@ -58,13 +58,13 @@ public final class Requantize<U extends Tensor> extends RawOp {
    * @return a new instance of Requantize
    */
   @Endpoint(describeByClass = true)
-  public static <U extends Tensor, T extends Tensor> Requantize<U> create(Scope scope, Operand<T> input, Operand<TFloat32> inputMin, Operand<TFloat32> inputMax, Operand<TFloat32> requestedOutputMin, Operand<TFloat32> requestedOutputMax, DataType<U> outType) {
+  public static <U extends TType, T extends TType> Requantize<U> create(Scope scope, Operand<T> input, Operand<TFloat32> inputMin, Operand<TFloat32> inputMax, Operand<TFloat32> requestedOutputMin, Operand<TFloat32> requestedOutputMax, DataType<U> outType) {
     OperationBuilder opBuilder = scope.env().opBuilder("Requantize", scope.makeOpName("Requantize"));
-    opBuilder.addInput(input.asOutput());
-    opBuilder.addInput(inputMin.asOutput());
-    opBuilder.addInput(inputMax.asOutput());
-    opBuilder.addInput(requestedOutputMin.asOutput());
-    opBuilder.addInput(requestedOutputMax.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
+    opBuilder.addInput(inputMin.asOutput(scope));
+    opBuilder.addInput(inputMax.asOutput(scope));
+    opBuilder.addInput(requestedOutputMin.asOutput(scope));
+    opBuilder.addInput(requestedOutputMax.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("out_type", outType);
     return new Requantize<U>(opBuilder.build());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/ragged/RaggedGather.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/ragged/RaggedGather.java
index d7d1d94d854..fe59b2e4373 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/ragged/RaggedGather.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/ragged/RaggedGather.java
@@ -23,13 +23,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.Operands;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * Gather ragged slices from `params` axis `0` according to `indices`.
@@ -65,7 +65,7 @@
  * @param <T> data type for {@code outputNestedSplits()} output
  * @param <U> data type for {@code outputDenseValues()} output
  */
-public final class RaggedGather<T extends Tensor & TNumber, U extends Tensor> extends RawOp {
+public final class RaggedGather<T extends TNumber, U extends TType> extends RawOp {
   
   /**
    * Factory method to create a class wrapping a new RaggedGather operation.
@@ -84,11 +84,11 @@ public final class RaggedGather<T extends Tensor & TNumber, U extends Tensor> ex
    * @return a new instance of RaggedGather
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber, U extends Tensor, V extends Tensor & TNumber> RaggedGather<T, U> create(Scope scope, Iterable<Operand<T>> paramsNestedSplits, Operand<U> paramsDenseValues, Operand<V> indices, Long OUTPUTRAGGEDRANK) {
+  public static <T extends TNumber, U extends TType, V extends TNumber> RaggedGather<T, U> create(Scope scope, Iterable<Operand<T>> paramsNestedSplits, Operand<U> paramsDenseValues, Operand<V> indices, Long OUTPUTRAGGEDRANK) {
     OperationBuilder opBuilder = scope.env().opBuilder("RaggedGather", scope.makeOpName("RaggedGather"));
-    opBuilder.addInputList(Operands.asOutputs(paramsNestedSplits));
-    opBuilder.addInput(paramsDenseValues.asOutput());
-    opBuilder.addInput(indices.asOutput());
+    opBuilder.addInputList(Operands.asOutputs(scope, paramsNestedSplits));
+    opBuilder.addInput(paramsDenseValues.asOutput(scope));
+    opBuilder.addInput(indices.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("OUTPUT_RAGGED_RANK", OUTPUTRAGGEDRANK);
     return new RaggedGather<T, U>(opBuilder.build());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/ragged/RaggedRange.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/ragged/RaggedRange.java
index c8c28878b19..69e86f2426f 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/ragged/RaggedRange.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/ragged/RaggedRange.java
@@ -22,7 +22,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -51,7 +50,7 @@
  * @param <U> data type for {@code rtNestedSplits()} output
  * @param <T> data type for {@code rtDenseValues()} output
  */
-public final class RaggedRange<U extends Tensor & TNumber, T extends Tensor & TNumber> extends RawOp {
+public final class RaggedRange<U extends TNumber, T extends TNumber> extends RawOp {
   
   /**
    * Factory method to create a class wrapping a new RaggedRange operation.
@@ -64,11 +63,11 @@ public final class RaggedRange<U extends Tensor & TNumber, T extends Tensor & TN
    * @return a new instance of RaggedRange
    */
   @Endpoint(describeByClass = true)
-  public static <U extends Tensor & TNumber, T extends Tensor & TNumber> RaggedRange<U, T> create(Scope scope, Operand<T> starts, Operand<T> limits, Operand<T> deltas, DataType<U> Tsplits) {
+  public static <U extends TNumber, T extends TNumber> RaggedRange<U, T> create(Scope scope, Operand<T> starts, Operand<T> limits, Operand<T> deltas, DataType<U> Tsplits) {
     OperationBuilder opBuilder = scope.env().opBuilder("RaggedRange", scope.makeOpName("RaggedRange"));
-    opBuilder.addInput(starts.asOutput());
-    opBuilder.addInput(limits.asOutput());
-    opBuilder.addInput(deltas.asOutput());
+    opBuilder.addInput(starts.asOutput(scope));
+    opBuilder.addInput(limits.asOutput(scope));
+    opBuilder.addInput(deltas.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("Tsplits", Tsplits);
     return new RaggedRange<U, T>(opBuilder.build());
@@ -84,7 +83,7 @@ public static <U extends Tensor & TNumber, T extends Tensor & TNumber> RaggedRan
    * @return a new instance of RaggedRange
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> RaggedRange<TInt64, T> create(Scope scope, Operand<T> starts, Operand<T> limits, Operand<T> deltas) {
+  public static <T extends TNumber> RaggedRange<TInt64, T> create(Scope scope, Operand<T> starts, Operand<T> limits, Operand<T> deltas) {
     return create(scope, starts, limits, deltas, TInt64.DTYPE);
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/ragged/RaggedTensorFromVariant.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/ragged/RaggedTensorFromVariant.java
index 36d159044e6..e2170e1e751 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/ragged/RaggedTensorFromVariant.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/ragged/RaggedTensorFromVariant.java
@@ -24,13 +24,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * Decodes a `variant` Tensor into a `RaggedTensor`.
@@ -51,7 +51,7 @@
  * @param <U> data type for {@code outputNestedSplits()} output
  * @param <T> data type for {@code outputDenseValues()} output
  */
-public final class RaggedTensorFromVariant<U extends Tensor & TNumber, T extends Tensor> extends RawOp {
+public final class RaggedTensorFromVariant<U extends TNumber, T extends TType> extends RawOp {
   
   /**
    * Factory method to create a class wrapping a new RaggedTensorFromVariant operation.
@@ -67,9 +67,9 @@ public final class RaggedTensorFromVariant<U extends Tensor & TNumber, T extends
    * @return a new instance of RaggedTensorFromVariant
    */
   @Endpoint(describeByClass = true)
-  public static <U extends Tensor & TNumber, T extends Tensor> RaggedTensorFromVariant<U, T> create(Scope scope, Operand<?> encodedRagged, Long inputRaggedRank, Long outputRaggedRank, DataType<T> Tvalues, DataType<U> Tsplits) {
+  public static <U extends TNumber, T extends TType> RaggedTensorFromVariant<U, T> create(Scope scope, Operand<?> encodedRagged, Long inputRaggedRank, Long outputRaggedRank, DataType<T> Tvalues, DataType<U> Tsplits) {
     OperationBuilder opBuilder = scope.env().opBuilder("RaggedTensorFromVariant", scope.makeOpName("RaggedTensorFromVariant"));
-    opBuilder.addInput(encodedRagged.asOutput());
+    opBuilder.addInput(encodedRagged.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("input_ragged_rank", inputRaggedRank);
     opBuilder.setAttr("output_ragged_rank", outputRaggedRank);
@@ -91,7 +91,7 @@ public static <U extends Tensor & TNumber, T extends Tensor> RaggedTensorFromVar
    * @return a new instance of RaggedTensorFromVariant
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> RaggedTensorFromVariant<TInt64, T> create(Scope scope, Operand<?> encodedRagged, Long inputRaggedRank, Long outputRaggedRank, DataType<T> Tvalues) {
+  public static <T extends TType> RaggedTensorFromVariant<TInt64, T> create(Scope scope, Operand<?> encodedRagged, Long inputRaggedRank, Long outputRaggedRank, DataType<T> Tvalues) {
     return create(scope, encodedRagged, inputRaggedRank, outputRaggedRank, Tvalues, TInt64.DTYPE);
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/ragged/RaggedTensorToSparse.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/ragged/RaggedTensorToSparse.java
index 33db48e4972..5ca5d799c67 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/ragged/RaggedTensorToSparse.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/ragged/RaggedTensorToSparse.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.Operands;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
@@ -29,6 +28,7 @@
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * Converts a `RaggedTensor` into a `SparseTensor` with the same values.
@@ -39,7 +39,7 @@
  * 
  * @param <U> data type for {@code sparseValues()} output
  */
-public final class RaggedTensorToSparse<U extends Tensor> extends RawOp {
+public final class RaggedTensorToSparse<U extends TType> extends RawOp {
   
   /**
    * Factory method to create a class wrapping a new RaggedTensorToSparse operation.
@@ -50,10 +50,10 @@ public final class RaggedTensorToSparse<U extends Tensor> extends RawOp {
    * @return a new instance of RaggedTensorToSparse
    */
   @Endpoint(describeByClass = true)
-  public static <U extends Tensor, T extends Tensor & TNumber> RaggedTensorToSparse<U> create(Scope scope, Iterable<Operand<T>> rtNestedSplits, Operand<U> rtDenseValues) {
+  public static <U extends TType, T extends TNumber> RaggedTensorToSparse<U> create(Scope scope, Iterable<Operand<T>> rtNestedSplits, Operand<U> rtDenseValues) {
     OperationBuilder opBuilder = scope.env().opBuilder("RaggedTensorToSparse", scope.makeOpName("RaggedTensorToSparse"));
-    opBuilder.addInputList(Operands.asOutputs(rtNestedSplits));
-    opBuilder.addInput(rtDenseValues.asOutput());
+    opBuilder.addInputList(Operands.asOutputs(scope, rtNestedSplits));
+    opBuilder.addInput(rtDenseValues.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new RaggedTensorToSparse<U>(opBuilder.build());
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/ragged/RaggedTensorToTensor.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/ragged/RaggedTensorToTensor.java
index 7b29dd7e156..589c0a5e100 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/ragged/RaggedTensorToTensor.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/ragged/RaggedTensorToTensor.java
@@ -22,13 +22,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.Operands;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * Create a dense tensor from a ragged tensor, possibly altering its shape.
@@ -58,7 +58,7 @@
  * 
  * @param <U> data type for {@code result()} output
  */
-public final class RaggedTensorToTensor<U extends Tensor> extends RawOp implements Operand<U> {
+public final class RaggedTensorToTensor<U extends TType> extends RawOp implements Operand<U> {
   
   /**
    * Factory method to create a class wrapping a new RaggedTensorToTensor operation.
@@ -106,12 +106,12 @@ public final class RaggedTensorToTensor<U extends Tensor> extends RawOp implemen
    * @return a new instance of RaggedTensorToTensor
    */
   @Endpoint(describeByClass = true)
-  public static <U extends Tensor, T extends Tensor & TNumber, V extends Tensor & TNumber> RaggedTensorToTensor<U> create(Scope scope, Operand<T> shape, Operand<U> values, Operand<U> defaultValue, Iterable<Operand<V>> rowPartitionTensors, List<String> rowPartitionTypes) {
+  public static <U extends TType, T extends TNumber, V extends TNumber> RaggedTensorToTensor<U> create(Scope scope, Operand<T> shape, Operand<U> values, Operand<U> defaultValue, Iterable<Operand<V>> rowPartitionTensors, List<String> rowPartitionTypes) {
     OperationBuilder opBuilder = scope.env().opBuilder("RaggedTensorToTensor", scope.makeOpName("RaggedTensorToTensor"));
-    opBuilder.addInput(shape.asOutput());
-    opBuilder.addInput(values.asOutput());
-    opBuilder.addInput(defaultValue.asOutput());
-    opBuilder.addInputList(Operands.asOutputs(rowPartitionTensors));
+    opBuilder.addInput(shape.asOutput(scope));
+    opBuilder.addInput(values.asOutput(scope));
+    opBuilder.addInput(defaultValue.asOutput(scope));
+    opBuilder.addInputList(Operands.asOutputs(scope, rowPartitionTensors));
     opBuilder = scope.applyControlDependencies(opBuilder);
     String[] rowPartitionTypesArray = new String[rowPartitionTypes.size()];
     for (int i = 0; i < rowPartitionTypesArray.length; ++i) {
@@ -129,7 +129,7 @@ public Output<U> result() {
   }
   
   @Override
-  public Output<U> asOutput() {
+  public Output<U> asOutput(Scope scope) {
     return result;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/ragged/RaggedTensorToVariant.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/ragged/RaggedTensorToVariant.java
index 761f25691b6..1a48b91b9f5 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/ragged/RaggedTensorToVariant.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/ragged/RaggedTensorToVariant.java
@@ -21,13 +21,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.Operands;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * Encodes a `RaggedTensor` into a `variant` Tensor.
@@ -45,7 +45,7 @@
  * corresponding decoding logic.
  * 
  */
-public final class RaggedTensorToVariant extends RawOp implements Operand<Tensor> {
+public final class RaggedTensorToVariant extends RawOp implements Operand<TType> {
   
   /**
    * Factory method to create a class wrapping a new RaggedTensorToVariant operation.
@@ -58,10 +58,10 @@ public final class RaggedTensorToVariant extends RawOp implements Operand<Tensor
    * @return a new instance of RaggedTensorToVariant
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber, U extends Tensor> RaggedTensorToVariant create(Scope scope, Iterable<Operand<T>> rtNestedSplits, Operand<U> rtDenseValues, Boolean batchedInput) {
+  public static <T extends TNumber, U extends TType> RaggedTensorToVariant create(Scope scope, Iterable<Operand<T>> rtNestedSplits, Operand<U> rtDenseValues, Boolean batchedInput) {
     OperationBuilder opBuilder = scope.env().opBuilder("RaggedTensorToVariant", scope.makeOpName("RaggedTensorToVariant"));
-    opBuilder.addInputList(Operands.asOutputs(rtNestedSplits));
-    opBuilder.addInput(rtDenseValues.asOutput());
+    opBuilder.addInputList(Operands.asOutputs(scope, rtNestedSplits));
+    opBuilder.addInput(rtDenseValues.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("batched_input", batchedInput);
     return new RaggedTensorToVariant(opBuilder.build());
@@ -76,8 +76,8 @@ public Output<?> encodedRagged() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<Tensor> asOutput() {
-    return (Output<Tensor>) encodedRagged;
+  public Output<TType> asOutput(Scope scope) {
+    return (Output<TType>) encodedRagged;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/AllCandidateSampler.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/AllCandidateSampler.java
index de24d12678f..1f495de2599 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/AllCandidateSampler.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/AllCandidateSampler.java
@@ -91,7 +91,7 @@ private Options() {
   @Endpoint(describeByClass = true)
   public static AllCandidateSampler create(Scope scope, Operand<TInt64> trueClasses, Long numTrue, Long numSampled, Boolean unique, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("AllCandidateSampler", scope.makeOpName("AllCandidateSampler"));
-    opBuilder.addInput(trueClasses.asOutput());
+    opBuilder.addInput(trueClasses.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("num_true", numTrue);
     opBuilder.setAttr("num_sampled", numSampled);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/AnonymousRandomSeedGenerator.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/AnonymousRandomSeedGenerator.java
index a87026cd7f1..3e5f52a8c9b 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/AnonymousRandomSeedGenerator.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/AnonymousRandomSeedGenerator.java
@@ -42,8 +42,8 @@ public final class AnonymousRandomSeedGenerator extends RawOp {
   @Endpoint(describeByClass = true)
   public static AnonymousRandomSeedGenerator create(Scope scope, Operand<TInt64> seed, Operand<TInt64> seed2) {
     OperationBuilder opBuilder = scope.env().opBuilder("AnonymousRandomSeedGenerator", scope.makeOpName("AnonymousRandomSeedGenerator"));
-    opBuilder.addInput(seed.asOutput());
-    opBuilder.addInput(seed2.asOutput());
+    opBuilder.addInput(seed.asOutput(scope));
+    opBuilder.addInput(seed2.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new AnonymousRandomSeedGenerator(opBuilder.build());
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/DeleteRandomSeedGenerator.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/DeleteRandomSeedGenerator.java
index 23b154f9d75..9b86275f453 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/DeleteRandomSeedGenerator.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/DeleteRandomSeedGenerator.java
@@ -40,8 +40,8 @@ public final class DeleteRandomSeedGenerator extends RawOp {
   @Endpoint(describeByClass = true)
   public static DeleteRandomSeedGenerator create(Scope scope, Operand<?> handle, Operand<?> deleter) {
     OperationBuilder opBuilder = scope.env().opBuilder("DeleteRandomSeedGenerator", scope.makeOpName("DeleteRandomSeedGenerator"));
-    opBuilder.addInput(handle.asOutput());
-    opBuilder.addInput(deleter.asOutput());
+    opBuilder.addInput(handle.asOutput(scope));
+    opBuilder.addInput(deleter.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new DeleteRandomSeedGenerator(opBuilder.build());
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/LogUniformCandidateSampler.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/LogUniformCandidateSampler.java
index 761f945de1a..2e493705a72 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/LogUniformCandidateSampler.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/LogUniformCandidateSampler.java
@@ -92,7 +92,7 @@ private Options() {
   @Endpoint(describeByClass = true)
   public static LogUniformCandidateSampler create(Scope scope, Operand<TInt64> trueClasses, Long numTrue, Long numSampled, Boolean unique, Long rangeMax, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("LogUniformCandidateSampler", scope.makeOpName("LogUniformCandidateSampler"));
-    opBuilder.addInput(trueClasses.asOutput());
+    opBuilder.addInput(trueClasses.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("num_true", numTrue);
     opBuilder.setAttr("num_sampled", numSampled);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/Multinomial.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/Multinomial.java
index 1d2d115ad87..3236ccf8bd7 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/Multinomial.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/Multinomial.java
@@ -22,7 +22,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -37,7 +36,7 @@
  * @param <U> data type for {@code output()} output
  */
 @Operator(group = "random")
-public final class Multinomial<U extends Tensor & TNumber> extends RawOp implements Operand<U> {
+public final class Multinomial<U extends TNumber> extends RawOp implements Operand<U> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.random.Multinomial}
@@ -80,10 +79,10 @@ private Options() {
    * @return a new instance of Multinomial
    */
   @Endpoint(describeByClass = true)
-  public static <U extends Tensor & TNumber, T extends Tensor & TNumber> Multinomial<U> create(Scope scope, Operand<T> logits, Operand<TInt32> numSamples, DataType<U> outputDtype, Options... options) {
+  public static <U extends TNumber, T extends TNumber> Multinomial<U> create(Scope scope, Operand<T> logits, Operand<TInt32> numSamples, DataType<U> outputDtype, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("Multinomial", scope.makeOpName("Multinomial"));
-    opBuilder.addInput(logits.asOutput());
-    opBuilder.addInput(numSamples.asOutput());
+    opBuilder.addInput(logits.asOutput(scope));
+    opBuilder.addInput(numSamples.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("output_dtype", outputDtype);
     if (options != null) {
@@ -110,7 +109,7 @@ public static <U extends Tensor & TNumber, T extends Tensor & TNumber> Multinomi
    * @return a new instance of Multinomial
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> Multinomial<TInt64> create(Scope scope, Operand<T> logits, Operand<TInt32> numSamples, Options... options) {
+  public static <T extends TNumber> Multinomial<TInt64> create(Scope scope, Operand<T> logits, Operand<TInt32> numSamples, Options... options) {
     return create(scope, logits, numSamples, TInt64.DTYPE, options);
   }
   
@@ -138,7 +137,7 @@ public Output<U> output() {
   }
   
   @Override
-  public Output<U> asOutput() {
+  public Output<U> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/NonDeterministicInts.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/NonDeterministicInts.java
index 48ad4545107..ea618bf846a 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/NonDeterministicInts.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/NonDeterministicInts.java
@@ -22,12 +22,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
 
 /**
  * Non-deterministically generates some integers.
@@ -36,7 +36,7 @@
  * 
  * @param <U> data type for {@code output()} output
  */
-public final class NonDeterministicInts<U extends Tensor> extends RawOp implements Operand<U> {
+public final class NonDeterministicInts<U extends TType> extends RawOp implements Operand<U> {
   
   /**
    * Factory method to create a class wrapping a new NonDeterministicInts operation.
@@ -47,9 +47,9 @@ public final class NonDeterministicInts<U extends Tensor> extends RawOp implemen
    * @return a new instance of NonDeterministicInts
    */
   @Endpoint(describeByClass = true)
-  public static <U extends Tensor, T extends Tensor> NonDeterministicInts<U> create(Scope scope, Operand<T> shape, DataType<U> dtype) {
+  public static <U extends TType, T extends TType> NonDeterministicInts<U> create(Scope scope, Operand<T> shape, DataType<U> dtype) {
     OperationBuilder opBuilder = scope.env().opBuilder("NonDeterministicInts", scope.makeOpName("NonDeterministicInts"));
-    opBuilder.addInput(shape.asOutput());
+    opBuilder.addInput(shape.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("dtype", dtype);
     return new NonDeterministicInts<U>(opBuilder.build());
@@ -63,7 +63,7 @@ public static <U extends Tensor, T extends Tensor> NonDeterministicInts<U> creat
    * @return a new instance of NonDeterministicInts
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> NonDeterministicInts<TInt64> create(Scope scope, Operand<T> shape) {
+  public static <T extends TType> NonDeterministicInts<TInt64> create(Scope scope, Operand<T> shape) {
     return create(scope, shape, TInt64.DTYPE);
   }
   
@@ -75,7 +75,7 @@ public Output<U> output() {
   }
   
   @Override
-  public Output<U> asOutput() {
+  public Output<U> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/ParameterizedTruncatedNormal.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/ParameterizedTruncatedNormal.java
index 0a8871772a9..76dc7d3417c 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/ParameterizedTruncatedNormal.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/ParameterizedTruncatedNormal.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -37,7 +36,7 @@
  * @param <U> data type for {@code output()} output
  */
 @Operator(group = "random")
-public final class ParameterizedTruncatedNormal<U extends Tensor & TNumber> extends RawOp implements Operand<U> {
+public final class ParameterizedTruncatedNormal<U extends TNumber> extends RawOp implements Operand<U> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.random.ParameterizedTruncatedNormal}
@@ -83,13 +82,13 @@ private Options() {
    * @return a new instance of ParameterizedTruncatedNormal
    */
   @Endpoint(describeByClass = true)
-  public static <U extends Tensor & TNumber, T extends Tensor & TNumber> ParameterizedTruncatedNormal<U> create(Scope scope, Operand<T> shape, Operand<U> means, Operand<U> stdevs, Operand<U> minvals, Operand<U> maxvals, Options... options) {
+  public static <U extends TNumber, T extends TNumber> ParameterizedTruncatedNormal<U> create(Scope scope, Operand<T> shape, Operand<U> means, Operand<U> stdevs, Operand<U> minvals, Operand<U> maxvals, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ParameterizedTruncatedNormal", scope.makeOpName("ParameterizedTruncatedNormal"));
-    opBuilder.addInput(shape.asOutput());
-    opBuilder.addInput(means.asOutput());
-    opBuilder.addInput(stdevs.asOutput());
-    opBuilder.addInput(minvals.asOutput());
-    opBuilder.addInput(maxvals.asOutput());
+    opBuilder.addInput(shape.asOutput(scope));
+    opBuilder.addInput(means.asOutput(scope));
+    opBuilder.addInput(stdevs.asOutput(scope));
+    opBuilder.addInput(minvals.asOutput(scope));
+    opBuilder.addInput(maxvals.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -129,7 +128,7 @@ public Output<U> output() {
   }
   
   @Override
-  public Output<U> asOutput() {
+  public Output<U> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RandomGamma.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RandomGamma.java
index 47aaf11c1a4..0f9fc373468 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RandomGamma.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RandomGamma.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -38,7 +37,7 @@
  * @param <U> data type for {@code output()} output
  */
 @Operator(group = "random")
-public final class RandomGamma<U extends Tensor & TNumber> extends RawOp implements Operand<U> {
+public final class RandomGamma<U extends TNumber> extends RawOp implements Operand<U> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.random.RandomGamma}
@@ -82,10 +81,10 @@ private Options() {
    * @return a new instance of RandomGamma
    */
   @Endpoint(describeByClass = true)
-  public static <U extends Tensor & TNumber, T extends Tensor & TNumber> RandomGamma<U> create(Scope scope, Operand<T> shape, Operand<U> alpha, Options... options) {
+  public static <U extends TNumber, T extends TNumber> RandomGamma<U> create(Scope scope, Operand<T> shape, Operand<U> alpha, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("RandomGamma", scope.makeOpName("RandomGamma"));
-    opBuilder.addInput(shape.asOutput());
-    opBuilder.addInput(alpha.asOutput());
+    opBuilder.addInput(shape.asOutput(scope));
+    opBuilder.addInput(alpha.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -126,7 +125,7 @@ public Output<U> output() {
   }
   
   @Override
-  public Output<U> asOutput() {
+  public Output<U> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RandomGammaGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RandomGammaGrad.java
index d66bc32b835..822cb158ebd 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RandomGammaGrad.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RandomGammaGrad.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -33,7 +32,7 @@
  * 
  * @param <T> data type for {@code output()} output
  */
-public final class RandomGammaGrad<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class RandomGammaGrad<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new RandomGammaGrad operation.
@@ -44,10 +43,10 @@ public final class RandomGammaGrad<T extends Tensor & TNumber> extends RawOp imp
    * @return a new instance of RandomGammaGrad
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> RandomGammaGrad<T> create(Scope scope, Operand<T> alpha, Operand<T> sample) {
+  public static <T extends TNumber> RandomGammaGrad<T> create(Scope scope, Operand<T> alpha, Operand<T> sample) {
     OperationBuilder opBuilder = scope.env().opBuilder("RandomGammaGrad", scope.makeOpName("RandomGammaGrad"));
-    opBuilder.addInput(alpha.asOutput());
-    opBuilder.addInput(sample.asOutput());
+    opBuilder.addInput(alpha.asOutput(scope));
+    opBuilder.addInput(sample.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new RandomGammaGrad<T>(opBuilder.build());
   }
@@ -59,7 +58,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RandomPoisson.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RandomPoisson.java
index c112d5a0def..f3934e43c6b 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RandomPoisson.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RandomPoisson.java
@@ -22,7 +22,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -46,7 +45,7 @@
  * @param <V> data type for {@code output()} output
  */
 @Operator(group = "random")
-public final class RandomPoisson<V extends Tensor & TNumber> extends RawOp implements Operand<V> {
+public final class RandomPoisson<V extends TNumber> extends RawOp implements Operand<V> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.random.RandomPoisson}
@@ -91,10 +90,10 @@ private Options() {
    * @return a new instance of RandomPoisson
    */
   @Endpoint(describeByClass = true)
-  public static <V extends Tensor & TNumber, T extends Tensor & TNumber, U extends Tensor & TNumber> RandomPoisson<V> create(Scope scope, Operand<T> shape, Operand<U> rate, DataType<V> dtype, Options... options) {
+  public static <V extends TNumber, T extends TNumber, U extends TNumber> RandomPoisson<V> create(Scope scope, Operand<T> shape, Operand<U> rate, DataType<V> dtype, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("RandomPoissonV2", scope.makeOpName("RandomPoisson"));
-    opBuilder.addInput(shape.asOutput());
-    opBuilder.addInput(rate.asOutput());
+    opBuilder.addInput(shape.asOutput(scope));
+    opBuilder.addInput(rate.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("dtype", dtype);
     if (options != null) {
@@ -122,7 +121,7 @@ public static <V extends Tensor & TNumber, T extends Tensor & TNumber, U extends
    * @return a new instance of RandomPoisson
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber, U extends Tensor & TNumber> RandomPoisson<TInt64> create(Scope scope, Operand<T> shape, Operand<U> rate, Options... options) {
+  public static <T extends TNumber, U extends TNumber> RandomPoisson<TInt64> create(Scope scope, Operand<T> shape, Operand<U> rate, Options... options) {
     return create(scope, shape, rate, TInt64.DTYPE, options);
   }
   
@@ -152,7 +151,7 @@ public Output<V> output() {
   }
   
   @Override
-  public Output<V> asOutput() {
+  public Output<V> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RandomShuffle.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RandomShuffle.java
index afe6ca061a4..2d7adf20727 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RandomShuffle.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RandomShuffle.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Randomly shuffles a tensor along its first dimension.
@@ -43,7 +43,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "random")
-public final class RandomShuffle<T extends Tensor> extends RawOp implements Operand<T> {
+public final class RandomShuffle<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.random.RandomShuffle}
@@ -84,9 +84,9 @@ private Options() {
    * @return a new instance of RandomShuffle
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> RandomShuffle<T> create(Scope scope, Operand<T> value, Options... options) {
+  public static <T extends TType> RandomShuffle<T> create(Scope scope, Operand<T> value, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("RandomShuffle", scope.makeOpName("RandomShuffle"));
-    opBuilder.addInput(value.asOutput());
+    opBuilder.addInput(value.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -126,7 +126,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RandomStandardNormal.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RandomStandardNormal.java
index 041a1ac0adb..a6f61546deb 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RandomStandardNormal.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RandomStandardNormal.java
@@ -22,7 +22,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -37,7 +36,7 @@
  * @param <U> data type for {@code output()} output
  */
 @Operator(group = "random")
-public final class RandomStandardNormal<U extends Tensor & TNumber> extends RawOp implements Operand<U> {
+public final class RandomStandardNormal<U extends TNumber> extends RawOp implements Operand<U> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.random.RandomStandardNormal}
@@ -79,9 +78,9 @@ private Options() {
    * @return a new instance of RandomStandardNormal
    */
   @Endpoint(describeByClass = true)
-  public static <U extends Tensor & TNumber, T extends Tensor & TNumber> RandomStandardNormal<U> create(Scope scope, Operand<T> shape, DataType<U> dtype, Options... options) {
+  public static <U extends TNumber, T extends TNumber> RandomStandardNormal<U> create(Scope scope, Operand<T> shape, DataType<U> dtype, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("RandomStandardNormal", scope.makeOpName("RandomStandardNormal"));
-    opBuilder.addInput(shape.asOutput());
+    opBuilder.addInput(shape.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("dtype", dtype);
     if (options != null) {
@@ -121,7 +120,7 @@ public Output<U> output() {
   }
   
   @Override
-  public Output<U> asOutput() {
+  public Output<U> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RandomUniform.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RandomUniform.java
index d82f0cda1bf..81ceb5de601 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RandomUniform.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RandomUniform.java
@@ -22,7 +22,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -38,7 +37,7 @@
  * @param <U> data type for {@code output()} output
  */
 @Operator(group = "random")
-public final class RandomUniform<U extends Tensor & TNumber> extends RawOp implements Operand<U> {
+public final class RandomUniform<U extends TNumber> extends RawOp implements Operand<U> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.random.RandomUniform}
@@ -80,9 +79,9 @@ private Options() {
    * @return a new instance of RandomUniform
    */
   @Endpoint(describeByClass = true)
-  public static <U extends Tensor & TNumber, T extends Tensor & TNumber> RandomUniform<U> create(Scope scope, Operand<T> shape, DataType<U> dtype, Options... options) {
+  public static <U extends TNumber, T extends TNumber> RandomUniform<U> create(Scope scope, Operand<T> shape, DataType<U> dtype, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("RandomUniform", scope.makeOpName("RandomUniform"));
-    opBuilder.addInput(shape.asOutput());
+    opBuilder.addInput(shape.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("dtype", dtype);
     if (options != null) {
@@ -122,7 +121,7 @@ public Output<U> output() {
   }
   
   @Override
-  public Output<U> asOutput() {
+  public Output<U> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RandomUniformInt.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RandomUniformInt.java
index 893b03fdfde..ecabd1fa6f8 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RandomUniformInt.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RandomUniformInt.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -42,7 +41,7 @@
  * @param <U> data type for {@code output()} output
  */
 @Operator(group = "random")
-public final class RandomUniformInt<U extends Tensor & TNumber> extends RawOp implements Operand<U> {
+public final class RandomUniformInt<U extends TNumber> extends RawOp implements Operand<U> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.random.RandomUniformInt}
@@ -85,11 +84,11 @@ private Options() {
    * @return a new instance of RandomUniformInt
    */
   @Endpoint(describeByClass = true)
-  public static <U extends Tensor & TNumber, T extends Tensor & TNumber> RandomUniformInt<U> create(Scope scope, Operand<T> shape, Operand<U> minval, Operand<U> maxval, Options... options) {
+  public static <U extends TNumber, T extends TNumber> RandomUniformInt<U> create(Scope scope, Operand<T> shape, Operand<U> minval, Operand<U> maxval, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("RandomUniformInt", scope.makeOpName("RandomUniformInt"));
-    opBuilder.addInput(shape.asOutput());
-    opBuilder.addInput(minval.asOutput());
-    opBuilder.addInput(maxval.asOutput());
+    opBuilder.addInput(shape.asOutput(scope));
+    opBuilder.addInput(minval.asOutput(scope));
+    opBuilder.addInput(maxval.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -128,7 +127,7 @@ public Output<U> output() {
   }
   
   @Override
-  public Output<U> asOutput() {
+  public Output<U> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RecordInput.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RecordInput.java
index 7e3550cb71c..b409b0f4674 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RecordInput.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RecordInput.java
@@ -189,7 +189,7 @@ public Output<TString> records() {
   }
   
   @Override
-  public Output<TString> asOutput() {
+  public Output<TString> asOutput(Scope scope) {
     return records;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RngSkip.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RngSkip.java
index f41cff35b04..a353b0e861c 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RngSkip.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RngSkip.java
@@ -48,9 +48,9 @@ public final class RngSkip extends RawOp {
   @Endpoint(describeByClass = true)
   public static RngSkip create(Scope scope, Operand<?> resource, Operand<TInt64> algorithm, Operand<TInt64> delta) {
     OperationBuilder opBuilder = scope.env().opBuilder("RngSkip", scope.makeOpName("RngSkip"));
-    opBuilder.addInput(resource.asOutput());
-    opBuilder.addInput(algorithm.asOutput());
-    opBuilder.addInput(delta.asOutput());
+    opBuilder.addInput(resource.asOutput(scope));
+    opBuilder.addInput(algorithm.asOutput(scope));
+    opBuilder.addInput(delta.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new RngSkip(opBuilder.build());
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatefulRandomBinomial.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatefulRandomBinomial.java
index 4a5539b1bb7..4c82bce1aac 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatefulRandomBinomial.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatefulRandomBinomial.java
@@ -22,7 +22,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -34,7 +33,7 @@
  * @param <V> data type for {@code output()} output
  */
 @Operator(group = "random")
-public final class StatefulRandomBinomial<V extends Tensor & TNumber> extends RawOp implements Operand<V> {
+public final class StatefulRandomBinomial<V extends TNumber> extends RawOp implements Operand<V> {
   
   /**
    * Factory method to create a class wrapping a new StatefulRandomBinomial operation.
@@ -49,13 +48,13 @@ public final class StatefulRandomBinomial<V extends Tensor & TNumber> extends Ra
    * @return a new instance of StatefulRandomBinomial
    */
   @Endpoint(describeByClass = true)
-  public static <V extends Tensor & TNumber, T extends Tensor & TNumber, U extends Tensor & TNumber> StatefulRandomBinomial<V> create(Scope scope, Operand<?> resource, Operand<TInt64> algorithm, Operand<T> shape, Operand<U> counts, Operand<U> probs, DataType<V> dtype) {
+  public static <V extends TNumber, T extends TNumber, U extends TNumber> StatefulRandomBinomial<V> create(Scope scope, Operand<?> resource, Operand<TInt64> algorithm, Operand<T> shape, Operand<U> counts, Operand<U> probs, DataType<V> dtype) {
     OperationBuilder opBuilder = scope.env().opBuilder("StatefulRandomBinomial", scope.makeOpName("StatefulRandomBinomial"));
-    opBuilder.addInput(resource.asOutput());
-    opBuilder.addInput(algorithm.asOutput());
-    opBuilder.addInput(shape.asOutput());
-    opBuilder.addInput(counts.asOutput());
-    opBuilder.addInput(probs.asOutput());
+    opBuilder.addInput(resource.asOutput(scope));
+    opBuilder.addInput(algorithm.asOutput(scope));
+    opBuilder.addInput(shape.asOutput(scope));
+    opBuilder.addInput(counts.asOutput(scope));
+    opBuilder.addInput(probs.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("dtype", dtype);
     return new StatefulRandomBinomial<V>(opBuilder.build());
@@ -73,7 +72,7 @@ public static <V extends Tensor & TNumber, T extends Tensor & TNumber, U extends
    * @return a new instance of StatefulRandomBinomial
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber, U extends Tensor & TNumber> StatefulRandomBinomial<TInt64> create(Scope scope, Operand<?> resource, Operand<TInt64> algorithm, Operand<T> shape, Operand<U> counts, Operand<U> probs) {
+  public static <T extends TNumber, U extends TNumber> StatefulRandomBinomial<TInt64> create(Scope scope, Operand<?> resource, Operand<TInt64> algorithm, Operand<T> shape, Operand<U> counts, Operand<U> probs) {
     return create(scope, resource, algorithm, shape, counts, probs, TInt64.DTYPE);
   }
   
@@ -84,7 +83,7 @@ public Output<V> output() {
   }
   
   @Override
-  public Output<V> asOutput() {
+  public Output<V> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatefulStandardNormal.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatefulStandardNormal.java
index 2db4f99804a..3de3c5211e3 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatefulStandardNormal.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatefulStandardNormal.java
@@ -22,13 +22,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TFloat32;
 import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
 
 /**
  * Outputs random values from a normal distribution.
@@ -38,7 +38,7 @@
  * @param <U> data type for {@code output()} output
  */
 @Operator(group = "random")
-public final class StatefulStandardNormal<U extends Tensor> extends RawOp implements Operand<U> {
+public final class StatefulStandardNormal<U extends TType> extends RawOp implements Operand<U> {
   
   /**
    * Factory method to create a class wrapping a new StatefulStandardNormal operation.
@@ -51,11 +51,11 @@ public final class StatefulStandardNormal<U extends Tensor> extends RawOp implem
    * @return a new instance of StatefulStandardNormal
    */
   @Endpoint(describeByClass = true)
-  public static <U extends Tensor, T extends Tensor> StatefulStandardNormal<U> create(Scope scope, Operand<?> resource, Operand<TInt64> algorithm, Operand<T> shape, DataType<U> dtype) {
+  public static <U extends TType, T extends TType> StatefulStandardNormal<U> create(Scope scope, Operand<?> resource, Operand<TInt64> algorithm, Operand<T> shape, DataType<U> dtype) {
     OperationBuilder opBuilder = scope.env().opBuilder("StatefulStandardNormalV2", scope.makeOpName("StatefulStandardNormal"));
-    opBuilder.addInput(resource.asOutput());
-    opBuilder.addInput(algorithm.asOutput());
-    opBuilder.addInput(shape.asOutput());
+    opBuilder.addInput(resource.asOutput(scope));
+    opBuilder.addInput(algorithm.asOutput(scope));
+    opBuilder.addInput(shape.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("dtype", dtype);
     return new StatefulStandardNormal<U>(opBuilder.build());
@@ -71,7 +71,7 @@ public static <U extends Tensor, T extends Tensor> StatefulStandardNormal<U> cre
    * @return a new instance of StatefulStandardNormal
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> StatefulStandardNormal<TFloat32> create(Scope scope, Operand<?> resource, Operand<TInt64> algorithm, Operand<T> shape) {
+  public static <T extends TType> StatefulStandardNormal<TFloat32> create(Scope scope, Operand<?> resource, Operand<TInt64> algorithm, Operand<T> shape) {
     return create(scope, resource, algorithm, shape, TFloat32.DTYPE);
   }
   
@@ -83,7 +83,7 @@ public Output<U> output() {
   }
   
   @Override
-  public Output<U> asOutput() {
+  public Output<U> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatefulTruncatedNormal.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatefulTruncatedNormal.java
index 334d9de6b1a..391462d9bbf 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatefulTruncatedNormal.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatefulTruncatedNormal.java
@@ -22,13 +22,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TFloat32;
 import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
 
 /**
  * Outputs random values from a truncated normal distribution.
@@ -39,7 +39,7 @@
  * 
  * @param <U> data type for {@code output()} output
  */
-public final class StatefulTruncatedNormal<U extends Tensor> extends RawOp implements Operand<U> {
+public final class StatefulTruncatedNormal<U extends TType> extends RawOp implements Operand<U> {
   
   /**
    * Factory method to create a class wrapping a new StatefulTruncatedNormal operation.
@@ -52,11 +52,11 @@ public final class StatefulTruncatedNormal<U extends Tensor> extends RawOp imple
    * @return a new instance of StatefulTruncatedNormal
    */
   @Endpoint(describeByClass = true)
-  public static <U extends Tensor, T extends Tensor> StatefulTruncatedNormal<U> create(Scope scope, Operand<?> resource, Operand<TInt64> algorithm, Operand<T> shape, DataType<U> dtype) {
+  public static <U extends TType, T extends TType> StatefulTruncatedNormal<U> create(Scope scope, Operand<?> resource, Operand<TInt64> algorithm, Operand<T> shape, DataType<U> dtype) {
     OperationBuilder opBuilder = scope.env().opBuilder("StatefulTruncatedNormal", scope.makeOpName("StatefulTruncatedNormal"));
-    opBuilder.addInput(resource.asOutput());
-    opBuilder.addInput(algorithm.asOutput());
-    opBuilder.addInput(shape.asOutput());
+    opBuilder.addInput(resource.asOutput(scope));
+    opBuilder.addInput(algorithm.asOutput(scope));
+    opBuilder.addInput(shape.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("dtype", dtype);
     return new StatefulTruncatedNormal<U>(opBuilder.build());
@@ -72,7 +72,7 @@ public static <U extends Tensor, T extends Tensor> StatefulTruncatedNormal<U> cr
    * @return a new instance of StatefulTruncatedNormal
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> StatefulTruncatedNormal<TFloat32> create(Scope scope, Operand<?> resource, Operand<TInt64> algorithm, Operand<T> shape) {
+  public static <T extends TType> StatefulTruncatedNormal<TFloat32> create(Scope scope, Operand<?> resource, Operand<TInt64> algorithm, Operand<T> shape) {
     return create(scope, resource, algorithm, shape, TFloat32.DTYPE);
   }
   
@@ -84,7 +84,7 @@ public Output<U> output() {
   }
   
   @Override
-  public Output<U> asOutput() {
+  public Output<U> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatefulUniform.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatefulUniform.java
index 6ca09b116a3..f82afca0163 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatefulUniform.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatefulUniform.java
@@ -22,13 +22,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TFloat32;
 import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
 
 /**
  * Outputs random values from a uniform distribution.
@@ -38,7 +38,7 @@
  * 
  * @param <U> data type for {@code output()} output
  */
-public final class StatefulUniform<U extends Tensor> extends RawOp implements Operand<U> {
+public final class StatefulUniform<U extends TType> extends RawOp implements Operand<U> {
   
   /**
    * Factory method to create a class wrapping a new StatefulUniform operation.
@@ -51,11 +51,11 @@ public final class StatefulUniform<U extends Tensor> extends RawOp implements Op
    * @return a new instance of StatefulUniform
    */
   @Endpoint(describeByClass = true)
-  public static <U extends Tensor, T extends Tensor> StatefulUniform<U> create(Scope scope, Operand<?> resource, Operand<TInt64> algorithm, Operand<T> shape, DataType<U> dtype) {
+  public static <U extends TType, T extends TType> StatefulUniform<U> create(Scope scope, Operand<?> resource, Operand<TInt64> algorithm, Operand<T> shape, DataType<U> dtype) {
     OperationBuilder opBuilder = scope.env().opBuilder("StatefulUniform", scope.makeOpName("StatefulUniform"));
-    opBuilder.addInput(resource.asOutput());
-    opBuilder.addInput(algorithm.asOutput());
-    opBuilder.addInput(shape.asOutput());
+    opBuilder.addInput(resource.asOutput(scope));
+    opBuilder.addInput(algorithm.asOutput(scope));
+    opBuilder.addInput(shape.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("dtype", dtype);
     return new StatefulUniform<U>(opBuilder.build());
@@ -71,7 +71,7 @@ public static <U extends Tensor, T extends Tensor> StatefulUniform<U> create(Sco
    * @return a new instance of StatefulUniform
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> StatefulUniform<TFloat32> create(Scope scope, Operand<?> resource, Operand<TInt64> algorithm, Operand<T> shape) {
+  public static <T extends TType> StatefulUniform<TFloat32> create(Scope scope, Operand<?> resource, Operand<TInt64> algorithm, Operand<T> shape) {
     return create(scope, resource, algorithm, shape, TFloat32.DTYPE);
   }
   
@@ -83,7 +83,7 @@ public Output<U> output() {
   }
   
   @Override
-  public Output<U> asOutput() {
+  public Output<U> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatefulUniformFullInt.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatefulUniformFullInt.java
index 3e3c4b5f0c5..9e59c5f6e15 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatefulUniformFullInt.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatefulUniformFullInt.java
@@ -22,12 +22,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
 
 /**
  * Outputs random integers from a uniform distribution.
@@ -36,7 +36,7 @@
  * 
  * @param <U> data type for {@code output()} output
  */
-public final class StatefulUniformFullInt<U extends Tensor> extends RawOp implements Operand<U> {
+public final class StatefulUniformFullInt<U extends TType> extends RawOp implements Operand<U> {
   
   /**
    * Factory method to create a class wrapping a new StatefulUniformFullInt operation.
@@ -49,11 +49,11 @@ public final class StatefulUniformFullInt<U extends Tensor> extends RawOp implem
    * @return a new instance of StatefulUniformFullInt
    */
   @Endpoint(describeByClass = true)
-  public static <U extends Tensor, T extends Tensor> StatefulUniformFullInt<U> create(Scope scope, Operand<?> resource, Operand<TInt64> algorithm, Operand<T> shape, DataType<U> dtype) {
+  public static <U extends TType, T extends TType> StatefulUniformFullInt<U> create(Scope scope, Operand<?> resource, Operand<TInt64> algorithm, Operand<T> shape, DataType<U> dtype) {
     OperationBuilder opBuilder = scope.env().opBuilder("StatefulUniformFullInt", scope.makeOpName("StatefulUniformFullInt"));
-    opBuilder.addInput(resource.asOutput());
-    opBuilder.addInput(algorithm.asOutput());
-    opBuilder.addInput(shape.asOutput());
+    opBuilder.addInput(resource.asOutput(scope));
+    opBuilder.addInput(algorithm.asOutput(scope));
+    opBuilder.addInput(shape.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("dtype", dtype);
     return new StatefulUniformFullInt<U>(opBuilder.build());
@@ -67,7 +67,7 @@ public Output<U> output() {
   }
   
   @Override
-  public Output<U> asOutput() {
+  public Output<U> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatefulUniformInt.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatefulUniformInt.java
index 56fa0646b35..14edb24fd00 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatefulUniformInt.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatefulUniformInt.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
 
 /**
  * Outputs random integers from a uniform distribution.
@@ -41,7 +41,7 @@
  * 
  * @param <U> data type for {@code output()} output
  */
-public final class StatefulUniformInt<U extends Tensor> extends RawOp implements Operand<U> {
+public final class StatefulUniformInt<U extends TType> extends RawOp implements Operand<U> {
   
   /**
    * Factory method to create a class wrapping a new StatefulUniformInt operation.
@@ -55,13 +55,13 @@ public final class StatefulUniformInt<U extends Tensor> extends RawOp implements
    * @return a new instance of StatefulUniformInt
    */
   @Endpoint(describeByClass = true)
-  public static <U extends Tensor, T extends Tensor> StatefulUniformInt<U> create(Scope scope, Operand<?> resource, Operand<TInt64> algorithm, Operand<T> shape, Operand<U> minval, Operand<U> maxval) {
+  public static <U extends TType, T extends TType> StatefulUniformInt<U> create(Scope scope, Operand<?> resource, Operand<TInt64> algorithm, Operand<T> shape, Operand<U> minval, Operand<U> maxval) {
     OperationBuilder opBuilder = scope.env().opBuilder("StatefulUniformInt", scope.makeOpName("StatefulUniformInt"));
-    opBuilder.addInput(resource.asOutput());
-    opBuilder.addInput(algorithm.asOutput());
-    opBuilder.addInput(shape.asOutput());
-    opBuilder.addInput(minval.asOutput());
-    opBuilder.addInput(maxval.asOutput());
+    opBuilder.addInput(resource.asOutput(scope));
+    opBuilder.addInput(algorithm.asOutput(scope));
+    opBuilder.addInput(shape.asOutput(scope));
+    opBuilder.addInput(minval.asOutput(scope));
+    opBuilder.addInput(maxval.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new StatefulUniformInt<U>(opBuilder.build());
   }
@@ -74,7 +74,7 @@ public Output<U> output() {
   }
   
   @Override
-  public Output<U> asOutput() {
+  public Output<U> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessMultinomial.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessMultinomial.java
index b9a23f53cfd..c9ee3523d5a 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessMultinomial.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessMultinomial.java
@@ -22,7 +22,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -37,7 +36,7 @@
  * @param <V> data type for {@code output()} output
  */
 @Operator(group = "random")
-public final class StatelessMultinomial<V extends Tensor & TNumber> extends RawOp implements Operand<V> {
+public final class StatelessMultinomial<V extends TNumber> extends RawOp implements Operand<V> {
   
   /**
    * Factory method to create a class wrapping a new StatelessMultinomial operation.
@@ -51,11 +50,11 @@ public final class StatelessMultinomial<V extends Tensor & TNumber> extends RawO
    * @return a new instance of StatelessMultinomial
    */
   @Endpoint(describeByClass = true)
-  public static <V extends Tensor & TNumber, T extends Tensor & TNumber, U extends Tensor & TNumber> StatelessMultinomial<V> create(Scope scope, Operand<T> logits, Operand<TInt32> numSamples, Operand<U> seed, DataType<V> outputDtype) {
+  public static <V extends TNumber, T extends TNumber, U extends TNumber> StatelessMultinomial<V> create(Scope scope, Operand<T> logits, Operand<TInt32> numSamples, Operand<U> seed, DataType<V> outputDtype) {
     OperationBuilder opBuilder = scope.env().opBuilder("StatelessMultinomial", scope.makeOpName("StatelessMultinomial"));
-    opBuilder.addInput(logits.asOutput());
-    opBuilder.addInput(numSamples.asOutput());
-    opBuilder.addInput(seed.asOutput());
+    opBuilder.addInput(logits.asOutput(scope));
+    opBuilder.addInput(numSamples.asOutput(scope));
+    opBuilder.addInput(seed.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("output_dtype", outputDtype);
     return new StatelessMultinomial<V>(opBuilder.build());
@@ -72,7 +71,7 @@ public static <V extends Tensor & TNumber, T extends Tensor & TNumber, U extends
    * @return a new instance of StatelessMultinomial
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber, U extends Tensor & TNumber> StatelessMultinomial<TInt64> create(Scope scope, Operand<T> logits, Operand<TInt32> numSamples, Operand<U> seed) {
+  public static <T extends TNumber, U extends TNumber> StatelessMultinomial<TInt64> create(Scope scope, Operand<T> logits, Operand<TInt32> numSamples, Operand<U> seed) {
     return create(scope, logits, numSamples, seed, TInt64.DTYPE);
   }
   
@@ -85,7 +84,7 @@ public Output<V> output() {
   }
   
   @Override
-  public Output<V> asOutput() {
+  public Output<V> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessRandomBinomial.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessRandomBinomial.java
index cfbf62891ba..d96613dee92 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessRandomBinomial.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessRandomBinomial.java
@@ -22,7 +22,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -39,7 +38,7 @@
  * 
  * @param <W> data type for {@code output()} output
  */
-public final class StatelessRandomBinomial<W extends Tensor & TNumber> extends RawOp implements Operand<W> {
+public final class StatelessRandomBinomial<W extends TNumber> extends RawOp implements Operand<W> {
   
   /**
    * Factory method to create a class wrapping a new StatelessRandomBinomial operation.
@@ -55,12 +54,12 @@ public final class StatelessRandomBinomial<W extends Tensor & TNumber> extends R
    * @return a new instance of StatelessRandomBinomial
    */
   @Endpoint(describeByClass = true)
-  public static <W extends Tensor & TNumber, T extends Tensor & TNumber, U extends Tensor & TNumber, V extends Tensor & TNumber> StatelessRandomBinomial<W> create(Scope scope, Operand<T> shape, Operand<U> seed, Operand<V> counts, Operand<V> probs, DataType<W> dtype) {
+  public static <W extends TNumber, T extends TNumber, U extends TNumber, V extends TNumber> StatelessRandomBinomial<W> create(Scope scope, Operand<T> shape, Operand<U> seed, Operand<V> counts, Operand<V> probs, DataType<W> dtype) {
     OperationBuilder opBuilder = scope.env().opBuilder("StatelessRandomBinomial", scope.makeOpName("StatelessRandomBinomial"));
-    opBuilder.addInput(shape.asOutput());
-    opBuilder.addInput(seed.asOutput());
-    opBuilder.addInput(counts.asOutput());
-    opBuilder.addInput(probs.asOutput());
+    opBuilder.addInput(shape.asOutput(scope));
+    opBuilder.addInput(seed.asOutput(scope));
+    opBuilder.addInput(counts.asOutput(scope));
+    opBuilder.addInput(probs.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("dtype", dtype);
     return new StatelessRandomBinomial<W>(opBuilder.build());
@@ -79,7 +78,7 @@ public static <W extends Tensor & TNumber, T extends Tensor & TNumber, U extends
    * @return a new instance of StatelessRandomBinomial
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber, U extends Tensor & TNumber, V extends Tensor & TNumber> StatelessRandomBinomial<TInt64> create(Scope scope, Operand<T> shape, Operand<U> seed, Operand<V> counts, Operand<V> probs) {
+  public static <T extends TNumber, U extends TNumber, V extends TNumber> StatelessRandomBinomial<TInt64> create(Scope scope, Operand<T> shape, Operand<U> seed, Operand<V> counts, Operand<V> probs) {
     return create(scope, shape, seed, counts, probs, TInt64.DTYPE);
   }
   
@@ -91,7 +90,7 @@ public Output<W> output() {
   }
   
   @Override
-  public Output<W> asOutput() {
+  public Output<W> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessRandomGamma.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessRandomGamma.java
index 0aabc88dd01..30800534761 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessRandomGamma.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessRandomGamma.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -37,7 +36,7 @@
  * 
  * @param <V> data type for {@code output()} output
  */
-public final class StatelessRandomGamma<V extends Tensor & TNumber> extends RawOp implements Operand<V> {
+public final class StatelessRandomGamma<V extends TNumber> extends RawOp implements Operand<V> {
   
   /**
    * Factory method to create a class wrapping a new StatelessRandomGamma operation.
@@ -50,11 +49,11 @@ public final class StatelessRandomGamma<V extends Tensor & TNumber> extends RawO
    * @return a new instance of StatelessRandomGamma
    */
   @Endpoint(describeByClass = true)
-  public static <V extends Tensor & TNumber, T extends Tensor & TNumber, U extends Tensor & TNumber> StatelessRandomGamma<V> create(Scope scope, Operand<T> shape, Operand<U> seed, Operand<V> alpha) {
+  public static <V extends TNumber, T extends TNumber, U extends TNumber> StatelessRandomGamma<V> create(Scope scope, Operand<T> shape, Operand<U> seed, Operand<V> alpha) {
     OperationBuilder opBuilder = scope.env().opBuilder("StatelessRandomGammaV2", scope.makeOpName("StatelessRandomGamma"));
-    opBuilder.addInput(shape.asOutput());
-    opBuilder.addInput(seed.asOutput());
-    opBuilder.addInput(alpha.asOutput());
+    opBuilder.addInput(shape.asOutput(scope));
+    opBuilder.addInput(seed.asOutput(scope));
+    opBuilder.addInput(alpha.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new StatelessRandomGamma<V>(opBuilder.build());
   }
@@ -67,7 +66,7 @@ public Output<V> output() {
   }
   
   @Override
-  public Output<V> asOutput() {
+  public Output<V> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessRandomNormal.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessRandomNormal.java
index ef2f6a1b70e..76c3aa9406a 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessRandomNormal.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessRandomNormal.java
@@ -22,7 +22,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -40,7 +39,7 @@
  * @param <V> data type for {@code output()} output
  */
 @Operator(group = "random")
-public final class StatelessRandomNormal<V extends Tensor & TNumber> extends RawOp implements Operand<V> {
+public final class StatelessRandomNormal<V extends TNumber> extends RawOp implements Operand<V> {
   
   /**
    * Factory method to create a class wrapping a new StatelessRandomNormal operation.
@@ -52,10 +51,10 @@ public final class StatelessRandomNormal<V extends Tensor & TNumber> extends Raw
    * @return a new instance of StatelessRandomNormal
    */
   @Endpoint(describeByClass = true)
-  public static <V extends Tensor & TNumber, T extends Tensor & TNumber, U extends Tensor & TNumber> StatelessRandomNormal<V> create(Scope scope, Operand<T> shape, Operand<U> seed, DataType<V> dtype) {
+  public static <V extends TNumber, T extends TNumber, U extends TNumber> StatelessRandomNormal<V> create(Scope scope, Operand<T> shape, Operand<U> seed, DataType<V> dtype) {
     OperationBuilder opBuilder = scope.env().opBuilder("StatelessRandomNormal", scope.makeOpName("StatelessRandomNormal"));
-    opBuilder.addInput(shape.asOutput());
-    opBuilder.addInput(seed.asOutput());
+    opBuilder.addInput(shape.asOutput(scope));
+    opBuilder.addInput(seed.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("dtype", dtype);
     return new StatelessRandomNormal<V>(opBuilder.build());
@@ -70,7 +69,7 @@ public static <V extends Tensor & TNumber, T extends Tensor & TNumber, U extends
    * @return a new instance of StatelessRandomNormal
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber, U extends Tensor & TNumber> StatelessRandomNormal<TFloat32> create(Scope scope, Operand<T> shape, Operand<U> seed) {
+  public static <T extends TNumber, U extends TNumber> StatelessRandomNormal<TFloat32> create(Scope scope, Operand<T> shape, Operand<U> seed) {
     return create(scope, shape, seed, TFloat32.DTYPE);
   }
   
@@ -82,7 +81,7 @@ public Output<V> output() {
   }
   
   @Override
-  public Output<V> asOutput() {
+  public Output<V> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessRandomPoisson.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessRandomPoisson.java
index d4249b20c59..10dc4f9bfab 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessRandomPoisson.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessRandomPoisson.java
@@ -22,7 +22,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -38,7 +37,7 @@
  * 
  * @param <W> data type for {@code output()} output
  */
-public final class StatelessRandomPoisson<W extends Tensor & TNumber> extends RawOp implements Operand<W> {
+public final class StatelessRandomPoisson<W extends TNumber> extends RawOp implements Operand<W> {
   
   /**
    * Factory method to create a class wrapping a new StatelessRandomPoisson operation.
@@ -52,11 +51,11 @@ public final class StatelessRandomPoisson<W extends Tensor & TNumber> extends Ra
    * @return a new instance of StatelessRandomPoisson
    */
   @Endpoint(describeByClass = true)
-  public static <W extends Tensor & TNumber, T extends Tensor & TNumber, U extends Tensor & TNumber, V extends Tensor & TNumber> StatelessRandomPoisson<W> create(Scope scope, Operand<T> shape, Operand<U> seed, Operand<V> lam, DataType<W> dtype) {
+  public static <W extends TNumber, T extends TNumber, U extends TNumber, V extends TNumber> StatelessRandomPoisson<W> create(Scope scope, Operand<T> shape, Operand<U> seed, Operand<V> lam, DataType<W> dtype) {
     OperationBuilder opBuilder = scope.env().opBuilder("StatelessRandomPoisson", scope.makeOpName("StatelessRandomPoisson"));
-    opBuilder.addInput(shape.asOutput());
-    opBuilder.addInput(seed.asOutput());
-    opBuilder.addInput(lam.asOutput());
+    opBuilder.addInput(shape.asOutput(scope));
+    opBuilder.addInput(seed.asOutput(scope));
+    opBuilder.addInput(lam.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("dtype", dtype);
     return new StatelessRandomPoisson<W>(opBuilder.build());
@@ -70,7 +69,7 @@ public Output<W> output() {
   }
   
   @Override
-  public Output<W> asOutput() {
+  public Output<W> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessRandomUniform.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessRandomUniform.java
index dafea214cda..d6718eeeca4 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessRandomUniform.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessRandomUniform.java
@@ -22,7 +22,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -41,7 +40,7 @@
  * @param <V> data type for {@code output()} output
  */
 @Operator(group = "random")
-public final class StatelessRandomUniform<V extends Tensor & TNumber> extends RawOp implements Operand<V> {
+public final class StatelessRandomUniform<V extends TNumber> extends RawOp implements Operand<V> {
   
   /**
    * Factory method to create a class wrapping a new StatelessRandomUniform operation.
@@ -53,10 +52,10 @@ public final class StatelessRandomUniform<V extends Tensor & TNumber> extends Ra
    * @return a new instance of StatelessRandomUniform
    */
   @Endpoint(describeByClass = true)
-  public static <V extends Tensor & TNumber, T extends Tensor & TNumber, U extends Tensor & TNumber> StatelessRandomUniform<V> create(Scope scope, Operand<T> shape, Operand<U> seed, DataType<V> dtype) {
+  public static <V extends TNumber, T extends TNumber, U extends TNumber> StatelessRandomUniform<V> create(Scope scope, Operand<T> shape, Operand<U> seed, DataType<V> dtype) {
     OperationBuilder opBuilder = scope.env().opBuilder("StatelessRandomUniform", scope.makeOpName("StatelessRandomUniform"));
-    opBuilder.addInput(shape.asOutput());
-    opBuilder.addInput(seed.asOutput());
+    opBuilder.addInput(shape.asOutput(scope));
+    opBuilder.addInput(seed.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("dtype", dtype);
     return new StatelessRandomUniform<V>(opBuilder.build());
@@ -71,7 +70,7 @@ public static <V extends Tensor & TNumber, T extends Tensor & TNumber, U extends
    * @return a new instance of StatelessRandomUniform
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber, U extends Tensor & TNumber> StatelessRandomUniform<TFloat32> create(Scope scope, Operand<T> shape, Operand<U> seed) {
+  public static <T extends TNumber, U extends TNumber> StatelessRandomUniform<TFloat32> create(Scope scope, Operand<T> shape, Operand<U> seed) {
     return create(scope, shape, seed, TFloat32.DTYPE);
   }
   
@@ -83,7 +82,7 @@ public Output<V> output() {
   }
   
   @Override
-  public Output<V> asOutput() {
+  public Output<V> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessRandomUniformFullInt.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessRandomUniformFullInt.java
index eff49a245ff..a1bcc25573f 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessRandomUniformFullInt.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessRandomUniformFullInt.java
@@ -22,7 +22,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -38,7 +37,7 @@
  * 
  * @param <V> data type for {@code output()} output
  */
-public final class StatelessRandomUniformFullInt<V extends Tensor & TNumber> extends RawOp implements Operand<V> {
+public final class StatelessRandomUniformFullInt<V extends TNumber> extends RawOp implements Operand<V> {
   
   /**
    * Factory method to create a class wrapping a new StatelessRandomUniformFullInt operation.
@@ -50,10 +49,10 @@ public final class StatelessRandomUniformFullInt<V extends Tensor & TNumber> ext
    * @return a new instance of StatelessRandomUniformFullInt
    */
   @Endpoint(describeByClass = true)
-  public static <V extends Tensor & TNumber, T extends Tensor & TNumber, U extends Tensor & TNumber> StatelessRandomUniformFullInt<V> create(Scope scope, Operand<T> shape, Operand<U> seed, DataType<V> dtype) {
+  public static <V extends TNumber, T extends TNumber, U extends TNumber> StatelessRandomUniformFullInt<V> create(Scope scope, Operand<T> shape, Operand<U> seed, DataType<V> dtype) {
     OperationBuilder opBuilder = scope.env().opBuilder("StatelessRandomUniformFullInt", scope.makeOpName("StatelessRandomUniformFullInt"));
-    opBuilder.addInput(shape.asOutput());
-    opBuilder.addInput(seed.asOutput());
+    opBuilder.addInput(shape.asOutput(scope));
+    opBuilder.addInput(seed.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("dtype", dtype);
     return new StatelessRandomUniformFullInt<V>(opBuilder.build());
@@ -67,7 +66,7 @@ public Output<V> output() {
   }
   
   @Override
-  public Output<V> asOutput() {
+  public Output<V> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessRandomUniformInt.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessRandomUniformInt.java
index f916718ebe9..29b5d589a76 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessRandomUniformInt.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessRandomUniformInt.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -37,7 +36,7 @@
  * 
  * @param <V> data type for {@code output()} output
  */
-public final class StatelessRandomUniformInt<V extends Tensor & TNumber> extends RawOp implements Operand<V> {
+public final class StatelessRandomUniformInt<V extends TNumber> extends RawOp implements Operand<V> {
   
   /**
    * Factory method to create a class wrapping a new StatelessRandomUniformInt operation.
@@ -50,12 +49,12 @@ public final class StatelessRandomUniformInt<V extends Tensor & TNumber> extends
    * @return a new instance of StatelessRandomUniformInt
    */
   @Endpoint(describeByClass = true)
-  public static <V extends Tensor & TNumber, T extends Tensor & TNumber, U extends Tensor & TNumber> StatelessRandomUniformInt<V> create(Scope scope, Operand<T> shape, Operand<U> seed, Operand<V> minval, Operand<V> maxval) {
+  public static <V extends TNumber, T extends TNumber, U extends TNumber> StatelessRandomUniformInt<V> create(Scope scope, Operand<T> shape, Operand<U> seed, Operand<V> minval, Operand<V> maxval) {
     OperationBuilder opBuilder = scope.env().opBuilder("StatelessRandomUniformInt", scope.makeOpName("StatelessRandomUniformInt"));
-    opBuilder.addInput(shape.asOutput());
-    opBuilder.addInput(seed.asOutput());
-    opBuilder.addInput(minval.asOutput());
-    opBuilder.addInput(maxval.asOutput());
+    opBuilder.addInput(shape.asOutput(scope));
+    opBuilder.addInput(seed.asOutput(scope));
+    opBuilder.addInput(minval.asOutput(scope));
+    opBuilder.addInput(maxval.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new StatelessRandomUniformInt<V>(opBuilder.build());
   }
@@ -68,7 +67,7 @@ public Output<V> output() {
   }
   
   @Override
-  public Output<V> asOutput() {
+  public Output<V> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessTruncatedNormal.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessTruncatedNormal.java
index d9e8bb3ce05..ca1ee733f62 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessTruncatedNormal.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessTruncatedNormal.java
@@ -22,7 +22,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -42,7 +41,7 @@
  * @param <V> data type for {@code output()} output
  */
 @Operator(group = "random")
-public final class StatelessTruncatedNormal<V extends Tensor & TNumber> extends RawOp implements Operand<V> {
+public final class StatelessTruncatedNormal<V extends TNumber> extends RawOp implements Operand<V> {
   
   /**
    * Factory method to create a class wrapping a new StatelessTruncatedNormal operation.
@@ -54,10 +53,10 @@ public final class StatelessTruncatedNormal<V extends Tensor & TNumber> extends
    * @return a new instance of StatelessTruncatedNormal
    */
   @Endpoint(describeByClass = true)
-  public static <V extends Tensor & TNumber, T extends Tensor & TNumber, U extends Tensor & TNumber> StatelessTruncatedNormal<V> create(Scope scope, Operand<T> shape, Operand<U> seed, DataType<V> dtype) {
+  public static <V extends TNumber, T extends TNumber, U extends TNumber> StatelessTruncatedNormal<V> create(Scope scope, Operand<T> shape, Operand<U> seed, DataType<V> dtype) {
     OperationBuilder opBuilder = scope.env().opBuilder("StatelessTruncatedNormal", scope.makeOpName("StatelessTruncatedNormal"));
-    opBuilder.addInput(shape.asOutput());
-    opBuilder.addInput(seed.asOutput());
+    opBuilder.addInput(shape.asOutput(scope));
+    opBuilder.addInput(seed.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("dtype", dtype);
     return new StatelessTruncatedNormal<V>(opBuilder.build());
@@ -72,7 +71,7 @@ public static <V extends Tensor & TNumber, T extends Tensor & TNumber, U extends
    * @return a new instance of StatelessTruncatedNormal
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber, U extends Tensor & TNumber> StatelessTruncatedNormal<TFloat32> create(Scope scope, Operand<T> shape, Operand<U> seed) {
+  public static <T extends TNumber, U extends TNumber> StatelessTruncatedNormal<TFloat32> create(Scope scope, Operand<T> shape, Operand<U> seed) {
     return create(scope, shape, seed, TFloat32.DTYPE);
   }
   
@@ -84,7 +83,7 @@ public Output<V> output() {
   }
   
   @Override
-  public Output<V> asOutput() {
+  public Output<V> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/TruncatedNormal.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/TruncatedNormal.java
index 1004c261237..e9588a4ab14 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/TruncatedNormal.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/TruncatedNormal.java
@@ -22,7 +22,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -39,7 +38,7 @@
  * @param <U> data type for {@code output()} output
  */
 @Operator(group = "random")
-public final class TruncatedNormal<U extends Tensor & TNumber> extends RawOp implements Operand<U> {
+public final class TruncatedNormal<U extends TNumber> extends RawOp implements Operand<U> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.random.TruncatedNormal}
@@ -81,9 +80,9 @@ private Options() {
    * @return a new instance of TruncatedNormal
    */
   @Endpoint(describeByClass = true)
-  public static <U extends Tensor & TNumber, T extends Tensor & TNumber> TruncatedNormal<U> create(Scope scope, Operand<T> shape, DataType<U> dtype, Options... options) {
+  public static <U extends TNumber, T extends TNumber> TruncatedNormal<U> create(Scope scope, Operand<T> shape, DataType<U> dtype, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("TruncatedNormal", scope.makeOpName("TruncatedNormal"));
-    opBuilder.addInput(shape.asOutput());
+    opBuilder.addInput(shape.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("dtype", dtype);
     if (options != null) {
@@ -124,7 +123,7 @@ public Output<U> output() {
   }
   
   @Override
-  public Output<U> asOutput() {
+  public Output<U> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/UniformCandidateSampler.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/UniformCandidateSampler.java
index 9e1f2b1abbd..9a0b724f8ff 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/UniformCandidateSampler.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/UniformCandidateSampler.java
@@ -92,7 +92,7 @@ private Options() {
   @Endpoint(describeByClass = true)
   public static UniformCandidateSampler create(Scope scope, Operand<TInt64> trueClasses, Long numTrue, Long numSampled, Boolean unique, Long rangeMax, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("UniformCandidateSampler", scope.makeOpName("UniformCandidateSampler"));
-    opBuilder.addInput(trueClasses.asOutput());
+    opBuilder.addInput(trueClasses.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("num_true", numTrue);
     opBuilder.setAttr("num_sampled", numSampled);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/BatchFft.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/BatchFft.java
index 80b665f059d..2e9d7f0da88 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/BatchFft.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/BatchFft.java
@@ -21,16 +21,16 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  */
 @Operator(group = "signal")
-public final class BatchFft extends RawOp implements Operand<Tensor> {
+public final class BatchFft extends RawOp implements Operand<TType> {
   
   /**
    * Factory method to create a class wrapping a new BatchFft operation.
@@ -42,7 +42,7 @@ public final class BatchFft extends RawOp implements Operand<Tensor> {
   @Endpoint(describeByClass = true)
   public static BatchFft create(Scope scope, Operand<?> input) {
     OperationBuilder opBuilder = scope.env().opBuilder("BatchFFT", scope.makeOpName("BatchFft"));
-    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new BatchFft(opBuilder.build());
   }
@@ -55,8 +55,8 @@ public Output<?> output() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<Tensor> asOutput() {
-    return (Output<Tensor>) output;
+  public Output<TType> asOutput(Scope scope) {
+    return (Output<TType>) output;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/BatchFft2d.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/BatchFft2d.java
index e8e48f76cd9..aa8d61a34ef 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/BatchFft2d.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/BatchFft2d.java
@@ -21,16 +21,16 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  */
 @Operator(group = "signal")
-public final class BatchFft2d extends RawOp implements Operand<Tensor> {
+public final class BatchFft2d extends RawOp implements Operand<TType> {
   
   /**
    * Factory method to create a class wrapping a new BatchFft2d operation.
@@ -42,7 +42,7 @@ public final class BatchFft2d extends RawOp implements Operand<Tensor> {
   @Endpoint(describeByClass = true)
   public static BatchFft2d create(Scope scope, Operand<?> input) {
     OperationBuilder opBuilder = scope.env().opBuilder("BatchFFT2D", scope.makeOpName("BatchFft2d"));
-    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new BatchFft2d(opBuilder.build());
   }
@@ -55,8 +55,8 @@ public Output<?> output() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<Tensor> asOutput() {
-    return (Output<Tensor>) output;
+  public Output<TType> asOutput(Scope scope) {
+    return (Output<TType>) output;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/BatchFft3d.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/BatchFft3d.java
index 538689e7ea7..57d61e6a060 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/BatchFft3d.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/BatchFft3d.java
@@ -21,16 +21,16 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  */
 @Operator(group = "signal")
-public final class BatchFft3d extends RawOp implements Operand<Tensor> {
+public final class BatchFft3d extends RawOp implements Operand<TType> {
   
   /**
    * Factory method to create a class wrapping a new BatchFft3d operation.
@@ -42,7 +42,7 @@ public final class BatchFft3d extends RawOp implements Operand<Tensor> {
   @Endpoint(describeByClass = true)
   public static BatchFft3d create(Scope scope, Operand<?> input) {
     OperationBuilder opBuilder = scope.env().opBuilder("BatchFFT3D", scope.makeOpName("BatchFft3d"));
-    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new BatchFft3d(opBuilder.build());
   }
@@ -55,8 +55,8 @@ public Output<?> output() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<Tensor> asOutput() {
-    return (Output<Tensor>) output;
+  public Output<TType> asOutput(Scope scope) {
+    return (Output<TType>) output;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/BatchIfft.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/BatchIfft.java
index 0f7529241fd..2033092ed75 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/BatchIfft.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/BatchIfft.java
@@ -21,16 +21,16 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  */
 @Operator(group = "signal")
-public final class BatchIfft extends RawOp implements Operand<Tensor> {
+public final class BatchIfft extends RawOp implements Operand<TType> {
   
   /**
    * Factory method to create a class wrapping a new BatchIfft operation.
@@ -42,7 +42,7 @@ public final class BatchIfft extends RawOp implements Operand<Tensor> {
   @Endpoint(describeByClass = true)
   public static BatchIfft create(Scope scope, Operand<?> input) {
     OperationBuilder opBuilder = scope.env().opBuilder("BatchIFFT", scope.makeOpName("BatchIfft"));
-    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new BatchIfft(opBuilder.build());
   }
@@ -55,8 +55,8 @@ public Output<?> output() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<Tensor> asOutput() {
-    return (Output<Tensor>) output;
+  public Output<TType> asOutput(Scope scope) {
+    return (Output<TType>) output;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/BatchIfft2d.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/BatchIfft2d.java
index 127dae4d1cd..fa6dcf4ba52 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/BatchIfft2d.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/BatchIfft2d.java
@@ -21,16 +21,16 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  */
 @Operator(group = "signal")
-public final class BatchIfft2d extends RawOp implements Operand<Tensor> {
+public final class BatchIfft2d extends RawOp implements Operand<TType> {
   
   /**
    * Factory method to create a class wrapping a new BatchIfft2d operation.
@@ -42,7 +42,7 @@ public final class BatchIfft2d extends RawOp implements Operand<Tensor> {
   @Endpoint(describeByClass = true)
   public static BatchIfft2d create(Scope scope, Operand<?> input) {
     OperationBuilder opBuilder = scope.env().opBuilder("BatchIFFT2D", scope.makeOpName("BatchIfft2d"));
-    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new BatchIfft2d(opBuilder.build());
   }
@@ -55,8 +55,8 @@ public Output<?> output() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<Tensor> asOutput() {
-    return (Output<Tensor>) output;
+  public Output<TType> asOutput(Scope scope) {
+    return (Output<TType>) output;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/BatchIfft3d.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/BatchIfft3d.java
index 3876316fd85..1ed5c57bb1b 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/BatchIfft3d.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/BatchIfft3d.java
@@ -21,16 +21,16 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  */
 @Operator(group = "signal")
-public final class BatchIfft3d extends RawOp implements Operand<Tensor> {
+public final class BatchIfft3d extends RawOp implements Operand<TType> {
   
   /**
    * Factory method to create a class wrapping a new BatchIfft3d operation.
@@ -42,7 +42,7 @@ public final class BatchIfft3d extends RawOp implements Operand<Tensor> {
   @Endpoint(describeByClass = true)
   public static BatchIfft3d create(Scope scope, Operand<?> input) {
     OperationBuilder opBuilder = scope.env().opBuilder("BatchIFFT3D", scope.makeOpName("BatchIfft3d"));
-    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new BatchIfft3d(opBuilder.build());
   }
@@ -55,8 +55,8 @@ public Output<?> output() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<Tensor> asOutput() {
-    return (Output<Tensor>) output;
+  public Output<TType> asOutput(Scope scope) {
+    return (Output<TType>) output;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Fft.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Fft.java
index 6599e4b2c5e..2c9fb8d16b9 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Fft.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Fft.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Fast Fourier transform.
@@ -36,7 +36,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "signal")
-public final class Fft<T extends Tensor> extends RawOp implements Operand<T> {
+public final class Fft<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Fft operation.
@@ -46,9 +46,9 @@ public final class Fft<T extends Tensor> extends RawOp implements Operand<T> {
    * @return a new instance of Fft
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> Fft<T> create(Scope scope, Operand<T> input) {
+  public static <T extends TType> Fft<T> create(Scope scope, Operand<T> input) {
     OperationBuilder opBuilder = scope.env().opBuilder("FFT", scope.makeOpName("Fft"));
-    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new Fft<T>(opBuilder.build());
   }
@@ -66,7 +66,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Fft2d.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Fft2d.java
index d605fd8c944..226d5b313d6 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Fft2d.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Fft2d.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * 2D fast Fourier transform.
@@ -36,7 +36,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "signal")
-public final class Fft2d<T extends Tensor> extends RawOp implements Operand<T> {
+public final class Fft2d<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Fft2d operation.
@@ -46,9 +46,9 @@ public final class Fft2d<T extends Tensor> extends RawOp implements Operand<T> {
    * @return a new instance of Fft2d
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> Fft2d<T> create(Scope scope, Operand<T> input) {
+  public static <T extends TType> Fft2d<T> create(Scope scope, Operand<T> input) {
     OperationBuilder opBuilder = scope.env().opBuilder("FFT2D", scope.makeOpName("Fft2d"));
-    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new Fft2d<T>(opBuilder.build());
   }
@@ -66,7 +66,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Fft3d.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Fft3d.java
index 819d584bcd9..4dc5e420356 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Fft3d.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Fft3d.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * 3D fast Fourier transform.
@@ -36,7 +36,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "signal")
-public final class Fft3d<T extends Tensor> extends RawOp implements Operand<T> {
+public final class Fft3d<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Fft3d operation.
@@ -46,9 +46,9 @@ public final class Fft3d<T extends Tensor> extends RawOp implements Operand<T> {
    * @return a new instance of Fft3d
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> Fft3d<T> create(Scope scope, Operand<T> input) {
+  public static <T extends TType> Fft3d<T> create(Scope scope, Operand<T> input) {
     OperationBuilder opBuilder = scope.env().opBuilder("FFT3D", scope.makeOpName("Fft3d"));
-    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new Fft3d<T>(opBuilder.build());
   }
@@ -66,7 +66,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Ifft.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Ifft.java
index 62af7dc0af4..fb65e5a5745 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Ifft.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Ifft.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Inverse fast Fourier transform.
@@ -36,7 +36,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "signal")
-public final class Ifft<T extends Tensor> extends RawOp implements Operand<T> {
+public final class Ifft<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Ifft operation.
@@ -46,9 +46,9 @@ public final class Ifft<T extends Tensor> extends RawOp implements Operand<T> {
    * @return a new instance of Ifft
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> Ifft<T> create(Scope scope, Operand<T> input) {
+  public static <T extends TType> Ifft<T> create(Scope scope, Operand<T> input) {
     OperationBuilder opBuilder = scope.env().opBuilder("IFFT", scope.makeOpName("Ifft"));
-    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new Ifft<T>(opBuilder.build());
   }
@@ -66,7 +66,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Ifft2d.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Ifft2d.java
index 03394267780..6d4bf5c8504 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Ifft2d.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Ifft2d.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Inverse 2D fast Fourier transform.
@@ -36,7 +36,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "signal")
-public final class Ifft2d<T extends Tensor> extends RawOp implements Operand<T> {
+public final class Ifft2d<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Ifft2d operation.
@@ -46,9 +46,9 @@ public final class Ifft2d<T extends Tensor> extends RawOp implements Operand<T>
    * @return a new instance of Ifft2d
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> Ifft2d<T> create(Scope scope, Operand<T> input) {
+  public static <T extends TType> Ifft2d<T> create(Scope scope, Operand<T> input) {
     OperationBuilder opBuilder = scope.env().opBuilder("IFFT2D", scope.makeOpName("Ifft2d"));
-    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new Ifft2d<T>(opBuilder.build());
   }
@@ -66,7 +66,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Ifft3d.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Ifft3d.java
index 1c2eb56e45e..1ee196c2a73 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Ifft3d.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Ifft3d.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Inverse 3D fast Fourier transform.
@@ -36,7 +36,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "signal")
-public final class Ifft3d<T extends Tensor> extends RawOp implements Operand<T> {
+public final class Ifft3d<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Ifft3d operation.
@@ -46,9 +46,9 @@ public final class Ifft3d<T extends Tensor> extends RawOp implements Operand<T>
    * @return a new instance of Ifft3d
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> Ifft3d<T> create(Scope scope, Operand<T> input) {
+  public static <T extends TType> Ifft3d<T> create(Scope scope, Operand<T> input) {
     OperationBuilder opBuilder = scope.env().opBuilder("IFFT3D", scope.makeOpName("Ifft3d"));
-    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new Ifft3d<T>(opBuilder.build());
   }
@@ -66,7 +66,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Irfft.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Irfft.java
index fb48c602bf7..3091bcdcff9 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Irfft.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Irfft.java
@@ -22,7 +22,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -30,6 +29,7 @@
 import org.tensorflow.types.TFloat32;
 import org.tensorflow.types.TInt32;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * Inverse real-valued fast Fourier transform.
@@ -51,7 +51,7 @@
  * @param <U> data type for {@code output()} output
  */
 @Operator(group = "signal")
-public final class Irfft<U extends Tensor & TNumber> extends RawOp implements Operand<U> {
+public final class Irfft<U extends TNumber> extends RawOp implements Operand<U> {
   
   /**
    * Factory method to create a class wrapping a new Irfft operation.
@@ -63,10 +63,10 @@ public final class Irfft<U extends Tensor & TNumber> extends RawOp implements Op
    * @return a new instance of Irfft
    */
   @Endpoint(describeByClass = true)
-  public static <U extends Tensor & TNumber, T extends Tensor> Irfft<U> create(Scope scope, Operand<T> input, Operand<TInt32> fftLength, DataType<U> Treal) {
+  public static <U extends TNumber, T extends TType> Irfft<U> create(Scope scope, Operand<T> input, Operand<TInt32> fftLength, DataType<U> Treal) {
     OperationBuilder opBuilder = scope.env().opBuilder("IRFFT", scope.makeOpName("Irfft"));
-    opBuilder.addInput(input.asOutput());
-    opBuilder.addInput(fftLength.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
+    opBuilder.addInput(fftLength.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("Treal", Treal);
     return new Irfft<U>(opBuilder.build());
@@ -81,7 +81,7 @@ public static <U extends Tensor & TNumber, T extends Tensor> Irfft<U> create(Sco
    * @return a new instance of Irfft
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> Irfft<TFloat32> create(Scope scope, Operand<T> input, Operand<TInt32> fftLength) {
+  public static <T extends TType> Irfft<TFloat32> create(Scope scope, Operand<T> input, Operand<TInt32> fftLength) {
     return create(scope, input, fftLength, TFloat32.DTYPE);
   }
   
@@ -99,7 +99,7 @@ public Output<U> output() {
   }
   
   @Override
-  public Output<U> asOutput() {
+  public Output<U> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Irfft2d.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Irfft2d.java
index 2b525b692ad..27af3c16aef 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Irfft2d.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Irfft2d.java
@@ -22,7 +22,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -30,6 +29,7 @@
 import org.tensorflow.types.TFloat32;
 import org.tensorflow.types.TInt32;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * Inverse 2D real-valued fast Fourier transform.
@@ -52,7 +52,7 @@
  * @param <U> data type for {@code output()} output
  */
 @Operator(group = "signal")
-public final class Irfft2d<U extends Tensor & TNumber> extends RawOp implements Operand<U> {
+public final class Irfft2d<U extends TNumber> extends RawOp implements Operand<U> {
   
   /**
    * Factory method to create a class wrapping a new Irfft2d operation.
@@ -64,10 +64,10 @@ public final class Irfft2d<U extends Tensor & TNumber> extends RawOp implements
    * @return a new instance of Irfft2d
    */
   @Endpoint(describeByClass = true)
-  public static <U extends Tensor & TNumber, T extends Tensor> Irfft2d<U> create(Scope scope, Operand<T> input, Operand<TInt32> fftLength, DataType<U> Treal) {
+  public static <U extends TNumber, T extends TType> Irfft2d<U> create(Scope scope, Operand<T> input, Operand<TInt32> fftLength, DataType<U> Treal) {
     OperationBuilder opBuilder = scope.env().opBuilder("IRFFT2D", scope.makeOpName("Irfft2d"));
-    opBuilder.addInput(input.asOutput());
-    opBuilder.addInput(fftLength.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
+    opBuilder.addInput(fftLength.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("Treal", Treal);
     return new Irfft2d<U>(opBuilder.build());
@@ -82,7 +82,7 @@ public static <U extends Tensor & TNumber, T extends Tensor> Irfft2d<U> create(S
    * @return a new instance of Irfft2d
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> Irfft2d<TFloat32> create(Scope scope, Operand<T> input, Operand<TInt32> fftLength) {
+  public static <T extends TType> Irfft2d<TFloat32> create(Scope scope, Operand<T> input, Operand<TInt32> fftLength) {
     return create(scope, input, fftLength, TFloat32.DTYPE);
   }
   
@@ -100,7 +100,7 @@ public Output<U> output() {
   }
   
   @Override
-  public Output<U> asOutput() {
+  public Output<U> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Irfft3d.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Irfft3d.java
index a36fc387339..25a5ade8b6f 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Irfft3d.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Irfft3d.java
@@ -22,7 +22,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -30,6 +29,7 @@
 import org.tensorflow.types.TFloat32;
 import org.tensorflow.types.TInt32;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * Inverse 3D real-valued fast Fourier transform.
@@ -52,7 +52,7 @@
  * @param <U> data type for {@code output()} output
  */
 @Operator(group = "signal")
-public final class Irfft3d<U extends Tensor & TNumber> extends RawOp implements Operand<U> {
+public final class Irfft3d<U extends TNumber> extends RawOp implements Operand<U> {
   
   /**
    * Factory method to create a class wrapping a new Irfft3d operation.
@@ -64,10 +64,10 @@ public final class Irfft3d<U extends Tensor & TNumber> extends RawOp implements
    * @return a new instance of Irfft3d
    */
   @Endpoint(describeByClass = true)
-  public static <U extends Tensor & TNumber, T extends Tensor> Irfft3d<U> create(Scope scope, Operand<T> input, Operand<TInt32> fftLength, DataType<U> Treal) {
+  public static <U extends TNumber, T extends TType> Irfft3d<U> create(Scope scope, Operand<T> input, Operand<TInt32> fftLength, DataType<U> Treal) {
     OperationBuilder opBuilder = scope.env().opBuilder("IRFFT3D", scope.makeOpName("Irfft3d"));
-    opBuilder.addInput(input.asOutput());
-    opBuilder.addInput(fftLength.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
+    opBuilder.addInput(fftLength.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("Treal", Treal);
     return new Irfft3d<U>(opBuilder.build());
@@ -82,7 +82,7 @@ public static <U extends Tensor & TNumber, T extends Tensor> Irfft3d<U> create(S
    * @return a new instance of Irfft3d
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> Irfft3d<TFloat32> create(Scope scope, Operand<T> input, Operand<TInt32> fftLength) {
+  public static <T extends TType> Irfft3d<TFloat32> create(Scope scope, Operand<T> input, Operand<TInt32> fftLength) {
     return create(scope, input, fftLength, TFloat32.DTYPE);
   }
   
@@ -100,7 +100,7 @@ public Output<U> output() {
   }
   
   @Override
-  public Output<U> asOutput() {
+  public Output<U> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Rfft.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Rfft.java
index c70d871ac9a..bbd122f47fc 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Rfft.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Rfft.java
@@ -22,13 +22,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * Real-valued fast Fourier transform.
@@ -47,7 +47,7 @@
  * @param <U> data type for {@code output()} output
  */
 @Operator(group = "signal")
-public final class Rfft<U extends Tensor> extends RawOp implements Operand<U> {
+public final class Rfft<U extends TType> extends RawOp implements Operand<U> {
   
   /**
    * Factory method to create a class wrapping a new Rfft operation.
@@ -59,10 +59,10 @@ public final class Rfft<U extends Tensor> extends RawOp implements Operand<U> {
    * @return a new instance of Rfft
    */
   @Endpoint(describeByClass = true)
-  public static <U extends Tensor, T extends Tensor & TNumber> Rfft<U> create(Scope scope, Operand<T> input, Operand<TInt32> fftLength, DataType<U> Tcomplex) {
+  public static <U extends TType, T extends TNumber> Rfft<U> create(Scope scope, Operand<T> input, Operand<TInt32> fftLength, DataType<U> Tcomplex) {
     OperationBuilder opBuilder = scope.env().opBuilder("RFFT", scope.makeOpName("Rfft"));
-    opBuilder.addInput(input.asOutput());
-    opBuilder.addInput(fftLength.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
+    opBuilder.addInput(fftLength.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("Tcomplex", Tcomplex);
     return new Rfft<U>(opBuilder.build());
@@ -82,7 +82,7 @@ public Output<U> output() {
   }
   
   @Override
-  public Output<U> asOutput() {
+  public Output<U> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Rfft2d.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Rfft2d.java
index 32fae79f26c..c206590bf30 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Rfft2d.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Rfft2d.java
@@ -22,13 +22,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * 2D real-valued fast Fourier transform.
@@ -48,7 +48,7 @@
  * @param <U> data type for {@code output()} output
  */
 @Operator(group = "signal")
-public final class Rfft2d<U extends Tensor> extends RawOp implements Operand<U> {
+public final class Rfft2d<U extends TType> extends RawOp implements Operand<U> {
   
   /**
    * Factory method to create a class wrapping a new Rfft2d operation.
@@ -60,10 +60,10 @@ public final class Rfft2d<U extends Tensor> extends RawOp implements Operand<U>
    * @return a new instance of Rfft2d
    */
   @Endpoint(describeByClass = true)
-  public static <U extends Tensor, T extends Tensor & TNumber> Rfft2d<U> create(Scope scope, Operand<T> input, Operand<TInt32> fftLength, DataType<U> Tcomplex) {
+  public static <U extends TType, T extends TNumber> Rfft2d<U> create(Scope scope, Operand<T> input, Operand<TInt32> fftLength, DataType<U> Tcomplex) {
     OperationBuilder opBuilder = scope.env().opBuilder("RFFT2D", scope.makeOpName("Rfft2d"));
-    opBuilder.addInput(input.asOutput());
-    opBuilder.addInput(fftLength.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
+    opBuilder.addInput(fftLength.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("Tcomplex", Tcomplex);
     return new Rfft2d<U>(opBuilder.build());
@@ -84,7 +84,7 @@ public Output<U> output() {
   }
   
   @Override
-  public Output<U> asOutput() {
+  public Output<U> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Rfft3d.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Rfft3d.java
index 5524ec596c4..36d3facf770 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Rfft3d.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Rfft3d.java
@@ -22,13 +22,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * 3D real-valued fast Fourier transform.
@@ -48,7 +48,7 @@
  * @param <U> data type for {@code output()} output
  */
 @Operator(group = "signal")
-public final class Rfft3d<U extends Tensor> extends RawOp implements Operand<U> {
+public final class Rfft3d<U extends TType> extends RawOp implements Operand<U> {
   
   /**
    * Factory method to create a class wrapping a new Rfft3d operation.
@@ -60,10 +60,10 @@ public final class Rfft3d<U extends Tensor> extends RawOp implements Operand<U>
    * @return a new instance of Rfft3d
    */
   @Endpoint(describeByClass = true)
-  public static <U extends Tensor, T extends Tensor & TNumber> Rfft3d<U> create(Scope scope, Operand<T> input, Operand<TInt32> fftLength, DataType<U> Tcomplex) {
+  public static <U extends TType, T extends TNumber> Rfft3d<U> create(Scope scope, Operand<T> input, Operand<TInt32> fftLength, DataType<U> Tcomplex) {
     OperationBuilder opBuilder = scope.env().opBuilder("RFFT3D", scope.makeOpName("Rfft3d"));
-    opBuilder.addInput(input.asOutput());
-    opBuilder.addInput(fftLength.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
+    opBuilder.addInput(fftLength.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("Tcomplex", Tcomplex);
     return new Rfft3d<U>(opBuilder.build());
@@ -84,7 +84,7 @@ public Output<U> output() {
   }
   
   @Override
-  public Output<U> asOutput() {
+  public Output<U> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/AddManySparseToTensorsMap.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/AddManySparseToTensorsMap.java
index 6a49325a74b..1a5adb2e8c0 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/AddManySparseToTensorsMap.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/AddManySparseToTensorsMap.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
 
 /**
  * Add an `N`-minibatch `SparseTensor` to a `SparseTensorsMap`, return `N` handles.
@@ -98,11 +98,11 @@ private Options() {
    * @return a new instance of AddManySparseToTensorsMap
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> AddManySparseToTensorsMap create(Scope scope, Operand<TInt64> sparseIndices, Operand<T> sparseValues, Operand<TInt64> sparseShape, Options... options) {
+  public static <T extends TType> AddManySparseToTensorsMap create(Scope scope, Operand<TInt64> sparseIndices, Operand<T> sparseValues, Operand<TInt64> sparseShape, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("AddManySparseToTensorsMap", scope.makeOpName("AddManySparseToTensorsMap"));
-    opBuilder.addInput(sparseIndices.asOutput());
-    opBuilder.addInput(sparseValues.asOutput());
-    opBuilder.addInput(sparseShape.asOutput());
+    opBuilder.addInput(sparseIndices.asOutput(scope));
+    opBuilder.addInput(sparseValues.asOutput(scope));
+    opBuilder.addInput(sparseShape.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -141,7 +141,7 @@ public Output<TInt64> sparseHandles() {
   }
   
   @Override
-  public Output<TInt64> asOutput() {
+  public Output<TInt64> asOutput(Scope scope) {
     return sparseHandles;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/AddSparseToTensorsMap.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/AddSparseToTensorsMap.java
index b8b14b04d91..936f21954db 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/AddSparseToTensorsMap.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/AddSparseToTensorsMap.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
 
 /**
  * Add a `SparseTensor` to a `SparseTensorsMap` return its handle.
@@ -89,11 +89,11 @@ private Options() {
    * @return a new instance of AddSparseToTensorsMap
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> AddSparseToTensorsMap create(Scope scope, Operand<TInt64> sparseIndices, Operand<T> sparseValues, Operand<TInt64> sparseShape, Options... options) {
+  public static <T extends TType> AddSparseToTensorsMap create(Scope scope, Operand<TInt64> sparseIndices, Operand<T> sparseValues, Operand<TInt64> sparseShape, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("AddSparseToTensorsMap", scope.makeOpName("AddSparseToTensorsMap"));
-    opBuilder.addInput(sparseIndices.asOutput());
-    opBuilder.addInput(sparseValues.asOutput());
-    opBuilder.addInput(sparseShape.asOutput());
+    opBuilder.addInput(sparseIndices.asOutput(scope));
+    opBuilder.addInput(sparseValues.asOutput(scope));
+    opBuilder.addInput(sparseShape.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -132,7 +132,7 @@ public Output<TInt64> sparseHandle() {
   }
   
   @Override
-  public Output<TInt64> asOutput() {
+  public Output<TInt64> asOutput(Scope scope) {
     return sparseHandle;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/DenseToDenseSetOperation.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/DenseToDenseSetOperation.java
index d8b50867170..dacf7725a02 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/DenseToDenseSetOperation.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/DenseToDenseSetOperation.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
 
 /**
  * Applies set operation along last dimension of 2 `Tensor` inputs.
@@ -42,7 +42,7 @@
  * @param <T> data type for {@code resultValues()} output
  */
 @Operator(group = "sparse")
-public final class DenseToDenseSetOperation<T extends Tensor> extends RawOp {
+public final class DenseToDenseSetOperation<T extends TType> extends RawOp {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.sparse.DenseToDenseSetOperation}
@@ -76,10 +76,10 @@ private Options() {
    * @return a new instance of DenseToDenseSetOperation
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> DenseToDenseSetOperation<T> create(Scope scope, Operand<T> set1, Operand<T> set2, String setOperation, Options... options) {
+  public static <T extends TType> DenseToDenseSetOperation<T> create(Scope scope, Operand<T> set1, Operand<T> set2, String setOperation, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("DenseToDenseSetOperation", scope.makeOpName("DenseToDenseSetOperation"));
-    opBuilder.addInput(set1.asOutput());
-    opBuilder.addInput(set2.asOutput());
+    opBuilder.addInput(set1.asOutput(scope));
+    opBuilder.addInput(set2.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("set_operation", setOperation);
     if (options != null) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/DenseToSparseSetOperation.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/DenseToSparseSetOperation.java
index bc83ee8e6b5..f47e96674f8 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/DenseToSparseSetOperation.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/DenseToSparseSetOperation.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
 
 /**
  * Applies set operation along last dimension of `Tensor` and `SparseTensor`.
@@ -50,7 +50,7 @@
  * @param <T> data type for {@code resultValues()} output
  */
 @Operator(group = "sparse")
-public final class DenseToSparseSetOperation<T extends Tensor> extends RawOp {
+public final class DenseToSparseSetOperation<T extends TType> extends RawOp {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.sparse.DenseToSparseSetOperation}
@@ -89,12 +89,12 @@ private Options() {
    * @return a new instance of DenseToSparseSetOperation
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> DenseToSparseSetOperation<T> create(Scope scope, Operand<T> set1, Operand<TInt64> set2Indices, Operand<T> set2Values, Operand<TInt64> set2Shape, String setOperation, Options... options) {
+  public static <T extends TType> DenseToSparseSetOperation<T> create(Scope scope, Operand<T> set1, Operand<TInt64> set2Indices, Operand<T> set2Values, Operand<TInt64> set2Shape, String setOperation, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("DenseToSparseSetOperation", scope.makeOpName("DenseToSparseSetOperation"));
-    opBuilder.addInput(set1.asOutput());
-    opBuilder.addInput(set2Indices.asOutput());
-    opBuilder.addInput(set2Values.asOutput());
-    opBuilder.addInput(set2Shape.asOutput());
+    opBuilder.addInput(set1.asOutput(scope));
+    opBuilder.addInput(set2Indices.asOutput(scope));
+    opBuilder.addInput(set2Values.asOutput(scope));
+    opBuilder.addInput(set2Shape.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("set_operation", setOperation);
     if (options != null) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/DeserializeSparse.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/DeserializeSparse.java
index 33c0f45fb6a..76be1dcb2d8 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/DeserializeSparse.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/DeserializeSparse.java
@@ -22,12 +22,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
 
 /**
  * Deserialize `SparseTensor` objects.
@@ -77,7 +77,7 @@
  * @param <U> data type for {@code sparseValues()} output
  */
 @Operator(group = "sparse")
-public final class DeserializeSparse<U extends Tensor> extends RawOp {
+public final class DeserializeSparse<U extends TType> extends RawOp {
   
   /**
    * Factory method to create a class wrapping a new DeserializeSparse operation.
@@ -89,9 +89,9 @@ public final class DeserializeSparse<U extends Tensor> extends RawOp {
    * @return a new instance of DeserializeSparse
    */
   @Endpoint(describeByClass = true)
-  public static <U extends Tensor, T extends Tensor> DeserializeSparse<U> create(Scope scope, Operand<T> serializedSparse, DataType<U> dtype) {
+  public static <U extends TType, T extends TType> DeserializeSparse<U> create(Scope scope, Operand<T> serializedSparse, DataType<U> dtype) {
     OperationBuilder opBuilder = scope.env().opBuilder("DeserializeSparse", scope.makeOpName("DeserializeSparse"));
-    opBuilder.addInput(serializedSparse.asOutput());
+    opBuilder.addInput(serializedSparse.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("dtype", dtype);
     return new DeserializeSparse<U>(opBuilder.build());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseAccumulatorApplyGradient.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseAccumulatorApplyGradient.java
index 6cedd5f287f..99bf0938930 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseAccumulatorApplyGradient.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseAccumulatorApplyGradient.java
@@ -20,13 +20,13 @@
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
 import org.tensorflow.types.TString;
+import org.tensorflow.types.family.TType;
 
 /**
  * Applies a sparse gradient to a given accumulator.
@@ -54,13 +54,13 @@ public final class SparseAccumulatorApplyGradient extends RawOp {
    * @return a new instance of SparseAccumulatorApplyGradient
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> SparseAccumulatorApplyGradient create(Scope scope, Operand<TString> handle, Operand<TInt64> localStep, Operand<TInt64> gradientIndices, Operand<T> gradientValues, Operand<TInt64> gradientShape, Boolean hasKnownShape) {
+  public static <T extends TType> SparseAccumulatorApplyGradient create(Scope scope, Operand<TString> handle, Operand<TInt64> localStep, Operand<TInt64> gradientIndices, Operand<T> gradientValues, Operand<TInt64> gradientShape, Boolean hasKnownShape) {
     OperationBuilder opBuilder = scope.env().opBuilder("SparseAccumulatorApplyGradient", scope.makeOpName("SparseAccumulatorApplyGradient"));
-    opBuilder.addInput(handle.asOutput());
-    opBuilder.addInput(localStep.asOutput());
-    opBuilder.addInput(gradientIndices.asOutput());
-    opBuilder.addInput(gradientValues.asOutput());
-    opBuilder.addInput(gradientShape.asOutput());
+    opBuilder.addInput(handle.asOutput(scope));
+    opBuilder.addInput(localStep.asOutput(scope));
+    opBuilder.addInput(gradientIndices.asOutput(scope));
+    opBuilder.addInput(gradientValues.asOutput(scope));
+    opBuilder.addInput(gradientShape.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("has_known_shape", hasKnownShape);
     return new SparseAccumulatorApplyGradient(opBuilder.build());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseAccumulatorTakeGradient.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseAccumulatorTakeGradient.java
index d226209221a..50836e4c0a4 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseAccumulatorTakeGradient.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseAccumulatorTakeGradient.java
@@ -22,7 +22,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -30,6 +29,7 @@
 import org.tensorflow.types.TInt32;
 import org.tensorflow.types.TInt64;
 import org.tensorflow.types.TString;
+import org.tensorflow.types.family.TType;
 
 /**
  * Extracts the average sparse gradient in a SparseConditionalAccumulator.
@@ -44,7 +44,7 @@
  * @param <T> data type for {@code values()} output
  */
 @Operator(group = "sparse")
-public final class SparseAccumulatorTakeGradient<T extends Tensor> extends RawOp {
+public final class SparseAccumulatorTakeGradient<T extends TType> extends RawOp {
   
   /**
    * Factory method to create a class wrapping a new SparseAccumulatorTakeGradient operation.
@@ -57,10 +57,10 @@ public final class SparseAccumulatorTakeGradient<T extends Tensor> extends RawOp
    * @return a new instance of SparseAccumulatorTakeGradient
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> SparseAccumulatorTakeGradient<T> create(Scope scope, Operand<TString> handle, Operand<TInt32> numRequired, DataType<T> dtype) {
+  public static <T extends TType> SparseAccumulatorTakeGradient<T> create(Scope scope, Operand<TString> handle, Operand<TInt32> numRequired, DataType<T> dtype) {
     OperationBuilder opBuilder = scope.env().opBuilder("SparseAccumulatorTakeGradient", scope.makeOpName("SparseAccumulatorTakeGradient"));
-    opBuilder.addInput(handle.asOutput());
-    opBuilder.addInput(numRequired.asOutput());
+    opBuilder.addInput(handle.asOutput(scope));
+    opBuilder.addInput(numRequired.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("dtype", dtype);
     return new SparseAccumulatorTakeGradient<T>(opBuilder.build());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseAdd.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseAdd.java
index 5882e508deb..7377c27b799 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseAdd.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseAdd.java
@@ -21,13 +21,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * Adds two `SparseTensor` objects to produce another `SparseTensor`.
@@ -49,7 +49,7 @@
  * @param <T> data type for {@code sumValues()} output
  */
 @Operator(group = "sparse")
-public final class SparseAdd<T extends Tensor> extends RawOp {
+public final class SparseAdd<T extends TType> extends RawOp {
   
   /**
    * Factory method to create a class wrapping a new SparseAdd operation.
@@ -66,15 +66,15 @@ public final class SparseAdd<T extends Tensor> extends RawOp {
    * @return a new instance of SparseAdd
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor, U extends Tensor & TNumber> SparseAdd<T> create(Scope scope, Operand<TInt64> aIndices, Operand<T> aValues, Operand<TInt64> aShape, Operand<TInt64> bIndices, Operand<T> bValues, Operand<TInt64> bShape, Operand<U> thresh) {
+  public static <T extends TType, U extends TNumber> SparseAdd<T> create(Scope scope, Operand<TInt64> aIndices, Operand<T> aValues, Operand<TInt64> aShape, Operand<TInt64> bIndices, Operand<T> bValues, Operand<TInt64> bShape, Operand<U> thresh) {
     OperationBuilder opBuilder = scope.env().opBuilder("SparseAdd", scope.makeOpName("SparseAdd"));
-    opBuilder.addInput(aIndices.asOutput());
-    opBuilder.addInput(aValues.asOutput());
-    opBuilder.addInput(aShape.asOutput());
-    opBuilder.addInput(bIndices.asOutput());
-    opBuilder.addInput(bValues.asOutput());
-    opBuilder.addInput(bShape.asOutput());
-    opBuilder.addInput(thresh.asOutput());
+    opBuilder.addInput(aIndices.asOutput(scope));
+    opBuilder.addInput(aValues.asOutput(scope));
+    opBuilder.addInput(aShape.asOutput(scope));
+    opBuilder.addInput(bIndices.asOutput(scope));
+    opBuilder.addInput(bValues.asOutput(scope));
+    opBuilder.addInput(bShape.asOutput(scope));
+    opBuilder.addInput(thresh.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new SparseAdd<T>(opBuilder.build());
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseAddGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseAddGrad.java
index b13a206294d..f7f84d2be77 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseAddGrad.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseAddGrad.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
 
 /**
  * The gradient operator for the SparseAdd op.
@@ -39,7 +39,7 @@
  * @param <T> data type for {@code aValGrad()} output
  */
 @Operator(group = "sparse")
-public final class SparseAddGrad<T extends Tensor> extends RawOp {
+public final class SparseAddGrad<T extends TType> extends RawOp {
   
   /**
    * Factory method to create a class wrapping a new SparseAddGrad operation.
@@ -54,12 +54,12 @@ public final class SparseAddGrad<T extends Tensor> extends RawOp {
    * @return a new instance of SparseAddGrad
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> SparseAddGrad<T> create(Scope scope, Operand<T> backpropValGrad, Operand<TInt64> aIndices, Operand<TInt64> bIndices, Operand<TInt64> sumIndices) {
+  public static <T extends TType> SparseAddGrad<T> create(Scope scope, Operand<T> backpropValGrad, Operand<TInt64> aIndices, Operand<TInt64> bIndices, Operand<TInt64> sumIndices) {
     OperationBuilder opBuilder = scope.env().opBuilder("SparseAddGrad", scope.makeOpName("SparseAddGrad"));
-    opBuilder.addInput(backpropValGrad.asOutput());
-    opBuilder.addInput(aIndices.asOutput());
-    opBuilder.addInput(bIndices.asOutput());
-    opBuilder.addInput(sumIndices.asOutput());
+    opBuilder.addInput(backpropValGrad.asOutput(scope));
+    opBuilder.addInput(aIndices.asOutput(scope));
+    opBuilder.addInput(bIndices.asOutput(scope));
+    opBuilder.addInput(sumIndices.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new SparseAddGrad<T>(opBuilder.build());
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseConcat.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseConcat.java
index b61b2a0a834..53eecb4b967 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseConcat.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseConcat.java
@@ -21,13 +21,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.Operands;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
 
 /**
  * Concatenates a list of `SparseTensor` along the specified dimension.
@@ -77,7 +77,7 @@
  * @param <T> data type for {@code outputValues()} output
  */
 @Operator(group = "sparse")
-public final class SparseConcat<T extends Tensor> extends RawOp {
+public final class SparseConcat<T extends TType> extends RawOp {
   
   /**
    * Factory method to create a class wrapping a new SparseConcat operation.
@@ -91,11 +91,11 @@ public final class SparseConcat<T extends Tensor> extends RawOp {
    * @return a new instance of SparseConcat
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> SparseConcat<T> create(Scope scope, Iterable<Operand<TInt64>> indices, Iterable<Operand<T>> values, Iterable<Operand<TInt64>> shapes, Long concatDim) {
+  public static <T extends TType> SparseConcat<T> create(Scope scope, Iterable<Operand<TInt64>> indices, Iterable<Operand<T>> values, Iterable<Operand<TInt64>> shapes, Long concatDim) {
     OperationBuilder opBuilder = scope.env().opBuilder("SparseConcat", scope.makeOpName("SparseConcat"));
-    opBuilder.addInputList(Operands.asOutputs(indices));
-    opBuilder.addInputList(Operands.asOutputs(values));
-    opBuilder.addInputList(Operands.asOutputs(shapes));
+    opBuilder.addInputList(Operands.asOutputs(scope, indices));
+    opBuilder.addInputList(Operands.asOutputs(scope, values));
+    opBuilder.addInputList(Operands.asOutputs(scope, shapes));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("concat_dim", concatDim);
     return new SparseConcat<T>(opBuilder.build());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseConditionalAccumulator.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseConditionalAccumulator.java
index 4a7a5ed2ce2..7be9dee9733 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseConditionalAccumulator.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseConditionalAccumulator.java
@@ -22,13 +22,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TString;
+import org.tensorflow.types.family.TType;
 
 /**
  * A conditional accumulator for aggregating sparse gradients.
@@ -92,7 +92,7 @@ private Options() {
    * @return a new instance of SparseConditionalAccumulator
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> SparseConditionalAccumulator create(Scope scope, DataType<T> dtype, Shape shape, Options... options) {
+  public static <T extends TType> SparseConditionalAccumulator create(Scope scope, DataType<T> dtype, Shape shape, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("SparseConditionalAccumulator", scope.makeOpName("SparseConditionalAccumulator"));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("dtype", dtype);
@@ -144,7 +144,7 @@ public Output<TString> handle() {
   }
   
   @Override
-  public Output<TString> asOutput() {
+  public Output<TString> asOutput(Scope scope) {
     return handle;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseCross.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseCross.java
index dfc030fc809..0c63e102925 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseCross.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseCross.java
@@ -22,13 +22,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.Operands;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
 
 /**
  * Generates sparse cross from a list of sparse and dense tensors.
@@ -73,7 +73,7 @@
  * @param <T> data type for {@code outputValues()} output
  */
 @Operator(group = "sparse")
-public final class SparseCross<T extends Tensor> extends RawOp {
+public final class SparseCross<T extends TType> extends RawOp {
   
   /**
    * Factory method to create a class wrapping a new SparseCross operation.
@@ -94,12 +94,12 @@ public final class SparseCross<T extends Tensor> extends RawOp {
    * @return a new instance of SparseCross
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor, U extends Tensor> SparseCross<T> create(Scope scope, Iterable<Operand<TInt64>> indices, Iterable<Operand<?>> values, Iterable<Operand<TInt64>> shapes, Iterable<Operand<?>> denseInputs, Boolean hashedOutput, Long numBuckets, Long hashKey, DataType<T> outType, DataType<U> internalType) {
+  public static <T extends TType, U extends TType> SparseCross<T> create(Scope scope, Iterable<Operand<TInt64>> indices, Iterable<Operand<?>> values, Iterable<Operand<TInt64>> shapes, Iterable<Operand<?>> denseInputs, Boolean hashedOutput, Long numBuckets, Long hashKey, DataType<T> outType, DataType<U> internalType) {
     OperationBuilder opBuilder = scope.env().opBuilder("SparseCross", scope.makeOpName("SparseCross"));
-    opBuilder.addInputList(Operands.asOutputs(indices));
-    opBuilder.addInputList(Operands.asOutputs(values));
-    opBuilder.addInputList(Operands.asOutputs(shapes));
-    opBuilder.addInputList(Operands.asOutputs(denseInputs));
+    opBuilder.addInputList(Operands.asOutputs(scope, indices));
+    opBuilder.addInputList(Operands.asOutputs(scope, values));
+    opBuilder.addInputList(Operands.asOutputs(scope, shapes));
+    opBuilder.addInputList(Operands.asOutputs(scope, denseInputs));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("hashed_output", hashedOutput);
     opBuilder.setAttr("num_buckets", numBuckets);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseDenseCwiseAdd.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseDenseCwiseAdd.java
index c07d4c50e22..f3ceaa81bbc 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseDenseCwiseAdd.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseDenseCwiseAdd.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
 
 /**
  * Adds up a SparseTensor and a dense Tensor, using these special rules:
@@ -43,7 +43,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "sparse")
-public final class SparseDenseCwiseAdd<T extends Tensor> extends RawOp implements Operand<T> {
+public final class SparseDenseCwiseAdd<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new SparseDenseCwiseAdd operation.
@@ -57,12 +57,12 @@ public final class SparseDenseCwiseAdd<T extends Tensor> extends RawOp implement
    * @return a new instance of SparseDenseCwiseAdd
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> SparseDenseCwiseAdd<T> create(Scope scope, Operand<TInt64> spIndices, Operand<T> spValues, Operand<TInt64> spShape, Operand<T> dense) {
+  public static <T extends TType> SparseDenseCwiseAdd<T> create(Scope scope, Operand<TInt64> spIndices, Operand<T> spValues, Operand<TInt64> spShape, Operand<T> dense) {
     OperationBuilder opBuilder = scope.env().opBuilder("SparseDenseCwiseAdd", scope.makeOpName("SparseDenseCwiseAdd"));
-    opBuilder.addInput(spIndices.asOutput());
-    opBuilder.addInput(spValues.asOutput());
-    opBuilder.addInput(spShape.asOutput());
-    opBuilder.addInput(dense.asOutput());
+    opBuilder.addInput(spIndices.asOutput(scope));
+    opBuilder.addInput(spValues.asOutput(scope));
+    opBuilder.addInput(spShape.asOutput(scope));
+    opBuilder.addInput(dense.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new SparseDenseCwiseAdd<T>(opBuilder.build());
   }
@@ -75,7 +75,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseDenseCwiseDiv.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseDenseCwiseDiv.java
index b2ac0fe6a0c..3d377ac1485 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseDenseCwiseDiv.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseDenseCwiseDiv.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
 
 /**
  * Component-wise divides a SparseTensor by a dense Tensor.
@@ -37,7 +37,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "sparse")
-public final class SparseDenseCwiseDiv<T extends Tensor> extends RawOp implements Operand<T> {
+public final class SparseDenseCwiseDiv<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new SparseDenseCwiseDiv operation.
@@ -51,12 +51,12 @@ public final class SparseDenseCwiseDiv<T extends Tensor> extends RawOp implement
    * @return a new instance of SparseDenseCwiseDiv
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> SparseDenseCwiseDiv<T> create(Scope scope, Operand<TInt64> spIndices, Operand<T> spValues, Operand<TInt64> spShape, Operand<T> dense) {
+  public static <T extends TType> SparseDenseCwiseDiv<T> create(Scope scope, Operand<TInt64> spIndices, Operand<T> spValues, Operand<TInt64> spShape, Operand<T> dense) {
     OperationBuilder opBuilder = scope.env().opBuilder("SparseDenseCwiseDiv", scope.makeOpName("SparseDenseCwiseDiv"));
-    opBuilder.addInput(spIndices.asOutput());
-    opBuilder.addInput(spValues.asOutput());
-    opBuilder.addInput(spShape.asOutput());
-    opBuilder.addInput(dense.asOutput());
+    opBuilder.addInput(spIndices.asOutput(scope));
+    opBuilder.addInput(spValues.asOutput(scope));
+    opBuilder.addInput(spShape.asOutput(scope));
+    opBuilder.addInput(dense.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new SparseDenseCwiseDiv<T>(opBuilder.build());
   }
@@ -69,7 +69,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseDenseCwiseMul.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseDenseCwiseMul.java
index cc36ff3cf9f..f6d0b868c90 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseDenseCwiseMul.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseDenseCwiseMul.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
 
 /**
  * Component-wise multiplies a SparseTensor by a dense Tensor.
@@ -41,7 +41,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "sparse")
-public final class SparseDenseCwiseMul<T extends Tensor> extends RawOp implements Operand<T> {
+public final class SparseDenseCwiseMul<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new SparseDenseCwiseMul operation.
@@ -55,12 +55,12 @@ public final class SparseDenseCwiseMul<T extends Tensor> extends RawOp implement
    * @return a new instance of SparseDenseCwiseMul
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> SparseDenseCwiseMul<T> create(Scope scope, Operand<TInt64> spIndices, Operand<T> spValues, Operand<TInt64> spShape, Operand<T> dense) {
+  public static <T extends TType> SparseDenseCwiseMul<T> create(Scope scope, Operand<TInt64> spIndices, Operand<T> spValues, Operand<TInt64> spShape, Operand<T> dense) {
     OperationBuilder opBuilder = scope.env().opBuilder("SparseDenseCwiseMul", scope.makeOpName("SparseDenseCwiseMul"));
-    opBuilder.addInput(spIndices.asOutput());
-    opBuilder.addInput(spValues.asOutput());
-    opBuilder.addInput(spShape.asOutput());
-    opBuilder.addInput(dense.asOutput());
+    opBuilder.addInput(spIndices.asOutput(scope));
+    opBuilder.addInput(spValues.asOutput(scope));
+    opBuilder.addInput(spShape.asOutput(scope));
+    opBuilder.addInput(dense.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new SparseDenseCwiseMul<T>(opBuilder.build());
   }
@@ -73,7 +73,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseFillEmptyRows.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseFillEmptyRows.java
index 6a05b2ddd9f..c54860b62ca 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseFillEmptyRows.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseFillEmptyRows.java
@@ -21,13 +21,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TBool;
 import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
 
 /**
  * Fills empty rows in the input 2-D `SparseTensor` with a default value.
@@ -72,7 +72,7 @@
  * @param <T> data type for {@code outputValues()} output
  */
 @Operator(group = "sparse")
-public final class SparseFillEmptyRows<T extends Tensor> extends RawOp {
+public final class SparseFillEmptyRows<T extends TType> extends RawOp {
   
   /**
    * Factory method to create a class wrapping a new SparseFillEmptyRows operation.
@@ -87,12 +87,12 @@ public final class SparseFillEmptyRows<T extends Tensor> extends RawOp {
    * @return a new instance of SparseFillEmptyRows
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> SparseFillEmptyRows<T> create(Scope scope, Operand<TInt64> indices, Operand<T> values, Operand<TInt64> denseShape, Operand<T> defaultValue) {
+  public static <T extends TType> SparseFillEmptyRows<T> create(Scope scope, Operand<TInt64> indices, Operand<T> values, Operand<TInt64> denseShape, Operand<T> defaultValue) {
     OperationBuilder opBuilder = scope.env().opBuilder("SparseFillEmptyRows", scope.makeOpName("SparseFillEmptyRows"));
-    opBuilder.addInput(indices.asOutput());
-    opBuilder.addInput(values.asOutput());
-    opBuilder.addInput(denseShape.asOutput());
-    opBuilder.addInput(defaultValue.asOutput());
+    opBuilder.addInput(indices.asOutput(scope));
+    opBuilder.addInput(values.asOutput(scope));
+    opBuilder.addInput(denseShape.asOutput(scope));
+    opBuilder.addInput(defaultValue.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new SparseFillEmptyRows<T>(opBuilder.build());
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseFillEmptyRowsGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseFillEmptyRowsGrad.java
index a8688193f67..bf1123de6a2 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseFillEmptyRowsGrad.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseFillEmptyRowsGrad.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
 
 /**
  * The gradient of SparseFillEmptyRows.
@@ -43,7 +43,7 @@
  * @param <T> data type for {@code dValues()} output
  */
 @Operator(group = "sparse")
-public final class SparseFillEmptyRowsGrad<T extends Tensor> extends RawOp {
+public final class SparseFillEmptyRowsGrad<T extends TType> extends RawOp {
   
   /**
    * Factory method to create a class wrapping a new SparseFillEmptyRowsGrad operation.
@@ -54,10 +54,10 @@ public final class SparseFillEmptyRowsGrad<T extends Tensor> extends RawOp {
    * @return a new instance of SparseFillEmptyRowsGrad
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> SparseFillEmptyRowsGrad<T> create(Scope scope, Operand<TInt64> reverseIndexMap, Operand<T> gradValues) {
+  public static <T extends TType> SparseFillEmptyRowsGrad<T> create(Scope scope, Operand<TInt64> reverseIndexMap, Operand<T> gradValues) {
     OperationBuilder opBuilder = scope.env().opBuilder("SparseFillEmptyRowsGrad", scope.makeOpName("SparseFillEmptyRowsGrad"));
-    opBuilder.addInput(reverseIndexMap.asOutput());
-    opBuilder.addInput(gradValues.asOutput());
+    opBuilder.addInput(reverseIndexMap.asOutput(scope));
+    opBuilder.addInput(gradValues.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new SparseFillEmptyRowsGrad<T>(opBuilder.build());
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseMatMul.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseMatMul.java
index 03ebba44774..45ba5453179 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseMatMul.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseMatMul.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -101,10 +100,10 @@ private Options() {
    * @return a new instance of SparseMatMul
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber, U extends Tensor & TNumber> SparseMatMul create(Scope scope, Operand<T> a, Operand<U> b, Options... options) {
+  public static <T extends TNumber, U extends TNumber> SparseMatMul create(Scope scope, Operand<T> a, Operand<U> b, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("SparseMatMul", scope.makeOpName("SparseMatMul"));
-    opBuilder.addInput(a.asOutput());
-    opBuilder.addInput(b.asOutput());
+    opBuilder.addInput(a.asOutput(scope));
+    opBuilder.addInput(b.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -160,7 +159,7 @@ public Output<TFloat32> product() {
   }
   
   @Override
-  public Output<TFloat32> asOutput() {
+  public Output<TFloat32> asOutput(Scope scope) {
     return product;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseReduceMax.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseReduceMax.java
index f21237e0352..1f4803c2d1a 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseReduceMax.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseReduceMax.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -49,7 +48,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "sparse")
-public final class SparseReduceMax<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class SparseReduceMax<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.sparse.SparseReduceMax}
@@ -83,12 +82,12 @@ private Options() {
    * @return a new instance of SparseReduceMax
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> SparseReduceMax<T> create(Scope scope, Operand<TInt64> inputIndices, Operand<T> inputValues, Operand<TInt64> inputShape, Operand<TInt32> reductionAxes, Options... options) {
+  public static <T extends TNumber> SparseReduceMax<T> create(Scope scope, Operand<TInt64> inputIndices, Operand<T> inputValues, Operand<TInt64> inputShape, Operand<TInt32> reductionAxes, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("SparseReduceMax", scope.makeOpName("SparseReduceMax"));
-    opBuilder.addInput(inputIndices.asOutput());
-    opBuilder.addInput(inputValues.asOutput());
-    opBuilder.addInput(inputShape.asOutput());
-    opBuilder.addInput(reductionAxes.asOutput());
+    opBuilder.addInput(inputIndices.asOutput(scope));
+    opBuilder.addInput(inputValues.asOutput(scope));
+    opBuilder.addInput(inputShape.asOutput(scope));
+    opBuilder.addInput(reductionAxes.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -115,7 +114,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseReduceMaxSparse.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseReduceMaxSparse.java
index a9d196426e9..448d70903b6 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseReduceMaxSparse.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseReduceMaxSparse.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -49,7 +48,7 @@
  * @param <T> data type for {@code outputValues()} output
  */
 @Operator(group = "sparse")
-public final class SparseReduceMaxSparse<T extends Tensor & TNumber> extends RawOp {
+public final class SparseReduceMaxSparse<T extends TNumber> extends RawOp {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.sparse.SparseReduceMaxSparse}
@@ -83,12 +82,12 @@ private Options() {
    * @return a new instance of SparseReduceMaxSparse
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> SparseReduceMaxSparse<T> create(Scope scope, Operand<TInt64> inputIndices, Operand<T> inputValues, Operand<TInt64> inputShape, Operand<TInt32> reductionAxes, Options... options) {
+  public static <T extends TNumber> SparseReduceMaxSparse<T> create(Scope scope, Operand<TInt64> inputIndices, Operand<T> inputValues, Operand<TInt64> inputShape, Operand<TInt32> reductionAxes, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("SparseReduceMaxSparse", scope.makeOpName("SparseReduceMaxSparse"));
-    opBuilder.addInput(inputIndices.asOutput());
-    opBuilder.addInput(inputValues.asOutput());
-    opBuilder.addInput(inputShape.asOutput());
-    opBuilder.addInput(reductionAxes.asOutput());
+    opBuilder.addInput(inputIndices.asOutput(scope));
+    opBuilder.addInput(inputValues.asOutput(scope));
+    opBuilder.addInput(inputShape.asOutput(scope));
+    opBuilder.addInput(reductionAxes.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseReduceSum.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseReduceSum.java
index daf9d4cca0e..1415dbdd867 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseReduceSum.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseReduceSum.java
@@ -21,13 +21,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
 import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
 
 /**
  * Computes the sum of elements across dimensions of a SparseTensor.
@@ -48,7 +48,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "sparse")
-public final class SparseReduceSum<T extends Tensor> extends RawOp implements Operand<T> {
+public final class SparseReduceSum<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.sparse.SparseReduceSum}
@@ -82,12 +82,12 @@ private Options() {
    * @return a new instance of SparseReduceSum
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> SparseReduceSum<T> create(Scope scope, Operand<TInt64> inputIndices, Operand<T> inputValues, Operand<TInt64> inputShape, Operand<TInt32> reductionAxes, Options... options) {
+  public static <T extends TType> SparseReduceSum<T> create(Scope scope, Operand<TInt64> inputIndices, Operand<T> inputValues, Operand<TInt64> inputShape, Operand<TInt32> reductionAxes, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("SparseReduceSum", scope.makeOpName("SparseReduceSum"));
-    opBuilder.addInput(inputIndices.asOutput());
-    opBuilder.addInput(inputValues.asOutput());
-    opBuilder.addInput(inputShape.asOutput());
-    opBuilder.addInput(reductionAxes.asOutput());
+    opBuilder.addInput(inputIndices.asOutput(scope));
+    opBuilder.addInput(inputValues.asOutput(scope));
+    opBuilder.addInput(inputShape.asOutput(scope));
+    opBuilder.addInput(reductionAxes.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -114,7 +114,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseReduceSumSparse.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseReduceSumSparse.java
index 4f4c839c694..b5ae71e226f 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseReduceSumSparse.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseReduceSumSparse.java
@@ -21,13 +21,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
 import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
 
 /**
  * Computes the sum of elements across dimensions of a SparseTensor.
@@ -48,7 +48,7 @@
  * @param <T> data type for {@code outputValues()} output
  */
 @Operator(group = "sparse")
-public final class SparseReduceSumSparse<T extends Tensor> extends RawOp {
+public final class SparseReduceSumSparse<T extends TType> extends RawOp {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.sparse.SparseReduceSumSparse}
@@ -82,12 +82,12 @@ private Options() {
    * @return a new instance of SparseReduceSumSparse
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> SparseReduceSumSparse<T> create(Scope scope, Operand<TInt64> inputIndices, Operand<T> inputValues, Operand<TInt64> inputShape, Operand<TInt32> reductionAxes, Options... options) {
+  public static <T extends TType> SparseReduceSumSparse<T> create(Scope scope, Operand<TInt64> inputIndices, Operand<T> inputValues, Operand<TInt64> inputShape, Operand<TInt32> reductionAxes, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("SparseReduceSumSparse", scope.makeOpName("SparseReduceSumSparse"));
-    opBuilder.addInput(inputIndices.asOutput());
-    opBuilder.addInput(inputValues.asOutput());
-    opBuilder.addInput(inputShape.asOutput());
-    opBuilder.addInput(reductionAxes.asOutput());
+    opBuilder.addInput(inputIndices.asOutput(scope));
+    opBuilder.addInput(inputValues.asOutput(scope));
+    opBuilder.addInput(inputShape.asOutput(scope));
+    opBuilder.addInput(reductionAxes.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseReorder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseReorder.java
index aacd2fab8e3..75ed7a3e04a 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseReorder.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseReorder.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
 
 /**
  * Reorders a SparseTensor into the canonical, row-major ordering.
@@ -43,7 +43,7 @@
  * @param <T> data type for {@code outputValues()} output
  */
 @Operator(group = "sparse")
-public final class SparseReorder<T extends Tensor> extends RawOp {
+public final class SparseReorder<T extends TType> extends RawOp {
   
   /**
    * Factory method to create a class wrapping a new SparseReorder operation.
@@ -56,11 +56,11 @@ public final class SparseReorder<T extends Tensor> extends RawOp {
    * @return a new instance of SparseReorder
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> SparseReorder<T> create(Scope scope, Operand<TInt64> inputIndices, Operand<T> inputValues, Operand<TInt64> inputShape) {
+  public static <T extends TType> SparseReorder<T> create(Scope scope, Operand<TInt64> inputIndices, Operand<T> inputValues, Operand<TInt64> inputShape) {
     OperationBuilder opBuilder = scope.env().opBuilder("SparseReorder", scope.makeOpName("SparseReorder"));
-    opBuilder.addInput(inputIndices.asOutput());
-    opBuilder.addInput(inputValues.asOutput());
-    opBuilder.addInput(inputShape.asOutput());
+    opBuilder.addInput(inputIndices.asOutput(scope));
+    opBuilder.addInput(inputValues.asOutput(scope));
+    opBuilder.addInput(inputShape.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new SparseReorder<T>(opBuilder.build());
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseReshape.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseReshape.java
index 772b19db9c5..e9a6673b759 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseReshape.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseReshape.java
@@ -62,9 +62,9 @@ public final class SparseReshape extends RawOp {
   @Endpoint(describeByClass = true)
   public static SparseReshape create(Scope scope, Operand<TInt64> inputIndices, Operand<TInt64> inputShape, Operand<TInt64> newShape) {
     OperationBuilder opBuilder = scope.env().opBuilder("SparseReshape", scope.makeOpName("SparseReshape"));
-    opBuilder.addInput(inputIndices.asOutput());
-    opBuilder.addInput(inputShape.asOutput());
-    opBuilder.addInput(newShape.asOutput());
+    opBuilder.addInput(inputIndices.asOutput(scope));
+    opBuilder.addInput(inputShape.asOutput(scope));
+    opBuilder.addInput(newShape.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new SparseReshape(opBuilder.build());
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentMean.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentMean.java
index 276076198c0..8cdeb61508f 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentMean.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentMean.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -40,7 +39,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "sparse")
-public final class SparseSegmentMean<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class SparseSegmentMean<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new SparseSegmentMean operation.
@@ -52,11 +51,11 @@ public final class SparseSegmentMean<T extends Tensor & TNumber> extends RawOp i
    * @return a new instance of SparseSegmentMean
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber, U extends Tensor & TNumber> SparseSegmentMean<T> create(Scope scope, Operand<T> data, Operand<U> indices, Operand<TInt32> segmentIds) {
+  public static <T extends TNumber, U extends TNumber> SparseSegmentMean<T> create(Scope scope, Operand<T> data, Operand<U> indices, Operand<TInt32> segmentIds) {
     OperationBuilder opBuilder = scope.env().opBuilder("SparseSegmentMean", scope.makeOpName("SparseSegmentMean"));
-    opBuilder.addInput(data.asOutput());
-    opBuilder.addInput(indices.asOutput());
-    opBuilder.addInput(segmentIds.asOutput());
+    opBuilder.addInput(data.asOutput(scope));
+    opBuilder.addInput(indices.asOutput(scope));
+    opBuilder.addInput(segmentIds.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new SparseSegmentMean<T>(opBuilder.build());
   }
@@ -70,7 +69,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentMeanGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentMeanGrad.java
index 5abe339711b..f12deda58e8 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentMeanGrad.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentMeanGrad.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -38,7 +37,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "sparse")
-public final class SparseSegmentMeanGrad<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class SparseSegmentMeanGrad<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new SparseSegmentMeanGrad operation.
@@ -51,12 +50,12 @@ public final class SparseSegmentMeanGrad<T extends Tensor & TNumber> extends Raw
    * @return a new instance of SparseSegmentMeanGrad
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber, U extends Tensor & TNumber> SparseSegmentMeanGrad<T> create(Scope scope, Operand<T> grad, Operand<U> indices, Operand<TInt32> segmentIds, Operand<TInt32> outputDim0) {
+  public static <T extends TNumber, U extends TNumber> SparseSegmentMeanGrad<T> create(Scope scope, Operand<T> grad, Operand<U> indices, Operand<TInt32> segmentIds, Operand<TInt32> outputDim0) {
     OperationBuilder opBuilder = scope.env().opBuilder("SparseSegmentMeanGrad", scope.makeOpName("SparseSegmentMeanGrad"));
-    opBuilder.addInput(grad.asOutput());
-    opBuilder.addInput(indices.asOutput());
-    opBuilder.addInput(segmentIds.asOutput());
-    opBuilder.addInput(outputDim0.asOutput());
+    opBuilder.addInput(grad.asOutput(scope));
+    opBuilder.addInput(indices.asOutput(scope));
+    opBuilder.addInput(segmentIds.asOutput(scope));
+    opBuilder.addInput(outputDim0.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new SparseSegmentMeanGrad<T>(opBuilder.build());
   }
@@ -68,7 +67,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentMeanWithNumSegments.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentMeanWithNumSegments.java
index 2e32109e991..2442e767106 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentMeanWithNumSegments.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentMeanWithNumSegments.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -42,7 +41,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "sparse")
-public final class SparseSegmentMeanWithNumSegments<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class SparseSegmentMeanWithNumSegments<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new SparseSegmentMeanWithNumSegments operation.
@@ -55,12 +54,12 @@ public final class SparseSegmentMeanWithNumSegments<T extends Tensor & TNumber>
    * @return a new instance of SparseSegmentMeanWithNumSegments
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber, U extends Tensor & TNumber, V extends Tensor & TNumber> SparseSegmentMeanWithNumSegments<T> create(Scope scope, Operand<T> data, Operand<U> indices, Operand<TInt32> segmentIds, Operand<V> numSegments) {
+  public static <T extends TNumber, U extends TNumber, V extends TNumber> SparseSegmentMeanWithNumSegments<T> create(Scope scope, Operand<T> data, Operand<U> indices, Operand<TInt32> segmentIds, Operand<V> numSegments) {
     OperationBuilder opBuilder = scope.env().opBuilder("SparseSegmentMeanWithNumSegments", scope.makeOpName("SparseSegmentMeanWithNumSegments"));
-    opBuilder.addInput(data.asOutput());
-    opBuilder.addInput(indices.asOutput());
-    opBuilder.addInput(segmentIds.asOutput());
-    opBuilder.addInput(numSegments.asOutput());
+    opBuilder.addInput(data.asOutput(scope));
+    opBuilder.addInput(indices.asOutput(scope));
+    opBuilder.addInput(segmentIds.asOutput(scope));
+    opBuilder.addInput(numSegments.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new SparseSegmentMeanWithNumSegments<T>(opBuilder.build());
   }
@@ -74,7 +73,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentSqrtN.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentSqrtN.java
index ce500c7a976..e2e2b56e4f7 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentSqrtN.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentSqrtN.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -40,7 +39,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "sparse")
-public final class SparseSegmentSqrtN<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class SparseSegmentSqrtN<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new SparseSegmentSqrtN operation.
@@ -52,11 +51,11 @@ public final class SparseSegmentSqrtN<T extends Tensor & TNumber> extends RawOp
    * @return a new instance of SparseSegmentSqrtN
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber, U extends Tensor & TNumber> SparseSegmentSqrtN<T> create(Scope scope, Operand<T> data, Operand<U> indices, Operand<TInt32> segmentIds) {
+  public static <T extends TNumber, U extends TNumber> SparseSegmentSqrtN<T> create(Scope scope, Operand<T> data, Operand<U> indices, Operand<TInt32> segmentIds) {
     OperationBuilder opBuilder = scope.env().opBuilder("SparseSegmentSqrtN", scope.makeOpName("SparseSegmentSqrtN"));
-    opBuilder.addInput(data.asOutput());
-    opBuilder.addInput(indices.asOutput());
-    opBuilder.addInput(segmentIds.asOutput());
+    opBuilder.addInput(data.asOutput(scope));
+    opBuilder.addInput(indices.asOutput(scope));
+    opBuilder.addInput(segmentIds.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new SparseSegmentSqrtN<T>(opBuilder.build());
   }
@@ -70,7 +69,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentSqrtNGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentSqrtNGrad.java
index 706e8cd9d15..93b737c96a8 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentSqrtNGrad.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentSqrtNGrad.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -38,7 +37,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "sparse")
-public final class SparseSegmentSqrtNGrad<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class SparseSegmentSqrtNGrad<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new SparseSegmentSqrtNGrad operation.
@@ -51,12 +50,12 @@ public final class SparseSegmentSqrtNGrad<T extends Tensor & TNumber> extends Ra
    * @return a new instance of SparseSegmentSqrtNGrad
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber, U extends Tensor & TNumber> SparseSegmentSqrtNGrad<T> create(Scope scope, Operand<T> grad, Operand<U> indices, Operand<TInt32> segmentIds, Operand<TInt32> outputDim0) {
+  public static <T extends TNumber, U extends TNumber> SparseSegmentSqrtNGrad<T> create(Scope scope, Operand<T> grad, Operand<U> indices, Operand<TInt32> segmentIds, Operand<TInt32> outputDim0) {
     OperationBuilder opBuilder = scope.env().opBuilder("SparseSegmentSqrtNGrad", scope.makeOpName("SparseSegmentSqrtNGrad"));
-    opBuilder.addInput(grad.asOutput());
-    opBuilder.addInput(indices.asOutput());
-    opBuilder.addInput(segmentIds.asOutput());
-    opBuilder.addInput(outputDim0.asOutput());
+    opBuilder.addInput(grad.asOutput(scope));
+    opBuilder.addInput(indices.asOutput(scope));
+    opBuilder.addInput(segmentIds.asOutput(scope));
+    opBuilder.addInput(outputDim0.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new SparseSegmentSqrtNGrad<T>(opBuilder.build());
   }
@@ -68,7 +67,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentSqrtNWithNumSegments.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentSqrtNWithNumSegments.java
index 0d6d1081a7d..a0b30e3fca1 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentSqrtNWithNumSegments.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentSqrtNWithNumSegments.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -44,7 +43,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "sparse")
-public final class SparseSegmentSqrtNWithNumSegments<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class SparseSegmentSqrtNWithNumSegments<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new SparseSegmentSqrtNWithNumSegments operation.
@@ -57,12 +56,12 @@ public final class SparseSegmentSqrtNWithNumSegments<T extends Tensor & TNumber>
    * @return a new instance of SparseSegmentSqrtNWithNumSegments
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber, U extends Tensor & TNumber, V extends Tensor & TNumber> SparseSegmentSqrtNWithNumSegments<T> create(Scope scope, Operand<T> data, Operand<U> indices, Operand<TInt32> segmentIds, Operand<V> numSegments) {
+  public static <T extends TNumber, U extends TNumber, V extends TNumber> SparseSegmentSqrtNWithNumSegments<T> create(Scope scope, Operand<T> data, Operand<U> indices, Operand<TInt32> segmentIds, Operand<V> numSegments) {
     OperationBuilder opBuilder = scope.env().opBuilder("SparseSegmentSqrtNWithNumSegments", scope.makeOpName("SparseSegmentSqrtNWithNumSegments"));
-    opBuilder.addInput(data.asOutput());
-    opBuilder.addInput(indices.asOutput());
-    opBuilder.addInput(segmentIds.asOutput());
-    opBuilder.addInput(numSegments.asOutput());
+    opBuilder.addInput(data.asOutput(scope));
+    opBuilder.addInput(indices.asOutput(scope));
+    opBuilder.addInput(segmentIds.asOutput(scope));
+    opBuilder.addInput(numSegments.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new SparseSegmentSqrtNWithNumSegments<T>(opBuilder.build());
   }
@@ -76,7 +75,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentSum.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentSum.java
index 13bcf882abe..f539b93ab86 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentSum.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentSum.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -65,7 +64,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "sparse")
-public final class SparseSegmentSum<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class SparseSegmentSum<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new SparseSegmentSum operation.
@@ -77,11 +76,11 @@ public final class SparseSegmentSum<T extends Tensor & TNumber> extends RawOp im
    * @return a new instance of SparseSegmentSum
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber, U extends Tensor & TNumber> SparseSegmentSum<T> create(Scope scope, Operand<T> data, Operand<U> indices, Operand<TInt32> segmentIds) {
+  public static <T extends TNumber, U extends TNumber> SparseSegmentSum<T> create(Scope scope, Operand<T> data, Operand<U> indices, Operand<TInt32> segmentIds) {
     OperationBuilder opBuilder = scope.env().opBuilder("SparseSegmentSum", scope.makeOpName("SparseSegmentSum"));
-    opBuilder.addInput(data.asOutput());
-    opBuilder.addInput(indices.asOutput());
-    opBuilder.addInput(segmentIds.asOutput());
+    opBuilder.addInput(data.asOutput(scope));
+    opBuilder.addInput(indices.asOutput(scope));
+    opBuilder.addInput(segmentIds.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new SparseSegmentSum<T>(opBuilder.build());
   }
@@ -95,7 +94,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentSumWithNumSegments.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentSumWithNumSegments.java
index 598d563c82b..ac8c400fa15 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentSumWithNumSegments.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentSumWithNumSegments.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -63,7 +62,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "sparse")
-public final class SparseSegmentSumWithNumSegments<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class SparseSegmentSumWithNumSegments<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new SparseSegmentSumWithNumSegments operation.
@@ -76,12 +75,12 @@ public final class SparseSegmentSumWithNumSegments<T extends Tensor & TNumber> e
    * @return a new instance of SparseSegmentSumWithNumSegments
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber, U extends Tensor & TNumber, V extends Tensor & TNumber> SparseSegmentSumWithNumSegments<T> create(Scope scope, Operand<T> data, Operand<U> indices, Operand<TInt32> segmentIds, Operand<V> numSegments) {
+  public static <T extends TNumber, U extends TNumber, V extends TNumber> SparseSegmentSumWithNumSegments<T> create(Scope scope, Operand<T> data, Operand<U> indices, Operand<TInt32> segmentIds, Operand<V> numSegments) {
     OperationBuilder opBuilder = scope.env().opBuilder("SparseSegmentSumWithNumSegments", scope.makeOpName("SparseSegmentSumWithNumSegments"));
-    opBuilder.addInput(data.asOutput());
-    opBuilder.addInput(indices.asOutput());
-    opBuilder.addInput(segmentIds.asOutput());
-    opBuilder.addInput(numSegments.asOutput());
+    opBuilder.addInput(data.asOutput(scope));
+    opBuilder.addInput(indices.asOutput(scope));
+    opBuilder.addInput(segmentIds.asOutput(scope));
+    opBuilder.addInput(numSegments.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new SparseSegmentSumWithNumSegments<T>(opBuilder.build());
   }
@@ -95,7 +94,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSlice.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSlice.java
index ce9c123ab64..3cebea9baa6 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSlice.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSlice.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
 
 /**
  * Slice a `SparseTensor` based on the `start` and `size`.
@@ -50,7 +50,7 @@
  * @param <T> data type for {@code outputValues()} output
  */
 @Operator(group = "sparse")
-public final class SparseSlice<T extends Tensor> extends RawOp {
+public final class SparseSlice<T extends TType> extends RawOp {
   
   /**
    * Factory method to create a class wrapping a new SparseSlice operation.
@@ -66,13 +66,13 @@ public final class SparseSlice<T extends Tensor> extends RawOp {
    * @return a new instance of SparseSlice
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> SparseSlice<T> create(Scope scope, Operand<TInt64> indices, Operand<T> values, Operand<TInt64> shape, Operand<TInt64> start, Operand<TInt64> size) {
+  public static <T extends TType> SparseSlice<T> create(Scope scope, Operand<TInt64> indices, Operand<T> values, Operand<TInt64> shape, Operand<TInt64> start, Operand<TInt64> size) {
     OperationBuilder opBuilder = scope.env().opBuilder("SparseSlice", scope.makeOpName("SparseSlice"));
-    opBuilder.addInput(indices.asOutput());
-    opBuilder.addInput(values.asOutput());
-    opBuilder.addInput(shape.asOutput());
-    opBuilder.addInput(start.asOutput());
-    opBuilder.addInput(size.asOutput());
+    opBuilder.addInput(indices.asOutput(scope));
+    opBuilder.addInput(values.asOutput(scope));
+    opBuilder.addInput(shape.asOutput(scope));
+    opBuilder.addInput(start.asOutput(scope));
+    opBuilder.addInput(size.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new SparseSlice<T>(opBuilder.build());
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSliceGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSliceGrad.java
index a5860715cbe..8b951621d2b 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSliceGrad.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSliceGrad.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
 
 /**
  * The gradient operator for the SparseSlice op.
@@ -38,7 +38,7 @@
  * @param <T> data type for {@code valGrad()} output
  */
 @Operator(group = "sparse")
-public final class SparseSliceGrad<T extends Tensor> extends RawOp implements Operand<T> {
+public final class SparseSliceGrad<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new SparseSliceGrad operation.
@@ -52,12 +52,12 @@ public final class SparseSliceGrad<T extends Tensor> extends RawOp implements Op
    * @return a new instance of SparseSliceGrad
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> SparseSliceGrad<T> create(Scope scope, Operand<T> backpropValGrad, Operand<TInt64> inputIndices, Operand<TInt64> inputStart, Operand<TInt64> outputIndices) {
+  public static <T extends TType> SparseSliceGrad<T> create(Scope scope, Operand<T> backpropValGrad, Operand<TInt64> inputIndices, Operand<TInt64> inputStart, Operand<TInt64> outputIndices) {
     OperationBuilder opBuilder = scope.env().opBuilder("SparseSliceGrad", scope.makeOpName("SparseSliceGrad"));
-    opBuilder.addInput(backpropValGrad.asOutput());
-    opBuilder.addInput(inputIndices.asOutput());
-    opBuilder.addInput(inputStart.asOutput());
-    opBuilder.addInput(outputIndices.asOutput());
+    opBuilder.addInput(backpropValGrad.asOutput(scope));
+    opBuilder.addInput(inputIndices.asOutput(scope));
+    opBuilder.addInput(inputStart.asOutput(scope));
+    opBuilder.addInput(outputIndices.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new SparseSliceGrad<T>(opBuilder.build());
   }
@@ -70,7 +70,7 @@ public Output<T> valGrad() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return valGrad;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSoftmax.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSoftmax.java
index 24e6a3553e9..e9a60b85f3d 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSoftmax.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSoftmax.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -51,7 +50,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "sparse")
-public final class SparseSoftmax<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class SparseSoftmax<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new SparseSoftmax operation.
@@ -64,11 +63,11 @@ public final class SparseSoftmax<T extends Tensor & TNumber> extends RawOp imple
    * @return a new instance of SparseSoftmax
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> SparseSoftmax<T> create(Scope scope, Operand<TInt64> spIndices, Operand<T> spValues, Operand<TInt64> spShape) {
+  public static <T extends TNumber> SparseSoftmax<T> create(Scope scope, Operand<TInt64> spIndices, Operand<T> spValues, Operand<TInt64> spShape) {
     OperationBuilder opBuilder = scope.env().opBuilder("SparseSoftmax", scope.makeOpName("SparseSoftmax"));
-    opBuilder.addInput(spIndices.asOutput());
-    opBuilder.addInput(spValues.asOutput());
-    opBuilder.addInput(spShape.asOutput());
+    opBuilder.addInput(spIndices.asOutput(scope));
+    opBuilder.addInput(spValues.asOutput(scope));
+    opBuilder.addInput(spShape.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new SparseSoftmax<T>(opBuilder.build());
   }
@@ -81,7 +80,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSparseMaximum.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSparseMaximum.java
index a993f2e3e7a..01cccab7b55 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSparseMaximum.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSparseMaximum.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -37,7 +36,7 @@
  * @param <T> data type for {@code outputValues()} output
  */
 @Operator(group = "sparse")
-public final class SparseSparseMaximum<T extends Tensor & TNumber> extends RawOp {
+public final class SparseSparseMaximum<T extends TNumber> extends RawOp {
   
   /**
    * Factory method to create a class wrapping a new SparseSparseMaximum operation.
@@ -53,14 +52,14 @@ public final class SparseSparseMaximum<T extends Tensor & TNumber> extends RawOp
    * @return a new instance of SparseSparseMaximum
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> SparseSparseMaximum<T> create(Scope scope, Operand<TInt64> aIndices, Operand<T> aValues, Operand<TInt64> aShape, Operand<TInt64> bIndices, Operand<T> bValues, Operand<TInt64> bShape) {
+  public static <T extends TNumber> SparseSparseMaximum<T> create(Scope scope, Operand<TInt64> aIndices, Operand<T> aValues, Operand<TInt64> aShape, Operand<TInt64> bIndices, Operand<T> bValues, Operand<TInt64> bShape) {
     OperationBuilder opBuilder = scope.env().opBuilder("SparseSparseMaximum", scope.makeOpName("SparseSparseMaximum"));
-    opBuilder.addInput(aIndices.asOutput());
-    opBuilder.addInput(aValues.asOutput());
-    opBuilder.addInput(aShape.asOutput());
-    opBuilder.addInput(bIndices.asOutput());
-    opBuilder.addInput(bValues.asOutput());
-    opBuilder.addInput(bShape.asOutput());
+    opBuilder.addInput(aIndices.asOutput(scope));
+    opBuilder.addInput(aValues.asOutput(scope));
+    opBuilder.addInput(aShape.asOutput(scope));
+    opBuilder.addInput(bIndices.asOutput(scope));
+    opBuilder.addInput(bValues.asOutput(scope));
+    opBuilder.addInput(bShape.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new SparseSparseMaximum<T>(opBuilder.build());
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSparseMinimum.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSparseMinimum.java
index 6a0d20c559b..cdfdaf03255 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSparseMinimum.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSparseMinimum.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
 
 /**
  * Returns the element-wise min of two SparseTensors.
@@ -36,7 +36,7 @@
  * @param <T> data type for {@code outputValues()} output
  */
 @Operator(group = "sparse")
-public final class SparseSparseMinimum<T extends Tensor> extends RawOp {
+public final class SparseSparseMinimum<T extends TType> extends RawOp {
   
   /**
    * Factory method to create a class wrapping a new SparseSparseMinimum operation.
@@ -52,14 +52,14 @@ public final class SparseSparseMinimum<T extends Tensor> extends RawOp {
    * @return a new instance of SparseSparseMinimum
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> SparseSparseMinimum<T> create(Scope scope, Operand<TInt64> aIndices, Operand<T> aValues, Operand<TInt64> aShape, Operand<TInt64> bIndices, Operand<T> bValues, Operand<TInt64> bShape) {
+  public static <T extends TType> SparseSparseMinimum<T> create(Scope scope, Operand<TInt64> aIndices, Operand<T> aValues, Operand<TInt64> aShape, Operand<TInt64> bIndices, Operand<T> bValues, Operand<TInt64> bShape) {
     OperationBuilder opBuilder = scope.env().opBuilder("SparseSparseMinimum", scope.makeOpName("SparseSparseMinimum"));
-    opBuilder.addInput(aIndices.asOutput());
-    opBuilder.addInput(aValues.asOutput());
-    opBuilder.addInput(aShape.asOutput());
-    opBuilder.addInput(bIndices.asOutput());
-    opBuilder.addInput(bValues.asOutput());
-    opBuilder.addInput(bShape.asOutput());
+    opBuilder.addInput(aIndices.asOutput(scope));
+    opBuilder.addInput(aValues.asOutput(scope));
+    opBuilder.addInput(aShape.asOutput(scope));
+    opBuilder.addInput(bIndices.asOutput(scope));
+    opBuilder.addInput(bValues.asOutput(scope));
+    opBuilder.addInput(bShape.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new SparseSparseMinimum<T>(opBuilder.build());
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSplit.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSplit.java
index 4970cca6cb2..ca60f9e528a 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSplit.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSplit.java
@@ -23,12 +23,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
 
 /**
  * Split a `SparseTensor` into `num_split` tensors along one dimension.
@@ -54,7 +54,7 @@
  * @param <T> data type for {@code outputValues()} output
  */
 @Operator(group = "sparse")
-public final class SparseSplit<T extends Tensor> extends RawOp {
+public final class SparseSplit<T extends TType> extends RawOp {
   
   /**
    * Factory method to create a class wrapping a new SparseSplit operation.
@@ -71,12 +71,12 @@ public final class SparseSplit<T extends Tensor> extends RawOp {
    * @return a new instance of SparseSplit
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> SparseSplit<T> create(Scope scope, Operand<TInt64> splitDim, Operand<TInt64> indices, Operand<T> values, Operand<TInt64> shape, Long numSplit) {
+  public static <T extends TType> SparseSplit<T> create(Scope scope, Operand<TInt64> splitDim, Operand<TInt64> indices, Operand<T> values, Operand<TInt64> shape, Long numSplit) {
     OperationBuilder opBuilder = scope.env().opBuilder("SparseSplit", scope.makeOpName("SparseSplit"));
-    opBuilder.addInput(splitDim.asOutput());
-    opBuilder.addInput(indices.asOutput());
-    opBuilder.addInput(values.asOutput());
-    opBuilder.addInput(shape.asOutput());
+    opBuilder.addInput(splitDim.asOutput(scope));
+    opBuilder.addInput(indices.asOutput(scope));
+    opBuilder.addInput(values.asOutput(scope));
+    opBuilder.addInput(shape.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("num_split", numSplit);
     return new SparseSplit<T>(opBuilder.build());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseTensorDenseAdd.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseTensorDenseAdd.java
index f1d0dc84d2b..527f2b3a759 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseTensorDenseAdd.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseTensorDenseAdd.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * Adds up a `SparseTensor` and a dense `Tensor`, producing a dense `Tensor`.
@@ -36,7 +36,7 @@
  * @param <U> data type for {@code output()} output
  */
 @Operator(group = "sparse")
-public final class SparseTensorDenseAdd<U extends Tensor> extends RawOp implements Operand<U> {
+public final class SparseTensorDenseAdd<U extends TType> extends RawOp implements Operand<U> {
   
   /**
    * Factory method to create a class wrapping a new SparseTensorDenseAdd operation.
@@ -49,12 +49,12 @@ public final class SparseTensorDenseAdd<U extends Tensor> extends RawOp implemen
    * @return a new instance of SparseTensorDenseAdd
    */
   @Endpoint(describeByClass = true)
-  public static <U extends Tensor, T extends Tensor & TNumber> SparseTensorDenseAdd<U> create(Scope scope, Operand<T> aIndices, Operand<U> aValues, Operand<T> aShape, Operand<U> b) {
+  public static <U extends TType, T extends TNumber> SparseTensorDenseAdd<U> create(Scope scope, Operand<T> aIndices, Operand<U> aValues, Operand<T> aShape, Operand<U> b) {
     OperationBuilder opBuilder = scope.env().opBuilder("SparseTensorDenseAdd", scope.makeOpName("SparseTensorDenseAdd"));
-    opBuilder.addInput(aIndices.asOutput());
-    opBuilder.addInput(aValues.asOutput());
-    opBuilder.addInput(aShape.asOutput());
-    opBuilder.addInput(b.asOutput());
+    opBuilder.addInput(aIndices.asOutput(scope));
+    opBuilder.addInput(aValues.asOutput(scope));
+    opBuilder.addInput(aShape.asOutput(scope));
+    opBuilder.addInput(b.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new SparseTensorDenseAdd<U>(opBuilder.build());
   }
@@ -66,7 +66,7 @@ public Output<U> output() {
   }
   
   @Override
-  public Output<U> asOutput() {
+  public Output<U> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseTensorDenseMatMul.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseTensorDenseMatMul.java
index ecfb1c780a4..c8b245d515d 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseTensorDenseMatMul.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseTensorDenseMatMul.java
@@ -21,13 +21,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * Multiply SparseTensor (of rank 2) "A" by dense matrix "B".
@@ -45,7 +45,7 @@
  * @param <U> data type for {@code product()} output
  */
 @Operator(group = "sparse")
-public final class SparseTensorDenseMatMul<U extends Tensor> extends RawOp implements Operand<U> {
+public final class SparseTensorDenseMatMul<U extends TType> extends RawOp implements Operand<U> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.sparse.SparseTensorDenseMatMul}
@@ -89,12 +89,12 @@ private Options() {
    * @return a new instance of SparseTensorDenseMatMul
    */
   @Endpoint(describeByClass = true)
-  public static <U extends Tensor, T extends Tensor & TNumber> SparseTensorDenseMatMul<U> create(Scope scope, Operand<T> aIndices, Operand<U> aValues, Operand<TInt64> aShape, Operand<U> b, Options... options) {
+  public static <U extends TType, T extends TNumber> SparseTensorDenseMatMul<U> create(Scope scope, Operand<T> aIndices, Operand<U> aValues, Operand<TInt64> aShape, Operand<U> b, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("SparseTensorDenseMatMul", scope.makeOpName("SparseTensorDenseMatMul"));
-    opBuilder.addInput(aIndices.asOutput());
-    opBuilder.addInput(aValues.asOutput());
-    opBuilder.addInput(aShape.asOutput());
-    opBuilder.addInput(b.asOutput());
+    opBuilder.addInput(aIndices.asOutput(scope));
+    opBuilder.addInput(aValues.asOutput(scope));
+    opBuilder.addInput(aShape.asOutput(scope));
+    opBuilder.addInput(b.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -132,7 +132,7 @@ public Output<U> product() {
   }
   
   @Override
-  public Output<U> asOutput() {
+  public Output<U> asOutput(Scope scope) {
     return product;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseToDense.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseToDense.java
index ff006e0cf5a..27f5f67a4e9 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseToDense.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseToDense.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * Converts a sparse representation into a dense tensor.
@@ -52,7 +52,7 @@
  * @param <U> data type for {@code dense()} output
  */
 @Operator(group = "sparse")
-public final class SparseToDense<U extends Tensor> extends RawOp implements Operand<U> {
+public final class SparseToDense<U extends TType> extends RawOp implements Operand<U> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.sparse.SparseToDense}
@@ -89,12 +89,12 @@ private Options() {
    * @return a new instance of SparseToDense
    */
   @Endpoint(describeByClass = true)
-  public static <U extends Tensor, T extends Tensor & TNumber> SparseToDense<U> create(Scope scope, Operand<T> sparseIndices, Operand<T> outputShape, Operand<U> sparseValues, Operand<U> defaultValue, Options... options) {
+  public static <U extends TType, T extends TNumber> SparseToDense<U> create(Scope scope, Operand<T> sparseIndices, Operand<T> outputShape, Operand<U> sparseValues, Operand<U> defaultValue, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("SparseToDense", scope.makeOpName("SparseToDense"));
-    opBuilder.addInput(sparseIndices.asOutput());
-    opBuilder.addInput(outputShape.asOutput());
-    opBuilder.addInput(sparseValues.asOutput());
-    opBuilder.addInput(defaultValue.asOutput());
+    opBuilder.addInput(sparseIndices.asOutput(scope));
+    opBuilder.addInput(outputShape.asOutput(scope));
+    opBuilder.addInput(sparseValues.asOutput(scope));
+    opBuilder.addInput(defaultValue.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -122,7 +122,7 @@ public Output<U> dense() {
   }
   
   @Override
-  public Output<U> asOutput() {
+  public Output<U> asOutput(Scope scope) {
     return dense;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseToSparseSetOperation.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseToSparseSetOperation.java
index c0284a23632..1d084cb04f7 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseToSparseSetOperation.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseToSparseSetOperation.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
 
 /**
  * Applies set operation along last dimension of 2 `SparseTensor` inputs.
@@ -58,7 +58,7 @@
  * @param <T> data type for {@code resultValues()} output
  */
 @Operator(group = "sparse")
-public final class SparseToSparseSetOperation<T extends Tensor> extends RawOp {
+public final class SparseToSparseSetOperation<T extends TType> extends RawOp {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.sparse.SparseToSparseSetOperation}
@@ -102,14 +102,14 @@ private Options() {
    * @return a new instance of SparseToSparseSetOperation
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> SparseToSparseSetOperation<T> create(Scope scope, Operand<TInt64> set1Indices, Operand<T> set1Values, Operand<TInt64> set1Shape, Operand<TInt64> set2Indices, Operand<T> set2Values, Operand<TInt64> set2Shape, String setOperation, Options... options) {
+  public static <T extends TType> SparseToSparseSetOperation<T> create(Scope scope, Operand<TInt64> set1Indices, Operand<T> set1Values, Operand<TInt64> set1Shape, Operand<TInt64> set2Indices, Operand<T> set2Values, Operand<TInt64> set2Shape, String setOperation, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("SparseToSparseSetOperation", scope.makeOpName("SparseToSparseSetOperation"));
-    opBuilder.addInput(set1Indices.asOutput());
-    opBuilder.addInput(set1Values.asOutput());
-    opBuilder.addInput(set1Shape.asOutput());
-    opBuilder.addInput(set2Indices.asOutput());
-    opBuilder.addInput(set2Values.asOutput());
-    opBuilder.addInput(set2Shape.asOutput());
+    opBuilder.addInput(set1Indices.asOutput(scope));
+    opBuilder.addInput(set1Values.asOutput(scope));
+    opBuilder.addInput(set1Shape.asOutput(scope));
+    opBuilder.addInput(set2Indices.asOutput(scope));
+    opBuilder.addInput(set2Values.asOutput(scope));
+    opBuilder.addInput(set2Shape.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("set_operation", setOperation);
     if (options != null) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/TakeManySparseFromTensorsMap.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/TakeManySparseFromTensorsMap.java
index 0909bf70863..c8af50dc2bc 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/TakeManySparseFromTensorsMap.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/TakeManySparseFromTensorsMap.java
@@ -22,12 +22,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
 
 /**
  * Read `SparseTensors` from a `SparseTensorsMap` and concatenate them.
@@ -80,7 +80,7 @@
  * @param <T> data type for {@code sparseValues()} output
  */
 @Operator(group = "sparse")
-public final class TakeManySparseFromTensorsMap<T extends Tensor> extends RawOp {
+public final class TakeManySparseFromTensorsMap<T extends TType> extends RawOp {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.sparse.TakeManySparseFromTensorsMap}
@@ -124,9 +124,9 @@ private Options() {
    * @return a new instance of TakeManySparseFromTensorsMap
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> TakeManySparseFromTensorsMap<T> create(Scope scope, Operand<TInt64> sparseHandles, DataType<T> dtype, Options... options) {
+  public static <T extends TType> TakeManySparseFromTensorsMap<T> create(Scope scope, Operand<TInt64> sparseHandles, DataType<T> dtype, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("TakeManySparseFromTensorsMap", scope.makeOpName("TakeManySparseFromTensorsMap"));
-    opBuilder.addInput(sparseHandles.asOutput());
+    opBuilder.addInput(sparseHandles.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("dtype", dtype);
     if (options != null) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/Join.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/Join.java
index 633457793a3..f52eff740ef 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/Join.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/Join.java
@@ -74,7 +74,7 @@ private Options() {
   @Endpoint(describeByClass = true)
   public static Join create(Scope scope, Iterable<Operand<TString>> inputs, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("StringJoin", scope.makeOpName("Join"));
-    opBuilder.addInputList(Operands.asOutputs(inputs));
+    opBuilder.addInputList(Operands.asOutputs(scope, inputs));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -100,7 +100,7 @@ public Output<TString> output() {
   }
   
   @Override
-  public Output<TString> asOutput() {
+  public Output<TString> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/Lower.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/Lower.java
index 0b214f47017..634fa32914e 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/Lower.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/Lower.java
@@ -69,7 +69,7 @@ private Options() {
   @Endpoint(describeByClass = true)
   public static Lower create(Scope scope, Operand<TString> input, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("StringLower", scope.makeOpName("Lower"));
-    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -95,7 +95,7 @@ public Output<TString> output() {
   }
   
   @Override
-  public Output<TString> asOutput() {
+  public Output<TString> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/ReduceJoin.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/ReduceJoin.java
index 23d90b3a2b1..b07f92d0ccc 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/ReduceJoin.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/ReduceJoin.java
@@ -100,8 +100,8 @@ private Options() {
   @Endpoint(describeByClass = true)
   public static ReduceJoin create(Scope scope, Operand<TString> inputs, Operand<TInt32> reductionIndices, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ReduceJoin", scope.makeOpName("ReduceJoin"));
-    opBuilder.addInput(inputs.asOutput());
-    opBuilder.addInput(reductionIndices.asOutput());
+    opBuilder.addInput(inputs.asOutput(scope));
+    opBuilder.addInput(reductionIndices.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -139,7 +139,7 @@ public Output<TString> output() {
   }
   
   @Override
-  public Output<TString> asOutput() {
+  public Output<TString> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/RegexFullMatch.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/RegexFullMatch.java
index 446166c6e29..a5462464baa 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/RegexFullMatch.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/RegexFullMatch.java
@@ -59,8 +59,8 @@ public final class RegexFullMatch extends RawOp implements Operand<TBool> {
   @Endpoint(describeByClass = true)
   public static RegexFullMatch create(Scope scope, Operand<TString> input, Operand<TString> pattern) {
     OperationBuilder opBuilder = scope.env().opBuilder("RegexFullMatch", scope.makeOpName("RegexFullMatch"));
-    opBuilder.addInput(input.asOutput());
-    opBuilder.addInput(pattern.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
+    opBuilder.addInput(pattern.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new RegexFullMatch(opBuilder.build());
   }
@@ -73,7 +73,7 @@ public Output<TBool> output() {
   }
   
   @Override
-  public Output<TBool> asOutput() {
+  public Output<TBool> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/RegexReplace.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/RegexReplace.java
index 3eaec903fee..475eaf4035c 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/RegexReplace.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/RegexReplace.java
@@ -71,9 +71,9 @@ private Options() {
   @Endpoint(describeByClass = true)
   public static RegexReplace create(Scope scope, Operand<TString> input, Operand<TString> pattern, Operand<TString> rewrite, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("RegexReplace", scope.makeOpName("RegexReplace"));
-    opBuilder.addInput(input.asOutput());
-    opBuilder.addInput(pattern.asOutput());
-    opBuilder.addInput(rewrite.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
+    opBuilder.addInput(pattern.asOutput(scope));
+    opBuilder.addInput(rewrite.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -102,7 +102,7 @@ public Output<TString> output() {
   }
   
   @Override
-  public Output<TString> asOutput() {
+  public Output<TString> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/StaticRegexFullMatch.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/StaticRegexFullMatch.java
index 3aa7b0827eb..db75da8afc5 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/StaticRegexFullMatch.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/StaticRegexFullMatch.java
@@ -51,7 +51,7 @@ public final class StaticRegexFullMatch extends RawOp implements Operand<TBool>
   @Endpoint(describeByClass = true)
   public static StaticRegexFullMatch create(Scope scope, Operand<TString> input, String pattern) {
     OperationBuilder opBuilder = scope.env().opBuilder("StaticRegexFullMatch", scope.makeOpName("StaticRegexFullMatch"));
-    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("pattern", pattern);
     return new StaticRegexFullMatch(opBuilder.build());
@@ -65,7 +65,7 @@ public Output<TBool> output() {
   }
   
   @Override
-  public Output<TBool> asOutput() {
+  public Output<TBool> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/StaticRegexReplace.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/StaticRegexReplace.java
index c1802f48001..ee7a1df0994 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/StaticRegexReplace.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/StaticRegexReplace.java
@@ -67,7 +67,7 @@ private Options() {
   @Endpoint(describeByClass = true)
   public static StaticRegexReplace create(Scope scope, Operand<TString> input, String pattern, String rewrite, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("StaticRegexReplace", scope.makeOpName("StaticRegexReplace"));
-    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("pattern", pattern);
     opBuilder.setAttr("rewrite", rewrite);
@@ -97,7 +97,7 @@ public Output<TString> output() {
   }
   
   @Override
-  public Output<TString> asOutput() {
+  public Output<TString> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/StringFormat.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/StringFormat.java
index 3decf1e5c93..706f9d769ea 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/StringFormat.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/StringFormat.java
@@ -84,7 +84,7 @@ private Options() {
   @Endpoint(describeByClass = true)
   public static StringFormat create(Scope scope, Iterable<Operand<?>> inputs, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("StringFormat", scope.makeOpName("StringFormat"));
-    opBuilder.addInputList(Operands.asOutputs(inputs));
+    opBuilder.addInputList(Operands.asOutputs(scope, inputs));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -131,7 +131,7 @@ public Output<TString> output() {
   }
   
   @Override
-  public Output<TString> asOutput() {
+  public Output<TString> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/StringLength.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/StringLength.java
index d9fc036ad22..c92ed01631b 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/StringLength.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/StringLength.java
@@ -77,7 +77,7 @@ private Options() {
   @Endpoint(describeByClass = true)
   public static StringLength create(Scope scope, Operand<TString> input, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("StringLength", scope.makeOpName("StringLength"));
-    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -109,7 +109,7 @@ public Output<TInt32> output() {
   }
   
   @Override
-  public Output<TInt32> asOutput() {
+  public Output<TInt32> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/StringNGrams.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/StringNGrams.java
index b420ae031f0..fe7b1463c46 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/StringNGrams.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/StringNGrams.java
@@ -22,7 +22,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -40,7 +39,7 @@
  * @param <T> data type for {@code ngramsSplits()} output
  */
 @Operator(group = "strings")
-public final class StringNGrams<T extends Tensor & TNumber> extends RawOp {
+public final class StringNGrams<T extends TNumber> extends RawOp {
   
   /**
    * Factory method to create a class wrapping a new StringNGrams operation.
@@ -63,10 +62,10 @@ public final class StringNGrams<T extends Tensor & TNumber> extends RawOp {
    * @return a new instance of StringNGrams
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> StringNGrams<T> create(Scope scope, Operand<TString> data, Operand<T> dataSplits, String separator, List<Long> ngramWidths, String leftPad, String rightPad, Long padWidth, Boolean preserveShortSequences) {
+  public static <T extends TNumber> StringNGrams<T> create(Scope scope, Operand<TString> data, Operand<T> dataSplits, String separator, List<Long> ngramWidths, String leftPad, String rightPad, Long padWidth, Boolean preserveShortSequences) {
     OperationBuilder opBuilder = scope.env().opBuilder("StringNGrams", scope.makeOpName("StringNGrams"));
-    opBuilder.addInput(data.asOutput());
-    opBuilder.addInput(dataSplits.asOutput());
+    opBuilder.addInput(data.asOutput(scope));
+    opBuilder.addInput(dataSplits.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("separator", separator);
     long[] ngramWidthsArray = new long[ngramWidths.size()];
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/StringSplit.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/StringSplit.java
index 87debf546cf..cc349f0aac3 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/StringSplit.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/StringSplit.java
@@ -89,8 +89,8 @@ private Options() {
   @Endpoint(describeByClass = true)
   public static StringSplit create(Scope scope, Operand<TString> input, Operand<TString> sep, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("StringSplitV2", scope.makeOpName("StringSplit"));
-    opBuilder.addInput(input.asOutput());
-    opBuilder.addInput(sep.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
+    opBuilder.addInput(sep.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/Strip.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/Strip.java
index fc6f37bbabb..0a6abf749c0 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/Strip.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/Strip.java
@@ -43,7 +43,7 @@ public final class Strip extends RawOp implements Operand<TString> {
   @Endpoint(describeByClass = true)
   public static Strip create(Scope scope, Operand<TString> input) {
     OperationBuilder opBuilder = scope.env().opBuilder("StringStrip", scope.makeOpName("Strip"));
-    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new Strip(opBuilder.build());
   }
@@ -61,7 +61,7 @@ public Output<TString> output() {
   }
   
   @Override
-  public Output<TString> asOutput() {
+  public Output<TString> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/Substr.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/Substr.java
index abb39bdd67a..ed30aef5aec 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/Substr.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/Substr.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -145,11 +144,11 @@ private Options() {
    * @return a new instance of Substr
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> Substr create(Scope scope, Operand<TString> input, Operand<T> pos, Operand<T> len, Options... options) {
+  public static <T extends TNumber> Substr create(Scope scope, Operand<TString> input, Operand<T> pos, Operand<T> len, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("Substr", scope.makeOpName("Substr"));
-    opBuilder.addInput(input.asOutput());
-    opBuilder.addInput(pos.asOutput());
-    opBuilder.addInput(len.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
+    opBuilder.addInput(pos.asOutput(scope));
+    opBuilder.addInput(len.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -180,7 +179,7 @@ public Output<TString> output() {
   }
   
   @Override
-  public Output<TString> asOutput() {
+  public Output<TString> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/ToHashBucket.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/ToHashBucket.java
index 11f025a4055..5b0d752274f 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/ToHashBucket.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/ToHashBucket.java
@@ -52,7 +52,7 @@ public final class ToHashBucket extends RawOp implements Operand<TInt64> {
   @Endpoint(describeByClass = true)
   public static ToHashBucket create(Scope scope, Operand<TString> stringTensor, Long numBuckets) {
     OperationBuilder opBuilder = scope.env().opBuilder("StringToHashBucket", scope.makeOpName("ToHashBucket"));
-    opBuilder.addInput(stringTensor.asOutput());
+    opBuilder.addInput(stringTensor.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("num_buckets", numBuckets);
     return new ToHashBucket(opBuilder.build());
@@ -66,7 +66,7 @@ public Output<TInt64> output() {
   }
   
   @Override
-  public Output<TInt64> asOutput() {
+  public Output<TInt64> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/ToHashBucketFast.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/ToHashBucketFast.java
index b1025fb9ccd..77085c6eca4 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/ToHashBucketFast.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/ToHashBucketFast.java
@@ -57,7 +57,7 @@ public final class ToHashBucketFast extends RawOp implements Operand<TInt64> {
   @Endpoint(describeByClass = true)
   public static ToHashBucketFast create(Scope scope, Operand<TString> input, Long numBuckets) {
     OperationBuilder opBuilder = scope.env().opBuilder("StringToHashBucketFast", scope.makeOpName("ToHashBucketFast"));
-    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("num_buckets", numBuckets);
     return new ToHashBucketFast(opBuilder.build());
@@ -71,7 +71,7 @@ public Output<TInt64> output() {
   }
   
   @Override
-  public Output<TInt64> asOutput() {
+  public Output<TInt64> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/ToHashBucketStrong.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/ToHashBucketStrong.java
index 28af44e377f..ce408cde7e2 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/ToHashBucketStrong.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/ToHashBucketStrong.java
@@ -67,7 +67,7 @@ public final class ToHashBucketStrong extends RawOp implements Operand<TInt64> {
   @Endpoint(describeByClass = true)
   public static ToHashBucketStrong create(Scope scope, Operand<TString> input, Long numBuckets, List<Long> key) {
     OperationBuilder opBuilder = scope.env().opBuilder("StringToHashBucketStrong", scope.makeOpName("ToHashBucketStrong"));
-    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("num_buckets", numBuckets);
     long[] keyArray = new long[key.size()];
@@ -86,7 +86,7 @@ public Output<TInt64> output() {
   }
   
   @Override
-  public Output<TInt64> asOutput() {
+  public Output<TInt64> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/ToNumber.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/ToNumber.java
index 8a6ae49eec3..604b8161cc6 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/ToNumber.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/ToNumber.java
@@ -22,7 +22,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -47,7 +46,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "strings")
-public final class ToNumber<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class ToNumber<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new ToNumber operation.
@@ -58,9 +57,9 @@ public final class ToNumber<T extends Tensor & TNumber> extends RawOp implements
    * @return a new instance of ToNumber
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> ToNumber<T> create(Scope scope, Operand<TString> stringTensor, DataType<T> outType) {
+  public static <T extends TNumber> ToNumber<T> create(Scope scope, Operand<TString> stringTensor, DataType<T> outType) {
     OperationBuilder opBuilder = scope.env().opBuilder("StringToNumber", scope.makeOpName("ToNumber"));
-    opBuilder.addInput(stringTensor.asOutput());
+    opBuilder.addInput(stringTensor.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("out_type", outType);
     return new ToNumber<T>(opBuilder.build());
@@ -86,7 +85,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/UnicodeDecode.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/UnicodeDecode.java
index 78e046da645..7e04b21ed62 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/UnicodeDecode.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/UnicodeDecode.java
@@ -22,7 +22,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -54,7 +53,7 @@
  * 
  * @param <T> data type for {@code rowSplits()} output
  */
-public final class UnicodeDecode<T extends Tensor & TNumber> extends RawOp {
+public final class UnicodeDecode<T extends TNumber> extends RawOp {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.strings.UnicodeDecode}
@@ -116,9 +115,9 @@ private Options() {
    * @return a new instance of UnicodeDecode
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> UnicodeDecode<T> create(Scope scope, Operand<TString> input, String inputEncoding, DataType<T> Tsplits, Options... options) {
+  public static <T extends TNumber> UnicodeDecode<T> create(Scope scope, Operand<TString> input, String inputEncoding, DataType<T> Tsplits, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("UnicodeDecode", scope.makeOpName("UnicodeDecode"));
-    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("input_encoding", inputEncoding);
     opBuilder.setAttr("Tsplits", Tsplits);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/UnicodeDecodeWithOffsets.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/UnicodeDecodeWithOffsets.java
index 8900040c0c3..987c0d6834c 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/UnicodeDecodeWithOffsets.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/UnicodeDecodeWithOffsets.java
@@ -22,7 +22,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -60,7 +59,7 @@
  * 
  * @param <T> data type for {@code rowSplits()} output
  */
-public final class UnicodeDecodeWithOffsets<T extends Tensor & TNumber> extends RawOp {
+public final class UnicodeDecodeWithOffsets<T extends TNumber> extends RawOp {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.strings.UnicodeDecodeWithOffsets}
@@ -122,9 +121,9 @@ private Options() {
    * @return a new instance of UnicodeDecodeWithOffsets
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> UnicodeDecodeWithOffsets<T> create(Scope scope, Operand<TString> input, String inputEncoding, DataType<T> Tsplits, Options... options) {
+  public static <T extends TNumber> UnicodeDecodeWithOffsets<T> create(Scope scope, Operand<TString> input, String inputEncoding, DataType<T> Tsplits, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("UnicodeDecodeWithOffsets", scope.makeOpName("UnicodeDecodeWithOffsets"));
-    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("input_encoding", inputEncoding);
     opBuilder.setAttr("Tsplits", Tsplits);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/UnicodeEncode.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/UnicodeEncode.java
index f6e29c08385..f0629e9e60a 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/UnicodeEncode.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/UnicodeEncode.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -102,10 +101,10 @@ private Options() {
    * @return a new instance of UnicodeEncode
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> UnicodeEncode create(Scope scope, Operand<TInt32> inputValues, Operand<T> inputSplits, String outputEncoding, Options... options) {
+  public static <T extends TNumber> UnicodeEncode create(Scope scope, Operand<TInt32> inputValues, Operand<T> inputSplits, String outputEncoding, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("UnicodeEncode", scope.makeOpName("UnicodeEncode"));
-    opBuilder.addInput(inputValues.asOutput());
-    opBuilder.addInput(inputSplits.asOutput());
+    opBuilder.addInput(inputValues.asOutput(scope));
+    opBuilder.addInput(inputSplits.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("output_encoding", outputEncoding);
     if (options != null) {
@@ -152,7 +151,7 @@ public Output<TString> output() {
   }
   
   @Override
-  public Output<TString> asOutput() {
+  public Output<TString> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/UnicodeScript.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/UnicodeScript.java
index 6d3600a1de3..39b43b34d45 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/UnicodeScript.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/UnicodeScript.java
@@ -54,7 +54,7 @@ public final class UnicodeScript extends RawOp implements Operand<TInt32> {
   @Endpoint(describeByClass = true)
   public static UnicodeScript create(Scope scope, Operand<TInt32> input) {
     OperationBuilder opBuilder = scope.env().opBuilder("UnicodeScript", scope.makeOpName("UnicodeScript"));
-    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new UnicodeScript(opBuilder.build());
   }
@@ -67,7 +67,7 @@ public Output<TInt32> output() {
   }
   
   @Override
-  public Output<TInt32> asOutput() {
+  public Output<TInt32> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/UnicodeTranscode.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/UnicodeTranscode.java
index 8f62db46152..8d666a585f1 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/UnicodeTranscode.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/UnicodeTranscode.java
@@ -135,7 +135,7 @@ private Options() {
   @Endpoint(describeByClass = true)
   public static UnicodeTranscode create(Scope scope, Operand<TString> input, String inputEncoding, String outputEncoding, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("UnicodeTranscode", scope.makeOpName("UnicodeTranscode"));
-    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("input_encoding", inputEncoding);
     opBuilder.setAttr("output_encoding", outputEncoding);
@@ -199,7 +199,7 @@ public Output<TString> output() {
   }
   
   @Override
-  public Output<TString> asOutput() {
+  public Output<TString> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/UnsortedSegmentJoin.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/UnsortedSegmentJoin.java
index 7ada46e3e7f..150511e769d 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/UnsortedSegmentJoin.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/UnsortedSegmentJoin.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -93,11 +92,11 @@ private Options() {
    * @return a new instance of UnsortedSegmentJoin
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber, U extends Tensor & TNumber> UnsortedSegmentJoin create(Scope scope, Operand<TString> inputs, Operand<T> segmentIds, Operand<U> numSegments, Options... options) {
+  public static <T extends TNumber, U extends TNumber> UnsortedSegmentJoin create(Scope scope, Operand<TString> inputs, Operand<T> segmentIds, Operand<U> numSegments, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("UnsortedSegmentJoin", scope.makeOpName("UnsortedSegmentJoin"));
-    opBuilder.addInput(inputs.asOutput());
-    opBuilder.addInput(segmentIds.asOutput());
-    opBuilder.addInput(numSegments.asOutput());
+    opBuilder.addInput(inputs.asOutput(scope));
+    opBuilder.addInput(segmentIds.asOutput(scope));
+    opBuilder.addInput(numSegments.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -123,7 +122,7 @@ public Output<TString> output() {
   }
   
   @Override
-  public Output<TString> asOutput() {
+  public Output<TString> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/Upper.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/Upper.java
index e90dc891dd8..44443dca328 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/Upper.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/Upper.java
@@ -69,7 +69,7 @@ private Options() {
   @Endpoint(describeByClass = true)
   public static Upper create(Scope scope, Operand<TString> input, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("StringUpper", scope.makeOpName("Upper"));
-    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -95,7 +95,7 @@ public Output<TString> output() {
   }
   
   @Override
-  public Output<TString> asOutput() {
+  public Output<TString> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/AudioSummary.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/AudioSummary.java
index 6743a4ef4c1..11d0a8cdf6a 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/AudioSummary.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/AudioSummary.java
@@ -81,9 +81,9 @@ private Options() {
   @Endpoint(describeByClass = true)
   public static AudioSummary create(Scope scope, Operand<TString> tag, Operand<TFloat32> tensor, Operand<TFloat32> sampleRate, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("AudioSummaryV2", scope.makeOpName("AudioSummary"));
-    opBuilder.addInput(tag.asOutput());
-    opBuilder.addInput(tensor.asOutput());
-    opBuilder.addInput(sampleRate.asOutput());
+    opBuilder.addInput(tag.asOutput(scope));
+    opBuilder.addInput(tensor.asOutput(scope));
+    opBuilder.addInput(sampleRate.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -110,7 +110,7 @@ public Output<TString> summary() {
   }
   
   @Override
-  public Output<TString> asOutput() {
+  public Output<TString> asOutput(Scope scope) {
     return summary;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/CloseSummaryWriter.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/CloseSummaryWriter.java
index f5d95d50976..defd036d9ea 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/CloseSummaryWriter.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/CloseSummaryWriter.java
@@ -39,7 +39,7 @@ public final class CloseSummaryWriter extends RawOp {
   @Endpoint(describeByClass = true)
   public static CloseSummaryWriter create(Scope scope, Operand<?> writer) {
     OperationBuilder opBuilder = scope.env().opBuilder("CloseSummaryWriter", scope.makeOpName("CloseSummaryWriter"));
-    opBuilder.addInput(writer.asOutput());
+    opBuilder.addInput(writer.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new CloseSummaryWriter(opBuilder.build());
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/CreateSummaryDbWriter.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/CreateSummaryDbWriter.java
index 8e40aa798d6..16dd23622d5 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/CreateSummaryDbWriter.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/CreateSummaryDbWriter.java
@@ -44,11 +44,11 @@ public final class CreateSummaryDbWriter extends RawOp {
   @Endpoint(describeByClass = true)
   public static CreateSummaryDbWriter create(Scope scope, Operand<?> writer, Operand<TString> dbUri, Operand<TString> experimentName, Operand<TString> runName, Operand<TString> userName) {
     OperationBuilder opBuilder = scope.env().opBuilder("CreateSummaryDbWriter", scope.makeOpName("CreateSummaryDbWriter"));
-    opBuilder.addInput(writer.asOutput());
-    opBuilder.addInput(dbUri.asOutput());
-    opBuilder.addInput(experimentName.asOutput());
-    opBuilder.addInput(runName.asOutput());
-    opBuilder.addInput(userName.asOutput());
+    opBuilder.addInput(writer.asOutput(scope));
+    opBuilder.addInput(dbUri.asOutput(scope));
+    opBuilder.addInput(experimentName.asOutput(scope));
+    opBuilder.addInput(runName.asOutput(scope));
+    opBuilder.addInput(userName.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new CreateSummaryDbWriter(opBuilder.build());
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/CreateSummaryFileWriter.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/CreateSummaryFileWriter.java
index e429fab20e2..a9fd623e23d 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/CreateSummaryFileWriter.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/CreateSummaryFileWriter.java
@@ -45,11 +45,11 @@ public final class CreateSummaryFileWriter extends RawOp {
   @Endpoint(describeByClass = true)
   public static CreateSummaryFileWriter create(Scope scope, Operand<?> writer, Operand<TString> logdir, Operand<TInt32> maxQueue, Operand<TInt32> flushMillis, Operand<TString> filenameSuffix) {
     OperationBuilder opBuilder = scope.env().opBuilder("CreateSummaryFileWriter", scope.makeOpName("CreateSummaryFileWriter"));
-    opBuilder.addInput(writer.asOutput());
-    opBuilder.addInput(logdir.asOutput());
-    opBuilder.addInput(maxQueue.asOutput());
-    opBuilder.addInput(flushMillis.asOutput());
-    opBuilder.addInput(filenameSuffix.asOutput());
+    opBuilder.addInput(writer.asOutput(scope));
+    opBuilder.addInput(logdir.asOutput(scope));
+    opBuilder.addInput(maxQueue.asOutput(scope));
+    opBuilder.addInput(flushMillis.asOutput(scope));
+    opBuilder.addInput(filenameSuffix.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new CreateSummaryFileWriter(opBuilder.build());
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/FlushSummaryWriter.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/FlushSummaryWriter.java
index e1586542972..05040764d4e 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/FlushSummaryWriter.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/FlushSummaryWriter.java
@@ -39,7 +39,7 @@ public final class FlushSummaryWriter extends RawOp {
   @Endpoint(describeByClass = true)
   public static FlushSummaryWriter create(Scope scope, Operand<?> writer) {
     OperationBuilder opBuilder = scope.env().opBuilder("FlushSummaryWriter", scope.makeOpName("FlushSummaryWriter"));
-    opBuilder.addInput(writer.asOutput());
+    opBuilder.addInput(writer.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new FlushSummaryWriter(opBuilder.build());
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/HistogramSummary.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/HistogramSummary.java
index f7ca3602773..271c2793c42 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/HistogramSummary.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/HistogramSummary.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -50,10 +49,10 @@ public final class HistogramSummary extends RawOp implements Operand<TString> {
    * @return a new instance of HistogramSummary
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> HistogramSummary create(Scope scope, Operand<TString> tag, Operand<T> values) {
+  public static <T extends TNumber> HistogramSummary create(Scope scope, Operand<TString> tag, Operand<T> values) {
     OperationBuilder opBuilder = scope.env().opBuilder("HistogramSummary", scope.makeOpName("HistogramSummary"));
-    opBuilder.addInput(tag.asOutput());
-    opBuilder.addInput(values.asOutput());
+    opBuilder.addInput(tag.asOutput(scope));
+    opBuilder.addInput(values.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new HistogramSummary(opBuilder.build());
   }
@@ -66,7 +65,7 @@ public Output<TString> summary() {
   }
   
   @Override
-  public Output<TString> asOutput() {
+  public Output<TString> asOutput(Scope scope) {
     return summary;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/ImageSummary.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/ImageSummary.java
index ac4702a0f6d..188074c9899 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/ImageSummary.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/ImageSummary.java
@@ -121,10 +121,10 @@ private Options() {
    * @return a new instance of ImageSummary
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> ImageSummary create(Scope scope, Operand<TString> tag, Operand<T> tensor, Options... options) {
+  public static <T extends TNumber> ImageSummary create(Scope scope, Operand<TString> tag, Operand<T> tensor, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ImageSummary", scope.makeOpName("ImageSummary"));
-    opBuilder.addInput(tag.asOutput());
-    opBuilder.addInput(tensor.asOutput());
+    opBuilder.addInput(tag.asOutput(scope));
+    opBuilder.addInput(tensor.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -161,7 +161,7 @@ public Output<TString> summary() {
   }
   
   @Override
-  public Output<TString> asOutput() {
+  public Output<TString> asOutput(Scope scope) {
     return summary;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/ImportEvent.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/ImportEvent.java
index 7bd97de571e..0a3276f246a 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/ImportEvent.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/ImportEvent.java
@@ -41,8 +41,8 @@ public final class ImportEvent extends RawOp {
   @Endpoint(describeByClass = true)
   public static ImportEvent create(Scope scope, Operand<?> writer, Operand<TString> event) {
     OperationBuilder opBuilder = scope.env().opBuilder("ImportEvent", scope.makeOpName("ImportEvent"));
-    opBuilder.addInput(writer.asOutput());
-    opBuilder.addInput(event.asOutput());
+    opBuilder.addInput(writer.asOutput(scope));
+    opBuilder.addInput(event.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new ImportEvent(opBuilder.build());
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/MergeSummary.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/MergeSummary.java
index e098bb8cf62..d561f574d8f 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/MergeSummary.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/MergeSummary.java
@@ -53,7 +53,7 @@ public final class MergeSummary extends RawOp implements Operand<TString> {
   @Endpoint(describeByClass = true)
   public static MergeSummary create(Scope scope, Iterable<Operand<TString>> inputs) {
     OperationBuilder opBuilder = scope.env().opBuilder("MergeSummary", scope.makeOpName("MergeSummary"));
-    opBuilder.addInputList(Operands.asOutputs(inputs));
+    opBuilder.addInputList(Operands.asOutputs(scope, inputs));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new MergeSummary(opBuilder.build());
   }
@@ -66,7 +66,7 @@ public Output<TString> summary() {
   }
   
   @Override
-  public Output<TString> asOutput() {
+  public Output<TString> asOutput(Scope scope) {
     return summary;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/ScalarSummary.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/ScalarSummary.java
index f2cb1135396..5e8f8ad82e2 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/ScalarSummary.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/ScalarSummary.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -47,10 +46,10 @@ public final class ScalarSummary extends RawOp implements Operand<TString> {
    * @return a new instance of ScalarSummary
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> ScalarSummary create(Scope scope, Operand<TString> tags, Operand<T> values) {
+  public static <T extends TNumber> ScalarSummary create(Scope scope, Operand<TString> tags, Operand<T> values) {
     OperationBuilder opBuilder = scope.env().opBuilder("ScalarSummary", scope.makeOpName("ScalarSummary"));
-    opBuilder.addInput(tags.asOutput());
-    opBuilder.addInput(values.asOutput());
+    opBuilder.addInput(tags.asOutput(scope));
+    opBuilder.addInput(values.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new ScalarSummary(opBuilder.build());
   }
@@ -63,7 +62,7 @@ public Output<TString> summary() {
   }
   
   @Override
-  public Output<TString> asOutput() {
+  public Output<TString> asOutput(Scope scope) {
     return summary;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/StatsAggregatorSummary.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/StatsAggregatorSummary.java
index 9cc5fe6d83b..576487c0972 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/StatsAggregatorSummary.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/StatsAggregatorSummary.java
@@ -42,7 +42,7 @@ public final class StatsAggregatorSummary extends RawOp implements Operand<TStri
   @Endpoint(describeByClass = true)
   public static StatsAggregatorSummary create(Scope scope, Operand<?> iterator) {
     OperationBuilder opBuilder = scope.env().opBuilder("StatsAggregatorSummary", scope.makeOpName("StatsAggregatorSummary"));
-    opBuilder.addInput(iterator.asOutput());
+    opBuilder.addInput(iterator.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new StatsAggregatorSummary(opBuilder.build());
   }
@@ -54,7 +54,7 @@ public Output<TString> summary() {
   }
   
   @Override
-  public Output<TString> asOutput() {
+  public Output<TString> asOutput(Scope scope) {
     return summary;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/SummaryWriter.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/SummaryWriter.java
index 3aff9146666..641683d7d38 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/SummaryWriter.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/SummaryWriter.java
@@ -21,15 +21,15 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  */
-public final class SummaryWriter extends RawOp implements Operand<Tensor> {
+public final class SummaryWriter extends RawOp implements Operand<TType> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.summary.SummaryWriter}
@@ -105,8 +105,8 @@ public Output<?> writer() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<Tensor> asOutput() {
-    return (Output<Tensor>) writer;
+  public Output<TType> asOutput(Scope scope) {
+    return (Output<TType>) writer;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/TensorSummary.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/TensorSummary.java
index 386b626778c..7bb21e42e5f 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/TensorSummary.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/TensorSummary.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TString;
+import org.tensorflow.types.family.TType;
 
 /**
  * Outputs a `Summary` protocol buffer with a tensor and per-plugin data.
@@ -45,11 +45,11 @@ public final class TensorSummary extends RawOp implements Operand<TString> {
    * @return a new instance of TensorSummary
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> TensorSummary create(Scope scope, Operand<TString> tag, Operand<T> tensor, Operand<TString> serializedSummaryMetadata) {
+  public static <T extends TType> TensorSummary create(Scope scope, Operand<TString> tag, Operand<T> tensor, Operand<TString> serializedSummaryMetadata) {
     OperationBuilder opBuilder = scope.env().opBuilder("TensorSummaryV2", scope.makeOpName("TensorSummary"));
-    opBuilder.addInput(tag.asOutput());
-    opBuilder.addInput(tensor.asOutput());
-    opBuilder.addInput(serializedSummaryMetadata.asOutput());
+    opBuilder.addInput(tag.asOutput(scope));
+    opBuilder.addInput(tensor.asOutput(scope));
+    opBuilder.addInput(serializedSummaryMetadata.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new TensorSummary(opBuilder.build());
   }
@@ -61,7 +61,7 @@ public Output<TString> summary() {
   }
   
   @Override
-  public Output<TString> asOutput() {
+  public Output<TString> asOutput(Scope scope) {
     return summary;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/WriteAudioSummary.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/WriteAudioSummary.java
index 8cdf23f4982..f2f444cf5c4 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/WriteAudioSummary.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/WriteAudioSummary.java
@@ -66,11 +66,11 @@ private Options() {
   @Endpoint(describeByClass = true)
   public static WriteAudioSummary create(Scope scope, Operand<?> writer, Operand<TInt64> step, Operand<TString> tag, Operand<TFloat32> tensor, Operand<TFloat32> sampleRate, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("WriteAudioSummary", scope.makeOpName("WriteAudioSummary"));
-    opBuilder.addInput(writer.asOutput());
-    opBuilder.addInput(step.asOutput());
-    opBuilder.addInput(tag.asOutput());
-    opBuilder.addInput(tensor.asOutput());
-    opBuilder.addInput(sampleRate.asOutput());
+    opBuilder.addInput(writer.asOutput(scope));
+    opBuilder.addInput(step.asOutput(scope));
+    opBuilder.addInput(tag.asOutput(scope));
+    opBuilder.addInput(tensor.asOutput(scope));
+    opBuilder.addInput(sampleRate.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/WriteGraphSummary.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/WriteGraphSummary.java
index dc0cbe0f222..36ccabbe82b 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/WriteGraphSummary.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/WriteGraphSummary.java
@@ -43,9 +43,9 @@ public final class WriteGraphSummary extends RawOp {
   @Endpoint(describeByClass = true)
   public static WriteGraphSummary create(Scope scope, Operand<?> writer, Operand<TInt64> step, Operand<TString> tensor) {
     OperationBuilder opBuilder = scope.env().opBuilder("WriteGraphSummary", scope.makeOpName("WriteGraphSummary"));
-    opBuilder.addInput(writer.asOutput());
-    opBuilder.addInput(step.asOutput());
-    opBuilder.addInput(tensor.asOutput());
+    opBuilder.addInput(writer.asOutput(scope));
+    opBuilder.addInput(step.asOutput(scope));
+    opBuilder.addInput(tensor.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new WriteGraphSummary(opBuilder.build());
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/WriteHistogramSummary.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/WriteHistogramSummary.java
index 59841f8cc7e..30bd22b8448 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/WriteHistogramSummary.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/WriteHistogramSummary.java
@@ -20,7 +20,6 @@
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -44,12 +43,12 @@ public final class WriteHistogramSummary extends RawOp {
    * @return a new instance of WriteHistogramSummary
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> WriteHistogramSummary create(Scope scope, Operand<?> writer, Operand<TInt64> step, Operand<TString> tag, Operand<T> values) {
+  public static <T extends TNumber> WriteHistogramSummary create(Scope scope, Operand<?> writer, Operand<TInt64> step, Operand<TString> tag, Operand<T> values) {
     OperationBuilder opBuilder = scope.env().opBuilder("WriteHistogramSummary", scope.makeOpName("WriteHistogramSummary"));
-    opBuilder.addInput(writer.asOutput());
-    opBuilder.addInput(step.asOutput());
-    opBuilder.addInput(tag.asOutput());
-    opBuilder.addInput(values.asOutput());
+    opBuilder.addInput(writer.asOutput(scope));
+    opBuilder.addInput(step.asOutput(scope));
+    opBuilder.addInput(tag.asOutput(scope));
+    opBuilder.addInput(values.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new WriteHistogramSummary(opBuilder.build());
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/WriteImageSummary.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/WriteImageSummary.java
index e66bafad757..80942e3b53f 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/WriteImageSummary.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/WriteImageSummary.java
@@ -20,7 +20,6 @@
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -66,13 +65,13 @@ private Options() {
    * @return a new instance of WriteImageSummary
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> WriteImageSummary create(Scope scope, Operand<?> writer, Operand<TInt64> step, Operand<TString> tag, Operand<T> tensor, Operand<TUint8> badColor, Options... options) {
+  public static <T extends TNumber> WriteImageSummary create(Scope scope, Operand<?> writer, Operand<TInt64> step, Operand<TString> tag, Operand<T> tensor, Operand<TUint8> badColor, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("WriteImageSummary", scope.makeOpName("WriteImageSummary"));
-    opBuilder.addInput(writer.asOutput());
-    opBuilder.addInput(step.asOutput());
-    opBuilder.addInput(tag.asOutput());
-    opBuilder.addInput(tensor.asOutput());
-    opBuilder.addInput(badColor.asOutput());
+    opBuilder.addInput(writer.asOutput(scope));
+    opBuilder.addInput(step.asOutput(scope));
+    opBuilder.addInput(tag.asOutput(scope));
+    opBuilder.addInput(tensor.asOutput(scope));
+    opBuilder.addInput(badColor.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/WriteRawProtoSummary.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/WriteRawProtoSummary.java
index 75499c1ff69..c1dd19a57f5 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/WriteRawProtoSummary.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/WriteRawProtoSummary.java
@@ -43,9 +43,9 @@ public final class WriteRawProtoSummary extends RawOp {
   @Endpoint(describeByClass = true)
   public static WriteRawProtoSummary create(Scope scope, Operand<?> writer, Operand<TInt64> step, Operand<TString> tensor) {
     OperationBuilder opBuilder = scope.env().opBuilder("WriteRawProtoSummary", scope.makeOpName("WriteRawProtoSummary"));
-    opBuilder.addInput(writer.asOutput());
-    opBuilder.addInput(step.asOutput());
-    opBuilder.addInput(tensor.asOutput());
+    opBuilder.addInput(writer.asOutput(scope));
+    opBuilder.addInput(step.asOutput(scope));
+    opBuilder.addInput(tensor.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new WriteRawProtoSummary(opBuilder.build());
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/WriteScalarSummary.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/WriteScalarSummary.java
index 63e61b54584..4878a15d68c 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/WriteScalarSummary.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/WriteScalarSummary.java
@@ -20,7 +20,6 @@
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -44,12 +43,12 @@ public final class WriteScalarSummary extends RawOp {
    * @return a new instance of WriteScalarSummary
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> WriteScalarSummary create(Scope scope, Operand<?> writer, Operand<TInt64> step, Operand<TString> tag, Operand<T> value) {
+  public static <T extends TNumber> WriteScalarSummary create(Scope scope, Operand<?> writer, Operand<TInt64> step, Operand<TString> tag, Operand<T> value) {
     OperationBuilder opBuilder = scope.env().opBuilder("WriteScalarSummary", scope.makeOpName("WriteScalarSummary"));
-    opBuilder.addInput(writer.asOutput());
-    opBuilder.addInput(step.asOutput());
-    opBuilder.addInput(tag.asOutput());
-    opBuilder.addInput(value.asOutput());
+    opBuilder.addInput(writer.asOutput(scope));
+    opBuilder.addInput(step.asOutput(scope));
+    opBuilder.addInput(tag.asOutput(scope));
+    opBuilder.addInput(value.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new WriteScalarSummary(opBuilder.build());
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/WriteSummary.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/WriteSummary.java
index eb431459f27..a5629306255 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/WriteSummary.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/WriteSummary.java
@@ -20,13 +20,13 @@
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
 import org.tensorflow.types.TString;
+import org.tensorflow.types.family.TType;
 
 /**
  */
@@ -44,13 +44,13 @@ public final class WriteSummary extends RawOp {
    * @return a new instance of WriteSummary
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> WriteSummary create(Scope scope, Operand<?> writer, Operand<TInt64> step, Operand<T> tensor, Operand<TString> tag, Operand<TString> summaryMetadata) {
+  public static <T extends TType> WriteSummary create(Scope scope, Operand<?> writer, Operand<TInt64> step, Operand<T> tensor, Operand<TString> tag, Operand<TString> summaryMetadata) {
     OperationBuilder opBuilder = scope.env().opBuilder("WriteSummary", scope.makeOpName("WriteSummary"));
-    opBuilder.addInput(writer.asOutput());
-    opBuilder.addInput(step.asOutput());
-    opBuilder.addInput(tensor.asOutput());
-    opBuilder.addInput(tag.asOutput());
-    opBuilder.addInput(summaryMetadata.asOutput());
+    opBuilder.addInput(writer.asOutput(scope));
+    opBuilder.addInput(step.asOutput(scope));
+    opBuilder.addInput(tensor.asOutput(scope));
+    opBuilder.addInput(tag.asOutput(scope));
+    opBuilder.addInput(summaryMetadata.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new WriteSummary(opBuilder.build());
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/AllToAll.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/AllToAll.java
index ca0c638cd10..71f38709d77 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/AllToAll.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/AllToAll.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TType;
 
 /**
  * An Op to exchange data across TPU replicas.
@@ -50,7 +50,7 @@
  * 
  * @param <T> data type for {@code output()} output
  */
-public final class AllToAll<T extends Tensor> extends RawOp implements Operand<T> {
+public final class AllToAll<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new AllToAll operation.
@@ -67,10 +67,10 @@ public final class AllToAll<T extends Tensor> extends RawOp implements Operand<T
    * @return a new instance of AllToAll
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> AllToAll<T> create(Scope scope, Operand<T> input, Operand<TInt32> groupAssignment, Long concatDimension, Long splitDimension, Long splitCount) {
+  public static <T extends TType> AllToAll<T> create(Scope scope, Operand<T> input, Operand<TInt32> groupAssignment, Long concatDimension, Long splitDimension, Long splitCount) {
     OperationBuilder opBuilder = scope.env().opBuilder("AllToAll", scope.makeOpName("AllToAll"));
-    opBuilder.addInput(input.asOutput());
-    opBuilder.addInput(groupAssignment.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
+    opBuilder.addInput(groupAssignment.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("concat_dimension", concatDimension);
     opBuilder.setAttr("split_dimension", splitDimension);
@@ -86,7 +86,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/CollectivePermute.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/CollectivePermute.java
index 6382cbfab4e..58a583fa5c8 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/CollectivePermute.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/CollectivePermute.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TType;
 
 /**
  * An Op to permute tensors across replicated TPU instances.
@@ -39,7 +39,7 @@
  * 
  * @param <T> data type for {@code output()} output
  */
-public final class CollectivePermute<T extends Tensor> extends RawOp implements Operand<T> {
+public final class CollectivePermute<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new CollectivePermute operation.
@@ -51,10 +51,10 @@ public final class CollectivePermute<T extends Tensor> extends RawOp implements
    * @return a new instance of CollectivePermute
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> CollectivePermute<T> create(Scope scope, Operand<T> input, Operand<TInt32> sourceTargetPairs) {
+  public static <T extends TType> CollectivePermute<T> create(Scope scope, Operand<T> input, Operand<TInt32> sourceTargetPairs) {
     OperationBuilder opBuilder = scope.env().opBuilder("CollectivePermute", scope.makeOpName("CollectivePermute"));
-    opBuilder.addInput(input.asOutput());
-    opBuilder.addInput(sourceTargetPairs.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
+    opBuilder.addInput(sourceTargetPairs.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new CollectivePermute<T>(opBuilder.build());
   }
@@ -67,7 +67,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/ConfigureDistributedTPU.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/ConfigureDistributedTPU.java
index bfdde5b98e2..dd01f7135a3 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/ConfigureDistributedTPU.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/ConfigureDistributedTPU.java
@@ -166,7 +166,7 @@ public Output<TString> topology() {
   }
   
   @Override
-  public Output<TString> asOutput() {
+  public Output<TString> asOutput(Scope scope) {
     return topology;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/CrossReplicaSum.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/CrossReplicaSum.java
index 7a1ec781270..1ddf061e4e9 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/CrossReplicaSum.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/CrossReplicaSum.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -41,7 +40,7 @@
  * 
  * @param <T> data type for {@code output()} output
  */
-public final class CrossReplicaSum<T extends Tensor & TNumber> extends RawOp implements Operand<T> {
+public final class CrossReplicaSum<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new CrossReplicaSum operation.
@@ -54,10 +53,10 @@ public final class CrossReplicaSum<T extends Tensor & TNumber> extends RawOp imp
    * @return a new instance of CrossReplicaSum
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber> CrossReplicaSum<T> create(Scope scope, Operand<T> input, Operand<TInt32> groupAssignment) {
+  public static <T extends TNumber> CrossReplicaSum<T> create(Scope scope, Operand<T> input, Operand<TInt32> groupAssignment) {
     OperationBuilder opBuilder = scope.env().opBuilder("CrossReplicaSum", scope.makeOpName("CrossReplicaSum"));
-    opBuilder.addInput(input.asOutput());
-    opBuilder.addInput(groupAssignment.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
+    opBuilder.addInput(groupAssignment.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new CrossReplicaSum<T>(opBuilder.build());
   }
@@ -70,7 +69,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/EnqueueTPUEmbeddingIntegerBatch.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/EnqueueTPUEmbeddingIntegerBatch.java
index 0a1a80c7a0a..e39002954f3 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/EnqueueTPUEmbeddingIntegerBatch.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/EnqueueTPUEmbeddingIntegerBatch.java
@@ -69,8 +69,8 @@ private Options() {
   @Endpoint(describeByClass = true)
   public static EnqueueTPUEmbeddingIntegerBatch create(Scope scope, Iterable<Operand<TInt32>> batch, Operand<TString> modeOverride, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("EnqueueTPUEmbeddingIntegerBatch", scope.makeOpName("EnqueueTPUEmbeddingIntegerBatch"));
-    opBuilder.addInputList(Operands.asOutputs(batch));
-    opBuilder.addInput(modeOverride.asOutput());
+    opBuilder.addInputList(Operands.asOutputs(scope, batch));
+    opBuilder.addInput(modeOverride.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/EnqueueTPUEmbeddingSparseBatch.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/EnqueueTPUEmbeddingSparseBatch.java
index 8e7f74431c2..a4607d1ae2a 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/EnqueueTPUEmbeddingSparseBatch.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/EnqueueTPUEmbeddingSparseBatch.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.Operands;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
@@ -99,12 +98,12 @@ private Options() {
    * @return a new instance of EnqueueTPUEmbeddingSparseBatch
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber, U extends Tensor & TNumber, V extends Tensor & TNumber> EnqueueTPUEmbeddingSparseBatch create(Scope scope, Iterable<Operand<T>> sampleIndices, Iterable<Operand<U>> embeddingIndices, Iterable<Operand<V>> aggregationWeights, Operand<TString> modeOverride, Options... options) {
+  public static <T extends TNumber, U extends TNumber, V extends TNumber> EnqueueTPUEmbeddingSparseBatch create(Scope scope, Iterable<Operand<T>> sampleIndices, Iterable<Operand<U>> embeddingIndices, Iterable<Operand<V>> aggregationWeights, Operand<TString> modeOverride, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("EnqueueTPUEmbeddingSparseBatch", scope.makeOpName("EnqueueTPUEmbeddingSparseBatch"));
-    opBuilder.addInputList(Operands.asOutputs(sampleIndices));
-    opBuilder.addInputList(Operands.asOutputs(embeddingIndices));
-    opBuilder.addInputList(Operands.asOutputs(aggregationWeights));
-    opBuilder.addInput(modeOverride.asOutput());
+    opBuilder.addInputList(Operands.asOutputs(scope, sampleIndices));
+    opBuilder.addInputList(Operands.asOutputs(scope, embeddingIndices));
+    opBuilder.addInputList(Operands.asOutputs(scope, aggregationWeights));
+    opBuilder.addInput(modeOverride.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/EnqueueTPUEmbeddingSparseTensorBatch.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/EnqueueTPUEmbeddingSparseTensorBatch.java
index b9b713fc110..cf0f2bf80f5 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/EnqueueTPUEmbeddingSparseTensorBatch.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/EnqueueTPUEmbeddingSparseTensorBatch.java
@@ -21,7 +21,6 @@
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.Operands;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
@@ -112,12 +111,12 @@ private Options() {
    * @return a new instance of EnqueueTPUEmbeddingSparseTensorBatch
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber, U extends Tensor & TNumber, V extends Tensor & TNumber> EnqueueTPUEmbeddingSparseTensorBatch create(Scope scope, Iterable<Operand<T>> sampleIndices, Iterable<Operand<U>> embeddingIndices, Iterable<Operand<V>> aggregationWeights, Operand<TString> modeOverride, List<Long> tableIds, Options... options) {
+  public static <T extends TNumber, U extends TNumber, V extends TNumber> EnqueueTPUEmbeddingSparseTensorBatch create(Scope scope, Iterable<Operand<T>> sampleIndices, Iterable<Operand<U>> embeddingIndices, Iterable<Operand<V>> aggregationWeights, Operand<TString> modeOverride, List<Long> tableIds, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("EnqueueTPUEmbeddingSparseTensorBatch", scope.makeOpName("EnqueueTPUEmbeddingSparseTensorBatch"));
-    opBuilder.addInputList(Operands.asOutputs(sampleIndices));
-    opBuilder.addInputList(Operands.asOutputs(embeddingIndices));
-    opBuilder.addInputList(Operands.asOutputs(aggregationWeights));
-    opBuilder.addInput(modeOverride.asOutput());
+    opBuilder.addInputList(Operands.asOutputs(scope, sampleIndices));
+    opBuilder.addInputList(Operands.asOutputs(scope, embeddingIndices));
+    opBuilder.addInputList(Operands.asOutputs(scope, aggregationWeights));
+    opBuilder.addInput(modeOverride.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     long[] tableIdsArray = new long[tableIds.size()];
     for (int i = 0; i < tableIdsArray.length; ++i) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/InfeedDequeue.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/InfeedDequeue.java
index e7008a92e99..db236d3ccca 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/InfeedDequeue.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/InfeedDequeue.java
@@ -22,19 +22,19 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * A placeholder op for a value that will be fed into the computation.
  * 
  * @param <T> data type for {@code output()} output
  */
-public final class InfeedDequeue<T extends Tensor> extends RawOp implements Operand<T> {
+public final class InfeedDequeue<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new InfeedDequeue operation.
@@ -45,7 +45,7 @@ public final class InfeedDequeue<T extends Tensor> extends RawOp implements Oper
    * @return a new instance of InfeedDequeue
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> InfeedDequeue<T> create(Scope scope, DataType<T> dtype, Shape shape) {
+  public static <T extends TType> InfeedDequeue<T> create(Scope scope, DataType<T> dtype, Shape shape) {
     OperationBuilder opBuilder = scope.env().opBuilder("InfeedDequeue", scope.makeOpName("InfeedDequeue"));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("dtype", dtype);
@@ -61,7 +61,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/InfeedDequeueTuple.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/InfeedDequeueTuple.java
index 2a2d5f7ed8c..f471da5b5d7 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/InfeedDequeueTuple.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/InfeedDequeueTuple.java
@@ -25,17 +25,17 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Fetches multiple values from infeed as an XLA tuple.
  */
-public final class InfeedDequeueTuple extends RawOp implements Iterable<Operand<Tensor>> {
+public final class InfeedDequeueTuple extends RawOp implements Iterable<Operand<TType>> {
   
   /**
    * Factory method to create a class wrapping a new InfeedDequeueTuple operation.
@@ -71,7 +71,7 @@ public List<Output<?>> outputs() {
   
   @Override
   @SuppressWarnings({"rawtypes", "unchecked"})
-  public Iterator<Operand<Tensor>> iterator() {
+  public Iterator<Operand<TType>> iterator() {
     return (Iterator) outputs.iterator();
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/InfeedEnqueue.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/InfeedEnqueue.java
index 4e816591eea..07c2452e051 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/InfeedEnqueue.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/InfeedEnqueue.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
-import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * An op which feeds a single Tensor value into the computation.
@@ -83,9 +83,9 @@ private Options() {
    * @return a new instance of InfeedEnqueue
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> InfeedEnqueue create(Scope scope, Operand<T> input, Options... options) {
+  public static <T extends TType> InfeedEnqueue create(Scope scope, Operand<T> input, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("InfeedEnqueue", scope.makeOpName("InfeedEnqueue"));
-    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/InfeedEnqueuePrelinearizedBuffer.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/InfeedEnqueuePrelinearizedBuffer.java
index 9344352791c..2d3b7277185 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/InfeedEnqueuePrelinearizedBuffer.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/InfeedEnqueuePrelinearizedBuffer.java
@@ -61,7 +61,7 @@ private Options() {
   @Endpoint(describeByClass = true)
   public static InfeedEnqueuePrelinearizedBuffer create(Scope scope, Operand<?> input, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("InfeedEnqueuePrelinearizedBuffer", scope.makeOpName("InfeedEnqueuePrelinearizedBuffer"));
-    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/InfeedEnqueueTuple.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/InfeedEnqueueTuple.java
index b439df84f71..a7a8d3e3be5 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/InfeedEnqueueTuple.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/InfeedEnqueueTuple.java
@@ -78,7 +78,7 @@ private Options() {
   @Endpoint(describeByClass = true)
   public static InfeedEnqueueTuple create(Scope scope, Iterable<Operand<?>> inputs, List<Shape> shapes, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("InfeedEnqueueTuple", scope.makeOpName("InfeedEnqueueTuple"));
-    opBuilder.addInputList(Operands.asOutputs(inputs));
+    opBuilder.addInputList(Operands.asOutputs(scope, inputs));
     opBuilder = scope.applyControlDependencies(opBuilder);
     Shape[] shapesArray = new Shape[shapes.size()];
     for (int i = 0; i < shapesArray.length; ++i) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingADAMParameters.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingADAMParameters.java
index 744688cee23..8b29e1022f1 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingADAMParameters.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingADAMParameters.java
@@ -89,9 +89,9 @@ private Options() {
   @Endpoint(describeByClass = true)
   public static LoadTPUEmbeddingADAMParameters create(Scope scope, Operand<TFloat32> parameters, Operand<TFloat32> momenta, Operand<TFloat32> velocities, Long numShards, Long shardId, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("LoadTPUEmbeddingADAMParameters", scope.makeOpName("LoadTPUEmbeddingADAMParameters"));
-    opBuilder.addInput(parameters.asOutput());
-    opBuilder.addInput(momenta.asOutput());
-    opBuilder.addInput(velocities.asOutput());
+    opBuilder.addInput(parameters.asOutput(scope));
+    opBuilder.addInput(momenta.asOutput(scope));
+    opBuilder.addInput(velocities.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("num_shards", numShards);
     opBuilder.setAttr("shard_id", shardId);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingADAMParametersGradAccumDebug.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingADAMParametersGradAccumDebug.java
index 63df2e6aa79..44117323874 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingADAMParametersGradAccumDebug.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingADAMParametersGradAccumDebug.java
@@ -90,10 +90,10 @@ private Options() {
   @Endpoint(describeByClass = true)
   public static LoadTPUEmbeddingADAMParametersGradAccumDebug create(Scope scope, Operand<TFloat32> parameters, Operand<TFloat32> momenta, Operand<TFloat32> velocities, Operand<TFloat32> gradientAccumulators, Long numShards, Long shardId, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("LoadTPUEmbeddingADAMParametersGradAccumDebug", scope.makeOpName("LoadTPUEmbeddingADAMParametersGradAccumDebug"));
-    opBuilder.addInput(parameters.asOutput());
-    opBuilder.addInput(momenta.asOutput());
-    opBuilder.addInput(velocities.asOutput());
-    opBuilder.addInput(gradientAccumulators.asOutput());
+    opBuilder.addInput(parameters.asOutput(scope));
+    opBuilder.addInput(momenta.asOutput(scope));
+    opBuilder.addInput(velocities.asOutput(scope));
+    opBuilder.addInput(gradientAccumulators.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("num_shards", numShards);
     opBuilder.setAttr("shard_id", shardId);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingAdadeltaParameters.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingAdadeltaParameters.java
index 43535a2aff8..43befd10f20 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingAdadeltaParameters.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingAdadeltaParameters.java
@@ -89,9 +89,9 @@ private Options() {
   @Endpoint(describeByClass = true)
   public static LoadTPUEmbeddingAdadeltaParameters create(Scope scope, Operand<TFloat32> parameters, Operand<TFloat32> accumulators, Operand<TFloat32> updates, Long numShards, Long shardId, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("LoadTPUEmbeddingAdadeltaParameters", scope.makeOpName("LoadTPUEmbeddingAdadeltaParameters"));
-    opBuilder.addInput(parameters.asOutput());
-    opBuilder.addInput(accumulators.asOutput());
-    opBuilder.addInput(updates.asOutput());
+    opBuilder.addInput(parameters.asOutput(scope));
+    opBuilder.addInput(accumulators.asOutput(scope));
+    opBuilder.addInput(updates.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("num_shards", numShards);
     opBuilder.setAttr("shard_id", shardId);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingAdadeltaParametersGradAccumDebug.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingAdadeltaParametersGradAccumDebug.java
index ce1b759ee60..3c9b92513da 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingAdadeltaParametersGradAccumDebug.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingAdadeltaParametersGradAccumDebug.java
@@ -90,10 +90,10 @@ private Options() {
   @Endpoint(describeByClass = true)
   public static LoadTPUEmbeddingAdadeltaParametersGradAccumDebug create(Scope scope, Operand<TFloat32> parameters, Operand<TFloat32> accumulators, Operand<TFloat32> updates, Operand<TFloat32> gradientAccumulators, Long numShards, Long shardId, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("LoadTPUEmbeddingAdadeltaParametersGradAccumDebug", scope.makeOpName("LoadTPUEmbeddingAdadeltaParametersGradAccumDebug"));
-    opBuilder.addInput(parameters.asOutput());
-    opBuilder.addInput(accumulators.asOutput());
-    opBuilder.addInput(updates.asOutput());
-    opBuilder.addInput(gradientAccumulators.asOutput());
+    opBuilder.addInput(parameters.asOutput(scope));
+    opBuilder.addInput(accumulators.asOutput(scope));
+    opBuilder.addInput(updates.asOutput(scope));
+    opBuilder.addInput(gradientAccumulators.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("num_shards", numShards);
     opBuilder.setAttr("shard_id", shardId);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingAdagradParameters.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingAdagradParameters.java
index f9e16c5b5d6..9695b8815e5 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingAdagradParameters.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingAdagradParameters.java
@@ -88,8 +88,8 @@ private Options() {
   @Endpoint(describeByClass = true)
   public static LoadTPUEmbeddingAdagradParameters create(Scope scope, Operand<TFloat32> parameters, Operand<TFloat32> accumulators, Long numShards, Long shardId, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("LoadTPUEmbeddingAdagradParameters", scope.makeOpName("LoadTPUEmbeddingAdagradParameters"));
-    opBuilder.addInput(parameters.asOutput());
-    opBuilder.addInput(accumulators.asOutput());
+    opBuilder.addInput(parameters.asOutput(scope));
+    opBuilder.addInput(accumulators.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("num_shards", numShards);
     opBuilder.setAttr("shard_id", shardId);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingAdagradParametersGradAccumDebug.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingAdagradParametersGradAccumDebug.java
index 7f8df653745..edc56619b76 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingAdagradParametersGradAccumDebug.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingAdagradParametersGradAccumDebug.java
@@ -89,9 +89,9 @@ private Options() {
   @Endpoint(describeByClass = true)
   public static LoadTPUEmbeddingAdagradParametersGradAccumDebug create(Scope scope, Operand<TFloat32> parameters, Operand<TFloat32> accumulators, Operand<TFloat32> gradientAccumulators, Long numShards, Long shardId, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("LoadTPUEmbeddingAdagradParametersGradAccumDebug", scope.makeOpName("LoadTPUEmbeddingAdagradParametersGradAccumDebug"));
-    opBuilder.addInput(parameters.asOutput());
-    opBuilder.addInput(accumulators.asOutput());
-    opBuilder.addInput(gradientAccumulators.asOutput());
+    opBuilder.addInput(parameters.asOutput(scope));
+    opBuilder.addInput(accumulators.asOutput(scope));
+    opBuilder.addInput(gradientAccumulators.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("num_shards", numShards);
     opBuilder.setAttr("shard_id", shardId);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingCenteredRMSPropParameters.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingCenteredRMSPropParameters.java
index f0b704cfaa1..d727ad24e1f 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingCenteredRMSPropParameters.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingCenteredRMSPropParameters.java
@@ -90,10 +90,10 @@ private Options() {
   @Endpoint(describeByClass = true)
   public static LoadTPUEmbeddingCenteredRMSPropParameters create(Scope scope, Operand<TFloat32> parameters, Operand<TFloat32> ms, Operand<TFloat32> mom, Operand<TFloat32> mg, Long numShards, Long shardId, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("LoadTPUEmbeddingCenteredRMSPropParameters", scope.makeOpName("LoadTPUEmbeddingCenteredRMSPropParameters"));
-    opBuilder.addInput(parameters.asOutput());
-    opBuilder.addInput(ms.asOutput());
-    opBuilder.addInput(mom.asOutput());
-    opBuilder.addInput(mg.asOutput());
+    opBuilder.addInput(parameters.asOutput(scope));
+    opBuilder.addInput(ms.asOutput(scope));
+    opBuilder.addInput(mom.asOutput(scope));
+    opBuilder.addInput(mg.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("num_shards", numShards);
     opBuilder.setAttr("shard_id", shardId);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingFTRLParameters.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingFTRLParameters.java
index c96edf58894..242282fdd7a 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingFTRLParameters.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingFTRLParameters.java
@@ -89,9 +89,9 @@ private Options() {
   @Endpoint(describeByClass = true)
   public static LoadTPUEmbeddingFTRLParameters create(Scope scope, Operand<TFloat32> parameters, Operand<TFloat32> accumulators, Operand<TFloat32> linears, Long numShards, Long shardId, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("LoadTPUEmbeddingFTRLParameters", scope.makeOpName("LoadTPUEmbeddingFTRLParameters"));
-    opBuilder.addInput(parameters.asOutput());
-    opBuilder.addInput(accumulators.asOutput());
-    opBuilder.addInput(linears.asOutput());
+    opBuilder.addInput(parameters.asOutput(scope));
+    opBuilder.addInput(accumulators.asOutput(scope));
+    opBuilder.addInput(linears.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("num_shards", numShards);
     opBuilder.setAttr("shard_id", shardId);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingFTRLParametersGradAccumDebug.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingFTRLParametersGradAccumDebug.java
index f0a85bd945a..f5fe30a6c27 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingFTRLParametersGradAccumDebug.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingFTRLParametersGradAccumDebug.java
@@ -90,10 +90,10 @@ private Options() {
   @Endpoint(describeByClass = true)
   public static LoadTPUEmbeddingFTRLParametersGradAccumDebug create(Scope scope, Operand<TFloat32> parameters, Operand<TFloat32> accumulators, Operand<TFloat32> linears, Operand<TFloat32> gradientAccumulators, Long numShards, Long shardId, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("LoadTPUEmbeddingFTRLParametersGradAccumDebug", scope.makeOpName("LoadTPUEmbeddingFTRLParametersGradAccumDebug"));
-    opBuilder.addInput(parameters.asOutput());
-    opBuilder.addInput(accumulators.asOutput());
-    opBuilder.addInput(linears.asOutput());
-    opBuilder.addInput(gradientAccumulators.asOutput());
+    opBuilder.addInput(parameters.asOutput(scope));
+    opBuilder.addInput(accumulators.asOutput(scope));
+    opBuilder.addInput(linears.asOutput(scope));
+    opBuilder.addInput(gradientAccumulators.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("num_shards", numShards);
     opBuilder.setAttr("shard_id", shardId);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingMDLAdagradLightParameters.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingMDLAdagradLightParameters.java
index f418a70cc8c..28983395b7c 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingMDLAdagradLightParameters.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingMDLAdagradLightParameters.java
@@ -90,10 +90,10 @@ private Options() {
   @Endpoint(describeByClass = true)
   public static LoadTPUEmbeddingMDLAdagradLightParameters create(Scope scope, Operand<TFloat32> parameters, Operand<TFloat32> accumulators, Operand<TFloat32> weights, Operand<TFloat32> benefits, Long numShards, Long shardId, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("LoadTPUEmbeddingMDLAdagradLightParameters", scope.makeOpName("LoadTPUEmbeddingMDLAdagradLightParameters"));
-    opBuilder.addInput(parameters.asOutput());
-    opBuilder.addInput(accumulators.asOutput());
-    opBuilder.addInput(weights.asOutput());
-    opBuilder.addInput(benefits.asOutput());
+    opBuilder.addInput(parameters.asOutput(scope));
+    opBuilder.addInput(accumulators.asOutput(scope));
+    opBuilder.addInput(weights.asOutput(scope));
+    opBuilder.addInput(benefits.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("num_shards", numShards);
     opBuilder.setAttr("shard_id", shardId);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingMomentumParameters.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingMomentumParameters.java
index 718bdc24f5c..27341769cf3 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingMomentumParameters.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingMomentumParameters.java
@@ -88,8 +88,8 @@ private Options() {
   @Endpoint(describeByClass = true)
   public static LoadTPUEmbeddingMomentumParameters create(Scope scope, Operand<TFloat32> parameters, Operand<TFloat32> momenta, Long numShards, Long shardId, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("LoadTPUEmbeddingMomentumParameters", scope.makeOpName("LoadTPUEmbeddingMomentumParameters"));
-    opBuilder.addInput(parameters.asOutput());
-    opBuilder.addInput(momenta.asOutput());
+    opBuilder.addInput(parameters.asOutput(scope));
+    opBuilder.addInput(momenta.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("num_shards", numShards);
     opBuilder.setAttr("shard_id", shardId);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingMomentumParametersGradAccumDebug.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingMomentumParametersGradAccumDebug.java
index 424c3c846c0..6da5dc0fe72 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingMomentumParametersGradAccumDebug.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingMomentumParametersGradAccumDebug.java
@@ -89,9 +89,9 @@ private Options() {
   @Endpoint(describeByClass = true)
   public static LoadTPUEmbeddingMomentumParametersGradAccumDebug create(Scope scope, Operand<TFloat32> parameters, Operand<TFloat32> momenta, Operand<TFloat32> gradientAccumulators, Long numShards, Long shardId, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("LoadTPUEmbeddingMomentumParametersGradAccumDebug", scope.makeOpName("LoadTPUEmbeddingMomentumParametersGradAccumDebug"));
-    opBuilder.addInput(parameters.asOutput());
-    opBuilder.addInput(momenta.asOutput());
-    opBuilder.addInput(gradientAccumulators.asOutput());
+    opBuilder.addInput(parameters.asOutput(scope));
+    opBuilder.addInput(momenta.asOutput(scope));
+    opBuilder.addInput(gradientAccumulators.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("num_shards", numShards);
     opBuilder.setAttr("shard_id", shardId);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingProximalAdagradParameters.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingProximalAdagradParameters.java
index 7b7265e9b82..9b332186332 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingProximalAdagradParameters.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingProximalAdagradParameters.java
@@ -88,8 +88,8 @@ private Options() {
   @Endpoint(describeByClass = true)
   public static LoadTPUEmbeddingProximalAdagradParameters create(Scope scope, Operand<TFloat32> parameters, Operand<TFloat32> accumulators, Long numShards, Long shardId, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("LoadTPUEmbeddingProximalAdagradParameters", scope.makeOpName("LoadTPUEmbeddingProximalAdagradParameters"));
-    opBuilder.addInput(parameters.asOutput());
-    opBuilder.addInput(accumulators.asOutput());
+    opBuilder.addInput(parameters.asOutput(scope));
+    opBuilder.addInput(accumulators.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("num_shards", numShards);
     opBuilder.setAttr("shard_id", shardId);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingProximalAdagradParametersGradAccumDebug.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingProximalAdagradParametersGradAccumDebug.java
index c18d2cf22f5..76c8193545b 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingProximalAdagradParametersGradAccumDebug.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingProximalAdagradParametersGradAccumDebug.java
@@ -89,9 +89,9 @@ private Options() {
   @Endpoint(describeByClass = true)
   public static LoadTPUEmbeddingProximalAdagradParametersGradAccumDebug create(Scope scope, Operand<TFloat32> parameters, Operand<TFloat32> accumulators, Operand<TFloat32> gradientAccumulators, Long numShards, Long shardId, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("LoadTPUEmbeddingProximalAdagradParametersGradAccumDebug", scope.makeOpName("LoadTPUEmbeddingProximalAdagradParametersGradAccumDebug"));
-    opBuilder.addInput(parameters.asOutput());
-    opBuilder.addInput(accumulators.asOutput());
-    opBuilder.addInput(gradientAccumulators.asOutput());
+    opBuilder.addInput(parameters.asOutput(scope));
+    opBuilder.addInput(accumulators.asOutput(scope));
+    opBuilder.addInput(gradientAccumulators.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("num_shards", numShards);
     opBuilder.setAttr("shard_id", shardId);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingRMSPropParameters.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingRMSPropParameters.java
index 1f747282d55..83753ee8b88 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingRMSPropParameters.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingRMSPropParameters.java
@@ -89,9 +89,9 @@ private Options() {
   @Endpoint(describeByClass = true)
   public static LoadTPUEmbeddingRMSPropParameters create(Scope scope, Operand<TFloat32> parameters, Operand<TFloat32> ms, Operand<TFloat32> mom, Long numShards, Long shardId, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("LoadTPUEmbeddingRMSPropParameters", scope.makeOpName("LoadTPUEmbeddingRMSPropParameters"));
-    opBuilder.addInput(parameters.asOutput());
-    opBuilder.addInput(ms.asOutput());
-    opBuilder.addInput(mom.asOutput());
+    opBuilder.addInput(parameters.asOutput(scope));
+    opBuilder.addInput(ms.asOutput(scope));
+    opBuilder.addInput(mom.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("num_shards", numShards);
     opBuilder.setAttr("shard_id", shardId);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingRMSPropParametersGradAccumDebug.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingRMSPropParametersGradAccumDebug.java
index a7c8ed4812c..f562849e113 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingRMSPropParametersGradAccumDebug.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingRMSPropParametersGradAccumDebug.java
@@ -90,10 +90,10 @@ private Options() {
   @Endpoint(describeByClass = true)
   public static LoadTPUEmbeddingRMSPropParametersGradAccumDebug create(Scope scope, Operand<TFloat32> parameters, Operand<TFloat32> ms, Operand<TFloat32> mom, Operand<TFloat32> gradientAccumulators, Long numShards, Long shardId, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("LoadTPUEmbeddingRMSPropParametersGradAccumDebug", scope.makeOpName("LoadTPUEmbeddingRMSPropParametersGradAccumDebug"));
-    opBuilder.addInput(parameters.asOutput());
-    opBuilder.addInput(ms.asOutput());
-    opBuilder.addInput(mom.asOutput());
-    opBuilder.addInput(gradientAccumulators.asOutput());
+    opBuilder.addInput(parameters.asOutput(scope));
+    opBuilder.addInput(ms.asOutput(scope));
+    opBuilder.addInput(mom.asOutput(scope));
+    opBuilder.addInput(gradientAccumulators.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("num_shards", numShards);
     opBuilder.setAttr("shard_id", shardId);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingStochasticGradientDescentParameters.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingStochasticGradientDescentParameters.java
index 769d3436eda..5c3fb5d662d 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingStochasticGradientDescentParameters.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingStochasticGradientDescentParameters.java
@@ -87,7 +87,7 @@ private Options() {
   @Endpoint(describeByClass = true)
   public static LoadTPUEmbeddingStochasticGradientDescentParameters create(Scope scope, Operand<TFloat32> parameters, Long numShards, Long shardId, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("LoadTPUEmbeddingStochasticGradientDescentParameters", scope.makeOpName("LoadTPUEmbeddingStochasticGradientDescentParameters"));
-    opBuilder.addInput(parameters.asOutput());
+    opBuilder.addInput(parameters.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("num_shards", numShards);
     opBuilder.setAttr("shard_id", shardId);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/OutfeedDequeue.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/OutfeedDequeue.java
index 645ff872328..fc4e30162d4 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/OutfeedDequeue.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/OutfeedDequeue.java
@@ -22,12 +22,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Retrieves a single tensor from the computation outfeed.
@@ -36,7 +36,7 @@
  * 
  * @param <T> data type for {@code output()} output
  */
-public final class OutfeedDequeue<T extends Tensor> extends RawOp implements Operand<T> {
+public final class OutfeedDequeue<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.tpu.OutfeedDequeue}
@@ -69,7 +69,7 @@ private Options() {
    * @return a new instance of OutfeedDequeue
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> OutfeedDequeue<T> create(Scope scope, DataType<T> dtype, Shape shape, Options... options) {
+  public static <T extends TType> OutfeedDequeue<T> create(Scope scope, DataType<T> dtype, Shape shape, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("OutfeedDequeue", scope.makeOpName("OutfeedDequeue"));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("dtype", dtype);
@@ -101,7 +101,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/OutfeedDequeueTuple.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/OutfeedDequeueTuple.java
index fe6bfa90ea4..6b9110232d8 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/OutfeedDequeueTuple.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/OutfeedDequeueTuple.java
@@ -25,12 +25,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Retrieve multiple values from the computation outfeed.
@@ -38,7 +38,7 @@
  * This operation will block indefinitely until data is available. Output `i`
  * corresponds to XLA tuple element `i`.
  */
-public final class OutfeedDequeueTuple extends RawOp implements Iterable<Operand<Tensor>> {
+public final class OutfeedDequeueTuple extends RawOp implements Iterable<Operand<TType>> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.tpu.OutfeedDequeueTuple}
@@ -112,7 +112,7 @@ public List<Output<?>> outputs() {
   
   @Override
   @SuppressWarnings({"rawtypes", "unchecked"})
-  public Iterator<Operand<Tensor>> iterator() {
+  public Iterator<Operand<TType>> iterator() {
     return (Iterator) outputs.iterator();
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/OutfeedEnqueue.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/OutfeedEnqueue.java
index 4192f8b8dbb..2cae7229fde 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/OutfeedEnqueue.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/OutfeedEnqueue.java
@@ -20,11 +20,11 @@
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Enqueue a Tensor on the computation outfeed.
@@ -39,9 +39,9 @@ public final class OutfeedEnqueue extends RawOp {
    * @return a new instance of OutfeedEnqueue
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> OutfeedEnqueue create(Scope scope, Operand<T> input) {
+  public static <T extends TType> OutfeedEnqueue create(Scope scope, Operand<T> input) {
     OperationBuilder opBuilder = scope.env().opBuilder("OutfeedEnqueue", scope.makeOpName("OutfeedEnqueue"));
-    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new OutfeedEnqueue(opBuilder.build());
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/OutfeedEnqueueTuple.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/OutfeedEnqueueTuple.java
index 8bfd04b9a2c..761b7f856ac 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/OutfeedEnqueueTuple.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/OutfeedEnqueueTuple.java
@@ -42,7 +42,7 @@ public final class OutfeedEnqueueTuple extends RawOp {
   @Endpoint(describeByClass = true)
   public static OutfeedEnqueueTuple create(Scope scope, Iterable<Operand<?>> inputs) {
     OperationBuilder opBuilder = scope.env().opBuilder("OutfeedEnqueueTuple", scope.makeOpName("OutfeedEnqueueTuple"));
-    opBuilder.addInputList(Operands.asOutputs(inputs));
+    opBuilder.addInputList(Operands.asOutputs(scope, inputs));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new OutfeedEnqueueTuple(opBuilder.build());
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/Prelinearize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/Prelinearize.java
index 27bb0405fa1..42eb810042a 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/Prelinearize.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/Prelinearize.java
@@ -22,17 +22,17 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * An op which linearizes one Tensor value to an opaque variant tensor.
  */
-public final class Prelinearize extends RawOp implements Operand<Tensor> {
+public final class Prelinearize extends RawOp implements Operand<TType> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.tpu.Prelinearize}
@@ -73,9 +73,9 @@ private Options() {
    * @return a new instance of Prelinearize
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> Prelinearize create(Scope scope, Operand<T> input, Options... options) {
+  public static <T extends TType> Prelinearize create(Scope scope, Operand<T> input, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("Prelinearize", scope.makeOpName("Prelinearize"));
-    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -118,8 +118,8 @@ public Output<?> output() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<Tensor> asOutput() {
-    return (Output<Tensor>) output;
+  public Output<TType> asOutput(Scope scope) {
+    return (Output<TType>) output;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/PrelinearizeTuple.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/PrelinearizeTuple.java
index 0b7b48778b9..427efc433bb 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/PrelinearizeTuple.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/PrelinearizeTuple.java
@@ -22,18 +22,18 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.Operands;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * An op which linearizes multiple Tensor values to an opaque variant tensor.
  */
-public final class PrelinearizeTuple extends RawOp implements Operand<Tensor> {
+public final class PrelinearizeTuple extends RawOp implements Operand<TType> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.tpu.PrelinearizeTuple}
@@ -69,7 +69,7 @@ private Options() {
   @Endpoint(describeByClass = true)
   public static PrelinearizeTuple create(Scope scope, Iterable<Operand<?>> inputs, List<Shape> shapes, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("PrelinearizeTuple", scope.makeOpName("PrelinearizeTuple"));
-    opBuilder.addInputList(Operands.asOutputs(inputs));
+    opBuilder.addInputList(Operands.asOutputs(scope, inputs));
     opBuilder = scope.applyControlDependencies(opBuilder);
     Shape[] shapesArray = new Shape[shapes.size()];
     for (int i = 0; i < shapesArray.length; ++i) {
@@ -108,8 +108,8 @@ public Output<?> output() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<Tensor> asOutput() {
-    return (Output<Tensor>) output;
+  public Output<TType> asOutput(Scope scope) {
+    return (Output<TType>) output;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingStochasticGradientDescentParameters.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingStochasticGradientDescentParameters.java
index e69247fa0ce..4af3eff2fa1 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingStochasticGradientDescentParameters.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingStochasticGradientDescentParameters.java
@@ -134,7 +134,7 @@ public Output<TFloat32> parameters() {
   }
   
   @Override
-  public Output<TFloat32> asOutput() {
+  public Output<TFloat32> asOutput(Scope scope) {
     return parameters;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/SendTPUEmbeddingGradients.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/SendTPUEmbeddingGradients.java
index 482080bde5d..2962d970bc3 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/SendTPUEmbeddingGradients.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/SendTPUEmbeddingGradients.java
@@ -54,8 +54,8 @@ public final class SendTPUEmbeddingGradients extends RawOp {
   @Endpoint(describeByClass = true)
   public static SendTPUEmbeddingGradients create(Scope scope, Iterable<Operand<TFloat32>> inputs, Iterable<Operand<TFloat32>> learningRates, String config) {
     OperationBuilder opBuilder = scope.env().opBuilder("SendTPUEmbeddingGradients", scope.makeOpName("SendTPUEmbeddingGradients"));
-    opBuilder.addInputList(Operands.asOutputs(inputs));
-    opBuilder.addInputList(Operands.asOutputs(learningRates));
+    opBuilder.addInputList(Operands.asOutputs(scope, inputs));
+    opBuilder.addInputList(Operands.asOutputs(scope, learningRates));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("config", config);
     return new SendTPUEmbeddingGradients(opBuilder.build());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/TPUCompilationResult.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/TPUCompilationResult.java
index 45280fbc0aa..31423e6d6c7 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/TPUCompilationResult.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/TPUCompilationResult.java
@@ -56,7 +56,7 @@ public Output<TString> output() {
   }
   
   @Override
-  public Output<TString> asOutput() {
+  public Output<TString> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/TPUEmbeddingActivations.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/TPUEmbeddingActivations.java
index 138a7434106..f215a06df79 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/TPUEmbeddingActivations.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/TPUEmbeddingActivations.java
@@ -53,8 +53,8 @@ public final class TPUEmbeddingActivations extends RawOp implements Operand<TFlo
   @Endpoint(describeByClass = true)
   public static TPUEmbeddingActivations create(Scope scope, Operand<TFloat32> embeddingVariable, Operand<TFloat32> slicedActivations, Long tableId, Long lookupId) {
     OperationBuilder opBuilder = scope.env().opBuilder("TPUEmbeddingActivations", scope.makeOpName("TPUEmbeddingActivations"));
-    opBuilder.addInput(embeddingVariable.asOutput());
-    opBuilder.addInput(slicedActivations.asOutput());
+    opBuilder.addInput(embeddingVariable.asOutput(scope));
+    opBuilder.addInput(slicedActivations.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("table_id", tableId);
     opBuilder.setAttr("lookup_id", lookupId);
@@ -68,7 +68,7 @@ public Output<TFloat32> output() {
   }
   
   @Override
-  public Output<TFloat32> asOutput() {
+  public Output<TFloat32> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/TPUOrdinalSelector.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/TPUOrdinalSelector.java
index 2df183cb793..64ff5519c73 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/TPUOrdinalSelector.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/TPUOrdinalSelector.java
@@ -57,7 +57,7 @@ public Output<TInt32> deviceOrdinals() {
   }
   
   @Override
-  public Output<TInt32> asOutput() {
+  public Output<TInt32> asOutput(Scope scope) {
     return deviceOrdinals;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/TPUReplicatedInput.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/TPUReplicatedInput.java
index 58959270d9b..597865ffa6d 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/TPUReplicatedInput.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/TPUReplicatedInput.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.Operands;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Connects N inputs to an N-way replicated TPU computation.
@@ -45,7 +45,7 @@
  * 
  * @param <T> data type for {@code output()} output
  */
-public final class TPUReplicatedInput<T extends Tensor> extends RawOp implements Operand<T> {
+public final class TPUReplicatedInput<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.tpu.TPUReplicatedInput}
@@ -93,9 +93,9 @@ private Options() {
    * @return a new instance of TPUReplicatedInput
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> TPUReplicatedInput<T> create(Scope scope, Iterable<Operand<T>> inputs, Options... options) {
+  public static <T extends TType> TPUReplicatedInput<T> create(Scope scope, Iterable<Operand<T>> inputs, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("TPUReplicatedInput", scope.makeOpName("TPUReplicatedInput"));
-    opBuilder.addInputList(Operands.asOutputs(inputs));
+    opBuilder.addInputList(Operands.asOutputs(scope, inputs));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -141,7 +141,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/TPUReplicatedOutput.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/TPUReplicatedOutput.java
index 36c52cbb6b4..d7d2487b90c 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/TPUReplicatedOutput.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/TPUReplicatedOutput.java
@@ -24,11 +24,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Connects N outputs from an N-way replicated TPU computation.
@@ -45,7 +45,7 @@
  * 
  * @param <T> data type for {@code outputs()} output
  */
-public final class TPUReplicatedOutput<T extends Tensor> extends RawOp implements Iterable<Operand<T>> {
+public final class TPUReplicatedOutput<T extends TType> extends RawOp implements Iterable<Operand<T>> {
   
   /**
    * Factory method to create a class wrapping a new TPUReplicatedOutput operation.
@@ -56,9 +56,9 @@ public final class TPUReplicatedOutput<T extends Tensor> extends RawOp implement
    * @return a new instance of TPUReplicatedOutput
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> TPUReplicatedOutput<T> create(Scope scope, Operand<T> input, Long numReplicas) {
+  public static <T extends TType> TPUReplicatedOutput<T> create(Scope scope, Operand<T> input, Long numReplicas) {
     OperationBuilder opBuilder = scope.env().opBuilder("TPUReplicatedOutput", scope.makeOpName("TPUReplicatedOutput"));
-    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("num_replicas", numReplicas);
     return new TPUReplicatedOutput<T>(opBuilder.build());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/WorkerHeartbeat.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/WorkerHeartbeat.java
index 49895008bd2..9ed34561956 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/WorkerHeartbeat.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/WorkerHeartbeat.java
@@ -45,7 +45,7 @@ public final class WorkerHeartbeat extends RawOp implements Operand<TString> {
   @Endpoint(describeByClass = true)
   public static WorkerHeartbeat create(Scope scope, Operand<TString> request) {
     OperationBuilder opBuilder = scope.env().opBuilder("WorkerHeartbeat", scope.makeOpName("WorkerHeartbeat"));
-    opBuilder.addInput(request.asOutput());
+    opBuilder.addInput(request.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new WorkerHeartbeat(opBuilder.build());
   }
@@ -58,7 +58,7 @@ public Output<TString> response() {
   }
   
   @Override
-  public Output<TString> asOutput() {
+  public Output<TString> asOutput(Scope scope) {
     return response;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/AccumulatorApplyGradient.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/AccumulatorApplyGradient.java
index eedb748560b..685ebe0de74 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/AccumulatorApplyGradient.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/AccumulatorApplyGradient.java
@@ -20,13 +20,13 @@
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
 import org.tensorflow.types.TString;
+import org.tensorflow.types.family.TType;
 
 /**
  * Applies a gradient to a given accumulator.
@@ -46,11 +46,11 @@ public final class AccumulatorApplyGradient extends RawOp {
    * @return a new instance of AccumulatorApplyGradient
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> AccumulatorApplyGradient create(Scope scope, Operand<TString> handle, Operand<TInt64> localStep, Operand<T> gradient) {
+  public static <T extends TType> AccumulatorApplyGradient create(Scope scope, Operand<TString> handle, Operand<TInt64> localStep, Operand<T> gradient) {
     OperationBuilder opBuilder = scope.env().opBuilder("AccumulatorApplyGradient", scope.makeOpName("AccumulatorApplyGradient"));
-    opBuilder.addInput(handle.asOutput());
-    opBuilder.addInput(localStep.asOutput());
-    opBuilder.addInput(gradient.asOutput());
+    opBuilder.addInput(handle.asOutput(scope));
+    opBuilder.addInput(localStep.asOutput(scope));
+    opBuilder.addInput(gradient.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new AccumulatorApplyGradient(opBuilder.build());
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/AccumulatorNumAccumulated.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/AccumulatorNumAccumulated.java
index 9fca915c550..cb8fc93f7aa 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/AccumulatorNumAccumulated.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/AccumulatorNumAccumulated.java
@@ -44,7 +44,7 @@ public final class AccumulatorNumAccumulated extends RawOp implements Operand<TI
   @Endpoint(describeByClass = true)
   public static AccumulatorNumAccumulated create(Scope scope, Operand<TString> handle) {
     OperationBuilder opBuilder = scope.env().opBuilder("AccumulatorNumAccumulated", scope.makeOpName("AccumulatorNumAccumulated"));
-    opBuilder.addInput(handle.asOutput());
+    opBuilder.addInput(handle.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new AccumulatorNumAccumulated(opBuilder.build());
   }
@@ -57,7 +57,7 @@ public Output<TInt32> numAccumulated() {
   }
   
   @Override
-  public Output<TInt32> asOutput() {
+  public Output<TInt32> asOutput(Scope scope) {
     return numAccumulated;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/AccumulatorSetGlobalStep.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/AccumulatorSetGlobalStep.java
index 9039d3a654d..8c354475de3 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/AccumulatorSetGlobalStep.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/AccumulatorSetGlobalStep.java
@@ -47,8 +47,8 @@ public final class AccumulatorSetGlobalStep extends RawOp {
   @Endpoint(describeByClass = true)
   public static AccumulatorSetGlobalStep create(Scope scope, Operand<TString> handle, Operand<TInt64> newGlobalStep) {
     OperationBuilder opBuilder = scope.env().opBuilder("AccumulatorSetGlobalStep", scope.makeOpName("AccumulatorSetGlobalStep"));
-    opBuilder.addInput(handle.asOutput());
-    opBuilder.addInput(newGlobalStep.asOutput());
+    opBuilder.addInput(handle.asOutput(scope));
+    opBuilder.addInput(newGlobalStep.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new AccumulatorSetGlobalStep(opBuilder.build());
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/AccumulatorTakeGradient.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/AccumulatorTakeGradient.java
index 95b394d3c45..e683c7c6496 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/AccumulatorTakeGradient.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/AccumulatorTakeGradient.java
@@ -22,13 +22,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
 import org.tensorflow.types.TString;
+import org.tensorflow.types.family.TType;
 
 /**
  * Extracts the average gradient in the given ConditionalAccumulator.
@@ -42,7 +42,7 @@
  * @param <T> data type for {@code average()} output
  */
 @Operator(group = "train")
-public final class AccumulatorTakeGradient<T extends Tensor> extends RawOp implements Operand<T> {
+public final class AccumulatorTakeGradient<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new AccumulatorTakeGradient operation.
@@ -55,10 +55,10 @@ public final class AccumulatorTakeGradient<T extends Tensor> extends RawOp imple
    * @return a new instance of AccumulatorTakeGradient
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> AccumulatorTakeGradient<T> create(Scope scope, Operand<TString> handle, Operand<TInt32> numRequired, DataType<T> dtype) {
+  public static <T extends TType> AccumulatorTakeGradient<T> create(Scope scope, Operand<TString> handle, Operand<TInt32> numRequired, DataType<T> dtype) {
     OperationBuilder opBuilder = scope.env().opBuilder("AccumulatorTakeGradient", scope.makeOpName("AccumulatorTakeGradient"));
-    opBuilder.addInput(handle.asOutput());
-    opBuilder.addInput(numRequired.asOutput());
+    opBuilder.addInput(handle.asOutput(scope));
+    opBuilder.addInput(numRequired.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("dtype", dtype);
     return new AccumulatorTakeGradient<T>(opBuilder.build());
@@ -72,7 +72,7 @@ public Output<T> average() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return average;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyAdaMax.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyAdaMax.java
index 62de9bcf1fc..8529a8d695b 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyAdaMax.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyAdaMax.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Update '*var' according to the AdaMax algorithm.
@@ -36,7 +36,7 @@
  * 
  * @param <T> data type for {@code out()} output
  */
-public final class ApplyAdaMax<T extends Tensor> extends RawOp implements Operand<T> {
+public final class ApplyAdaMax<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.train.ApplyAdaMax}
@@ -76,17 +76,17 @@ private Options() {
    * @return a new instance of ApplyAdaMax
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> ApplyAdaMax<T> create(Scope scope, Operand<T> var, Operand<T> m, Operand<T> v, Operand<T> beta1Power, Operand<T> lr, Operand<T> beta1, Operand<T> beta2, Operand<T> epsilon, Operand<T> grad, Options... options) {
+  public static <T extends TType> ApplyAdaMax<T> create(Scope scope, Operand<T> var, Operand<T> m, Operand<T> v, Operand<T> beta1Power, Operand<T> lr, Operand<T> beta1, Operand<T> beta2, Operand<T> epsilon, Operand<T> grad, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ApplyAdaMax", scope.makeOpName("ApplyAdaMax"));
-    opBuilder.addInput(var.asOutput());
-    opBuilder.addInput(m.asOutput());
-    opBuilder.addInput(v.asOutput());
-    opBuilder.addInput(beta1Power.asOutput());
-    opBuilder.addInput(lr.asOutput());
-    opBuilder.addInput(beta1.asOutput());
-    opBuilder.addInput(beta2.asOutput());
-    opBuilder.addInput(epsilon.asOutput());
-    opBuilder.addInput(grad.asOutput());
+    opBuilder.addInput(var.asOutput(scope));
+    opBuilder.addInput(m.asOutput(scope));
+    opBuilder.addInput(v.asOutput(scope));
+    opBuilder.addInput(beta1Power.asOutput(scope));
+    opBuilder.addInput(lr.asOutput(scope));
+    opBuilder.addInput(beta1.asOutput(scope));
+    opBuilder.addInput(beta2.asOutput(scope));
+    opBuilder.addInput(epsilon.asOutput(scope));
+    opBuilder.addInput(grad.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -115,7 +115,7 @@ public Output<T> out() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return out;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyAdadelta.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyAdadelta.java
index 0a75fc0d386..cef6fc27612 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyAdadelta.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyAdadelta.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Update '*var' according to the adadelta scheme.
@@ -38,7 +38,7 @@
  * @param <T> data type for {@code out()} output
  */
 @Operator(group = "train")
-public final class ApplyAdadelta<T extends Tensor> extends RawOp implements Operand<T> {
+public final class ApplyAdadelta<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.train.ApplyAdadelta}
@@ -75,15 +75,15 @@ private Options() {
    * @return a new instance of ApplyAdadelta
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> ApplyAdadelta<T> create(Scope scope, Operand<T> var, Operand<T> accum, Operand<T> accumUpdate, Operand<T> lr, Operand<T> rho, Operand<T> epsilon, Operand<T> grad, Options... options) {
+  public static <T extends TType> ApplyAdadelta<T> create(Scope scope, Operand<T> var, Operand<T> accum, Operand<T> accumUpdate, Operand<T> lr, Operand<T> rho, Operand<T> epsilon, Operand<T> grad, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ApplyAdadelta", scope.makeOpName("ApplyAdadelta"));
-    opBuilder.addInput(var.asOutput());
-    opBuilder.addInput(accum.asOutput());
-    opBuilder.addInput(accumUpdate.asOutput());
-    opBuilder.addInput(lr.asOutput());
-    opBuilder.addInput(rho.asOutput());
-    opBuilder.addInput(epsilon.asOutput());
-    opBuilder.addInput(grad.asOutput());
+    opBuilder.addInput(var.asOutput(scope));
+    opBuilder.addInput(accum.asOutput(scope));
+    opBuilder.addInput(accumUpdate.asOutput(scope));
+    opBuilder.addInput(lr.asOutput(scope));
+    opBuilder.addInput(rho.asOutput(scope));
+    opBuilder.addInput(epsilon.asOutput(scope));
+    opBuilder.addInput(grad.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -111,7 +111,7 @@ public Output<T> out() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return out;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyAdagrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyAdagrad.java
index 0cf6eed8c09..725ebde8e91 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyAdagrad.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyAdagrad.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Update '*var' according to the adagrad scheme.
@@ -36,7 +36,7 @@
  * @param <T> data type for {@code out()} output
  */
 @Operator(group = "train")
-public final class ApplyAdagrad<T extends Tensor> extends RawOp implements Operand<T> {
+public final class ApplyAdagrad<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.train.ApplyAdagrad}
@@ -80,12 +80,12 @@ private Options() {
    * @return a new instance of ApplyAdagrad
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> ApplyAdagrad<T> create(Scope scope, Operand<T> var, Operand<T> accum, Operand<T> lr, Operand<T> grad, Options... options) {
+  public static <T extends TType> ApplyAdagrad<T> create(Scope scope, Operand<T> var, Operand<T> accum, Operand<T> lr, Operand<T> grad, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ApplyAdagrad", scope.makeOpName("ApplyAdagrad"));
-    opBuilder.addInput(var.asOutput());
-    opBuilder.addInput(accum.asOutput());
-    opBuilder.addInput(lr.asOutput());
-    opBuilder.addInput(grad.asOutput());
+    opBuilder.addInput(var.asOutput(scope));
+    opBuilder.addInput(accum.asOutput(scope));
+    opBuilder.addInput(lr.asOutput(scope));
+    opBuilder.addInput(grad.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -124,7 +124,7 @@ public Output<T> out() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return out;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyAdagradDa.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyAdagradDa.java
index 7b09995fa1c..4c53fb860fe 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyAdagradDa.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyAdagradDa.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
 
 /**
  * Update '*var' according to the proximal adagrad scheme.
@@ -34,7 +34,7 @@
  * @param <T> data type for {@code out()} output
  */
 @Operator(group = "train")
-public final class ApplyAdagradDa<T extends Tensor> extends RawOp implements Operand<T> {
+public final class ApplyAdagradDa<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.train.ApplyAdagradDa}
@@ -72,16 +72,16 @@ private Options() {
    * @return a new instance of ApplyAdagradDa
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> ApplyAdagradDa<T> create(Scope scope, Operand<T> var, Operand<T> gradientAccumulator, Operand<T> gradientSquaredAccumulator, Operand<T> grad, Operand<T> lr, Operand<T> l1, Operand<T> l2, Operand<TInt64> globalStep, Options... options) {
+  public static <T extends TType> ApplyAdagradDa<T> create(Scope scope, Operand<T> var, Operand<T> gradientAccumulator, Operand<T> gradientSquaredAccumulator, Operand<T> grad, Operand<T> lr, Operand<T> l1, Operand<T> l2, Operand<TInt64> globalStep, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ApplyAdagradDA", scope.makeOpName("ApplyAdagradDa"));
-    opBuilder.addInput(var.asOutput());
-    opBuilder.addInput(gradientAccumulator.asOutput());
-    opBuilder.addInput(gradientSquaredAccumulator.asOutput());
-    opBuilder.addInput(grad.asOutput());
-    opBuilder.addInput(lr.asOutput());
-    opBuilder.addInput(l1.asOutput());
-    opBuilder.addInput(l2.asOutput());
-    opBuilder.addInput(globalStep.asOutput());
+    opBuilder.addInput(var.asOutput(scope));
+    opBuilder.addInput(gradientAccumulator.asOutput(scope));
+    opBuilder.addInput(gradientSquaredAccumulator.asOutput(scope));
+    opBuilder.addInput(grad.asOutput(scope));
+    opBuilder.addInput(lr.asOutput(scope));
+    opBuilder.addInput(l1.asOutput(scope));
+    opBuilder.addInput(l2.asOutput(scope));
+    opBuilder.addInput(globalStep.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -109,7 +109,7 @@ public Output<T> out() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return out;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyAdagradV2.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyAdagradV2.java
index 6b50f9ffed3..3d0e00bcfea 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyAdagradV2.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyAdagradV2.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Update '*var' according to the adagrad scheme.
@@ -35,7 +35,7 @@
  * 
  * @param <T> data type for {@code out()} output
  */
-public final class ApplyAdagradV2<T extends Tensor> extends RawOp implements Operand<T> {
+public final class ApplyAdagradV2<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.train.ApplyAdagradV2}
@@ -80,13 +80,13 @@ private Options() {
    * @return a new instance of ApplyAdagradV2
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> ApplyAdagradV2<T> create(Scope scope, Operand<T> var, Operand<T> accum, Operand<T> lr, Operand<T> epsilon, Operand<T> grad, Options... options) {
+  public static <T extends TType> ApplyAdagradV2<T> create(Scope scope, Operand<T> var, Operand<T> accum, Operand<T> lr, Operand<T> epsilon, Operand<T> grad, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ApplyAdagradV2", scope.makeOpName("ApplyAdagradV2"));
-    opBuilder.addInput(var.asOutput());
-    opBuilder.addInput(accum.asOutput());
-    opBuilder.addInput(lr.asOutput());
-    opBuilder.addInput(epsilon.asOutput());
-    opBuilder.addInput(grad.asOutput());
+    opBuilder.addInput(var.asOutput(scope));
+    opBuilder.addInput(accum.asOutput(scope));
+    opBuilder.addInput(lr.asOutput(scope));
+    opBuilder.addInput(epsilon.asOutput(scope));
+    opBuilder.addInput(grad.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -125,7 +125,7 @@ public Output<T> out() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return out;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyAdam.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyAdam.java
index b662b1f0203..1a212a4f9fd 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyAdam.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyAdam.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Update '*var' according to the Adam algorithm.
@@ -38,7 +38,7 @@
  * @param <T> data type for {@code out()} output
  */
 @Operator(group = "train")
-public final class ApplyAdam<T extends Tensor> extends RawOp implements Operand<T> {
+public final class ApplyAdam<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.train.ApplyAdam}
@@ -88,18 +88,18 @@ private Options() {
    * @return a new instance of ApplyAdam
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> ApplyAdam<T> create(Scope scope, Operand<T> var, Operand<T> m, Operand<T> v, Operand<T> beta1Power, Operand<T> beta2Power, Operand<T> lr, Operand<T> beta1, Operand<T> beta2, Operand<T> epsilon, Operand<T> grad, Options... options) {
+  public static <T extends TType> ApplyAdam<T> create(Scope scope, Operand<T> var, Operand<T> m, Operand<T> v, Operand<T> beta1Power, Operand<T> beta2Power, Operand<T> lr, Operand<T> beta1, Operand<T> beta2, Operand<T> epsilon, Operand<T> grad, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ApplyAdam", scope.makeOpName("ApplyAdam"));
-    opBuilder.addInput(var.asOutput());
-    opBuilder.addInput(m.asOutput());
-    opBuilder.addInput(v.asOutput());
-    opBuilder.addInput(beta1Power.asOutput());
-    opBuilder.addInput(beta2Power.asOutput());
-    opBuilder.addInput(lr.asOutput());
-    opBuilder.addInput(beta1.asOutput());
-    opBuilder.addInput(beta2.asOutput());
-    opBuilder.addInput(epsilon.asOutput());
-    opBuilder.addInput(grad.asOutput());
+    opBuilder.addInput(var.asOutput(scope));
+    opBuilder.addInput(m.asOutput(scope));
+    opBuilder.addInput(v.asOutput(scope));
+    opBuilder.addInput(beta1Power.asOutput(scope));
+    opBuilder.addInput(beta2Power.asOutput(scope));
+    opBuilder.addInput(lr.asOutput(scope));
+    opBuilder.addInput(beta1.asOutput(scope));
+    opBuilder.addInput(beta2.asOutput(scope));
+    opBuilder.addInput(epsilon.asOutput(scope));
+    opBuilder.addInput(grad.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -138,7 +138,7 @@ public Output<T> out() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return out;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyAddSign.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyAddSign.java
index 970172dcf14..ee3cb5e20a2 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyAddSign.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyAddSign.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Update '*var' according to the AddSign update.
@@ -37,7 +37,7 @@
  * @param <T> data type for {@code out()} output
  */
 @Operator(group = "train")
-public final class ApplyAddSign<T extends Tensor> extends RawOp implements Operand<T> {
+public final class ApplyAddSign<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.train.ApplyAddSign}
@@ -75,15 +75,15 @@ private Options() {
    * @return a new instance of ApplyAddSign
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> ApplyAddSign<T> create(Scope scope, Operand<T> var, Operand<T> m, Operand<T> lr, Operand<T> alpha, Operand<T> signDecay, Operand<T> beta, Operand<T> grad, Options... options) {
+  public static <T extends TType> ApplyAddSign<T> create(Scope scope, Operand<T> var, Operand<T> m, Operand<T> lr, Operand<T> alpha, Operand<T> signDecay, Operand<T> beta, Operand<T> grad, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ApplyAddSign", scope.makeOpName("ApplyAddSign"));
-    opBuilder.addInput(var.asOutput());
-    opBuilder.addInput(m.asOutput());
-    opBuilder.addInput(lr.asOutput());
-    opBuilder.addInput(alpha.asOutput());
-    opBuilder.addInput(signDecay.asOutput());
-    opBuilder.addInput(beta.asOutput());
-    opBuilder.addInput(grad.asOutput());
+    opBuilder.addInput(var.asOutput(scope));
+    opBuilder.addInput(m.asOutput(scope));
+    opBuilder.addInput(lr.asOutput(scope));
+    opBuilder.addInput(alpha.asOutput(scope));
+    opBuilder.addInput(signDecay.asOutput(scope));
+    opBuilder.addInput(beta.asOutput(scope));
+    opBuilder.addInput(grad.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -112,7 +112,7 @@ public Output<T> out() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return out;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyCenteredRmsProp.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyCenteredRmsProp.java
index 97a43dedd51..7150e30241b 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyCenteredRmsProp.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyCenteredRmsProp.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Update '*var' according to the centered RMSProp algorithm.
@@ -52,7 +52,7 @@
  * @param <T> data type for {@code out()} output
  */
 @Operator(group = "train")
-public final class ApplyCenteredRmsProp<T extends Tensor> extends RawOp implements Operand<T> {
+public final class ApplyCenteredRmsProp<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.train.ApplyCenteredRmsProp}
@@ -92,17 +92,17 @@ private Options() {
    * @return a new instance of ApplyCenteredRmsProp
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> ApplyCenteredRmsProp<T> create(Scope scope, Operand<T> var, Operand<T> mg, Operand<T> ms, Operand<T> mom, Operand<T> lr, Operand<T> rho, Operand<T> momentum, Operand<T> epsilon, Operand<T> grad, Options... options) {
+  public static <T extends TType> ApplyCenteredRmsProp<T> create(Scope scope, Operand<T> var, Operand<T> mg, Operand<T> ms, Operand<T> mom, Operand<T> lr, Operand<T> rho, Operand<T> momentum, Operand<T> epsilon, Operand<T> grad, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ApplyCenteredRMSProp", scope.makeOpName("ApplyCenteredRmsProp"));
-    opBuilder.addInput(var.asOutput());
-    opBuilder.addInput(mg.asOutput());
-    opBuilder.addInput(ms.asOutput());
-    opBuilder.addInput(mom.asOutput());
-    opBuilder.addInput(lr.asOutput());
-    opBuilder.addInput(rho.asOutput());
-    opBuilder.addInput(momentum.asOutput());
-    opBuilder.addInput(epsilon.asOutput());
-    opBuilder.addInput(grad.asOutput());
+    opBuilder.addInput(var.asOutput(scope));
+    opBuilder.addInput(mg.asOutput(scope));
+    opBuilder.addInput(ms.asOutput(scope));
+    opBuilder.addInput(mom.asOutput(scope));
+    opBuilder.addInput(lr.asOutput(scope));
+    opBuilder.addInput(rho.asOutput(scope));
+    opBuilder.addInput(momentum.asOutput(scope));
+    opBuilder.addInput(epsilon.asOutput(scope));
+    opBuilder.addInput(grad.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -131,7 +131,7 @@ public Output<T> out() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return out;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyFtrl.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyFtrl.java
index 368fa40cb87..b4e00bd8cc1 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyFtrl.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyFtrl.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Update '*var' according to the Ftrl-proximal scheme.
@@ -41,7 +41,7 @@
  * @param <T> data type for {@code out()} output
  */
 @Operator(group = "train")
-public final class ApplyFtrl<T extends Tensor> extends RawOp implements Operand<T> {
+public final class ApplyFtrl<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.train.ApplyFtrl}
@@ -90,17 +90,17 @@ private Options() {
    * @return a new instance of ApplyFtrl
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> ApplyFtrl<T> create(Scope scope, Operand<T> var, Operand<T> accum, Operand<T> linear, Operand<T> grad, Operand<T> lr, Operand<T> l1, Operand<T> l2, Operand<T> l2Shrinkage, Operand<T> lrPower, Options... options) {
+  public static <T extends TType> ApplyFtrl<T> create(Scope scope, Operand<T> var, Operand<T> accum, Operand<T> linear, Operand<T> grad, Operand<T> lr, Operand<T> l1, Operand<T> l2, Operand<T> l2Shrinkage, Operand<T> lrPower, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ApplyFtrlV2", scope.makeOpName("ApplyFtrl"));
-    opBuilder.addInput(var.asOutput());
-    opBuilder.addInput(accum.asOutput());
-    opBuilder.addInput(linear.asOutput());
-    opBuilder.addInput(grad.asOutput());
-    opBuilder.addInput(lr.asOutput());
-    opBuilder.addInput(l1.asOutput());
-    opBuilder.addInput(l2.asOutput());
-    opBuilder.addInput(l2Shrinkage.asOutput());
-    opBuilder.addInput(lrPower.asOutput());
+    opBuilder.addInput(var.asOutput(scope));
+    opBuilder.addInput(accum.asOutput(scope));
+    opBuilder.addInput(linear.asOutput(scope));
+    opBuilder.addInput(grad.asOutput(scope));
+    opBuilder.addInput(lr.asOutput(scope));
+    opBuilder.addInput(l1.asOutput(scope));
+    opBuilder.addInput(l2.asOutput(scope));
+    opBuilder.addInput(l2Shrinkage.asOutput(scope));
+    opBuilder.addInput(lrPower.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -139,7 +139,7 @@ public Output<T> out() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return out;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyGradientDescent.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyGradientDescent.java
index 4cfb6878c25..450adff5136 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyGradientDescent.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyGradientDescent.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Update '*var' by subtracting 'alpha' * 'delta' from it.
@@ -33,7 +33,7 @@
  * @param <T> data type for {@code out()} output
  */
 @Operator(group = "train")
-public final class ApplyGradientDescent<T extends Tensor> extends RawOp implements Operand<T> {
+public final class ApplyGradientDescent<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.train.ApplyGradientDescent}
@@ -66,11 +66,11 @@ private Options() {
    * @return a new instance of ApplyGradientDescent
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> ApplyGradientDescent<T> create(Scope scope, Operand<T> var, Operand<T> alpha, Operand<T> delta, Options... options) {
+  public static <T extends TType> ApplyGradientDescent<T> create(Scope scope, Operand<T> var, Operand<T> alpha, Operand<T> delta, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ApplyGradientDescent", scope.makeOpName("ApplyGradientDescent"));
-    opBuilder.addInput(var.asOutput());
-    opBuilder.addInput(alpha.asOutput());
-    opBuilder.addInput(delta.asOutput());
+    opBuilder.addInput(var.asOutput(scope));
+    opBuilder.addInput(alpha.asOutput(scope));
+    opBuilder.addInput(delta.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -98,7 +98,7 @@ public Output<T> out() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return out;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyMomentum.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyMomentum.java
index 24e221691e1..a3a4e1cde3c 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyMomentum.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyMomentum.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Update '*var' according to the momentum scheme.
@@ -38,7 +38,7 @@
  * @param <T> data type for {@code out()} output
  */
 @Operator(group = "train")
-public final class ApplyMomentum<T extends Tensor> extends RawOp implements Operand<T> {
+public final class ApplyMomentum<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.train.ApplyMomentum}
@@ -85,13 +85,13 @@ private Options() {
    * @return a new instance of ApplyMomentum
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> ApplyMomentum<T> create(Scope scope, Operand<T> var, Operand<T> accum, Operand<T> lr, Operand<T> grad, Operand<T> momentum, Options... options) {
+  public static <T extends TType> ApplyMomentum<T> create(Scope scope, Operand<T> var, Operand<T> accum, Operand<T> lr, Operand<T> grad, Operand<T> momentum, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ApplyMomentum", scope.makeOpName("ApplyMomentum"));
-    opBuilder.addInput(var.asOutput());
-    opBuilder.addInput(accum.asOutput());
-    opBuilder.addInput(lr.asOutput());
-    opBuilder.addInput(grad.asOutput());
-    opBuilder.addInput(momentum.asOutput());
+    opBuilder.addInput(var.asOutput(scope));
+    opBuilder.addInput(accum.asOutput(scope));
+    opBuilder.addInput(lr.asOutput(scope));
+    opBuilder.addInput(grad.asOutput(scope));
+    opBuilder.addInput(momentum.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -132,7 +132,7 @@ public Output<T> out() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return out;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyPowerSign.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyPowerSign.java
index a1a3d8000bc..e13fd58022d 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyPowerSign.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyPowerSign.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Update '*var' according to the AddSign update.
@@ -37,7 +37,7 @@
  * @param <T> data type for {@code out()} output
  */
 @Operator(group = "train")
-public final class ApplyPowerSign<T extends Tensor> extends RawOp implements Operand<T> {
+public final class ApplyPowerSign<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.train.ApplyPowerSign}
@@ -75,15 +75,15 @@ private Options() {
    * @return a new instance of ApplyPowerSign
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> ApplyPowerSign<T> create(Scope scope, Operand<T> var, Operand<T> m, Operand<T> lr, Operand<T> logbase, Operand<T> signDecay, Operand<T> beta, Operand<T> grad, Options... options) {
+  public static <T extends TType> ApplyPowerSign<T> create(Scope scope, Operand<T> var, Operand<T> m, Operand<T> lr, Operand<T> logbase, Operand<T> signDecay, Operand<T> beta, Operand<T> grad, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ApplyPowerSign", scope.makeOpName("ApplyPowerSign"));
-    opBuilder.addInput(var.asOutput());
-    opBuilder.addInput(m.asOutput());
-    opBuilder.addInput(lr.asOutput());
-    opBuilder.addInput(logbase.asOutput());
-    opBuilder.addInput(signDecay.asOutput());
-    opBuilder.addInput(beta.asOutput());
-    opBuilder.addInput(grad.asOutput());
+    opBuilder.addInput(var.asOutput(scope));
+    opBuilder.addInput(m.asOutput(scope));
+    opBuilder.addInput(lr.asOutput(scope));
+    opBuilder.addInput(logbase.asOutput(scope));
+    opBuilder.addInput(signDecay.asOutput(scope));
+    opBuilder.addInput(beta.asOutput(scope));
+    opBuilder.addInput(grad.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -112,7 +112,7 @@ public Output<T> out() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return out;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyProximalAdagrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyProximalAdagrad.java
index afec66ccab6..ec8e57b7c5a 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyProximalAdagrad.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyProximalAdagrad.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Update '*var' and '*accum' according to FOBOS with Adagrad learning rate.
@@ -37,7 +37,7 @@
  * @param <T> data type for {@code out()} output
  */
 @Operator(group = "train")
-public final class ApplyProximalAdagrad<T extends Tensor> extends RawOp implements Operand<T> {
+public final class ApplyProximalAdagrad<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.train.ApplyProximalAdagrad}
@@ -73,14 +73,14 @@ private Options() {
    * @return a new instance of ApplyProximalAdagrad
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> ApplyProximalAdagrad<T> create(Scope scope, Operand<T> var, Operand<T> accum, Operand<T> lr, Operand<T> l1, Operand<T> l2, Operand<T> grad, Options... options) {
+  public static <T extends TType> ApplyProximalAdagrad<T> create(Scope scope, Operand<T> var, Operand<T> accum, Operand<T> lr, Operand<T> l1, Operand<T> l2, Operand<T> grad, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ApplyProximalAdagrad", scope.makeOpName("ApplyProximalAdagrad"));
-    opBuilder.addInput(var.asOutput());
-    opBuilder.addInput(accum.asOutput());
-    opBuilder.addInput(lr.asOutput());
-    opBuilder.addInput(l1.asOutput());
-    opBuilder.addInput(l2.asOutput());
-    opBuilder.addInput(grad.asOutput());
+    opBuilder.addInput(var.asOutput(scope));
+    opBuilder.addInput(accum.asOutput(scope));
+    opBuilder.addInput(lr.asOutput(scope));
+    opBuilder.addInput(l1.asOutput(scope));
+    opBuilder.addInput(l2.asOutput(scope));
+    opBuilder.addInput(grad.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -108,7 +108,7 @@ public Output<T> out() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return out;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyProximalGradientDescent.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyProximalGradientDescent.java
index b3040099f58..8ea81694471 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyProximalGradientDescent.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyProximalGradientDescent.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Update '*var' as FOBOS algorithm with fixed learning rate.
@@ -36,7 +36,7 @@
  * @param <T> data type for {@code out()} output
  */
 @Operator(group = "train")
-public final class ApplyProximalGradientDescent<T extends Tensor> extends RawOp implements Operand<T> {
+public final class ApplyProximalGradientDescent<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.train.ApplyProximalGradientDescent}
@@ -71,13 +71,13 @@ private Options() {
    * @return a new instance of ApplyProximalGradientDescent
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> ApplyProximalGradientDescent<T> create(Scope scope, Operand<T> var, Operand<T> alpha, Operand<T> l1, Operand<T> l2, Operand<T> delta, Options... options) {
+  public static <T extends TType> ApplyProximalGradientDescent<T> create(Scope scope, Operand<T> var, Operand<T> alpha, Operand<T> l1, Operand<T> l2, Operand<T> delta, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ApplyProximalGradientDescent", scope.makeOpName("ApplyProximalGradientDescent"));
-    opBuilder.addInput(var.asOutput());
-    opBuilder.addInput(alpha.asOutput());
-    opBuilder.addInput(l1.asOutput());
-    opBuilder.addInput(l2.asOutput());
-    opBuilder.addInput(delta.asOutput());
+    opBuilder.addInput(var.asOutput(scope));
+    opBuilder.addInput(alpha.asOutput(scope));
+    opBuilder.addInput(l1.asOutput(scope));
+    opBuilder.addInput(l2.asOutput(scope));
+    opBuilder.addInput(delta.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -105,7 +105,7 @@ public Output<T> out() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return out;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyRmsProp.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyRmsProp.java
index 2b1bd6bb0b1..9d5e1e4e449 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyRmsProp.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyRmsProp.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Update '*var' according to the RMSProp algorithm.
@@ -44,7 +44,7 @@
  * @param <T> data type for {@code out()} output
  */
 @Operator(group = "train")
-public final class ApplyRmsProp<T extends Tensor> extends RawOp implements Operand<T> {
+public final class ApplyRmsProp<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.train.ApplyRmsProp}
@@ -83,16 +83,16 @@ private Options() {
    * @return a new instance of ApplyRmsProp
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> ApplyRmsProp<T> create(Scope scope, Operand<T> var, Operand<T> ms, Operand<T> mom, Operand<T> lr, Operand<T> rho, Operand<T> momentum, Operand<T> epsilon, Operand<T> grad, Options... options) {
+  public static <T extends TType> ApplyRmsProp<T> create(Scope scope, Operand<T> var, Operand<T> ms, Operand<T> mom, Operand<T> lr, Operand<T> rho, Operand<T> momentum, Operand<T> epsilon, Operand<T> grad, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ApplyRMSProp", scope.makeOpName("ApplyRmsProp"));
-    opBuilder.addInput(var.asOutput());
-    opBuilder.addInput(ms.asOutput());
-    opBuilder.addInput(mom.asOutput());
-    opBuilder.addInput(lr.asOutput());
-    opBuilder.addInput(rho.asOutput());
-    opBuilder.addInput(momentum.asOutput());
-    opBuilder.addInput(epsilon.asOutput());
-    opBuilder.addInput(grad.asOutput());
+    opBuilder.addInput(var.asOutput(scope));
+    opBuilder.addInput(ms.asOutput(scope));
+    opBuilder.addInput(mom.asOutput(scope));
+    opBuilder.addInput(lr.asOutput(scope));
+    opBuilder.addInput(rho.asOutput(scope));
+    opBuilder.addInput(momentum.asOutput(scope));
+    opBuilder.addInput(epsilon.asOutput(scope));
+    opBuilder.addInput(grad.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -121,7 +121,7 @@ public Output<T> out() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return out;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/BatchMatMul.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/BatchMatMul.java
index beac69ae46d..b362fe5447a 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/BatchMatMul.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/BatchMatMul.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Multiplies slices of two tensors in batches.
@@ -57,7 +57,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "train")
-public final class BatchMatMul<T extends Tensor> extends RawOp implements Operand<T> {
+public final class BatchMatMul<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.train.BatchMatMul}
@@ -97,10 +97,10 @@ private Options() {
    * @return a new instance of BatchMatMul
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> BatchMatMul<T> create(Scope scope, Operand<T> x, Operand<T> y, Options... options) {
+  public static <T extends TType> BatchMatMul<T> create(Scope scope, Operand<T> x, Operand<T> y, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("BatchMatMulV2", scope.makeOpName("BatchMatMul"));
-    opBuilder.addInput(x.asOutput());
-    opBuilder.addInput(y.asOutput());
+    opBuilder.addInput(x.asOutput(scope));
+    opBuilder.addInput(y.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -137,7 +137,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ConditionalAccumulator.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ConditionalAccumulator.java
index 4cad6d75ff2..9bd06537e1f 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ConditionalAccumulator.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ConditionalAccumulator.java
@@ -22,13 +22,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TString;
+import org.tensorflow.types.family.TType;
 
 /**
  * A conditional accumulator for aggregating gradients.
@@ -92,7 +92,7 @@ private Options() {
    * @return a new instance of ConditionalAccumulator
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> ConditionalAccumulator create(Scope scope, DataType<T> dtype, Shape shape, Options... options) {
+  public static <T extends TType> ConditionalAccumulator create(Scope scope, DataType<T> dtype, Shape shape, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ConditionalAccumulator", scope.makeOpName("ConditionalAccumulator"));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("dtype", dtype);
@@ -144,7 +144,7 @@ public Output<TString> handle() {
   }
   
   @Override
-  public Output<TString> asOutput() {
+  public Output<TString> asOutput(Scope scope) {
     return handle;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/GenerateVocabRemapping.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/GenerateVocabRemapping.java
index e91afa44cd8..7204dc32b0e 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/GenerateVocabRemapping.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/GenerateVocabRemapping.java
@@ -98,8 +98,8 @@ private Options() {
   @Endpoint(describeByClass = true)
   public static GenerateVocabRemapping create(Scope scope, Operand<TString> newVocabFile, Operand<TString> oldVocabFile, Long newVocabOffset, Long numNewVocab, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("GenerateVocabRemapping", scope.makeOpName("GenerateVocabRemapping"));
-    opBuilder.addInput(newVocabFile.asOutput());
-    opBuilder.addInput(oldVocabFile.asOutput());
+    opBuilder.addInput(newVocabFile.asOutput(scope));
+    opBuilder.addInput(oldVocabFile.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("new_vocab_offset", newVocabOffset);
     opBuilder.setAttr("num_new_vocab", numNewVocab);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/MergeV2Checkpoints.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/MergeV2Checkpoints.java
index 986553a2d8e..dbd84af2f49 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/MergeV2Checkpoints.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/MergeV2Checkpoints.java
@@ -73,8 +73,8 @@ private Options() {
   @Endpoint(describeByClass = true)
   public static MergeV2Checkpoints create(Scope scope, Operand<TString> checkpointPrefixes, Operand<TString> destinationPrefix, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("MergeV2Checkpoints", scope.makeOpName("MergeV2Checkpoints"));
-    opBuilder.addInput(checkpointPrefixes.asOutput());
-    opBuilder.addInput(destinationPrefix.asOutput());
+    opBuilder.addInput(checkpointPrefixes.asOutput(scope));
+    opBuilder.addInput(destinationPrefix.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/NegTrain.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/NegTrain.java
index b3e5316ad7e..fc0cc638422 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/NegTrain.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/NegTrain.java
@@ -50,11 +50,11 @@ public final class NegTrain extends RawOp {
   @Endpoint(describeByClass = true)
   public static NegTrain create(Scope scope, Operand<TFloat32> wIn, Operand<TFloat32> wOut, Operand<TInt32> examples, Operand<TInt32> labels, Operand<TFloat32> lr, List<Long> vocabCount, Long numNegativeSamples) {
     OperationBuilder opBuilder = scope.env().opBuilder("NegTrain", scope.makeOpName("NegTrain"));
-    opBuilder.addInput(wIn.asOutput());
-    opBuilder.addInput(wOut.asOutput());
-    opBuilder.addInput(examples.asOutput());
-    opBuilder.addInput(labels.asOutput());
-    opBuilder.addInput(lr.asOutput());
+    opBuilder.addInput(wIn.asOutput(scope));
+    opBuilder.addInput(wOut.asOutput(scope));
+    opBuilder.addInput(examples.asOutput(scope));
+    opBuilder.addInput(labels.asOutput(scope));
+    opBuilder.addInput(lr.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     long[] vocabCountArray = new long[vocabCount.size()];
     for (int i = 0; i < vocabCountArray.length; ++i) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/PreventGradient.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/PreventGradient.java
index 8911db15871..46f03708cce 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/PreventGradient.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/PreventGradient.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * An identity op that triggers an error if a gradient is requested.
@@ -41,7 +41,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "train")
-public final class PreventGradient<T extends Tensor> extends RawOp implements Operand<T> {
+public final class PreventGradient<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.train.PreventGradient}
@@ -72,9 +72,9 @@ private Options() {
    * @return a new instance of PreventGradient
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> PreventGradient<T> create(Scope scope, Operand<T> input, Options... options) {
+  public static <T extends TType> PreventGradient<T> create(Scope scope, Operand<T> input, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("PreventGradient", scope.makeOpName("PreventGradient"));
-    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -102,7 +102,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceAccumulatorApplyGradient.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceAccumulatorApplyGradient.java
index 7631a6f32b4..035969ce383 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceAccumulatorApplyGradient.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceAccumulatorApplyGradient.java
@@ -20,12 +20,12 @@
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
 
 /**
  * Applies a gradient to a given accumulator.
@@ -44,11 +44,11 @@ public final class ResourceAccumulatorApplyGradient extends RawOp {
    * @return a new instance of ResourceAccumulatorApplyGradient
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> ResourceAccumulatorApplyGradient create(Scope scope, Operand<?> handle, Operand<TInt64> localStep, Operand<T> gradient) {
+  public static <T extends TType> ResourceAccumulatorApplyGradient create(Scope scope, Operand<?> handle, Operand<TInt64> localStep, Operand<T> gradient) {
     OperationBuilder opBuilder = scope.env().opBuilder("ResourceAccumulatorApplyGradient", scope.makeOpName("ResourceAccumulatorApplyGradient"));
-    opBuilder.addInput(handle.asOutput());
-    opBuilder.addInput(localStep.asOutput());
-    opBuilder.addInput(gradient.asOutput());
+    opBuilder.addInput(handle.asOutput(scope));
+    opBuilder.addInput(localStep.asOutput(scope));
+    opBuilder.addInput(gradient.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new ResourceAccumulatorApplyGradient(opBuilder.build());
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceAccumulatorNumAccumulated.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceAccumulatorNumAccumulated.java
index 628a916f659..41b4ddbd718 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceAccumulatorNumAccumulated.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceAccumulatorNumAccumulated.java
@@ -42,7 +42,7 @@ public final class ResourceAccumulatorNumAccumulated extends RawOp implements Op
   @Endpoint(describeByClass = true)
   public static ResourceAccumulatorNumAccumulated create(Scope scope, Operand<?> handle) {
     OperationBuilder opBuilder = scope.env().opBuilder("ResourceAccumulatorNumAccumulated", scope.makeOpName("ResourceAccumulatorNumAccumulated"));
-    opBuilder.addInput(handle.asOutput());
+    opBuilder.addInput(handle.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new ResourceAccumulatorNumAccumulated(opBuilder.build());
   }
@@ -55,7 +55,7 @@ public Output<TInt32> numAccumulated() {
   }
   
   @Override
-  public Output<TInt32> asOutput() {
+  public Output<TInt32> asOutput(Scope scope) {
     return numAccumulated;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceAccumulatorSetGlobalStep.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceAccumulatorSetGlobalStep.java
index 37570909340..46c8f2c9108 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceAccumulatorSetGlobalStep.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceAccumulatorSetGlobalStep.java
@@ -45,8 +45,8 @@ public final class ResourceAccumulatorSetGlobalStep extends RawOp {
   @Endpoint(describeByClass = true)
   public static ResourceAccumulatorSetGlobalStep create(Scope scope, Operand<?> handle, Operand<TInt64> newGlobalStep) {
     OperationBuilder opBuilder = scope.env().opBuilder("ResourceAccumulatorSetGlobalStep", scope.makeOpName("ResourceAccumulatorSetGlobalStep"));
-    opBuilder.addInput(handle.asOutput());
-    opBuilder.addInput(newGlobalStep.asOutput());
+    opBuilder.addInput(handle.asOutput(scope));
+    opBuilder.addInput(newGlobalStep.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new ResourceAccumulatorSetGlobalStep(opBuilder.build());
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceAccumulatorTakeGradient.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceAccumulatorTakeGradient.java
index 2f31831fff8..91314a0f00a 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceAccumulatorTakeGradient.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceAccumulatorTakeGradient.java
@@ -22,12 +22,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TType;
 
 /**
  * Extracts the average gradient in the given ConditionalAccumulator.
@@ -40,7 +40,7 @@
  * 
  * @param <T> data type for {@code average()} output
  */
-public final class ResourceAccumulatorTakeGradient<T extends Tensor> extends RawOp implements Operand<T> {
+public final class ResourceAccumulatorTakeGradient<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new ResourceAccumulatorTakeGradient operation.
@@ -53,10 +53,10 @@ public final class ResourceAccumulatorTakeGradient<T extends Tensor> extends Raw
    * @return a new instance of ResourceAccumulatorTakeGradient
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> ResourceAccumulatorTakeGradient<T> create(Scope scope, Operand<?> handle, Operand<TInt32> numRequired, DataType<T> dtype) {
+  public static <T extends TType> ResourceAccumulatorTakeGradient<T> create(Scope scope, Operand<?> handle, Operand<TInt32> numRequired, DataType<T> dtype) {
     OperationBuilder opBuilder = scope.env().opBuilder("ResourceAccumulatorTakeGradient", scope.makeOpName("ResourceAccumulatorTakeGradient"));
-    opBuilder.addInput(handle.asOutput());
-    opBuilder.addInput(numRequired.asOutput());
+    opBuilder.addInput(handle.asOutput(scope));
+    opBuilder.addInput(numRequired.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("dtype", dtype);
     return new ResourceAccumulatorTakeGradient<T>(opBuilder.build());
@@ -70,7 +70,7 @@ public Output<T> average() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return average;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyAdaMax.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyAdaMax.java
index eb4720db721..67d59ac422b 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyAdaMax.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyAdaMax.java
@@ -20,11 +20,11 @@
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Update '*var' according to the AdaMax algorithm.
@@ -73,17 +73,17 @@ private Options() {
    * @return a new instance of ResourceApplyAdaMax
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> ResourceApplyAdaMax create(Scope scope, Operand<?> var, Operand<?> m, Operand<?> v, Operand<T> beta1Power, Operand<T> lr, Operand<T> beta1, Operand<T> beta2, Operand<T> epsilon, Operand<T> grad, Options... options) {
+  public static <T extends TType> ResourceApplyAdaMax create(Scope scope, Operand<?> var, Operand<?> m, Operand<?> v, Operand<T> beta1Power, Operand<T> lr, Operand<T> beta1, Operand<T> beta2, Operand<T> epsilon, Operand<T> grad, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ResourceApplyAdaMax", scope.makeOpName("ResourceApplyAdaMax"));
-    opBuilder.addInput(var.asOutput());
-    opBuilder.addInput(m.asOutput());
-    opBuilder.addInput(v.asOutput());
-    opBuilder.addInput(beta1Power.asOutput());
-    opBuilder.addInput(lr.asOutput());
-    opBuilder.addInput(beta1.asOutput());
-    opBuilder.addInput(beta2.asOutput());
-    opBuilder.addInput(epsilon.asOutput());
-    opBuilder.addInput(grad.asOutput());
+    opBuilder.addInput(var.asOutput(scope));
+    opBuilder.addInput(m.asOutput(scope));
+    opBuilder.addInput(v.asOutput(scope));
+    opBuilder.addInput(beta1Power.asOutput(scope));
+    opBuilder.addInput(lr.asOutput(scope));
+    opBuilder.addInput(beta1.asOutput(scope));
+    opBuilder.addInput(beta2.asOutput(scope));
+    opBuilder.addInput(epsilon.asOutput(scope));
+    opBuilder.addInput(grad.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyAdadelta.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyAdadelta.java
index 19687ce7090..fdf35b45ce7 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyAdadelta.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyAdadelta.java
@@ -20,11 +20,11 @@
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Update '*var' according to the adadelta scheme.
@@ -72,15 +72,15 @@ private Options() {
    * @return a new instance of ResourceApplyAdadelta
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> ResourceApplyAdadelta create(Scope scope, Operand<?> var, Operand<?> accum, Operand<?> accumUpdate, Operand<T> lr, Operand<T> rho, Operand<T> epsilon, Operand<T> grad, Options... options) {
+  public static <T extends TType> ResourceApplyAdadelta create(Scope scope, Operand<?> var, Operand<?> accum, Operand<?> accumUpdate, Operand<T> lr, Operand<T> rho, Operand<T> epsilon, Operand<T> grad, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ResourceApplyAdadelta", scope.makeOpName("ResourceApplyAdadelta"));
-    opBuilder.addInput(var.asOutput());
-    opBuilder.addInput(accum.asOutput());
-    opBuilder.addInput(accumUpdate.asOutput());
-    opBuilder.addInput(lr.asOutput());
-    opBuilder.addInput(rho.asOutput());
-    opBuilder.addInput(epsilon.asOutput());
-    opBuilder.addInput(grad.asOutput());
+    opBuilder.addInput(var.asOutput(scope));
+    opBuilder.addInput(accum.asOutput(scope));
+    opBuilder.addInput(accumUpdate.asOutput(scope));
+    opBuilder.addInput(lr.asOutput(scope));
+    opBuilder.addInput(rho.asOutput(scope));
+    opBuilder.addInput(epsilon.asOutput(scope));
+    opBuilder.addInput(grad.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyAdagrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyAdagrad.java
index 2ab492152d6..8b6817db364 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyAdagrad.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyAdagrad.java
@@ -20,11 +20,11 @@
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Update '*var' according to the adagrad scheme.
@@ -77,13 +77,13 @@ private Options() {
    * @return a new instance of ResourceApplyAdagrad
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> ResourceApplyAdagrad create(Scope scope, Operand<?> var, Operand<?> accum, Operand<T> lr, Operand<T> epsilon, Operand<T> grad, Options... options) {
+  public static <T extends TType> ResourceApplyAdagrad create(Scope scope, Operand<?> var, Operand<?> accum, Operand<T> lr, Operand<T> epsilon, Operand<T> grad, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ResourceApplyAdagradV2", scope.makeOpName("ResourceApplyAdagrad"));
-    opBuilder.addInput(var.asOutput());
-    opBuilder.addInput(accum.asOutput());
-    opBuilder.addInput(lr.asOutput());
-    opBuilder.addInput(epsilon.asOutput());
-    opBuilder.addInput(grad.asOutput());
+    opBuilder.addInput(var.asOutput(scope));
+    opBuilder.addInput(accum.asOutput(scope));
+    opBuilder.addInput(lr.asOutput(scope));
+    opBuilder.addInput(epsilon.asOutput(scope));
+    opBuilder.addInput(grad.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyAdagradDa.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyAdagradDa.java
index a2ea0b0b917..a79c4e454b1 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyAdagradDa.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyAdagradDa.java
@@ -20,12 +20,12 @@
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
 
 /**
  * Update '*var' according to the proximal adagrad scheme.
@@ -69,16 +69,16 @@ private Options() {
    * @return a new instance of ResourceApplyAdagradDa
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> ResourceApplyAdagradDa create(Scope scope, Operand<?> var, Operand<?> gradientAccumulator, Operand<?> gradientSquaredAccumulator, Operand<T> grad, Operand<T> lr, Operand<T> l1, Operand<T> l2, Operand<TInt64> globalStep, Options... options) {
+  public static <T extends TType> ResourceApplyAdagradDa create(Scope scope, Operand<?> var, Operand<?> gradientAccumulator, Operand<?> gradientSquaredAccumulator, Operand<T> grad, Operand<T> lr, Operand<T> l1, Operand<T> l2, Operand<TInt64> globalStep, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ResourceApplyAdagradDA", scope.makeOpName("ResourceApplyAdagradDa"));
-    opBuilder.addInput(var.asOutput());
-    opBuilder.addInput(gradientAccumulator.asOutput());
-    opBuilder.addInput(gradientSquaredAccumulator.asOutput());
-    opBuilder.addInput(grad.asOutput());
-    opBuilder.addInput(lr.asOutput());
-    opBuilder.addInput(l1.asOutput());
-    opBuilder.addInput(l2.asOutput());
-    opBuilder.addInput(globalStep.asOutput());
+    opBuilder.addInput(var.asOutput(scope));
+    opBuilder.addInput(gradientAccumulator.asOutput(scope));
+    opBuilder.addInput(gradientSquaredAccumulator.asOutput(scope));
+    opBuilder.addInput(grad.asOutput(scope));
+    opBuilder.addInput(lr.asOutput(scope));
+    opBuilder.addInput(l1.asOutput(scope));
+    opBuilder.addInput(l2.asOutput(scope));
+    opBuilder.addInput(globalStep.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyAdam.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyAdam.java
index 7acf59a7c16..8867adfa7f6 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyAdam.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyAdam.java
@@ -20,11 +20,11 @@
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Update '*var' according to the Adam algorithm.
@@ -85,18 +85,18 @@ private Options() {
    * @return a new instance of ResourceApplyAdam
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> ResourceApplyAdam create(Scope scope, Operand<?> var, Operand<?> m, Operand<?> v, Operand<T> beta1Power, Operand<T> beta2Power, Operand<T> lr, Operand<T> beta1, Operand<T> beta2, Operand<T> epsilon, Operand<T> grad, Options... options) {
+  public static <T extends TType> ResourceApplyAdam create(Scope scope, Operand<?> var, Operand<?> m, Operand<?> v, Operand<T> beta1Power, Operand<T> beta2Power, Operand<T> lr, Operand<T> beta1, Operand<T> beta2, Operand<T> epsilon, Operand<T> grad, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ResourceApplyAdam", scope.makeOpName("ResourceApplyAdam"));
-    opBuilder.addInput(var.asOutput());
-    opBuilder.addInput(m.asOutput());
-    opBuilder.addInput(v.asOutput());
-    opBuilder.addInput(beta1Power.asOutput());
-    opBuilder.addInput(beta2Power.asOutput());
-    opBuilder.addInput(lr.asOutput());
-    opBuilder.addInput(beta1.asOutput());
-    opBuilder.addInput(beta2.asOutput());
-    opBuilder.addInput(epsilon.asOutput());
-    opBuilder.addInput(grad.asOutput());
+    opBuilder.addInput(var.asOutput(scope));
+    opBuilder.addInput(m.asOutput(scope));
+    opBuilder.addInput(v.asOutput(scope));
+    opBuilder.addInput(beta1Power.asOutput(scope));
+    opBuilder.addInput(beta2Power.asOutput(scope));
+    opBuilder.addInput(lr.asOutput(scope));
+    opBuilder.addInput(beta1.asOutput(scope));
+    opBuilder.addInput(beta2.asOutput(scope));
+    opBuilder.addInput(epsilon.asOutput(scope));
+    opBuilder.addInput(grad.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyAdamWithAmsgrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyAdamWithAmsgrad.java
index 6d1806cd6c8..a9d5b7b72d2 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyAdamWithAmsgrad.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyAdamWithAmsgrad.java
@@ -20,11 +20,11 @@
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Update '*var' according to the Adam algorithm.
@@ -78,19 +78,19 @@ private Options() {
    * @return a new instance of ResourceApplyAdamWithAmsgrad
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> ResourceApplyAdamWithAmsgrad create(Scope scope, Operand<?> var, Operand<?> m, Operand<?> v, Operand<?> vhat, Operand<T> beta1Power, Operand<T> beta2Power, Operand<T> lr, Operand<T> beta1, Operand<T> beta2, Operand<T> epsilon, Operand<T> grad, Options... options) {
+  public static <T extends TType> ResourceApplyAdamWithAmsgrad create(Scope scope, Operand<?> var, Operand<?> m, Operand<?> v, Operand<?> vhat, Operand<T> beta1Power, Operand<T> beta2Power, Operand<T> lr, Operand<T> beta1, Operand<T> beta2, Operand<T> epsilon, Operand<T> grad, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ResourceApplyAdamWithAmsgrad", scope.makeOpName("ResourceApplyAdamWithAmsgrad"));
-    opBuilder.addInput(var.asOutput());
-    opBuilder.addInput(m.asOutput());
-    opBuilder.addInput(v.asOutput());
-    opBuilder.addInput(vhat.asOutput());
-    opBuilder.addInput(beta1Power.asOutput());
-    opBuilder.addInput(beta2Power.asOutput());
-    opBuilder.addInput(lr.asOutput());
-    opBuilder.addInput(beta1.asOutput());
-    opBuilder.addInput(beta2.asOutput());
-    opBuilder.addInput(epsilon.asOutput());
-    opBuilder.addInput(grad.asOutput());
+    opBuilder.addInput(var.asOutput(scope));
+    opBuilder.addInput(m.asOutput(scope));
+    opBuilder.addInput(v.asOutput(scope));
+    opBuilder.addInput(vhat.asOutput(scope));
+    opBuilder.addInput(beta1Power.asOutput(scope));
+    opBuilder.addInput(beta2Power.asOutput(scope));
+    opBuilder.addInput(lr.asOutput(scope));
+    opBuilder.addInput(beta1.asOutput(scope));
+    opBuilder.addInput(beta2.asOutput(scope));
+    opBuilder.addInput(epsilon.asOutput(scope));
+    opBuilder.addInput(grad.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyAddSign.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyAddSign.java
index f8670ff14c5..e23409074f7 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyAddSign.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyAddSign.java
@@ -20,11 +20,11 @@
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Update '*var' according to the AddSign update.
@@ -72,15 +72,15 @@ private Options() {
    * @return a new instance of ResourceApplyAddSign
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> ResourceApplyAddSign create(Scope scope, Operand<?> var, Operand<?> m, Operand<T> lr, Operand<T> alpha, Operand<T> signDecay, Operand<T> beta, Operand<T> grad, Options... options) {
+  public static <T extends TType> ResourceApplyAddSign create(Scope scope, Operand<?> var, Operand<?> m, Operand<T> lr, Operand<T> alpha, Operand<T> signDecay, Operand<T> beta, Operand<T> grad, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ResourceApplyAddSign", scope.makeOpName("ResourceApplyAddSign"));
-    opBuilder.addInput(var.asOutput());
-    opBuilder.addInput(m.asOutput());
-    opBuilder.addInput(lr.asOutput());
-    opBuilder.addInput(alpha.asOutput());
-    opBuilder.addInput(signDecay.asOutput());
-    opBuilder.addInput(beta.asOutput());
-    opBuilder.addInput(grad.asOutput());
+    opBuilder.addInput(var.asOutput(scope));
+    opBuilder.addInput(m.asOutput(scope));
+    opBuilder.addInput(lr.asOutput(scope));
+    opBuilder.addInput(alpha.asOutput(scope));
+    opBuilder.addInput(signDecay.asOutput(scope));
+    opBuilder.addInput(beta.asOutput(scope));
+    opBuilder.addInput(grad.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyCenteredRmsProp.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyCenteredRmsProp.java
index e1499e57edd..683dbef6a0f 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyCenteredRmsProp.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyCenteredRmsProp.java
@@ -20,11 +20,11 @@
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Update '*var' according to the centered RMSProp algorithm.
@@ -89,17 +89,17 @@ private Options() {
    * @return a new instance of ResourceApplyCenteredRmsProp
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> ResourceApplyCenteredRmsProp create(Scope scope, Operand<?> var, Operand<?> mg, Operand<?> ms, Operand<?> mom, Operand<T> lr, Operand<T> rho, Operand<T> momentum, Operand<T> epsilon, Operand<T> grad, Options... options) {
+  public static <T extends TType> ResourceApplyCenteredRmsProp create(Scope scope, Operand<?> var, Operand<?> mg, Operand<?> ms, Operand<?> mom, Operand<T> lr, Operand<T> rho, Operand<T> momentum, Operand<T> epsilon, Operand<T> grad, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ResourceApplyCenteredRMSProp", scope.makeOpName("ResourceApplyCenteredRmsProp"));
-    opBuilder.addInput(var.asOutput());
-    opBuilder.addInput(mg.asOutput());
-    opBuilder.addInput(ms.asOutput());
-    opBuilder.addInput(mom.asOutput());
-    opBuilder.addInput(lr.asOutput());
-    opBuilder.addInput(rho.asOutput());
-    opBuilder.addInput(momentum.asOutput());
-    opBuilder.addInput(epsilon.asOutput());
-    opBuilder.addInput(grad.asOutput());
+    opBuilder.addInput(var.asOutput(scope));
+    opBuilder.addInput(mg.asOutput(scope));
+    opBuilder.addInput(ms.asOutput(scope));
+    opBuilder.addInput(mom.asOutput(scope));
+    opBuilder.addInput(lr.asOutput(scope));
+    opBuilder.addInput(rho.asOutput(scope));
+    opBuilder.addInput(momentum.asOutput(scope));
+    opBuilder.addInput(epsilon.asOutput(scope));
+    opBuilder.addInput(grad.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyFtrl.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyFtrl.java
index acb8879bc29..93d78c3a498 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyFtrl.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyFtrl.java
@@ -20,11 +20,11 @@
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Update '*var' according to the Ftrl-proximal scheme.
@@ -87,17 +87,17 @@ private Options() {
    * @return a new instance of ResourceApplyFtrl
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> ResourceApplyFtrl create(Scope scope, Operand<?> var, Operand<?> accum, Operand<?> linear, Operand<T> grad, Operand<T> lr, Operand<T> l1, Operand<T> l2, Operand<T> l2Shrinkage, Operand<T> lrPower, Options... options) {
+  public static <T extends TType> ResourceApplyFtrl create(Scope scope, Operand<?> var, Operand<?> accum, Operand<?> linear, Operand<T> grad, Operand<T> lr, Operand<T> l1, Operand<T> l2, Operand<T> l2Shrinkage, Operand<T> lrPower, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ResourceApplyFtrlV2", scope.makeOpName("ResourceApplyFtrl"));
-    opBuilder.addInput(var.asOutput());
-    opBuilder.addInput(accum.asOutput());
-    opBuilder.addInput(linear.asOutput());
-    opBuilder.addInput(grad.asOutput());
-    opBuilder.addInput(lr.asOutput());
-    opBuilder.addInput(l1.asOutput());
-    opBuilder.addInput(l2.asOutput());
-    opBuilder.addInput(l2Shrinkage.asOutput());
-    opBuilder.addInput(lrPower.asOutput());
+    opBuilder.addInput(var.asOutput(scope));
+    opBuilder.addInput(accum.asOutput(scope));
+    opBuilder.addInput(linear.asOutput(scope));
+    opBuilder.addInput(grad.asOutput(scope));
+    opBuilder.addInput(lr.asOutput(scope));
+    opBuilder.addInput(l1.asOutput(scope));
+    opBuilder.addInput(l2.asOutput(scope));
+    opBuilder.addInput(l2Shrinkage.asOutput(scope));
+    opBuilder.addInput(lrPower.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyGradientDescent.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyGradientDescent.java
index 701d8691eff..0098692cf89 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyGradientDescent.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyGradientDescent.java
@@ -20,11 +20,11 @@
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Update '*var' by subtracting 'alpha' * 'delta' from it.
@@ -63,11 +63,11 @@ private Options() {
    * @return a new instance of ResourceApplyGradientDescent
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> ResourceApplyGradientDescent create(Scope scope, Operand<?> var, Operand<T> alpha, Operand<T> delta, Options... options) {
+  public static <T extends TType> ResourceApplyGradientDescent create(Scope scope, Operand<?> var, Operand<T> alpha, Operand<T> delta, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ResourceApplyGradientDescent", scope.makeOpName("ResourceApplyGradientDescent"));
-    opBuilder.addInput(var.asOutput());
-    opBuilder.addInput(alpha.asOutput());
-    opBuilder.addInput(delta.asOutput());
+    opBuilder.addInput(var.asOutput(scope));
+    opBuilder.addInput(alpha.asOutput(scope));
+    opBuilder.addInput(delta.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyKerasMomentum.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyKerasMomentum.java
index bb00f90ece5..2f25475a048 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyKerasMomentum.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyKerasMomentum.java
@@ -20,11 +20,11 @@
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Update '*var' according to the momentum scheme.
@@ -82,13 +82,13 @@ private Options() {
    * @return a new instance of ResourceApplyKerasMomentum
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> ResourceApplyKerasMomentum create(Scope scope, Operand<?> var, Operand<?> accum, Operand<T> lr, Operand<T> grad, Operand<T> momentum, Options... options) {
+  public static <T extends TType> ResourceApplyKerasMomentum create(Scope scope, Operand<?> var, Operand<?> accum, Operand<T> lr, Operand<T> grad, Operand<T> momentum, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ResourceApplyKerasMomentum", scope.makeOpName("ResourceApplyKerasMomentum"));
-    opBuilder.addInput(var.asOutput());
-    opBuilder.addInput(accum.asOutput());
-    opBuilder.addInput(lr.asOutput());
-    opBuilder.addInput(grad.asOutput());
-    opBuilder.addInput(momentum.asOutput());
+    opBuilder.addInput(var.asOutput(scope));
+    opBuilder.addInput(accum.asOutput(scope));
+    opBuilder.addInput(lr.asOutput(scope));
+    opBuilder.addInput(grad.asOutput(scope));
+    opBuilder.addInput(momentum.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyMomentum.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyMomentum.java
index 4ab1f15ca92..b2e93b6dc17 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyMomentum.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyMomentum.java
@@ -20,11 +20,11 @@
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Update '*var' according to the momentum scheme.
@@ -82,13 +82,13 @@ private Options() {
    * @return a new instance of ResourceApplyMomentum
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> ResourceApplyMomentum create(Scope scope, Operand<?> var, Operand<?> accum, Operand<T> lr, Operand<T> grad, Operand<T> momentum, Options... options) {
+  public static <T extends TType> ResourceApplyMomentum create(Scope scope, Operand<?> var, Operand<?> accum, Operand<T> lr, Operand<T> grad, Operand<T> momentum, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ResourceApplyMomentum", scope.makeOpName("ResourceApplyMomentum"));
-    opBuilder.addInput(var.asOutput());
-    opBuilder.addInput(accum.asOutput());
-    opBuilder.addInput(lr.asOutput());
-    opBuilder.addInput(grad.asOutput());
-    opBuilder.addInput(momentum.asOutput());
+    opBuilder.addInput(var.asOutput(scope));
+    opBuilder.addInput(accum.asOutput(scope));
+    opBuilder.addInput(lr.asOutput(scope));
+    opBuilder.addInput(grad.asOutput(scope));
+    opBuilder.addInput(momentum.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyPowerSign.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyPowerSign.java
index a7444fe0fa5..b31285fb631 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyPowerSign.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyPowerSign.java
@@ -20,11 +20,11 @@
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Update '*var' according to the AddSign update.
@@ -72,15 +72,15 @@ private Options() {
    * @return a new instance of ResourceApplyPowerSign
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> ResourceApplyPowerSign create(Scope scope, Operand<?> var, Operand<?> m, Operand<T> lr, Operand<T> logbase, Operand<T> signDecay, Operand<T> beta, Operand<T> grad, Options... options) {
+  public static <T extends TType> ResourceApplyPowerSign create(Scope scope, Operand<?> var, Operand<?> m, Operand<T> lr, Operand<T> logbase, Operand<T> signDecay, Operand<T> beta, Operand<T> grad, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ResourceApplyPowerSign", scope.makeOpName("ResourceApplyPowerSign"));
-    opBuilder.addInput(var.asOutput());
-    opBuilder.addInput(m.asOutput());
-    opBuilder.addInput(lr.asOutput());
-    opBuilder.addInput(logbase.asOutput());
-    opBuilder.addInput(signDecay.asOutput());
-    opBuilder.addInput(beta.asOutput());
-    opBuilder.addInput(grad.asOutput());
+    opBuilder.addInput(var.asOutput(scope));
+    opBuilder.addInput(m.asOutput(scope));
+    opBuilder.addInput(lr.asOutput(scope));
+    opBuilder.addInput(logbase.asOutput(scope));
+    opBuilder.addInput(signDecay.asOutput(scope));
+    opBuilder.addInput(beta.asOutput(scope));
+    opBuilder.addInput(grad.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyProximalAdagrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyProximalAdagrad.java
index 53e885b6324..38528f4054b 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyProximalAdagrad.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyProximalAdagrad.java
@@ -20,11 +20,11 @@
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Update '*var' and '*accum' according to FOBOS with Adagrad learning rate.
@@ -70,14 +70,14 @@ private Options() {
    * @return a new instance of ResourceApplyProximalAdagrad
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> ResourceApplyProximalAdagrad create(Scope scope, Operand<?> var, Operand<?> accum, Operand<T> lr, Operand<T> l1, Operand<T> l2, Operand<T> grad, Options... options) {
+  public static <T extends TType> ResourceApplyProximalAdagrad create(Scope scope, Operand<?> var, Operand<?> accum, Operand<T> lr, Operand<T> l1, Operand<T> l2, Operand<T> grad, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ResourceApplyProximalAdagrad", scope.makeOpName("ResourceApplyProximalAdagrad"));
-    opBuilder.addInput(var.asOutput());
-    opBuilder.addInput(accum.asOutput());
-    opBuilder.addInput(lr.asOutput());
-    opBuilder.addInput(l1.asOutput());
-    opBuilder.addInput(l2.asOutput());
-    opBuilder.addInput(grad.asOutput());
+    opBuilder.addInput(var.asOutput(scope));
+    opBuilder.addInput(accum.asOutput(scope));
+    opBuilder.addInput(lr.asOutput(scope));
+    opBuilder.addInput(l1.asOutput(scope));
+    opBuilder.addInput(l2.asOutput(scope));
+    opBuilder.addInput(grad.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyProximalGradientDescent.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyProximalGradientDescent.java
index 9df791f64ce..bf781ae05f2 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyProximalGradientDescent.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyProximalGradientDescent.java
@@ -20,11 +20,11 @@
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Update '*var' as FOBOS algorithm with fixed learning rate.
@@ -68,13 +68,13 @@ private Options() {
    * @return a new instance of ResourceApplyProximalGradientDescent
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> ResourceApplyProximalGradientDescent create(Scope scope, Operand<?> var, Operand<T> alpha, Operand<T> l1, Operand<T> l2, Operand<T> delta, Options... options) {
+  public static <T extends TType> ResourceApplyProximalGradientDescent create(Scope scope, Operand<?> var, Operand<T> alpha, Operand<T> l1, Operand<T> l2, Operand<T> delta, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ResourceApplyProximalGradientDescent", scope.makeOpName("ResourceApplyProximalGradientDescent"));
-    opBuilder.addInput(var.asOutput());
-    opBuilder.addInput(alpha.asOutput());
-    opBuilder.addInput(l1.asOutput());
-    opBuilder.addInput(l2.asOutput());
-    opBuilder.addInput(delta.asOutput());
+    opBuilder.addInput(var.asOutput(scope));
+    opBuilder.addInput(alpha.asOutput(scope));
+    opBuilder.addInput(l1.asOutput(scope));
+    opBuilder.addInput(l2.asOutput(scope));
+    opBuilder.addInput(delta.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyRmsProp.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyRmsProp.java
index 302afd570d1..46ec2ea3ca3 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyRmsProp.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyRmsProp.java
@@ -20,11 +20,11 @@
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Update '*var' according to the RMSProp algorithm.
@@ -80,16 +80,16 @@ private Options() {
    * @return a new instance of ResourceApplyRmsProp
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> ResourceApplyRmsProp create(Scope scope, Operand<?> var, Operand<?> ms, Operand<?> mom, Operand<T> lr, Operand<T> rho, Operand<T> momentum, Operand<T> epsilon, Operand<T> grad, Options... options) {
+  public static <T extends TType> ResourceApplyRmsProp create(Scope scope, Operand<?> var, Operand<?> ms, Operand<?> mom, Operand<T> lr, Operand<T> rho, Operand<T> momentum, Operand<T> epsilon, Operand<T> grad, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ResourceApplyRMSProp", scope.makeOpName("ResourceApplyRmsProp"));
-    opBuilder.addInput(var.asOutput());
-    opBuilder.addInput(ms.asOutput());
-    opBuilder.addInput(mom.asOutput());
-    opBuilder.addInput(lr.asOutput());
-    opBuilder.addInput(rho.asOutput());
-    opBuilder.addInput(momentum.asOutput());
-    opBuilder.addInput(epsilon.asOutput());
-    opBuilder.addInput(grad.asOutput());
+    opBuilder.addInput(var.asOutput(scope));
+    opBuilder.addInput(ms.asOutput(scope));
+    opBuilder.addInput(mom.asOutput(scope));
+    opBuilder.addInput(lr.asOutput(scope));
+    opBuilder.addInput(rho.asOutput(scope));
+    opBuilder.addInput(momentum.asOutput(scope));
+    opBuilder.addInput(epsilon.asOutput(scope));
+    opBuilder.addInput(grad.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceConditionalAccumulator.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceConditionalAccumulator.java
index d03834634ed..47bb1bb3027 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceConditionalAccumulator.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceConditionalAccumulator.java
@@ -22,12 +22,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * A conditional accumulator for aggregating gradients.
@@ -41,7 +41,7 @@
  * This is a resource version of ConditionalAccumulator that will work in TF2.0
  * with tf.cond version 2.
  */
-public final class ResourceConditionalAccumulator extends RawOp implements Operand<Tensor> {
+public final class ResourceConditionalAccumulator extends RawOp implements Operand<TType> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.train.ResourceConditionalAccumulator}
@@ -92,7 +92,7 @@ private Options() {
    * @return a new instance of ResourceConditionalAccumulator
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> ResourceConditionalAccumulator create(Scope scope, DataType<T> dtype, Shape shape, Options... options) {
+  public static <T extends TType> ResourceConditionalAccumulator create(Scope scope, DataType<T> dtype, Shape shape, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ResourceConditionalAccumulator", scope.makeOpName("ResourceConditionalAccumulator"));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("dtype", dtype);
@@ -145,8 +145,8 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<Tensor> asOutput() {
-    return (Output<Tensor>) handle;
+  public Output<TType> asOutput(Scope scope) {
+    return (Output<TType>) handle;
   }
   
   /** The name of this op, as known by TensorFlow core engine */
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyAdadelta.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyAdadelta.java
index 78056617597..41813e0faac 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyAdadelta.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyAdadelta.java
@@ -20,12 +20,12 @@
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * var: Should be from a Variable().
@@ -69,16 +69,16 @@ private Options() {
    * @return a new instance of ResourceSparseApplyAdadelta
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor, U extends Tensor & TNumber> ResourceSparseApplyAdadelta create(Scope scope, Operand<?> var, Operand<?> accum, Operand<?> accumUpdate, Operand<T> lr, Operand<T> rho, Operand<T> epsilon, Operand<T> grad, Operand<U> indices, Options... options) {
+  public static <T extends TType, U extends TNumber> ResourceSparseApplyAdadelta create(Scope scope, Operand<?> var, Operand<?> accum, Operand<?> accumUpdate, Operand<T> lr, Operand<T> rho, Operand<T> epsilon, Operand<T> grad, Operand<U> indices, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ResourceSparseApplyAdadelta", scope.makeOpName("ResourceSparseApplyAdadelta"));
-    opBuilder.addInput(var.asOutput());
-    opBuilder.addInput(accum.asOutput());
-    opBuilder.addInput(accumUpdate.asOutput());
-    opBuilder.addInput(lr.asOutput());
-    opBuilder.addInput(rho.asOutput());
-    opBuilder.addInput(epsilon.asOutput());
-    opBuilder.addInput(grad.asOutput());
-    opBuilder.addInput(indices.asOutput());
+    opBuilder.addInput(var.asOutput(scope));
+    opBuilder.addInput(accum.asOutput(scope));
+    opBuilder.addInput(accumUpdate.asOutput(scope));
+    opBuilder.addInput(lr.asOutput(scope));
+    opBuilder.addInput(rho.asOutput(scope));
+    opBuilder.addInput(epsilon.asOutput(scope));
+    opBuilder.addInput(grad.asOutput(scope));
+    opBuilder.addInput(indices.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyAdagrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyAdagrad.java
index b04a01c5411..68a3a21b4a5 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyAdagrad.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyAdagrad.java
@@ -20,12 +20,12 @@
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * Update relevant entries in '*var' and '*accum' according to the adagrad scheme.
@@ -80,13 +80,13 @@ private Options() {
    * @return a new instance of ResourceSparseApplyAdagrad
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor, U extends Tensor & TNumber> ResourceSparseApplyAdagrad create(Scope scope, Operand<?> var, Operand<?> accum, Operand<T> lr, Operand<T> grad, Operand<U> indices, Options... options) {
+  public static <T extends TType, U extends TNumber> ResourceSparseApplyAdagrad create(Scope scope, Operand<?> var, Operand<?> accum, Operand<T> lr, Operand<T> grad, Operand<U> indices, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ResourceSparseApplyAdagrad", scope.makeOpName("ResourceSparseApplyAdagrad"));
-    opBuilder.addInput(var.asOutput());
-    opBuilder.addInput(accum.asOutput());
-    opBuilder.addInput(lr.asOutput());
-    opBuilder.addInput(grad.asOutput());
-    opBuilder.addInput(indices.asOutput());
+    opBuilder.addInput(var.asOutput(scope));
+    opBuilder.addInput(accum.asOutput(scope));
+    opBuilder.addInput(lr.asOutput(scope));
+    opBuilder.addInput(grad.asOutput(scope));
+    opBuilder.addInput(indices.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyAdagradDa.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyAdagradDa.java
index 64450eadd93..760e2cf1b47 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyAdagradDa.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyAdagradDa.java
@@ -20,13 +20,13 @@
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * Update entries in '*var' and '*accum' according to the proximal adagrad scheme.
@@ -71,17 +71,17 @@ private Options() {
    * @return a new instance of ResourceSparseApplyAdagradDa
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor, U extends Tensor & TNumber> ResourceSparseApplyAdagradDa create(Scope scope, Operand<?> var, Operand<?> gradientAccumulator, Operand<?> gradientSquaredAccumulator, Operand<T> grad, Operand<U> indices, Operand<T> lr, Operand<T> l1, Operand<T> l2, Operand<TInt64> globalStep, Options... options) {
+  public static <T extends TType, U extends TNumber> ResourceSparseApplyAdagradDa create(Scope scope, Operand<?> var, Operand<?> gradientAccumulator, Operand<?> gradientSquaredAccumulator, Operand<T> grad, Operand<U> indices, Operand<T> lr, Operand<T> l1, Operand<T> l2, Operand<TInt64> globalStep, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ResourceSparseApplyAdagradDA", scope.makeOpName("ResourceSparseApplyAdagradDa"));
-    opBuilder.addInput(var.asOutput());
-    opBuilder.addInput(gradientAccumulator.asOutput());
-    opBuilder.addInput(gradientSquaredAccumulator.asOutput());
-    opBuilder.addInput(grad.asOutput());
-    opBuilder.addInput(indices.asOutput());
-    opBuilder.addInput(lr.asOutput());
-    opBuilder.addInput(l1.asOutput());
-    opBuilder.addInput(l2.asOutput());
-    opBuilder.addInput(globalStep.asOutput());
+    opBuilder.addInput(var.asOutput(scope));
+    opBuilder.addInput(gradientAccumulator.asOutput(scope));
+    opBuilder.addInput(gradientSquaredAccumulator.asOutput(scope));
+    opBuilder.addInput(grad.asOutput(scope));
+    opBuilder.addInput(indices.asOutput(scope));
+    opBuilder.addInput(lr.asOutput(scope));
+    opBuilder.addInput(l1.asOutput(scope));
+    opBuilder.addInput(l2.asOutput(scope));
+    opBuilder.addInput(globalStep.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyAdagradV2.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyAdagradV2.java
index 8ea0ceb8390..7572b71cd56 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyAdagradV2.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyAdagradV2.java
@@ -20,12 +20,12 @@
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * Update relevant entries in '*var' and '*accum' according to the adagrad scheme.
@@ -80,14 +80,14 @@ private Options() {
    * @return a new instance of ResourceSparseApplyAdagradV2
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor, U extends Tensor & TNumber> ResourceSparseApplyAdagradV2 create(Scope scope, Operand<?> var, Operand<?> accum, Operand<T> lr, Operand<T> epsilon, Operand<T> grad, Operand<U> indices, Options... options) {
+  public static <T extends TType, U extends TNumber> ResourceSparseApplyAdagradV2 create(Scope scope, Operand<?> var, Operand<?> accum, Operand<T> lr, Operand<T> epsilon, Operand<T> grad, Operand<U> indices, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ResourceSparseApplyAdagradV2", scope.makeOpName("ResourceSparseApplyAdagradV2"));
-    opBuilder.addInput(var.asOutput());
-    opBuilder.addInput(accum.asOutput());
-    opBuilder.addInput(lr.asOutput());
-    opBuilder.addInput(epsilon.asOutput());
-    opBuilder.addInput(grad.asOutput());
-    opBuilder.addInput(indices.asOutput());
+    opBuilder.addInput(var.asOutput(scope));
+    opBuilder.addInput(accum.asOutput(scope));
+    opBuilder.addInput(lr.asOutput(scope));
+    opBuilder.addInput(epsilon.asOutput(scope));
+    opBuilder.addInput(grad.asOutput(scope));
+    opBuilder.addInput(indices.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyCenteredRmsProp.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyCenteredRmsProp.java
index a71f6524b61..a1015cadb11 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyCenteredRmsProp.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyCenteredRmsProp.java
@@ -20,12 +20,12 @@
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * Update '*var' according to the centered RMSProp algorithm.
@@ -89,18 +89,18 @@ private Options() {
    * @return a new instance of ResourceSparseApplyCenteredRmsProp
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor, U extends Tensor & TNumber> ResourceSparseApplyCenteredRmsProp create(Scope scope, Operand<?> var, Operand<?> mg, Operand<?> ms, Operand<?> mom, Operand<T> lr, Operand<T> rho, Operand<T> momentum, Operand<T> epsilon, Operand<T> grad, Operand<U> indices, Options... options) {
+  public static <T extends TType, U extends TNumber> ResourceSparseApplyCenteredRmsProp create(Scope scope, Operand<?> var, Operand<?> mg, Operand<?> ms, Operand<?> mom, Operand<T> lr, Operand<T> rho, Operand<T> momentum, Operand<T> epsilon, Operand<T> grad, Operand<U> indices, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ResourceSparseApplyCenteredRMSProp", scope.makeOpName("ResourceSparseApplyCenteredRmsProp"));
-    opBuilder.addInput(var.asOutput());
-    opBuilder.addInput(mg.asOutput());
-    opBuilder.addInput(ms.asOutput());
-    opBuilder.addInput(mom.asOutput());
-    opBuilder.addInput(lr.asOutput());
-    opBuilder.addInput(rho.asOutput());
-    opBuilder.addInput(momentum.asOutput());
-    opBuilder.addInput(epsilon.asOutput());
-    opBuilder.addInput(grad.asOutput());
-    opBuilder.addInput(indices.asOutput());
+    opBuilder.addInput(var.asOutput(scope));
+    opBuilder.addInput(mg.asOutput(scope));
+    opBuilder.addInput(ms.asOutput(scope));
+    opBuilder.addInput(mom.asOutput(scope));
+    opBuilder.addInput(lr.asOutput(scope));
+    opBuilder.addInput(rho.asOutput(scope));
+    opBuilder.addInput(momentum.asOutput(scope));
+    opBuilder.addInput(epsilon.asOutput(scope));
+    opBuilder.addInput(grad.asOutput(scope));
+    opBuilder.addInput(indices.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyFtrl.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyFtrl.java
index c6b83eb9091..a04ab7aab5b 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyFtrl.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyFtrl.java
@@ -20,12 +20,12 @@
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * Update relevant entries in '*var' according to the Ftrl-proximal scheme.
@@ -90,18 +90,18 @@ private Options() {
    * @return a new instance of ResourceSparseApplyFtrl
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor, U extends Tensor & TNumber> ResourceSparseApplyFtrl create(Scope scope, Operand<?> var, Operand<?> accum, Operand<?> linear, Operand<T> grad, Operand<U> indices, Operand<T> lr, Operand<T> l1, Operand<T> l2, Operand<T> l2Shrinkage, Operand<T> lrPower, Options... options) {
+  public static <T extends TType, U extends TNumber> ResourceSparseApplyFtrl create(Scope scope, Operand<?> var, Operand<?> accum, Operand<?> linear, Operand<T> grad, Operand<U> indices, Operand<T> lr, Operand<T> l1, Operand<T> l2, Operand<T> l2Shrinkage, Operand<T> lrPower, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ResourceSparseApplyFtrlV2", scope.makeOpName("ResourceSparseApplyFtrl"));
-    opBuilder.addInput(var.asOutput());
-    opBuilder.addInput(accum.asOutput());
-    opBuilder.addInput(linear.asOutput());
-    opBuilder.addInput(grad.asOutput());
-    opBuilder.addInput(indices.asOutput());
-    opBuilder.addInput(lr.asOutput());
-    opBuilder.addInput(l1.asOutput());
-    opBuilder.addInput(l2.asOutput());
-    opBuilder.addInput(l2Shrinkage.asOutput());
-    opBuilder.addInput(lrPower.asOutput());
+    opBuilder.addInput(var.asOutput(scope));
+    opBuilder.addInput(accum.asOutput(scope));
+    opBuilder.addInput(linear.asOutput(scope));
+    opBuilder.addInput(grad.asOutput(scope));
+    opBuilder.addInput(indices.asOutput(scope));
+    opBuilder.addInput(lr.asOutput(scope));
+    opBuilder.addInput(l1.asOutput(scope));
+    opBuilder.addInput(l2.asOutput(scope));
+    opBuilder.addInput(l2Shrinkage.asOutput(scope));
+    opBuilder.addInput(lrPower.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyKerasMomentum.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyKerasMomentum.java
index 648217b3d8b..05b6d72816c 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyKerasMomentum.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyKerasMomentum.java
@@ -20,12 +20,12 @@
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * Update relevant entries in '*var' and '*accum' according to the momentum scheme.
@@ -86,14 +86,14 @@ private Options() {
    * @return a new instance of ResourceSparseApplyKerasMomentum
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor, U extends Tensor & TNumber> ResourceSparseApplyKerasMomentum create(Scope scope, Operand<?> var, Operand<?> accum, Operand<T> lr, Operand<T> grad, Operand<U> indices, Operand<T> momentum, Options... options) {
+  public static <T extends TType, U extends TNumber> ResourceSparseApplyKerasMomentum create(Scope scope, Operand<?> var, Operand<?> accum, Operand<T> lr, Operand<T> grad, Operand<U> indices, Operand<T> momentum, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ResourceSparseApplyKerasMomentum", scope.makeOpName("ResourceSparseApplyKerasMomentum"));
-    opBuilder.addInput(var.asOutput());
-    opBuilder.addInput(accum.asOutput());
-    opBuilder.addInput(lr.asOutput());
-    opBuilder.addInput(grad.asOutput());
-    opBuilder.addInput(indices.asOutput());
-    opBuilder.addInput(momentum.asOutput());
+    opBuilder.addInput(var.asOutput(scope));
+    opBuilder.addInput(accum.asOutput(scope));
+    opBuilder.addInput(lr.asOutput(scope));
+    opBuilder.addInput(grad.asOutput(scope));
+    opBuilder.addInput(indices.asOutput(scope));
+    opBuilder.addInput(momentum.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyMomentum.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyMomentum.java
index 0b217881abf..160ddf01f60 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyMomentum.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyMomentum.java
@@ -20,12 +20,12 @@
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * Update relevant entries in '*var' and '*accum' according to the momentum scheme.
@@ -86,14 +86,14 @@ private Options() {
    * @return a new instance of ResourceSparseApplyMomentum
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor, U extends Tensor & TNumber> ResourceSparseApplyMomentum create(Scope scope, Operand<?> var, Operand<?> accum, Operand<T> lr, Operand<T> grad, Operand<U> indices, Operand<T> momentum, Options... options) {
+  public static <T extends TType, U extends TNumber> ResourceSparseApplyMomentum create(Scope scope, Operand<?> var, Operand<?> accum, Operand<T> lr, Operand<T> grad, Operand<U> indices, Operand<T> momentum, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ResourceSparseApplyMomentum", scope.makeOpName("ResourceSparseApplyMomentum"));
-    opBuilder.addInput(var.asOutput());
-    opBuilder.addInput(accum.asOutput());
-    opBuilder.addInput(lr.asOutput());
-    opBuilder.addInput(grad.asOutput());
-    opBuilder.addInput(indices.asOutput());
-    opBuilder.addInput(momentum.asOutput());
+    opBuilder.addInput(var.asOutput(scope));
+    opBuilder.addInput(accum.asOutput(scope));
+    opBuilder.addInput(lr.asOutput(scope));
+    opBuilder.addInput(grad.asOutput(scope));
+    opBuilder.addInput(indices.asOutput(scope));
+    opBuilder.addInput(momentum.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyProximalAdagrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyProximalAdagrad.java
index efbc4f836c2..af51baac80e 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyProximalAdagrad.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyProximalAdagrad.java
@@ -20,12 +20,12 @@
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * Sparse update entries in '*var' and '*accum' according to FOBOS algorithm.
@@ -74,15 +74,15 @@ private Options() {
    * @return a new instance of ResourceSparseApplyProximalAdagrad
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor, U extends Tensor & TNumber> ResourceSparseApplyProximalAdagrad create(Scope scope, Operand<?> var, Operand<?> accum, Operand<T> lr, Operand<T> l1, Operand<T> l2, Operand<T> grad, Operand<U> indices, Options... options) {
+  public static <T extends TType, U extends TNumber> ResourceSparseApplyProximalAdagrad create(Scope scope, Operand<?> var, Operand<?> accum, Operand<T> lr, Operand<T> l1, Operand<T> l2, Operand<T> grad, Operand<U> indices, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ResourceSparseApplyProximalAdagrad", scope.makeOpName("ResourceSparseApplyProximalAdagrad"));
-    opBuilder.addInput(var.asOutput());
-    opBuilder.addInput(accum.asOutput());
-    opBuilder.addInput(lr.asOutput());
-    opBuilder.addInput(l1.asOutput());
-    opBuilder.addInput(l2.asOutput());
-    opBuilder.addInput(grad.asOutput());
-    opBuilder.addInput(indices.asOutput());
+    opBuilder.addInput(var.asOutput(scope));
+    opBuilder.addInput(accum.asOutput(scope));
+    opBuilder.addInput(lr.asOutput(scope));
+    opBuilder.addInput(l1.asOutput(scope));
+    opBuilder.addInput(l2.asOutput(scope));
+    opBuilder.addInput(grad.asOutput(scope));
+    opBuilder.addInput(indices.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyProximalGradientDescent.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyProximalGradientDescent.java
index 77194576cfe..eda3f87707b 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyProximalGradientDescent.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyProximalGradientDescent.java
@@ -20,12 +20,12 @@
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * Sparse update '*var' as FOBOS algorithm with fixed learning rate.
@@ -71,14 +71,14 @@ private Options() {
    * @return a new instance of ResourceSparseApplyProximalGradientDescent
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor, U extends Tensor & TNumber> ResourceSparseApplyProximalGradientDescent create(Scope scope, Operand<?> var, Operand<T> alpha, Operand<T> l1, Operand<T> l2, Operand<T> grad, Operand<U> indices, Options... options) {
+  public static <T extends TType, U extends TNumber> ResourceSparseApplyProximalGradientDescent create(Scope scope, Operand<?> var, Operand<T> alpha, Operand<T> l1, Operand<T> l2, Operand<T> grad, Operand<U> indices, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ResourceSparseApplyProximalGradientDescent", scope.makeOpName("ResourceSparseApplyProximalGradientDescent"));
-    opBuilder.addInput(var.asOutput());
-    opBuilder.addInput(alpha.asOutput());
-    opBuilder.addInput(l1.asOutput());
-    opBuilder.addInput(l2.asOutput());
-    opBuilder.addInput(grad.asOutput());
-    opBuilder.addInput(indices.asOutput());
+    opBuilder.addInput(var.asOutput(scope));
+    opBuilder.addInput(alpha.asOutput(scope));
+    opBuilder.addInput(l1.asOutput(scope));
+    opBuilder.addInput(l2.asOutput(scope));
+    opBuilder.addInput(grad.asOutput(scope));
+    opBuilder.addInput(indices.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyRmsProp.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyRmsProp.java
index 76a1623708b..ae2678b3120 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyRmsProp.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyRmsProp.java
@@ -20,12 +20,12 @@
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * Update '*var' according to the RMSProp algorithm.
@@ -82,17 +82,17 @@ private Options() {
    * @return a new instance of ResourceSparseApplyRmsProp
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor, U extends Tensor & TNumber> ResourceSparseApplyRmsProp create(Scope scope, Operand<?> var, Operand<?> ms, Operand<?> mom, Operand<T> lr, Operand<T> rho, Operand<T> momentum, Operand<T> epsilon, Operand<T> grad, Operand<U> indices, Options... options) {
+  public static <T extends TType, U extends TNumber> ResourceSparseApplyRmsProp create(Scope scope, Operand<?> var, Operand<?> ms, Operand<?> mom, Operand<T> lr, Operand<T> rho, Operand<T> momentum, Operand<T> epsilon, Operand<T> grad, Operand<U> indices, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ResourceSparseApplyRMSProp", scope.makeOpName("ResourceSparseApplyRmsProp"));
-    opBuilder.addInput(var.asOutput());
-    opBuilder.addInput(ms.asOutput());
-    opBuilder.addInput(mom.asOutput());
-    opBuilder.addInput(lr.asOutput());
-    opBuilder.addInput(rho.asOutput());
-    opBuilder.addInput(momentum.asOutput());
-    opBuilder.addInput(epsilon.asOutput());
-    opBuilder.addInput(grad.asOutput());
-    opBuilder.addInput(indices.asOutput());
+    opBuilder.addInput(var.asOutput(scope));
+    opBuilder.addInput(ms.asOutput(scope));
+    opBuilder.addInput(mom.asOutput(scope));
+    opBuilder.addInput(lr.asOutput(scope));
+    opBuilder.addInput(rho.asOutput(scope));
+    opBuilder.addInput(momentum.asOutput(scope));
+    opBuilder.addInput(epsilon.asOutput(scope));
+    opBuilder.addInput(grad.asOutput(scope));
+    opBuilder.addInput(indices.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/Restore.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/Restore.java
index 29e595675f8..6d613546764 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/Restore.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/Restore.java
@@ -25,12 +25,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TString;
+import org.tensorflow.types.family.TType;
 
 /**
  * Restores tensors from a V2 checkpoint.
@@ -50,7 +50,7 @@
  * Callers must ensure all the named tensors are indeed stored in the checkpoint.
  */
 @Operator(group = "train")
-public final class Restore extends RawOp implements Iterable<Operand<Tensor>> {
+public final class Restore extends RawOp implements Iterable<Operand<TType>> {
   
   /**
    * Factory method to create a class wrapping a new Restore operation.
@@ -67,9 +67,9 @@ public final class Restore extends RawOp implements Iterable<Operand<Tensor>> {
   @Endpoint(describeByClass = true)
   public static Restore create(Scope scope, Operand<TString> prefix, Operand<TString> tensorNames, Operand<TString> shapeAndSlices, List<DataType<?>> dtypes) {
     OperationBuilder opBuilder = scope.env().opBuilder("RestoreV2", scope.makeOpName("Restore"));
-    opBuilder.addInput(prefix.asOutput());
-    opBuilder.addInput(tensorNames.asOutput());
-    opBuilder.addInput(shapeAndSlices.asOutput());
+    opBuilder.addInput(prefix.asOutput(scope));
+    opBuilder.addInput(tensorNames.asOutput(scope));
+    opBuilder.addInput(shapeAndSlices.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     DataType[] dtypesArray = new DataType[dtypes.size()];
     for (int i = 0; i < dtypesArray.length; ++i) {
@@ -89,7 +89,7 @@ public List<Output<?>> tensors() {
   
   @Override
   @SuppressWarnings({"rawtypes", "unchecked"})
-  public Iterator<Operand<Tensor>> iterator() {
+  public Iterator<Operand<TType>> iterator() {
     return (Iterator) tensors.iterator();
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/RestoreSlice.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/RestoreSlice.java
index 7a9e8382ca1..4e1cfc35405 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/RestoreSlice.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/RestoreSlice.java
@@ -22,12 +22,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TString;
+import org.tensorflow.types.family.TType;
 
 /**
  * Restores a tensor from checkpoint files.
@@ -42,7 +42,7 @@
  * @param <T> data type for {@code tensor()} output
  */
 @Operator(group = "train")
-public final class RestoreSlice<T extends Tensor> extends RawOp implements Operand<T> {
+public final class RestoreSlice<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.train.RestoreSlice}
@@ -79,11 +79,11 @@ private Options() {
    * @return a new instance of RestoreSlice
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> RestoreSlice<T> create(Scope scope, Operand<TString> filePattern, Operand<TString> tensorName, Operand<TString> shapeAndSlice, DataType<T> dt, Options... options) {
+  public static <T extends TType> RestoreSlice<T> create(Scope scope, Operand<TString> filePattern, Operand<TString> tensorName, Operand<TString> shapeAndSlice, DataType<T> dt, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("RestoreSlice", scope.makeOpName("RestoreSlice"));
-    opBuilder.addInput(filePattern.asOutput());
-    opBuilder.addInput(tensorName.asOutput());
-    opBuilder.addInput(shapeAndSlice.asOutput());
+    opBuilder.addInput(filePattern.asOutput(scope));
+    opBuilder.addInput(tensorName.asOutput(scope));
+    opBuilder.addInput(shapeAndSlice.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("dt", dt);
     if (options != null) {
@@ -112,7 +112,7 @@ public Output<T> tensor() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return tensor;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/Save.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/Save.java
index c5de40fc91b..21983a8fdc0 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/Save.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/Save.java
@@ -52,10 +52,10 @@ public final class Save extends RawOp {
   @Endpoint(describeByClass = true)
   public static Save create(Scope scope, Operand<TString> prefix, Operand<TString> tensorNames, Operand<TString> shapeAndSlices, Iterable<Operand<?>> tensors) {
     OperationBuilder opBuilder = scope.env().opBuilder("SaveV2", scope.makeOpName("Save"));
-    opBuilder.addInput(prefix.asOutput());
-    opBuilder.addInput(tensorNames.asOutput());
-    opBuilder.addInput(shapeAndSlices.asOutput());
-    opBuilder.addInputList(Operands.asOutputs(tensors));
+    opBuilder.addInput(prefix.asOutput(scope));
+    opBuilder.addInput(tensorNames.asOutput(scope));
+    opBuilder.addInput(shapeAndSlices.asOutput(scope));
+    opBuilder.addInputList(Operands.asOutputs(scope, tensors));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new Save(opBuilder.build());
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SaveSlices.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SaveSlices.java
index 73325d1d1bc..348f8d983c2 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SaveSlices.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SaveSlices.java
@@ -78,10 +78,10 @@ public final class SaveSlices extends RawOp {
   @Endpoint(describeByClass = true)
   public static SaveSlices create(Scope scope, Operand<TString> filename, Operand<TString> tensorNames, Operand<TString> shapesAndSlices, Iterable<Operand<?>> data) {
     OperationBuilder opBuilder = scope.env().opBuilder("SaveSlices", scope.makeOpName("SaveSlices"));
-    opBuilder.addInput(filename.asOutput());
-    opBuilder.addInput(tensorNames.asOutput());
-    opBuilder.addInput(shapesAndSlices.asOutput());
-    opBuilder.addInputList(Operands.asOutputs(data));
+    opBuilder.addInput(filename.asOutput(scope));
+    opBuilder.addInput(tensorNames.asOutput(scope));
+    opBuilder.addInput(shapesAndSlices.asOutput(scope));
+    opBuilder.addInputList(Operands.asOutputs(scope, data));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new SaveSlices(opBuilder.build());
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SdcaFprint.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SdcaFprint.java
index cdee1af05f5..b162a071713 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SdcaFprint.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SdcaFprint.java
@@ -44,7 +44,7 @@ public final class SdcaFprint extends RawOp implements Operand<TInt64> {
   @Endpoint(describeByClass = true)
   public static SdcaFprint create(Scope scope, Operand<TString> input) {
     OperationBuilder opBuilder = scope.env().opBuilder("SdcaFprint", scope.makeOpName("SdcaFprint"));
-    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new SdcaFprint(opBuilder.build());
   }
@@ -58,7 +58,7 @@ public Output<TInt64> output() {
   }
   
   @Override
-  public Output<TInt64> asOutput() {
+  public Output<TInt64> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SdcaOptimizer.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SdcaOptimizer.java
index a38f6773d7f..a97a4589134 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SdcaOptimizer.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SdcaOptimizer.java
@@ -106,16 +106,16 @@ private Options() {
   @Endpoint(describeByClass = true)
   public static SdcaOptimizer create(Scope scope, Iterable<Operand<TInt64>> sparseExampleIndices, Iterable<Operand<TInt64>> sparseFeatureIndices, Iterable<Operand<TFloat32>> sparseFeatureValues, Iterable<Operand<TFloat32>> denseFeatures, Operand<TFloat32> exampleWeights, Operand<TFloat32> exampleLabels, Iterable<Operand<TInt64>> sparseIndices, Iterable<Operand<TFloat32>> sparseWeights, Iterable<Operand<TFloat32>> denseWeights, Operand<TFloat32> exampleStateData, String lossType, Float l1, Float l2, Long numLossPartitions, Long numInnerIterations, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("SdcaOptimizerV2", scope.makeOpName("SdcaOptimizer"));
-    opBuilder.addInputList(Operands.asOutputs(sparseExampleIndices));
-    opBuilder.addInputList(Operands.asOutputs(sparseFeatureIndices));
-    opBuilder.addInputList(Operands.asOutputs(sparseFeatureValues));
-    opBuilder.addInputList(Operands.asOutputs(denseFeatures));
-    opBuilder.addInput(exampleWeights.asOutput());
-    opBuilder.addInput(exampleLabels.asOutput());
-    opBuilder.addInputList(Operands.asOutputs(sparseIndices));
-    opBuilder.addInputList(Operands.asOutputs(sparseWeights));
-    opBuilder.addInputList(Operands.asOutputs(denseWeights));
-    opBuilder.addInput(exampleStateData.asOutput());
+    opBuilder.addInputList(Operands.asOutputs(scope, sparseExampleIndices));
+    opBuilder.addInputList(Operands.asOutputs(scope, sparseFeatureIndices));
+    opBuilder.addInputList(Operands.asOutputs(scope, sparseFeatureValues));
+    opBuilder.addInputList(Operands.asOutputs(scope, denseFeatures));
+    opBuilder.addInput(exampleWeights.asOutput(scope));
+    opBuilder.addInput(exampleLabels.asOutput(scope));
+    opBuilder.addInputList(Operands.asOutputs(scope, sparseIndices));
+    opBuilder.addInputList(Operands.asOutputs(scope, sparseWeights));
+    opBuilder.addInputList(Operands.asOutputs(scope, denseWeights));
+    opBuilder.addInput(exampleStateData.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("loss_type", lossType);
     opBuilder.setAttr("l1", l1);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SdcaShrinkL1.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SdcaShrinkL1.java
index 748a2eacaec..54fad2cdcf6 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SdcaShrinkL1.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SdcaShrinkL1.java
@@ -46,7 +46,7 @@ public final class SdcaShrinkL1 extends RawOp {
   @Endpoint(describeByClass = true)
   public static SdcaShrinkL1 create(Scope scope, Iterable<Operand<TFloat32>> weights, Float l1, Float l2) {
     OperationBuilder opBuilder = scope.env().opBuilder("SdcaShrinkL1", scope.makeOpName("SdcaShrinkL1"));
-    opBuilder.addInputList(Operands.asOutputs(weights));
+    opBuilder.addInputList(Operands.asOutputs(scope, weights));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("l1", l1);
     opBuilder.setAttr("l2", l2);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyAdadelta.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyAdadelta.java
index 4cf10cb7fcd..5f44aca4160 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyAdadelta.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyAdadelta.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * var: Should be from a Variable().
@@ -34,7 +34,7 @@
  * @param <T> data type for {@code out()} output
  */
 @Operator(group = "train")
-public final class SparseApplyAdadelta<T extends Tensor> extends RawOp implements Operand<T> {
+public final class SparseApplyAdadelta<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.train.SparseApplyAdadelta}
@@ -72,16 +72,16 @@ private Options() {
    * @return a new instance of SparseApplyAdadelta
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor, U extends Tensor & TNumber> SparseApplyAdadelta<T> create(Scope scope, Operand<T> var, Operand<T> accum, Operand<T> accumUpdate, Operand<T> lr, Operand<T> rho, Operand<T> epsilon, Operand<T> grad, Operand<U> indices, Options... options) {
+  public static <T extends TType, U extends TNumber> SparseApplyAdadelta<T> create(Scope scope, Operand<T> var, Operand<T> accum, Operand<T> accumUpdate, Operand<T> lr, Operand<T> rho, Operand<T> epsilon, Operand<T> grad, Operand<U> indices, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("SparseApplyAdadelta", scope.makeOpName("SparseApplyAdadelta"));
-    opBuilder.addInput(var.asOutput());
-    opBuilder.addInput(accum.asOutput());
-    opBuilder.addInput(accumUpdate.asOutput());
-    opBuilder.addInput(lr.asOutput());
-    opBuilder.addInput(rho.asOutput());
-    opBuilder.addInput(epsilon.asOutput());
-    opBuilder.addInput(grad.asOutput());
-    opBuilder.addInput(indices.asOutput());
+    opBuilder.addInput(var.asOutput(scope));
+    opBuilder.addInput(accum.asOutput(scope));
+    opBuilder.addInput(accumUpdate.asOutput(scope));
+    opBuilder.addInput(lr.asOutput(scope));
+    opBuilder.addInput(rho.asOutput(scope));
+    opBuilder.addInput(epsilon.asOutput(scope));
+    opBuilder.addInput(grad.asOutput(scope));
+    opBuilder.addInput(indices.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -109,7 +109,7 @@ public Output<T> out() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return out;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyAdagrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyAdagrad.java
index 8745fd9f4ae..a4c0eeb942a 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyAdagrad.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyAdagrad.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * Update relevant entries in '*var' and '*accum' according to the adagrad scheme.
@@ -37,7 +37,7 @@
  * 
  * @param <T> data type for {@code out()} output
  */
-public final class SparseApplyAdagrad<T extends Tensor> extends RawOp implements Operand<T> {
+public final class SparseApplyAdagrad<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.train.SparseApplyAdagrad}
@@ -83,14 +83,14 @@ private Options() {
    * @return a new instance of SparseApplyAdagrad
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor, U extends Tensor & TNumber> SparseApplyAdagrad<T> create(Scope scope, Operand<T> var, Operand<T> accum, Operand<T> lr, Operand<T> epsilon, Operand<T> grad, Operand<U> indices, Options... options) {
+  public static <T extends TType, U extends TNumber> SparseApplyAdagrad<T> create(Scope scope, Operand<T> var, Operand<T> accum, Operand<T> lr, Operand<T> epsilon, Operand<T> grad, Operand<U> indices, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("SparseApplyAdagradV2", scope.makeOpName("SparseApplyAdagrad"));
-    opBuilder.addInput(var.asOutput());
-    opBuilder.addInput(accum.asOutput());
-    opBuilder.addInput(lr.asOutput());
-    opBuilder.addInput(epsilon.asOutput());
-    opBuilder.addInput(grad.asOutput());
-    opBuilder.addInput(indices.asOutput());
+    opBuilder.addInput(var.asOutput(scope));
+    opBuilder.addInput(accum.asOutput(scope));
+    opBuilder.addInput(lr.asOutput(scope));
+    opBuilder.addInput(epsilon.asOutput(scope));
+    opBuilder.addInput(grad.asOutput(scope));
+    opBuilder.addInput(indices.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -129,7 +129,7 @@ public Output<T> out() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return out;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyAdagradDa.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyAdagradDa.java
index e2952193b28..6264bfaadd7 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyAdagradDa.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyAdagradDa.java
@@ -21,13 +21,13 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * Update entries in '*var' and '*accum' according to the proximal adagrad scheme.
@@ -35,7 +35,7 @@
  * @param <T> data type for {@code out()} output
  */
 @Operator(group = "train")
-public final class SparseApplyAdagradDa<T extends Tensor> extends RawOp implements Operand<T> {
+public final class SparseApplyAdagradDa<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.train.SparseApplyAdagradDa}
@@ -74,17 +74,17 @@ private Options() {
    * @return a new instance of SparseApplyAdagradDa
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor, U extends Tensor & TNumber> SparseApplyAdagradDa<T> create(Scope scope, Operand<T> var, Operand<T> gradientAccumulator, Operand<T> gradientSquaredAccumulator, Operand<T> grad, Operand<U> indices, Operand<T> lr, Operand<T> l1, Operand<T> l2, Operand<TInt64> globalStep, Options... options) {
+  public static <T extends TType, U extends TNumber> SparseApplyAdagradDa<T> create(Scope scope, Operand<T> var, Operand<T> gradientAccumulator, Operand<T> gradientSquaredAccumulator, Operand<T> grad, Operand<U> indices, Operand<T> lr, Operand<T> l1, Operand<T> l2, Operand<TInt64> globalStep, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("SparseApplyAdagradDA", scope.makeOpName("SparseApplyAdagradDa"));
-    opBuilder.addInput(var.asOutput());
-    opBuilder.addInput(gradientAccumulator.asOutput());
-    opBuilder.addInput(gradientSquaredAccumulator.asOutput());
-    opBuilder.addInput(grad.asOutput());
-    opBuilder.addInput(indices.asOutput());
-    opBuilder.addInput(lr.asOutput());
-    opBuilder.addInput(l1.asOutput());
-    opBuilder.addInput(l2.asOutput());
-    opBuilder.addInput(globalStep.asOutput());
+    opBuilder.addInput(var.asOutput(scope));
+    opBuilder.addInput(gradientAccumulator.asOutput(scope));
+    opBuilder.addInput(gradientSquaredAccumulator.asOutput(scope));
+    opBuilder.addInput(grad.asOutput(scope));
+    opBuilder.addInput(indices.asOutput(scope));
+    opBuilder.addInput(lr.asOutput(scope));
+    opBuilder.addInput(l1.asOutput(scope));
+    opBuilder.addInput(l2.asOutput(scope));
+    opBuilder.addInput(globalStep.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -112,7 +112,7 @@ public Output<T> out() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return out;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyCenteredRmsProp.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyCenteredRmsProp.java
index ca7cab533b3..8b4d0d9d78c 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyCenteredRmsProp.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyCenteredRmsProp.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * Update '*var' according to the centered RMSProp algorithm.
@@ -51,7 +51,7 @@
  * @param <T> data type for {@code out()} output
  */
 @Operator(group = "train")
-public final class SparseApplyCenteredRmsProp<T extends Tensor> extends RawOp implements Operand<T> {
+public final class SparseApplyCenteredRmsProp<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.train.SparseApplyCenteredRmsProp}
@@ -92,18 +92,18 @@ private Options() {
    * @return a new instance of SparseApplyCenteredRmsProp
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor, U extends Tensor & TNumber> SparseApplyCenteredRmsProp<T> create(Scope scope, Operand<T> var, Operand<T> mg, Operand<T> ms, Operand<T> mom, Operand<T> lr, Operand<T> rho, Operand<T> momentum, Operand<T> epsilon, Operand<T> grad, Operand<U> indices, Options... options) {
+  public static <T extends TType, U extends TNumber> SparseApplyCenteredRmsProp<T> create(Scope scope, Operand<T> var, Operand<T> mg, Operand<T> ms, Operand<T> mom, Operand<T> lr, Operand<T> rho, Operand<T> momentum, Operand<T> epsilon, Operand<T> grad, Operand<U> indices, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("SparseApplyCenteredRMSProp", scope.makeOpName("SparseApplyCenteredRmsProp"));
-    opBuilder.addInput(var.asOutput());
-    opBuilder.addInput(mg.asOutput());
-    opBuilder.addInput(ms.asOutput());
-    opBuilder.addInput(mom.asOutput());
-    opBuilder.addInput(lr.asOutput());
-    opBuilder.addInput(rho.asOutput());
-    opBuilder.addInput(momentum.asOutput());
-    opBuilder.addInput(epsilon.asOutput());
-    opBuilder.addInput(grad.asOutput());
-    opBuilder.addInput(indices.asOutput());
+    opBuilder.addInput(var.asOutput(scope));
+    opBuilder.addInput(mg.asOutput(scope));
+    opBuilder.addInput(ms.asOutput(scope));
+    opBuilder.addInput(mom.asOutput(scope));
+    opBuilder.addInput(lr.asOutput(scope));
+    opBuilder.addInput(rho.asOutput(scope));
+    opBuilder.addInput(momentum.asOutput(scope));
+    opBuilder.addInput(epsilon.asOutput(scope));
+    opBuilder.addInput(grad.asOutput(scope));
+    opBuilder.addInput(indices.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -132,7 +132,7 @@ public Output<T> out() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return out;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyFtrl.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyFtrl.java
index 44967501a85..8102aea01ee 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyFtrl.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyFtrl.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * Update relevant entries in '*var' according to the Ftrl-proximal scheme.
@@ -43,7 +43,7 @@
  * @param <T> data type for {@code out()} output
  */
 @Operator(group = "train")
-public final class SparseApplyFtrl<T extends Tensor> extends RawOp implements Operand<T> {
+public final class SparseApplyFtrl<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.train.SparseApplyFtrl}
@@ -93,18 +93,18 @@ private Options() {
    * @return a new instance of SparseApplyFtrl
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor, U extends Tensor & TNumber> SparseApplyFtrl<T> create(Scope scope, Operand<T> var, Operand<T> accum, Operand<T> linear, Operand<T> grad, Operand<U> indices, Operand<T> lr, Operand<T> l1, Operand<T> l2, Operand<T> l2Shrinkage, Operand<T> lrPower, Options... options) {
+  public static <T extends TType, U extends TNumber> SparseApplyFtrl<T> create(Scope scope, Operand<T> var, Operand<T> accum, Operand<T> linear, Operand<T> grad, Operand<U> indices, Operand<T> lr, Operand<T> l1, Operand<T> l2, Operand<T> l2Shrinkage, Operand<T> lrPower, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("SparseApplyFtrlV2", scope.makeOpName("SparseApplyFtrl"));
-    opBuilder.addInput(var.asOutput());
-    opBuilder.addInput(accum.asOutput());
-    opBuilder.addInput(linear.asOutput());
-    opBuilder.addInput(grad.asOutput());
-    opBuilder.addInput(indices.asOutput());
-    opBuilder.addInput(lr.asOutput());
-    opBuilder.addInput(l1.asOutput());
-    opBuilder.addInput(l2.asOutput());
-    opBuilder.addInput(l2Shrinkage.asOutput());
-    opBuilder.addInput(lrPower.asOutput());
+    opBuilder.addInput(var.asOutput(scope));
+    opBuilder.addInput(accum.asOutput(scope));
+    opBuilder.addInput(linear.asOutput(scope));
+    opBuilder.addInput(grad.asOutput(scope));
+    opBuilder.addInput(indices.asOutput(scope));
+    opBuilder.addInput(lr.asOutput(scope));
+    opBuilder.addInput(l1.asOutput(scope));
+    opBuilder.addInput(l2.asOutput(scope));
+    opBuilder.addInput(l2Shrinkage.asOutput(scope));
+    opBuilder.addInput(lrPower.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -143,7 +143,7 @@ public Output<T> out() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return out;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyMomentum.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyMomentum.java
index 65de971b0ee..84068e2d518 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyMomentum.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyMomentum.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * Update relevant entries in '*var' and '*accum' according to the momentum scheme.
@@ -41,7 +41,7 @@
  * @param <T> data type for {@code out()} output
  */
 @Operator(group = "train")
-public final class SparseApplyMomentum<T extends Tensor> extends RawOp implements Operand<T> {
+public final class SparseApplyMomentum<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.train.SparseApplyMomentum}
@@ -89,14 +89,14 @@ private Options() {
    * @return a new instance of SparseApplyMomentum
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor, U extends Tensor & TNumber> SparseApplyMomentum<T> create(Scope scope, Operand<T> var, Operand<T> accum, Operand<T> lr, Operand<T> grad, Operand<U> indices, Operand<T> momentum, Options... options) {
+  public static <T extends TType, U extends TNumber> SparseApplyMomentum<T> create(Scope scope, Operand<T> var, Operand<T> accum, Operand<T> lr, Operand<T> grad, Operand<U> indices, Operand<T> momentum, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("SparseApplyMomentum", scope.makeOpName("SparseApplyMomentum"));
-    opBuilder.addInput(var.asOutput());
-    opBuilder.addInput(accum.asOutput());
-    opBuilder.addInput(lr.asOutput());
-    opBuilder.addInput(grad.asOutput());
-    opBuilder.addInput(indices.asOutput());
-    opBuilder.addInput(momentum.asOutput());
+    opBuilder.addInput(var.asOutput(scope));
+    opBuilder.addInput(accum.asOutput(scope));
+    opBuilder.addInput(lr.asOutput(scope));
+    opBuilder.addInput(grad.asOutput(scope));
+    opBuilder.addInput(indices.asOutput(scope));
+    opBuilder.addInput(momentum.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -137,7 +137,7 @@ public Output<T> out() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return out;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyProximalAdagrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyProximalAdagrad.java
index b2f420a9922..17933470a49 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyProximalAdagrad.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyProximalAdagrad.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * Sparse update entries in '*var' and '*accum' according to FOBOS algorithm.
@@ -40,7 +40,7 @@
  * @param <T> data type for {@code out()} output
  */
 @Operator(group = "train")
-public final class SparseApplyProximalAdagrad<T extends Tensor> extends RawOp implements Operand<T> {
+public final class SparseApplyProximalAdagrad<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.train.SparseApplyProximalAdagrad}
@@ -77,15 +77,15 @@ private Options() {
    * @return a new instance of SparseApplyProximalAdagrad
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor, U extends Tensor & TNumber> SparseApplyProximalAdagrad<T> create(Scope scope, Operand<T> var, Operand<T> accum, Operand<T> lr, Operand<T> l1, Operand<T> l2, Operand<T> grad, Operand<U> indices, Options... options) {
+  public static <T extends TType, U extends TNumber> SparseApplyProximalAdagrad<T> create(Scope scope, Operand<T> var, Operand<T> accum, Operand<T> lr, Operand<T> l1, Operand<T> l2, Operand<T> grad, Operand<U> indices, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("SparseApplyProximalAdagrad", scope.makeOpName("SparseApplyProximalAdagrad"));
-    opBuilder.addInput(var.asOutput());
-    opBuilder.addInput(accum.asOutput());
-    opBuilder.addInput(lr.asOutput());
-    opBuilder.addInput(l1.asOutput());
-    opBuilder.addInput(l2.asOutput());
-    opBuilder.addInput(grad.asOutput());
-    opBuilder.addInput(indices.asOutput());
+    opBuilder.addInput(var.asOutput(scope));
+    opBuilder.addInput(accum.asOutput(scope));
+    opBuilder.addInput(lr.asOutput(scope));
+    opBuilder.addInput(l1.asOutput(scope));
+    opBuilder.addInput(l2.asOutput(scope));
+    opBuilder.addInput(grad.asOutput(scope));
+    opBuilder.addInput(indices.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -113,7 +113,7 @@ public Output<T> out() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return out;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyProximalGradientDescent.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyProximalGradientDescent.java
index 8e4a87f1b6f..5f3a6776c79 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyProximalGradientDescent.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyProximalGradientDescent.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * Sparse update '*var' as FOBOS algorithm with fixed learning rate.
@@ -38,7 +38,7 @@
  * @param <T> data type for {@code out()} output
  */
 @Operator(group = "train")
-public final class SparseApplyProximalGradientDescent<T extends Tensor> extends RawOp implements Operand<T> {
+public final class SparseApplyProximalGradientDescent<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.train.SparseApplyProximalGradientDescent}
@@ -74,14 +74,14 @@ private Options() {
    * @return a new instance of SparseApplyProximalGradientDescent
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor, U extends Tensor & TNumber> SparseApplyProximalGradientDescent<T> create(Scope scope, Operand<T> var, Operand<T> alpha, Operand<T> l1, Operand<T> l2, Operand<T> grad, Operand<U> indices, Options... options) {
+  public static <T extends TType, U extends TNumber> SparseApplyProximalGradientDescent<T> create(Scope scope, Operand<T> var, Operand<T> alpha, Operand<T> l1, Operand<T> l2, Operand<T> grad, Operand<U> indices, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("SparseApplyProximalGradientDescent", scope.makeOpName("SparseApplyProximalGradientDescent"));
-    opBuilder.addInput(var.asOutput());
-    opBuilder.addInput(alpha.asOutput());
-    opBuilder.addInput(l1.asOutput());
-    opBuilder.addInput(l2.asOutput());
-    opBuilder.addInput(grad.asOutput());
-    opBuilder.addInput(indices.asOutput());
+    opBuilder.addInput(var.asOutput(scope));
+    opBuilder.addInput(alpha.asOutput(scope));
+    opBuilder.addInput(l1.asOutput(scope));
+    opBuilder.addInput(l2.asOutput(scope));
+    opBuilder.addInput(grad.asOutput(scope));
+    opBuilder.addInput(indices.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -109,7 +109,7 @@ public Output<T> out() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return out;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyRmsProp.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyRmsProp.java
index 3528d012eb1..a56246ca83f 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyRmsProp.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyRmsProp.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * Update '*var' according to the RMSProp algorithm.
@@ -45,7 +45,7 @@
  * @param <T> data type for {@code out()} output
  */
 @Operator(group = "train")
-public final class SparseApplyRmsProp<T extends Tensor> extends RawOp implements Operand<T> {
+public final class SparseApplyRmsProp<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Optional attributes for {@link org.tensorflow.op.train.SparseApplyRmsProp}
@@ -85,17 +85,17 @@ private Options() {
    * @return a new instance of SparseApplyRmsProp
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor, U extends Tensor & TNumber> SparseApplyRmsProp<T> create(Scope scope, Operand<T> var, Operand<T> ms, Operand<T> mom, Operand<T> lr, Operand<T> rho, Operand<T> momentum, Operand<T> epsilon, Operand<T> grad, Operand<U> indices, Options... options) {
+  public static <T extends TType, U extends TNumber> SparseApplyRmsProp<T> create(Scope scope, Operand<T> var, Operand<T> ms, Operand<T> mom, Operand<T> lr, Operand<T> rho, Operand<T> momentum, Operand<T> epsilon, Operand<T> grad, Operand<U> indices, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("SparseApplyRMSProp", scope.makeOpName("SparseApplyRmsProp"));
-    opBuilder.addInput(var.asOutput());
-    opBuilder.addInput(ms.asOutput());
-    opBuilder.addInput(mom.asOutput());
-    opBuilder.addInput(lr.asOutput());
-    opBuilder.addInput(rho.asOutput());
-    opBuilder.addInput(momentum.asOutput());
-    opBuilder.addInput(epsilon.asOutput());
-    opBuilder.addInput(grad.asOutput());
-    opBuilder.addInput(indices.asOutput());
+    opBuilder.addInput(var.asOutput(scope));
+    opBuilder.addInput(ms.asOutput(scope));
+    opBuilder.addInput(mom.asOutput(scope));
+    opBuilder.addInput(lr.asOutput(scope));
+    opBuilder.addInput(rho.asOutput(scope));
+    opBuilder.addInput(momentum.asOutput(scope));
+    opBuilder.addInput(epsilon.asOutput(scope));
+    opBuilder.addInput(grad.asOutput(scope));
+    opBuilder.addInput(indices.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -124,7 +124,7 @@ public Output<T> out() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return out;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/TileGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/TileGrad.java
index 46a7eac1e27..31b6a1dd486 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/TileGrad.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/TileGrad.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TType;
 
 /**
  * Returns the gradient of `Tile`.
@@ -38,7 +38,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "train")
-public final class TileGrad<T extends Tensor> extends RawOp implements Operand<T> {
+public final class TileGrad<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new TileGrad operation.
@@ -49,10 +49,10 @@ public final class TileGrad<T extends Tensor> extends RawOp implements Operand<T
    * @return a new instance of TileGrad
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> TileGrad<T> create(Scope scope, Operand<T> input, Operand<TInt32> multiples) {
+  public static <T extends TType> TileGrad<T> create(Scope scope, Operand<T> input, Operand<TInt32> multiples) {
     OperationBuilder opBuilder = scope.env().opBuilder("TileGrad", scope.makeOpName("TileGrad"));
-    opBuilder.addInput(input.asOutput());
-    opBuilder.addInput(multiples.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
+    opBuilder.addInput(multiples.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new TileGrad<T>(opBuilder.build());
   }
@@ -64,7 +64,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/BroadcastHelper.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/BroadcastHelper.java
index 770f551d5ae..59c75e98fa2 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/BroadcastHelper.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/BroadcastHelper.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * Helper operator for performing XLA-style broadcasts
@@ -38,7 +38,7 @@
  * @param <T> data type for {@code lhsOutput()} output
  */
 @Operator(group = "xla")
-public final class BroadcastHelper<T extends Tensor> extends RawOp {
+public final class BroadcastHelper<T extends TType> extends RawOp {
   
   /**
    * Factory method to create a class wrapping a new BroadcastHelper operation.
@@ -50,11 +50,11 @@ public final class BroadcastHelper<T extends Tensor> extends RawOp {
    * @return a new instance of BroadcastHelper
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor, U extends Tensor & TNumber> BroadcastHelper<T> create(Scope scope, Operand<T> lhs, Operand<T> rhs, Operand<U> broadcastDims) {
+  public static <T extends TType, U extends TNumber> BroadcastHelper<T> create(Scope scope, Operand<T> lhs, Operand<T> rhs, Operand<U> broadcastDims) {
     OperationBuilder opBuilder = scope.env().opBuilder("XlaBroadcastHelper", scope.makeOpName("BroadcastHelper"));
-    opBuilder.addInput(lhs.asOutput());
-    opBuilder.addInput(rhs.asOutput());
-    opBuilder.addInput(broadcastDims.asOutput());
+    opBuilder.addInput(lhs.asOutput(scope));
+    opBuilder.addInput(rhs.asOutput(scope));
+    opBuilder.addInput(broadcastDims.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new BroadcastHelper<T>(opBuilder.build());
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/ClusterOutput.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/ClusterOutput.java
index 1a9b7fa18f4..e9219f14b2a 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/ClusterOutput.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/ClusterOutput.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Operator that connects the output of an XLA computation to other consumer graph nodes.
@@ -33,7 +33,7 @@
  * @param <T> data type for {@code outputs()} output
  */
 @Operator(group = "xla")
-public final class ClusterOutput<T extends Tensor> extends RawOp implements Operand<T> {
+public final class ClusterOutput<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new ClusterOutput operation.
@@ -43,9 +43,9 @@ public final class ClusterOutput<T extends Tensor> extends RawOp implements Oper
    * @return a new instance of ClusterOutput
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> ClusterOutput<T> create(Scope scope, Operand<T> input) {
+  public static <T extends TType> ClusterOutput<T> create(Scope scope, Operand<T> input) {
     OperationBuilder opBuilder = scope.env().opBuilder("XlaClusterOutput", scope.makeOpName("ClusterOutput"));
-    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new ClusterOutput<T>(opBuilder.build());
   }
@@ -57,7 +57,7 @@ public Output<T> outputs() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return outputs;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Conv.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Conv.java
index e2d53e58b2f..b6b94c44624 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Conv.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Conv.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * Wraps the XLA ConvGeneralDilated operator, documented at
@@ -37,7 +37,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "xla")
-public final class Conv<T extends Tensor> extends RawOp implements Operand<T> {
+public final class Conv<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Conv operation.
@@ -55,15 +55,15 @@ public final class Conv<T extends Tensor> extends RawOp implements Operand<T> {
    * @return a new instance of Conv
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor, U extends Tensor & TNumber> Conv<T> create(Scope scope, Operand<T> lhs, Operand<T> rhs, Operand<U> windowStrides, Operand<U> padding, Operand<U> lhsDilation, Operand<U> rhsDilation, Operand<U> featureGroupCount, String dimensionNumbers, String precisionConfig) {
+  public static <T extends TType, U extends TNumber> Conv<T> create(Scope scope, Operand<T> lhs, Operand<T> rhs, Operand<U> windowStrides, Operand<U> padding, Operand<U> lhsDilation, Operand<U> rhsDilation, Operand<U> featureGroupCount, String dimensionNumbers, String precisionConfig) {
     OperationBuilder opBuilder = scope.env().opBuilder("XlaConv", scope.makeOpName("Conv"));
-    opBuilder.addInput(lhs.asOutput());
-    opBuilder.addInput(rhs.asOutput());
-    opBuilder.addInput(windowStrides.asOutput());
-    opBuilder.addInput(padding.asOutput());
-    opBuilder.addInput(lhsDilation.asOutput());
-    opBuilder.addInput(rhsDilation.asOutput());
-    opBuilder.addInput(featureGroupCount.asOutput());
+    opBuilder.addInput(lhs.asOutput(scope));
+    opBuilder.addInput(rhs.asOutput(scope));
+    opBuilder.addInput(windowStrides.asOutput(scope));
+    opBuilder.addInput(padding.asOutput(scope));
+    opBuilder.addInput(lhsDilation.asOutput(scope));
+    opBuilder.addInput(rhsDilation.asOutput(scope));
+    opBuilder.addInput(featureGroupCount.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("dimension_numbers", dimensionNumbers);
     opBuilder.setAttr("precision_config", precisionConfig);
@@ -77,7 +77,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Dequantize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Dequantize.java
index f8cdec66d1f..9cc4ca5a4f9 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Dequantize.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Dequantize.java
@@ -50,7 +50,7 @@ public final class Dequantize extends RawOp implements Operand<TBfloat16> {
   @Endpoint(describeByClass = true)
   public static Dequantize create(Scope scope, Operand<?> input, Float minRange, Float maxRange, String mode, Boolean transposeOutput) {
     OperationBuilder opBuilder = scope.env().opBuilder("XlaDequantize", scope.makeOpName("Dequantize"));
-    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("min_range", minRange);
     opBuilder.setAttr("max_range", maxRange);
@@ -69,7 +69,7 @@ public Output<TBfloat16> output() {
   }
   
   @Override
-  public Output<TBfloat16> asOutput() {
+  public Output<TBfloat16> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Dot.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Dot.java
index e4cb77489a6..ee69f7bccac 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Dot.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Dot.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Wraps the XLA DotGeneral operator, documented at
@@ -36,7 +36,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "xla")
-public final class Dot<T extends Tensor> extends RawOp implements Operand<T> {
+public final class Dot<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Dot operation.
@@ -49,10 +49,10 @@ public final class Dot<T extends Tensor> extends RawOp implements Operand<T> {
    * @return a new instance of Dot
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> Dot<T> create(Scope scope, Operand<T> lhs, Operand<T> rhs, String dimensionNumbers, String precisionConfig) {
+  public static <T extends TType> Dot<T> create(Scope scope, Operand<T> lhs, Operand<T> rhs, String dimensionNumbers, String precisionConfig) {
     OperationBuilder opBuilder = scope.env().opBuilder("XlaDot", scope.makeOpName("Dot"));
-    opBuilder.addInput(lhs.asOutput());
-    opBuilder.addInput(rhs.asOutput());
+    opBuilder.addInput(lhs.asOutput(scope));
+    opBuilder.addInput(rhs.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("dimension_numbers", dimensionNumbers);
     opBuilder.setAttr("precision_config", precisionConfig);
@@ -66,7 +66,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/DynamicSlice.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/DynamicSlice.java
index e3ea1be1b5e..249358166e1 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/DynamicSlice.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/DynamicSlice.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * Wraps the XLA DynamicSlice operator, documented at
@@ -43,7 +43,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "xla")
-public final class DynamicSlice<T extends Tensor> extends RawOp implements Operand<T> {
+public final class DynamicSlice<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new DynamicSlice operation.
@@ -58,11 +58,11 @@ public final class DynamicSlice<T extends Tensor> extends RawOp implements Opera
    * @return a new instance of DynamicSlice
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor, U extends Tensor & TNumber> DynamicSlice<T> create(Scope scope, Operand<T> input, Operand<U> startIndices, Operand<U> sizeIndices) {
+  public static <T extends TType, U extends TNumber> DynamicSlice<T> create(Scope scope, Operand<T> input, Operand<U> startIndices, Operand<U> sizeIndices) {
     OperationBuilder opBuilder = scope.env().opBuilder("XlaDynamicSlice", scope.makeOpName("DynamicSlice"));
-    opBuilder.addInput(input.asOutput());
-    opBuilder.addInput(startIndices.asOutput());
-    opBuilder.addInput(sizeIndices.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
+    opBuilder.addInput(startIndices.asOutput(scope));
+    opBuilder.addInput(sizeIndices.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new DynamicSlice<T>(opBuilder.build());
   }
@@ -74,7 +74,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/DynamicUpdateSlice.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/DynamicUpdateSlice.java
index 7782c2d9587..b94824d745e 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/DynamicUpdateSlice.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/DynamicUpdateSlice.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * Wraps the XLA DynamicUpdateSlice operator, documented at
@@ -44,7 +44,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "xla")
-public final class DynamicUpdateSlice<T extends Tensor> extends RawOp implements Operand<T> {
+public final class DynamicUpdateSlice<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new DynamicUpdateSlice operation.
@@ -57,11 +57,11 @@ public final class DynamicUpdateSlice<T extends Tensor> extends RawOp implements
    * @return a new instance of DynamicUpdateSlice
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor, U extends Tensor & TNumber> DynamicUpdateSlice<T> create(Scope scope, Operand<T> input, Operand<T> update, Operand<U> indices) {
+  public static <T extends TType, U extends TNumber> DynamicUpdateSlice<T> create(Scope scope, Operand<T> input, Operand<T> update, Operand<U> indices) {
     OperationBuilder opBuilder = scope.env().opBuilder("XlaDynamicUpdateSlice", scope.makeOpName("DynamicUpdateSlice"));
-    opBuilder.addInput(input.asOutput());
-    opBuilder.addInput(update.asOutput());
-    opBuilder.addInput(indices.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
+    opBuilder.addInput(update.asOutput(scope));
+    opBuilder.addInput(indices.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new DynamicUpdateSlice<T>(opBuilder.build());
   }
@@ -74,7 +74,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Einsum.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Einsum.java
index d83b2a7b487..28d2a86ed00 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Einsum.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Einsum.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * An op which supports basic einsum op with 2 inputs and 1 output.
@@ -36,7 +36,7 @@
  * @param <T> data type for {@code product()} output
  */
 @Operator(group = "xla")
-public final class Einsum<T extends Tensor> extends RawOp implements Operand<T> {
+public final class Einsum<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Einsum operation.
@@ -48,10 +48,10 @@ public final class Einsum<T extends Tensor> extends RawOp implements Operand<T>
    * @return a new instance of Einsum
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> Einsum<T> create(Scope scope, Operand<T> a, Operand<T> b, String equation) {
+  public static <T extends TType> Einsum<T> create(Scope scope, Operand<T> a, Operand<T> b, String equation) {
     OperationBuilder opBuilder = scope.env().opBuilder("XlaEinsum", scope.makeOpName("Einsum"));
-    opBuilder.addInput(a.asOutput());
-    opBuilder.addInput(b.asOutput());
+    opBuilder.addInput(a.asOutput(scope));
+    opBuilder.addInput(b.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("equation", equation);
     return new Einsum<T>(opBuilder.build());
@@ -64,7 +64,7 @@ public Output<T> product() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return product;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Gather.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Gather.java
index f1fd72ba2a5..1b7d3e4e627 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Gather.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Gather.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * Wraps the XLA Gather operator documented at
@@ -36,7 +36,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "xla")
-public final class Gather<T extends Tensor> extends RawOp implements Operand<T> {
+public final class Gather<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Gather operation.
@@ -50,11 +50,11 @@ public final class Gather<T extends Tensor> extends RawOp implements Operand<T>
    * @return a new instance of Gather
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor, U extends Tensor & TNumber> Gather<T> create(Scope scope, Operand<T> operand, Operand<U> startIndices, Operand<U> sliceSizes, String dimensionNumbers, Boolean indicesAreSorted) {
+  public static <T extends TType, U extends TNumber> Gather<T> create(Scope scope, Operand<T> operand, Operand<U> startIndices, Operand<U> sliceSizes, String dimensionNumbers, Boolean indicesAreSorted) {
     OperationBuilder opBuilder = scope.env().opBuilder("XlaGather", scope.makeOpName("Gather"));
-    opBuilder.addInput(operand.asOutput());
-    opBuilder.addInput(startIndices.asOutput());
-    opBuilder.addInput(sliceSizes.asOutput());
+    opBuilder.addInput(operand.asOutput(scope));
+    opBuilder.addInput(startIndices.asOutput(scope));
+    opBuilder.addInput(sliceSizes.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("dimension_numbers", dimensionNumbers);
     opBuilder.setAttr("indices_are_sorted", indicesAreSorted);
@@ -68,7 +68,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/KeyValueSort.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/KeyValueSort.java
index 5f60de49d10..60b3de8d550 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/KeyValueSort.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/KeyValueSort.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * Wraps the XLA Sort operator, documented at
@@ -40,7 +40,7 @@
  * @param <U> data type for {@code sortedValues()} output
  */
 @Operator(group = "xla")
-public final class KeyValueSort<T extends Tensor & TNumber, U extends Tensor> extends RawOp {
+public final class KeyValueSort<T extends TNumber, U extends TType> extends RawOp {
   
   /**
    * Factory method to create a class wrapping a new KeyValueSort operation.
@@ -51,10 +51,10 @@ public final class KeyValueSort<T extends Tensor & TNumber, U extends Tensor> ex
    * @return a new instance of KeyValueSort
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor & TNumber, U extends Tensor> KeyValueSort<T, U> create(Scope scope, Operand<T> keys, Operand<U> values) {
+  public static <T extends TNumber, U extends TType> KeyValueSort<T, U> create(Scope scope, Operand<T> keys, Operand<U> values) {
     OperationBuilder opBuilder = scope.env().opBuilder("XlaKeyValueSort", scope.makeOpName("KeyValueSort"));
-    opBuilder.addInput(keys.asOutput());
-    opBuilder.addInput(values.asOutput());
+    opBuilder.addInput(keys.asOutput(scope));
+    opBuilder.addInput(values.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new KeyValueSort<T, U>(opBuilder.build());
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Pad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Pad.java
index b2b7b511256..65c9d257ad5 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Pad.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Pad.java
@@ -21,12 +21,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * Wraps the XLA Pad operator, documented at
@@ -37,7 +37,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "xla")
-public final class Pad<T extends Tensor> extends RawOp implements Operand<T> {
+public final class Pad<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Pad operation.
@@ -51,13 +51,13 @@ public final class Pad<T extends Tensor> extends RawOp implements Operand<T> {
    * @return a new instance of Pad
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor, U extends Tensor & TNumber> Pad<T> create(Scope scope, Operand<T> input, Operand<T> paddingValue, Operand<U> paddingLow, Operand<U> paddingHigh, Operand<U> paddingInterior) {
+  public static <T extends TType, U extends TNumber> Pad<T> create(Scope scope, Operand<T> input, Operand<T> paddingValue, Operand<U> paddingLow, Operand<U> paddingHigh, Operand<U> paddingInterior) {
     OperationBuilder opBuilder = scope.env().opBuilder("XlaPad", scope.makeOpName("Pad"));
-    opBuilder.addInput(input.asOutput());
-    opBuilder.addInput(paddingValue.asOutput());
-    opBuilder.addInput(paddingLow.asOutput());
-    opBuilder.addInput(paddingHigh.asOutput());
-    opBuilder.addInput(paddingInterior.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
+    opBuilder.addInput(paddingValue.asOutput(scope));
+    opBuilder.addInput(paddingLow.asOutput(scope));
+    opBuilder.addInput(paddingHigh.asOutput(scope));
+    opBuilder.addInput(paddingInterior.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new Pad<T>(opBuilder.build());
   }
@@ -70,7 +70,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Recv.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Recv.java
index b53d6f36c55..602e85a5b90 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Recv.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Recv.java
@@ -22,12 +22,12 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Receives the named tensor from another XLA computation. Wraps the XLA Recv
@@ -38,7 +38,7 @@
  * @param <T> data type for {@code tensor()} output
  */
 @Operator(group = "xla")
-public final class Recv<T extends Tensor> extends RawOp implements Operand<T> {
+public final class Recv<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Recv operation.
@@ -50,7 +50,7 @@ public final class Recv<T extends Tensor> extends RawOp implements Operand<T> {
    * @return a new instance of Recv
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> Recv<T> create(Scope scope, DataType<T> dtype, String tensorName, Shape shape) {
+  public static <T extends TType> Recv<T> create(Scope scope, DataType<T> dtype, String tensorName, Shape shape) {
     OperationBuilder opBuilder = scope.env().opBuilder("XlaRecv", scope.makeOpName("Recv"));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("dtype", dtype);
@@ -67,7 +67,7 @@ public Output<T> tensor() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return tensor;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/ReplicaId.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/ReplicaId.java
index 886cc40b3ab..4d1b8e69a2b 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/ReplicaId.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/ReplicaId.java
@@ -53,7 +53,7 @@ public Output<TInt32> id() {
   }
   
   @Override
-  public Output<TInt32> asOutput() {
+  public Output<TInt32> asOutput(Scope scope) {
     return id;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/SelfAdjointEig.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/SelfAdjointEig.java
index b36fc427d4f..7acb1da628b 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/SelfAdjointEig.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/SelfAdjointEig.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Computes the eigen decomposition of a batch of self-adjoint matrices
@@ -39,7 +39,7 @@
  * @param <T> data type for {@code w()} output
  */
 @Operator(group = "xla")
-public final class SelfAdjointEig<T extends Tensor> extends RawOp {
+public final class SelfAdjointEig<T extends TType> extends RawOp {
   
   /**
    * Factory method to create a class wrapping a new SelfAdjointEig operation.
@@ -56,9 +56,9 @@ public final class SelfAdjointEig<T extends Tensor> extends RawOp {
    * @return a new instance of SelfAdjointEig
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> SelfAdjointEig<T> create(Scope scope, Operand<T> a, Boolean lower, Long maxIter, Float epsilon) {
+  public static <T extends TType> SelfAdjointEig<T> create(Scope scope, Operand<T> a, Boolean lower, Long maxIter, Float epsilon) {
     OperationBuilder opBuilder = scope.env().opBuilder("XlaSelfAdjointEig", scope.makeOpName("SelfAdjointEig"));
-    opBuilder.addInput(a.asOutput());
+    opBuilder.addInput(a.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("lower", lower);
     opBuilder.setAttr("max_iter", maxIter);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Send.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Send.java
index 7e8287d8766..73cc5acbd91 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Send.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Send.java
@@ -20,11 +20,11 @@
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Sends the named tensor to another XLA computation. Wraps the XLA Send operator
@@ -44,9 +44,9 @@ public final class Send extends RawOp {
    * @return a new instance of Send
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> Send create(Scope scope, Operand<T> tensor, String tensorName) {
+  public static <T extends TType> Send create(Scope scope, Operand<T> tensor, String tensorName) {
     OperationBuilder opBuilder = scope.env().opBuilder("XlaSend", scope.makeOpName("Send"));
-    opBuilder.addInput(tensor.asOutput());
+    opBuilder.addInput(tensor.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("tensor_name", tensorName);
     return new Send(opBuilder.build());
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Sharding.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Sharding.java
index cb483010a40..f3a982379fc 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Sharding.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Sharding.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * An op which shards the input based on the given sharding attribute.
@@ -33,7 +33,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "xla")
-public final class Sharding<T extends Tensor> extends RawOp implements Operand<T> {
+public final class Sharding<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Sharding operation.
@@ -43,9 +43,9 @@ public final class Sharding<T extends Tensor> extends RawOp implements Operand<T
    * @return a new instance of Sharding
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> Sharding<T> create(Scope scope, Operand<T> input) {
+  public static <T extends TType> Sharding<T> create(Scope scope, Operand<T> input) {
     OperationBuilder opBuilder = scope.env().opBuilder("XlaSharding", scope.makeOpName("Sharding"));
-    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new Sharding<T>(opBuilder.build());
   }
@@ -57,7 +57,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Sort.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Sort.java
index 11b15492709..3beaee0a3e9 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Sort.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Sort.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Wraps the XLA Sort operator, documented at
@@ -38,7 +38,7 @@
  * @param <T> data type for {@code output()} output
  */
 @Operator(group = "xla")
-public final class Sort<T extends Tensor> extends RawOp implements Operand<T> {
+public final class Sort<T extends TType> extends RawOp implements Operand<T> {
   
   /**
    * Factory method to create a class wrapping a new Sort operation.
@@ -48,9 +48,9 @@ public final class Sort<T extends Tensor> extends RawOp implements Operand<T> {
    * @return a new instance of Sort
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> Sort<T> create(Scope scope, Operand<T> input) {
+  public static <T extends TType> Sort<T> create(Scope scope, Operand<T> input) {
     OperationBuilder opBuilder = scope.env().opBuilder("XlaSort", scope.makeOpName("Sort"));
-    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new Sort<T>(opBuilder.build());
   }
@@ -63,7 +63,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Svd.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Svd.java
index b5dc57f6531..1e21a00b7f8 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Svd.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Svd.java
@@ -21,11 +21,11 @@
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.RawOp;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Computes the eigen decomposition of a batch of self-adjoint matrices
@@ -38,7 +38,7 @@
  * @param <T> data type for {@code s()} output
  */
 @Operator(group = "xla")
-public final class Svd<T extends Tensor> extends RawOp {
+public final class Svd<T extends TType> extends RawOp {
   
   /**
    * Factory method to create a class wrapping a new Svd operation.
@@ -54,9 +54,9 @@ public final class Svd<T extends Tensor> extends RawOp {
    * @return a new instance of Svd
    */
   @Endpoint(describeByClass = true)
-  public static <T extends Tensor> Svd<T> create(Scope scope, Operand<T> a, Long maxIter, Float epsilon, String precisionConfig) {
+  public static <T extends TType> Svd<T> create(Scope scope, Operand<T> a, Long maxIter, Float epsilon, String precisionConfig) {
     OperationBuilder opBuilder = scope.env().opBuilder("XlaSvd", scope.makeOpName("Svd"));
-    opBuilder.addInput(a.asOutput());
+    opBuilder.addInput(a.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("max_iter", maxIter);
     opBuilder.setAttr("epsilon", epsilon);
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/AbstractOperation.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/AbstractOperation.java
index 71a9bc45bcf..96da6bc5ff4 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/AbstractOperation.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/AbstractOperation.java
@@ -17,6 +17,7 @@
 
 import org.bytedeco.javacpp.Pointer;
 import org.tensorflow.ndarray.Shape;
+import org.tensorflow.types.family.TType;
 
 /**
  * Base class for {@link Operation} implementations.
@@ -36,7 +37,7 @@ public Output<?>[] outputList(int idx, int length) {
   }
 
   @Override
-  public <T extends Tensor> Output<T> output(int idx) {
+  public <T extends TType> Output<T> output(int idx) {
     return new Output<>(this, idx);
   }
 
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/AbstractTensor.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/AbstractTensor.java
new file mode 100644
index 00000000000..71fd952bd6a
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/AbstractTensor.java
@@ -0,0 +1,105 @@
+/* Copyright 2016 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+==============================================================================*/
+
+package org.tensorflow;
+
+import static org.tensorflow.internal.c_api.global.tensorflow.TF_Dim;
+import static org.tensorflow.internal.c_api.global.tensorflow.TF_NumDims;
+import static org.tensorflow.internal.c_api.global.tensorflow.TF_TensorByteSize;
+import static org.tensorflow.internal.c_api.global.tensorflow.TF_TensorType;
+
+import java.util.function.Consumer;
+import org.bytedeco.javacpp.PointerScope;
+import org.tensorflow.internal.c_api.TF_Tensor;
+import org.tensorflow.internal.tensor.buffer.TensorBuffers;
+import org.tensorflow.ndarray.NdArray;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.ndarray.buffer.ByteDataBuffer;
+import org.tensorflow.types.family.TType;
+
+/**
+ * A basic implementation of a dense tensor.
+ *
+ * <p>Dense tensors are stored as a continuous segment of native memory, mapped in a n-dimensional
+ * space, normally by delegating all I/O operations to an {@link NdArray} instance.
+ *
+ * <p>Usage of this class is meant to be kept internal.
+ */
+public abstract class AbstractDenseTensor<T> implements Tensor<T> {
+
+  @Override
+  @SuppressWarnings("unchecked")
+  public <U extends TType> U expect(DataType<U> dt) {
+    if (!dt.equals(this.dtype)) {
+      throw new IllegalArgumentException(
+          "Cannot cast from tensor of " + dtype + " to tensor of " + dt);
+    }
+    return (U)this;
+  }
+
+  @Override
+  public void close() {
+    nativeHandle().close();
+  }
+
+  @Override
+  public DataType<?> dataType() {
+    return dtype;
+  }
+
+  @Override
+  public long numBytes() {
+    return TF_TensorByteSize(nativeHandle());
+  }
+
+  @Override
+  public ByteDataBuffer rawData() {
+    return TensorBuffers.toBytes(nativeHandle(), true);
+  }
+
+  @Override
+  public String toString() {
+    return String.format("%s tensor with shape %s", dtype.toString(), shape());
+  }
+
+  protected AbstractDenseTensor(TF_Tensor nativeHandle, DataType<?> dtype) {
+    this.nativeHandle = nativeHandle;
+    this.dtype = dtype;
+    nativeHandle.retainReference();
+  }
+
+  /**
+   * @return native handle to this tensor
+   * @throws IllegalStateException if tensor has been closed
+   */
+  protected TF_Tensor nativeHandle() {
+    return requireHandle(nativeHandle);
+  }
+
+  void attachTo(EagerSession session) {
+    session.attach(nativeHandle());
+    nativeHandle.releaseReference();
+  }
+
+  protected static TF_Tensor requireHandle(TF_Tensor handle) {
+    if (handle == null || handle.isNull()) {
+      throw new IllegalStateException("close() was called on the Tensor");
+    }
+    return handle;
+  }
+
+  private final TF_Tensor nativeHandle;
+  private final DataType<?> dtype;
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/DataType.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/DataType.java
index 974a04f3333..70edafa391a 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/DataType.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/DataType.java
@@ -17,9 +17,10 @@
 
 import org.tensorflow.internal.c_api.TF_Tensor;
 import org.tensorflow.ndarray.Shape;
+import org.tensorflow.types.family.TType;
 
 /** Represents a type of elements in a {@link Tensor} */
-public final class DataType<T extends Tensor> {
+public final class DataType<T extends TType> {
 
   @FunctionalInterface
   public interface TensorInstantiator<T> {
@@ -42,7 +43,7 @@ public interface TensorInstantiator<T> {
    * @param byteSize size of an element of this type, in bytes, -1 if unknown
    * @param tensorMapper method for instantiating tensor from a native reference
    */
-  public static <T extends Tensor> DataType<T> create(String name, int value, int byteSize, TensorInstantiator<T> instantiator) {
+  public static <T extends TType> DataType<T> create(String name, int value, int byteSize, TensorInstantiator<T> instantiator) {
     return new DataType<>(name, value, byteSize, instantiator);
   }
 
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/EagerOperation.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/EagerOperation.java
index 690d5d1a4cb..cc0d93d6bac 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/EagerOperation.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/EagerOperation.java
@@ -135,7 +135,7 @@ private Tensor<?> resolveTensor(int outputIndex) {
     // instead.
     Tensor<?> tensor = resolveTensorHandle(getUnsafeNativeHandle(outputIndex), session);
     if (!outputTensors.compareAndSet(outputIndex, null, tensor)) {
-      session.detach(tensor.nativeHandle());
+      session.detach(((AbstractTensor)tensor).nativeHandle());
       tensor = outputTensors.get(outputIndex);
     }
     return tensor;
@@ -160,14 +160,10 @@ private static Tensor<?> resolveTensorHandle(TFE_TensorHandle handle, EagerSessi
     requireTensorHandle(handle);
     try (PointerScope scope = new PointerScope()) {
       TF_Status status = TF_Status.newStatus();
-      TF_Tensor tensor = TFE_TensorHandleResolve(handle, status).withDeallocator();
+      TF_Tensor nativeTensor = TFE_TensorHandleResolve(handle, status).withDeallocator();
       status.throwExceptionIfNotOK();
-      Tensor<?> t = Tensors.fromHandle(tensor);
-      session.attach(t.nativeHandle());
-      // Now that the tensor life is attached to the eager session scope, closing it will
-      // implicitly detach it from its previous internal scope
-      // FIXME TENSOR REFACT : is that right and ok?
-      t.close();
+      Tensor<?> t = Tensors.fromHandle(nativeTensor);
+      ((AbstractTensor)t).attachTo(session);
       return t;
     }
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/EagerOperationBuilder.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/EagerOperationBuilder.java
index f14795df55a..8bebb554e57 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/EagerOperationBuilder.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/EagerOperationBuilder.java
@@ -176,7 +176,7 @@ public EagerOperationBuilder setAttr(String name, DataType<?>[] values) {
 
   @Override
   public EagerOperationBuilder setAttr(String name, Tensor<?> value) {
-    setAttrTensor(opHandle, name, value.nativeHandle());
+    setAttrTensor(opHandle, name, ((AbstractTensor)value).nativeHandle());
     return this;
   }
 
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/GraphOperationBuilder.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/GraphOperationBuilder.java
index 2ef5c9010a1..e57c997b4b0 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/GraphOperationBuilder.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/GraphOperationBuilder.java
@@ -250,7 +250,7 @@ public GraphOperationBuilder setAttr(String name, DataType[] value) {
   public GraphOperationBuilder setAttr(String name, Tensor<?> value) {
     Graph.Reference r = graph.ref();
     try {
-      setAttrTensor(unsafeNativeHandle, name, value.nativeHandle());
+      setAttrTensor(unsafeNativeHandle, name, ((AbstractTensor)value).nativeHandle());
     } finally {
       r.close();
     }
@@ -262,7 +262,7 @@ public GraphOperationBuilder setAttr(String name, Tensor<?>[] value) {
     TF_Tensor[] handles = new TF_Tensor[value.length];
     int idx = 0;
     for (Tensor<?> t : value) {
-      handles[idx++] = t.nativeHandle();
+      handles[idx++] = ((AbstractTensor)t).nativeHandle();
     }
     Graph.Reference r = graph.ref();
     try {
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Operand.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Operand.java
index 78228d1a5ec..d80ce85dab2 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Operand.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Operand.java
@@ -16,6 +16,7 @@
 package org.tensorflow;
 
 import org.tensorflow.op.Op;
+import org.tensorflow.op.Scope;
 import org.tensorflow.types.family.TType;
 
 /**
@@ -39,7 +40,7 @@
  * tf.concat(split, tf.constant(0));
  * }</pre>
  */
-public interface Operand<T extends Tensor> extends Op {
+public interface Operand<T extends TType> extends Op {
 
   /**
    * Returns the symbolic handle of the tensor.
@@ -47,9 +48,26 @@ public interface Operand<T extends Tensor> extends Op {
    * <p>Inputs to TensorFlow operations are outputs of another TensorFlow operation. This method is
    * used to obtain a symbolic handle that represents the computation of the input.
    *
+   * <p>If a valid non-null value is provided as the {@code scope} of this operand, the implementor
+   * can program additional computations before returning its output. This might be required if,
+   * for instance, the implementor has not yet been added as a node to the graph.</p>
+   *
+   * @param scope if non-null, scope that can be used by the operand to produce a new output
    * @see OperationBuilder#addInput(Output)
    */
-  Output<T> asOutput();
+  Output<T> asOutput(Scope scope);
+
+  /**
+   * Returns the symbolic handle of the tensor.
+   *
+   * <p>Inputs to TensorFlow operations are outputs of another TensorFlow operation. This method is
+   * used to obtain a symbolic handle that represents the computation of the input.
+   *
+   * @see OperationBuilder#addInput(Output)
+   */
+  default Output<T> asOutput() {
+    return asOutput(null);
+  }
 
   /**
    * Returns this operand as a tensor.
@@ -65,17 +83,16 @@ default T asTensor() {
   }
 
   /**
-   * Returns the data of this operand.
+   * Returns this operand as a tensor of the given type.
    *
    * <i>Only works when running in an eager execution</i>
-   * <p>This helper method is equivalent to {@code asTensor().data()}
+   * <p>This helper method is equivalent to {@code asOutput().tensor()}
    *
-   * @return the tensor data
+   * @return the tensor
    * @throws IllegalStateException if this is an operand of a graph
-   * @deprecated use {@link #asTensor()} instead
+   * @throws IllegalArgumentException if the {@code dataType} is incompatible with tensor type
    */
-  @Deprecated
-  default T data() {
-    return asTensor();
+  default <U extends TType> U asTensor(DataType<U> dataType) {
+    return (U)asOutput().tensor().expect(dataType);
   }
 }
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Operation.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Operation.java
index dd7903e07a6..1cc175da161 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Operation.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Operation.java
@@ -15,6 +15,8 @@
 
 package org.tensorflow;
 
+import org.tensorflow.types.family.TType;
+
 /**
  * Performs computation on Tensors.
  *
@@ -68,7 +70,7 @@ public interface Operation {
    * @param <T> The expected element type of the tensors produced by this output.
    * @param idx The index of the output among the outputs produced by this operation.
    */
-  <T extends Tensor> Output<T> output(int idx);
+  <T extends TType> Output<T> output(int idx);
 
   /**
    * Returns the size of the given inputs list of Tensors for this operation.
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Output.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Output.java
index 944ed507ee1..55979a5f6da 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Output.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Output.java
@@ -18,6 +18,8 @@
 import java.util.Objects;
 import org.bytedeco.javacpp.Pointer;
 import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.Scope;
+import org.tensorflow.types.family.TType;
 
 /**
  * A symbolic handle to a tensor produced by an {@link Operation}.
@@ -28,7 +30,7 @@
  * <p>By implementing the {@link Operand} interface, instances of this class also act as operands to
  * {@link org.tensorflow.op.Op Op} instances.
  */
-public final class Output<T extends Tensor> implements Operand<T> {
+public final class Output<T extends TType> implements Operand<T> {
 
   /** Returns the index into the outputs of the Operation. */
   public int index() {
@@ -55,7 +57,7 @@ public DataType<T> dataType() {
    *     {@code U}.
    */
   @SuppressWarnings("unchecked")
-  public <U extends Tensor> Output<U> expect(DataType<U> dt) {
+  public <U extends TType> Output<U> expect(DataType<U> dt) {
     if (!dt.equals(this.dataType())) {
       throw new IllegalArgumentException(
           "Cannot cast from output of " + this.dataType() + " to output of " + dt);
@@ -89,7 +91,7 @@ public Operation op() {
   }
 
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return this;
   }
 
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Session.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Session.java
index b486c0b5dd7..ce2ad5cbbd1 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Session.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Session.java
@@ -355,7 +355,7 @@ private Run runHelper(boolean wantMetadata) {
       // validity of the Graph and graphRef ensures that.
       int idx = 0;
       for (Tensor<?> t : inputTensors) {
-        inputTensorHandles[idx++] = t.nativeHandle();
+        inputTensorHandles[idx++] = ((AbstractTensor)t).nativeHandle();
       }
       idx = 0;
       for (Output<?> o : inputs) {
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Tensor.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Tensor.java
index 64ba11d2e14..feb308a513e 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Tensor.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Tensor.java
@@ -15,12 +15,11 @@
 
 package org.tensorflow;
 
-import java.util.function.Consumer;
-import org.tensorflow.internal.tensor.buffer.TensorBuffers;
-import org.tensorflow.internal.c_api.TF_Tensor;
+import org.checkerframework.checker.units.qual.C;
 import org.tensorflow.ndarray.NdArray;
-import org.tensorflow.ndarray.Shape;
 import org.tensorflow.ndarray.buffer.ByteDataBuffer;
+import org.tensorflow.ndarray.buffer.DataBuffer;
+import org.tensorflow.types.family.TType;
 
 /**
  * A statically typed multi-dimensional array whose elements are of a type described by T.
@@ -39,133 +38,6 @@
  */
 public interface Tensor<T> extends NdArray<T>, AutoCloseable {
 
-  /**
-   * Allocates a tensor of a given datatype and shape.
-   *
-   * <p>The amount of memory to allocate is derived from the datatype and the shape of the tensor.
-   * Memory is left uninitialized after this method returns, so it is the responsibility of the
-   * caller to initialize the tensor data before it is used, via the {@link #data()} accessor.
-   * For example:
-   *
-   * <pre>{@code
-   * FloatNdArray data = ...
-   * try (Tensor<TFloat32> t = Tensor.of(TFloat32.DTYPE, Shape.of(2, 2))) {
-   *   data.copyTo(t.data());
-   *   ...
-   * }
-   * }</pre>
-   *
-   * @param <T> the tensor element type
-   * @param dtype datatype of the tensor
-   * @param shape shape of the tensor
-   * @return an allocated but uninitialized tensor
-   * @throws IllegalStateException if tensor failed to be allocated
-   */
-  static <T extends Tensor> T of(DataType<T> dtype, Shape shape) {
-    return of(dtype, shape, shape.size() * dtype.byteSize());
-  }
-
-  /**
-   * Allocates a tensor of a given datatype, shape and size.
-   *
-   * <p>This method is identical to {@link #of(DataType, Shape)}, except that the final size of the
-   * tensor is explicitly set instead of computing it from the datatype and shape.
-   *
-   * <p>This could be useful for tensor types that stores data but also metadata in the tensor memory,
-   * like {@link org.tensorflow.types.TString TString}.
-   *
-   * @param <T> the tensor element type
-   * @param dtype datatype of the tensor
-   * @param shape shape of the tensor
-   * @param size size, in bytes, of the tensor
-   * @return an allocated but uninitialized tensor
-   * @see #of(DataType, Shape)
-   * @throws IllegalArgumentException if {@code size} is smaller than the minimum space required to
-   *                                  store the tensor data
-   * @throws IllegalStateException if tensor failed to be allocated
-   */
-  static <T extends Tensor> T of(DataType<T> dtype, Shape shape, long size) {
-    return Tensors.allocate(dtype, shape, size);
-  }
-
-  /**
-   * Allocates and initialize a tensor of a given datatype and shape.
-   *
-   * <p>The amount of memory to allocate is derived from the datatype and the shape of the tensor.
-   * Tensor data is initialized by calling the {@code dataInitializer}, which receives in argument
-   * the value returned by {@link #data()} on the allocated tensor. For example:
-   *
-   * <pre>{@code
-   * FloatNdArray data = ...
-   * try (Tensor<TFloat32> t = Tensor.of(TFloat32.DTYPE, Shape.of(2, 2), data::copyTo)) {
-   *   ...
-   * }
-   * }</pre>
-   *
-   * <p>If {@code dataInitializer} fails and throws an exception, the allocated tensor will be
-   * automatically released before rethrowing the same exception.
-   *
-   * @param <T> the tensor element type
-   * @param dtype datatype of the tensor
-   * @param shape shape of the tensor
-   * @param dataInitializer method receiving accessor to the allocated tensor data for initialization
-   * @return an allocated and initialized tensor
-   * @throws IllegalStateException if tensor failed to be allocated
-   */
-  static <T extends Tensor> T of(DataType<T> dtype, Shape shape, Consumer<T> dataInitializer) {
-    return of(dtype, shape, shape.size() * dtype.byteSize(), dataInitializer);
-  }
-
-  /**
-   * Allocates a tensor of a given datatype, shape and size.
-   *
-   * <p>This method is identical to {@link #of(DataType, Shape, Consumer)}, except that the final
-   * size for the tensor is explicitly set instead of being computed from the datatype and shape.
-   *
-   * <p>This could be useful for tensor types that stores data but also metadata in the tensor memory,
-   * such as {@link org.tensorflow.types.TString TString}.
-   *
-   * @param <T> the tensor element type
-   * @param dtype datatype of the tensor
-   * @param shape shape of the tensor
-   * @param size size, in bytes, of the tensor
-   * @param tensorInit method receiving accessor to the allocated tensor data for initialization
-   * @return an allocated and initialized tensor
-   * @see #of(DataType, Shape, long, Consumer)
-   * @throws IllegalArgumentException if {@code size} is smaller than the minimum space required to
-   *                                  store the tensor data
-   * @throws IllegalStateException if tensor failed to be allocated
-   */
-  static <T extends Tensor> T of(DataType<T> dtype, Shape shape, long size, Consumer<T> tensorInit) {
-    T tensor = of(dtype, shape, size);
-    try {
-      tensorInit.accept(tensor);
-      return tensor;
-    } catch (Throwable t) {
-      tensor.close();
-      throw t;
-    }
-  }
-
-  /**
-   * Creates a Tensor of any type from the raw data provided by the given buffer.
-   *
-   * <p>Data must have been encoded into {@code data} as per the specification of the TensorFlow <a
-   * href="https://www.tensorflow.org/code/tensorflow/c/c_api.h">C API</a>.
-   *
-   * @param <T> the tensor element type
-   * @param dtype the tensor element data type
-   * @param shape the tensor shape.
-   * @param rawData a buffer containing the tensor raw data.
-   * @throws IllegalArgumentException if {@code rawData} is not large enough to contain the tensor data
-   * @throws IllegalStateException if tensor failed to be allocated with the given parameters
-   */
-  static <T extends Tensor> T of(DataType<T> dtype, Shape shape, ByteDataBuffer rawData) {
-    T tensor = of(dtype, shape, rawData.size());
-    rawData.copyTo(TensorBuffers.toBytes(tensor.nativeHandle()), rawData.size());
-    return tensor;
-  }
-
   /**
    * Returns this Tensor object with the type {@code Tensor<U>}. This method is useful when given a
    * value of type {@code Tensor<?>}.
@@ -176,14 +48,7 @@ static <T extends Tensor> T of(DataType<T> dtype, Shape shape, ByteDataBuffer ra
    * @throws IllegalArgumentException if the actual data type of this object does not match the type
    *     {@code U}.
    */
-  @SuppressWarnings("unchecked")
-  default <U extends Tensor> U expect(DataType<U> dt) {
-    if (!dt.equals(dataType())) {
-      throw new IllegalArgumentException(
-          "Cannot cast from tensor of " + dataType() + " to tensor of " + dt);
-    }
-    return (U)this;
-  }
+  <U extends TType> U expect(DataType<U> dt);
 
   /**
    * Release resources associated with the Tensor.
@@ -202,53 +67,6 @@ default <U extends Tensor> U expect(DataType<U> dt) {
   /** Returns the size, in bytes, of the tensor data. */
   long numBytes();
 
-  /**
-   * Returns the <a href="https://www.tensorflow.org/resources/dims_types.html#shape">shape</a> of
-   * the Tensor, i.e., the sizes of each dimension.
-   *
-   * @return shape of this tensor
-   */
-  Shape shape();
-
-  /**
-   * Returns the data of this tensor.
-   *
-   * <p>This method returns an accessor to the tensor data as an instance of {@code T}, which
-   * commonly maps this data to an {@link NdArray NdArray}. Input and
-   * output operations performed on the returned n-dimensional array are applied directly to the
-   * tensor native memory. For example:
-   *
-   * <pre>{@code
-   * Ops tf = Ops.create();
-   * try (Tensor<TFloat32> t = TFloat32.tensorOf(Shape.of(2, 2))) {
-   *   TFloat32 data = t.data();
-   *
-   *   StdArrays.copyTo(data, new float[][] {
-   *     {1.0f, 2.0f},
-   *     {3.0f, 4.0f}
-   *   });
-   *   assertEquals(NdArrays.vectorOf(3.0f, 4.0f), data.getFloat(1));
-   *
-   *   Constant<TFloat32> c = tf.constant(t);
-   *   assertEquals(4.0f, c.data().getFloat(1, 1));
-   * }
-   * }</pre>
-   *
-   * <p>Please refer to the documentation of the {@link NdArray NdArray}
-   * classes for more information on the various techniques to read or write data in an
-   * n-dimensional space using this data structure.
-   *
-   * @return the tensor data mapped to an n-dimensional space
-   * @throws IllegalStateException if the tensor has been closed
-   * @see NdArray
-   * @deprecated since tensor implements {@code NdArray}
-   */
-  @Deprecated
-  default Tensor<T> data() {
-    nativeHandle(); // make sure native handle is still valid
-    return this;
-  }
-
   /**
    * Returns the raw data of this tensor as a buffer of bytes.
    *
@@ -261,11 +79,20 @@ default Tensor<T> data() {
    */
   ByteDataBuffer rawData();
 
-  /**
-   * @return native handle to this tensor
-   * @throws IllegalStateException if tensor has been closed
-   */
-  TF_Tensor nativeHandle();
+  @Override
+  Tensor<T> set(NdArray<T> src, long... coordinates);
+
+  @Override
+  Tensor<T> setObject(T value, long... coordinates);
+
+  @Override
+  Tensor<T> copyTo(NdArray<T> dst);
+
+  @Override
+  Tensor<T> read(DataBuffer<T> dst);
+
+  @Override
+  Tensor<T> write(DataBuffer<T> src);
 }
 
 
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Tensors.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Tensors.java
index b04a0f297b9..3df1df81ddf 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Tensors.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Tensors.java
@@ -4,14 +4,144 @@
 import static org.tensorflow.internal.c_api.global.tensorflow.TF_NumDims;
 import static org.tensorflow.internal.c_api.global.tensorflow.TF_TensorType;
 
+import java.util.function.Consumer;
 import org.bytedeco.javacpp.PointerScope;
 import org.tensorflow.internal.c_api.TF_Tensor;
+import org.tensorflow.internal.tensor.buffer.TensorBuffers;
 import org.tensorflow.ndarray.Shape;
+import org.tensorflow.ndarray.buffer.ByteDataBuffer;
 import org.tensorflow.types.family.TType;
 
-final class Tensors {
+public final class Tensors {
 
-  static <T extends Tensor> T allocate(DataType<T> dataType, Shape shape, long size) {
+  /**
+   * Allocates a tensor of a given datatype and shape.
+   *
+   * <p>The amount of memory to allocate is derived from the datatype and the shape of the tensor.
+   * Memory is left uninitialized after this method returns, so it is the responsibility of the
+   * caller to initialize the tensor data before it is used, via the {@link #data()} accessor.
+   * For example:
+   *
+   * <pre>{@code
+   * FloatNdArray data = ...
+   * try (Tensor<TFloat32> t = Tensor.of(TFloat32.DTYPE, Shape.of(2, 2))) {
+   *   data.copyTo(t.data());
+   *   ...
+   * }
+   * }</pre>
+   *
+   * @param <T> the tensor element type
+   * @param dtype datatype of the tensor
+   * @param shape shape of the tensor
+   * @return an allocated but uninitialized tensor
+   * @throws IllegalStateException if tensor failed to be allocated
+   */
+  public static <T extends TType> T of(DataType<T> dtype, Shape shape) {
+    return of(dtype, shape, shape.size() * dtype.byteSize());
+  }
+
+  /**
+   * Allocates a tensor of a given datatype, shape and size.
+   *
+   * <p>This method is identical to {@link #of(DataType, Shape)}, except that the final size of the
+   * tensor is explicitly set instead of computing it from the datatype and shape.
+   *
+   * <p>This could be useful for tensor types that stores data but also metadata in the tensor memory,
+   * like {@link org.tensorflow.types.TString TString}.
+   *
+   * @param <T> the tensor element type
+   * @param dtype datatype of the tensor
+   * @param shape shape of the tensor
+   * @param size size, in bytes, of the tensor
+   * @return an allocated but uninitialized tensor
+   * @see #of(DataType, Shape)
+   * @throws IllegalArgumentException if {@code size} is smaller than the minimum space required to
+   *                                  store the tensor data
+   * @throws IllegalStateException if tensor failed to be allocated
+   */
+  public static <T extends TType> T of(DataType<T> dtype, Shape shape, long size) {
+    return Tensors.allocate(dtype, shape, size);
+  }
+
+  /**
+   * Allocates and initialize a tensor of a given datatype and shape.
+   *
+   * <p>The amount of memory to allocate is derived from the datatype and the shape of the tensor.
+   * Tensor data is initialized by calling the {@code dataInitializer}, which receives in argument
+   * the value returned by {@link #data()} on the allocated tensor. For example:
+   *
+   * <pre>{@code
+   * FloatNdArray data = ...
+   * try (Tensor<TFloat32> t = Tensor.of(TFloat32.DTYPE, Shape.of(2, 2), data::copyTo)) {
+   *   ...
+   * }
+   * }</pre>
+   *
+   * <p>If {@code dataInitializer} fails and throws an exception, the allocated tensor will be
+   * automatically released before rethrowing the same exception.
+   *
+   * @param <T> the tensor element type
+   * @param dtype datatype of the tensor
+   * @param shape shape of the tensor
+   * @param dataInitializer method receiving accessor to the allocated tensor data for initialization
+   * @return an allocated and initialized tensor
+   * @throws IllegalStateException if tensor failed to be allocated
+   */
+  public static <T extends TType> T of(DataType<T> dtype, Shape shape, Consumer<T> dataInitializer) {
+    return of(dtype, shape, shape.size() * dtype.byteSize(), dataInitializer);
+  }
+
+  /**
+   * Allocates a tensor of a given datatype, shape and size.
+   *
+   * <p>This method is identical to {@link #of(DataType, Shape, Consumer)}, except that the final
+   * size for the tensor is explicitly set instead of being computed from the datatype and shape.
+   *
+   * <p>This could be useful for tensor types that stores data but also metadata in the tensor memory,
+   * such as {@link org.tensorflow.types.TString TString}.
+   *
+   * @param <T> the tensor element type
+   * @param dtype datatype of the tensor
+   * @param shape shape of the tensor
+   * @param size size, in bytes, of the tensor
+   * @param tensorInit method receiving accessor to the allocated tensor data for initialization
+   * @return an allocated and initialized tensor
+   * @see #of(DataType, Shape, long, Consumer)
+   * @throws IllegalArgumentException if {@code size} is smaller than the minimum space required to
+   *                                  store the tensor data
+   * @throws IllegalStateException if tensor failed to be allocated
+   */
+  public static <T extends TType> T of(DataType<T> dtype, Shape shape, long size, Consumer<T> tensorInit) {
+    T tensor = of(dtype, shape, size);
+    try {
+      tensorInit.accept(tensor);
+      return tensor;
+    } catch (Throwable t) {
+      tensor.close();
+      throw t;
+    }
+  }
+
+  /**
+   * Creates a Tensor of any type from the raw data provided by the given buffer.
+   *
+   * <p>Data must have been encoded into {@code data} as per the specification of the TensorFlow <a
+   * href="https://www.tensorflow.org/code/tensorflow/c/c_api.h">C API</a>.
+   *
+   * @param <T> the tensor element type
+   * @param dtype the tensor element data type
+   * @param shape the tensor shape.
+   * @param rawData a buffer containing the tensor raw data.
+   * @throws IllegalArgumentException if {@code rawData} is not large enough to contain the tensor data
+   * @throws IllegalStateException if tensor failed to be allocated with the given parameters
+   */
+  public static <T extends TType> T of(DataType<T> dtype, Shape shape, ByteDataBuffer rawData) {
+    T tensor = of(dtype, shape, rawData.size());
+    rawData.copyTo(TensorBuffers.toBytes(((AbstractTensor)tensor).nativeHandle()), rawData.size());
+    return tensor;
+  }
+
+  static <T extends TType> T allocate(DataType<T> dataType, Shape shape, long size) {
     // Minimum requirements for datatypes of variable length cannot be verified in a relevant way so
     // we only validate them for fixed length datatypes
     if (!dataType.isVariableLength() && shape.size() * dataType.byteSize() > size) {
@@ -30,12 +160,9 @@ static <T extends Tensor> T allocate(DataType<T> dataType, Shape shape, long siz
    * <p>Takes ownership of the handle.
    */
   static Tensor<?> fromHandle(TF_Tensor handle) {
-    DataType<? extends Tensor> dataType = DataTypes.fromNativeCode(dtype(handle));
+    DataType<?> dataType = DataTypes.fromNativeCode(dtype(handle));
     Shape shape = Shape.of(shape(handle));
-    try (PointerScope scope = new PointerScope()) {
-      scope.attach(handle);
-      return dataType.instantiateTensor(handle, shape);
-    }
+    return dataType.instantiateTensor(handle, shape);
   }
 
   private static TF_Tensor requireHandle(TF_Tensor handle) {
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/BooleanTensorImpl.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/BooleanTensorImpl.java
index 09318072a7f..9c7c059691d 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/BooleanTensorImpl.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/BooleanTensorImpl.java
@@ -1,96 +1,143 @@
 package org.tensorflow.internal.tensor;
 
+import org.tensorflow.AbstractTensor;
 import org.tensorflow.DataType;
+import org.tensorflow.Tensor;
 import org.tensorflow.internal.c_api.TF_Tensor;
+import org.tensorflow.ndarray.BooleanNdArray;
 import org.tensorflow.ndarray.NdArray;
+import org.tensorflow.ndarray.NdArraySequence;
+import org.tensorflow.ndarray.NdArrays;
 import org.tensorflow.ndarray.Shape;
-import org.tensorflow.ndarray.buffer.BooleanDataBuffer;
-import org.tensorflow.ndarray.buffer.ByteDataBuffer;
 import org.tensorflow.ndarray.buffer.DataBuffer;
-import org.tensorflow.ndarray.impl.dense.BooleanDenseNdArray;
+import org.tensorflow.ndarray.buffer.BooleanDataBuffer;
+import org.tensorflow.ndarray.index.Index;
 import org.tensorflow.tensor.BooleanTensor;
 
-public class BooleanTensorImpl extends BooleanDenseNdArray implements BooleanTensor {
+public class BooleanTensorImpl extends AbstractTensor<Boolean> implements BooleanTensor {
 
-  @Override
-  public TF_Tensor nativeHandle() {
-    return rawTensor.nativeHandle();
+  private final BooleanNdArray data;
+
+  public BooleanTensorImpl(TF_Tensor nativeHandle, DataType<?> dataType, Shape shape,
+      BooleanDataBuffer buffer) {
+    super(nativeHandle, dataType);
+    data = NdArrays.wrap(shape, buffer);
   }
 
   @Override
-  public DataType<?> dataType() {
-    return rawTensor.dataType();
+  public boolean getBoolean(long... coordinates) {
+    requireHandle(nativeHandle);
+    return data.getBoolean(coordinates);
   }
 
   @Override
-  public Shape shape() {
-    return rawTensor.shape();
+  public BooleanTensor setBoolean(boolean value, long... coordinates) {
+    requireHandle(nativeHandle);
+    data.setBoolean(value, coordinates);
+    return this;
   }
 
   @Override
-  public long numBytes() {
-    return rawTensor.numBytes();
+  public BooleanNdArray slice(Index... coordinates) {
+    requireHandle(nativeHandle);
+    return data.slice(coordinates);
   }
 
   @Override
-  public ByteDataBuffer rawData() {
-    return rawTensor.rawData();
+  public BooleanTensor get(long... coordinates) {
+    requireHandle(nativeHandle);
+    data.get(coordinates);
+    return this;
   }
 
   @Override
-  public void close() {
-    rawTensor.close();
+  public BooleanTensor set(NdArray<Boolean> src, long... coordinates) {
+    requireHandle(nativeHandle);
+    data.set(src, coordinates);
+    return this;
   }
 
   @Override
-  public String toString() {
-    return rawTensor.toString();
+  public Boolean getObject(long... coordinates) {
+    requireHandle(nativeHandle);
+    return data.getObject(coordinates);
   }
 
   @Override
-  public BooleanTensor setBoolean(boolean value, long... coordinates) {
-    return (BooleanTensor)super.setBoolean(value, coordinates);
+  public BooleanTensor setObject(Boolean value, long... coordinates) {
+    requireHandle(nativeHandle);
+    data.setObject(value, coordinates);
+    return this;
   }
 
   @Override
-  public BooleanTensor setObject(Boolean value, long... coordinates) {
-    return (BooleanTensor)super.setObject(value, coordinates);
+  public NdArraySequence<BooleanNdArray> elements(int dimensionIdx) {
+    requireHandle(nativeHandle);
+    return data.elements(dimensionIdx);
   }
 
   @Override
-  public BooleanTensor set(NdArray<Boolean> src, long... coordinates) {
-    return (BooleanTensor)super.set(src, coordinates);
+  public NdArraySequence<BooleanNdArray> scalars() {
+    requireHandle(nativeHandle);
+    return data.scalars();
   }
 
   @Override
   public BooleanTensor copyTo(NdArray<Boolean> dst) {
-    return (BooleanTensor)super.copyTo(dst);
+    requireHandle(nativeHandle);
+    data.copyTo(dst);
+    return this;
   }
 
   @Override
   public BooleanTensor read(DataBuffer<Boolean> dst) {
-    return (BooleanTensor)super.read(dst);
+    requireHandle(nativeHandle);
+    data.read(dst);
+    return this;
   }
 
   @Override
   public BooleanTensor read(BooleanDataBuffer dst) {
-    return (BooleanTensor)super.read(dst);
+    requireHandle(nativeHandle);
+    data.read(dst);
+    return this;
   }
 
   @Override
   public BooleanTensor write(DataBuffer<Boolean> src) {
-    return (BooleanTensor)super.write(src);
+    requireHandle(nativeHandle);
+    data.write(src);
+    return this;
   }
 
   @Override
   public BooleanTensor write(BooleanDataBuffer src) {
-    return (BooleanTensor)super.write(src);
+    requireHandle(nativeHandle);
+    data.write(src);
+    return this;
   }
 
-  public BooleanTensorImpl(TF_Tensor nativeHandle, DataType<?> dataType, Shape shape, BooleanDataBuffer buffer) {
-    super(buffer, shape);
-    this.rawTensor = new RawTensor(nativeHandle, dataType, shape);
+  @Override
+  public Shape shape() {
+    requireHandle(nativeHandle);
+    return data.shape();
+  }
+
+  @Override
+  public int rank() {
+    requireHandle(nativeHandle);
+    return data.rank();
   }
 
-  private final RawTensor rawTensor;
+  @Override
+  public long size() {
+    requireHandle(nativeHandle);
+    return data.size();
+  }
+
+  @Override
+  public boolean equals(Object obj) {
+    requireHandle(nativeHandle);
+    return data.equals(obj);
+  }
 }
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/ByteTensorImpl.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/ByteTensorImpl.java
index 7f4f4742ff6..18d993219c6 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/ByteTensorImpl.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/ByteTensorImpl.java
@@ -1,95 +1,143 @@
 package org.tensorflow.internal.tensor;
 
+import org.tensorflow.AbstractTensor;
 import org.tensorflow.DataType;
+import org.tensorflow.Tensor;
 import org.tensorflow.internal.c_api.TF_Tensor;
+import org.tensorflow.ndarray.ByteNdArray;
 import org.tensorflow.ndarray.NdArray;
+import org.tensorflow.ndarray.NdArraySequence;
+import org.tensorflow.ndarray.NdArrays;
 import org.tensorflow.ndarray.Shape;
-import org.tensorflow.ndarray.buffer.ByteDataBuffer;
 import org.tensorflow.ndarray.buffer.DataBuffer;
-import org.tensorflow.ndarray.impl.dense.ByteDenseNdArray;
+import org.tensorflow.ndarray.buffer.ByteDataBuffer;
+import org.tensorflow.ndarray.index.Index;
 import org.tensorflow.tensor.ByteTensor;
 
-public class ByteTensorImpl extends ByteDenseNdArray implements ByteTensor {
+public class ByteTensorImpl extends AbstractTensor<Byte> implements ByteTensor {
 
-  @Override
-  public TF_Tensor nativeHandle() {
-    return rawTensor.nativeHandle();
+  private final ByteNdArray data;
+
+  public ByteTensorImpl(TF_Tensor nativeHandle, DataType<?> dataType, Shape shape,
+      ByteDataBuffer buffer) {
+    super(nativeHandle, dataType);
+    data = NdArrays.wrap(shape, buffer);
   }
 
   @Override
-  public DataType<?> dataType() {
-    return rawTensor.dataType();
+  public byte getByte(long... coordinates) {
+    requireHandle(nativeHandle);
+    return data.getByte(coordinates);
   }
 
   @Override
-  public Shape shape() {
-    return rawTensor.shape();
+  public ByteTensor setByte(byte value, long... coordinates) {
+    requireHandle(nativeHandle);
+    data.setByte(value, coordinates);
+    return this;
   }
 
   @Override
-  public long numBytes() {
-    return rawTensor.numBytes();
+  public ByteNdArray slice(Index... coordinates) {
+    requireHandle(nativeHandle);
+    return data.slice(coordinates);
   }
 
   @Override
-  public ByteDataBuffer rawData() {
-    return rawTensor.rawData();
+  public ByteTensor get(long... coordinates) {
+    requireHandle(nativeHandle);
+    data.get(coordinates);
+    return this;
   }
 
   @Override
-  public void close() {
-    rawTensor.close();
+  public ByteTensor set(NdArray<Byte> src, long... coordinates) {
+    requireHandle(nativeHandle);
+    data.set(src, coordinates);
+    return this;
   }
 
   @Override
-  public String toString() {
-    return rawTensor.toString();
+  public Byte getObject(long... coordinates) {
+    requireHandle(nativeHandle);
+    return data.getObject(coordinates);
   }
 
   @Override
-  public ByteTensor setByte(byte value, long... coordinates) {
-    return (ByteTensor)super.setByte(value, coordinates);
+  public ByteTensor setObject(Byte value, long... coordinates) {
+    requireHandle(nativeHandle);
+    data.setObject(value, coordinates);
+    return this;
   }
 
   @Override
-  public ByteTensor setObject(Byte value, long... coordinates) {
-    return (ByteTensor)super.setObject(value, coordinates);
+  public NdArraySequence<ByteNdArray> elements(int dimensionIdx) {
+    requireHandle(nativeHandle);
+    return data.elements(dimensionIdx);
   }
 
   @Override
-  public ByteTensor set(NdArray<Byte> src, long... coordinates) {
-    return (ByteTensor)super.set(src, coordinates);
+  public NdArraySequence<ByteNdArray> scalars() {
+    requireHandle(nativeHandle);
+    return data.scalars();
   }
 
   @Override
   public ByteTensor copyTo(NdArray<Byte> dst) {
-    return (ByteTensor)super.copyTo(dst);
+    requireHandle(nativeHandle);
+    data.copyTo(dst);
+    return this;
   }
 
   @Override
   public ByteTensor read(DataBuffer<Byte> dst) {
-    return (ByteTensor)super.read(dst);
+    requireHandle(nativeHandle);
+    data.read(dst);
+    return this;
   }
 
   @Override
   public ByteTensor read(ByteDataBuffer dst) {
-    return (ByteTensor)super.read(dst);
+    requireHandle(nativeHandle);
+    data.read(dst);
+    return this;
   }
 
   @Override
   public ByteTensor write(DataBuffer<Byte> src) {
-    return (ByteTensor)super.write(src);
+    requireHandle(nativeHandle);
+    data.write(src);
+    return this;
   }
 
   @Override
   public ByteTensor write(ByteDataBuffer src) {
-    return (ByteTensor)super.write(src);
+    requireHandle(nativeHandle);
+    data.write(src);
+    return this;
   }
 
-  public ByteTensorImpl(TF_Tensor nativeHandle, DataType<?> dataType, Shape shape, ByteDataBuffer buffer) {
-    super(buffer, shape);
-    this.rawTensor = new RawTensor(nativeHandle, dataType, shape);
+  @Override
+  public Shape shape() {
+    requireHandle(nativeHandle);
+    return data.shape();
+  }
+
+  @Override
+  public int rank() {
+    requireHandle(nativeHandle);
+    return data.rank();
   }
 
-  private final RawTensor rawTensor;
+  @Override
+  public long size() {
+    requireHandle(nativeHandle);
+    return data.size();
+  }
+
+  @Override
+  public boolean equals(Object obj) {
+    requireHandle(nativeHandle);
+    return data.equals(obj);
+  }
 }
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/DoubleTensorImpl.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/DoubleTensorImpl.java
index f4da880c976..7f85d86c372 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/DoubleTensorImpl.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/DoubleTensorImpl.java
@@ -1,96 +1,143 @@
 package org.tensorflow.internal.tensor;
 
+import org.tensorflow.AbstractTensor;
 import org.tensorflow.DataType;
+import org.tensorflow.Tensor;
 import org.tensorflow.internal.c_api.TF_Tensor;
+import org.tensorflow.ndarray.DoubleNdArray;
 import org.tensorflow.ndarray.NdArray;
+import org.tensorflow.ndarray.NdArraySequence;
+import org.tensorflow.ndarray.NdArrays;
 import org.tensorflow.ndarray.Shape;
-import org.tensorflow.ndarray.buffer.ByteDataBuffer;
 import org.tensorflow.ndarray.buffer.DataBuffer;
 import org.tensorflow.ndarray.buffer.DoubleDataBuffer;
-import org.tensorflow.ndarray.impl.dense.DoubleDenseNdArray;
+import org.tensorflow.ndarray.index.Index;
 import org.tensorflow.tensor.DoubleTensor;
 
-public class DoubleTensorImpl extends DoubleDenseNdArray implements DoubleTensor {
+public class DoubleTensorImpl extends AbstractTensor<Double> implements DoubleTensor {
 
-  @Override
-  public TF_Tensor nativeHandle() {
-    return rawTensor.nativeHandle();
+  private final DoubleNdArray data;
+
+  public DoubleTensorImpl(TF_Tensor nativeHandle, DataType<?> dataType, Shape shape,
+      DoubleDataBuffer buffer) {
+    super(nativeHandle, dataType);
+    data = NdArrays.wrap(shape, buffer);
   }
 
   @Override
-  public DataType<?> dataType() {
-    return rawTensor.dataType();
+  public double getDouble(long... coordinates) {
+    requireHandle(nativeHandle);
+    return data.getDouble(coordinates);
   }
 
   @Override
-  public Shape shape() {
-    return rawTensor.shape();
+  public DoubleTensor setDouble(double value, long... coordinates) {
+    requireHandle(nativeHandle);
+    data.setDouble(value, coordinates);
+    return this;
   }
 
   @Override
-  public long numBytes() {
-    return rawTensor.numBytes();
+  public DoubleNdArray slice(Index... coordinates) {
+    requireHandle(nativeHandle);
+    return data.slice(coordinates);
   }
 
   @Override
-  public ByteDataBuffer rawData() {
-    return rawTensor.rawData();
+  public DoubleTensor get(long... coordinates) {
+    requireHandle(nativeHandle);
+    data.get(coordinates);
+    return this;
   }
 
   @Override
-  public void close() {
-    rawTensor.close();
+  public DoubleTensor set(NdArray<Double> src, long... coordinates) {
+    requireHandle(nativeHandle);
+    data.set(src, coordinates);
+    return this;
   }
 
   @Override
-  public String toString() {
-    return rawTensor.toString();
+  public Double getObject(long... coordinates) {
+    requireHandle(nativeHandle);
+    return data.getObject(coordinates);
   }
 
   @Override
-  public DoubleTensor setDouble(double value, long... coordinates) {
-    return (DoubleTensor)super.setDouble(value, coordinates);
+  public DoubleTensor setObject(Double value, long... coordinates) {
+    requireHandle(nativeHandle);
+    data.setObject(value, coordinates);
+    return this;
   }
 
   @Override
-  public DoubleTensor setObject(Double value, long... coordinates) {
-    return (DoubleTensor)super.setObject(value, coordinates);
+  public NdArraySequence<DoubleNdArray> elements(int dimensionIdx) {
+    requireHandle(nativeHandle);
+    return data.elements(dimensionIdx);
   }
 
   @Override
-  public DoubleTensor set(NdArray<Double> src, long... coordinates) {
-    return (DoubleTensor)super.set(src, coordinates);
+  public NdArraySequence<DoubleNdArray> scalars() {
+    requireHandle(nativeHandle);
+    return data.scalars();
   }
 
   @Override
   public DoubleTensor copyTo(NdArray<Double> dst) {
-    return (DoubleTensor)super.copyTo(dst);
+    requireHandle(nativeHandle);
+    data.copyTo(dst);
+    return this;
   }
 
   @Override
   public DoubleTensor read(DataBuffer<Double> dst) {
-    return (DoubleTensor)super.read(dst);
+    requireHandle(nativeHandle);
+    data.read(dst);
+    return this;
   }
 
   @Override
   public DoubleTensor read(DoubleDataBuffer dst) {
-    return (DoubleTensor)super.read(dst);
+    requireHandle(nativeHandle);
+    data.read(dst);
+    return this;
   }
 
   @Override
   public DoubleTensor write(DataBuffer<Double> src) {
-    return (DoubleTensor)super.write(src);
+    requireHandle(nativeHandle);
+    data.write(src);
+    return this;
   }
 
   @Override
   public DoubleTensor write(DoubleDataBuffer src) {
-    return (DoubleTensor)super.write(src);
+    requireHandle(nativeHandle);
+    data.write(src);
+    return this;
   }
 
-  public DoubleTensorImpl(TF_Tensor nativeHandle, DataType<?> dataType, Shape shape, DoubleDataBuffer buffer) {
-    super(buffer, shape);
-    this.rawTensor = new RawTensor(nativeHandle, dataType, shape);
+  @Override
+  public Shape shape() {
+    requireHandle(nativeHandle);
+    return data.shape();
   }
 
-  private final RawTensor rawTensor;
+  @Override
+  public int rank() {
+    requireHandle(nativeHandle);
+    return data.rank();
+  }
+
+  @Override
+  public long size() {
+    requireHandle(nativeHandle);
+    return data.size();
+  }
+
+  @Override
+  public boolean equals(Object obj) {
+    requireHandle(nativeHandle);
+    return data.equals(obj);
+  }
 }
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/FloatTensorImpl.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/FloatTensorImpl.java
index 7aa4e824123..e2d42bba37c 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/FloatTensorImpl.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/FloatTensorImpl.java
@@ -1,96 +1,142 @@
 package org.tensorflow.internal.tensor;
 
+import org.tensorflow.AbstractTensor;
 import org.tensorflow.DataType;
 import org.tensorflow.internal.c_api.TF_Tensor;
+import org.tensorflow.ndarray.FloatNdArray;
 import org.tensorflow.ndarray.NdArray;
+import org.tensorflow.ndarray.NdArraySequence;
+import org.tensorflow.ndarray.NdArrays;
 import org.tensorflow.ndarray.Shape;
-import org.tensorflow.ndarray.buffer.ByteDataBuffer;
 import org.tensorflow.ndarray.buffer.DataBuffer;
 import org.tensorflow.ndarray.buffer.FloatDataBuffer;
-import org.tensorflow.ndarray.impl.dense.FloatDenseNdArray;
+import org.tensorflow.ndarray.index.Index;
 import org.tensorflow.tensor.FloatTensor;
 
-public class FloatTensorImpl extends FloatDenseNdArray implements FloatTensor {
+public class FloatTensorImpl extends AbstractTensor<Float> implements FloatTensor {
 
-  @Override
-  public TF_Tensor nativeHandle() {
-    return rawTensor.nativeHandle();
+  private final FloatNdArray data;
+
+  public FloatTensorImpl(TF_Tensor nativeHandle, DataType<?> dataType, Shape shape,
+      FloatDataBuffer buffer) {
+    super(nativeHandle, dataType);
+    data = NdArrays.wrap(shape, buffer);
   }
 
   @Override
-  public DataType<?> dataType() {
-    return rawTensor.dataType();
+  public float getFloat(long... coordinates) {
+    requireHandle(nativeHandle);
+    return data.getFloat(coordinates);
   }
 
   @Override
-  public Shape shape() {
-    return rawTensor.shape();
+  public FloatTensor setFloat(float value, long... coordinates) {
+    requireHandle(nativeHandle);
+    data.setFloat(value, coordinates);
+    return this;
   }
 
   @Override
-  public long numBytes() {
-    return rawTensor.numBytes();
+  public FloatNdArray slice(Index... coordinates) {
+    requireHandle(nativeHandle);
+    return data.slice(coordinates);
   }
 
   @Override
-  public ByteDataBuffer rawData() {
-    return rawTensor.rawData();
+  public FloatTensor get(long... coordinates) {
+    requireHandle(nativeHandle);
+    data.get(coordinates);
+    return this;
   }
 
   @Override
-  public void close() {
-    rawTensor.close();
+  public FloatTensor set(NdArray<Float> src, long... coordinates) {
+    requireHandle(nativeHandle);
+    data.set(src, coordinates);
+    return this;
   }
 
   @Override
-  public String toString() {
-    return rawTensor.toString();
+  public Float getObject(long... coordinates) {
+    requireHandle(nativeHandle);
+    return data.getObject(coordinates);
   }
 
   @Override
-  public FloatTensor setFloat(float value, long... coordinates) {
-    return (FloatTensor)super.setFloat(value, coordinates);
+  public FloatTensor setObject(Float value, long... coordinates) {
+    requireHandle(nativeHandle);
+    data.setObject(value, coordinates);
+    return this;
   }
 
   @Override
-  public FloatTensor setObject(Float value, long... coordinates) {
-    return (FloatTensor)super.setObject(value, coordinates);
+  public NdArraySequence<FloatNdArray> elements(int dimensionIdx) {
+    requireHandle(nativeHandle);
+    return data.elements(dimensionIdx);
   }
 
   @Override
-  public FloatTensor set(NdArray<Float> src, long... coordinates) {
-    return (FloatTensor)super.set(src, coordinates);
+  public NdArraySequence<FloatNdArray> scalars() {
+    requireHandle(nativeHandle);
+    return data.scalars();
   }
 
   @Override
   public FloatTensor copyTo(NdArray<Float> dst) {
-    return (FloatTensor)super.copyTo(dst);
+    requireHandle(nativeHandle);
+    data.copyTo(dst);
+    return this;
   }
 
   @Override
   public FloatTensor read(DataBuffer<Float> dst) {
-    return (FloatTensor)super.read(dst);
+    requireHandle(nativeHandle);
+    data.read(dst);
+    return this;
   }
 
   @Override
   public FloatTensor read(FloatDataBuffer dst) {
-    return (FloatTensor)super.read(dst);
+    requireHandle(nativeHandle);
+    data.read(dst);
+    return this;
   }
 
   @Override
   public FloatTensor write(DataBuffer<Float> src) {
-    return (FloatTensor)super.write(src);
+    requireHandle(nativeHandle);
+    data.write(src);
+    return this;
   }
 
   @Override
   public FloatTensor write(FloatDataBuffer src) {
-    return (FloatTensor)super.write(src);
+    requireHandle(nativeHandle);
+    data.write(src);
+    return this;
   }
 
-  public FloatTensorImpl(TF_Tensor nativeHandle, DataType<?> dataType, Shape shape, FloatDataBuffer buffer) {
-    super(buffer, shape);
-    this.rawTensor = new RawTensor(nativeHandle, dataType, shape);
+  @Override
+  public Shape shape() {
+    requireHandle(nativeHandle);
+    return data.shape();
   }
 
-  private final RawTensor rawTensor;
+  @Override
+  public int rank() {
+    requireHandle(nativeHandle);
+    return data.rank();
+  }
+
+  @Override
+  public long size() {
+    requireHandle(nativeHandle);
+    return data.size();
+  }
+
+  @Override
+  public boolean equals(Object obj) {
+    requireHandle(nativeHandle);
+    return data.equals(obj);
+  }
 }
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/IntTensorImpl.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/IntTensorImpl.java
index 6febdda8c93..bff637e38ab 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/IntTensorImpl.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/IntTensorImpl.java
@@ -1,96 +1,143 @@
 package org.tensorflow.internal.tensor;
 
+import org.tensorflow.AbstractTensor;
 import org.tensorflow.DataType;
+import org.tensorflow.Tensor;
 import org.tensorflow.internal.c_api.TF_Tensor;
+import org.tensorflow.ndarray.IntNdArray;
 import org.tensorflow.ndarray.NdArray;
+import org.tensorflow.ndarray.NdArraySequence;
+import org.tensorflow.ndarray.NdArrays;
 import org.tensorflow.ndarray.Shape;
-import org.tensorflow.ndarray.buffer.ByteDataBuffer;
 import org.tensorflow.ndarray.buffer.DataBuffer;
 import org.tensorflow.ndarray.buffer.IntDataBuffer;
-import org.tensorflow.ndarray.impl.dense.IntDenseNdArray;
+import org.tensorflow.ndarray.index.Index;
 import org.tensorflow.tensor.IntTensor;
 
-public class IntTensorImpl extends IntDenseNdArray implements IntTensor {
+public class IntTensorImpl extends AbstractTensor<Integer> implements IntTensor {
 
-  @Override
-  public TF_Tensor nativeHandle() {
-    return rawTensor.nativeHandle();
+  private final IntNdArray data;
+
+  public IntTensorImpl(TF_Tensor nativeHandle, DataType<?> dataType, Shape shape,
+      IntDataBuffer buffer) {
+    super(nativeHandle, dataType);
+    data = NdArrays.wrap(shape, buffer);
   }
 
   @Override
-  public DataType<?> dataType() {
-    return rawTensor.dataType();
+  public int getInt(long... coordinates) {
+    requireHandle(nativeHandle);
+    return data.getInt(coordinates);
   }
 
   @Override
-  public Shape shape() {
-    return rawTensor.shape();
+  public IntTensor setInt(int value, long... coordinates) {
+    requireHandle(nativeHandle);
+    data.setInt(value, coordinates);
+    return this;
   }
 
   @Override
-  public long numBytes() {
-    return rawTensor.numBytes();
+  public IntNdArray slice(Index... coordinates) {
+    requireHandle(nativeHandle);
+    return data.slice(coordinates);
   }
 
   @Override
-  public ByteDataBuffer rawData() {
-    return rawTensor.rawData();
+  public IntTensor get(long... coordinates) {
+    requireHandle(nativeHandle);
+    data.get(coordinates);
+    return this;
   }
 
   @Override
-  public void close() {
-    rawTensor.close();
+  public IntTensor set(NdArray<Integer> src, long... coordinates) {
+    requireHandle(nativeHandle);
+    data.set(src, coordinates);
+    return this;
   }
 
   @Override
-  public String toString() {
-    return rawTensor.toString();
+  public Integer getObject(long... coordinates) {
+    requireHandle(nativeHandle);
+    return data.getObject(coordinates);
   }
 
   @Override
-  public IntTensor setInt(int value, long... coordinates) {
-    return (IntTensor)super.setInt(value, coordinates);
+  public IntTensor setObject(Integer value, long... coordinates) {
+    requireHandle(nativeHandle);
+    data.setObject(value, coordinates);
+    return this;
   }
 
   @Override
-  public IntTensor setObject(Integer value, long... coordinates) {
-    return (IntTensor)super.setObject(value, coordinates);
+  public NdArraySequence<IntNdArray> elements(int dimensionIdx) {
+    requireHandle(nativeHandle);
+    return data.elements(dimensionIdx);
   }
 
   @Override
-  public IntTensor set(NdArray<Integer> src, long... coordinates) {
-    return (IntTensor)super.set(src, coordinates);
+  public NdArraySequence<IntNdArray> scalars() {
+    requireHandle(nativeHandle);
+    return data.scalars();
   }
 
   @Override
   public IntTensor copyTo(NdArray<Integer> dst) {
-    return (IntTensor)super.copyTo(dst);
+    requireHandle(nativeHandle);
+    data.copyTo(dst);
+    return this;
   }
 
   @Override
   public IntTensor read(DataBuffer<Integer> dst) {
-    return (IntTensor)super.read(dst);
+    requireHandle(nativeHandle);
+    data.read(dst);
+    return this;
   }
 
   @Override
   public IntTensor read(IntDataBuffer dst) {
-    return (IntTensor)super.read(dst);
+    requireHandle(nativeHandle);
+    data.read(dst);
+    return this;
   }
 
   @Override
   public IntTensor write(DataBuffer<Integer> src) {
-    return (IntTensor)super.write(src);
+    requireHandle(nativeHandle);
+    data.write(src);
+    return this;
   }
 
   @Override
   public IntTensor write(IntDataBuffer src) {
-    return (IntTensor)super.write(src);
+    requireHandle(nativeHandle);
+    data.write(src);
+    return this;
   }
 
-  public IntTensorImpl(TF_Tensor nativeHandle, DataType<?> dataType, Shape shape, IntDataBuffer buffer) {
-    super(buffer, shape);
-    this.rawTensor = new RawTensor(nativeHandle, dataType, shape);
+  @Override
+  public Shape shape() {
+    requireHandle(nativeHandle);
+    return data.shape();
   }
 
-  private final RawTensor rawTensor;
+  @Override
+  public int rank() {
+    requireHandle(nativeHandle);
+    return data.rank();
+  }
+
+  @Override
+  public long size() {
+    requireHandle(nativeHandle);
+    return data.size();
+  }
+
+  @Override
+  public boolean equals(Object obj) {
+    requireHandle(nativeHandle);
+    return data.equals(obj);
+  }
 }
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/LongTensorImpl.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/LongTensorImpl.java
index 7dc68cf58ce..8b3436b4e77 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/LongTensorImpl.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/LongTensorImpl.java
@@ -1,96 +1,143 @@
 package org.tensorflow.internal.tensor;
 
+import org.tensorflow.AbstractTensor;
 import org.tensorflow.DataType;
+import org.tensorflow.Tensor;
 import org.tensorflow.internal.c_api.TF_Tensor;
+import org.tensorflow.ndarray.LongNdArray;
 import org.tensorflow.ndarray.NdArray;
+import org.tensorflow.ndarray.NdArraySequence;
+import org.tensorflow.ndarray.NdArrays;
 import org.tensorflow.ndarray.Shape;
-import org.tensorflow.ndarray.buffer.ByteDataBuffer;
 import org.tensorflow.ndarray.buffer.DataBuffer;
 import org.tensorflow.ndarray.buffer.LongDataBuffer;
-import org.tensorflow.ndarray.impl.dense.LongDenseNdArray;
+import org.tensorflow.ndarray.index.Index;
 import org.tensorflow.tensor.LongTensor;
 
-public class LongTensorImpl extends LongDenseNdArray implements LongTensor {
+public class LongTensorImpl extends AbstractTensor<Long> implements LongTensor {
 
-  @Override
-  public TF_Tensor nativeHandle() {
-    return rawTensor.nativeHandle();
+  private final LongNdArray data;
+
+  public LongTensorImpl(TF_Tensor nativeHandle, DataType<?> dataType, Shape shape,
+      LongDataBuffer buffer) {
+    super(nativeHandle, dataType);
+    data = NdArrays.wrap(shape, buffer);
   }
 
   @Override
-  public DataType<?> dataType() {
-    return rawTensor.dataType();
+  public long getLong(long... coordinates) {
+    requireHandle(nativeHandle);
+    return data.getLong(coordinates);
   }
 
   @Override
-  public Shape shape() {
-    return rawTensor.shape();
+  public LongTensor setLong(long value, long... coordinates) {
+    requireHandle(nativeHandle);
+    data.setLong(value, coordinates);
+    return this;
   }
 
   @Override
-  public long numBytes() {
-    return rawTensor.numBytes();
+  public LongNdArray slice(Index... coordinates) {
+    requireHandle(nativeHandle);
+    return data.slice(coordinates);
   }
 
   @Override
-  public ByteDataBuffer rawData() {
-    return rawTensor.rawData();
+  public LongTensor get(long... coordinates) {
+    requireHandle(nativeHandle);
+    data.get(coordinates);
+    return this;
   }
 
   @Override
-  public void close() {
-    rawTensor.close();
+  public LongTensor set(NdArray<Long> src, long... coordinates) {
+    requireHandle(nativeHandle);
+    data.set(src, coordinates);
+    return this;
   }
 
   @Override
-  public String toString() {
-    return rawTensor.toString();
+  public Long getObject(long... coordinates) {
+    requireHandle(nativeHandle);
+    return data.getObject(coordinates);
   }
 
   @Override
-  public LongTensor setLong(long value, long... coordinates) {
-    return (LongTensor)super.setLong(value, coordinates);
+  public LongTensor setObject(Long value, long... coordinates) {
+    requireHandle(nativeHandle);
+    data.setObject(value, coordinates);
+    return this;
   }
 
   @Override
-  public LongTensor setObject(Long value, long... coordinates) {
-    return (LongTensor)super.setObject(value, coordinates);
+  public NdArraySequence<LongNdArray> elements(int dimensionIdx) {
+    requireHandle(nativeHandle);
+    return data.elements(dimensionIdx);
   }
 
   @Override
-  public LongTensor set(NdArray<Long> src, long... coordinates) {
-    return (LongTensor)super.set(src, coordinates);
+  public NdArraySequence<LongNdArray> scalars() {
+    requireHandle(nativeHandle);
+    return data.scalars();
   }
 
   @Override
   public LongTensor copyTo(NdArray<Long> dst) {
-    return (LongTensor)super.copyTo(dst);
+    requireHandle(nativeHandle);
+    data.copyTo(dst);
+    return this;
   }
 
   @Override
   public LongTensor read(DataBuffer<Long> dst) {
-    return (LongTensor)super.read(dst);
+    requireHandle(nativeHandle);
+    data.read(dst);
+    return this;
   }
 
   @Override
   public LongTensor read(LongDataBuffer dst) {
-    return (LongTensor)super.read(dst);
+    requireHandle(nativeHandle);
+    data.read(dst);
+    return this;
   }
 
   @Override
   public LongTensor write(DataBuffer<Long> src) {
-    return (LongTensor)super.write(src);
+    requireHandle(nativeHandle);
+    data.write(src);
+    return this;
   }
 
   @Override
   public LongTensor write(LongDataBuffer src) {
-    return (LongTensor)super.write(src);
+    requireHandle(nativeHandle);
+    data.write(src);
+    return this;
   }
 
-  public LongTensorImpl(TF_Tensor nativeHandle, DataType<?> dataType, Shape shape, LongDataBuffer buffer) {
-    super(buffer, shape);
-    this.rawTensor = new RawTensor(nativeHandle, dataType, shape);
+  @Override
+  public Shape shape() {
+    requireHandle(nativeHandle);
+    return data.shape();
   }
 
-  private final RawTensor rawTensor;
+  @Override
+  public int rank() {
+    requireHandle(nativeHandle);
+    return data.rank();
+  }
+
+  @Override
+  public long size() {
+    requireHandle(nativeHandle);
+    return data.size();
+  }
+
+  @Override
+  public boolean equals(Object obj) {
+    requireHandle(nativeHandle);
+    return data.equals(obj);
+  }
 }
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/RawTensor.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/RawTensor.java
deleted file mode 100644
index 98b9720b727..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/RawTensor.java
+++ /dev/null
@@ -1,58 +0,0 @@
-package org.tensorflow.internal.tensor;
-
-import static org.tensorflow.internal.c_api.global.tensorflow.TF_TensorByteSize;
-
-import org.bytedeco.javacpp.PointerScope;
-import org.tensorflow.DataType;
-import org.tensorflow.internal.tensor.buffer.TensorBuffers;
-import org.tensorflow.internal.c_api.TF_Tensor;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.ndarray.buffer.ByteDataBuffer;
-
-class RawTensor<T> {
-
-  public TF_Tensor nativeHandle() {
-    if (nativeHandle.isNull()) {
-      throw new IllegalStateException("Tensor has been already released");
-    }
-    return nativeHandle;
-  }
-
-  public DataType<?> dataType() {
-    return dtype;
-  }
-
-  public Shape shape() {
-    return shape;
-  }
-
-  public long numBytes() {
-    return TF_TensorByteSize(nativeHandle);
-  }
-
-  public ByteDataBuffer rawData() {
-    return TensorBuffers.toBytes(nativeHandle, true);
-  }
-
-  public void close() {
-    tensorScope.close();
-  }
-
-  /** Returns a string describing the type and shape of the Tensor. */
-  @Override
-  public String toString() {
-    return String.format("%s tensor with shape %s", dataType().toString(), shape());
-  }
-
-  private final PointerScope tensorScope = new PointerScope();
-  private final TF_Tensor nativeHandle;
-  private final DataType<?> dtype;
-  private final Shape shape;
-
-  RawTensor(TF_Tensor nativeHandle, DataType<?> dtype, Shape shape) {
-    this.nativeHandle = nativeHandle;
-    this.dtype = dtype;
-    this.shape = shape;
-    this.tensorScope.attach(nativeHandle);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/StringTensorImpl.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/StringTensorImpl.java
index 3ee3966a22f..4a1cd3333e5 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/StringTensorImpl.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/StringTensorImpl.java
@@ -1,101 +1,132 @@
 package org.tensorflow.internal.tensor;
 
+import org.tensorflow.AbstractTensor;
 import org.tensorflow.DataType;
 import org.tensorflow.Tensor;
-import org.tensorflow.internal.tensor.buffer.ByteSequenceTensorBuffer;
 import org.tensorflow.internal.c_api.TF_Tensor;
+import org.tensorflow.internal.tensor.buffer.ByteSequenceTensorBuffer;
 import org.tensorflow.ndarray.NdArray;
+import org.tensorflow.ndarray.NdArraySequence;
 import org.tensorflow.ndarray.NdArrays;
 import org.tensorflow.ndarray.Shape;
-import org.tensorflow.ndarray.buffer.ByteDataBuffer;
 import org.tensorflow.ndarray.buffer.DataBuffer;
 import org.tensorflow.ndarray.buffer.layout.DataLayout;
-import org.tensorflow.ndarray.impl.dense.DenseNdArray;
+import org.tensorflow.ndarray.index.Index;
 import org.tensorflow.tensor.StringTensor;
 
-public class StringTensorImpl extends DenseNdArray<String> implements StringTensor {
+public class StringTensorImpl extends AbstractTensor<String> implements StringTensor {
+
+  private final ByteSequenceTensorBuffer rawBuffer;
+  private final NdArray<String> data;
+
+  public StringTensorImpl(
+      TF_Tensor nativeHandle,
+      DataType<?> dataType,
+      Shape shape,
+      DataLayout<DataBuffer<byte[]>, String> layout,
+      ByteSequenceTensorBuffer buffer
+  ) {
+    super(nativeHandle, dataType);
+    this.rawBuffer = buffer;
+    data = NdArrays.wrap(shape, layout.applyTo(buffer));
+  }
 
   @Override
   public NdArray<byte[]> asBytes() {
+    requireHandle(nativeHandle);
     return NdArrays.wrap(shape(), rawBuffer);
   }
 
   @Override
-  public TF_Tensor nativeHandle() {
-    return rawTensor.nativeHandle();
+  public NdArraySequence<? extends NdArray<String>> elements(int dimensionIdx) {
+    requireHandle(nativeHandle);
+    return data.elements(dimensionIdx);
   }
 
   @Override
-  public DataType<Tensor<String>> dataType() {
-    return rawTensor.dataType();
+  public NdArraySequence<? extends NdArray<String>> scalars() {
+    requireHandle(nativeHandle);
+    return data.scalars();
   }
 
   @Override
-  public Shape shape() {
-    return rawTensor.shape();
+  public NdArray<String> slice(Index... indices) {
+    requireHandle(nativeHandle);
+    return data.slice(indices);
   }
 
   @Override
-  public long numBytes() {
-    return rawTensor.numBytes();
+  public NdArray<String> get(long... coordinates) {
+    requireHandle(nativeHandle);
+    return data.get(coordinates);
   }
 
   @Override
-  public ByteDataBuffer rawData() {
-    return rawTensor.rawData();
+  public StringTensor set(NdArray<String> src, long... coordinates) {
+    requireHandle(nativeHandle);
+    data.set(src, coordinates);
+    return this;
   }
 
   @Override
-  public void close() {
-    rawTensor.close();
+  public String getObject(long... coordinates) {
+    requireHandle(nativeHandle);
+    return data.getObject(coordinates);
   }
 
   @Override
-  public String toString() {
-    return rawTensor.toString();
+  public StringTensor setObject(String value, long... coordinates) {
+    requireHandle(nativeHandle);
+    data.setObject(value, coordinates);
+    return this;
   }
 
   @Override
-  public Tensor<String> setObject(String value, long... coordinates) {
-    return (Tensor<String>)super.setObject(value, coordinates);
+  public StringTensor copyTo(NdArray<String> dst) {
+    requireHandle(nativeHandle);
+    data.copyTo(dst);
+    return this;
   }
 
   @Override
-  public Tensor<String> set(NdArray<String> src, long... coordinates) {
-    return (Tensor<String>)super.set(src, coordinates);
+  public StringTensor read(DataBuffer<String> dst) {
+    requireHandle(nativeHandle);
+    data.read(dst);
+    return this;
   }
 
   @Override
-  public Tensor<String> copyTo(NdArray<String> dst) {
-    return (Tensor<String>)super.copyTo(dst);
+  public StringTensor write(DataBuffer<String> src) {
+    requireHandle(nativeHandle);
+    data.write(src);
+    return this;
   }
 
   @Override
-  public Tensor<String> read(DataBuffer<String> dst) {
-    return (Tensor<String>)super.read(dst);
+  public Shape shape() {
+    requireHandle(nativeHandle);
+    return data.shape();
   }
 
   @Override
-  public Tensor<String> write(DataBuffer<String> src) {
-    return (Tensor<String>)super.write(src);
+  public int rank() {
+    requireHandle(nativeHandle);
+    return data.rank();
   }
 
-  protected ByteSequenceTensorBuffer rawBuffer() {
-    return rawBuffer;
+  @Override
+  public long size() {
+    requireHandle(nativeHandle);
+    return data.size();
   }
 
-  public StringTensorImpl(
-      TF_Tensor nativeHandle,
-      DataType<?> dataType,
-      Shape shape,
-      DataLayout<DataBuffer<byte[]>, String> layout,
-      ByteSequenceTensorBuffer buffer
-  ) {
-    super(layout.applyTo(buffer), shape);
-    this.rawTensor = new RawTensor(nativeHandle, dataType, shape);
-    this.rawBuffer = buffer;
+  @Override
+  public boolean equals(Object obj) {
+    requireHandle(nativeHandle);
+    return data.equals(obj);
   }
 
-  private final RawTensor rawTensor;
-  private final ByteSequenceTensorBuffer rawBuffer;
+  protected ByteSequenceTensorBuffer rawBuffer() {
+    return rawBuffer;
+  }
 }
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/op/Operands.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/op/Operands.java
index ac48da80326..5496eaa3d15 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/op/Operands.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/op/Operands.java
@@ -30,13 +30,18 @@ public final class Operands {
    * <p>Operation wrappers need to convert back a list of inputs into an array of outputs in order
    * to build an operation, see {@link OperationBuilder#addInputList(Output[])}.
    *
+   * <p>If a valid non-null value is provided as the {@code scope} of this operand, the implementor
+   * can program additional computations before returning its output. This might be required if,
+   * for instance, the implementor has not yet been added as a node to the graph.</p>
+   *
+   * @param scope if non-null, scope that can be used an operand to produce a new output
    * @param inputs an iteration of input operands
    * @return an array of outputs
    */
-  public static Output<?>[] asOutputs(Iterable<? extends Operand<?>> inputs) {
+  public static Output<?>[] asOutputs(Scope scope, Iterable<? extends Operand<?>> inputs) {
     List<Output<?>> outputList = new ArrayList<>();
     for (Operand<?> input : inputs) {
-      outputList.add(input.asOutput());
+      outputList.add(input.asOutput(scope));
     }
     return outputList.toArray(new Output<?>[outputList.size()]);
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/op/core/Constant.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/op/core/Constant.java
index 2fc69dd9188..e7ce5158ff0 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/op/core/Constant.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/op/core/Constant.java
@@ -21,6 +21,7 @@
 import org.tensorflow.Operation;
 import org.tensorflow.Output;
 import org.tensorflow.Tensor;
+import org.tensorflow.Tensors;
 import org.tensorflow.ndarray.BooleanNdArray;
 import org.tensorflow.ndarray.ByteNdArray;
 import org.tensorflow.ndarray.DoubleNdArray;
@@ -42,6 +43,7 @@
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.tensor.BooleanTensor;
 import org.tensorflow.types.TBool;
 import org.tensorflow.types.TFloat32;
 import org.tensorflow.types.TFloat64;
@@ -66,7 +68,7 @@
  * }</pre>
  */
 @Operator
-public final class Constant<T extends Tensor> extends RawOp implements Operand<T> {
+public final class Constant<T extends TType> extends RawOp implements Operand<T> {
 
   /**
    * Creates a constant containing a single {@code int} element.
@@ -1004,9 +1006,9 @@ public static Constant<TUint8> tensorOf(Scope scope, Shape shape, ByteDataBuffer
    *     buffer
    */
   @Endpoint
-  public static <T extends Tensor & TType> Constant<T> tensorOf(Scope scope, DataType<T> type, Shape shape,
+  public static <T extends TType> Constant<T> tensorOf(Scope scope, DataType<T> type, Shape shape,
       ByteDataBuffer data) {
-    try (T value = Tensor.of(type, shape, data)) {
+    try (T value = Tensors.of(type, shape, data)) {
       return create(scope, value);
     }
   }
@@ -1263,8 +1265,8 @@ public static Constant<TInt64> tensorOf(Scope scope, Shape shape) {
    * @param tensor a Tensor holding the constant value
    * @return a constant of the same data type as `tensor`
    */
-  @Endpoint
-  public static <T extends Tensor & TType> Constant<T> create(Scope scope, T tensor) {
+  @Endpoint(name = "capture")
+  public static <T extends TType> Constant<T> create(Scope scope, T tensor) {
     return new Constant<>(
         scope
             .env()
@@ -1275,7 +1277,7 @@ public static <T extends Tensor & TType> Constant<T> create(Scope scope, T tenso
   }
 
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
 
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/op/core/Gradients.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/op/core/Gradients.java
index f5192697628..bda5f97cd96 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/op/core/Gradients.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/op/core/Gradients.java
@@ -27,6 +27,7 @@
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Adds operations to compute the partial derivatives of sum of {@code y}s w.r.t {@code x}s,
@@ -96,13 +97,15 @@ public static Gradients create(
     if (options != null) {
       for (Options opts : options) {
         if (opts.dx != null) {
-          dx = Operands.asOutputs(opts.dx);
+          dx = Operands.asOutputs(scope, opts.dx);
         }
       }
     }
-    Output<?>[] dy =
-        graph.addGradients(
-            scope.makeOpName("Gradients"), Operands.asOutputs(y), Operands.asOutputs(x), dx);
+    Output<?>[] dy = graph.addGradients(
+        scope.makeOpName("Gradients"),
+        Operands.asOutputs(scope, y),
+        Operands.asOutputs(scope, x), dx
+    );
     return new Gradients(Arrays.asList(dy));
   }
 
@@ -157,7 +160,7 @@ public List<Output<?>> dy() {
    * @param index The index of the output among the gradients added by this operation
    */
   @SuppressWarnings("unchecked")
-  public <T extends Tensor> Output<T> dy(int index) {
+  public <T extends TType> Output<T> dy(int index) {
     return (Output<T>) dy.get(index);
   }
 
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/op/core/Helpers.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/op/core/Helpers.java
index 123f7eb826b..f9ce837fe60 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/op/core/Helpers.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/op/core/Helpers.java
@@ -17,10 +17,10 @@
 
 import org.tensorflow.Operand;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
 
 /**
  * Container class for core methods which add or perform several operations
@@ -45,7 +45,7 @@ private Helpers() {}
      * @return a new instance of Variable
      */
     @Endpoint(name = "variable")
-    public static <T extends Tensor> Variable<T> createVariableWithInit(Scope scope, Operand<T> init, Variable.Options... options) {
+    public static <T extends TType> Variable<T> createVariableWithInit(Scope scope, Operand<T> init, Variable.Options... options) {
         Output<T> initOutput = init.asOutput();
         Variable<T> newVar = Variable.create(scope,initOutput.shape(), initOutput.dataType(), options);
         Assign<T> assignOp = Assign.create(scope, newVar, init);
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/op/core/Shapes.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/op/core/Shapes.java
index 14bc0604ce6..a2dc0cc6b5d 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/op/core/Shapes.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/op/core/Shapes.java
@@ -15,23 +15,21 @@
 package org.tensorflow.op.core;
 
 import java.util.Arrays;
-
 import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Tensor;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-
-import org.tensorflow.op.math.FloorMod;
-
 import org.tensorflow.op.dtypes.Cast;
+import org.tensorflow.op.math.FloorMod;
 import org.tensorflow.op.math.NotEqual;
 import org.tensorflow.op.math.Sub;
 import org.tensorflow.types.TBool;
 import org.tensorflow.types.TInt32;
 import org.tensorflow.types.TInt64;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * An operator providing methods on org.tensorflow.op.core.Shape tensors and 1d operands that
@@ -67,7 +65,7 @@ public abstract class Shapes {
    * @return the reshaped operand
    */
   @Endpoint(name = "flatten")
-  public static <T extends Tensor> Operand<T> flatten(Scope scope, Operand<T> operand) {
+  public static <T extends TType> Operand<T> flatten(Scope scope, Operand<T> operand) {
     return flatten(scope, operand, TInt32.DTYPE);
   }
 
@@ -82,7 +80,7 @@ public static <T extends Tensor> Operand<T> flatten(Scope scope, Operand<T> oper
    * @return the reshaped operand
    */
   @Endpoint(name = "flatten")
-  public static <T extends Tensor, U extends Tensor & TNumber> Operand<T> flatten(
+  public static <T extends TType, U extends TNumber> Operand<T> flatten(
       Scope scope, Operand<T> operand, DataType<U> dType) {
     Operand<U> flatShape = flatten(scope, Shape.create(scope, operand, dType), dType);
     return Reshape.create(scope, operand, flatShape);
@@ -110,7 +108,7 @@ public static Operand<TInt32> flatten(Scope scope, Shape<TInt32> shape) {
    * @return the flattened shape
    */
   @Endpoint(name = "flatten")
-  public static <U extends Tensor & TNumber> Operand<U> flatten(
+  public static <U extends TNumber> Operand<U> flatten(
       Scope scope, Shape<U> shape, DataType<U> dType) {
     return ExpandDims.create(
         scope,
@@ -140,7 +138,7 @@ public static Operand<TInt32> size(Scope scope, Shape<TInt32> shape) {
    * @return the size
    */
   @Endpoint(name = "size")
-  public static <U extends Tensor & TNumber> Operand<U> size(
+  public static <U extends TNumber> Operand<U> size(
       Scope scope, Shape<U> shape, DataType<U> dType) {
     Slice<U> dims =
         Slice.create(
@@ -178,7 +176,7 @@ public static Operand<TInt32> size(Scope scope, Shape<TInt32> shape, Operand<TIn
    * @return the size of the specified dimension
    */
   @Endpoint(name = "size")
-  public static <U extends Tensor & TNumber> Operand<U> size(
+  public static <U extends TNumber> Operand<U> size(
       Scope scope, Shape<U> shape, Operand<U> dim, DataType<U> dType) {
     return Slice.create(
         scope,
@@ -199,7 +197,7 @@ public static <U extends Tensor & TNumber> Operand<U> size(
    * @return the size of the specified dimension
    */
   @Endpoint(name = "size")
-  public static <T extends Tensor> Operand<TInt32> size(
+  public static <T extends TType> Operand<TInt32> size(
       Scope scope, Operand<T> input, Operand<TInt32> dim) {
     return size(scope, input, dim, TInt32.DTYPE);
   }
@@ -215,7 +213,7 @@ public static <T extends Tensor> Operand<TInt32> size(
    * @return the size of the specified dimension
    */
   @Endpoint(name = "size")
-  public static <T extends Tensor, U extends Tensor & TNumber> Operand<U> size(
+  public static <T extends TType, U extends TNumber> Operand<U> size(
       Scope scope, Operand<T> input, Operand<U> dim, DataType<U> dType) {
     return size(scope, Shape.create(scope, input, dType), dim, dType);
   }
@@ -242,7 +240,7 @@ public static Operand<TInt32> numDimensions(Scope scope, Shape<TInt32> shape) {
    * @return the number of dimensions
    */
   @Endpoint(name = "numDimensions")
-  public static <U extends Tensor & TNumber> Operand<U> numDimensions(
+  public static <U extends TNumber> Operand<U> numDimensions(
       Scope scope, Shape<U> shape, DataType<U> dType) {
     return Size.create(scope, shape, dType);
   }
@@ -257,7 +255,7 @@ public static <U extends Tensor & TNumber> Operand<U> numDimensions(
    * @return the reshaped operand
    */
   @Endpoint(name = "reduceDims")
-  public static <T extends Tensor> Operand<T> reduceDims(
+  public static <T extends TType> Operand<T> reduceDims(
       Scope scope, Operand<T> operand, Operand<TInt32> axis) {
     return reduceDims(scope, operand, axis, TInt32.DTYPE);
   }
@@ -274,7 +272,7 @@ public static <T extends Tensor> Operand<T> reduceDims(
    * @return the reshaped operand
    */
   @Endpoint(name = "reduceDims")
-  public static <T extends Tensor, U extends Tensor & TNumber> Operand<T> reduceDims(
+  public static <T extends TType, U extends TNumber> Operand<T> reduceDims(
       Scope scope, Operand<T> operand, Operand<U> axis, DataType<U> dType) {
     Shape<U> newShape = Shape.create(scope, operand, dType);
     return Reshape.create(scope, operand, reduceDims(scope, newShape, axis, dType));
@@ -304,7 +302,7 @@ public static Operand<TInt32> reduceDims(Scope scope, Shape<TInt32> shape, Opera
    * @return the reduced shape
    */
   @Endpoint(name = "reduceDims")
-  public static <U extends Tensor & TNumber> Operand<U> reduceDims(
+  public static <U extends TNumber> Operand<U> reduceDims(
       Scope scope, Shape<U> shape, Operand<U> axis, DataType<U> dType) {
     Size<U> rank = Size.create(scope, shape, dType);
     axis = FloorMod.create(scope, axis, rank);
@@ -356,7 +354,7 @@ public static Operand<TInt32> squeeze(Scope scope, Shape<TInt32> shape) {
    * @return the squeezed shape
    */
   @Endpoint(name = "squeeze")
-  public static <U extends Tensor & TNumber> Operand<U> squeeze(
+  public static <U extends TNumber> Operand<U> squeeze(
       Scope scope, Shape<U> shape, DataType<U> dType) {
     Operand<TBool> mask =
         NotEqual.create(scope, shape, Cast.create(scope, OnesLike.create(scope, shape), dType));
@@ -386,7 +384,7 @@ public static Operand<TInt32> head(Scope scope, Shape<TInt32> shape) {
    * @return a 1-dimensional Operand containing the Shape's first dimension
    */
   @Endpoint(name = "head")
-  public static <U extends Tensor & TNumber> Operand<U> head(
+  public static <U extends TNumber> Operand<U> head(
       Scope scope, Shape<U> shape, DataType<U> dType) {
     return take(scope, shape, Cast.create(scope, Constant.scalarOf(scope, 1), dType), dType);
   }
@@ -419,7 +417,7 @@ public static Operand<TInt32> take(Scope scope, Shape<TInt32> shape, Operand<TIn
    *     shape
    */
   @Endpoint(name = "take")
-  public static <U extends Tensor & TNumber> Operand<U> take(
+  public static <U extends TNumber> Operand<U> take(
       Scope scope, Shape<U> shape, Operand<U> n, DataType<U> dType) {
     return Slice.create(
         scope,
@@ -454,7 +452,7 @@ public static Operand<TInt32> tail(Scope scope, Shape<TInt32> shape) {
    *     Shape
    */
   @Endpoint(name = "tail")
-  public static <U extends Tensor & TNumber> Operand<U> tail(
+  public static <U extends TNumber> Operand<U> tail(
       Scope scope, Shape<U> shape, DataType<U> dType) {
     return takeLast(scope, shape, Cast.create(scope, Constant.scalarOf(scope, 1), dType), dType);
   }
@@ -470,7 +468,7 @@ public static <U extends Tensor & TNumber> Operand<U> tail(
    *     shape
    */
   @Endpoint(name = "takeLast")
-  public static <U extends Tensor & TNumber> Operand<TInt32> takeLast(
+  public static <U extends TNumber> Operand<TInt32> takeLast(
       Scope scope, Shape<TInt32> shape, Operand<TInt32> n) {
     return takeLast(scope, shape, n, TInt32.DTYPE);
   }
@@ -488,7 +486,7 @@ public static <U extends Tensor & TNumber> Operand<TInt32> takeLast(
    *     shape
    */
   @Endpoint(name = "takeLast")
-  public static <U extends Tensor & TNumber> Operand<U> takeLast(
+  public static <U extends TNumber> Operand<U> takeLast(
       Scope scope, Shape<U> shape, Operand<U> n, DataType<U> dType) {
 
     Size<U> rank = Size.create(scope, shape, dType);
@@ -548,7 +546,7 @@ public static Operand<TInt64> prepend(Scope scope, Shape<TInt64> shape, long fir
    *     representing the shape
    */
   @Endpoint(name = "prepend")
-  public static <T extends Tensor & TNumber> Operand<T> prepend(
+  public static <T extends TNumber> Operand<T> prepend(
       Scope scope, Operand<T> shape, Operand<T> shapeToPrepend) {
 
     return Concat.create(scope, Arrays.asList(shapeToPrepend, shape), Constant.scalarOf(scope, 0));
@@ -600,7 +598,7 @@ public static Operand<TInt64> append(Scope scope, Shape<TInt64> shape, long last
    *     to append
    */
   @Endpoint(name = "append")
-  public static <T extends Tensor & TNumber> Operand<T> append(
+  public static <T extends TNumber> Operand<T> append(
       Scope scope, Operand<T> shape, Operand<T> shapeToAppend) {
     return Concat.create(scope, Arrays.asList(shape, shapeToAppend), Constant.scalarOf(scope, 0));
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/op/core/Zeros.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/op/core/Zeros.java
index 389be13e59d..db8fb8b9671 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/op/core/Zeros.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/op/core/Zeros.java
@@ -18,7 +18,6 @@
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.Op;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
@@ -26,6 +25,7 @@
 import org.tensorflow.op.dtypes.Cast;
 import org.tensorflow.types.TString;
 import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
 
 /**
  * An operator creating a constant initialized with zeros of the shape given by `dims`.
@@ -38,7 +38,7 @@
  * @param <T> constant type
  */
 @Operator
-public final class Zeros<T extends Tensor> implements Op, Operand<T> {
+public final class Zeros<T extends TType> implements Op, Operand<T> {
 
   /**
    * Creates a zeroed tensor given its type and shape.
@@ -51,7 +51,7 @@ public final class Zeros<T extends Tensor> implements Op, Operand<T> {
    */
   @Endpoint
   @SuppressWarnings("unchecked")
-  public static <T extends Tensor, U extends Tensor & TNumber> Zeros<T> create(Scope scope, Operand<U> dims, DataType<T> type) {
+  public static <T extends TType, U extends TNumber> Zeros<T> create(Scope scope, Operand<U> dims, DataType<T> type) {
     Scope zerosScope = scope.withSubScope("Zeros");
     Operand<T> zero;
     if (type == TString.DTYPE) {
@@ -68,7 +68,7 @@ public Operation op() {
   }
 
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return fill.asOutput();
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/tensor/BooleanTensor.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/tensor/BooleanTensor.java
index ac270e7ad0e..106e18de8f6 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/tensor/BooleanTensor.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/tensor/BooleanTensor.java
@@ -1,16 +1,14 @@
-package org.tensorflow.tensor;
+package org.tensorflow.types.tensor;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Tensor;
-import org.tensorflow.internal.c_api.TF_Tensor;
 import org.tensorflow.ndarray.BooleanNdArray;
 import org.tensorflow.ndarray.NdArray;
-import org.tensorflow.ndarray.Shape;
 import org.tensorflow.ndarray.buffer.BooleanDataBuffer;
-import org.tensorflow.ndarray.buffer.ByteDataBuffer;
 import org.tensorflow.ndarray.buffer.DataBuffer;
-import org.tensorflow.ndarray.impl.dense.BooleanDenseNdArray;
 
+/**
+ * A {@link Tensor} of booleans.
+ */
 public interface BooleanTensor extends BooleanNdArray, Tensor<Boolean> {
 
   @Override
@@ -37,4 +35,3 @@ public interface BooleanTensor extends BooleanNdArray, Tensor<Boolean> {
   @Override
   BooleanTensor write(BooleanDataBuffer src);
 }
-
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/tensor/ByteTensor.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/tensor/ByteTensor.java
index ba18867765b..16b7825422d 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/tensor/ByteTensor.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/tensor/ByteTensor.java
@@ -1,15 +1,14 @@
-package org.tensorflow.tensor;
+package org.tensorflow.types.tensor;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Tensor;
-import org.tensorflow.internal.c_api.TF_Tensor;
 import org.tensorflow.ndarray.ByteNdArray;
 import org.tensorflow.ndarray.NdArray;
-import org.tensorflow.ndarray.Shape;
 import org.tensorflow.ndarray.buffer.ByteDataBuffer;
 import org.tensorflow.ndarray.buffer.DataBuffer;
-import org.tensorflow.ndarray.impl.dense.ByteDenseNdArray;
 
+/**
+ * A {@link Tensor} of bytes.
+ */
 public interface ByteTensor extends ByteNdArray, Tensor<Byte> {
 
   @Override
@@ -36,4 +35,3 @@ public interface ByteTensor extends ByteNdArray, Tensor<Byte> {
   @Override
   ByteTensor write(ByteDataBuffer src);
 }
-
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/tensor/DoubleTensor.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/tensor/DoubleTensor.java
index 4f8b1b5b0c2..5f6036fa08f 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/tensor/DoubleTensor.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/tensor/DoubleTensor.java
@@ -1,16 +1,14 @@
-package org.tensorflow.tensor;
+package org.tensorflow.types.tensor;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Tensor;
-import org.tensorflow.internal.c_api.TF_Tensor;
 import org.tensorflow.ndarray.DoubleNdArray;
 import org.tensorflow.ndarray.NdArray;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.ndarray.buffer.ByteDataBuffer;
 import org.tensorflow.ndarray.buffer.DataBuffer;
 import org.tensorflow.ndarray.buffer.DoubleDataBuffer;
-import org.tensorflow.ndarray.impl.dense.DoubleDenseNdArray;
 
+/**
+ * A {@link Tensor} of doubles.
+ */
 public interface DoubleTensor extends DoubleNdArray, Tensor<Double> {
 
   @Override
@@ -37,4 +35,3 @@ public interface DoubleTensor extends DoubleNdArray, Tensor<Double> {
   @Override
   DoubleTensor write(DoubleDataBuffer src);
 }
-
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/tensor/FloatTensor.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/tensor/FloatTensor.java
index c06e9c5f0d9..c4222d8c1d6 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/tensor/FloatTensor.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/tensor/FloatTensor.java
@@ -1,16 +1,14 @@
-package org.tensorflow.tensor;
+package org.tensorflow.types.tensor;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Tensor;
-import org.tensorflow.internal.c_api.TF_Tensor;
 import org.tensorflow.ndarray.FloatNdArray;
 import org.tensorflow.ndarray.NdArray;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.ndarray.buffer.ByteDataBuffer;
 import org.tensorflow.ndarray.buffer.DataBuffer;
 import org.tensorflow.ndarray.buffer.FloatDataBuffer;
-import org.tensorflow.ndarray.impl.dense.FloatDenseNdArray;
 
+/**
+ * A {@link Tensor} of floats.
+ */
 public interface FloatTensor extends FloatNdArray, Tensor<Float> {
 
   @Override
@@ -37,4 +35,3 @@ public interface FloatTensor extends FloatNdArray, Tensor<Float> {
   @Override
   FloatTensor write(FloatDataBuffer src);
 }
-
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/tensor/IntTensor.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/tensor/IntTensor.java
index 52d3930a844..67aab65f309 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/tensor/IntTensor.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/tensor/IntTensor.java
@@ -1,16 +1,14 @@
-package org.tensorflow.tensor;
+package org.tensorflow.types.tensor;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Tensor;
-import org.tensorflow.internal.c_api.TF_Tensor;
 import org.tensorflow.ndarray.IntNdArray;
 import org.tensorflow.ndarray.NdArray;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.ndarray.buffer.ByteDataBuffer;
 import org.tensorflow.ndarray.buffer.DataBuffer;
 import org.tensorflow.ndarray.buffer.IntDataBuffer;
-import org.tensorflow.ndarray.impl.dense.IntDenseNdArray;
 
+/**
+ * A {@link Tensor} of integers.
+ */
 public interface IntTensor extends IntNdArray, Tensor<Integer> {
 
   @Override
@@ -37,4 +35,3 @@ public interface IntTensor extends IntNdArray, Tensor<Integer> {
   @Override
   IntTensor write(IntDataBuffer src);
 }
-
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/tensor/LongTensor.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/tensor/LongTensor.java
index 1a29bb278d6..3d708bf869e 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/tensor/LongTensor.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/tensor/LongTensor.java
@@ -1,16 +1,14 @@
-package org.tensorflow.tensor;
+package org.tensorflow.types.tensor;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Tensor;
-import org.tensorflow.internal.c_api.TF_Tensor;
 import org.tensorflow.ndarray.LongNdArray;
 import org.tensorflow.ndarray.NdArray;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.ndarray.buffer.ByteDataBuffer;
 import org.tensorflow.ndarray.buffer.DataBuffer;
 import org.tensorflow.ndarray.buffer.LongDataBuffer;
-import org.tensorflow.ndarray.impl.dense.LongDenseNdArray;
 
+/**
+ * A {@link Tensor} of longs.
+ */
 public interface LongTensor extends LongNdArray, Tensor<Long> {
 
   @Override
@@ -37,4 +35,3 @@ public interface LongTensor extends LongNdArray, Tensor<Long> {
   @Override
   LongTensor write(LongDataBuffer src);
 }
-
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/tensor/StringTensor.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/tensor/StringTensor.java
index 78a9c2d43a0..7a74276e8df 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/tensor/StringTensor.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/tensor/StringTensor.java
@@ -1,9 +1,12 @@
-package org.tensorflow.tensor;
+package org.tensorflow.types.tensor;
 
 import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.NdArray;
 import org.tensorflow.ndarray.buffer.DataBuffer;
 
+/**
+ * A {@link Tensor} of strings.
+ */
 public interface StringTensor extends Tensor<String> {
 
   /**
@@ -12,18 +15,17 @@ public interface StringTensor extends Tensor<String> {
   NdArray<byte[]> asBytes();
 
   @Override
-  Tensor<String> set(NdArray<String> src, long... coordinates);
+  StringTensor set(NdArray<String> src, long... coordinates);
 
   @Override
-  Tensor<String> setObject(String value, long... coordinates);
+  StringTensor setObject(String value, long... coordinates);
 
   @Override
-  Tensor<String> copyTo(NdArray<String> dst);
+  StringTensor copyTo(NdArray<String> dst);
 
   @Override
-  Tensor<String> read(DataBuffer<String> dst);
+  StringTensor read(DataBuffer<String> dst);
 
   @Override
-  Tensor<String> write(DataBuffer<String> src);
+  StringTensor write(DataBuffer<String> src);
 }
-
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TBfloat16.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TBfloat16.java
index efd69dc1db1..402e7d61747 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TBfloat16.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TBfloat16.java
@@ -19,11 +19,11 @@
 
 import java.util.function.Consumer;
 import org.tensorflow.DataType;
-import org.tensorflow.Tensor;
+import org.tensorflow.Tensors;
 import org.tensorflow.exceptions.TensorFlowException;
-import org.tensorflow.internal.tensor.buffer.TensorBuffers;
 import org.tensorflow.internal.c_api.TF_Tensor;
 import org.tensorflow.internal.tensor.FloatTensorImpl;
+import org.tensorflow.internal.tensor.buffer.TensorBuffers;
 import org.tensorflow.ndarray.NdArray;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.ndarray.StdArrays;
@@ -48,7 +48,7 @@
  * <p>Note that some CPUs support the bfloat16 format natively, which can result in faster
  * computation compared to {@link TFloat16} when GPUs are not used.
  */
-public interface TBfloat16 extends FloatTensor, TFloating {
+public interface TBfloat16 extends FloatTensor, TFloating<TBfloat16, Float> {
 
   /** readable-name for the data type */
   static final String NAME = "BFLOAT16";
@@ -63,7 +63,7 @@ public interface TBfloat16 extends FloatTensor, TFloating {
    * @return the new tensor
    */
   static TBfloat16 scalarOf(float value) {
-    return Tensor.of(DTYPE, Shape.scalar(), t -> t.setFloat(value));
+    return Tensors.of(DTYPE, Shape.scalar(), t -> t.setFloat(value));
   }
 
   /**
@@ -76,7 +76,7 @@ static TBfloat16 vectorOf(float... values) {
     if (values == null) {
       throw new IllegalArgumentException();
     }
-    return Tensor.of(DTYPE, Shape.of(values.length), t -> StdArrays.copyTo(values, t));
+    return Tensors.of(DTYPE, Shape.of(values.length), t -> StdArrays.copyTo(values, t));
   }
 
   /**
@@ -88,7 +88,7 @@ static TBfloat16 vectorOf(float... values) {
    * @return the new tensor
    */
   static TBfloat16 tensorOf(NdArray<Float> src) {
-    return Tensor.of(DTYPE, src.shape(), src::copyTo);
+    return Tensors.of(DTYPE, src.shape(), src::copyTo);
   }
 
   /**
@@ -98,7 +98,7 @@ static TBfloat16 tensorOf(NdArray<Float> src) {
    * @return the new tensor
    */
   static TBfloat16 tensorOf(Shape shape) {
-    return Tensor.of(DTYPE, shape);
+    return Tensors.of(DTYPE, shape);
   }
 
   /**
@@ -109,7 +109,7 @@ static TBfloat16 tensorOf(Shape shape) {
    * @return the new tensor
    */
   static TBfloat16 tensorOf(Shape shape, FloatDataBuffer data) {
-    return Tensor.of(DTYPE, shape, t -> t.write(data));
+    return Tensors.of(DTYPE, shape, t -> t.write(data));
   }
 
   /**
@@ -121,7 +121,7 @@ static TBfloat16 tensorOf(Shape shape, FloatDataBuffer data) {
    * @throws TensorFlowException if the tensor cannot be allocated or initialized
    */
   static TBfloat16 tensorOf(Shape shape, Consumer<TBfloat16> tensorInit) {
-    return Tensor.of(DTYPE, shape, tensorInit);
+    return Tensors.of(DTYPE, shape, tensorInit);
   }
 }
 
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TBool.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TBool.java
index ad016a9db18..3d95b593d3b 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TBool.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TBool.java
@@ -19,11 +19,11 @@
 
 import java.util.function.Consumer;
 import org.tensorflow.DataType;
-import org.tensorflow.Tensor;
+import org.tensorflow.Tensors;
 import org.tensorflow.exceptions.TensorFlowException;
-import org.tensorflow.internal.tensor.buffer.TensorBuffers;
 import org.tensorflow.internal.c_api.TF_Tensor;
 import org.tensorflow.internal.tensor.BooleanTensorImpl;
+import org.tensorflow.internal.tensor.buffer.TensorBuffers;
 import org.tensorflow.ndarray.NdArray;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.ndarray.StdArrays;
@@ -39,7 +39,7 @@
  * explicit mapping between Java boolean values and byte buffers using the {@link DataLayouts#BOOL
  * BOOL} layout, which may impact I/O performances.
  */
-public interface TBool extends BooleanTensor, TType {
+public interface TBool extends BooleanTensor, TType<TBool, Boolean> {
 
   /** readable-name for the data type */
   static final String NAME = "BOOL";
@@ -54,7 +54,7 @@ public interface TBool extends BooleanTensor, TType {
    * @return the new tensor
    */
   static TBool scalarOf(boolean value) {
-    return Tensor.of(DTYPE, Shape.scalar(), t -> t.setBoolean(value));
+    return Tensors.of(DTYPE, Shape.scalar(), t -> t.setBoolean(value));
   }
 
   /**
@@ -67,7 +67,7 @@ static TBool vectorOf(boolean... values) {
     if (values == null) {
       throw new IllegalArgumentException();
     }
-    return Tensor.of(DTYPE, Shape.of(values.length), t -> StdArrays.copyTo(values, t));
+    return Tensors.of(DTYPE, Shape.of(values.length), t -> StdArrays.copyTo(values, t));
   }
 
   /**
@@ -79,7 +79,7 @@ static TBool vectorOf(boolean... values) {
    * @return the new tensor
    */
   static TBool tensorOf(NdArray<Boolean> src) {
-    return Tensor.of(DTYPE, src.shape(), src::copyTo);
+    return Tensors.of(DTYPE, src.shape(), src::copyTo);
   }
 
   /**
@@ -89,7 +89,7 @@ static TBool tensorOf(NdArray<Boolean> src) {
    * @return the new tensor
    */
   static TBool tensorOf(Shape shape) {
-    return Tensor.of(DTYPE, shape);
+    return Tensors.of(DTYPE, shape);
   }
 
   /**
@@ -100,7 +100,7 @@ static TBool tensorOf(Shape shape) {
    * @return the new tensor
    */
   static TBool tensorOf(Shape shape, BooleanDataBuffer data) {
-    return Tensor.of(DTYPE, shape, d -> d.write(data));
+    return Tensors.of(DTYPE, shape, d -> d.write(data));
   }
 
   /**
@@ -112,7 +112,7 @@ static TBool tensorOf(Shape shape, BooleanDataBuffer data) {
    * @throws TensorFlowException if the tensor cannot be allocated or initialized
    */
   static TBool tensorOf(Shape shape, Consumer<TBool> tensorInit) {
-    return Tensor.of(DTYPE, shape, tensorInit);
+    return Tensors.of(DTYPE, shape, tensorInit);
   }
 }
 
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TFloat16.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TFloat16.java
index db86071514c..1e8a3d2a1e1 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TFloat16.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TFloat16.java
@@ -19,11 +19,11 @@
 
 import java.util.function.Consumer;
 import org.tensorflow.DataType;
-import org.tensorflow.Tensor;
+import org.tensorflow.Tensors;
 import org.tensorflow.exceptions.TensorFlowException;
-import org.tensorflow.internal.tensor.buffer.TensorBuffers;
 import org.tensorflow.internal.c_api.TF_Tensor;
 import org.tensorflow.internal.tensor.FloatTensorImpl;
+import org.tensorflow.internal.tensor.buffer.TensorBuffers;
 import org.tensorflow.ndarray.NdArray;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.ndarray.StdArrays;
@@ -45,7 +45,7 @@
  * most CPUs do not support this format natively. For CPU computation on 16-bit floats, the {@link
  * TBfloat16} tensor type might be a better option.
  */
-public interface TFloat16 extends FloatTensor, TFloating {
+public interface TFloat16 extends FloatTensor, TFloating<TFloat16, Float> {
 
   /** readable-name for the data type */
   static final String NAME = "FLOAT16";
@@ -60,7 +60,7 @@ public interface TFloat16 extends FloatTensor, TFloating {
    * @return the new tensor
    */
   static TFloat16 scalarOf(float value) {
-    return Tensor.of(DTYPE, Shape.scalar(), t -> t.setFloat(value));
+    return Tensors.of(DTYPE, Shape.scalar(), t -> t.setFloat(value));
   }
 
   /**
@@ -73,7 +73,7 @@ static TFloat16 vectorOf(float... values) {
     if (values == null) {
       throw new IllegalArgumentException();
     }
-    return Tensor.of(DTYPE, Shape.of(values.length), t -> StdArrays.copyTo(values, t));
+    return Tensors.of(DTYPE, Shape.of(values.length), t -> StdArrays.copyTo(values, t));
   }
 
   /**
@@ -85,7 +85,7 @@ static TFloat16 vectorOf(float... values) {
    * @return the new tensor
    */
   static TFloat16 tensorOf(NdArray<Float> src) {
-    return Tensor.of(DTYPE, src.shape(), src::copyTo);
+    return Tensors.of(DTYPE, src.shape(), src::copyTo);
   }
 
   /**
@@ -95,7 +95,7 @@ static TFloat16 tensorOf(NdArray<Float> src) {
    * @return the new tensor
    */
   static TFloat16 tensorOf(Shape shape) {
-    return Tensor.of(DTYPE, shape);
+    return Tensors.of(DTYPE, shape);
   }
 
   /**
@@ -106,7 +106,7 @@ static TFloat16 tensorOf(Shape shape) {
    * @return the new tensor
    */
   static TFloat16 tensorOf(Shape shape, FloatDataBuffer data) {
-    return Tensor.of(DTYPE, shape, t -> t.write(data));
+    return Tensors.of(DTYPE, shape, t -> t.write(data));
   }
 
   /**
@@ -118,7 +118,7 @@ static TFloat16 tensorOf(Shape shape, FloatDataBuffer data) {
    * @throws TensorFlowException if the tensor cannot be allocated or initialized
    */
   static TFloat16 tensorOf(Shape shape, Consumer<TFloat16> tensorInit) {
-    return Tensor.of(DTYPE, shape, tensorInit);
+    return Tensors.of(DTYPE, shape, tensorInit);
   }
 }
 
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TFloat32.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TFloat32.java
index 96149a32e0e..378ed62d2a4 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TFloat32.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TFloat32.java
@@ -19,11 +19,11 @@
 
 import java.util.function.Consumer;
 import org.tensorflow.DataType;
-import org.tensorflow.Tensor;
+import org.tensorflow.Tensors;
 import org.tensorflow.exceptions.TensorFlowException;
-import org.tensorflow.internal.tensor.buffer.TensorBuffers;
 import org.tensorflow.internal.c_api.TF_Tensor;
 import org.tensorflow.internal.tensor.FloatTensorImpl;
+import org.tensorflow.internal.tensor.buffer.TensorBuffers;
 import org.tensorflow.ndarray.NdArray;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.ndarray.StdArrays;
@@ -34,7 +34,7 @@
 /**
  * IEEE-754 single-precision 32-bit float tensor type.
  */
-public interface TFloat32 extends FloatTensor, TFloating {
+public interface TFloat32 extends FloatTensor, TFloating<TFloat32, Float> {
 
   /** readable-name for the data type */
   static final String NAME = "FLOAT";
@@ -49,7 +49,7 @@ public interface TFloat32 extends FloatTensor, TFloating {
    * @return the new tensor
    */
   static TFloat32 scalarOf(float value) {
-    return Tensor.of(DTYPE, Shape.scalar(), t -> t.setFloat(value));
+    return Tensors.of(DTYPE, Shape.scalar(), t -> t.setFloat(value));
   }
 
   /**
@@ -62,7 +62,7 @@ static TFloat32 vectorOf(float... values) {
     if (values == null) {
       throw new IllegalArgumentException();
     }
-    return Tensor.of(DTYPE, Shape.of(values.length), t -> StdArrays.copyTo(values, t));
+    return Tensors.of(DTYPE, Shape.of(values.length), t -> StdArrays.copyTo(values, t));
   }
 
   /**
@@ -74,7 +74,7 @@ static TFloat32 vectorOf(float... values) {
    * @return the new tensor
    */
   static TFloat32 tensorOf(NdArray<Float> src) {
-    return Tensor.of(DTYPE, src.shape(), src::copyTo);
+    return Tensors.of(DTYPE, src.shape(), src::copyTo);
   }
 
   /**
@@ -84,7 +84,7 @@ static TFloat32 tensorOf(NdArray<Float> src) {
    * @return the new tensor
    */
   static TFloat32 tensorOf(Shape shape) {
-    return Tensor.of(DTYPE, shape);
+    return Tensors.of(DTYPE, shape);
   }
 
   /**
@@ -95,7 +95,7 @@ static TFloat32 tensorOf(Shape shape) {
    * @return the new tensor
    */
   static TFloat32 tensorOf(Shape shape, FloatDataBuffer data) {
-    return Tensor.of(DTYPE, shape, t -> t.write(data));
+    return Tensors.of(DTYPE, shape, t -> t.write(data));
   }
 
   /**
@@ -107,7 +107,7 @@ static TFloat32 tensorOf(Shape shape, FloatDataBuffer data) {
    * @throws TensorFlowException if the tensor cannot be allocated or initialized
    */
   static TFloat32 tensorOf(Shape shape, Consumer<TFloat32> tensorInit) {
-    return Tensor.of(DTYPE, shape, tensorInit);
+    return Tensors.of(DTYPE, shape, tensorInit);
   }
 }
 
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TFloat64.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TFloat64.java
index db455552950..3c55d64fbaf 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TFloat64.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TFloat64.java
@@ -19,11 +19,11 @@
 
 import java.util.function.Consumer;
 import org.tensorflow.DataType;
-import org.tensorflow.Tensor;
+import org.tensorflow.Tensors;
 import org.tensorflow.exceptions.TensorFlowException;
-import org.tensorflow.internal.tensor.buffer.TensorBuffers;
 import org.tensorflow.internal.c_api.TF_Tensor;
 import org.tensorflow.internal.tensor.DoubleTensorImpl;
+import org.tensorflow.internal.tensor.buffer.TensorBuffers;
 import org.tensorflow.ndarray.NdArray;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.ndarray.StdArrays;
@@ -34,7 +34,7 @@
 /**
  * IEEE-754 double-precision 64-bit float tensor type.
  */
-public interface TFloat64 extends DoubleTensor, TFloating {
+public interface TFloat64 extends DoubleTensor, TFloating<TFloat64, Double> {
 
   /** readable-name for the data type */
   static final String NAME = "DOUBLE";
@@ -49,7 +49,7 @@ public interface TFloat64 extends DoubleTensor, TFloating {
    * @return the new tensor
    */
   static TFloat64 scalarOf(double value) {
-    return Tensor.of(DTYPE, Shape.scalar(), t -> t.setDouble(value));
+    return Tensors.of(DTYPE, Shape.scalar(), t -> t.setDouble(value));
   }
 
   /**
@@ -62,7 +62,7 @@ static TFloat64 vectorOf(double... values) {
     if (values == null) {
       throw new IllegalArgumentException();
     }
-    return Tensor.of(DTYPE, Shape.of(values.length), t -> StdArrays.copyTo(values, t));
+    return Tensors.of(DTYPE, Shape.of(values.length), t -> StdArrays.copyTo(values, t));
   }
 
   /**
@@ -74,7 +74,7 @@ static TFloat64 vectorOf(double... values) {
    * @return the new tensor
    */
   static TFloat64 tensorOf(NdArray<Double> src) {
-    return Tensor.of(DTYPE, src.shape(), src::copyTo);
+    return Tensors.of(DTYPE, src.shape(), src::copyTo);
   }
 
   /**
@@ -84,7 +84,7 @@ static TFloat64 tensorOf(NdArray<Double> src) {
    * @return the new tensor
    */
   static TFloat64 tensorOf(Shape shape) {
-    return Tensor.of(DTYPE, shape);
+    return Tensors.of(DTYPE, shape);
   }
 
   /**
@@ -95,7 +95,7 @@ static TFloat64 tensorOf(Shape shape) {
    * @return the new tensor
    */
   static TFloat64 tensorOf(Shape shape, DoubleDataBuffer data) {
-    return Tensor.of(DTYPE, shape, t -> t.write(data));
+    return Tensors.of(DTYPE, shape, t -> t.write(data));
   }
 
   /**
@@ -107,7 +107,7 @@ static TFloat64 tensorOf(Shape shape, DoubleDataBuffer data) {
    * @throws TensorFlowException if the tensor cannot be allocated or initialized
    */
   static TFloat64 tensorOf(Shape shape, Consumer<TFloat64> tensorInit) {
-    return Tensor.of(DTYPE, shape, tensorInit);
+    return Tensors.of(DTYPE, shape, tensorInit);
   }
 }
 
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TInt32.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TInt32.java
index 6218ac06398..5ae4b9a388e 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TInt32.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TInt32.java
@@ -19,10 +19,10 @@
 
 import java.util.function.Consumer;
 import org.tensorflow.DataType;
-import org.tensorflow.Tensor;
-import org.tensorflow.internal.tensor.buffer.TensorBuffers;
+import org.tensorflow.Tensors;
 import org.tensorflow.internal.c_api.TF_Tensor;
 import org.tensorflow.internal.tensor.IntTensorImpl;
+import org.tensorflow.internal.tensor.buffer.TensorBuffers;
 import org.tensorflow.ndarray.NdArray;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.ndarray.StdArrays;
@@ -33,7 +33,7 @@
 /**
  * 32-bit signed integer tensor type.
  */
-public interface TInt32 extends IntTensor, TNumber {
+public interface TInt32 extends IntTensor, TNumber<TInt32, Integer> {
 
   /** readable-name for the data type */
   static final String NAME = "INT32";
@@ -48,7 +48,7 @@ public interface TInt32 extends IntTensor, TNumber {
    * @return the new tensor
    */
   static TInt32 scalarOf(int value) {
-    return Tensor.of(DTYPE, Shape.scalar(), t -> t.setInt(value));
+    return Tensors.of(DTYPE, Shape.scalar(), t -> t.setInt(value));
   }
 
   /**
@@ -62,7 +62,7 @@ static TInt32 vectorOf(int... values) {
     if (values == null) {
       throw new IllegalArgumentException();
     }
-    return Tensor.of(DTYPE, Shape.of(values.length), t -> StdArrays.copyTo(values, t));
+    return Tensors.of(DTYPE, Shape.of(values.length), t -> StdArrays.copyTo(values, t));
   }
 
   /**
@@ -74,7 +74,7 @@ static TInt32 vectorOf(int... values) {
    * @return the new tensor
    */
   static TInt32 tensorOf(NdArray<Integer> src) {
-    return Tensor.of(DTYPE, src.shape(), src::copyTo);
+    return Tensors.of(DTYPE, src.shape(), src::copyTo);
   }
 
   /**
@@ -84,7 +84,7 @@ static TInt32 tensorOf(NdArray<Integer> src) {
    * @return the new tensor
    */
   static TInt32 tensorOf(Shape shape) {
-    return Tensor.of(DTYPE, shape);
+    return Tensors.of(DTYPE, shape);
   }
 
   /**
@@ -95,7 +95,7 @@ static TInt32 tensorOf(Shape shape) {
    * @return the new tensor
    */
   static TInt32 tensorOf(Shape shape, IntDataBuffer data) {
-    return Tensor.of(DTYPE, shape, t -> t.write(data));
+    return Tensors.of(DTYPE, shape, t -> t.write(data));
   }
 
   /**
@@ -106,7 +106,7 @@ static TInt32 tensorOf(Shape shape, IntDataBuffer data) {
    * @return the new tensor
    */
   static TInt32 tensorOf(Shape shape, Consumer<TInt32> tensorInit) {
-    return Tensor.of(DTYPE, shape, tensorInit);
+    return Tensors.of(DTYPE, shape, tensorInit);
   }
 }
 
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TInt64.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TInt64.java
index 1a55a19458f..e91c8a7703e 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TInt64.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TInt64.java
@@ -19,11 +19,11 @@
 
 import java.util.function.Consumer;
 import org.tensorflow.DataType;
-import org.tensorflow.Tensor;
+import org.tensorflow.Tensors;
 import org.tensorflow.exceptions.TensorFlowException;
-import org.tensorflow.internal.tensor.buffer.TensorBuffers;
 import org.tensorflow.internal.c_api.TF_Tensor;
 import org.tensorflow.internal.tensor.LongTensorImpl;
+import org.tensorflow.internal.tensor.buffer.TensorBuffers;
 import org.tensorflow.ndarray.NdArray;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.ndarray.StdArrays;
@@ -34,7 +34,7 @@
 /**
  * 64-bit signed integer tensor type.
  */
-public interface TInt64 extends LongTensor, TNumber {
+public interface TInt64 extends LongTensor, TNumber<TInt64, Long> {
 
   /** readable-name for the data type */
   static final String NAME = "INT64";
@@ -49,7 +49,7 @@ public interface TInt64 extends LongTensor, TNumber {
    * @return the new tensor
    */
   static TInt64 scalarOf(long value) {
-    return Tensor.of(DTYPE, Shape.scalar(), t -> t.setLong(value));
+    return Tensors.of(DTYPE, Shape.scalar(), t -> t.setLong(value));
   }
 
   /**
@@ -62,7 +62,7 @@ static TInt64 vectorOf(long... values) {
     if (values == null) {
       throw new IllegalArgumentException();
     }
-    return Tensor.of(DTYPE, Shape.of(values.length), t -> StdArrays.copyTo(values, t));
+    return Tensors.of(DTYPE, Shape.of(values.length), t -> StdArrays.copyTo(values, t));
   }
 
   /**
@@ -74,7 +74,7 @@ static TInt64 vectorOf(long... values) {
    * @return the new tensor
    */
   static TInt64 tensorOf(NdArray<Long> src) {
-    return Tensor.of(DTYPE, src.shape(), src::copyTo);
+    return Tensors.of(DTYPE, src.shape(), src::copyTo);
   }
 
   /**
@@ -84,7 +84,7 @@ static TInt64 tensorOf(NdArray<Long> src) {
    * @return the new tensor
    */
   static TInt64 tensorOf(Shape shape) {
-    return Tensor.of(DTYPE, shape);
+    return Tensors.of(DTYPE, shape);
   }
 
   /**
@@ -95,7 +95,7 @@ static TInt64 tensorOf(Shape shape) {
    * @return the new tensor
    */
   static TInt64 tensorOf(Shape shape, LongDataBuffer data) {
-    return Tensor.of(DTYPE, shape, t -> t.write(data));
+    return Tensors.of(DTYPE, shape, t -> t.write(data));
   }
 
   /**
@@ -107,7 +107,7 @@ static TInt64 tensorOf(Shape shape, LongDataBuffer data) {
    * @throws TensorFlowException if the tensor cannot be allocated or initialized
    */
   static TInt64 tensorOf(Shape shape, Consumer<TInt64> tensorInit) {
-    return Tensor.of(DTYPE, shape, tensorInit);
+    return Tensors.of(DTYPE, shape, tensorInit);
   }
 }
 
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TString.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TString.java
index 901857f24a8..69dc26806c5 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TString.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TString.java
@@ -21,11 +21,11 @@
 import java.nio.charset.StandardCharsets;
 import java.util.function.Function;
 import org.tensorflow.DataType;
-import org.tensorflow.Tensor;
-import org.tensorflow.internal.tensor.buffer.ByteSequenceTensorBuffer;
-import org.tensorflow.internal.tensor.buffer.TensorBuffers;
+import org.tensorflow.Tensors;
 import org.tensorflow.internal.c_api.TF_Tensor;
 import org.tensorflow.internal.tensor.StringTensorImpl;
+import org.tensorflow.internal.tensor.buffer.ByteSequenceTensorBuffer;
+import org.tensorflow.internal.tensor.buffer.TensorBuffers;
 import org.tensorflow.ndarray.NdArray;
 import org.tensorflow.ndarray.NdArrays;
 import org.tensorflow.ndarray.Shape;
@@ -44,7 +44,7 @@
  * its values initially, so TensorFlow can compute and allocate the right amount of memory. Then the
  * data in the tensor is initialized once and cannot be modified afterwards.
  */
-public interface TString extends StringTensor, TType {
+public interface TString extends StringTensor, TType<TString, String> {
 
   /** readable-name for the data type */
   static final String NAME = "STRING";
@@ -225,12 +225,12 @@ class TStringImpl extends StringTensorImpl implements TString {
 
   @Override
   public TString using(Charset charset) {
-    return new TStringImpl(nativeHandle(), shape(), DataLayouts.ofStrings(charset), rawBuffer());
+    return new TStringImpl(nativeHandle, shape(), DataLayouts.ofStrings(charset), rawBuffer());
   }
 
   static <T> TString createTensor(NdArray<T> src, Function<T, byte[]> getBytes) {
     long size = ByteSequenceTensorBuffer.computeSize(src, getBytes);
-    return Tensor.of(TString.DTYPE, src.shape(), size, t ->
+    return Tensors.of(TString.DTYPE, src.shape(), size, t ->
         ((TStringImpl)t).rawBuffer().init(src, getBytes)
     );
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TUint8.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TUint8.java
index 5907cfbfc12..e5780a18ab8 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TUint8.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TUint8.java
@@ -19,11 +19,11 @@
 
 import java.util.function.Consumer;
 import org.tensorflow.DataType;
-import org.tensorflow.Tensor;
+import org.tensorflow.Tensors;
 import org.tensorflow.exceptions.TensorFlowException;
-import org.tensorflow.internal.tensor.buffer.TensorBuffers;
 import org.tensorflow.internal.c_api.TF_Tensor;
 import org.tensorflow.internal.tensor.ByteTensorImpl;
+import org.tensorflow.internal.tensor.buffer.TensorBuffers;
 import org.tensorflow.ndarray.NdArray;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.ndarray.StdArrays;
@@ -34,7 +34,7 @@
 /**
  * 8-bit unsigned integer tensor type.
  */
-public interface TUint8 extends ByteTensor, TNumber {
+public interface TUint8 extends ByteTensor, TNumber<TUint8, Byte> {
 
   /** readable-name for the data type */
   static final String NAME = "UINT8";
@@ -49,7 +49,7 @@ public interface TUint8 extends ByteTensor, TNumber {
    * @return the new tensor
    */
   static TUint8 scalarOf(byte value) {
-    return Tensor.of(DTYPE, Shape.scalar(), t -> t.setByte(value));
+    return Tensors.of(DTYPE, Shape.scalar(), t -> t.setByte(value));
   }
 
   /**
@@ -62,7 +62,7 @@ static TUint8 vectorOf(byte... values) {
     if (values == null) {
       throw new IllegalArgumentException();
     }
-    return Tensor.of(DTYPE, Shape.of(values.length), t -> StdArrays.copyTo(values, t));
+    return Tensors.of(DTYPE, Shape.of(values.length), t -> StdArrays.copyTo(values, t));
   }
 
   /**
@@ -74,7 +74,7 @@ static TUint8 vectorOf(byte... values) {
    * @return the new tensor
    */
   static TUint8 tensorOf(NdArray<Byte> src) {
-    return Tensor.of(DTYPE, src.shape(), src::copyTo);
+    return Tensors.of(DTYPE, src.shape(), src::copyTo);
   }
 
   /**
@@ -84,7 +84,7 @@ static TUint8 tensorOf(NdArray<Byte> src) {
    * @return the new tensor
    */
   static TUint8 tensorOf(Shape shape) {
-    return Tensor.of(DTYPE, shape);
+    return Tensors.of(DTYPE, shape);
   }
 
   /**
@@ -95,7 +95,7 @@ static TUint8 tensorOf(Shape shape) {
    * @return the new tensor
    */
   static TUint8 tensorOf(Shape shape, ByteDataBuffer data) {
-    return Tensor.of(DTYPE, shape, d -> d.write(data));
+    return Tensors.of(DTYPE, shape, d -> d.write(data));
   }
 
   /**
@@ -107,7 +107,7 @@ static TUint8 tensorOf(Shape shape, ByteDataBuffer data) {
    * @throws TensorFlowException if the tensor cannot be allocated or initialized
    */
   static TUint8 tensorOf(Shape shape, Consumer<TUint8> tensorInit) {
-    return Tensor.of(DTYPE, shape, tensorInit);
+    return Tensors.of(DTYPE, shape, tensorInit);
   }
 }
 
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/family/TNumber.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/family/TNumber.java
index 97ee59af095..4fa2b563cae 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/family/TNumber.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/family/TNumber.java
@@ -32,4 +32,4 @@
  * tf.nn.softmax(tf.constant(tensor2));  // Compilation failure
  * }</pre>
  */
-public interface TNumber extends TType {}
+public interface TNumber<T extends TNumber, U> extends TType<T, U> {}
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/family/TType.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/family/TType.java
index 8f3451b9a68..d8a8063eeb9 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/family/TType.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/family/TType.java
@@ -17,6 +17,14 @@
 
 package org.tensorflow.types.family;
 
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.Output;
+import org.tensorflow.Tensor;
+import org.tensorflow.op.Ops;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.core.Constant;
+
 /**
  * Marker interface for all tensor types.
  *
@@ -25,13 +33,31 @@
  * between operands of a computation at compile-time. For example:
  *
  * <pre>{@code
- * Tensor<TFloat32> tensor1 = TFloat32.ofShape(2, 3, 2);
- * Tensor<TFloat32> tensor2 = TFloat32.ofShape(2, 3, 2);
- * Tensor<TInt32> tensor3 = TInt32.ofShape(2, 3, 2);
+ * TFloat32 tensor1 = TFloat32.ofShape(2, 3, 2);
+ * TFloat32 tensor2 = TFloat32.ofShape(2, 3, 2);
+ * TInt32 tensor3 = TInt32.ofShape(2, 3, 2);
  *
  * Ops tf = Ops.create();
- * tf.math.add(tf.constant(tensor1), tf.constant(tensor2));  // OK
- * tf.math.add(tf.constant(tensor1), tf.constant(tensor3));  // Compilation failure
+ * tf.math.add(tensor1, tensor2);  // OK
+ * tf.math.add(tensor1, tensor3);  // Compilation failure
  * }</pre>
  */
-public interface TType {}
+public interface TType<T extends TType, U> extends Tensor<U>, Operand<T> {
+
+  default Constant<T> toConstant(Scope scope) {
+    if (scope == null) {
+      throw new IllegalArgumentException("Scope is required");
+    }
+    return Constant.create(scope, (T)this);
+  }
+
+  @Override
+  default Output<T> asOutput(Scope scope) {
+    return toConstant(scope).asOutput();
+  }
+
+  @Override
+  default Operation op() {
+    throw new IllegalStateException("A tensor is not attached to a specific operation");
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/TensorTest.java b/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/TensorTest.java
index 162f1ac1c95..6aa33ef0a73 100644
--- a/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/TensorTest.java
+++ b/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/TensorTest.java
@@ -30,9 +30,6 @@
 import java.nio.IntBuffer;
 import java.nio.LongBuffer;
 import org.junit.jupiter.api.Test;
-import org.tensorflow.op.Ops;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.ndarray.buffer.DataBuffers;
 import org.tensorflow.ndarray.BooleanNdArray;
 import org.tensorflow.ndarray.DoubleNdArray;
 import org.tensorflow.ndarray.FloatNdArray;
@@ -40,7 +37,10 @@
 import org.tensorflow.ndarray.LongNdArray;
 import org.tensorflow.ndarray.NdArray;
 import org.tensorflow.ndarray.NdArrays;
+import org.tensorflow.ndarray.Shape;
 import org.tensorflow.ndarray.StdArrays;
+import org.tensorflow.ndarray.buffer.DataBuffers;
+import org.tensorflow.op.Ops;
 import org.tensorflow.types.TBool;
 import org.tensorflow.types.TFloat32;
 import org.tensorflow.types.TFloat64;
@@ -71,14 +71,14 @@ public void createWithRawData() {
 
     // validate creating a tensor using a raw data byte buffers
     {
-      try (TBool t = Tensor.of(TBool.DTYPE, bools_shape, DataBuffers.of(bools_))) {
+      try (TBool t = Tensors.of(TBool.DTYPE, bools_shape, DataBuffers.of(bools_))) {
         boolean[] actual = new boolean[bools_.length];
         t.read(DataBuffers.of(actual));
         assertArrayEquals(bools, actual);
       }
 
       // note: the buffer is expected to contain raw TF_STRING (as per C API)
-      try (TString t = Tensor.of(TString.DTYPE, strings_shape, DataBuffers.of(strings_))) {
+      try (TString t = Tensors.of(TString.DTYPE, strings_shape, DataBuffers.of(strings_))) {
         assertEquals(strings, t.getObject());
       }
     }
@@ -95,7 +95,7 @@ public void createWithRawData() {
     }
 
     // validate shape checking
-    try (TBool t = Tensor.of(TBool.DTYPE, Shape.of(bools_.length * 2), DataBuffers.of(bools_))) {
+    try (TBool t = Tensors.of(TBool.DTYPE, Shape.of(bools_.length * 2), DataBuffers.of(bools_))) {
       fail("should have failed on incompatible buffer");
     } catch (IllegalArgumentException e) {
       // expected
@@ -445,43 +445,31 @@ public void tensorWithZeroDimension() {
 
   @Test
   public void allocateTensorWithSize() {
-    try (TInt32 t = Tensor.of(TInt32.DTYPE, Shape.of(2, 2, 2), 8 * TInt32.DTYPE.byteSize())) {
+    try (TInt32 t = Tensors.of(TInt32.DTYPE, Shape.of(2, 2, 2), 8 * TInt32.DTYPE.byteSize())) {
       // ok
     }
-    try (TInt32 t = Tensor.of(TInt32.DTYPE, Shape.of(2, 2, 2), 9 * TInt32.DTYPE.byteSize())) {
+    try (TInt32 t = Tensors.of(TInt32.DTYPE, Shape.of(2, 2, 2), 9 * TInt32.DTYPE.byteSize())) {
       // ok (size requested is larger that minimum space required)
     }
     try {
-      Tensor.of(TInt32.DTYPE, Shape.of(2, 2, 2), 8 * TInt32.DTYPE.byteSize() - 1);
+      Tensors.of(TInt32.DTYPE, Shape.of(2, 2, 2), 8 * TInt32.DTYPE.byteSize() - 1);
       fail();
     } catch (IllegalArgumentException e) {
       // as expected
     }
   }
 
-  @Test
-  public void useAfterClose() {
-    int n = 4;
-    Tensor<?> t = TInt32.scalarOf(n);
-    t.close();
-    try {
-      t.data();
-    } catch (IllegalStateException e) {
-      // The expected exception.
-    }
-  }
-
   @Test
   public void eagerTensorIsReleasedAfterSessionIsClosed() {
     TInt32 sum;
     try (EagerSession session = EagerSession.create()) {
       Ops tf = Ops.create(session);
       sum = tf.math.add(tf.constant(10), tf.constant(20)).asTensor();
-      sum.nativeHandle(); // does not throw
+      ((AbstractTensor)sum).nativeHandle(); // does not throw
       assertEquals(30, sum.getInt());
     }
     try {
-      sum.nativeHandle();
+      ((AbstractTensor)sum).nativeHandle();
       fail("Tensor native handle should have been closed by ending eager session");
     } catch (IllegalStateException e) {
       // as expected
@@ -504,7 +492,7 @@ public void fromHandle() {
     // close() on both Tensors.
     final FloatNdArray matrix = StdArrays.ndCopyOf(new float[][]{{1, 2, 3}, {4, 5, 6}});
     try (TFloat32 src = TFloat32.tensorOf(matrix)) {
-      TFloat32 cpy = Tensors.fromHandle(src.nativeHandle()).expect(TFloat32.DTYPE);
+      TFloat32 cpy = Tensors.fromHandle(((AbstractTensor)src).nativeHandle()).expect(TFloat32.DTYPE);
       assertEquals(src.dataType(), cpy.dataType());
       assertEquals(src.shape().numDimensions(), cpy.shape().numDimensions());
       assertEquals(src.shape(), cpy.shape());
diff --git a/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/op/OperandsTest.java b/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/op/OperandsTest.java
index bc3e418e6f4..f2c27c4cdf0 100644
--- a/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/op/OperandsTest.java
+++ b/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/op/OperandsTest.java
@@ -31,7 +31,7 @@ public void createOutputArrayFromOperandList() {
     try (Graph g = new Graph()) {
       Ops tf = Ops.create(g);
       Split<?> split = tf.split(tf.constant(0), tf.array(0, 1, 2), 3L);
-      Output<?>[] array = Operands.asOutputs(split.output());
+      Output<?>[] array = Operands.asOutputs(tf.scope(), split.output());
       assertEquals(split.output().size(), array.length);
       assertSame(array[0], split.output().get(0));
       assertSame(array[1], split.output().get(1));
diff --git a/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/op/ScopeTest.java b/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/op/ScopeTest.java
index 288c7d7b18b..af3523c9e28 100644
--- a/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/op/ScopeTest.java
+++ b/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/op/ScopeTest.java
@@ -25,6 +25,7 @@
 import org.tensorflow.Session;
 import org.tensorflow.Tensor;
 import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TType;
 
 /** Unit tests for {@link org.tensorflow.op.Scope}. */
 public class ScopeTest {
@@ -177,7 +178,7 @@ public void composite() {
   }
 
   // "handwritten" sample operator classes
-  private static final class Const<T extends Tensor> {
+  private static final class Const<T extends TType> {
     private final Output<T> output;
 
     static Const<TInt32> create(Scope s, int v) {
@@ -188,7 +189,7 @@ static Const<TInt32> create(Scope s, int[] v) {
       return create(s, TInt32.vectorOf(v));
     }
 
-    static <T extends Tensor> Const<T> create(Scope s, T value) {
+    static <T extends TType> Const<T> create(Scope s, T value) {
       return new Const<>(
           s.env()
               .opBuilder("Const", s.makeOpName("Const"))
@@ -207,10 +208,10 @@ Output<T> output() {
     }
   }
 
-  private static final class Mean<T extends Tensor> {
+  private static final class Mean<T extends TType> {
     private final Output<T> output;
 
-    static <T extends Tensor> Mean<T> create(Scope s, Output<T> input, Output<T> reductionIndices) {
+    static <T extends TType> Mean<T> create(Scope s, Output<T> input, Output<T> reductionIndices) {
       return new Mean<>(
           s.env()
               .opBuilder("Mean", s.makeOpName("Mean"))
@@ -229,10 +230,10 @@ Output<T> output() {
     }
   }
 
-  private static final class SquaredDifference<T extends Tensor> {
+  private static final class SquaredDifference<T extends TType> {
     private final Output<T> output;
 
-    static <T extends Tensor> SquaredDifference<T> create(Scope s, Output<T> x, Output<T> y) {
+    static <T extends TType> SquaredDifference<T> create(Scope s, Output<T> x, Output<T> y) {
       return new SquaredDifference<>(
           s.env()
               .opBuilder("SquaredDifference", s.makeOpName("SquaredDifference"))
@@ -251,7 +252,7 @@ Output<T> output() {
     }
   }
 
-  private static final class Variance<T extends Tensor> {
+  private static final class Variance<T extends TType> {
     private final Output<T> output;
 
     static Variance<TInt32> create(Scope base, Output<TInt32> x) {
diff --git a/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/op/core/ConstantTest.java b/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/op/core/ConstantTest.java
index 266a62bd1ed..cd9f4c94a24 100644
--- a/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/op/core/ConstantTest.java
+++ b/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/op/core/ConstantTest.java
@@ -18,11 +18,6 @@
 import static org.junit.jupiter.api.Assertions.assertEquals;
 
 import java.io.IOException;
-import java.nio.ByteBuffer;
-import java.nio.DoubleBuffer;
-import java.nio.FloatBuffer;
-import java.nio.IntBuffer;
-import java.nio.LongBuffer;
 import org.junit.jupiter.api.Test;
 import org.tensorflow.AutoCloseableList;
 import org.tensorflow.Graph;
@@ -64,8 +59,8 @@ public void createInts() {
       Constant<TInt32> op2 = Constant.tensorOf(scope, array);
       try (AutoCloseableList<Tensor<?>> t =
           new AutoCloseableList<>(sess.runner().fetch(op1).fetch(op2).run())) {
-        assertEquals(array, t.get(0).expect(TInt32.DTYPE).data());
-        assertEquals(array, t.get(1).expect(TInt32.DTYPE).data());
+        assertEquals(array, t.get(0).expect(TInt32.DTYPE));
+        assertEquals(array, t.get(1).expect(TInt32.DTYPE));
       }
     }
   }
@@ -83,8 +78,8 @@ public void createFloats() {
       Constant<TFloat32> op2 = Constant.tensorOf(scope, array);
       try (AutoCloseableList<Tensor<?>> t =
           new AutoCloseableList<>(sess.runner().fetch(op1).fetch(op2).run())) {
-        assertEquals(array, t.get(0).expect(TFloat32.DTYPE).data());
-        assertEquals(array, t.get(1).expect(TFloat32.DTYPE).data());
+        assertEquals(array, t.get(0).expect(TFloat32.DTYPE));
+        assertEquals(array, t.get(1).expect(TFloat32.DTYPE));
       }
     }
   }
@@ -102,8 +97,8 @@ public void createDoubles() {
       Constant<TFloat64> op2 = Constant.tensorOf(scope, array);
       try (AutoCloseableList<Tensor<?>> t =
           new AutoCloseableList<>(sess.runner().fetch(op1).fetch(op2).run())) {
-        assertEquals(array, t.get(0).expect(TFloat64.DTYPE).data());
-        assertEquals(array, t.get(1).expect(TFloat64.DTYPE).data());
+        assertEquals(array, t.get(0).expect(TFloat64.DTYPE));
+        assertEquals(array, t.get(1).expect(TFloat64.DTYPE));
       }
     }
   }
@@ -121,8 +116,8 @@ public void createLongs() {
       Constant<TInt64> op2 = Constant.tensorOf(scope, array);
       try (AutoCloseableList<Tensor<?>> t =
           new AutoCloseableList<>(sess.runner().fetch(op1).fetch(op2).run())) {
-        assertEquals(array, t.get(0).expect(TInt64.DTYPE).data());
-        assertEquals(array, t.get(1).expect(TInt64.DTYPE).data());
+        assertEquals(array, t.get(0).expect(TInt64.DTYPE));
+        assertEquals(array, t.get(1).expect(TInt64.DTYPE));
       }
     }
   }
@@ -140,8 +135,8 @@ public void createStrings() throws IOException {
       Constant<TString> op2 = Constant.tensorOf(scope, array);
       try (AutoCloseableList<Tensor<?>> t =
           new AutoCloseableList<>(sess.runner().fetch(op1).fetch(op2).run())) {
-        assertEquals(array, t.get(0).expect(TString.DTYPE).data());
-        assertEquals(array, t.get(1).expect(TString.DTYPE).data());
+        assertEquals(array, t.get(0).expect(TString.DTYPE));
+        assertEquals(array, t.get(1).expect(TString.DTYPE));
       }
     }
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/types/NumericTypesTestBase.java b/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/types/NumericTypesTestBase.java
index de4dc09035f..3c7dbea1450 100644
--- a/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/types/NumericTypesTestBase.java
+++ b/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/types/NumericTypesTestBase.java
@@ -20,19 +20,22 @@
 import static org.junit.jupiter.api.Assertions.assertEquals;
 
 import org.junit.jupiter.api.Test;
+import org.tensorflow.DataType;
 import org.tensorflow.EagerSession;
-import org.tensorflow.Tensor;
-import org.tensorflow.op.Ops;
-import org.tensorflow.op.core.Constant;
-import org.tensorflow.op.math.Sub;
-import org.tensorflow.ndarray.Shape;
 import org.tensorflow.ndarray.IntNdArray;
+import org.tensorflow.ndarray.NdArray;
 import org.tensorflow.ndarray.NdArrays;
+import org.tensorflow.ndarray.Shape;
 import org.tensorflow.ndarray.index.Indices;
+import org.tensorflow.op.Ops;
+import org.tensorflow.op.core.Constant;
+import org.tensorflow.op.math.Add;
+import org.tensorflow.op.math.Pow;
+import org.tensorflow.op.math.Sub;
 import org.tensorflow.tensor.IntTensor;
-import org.tensorflow.types.family.TType;
+import org.tensorflow.types.family.TNumber;
 
-abstract class NumericTypesTestBase<T extends Tensor<U> & TType, U> {
+abstract class NumericTypesTestBase<T extends TNumber, U> {
 
   @Test
   public void initializeTensorsWithZeros() {
@@ -46,27 +49,31 @@ public void initializeTensorsWithZeros() {
       Ops tf = Ops.create(session);
 
       // Initialize tensor memory with zeros and take a snapshot
-      tensor.scalars().forEach(scalar -> scalar.setObject(valueOf(0)));
-      Constant<T> x = tf.constant(tensor);
+      tensor.scalars().forEach(scalar -> ((NdArray<U>)scalar).setObject(valueOf(0)));
+      Constant<T> x = tf.capture(tensor);
 
       // Initialize the same tensor memory with ones and take a snapshot
-      tensor.scalars().forEach(scalar -> scalar.setObject(valueOf(1)));
-      Constant<T> y = tf.constant(tensor);
+      tensor.scalars().forEach(scalar -> ((NdArray<U>)scalar).setObject(valueOf(1)));
+      Constant<T> y = tf.capture(tensor);
 
       // Subtract y from x and validate the result
       Sub<T> sub = tf.math.sub(x, y);
       sub.asTensor().scalars().forEach(scalar ->
-          assertEquals(valueOf(-1), scalar.getObject())
+          assertEquals(valueOf(-1), ((NdArray<U>)scalar).getObject())
       );
     }
   }
 
   @Test
-  public void genericTest() {
-    IntNdArray heapData = NdArrays.vectorOf(0, 1, 2, 3);
+  public void setAndCompute() {
+    NdArray<U> heapData = allocateNdArray(Shape.of(4))
+        .setObject(valueOf(0), 0)
+        .setObject(valueOf(1), 1)
+        .setObject(valueOf(2), 2)
+        .setObject(valueOf(3), 3);
 
     // Creates a 2x2 matrix
-    try (IntTensor tensor = TInt32.tensorOf(Shape.of(2, 2))) {
+    try (T tensor = allocateTensor(Shape.of(2, 2))) {
 
       // Copy first 2 values of the vector to the first row of the matrix
       tensor.set(heapData.slice(Indices.range(0, 2)), 0);
@@ -74,35 +81,36 @@ public void genericTest() {
       // Copy values at an odd position in the vector as the second row of the matrix
       tensor.set(heapData.slice(Indices.odd()), 1);
 
-      assertEquals(0, tensor.getInt(0, 0));
-      assertEquals(1, tensor.getInt(0, 1));
-      assertEquals(1, tensor.getInt(1, 0));
-      assertEquals(3, tensor.getInt(1, 1));
+      assertEquals(valueOf(0), tensor.getObject(0, 0));
+      assertEquals(valueOf(1), tensor.getObject(0, 1));
+      assertEquals(valueOf(1), tensor.getObject(1, 0));
+      assertEquals(valueOf(3), tensor.getObject(1, 1));
 
       // Read rows of the tensor in reverse order
-      IntNdArray reversedTensorData = tensor.slice(Indices.all(), Indices.flip());
+      NdArray<U> flippedData = tensor.slice(Indices.flip(), Indices.flip());
 
-      assertEquals(1, reversedTensorData.getInt(0, 0));
-      assertEquals(0, reversedTensorData.getInt(0, 1));
-      assertEquals(3, reversedTensorData.getInt(1, 0));
-      assertEquals(1, reversedTensorData.getInt(1, 1));
+      assertEquals(valueOf(3), flippedData.getObject(0, 0));
+      assertEquals(valueOf(1), flippedData.getObject(0, 1));
+      assertEquals(valueOf(1), flippedData.getObject(1, 0));
+      assertEquals(valueOf(0), flippedData.getObject(1, 1));
 
       try (EagerSession session = EagerSession.create()) {
         Ops tf = Ops.create(session);
 
-        // Compute the power of the tensor by itself
-        Constant<TInt32> x = tf.constant(tensor);
-        IntNdArray result = tf.math.pow(x, x).data();
+        Add<T> add = tf.math.add(tensor, tensor);
+        T result = add.asTensor();
 
-        // Validate result by computing the same operation in Java
-        tensor.scalars().forEachIndexed((coords, s) ->
-          assertEquals(Math.pow(s.getInt(), s.getInt()), result.getInt(coords), 1e-7f)
-        );
+        assertEquals(valueOf(0), result.getObject(0, 0));
+        assertEquals(valueOf(2), result.getObject(0, 1));
+        assertEquals(valueOf(2), result.getObject(1, 0));
+        assertEquals(valueOf(6), result.getObject(1, 1));
       }
     }
   }
 
   abstract T allocateTensor(Shape shape);
 
+  abstract NdArray<U> allocateNdArray(Shape shape);
+
   abstract U valueOf(Integer value);
 }
diff --git a/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/types/TBfloat16Test.java b/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/types/TBfloat16Test.java
index 1ef6a24fc8b..70051fd1fc7 100644
--- a/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/types/TBfloat16Test.java
+++ b/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/types/TBfloat16Test.java
@@ -18,6 +18,8 @@
 package org.tensorflow.types;
 
 import org.tensorflow.Tensor;
+import org.tensorflow.ndarray.NdArray;
+import org.tensorflow.ndarray.NdArrays;
 import org.tensorflow.ndarray.Shape;
 
 public class TBfloat16Test extends NumericTypesTestBase<TBfloat16, Float> {
@@ -27,6 +29,11 @@ TBfloat16 allocateTensor(Shape shape) {
     return TBfloat16.tensorOf(shape);
   }
 
+  @Override
+  NdArray<Float> allocateNdArray(Shape shape) {
+    return NdArrays.ofFloats(shape);
+  }
+
   @Override
   Float valueOf(Integer value) {
     return value.floatValue();
diff --git a/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/types/TFloat16Test.java b/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/types/TFloat16Test.java
index b0b2fa908f0..7fd65ba9654 100644
--- a/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/types/TFloat16Test.java
+++ b/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/types/TFloat16Test.java
@@ -18,6 +18,8 @@
 package org.tensorflow.types;
 
 import org.tensorflow.Tensor;
+import org.tensorflow.ndarray.NdArray;
+import org.tensorflow.ndarray.NdArrays;
 import org.tensorflow.ndarray.Shape;
 
 public class TFloat16Test extends NumericTypesTestBase<TFloat16, Float> {
@@ -27,6 +29,11 @@ TFloat16 allocateTensor(Shape shape) {
     return TFloat16.tensorOf(shape);
   }
 
+  @Override
+  NdArray<Float> allocateNdArray(Shape shape) {
+    return NdArrays.ofFloats(shape);
+  }
+
   @Override
   Float valueOf(Integer value) {
     return value.floatValue();
diff --git a/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/types/TFloat32Test.java b/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/types/TFloat32Test.java
index 21ef1586d12..938179ab8f3 100644
--- a/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/types/TFloat32Test.java
+++ b/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/types/TFloat32Test.java
@@ -18,6 +18,8 @@
 package org.tensorflow.types;
 
 import org.tensorflow.Tensor;
+import org.tensorflow.ndarray.NdArray;
+import org.tensorflow.ndarray.NdArrays;
 import org.tensorflow.ndarray.Shape;
 
 public class TFloat32Test extends NumericTypesTestBase<TFloat32, Float> {
@@ -27,6 +29,11 @@ TFloat32 allocateTensor(Shape shape) {
     return TFloat32.tensorOf(shape);
   }
 
+  @Override
+  NdArray<Float> allocateNdArray(Shape shape) {
+    return NdArrays.ofFloats(shape);
+  }
+
   @Override
   Float valueOf(Integer value) {
     return value.floatValue();
diff --git a/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/types/TFloat64Test.java b/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/types/TFloat64Test.java
index cb9de0bea19..5eaf391692c 100644
--- a/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/types/TFloat64Test.java
+++ b/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/types/TFloat64Test.java
@@ -18,6 +18,8 @@
 package org.tensorflow.types;
 
 import org.tensorflow.Tensor;
+import org.tensorflow.ndarray.NdArray;
+import org.tensorflow.ndarray.NdArrays;
 import org.tensorflow.ndarray.Shape;
 
 public class TFloat64Test extends NumericTypesTestBase<TFloat64, Double> {
@@ -27,6 +29,11 @@ TFloat64 allocateTensor(Shape shape) {
     return TFloat64.tensorOf(shape);
   }
 
+  @Override
+  NdArray<Double> allocateNdArray(Shape shape) {
+    return NdArrays.ofDoubles(shape);
+  }
+
   @Override
   Double valueOf(Integer value) {
     return value.doubleValue();
diff --git a/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/types/TInt32Test.java b/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/types/TInt32Test.java
index 8388c2834b2..5cae9392122 100644
--- a/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/types/TInt32Test.java
+++ b/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/types/TInt32Test.java
@@ -17,8 +17,19 @@
 
 package org.tensorflow.types;
 
+import static org.junit.jupiter.api.Assertions.assertEquals;
+
+import org.junit.jupiter.api.Test;
+import org.tensorflow.EagerSession;
 import org.tensorflow.Tensor;
+import org.tensorflow.ndarray.IntNdArray;
+import org.tensorflow.ndarray.NdArray;
+import org.tensorflow.ndarray.NdArrays;
 import org.tensorflow.ndarray.Shape;
+import org.tensorflow.ndarray.index.Indices;
+import org.tensorflow.op.Ops;
+import org.tensorflow.op.core.Constant;
+import org.tensorflow.tensor.IntTensor;
 
 public class TInt32Test extends NumericTypesTestBase<TInt32, Integer> {
 
@@ -27,6 +38,11 @@ TInt32 allocateTensor(Shape shape) {
     return TInt32.tensorOf(shape);
   }
 
+  @Override
+  NdArray<Integer> allocateNdArray(Shape shape) {
+    return NdArrays.ofInts(shape);
+  }
+
   @Override
   Integer valueOf(Integer value) {
     return value;
diff --git a/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/types/TInt64Test.java b/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/types/TInt64Test.java
index 55e4c7ead87..23af4f10c6a 100644
--- a/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/types/TInt64Test.java
+++ b/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/types/TInt64Test.java
@@ -18,6 +18,8 @@
 package org.tensorflow.types;
 
 import org.tensorflow.Tensor;
+import org.tensorflow.ndarray.NdArray;
+import org.tensorflow.ndarray.NdArrays;
 import org.tensorflow.ndarray.Shape;
 
 public class TInt64Test extends NumericTypesTestBase<TInt64, Long> {
@@ -27,6 +29,11 @@ TInt64 allocateTensor(Shape shape) {
     return TInt64.tensorOf(shape);
   }
 
+  @Override
+  NdArray<Long> allocateNdArray(Shape shape) {
+    return NdArrays.ofLongs(shape);
+  }
+
   @Override
   Long valueOf(Integer value) {
     return value.longValue();
diff --git a/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/types/TStringTest.java b/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/types/TStringTest.java
index 6d2b293be8b..4872f9be11f 100644
--- a/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/types/TStringTest.java
+++ b/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/types/TStringTest.java
@@ -79,7 +79,7 @@ public void defaultCharsetIsUtf8() {
     TString tensor = TString.tensorOf(NdArrays.scalarOfObject(BABY_CHICK));
     byte[] bytes = tensor.asBytes().getObject();
     assertArrayEquals(new byte[] { (byte)0xF0, (byte)0x9F, (byte)0x90, (byte)0xA5 }, bytes);
-    assertEquals(BABY_CHICK, tensor.data().getObject());
+    assertEquals(BABY_CHICK, tensor.getObject());
   }
 
   @Test
diff --git a/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/types/TUint8Test.java b/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/types/TUint8Test.java
index eb2ddd5bd38..e2ba52ef10b 100644
--- a/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/types/TUint8Test.java
+++ b/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/types/TUint8Test.java
@@ -18,6 +18,8 @@
 package org.tensorflow.types;
 
 import org.tensorflow.Tensor;
+import org.tensorflow.ndarray.NdArray;
+import org.tensorflow.ndarray.NdArrays;
 import org.tensorflow.ndarray.Shape;
 
 public class TUint8Test extends NumericTypesTestBase<TUint8, Byte> {
@@ -27,6 +29,11 @@ TUint8 allocateTensor(Shape shape) {
     return TUint8.tensorOf(shape);
   }
 
+  @Override
+  NdArray<Byte> allocateNdArray(Shape shape) {
+    return NdArrays.ofBytes(shape);
+  }
+
   @Override
   Byte valueOf(Integer value) {
     return value.byteValue();
diff --git a/tensorflow-framework/src/main/java/org/tensorflow/framework/data/DatasetIterator.java b/tensorflow-framework/src/main/java/org/tensorflow/framework/data/DatasetIterator.java
index f4c4b681715..0bad6a41214 100644
--- a/tensorflow-framework/src/main/java/org/tensorflow/framework/data/DatasetIterator.java
+++ b/tensorflow-framework/src/main/java/org/tensorflow/framework/data/DatasetIterator.java
@@ -272,7 +272,7 @@ public Iterator<List<Operand<?>>> iterator() {
 
       @Override
       public boolean hasNext() {
-        return nextOptional.hasValue().data().getBoolean();
+        return nextOptional.hasValue().asTensor().getBoolean();
       }
 
       @Override
diff --git a/tensorflow-framework/src/test/java/org/tensorflow/framework/data/BatchDatasetTest.java b/tensorflow-framework/src/test/java/org/tensorflow/framework/data/BatchDatasetTest.java
index 6a54cb08de6..e3c5c6ad06b 100644
--- a/tensorflow-framework/src/test/java/org/tensorflow/framework/data/BatchDatasetTest.java
+++ b/tensorflow-framework/src/test/java/org/tensorflow/framework/data/BatchDatasetTest.java
@@ -45,14 +45,10 @@ public void testEagerBatchDataset() {
 
     int count = 0;
     for (List<Operand<?>> components : dataset) {
-      try (Tensor<TInt32> batch1 =
-               components.get(0).asTensor().expect(TInt32.DTYPE);
-           Tensor<TInt32> batch2 =
-               components.get(1).asTensor().expect(TInt32.DTYPE);) {
-
-        assertEquals(testMatrix1.slice(range(count, count + 2)), batch1.data());
-        assertEquals(testMatrix2.slice(range(count, count + 2)), batch2.data());
-
+      try (TInt32 batch1 = components.get(0).asTensor(TInt32.DTYPE);
+           TInt32 batch2 = components.get(1).asTensor(TInt32.DTYPE);) {
+        assertEquals(testMatrix1.slice(range(count, count + 2)), batch1);
+        assertEquals(testMatrix2.slice(range(count, count + 2)), batch2);
         count += 2;
       }
     }
@@ -63,23 +59,16 @@ public void testDropLastBatch() {
     Ops tf = Ops.create();
     Dataset dataset = Dataset
         .fromTensorSlices(tf,
-            Arrays.asList(
-                tf.constant(testMatrix1),
-                tf.constant(testMatrix2)),
+            Arrays.asList(tf.constant(testMatrix1), tf.constant(testMatrix2)),
             Arrays.asList(TInt32.DTYPE, TInt32.DTYPE))
         .batch(3, true);
 
     int count = 0;
     for (List<Operand<?>> components : dataset) {
-
-      try (Tensor<TInt32> batch1 =
-               components.get(0).asTensor().expect(TInt32.DTYPE);
-           Tensor<TInt32> batch2 =
-               components.get(1).asTensor().expect(TInt32.DTYPE);) {
-
-        assertEquals(testMatrix1.slice(range(count, count + 3)), batch1.data());
-        assertEquals(testMatrix2.slice(range(count, count + 3)), batch2.data());
-
+      try (TInt32 batch1 = components.get(0).asTensor(TInt32.DTYPE);
+          TInt32 batch2 = components.get(1).asTensor(TInt32.DTYPE);) {
+        assertEquals(testMatrix1.slice(range(count, count + 3)), batch1);
+        assertEquals(testMatrix2.slice(range(count, count + 3)), batch2);
         count += 3;
       }
     }
@@ -90,9 +79,7 @@ public void testKeepLastBatch() {
     Ops tf = Ops.create();
     Dataset dataset = Dataset
         .fromTensorSlices(tf,
-            Arrays.asList(
-                tf.constant(testMatrix1),
-                tf.constant(testMatrix2)),
+            Arrays.asList(tf.constant(testMatrix1), tf.constant(testMatrix2)),
             Arrays.asList(TInt32.DTYPE, TInt32.DTYPE))
         .batch(3, false);
 
@@ -100,21 +87,15 @@ public void testKeepLastBatch() {
     boolean foundLastBatch = false;
 
     for (List<Operand<?>> components : dataset) {
-      try (Tensor<TInt32> batch1 =
-               components.get(0).asTensor().expect(TInt32.DTYPE);
-           Tensor<TInt32> batch2 =
-               components.get(1).asTensor().expect(TInt32.DTYPE);) {
+      try (TInt32 batch1 = components.get(0).asTensor(TInt32.DTYPE);
+           TInt32 batch2 = components.get(1).asTensor(TInt32.DTYPE);) {
         if (count == 0) {
-          assertEquals(testMatrix1.slice(range(count, count + 3)),
-              batch1.data());
-          assertEquals(testMatrix2.slice(range(count, count + 3)),
-              batch2.data());
+          assertEquals(testMatrix1.slice(range(count, count + 3)), batch1);
+          assertEquals(testMatrix2.slice(range(count, count + 3)), batch2);
           count += 3;
         } else {
-          assertEquals(testMatrix1.slice(range(count, count + 1)),
-              batch1.data());
-          assertEquals(testMatrix2.slice(range(count, count + 1)),
-              batch2.data());
+          assertEquals(testMatrix1.slice(range(count, count + 1)), batch1);
+          assertEquals(testMatrix2.slice(range(count, count + 1)), batch2);
           foundLastBatch = true;
         }
       }
diff --git a/tensorflow-framework/src/test/java/org/tensorflow/framework/data/DatasetIteratorTest.java b/tensorflow-framework/src/test/java/org/tensorflow/framework/data/DatasetIteratorTest.java
index 6bb6e21f330..d89419e8a3d 100644
--- a/tensorflow-framework/src/test/java/org/tensorflow/framework/data/DatasetIteratorTest.java
+++ b/tensorflow-framework/src/test/java/org/tensorflow/framework/data/DatasetIteratorTest.java
@@ -56,10 +56,10 @@ public void testGraphIteration() {
           try {
             List<Tensor<?>> outputs = session.runner().fetch(x).fetch(y).run();
 
-            try (Tensor<TInt32> xBatch = outputs.get(0).expect(TInt32.DTYPE);
-                Tensor<TInt32> yBatch = outputs.get(1).expect(TInt32.DTYPE)) {
-              assertEquals(testMatrix1.get(batches), xBatch.data());
-              assertEquals(testMatrix2.get(batches), yBatch.data());
+            try (TInt32 xBatch = outputs.get(0).expect(TInt32.DTYPE);
+                TInt32 yBatch = outputs.get(1).expect(TInt32.DTYPE)) {
+              assertEquals(testMatrix1.get(batches), xBatch);
+              assertEquals(testMatrix2.get(batches), yBatch);
               batches++;
             }
           } catch (TFOutOfRangeException e) {
@@ -82,12 +82,10 @@ public void testEagerIteration() {
     Dataset dataset = Dataset.fromTensorSlices(tf, tensors, dataTypes);
     int count = 0;
     for (List<Operand<?>> outputs : dataset) {
-      try (Tensor<TInt32> batch1 = outputs.get(0).asTensor().expect(TInt32.DTYPE);
-          Tensor<TInt32> batch2 = outputs.get(1).asTensor().expect(TInt32.DTYPE); ) {
-
-        assertEquals(testMatrix1.get(count), batch1.data());
-        assertEquals(testMatrix2.get(count), batch2.data());
-
+      try (TInt32 batch1 = outputs.get(0).asTensor(TInt32.DTYPE);
+          TInt32 batch2 = outputs.get(1).asTensor(TInt32.DTYPE); ) {
+        assertEquals(testMatrix1.get(count), batch1);
+        assertEquals(testMatrix2.get(count), batch2);
         count++;
       }
     }
diff --git a/tensorflow-framework/src/test/java/org/tensorflow/framework/data/MapDatasetTest.java b/tensorflow-framework/src/test/java/org/tensorflow/framework/data/MapDatasetTest.java
index 5960442ff70..c1dc0a98097 100644
--- a/tensorflow-framework/src/test/java/org/tensorflow/framework/data/MapDatasetTest.java
+++ b/tensorflow-framework/src/test/java/org/tensorflow/framework/data/MapDatasetTest.java
@@ -81,12 +81,10 @@ public void testGraphIteration() {
           try {
             List<Tensor<?>> outputs = session.runner().fetch(X).fetch(y).run();
 
-            try (Tensor<TInt32> XBatch = outputs.get(0).expect(TInt32.DTYPE);
-                Tensor<TInt32> yBatch = outputs.get(1).expect(TInt32.DTYPE)) {
-
-              assertEquals(mapped1.get(batches), XBatch.data());
-              assertEquals(mapped2.get(batches), yBatch.data());
-
+            try (TInt32 XBatch = outputs.get(0).expect(TInt32.DTYPE);
+                TInt32 yBatch = outputs.get(1).expect(TInt32.DTYPE)) {
+              assertEquals(mapped1.get(batches), XBatch);
+              assertEquals(mapped2.get(batches), yBatch);
               batches++;
             }
           } catch (TFOutOfRangeException e) {
@@ -114,12 +112,10 @@ public void testEagerIteration() {
 
     int count = 0;
     for (List<Operand<?>> outputs : dataset) {
-      try (Tensor<TInt32> XBatch = outputs.get(0).asTensor().expect(TInt32.DTYPE);
-          Tensor<TInt32> yBatch = outputs.get(1).asTensor().expect(TInt32.DTYPE); ) {
-
-        assertEquals(mapped1.get(count), XBatch.data());
-        assertEquals(mapped2.get(count), yBatch.data());
-
+      try (TInt32 XBatch = outputs.get(0).asTensor(TInt32.DTYPE);
+          TInt32 yBatch = outputs.get(1).asTensor(TInt32.DTYPE); ) {
+        assertEquals(mapped1.get(count), XBatch);
+        assertEquals(mapped2.get(count), yBatch);
         count++;
       }
     }
diff --git a/tensorflow-framework/src/test/java/org/tensorflow/framework/data/SkipDatasetTest.java b/tensorflow-framework/src/test/java/org/tensorflow/framework/data/SkipDatasetTest.java
index 9ff8080034d..eba3a273380 100644
--- a/tensorflow-framework/src/test/java/org/tensorflow/framework/data/SkipDatasetTest.java
+++ b/tensorflow-framework/src/test/java/org/tensorflow/framework/data/SkipDatasetTest.java
@@ -40,11 +40,10 @@ public void testEagerSkipDataset() {
 
     int count = 2;
     for (List<Operand<?>> components : dataset) {
-      try (Tensor<TInt32> batch1 = components.get(0).asTensor().expect(TInt32.DTYPE);
-          Tensor<TInt32> batch2 =
-              components.get(1).asTensor().expect(TInt32.DTYPE); ) {
-        assertEquals(testMatrix1.get(count), batch1.data());
-        assertEquals(testMatrix2.get(count), batch2.data());
+      try (TInt32 batch1 = components.get(0).asTensor(TInt32.DTYPE);
+          TInt32 batch2 = components.get(1).asTensor(TInt32.DTYPE); ) {
+        assertEquals(testMatrix1.get(count), batch1);
+        assertEquals(testMatrix2.get(count), batch2);
         count++;
       }
     }
diff --git a/tensorflow-framework/src/test/java/org/tensorflow/framework/data/TakeDatasetTest.java b/tensorflow-framework/src/test/java/org/tensorflow/framework/data/TakeDatasetTest.java
index 4419f4660db..e07272e3718 100644
--- a/tensorflow-framework/src/test/java/org/tensorflow/framework/data/TakeDatasetTest.java
+++ b/tensorflow-framework/src/test/java/org/tensorflow/framework/data/TakeDatasetTest.java
@@ -41,11 +41,10 @@ public void testEagerTakeDataset() {
 
     int count = 0;
     for (List<Operand<?>> components : dataset) {
-      try (Tensor<TInt32> batch1 = components.get(0).asTensor().expect(TInt32.DTYPE);
-          Tensor<TInt32> batch2 = components.get(1).asTensor().expect(TInt32.DTYPE); ) {
-
-        assertEquals(testMatrix1.get(count), batch1.data());
-        assertEquals(testMatrix2.get(count), batch2.data());
+      try (TInt32 batch1 = components.get(0).asTensor(TInt32.DTYPE);
+          TInt32 batch2 = components.get(1).asTensor(TInt32.DTYPE); ) {
+        assertEquals(testMatrix1.get(count), batch1);
+        assertEquals(testMatrix2.get(count), batch2);
         count++;
       }
     }

From 597cdb7923f0ce9efd6b6debbac55e80f690d884 Mon Sep 17 00:00:00 2001
From: klessard <klessard@expedia.com>
Date: Sat, 24 Oct 2020 23:55:05 -0400
Subject: [PATCH 05/11] Change T*.DTYPE for T*.class

---
 .../src/bazel/op_generator/java_defs.h        |    3 -
 .../src/bazel/op_generator/op_generator.cc    |    2 +-
 .../src/bazel/op_generator/op_specs.cc        |    4 +-
 .../src/bazel/op_generator/source_writer.cc   |   27 +-
 .../src/bazel/op_generator/source_writer.h    |    1 +
 .../org/tensorflow/op/AudioOps.java           |  147 -
 .../org/tensorflow/op/BitwiseOps.java         |  271 -
 .../tensorflow/op/DataExperimentalOps.java    |   61 -
 .../org/tensorflow/op/DataOps.java            |  413 -
 .../org/tensorflow/op/DebuggingOps.java       |   50 -
 .../org/tensorflow/op/DtypesOps.java          |  105 -
 .../org/tensorflow/op/ImageOps.java           |  945 --
 .../annotations/org/tensorflow/op/IoOps.java  | 1033 ---
 .../org/tensorflow/op/LinalgOps.java          | 1609 ----
 .../org/tensorflow/op/LinalgSparseOps.java    |  463 -
 .../org/tensorflow/op/MathOps.java            | 2402 ------
 .../org/tensorflow/op/NnRawOps.java           |   74 -
 .../annotations/org/tensorflow/op/Ops.java    | 7637 -----------------
 .../org/tensorflow/op/QuantizationOps.java    |  631 --
 .../org/tensorflow/op/RaggedOps.java          |   63 -
 .../org/tensorflow/op/RandomOps.java          |  586 --
 .../org/tensorflow/op/ShapeOps.java           |  472 -
 .../org/tensorflow/op/SignalOps.java          |  437 -
 .../org/tensorflow/op/SparseOps.java          | 1422 ---
 .../org/tensorflow/op/StringsOps.java         |  620 --
 .../org/tensorflow/op/SummaryOps.java         |  198 -
 .../org/tensorflow/op/TrainOps.java           | 1661 ----
 .../annotations/org/tensorflow/op/XlaOps.java |  382 -
 .../op/collective/BroadcastRecv.java          |    3 +-
 .../java/org/tensorflow/op/core/Barrier.java  |    5 +-
 .../tensorflow/op/core/BarrierTakeMany.java   |    5 +-
 .../java/org/tensorflow/op/core/Bitcast.java  |    3 +-
 .../org/tensorflow/op/core/DecodeProto.java   |    5 +-
 .../org/tensorflow/op/core/DeviceIndex.java   |    2 +-
 .../op/core/DummySeedGenerator.java           |   66 -
 .../java/org/tensorflow/op/core/Empty.java    |    3 +-
 .../tensorflow/op/core/EmptyTensorList.java   |    3 +-
 .../tensorflow/op/core/GetSessionTensor.java  |    3 +-
 .../org/tensorflow/op/core/HashTable.java     |    3 +-
 .../op/core/HistogramFixedWidth.java          |    5 +-
 .../tensorflow/op/core/ImmutableConst.java    |    3 +-
 .../java/org/tensorflow/op/core/LinSpace.java |    4 +-
 .../tensorflow/op/core/LookupTableExport.java |    3 +-
 .../org/tensorflow/op/core/LowerBound.java    |    5 +-
 .../java/org/tensorflow/op/core/MapClear.java |    5 +-
 .../tensorflow/op/core/MapIncompleteSize.java |    5 +-
 .../java/org/tensorflow/op/core/MapPeek.java  |    5 +-
 .../java/org/tensorflow/op/core/MapSize.java  |    5 +-
 .../java/org/tensorflow/op/core/MapStage.java |    5 +-
 .../org/tensorflow/op/core/MapUnstage.java    |    5 +-
 .../tensorflow/op/core/MapUnstageNoKey.java   |    5 +-
 .../tensorflow/op/core/MlirPassthroughOp.java |    5 +-
 .../op/core/MutableDenseHashTable.java        |    3 +-
 .../tensorflow/op/core/MutableHashTable.java  |    3 +-
 .../op/core/MutableHashTableOfTensors.java    |    3 +-
 .../tensorflow/op/core/OrderedMapClear.java   |    5 +-
 .../op/core/OrderedMapIncompleteSize.java     |    5 +-
 .../tensorflow/op/core/OrderedMapPeek.java    |    5 +-
 .../tensorflow/op/core/OrderedMapSize.java    |    5 +-
 .../tensorflow/op/core/OrderedMapStage.java   |    5 +-
 .../tensorflow/op/core/OrderedMapUnstage.java |    5 +-
 .../op/core/OrderedMapUnstageNoKey.java       |    5 +-
 .../org/tensorflow/op/core/Placeholder.java   |    3 +-
 .../tensorflow/op/core/ReadVariableOp.java    |    3 +-
 .../gen/java/org/tensorflow/op/core/Recv.java |    3 +-
 .../op/core/RemoteFusedGraphExecute.java      |    5 +-
 .../tensorflow/op/core/ResourceCountUpTo.java |    3 +-
 .../tensorflow/op/core/ResourceGather.java    |    3 +-
 .../tensorflow/op/core/ResourceGatherNd.java  |    3 +-
 .../op/core/ResourceScatterNdMax.java         |    6 +-
 .../op/core/ResourceScatterNdMin.java         |    6 +-
 .../org/tensorflow/op/core/ScatterNdMax.java  |    8 +-
 .../org/tensorflow/op/core/ScatterNdMin.java  |    8 +-
 .../org/tensorflow/op/core/SetDiff1d.java     |    5 +-
 .../java/org/tensorflow/op/core/Shape.java    |    5 +-
 .../java/org/tensorflow/op/core/ShapeN.java   |    5 +-
 .../gen/java/org/tensorflow/op/core/Size.java |    5 +-
 .../org/tensorflow/op/core/StageClear.java    |    5 +-
 .../org/tensorflow/op/core/StagePeek.java     |    5 +-
 .../org/tensorflow/op/core/StageSize.java     |    5 +-
 .../tensorflow/op/core/TemporaryVariable.java |    3 +-
 .../org/tensorflow/op/core/TensorArray.java   |    3 +-
 .../tensorflow/op/core/TensorArrayConcat.java |    3 +-
 .../tensorflow/op/core/TensorArrayGather.java |    3 +-
 .../tensorflow/op/core/TensorArrayPack.java   |    3 +-
 .../tensorflow/op/core/TensorArrayRead.java   |    3 +-
 .../tensorflow/op/core/TensorListConcat.java  |    3 +-
 .../op/core/TensorListConcatLists.java        |    3 +-
 .../op/core/TensorListElementShape.java       |    3 +-
 .../tensorflow/op/core/TensorListGather.java  |    3 +-
 .../tensorflow/op/core/TensorListGetItem.java |    3 +-
 .../tensorflow/op/core/TensorListPopBack.java |    3 +-
 .../tensorflow/op/core/TensorListReserve.java |    3 +-
 .../tensorflow/op/core/TensorListStack.java   |    3 +-
 .../tensorflow/op/core/TensorScatterMax.java  |   75 -
 .../tensorflow/op/core/TensorScatterMin.java  |   75 -
 .../op/core/TensorScatterNdMax.java           |    8 +-
 .../op/core/TensorScatterNdMin.java           |    8 +-
 .../java/org/tensorflow/op/core/Unique.java   |    5 +-
 .../tensorflow/op/core/UniqueWithCounts.java  |    5 +-
 .../java/org/tensorflow/op/core/Unstage.java  |    5 +-
 .../org/tensorflow/op/core/UpperBound.java    |    5 +-
 .../org/tensorflow/op/core/VarHandleOp.java   |    3 +-
 .../java/org/tensorflow/op/core/Variable.java |    3 +-
 .../org/tensorflow/op/core/VariableShape.java |    5 +-
 .../op/core/XlaSpmdFullToShardShape.java      |    4 +-
 .../op/core/XlaSpmdShardToFullShape.java      |    4 +-
 .../tensorflow/op/data/AnonymousIterator.java |    5 +-
 .../op/data/AnonymousMultiDeviceIterator.java |    5 +-
 .../tensorflow/op/data/AssertNextDataset.java |    5 +-
 .../tensorflow/op/data/AutoShardDataset.java  |    5 +-
 .../org/tensorflow/op/data/BatchDataset.java  |    5 +-
 .../op/data/BytesProducedStatsDataset.java    |    5 +-
 .../org/tensorflow/op/data/CacheDataset.java  |    5 +-
 .../tensorflow/op/data/CacheDatasetV2.java    |    5 +-
 .../op/data/ChooseFastestDataset.java         |    5 +-
 .../op/data/ConcatenateDataset.java           |    5 +-
 .../op/data/DatasetToSingleElement.java       |    5 +-
 .../op/data/DenseToSparseBatchDataset.java    |    5 +-
 .../op/data/DirectedInterleaveDataset.java    |    5 +-
 .../op/data/FilterByLastComponentDataset.java |    5 +-
 .../op/data/IgnoreErrorsDataset.java          |    5 +-
 .../op/data/InitializeTableFromDataset.java   |    4 +-
 .../java/org/tensorflow/op/data/Iterator.java |    5 +-
 .../op/data/IteratorFromStringHandle.java     |    5 +-
 .../tensorflow/op/data/IteratorGetNext.java   |    5 +-
 .../op/data/IteratorGetNextAsOptional.java    |    5 +-
 .../op/data/IteratorGetNextSync.java          |    5 +-
 .../org/tensorflow/op/data/LMDBDataset.java   |    5 +-
 .../op/data/LatencyStatsDataset.java          |    5 +-
 .../op/data/MaxIntraOpParallelismDataset.java |    5 +-
 .../org/tensorflow/op/data/ModelDataset.java  |    5 +-
 .../op/data/MultiDeviceIterator.java          |    5 +-
 .../MultiDeviceIteratorFromStringHandle.java  |    5 +-
 .../MultiDeviceIteratorGetNextFromShard.java  |    5 +-
 .../op/data/NonSerializableDataset.java       |    5 +-
 .../tensorflow/op/data/OptimizeDataset.java   |    5 +-
 .../tensorflow/op/data/OptionalGetValue.java  |    5 +-
 .../tensorflow/op/data/PrefetchDataset.java   |    5 +-
 .../op/data/PrivateThreadPoolDataset.java     |    5 +-
 .../org/tensorflow/op/data/RandomDataset.java |    5 +-
 .../org/tensorflow/op/data/RangeDataset.java  |    5 +-
 .../tensorflow/op/data/RebatchDataset.java    |    5 +-
 .../tensorflow/op/data/RegisterDataset.java   |    8 +-
 .../org/tensorflow/op/data/RepeatDataset.java |    5 +-
 .../tensorflow/op/data/SamplingDataset.java   |    5 +-
 .../op/data/SetStatsAggregatorDataset.java    |    5 +-
 .../org/tensorflow/op/data/ShardDataset.java  |    5 +-
 .../op/data/ShuffleAndRepeatDataset.java      |   62 +-
 .../tensorflow/op/data/ShuffleDataset.java    |   48 +-
 .../org/tensorflow/op/data/SkipDataset.java   |    5 +-
 .../org/tensorflow/op/data/SleepDataset.java  |    5 +-
 .../op/data/SlidingWindowDataset.java         |    5 +-
 .../tensorflow/op/data/SnapshotDataset.java   |  376 -
 .../org/tensorflow/op/data/SqlDataset.java    |    5 +-
 .../org/tensorflow/op/data/TakeDataset.java   |    5 +-
 .../tensorflow/op/data/ThreadPoolDataset.java |    5 +-
 .../tensorflow/op/data/UnbatchDataset.java    |    5 +-
 .../org/tensorflow/op/data/UniqueDataset.java |    5 +-
 .../org/tensorflow/op/data/WindowDataset.java |    5 +-
 .../org/tensorflow/op/data/ZipDataset.java    |    5 +-
 .../AssertCardinalityDataset.java             |    5 +-
 .../data/experimental/AssertNextDataset.java  |    5 +-
 .../data/experimental/AutoShardDataset.java   |    5 +-
 .../BytesProducedStatsDataset.java            |    5 +-
 .../experimental/ChooseFastestDataset.java    |    5 +-
 .../op/data/experimental/CompressElement.java |    4 +-
 .../data/experimental/DataServiceDataset.java |   21 +-
 .../DenseToSparseBatchDataset.java            |    5 +-
 .../DirectedInterleaveDataset.java            |    5 +-
 .../experimental/DummyIterationCounter.java   |    2 +-
 .../experimental/IgnoreErrorsDataset.java     |    5 +-
 .../experimental/LatencyStatsDataset.java     |    5 +-
 .../op/data/experimental/LmdbDataset.java     |    5 +-
 .../MaxIntraOpParallelismDataset.java         |    5 +-
 .../experimental/NonSerializableDataset.java  |    5 +-
 .../experimental/ParseExampleDataset.java     |   11 +-
 .../PrivateThreadPoolDataset.java             |    5 +-
 .../op/data/experimental/RandomDataset.java   |    5 +-
 .../op/data/experimental/RebatchDataset.java  |    5 +-
 .../SetStatsAggregatorDataset.java            |    5 +-
 .../op/data/experimental/SleepDataset.java    |    5 +-
 .../experimental/SlidingWindowDataset.java    |    5 +-
 .../op/data/experimental/SqlDataset.java      |    5 +-
 .../data/experimental/ThreadPoolDataset.java  |    5 +-
 .../op/data/experimental/UnbatchDataset.java  |    5 +-
 .../data/experimental/UncompressElement.java  |    7 +-
 .../op/data/experimental/UniqueDataset.java   |    5 +-
 .../op/debugging/DebugNumericsSummary.java    |    5 +-
 .../java/org/tensorflow/op/dtypes/Cast.java   |    3 +-
 .../org/tensorflow/op/dtypes/Complex.java     |    3 +-
 .../op/image/CropAndResizeGradImage.java      |    3 +-
 .../org/tensorflow/op/image/DecodePng.java    |    5 +-
 .../tensorflow/op/image/ExtractGlimpse.java   |    6 +-
 .../tensorflow/op/image/ExtractJpegShape.java |    5 +-
 .../org/tensorflow/op/io/DecodePaddedRaw.java |    3 +-
 .../java/org/tensorflow/op/io/DecodeRaw.java  |    3 +-
 .../op/io/DeserializeManySparse.java          |    3 +-
 .../java/org/tensorflow/op/io/FifoQueue.java  |    5 +-
 .../tensorflow/op/io/PaddingFifoQueue.java    |    5 +-
 .../org/tensorflow/op/io/ParseExample.java    |    9 +-
 .../op/io/ParseSequenceExample.java           |   17 +-
 .../tensorflow/op/io/ParseSingleExample.java  |    5 +-
 .../op/io/ParseSingleSequenceExample.java     |    9 +-
 .../org/tensorflow/op/io/ParseTensor.java     |    3 +-
 .../org/tensorflow/op/io/PriorityQueue.java   |    5 +-
 .../org/tensorflow/op/io/QueueDequeue.java    |    5 +-
 .../tensorflow/op/io/QueueDequeueMany.java    |    5 +-
 .../tensorflow/op/io/QueueDequeueUpTo.java    |    5 +-
 .../tensorflow/op/io/RandomShuffleQueue.java  |    5 +-
 .../tensorflow/op/io/SerializeManySparse.java |    5 +-
 .../org/tensorflow/op/io/SerializeSparse.java |    5 +-
 .../op/linalg/BandedTriangularSolve.java      |    6 +-
 .../java/org/tensorflow/op/linalg/Eig.java    |    3 +-
 .../gen/java/org/tensorflow/op/linalg/Lu.java |    5 +-
 .../tensorflow/op/linalg/QuantizedMatMul.java |    3 +-
 .../op/linalg/QuantizedMatMulWithBias.java    |    3 +-
 .../QuantizedMatMulWithBiasAndRelu.java       |    3 +-
 ...zedMatMulWithBiasAndReluAndRequantize.java |    3 +-
 .../sparse/CSRSparseMatrixComponents.java     |    3 +-
 .../linalg/sparse/CSRSparseMatrixToDense.java |    3 +-
 .../sparse/CSRSparseMatrixToSparseTensor.java |    3 +-
 .../op/linalg/sparse/SparseMatrixSoftmax.java |    3 +-
 .../sparse/SparseMatrixSoftmaxGrad.java       |    3 +-
 .../sparse/SparseMatrixSparseCholesky.java    |    3 +-
 .../sparse/SparseMatrixSparseMatMul.java      |    3 +-
 .../linalg/sparse/SparseMatrixTranspose.java  |    3 +-
 .../op/linalg/sparse/SparseMatrixZeros.java   |    3 +-
 .../java/org/tensorflow/op/math/Angle.java    |    5 +-
 .../java/org/tensorflow/op/math/ArgMax.java   |    5 +-
 .../java/org/tensorflow/op/math/ArgMin.java   |    5 +-
 .../java/org/tensorflow/op/math/BesselI0.java |    5 +-
 .../org/tensorflow/op/math/BesselI0e.java     |   11 +-
 .../java/org/tensorflow/op/math/BesselI1.java |    5 +-
 .../org/tensorflow/op/math/BesselI1e.java     |   11 +-
 .../org/tensorflow/op/math/ComplexAbs.java    |    5 +-
 .../org/tensorflow/op/math/DenseBincount.java |    9 +-
 .../gen/java/org/tensorflow/op/math/Imag.java |    5 +-
 .../org/tensorflow/op/math/QuantizedAdd.java  |    3 +-
 .../org/tensorflow/op/math/QuantizedMul.java  |    3 +-
 .../gen/java/org/tensorflow/op/math/Real.java |    5 +-
 .../op/math/RequantizePerChannel.java         |    3 +-
 .../org/tensorflow/op/math/SobolSample.java   |    5 +-
 .../tensorflow/op/math/special/BesselJ0.java  |    5 +-
 .../tensorflow/op/math/special/BesselJ1.java  |    5 +-
 .../tensorflow/op/math/special/BesselK0.java  |    5 +-
 .../tensorflow/op/math/special/BesselK0e.java |    5 +-
 .../tensorflow/op/math/special/BesselK1.java  |    5 +-
 .../tensorflow/op/math/special/BesselK1e.java |    5 +-
 .../tensorflow/op/math/special/BesselY0.java  |    5 +-
 .../tensorflow/op/math/special/BesselY1.java  |    5 +-
 .../tensorflow/op/nn/CudnnRnnParamsSize.java  |    3 +-
 .../tensorflow/op/nn/MaxPoolWithArgmax.java   |    5 +-
 ...tizedBatchNormWithGlobalNormalization.java |    3 +-
 .../tensorflow/op/nn/QuantizedBiasAdd.java    |    3 +-
 .../op/nn/QuantizedConv2DAndRelu.java         |    3 +-
 .../QuantizedConv2DAndReluAndRequantize.java  |    3 +-
 .../op/nn/QuantizedConv2DAndRequantize.java   |    3 +-
 .../op/nn/QuantizedConv2DPerChannel.java      |    3 +-
 .../op/nn/QuantizedConv2DWithBias.java        |    3 +-
 .../op/nn/QuantizedConv2DWithBiasAndRelu.java |    3 +-
 ...zedConv2DWithBiasAndReluAndRequantize.java |    3 +-
 .../QuantizedConv2DWithBiasAndRequantize.java |    3 +-
 ...WithBiasSignedSumAndReluAndRequantize.java |    3 +-
 .../nn/QuantizedConv2DWithBiasSumAndRelu.java |    3 +-
 ...Conv2DWithBiasSumAndReluAndRequantize.java |    3 +-
 .../org/tensorflow/op/nn/QuantizedConv2d.java |    3 +-
 .../op/nn/QuantizedDepthwiseConv2D.java       |    3 +-
 .../nn/QuantizedDepthwiseConv2DWithBias.java  |    3 +-
 ...antizedDepthwiseConv2DWithBiasAndRelu.java |    3 +-
 ...iseConv2DWithBiasAndReluAndRequantize.java |    3 +-
 .../org/tensorflow/op/nn/QuantizedRelu.java   |    3 +-
 .../org/tensorflow/op/nn/QuantizedRelu6.java  |    3 +-
 .../org/tensorflow/op/nn/QuantizedReluX.java  |    3 +-
 .../nn/raw/SoftmaxCrossEntropyWithLogits.java |    7 +-
 .../SparseSoftmaxCrossEntropyWithLogits.java  |    7 +-
 .../op/quantization/Dequantize.java           |    5 +-
 .../tensorflow/op/quantization/Quantize.java  |    3 +-
 .../QuantizeDownAndShrinkRange.java           |    3 +-
 .../QuantizedMatMulWithBiasAndDequantize.java |    3 +-
 .../QuantizedMatMulWithBiasAndRequantize.java |    3 +-
 .../op/quantization/Requantize.java           |    3 +-
 .../tensorflow/op/ragged/RaggedBincount.java  |   11 +-
 .../op/ragged/RaggedCountSparseOutput.java    |    7 +-
 .../org/tensorflow/op/ragged/RaggedCross.java |   15 +-
 .../org/tensorflow/op/ragged/RaggedRange.java |    5 +-
 .../op/ragged/RaggedTensorFromVariant.java    |    7 +-
 .../op/random/AnonymousSeedGenerator.java     |    6 +-
 .../op/random/DeleteSeedGenerator.java        |    4 +-
 .../org/tensorflow/op/random/Multinomial.java |    5 +-
 .../op/random/NonDeterministicInts.java       |    5 +-
 .../tensorflow/op/random/RandomPoisson.java   |    5 +-
 .../op/random/RandomStandardNormal.java       |    3 +-
 .../tensorflow/op/random/RandomUniform.java   |    3 +-
 .../op/random/StatefulRandomBinomial.java     |    5 +-
 .../op/random/StatefulStandardNormal.java     |    5 +-
 .../op/random/StatefulTruncatedNormal.java    |    5 +-
 .../tensorflow/op/random/StatefulUniform.java |    5 +-
 .../op/random/StatefulUniformFullInt.java     |    3 +-
 .../op/random/StatelessMultinomial.java       |    5 +-
 ...StatelessParameterizedTruncatedNormal.java |   15 +-
 .../op/random/StatelessRandomBinomial.java    |    5 +-
 .../op/random/StatelessRandomNormal.java      |    5 +-
 .../op/random/StatelessRandomPoisson.java     |    3 +-
 .../op/random/StatelessRandomUniform.java     |    5 +-
 .../random/StatelessRandomUniformFullInt.java |    3 +-
 .../op/random/StatelessTruncatedNormal.java   |    5 +-
 .../tensorflow/op/random/TruncatedNormal.java |    3 +-
 .../experimental/DummySeedGenerator.java      |    2 +-
 .../java/org/tensorflow/op/signal/Irfft.java  |    5 +-
 .../org/tensorflow/op/signal/Irfft2d.java     |    5 +-
 .../org/tensorflow/op/signal/Irfft3d.java     |    5 +-
 .../java/org/tensorflow/op/signal/Rfft.java   |    3 +-
 .../java/org/tensorflow/op/signal/Rfft2d.java |    3 +-
 .../java/org/tensorflow/op/signal/Rfft3d.java |    3 +-
 .../op/sparse/DenseCountSparseOutput.java     |    5 +-
 .../op/sparse/DeserializeSparse.java          |    3 +-
 .../sparse/SparseAccumulatorTakeGradient.java |    3 +-
 .../tensorflow/op/sparse/SparseBincount.java  |   13 +-
 .../sparse/SparseConditionalAccumulator.java  |    3 +-
 .../op/sparse/SparseCountSparseOutput.java    |    9 +-
 .../org/tensorflow/op/sparse/SparseCross.java |   35 +-
 .../op/sparse/SparseCrossHashed.java          |   14 +-
 .../op/sparse/SparseSegmentMean.java          |    6 +-
 .../op/sparse/SparseSegmentMeanGrad.java      |    5 +-
 .../SparseSegmentMeanWithNumSegments.java     |    8 +-
 .../op/sparse/SparseSegmentSqrtN.java         |    6 +-
 .../op/sparse/SparseSegmentSqrtNGrad.java     |    5 +-
 .../SparseSegmentSqrtNWithNumSegments.java    |    8 +-
 .../op/sparse/SparseSegmentSum.java           |    6 +-
 .../SparseSegmentSumWithNumSegments.java      |    8 +-
 .../sparse/TakeManySparseFromTensorsMap.java  |    3 +-
 .../org/tensorflow/op/strings/ToNumber.java   |    5 +-
 .../tensorflow/op/strings/UnicodeDecode.java  |    5 +-
 .../op/strings/UnicodeDecodeWithOffsets.java  |    5 +-
 .../EnqueueTPUEmbeddingRaggedTensorBatch.java |    9 +-
 .../org/tensorflow/op/tpu/InfeedDequeue.java  |    3 +-
 .../tensorflow/op/tpu/InfeedDequeueTuple.java |    5 +-
 ...oadTPUEmbeddingProximalYogiParameters.java |    6 +-
 ...gProximalYogiParametersGradAccumDebug.java |    8 +-
 ...adientDescentParametersGradAccumDebug.java |    4 +-
 .../org/tensorflow/op/tpu/OutfeedDequeue.java |    3 +-
 .../op/tpu/OutfeedDequeueTuple.java           |    5 +-
 .../op/train/AccumulatorTakeGradient.java     |    3 +-
 .../op/train/ConditionalAccumulator.java      |    3 +-
 .../ResourceAccumulatorTakeGradient.java      |    3 +-
 .../train/ResourceConditionalAccumulator.java |    3 +-
 .../java/org/tensorflow/op/train/Restore.java |    5 +-
 .../org/tensorflow/op/train/RestoreSlice.java |    3 +-
 .../gen/java/org/tensorflow/op/xla/Recv.java  |    3 +-
 .../org/tensorflow/AbstractOperation.java     |    4 +-
 .../java/org/tensorflow/AbstractTensor.java   |   31 +-
 .../main/java/org/tensorflow/DataType.java    |   99 +-
 .../main/java/org/tensorflow/DataTypes.java   |   75 -
 .../java/org/tensorflow/EagerOperation.java   |   11 +-
 .../org/tensorflow/EagerOperationBuilder.java |   13 +-
 .../src/main/java/org/tensorflow/Graph.java   |    6 +-
 .../java/org/tensorflow/GraphOperation.java   |    4 +-
 .../org/tensorflow/GraphOperationBuilder.java |   13 +-
 .../src/main/java/org/tensorflow/Operand.java |   14 -
 .../java/org/tensorflow/OperationBuilder.java |    5 +-
 .../src/main/java/org/tensorflow/Output.java  |   15 +-
 .../main/java/org/tensorflow/Signature.java   |    2 +-
 .../src/main/java/org/tensorflow/Tensor.java  |   16 -
 .../main/java/org/tensorflow/TensorType.java  |   40 +
 .../main/java/org/tensorflow/TensorTypes.java |  114 +
 .../src/main/java/org/tensorflow/Tensors.java |   70 +-
 .../internal/tensor/BooleanTensorImpl.java    |    7 +-
 .../internal/tensor/ByteTensorImpl.java       |    7 +-
 .../internal/tensor/DoubleTensorImpl.java     |    7 +-
 .../internal/tensor/FloatTensorImpl.java      |    7 +-
 .../internal/tensor/IntTensorImpl.java        |    7 +-
 .../internal/tensor/LongTensorImpl.java       |    7 +-
 .../internal/tensor/StringTensorImpl.java     |    5 +-
 .../java/org/tensorflow/op/core/Constant.java |    7 +-
 .../java/org/tensorflow/op/core/Helpers.java  |    2 +-
 .../java/org/tensorflow/op/core/Shapes.java   |   59 +-
 .../java/org/tensorflow/op/core/Zeros.java    |    4 +-
 .../op/nn/SigmoidCrossEntropyWithLogits.java  |    2 +-
 .../op/nn/SoftmaxCrossEntropyWithLogits.java  |   24 +-
 .../SparseSoftmaxCrossEntropyWithLogits.java  |   15 +-
 .../java/org/tensorflow/types/TBfloat16.java  |   29 +-
 .../main/java/org/tensorflow/types/TBool.java |   29 +-
 .../java/org/tensorflow/types/TFloat16.java   |   29 +-
 .../java/org/tensorflow/types/TFloat32.java   |   29 +-
 .../java/org/tensorflow/types/TFloat64.java   |   29 +-
 .../java/org/tensorflow/types/TInt32.java     |   29 +-
 .../java/org/tensorflow/types/TInt64.java     |   29 +-
 .../java/org/tensorflow/types/TString.java    |   21 +-
 .../java/org/tensorflow/types/TUint8.java     |   29 +-
 .../types/annotation/TensorType.java          |   30 +
 .../org/tensorflow/types/family/TType.java    |    2 +
 .../{ => types}/tensor/BooleanTensor.java     |    0
 .../{ => types}/tensor/ByteTensor.java        |    0
 .../{ => types}/tensor/DoubleTensor.java      |    0
 .../{ => types}/tensor/FloatTensor.java       |    0
 .../{ => types}/tensor/IntTensor.java         |    0
 .../{ => types}/tensor/LongTensor.java        |    0
 .../{ => types}/tensor/StringTensor.java      |    0
 399 files changed, 1226 insertions(+), 23711 deletions(-)
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/AudioOps.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/BitwiseOps.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/DataExperimentalOps.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/DataOps.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/DebuggingOps.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/DtypesOps.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/ImageOps.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/IoOps.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/LinalgOps.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/LinalgSparseOps.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/MathOps.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/NnRawOps.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/Ops.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/QuantizationOps.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/RaggedOps.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/RandomOps.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/ShapeOps.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/SignalOps.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/SparseOps.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/StringsOps.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/SummaryOps.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/TrainOps.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/XlaOps.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/DummySeedGenerator.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorScatterMax.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorScatterMin.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SnapshotDataset.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/DataTypes.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/TensorType.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/TensorTypes.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/annotation/TensorType.java
 rename tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/{ => types}/tensor/BooleanTensor.java (100%)
 rename tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/{ => types}/tensor/ByteTensor.java (100%)
 rename tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/{ => types}/tensor/DoubleTensor.java (100%)
 rename tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/{ => types}/tensor/FloatTensor.java (100%)
 rename tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/{ => types}/tensor/IntTensor.java (100%)
 rename tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/{ => types}/tensor/LongTensor.java (100%)
 rename tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/{ => types}/tensor/StringTensor.java (100%)

diff --git a/tensorflow-core/tensorflow-core-api/src/bazel/op_generator/java_defs.h b/tensorflow-core/tensorflow-core-api/src/bazel/op_generator/java_defs.h
index e41dc2dd9df..6028c4ea71d 100644
--- a/tensorflow-core/tensorflow-core-api/src/bazel/op_generator/java_defs.h
+++ b/tensorflow-core/tensorflow-core-api/src/bazel/op_generator/java_defs.h
@@ -97,9 +97,6 @@ class Type {
   static Type IterableOf(const Type& type) {
     return Interface("Iterable").add_parameter(type);
   }
-  static Type DataTypeOf(const Type& type) {
-    return Class("DataType", "org.tensorflow").add_parameter(type);
-  }
   static Type ForDataType(DataType data_type) {
     switch (data_type) {
       case DataType::DT_BOOL:
diff --git a/tensorflow-core/tensorflow-core-api/src/bazel/op_generator/op_generator.cc b/tensorflow-core/tensorflow-core-api/src/bazel/op_generator/op_generator.cc
index 667ce463d26..4ad39a07a2a 100644
--- a/tensorflow-core/tensorflow-core-api/src/bazel/op_generator/op_generator.cc
+++ b/tensorflow-core/tensorflow-core-api/src/bazel/op_generator/op_generator.cc
@@ -177,7 +177,7 @@ void RenderSecondaryFactoryMethod(const OpSpec& op, const Type& op_class,
     if (attr.type().kind() == Type::GENERIC &&
         default_types.find(attr.type().name()) != default_types.end()) {
       factory_statement << default_types.at(attr.type().name()).name()
-                        << ".DTYPE";
+                        << ".class";
     } else {
       AddArgument(attr.var(), attr.description(), &factory, &factory_doc);
       factory_statement << attr.var().name();
diff --git a/tensorflow-core/tensorflow-core-api/src/bazel/op_generator/op_specs.cc b/tensorflow-core/tensorflow-core-api/src/bazel/op_generator/op_specs.cc
index 7906613569a..b7452c8f5b6 100644
--- a/tensorflow-core/tensorflow-core-api/src/bazel/op_generator/op_specs.cc
+++ b/tensorflow-core/tensorflow-core-api/src/bazel/op_generator/op_specs.cc
@@ -163,7 +163,7 @@ std::pair<Type, Type> TypeResolver::TypesOf(const OpDef_AttrDef& attr_def,
 
   } else if (attr_type == "type") {
     Type type = *iterable_out ? Type::Wildcard() : NextGeneric(FamilyOf(attr_def));
-    types = MakeTypePair(type, Type::Enum("DataType", "org.tensorflow"));
+    types = MakeTypePair(type, Type::Class("Class"));
 
   } else {
     LOG(FATAL) << "Cannot resolve data type for attribute \"" << attr_type
@@ -317,7 +317,7 @@ AttributeSpec CreateAttribute(const OpDef_AttrDef& attr_def,
   bool iterable = false;
   std::pair<Type, Type> types = type_resolver->TypesOf(attr_def, &iterable);
   Type var_type = types.first.kind() == Type::GENERIC
-                      ? Type::DataTypeOf(types.first)
+                      ? Type::ClassOf(types.first)
                       : types.first;
   if (iterable) {
     var_type = Type::ListOf(var_type);
diff --git a/tensorflow-core/tensorflow-core-api/src/bazel/op_generator/source_writer.cc b/tensorflow-core/tensorflow-core-api/src/bazel/op_generator/source_writer.cc
index 9c36405c44c..37315f0dff3 100644
--- a/tensorflow-core/tensorflow-core-api/src/bazel/op_generator/source_writer.cc
+++ b/tensorflow-core/tensorflow-core-api/src/bazel/op_generator/source_writer.cc
@@ -85,6 +85,7 @@ SourceWriter& SourceWriter::Append(const StringPiece& str) {
 SourceWriter& SourceWriter::AppendType(const Type& type) {
   if (type.wildcard()) {
     Append("?");
+    WriteTypeBounds(type.supertypes());
   } else {
     Append(type.name());
     if (!type.parameters().empty()) {
@@ -321,21 +322,27 @@ SourceWriter& SourceWriter::WriteGenerics(
       Append(", ");
     }
     Append(pt->name());
-    bool first_bound = true;
-    for (const Type& bound : pt->supertypes()) {
-      if (first_bound) {
-        Append(" extends ");
-        first_bound = false;
-      } else {
-        Append(" & ");
-      }
-      AppendType(bound);
-    }
+    WriteTypeBounds(pt->supertypes());
     first = false;
   }
   return Append(">");
 }
 
+SourceWriter& SourceWriter::WriteTypeBounds(
+    const std::list<Type>& bounds) {
+  bool first = true;
+  for (const Type& bound : bounds) {
+    if (first) {
+      Append(" extends ");
+      first = false;
+    } else {
+      Append(" & ");
+    }
+    AppendType(bound);
+  }
+  return *this;
+}
+
 SourceWriter::GenericNamespace* SourceWriter::PushGenericNamespace(
     int modifiers) {
   GenericNamespace* generic_namespace;
diff --git a/tensorflow-core/tensorflow-core-api/src/bazel/op_generator/source_writer.h b/tensorflow-core/tensorflow-core-api/src/bazel/op_generator/source_writer.h
index 097887083e7..26b97f7a9c4 100644
--- a/tensorflow-core/tensorflow-core-api/src/bazel/op_generator/source_writer.h
+++ b/tensorflow-core/tensorflow-core-api/src/bazel/op_generator/source_writer.h
@@ -213,6 +213,7 @@ class SourceWriter {
   SourceWriter& WriteJavadoc(const Javadoc& javadoc);
   SourceWriter& WriteAnnotations(const std::list<Annotation>& annotations);
   SourceWriter& WriteGenerics(const std::list<const Type*>& generics);
+  SourceWriter& WriteTypeBounds(const std::list<Type>& bounds);
   GenericNamespace* PushGenericNamespace(int modifiers);
   void PopGenericNamespace();
 };
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/AudioOps.java b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/AudioOps.java
deleted file mode 100644
index 16770394378..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/AudioOps.java
+++ /dev/null
@@ -1,147 +0,0 @@
-// Copyright 2020 The TensorFlow Authors. All Rights Reserved.
-//
-// Licensed under the Apache License, Version 2.0 (the "License");
-// you may not use this file except in compliance with the License.
-// You may obtain a copy of the License at
-//
-//     http://www.apache.org/licenses/LICENSE-2.0
-//
-// Unless required by applicable law or agreed to in writing, software
-// distributed under the License is distributed on an "AS IS" BASIS,
-// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-// See the License for the specific language governing permissions and
-// limitations under the License.
-// ==============================================================================
-//
-// This class has been generated, DO NOT EDIT!
-//
-package org.tensorflow.op;
-
-import org.tensorflow.Operand;
-import org.tensorflow.op.audio.AudioSpectrogram;
-import org.tensorflow.op.audio.DecodeWav;
-import org.tensorflow.op.audio.EncodeWav;
-import org.tensorflow.op.audio.Mfcc;
-import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.TString;
-
-/**
- * An API for building {@code audio} operations as {@link Op Op}s
- *
- * @see {@link Ops}
- */
-public final class AudioOps {
-  private final Scope scope;
-
-  AudioOps(Scope scope) {
-    this.scope = scope;
-  }
-
-  /**
-   * Produces a visualization of audio data over time.
-   *  <p>
-   *  Spectrograms are a standard way of representing audio information as a series of
-   *  slices of frequency information, one slice for each window of time. By joining
-   *  these together into a sequence, they form a distinctive fingerprint of the sound
-   *  over time.
-   *  <p>
-   *  This op expects to receive audio data as an input, stored as floats in the range
-   *  -1 to 1, together with a window width in samples, and a stride specifying how
-   *  far to move the window between slices. From this it generates a three
-   *  dimensional output. The first dimension is for the channels in the input, so a
-   *  stereo audio input would have two here for example. The second dimension is time,
-   *  with successive frequency slices. The third dimension has an amplitude value for
-   *  each frequency during that time slice.
-   *  <p>
-   *  This means the layout when converted and saved as an image is rotated 90 degrees
-   *  clockwise from a typical spectrogram. Time is descending down the Y axis, and
-   *  the frequency decreases from left to right.
-   *  <p>
-   *  Each value in the result represents the square root of the sum of the real and
-   *  imaginary parts of an FFT on the current window of samples. In this way, the
-   *  lowest dimension represents the power of each frequency in the current window,
-   *  and adjacent windows are concatenated in the next dimension.
-   *  <p>
-   *  To get a more intuitive and visual look at what this operation does, you can run
-   *  tensorflow/examples/wav_to_spectrogram to read in an audio file and save out the
-   *  resulting spectrogram as a PNG image.
-   *
-   * @param input Float representation of audio data.
-   * @param windowSize How wide the input window is in samples. For the highest efficiency
-   *  this should be a power of two, but other values are accepted.
-   * @param stride How widely apart the center of adjacent sample windows should be.
-   * @param options carries optional attributes values
-   * @return a new instance of AudioSpectrogram
-   */
-  public AudioSpectrogram audioSpectrogram(Operand<TFloat32> input, Long windowSize, Long stride,
-      AudioSpectrogram.Options... options) {
-    return AudioSpectrogram.create(scope, input, windowSize, stride, options);
-  }
-
-  /**
-   * Decode a 16-bit PCM WAV file to a float tensor.
-   *  <p>
-   *  The -32768 to 32767 signed 16-bit values will be scaled to -1.0 to 1.0 in float.
-   *  <p>
-   *  When desired_channels is set, if the input contains fewer channels than this
-   *  then the last channel will be duplicated to give the requested number, else if
-   *  the input has more channels than requested then the additional channels will be
-   *  ignored.
-   *  <p>
-   *  If desired_samples is set, then the audio will be cropped or padded with zeroes
-   *  to the requested length.
-   *  <p>
-   *  The first output contains a Tensor with the content of the audio samples. The
-   *  lowest dimension will be the number of channels, and the second will be the
-   *  number of samples. For example, a ten-sample-long stereo WAV file should give an
-   *  output shape of [10, 2].
-   *
-   * @param contents The WAV-encoded audio, usually from a file.
-   * @param options carries optional attributes values
-   * @return a new instance of DecodeWav
-   */
-  public DecodeWav decodeWav(Operand<TString> contents, DecodeWav.Options... options) {
-    return DecodeWav.create(scope, contents, options);
-  }
-
-  /**
-   * Encode audio data using the WAV file format.
-   *  <p>
-   *  This operation will generate a string suitable to be saved out to create a .wav
-   *  audio file. It will be encoded in the 16-bit PCM format. It takes in float
-   *  values in the range -1.0f to 1.0f, and any outside that value will be clamped to
-   *  that range.
-   *  <p>
-   *  `audio` is a 2-D float Tensor of shape `[length, channels]`.
-   *  `sample_rate` is a scalar Tensor holding the rate to use (e.g. 44100).
-   *
-   * @param audio 2-D with shape `[length, channels]`.
-   * @param sampleRate Scalar containing the sample frequency.
-   * @return a new instance of EncodeWav
-   */
-  public EncodeWav encodeWav(Operand<TFloat32> audio, Operand<TInt32> sampleRate) {
-    return EncodeWav.create(scope, audio, sampleRate);
-  }
-
-  /**
-   * Transforms a spectrogram into a form that's useful for speech recognition.
-   *  <p>
-   *  Mel Frequency Cepstral Coefficients are a way of representing audio data that's
-   *  been effective as an input feature for machine learning. They are created by
-   *  taking the spectrum of a spectrogram (a 'cepstrum'), and discarding some of the
-   *  higher frequencies that are less significant to the human ear. They have a long
-   *  history in the speech recognition world, and https://en.wikipedia.org/wiki/Mel-frequency_cepstrum
-   *  is a good resource to learn more.
-   *
-   * @param spectrogram Typically produced by the Spectrogram op, with magnitude_squared
-   *  set to true.
-   * @param sampleRate How many samples per second the source audio used.
-   * @param options carries optional attributes values
-   * @return a new instance of Mfcc
-   */
-  public Mfcc mfcc(Operand<TFloat32> spectrogram, Operand<TInt32> sampleRate,
-      Mfcc.Options... options) {
-    return Mfcc.create(scope, spectrogram, sampleRate, options);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/BitwiseOps.java b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/BitwiseOps.java
deleted file mode 100644
index 8ac6d565e51..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/BitwiseOps.java
+++ /dev/null
@@ -1,271 +0,0 @@
-// Copyright 2020 The TensorFlow Authors. All Rights Reserved.
-//
-// Licensed under the Apache License, Version 2.0 (the "License");
-// you may not use this file except in compliance with the License.
-// You may obtain a copy of the License at
-//
-//     http://www.apache.org/licenses/LICENSE-2.0
-//
-// Unless required by applicable law or agreed to in writing, software
-// distributed under the License is distributed on an "AS IS" BASIS,
-// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-// See the License for the specific language governing permissions and
-// limitations under the License.
-// ==============================================================================
-//
-// This class has been generated, DO NOT EDIT!
-//
-package org.tensorflow.op;
-
-import org.tensorflow.Operand;
-import org.tensorflow.op.bitwise.BitwiseAnd;
-import org.tensorflow.op.bitwise.BitwiseOr;
-import org.tensorflow.op.bitwise.BitwiseXor;
-import org.tensorflow.op.bitwise.Invert;
-import org.tensorflow.op.bitwise.LeftShift;
-import org.tensorflow.op.bitwise.RightShift;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * An API for building {@code bitwise} operations as {@link Op Op}s
- *
- * @see {@link Ops}
- */
-public final class BitwiseOps {
-  private final Scope scope;
-
-  BitwiseOps(Scope scope) {
-    this.scope = scope;
-  }
-
-  /**
-   * Elementwise computes the bitwise AND of `x` and `y`.
-   *  <p>
-   *  The result will have those bits set, that are set in both `x` and `y`. The
-   *  computation is performed on the underlying representations of `x` and `y`.
-   *  <p>
-   *  For example:
-   *  <pre>{@code
-   *  import tensorflow as tf
-   *  from tensorflow.python.ops import bitwise_ops
-   *  dtype_list = [tf.int8, tf.int16, tf.int32, tf.int64,
-   *                tf.uint8, tf.uint16, tf.uint32, tf.uint64]
-   *
-   *  for dtype in dtype_list:
-   *    lhs = tf.constant([0, 5, 3, 14], dtype=dtype)
-   *    rhs = tf.constant([5, 0, 7, 11], dtype=dtype)
-   *    exp = tf.constant([0, 0, 3, 10], dtype=tf.float32)
-   *
-   *    res = bitwise_ops.bitwise_and(lhs, rhs)
-   *    tf.assert_equal(tf.cast(res, tf.float32), exp) # TRUE
-   *  }</pre>
-   *
-   * @param <T> data type for {@code z()} output
-   * @param x
-   * @param y
-   * @return a new instance of BitwiseAnd
-   */
-  public <T extends TNumber> BitwiseAnd<T> bitwiseAnd(Operand<T> x, Operand<T> y) {
-    return BitwiseAnd.create(scope, x, y);
-  }
-
-  /**
-   * Elementwise computes the bitwise OR of `x` and `y`.
-   *  <p>
-   *  The result will have those bits set, that are set in `x`, `y` or both. The
-   *  computation is performed on the underlying representations of `x` and `y`.
-   *  <p>
-   *  For example:
-   *  <pre>{@code
-   *  import tensorflow as tf
-   *  from tensorflow.python.ops import bitwise_ops
-   *  dtype_list = [tf.int8, tf.int16, tf.int32, tf.int64,
-   *                tf.uint8, tf.uint16, tf.uint32, tf.uint64]
-   *
-   *  for dtype in dtype_list:
-   *    lhs = tf.constant([0, 5, 3, 14], dtype=dtype)
-   *    rhs = tf.constant([5, 0, 7, 11], dtype=dtype)
-   *    exp = tf.constant([5, 5, 7, 15], dtype=tf.float32)
-   *
-   *    res = bitwise_ops.bitwise_or(lhs, rhs)
-   *    tf.assert_equal(tf.cast(res,  tf.float32), exp)  # TRUE
-   *  }</pre>
-   *
-   * @param <T> data type for {@code z()} output
-   * @param x
-   * @param y
-   * @return a new instance of BitwiseOr
-   */
-  public <T extends TNumber> BitwiseOr<T> bitwiseOr(Operand<T> x, Operand<T> y) {
-    return BitwiseOr.create(scope, x, y);
-  }
-
-  /**
-   * Elementwise computes the bitwise XOR of `x` and `y`.
-   *  <p>
-   *  The result will have those bits set, that are different in `x` and `y`. The
-   *  computation is performed on the underlying representations of `x` and `y`.
-   *  <p>
-   *  For example:
-   *  <pre>{@code
-   *  import tensorflow as tf
-   *  from tensorflow.python.ops import bitwise_ops
-   *  dtype_list = [tf.int8, tf.int16, tf.int32, tf.int64,
-   *                tf.uint8, tf.uint16, tf.uint32, tf.uint64]
-   *
-   *  for dtype in dtype_list:
-   *    lhs = tf.constant([0, 5, 3, 14], dtype=dtype)
-   *    rhs = tf.constant([5, 0, 7, 11], dtype=dtype)
-   *    exp = tf.constant([5, 5, 4, 5],  dtype=tf.float32)
-   *
-   *    res = bitwise_ops.bitwise_xor(lhs, rhs)
-   *    tf.assert_equal(tf.cast(res, tf.float32), exp) # TRUE
-   *  }</pre>
-   *
-   * @param <T> data type for {@code z()} output
-   * @param x
-   * @param y
-   * @return a new instance of BitwiseXor
-   */
-  public <T extends TNumber> BitwiseXor<T> bitwiseXor(Operand<T> x, Operand<T> y) {
-    return BitwiseXor.create(scope, x, y);
-  }
-
-  /**
-   * Invert (flip) each bit of supported types; for example, type `uint8` value 01010101 becomes 10101010.
-   *  <p>
-   *  Flip each bit of supported types.  For example, type `int8` (decimal 2) binary 00000010 becomes (decimal -3) binary 11111101.
-   *  This operation is performed on each element of the tensor argument `x`.
-   *  <p>
-   *  Example:
-   *  <pre>{@code
-   *  import tensorflow as tf
-   *  from tensorflow.python.ops import bitwise_ops
-   *
-   *  # flip 2 (00000010) to -3 (11111101)
-   *  tf.assert_equal(-3, bitwise_ops.invert(2))
-   *
-   *  dtype_list = [dtypes.int8, dtypes.int16, dtypes.int32, dtypes.int64,
-   *                dtypes.uint8, dtypes.uint16, dtypes.uint32, dtypes.uint64]
-   *
-   *  inputs = [0, 5, 3, 14]
-   *  for dtype in dtype_list:
-   *    # Because of issues with negative numbers, let's test this indirectly.
-   *    # 1. invert(a) and a = 0
-   *    # 2. invert(a) or a = invert(0)
-   *    input_tensor = tf.constant([0, 5, 3, 14], dtype=dtype)
-   *    not_a_and_a, not_a_or_a, not_0 = [bitwise_ops.bitwise_and(
-   *                                        input_tensor, bitwise_ops.invert(input_tensor)),
-   *                                      bitwise_ops.bitwise_or(
-   *                                        input_tensor, bitwise_ops.invert(input_tensor)),
-   *                                      bitwise_ops.invert(
-   *                                        tf.constant(0, dtype=dtype))]
-   *
-   *    expected = tf.constant([0, 0, 0, 0], dtype=tf.float32)
-   *    tf.assert_equal(tf.cast(not_a_and_a, tf.float32), expected)
-   *
-   *    expected = tf.cast([not_0] * 4, tf.float32)
-   *    tf.assert_equal(tf.cast(not_a_or_a, tf.float32), expected)
-   *
-   *    # For unsigned dtypes let's also check the result directly.
-   *    if dtype.is_unsigned:
-   *      inverted = bitwise_ops.invert(input_tensor)
-   *      expected = tf.constant([dtype.max - x for x in inputs], dtype=tf.float32)
-   *      tf.assert_equal(tf.cast(inverted, tf.float32), tf.cast(expected, tf.float32))
-   *  }</pre>
-   *
-   * @param <T> data type for {@code y()} output
-   * @param x
-   * @return a new instance of Invert
-   */
-  public <T extends TNumber> Invert<T> invert(Operand<T> x) {
-    return Invert.create(scope, x);
-  }
-
-  /**
-   * Elementwise computes the bitwise left-shift of `x` and `y`.
-   *  <p>
-   *  If `y` is negative, or greater than or equal to the width of `x` in bits the
-   *  result is implementation defined.
-   *  <p>
-   *  Example:
-   *  <pre>{@code
-   *  import tensorflow as tf
-   *  from tensorflow.python.ops import bitwise_ops
-   *  import numpy as np
-   *  dtype_list = [tf.int8, tf.int16, tf.int32, tf.int64]
-   *
-   *  for dtype in dtype_list:
-   *    lhs = tf.constant([-1, -5, -3, -14], dtype=dtype)
-   *    rhs = tf.constant([5, 0, 7, 11], dtype=dtype)
-   *
-   *    left_shift_result = bitwise_ops.left_shift(lhs, rhs)
-   *
-   *    print(left_shift_result)
-   *
-   *  # This will print:
-   *  # tf.Tensor([ -32   -5 -128    0], shape=(4,), dtype=int8)
-   *  # tf.Tensor([   -32     -5   -384 -28672], shape=(4,), dtype=int16)
-   *  # tf.Tensor([   -32     -5   -384 -28672], shape=(4,), dtype=int32)
-   *  # tf.Tensor([   -32     -5   -384 -28672], shape=(4,), dtype=int64)
-   *
-   *  lhs = np.array([-2, 64, 101, 32], dtype=np.int8)
-   *  rhs = np.array([-1, -5, -3, -14], dtype=np.int8)
-   *  bitwise_ops.left_shift(lhs, rhs)
-   *  # <tf.Tensor: shape=(4,), dtype=int8, numpy=array([ -2,  64, 101,  32], dtype=int8)>
-   *  }</pre>
-   *
-   * @param <T> data type for {@code z()} output
-   * @param x
-   * @param y
-   * @return a new instance of LeftShift
-   */
-  public <T extends TNumber> LeftShift<T> leftShift(Operand<T> x, Operand<T> y) {
-    return LeftShift.create(scope, x, y);
-  }
-
-  /**
-   * Elementwise computes the bitwise right-shift of `x` and `y`.
-   *  <p>
-   *  Performs a logical shift for unsigned integer types, and an arithmetic shift
-   *  for signed integer types.
-   *  <p>
-   *  If `y` is negative, or greater than or equal to than the width of `x` in bits
-   *  the result is implementation defined.
-   *  <p>
-   *  Example:
-   *  <pre>{@code
-   *  import tensorflow as tf
-   *  from tensorflow.python.ops import bitwise_ops
-   *  import numpy as np
-   *  dtype_list = [tf.int8, tf.int16, tf.int32, tf.int64]
-   *
-   *  for dtype in dtype_list:
-   *    lhs = tf.constant([-1, -5, -3, -14], dtype=dtype)
-   *    rhs = tf.constant([5, 0, 7, 11], dtype=dtype)
-   *
-   *    right_shift_result = bitwise_ops.right_shift(lhs, rhs)
-   *
-   *    print(right_shift_result)
-   *
-   *  # This will print:
-   *  # tf.Tensor([-1 -5 -1 -1], shape=(4,), dtype=int8)
-   *  # tf.Tensor([-1 -5 -1 -1], shape=(4,), dtype=int16)
-   *  # tf.Tensor([-1 -5 -1 -1], shape=(4,), dtype=int32)
-   *  # tf.Tensor([-1 -5 -1 -1], shape=(4,), dtype=int64)
-   *
-   *  lhs = np.array([-2, 64, 101, 32], dtype=np.int8)
-   *  rhs = np.array([-1, -5, -3, -14], dtype=np.int8)
-   *  bitwise_ops.right_shift(lhs, rhs)
-   *  # <tf.Tensor: shape=(4,), dtype=int8, numpy=array([ -2,  64, 101,  32], dtype=int8)>
-   *  }</pre>
-   *
-   * @param <T> data type for {@code z()} output
-   * @param x
-   * @param y
-   * @return a new instance of RightShift
-   */
-  public <T extends TNumber> RightShift<T> rightShift(Operand<T> x, Operand<T> y) {
-    return RightShift.create(scope, x, y);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/DataExperimentalOps.java b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/DataExperimentalOps.java
deleted file mode 100644
index cccc4ac8dcb..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/DataExperimentalOps.java
+++ /dev/null
@@ -1,61 +0,0 @@
-// Copyright 2020 The TensorFlow Authors. All Rights Reserved.
-//
-// Licensed under the Apache License, Version 2.0 (the "License");
-// you may not use this file except in compliance with the License.
-// You may obtain a copy of the License at
-//
-//     http://www.apache.org/licenses/LICENSE-2.0
-//
-// Unless required by applicable law or agreed to in writing, software
-// distributed under the License is distributed on an "AS IS" BASIS,
-// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-// See the License for the specific language governing permissions and
-// limitations under the License.
-// ==============================================================================
-//
-// This class has been generated, DO NOT EDIT!
-//
-package org.tensorflow.op;
-
-import java.util.List;
-import org.tensorflow.DataType;
-import org.tensorflow.Operand;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.op.data.experimental.DataServiceDataset;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.TString;
-
-/**
- * An API for building {@code data.experimental} operations as {@link Op Op}s
- *
- * @see {@link Ops}
- */
-public final class DataExperimentalOps {
-  private final Scope scope;
-
-  DataExperimentalOps(Scope scope) {
-    this.scope = scope;
-  }
-
-  /**
-   *
-   * @param datasetId
-   * @param processingMode
-   * @param address
-   * @param protocol
-   * @param jobName
-   * @param maxOutstandingRequests
-   * @param iterationCounter
-   * @param outputTypes
-   * @param outputShapes
-   * @param options carries optional attributes values
-   * @return a new instance of DataServiceDataset
-   */
-  public DataServiceDataset dataServiceDataset(Operand<TInt64> datasetId,
-      Operand<TString> processingMode, Operand<TString> address, Operand<TString> protocol,
-      Operand<TString> jobName, Operand<TInt64> maxOutstandingRequests, Operand<?> iterationCounter,
-      List<DataType<?>> outputTypes, List<Shape> outputShapes,
-      DataServiceDataset.Options... options) {
-    return DataServiceDataset.create(scope, datasetId, processingMode, address, protocol, jobName, maxOutstandingRequests, iterationCounter, outputTypes, outputShapes, options);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/DataOps.java b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/DataOps.java
deleted file mode 100644
index 273025ef6bd..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/DataOps.java
+++ /dev/null
@@ -1,413 +0,0 @@
-// Copyright 2020 The TensorFlow Authors. All Rights Reserved.
-//
-// Licensed under the Apache License, Version 2.0 (the "License");
-// you may not use this file except in compliance with the License.
-// You may obtain a copy of the License at
-//
-//     http://www.apache.org/licenses/LICENSE-2.0
-//
-// Unless required by applicable law or agreed to in writing, software
-// distributed under the License is distributed on an "AS IS" BASIS,
-// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-// See the License for the specific language governing permissions and
-// limitations under the License.
-// ==============================================================================
-//
-// This class has been generated, DO NOT EDIT!
-//
-package org.tensorflow.op;
-
-import java.util.List;
-import org.tensorflow.DataType;
-import org.tensorflow.Operand;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.op.data.AnonymousIterator;
-import org.tensorflow.op.data.BatchDataset;
-import org.tensorflow.op.data.CSVDataset;
-import org.tensorflow.op.data.ConcatenateDataset;
-import org.tensorflow.op.data.DeleteIterator;
-import org.tensorflow.op.data.DeserializeIterator;
-import org.tensorflow.op.data.Iterator;
-import org.tensorflow.op.data.IteratorGetNext;
-import org.tensorflow.op.data.IteratorGetNextAsOptional;
-import org.tensorflow.op.data.IteratorGetNextSync;
-import org.tensorflow.op.data.IteratorToStringHandle;
-import org.tensorflow.op.data.MakeIterator;
-import org.tensorflow.op.data.OptionalFromValue;
-import org.tensorflow.op.data.OptionalGetValue;
-import org.tensorflow.op.data.OptionalHasValue;
-import org.tensorflow.op.data.OptionalNone;
-import org.tensorflow.op.data.RangeDataset;
-import org.tensorflow.op.data.RepeatDataset;
-import org.tensorflow.op.data.SerializeIterator;
-import org.tensorflow.op.data.SkipDataset;
-import org.tensorflow.op.data.TakeDataset;
-import org.tensorflow.op.data.TensorSliceDataset;
-import org.tensorflow.op.data.TextLineDataset;
-import org.tensorflow.op.data.TfRecordDataset;
-import org.tensorflow.op.data.ZipDataset;
-import org.tensorflow.types.TBool;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.TString;
-
-/**
- * An API for building {@code data} operations as {@link Op Op}s
- *
- * @see {@link Ops}
- */
-public final class DataOps {
-  public final DataExperimentalOps experimental;
-
-  private final Scope scope;
-
-  DataOps(Scope scope) {
-    this.scope = scope;
-    experimental = new DataExperimentalOps(scope);
-  }
-
-  /**
-   * A container for an iterator resource.
-   *
-   * @param outputTypes
-   * @param outputShapes
-   * @return a new instance of AnonymousIterator
-   */
-  public AnonymousIterator anonymousIterator(List<DataType<?>> outputTypes,
-      List<Shape> outputShapes) {
-    return AnonymousIterator.create(scope, outputTypes, outputShapes);
-  }
-
-  /**
-   * Creates a dataset that batches `batch_size` elements from `input_dataset`.
-   *
-   * @param inputDataset
-   * @param batchSize A scalar representing the number of elements to accumulate in a batch.
-   * @param dropRemainder A scalar representing whether the last batch should be dropped in case its size
-   *  is smaller than desired.
-   * @param outputTypes
-   * @param outputShapes
-   * @param options carries optional attributes values
-   * @return a new instance of BatchDataset
-   */
-  public BatchDataset batchDataset(Operand<?> inputDataset, Operand<TInt64> batchSize,
-      Operand<TBool> dropRemainder, List<DataType<?>> outputTypes, List<Shape> outputShapes,
-      BatchDataset.Options... options) {
-    return BatchDataset.create(scope, inputDataset, batchSize, dropRemainder, outputTypes, outputShapes, options);
-  }
-
-  /**
-   *
-   * @param filenames
-   * @param compressionType
-   * @param bufferSize
-   * @param header
-   * @param fieldDelim
-   * @param useQuoteDelim
-   * @param naValue
-   * @param selectCols
-   * @param recordDefaults
-   * @param outputShapes
-   * @return a new instance of CSVDataset
-   */
-  public CSVDataset cSVDataset(Operand<TString> filenames, Operand<TString> compressionType,
-      Operand<TInt64> bufferSize, Operand<TBool> header, Operand<TString> fieldDelim,
-      Operand<TBool> useQuoteDelim, Operand<TString> naValue, Operand<TInt64> selectCols,
-      Iterable<Operand<?>> recordDefaults, List<Shape> outputShapes) {
-    return CSVDataset.create(scope, filenames, compressionType, bufferSize, header, fieldDelim, useQuoteDelim, naValue, selectCols, recordDefaults, outputShapes);
-  }
-
-  /**
-   * Creates a dataset that concatenates `input_dataset` with `another_dataset`.
-   *
-   * @param inputDataset
-   * @param anotherDataset
-   * @param outputTypes
-   * @param outputShapes
-   * @return a new instance of ConcatenateDataset
-   */
-  public ConcatenateDataset concatenateDataset(Operand<?> inputDataset, Operand<?> anotherDataset,
-      List<DataType<?>> outputTypes, List<Shape> outputShapes) {
-    return ConcatenateDataset.create(scope, inputDataset, anotherDataset, outputTypes, outputShapes);
-  }
-
-  /**
-   * A container for an iterator resource.
-   *
-   * @param handle A handle to the iterator to delete.
-   * @param deleter A variant deleter.
-   * @return a new instance of DeleteIterator
-   */
-  public DeleteIterator deleteIterator(Operand<?> handle, Operand<?> deleter) {
-    return DeleteIterator.create(scope, handle, deleter);
-  }
-
-  /**
-   * Converts the given variant tensor to an iterator and stores it in the given resource.
-   *
-   * @param resourceHandle A handle to an iterator resource.
-   * @param serialized A variant tensor storing the state of the iterator contained in the
-   *  resource.
-   * @return a new instance of DeserializeIterator
-   */
-  public DeserializeIterator deserializeIterator(Operand<?> resourceHandle, Operand<?> serialized) {
-    return DeserializeIterator.create(scope, resourceHandle, serialized);
-  }
-
-  /**
-   *
-   * @param sharedName
-   * @param container
-   * @param outputTypes
-   * @param outputShapes
-   * @return a new instance of Iterator
-   */
-  public Iterator iterator(String sharedName, String container, List<DataType<?>> outputTypes,
-      List<Shape> outputShapes) {
-    return Iterator.create(scope, sharedName, container, outputTypes, outputShapes);
-  }
-
-  /**
-   * Gets the next output from the given iterator .
-   *
-   * @param iterator
-   * @param outputTypes
-   * @param outputShapes
-   * @return a new instance of IteratorGetNext
-   */
-  public IteratorGetNext iteratorGetNext(Operand<?> iterator, List<DataType<?>> outputTypes,
-      List<Shape> outputShapes) {
-    return IteratorGetNext.create(scope, iterator, outputTypes, outputShapes);
-  }
-
-  /**
-   * Gets the next output from the given iterator as an Optional variant.
-   *
-   * @param iterator
-   * @param outputTypes
-   * @param outputShapes
-   * @return a new instance of IteratorGetNextAsOptional
-   */
-  public IteratorGetNextAsOptional iteratorGetNextAsOptional(Operand<?> iterator,
-      List<DataType<?>> outputTypes, List<Shape> outputShapes) {
-    return IteratorGetNextAsOptional.create(scope, iterator, outputTypes, outputShapes);
-  }
-
-  /**
-   * Gets the next output from the given iterator.
-   *  <p>
-   *  This operation is a synchronous version IteratorGetNext. It should only be used
-   *  in situations where the iterator does not block the calling thread, or where
-   *  the calling thread is not a member of the thread pool used to execute parallel
-   *  operations (e.g. in eager mode).
-   *
-   * @param iterator
-   * @param outputTypes
-   * @param outputShapes
-   * @return a new instance of IteratorGetNextSync
-   */
-  public IteratorGetNextSync iteratorGetNextSync(Operand<?> iterator, List<DataType<?>> outputTypes,
-      List<Shape> outputShapes) {
-    return IteratorGetNextSync.create(scope, iterator, outputTypes, outputShapes);
-  }
-
-  /**
-   * Converts the given `resource_handle` representing an iterator to a string.
-   *
-   * @param resourceHandle A handle to an iterator resource.
-   * @return a new instance of IteratorToStringHandle
-   */
-  public IteratorToStringHandle iteratorToStringHandle(Operand<?> resourceHandle) {
-    return IteratorToStringHandle.create(scope, resourceHandle);
-  }
-
-  /**
-   * Makes a new iterator from the given `dataset` and stores it in `iterator`.
-   *  <p>
-   *  This operation may be executed multiple times. Each execution will reset the
-   *  iterator in `iterator` to the first element of `dataset`.
-   *
-   * @param dataset
-   * @param iterator
-   * @return a new instance of MakeIterator
-   */
-  public MakeIterator makeIterator(Operand<?> dataset, Operand<?> iterator) {
-    return MakeIterator.create(scope, dataset, iterator);
-  }
-
-  /**
-   * Constructs an Optional variant from a tuple of tensors.
-   *
-   * @param components
-   * @return a new instance of OptionalFromValue
-   */
-  public OptionalFromValue optionalFromValue(Iterable<Operand<?>> components) {
-    return OptionalFromValue.create(scope, components);
-  }
-
-  /**
-   * Returns the value stored in an Optional variant or raises an error if none exists.
-   *
-   * @param optional
-   * @param outputTypes
-   * @param outputShapes
-   * @return a new instance of OptionalGetValue
-   */
-  public OptionalGetValue optionalGetValue(Operand<?> optional, List<DataType<?>> outputTypes,
-      List<Shape> outputShapes) {
-    return OptionalGetValue.create(scope, optional, outputTypes, outputShapes);
-  }
-
-  /**
-   * Returns true if and only if the given Optional variant has a value.
-   *
-   * @param optional
-   * @return a new instance of OptionalHasValue
-   */
-  public OptionalHasValue optionalHasValue(Operand<?> optional) {
-    return OptionalHasValue.create(scope, optional);
-  }
-
-  /**
-   * Creates an Optional variant with no value.
-   *
-   * @return a new instance of OptionalNone
-   */
-  public OptionalNone optionalNone() {
-    return OptionalNone.create(scope);
-  }
-
-  /**
-   * Creates a dataset with a range of values. Corresponds to python's xrange.
-   *
-   * @param start corresponds to start in python's xrange().
-   * @param stop corresponds to stop in python's xrange().
-   * @param step corresponds to step in python's xrange().
-   * @param outputTypes
-   * @param outputShapes
-   * @return a new instance of RangeDataset
-   */
-  public RangeDataset rangeDataset(Operand<TInt64> start, Operand<TInt64> stop,
-      Operand<TInt64> step, List<DataType<?>> outputTypes, List<Shape> outputShapes) {
-    return RangeDataset.create(scope, start, stop, step, outputTypes, outputShapes);
-  }
-
-  /**
-   * Creates a dataset that emits the outputs of `input_dataset` `count` times.
-   *
-   * @param inputDataset
-   * @param count A scalar representing the number of times that `input_dataset` should
-   *  be repeated. A value of `-1` indicates that it should be repeated infinitely.
-   * @param outputTypes
-   * @param outputShapes
-   * @return a new instance of RepeatDataset
-   */
-  public RepeatDataset repeatDataset(Operand<?> inputDataset, Operand<TInt64> count,
-      List<DataType<?>> outputTypes, List<Shape> outputShapes) {
-    return RepeatDataset.create(scope, inputDataset, count, outputTypes, outputShapes);
-  }
-
-  /**
-   * Converts the given `resource_handle` representing an iterator to a variant tensor.
-   *
-   * @param resourceHandle A handle to an iterator resource.
-   * @param options carries optional attributes values
-   * @return a new instance of SerializeIterator
-   */
-  public SerializeIterator serializeIterator(Operand<?> resourceHandle,
-      SerializeIterator.Options... options) {
-    return SerializeIterator.create(scope, resourceHandle, options);
-  }
-
-  /**
-   * Creates a dataset that skips `count` elements from the `input_dataset`.
-   *
-   * @param inputDataset
-   * @param count A scalar representing the number of elements from the `input_dataset`
-   *  that should be skipped.  If count is -1, skips everything.
-   * @param outputTypes
-   * @param outputShapes
-   * @return a new instance of SkipDataset
-   */
-  public SkipDataset skipDataset(Operand<?> inputDataset, Operand<TInt64> count,
-      List<DataType<?>> outputTypes, List<Shape> outputShapes) {
-    return SkipDataset.create(scope, inputDataset, count, outputTypes, outputShapes);
-  }
-
-  /**
-   * Creates a dataset that contains `count` elements from the `input_dataset`.
-   *
-   * @param inputDataset
-   * @param count A scalar representing the number of elements from the `input_dataset`
-   *  that should be taken. A value of `-1` indicates that all of `input_dataset`
-   *  is taken.
-   * @param outputTypes
-   * @param outputShapes
-   * @return a new instance of TakeDataset
-   */
-  public TakeDataset takeDataset(Operand<?> inputDataset, Operand<TInt64> count,
-      List<DataType<?>> outputTypes, List<Shape> outputShapes) {
-    return TakeDataset.create(scope, inputDataset, count, outputTypes, outputShapes);
-  }
-
-  /**
-   * Creates a dataset that emits each dim-0 slice of `components` once.
-   *
-   * @param components
-   * @param outputShapes
-   * @return a new instance of TensorSliceDataset
-   */
-  public TensorSliceDataset tensorSliceDataset(Iterable<Operand<?>> components,
-      List<Shape> outputShapes) {
-    return TensorSliceDataset.create(scope, components, outputShapes);
-  }
-
-  /**
-   * Creates a dataset that emits the lines of one or more text files.
-   *
-   * @param filenames A scalar or a vector containing the name(s) of the file(s) to be
-   *  read.
-   * @param compressionType A scalar containing either (i) the empty string (no
-   *  compression), (ii) "ZLIB", or (iii) "GZIP".
-   * @param bufferSize A scalar containing the number of bytes to buffer.
-   * @return a new instance of TextLineDataset
-   */
-  public TextLineDataset textLineDataset(Operand<TString> filenames,
-      Operand<TString> compressionType, Operand<TInt64> bufferSize) {
-    return TextLineDataset.create(scope, filenames, compressionType, bufferSize);
-  }
-
-  /**
-   * Creates a dataset that emits the records from one or more TFRecord files.
-   *
-   * @param filenames A scalar or vector containing the name(s) of the file(s) to be
-   *  read.
-   * @param compressionType A scalar containing either (i) the empty string (no
-   *  compression), (ii) "ZLIB", or (iii) "GZIP".
-   * @param bufferSize A scalar representing the number of bytes to buffer. A value of
-   *  0 means no buffering will be performed.
-   * @return a new instance of TfRecordDataset
-   */
-  public TfRecordDataset tfRecordDataset(Operand<TString> filenames,
-      Operand<TString> compressionType, Operand<TInt64> bufferSize) {
-    return TfRecordDataset.create(scope, filenames, compressionType, bufferSize);
-  }
-
-  /**
-   * Creates a dataset that zips together `input_datasets`.
-   *  <p>
-   *  The elements of the resulting dataset are created by zipping corresponding
-   *  elements from each of the input datasets.
-   *  <p>
-   *  The size of the resulting dataset will match the size of the smallest input
-   *  dataset, and no error will be raised if input datasets have different sizes.
-   *
-   * @param inputDatasets List of `N` variant Tensors representing datasets to be zipped together.
-   * @param outputTypes
-   * @param outputShapes
-   * @return a new instance of ZipDataset
-   */
-  public ZipDataset zipDataset(Iterable<Operand<?>> inputDatasets, List<DataType<?>> outputTypes,
-      List<Shape> outputShapes) {
-    return ZipDataset.create(scope, inputDatasets, outputTypes, outputShapes);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/DebuggingOps.java b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/DebuggingOps.java
deleted file mode 100644
index f12d18f925b..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/DebuggingOps.java
+++ /dev/null
@@ -1,50 +0,0 @@
-// Copyright 2020 The TensorFlow Authors. All Rights Reserved.
-//
-// Licensed under the Apache License, Version 2.0 (the "License");
-// you may not use this file except in compliance with the License.
-// You may obtain a copy of the License at
-//
-//     http://www.apache.org/licenses/LICENSE-2.0
-//
-// Unless required by applicable law or agreed to in writing, software
-// distributed under the License is distributed on an "AS IS" BASIS,
-// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-// See the License for the specific language governing permissions and
-// limitations under the License.
-// ==============================================================================
-//
-// This class has been generated, DO NOT EDIT!
-//
-package org.tensorflow.op;
-
-import org.tensorflow.Operand;
-import org.tensorflow.op.debugging.CheckNumerics;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * An API for building {@code debugging} operations as {@link Op Op}s
- *
- * @see {@link Ops}
- */
-public final class DebuggingOps {
-  private final Scope scope;
-
-  DebuggingOps(Scope scope) {
-    this.scope = scope;
-  }
-
-  /**
-   * Checks a tensor for NaN and Inf values.
-   *  <p>
-   *  When run, reports an `InvalidArgument` error if `tensor` has any values
-   *  that are not a number (NaN) or infinity (Inf). Otherwise, passes `tensor` as-is.
-   *
-   * @param <T> data type for {@code output()} output
-   * @param tensor
-   * @param message Prefix of the error message.
-   * @return a new instance of CheckNumerics
-   */
-  public <T extends TNumber> CheckNumerics<T> checkNumerics(Operand<T> tensor, String message) {
-    return CheckNumerics.create(scope, tensor, message);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/DtypesOps.java b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/DtypesOps.java
deleted file mode 100644
index 16d571a6428..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/DtypesOps.java
+++ /dev/null
@@ -1,105 +0,0 @@
-// Copyright 2020 The TensorFlow Authors. All Rights Reserved.
-//
-// Licensed under the Apache License, Version 2.0 (the "License");
-// you may not use this file except in compliance with the License.
-// You may obtain a copy of the License at
-//
-//     http://www.apache.org/licenses/LICENSE-2.0
-//
-// Unless required by applicable law or agreed to in writing, software
-// distributed under the License is distributed on an "AS IS" BASIS,
-// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-// See the License for the specific language governing permissions and
-// limitations under the License.
-// ==============================================================================
-//
-// This class has been generated, DO NOT EDIT!
-//
-package org.tensorflow.op;
-
-import org.tensorflow.DataType;
-import org.tensorflow.Operand;
-import org.tensorflow.op.dtypes.AsString;
-import org.tensorflow.op.dtypes.Cast;
-import org.tensorflow.op.dtypes.Complex;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * An API for building {@code dtypes} operations as {@link Op Op}s
- *
- * @see {@link Ops}
- */
-public final class DtypesOps {
-  private final Scope scope;
-
-  DtypesOps(Scope scope) {
-    this.scope = scope;
-  }
-
-  /**
-   * Converts each entry in the given tensor to strings.
-   *  <p>
-   *  Supports many numeric types and boolean.
-   *  <p>
-   *  For Unicode, see the
-   *  [https://www.tensorflow.org/tutorials/representation/unicode](Working with Unicode text)
-   *  tutorial.
-   *  <p>
-   *  Examples:
-   *  <p>
-   *  >>> tf.strings.as_string([3, 2])
-   *  <tf.Tensor: shape=(2,), dtype=string, numpy=array([b'3', b'2'], dtype=object)>
-   *  >>> tf.strings.as_string([3.1415926, 2.71828], precision=2).numpy()
-   *  array([b'3.14', b'2.72'], dtype=object)
-   *
-   * @param input
-   * @param options carries optional attributes values
-   * @return a new instance of AsString
-   */
-  public <T extends TType> AsString asString(Operand<T> input, AsString.Options... options) {
-    return AsString.create(scope, input, options);
-  }
-
-  /**
-   * Cast x of type SrcT to y of DstT.
-   *
-   * @param <U> data type for {@code y()} output
-   * @param x
-   * @param DstT
-   * @param options carries optional attributes values
-   * @return a new instance of Cast
-   */
-  public <U extends TType, T extends TType> Cast<U> cast(Operand<T> x, DataType<U> DstT,
-      Cast.Options... options) {
-    return Cast.create(scope, x, DstT, options);
-  }
-
-  /**
-   * Converts two real numbers to a complex number.
-   *  <p>
-   *  Given a tensor `real` representing the real part of a complex number, and a
-   *  tensor `imag` representing the imaginary part of a complex number, this
-   *  operation returns complex numbers elementwise of the form \\(a + bj\\), where
-   *  <i>a</i> represents the `real` part and <i>b</i> represents the `imag` part.
-   *  <p>
-   *  The input tensors `real` and `imag` must have the same shape.
-   *  <p>
-   *  For example:
-   *  <pre>{@code
-   *  # tensor 'real' is [2.25, 3.25]
-   *  # tensor `imag` is [4.75, 5.75]
-   *  tf.complex(real, imag) ==> [[2.25 + 4.75j], [3.25 + 5.75j]]
-   *  }</pre>
-   *
-   * @param <U> data type for {@code out()} output
-   * @param real
-   * @param imag
-   * @param Tout
-   * @return a new instance of Complex
-   */
-  public <U extends TType, T extends TNumber> Complex<U> complex(Operand<T> real, Operand<T> imag,
-      DataType<U> Tout) {
-    return Complex.create(scope, real, imag, Tout);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/ImageOps.java b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/ImageOps.java
deleted file mode 100644
index eea2fc4b8f1..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/ImageOps.java
+++ /dev/null
@@ -1,945 +0,0 @@
-// Copyright 2020 The TensorFlow Authors. All Rights Reserved.
-//
-// Licensed under the Apache License, Version 2.0 (the "License");
-// you may not use this file except in compliance with the License.
-// You may obtain a copy of the License at
-//
-//     http://www.apache.org/licenses/LICENSE-2.0
-//
-// Unless required by applicable law or agreed to in writing, software
-// distributed under the License is distributed on an "AS IS" BASIS,
-// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-// See the License for the specific language governing permissions and
-// limitations under the License.
-// ==============================================================================
-//
-// This class has been generated, DO NOT EDIT!
-//
-package org.tensorflow.op;
-
-import java.util.List;
-import org.tensorflow.DataType;
-import org.tensorflow.Operand;
-import org.tensorflow.op.image.AdjustContrast;
-import org.tensorflow.op.image.AdjustHue;
-import org.tensorflow.op.image.AdjustSaturation;
-import org.tensorflow.op.image.CombinedNonMaxSuppression;
-import org.tensorflow.op.image.CropAndResize;
-import org.tensorflow.op.image.CropAndResizeGradBoxes;
-import org.tensorflow.op.image.CropAndResizeGradImage;
-import org.tensorflow.op.image.DecodeAndCropJpeg;
-import org.tensorflow.op.image.DecodeBmp;
-import org.tensorflow.op.image.DecodeGif;
-import org.tensorflow.op.image.DecodeJpeg;
-import org.tensorflow.op.image.DecodePng;
-import org.tensorflow.op.image.DrawBoundingBoxes;
-import org.tensorflow.op.image.EncodeJpeg;
-import org.tensorflow.op.image.EncodeJpegVariableQuality;
-import org.tensorflow.op.image.EncodePng;
-import org.tensorflow.op.image.ExtractImagePatches;
-import org.tensorflow.op.image.ExtractJpegShape;
-import org.tensorflow.op.image.HsvToRgb;
-import org.tensorflow.op.image.NonMaxSuppression;
-import org.tensorflow.op.image.NonMaxSuppressionWithOverlaps;
-import org.tensorflow.op.image.QuantizedResizeBilinear;
-import org.tensorflow.op.image.RandomCrop;
-import org.tensorflow.op.image.ResizeArea;
-import org.tensorflow.op.image.ResizeBicubic;
-import org.tensorflow.op.image.ResizeBilinear;
-import org.tensorflow.op.image.ResizeNearestNeighbor;
-import org.tensorflow.op.image.RgbToHsv;
-import org.tensorflow.op.image.SampleDistortedBoundingBox;
-import org.tensorflow.op.image.ScaleAndTranslate;
-import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.TString;
-import org.tensorflow.types.TUint8;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * An API for building {@code image} operations as {@link Op Op}s
- *
- * @see {@link Ops}
- */
-public final class ImageOps {
-  private final Scope scope;
-
-  ImageOps(Scope scope) {
-    this.scope = scope;
-  }
-
-  /**
-   * Adjust the contrast of one or more images.
-   *  <p>
-   *  `images` is a tensor of at least 3 dimensions.  The last 3 dimensions are
-   *  interpreted as `[height, width, channels]`.  The other dimensions only
-   *  represent a collection of images, such as `[batch, height, width, channels].`
-   *  <p>
-   *  Contrast is adjusted independently for each channel of each image.
-   *  <p>
-   *  For each channel, the Op first computes the mean of the image pixels in the
-   *  channel and then adjusts each component of each pixel to
-   *  `(x - mean) * contrast_factor + mean`.
-   *
-   * @param <T> data type for {@code output()} output
-   * @param images Images to adjust.  At least 3-D.
-   * @param contrastFactor A float multiplier for adjusting contrast.
-   * @return a new instance of AdjustContrast
-   */
-  public <T extends TNumber> AdjustContrast<T> adjustContrast(Operand<T> images,
-      Operand<TFloat32> contrastFactor) {
-    return AdjustContrast.create(scope, images, contrastFactor);
-  }
-
-  /**
-   * Adjust the hue of one or more images.
-   *  <p>
-   *  `images` is a tensor of at least 3 dimensions.  The last dimension is
-   *  interpreted as channels, and must be three.
-   *  <p>
-   *  The input image is considered in the RGB colorspace. Conceptually, the RGB
-   *  colors are first mapped into HSV. A delta is then applied all the hue values,
-   *  and then remapped back to RGB colorspace.
-   *
-   * @param <T> data type for {@code output()} output
-   * @param images Images to adjust.  At least 3-D.
-   * @param delta A float delta to add to the hue.
-   * @return a new instance of AdjustHue
-   */
-  public <T extends TNumber> AdjustHue<T> adjustHue(Operand<T> images, Operand<TFloat32> delta) {
-    return AdjustHue.create(scope, images, delta);
-  }
-
-  /**
-   * Adjust the saturation of one or more images.
-   *  <p>
-   *  `images` is a tensor of at least 3 dimensions.  The last dimension is
-   *  interpreted as channels, and must be three.
-   *  <p>
-   *  The input image is considered in the RGB colorspace. Conceptually, the RGB
-   *  colors are first mapped into HSV. A scale is then applied all the saturation
-   *  values, and then remapped back to RGB colorspace.
-   *
-   * @param <T> data type for {@code output()} output
-   * @param images Images to adjust.  At least 3-D.
-   * @param scale A float scale to add to the saturation.
-   * @return a new instance of AdjustSaturation
-   */
-  public <T extends TNumber> AdjustSaturation<T> adjustSaturation(Operand<T> images,
-      Operand<TFloat32> scale) {
-    return AdjustSaturation.create(scope, images, scale);
-  }
-
-  /**
-   * Greedily selects a subset of bounding boxes in descending order of score,
-   *  <p>
-   *  This operation performs non_max_suppression on the inputs per batch, across
-   *  all classes.
-   *  Prunes away boxes that have high intersection-over-union (IOU) overlap
-   *  with previously selected boxes.  Bounding boxes are supplied as
-   *  [y1, x1, y2, x2], where (y1, x1) and (y2, x2) are the coordinates of any
-   *  diagonal pair of box corners and the coordinates can be provided as normalized
-   *  (i.e., lying in the interval [0, 1]) or absolute.  Note that this algorithm
-   *  is agnostic to where the origin is in the coordinate system. Also note that
-   *  this algorithm is invariant to orthogonal transformations and translations
-   *  of the coordinate system; thus translating or reflections of the coordinate
-   *  system result in the same boxes being selected by the algorithm.
-   *  The output of this operation is the final boxes, scores and classes tensor
-   *  returned after performing non_max_suppression.
-   *
-   * @param boxes A 4-D float tensor of shape `[batch_size, num_boxes, q, 4]`. If `q` is 1 then
-   *  same boxes are used for all classes otherwise, if `q` is equal to number of
-   *  classes, class-specific boxes are used.
-   * @param scores A 3-D float tensor of shape `[batch_size, num_boxes, num_classes]`
-   *  representing a single score corresponding to each box (each row of boxes).
-   * @param maxOutputSizePerClass A scalar integer tensor representing the maximum number of
-   *  boxes to be selected by non max suppression per class
-   * @param maxTotalSize A scalar representing maximum number of boxes retained over all classes.
-   * @param iouThreshold A 0-D float tensor representing the threshold for deciding whether
-   *  boxes overlap too much with respect to IOU.
-   * @param scoreThreshold A 0-D float tensor representing the threshold for deciding when to remove
-   *  boxes based on score.
-   * @param options carries optional attributes values
-   * @return a new instance of CombinedNonMaxSuppression
-   */
-  public CombinedNonMaxSuppression combinedNonMaxSuppression(Operand<TFloat32> boxes,
-      Operand<TFloat32> scores, Operand<TInt32> maxOutputSizePerClass, Operand<TInt32> maxTotalSize,
-      Operand<TFloat32> iouThreshold, Operand<TFloat32> scoreThreshold,
-      CombinedNonMaxSuppression.Options... options) {
-    return CombinedNonMaxSuppression.create(scope, boxes, scores, maxOutputSizePerClass, maxTotalSize, iouThreshold, scoreThreshold, options);
-  }
-
-  /**
-   * Extracts crops from the input image tensor and resizes them.
-   *  <p>
-   *  Extracts crops from the input image tensor and resizes them using bilinear
-   *  sampling or nearest neighbor sampling (possibly with aspect ratio change) to a
-   *  common output size specified by `crop_size`. This is more general than the
-   *  `crop_to_bounding_box` op which extracts a fixed size slice from the input image
-   *  and does not allow resizing or aspect ratio change.
-   *  <p>
-   *  Returns a tensor with `crops` from the input `image` at positions defined at the
-   *  bounding box locations in `boxes`. The cropped boxes are all resized (with
-   *  bilinear or nearest neighbor interpolation) to a fixed
-   *  `size = [crop_height, crop_width]`. The result is a 4-D tensor
-   *  `[num_boxes, crop_height, crop_width, depth]`. The resizing is corner aligned.
-   *  In particular, if `boxes = [[0, 0, 1, 1]]`, the method will give identical
-   *  results to using `tf.image.resize_bilinear()` or
-   *  `tf.image.resize_nearest_neighbor()`(depends on the `method` argument) with
-   *  `align_corners=True`.
-   *
-   * @param image A 4-D tensor of shape `[batch, image_height, image_width, depth]`.
-   *  Both `image_height` and `image_width` need to be positive.
-   * @param boxes A 2-D tensor of shape `[num_boxes, 4]`. The `i`-th row of the tensor
-   *  specifies the coordinates of a box in the `box_ind[i]` image and is specified
-   *  in normalized coordinates `[y1, x1, y2, x2]`. A normalized coordinate value of
-   *  `y` is mapped to the image coordinate at `y * (image_height - 1)`, so as the
-   *  `[0, 1]` interval of normalized image height is mapped to
-   *  `[0, image_height - 1]` in image height coordinates. We do allow `y1` > `y2`, in
-   *  which case the sampled crop is an up-down flipped version of the original
-   *  image. The width dimension is treated similarly. Normalized coordinates
-   *  outside the `[0, 1]` range are allowed, in which case we use
-   *  `extrapolation_value` to extrapolate the input image values.
-   * @param boxInd A 1-D tensor of shape `[num_boxes]` with int32 values in `[0, batch)`.
-   *  The value of `box_ind[i]` specifies the image that the `i`-th box refers to.
-   * @param cropSize A 1-D tensor of 2 elements, `size = [crop_height, crop_width]`. All
-   *  cropped image patches are resized to this size. The aspect ratio of the image
-   *  content is not preserved. Both `crop_height` and `crop_width` need to be
-   *  positive.
-   * @param options carries optional attributes values
-   * @return a new instance of CropAndResize
-   */
-  public <T extends TNumber> CropAndResize cropAndResize(Operand<T> image, Operand<TFloat32> boxes,
-      Operand<TInt32> boxInd, Operand<TInt32> cropSize, CropAndResize.Options... options) {
-    return CropAndResize.create(scope, image, boxes, boxInd, cropSize, options);
-  }
-
-  /**
-   * Computes the gradient of the crop_and_resize op wrt the input boxes tensor.
-   *
-   * @param grads A 4-D tensor of shape `[num_boxes, crop_height, crop_width, depth]`.
-   * @param image A 4-D tensor of shape `[batch, image_height, image_width, depth]`.
-   *  Both `image_height` and `image_width` need to be positive.
-   * @param boxes A 2-D tensor of shape `[num_boxes, 4]`. The `i`-th row of the tensor
-   *  specifies the coordinates of a box in the `box_ind[i]` image and is specified
-   *  in normalized coordinates `[y1, x1, y2, x2]`. A normalized coordinate value of
-   *  `y` is mapped to the image coordinate at `y * (image_height - 1)`, so as the
-   *  `[0, 1]` interval of normalized image height is mapped to
-   *  `[0, image_height - 1] in image height coordinates. We do allow y1 > y2, in
-   *  which case the sampled crop is an up-down flipped version of the original
-   *  image. The width dimension is treated similarly. Normalized coordinates
-   *  outside the `[0, 1]` range are allowed, in which case we use
-   *  `extrapolation_value` to extrapolate the input image values.
-   * @param boxInd A 1-D tensor of shape `[num_boxes]` with int32 values in `[0, batch)`.
-   *  The value of `box_ind[i]` specifies the image that the `i`-th box refers to.
-   * @param options carries optional attributes values
-   * @return a new instance of CropAndResizeGradBoxes
-   */
-  public <T extends TNumber> CropAndResizeGradBoxes cropAndResizeGradBoxes(Operand<TFloat32> grads,
-      Operand<T> image, Operand<TFloat32> boxes, Operand<TInt32> boxInd,
-      CropAndResizeGradBoxes.Options... options) {
-    return CropAndResizeGradBoxes.create(scope, grads, image, boxes, boxInd, options);
-  }
-
-  /**
-   * Computes the gradient of the crop_and_resize op wrt the input image tensor.
-   *
-   * @param <T> data type for {@code output()} output
-   * @param grads A 4-D tensor of shape `[num_boxes, crop_height, crop_width, depth]`.
-   * @param boxes A 2-D tensor of shape `[num_boxes, 4]`. The `i`-th row of the tensor
-   *  specifies the coordinates of a box in the `box_ind[i]` image and is specified
-   *  in normalized coordinates `[y1, x1, y2, x2]`. A normalized coordinate value of
-   *  `y` is mapped to the image coordinate at `y * (image_height - 1)`, so as the
-   *  `[0, 1]` interval of normalized image height is mapped to
-   *  `[0, image_height - 1] in image height coordinates. We do allow y1 > y2, in
-   *  which case the sampled crop is an up-down flipped version of the original
-   *  image. The width dimension is treated similarly. Normalized coordinates
-   *  outside the `[0, 1]` range are allowed, in which case we use
-   *  `extrapolation_value` to extrapolate the input image values.
-   * @param boxInd A 1-D tensor of shape `[num_boxes]` with int32 values in `[0, batch)`.
-   *  The value of `box_ind[i]` specifies the image that the `i`-th box refers to.
-   * @param imageSize A 1-D tensor with value `[batch, image_height, image_width, depth]`
-   *  containing the original image size. Both `image_height` and `image_width` need
-   *  to be positive.
-   * @param T
-   * @param options carries optional attributes values
-   * @return a new instance of CropAndResizeGradImage
-   */
-  public <T extends TNumber> CropAndResizeGradImage<T> cropAndResizeGradImage(
-      Operand<TFloat32> grads, Operand<TFloat32> boxes, Operand<TInt32> boxInd,
-      Operand<TInt32> imageSize, DataType<T> T, CropAndResizeGradImage.Options... options) {
-    return CropAndResizeGradImage.create(scope, grads, boxes, boxInd, imageSize, T, options);
-  }
-
-  /**
-   * Decode and Crop a JPEG-encoded image to a uint8 tensor.
-   *  <p>
-   *  The attr `channels` indicates the desired number of color channels for the
-   *  decoded image.
-   *  <p>
-   *  Accepted values are:
-   *  <ul>
-   *  <li>
-   *  0: Use the number of channels in the JPEG-encoded image.
-   *  </li>
-   *  <li>
-   *  1: output a grayscale image.
-   *  </li>
-   *  <li>
-   *  3: output an RGB image.
-   *  </li>
-   *  </ul>
-   *  If needed, the JPEG-encoded image is transformed to match the requested number
-   *  of color channels.
-   *  <p>
-   *  The attr `ratio` allows downscaling the image by an integer factor during
-   *  decoding.  Allowed values are: 1, 2, 4, and 8.  This is much faster than
-   *  downscaling the image later.
-   *  <p>
-   *  It is equivalent to a combination of decode and crop, but much faster by only
-   *  decoding partial jpeg image.
-   *
-   * @param contents 0-D.  The JPEG-encoded image.
-   * @param cropWindow 1-D.  The crop window: [crop_y, crop_x, crop_height, crop_width].
-   * @param options carries optional attributes values
-   * @return a new instance of DecodeAndCropJpeg
-   */
-  public DecodeAndCropJpeg decodeAndCropJpeg(Operand<TString> contents, Operand<TInt32> cropWindow,
-      DecodeAndCropJpeg.Options... options) {
-    return DecodeAndCropJpeg.create(scope, contents, cropWindow, options);
-  }
-
-  /**
-   * Decode the first frame of a BMP-encoded image to a uint8 tensor.
-   *  <p>
-   *  The attr `channels` indicates the desired number of color channels for the
-   *  decoded image.
-   *  <p>
-   *  Accepted values are:
-   *  <ul>
-   *  <li>
-   *  0: Use the number of channels in the BMP-encoded image.
-   *  </li>
-   *  <li>
-   *  3: output an RGB image.
-   *  </li>
-   *  <li>
-   *  4: output an RGBA image.
-   *
-   * @param contents 0-D.  The BMP-encoded image.
-   * @param options carries optional attributes values
-   * @return a new instance of DecodeBmp
-   */
-  public DecodeBmp decodeBmp(Operand<TString> contents, DecodeBmp.Options... options) {
-    return DecodeBmp.create(scope, contents, options);
-  }
-
-  /**
-   * Decode the frame(s) of a GIF-encoded image to a uint8 tensor.
-   *  <p>
-   *  GIF images with frame or transparency compression are not supported.
-   *  On Linux and MacOS systems, convert animated GIFs from compressed to
-   *  uncompressed by running:
-   *  <p>
-   *      convert $src.gif -coalesce $dst.gif
-   *  <p>
-   *  This op also supports decoding JPEGs and PNGs, though it is cleaner to use
-   *  `tf.io.decode_image`.
-   *
-   * @param contents 0-D.  The GIF-encoded image.
-   * @return a new instance of DecodeGif
-   */
-  public DecodeGif decodeGif(Operand<TString> contents) {
-    return DecodeGif.create(scope, contents);
-  }
-
-  /**
-   * Decode a JPEG-encoded image to a uint8 tensor.
-   *  <p>
-   *  The attr `channels` indicates the desired number of color channels for the
-   *  decoded image.
-   *  <p>
-   *  Accepted values are:
-   *  <ul>
-   *  <li>
-   *  0: Use the number of channels in the JPEG-encoded image.
-   *  </li>
-   *  <li>
-   *  1: output a grayscale image.
-   *  </li>
-   *  <li>
-   *  3: output an RGB image.
-   *  </li>
-   *  </ul>
-   *  If needed, the JPEG-encoded image is transformed to match the requested number
-   *  of color channels.
-   *  <p>
-   *  The attr `ratio` allows downscaling the image by an integer factor during
-   *  decoding.  Allowed values are: 1, 2, 4, and 8.  This is much faster than
-   *  downscaling the image later.
-   *  <p>
-   *  This op also supports decoding PNGs and non-animated GIFs since the interface is
-   *  the same, though it is cleaner to use `tf.io.decode_image`.
-   *
-   * @param contents 0-D.  The JPEG-encoded image.
-   * @param options carries optional attributes values
-   * @return a new instance of DecodeJpeg
-   */
-  public DecodeJpeg decodeJpeg(Operand<TString> contents, DecodeJpeg.Options... options) {
-    return DecodeJpeg.create(scope, contents, options);
-  }
-
-  /**
-   * Decode a PNG-encoded image to a uint8 or uint16 tensor.
-   *  <p>
-   *  The attr `channels` indicates the desired number of color channels for the
-   *  decoded image.
-   *  <p>
-   *  Accepted values are:
-   *  <ul>
-   *  <li>
-   *  0: Use the number of channels in the PNG-encoded image.
-   *  </li>
-   *  <li>
-   *  1: output a grayscale image.
-   *  </li>
-   *  <li>
-   *  3: output an RGB image.
-   *  </li>
-   *  <li>
-   *  4: output an RGBA image.
-   *  </li>
-   *  </ul>
-   *  If needed, the PNG-encoded image is transformed to match the requested number
-   *  of color channels.
-   *  <p>
-   *  This op also supports decoding JPEGs and non-animated GIFs since the interface
-   *  is the same, though it is cleaner to use `tf.io.decode_image`.
-   *
-   * @param <T> data type for {@code image()} output
-   * @param contents 0-D.  The PNG-encoded image.
-   * @param options carries optional attributes values
-   * @return a new instance of DecodePng
-   */
-  public DecodePng<TUint8> decodePng(Operand<TString> contents, DecodePng.Options... options) {
-    return DecodePng.create(scope, contents, options);
-  }
-
-  /**
-   * Decode a PNG-encoded image to a uint8 or uint16 tensor.
-   *  <p>
-   *  The attr `channels` indicates the desired number of color channels for the
-   *  decoded image.
-   *  <p>
-   *  Accepted values are:
-   *  <ul>
-   *  <li>
-   *  0: Use the number of channels in the PNG-encoded image.
-   *  </li>
-   *  <li>
-   *  1: output a grayscale image.
-   *  </li>
-   *  <li>
-   *  3: output an RGB image.
-   *  </li>
-   *  <li>
-   *  4: output an RGBA image.
-   *  </li>
-   *  </ul>
-   *  If needed, the PNG-encoded image is transformed to match the requested number
-   *  of color channels.
-   *  <p>
-   *  This op also supports decoding JPEGs and non-animated GIFs since the interface
-   *  is the same, though it is cleaner to use `tf.io.decode_image`.
-   *
-   * @param <T> data type for {@code image()} output
-   * @param contents 0-D.  The PNG-encoded image.
-   * @param dtype
-   * @param options carries optional attributes values
-   * @return a new instance of DecodePng
-   */
-  public <T extends TNumber> DecodePng<T> decodePng(Operand<TString> contents, DataType<T> dtype,
-      DecodePng.Options... options) {
-    return DecodePng.create(scope, contents, dtype, options);
-  }
-
-  /**
-   * Draw bounding boxes on a batch of images.
-   *  <p>
-   *  Outputs a copy of `images` but draws on top of the pixels zero or more bounding
-   *  boxes specified by the locations in `boxes`. The coordinates of the each
-   *  bounding box in `boxes` are encoded as `[y_min, x_min, y_max, x_max]`. The
-   *  bounding box coordinates are floats in `[0.0, 1.0]` relative to the width and
-   *  height of the underlying image.
-   *  <p>
-   *  For example, if an image is 100 x 200 pixels (height x width) and the bounding
-   *  box is `[0.1, 0.2, 0.5, 0.9]`, the upper-left and bottom-right coordinates of
-   *  the bounding box will be `(40, 10)` to `(100, 50)` (in (x,y) coordinates).
-   *  <p>
-   *  Parts of the bounding box may fall outside the image.
-   *
-   * @param <T> data type for {@code output()} output
-   * @param images 4-D with shape `[batch, height, width, depth]`. A batch of images.
-   * @param boxes 3-D with shape `[batch, num_bounding_boxes, 4]` containing bounding
-   *  boxes.
-   * @param colors 2-D. A list of RGBA colors to cycle through for the boxes.
-   * @return a new instance of DrawBoundingBoxes
-   */
-  public <T extends TNumber> DrawBoundingBoxes<T> drawBoundingBoxes(Operand<T> images,
-      Operand<TFloat32> boxes, Operand<TFloat32> colors) {
-    return DrawBoundingBoxes.create(scope, images, boxes, colors);
-  }
-
-  /**
-   * JPEG-encode an image.
-   *  <p>
-   *  `image` is a 3-D uint8 Tensor of shape `[height, width, channels]`.
-   *  <p>
-   *  The attr `format` can be used to override the color format of the encoded
-   *  output.  Values can be:
-   *  <ul>
-   *  <li>
-   *  `''`: Use a default format based on the number of channels in the image.
-   *  </li>
-   *  <li>
-   *  `grayscale`: Output a grayscale JPEG image.  The `channels` dimension
-   *      of `image` must be 1.
-   *  </li>
-   *  <li>
-   *  `rgb`: Output an RGB JPEG image. The `channels` dimension
-   *      of `image` must be 3.
-   *  </li>
-   *  </ul>
-   *  If `format` is not specified or is the empty string, a default format is picked
-   *  in function of the number of channels in `image`:
-   *  <ul>
-   *  <li>
-   *  1: Output a grayscale image.
-   *  </li>
-   *  <li>
-   *  3: Output an RGB image.
-   *
-   * @param image 3-D with shape `[height, width, channels]`.
-   * @param options carries optional attributes values
-   * @return a new instance of EncodeJpeg
-   */
-  public EncodeJpeg encodeJpeg(Operand<TUint8> image, EncodeJpeg.Options... options) {
-    return EncodeJpeg.create(scope, image, options);
-  }
-
-  /**
-   * JPEG encode input image with provided compression quality.
-   *  <p>
-   *  `image` is a 3-D uint8 Tensor of shape `[height, width, channels]`.
-   *  `quality` is an int32 jpeg compression quality value between 0 and 100.
-   *
-   * @param images Images to adjust.  At least 3-D.
-   * @param quality An int quality to encode to.
-   * @return a new instance of EncodeJpegVariableQuality
-   */
-  public EncodeJpegVariableQuality encodeJpegVariableQuality(Operand<TUint8> images,
-      Operand<TInt32> quality) {
-    return EncodeJpegVariableQuality.create(scope, images, quality);
-  }
-
-  /**
-   * PNG-encode an image.
-   *  <p>
-   *  `image` is a 3-D uint8 or uint16 Tensor of shape `[height, width, channels]`
-   *  where `channels` is:
-   *  <ul>
-   *  <li>
-   *  1: for grayscale.
-   *  </li>
-   *  <li>
-   *  2: for grayscale + alpha.
-   *  </li>
-   *  <li>
-   *  3: for RGB.
-   *  </li>
-   *  <li>
-   *  4: for RGBA.
-   *  </li>
-   *  </ul>
-   *  The ZLIB compression level, `compression`, can be -1 for the PNG-encoder
-   *  default or a value from 0 to 9.  9 is the highest compression level, generating
-   *  the smallest output, but is slower.
-   *
-   * @param image 3-D with shape `[height, width, channels]`.
-   * @param options carries optional attributes values
-   * @return a new instance of EncodePng
-   */
-  public <T extends TNumber> EncodePng encodePng(Operand<T> image, EncodePng.Options... options) {
-    return EncodePng.create(scope, image, options);
-  }
-
-  /**
-   * Extract `patches` from `images` and put them in the "depth" output dimension.
-   *
-   * @param <T> data type for {@code patches()} output
-   * @param images 4-D Tensor with shape `[batch, in_rows, in_cols, depth]`.
-   * @param ksizes The size of the sliding window for each dimension of `images`.
-   * @param strides How far the centers of two consecutive patches are in
-   *  the images. Must be: `[1, stride_rows, stride_cols, 1]`.
-   * @param rates Must be: `[1, rate_rows, rate_cols, 1]`. This is the
-   *  input stride, specifying how far two consecutive patch samples are in the
-   *  input. Equivalent to extracting patches with
-   *  `patch_sizes_eff = patch_sizes + (patch_sizes - 1) * (rates - 1)`, followed by
-   *  subsampling them spatially by a factor of `rates`. This is equivalent to
-   *  `rate` in dilated (a.k.a. Atrous) convolutions.
-   * @param padding The type of padding algorithm to use.
-   * @return a new instance of ExtractImagePatches
-   */
-  public <T extends TType> ExtractImagePatches<T> extractImagePatches(Operand<T> images,
-      List<Long> ksizes, List<Long> strides, List<Long> rates, String padding) {
-    return ExtractImagePatches.create(scope, images, ksizes, strides, rates, padding);
-  }
-
-  /**
-   * Extract the shape information of a JPEG-encoded image.
-   *  <p>
-   *  This op only parses the image header, so it is much faster than DecodeJpeg.
-   *
-   * @param <T> data type for {@code imageShape()} output
-   * @param contents 0-D. The JPEG-encoded image.
-   * @return a new instance of ExtractJpegShape
-   */
-  public ExtractJpegShape<TInt32> extractJpegShape(Operand<TString> contents) {
-    return ExtractJpegShape.create(scope, contents);
-  }
-
-  /**
-   * Extract the shape information of a JPEG-encoded image.
-   *  <p>
-   *  This op only parses the image header, so it is much faster than DecodeJpeg.
-   *
-   * @param <T> data type for {@code imageShape()} output
-   * @param contents 0-D. The JPEG-encoded image.
-   * @param outputType (Optional) The output type of the operation (int32 or int64).
-   *  Defaults to int32.
-   * @return a new instance of ExtractJpegShape
-   */
-  public <T extends TNumber> ExtractJpegShape<T> extractJpegShape(Operand<TString> contents,
-      DataType<T> outputType) {
-    return ExtractJpegShape.create(scope, contents, outputType);
-  }
-
-  /**
-   * Convert one or more images from HSV to RGB.
-   *  <p>
-   *  Outputs a tensor of the same shape as the `images` tensor, containing the RGB
-   *  value of the pixels. The output is only well defined if the value in `images`
-   *  are in `[0,1]`.
-   *  <p>
-   *  See `rgb_to_hsv` for a description of the HSV encoding.
-   *
-   * @param <T> data type for {@code output()} output
-   * @param images 1-D or higher rank. HSV data to convert. Last dimension must be size 3.
-   * @return a new instance of HsvToRgb
-   */
-  public <T extends TNumber> HsvToRgb<T> hsvToRgb(Operand<T> images) {
-    return HsvToRgb.create(scope, images);
-  }
-
-  /**
-   * Greedily selects a subset of bounding boxes in descending order of score,
-   *  <p>
-   *  pruning away boxes that have high intersection-over-union (IOU) overlap
-   *  with previously selected boxes.  Bounding boxes with score less than
-   *  `score_threshold` are removed.  Bounding boxes are supplied as
-   *  [y1, x1, y2, x2], where (y1, x1) and (y2, x2) are the coordinates of any
-   *  diagonal pair of box corners and the coordinates can be provided as normalized
-   *  (i.e., lying in the interval [0, 1]) or absolute.  Note that this algorithm
-   *  is agnostic to where the origin is in the coordinate system and more
-   *  generally is invariant to orthogonal transformations and translations
-   *  of the coordinate system; thus translating or reflections of the coordinate
-   *  system result in the same boxes being selected by the algorithm.
-   *  The output of this operation is a set of integers indexing into the input
-   *  collection of bounding boxes representing the selected boxes.  The bounding
-   *  box coordinates corresponding to the selected indices can then be obtained
-   *  using the `tf.gather operation`.  For example:
-   *    selected_indices = tf.image.non_max_suppression_v2(
-   *        boxes, scores, max_output_size, iou_threshold, score_threshold)
-   *    selected_boxes = tf.gather(boxes, selected_indices)
-   *  This op also supports a Soft-NMS (with Gaussian weighting) mode (c.f.
-   *  Bodla et al, https://arxiv.org/abs/1704.04503) where boxes reduce the score
-   *  of other overlapping boxes instead of directly causing them to be pruned.
-   *  To enable this Soft-NMS mode, set the `soft_nms_sigma` parameter to be
-   *  larger than 0.
-   *
-   * @param <T> data type for {@code selectedScores()} output
-   * @param boxes A 2-D float tensor of shape `[num_boxes, 4]`.
-   * @param scores A 1-D float tensor of shape `[num_boxes]` representing a single
-   *  score corresponding to each box (each row of boxes).
-   * @param maxOutputSize A scalar integer tensor representing the maximum number of
-   *  boxes to be selected by non max suppression.
-   * @param iouThreshold A 0-D float tensor representing the threshold for deciding whether
-   *  boxes overlap too much with respect to IOU.
-   * @param scoreThreshold A 0-D float tensor representing the threshold for deciding when to remove
-   *  boxes based on score.
-   * @param softNmsSigma A 0-D float tensor representing the sigma parameter for Soft NMS; see Bodla et
-   *  al (c.f. https://arxiv.org/abs/1704.04503).  When `soft_nms_sigma=0.0` (which
-   *  is default), we fall back to standard (hard) NMS.
-   * @param options carries optional attributes values
-   * @return a new instance of NonMaxSuppression
-   */
-  public <T extends TNumber> NonMaxSuppression<T> nonMaxSuppression(Operand<T> boxes,
-      Operand<T> scores, Operand<TInt32> maxOutputSize, Operand<T> iouThreshold,
-      Operand<T> scoreThreshold, Operand<T> softNmsSigma, NonMaxSuppression.Options... options) {
-    return NonMaxSuppression.create(scope, boxes, scores, maxOutputSize, iouThreshold, scoreThreshold, softNmsSigma, options);
-  }
-
-  /**
-   * Greedily selects a subset of bounding boxes in descending order of score,
-   *  <p>
-   *  pruning away boxes that have high overlaps
-   *  with previously selected boxes.  Bounding boxes with score less than
-   *  `score_threshold` are removed. N-by-n overlap values are supplied as square matrix,
-   *  which allows for defining a custom overlap criterium (eg. intersection over union,
-   *  intersection over area, etc.).
-   *  <p>
-   *  The output of this operation is a set of integers indexing into the input
-   *  collection of bounding boxes representing the selected boxes.  The bounding
-   *  box coordinates corresponding to the selected indices can then be obtained
-   *  using the `tf.gather operation`.  For example:
-   *  <p>
-   *    selected_indices = tf.image.non_max_suppression_with_overlaps(
-   *        overlaps, scores, max_output_size, overlap_threshold, score_threshold)
-   *    selected_boxes = tf.gather(boxes, selected_indices)
-   *
-   * @param overlaps A 2-D float tensor of shape `[num_boxes, num_boxes]` representing
-   *  the n-by-n box overlap values.
-   * @param scores A 1-D float tensor of shape `[num_boxes]` representing a single
-   *  score corresponding to each box (each row of boxes).
-   * @param maxOutputSize A scalar integer tensor representing the maximum number of
-   *  boxes to be selected by non max suppression.
-   * @param overlapThreshold A 0-D float tensor representing the threshold for deciding whether
-   *  boxes overlap too.
-   * @param scoreThreshold A 0-D float tensor representing the threshold for deciding when to remove
-   *  boxes based on score.
-   * @return a new instance of NonMaxSuppressionWithOverlaps
-   */
-  public NonMaxSuppressionWithOverlaps nonMaxSuppressionWithOverlaps(Operand<TFloat32> overlaps,
-      Operand<TFloat32> scores, Operand<TInt32> maxOutputSize, Operand<TFloat32> overlapThreshold,
-      Operand<TFloat32> scoreThreshold) {
-    return NonMaxSuppressionWithOverlaps.create(scope, overlaps, scores, maxOutputSize, overlapThreshold, scoreThreshold);
-  }
-
-  /**
-   * Resize quantized `images` to `size` using quantized bilinear interpolation.
-   *  <p>
-   *  Input images and output images must be quantized types.
-   *
-   * @param <T> data type for {@code resizedImages()} output
-   * @param images 4-D with shape `[batch, height, width, channels]`.
-   * @param size = A 1-D int32 Tensor of 2 elements: `new_height, new_width`.  The
-   *  new size for the images.
-   * @param min
-   * @param max
-   * @param options carries optional attributes values
-   * @return a new instance of QuantizedResizeBilinear
-   */
-  public <T extends TType> QuantizedResizeBilinear<T> quantizedResizeBilinear(Operand<T> images,
-      Operand<TInt32> size, Operand<TFloat32> min, Operand<TFloat32> max,
-      QuantizedResizeBilinear.Options... options) {
-    return QuantizedResizeBilinear.create(scope, images, size, min, max, options);
-  }
-
-  /**
-   * Randomly crop `image`.
-   *  <p>
-   *  `size` is a 1-D int64 tensor with 2 elements representing the crop height and
-   *  width.  The values must be non negative.
-   *  <p>
-   *  This Op picks a random location in `image` and crops a `height` by `width`
-   *  rectangle from that location.  The random location is picked so the cropped
-   *  area will fit inside the original image.
-   *
-   * @param <T> data type for {@code output()} output
-   * @param image 3-D of shape `[height, width, channels]`.
-   * @param size 1-D of length 2 containing: `crop_height`, `crop_width`..
-   * @param options carries optional attributes values
-   * @return a new instance of RandomCrop
-   */
-  public <T extends TNumber> RandomCrop<T> randomCrop(Operand<T> image, Operand<TInt64> size,
-      RandomCrop.Options... options) {
-    return RandomCrop.create(scope, image, size, options);
-  }
-
-  /**
-   * Resize `images` to `size` using area interpolation.
-   *  <p>
-   *  Input images can be of different types but output images are always float.
-   *  <p>
-   *  The range of pixel values for the output image might be slightly different
-   *  from the range for the input image because of limited numerical precision.
-   *  To guarantee an output range, for example `[0.0, 1.0]`, apply
-   *  `tf.clip_by_value` to the output.
-   *  <p>
-   *  Each output pixel is computed by first transforming the pixel's footprint into
-   *  the input tensor and then averaging the pixels that intersect the footprint. An
-   *  input pixel's contribution to the average is weighted by the fraction of its
-   *  area that intersects the footprint.  This is the same as OpenCV's INTER_AREA.
-   *
-   * @param images 4-D with shape `[batch, height, width, channels]`.
-   * @param size = A 1-D int32 Tensor of 2 elements: `new_height, new_width`.  The
-   *  new size for the images.
-   * @param options carries optional attributes values
-   * @return a new instance of ResizeArea
-   */
-  public <T extends TNumber> ResizeArea resizeArea(Operand<T> images, Operand<TInt32> size,
-      ResizeArea.Options... options) {
-    return ResizeArea.create(scope, images, size, options);
-  }
-
-  /**
-   * Resize `images` to `size` using bicubic interpolation.
-   *  <p>
-   *  Input images can be of different types but output images are always float.
-   *
-   * @param images 4-D with shape `[batch, height, width, channels]`.
-   * @param size = A 1-D int32 Tensor of 2 elements: `new_height, new_width`.  The
-   *  new size for the images.
-   * @param options carries optional attributes values
-   * @return a new instance of ResizeBicubic
-   */
-  public <T extends TNumber> ResizeBicubic resizeBicubic(Operand<T> images, Operand<TInt32> size,
-      ResizeBicubic.Options... options) {
-    return ResizeBicubic.create(scope, images, size, options);
-  }
-
-  /**
-   * Resize `images` to `size` using bilinear interpolation.
-   *  <p>
-   *  Input images can be of different types but output images are always float.
-   *
-   * @param images 4-D with shape `[batch, height, width, channels]`.
-   * @param size = A 1-D int32 Tensor of 2 elements: `new_height, new_width`.  The
-   *  new size for the images.
-   * @param options carries optional attributes values
-   * @return a new instance of ResizeBilinear
-   */
-  public <T extends TNumber> ResizeBilinear resizeBilinear(Operand<T> images, Operand<TInt32> size,
-      ResizeBilinear.Options... options) {
-    return ResizeBilinear.create(scope, images, size, options);
-  }
-
-  /**
-   * Resize `images` to `size` using nearest neighbor interpolation.
-   *
-   * @param <T> data type for {@code resizedImages()} output
-   * @param images 4-D with shape `[batch, height, width, channels]`.
-   * @param size = A 1-D int32 Tensor of 2 elements: `new_height, new_width`.  The
-   *  new size for the images.
-   * @param options carries optional attributes values
-   * @return a new instance of ResizeNearestNeighbor
-   */
-  public <T extends TNumber> ResizeNearestNeighbor<T> resizeNearestNeighbor(Operand<T> images,
-      Operand<TInt32> size, ResizeNearestNeighbor.Options... options) {
-    return ResizeNearestNeighbor.create(scope, images, size, options);
-  }
-
-  /**
-   * Converts one or more images from RGB to HSV.
-   *  <p>
-   *  Outputs a tensor of the same shape as the `images` tensor, containing the HSV
-   *  value of the pixels. The output is only well defined if the value in `images`
-   *  are in `[0,1]`.
-   *  <p>
-   *  `output[..., 0]` contains hue, `output[..., 1]` contains saturation, and
-   *  `output[..., 2]` contains value. All HSV values are in `[0,1]`. A hue of 0
-   *  corresponds to pure red, hue 1/3 is pure green, and 2/3 is pure blue.
-   *  <p>
-   *  Usage Example:
-   *  <p>
-   *  >>> blue_image = tf.stack([
-   *  ...    tf.zeros([5,5]),
-   *  ...    tf.zeros([5,5]),
-   *  ...    tf.ones([5,5])],
-   *  ...    axis=-1)
-   *  >>> blue_hsv_image = tf.image.rgb_to_hsv(blue_image)
-   *  >>> blue_hsv_image[0,0].numpy()
-   *  array([0.6666667, 1. , 1. ], dtype=float32)
-   *
-   * @param <T> data type for {@code output()} output
-   * @param images 1-D or higher rank. RGB data to convert. Last dimension must be size 3.
-   * @return a new instance of RgbToHsv
-   */
-  public <T extends TNumber> RgbToHsv<T> rgbToHsv(Operand<T> images) {
-    return RgbToHsv.create(scope, images);
-  }
-
-  /**
-   * Generate a single randomly distorted bounding box for an image.
-   *  <p>
-   *  Bounding box annotations are often supplied in addition to ground-truth labels
-   *  in image recognition or object localization tasks. A common technique for
-   *  training such a system is to randomly distort an image while preserving
-   *  its content, i.e. <i>data augmentation</i>. This Op outputs a randomly distorted
-   *  localization of an object, i.e. bounding box, given an `image_size`,
-   *  `bounding_boxes` and a series of constraints.
-   *  <p>
-   *  The output of this Op is a single bounding box that may be used to crop the
-   *  original image. The output is returned as 3 tensors: `begin`, `size` and
-   *  `bboxes`. The first 2 tensors can be fed directly into `tf.slice` to crop the
-   *  image. The latter may be supplied to `tf.image.draw_bounding_boxes` to visualize
-   *  what the bounding box looks like.
-   *  <p>
-   *  Bounding boxes are supplied and returned as `[y_min, x_min, y_max, x_max]`. The
-   *  bounding box coordinates are floats in `[0.0, 1.0]` relative to the width and
-   *  height of the underlying image.
-   *  <p>
-   *  For example,
-   *  <pre>{@code
-   *      # Generate a single distorted bounding box.
-   *      begin, size, bbox_for_draw = tf.image.sample_distorted_bounding_box(
-   *          tf.shape(image),
-   *          bounding_boxes=bounding_boxes)
-   *
-   *      # Draw the bounding box in an image summary.
-   *      image_with_box = tf.image.draw_bounding_boxes(tf.expand_dims(image, 0),
-   *                                                    bbox_for_draw)
-   *      tf.summary.image('images_with_box', image_with_box)
-   *
-   *      # Employ the bounding box to distort the image.
-   *      distorted_image = tf.slice(image, begin, size)
-   *  }</pre>
-   *  Note that if no bounding box information is available, setting
-   *  `use_image_if_no_bounding_boxes = true` will assume there is a single implicit
-   *  bounding box covering the whole image. If `use_image_if_no_bounding_boxes` is
-   *  false and no bounding boxes are supplied, an error is raised.
-   *
-   * @param <T> data type for {@code begin()} output
-   * @param imageSize 1-D, containing `[height, width, channels]`.
-   * @param boundingBoxes 3-D with shape `[batch, N, 4]` describing the N bounding boxes
-   *  associated with the image.
-   * @param minObjectCovered The cropped area of the image must contain at least this
-   *  fraction of any bounding box supplied. The value of this parameter should be
-   *  non-negative. In the case of 0, the cropped area does not need to overlap
-   *  any of the bounding boxes supplied.
-   * @param options carries optional attributes values
-   * @return a new instance of SampleDistortedBoundingBox
-   */
-  public <T extends TNumber> SampleDistortedBoundingBox<T> sampleDistortedBoundingBox(
-      Operand<T> imageSize, Operand<TFloat32> boundingBoxes, Operand<TFloat32> minObjectCovered,
-      SampleDistortedBoundingBox.Options... options) {
-    return SampleDistortedBoundingBox.create(scope, imageSize, boundingBoxes, minObjectCovered, options);
-  }
-
-  /**
-   *
-   * @param images
-   * @param size
-   * @param scale
-   * @param translation
-   * @param options carries optional attributes values
-   * @return a new instance of ScaleAndTranslate
-   */
-  public <T extends TNumber> ScaleAndTranslate scaleAndTranslate(Operand<T> images,
-      Operand<TInt32> size, Operand<TFloat32> scale, Operand<TFloat32> translation,
-      ScaleAndTranslate.Options... options) {
-    return ScaleAndTranslate.create(scope, images, size, scale, translation, options);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/IoOps.java b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/IoOps.java
deleted file mode 100644
index adc656dc5af..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/IoOps.java
+++ /dev/null
@@ -1,1033 +0,0 @@
-// Copyright 2020 The TensorFlow Authors. All Rights Reserved.
-//
-// Licensed under the Apache License, Version 2.0 (the "License");
-// you may not use this file except in compliance with the License.
-// You may obtain a copy of the License at
-//
-//     http://www.apache.org/licenses/LICENSE-2.0
-//
-// Unless required by applicable law or agreed to in writing, software
-// distributed under the License is distributed on an "AS IS" BASIS,
-// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-// See the License for the specific language governing permissions and
-// limitations under the License.
-// ==============================================================================
-//
-// This class has been generated, DO NOT EDIT!
-//
-package org.tensorflow.op;
-
-import java.util.List;
-import org.tensorflow.DataType;
-import org.tensorflow.Operand;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.op.io.DecodeBase64;
-import org.tensorflow.op.io.DecodeCompressed;
-import org.tensorflow.op.io.DecodeCsv;
-import org.tensorflow.op.io.DecodeJsonExample;
-import org.tensorflow.op.io.DecodePaddedRaw;
-import org.tensorflow.op.io.DecodeRaw;
-import org.tensorflow.op.io.DeserializeManySparse;
-import org.tensorflow.op.io.EncodeBase64;
-import org.tensorflow.op.io.FifoQueue;
-import org.tensorflow.op.io.FixedLengthRecordReader;
-import org.tensorflow.op.io.IdentityReader;
-import org.tensorflow.op.io.LmdbReader;
-import org.tensorflow.op.io.MatchingFiles;
-import org.tensorflow.op.io.PaddingFifoQueue;
-import org.tensorflow.op.io.ParseExample;
-import org.tensorflow.op.io.ParseSequenceExample;
-import org.tensorflow.op.io.ParseSingleExample;
-import org.tensorflow.op.io.ParseSingleSequenceExample;
-import org.tensorflow.op.io.ParseTensor;
-import org.tensorflow.op.io.PriorityQueue;
-import org.tensorflow.op.io.QueueClose;
-import org.tensorflow.op.io.QueueDequeue;
-import org.tensorflow.op.io.QueueDequeueMany;
-import org.tensorflow.op.io.QueueDequeueUpTo;
-import org.tensorflow.op.io.QueueEnqueue;
-import org.tensorflow.op.io.QueueEnqueueMany;
-import org.tensorflow.op.io.QueueIsClosed;
-import org.tensorflow.op.io.QueueSize;
-import org.tensorflow.op.io.RandomShuffleQueue;
-import org.tensorflow.op.io.ReadFile;
-import org.tensorflow.op.io.ReaderNumRecordsProduced;
-import org.tensorflow.op.io.ReaderNumWorkUnitsCompleted;
-import org.tensorflow.op.io.ReaderRead;
-import org.tensorflow.op.io.ReaderReadUpTo;
-import org.tensorflow.op.io.ReaderReset;
-import org.tensorflow.op.io.ReaderRestoreState;
-import org.tensorflow.op.io.ReaderSerializeState;
-import org.tensorflow.op.io.SerializeManySparse;
-import org.tensorflow.op.io.SerializeSparse;
-import org.tensorflow.op.io.SerializeTensor;
-import org.tensorflow.op.io.ShardedFilename;
-import org.tensorflow.op.io.ShardedFilespec;
-import org.tensorflow.op.io.TextLineReader;
-import org.tensorflow.op.io.TfRecordReader;
-import org.tensorflow.op.io.WholeFileReader;
-import org.tensorflow.op.io.WriteFile;
-import org.tensorflow.types.TBool;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.TString;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * An API for building {@code io} operations as {@link Op Op}s
- *
- * @see {@link Ops}
- */
-public final class IoOps {
-  private final Scope scope;
-
-  IoOps(Scope scope) {
-    this.scope = scope;
-  }
-
-  /**
-   * Decode web-safe base64-encoded strings.
-   *  <p>
-   *  Input may or may not have padding at the end. See EncodeBase64 for padding.
-   *  Web-safe means that input must use - and _ instead of + and /.
-   *
-   * @param input Base64 strings to decode.
-   * @return a new instance of DecodeBase64
-   */
-  public DecodeBase64 decodeBase64(Operand<TString> input) {
-    return DecodeBase64.create(scope, input);
-  }
-
-  /**
-   * Decompress strings.
-   *  <p>
-   *  This op decompresses each element of the `bytes` input `Tensor`, which
-   *  is assumed to be compressed using the given `compression_type`.
-   *  <p>
-   *  The `output` is a string `Tensor` of the same shape as `bytes`,
-   *  each element containing the decompressed data from the corresponding
-   *  element in `bytes`.
-   *
-   * @param bytes A Tensor of string which is compressed.
-   * @param options carries optional attributes values
-   * @return a new instance of DecodeCompressed
-   */
-  public DecodeCompressed decodeCompressed(Operand<TString> bytes,
-      DecodeCompressed.Options... options) {
-    return DecodeCompressed.create(scope, bytes, options);
-  }
-
-  /**
-   * Convert CSV records to tensors. Each column maps to one tensor.
-   *  <p>
-   *  RFC 4180 format is expected for the CSV records.
-   *  (https://tools.ietf.org/html/rfc4180)
-   *  Note that we allow leading and trailing spaces with int or float field.
-   *
-   * @param records Each string is a record/row in the csv and all records should have
-   *  the same format.
-   * @param recordDefaults One tensor per column of the input record, with either a
-   *  scalar default value for that column or an empty vector if the column is
-   *  required.
-   * @param options carries optional attributes values
-   * @return a new instance of DecodeCsv
-   */
-  public DecodeCsv decodeCsv(Operand<TString> records, Iterable<Operand<?>> recordDefaults,
-      DecodeCsv.Options... options) {
-    return DecodeCsv.create(scope, records, recordDefaults, options);
-  }
-
-  /**
-   * Convert JSON-encoded Example records to binary protocol buffer strings.
-   *  <p>
-   *  This op translates a tensor containing Example records, encoded using
-   *  the [standard JSON
-   *  mapping](https://developers.google.com/protocol-buffers/docs/proto3#json),
-   *  into a tensor containing the same records encoded as binary protocol
-   *  buffers. The resulting tensor can then be fed to any of the other
-   *  Example-parsing ops.
-   *
-   * @param jsonExamples Each string is a JSON object serialized according to the JSON
-   *  mapping of the Example proto.
-   * @return a new instance of DecodeJsonExample
-   */
-  public DecodeJsonExample decodeJsonExample(Operand<TString> jsonExamples) {
-    return DecodeJsonExample.create(scope, jsonExamples);
-  }
-
-  /**
-   * Reinterpret the bytes of a string as a vector of numbers.
-   *
-   * @param <T> data type for {@code output()} output
-   * @param inputBytes Tensor of string to be decoded.
-   * @param fixedLength Length in bytes for each element of the decoded output. Must be a multiple
-   *  of the size of the output type.
-   * @param outType
-   * @param options carries optional attributes values
-   * @return a new instance of DecodePaddedRaw
-   */
-  public <T extends TNumber> DecodePaddedRaw<T> decodePaddedRaw(Operand<TString> inputBytes,
-      Operand<TInt32> fixedLength, DataType<T> outType, DecodePaddedRaw.Options... options) {
-    return DecodePaddedRaw.create(scope, inputBytes, fixedLength, outType, options);
-  }
-
-  /**
-   * Reinterpret the bytes of a string as a vector of numbers.
-   *
-   * @param <T> data type for {@code output()} output
-   * @param bytes All the elements must have the same length.
-   * @param outType
-   * @param options carries optional attributes values
-   * @return a new instance of DecodeRaw
-   */
-  public <T extends TType> DecodeRaw<T> decodeRaw(Operand<TString> bytes, DataType<T> outType,
-      DecodeRaw.Options... options) {
-    return DecodeRaw.create(scope, bytes, outType, options);
-  }
-
-  /**
-   * Deserialize and concatenate `SparseTensors` from a serialized minibatch.
-   *  <p>
-   *  The input `serialized_sparse` must be a string matrix of shape `[N x 3]` where
-   *  `N` is the minibatch size and the rows correspond to packed outputs of
-   *  `SerializeSparse`.  The ranks of the original `SparseTensor` objects
-   *  must all match.  When the final `SparseTensor` is created, it has rank one
-   *  higher than the ranks of the incoming `SparseTensor` objects
-   *  (they have been concatenated along a new row dimension).
-   *  <p>
-   *  The output `SparseTensor` object's shape values for all dimensions but the
-   *  first are the max across the input `SparseTensor` objects' shape values
-   *  for the corresponding dimensions.  Its first shape value is `N`, the minibatch
-   *  size.
-   *  <p>
-   *  The input `SparseTensor` objects' indices are assumed ordered in
-   *  standard lexicographic order.  If this is not the case, after this
-   *  step run `SparseReorder` to restore index ordering.
-   *  <p>
-   *  For example, if the serialized input is a `[2 x 3]` matrix representing two
-   *  original `SparseTensor` objects:
-   *  <p>
-   *      index = [ 0]
-   *              [10]
-   *              [20]
-   *      values = [1, 2, 3]
-   *      shape = [50]
-   *  <p>
-   *  and
-   *  <p>
-   *      index = [ 2]
-   *              [10]
-   *      values = [4, 5]
-   *      shape = [30]
-   *  <p>
-   *  then the final deserialized `SparseTensor` will be:
-   *  <p>
-   *      index = [0  0]
-   *              [0 10]
-   *              [0 20]
-   *              [1  2]
-   *              [1 10]
-   *      values = [1, 2, 3, 4, 5]
-   *      shape = [2 50]
-   *
-   * @param <T> data type for {@code sparseValues()} output
-   * @param serializedSparse 2-D, The `N` serialized `SparseTensor` objects.
-   *  Must have 3 columns.
-   * @param dtype The `dtype` of the serialized `SparseTensor` objects.
-   * @return a new instance of DeserializeManySparse
-   */
-  public <T extends TType> DeserializeManySparse<T> deserializeManySparse(
-      Operand<TString> serializedSparse, DataType<T> dtype) {
-    return DeserializeManySparse.create(scope, serializedSparse, dtype);
-  }
-
-  /**
-   * Encode strings into web-safe base64 format.
-   *  <p>
-   *  Refer to the following article for more information on base64 format:
-   *  en.wikipedia.org/wiki/Base64. Base64 strings may have padding with '=' at the
-   *  end so that the encoded has length multiple of 4. See Padding section of the
-   *  link above.
-   *  <p>
-   *  Web-safe means that the encoder uses - and _ instead of + and /.
-   *
-   * @param input Strings to be encoded.
-   * @param options carries optional attributes values
-   * @return a new instance of EncodeBase64
-   */
-  public EncodeBase64 encodeBase64(Operand<TString> input, EncodeBase64.Options... options) {
-    return EncodeBase64.create(scope, input, options);
-  }
-
-  /**
-   * A queue that produces elements in first-in first-out order.
-   *
-   * @param componentTypes The type of each component in a value.
-   * @param options carries optional attributes values
-   * @return a new instance of FifoQueue
-   */
-  public FifoQueue fifoQueue(List<DataType<?>> componentTypes, FifoQueue.Options... options) {
-    return FifoQueue.create(scope, componentTypes, options);
-  }
-
-  /**
-   * A Reader that outputs fixed-length records from a file.
-   *
-   * @param recordBytes Number of bytes in the record.
-   * @param options carries optional attributes values
-   * @return a new instance of FixedLengthRecordReader
-   */
-  public FixedLengthRecordReader fixedLengthRecordReader(Long recordBytes,
-      FixedLengthRecordReader.Options... options) {
-    return FixedLengthRecordReader.create(scope, recordBytes, options);
-  }
-
-  /**
-   * A Reader that outputs the queued work as both the key and value.
-   *  <p>
-   *  To use, enqueue strings in a Queue.  ReaderRead will take the front
-   *  work string and output (work, work).
-   *
-   * @param options carries optional attributes values
-   * @return a new instance of IdentityReader
-   */
-  public IdentityReader identityReader(IdentityReader.Options... options) {
-    return IdentityReader.create(scope, options);
-  }
-
-  /**
-   * A Reader that outputs the records from a LMDB file.
-   *
-   * @param options carries optional attributes values
-   * @return a new instance of LmdbReader
-   */
-  public LmdbReader lmdbReader(LmdbReader.Options... options) {
-    return LmdbReader.create(scope, options);
-  }
-
-  /**
-   * Returns the set of files matching one or more glob patterns.
-   *  <p>
-   *  Note that this routine only supports wildcard characters in the
-   *  basename portion of the pattern, not in the directory portion.
-   *  Note also that the order of filenames returned is deterministic.
-   *
-   * @param pattern Shell wildcard pattern(s). Scalar or vector of type string.
-   * @return a new instance of MatchingFiles
-   */
-  public MatchingFiles matchingFiles(Operand<TString> pattern) {
-    return MatchingFiles.create(scope, pattern);
-  }
-
-  /**
-   * A queue that produces elements in first-in first-out order.
-   *  <p>
-   *  Variable-size shapes are allowed by setting the corresponding shape dimensions
-   *  to 0 in the shape attr.  In this case DequeueMany will pad up to the maximum
-   *  size of any given element in the minibatch.  See below for details.
-   *
-   * @param componentTypes The type of each component in a value.
-   * @param options carries optional attributes values
-   * @return a new instance of PaddingFifoQueue
-   */
-  public PaddingFifoQueue paddingFifoQueue(List<DataType<?>> componentTypes,
-      PaddingFifoQueue.Options... options) {
-    return PaddingFifoQueue.create(scope, componentTypes, options);
-  }
-
-  /**
-   * Transforms a vector of tf.Example protos (as strings) into typed tensors.
-   *
-   * @param serialized A scalar or vector containing binary serialized Example protos.
-   * @param names A tensor containing the names of the serialized protos.
-   *  Corresponds 1:1 with the `serialized` tensor.
-   *  May contain, for example, table key (descriptive) names for the
-   *  corresponding serialized protos.  These are purely useful for debugging
-   *  purposes, and the presence of values here has no effect on the output.
-   *  May also be an empty vector if no names are available.
-   *  If non-empty, this tensor must have the same shape as "serialized".
-   * @param sparseKeys Vector of strings.
-   *  The keys expected in the Examples' features associated with sparse values.
-   * @param denseKeys Vector of strings.
-   *  The keys expected in the Examples' features associated with dense values.
-   * @param raggedKeys Vector of strings.
-   *  The keys expected in the Examples' features associated with ragged values.
-   * @param denseDefaults A list of Tensors (some may be empty).  Corresponds 1:1 with `dense_keys`.
-   *  dense_defaults[j] provides default values
-   *  when the example's feature_map lacks dense_key[j].  If an empty Tensor is
-   *  provided for dense_defaults[j], then the Feature dense_keys[j] is required.
-   *  The input type is inferred from dense_defaults[j], even when it's empty.
-   *  If dense_defaults[j] is not empty, and dense_shapes[j] is fully defined,
-   *  then the shape of dense_defaults[j] must match that of dense_shapes[j].
-   *  If dense_shapes[j] has an undefined major dimension (variable strides dense
-   *  feature), dense_defaults[j] must contain a single element:
-   *  the padding element.
-   * @param numSparse The number of sparse keys.
-   * @param sparseTypes A list of `num_sparse` types; the data types of data in each Feature
-   *  given in sparse_keys.
-   *  Currently the ParseExample supports DT_FLOAT (FloatList),
-   *  DT_INT64 (Int64List), and DT_STRING (BytesList).
-   * @param raggedValueTypes A list of `num_ragged` types; the data types of data in each Feature
-   *  given in ragged_keys (where `num_ragged = sparse_keys.size()`).
-   *  Currently the ParseExample supports DT_FLOAT (FloatList),
-   *  DT_INT64 (Int64List), and DT_STRING (BytesList).
-   * @param raggedSplitTypes A list of `num_ragged` types; the data types of row_splits in each Feature
-   *  given in ragged_keys (where `num_ragged = sparse_keys.size()`).
-   *  May be DT_INT32 or DT_INT64.
-   * @param denseShapes A list of `num_dense` shapes; the shapes of data in each Feature
-   *  given in dense_keys (where `num_dense = dense_keys.size()`).
-   *  The number of elements in the Feature corresponding to dense_key[j]
-   *  must always equal dense_shapes[j].NumEntries().
-   *  If dense_shapes[j] == (D0, D1, ..., DN) then the shape of output
-   *  Tensor dense_values[j] will be (|serialized|, D0, D1, ..., DN):
-   *  The dense outputs are just the inputs row-stacked by batch.
-   *  This works for dense_shapes[j] = (-1, D1, ..., DN).  In this case
-   *  the shape of the output Tensor dense_values[j] will be
-   *  (|serialized|, M, D1, .., DN), where M is the maximum number of blocks
-   *  of elements of length D1 * .... * DN, across all minibatch entries
-   *  in the input.  Any minibatch entry with less than M blocks of elements of
-   *  length D1 * ... * DN will be padded with the corresponding default_value
-   *  scalar element along the second dimension.
-   * @return a new instance of ParseExample
-   */
-  public ParseExample parseExample(Operand<TString> serialized, Operand<TString> names,
-      Operand<TString> sparseKeys, Operand<TString> denseKeys, Operand<TString> raggedKeys,
-      Iterable<Operand<?>> denseDefaults, Long numSparse, List<DataType<?>> sparseTypes,
-      List<DataType<?>> raggedValueTypes, List<DataType<?>> raggedSplitTypes,
-      List<Shape> denseShapes) {
-    return ParseExample.create(scope, serialized, names, sparseKeys, denseKeys, raggedKeys, denseDefaults, numSparse, sparseTypes, raggedValueTypes, raggedSplitTypes, denseShapes);
-  }
-
-  /**
-   * Transforms a vector of tf.io.SequenceExample protos (as strings) into
-   *  typed tensors.
-   *
-   * @param serialized A scalar or vector containing binary serialized SequenceExample protos.
-   * @param debugName A scalar or vector containing the names of the serialized protos.
-   *  May contain, for example, table key (descriptive) name for the
-   *  corresponding serialized proto.  This is purely useful for debugging
-   *  purposes, and the presence of values here has no effect on the output.
-   *  May also be an empty vector if no name is available.
-   * @param contextSparseKeys The keys expected in the Examples' features associated with context_sparse
-   *  values.
-   * @param contextDenseKeys The keys expected in the SequenceExamples' context features associated with
-   *  dense values.
-   * @param contextRaggedKeys The keys expected in the Examples' features associated with context_ragged
-   *  values.
-   * @param featureListSparseKeys The keys expected in the FeatureLists associated with sparse values.
-   * @param featureListDenseKeys The keys expected in the SequenceExamples' feature_lists associated
-   *  with lists of dense values.
-   * @param featureListRaggedKeys The keys expected in the FeatureLists associated with ragged values.
-   * @param featureListDenseMissingAssumedEmpty A vector corresponding 1:1 with feature_list_dense_keys, indicating which
-   *  features may be missing from the SequenceExamples.  If the associated
-   *  FeatureList is missing, it is treated as empty.
-   * @param contextDenseDefaults A list of Ncontext_dense Tensors (some may be empty).
-   *  context_dense_defaults[j] provides default values
-   *  when the SequenceExample's context map lacks context_dense_key[j].
-   *  If an empty Tensor is provided for context_dense_defaults[j],
-   *  then the Feature context_dense_keys[j] is required.
-   *  The input type is inferred from context_dense_defaults[j], even when it's
-   *  empty.  If context_dense_defaults[j] is not empty, its shape must match
-   *  context_dense_shapes[j].
-   * @param contextSparseTypes A list of Ncontext_sparse types; the data types of data in
-   *  each context Feature given in context_sparse_keys.
-   *  Currently the ParseSingleSequenceExample supports DT_FLOAT (FloatList),
-   *  DT_INT64 (Int64List), and DT_STRING (BytesList).
-   * @param contextRaggedValueTypes RaggedTensor.value dtypes for the ragged context features.
-   * @param contextRaggedSplitTypes RaggedTensor.row_split dtypes for the ragged context features.
-   * @param featureListDenseTypes
-   * @param featureListSparseTypes A list of Nfeature_list_sparse types; the data types
-   *  of data in each FeatureList given in feature_list_sparse_keys.
-   *  Currently the ParseSingleSequenceExample supports DT_FLOAT (FloatList),
-   *  DT_INT64 (Int64List), and DT_STRING (BytesList).
-   * @param featureListRaggedValueTypes RaggedTensor.value dtypes for the ragged FeatureList features.
-   * @param featureListRaggedSplitTypes RaggedTensor.row_split dtypes for the ragged FeatureList features.
-   * @param options carries optional attributes values
-   * @return a new instance of ParseSequenceExample
-   */
-  public ParseSequenceExample parseSequenceExample(Operand<TString> serialized,
-      Operand<TString> debugName, Operand<TString> contextSparseKeys,
-      Operand<TString> contextDenseKeys, Operand<TString> contextRaggedKeys,
-      Operand<TString> featureListSparseKeys, Operand<TString> featureListDenseKeys,
-      Operand<TString> featureListRaggedKeys, Operand<TBool> featureListDenseMissingAssumedEmpty,
-      Iterable<Operand<?>> contextDenseDefaults, List<DataType<?>> contextSparseTypes,
-      List<DataType<?>> contextRaggedValueTypes, List<DataType<?>> contextRaggedSplitTypes,
-      List<DataType<?>> featureListDenseTypes, List<DataType<?>> featureListSparseTypes,
-      List<DataType<?>> featureListRaggedValueTypes, List<DataType<?>> featureListRaggedSplitTypes,
-      ParseSequenceExample.Options... options) {
-    return ParseSequenceExample.create(scope, serialized, debugName, contextSparseKeys, contextDenseKeys, contextRaggedKeys, featureListSparseKeys, featureListDenseKeys, featureListRaggedKeys, featureListDenseMissingAssumedEmpty, contextDenseDefaults, contextSparseTypes, contextRaggedValueTypes, contextRaggedSplitTypes, featureListDenseTypes, featureListSparseTypes, featureListRaggedValueTypes, featureListRaggedSplitTypes, options);
-  }
-
-  /**
-   * Transforms a tf.Example proto (as a string) into typed tensors.
-   *
-   * @param serialized A vector containing a batch of binary serialized Example protos.
-   * @param denseDefaults A list of Tensors (some may be empty), whose length matches
-   *  the length of `dense_keys`. dense_defaults[j] provides default values
-   *  when the example's feature_map lacks dense_key[j].  If an empty Tensor is
-   *  provided for dense_defaults[j], then the Feature dense_keys[j] is required.
-   *  The input type is inferred from dense_defaults[j], even when it's empty.
-   *  If dense_defaults[j] is not empty, and dense_shapes[j] is fully defined,
-   *  then the shape of dense_defaults[j] must match that of dense_shapes[j].
-   *  If dense_shapes[j] has an undefined major dimension (variable strides dense
-   *  feature), dense_defaults[j] must contain a single element:
-   *  the padding element.
-   * @param numSparse The number of sparse features to be parsed from the example. This
-   *  must match the lengths of `sparse_keys` and `sparse_types`.
-   * @param sparseKeys A list of `num_sparse` strings.
-   *  The keys expected in the Examples' features associated with sparse values.
-   * @param denseKeys The keys expected in the Examples' features associated with dense
-   *  values.
-   * @param sparseTypes A list of `num_sparse` types; the data types of data in each
-   *  Feature given in sparse_keys.
-   *  Currently the ParseSingleExample op supports DT_FLOAT (FloatList),
-   *  DT_INT64 (Int64List), and DT_STRING (BytesList).
-   * @param denseShapes The shapes of data in each Feature given in dense_keys.
-   *  The length of this list must match the length of `dense_keys`.  The
-   *  number of elements in the Feature corresponding to dense_key[j] must
-   *  always equal dense_shapes[j].NumEntries().  If dense_shapes[j] ==
-   *  (D0, D1, ..., DN) then the shape of output Tensor dense_values[j]
-   *  will be (D0, D1, ..., DN): In the case dense_shapes[j] = (-1, D1,
-   *  ..., DN), the shape of the output Tensor dense_values[j] will be (M,
-   *  D1, .., DN), where M is the number of blocks of elements of length
-   *  D1 * .... * DN, in the input.
-   * @return a new instance of ParseSingleExample
-   */
-  public ParseSingleExample parseSingleExample(Operand<TString> serialized,
-      Iterable<Operand<?>> denseDefaults, Long numSparse, List<String> sparseKeys,
-      List<String> denseKeys, List<DataType<?>> sparseTypes, List<Shape> denseShapes) {
-    return ParseSingleExample.create(scope, serialized, denseDefaults, numSparse, sparseKeys, denseKeys, sparseTypes, denseShapes);
-  }
-
-  /**
-   * Transforms a scalar brain.SequenceExample proto (as strings) into typed tensors.
-   *
-   * @param serialized A scalar containing a binary serialized SequenceExample proto.
-   * @param featureListDenseMissingAssumedEmpty A vector listing the
-   *  FeatureList keys which may be missing from the SequenceExample.  If the
-   *  associated FeatureList is missing, it is treated as empty.  By default,
-   *  any FeatureList not listed in this vector must exist in the SequenceExample.
-   * @param contextSparseKeys A list of Ncontext_sparse string Tensors (scalars).
-   *  The keys expected in the Examples' features associated with context_sparse
-   *  values.
-   * @param contextDenseKeys A list of Ncontext_dense string Tensors (scalars).
-   *  The keys expected in the SequenceExamples' context features associated with
-   *  dense values.
-   * @param featureListSparseKeys A list of Nfeature_list_sparse string Tensors
-   *  (scalars).  The keys expected in the FeatureLists associated with sparse
-   *  values.
-   * @param featureListDenseKeys A list of Nfeature_list_dense string Tensors (scalars).
-   *  The keys expected in the SequenceExamples' feature_lists associated
-   *  with lists of dense values.
-   * @param contextDenseDefaults A list of Ncontext_dense Tensors (some may be empty).
-   *  context_dense_defaults[j] provides default values
-   *  when the SequenceExample's context map lacks context_dense_key[j].
-   *  If an empty Tensor is provided for context_dense_defaults[j],
-   *  then the Feature context_dense_keys[j] is required.
-   *  The input type is inferred from context_dense_defaults[j], even when it's
-   *  empty.  If context_dense_defaults[j] is not empty, its shape must match
-   *  context_dense_shapes[j].
-   * @param debugName A scalar containing the name of the serialized proto.
-   *  May contain, for example, table key (descriptive) name for the
-   *  corresponding serialized proto.  This is purely useful for debugging
-   *  purposes, and the presence of values here has no effect on the output.
-   *  May also be an empty scalar if no name is available.
-   * @param contextSparseTypes A list of Ncontext_sparse types; the data types of data in
-   *  each context Feature given in context_sparse_keys.
-   *  Currently the ParseSingleSequenceExample supports DT_FLOAT (FloatList),
-   *  DT_INT64 (Int64List), and DT_STRING (BytesList).
-   * @param featureListDenseTypes
-   * @param featureListSparseTypes A list of Nfeature_list_sparse types; the data types
-   *  of data in each FeatureList given in feature_list_sparse_keys.
-   *  Currently the ParseSingleSequenceExample supports DT_FLOAT (FloatList),
-   *  DT_INT64 (Int64List), and DT_STRING (BytesList).
-   * @param options carries optional attributes values
-   * @return a new instance of ParseSingleSequenceExample
-   */
-  public ParseSingleSequenceExample parseSingleSequenceExample(Operand<TString> serialized,
-      Operand<TString> featureListDenseMissingAssumedEmpty,
-      Iterable<Operand<TString>> contextSparseKeys, Iterable<Operand<TString>> contextDenseKeys,
-      Iterable<Operand<TString>> featureListSparseKeys,
-      Iterable<Operand<TString>> featureListDenseKeys, Iterable<Operand<?>> contextDenseDefaults,
-      Operand<TString> debugName, List<DataType<?>> contextSparseTypes,
-      List<DataType<?>> featureListDenseTypes, List<DataType<?>> featureListSparseTypes,
-      ParseSingleSequenceExample.Options... options) {
-    return ParseSingleSequenceExample.create(scope, serialized, featureListDenseMissingAssumedEmpty, contextSparseKeys, contextDenseKeys, featureListSparseKeys, featureListDenseKeys, contextDenseDefaults, debugName, contextSparseTypes, featureListDenseTypes, featureListSparseTypes, options);
-  }
-
-  /**
-   * Transforms a serialized tensorflow.TensorProto proto into a Tensor.
-   *
-   * @param <T> data type for {@code output()} output
-   * @param serialized A scalar string containing a serialized TensorProto proto.
-   * @param outType The type of the serialized tensor.  The provided type must match the
-   *  type of the serialized tensor and no implicit conversion will take place.
-   * @return a new instance of ParseTensor
-   */
-  public <T extends TType> ParseTensor<T> parseTensor(Operand<TString> serialized,
-      DataType<T> outType) {
-    return ParseTensor.create(scope, serialized, outType);
-  }
-
-  /**
-   * A queue that produces elements sorted by the first component value.
-   *  <p>
-   *  Note that the PriorityQueue requires the first component of any element
-   *  to be a scalar int64, in addition to the other elements declared by
-   *  component_types.  Therefore calls to Enqueue and EnqueueMany (resp. Dequeue
-   *  and DequeueMany) on a PriorityQueue will all require (resp. output) one extra
-   *  entry in their input (resp. output) lists.
-   *
-   * @param componentTypes The type of each component in a value.
-   * @param shapes The shape of each component in a value. The length of this attr must
-   *  be either 0 or the same as the length of component_types. If the length of
-   *  this attr is 0, the shapes of queue elements are not constrained, and
-   *  only one element may be dequeued at a time.
-   * @param options carries optional attributes values
-   * @return a new instance of PriorityQueue
-   */
-  public PriorityQueue priorityQueue(List<DataType<?>> componentTypes, List<Shape> shapes,
-      PriorityQueue.Options... options) {
-    return PriorityQueue.create(scope, componentTypes, shapes, options);
-  }
-
-  /**
-   * Closes the given queue.
-   *  <p>
-   *  This operation signals that no more elements will be enqueued in the
-   *  given queue. Subsequent Enqueue(Many) operations will fail.
-   *  Subsequent Dequeue(Many) operations will continue to succeed if
-   *  sufficient elements remain in the queue. Subsequent Dequeue(Many)
-   *  operations that would block will fail immediately.
-   *
-   * @param handle The handle to a queue.
-   * @param options carries optional attributes values
-   * @return a new instance of QueueClose
-   */
-  public QueueClose queueClose(Operand<?> handle, QueueClose.Options... options) {
-    return QueueClose.create(scope, handle, options);
-  }
-
-  /**
-   * Dequeues a tuple of one or more tensors from the given queue.
-   *  <p>
-   *  This operation has k outputs, where k is the number of components
-   *  in the tuples stored in the given queue, and output i is the ith
-   *  component of the dequeued tuple.
-   *  <p>
-   *  N.B. If the queue is empty, this operation will block until an element
-   *  has been dequeued (or 'timeout_ms' elapses, if specified).
-   *
-   * @param handle The handle to a queue.
-   * @param componentTypes The type of each component in a tuple.
-   * @param options carries optional attributes values
-   * @return a new instance of QueueDequeue
-   */
-  public QueueDequeue queueDequeue(Operand<?> handle, List<DataType<?>> componentTypes,
-      QueueDequeue.Options... options) {
-    return QueueDequeue.create(scope, handle, componentTypes, options);
-  }
-
-  /**
-   * Dequeues `n` tuples of one or more tensors from the given queue.
-   *  <p>
-   *  If the queue is closed and there are fewer than `n` elements, then an
-   *  OutOfRange error is returned.
-   *  <p>
-   *  This operation concatenates queue-element component tensors along the
-   *  0th dimension to make a single component tensor.  All of the components
-   *  in the dequeued tuple will have size `n` in the 0th dimension.
-   *  <p>
-   *  This operation has `k` outputs, where `k` is the number of components in
-   *  the tuples stored in the given queue, and output `i` is the ith
-   *  component of the dequeued tuple.
-   *  <p>
-   *  N.B. If the queue is empty, this operation will block until `n` elements
-   *  have been dequeued (or 'timeout_ms' elapses, if specified).
-   *
-   * @param handle The handle to a queue.
-   * @param n The number of tuples to dequeue.
-   * @param componentTypes The type of each component in a tuple.
-   * @param options carries optional attributes values
-   * @return a new instance of QueueDequeueMany
-   */
-  public QueueDequeueMany queueDequeueMany(Operand<?> handle, Operand<TInt32> n,
-      List<DataType<?>> componentTypes, QueueDequeueMany.Options... options) {
-    return QueueDequeueMany.create(scope, handle, n, componentTypes, options);
-  }
-
-  /**
-   * Dequeues `n` tuples of one or more tensors from the given queue.
-   *  <p>
-   *  This operation is not supported by all queues.  If a queue does not support
-   *  DequeueUpTo, then an Unimplemented error is returned.
-   *  <p>
-   *  If the queue is closed and there are more than 0 but less than `n`
-   *  elements remaining, then instead of returning an OutOfRange error like
-   *  QueueDequeueMany, less than `n` elements are returned immediately.  If
-   *  the queue is closed and there are 0 elements left in the queue, then
-   *  an OutOfRange error is returned just like in QueueDequeueMany.
-   *  Otherwise the behavior is identical to QueueDequeueMany:
-   *  <p>
-   *  This operation concatenates queue-element component tensors along the
-   *  0th dimension to make a single component tensor.  All of the components
-   *  in the dequeued tuple will have size n in the 0th dimension.
-   *  <p>
-   *  This operation has `k` outputs, where `k` is the number of components in
-   *  the tuples stored in the given queue, and output `i` is the ith
-   *  component of the dequeued tuple.
-   *
-   * @param handle The handle to a queue.
-   * @param n The number of tuples to dequeue.
-   * @param componentTypes The type of each component in a tuple.
-   * @param options carries optional attributes values
-   * @return a new instance of QueueDequeueUpTo
-   */
-  public QueueDequeueUpTo queueDequeueUpTo(Operand<?> handle, Operand<TInt32> n,
-      List<DataType<?>> componentTypes, QueueDequeueUpTo.Options... options) {
-    return QueueDequeueUpTo.create(scope, handle, n, componentTypes, options);
-  }
-
-  /**
-   * Enqueues a tuple of one or more tensors in the given queue.
-   *  <p>
-   *  The components input has k elements, which correspond to the components of
-   *  tuples stored in the given queue.
-   *  <p>
-   *  N.B. If the queue is full, this operation will block until the given
-   *  element has been enqueued (or 'timeout_ms' elapses, if specified).
-   *
-   * @param handle The handle to a queue.
-   * @param components One or more tensors from which the enqueued tensors should be taken.
-   * @param options carries optional attributes values
-   * @return a new instance of QueueEnqueue
-   */
-  public QueueEnqueue queueEnqueue(Operand<?> handle, Iterable<Operand<?>> components,
-      QueueEnqueue.Options... options) {
-    return QueueEnqueue.create(scope, handle, components, options);
-  }
-
-  /**
-   * Enqueues zero or more tuples of one or more tensors in the given queue.
-   *  <p>
-   *  This operation slices each component tensor along the 0th dimension to
-   *  make multiple queue elements. All of the tuple components must have the
-   *  same size in the 0th dimension.
-   *  <p>
-   *  The components input has k elements, which correspond to the components of
-   *  tuples stored in the given queue.
-   *  <p>
-   *  N.B. If the queue is full, this operation will block until the given
-   *  elements have been enqueued (or 'timeout_ms' elapses, if specified).
-   *
-   * @param handle The handle to a queue.
-   * @param components One or more tensors from which the enqueued tensors should
-   *  be taken.
-   * @param options carries optional attributes values
-   * @return a new instance of QueueEnqueueMany
-   */
-  public QueueEnqueueMany queueEnqueueMany(Operand<?> handle, Iterable<Operand<?>> components,
-      QueueEnqueueMany.Options... options) {
-    return QueueEnqueueMany.create(scope, handle, components, options);
-  }
-
-  /**
-   * Returns true if queue is closed.
-   *  <p>
-   *  This operation returns true if the queue is closed and false if the queue
-   *  is open.
-   *
-   * @param handle The handle to a queue.
-   * @return a new instance of QueueIsClosed
-   */
-  public QueueIsClosed queueIsClosed(Operand<?> handle) {
-    return QueueIsClosed.create(scope, handle);
-  }
-
-  /**
-   * Computes the number of elements in the given queue.
-   *
-   * @param handle The handle to a queue.
-   * @return a new instance of QueueSize
-   */
-  public QueueSize queueSize(Operand<?> handle) {
-    return QueueSize.create(scope, handle);
-  }
-
-  /**
-   * A queue that randomizes the order of elements.
-   *
-   * @param componentTypes The type of each component in a value.
-   * @param options carries optional attributes values
-   * @return a new instance of RandomShuffleQueue
-   */
-  public RandomShuffleQueue randomShuffleQueue(List<DataType<?>> componentTypes,
-      RandomShuffleQueue.Options... options) {
-    return RandomShuffleQueue.create(scope, componentTypes, options);
-  }
-
-  /**
-   * Reads and outputs the entire contents of the input filename.
-   *
-   * @param filename
-   * @return a new instance of ReadFile
-   */
-  public ReadFile readFile(Operand<TString> filename) {
-    return ReadFile.create(scope, filename);
-  }
-
-  /**
-   * Returns the number of records this Reader has produced.
-   *  <p>
-   *  This is the same as the number of ReaderRead executions that have
-   *  succeeded.
-   *
-   * @param readerHandle Handle to a Reader.
-   * @return a new instance of ReaderNumRecordsProduced
-   */
-  public ReaderNumRecordsProduced readerNumRecordsProduced(Operand<?> readerHandle) {
-    return ReaderNumRecordsProduced.create(scope, readerHandle);
-  }
-
-  /**
-   * Returns the number of work units this Reader has finished processing.
-   *
-   * @param readerHandle Handle to a Reader.
-   * @return a new instance of ReaderNumWorkUnitsCompleted
-   */
-  public ReaderNumWorkUnitsCompleted readerNumWorkUnitsCompleted(Operand<?> readerHandle) {
-    return ReaderNumWorkUnitsCompleted.create(scope, readerHandle);
-  }
-
-  /**
-   * Returns the next record (key, value pair) produced by a Reader.
-   *  <p>
-   *  Will dequeue from the input queue if necessary (e.g. when the
-   *  Reader needs to start reading from a new file since it has finished
-   *  with the previous file).
-   *
-   * @param readerHandle Handle to a Reader.
-   * @param queueHandle Handle to a Queue, with string work items.
-   * @return a new instance of ReaderRead
-   */
-  public ReaderRead readerRead(Operand<?> readerHandle, Operand<?> queueHandle) {
-    return ReaderRead.create(scope, readerHandle, queueHandle);
-  }
-
-  /**
-   * Returns up to `num_records` (key, value) pairs produced by a Reader.
-   *  <p>
-   *  Will dequeue from the input queue if necessary (e.g. when the
-   *  Reader needs to start reading from a new file since it has finished
-   *  with the previous file).
-   *  It may return less than `num_records` even before the last batch.
-   *
-   * @param readerHandle Handle to a `Reader`.
-   * @param queueHandle Handle to a `Queue`, with string work items.
-   * @param numRecords number of records to read from `Reader`.
-   * @return a new instance of ReaderReadUpTo
-   */
-  public ReaderReadUpTo readerReadUpTo(Operand<?> readerHandle, Operand<?> queueHandle,
-      Operand<TInt64> numRecords) {
-    return ReaderReadUpTo.create(scope, readerHandle, queueHandle, numRecords);
-  }
-
-  /**
-   * Restore a Reader to its initial clean state.
-   *
-   * @param readerHandle Handle to a Reader.
-   * @return a new instance of ReaderReset
-   */
-  public ReaderReset readerReset(Operand<?> readerHandle) {
-    return ReaderReset.create(scope, readerHandle);
-  }
-
-  /**
-   * Restore a reader to a previously saved state.
-   *  <p>
-   *  Not all Readers support being restored, so this can produce an
-   *  Unimplemented error.
-   *
-   * @param readerHandle Handle to a Reader.
-   * @param state Result of a ReaderSerializeState of a Reader with type
-   *  matching reader_handle.
-   * @return a new instance of ReaderRestoreState
-   */
-  public ReaderRestoreState readerRestoreState(Operand<?> readerHandle, Operand<TString> state) {
-    return ReaderRestoreState.create(scope, readerHandle, state);
-  }
-
-  /**
-   * Produce a string tensor that encodes the state of a Reader.
-   *  <p>
-   *  Not all Readers support being serialized, so this can produce an
-   *  Unimplemented error.
-   *
-   * @param readerHandle Handle to a Reader.
-   * @return a new instance of ReaderSerializeState
-   */
-  public ReaderSerializeState readerSerializeState(Operand<?> readerHandle) {
-    return ReaderSerializeState.create(scope, readerHandle);
-  }
-
-  /**
-   * Serialize an `N`-minibatch `SparseTensor` into an `[N, 3]` `Tensor` object.
-   *  <p>
-   *  The `SparseTensor` must have rank `R` greater than 1, and the first dimension
-   *  is treated as the minibatch dimension.  Elements of the `SparseTensor`
-   *  must be sorted in increasing order of this first dimension.  The serialized
-   *  `SparseTensor` objects going into each row of `serialized_sparse` will have
-   *  rank `R-1`.
-   *  <p>
-   *  The minibatch size `N` is extracted from `sparse_shape[0]`.
-   *
-   * @param <U> data type for {@code serializedSparse()} output
-   * @param sparseIndices 2-D.  The `indices` of the minibatch `SparseTensor`.
-   * @param sparseValues 1-D.  The `values` of the minibatch `SparseTensor`.
-   * @param sparseShape 1-D.  The `shape` of the minibatch `SparseTensor`.
-   * @return a new instance of SerializeManySparse
-   */
-  public <T extends TType> SerializeManySparse<TString> serializeManySparse(
-      Operand<TInt64> sparseIndices, Operand<T> sparseValues, Operand<TInt64> sparseShape) {
-    return SerializeManySparse.create(scope, sparseIndices, sparseValues, sparseShape);
-  }
-
-  /**
-   * Serialize an `N`-minibatch `SparseTensor` into an `[N, 3]` `Tensor` object.
-   *  <p>
-   *  The `SparseTensor` must have rank `R` greater than 1, and the first dimension
-   *  is treated as the minibatch dimension.  Elements of the `SparseTensor`
-   *  must be sorted in increasing order of this first dimension.  The serialized
-   *  `SparseTensor` objects going into each row of `serialized_sparse` will have
-   *  rank `R-1`.
-   *  <p>
-   *  The minibatch size `N` is extracted from `sparse_shape[0]`.
-   *
-   * @param <U> data type for {@code serializedSparse()} output
-   * @param sparseIndices 2-D.  The `indices` of the minibatch `SparseTensor`.
-   * @param sparseValues 1-D.  The `values` of the minibatch `SparseTensor`.
-   * @param sparseShape 1-D.  The `shape` of the minibatch `SparseTensor`.
-   * @param outType The `dtype` to use for serialization; the supported types are `string`
-   *  (default) and `variant`.
-   * @return a new instance of SerializeManySparse
-   */
-  public <U extends TType, T extends TType> SerializeManySparse<U> serializeManySparse(
-      Operand<TInt64> sparseIndices, Operand<T> sparseValues, Operand<TInt64> sparseShape,
-      DataType<U> outType) {
-    return SerializeManySparse.create(scope, sparseIndices, sparseValues, sparseShape, outType);
-  }
-
-  /**
-   * Serialize a `SparseTensor` into a `[3]` `Tensor` object.
-   *
-   * @param <U> data type for {@code serializedSparse()} output
-   * @param sparseIndices 2-D.  The `indices` of the `SparseTensor`.
-   * @param sparseValues 1-D.  The `values` of the `SparseTensor`.
-   * @param sparseShape 1-D.  The `shape` of the `SparseTensor`.
-   * @return a new instance of SerializeSparse
-   */
-  public <T extends TType> SerializeSparse<TString> serializeSparse(Operand<TInt64> sparseIndices,
-      Operand<T> sparseValues, Operand<TInt64> sparseShape) {
-    return SerializeSparse.create(scope, sparseIndices, sparseValues, sparseShape);
-  }
-
-  /**
-   * Serialize a `SparseTensor` into a `[3]` `Tensor` object.
-   *
-   * @param <U> data type for {@code serializedSparse()} output
-   * @param sparseIndices 2-D.  The `indices` of the `SparseTensor`.
-   * @param sparseValues 1-D.  The `values` of the `SparseTensor`.
-   * @param sparseShape 1-D.  The `shape` of the `SparseTensor`.
-   * @param outType The `dtype` to use for serialization; the supported types are `string`
-   *  (default) and `variant`.
-   * @return a new instance of SerializeSparse
-   */
-  public <U extends TType, T extends TType> SerializeSparse<U> serializeSparse(
-      Operand<TInt64> sparseIndices, Operand<T> sparseValues, Operand<TInt64> sparseShape,
-      DataType<U> outType) {
-    return SerializeSparse.create(scope, sparseIndices, sparseValues, sparseShape, outType);
-  }
-
-  /**
-   * Transforms a Tensor into a serialized TensorProto proto.
-   *
-   * @param tensor A Tensor of type `T`.
-   * @return a new instance of SerializeTensor
-   */
-  public <T extends TType> SerializeTensor serializeTensor(Operand<T> tensor) {
-    return SerializeTensor.create(scope, tensor);
-  }
-
-  /**
-   * Generate a sharded filename. The filename is printf formatted as
-   *  <p>
-   *     %s-%05d-of-%05d, basename, shard, num_shards.
-   *
-   * @param basename
-   * @param shard
-   * @param numShards
-   * @return a new instance of ShardedFilename
-   */
-  public ShardedFilename shardedFilename(Operand<TString> basename, Operand<TInt32> shard,
-      Operand<TInt32> numShards) {
-    return ShardedFilename.create(scope, basename, shard, numShards);
-  }
-
-  /**
-   * Generate a glob pattern matching all sharded file names.
-   *
-   * @param basename
-   * @param numShards
-   * @return a new instance of ShardedFilespec
-   */
-  public ShardedFilespec shardedFilespec(Operand<TString> basename, Operand<TInt32> numShards) {
-    return ShardedFilespec.create(scope, basename, numShards);
-  }
-
-  /**
-   * A Reader that outputs the lines of a file delimited by '\n'.
-   *
-   * @param options carries optional attributes values
-   * @return a new instance of TextLineReader
-   */
-  public TextLineReader textLineReader(TextLineReader.Options... options) {
-    return TextLineReader.create(scope, options);
-  }
-
-  /**
-   * A Reader that outputs the records from a TensorFlow Records file.
-   *
-   * @param options carries optional attributes values
-   * @return a new instance of TfRecordReader
-   */
-  public TfRecordReader tfRecordReader(TfRecordReader.Options... options) {
-    return TfRecordReader.create(scope, options);
-  }
-
-  /**
-   * A Reader that outputs the entire contents of a file as a value.
-   *  <p>
-   *  To use, enqueue filenames in a Queue.  The output of ReaderRead will
-   *  be a filename (key) and the contents of that file (value).
-   *
-   * @param options carries optional attributes values
-   * @return a new instance of WholeFileReader
-   */
-  public WholeFileReader wholeFileReader(WholeFileReader.Options... options) {
-    return WholeFileReader.create(scope, options);
-  }
-
-  /**
-   * Writes contents to the file at input filename. Creates file and recursively
-   *  <p>
-   *  creates directory if not existing.
-   *
-   * @param filename scalar. The name of the file to which we write the contents.
-   * @param contents scalar. The content to be written to the output file.
-   * @return a new instance of WriteFile
-   */
-  public WriteFile writeFile(Operand<TString> filename, Operand<TString> contents) {
-    return WriteFile.create(scope, filename, contents);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/LinalgOps.java b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/LinalgOps.java
deleted file mode 100644
index b2242f1068e..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/LinalgOps.java
+++ /dev/null
@@ -1,1609 +0,0 @@
-// Copyright 2020 The TensorFlow Authors. All Rights Reserved.
-//
-// Licensed under the Apache License, Version 2.0 (the "License");
-// you may not use this file except in compliance with the License.
-// You may obtain a copy of the License at
-//
-//     http://www.apache.org/licenses/LICENSE-2.0
-//
-// Unless required by applicable law or agreed to in writing, software
-// distributed under the License is distributed on an "AS IS" BASIS,
-// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-// See the License for the specific language governing permissions and
-// limitations under the License.
-// ==============================================================================
-//
-// This class has been generated, DO NOT EDIT!
-//
-package org.tensorflow.op;
-
-import org.tensorflow.DataType;
-import org.tensorflow.Operand;
-import org.tensorflow.op.linalg.BandPart;
-import org.tensorflow.op.linalg.BatchCholesky;
-import org.tensorflow.op.linalg.BatchCholeskyGrad;
-import org.tensorflow.op.linalg.BatchMatrixBandPart;
-import org.tensorflow.op.linalg.BatchMatrixDeterminant;
-import org.tensorflow.op.linalg.BatchMatrixDiag;
-import org.tensorflow.op.linalg.BatchMatrixDiagPart;
-import org.tensorflow.op.linalg.BatchMatrixInverse;
-import org.tensorflow.op.linalg.BatchMatrixSetDiag;
-import org.tensorflow.op.linalg.BatchMatrixSolve;
-import org.tensorflow.op.linalg.BatchMatrixSolveLs;
-import org.tensorflow.op.linalg.BatchMatrixTriangularSolve;
-import org.tensorflow.op.linalg.BatchSelfAdjointEig;
-import org.tensorflow.op.linalg.BatchSvd;
-import org.tensorflow.op.linalg.Cholesky;
-import org.tensorflow.op.linalg.CholeskyGrad;
-import org.tensorflow.op.linalg.ConjugateTranspose;
-import org.tensorflow.op.linalg.Cross;
-import org.tensorflow.op.linalg.Det;
-import org.tensorflow.op.linalg.Eig;
-import org.tensorflow.op.linalg.Einsum;
-import org.tensorflow.op.linalg.EuclideanNorm;
-import org.tensorflow.op.linalg.Inv;
-import org.tensorflow.op.linalg.LoadAndRemapMatrix;
-import org.tensorflow.op.linalg.LogMatrixDeterminant;
-import org.tensorflow.op.linalg.Lu;
-import org.tensorflow.op.linalg.MatMul;
-import org.tensorflow.op.linalg.MatrixDiag;
-import org.tensorflow.op.linalg.MatrixDiagPart;
-import org.tensorflow.op.linalg.MatrixDiagPartV3;
-import org.tensorflow.op.linalg.MatrixDiagV3;
-import org.tensorflow.op.linalg.MatrixSetDiag;
-import org.tensorflow.op.linalg.MatrixSolveLs;
-import org.tensorflow.op.linalg.Qr;
-import org.tensorflow.op.linalg.QuantizedMatMul;
-import org.tensorflow.op.linalg.SelfAdjointEig;
-import org.tensorflow.op.linalg.Solve;
-import org.tensorflow.op.linalg.Sqrtm;
-import org.tensorflow.op.linalg.Svd;
-import org.tensorflow.op.linalg.TensorDiag;
-import org.tensorflow.op.linalg.TensorDiagPart;
-import org.tensorflow.op.linalg.Transpose;
-import org.tensorflow.op.linalg.TriangularSolve;
-import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.TFloat64;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.TString;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * An API for building {@code linalg} operations as {@link Op Op}s
- *
- * @see {@link Ops}
- */
-public final class LinalgOps {
-  private final Scope scope;
-
-  LinalgOps(Scope scope) {
-    this.scope = scope;
-  }
-
-  /**
-   * Copy a tensor setting everything outside a central band in each innermost matrix to zero.
-   *  <p>
-   *  The `band` part is computed as follows:
-   *  Assume `input` has `k` dimensions `[I, J, K, ..., M, N]`, then the output is a
-   *  tensor with the same shape where
-   *  <p>
-   *  `band[i, j, k, ..., m, n] = in_band(m, n) * input[i, j, k, ..., m, n]`.
-   *  <p>
-   *  The indicator function
-   *  <p>
-   *  `in_band(m, n) = (num_lower < 0 || (m-n) <= num_lower)) &&
-   *                   (num_upper < 0 || (n-m) <= num_upper)`.
-   *  <p>
-   *  For example:
-   *  <pre>{@code
-   *  # if 'input' is [[ 0,  1,  2, 3]
-   *                   [-1,  0,  1, 2]
-   *                   [-2, -1,  0, 1]
-   *                   [-3, -2, -1, 0]],
-   *
-   *  tf.matrix_band_part(input, 1, -1) ==> [[ 0,  1,  2, 3]
-   *                                         [-1,  0,  1, 2]
-   *                                         [ 0, -1,  0, 1]
-   *                                         [ 0,  0, -1, 0]],
-   *
-   *  tf.matrix_band_part(input, 2, 1) ==> [[ 0,  1,  0, 0]
-   *                                        [-1,  0,  1, 0]
-   *                                        [-2, -1,  0, 1]
-   *                                        [ 0, -2, -1, 0]]
-   *  }</pre>
-   *  Useful special cases:
-   *  <pre>{@code
-   *   tf.matrix_band_part(input, 0, -1) ==> Upper triangular part.
-   *   tf.matrix_band_part(input, -1, 0) ==> Lower triangular part.
-   *   tf.matrix_band_part(input, 0, 0) ==> Diagonal.
-   *  }</pre>
-   *
-   * @param <T> data type for {@code band()} output
-   * @param input Rank `k` tensor.
-   * @param numLower 0-D tensor. Number of subdiagonals to keep. If negative, keep entire
-   *  lower triangle.
-   * @param numUpper 0-D tensor. Number of superdiagonals to keep. If negative, keep
-   *  entire upper triangle.
-   * @return a new instance of BandPart
-   */
-  public <T extends TType, U extends TNumber> BandPart<T> bandPart(Operand<T> input,
-      Operand<U> numLower, Operand<U> numUpper) {
-    return BandPart.create(scope, input, numLower, numUpper);
-  }
-
-  /**
-   *
-   * @param <T> data type for {@code output()} output
-   * @param input
-   * @return a new instance of BatchCholesky
-   */
-  public <T extends TNumber> BatchCholesky<T> batchCholesky(Operand<T> input) {
-    return BatchCholesky.create(scope, input);
-  }
-
-  /**
-   *
-   * @param <T> data type for {@code output()} output
-   * @param l
-   * @param grad
-   * @return a new instance of BatchCholeskyGrad
-   */
-  public <T extends TNumber> BatchCholeskyGrad<T> batchCholeskyGrad(Operand<T> l, Operand<T> grad) {
-    return BatchCholeskyGrad.create(scope, l, grad);
-  }
-
-  /**
-   *
-   * @param <T> data type for {@code band()} output
-   * @param input
-   * @param numLower
-   * @param numUpper
-   * @return a new instance of BatchMatrixBandPart
-   */
-  public <T extends TType> BatchMatrixBandPart<T> batchMatrixBandPart(Operand<T> input,
-      Operand<TInt64> numLower, Operand<TInt64> numUpper) {
-    return BatchMatrixBandPart.create(scope, input, numLower, numUpper);
-  }
-
-  /**
-   *
-   * @param <T> data type for {@code output()} output
-   * @param input
-   * @return a new instance of BatchMatrixDeterminant
-   */
-  public <T extends TType> BatchMatrixDeterminant<T> batchMatrixDeterminant(Operand<T> input) {
-    return BatchMatrixDeterminant.create(scope, input);
-  }
-
-  /**
-   *
-   * @param <T> data type for {@code output()} output
-   * @param diagonal
-   * @return a new instance of BatchMatrixDiag
-   */
-  public <T extends TType> BatchMatrixDiag<T> batchMatrixDiag(Operand<T> diagonal) {
-    return BatchMatrixDiag.create(scope, diagonal);
-  }
-
-  /**
-   *
-   * @param <T> data type for {@code diagonal()} output
-   * @param input
-   * @return a new instance of BatchMatrixDiagPart
-   */
-  public <T extends TType> BatchMatrixDiagPart<T> batchMatrixDiagPart(Operand<T> input) {
-    return BatchMatrixDiagPart.create(scope, input);
-  }
-
-  /**
-   *
-   * @param <T> data type for {@code output()} output
-   * @param input
-   * @param options carries optional attributes values
-   * @return a new instance of BatchMatrixInverse
-   */
-  public <T extends TNumber> BatchMatrixInverse<T> batchMatrixInverse(Operand<T> input,
-      BatchMatrixInverse.Options... options) {
-    return BatchMatrixInverse.create(scope, input, options);
-  }
-
-  /**
-   *
-   * @param <T> data type for {@code output()} output
-   * @param input
-   * @param diagonal
-   * @return a new instance of BatchMatrixSetDiag
-   */
-  public <T extends TType> BatchMatrixSetDiag<T> batchMatrixSetDiag(Operand<T> input,
-      Operand<T> diagonal) {
-    return BatchMatrixSetDiag.create(scope, input, diagonal);
-  }
-
-  /**
-   *
-   * @param <T> data type for {@code output()} output
-   * @param matrix
-   * @param rhs
-   * @param options carries optional attributes values
-   * @return a new instance of BatchMatrixSolve
-   */
-  public <T extends TNumber> BatchMatrixSolve<T> batchMatrixSolve(Operand<T> matrix, Operand<T> rhs,
-      BatchMatrixSolve.Options... options) {
-    return BatchMatrixSolve.create(scope, matrix, rhs, options);
-  }
-
-  /**
-   *
-   * @param <T> data type for {@code output()} output
-   * @param matrix
-   * @param rhs
-   * @param l2Regularizer
-   * @param options carries optional attributes values
-   * @return a new instance of BatchMatrixSolveLs
-   */
-  public <T extends TNumber> BatchMatrixSolveLs<T> batchMatrixSolveLs(Operand<T> matrix,
-      Operand<T> rhs, Operand<TFloat64> l2Regularizer, BatchMatrixSolveLs.Options... options) {
-    return BatchMatrixSolveLs.create(scope, matrix, rhs, l2Regularizer, options);
-  }
-
-  /**
-   *
-   * @param <T> data type for {@code output()} output
-   * @param matrix
-   * @param rhs
-   * @param options carries optional attributes values
-   * @return a new instance of BatchMatrixTriangularSolve
-   */
-  public <T extends TNumber> BatchMatrixTriangularSolve<T> batchMatrixTriangularSolve(
-      Operand<T> matrix, Operand<T> rhs, BatchMatrixTriangularSolve.Options... options) {
-    return BatchMatrixTriangularSolve.create(scope, matrix, rhs, options);
-  }
-
-  /**
-   *
-   * @param <T> data type for {@code e()} output
-   * @param input
-   * @param options carries optional attributes values
-   * @return a new instance of BatchSelfAdjointEig
-   */
-  public <T extends TNumber> BatchSelfAdjointEig<T> batchSelfAdjointEig(Operand<T> input,
-      BatchSelfAdjointEig.Options... options) {
-    return BatchSelfAdjointEig.create(scope, input, options);
-  }
-
-  /**
-   *
-   * @param <T> data type for {@code s()} output
-   * @param input
-   * @param options carries optional attributes values
-   * @return a new instance of BatchSvd
-   */
-  public <T extends TType> BatchSvd<T> batchSvd(Operand<T> input, BatchSvd.Options... options) {
-    return BatchSvd.create(scope, input, options);
-  }
-
-  /**
-   * Computes the Cholesky decomposition of one or more square matrices.
-   *  <p>
-   *  The input is a tensor of shape `[..., M, M]` whose inner-most 2 dimensions
-   *  form square matrices.
-   *  <p>
-   *  The input has to be symmetric and positive definite. Only the lower-triangular
-   *  part of the input will be used for this operation. The upper-triangular part
-   *  will not be read.
-   *  <p>
-   *  The output is a tensor of the same shape as the input
-   *  containing the Cholesky decompositions for all input submatrices `[..., :, :]`.
-   *  <p>
-   *  <b>Note</b>: The gradient computation on GPU is faster for large matrices but
-   *  not for large batch dimensions when the submatrices are small. In this
-   *  case it might be faster to use the CPU.
-   *
-   * @param <T> data type for {@code output()} output
-   * @param input Shape is `[..., M, M]`.
-   * @return a new instance of Cholesky
-   */
-  public <T extends TType> Cholesky<T> cholesky(Operand<T> input) {
-    return Cholesky.create(scope, input);
-  }
-
-  /**
-   * Computes the reverse mode backpropagated gradient of the Cholesky algorithm.
-   *  <p>
-   *  For an explanation see "Differentiation of the Cholesky algorithm" by
-   *  Iain Murray http://arxiv.org/abs/1602.07527.
-   *
-   * @param <T> data type for {@code output()} output
-   * @param l Output of batch Cholesky algorithm l = cholesky(A). Shape is `[..., M, M]`.
-   *  Algorithm depends only on lower triangular part of the innermost matrices of
-   *  this tensor.
-   * @param grad df/dl where f is some scalar function. Shape is `[..., M, M]`.
-   *  Algorithm depends only on lower triangular part of the innermost matrices of
-   *  this tensor.
-   * @return a new instance of CholeskyGrad
-   */
-  public <T extends TNumber> CholeskyGrad<T> choleskyGrad(Operand<T> l, Operand<T> grad) {
-    return CholeskyGrad.create(scope, l, grad);
-  }
-
-  /**
-   * Shuffle dimensions of x according to a permutation and conjugate the result.
-   *  <p>
-   *  The output `y` has the same rank as `x`. The shapes of `x` and `y` satisfy:
-   *    `y.shape[i] == x.shape[perm[i]] for i in [0, 1, ..., rank(x) - 1]`
-   *    `y[i,j,k,...,s,t,u] == conj(x[perm[i], perm[j], perm[k],...,perm[s], perm[t], perm[u]])`
-   *
-   * @param <T> data type for {@code y()} output
-   * @param x
-   * @param perm
-   * @return a new instance of ConjugateTranspose
-   */
-  public <T extends TType, U extends TNumber> ConjugateTranspose<T> conjugateTranspose(Operand<T> x,
-      Operand<U> perm) {
-    return ConjugateTranspose.create(scope, x, perm);
-  }
-
-  /**
-   * Compute the pairwise cross product.
-   *  <p>
-   *  `a` and `b` must be the same shape; they can either be simple 3-element vectors,
-   *  or any shape where the innermost dimension is 3. In the latter case, each pair
-   *  of corresponding 3-element vectors is cross-multiplied independently.
-   *
-   * @param <T> data type for {@code product()} output
-   * @param a A tensor containing 3-element vectors.
-   * @param b Another tensor, of same type and shape as `a`.
-   * @return a new instance of Cross
-   */
-  public <T extends TNumber> Cross<T> cross(Operand<T> a, Operand<T> b) {
-    return Cross.create(scope, a, b);
-  }
-
-  /**
-   * Computes the determinant of one or more square matrices.
-   *  <p>
-   *  The input is a tensor of shape `[..., M, M]` whose inner-most 2 dimensions
-   *  form square matrices. The output is a tensor containing the determinants
-   *  for all input submatrices `[..., :, :]`.
-   *
-   * @param <T> data type for {@code output()} output
-   * @param input Shape is `[..., M, M]`.
-   * @return a new instance of Det
-   */
-  public <T extends TType> Det<T> det(Operand<T> input) {
-    return Det.create(scope, input);
-  }
-
-  /**
-   * Computes the eigen decomposition of one or more square matrices.
-   *  <p>
-   *  Computes the eigenvalues and (optionally) right eigenvectors of each inner matrix in
-   *  `input` such that `input[..., :, :] = v[..., :, :] * diag(e[..., :])`. The eigenvalues
-   *  are sorted in non-decreasing order.
-   *  <pre>{@code
-   *  # a is a tensor.
-   *  # e is a tensor of eigenvalues.
-   *  # v is a tensor of eigenvectors.
-   *  e, v = eig(a)
-   *  e = eig(a, compute_v=False)
-   *  }</pre>
-   *
-   * @param <U> data type for {@code e()} output
-   * @param input `Tensor` input of shape `[N, N]`.
-   * @param Tout
-   * @param options carries optional attributes values
-   * @return a new instance of Eig
-   */
-  public <U extends TType, T extends TType> Eig<U> eig(Operand<T> input, DataType<U> Tout,
-      Eig.Options... options) {
-    return Eig.create(scope, input, Tout, options);
-  }
-
-  /**
-   * Tensor contraction according to Einstein summation convention.
-   *  <p>
-   *  Implements generalized Tensor contraction and reduction. Each input Tensor must
-   *  have a corresponding input subscript appearing in the comma-separated left-hand
-   *  side of the equation. The right-hand side of the equation consists of the
-   *  output subscript. The input subscripts and the output subscript should consist
-   *  of zero or more named axis labels and at most one ellipsis (`...`).
-   *  <p>
-   *  The named axis labels may be any single character other than those having
-   *  special meaning, namely `,.->`. The behavior of this Op is undefined if it
-   *  receives an ill-formatted equation; since the validation is done at
-   *  graph-building time, we omit format validation checks at runtime.
-   *  <p>
-   *  Note: This Op is <i>not</i> intended to be called by the user; instead users should
-   *  call `tf.einsum` directly. It is a hidden Op used by `tf.einsum`.
-   *  <p>
-   *  Operations are applied to the input(s) according to the following rules:
-   *  <p>
-   *   (a) Generalized Diagonals: For input dimensions corresponding to axis labels
-   *       appearing more than once in the same input subscript, we take the
-   *       generalized (`k`-dimensional) diagonal.
-   *       For example, in the equation `iii->i` with input shape `[3, 3, 3]`, the
-   *       generalized diagonal would consist of `3` elements at indices `(0, 0, 0)`,
-   *       `(1, 1, 1)` and `(2, 2, 2)` to create a Tensor of shape `[3]`.
-   *  <p>
-   *   (b) Reduction: Axes corresponding to labels appearing only in one input
-   *       subscript but not in the output subscript are summed over prior to Tensor
-   *       contraction.
-   *       For example, in the equation `ab,bc->b`, the axis labels `a` and `c` are
-   *       the reduction axis labels.
-   *  <p>
-   *   (c) Batch Dimensions: Axes corresponding to labels appearing in each of the
-   *       input subscripts and also in the output subscript make up the batch
-   *       dimensions in Tensor contraction. Unnamed axis labels corresponding to
-   *       ellipsis (`...`) also correspond to batch dimensions.
-   *       For example, for the equation denoting batch matrix multiplication,
-   *       `bij,bjk->bik`, the axis label `b` corresponds to a batch dimension.
-   *  <p>
-   *   (d) Contraction: In case of binary einsum, axes corresponding to labels
-   *       appearing in two different inputs (and not in the output) are contracted
-   *       against each other.
-   *       Considering the batch matrix multiplication equation again
-   *       (`bij,bjk->bik`), the contracted axis label is `j`.
-   *  <p>
-   *   (e) Expand Diagonal: If the output subscripts contain repeated (explicit) axis
-   *       labels, the opposite operation of (a) is applied. For example, in the
-   *       equation `i->iii`, and input shape `[3]`, the output of shape `[3, 3, 3]`
-   *       are all zeros, except for the (generalized) diagonal which is populated
-   *       with values from the input.
-   *       Note: This operation is not supported by `np.einsum` or `tf.einsum`; it is
-   *       provided to enable computing the symbolic gradient of `tf.einsum`.
-   *  <p>
-   *  The output subscripts must contain only labels appearing in at least one of the
-   *  input subscripts. Furthermore, all dimensions mapping to the same axis label
-   *  must be equal.
-   *  <p>
-   *  Any of the input and output subscripts may contain at most a single ellipsis
-   *  (`...`). These ellipsis are mapped against dimensions not corresponding to any
-   *  named axis label. If two inputs contain ellipsis, then they are broadcasted
-   *  according to standard NumPy broadcasting
-   *  [rules](http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html).
-   *  <p>
-   *  The broadcasted dimensions are placed in the corresponding location of the
-   *  ellipsis in the output subscript. If the broadcasted dimensions are non-empty
-   *  and the output subscripts do not contain ellipsis, then an InvalidArgument error
-   *  is raised.
-   *  <p>
-   *
-   * @compatibility(numpy) Similar to [`numpy.einsum`](https://docs.scipy.org/doc/numpy/reference/generated/numpy.einsum.html).
-   *  <p>
-   *  Comparison with `numpy.einsum`:
-   *  <p>
-   * This Op only supports unary and binary forms of `numpy.einsum`.
-   * This Op does not support implicit form. (i.e. equations without `->`).
-   * This Op also supports repeated indices in the output subscript, which is not
-   *     supported by `numpy.einsum`.
-   * @end_compatibility
-   * @param <T> data type for {@code output()} output
-   * @param inputs List of 1 or 2 Tensors.
-   * @param equation String describing the Einstein Summation operation; in the format of np.einsum.
-   * @return a new instance of Einsum
-   */
-  public <T extends TType> Einsum<T> einsum(Iterable<Operand<T>> inputs, String equation) {
-    return Einsum.create(scope, inputs, equation);
-  }
-
-  /**
-   * Computes the euclidean norm of elements across dimensions of a tensor.
-   *  <p>
-   *  Reduces `input` along the dimensions given in `axis`. Unless
-   *  `keep_dims` is true, the rank of the tensor is reduced by 1 for each entry in
-   *  `axis`. If `keep_dims` is true, the reduced dimensions are
-   *  retained with length 1.
-   *
-   * @param <T> data type for {@code output()} output
-   * @param input The tensor to reduce.
-   * @param axis The dimensions to reduce. Must be in the range
-   *  `[-rank(input), rank(input))`.
-   * @param options carries optional attributes values
-   * @return a new instance of EuclideanNorm
-   */
-  public <T extends TType, U extends TNumber> EuclideanNorm<T> euclideanNorm(Operand<T> input,
-      Operand<U> axis, EuclideanNorm.Options... options) {
-    return EuclideanNorm.create(scope, input, axis, options);
-  }
-
-  /**
-   * Computes the inverse of one or more square invertible matrices or their
-   *  <p>
-   *  adjoints (conjugate transposes).
-   *  <p>
-   *  The input is a tensor of shape `[..., M, M]` whose inner-most 2 dimensions
-   *  form square matrices. The output is a tensor of the same shape as the input
-   *  containing the inverse for all input submatrices `[..., :, :]`.
-   *  <p>
-   *  The op uses LU decomposition with partial pivoting to compute the inverses.
-   *  <p>
-   *  If a matrix is not invertible there is no guarantee what the op does. It
-   *  may detect the condition and raise an exception or it may simply return a
-   *  garbage result.
-   *
-   * @param <T> data type for {@code output()} output
-   * @param input Shape is `[..., M, M]`.
-   * @param options carries optional attributes values
-   * @return a new instance of Inv
-   */
-  public <T extends TType> Inv<T> inv(Operand<T> input, Inv.Options... options) {
-    return Inv.create(scope, input, options);
-  }
-
-  /**
-   * Loads a 2-D (matrix) `Tensor` with name `old_tensor_name` from the checkpoint
-   *  <p>
-   *  at `ckpt_path` and potentially reorders its rows and columns using the
-   *  specified remappings.
-   *  <p>
-   *  Most users should use one of the wrapper initializers (such as
-   *  `tf.contrib.framework.load_and_remap_matrix_initializer`) instead of this
-   *  function directly.
-   *  <p>
-   *  The remappings are 1-D tensors with the following properties:
-   *  <ul>
-   *  <li>
-   *  `row_remapping` must have exactly `num_rows` entries. Row `i` of the output
-   *    matrix will be initialized from the row corresponding to index
-   *    `row_remapping[i]` in the old `Tensor` from the checkpoint.
-   *  </li>
-   *  <li>
-   *  `col_remapping` must have either 0 entries (indicating that no column
-   *    reordering is needed) or `num_cols` entries. If specified, column `j` of the
-   *    output matrix will be initialized from the column corresponding to index
-   *    `col_remapping[j]` in the old `Tensor` from the checkpoint.
-   *  </li>
-   *  <li>
-   *  A value of -1 in either of the remappings signifies a "missing" entry. In that
-   *    case, values from the `initializing_values` tensor will be used to fill that
-   *    missing row or column. If `row_remapping` has `r` missing entries and
-   *    `col_remapping` has `c` missing entries, then the following condition must be
-   *    true:
-   *  </li>
-   *  </ul>
-   *  `(r * num_cols) + (c * num_rows) - (r * c) == len(initializing_values)`
-   *  <p>
-   *  The remapping tensors can be generated using the GenerateVocabRemapping op.
-   *  <p>
-   *  As an example, with row_remapping = [1, 0, -1], col_remapping = [0, 2, -1],
-   *  initializing_values = [0.5, -0.5, 0.25, -0.25, 42], and w(i, j) representing
-   *  the value from row i, column j of the old tensor in the checkpoint, the output
-   *  matrix will look like the following:
-   *  <p>
-   *  [[w(1, 0),  w(1, 2),  0.5],
-   *   [w(0, 0),  w(0, 2), -0.5],
-   *   [0.25,    -0.25,      42]]
-   *
-   * @param ckptPath Path to the TensorFlow checkpoint (version 2, `TensorBundle`) from
-   *  which the old matrix `Tensor` will be loaded.
-   * @param oldTensorName Name of the 2-D `Tensor` to load from checkpoint.
-   * @param rowRemapping An int `Tensor` of row remappings (generally created by
-   *  `generate_vocab_remapping`).  Even if no row remapping is needed, this must
-   *  still be an index-valued Tensor (e.g. [0, 1, 2, ...]), or a shifted
-   *  index-valued `Tensor` (e.g. [8, 9, 10, ...], for partitioned `Variables`).
-   * @param colRemapping An int `Tensor` of column remappings (generally created by
-   *  `generate_vocab_remapping`).  May be a size-0 `Tensor` if only row remapping
-   *  is to be done (e.g. column ordering is the same).
-   * @param initializingValues A float `Tensor` containing  values to fill in for cells
-   *  in the output matrix that are not loaded from the checkpoint. Length must be
-   *  exactly the same as the number of missing / new cells.
-   * @param numRows Number of rows (length of the 1st dimension) in the output matrix.
-   * @param numCols Number of columns (length of the 2nd dimension) in the output matrix.
-   * @param options carries optional attributes values
-   * @return a new instance of LoadAndRemapMatrix
-   */
-  public LoadAndRemapMatrix loadAndRemapMatrix(Operand<TString> ckptPath,
-      Operand<TString> oldTensorName, Operand<TInt64> rowRemapping, Operand<TInt64> colRemapping,
-      Operand<TFloat32> initializingValues, Long numRows, Long numCols,
-      LoadAndRemapMatrix.Options... options) {
-    return LoadAndRemapMatrix.create(scope, ckptPath, oldTensorName, rowRemapping, colRemapping, initializingValues, numRows, numCols, options);
-  }
-
-  /**
-   * Computes the sign and the log of the absolute value of the determinant of
-   *  <p>
-   *  one or more square matrices.
-   *  <p>
-   *  The input is a tensor of shape `[N, M, M]` whose inner-most 2 dimensions
-   *  form square matrices. The outputs are two tensors containing the signs and
-   *  absolute values of the log determinants for all N input submatrices
-   *  `[..., :, :]` such that the determinant = sign*exp(log_abs_determinant).
-   *  The log_abs_determinant is computed as det(P)*sum(log(diag(LU))) where LU
-   *  is the LU decomposition of the input and P is the corresponding
-   *  permutation matrix.
-   *
-   * @param <T> data type for {@code sign()} output
-   * @param input Shape is `[N, M, M]`.
-   * @return a new instance of LogMatrixDeterminant
-   */
-  public <T extends TType> LogMatrixDeterminant<T> logMatrixDeterminant(Operand<T> input) {
-    return LogMatrixDeterminant.create(scope, input);
-  }
-
-  /**
-   * Computes the LU decomposition of one or more square matrices.
-   *  <p>
-   *  The input is a tensor of shape `[..., M, M]` whose inner-most 2 dimensions
-   *  form square matrices.
-   *  <p>
-   *  The input has to be invertible.
-   *  <p>
-   *  The output consists of two tensors LU and P containing the LU decomposition
-   *  of all input submatrices `[..., :, :]`. LU encodes the lower triangular and
-   *  upper triangular factors.
-   *  <p>
-   *  For each input submatrix of shape `[M, M]`, L is a lower triangular matrix of
-   *  shape `[M, M]` with unit diagonal whose entries correspond to the strictly lower
-   *  triangular part of LU. U is a upper triangular matrix of shape `[M, M]` whose
-   *  entries correspond to the upper triangular part, including the diagonal, of LU.
-   *  <p>
-   *  P represents a permutation matrix encoded as a list of indices each between `0`
-   *  and `M-1`, inclusive. If P_mat denotes the permutation matrix corresponding to
-   *  P, then the L, U and P satisfies P_mat * input = L * U.
-   *
-   * @param <T> data type for {@code lu()} output
-   * @param <U> data type for {@code p()} output
-   * @param input A tensor of shape `[..., M, M]` whose inner-most 2 dimensions form matrices of
-   *  size `[M, M]`.
-   * @return a new instance of Lu
-   */
-  public <T extends TType> Lu<T, TInt32> lu(Operand<T> input) {
-    return Lu.create(scope, input);
-  }
-
-  /**
-   * Computes the LU decomposition of one or more square matrices.
-   *  <p>
-   *  The input is a tensor of shape `[..., M, M]` whose inner-most 2 dimensions
-   *  form square matrices.
-   *  <p>
-   *  The input has to be invertible.
-   *  <p>
-   *  The output consists of two tensors LU and P containing the LU decomposition
-   *  of all input submatrices `[..., :, :]`. LU encodes the lower triangular and
-   *  upper triangular factors.
-   *  <p>
-   *  For each input submatrix of shape `[M, M]`, L is a lower triangular matrix of
-   *  shape `[M, M]` with unit diagonal whose entries correspond to the strictly lower
-   *  triangular part of LU. U is a upper triangular matrix of shape `[M, M]` whose
-   *  entries correspond to the upper triangular part, including the diagonal, of LU.
-   *  <p>
-   *  P represents a permutation matrix encoded as a list of indices each between `0`
-   *  and `M-1`, inclusive. If P_mat denotes the permutation matrix corresponding to
-   *  P, then the L, U and P satisfies P_mat * input = L * U.
-   *
-   * @param <T> data type for {@code lu()} output
-   * @param <U> data type for {@code p()} output
-   * @param input A tensor of shape `[..., M, M]` whose inner-most 2 dimensions form matrices of
-   *  size `[M, M]`.
-   * @param outputIdxType
-   * @return a new instance of Lu
-   */
-  public <T extends TType, U extends TNumber> Lu<T, U> lu(Operand<T> input,
-      DataType<U> outputIdxType) {
-    return Lu.create(scope, input, outputIdxType);
-  }
-
-  /**
-   * Multiply the matrix "a" by the matrix "b".
-   *  <p>
-   *  The inputs must be two-dimensional matrices and the inner dimension of
-   *  "a" (after being transposed if transpose_a is true) must match the
-   *  outer dimension of "b" (after being transposed if transposed_b is
-   *  true).
-   *  <p>
-   *  <i>Note</i>: The default kernel implementation for MatMul on GPUs uses
-   *  cublas.
-   *
-   * @param <T> data type for {@code product()} output
-   * @param a
-   * @param b
-   * @param options carries optional attributes values
-   * @return a new instance of MatMul
-   */
-  public <T extends TType> MatMul<T> matMul(Operand<T> a, Operand<T> b, MatMul.Options... options) {
-    return MatMul.create(scope, a, b, options);
-  }
-
-  /**
-   * Returns a batched diagonal tensor with given batched diagonal values.
-   *  <p>
-   *  Returns a tensor with the contents in `diagonal` as `k[0]`-th to `k[1]`-th
-   *  diagonals of a matrix, with everything else padded with `padding`. `num_rows`
-   *  and `num_cols` specify the dimension of the innermost matrix of the output. If
-   *  both are not specified, the op assumes the innermost matrix is square and infers
-   *  its size from `k` and the innermost dimension of `diagonal`. If only one of them
-   *  is specified, the op assumes the unspecified value is the smallest possible
-   *  based on other criteria.
-   *  <p>
-   *  Let `diagonal` have `r` dimensions `[I, J, ..., L, M, N]`. The output tensor has
-   *  rank `r+1` with shape `[I, J, ..., L, M, num_rows, num_cols]` when only one
-   *  diagonal is given (`k` is an integer or `k[0] == k[1]`). Otherwise, it has rank
-   *  `r` with shape `[I, J, ..., L, num_rows, num_cols]`.
-   *  <p>
-   *  The second innermost dimension of `diagonal` has double meaning.
-   *  When `k` is scalar or `k[0] == k[1]`, `M` is part of the batch size
-   *  [I, J, ..., M], and the output tensor is:
-   *  <pre>{@code
-   *  output[i, j, ..., l, m, n]
-   *    = diagonal[i, j, ..., l, n-max(d_upper, 0)] ; if n - m == d_upper
-   *      padding_value                             ; otherwise
-   *  }</pre>
-   *  Otherwise, `M` is treated as the number of diagonals for the matrix in the
-   *  same batch (`M = k[1]-k[0]+1`), and the output tensor is:
-   *  <pre>{@code
-   *  output[i, j, ..., l, m, n]
-   *    = diagonal[i, j, ..., l, diag_index, index_in_diag] ; if k[0] <= d <= k[1]
-   *      padding_value                                     ; otherwise
-   *  }</pre>
-   *  where `d = n - m`, `diag_index = k[1] - d`, and `index_in_diag = n - max(d, 0)`.
-   *  <p>
-   *  For example:
-   *  <pre>{@code
-   *  # The main diagonal.
-   *  diagonal = np.array([[1, 2, 3, 4],            # Input shape: (2, 4)
-   *                       [5, 6, 7, 8]])
-   *  tf.matrix_diag(diagonal) ==> [[[1, 0, 0, 0],  # Output shape: (2, 4, 4)
-   *                                 [0, 2, 0, 0],
-   *                                 [0, 0, 3, 0],
-   *                                 [0, 0, 0, 4]],
-   *                                [[5, 0, 0, 0],
-   *                                 [0, 6, 0, 0],
-   *                                 [0, 0, 7, 0],
-   *                                 [0, 0, 0, 8]]]
-   *
-   *  # A superdiagonal (per batch).
-   *  diagonal = np.array([[1, 2, 3],  # Input shape: (2, 3)
-   *                       [4, 5, 6]])
-   *  tf.matrix_diag(diagonal, k = 1)
-   *    ==> [[[0, 1, 0, 0],  # Output shape: (2, 4, 4)
-   *          [0, 0, 2, 0],
-   *          [0, 0, 0, 3],
-   *          [0, 0, 0, 0]],
-   *         [[0, 4, 0, 0],
-   *          [0, 0, 5, 0],
-   *          [0, 0, 0, 6],
-   *          [0, 0, 0, 0]]]
-   *
-   *  # A band of diagonals.
-   *  diagonals = np.array([[[1, 2, 3],  # Input shape: (2, 2, 3)
-   *                         [4, 5, 0]],
-   *                        [[6, 7, 9],
-   *                         [9, 1, 0]]])
-   *  tf.matrix_diag(diagonals, k = (-1, 0))
-   *    ==> [[[1, 0, 0],  # Output shape: (2, 3, 3)
-   *          [4, 2, 0],
-   *          [0, 5, 3]],
-   *         [[6, 0, 0],
-   *          [9, 7, 0],
-   *          [0, 1, 9]]]
-   *
-   *  # Rectangular matrix.
-   *  diagonal = np.array([1, 2])  # Input shape: (2)
-   *  tf.matrix_diag(diagonal, k = -1, num_rows = 3, num_cols = 4)
-   *    ==> [[0, 0, 0, 0],  # Output shape: (3, 4)
-   *         [1, 0, 0, 0],
-   *         [0, 2, 0, 0]]
-   *
-   *  # Rectangular matrix with inferred num_cols and padding_value = 9.
-   *  tf.matrix_diag(diagonal, k = -1, num_rows = 3, padding_value = 9)
-   *    ==> [[9, 9],  # Output shape: (3, 2)
-   *         [1, 9],
-   *         [9, 2]]
-   *  }</pre>
-   *
-   * @param <T> data type for {@code output()} output
-   * @param diagonal Rank `r`, where `r >= 1`
-   * @param k Diagonal offset(s). Positive value means superdiagonal, 0 refers to the main
-   *  diagonal, and negative value means subdiagonals. `k` can be a single integer
-   *  (for a single diagonal) or a pair of integers specifying the low and high ends
-   *  of a matrix band. `k[0]` must not be larger than `k[1]`.
-   * @param numRows The number of rows of the output matrix. If it is not provided, the op assumes
-   *  the output matrix is a square matrix and infers the matrix size from k and the
-   *  innermost dimension of `diagonal`.
-   * @param numCols The number of columns of the output matrix. If it is not provided, the op
-   *  assumes the output matrix is a square matrix and infers the matrix size from
-   *  k and the innermost dimension of `diagonal`.
-   * @param paddingValue The number to fill the area outside the specified diagonal band with.
-   *  Default is 0.
-   * @return a new instance of MatrixDiag
-   */
-  public <T extends TType> MatrixDiag<T> matrixDiag(Operand<T> diagonal, Operand<TInt32> k,
-      Operand<TInt32> numRows, Operand<TInt32> numCols, Operand<T> paddingValue) {
-    return MatrixDiag.create(scope, diagonal, k, numRows, numCols, paddingValue);
-  }
-
-  /**
-   * Returns the batched diagonal part of a batched tensor.
-   *  <p>
-   *  Returns a tensor with the `k[0]`-th to `k[1]`-th diagonals of the batched
-   *  `input`.
-   *  <p>
-   *  Assume `input` has `r` dimensions `[I, J, ..., L, M, N]`.
-   *  Let `max_diag_len` be the maximum length among all diagonals to be extracted,
-   *  `max_diag_len = min(M + min(k[1], 0), N + min(-k[0], 0))`
-   *  Let `num_diags` be the number of diagonals to extract,
-   *  `num_diags = k[1] - k[0] + 1`.
-   *  <p>
-   *  If `num_diags == 1`, the output tensor is of rank `r - 1` with shape
-   *  `[I, J, ..., L, max_diag_len]` and values:
-   *  <pre>{@code
-   *  diagonal[i, j, ..., l, n]
-   *    = input[i, j, ..., l, n+y, n+x] ; if 0 <= n+y < M and 0 <= n+x < N,
-   *      padding_value                 ; otherwise.
-   *  }</pre>
-   *  where `y = max(-k[1], 0)`, `x = max(k[1], 0)`.
-   *  <p>
-   *  Otherwise, the output tensor has rank `r` with dimensions
-   *  `[I, J, ..., L, num_diags, max_diag_len]` with values:
-   *  <pre>{@code
-   *  diagonal[i, j, ..., l, m, n]
-   *    = input[i, j, ..., l, n+y, n+x] ; if 0 <= n+y < M and 0 <= n+x < N,
-   *      padding_value                 ; otherwise.
-   *  }</pre>
-   *  where `d = k[1] - m`, `y = max(-d, 0)`, and `x = max(d, 0)`.
-   *  <p>
-   *  The input must be at least a matrix.
-   *  <p>
-   *  For example:
-   *  <pre>{@code
-   *  input = np.array([[[1, 2, 3, 4],  # Input shape: (2, 3, 4)
-   *                     [5, 6, 7, 8],
-   *                     [9, 8, 7, 6]],
-   *                    [[5, 4, 3, 2],
-   *                     [1, 2, 3, 4],
-   *                     [5, 6, 7, 8]]])
-   *
-   *  # A main diagonal from each batch.
-   *  tf.matrix_diag_part(input) ==> [[1, 6, 7],  # Output shape: (2, 3)
-   *                                  [5, 2, 7]]
-   *
-   *  # A superdiagonal from each batch.
-   *  tf.matrix_diag_part(input, k = 1)
-   *    ==> [[2, 7, 6],  # Output shape: (2, 3)
-   *         [4, 3, 8]]
-   *
-   *  # A tridiagonal band from each batch.
-   *  tf.matrix_diag_part(input, k = (-1, 1))
-   *    ==> [[[2, 7, 6],  # Output shape: (2, 3, 3)
-   *          [1, 6, 7],
-   *          [5, 8, 0]],
-   *         [[4, 3, 8],
-   *          [5, 2, 7],
-   *          [1, 6, 0]]]
-   *
-   *  # Padding value = 9
-   *  tf.matrix_diag_part(input, k = (1, 3), padding_value = 9)
-   *    ==> [[[4, 9, 9],  # Output shape: (2, 3, 3)
-   *          [3, 8, 9],
-   *          [2, 7, 6]],
-   *         [[2, 9, 9],
-   *          [3, 4, 9],
-   *          [4, 3, 8]]]
-   *  }</pre>
-   *
-   * @param <T> data type for {@code diagonal()} output
-   * @param input Rank `r` tensor where `r >= 2`.
-   * @param k Diagonal offset(s). Positive value means superdiagonal, 0 refers to the main
-   *  diagonal, and negative value means subdiagonals. `k` can be a single integer
-   *  (for a single diagonal) or a pair of integers specifying the low and high ends
-   *  of a matrix band. `k[0]` must not be larger than `k[1]`.
-   * @param paddingValue The value to fill the area outside the specified diagonal band with.
-   *  Default is 0.
-   * @return a new instance of MatrixDiagPart
-   */
-  public <T extends TType> MatrixDiagPart<T> matrixDiagPart(Operand<T> input, Operand<TInt32> k,
-      Operand<T> paddingValue) {
-    return MatrixDiagPart.create(scope, input, k, paddingValue);
-  }
-
-  /**
-   * Returns the batched diagonal part of a batched tensor.
-   *  <p>
-   *  Returns a tensor with the `k[0]`-th to `k[1]`-th diagonals of the batched
-   *  `input`.
-   *  <p>
-   *  Assume `input` has `r` dimensions `[I, J, ..., L, M, N]`.
-   *  Let `max_diag_len` be the maximum length among all diagonals to be extracted,
-   *  `max_diag_len = min(M + min(k[1], 0), N + min(-k[0], 0))`
-   *  Let `num_diags` be the number of diagonals to extract,
-   *  `num_diags = k[1] - k[0] + 1`.
-   *  <p>
-   *  If `num_diags == 1`, the output tensor is of rank `r - 1` with shape
-   *  `[I, J, ..., L, max_diag_len]` and values:
-   *  <pre>{@code
-   *  diagonal[i, j, ..., l, n]
-   *    = input[i, j, ..., l, n+y, n+x] ; if 0 <= n+y < M and 0 <= n+x < N,
-   *      padding_value                 ; otherwise.
-   *  }</pre>
-   *  where `y = max(-k[1], 0)`, `x = max(k[1], 0)`.
-   *  <p>
-   *  Otherwise, the output tensor has rank `r` with dimensions
-   *  `[I, J, ..., L, num_diags, max_diag_len]` with values:
-   *  <pre>{@code
-   *  diagonal[i, j, ..., l, m, n]
-   *    = input[i, j, ..., l, n+y, n+x] ; if 0 <= n+y < M and 0 <= n+x < N,
-   *      padding_value                 ; otherwise.
-   *  }</pre>
-   *  where `d = k[1] - m`, `y = max(-d, 0) - offset`, and `x = max(d, 0) - offset`.
-   *  <p>
-   *  `offset` is zero except when the alignment of the diagonal is to the right.
-   *  <pre>{@code
-   *  offset = max_diag_len - diag_len(d) ; if (`align` in {RIGHT_LEFT, RIGHT_RIGHT}
-   *                                             and `d >= 0`) or
-   *                                           (`align` in {LEFT_RIGHT, RIGHT_RIGHT}
-   *                                             and `d <= 0`)
-   *           0                          ; otherwise
-   *  }</pre>
-   *  where `diag_len(d) = min(cols - max(d, 0), rows + min(d, 0))`.
-   *  <p>
-   *  The input must be at least a matrix.
-   *  <p>
-   *  For example:
-   *  <pre>{@code
-   *  input = np.array([[[1, 2, 3, 4],  # Input shape: (2, 3, 4)
-   *                     [5, 6, 7, 8],
-   *                     [9, 8, 7, 6]],
-   *                    [[5, 4, 3, 2],
-   *                     [1, 2, 3, 4],
-   *                     [5, 6, 7, 8]]])
-   *
-   *  # A main diagonal from each batch.
-   *  tf.matrix_diag_part(input) ==> [[1, 6, 7],  # Output shape: (2, 3)
-   *                                  [5, 2, 7]]
-   *
-   *  # A superdiagonal from each batch.
-   *  tf.matrix_diag_part(input, k = 1)
-   *    ==> [[2, 7, 6],  # Output shape: (2, 3)
-   *         [4, 3, 8]]
-   *
-   *  # A band from each batch.
-   *  tf.matrix_diag_part(input, k = (-1, 2))
-   *    ==> [[[0, 3, 8],  # Output shape: (2, 4, 3)
-   *          [2, 7, 6],
-   *          [1, 6, 7],
-   *          [5, 8, 0]],
-   *         [[0, 3, 4],
-   *          [4, 3, 8],
-   *          [5, 2, 7],
-   *          [1, 6, 0]]]
-   *
-   *  # LEFT_RIGHT alignment.
-   *  tf.matrix_diag_part(input, k = (-1, 2), align="LEFT_RIGHT")
-   *    ==> [[[3, 8, 0],  # Output shape: (2, 4, 3)
-   *          [2, 7, 6],
-   *          [1, 6, 7],
-   *          [0, 5, 8]],
-   *         [[3, 4, 0],
-   *          [4, 3, 8],
-   *          [5, 2, 7],
-   *          [0, 1, 6]]]
-   *
-   *  # max_diag_len can be shorter than the main diagonal.
-   *  tf.matrix_diag_part(input, k = (-2, -1))
-   *    ==> [[[5, 8],
-   *          [9, 0]],
-   *         [[1, 6],
-   *          [5, 0]]]
-   *
-   *  # padding_value = 9
-   *  tf.matrix_diag_part(input, k = (1, 3), padding_value = 9)
-   *    ==> [[[9, 9, 4],  # Output shape: (2, 3, 3)
-   *          [9, 3, 8],
-   *          [2, 7, 6]],
-   *         [[9, 9, 2],
-   *          [9, 3, 4],
-   *          [4, 3, 8]]]
-   *
-   *  }</pre>
-   *
-   * @param <T> data type for {@code diagonal()} output
-   * @param input Rank `r` tensor where `r >= 2`.
-   * @param k Diagonal offset(s). Positive value means superdiagonal, 0 refers to the main
-   *  diagonal, and negative value means subdiagonals. `k` can be a single integer
-   *  (for a single diagonal) or a pair of integers specifying the low and high ends
-   *  of a matrix band. `k[0]` must not be larger than `k[1]`.
-   * @param paddingValue The value to fill the area outside the specified diagonal band with.
-   *  Default is 0.
-   * @param options carries optional attributes values
-   * @return a new instance of MatrixDiagPartV3
-   */
-  public <T extends TType> MatrixDiagPartV3<T> matrixDiagPartV3(Operand<T> input, Operand<TInt32> k,
-      Operand<T> paddingValue, MatrixDiagPartV3.Options... options) {
-    return MatrixDiagPartV3.create(scope, input, k, paddingValue, options);
-  }
-
-  /**
-   * Returns a batched diagonal tensor with given batched diagonal values.
-   *  <p>
-   *  Returns a tensor with the contents in `diagonal` as `k[0]`-th to `k[1]`-th
-   *  diagonals of a matrix, with everything else padded with `padding`. `num_rows`
-   *  and `num_cols` specify the dimension of the innermost matrix of the output. If
-   *  both are not specified, the op assumes the innermost matrix is square and infers
-   *  its size from `k` and the innermost dimension of `diagonal`. If only one of them
-   *  is specified, the op assumes the unspecified value is the smallest possible
-   *  based on other criteria.
-   *  <p>
-   *  Let `diagonal` have `r` dimensions `[I, J, ..., L, M, N]`. The output tensor has
-   *  rank `r+1` with shape `[I, J, ..., L, M, num_rows, num_cols]` when only one
-   *  diagonal is given (`k` is an integer or `k[0] == k[1]`). Otherwise, it has rank
-   *  `r` with shape `[I, J, ..., L, num_rows, num_cols]`.
-   *  <p>
-   *  The second innermost dimension of `diagonal` has double meaning.
-   *  When `k` is scalar or `k[0] == k[1]`, `M` is part of the batch size
-   *  [I, J, ..., M], and the output tensor is:
-   *  <pre>{@code
-   *  output[i, j, ..., l, m, n]
-   *    = diagonal[i, j, ..., l, n-max(d_upper, 0)] ; if n - m == d_upper
-   *      padding_value                             ; otherwise
-   *  }</pre>
-   *  Otherwise, `M` is treated as the number of diagonals for the matrix in the
-   *  same batch (`M = k[1]-k[0]+1`), and the output tensor is:
-   *  <pre>{@code
-   *  output[i, j, ..., l, m, n]
-   *    = diagonal[i, j, ..., l, diag_index, index_in_diag] ; if k[0] <= d <= k[1]
-   *      padding_value                                     ; otherwise
-   *  }</pre>
-   *  where `d = n - m`, `diag_index = [k] - d`, and
-   *  `index_in_diag = n - max(d, 0) + offset`.
-   *  <p>
-   *  `offset` is zero except when the alignment of the diagonal is to the right.
-   *  <pre>{@code
-   *  offset = max_diag_len - diag_len(d) ; if (`align` in {RIGHT_LEFT, RIGHT_RIGHT}
-   *                                             and `d >= 0`) or
-   *                                           (`align` in {LEFT_RIGHT, RIGHT_RIGHT}
-   *                                             and `d <= 0`)
-   *           0                          ; otherwise
-   *  }</pre>
-   *  where `diag_len(d) = min(cols - max(d, 0), rows + min(d, 0))`.
-   *  <p>
-   *  For example:
-   *  <pre>{@code
-   *  # The main diagonal.
-   *  diagonal = np.array([[1, 2, 3, 4],            # Input shape: (2, 4)
-   *                       [5, 6, 7, 8]])
-   *  tf.matrix_diag(diagonal) ==> [[[1, 0, 0, 0],  # Output shape: (2, 4, 4)
-   *                                 [0, 2, 0, 0],
-   *                                 [0, 0, 3, 0],
-   *                                 [0, 0, 0, 4]],
-   *                                [[5, 0, 0, 0],
-   *                                 [0, 6, 0, 0],
-   *                                 [0, 0, 7, 0],
-   *                                 [0, 0, 0, 8]]]
-   *
-   *  # A superdiagonal (per batch).
-   *  diagonal = np.array([[1, 2, 3],  # Input shape: (2, 3)
-   *                       [4, 5, 6]])
-   *  tf.matrix_diag(diagonal, k = 1)
-   *    ==> [[[0, 1, 0, 0],  # Output shape: (2, 4, 4)
-   *          [0, 0, 2, 0],
-   *          [0, 0, 0, 3],
-   *          [0, 0, 0, 0]],
-   *         [[0, 4, 0, 0],
-   *          [0, 0, 5, 0],
-   *          [0, 0, 0, 6],
-   *          [0, 0, 0, 0]]]
-   *
-   *  # A tridiagonal band (per batch).
-   *  diagonals = np.array([[[0, 8, 9],  # Input shape: (2, 2, 3)
-   *                         [1, 2, 3],
-   *                         [4, 5, 0]],
-   *                        [[0, 2, 3],
-   *                         [6, 7, 9],
-   *                         [9, 1, 0]]])
-   *  tf.matrix_diag(diagonals, k = (-1, 1))
-   *    ==> [[[1, 8, 0],  # Output shape: (2, 3, 3)
-   *          [4, 2, 9],
-   *          [0, 5, 3]],
-   *         [[6, 2, 0],
-   *          [9, 7, 3],
-   *          [0, 1, 9]]]
-   *
-   *  # LEFT_RIGHT alignment.
-   *  diagonals = np.array([[[8, 9, 0],  # Input shape: (2, 2, 3)
-   *                         [1, 2, 3],
-   *                         [0, 4, 5]],
-   *                        [[2, 3, 0],
-   *                         [6, 7, 9],
-   *                         [0, 9, 1]]])
-   *  tf.matrix_diag(diagonals, k = (-1, 1), align="LEFT_RIGHT")
-   *    ==> [[[1, 8, 0],  # Output shape: (2, 3, 3)
-   *          [4, 2, 9],
-   *          [0, 5, 3]],
-   *         [[6, 2, 0],
-   *          [9, 7, 3],
-   *          [0, 1, 9]]]
-   *
-   *  # Rectangular matrix.
-   *  diagonal = np.array([1, 2])  # Input shape: (2)
-   *  tf.matrix_diag(diagonal, k = -1, num_rows = 3, num_cols = 4)
-   *    ==> [[0, 0, 0, 0],  # Output shape: (3, 4)
-   *         [1, 0, 0, 0],
-   *         [0, 2, 0, 0]]
-   *
-   *  # Rectangular matrix with inferred num_cols and padding_value = 9.
-   *  tf.matrix_diag(diagonal, k = -1, num_rows = 3, padding_value = 9)
-   *    ==> [[9, 9],  # Output shape: (3, 2)
-   *         [1, 9],
-   *         [9, 2]]
-   *
-   *  }</pre>
-   *
-   * @param <T> data type for {@code output()} output
-   * @param diagonal Rank `r`, where `r >= 1`
-   * @param k Diagonal offset(s). Positive value means superdiagonal, 0 refers to the main
-   *  diagonal, and negative value means subdiagonals. `k` can be a single integer
-   *  (for a single diagonal) or a pair of integers specifying the low and high ends
-   *  of a matrix band. `k[0]` must not be larger than `k[1]`.
-   * @param numRows The number of rows of the output matrix. If it is not provided, the op assumes
-   *  the output matrix is a square matrix and infers the matrix size from k and the
-   *  innermost dimension of `diagonal`.
-   * @param numCols The number of columns of the output matrix. If it is not provided, the op
-   *  assumes the output matrix is a square matrix and infers the matrix size from
-   *  k and the innermost dimension of `diagonal`.
-   * @param paddingValue The number to fill the area outside the specified diagonal band with.
-   *  Default is 0.
-   * @param options carries optional attributes values
-   * @return a new instance of MatrixDiagV3
-   */
-  public <T extends TType> MatrixDiagV3<T> matrixDiagV3(Operand<T> diagonal, Operand<TInt32> k,
-      Operand<TInt32> numRows, Operand<TInt32> numCols, Operand<T> paddingValue,
-      MatrixDiagV3.Options... options) {
-    return MatrixDiagV3.create(scope, diagonal, k, numRows, numCols, paddingValue, options);
-  }
-
-  /**
-   * Returns a batched matrix tensor with new batched diagonal values.
-   *  <p>
-   *  Given `input` and `diagonal`, this operation returns a tensor with the
-   *  same shape and values as `input`, except for the specified diagonals of the
-   *  innermost matrices. These will be overwritten by the values in `diagonal`.
-   *  <p>
-   *  `input` has `r+1` dimensions `[I, J, ..., L, M, N]`. When `k` is scalar or
-   *  `k[0] == k[1]`, `diagonal` has `r` dimensions `[I, J, ..., L, max_diag_len]`.
-   *  Otherwise, it has `r+1` dimensions `[I, J, ..., L, num_diags, max_diag_len]`.
-   *  `num_diags` is the number of diagonals, `num_diags = k[1] - k[0] + 1`.
-   *  `max_diag_len` is the longest diagonal in the range `[k[0], k[1]]`,
-   *  `max_diag_len = min(M + min(k[1], 0), N + min(-k[0], 0))`
-   *  <p>
-   *  The output is a tensor of rank `k+1` with dimensions `[I, J, ..., L, M, N]`.
-   *  If `k` is scalar or `k[0] == k[1]`:
-   *  <pre>{@code
-   *  output[i, j, ..., l, m, n]
-   *    = diagonal[i, j, ..., l, n-max(k[1], 0)] ; if n - m == k[1]
-   *      input[i, j, ..., l, m, n]              ; otherwise
-   *  }</pre>
-   *  Otherwise,
-   *  <pre>{@code
-   *  output[i, j, ..., l, m, n]
-   *    = diagonal[i, j, ..., l, diag_index, index_in_diag] ; if k[0] <= d <= k[1]
-   *      input[i, j, ..., l, m, n]                         ; otherwise
-   *  }</pre>
-   *  where `d = n - m`, `diag_index = k[1] - d`, and
-   *  `index_in_diag = n - max(d, 0) + offset`.
-   *  <p>
-   *  `offset` is zero except when the alignment of the diagonal is to the right.
-   *  <pre>{@code
-   *  offset = max_diag_len - diag_len(d) ; if (`align` in {RIGHT_LEFT, RIGHT_RIGHT}
-   *                                             and `d >= 0`) or
-   *                                           (`align` in {LEFT_RIGHT, RIGHT_RIGHT}
-   *                                             and `d <= 0`)
-   *           0                          ; otherwise
-   *  }</pre>
-   *  where `diag_len(d) = min(cols - max(d, 0), rows + min(d, 0))`.
-   *  <p>
-   *  For example:
-   *  <pre>{@code
-   *  # The main diagonal.
-   *  input = np.array([[[7, 7, 7, 7],              # Input shape: (2, 3, 4)
-   *                     [7, 7, 7, 7],
-   *                     [7, 7, 7, 7]],
-   *                    [[7, 7, 7, 7],
-   *                     [7, 7, 7, 7],
-   *                     [7, 7, 7, 7]]])
-   *  diagonal = np.array([[1, 2, 3],               # Diagonal shape: (2, 3)
-   *                       [4, 5, 6]])
-   *  tf.matrix_set_diag(input, diagonal)
-   *    ==> [[[1, 7, 7, 7],  # Output shape: (2, 3, 4)
-   *          [7, 2, 7, 7],
-   *          [7, 7, 3, 7]],
-   *         [[4, 7, 7, 7],
-   *          [7, 5, 7, 7],
-   *          [7, 7, 6, 7]]]
-   *
-   *  # A superdiagonal (per batch).
-   *  tf.matrix_set_diag(input, diagonal, k = 1)
-   *    ==> [[[7, 1, 7, 7],  # Output shape: (2, 3, 4)
-   *          [7, 7, 2, 7],
-   *          [7, 7, 7, 3]],
-   *         [[7, 4, 7, 7],
-   *          [7, 7, 5, 7],
-   *          [7, 7, 7, 6]]]
-   *
-   *  # A band of diagonals.
-   *  diagonals = np.array([[[0, 9, 1],  # Diagonal shape: (2, 4, 3)
-   *                         [6, 5, 8],
-   *                         [1, 2, 3],
-   *                         [4, 5, 0]],
-   *                        [[0, 1, 2],
-   *                         [5, 6, 4],
-   *                         [6, 1, 2],
-   *                         [3, 4, 0]]])
-   *  tf.matrix_set_diag(input, diagonals, k = (-1, 2))
-   *    ==> [[[1, 6, 9, 7],  # Output shape: (2, 3, 4)
-   *          [4, 2, 5, 1],
-   *          [7, 5, 3, 8]],
-   *         [[6, 5, 1, 7],
-   *          [3, 1, 6, 2],
-   *          [7, 4, 2, 4]]]
-   *
-   *  # LEFT_RIGHT alignment.
-   *  diagonals = np.array([[[9, 1, 0],  # Diagonal shape: (2, 4, 3)
-   *                         [6, 5, 8],
-   *                         [1, 2, 3],
-   *                         [0, 4, 5]],
-   *                        [[1, 2, 0],
-   *                         [5, 6, 4],
-   *                         [6, 1, 2],
-   *                         [0, 3, 4]]])
-   *  tf.matrix_set_diag(input, diagonals, k = (-1, 2), align="LEFT_RIGHT")
-   *    ==> [[[1, 6, 9, 7],  # Output shape: (2, 3, 4)
-   *          [4, 2, 5, 1],
-   *          [7, 5, 3, 8]],
-   *         [[6, 5, 1, 7],
-   *          [3, 1, 6, 2],
-   *          [7, 4, 2, 4]]]
-   *
-   *  }</pre>
-   *
-   * @param <T> data type for {@code output()} output
-   * @param input Rank `r+1`, where `r >= 1`.
-   * @param diagonal Rank `r` when `k` is an integer or `k[0] == k[1]`. Otherwise, it has rank `r+1`.
-   *  `k >= 1`.
-   * @param k Diagonal offset(s). Positive value means superdiagonal, 0 refers to the main
-   *  diagonal, and negative value means subdiagonals. `k` can be a single integer
-   *  (for a single diagonal) or a pair of integers specifying the low and high ends
-   *  of a matrix band. `k[0]` must not be larger than `k[1]`.
-   * @param options carries optional attributes values
-   * @return a new instance of MatrixSetDiag
-   */
-  public <T extends TType> MatrixSetDiag<T> matrixSetDiag(Operand<T> input, Operand<T> diagonal,
-      Operand<TInt32> k, MatrixSetDiag.Options... options) {
-    return MatrixSetDiag.create(scope, input, diagonal, k, options);
-  }
-
-  /**
-   * Solves one or more linear least-squares problems.
-   *  <p>
-   *  `matrix` is a tensor of shape `[..., M, N]` whose inner-most 2 dimensions
-   *  form real or complex matrices of size `[M, N]`. `Rhs` is a tensor of the same
-   *  type as `matrix` and shape `[..., M, K]`.
-   *  The output is a tensor shape `[..., N, K]` where each output matrix solves
-   *  each of the equations
-   *  `matrix[..., :, :]` * `output[..., :, :]` = `rhs[..., :, :]`
-   *  in the least squares sense.
-   *  <p>
-   *  We use the following notation for (complex) matrix and right-hand sides
-   *  in the batch:
-   *  <p>
-   *  `matrix`=\\(A \in \mathbb{C}^{m \times n}\\),
-   *  `rhs`=\\(B  \in \mathbb{C}^{m \times k}\\),
-   *  `output`=\\(X  \in \mathbb{C}^{n \times k}\\),
-   *  `l2_regularizer`=\\(\lambda \in \mathbb{R}\\).
-   *  <p>
-   *  If `fast` is `True`, then the solution is computed by solving the normal
-   *  equations using Cholesky decomposition. Specifically, if \\(m \ge n\\) then
-   *  \\(X = (A^H A + \lambda I)^{-1} A^H B\\), which solves the least-squares
-   *  problem \\(X = \mathrm{argmin}_{Z \in \Re^{n \times k} } ||A Z - B||_F^2 + \lambda ||Z||_F^2\\).
-   *  If \\(m \lt n\\) then `output` is computed as
-   *  \\(X = A^H (A A^H + \lambda I)^{-1} B\\), which (for \\(\lambda = 0\\)) is the
-   *  minimum-norm solution to the under-determined linear system, i.e.
-   *  \\(X = \mathrm{argmin}_{Z \in \mathbb{C}^{n \times k} } ||Z||_F^2 \\),
-   *  subject to \\(A Z = B\\). Notice that the fast path is only numerically stable
-   *  when \\(A\\) is numerically full rank and has a condition number
-   *  \\(\mathrm{cond}(A) \lt \frac{1}{\sqrt{\epsilon_{mach} } }\\) or \\(\lambda\\) is
-   *  sufficiently large.
-   *  <p>
-   *  If `fast` is `False` an algorithm based on the numerically robust complete
-   *  orthogonal decomposition is used. This computes the minimum-norm
-   *  least-squares solution, even when \\(A\\) is rank deficient. This path is
-   *  typically 6-7 times slower than the fast path. If `fast` is `False` then
-   *  `l2_regularizer` is ignored.
-   *
-   * @param <T> data type for {@code output()} output
-   * @param matrix Shape is `[..., M, N]`.
-   * @param rhs Shape is `[..., M, K]`.
-   * @param l2Regularizer Scalar tensor.
-   *  <p>
-   * @compatibility(numpy) Equivalent to np.linalg.lstsq
-   * @end_compatibility
-   * @param options carries optional attributes values
-   * @return a new instance of MatrixSolveLs
-   */
-  public <T extends TType> MatrixSolveLs<T> matrixSolveLs(Operand<T> matrix, Operand<T> rhs,
-      Operand<TFloat64> l2Regularizer, MatrixSolveLs.Options... options) {
-    return MatrixSolveLs.create(scope, matrix, rhs, l2Regularizer, options);
-  }
-
-  /**
-   * Computes the QR decompositions of one or more matrices.
-   *  <p>
-   *  Computes the QR decomposition of each inner matrix in `tensor` such that
-   *  `tensor[..., :, :] = q[..., :, :] * r[..., :,:])`
-   *  <pre>{@code
-   *  # a is a tensor.
-   *  # q is a tensor of orthonormal matrices.
-   *  # r is a tensor of upper triangular matrices.
-   *  q, r = qr(a)
-   *  q_full, r_full = qr(a, full_matrices=True)
-   *  }</pre>
-   *
-   * @param <T> data type for {@code q()} output
-   * @param input A tensor of shape `[..., M, N]` whose inner-most 2 dimensions
-   *  form matrices of size `[M, N]`. Let `P` be the minimum of `M` and `N`.
-   * @param options carries optional attributes values
-   * @return a new instance of Qr
-   */
-  public <T extends TType> Qr<T> qr(Operand<T> input, Qr.Options... options) {
-    return Qr.create(scope, input, options);
-  }
-
-  /**
-   * Perform a quantized matrix multiplication of  `a` by the matrix `b`.
-   *  <p>
-   *  The inputs must be two-dimensional matrices and the inner dimension of
-   *  `a` (after being transposed if `transpose_a` is non-zero) must match the
-   *  outer dimension of `b` (after being transposed if `transposed_b` is
-   *  non-zero).
-   *
-   * @param <V> data type for {@code out()} output
-   * @param a Must be a two-dimensional tensor.
-   * @param b Must be a two-dimensional tensor.
-   * @param minA The float value that the lowest quantized `a` value represents.
-   * @param maxA The float value that the highest quantized `a` value represents.
-   * @param minB The float value that the lowest quantized `b` value represents.
-   * @param maxB The float value that the highest quantized `b` value represents.
-   * @param Toutput
-   * @param Tactivation The type of output produced by activation function
-   *  following this operation.
-   * @param options carries optional attributes values
-   * @return a new instance of QuantizedMatMul
-   */
-  public <V extends TType, T extends TType, U extends TType, W extends TType> QuantizedMatMul<V> quantizedMatMul(
-      Operand<T> a, Operand<U> b, Operand<TFloat32> minA, Operand<TFloat32> maxA,
-      Operand<TFloat32> minB, Operand<TFloat32> maxB, DataType<V> Toutput, DataType<W> Tactivation,
-      QuantizedMatMul.Options... options) {
-    return QuantizedMatMul.create(scope, a, b, minA, maxA, minB, maxB, Toutput, Tactivation, options);
-  }
-
-  /**
-   * Computes the eigen decomposition of one or more square self-adjoint matrices.
-   *  <p>
-   *  Computes the eigenvalues and (optionally) eigenvectors of each inner matrix in
-   *  `input` such that `input[..., :, :] = v[..., :, :] * diag(e[..., :])`. The eigenvalues
-   *  are sorted in non-decreasing order.
-   *  <pre>{@code
-   *  # a is a tensor.
-   *  # e is a tensor of eigenvalues.
-   *  # v is a tensor of eigenvectors.
-   *  e, v = self_adjoint_eig(a)
-   *  e = self_adjoint_eig(a, compute_v=False)
-   *  }</pre>
-   *
-   * @param <T> data type for {@code e()} output
-   * @param input `Tensor` input of shape `[N, N]`.
-   * @param options carries optional attributes values
-   * @return a new instance of SelfAdjointEig
-   */
-  public <T extends TType> SelfAdjointEig<T> selfAdjointEig(Operand<T> input,
-      SelfAdjointEig.Options... options) {
-    return SelfAdjointEig.create(scope, input, options);
-  }
-
-  /**
-   * Solves systems of linear equations.
-   *  <p>
-   *  `Matrix` is a tensor of shape `[..., M, M]` whose inner-most 2 dimensions
-   *  form square matrices. `Rhs` is a tensor of shape `[..., M, K]`. The `output` is
-   *  a tensor shape `[..., M, K]`.  If `adjoint` is `False` then each output matrix
-   *  satisfies `matrix[..., :, :] * output[..., :, :] = rhs[..., :, :]`.
-   *  If `adjoint` is `True` then each output matrix satisfies
-   *  `adjoint(matrix[..., :, :]) * output[..., :, :] = rhs[..., :, :]`.
-   *
-   * @param <T> data type for {@code output()} output
-   * @param matrix Shape is `[..., M, M]`.
-   * @param rhs Shape is `[..., M, K]`.
-   * @param options carries optional attributes values
-   * @return a new instance of Solve
-   */
-  public <T extends TType> Solve<T> solve(Operand<T> matrix, Operand<T> rhs,
-      Solve.Options... options) {
-    return Solve.create(scope, matrix, rhs, options);
-  }
-
-  /**
-   * Computes the matrix square root of one or more square matrices:
-   *  <p>
-   *  matmul(sqrtm(A), sqrtm(A)) = A
-   *  <p>
-   *  The input matrix should be invertible. If the input matrix is real, it should
-   *  have no eigenvalues which are real and negative (pairs of complex conjugate
-   *  eigenvalues are allowed).
-   *  <p>
-   *  The matrix square root is computed by first reducing the matrix to
-   *  quasi-triangular form with the real Schur decomposition. The square root
-   *  of the quasi-triangular matrix is then computed directly. Details of
-   *  the algorithm can be found in: Nicholas J. Higham, "Computing real
-   *  square roots of a real matrix", Linear Algebra Appl., 1987.
-   *  <p>
-   *  The input is a tensor of shape `[..., M, M]` whose inner-most 2 dimensions
-   *  form square matrices. The output is a tensor of the same shape as the input
-   *  containing the matrix square root for all input submatrices `[..., :, :]`.
-   *
-   * @param <T> data type for {@code output()} output
-   * @param input Shape is `[..., M, M]`.
-   * @return a new instance of Sqrtm
-   */
-  public <T extends TType> Sqrtm<T> sqrtm(Operand<T> input) {
-    return Sqrtm.create(scope, input);
-  }
-
-  /**
-   * Computes the singular value decompositions of one or more matrices.
-   *  <p>
-   *  Computes the SVD of each inner matrix in `input` such that
-   *  `input[..., :, :] = u[..., :, :] * diag(s[..., :, :]) * transpose(v[..., :, :])`
-   *  <pre>{@code
-   *  # a is a tensor containing a batch of matrices.
-   *  # s is a tensor of singular values for each matrix.
-   *  # u is the tensor containing the left singular vectors for each matrix.
-   *  # v is the tensor containing the right singular vectors for each matrix.
-   *  s, u, v = svd(a)
-   *  s, _, _ = svd(a, compute_uv=False)
-   *  }</pre>
-   *
-   * @param <T> data type for {@code s()} output
-   * @param input A tensor of shape `[..., M, N]` whose inner-most 2 dimensions
-   *  form matrices of size `[M, N]`. Let `P` be the minimum of `M` and `N`.
-   * @param options carries optional attributes values
-   * @return a new instance of Svd
-   */
-  public <T extends TType> Svd<T> svd(Operand<T> input, Svd.Options... options) {
-    return Svd.create(scope, input, options);
-  }
-
-  /**
-   * Returns a diagonal tensor with a given diagonal values.
-   *  <p>
-   *  Given a `diagonal`, this operation returns a tensor with the `diagonal` and
-   *  everything else padded with zeros. The diagonal is computed as follows:
-   *  <p>
-   *  Assume `diagonal` has dimensions [D1,..., Dk], then the output is a tensor of
-   *  rank 2k with dimensions [D1,..., Dk, D1,..., Dk] where:
-   *  <p>
-   *  `output[i1,..., ik, i1,..., ik] = diagonal[i1, ..., ik]` and 0 everywhere else.
-   *  <p>
-   *  For example:
-   *  <pre>{@code
-   *  # 'diagonal' is [1, 2, 3, 4]
-   *  tf.diag(diagonal) ==> [[1, 0, 0, 0]
-   *                         [0, 2, 0, 0]
-   *                         [0, 0, 3, 0]
-   *                         [0, 0, 0, 4]]
-   *  }</pre>
-   *
-   * @param <T> data type for {@code output()} output
-   * @param diagonal Rank k tensor where k is at most 1.
-   * @return a new instance of TensorDiag
-   */
-  public <T extends TType> TensorDiag<T> tensorDiag(Operand<T> diagonal) {
-    return TensorDiag.create(scope, diagonal);
-  }
-
-  /**
-   * Returns the diagonal part of the tensor.
-   *  <p>
-   *  This operation returns a tensor with the `diagonal` part
-   *  of the `input`. The `diagonal` part is computed as follows:
-   *  <p>
-   *  Assume `input` has dimensions `[D1,..., Dk, D1,..., Dk]`, then the output is a
-   *  tensor of rank `k` with dimensions `[D1,..., Dk]` where:
-   *  <p>
-   *  `diagonal[i1,..., ik] = input[i1, ..., ik, i1,..., ik]`.
-   *  <p>
-   *  For example:
-   *  <pre>{@code
-   *  # 'input' is [[1, 0, 0, 0]
-   *                [0, 2, 0, 0]
-   *                [0, 0, 3, 0]
-   *                [0, 0, 0, 4]]
-   *
-   *  tf.diag_part(input) ==> [1, 2, 3, 4]
-   *  }</pre>
-   *
-   * @param <T> data type for {@code diagonal()} output
-   * @param input Rank k tensor where k is even and not zero.
-   * @return a new instance of TensorDiagPart
-   */
-  public <T extends TType> TensorDiagPart<T> tensorDiagPart(Operand<T> input) {
-    return TensorDiagPart.create(scope, input);
-  }
-
-  /**
-   * Shuffle dimensions of x according to a permutation.
-   *  <p>
-   *  The output `y` has the same rank as `x`. The shapes of `x` and `y` satisfy:
-   *    `y.shape[i] == x.shape[perm[i]] for i in [0, 1, ..., rank(x) - 1]`
-   *
-   * @param <T> data type for {@code y()} output
-   * @param x
-   * @param perm
-   * @return a new instance of Transpose
-   */
-  public <T extends TType, U extends TNumber> Transpose<T> transpose(Operand<T> x,
-      Operand<U> perm) {
-    return Transpose.create(scope, x, perm);
-  }
-
-  /**
-   * Solves systems of linear equations with upper or lower triangular matrices by backsubstitution.
-   *  <p>
-   *
-   *  `matrix` is a tensor of shape `[..., M, M]` whose inner-most 2 dimensions form
-   *  square matrices. If `lower` is `True` then the strictly upper triangular part
-   *  of each inner-most matrix is assumed to be zero and not accessed.
-   *  If `lower` is False then the strictly lower triangular part of each inner-most
-   *  matrix is assumed to be zero and not accessed.
-   *  `rhs` is a tensor of shape `[..., M, N]`.
-   *  <p>
-   *  The output is a tensor of shape `[..., M, N]`. If `adjoint` is
-   *  `True` then the innermost matrices in `output` satisfy matrix equations
-   *  `matrix[..., :, :] * output[..., :, :] = rhs[..., :, :]`.
-   *  If `adjoint` is `False` then the strictly then the  innermost matrices in
-   *  `output` satisfy matrix equations
-   *  `adjoint(matrix[..., i, k]) * output[..., k, j] = rhs[..., i, j]`.
-   *  <p>
-   *  Note, the batch shapes for the inputs only need to broadcast.
-   *  <p>
-   *  Example:
-   *  <pre>{@code
-   *  a = tf.constant([[3,  0,  0,  0],
-   *                   [2,  1,  0,  0],
-   *                   [1,  0,  1,  0],
-   *                   [1,  1,  1,  1]], dtype=tf.float32)
-   *
-   *  b = tf.constant([[4],
-   *                   [2],
-   *                   [4],
-   *                   [2]], dtype=tf.float32)
-   *
-   *  x = tf.linalg.triangular_solve(a, b, lower=True)
-   *  x
-   *  # <tf.Tensor: shape=(4, 1), dtype=float32, numpy=
-   *  # array([[ 1.3333334 ],
-   *  #        [-0.66666675],
-   *  #        [ 2.6666665 ],
-   *  #        [-1.3333331 ]], dtype=float32)>
-   *
-   *  # in python3 one can use `a@x`
-   *  tf.matmul(a, x)
-   *  # <tf.Tensor: shape=(4, 1), dtype=float32, numpy=
-   *  # array([[4.       ],
-   *  #        [2.       ],
-   *  #        [4.       ],
-   *  #        [1.9999999]], dtype=float32)>
-   *  }</pre>
-   *
-   * @param <T> data type for {@code output()} output
-   * @param matrix Shape is `[..., M, M]`.
-   * @param rhs Shape is `[..., M, K]`.
-   * @param options carries optional attributes values
-   * @return a new instance of TriangularSolve
-   */
-  public <T extends TType> TriangularSolve<T> triangularSolve(Operand<T> matrix, Operand<T> rhs,
-      TriangularSolve.Options... options) {
-    return TriangularSolve.create(scope, matrix, rhs, options);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/LinalgSparseOps.java b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/LinalgSparseOps.java
deleted file mode 100644
index 7f8777c883a..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/LinalgSparseOps.java
+++ /dev/null
@@ -1,463 +0,0 @@
-// Copyright 2020 The TensorFlow Authors. All Rights Reserved.
-//
-// Licensed under the Apache License, Version 2.0 (the "License");
-// you may not use this file except in compliance with the License.
-// You may obtain a copy of the License at
-//
-//     http://www.apache.org/licenses/LICENSE-2.0
-//
-// Unless required by applicable law or agreed to in writing, software
-// distributed under the License is distributed on an "AS IS" BASIS,
-// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-// See the License for the specific language governing permissions and
-// limitations under the License.
-// ==============================================================================
-//
-// This class has been generated, DO NOT EDIT!
-//
-package org.tensorflow.op;
-
-import org.tensorflow.DataType;
-import org.tensorflow.Operand;
-import org.tensorflow.op.linalg.sparse.CSRSparseMatrixToSparseTensor;
-import org.tensorflow.op.linalg.sparse.DenseToCSRSparseMatrix;
-import org.tensorflow.op.linalg.sparse.SparseMatrixAdd;
-import org.tensorflow.op.linalg.sparse.SparseMatrixMatMul;
-import org.tensorflow.op.linalg.sparse.SparseMatrixMul;
-import org.tensorflow.op.linalg.sparse.SparseMatrixNNZ;
-import org.tensorflow.op.linalg.sparse.SparseMatrixOrderingAMD;
-import org.tensorflow.op.linalg.sparse.SparseMatrixSoftmax;
-import org.tensorflow.op.linalg.sparse.SparseMatrixSoftmaxGrad;
-import org.tensorflow.op.linalg.sparse.SparseMatrixSparseCholesky;
-import org.tensorflow.op.linalg.sparse.SparseMatrixSparseMatMul;
-import org.tensorflow.op.linalg.sparse.SparseMatrixTranspose;
-import org.tensorflow.op.linalg.sparse.SparseMatrixZeros;
-import org.tensorflow.op.linalg.sparse.SparseTensorToCSRSparseMatrix;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * An API for building {@code linalg.sparse} operations as {@link Op Op}s
- *
- * @see {@link Ops}
- */
-public final class LinalgSparseOps {
-  private final Scope scope;
-
-  LinalgSparseOps(Scope scope) {
-    this.scope = scope;
-  }
-
-  /**
-   * Converts a (possibly batched) CSRSparesMatrix to a SparseTensor.
-   *
-   * @param <T> data type for {@code values()} output
-   * @param sparseMatrix A (possibly batched) CSRSparseMatrix.
-   * @param type
-   * @return a new instance of CSRSparseMatrixToSparseTensor
-   */
-  public <T extends TType> CSRSparseMatrixToSparseTensor<T> cSRSparseMatrixToSparseTensor(
-      Operand<?> sparseMatrix, DataType<T> type) {
-    return CSRSparseMatrixToSparseTensor.create(scope, sparseMatrix, type);
-  }
-
-  /**
-   * Converts a dense tensor to a (possibly batched) CSRSparseMatrix.
-   *
-   * @param denseInput A Dense tensor.
-   * @param indices Indices of nonzero elements.
-   * @return a new instance of DenseToCSRSparseMatrix
-   */
-  public <T extends TType> DenseToCSRSparseMatrix denseToCSRSparseMatrix(Operand<T> denseInput,
-      Operand<TInt64> indices) {
-    return DenseToCSRSparseMatrix.create(scope, denseInput, indices);
-  }
-
-  /**
-   * Sparse addition of two CSR matrices, C = alpha * A + beta * B.
-   *  <p>
-   *  The gradients of SparseMatrixAdd outputs with respect to alpha and beta are not
-   *  currently defined (TensorFlow will return zeros for these entries).
-   *
-   * @param a A CSRSparseMatrix.
-   * @param b A CSRSparseMatrix.
-   * @param alpha A constant scalar.
-   * @param beta A constant scalar.
-   * @return a new instance of SparseMatrixAdd
-   */
-  public <T extends TType> SparseMatrixAdd sparseMatrixAdd(Operand<?> a, Operand<?> b,
-      Operand<T> alpha, Operand<T> beta) {
-    return SparseMatrixAdd.create(scope, a, b, alpha, beta);
-  }
-
-  /**
-   * Matrix-multiplies a sparse matrix with a dense matrix.
-   *  <p>
-   *  Returns a dense matrix.
-   *  For inputs A and B, where A is CSR and B is dense; this op returns a dense C;
-   *  <p>
-   *  If transpose_output is false, returns:
-   *  <pre>{@code
-   *    C = A . B
-   *  }</pre>
-   *  If transpose_output is `true`, returns:
-   *  <pre>{@code
-   *    C = transpose(A . B) = transpose(B) . transpose(A)
-   *  }</pre>
-   *  where the transposition is performed along the two innermost (matrix)
-   *  dimensions.
-   *  <p>
-   *  If conjugate_output is `true`, returns:
-   *  <pre>{@code
-   *    C = conjugate(A . B) = conjugate(A) . conjugate(B)
-   *  }</pre>
-   *  If both conjugate_output and transpose_output are `true`, returns:
-   *  <pre>{@code
-   *    C = conjugate(transpose(A . B)) = conjugate(transpose(B)) .
-   *                                      conjugate(transpose(A))
-   *  }</pre>
-   *
-   * @param <T> data type for {@code output()} output
-   * @param a A CSRSparseMatrix.
-   * @param b A dense tensor.
-   * @param options carries optional attributes values
-   * @return a new instance of SparseMatrixMatMul
-   */
-  public <T extends TType> SparseMatrixMatMul<T> sparseMatrixMatMul(Operand<?> a, Operand<T> b,
-      SparseMatrixMatMul.Options... options) {
-    return SparseMatrixMatMul.create(scope, a, b, options);
-  }
-
-  /**
-   * Element-wise multiplication of a sparse matrix with a dense tensor.
-   *  <p>
-   *  Returns a sparse matrix.
-   *  <p>
-   *  The dense tensor `b` may be either a scalar; otherwise `a` must be a rank-3
-   *  `SparseMatrix`; in this case `b` must be shaped `[batch_size, 1, 1]` and the
-   *  multiply operation broadcasts.
-   *  <p>
-   *  <b>NOTE</b> even if `b` is zero, the sparsity structure of the output does not
-   *  change.
-   *
-   * @param a A CSRSparseMatrix.
-   * @param b A dense tensor.
-   * @return a new instance of SparseMatrixMul
-   */
-  public <T extends TType> SparseMatrixMul sparseMatrixMul(Operand<?> a, Operand<T> b) {
-    return SparseMatrixMul.create(scope, a, b);
-  }
-
-  /**
-   * Returns the number of nonzeroes of `sparse_matrix`.
-   *
-   * @param sparseMatrix A CSRSparseMatrix.
-   * @return a new instance of SparseMatrixNNZ
-   */
-  public SparseMatrixNNZ sparseMatrixNNZ(Operand<?> sparseMatrix) {
-    return SparseMatrixNNZ.create(scope, sparseMatrix);
-  }
-
-  /**
-   * Computes the Approximate Minimum Degree (AMD) ordering of `input`.
-   *  <p>
-   *  Computes the Approximate Minimum Degree (AMD) ordering for a sparse matrix.
-   *  <p>
-   *  The returned permutation may be used to permute the rows and columns of the
-   *  given sparse matrix. This typically results in permuted sparse matrix's sparse
-   *  Cholesky (or other decompositions) in having fewer zero fill-in compared to
-   *  decomposition of the original matrix.
-   *  <p>
-   *  The input sparse matrix may have rank 2 or rank 3. The output Tensor,
-   *  representing would then have rank 1 or 2 respectively, with the same batch
-   *  shape as the input.
-   *  <p>
-   *  Each component of the input sparse matrix must represent a square symmetric
-   *  matrix; only the lower triangular part of the matrix is read. The values of the
-   *  sparse matrix does not affect the returned permutation, only the sparsity
-   *  pattern of the sparse matrix is used. Hence, a single AMD ordering may be
-   *  reused for the Cholesky decompositions of sparse matrices with the same sparsity
-   *  pattern but with possibly different values.
-   *  <p>
-   *  Each batch component of the output permutation represents a permutation of `N`
-   *  elements, where the input sparse matrix components each have `N` rows. That is,
-   *  the component contains each of the integers `{0, .. N-1}` exactly once. The
-   *  `i`th element represents the row index that the `i`th row maps to.
-   *  <p>
-   *  Usage example:
-   *  <pre>{@code
-   *      from tensorflow.python.ops.linalg.sparse import sparse_csr_matrix_ops
-   *
-   *      a_indices = np.array([[0, 0], [1, 1], [2, 1], [2, 2], [3, 3]])
-   *      a_values = np.array([1.0, 2.0, 1.0, 3.0, 4.0], np.float32)
-   *      a_dense_shape = [4, 4]
-   *
-   *      with tf.Session() as sess:
-   *        # Define (COO format) SparseTensor over Numpy array.
-   *        a_st = tf.SparseTensor(a_indices, a_values, a_dense_shape)
-   *
-   *        # Convert SparseTensors to CSR SparseMatrix.
-   *        a_sm = sparse_csr_matrix_ops.sparse_tensor_to_csr_sparse_matrix(
-   *            a_st.indices, a_st.values, a_st.dense_shape)
-   *
-   *        # Obtain the AMD Ordering for the CSR SparseMatrix.
-   *        ordering_amd = sparse_csr_matrix_ops.sparse_matrix_ordering_amd(sparse_matrix)
-   *
-   *        ordering_amd_value = sess.run(ordering_amd)
-   *  }</pre>
-   *  `ordering_amd_value` stores the AMD ordering: `[1 2 3 0]`.
-   *  <p>
-   *  input: A `CSRSparseMatrix`.
-   *
-   * @param input A `CSRSparseMatrix`.
-   * @return a new instance of SparseMatrixOrderingAMD
-   */
-  public SparseMatrixOrderingAMD sparseMatrixOrderingAMD(Operand<?> input) {
-    return SparseMatrixOrderingAMD.create(scope, input);
-  }
-
-  /**
-   * Calculates the softmax of a CSRSparseMatrix.
-   *  <p>
-   *  Calculate the softmax of the innermost dimensions of a SparseMatrix.
-   *  <p>
-   *  Missing values are treated as `-inf` (i.e., logits of zero probability); and
-   *  the output has the same sparsity structure as the input (though missing values
-   *  in the output may now be treated as having probability zero).
-   *
-   * @param logits A CSRSparseMatrix.
-   * @param type
-   * @return a new instance of SparseMatrixSoftmax
-   */
-  public <T extends TNumber> SparseMatrixSoftmax sparseMatrixSoftmax(Operand<?> logits,
-      DataType<T> type) {
-    return SparseMatrixSoftmax.create(scope, logits, type);
-  }
-
-  /**
-   * Calculates the gradient of the SparseMatrixSoftmax op.
-   *
-   * @param softmax A CSRSparseMatrix.
-   * @param gradSoftmax The gradient of `softmax`.
-   * @param type
-   * @return a new instance of SparseMatrixSoftmaxGrad
-   */
-  public <T extends TNumber> SparseMatrixSoftmaxGrad sparseMatrixSoftmaxGrad(Operand<?> softmax,
-      Operand<?> gradSoftmax, DataType<T> type) {
-    return SparseMatrixSoftmaxGrad.create(scope, softmax, gradSoftmax, type);
-  }
-
-  /**
-   * Computes the sparse Cholesky decomposition of `input`.
-   *  <p>
-   *  Computes the Sparse Cholesky decomposition of a sparse matrix, with the given
-   *  fill-in reducing permutation.
-   *  <p>
-   *  The input sparse matrix and the fill-in reducing permutation `permutation` must
-   *  have compatible shapes. If the sparse matrix has rank 3; with the batch
-   *  dimension `B`, then the `permutation` must be of rank 2; with the same batch
-   *  dimension `B`. There is no support for broadcasting.
-   *  <p>
-   *  Furthermore, each component vector of `permutation` must be of length `N`,
-   *  containing each of the integers {0, 1, ..., N - 1} exactly once, where `N` is
-   *  the number of rows of each component of the sparse matrix.
-   *  <p>
-   *  Each component of the input sparse matrix must represent a symmetric positive
-   *  definite (SPD) matrix; although only the lower triangular part of the matrix is
-   *  read. If any individual component is not SPD, then an InvalidArgument error is
-   *  thrown.
-   *  <p>
-   *  The returned sparse matrix has the same dense shape as the input sparse matrix.
-   *  For each component `A` of the input sparse matrix, the corresponding output
-   *  sparse matrix represents `L`, the lower triangular Cholesky factor satisfying
-   *  the following identity:
-   *  <pre>{@code
-   *    A = L * Lt
-   *  }</pre>
-   *  where Lt denotes the transpose of L (or its conjugate transpose, if `type` is
-   *  `complex64` or `complex128`).
-   *  <p>
-   *  The `type` parameter denotes the type of the matrix elements. The supported
-   *  types are: `float32`, `float64`, `complex64` and `complex128`.
-   *  <p>
-   *  Usage example:
-   *  <pre>{@code
-   *      from tensorflow.python.ops.linalg.sparse import sparse_csr_matrix_ops
-   *
-   *      a_indices = np.array([[0, 0], [1, 1], [2, 1], [2, 2], [3, 3]])
-   *      a_values = np.array([1.0, 2.0, 1.0, 3.0, 4.0], np.float32)
-   *      a_dense_shape = [4, 4]
-   *
-   *      with tf.Session() as sess:
-   *        # Define (COO format) SparseTensor over Numpy array.
-   *        a_st = tf.SparseTensor(a_indices, a_values, a_dense_shape)
-   *
-   *        # Convert SparseTensors to CSR SparseMatrix.
-   *        a_sm = sparse_csr_matrix_ops.sparse_tensor_to_csr_sparse_matrix(
-   *            a_st.indices, a_st.values, a_st.dense_shape)
-   *
-   *        # Obtain the Sparse Cholesky factor using AMD Ordering for reducing zero
-   *        # fill-in (number of structural non-zeros in the sparse Cholesky factor).
-   *        ordering_amd = sparse_csr_matrix_ops.sparse_matrix_ordering_amd(sparse_matrix)
-   *        cholesky_sparse_matrices = (
-   *            sparse_csr_matrix_ops.sparse_matrix_sparse_cholesky(
-   *                sparse_matrix, ordering_amd, type=tf.float32))
-   *
-   *        # Convert the CSRSparseMatrix Cholesky factor to a dense Tensor
-   *        dense_cholesky = sparse_csr_matrix_ops.csr_sparse_matrix_to_dense(
-   *            cholesky_sparse_matrices, tf.float32)
-   *
-   *        # Evaluate the dense Tensor value.
-   *        dense_cholesky_value = sess.run(dense_cholesky)
-   *  }</pre>
-   *  `dense_cholesky_value` stores the dense Cholesky factor:
-   *  <pre>{@code
-   *      [[  1.  0.    0.    0.]
-   *       [  0.  1.41  0.    0.]
-   *       [  0.  0.70  1.58  0.]
-   *       [  0.  0.    0.    2.]]
-   *  }</pre>
-   *  input: A `CSRSparseMatrix`.
-   *  permutation: A `Tensor`.
-   *  type: The type of `input`.
-   *
-   * @param input A `CSRSparseMatrix`.
-   * @param permutation A fill-in reducing permutation matrix.
-   * @param type
-   * @return a new instance of SparseMatrixSparseCholesky
-   */
-  public <T extends TType> SparseMatrixSparseCholesky sparseMatrixSparseCholesky(Operand<?> input,
-      Operand<TInt32> permutation, DataType<T> type) {
-    return SparseMatrixSparseCholesky.create(scope, input, permutation, type);
-  }
-
-  /**
-   * Sparse-matrix-multiplies two CSR matrices `a` and `b`.
-   *  <p>
-   *  Performs a matrix multiplication of a sparse matrix `a` with a sparse matrix
-   *  `b`; returns a sparse matrix `a * b`, unless either `a` or `b` is transposed or
-   *  adjointed.
-   *  <p>
-   *  Each matrix may be transposed or adjointed (conjugated and transposed)
-   *  according to the Boolean parameters `transpose_a`, `adjoint_a`, `transpose_b`
-   *  and `adjoint_b`. At most one of `transpose_a` or `adjoint_a` may be True.
-   *  Similarly, at most one of `transpose_b` or `adjoint_b` may be True.
-   *  <p>
-   *  The inputs must have compatible shapes. That is, the inner dimension of `a`
-   *  must be equal to the outer dimension of `b`. This requirement is adjusted
-   *  according to whether either `a` or `b` is transposed or adjointed.
-   *  <p>
-   *  The `type` parameter denotes the type of the matrix elements. Both `a` and `b`
-   *  must have the same type. The supported types are: `float32`, `float64`,
-   *  `complex64` and `complex128`.
-   *  <p>
-   *  Both `a` and `b` must have the same rank. Broadcasting is not supported. If they
-   *  have rank 3, each batch of 2D CSRSparseMatrices within `a` and `b` must have the
-   *  same dense shape.
-   *  <p>
-   *  The sparse matrix product may have numeric (non-structural) zeros.
-   *  TODO(anudhyan): Consider adding a boolean attribute to control whether to prune
-   *  zeros.
-   *  <p>
-   *  Usage example:
-   *  <pre>{@code
-   *      from tensorflow.python.ops.linalg.sparse import sparse_csr_matrix_ops
-   *
-   *      a_indices = np.array([[0, 0], [2, 3], [2, 4], [3, 0]])
-   *      a_values = np.array([1.0, 5.0, -1.0, -2.0], np.float32)
-   *      a_dense_shape = [4, 5]
-   *
-   *      b_indices = np.array([[0, 0], [3, 0], [3, 1]])
-   *      b_values = np.array([2.0, 7.0, 8.0], np.float32)
-   *      b_dense_shape = [5, 3]
-   *
-   *      with tf.Session() as sess:
-   *        # Define (COO format) Sparse Tensors over Numpy arrays
-   *        a_st = tf.SparseTensor(a_indices, a_values, a_dense_shape)
-   *        b_st = tf.SparseTensor(b_indices, b_values, b_dense_shape)
-   *
-   *        # Convert SparseTensors to CSR SparseMatrix
-   *        a_sm = sparse_csr_matrix_ops.sparse_tensor_to_csr_sparse_matrix(
-   *            a_st.indices, a_st.values, a_st.dense_shape)
-   *        b_sm = sparse_csr_matrix_ops.sparse_tensor_to_csr_sparse_matrix(
-   *            b_st.indices, b_st.values, b_st.dense_shape)
-   *
-   *        # Compute the CSR SparseMatrix matrix multiplication
-   *        c_sm = sparse_csr_matrix_ops.sparse_matrix_sparse_mat_mul(
-   *            a=a_sm, b=b_sm, type=tf.float32)
-   *
-   *        # Convert the CSR SparseMatrix product to a dense Tensor
-   *        c_sm_dense = sparse_csr_matrix_ops.csr_sparse_matrix_to_dense(
-   *            c_sm, tf.float32)
-   *        # Evaluate the dense Tensor value
-   *        c_sm_dense_value = sess.run(c_sm_dense)
-   *  }</pre>
-   *  `c_sm_dense_value` stores the dense matrix product:
-   *  <pre>{@code
-   *      [[  2.   0.   0.]
-   *       [  0.   0.   0.]
-   *       [ 35.  40.   0.]
-   *       [ -4.   0.   0.]]
-   *  }</pre>
-   *  a: A `CSRSparseMatrix`.
-   *  b: A `CSRSparseMatrix` with the same type and rank as `a`.
-   *  type: The type of both `a` and `b`.
-   *  transpose_a: If True, `a` transposed before multiplication.
-   *  transpose_b: If True, `b` transposed before multiplication.
-   *  adjoint_a: If True, `a` adjointed before multiplication.
-   *  adjoint_b: If True, `b` adjointed before multiplication.
-   *
-   * @param a A CSRSparseMatrix.
-   * @param b A CSRSparseMatrix.
-   * @param type
-   * @param options carries optional attributes values
-   * @return a new instance of SparseMatrixSparseMatMul
-   */
-  public <T extends TType> SparseMatrixSparseMatMul sparseMatrixSparseMatMul(Operand<?> a,
-      Operand<?> b, DataType<T> type, SparseMatrixSparseMatMul.Options... options) {
-    return SparseMatrixSparseMatMul.create(scope, a, b, type, options);
-  }
-
-  /**
-   * Transposes the inner (matrix) dimensions of a CSRSparseMatrix.
-   *  <p>
-   *  Transposes the inner (matrix) dimensions of a SparseMatrix and optionally
-   *  conjugates its values.
-   *
-   * @param input A CSRSparseMatrix.
-   * @param type
-   * @param options carries optional attributes values
-   * @return a new instance of SparseMatrixTranspose
-   */
-  public <T extends TType> SparseMatrixTranspose sparseMatrixTranspose(Operand<?> input,
-      DataType<T> type, SparseMatrixTranspose.Options... options) {
-    return SparseMatrixTranspose.create(scope, input, type, options);
-  }
-
-  /**
-   * Creates an all-zeros CSRSparseMatrix with shape `dense_shape`.
-   *
-   * @param denseShape The desired matrix shape.
-   * @param type
-   * @return a new instance of SparseMatrixZeros
-   */
-  public <T extends TType> SparseMatrixZeros sparseMatrixZeros(Operand<TInt64> denseShape,
-      DataType<T> type) {
-    return SparseMatrixZeros.create(scope, denseShape, type);
-  }
-
-  /**
-   * Converts a SparseTensor to a (possibly batched) CSRSparseMatrix.
-   *
-   * @param indices SparseTensor indices.
-   * @param values SparseTensor values.
-   * @param denseShape SparseTensor dense shape.
-   * @return a new instance of SparseTensorToCSRSparseMatrix
-   */
-  public <T extends TType> SparseTensorToCSRSparseMatrix sparseTensorToCSRSparseMatrix(
-      Operand<TInt64> indices, Operand<T> values, Operand<TInt64> denseShape) {
-    return SparseTensorToCSRSparseMatrix.create(scope, indices, values, denseShape);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/MathOps.java b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/MathOps.java
deleted file mode 100644
index 8da2c9ee3cd..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/MathOps.java
+++ /dev/null
@@ -1,2402 +0,0 @@
-// Copyright 2020 The TensorFlow Authors. All Rights Reserved.
-//
-// Licensed under the Apache License, Version 2.0 (the "License");
-// you may not use this file except in compliance with the License.
-// You may obtain a copy of the License at
-//
-//     http://www.apache.org/licenses/LICENSE-2.0
-//
-// Unless required by applicable law or agreed to in writing, software
-// distributed under the License is distributed on an "AS IS" BASIS,
-// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-// See the License for the specific language governing permissions and
-// limitations under the License.
-// ==============================================================================
-//
-// This class has been generated, DO NOT EDIT!
-//
-package org.tensorflow.op;
-
-import org.tensorflow.DataType;
-import org.tensorflow.Operand;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.op.math.Abs;
-import org.tensorflow.op.math.AccumulateN;
-import org.tensorflow.op.math.Acos;
-import org.tensorflow.op.math.Acosh;
-import org.tensorflow.op.math.Add;
-import org.tensorflow.op.math.AddN;
-import org.tensorflow.op.math.Angle;
-import org.tensorflow.op.math.ApproximateEqual;
-import org.tensorflow.op.math.ArgMax;
-import org.tensorflow.op.math.ArgMin;
-import org.tensorflow.op.math.Asin;
-import org.tensorflow.op.math.Asinh;
-import org.tensorflow.op.math.Atan;
-import org.tensorflow.op.math.Atan2;
-import org.tensorflow.op.math.Atanh;
-import org.tensorflow.op.math.BesselI0e;
-import org.tensorflow.op.math.BesselI1e;
-import org.tensorflow.op.math.Betainc;
-import org.tensorflow.op.math.Bincount;
-import org.tensorflow.op.math.Ceil;
-import org.tensorflow.op.math.CompareAndBitpack;
-import org.tensorflow.op.math.ComplexAbs;
-import org.tensorflow.op.math.Conj;
-import org.tensorflow.op.math.Cos;
-import org.tensorflow.op.math.Cosh;
-import org.tensorflow.op.math.Cumprod;
-import org.tensorflow.op.math.Cumsum;
-import org.tensorflow.op.math.Digamma;
-import org.tensorflow.op.math.Div;
-import org.tensorflow.op.math.DivNoNan;
-import org.tensorflow.op.math.Equal;
-import org.tensorflow.op.math.Erf;
-import org.tensorflow.op.math.Erfc;
-import org.tensorflow.op.math.Exp;
-import org.tensorflow.op.math.Expm1;
-import org.tensorflow.op.math.Fact;
-import org.tensorflow.op.math.Floor;
-import org.tensorflow.op.math.FloorDiv;
-import org.tensorflow.op.math.FloorMod;
-import org.tensorflow.op.math.Greater;
-import org.tensorflow.op.math.GreaterEqual;
-import org.tensorflow.op.math.Igamma;
-import org.tensorflow.op.math.Igammac;
-import org.tensorflow.op.math.Imag;
-import org.tensorflow.op.math.InvertPermutation;
-import org.tensorflow.op.math.IsFinite;
-import org.tensorflow.op.math.IsInf;
-import org.tensorflow.op.math.IsNan;
-import org.tensorflow.op.math.Less;
-import org.tensorflow.op.math.LessEqual;
-import org.tensorflow.op.math.Lgamma;
-import org.tensorflow.op.math.Log;
-import org.tensorflow.op.math.Log1p;
-import org.tensorflow.op.math.LogicalAnd;
-import org.tensorflow.op.math.LogicalNot;
-import org.tensorflow.op.math.LogicalOr;
-import org.tensorflow.op.math.Maximum;
-import org.tensorflow.op.math.Mean;
-import org.tensorflow.op.math.Minimum;
-import org.tensorflow.op.math.Mod;
-import org.tensorflow.op.math.Mul;
-import org.tensorflow.op.math.MulNoNan;
-import org.tensorflow.op.math.Ndtri;
-import org.tensorflow.op.math.Neg;
-import org.tensorflow.op.math.NextAfter;
-import org.tensorflow.op.math.NotEqual;
-import org.tensorflow.op.math.Polygamma;
-import org.tensorflow.op.math.PopulationCount;
-import org.tensorflow.op.math.Pow;
-import org.tensorflow.op.math.QuantizedAdd;
-import org.tensorflow.op.math.QuantizedMul;
-import org.tensorflow.op.math.Real;
-import org.tensorflow.op.math.RealDiv;
-import org.tensorflow.op.math.Reciprocal;
-import org.tensorflow.op.math.Rint;
-import org.tensorflow.op.math.Round;
-import org.tensorflow.op.math.Rsqrt;
-import org.tensorflow.op.math.SegmentMax;
-import org.tensorflow.op.math.SegmentMean;
-import org.tensorflow.op.math.SegmentMin;
-import org.tensorflow.op.math.SegmentProd;
-import org.tensorflow.op.math.SegmentSum;
-import org.tensorflow.op.math.Sigmoid;
-import org.tensorflow.op.math.Sign;
-import org.tensorflow.op.math.Sin;
-import org.tensorflow.op.math.Sinh;
-import org.tensorflow.op.math.Softplus;
-import org.tensorflow.op.math.Sqrt;
-import org.tensorflow.op.math.Square;
-import org.tensorflow.op.math.SquaredDifference;
-import org.tensorflow.op.math.Sub;
-import org.tensorflow.op.math.Tan;
-import org.tensorflow.op.math.Tanh;
-import org.tensorflow.op.math.TruncateDiv;
-import org.tensorflow.op.math.TruncateMod;
-import org.tensorflow.op.math.UnsortedSegmentMax;
-import org.tensorflow.op.math.UnsortedSegmentMin;
-import org.tensorflow.op.math.UnsortedSegmentProd;
-import org.tensorflow.op.math.UnsortedSegmentSum;
-import org.tensorflow.op.math.Xdivy;
-import org.tensorflow.op.math.Xlog1py;
-import org.tensorflow.op.math.Xlogy;
-import org.tensorflow.op.math.Zeta;
-import org.tensorflow.op.math.erfinv;
-import org.tensorflow.types.TBool;
-import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * An API for building {@code math} operations as {@link Op Op}s
- *
- * @see {@link Ops}
- */
-public final class MathOps {
-  private final Scope scope;
-
-  MathOps(Scope scope) {
-    this.scope = scope;
-  }
-
-  /**
-   * Computes the absolute value of a tensor.
-   *  <p>
-   *  Given a tensor `x`, this operation returns a tensor containing the absolute
-   *  value of each element in `x`. For example, if x is an input element and y is
-   *  an output element, this operation computes \\(y = |x|\\).
-   *
-   * @param <T> data type for {@code y()} output
-   * @param x
-   * @return a new instance of Abs
-   */
-  public <T extends TNumber> Abs<T> abs(Operand<T> x) {
-    return Abs.create(scope, x);
-  }
-
-  /**
-   * Returns the element-wise sum of a list of tensors.
-   *  <p>
-   *  `tf.accumulate_n_v2` performs the same operation as `tf.add_n`, but does not
-   *  wait for all of its inputs to be ready before beginning to sum. This can
-   *  save memory if inputs are ready at different times, since minimum temporary
-   *  storage is proportional to the output size rather than the inputs size.
-   *  <p>
-   *  Unlike the original `accumulate_n`, `accumulate_n_v2` is differentiable.
-   *  <p>
-   *  Returns a `Tensor` of same shape and type as the elements of `inputs`.
-   *
-   * @param <T> data type for {@code sum()} output
-   * @param inputs A list of `Tensor` objects, each with same shape and type.
-   * @param shape Shape of elements of `inputs`.
-   * @return a new instance of AccumulateN
-   */
-  public <T extends TType> AccumulateN<T> accumulateN(Iterable<Operand<T>> inputs, Shape shape) {
-    return AccumulateN.create(scope, inputs, shape);
-  }
-
-  /**
-   * Computes acos of x element-wise.
-   *
-   * @param <T> data type for {@code y()} output
-   * @param x
-   * @return a new instance of Acos
-   */
-  public <T extends TType> Acos<T> acos(Operand<T> x) {
-    return Acos.create(scope, x);
-  }
-
-  /**
-   * Computes inverse hyperbolic cosine of x element-wise.
-   *  <p>
-   *  Given an input tensor, the function computes inverse hyperbolic cosine of every element.
-   *  Input range is `[1, inf]`. It returns `nan` if the input lies outside the range.
-   *  <pre>{@code
-   *  x = tf.constant([-2, -0.5, 1, 1.2, 200, 10000, float("inf")])
-   *  tf.math.acosh(x) ==> [nan nan 0. 0.62236255 5.9914584 9.903487 inf]
-   *  }</pre>
-   *
-   * @param <T> data type for {@code y()} output
-   * @param x
-   * @return a new instance of Acosh
-   */
-  public <T extends TType> Acosh<T> acosh(Operand<T> x) {
-    return Acosh.create(scope, x);
-  }
-
-  /**
-   * Returns x + y element-wise.
-   *  <p>
-   *  <i>NOTE</i>: `math.Add` supports broadcasting. `AddN` does not. More about broadcasting
-   *  [here](http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html)
-   *
-   * @param <T> data type for {@code z()} output
-   * @param x
-   * @param y
-   * @return a new instance of Add
-   */
-  public <T extends TType> Add<T> add(Operand<T> x, Operand<T> y) {
-    return Add.create(scope, x, y);
-  }
-
-  /**
-   * Add all input tensors element wise.
-   *  <p>
-   *    Inputs must be of same size and shape.
-   *  <p>
-   *    <pre>{@code
-   *    x = [9, 7, 10]
-   *    tf.math.add_n(x) ==> 26
-   *    }</pre>
-   *
-   * @param <T> data type for {@code sum()} output
-   * @param inputs
-   * @return a new instance of AddN
-   */
-  public <T extends TType> AddN<T> addN(Iterable<Operand<T>> inputs) {
-    return AddN.create(scope, inputs);
-  }
-
-  /**
-   * Returns the argument of a complex number.
-   *  <p>
-   *  Given a tensor `input` of complex numbers, this operation returns a tensor of
-   *  type `float` that is the argument of each element in `input`. All elements in
-   *  `input` must be complex numbers of the form \\(a + bj\\), where <i>a</i>
-   *  is the real part and <i>b</i> is the imaginary part.
-   *  <p>
-   *  The argument returned by this operation is of the form \\(atan2(b, a)\\).
-   *  <p>
-   *  For example:
-   *  <pre>{@code
-   *  # tensor 'input' is [-2.25 + 4.75j, 3.25 + 5.75j]
-   *  tf.angle(input) ==> [2.0132, 1.056]
-   *  }</pre>
-   *
-   * @compatibility(numpy) Equivalent to np.angle.
-   * @end_compatibility
-   * @param <U> data type for {@code output()} output
-   * @param input
-   * @return a new instance of Angle
-   */
-  public <T extends TType> Angle<TFloat32> angle(Operand<T> input) {
-    return Angle.create(scope, input);
-  }
-
-  /**
-   * Returns the argument of a complex number.
-   *  <p>
-   *  Given a tensor `input` of complex numbers, this operation returns a tensor of
-   *  type `float` that is the argument of each element in `input`. All elements in
-   *  `input` must be complex numbers of the form \\(a + bj\\), where <i>a</i>
-   *  is the real part and <i>b</i> is the imaginary part.
-   *  <p>
-   *  The argument returned by this operation is of the form \\(atan2(b, a)\\).
-   *  <p>
-   *  For example:
-   *  <pre>{@code
-   *  # tensor 'input' is [-2.25 + 4.75j, 3.25 + 5.75j]
-   *  tf.angle(input) ==> [2.0132, 1.056]
-   *  }</pre>
-   *
-   * @compatibility(numpy) Equivalent to np.angle.
-   * @end_compatibility
-   * @param <U> data type for {@code output()} output
-   * @param input
-   * @param Tout
-   * @return a new instance of Angle
-   */
-  public <U extends TNumber, T extends TType> Angle<U> angle(Operand<T> input, DataType<U> Tout) {
-    return Angle.create(scope, input, Tout);
-  }
-
-  /**
-   * Returns the truth value of abs(x-y) < tolerance element-wise.
-   *
-   * @param x
-   * @param y
-   * @param options carries optional attributes values
-   * @return a new instance of ApproximateEqual
-   */
-  public <T extends TType> ApproximateEqual approximateEqual(Operand<T> x, Operand<T> y,
-      ApproximateEqual.Options... options) {
-    return ApproximateEqual.create(scope, x, y, options);
-  }
-
-  /**
-   * Returns the index with the largest value across dimensions of a tensor.
-   *  <p>
-   *  Note that in case of ties the identity of the return value is not guaranteed.
-   *  <p>
-   *  Usage:
-   *    <pre>{@code
-   *    import tensorflow as tf
-   *    a = [1, 10, 26.9, 2.8, 166.32, 62.3]
-   *    b = tf.math.argmax(input = a)
-   *    c = tf.keras.backend.eval(b)
-   *    # c = 4
-   *    # here a[4] = 166.32 which is the largest element of a across axis 0
-   *    }</pre>
-   *
-   * @param <V> data type for {@code output()} output
-   * @param input
-   * @param dimension int32 or int64, must be in the range `[-rank(input), rank(input))`.
-   *  Describes which dimension of the input Tensor to reduce across. For vectors,
-   *  use dimension = 0.
-   * @return a new instance of ArgMax
-   */
-  public <T extends TType, U extends TNumber> ArgMax<TInt64> argMax(Operand<T> input,
-      Operand<U> dimension) {
-    return ArgMax.create(scope, input, dimension);
-  }
-
-  /**
-   * Returns the index with the largest value across dimensions of a tensor.
-   *  <p>
-   *  Note that in case of ties the identity of the return value is not guaranteed.
-   *  <p>
-   *  Usage:
-   *    <pre>{@code
-   *    import tensorflow as tf
-   *    a = [1, 10, 26.9, 2.8, 166.32, 62.3]
-   *    b = tf.math.argmax(input = a)
-   *    c = tf.keras.backend.eval(b)
-   *    # c = 4
-   *    # here a[4] = 166.32 which is the largest element of a across axis 0
-   *    }</pre>
-   *
-   * @param <V> data type for {@code output()} output
-   * @param input
-   * @param dimension int32 or int64, must be in the range `[-rank(input), rank(input))`.
-   *  Describes which dimension of the input Tensor to reduce across. For vectors,
-   *  use dimension = 0.
-   * @param outputType
-   * @return a new instance of ArgMax
-   */
-  public <V extends TNumber, T extends TType, U extends TNumber> ArgMax<V> argMax(Operand<T> input,
-      Operand<U> dimension, DataType<V> outputType) {
-    return ArgMax.create(scope, input, dimension, outputType);
-  }
-
-  /**
-   * Returns the index with the smallest value across dimensions of a tensor.
-   *  <p>
-   *  Note that in case of ties the identity of the return value is not guaranteed.
-   *  <p>
-   *  Usage:
-   *    <pre>{@code
-   *    import tensorflow as tf
-   *    a = [1, 10, 26.9, 2.8, 166.32, 62.3]
-   *    b = tf.math.argmin(input = a)
-   *    c = tf.keras.backend.eval(b)
-   *    # c = 0
-   *    # here a[0] = 1 which is the smallest element of a across axis 0
-   *    }</pre>
-   *
-   * @param <V> data type for {@code output()} output
-   * @param input
-   * @param dimension int32 or int64, must be in the range `[-rank(input), rank(input))`.
-   *  Describes which dimension of the input Tensor to reduce across. For vectors,
-   *  use dimension = 0.
-   * @return a new instance of ArgMin
-   */
-  public <T extends TType, U extends TNumber> ArgMin<TInt64> argMin(Operand<T> input,
-      Operand<U> dimension) {
-    return ArgMin.create(scope, input, dimension);
-  }
-
-  /**
-   * Returns the index with the smallest value across dimensions of a tensor.
-   *  <p>
-   *  Note that in case of ties the identity of the return value is not guaranteed.
-   *  <p>
-   *  Usage:
-   *    <pre>{@code
-   *    import tensorflow as tf
-   *    a = [1, 10, 26.9, 2.8, 166.32, 62.3]
-   *    b = tf.math.argmin(input = a)
-   *    c = tf.keras.backend.eval(b)
-   *    # c = 0
-   *    # here a[0] = 1 which is the smallest element of a across axis 0
-   *    }</pre>
-   *
-   * @param <V> data type for {@code output()} output
-   * @param input
-   * @param dimension int32 or int64, must be in the range `[-rank(input), rank(input))`.
-   *  Describes which dimension of the input Tensor to reduce across. For vectors,
-   *  use dimension = 0.
-   * @param outputType
-   * @return a new instance of ArgMin
-   */
-  public <V extends TNumber, T extends TType, U extends TNumber> ArgMin<V> argMin(Operand<T> input,
-      Operand<U> dimension, DataType<V> outputType) {
-    return ArgMin.create(scope, input, dimension, outputType);
-  }
-
-  /**
-   * Computes the trignometric inverse sine of x element-wise.
-   *  <p>
-   *  The `tf.math.asin` operation returns the inverse of `tf.math.sin`, such that
-   *  if `y = tf.math.sin(x)` then, `x = tf.math.asin(y)`.
-   *  <p>
-   *  <b>Note</b>: The output of `tf.math.asin` will lie within the invertible range
-   *  of sine, i.e [-pi/2, pi/2].
-   *  <p>
-   *  For example:
-   *  <pre>{@code
-   *  # Note: [1.047, 0.785] ~= [(pi/3), (pi/4)]
-   *  x = tf.constant([1.047, 0.785])
-   *  y = tf.math.sin(x) # [0.8659266, 0.7068252]
-   *
-   *  tf.math.asin(y) # [1.047, 0.785] = x
-   *  }</pre>
-   *
-   * @param <T> data type for {@code y()} output
-   * @param x
-   * @return a new instance of Asin
-   */
-  public <T extends TType> Asin<T> asin(Operand<T> x) {
-    return Asin.create(scope, x);
-  }
-
-  /**
-   * Computes inverse hyperbolic sine of x element-wise.
-   *  <p>
-   *    Given an input tensor, this function computes inverse hyperbolic sine
-   *    for every element in the tensor. Both input and output has a range of
-   *    `[-inf, inf]`.
-   *  <p>
-   *    <pre>{@code
-   *    x = tf.constant([-float("inf"), -2, -0.5, 1, 1.2, 200, 10000, float("inf")])
-   *    tf.math.asinh(x) ==> [-inf -1.4436355 -0.4812118 0.8813736 1.0159732 5.991471 9.903487 inf]
-   *    }</pre>
-   *
-   * @param <T> data type for {@code y()} output
-   * @param x
-   * @return a new instance of Asinh
-   */
-  public <T extends TType> Asinh<T> asinh(Operand<T> x) {
-    return Asinh.create(scope, x);
-  }
-
-  /**
-   * Computes the trignometric inverse tangent of x element-wise.
-   *  <p>
-   *  The `tf.math.atan` operation returns the inverse of `tf.math.tan`, such that
-   *  if `y = tf.math.tan(x)` then, `x = tf.math.atan(y)`.
-   *  <p>
-   *  <b>Note</b>: The output of `tf.math.atan` will lie within the invertible range
-   *  of tan, i.e (-pi/2, pi/2).
-   *  <p>
-   *  For example:
-   *  <pre>{@code
-   *  # Note: [1.047, 0.785] ~= [(pi/3), (pi/4)]
-   *  x = tf.constant([1.047, 0.785])
-   *  y = tf.math.tan(x) # [1.731261, 0.99920404]
-   *
-   *  tf.math.atan(y) # [1.047, 0.785] = x
-   *  }</pre>
-   *
-   * @param <T> data type for {@code y()} output
-   * @param x
-   * @return a new instance of Atan
-   */
-  public <T extends TType> Atan<T> atan(Operand<T> x) {
-    return Atan.create(scope, x);
-  }
-
-  /**
-   * Computes arctangent of `y/x` element-wise, respecting signs of the arguments.
-   *  <p>
-   *  This is the angle \( \theta \in [-\pi, \pi] \) such that
-   *  \[ x = r \cos(\theta) \]
-   *  and
-   *  \[ y = r \sin(\theta) \]
-   *  where \(r = \sqrt(x^2 + y^2) \).
-   *
-   * @param <T> data type for {@code z()} output
-   * @param y
-   * @param x
-   * @return a new instance of Atan2
-   */
-  public <T extends TNumber> Atan2<T> atan2(Operand<T> y, Operand<T> x) {
-    return Atan2.create(scope, y, x);
-  }
-
-  /**
-   * Computes inverse hyperbolic tangent of x element-wise.
-   *  <p>
-   *    Given an input tensor, this function computes inverse hyperbolic tangent
-   *    for every element in the tensor. Input range is `[-1,1]` and output range is
-   *    `[-inf, inf]`. If input is `-1`, output will be `-inf` and if the
-   *    input is `1`, output will be `inf`. Values outside the range will have
-   *    `nan` as output.
-   *  <p>
-   *    <pre>{@code
-   *    x = tf.constant([-float("inf"), -1, -0.5, 1, 0, 0.5, 10, float("inf")])
-   *    tf.math.atanh(x) ==> [nan -inf -0.54930615 inf  0. 0.54930615 nan nan]
-   *    }</pre>
-   *
-   * @param <T> data type for {@code y()} output
-   * @param x
-   * @return a new instance of Atanh
-   */
-  public <T extends TType> Atanh<T> atanh(Operand<T> x) {
-    return Atanh.create(scope, x);
-  }
-
-  /**
-   * Computes the Bessel i0e function of `x` element-wise.
-   *  <p>
-   *  Exponentially scaled modified Bessel function of order 0 defined as
-   *  `bessel_i0e(x) = exp(-abs(x)) bessel_i0(x)`.
-   *  <p>
-   *  This function is faster and numerically stabler than `bessel_i0(x)`.
-   *
-   * @param <T> data type for {@code y()} output
-   * @param x
-   * @return a new instance of BesselI0e
-   */
-  public <T extends TNumber> BesselI0e<T> besselI0e(Operand<T> x) {
-    return BesselI0e.create(scope, x);
-  }
-
-  /**
-   * Computes the Bessel i1e function of `x` element-wise.
-   *  <p>
-   *  Exponentially scaled modified Bessel function of order 0 defined as
-   *  `bessel_i1e(x) = exp(-abs(x)) bessel_i1(x)`.
-   *  <p>
-   *  This function is faster and numerically stabler than `bessel_i1(x)`.
-   *
-   * @param <T> data type for {@code y()} output
-   * @param x
-   * @return a new instance of BesselI1e
-   */
-  public <T extends TNumber> BesselI1e<T> besselI1e(Operand<T> x) {
-    return BesselI1e.create(scope, x);
-  }
-
-  /**
-   * Compute the regularized incomplete beta integral \\(I_x(a, b)\\).
-   *  <p>
-   *  The regularized incomplete beta integral is defined as:
-   *  <p>
-   *  \\(I_x(a, b) = \frac{B(x; a, b)}{B(a, b)}\\)
-   *  <p>
-   *  where
-   *  <p>
-   *  \\(B(x; a, b) = \int_0^x t^{a-1} (1 - t)^{b-1} dt\\)
-   *  <p>
-   *  is the incomplete beta function and \\(B(a, b)\\) is the <i>complete</i>
-   *  beta function.
-   *
-   * @param <T> data type for {@code z()} output
-   * @param a
-   * @param b
-   * @param x
-   * @return a new instance of Betainc
-   */
-  public <T extends TNumber> Betainc<T> betainc(Operand<T> a, Operand<T> b, Operand<T> x) {
-    return Betainc.create(scope, a, b, x);
-  }
-
-  /**
-   * Counts the number of occurrences of each value in an integer array.
-   *  <p>
-   *  Outputs a vector with length `size` and the same dtype as `weights`. If
-   *  `weights` are empty, then index `i` stores the number of times the value `i` is
-   *  counted in `arr`. If `weights` are non-empty, then index `i` stores the sum of
-   *  the value in `weights` at each index where the corresponding value in `arr` is
-   *  `i`.
-   *  <p>
-   *  Values in `arr` outside of the range [0, size) are ignored.
-   *
-   * @param <T> data type for {@code bins()} output
-   * @param arr int32 `Tensor`.
-   * @param size non-negative int32 scalar `Tensor`.
-   * @param weights is an int32, int64, float32, or float64 `Tensor` with the same
-   *  shape as `arr`, or a length-0 `Tensor`, in which case it acts as all weights
-   *  equal to 1.
-   * @return a new instance of Bincount
-   */
-  public <T extends TNumber> Bincount<T> bincount(Operand<TInt32> arr, Operand<TInt32> size,
-      Operand<T> weights) {
-    return Bincount.create(scope, arr, size, weights);
-  }
-
-  /**
-   * Returns element-wise smallest integer not less than x.
-   *
-   * @param <T> data type for {@code y()} output
-   * @param x
-   * @return a new instance of Ceil
-   */
-  public <T extends TNumber> Ceil<T> ceil(Operand<T> x) {
-    return Ceil.create(scope, x);
-  }
-
-  /**
-   * Compare values of `input` to `threshold` and pack resulting bits into a `uint8`.
-   *  <p>
-   *  Each comparison returns a boolean `true` (if `input_value > threshold`)
-   *  or and `false` otherwise.
-   *  <p>
-   *  This operation is useful for Locality-Sensitive-Hashing (LSH) and other
-   *  algorithms that use hashing approximations of cosine and `L2` distances;
-   *  codes can be generated from an input via:
-   *  <pre>{@code
-   *  codebook_size = 50
-   *  codebook_bits = codebook_size * 32
-   *  codebook = tf.get_variable('codebook', [x.shape[-1].value, codebook_bits],
-   *                             dtype=x.dtype,
-   *                             initializer=tf.orthogonal_initializer())
-   *  codes = compare_and_threshold(tf.matmul(x, codebook), threshold=0.)
-   *  codes = tf.bitcast(codes, tf.int32)  # go from uint8 to int32
-   *  # now codes has shape x.shape[:-1] + [codebook_size]
-   *  }</pre>
-   *  <b>NOTE</b>: Currently, the innermost dimension of the tensor must be divisible
-   *  by 8.
-   *  <p>
-   *  Given an `input` shaped `[s0, s1, ..., s_n]`, the output is
-   *  a `uint8` tensor shaped `[s0, s1, ..., s_n / 8]`.
-   *
-   * @param input Values to compare against `threshold` and bitpack.
-   * @param threshold Threshold to compare against.
-   * @return a new instance of CompareAndBitpack
-   */
-  public <T extends TType> CompareAndBitpack compareAndBitpack(Operand<T> input,
-      Operand<T> threshold) {
-    return CompareAndBitpack.create(scope, input, threshold);
-  }
-
-  /**
-   * Computes the complex absolute value of a tensor.
-   *  <p>
-   *  Given a tensor `x` of complex numbers, this operation returns a tensor of type
-   *  `float` or `double` that is the absolute value of each element in `x`. All
-   *  elements in `x` must be complex numbers of the form \\(a + bj\\). The absolute
-   *  value is computed as \\( \sqrt{a^2 + b^2}\\).
-   *
-   * @param <U> data type for {@code y()} output
-   * @param x
-   * @return a new instance of ComplexAbs
-   */
-  public <T extends TType> ComplexAbs<TFloat32> complexAbs(Operand<T> x) {
-    return ComplexAbs.create(scope, x);
-  }
-
-  /**
-   * Computes the complex absolute value of a tensor.
-   *  <p>
-   *  Given a tensor `x` of complex numbers, this operation returns a tensor of type
-   *  `float` or `double` that is the absolute value of each element in `x`. All
-   *  elements in `x` must be complex numbers of the form \\(a + bj\\). The absolute
-   *  value is computed as \\( \sqrt{a^2 + b^2}\\).
-   *
-   * @param <U> data type for {@code y()} output
-   * @param x
-   * @param Tout
-   * @return a new instance of ComplexAbs
-   */
-  public <U extends TNumber, T extends TType> ComplexAbs<U> complexAbs(Operand<T> x,
-      DataType<U> Tout) {
-    return ComplexAbs.create(scope, x, Tout);
-  }
-
-  /**
-   * Returns the complex conjugate of a complex number.
-   *  <p>
-   *  Given a tensor `input` of complex numbers, this operation returns a tensor of
-   *  complex numbers that are the complex conjugate of each element in `input`. The
-   *  complex numbers in `input` must be of the form \\(a + bj\\), where <i>a</i> is the
-   *  real part and <i>b</i> is the imaginary part.
-   *  <p>
-   *  The complex conjugate returned by this operation is of the form \\(a - bj\\).
-   *  <p>
-   *  For example:
-   *  <pre>{@code
-   *  # tensor 'input' is [-2.25 + 4.75j, 3.25 + 5.75j]
-   *  tf.conj(input) ==> [-2.25 - 4.75j, 3.25 - 5.75j]
-   *  }</pre>
-   *
-   * @param <T> data type for {@code output()} output
-   * @param input
-   * @return a new instance of Conj
-   */
-  public <T extends TType> Conj<T> conj(Operand<T> input) {
-    return Conj.create(scope, input);
-  }
-
-  /**
-   * Computes cos of x element-wise.
-   *  <p>
-   *    Given an input tensor, this function computes cosine of every
-   *    element in the tensor. Input range is `(-inf, inf)` and
-   *    output range is `[-1,1]`. If input lies outside the boundary, `nan`
-   *    is returned.
-   *  <p>
-   *    <pre>{@code
-   *    x = tf.constant([-float("inf"), -9, -0.5, 1, 1.2, 200, 10000, float("inf")])
-   *    tf.math.cos(x) ==> [nan -0.91113025 0.87758255 0.5403023 0.36235774 0.48718765 -0.95215535 nan]
-   *    }</pre>
-   *
-   * @param <T> data type for {@code y()} output
-   * @param x
-   * @return a new instance of Cos
-   */
-  public <T extends TType> Cos<T> cos(Operand<T> x) {
-    return Cos.create(scope, x);
-  }
-
-  /**
-   * Computes hyperbolic cosine of x element-wise.
-   *  <p>
-   *    Given an input tensor, this function computes hyperbolic cosine of every
-   *    element in the tensor. Input range is `[-inf, inf]` and output range
-   *    is `[1, inf]`.
-   *  <p>
-   *    <pre>{@code
-   *    x = tf.constant([-float("inf"), -9, -0.5, 1, 1.2, 2, 10, float("inf")])
-   *    tf.math.cosh(x) ==> [inf 4.0515420e+03 1.1276259e+00 1.5430807e+00 1.8106556e+00 3.7621956e+00 1.1013233e+04 inf]
-   *    }</pre>
-   *
-   * @param <T> data type for {@code y()} output
-   * @param x
-   * @return a new instance of Cosh
-   */
-  public <T extends TType> Cosh<T> cosh(Operand<T> x) {
-    return Cosh.create(scope, x);
-  }
-
-  /**
-   * Compute the cumulative product of the tensor `x` along `axis`.
-   *  <p>
-   *  By default, this op performs an inclusive cumprod, which means that the first
-   *  element of the input is identical to the first element of the output:
-   *  <pre>{@code
-   *  tf.cumprod([a, b, c])  # => [a, a * b, a * b * c]
-   *  }</pre>
-   *  By setting the `exclusive` kwarg to `True`, an exclusive cumprod is
-   *  performed instead:
-   *  <pre>{@code
-   *  tf.cumprod([a, b, c], exclusive=True)  # => [1, a, a * b]
-   *  }</pre>
-   *  By setting the `reverse` kwarg to `True`, the cumprod is performed in the
-   *  opposite direction:
-   *  <pre>{@code
-   *  tf.cumprod([a, b, c], reverse=True)  # => [a * b * c, b * c, c]
-   *  }</pre>
-   *  This is more efficient than using separate `tf.reverse` ops.
-   *  <p>
-   *  The `reverse` and `exclusive` kwargs can also be combined:
-   *  <pre>{@code
-   *  tf.cumprod([a, b, c], exclusive=True, reverse=True)  # => [b * c, c, 1]
-   *  }</pre>
-   *
-   * @param <T> data type for {@code out()} output
-   * @param x A `Tensor`. Must be one of the following types: `float32`, `float64`,
-   *  `int64`, `int32`, `uint8`, `uint16`, `int16`, `int8`, `complex64`,
-   *  `complex128`, `qint8`, `quint8`, `qint32`, `half`.
-   * @param axis A `Tensor` of type `int32` (default: 0). Must be in the range
-   *  `[-rank(x), rank(x))`.
-   * @param options carries optional attributes values
-   * @return a new instance of Cumprod
-   */
-  public <T extends TType, U extends TNumber> Cumprod<T> cumprod(Operand<T> x, Operand<U> axis,
-      Cumprod.Options... options) {
-    return Cumprod.create(scope, x, axis, options);
-  }
-
-  /**
-   * Compute the cumulative sum of the tensor `x` along `axis`.
-   *  <p>
-   *  By default, this op performs an inclusive cumsum, which means that the first
-   *  element of the input is identical to the first element of the output:
-   *  <pre>{@code
-   *  tf.cumsum([a, b, c])  # => [a, a + b, a + b + c]
-   *  }</pre>
-   *  By setting the `exclusive` kwarg to `True`, an exclusive cumsum is
-   *  performed instead:
-   *  <pre>{@code
-   *  tf.cumsum([a, b, c], exclusive=True)  # => [0, a, a + b]
-   *  }</pre>
-   *  By setting the `reverse` kwarg to `True`, the cumsum is performed in the
-   *  opposite direction:
-   *  <pre>{@code
-   *  tf.cumsum([a, b, c], reverse=True)  # => [a + b + c, b + c, c]
-   *  }</pre>
-   *  This is more efficient than using separate `tf.reverse` ops.
-   *  <p>
-   *  The `reverse` and `exclusive` kwargs can also be combined:
-   *  <pre>{@code
-   *  tf.cumsum([a, b, c], exclusive=True, reverse=True)  # => [b + c, c, 0]
-   *  }</pre>
-   *
-   * @param <T> data type for {@code out()} output
-   * @param x A `Tensor`. Must be one of the following types: `float32`, `float64`,
-   *  `int64`, `int32`, `uint8`, `uint16`, `int16`, `int8`, `complex64`,
-   *  `complex128`, `qint8`, `quint8`, `qint32`, `half`.
-   * @param axis A `Tensor` of type `int32` (default: 0). Must be in the range
-   *  `[-rank(x), rank(x))`.
-   * @param options carries optional attributes values
-   * @return a new instance of Cumsum
-   */
-  public <T extends TType, U extends TNumber> Cumsum<T> cumsum(Operand<T> x, Operand<U> axis,
-      Cumsum.Options... options) {
-    return Cumsum.create(scope, x, axis, options);
-  }
-
-  /**
-   * Computes Psi, the derivative of Lgamma (the log of the absolute value of
-   *  <p>
-   *  `Gamma(x)`), element-wise.
-   *
-   * @param <T> data type for {@code y()} output
-   * @param x
-   * @return a new instance of Digamma
-   */
-  public <T extends TNumber> Digamma<T> digamma(Operand<T> x) {
-    return Digamma.create(scope, x);
-  }
-
-  /**
-   * Returns x / y element-wise.
-   *  <p>
-   *  <i>NOTE</i>: `math.Div` supports broadcasting. More about broadcasting
-   *  [here](http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html)
-   *
-   * @param <T> data type for {@code z()} output
-   * @param x
-   * @param y
-   * @return a new instance of Div
-   */
-  public <T extends TType> Div<T> div(Operand<T> x, Operand<T> y) {
-    return Div.create(scope, x, y);
-  }
-
-  /**
-   * Returns 0 if the denominator is zero.
-   *  <p>
-   *
-   *  <i>NOTE</i>: `math.DivNoNan` supports broadcasting. More about broadcasting
-   *  [here](http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html)
-   *
-   * @param <T> data type for {@code z()} output
-   * @param x
-   * @param y
-   * @return a new instance of DivNoNan
-   */
-  public <T extends TType> DivNoNan<T> divNoNan(Operand<T> x, Operand<T> y) {
-    return DivNoNan.create(scope, x, y);
-  }
-
-  /**
-   * Returns the truth value of (x == y) element-wise.
-   *  <p>
-   *  <i>NOTE</i>: `math.Equal` supports broadcasting. More about broadcasting
-   *  [here](http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html)
-   *  <pre>{@code
-   *  x = tf.constant([2, 4])
-   *  y = tf.constant(2)
-   *  tf.math.equal(x, y) ==> array([True, False])
-   *
-   *  x = tf.constant([2, 4])
-   *  y = tf.constant([2, 4])
-   *  tf.math.equal(x, y) ==> array([True,  True])
-   *  }</pre>
-   *
-   * @param x
-   * @param y
-   * @param options carries optional attributes values
-   * @return a new instance of Equal
-   */
-  public <T extends TType> Equal equal(Operand<T> x, Operand<T> y, Equal.Options... options) {
-    return Equal.create(scope, x, y, options);
-  }
-
-  /**
-   * Computes the Gauss error function of `x` element-wise.
-   *
-   * @param <T> data type for {@code y()} output
-   * @param x
-   * @return a new instance of Erf
-   */
-  public <T extends TNumber> Erf<T> erf(Operand<T> x) {
-    return Erf.create(scope, x);
-  }
-
-  /**
-   * Computes the complementary error function of `x` element-wise.
-   *
-   * @param <T> data type for {@code y()} output
-   * @param x
-   * @return a new instance of Erfc
-   */
-  public <T extends TNumber> Erfc<T> erfc(Operand<T> x) {
-    return Erfc.create(scope, x);
-  }
-
-  /**
-   *
-   * @param <T> data type for {@code y()} output
-   * @param x
-   * @return a new instance of erfinv
-   */
-  public <T extends TNumber> erfinv<T> erfinv(Operand<T> x) {
-    return erfinv.create(scope, x);
-  }
-
-  /**
-   * Computes exponential of x element-wise.  \\(y = e^x\\).
-   *  <p>
-   *    This function computes the exponential of every element in the input tensor.
-   *    i.e. `exp(x)` or `e^(x)`, where `x` is the input tensor.
-   *    `e` denotes Euler's number and is approximately equal to 2.718281.
-   *    Output is positive for any real input.
-   *  <p>
-   *    <pre>{@code
-   *    x = tf.constant(2.0)
-   *    tf.math.exp(x) ==> 7.389056
-   *
-   *    x = tf.constant([2.0, 8.0])
-   *    tf.math.exp(x) ==> array([7.389056, 2980.958], dtype=float32)
-   *    }</pre>
-   *  For complex numbers, the exponential value is calculated as follows:
-   *  <p>
-   *    <pre>{@code
-   *    e^(x+iy) = e^x * e^iy = e^x * (cos y + i sin y)
-   *    }</pre>
-   *  Let's consider complex number 1+1j as an example.
-   *    e^1 * (cos 1 + i sin 1) = 2.7182818284590 * (0.54030230586+0.8414709848j)
-   *  <p>
-   *    <pre>{@code
-   *    x = tf.constant(1 + 1j)
-   *    tf.math.exp(x) ==> 1.4686939399158851+2.2873552871788423j
-   *    }</pre>
-   *
-   * @param <T> data type for {@code y()} output
-   * @param x
-   * @return a new instance of Exp
-   */
-  public <T extends TType> Exp<T> exp(Operand<T> x) {
-    return Exp.create(scope, x);
-  }
-
-  /**
-   * Computes `exp(x) - 1` element-wise.
-   *  <p>
-   *    i.e. `exp(x) - 1` or `e^(x) - 1`, where `x` is the input tensor.
-   *    `e` denotes Euler's number and is approximately equal to 2.718281.
-   *  <p>
-   *    <pre>{@code
-   *    x = tf.constant(2.0)
-   *    tf.math.expm1(x) ==> 6.389056
-   *
-   *    x = tf.constant([2.0, 8.0])
-   *    tf.math.expm1(x) ==> array([6.389056, 2979.958], dtype=float32)
-   *
-   *    x = tf.constant(1 + 1j)
-   *    tf.math.expm1(x) ==> (0.46869393991588515+2.2873552871788423j)
-   *    }</pre>
-   *
-   * @param <T> data type for {@code y()} output
-   * @param x
-   * @return a new instance of Expm1
-   */
-  public <T extends TType> Expm1<T> expm1(Operand<T> x) {
-    return Expm1.create(scope, x);
-  }
-
-  /**
-   * Output a fact about factorials.
-   *
-   * @return a new instance of Fact
-   */
-  public Fact fact() {
-    return Fact.create(scope);
-  }
-
-  /**
-   * Returns element-wise largest integer not greater than x.
-   *
-   * @param <T> data type for {@code y()} output
-   * @param x
-   * @return a new instance of Floor
-   */
-  public <T extends TNumber> Floor<T> floor(Operand<T> x) {
-    return Floor.create(scope, x);
-  }
-
-  /**
-   * Returns x // y element-wise.
-   *  <p>
-   *  <i>NOTE</i>: `math.FloorDiv` supports broadcasting. More about broadcasting
-   *  [here](http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html)
-   *
-   * @param <T> data type for {@code z()} output
-   * @param x
-   * @param y
-   * @return a new instance of FloorDiv
-   */
-  public <T extends TType> FloorDiv<T> floorDiv(Operand<T> x, Operand<T> y) {
-    return FloorDiv.create(scope, x, y);
-  }
-
-  /**
-   * Returns element-wise remainder of division. When `x < 0` xor `y < 0` is
-   *  <p>
-   *  true, this follows Python semantics in that the result here is consistent
-   *  with a flooring divide. E.g. `floor(x / y) * y + mod(x, y) = x`.
-   *  <p>
-   *  <i>NOTE</i>: `math.FloorMod` supports broadcasting. More about broadcasting
-   *  [here](http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html)
-   *
-   * @param <T> data type for {@code z()} output
-   * @param x
-   * @param y
-   * @return a new instance of FloorMod
-   */
-  public <T extends TNumber> FloorMod<T> floorMod(Operand<T> x, Operand<T> y) {
-    return FloorMod.create(scope, x, y);
-  }
-
-  /**
-   * Returns the truth value of (x > y) element-wise.
-   *  <p>
-   *  <i>NOTE</i>: `math.Greater` supports broadcasting. More about broadcasting
-   *  [here](http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html)
-   *  <p>
-   *  Example:
-   *  <pre>{@code
-   *  x = tf.constant([5, 4, 6])
-   *  y = tf.constant([5, 2, 5])
-   *  tf.math.greater(x, y) ==> [False, True, True]
-   *
-   *  x = tf.constant([5, 4, 6])
-   *  y = tf.constant([5])
-   *  tf.math.greater(x, y) ==> [False, False, True]
-   *  }</pre>
-   *
-   * @param x
-   * @param y
-   * @return a new instance of Greater
-   */
-  public <T extends TNumber> Greater greater(Operand<T> x, Operand<T> y) {
-    return Greater.create(scope, x, y);
-  }
-
-  /**
-   * Returns the truth value of (x >= y) element-wise.
-   *  <p>
-   *  <i>NOTE</i>: `math.GreaterEqual` supports broadcasting. More about broadcasting
-   *  [here](http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html)
-   *  <p>
-   *  Example:
-   *  <pre>{@code
-   *  x = tf.constant([5, 4, 6, 7])
-   *  y = tf.constant([5, 2, 5, 10])
-   *  tf.math.greater_equal(x, y) ==> [True, True, True, False]
-   *
-   *  x = tf.constant([5, 4, 6, 7])
-   *  y = tf.constant([5])
-   *  tf.math.greater_equal(x, y) ==> [True, False, True, True]
-   *  }</pre>
-   *
-   * @param x
-   * @param y
-   * @return a new instance of GreaterEqual
-   */
-  public <T extends TNumber> GreaterEqual greaterEqual(Operand<T> x, Operand<T> y) {
-    return GreaterEqual.create(scope, x, y);
-  }
-
-  /**
-   * Compute the lower regularized incomplete Gamma function `P(a, x)`.
-   *  <p>
-   *  The lower regularized incomplete Gamma function is defined as:
-   *  <p>
-   *  \\(P(a, x) = gamma(a, x) / Gamma(a) = 1 - Q(a, x)\\)
-   *  <p>
-   *  where
-   *  <p>
-   *  \\(gamma(a, x) = \\int_{0}^{x} t^{a-1} exp(-t) dt\\)
-   *  <p>
-   *  is the lower incomplete Gamma function.
-   *  <p>
-   *  Note, above `Q(a, x)` (`Igammac`) is the upper regularized complete
-   *  Gamma function.
-   *
-   * @param <T> data type for {@code z()} output
-   * @param a
-   * @param x
-   * @return a new instance of Igamma
-   */
-  public <T extends TNumber> Igamma<T> igamma(Operand<T> a, Operand<T> x) {
-    return Igamma.create(scope, a, x);
-  }
-
-  /**
-   * Compute the upper regularized incomplete Gamma function `Q(a, x)`.
-   *  <p>
-   *  The upper regularized incomplete Gamma function is defined as:
-   *  <p>
-   *  \\(Q(a, x) = Gamma(a, x) / Gamma(a) = 1 - P(a, x)\\)
-   *  <p>
-   *  where
-   *  <p>
-   *  \\(Gamma(a, x) = int_{x}^{\infty} t^{a-1} exp(-t) dt\\)
-   *  <p>
-   *  is the upper incomplete Gama function.
-   *  <p>
-   *  Note, above `P(a, x)` (`Igamma`) is the lower regularized complete
-   *  Gamma function.
-   *
-   * @param <T> data type for {@code z()} output
-   * @param a
-   * @param x
-   * @return a new instance of Igammac
-   */
-  public <T extends TNumber> Igammac<T> igammac(Operand<T> a, Operand<T> x) {
-    return Igammac.create(scope, a, x);
-  }
-
-  /**
-   * Returns the imaginary part of a complex number.
-   *  <p>
-   *  Given a tensor `input` of complex numbers, this operation returns a tensor of
-   *  type `float` that is the imaginary part of each element in `input`. All
-   *  elements in `input` must be complex numbers of the form \\(a + bj\\), where <i>a</i>
-   *  is the real part and <i>b</i> is the imaginary part returned by this operation.
-   *  <p>
-   *  For example:
-   *  <pre>{@code
-   *  # tensor 'input' is [-2.25 + 4.75j, 3.25 + 5.75j]
-   *  tf.imag(input) ==> [4.75, 5.75]
-   *  }</pre>
-   *
-   * @param <U> data type for {@code output()} output
-   * @param input
-   * @return a new instance of Imag
-   */
-  public <T extends TType> Imag<TFloat32> imag(Operand<T> input) {
-    return Imag.create(scope, input);
-  }
-
-  /**
-   * Returns the imaginary part of a complex number.
-   *  <p>
-   *  Given a tensor `input` of complex numbers, this operation returns a tensor of
-   *  type `float` that is the imaginary part of each element in `input`. All
-   *  elements in `input` must be complex numbers of the form \\(a + bj\\), where <i>a</i>
-   *  is the real part and <i>b</i> is the imaginary part returned by this operation.
-   *  <p>
-   *  For example:
-   *  <pre>{@code
-   *  # tensor 'input' is [-2.25 + 4.75j, 3.25 + 5.75j]
-   *  tf.imag(input) ==> [4.75, 5.75]
-   *  }</pre>
-   *
-   * @param <U> data type for {@code output()} output
-   * @param input
-   * @param Tout
-   * @return a new instance of Imag
-   */
-  public <U extends TNumber, T extends TType> Imag<U> imag(Operand<T> input, DataType<U> Tout) {
-    return Imag.create(scope, input, Tout);
-  }
-
-  /**
-   * Computes the inverse permutation of a tensor.
-   *  <p>
-   *  This operation computes the inverse of an index permutation. It takes a 1-D
-   *  integer tensor `x`, which represents the indices of a zero-based array, and
-   *  swaps each value with its index position. In other words, for an output tensor
-   *  `y` and an input tensor `x`, this operation computes the following:
-   *  <p>
-   *  `y[x[i]] = i for i in [0, 1, ..., len(x) - 1]`
-   *  <p>
-   *  The values must include 0. There can be no duplicate values or negative values.
-   *  <p>
-   *  For example:
-   *  <pre>{@code
-   *  # tensor `x` is [3, 4, 0, 2, 1]
-   *  invert_permutation(x) ==> [2, 4, 3, 0, 1]
-   *  }</pre>
-   *
-   * @param <T> data type for {@code y()} output
-   * @param x 1-D.
-   * @return a new instance of InvertPermutation
-   */
-  public <T extends TNumber> InvertPermutation<T> invertPermutation(Operand<T> x) {
-    return InvertPermutation.create(scope, x);
-  }
-
-  /**
-   * Returns which elements of x are finite.
-   *  <p>
-   *
-   * @compatibility(numpy) Equivalent to np.isfinite
-   * @end_compatibility <p>
-   *  Example:
-   *  <pre>{@code
-   *  x = tf.constant([5.0, 4.8, 6.8, np.inf, np.nan])
-   *  tf.math.is_finite(x) ==> [True, True, True, False, False]
-   *  }</pre>
-   * @param x
-   * @return a new instance of IsFinite
-   */
-  public <T extends TNumber> IsFinite isFinite(Operand<T> x) {
-    return IsFinite.create(scope, x);
-  }
-
-  /**
-   * Returns which elements of x are Inf.
-   *  <p>
-   *
-   * @compatibility(numpy) Equivalent to np.isinf
-   * @end_compatibility <p>
-   *  Example:
-   *  <pre>{@code
-   *  x = tf.constant([5.0, np.inf, 6.8, np.inf])
-   *  tf.math.is_inf(x) ==> [False, True, False, True]
-   *  }</pre>
-   * @param x
-   * @return a new instance of IsInf
-   */
-  public <T extends TNumber> IsInf isInf(Operand<T> x) {
-    return IsInf.create(scope, x);
-  }
-
-  /**
-   * Returns which elements of x are NaN.
-   *  <p>
-   *
-   * @compatibility(numpy) Equivalent to np.isnan
-   * @end_compatibility <p>
-   *  Example:
-   *  <pre>{@code
-   *  x = tf.constant([5.0, np.nan, 6.8, np.nan, np.inf])
-   *  tf.math.is_nan(x) ==> [False, True, False, True, False]
-   *  }</pre>
-   * @param x
-   * @return a new instance of IsNan
-   */
-  public <T extends TNumber> IsNan isNan(Operand<T> x) {
-    return IsNan.create(scope, x);
-  }
-
-  /**
-   * Returns the truth value of (x < y) element-wise.
-   *  <p>
-   *  <i>NOTE</i>: `math.Less` supports broadcasting. More about broadcasting
-   *  [here](http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html)
-   *  <p>
-   *  Example:
-   *  <pre>{@code
-   *  x = tf.constant([5, 4, 6])
-   *  y = tf.constant([5])
-   *  tf.math.less(x, y) ==> [False, True, False]
-   *
-   *  x = tf.constant([5, 4, 6])
-   *  y = tf.constant([5, 6, 7])
-   *  tf.math.less(x, y) ==> [False, True, True]
-   *  }</pre>
-   *
-   * @param x
-   * @param y
-   * @return a new instance of Less
-   */
-  public <T extends TNumber> Less less(Operand<T> x, Operand<T> y) {
-    return Less.create(scope, x, y);
-  }
-
-  /**
-   * Returns the truth value of (x <= y) element-wise.
-   *  <p>
-   *  <i>NOTE</i>: `math.LessEqual` supports broadcasting. More about broadcasting
-   *  [here](http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html)
-   *  <p>
-   *  Example:
-   *  <pre>{@code
-   *  x = tf.constant([5, 4, 6])
-   *  y = tf.constant([5])
-   *  tf.math.less_equal(x, y) ==> [True, True, False]
-   *
-   *  x = tf.constant([5, 4, 6])
-   *  y = tf.constant([5, 6, 6])
-   *  tf.math.less_equal(x, y) ==> [True, True, True]
-   *  }</pre>
-   *
-   * @param x
-   * @param y
-   * @return a new instance of LessEqual
-   */
-  public <T extends TNumber> LessEqual lessEqual(Operand<T> x, Operand<T> y) {
-    return LessEqual.create(scope, x, y);
-  }
-
-  /**
-   * Computes the log of the absolute value of `Gamma(x)` element-wise.
-   *  <p>
-   *    For positive numbers, this function computes log((input - 1)!) for every element in the tensor.
-   *    `lgamma(5) = log((5-1)!) = log(4!) = log(24) = 3.1780539`
-   *  <p>
-   *  Example:
-   *  <pre>{@code
-   *  x = tf.constant([0, 0.5, 1, 4.5, -4, -5.6])
-   *  tf.math.lgamma(x) ==> [inf, 0.5723649, 0., 2.4537368, inf, -4.6477685]
-   *  }</pre>
-   *
-   * @param <T> data type for {@code y()} output
-   * @param x
-   * @return a new instance of Lgamma
-   */
-  public <T extends TNumber> Lgamma<T> lgamma(Operand<T> x) {
-    return Lgamma.create(scope, x);
-  }
-
-  /**
-   * Computes natural logarithm of x element-wise.
-   *  <p>
-   *  I.e., \\(y = \log_e x\\).
-   *  <p>
-   *  Example:
-   *  <pre>{@code
-   *  x = tf.constant([0, 0.5, 1, 5])
-   *  tf.math.log(x) ==> [-inf, -0.6931472,  0. ,  1.609438]
-   *  }</pre>
-   *
-   * @param <T> data type for {@code y()} output
-   * @param x
-   * @return a new instance of Log
-   */
-  public <T extends TType> Log<T> log(Operand<T> x) {
-    return Log.create(scope, x);
-  }
-
-  /**
-   * Computes natural logarithm of (1 + x) element-wise.
-   *  <p>
-   *  I.e., \\(y = \log_e (1 + x)\\).
-   *  <p>
-   *  Example:
-   *  <pre>{@code
-   *  x = tf.constant([0, 0.5, 1, 5])
-   *  tf.math.log1p(x) ==> [0., 0.4054651, 0.6931472, 1.7917595]
-   *  }</pre>
-   *
-   * @param <T> data type for {@code y()} output
-   * @param x
-   * @return a new instance of Log1p
-   */
-  public <T extends TType> Log1p<T> log1p(Operand<T> x) {
-    return Log1p.create(scope, x);
-  }
-
-  /**
-   * Returns the truth value of x AND y element-wise.
-   *  <p>
-   *  <i>NOTE</i>: `math.LogicalAnd` supports broadcasting. More about broadcasting
-   *  [here](http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html)
-   *
-   * @param x
-   * @param y
-   * @return a new instance of LogicalAnd
-   */
-  public LogicalAnd logicalAnd(Operand<TBool> x, Operand<TBool> y) {
-    return LogicalAnd.create(scope, x, y);
-  }
-
-  /**
-   * Returns the truth value of `NOT x` element-wise.
-   *
-   * @param x A `Tensor` of type `bool`.
-   * @return a new instance of LogicalNot
-   */
-  public LogicalNot logicalNot(Operand<TBool> x) {
-    return LogicalNot.create(scope, x);
-  }
-
-  /**
-   * Returns the truth value of x OR y element-wise.
-   *  <p>
-   *  <i>NOTE</i>: `math.LogicalOr` supports broadcasting. More about broadcasting
-   *  [here](http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html)
-   *
-   * @param x
-   * @param y
-   * @return a new instance of LogicalOr
-   */
-  public LogicalOr logicalOr(Operand<TBool> x, Operand<TBool> y) {
-    return LogicalOr.create(scope, x, y);
-  }
-
-  /**
-   * Returns the max of x and y (i.e. x > y ? x : y) element-wise.
-   *  <p>
-   *  <i>NOTE</i>: `math.Maximum` supports broadcasting. More about broadcasting
-   *  [here](http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html)
-   *
-   * @param <T> data type for {@code z()} output
-   * @param x
-   * @param y
-   * @return a new instance of Maximum
-   */
-  public <T extends TNumber> Maximum<T> maximum(Operand<T> x, Operand<T> y) {
-    return Maximum.create(scope, x, y);
-  }
-
-  /**
-   * Computes the mean of elements across dimensions of a tensor.
-   *  <p>
-   *  Reduces `input` along the dimensions given in `axis`. Unless
-   *  `keep_dims` is true, the rank of the tensor is reduced by 1 for each entry in
-   *  `axis`. If `keep_dims` is true, the reduced dimensions are
-   *  retained with length 1.
-   *
-   * @param <T> data type for {@code output()} output
-   * @param input The tensor to reduce.
-   * @param axis The dimensions to reduce. Must be in the range
-   *  `[-rank(input), rank(input))`.
-   * @param options carries optional attributes values
-   * @return a new instance of Mean
-   */
-  public <T extends TType, U extends TNumber> Mean<T> mean(Operand<T> input, Operand<U> axis,
-      Mean.Options... options) {
-    return Mean.create(scope, input, axis, options);
-  }
-
-  /**
-   * Returns the min of x and y (i.e. x < y ? x : y) element-wise.
-   *  <p>
-   *  <i>NOTE</i>: `math.Minimum` supports broadcasting. More about broadcasting
-   *  [here](http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html)
-   *
-   * @param <T> data type for {@code z()} output
-   * @param x
-   * @param y
-   * @return a new instance of Minimum
-   */
-  public <T extends TNumber> Minimum<T> minimum(Operand<T> x, Operand<T> y) {
-    return Minimum.create(scope, x, y);
-  }
-
-  /**
-   * Returns element-wise remainder of division. This emulates C semantics in that
-   *  <p>
-   *  the result here is consistent with a truncating divide. E.g.
-   *  `tf.truncatediv(x, y) * y + truncate_mod(x, y) = x`.
-   *  <p>
-   *  <i>NOTE</i>: `math.Mod` supports broadcasting. More about broadcasting
-   *  [here](http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html)
-   *
-   * @param <T> data type for {@code z()} output
-   * @param x
-   * @param y
-   * @return a new instance of Mod
-   */
-  public <T extends TNumber> Mod<T> mod(Operand<T> x, Operand<T> y) {
-    return Mod.create(scope, x, y);
-  }
-
-  /**
-   * Returns x * y element-wise.
-   *  <p>
-   *  <i>NOTE</i>: `math.Mul` supports broadcasting. More about broadcasting
-   *  [here](http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html)
-   *
-   * @param <T> data type for {@code z()} output
-   * @param x
-   * @param y
-   * @return a new instance of Mul
-   */
-  public <T extends TType> Mul<T> mul(Operand<T> x, Operand<T> y) {
-    return Mul.create(scope, x, y);
-  }
-
-  /**
-   * Returns x * y element-wise. Returns zero if y is zero, even if x if infinite or NaN.
-   *  <p>
-   *  <i>NOTE</i>: `math.MulNoNan` supports broadcasting. More about broadcasting
-   *  [here](http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html)
-   *
-   * @param <T> data type for {@code z()} output
-   * @param x
-   * @param y
-   * @return a new instance of MulNoNan
-   */
-  public <T extends TType> MulNoNan<T> mulNoNan(Operand<T> x, Operand<T> y) {
-    return MulNoNan.create(scope, x, y);
-  }
-
-  /**
-   *
-   * @param <T> data type for {@code y()} output
-   * @param x
-   * @return a new instance of Ndtri
-   */
-  public <T extends TNumber> Ndtri<T> ndtri(Operand<T> x) {
-    return Ndtri.create(scope, x);
-  }
-
-  /**
-   * Computes numerical negative value element-wise.
-   *  <p>
-   *  I.e., \\(y = -x\\).
-   *
-   * @param <T> data type for {@code y()} output
-   * @param x
-   * @return a new instance of Neg
-   */
-  public <T extends TType> Neg<T> neg(Operand<T> x) {
-    return Neg.create(scope, x);
-  }
-
-  /**
-   * Returns the next representable value of `x1` in the direction of `x2`, element-wise.
-   *  <p>
-   *  This operation returns the same result as the C++ std::nextafter function.
-   *  <p>
-   *  It can also return a subnormal number.
-   *  <p>
-   *
-   * @compatibility(cpp) Equivalent to C++ std::nextafter function.
-   * @end_compatibility
-   * @param <T> data type for {@code output()} output
-   * @param x1
-   * @param x2
-   * @return a new instance of NextAfter
-   */
-  public <T extends TNumber> NextAfter<T> nextAfter(Operand<T> x1, Operand<T> x2) {
-    return NextAfter.create(scope, x1, x2);
-  }
-
-  /**
-   * Returns the truth value of (x != y) element-wise.
-   *  <p>
-   *  <i>NOTE</i>: `math.NotEqual` supports broadcasting. More about broadcasting
-   *  [here](http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html)
-   *
-   * @param x
-   * @param y
-   * @param options carries optional attributes values
-   * @return a new instance of NotEqual
-   */
-  public <T extends TType> NotEqual notEqual(Operand<T> x, Operand<T> y,
-      NotEqual.Options... options) {
-    return NotEqual.create(scope, x, y, options);
-  }
-
-  /**
-   * Compute the polygamma function \\(\psi^{(n)}(x)\\).
-   *  <p>
-   *  The polygamma function is defined as:
-   *  <p>
-   *  \\(\psi^{(a)}(x) = \frac{d^a}{dx^a} \psi(x)\\)
-   *  <p>
-   *  where \\(\psi(x)\\) is the digamma function.
-   *  The polygamma function is defined only for non-negative integer orders \\a\\.
-   *
-   * @param <T> data type for {@code z()} output
-   * @param a
-   * @param x
-   * @return a new instance of Polygamma
-   */
-  public <T extends TNumber> Polygamma<T> polygamma(Operand<T> a, Operand<T> x) {
-    return Polygamma.create(scope, a, x);
-  }
-
-  /**
-   * Computes element-wise population count (a.k.a. popcount, bitsum, bitcount).
-   *  <p>
-   *  For each entry in `x`, calculates the number of `1` (on) bits in the binary
-   *  representation of that entry.
-   *  <p>
-   *  <b>NOTE</b>: It is more efficient to first `tf.bitcast` your tensors into
-   *  `int32` or `int64` and perform the bitcount on the result, than to feed in
-   *  8- or 16-bit inputs and then aggregate the resulting counts.
-   *
-   * @param x
-   * @return a new instance of PopulationCount
-   */
-  public <T extends TNumber> PopulationCount populationCount(Operand<T> x) {
-    return PopulationCount.create(scope, x);
-  }
-
-  /**
-   * Computes the power of one value to another.
-   *  <p>
-   *  Given a tensor `x` and a tensor `y`, this operation computes \\(x^y\\) for
-   *  corresponding elements in `x` and `y`. For example:
-   *  <pre>{@code
-   *  # tensor 'x' is [[2, 2]], [3, 3]]
-   *  # tensor 'y' is [[8, 16], [2, 3]]
-   *  tf.pow(x, y) ==> [[256, 65536], [9, 27]]
-   *  }</pre>
-   *
-   * @param <T> data type for {@code z()} output
-   * @param x
-   * @param y
-   * @return a new instance of Pow
-   */
-  public <T extends TType> Pow<T> pow(Operand<T> x, Operand<T> y) {
-    return Pow.create(scope, x, y);
-  }
-
-  /**
-   * Returns x + y element-wise, working on quantized buffers.
-   *
-   * @param <V> data type for {@code z()} output
-   * @param x
-   * @param y
-   * @param minX The float value that the lowest quantized `x` value represents.
-   * @param maxX The float value that the highest quantized `x` value represents.
-   * @param minY The float value that the lowest quantized `y` value represents.
-   * @param maxY The float value that the highest quantized `y` value represents.
-   * @param Toutput
-   * @return a new instance of QuantizedAdd
-   */
-  public <V extends TType, T extends TType, U extends TType> QuantizedAdd<V> quantizedAdd(
-      Operand<T> x, Operand<U> y, Operand<TFloat32> minX, Operand<TFloat32> maxX,
-      Operand<TFloat32> minY, Operand<TFloat32> maxY, DataType<V> Toutput) {
-    return QuantizedAdd.create(scope, x, y, minX, maxX, minY, maxY, Toutput);
-  }
-
-  /**
-   * Returns x * y element-wise, working on quantized buffers.
-   *
-   * @param <V> data type for {@code z()} output
-   * @param x
-   * @param y
-   * @param minX The float value that the lowest quantized `x` value represents.
-   * @param maxX The float value that the highest quantized `x` value represents.
-   * @param minY The float value that the lowest quantized `y` value represents.
-   * @param maxY The float value that the highest quantized `y` value represents.
-   * @param Toutput
-   * @return a new instance of QuantizedMul
-   */
-  public <V extends TType, T extends TType, U extends TType> QuantizedMul<V> quantizedMul(
-      Operand<T> x, Operand<U> y, Operand<TFloat32> minX, Operand<TFloat32> maxX,
-      Operand<TFloat32> minY, Operand<TFloat32> maxY, DataType<V> Toutput) {
-    return QuantizedMul.create(scope, x, y, minX, maxX, minY, maxY, Toutput);
-  }
-
-  /**
-   * Returns the real part of a complex number.
-   *  <p>
-   *  Given a tensor `input` of complex numbers, this operation returns a tensor of
-   *  type `float` that is the real part of each element in `input`. All elements in
-   *  `input` must be complex numbers of the form \\(a + bj\\), where <i>a</i> is the real
-   *   part returned by this operation and <i>b</i> is the imaginary part.
-   *  <p>
-   *  For example:
-   *  <pre>{@code
-   *  # tensor 'input' is [-2.25 + 4.75j, 3.25 + 5.75j]
-   *  tf.real(input) ==> [-2.25, 3.25]
-   *  }</pre>
-   *
-   * @param <U> data type for {@code output()} output
-   * @param input
-   * @return a new instance of Real
-   */
-  public <T extends TType> Real<TFloat32> real(Operand<T> input) {
-    return Real.create(scope, input);
-  }
-
-  /**
-   * Returns the real part of a complex number.
-   *  <p>
-   *  Given a tensor `input` of complex numbers, this operation returns a tensor of
-   *  type `float` that is the real part of each element in `input`. All elements in
-   *  `input` must be complex numbers of the form \\(a + bj\\), where <i>a</i> is the real
-   *   part returned by this operation and <i>b</i> is the imaginary part.
-   *  <p>
-   *  For example:
-   *  <pre>{@code
-   *  # tensor 'input' is [-2.25 + 4.75j, 3.25 + 5.75j]
-   *  tf.real(input) ==> [-2.25, 3.25]
-   *  }</pre>
-   *
-   * @param <U> data type for {@code output()} output
-   * @param input
-   * @param Tout
-   * @return a new instance of Real
-   */
-  public <U extends TNumber, T extends TType> Real<U> real(Operand<T> input, DataType<U> Tout) {
-    return Real.create(scope, input, Tout);
-  }
-
-  /**
-   * Returns x / y element-wise for real types.
-   *  <p>
-   *  If `x` and `y` are reals, this will return the floating-point division.
-   *  <p>
-   *  <i>NOTE</i>: `Div` supports broadcasting. More about broadcasting
-   *  [here](http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html)
-   *
-   * @param <T> data type for {@code z()} output
-   * @param x
-   * @param y
-   * @return a new instance of RealDiv
-   */
-  public <T extends TType> RealDiv<T> realDiv(Operand<T> x, Operand<T> y) {
-    return RealDiv.create(scope, x, y);
-  }
-
-  /**
-   * Computes the reciprocal of x element-wise.
-   *  <p>
-   *  I.e., \\(y = 1 / x\\).
-   *
-   * @param <T> data type for {@code y()} output
-   * @param x
-   * @return a new instance of Reciprocal
-   */
-  public <T extends TType> Reciprocal<T> reciprocal(Operand<T> x) {
-    return Reciprocal.create(scope, x);
-  }
-
-  /**
-   * Returns element-wise integer closest to x.
-   *  <p>
-   *  If the result is midway between two representable values,
-   *  the even representable is chosen.
-   *  For example:
-   *  <pre>{@code
-   *  rint(-1.5) ==> -2.0
-   *  rint(0.5000001) ==> 1.0
-   *  rint([-1.7, -1.5, -0.2, 0.2, 1.5, 1.7, 2.0]) ==> [-2., -2., -0., 0., 2., 2., 2.]
-   *  }</pre>
-   *
-   * @param <T> data type for {@code y()} output
-   * @param x
-   * @return a new instance of Rint
-   */
-  public <T extends TNumber> Rint<T> rint(Operand<T> x) {
-    return Rint.create(scope, x);
-  }
-
-  /**
-   * Rounds the values of a tensor to the nearest integer, element-wise.
-   *  <p>
-   *  Rounds half to even.  Also known as bankers rounding. If you want to round
-   *  according to the current system rounding mode use std::cint.
-   *
-   * @param <T> data type for {@code y()} output
-   * @param x
-   * @return a new instance of Round
-   */
-  public <T extends TType> Round<T> round(Operand<T> x) {
-    return Round.create(scope, x);
-  }
-
-  /**
-   * Computes reciprocal of square root of x element-wise.
-   *  <p>
-   *  I.e., \\(y = 1 / \sqrt{x}\\).
-   *
-   * @param <T> data type for {@code y()} output
-   * @param x
-   * @return a new instance of Rsqrt
-   */
-  public <T extends TType> Rsqrt<T> rsqrt(Operand<T> x) {
-    return Rsqrt.create(scope, x);
-  }
-
-  /**
-   * Computes the maximum along segments of a tensor.
-   *  <p>
-   *  Read
-   *  [the section on segmentation](https://tensorflow.org/api_docs/python/tf/math#Segmentation)
-   *  for an explanation of segments.
-   *  <p>
-   *  Computes a tensor such that
-   *  \\(output_i = \max_j(data_j)\\) where `max` is over `j` such
-   *  that `segment_ids[j] == i`.
-   *  <p>
-   *  If the max is empty for a given segment ID `i`, `output[i] = 0`.
-   *  <p>
-   *  <div style="width:70%; margin:auto; margin-bottom:10px; margin-top:20px;">
-   *  <img style="width:100%" src="https://www.tensorflow.org/images/SegmentMax.png" alt>
-   *  </div>
-   *  <p>
-   *  For example:
-   *  <pre>{@code
-   *  c = tf.constant([[1,2,3,4], [4, 3, 2, 1], [5,6,7,8]])
-   *  tf.segment_max(c, tf.constant([0, 0, 1]))
-   *  # ==> [[4, 3, 3, 4],
-   *  #      [5, 6, 7, 8]]
-   *  }</pre>
-   *
-   * @param <T> data type for {@code output()} output
-   * @param data
-   * @param segmentIds A 1-D tensor whose size is equal to the size of `data`'s
-   *  first dimension.  Values should be sorted and can be repeated.
-   * @return a new instance of SegmentMax
-   */
-  public <T extends TNumber, U extends TNumber> SegmentMax<T> segmentMax(Operand<T> data,
-      Operand<U> segmentIds) {
-    return SegmentMax.create(scope, data, segmentIds);
-  }
-
-  /**
-   * Computes the mean along segments of a tensor.
-   *  <p>
-   *  Read
-   *  [the section on segmentation](https://tensorflow.org/api_docs/python/tf/math#Segmentation)
-   *  for an explanation of segments.
-   *  <p>
-   *  Computes a tensor such that
-   *  \\(output_i = \frac{\sum_j data_j}{N}\\) where `mean` is
-   *  over `j` such that `segment_ids[j] == i` and `N` is the total number of
-   *  values summed.
-   *  <p>
-   *  If the mean is empty for a given segment ID `i`, `output[i] = 0`.
-   *  <p>
-   *  <div style="width:70%; margin:auto; margin-bottom:10px; margin-top:20px;">
-   *  <img style="width:100%" src="https://www.tensorflow.org/images/SegmentMean.png" alt>
-   *  </div>
-   *  <p>
-   *  For example:
-   *  <pre>{@code
-   *  c = tf.constant([[1.0,2,3,4], [4, 3, 2, 1], [5,6,7,8]])
-   *  tf.segment_mean(c, tf.constant([0, 0, 1]))
-   *  # ==> [[2.5, 2.5, 2.5, 2.5],
-   *  #      [5, 6, 7, 8]]
-   *  }</pre>
-   *
-   * @param <T> data type for {@code output()} output
-   * @param data
-   * @param segmentIds A 1-D tensor whose size is equal to the size of `data`'s
-   *  first dimension.  Values should be sorted and can be repeated.
-   * @return a new instance of SegmentMean
-   */
-  public <T extends TType, U extends TNumber> SegmentMean<T> segmentMean(Operand<T> data,
-      Operand<U> segmentIds) {
-    return SegmentMean.create(scope, data, segmentIds);
-  }
-
-  /**
-   * Computes the minimum along segments of a tensor.
-   *  <p>
-   *  Read
-   *  [the section on segmentation](https://tensorflow.org/api_docs/python/tf/math#Segmentation)
-   *  for an explanation of segments.
-   *  <p>
-   *  Computes a tensor such that
-   *  \\(output_i = \min_j(data_j)\\) where `min` is over `j` such
-   *  that `segment_ids[j] == i`.
-   *  <p>
-   *  If the min is empty for a given segment ID `i`, `output[i] = 0`.
-   *  <p>
-   *  <div style="width:70%; margin:auto; margin-bottom:10px; margin-top:20px;">
-   *  <img style="width:100%" src="https://www.tensorflow.org/images/SegmentMin.png" alt>
-   *  </div>
-   *  <p>
-   *  For example:
-   *  <pre>{@code
-   *  c = tf.constant([[1,2,3,4], [4, 3, 2, 1], [5,6,7,8]])
-   *  tf.segment_min(c, tf.constant([0, 0, 1]))
-   *  # ==> [[1, 2, 2, 1],
-   *  #      [5, 6, 7, 8]]
-   *  }</pre>
-   *
-   * @param <T> data type for {@code output()} output
-   * @param data
-   * @param segmentIds A 1-D tensor whose size is equal to the size of `data`'s
-   *  first dimension.  Values should be sorted and can be repeated.
-   * @return a new instance of SegmentMin
-   */
-  public <T extends TNumber, U extends TNumber> SegmentMin<T> segmentMin(Operand<T> data,
-      Operand<U> segmentIds) {
-    return SegmentMin.create(scope, data, segmentIds);
-  }
-
-  /**
-   * Computes the product along segments of a tensor.
-   *  <p>
-   *  Read
-   *  [the section on segmentation](https://tensorflow.org/api_docs/python/tf/math#Segmentation)
-   *  for an explanation of segments.
-   *  <p>
-   *  Computes a tensor such that
-   *  \\(output_i = \prod_j data_j\\) where the product is over `j` such
-   *  that `segment_ids[j] == i`.
-   *  <p>
-   *  If the product is empty for a given segment ID `i`, `output[i] = 1`.
-   *  <p>
-   *  <div style="width:70%; margin:auto; margin-bottom:10px; margin-top:20px;">
-   *  <img style="width:100%" src="https://www.tensorflow.org/images/SegmentProd.png" alt>
-   *  </div>
-   *  <p>
-   *  For example:
-   *  <pre>{@code
-   *  c = tf.constant([[1,2,3,4], [4, 3, 2, 1], [5,6,7,8]])
-   *  tf.segment_prod(c, tf.constant([0, 0, 1]))
-   *  # ==> [[4, 6, 6, 4],
-   *  #      [5, 6, 7, 8]]
-   *  }</pre>
-   *
-   * @param <T> data type for {@code output()} output
-   * @param data
-   * @param segmentIds A 1-D tensor whose size is equal to the size of `data`'s
-   *  first dimension.  Values should be sorted and can be repeated.
-   * @return a new instance of SegmentProd
-   */
-  public <T extends TType, U extends TNumber> SegmentProd<T> segmentProd(Operand<T> data,
-      Operand<U> segmentIds) {
-    return SegmentProd.create(scope, data, segmentIds);
-  }
-
-  /**
-   * Computes the sum along segments of a tensor.
-   *  <p>
-   *  Read
-   *  [the section on segmentation](https://tensorflow.org/api_docs/python/tf/math#Segmentation)
-   *  for an explanation of segments.
-   *  <p>
-   *  Computes a tensor such that
-   *  \\(output_i = \sum_j data_j\\) where sum is over `j` such
-   *  that `segment_ids[j] == i`.
-   *  <p>
-   *  If the sum is empty for a given segment ID `i`, `output[i] = 0`.
-   *  <p>
-   *  <div style="width:70%; margin:auto; margin-bottom:10px; margin-top:20px;">
-   *  <img style="width:100%" src="https://www.tensorflow.org/images/SegmentSum.png" alt>
-   *  </div>
-   *  <p>
-   *  For example:
-   *  <pre>{@code
-   *  c = tf.constant([[1,2,3,4], [4, 3, 2, 1], [5,6,7,8]])
-   *  tf.segment_sum(c, tf.constant([0, 0, 1]))
-   *  # ==> [[5, 5, 5, 5],
-   *  #      [5, 6, 7, 8]]
-   *  }</pre>
-   *
-   * @param <T> data type for {@code output()} output
-   * @param data
-   * @param segmentIds A 1-D tensor whose size is equal to the size of `data`'s
-   *  first dimension.  Values should be sorted and can be repeated.
-   * @return a new instance of SegmentSum
-   */
-  public <T extends TType, U extends TNumber> SegmentSum<T> segmentSum(Operand<T> data,
-      Operand<U> segmentIds) {
-    return SegmentSum.create(scope, data, segmentIds);
-  }
-
-  /**
-   * Computes sigmoid of `x` element-wise.
-   *  <p>
-   *  Specifically, `y = 1 / (1 + exp(-x))`.
-   *
-   * @param <T> data type for {@code y()} output
-   * @param x
-   * @return a new instance of Sigmoid
-   */
-  public <T extends TType> Sigmoid<T> sigmoid(Operand<T> x) {
-    return Sigmoid.create(scope, x);
-  }
-
-  /**
-   * Returns an element-wise indication of the sign of a number.
-   *  <p>
-   *  `y = sign(x) = -1` if `x < 0`; 0 if `x == 0`; 1 if `x > 0`.
-   *  <p>
-   *  For complex numbers, `y = sign(x) = x / |x|` if `x != 0`, otherwise `y = 0`.
-   *  <p>
-   *  Example usage:
-   *  >>> tf.math.sign([0., 2., -3.])
-   *  <tf.Tensor: shape=(3,), dtype=float32, numpy=array([ 0.,  1., -1.], dtype=float32)>
-   *
-   * @param <T> data type for {@code y()} output
-   * @param x
-   * @return a new instance of Sign
-   */
-  public <T extends TType> Sign<T> sign(Operand<T> x) {
-    return Sign.create(scope, x);
-  }
-
-  /**
-   * Computes sine of x element-wise.
-   *  <p>
-   *    Given an input tensor, this function computes sine of every
-   *    element in the tensor. Input range is `(-inf, inf)` and
-   *    output range is `[-1,1]`.
-   *  <p>
-   *    <pre>{@code
-   *    x = tf.constant([-float("inf"), -9, -0.5, 1, 1.2, 200, 10, float("inf")])
-   *    tf.math.sin(x) ==> [nan -0.4121185 -0.47942555 0.84147096 0.9320391 -0.87329733 -0.54402107 nan]
-   *    }</pre>
-   *
-   * @param <T> data type for {@code y()} output
-   * @param x
-   * @return a new instance of Sin
-   */
-  public <T extends TType> Sin<T> sin(Operand<T> x) {
-    return Sin.create(scope, x);
-  }
-
-  /**
-   * Computes hyperbolic sine of x element-wise.
-   *  <p>
-   *    Given an input tensor, this function computes hyperbolic sine of every
-   *    element in the tensor. Input range is `[-inf,inf]` and output range
-   *    is `[-inf,inf]`.
-   *  <p>
-   *    <pre>{@code
-   *    x = tf.constant([-float("inf"), -9, -0.5, 1, 1.2, 2, 10, float("inf")])
-   *    tf.math.sinh(x) ==> [-inf -4.0515420e+03 -5.2109528e-01 1.1752012e+00 1.5094614e+00 3.6268604e+00 1.1013232e+04 inf]
-   *    }</pre>
-   *
-   * @param <T> data type for {@code y()} output
-   * @param x
-   * @return a new instance of Sinh
-   */
-  public <T extends TType> Sinh<T> sinh(Operand<T> x) {
-    return Sinh.create(scope, x);
-  }
-
-  /**
-   * Computes softplus: `log(exp(features) + 1)`.
-   *
-   * @param <T> data type for {@code activations()} output
-   * @param features
-   * @return a new instance of Softplus
-   */
-  public <T extends TNumber> Softplus<T> softplus(Operand<T> features) {
-    return Softplus.create(scope, features);
-  }
-
-  /**
-   * Computes square root of x element-wise.
-   *  <p>
-   *  I.e., \\(y = \sqrt{x} = x^{1/2}\\).
-   *
-   * @param <T> data type for {@code y()} output
-   * @param x
-   * @return a new instance of Sqrt
-   */
-  public <T extends TType> Sqrt<T> sqrt(Operand<T> x) {
-    return Sqrt.create(scope, x);
-  }
-
-  /**
-   * Computes square of x element-wise.
-   *  <p>
-   *  I.e., \\(y = x * x = x^2\\).
-   *
-   * @param <T> data type for {@code y()} output
-   * @param x
-   * @return a new instance of Square
-   */
-  public <T extends TType> Square<T> square(Operand<T> x) {
-    return Square.create(scope, x);
-  }
-
-  /**
-   * Returns (x - y)(x - y) element-wise.
-   *  <p>
-   *  <i>NOTE</i>: `math.SquaredDifference` supports broadcasting. More about broadcasting
-   *  [here](http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html)
-   *
-   * @param <T> data type for {@code z()} output
-   * @param x
-   * @param y
-   * @return a new instance of SquaredDifference
-   */
-  public <T extends TType> SquaredDifference<T> squaredDifference(Operand<T> x, Operand<T> y) {
-    return SquaredDifference.create(scope, x, y);
-  }
-
-  /**
-   * Returns x - y element-wise.
-   *  <p>
-   *  <i>NOTE</i>: `math.Sub` supports broadcasting. More about broadcasting
-   *  [here](http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html)
-   *
-   * @param <T> data type for {@code z()} output
-   * @param x
-   * @param y
-   * @return a new instance of Sub
-   */
-  public <T extends TType> Sub<T> sub(Operand<T> x, Operand<T> y) {
-    return Sub.create(scope, x, y);
-  }
-
-  /**
-   * Computes tan of x element-wise.
-   *  <p>
-   *    Given an input tensor, this function computes tangent of every
-   *    element in the tensor. Input range is `(-inf, inf)` and
-   *    output range is `(-inf, inf)`. If input lies outside the boundary, `nan`
-   *    is returned.
-   *  <p>
-   *    <pre>{@code
-   *    x = tf.constant([-float("inf"), -9, -0.5, 1, 1.2, 200, 10000, float("inf")])
-   *    tf.math.tan(x) ==> [nan 0.45231566 -0.5463025 1.5574077 2.572152 -1.7925274 0.32097113 nan]
-   *    }</pre>
-   *
-   * @param <T> data type for {@code y()} output
-   * @param x
-   * @return a new instance of Tan
-   */
-  public <T extends TType> Tan<T> tan(Operand<T> x) {
-    return Tan.create(scope, x);
-  }
-
-  /**
-   * Computes hyperbolic tangent of `x` element-wise.
-   *  <p>
-   *    Given an input tensor, this function computes hyperbolic tangent of every
-   *    element in the tensor. Input range is `[-inf, inf]` and
-   *    output range is `[-1,1]`.
-   *  <p>
-   *    <pre>{@code
-   *    x = tf.constant([-float("inf"), -5, -0.5, 1, 1.2, 2, 3, float("inf")])
-   *    tf.math.tanh(x) ==> [-1. -0.99990916 -0.46211717 0.7615942 0.8336547 0.9640276 0.9950547 1.]
-   *    }</pre>
-   *
-   * @param <T> data type for {@code y()} output
-   * @param x
-   * @return a new instance of Tanh
-   */
-  public <T extends TType> Tanh<T> tanh(Operand<T> x) {
-    return Tanh.create(scope, x);
-  }
-
-  /**
-   * Returns x / y element-wise for integer types.
-   *  <p>
-   *  Truncation designates that negative numbers will round fractional quantities
-   *  toward zero. I.e. -7 / 5 = -1. This matches C semantics but it is different
-   *  than Python semantics. See `FloorDiv` for a division function that matches
-   *  Python Semantics.
-   *  <p>
-   *  <i>NOTE</i>: `math.TruncateDiv` supports broadcasting. More about broadcasting
-   *  [here](http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html)
-   *
-   * @param <T> data type for {@code z()} output
-   * @param x
-   * @param y
-   * @return a new instance of TruncateDiv
-   */
-  public <T extends TType> TruncateDiv<T> truncateDiv(Operand<T> x, Operand<T> y) {
-    return TruncateDiv.create(scope, x, y);
-  }
-
-  /**
-   * Returns element-wise remainder of division. This emulates C semantics in that
-   *  <p>
-   *  the result here is consistent with a truncating divide. E.g. `truncate(x / y) *
-   *  y + truncate_mod(x, y) = x`.
-   *  <p>
-   *  <i>NOTE</i>: `math.TruncateMod` supports broadcasting. More about broadcasting
-   *  [here](http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html)
-   *
-   * @param <T> data type for {@code z()} output
-   * @param x
-   * @param y
-   * @return a new instance of TruncateMod
-   */
-  public <T extends TNumber> TruncateMod<T> truncateMod(Operand<T> x, Operand<T> y) {
-    return TruncateMod.create(scope, x, y);
-  }
-
-  /**
-   * Computes the maximum along segments of a tensor.
-   *  <p>
-   *  Read
-   *  [the section on segmentation](https://tensorflow.org/api_docs/python/tf/math#Segmentation)
-   *  for an explanation of segments.
-   *  <p>
-   *  This operator is similar to the unsorted segment sum operator found
-   *  [(here)](../../../api_docs/python/math_ops.md#UnsortedSegmentSum).
-   *  Instead of computing the sum over segments, it computes the maximum such that:
-   *  <p>
-   *  \\(output_i = \max_{j...} data[j...]\\) where max is over tuples `j...` such
-   *  that `segment_ids[j...] == i`.
-   *  <p>
-   *  If the maximum is empty for a given segment ID `i`, it outputs the smallest
-   *  possible value for the specific numeric type,
-   *  `output[i] = numeric_limits<T>::lowest()`.
-   *  <p>
-   *  If the given segment ID `i` is negative, then the corresponding value is
-   *  dropped, and will not be included in the result.
-   *  <p>
-   *  <div style="width:70%; margin:auto; margin-bottom:10px; margin-top:20px;">
-   *  <img style="width:100%" src="https://www.tensorflow.org/images/UnsortedSegmentMax.png" alt>
-   *  </div>
-   *  <p>
-   *  For example:
-   *  <pre>{@code
-   *  c = tf.constant([[1,2,3,4], [5,6,7,8], [4,3,2,1]])
-   *  tf.unsorted_segment_max(c, tf.constant([0, 1, 0]), num_segments=2)
-   *  # ==> [[ 4,  3, 3, 4],
-   *  #       [5,  6, 7, 8]]
-   *  }</pre>
-   *
-   * @param <T> data type for {@code output()} output
-   * @param data
-   * @param segmentIds A tensor whose shape is a prefix of `data.shape`.
-   * @param numSegments
-   * @return a new instance of UnsortedSegmentMax
-   */
-  public <T extends TNumber, U extends TNumber, V extends TNumber> UnsortedSegmentMax<T> unsortedSegmentMax(
-      Operand<T> data, Operand<U> segmentIds, Operand<V> numSegments) {
-    return UnsortedSegmentMax.create(scope, data, segmentIds, numSegments);
-  }
-
-  /**
-   * Computes the minimum along segments of a tensor.
-   *  <p>
-   *  Read
-   *  [the section on segmentation](https://tensorflow.org/api_docs/python/tf/math#Segmentation)
-   *  for an explanation of segments.
-   *  <p>
-   *  This operator is similar to the unsorted segment sum operator found
-   *  [(here)](../../../api_docs/python/math_ops.md#UnsortedSegmentSum).
-   *  Instead of computing the sum over segments, it computes the minimum such that:
-   *  <p>
-   *  \\(output_i = \min_{j...} data_[j...]\\) where min is over tuples `j...` such
-   *  that `segment_ids[j...] == i`.
-   *  <p>
-   *  If the minimum is empty for a given segment ID `i`, it outputs the largest
-   *  possible value for the specific numeric type,
-   *  `output[i] = numeric_limits<T>::max()`.
-   *  <p>
-   *  For example:
-   *  <pre>{@code
-   *  c = tf.constant([[1,2,3,4], [5,6,7,8], [4,3,2,1]])
-   *  tf.unsorted_segment_min(c, tf.constant([0, 1, 0]), num_segments=2)
-   *  # ==> [[ 1,  2, 2, 1],
-   *  #       [5,  6, 7, 8]]
-   *  }</pre>
-   *  If the given segment ID `i` is negative, then the corresponding value is
-   *  dropped, and will not be included in the result.
-   *
-   * @param <T> data type for {@code output()} output
-   * @param data
-   * @param segmentIds A tensor whose shape is a prefix of `data.shape`.
-   * @param numSegments
-   * @return a new instance of UnsortedSegmentMin
-   */
-  public <T extends TNumber, U extends TNumber, V extends TNumber> UnsortedSegmentMin<T> unsortedSegmentMin(
-      Operand<T> data, Operand<U> segmentIds, Operand<V> numSegments) {
-    return UnsortedSegmentMin.create(scope, data, segmentIds, numSegments);
-  }
-
-  /**
-   * Computes the product along segments of a tensor.
-   *  <p>
-   *  Read
-   *  [the section on segmentation](https://tensorflow.org/api_docs/python/tf/math#Segmentation)
-   *  for an explanation of segments.
-   *  <p>
-   *  This operator is similar to the unsorted segment sum operator found
-   *  [(here)](../../../api_docs/python/math_ops.md#UnsortedSegmentSum).
-   *  Instead of computing the sum over segments, it computes the product of all
-   *  entries belonging to a segment such that:
-   *  <p>
-   *  \\(output_i = \prod_{j...} data[j...]\\) where the product is over tuples
-   *  `j...` such that `segment_ids[j...] == i`.
-   *  <p>
-   *  For example:
-   *  <pre>{@code
-   *  c = tf.constant([[1,2,3,4], [5,6,7,8], [4,3,2,1]])
-   *  tf.unsorted_segment_prod(c, tf.constant([0, 1, 0]), num_segments=2)
-   *  # ==> [[ 4,  6, 6, 4],
-   *  #       [5,  6, 7, 8]]
-   *  }</pre>
-   *  If there is no entry for a given segment ID `i`, it outputs 1.
-   *  <p>
-   *  If the given segment ID `i` is negative, then the corresponding value is
-   *  dropped, and will not be included in the result.
-   *
-   * @param <T> data type for {@code output()} output
-   * @param data
-   * @param segmentIds A tensor whose shape is a prefix of `data.shape`.
-   * @param numSegments
-   * @return a new instance of UnsortedSegmentProd
-   */
-  public <T extends TType, U extends TNumber, V extends TNumber> UnsortedSegmentProd<T> unsortedSegmentProd(
-      Operand<T> data, Operand<U> segmentIds, Operand<V> numSegments) {
-    return UnsortedSegmentProd.create(scope, data, segmentIds, numSegments);
-  }
-
-  /**
-   * Computes the sum along segments of a tensor.
-   *  <p>
-   *  Read
-   *  [the section on segmentation](https://tensorflow.org/api_docs/python/tf/math#Segmentation)
-   *  for an explanation of segments.
-   *  <p>
-   *  Computes a tensor such that
-   *  \\(output[i] = \sum_{j...} data[j...]\\) where the sum is over tuples `j...` such
-   *  that `segment_ids[j...] == i`.  Unlike `SegmentSum`, `segment_ids`
-   *  need not be sorted and need not cover all values in the full
-   *  range of valid values.
-   *  <p>
-   *  If the sum is empty for a given segment ID `i`, `output[i] = 0`.
-   *  If the given segment ID `i` is negative, the value is dropped and will not be
-   *  added to the sum of the segment.
-   *  <p>
-   *  `num_segments` should equal the number of distinct segment IDs.
-   *  <p>
-   *  <div style="width:70%; margin:auto; margin-bottom:10px; margin-top:20px;">
-   *  <img style="width:100%" src="https://www.tensorflow.org/images/UnsortedSegmentSum.png" alt>
-   *  </div>
-   *  <pre>{@code
-   *  c = tf.constant([[1,2,3,4], [5,6,7,8], [4,3,2,1]])
-   *  tf.unsorted_segment_sum(c, tf.constant([0, 1, 0]), num_segments=2)
-   *  # ==> [[ 5,  5, 5, 5],
-   *  #       [5,  6, 7, 8]]
-   *  }</pre>
-   *
-   * @param <T> data type for {@code output()} output
-   * @param data
-   * @param segmentIds A tensor whose shape is a prefix of `data.shape`.
-   * @param numSegments
-   * @return a new instance of UnsortedSegmentSum
-   */
-  public <T extends TType, U extends TNumber, V extends TNumber> UnsortedSegmentSum<T> unsortedSegmentSum(
-      Operand<T> data, Operand<U> segmentIds, Operand<V> numSegments) {
-    return UnsortedSegmentSum.create(scope, data, segmentIds, numSegments);
-  }
-
-  /**
-   * Returns 0 if x == 0, and x / y otherwise, elementwise.
-   *
-   * @param <T> data type for {@code z()} output
-   * @param x
-   * @param y
-   * @return a new instance of Xdivy
-   */
-  public <T extends TType> Xdivy<T> xdivy(Operand<T> x, Operand<T> y) {
-    return Xdivy.create(scope, x, y);
-  }
-
-  /**
-   * Returns 0 if x == 0, and x * log1p(y) otherwise, elementwise.
-   *
-   * @param <T> data type for {@code z()} output
-   * @param x
-   * @param y
-   * @return a new instance of Xlog1py
-   */
-  public <T extends TType> Xlog1py<T> xlog1py(Operand<T> x, Operand<T> y) {
-    return Xlog1py.create(scope, x, y);
-  }
-
-  /**
-   * Returns 0 if x == 0, and x * log(y) otherwise, elementwise.
-   *
-   * @param <T> data type for {@code z()} output
-   * @param x
-   * @param y
-   * @return a new instance of Xlogy
-   */
-  public <T extends TType> Xlogy<T> xlogy(Operand<T> x, Operand<T> y) {
-    return Xlogy.create(scope, x, y);
-  }
-
-  /**
-   * Compute the Hurwitz zeta function \\(\zeta(x, q)\\).
-   *  <p>
-   *  The Hurwitz zeta function is defined as:
-   *  <p>
-   *  \\(\zeta(x, q) = \sum_{n=0}^{\infty} (q + n)^{-x}\\)
-   *
-   * @param <T> data type for {@code z()} output
-   * @param x
-   * @param q
-   * @return a new instance of Zeta
-   */
-  public <T extends TNumber> Zeta<T> zeta(Operand<T> x, Operand<T> q) {
-    return Zeta.create(scope, x, q);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/NnRawOps.java b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/NnRawOps.java
deleted file mode 100644
index d9147af3934..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/NnRawOps.java
+++ /dev/null
@@ -1,74 +0,0 @@
-// Copyright 2020 The TensorFlow Authors. All Rights Reserved.
-//
-// Licensed under the Apache License, Version 2.0 (the "License");
-// you may not use this file except in compliance with the License.
-// You may obtain a copy of the License at
-//
-//     http://www.apache.org/licenses/LICENSE-2.0
-//
-// Unless required by applicable law or agreed to in writing, software
-// distributed under the License is distributed on an "AS IS" BASIS,
-// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-// See the License for the specific language governing permissions and
-// limitations under the License.
-// ==============================================================================
-//
-// This class has been generated, DO NOT EDIT!
-//
-package org.tensorflow.op;
-
-import org.tensorflow.Operand;
-import org.tensorflow.op.nn.raw.SoftmaxCrossEntropyWithLogits;
-import org.tensorflow.op.nn.raw.SparseSoftmaxCrossEntropyWithLogits;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * An API for building {@code nn.raw} operations as {@link Op Op}s
- *
- * @see {@link Ops}
- */
-public final class NnRawOps {
-  private final Scope scope;
-
-  NnRawOps(Scope scope) {
-    this.scope = scope;
-  }
-
-  /**
-   * Computes softmax cross entropy cost and gradients to backpropagate.
-   *  <p>
-   *  Inputs are the logits, not probabilities.
-   *
-   * @param <T> data type for {@code loss()} output
-   * @param features batch_size x num_classes matrix
-   * @param labels batch_size x num_classes matrix
-   *  The caller must ensure that each batch of labels represents a valid
-   *  probability distribution.
-   * @return a new instance of SoftmaxCrossEntropyWithLogits
-   */
-  public <T extends TNumber> SoftmaxCrossEntropyWithLogits<T> softmaxCrossEntropyWithLogits(
-      Operand<T> features, Operand<T> labels) {
-    return SoftmaxCrossEntropyWithLogits.create(scope, features, labels);
-  }
-
-  /**
-   * Computes softmax cross entropy cost and gradients to backpropagate.
-   *  <p>
-   *  Unlike `SoftmaxCrossEntropyWithLogits`, this operation does not accept
-   *  a matrix of label probabilities, but rather a single label per row
-   *  of features.  This label is considered to have probability 1.0 for the
-   *  given row.
-   *  <p>
-   *  Inputs are the logits, not probabilities.
-   *
-   * @param <T> data type for {@code loss()} output
-   * @param features batch_size x num_classes matrix
-   * @param labels batch_size vector with values in [0, num_classes).
-   *  This is the label for the given minibatch entry.
-   * @return a new instance of SparseSoftmaxCrossEntropyWithLogits
-   */
-  public <T extends TNumber, U extends TNumber> SparseSoftmaxCrossEntropyWithLogits<T> sparseSoftmaxCrossEntropyWithLogits(
-      Operand<T> features, Operand<U> labels) {
-    return SparseSoftmaxCrossEntropyWithLogits.create(scope, features, labels);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/Ops.java b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/Ops.java
deleted file mode 100644
index b17d3c81d31..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/Ops.java
+++ /dev/null
@@ -1,7637 +0,0 @@
-// Copyright 2020 The TensorFlow Authors. All Rights Reserved.
-//
-// Licensed under the Apache License, Version 2.0 (the "License");
-// you may not use this file except in compliance with the License.
-// You may obtain a copy of the License at
-//
-//     http://www.apache.org/licenses/LICENSE-2.0
-//
-// Unless required by applicable law or agreed to in writing, software
-// distributed under the License is distributed on an "AS IS" BASIS,
-// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-// See the License for the specific language governing permissions and
-// limitations under the License.
-// ==============================================================================
-//
-// This class has been generated, DO NOT EDIT!
-//
-package org.tensorflow.op;
-
-import java.nio.charset.Charset;
-import java.util.List;
-import org.tensorflow.DataType;
-import org.tensorflow.EagerSession;
-import org.tensorflow.ExecutionEnvironment;
-import org.tensorflow.Operand;
-import org.tensorflow.ndarray.BooleanNdArray;
-import org.tensorflow.ndarray.ByteNdArray;
-import org.tensorflow.ndarray.DoubleNdArray;
-import org.tensorflow.ndarray.FloatNdArray;
-import org.tensorflow.ndarray.IntNdArray;
-import org.tensorflow.ndarray.LongNdArray;
-import org.tensorflow.ndarray.NdArray;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.ndarray.buffer.BooleanDataBuffer;
-import org.tensorflow.ndarray.buffer.ByteDataBuffer;
-import org.tensorflow.ndarray.buffer.DataBuffer;
-import org.tensorflow.ndarray.buffer.DoubleDataBuffer;
-import org.tensorflow.ndarray.buffer.FloatDataBuffer;
-import org.tensorflow.ndarray.buffer.IntDataBuffer;
-import org.tensorflow.ndarray.buffer.LongDataBuffer;
-import org.tensorflow.op.core.Abort;
-import org.tensorflow.op.core.All;
-import org.tensorflow.op.core.Any;
-import org.tensorflow.op.core.AssertThat;
-import org.tensorflow.op.core.Assign;
-import org.tensorflow.op.core.AssignAdd;
-import org.tensorflow.op.core.AssignAddVariableOp;
-import org.tensorflow.op.core.AssignSub;
-import org.tensorflow.op.core.AssignSubVariableOp;
-import org.tensorflow.op.core.AssignVariableOp;
-import org.tensorflow.op.core.Barrier;
-import org.tensorflow.op.core.BarrierClose;
-import org.tensorflow.op.core.BarrierIncompleteSize;
-import org.tensorflow.op.core.BarrierInsertMany;
-import org.tensorflow.op.core.BarrierReadySize;
-import org.tensorflow.op.core.BarrierTakeMany;
-import org.tensorflow.op.core.Batch;
-import org.tensorflow.op.core.BatchToSpace;
-import org.tensorflow.op.core.BatchToSpaceNd;
-import org.tensorflow.op.core.Bitcast;
-import org.tensorflow.op.core.BroadcastDynamicShape;
-import org.tensorflow.op.core.BroadcastTo;
-import org.tensorflow.op.core.Bucketize;
-import org.tensorflow.op.core.ClipByValue;
-import org.tensorflow.op.core.Concat;
-import org.tensorflow.op.core.Constant;
-import org.tensorflow.op.core.ConsumeMutexLock;
-import org.tensorflow.op.core.ControlTrigger;
-import org.tensorflow.op.core.CountUpTo;
-import org.tensorflow.op.core.DeepCopy;
-import org.tensorflow.op.core.DeleteSessionTensor;
-import org.tensorflow.op.core.DestroyResourceOp;
-import org.tensorflow.op.core.DestroyTemporaryVariable;
-import org.tensorflow.op.core.DynamicPartition;
-import org.tensorflow.op.core.DynamicStitch;
-import org.tensorflow.op.core.EditDistance;
-import org.tensorflow.op.core.Empty;
-import org.tensorflow.op.core.EmptyTensorList;
-import org.tensorflow.op.core.EnsureShape;
-import org.tensorflow.op.core.ExpandDims;
-import org.tensorflow.op.core.ExtractVolumePatches;
-import org.tensorflow.op.core.Fill;
-import org.tensorflow.op.core.Fingerprint;
-import org.tensorflow.op.core.Gather;
-import org.tensorflow.op.core.GatherNd;
-import org.tensorflow.op.core.GetSessionHandle;
-import org.tensorflow.op.core.GetSessionTensor;
-import org.tensorflow.op.core.Gradients;
-import org.tensorflow.op.core.GuaranteeConst;
-import org.tensorflow.op.core.HashTable;
-import org.tensorflow.op.core.Helpers;
-import org.tensorflow.op.core.HistogramFixedWidth;
-import org.tensorflow.op.core.Identity;
-import org.tensorflow.op.core.IdentityN;
-import org.tensorflow.op.core.ImmutableConst;
-import org.tensorflow.op.core.Init;
-import org.tensorflow.op.core.InitializeTable;
-import org.tensorflow.op.core.InitializeTableFromTextFile;
-import org.tensorflow.op.core.InplaceAdd;
-import org.tensorflow.op.core.InplaceSub;
-import org.tensorflow.op.core.InplaceUpdate;
-import org.tensorflow.op.core.IsVariableInitialized;
-import org.tensorflow.op.core.LinSpace;
-import org.tensorflow.op.core.LookupTableExport;
-import org.tensorflow.op.core.LookupTableFind;
-import org.tensorflow.op.core.LookupTableImport;
-import org.tensorflow.op.core.LookupTableInsert;
-import org.tensorflow.op.core.LookupTableSize;
-import org.tensorflow.op.core.LoopCond;
-import org.tensorflow.op.core.MapClear;
-import org.tensorflow.op.core.MapIncompleteSize;
-import org.tensorflow.op.core.MapPeek;
-import org.tensorflow.op.core.MapSize;
-import org.tensorflow.op.core.MapStage;
-import org.tensorflow.op.core.MapUnstage;
-import org.tensorflow.op.core.MapUnstageNoKey;
-import org.tensorflow.op.core.Max;
-import org.tensorflow.op.core.Merge;
-import org.tensorflow.op.core.Min;
-import org.tensorflow.op.core.MirrorPad;
-import org.tensorflow.op.core.MlirPassthroughOp;
-import org.tensorflow.op.core.MutableDenseHashTable;
-import org.tensorflow.op.core.MutableHashTable;
-import org.tensorflow.op.core.MutableHashTableOfTensors;
-import org.tensorflow.op.core.Mutex;
-import org.tensorflow.op.core.MutexLock;
-import org.tensorflow.op.core.NextIteration;
-import org.tensorflow.op.core.NoOp;
-import org.tensorflow.op.core.OneHot;
-import org.tensorflow.op.core.OnesLike;
-import org.tensorflow.op.core.OrderedMapClear;
-import org.tensorflow.op.core.OrderedMapIncompleteSize;
-import org.tensorflow.op.core.OrderedMapPeek;
-import org.tensorflow.op.core.OrderedMapSize;
-import org.tensorflow.op.core.OrderedMapStage;
-import org.tensorflow.op.core.OrderedMapUnstage;
-import org.tensorflow.op.core.OrderedMapUnstageNoKey;
-import org.tensorflow.op.core.Pad;
-import org.tensorflow.op.core.ParallelConcat;
-import org.tensorflow.op.core.ParallelDynamicStitch;
-import org.tensorflow.op.core.Placeholder;
-import org.tensorflow.op.core.PlaceholderWithDefault;
-import org.tensorflow.op.core.Print;
-import org.tensorflow.op.core.Prod;
-import org.tensorflow.op.core.QuantizedReshape;
-import org.tensorflow.op.core.Range;
-import org.tensorflow.op.core.Rank;
-import org.tensorflow.op.core.ReadVariableOp;
-import org.tensorflow.op.core.ReduceAll;
-import org.tensorflow.op.core.ReduceAny;
-import org.tensorflow.op.core.ReduceMax;
-import org.tensorflow.op.core.ReduceMin;
-import org.tensorflow.op.core.ReduceProd;
-import org.tensorflow.op.core.ReduceSum;
-import org.tensorflow.op.core.RefNextIteration;
-import org.tensorflow.op.core.RefSelect;
-import org.tensorflow.op.core.RefSwitch;
-import org.tensorflow.op.core.RemoteFusedGraphExecute;
-import org.tensorflow.op.core.Reshape;
-import org.tensorflow.op.core.ResourceCountUpTo;
-import org.tensorflow.op.core.ResourceGather;
-import org.tensorflow.op.core.ResourceGatherNd;
-import org.tensorflow.op.core.ResourceScatterAdd;
-import org.tensorflow.op.core.ResourceScatterDiv;
-import org.tensorflow.op.core.ResourceScatterMax;
-import org.tensorflow.op.core.ResourceScatterMin;
-import org.tensorflow.op.core.ResourceScatterMul;
-import org.tensorflow.op.core.ResourceScatterNdAdd;
-import org.tensorflow.op.core.ResourceScatterNdSub;
-import org.tensorflow.op.core.ResourceScatterNdUpdate;
-import org.tensorflow.op.core.ResourceScatterSub;
-import org.tensorflow.op.core.ResourceScatterUpdate;
-import org.tensorflow.op.core.ResourceStridedSliceAssign;
-import org.tensorflow.op.core.Reverse;
-import org.tensorflow.op.core.ReverseSequence;
-import org.tensorflow.op.core.Roll;
-import org.tensorflow.op.core.Rpc;
-import org.tensorflow.op.core.ScatterAdd;
-import org.tensorflow.op.core.ScatterDiv;
-import org.tensorflow.op.core.ScatterMax;
-import org.tensorflow.op.core.ScatterMin;
-import org.tensorflow.op.core.ScatterMul;
-import org.tensorflow.op.core.ScatterNd;
-import org.tensorflow.op.core.ScatterNdAdd;
-import org.tensorflow.op.core.ScatterNdNonAliasingAdd;
-import org.tensorflow.op.core.ScatterNdSub;
-import org.tensorflow.op.core.ScatterNdUpdate;
-import org.tensorflow.op.core.ScatterSub;
-import org.tensorflow.op.core.ScatterUpdate;
-import org.tensorflow.op.core.Select;
-import org.tensorflow.op.core.SetDiff1d;
-import org.tensorflow.op.core.SetSize;
-import org.tensorflow.op.core.ShapeN;
-import org.tensorflow.op.core.Size;
-import org.tensorflow.op.core.Skipgram;
-import org.tensorflow.op.core.Slice;
-import org.tensorflow.op.core.Snapshot;
-import org.tensorflow.op.core.SpaceToBatchNd;
-import org.tensorflow.op.core.Split;
-import org.tensorflow.op.core.SplitV;
-import org.tensorflow.op.core.Squeeze;
-import org.tensorflow.op.core.Stack;
-import org.tensorflow.op.core.Stage;
-import org.tensorflow.op.core.StageClear;
-import org.tensorflow.op.core.StagePeek;
-import org.tensorflow.op.core.StageSize;
-import org.tensorflow.op.core.StopGradient;
-import org.tensorflow.op.core.StridedSlice;
-import org.tensorflow.op.core.StridedSliceAssign;
-import org.tensorflow.op.core.StridedSliceGrad;
-import org.tensorflow.op.core.Sum;
-import org.tensorflow.op.core.SwitchCond;
-import org.tensorflow.op.core.TemporaryVariable;
-import org.tensorflow.op.core.TensorArray;
-import org.tensorflow.op.core.TensorArrayClose;
-import org.tensorflow.op.core.TensorArrayConcat;
-import org.tensorflow.op.core.TensorArrayGather;
-import org.tensorflow.op.core.TensorArrayGrad;
-import org.tensorflow.op.core.TensorArrayGradWithShape;
-import org.tensorflow.op.core.TensorArrayPack;
-import org.tensorflow.op.core.TensorArrayRead;
-import org.tensorflow.op.core.TensorArrayScatter;
-import org.tensorflow.op.core.TensorArraySize;
-import org.tensorflow.op.core.TensorArraySplit;
-import org.tensorflow.op.core.TensorArrayUnpack;
-import org.tensorflow.op.core.TensorArrayWrite;
-import org.tensorflow.op.core.TensorListConcat;
-import org.tensorflow.op.core.TensorListConcatLists;
-import org.tensorflow.op.core.TensorListElementShape;
-import org.tensorflow.op.core.TensorListFromTensor;
-import org.tensorflow.op.core.TensorListGather;
-import org.tensorflow.op.core.TensorListGetItem;
-import org.tensorflow.op.core.TensorListLength;
-import org.tensorflow.op.core.TensorListPopBack;
-import org.tensorflow.op.core.TensorListPushBack;
-import org.tensorflow.op.core.TensorListPushBackBatch;
-import org.tensorflow.op.core.TensorListReserve;
-import org.tensorflow.op.core.TensorListResize;
-import org.tensorflow.op.core.TensorListScatter;
-import org.tensorflow.op.core.TensorListScatterIntoExistingList;
-import org.tensorflow.op.core.TensorListSetItem;
-import org.tensorflow.op.core.TensorListSplit;
-import org.tensorflow.op.core.TensorListStack;
-import org.tensorflow.op.core.TensorScatterNdAdd;
-import org.tensorflow.op.core.TensorScatterNdSub;
-import org.tensorflow.op.core.TensorScatterNdUpdate;
-import org.tensorflow.op.core.TensorStridedSliceUpdate;
-import org.tensorflow.op.core.Tile;
-import org.tensorflow.op.core.Timestamp;
-import org.tensorflow.op.core.TryRpc;
-import org.tensorflow.op.core.Unbatch;
-import org.tensorflow.op.core.UnbatchGrad;
-import org.tensorflow.op.core.Unique;
-import org.tensorflow.op.core.UniqueWithCounts;
-import org.tensorflow.op.core.UnravelIndex;
-import org.tensorflow.op.core.Unstack;
-import org.tensorflow.op.core.Unstage;
-import org.tensorflow.op.core.VarHandleOp;
-import org.tensorflow.op.core.VarIsInitializedOp;
-import org.tensorflow.op.core.Variable;
-import org.tensorflow.op.core.VariableShape;
-import org.tensorflow.op.core.Where;
-import org.tensorflow.op.core.Zeros;
-import org.tensorflow.op.core.ZerosLike;
-import org.tensorflow.types.TBool;
-import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.TFloat64;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.TString;
-import org.tensorflow.types.TUint8;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * An API for building operations as {@link Op Op}s
- * <p>
- * Any operation wrapper found in the classpath properly annotated as an{@link org.tensorflow.op.annotation.Operator @Operator} is exposed
- * by this API or one of its subgroup.
- * <p>Example usage:
- * <pre>{@code
- * try (Graph g = new Graph()) {
- *   Ops tf = Ops.create(g);
- *   // Operations are typed classes with convenience
- *   // builders in Ops.
- *   Constant<TInt32> three = tf.constant(3);
- *   // Single-result operations implement the Operand
- *   // interface, so this works too.
- *   Operand<TInt32> four = tf.constant(4);
- *   // Most builders are found within a group, and accept
- *   // Operand types as operands
- *   Operand<TInt32> nine = tf.math.add(four, tf.constant(5));
- *   // Multi-result operations however offer methods to
- *   // select a particular result for use.
- *   Operand<TInt32> result = 
- *       tf.math.add(tf.unique(s, a).y(), b);
- *   // Optional attributes
- *   tf.linalg.matMul(a, b, MatMul.transposeA(true));
- *   // Naming operators
- *   tf.withName("foo").constant(5); // name "foo"
- *   // Names can exist in a hierarchy
- *   Ops sub = tf.withSubScope("sub");
- *   sub.withName("bar").constant(4); // "sub/bar"
- * }
- * }</pre>
- */
-public final class Ops {
-  public final NnOps nn;
-
-  public final SummaryOps summary;
-
-  public final ImageOps image;
-
-  public final DataOps data;
-
-  public final ShapeOps shape;
-
-  public final IoOps io;
-
-  public final DtypesOps dtypes;
-
-  public final XlaOps xla;
-
-  public final LinalgOps linalg;
-
-  public final RandomOps random;
-
-  public final StringsOps strings;
-
-  public final SparseOps sparse;
-
-  public final BitwiseOps bitwise;
-
-  public final MathOps math;
-
-  public final AudioOps audio;
-
-  public final SignalOps signal;
-
-  public final QuantizationOps quantization;
-
-  public final TrainOps train;
-
-  private final Scope scope;
-
-  private Ops(Scope scope) {
-    this.scope = scope;
-    nn = new NnOps(scope);
-    summary = new SummaryOps(scope);
-    image = new ImageOps(scope);
-    data = new DataOps(scope);
-    shape = new ShapeOps(scope);
-    io = new IoOps(scope);
-    dtypes = new DtypesOps(scope);
-    xla = new XlaOps(scope);
-    linalg = new LinalgOps(scope);
-    random = new RandomOps(scope);
-    strings = new StringsOps(scope);
-    sparse = new SparseOps(scope);
-    bitwise = new BitwiseOps(scope);
-    math = new MathOps(scope);
-    audio = new AudioOps(scope);
-    signal = new SignalOps(scope);
-    quantization = new QuantizationOps(scope);
-    train = new TrainOps(scope);
-  }
-
-  /**
-   * Raise a exception to abort the process when called.
-   *  <p>
-   *  If exit_without_error is true, the process will exit normally,
-   *  otherwise it will exit with a SIGABORT signal.
-   *  <p>
-   *  Returns nothing but an exception.
-   *
-   * @param options carries optional attributes values
-   * @return a new instance of Abort
-   */
-  public Abort abort(Abort.Options... options) {
-    return Abort.create(scope, options);
-  }
-
-  /**
-   * Computes the "logical and" of elements across dimensions of a tensor.
-   *  <p>
-   *  Reduces `input` along the dimensions given in `axis`. Unless
-   *  `keep_dims` is true, the rank of the tensor is reduced by 1 for each entry in
-   *  `axis`. If `keep_dims` is true, the reduced dimensions are
-   *  retained with length 1.
-   *
-   * @param input The tensor to reduce.
-   * @param axis The dimensions to reduce. Must be in the range
-   *  `[-rank(input), rank(input))`.
-   * @param options carries optional attributes values
-   * @return a new instance of All
-   */
-  public <T extends TNumber> All all(Operand<TBool> input, Operand<T> axis,
-      All.Options... options) {
-    return All.create(scope, input, axis, options);
-  }
-
-  /**
-   * Computes the "logical or" of elements across dimensions of a tensor.
-   *  <p>
-   *  Reduces `input` along the dimensions given in `axis`. Unless
-   *  `keep_dims` is true, the rank of the tensor is reduced by 1 for each entry in
-   *  `axis`. If `keep_dims` is true, the reduced dimensions are
-   *  retained with length 1.
-   *
-   * @param input The tensor to reduce.
-   * @param axis The dimensions to reduce. Must be in the range
-   *  `[-rank(input), rank(input))`.
-   * @param options carries optional attributes values
-   * @return a new instance of Any
-   */
-  public <T extends TNumber> Any any(Operand<TBool> input, Operand<T> axis,
-      Any.Options... options) {
-    return Any.create(scope, input, axis, options);
-  }
-
-  /**
-   * Creates a constant of {@code int} elements.
-   *
-   * @param scope is a scope used to add the underlying operation.
-   * @param data An array containing the values to put into the new constant.
-   * @return a float constant
-   */
-  public Constant<TInt32> array(int... data) {
-    return Constant.arrayOf(scope, data);
-  }
-
-  /**
-   * Creates a constant of {@code String} elements, using the default UTF-8 charset.
-   *
-   * @param scope is a scope used to add the underlying operation.
-   * @param data An array containing the values to put into the new constant.
-   * @return the {@code String} constant
-   */
-  public Constant<TString> array(String... data) {
-    return Constant.arrayOf(scope, data);
-  }
-
-  /**
-   * Creates a constant of {@code boolean} elements.
-   *
-   * @param scope is a scope used to add the underlying operation.
-   * @param data An array containing the values to put into the new constant.
-   * @return a boolean constant
-   */
-  public Constant<TBool> array(boolean... data) {
-    return Constant.arrayOf(scope, data);
-  }
-
-  /**
-   * Creates a constant of {@code long} elements.
-   *
-   * @param scope is a scope used to add the underlying operation.
-   * @param data An array containing the values to put into the new constant.
-   * @return a long constant
-   */
-  public Constant<TInt64> array(long... data) {
-    return Constant.arrayOf(scope, data);
-  }
-
-  /**
-   * Creates a constant of {@code float} elements.
-   *
-   * @param scope is a scope used to add the underlying operation.
-   * @param data An array containing the values to put into the new constant.
-   * @return a float constant
-   */
-  public Constant<TFloat32> array(float... data) {
-    return Constant.arrayOf(scope, data);
-  }
-
-  /**
-   * Creates a constant of {@code double} elements.
-   *
-   * @param scope is a scope used to add the underlying operation.
-   * @param data An array containing the values to put into the new constant.
-   * @return a double constant
-   */
-  public Constant<TFloat64> array(double... data) {
-    return Constant.arrayOf(scope, data);
-  }
-
-  /**
-   * Creates a constant of {@code byte} elements.
-   *
-   * @param scope is a scope used to add the underlying operation.
-   * @param data An array containing the values to put into the new constant.
-   * @return a byte constant
-   */
-  public Constant<TUint8> array(byte... data) {
-    return Constant.arrayOf(scope, data);
-  }
-
-  /**
-   * Creates a constant of {@code String} elements, using the given charset.
-   *
-   * @param scope is a scope used to add the underlying operation.
-   * @param charset charset for encoding/decoding strings bytes.
-   * @param data An array containing the values to put into the new constant. String elements are
-   *      sequences of bytes from the last array dimension.
-   * @return the {@code String} constant
-   */
-  public Constant<TString> array(Charset charset, String... data) {
-    return Constant.arrayOf(scope, charset, data);
-  }
-
-  /**
-   * Asserts that the given condition is true.
-   *  <p>
-   *  If `condition` evaluates to false, print the list of tensors in `data`.
-   *  `summarize` determines how many entries of the tensors to print.
-   *
-   * @param condition The condition to evaluate.
-   * @param data The tensors to print out when condition is false.
-   * @param options carries optional attributes values
-   * @return a new instance of AssertThat
-   */
-  public AssertThat assertThat(Operand<TBool> condition, Iterable<Operand<?>> data,
-      AssertThat.Options... options) {
-    return AssertThat.create(scope, condition, data, options);
-  }
-
-  /**
-   * Update 'ref' by assigning 'value' to it.
-   *  <p>
-   *  This operation outputs "ref" after the assignment is done.
-   *  This makes it easier to chain operations that need to use the reset value.
-   *
-   * @param <T> data type for {@code outputRef()} output
-   * @param ref Should be from a `Variable` node. May be uninitialized.
-   * @param value The value to be assigned to the variable.
-   * @param options carries optional attributes values
-   * @return a new instance of Assign
-   */
-  public <T extends TType> Assign<T> assign(Operand<T> ref, Operand<T> value,
-      Assign.Options... options) {
-    return Assign.create(scope, ref, value, options);
-  }
-
-  /**
-   * Update 'ref' by adding 'value' to it.
-   *  <p>
-   *  This operation outputs "ref" after the update is done.
-   *  This makes it easier to chain operations that need to use the reset value.
-   *
-   * @param <T> data type for {@code outputRef()} output
-   * @param ref Should be from a `Variable` node.
-   * @param value The value to be added to the variable.
-   * @param options carries optional attributes values
-   * @return a new instance of AssignAdd
-   */
-  public <T extends TType> AssignAdd<T> assignAdd(Operand<T> ref, Operand<T> value,
-      AssignAdd.Options... options) {
-    return AssignAdd.create(scope, ref, value, options);
-  }
-
-  /**
-   * Adds a value to the current value of a variable.
-   *  <p>
-   *  Any ReadVariableOp with a control dependency on this op is guaranteed to
-   *  see the incremented value or a subsequent newer one.
-   *
-   * @param resource handle to the resource in which to store the variable.
-   * @param value the value by which the variable will be incremented.
-   * @return a new instance of AssignAddVariableOp
-   */
-  public <T extends TType> AssignAddVariableOp assignAddVariableOp(Operand<?> resource,
-      Operand<T> value) {
-    return AssignAddVariableOp.create(scope, resource, value);
-  }
-
-  /**
-   * Update 'ref' by subtracting 'value' from it.
-   *  <p>
-   *  This operation outputs "ref" after the update is done.
-   *  This makes it easier to chain operations that need to use the reset value.
-   *
-   * @param <T> data type for {@code outputRef()} output
-   * @param ref Should be from a `Variable` node.
-   * @param value The value to be subtracted to the variable.
-   * @param options carries optional attributes values
-   * @return a new instance of AssignSub
-   */
-  public <T extends TType> AssignSub<T> assignSub(Operand<T> ref, Operand<T> value,
-      AssignSub.Options... options) {
-    return AssignSub.create(scope, ref, value, options);
-  }
-
-  /**
-   * Subtracts a value from the current value of a variable.
-   *  <p>
-   *  Any ReadVariableOp with a control dependency on this op is guaranteed to
-   *  see the decremented value or a subsequent newer one.
-   *
-   * @param resource handle to the resource in which to store the variable.
-   * @param value the value by which the variable will be incremented.
-   * @return a new instance of AssignSubVariableOp
-   */
-  public <T extends TType> AssignSubVariableOp assignSubVariableOp(Operand<?> resource,
-      Operand<T> value) {
-    return AssignSubVariableOp.create(scope, resource, value);
-  }
-
-  /**
-   * Assigns a new value to a variable.
-   *  <p>
-   *  Any ReadVariableOp with a control dependency on this op is guaranteed to return
-   *  this value or a subsequent newer value of the variable.
-   *
-   * @param resource handle to the resource in which to store the variable.
-   * @param value the value to set the new tensor to use.
-   * @return a new instance of AssignVariableOp
-   */
-  public <T extends TType> AssignVariableOp assignVariableOp(Operand<?> resource,
-      Operand<T> value) {
-    return AssignVariableOp.create(scope, resource, value);
-  }
-
-  /**
-   * Defines a barrier that persists across different graph executions.
-   *  <p>
-   *  A barrier represents a key-value map, where each key is a string, and
-   *  each value is a tuple of tensors.
-   *  <p>
-   *  At runtime, the barrier contains 'complete' and 'incomplete'
-   *  elements. A complete element has defined tensors for all components of
-   *  its value tuple, and may be accessed using BarrierTakeMany. An
-   *  incomplete element has some undefined components in its value tuple,
-   *  and may be updated using BarrierInsertMany.
-   *
-   * @param componentTypes The type of each component in a value.
-   * @param options carries optional attributes values
-   * @return a new instance of Barrier
-   */
-  public Barrier barrier(List<DataType<?>> componentTypes, Barrier.Options... options) {
-    return Barrier.create(scope, componentTypes, options);
-  }
-
-  /**
-   * Closes the given barrier.
-   *  <p>
-   *  This operation signals that no more new elements will be inserted in the
-   *  given barrier. Subsequent InsertMany that try to introduce a new key will fail.
-   *  Subsequent InsertMany operations that just add missing components to already
-   *  existing elements will continue to succeed. Subsequent TakeMany operations will
-   *  continue to succeed if sufficient completed elements remain in the barrier.
-   *  Subsequent TakeMany operations that would block will fail immediately.
-   *
-   * @param handle The handle to a barrier.
-   * @param options carries optional attributes values
-   * @return a new instance of BarrierClose
-   */
-  public BarrierClose barrierClose(Operand<TString> handle, BarrierClose.Options... options) {
-    return BarrierClose.create(scope, handle, options);
-  }
-
-  /**
-   * Computes the number of incomplete elements in the given barrier.
-   *
-   * @param handle The handle to a barrier.
-   * @return a new instance of BarrierIncompleteSize
-   */
-  public BarrierIncompleteSize barrierIncompleteSize(Operand<TString> handle) {
-    return BarrierIncompleteSize.create(scope, handle);
-  }
-
-  /**
-   * For each key, assigns the respective value to the specified component.
-   *  <p>
-   *  If a key is not found in the barrier, this operation will create a new
-   *  incomplete element. If a key is found in the barrier, and the element
-   *  already has a value at component_index, this operation will fail with
-   *  INVALID_ARGUMENT, and leave the barrier in an undefined state.
-   *
-   * @param handle The handle to a barrier.
-   * @param keys A one-dimensional tensor of keys, with length n.
-   * @param values An any-dimensional tensor of values, which are associated with the
-   *  respective keys. The 0th dimension must have length n.
-   * @param componentIndex The component of the barrier elements that is being assigned.
-   * @return a new instance of BarrierInsertMany
-   */
-  public <T extends TType> BarrierInsertMany barrierInsertMany(Operand<TString> handle,
-      Operand<TString> keys, Operand<T> values, Long componentIndex) {
-    return BarrierInsertMany.create(scope, handle, keys, values, componentIndex);
-  }
-
-  /**
-   * Computes the number of complete elements in the given barrier.
-   *
-   * @param handle The handle to a barrier.
-   * @return a new instance of BarrierReadySize
-   */
-  public BarrierReadySize barrierReadySize(Operand<TString> handle) {
-    return BarrierReadySize.create(scope, handle);
-  }
-
-  /**
-   * Takes the given number of completed elements from a barrier.
-   *  <p>
-   *  This operation concatenates completed-element component tensors along
-   *  the 0th dimension to make a single component tensor.
-   *  <p>
-   *  Elements come out of the barrier when they are complete, and in the order
-   *  in which they were placed into the barrier.  The indices output provides
-   *  information about the batch in which each element was originally inserted
-   *  into the barrier.
-   *
-   * @param handle The handle to a barrier.
-   * @param numElements A single-element tensor containing the number of elements to
-   *  take.
-   * @param componentTypes The type of each component in a value.
-   * @param options carries optional attributes values
-   * @return a new instance of BarrierTakeMany
-   */
-  public BarrierTakeMany barrierTakeMany(Operand<TString> handle, Operand<TInt32> numElements,
-      List<DataType<?>> componentTypes, BarrierTakeMany.Options... options) {
-    return BarrierTakeMany.create(scope, handle, numElements, componentTypes, options);
-  }
-
-  /**
-   * Batches all input tensors nondeterministically.
-   *  <p>
-   *  When many instances of this Op are being run concurrently with the same
-   *  container/shared_name in the same device, some will output zero-shaped Tensors
-   *  and others will output Tensors of size up to max_batch_size.
-   *  <p>
-   *  All Tensors in in_tensors are batched together (so, for example, labels and
-   *  features should be batched with a single instance of this operation.
-   *  <p>
-   *  Each invocation of batch emits an `id` scalar which will be used to identify
-   *  this particular invocation when doing unbatch or its gradient.
-   *  <p>
-   *  Each op which emits a non-empty batch will also emit a non-empty batch_index
-   *  Tensor, which, is a [K, 3] matrix where each row contains the invocation's id,
-   *  start, and length of elements of each set of Tensors present in batched_tensors.
-   *  <p>
-   *  Batched tensors are concatenated along the first dimension, and all tensors in
-   *  in_tensors must have the first dimension of the same size.
-   *  <p>
-   *  in_tensors: The tensors to be batched.
-   *  num_batch_threads: Number of scheduling threads for processing batches of work.
-   *   Determines the number of batches processed in parallel.
-   *  max_batch_size: Batch sizes will never be bigger than this.
-   *  batch_timeout_micros: Maximum number of microseconds to wait before outputting
-   *   an incomplete batch.
-   *  allowed_batch_sizes: Optional list of allowed batch sizes. If left empty, does
-   *   nothing. Otherwise, supplies a list of batch sizes, causing the op to pad
-   *   batches up to one of those sizes. The entries must increase monotonically, and
-   *   the final entry must equal max_batch_size.
-   *  grad_timeout_micros: The timeout to use for the gradient. See Unbatch.
-   *  batched_tensors: Either empty tensors or a batch of concatenated Tensors.
-   *  batch_index: If out_tensors is non-empty, has information to invert it.
-   *  container: Controls the scope of sharing of this batch.
-   *  id: always contains a scalar with a unique ID for this invocation of Batch.
-   *  shared_name: Concurrently running instances of batch in the same device with the
-   *   same container and shared_name will batch their elements together. If left
-   *   empty, the op name will be used as the shared name.
-   *  T: the types of tensors to be batched.
-   *
-   * @param inTensors
-   * @param numBatchThreads
-   * @param maxBatchSize
-   * @param batchTimeoutMicros
-   * @param gradTimeoutMicros
-   * @param options carries optional attributes values
-   * @return a new instance of Batch
-   */
-  public Batch batch(Iterable<Operand<?>> inTensors, Long numBatchThreads, Long maxBatchSize,
-      Long batchTimeoutMicros, Long gradTimeoutMicros, Batch.Options... options) {
-    return Batch.create(scope, inTensors, numBatchThreads, maxBatchSize, batchTimeoutMicros, gradTimeoutMicros, options);
-  }
-
-  /**
-   * BatchToSpace for 4-D tensors of type T.
-   *  <p>
-   *  This is a legacy version of the more general BatchToSpaceND.
-   *  <p>
-   *  Rearranges (permutes) data from batch into blocks of spatial data, followed by
-   *  cropping. This is the reverse transformation of SpaceToBatch. More specifically,
-   *  this op outputs a copy of the input tensor where values from the `batch`
-   *  dimension are moved in spatial blocks to the `height` and `width` dimensions,
-   *  followed by cropping along the `height` and `width` dimensions.
-   *
-   * @param <T> data type for {@code output()} output
-   * @param input 4-D tensor with shape
-   *  `[batch<i>block_size</i>block_size, height_pad/block_size, width_pad/block_size,
-   *    depth]`. Note that the batch size of the input tensor must be divisible by
-   *  `block_size * block_size`.
-   * @param crops 2-D tensor of non-negative integers with shape `[2, 2]`. It specifies
-   *  how many elements to crop from the intermediate result across the spatial
-   *  dimensions as follows:
-   *  <p>
-   *      crops = [[crop_top, crop_bottom], [crop_left, crop_right]]
-   * @param blockSize
-   * @return a new instance of BatchToSpace
-   */
-  public <T extends TType, U extends TNumber> BatchToSpace<T> batchToSpace(Operand<T> input,
-      Operand<U> crops, Long blockSize) {
-    return BatchToSpace.create(scope, input, crops, blockSize);
-  }
-
-  /**
-   * BatchToSpace for N-D tensors of type T.
-   *  <p>
-   *  This operation reshapes the "batch" dimension 0 into `M + 1` dimensions of shape
-   *  `block_shape + [batch]`, interleaves these blocks back into the grid defined by
-   *  the spatial dimensions `[1, ..., M]`, to obtain a result with the same rank as
-   *  the input.  The spatial dimensions of this intermediate result are then
-   *  optionally cropped according to `crops` to produce the output.  This is the
-   *  reverse of SpaceToBatch.  See below for a precise description.
-   *
-   * @param <T> data type for {@code output()} output
-   * @param input N-D with shape `input_shape = [batch] + spatial_shape + remaining_shape`,
-   *  where spatial_shape has M dimensions.
-   * @param blockShape 1-D with shape `[M]`, all values must be >= 1.
-   * @param crops 2-D with shape `[M, 2]`, all values must be >= 0.
-   *    `crops[i] = [crop_start, crop_end]` specifies the amount to crop from input
-   *    dimension `i + 1`, which corresponds to spatial dimension `i`.  It is
-   *    required that
-   *    `crop_start[i] + crop_end[i] <= block_shape[i] * input_shape[i + 1]`.
-   *  <p>
-   *  This operation is equivalent to the following steps:
-   *  <p>
-   *  1. Reshape `input` to `reshaped` of shape:
-   *       [block_shape[0], ..., block_shape[M-1],
-   *        batch / prod(block_shape),
-   *        input_shape[1], ..., input_shape[N-1]]
-   *  <p>
-   *  2. Permute dimensions of `reshaped` to produce `permuted` of shape
-   *       [batch / prod(block_shape),
-   *  <p>
-   *        input_shape[1], block_shape[0],
-   *        ...,
-   *        input_shape[M], block_shape[M-1],
-   *  <p>
-   *        input_shape[M+1], ..., input_shape[N-1]]
-   *  <p>
-   *  3. Reshape `permuted` to produce `reshaped_permuted` of shape
-   *       [batch / prod(block_shape),
-   *  <p>
-   *        input_shape[1] * block_shape[0],
-   *        ...,
-   *        input_shape[M] * block_shape[M-1],
-   *  <p>
-   *        input_shape[M+1],
-   *        ...,
-   *        input_shape[N-1]]
-   *  <p>
-   *  4. Crop the start and end of dimensions `[1, ..., M]` of
-   *     `reshaped_permuted` according to `crops` to produce the output of shape:
-   *       [batch / prod(block_shape),
-   *  <p>
-   *        input_shape[1] * block_shape[0] - crops[0,0] - crops[0,1],
-   *        ...,
-   *        input_shape[M] * block_shape[M-1] - crops[M-1,0] - crops[M-1,1],
-   *  <p>
-   *        input_shape[M+1], ..., input_shape[N-1]]
-   *  <p>
-   *  Some examples:
-   *  <p>
-   *  (1) For the following input of shape `[4, 1, 1, 1]`, `block_shape = [2, 2]`, and
-   *      `crops = [[0, 0], [0, 0]]`:
-   *  <pre>{@code
-   *  [[[[1]]], [[[2]]], [[[3]]], [[[4]]]]
-   *  }</pre>
-   *  The output tensor has shape `[1, 2, 2, 1]` and value:
-   *  <pre>{@code
-   *  x = [[[[1], [2]], [[3], [4]]]]
-   *  }</pre>
-   *  (2) For the following input of shape `[4, 1, 1, 3]`, `block_shape = [2, 2]`, and
-   *      `crops = [[0, 0], [0, 0]]`:
-   *  <pre>{@code
-   *  [[[[1, 2, 3]]], [[[4, 5, 6]]], [[[7, 8, 9]]], [[[10, 11, 12]]]]
-   *  }</pre>
-   *  The output tensor has shape `[1, 2, 2, 3]` and value:
-   *  <pre>{@code
-   *  x = [[[[1, 2, 3], [4, 5, 6]],
-   *        [[7, 8, 9], [10, 11, 12]]]]
-   *  }</pre>
-   *  (3) For the following input of shape `[4, 2, 2, 1]`, `block_shape = [2, 2]`, and
-   *      `crops = [[0, 0], [0, 0]]`:
-   *  <pre>{@code
-   *  x = [[[[1], [3]], [[9], [11]]],
-   *       [[[2], [4]], [[10], [12]]],
-   *       [[[5], [7]], [[13], [15]]],
-   *       [[[6], [8]], [[14], [16]]]]
-   *  }</pre>
-   *  The output tensor has shape `[1, 4, 4, 1]` and value:
-   *  <pre>{@code
-   *  x = [[[[1],   [2],  [3],  [4]],
-   *       [[5],   [6],  [7],  [8]],
-   *       [[9],  [10], [11],  [12]],
-   *       [[13], [14], [15],  [16]]]]
-   *  }</pre>
-   *  (4) For the following input of shape `[8, 1, 3, 1]`, `block_shape = [2, 2]`, and
-   *      `crops = [[0, 0], [2, 0]]`:
-   *  <pre>{@code
-   *  x = [[[[0], [1], [3]]], [[[0], [9], [11]]],
-   *       [[[0], [2], [4]]], [[[0], [10], [12]]],
-   *       [[[0], [5], [7]]], [[[0], [13], [15]]],
-   *       [[[0], [6], [8]]], [[[0], [14], [16]]]]
-   *  }</pre>
-   *  The output tensor has shape `[2, 2, 4, 1]` and value:
-   *  <pre>{@code
-   *  x = [[[[1],   [2],  [3],  [4]],
-   *        [[5],   [6],  [7],  [8]]],
-   *       [[[9],  [10], [11],  [12]],
-   *        [[13], [14], [15],  [16]]]]
-   *  }</pre>
-   * @return a new instance of BatchToSpaceNd
-   */
-  public <T extends TType, U extends TNumber, V extends TNumber> BatchToSpaceNd<T> batchToSpaceNd(
-      Operand<T> input, Operand<U> blockShape, Operand<V> crops) {
-    return BatchToSpaceNd.create(scope, input, blockShape, crops);
-  }
-
-  /**
-   * Bitcasts a tensor from one type to another without copying data.
-   *  <p>
-   *  Given a tensor `input`, this operation returns a tensor that has the same buffer
-   *  data as `input` with datatype `type`.
-   *  <p>
-   *  If the input datatype `T` is larger than the output datatype `type` then the
-   *  shape changes from [...] to [..., sizeof(`T`)/sizeof(`type`)].
-   *  <p>
-   *  If `T` is smaller than `type`, the operator requires that the rightmost
-   *  dimension be equal to sizeof(`type`)/sizeof(`T`). The shape then goes from
-   *  [..., sizeof(`type`)/sizeof(`T`)] to [...].
-   *  <p>
-   *  tf.bitcast() and tf.cast() work differently when real dtype is casted as a complex dtype
-   *  (e.g. tf.complex64 or tf.complex128) as tf.cast() make imaginary part 0 while tf.bitcast()
-   *  gives module error.
-   *  For example,
-   *  <p>
-   *  Example 1:
-   *  <p>
-   *  >>> a = [1., 2., 3.]
-   *  >>> equality_bitcast = tf.bitcast(a, tf.complex128)
-   *  Traceback (most recent call last):
-   *  ...
-   *  InvalidArgumentError: Cannot bitcast from 1 to 18 [Op:Bitcast]
-   *  >>> equality_cast = tf.cast(a, tf.complex128)
-   *  >>> print(equality_cast)
-   *  tf.Tensor([1.+0.j 2.+0.j 3.+0.j], shape=(3,), dtype=complex128)
-   *  <p>
-   *  Example 2:
-   *  <p>
-   *  >>> tf.bitcast(tf.constant(0xffffffff, dtype=tf.uint32), tf.uint8)
-   *  <tf.Tensor: shape=(4,), dtype=uint8, numpy=array([255, 255, 255, 255], dtype=uint8)>
-   *  <p>
-   *  Example 3:
-   *  <p>
-   *  >>> x = [1., 2., 3.]
-   *  >>> y = [0., 2., 3.]
-   *  >>> equality= tf.equal(x,y)
-   *  >>> equality_cast = tf.cast(equality,tf.float32)
-   *  >>> equality_bitcast = tf.bitcast(equality_cast,tf.uint8)
-   *  >>> print(equality)
-   *  tf.Tensor([False True True], shape=(3,), dtype=bool)
-   *  >>> print(equality_cast)
-   *  tf.Tensor([0. 1. 1.], shape=(3,), dtype=float32)
-   *  >>> print(equality_bitcast)
-   *  tf.Tensor(
-   *      [[  0   0   0   0]
-   *       [  0   0 128  63]
-   *       [  0   0 128  63]], shape=(3, 4), dtype=uint8)
-   *  <p>
-   *  <i>NOTE</i>: Bitcast is implemented as a low-level cast, so machines with different
-   *  endian orderings will give different results.
-   *
-   * @param <U> data type for {@code output()} output
-   * @param input
-   * @param type
-   * @return a new instance of Bitcast
-   */
-  public <U extends TType, T extends TType> Bitcast<U> bitcast(Operand<T> input, DataType<U> type) {
-    return Bitcast.create(scope, input, type);
-  }
-
-  /**
-   * Return the shape of s0 op s1 with broadcast.
-   *  <p>
-   *  Given `s0` and `s1`, tensors that represent shapes, compute `r0`, the
-   *  broadcasted shape. `s0`, `s1` and `r0` are all integer vectors.
-   *
-   * @param <T> data type for {@code r0()} output
-   * @param s0
-   * @param s1
-   * @return a new instance of BroadcastDynamicShape
-   */
-  public <T extends TNumber> BroadcastDynamicShape<T> broadcastDynamicShape(Operand<T> s0,
-      Operand<T> s1) {
-    return BroadcastDynamicShape.create(scope, s0, s1);
-  }
-
-  /**
-   * Broadcast an array for a compatible shape.
-   *  <p>
-   *  Broadcasting is the process of making arrays to have compatible shapes
-   *  for arithmetic operations. Two shapes are compatible if for each
-   *  dimension pair they are either equal or one of them is one. When trying
-   *  to broadcast a Tensor to a shape, it starts with the trailing dimensions,
-   *  and works its way forward.
-   *  <p>
-   *  For example,
-   *  <p>
-   *  >>> x = tf.constant([1, 2, 3])
-   *  >>> y = tf.broadcast_to(x, [3, 3])
-   *  >>> print(y)
-   *  tf.Tensor(
-   *      [[1 2 3]
-   *       [1 2 3]
-   *       [1 2 3]], shape=(3, 3), dtype=int32)
-   *  <p>
-   *  In the above example, the input Tensor with the shape of `[1, 3]`
-   *  is broadcasted to output Tensor with shape of `[3, 3]`.
-   *
-   * @param <T> data type for {@code output()} output
-   * @param input A Tensor to broadcast.
-   * @param shape An 1-D `int` Tensor. The shape of the desired output.
-   * @return a new instance of BroadcastTo
-   */
-  public <T extends TType, U extends TNumber> BroadcastTo<T> broadcastTo(Operand<T> input,
-      Operand<U> shape) {
-    return BroadcastTo.create(scope, input, shape);
-  }
-
-  /**
-   * Bucketizes 'input' based on 'boundaries'.
-   *  <p>
-   *  For example, if the inputs are
-   *      boundaries = [0, 10, 100]
-   *      input = [[-5, 10000]
-   *               [150,   10]
-   *               [5,    100]]
-   *  <p>
-   *  then the output will be
-   *      output = [[0, 3]
-   *                [3, 2]
-   *                [1, 3]]
-   *
-   * @param input Any shape of Tensor contains with int or float type.
-   * @param boundaries A sorted list of floats gives the boundary of the buckets.
-   * @return a new instance of Bucketize
-   */
-  public <T extends TNumber> Bucketize bucketize(Operand<T> input, List<Float> boundaries) {
-    return Bucketize.create(scope, input, boundaries);
-  }
-
-  /**
-   * Create a constant from a Tensor.
-   *
-   * @param scope is a scope used to add the underlying operation.
-   * @param tensor a Tensor holding the constant value
-   * @return a constant of the same data type as `tensor`
-   */
-  public <T extends TType> Constant<T> capture(T tensor) {
-    return Constant.create(scope, tensor);
-  }
-
-  /**
-   * Clips tensor values to a specified min and max.
-   *  <p>
-   *  Given a tensor `t`, this operation returns a tensor of the same type and
-   *  shape as `t` with its values clipped to `clip_value_min` and `clip_value_max`.
-   *  Any values less than `clip_value_min` are set to `clip_value_min`. Any values
-   *  greater than `clip_value_max` are set to `clip_value_max`.
-   *
-   * @param <T> data type for {@code output()} output
-   * @param t A `Tensor`.
-   * @param clipValueMin A 0-D (scalar) `Tensor`, or a `Tensor` with the same shape
-   *  as `t`. The minimum value to clip by.
-   * @param clipValueMax A 0-D (scalar) `Tensor`, or a `Tensor` with the same shape
-   *  as `t`. The maximum value to clip by.
-   * @return a new instance of ClipByValue
-   */
-  public <T extends TType> ClipByValue<T> clipByValue(Operand<T> t, Operand<T> clipValueMin,
-      Operand<T> clipValueMax) {
-    return ClipByValue.create(scope, t, clipValueMin, clipValueMax);
-  }
-
-  /**
-   * Concatenates tensors along one dimension.
-   *
-   * @param <T> data type for {@code output()} output
-   * @param values List of `N` Tensors to concatenate. Their ranks and types must match,
-   *  and their sizes must match in all dimensions except `concat_dim`.
-   * @param axis 0-D.  The dimension along which to concatenate.  Must be in the
-   *  range [-rank(values), rank(values)).
-   * @return a new instance of Concat
-   */
-  public <T extends TType, U extends TNumber> Concat<T> concat(Iterable<Operand<T>> values,
-      Operand<U> axis) {
-    return Concat.create(scope, values, axis);
-  }
-
-  /**
-   * Creates a constant of {@code long} elements that is a copy of a given n-dimensional array.
-   *
-   * @param scope is a scope used to add the underlying operation.
-   * @param data an n-dimensional array of {@code long} elements.
-   * @return a long constant
-   */
-  public Constant<TInt64> constant(LongNdArray data) {
-    return Constant.tensorOf(scope, data);
-  }
-
-  /**
-   * Creates a rank-1 constant of {@code int} elements.
-   *
-   * @param scope is a scope used to add the underlying operation.
-   * @param data An array containing the values to put into the new constant. The dimensions of the
-   *      new constant will match those of the array.
-   * @return an integer constant
-   */
-  public Constant<TInt32> constant(int[] data) {
-    return Constant.vectorOf(scope, data);
-  }
-
-  /**
-   * Creates a rank-3 constant of {@code int} elements.
-   *
-   * @param scope is a scope used to add the underlying operation.
-   * @param data An array containing the values to put into the new constant. The dimensions of the
-   *      new constant will match those of the array.
-   * @return an integer constant
-   */
-  public Constant<TInt32> constant(int[][][] data) {
-    return Constant.tensorOf(scope, data);
-  }
-
-  /**
-   * Creates a constant containing a single {@code double} element.
-   *
-   * @param scope is a scope used to add the underlying operation.
-   * @param data The value to put into the new constant.
-   * @return a double constant
-   */
-  public Constant<TFloat64> constant(double data) {
-    return Constant.scalarOf(scope, data);
-  }
-
-  /**
-   * Creates a rank-5 constant of {@code long} elements.
-   *
-   * @param scope is a scope used to add the underlying operation.
-   * @param data An array containing the values to put into the new constant. The dimensions of the
-   *      new constant will match those of the array.
-   * @return a long constant
-   */
-  public Constant<TInt64> constant(long[][][][][] data) {
-    return Constant.tensorOf(scope, data);
-  }
-
-  /**
-   * Creates a rank-5 constant of {@code boolean} elements.
-   *
-   * @param scope is a scope used to add the underlying operation.
-   * @param data An array containing the values to put into the new constant. The dimensions of the
-   *      new constant will match those of the array.
-   * @return a boolean constant
-   */
-  public Constant<TBool> constant(boolean[][][][][] data) {
-    return Constant.tensorOf(scope, data);
-  }
-
-  /**
-   * Creates a constant of {@code int} elements that is a copy of a given n-dimensional array.
-   *
-   * @param scope is a scope used to add the underlying operation.
-   * @param data an n-dimensional array of {@code int} elements.
-   * @return an integer constant
-   */
-  public Constant<TInt32> constant(IntNdArray data) {
-    return Constant.tensorOf(scope, data);
-  }
-
-  /**
-   * Creates a constant of {@code double} elements that is a copy of a given n-dimensional array.
-   *
-   * @param scope is a scope used to add the underlying operation.
-   * @param data an n-dimensional array of {@code double} elements.
-   * @return a double constant
-   */
-  public Constant<TFloat64> constant(DoubleNdArray data) {
-    return Constant.tensorOf(scope, data);
-  }
-
-  /**
-   * Creates a rank-4 constant of {@code int} elements.
-   *
-   * @param scope is a scope used to add the underlying operation.
-   * @param data An array containing the values to put into the new constant. The dimensions of the
-   *      new constant will match those of the array.
-   * @return an integer constant
-   */
-  public Constant<TInt32> constant(int[][][][] data) {
-    return Constant.tensorOf(scope, data);
-  }
-
-  /**
-   * Creates a rank-6 constant of {@code float} elements.
-   *
-   * @param scope is a scope used to add the underlying operation.
-   * @param data An array containing the values to put into the new constant. The dimensions of the
-   *      new constant will match those of the array.
-   * @return a float constant
-   */
-  public Constant<TFloat32> constant(float[][][][][][] data) {
-    return Constant.tensorOf(scope, data);
-  }
-
-  /**
-   * Creates a constant containing a single {@code byte} element.
-   *
-   * @param scope is a scope used to add the underlying operation.
-   * @param data The value to put into the new constant.
-   * @return a byte constant
-   */
-  public Constant<TUint8> constant(byte data) {
-    return Constant.scalarOf(scope, data);
-  }
-
-  /**
-   * Creates a rank-3 constant of {@code boolean} elements.
-   *
-   * @param scope is a scope used to add the underlying operation.
-   * @param data An array containing the values to put into the new constant. The dimensions of the
-   *      new constant will match those of the array.
-   * @return a boolean constant
-   */
-  public Constant<TBool> constant(boolean[][][] data) {
-    return Constant.tensorOf(scope, data);
-  }
-
-  /**
-   * Creates a rank-4 constant of {@code float} elements.
-   *
-   * @param scope is a scope used to add the underlying operation.
-   * @param data An array containing the values to put into the new constant. The dimensions of the
-   *      new constant will match those of the array.
-   * @return a float constant
-   */
-  public Constant<TFloat32> constant(float[][][][] data) {
-    return Constant.tensorOf(scope, data);
-  }
-
-  /**
-   * Creates a rank-2 constant of {@code long} elements.
-   *
-   * @param scope is a scope used to add the underlying operation.
-   * @param data An array containing the values to put into the new constant. The dimensions of the
-   *      new constant will match those of the array.
-   * @return a long constant
-   */
-  public Constant<TInt64> constant(long[][] data) {
-    return Constant.tensorOf(scope, data);
-  }
-
-  /**
-   * Creates a rank-5 constant of {@code byte} elements.
-   *
-   * @param scope is a scope used to add the underlying operation.
-   * @param data An array containing the values to put into the new constant. The dimensions of the
-   *      new constant will match those of the array.
-   * @return a byte constant
-   */
-  public Constant<TUint8> constant(byte[][][][][] data) {
-    return Constant.tensorOf(scope, data);
-  }
-
-  /**
-   * Creates a constant of {@code boolean} elements that is a copy of a given n-dimensional array.
-   *
-   * @param scope is a scope used to add the underlying operation.
-   * @param data an n-dimensional array of {@code boolean} elements.
-   * @return a boolean constant
-   */
-  public Constant<TBool> constant(BooleanNdArray data) {
-    return Constant.tensorOf(scope, data);
-  }
-
-  /**
-   * Creates a rank-2 constant of {@code float} elements.
-   *
-   * @param scope is a scope used to add the underlying operation.
-   * @param data An array containing the values to put into the new constant. The dimensions of the
-   *      new constant will match those of the array.
-   * @return a float constant
-   */
-  public Constant<TFloat32> constant(float[][] data) {
-    return Constant.tensorOf(scope, data);
-  }
-
-  /**
-   * Creates a constant of {@code byte} elements that is a copy of a given n-dimensional array.
-   *
-   * @param scope is a scope used to add the underlying operation.
-   * @param data an n-dimensional array of {@code byte} elements.
-   * @return a byte constant
-   */
-  public Constant<TUint8> constant(ByteNdArray data) {
-    return Constant.tensorOf(scope, data);
-  }
-
-  /**
-   * Creates a rank-2 constant of {@code byte} elements.
-   *
-   * @param scope is a scope used to add the underlying operation.
-   * @param data An array containing the values to put into the new constant. The dimensions of the
-   *      new constant will match those of the array.
-   * @return a byte constant
-   */
-  public Constant<TUint8> constant(byte[][] data) {
-    return Constant.tensorOf(scope, data);
-  }
-
-  /**
-   * Creates a rank-5 constant of {@code double} elements.
-   *
-   * @param scope is a scope used to add the underlying operation.
-   * @param data An array containing the values to put into the new constant. The dimensions of the
-   *      new constant will match those of the array.
-   * @return a double constant
-   */
-  public Constant<TFloat64> constant(double[][][][][] data) {
-    return Constant.tensorOf(scope, data);
-  }
-
-  /**
-   * Creates a rank-3 constant of {@code float} elements.
-   *
-   * @param scope is a scope used to add the underlying operation.
-   * @param data An array containing the values to put into the new constant. The dimensions of the
-   *      new constant will match those of the array.
-   * @return a float constant
-   */
-  public Constant<TFloat32> constant(float[][][] data) {
-    return Constant.tensorOf(scope, data);
-  }
-
-  /**
-   * Creates a rank-1 constant of {@code byte} elements.
-   *
-   * @param scope is a scope used to add the underlying operation.
-   * @param data An array containing the values to put into the new constant. The dimensions of the
-   *      new constant will match those of the array.
-   * @return a byte constant
-   */
-  public Constant<TUint8> constant(byte[] data) {
-    return Constant.vectorOf(scope, data);
-  }
-
-  /**
-   * Creates a rank-1 constant of {@code float} elements.
-   *
-   * @param scope is a scope used to add the underlying operation.
-   * @param data An array containing the values to put into the new constant. The dimensions of the
-   *      new constant will match those of the array.
-   * @return a float constant
-   */
-  public Constant<TFloat32> constant(float[] data) {
-    return Constant.vectorOf(scope, data);
-  }
-
-  /**
-   * Creates a rank-2 constant of {@code boolean} elements.
-   *
-   * @param scope is a scope used to add the underlying operation.
-   * @param data An array containing the values to put into the new constant. The dimensions of the
-   *      new constant will match those of the array.
-   * @return a boolean constant
-   */
-  public Constant<TBool> constant(boolean[][] data) {
-    return Constant.tensorOf(scope, data);
-  }
-
-  /**
-   * Creates a constant of {@code String} elements that is a copy of a given n-dimensional array,
-   *  using the default UTF-8 encoding.
-   *
-   * @param scope is a scope used to add the underlying operation.
-   * @param data an n-dimensional array of {@code String} elements.
-   * @return a string constant
-   */
-  public Constant<TString> constant(NdArray<String> data) {
-    return Constant.tensorOf(scope, data);
-  }
-
-  /**
-   * Creates a {@code String} constant using the default, UTF-8 encoding.
-   *
-   * @param scope is a scope used to add the underlying operation.
-   * @param data The string to put into the new constant.
-   * @return a string constant
-   */
-  public Constant<TString> constant(String data) {
-    return Constant.scalarOf(scope, data);
-  }
-
-  /**
-   * Creates a rank-4 constant of {@code double} elements.
-   *
-   * @param scope is a scope used to add the underlying operation.
-   * @param data An array containing the values to put into the new constant. The dimensions of the
-   *      new constant will match those of the array.
-   * @return a double constant
-   */
-  public Constant<TFloat64> constant(double[][][][] data) {
-    return Constant.tensorOf(scope, data);
-  }
-
-  /**
-   * Creates a rank-2 constant of {@code double} elements.
-   *
-   * @param scope is a scope used to add the underlying operation.
-   * @param data An array containing the values to put into the new constant. The dimensions of the
-   *      new constant will match those of the array.
-   * @return a double constant
-   */
-  public Constant<TFloat64> constant(double[][] data) {
-    return Constant.tensorOf(scope, data);
-  }
-
-  /**
-   * Creates a constant containing a single {@code int} element.
-   *
-   * @param scope is a scope used to add the underlying operation.
-   * @param data The value to put into the new constant.
-   * @return an integer constant
-   */
-  public Constant<TInt32> constant(int data) {
-    return Constant.scalarOf(scope, data);
-  }
-
-  /**
-   * Creates a rank-4 constant of {@code byte} elements.
-   *
-   * @param scope is a scope used to add the underlying operation.
-   * @param data An array containing the values to put into the new constant. The dimensions of the
-   *      new constant will match those of the array.
-   * @return a byte constant
-   */
-  public Constant<TUint8> constant(byte[][][][] data) {
-    return Constant.tensorOf(scope, data);
-  }
-
-  /**
-   * Creates a rank-6 constant of {@code int} elements.
-   *
-   * @param scope is a scope used to add the underlying operation.
-   * @param data An array containing the values to put into the new constant. The dimensions of the
-   *      new constant will match those of the array.
-   * @return an integer constant
-   */
-  public Constant<TInt32> constant(int[][][][][][] data) {
-    return Constant.tensorOf(scope, data);
-  }
-
-  /**
-   * Creates a constant containing a single {@code long} element.
-   *
-   * @param scope is a scope used to add the underlying operation.
-   * @param data The value to put into the new constant.
-   * @return a long constant
-   */
-  public Constant<TInt64> constant(long data) {
-    return Constant.scalarOf(scope, data);
-  }
-
-  /**
-   * Creates a constant containing a single {@code float} element.
-   *
-   * @param scope is a scope used to add the underlying operation.
-   * @param data The value to put into the new constant.
-   * @return a float constant
-   */
-  public Constant<TFloat32> constant(float data) {
-    return Constant.scalarOf(scope, data);
-  }
-
-  /**
-   * Creates a rank-5 constant of {@code float} elements.
-   *
-   * @param scope is a scope used to add the underlying operation.
-   * @param data An array containing the values to put into the new constant. The dimensions of the
-   *      new constant will match those of the array.
-   * @return a float constant
-   */
-  public Constant<TFloat32> constant(float[][][][][] data) {
-    return Constant.tensorOf(scope, data);
-  }
-
-  /**
-   * Creates a rank-3 constant of {@code double} elements.
-   *
-   * @param scope is a scope used to add the underlying operation.
-   * @param data An array containing the values to put into the new constant. The dimensions of the
-   *      new constant will match those of the array.
-   * @return a double constant
-   */
-  public Constant<TFloat64> constant(double[][][] data) {
-    return Constant.tensorOf(scope, data);
-  }
-
-  /**
-   * Creates a rank-6 constant of {@code long} elements.
-   *
-   * @param scope is a scope used to add the underlying operation.
-   * @param data An array containing the values to put into the new constant. The dimensions of the
-   *      new constant will match those of the array.
-   * @return a long constant
-   */
-  public Constant<TInt64> constant(long[][][][][][] data) {
-    return Constant.tensorOf(scope, data);
-  }
-
-  /**
-   * Creates a rank-4 constant of {@code long} elements.
-   *
-   * @param scope is a scope used to add the underlying operation.
-   * @param data An array containing the values to put into the new constant. The dimensions of the
-   *      new constant will match those of the array.
-   * @return a long constant
-   */
-  public Constant<TInt64> constant(long[][][][] data) {
-    return Constant.tensorOf(scope, data);
-  }
-
-  /**
-   * Creates a rank-1 constant of {@code long} elements.
-   *
-   * @param scope is a scope used to add the underlying operation.
-   * @param data An array containing the values to put into the new constant. The dimensions of the
-   *      new constant will match those of the array.
-   * @return a long constant
-   */
-  public Constant<TInt64> constant(long[] data) {
-    return Constant.vectorOf(scope, data);
-  }
-
-  /**
-   * Creates a rank-1 constant of {@code boolean} elements.
-   *
-   * @param scope is a scope used to add the underlying operation.
-   * @param data An array containing the values to put into the new constant. The dimensions of the
-   *      new constant will match those of the array.
-   * @return a boolean constant
-   */
-  public Constant<TBool> constant(boolean[] data) {
-    return Constant.vectorOf(scope, data);
-  }
-
-  /**
-   * Creates a rank-3 constant of {@code byte} elements.
-   *
-   * @param scope is a scope used to add the underlying operation.
-   * @param data An array containing the values to put into the new constant. The dimensions of the
-   *      new constant will match those of the array.
-   * @return a byte constant
-   */
-  public Constant<TUint8> constant(byte[][][] data) {
-    return Constant.tensorOf(scope, data);
-  }
-
-  /**
-   * Creates a rank-6 constant of {@code byte} elements.
-   *
-   * @param scope is a scope used to add the underlying operation.
-   * @param data An array containing the values to put into the new constant. The dimensions of the
-   *      new constant will match those of the array.
-   * @return a byte constant
-   */
-  public Constant<TUint8> constant(byte[][][][][][] data) {
-    return Constant.tensorOf(scope, data);
-  }
-
-  /**
-   * Creates a rank-2 constant of {@code int} elements.
-   *
-   * @param scope is a scope used to add the underlying operation.
-   * @param data An array containing the values to put into the new constant. The dimensions of the
-   *      new constant will match those of the array.
-   * @return an integer constant
-   */
-  public Constant<TInt32> constant(int[][] data) {
-    return Constant.tensorOf(scope, data);
-  }
-
-  /**
-   * Creates a constant of {@code float} elements that is a copy of a given n-dimensional array.
-   *
-   * @param scope is a scope used to add the underlying operation.
-   * @param data an n-dimensional array of {@code float} elements.
-   * @return a float constant
-   */
-  public Constant<TFloat32> constant(FloatNdArray data) {
-    return Constant.tensorOf(scope, data);
-  }
-
-  /**
-   * Creates a rank-5 constant of {@code int} elements.
-   *
-   * @param scope is a scope used to add the underlying operation.
-   * @param data An array containing the values to put into the new constant. The dimensions of the
-   *      new constant will match those of the array.
-   * @return an integer constant
-   */
-  public Constant<TInt32> constant(int[][][][][] data) {
-    return Constant.tensorOf(scope, data);
-  }
-
-  /**
-   * Creates a rank-1 constant of {@code double} elements.
-   *
-   * @param scope is a scope used to add the underlying operation.
-   * @param data An array containing the values to put into the new constant. The dimensions of the
-   *      new constant will match those of the array.
-   * @return a double constant
-   */
-  public Constant<TFloat64> constant(double[] data) {
-    return Constant.vectorOf(scope, data);
-  }
-
-  /**
-   * Creates a rank-6 constant of {@code boolean} elements.
-   *
-   * @param scope is a scope used to add the underlying operation.
-   * @param data An array containing the values to put into the new constant. The dimensions of the
-   *      new constant will match those of the array.
-   * @return a boolean constant
-   */
-  public Constant<TBool> constant(boolean[][][][][][] data) {
-    return Constant.tensorOf(scope, data);
-  }
-
-  /**
-   * Creates a rank-6 constant of {@code double} elements.
-   *
-   * @param scope is a scope used to add the underlying operation.
-   * @param data An array containing the values to put into the new constant. The dimensions of the
-   *      new constant will match those of the array.
-   * @return a double constant
-   */
-  public Constant<TFloat64> constant(double[][][][][][] data) {
-    return Constant.tensorOf(scope, data);
-  }
-
-  /**
-   * Creates a constant containing a single {@code boolean} element.
-   *
-   * @param scope is a scope used to add the underlying operation.
-   * @param data The value to put into the new constant.
-   * @return a boolean constant
-   */
-  public Constant<TBool> constant(boolean data) {
-    return Constant.scalarOf(scope, data);
-  }
-
-  /**
-   * Creates a rank-4 constant of {@code boolean} elements.
-   *
-   * @param scope is a scope used to add the underlying operation.
-   * @param data An array containing the values to put into the new constant. The dimensions of the
-   *      new constant will match those of the array.
-   * @return a boolean constant
-   */
-  public Constant<TBool> constant(boolean[][][][] data) {
-    return Constant.tensorOf(scope, data);
-  }
-
-  /**
-   * Creates a rank-3 constant of {@code long} elements.
-   *
-   * @param scope is a scope used to add the underlying operation.
-   * @param data An array containing the values to put into the new constant. The dimensions of the
-   *      new constant will match those of the array.
-   * @return a long constant
-   */
-  public Constant<TInt64> constant(long[][][] data) {
-    return Constant.tensorOf(scope, data);
-  }
-
-  /**
-   * Creates a rank-1 constant of {@code long} elements representing the size of each dimensions of
-   *  the given shape.
-   *
-   * @param scope is a scope used to add the underlying operation.
-   * @param shape a shape
-   * @return a long constant
-   */
-  public Constant<TInt64> constant(Shape shape) {
-    return Constant.tensorOf(scope, shape);
-  }
-
-  /**
-   * Creates a constant of {@code String} elements, using the given charset.
-   *
-   * @param scope is a scope used to add the underlying operation.
-   * @param charset charset for encoding/decoding strings bytes.
-   * @param data An array containing the values to put into the new constant. String elements are
-   *      sequences of bytes from the last array dimension.
-   * @return the {@code String} constant
-   */
-  public Constant<TString> constant(Charset charset, String[] data) {
-    return Constant.vectorOf(scope, charset, data);
-  }
-
-  /**
-   * Creates a {@code String} constant using a specified encoding.
-   *
-   * @param scope is a scope used to add the underlying operation.
-   * @param charset The encoding from String to bytes.
-   * @param data The string to put into the new constant.
-   * @return a string constant
-   */
-  public Constant<TString> constant(Charset charset, String data) {
-    return Constant.scalarOf(scope, charset, data);
-  }
-
-  /**
-   * Creates a constant of {@code String} elements that is a copy of a given n-dimensional array,
-   *  using the given encoding.
-   *
-   * @param scope is a scope used to add the underlying operation.
-   * @param charset charset used to encode/decode string bytes.
-   * @param data an n-dimensional array of {@code String} elements.
-   * @return a string constant
-   */
-  public Constant<TString> constant(Charset charset, NdArray<String> data) {
-    return Constant.tensorOf(scope, charset, data);
-  }
-
-  /**
-   * Create a {@link TFloat32} constant with data from the given buffer.
-   *
-   * @param scope is a scope used to add the underlying operation.
-   * @param shape the tensor shape.
-   * @param data a buffer containing the tensor data.
-   * @return a float constant
-   * @throws IllegalArgumentException If the tensor shape is not compatible with the buffer
-   */
-  public Constant<TFloat32> constant(Shape shape, FloatDataBuffer data) {
-    return Constant.tensorOf(scope, shape, data);
-  }
-
-  /**
-   * Create a {@link TBool} constant with data from the given buffer.
-   *
-   * @param scope is a scope used to add the underlying operation.
-   * @param shape the tensor shape.
-   * @param data a buffer containing the tensor data.
-   * @return an boolean constant
-   * @throws IllegalArgumentException If the tensor shape is not compatible with the buffer
-   */
-  public Constant<TBool> constant(Shape shape, BooleanDataBuffer data) {
-    return Constant.tensorOf(scope, shape, data);
-  }
-
-  /**
-   * Create a {@link TUint8} constant with data from the given buffer.
-   *
-   * @param scope is a scope used to add the underlying operation.
-   * @param shape the tensor shape.
-   * @param data a buffer containing the tensor data.
-   * @return a byte constant
-   * @throws IllegalArgumentException If the tensor shape is not compatible with the buffer
-   */
-  public Constant<TUint8> constant(Shape shape, ByteDataBuffer data) {
-    return Constant.tensorOf(scope, shape, data);
-  }
-
-  /**
-   * Create a {@link TInt64} constant with data from the given buffer.
-   *
-   * @param scope is a scope used to add the underlying operation.
-   * @param shape the tensor shape.
-   * @param data a buffer containing the tensor data.
-   * @return a long constant
-   * @throws IllegalArgumentException If the tensor shape is not compatible with the buffer
-   */
-  public Constant<TInt64> constant(Shape shape, LongDataBuffer data) {
-    return Constant.tensorOf(scope, shape, data);
-  }
-
-  /**
-   * Create a {@link TString} constant with data from the given buffer, using the default UTF-8
-   *  encoding.
-   *
-   * @param scope is a scope used to add the underlying operation.
-   * @param shape the tensor shape.
-   * @param data a buffer containing the tensor data.
-   * @return a string constant
-   * @throws IllegalArgumentException If the tensor shape is not compatible with the buffer
-   */
-  public Constant<TString> constant(Shape shape, DataBuffer<String> data) {
-    return Constant.tensorOf(scope, shape, data);
-  }
-
-  /**
-   * Create a {@link TFloat64} constant with data from the given buffer.
-   *
-   * @param scope is a scope used to add the underlying operation.
-   * @param shape the tensor shape.
-   * @param data a buffer containing the tensor data.
-   * @return a double constant
-   * @throws IllegalArgumentException If the tensor shape is not compatible with the buffer
-   */
-  public Constant<TFloat64> constant(Shape shape, DoubleDataBuffer data) {
-    return Constant.tensorOf(scope, shape, data);
-  }
-
-  /**
-   * Create a {@link TInt32} constant with data from the given buffer.
-   *
-   * @param scope is a scope used to add the underlying operation.
-   * @param shape the tensor shape.
-   * @param data a buffer containing the tensor data.
-   * @return an integer constant
-   * @throws IllegalArgumentException If the tensor shape is not compatible with the buffer
-   */
-  public Constant<TInt32> constant(Shape shape, IntDataBuffer data) {
-    return Constant.tensorOf(scope, shape, data);
-  }
-
-  /**
-   * Create a {@link TString} constant with data from the given buffer, using the given encoding.
-   *
-   * @param scope is a scope used to add the underlying operation.
-   * @param charset charset used to encode/decode string bytes.
-   * @param shape the tensor shape.
-   * @param data a buffer containing the tensor data.
-   * @return a string constant
-   * @throws IllegalArgumentException If the tensor shape is not compatible with the buffer
-   */
-  public Constant<TString> constant(Charset charset, Shape shape, DataBuffer<String> data) {
-    return Constant.tensorOf(scope, charset, shape, data);
-  }
-
-  /**
-   * Create a constant with data from the given buffer.
-   *
-   * @param scope is a scope used to add the underlying operation.
-   * @param type the tensor datatype.
-   * @param shape the tensor shape.
-   * @param data a buffer containing the tensor data.
-   * @return a constant of type `type`
-   * @throws IllegalArgumentException If the tensor datatype or shape is not compatible with the
-   *      buffer
-   */
-  public <T extends TType> Constant<T> constant(DataType<T> type, Shape shape,
-      ByteDataBuffer data) {
-    return Constant.tensorOf(scope, type, shape, data);
-  }
-
-  /**
-   * This op consumes a lock created by `MutexLock`.
-   *  <p>
-   *  This op exists to consume a tensor created by `MutexLock` (other than
-   *  direct control dependencies).  It should be the only that consumes the tensor,
-   *  and will raise an error if it is not.  Its only purpose is to keep the
-   *  mutex lock tensor alive until it is consumed by this op.
-   *  <p>
-   *  <b>NOTE</b>: This operation must run on the same device as its input.  This may
-   *  be enforced via the `colocate_with` mechanism.
-   *
-   * @param mutexLock A tensor returned by `MutexLock`.
-   * @return a new instance of ConsumeMutexLock
-   */
-  public ConsumeMutexLock consumeMutexLock(Operand<?> mutexLock) {
-    return ConsumeMutexLock.create(scope, mutexLock);
-  }
-
-  /**
-   * Does nothing. Serves as a control trigger for scheduling.
-   *  <p>
-   *  Only useful as a placeholder for control edges.
-   *
-   * @return a new instance of ControlTrigger
-   */
-  public ControlTrigger controlTrigger() {
-    return ControlTrigger.create(scope);
-  }
-
-  /**
-   * Increments 'ref' until it reaches 'limit'.
-   *
-   * @param <T> data type for {@code output()} output
-   * @param ref Should be from a scalar `Variable` node.
-   * @param limit If incrementing ref would bring it above limit, instead generates an
-   *  'OutOfRange' error.
-   * @return a new instance of CountUpTo
-   */
-  public <T extends TNumber> CountUpTo<T> countUpTo(Operand<T> ref, Long limit) {
-    return CountUpTo.create(scope, ref, limit);
-  }
-
-  /**
-   * Makes a copy of `x`.
-   *
-   * @param <T> data type for {@code y()} output
-   * @param x The source tensor of type `T`.
-   * @return a new instance of DeepCopy
-   */
-  public <T extends TType> DeepCopy<T> deepCopy(Operand<T> x) {
-    return DeepCopy.create(scope, x);
-  }
-
-  /**
-   * Delete the tensor specified by its handle in the session.
-   *
-   * @param handle The handle for a tensor stored in the session state.
-   * @return a new instance of DeleteSessionTensor
-   */
-  public DeleteSessionTensor deleteSessionTensor(Operand<TString> handle) {
-    return DeleteSessionTensor.create(scope, handle);
-  }
-
-  /**
-   * Deletes the resource specified by the handle.
-   *  <p>
-   *  All subsequent operations using the resource will result in a NotFound
-   *  error status.
-   *
-   * @param resource handle to the resource to delete.
-   * @param options carries optional attributes values
-   * @return a new instance of DestroyResourceOp
-   */
-  public DestroyResourceOp destroyResourceOp(Operand<?> resource,
-      DestroyResourceOp.Options... options) {
-    return DestroyResourceOp.create(scope, resource, options);
-  }
-
-  /**
-   * Destroys the temporary variable and returns its final value.
-   *  <p>
-   *  Sets output to the value of the Tensor pointed to by 'ref', then destroys
-   *  the temporary variable called 'var_name'.
-   *  All other uses of 'ref' <i>must</i> have executed before this op.
-   *  This is typically achieved by chaining the ref through each assign op, or by
-   *  using control dependencies.
-   *  <p>
-   *  Outputs the final value of the tensor pointed to by 'ref'.
-   *
-   * @param <T> data type for {@code value()} output
-   * @param ref A reference to the temporary variable tensor.
-   * @param varName Name of the temporary variable, usually the name of the matching
-   *  'TemporaryVariable' op.
-   * @return a new instance of DestroyTemporaryVariable
-   */
-  public <T extends TType> DestroyTemporaryVariable<T> destroyTemporaryVariable(Operand<T> ref,
-      String varName) {
-    return DestroyTemporaryVariable.create(scope, ref, varName);
-  }
-
-  /**
-   * Partitions `data` into `num_partitions` tensors using indices from `partitions`.
-   *  <p>
-   *  For each index tuple `js` of size `partitions.ndim`, the slice `data[js, ...]`
-   *  becomes part of `outputs[partitions[js]]`.  The slices with `partitions[js] = i`
-   *  are placed in `outputs[i]` in lexicographic order of `js`, and the first
-   *  dimension of `outputs[i]` is the number of entries in `partitions` equal to `i`.
-   *  In detail,
-   *  <pre>{@code
-   *      outputs[i].shape = [sum(partitions == i)] + data.shape[partitions.ndim:]
-   *
-   *      outputs[i] = pack([data[js, ...] for js if partitions[js] == i])
-   *  }</pre>
-   *  `data.shape` must start with `partitions.shape`.
-   *  <p>
-   *  For example:
-   *  <pre>{@code
-   *      # Scalar partitions.
-   *      partitions = 1
-   *      num_partitions = 2
-   *      data = [10, 20]
-   *      outputs[0] = []  # Empty with shape [0, 2]
-   *      outputs[1] = [[10, 20]]
-   *
-   *      # Vector partitions.
-   *      partitions = [0, 0, 1, 1, 0]
-   *      num_partitions = 2
-   *      data = [10, 20, 30, 40, 50]
-   *      outputs[0] = [10, 20, 50]
-   *      outputs[1] = [30, 40]
-   *  }</pre>
-   *  See `dynamic_stitch` for an example on how to merge partitions back.
-   *  <p>
-   *  <div style="width:70%; margin:auto; margin-bottom:10px; margin-top:20px;">
-   *  <img style="width:100%" src="https://www.tensorflow.org/images/DynamicPartition.png" alt>
-   *  </div>
-   *
-   * @param <T> data type for {@code outputs()} output
-   * @param data
-   * @param partitions Any shape.  Indices in the range `[0, num_partitions)`.
-   * @param numPartitions The number of partitions to output.
-   * @return a new instance of DynamicPartition
-   */
-  public <T extends TType> DynamicPartition<T> dynamicPartition(Operand<T> data,
-      Operand<TInt32> partitions, Long numPartitions) {
-    return DynamicPartition.create(scope, data, partitions, numPartitions);
-  }
-
-  /**
-   * Interleave the values from the `data` tensors into a single tensor.
-   *  <p>
-   *  Builds a merged tensor such that
-   *  <pre>{@code
-   *      merged[indices[m][i, ..., j], ...] = data[m][i, ..., j, ...]
-   *  }</pre>
-   *  For example, if each `indices[m]` is scalar or vector, we have
-   *  <pre>{@code
-   *      # Scalar indices:
-   *      merged[indices[m], ...] = data[m][...]
-   *
-   *      # Vector indices:
-   *      merged[indices[m][i], ...] = data[m][i, ...]
-   *  }</pre>
-   *  Each `data[i].shape` must start with the corresponding `indices[i].shape`,
-   *  and the rest of `data[i].shape` must be constant w.r.t. `i`.  That is, we
-   *  must have `data[i].shape = indices[i].shape + constant`.  In terms of this
-   *  `constant`, the output shape is
-   *  <p>
-   *      merged.shape = [max(indices)] + constant
-   *  <p>
-   *  Values are merged in order, so if an index appears in both `indices[m][i]` and
-   *  `indices[n][j]` for `(m,i) < (n,j)` the slice `data[n][j]` will appear in the
-   *  merged result. If you do not need this guarantee, ParallelDynamicStitch might
-   *  perform better on some devices.
-   *  <p>
-   *  For example:
-   *  <pre>{@code
-   *      indices[0] = 6
-   *      indices[1] = [4, 1]
-   *      indices[2] = [[5, 2], [0, 3]]
-   *      data[0] = [61, 62]
-   *      data[1] = [[41, 42], [11, 12]]
-   *      data[2] = [[[51, 52], [21, 22]], [[1, 2], [31, 32]]]
-   *      merged = [[1, 2], [11, 12], [21, 22], [31, 32], [41, 42],
-   *                [51, 52], [61, 62]]
-   *  }</pre>
-   *  This method can be used to merge partitions created by `dynamic_partition`
-   *  as illustrated on the following example:
-   *  <pre>{@code
-   *      # Apply function (increments x_i) on elements for which a certain condition
-   *      # apply (x_i != -1 in this example).
-   *      x=tf.constant([0.1, -1., 5.2, 4.3, -1., 7.4])
-   *      condition_mask=tf.not_equal(x,tf.constant(-1.))
-   *      partitioned_data = tf.dynamic_partition(
-   *          x, tf.cast(condition_mask, tf.int32) , 2)
-   *      partitioned_data[1] = partitioned_data[1] + 1.0
-   *      condition_indices = tf.dynamic_partition(
-   *          tf.range(tf.shape(x)[0]), tf.cast(condition_mask, tf.int32) , 2)
-   *      x = tf.dynamic_stitch(condition_indices, partitioned_data)
-   *      # Here x=[1.1, -1., 6.2, 5.3, -1, 8.4], the -1. values remain
-   *      # unchanged.
-   *  }</pre>
-   *  <div style="width:70%; margin:auto; margin-bottom:10px; margin-top:20px;">
-   *  <img style="width:100%" src="https://www.tensorflow.org/images/DynamicStitch.png" alt>
-   *  </div>
-   *
-   * @param <T> data type for {@code merged()} output
-   * @param indices
-   * @param data
-   * @return a new instance of DynamicStitch
-   */
-  public <T extends TType> DynamicStitch<T> dynamicStitch(Iterable<Operand<TInt32>> indices,
-      Iterable<Operand<T>> data) {
-    return DynamicStitch.create(scope, indices, data);
-  }
-
-  /**
-   * Computes the (possibly normalized) Levenshtein Edit Distance.
-   *  <p>
-   *  The inputs are variable-length sequences provided by SparseTensors
-   *    (hypothesis_indices, hypothesis_values, hypothesis_shape)
-   *  and
-   *    (truth_indices, truth_values, truth_shape).
-   *  <p>
-   *  The inputs are:
-   *
-   * @param hypothesisIndices The indices of the hypothesis list SparseTensor.
-   *  This is an N x R int64 matrix.
-   * @param hypothesisValues The values of the hypothesis list SparseTensor.
-   *  This is an N-length vector.
-   * @param hypothesisShape The shape of the hypothesis list SparseTensor.
-   *  This is an R-length vector.
-   * @param truthIndices The indices of the truth list SparseTensor.
-   *  This is an M x R int64 matrix.
-   * @param truthValues The values of the truth list SparseTensor.
-   *  This is an M-length vector.
-   * @param truthShape truth indices, vector.
-   * @param options carries optional attributes values
-   * @return a new instance of EditDistance
-   */
-  public <T extends TType> EditDistance editDistance(Operand<TInt64> hypothesisIndices,
-      Operand<T> hypothesisValues, Operand<TInt64> hypothesisShape, Operand<TInt64> truthIndices,
-      Operand<T> truthValues, Operand<TInt64> truthShape, EditDistance.Options... options) {
-    return EditDistance.create(scope, hypothesisIndices, hypothesisValues, hypothesisShape, truthIndices, truthValues, truthShape, options);
-  }
-
-  /**
-   * Creates a tensor with the given shape.
-   *  <p>
-   *  This operation creates a tensor of `shape` and `dtype`.
-   *
-   * @param <T> data type for {@code output()} output
-   * @param shape 1-D. Represents the shape of the output tensor.
-   * @param dtype
-   * @param options carries optional attributes values
-   * @return a new instance of Empty
-   */
-  public <T extends TType> Empty<T> empty(Operand<TInt32> shape, DataType<T> dtype,
-      Empty.Options... options) {
-    return Empty.create(scope, shape, dtype, options);
-  }
-
-  /**
-   * Creates and returns an empty tensor list.
-   *  <p>
-   *  All list elements must be tensors of dtype element_dtype and shape compatible
-   *  with element_shape.
-   *  <p>
-   *  handle: an empty tensor list.
-   *  element_dtype: the type of elements in the list.
-   *  element_shape: a shape compatible with that of elements in the list.
-   *
-   * @param elementShape
-   * @param maxNumElements
-   * @param elementDtype
-   * @return a new instance of EmptyTensorList
-   */
-  public <T extends TNumber, U extends TType> EmptyTensorList emptyTensorList(
-      Operand<T> elementShape, Operand<TInt32> maxNumElements, DataType<U> elementDtype) {
-    return EmptyTensorList.create(scope, elementShape, maxNumElements, elementDtype);
-  }
-
-  /**
-   * Ensures that the tensor's shape matches the expected shape.
-   *  <p>
-   *  Raises an error if the input tensor's shape does not match the specified shape.
-   *  Returns the input tensor otherwise.
-   *
-   * @param <T> data type for {@code output()} output
-   * @param input A tensor, whose shape is to be validated.
-   * @param shape The expected (possibly partially specified) shape of the input tensor.
-   * @return a new instance of EnsureShape
-   */
-  public <T extends TType> EnsureShape<T> ensureShape(Operand<T> input, Shape shape) {
-    return EnsureShape.create(scope, input, shape);
-  }
-
-  /**
-   * Inserts a dimension of 1 into a tensor's shape.
-   *  <p>
-   *  Given a tensor `input`, this operation inserts a dimension of 1 at the
-   *  dimension index `axis` of `input`'s shape. The dimension index `axis` starts at
-   *  zero; if you specify a negative number for `axis` it is counted backward from
-   *  the end.
-   *  <p>
-   *  This operation is useful if you want to add a batch dimension to a single
-   *  element. For example, if you have a single image of shape `[height, width,
-   *  channels]`, you can make it a batch of 1 image with `expand_dims(image, 0)`,
-   *  which will make the shape `[1, height, width, channels]`.
-   *  <p>
-   *  Other examples:
-   *  <pre>{@code
-   *  # 't' is a tensor of shape [2]
-   *  shape(expand_dims(t, 0)) ==> [1, 2]
-   *  shape(expand_dims(t, 1)) ==> [2, 1]
-   *  shape(expand_dims(t, -1)) ==> [2, 1]
-   *
-   *  # 't2' is a tensor of shape [2, 3, 5]
-   *  shape(expand_dims(t2, 0)) ==> [1, 2, 3, 5]
-   *  shape(expand_dims(t2, 2)) ==> [2, 3, 1, 5]
-   *  shape(expand_dims(t2, 3)) ==> [2, 3, 5, 1]
-   *  }</pre>
-   *  This operation requires that:
-   *  <p>
-   *  `-1-input.dims() <= dim <= input.dims()`
-   *  <p>
-   *  This operation is related to `squeeze()`, which removes dimensions of
-   *  size 1.
-   *
-   * @param <T> data type for {@code output()} output
-   * @param input
-   * @param axis 0-D (scalar). Specifies the dimension index at which to
-   *  expand the shape of `input`. Must be in the range
-   *  `[-rank(input) - 1, rank(input)]`.
-   * @return a new instance of ExpandDims
-   */
-  public <T extends TType, U extends TNumber> ExpandDims<T> expandDims(Operand<T> input,
-      Operand<U> axis) {
-    return ExpandDims.create(scope, input, axis);
-  }
-
-  /**
-   * Extract `patches` from `input` and put them in the "depth" output dimension. 3D extension of `extract_image_patches`.
-   *
-   * @param <T> data type for {@code patches()} output
-   * @param input 5-D Tensor with shape `[batch, in_planes, in_rows, in_cols, depth]`.
-   * @param ksizes The size of the sliding window for each dimension of `input`.
-   * @param strides 1-D of length 5. How far the centers of two consecutive patches are in
-   *  `input`. Must be: `[1, stride_planes, stride_rows, stride_cols, 1]`.
-   * @param padding The type of padding algorithm to use.
-   *  <p>
-   *  We specify the size-related attributes as:
-   *  <pre>{@code
-   *        ksizes = [1, ksize_planes, ksize_rows, ksize_cols, 1]
-   *        strides = [1, stride_planes, strides_rows, strides_cols, 1]
-   *  }</pre>
-   * @return a new instance of ExtractVolumePatches
-   */
-  public <T extends TNumber> ExtractVolumePatches<T> extractVolumePatches(Operand<T> input,
-      List<Long> ksizes, List<Long> strides, String padding) {
-    return ExtractVolumePatches.create(scope, input, ksizes, strides, padding);
-  }
-
-  /**
-   * Creates a tensor filled with a scalar value.
-   *  <p>
-   *  This operation creates a tensor of shape `dims` and fills it with `value`.
-   *  <p>
-   *  For example:
-   *  <pre>{@code
-   *  # Output tensor has shape [2, 3].
-   *  fill([2, 3], 9) ==> [[9, 9, 9]
-   *                       [9, 9, 9]]
-   *  }</pre>
-   *  `tf.fill` differs from `tf.constant` in a few ways:
-   *  <ul>
-   *  <li>
-   *  `tf.fill` only supports scalar contents, whereas `tf.constant` supports
-   *      Tensor values.
-   *  </li>
-   *  <li>
-   *  `tf.fill` creates an Op in the computation graph that constructs the actual
-   *      Tensor value at runtime. This is in contrast to `tf.constant` which embeds
-   *      the entire Tensor into the graph with a `Const` node.
-   *  </li>
-   *  <li>
-   *  Because `tf.fill` evaluates at graph runtime, it supports dynamic shapes
-   *      based on other runtime Tensors, unlike `tf.constant`.
-   *
-   * @param <U> data type for {@code output()} output
-   * @param dims 1-D. Represents the shape of the output tensor.
-   * @param value 0-D (scalar). Value to fill the returned tensor.
-   *  <p>
-   * @compatibility(numpy) Equivalent to np.full
-   * @end_compatibility
-   * @return a new instance of Fill
-   */
-  public <U extends TType, T extends TNumber> Fill<U> fill(Operand<T> dims, Operand<U> value) {
-    return Fill.create(scope, dims, value);
-  }
-
-  /**
-   * Generates fingerprint values.
-   *  <p>
-   *  Generates fingerprint values of `data`.
-   *  <p>
-   *  Fingerprint op considers the first dimension of `data` as the batch dimension,
-   *  and `output[i]` contains the fingerprint value generated from contents in
-   *  `data[i, ...]` for all `i`.
-   *  <p>
-   *  Fingerprint op writes fingerprint values as byte arrays. For example, the
-   *  default method `farmhash64` generates a 64-bit fingerprint value at a time.
-   *  This 8-byte value is written out as an `uint8` array of size 8, in little-endian
-   *  order.
-   *  <p>
-   *  For example, suppose that `data` has data type `DT_INT32` and shape (2, 3, 4),
-   *  and that the fingerprint method is `farmhash64`. In this case, the output shape
-   *  is (2, 8), where 2 is the batch dimension size of `data`, and 8 is the size of
-   *  each fingerprint value in bytes. `output[0, :]` is generated from 12 integers in
-   *  `data[0, :, :]` and similarly `output[1, :]` is generated from other 12 integers
-   *  in `data[1, :, :]`.
-   *  <p>
-   *  Note that this op fingerprints the raw underlying buffer, and it does not
-   *  fingerprint Tensor's metadata such as data type and/or shape. For example, the
-   *  fingerprint values are invariant under reshapes and bitcasts as long as the
-   *  batch dimension remain the same:
-   *  <pre>{@code
-   *  Fingerprint(data) == Fingerprint(Reshape(data, ...))
-   *  Fingerprint(data) == Fingerprint(Bitcast(data, ...))
-   *  }</pre>
-   *  For string data, one should expect `Fingerprint(data) !=
-   *  Fingerprint(ReduceJoin(data))` in general.
-   *
-   * @param data Must have rank 1 or higher.
-   * @param method Fingerprint method used by this op. Currently available method is
-   *  `farmhash::fingerprint64`.
-   * @return a new instance of Fingerprint
-   */
-  public <T extends TType> Fingerprint fingerprint(Operand<T> data, Operand<TString> method) {
-    return Fingerprint.create(scope, data, method);
-  }
-
-  /**
-   * Gather slices from `params` axis `axis` according to `indices`.
-   *  <p>
-   *  `indices` must be an integer tensor of any dimension (usually 0-D or 1-D).
-   *  Produces an output tensor with shape `params.shape[:axis] + indices.shape +
-   *  params.shape[axis + 1:]` where:
-   *  <pre>{@code
-   *      # Scalar indices (output is rank(params) - 1).
-   *      output[a_0, ..., a_n, b_0, ..., b_n] =
-   *        params[a_0, ..., a_n, indices, b_0, ..., b_n]
-   *
-   *      # Vector indices (output is rank(params)).
-   *      output[a_0, ..., a_n, i, b_0, ..., b_n] =
-   *        params[a_0, ..., a_n, indices[i], b_0, ..., b_n]
-   *
-   *      # Higher rank indices (output is rank(params) + rank(indices) - 1).
-   *      output[a_0, ..., a_n, i, ..., j, b_0, ... b_n] =
-   *        params[a_0, ..., a_n, indices[i, ..., j], b_0, ..., b_n]
-   *  }</pre>
-   *  <div style="width:70%; margin:auto; margin-bottom:10px; margin-top:20px;">
-   *  <img style="width:100%" src="https://www.tensorflow.org/images/Gather.png" alt>
-   *  </div>
-   *  <p>
-   *  Note that on CPU, if an out of bound index is found, an error is returned.
-   *  On GPU, if an out of bound index is found, a 0 is stored in the
-   *  corresponding output value.
-   *  <p>
-   *  See also `tf.batch_gather` and `tf.gather_nd`.
-   *
-   * @param <T> data type for {@code output()} output
-   * @param params The tensor from which to gather values. Must be at least rank
-   *  `axis + 1`.
-   * @param indices Index tensor. Must be in range `[0, params.shape[axis])`.
-   * @param axis The axis in `params` to gather `indices` from. Defaults to the first
-   *  dimension. Supports negative indexes.
-   * @param options carries optional attributes values
-   * @return a new instance of Gather
-   */
-  public <T extends TType, U extends TNumber, V extends TNumber> Gather<T> gather(Operand<T> params,
-      Operand<U> indices, Operand<V> axis, Gather.Options... options) {
-    return Gather.create(scope, params, indices, axis, options);
-  }
-
-  /**
-   * Gather slices from `params` into a Tensor with shape specified by `indices`.
-   *  <p>
-   *  `indices` is a K-dimensional integer tensor, best thought of as a
-   *  (K-1)-dimensional tensor of indices into `params`, where each element defines a
-   *  slice of `params`:
-   *  <p>
-   *      output[\\(i_0, ..., i_{K-2}\\)] = params[indices[\\(i_0, ..., i_{K-2}\\)]]
-   *  <p>
-   *  Whereas in `tf.gather` `indices` defines slices into the `axis`
-   *  dimension of `params`, in `tf.gather_nd`, `indices` defines slices into the
-   *  first `N` dimensions of `params`, where `N = indices.shape[-1]`.
-   *  <p>
-   *  The last dimension of `indices` can be at most the rank of
-   *  `params`:
-   *  <p>
-   *      indices.shape[-1] <= params.rank
-   *  <p>
-   *  The last dimension of `indices` corresponds to elements
-   *  (if `indices.shape[-1] == params.rank`) or slices
-   *  (if `indices.shape[-1] < params.rank`) along dimension `indices.shape[-1]`
-   *  of `params`.  The output tensor has shape
-   *  <p>
-   *      indices.shape[:-1] + params.shape[indices.shape[-1]:]
-   *  <p>
-   *  Note that on CPU, if an out of bound index is found, an error is returned.
-   *  On GPU, if an out of bound index is found, a 0 is stored in the
-   *  corresponding output value.
-   *  <p>
-   *  Some examples below.
-   *  <p>
-   *  Simple indexing into a matrix:
-   *  <pre>{@code
-   *      indices = [[0, 0], [1, 1]]
-   *      params = [['a', 'b'], ['c', 'd']]
-   *      output = ['a', 'd']
-   *  }</pre>
-   *  Slice indexing into a matrix:
-   *  <pre>{@code
-   *      indices = [[1], [0]]
-   *      params = [['a', 'b'], ['c', 'd']]
-   *      output = [['c', 'd'], ['a', 'b']]
-   *  }</pre>
-   *  Indexing into a 3-tensor:
-   *  <pre>{@code
-   *      indices = [[1]]
-   *      params = [[['a0', 'b0'], ['c0', 'd0']],
-   *                [['a1', 'b1'], ['c1', 'd1']]]
-   *      output = [[['a1', 'b1'], ['c1', 'd1']]]
-   *
-   *
-   *      indices = [[0, 1], [1, 0]]
-   *      params = [[['a0', 'b0'], ['c0', 'd0']],
-   *                [['a1', 'b1'], ['c1', 'd1']]]
-   *      output = [['c0', 'd0'], ['a1', 'b1']]
-   *
-   *
-   *      indices = [[0, 0, 1], [1, 0, 1]]
-   *      params = [[['a0', 'b0'], ['c0', 'd0']],
-   *                [['a1', 'b1'], ['c1', 'd1']]]
-   *      output = ['b0', 'b1']
-   *  }</pre>
-   *  Batched indexing into a matrix:
-   *  <pre>{@code
-   *      indices = [[[0, 0]], [[0, 1]]]
-   *      params = [['a', 'b'], ['c', 'd']]
-   *      output = [['a'], ['b']]
-   *  }</pre>
-   *  Batched slice indexing into a matrix:
-   *  <pre>{@code
-   *      indices = [[[1]], [[0]]]
-   *      params = [['a', 'b'], ['c', 'd']]
-   *      output = [[['c', 'd']], [['a', 'b']]]
-   *  }</pre>
-   *  Batched indexing into a 3-tensor:
-   *  <pre>{@code
-   *      indices = [[[1]], [[0]]]
-   *      params = [[['a0', 'b0'], ['c0', 'd0']],
-   *                [['a1', 'b1'], ['c1', 'd1']]]
-   *      output = [[[['a1', 'b1'], ['c1', 'd1']]],
-   *                [[['a0', 'b0'], ['c0', 'd0']]]]
-   *
-   *      indices = [[[0, 1], [1, 0]], [[0, 0], [1, 1]]]
-   *      params = [[['a0', 'b0'], ['c0', 'd0']],
-   *                [['a1', 'b1'], ['c1', 'd1']]]
-   *      output = [[['c0', 'd0'], ['a1', 'b1']],
-   *                [['a0', 'b0'], ['c1', 'd1']]]
-   *
-   *
-   *      indices = [[[0, 0, 1], [1, 0, 1]], [[0, 1, 1], [1, 1, 0]]]
-   *      params = [[['a0', 'b0'], ['c0', 'd0']],
-   *                [['a1', 'b1'], ['c1', 'd1']]]
-   *      output = [['b0', 'b1'], ['d0', 'c1']]
-   *  }</pre>
-   *  See also `tf.gather` and `tf.batch_gather`.
-   *
-   * @param <T> data type for {@code output()} output
-   * @param params The tensor from which to gather values.
-   * @param indices Index tensor.
-   * @return a new instance of GatherNd
-   */
-  public <T extends TType, U extends TNumber> GatherNd<T> gatherNd(Operand<T> params,
-      Operand<U> indices) {
-    return GatherNd.create(scope, params, indices);
-  }
-
-  /**
-   * Store the input tensor in the state of the current session.
-   *
-   * @param value The tensor to be stored.
-   * @return a new instance of GetSessionHandle
-   */
-  public <T extends TType> GetSessionHandle getSessionHandle(Operand<T> value) {
-    return GetSessionHandle.create(scope, value);
-  }
-
-  /**
-   * Get the value of the tensor specified by its handle.
-   *
-   * @param <T> data type for {@code value()} output
-   * @param handle The handle for a tensor stored in the session state.
-   * @param dtype The type of the output value.
-   * @return a new instance of GetSessionTensor
-   */
-  public <T extends TType> GetSessionTensor<T> getSessionTensor(Operand<TString> handle,
-      DataType<T> dtype) {
-    return GetSessionTensor.create(scope, handle, dtype);
-  }
-
-  /**
-   * Adds gradients computation ops to the graph according to scope.
-   *
-   * @param scope current graph scope
-   * @param y outputs of the function to derive
-   * @param x inputs of the function for which partial derivatives are computed
-   * @param options carries optional attributes values
-   * @return a new instance of {@code Gradients}
-   * @throws IllegalArgumentException if execution environment is not a graph
-   */
-  public Gradients gradients(Iterable<? extends Operand<?>> y, Iterable<? extends Operand<?>> x,
-      Gradients.Options... options) {
-    return Gradients.create(scope, y, x, options);
-  }
-
-  /**
-   * Adds operations to compute the partial derivatives of sum of {@code y}s w.r.t {@code x}s,
-   *  i.e., {@code d(y_1 + y_2 + ...)/dx_1, d(y_1 + y_2 + ...)/dx_2...}
-   *  <p> 
-   *  If {@code Options.dx()} values are set, they are as the initial symbolic partial derivatives of some loss 
-   *  function {@code L} w.r.t. {@code y}. {@code Options.dx()} must have the size of {@code y}.
-   *  <p>
-   *  If {@code Options.dx()} is not set, the implementation will use dx of {@code OnesLike} for all
-   *  shapes in {@code y}.
-   *  <p>
-   *  The partial derivatives are returned in output {@code dy}, with the size of {@code x}.
-   *  <p>
-   *  Example of usage:
-   *  <pre>{@code
-   *  Gradients gradients = tf.gradients(loss, Arrays.asList(w, b));
-   *  Constant<TFloat32> alpha = tf.constant(1.0f);
-   *  tf.train.applyGradientDescent(w, alpha, gradients.<Float>dy(0));
-   *  tf.train.applyGradientDescent(b, alpha, gradients.<Float>dy(1));
-   *  }</pre>
-   *
-   * @param y output of the function to derive
-   * @param x inputs of the function for which partial derivatives are computed
-   * @param options carries optional attributes values
-   * @return a new instance of {@code Gradients}
-   * @throws IllegalArgumentException if execution environment is not a graph
-   */
-  public Gradients gradients(Operand<?> y, Iterable<? extends Operand<?>> x,
-      Gradients.Options... options) {
-    return Gradients.create(scope, y, x, options);
-  }
-
-  /**
-   * Gives a guarantee to the TF runtime that the input tensor is a constant.
-   *  <p>
-   *  The runtime is then free to make optimizations based on this.
-   *  <p>
-   *  Only accepts value typed tensors as inputs and rejects resource variable handles
-   *  as input.
-   *  <p>
-   *  Returns the input tensor without modification.
-   *
-   * @param <T> data type for {@code output()} output
-   * @param input
-   * @return a new instance of GuaranteeConst
-   */
-  public <T extends TType> GuaranteeConst<T> guaranteeConst(Operand<T> input) {
-    return GuaranteeConst.create(scope, input);
-  }
-
-  /**
-   * Creates a non-initialized hash table.
-   *  <p>
-   *  This op creates a hash table, specifying the type of its keys and values.
-   *  Before using the table you will have to initialize it.  After initialization the
-   *  table will be immutable.
-   *
-   * @param keyDtype Type of the table keys.
-   * @param valueDtype Type of the table values.
-   * @param options carries optional attributes values
-   * @return a new instance of HashTable
-   */
-  public <T extends TType, U extends TType> HashTable hashTable(DataType<T> keyDtype,
-      DataType<U> valueDtype, HashTable.Options... options) {
-    return HashTable.create(scope, keyDtype, valueDtype, options);
-  }
-
-  /**
-   * Return histogram of values.
-   *  <p>
-   *  Given the tensor `values`, this operation returns a rank 1 histogram counting
-   *  the number of entries in `values` that fall into every bin.  The bins are
-   *  equal width and determined by the arguments `value_range` and `nbins`.
-   *  <pre>{@code
-   *  # Bins will be:  (-inf, 1), [1, 2), [2, 3), [3, 4), [4, inf)
-   *  nbins = 5
-   *  value_range = [0.0, 5.0]
-   *  new_values = [-1.0, 0.0, 1.5, 2.0, 5.0, 15]
-   *
-   *  with tf.get_default_session() as sess:
-   *    hist = tf.histogram_fixed_width(new_values, value_range, nbins=5)
-   *    variables.global_variables_initializer().run()
-   *    sess.run(hist) => [2, 1, 1, 0, 2]
-   *  }</pre>
-   *
-   * @param <U> data type for {@code out()} output
-   * @param values Numeric `Tensor`.
-   * @param valueRange Shape [2] `Tensor` of same `dtype` as `values`.
-   *  values <= value_range[0] will be mapped to hist[0],
-   *  values >= value_range[1] will be mapped to hist[-1].
-   * @param nbins Scalar `int32 Tensor`.  Number of histogram bins.
-   * @return a new instance of HistogramFixedWidth
-   */
-  public <T extends TNumber> HistogramFixedWidth<TInt32> histogramFixedWidth(Operand<T> values,
-      Operand<T> valueRange, Operand<TInt32> nbins) {
-    return HistogramFixedWidth.create(scope, values, valueRange, nbins);
-  }
-
-  /**
-   * Return histogram of values.
-   *  <p>
-   *  Given the tensor `values`, this operation returns a rank 1 histogram counting
-   *  the number of entries in `values` that fall into every bin.  The bins are
-   *  equal width and determined by the arguments `value_range` and `nbins`.
-   *  <pre>{@code
-   *  # Bins will be:  (-inf, 1), [1, 2), [2, 3), [3, 4), [4, inf)
-   *  nbins = 5
-   *  value_range = [0.0, 5.0]
-   *  new_values = [-1.0, 0.0, 1.5, 2.0, 5.0, 15]
-   *
-   *  with tf.get_default_session() as sess:
-   *    hist = tf.histogram_fixed_width(new_values, value_range, nbins=5)
-   *    variables.global_variables_initializer().run()
-   *    sess.run(hist) => [2, 1, 1, 0, 2]
-   *  }</pre>
-   *
-   * @param <U> data type for {@code out()} output
-   * @param values Numeric `Tensor`.
-   * @param valueRange Shape [2] `Tensor` of same `dtype` as `values`.
-   *  values <= value_range[0] will be mapped to hist[0],
-   *  values >= value_range[1] will be mapped to hist[-1].
-   * @param nbins Scalar `int32 Tensor`.  Number of histogram bins.
-   * @param dtype
-   * @return a new instance of HistogramFixedWidth
-   */
-  public <U extends TNumber, T extends TNumber> HistogramFixedWidth<U> histogramFixedWidth(
-      Operand<T> values, Operand<T> valueRange, Operand<TInt32> nbins, DataType<U> dtype) {
-    return HistogramFixedWidth.create(scope, values, valueRange, nbins, dtype);
-  }
-
-  /**
-   * Return a tensor with the same shape and contents as the input tensor or value.
-   *
-   * @param <T> data type for {@code output()} output
-   * @param input
-   * @return a new instance of Identity
-   */
-  public <T extends TType> Identity<T> identity(Operand<T> input) {
-    return Identity.create(scope, input);
-  }
-
-  /**
-   * Returns a list of tensors with the same shapes and contents as the input
-   *  <p>
-   *  tensors.
-   *  <p>
-   *  This op can be used to override the gradient for complicated functions. For
-   *  example, suppose y = f(x) and we wish to apply a custom function g for backprop
-   *  such that dx = g(dy). In Python,
-   *  <pre>{@code
-   *  with tf.get_default_graph().gradient_override_map(
-   *      {'IdentityN': 'OverrideGradientWithG'}):
-   *    y, _ = identity_n([f(x), x])
-   *
-   * @tf.RegisterGradient('OverrideGradientWithG') def ApplyG(op, dy, _):
-   *    return [None, g(dy)]  # Do not backprop to f(x).
-   *  }</pre>
-   * @param input
-   * @return a new instance of IdentityN
-   */
-  public IdentityN identityN(Iterable<Operand<?>> input) {
-    return IdentityN.create(scope, input);
-  }
-
-  /**
-   * Returns immutable tensor from memory region.
-   *  <p>
-   *  The current implementation memmaps the tensor from a file.
-   *
-   * @param <T> data type for {@code tensor()} output
-   * @param dtype Type of the returned tensor.
-   * @param shape Shape of the returned tensor.
-   * @param memoryRegionName Name of readonly memory region used by the tensor, see
-   *  NewReadOnlyMemoryRegionFromFile in tensorflow::Env.
-   * @return a new instance of ImmutableConst
-   */
-  public <T extends TType> ImmutableConst<T> immutableConst(DataType<T> dtype, Shape shape,
-      String memoryRegionName) {
-    return ImmutableConst.create(scope, dtype, shape, memoryRegionName);
-  }
-
-  /**
-   * Factory method to create an operation executing all initializers of a graph.
-   *
-   *  <p>All initializers added to a graph via
-   *  {@link org.tensorflow.op.core.Init#add(Scope, Op) tf.initAdd} are grouped together as a single
-   *  unit of computation in the graph. This operation must then be added to any graph using one or
-   *  more {@link Variable variables} and executed once before running the graph so the variable
-   *  states are initialized properly.</p>
-   *
-   *  <p>When the graph is built by the same process that is running the session, the initializers
-   *  can be invoked by executing this single endpoint. For example:</p>
-   *  <pre>{@code
-   *  try (Graph g = new Graph()) {
-   *    Variable<TInt32> x = tf.variable(tf.constant(10));  // initAdd is called implicitly
-   *    Variable<TInt32> y = tf.variable(tf.constant(20));  // idem
-   *    Add<TInt32> z = tf.math.add(x, y);
-   *
-   *    try (Session s = new Session(g)) {
-   *      s.run(tf.init());  // initialize all variables
-   *
-   *      try (Tensor<TInt32> t = s.runner().fetch(z).run().get(0).expect(TInt32.DTYPE)) {
-   *        assertEquals(30, t.data().getInt());
-   *      }
-   *    }
-   *  }
-   *  }</pre>
-   *
-   *  <p>When the graph is built by a separate process, the initializers can be invoked by running
-   *  the init op by its name, which defaults to {@link org.tensorflow.op.core.Init#DEFAULT_NAME}.
-   *  For example:</p>
-   *  <pre>{@code
-   *  // Building the model
-   *  try (Graph g = new Graph()) {
-   *    Variable<TInt32> x = tf.variable(tf.constant(10));  // initAdd is called implicitly
-   *    Variable<TInt32> y = tf.variable(tf.constant(20));  // idem
-   *    Add<TInt32> z = tf.withName("z").math.add(x, y);
-   *
-   *    tf.init();  // add variables initializers to the graph, as Init.DEFAULT_NAME
-   *    // ...exporting graph as a saved model...
-   *  }
-   *
-   *  ...
-   *
-   *  // Running the model
-   *  try (SavedModelBundle model = SavedModelBundle.load("/path/to/model", "train")) {
-   *    model.session().run(Init.DEFAULT_NAME);
-   *
-   *    try (Tensor<TInt32> t = s.runner().fetch("z").run().get(0).expect(TInt32.DTYPE)) {
-   *      assertEquals(30, t.data().getInt());
-   *    }
-   *  }
-   *  }</pre>
-   *
-   * @param scope current scope
-   * @return an op grouping all initializers added to the graph
-   * @throws IllegalArgumentException if the execution environment in scope is not a graph
-   */
-  public Init init() {
-    return Init.create(scope);
-  }
-
-  /**
-   * Register an op as an initializer of the graph.
-   *
-   *  <p>Registered initializers are then grouped as a single unit of computation by adding
-   *  and executing an {@link org.tensorflow.op.core.Init#create(Scope) init} operation from a graph
-   *  session.
-   *
-   * @param scope
-   * @param initializer
-   * @throws IllegalArgumentException if the execution environment in scope is not a graph
-   * @see org.tensorflow.op.core.Init#create(Scope) init
-   */
-  public void initAdd(Op initializer) {
-    Init.add(scope, initializer);
-  }
-
-  /**
-   * Table initializer that takes two tensors for keys and values respectively.
-   *
-   * @param tableHandle Handle to a table which will be initialized.
-   * @param keys Keys of type Tkey.
-   * @param values Values of type Tval.
-   * @return a new instance of InitializeTable
-   */
-  public <T extends TType, U extends TType> InitializeTable initializeTable(Operand<?> tableHandle,
-      Operand<T> keys, Operand<U> values) {
-    return InitializeTable.create(scope, tableHandle, keys, values);
-  }
-
-  /**
-   * Initializes a table from a text file.
-   *  <p>
-   *  It inserts one key-value pair into the table for each line of the file.
-   *  The key and value is extracted from the whole line content, elements from the
-   *  split line based on `delimiter` or the line number (starting from zero).
-   *  Where to extract the key and value from a line is specified by `key_index` and
-   *  `value_index`.
-   *  <p>
-   *  - A value of -1 means use the line number(starting from zero), expects `int64`.
-   *  - A value of -2 means use the whole line content, expects `string`.
-   *  - A value >= 0 means use the index (starting at zero) of the split line based
-   *    on `delimiter`.
-   *
-   * @param tableHandle Handle to a table which will be initialized.
-   * @param filename Filename of a vocabulary text file.
-   * @param keyIndex Column index in a line to get the table `key` values from.
-   * @param valueIndex Column index that represents information of a line to get the table
-   *  `value` values from.
-   * @param options carries optional attributes values
-   * @return a new instance of InitializeTableFromTextFile
-   */
-  public InitializeTableFromTextFile initializeTableFromTextFile(Operand<?> tableHandle,
-      Operand<TString> filename, Long keyIndex, Long valueIndex,
-      InitializeTableFromTextFile.Options... options) {
-    return InitializeTableFromTextFile.create(scope, tableHandle, filename, keyIndex, valueIndex, options);
-  }
-
-  /**
-   *     Adds v into specified rows of x.
-   *  <p>
-   *      Computes y = x; y[i, :] += v; return y.
-   *
-   * @param <T> data type for {@code y()} output
-   * @param x A `Tensor` of type T.
-   * @param i A vector. Indices into the left-most dimension of `x`.
-   * @param v A `Tensor` of type T. Same dimension sizes as x except the first dimension, which must be the same as i's size.
-   * @return a new instance of InplaceAdd
-   */
-  public <T extends TType> InplaceAdd<T> inplaceAdd(Operand<T> x, Operand<TInt32> i, Operand<T> v) {
-    return InplaceAdd.create(scope, x, i, v);
-  }
-
-  /**
-   *     Subtracts `v` into specified rows of `x`.
-   *  <p>
-   *      Computes y = x; y[i, :] -= v; return y.
-   *
-   * @param <T> data type for {@code y()} output
-   * @param x A `Tensor` of type T.
-   * @param i A vector. Indices into the left-most dimension of `x`.
-   * @param v A `Tensor` of type T. Same dimension sizes as x except the first dimension, which must be the same as i's size.
-   * @return a new instance of InplaceSub
-   */
-  public <T extends TType> InplaceSub<T> inplaceSub(Operand<T> x, Operand<TInt32> i, Operand<T> v) {
-    return InplaceSub.create(scope, x, i, v);
-  }
-
-  /**
-   *     Updates specified rows with values in `v`.
-   *  <p>
-   *      Computes `x[i, :] = v; return x`.
-   *
-   * @param <T> data type for {@code y()} output
-   * @param x A tensor of type `T`.
-   * @param i A vector. Indices into the left-most dimension of `x`.
-   * @param v A `Tensor` of type T. Same dimension sizes as x except the first dimension, which must be the same as i's size.
-   * @return a new instance of InplaceUpdate
-   */
-  public <T extends TType> InplaceUpdate<T> inplaceUpdate(Operand<T> x, Operand<TInt32> i,
-      Operand<T> v) {
-    return InplaceUpdate.create(scope, x, i, v);
-  }
-
-  /**
-   * Checks whether a tensor has been initialized.
-   *  <p>
-   *  Outputs boolean scalar indicating whether the tensor has been initialized.
-   *
-   * @param ref Should be from a `Variable` node. May be uninitialized.
-   * @return a new instance of IsVariableInitialized
-   */
-  public <T extends TType> IsVariableInitialized isVariableInitialized(Operand<T> ref) {
-    return IsVariableInitialized.create(scope, ref);
-  }
-
-  /**
-   * Generates values in an interval.
-   *  <p>
-   *  A sequence of `num` evenly-spaced values are generated beginning at `start`.
-   *  If `num > 1`, the values in the sequence increase by `stop - start / num - 1`,
-   *  so that the last one is exactly `stop`.
-   *  <p>
-   *  For example:
-   *  <pre>{@code
-   *  tf.linspace(10.0, 12.0, 3, name="linspace") => [ 10.0  11.0  12.0]
-   *  }</pre>
-   *
-   * @param <T> data type for {@code output()} output
-   * @param start 0-D tensor. First entry in the range.
-   * @param stop 0-D tensor. Last entry in the range.
-   * @param num 0-D tensor. Number of values to generate.
-   * @return a new instance of LinSpace
-   */
-  public <T extends TNumber, U extends TNumber> LinSpace<T> linSpace(Operand<T> start,
-      Operand<T> stop, Operand<U> num) {
-    return LinSpace.create(scope, start, stop, num);
-  }
-
-  /**
-   * Outputs all keys and values in the table.
-   *
-   * @param <T> data type for {@code keys()} output
-   * @param <U> data type for {@code values()} output
-   * @param tableHandle Handle to the table.
-   * @param Tkeys
-   * @param Tvalues
-   * @return a new instance of LookupTableExport
-   */
-  public <T extends TType, U extends TType> LookupTableExport<T, U> lookupTableExport(
-      Operand<?> tableHandle, DataType<T> Tkeys, DataType<U> Tvalues) {
-    return LookupTableExport.create(scope, tableHandle, Tkeys, Tvalues);
-  }
-
-  /**
-   * Looks up keys in a table, outputs the corresponding values.
-   *  <p>
-   *  The tensor `keys` must of the same type as the keys of the table.
-   *  The output `values` is of the type of the table values.
-   *  <p>
-   *  The scalar `default_value` is the value output for keys not present in the
-   *  table. It must also be of the same type as the table values.
-   *
-   * @param <U> data type for {@code values()} output
-   * @param tableHandle Handle to the table.
-   * @param keys Any shape.  Keys to look up.
-   * @param defaultValue
-   * @return a new instance of LookupTableFind
-   */
-  public <U extends TType, T extends TType> LookupTableFind<U> lookupTableFind(
-      Operand<?> tableHandle, Operand<T> keys, Operand<U> defaultValue) {
-    return LookupTableFind.create(scope, tableHandle, keys, defaultValue);
-  }
-
-  /**
-   * Replaces the contents of the table with the specified keys and values.
-   *  <p>
-   *  The tensor `keys` must be of the same type as the keys of the table.
-   *  The tensor `values` must be of the type of the table values.
-   *
-   * @param tableHandle Handle to the table.
-   * @param keys Any shape.  Keys to look up.
-   * @param values Values to associate with keys.
-   * @return a new instance of LookupTableImport
-   */
-  public <T extends TType, U extends TType> LookupTableImport lookupTableImport(
-      Operand<?> tableHandle, Operand<T> keys, Operand<U> values) {
-    return LookupTableImport.create(scope, tableHandle, keys, values);
-  }
-
-  /**
-   * Updates the table to associates keys with values.
-   *  <p>
-   *  The tensor `keys` must be of the same type as the keys of the table.
-   *  The tensor `values` must be of the type of the table values.
-   *
-   * @param tableHandle Handle to the table.
-   * @param keys Any shape.  Keys to look up.
-   * @param values Values to associate with keys.
-   * @return a new instance of LookupTableInsert
-   */
-  public <T extends TType, U extends TType> LookupTableInsert lookupTableInsert(
-      Operand<?> tableHandle, Operand<T> keys, Operand<U> values) {
-    return LookupTableInsert.create(scope, tableHandle, keys, values);
-  }
-
-  /**
-   * Computes the number of elements in the given table.
-   *
-   * @param tableHandle Handle to the table.
-   * @return a new instance of LookupTableSize
-   */
-  public LookupTableSize lookupTableSize(Operand<?> tableHandle) {
-    return LookupTableSize.create(scope, tableHandle);
-  }
-
-  /**
-   * Forwards the input to the output.
-   *  <p>
-   *  This operator represents the loop termination condition used by the
-   *  "pivot" switches of a loop.
-   *
-   * @param input A boolean scalar, representing the branch predicate of the Switch op.
-   * @return a new instance of LoopCond
-   */
-  public LoopCond loopCond(Operand<TBool> input) {
-    return LoopCond.create(scope, input);
-  }
-
-  /**
-   * Op removes all elements in the underlying container.
-   *
-   * @param dtypes
-   * @param options carries optional attributes values
-   * @return a new instance of MapClear
-   */
-  public MapClear mapClear(List<DataType<?>> dtypes, MapClear.Options... options) {
-    return MapClear.create(scope, dtypes, options);
-  }
-
-  /**
-   * Op returns the number of incomplete elements in the underlying container.
-   *
-   * @param dtypes
-   * @param options carries optional attributes values
-   * @return a new instance of MapIncompleteSize
-   */
-  public MapIncompleteSize mapIncompleteSize(List<DataType<?>> dtypes,
-      MapIncompleteSize.Options... options) {
-    return MapIncompleteSize.create(scope, dtypes, options);
-  }
-
-  /**
-   * Op peeks at the values at the specified key.  If the
-   *  <p>
-   *  underlying container does not contain this key
-   *  this op will block until it does.
-   *
-   * @param key
-   * @param indices
-   * @param dtypes
-   * @param options carries optional attributes values
-   * @return a new instance of MapPeek
-   */
-  public MapPeek mapPeek(Operand<TInt64> key, Operand<TInt32> indices, List<DataType<?>> dtypes,
-      MapPeek.Options... options) {
-    return MapPeek.create(scope, key, indices, dtypes, options);
-  }
-
-  /**
-   * Op returns the number of elements in the underlying container.
-   *
-   * @param dtypes
-   * @param options carries optional attributes values
-   * @return a new instance of MapSize
-   */
-  public MapSize mapSize(List<DataType<?>> dtypes, MapSize.Options... options) {
-    return MapSize.create(scope, dtypes, options);
-  }
-
-  /**
-   * Stage (key, values) in the underlying container which behaves like a hashtable.
-   *
-   * @param key int64
-   * @param indices
-   * @param values a list of tensors
-   *  dtypes A list of data types that inserted values should adhere to.
-   * @param dtypes
-   * @param options carries optional attributes values
-   * @return a new instance of MapStage
-   */
-  public MapStage mapStage(Operand<TInt64> key, Operand<TInt32> indices,
-      Iterable<Operand<?>> values, List<DataType<?>> dtypes, MapStage.Options... options) {
-    return MapStage.create(scope, key, indices, values, dtypes, options);
-  }
-
-  /**
-   * Op removes and returns the values associated with the key
-   *  <p>
-   *  from the underlying container.   If the underlying container
-   *  does not contain this key, the op will block until it does.
-   *
-   * @param key
-   * @param indices
-   * @param dtypes
-   * @param options carries optional attributes values
-   * @return a new instance of MapUnstage
-   */
-  public MapUnstage mapUnstage(Operand<TInt64> key, Operand<TInt32> indices,
-      List<DataType<?>> dtypes, MapUnstage.Options... options) {
-    return MapUnstage.create(scope, key, indices, dtypes, options);
-  }
-
-  /**
-   * Op removes and returns a random (key, value)
-   *  <p>
-   *  from the underlying container.   If the underlying container
-   *  does not contain elements, the op will block until it does.
-   *
-   * @param indices
-   * @param dtypes
-   * @param options carries optional attributes values
-   * @return a new instance of MapUnstageNoKey
-   */
-  public MapUnstageNoKey mapUnstageNoKey(Operand<TInt32> indices, List<DataType<?>> dtypes,
-      MapUnstageNoKey.Options... options) {
-    return MapUnstageNoKey.create(scope, indices, dtypes, options);
-  }
-
-  /**
-   * Computes the maximum of elements across dimensions of a tensor.
-   *  <p>
-   *  Reduces `input` along the dimensions given in `axis`. Unless
-   *  `keep_dims` is true, the rank of the tensor is reduced by 1 for each entry in
-   *  `axis`. If `keep_dims` is true, the reduced dimensions are
-   *  retained with length 1.
-   *
-   * @param <T> data type for {@code output()} output
-   * @param input The tensor to reduce.
-   * @param axis The dimensions to reduce. Must be in the range
-   *  `[-rank(input), rank(input))`.
-   * @param options carries optional attributes values
-   * @return a new instance of Max
-   */
-  public <T extends TType, U extends TNumber> Max<T> max(Operand<T> input, Operand<U> axis,
-      Max.Options... options) {
-    return Max.create(scope, input, axis, options);
-  }
-
-  /**
-   * Forwards the value of an available tensor from `inputs` to `output`.
-   *  <p>
-   *  `Merge` waits for at least one of the tensors in `inputs` to become available.
-   *  It is usually combined with `Switch` to implement branching.
-   *  <p>
-   *  `Merge` forwards the first tensor to become available to `output`, and sets
-   *  `value_index` to its index in `inputs`.
-   *
-   * @param <T> data type for {@code output()} output
-   * @param inputs The input tensors, exactly one of which will become available.
-   * @return a new instance of Merge
-   */
-  public <T extends TType> Merge<T> merge(Iterable<Operand<T>> inputs) {
-    return Merge.create(scope, inputs);
-  }
-
-  /**
-   * Computes the minimum of elements across dimensions of a tensor.
-   *  <p>
-   *  Reduces `input` along the dimensions given in `axis`. Unless
-   *  `keep_dims` is true, the rank of the tensor is reduced by 1 for each entry in
-   *  `axis`. If `keep_dims` is true, the reduced dimensions are
-   *  retained with length 1.
-   *
-   * @param <T> data type for {@code output()} output
-   * @param input The tensor to reduce.
-   * @param axis The dimensions to reduce. Must be in the range
-   *  `[-rank(input), rank(input))`.
-   * @param options carries optional attributes values
-   * @return a new instance of Min
-   */
-  public <T extends TType, U extends TNumber> Min<T> min(Operand<T> input, Operand<U> axis,
-      Min.Options... options) {
-    return Min.create(scope, input, axis, options);
-  }
-
-  /**
-   * Pads a tensor with mirrored values.
-   *  <p>
-   *  This operation pads a `input` with mirrored values according to the `paddings`
-   *  you specify. `paddings` is an integer tensor with shape `[n, 2]`, where n is
-   *  the rank of `input`. For each dimension D of `input`, `paddings[D, 0]` indicates
-   *  how many values to add before the contents of `input` in that dimension, and
-   *  `paddings[D, 1]` indicates how many values to add after the contents of `input`
-   *  in that dimension. Both `paddings[D, 0]` and `paddings[D, 1]` must be no greater
-   *  than `input.dim_size(D)` (or `input.dim_size(D) - 1`) if `copy_border` is true
-   *  (if false, respectively).
-   *  <p>
-   *  The padded size of each dimension D of the output is:
-   *  <p>
-   *  `paddings(D, 0) + input.dim_size(D) + paddings(D, 1)`
-   *  <p>
-   *  For example:
-   *  <pre>{@code
-   *  # 't' is [[1, 2, 3], [4, 5, 6]].
-   *  # 'paddings' is [[1, 1]], [2, 2]].
-   *  # 'mode' is SYMMETRIC.
-   *  # rank of 't' is 2.
-   *  pad(t, paddings) ==> [[2, 1, 1, 2, 3, 3, 2]
-   *                        [2, 1, 1, 2, 3, 3, 2]
-   *                        [5, 4, 4, 5, 6, 6, 5]
-   *                        [5, 4, 4, 5, 6, 6, 5]]
-   *  }</pre>
-   *
-   * @param <T> data type for {@code output()} output
-   * @param input The input tensor to be padded.
-   * @param paddings A two-column matrix specifying the padding sizes. The number of
-   *  rows must be the same as the rank of `input`.
-   * @param mode Either `REFLECT` or `SYMMETRIC`. In reflect mode the padded regions
-   *  do not include the borders, while in symmetric mode the padded regions
-   *  do include the borders. For example, if `input` is `[1, 2, 3]` and `paddings`
-   *  is `[0, 2]`, then the output is `[1, 2, 3, 2, 1]` in reflect mode, and
-   *  it is `[1, 2, 3, 3, 2]` in symmetric mode.
-   * @return a new instance of MirrorPad
-   */
-  public <T extends TType, U extends TNumber> MirrorPad<T> mirrorPad(Operand<T> input,
-      Operand<U> paddings, String mode) {
-    return MirrorPad.create(scope, input, paddings, mode);
-  }
-
-  /**
-   * Wraps an arbitrary MLIR computation expressed as a module with a main() function.
-   *  <p>
-   *  This operation does not have an associated kernel and is not intended to be
-   *  executed in a regular TensorFlow session. Instead it is intended to be used for
-   *  testing or for special case where a user intends to pass custom MLIR computation
-   *  through a TensorFlow graph with the intent of having custom tooling processing
-   *  it downstream (when targeting a different environment, like TensorFlow lite for
-   *  example).
-   *  The MLIR module is expected to have a main() function that will be used as an
-   *  entry point. The inputs to the operations will be passed as argument to the
-   *  main() function and the returned values of the main function mapped to the
-   *  outputs.
-   *  Example usage:
-   *  <pre>{@code
-   *  import tensorflow as tf
-   *  from tensorflow.compiler.mlir.tensorflow.gen_mlir_passthrough_op import mlir_passthrough_op
-   *
-   *  mlir_module = '''python
-   *  func @main(%arg0 : tensor<10xf32>, %arg1 : tensor<10xf32>) -> tensor<10x10xf32> {
-   *     %add = "magic.op"(%arg0, %arg1) : (tensor<10xf32>, tensor<10xf32>) -> tensor<10x10xf32>
-   *     return %ret : tensor<10x10xf32>
-   *  }
-   *  '''
-   *
-   * @tf.function def foo(x, y):
-   *    return = mlir_passthrough_op([x, y], mlir_module, Toutputs=[tf.float32])
-   *
-   *  graph_def = foo.get_concrete_function(tf.TensorSpec([10], tf.float32), tf.TensorSpec([10], tf.float32)).graph.as_graph_def()
-   *  }</pre>
-   * @param inputs
-   * @param mlirModule
-   * @param Toutputs
-   * @return a new instance of MlirPassthroughOp
-   */
-  public MlirPassthroughOp mlirPassthroughOp(Iterable<Operand<?>> inputs, String mlirModule,
-      List<DataType<?>> Toutputs) {
-    return MlirPassthroughOp.create(scope, inputs, mlirModule, Toutputs);
-  }
-
-  /**
-   * Creates an empty hash table that uses tensors as the backing store.
-   *  <p>
-   *  It uses "open addressing" with quadratic reprobing to resolve
-   *  collisions.
-   *  <p>
-   *  This op creates a mutable hash table, specifying the type of its keys and
-   *  values. Each value must be a scalar. Data can be inserted into the table using
-   *  the insert operations. It does not support the initialization operation.
-   *
-   * @param emptyKey The key used to represent empty key buckets internally. Must not
-   *  be used in insert or lookup operations.
-   * @param deletedKey
-   * @param valueDtype Type of the table values.
-   * @param options carries optional attributes values
-   * @return a new instance of MutableDenseHashTable
-   */
-  public <T extends TType, U extends TType> MutableDenseHashTable mutableDenseHashTable(
-      Operand<T> emptyKey, Operand<T> deletedKey, DataType<U> valueDtype,
-      MutableDenseHashTable.Options... options) {
-    return MutableDenseHashTable.create(scope, emptyKey, deletedKey, valueDtype, options);
-  }
-
-  /**
-   * Creates an empty hash table.
-   *  <p>
-   *  This op creates a mutable hash table, specifying the type of its keys and
-   *  values. Each value must be a scalar. Data can be inserted into the table using
-   *  the insert operations. It does not support the initialization operation.
-   *
-   * @param keyDtype Type of the table keys.
-   * @param valueDtype Type of the table values.
-   * @param options carries optional attributes values
-   * @return a new instance of MutableHashTable
-   */
-  public <T extends TType, U extends TType> MutableHashTable mutableHashTable(DataType<T> keyDtype,
-      DataType<U> valueDtype, MutableHashTable.Options... options) {
-    return MutableHashTable.create(scope, keyDtype, valueDtype, options);
-  }
-
-  /**
-   * Creates an empty hash table.
-   *  <p>
-   *  This op creates a mutable hash table, specifying the type of its keys and
-   *  values. Each value must be a vector. Data can be inserted into the table using
-   *  the insert operations. It does not support the initialization operation.
-   *
-   * @param keyDtype Type of the table keys.
-   * @param valueDtype Type of the table values.
-   * @param options carries optional attributes values
-   * @return a new instance of MutableHashTableOfTensors
-   */
-  public <T extends TType, U extends TType> MutableHashTableOfTensors mutableHashTableOfTensors(
-      DataType<T> keyDtype, DataType<U> valueDtype, MutableHashTableOfTensors.Options... options) {
-    return MutableHashTableOfTensors.create(scope, keyDtype, valueDtype, options);
-  }
-
-  /**
-   * Creates a Mutex resource that can be locked by `MutexLock`.
-   *
-   * @param options carries optional attributes values
-   * @return a new instance of Mutex
-   */
-  public Mutex mutex(Mutex.Options... options) {
-    return Mutex.create(scope, options);
-  }
-
-  /**
-   * Locks a mutex resource.  The output is the lock.  So long as the lock tensor
-   *  <p>
-   *  is alive, any other request to use `MutexLock` with this mutex will wait.
-   *  <p>
-   *  This is particularly useful for creating a critical section when used in
-   *  conjunction with `MutexLockIdentity`:
-   *  <pre>{@code
-   *  mutex = mutex_v2(
-   *    shared_name=handle_name, container=container, name=name)
-   *
-   *  def execute_in_critical_section(fn, *args, **kwargs):
-   *    lock = gen_resource_variable_ops.mutex_lock(mutex)
-   *
-   *    with ops.control_dependencies([lock]):
-   *      r = fn(*args, **kwargs)
-   *
-   *    with ops.control_dependencies(nest.flatten(r)):
-   *      with ops.colocate_with(mutex):
-   *        ensure_lock_exists = mutex_lock_identity(lock)
-   *
-   *      # Make sure that if any element of r is accessed, all of
-   *      # them are executed together.
-   *      r = nest.map_structure(tf.identity, r)
-   *
-   *    with ops.control_dependencies([ensure_lock_exists]):
-   *      return nest.map_structure(tf.identity, r)
-   *  }</pre>
-   *  While `fn` is running in the critical section, no other functions which wish to
-   *  use this critical section may run.
-   *  <p>
-   *  Often the use case is that two executions of the same graph, in parallel,
-   *  wish to run `fn`; and we wish to ensure that only one of them executes
-   *  at a time.  This is especially important if `fn` modifies one or more
-   *  variables at a time.
-   *  <p>
-   *  It is also useful if two separate functions must share a resource, but we
-   *  wish to ensure the usage is exclusive.
-   *
-   * @param mutex The mutex resource to lock.
-   * @return a new instance of MutexLock
-   */
-  public MutexLock mutexLock(Operand<?> mutex) {
-    return MutexLock.create(scope, mutex);
-  }
-
-  /**
-   * Makes its input available to the next iteration.
-   *
-   * @param <T> data type for {@code output()} output
-   * @param data The tensor to be made available to the next iteration.
-   * @return a new instance of NextIteration
-   */
-  public <T extends TType> NextIteration<T> nextIteration(Operand<T> data) {
-    return NextIteration.create(scope, data);
-  }
-
-  /**
-   * Does nothing. Only useful as a placeholder for control edges.
-   *
-   * @return a new instance of NoOp
-   */
-  public NoOp noOp() {
-    return NoOp.create(scope);
-  }
-
-  /**
-   * Returns a one-hot tensor.
-   *  <p>
-   *  The locations represented by indices in `indices` take value `on_value`,
-   *  while all other locations take value `off_value`.
-   *  <p>
-   *  If the input `indices` is rank `N`, the output will have rank `N+1`,
-   *  The new axis is created at dimension `axis` (default: the new axis is
-   *  appended at the end).
-   *  <p>
-   *  If `indices` is a scalar the output shape will be a vector of length `depth`.
-   *  <p>
-   *  If `indices` is a vector of length `features`, the output shape will be:
-   *  <pre>{@code
-   *    features x depth if axis == -1
-   *    depth x features if axis == 0
-   *  }</pre>
-   *  If `indices` is a matrix (batch) with shape `[batch, features]`,
-   *  the output shape will be:
-   *  <pre>{@code
-   *    batch x features x depth if axis == -1
-   *    batch x depth x features if axis == 1
-   *    depth x batch x features if axis == 0
-   *  }</pre>
-   *  Examples
-   *  =========
-   *  <p>
-   *  Suppose that
-   *  <pre>{@code
-   *    indices = [0, 2, -1, 1]
-   *    depth = 3
-   *    on_value = 5.0
-   *    off_value = 0.0
-   *    axis = -1
-   *  }</pre>
-   *  Then output is `[4 x 3]`:
-   *  <pre>{@code
-   *  output =
-   *    [5.0 0.0 0.0]  // one_hot(0)
-   *    [0.0 0.0 5.0]  // one_hot(2)
-   *    [0.0 0.0 0.0]  // one_hot(-1)
-   *    [0.0 5.0 0.0]  // one_hot(1)
-   *  }</pre>
-   *  Suppose that
-   *  <pre>{@code
-   *    indices = [0, 2, -1, 1]
-   *    depth = 3
-   *    on_value = 0.0
-   *    off_value = 3.0
-   *    axis = 0
-   *  }</pre>
-   *  Then output is `[3 x 4]`:
-   *  <pre>{@code
-   *  output =
-   *    [0.0 3.0 3.0 3.0]
-   *    [3.0 3.0 3.0 0.0]
-   *    [3.0 3.0 3.0 3.0]
-   *    [3.0 0.0 3.0 3.0]
-   *  //  ^                one_hot(0)
-   *  //      ^            one_hot(2)
-   *  //          ^        one_hot(-1)
-   *  //              ^    one_hot(1)
-   *  }</pre>
-   *  Suppose that
-   *  <pre>{@code
-   *    indices = [[0, 2], [1, -1]]
-   *    depth = 3
-   *    on_value = 1.0
-   *    off_value = 0.0
-   *    axis = -1
-   *  }</pre>
-   *  Then output is `[2 x 2 x 3]`:
-   *  <pre>{@code
-   *  output =
-   *    [
-   *      [1.0, 0.0, 0.0]  // one_hot(0)
-   *      [0.0, 0.0, 1.0]  // one_hot(2)
-   *    ][
-   *      [0.0, 1.0, 0.0]  // one_hot(1)
-   *      [0.0, 0.0, 0.0]  // one_hot(-1)
-   *    ]
-   *  }</pre>
-   *
-   * @param <U> data type for {@code output()} output
-   * @param indices A tensor of indices.
-   * @param depth A scalar defining the depth of the one hot dimension.
-   * @param onValue A scalar defining the value to fill in output when `indices[j] = i`.
-   * @param offValue A scalar defining the value to fill in output when `indices[j] != i`.
-   * @param options carries optional attributes values
-   * @return a new instance of OneHot
-   */
-  public <U extends TType, T extends TNumber> OneHot<U> oneHot(Operand<T> indices,
-      Operand<TInt32> depth, Operand<U> onValue, Operand<U> offValue, OneHot.Options... options) {
-    return OneHot.create(scope, indices, depth, onValue, offValue, options);
-  }
-
-  /**
-   * Returns a tensor of ones with the same shape and type as x.
-   *
-   * @param <T> data type for {@code y()} output
-   * @param x a tensor of type T.
-   * @return a new instance of OnesLike
-   */
-  public <T extends TType> OnesLike<T> onesLike(Operand<T> x) {
-    return OnesLike.create(scope, x);
-  }
-
-  /**
-   * Op removes all elements in the underlying container.
-   *
-   * @param dtypes
-   * @param options carries optional attributes values
-   * @return a new instance of OrderedMapClear
-   */
-  public OrderedMapClear orderedMapClear(List<DataType<?>> dtypes,
-      OrderedMapClear.Options... options) {
-    return OrderedMapClear.create(scope, dtypes, options);
-  }
-
-  /**
-   * Op returns the number of incomplete elements in the underlying container.
-   *
-   * @param dtypes
-   * @param options carries optional attributes values
-   * @return a new instance of OrderedMapIncompleteSize
-   */
-  public OrderedMapIncompleteSize orderedMapIncompleteSize(List<DataType<?>> dtypes,
-      OrderedMapIncompleteSize.Options... options) {
-    return OrderedMapIncompleteSize.create(scope, dtypes, options);
-  }
-
-  /**
-   * Op peeks at the values at the specified key.  If the
-   *  <p>
-   *  underlying container does not contain this key
-   *  this op will block until it does.   This Op is optimized for
-   *  performance.
-   *
-   * @param key
-   * @param indices
-   * @param dtypes
-   * @param options carries optional attributes values
-   * @return a new instance of OrderedMapPeek
-   */
-  public OrderedMapPeek orderedMapPeek(Operand<TInt64> key, Operand<TInt32> indices,
-      List<DataType<?>> dtypes, OrderedMapPeek.Options... options) {
-    return OrderedMapPeek.create(scope, key, indices, dtypes, options);
-  }
-
-  /**
-   * Op returns the number of elements in the underlying container.
-   *
-   * @param dtypes
-   * @param options carries optional attributes values
-   * @return a new instance of OrderedMapSize
-   */
-  public OrderedMapSize orderedMapSize(List<DataType<?>> dtypes,
-      OrderedMapSize.Options... options) {
-    return OrderedMapSize.create(scope, dtypes, options);
-  }
-
-  /**
-   * Stage (key, values) in the underlying container which behaves like a ordered
-   *  <p>
-   *  associative container.   Elements are ordered by key.
-   *
-   * @param key int64
-   * @param indices
-   * @param values a list of tensors
-   *  dtypes A list of data types that inserted values should adhere to.
-   * @param dtypes
-   * @param options carries optional attributes values
-   * @return a new instance of OrderedMapStage
-   */
-  public OrderedMapStage orderedMapStage(Operand<TInt64> key, Operand<TInt32> indices,
-      Iterable<Operand<?>> values, List<DataType<?>> dtypes, OrderedMapStage.Options... options) {
-    return OrderedMapStage.create(scope, key, indices, values, dtypes, options);
-  }
-
-  /**
-   * Op removes and returns the values associated with the key
-   *  <p>
-   *  from the underlying container.   If the underlying container
-   *  does not contain this key, the op will block until it does.
-   *
-   * @param key
-   * @param indices
-   * @param dtypes
-   * @param options carries optional attributes values
-   * @return a new instance of OrderedMapUnstage
-   */
-  public OrderedMapUnstage orderedMapUnstage(Operand<TInt64> key, Operand<TInt32> indices,
-      List<DataType<?>> dtypes, OrderedMapUnstage.Options... options) {
-    return OrderedMapUnstage.create(scope, key, indices, dtypes, options);
-  }
-
-  /**
-   * Op removes and returns the (key, value) element with the smallest
-   *  <p>
-   *  key from the underlying container.   If the underlying container
-   *  does not contain elements, the op will block until it does.
-   *
-   * @param indices
-   * @param dtypes
-   * @param options carries optional attributes values
-   * @return a new instance of OrderedMapUnstageNoKey
-   */
-  public OrderedMapUnstageNoKey orderedMapUnstageNoKey(Operand<TInt32> indices,
-      List<DataType<?>> dtypes, OrderedMapUnstageNoKey.Options... options) {
-    return OrderedMapUnstageNoKey.create(scope, indices, dtypes, options);
-  }
-
-  /**
-   * Pads a tensor.
-   *  <p>
-   *  This operation pads `input` according to the `paddings` and `constant_values`
-   *  you specify. `paddings` is an integer tensor with shape `[Dn, 2]`, where n is
-   *  the rank of `input`. For each dimension D of `input`, `paddings[D, 0]` indicates
-   *  how many padding values to add before the contents of `input` in that dimension,
-   *  and `paddings[D, 1]` indicates how many padding values to add after the contents
-   *  of `input` in that dimension. `constant_values` is a scalar tensor of the same
-   *  type as `input` that indicates the value to use for padding `input`.
-   *  <p>
-   *  The padded size of each dimension D of the output is:
-   *  <p>
-   *  `paddings(D, 0) + input.dim_size(D) + paddings(D, 1)`
-   *  <p>
-   *  For example:
-   *  <pre>{@code
-   *  # 't' is [[1, 1], [2, 2]]
-   *  # 'paddings' is [[1, 1], [2, 2]]
-   *  # 'constant_values' is 0
-   *  # rank of 't' is 2
-   *  pad(t, paddings) ==> [[0, 0, 0, 0, 0, 0]
-   *                        [0, 0, 1, 1, 0, 0]
-   *                        [0, 0, 2, 2, 0, 0]
-   *                        [0, 0, 0, 0, 0, 0]]
-   *  }</pre>
-   *
-   * @param <T> data type for {@code output()} output
-   * @param input
-   * @param paddings
-   * @param constantValues
-   * @return a new instance of Pad
-   */
-  public <T extends TType, U extends TNumber> Pad<T> pad(Operand<T> input, Operand<U> paddings,
-      Operand<T> constantValues) {
-    return Pad.create(scope, input, paddings, constantValues);
-  }
-
-  /**
-   * Concatenates a list of `N` tensors along the first dimension.
-   *  <p>
-   *  The input tensors are all required to have size 1 in the first dimension.
-   *  <p>
-   *  For example:
-   *  <pre>{@code
-   *  # 'x' is [[1, 4]]
-   *  # 'y' is [[2, 5]]
-   *  # 'z' is [[3, 6]]
-   *  parallel_concat([x, y, z]) => [[1, 4], [2, 5], [3, 6]]  # Pack along first dim.
-   *  }</pre>
-   *  The difference between concat and parallel_concat is that concat requires all
-   *  of the inputs be computed before the operation will begin but doesn't require
-   *  that the input shapes be known during graph construction.  Parallel concat
-   *  will copy pieces of the input into the output as they become available, in
-   *  some situations this can provide a performance benefit.
-   *
-   * @param <T> data type for {@code output()} output
-   * @param values Tensors to be concatenated. All must have size 1 in the first dimension
-   *  and same shape.
-   * @param shape the final shape of the result; should be equal to the shapes of any input
-   *  but with the number of input values in the first dimension.
-   * @return a new instance of ParallelConcat
-   */
-  public <T extends TType> ParallelConcat<T> parallelConcat(Iterable<Operand<T>> values,
-      Shape shape) {
-    return ParallelConcat.create(scope, values, shape);
-  }
-
-  /**
-   * Interleave the values from the `data` tensors into a single tensor.
-   *  <p>
-   *  Builds a merged tensor such that
-   *  <pre>{@code
-   *      merged[indices[m][i, ..., j], ...] = data[m][i, ..., j, ...]
-   *  }</pre>
-   *  For example, if each `indices[m]` is scalar or vector, we have
-   *  <pre>{@code
-   *      # Scalar indices:
-   *      merged[indices[m], ...] = data[m][...]
-   *
-   *      # Vector indices:
-   *      merged[indices[m][i], ...] = data[m][i, ...]
-   *  }</pre>
-   *  Each `data[i].shape` must start with the corresponding `indices[i].shape`,
-   *  and the rest of `data[i].shape` must be constant w.r.t. `i`.  That is, we
-   *  must have `data[i].shape = indices[i].shape + constant`.  In terms of this
-   *  `constant`, the output shape is
-   *  <p>
-   *      merged.shape = [max(indices)] + constant
-   *  <p>
-   *  Values may be merged in parallel, so if an index appears in both `indices[m][i]`
-   *  and `indices[n][j]`, the result may be invalid. This differs from the normal
-   *  DynamicStitch operator that defines the behavior in that case.
-   *  <p>
-   *  For example:
-   *  <pre>{@code
-   *      indices[0] = 6
-   *      indices[1] = [4, 1]
-   *      indices[2] = [[5, 2], [0, 3]]
-   *      data[0] = [61, 62]
-   *      data[1] = [[41, 42], [11, 12]]
-   *      data[2] = [[[51, 52], [21, 22]], [[1, 2], [31, 32]]]
-   *      merged = [[1, 2], [11, 12], [21, 22], [31, 32], [41, 42],
-   *                [51, 52], [61, 62]]
-   *  }</pre>
-   *  This method can be used to merge partitions created by `dynamic_partition`
-   *  as illustrated on the following example:
-   *  <pre>{@code
-   *      # Apply function (increments x_i) on elements for which a certain condition
-   *      # apply (x_i != -1 in this example).
-   *      x=tf.constant([0.1, -1., 5.2, 4.3, -1., 7.4])
-   *      condition_mask=tf.not_equal(x,tf.constant(-1.))
-   *      partitioned_data = tf.dynamic_partition(
-   *          x, tf.cast(condition_mask, tf.int32) , 2)
-   *      partitioned_data[1] = partitioned_data[1] + 1.0
-   *      condition_indices = tf.dynamic_partition(
-   *          tf.range(tf.shape(x)[0]), tf.cast(condition_mask, tf.int32) , 2)
-   *      x = tf.dynamic_stitch(condition_indices, partitioned_data)
-   *      # Here x=[1.1, -1., 6.2, 5.3, -1, 8.4], the -1. values remain
-   *      # unchanged.
-   *  }</pre>
-   *  <div style="width:70%; margin:auto; margin-bottom:10px; margin-top:20px;">
-   *  <img style="width:100%" src="https://www.tensorflow.org/images/DynamicStitch.png" alt>
-   *  </div>
-   *
-   * @param <T> data type for {@code merged()} output
-   * @param indices
-   * @param data
-   * @return a new instance of ParallelDynamicStitch
-   */
-  public <T extends TType> ParallelDynamicStitch<T> parallelDynamicStitch(
-      Iterable<Operand<TInt32>> indices, Iterable<Operand<T>> data) {
-    return ParallelDynamicStitch.create(scope, indices, data);
-  }
-
-  /**
-   * A placeholder op for a value that will be fed into the computation.
-   *  <p>
-   *  N.B. This operation will fail with an error if it is executed. It is
-   *  intended as a way to represent a value that will always be fed, and to
-   *  provide attrs that enable the fed value to be checked at runtime.
-   *
-   * @param <T> data type for {@code output()} output
-   * @param dtype The type of elements in the tensor.
-   * @param options carries optional attributes values
-   * @return a new instance of Placeholder
-   */
-  public <T extends TType> Placeholder<T> placeholder(DataType<T> dtype,
-      Placeholder.Options... options) {
-    return Placeholder.create(scope, dtype, options);
-  }
-
-  /**
-   * A placeholder op that passes through `input` when its output is not fed.
-   *
-   * @param <T> data type for {@code output()} output
-   * @param input The default value to produce when `output` is not fed.
-   * @param shape The (possibly partial) shape of the tensor.
-   * @return a new instance of PlaceholderWithDefault
-   */
-  public <T extends TType> PlaceholderWithDefault<T> placeholderWithDefault(Operand<T> input,
-      Shape shape) {
-    return PlaceholderWithDefault.create(scope, input, shape);
-  }
-
-  /**
-   * Prints a string scalar.
-   *  <p>
-   *  Prints a string scalar to the desired output_stream.
-   *
-   * @param input The string scalar to print.
-   * @param options carries optional attributes values
-   * @return a new instance of Print
-   */
-  public Print print(Operand<TString> input, Print.Options... options) {
-    return Print.create(scope, input, options);
-  }
-
-  /**
-   * Computes the product of elements across dimensions of a tensor.
-   *  <p>
-   *  Reduces `input` along the dimensions given in `axis`. Unless
-   *  `keep_dims` is true, the rank of the tensor is reduced by 1 for each entry in
-   *  `axis`. If `keep_dims` is true, the reduced dimensions are
-   *  retained with length 1.
-   *
-   * @param <T> data type for {@code output()} output
-   * @param input The tensor to reduce.
-   * @param axis The dimensions to reduce. Must be in the range
-   *  `[-rank(input), rank(input))`.
-   * @param options carries optional attributes values
-   * @return a new instance of Prod
-   */
-  public <T extends TType, U extends TNumber> Prod<T> prod(Operand<T> input, Operand<U> axis,
-      Prod.Options... options) {
-    return Prod.create(scope, input, axis, options);
-  }
-
-  /**
-   * Reshapes a quantized tensor as per the Reshape op.
-   *  <p>
-   *  ```
-   *
-   * @param <T> data type for {@code output()} output
-   * @param tensor
-   * @param shape Defines the shape of the output tensor.
-   * @param inputMin The minimum value of the input.
-   * @param inputMax The maximum value of the input.
-   * @return a new instance of QuantizedReshape
-   */
-  public <T extends TType, U extends TNumber> QuantizedReshape<T> quantizedReshape(
-      Operand<T> tensor, Operand<U> shape, Operand<TFloat32> inputMin, Operand<TFloat32> inputMax) {
-    return QuantizedReshape.create(scope, tensor, shape, inputMin, inputMax);
-  }
-
-  /**
-   * Creates a sequence of numbers.
-   *  <p>
-   *  This operation creates a sequence of numbers that begins at `start` and
-   *  extends by increments of `delta` up to but not including `limit`.
-   *  <p>
-   *  For example:
-   *  <pre>{@code
-   *  # 'start' is 3
-   *  # 'limit' is 18
-   *  # 'delta' is 3
-   *  tf.range(start, limit, delta) ==> [3, 6, 9, 12, 15]
-   *  }</pre>
-   *
-   * @param <T> data type for {@code output()} output
-   * @param start 0-D (scalar). First entry in the sequence.
-   * @param limit 0-D (scalar). Upper limit of sequence, exclusive.
-   * @param delta 0-D (scalar). Optional. Default is 1. Number that increments `start`.
-   * @return a new instance of Range
-   */
-  public <T extends TNumber> Range<T> range(Operand<T> start, Operand<T> limit, Operand<T> delta) {
-    return Range.create(scope, start, limit, delta);
-  }
-
-  /**
-   * Returns the rank of a tensor.
-   *  <p>
-   *  This operation returns an integer representing the rank of `input`.
-   *  <p>
-   *  For example:
-   *  <pre>{@code
-   *  # 't' is [[[1, 1, 1], [2, 2, 2]], [[3, 3, 3], [4, 4, 4]]]
-   *  # shape of tensor 't' is [2, 2, 3]
-   *  rank(t) ==> 3
-   *  }</pre>
-   *  <b>Note</b>: The rank of a tensor is not the same as the rank of a matrix. The rank
-   *  of a tensor is the number of indices required to uniquely select each element
-   *  of the tensor. Rank is also known as "order", "degree", or "ndims."
-   *
-   * @param input
-   * @return a new instance of Rank
-   */
-  public <T extends TType> Rank rank(Operand<T> input) {
-    return Rank.create(scope, input);
-  }
-
-  /**
-   * Reads the value of a variable.
-   *  <p>
-   *  The tensor returned by this operation is immutable.
-   *  <p>
-   *  The value returned by this operation is guaranteed to be influenced by all the
-   *  writes on which this operation depends directly or indirectly, and to not be
-   *  influenced by any of the writes which depend directly or indirectly on this
-   *  operation.
-   *
-   * @param <T> data type for {@code value()} output
-   * @param resource handle to the resource in which to store the variable.
-   * @param dtype the dtype of the value.
-   * @return a new instance of ReadVariableOp
-   */
-  public <T extends TType> ReadVariableOp<T> readVariableOp(Operand<?> resource,
-      DataType<T> dtype) {
-    return ReadVariableOp.create(scope, resource, dtype);
-  }
-
-  /**
-   * Computes the "logical and" of elements across dimensions of a tensor.
-   *  <p>
-   *  Reduces `input` along the dimensions given in `axis`. Unless
-   *  `keep_dims` is true, the rank of the tensor is reduced by 1 for each entry in
-   *  `axis`. If `keep_dims` is true, the reduced dimensions are
-   *  retained with length 1.
-   *
-   * @param input The tensor to reduce.
-   * @param axis The dimensions to reduce. Must be in the range
-   *  `[-rank(input), rank(input))`.
-   * @param options carries optional attributes values
-   * @return a new instance of ReduceAll
-   */
-  public <T extends TNumber> ReduceAll reduceAll(Operand<TBool> input, Operand<T> axis,
-      ReduceAll.Options... options) {
-    return ReduceAll.create(scope, input, axis, options);
-  }
-
-  /**
-   * Computes the "logical or" of elements across dimensions of a tensor.
-   *  <p>
-   *  Reduces `input` along the dimensions given in `axis`. Unless
-   *  `keep_dims` is true, the rank of the tensor is reduced by 1 for each entry in
-   *  `axis`. If `keep_dims` is true, the reduced dimensions are
-   *  retained with length 1.
-   *
-   * @param input The tensor to reduce.
-   * @param axis The dimensions to reduce. Must be in the range
-   *  `[-rank(input), rank(input))`.
-   * @param options carries optional attributes values
-   * @return a new instance of ReduceAny
-   */
-  public <T extends TNumber> ReduceAny reduceAny(Operand<TBool> input, Operand<T> axis,
-      ReduceAny.Options... options) {
-    return ReduceAny.create(scope, input, axis, options);
-  }
-
-  /**
-   * Computes the maximum of elements across dimensions of a tensor.
-   *  <p>
-   *  Reduces `input` along the dimensions given in `axis`. Unless
-   *  `keep_dims` is true, the rank of the tensor is reduced by 1 for each entry in
-   *  `axis`. If `keep_dims` is true, the reduced dimensions are
-   *  retained with length 1.
-   *
-   * @param <T> data type for {@code output()} output
-   * @param input The tensor to reduce.
-   * @param axis The dimensions to reduce. Must be in the range
-   *  `[-rank(input), rank(input))`.
-   * @param options carries optional attributes values
-   * @return a new instance of ReduceMax
-   */
-  public <T extends TType, U extends TNumber> ReduceMax<T> reduceMax(Operand<T> input,
-      Operand<U> axis, ReduceMax.Options... options) {
-    return ReduceMax.create(scope, input, axis, options);
-  }
-
-  /**
-   * Computes the minimum of elements across dimensions of a tensor.
-   *  <p>
-   *  Reduces `input` along the dimensions given in `axis`. Unless
-   *  `keep_dims` is true, the rank of the tensor is reduced by 1 for each entry in
-   *  `axis`. If `keep_dims` is true, the reduced dimensions are
-   *  retained with length 1.
-   *
-   * @param <T> data type for {@code output()} output
-   * @param input The tensor to reduce.
-   * @param axis The dimensions to reduce. Must be in the range
-   *  `[-rank(input), rank(input))`.
-   * @param options carries optional attributes values
-   * @return a new instance of ReduceMin
-   */
-  public <T extends TType, U extends TNumber> ReduceMin<T> reduceMin(Operand<T> input,
-      Operand<U> axis, ReduceMin.Options... options) {
-    return ReduceMin.create(scope, input, axis, options);
-  }
-
-  /**
-   * Computes the product of elements across dimensions of a tensor.
-   *  <p>
-   *  Reduces `input` along the dimensions given in `axis`. Unless
-   *  `keep_dims` is true, the rank of the tensor is reduced by 1 for each entry in
-   *  `axis`. If `keep_dims` is true, the reduced dimensions are
-   *  retained with length 1.
-   *
-   * @param <T> data type for {@code output()} output
-   * @param input The tensor to reduce.
-   * @param axis The dimensions to reduce. Must be in the range
-   *  `[-rank(input), rank(input))`.
-   * @param options carries optional attributes values
-   * @return a new instance of ReduceProd
-   */
-  public <T extends TType, U extends TNumber> ReduceProd<T> reduceProd(Operand<T> input,
-      Operand<U> axis, ReduceProd.Options... options) {
-    return ReduceProd.create(scope, input, axis, options);
-  }
-
-  /**
-   * Computes the sum of elements across dimensions of a tensor.
-   *  <p>
-   *  Reduces `input` along the dimensions given in `axis`. Unless
-   *  `keep_dims` is true, the rank of the tensor is reduced by 1 for each entry in
-   *  `axis`. If `keep_dims` is true, the reduced dimensions are
-   *  retained with length 1.
-   *
-   * @param <T> data type for {@code output()} output
-   * @param input The tensor to reduce.
-   * @param axis The dimensions to reduce. Must be in the range
-   *  `[-rank(input), rank(input))`.
-   * @param options carries optional attributes values
-   * @return a new instance of ReduceSum
-   */
-  public <T extends TType, U extends TNumber> ReduceSum<T> reduceSum(Operand<T> input,
-      Operand<U> axis, ReduceSum.Options... options) {
-    return ReduceSum.create(scope, input, axis, options);
-  }
-
-  /**
-   * Makes its input available to the next iteration.
-   *
-   * @param <T> data type for {@code output()} output
-   * @param data The tensor to be made available to the next iteration.
-   * @return a new instance of RefNextIteration
-   */
-  public <T extends TType> RefNextIteration<T> refNextIteration(Operand<T> data) {
-    return RefNextIteration.create(scope, data);
-  }
-
-  /**
-   * Forwards the `index`th element of `inputs` to `output`.
-   *
-   * @param <T> data type for {@code output()} output
-   * @param index A scalar that determines the input that gets selected.
-   * @param inputs A list of ref tensors, one of which will be forwarded to `output`.
-   * @return a new instance of RefSelect
-   */
-  public <T extends TType> RefSelect<T> refSelect(Operand<TInt32> index,
-      Iterable<Operand<T>> inputs) {
-    return RefSelect.create(scope, index, inputs);
-  }
-
-  /**
-   * Forwards the ref tensor `data` to the output port determined by `pred`.
-   *  <p>
-   *  If `pred` is true, the `data` input is forwarded to `output_true`. Otherwise,
-   *  the data goes to `output_false`.
-   *  <p>
-   *  See also `Switch` and `Merge`.
-   *
-   * @param <T> data type for {@code outputFalse()} output
-   * @param data The ref tensor to be forwarded to the appropriate output.
-   * @param pred A scalar that specifies which output port will receive data.
-   * @return a new instance of RefSwitch
-   */
-  public <T extends TType> RefSwitch<T> refSwitch(Operand<T> data, Operand<TBool> pred) {
-    return RefSwitch.create(scope, data, pred);
-  }
-
-  /**
-   * Execute a sub graph on a remote processor.
-   *  <p>
-   *  The graph specifications(such as graph itself, input tensors and output names)
-   *  are stored as a serialized protocol buffer of RemoteFusedGraphExecuteInfo
-   *  as serialized_remote_fused_graph_execute_info.
-   *  The specifications will be passed to a dedicated registered
-   *  remote fused graph executor.  The executor will send the graph specifications
-   *  to a remote processor and execute that graph.  The execution results
-   *  will be passed to consumer nodes as outputs of this node.
-   *
-   * @param inputs Arbitrary number of tensors with arbitrary data types
-   * @param Toutputs
-   * @param serializedRemoteFusedGraphExecuteInfo Serialized protocol buffer
-   *  of RemoteFusedGraphExecuteInfo which contains graph specifications.
-   * @return a new instance of RemoteFusedGraphExecute
-   */
-  public RemoteFusedGraphExecute remoteFusedGraphExecute(Iterable<Operand<?>> inputs,
-      List<DataType<?>> Toutputs, String serializedRemoteFusedGraphExecuteInfo) {
-    return RemoteFusedGraphExecute.create(scope, inputs, Toutputs, serializedRemoteFusedGraphExecuteInfo);
-  }
-
-  /**
-   * Reshapes a tensor.
-   *  <p>
-   *  Given `tensor`, this operation returns a tensor that has the same values
-   *  as `tensor` with shape `shape`.
-   *  <p>
-   *  If one component of 1-D tensor `shape` is the special value -1, the size of that
-   *  dimension is computed so that the total size remains constant.  In particular, a
-   *  `shape` of `[-1]` flattens into 1-D.  At most one component of `shape` may be
-   *  unknown.
-   *  <p>
-   *  The `shape` must be 1-D and the operation returns a tensor with shape
-   *  `shape` filled with the values of `tensor`. In this case, the number of elements
-   *  implied by `shape` must be the same as the number of elements in `tensor`.
-   *  <p>
-   *  It is an error if `shape` is not 1-D.
-   *  <p>
-   *  For example:
-   *  <pre>{@code
-   *  # tensor 't' is [1, 2, 3, 4, 5, 6, 7, 8, 9]
-   *  # tensor 't' has shape [9]
-   *  reshape(t, [3, 3]) ==> [[1, 2, 3],
-   *                          [4, 5, 6],
-   *                          [7, 8, 9]]
-   *
-   *  # tensor 't' is [[[1, 1], [2, 2]],
-   *  #                [[3, 3], [4, 4]]]
-   *  # tensor 't' has shape [2, 2, 2]
-   *  reshape(t, [2, 4]) ==> [[1, 1, 2, 2],
-   *                          [3, 3, 4, 4]]
-   *
-   *  # tensor 't' is [[[1, 1, 1],
-   *  #                 [2, 2, 2]],
-   *  #                [[3, 3, 3],
-   *  #                 [4, 4, 4]],
-   *  #                [[5, 5, 5],
-   *  #                 [6, 6, 6]]]
-   *  # tensor 't' has shape [3, 2, 3]
-   *  # pass '[-1]' to flatten 't'
-   *  reshape(t, [-1]) ==> [1, 1, 1, 2, 2, 2, 3, 3, 3, 4, 4, 4, 5, 5, 5, 6, 6, 6]
-   *
-   *  # -1 can also be used to infer the shape
-   *
-   *  # -1 is inferred to be 9:
-   *  reshape(t, [2, -1]) ==> [[1, 1, 1, 2, 2, 2, 3, 3, 3],
-   *                           [4, 4, 4, 5, 5, 5, 6, 6, 6]]
-   *  # -1 is inferred to be 2:
-   *  reshape(t, [-1, 9]) ==> [[1, 1, 1, 2, 2, 2, 3, 3, 3],
-   *                           [4, 4, 4, 5, 5, 5, 6, 6, 6]]
-   *  # -1 is inferred to be 3:
-   *  reshape(t, [ 2, -1, 3]) ==> [[[1, 1, 1],
-   *                                [2, 2, 2],
-   *                                [3, 3, 3]],
-   *                               [[4, 4, 4],
-   *                                [5, 5, 5],
-   *                                [6, 6, 6]]]
-   *
-   *  # tensor 't' is [7]
-   *  # shape `[]` reshapes to a scalar
-   *  reshape(t, []) ==> 7
-   *  }</pre>
-   *
-   * @param <T> data type for {@code output()} output
-   * @param tensor
-   * @param shape Defines the shape of the output tensor.
-   * @return a new instance of Reshape
-   */
-  public <T extends TType, U extends TNumber> Reshape<T> reshape(Operand<T> tensor,
-      Operand<U> shape) {
-    return Reshape.create(scope, tensor, shape);
-  }
-
-  /**
-   * Increments variable pointed to by 'resource' until it reaches 'limit'.
-   *
-   * @param <T> data type for {@code output()} output
-   * @param resource Should be from a scalar `Variable` node.
-   * @param limit If incrementing ref would bring it above limit, instead generates an
-   *  'OutOfRange' error.
-   * @param T
-   * @return a new instance of ResourceCountUpTo
-   */
-  public <T extends TNumber> ResourceCountUpTo<T> resourceCountUpTo(Operand<?> resource, Long limit,
-      DataType<T> T) {
-    return ResourceCountUpTo.create(scope, resource, limit, T);
-  }
-
-  /**
-   * Gather slices from the variable pointed to by `resource` according to `indices`.
-   *  <p>
-   *  `indices` must be an integer tensor of any dimension (usually 0-D or 1-D).
-   *  Produces an output tensor with shape `indices.shape + params.shape[1:]` where:
-   *  <pre>{@code
-   *      # Scalar indices
-   *      output[:, ..., :] = params[indices, :, ... :]
-   *
-   *      # Vector indices
-   *      output[i, :, ..., :] = params[indices[i], :, ... :]
-   *
-   *      # Higher rank indices
-   *      output[i, ..., j, :, ... :] = params[indices[i, ..., j], :, ..., :]
-   *  }</pre>
-   *
-   * @param <U> data type for {@code output()} output
-   * @param resource
-   * @param indices
-   * @param dtype
-   * @param options carries optional attributes values
-   * @return a new instance of ResourceGather
-   */
-  public <U extends TType, T extends TNumber> ResourceGather<U> resourceGather(Operand<?> resource,
-      Operand<T> indices, DataType<U> dtype, ResourceGather.Options... options) {
-    return ResourceGather.create(scope, resource, indices, dtype, options);
-  }
-
-  /**
-   *
-   * @param <U> data type for {@code output()} output
-   * @param resource
-   * @param indices
-   * @param dtype
-   * @return a new instance of ResourceGatherNd
-   */
-  public <U extends TType, T extends TNumber> ResourceGatherNd<U> resourceGatherNd(
-      Operand<?> resource, Operand<T> indices, DataType<U> dtype) {
-    return ResourceGatherNd.create(scope, resource, indices, dtype);
-  }
-
-  /**
-   * Adds sparse updates to the variable referenced by `resource`.
-   *  <p>
-   *  This operation computes
-   *  <p>
-   *      # Scalar indices
-   *      ref[indices, ...] += updates[...]
-   *  <p>
-   *      # Vector indices (for each i)
-   *      ref[indices[i], ...] += updates[i, ...]
-   *  <p>
-   *      # High rank indices (for each i, ..., j)
-   *      ref[indices[i, ..., j], ...] += updates[i, ..., j, ...]
-   *  <p>
-   *  Duplicate entries are handled correctly: if multiple `indices` reference
-   *  the same location, their contributions add.
-   *  <p>
-   *  Requires `updates.shape = indices.shape + ref.shape[1:]` or `updates.shape = []`.
-   *  <p>
-   *  <div style="width:70%; margin:auto; margin-bottom:10px; margin-top:20px;">
-   *  <img style="width:100%" src='https://www.tensorflow.org/images/ScatterAdd.png' alt>
-   *  </div>
-   *
-   * @param resource Should be from a `Variable` node.
-   * @param indices A tensor of indices into the first dimension of `ref`.
-   * @param updates A tensor of updated values to add to `ref`.
-   * @return a new instance of ResourceScatterAdd
-   */
-  public <T extends TNumber, U extends TType> ResourceScatterAdd resourceScatterAdd(
-      Operand<?> resource, Operand<T> indices, Operand<U> updates) {
-    return ResourceScatterAdd.create(scope, resource, indices, updates);
-  }
-
-  /**
-   * Divides sparse updates into the variable referenced by `resource`.
-   *  <p>
-   *  This operation computes
-   *  <p>
-   *      # Scalar indices
-   *      ref[indices, ...] /= updates[...]
-   *  <p>
-   *      # Vector indices (for each i)
-   *      ref[indices[i], ...] /= updates[i, ...]
-   *  <p>
-   *      # High rank indices (for each i, ..., j)
-   *      ref[indices[i, ..., j], ...] /= updates[i, ..., j, ...]
-   *  <p>
-   *  Duplicate entries are handled correctly: if multiple `indices` reference
-   *  the same location, their contributions multiply.
-   *  <p>
-   *  Requires `updates.shape = indices.shape + ref.shape[1:]` or `updates.shape = []`.
-   *  <p>
-   *  <div style="width:70%; margin:auto; margin-bottom:10px; margin-top:20px;">
-   *  <img style="width:100%" src='https://www.tensorflow.org/images/ScatterAdd.png' alt>
-   *  </div>
-   *
-   * @param resource Should be from a `Variable` node.
-   * @param indices A tensor of indices into the first dimension of `ref`.
-   * @param updates A tensor of updated values to add to `ref`.
-   * @return a new instance of ResourceScatterDiv
-   */
-  public <T extends TNumber, U extends TType> ResourceScatterDiv resourceScatterDiv(
-      Operand<?> resource, Operand<T> indices, Operand<U> updates) {
-    return ResourceScatterDiv.create(scope, resource, indices, updates);
-  }
-
-  /**
-   * Reduces sparse updates into the variable referenced by `resource` using the `max` operation.
-   *  <p>
-   *  This operation computes
-   *  <p>
-   *      # Scalar indices
-   *      ref[indices, ...] = max(ref[indices, ...], updates[...])
-   *  <p>
-   *      # Vector indices (for each i)
-   *      ref[indices[i], ...] = max(ref[indices[i], ...], updates[i, ...])
-   *  <p>
-   *      # High rank indices (for each i, ..., j)
-   *      ref[indices[i, ..., j], ...] = max(ref[indices[i, ..., j], ...], updates[i, ..., j, ...])
-   *  <p>
-   *  Duplicate entries are handled correctly: if multiple `indices` reference
-   *  the same location, their contributions are combined.
-   *  <p>
-   *  Requires `updates.shape = indices.shape + ref.shape[1:]` or `updates.shape = []`.
-   *  <p>
-   *  <div style="width:70%; margin:auto; margin-bottom:10px; margin-top:20px;">
-   *  <img style="width:100%" src='https://www.tensorflow.org/images/ScatterAdd.png' alt>
-   *  </div>
-   *
-   * @param resource Should be from a `Variable` node.
-   * @param indices A tensor of indices into the first dimension of `ref`.
-   * @param updates A tensor of updated values to add to `ref`.
-   * @return a new instance of ResourceScatterMax
-   */
-  public <T extends TNumber, U extends TType> ResourceScatterMax resourceScatterMax(
-      Operand<?> resource, Operand<T> indices, Operand<U> updates) {
-    return ResourceScatterMax.create(scope, resource, indices, updates);
-  }
-
-  /**
-   * Reduces sparse updates into the variable referenced by `resource` using the `min` operation.
-   *  <p>
-   *  This operation computes
-   *  <p>
-   *      # Scalar indices
-   *      ref[indices, ...] = min(ref[indices, ...], updates[...])
-   *  <p>
-   *      # Vector indices (for each i)
-   *      ref[indices[i], ...] = min(ref[indices[i], ...], updates[i, ...])
-   *  <p>
-   *      # High rank indices (for each i, ..., j)
-   *      ref[indices[i, ..., j], ...] = min(ref[indices[i, ..., j], ...], updates[i, ..., j, ...])
-   *  <p>
-   *  Duplicate entries are handled correctly: if multiple `indices` reference
-   *  the same location, their contributions are combined.
-   *  <p>
-   *  Requires `updates.shape = indices.shape + ref.shape[1:]` or `updates.shape = []`.
-   *  <p>
-   *  <div style="width:70%; margin:auto; margin-bottom:10px; margin-top:20px;">
-   *  <img style="width:100%" src='https://www.tensorflow.org/images/ScatterAdd.png' alt>
-   *  </div>
-   *
-   * @param resource Should be from a `Variable` node.
-   * @param indices A tensor of indices into the first dimension of `ref`.
-   * @param updates A tensor of updated values to add to `ref`.
-   * @return a new instance of ResourceScatterMin
-   */
-  public <T extends TNumber, U extends TType> ResourceScatterMin resourceScatterMin(
-      Operand<?> resource, Operand<T> indices, Operand<U> updates) {
-    return ResourceScatterMin.create(scope, resource, indices, updates);
-  }
-
-  /**
-   * Multiplies sparse updates into the variable referenced by `resource`.
-   *  <p>
-   *  This operation computes
-   *  <p>
-   *      # Scalar indices
-   *      ref[indices, ...] *= updates[...]
-   *  <p>
-   *      # Vector indices (for each i)
-   *      ref[indices[i], ...] *= updates[i, ...]
-   *  <p>
-   *      # High rank indices (for each i, ..., j)
-   *      ref[indices[i, ..., j], ...] *= updates[i, ..., j, ...]
-   *  <p>
-   *  Duplicate entries are handled correctly: if multiple `indices` reference
-   *  the same location, their contributions multiply.
-   *  <p>
-   *  Requires `updates.shape = indices.shape + ref.shape[1:]` or `updates.shape = []`.
-   *  <p>
-   *  <div style="width:70%; margin:auto; margin-bottom:10px; margin-top:20px;">
-   *  <img style="width:100%" src='https://www.tensorflow.org/images/ScatterAdd.png' alt>
-   *  </div>
-   *
-   * @param resource Should be from a `Variable` node.
-   * @param indices A tensor of indices into the first dimension of `ref`.
-   * @param updates A tensor of updated values to add to `ref`.
-   * @return a new instance of ResourceScatterMul
-   */
-  public <T extends TNumber, U extends TType> ResourceScatterMul resourceScatterMul(
-      Operand<?> resource, Operand<T> indices, Operand<U> updates) {
-    return ResourceScatterMul.create(scope, resource, indices, updates);
-  }
-
-  /**
-   * Applies sparse addition to individual values or slices in a Variable.
-   *  <p>
-   *  `ref` is a `Tensor` with rank `P` and `indices` is a `Tensor` of rank `Q`.
-   *  <p>
-   *  `indices` must be integer tensor, containing indices into `ref`.
-   *  It must be shape `[d_0, ..., d_{Q-2}, K]` where `0 < K <= P`.
-   *  <p>
-   *  The innermost dimension of `indices` (with length `K`) corresponds to
-   *  indices into elements (if `K = P`) or slices (if `K < P`) along the `K`th
-   *  dimension of `ref`.
-   *  <p>
-   *  `updates` is `Tensor` of rank `Q-1+P-K` with shape:
-   *  <pre>{@code
-   *  [d_0, ..., d_{Q-2}, ref.shape[K], ..., ref.shape[P-1]]
-   *  }</pre>
-   *  For example, say we want to add 4 scattered elements to a rank-1 tensor to
-   *  8 elements. In Python, that addition would look like this:
-   *  <pre>{@code
-   *  ref = tf.Variable([1, 2, 3, 4, 5, 6, 7, 8], use_resource=True)
-   *  indices = tf.constant([[4], [3], [1], [7]])
-   *  updates = tf.constant([9, 10, 11, 12])
-   *  add = tf.scatter_nd_add(ref, indices, updates)
-   *  with tf.Session() as sess:
-   *    print sess.run(add)
-   *  }</pre>
-   *  The resulting update to ref would look like this:
-   *  <p>
-   *      [1, 13, 3, 14, 14, 6, 7, 20]
-   *  <p>
-   *  See `tf.scatter_nd` for more details about how to make updates to
-   *  slices.
-   *
-   * @param ref A resource handle. Must be from a VarHandleOp.
-   * @param indices A Tensor. Must be one of the following types: int32, int64.
-   *  A tensor of indices into ref.
-   * @param updates A Tensor. Must have the same type as ref. A tensor of
-   *  values to add to ref.
-   * @param options carries optional attributes values
-   * @return a new instance of ResourceScatterNdAdd
-   */
-  public <T extends TNumber, U extends TType> ResourceScatterNdAdd resourceScatterNdAdd(
-      Operand<?> ref, Operand<T> indices, Operand<U> updates,
-      ResourceScatterNdAdd.Options... options) {
-    return ResourceScatterNdAdd.create(scope, ref, indices, updates, options);
-  }
-
-  /**
-   * Applies sparse subtraction to individual values or slices in a Variable.
-   *  <p>
-   *  `ref` is a `Tensor` with rank `P` and `indices` is a `Tensor` of rank `Q`.
-   *  <p>
-   *  `indices` must be integer tensor, containing indices into `ref`.
-   *  It must be shape `[d_0, ..., d_{Q-2}, K]` where `0 < K <= P`.
-   *  <p>
-   *  The innermost dimension of `indices` (with length `K`) corresponds to
-   *  indices into elements (if `K = P`) or slices (if `K < P`) along the `K`th
-   *  dimension of `ref`.
-   *  <p>
-   *  `updates` is `Tensor` of rank `Q-1+P-K` with shape:
-   *  <pre>{@code
-   *  [d_0, ..., d_{Q-2}, ref.shape[K], ..., ref.shape[P-1]]
-   *  }</pre>
-   *  For example, say we want to subtract 4 scattered elements from a rank-1 tensor
-   *  with 8 elements. In Python, that subtraction would look like this:
-   *  <pre>{@code
-   *  ref = tf.Variable([1, 2, 3, 4, 5, 6, 7, 8], use_resource=True)
-   *  indices = tf.constant([[4], [3], [1], [7]])
-   *  updates = tf.constant([9, 10, 11, 12])
-   *  sub = tf.scatter_nd_sub(ref, indices, updates)
-   *  with tf.Session() as sess:
-   *    print sess.run(sub)
-   *  }</pre>
-   *  The resulting update to ref would look like this:
-   *  <p>
-   *      [1, -9, 3, -6, -4, 6, 7, -4]
-   *  <p>
-   *  See `tf.scatter_nd` for more details about how to make updates to
-   *  slices.
-   *
-   * @param ref A resource handle. Must be from a VarHandleOp.
-   * @param indices A Tensor. Must be one of the following types: int32, int64.
-   *  A tensor of indices into ref.
-   * @param updates A Tensor. Must have the same type as ref. A tensor of
-   *  values to add to ref.
-   * @param options carries optional attributes values
-   * @return a new instance of ResourceScatterNdSub
-   */
-  public <T extends TNumber, U extends TType> ResourceScatterNdSub resourceScatterNdSub(
-      Operand<?> ref, Operand<T> indices, Operand<U> updates,
-      ResourceScatterNdSub.Options... options) {
-    return ResourceScatterNdSub.create(scope, ref, indices, updates, options);
-  }
-
-  /**
-   * Applies sparse `updates` to individual values or slices within a given
-   *  <p>
-   *  variable according to `indices`.
-   *  <p>
-   *  `ref` is a `Tensor` with rank `P` and `indices` is a `Tensor` of rank `Q`.
-   *  <p>
-   *  `indices` must be integer tensor, containing indices into `ref`.
-   *  It must be shape `[d_0, ..., d_{Q-2}, K]` where `0 < K <= P`.
-   *  <p>
-   *  The innermost dimension of `indices` (with length `K`) corresponds to
-   *  indices into elements (if `K = P`) or slices (if `K < P`) along the `K`th
-   *  dimension of `ref`.
-   *  <p>
-   *  `updates` is `Tensor` of rank `Q-1+P-K` with shape:
-   *  <pre>{@code
-   *  [d_0, ..., d_{Q-2}, ref.shape[K], ..., ref.shape[P-1]].
-   *  }</pre>
-   *  For example, say we want to update 4 scattered elements to a rank-1 tensor to
-   *  8 elements. In Python, that update would look like this:
-   *  <pre>{@code
-   *      ref = tf.Variable([1, 2, 3, 4, 5, 6, 7, 8])
-   *      indices = tf.constant([[4], [3], [1] ,[7]])
-   *      updates = tf.constant([9, 10, 11, 12])
-   *      update = tf.scatter_nd_update(ref, indices, updates)
-   *      with tf.Session() as sess:
-   *        print sess.run(update)
-   *  }</pre>
-   *  The resulting update to ref would look like this:
-   *  <p>
-   *      [1, 11, 3, 10, 9, 6, 7, 12]
-   *  <p>
-   *  See `tf.scatter_nd` for more details about how to make updates to
-   *  slices.
-   *
-   * @param ref A resource handle. Must be from a VarHandleOp.
-   * @param indices A Tensor. Must be one of the following types: int32, int64.
-   *  A tensor of indices into ref.
-   * @param updates A Tensor. Must have the same type as ref. A tensor of updated
-   *  values to add to ref.
-   * @param options carries optional attributes values
-   * @return a new instance of ResourceScatterNdUpdate
-   */
-  public <T extends TNumber, U extends TType> ResourceScatterNdUpdate resourceScatterNdUpdate(
-      Operand<?> ref, Operand<T> indices, Operand<U> updates,
-      ResourceScatterNdUpdate.Options... options) {
-    return ResourceScatterNdUpdate.create(scope, ref, indices, updates, options);
-  }
-
-  /**
-   * Subtracts sparse updates from the variable referenced by `resource`.
-   *  <p>
-   *  This operation computes
-   *  <p>
-   *      # Scalar indices
-   *      ref[indices, ...] -= updates[...]
-   *  <p>
-   *      # Vector indices (for each i)
-   *      ref[indices[i], ...] -= updates[i, ...]
-   *  <p>
-   *      # High rank indices (for each i, ..., j)
-   *      ref[indices[i, ..., j], ...] -= updates[i, ..., j, ...]
-   *  <p>
-   *  Duplicate entries are handled correctly: if multiple `indices` reference
-   *  the same location, their contributions add.
-   *  <p>
-   *  Requires `updates.shape = indices.shape + ref.shape[1:]` or `updates.shape = []`.
-   *  <p>
-   *  <div style="width:70%; margin:auto; margin-bottom:10px; margin-top:20px;">
-   *  <img style="width:100%" src='https://www.tensorflow.org/images/ScatterAdd.png' alt>
-   *  </div>
-   *
-   * @param resource Should be from a `Variable` node.
-   * @param indices A tensor of indices into the first dimension of `ref`.
-   * @param updates A tensor of updated values to add to `ref`.
-   * @return a new instance of ResourceScatterSub
-   */
-  public <T extends TNumber, U extends TType> ResourceScatterSub resourceScatterSub(
-      Operand<?> resource, Operand<T> indices, Operand<U> updates) {
-    return ResourceScatterSub.create(scope, resource, indices, updates);
-  }
-
-  /**
-   * Assigns sparse updates to the variable referenced by `resource`.
-   *  <p>
-   *  This operation computes
-   *  <p>
-   *      # Scalar indices
-   *      ref[indices, ...] = updates[...]
-   *  <p>
-   *      # Vector indices (for each i)
-   *      ref[indices[i], ...] = updates[i, ...]
-   *  <p>
-   *      # High rank indices (for each i, ..., j)
-   *      ref[indices[i, ..., j], ...] = updates[i, ..., j, ...]
-   *
-   * @param resource Should be from a `Variable` node.
-   * @param indices A tensor of indices into the first dimension of `ref`.
-   * @param updates A tensor of updated values to add to `ref`.
-   * @return a new instance of ResourceScatterUpdate
-   */
-  public <T extends TNumber, U extends TType> ResourceScatterUpdate resourceScatterUpdate(
-      Operand<?> resource, Operand<T> indices, Operand<U> updates) {
-    return ResourceScatterUpdate.create(scope, resource, indices, updates);
-  }
-
-  /**
-   * Assign `value` to the sliced l-value reference of `ref`.
-   *  <p>
-   *  The values of `value` are assigned to the positions in the variable
-   *  `ref` that are selected by the slice parameters. The slice parameters
-   *  `begin, `end`, `strides`, etc. work exactly as in `StridedSlice`.
-   *  <p>
-   *  NOTE this op currently does not support broadcasting and so `value`'s
-   *  shape must be exactly the shape produced by the slice of `ref`.
-   *
-   * @param ref
-   * @param begin
-   * @param end
-   * @param strides
-   * @param value
-   * @param options carries optional attributes values
-   * @return a new instance of ResourceStridedSliceAssign
-   */
-  public <T extends TNumber, U extends TType> ResourceStridedSliceAssign resourceStridedSliceAssign(
-      Operand<?> ref, Operand<T> begin, Operand<T> end, Operand<T> strides, Operand<U> value,
-      ResourceStridedSliceAssign.Options... options) {
-    return ResourceStridedSliceAssign.create(scope, ref, begin, end, strides, value, options);
-  }
-
-  /**
-   * Reverses specific dimensions of a tensor.
-   *  <p>
-   *  NOTE `tf.reverse` has now changed behavior in preparation for 1.0.
-   *  `tf.reverse_v2` is currently an alias that will be deprecated before TF 1.0.
-   *  <p>
-   *  Given a `tensor`, and a `int32` tensor `axis` representing the set of
-   *  dimensions of `tensor` to reverse. This operation reverses each dimension
-   *  `i` for which there exists `j` s.t. `axis[j] == i`.
-   *  <p>
-   *  `tensor` can have up to 8 dimensions. The number of dimensions specified
-   *  in `axis` may be 0 or more entries. If an index is specified more than
-   *  once, a InvalidArgument error is raised.
-   *  <p>
-   *  For example:
-   *  <pre>{@code
-   *  # tensor 't' is [[[[ 0,  1,  2,  3],
-   *  #                  [ 4,  5,  6,  7],
-   *  #                  [ 8,  9, 10, 11]],
-   *  #                 [[12, 13, 14, 15],
-   *  #                  [16, 17, 18, 19],
-   *  #                  [20, 21, 22, 23]]]]
-   *  # tensor 't' shape is [1, 2, 3, 4]
-   *
-   *  # 'dims' is [3] or 'dims' is [-1]
-   *  reverse(t, dims) ==> [[[[ 3,  2,  1,  0],
-   *                          [ 7,  6,  5,  4],
-   *                          [ 11, 10, 9, 8]],
-   *                         [[15, 14, 13, 12],
-   *                          [19, 18, 17, 16],
-   *                          [23, 22, 21, 20]]]]
-   *
-   *  # 'dims' is '[1]' (or 'dims' is '[-3]')
-   *  reverse(t, dims) ==> [[[[12, 13, 14, 15],
-   *                          [16, 17, 18, 19],
-   *                          [20, 21, 22, 23]
-   *                         [[ 0,  1,  2,  3],
-   *                          [ 4,  5,  6,  7],
-   *                          [ 8,  9, 10, 11]]]]
-   *
-   *  # 'dims' is '[2]' (or 'dims' is '[-2]')
-   *  reverse(t, dims) ==> [[[[8, 9, 10, 11],
-   *                          [4, 5, 6, 7],
-   *                          [0, 1, 2, 3]]
-   *                         [[20, 21, 22, 23],
-   *                          [16, 17, 18, 19],
-   *                          [12, 13, 14, 15]]]]
-   *  }</pre>
-   *
-   * @param <T> data type for {@code output()} output
-   * @param tensor Up to 8-D.
-   * @param axis 1-D. The indices of the dimensions to reverse. Must be in the range
-   *  `[-rank(tensor), rank(tensor))`.
-   * @return a new instance of Reverse
-   */
-  public <T extends TType, U extends TNumber> Reverse<T> reverse(Operand<T> tensor,
-      Operand<U> axis) {
-    return Reverse.create(scope, tensor, axis);
-  }
-
-  /**
-   * Reverses variable length slices.
-   *  <p>
-   *  This op first slices `input` along the dimension `batch_dim`, and for each
-   *  slice `i`, reverses the first `seq_lengths[i]` elements along
-   *  the dimension `seq_dim`.
-   *  <p>
-   *  The elements of `seq_lengths` must obey `seq_lengths[i] <= input.dims[seq_dim]`,
-   *  and `seq_lengths` must be a vector of length `input.dims[batch_dim]`.
-   *  <p>
-   *  The output slice `i` along dimension `batch_dim` is then given by input
-   *  slice `i`, with the first `seq_lengths[i]` slices along dimension
-   *  `seq_dim` reversed.
-   *  <p>
-   *  For example:
-   *  <pre>{@code
-   *  # Given this:
-   *  batch_dim = 0
-   *  seq_dim = 1
-   *  input.dims = (4, 8, ...)
-   *  seq_lengths = [7, 2, 3, 5]
-   *
-   *  # then slices of input are reversed on seq_dim, but only up to seq_lengths:
-   *  output[0, 0:7, :, ...] = input[0, 7:0:-1, :, ...]
-   *  output[1, 0:2, :, ...] = input[1, 2:0:-1, :, ...]
-   *  output[2, 0:3, :, ...] = input[2, 3:0:-1, :, ...]
-   *  output[3, 0:5, :, ...] = input[3, 5:0:-1, :, ...]
-   *
-   *  # while entries past seq_lens are copied through:
-   *  output[0, 7:, :, ...] = input[0, 7:, :, ...]
-   *  output[1, 2:, :, ...] = input[1, 2:, :, ...]
-   *  output[2, 3:, :, ...] = input[2, 3:, :, ...]
-   *  output[3, 2:, :, ...] = input[3, 2:, :, ...]
-   *  }</pre>
-   *  In contrast, if:
-   *  <pre>{@code
-   *  # Given this:
-   *  batch_dim = 2
-   *  seq_dim = 0
-   *  input.dims = (8, ?, 4, ...)
-   *  seq_lengths = [7, 2, 3, 5]
-   *
-   *  # then slices of input are reversed on seq_dim, but only up to seq_lengths:
-   *  output[0:7, :, 0, :, ...] = input[7:0:-1, :, 0, :, ...]
-   *  output[0:2, :, 1, :, ...] = input[2:0:-1, :, 1, :, ...]
-   *  output[0:3, :, 2, :, ...] = input[3:0:-1, :, 2, :, ...]
-   *  output[0:5, :, 3, :, ...] = input[5:0:-1, :, 3, :, ...]
-   *
-   *  # while entries past seq_lens are copied through:
-   *  output[7:, :, 0, :, ...] = input[7:, :, 0, :, ...]
-   *  output[2:, :, 1, :, ...] = input[2:, :, 1, :, ...]
-   *  output[3:, :, 2, :, ...] = input[3:, :, 2, :, ...]
-   *  output[2:, :, 3, :, ...] = input[2:, :, 3, :, ...]
-   *  }</pre>
-   *
-   * @param <T> data type for {@code output()} output
-   * @param input The input to reverse.
-   * @param seqLengths 1-D with length `input.dims(batch_dim)` and
-   *  `max(seq_lengths) <= input.dims(seq_dim)`
-   * @param seqDim The dimension which is partially reversed.
-   * @param options carries optional attributes values
-   * @return a new instance of ReverseSequence
-   */
-  public <T extends TType, U extends TNumber> ReverseSequence<T> reverseSequence(Operand<T> input,
-      Operand<U> seqLengths, Long seqDim, ReverseSequence.Options... options) {
-    return ReverseSequence.create(scope, input, seqLengths, seqDim, options);
-  }
-
-  /**
-   * Rolls the elements of a tensor along an axis.
-   *  <p>
-   *  The elements are shifted positively (towards larger indices) by the offset of
-   *  `shift` along the dimension of `axis`. Negative `shift` values will shift
-   *  elements in the opposite direction. Elements that roll passed the last position
-   *  will wrap around to the first and vice versa. Multiple shifts along multiple
-   *  axes may be specified.
-   *  <p>
-   *  For example:
-   *  <pre>{@code
-   *  # 't' is [0, 1, 2, 3, 4]
-   *  roll(t, shift=2, axis=0) ==> [3, 4, 0, 1, 2]
-   *
-   *  # shifting along multiple dimensions
-   *  # 't' is [[0, 1, 2, 3, 4], [5, 6, 7, 8, 9]]
-   *  roll(t, shift=[1, -2], axis=[0, 1]) ==> [[7, 8, 9, 5, 6], [2, 3, 4, 0, 1]]
-   *
-   *  # shifting along the same axis multiple times
-   *  # 't' is [[0, 1, 2, 3, 4], [5, 6, 7, 8, 9]]
-   *  roll(t, shift=[2, -3], axis=[1, 1]) ==> [[1, 2, 3, 4, 0], [6, 7, 8, 9, 5]]
-   *  }</pre>
-   *
-   * @param <T> data type for {@code output()} output
-   * @param input
-   * @param shift Dimension must be 0-D or 1-D. `shift[i]` specifies the number of places by which
-   *  elements are shifted positively (towards larger indices) along the dimension
-   *  specified by `axis[i]`. Negative shifts will roll the elements in the opposite
-   *  direction.
-   * @param axis Dimension must be 0-D or 1-D. `axis[i]` specifies the dimension that the shift
-   *  `shift[i]` should occur. If the same axis is referenced more than once, the
-   *  total shift for that axis will be the sum of all the shifts that belong to that
-   *  axis.
-   * @return a new instance of Roll
-   */
-  public <T extends TType, U extends TNumber, V extends TNumber> Roll<T> roll(Operand<T> input,
-      Operand<U> shift, Operand<V> axis) {
-    return Roll.create(scope, input, shift, axis);
-  }
-
-  /**
-   * Perform batches of RPC requests.
-   *  <p>
-   *  This op asynchronously performs either a single RPC request, or a batch
-   *  of requests.  RPC requests are defined by three main parameters:
-   *  <p>
-   *    - `address` (the host+port or BNS address of the request)
-   *    - `method` (the RPC method name for the request)
-   *    - `request` (the serialized proto string, or vector of strings,
-   *       of the RPC request argument).
-   *  <p>
-   *  For example, if you have an RPC service running on port localhost:2345,
-   *  and its interface is configured with the following proto declaration:
-   *  <pre>{@code
-   *  service MyService {
-   *    rpc MyMethod(MyRequestProto) returns (MyResponseProto) {
-   *    }
-   *  };
-   *  }</pre>
-   *  then call this op with arguments:
-   *  <pre>{@code
-   *  address = "localhost:2345"
-   *  method = "MyService/MyMethod"
-   *  }</pre>
-   *  The `request` tensor is a string tensor representing serialized `MyRequestProto`
-   *  strings; and the output string tensor `response` will have the same shape
-   *  and contain (upon successful completion) corresponding serialized
-   *  `MyResponseProto` strings.
-   *  <p>
-   *  For example, to send a single, empty, `MyRequestProto`, call
-   *  this op with `request = ""`.  To send 5 <b>parallel</b> empty requests,
-   *  call this op with `request = ["", "", "", "", ""]`.
-   *  <p>
-   *  More generally, one can create a batch of `MyRequestProto` serialized protos
-   *  from regular batched tensors using the `encode_proto` op, and convert
-   *  the response `MyResponseProto` serialized protos to batched tensors
-   *  using the `decode_proto` op.
-   *  <p>
-   *  <b>NOTE</b> Working with serialized proto strings is faster than instantiating
-   *  actual proto objects in memory, so no performance degradation is expected
-   *  compared to writing custom kernels for this workflow.
-   *  <p>
-   *  If the connection fails or the remote worker returns an error
-   *  status, the op reraises this exception locally.
-   *  <p>
-   *  See the `TryRpc` op if you prefer to handle RPC failures manually in the graph.
-   *
-   * @param address `0-D` or `1-D`.  The address (i.e. host_name:port) of the RPC server.
-   *  If this tensor has more than 1 element, then multiple parallel rpc requests
-   *  are sent.  This argument broadcasts with `method` and `request`.
-   * @param method `0-D` or `1-D`.  The method address on the RPC server.
-   *  If this tensor has more than 1 element, then multiple parallel rpc requests
-   *  are sent.  This argument broadcasts with `address` and `request`.
-   * @param request `0-D` or `1-D`.  Serialized proto strings: the rpc request argument.
-   *  If this tensor has more than 1 element, then multiple parallel rpc requests
-   *  are sent.  This argument broadcasts with `address` and `method`.
-   * @param options carries optional attributes values
-   * @return a new instance of Rpc
-   */
-  public Rpc rpc(Operand<TString> address, Operand<TString> method, Operand<TString> request,
-      Rpc.Options... options) {
-    return Rpc.create(scope, address, method, request, options);
-  }
-
-  /**
-   * Adds sparse updates to a variable reference.
-   *  <p>
-   *  This operation computes
-   *  <p>
-   *      # Scalar indices
-   *      ref[indices, ...] += updates[...]
-   *  <p>
-   *      # Vector indices (for each i)
-   *      ref[indices[i], ...] += updates[i, ...]
-   *  <p>
-   *      # High rank indices (for each i, ..., j)
-   *      ref[indices[i, ..., j], ...] += updates[i, ..., j, ...]
-   *  <p>
-   *  This operation outputs `ref` after the update is done.
-   *  This makes it easier to chain operations that need to use the reset value.
-   *  <p>
-   *  Duplicate entries are handled correctly: if multiple `indices` reference
-   *  the same location, their contributions add.
-   *  <p>
-   *  Requires `updates.shape = indices.shape + ref.shape[1:]` or `updates.shape = []`.
-   *  <p>
-   *  <div style="width:70%; margin:auto; margin-bottom:10px; margin-top:20px;">
-   *  <img style="width:100%" src="https://www.tensorflow.org/images/ScatterAdd.png" alt>
-   *  </div>
-   *
-   * @param <T> data type for {@code outputRef()} output
-   * @param ref Should be from a `Variable` node.
-   * @param indices A tensor of indices into the first dimension of `ref`.
-   * @param updates A tensor of updated values to add to `ref`.
-   * @param options carries optional attributes values
-   * @return a new instance of ScatterAdd
-   */
-  public <T extends TType, U extends TNumber> ScatterAdd<T> scatterAdd(Operand<T> ref,
-      Operand<U> indices, Operand<T> updates, ScatterAdd.Options... options) {
-    return ScatterAdd.create(scope, ref, indices, updates, options);
-  }
-
-  /**
-   * Divides a variable reference by sparse updates.
-   *  <p>
-   *  This operation computes
-   *  <pre>{@code
-   *      # Scalar indices
-   *      ref[indices, ...] /= updates[...]
-   *
-   *      # Vector indices (for each i)
-   *      ref[indices[i], ...] /= updates[i, ...]
-   *
-   *      # High rank indices (for each i, ..., j)
-   *      ref[indices[i, ..., j], ...] /= updates[i, ..., j, ...]
-   *  }</pre>
-   *  This operation outputs `ref` after the update is done.
-   *  This makes it easier to chain operations that need to use the reset value.
-   *  <p>
-   *  Duplicate entries are handled correctly: if multiple `indices` reference
-   *  the same location, their contributions divide.
-   *  <p>
-   *  Requires `updates.shape = indices.shape + ref.shape[1:]` or `updates.shape = []`.
-   *
-   * @param <T> data type for {@code outputRef()} output
-   * @param ref Should be from a `Variable` node.
-   * @param indices A tensor of indices into the first dimension of `ref`.
-   * @param updates A tensor of values that `ref` is divided by.
-   * @param options carries optional attributes values
-   * @return a new instance of ScatterDiv
-   */
-  public <T extends TType, U extends TNumber> ScatterDiv<T> scatterDiv(Operand<T> ref,
-      Operand<U> indices, Operand<T> updates, ScatterDiv.Options... options) {
-    return ScatterDiv.create(scope, ref, indices, updates, options);
-  }
-
-  /**
-   * Reduces sparse updates into a variable reference using the `max` operation.
-   *  <p>
-   *  This operation computes
-   *  <p>
-   *      # Scalar indices
-   *      ref[indices, ...] = max(ref[indices, ...], updates[...])
-   *  <p>
-   *      # Vector indices (for each i)
-   *      ref[indices[i], ...] = max(ref[indices[i], ...], updates[i, ...])
-   *  <p>
-   *      # High rank indices (for each i, ..., j)
-   *      ref[indices[i, ..., j], ...] = max(ref[indices[i, ..., j], ...], updates[i, ..., j, ...])
-   *  <p>
-   *  This operation outputs `ref` after the update is done.
-   *  This makes it easier to chain operations that need to use the reset value.
-   *  <p>
-   *  Duplicate entries are handled correctly: if multiple `indices` reference
-   *  the same location, their contributions combine.
-   *  <p>
-   *  Requires `updates.shape = indices.shape + ref.shape[1:]` or `updates.shape = []`.
-   *  <p>
-   *  <div style="width:70%; margin:auto; margin-bottom:10px; margin-top:20px;">
-   *  <img style="width:100%" src="https://www.tensorflow.org/images/ScatterAdd.png" alt>
-   *  </div>
-   *
-   * @param <T> data type for {@code outputRef()} output
-   * @param ref Should be from a `Variable` node.
-   * @param indices A tensor of indices into the first dimension of `ref`.
-   * @param updates A tensor of updated values to reduce into `ref`.
-   * @param options carries optional attributes values
-   * @return a new instance of ScatterMax
-   */
-  public <T extends TNumber, U extends TNumber> ScatterMax<T> scatterMax(Operand<T> ref,
-      Operand<U> indices, Operand<T> updates, ScatterMax.Options... options) {
-    return ScatterMax.create(scope, ref, indices, updates, options);
-  }
-
-  /**
-   * Reduces sparse updates into a variable reference using the `min` operation.
-   *  <p>
-   *  This operation computes
-   *  <p>
-   *      # Scalar indices
-   *      ref[indices, ...] = min(ref[indices, ...], updates[...])
-   *  <p>
-   *      # Vector indices (for each i)
-   *      ref[indices[i], ...] = min(ref[indices[i], ...], updates[i, ...])
-   *  <p>
-   *      # High rank indices (for each i, ..., j)
-   *      ref[indices[i, ..., j], ...] = min(ref[indices[i, ..., j], ...], updates[i, ..., j, ...])
-   *  <p>
-   *  This operation outputs `ref` after the update is done.
-   *  This makes it easier to chain operations that need to use the reset value.
-   *  <p>
-   *  Duplicate entries are handled correctly: if multiple `indices` reference
-   *  the same location, their contributions combine.
-   *  <p>
-   *  Requires `updates.shape = indices.shape + ref.shape[1:]` or `updates.shape = []`.
-   *  <p>
-   *  <div style="width:70%; margin:auto; margin-bottom:10px; margin-top:20px;">
-   *  <img style="width:100%" src="https://www.tensorflow.org/images/ScatterAdd.png" alt>
-   *  </div>
-   *
-   * @param <T> data type for {@code outputRef()} output
-   * @param ref Should be from a `Variable` node.
-   * @param indices A tensor of indices into the first dimension of `ref`.
-   * @param updates A tensor of updated values to reduce into `ref`.
-   * @param options carries optional attributes values
-   * @return a new instance of ScatterMin
-   */
-  public <T extends TNumber, U extends TNumber> ScatterMin<T> scatterMin(Operand<T> ref,
-      Operand<U> indices, Operand<T> updates, ScatterMin.Options... options) {
-    return ScatterMin.create(scope, ref, indices, updates, options);
-  }
-
-  /**
-   * Multiplies sparse updates into a variable reference.
-   *  <p>
-   *  This operation computes
-   *  <pre>{@code
-   *      # Scalar indices
-   *      ref[indices, ...] *= updates[...]
-   *
-   *      # Vector indices (for each i)
-   *      ref[indices[i], ...] *= updates[i, ...]
-   *
-   *      # High rank indices (for each i, ..., j)
-   *      ref[indices[i, ..., j], ...] *= updates[i, ..., j, ...]
-   *  }</pre>
-   *  This operation outputs `ref` after the update is done.
-   *  This makes it easier to chain operations that need to use the reset value.
-   *  <p>
-   *  Duplicate entries are handled correctly: if multiple `indices` reference
-   *  the same location, their contributions multiply.
-   *  <p>
-   *  Requires `updates.shape = indices.shape + ref.shape[1:]` or `updates.shape = []`.
-   *
-   * @param <T> data type for {@code outputRef()} output
-   * @param ref Should be from a `Variable` node.
-   * @param indices A tensor of indices into the first dimension of `ref`.
-   * @param updates A tensor of updated values to multiply to `ref`.
-   * @param options carries optional attributes values
-   * @return a new instance of ScatterMul
-   */
-  public <T extends TType, U extends TNumber> ScatterMul<T> scatterMul(Operand<T> ref,
-      Operand<U> indices, Operand<T> updates, ScatterMul.Options... options) {
-    return ScatterMul.create(scope, ref, indices, updates, options);
-  }
-
-  /**
-   * Scatter `updates` into a new tensor according to `indices`.
-   *  <p>
-   *  Creates a new tensor by applying sparse `updates` to individual values or
-   *  slices within a tensor (initially zero for numeric, empty for string) of
-   *  the given `shape` according to indices.  This operator is the inverse of the
-   *  `tf.gather_nd` operator which extracts values or slices from a given tensor.
-   *  <p>
-   *  This operation is similar to tensor_scatter_add, except that the tensor is
-   *  zero-initialized. Calling `tf.scatter_nd(indices, values, shape)` is identical
-   *  to `tensor_scatter_add(tf.zeros(shape, values.dtype), indices, values)`
-   *  <p>
-   *  If `indices` contains duplicates, then their updates are accumulated (summed).
-   *  <p>
-   *  <b>WARNING</b>: The order in which updates are applied is nondeterministic, so the
-   *  output will be nondeterministic if `indices` contains duplicates -- because
-   *  of some numerical approximation issues, numbers summed in different order
-   *  may yield different results.
-   *  <p>
-   *  `indices` is an integer tensor containing indices into a new tensor of shape
-   *  `shape`.  The last dimension of `indices` can be at most the rank of `shape`:
-   *  <p>
-   *      indices.shape[-1] <= shape.rank
-   *  <p>
-   *  The last dimension of `indices` corresponds to indices into elements
-   *  (if `indices.shape[-1] = shape.rank`) or slices
-   *  (if `indices.shape[-1] < shape.rank`) along dimension `indices.shape[-1]` of
-   *  `shape`.  `updates` is a tensor with shape
-   *  <p>
-   *      indices.shape[:-1] + shape[indices.shape[-1]:]
-   *  <p>
-   *  The simplest form of scatter is to insert individual elements in a tensor by
-   *  index. For example, say we want to insert 4 scattered elements in a rank-1
-   *  tensor with 8 elements.
-   *  <p>
-   *  <div style="width:70%; margin:auto; margin-bottom:10px; margin-top:20px;">
-   *  <img style="width:100%" src="https://www.tensorflow.org/images/ScatterNd1.png" alt>
-   *  </div>
-   *  <p>
-   *  In Python, this scatter operation would look like this:
-   *  <pre>{@code
-   *      indices = tf.constant([[4], [3], [1], [7]])
-   *      updates = tf.constant([9, 10, 11, 12])
-   *      shape = tf.constant([8])
-   *      scatter = tf.scatter_nd(indices, updates, shape)
-   *      print(scatter)
-   *  }</pre>
-   *  The resulting tensor would look like this:
-   *  <p>
-   *      [0, 11, 0, 10, 9, 0, 0, 12]
-   *  <p>
-   *  We can also, insert entire slices of a higher rank tensor all at once. For
-   *  example, if we wanted to insert two slices in the first dimension of a
-   *  rank-3 tensor with two matrices of new values.
-   *  <p>
-   *  <div style="width:70%; margin:auto; margin-bottom:10px; margin-top:20px;">
-   *  <img style="width:100%" src="https://www.tensorflow.org/images/ScatterNd2.png" alt>
-   *  </div>
-   *  <p>
-   *  In Python, this scatter operation would look like this:
-   *  <pre>{@code
-   *      indices = tf.constant([[0], [2]])
-   *      updates = tf.constant([[[5, 5, 5, 5], [6, 6, 6, 6],
-   *                              [7, 7, 7, 7], [8, 8, 8, 8]],
-   *                             [[5, 5, 5, 5], [6, 6, 6, 6],
-   *                              [7, 7, 7, 7], [8, 8, 8, 8]]])
-   *      shape = tf.constant([4, 4, 4])
-   *      scatter = tf.scatter_nd(indices, updates, shape)
-   *      print(scatter)
-   *  }</pre>
-   *  The resulting tensor would look like this:
-   *  <p>
-   *      [[[5, 5, 5, 5], [6, 6, 6, 6], [7, 7, 7, 7], [8, 8, 8, 8]],
-   *       [[0, 0, 0, 0], [0, 0, 0, 0], [0, 0, 0, 0], [0, 0, 0, 0]],
-   *       [[5, 5, 5, 5], [6, 6, 6, 6], [7, 7, 7, 7], [8, 8, 8, 8]],
-   *       [[0, 0, 0, 0], [0, 0, 0, 0], [0, 0, 0, 0], [0, 0, 0, 0]]]
-   *  <p>
-   *  Note that on CPU, if an out of bound index is found, an error is returned.
-   *  On GPU, if an out of bound index is found, the index is ignored.
-   *
-   * @param <U> data type for {@code output()} output
-   * @param indices Index tensor.
-   * @param updates Updates to scatter into output.
-   * @param shape 1-D. The shape of the resulting tensor.
-   * @return a new instance of ScatterNd
-   */
-  public <U extends TType, T extends TNumber> ScatterNd<U> scatterNd(Operand<T> indices,
-      Operand<U> updates, Operand<T> shape) {
-    return ScatterNd.create(scope, indices, updates, shape);
-  }
-
-  /**
-   * Applies sparse addition to individual values or slices in a Variable.
-   *  <p>
-   *  `ref` is a `Tensor` with rank `P` and `indices` is a `Tensor` of rank `Q`.
-   *  <p>
-   *  `indices` must be integer tensor, containing indices into `ref`.
-   *  It must be shape `[d_0, ..., d_{Q-2}, K]` where `0 < K <= P`.
-   *  <p>
-   *  The innermost dimension of `indices` (with length `K`) corresponds to
-   *  indices into elements (if `K = P`) or slices (if `K < P`) along the `K`th
-   *  dimension of `ref`.
-   *  <p>
-   *  `updates` is `Tensor` of rank `Q-1+P-K` with shape:
-   *  <pre>{@code
-   *  [d_0, ..., d_{Q-2}, ref.shape[K], ..., ref.shape[P-1]]
-   *  }</pre>
-   *  For example, say we want to add 4 scattered elements to a rank-1 tensor to
-   *  8 elements. In Python, that addition would look like this:
-   *  <pre>{@code
-   *  ref = tf.Variable([1, 2, 3, 4, 5, 6, 7, 8])
-   *  indices = tf.constant([[4], [3], [1], [7]])
-   *  updates = tf.constant([9, 10, 11, 12])
-   *  add = tf.scatter_nd_add(ref, indices, updates)
-   *  with tf.Session() as sess:
-   *    print sess.run(add)
-   *  }</pre>
-   *  The resulting update to ref would look like this:
-   *  <p>
-   *      [1, 13, 3, 14, 14, 6, 7, 20]
-   *  <p>
-   *  See `tf.scatter_nd` for more details about how to make updates to
-   *  slices.
-   *
-   * @param <T> data type for {@code outputRef()} output
-   * @param ref A mutable Tensor. Should be from a Variable node.
-   * @param indices A Tensor. Must be one of the following types: int32, int64.
-   *  A tensor of indices into ref.
-   * @param updates A Tensor. Must have the same type as ref. A tensor of updated values
-   *  to add to ref.
-   * @param options carries optional attributes values
-   * @return a new instance of ScatterNdAdd
-   */
-  public <T extends TType, U extends TNumber> ScatterNdAdd<T> scatterNdAdd(Operand<T> ref,
-      Operand<U> indices, Operand<T> updates, ScatterNdAdd.Options... options) {
-    return ScatterNdAdd.create(scope, ref, indices, updates, options);
-  }
-
-  /**
-   * Applies sparse addition to `input` using individual values or slices
-   *  <p>
-   *  from `updates` according to indices `indices`.  The updates are non-aliasing:
-   *  `input` is only modified in-place if no other operations will use it.
-   *  Otherwise, a copy of `input` is made.  This operation has a gradient with
-   *  respect to both `input` and `updates`.
-   *  <p>
-   *  `input` is a `Tensor` with rank `P` and `indices` is a `Tensor` of rank `Q`.
-   *  <p>
-   *  `indices` must be integer tensor, containing indices into `input`.
-   *  It must be shape \\([d_0, ..., d_{Q-2}, K]\\) where `0 < K <= P`.
-   *  <p>
-   *  The innermost dimension of `indices` (with length `K`) corresponds to
-   *  indices into elements (if `K = P`) or `(P-K)`-dimensional slices
-   *  (if `K < P`) along the `K`th dimension of `input`.
-   *  <p>
-   *  `updates` is `Tensor` of rank `Q-1+P-K` with shape:
-   *  <p>
-   *  $$[d_0, ..., d_{Q-2}, input.shape[K], ..., input.shape[P-1]].$$
-   *  <p>
-   *  For example, say we want to add 4 scattered elements to a rank-1 tensor to 8
-   *  elements. In Python, that addition would look like this:
-   *  <p>
-   *      input = tf.constant([1, 2, 3, 4, 5, 6, 7, 8])
-   *      indices = tf.constant([[4], [3], [1], [7]])
-   *      updates = tf.constant([9, 10, 11, 12])
-   *      output = tf.scatter_nd_non_aliasing_add(input, indices, updates)
-   *      with tf.Session() as sess:
-   *        print(sess.run(output))
-   *  <p>
-   *  The resulting value `output` would look like this:
-   *  <p>
-   *      [1, 13, 3, 14, 14, 6, 7, 20]
-   *  <p>
-   *  See `tf.scatter_nd` for more details about how to make updates to slices.
-   *
-   * @param <T> data type for {@code output()} output
-   * @param input A Tensor.
-   * @param indices A Tensor. Must be one of the following types: `int32`, `int64`.
-   *  A tensor of indices into `input`.
-   * @param updates A Tensor. Must have the same type as ref. A tensor of updated values
-   *  to add to `input`.
-   * @return a new instance of ScatterNdNonAliasingAdd
-   */
-  public <T extends TType, U extends TNumber> ScatterNdNonAliasingAdd<T> scatterNdNonAliasingAdd(
-      Operand<T> input, Operand<U> indices, Operand<T> updates) {
-    return ScatterNdNonAliasingAdd.create(scope, input, indices, updates);
-  }
-
-  /**
-   * Applies sparse subtraction to individual values or slices in a Variable.
-   *  <p>
-   *  within a given variable according to `indices`.
-   *  <p>
-   *  `ref` is a `Tensor` with rank `P` and `indices` is a `Tensor` of rank `Q`.
-   *  <p>
-   *  `indices` must be integer tensor, containing indices into `ref`.
-   *  It must be shape `[d_0, ..., d_{Q-2}, K]` where `0 < K <= P`.
-   *  <p>
-   *  The innermost dimension of `indices` (with length `K`) corresponds to
-   *  indices into elements (if `K = P`) or slices (if `K < P`) along the `K`th
-   *  dimension of `ref`.
-   *  <p>
-   *  `updates` is `Tensor` of rank `Q-1+P-K` with shape:
-   *  <pre>{@code
-   *  [d_0, ..., d_{Q-2}, ref.shape[K], ..., ref.shape[P-1]]
-   *  }</pre>
-   *  For example, say we want to subtract 4 scattered elements from a rank-1 tensor
-   *  with 8 elements. In Python, that subtraction would look like this:
-   *  <pre>{@code
-   *  ref = tf.Variable([1, 2, 3, 4, 5, 6, 7, 8])
-   *  indices = tf.constant([[4], [3], [1], [7]])
-   *  updates = tf.constant([9, 10, 11, 12])
-   *  sub = tf.scatter_nd_sub(ref, indices, updates)
-   *  with tf.Session() as sess:
-   *    print sess.run(sub)
-   *  }</pre>
-   *  The resulting update to ref would look like this:
-   *  <p>
-   *      [1, -9, 3, -6, -4, 6, 7, -4]
-   *  <p>
-   *  See `tf.scatter_nd` for more details about how to make updates to
-   *  slices.
-   *
-   * @param <T> data type for {@code outputRef()} output
-   * @param ref A mutable Tensor. Should be from a Variable node.
-   * @param indices A Tensor. Must be one of the following types: int32, int64.
-   *  A tensor of indices into ref.
-   * @param updates A Tensor. Must have the same type as ref. A tensor of updated values
-   *  to subtract from ref.
-   * @param options carries optional attributes values
-   * @return a new instance of ScatterNdSub
-   */
-  public <T extends TType, U extends TNumber> ScatterNdSub<T> scatterNdSub(Operand<T> ref,
-      Operand<U> indices, Operand<T> updates, ScatterNdSub.Options... options) {
-    return ScatterNdSub.create(scope, ref, indices, updates, options);
-  }
-
-  /**
-   * Applies sparse `updates` to individual values or slices within a given
-   *  <p>
-   *  variable according to `indices`.
-   *  <p>
-   *  `ref` is a `Tensor` with rank `P` and `indices` is a `Tensor` of rank `Q`.
-   *  <p>
-   *  `indices` must be integer tensor, containing indices into `ref`.
-   *  It must be shape \\([d_0, ..., d_{Q-2}, K]\\) where `0 < K <= P`.
-   *  <p>
-   *  The innermost dimension of `indices` (with length `K`) corresponds to
-   *  indices into elements (if `K = P`) or slices (if `K < P`) along the `K`th
-   *  dimension of `ref`.
-   *  <p>
-   *  `updates` is `Tensor` of rank `Q-1+P-K` with shape:
-   *  <p>
-   *  $$[d_0, ..., d_{Q-2}, ref.shape[K], ..., ref.shape[P-1]].$$
-   *  <p>
-   *  For example, say we want to update 4 scattered elements to a rank-1 tensor to
-   *  8 elements. In Python, that update would look like this:
-   *  <pre>{@code
-   *      ref = tf.Variable([1, 2, 3, 4, 5, 6, 7, 8])
-   *      indices = tf.constant([[4], [3], [1] ,[7]])
-   *      updates = tf.constant([9, 10, 11, 12])
-   *      update = tf.scatter_nd_update(ref, indices, updates)
-   *      with tf.Session() as sess:
-   *        print sess.run(update)
-   *  }</pre>
-   *  The resulting update to ref would look like this:
-   *  <p>
-   *      [1, 11, 3, 10, 9, 6, 7, 12]
-   *  <p>
-   *  See `tf.scatter_nd` for more details about how to make updates to
-   *  slices.
-   *  <p>
-   *  See also `tf.scatter_update` and `tf.batch_scatter_update`.
-   *
-   * @param <T> data type for {@code outputRef()} output
-   * @param ref A mutable Tensor. Should be from a Variable node.
-   * @param indices A Tensor. Must be one of the following types: int32, int64.
-   *  A tensor of indices into ref.
-   * @param updates A Tensor. Must have the same type as ref. A tensor of updated
-   *  values to add to ref.
-   * @param options carries optional attributes values
-   * @return a new instance of ScatterNdUpdate
-   */
-  public <T extends TType, U extends TNumber> ScatterNdUpdate<T> scatterNdUpdate(Operand<T> ref,
-      Operand<U> indices, Operand<T> updates, ScatterNdUpdate.Options... options) {
-    return ScatterNdUpdate.create(scope, ref, indices, updates, options);
-  }
-
-  /**
-   * Subtracts sparse updates to a variable reference.
-   *  <p>
-   *  <pre>{@code
-   *      # Scalar indices
-   *      ref[indices, ...] -= updates[...]
-   *
-   *      # Vector indices (for each i)
-   *      ref[indices[i], ...] -= updates[i, ...]
-   *
-   *      # High rank indices (for each i, ..., j)
-   *      ref[indices[i, ..., j], ...] -= updates[i, ..., j, ...]
-   *  }</pre>
-   *  This operation outputs `ref` after the update is done.
-   *  This makes it easier to chain operations that need to use the reset value.
-   *  <p>
-   *  Duplicate entries are handled correctly: if multiple `indices` reference
-   *  the same location, their (negated) contributions add.
-   *  <p>
-   *  Requires `updates.shape = indices.shape + ref.shape[1:]` or `updates.shape = []`.
-   *  <p>
-   *  <div style="width:70%; margin:auto; margin-bottom:10px; margin-top:20px;">
-   *  <img style="width:100%" src="https://www.tensorflow.org/images/ScatterSub.png" alt>
-   *  </div>
-   *
-   * @param <T> data type for {@code outputRef()} output
-   * @param ref Should be from a `Variable` node.
-   * @param indices A tensor of indices into the first dimension of `ref`.
-   * @param updates A tensor of updated values to subtract from `ref`.
-   * @param options carries optional attributes values
-   * @return a new instance of ScatterSub
-   */
-  public <T extends TType, U extends TNumber> ScatterSub<T> scatterSub(Operand<T> ref,
-      Operand<U> indices, Operand<T> updates, ScatterSub.Options... options) {
-    return ScatterSub.create(scope, ref, indices, updates, options);
-  }
-
-  /**
-   * Applies sparse updates to a variable reference.
-   *  <p>
-   *  This operation computes
-   *  <pre>{@code
-   *      # Scalar indices
-   *      ref[indices, ...] = updates[...]
-   *
-   *      # Vector indices (for each i)
-   *      ref[indices[i], ...] = updates[i, ...]
-   *
-   *      # High rank indices (for each i, ..., j)
-   *      ref[indices[i, ..., j], ...] = updates[i, ..., j, ...]
-   *  }</pre>
-   *  This operation outputs `ref` after the update is done.
-   *  This makes it easier to chain operations that need to use the reset value.
-   *  <p>
-   *  If values in `ref` is to be updated more than once, because there are
-   *  duplicate entries in `indices`, the order at which the updates happen
-   *  for each value is undefined.
-   *  <p>
-   *  Requires `updates.shape = indices.shape + ref.shape[1:]` or `updates.shape = []`.
-   *  <p>
-   *  <div style="width:70%; margin:auto; margin-bottom:10px; margin-top:20px;">
-   *  <img style="width:100%" src="https://www.tensorflow.org/images/ScatterUpdate.png" alt>
-   *  </div>
-   *  <p>
-   *  See also `tf.batch_scatter_update` and `tf.scatter_nd_update`.
-   *
-   * @param <T> data type for {@code outputRef()} output
-   * @param ref Should be from a `Variable` node.
-   * @param indices A tensor of indices into the first dimension of `ref`.
-   * @param updates A tensor of updated values to store in `ref`.
-   * @param options carries optional attributes values
-   * @return a new instance of ScatterUpdate
-   */
-  public <T extends TType, U extends TNumber> ScatterUpdate<T> scatterUpdate(Operand<T> ref,
-      Operand<U> indices, Operand<T> updates, ScatterUpdate.Options... options) {
-    return ScatterUpdate.create(scope, ref, indices, updates, options);
-  }
-
-  /**
-   *
-   * @param <T> data type for {@code output()} output
-   * @param condition
-   * @param t
-   * @param e
-   * @return a new instance of Select
-   */
-  public <T extends TType> Select<T> select(Operand<TBool> condition, Operand<T> t, Operand<T> e) {
-    return Select.create(scope, condition, t, e);
-  }
-
-  /**
-   * Computes the difference between two lists of numbers or strings.
-   *  <p>
-   *  Given a list `x` and a list `y`, this operation returns a list `out` that
-   *  represents all values that are in `x` but not in `y`. The returned list `out`
-   *  is sorted in the same order that the numbers appear in `x` (duplicates are
-   *  preserved). This operation also returns a list `idx` that represents the
-   *  position of each `out` element in `x`. In other words:
-   *  <p>
-   *  `out[i] = x[idx[i]] for i in [0, 1, ..., len(out) - 1]`
-   *  <p>
-   *  For example, given this input:
-   *  <pre>{@code
-   *  x = [1, 2, 3, 4, 5, 6]
-   *  y = [1, 3, 5]
-   *  }</pre>
-   *  This operation would return:
-   *  <pre>{@code
-   *  out ==> [2, 4, 6]
-   *  idx ==> [1, 3, 5]
-   *  }</pre>
-   *
-   * @param <T> data type for {@code out()} output
-   * @param <U> data type for {@code idx()} output
-   * @param x 1-D. Values to keep.
-   * @param y 1-D. Values to remove.
-   * @return a new instance of SetDiff1d
-   */
-  public <T extends TType> SetDiff1d<T, TInt32> setDiff1d(Operand<T> x, Operand<T> y) {
-    return SetDiff1d.create(scope, x, y);
-  }
-
-  /**
-   * Computes the difference between two lists of numbers or strings.
-   *  <p>
-   *  Given a list `x` and a list `y`, this operation returns a list `out` that
-   *  represents all values that are in `x` but not in `y`. The returned list `out`
-   *  is sorted in the same order that the numbers appear in `x` (duplicates are
-   *  preserved). This operation also returns a list `idx` that represents the
-   *  position of each `out` element in `x`. In other words:
-   *  <p>
-   *  `out[i] = x[idx[i]] for i in [0, 1, ..., len(out) - 1]`
-   *  <p>
-   *  For example, given this input:
-   *  <pre>{@code
-   *  x = [1, 2, 3, 4, 5, 6]
-   *  y = [1, 3, 5]
-   *  }</pre>
-   *  This operation would return:
-   *  <pre>{@code
-   *  out ==> [2, 4, 6]
-   *  idx ==> [1, 3, 5]
-   *  }</pre>
-   *
-   * @param <T> data type for {@code out()} output
-   * @param <U> data type for {@code idx()} output
-   * @param x 1-D. Values to keep.
-   * @param y 1-D. Values to remove.
-   * @param outIdx
-   * @return a new instance of SetDiff1d
-   */
-  public <T extends TType, U extends TNumber> SetDiff1d<T, U> setDiff1d(Operand<T> x, Operand<T> y,
-      DataType<U> outIdx) {
-    return SetDiff1d.create(scope, x, y, outIdx);
-  }
-
-  /**
-   * Number of unique elements along last dimension of input `set`.
-   *  <p>
-   *  Input `set` is a `SparseTensor` represented by `set_indices`, `set_values`,
-   *  and `set_shape`. The last dimension contains values in a set, duplicates are
-   *  allowed but ignored.
-   *  <p>
-   *  If `validate_indices` is `True`, this op validates the order and range of `set`
-   *  indices.
-   *
-   * @param setIndices 2D `Tensor`, indices of a `SparseTensor`.
-   * @param setValues 1D `Tensor`, values of a `SparseTensor`.
-   * @param setShape 1D `Tensor`, shape of a `SparseTensor`.
-   * @param options carries optional attributes values
-   * @return a new instance of SetSize
-   */
-  public <T extends TType> SetSize setSize(Operand<TInt64> setIndices, Operand<T> setValues,
-      Operand<TInt64> setShape, SetSize.Options... options) {
-    return SetSize.create(scope, setIndices, setValues, setShape, options);
-  }
-
-  /**
-   * Returns the shape of a tensor.
-   *  <p>
-   *  This operation returns a 1-D integer tensor representing the shape of `input`.
-   *  <p>
-   *  For example:
-   *  <pre>{@code
-   *  # 't' is [[[1, 1, 1], [2, 2, 2]], [[3, 3, 3], [4, 4, 4]]]
-   *  shape(t) ==> [2, 2, 3]
-   *  }</pre>
-   *
-   * @param <U> data type for {@code output()} output
-   * @param input
-   * @return a new instance of Shape
-   */
-  public <T extends TType> org.tensorflow.op.core.Shape<TInt32> shape(Operand<T> input) {
-    return org.tensorflow.op.core.Shape.create(scope, input);
-  }
-
-  /**
-   * Returns the shape of a tensor.
-   *  <p>
-   *  This operation returns a 1-D integer tensor representing the shape of `input`.
-   *  <p>
-   *  For example:
-   *  <pre>{@code
-   *  # 't' is [[[1, 1, 1], [2, 2, 2]], [[3, 3, 3], [4, 4, 4]]]
-   *  shape(t) ==> [2, 2, 3]
-   *  }</pre>
-   *
-   * @param <U> data type for {@code output()} output
-   * @param input
-   * @param outType
-   * @return a new instance of Shape
-   */
-  public <U extends TNumber, T extends TType> org.tensorflow.op.core.Shape<U> shape(
-      Operand<T> input, DataType<U> outType) {
-    return org.tensorflow.op.core.Shape.create(scope, input, outType);
-  }
-
-  /**
-   * Returns shape of tensors.
-   *  <p>
-   *  This operation returns N 1-D integer tensors representing shape of `input[i]s`.
-   *
-   * @param <U> data type for {@code output()} output
-   * @param input
-   * @return a new instance of ShapeN
-   */
-  public <T extends TType> ShapeN<TInt32> shapeN(Iterable<Operand<T>> input) {
-    return ShapeN.create(scope, input);
-  }
-
-  /**
-   * Returns shape of tensors.
-   *  <p>
-   *  This operation returns N 1-D integer tensors representing shape of `input[i]s`.
-   *
-   * @param <U> data type for {@code output()} output
-   * @param input
-   * @param outType
-   * @return a new instance of ShapeN
-   */
-  public <U extends TNumber, T extends TType> ShapeN<U> shapeN(Iterable<Operand<T>> input,
-      DataType<U> outType) {
-    return ShapeN.create(scope, input, outType);
-  }
-
-  /**
-   * Returns the size of a tensor.
-   *  <p>
-   *  This operation returns an integer representing the number of elements in
-   *  `input`.
-   *  <p>
-   *  For example:
-   *  <pre>{@code
-   *  # 't' is [[[1, 1,, 1], [2, 2, 2]], [[3, 3, 3], [4, 4, 4]]]]
-   *  size(t) ==> 12
-   *  }</pre>
-   *
-   * @param <U> data type for {@code output()} output
-   * @param input
-   * @return a new instance of Size
-   */
-  public <T extends TType> Size<TInt32> size(Operand<T> input) {
-    return Size.create(scope, input);
-  }
-
-  /**
-   * Returns the size of a tensor.
-   *  <p>
-   *  This operation returns an integer representing the number of elements in
-   *  `input`.
-   *  <p>
-   *  For example:
-   *  <pre>{@code
-   *  # 't' is [[[1, 1,, 1], [2, 2, 2]], [[3, 3, 3], [4, 4, 4]]]]
-   *  size(t) ==> 12
-   *  }</pre>
-   *
-   * @param <U> data type for {@code output()} output
-   * @param input
-   * @param outType
-   * @return a new instance of Size
-   */
-  public <U extends TNumber, T extends TType> Size<U> size(Operand<T> input, DataType<U> outType) {
-    return Size.create(scope, input, outType);
-  }
-
-  /**
-   * Parses a text file and creates a batch of examples.
-   *
-   * @param filename The corpus's text file name.
-   * @param batchSize The size of produced batch.
-   * @param options carries optional attributes values
-   * @return a new instance of Skipgram
-   */
-  public Skipgram skipgram(String filename, Long batchSize, Skipgram.Options... options) {
-    return Skipgram.create(scope, filename, batchSize, options);
-  }
-
-  /**
-   * Return a slice from 'input'.
-   *  <p>
-   *  The output tensor is a tensor with dimensions described by 'size'
-   *  whose values are extracted from 'input' starting at the offsets in
-   *  'begin'.
-   *  <p>
-   *  <i>Requirements</i>:
-   *    0 <= begin[i] <= begin[i] + size[i] <= Di  for i in [0, n)
-   *
-   * @param <T> data type for {@code output()} output
-   * @param input
-   * @param begin begin[i] specifies the offset into the 'i'th dimension of
-   *  'input' to slice from.
-   * @param size size[i] specifies the number of elements of the 'i'th dimension
-   *  of 'input' to slice. If size[i] is -1, all remaining elements in dimension
-   *  i are included in the slice (i.e. this is equivalent to setting
-   *  size[i] = input.dim_size(i) - begin[i]).
-   * @return a new instance of Slice
-   */
-  public <T extends TType, U extends TNumber> Slice<T> slice(Operand<T> input, Operand<U> begin,
-      Operand<U> size) {
-    return Slice.create(scope, input, begin, size);
-  }
-
-  /**
-   * Returns a copy of the input tensor.
-   *
-   * @param <T> data type for {@code output()} output
-   * @param input
-   * @return a new instance of Snapshot
-   */
-  public <T extends TType> Snapshot<T> snapshot(Operand<T> input) {
-    return Snapshot.create(scope, input);
-  }
-
-  /**
-   * SpaceToBatch for N-D tensors of type T.
-   *  <p>
-   *  This operation divides "spatial" dimensions `[1, ..., M]` of the input into a
-   *  grid of blocks of shape `block_shape`, and interleaves these blocks with the
-   *  "batch" dimension (0) such that in the output, the spatial dimensions
-   *  `[1, ..., M]` correspond to the position within the grid, and the batch
-   *  dimension combines both the position within a spatial block and the original
-   *  batch position.  Prior to division into blocks, the spatial dimensions of the
-   *  input are optionally zero padded according to `paddings`.  See below for a
-   *  precise description.
-   *
-   * @param <T> data type for {@code output()} output
-   * @param input N-D with shape `input_shape = [batch] + spatial_shape + remaining_shape`,
-   *  where spatial_shape has `M` dimensions.
-   * @param blockShape 1-D with shape `[M]`, all values must be >= 1.
-   * @param paddings 2-D with shape `[M, 2]`, all values must be >= 0.
-   *    `paddings[i] = [pad_start, pad_end]` specifies the padding for input dimension
-   *    `i + 1`, which corresponds to spatial dimension `i`.  It is required that
-   *    `block_shape[i]` divides `input_shape[i + 1] + pad_start + pad_end`.
-   *  <p>
-   *  This operation is equivalent to the following steps:
-   *  <p>
-   *  1. Zero-pad the start and end of dimensions `[1, ..., M]` of the
-   *     input according to `paddings` to produce `padded` of shape `padded_shape`.
-   *  <p>
-   *  2. Reshape `padded` to `reshaped_padded` of shape:
-   *  <p>
-   *       [batch] +
-   *       [padded_shape[1] / block_shape[0],
-   *         block_shape[0],
-   *        ...,
-   *        padded_shape[M] / block_shape[M-1],
-   *        block_shape[M-1]] +
-   *       remaining_shape
-   *  <p>
-   *  3. Permute dimensions of `reshaped_padded` to produce
-   *     `permuted_reshaped_padded` of shape:
-   *  <p>
-   *       block_shape +
-   *       [batch] +
-   *       [padded_shape[1] / block_shape[0],
-   *        ...,
-   *        padded_shape[M] / block_shape[M-1]] +
-   *       remaining_shape
-   *  <p>
-   *  4. Reshape `permuted_reshaped_padded` to flatten `block_shape` into the batch
-   *     dimension, producing an output tensor of shape:
-   *  <p>
-   *       [batch * prod(block_shape)] +
-   *       [padded_shape[1] / block_shape[0],
-   *        ...,
-   *        padded_shape[M] / block_shape[M-1]] +
-   *       remaining_shape
-   *  <p>
-   *  Some examples:
-   *  <p>
-   *  (1) For the following input of shape `[1, 2, 2, 1]`, `block_shape = [2, 2]`, and
-   *      `paddings = [[0, 0], [0, 0]]`:
-   *  <pre>{@code
-   *  x = [[[[1], [2]], [[3], [4]]]]
-   *  }</pre>
-   *  The output tensor has shape `[4, 1, 1, 1]` and value:
-   *  <pre>{@code
-   *  [[[[1]]], [[[2]]], [[[3]]], [[[4]]]]
-   *  }</pre>
-   *  (2) For the following input of shape `[1, 2, 2, 3]`, `block_shape = [2, 2]`, and
-   *      `paddings = [[0, 0], [0, 0]]`:
-   *  <pre>{@code
-   *  x = [[[[1, 2, 3], [4, 5, 6]],
-   *        [[7, 8, 9], [10, 11, 12]]]]
-   *  }</pre>
-   *  The output tensor has shape `[4, 1, 1, 3]` and value:
-   *  <pre>{@code
-   *  [[[[1, 2, 3]]], [[[4, 5, 6]]], [[[7, 8, 9]]], [[[10, 11, 12]]]]
-   *  }</pre>
-   *  (3) For the following input of shape `[1, 4, 4, 1]`, `block_shape = [2, 2]`, and
-   *      `paddings = [[0, 0], [0, 0]]`:
-   *  <pre>{@code
-   *  x = [[[[1],   [2],  [3],  [4]],
-   *        [[5],   [6],  [7],  [8]],
-   *        [[9],  [10], [11],  [12]],
-   *        [[13], [14], [15],  [16]]]]
-   *  }</pre>
-   *  The output tensor has shape `[4, 2, 2, 1]` and value:
-   *  <pre>{@code
-   *  x = [[[[1], [3]], [[9], [11]]],
-   *       [[[2], [4]], [[10], [12]]],
-   *       [[[5], [7]], [[13], [15]]],
-   *       [[[6], [8]], [[14], [16]]]]
-   *  }</pre>
-   *  (4) For the following input of shape `[2, 2, 4, 1]`, block_shape = `[2, 2]`, and
-   *      paddings = `[[0, 0], [2, 0]]`:
-   *  <pre>{@code
-   *  x = [[[[1],   [2],  [3],  [4]],
-   *        [[5],   [6],  [7],  [8]]],
-   *       [[[9],  [10], [11],  [12]],
-   *        [[13], [14], [15],  [16]]]]
-   *  }</pre>
-   *  The output tensor has shape `[8, 1, 3, 1]` and value:
-   *  <pre>{@code
-   *  x = [[[[0], [1], [3]]], [[[0], [9], [11]]],
-   *       [[[0], [2], [4]]], [[[0], [10], [12]]],
-   *       [[[0], [5], [7]]], [[[0], [13], [15]]],
-   *       [[[0], [6], [8]]], [[[0], [14], [16]]]]
-   *  }</pre>
-   *  Among others, this operation is useful for reducing atrous convolution into
-   *  regular convolution.
-   * @return a new instance of SpaceToBatchNd
-   */
-  public <T extends TType, U extends TNumber, V extends TNumber> SpaceToBatchNd<T> spaceToBatchNd(
-      Operand<T> input, Operand<U> blockShape, Operand<V> paddings) {
-    return SpaceToBatchNd.create(scope, input, blockShape, paddings);
-  }
-
-  /**
-   * Splits a tensor into `num_split` tensors along one dimension.
-   *
-   * @param <T> data type for {@code output()} output
-   * @param axis 0-D.  The dimension along which to split.  Must be in the range
-   *  `[-rank(value), rank(value))`.
-   * @param value The tensor to split.
-   * @param numSplit The number of ways to split.  Must evenly divide
-   *  `value.shape[split_dim]`.
-   * @return a new instance of Split
-   */
-  public <T extends TType> Split<T> split(Operand<TInt32> axis, Operand<T> value, Long numSplit) {
-    return Split.create(scope, axis, value, numSplit);
-  }
-
-  /**
-   * Splits a tensor into `num_split` tensors along one dimension.
-   *
-   * @param <T> data type for {@code output()} output
-   * @param value The tensor to split.
-   * @param sizeSplits list containing the sizes of each output tensor along the split
-   *  dimension. Must sum to the dimension of value along split_dim.
-   *  Can contain one -1 indicating that dimension is to be inferred.
-   * @param axis 0-D.  The dimension along which to split.  Must be in the range
-   *  `[-rank(value), rank(value))`.
-   * @param numSplit
-   * @return a new instance of SplitV
-   */
-  public <T extends TType, U extends TNumber> SplitV<T> splitV(Operand<T> value,
-      Operand<U> sizeSplits, Operand<TInt32> axis, Long numSplit) {
-    return SplitV.create(scope, value, sizeSplits, axis, numSplit);
-  }
-
-  /**
-   * Removes dimensions of size 1 from the shape of a tensor.
-   *  <p>
-   *  Given a tensor `input`, this operation returns a tensor of the same type with
-   *  all dimensions of size 1 removed. If you don't want to remove all size 1
-   *  dimensions, you can remove specific size 1 dimensions by specifying
-   *  `axis`.
-   *  <p>
-   *  For example:
-   *  <pre>{@code
-   *  # 't' is a tensor of shape [1, 2, 1, 3, 1, 1]
-   *  shape(squeeze(t)) ==> [2, 3]
-   *  }</pre>
-   *  Or, to remove specific size 1 dimensions:
-   *  <pre>{@code
-   *  # 't' is a tensor of shape [1, 2, 1, 3, 1, 1]
-   *  shape(squeeze(t, [2, 4])) ==> [1, 2, 3, 1]
-   *  }</pre>
-   *
-   * @param <T> data type for {@code output()} output
-   * @param input The `input` to squeeze.
-   * @param options carries optional attributes values
-   * @return a new instance of Squeeze
-   */
-  public <T extends TType> Squeeze<T> squeeze(Operand<T> input, Squeeze.Options... options) {
-    return Squeeze.create(scope, input, options);
-  }
-
-  /**
-   * Packs a list of `N` rank-`R` tensors into one rank-`(R+1)` tensor.
-   *  <p>
-   *  Packs the `N` tensors in `values` into a tensor with rank one higher than each
-   *  tensor in `values`, by packing them along the `axis` dimension.
-   *  Given a list of tensors of shape `(A, B, C)`;
-   *  <p>
-   *  if `axis == 0` then the `output` tensor will have the shape `(N, A, B, C)`.
-   *  if `axis == 1` then the `output` tensor will have the shape `(A, N, B, C)`.
-   *  Etc.
-   *  <p>
-   *  For example:
-   *  <pre>{@code
-   *  # 'x' is [1, 4]
-   *  # 'y' is [2, 5]
-   *  # 'z' is [3, 6]
-   *  pack([x, y, z]) => [[1, 4], [2, 5], [3, 6]]  # Pack along first dim.
-   *  pack([x, y, z], axis=1) => [[1, 2, 3], [4, 5, 6]]
-   *  }</pre>
-   *  This is the opposite of `unpack`.
-   *
-   * @param <T> data type for {@code output()} output
-   * @param values Must be of same shape and type.
-   * @param options carries optional attributes values
-   * @return a new instance of Stack
-   */
-  public <T extends TType> Stack<T> stack(Iterable<Operand<T>> values, Stack.Options... options) {
-    return Stack.create(scope, values, options);
-  }
-
-  /**
-   * Stage values similar to a lightweight Enqueue.
-   *  <p>
-   *  The basic functionality of this Op is similar to a queue with many
-   *  fewer capabilities and options.  This Op is optimized for performance.
-   *
-   * @param values a list of tensors
-   *  dtypes A list of data types that inserted values should adhere to.
-   * @param options carries optional attributes values
-   * @return a new instance of Stage
-   */
-  public Stage stage(Iterable<Operand<?>> values, Stage.Options... options) {
-    return Stage.create(scope, values, options);
-  }
-
-  /**
-   * Op removes all elements in the underlying container.
-   *
-   * @param dtypes
-   * @param options carries optional attributes values
-   * @return a new instance of StageClear
-   */
-  public StageClear stageClear(List<DataType<?>> dtypes, StageClear.Options... options) {
-    return StageClear.create(scope, dtypes, options);
-  }
-
-  /**
-   * Op peeks at the values at the specified index.  If the
-   *  <p>
-   *  underlying container does not contain sufficient elements
-   *  this op will block until it does.   This Op is optimized for
-   *  performance.
-   *
-   * @param index
-   * @param dtypes
-   * @param options carries optional attributes values
-   * @return a new instance of StagePeek
-   */
-  public StagePeek stagePeek(Operand<TInt32> index, List<DataType<?>> dtypes,
-      StagePeek.Options... options) {
-    return StagePeek.create(scope, index, dtypes, options);
-  }
-
-  /**
-   * Op returns the number of elements in the underlying container.
-   *
-   * @param dtypes
-   * @param options carries optional attributes values
-   * @return a new instance of StageSize
-   */
-  public StageSize stageSize(List<DataType<?>> dtypes, StageSize.Options... options) {
-    return StageSize.create(scope, dtypes, options);
-  }
-
-  /**
-   * Stops gradient computation.
-   *  <p>
-   *  When executed in a graph, this op outputs its input tensor as-is.
-   *  <p>
-   *  When building ops to compute gradients, this op prevents the contribution of
-   *  its inputs to be taken into account.  Normally, the gradient generator adds ops
-   *  to a graph to compute the derivatives of a specified 'loss' by recursively
-   *  finding out inputs that contributed to its computation.  If you insert this op
-   *  in the graph it inputs are masked from the gradient generator.  They are not
-   *  taken into account for computing gradients.
-   *  <p>
-   *  This is useful any time you want to compute a value with TensorFlow but need
-   *  to pretend that the value was a constant. Some examples include:
-   *  <ul>
-   *  <li>
-   *  The <i>EM</i> algorithm where the <i>M-step</i> should not involve backpropagation
-   *     through the output of the <i>E-step</i>.
-   *  </li>
-   *  <li>
-   *  Contrastive divergence training of Boltzmann machines where, when
-   *     differentiating the energy function, the training must not backpropagate
-   *     through the graph that generated the samples from the model.
-   *  </li>
-   *  <li>
-   *  Adversarial training, where no backprop should happen through the adversarial
-   *     example generation process.
-   *
-   * @param <T> data type for {@code output()} output
-   * @param input
-   * @return a new instance of StopGradient
-   */
-  public <T extends TType> StopGradient<T> stopGradient(Operand<T> input) {
-    return StopGradient.create(scope, input);
-  }
-
-  /**
-   * Return a strided slice from `input`.
-   *  <p>
-   *  Note, most python users will want to use the Python `Tensor.__getitem__`
-   *  or `Variable.__getitem__` rather than this op directly.
-   *  <p>
-   *  The goal of this op is to produce a new tensor with a subset of
-   *  the elements from the `n` dimensional `input` tensor. The subset is chosen using
-   *  a sequence of `m` sparse range specifications encoded into the arguments
-   *  of this function. Note, in some cases
-   *  `m` could be equal to `n`, but this need not be the case. Each
-   *  range specification entry can be one of the following:
-   *  <p>
-   *  - An ellipsis (...). Ellipses are used to imply zero or more
-   *    dimensions of full-dimension selection and are produced using
-   *    `ellipsis_mask`. For example, `foo[...]` is the identity slice.
-   *  <p>
-   *  - A new axis. This is used to insert a new shape=1 dimension and is
-   *    produced using `new_axis_mask`. For example, `foo[:, ...]` where
-   *    `foo` is shape `(3, 4)` produces a `(1, 3, 4)` tensor.
-   *  <p>
-   *  - A range `begin:end:stride`. This is used to specify how much to choose from
-   *    a given dimension. `stride` can be any integer but 0.  `begin` is an integer
-   *    which represents the index of the first value to select while `end` represents
-   *    the index of the last value to select. The number of values selected in each
-   *    dimension is `end - begin` if `stride > 0` and `begin - end` if `stride < 0`.
-   *    `begin` and `end` can be negative where `-1` is the last element, `-2` is
-   *    the second to last. `begin_mask` controls whether to replace the explicitly
-   *    given `begin` with an implicit effective value of `0` if `stride > 0` and
-   *    `-1` if `stride < 0`. `end_mask` is analogous but produces the number
-   *    required to create the largest open interval. For example, given a shape
-   *    `(3,)` tensor `foo[:]`, the effective `begin` and `end` are `0` and `3`. Do
-   *    not assume this is equivalent to `foo[0:-1]` which has an effective `begin`
-   *    and `end` of `0` and `2`. Another example is `foo[-2::-1]` which reverses the
-   *    first dimension of a tensor while dropping the last two (in the original
-   *    order elements). For example `foo = [1,2,3,4]; foo[-2::-1]` is `[4,3]`.
-   *  <p>
-   *  - A single index. This is used to keep only elements that have a given
-   *    index. For example (`foo[2, :]` on a shape `(5,6)` tensor produces a
-   *    shape `(6,)` tensor. This is encoded in `begin` and `end` and
-   *    `shrink_axis_mask`.
-   *  <p>
-   *  Each conceptual range specification is encoded in the op's argument. This
-   *  encoding is best understand by considering a non-trivial example. In
-   *  particular,
-   *  `foo[1, 2:4, None, ..., :-3:-1, :]` will be encoded as
-   *  <pre>{@code
-   *  begin = [1, 2, x, x, 0, x] # x denotes don't care (usually 0)
-   *  end = [2, 4, x, x, -3, x]
-   *  strides = [1, 1, x, x, -1, 1]
-   *  begin_mask = 1<<4 | 1 << 5 = 48
-   *  end_mask = 1<<5 = 32
-   *  ellipsis_mask = 1<<3 = 8
-   *  new_axis_mask = 1<<2 4
-   *  shrink_axis_mask = 1<<0
-   *  }</pre>
-   *  In this case if `foo.shape` is (5, 5, 5, 5, 5, 5) the final shape of
-   *  the slice becomes (2, 1, 5, 5, 2, 5).
-   *  Let us walk step by step through each argument specification.
-   *  <p>
-   *  1.  The first argument in the example slice is turned into `begin = 1` and
-   *  `end = begin + 1 = 2`. To disambiguate from the original spec `2:4` we
-   *  also set the appropriate bit in `shrink_axis_mask`.
-   *  <p>
-   *  2. `2:4` is contributes 2, 4, 1 to begin, end, and stride. All masks have
-   *  zero bits contributed.
-   *  <p>
-   *  3. None is a synonym for `tf.newaxis`. This means insert a dimension of size 1
-   *  dimension in the final shape. Dummy values are contributed to begin,
-   *  end and stride, while the new_axis_mask bit is set.
-   *  <p>
-   *  4. `...` grab the full ranges from as many dimensions as needed to
-   *  fully specify a slice for every dimension of the input shape.
-   *  <p>
-   *  5. `:-3:-1` shows the use of negative indices. A negative index `i` associated
-   *  with a dimension that has shape `s` is converted to a positive index
-   *  `s + i`. So `-1` becomes `s-1` (i.e. the last element). This conversion
-   *  is done internally so begin, end and strides receive x, -3, and -1.
-   *  The appropriate begin_mask bit is set to indicate the start range is the
-   *  full range (ignoring the x).
-   *  <p>
-   *  6. `:` indicates that the entire contents of the corresponding dimension
-   *  is selected. This is equivalent to `::` or `0::1`. begin, end, and strides
-   *  receive 0, 0, and 1, respectively. The appropriate bits in `begin_mask` and
-   *  `end_mask` are also set.
-   *  <p>
-   *  <i>Requirements</i>:
-   *    `0 != strides[i] for i in [0, m)`
-   *    `ellipsis_mask must be a power of two (only one ellipsis)`
-   *
-   * @param <T> data type for {@code output()} output
-   * @param input
-   * @param begin `begin[k]` specifies the offset into the `k`th range specification.
-   *  The exact dimension this corresponds to will be determined by context.
-   *  Out-of-bounds values will be silently clamped. If the `k`th bit of
-   *  `begin_mask` then `begin[k]` is ignored and the full range of the
-   *  appropriate dimension is used instead. Negative values causes indexing
-   *  to start from the highest element e.g. If `foo==[1,2,3]` then `foo[-1]==3`.
-   * @param end `end[i]` is like `begin` with the exception that `end_mask` is
-   *  used to determine full ranges.
-   * @param strides `strides[i]` specifies the increment in the `i`th specification
-   *  after extracting a given element. Negative indices will reverse
-   *  the original order. Out or range values are
-   *  clamped to `[0,dim[i]) if slice[i]>0` or `[-1,dim[i]-1] if slice[i] < 0`
-   * @param options carries optional attributes values
-   * @return a new instance of StridedSlice
-   */
-  public <T extends TType, U extends TNumber> StridedSlice<T> stridedSlice(Operand<T> input,
-      Operand<U> begin, Operand<U> end, Operand<U> strides, StridedSlice.Options... options) {
-    return StridedSlice.create(scope, input, begin, end, strides, options);
-  }
-
-  /**
-   * Assign `value` to the sliced l-value reference of `ref`.
-   *  <p>
-   *  The values of `value` are assigned to the positions in the variable
-   *  `ref` that are selected by the slice parameters. The slice parameters
-   *  `begin`, `end`, `strides`, etc. work exactly as in `StridedSlice`.
-   *  <p>
-   *  NOTE this op currently does not support broadcasting and so `value`'s
-   *  shape must be exactly the shape produced by the slice of `ref`.
-   *
-   * @param <T> data type for {@code outputRef()} output
-   * @param ref
-   * @param begin
-   * @param end
-   * @param strides
-   * @param value
-   * @param options carries optional attributes values
-   * @return a new instance of StridedSliceAssign
-   */
-  public <T extends TType, U extends TNumber> StridedSliceAssign<T> stridedSliceAssign(
-      Operand<T> ref, Operand<U> begin, Operand<U> end, Operand<U> strides, Operand<T> value,
-      StridedSliceAssign.Options... options) {
-    return StridedSliceAssign.create(scope, ref, begin, end, strides, value, options);
-  }
-
-  /**
-   * Returns the gradient of `StridedSlice`.
-   *  <p>
-   *  Since `StridedSlice` cuts out pieces of its `input` which is size
-   *  `shape`, its gradient will have the same shape (which is passed here
-   *  as `shape`). The gradient will be zero in any element that the slice
-   *  does not select.
-   *  <p>
-   *  Arguments are the same as StridedSliceGrad with the exception that
-   *  `dy` is the input gradient to be propagated and `shape` is the
-   *  shape of `StridedSlice`'s `input`.
-   *
-   * @param <U> data type for {@code output()} output
-   * @param shape
-   * @param begin
-   * @param end
-   * @param strides
-   * @param dy
-   * @param options carries optional attributes values
-   * @return a new instance of StridedSliceGrad
-   */
-  public <U extends TType, T extends TNumber> StridedSliceGrad<U> stridedSliceGrad(Operand<T> shape,
-      Operand<T> begin, Operand<T> end, Operand<T> strides, Operand<U> dy,
-      StridedSliceGrad.Options... options) {
-    return StridedSliceGrad.create(scope, shape, begin, end, strides, dy, options);
-  }
-
-  /**
-   * Computes the sum of elements across dimensions of a tensor.
-   *  <p>
-   *  Reduces `input` along the dimensions given in `axis`. Unless
-   *  `keep_dims` is true, the rank of the tensor is reduced by 1 for each entry in
-   *  `axis`. If `keep_dims` is true, the reduced dimensions are
-   *  retained with length 1.
-   *
-   * @param <T> data type for {@code output()} output
-   * @param input The tensor to reduce.
-   * @param axis The dimensions to reduce. Must be in the range
-   *  `[-rank(input), rank(input))`.
-   * @param options carries optional attributes values
-   * @return a new instance of Sum
-   */
-  public <T extends TType, U extends TNumber> Sum<T> sum(Operand<T> input, Operand<U> axis,
-      Sum.Options... options) {
-    return Sum.create(scope, input, axis, options);
-  }
-
-  /**
-   * Forwards `data` to the output port determined by `pred`.
-   *  <p>
-   *  If `pred` is true, the `data` input is forwarded to `output_true`. Otherwise,
-   *  the data goes to `output_false`.
-   *  <p>
-   *  See also `RefSwitch` and `Merge`.
-   *
-   * @param <T> data type for {@code outputFalse()} output
-   * @param data The tensor to be forwarded to the appropriate output.
-   * @param pred A scalar that specifies which output port will receive data.
-   * @return a new instance of SwitchCond
-   */
-  public <T extends TType> SwitchCond<T> switchCond(Operand<T> data, Operand<TBool> pred) {
-    return SwitchCond.create(scope, data, pred);
-  }
-
-  /**
-   * Returns a tensor that may be mutated, but only persists within a single step.
-   *  <p>
-   *  This is an experimental op for internal use only and it is possible to use this
-   *  op in unsafe ways.  DO NOT USE unless you fully understand the risks.
-   *  <p>
-   *  It is the caller's responsibility to ensure that 'ref' is eventually passed to a
-   *  matching 'DestroyTemporaryVariable' op after all other uses have completed.
-   *  <p>
-   *  Outputs a ref to the tensor state so it may be read or modified.
-   *  <p>
-   *    E.g.
-   *        var = state_ops._temporary_variable([1, 2], types.float_)
-   *        var_name = var.op.name
-   *        var = state_ops.assign(var, [[4.0, 5.0]])
-   *        var = state_ops.assign_add(var, [[6.0, 7.0]])
-   *        final = state_ops._destroy_temporary_variable(var, var_name=var_name)
-   *
-   * @param <T> data type for {@code ref()} output
-   * @param shape The shape of the variable tensor.
-   * @param dtype The type of elements in the variable tensor.
-   * @param options carries optional attributes values
-   * @return a new instance of TemporaryVariable
-   */
-  public <T extends TType> TemporaryVariable<T> temporaryVariable(Shape shape, DataType<T> dtype,
-      TemporaryVariable.Options... options) {
-    return TemporaryVariable.create(scope, shape, dtype, options);
-  }
-
-  /**
-   * An array of Tensors of given size.
-   *  <p>
-   *  Write data via Write and read via Read or Pack.
-   *
-   * @param size The size of the array.
-   * @param dtype The type of the elements on the tensor_array.
-   * @param options carries optional attributes values
-   * @return a new instance of TensorArray
-   */
-  public <T extends TType> TensorArray tensorArray(Operand<TInt32> size, DataType<T> dtype,
-      TensorArray.Options... options) {
-    return TensorArray.create(scope, size, dtype, options);
-  }
-
-  /**
-   * Delete the TensorArray from its resource container.
-   *  <p>
-   *  This enables the user to close and release the resource in the middle
-   *  of a step/run.
-   *
-   * @param handle The handle to a TensorArray (output of TensorArray or TensorArrayGrad).
-   * @return a new instance of TensorArrayClose
-   */
-  public TensorArrayClose tensorArrayClose(Operand<?> handle) {
-    return TensorArrayClose.create(scope, handle);
-  }
-
-  /**
-   * Concat the elements from the TensorArray into value `value`.
-   *  <p>
-   *  Takes `T` elements of shapes
-   *  <p>
-   *    <pre>{@code
-   *    (n0 x d0 x d1 x ...), (n1 x d0 x d1 x ...), ..., (n(T-1) x d0 x d1 x ...)
-   *    }</pre>
-   *  and concatenates them into a Tensor of shape:
-   *  <p>
-   *    <pre>{@code
-   *  (n0 + n1 + ... + n(T-1) x d0 x d1 x ...)}</pre>
-   *  All elements must have the same shape (excepting the first dimension).
-   *
-   * @param <T> data type for {@code value()} output
-   * @param handle The handle to a TensorArray.
-   * @param flowIn A float scalar that enforces proper chaining of operations.
-   * @param dtype The type of the elem that is returned.
-   * @param options carries optional attributes values
-   * @return a new instance of TensorArrayConcat
-   */
-  public <T extends TType> TensorArrayConcat<T> tensorArrayConcat(Operand<?> handle,
-      Operand<TFloat32> flowIn, DataType<T> dtype, TensorArrayConcat.Options... options) {
-    return TensorArrayConcat.create(scope, handle, flowIn, dtype, options);
-  }
-
-  /**
-   * Gather specific elements from the TensorArray into output `value`.
-   *  <p>
-   *  All elements selected by `indices` must have the same shape.
-   *
-   * @param <T> data type for {@code value()} output
-   * @param handle The handle to a TensorArray.
-   * @param indices The locations in the TensorArray from which to read tensor elements.
-   * @param flowIn A float scalar that enforces proper chaining of operations.
-   * @param dtype The type of the elem that is returned.
-   * @param options carries optional attributes values
-   * @return a new instance of TensorArrayGather
-   */
-  public <T extends TType> TensorArrayGather<T> tensorArrayGather(Operand<?> handle,
-      Operand<TInt32> indices, Operand<TFloat32> flowIn, DataType<T> dtype,
-      TensorArrayGather.Options... options) {
-    return TensorArrayGather.create(scope, handle, indices, flowIn, dtype, options);
-  }
-
-  /**
-   * Creates a TensorArray for storing the gradients of values in the given handle.
-   *  <p>
-   *  If the given TensorArray gradient already exists, returns a reference to it.
-   *  <p>
-   *  Locks the size of the original TensorArray by disabling its dynamic size flag.
-   *  <p>
-   * *A note about the input flow_in:**
-   *  <p>
-   *  The handle flow_in forces the execution of the gradient lookup to occur
-   *  only after certain other operations have occurred.  For example, when
-   *  the forward TensorArray is dynamically sized, writes to this TensorArray
-   *  may resize the object.  The gradient TensorArray is statically sized based
-   *  on the size of the forward TensorArray when this operation executes.
-   *  Furthermore, the size of the forward TensorArray is frozen by this call.
-   *  As a result, the flow is used to ensure that the call to generate the gradient
-   *  TensorArray only happens after all writes are executed.
-   *  <p>
-   *  In the case of dynamically sized TensorArrays, gradient computation should
-   *  only be performed on read operations that have themselves been chained via
-   *  flow to occur only after all writes have executed. That way the final size
-   *  of the forward TensorArray is known when this operation is called.
-   *  <p>
-   * *A note about the source attribute:**
-   *  <p>
-   *  TensorArray gradient calls use an accumulator TensorArray object.  If
-   *  multiple gradients are calculated and run in the same session, the multiple
-   *  gradient nodes may accidentally flow through the same accumulator TensorArray.
-   *  This double counts and generally breaks the TensorArray gradient flow.
-   *  <p>
-   *  The solution is to identify which gradient call this particular
-   *  TensorArray gradient is being called in.  This is performed by identifying
-   *  a unique string (e.g. "gradients", "gradients_1", ...) from the input
-   *  gradient Tensor's name.  This string is used as a suffix when creating
-   *  the TensorArray gradient object here (the attribute `source`).
-   *  <p>
-   *  The attribute `source` is added as a suffix to the forward TensorArray's
-   *  name when performing the creation / lookup, so that each separate gradient
-   *  calculation gets its own TensorArray accumulator.
-   *
-   * @param handle The handle to the forward TensorArray.
-   * @param flowIn A float scalar that enforces proper chaining of operations.
-   * @param source The gradient source string, used to decide which gradient TensorArray
-   *  to return.
-   * @return a new instance of TensorArrayGrad
-   */
-  public TensorArrayGrad tensorArrayGrad(Operand<?> handle, Operand<TFloat32> flowIn,
-      String source) {
-    return TensorArrayGrad.create(scope, handle, flowIn, source);
-  }
-
-  /**
-   * Creates a TensorArray for storing multiple gradients of values in the given handle.
-   *  <p>
-   *  Similar to TensorArrayGradV3. However it creates an accumulator with an
-   *  expanded shape compared to the input TensorArray whose gradient is being
-   *  computed. This enables multiple gradients for the same TensorArray to be
-   *  calculated using the same accumulator.
-   *
-   * @param handle The handle to the forward TensorArray.
-   * @param flowIn A float scalar that enforces proper chaining of operations.
-   * @param shapeToPrepend An int32 vector representing a shape. Elements in the gradient accumulator will
-   *  have shape which is this shape_to_prepend value concatenated with shape of the
-   *  elements in the TensorArray corresponding to the input handle.
-   * @param source The gradient source string, used to decide which gradient TensorArray
-   *  to return.
-   * @return a new instance of TensorArrayGradWithShape
-   */
-  public TensorArrayGradWithShape tensorArrayGradWithShape(Operand<?> handle,
-      Operand<TFloat32> flowIn, Operand<TInt32> shapeToPrepend, String source) {
-    return TensorArrayGradWithShape.create(scope, handle, flowIn, shapeToPrepend, source);
-  }
-
-  /**
-   *
-   * @param <T> data type for {@code value()} output
-   * @param handle
-   * @param flowIn
-   * @param dtype
-   * @param options carries optional attributes values
-   * @return a new instance of TensorArrayPack
-   */
-  public <T extends TType> TensorArrayPack<T> tensorArrayPack(Operand<TString> handle,
-      Operand<TFloat32> flowIn, DataType<T> dtype, TensorArrayPack.Options... options) {
-    return TensorArrayPack.create(scope, handle, flowIn, dtype, options);
-  }
-
-  /**
-   * Read an element from the TensorArray into output `value`.
-   *
-   * @param <T> data type for {@code value()} output
-   * @param handle The handle to a TensorArray.
-   * @param index
-   * @param flowIn A float scalar that enforces proper chaining of operations.
-   * @param dtype The type of the elem that is returned.
-   * @return a new instance of TensorArrayRead
-   */
-  public <T extends TType> TensorArrayRead<T> tensorArrayRead(Operand<?> handle,
-      Operand<TInt32> index, Operand<TFloat32> flowIn, DataType<T> dtype) {
-    return TensorArrayRead.create(scope, handle, index, flowIn, dtype);
-  }
-
-  /**
-   * Scatter the data from the input value into specific TensorArray elements.
-   *  <p>
-   *  `indices` must be a vector, its length must match the first dim of `value`.
-   *
-   * @param handle The handle to a TensorArray.
-   * @param indices The locations at which to write the tensor elements.
-   * @param value The concatenated tensor to write to the TensorArray.
-   * @param flowIn A float scalar that enforces proper chaining of operations.
-   * @return a new instance of TensorArrayScatter
-   */
-  public <T extends TType> TensorArrayScatter tensorArrayScatter(Operand<?> handle,
-      Operand<TInt32> indices, Operand<T> value, Operand<TFloat32> flowIn) {
-    return TensorArrayScatter.create(scope, handle, indices, value, flowIn);
-  }
-
-  /**
-   * Get the current size of the TensorArray.
-   *
-   * @param handle The handle to a TensorArray (output of TensorArray or TensorArrayGrad).
-   * @param flowIn A float scalar that enforces proper chaining of operations.
-   * @return a new instance of TensorArraySize
-   */
-  public TensorArraySize tensorArraySize(Operand<?> handle, Operand<TFloat32> flowIn) {
-    return TensorArraySize.create(scope, handle, flowIn);
-  }
-
-  /**
-   * Split the data from the input value into TensorArray elements.
-   *  <p>
-   *  Assuming that `lengths` takes on values
-   *  <p>
-   *    <pre>{@code
-   *  (n0, n1, ..., n(T-1))}</pre>
-   *  and that `value` has shape
-   *  <p>
-   *    <pre>{@code
-   *  (n0 + n1 + ... + n(T-1) x d0 x d1 x ...)}</pre>
-   *  ,
-   *  <p>
-   *  this splits values into a TensorArray with T tensors.
-   *  <p>
-   *  TensorArray index t will be the subtensor of values with starting position
-   *  <p>
-   *    <pre>{@code
-   *  (n0 + n1 + ... + n(t-1), 0, 0, ...)}</pre>
-   *  and having size
-   *  <p>
-   *    <pre>{@code
-   *  nt x d0 x d1 x ...}</pre>
-   *
-   * @param handle The handle to a TensorArray.
-   * @param value The concatenated tensor to write to the TensorArray.
-   * @param lengths The vector of lengths, how to split the rows of value into the
-   *  TensorArray.
-   * @param flowIn A float scalar that enforces proper chaining of operations.
-   * @return a new instance of TensorArraySplit
-   */
-  public <T extends TType> TensorArraySplit tensorArraySplit(Operand<?> handle, Operand<T> value,
-      Operand<TInt64> lengths, Operand<TFloat32> flowIn) {
-    return TensorArraySplit.create(scope, handle, value, lengths, flowIn);
-  }
-
-  /**
-   *
-   * @param handle
-   * @param value
-   * @param flowIn
-   * @return a new instance of TensorArrayUnpack
-   */
-  public <T extends TType> TensorArrayUnpack tensorArrayUnpack(Operand<TString> handle,
-      Operand<T> value, Operand<TFloat32> flowIn) {
-    return TensorArrayUnpack.create(scope, handle, value, flowIn);
-  }
-
-  /**
-   * Push an element onto the tensor_array.
-   *
-   * @param handle The handle to a TensorArray.
-   * @param index The position to write to inside the TensorArray.
-   * @param value The tensor to write to the TensorArray.
-   * @param flowIn A float scalar that enforces proper chaining of operations.
-   * @return a new instance of TensorArrayWrite
-   */
-  public <T extends TType> TensorArrayWrite tensorArrayWrite(Operand<?> handle,
-      Operand<TInt32> index, Operand<T> value, Operand<TFloat32> flowIn) {
-    return TensorArrayWrite.create(scope, handle, index, value, flowIn);
-  }
-
-  /**
-   * Concats all tensors in the list along the 0th dimension.
-   *  <p>
-   *  Requires that all tensors have the same shape except the first dimension.
-   *  <p>
-   *  input_handle: The input list.
-   *  element_shape: The shape of the uninitialized elements in the list. If the first
-   *    dimension is not -1, it is assumed that all list elements have the same
-   *    leading dim.
-   *  leading_dims: The list of leading dims of uninitialized list elements. Used if
-   *    the leading dim of input_handle.element_shape or the element_shape input arg
-   *    is not already set.
-   *  tensor: The concated result.
-   *  lengths: Output tensor containing sizes of the 0th dimension of tensors in the list, used for computing the gradient.
-   *
-   * @param <U> data type for {@code tensor()} output
-   * @param inputHandle
-   * @param elementShape
-   * @param leadingDims
-   * @param elementDtype
-   * @return a new instance of TensorListConcat
-   */
-  public <U extends TType, T extends TNumber> TensorListConcat<U> tensorListConcat(
-      Operand<?> inputHandle, Operand<T> elementShape, Operand<TInt64> leadingDims,
-      DataType<U> elementDtype) {
-    return TensorListConcat.create(scope, inputHandle, elementShape, leadingDims, elementDtype);
-  }
-
-  /**
-   *
-   * @param inputA
-   * @param inputB
-   * @param elementDtype
-   * @return a new instance of TensorListConcatLists
-   */
-  public <T extends TType> TensorListConcatLists tensorListConcatLists(Operand<?> inputA,
-      Operand<?> inputB, DataType<T> elementDtype) {
-    return TensorListConcatLists.create(scope, inputA, inputB, elementDtype);
-  }
-
-  /**
-   * The shape of the elements of the given list, as a tensor.
-   *  <p>
-   *    input_handle: the list
-   *    element_shape: the shape of elements of the list
-   *
-   * @param <T> data type for {@code elementShape()} output
-   * @param inputHandle
-   * @param shapeType
-   * @return a new instance of TensorListElementShape
-   */
-  public <T extends TNumber> TensorListElementShape<T> tensorListElementShape(
-      Operand<?> inputHandle, DataType<T> shapeType) {
-    return TensorListElementShape.create(scope, inputHandle, shapeType);
-  }
-
-  /**
-   * Creates a TensorList which, when stacked, has the value of `tensor`.
-   *  <p>
-   *  Each tensor in the result list corresponds to one row of the input tensor.
-   *  <p>
-   *  tensor: The input tensor.
-   *  output_handle: The list.
-   *
-   * @param tensor
-   * @param elementShape
-   * @return a new instance of TensorListFromTensor
-   */
-  public <T extends TType, U extends TNumber> TensorListFromTensor tensorListFromTensor(
-      Operand<T> tensor, Operand<U> elementShape) {
-    return TensorListFromTensor.create(scope, tensor, elementShape);
-  }
-
-  /**
-   * Creates a Tensor by indexing into the TensorList.
-   *  <p>
-   *  Each row in the produced Tensor corresponds to the element in the TensorList
-   *  specified by the given index (see `tf.gather`).
-   *  <p>
-   *  input_handle: The input tensor list.
-   *  indices: The indices used to index into the list.
-   *  values: The tensor.
-   *
-   * @param <T> data type for {@code values()} output
-   * @param inputHandle
-   * @param indices
-   * @param elementShape
-   * @param elementDtype
-   * @return a new instance of TensorListGather
-   */
-  public <T extends TType> TensorListGather<T> tensorListGather(Operand<?> inputHandle,
-      Operand<TInt32> indices, Operand<TInt32> elementShape, DataType<T> elementDtype) {
-    return TensorListGather.create(scope, inputHandle, indices, elementShape, elementDtype);
-  }
-
-  /**
-   *
-   * @param <T> data type for {@code item()} output
-   * @param inputHandle
-   * @param index
-   * @param elementShape
-   * @param elementDtype
-   * @return a new instance of TensorListGetItem
-   */
-  public <T extends TType> TensorListGetItem<T> tensorListGetItem(Operand<?> inputHandle,
-      Operand<TInt32> index, Operand<TInt32> elementShape, DataType<T> elementDtype) {
-    return TensorListGetItem.create(scope, inputHandle, index, elementShape, elementDtype);
-  }
-
-  /**
-   * Returns the number of tensors in the input tensor list.
-   *  <p>
-   *  input_handle: the input list
-   *  length: the number of tensors in the list
-   *
-   * @param inputHandle
-   * @return a new instance of TensorListLength
-   */
-  public TensorListLength tensorListLength(Operand<?> inputHandle) {
-    return TensorListLength.create(scope, inputHandle);
-  }
-
-  /**
-   * Returns the last element of the input list as well as a list with all but that element.
-   *  <p>
-   *  Fails if the list is empty.
-   *  <p>
-   *  input_handle: the input list
-   *  tensor: the withdrawn last element of the list
-   *  element_dtype: the type of elements in the list
-   *  element_shape: the shape of the output tensor
-   *
-   * @param <T> data type for {@code tensor()} output
-   * @param inputHandle
-   * @param elementShape
-   * @param elementDtype
-   * @return a new instance of TensorListPopBack
-   */
-  public <T extends TType> TensorListPopBack<T> tensorListPopBack(Operand<?> inputHandle,
-      Operand<TInt32> elementShape, DataType<T> elementDtype) {
-    return TensorListPopBack.create(scope, inputHandle, elementShape, elementDtype);
-  }
-
-  /**
-   * Returns a list which has the passed-in `Tensor` as last element and the other elements of the given list in `input_handle`.
-   *  <p>
-   *  tensor: The tensor to put on the list.
-   *  input_handle: The old list.
-   *  output_handle: A list with the elements of the old list followed by tensor.
-   *  element_dtype: the type of elements in the list.
-   *  element_shape: a shape compatible with that of elements in the list.
-   *
-   * @param inputHandle
-   * @param tensor
-   * @return a new instance of TensorListPushBack
-   */
-  public <T extends TType> TensorListPushBack tensorListPushBack(Operand<?> inputHandle,
-      Operand<T> tensor) {
-    return TensorListPushBack.create(scope, inputHandle, tensor);
-  }
-
-  /**
-   *
-   * @param inputHandles
-   * @param tensor
-   * @return a new instance of TensorListPushBackBatch
-   */
-  public <T extends TType> TensorListPushBackBatch tensorListPushBackBatch(Operand<?> inputHandles,
-      Operand<T> tensor) {
-    return TensorListPushBackBatch.create(scope, inputHandles, tensor);
-  }
-
-  /**
-   * List of the given size with empty elements.
-   *  <p>
-   *  element_shape: the shape of the future elements of the list
-   *  num_elements: the number of elements to reserve
-   *  handle: the output list
-   *  element_dtype: the desired type of elements in the list.
-   *
-   * @param elementShape
-   * @param numElements
-   * @param elementDtype
-   * @return a new instance of TensorListReserve
-   */
-  public <T extends TNumber, U extends TType> TensorListReserve tensorListReserve(
-      Operand<T> elementShape, Operand<TInt32> numElements, DataType<U> elementDtype) {
-    return TensorListReserve.create(scope, elementShape, numElements, elementDtype);
-  }
-
-  /**
-   * Resizes the list.
-   *  <p>
-   *
-   *  input_handle: the input list
-   *  size: size of the output list
-   *
-   * @param inputHandle
-   * @param size
-   * @return a new instance of TensorListResize
-   */
-  public TensorListResize tensorListResize(Operand<?> inputHandle, Operand<TInt32> size) {
-    return TensorListResize.create(scope, inputHandle, size);
-  }
-
-  /**
-   * Creates a TensorList by indexing into a Tensor.
-   *  <p>
-   *  Each member of the TensorList corresponds to one row of the input tensor,
-   *  specified by the given index (see `tf.gather`).
-   *  <p>
-   *  tensor: The input tensor.
-   *  indices: The indices used to index into the list.
-   *  element_shape: The shape of the elements in the list (can be less specified than
-   *    the shape of the tensor).
-   *  num_elements: The size of the output list. Must be large enough to accommodate
-   *    the largest index in indices. If -1, the list is just large enough to include
-   *    the largest index in indices.
-   *  output_handle: The TensorList.
-   *
-   * @param tensor
-   * @param indices
-   * @param elementShape
-   * @param numElements
-   * @return a new instance of TensorListScatter
-   */
-  public <T extends TType, U extends TNumber> TensorListScatter tensorListScatter(Operand<T> tensor,
-      Operand<TInt32> indices, Operand<U> elementShape, Operand<TInt32> numElements) {
-    return TensorListScatter.create(scope, tensor, indices, elementShape, numElements);
-  }
-
-  /**
-   * Scatters tensor at indices in an input list.
-   *  <p>
-   *  Each member of the TensorList corresponds to one row of the input tensor,
-   *  specified by the given index (see `tf.gather`).
-   *  <p>
-   *  input_handle: The list to scatter into.
-   *  tensor: The input tensor.
-   *  indices: The indices used to index into the list.
-   *  output_handle: The TensorList.
-   *
-   * @param inputHandle
-   * @param tensor
-   * @param indices
-   * @return a new instance of TensorListScatterIntoExistingList
-   */
-  public <T extends TType> TensorListScatterIntoExistingList tensorListScatterIntoExistingList(
-      Operand<?> inputHandle, Operand<T> tensor, Operand<TInt32> indices) {
-    return TensorListScatterIntoExistingList.create(scope, inputHandle, tensor, indices);
-  }
-
-  /**
-   *
-   * @param inputHandle
-   * @param index
-   * @param item
-   * @return a new instance of TensorListSetItem
-   */
-  public <T extends TType> TensorListSetItem tensorListSetItem(Operand<?> inputHandle,
-      Operand<TInt32> index, Operand<T> item) {
-    return TensorListSetItem.create(scope, inputHandle, index, item);
-  }
-
-  /**
-   * Splits a tensor into a list.
-   *  <p>
-   *  list[i] corresponds to lengths[i] tensors from the input tensor.
-   *  The tensor must have rank at least 1 and contain exactly sum(lengths) elements.
-   *  <p>
-   *  tensor: The input tensor.
-   *  element_shape: A shape compatible with that of elements in the tensor.
-   *  lengths: Vector of sizes of the 0th dimension of tensors in the list.
-   *  output_handle: The list.
-   *
-   * @param tensor
-   * @param elementShape
-   * @param lengths
-   * @return a new instance of TensorListSplit
-   */
-  public <T extends TType, U extends TNumber> TensorListSplit tensorListSplit(Operand<T> tensor,
-      Operand<U> elementShape, Operand<TInt64> lengths) {
-    return TensorListSplit.create(scope, tensor, elementShape, lengths);
-  }
-
-  /**
-   * Stacks all tensors in the list.
-   *  <p>
-   *  Requires that all tensors have the same shape.
-   *  <p>
-   *  input_handle: the input list
-   *  tensor: the gathered result
-   *  num_elements: optional. If not -1, the number of elements in the list.
-   *
-   * @param <T> data type for {@code tensor()} output
-   * @param inputHandle
-   * @param elementShape
-   * @param elementDtype
-   * @param options carries optional attributes values
-   * @return a new instance of TensorListStack
-   */
-  public <T extends TType> TensorListStack<T> tensorListStack(Operand<?> inputHandle,
-      Operand<TInt32> elementShape, DataType<T> elementDtype, TensorListStack.Options... options) {
-    return TensorListStack.create(scope, inputHandle, elementShape, elementDtype, options);
-  }
-
-  /**
-   * Adds sparse `updates` to an existing tensor according to `indices`.
-   *  <p>
-   *  This operation creates a new tensor by adding sparse `updates` to the passed
-   *  in `tensor`.
-   *  This operation is very similar to `tf.scatter_nd_add`, except that the updates
-   *  are added onto an existing tensor (as opposed to a variable). If the memory
-   *  for the existing tensor cannot be re-used, a copy is made and updated.
-   *  <p>
-   *  `indices` is an integer tensor containing indices into a new tensor of shape
-   *  `shape`.  The last dimension of `indices` can be at most the rank of `shape`:
-   *  <p>
-   *      indices.shape[-1] <= shape.rank
-   *  <p>
-   *  The last dimension of `indices` corresponds to indices into elements
-   *  (if `indices.shape[-1] = shape.rank`) or slices
-   *  (if `indices.shape[-1] < shape.rank`) along dimension `indices.shape[-1]` of
-   *  `shape`.  `updates` is a tensor with shape
-   *  <p>
-   *      indices.shape[:-1] + shape[indices.shape[-1]:]
-   *  <p>
-   *  The simplest form of tensor_scatter_add is to add individual elements to a
-   *  tensor by index. For example, say we want to add 4 elements in a rank-1
-   *  tensor with 8 elements.
-   *  <p>
-   *  In Python, this scatter add operation would look like this:
-   *  <pre>{@code
-   *      indices = tf.constant([[4], [3], [1], [7]])
-   *      updates = tf.constant([9, 10, 11, 12])
-   *      tensor = tf.ones([8], dtype=tf.int32)
-   *      updated = tf.tensor_scatter_nd_add(tensor, indices, updates)
-   *      print(updated)
-   *  }</pre>
-   *  The resulting tensor would look like this:
-   *  <p>
-   *      [1, 12, 1, 11, 10, 1, 1, 13]
-   *  <p>
-   *  We can also, insert entire slices of a higher rank tensor all at once. For
-   *  example, if we wanted to insert two slices in the first dimension of a
-   *  rank-3 tensor with two matrices of new values.
-   *  <p>
-   *  In Python, this scatter add operation would look like this:
-   *  <pre>{@code
-   *      indices = tf.constant([[0], [2]])
-   *      updates = tf.constant([[[5, 5, 5, 5], [6, 6, 6, 6],
-   *                              [7, 7, 7, 7], [8, 8, 8, 8]],
-   *                             [[5, 5, 5, 5], [6, 6, 6, 6],
-   *                              [7, 7, 7, 7], [8, 8, 8, 8]]])
-   *      tensor = tf.ones([4, 4, 4],dtype=tf.int32)
-   *      updated = tf.tensor_scatter_nd_add(tensor, indices, updates)
-   *      print(updated)
-   *  }</pre>
-   *  The resulting tensor would look like this:
-   *  <p>
-   *      [[[6, 6, 6, 6], [7, 7, 7, 7], [8, 8, 8, 8], [9, 9, 9, 9]],
-   *       [[1, 1, 1, 1], [1, 1, 1, 1], [1, 1, 1, 1], [1, 1, 1, 1]],
-   *       [[6, 6, 6, 6], [7, 7, 7, 7], [8, 8, 8, 8], [9, 9, 9, 9]],
-   *       [[1, 1, 1, 1], [1, 1, 1, 1], [1, 1, 1, 1], [1, 1, 1, 1]]]
-   *  <p>
-   *  Note that on CPU, if an out of bound index is found, an error is returned.
-   *  On GPU, if an out of bound index is found, the index is ignored.
-   *
-   * @param <T> data type for {@code output()} output
-   * @param tensor Tensor to copy/update.
-   * @param indices Index tensor.
-   * @param updates Updates to scatter into output.
-   * @return a new instance of TensorScatterNdAdd
-   */
-  public <T extends TType, U extends TNumber> TensorScatterNdAdd<T> tensorScatterNdAdd(
-      Operand<T> tensor, Operand<U> indices, Operand<T> updates) {
-    return TensorScatterNdAdd.create(scope, tensor, indices, updates);
-  }
-
-  /**
-   * Subtracts sparse `updates` from an existing tensor according to `indices`.
-   *  <p>
-   *  This operation creates a new tensor by subtracting sparse `updates` from the
-   *  passed in `tensor`.
-   *  This operation is very similar to `tf.scatter_nd_sub`, except that the updates
-   *  are subtracted from an existing tensor (as opposed to a variable). If the memory
-   *  for the existing tensor cannot be re-used, a copy is made and updated.
-   *  <p>
-   *  `indices` is an integer tensor containing indices into a new tensor of shape
-   *  `shape`.  The last dimension of `indices` can be at most the rank of `shape`:
-   *  <p>
-   *      indices.shape[-1] <= shape.rank
-   *  <p>
-   *  The last dimension of `indices` corresponds to indices into elements
-   *  (if `indices.shape[-1] = shape.rank`) or slices
-   *  (if `indices.shape[-1] < shape.rank`) along dimension `indices.shape[-1]` of
-   *  `shape`.  `updates` is a tensor with shape
-   *  <p>
-   *      indices.shape[:-1] + shape[indices.shape[-1]:]
-   *  <p>
-   *  The simplest form of tensor_scatter_sub is to subtract individual elements
-   *  from a tensor by index. For example, say we want to insert 4 scattered elements
-   *  in a rank-1 tensor with 8 elements.
-   *  <p>
-   *  In Python, this scatter subtract operation would look like this:
-   *  <pre>{@code
-   *      indices = tf.constant([[4], [3], [1], [7]])
-   *      updates = tf.constant([9, 10, 11, 12])
-   *      tensor = tf.ones([8], dtype=tf.int32)
-   *      updated = tf.tensor_scatter_nd_sub(tensor, indices, updates)
-   *      print(updated)
-   *  }</pre>
-   *  The resulting tensor would look like this:
-   *  <p>
-   *      [1, -10, 1, -9, -8, 1, 1, -11]
-   *  <p>
-   *  We can also, insert entire slices of a higher rank tensor all at once. For
-   *  example, if we wanted to insert two slices in the first dimension of a
-   *  rank-3 tensor with two matrices of new values.
-   *  <p>
-   *  In Python, this scatter add operation would look like this:
-   *  <pre>{@code
-   *      indices = tf.constant([[0], [2]])
-   *      updates = tf.constant([[[5, 5, 5, 5], [6, 6, 6, 6],
-   *                              [7, 7, 7, 7], [8, 8, 8, 8]],
-   *                             [[5, 5, 5, 5], [6, 6, 6, 6],
-   *                              [7, 7, 7, 7], [8, 8, 8, 8]]])
-   *      tensor = tf.ones([4, 4, 4],dtype=tf.int32)
-   *      updated = tf.tensor_scatter_nd_sub(tensor, indices, updates)
-   *      print(updated)
-   *  }</pre>
-   *  The resulting tensor would look like this:
-   *  <p>
-   *      [[[-4, -4, -4, -4], [-5, -5, -5, -5], [-6, -6, -6, -6], [-7, -7, -7, -7]],
-   *       [[1, 1, 1, 1], [1, 1, 1, 1], [1, 1, 1, 1], [1, 1, 1, 1]],
-   *       [[-4, -4, -4, -4], [-5, -5, -5, -5], [-6, -6, -6, -6], [-7, -7, -7, -7]],
-   *       [[1, 1, 1, 1], [1, 1, 1, 1], [1, 1, 1, 1], [1, 1, 1, 1]]]
-   *  <p>
-   *  Note that on CPU, if an out of bound index is found, an error is returned.
-   *  On GPU, if an out of bound index is found, the index is ignored.
-   *
-   * @param <T> data type for {@code output()} output
-   * @param tensor Tensor to copy/update.
-   * @param indices Index tensor.
-   * @param updates Updates to scatter into output.
-   * @return a new instance of TensorScatterNdSub
-   */
-  public <T extends TType, U extends TNumber> TensorScatterNdSub<T> tensorScatterNdSub(
-      Operand<T> tensor, Operand<U> indices, Operand<T> updates) {
-    return TensorScatterNdSub.create(scope, tensor, indices, updates);
-  }
-
-  /**
-   * Scatter `updates` into an existing tensor according to `indices`.
-   *  <p>
-   *  This operation creates a new tensor by applying sparse `updates` to the passed
-   *  in `tensor`.
-   *  This operation is very similar to `tf.scatter_nd`, except that the updates are
-   *  scattered onto an existing tensor (as opposed to a zero-tensor). If the memory
-   *  for the existing tensor cannot be re-used, a copy is made and updated.
-   *  <p>
-   *  If `indices` contains duplicates, then their updates are accumulated (summed).
-   *  <p>
-   *  <b>WARNING</b>: The order in which updates are applied is nondeterministic, so the
-   *  output will be nondeterministic if `indices` contains duplicates -- because
-   *  of some numerical approximation issues, numbers summed in different order
-   *  may yield different results.
-   *  <p>
-   *  `indices` is an integer tensor containing indices into a new tensor of shape
-   *  `shape`.  The last dimension of `indices` can be at most the rank of `shape`:
-   *  <p>
-   *      indices.shape[-1] <= shape.rank
-   *  <p>
-   *  The last dimension of `indices` corresponds to indices into elements
-   *  (if `indices.shape[-1] = shape.rank`) or slices
-   *  (if `indices.shape[-1] < shape.rank`) along dimension `indices.shape[-1]` of
-   *  `shape`.  `updates` is a tensor with shape
-   *  <p>
-   *      indices.shape[:-1] + shape[indices.shape[-1]:]
-   *  <p>
-   *  The simplest form of scatter is to insert individual elements in a tensor by
-   *  index. For example, say we want to insert 4 scattered elements in a rank-1
-   *  tensor with 8 elements.
-   *  <p>
-   *  <div style="width:70%; margin:auto; margin-bottom:10px; margin-top:20px;">
-   *  <img style="width:100%" src="https://www.tensorflow.org/images/ScatterNd1.png" alt>
-   *  </div>
-   *  <p>
-   *  In Python, this scatter operation would look like this:
-   *  <p>
-   *      >>> indices = tf.constant([[4], [3], [1], [7]])
-   *      >>> updates = tf.constant([9, 10, 11, 12])
-   *      >>> tensor = tf.ones([8], dtype=tf.int32)
-   *      >>> print(tf.tensor_scatter_nd_update(tensor, indices, updates))
-   *      tf.Tensor([ 1 11  1 10  9  1  1 12], shape=(8,), dtype=int32)
-   *  <p>
-   *  We can also, insert entire slices of a higher rank tensor all at once. For
-   *  example, if we wanted to insert two slices in the first dimension of a
-   *  rank-3 tensor with two matrices of new values.
-   *  <p>
-   *  In Python, this scatter operation would look like this:
-   *  <p>
-   *      >>> indices = tf.constant([[0], [2]])
-   *      >>> updates = tf.constant([[[5, 5, 5, 5], [6, 6, 6, 6],
-   *      ...                         [7, 7, 7, 7], [8, 8, 8, 8]],
-   *      ...                        [[5, 5, 5, 5], [6, 6, 6, 6],
-   *      ...                         [7, 7, 7, 7], [8, 8, 8, 8]]])
-   *      >>> tensor = tf.ones([4, 4, 4], dtype=tf.int32)
-   *      >>> print(tf.tensor_scatter_nd_update(tensor, indices, updates).numpy())
-   *      [[[5 5 5 5]
-   *        [6 6 6 6]
-   *        [7 7 7 7]
-   *        [8 8 8 8]]
-   *       [[1 1 1 1]
-   *        [1 1 1 1]
-   *        [1 1 1 1]
-   *        [1 1 1 1]]
-   *       [[5 5 5 5]
-   *        [6 6 6 6]
-   *        [7 7 7 7]
-   *        [8 8 8 8]]
-   *       [[1 1 1 1]
-   *        [1 1 1 1]
-   *        [1 1 1 1]
-   *        [1 1 1 1]]]
-   *  <p>
-   *  Note that on CPU, if an out of bound index is found, an error is returned.
-   *  On GPU, if an out of bound index is found, the index is ignored.
-   *
-   * @param <T> data type for {@code output()} output
-   * @param tensor Tensor to copy/update.
-   * @param indices Index tensor.
-   * @param updates Updates to scatter into output.
-   * @return a new instance of TensorScatterNdUpdate
-   */
-  public <T extends TType, U extends TNumber> TensorScatterNdUpdate<T> tensorScatterNdUpdate(
-      Operand<T> tensor, Operand<U> indices, Operand<T> updates) {
-    return TensorScatterNdUpdate.create(scope, tensor, indices, updates);
-  }
-
-  /**
-   * Assign `value` to the sliced l-value reference of `input`.
-   *  <p>
-   *  The values of `value` are assigned to the positions in the tensor `input` that
-   *  are selected by the slice parameters. The slice parameters `begin` `end`
-   *  `strides` etc. work exactly as in `StridedSlice`.
-   *  <p>
-   *  NOTE this op currently does not support broadcasting and so `value`'s shape
-   *  must be exactly the shape produced by the slice of `input`.
-   *
-   * @param <T> data type for {@code output()} output
-   * @param input
-   * @param begin
-   * @param end
-   * @param strides
-   * @param value
-   * @param options carries optional attributes values
-   * @return a new instance of TensorStridedSliceUpdate
-   */
-  public <T extends TType, U extends TNumber> TensorStridedSliceUpdate<T> tensorStridedSliceUpdate(
-      Operand<T> input, Operand<U> begin, Operand<U> end, Operand<U> strides, Operand<T> value,
-      TensorStridedSliceUpdate.Options... options) {
-    return TensorStridedSliceUpdate.create(scope, input, begin, end, strides, value, options);
-  }
-
-  /**
-   * Constructs a tensor by tiling a given tensor.
-   *  <p>
-   *  This operation creates a new tensor by replicating `input` `multiples` times.
-   *  The output tensor's i'th dimension has `input.dims(i) * multiples[i]` elements,
-   *  and the values of `input` are replicated `multiples[i]` times along the 'i'th
-   *  dimension. For example, tiling `[a b c d]` by `[2]` produces
-   *  `[a b c d a b c d]`.
-   *  <p>
-   *  >>> a = tf.constant([[1,2,3],[4,5,6]], tf.int32)
-   *  >>> b = tf.constant([1,2], tf.int32)
-   *  >>> tf.tile(a, b)
-   *  <tf.Tensor: shape=(2, 6), dtype=int32, numpy=
-   *  array([[1, 2, 3, 1, 2, 3],
-   *         [4, 5, 6, 4, 5, 6]], dtype=int32)>
-   *  >>> c = tf.constant([2,1], tf.int32)
-   *  >>> tf.tile(a, c)
-   *  <tf.Tensor: shape=(4, 3), dtype=int32, numpy=
-   *  array([[1, 2, 3],
-   *         [4, 5, 6],
-   *         [1, 2, 3],
-   *         [4, 5, 6]], dtype=int32)>
-   *  >>> d = tf.constant([2,2], tf.int32)
-   *  >>> tf.tile(a, d)
-   *  <tf.Tensor: shape=(4, 6), dtype=int32, numpy=
-   *  array([[1, 2, 3, 1, 2, 3],
-   *         [4, 5, 6, 4, 5, 6],
-   *         [1, 2, 3, 1, 2, 3],
-   *         [4, 5, 6, 4, 5, 6]], dtype=int32)>
-   *
-   * @param <T> data type for {@code output()} output
-   * @param input 1-D or higher.
-   * @param multiples 1-D. Length must be the same as the number of dimensions in `input`
-   * @return a new instance of Tile
-   */
-  public <T extends TType, U extends TNumber> Tile<T> tile(Operand<T> input, Operand<U> multiples) {
-    return Tile.create(scope, input, multiples);
-  }
-
-  /**
-   * Provides the time since epoch in seconds.
-   *  <p>
-   *  Returns the timestamp as a `float64` for seconds since the Unix epoch.
-   *  <p>
-   *  Note: the timestamp is computed when the op is executed, not when it is added
-   *  to the graph.
-   *
-   * @return a new instance of Timestamp
-   */
-  public Timestamp timestamp() {
-    return Timestamp.create(scope);
-  }
-
-  /**
-   * Perform batches of RPC requests.
-   *  <p>
-   *  This op asynchronously performs either a single RPC request, or a batch
-   *  of requests.  RPC requests are defined by three main parameters:
-   *  <p>
-   *    - `address` (the host+port or BNS address of the request)
-   *    - `method` (the method name for the request)
-   *    - `request` (the serialized proto string, or vector of strings,
-   *       of the RPC request argument).
-   *  <p>
-   *  For example, if you have an RPC service running on port localhost:2345,
-   *  and its interface is configured with the following proto declaration:
-   *  <pre>{@code
-   *  service MyService {
-   *    rpc MyMethod(MyRequestProto) returns (MyResponseProto) {
-   *    }
-   *  };
-   *  }</pre>
-   *  then call this op with arguments:
-   *  <pre>{@code
-   *  address = "localhost:2345"
-   *  method = "MyService/MyMethod"
-   *  }</pre>
-   *  The `request` tensor is a string tensor representing serialized `MyRequestProto`
-   *  strings; and the output string tensor `response` will have the same shape
-   *  and contain (upon successful completion) corresponding serialized
-   *  `MyResponseProto` strings.
-   *  <p>
-   *  For example, to send a single, empty, `MyRequestProto`, call
-   *  this op with `request = ""`.  To send 5 <b>parallel</b> empty requests,
-   *  call this op with `request = ["", "", "", "", ""]`.
-   *  <p>
-   *  More generally, one can create a batch of `MyRequestProto` serialized protos
-   *  from regular batched tensors using the `encode_proto` op, and convert
-   *  the response `MyResponseProto` serialized protos to batched tensors
-   *  using the `decode_proto` op.
-   *  <p>
-   *  <b>NOTE</b> Working with serialized proto strings is faster than instantiating
-   *  actual proto objects in memory, so no performance degradation is expected
-   *  compared to writing custom kernels for this workflow.
-   *  <p>
-   *  Unlike the standard `Rpc` op, if the connection fails or the remote worker
-   *  returns an error status, this op does <b>not</b> reraise the exception.
-   *  Instead, the `status_code` and `status_message` entry for the corresponding RPC
-   *  call is set with the error returned from the RPC call.  The `response` tensor
-   *  will contain valid response values for those minibatch entries whose RPCs did
-   *  not fail; the rest of the entries will have empty strings.
-   *
-   * @param address `0-D` or `1-D`.  The address (i.e. host_name:port) of the RPC server.
-   *  If this tensor has more than 1 element, then multiple parallel rpc requests
-   *  are sent.  This argument broadcasts with `method` and `request`.
-   * @param method `0-D` or `1-D`.  The method address on the RPC server.
-   *  If this tensor has more than 1 element, then multiple parallel rpc requests
-   *  are sent.  This argument broadcasts with `address` and `request`.
-   * @param request `0-D` or `1-D`.  Serialized proto strings: the rpc request argument.
-   *  If this tensor has more than 1 element, then multiple parallel rpc requests
-   *  are sent.  This argument broadcasts with `address` and `method`.
-   * @param options carries optional attributes values
-   * @return a new instance of TryRpc
-   */
-  public TryRpc tryRpc(Operand<TString> address, Operand<TString> method, Operand<TString> request,
-      TryRpc.Options... options) {
-    return TryRpc.create(scope, address, method, request, options);
-  }
-
-  /**
-   * Reverses the operation of Batch for a single output Tensor.
-   *  <p>
-   *  An instance of Unbatch either receives an empty batched_tensor, in which case it
-   *  asynchronously waits until the values become available from a concurrently
-   *  running instance of Unbatch with the same container and shared_name, or receives
-   *  a non-empty batched_tensor in which case it finalizes all other concurrently
-   *  running instances and outputs its own element from the batch.
-   *  <p>
-   *  batched_tensor: The possibly transformed output of Batch. The size of the first
-   *   dimension should remain unchanged by the transformations for the operation to
-   *   work.
-   *  batch_index: The matching batch_index obtained from Batch.
-   *  id: The id scalar emitted by Batch.
-   *  unbatched_tensor: The Tensor corresponding to this execution.
-   *  timeout_micros: Maximum amount of time (in microseconds) to wait to receive the
-   *   batched input tensor associated with a given invocation of the op.
-   *  container: Container to control resource sharing.
-   *  shared_name: Instances of Unbatch with the same container and shared_name are
-   *   assumed to possibly belong to the same batch. If left empty, the op name will
-   *   be used as the shared name.
-   *
-   * @param <T> data type for {@code unbatchedTensor()} output
-   * @param batchedTensor
-   * @param batchIndex
-   * @param id
-   * @param timeoutMicros
-   * @param options carries optional attributes values
-   * @return a new instance of Unbatch
-   */
-  public <T extends TType> Unbatch<T> unbatch(Operand<T> batchedTensor, Operand<TInt64> batchIndex,
-      Operand<TInt64> id, Long timeoutMicros, Unbatch.Options... options) {
-    return Unbatch.create(scope, batchedTensor, batchIndex, id, timeoutMicros, options);
-  }
-
-  /**
-   * Gradient of Unbatch.
-   *  <p>
-   *  Acts like Batch but using the given batch_index index of batching things as they
-   *  become available. This ensures that the gradients are propagated back in the
-   *  same session which did the forward pass.
-   *  <p>
-   *  original_input: The input to the Unbatch operation this is the gradient of.
-   *  batch_index: The batch_index given to the Unbatch operation this is the gradient
-   *  of.
-   *  grad: The downstream gradient.
-   *  id: The id scalar emitted by Batch.
-   *  batched_grad: The return value, either an empty tensor or the batched gradient.
-   *  container: Container to control resource sharing.
-   *  shared_name: Instances of UnbatchGrad with the same container and shared_name
-   *   are assumed to possibly belong to the same batch. If left empty, the op name
-   *   will be used as the shared name.
-   *
-   * @param <T> data type for {@code batchedGrad()} output
-   * @param originalInput
-   * @param batchIndex
-   * @param grad
-   * @param id
-   * @param options carries optional attributes values
-   * @return a new instance of UnbatchGrad
-   */
-  public <T extends TType> UnbatchGrad<T> unbatchGrad(Operand<T> originalInput,
-      Operand<TInt64> batchIndex, Operand<T> grad, Operand<TInt64> id,
-      UnbatchGrad.Options... options) {
-    return UnbatchGrad.create(scope, originalInput, batchIndex, grad, id, options);
-  }
-
-  /**
-   * Finds unique elements along an axis of a tensor.
-   *  <p>
-   *  This operation either returns a tensor `y` containing unique elements
-   *  along the `axis` of a tensor. The returned unique elements is sorted
-   *  in the same order as they occur along `axis` in `x`.
-   *  This operation also returns a tensor `idx` that is the same size as
-   *  the number of the elements in `x` along the `axis` dimension. It
-   *  contains the index in the unique output `y`.
-   *  In other words, for an `1-D` tensor `x` with `axis = None:
-   *  <p>
-   *  `y[idx[i]] = x[i] for i in [0, 1,...,rank(x) - 1]`
-   *  <p>
-   *  For example:
-   *  <pre>{@code
-   *  # tensor 'x' is [1, 1, 2, 4, 4, 4, 7, 8, 8]
-   *  y, idx = unique(x)
-   *  y ==> [1, 2, 4, 7, 8]
-   *  idx ==> [0, 0, 1, 2, 2, 2, 3, 4, 4]
-   *  }</pre>
-   *  For an `2-D` tensor `x` with `axis = 0`:
-   *  <pre>{@code
-   *  # tensor 'x' is [[1, 0, 0],
-   *  #                [1, 0, 0],
-   *  #                [2, 0, 0]]
-   *  y, idx = unique(x, axis=0)
-   *  y ==> [[1, 0, 0],
-   *         [2, 0, 0]]
-   *  idx ==> [0, 0, 1]
-   *  }</pre>
-   *  For an `2-D` tensor `x` with `axis = 1`:
-   *  <pre>{@code
-   *  # tensor 'x' is [[1, 0, 0],
-   *  #                [1, 0, 0],
-   *  #                [2, 0, 0]]
-   *  y, idx = unique(x, axis=1)
-   *  y ==> [[1, 0],
-   *         [1, 0],
-   *         [2, 0]]
-   *  idx ==> [0, 1, 1]
-   *  }</pre>
-   *
-   * @param <T> data type for {@code y()} output
-   * @param <V> data type for {@code idx()} output
-   * @param x A `Tensor`.
-   * @param axis A `Tensor` of type `int32` (default: None). The axis of the Tensor to
-   *  find the unique elements.
-   * @return a new instance of Unique
-   */
-  public <T extends TType, U extends TNumber> Unique<T, TInt32> unique(Operand<T> x,
-      Operand<U> axis) {
-    return Unique.create(scope, x, axis);
-  }
-
-  /**
-   * Finds unique elements along an axis of a tensor.
-   *  <p>
-   *  This operation either returns a tensor `y` containing unique elements
-   *  along the `axis` of a tensor. The returned unique elements is sorted
-   *  in the same order as they occur along `axis` in `x`.
-   *  This operation also returns a tensor `idx` that is the same size as
-   *  the number of the elements in `x` along the `axis` dimension. It
-   *  contains the index in the unique output `y`.
-   *  In other words, for an `1-D` tensor `x` with `axis = None:
-   *  <p>
-   *  `y[idx[i]] = x[i] for i in [0, 1,...,rank(x) - 1]`
-   *  <p>
-   *  For example:
-   *  <pre>{@code
-   *  # tensor 'x' is [1, 1, 2, 4, 4, 4, 7, 8, 8]
-   *  y, idx = unique(x)
-   *  y ==> [1, 2, 4, 7, 8]
-   *  idx ==> [0, 0, 1, 2, 2, 2, 3, 4, 4]
-   *  }</pre>
-   *  For an `2-D` tensor `x` with `axis = 0`:
-   *  <pre>{@code
-   *  # tensor 'x' is [[1, 0, 0],
-   *  #                [1, 0, 0],
-   *  #                [2, 0, 0]]
-   *  y, idx = unique(x, axis=0)
-   *  y ==> [[1, 0, 0],
-   *         [2, 0, 0]]
-   *  idx ==> [0, 0, 1]
-   *  }</pre>
-   *  For an `2-D` tensor `x` with `axis = 1`:
-   *  <pre>{@code
-   *  # tensor 'x' is [[1, 0, 0],
-   *  #                [1, 0, 0],
-   *  #                [2, 0, 0]]
-   *  y, idx = unique(x, axis=1)
-   *  y ==> [[1, 0],
-   *         [1, 0],
-   *         [2, 0]]
-   *  idx ==> [0, 1, 1]
-   *  }</pre>
-   *
-   * @param <T> data type for {@code y()} output
-   * @param <V> data type for {@code idx()} output
-   * @param x A `Tensor`.
-   * @param axis A `Tensor` of type `int32` (default: None). The axis of the Tensor to
-   *  find the unique elements.
-   * @param outIdx
-   * @return a new instance of Unique
-   */
-  public <T extends TType, V extends TNumber, U extends TNumber> Unique<T, V> unique(Operand<T> x,
-      Operand<U> axis, DataType<V> outIdx) {
-    return Unique.create(scope, x, axis, outIdx);
-  }
-
-  /**
-   * Finds unique elements along an axis of a tensor.
-   *  <p>
-   *  This operation either returns a tensor `y` containing unique elements
-   *  along the `axis` of a tensor. The returned unique elements is sorted
-   *  in the same order as they occur along `axis` in `x`.
-   *  This operation also returns a tensor `idx` and a tensor `count`
-   *  that are the same size as the number of the elements in `x` along the
-   *  `axis` dimension. The `idx` contains the index in the unique output `y`
-   *  and the `count` contains the count in the unique output `y`.
-   *  In other words, for an `1-D` tensor `x` with `axis = None:
-   *  <p>
-   *  `y[idx[i]] = x[i] for i in [0, 1,...,rank(x) - 1]`
-   *  <p>
-   *  For example:
-   *  <pre>{@code
-   *  # tensor 'x' is [1, 1, 2, 4, 4, 4, 7, 8, 8]
-   *  y, idx, count = unique_with_counts(x)
-   *  y ==> [1, 2, 4, 7, 8]
-   *  idx ==> [0, 0, 1, 2, 2, 2, 3, 4, 4]
-   *  count ==> [2, 1, 3, 1, 2]
-   *  }</pre>
-   *  For an `2-D` tensor `x` with `axis = 0`:
-   *  <pre>{@code
-   *  # tensor 'x' is [[1, 0, 0],
-   *  #                [1, 0, 0],
-   *  #                [2, 0, 0]]
-   *  y, idx, count = unique_with_counts(x, axis=0)
-   *  y ==> [[1, 0, 0],
-   *         [2, 0, 0]]
-   *  idx ==> [0, 0, 1]
-   *  count ==> [2, 1]
-   *  }</pre>
-   *  For an `2-D` tensor `x` with `axis = 1`:
-   *  <pre>{@code
-   *  # tensor 'x' is [[1, 0, 0],
-   *  #                [1, 0, 0],
-   *  #                [2, 0, 0]]
-   *  y, idx, count = unique_with_counts(x, axis=1)
-   *  y ==> [[1, 0],
-   *         [1, 0],
-   *         [2, 0]]
-   *  idx ==> [0, 1, 1]
-   *  count ==> [1, 2]
-   *  }</pre>
-   *
-   * @param <T> data type for {@code y()} output
-   * @param <V> data type for {@code idx()} output
-   * @param x A `Tensor`.
-   * @param axis A `Tensor` of type `int32` (default: None). The axis of the Tensor to
-   *  find the unique elements.
-   * @return a new instance of UniqueWithCounts
-   */
-  public <T extends TType, U extends TNumber> UniqueWithCounts<T, TInt32> uniqueWithCounts(
-      Operand<T> x, Operand<U> axis) {
-    return UniqueWithCounts.create(scope, x, axis);
-  }
-
-  /**
-   * Finds unique elements along an axis of a tensor.
-   *  <p>
-   *  This operation either returns a tensor `y` containing unique elements
-   *  along the `axis` of a tensor. The returned unique elements is sorted
-   *  in the same order as they occur along `axis` in `x`.
-   *  This operation also returns a tensor `idx` and a tensor `count`
-   *  that are the same size as the number of the elements in `x` along the
-   *  `axis` dimension. The `idx` contains the index in the unique output `y`
-   *  and the `count` contains the count in the unique output `y`.
-   *  In other words, for an `1-D` tensor `x` with `axis = None:
-   *  <p>
-   *  `y[idx[i]] = x[i] for i in [0, 1,...,rank(x) - 1]`
-   *  <p>
-   *  For example:
-   *  <pre>{@code
-   *  # tensor 'x' is [1, 1, 2, 4, 4, 4, 7, 8, 8]
-   *  y, idx, count = unique_with_counts(x)
-   *  y ==> [1, 2, 4, 7, 8]
-   *  idx ==> [0, 0, 1, 2, 2, 2, 3, 4, 4]
-   *  count ==> [2, 1, 3, 1, 2]
-   *  }</pre>
-   *  For an `2-D` tensor `x` with `axis = 0`:
-   *  <pre>{@code
-   *  # tensor 'x' is [[1, 0, 0],
-   *  #                [1, 0, 0],
-   *  #                [2, 0, 0]]
-   *  y, idx, count = unique_with_counts(x, axis=0)
-   *  y ==> [[1, 0, 0],
-   *         [2, 0, 0]]
-   *  idx ==> [0, 0, 1]
-   *  count ==> [2, 1]
-   *  }</pre>
-   *  For an `2-D` tensor `x` with `axis = 1`:
-   *  <pre>{@code
-   *  # tensor 'x' is [[1, 0, 0],
-   *  #                [1, 0, 0],
-   *  #                [2, 0, 0]]
-   *  y, idx, count = unique_with_counts(x, axis=1)
-   *  y ==> [[1, 0],
-   *         [1, 0],
-   *         [2, 0]]
-   *  idx ==> [0, 1, 1]
-   *  count ==> [1, 2]
-   *  }</pre>
-   *
-   * @param <T> data type for {@code y()} output
-   * @param <V> data type for {@code idx()} output
-   * @param x A `Tensor`.
-   * @param axis A `Tensor` of type `int32` (default: None). The axis of the Tensor to
-   *  find the unique elements.
-   * @param outIdx
-   * @return a new instance of UniqueWithCounts
-   */
-  public <T extends TType, V extends TNumber, U extends TNumber> UniqueWithCounts<T, V> uniqueWithCounts(
-      Operand<T> x, Operand<U> axis, DataType<V> outIdx) {
-    return UniqueWithCounts.create(scope, x, axis, outIdx);
-  }
-
-  /**
-   * Converts an array of flat indices into a tuple of coordinate arrays.
-   *  <p>
-   *
-   *  Example:
-   *  <pre>{@code
-   *  y = tf.unravel_index(indices=[2, 5, 7], dims=[3, 3])
-   *  # 'dims' represent a hypothetical (3, 3) tensor of indices:
-   *  # [[0, 1, *2*],
-   *  #  [3, 4, *5*],
-   *  #  [6, *7*, 8]]
-   *  # For each entry from 'indices', this operation returns
-   *  # its coordinates (marked with '*'), such as
-   *  # 2 ==> (0, 2)
-   *  # 5 ==> (1, 2)
-   *  # 7 ==> (2, 1)
-   *  y ==> [[0, 1, 2], [2, 2, 1]]
-   *  }</pre>
-   *
-   * @compatibility(numpy) Equivalent to np.unravel_index
-   * @end_compatibility
-   * @param <T> data type for {@code output()} output
-   * @param indices An 0-D or 1-D `int` Tensor whose elements are indices into the
-   *  flattened version of an array of dimensions dims.
-   * @param dims An 1-D `int` Tensor. The shape of the array to use for unraveling
-   *  indices.
-   * @return a new instance of UnravelIndex
-   */
-  public <T extends TNumber> UnravelIndex<T> unravelIndex(Operand<T> indices, Operand<T> dims) {
-    return UnravelIndex.create(scope, indices, dims);
-  }
-
-  /**
-   * Unpacks a given dimension of a rank-`R` tensor into `num` rank-`(R-1)` tensors.
-   *  <p>
-   *  Unpacks `num` tensors from `value` by chipping it along the `axis` dimension.
-   *  For example, given a tensor of shape `(A, B, C, D)`;
-   *  <p>
-   *  If `axis == 0` then the i'th tensor in `output` is the slice `value[i, :, :, :]`
-   *    and each tensor in `output` will have shape `(B, C, D)`. (Note that the
-   *    dimension unpacked along is gone, unlike `split`).
-   *  <p>
-   *  If `axis == 1` then the i'th tensor in `output` is the slice `value[:, i, :, :]`
-   *    and each tensor in `output` will have shape `(A, C, D)`.
-   *  Etc.
-   *  <p>
-   *  This is the opposite of `pack`.
-   *
-   * @param <T> data type for {@code output()} output
-   * @param value 1-D or higher, with `axis` dimension size equal to `num`.
-   * @param num
-   * @param options carries optional attributes values
-   * @return a new instance of Unstack
-   */
-  public <T extends TType> Unstack<T> unstack(Operand<T> value, Long num,
-      Unstack.Options... options) {
-    return Unstack.create(scope, value, num, options);
-  }
-
-  /**
-   * Op is similar to a lightweight Dequeue.
-   *  <p>
-   *  The basic functionality is similar to dequeue with many fewer
-   *  capabilities and options.  This Op is optimized for performance.
-   *
-   * @param dtypes
-   * @param options carries optional attributes values
-   * @return a new instance of Unstage
-   */
-  public Unstage unstage(List<DataType<?>> dtypes, Unstage.Options... options) {
-    return Unstage.create(scope, dtypes, options);
-  }
-
-  /**
-   * Creates a handle to a Variable resource.
-   *
-   * @param dtype the type of this variable. Must agree with the dtypes
-   *  of all ops using this variable.
-   * @param shape The (possibly partially specified) shape of this variable.
-   * @param options carries optional attributes values
-   * @return a new instance of VarHandleOp
-   */
-  public <T extends TType> VarHandleOp varHandleOp(DataType<T> dtype, Shape shape,
-      VarHandleOp.Options... options) {
-    return VarHandleOp.create(scope, dtype, shape, options);
-  }
-
-  /**
-   * Checks whether a resource handle-based variable has been initialized.
-   *
-   * @param resource the input resource handle.
-   * @return a new instance of VarIsInitializedOp
-   */
-  public VarIsInitializedOp varIsInitializedOp(Operand<?> resource) {
-    return VarIsInitializedOp.create(scope, resource);
-  }
-
-  /**
-   * Factory method to create a new Variable with it's initializer.
-   *  <p>
-   *  Only supported on Graph sessions as the {@link org.tensorflow.op.core.Assign} op
-   *  does not work in an EagerSession.
-   *
-   * @param scope current scope
-   * @param init The op to use to initialise this variable.
-   * @param options carries optional attributes values
-   * @return a new instance of Variable
-   */
-  public <T extends TType> Variable<T> variable(Operand<T> init, Variable.Options... options) {
-    return Helpers.createVariableWithInit(scope, init, options);
-  }
-
-  /**
-   * Holds state in the form of a tensor that persists across steps.
-   *  <p>
-   *  Outputs a ref to the tensor state so it may be read or modified.
-   *  TODO(zhifengc/mrry): Adds a pointer to a more detail document
-   *  about sharing states in tensorflow.
-   *
-   * @param <T> data type for {@code ref()} output
-   * @param shape The shape of the variable tensor.
-   * @param dtype The type of elements in the variable tensor.
-   * @param options carries optional attributes values
-   * @return a new instance of Variable
-   */
-  public <T extends TType> Variable<T> variable(Shape shape, DataType<T> dtype,
-      Variable.Options... options) {
-    return Variable.create(scope, shape, dtype, options);
-  }
-
-  /**
-   * Returns the shape of the variable pointed to by `resource`.
-   *  <p>
-   *  This operation returns a 1-D integer tensor representing the shape of `input`.
-   *  <p>
-   *  For example:
-   *  <pre>{@code
-   *  # 't' is [[[1, 1, 1], [2, 2, 2]], [[3, 3, 3], [4, 4, 4]]]
-   *  shape(t) ==> [2, 2, 3]
-   *  }</pre>
-   *
-   * @param <T> data type for {@code output()} output
-   * @param input
-   * @return a new instance of VariableShape
-   */
-  public VariableShape<TInt32> variableShape(Operand<?> input) {
-    return VariableShape.create(scope, input);
-  }
-
-  /**
-   * Returns the shape of the variable pointed to by `resource`.
-   *  <p>
-   *  This operation returns a 1-D integer tensor representing the shape of `input`.
-   *  <p>
-   *  For example:
-   *  <pre>{@code
-   *  # 't' is [[[1, 1, 1], [2, 2, 2]], [[3, 3, 3], [4, 4, 4]]]
-   *  shape(t) ==> [2, 2, 3]
-   *  }</pre>
-   *
-   * @param <T> data type for {@code output()} output
-   * @param input
-   * @param outType
-   * @return a new instance of VariableShape
-   */
-  public <T extends TNumber> VariableShape<T> variableShape(Operand<?> input, DataType<T> outType) {
-    return VariableShape.create(scope, input, outType);
-  }
-
-  /**
-   * Returns locations of nonzero / true values in a tensor.
-   *  <p>
-   *  This operation returns the coordinates of true elements in `condition`. The
-   *  coordinates are returned in a 2-D tensor where the first dimension (rows)
-   *  represents the number of true elements, and the second dimension (columns)
-   *  represents the coordinates of the true elements. Keep in mind, the shape of
-   *  the output tensor can vary depending on how many true values there are in
-   *  `condition`. Indices are output in row-major order.
-   *  <p>
-   *  For example:
-   *  <pre>{@code
-   *  # 'input' tensor is [[True, False]
-   *  #                    [True, False]]
-   *  # 'input' has two true values, so output has two coordinates.
-   *  # 'input' has rank of 2, so coordinates have two indices.
-   *  where(input) ==> [[0, 0],
-   *                    [1, 0]]
-   *
-   *  # `condition` tensor is [[[True, False]
-   *  #                     [True, False]]
-   *  #                    [[False, True]
-   *  #                     [False, True]]
-   *  #                    [[False, False]
-   *  #                     [False, True]]]
-   *  # 'input' has 5 true values, so output has 5 coordinates.
-   *  # 'input' has rank of 3, so coordinates have three indices.
-   *  where(input) ==> [[0, 0, 0],
-   *                    [0, 1, 0],
-   *                    [1, 0, 1],
-   *                    [1, 1, 1],
-   *                    [2, 1, 1]]
-   *
-   *  # `condition` tensor is [[[1.5,  0.0]
-   *  #                     [-0.5, 0.0]]
-   *  #                    [[0.0,  0.25]
-   *  #                     [0.0,  0.75]]
-   *  #                    [[0.0,  0.0]
-   *  #                     [0.0,  0.01]]]
-   *  # 'input' has 5 nonzero values, so output has 5 coordinates.
-   *  # 'input' has rank of 3, so coordinates have three indices.
-   *  where(input) ==> [[0, 0, 0],
-   *                    [0, 1, 0],
-   *                    [1, 0, 1],
-   *                    [1, 1, 1],
-   *                    [2, 1, 1]]
-   *
-   *  # `condition` tensor is [[[1.5 + 0.0j, 0.0  + 0.0j]
-   *  #                     [0.0 + 0.5j, 0.0  + 0.0j]]
-   *  #                    [[0.0 + 0.0j, 0.25 + 1.5j]
-   *  #                     [0.0 + 0.0j, 0.75 + 0.0j]]
-   *  #                    [[0.0 + 0.0j, 0.0  + 0.0j]
-   *  #                     [0.0 + 0.0j, 0.01 + 0.0j]]]
-   *  # 'input' has 5 nonzero magnitude values, so output has 5 coordinates.
-   *  # 'input' has rank of 3, so coordinates have three indices.
-   *  where(input) ==> [[0, 0, 0],
-   *                    [0, 1, 0],
-   *                    [1, 0, 1],
-   *                    [1, 1, 1],
-   *                    [2, 1, 1]]
-   *  }</pre>
-   *
-   * @param condition
-   * @return a new instance of Where
-   */
-  public <T extends TType> Where where(Operand<T> condition) {
-    return Where.create(scope, condition);
-  }
-
-  /**
-   * Creates a zeroed tensor given its type and shape.
-   *
-   * @param scope is a scope used to add the underlying operation
-   * @param dims a 1-D operand that represents the shape of the output tensor
-   * @param type the output tensor datatype
-   * @return a constant tensor initialized with zeros
-   * @throws IllegalArgumentException if the tensor type or shape cannot be initialized with zeros.
-   */
-  public <T extends TType, U extends TNumber> Zeros<T> zeros(Operand<U> dims, DataType<T> type) {
-    return Zeros.create(scope, dims, type);
-  }
-
-  /**
-   * Returns a tensor of zeros with the same shape and type as x.
-   *
-   * @param <T> data type for {@code y()} output
-   * @param x a tensor of type T.
-   * @return a new instance of ZerosLike
-   */
-  public <T extends TType> ZerosLike<T> zerosLike(Operand<T> x) {
-    return ZerosLike.create(scope, x);
-  }
-
-  /**
-   * Returns an API that builds operations with the provided name prefix.
-   *
-   * @see {@link Scope#withSubScope(String)}
-   */
-  public Ops withSubScope(String childScopeName) {
-    return new Ops(scope.withSubScope(childScopeName));
-  }
-
-  /**
-   * Returns an API that uses the provided name for an op.
-   *
-   * @see {@link Scope#withName(String)}
-   */
-  public Ops withName(String opName) {
-    return new Ops(scope.withName(opName));
-  }
-
-  /**
-   * Returns an API that adds operations to the graph with the provided control dependencies.
-   *
-   * @see {@link Scope#withControlDependencies(Iterable<Op<?>>)}
-   */
-  public Ops withControlDependencies(Iterable<Op> controls) {
-    return new Ops(scope.withControlDependencies(controls));
-  }
-
-  /**
-   * Returns the current {@link Scope scope} of this API
-   */
-  public final Scope scope() {
-    return scope;
-  }
-
-  /**
-   * Creates an API for building operations in the provided execution environment
-   */
-  public static Ops create(ExecutionEnvironment env) {
-    return new Ops(new Scope(env));
-  }
-
-  /**
-   * Creates an API for building operations in the default eager execution environment
-   *
-   * <p>Invoking this method is equivalent to {@code Ops.create(EagerSession.getDefault())}.
-   */
-  public static Ops create() {
-    return new Ops(new Scope(EagerSession.getDefault()));
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/QuantizationOps.java b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/QuantizationOps.java
deleted file mode 100644
index aec0d667c65..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/QuantizationOps.java
+++ /dev/null
@@ -1,631 +0,0 @@
-// Copyright 2020 The TensorFlow Authors. All Rights Reserved.
-//
-// Licensed under the Apache License, Version 2.0 (the "License");
-// you may not use this file except in compliance with the License.
-// You may obtain a copy of the License at
-//
-//     http://www.apache.org/licenses/LICENSE-2.0
-//
-// Unless required by applicable law or agreed to in writing, software
-// distributed under the License is distributed on an "AS IS" BASIS,
-// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-// See the License for the specific language governing permissions and
-// limitations under the License.
-// ==============================================================================
-//
-// This class has been generated, DO NOT EDIT!
-//
-package org.tensorflow.op;
-
-import org.tensorflow.DataType;
-import org.tensorflow.Operand;
-import org.tensorflow.op.quantization.Dequantize;
-import org.tensorflow.op.quantization.FakeQuantWithMinMaxArgs;
-import org.tensorflow.op.quantization.FakeQuantWithMinMaxArgsGradient;
-import org.tensorflow.op.quantization.FakeQuantWithMinMaxVars;
-import org.tensorflow.op.quantization.FakeQuantWithMinMaxVarsGradient;
-import org.tensorflow.op.quantization.FakeQuantWithMinMaxVarsPerChannel;
-import org.tensorflow.op.quantization.FakeQuantWithMinMaxVarsPerChannelGradient;
-import org.tensorflow.op.quantization.Quantize;
-import org.tensorflow.op.quantization.QuantizeAndDequantize;
-import org.tensorflow.op.quantization.QuantizeDownAndShrinkRange;
-import org.tensorflow.op.quantization.QuantizedConcat;
-import org.tensorflow.op.quantization.RequantizationRange;
-import org.tensorflow.op.quantization.Requantize;
-import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * An API for building {@code quantization} operations as {@link Op Op}s
- *
- * @see {@link Ops}
- */
-public final class QuantizationOps {
-  private final Scope scope;
-
-  QuantizationOps(Scope scope) {
-    this.scope = scope;
-  }
-
-  /**
-   * Dequantize the 'input' tensor into a float or bfloat16 Tensor.
-   *  <p>
-   *  [min_range, max_range] are scalar floats that specify the range for
-   *  the output. The 'mode' attribute controls exactly which calculations are
-   *  used to convert the float values to their quantized equivalents.
-   *  <p>
-   *  In 'MIN_COMBINED' mode, each value of the tensor will undergo the following:
-   *  <pre>{@code
-   *  if T == qint8: in[i] += (range(T) + 1)/ 2.0
-   *  out[i] = min_range + (in[i]* (max_range - min_range) / range(T))
-   *  }</pre>
-   *  here `range(T) = numeric_limits<T>::max() - numeric_limits<T>::min()`
-   *  <p>
-   *  <i>MIN_COMBINED Mode Example</i>
-   *  <p>
-   *  If the input comes from a QuantizedRelu6, the output type is
-   *  quint8 (range of 0-255) but the possible range of QuantizedRelu6 is
-   *  0-6.  The min_range and max_range values are therefore 0.0 and 6.0.
-   *  Dequantize on quint8 will take each value, cast to float, and multiply
-   *  by 6 / 255.
-   *  Note that if quantizedtype is qint8, the operation will additionally add
-   *  each value by 128 prior to casting.
-   *  <p>
-   *  If the mode is 'MIN_FIRST', then this approach is used:
-   *  <pre>{@code
-   *  num_discrete_values = 1 << (# of bits in T)
-   *  range_adjust = num_discrete_values / (num_discrete_values - 1)
-   *  range = (range_max - range_min) * range_adjust
-   *  range_scale = range / num_discrete_values
-   *  const double offset_input = static_cast<double>(input) - lowest_quantized;
-   *  result = range_min + ((input - numeric_limits<T>::min()) * range_scale)
-   *  }</pre>
-   *  If the mode is `SCALED`, dequantization is performed by multiplying each
-   *  input value by a scaling_factor. (Thus an input of 0 always maps to 0.0).
-   *  <p>
-   *  The scaling_factor is determined from `min_range`, `max_range`, and
-   *  `narrow_range` in a way that is compatible with `QuantizeAndDequantize{V2|V3}`
-   *  and `QuantizeV2`, using the following algorithm:
-   *  <pre>{@code
-   *    const int min_expected_T = std::numeric_limits<T>::min() +
-   *      (narrow_range ? 1 : 0);
-   *    const int max_expected_T = std::numeric_limits<T>::max();
-   *    const float max_expected_T = std::numeric_limits<float>::max();
-   *
-   *    const float scale_factor =
-   *      (std::numeric_limits<T>::min() == 0) ? (max_range / max_expected_T)
-   *                                           : std::max(min_range / min_expected_T,
-   *                                                      max_range / max_expected_T);
-   *  }</pre>
-   *
-   * @param <U> data type for {@code output()} output
-   * @param input
-   * @param minRange The minimum scalar value possibly produced for the input.
-   * @param maxRange The maximum scalar value possibly produced for the input.
-   * @param options carries optional attributes values
-   * @return a new instance of Dequantize
-   */
-  public <T extends TType> Dequantize<TFloat32> dequantize(Operand<T> input,
-      Operand<TFloat32> minRange, Operand<TFloat32> maxRange, Dequantize.Options... options) {
-    return Dequantize.create(scope, input, minRange, maxRange, options);
-  }
-
-  /**
-   * Dequantize the 'input' tensor into a float or bfloat16 Tensor.
-   *  <p>
-   *  [min_range, max_range] are scalar floats that specify the range for
-   *  the output. The 'mode' attribute controls exactly which calculations are
-   *  used to convert the float values to their quantized equivalents.
-   *  <p>
-   *  In 'MIN_COMBINED' mode, each value of the tensor will undergo the following:
-   *  <pre>{@code
-   *  if T == qint8: in[i] += (range(T) + 1)/ 2.0
-   *  out[i] = min_range + (in[i]* (max_range - min_range) / range(T))
-   *  }</pre>
-   *  here `range(T) = numeric_limits<T>::max() - numeric_limits<T>::min()`
-   *  <p>
-   *  <i>MIN_COMBINED Mode Example</i>
-   *  <p>
-   *  If the input comes from a QuantizedRelu6, the output type is
-   *  quint8 (range of 0-255) but the possible range of QuantizedRelu6 is
-   *  0-6.  The min_range and max_range values are therefore 0.0 and 6.0.
-   *  Dequantize on quint8 will take each value, cast to float, and multiply
-   *  by 6 / 255.
-   *  Note that if quantizedtype is qint8, the operation will additionally add
-   *  each value by 128 prior to casting.
-   *  <p>
-   *  If the mode is 'MIN_FIRST', then this approach is used:
-   *  <pre>{@code
-   *  num_discrete_values = 1 << (# of bits in T)
-   *  range_adjust = num_discrete_values / (num_discrete_values - 1)
-   *  range = (range_max - range_min) * range_adjust
-   *  range_scale = range / num_discrete_values
-   *  const double offset_input = static_cast<double>(input) - lowest_quantized;
-   *  result = range_min + ((input - numeric_limits<T>::min()) * range_scale)
-   *  }</pre>
-   *  If the mode is `SCALED`, dequantization is performed by multiplying each
-   *  input value by a scaling_factor. (Thus an input of 0 always maps to 0.0).
-   *  <p>
-   *  The scaling_factor is determined from `min_range`, `max_range`, and
-   *  `narrow_range` in a way that is compatible with `QuantizeAndDequantize{V2|V3}`
-   *  and `QuantizeV2`, using the following algorithm:
-   *  <pre>{@code
-   *    const int min_expected_T = std::numeric_limits<T>::min() +
-   *      (narrow_range ? 1 : 0);
-   *    const int max_expected_T = std::numeric_limits<T>::max();
-   *    const float max_expected_T = std::numeric_limits<float>::max();
-   *
-   *    const float scale_factor =
-   *      (std::numeric_limits<T>::min() == 0) ? (max_range / max_expected_T)
-   *                                           : std::max(min_range / min_expected_T,
-   *                                                      max_range / max_expected_T);
-   *  }</pre>
-   *
-   * @param <U> data type for {@code output()} output
-   * @param input
-   * @param minRange The minimum scalar value possibly produced for the input.
-   * @param maxRange The maximum scalar value possibly produced for the input.
-   * @param dtype Type of the output tensor. Currently Dequantize supports float and bfloat16.
-   *  If 'dtype' is 'bfloat16', it only supports 'MIN_COMBINED' mode.
-   * @param options carries optional attributes values
-   * @return a new instance of Dequantize
-   */
-  public <U extends TNumber, T extends TType> Dequantize<U> dequantize(Operand<T> input,
-      Operand<TFloat32> minRange, Operand<TFloat32> maxRange, DataType<U> dtype,
-      Dequantize.Options... options) {
-    return Dequantize.create(scope, input, minRange, maxRange, dtype, options);
-  }
-
-  /**
-   * Fake-quantize the 'inputs' tensor, type float to 'outputs' tensor of same type.
-   *  <p>
-   *  Attributes
-   *  <ul>
-   *  <li>
-   *  `[min; max]` define the clamping range for the `inputs` data.
-   *  </li>
-   *  <li>
-   *  `inputs` values are quantized into the quantization range (
-   *  `[0; 2^num_bits - 1]` when `narrow_range` is false and `[1; 2^num_bits - 1]`
-   *  when it is true) and then de-quantized and output as floats in `[min; max]`
-   *  interval.
-   *  </li>
-   *  <li>
-   *  `num_bits` is the bitwidth of the quantization; between 2 and 16, inclusive.
-   *  </li>
-   *  </ul>
-   *  Before quantization, `min` and `max` values are adjusted with the following
-   *  logic.
-   *  It is suggested to have `min <= 0 <= max`. If `0` is not in the range of values,
-   *  the behavior can be unexpected:
-   *  <ul>
-   *  <li>
-   *  If `0 < min < max`: `min_adj = 0` and `max_adj = max - min`.
-   *  </li>
-   *  <li>
-   *  If `min < max < 0`: `min_adj = min - max` and `max_adj = 0`.
-   *  </li>
-   *  <li>
-   *  If `min <= 0 <= max`: `scale = (max - min) / (2^num_bits - 1) `,
-   *  `min_adj = scale * round(min / scale)` and `max_adj = max + min_adj - min`.
-   *  </li>
-   *  </ul>
-   *  Quantization is called fake since the output is still in floating point.
-   *
-   * @param inputs
-   * @param options carries optional attributes values
-   * @return a new instance of FakeQuantWithMinMaxArgs
-   */
-  public FakeQuantWithMinMaxArgs fakeQuantWithMinMaxArgs(Operand<TFloat32> inputs,
-      FakeQuantWithMinMaxArgs.Options... options) {
-    return FakeQuantWithMinMaxArgs.create(scope, inputs, options);
-  }
-
-  /**
-   * Compute gradients for a FakeQuantWithMinMaxArgs operation.
-   *
-   * @param gradients Backpropagated gradients above the FakeQuantWithMinMaxArgs operation.
-   * @param inputs Values passed as inputs to the FakeQuantWithMinMaxArgs operation.
-   * @param options carries optional attributes values
-   * @return a new instance of FakeQuantWithMinMaxArgsGradient
-   */
-  public FakeQuantWithMinMaxArgsGradient fakeQuantWithMinMaxArgsGradient(
-      Operand<TFloat32> gradients, Operand<TFloat32> inputs,
-      FakeQuantWithMinMaxArgsGradient.Options... options) {
-    return FakeQuantWithMinMaxArgsGradient.create(scope, gradients, inputs, options);
-  }
-
-  /**
-   * Fake-quantize the 'inputs' tensor of type float via global float scalars
-   *  <p>
-   *  Fake-quantize the `inputs` tensor of type float via global float scalars
-   *  `min` and `max` to `outputs` tensor of same shape as `inputs`.
-   *  <p>
-   *  Attributes
-   *  <ul>
-   *  <li>
-   *  `[min; max]` define the clamping range for the `inputs` data.
-   *  </li>
-   *  <li>
-   *  `inputs` values are quantized into the quantization range (
-   *  `[0; 2^num_bits - 1]` when `narrow_range` is false and `[1; 2^num_bits - 1]`
-   *  when it is true) and then de-quantized and output as floats in `[min; max]`
-   *  interval.
-   *  </li>
-   *  <li>
-   *  `num_bits` is the bitwidth of the quantization; between 2 and 16, inclusive.
-   *  </li>
-   *  </ul>
-   *  Before quantization, `min` and `max` values are adjusted with the following
-   *  logic.
-   *  It is suggested to have `min <= 0 <= max`. If `0` is not in the range of values,
-   *  the behavior can be unexpected:
-   *  <ul>
-   *  <li>
-   *  If `0 < min < max`: `min_adj = 0` and `max_adj = max - min`.
-   *  </li>
-   *  <li>
-   *  If `min < max < 0`: `min_adj = min - max` and `max_adj = 0`.
-   *  </li>
-   *  <li>
-   *  If `min <= 0 <= max`: `scale = (max - min) / (2^num_bits - 1) `,
-   *  `min_adj = scale * round(min / scale)` and `max_adj = max + min_adj - min`.
-   *  </li>
-   *  </ul>
-   *  This operation has a gradient and thus allows for training `min` and `max`
-   *  values.
-   *
-   * @param inputs
-   * @param min
-   * @param max
-   * @param options carries optional attributes values
-   * @return a new instance of FakeQuantWithMinMaxVars
-   */
-  public FakeQuantWithMinMaxVars fakeQuantWithMinMaxVars(Operand<TFloat32> inputs,
-      Operand<TFloat32> min, Operand<TFloat32> max, FakeQuantWithMinMaxVars.Options... options) {
-    return FakeQuantWithMinMaxVars.create(scope, inputs, min, max, options);
-  }
-
-  /**
-   * Compute gradients for a FakeQuantWithMinMaxVars operation.
-   *
-   * @param gradients Backpropagated gradients above the FakeQuantWithMinMaxVars operation.
-   * @param inputs Values passed as inputs to the FakeQuantWithMinMaxVars operation.
-   *  min, max: Quantization interval, scalar floats.
-   * @param min
-   * @param max
-   * @param options carries optional attributes values
-   * @return a new instance of FakeQuantWithMinMaxVarsGradient
-   */
-  public FakeQuantWithMinMaxVarsGradient fakeQuantWithMinMaxVarsGradient(
-      Operand<TFloat32> gradients, Operand<TFloat32> inputs, Operand<TFloat32> min,
-      Operand<TFloat32> max, FakeQuantWithMinMaxVarsGradient.Options... options) {
-    return FakeQuantWithMinMaxVarsGradient.create(scope, gradients, inputs, min, max, options);
-  }
-
-  /**
-   * Fake-quantize the 'inputs' tensor of type float via per-channel floats
-   *  <p>
-   *  Fake-quantize the `inputs` tensor of type float per-channel and one of the
-   *  shapes: `[d]`, `[b, d]` `[b, h, w, d]` via per-channel floats `min` and `max`
-   *  of shape `[d]` to `outputs` tensor of same shape as `inputs`.
-   *  <p>
-   *  Attributes
-   *  <ul>
-   *  <li>
-   *  `[min; max]` define the clamping range for the `inputs` data.
-   *  </li>
-   *  <li>
-   *  `inputs` values are quantized into the quantization range (
-   *  `[0; 2^num_bits - 1]` when `narrow_range` is false and `[1; 2^num_bits - 1]`
-   *  when it is true) and then de-quantized and output as floats in `[min; max]`
-   *  interval.
-   *  </li>
-   *  <li>
-   *  `num_bits` is the bitwidth of the quantization; between 2 and 16, inclusive.
-   *  </li>
-   *  </ul>
-   *  Before quantization, `min` and `max` values are adjusted with the following
-   *  logic.
-   *  It is suggested to have `min <= 0 <= max`. If `0` is not in the range of values,
-   *  the behavior can be unexpected:
-   *  <ul>
-   *  <li>
-   *  If `0 < min < max`: `min_adj = 0` and `max_adj = max - min`.
-   *  </li>
-   *  <li>
-   *  If `min < max < 0`: `min_adj = min - max` and `max_adj = 0`.
-   *  </li>
-   *  <li>
-   *  If `min <= 0 <= max`: `scale = (max - min) / (2^num_bits - 1) `,
-   *  `min_adj = scale * round(min / scale)` and `max_adj = max + min_adj - min`.
-   *  </li>
-   *  </ul>
-   *  This operation has a gradient and thus allows for training `min` and `max`
-   *  values.
-   *
-   * @param inputs
-   * @param min
-   * @param max
-   * @param options carries optional attributes values
-   * @return a new instance of FakeQuantWithMinMaxVarsPerChannel
-   */
-  public FakeQuantWithMinMaxVarsPerChannel fakeQuantWithMinMaxVarsPerChannel(
-      Operand<TFloat32> inputs, Operand<TFloat32> min, Operand<TFloat32> max,
-      FakeQuantWithMinMaxVarsPerChannel.Options... options) {
-    return FakeQuantWithMinMaxVarsPerChannel.create(scope, inputs, min, max, options);
-  }
-
-  /**
-   * Compute gradients for a FakeQuantWithMinMaxVarsPerChannel operation.
-   *
-   * @param gradients Backpropagated gradients above the FakeQuantWithMinMaxVars operation,
-   *  shape one of: `[d]`, `[b, d]`,  `[b, h, w, d]`.
-   * @param inputs Values passed as inputs to the FakeQuantWithMinMaxVars operation, shape
-   *    same as `gradients`.
-   *  min, max: Quantization interval, floats of shape `[d]`.
-   * @param min
-   * @param max
-   * @param options carries optional attributes values
-   * @return a new instance of FakeQuantWithMinMaxVarsPerChannelGradient
-   */
-  public FakeQuantWithMinMaxVarsPerChannelGradient fakeQuantWithMinMaxVarsPerChannelGradient(
-      Operand<TFloat32> gradients, Operand<TFloat32> inputs, Operand<TFloat32> min,
-      Operand<TFloat32> max, FakeQuantWithMinMaxVarsPerChannelGradient.Options... options) {
-    return FakeQuantWithMinMaxVarsPerChannelGradient.create(scope, gradients, inputs, min, max, options);
-  }
-
-  /**
-   * Quantize the 'input' tensor of type float to 'output' tensor of type 'T'.
-   *  <p>
-   *  [min_range, max_range] are scalar floats that specify the range for
-   *  the 'input' data. The 'mode' attribute controls exactly which calculations are
-   *  used to convert the float values to their quantized equivalents.  The
-   *  'round_mode' attribute controls which rounding tie-breaking algorithm is used
-   *  when rounding float values to their quantized equivalents.
-   *  <p>
-   *  In 'MIN_COMBINED' mode, each value of the tensor will undergo the following:
-   *  <pre>{@code
-   *  out[i] = (in[i] - min_range) * range(T) / (max_range - min_range)
-   *  if T == qint8: out[i] -= (range(T) + 1) / 2.0
-   *  }</pre>
-   *  here `range(T) = numeric_limits<T>::max() - numeric_limits<T>::min()`
-   *  <p>
-   *  <i>MIN_COMBINED Mode Example</i>
-   *  <p>
-   *  Assume the input is type float and has a possible range of [0.0, 6.0] and the
-   *  output type is quint8 ([0, 255]). The min_range and max_range values should be
-   *  specified as 0.0 and 6.0. Quantizing from float to quint8 will multiply each
-   *  value of the input by 255/6 and cast to quint8.
-   *  <p>
-   *  If the output type was qint8 ([-128, 127]), the operation will additionally
-   *  subtract each value by 128 prior to casting, so that the range of values aligns
-   *  with the range of qint8.
-   *  <p>
-   *  If the mode is 'MIN_FIRST', then this approach is used:
-   *  <pre>{@code
-   *  num_discrete_values = 1 << (# of bits in T)
-   *  range_adjust = num_discrete_values / (num_discrete_values - 1)
-   *  range = (range_max - range_min) * range_adjust
-   *  range_scale = num_discrete_values / range
-   *  quantized = round(input * range_scale) - round(range_min * range_scale) +
-   *    numeric_limits<T>::min()
-   *  quantized = max(quantized, numeric_limits<T>::min())
-   *  quantized = min(quantized, numeric_limits<T>::max())
-   *  }</pre>
-   *  The biggest difference between this and MIN_COMBINED is that the minimum range
-   *  is rounded first, before it's subtracted from the rounded value. With
-   *  MIN_COMBINED, a small bias is introduced where repeated iterations of quantizing
-   *  and dequantizing will introduce a larger and larger error.
-   *  <p>
-   *  <i>SCALED mode Example</i>
-   *  <p>
-   *  `SCALED` mode matches the quantization approach used in
-   *  `QuantizeAndDequantize{V2|V3}`.
-   *  <p>
-   *  If the mode is `SCALED`, the quantization is performed by multiplying each
-   *  input value by a scaling_factor.
-   *  The scaling_factor is determined from `min_range` and `max_range` to be as large
-   *  as possible such that the range from `min_range` to `max_range` is representable
-   *  within values of type T.
-   *  <pre>{@code
-   *    const int min_T = std::numeric_limits<T>::min();
-   *    const int max_T = std::numeric_limits<T>::max();
-   *    const float max_float = std::numeric_limits<float>::max();
-   *
-   *    const float scale_factor_from_min_side =
-   *        (min_T * min_range > 0) ? min_T / min_range : max_float;
-   *    const float scale_factor_from_max_side =
-   *        (max_T * max_range > 0) ? max_T / max_range : max_float;
-   *
-   *    const float scale_factor = std::min(scale_factor_from_min_side,
-   *                                        scale_factor_from_max_side);
-   *  }</pre>
-   *  We next use the scale_factor to adjust min_range and max_range as follows:
-   *  <pre>{@code
-   *        min_range = min_T / scale_factor;
-   *        max_range = max_T / scale_factor;
-   *  }</pre>
-   *  e.g. if T = qint8, and initially min_range = -10, and max_range = 9, we would
-   *  compare -128/-10.0 = 12.8 to 127/9.0 = 14.11, and set scaling_factor = 12.8
-   *  In this case, min_range would remain -10, but max_range would be adjusted to
-   *  127 / 12.8 = 9.921875
-   *  <p>
-   *  So we will quantize input values in the range (-10, 9.921875) to (-128, 127).
-   *  <p>
-   *  The input tensor can now be quantized by clipping values to the range
-   *  `min_range` to `max_range`, then multiplying by scale_factor as follows:
-   *  <pre>{@code
-   *  result = round(min(max_range, max(min_range, input)) * scale_factor)
-   *  }</pre>
-   *  The adjusted `min_range` and `max_range` are returned as outputs 2 and 3 of
-   *  this operation. These outputs should be used as the range for any further
-   *  calculations.
-   *  <p>
-   *  <i>narrow_range (bool) attribute</i>
-   *  <p>
-   *  If true, we do not use the minimum quantized value.
-   *  i.e. for int8 the quantized output, it would be restricted to the range
-   *  -127..127 instead of the full -128..127 range.
-   *  This is provided for compatibility with certain inference backends.
-   *  (Only applies to SCALED mode)
-   *  <p>
-   *  <i>axis (int) attribute</i>
-   *  <p>
-   *  An optional `axis` attribute can specify a dimension index of the input tensor,
-   *  such that quantization ranges will be calculated and applied separately for each
-   *  slice of the tensor along that dimension. This is useful for per-channel
-   *  quantization.
-   *  <p>
-   *  If axis is specified, min_range and max_range
-   *  <p>
-   *  if `axis`=None, per-tensor quantization is performed as normal.
-   *  <p>
-   *  <i>ensure_minimum_range (float) attribute</i>
-   *  <p>
-   *  Ensures the minimum quantization range is at least this value.
-   *  The legacy default value for this is 0.01, but it is strongly suggested to
-   *  set it to 0 for new uses.
-   *
-   * @param <T> data type for {@code output()} output
-   * @param input
-   * @param minRange The minimum value of the quantization range. This value may be adjusted by the
-   *  op depending on other parameters. The adjusted value is written to `output_min`.
-   *  If the `axis` attribute is specified, this must be a 1-D tensor whose size
-   *  matches the `axis` dimension of the input and output tensors.
-   * @param maxRange The maximum value of the quantization range. This value may be adjusted by the
-   *  op depending on other parameters. The adjusted value is written to `output_max`.
-   *  If the `axis` attribute is specified, this must be a 1-D tensor whose size
-   *  matches the `axis` dimension of the input and output tensors.
-   * @param T
-   * @param options carries optional attributes values
-   * @return a new instance of Quantize
-   */
-  public <T extends TType> Quantize<T> quantize(Operand<TFloat32> input, Operand<TFloat32> minRange,
-      Operand<TFloat32> maxRange, DataType<T> T, Quantize.Options... options) {
-    return Quantize.create(scope, input, minRange, maxRange, T, options);
-  }
-
-  /**
-   * Quantizes then dequantizes a tensor.
-   *  <p>
-   *  This is almost identical to QuantizeAndDequantizeV2, except that num_bits is a
-   *  tensor, so its value can change during training.
-   *
-   * @param <T> data type for {@code output()} output
-   * @param input
-   * @param inputMin
-   * @param inputMax
-   * @param numBits
-   * @param options carries optional attributes values
-   * @return a new instance of QuantizeAndDequantize
-   */
-  public <T extends TNumber> QuantizeAndDequantize<T> quantizeAndDequantize(Operand<T> input,
-      Operand<T> inputMin, Operand<T> inputMax, Operand<TInt32> numBits,
-      QuantizeAndDequantize.Options... options) {
-    return QuantizeAndDequantize.create(scope, input, inputMin, inputMax, numBits, options);
-  }
-
-  /**
-   * Convert the quantized 'input' tensor into a lower-precision 'output', using the
-   *  <p>
-   *  actual distribution of the values to maximize the usage of the lower bit depth
-   *  and adjusting the output min and max ranges accordingly.
-   *  <p>
-   *  [input_min, input_max] are scalar floats that specify the range for the float
-   *  interpretation of the 'input' data. For example, if input_min is -1.0f and
-   *  input_max is 1.0f, and we are dealing with quint16 quantized data, then a 0
-   *  value in the 16-bit data should be interpreted as -1.0f, and a 65535 means 1.0f.
-   *  <p>
-   *  This operator tries to squeeze as much precision as possible into an output with
-   *  a lower bit depth by calculating the actual min and max values found in the
-   *  data. For example, maybe that quint16 input has no values lower than 16,384 and
-   *  none higher than 49,152. That means only half the range is actually needed, all
-   *  the float interpretations are between -0.5f and 0.5f, so if we want to compress
-   *  the data into a quint8 output, we can use that range rather than the theoretical
-   *  -1.0f to 1.0f that is suggested by the input min and max.
-   *  <p>
-   *  In practice, this is most useful for taking output from operations like
-   *  QuantizedMatMul that can produce higher bit-depth outputs than their inputs and
-   *  may have large potential output ranges, but in practice have a distribution of
-   *  input values that only uses a small fraction of the possible range. By feeding
-   *  that output into this operator, we can reduce it from 32 bits down to 8 with
-   *  minimal loss of accuracy.
-   *
-   * @param <U> data type for {@code output()} output
-   * @param input
-   * @param inputMin The float value that the minimum quantized input value represents.
-   * @param inputMax The float value that the maximum quantized input value represents.
-   * @param outType The type of the output. Should be a lower bit depth than Tinput.
-   * @return a new instance of QuantizeDownAndShrinkRange
-   */
-  public <U extends TType, T extends TType> QuantizeDownAndShrinkRange<U> quantizeDownAndShrinkRange(
-      Operand<T> input, Operand<TFloat32> inputMin, Operand<TFloat32> inputMax,
-      DataType<U> outType) {
-    return QuantizeDownAndShrinkRange.create(scope, input, inputMin, inputMax, outType);
-  }
-
-  /**
-   * Concatenates quantized tensors along one dimension.
-   *
-   * @param <T> data type for {@code output()} output
-   * @param concatDim 0-D.  The dimension along which to concatenate.  Must be in the
-   *  range [0, rank(values)).
-   * @param values The `N` Tensors to concatenate. Their ranks and types must match,
-   *  and their sizes must match in all dimensions except `concat_dim`.
-   * @param inputMins The minimum scalar values for each of the input tensors.
-   * @param inputMaxes The maximum scalar values for each of the input tensors.
-   * @return a new instance of QuantizedConcat
-   */
-  public <T extends TType> QuantizedConcat<T> quantizedConcat(Operand<TInt32> concatDim,
-      Iterable<Operand<T>> values, Iterable<Operand<TFloat32>> inputMins,
-      Iterable<Operand<TFloat32>> inputMaxes) {
-    return QuantizedConcat.create(scope, concatDim, values, inputMins, inputMaxes);
-  }
-
-  /**
-   * Computes a range that covers the actual values present in a quantized tensor.
-   *  <p>
-   *  Given a quantized tensor described by `(input, input_min, input_max)`, outputs a
-   *  range that covers the actual values present in that tensor. This op is typically
-   *  used to produce the `requested_output_min` and `requested_output_max` for
-   *  `Requantize`.
-   *
-   * @param input
-   * @param inputMin The float value that the minimum quantized input value represents.
-   * @param inputMax The float value that the maximum quantized input value represents.
-   * @return a new instance of RequantizationRange
-   */
-  public <T extends TType> RequantizationRange requantizationRange(Operand<T> input,
-      Operand<TFloat32> inputMin, Operand<TFloat32> inputMax) {
-    return RequantizationRange.create(scope, input, inputMin, inputMax);
-  }
-
-  /**
-   * Converts the quantized `input` tensor into a lower-precision `output`.
-   *  <p>
-   *  Converts the quantized `input` tensor into a lower-precision `output`, using the
-   *  output range specified with `requested_output_min` and `requested_output_max`.
-   *  <p>
-   *  `[input_min, input_max]` are scalar floats that specify the range for the float
-   *  interpretation of the `input` data. For example, if `input_min` is -1.0f and
-   *  `input_max` is 1.0f, and we are dealing with `quint16` quantized data, then a 0
-   *  value in the 16-bit data should be interpreted as -1.0f, and a 65535 means 1.0f.
-   *
-   * @param <U> data type for {@code output()} output
-   * @param input
-   * @param inputMin The float value that the minimum quantized input value represents.
-   * @param inputMax The float value that the maximum quantized input value represents.
-   * @param requestedOutputMin The float value that the minimum quantized output value represents.
-   * @param requestedOutputMax The float value that the maximum quantized output value represents.
-   * @param outType The type of the output. Should be a lower bit depth than Tinput.
-   * @return a new instance of Requantize
-   */
-  public <U extends TType, T extends TType> Requantize<U> requantize(Operand<T> input,
-      Operand<TFloat32> inputMin, Operand<TFloat32> inputMax, Operand<TFloat32> requestedOutputMin,
-      Operand<TFloat32> requestedOutputMax, DataType<U> outType) {
-    return Requantize.create(scope, input, inputMin, inputMax, requestedOutputMin, requestedOutputMax, outType);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/RaggedOps.java b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/RaggedOps.java
deleted file mode 100644
index 7d585455bdd..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/RaggedOps.java
+++ /dev/null
@@ -1,63 +0,0 @@
-// Copyright 2020 The TensorFlow Authors. All Rights Reserved.
-//
-// Licensed under the Apache License, Version 2.0 (the "License");
-// you may not use this file except in compliance with the License.
-// You may obtain a copy of the License at
-//
-//     http://www.apache.org/licenses/LICENSE-2.0
-//
-// Unless required by applicable law or agreed to in writing, software
-// distributed under the License is distributed on an "AS IS" BASIS,
-// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-// See the License for the specific language governing permissions and
-// limitations under the License.
-// ==============================================================================
-//
-// This class has been generated, DO NOT EDIT!
-//
-package org.tensorflow.op;
-
-import org.tensorflow.Operand;
-import org.tensorflow.op.ragged.RaggedBincount;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * An API for building {@code ragged} operations as {@link Op Op}s
- *
- * @see {@link Ops}
- */
-public final class RaggedOps {
-  private final Scope scope;
-
-  RaggedOps(Scope scope) {
-    this.scope = scope;
-  }
-
-  /**
-   * Counts the number of occurrences of each value in an integer array.
-   *  <p>
-   *  Outputs a vector with length `size` and the same dtype as `weights`. If
-   *  `weights` are empty, then index `i` stores the number of times the value `i` is
-   *  counted in `arr`. If `weights` are non-empty, then index `i` stores the sum of
-   *  the value in `weights` at each index where the corresponding value in `arr` is
-   *  `i`.
-   *  <p>
-   *  Values in `arr` outside of the range [0, size) are ignored.
-   *
-   * @param <U> data type for {@code output()} output
-   * @param splits 1D int64 `Tensor`.
-   * @param values 2D int `Tensor`.
-   * @param size non-negative int scalar `Tensor`.
-   * @param weights is an int32, int64, float32, or float64 `Tensor` with the same
-   *  shape as `input`, or a length-0 `Tensor`, in which case it acts as all weights
-   *  equal to 1.
-   * @param options carries optional attributes values
-   * @return a new instance of RaggedBincount
-   */
-  public <U extends TNumber, T extends TNumber> RaggedBincount<U> raggedBincount(
-      Operand<TInt64> splits, Operand<T> values, Operand<T> size, Operand<U> weights,
-      RaggedBincount.Options... options) {
-    return RaggedBincount.create(scope, splits, values, size, weights, options);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/RandomOps.java b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/RandomOps.java
deleted file mode 100644
index 071e77c7a70..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/RandomOps.java
+++ /dev/null
@@ -1,586 +0,0 @@
-// Copyright 2020 The TensorFlow Authors. All Rights Reserved.
-//
-// Licensed under the Apache License, Version 2.0 (the "License");
-// you may not use this file except in compliance with the License.
-// You may obtain a copy of the License at
-//
-//     http://www.apache.org/licenses/LICENSE-2.0
-//
-// Unless required by applicable law or agreed to in writing, software
-// distributed under the License is distributed on an "AS IS" BASIS,
-// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-// See the License for the specific language governing permissions and
-// limitations under the License.
-// ==============================================================================
-//
-// This class has been generated, DO NOT EDIT!
-//
-package org.tensorflow.op;
-
-import org.tensorflow.DataType;
-import org.tensorflow.Operand;
-import org.tensorflow.op.random.AllCandidateSampler;
-import org.tensorflow.op.random.LogUniformCandidateSampler;
-import org.tensorflow.op.random.Multinomial;
-import org.tensorflow.op.random.ParameterizedTruncatedNormal;
-import org.tensorflow.op.random.RandomGamma;
-import org.tensorflow.op.random.RandomPoisson;
-import org.tensorflow.op.random.RandomShuffle;
-import org.tensorflow.op.random.RandomStandardNormal;
-import org.tensorflow.op.random.RandomUniform;
-import org.tensorflow.op.random.RandomUniformInt;
-import org.tensorflow.op.random.RecordInput;
-import org.tensorflow.op.random.StatefulRandomBinomial;
-import org.tensorflow.op.random.StatefulStandardNormal;
-import org.tensorflow.op.random.StatelessMultinomial;
-import org.tensorflow.op.random.StatelessRandomNormal;
-import org.tensorflow.op.random.StatelessRandomUniform;
-import org.tensorflow.op.random.StatelessTruncatedNormal;
-import org.tensorflow.op.random.TruncatedNormal;
-import org.tensorflow.op.random.UniformCandidateSampler;
-import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * An API for building {@code random} operations as {@link Op Op}s
- *
- * @see {@link Ops}
- */
-public final class RandomOps {
-  private final Scope scope;
-
-  RandomOps(Scope scope) {
-    this.scope = scope;
-  }
-
-  /**
-   * Generates labels for candidate sampling with a learned unigram distribution.
-   *  <p>
-   *  See explanations of candidate sampling and the data formats at
-   *  go/candidate-sampling.
-   *  <p>
-   *  For each batch, this op picks a single set of sampled candidate labels.
-   *  <p>
-   *  The advantages of sampling candidates per-batch are simplicity and the
-   *  possibility of efficient dense matrix multiplication. The disadvantage is that
-   *  the sampled candidates must be chosen independently of the context and of the
-   *  true labels.
-   *
-   * @param trueClasses A batch_size * num_true matrix, in which each row contains the
-   *  IDs of the num_true target_classes in the corresponding original label.
-   * @param numTrue Number of true labels per context.
-   * @param numSampled Number of candidates to produce.
-   * @param unique If unique is true, we sample with rejection, so that all sampled
-   *  candidates in a batch are unique. This requires some approximation to
-   *  estimate the post-rejection sampling probabilities.
-   * @param options carries optional attributes values
-   * @return a new instance of AllCandidateSampler
-   */
-  public AllCandidateSampler allCandidateSampler(Operand<TInt64> trueClasses, Long numTrue,
-      Long numSampled, Boolean unique, AllCandidateSampler.Options... options) {
-    return AllCandidateSampler.create(scope, trueClasses, numTrue, numSampled, unique, options);
-  }
-
-  /**
-   * Generates labels for candidate sampling with a log-uniform distribution.
-   *  <p>
-   *  See explanations of candidate sampling and the data formats at
-   *  go/candidate-sampling.
-   *  <p>
-   *  For each batch, this op picks a single set of sampled candidate labels.
-   *  <p>
-   *  The advantages of sampling candidates per-batch are simplicity and the
-   *  possibility of efficient dense matrix multiplication. The disadvantage is that
-   *  the sampled candidates must be chosen independently of the context and of the
-   *  true labels.
-   *
-   * @param trueClasses A batch_size * num_true matrix, in which each row contains the
-   *  IDs of the num_true target_classes in the corresponding original label.
-   * @param numTrue Number of true labels per context.
-   * @param numSampled Number of candidates to randomly sample.
-   * @param unique If unique is true, we sample with rejection, so that all sampled
-   *  candidates in a batch are unique. This requires some approximation to
-   *  estimate the post-rejection sampling probabilities.
-   * @param rangeMax The sampler will sample integers from the interval [0, range_max).
-   * @param options carries optional attributes values
-   * @return a new instance of LogUniformCandidateSampler
-   */
-  public LogUniformCandidateSampler logUniformCandidateSampler(Operand<TInt64> trueClasses,
-      Long numTrue, Long numSampled, Boolean unique, Long rangeMax,
-      LogUniformCandidateSampler.Options... options) {
-    return LogUniformCandidateSampler.create(scope, trueClasses, numTrue, numSampled, unique, rangeMax, options);
-  }
-
-  /**
-   * Draws samples from a multinomial distribution.
-   *
-   * @param <U> data type for {@code output()} output
-   * @param logits 2-D Tensor with shape `[batch_size, num_classes]`.  Each slice `[i, :]`
-   *  represents the unnormalized log probabilities for all classes.
-   * @param numSamples 0-D.  Number of independent samples to draw for each row slice.
-   * @param options carries optional attributes values
-   * @return a new instance of Multinomial
-   */
-  public <T extends TNumber> Multinomial<TInt64> multinomial(Operand<T> logits,
-      Operand<TInt32> numSamples, Multinomial.Options... options) {
-    return Multinomial.create(scope, logits, numSamples, options);
-  }
-
-  /**
-   * Draws samples from a multinomial distribution.
-   *
-   * @param <U> data type for {@code output()} output
-   * @param logits 2-D Tensor with shape `[batch_size, num_classes]`.  Each slice `[i, :]`
-   *  represents the unnormalized log probabilities for all classes.
-   * @param numSamples 0-D.  Number of independent samples to draw for each row slice.
-   * @param outputDtype
-   * @param options carries optional attributes values
-   * @return a new instance of Multinomial
-   */
-  public <U extends TNumber, T extends TNumber> Multinomial<U> multinomial(Operand<T> logits,
-      Operand<TInt32> numSamples, DataType<U> outputDtype, Multinomial.Options... options) {
-    return Multinomial.create(scope, logits, numSamples, outputDtype, options);
-  }
-
-  /**
-   * Outputs random values from a normal distribution. The parameters may each be a
-   *  <p>
-   *  scalar which applies to the entire output, or a vector of length shape[0] which
-   *  stores the parameters for each batch.
-   *
-   * @param <U> data type for {@code output()} output
-   * @param shape The shape of the output tensor. Batches are indexed by the 0th dimension.
-   * @param means The mean parameter of each batch.
-   * @param stdevs The standard deviation parameter of each batch. Must be greater than 0.
-   * @param minvals The minimum cutoff. May be -infinity.
-   * @param maxvals The maximum cutoff. May be +infinity, and must be more than the minval
-   *  for each batch.
-   * @param options carries optional attributes values
-   * @return a new instance of ParameterizedTruncatedNormal
-   */
-  public <U extends TNumber, T extends TNumber> ParameterizedTruncatedNormal<U> parameterizedTruncatedNormal(
-      Operand<T> shape, Operand<U> means, Operand<U> stdevs, Operand<U> minvals, Operand<U> maxvals,
-      ParameterizedTruncatedNormal.Options... options) {
-    return ParameterizedTruncatedNormal.create(scope, shape, means, stdevs, minvals, maxvals, options);
-  }
-
-  /**
-   * Outputs random values from the Gamma distribution(s) described by alpha.
-   *  <p>
-   *  This op uses the algorithm by Marsaglia et al. to acquire samples via
-   *  transformation-rejection from pairs of uniform and normal random variables.
-   *  See http://dl.acm.org/citation.cfm?id=358414
-   *
-   * @param <U> data type for {@code output()} output
-   * @param shape 1-D integer tensor. Shape of independent samples to draw from each
-   *  distribution described by the shape parameters given in alpha.
-   * @param alpha A tensor in which each scalar is a "shape" parameter describing the
-   *  associated gamma distribution.
-   * @param options carries optional attributes values
-   * @return a new instance of RandomGamma
-   */
-  public <U extends TNumber, T extends TNumber> RandomGamma<U> randomGamma(Operand<T> shape,
-      Operand<U> alpha, RandomGamma.Options... options) {
-    return RandomGamma.create(scope, shape, alpha, options);
-  }
-
-  /**
-   * Outputs random values from the Poisson distribution(s) described by rate.
-   *  <p>
-   *  This op uses two algorithms, depending on rate. If rate >= 10, then
-   *  the algorithm by Hormann is used to acquire samples via
-   *  transformation-rejection.
-   *  See http://www.sciencedirect.com/science/article/pii/0167668793909974.
-   *  <p>
-   *  Otherwise, Knuth's algorithm is used to acquire samples via multiplying uniform
-   *  random variables.
-   *  See Donald E. Knuth (1969). Seminumerical Algorithms. The Art of Computer
-   *  Programming, Volume 2. Addison Wesley
-   *
-   * @param <V> data type for {@code output()} output
-   * @param shape 1-D integer tensor. Shape of independent samples to draw from each
-   *  distribution described by the shape parameters given in rate.
-   * @param rate A tensor in which each scalar is a "rate" parameter describing the
-   *  associated poisson distribution.
-   * @param options carries optional attributes values
-   * @return a new instance of RandomPoisson
-   */
-  public <T extends TNumber, U extends TNumber> RandomPoisson<TInt64> randomPoisson(
-      Operand<T> shape, Operand<U> rate, RandomPoisson.Options... options) {
-    return RandomPoisson.create(scope, shape, rate, options);
-  }
-
-  /**
-   * Outputs random values from the Poisson distribution(s) described by rate.
-   *  <p>
-   *  This op uses two algorithms, depending on rate. If rate >= 10, then
-   *  the algorithm by Hormann is used to acquire samples via
-   *  transformation-rejection.
-   *  See http://www.sciencedirect.com/science/article/pii/0167668793909974.
-   *  <p>
-   *  Otherwise, Knuth's algorithm is used to acquire samples via multiplying uniform
-   *  random variables.
-   *  See Donald E. Knuth (1969). Seminumerical Algorithms. The Art of Computer
-   *  Programming, Volume 2. Addison Wesley
-   *
-   * @param <V> data type for {@code output()} output
-   * @param shape 1-D integer tensor. Shape of independent samples to draw from each
-   *  distribution described by the shape parameters given in rate.
-   * @param rate A tensor in which each scalar is a "rate" parameter describing the
-   *  associated poisson distribution.
-   * @param dtype
-   * @param options carries optional attributes values
-   * @return a new instance of RandomPoisson
-   */
-  public <V extends TNumber, T extends TNumber, U extends TNumber> RandomPoisson<V> randomPoisson(
-      Operand<T> shape, Operand<U> rate, DataType<V> dtype, RandomPoisson.Options... options) {
-    return RandomPoisson.create(scope, shape, rate, dtype, options);
-  }
-
-  /**
-   * Randomly shuffles a tensor along its first dimension.
-   *  <p>
-   *    The tensor is shuffled along dimension 0, such that each `value[j]` is mapped
-   *    to one and only one `output[i]`. For example, a mapping that might occur for a
-   *    3x2 tensor is:
-   *  <pre>{@code
-   *  [[1, 2],       [[5, 6],
-   *   [3, 4],  ==>   [1, 2],
-   *   [5, 6]]        [3, 4]]
-   *  }</pre>
-   *
-   * @param <T> data type for {@code output()} output
-   * @param value The tensor to be shuffled.
-   * @param options carries optional attributes values
-   * @return a new instance of RandomShuffle
-   */
-  public <T extends TType> RandomShuffle<T> randomShuffle(Operand<T> value,
-      RandomShuffle.Options... options) {
-    return RandomShuffle.create(scope, value, options);
-  }
-
-  /**
-   * Outputs random values from a normal distribution.
-   *  <p>
-   *  The generated values will have mean 0 and standard deviation 1.
-   *
-   * @param <U> data type for {@code output()} output
-   * @param shape The shape of the output tensor.
-   * @param dtype The type of the output.
-   * @param options carries optional attributes values
-   * @return a new instance of RandomStandardNormal
-   */
-  public <U extends TNumber, T extends TNumber> RandomStandardNormal<U> randomStandardNormal(
-      Operand<T> shape, DataType<U> dtype, RandomStandardNormal.Options... options) {
-    return RandomStandardNormal.create(scope, shape, dtype, options);
-  }
-
-  /**
-   * Outputs random values from a uniform distribution.
-   *  <p>
-   *  The generated values follow a uniform distribution in the range `[0, 1)`. The
-   *  lower bound 0 is included in the range, while the upper bound 1 is excluded.
-   *
-   * @param <U> data type for {@code output()} output
-   * @param shape The shape of the output tensor.
-   * @param dtype The type of the output.
-   * @param options carries optional attributes values
-   * @return a new instance of RandomUniform
-   */
-  public <U extends TNumber, T extends TNumber> RandomUniform<U> randomUniform(Operand<T> shape,
-      DataType<U> dtype, RandomUniform.Options... options) {
-    return RandomUniform.create(scope, shape, dtype, options);
-  }
-
-  /**
-   * Outputs random integers from a uniform distribution.
-   *  <p>
-   *  The generated values are uniform integers in the range `[minval, maxval)`.
-   *  The lower bound `minval` is included in the range, while the upper bound
-   *  `maxval` is excluded.
-   *  <p>
-   *  The random integers are slightly biased unless `maxval - minval` is an exact
-   *  power of two.  The bias is small for values of `maxval - minval` significantly
-   *  smaller than the range of the output (either `2^32` or `2^64`).
-   *
-   * @param <U> data type for {@code output()} output
-   * @param shape The shape of the output tensor.
-   * @param minval 0-D.  Inclusive lower bound on the generated integers.
-   * @param maxval 0-D.  Exclusive upper bound on the generated integers.
-   * @param options carries optional attributes values
-   * @return a new instance of RandomUniformInt
-   */
-  public <U extends TNumber, T extends TNumber> RandomUniformInt<U> randomUniformInt(
-      Operand<T> shape, Operand<U> minval, Operand<U> maxval, RandomUniformInt.Options... options) {
-    return RandomUniformInt.create(scope, shape, minval, maxval, options);
-  }
-
-  /**
-   * Emits randomized records.
-   *
-   * @param filePattern Glob pattern for the data files.
-   * @param options carries optional attributes values
-   * @return a new instance of RecordInput
-   */
-  public RecordInput recordInput(String filePattern, RecordInput.Options... options) {
-    return RecordInput.create(scope, filePattern, options);
-  }
-
-  /**
-   *
-   * @param <V> data type for {@code output()} output
-   * @param resource
-   * @param algorithm
-   * @param shape
-   * @param counts
-   * @param probs
-   * @return a new instance of StatefulRandomBinomial
-   */
-  public <T extends TNumber, U extends TNumber> StatefulRandomBinomial<TInt64> statefulRandomBinomial(
-      Operand<?> resource, Operand<TInt64> algorithm, Operand<T> shape, Operand<U> counts,
-      Operand<U> probs) {
-    return StatefulRandomBinomial.create(scope, resource, algorithm, shape, counts, probs);
-  }
-
-  /**
-   *
-   * @param <V> data type for {@code output()} output
-   * @param resource
-   * @param algorithm
-   * @param shape
-   * @param counts
-   * @param probs
-   * @param dtype
-   * @return a new instance of StatefulRandomBinomial
-   */
-  public <V extends TNumber, T extends TNumber, U extends TNumber> StatefulRandomBinomial<V> statefulRandomBinomial(
-      Operand<?> resource, Operand<TInt64> algorithm, Operand<T> shape, Operand<U> counts,
-      Operand<U> probs, DataType<V> dtype) {
-    return StatefulRandomBinomial.create(scope, resource, algorithm, shape, counts, probs, dtype);
-  }
-
-  /**
-   * Outputs random values from a normal distribution.
-   *  <p>
-   *  The generated values will have mean 0 and standard deviation 1.
-   *
-   * @param <U> data type for {@code output()} output
-   * @param resource The handle of the resource variable that stores the state of the RNG.
-   * @param algorithm The RNG algorithm.
-   * @param shape The shape of the output tensor.
-   * @return a new instance of StatefulStandardNormal
-   */
-  public <T extends TType> StatefulStandardNormal<TFloat32> statefulStandardNormal(
-      Operand<?> resource, Operand<TInt64> algorithm, Operand<T> shape) {
-    return StatefulStandardNormal.create(scope, resource, algorithm, shape);
-  }
-
-  /**
-   * Outputs random values from a normal distribution.
-   *  <p>
-   *  The generated values will have mean 0 and standard deviation 1.
-   *
-   * @param <U> data type for {@code output()} output
-   * @param resource The handle of the resource variable that stores the state of the RNG.
-   * @param algorithm The RNG algorithm.
-   * @param shape The shape of the output tensor.
-   * @param dtype The type of the output.
-   * @return a new instance of StatefulStandardNormal
-   */
-  public <U extends TType, T extends TType> StatefulStandardNormal<U> statefulStandardNormal(
-      Operand<?> resource, Operand<TInt64> algorithm, Operand<T> shape, DataType<U> dtype) {
-    return StatefulStandardNormal.create(scope, resource, algorithm, shape, dtype);
-  }
-
-  /**
-   * Draws samples from a multinomial distribution.
-   *
-   * @param <V> data type for {@code output()} output
-   * @param logits 2-D Tensor with shape `[batch_size, num_classes]`.  Each slice `[i, :]`
-   *  represents the unnormalized log probabilities for all classes.
-   * @param numSamples 0-D.  Number of independent samples to draw for each row slice.
-   * @param seed 2 seeds (shape [2]).
-   * @return a new instance of StatelessMultinomial
-   */
-  public <T extends TNumber, U extends TNumber> StatelessMultinomial<TInt64> statelessMultinomial(
-      Operand<T> logits, Operand<TInt32> numSamples, Operand<U> seed) {
-    return StatelessMultinomial.create(scope, logits, numSamples, seed);
-  }
-
-  /**
-   * Draws samples from a multinomial distribution.
-   *
-   * @param <V> data type for {@code output()} output
-   * @param logits 2-D Tensor with shape `[batch_size, num_classes]`.  Each slice `[i, :]`
-   *  represents the unnormalized log probabilities for all classes.
-   * @param numSamples 0-D.  Number of independent samples to draw for each row slice.
-   * @param seed 2 seeds (shape [2]).
-   * @param outputDtype
-   * @return a new instance of StatelessMultinomial
-   */
-  public <V extends TNumber, T extends TNumber, U extends TNumber> StatelessMultinomial<V> statelessMultinomial(
-      Operand<T> logits, Operand<TInt32> numSamples, Operand<U> seed, DataType<V> outputDtype) {
-    return StatelessMultinomial.create(scope, logits, numSamples, seed, outputDtype);
-  }
-
-  /**
-   * Outputs deterministic pseudorandom values from a normal distribution.
-   *  <p>
-   *  The generated values will have mean 0 and standard deviation 1.
-   *  <p>
-   *  The outputs are a deterministic function of `shape` and `seed`.
-   *
-   * @param <V> data type for {@code output()} output
-   * @param shape The shape of the output tensor.
-   * @param seed 2 seeds (shape [2]).
-   * @return a new instance of StatelessRandomNormal
-   */
-  public <T extends TNumber, U extends TNumber> StatelessRandomNormal<TFloat32> statelessRandomNormal(
-      Operand<T> shape, Operand<U> seed) {
-    return StatelessRandomNormal.create(scope, shape, seed);
-  }
-
-  /**
-   * Outputs deterministic pseudorandom values from a normal distribution.
-   *  <p>
-   *  The generated values will have mean 0 and standard deviation 1.
-   *  <p>
-   *  The outputs are a deterministic function of `shape` and `seed`.
-   *
-   * @param <V> data type for {@code output()} output
-   * @param shape The shape of the output tensor.
-   * @param seed 2 seeds (shape [2]).
-   * @param dtype The type of the output.
-   * @return a new instance of StatelessRandomNormal
-   */
-  public <V extends TNumber, T extends TNumber, U extends TNumber> StatelessRandomNormal<V> statelessRandomNormal(
-      Operand<T> shape, Operand<U> seed, DataType<V> dtype) {
-    return StatelessRandomNormal.create(scope, shape, seed, dtype);
-  }
-
-  /**
-   * Outputs deterministic pseudorandom random values from a uniform distribution.
-   *  <p>
-   *  The generated values follow a uniform distribution in the range `[0, 1)`. The
-   *  lower bound 0 is included in the range, while the upper bound 1 is excluded.
-   *  <p>
-   *  The outputs are a deterministic function of `shape` and `seed`.
-   *
-   * @param <V> data type for {@code output()} output
-   * @param shape The shape of the output tensor.
-   * @param seed 2 seeds (shape [2]).
-   * @return a new instance of StatelessRandomUniform
-   */
-  public <T extends TNumber, U extends TNumber> StatelessRandomUniform<TFloat32> statelessRandomUniform(
-      Operand<T> shape, Operand<U> seed) {
-    return StatelessRandomUniform.create(scope, shape, seed);
-  }
-
-  /**
-   * Outputs deterministic pseudorandom random values from a uniform distribution.
-   *  <p>
-   *  The generated values follow a uniform distribution in the range `[0, 1)`. The
-   *  lower bound 0 is included in the range, while the upper bound 1 is excluded.
-   *  <p>
-   *  The outputs are a deterministic function of `shape` and `seed`.
-   *
-   * @param <V> data type for {@code output()} output
-   * @param shape The shape of the output tensor.
-   * @param seed 2 seeds (shape [2]).
-   * @param dtype The type of the output.
-   * @return a new instance of StatelessRandomUniform
-   */
-  public <V extends TNumber, T extends TNumber, U extends TNumber> StatelessRandomUniform<V> statelessRandomUniform(
-      Operand<T> shape, Operand<U> seed, DataType<V> dtype) {
-    return StatelessRandomUniform.create(scope, shape, seed, dtype);
-  }
-
-  /**
-   * Outputs deterministic pseudorandom values from a truncated normal distribution.
-   *  <p>
-   *  The generated values follow a normal distribution with mean 0 and standard
-   *  deviation 1, except that values whose magnitude is more than 2 standard
-   *  deviations from the mean are dropped and re-picked.
-   *  <p>
-   *  The outputs are a deterministic function of `shape` and `seed`.
-   *
-   * @param <V> data type for {@code output()} output
-   * @param shape The shape of the output tensor.
-   * @param seed 2 seeds (shape [2]).
-   * @return a new instance of StatelessTruncatedNormal
-   */
-  public <T extends TNumber, U extends TNumber> StatelessTruncatedNormal<TFloat32> statelessTruncatedNormal(
-      Operand<T> shape, Operand<U> seed) {
-    return StatelessTruncatedNormal.create(scope, shape, seed);
-  }
-
-  /**
-   * Outputs deterministic pseudorandom values from a truncated normal distribution.
-   *  <p>
-   *  The generated values follow a normal distribution with mean 0 and standard
-   *  deviation 1, except that values whose magnitude is more than 2 standard
-   *  deviations from the mean are dropped and re-picked.
-   *  <p>
-   *  The outputs are a deterministic function of `shape` and `seed`.
-   *
-   * @param <V> data type for {@code output()} output
-   * @param shape The shape of the output tensor.
-   * @param seed 2 seeds (shape [2]).
-   * @param dtype The type of the output.
-   * @return a new instance of StatelessTruncatedNormal
-   */
-  public <V extends TNumber, T extends TNumber, U extends TNumber> StatelessTruncatedNormal<V> statelessTruncatedNormal(
-      Operand<T> shape, Operand<U> seed, DataType<V> dtype) {
-    return StatelessTruncatedNormal.create(scope, shape, seed, dtype);
-  }
-
-  /**
-   * Outputs random values from a truncated normal distribution.
-   *  <p>
-   *  The generated values follow a normal distribution with mean 0 and standard
-   *  deviation 1, except that values whose magnitude is more than 2 standard
-   *  deviations from the mean are dropped and re-picked.
-   *
-   * @param <U> data type for {@code output()} output
-   * @param shape The shape of the output tensor.
-   * @param dtype The type of the output.
-   * @param options carries optional attributes values
-   * @return a new instance of TruncatedNormal
-   */
-  public <U extends TNumber, T extends TNumber> TruncatedNormal<U> truncatedNormal(Operand<T> shape,
-      DataType<U> dtype, TruncatedNormal.Options... options) {
-    return TruncatedNormal.create(scope, shape, dtype, options);
-  }
-
-  /**
-   * Generates labels for candidate sampling with a uniform distribution.
-   *  <p>
-   *  See explanations of candidate sampling and the data formats at
-   *  go/candidate-sampling.
-   *  <p>
-   *  For each batch, this op picks a single set of sampled candidate labels.
-   *  <p>
-   *  The advantages of sampling candidates per-batch are simplicity and the
-   *  possibility of efficient dense matrix multiplication. The disadvantage is that
-   *  the sampled candidates must be chosen independently of the context and of the
-   *  true labels.
-   *
-   * @param trueClasses A batch_size * num_true matrix, in which each row contains the
-   *  IDs of the num_true target_classes in the corresponding original label.
-   * @param numTrue Number of true labels per context.
-   * @param numSampled Number of candidates to randomly sample.
-   * @param unique If unique is true, we sample with rejection, so that all sampled
-   *  candidates in a batch are unique. This requires some approximation to
-   *  estimate the post-rejection sampling probabilities.
-   * @param rangeMax The sampler will sample integers from the interval [0, range_max).
-   * @param options carries optional attributes values
-   * @return a new instance of UniformCandidateSampler
-   */
-  public UniformCandidateSampler uniformCandidateSampler(Operand<TInt64> trueClasses, Long numTrue,
-      Long numSampled, Boolean unique, Long rangeMax, UniformCandidateSampler.Options... options) {
-    return UniformCandidateSampler.create(scope, trueClasses, numTrue, numSampled, unique, rangeMax, options);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/ShapeOps.java b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/ShapeOps.java
deleted file mode 100644
index 81c692571f1..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/ShapeOps.java
+++ /dev/null
@@ -1,472 +0,0 @@
-// Copyright 2020 The TensorFlow Authors. All Rights Reserved.
-//
-// Licensed under the Apache License, Version 2.0 (the "License");
-// you may not use this file except in compliance with the License.
-// You may obtain a copy of the License at
-//
-//     http://www.apache.org/licenses/LICENSE-2.0
-//
-// Unless required by applicable law or agreed to in writing, software
-// distributed under the License is distributed on an "AS IS" BASIS,
-// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-// See the License for the specific language governing permissions and
-// limitations under the License.
-// ==============================================================================
-//
-// This class has been generated, DO NOT EDIT!
-//
-package org.tensorflow.op;
-
-import org.tensorflow.DataType;
-import org.tensorflow.Operand;
-import org.tensorflow.op.core.Shape;
-import org.tensorflow.op.core.Shapes;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * An API for building {@code shape} operations as {@link Op Op}s
- *
- * @see {@link Ops}
- */
-public final class ShapeOps {
-  private final Scope scope;
-
-  ShapeOps(Scope scope) {
-    this.scope = scope;
-  }
-
-  /**
-   * Creates a 1-dimensional operand containing the dimensions of a shape followed by the last
-   *  dimension.
-   *
-   * @param scope current scope
-   * @param shape the TensorFlow shape
-   * @param lastDimension the dimension(s) to append
-   * @return a 1-dimensional operand containing the dimensions of a shape followed by the last
-   *      dimension
-   */
-  public Operand<TInt64> append(Shape<TInt64> shape, long lastDimension) {
-    return Shapes.append(scope, shape, lastDimension);
-  }
-
-  /**
-   * Creates a 1-dimensional operand containing the dimensions of a shape followed by the last
-   *  dimension.
-   *
-   * @param scope current scope
-   * @param shape the TensorFlow shape
-   * @param lastDimension the dimension(s) to append
-   * @return a 1-dimensional operand containing the dimensions of a shape followed by the last
-   *      dimension
-   */
-  public Operand<TInt32> append(Shape<TInt32> shape, int lastDimension) {
-    return Shapes.append(scope, shape, lastDimension);
-  }
-
-  /**
-   * Creates a 1-dimensional operand that represents a new shape containing the dimensions of the
-   *  operand representing a shape, followed by the dimensions of an operand representing a shape to
-   *  append.
-   *
-   * @param scope current scope
-   * @param shape the TensorFlow shape
-   * @param shapeToAppend the other shape to append
-   * @return a 1-dimensional operand that represents a new shape containing the dimensions of the
-   *      operand representing a shape, followed by the dimensions of an operand representing a shape
-   *      to append
-   */
-  public <T extends TNumber> Operand<T> append(Operand<T> shape, Operand<T> shapeToAppend) {
-    return Shapes.append(scope, shape, shapeToAppend);
-  }
-
-  /**
-   * Flatten the operand to 1 dimension.
-   *
-   * @param <T> the type of operand
-   * @param scope current scope
-   * @param operand the operand to flatten
-   * @return the reshaped operand
-   */
-  public <T extends TType> Operand<T> flatten(Operand<T> operand) {
-    return Shapes.flatten(scope, operand);
-  }
-
-  /**
-   * Flatten the shape to 1 dimension.
-   *
-   * @param scope current scope
-   * @param shape the TensorFlow shape
-   * @return the flattened shape
-   */
-  public Operand<TInt32> flatten(Shape<TInt32> shape) {
-    return Shapes.flatten(scope, shape);
-  }
-
-  /**
-   * Flatten the operand to 1 dimension
-   *
-   * @param <T> the type of operand
-   * @param <U> the shape datatype
-   * @param scope current scope
-   * @param operand the operand to flatten
-   * @param dType the shape datatype
-   * @return the reshaped operand
-   */
-  public <T extends TType, U extends TNumber> Operand<T> flatten(Operand<T> operand,
-      DataType<U> dType) {
-    return Shapes.flatten(scope, operand, dType);
-  }
-
-  /**
-   * Flatten the shape to 1 dimension.
-   *
-   * @param <U> the shape datatype
-   * @param scope current scope
-   * @param shape the TensorFlow shape
-   * @param dType the shape datatype
-   * @return the flattened shape
-   */
-  public <U extends TNumber> Operand<U> flatten(Shape<U> shape, DataType<U> dType) {
-    return Shapes.flatten(scope, shape, dType);
-  }
-
-  /**
-   * Creates a 1-dimensional Operand containing the Shape's first dimension.
-   *
-   * @param scope current scope
-   * @param shape the TensorFlow shape
-   * @return a 1-dimensional Operand containing the Shape's first dimension
-   */
-  public Operand<TInt32> head(Shape<TInt32> shape) {
-    return Shapes.head(scope, shape);
-  }
-
-  /**
-   * Creates a 1-dimensional Operand containing the Shape's first dimension.
-   *
-   * @param scope current scope
-   * @param shape the TensorFlow shape
-   * @param dType the shape datatype.
-   * @param <U> the shape datatype.
-   * @return a 1-dimensional Operand containing the Shape's first dimension
-   */
-  public <U extends TNumber> Operand<U> head(Shape<U> shape, DataType<U> dType) {
-    return Shapes.head(scope, shape, dType);
-  }
-
-  /**
-   * Get the number of dimensions of the shape object.
-   *
-   * @param scope current scope
-   * @param shape the shape
-   * @return the number of dimensions
-   */
-  public Operand<TInt32> numDimensions(Shape<TInt32> shape) {
-    return Shapes.numDimensions(scope, shape);
-  }
-
-  /**
-   * Get the number of dimensions of the shape object.
-   *
-   * @param <U> the shape datatype
-   * @param scope the curren scope
-   * @param shape the shape
-   * @param dType the shape datatype
-   * @return the number of dimensions
-   */
-  public <U extends TNumber> Operand<U> numDimensions(Shape<U> shape, DataType<U> dType) {
-    return Shapes.numDimensions(scope, shape, dType);
-  }
-
-  /**
-   * Creates a 1-dimensional operand containing the first dimension followed by the dimensions of
-   *  the shape.
-   *
-   * @param scope current scope
-   * @param shape the TensorFlow shape
-   * @param firstDimension the dimension to prepend
-   * @return a 1-dimensional operand containing the first dimension followed by the dimensions of
-   *      the shape
-   */
-  public Operand<TInt64> prepend(Shape<TInt64> shape, long firstDimension) {
-    return Shapes.prepend(scope, shape, firstDimension);
-  }
-
-  /**
-   * Creates a 1-dimensional operand containing the first dimension followed by the dimensions of
-   *  the shape.
-   *
-   * @param scope current scope
-   * @param shape the TensorFlow shape
-   * @param firstDimension the dimension to prepend
-   * @return a 1-dimensional operand containing the first dimension followed by the dimensions of
-   *      the shape
-   */
-  public Operand<TInt32> prepend(Shape<TInt32> shape, int firstDimension) {
-    return Shapes.prepend(scope, shape, firstDimension);
-  }
-
-  /**
-   * Creates a 1-dimensional operand that represents a new shape containing the dimensions of an
-   *  operand representing the shape to prepend, followed by the dimensions of an operand
-   *  representing a shape.
-   *
-   * @param scope current scope
-   * @param shape an operand containing the dimensions of a shape
-   * @param shapeToPrepend an operand containing the dimensions of the shape to prepend
-   * @return a 1-dimensional operand that represents a new shape containing the dimensions of an
-   *      operand representing the shape to prepend, followed by the dimensions of an operand
-   *      representing the shape
-   */
-  public <T extends TNumber> Operand<T> prepend(Operand<T> shape, Operand<T> shapeToPrepend) {
-    return Shapes.prepend(scope, shape, shapeToPrepend);
-  }
-
-  /**
-   * Reshapes the operand by reducing the shape to the specified axis.
-   *
-   * @param <T> the type of Operand
-   * @param scope current scope
-   * @param operand the operand
-   * @param axis the axis
-   * @return the reshaped operand
-   */
-  public <T extends TType> Operand<T> reduceDims(Operand<T> operand, Operand<TInt32> axis) {
-    return Shapes.reduceDims(scope, operand, axis);
-  }
-
-  /**
-   * Reduces the shape to the specified axis.
-   *
-   * @param scope current scope
-   * @param shape the TensorFlow shape
-   * @param axis the axis
-   * @return an operand containing the dimensions for the reduced shape
-   */
-  public Operand<TInt32> reduceDims(Shape<TInt32> shape, Operand<TInt32> axis) {
-    return Shapes.reduceDims(scope, shape, axis);
-  }
-
-  /**
-   * Reshapes the operand by reducing the shape to the specified axis.
-   *
-   * @param <T> the type of Operand
-   * @param <U> the shape datatype
-   * @param scope current scope
-   * @param operand the operand
-   * @param axis the axis
-   * @param dType the shape datatype
-   * @return the reshaped operand
-   */
-  public <T extends TType, U extends TNumber> Operand<T> reduceDims(Operand<T> operand,
-      Operand<U> axis, DataType<U> dType) {
-    return Shapes.reduceDims(scope, operand, axis, dType);
-  }
-
-  /**
-   * Reduces the shape to the specified axis.
-   *
-   * @param <U> the shape datatype
-   * @param scope current scope
-   * @param shape the TensorFlow shape
-   * @param axis the axis
-   * @param dType the shape datatype
-   * @return the reduced shape
-   */
-  public <U extends TNumber> Operand<U> reduceDims(Shape<U> shape, Operand<U> axis,
-      DataType<U> dType) {
-    return Shapes.reduceDims(scope, shape, axis, dType);
-  }
-
-  /**
-   * Get the size represented by the TensorFlow shape.
-   *
-   * @param scope current scope
-   * @param shape the TensorFlow shape
-   * @return the size
-   */
-  public Operand<TInt32> size(Shape<TInt32> shape) {
-    return Shapes.size(scope, shape);
-  }
-
-  /**
-   * Get the size of the specified dimension for the shape of the tensor.
-   *
-   * @param scope current scope
-   * @param input the operand
-   * @param dim the dimension
-   * @return the size of the specified dimension
-   */
-  public <T extends TType> Operand<TInt32> size(Operand<T> input, Operand<TInt32> dim) {
-    return Shapes.size(scope, input, dim);
-  }
-
-  /**
-   * Get the size represented by the TensorFlow shape.
-   *
-   * @param <U> the type of the shape
-   * @param scope current scope
-   * @param shape the TensorFlow shape
-   * @param dType the shape datatype
-   * @return the size
-   */
-  public <U extends TNumber> Operand<U> size(Shape<U> shape, DataType<U> dType) {
-    return Shapes.size(scope, shape, dType);
-  }
-
-  /**
-   * Get the size of the specified dimension in the shape.
-   *
-   * @param scope current scope
-   * @param shape the TensorFlow shape
-   * @param dim the dimension
-   * @return the size of the specified dimension
-   */
-  public Operand<TInt32> size(Shape<TInt32> shape, Operand<TInt32> dim) {
-    return Shapes.size(scope, shape, dim);
-  }
-
-  /**
-   * Get the size of the specified dimension for the shape of the tensor.
-   *
-   * @param <U> the shape datatype
-   * @param scope current scope
-   * @param input the operand
-   * @param dim the dimension
-   * @param dType the shape datatype
-   * @return the size of the specified dimension
-   */
-  public <T extends TType, U extends TNumber> Operand<U> size(Operand<T> input, Operand<U> dim,
-      DataType<U> dType) {
-    return Shapes.size(scope, input, dim, dType);
-  }
-
-  /**
-   * Get the size of the specified dimension in the shape.
-   *
-   * @param <U> the shape datatype
-   * @param scope current scope
-   * @param shape the TensorFlow shape
-   * @param dim the dimension
-   * @param dType the shape datatype
-   * @return the size of the specified dimension
-   */
-  public <U extends TNumber> Operand<U> size(Shape<U> shape, Operand<U> dim, DataType<U> dType) {
-    return Shapes.size(scope, shape, dim, dType);
-  }
-
-  /**
-   * Removes dimensions of size 1 from the shape.
-   *
-   * @param scope current scope
-   * @param shape the TensorFlow shape
-   * @return the squeezed shape
-   */
-  public Operand<TInt32> squeeze(Shape<TInt32> shape) {
-    return Shapes.squeeze(scope, shape);
-  }
-
-  /**
-   * Removes dimensions of size 1 from the shape.
-   *
-   * @param <U> the shape datatype.
-   * @param scope current scope
-   * @param shape the TensorFlow shape
-   * @param dType the shape datatype.
-   * @return the squeezed shape
-   */
-  public <U extends TNumber> Operand<U> squeeze(Shape<U> shape, DataType<U> dType) {
-    return Shapes.squeeze(scope, shape, dType);
-  }
-
-  /**
-   * Creates a 1-dimensional Operand that contains the dimension matching the last dimension of the
-   *  Shape.
-   *
-   * @param scope current scope
-   * @param shape the TensorFlow shape
-   * @return a 1-dimensional Operand that contains the dimension matching the last dimension of the
-   *      Shape
-   */
-  public Operand<TInt32> tail(Shape<TInt32> shape) {
-    return Shapes.tail(scope, shape);
-  }
-
-  /**
-   * Creates a 1-dimensional Operand that contains the dimension matching the last dimension of *
-   *  the Shape.
-   *
-   * @param scope current scope
-   * @param shape the TensorFlow shape
-   * @param dType the shape datatype.
-   * @param <U> the shape datatype.
-   * @return a 1-dimensional Operand that contains the dimension matching the last dimension of the
-   *      Shape
-   */
-  public <U extends TNumber> Operand<U> tail(Shape<U> shape, DataType<U> dType) {
-    return Shapes.tail(scope, shape, dType);
-  }
-
-  /**
-   * Creates a 1-dimensional operand with the dimensions matching the first n dimensions of the
-   *  shape.
-   *
-   * @param scope current scope
-   * @param shape the TensorFlow shape
-   * @param n the number of leading dimensions to get, must be <= than the shape's numDimensions()
-   * @return a 1-dimensional operand with the dimensions matching the first n dimensions of the
-   *      shape
-   */
-  public Operand<TInt32> take(Shape<TInt32> shape, Operand<TInt32> n) {
-    return Shapes.take(scope, shape, n);
-  }
-
-  /**
-   * Creates a 1-dimensional operand containin the dimensions matching the first n dimensions of the
-   *  shape.
-   *
-   * @param scope current scope
-   * @param shape the TensorFlow shape
-   * @param n the number of leading dimensions to get, must be <= than the shape's numDimensions()
-   * @param dType the shape datatype.
-   * @param <U> the shape datatype.
-   * @return a 1-dimensional operand with the dimensions matching * the first n dimensions of the
-   *      shape
-   */
-  public <U extends TNumber> Operand<U> take(Shape<U> shape, Operand<U> n, DataType<U> dType) {
-    return Shapes.take(scope, shape, n, dType);
-  }
-
-  /**
-   * Creates a 1-dimensional operand containing the dimensions matching the last n dimensions of the
-   *  shape.
-   *
-   * @param scope current scope
-   * @param shape the TensorFlow shape
-   * @param n the number of leading dimensions to get, must be <= than the shape's numDimensions()
-   * @return a 1-dimensional operand containing the dimensions matching the last n dimensions of the
-   *      shape
-   */
-  public <U extends TNumber> Operand<TInt32> takeLast(Shape<TInt32> shape, Operand<TInt32> n) {
-    return Shapes.takeLast(scope, shape, n);
-  }
-
-  /**
-   * Creates a 1-dimensional operand containing the dimensions matching the last n dimensions of the
-   *  shape.
-   *
-   * @param scope current scope
-   * @param shape the TensorFlow shape
-   * @param n the number of leading dimensions to get, must be <= than the shape's numDimensions()
-   * @param dType the shape datatype.
-   * @param <U> the shape datatype.
-   * @return a 1-dimensional operand containing the dimensions matching the last n dimensions of the
-   *      shape
-   */
-  public <U extends TNumber> Operand<U> takeLast(Shape<U> shape, Operand<U> n, DataType<U> dType) {
-    return Shapes.takeLast(scope, shape, n, dType);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/SignalOps.java b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/SignalOps.java
deleted file mode 100644
index f4ec7bdb48d..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/SignalOps.java
+++ /dev/null
@@ -1,437 +0,0 @@
-// Copyright 2020 The TensorFlow Authors. All Rights Reserved.
-//
-// Licensed under the Apache License, Version 2.0 (the "License");
-// you may not use this file except in compliance with the License.
-// You may obtain a copy of the License at
-//
-//     http://www.apache.org/licenses/LICENSE-2.0
-//
-// Unless required by applicable law or agreed to in writing, software
-// distributed under the License is distributed on an "AS IS" BASIS,
-// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-// See the License for the specific language governing permissions and
-// limitations under the License.
-// ==============================================================================
-//
-// This class has been generated, DO NOT EDIT!
-//
-package org.tensorflow.op;
-
-import org.tensorflow.DataType;
-import org.tensorflow.Operand;
-import org.tensorflow.op.signal.BatchFft;
-import org.tensorflow.op.signal.BatchFft2d;
-import org.tensorflow.op.signal.BatchFft3d;
-import org.tensorflow.op.signal.BatchIfft;
-import org.tensorflow.op.signal.BatchIfft2d;
-import org.tensorflow.op.signal.BatchIfft3d;
-import org.tensorflow.op.signal.Fft;
-import org.tensorflow.op.signal.Fft2d;
-import org.tensorflow.op.signal.Fft3d;
-import org.tensorflow.op.signal.Ifft;
-import org.tensorflow.op.signal.Ifft2d;
-import org.tensorflow.op.signal.Ifft3d;
-import org.tensorflow.op.signal.Irfft;
-import org.tensorflow.op.signal.Irfft2d;
-import org.tensorflow.op.signal.Irfft3d;
-import org.tensorflow.op.signal.Rfft;
-import org.tensorflow.op.signal.Rfft2d;
-import org.tensorflow.op.signal.Rfft3d;
-import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * An API for building {@code signal} operations as {@link Op Op}s
- *
- * @see {@link Ops}
- */
-public final class SignalOps {
-  private final Scope scope;
-
-  SignalOps(Scope scope) {
-    this.scope = scope;
-  }
-
-  /**
-   *
-   * @param input
-   * @return a new instance of BatchFft
-   */
-  public BatchFft batchFft(Operand<?> input) {
-    return BatchFft.create(scope, input);
-  }
-
-  /**
-   *
-   * @param input
-   * @return a new instance of BatchFft2d
-   */
-  public BatchFft2d batchFft2d(Operand<?> input) {
-    return BatchFft2d.create(scope, input);
-  }
-
-  /**
-   *
-   * @param input
-   * @return a new instance of BatchFft3d
-   */
-  public BatchFft3d batchFft3d(Operand<?> input) {
-    return BatchFft3d.create(scope, input);
-  }
-
-  /**
-   *
-   * @param input
-   * @return a new instance of BatchIfft
-   */
-  public BatchIfft batchIfft(Operand<?> input) {
-    return BatchIfft.create(scope, input);
-  }
-
-  /**
-   *
-   * @param input
-   * @return a new instance of BatchIfft2d
-   */
-  public BatchIfft2d batchIfft2d(Operand<?> input) {
-    return BatchIfft2d.create(scope, input);
-  }
-
-  /**
-   *
-   * @param input
-   * @return a new instance of BatchIfft3d
-   */
-  public BatchIfft3d batchIfft3d(Operand<?> input) {
-    return BatchIfft3d.create(scope, input);
-  }
-
-  /**
-   * Fast Fourier transform.
-   *  <p>
-   *  Computes the 1-dimensional discrete Fourier transform over the inner-most
-   *  dimension of `input`.
-   *
-   * @param <T> data type for {@code output()} output
-   * @param input A complex tensor.
-   * @return a new instance of Fft
-   */
-  public <T extends TType> Fft<T> fft(Operand<T> input) {
-    return Fft.create(scope, input);
-  }
-
-  /**
-   * 2D fast Fourier transform.
-   *  <p>
-   *  Computes the 2-dimensional discrete Fourier transform over the inner-most
-   *  2 dimensions of `input`.
-   *
-   * @param <T> data type for {@code output()} output
-   * @param input A complex tensor.
-   * @return a new instance of Fft2d
-   */
-  public <T extends TType> Fft2d<T> fft2d(Operand<T> input) {
-    return Fft2d.create(scope, input);
-  }
-
-  /**
-   * 3D fast Fourier transform.
-   *  <p>
-   *  Computes the 3-dimensional discrete Fourier transform over the inner-most 3
-   *  dimensions of `input`.
-   *
-   * @param <T> data type for {@code output()} output
-   * @param input A complex tensor.
-   * @return a new instance of Fft3d
-   */
-  public <T extends TType> Fft3d<T> fft3d(Operand<T> input) {
-    return Fft3d.create(scope, input);
-  }
-
-  /**
-   * Inverse fast Fourier transform.
-   *  <p>
-   *  Computes the inverse 1-dimensional discrete Fourier transform over the
-   *  inner-most dimension of `input`.
-   *
-   * @param <T> data type for {@code output()} output
-   * @param input A complex tensor.
-   * @return a new instance of Ifft
-   */
-  public <T extends TType> Ifft<T> ifft(Operand<T> input) {
-    return Ifft.create(scope, input);
-  }
-
-  /**
-   * Inverse 2D fast Fourier transform.
-   *  <p>
-   *  Computes the inverse 2-dimensional discrete Fourier transform over the
-   *  inner-most 2 dimensions of `input`.
-   *
-   * @param <T> data type for {@code output()} output
-   * @param input A complex tensor.
-   * @return a new instance of Ifft2d
-   */
-  public <T extends TType> Ifft2d<T> ifft2d(Operand<T> input) {
-    return Ifft2d.create(scope, input);
-  }
-
-  /**
-   * Inverse 3D fast Fourier transform.
-   *  <p>
-   *  Computes the inverse 3-dimensional discrete Fourier transform over the
-   *  inner-most 3 dimensions of `input`.
-   *
-   * @param <T> data type for {@code output()} output
-   * @param input A complex tensor.
-   * @return a new instance of Ifft3d
-   */
-  public <T extends TType> Ifft3d<T> ifft3d(Operand<T> input) {
-    return Ifft3d.create(scope, input);
-  }
-
-  /**
-   * Inverse real-valued fast Fourier transform.
-   *  <p>
-   *  Computes the inverse 1-dimensional discrete Fourier transform of a real-valued
-   *  signal over the inner-most dimension of `input`.
-   *  <p>
-   *  The inner-most dimension of `input` is assumed to be the result of `RFFT`: the
-   *  `fft_length / 2 + 1` unique components of the DFT of a real-valued signal. If
-   *  `fft_length` is not provided, it is computed from the size of the inner-most
-   *  dimension of `input` (`fft_length = 2 * (inner - 1)`). If the FFT length used to
-   *  compute `input` is odd, it should be provided since it cannot be inferred
-   *  properly.
-   *  <p>
-   *  Along the axis `signal.Irfft` is computed on, if `fft_length / 2 + 1` is smaller
-   *  than the corresponding dimension of `input`, the dimension is cropped. If it is
-   *  larger, the dimension is padded with zeros.
-   *
-   * @param <U> data type for {@code output()} output
-   * @param input A complex tensor.
-   * @param fftLength An int32 tensor of shape [1]. The FFT length.
-   * @return a new instance of Irfft
-   */
-  public <T extends TType> Irfft<TFloat32> irfft(Operand<T> input, Operand<TInt32> fftLength) {
-    return Irfft.create(scope, input, fftLength);
-  }
-
-  /**
-   * Inverse real-valued fast Fourier transform.
-   *  <p>
-   *  Computes the inverse 1-dimensional discrete Fourier transform of a real-valued
-   *  signal over the inner-most dimension of `input`.
-   *  <p>
-   *  The inner-most dimension of `input` is assumed to be the result of `RFFT`: the
-   *  `fft_length / 2 + 1` unique components of the DFT of a real-valued signal. If
-   *  `fft_length` is not provided, it is computed from the size of the inner-most
-   *  dimension of `input` (`fft_length = 2 * (inner - 1)`). If the FFT length used to
-   *  compute `input` is odd, it should be provided since it cannot be inferred
-   *  properly.
-   *  <p>
-   *  Along the axis `signal.Irfft` is computed on, if `fft_length / 2 + 1` is smaller
-   *  than the corresponding dimension of `input`, the dimension is cropped. If it is
-   *  larger, the dimension is padded with zeros.
-   *
-   * @param <U> data type for {@code output()} output
-   * @param input A complex tensor.
-   * @param fftLength An int32 tensor of shape [1]. The FFT length.
-   * @param Treal
-   * @return a new instance of Irfft
-   */
-  public <U extends TNumber, T extends TType> Irfft<U> irfft(Operand<T> input,
-      Operand<TInt32> fftLength, DataType<U> Treal) {
-    return Irfft.create(scope, input, fftLength, Treal);
-  }
-
-  /**
-   * Inverse 2D real-valued fast Fourier transform.
-   *  <p>
-   *  Computes the inverse 2-dimensional discrete Fourier transform of a real-valued
-   *  signal over the inner-most 2 dimensions of `input`.
-   *  <p>
-   *  The inner-most 2 dimensions of `input` are assumed to be the result of `RFFT2D`:
-   *  The inner-most dimension contains the `fft_length / 2 + 1` unique components of
-   *  the DFT of a real-valued signal. If `fft_length` is not provided, it is computed
-   *  from the size of the inner-most 2 dimensions of `input`. If the FFT length used
-   *  to compute `input` is odd, it should be provided since it cannot be inferred
-   *  properly.
-   *  <p>
-   *  Along each axis `signal.Irfft2d` is computed on, if `fft_length` (or
-   *  `fft_length / 2 + 1` for the inner-most dimension) is smaller than the
-   *  corresponding dimension of `input`, the dimension is cropped. If it is larger,
-   *  the dimension is padded with zeros.
-   *
-   * @param <U> data type for {@code output()} output
-   * @param input A complex tensor.
-   * @param fftLength An int32 tensor of shape [2]. The FFT length for each dimension.
-   * @return a new instance of Irfft2d
-   */
-  public <T extends TType> Irfft2d<TFloat32> irfft2d(Operand<T> input, Operand<TInt32> fftLength) {
-    return Irfft2d.create(scope, input, fftLength);
-  }
-
-  /**
-   * Inverse 2D real-valued fast Fourier transform.
-   *  <p>
-   *  Computes the inverse 2-dimensional discrete Fourier transform of a real-valued
-   *  signal over the inner-most 2 dimensions of `input`.
-   *  <p>
-   *  The inner-most 2 dimensions of `input` are assumed to be the result of `RFFT2D`:
-   *  The inner-most dimension contains the `fft_length / 2 + 1` unique components of
-   *  the DFT of a real-valued signal. If `fft_length` is not provided, it is computed
-   *  from the size of the inner-most 2 dimensions of `input`. If the FFT length used
-   *  to compute `input` is odd, it should be provided since it cannot be inferred
-   *  properly.
-   *  <p>
-   *  Along each axis `signal.Irfft2d` is computed on, if `fft_length` (or
-   *  `fft_length / 2 + 1` for the inner-most dimension) is smaller than the
-   *  corresponding dimension of `input`, the dimension is cropped. If it is larger,
-   *  the dimension is padded with zeros.
-   *
-   * @param <U> data type for {@code output()} output
-   * @param input A complex tensor.
-   * @param fftLength An int32 tensor of shape [2]. The FFT length for each dimension.
-   * @param Treal
-   * @return a new instance of Irfft2d
-   */
-  public <U extends TNumber, T extends TType> Irfft2d<U> irfft2d(Operand<T> input,
-      Operand<TInt32> fftLength, DataType<U> Treal) {
-    return Irfft2d.create(scope, input, fftLength, Treal);
-  }
-
-  /**
-   * Inverse 3D real-valued fast Fourier transform.
-   *  <p>
-   *  Computes the inverse 3-dimensional discrete Fourier transform of a real-valued
-   *  signal over the inner-most 3 dimensions of `input`.
-   *  <p>
-   *  The inner-most 3 dimensions of `input` are assumed to be the result of `RFFT3D`:
-   *  The inner-most dimension contains the `fft_length / 2 + 1` unique components of
-   *  the DFT of a real-valued signal. If `fft_length` is not provided, it is computed
-   *  from the size of the inner-most 3 dimensions of `input`. If the FFT length used
-   *  to compute `input` is odd, it should be provided since it cannot be inferred
-   *  properly.
-   *  <p>
-   *  Along each axis `signal.Irfft3d` is computed on, if `fft_length` (or
-   *  `fft_length / 2 + 1` for the inner-most dimension) is smaller than the
-   *  corresponding dimension of `input`, the dimension is cropped. If it is larger,
-   *  the dimension is padded with zeros.
-   *
-   * @param <U> data type for {@code output()} output
-   * @param input A complex tensor.
-   * @param fftLength An int32 tensor of shape [3]. The FFT length for each dimension.
-   * @return a new instance of Irfft3d
-   */
-  public <T extends TType> Irfft3d<TFloat32> irfft3d(Operand<T> input, Operand<TInt32> fftLength) {
-    return Irfft3d.create(scope, input, fftLength);
-  }
-
-  /**
-   * Inverse 3D real-valued fast Fourier transform.
-   *  <p>
-   *  Computes the inverse 3-dimensional discrete Fourier transform of a real-valued
-   *  signal over the inner-most 3 dimensions of `input`.
-   *  <p>
-   *  The inner-most 3 dimensions of `input` are assumed to be the result of `RFFT3D`:
-   *  The inner-most dimension contains the `fft_length / 2 + 1` unique components of
-   *  the DFT of a real-valued signal. If `fft_length` is not provided, it is computed
-   *  from the size of the inner-most 3 dimensions of `input`. If the FFT length used
-   *  to compute `input` is odd, it should be provided since it cannot be inferred
-   *  properly.
-   *  <p>
-   *  Along each axis `signal.Irfft3d` is computed on, if `fft_length` (or
-   *  `fft_length / 2 + 1` for the inner-most dimension) is smaller than the
-   *  corresponding dimension of `input`, the dimension is cropped. If it is larger,
-   *  the dimension is padded with zeros.
-   *
-   * @param <U> data type for {@code output()} output
-   * @param input A complex tensor.
-   * @param fftLength An int32 tensor of shape [3]. The FFT length for each dimension.
-   * @param Treal
-   * @return a new instance of Irfft3d
-   */
-  public <U extends TNumber, T extends TType> Irfft3d<U> irfft3d(Operand<T> input,
-      Operand<TInt32> fftLength, DataType<U> Treal) {
-    return Irfft3d.create(scope, input, fftLength, Treal);
-  }
-
-  /**
-   * Real-valued fast Fourier transform.
-   *  <p>
-   *  Computes the 1-dimensional discrete Fourier transform of a real-valued signal
-   *  over the inner-most dimension of `input`.
-   *  <p>
-   *  Since the DFT of a real signal is Hermitian-symmetric, `signal.Rfft` only returns the
-   *  `fft_length / 2 + 1` unique components of the FFT: the zero-frequency term,
-   *  followed by the `fft_length / 2` positive-frequency terms.
-   *  <p>
-   *  Along the axis `signal.Rfft` is computed on, if `fft_length` is smaller than the
-   *  corresponding dimension of `input`, the dimension is cropped. If it is larger,
-   *  the dimension is padded with zeros.
-   *
-   * @param <U> data type for {@code output()} output
-   * @param input A float32 tensor.
-   * @param fftLength An int32 tensor of shape [1]. The FFT length.
-   * @param Tcomplex
-   * @return a new instance of Rfft
-   */
-  public <U extends TType, T extends TNumber> Rfft<U> rfft(Operand<T> input,
-      Operand<TInt32> fftLength, DataType<U> Tcomplex) {
-    return Rfft.create(scope, input, fftLength, Tcomplex);
-  }
-
-  /**
-   * 2D real-valued fast Fourier transform.
-   *  <p>
-   *  Computes the 2-dimensional discrete Fourier transform of a real-valued signal
-   *  over the inner-most 2 dimensions of `input`.
-   *  <p>
-   *  Since the DFT of a real signal is Hermitian-symmetric, `signal.Rfft2d` only returns the
-   *  `fft_length / 2 + 1` unique components of the FFT for the inner-most dimension
-   *  of `output`: the zero-frequency term, followed by the `fft_length / 2`
-   *  positive-frequency terms.
-   *  <p>
-   *  Along each axis `signal.Rfft2d` is computed on, if `fft_length` is smaller than the
-   *  corresponding dimension of `input`, the dimension is cropped. If it is larger,
-   *  the dimension is padded with zeros.
-   *
-   * @param <U> data type for {@code output()} output
-   * @param input A float32 tensor.
-   * @param fftLength An int32 tensor of shape [2]. The FFT length for each dimension.
-   * @param Tcomplex
-   * @return a new instance of Rfft2d
-   */
-  public <U extends TType, T extends TNumber> Rfft2d<U> rfft2d(Operand<T> input,
-      Operand<TInt32> fftLength, DataType<U> Tcomplex) {
-    return Rfft2d.create(scope, input, fftLength, Tcomplex);
-  }
-
-  /**
-   * 3D real-valued fast Fourier transform.
-   *  <p>
-   *  Computes the 3-dimensional discrete Fourier transform of a real-valued signal
-   *  over the inner-most 3 dimensions of `input`.
-   *  <p>
-   *  Since the DFT of a real signal is Hermitian-symmetric, `signal.Rfft3d` only returns the
-   *  `fft_length / 2 + 1` unique components of the FFT for the inner-most dimension
-   *  of `output`: the zero-frequency term, followed by the `fft_length / 2`
-   *  positive-frequency terms.
-   *  <p>
-   *  Along each axis `signal.Rfft3d` is computed on, if `fft_length` is smaller than the
-   *  corresponding dimension of `input`, the dimension is cropped. If it is larger,
-   *  the dimension is padded with zeros.
-   *
-   * @param <U> data type for {@code output()} output
-   * @param input A float32 tensor.
-   * @param fftLength An int32 tensor of shape [3]. The FFT length for each dimension.
-   * @param Tcomplex
-   * @return a new instance of Rfft3d
-   */
-  public <U extends TType, T extends TNumber> Rfft3d<U> rfft3d(Operand<T> input,
-      Operand<TInt32> fftLength, DataType<U> Tcomplex) {
-    return Rfft3d.create(scope, input, fftLength, Tcomplex);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/SparseOps.java b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/SparseOps.java
deleted file mode 100644
index 6e26fec2275..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/SparseOps.java
+++ /dev/null
@@ -1,1422 +0,0 @@
-// Copyright 2020 The TensorFlow Authors. All Rights Reserved.
-//
-// Licensed under the Apache License, Version 2.0 (the "License");
-// you may not use this file except in compliance with the License.
-// You may obtain a copy of the License at
-//
-//     http://www.apache.org/licenses/LICENSE-2.0
-//
-// Unless required by applicable law or agreed to in writing, software
-// distributed under the License is distributed on an "AS IS" BASIS,
-// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-// See the License for the specific language governing permissions and
-// limitations under the License.
-// ==============================================================================
-//
-// This class has been generated, DO NOT EDIT!
-//
-package org.tensorflow.op;
-
-import org.tensorflow.DataType;
-import org.tensorflow.Operand;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.op.sparse.AddManySparseToTensorsMap;
-import org.tensorflow.op.sparse.AddSparseToTensorsMap;
-import org.tensorflow.op.sparse.DenseToDenseSetOperation;
-import org.tensorflow.op.sparse.DenseToSparseSetOperation;
-import org.tensorflow.op.sparse.DeserializeSparse;
-import org.tensorflow.op.sparse.SparseAccumulatorApplyGradient;
-import org.tensorflow.op.sparse.SparseAccumulatorTakeGradient;
-import org.tensorflow.op.sparse.SparseAdd;
-import org.tensorflow.op.sparse.SparseAddGrad;
-import org.tensorflow.op.sparse.SparseConcat;
-import org.tensorflow.op.sparse.SparseConditionalAccumulator;
-import org.tensorflow.op.sparse.SparseCross;
-import org.tensorflow.op.sparse.SparseDenseCwiseAdd;
-import org.tensorflow.op.sparse.SparseDenseCwiseDiv;
-import org.tensorflow.op.sparse.SparseDenseCwiseMul;
-import org.tensorflow.op.sparse.SparseFillEmptyRows;
-import org.tensorflow.op.sparse.SparseFillEmptyRowsGrad;
-import org.tensorflow.op.sparse.SparseMatMul;
-import org.tensorflow.op.sparse.SparseReduceMax;
-import org.tensorflow.op.sparse.SparseReduceMaxSparse;
-import org.tensorflow.op.sparse.SparseReduceSum;
-import org.tensorflow.op.sparse.SparseReduceSumSparse;
-import org.tensorflow.op.sparse.SparseReorder;
-import org.tensorflow.op.sparse.SparseReshape;
-import org.tensorflow.op.sparse.SparseSegmentMean;
-import org.tensorflow.op.sparse.SparseSegmentMeanGrad;
-import org.tensorflow.op.sparse.SparseSegmentMeanWithNumSegments;
-import org.tensorflow.op.sparse.SparseSegmentSqrtN;
-import org.tensorflow.op.sparse.SparseSegmentSqrtNGrad;
-import org.tensorflow.op.sparse.SparseSegmentSqrtNWithNumSegments;
-import org.tensorflow.op.sparse.SparseSegmentSum;
-import org.tensorflow.op.sparse.SparseSegmentSumWithNumSegments;
-import org.tensorflow.op.sparse.SparseSlice;
-import org.tensorflow.op.sparse.SparseSliceGrad;
-import org.tensorflow.op.sparse.SparseSoftmax;
-import org.tensorflow.op.sparse.SparseSparseMaximum;
-import org.tensorflow.op.sparse.SparseSparseMinimum;
-import org.tensorflow.op.sparse.SparseSplit;
-import org.tensorflow.op.sparse.SparseTensorDenseAdd;
-import org.tensorflow.op.sparse.SparseTensorDenseMatMul;
-import org.tensorflow.op.sparse.SparseToDense;
-import org.tensorflow.op.sparse.SparseToSparseSetOperation;
-import org.tensorflow.op.sparse.TakeManySparseFromTensorsMap;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.TString;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * An API for building {@code sparse} operations as {@link Op Op}s
- *
- * @see {@link Ops}
- */
-public final class SparseOps {
-  private final Scope scope;
-
-  SparseOps(Scope scope) {
-    this.scope = scope;
-  }
-
-  /**
-   * Add an `N`-minibatch `SparseTensor` to a `SparseTensorsMap`, return `N` handles.
-   *  <p>
-   *  A `SparseTensor` of rank `R` is represented by three tensors: `sparse_indices`,
-   *  `sparse_values`, and `sparse_shape`, where
-   *  <pre>{@code
-   *  sparse_indices.shape[1] == sparse_shape.shape[0] == R}</pre>
-   *  An `N`-minibatch of `SparseTensor` objects is represented as a `SparseTensor`
-   *  having a first `sparse_indices` column taking values between `[0, N)`, where
-   *  the minibatch size `N == sparse_shape[0]`.
-   *  <p>
-   *  The input `SparseTensor` must have rank `R` greater than 1, and the first
-   *  dimension is treated as the minibatch dimension.  Elements of the `SparseTensor`
-   *  must be sorted in increasing order of this first dimension.  The stored
-   *  `SparseTensor` objects pointed to by each row of the output `sparse_handles`
-   *  will have rank `R-1`.
-   *  <p>
-   *  The `SparseTensor` values can then be read out as part of a minibatch by passing
-   *  the given keys as vector elements to `TakeManySparseFromTensorsMap`.  To ensure
-   *  the correct `SparseTensorsMap` is accessed, ensure that the same
-   *  `container` and `shared_name` are passed to that Op.  If no `shared_name`
-   *  is provided here, instead use the <i>name</i> of the Operation created by calling
-   *  `sparse.AddManySparseToTensorsMap` as the `shared_name` passed to
-   *  `TakeManySparseFromTensorsMap`.  Ensure the Operations are colocated.
-   *
-   * @param sparseIndices 2-D.  The `indices` of the minibatch `SparseTensor`.
-   *  `sparse_indices[:, 0]` must be ordered values in `[0, N)`.
-   * @param sparseValues 1-D.  The `values` of the minibatch `SparseTensor`.
-   * @param sparseShape 1-D.  The `shape` of the minibatch `SparseTensor`.
-   *  The minibatch size `N == sparse_shape[0]`.
-   * @param options carries optional attributes values
-   * @return a new instance of AddManySparseToTensorsMap
-   */
-  public <T extends TType> AddManySparseToTensorsMap addManySparseToTensorsMap(
-      Operand<TInt64> sparseIndices, Operand<T> sparseValues, Operand<TInt64> sparseShape,
-      AddManySparseToTensorsMap.Options... options) {
-    return AddManySparseToTensorsMap.create(scope, sparseIndices, sparseValues, sparseShape, options);
-  }
-
-  /**
-   * Add a `SparseTensor` to a `SparseTensorsMap` return its handle.
-   *  <p>
-   *  A `SparseTensor` is represented by three tensors: `sparse_indices`,
-   *  `sparse_values`, and `sparse_shape`.
-   *  <p>
-   *  This operator takes the given `SparseTensor` and adds it to a container
-   *  object (a `SparseTensorsMap`).  A unique key within this container is generated
-   *  in the form of an `int64`, and this is the value that is returned.
-   *  <p>
-   *  The `SparseTensor` can then be read out as part of a minibatch by passing
-   *  the key as a vector element to `TakeManySparseFromTensorsMap`.  To ensure
-   *  the correct `SparseTensorsMap` is accessed, ensure that the same
-   *  `container` and `shared_name` are passed to that Op.  If no `shared_name`
-   *  is provided here, instead use the <i>name</i> of the Operation created by calling
-   *  `sparse.AddSparseToTensorsMap` as the `shared_name` passed to
-   *  `TakeManySparseFromTensorsMap`.  Ensure the Operations are colocated.
-   *
-   * @param sparseIndices 2-D.  The `indices` of the `SparseTensor`.
-   * @param sparseValues 1-D.  The `values` of the `SparseTensor`.
-   * @param sparseShape 1-D.  The `shape` of the `SparseTensor`.
-   * @param options carries optional attributes values
-   * @return a new instance of AddSparseToTensorsMap
-   */
-  public <T extends TType> AddSparseToTensorsMap addSparseToTensorsMap(
-      Operand<TInt64> sparseIndices, Operand<T> sparseValues, Operand<TInt64> sparseShape,
-      AddSparseToTensorsMap.Options... options) {
-    return AddSparseToTensorsMap.create(scope, sparseIndices, sparseValues, sparseShape, options);
-  }
-
-  /**
-   * Applies set operation along last dimension of 2 `Tensor` inputs.
-   *  <p>
-   *  See SetOperationOp::SetOperationFromContext for values of `set_operation`.
-   *  <p>
-   *  Output `result` is a `SparseTensor` represented by `result_indices`,
-   *  `result_values`, and `result_shape`. For `set1` and `set2` ranked `n`, this
-   *  has rank `n` and the same 1st `n-1` dimensions as `set1` and `set2`. The `nth`
-   *  dimension contains the result of `set_operation` applied to the corresponding
-   *  `[0...n-1]` dimension of `set`.
-   *
-   * @param <T> data type for {@code resultValues()} output
-   * @param set1 `Tensor` with rank `n`. 1st `n-1` dimensions must be the same as `set2`.
-   *  Dimension `n` contains values in a set, duplicates are allowed but ignored.
-   * @param set2 `Tensor` with rank `n`. 1st `n-1` dimensions must be the same as `set1`.
-   *  Dimension `n` contains values in a set, duplicates are allowed but ignored.
-   * @param setOperation
-   * @param options carries optional attributes values
-   * @return a new instance of DenseToDenseSetOperation
-   */
-  public <T extends TType> DenseToDenseSetOperation<T> denseToDenseSetOperation(Operand<T> set1,
-      Operand<T> set2, String setOperation, DenseToDenseSetOperation.Options... options) {
-    return DenseToDenseSetOperation.create(scope, set1, set2, setOperation, options);
-  }
-
-  /**
-   * Applies set operation along last dimension of `Tensor` and `SparseTensor`.
-   *  <p>
-   *  See SetOperationOp::SetOperationFromContext for values of `set_operation`.
-   *  <p>
-   *  Input `set2` is a `SparseTensor` represented by `set2_indices`, `set2_values`,
-   *  and `set2_shape`. For `set2` ranked `n`, 1st `n-1` dimensions must be the same
-   *  as `set1`. Dimension `n` contains values in a set, duplicates are allowed but
-   *  ignored.
-   *  <p>
-   *  If `validate_indices` is `True`, this op validates the order and range of `set2`
-   *  indices.
-   *  <p>
-   *  Output `result` is a `SparseTensor` represented by `result_indices`,
-   *  `result_values`, and `result_shape`. For `set1` and `set2` ranked `n`, this
-   *  has rank `n` and the same 1st `n-1` dimensions as `set1` and `set2`. The `nth`
-   *  dimension contains the result of `set_operation` applied to the corresponding
-   *  `[0...n-1]` dimension of `set`.
-   *
-   * @param <T> data type for {@code resultValues()} output
-   * @param set1 `Tensor` with rank `n`. 1st `n-1` dimensions must be the same as `set2`.
-   *  Dimension `n` contains values in a set, duplicates are allowed but ignored.
-   * @param set2Indices 2D `Tensor`, indices of a `SparseTensor`. Must be in row-major
-   *  order.
-   * @param set2Values 1D `Tensor`, values of a `SparseTensor`. Must be in row-major
-   *  order.
-   * @param set2Shape 1D `Tensor`, shape of a `SparseTensor`. `set2_shape[0...n-1]` must
-   *  be the same as the 1st `n-1` dimensions of `set1`, `result_shape[n]` is the
-   *  max set size across `n-1` dimensions.
-   * @param setOperation
-   * @param options carries optional attributes values
-   * @return a new instance of DenseToSparseSetOperation
-   */
-  public <T extends TType> DenseToSparseSetOperation<T> denseToSparseSetOperation(Operand<T> set1,
-      Operand<TInt64> set2Indices, Operand<T> set2Values, Operand<TInt64> set2Shape,
-      String setOperation, DenseToSparseSetOperation.Options... options) {
-    return DenseToSparseSetOperation.create(scope, set1, set2Indices, set2Values, set2Shape, setOperation, options);
-  }
-
-  /**
-   * Deserialize `SparseTensor` objects.
-   *  <p>
-   *  The input `serialized_sparse` must have the shape `[?, ?, ..., ?, 3]` where
-   *  the last dimension stores serialized `SparseTensor` objects and the other N
-   *  dimensions (N >= 0) correspond to a batch. The ranks of the original
-   *  `SparseTensor` objects must all match. When the final `SparseTensor` is
-   *  created, its rank is the rank of the incoming `SparseTensor` objects plus N;
-   *  the sparse tensors have been concatenated along new dimensions, one for each
-   *  batch.
-   *  <p>
-   *  The output `SparseTensor` object's shape values for the original dimensions
-   *  are the max across the input `SparseTensor` objects' shape values for the
-   *  corresponding dimensions. The new dimensions match the size of the batch.
-   *  <p>
-   *  The input `SparseTensor` objects' indices are assumed ordered in
-   *  standard lexicographic order.  If this is not the case, after this
-   *  step run `SparseReorder` to restore index ordering.
-   *  <p>
-   *  For example, if the serialized input is a `[2 x 3]` matrix representing two
-   *  original `SparseTensor` objects:
-   *  <p>
-   *      index = [ 0]
-   *              [10]
-   *              [20]
-   *      values = [1, 2, 3]
-   *      shape = [50]
-   *  <p>
-   *  and
-   *  <p>
-   *      index = [ 2]
-   *              [10]
-   *      values = [4, 5]
-   *      shape = [30]
-   *  <p>
-   *  then the final deserialized `SparseTensor` will be:
-   *  <p>
-   *      index = [0  0]
-   *              [0 10]
-   *              [0 20]
-   *              [1  2]
-   *              [1 10]
-   *      values = [1, 2, 3, 4, 5]
-   *      shape = [2 50]
-   *
-   * @param <U> data type for {@code sparseValues()} output
-   * @param serializedSparse The serialized `SparseTensor` objects. The last dimension
-   *  must have 3 columns.
-   * @param dtype The `dtype` of the serialized `SparseTensor` objects.
-   * @return a new instance of DeserializeSparse
-   */
-  public <U extends TType, T extends TType> DeserializeSparse<U> deserializeSparse(
-      Operand<T> serializedSparse, DataType<U> dtype) {
-    return DeserializeSparse.create(scope, serializedSparse, dtype);
-  }
-
-  /**
-   * Applies a sparse gradient to a given accumulator.
-   *  <p>
-   *  Does not add if local_step is smaller than the accumulator's
-   *  global_step.
-   *
-   * @param handle The handle to a accumulator.
-   * @param localStep The local_step value at which the sparse gradient was computed.
-   * @param gradientIndices Indices of the sparse gradient to be accumulated. Must be a
-   *  vector.
-   * @param gradientValues Values are the non-zero slices of the gradient, and must have
-   *  the same first dimension as indices, i.e., the nnz represented by indices and
-   *  values must be consistent.
-   * @param gradientShape Shape of the sparse gradient to be accumulated.
-   * @param hasKnownShape Boolean indicating whether gradient_shape is unknown, in which
-   *  case the input is ignored during validation.
-   * @return a new instance of SparseAccumulatorApplyGradient
-   */
-  public <T extends TType> SparseAccumulatorApplyGradient sparseAccumulatorApplyGradient(
-      Operand<TString> handle, Operand<TInt64> localStep, Operand<TInt64> gradientIndices,
-      Operand<T> gradientValues, Operand<TInt64> gradientShape, Boolean hasKnownShape) {
-    return SparseAccumulatorApplyGradient.create(scope, handle, localStep, gradientIndices, gradientValues, gradientShape, hasKnownShape);
-  }
-
-  /**
-   * Extracts the average sparse gradient in a SparseConditionalAccumulator.
-   *  <p>
-   *  The op will blocks until sufficient (i.e., more than num_required)
-   *  gradients have been accumulated. If the accumulator has already
-   *  aggregated more than num_required gradients, it will return its
-   *  average of the accumulated gradients.  Also automatically increments
-   *  the recorded global_step in the accumulator by 1, and resets the
-   *  aggregate to 0.
-   *
-   * @param <T> data type for {@code values()} output
-   * @param handle The handle to a SparseConditionalAccumulator.
-   * @param numRequired Number of gradients required before we return an aggregate.
-   * @param dtype The data type of accumulated gradients. Needs to correspond to the type
-   *  of the accumulator.
-   * @return a new instance of SparseAccumulatorTakeGradient
-   */
-  public <T extends TType> SparseAccumulatorTakeGradient<T> sparseAccumulatorTakeGradient(
-      Operand<TString> handle, Operand<TInt32> numRequired, DataType<T> dtype) {
-    return SparseAccumulatorTakeGradient.create(scope, handle, numRequired, dtype);
-  }
-
-  /**
-   * Adds two `SparseTensor` objects to produce another `SparseTensor`.
-   *  <p>
-   *  The input `SparseTensor` objects' indices are assumed ordered in standard
-   *  lexicographic order.  If this is not the case, before this step run
-   *  `SparseReorder` to restore index ordering.
-   *  <p>
-   *  By default, if two values sum to zero at some index, the output `SparseTensor`
-   *  would still include that particular location in its index, storing a zero in the
-   *  corresponding value slot.  To override this, callers can specify `thresh`,
-   *  indicating that if the sum has a magnitude strictly smaller than `thresh`, its
-   *  corresponding value and index would then not be included.  In particular,
-   *  `thresh == 0` (default) means everything is kept and actual thresholding happens
-   *  only for a positive value.
-   *  <p>
-   *  In the following shapes, `nnz` is the count after taking `thresh` into account.
-   *
-   * @param <T> data type for {@code sumValues()} output
-   * @param aIndices 2-D.  The `indices` of the first `SparseTensor`, size `[nnz, ndims]` Matrix.
-   * @param aValues 1-D.  The `values` of the first `SparseTensor`, size `[nnz]` Vector.
-   * @param aShape 1-D.  The `shape` of the first `SparseTensor`, size `[ndims]` Vector.
-   * @param bIndices 2-D.  The `indices` of the second `SparseTensor`, size `[nnz, ndims]` Matrix.
-   * @param bValues 1-D.  The `values` of the second `SparseTensor`, size `[nnz]` Vector.
-   * @param bShape 1-D.  The `shape` of the second `SparseTensor`, size `[ndims]` Vector.
-   * @param thresh 0-D.  The magnitude threshold that determines if an output value/index
-   *  pair takes space.
-   * @return a new instance of SparseAdd
-   */
-  public <T extends TType, U extends TNumber> SparseAdd<T> sparseAdd(Operand<TInt64> aIndices,
-      Operand<T> aValues, Operand<TInt64> aShape, Operand<TInt64> bIndices, Operand<T> bValues,
-      Operand<TInt64> bShape, Operand<U> thresh) {
-    return SparseAdd.create(scope, aIndices, aValues, aShape, bIndices, bValues, bShape, thresh);
-  }
-
-  /**
-   * The gradient operator for the SparseAdd op.
-   *  <p>
-   *  The SparseAdd op calculates A + B, where A, B, and the sum are all represented
-   *  as `SparseTensor` objects.  This op takes in the upstream gradient w.r.t.
-   *  non-empty values of the sum, and outputs the gradients w.r.t. the non-empty
-   *  values of A and B.
-   *
-   * @param <T> data type for {@code aValGrad()} output
-   * @param backpropValGrad 1-D with shape `[nnz(sum)]`.  The gradient with respect to
-   *  the non-empty values of the sum.
-   * @param aIndices 2-D.  The `indices` of the `SparseTensor` A, size `[nnz(A), ndims]`.
-   * @param bIndices 2-D.  The `indices` of the `SparseTensor` B, size `[nnz(B), ndims]`.
-   * @param sumIndices 2-D.  The `indices` of the sum `SparseTensor`, size
-   *  `[nnz(sum), ndims]`.
-   * @return a new instance of SparseAddGrad
-   */
-  public <T extends TType> SparseAddGrad<T> sparseAddGrad(Operand<T> backpropValGrad,
-      Operand<TInt64> aIndices, Operand<TInt64> bIndices, Operand<TInt64> sumIndices) {
-    return SparseAddGrad.create(scope, backpropValGrad, aIndices, bIndices, sumIndices);
-  }
-
-  /**
-   * Concatenates a list of `SparseTensor` along the specified dimension.
-   *  <p>
-   *  Concatenation is with respect to the dense versions of these sparse tensors.
-   *  It is assumed that each input is a `SparseTensor` whose elements are ordered
-   *  along increasing dimension number.
-   *  <p>
-   *  All inputs' shapes must match, except for the concat dimension.  The
-   *  `indices`, `values`, and `shapes` lists must have the same length.
-   *  <p>
-   *  The output shape is identical to the inputs', except along the concat
-   *  dimension, where it is the sum of the inputs' sizes along that dimension.
-   *  <p>
-   *  The output elements will be resorted to preserve the sort order along
-   *  increasing dimension number.
-   *  <p>
-   *  This op runs in `O(M log M)` time, where `M` is the total number of non-empty
-   *  values across all inputs. This is due to the need for an internal sort in
-   *  order to concatenate efficiently across an arbitrary dimension.
-   *  <p>
-   *  For example, if `concat_dim = 1` and the inputs are
-   *  <p>
-   *      sp_inputs[0]: shape = [2, 3]
-   *      [0, 2]: "a"
-   *      [1, 0]: "b"
-   *      [1, 1]: "c"
-   *  <p>
-   *      sp_inputs[1]: shape = [2, 4]
-   *      [0, 1]: "d"
-   *      [0, 2]: "e"
-   *  <p>
-   *  then the output will be
-   *  <p>
-   *      shape = [2, 7]
-   *      [0, 2]: "a"
-   *      [0, 4]: "d"
-   *      [0, 5]: "e"
-   *      [1, 0]: "b"
-   *      [1, 1]: "c"
-   *  <p>
-   *  Graphically this is equivalent to doing
-   *  <p>
-   *      [    a] concat [  d e  ] = [    a   d e  ]
-   *      [b c  ]        [       ]   [b c          ]
-   *
-   * @param <T> data type for {@code outputValues()} output
-   * @param indices 2-D.  Indices of each input `SparseTensor`.
-   * @param values 1-D.  Non-empty values of each `SparseTensor`.
-   * @param shapes 1-D.  Shapes of each `SparseTensor`.
-   * @param concatDim Dimension to concatenate along. Must be in range [-rank, rank),
-   *  where rank is the number of dimensions in each input `SparseTensor`.
-   * @return a new instance of SparseConcat
-   */
-  public <T extends TType> SparseConcat<T> sparseConcat(Iterable<Operand<TInt64>> indices,
-      Iterable<Operand<T>> values, Iterable<Operand<TInt64>> shapes, Long concatDim) {
-    return SparseConcat.create(scope, indices, values, shapes, concatDim);
-  }
-
-  /**
-   * A conditional accumulator for aggregating sparse gradients.
-   *  <p>
-   *  The accumulator accepts gradients marked with local_step greater or
-   *  equal to the most recent global_step known to the accumulator. The
-   *  average can be extracted from the accumulator, provided sufficient
-   *  gradients have been accumulated. Extracting the average automatically
-   *  resets the aggregate to 0, and increments the global_step recorded by
-   *  the accumulator.
-   *
-   * @param dtype The type of the value being accumulated.
-   * @param shape The shape of the values.
-   * @param options carries optional attributes values
-   * @return a new instance of SparseConditionalAccumulator
-   */
-  public <T extends TType> SparseConditionalAccumulator sparseConditionalAccumulator(
-      DataType<T> dtype, Shape shape, SparseConditionalAccumulator.Options... options) {
-    return SparseConditionalAccumulator.create(scope, dtype, shape, options);
-  }
-
-  /**
-   * Generates sparse cross from a list of sparse and dense tensors.
-   *  <p>
-   *  The op takes two lists, one of 2D `SparseTensor` and one of 2D `Tensor`, each
-   *  representing features of one feature column. It outputs a 2D `SparseTensor` with
-   *  the batchwise crosses of these features.
-   *  <p>
-   *  For example, if the inputs are
-   *  <p>
-   *      inputs[0]: SparseTensor with shape = [2, 2]
-   *      [0, 0]: "a"
-   *      [1, 0]: "b"
-   *      [1, 1]: "c"
-   *  <p>
-   *      inputs[1]: SparseTensor with shape = [2, 1]
-   *      [0, 0]: "d"
-   *      [1, 0]: "e"
-   *  <p>
-   *      inputs[2]: Tensor [["f"], ["g"]]
-   *  <p>
-   *  then the output will be
-   *  <p>
-   *      shape = [2, 2]
-   *      [0, 0]: "a_X_d_X_f"
-   *      [1, 0]: "b_X_e_X_g"
-   *      [1, 1]: "c_X_e_X_g"
-   *  <p>
-   *  if hashed_output=true then the output will be
-   *  <p>
-   *      shape = [2, 2]
-   *      [0, 0]: FingerprintCat64(
-   *                  Fingerprint64("f"), FingerprintCat64(
-   *                      Fingerprint64("d"), Fingerprint64("a")))
-   *      [1, 0]: FingerprintCat64(
-   *                  Fingerprint64("g"), FingerprintCat64(
-   *                      Fingerprint64("e"), Fingerprint64("b")))
-   *      [1, 1]: FingerprintCat64(
-   *                  Fingerprint64("g"), FingerprintCat64(
-   *                      Fingerprint64("e"), Fingerprint64("c")))
-   *
-   * @param <T> data type for {@code outputValues()} output
-   * @param indices 2-D.  Indices of each input `SparseTensor`.
-   * @param values 1-D.   values of each `SparseTensor`.
-   * @param shapes 1-D.   Shapes of each `SparseTensor`.
-   * @param denseInputs 2-D.    Columns represented by dense `Tensor`.
-   * @param hashedOutput If true, returns the hash of the cross instead of the string.
-   *  This will allow us avoiding string manipulations.
-   * @param numBuckets It is used if hashed_output is true.
-   *  output = hashed_value%num_buckets if num_buckets > 0 else hashed_value.
-   * @param hashKey Specify the hash_key that will be used by the `FingerprintCat64`
-   *  function to combine the crosses fingerprints.
-   * @param outType
-   * @param internalType
-   * @return a new instance of SparseCross
-   */
-  public <T extends TType, U extends TType> SparseCross<T> sparseCross(
-      Iterable<Operand<TInt64>> indices, Iterable<Operand<?>> values,
-      Iterable<Operand<TInt64>> shapes, Iterable<Operand<?>> denseInputs, Boolean hashedOutput,
-      Long numBuckets, Long hashKey, DataType<T> outType, DataType<U> internalType) {
-    return SparseCross.create(scope, indices, values, shapes, denseInputs, hashedOutput, numBuckets, hashKey, outType, internalType);
-  }
-
-  /**
-   * Adds up a SparseTensor and a dense Tensor, using these special rules:
-   *  <p>
-   *  (1) Broadcasts the dense side to have the same shape as the sparse side, if
-   *      eligible;
-   *  (2) Then, only the dense values pointed to by the indices of the SparseTensor
-   *      participate in the cwise addition.
-   *  <p>
-   *  By these rules, the result is a logical SparseTensor with exactly the same
-   *  indices and shape, but possibly with different non-zero values.  The output of
-   *  this Op is the resultant non-zero values.
-   *
-   * @param <T> data type for {@code output()} output
-   * @param spIndices 2-D.  `N x R` matrix with the indices of non-empty values in a
-   *  SparseTensor, possibly not in canonical ordering.
-   * @param spValues 1-D.  `N` non-empty values corresponding to `sp_indices`.
-   * @param spShape 1-D.  Shape of the input SparseTensor.
-   * @param dense `R`-D.  The dense Tensor operand.
-   * @return a new instance of SparseDenseCwiseAdd
-   */
-  public <T extends TType> SparseDenseCwiseAdd<T> sparseDenseCwiseAdd(Operand<TInt64> spIndices,
-      Operand<T> spValues, Operand<TInt64> spShape, Operand<T> dense) {
-    return SparseDenseCwiseAdd.create(scope, spIndices, spValues, spShape, dense);
-  }
-
-  /**
-   * Component-wise divides a SparseTensor by a dense Tensor.
-   *  <p>
-   *  <i>Limitation</i>: this Op only broadcasts the dense side to the sparse side, but not
-   *  the other direction.
-   *
-   * @param <T> data type for {@code output()} output
-   * @param spIndices 2-D.  `N x R` matrix with the indices of non-empty values in a
-   *  SparseTensor, possibly not in canonical ordering.
-   * @param spValues 1-D.  `N` non-empty values corresponding to `sp_indices`.
-   * @param spShape 1-D.  Shape of the input SparseTensor.
-   * @param dense `R`-D.  The dense Tensor operand.
-   * @return a new instance of SparseDenseCwiseDiv
-   */
-  public <T extends TType> SparseDenseCwiseDiv<T> sparseDenseCwiseDiv(Operand<TInt64> spIndices,
-      Operand<T> spValues, Operand<TInt64> spShape, Operand<T> dense) {
-    return SparseDenseCwiseDiv.create(scope, spIndices, spValues, spShape, dense);
-  }
-
-  /**
-   * Component-wise multiplies a SparseTensor by a dense Tensor.
-   *  <p>
-   *  The output locations corresponding to the implicitly zero elements in the sparse
-   *  tensor will be zero (i.e., will not take up storage space), regardless of the
-   *  contents of the dense tensor (even if it's +/-INF and that INF<i>0 == NaN).
-   *  <p>
-   *  </i>Limitation*: this Op only broadcasts the dense side to the sparse side, but not
-   *  the other direction.
-   *
-   * @param <T> data type for {@code output()} output
-   * @param spIndices 2-D.  `N x R` matrix with the indices of non-empty values in a
-   *  SparseTensor, possibly not in canonical ordering.
-   * @param spValues 1-D.  `N` non-empty values corresponding to `sp_indices`.
-   * @param spShape 1-D.  Shape of the input SparseTensor.
-   * @param dense `R`-D.  The dense Tensor operand.
-   * @return a new instance of SparseDenseCwiseMul
-   */
-  public <T extends TType> SparseDenseCwiseMul<T> sparseDenseCwiseMul(Operand<TInt64> spIndices,
-      Operand<T> spValues, Operand<TInt64> spShape, Operand<T> dense) {
-    return SparseDenseCwiseMul.create(scope, spIndices, spValues, spShape, dense);
-  }
-
-  /**
-   * Fills empty rows in the input 2-D `SparseTensor` with a default value.
-   *  <p>
-   *  The input `SparseTensor` is represented via the tuple of inputs
-   *  (`indices`, `values`, `dense_shape`).  The output `SparseTensor` has the
-   *  same `dense_shape` but with indices `output_indices` and values
-   *  `output_values`.
-   *  <p>
-   *  This op inserts a single entry for every row that doesn't have any values.
-   *  The index is created as `[row, 0, ..., 0]` and the inserted value
-   *  is `default_value`.
-   *  <p>
-   *  For example, suppose `sp_input` has shape `[5, 6]` and non-empty values:
-   *  <p>
-   *      [0, 1]: a
-   *      [0, 3]: b
-   *      [2, 0]: c
-   *      [3, 1]: d
-   *  <p>
-   *  Rows 1 and 4 are empty, so the output will be of shape `[5, 6]` with values:
-   *  <p>
-   *      [0, 1]: a
-   *      [0, 3]: b
-   *      [1, 0]: default_value
-   *      [2, 0]: c
-   *      [3, 1]: d
-   *      [4, 0]: default_value
-   *  <p>
-   *  The output `SparseTensor` will be in row-major order and will have the
-   *  same shape as the input.
-   *  <p>
-   *  This op also returns an indicator vector shaped `[dense_shape[0]]` such that
-   *  <p>
-   *      empty_row_indicator[i] = True iff row i was an empty row.
-   *  <p>
-   *  And a reverse index map vector shaped `[indices.shape[0]]` that is used during
-   *  backpropagation,
-   *  <p>
-   *      reverse_index_map[j] = out_j s.t. indices[j, :] == output_indices[out_j, :]
-   *
-   * @param <T> data type for {@code outputValues()} output
-   * @param indices 2-D. the indices of the sparse tensor.
-   * @param values 1-D. the values of the sparse tensor.
-   * @param denseShape 1-D. the shape of the sparse tensor.
-   * @param defaultValue 0-D. default value to insert into location `[row, 0, ..., 0]`
-   *    for rows missing from the input sparse tensor.
-   *  output indices: 2-D. the indices of the filled sparse tensor.
-   * @return a new instance of SparseFillEmptyRows
-   */
-  public <T extends TType> SparseFillEmptyRows<T> sparseFillEmptyRows(Operand<TInt64> indices,
-      Operand<T> values, Operand<TInt64> denseShape, Operand<T> defaultValue) {
-    return SparseFillEmptyRows.create(scope, indices, values, denseShape, defaultValue);
-  }
-
-  /**
-   * The gradient of SparseFillEmptyRows.
-   *  <p>
-   *  Takes vectors reverse_index_map, shaped `[N]`, and grad_values,
-   *  shaped `[N_full]`, where `N_full >= N` and copies data into either
-   *  `d_values` or `d_default_value`.  Here `d_values` is shaped `[N]` and
-   *  `d_default_value` is a scalar.
-   *  <p>
-   *    d_values[j] = grad_values[reverse_index_map[j]]
-   *    d_default_value = sum_{k : 0 .. N_full - 1} (
-   *       grad_values[k] * 1{k not in reverse_index_map})
-   *
-   * @param <T> data type for {@code dValues()} output
-   * @param reverseIndexMap 1-D.  The reverse index map from SparseFillEmptyRows.
-   * @param gradValues 1-D.  The gradients from backprop.
-   * @return a new instance of SparseFillEmptyRowsGrad
-   */
-  public <T extends TType> SparseFillEmptyRowsGrad<T> sparseFillEmptyRowsGrad(
-      Operand<TInt64> reverseIndexMap, Operand<T> gradValues) {
-    return SparseFillEmptyRowsGrad.create(scope, reverseIndexMap, gradValues);
-  }
-
-  /**
-   * Multiply matrix "a" by matrix "b".
-   *  <p>
-   *  The inputs must be two-dimensional matrices and the inner dimension of "a" must
-   *  match the outer dimension of "b". Both "a" and "b" must be `Tensor`s not
-   *  `SparseTensor`s.  This op is optimized for the case where at least one of "a" or
-   *  "b" is sparse, in the sense that they have a large proportion of zero values.
-   *  The breakeven for using this versus a dense matrix multiply on one platform was
-   *  30% zero values in the sparse matrix.
-   *  <p>
-   *  The gradient computation of this operation will only take advantage of sparsity
-   *  in the input gradient when that gradient comes from a Relu.
-   *
-   * @param a
-   * @param b
-   * @param options carries optional attributes values
-   * @return a new instance of SparseMatMul
-   */
-  public <T extends TNumber, U extends TNumber> SparseMatMul sparseMatMul(Operand<T> a,
-      Operand<U> b, SparseMatMul.Options... options) {
-    return SparseMatMul.create(scope, a, b, options);
-  }
-
-  /**
-   * Computes the max of elements across dimensions of a SparseTensor.
-   *  <p>
-   *  This Op takes a SparseTensor and is the sparse counterpart to
-   *  `tf.reduce_max()`.  In particular, this Op also returns a dense `Tensor`
-   *  instead of a sparse one.
-   *  <p>
-   *  Reduces `sp_input` along the dimensions given in `reduction_axes`.  Unless
-   *  `keep_dims` is true, the rank of the tensor is reduced by 1 for each entry in
-   *  `reduction_axes`. If `keep_dims` is true, the reduced dimensions are retained
-   *  with length 1.
-   *  <p>
-   *  If `reduction_axes` has no entries, all dimensions are reduced, and a tensor
-   *  with a single element is returned.  Additionally, the axes can be negative,
-   *  which are interpreted according to the indexing rules in Python.
-   *
-   * @param <T> data type for {@code output()} output
-   * @param inputIndices 2-D.  `N x R` matrix with the indices of non-empty values in a
-   *  SparseTensor, possibly not in canonical ordering.
-   * @param inputValues 1-D.  `N` non-empty values corresponding to `input_indices`.
-   * @param inputShape 1-D.  Shape of the input SparseTensor.
-   * @param reductionAxes 1-D.  Length-`K` vector containing the reduction axes.
-   * @param options carries optional attributes values
-   * @return a new instance of SparseReduceMax
-   */
-  public <T extends TNumber> SparseReduceMax<T> sparseReduceMax(Operand<TInt64> inputIndices,
-      Operand<T> inputValues, Operand<TInt64> inputShape, Operand<TInt32> reductionAxes,
-      SparseReduceMax.Options... options) {
-    return SparseReduceMax.create(scope, inputIndices, inputValues, inputShape, reductionAxes, options);
-  }
-
-  /**
-   * Computes the max of elements across dimensions of a SparseTensor.
-   *  <p>
-   *  This Op takes a SparseTensor and is the sparse counterpart to
-   *  `tf.reduce_max()`.  In contrast to SparseReduceMax, this Op returns a
-   *  SparseTensor.
-   *  <p>
-   *  Reduces `sp_input` along the dimensions given in `reduction_axes`.  Unless
-   *  `keep_dims` is true, the rank of the tensor is reduced by 1 for each entry in
-   *  `reduction_axes`. If `keep_dims` is true, the reduced dimensions are retained
-   *  with length 1.
-   *  <p>
-   *  If `reduction_axes` has no entries, all dimensions are reduced, and a tensor
-   *  with a single element is returned.  Additionally, the axes can be negative,
-   *  which are interpreted according to the indexing rules in Python.
-   *
-   * @param <T> data type for {@code outputValues()} output
-   * @param inputIndices 2-D.  `N x R` matrix with the indices of non-empty values in a
-   *  SparseTensor, possibly not in canonical ordering.
-   * @param inputValues 1-D.  `N` non-empty values corresponding to `input_indices`.
-   * @param inputShape 1-D.  Shape of the input SparseTensor.
-   * @param reductionAxes 1-D.  Length-`K` vector containing the reduction axes.
-   * @param options carries optional attributes values
-   * @return a new instance of SparseReduceMaxSparse
-   */
-  public <T extends TNumber> SparseReduceMaxSparse<T> sparseReduceMaxSparse(
-      Operand<TInt64> inputIndices, Operand<T> inputValues, Operand<TInt64> inputShape,
-      Operand<TInt32> reductionAxes, SparseReduceMaxSparse.Options... options) {
-    return SparseReduceMaxSparse.create(scope, inputIndices, inputValues, inputShape, reductionAxes, options);
-  }
-
-  /**
-   * Computes the sum of elements across dimensions of a SparseTensor.
-   *  <p>
-   *  This Op takes a SparseTensor and is the sparse counterpart to
-   *  `tf.reduce_sum()`.  In particular, this Op also returns a dense `Tensor`
-   *  instead of a sparse one.
-   *  <p>
-   *  Reduces `sp_input` along the dimensions given in `reduction_axes`.  Unless
-   *  `keep_dims` is true, the rank of the tensor is reduced by 1 for each entry in
-   *  `reduction_axes`. If `keep_dims` is true, the reduced dimensions are retained
-   *  with length 1.
-   *  <p>
-   *  If `reduction_axes` has no entries, all dimensions are reduced, and a tensor
-   *  with a single element is returned.  Additionally, the axes can be negative,
-   *  which are interpreted according to the indexing rules in Python.
-   *
-   * @param <T> data type for {@code output()} output
-   * @param inputIndices 2-D.  `N x R` matrix with the indices of non-empty values in a
-   *  SparseTensor, possibly not in canonical ordering.
-   * @param inputValues 1-D.  `N` non-empty values corresponding to `input_indices`.
-   * @param inputShape 1-D.  Shape of the input SparseTensor.
-   * @param reductionAxes 1-D.  Length-`K` vector containing the reduction axes.
-   * @param options carries optional attributes values
-   * @return a new instance of SparseReduceSum
-   */
-  public <T extends TType> SparseReduceSum<T> sparseReduceSum(Operand<TInt64> inputIndices,
-      Operand<T> inputValues, Operand<TInt64> inputShape, Operand<TInt32> reductionAxes,
-      SparseReduceSum.Options... options) {
-    return SparseReduceSum.create(scope, inputIndices, inputValues, inputShape, reductionAxes, options);
-  }
-
-  /**
-   * Computes the sum of elements across dimensions of a SparseTensor.
-   *  <p>
-   *  This Op takes a SparseTensor and is the sparse counterpart to
-   *  `tf.reduce_sum()`.  In contrast to SparseReduceSum, this Op returns a
-   *  SparseTensor.
-   *  <p>
-   *  Reduces `sp_input` along the dimensions given in `reduction_axes`.  Unless
-   *  `keep_dims` is true, the rank of the tensor is reduced by 1 for each entry in
-   *  `reduction_axes`. If `keep_dims` is true, the reduced dimensions are retained
-   *  with length 1.
-   *  <p>
-   *  If `reduction_axes` has no entries, all dimensions are reduced, and a tensor
-   *  with a single element is returned.  Additionally, the axes can be negative,
-   *  which are interpreted according to the indexing rules in Python.
-   *
-   * @param <T> data type for {@code outputValues()} output
-   * @param inputIndices 2-D.  `N x R` matrix with the indices of non-empty values in a
-   *  SparseTensor, possibly not in canonical ordering.
-   * @param inputValues 1-D.  `N` non-empty values corresponding to `input_indices`.
-   * @param inputShape 1-D.  Shape of the input SparseTensor.
-   * @param reductionAxes 1-D.  Length-`K` vector containing the reduction axes.
-   * @param options carries optional attributes values
-   * @return a new instance of SparseReduceSumSparse
-   */
-  public <T extends TType> SparseReduceSumSparse<T> sparseReduceSumSparse(
-      Operand<TInt64> inputIndices, Operand<T> inputValues, Operand<TInt64> inputShape,
-      Operand<TInt32> reductionAxes, SparseReduceSumSparse.Options... options) {
-    return SparseReduceSumSparse.create(scope, inputIndices, inputValues, inputShape, reductionAxes, options);
-  }
-
-  /**
-   * Reorders a SparseTensor into the canonical, row-major ordering.
-   *  <p>
-   *  Note that by convention, all sparse ops preserve the canonical ordering along
-   *  increasing dimension number. The only time ordering can be violated is during
-   *  manual manipulation of the indices and values vectors to add entries.
-   *  <p>
-   *  Reordering does not affect the shape of the SparseTensor.
-   *  <p>
-   *  If the tensor has rank `R` and `N` non-empty values, `input_indices` has
-   *  shape `[N, R]`, input_values has length `N`, and input_shape has length `R`.
-   *
-   * @param <T> data type for {@code outputValues()} output
-   * @param inputIndices 2-D.  `N x R` matrix with the indices of non-empty values in a
-   *  SparseTensor, possibly not in canonical ordering.
-   * @param inputValues 1-D.  `N` non-empty values corresponding to `input_indices`.
-   * @param inputShape 1-D.  Shape of the input SparseTensor.
-   * @return a new instance of SparseReorder
-   */
-  public <T extends TType> SparseReorder<T> sparseReorder(Operand<TInt64> inputIndices,
-      Operand<T> inputValues, Operand<TInt64> inputShape) {
-    return SparseReorder.create(scope, inputIndices, inputValues, inputShape);
-  }
-
-  /**
-   * Reshapes a SparseTensor to represent values in a new dense shape.
-   *  <p>
-   *  This operation has the same semantics as reshape on the represented dense
-   *  tensor.  The `input_indices` are recomputed based on the requested `new_shape`.
-   *  <p>
-   *  If one component of `new_shape` is the special value -1, the size of that
-   *  dimension is computed so that the total dense size remains constant.  At
-   *  most one component of `new_shape` can be -1.  The number of dense elements
-   *  implied by `new_shape` must be the same as the number of dense elements
-   *  originally implied by `input_shape`.
-   *  <p>
-   *  Reshaping does not affect the order of values in the SparseTensor.
-   *  <p>
-   *  If the input tensor has rank `R_in` and `N` non-empty values, and `new_shape`
-   *  has length `R_out`, then `input_indices` has shape `[N, R_in]`,
-   *  `input_shape` has length `R_in`, `output_indices` has shape `[N, R_out]`, and
-   *  `output_shape` has length `R_out`.
-   *
-   * @param inputIndices 2-D.  `N x R_in` matrix with the indices of non-empty values in a
-   *  SparseTensor.
-   * @param inputShape 1-D.  `R_in` vector with the input SparseTensor's dense shape.
-   * @param newShape 1-D.  `R_out` vector with the requested new dense shape.
-   * @return a new instance of SparseReshape
-   */
-  public SparseReshape sparseReshape(Operand<TInt64> inputIndices, Operand<TInt64> inputShape,
-      Operand<TInt64> newShape) {
-    return SparseReshape.create(scope, inputIndices, inputShape, newShape);
-  }
-
-  /**
-   * Computes the mean along sparse segments of a tensor.
-   *  <p>
-   *  See `tf.sparse.segment_sum` for usage examples.
-   *  <p>
-   *  Like `SegmentMean`, but `segment_ids` can have rank less than `data`'s first
-   *  dimension, selecting a subset of dimension 0, specified by `indices`.
-   *
-   * @param <T> data type for {@code output()} output
-   * @param data
-   * @param indices A 1-D tensor. Has same rank as `segment_ids`.
-   * @param segmentIds A 1-D tensor. Values should be sorted and can be repeated.
-   * @return a new instance of SparseSegmentMean
-   */
-  public <T extends TNumber, U extends TNumber> SparseSegmentMean<T> sparseSegmentMean(
-      Operand<T> data, Operand<U> indices, Operand<TInt32> segmentIds) {
-    return SparseSegmentMean.create(scope, data, indices, segmentIds);
-  }
-
-  /**
-   * Computes gradients for SparseSegmentMean.
-   *  <p>
-   *  Returns tensor "output" with same shape as grad, except for dimension 0 whose
-   *  value is output_dim0.
-   *
-   * @param <T> data type for {@code output()} output
-   * @param grad gradient propagated to the SparseSegmentMean op.
-   * @param indices indices passed to the corresponding SparseSegmentMean op.
-   * @param segmentIds segment_ids passed to the corresponding SparseSegmentMean op.
-   * @param outputDim0 dimension 0 of "data" passed to SparseSegmentMean op.
-   * @return a new instance of SparseSegmentMeanGrad
-   */
-  public <T extends TNumber, U extends TNumber> SparseSegmentMeanGrad<T> sparseSegmentMeanGrad(
-      Operand<T> grad, Operand<U> indices, Operand<TInt32> segmentIds, Operand<TInt32> outputDim0) {
-    return SparseSegmentMeanGrad.create(scope, grad, indices, segmentIds, outputDim0);
-  }
-
-  /**
-   * Computes the mean along sparse segments of a tensor.
-   *  <p>
-   *  Like `SparseSegmentMean`, but allows missing ids in `segment_ids`. If an id is
-   *  misisng, the `output` tensor at that position will be zeroed.
-   *  <p>
-   *  Read
-   *  [the section on segmentation](https://tensorflow.org/api_docs/python/tf/math#Segmentation)
-   *  for an explanation of segments.
-   *
-   * @param <T> data type for {@code output()} output
-   * @param data
-   * @param indices A 1-D tensor. Has same rank as `segment_ids`.
-   * @param segmentIds A 1-D tensor. Values should be sorted and can be repeated.
-   * @param numSegments Should equal the number of distinct segment IDs.
-   * @return a new instance of SparseSegmentMeanWithNumSegments
-   */
-  public <T extends TNumber, U extends TNumber, V extends TNumber> SparseSegmentMeanWithNumSegments<T> sparseSegmentMeanWithNumSegments(
-      Operand<T> data, Operand<U> indices, Operand<TInt32> segmentIds, Operand<V> numSegments) {
-    return SparseSegmentMeanWithNumSegments.create(scope, data, indices, segmentIds, numSegments);
-  }
-
-  /**
-   * Computes the sum along sparse segments of a tensor divided by the sqrt of N.
-   *  <p>
-   *  N is the size of the segment being reduced.
-   *  <p>
-   *  See `tf.sparse.segment_sum` for usage examples.
-   *
-   * @param <T> data type for {@code output()} output
-   * @param data
-   * @param indices A 1-D tensor. Has same rank as `segment_ids`.
-   * @param segmentIds A 1-D tensor. Values should be sorted and can be repeated.
-   * @return a new instance of SparseSegmentSqrtN
-   */
-  public <T extends TNumber, U extends TNumber> SparseSegmentSqrtN<T> sparseSegmentSqrtN(
-      Operand<T> data, Operand<U> indices, Operand<TInt32> segmentIds) {
-    return SparseSegmentSqrtN.create(scope, data, indices, segmentIds);
-  }
-
-  /**
-   * Computes gradients for SparseSegmentSqrtN.
-   *  <p>
-   *  Returns tensor "output" with same shape as grad, except for dimension 0 whose
-   *  value is output_dim0.
-   *
-   * @param <T> data type for {@code output()} output
-   * @param grad gradient propagated to the SparseSegmentSqrtN op.
-   * @param indices indices passed to the corresponding SparseSegmentSqrtN op.
-   * @param segmentIds segment_ids passed to the corresponding SparseSegmentSqrtN op.
-   * @param outputDim0 dimension 0 of "data" passed to SparseSegmentSqrtN op.
-   * @return a new instance of SparseSegmentSqrtNGrad
-   */
-  public <T extends TNumber, U extends TNumber> SparseSegmentSqrtNGrad<T> sparseSegmentSqrtNGrad(
-      Operand<T> grad, Operand<U> indices, Operand<TInt32> segmentIds, Operand<TInt32> outputDim0) {
-    return SparseSegmentSqrtNGrad.create(scope, grad, indices, segmentIds, outputDim0);
-  }
-
-  /**
-   * Computes the sum along sparse segments of a tensor divided by the sqrt of N.
-   *  <p>
-   *  N is the size of the segment being reduced.
-   *  <p>
-   *  Like `SparseSegmentSqrtN`, but allows missing ids in `segment_ids`. If an id is
-   *  misisng, the `output` tensor at that position will be zeroed.
-   *  <p>
-   *  Read
-   *  [the section on segmentation](https://tensorflow.org/api_docs/python/tf/math#Segmentation)
-   *  for an explanation of segments.
-   *
-   * @param <T> data type for {@code output()} output
-   * @param data
-   * @param indices A 1-D tensor. Has same rank as `segment_ids`.
-   * @param segmentIds A 1-D tensor. Values should be sorted and can be repeated.
-   * @param numSegments Should equal the number of distinct segment IDs.
-   * @return a new instance of SparseSegmentSqrtNWithNumSegments
-   */
-  public <T extends TNumber, U extends TNumber, V extends TNumber> SparseSegmentSqrtNWithNumSegments<T> sparseSegmentSqrtNWithNumSegments(
-      Operand<T> data, Operand<U> indices, Operand<TInt32> segmentIds, Operand<V> numSegments) {
-    return SparseSegmentSqrtNWithNumSegments.create(scope, data, indices, segmentIds, numSegments);
-  }
-
-  /**
-   * Computes the sum along sparse segments of a tensor.
-   *  <p>
-   *  Read
-   *  [the section on segmentation](https://tensorflow.org/api_docs/python/tf/math#Segmentation)
-   *  for an explanation of segments.
-   *  <p>
-   *  Like `SegmentSum`, but `segment_ids` can have rank less than `data`'s first
-   *  dimension, selecting a subset of dimension 0, specified by `indices`.
-   *  <p>
-   *  For example:
-   *  <pre>{@code
-   *  c = tf.constant([[1,2,3,4], [-1,-2,-3,-4], [5,6,7,8]])
-   *
-   *  # Select two rows, one segment.
-   *  tf.sparse_segment_sum(c, tf.constant([0, 1]), tf.constant([0, 0]))
-   *  # => [[0 0 0 0]]
-   *
-   *  # Select two rows, two segment.
-   *  tf.sparse_segment_sum(c, tf.constant([0, 1]), tf.constant([0, 1]))
-   *  # => [[ 1  2  3  4]
-   *  #     [-1 -2 -3 -4]]
-   *
-   *  # Select all rows, two segments.
-   *  tf.sparse_segment_sum(c, tf.constant([0, 1, 2]), tf.constant([0, 0, 1]))
-   *  # => [[0 0 0 0]
-   *  #     [5 6 7 8]]
-   *
-   *  # Which is equivalent to:
-   *  tf.segment_sum(c, tf.constant([0, 0, 1]))
-   *  }</pre>
-   *
-   * @param <T> data type for {@code output()} output
-   * @param data
-   * @param indices A 1-D tensor. Has same rank as `segment_ids`.
-   * @param segmentIds A 1-D tensor. Values should be sorted and can be repeated.
-   * @return a new instance of SparseSegmentSum
-   */
-  public <T extends TNumber, U extends TNumber> SparseSegmentSum<T> sparseSegmentSum(
-      Operand<T> data, Operand<U> indices, Operand<TInt32> segmentIds) {
-    return SparseSegmentSum.create(scope, data, indices, segmentIds);
-  }
-
-  /**
-   * Computes the sum along sparse segments of a tensor.
-   *  <p>
-   *  Like `SparseSegmentSum`, but allows missing ids in `segment_ids`. If an id is
-   *  misisng, the `output` tensor at that position will be zeroed.
-   *  <p>
-   *  Read
-   *  [the section on segmentation](https://tensorflow.org/api_docs/python/tf/sparse#Segmentation)
-   *  for an explanation of segments.
-   *  <p>
-   *  For example:
-   *  <pre>{@code
-   *  c = tf.constant([[1,2,3,4], [-1,-2,-3,-4], [5,6,7,8]])
-   *
-   *  tf.sparse_segment_sum_with_num_segments(
-   *      c, tf.constant([0, 1]), tf.constant([0, 0]), num_segments=3)
-   *  # => [[0 0 0 0]
-   *  #     [0 0 0 0]
-   *  #     [0 0 0 0]]
-   *
-   *  tf.sparse_segment_sum_with_num_segments(c,
-   *                                          tf.constant([0, 1]),
-   *                                          tf.constant([0, 2],
-   *                                          num_segments=4))
-   *  # => [[ 1  2  3  4]
-   *  #     [ 0  0  0  0]
-   *  #     [-1 -2 -3 -4]
-   *  #     [ 0  0  0  0]]
-   *  }</pre>
-   *
-   * @param <T> data type for {@code output()} output
-   * @param data
-   * @param indices A 1-D tensor. Has same rank as `segment_ids`.
-   * @param segmentIds A 1-D tensor. Values should be sorted and can be repeated.
-   * @param numSegments Should equal the number of distinct segment IDs.
-   * @return a new instance of SparseSegmentSumWithNumSegments
-   */
-  public <T extends TNumber, U extends TNumber, V extends TNumber> SparseSegmentSumWithNumSegments<T> sparseSegmentSumWithNumSegments(
-      Operand<T> data, Operand<U> indices, Operand<TInt32> segmentIds, Operand<V> numSegments) {
-    return SparseSegmentSumWithNumSegments.create(scope, data, indices, segmentIds, numSegments);
-  }
-
-  /**
-   * Slice a `SparseTensor` based on the `start` and `size`.
-   *  <p>
-   *  For example, if the input is
-   *  <p>
-   *      input_tensor = shape = [2, 7]
-   *      [    a   d e  ]
-   *      [b c          ]
-   *  <p>
-   *  Graphically the output tensors are:
-   *  <p>
-   *      sparse_slice([0, 0], [2, 4]) = shape = [2, 4]
-   *      [    a  ]
-   *      [b c    ]
-   *  <p>
-   *      sparse_slice([0, 4], [2, 3]) = shape = [2, 3]
-   *      [ d e  ]
-   *      [      ]
-   *
-   * @param <T> data type for {@code outputValues()} output
-   * @param indices 2-D tensor represents the indices of the sparse tensor.
-   * @param values 1-D tensor represents the values of the sparse tensor.
-   * @param shape 1-D. tensor represents the shape of the sparse tensor.
-   * @param start 1-D. tensor represents the start of the slice.
-   * @param size 1-D. tensor represents the size of the slice.
-   *  output indices: A list of 1-D tensors represents the indices of the output
-   *  sparse tensors.
-   * @return a new instance of SparseSlice
-   */
-  public <T extends TType> SparseSlice<T> sparseSlice(Operand<TInt64> indices, Operand<T> values,
-      Operand<TInt64> shape, Operand<TInt64> start, Operand<TInt64> size) {
-    return SparseSlice.create(scope, indices, values, shape, start, size);
-  }
-
-  /**
-   * The gradient operator for the SparseSlice op.
-   *  <p>
-   *  This op takes in the upstream gradient w.r.t. non-empty values of
-   *  the sliced `SparseTensor`, and outputs the gradients w.r.t.
-   *  the non-empty values of input `SparseTensor`.
-   *
-   * @param <T> data type for {@code valGrad()} output
-   * @param backpropValGrad 1-D. The gradient with respect to
-   *  the non-empty values of the sliced `SparseTensor`.
-   * @param inputIndices 2-D.  The `indices` of the input `SparseTensor`.
-   * @param inputStart 1-D. tensor represents the start of the slice.
-   * @param outputIndices 2-D.  The `indices` of the sliced `SparseTensor`.
-   * @return a new instance of SparseSliceGrad
-   */
-  public <T extends TType> SparseSliceGrad<T> sparseSliceGrad(Operand<T> backpropValGrad,
-      Operand<TInt64> inputIndices, Operand<TInt64> inputStart, Operand<TInt64> outputIndices) {
-    return SparseSliceGrad.create(scope, backpropValGrad, inputIndices, inputStart, outputIndices);
-  }
-
-  /**
-   * Applies softmax to a batched N-D `SparseTensor`.
-   *  <p>
-   *  The inputs represent an N-D SparseTensor  with logical shape `[..., B, C]`
-   *  (where `N >= 2`), and with indices sorted in the canonical lexicographic order.
-   *  <p>
-   *  This op is equivalent to applying the normal `tf.nn.softmax()` to each innermost
-   *  logical submatrix with shape `[B, C]`, but with the catch that <i>the implicitly
-   *  zero elements do not participate</i>.  Specifically, the algorithm is equivalent
-   *  to the following:
-   *  <p>
-   *    (1) Applies `tf.nn.softmax()` to a densified view of each innermost submatrix
-   *        with shape `[B, C]`, along the size-C dimension;
-   *    (2) Masks out the original implicitly-zero locations;
-   *    (3) Renormalizes the remaining elements.
-   *  <p>
-   *  Hence, the `SparseTensor` result has exactly the same non-zero indices and
-   *  shape.
-   *
-   * @param <T> data type for {@code output()} output
-   * @param spIndices 2-D.  `NNZ x R` matrix with the indices of non-empty values in a
-   *  SparseTensor, in canonical ordering.
-   * @param spValues 1-D.  `NNZ` non-empty values corresponding to `sp_indices`.
-   * @param spShape 1-D.  Shape of the input SparseTensor.
-   * @return a new instance of SparseSoftmax
-   */
-  public <T extends TNumber> SparseSoftmax<T> sparseSoftmax(Operand<TInt64> spIndices,
-      Operand<T> spValues, Operand<TInt64> spShape) {
-    return SparseSoftmax.create(scope, spIndices, spValues, spShape);
-  }
-
-  /**
-   * Returns the element-wise max of two SparseTensors.
-   *  <p>
-   *  Assumes the two SparseTensors have the same shape, i.e., no broadcasting.
-   *
-   * @param <T> data type for {@code outputValues()} output
-   * @param aIndices 2-D.  `N x R` matrix with the indices of non-empty values in a
-   *  SparseTensor, in the canonical lexicographic ordering.
-   * @param aValues 1-D.  `N` non-empty values corresponding to `a_indices`.
-   * @param aShape 1-D.  Shape of the input SparseTensor.
-   * @param bIndices counterpart to `a_indices` for the other operand.
-   * @param bValues counterpart to `a_values` for the other operand; must be of the same dtype.
-   * @param bShape counterpart to `a_shape` for the other operand; the two shapes must be equal.
-   * @return a new instance of SparseSparseMaximum
-   */
-  public <T extends TNumber> SparseSparseMaximum<T> sparseSparseMaximum(Operand<TInt64> aIndices,
-      Operand<T> aValues, Operand<TInt64> aShape, Operand<TInt64> bIndices, Operand<T> bValues,
-      Operand<TInt64> bShape) {
-    return SparseSparseMaximum.create(scope, aIndices, aValues, aShape, bIndices, bValues, bShape);
-  }
-
-  /**
-   * Returns the element-wise min of two SparseTensors.
-   *  <p>
-   *  Assumes the two SparseTensors have the same shape, i.e., no broadcasting.
-   *
-   * @param <T> data type for {@code outputValues()} output
-   * @param aIndices 2-D.  `N x R` matrix with the indices of non-empty values in a
-   *  SparseTensor, in the canonical lexicographic ordering.
-   * @param aValues 1-D.  `N` non-empty values corresponding to `a_indices`.
-   * @param aShape 1-D.  Shape of the input SparseTensor.
-   * @param bIndices counterpart to `a_indices` for the other operand.
-   * @param bValues counterpart to `a_values` for the other operand; must be of the same dtype.
-   * @param bShape counterpart to `a_shape` for the other operand; the two shapes must be equal.
-   * @return a new instance of SparseSparseMinimum
-   */
-  public <T extends TType> SparseSparseMinimum<T> sparseSparseMinimum(Operand<TInt64> aIndices,
-      Operand<T> aValues, Operand<TInt64> aShape, Operand<TInt64> bIndices, Operand<T> bValues,
-      Operand<TInt64> bShape) {
-    return SparseSparseMinimum.create(scope, aIndices, aValues, aShape, bIndices, bValues, bShape);
-  }
-
-  /**
-   * Split a `SparseTensor` into `num_split` tensors along one dimension.
-   *  <p>
-   *  If the `shape[split_dim]` is not an integer multiple of `num_split`. Slices
-   *  `[0 : shape[split_dim] % num_split]` gets one extra dimension.
-   *  For example, if `split_dim = 1` and `num_split = 2` and the input is
-   *  <p>
-   *      input_tensor = shape = [2, 7]
-   *      [    a   d e  ]
-   *      [b c          ]
-   *  <p>
-   *  Graphically the output tensors are:
-   *  <p>
-   *      output_tensor[0] = shape = [2, 4]
-   *      [    a  ]
-   *      [b c    ]
-   *  <p>
-   *      output_tensor[1] = shape = [2, 3]
-   *      [ d e  ]
-   *      [      ]
-   *
-   * @param <T> data type for {@code outputValues()} output
-   * @param splitDim 0-D.  The dimension along which to split.  Must be in the range
-   *  `[0, rank(shape))`.
-   * @param indices 2-D tensor represents the indices of the sparse tensor.
-   * @param values 1-D tensor represents the values of the sparse tensor.
-   * @param shape 1-D. tensor represents the shape of the sparse tensor.
-   *  output indices: A list of 1-D tensors represents the indices of the output
-   *  sparse tensors.
-   * @param numSplit The number of ways to split.
-   * @return a new instance of SparseSplit
-   */
-  public <T extends TType> SparseSplit<T> sparseSplit(Operand<TInt64> splitDim,
-      Operand<TInt64> indices, Operand<T> values, Operand<TInt64> shape, Long numSplit) {
-    return SparseSplit.create(scope, splitDim, indices, values, shape, numSplit);
-  }
-
-  /**
-   * Adds up a `SparseTensor` and a dense `Tensor`, producing a dense `Tensor`.
-   *  <p>
-   *  This Op does not require `a_indices` be sorted in standard lexicographic order.
-   *
-   * @param <U> data type for {@code output()} output
-   * @param aIndices 2-D.  The `indices` of the `SparseTensor`, with shape `[nnz, ndims]`.
-   * @param aValues 1-D.  The `values` of the `SparseTensor`, with shape `[nnz]`.
-   * @param aShape 1-D.  The `shape` of the `SparseTensor`, with shape `[ndims]`.
-   * @param b `ndims`-D Tensor.  With shape `a_shape`.
-   * @return a new instance of SparseTensorDenseAdd
-   */
-  public <U extends TType, T extends TNumber> SparseTensorDenseAdd<U> sparseTensorDenseAdd(
-      Operand<T> aIndices, Operand<U> aValues, Operand<T> aShape, Operand<U> b) {
-    return SparseTensorDenseAdd.create(scope, aIndices, aValues, aShape, b);
-  }
-
-  /**
-   * Multiply SparseTensor (of rank 2) "A" by dense matrix "B".
-   *  <p>
-   *  No validity checking is performed on the indices of A.  However, the following
-   *  input format is recommended for optimal behavior:
-   *  <p>
-   *  if adjoint_a == false:
-   *    A should be sorted in lexicographically increasing order.  Use SparseReorder
-   *    if you're not sure.
-   *  if adjoint_a == true:
-   *    A should be sorted in order of increasing dimension 1 (i.e., "column major"
-   *    order instead of "row major" order).
-   *
-   * @param <U> data type for {@code product()} output
-   * @param aIndices 2-D.  The `indices` of the `SparseTensor`, size `[nnz, 2]` Matrix.
-   * @param aValues 1-D.  The `values` of the `SparseTensor`, size `[nnz]` Vector.
-   * @param aShape 1-D.  The `shape` of the `SparseTensor`, size `[2]` Vector.
-   * @param b 2-D.  A dense Matrix.
-   * @param options carries optional attributes values
-   * @return a new instance of SparseTensorDenseMatMul
-   */
-  public <U extends TType, T extends TNumber> SparseTensorDenseMatMul<U> sparseTensorDenseMatMul(
-      Operand<T> aIndices, Operand<U> aValues, Operand<TInt64> aShape, Operand<U> b,
-      SparseTensorDenseMatMul.Options... options) {
-    return SparseTensorDenseMatMul.create(scope, aIndices, aValues, aShape, b, options);
-  }
-
-  /**
-   * Converts a sparse representation into a dense tensor.
-   *  <p>
-   *  Builds an array `dense` with shape `output_shape` such that
-   *  <pre>{@code
-   *  # If sparse_indices is scalar
-   *  dense[i] = (i == sparse_indices ? sparse_values : default_value)
-   *
-   *  # If sparse_indices is a vector, then for each i
-   *  dense[sparse_indices[i]] = sparse_values[i]
-   *
-   *  # If sparse_indices is an n by d matrix, then for each i in [0, n)
-   *  dense[sparse_indices[i][0], ..., sparse_indices[i][d-1]] = sparse_values[i]
-   *  }</pre>
-   *  All other values in `dense` are set to `default_value`.  If `sparse_values` is a
-   *  scalar, all sparse indices are set to this single value.
-   *  <p>
-   *  Indices should be sorted in lexicographic order, and indices must not
-   *  contain any repeats. If `validate_indices` is true, these properties
-   *  are checked during execution.
-   *
-   * @param <U> data type for {@code dense()} output
-   * @param sparseIndices 0-D, 1-D, or 2-D.  `sparse_indices[i]` contains the complete
-   *  index where `sparse_values[i]` will be placed.
-   * @param outputShape 1-D.  Shape of the dense output tensor.
-   * @param sparseValues 1-D.  Values corresponding to each row of `sparse_indices`,
-   *  or a scalar value to be used for all sparse indices.
-   * @param defaultValue Scalar value to set for indices not specified in
-   *  `sparse_indices`.
-   * @param options carries optional attributes values
-   * @return a new instance of SparseToDense
-   */
-  public <U extends TType, T extends TNumber> SparseToDense<U> sparseToDense(
-      Operand<T> sparseIndices, Operand<T> outputShape, Operand<U> sparseValues,
-      Operand<U> defaultValue, SparseToDense.Options... options) {
-    return SparseToDense.create(scope, sparseIndices, outputShape, sparseValues, defaultValue, options);
-  }
-
-  /**
-   * Applies set operation along last dimension of 2 `SparseTensor` inputs.
-   *  <p>
-   *  See SetOperationOp::SetOperationFromContext for values of `set_operation`.
-   *  <p>
-   *  If `validate_indices` is `True`, `sparse.SparseToSparseSetOperation` validates the
-   *  order and range of `set1` and `set2` indices.
-   *  <p>
-   *  Input `set1` is a `SparseTensor` represented by `set1_indices`, `set1_values`,
-   *  and `set1_shape`. For `set1` ranked `n`, 1st `n-1` dimensions must be the same
-   *  as `set2`. Dimension `n` contains values in a set, duplicates are allowed but
-   *  ignored.
-   *  <p>
-   *  Input `set2` is a `SparseTensor` represented by `set2_indices`, `set2_values`,
-   *  and `set2_shape`. For `set2` ranked `n`, 1st `n-1` dimensions must be the same
-   *  as `set1`. Dimension `n` contains values in a set, duplicates are allowed but
-   *  ignored.
-   *  <p>
-   *  If `validate_indices` is `True`, this op validates the order and range of `set1`
-   *  and `set2` indices.
-   *  <p>
-   *  Output `result` is a `SparseTensor` represented by `result_indices`,
-   *  `result_values`, and `result_shape`. For `set1` and `set2` ranked `n`, this
-   *  has rank `n` and the same 1st `n-1` dimensions as `set1` and `set2`. The `nth`
-   *  dimension contains the result of `set_operation` applied to the corresponding
-   *  `[0...n-1]` dimension of `set`.
-   *
-   * @param <T> data type for {@code resultValues()} output
-   * @param set1Indices 2D `Tensor`, indices of a `SparseTensor`. Must be in row-major
-   *  order.
-   * @param set1Values 1D `Tensor`, values of a `SparseTensor`. Must be in row-major
-   *  order.
-   * @param set1Shape 1D `Tensor`, shape of a `SparseTensor`. `set1_shape[0...n-1]` must
-   *  be the same as `set2_shape[0...n-1]`, `set1_shape[n]` is the
-   *  max set size across `0...n-1` dimensions.
-   * @param set2Indices 2D `Tensor`, indices of a `SparseTensor`. Must be in row-major
-   *  order.
-   * @param set2Values 1D `Tensor`, values of a `SparseTensor`. Must be in row-major
-   *  order.
-   * @param set2Shape 1D `Tensor`, shape of a `SparseTensor`. `set2_shape[0...n-1]` must
-   *  be the same as `set1_shape[0...n-1]`, `set2_shape[n]` is the
-   *  max set size across `0...n-1` dimensions.
-   * @param setOperation
-   * @param options carries optional attributes values
-   * @return a new instance of SparseToSparseSetOperation
-   */
-  public <T extends TType> SparseToSparseSetOperation<T> sparseToSparseSetOperation(
-      Operand<TInt64> set1Indices, Operand<T> set1Values, Operand<TInt64> set1Shape,
-      Operand<TInt64> set2Indices, Operand<T> set2Values, Operand<TInt64> set2Shape,
-      String setOperation, SparseToSparseSetOperation.Options... options) {
-    return SparseToSparseSetOperation.create(scope, set1Indices, set1Values, set1Shape, set2Indices, set2Values, set2Shape, setOperation, options);
-  }
-
-  /**
-   * Read `SparseTensors` from a `SparseTensorsMap` and concatenate them.
-   *  <p>
-   *  The input `sparse_handles` must be an `int64` matrix of shape `[N, 1]` where
-   *  `N` is the minibatch size and the rows correspond to the output handles of
-   *  `AddSparseToTensorsMap` or `AddManySparseToTensorsMap`.  The ranks of the
-   *  original `SparseTensor` objects that went into the given input ops must all
-   *  match.  When the final `SparseTensor` is created, it has rank one
-   *  higher than the ranks of the incoming `SparseTensor` objects
-   *  (they have been concatenated along a new row dimension on the left).
-   *  <p>
-   *  The output `SparseTensor` object's shape values for all dimensions but the
-   *  first are the max across the input `SparseTensor` objects' shape values
-   *  for the corresponding dimensions.  Its first shape value is `N`, the minibatch
-   *  size.
-   *  <p>
-   *  The input `SparseTensor` objects' indices are assumed ordered in
-   *  standard lexicographic order.  If this is not the case, after this
-   *  step run `SparseReorder` to restore index ordering.
-   *  <p>
-   *  For example, if the handles represent an input, which is a `[2, 3]` matrix
-   *  representing two original `SparseTensor` objects:
-   *  <pre>{@code
-   *      index = [ 0]
-   *              [10]
-   *              [20]
-   *      values = [1, 2, 3]
-   *      shape = [50]
-   *  }</pre>
-   *  and
-   *  <pre>{@code
-   *      index = [ 2]
-   *              [10]
-   *      values = [4, 5]
-   *      shape = [30]
-   *  }</pre>
-   *  then the final `SparseTensor` will be:
-   *  <pre>{@code
-   *      index = [0  0]
-   *              [0 10]
-   *              [0 20]
-   *              [1  2]
-   *              [1 10]
-   *      values = [1, 2, 3, 4, 5]
-   *      shape = [2 50]
-   *  }</pre>
-   *
-   * @param <T> data type for {@code sparseValues()} output
-   * @param sparseHandles 1-D, The `N` serialized `SparseTensor` objects.
-   *  Shape: `[N]`.
-   * @param dtype The `dtype` of the `SparseTensor` objects stored in the
-   *  `SparseTensorsMap`.
-   * @param options carries optional attributes values
-   * @return a new instance of TakeManySparseFromTensorsMap
-   */
-  public <T extends TType> TakeManySparseFromTensorsMap<T> takeManySparseFromTensorsMap(
-      Operand<TInt64> sparseHandles, DataType<T> dtype,
-      TakeManySparseFromTensorsMap.Options... options) {
-    return TakeManySparseFromTensorsMap.create(scope, sparseHandles, dtype, options);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/StringsOps.java b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/StringsOps.java
deleted file mode 100644
index f6491843332..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/StringsOps.java
+++ /dev/null
@@ -1,620 +0,0 @@
-// Copyright 2020 The TensorFlow Authors. All Rights Reserved.
-//
-// Licensed under the Apache License, Version 2.0 (the "License");
-// you may not use this file except in compliance with the License.
-// You may obtain a copy of the License at
-//
-//     http://www.apache.org/licenses/LICENSE-2.0
-//
-// Unless required by applicable law or agreed to in writing, software
-// distributed under the License is distributed on an "AS IS" BASIS,
-// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-// See the License for the specific language governing permissions and
-// limitations under the License.
-// ==============================================================================
-//
-// This class has been generated, DO NOT EDIT!
-//
-package org.tensorflow.op;
-
-import java.util.List;
-import org.tensorflow.DataType;
-import org.tensorflow.Operand;
-import org.tensorflow.op.strings.Join;
-import org.tensorflow.op.strings.Lower;
-import org.tensorflow.op.strings.ReduceJoin;
-import org.tensorflow.op.strings.RegexFullMatch;
-import org.tensorflow.op.strings.RegexReplace;
-import org.tensorflow.op.strings.StringFormat;
-import org.tensorflow.op.strings.StringLength;
-import org.tensorflow.op.strings.StringNGrams;
-import org.tensorflow.op.strings.StringSplit;
-import org.tensorflow.op.strings.Strip;
-import org.tensorflow.op.strings.Substr;
-import org.tensorflow.op.strings.ToHashBucket;
-import org.tensorflow.op.strings.ToHashBucketFast;
-import org.tensorflow.op.strings.ToHashBucketStrong;
-import org.tensorflow.op.strings.ToNumber;
-import org.tensorflow.op.strings.UnicodeScript;
-import org.tensorflow.op.strings.UnicodeTranscode;
-import org.tensorflow.op.strings.UnsortedSegmentJoin;
-import org.tensorflow.op.strings.Upper;
-import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.TString;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * An API for building {@code strings} operations as {@link Op Op}s
- *
- * @see {@link Ops}
- */
-public final class StringsOps {
-  private final Scope scope;
-
-  StringsOps(Scope scope) {
-    this.scope = scope;
-  }
-
-  /**
-   * Joins the strings in the given list of string tensors into one tensor;
-   *  <p>
-   *  with the given separator (default is an empty separator).
-   *  <p>
-   *  Examples:
-   *  <p>
-   *  >>> s = ["hello", "world", "tensorflow"]
-   *  >>> tf.strings.join(s, " ")
-   *  <tf.Tensor: shape=(), dtype=string, numpy=b'hello world tensorflow'>
-   *
-   * @param inputs A list of string tensors.  The tensors must all have the same shape,
-   *  or be scalars.  Scalars may be mixed in; these will be broadcast to the shape
-   *  of non-scalar inputs.
-   * @param options carries optional attributes values
-   * @return a new instance of Join
-   */
-  public Join join(Iterable<Operand<TString>> inputs, Join.Options... options) {
-    return Join.create(scope, inputs, options);
-  }
-
-  /**
-   * Converts all uppercase characters into their respective lowercase replacements.
-   *  <p>
-   *  Example:
-   *  <p>
-   *  >>> tf.strings.lower("CamelCase string and ALL CAPS")
-   *  <tf.Tensor: shape=(), dtype=string, numpy=b'camelcase string and all caps'>
-   *
-   * @param input
-   * @param options carries optional attributes values
-   * @return a new instance of Lower
-   */
-  public Lower lower(Operand<TString> input, Lower.Options... options) {
-    return Lower.create(scope, input, options);
-  }
-
-  /**
-   * Joins a string Tensor across the given dimensions.
-   *  <p>
-   *  Computes the string join across dimensions in the given string Tensor of shape
-   *  `[\\(d_0, d_1, ..., d_{n-1}\\)]`.  Returns a new Tensor created by joining the input
-   *  strings with the given separator (default: empty string).  Negative indices are
-   *  counted backwards from the end, with `-1` being equivalent to `n - 1`.  If
-   *  indices are not specified, joins across all dimensions beginning from `n - 1`
-   *  through `0`.
-   *  <p>
-   *  For example:
-   *  <pre>{@code
-   *  # tensor `a` is [["a", "b"], ["c", "d"]]
-   *  tf.reduce_join(a, 0) ==> ["ac", "bd"]
-   *  tf.reduce_join(a, 1) ==> ["ab", "cd"]
-   *  tf.reduce_join(a, -2) = tf.reduce_join(a, 0) ==> ["ac", "bd"]
-   *  tf.reduce_join(a, -1) = tf.reduce_join(a, 1) ==> ["ab", "cd"]
-   *  tf.reduce_join(a, 0, keep_dims=True) ==> [["ac", "bd"]]
-   *  tf.reduce_join(a, 1, keep_dims=True) ==> [["ab"], ["cd"]]
-   *  tf.reduce_join(a, 0, separator=".") ==> ["a.c", "b.d"]
-   *  tf.reduce_join(a, [0, 1]) ==> "acbd"
-   *  tf.reduce_join(a, [1, 0]) ==> "abcd"
-   *  tf.reduce_join(a, []) ==> [["a", "b"], ["c", "d"]]
-   *  tf.reduce_join(a) = tf.reduce_join(a, [1, 0]) ==> "abcd"
-   *  }</pre>
-   *
-   * @param inputs The input to be joined.  All reduced indices must have non-zero size.
-   * @param reductionIndices The dimensions to reduce over.  Dimensions are reduced in the
-   *  order specified.  Omitting `reduction_indices` is equivalent to passing
-   *  `[n-1, n-2, ..., 0]`.  Negative indices from `-n` to `-1` are supported.
-   * @param options carries optional attributes values
-   * @return a new instance of ReduceJoin
-   */
-  public ReduceJoin reduceJoin(Operand<TString> inputs, Operand<TInt32> reductionIndices,
-      ReduceJoin.Options... options) {
-    return ReduceJoin.create(scope, inputs, reductionIndices, options);
-  }
-
-  /**
-   * Check if the input matches the regex pattern.
-   *  <p>
-   *  The input is a string tensor of any shape. The pattern is a scalar
-   *  string tensor which is applied to every element of the input tensor.
-   *  The boolean values (True or False) of the output tensor indicate
-   *  if the input matches the regex pattern provided.
-   *  <p>
-   *  The pattern follows the re2 syntax (https://github.com/google/re2/wiki/Syntax)
-   *  <p>
-   *  Examples:
-   *  <p>
-   *  >>> tf.strings.regex_full_match(["TF lib", "lib TF"], ".*lib$")
-   *  <tf.Tensor: shape=(2,), dtype=bool, numpy=array([ True, False])>
-   *  >>> tf.strings.regex_full_match(["TF lib", "lib TF"], ".*TF$")
-   *  <tf.Tensor: shape=(2,), dtype=bool, numpy=array([False,  True])>
-   *
-   * @param input A string tensor of the text to be processed.
-   * @param pattern A scalar string tensor containing the regular expression to match the input.
-   * @return a new instance of RegexFullMatch
-   */
-  public RegexFullMatch regexFullMatch(Operand<TString> input, Operand<TString> pattern) {
-    return RegexFullMatch.create(scope, input, pattern);
-  }
-
-  /**
-   * Replaces matches of the `pattern` regular expression in `input` with the
-   *  replacement string provided in `rewrite`.
-   *  <p>
-   *  It follows the re2 syntax (https://github.com/google/re2/wiki/Syntax)
-   *
-   * @param input The text to be processed.
-   * @param pattern The regular expression to be matched in the `input` strings.
-   * @param rewrite The rewrite string to be substituted for the `pattern` expression where it is
-   *  matched in the `input` strings.
-   * @param options carries optional attributes values
-   * @return a new instance of RegexReplace
-   */
-  public RegexReplace regexReplace(Operand<TString> input, Operand<TString> pattern,
-      Operand<TString> rewrite, RegexReplace.Options... options) {
-    return RegexReplace.create(scope, input, pattern, rewrite, options);
-  }
-
-  /**
-   * Formats a string template using a list of tensors.
-   *  <p>
-   *  Formats a string template using a list of tensors, pretty-printing tensor summaries.
-   *
-   * @param inputs The list of tensors to format into the placeholder string.
-   * @param options carries optional attributes values
-   * @return a new instance of StringFormat
-   */
-  public StringFormat stringFormat(Iterable<Operand<?>> inputs, StringFormat.Options... options) {
-    return StringFormat.create(scope, inputs, options);
-  }
-
-  /**
-   * String lengths of `input`.
-   *  <p>
-   *  Computes the length of each string given in the input tensor.
-   *  <p>
-   *  >>> strings = tf.constant(['Hello','TensorFlow', '\U0001F642'])
-   *  >>> tf.strings.length(strings).numpy() # default counts bytes
-   *  array([ 5, 10, 4], dtype=int32)
-   *  >>> tf.strings.length(strings, unit="UTF8_CHAR").numpy()
-   *  array([ 5, 10, 1], dtype=int32)
-   *
-   * @param input The strings for which to compute the length for each element.
-   * @param options carries optional attributes values
-   * @return a new instance of StringLength
-   */
-  public StringLength stringLength(Operand<TString> input, StringLength.Options... options) {
-    return StringLength.create(scope, input, options);
-  }
-
-  /**
-   * Creates ngrams from ragged string data.
-   *  <p>
-   *  This op accepts a ragged tensor with 1 ragged dimension containing only
-   *  strings and outputs a ragged tensor with 1 ragged dimension containing ngrams
-   *  of that string, joined along the innermost axis.
-   *
-   * @param <T> data type for {@code ngramsSplits()} output
-   * @param data The values tensor of the ragged string tensor to make ngrams out of. Must be a
-   *  1D string tensor.
-   * @param dataSplits The splits tensor of the ragged string tensor to make ngrams out of.
-   * @param separator The string to append between elements of the token. Use "" for no separator.
-   * @param ngramWidths The sizes of the ngrams to create.
-   * @param leftPad The string to use to pad the left side of the ngram sequence. Only used if
-   *  pad_width != 0.
-   * @param rightPad The string to use to pad the right side of the ngram sequence. Only used if
-   *  pad_width != 0.
-   * @param padWidth The number of padding elements to add to each side of each
-   *  sequence. Note that padding will never be greater than 'ngram_widths'-1
-   *  regardless of this value. If `pad_width=-1`, then add `max(ngram_widths)-1`
-   *  elements.
-   * @param preserveShortSequences
-   * @return a new instance of StringNGrams
-   */
-  public <T extends TNumber> StringNGrams<T> stringNGrams(Operand<TString> data,
-      Operand<T> dataSplits, String separator, List<Long> ngramWidths, String leftPad,
-      String rightPad, Long padWidth, Boolean preserveShortSequences) {
-    return StringNGrams.create(scope, data, dataSplits, separator, ngramWidths, leftPad, rightPad, padWidth, preserveShortSequences);
-  }
-
-  /**
-   * Split elements of `source` based on `sep` into a `SparseTensor`.
-   *  <p>
-   *  Let N be the size of source (typically N will be the batch size). Split each
-   *  element of `source` based on `sep` and return a `SparseTensor`
-   *  containing the split tokens. Empty tokens are ignored.
-   *  <p>
-   *  For example, N = 2, source[0] is 'hello world' and source[1] is 'a b c',
-   *  then the output will be
-   *  <pre>{@code
-   *  st.indices = [0, 0;
-   *                0, 1;
-   *                1, 0;
-   *                1, 1;
-   *                1, 2]
-   *  st.shape = [2, 3]
-   *  st.values = ['hello', 'world', 'a', 'b', 'c']
-   *  }</pre>
-   *  If `sep` is given, consecutive delimiters are not grouped together and are
-   *  deemed to delimit empty strings. For example, source of `"1<>2<><>3"` and
-   *  sep of `"<>"` returns `["1", "2", "", "3"]`. If `sep` is None or an empty
-   *  string, consecutive whitespace are regarded as a single separator, and the
-   *  result will contain no empty strings at the startor end if the string has
-   *  leading or trailing whitespace.
-   *  <p>
-   *  Note that the above mentioned behavior matches python's str.split.
-   *
-   * @param input `1-D` string `Tensor`, the strings to split.
-   * @param sep `0-D` string `Tensor`, the delimiter character.
-   * @param options carries optional attributes values
-   * @return a new instance of StringSplit
-   */
-  public StringSplit stringSplit(Operand<TString> input, Operand<TString> sep,
-      StringSplit.Options... options) {
-    return StringSplit.create(scope, input, sep, options);
-  }
-
-  /**
-   * Strip leading and trailing whitespaces from the Tensor.
-   *
-   * @param input A string `Tensor` of any shape.
-   * @return a new instance of Strip
-   */
-  public Strip strip(Operand<TString> input) {
-    return Strip.create(scope, input);
-  }
-
-  /**
-   * Return substrings from `Tensor` of strings.
-   *  <p>
-   *  For each string in the input `Tensor`, creates a substring starting at index
-   *  `pos` with a total length of `len`.
-   *  <p>
-   *  If `len` defines a substring that would extend beyond the length of the input
-   *  string, or if `len` is negative, then as many characters as possible are used.
-   *  <p>
-   *  A negative `pos` indicates distance within the string backwards from the end.
-   *  <p>
-   *  If `pos` specifies an index which is out of range for any of the input strings,
-   *  then an `InvalidArgumentError` is thrown.
-   *  <p>
-   *  `pos` and `len` must have the same shape, otherwise a `ValueError` is thrown on
-   *  Op creation.
-   *  <p>
-   *  <i>NOTE</i>: `strings.Substr` supports broadcasting up to two dimensions. More about
-   *  broadcasting
-   *  [here](http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html)
-   *  <p>
-   *  ---
-   *  <p>
-   *  Examples
-   *  <p>
-   *  Using scalar `pos` and `len`:
-   *  <pre>{@code
-   *  input = [b'Hello', b'World']
-   *  position = 1
-   *  length = 3
-   *
-   *  output = [b'ell', b'orl']
-   *  }</pre>
-   *  Using `pos` and `len` with same shape as `input`:
-   *  <pre>{@code
-   *  input = [[b'ten', b'eleven', b'twelve'],
-   *           [b'thirteen', b'fourteen', b'fifteen'],
-   *           [b'sixteen', b'seventeen', b'eighteen']]
-   *  position = [[1, 2, 3],
-   *              [1, 2, 3],
-   *              [1, 2, 3]]
-   *  length =   [[2, 3, 4],
-   *              [4, 3, 2],
-   *              [5, 5, 5]]
-   *
-   *  output = [[b'en', b'eve', b'lve'],
-   *            [b'hirt', b'urt', b'te'],
-   *            [b'ixtee', b'vente', b'hteen']]
-   *  }</pre>
-   *  Broadcasting `pos` and `len` onto `input`:
-   *  <pre>{@code
-   *  input = [[b'ten', b'eleven', b'twelve'],
-   *           [b'thirteen', b'fourteen', b'fifteen'],
-   *           [b'sixteen', b'seventeen', b'eighteen'],
-   *           [b'nineteen', b'twenty', b'twentyone']]
-   *  position = [1, 2, 3]
-   *  length =   [1, 2, 3]
-   *
-   *  output = [[b'e', b'ev', b'lve'],
-   *            [b'h', b'ur', b'tee'],
-   *            [b'i', b've', b'hte'],
-   *            [b'i', b'en', b'nty']]
-   *  }</pre>
-   *  Broadcasting `input` onto `pos` and `len`:
-   *  <pre>{@code
-   *  input = b'thirteen'
-   *  position = [1, 5, 7]
-   *  length =   [3, 2, 1]
-   *
-   *  output = [b'hir', b'ee', b'n']
-   *  }</pre>
-   *  Raises:
-   *  <p>
-   * `ValueError`: If the first argument cannot be converted to a
-   *       Tensor of `dtype string`.
-   * `InvalidArgumentError`: If indices are out of range.
-   * `ValueError`: If `pos` and `len` are not the same shape.
-   *
-   * @param input Tensor of strings
-   * @param pos Scalar defining the position of first character in each substring
-   * @param len Scalar defining the number of characters to include in each substring
-   * @param options carries optional attributes values
-   * @return a new instance of Substr
-   */
-  public <T extends TNumber> Substr substr(Operand<TString> input, Operand<T> pos, Operand<T> len,
-      Substr.Options... options) {
-    return Substr.create(scope, input, pos, len, options);
-  }
-
-  /**
-   * Converts each string in the input Tensor to its hash mod by a number of buckets.
-   *  <p>
-   *  The hash function is deterministic on the content of the string within the
-   *  process.
-   *  <p>
-   *  Note that the hash function may change from time to time.
-   *  This functionality will be deprecated and it's recommended to use
-   *  `tf.string_to_hash_bucket_fast()` or `tf.string_to_hash_bucket_strong()`.
-   *
-   * @param stringTensor
-   * @param numBuckets The number of buckets.
-   * @return a new instance of ToHashBucket
-   */
-  public ToHashBucket toHashBucket(Operand<TString> stringTensor, Long numBuckets) {
-    return ToHashBucket.create(scope, stringTensor, numBuckets);
-  }
-
-  /**
-   * Converts each string in the input Tensor to its hash mod by a number of buckets.
-   *  <p>
-   *  The hash function is deterministic on the content of the string within the
-   *  process and will never change. However, it is not suitable for cryptography.
-   *  This function may be used when CPU time is scarce and inputs are trusted or
-   *  unimportant. There is a risk of adversaries constructing inputs that all hash
-   *  to the same bucket. To prevent this problem, use a strong hash function with
-   *  `tf.string_to_hash_bucket_strong`.
-   *  <p>
-   *  Examples:
-   *  <p>
-   *  >>> tf.strings.to_hash_bucket_fast(["Hello", "TensorFlow", "2.x"], 3).numpy()
-   *  array([0, 2, 2])
-   *
-   * @param input The strings to assign a hash bucket.
-   * @param numBuckets The number of buckets.
-   * @return a new instance of ToHashBucketFast
-   */
-  public ToHashBucketFast toHashBucketFast(Operand<TString> input, Long numBuckets) {
-    return ToHashBucketFast.create(scope, input, numBuckets);
-  }
-
-  /**
-   * Converts each string in the input Tensor to its hash mod by a number of buckets.
-   *  <p>
-   *  The hash function is deterministic on the content of the string within the
-   *  process. The hash function is a keyed hash function, where attribute `key`
-   *  defines the key of the hash function. `key` is an array of 2 elements.
-   *  <p>
-   *  A strong hash is important when inputs may be malicious, e.g. URLs with
-   *  additional components. Adversaries could try to make their inputs hash to the
-   *  same bucket for a denial-of-service attack or to skew the results. A strong
-   *  hash can be used to make it difficult to find inputs with a skewed hash value
-   *  distribution over buckets. This requires that the hash function is
-   *  seeded by a high-entropy (random) "key" unknown to the adversary.
-   *  <p>
-   *  The additional robustness comes at a cost of roughly 4x higher compute
-   *  time than `tf.string_to_hash_bucket_fast`.
-   *  <p>
-   *  Examples:
-   *  <p>
-   *  >>> tf.strings.to_hash_bucket_strong(["Hello", "TF"], 3, [1, 2]).numpy()
-   *  array([2, 0])
-   *
-   * @param input The strings to assign a hash bucket.
-   * @param numBuckets The number of buckets.
-   * @param key The key used to seed the hash function, passed as a list of two uint64
-   *  elements.
-   * @return a new instance of ToHashBucketStrong
-   */
-  public ToHashBucketStrong toHashBucketStrong(Operand<TString> input, Long numBuckets,
-      List<Long> key) {
-    return ToHashBucketStrong.create(scope, input, numBuckets, key);
-  }
-
-  /**
-   * Converts each string in the input Tensor to the specified numeric type.
-   *  <p>
-   *  (Note that int32 overflow results in an error while float overflow
-   *  results in a rounded value.)
-   *  <p>
-   *  Example:
-   *  <p>
-   *  >>> strings = ["5.0", "3.0", "7.0"]
-   *  >>> tf.strings.to_number(strings)
-   *  <tf.Tensor: shape=(3,), dtype=float32, numpy=array([5., 3., 7.], dtype=float32)>
-   *
-   * @param <T> data type for {@code output()} output
-   * @param stringTensor
-   * @return a new instance of ToNumber
-   */
-  public ToNumber<TFloat32> toNumber(Operand<TString> stringTensor) {
-    return ToNumber.create(scope, stringTensor);
-  }
-
-  /**
-   * Converts each string in the input Tensor to the specified numeric type.
-   *  <p>
-   *  (Note that int32 overflow results in an error while float overflow
-   *  results in a rounded value.)
-   *  <p>
-   *  Example:
-   *  <p>
-   *  >>> strings = ["5.0", "3.0", "7.0"]
-   *  >>> tf.strings.to_number(strings)
-   *  <tf.Tensor: shape=(3,), dtype=float32, numpy=array([5., 3., 7.], dtype=float32)>
-   *
-   * @param <T> data type for {@code output()} output
-   * @param stringTensor
-   * @param outType The numeric type to interpret each string in `string_tensor` as.
-   * @return a new instance of ToNumber
-   */
-  public <T extends TNumber> ToNumber<T> toNumber(Operand<TString> stringTensor,
-      DataType<T> outType) {
-    return ToNumber.create(scope, stringTensor, outType);
-  }
-
-  /**
-   * Determine the script codes of a given tensor of Unicode integer code points.
-   *  <p>
-   *  This operation converts Unicode code points to script codes corresponding to
-   *  each code point. Script codes correspond to International Components for
-   *  Unicode (ICU) UScriptCode values. See http://icu-project.org/apiref/icu4c/uscript_8h.html.
-   *  Returns -1 (USCRIPT_INVALID_CODE) for invalid codepoints. Output shape will
-   *  match input shape.
-   *  <p>
-   *  Examples:
-   *  <p>
-   *  >>> tf.strings.unicode_script([1, 31, 38])
-   *  <tf.Tensor: shape=(3,), dtype=int32, numpy=array([0, 0, 0], dtype=int32)>
-   *
-   * @param input A Tensor of int32 Unicode code points.
-   * @return a new instance of UnicodeScript
-   */
-  public UnicodeScript unicodeScript(Operand<TInt32> input) {
-    return UnicodeScript.create(scope, input);
-  }
-
-  /**
-   * Transcode the input text from a source encoding to a destination encoding.
-   *  <p>
-   *  The input is a string tensor of any shape. The output is a string tensor of
-   *  the same shape containing the transcoded strings. Output strings are always
-   *  valid unicode. If the input contains invalid encoding positions, the
-   *  `errors` attribute sets the policy for how to deal with them. If the default
-   *  error-handling policy is used, invalid formatting will be substituted in the
-   *  output by the `replacement_char`. If the errors policy is to `ignore`, any
-   *  invalid encoding positions in the input are skipped and not included in the
-   *  output. If it set to `strict` then any invalid formatting will result in an
-   *  InvalidArgument error.
-   *  <p>
-   *  This operation can be used with `output_encoding = input_encoding` to enforce
-   *  correct formatting for inputs even if they are already in the desired encoding.
-   *  <p>
-   *  If the input is prefixed by a Byte Order Mark needed to determine encoding
-   *  (e.g. if the encoding is UTF-16 and the BOM indicates big-endian), then that
-   *  BOM will be consumed and not emitted into the output. If the input encoding
-   *  is marked with an explicit endianness (e.g. UTF-16-BE), then the BOM is
-   *  interpreted as a non-breaking-space and is preserved in the output (including
-   *  always for UTF-8).
-   *  <p>
-   *  The end result is that if the input is marked as an explicit endianness the
-   *  transcoding is faithful to all codepoints in the source. If it is not marked
-   *  with an explicit endianness, the BOM is not considered part of the string itself
-   *  but as metadata, and so is not preserved in the output.
-   *  <p>
-   *  Examples:
-   *  <p>
-   *  >>> tf.strings.unicode_transcode(["Hello", "TensorFlow", "2.x"], "UTF-8", "UTF-16-BE")
-   *  <tf.Tensor: shape=(3,), dtype=string, numpy=
-   *  array([b'\x00H\x00e\x00l\x00l\x00o',
-   *         b'\x00T\x00e\x00n\x00s\x00o\x00r\x00F\x00l\x00o\x00w',
-   *         b'\x002\x00.\x00x'], dtype=object)>
-   *  >>> tf.strings.unicode_transcode(["A", "B", "C"], "US ASCII", "UTF-8").numpy()
-   *  array([b'A', b'B', b'C'], dtype=object)
-   *
-   * @param input The text to be processed. Can have any shape.
-   * @param inputEncoding Text encoding of the input strings. This is any of the encodings supported
-   *  by ICU ucnv algorithmic converters. Examples: `"UTF-16", "US ASCII", "UTF-8"`.
-   * @param outputEncoding The unicode encoding to use in the output. Must be one of
-   *  `"UTF-8", "UTF-16-BE", "UTF-32-BE"`. Multi-byte encodings will be big-endian.
-   * @param options carries optional attributes values
-   * @return a new instance of UnicodeTranscode
-   */
-  public UnicodeTranscode unicodeTranscode(Operand<TString> input, String inputEncoding,
-      String outputEncoding, UnicodeTranscode.Options... options) {
-    return UnicodeTranscode.create(scope, input, inputEncoding, outputEncoding, options);
-  }
-
-  /**
-   * Joins the elements of `inputs` based on `segment_ids`.
-   *  <p>
-   *  Computes the string join along segments of a tensor.
-   *  Given `segment_ids` with rank `N` and `data` with rank `N+M`:
-   *  <p>
-   *      `output[i, k1...kM] = strings.join([data[j1...jN, k1...kM])`
-   *  <p>
-   *  where the join is over all [j1...jN] such that segment_ids[j1...jN] = i.
-   *  Strings are joined in row-major order.
-   *  <p>
-   *  For example:
-   *  <pre>{@code
-   *  inputs = [['Y', 'q', 'c'], ['Y', '6', '6'], ['p', 'G', 'a']]
-   *  output_array = string_ops.unsorted_segment_join(inputs=inputs,
-   *                                                  segment_ids=[1, 0, 1],
-   *                                                  num_segments=2,
-   *                                                  separator=':'))
-   *  # output_array ==> [['Y', '6', '6'], ['Y:p', 'q:G', 'c:a']]
-   *
-   *
-   *  inputs = ['this', 'is', 'a', 'test']
-   *  output_array = string_ops.unsorted_segment_join(inputs=inputs,
-   *                                                  segment_ids=[0, 0, 0, 0],
-   *                                                  num_segments=1,
-   *                                                  separator=':'))
-   *  # output_array ==> ['this:is:a:test']
-   *  }</pre>
-   *
-   * @param inputs The input to be joined.
-   * @param segmentIds A tensor whose shape is a prefix of data.shape.  Negative segment ids are not
-   *  supported.
-   * @param numSegments A scalar.
-   * @param options carries optional attributes values
-   * @return a new instance of UnsortedSegmentJoin
-   */
-  public <T extends TNumber, U extends TNumber> UnsortedSegmentJoin unsortedSegmentJoin(
-      Operand<TString> inputs, Operand<T> segmentIds, Operand<U> numSegments,
-      UnsortedSegmentJoin.Options... options) {
-    return UnsortedSegmentJoin.create(scope, inputs, segmentIds, numSegments, options);
-  }
-
-  /**
-   * Converts all lowercase characters into their respective uppercase replacements.
-   *  <p>
-   *  Example:
-   *  <p>
-   *  >>> tf.strings.upper("CamelCase string and ALL CAPS")
-   *  <tf.Tensor: shape=(), dtype=string, numpy=b'CAMELCASE STRING AND ALL CAPS'>
-   *
-   * @param input
-   * @param options carries optional attributes values
-   * @return a new instance of Upper
-   */
-  public Upper upper(Operand<TString> input, Upper.Options... options) {
-    return Upper.create(scope, input, options);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/SummaryOps.java b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/SummaryOps.java
deleted file mode 100644
index 6143335b8fd..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/SummaryOps.java
+++ /dev/null
@@ -1,198 +0,0 @@
-// Copyright 2020 The TensorFlow Authors. All Rights Reserved.
-//
-// Licensed under the Apache License, Version 2.0 (the "License");
-// you may not use this file except in compliance with the License.
-// You may obtain a copy of the License at
-//
-//     http://www.apache.org/licenses/LICENSE-2.0
-//
-// Unless required by applicable law or agreed to in writing, software
-// distributed under the License is distributed on an "AS IS" BASIS,
-// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-// See the License for the specific language governing permissions and
-// limitations under the License.
-// ==============================================================================
-//
-// This class has been generated, DO NOT EDIT!
-//
-package org.tensorflow.op;
-
-import org.tensorflow.Operand;
-import org.tensorflow.op.summary.AudioSummary;
-import org.tensorflow.op.summary.HistogramSummary;
-import org.tensorflow.op.summary.ImageSummary;
-import org.tensorflow.op.summary.MergeSummary;
-import org.tensorflow.op.summary.ScalarSummary;
-import org.tensorflow.op.summary.TensorSummary;
-import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.TString;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * An API for building {@code summary} operations as {@link Op Op}s
- *
- * @see {@link Ops}
- */
-public final class SummaryOps {
-  private final Scope scope;
-
-  SummaryOps(Scope scope) {
-    this.scope = scope;
-  }
-
-  /**
-   * Outputs a `Summary` protocol buffer with audio.
-   *  <p>
-   *  The summary has up to `max_outputs` summary values containing audio. The
-   *  audio is built from `tensor` which must be 3-D with shape `[batch_size,
-   *  frames, channels]` or 2-D with shape `[batch_size, frames]`. The values are
-   *  assumed to be in the range of `[-1.0, 1.0]` with a sample rate of `sample_rate`.
-   *  <p>
-   *  The `tag` argument is a scalar `Tensor` of type `string`.  It is used to
-   *  build the `tag` of the summary values:
-   *  <ul>
-   *  <li>
-   *  If `max_outputs` is 1, the summary value tag is '<i>tag</i>/audio'.
-   *  </li>
-   *  <li>
-   *  If `max_outputs` is greater than 1, the summary value tags are
-   *     generated sequentially as '<i>tag</i>/audio/0', '<i>tag</i>/audio/1', etc.
-   *
-   * @param tag Scalar. Used to build the `tag` attribute of the summary values.
-   * @param tensor 2-D of shape `[batch_size, frames]`.
-   * @param sampleRate The sample rate of the signal in hertz.
-   * @param options carries optional attributes values
-   * @return a new instance of AudioSummary
-   */
-  public AudioSummary audioSummary(Operand<TString> tag, Operand<TFloat32> tensor,
-      Operand<TFloat32> sampleRate, AudioSummary.Options... options) {
-    return AudioSummary.create(scope, tag, tensor, sampleRate, options);
-  }
-
-  /**
-   * Outputs a `Summary` protocol buffer with a histogram.
-   *  <p>
-   *  The generated
-   *  [`Summary`](https://www.tensorflow.org/code/tensorflow/core/framework/summary.proto)
-   *  has one summary value containing a histogram for `values`.
-   *  <p>
-   *  This op reports an `InvalidArgument` error if any value is not finite.
-   *
-   * @param tag Scalar.  Tag to use for the `Summary.Value`.
-   * @param values Any shape. Values to use to build the histogram.
-   * @return a new instance of HistogramSummary
-   */
-  public <T extends TNumber> HistogramSummary histogramSummary(Operand<TString> tag,
-      Operand<T> values) {
-    return HistogramSummary.create(scope, tag, values);
-  }
-
-  /**
-   * Outputs a `Summary` protocol buffer with images.
-   *  <p>
-   *  The summary has up to `max_images` summary values containing images. The
-   *  images are built from `tensor` which must be 4-D with shape `[batch_size,
-   *  height, width, channels]` and where `channels` can be:
-   *  <ul>
-   *  <li>
-   *  1: `tensor` is interpreted as Grayscale.
-   *  </li>
-   *  <li>
-   *  3: `tensor` is interpreted as RGB.
-   *  </li>
-   *  <li>
-   *  4: `tensor` is interpreted as RGBA.
-   *  </li>
-   *  </ul>
-   *  The images have the same number of channels as the input tensor. For float
-   *  input, the values are normalized one image at a time to fit in the range
-   *  `[0, 255]`.  `uint8` values are unchanged.  The op uses two different
-   *  normalization algorithms:
-   *  <ul>
-   *  <li>
-   *  If the input values are all positive, they are rescaled so the largest one
-   *     is 255.
-   *  </li>
-   *  <li>
-   *  If any input value is negative, the values are shifted so input value 0.0
-   *     is at 127.  They are then rescaled so that either the smallest value is 0,
-   *     or the largest one is 255.
-   *  </li>
-   *  </ul>
-   *  The `tag` argument is a scalar `Tensor` of type `string`.  It is used to
-   *  build the `tag` of the summary values:
-   *  <ul>
-   *  <li>
-   *  If `max_images` is 1, the summary value tag is '<i>tag</i>/image'.
-   *  </li>
-   *  <li>
-   *  If `max_images` is greater than 1, the summary value tags are
-   *     generated sequentially as '<i>tag</i>/image/0', '<i>tag</i>/image/1', etc.
-   *  </li>
-   *  </ul>
-   *  The `bad_color` argument is the color to use in the generated images for
-   *  non-finite input values.  It is a `uint8` 1-D tensor of length `channels`.
-   *  Each element must be in the range `[0, 255]` (It represents the value of a
-   *  pixel in the output image).  Non-finite values in the input tensor are
-   *  replaced by this tensor in the output image.  The default value is the color
-   *  red.
-   *
-   * @param tag Scalar. Used to build the `tag` attribute of the summary values.
-   * @param tensor 4-D of shape `[batch_size, height, width, channels]` where
-   *  `channels` is 1, 3, or 4.
-   * @param options carries optional attributes values
-   * @return a new instance of ImageSummary
-   */
-  public <T extends TNumber> ImageSummary imageSummary(Operand<TString> tag, Operand<T> tensor,
-      ImageSummary.Options... options) {
-    return ImageSummary.create(scope, tag, tensor, options);
-  }
-
-  /**
-   * Merges summaries.
-   *  <p>
-   *  This op creates a
-   *  [`Summary`](https://www.tensorflow.org/code/tensorflow/core/framework/summary.proto)
-   *  protocol buffer that contains the union of all the values in the input
-   *  summaries.
-   *  <p>
-   *  When the Op is run, it reports an `InvalidArgument` error if multiple values
-   *  in the summaries to merge use the same tag.
-   *
-   * @param inputs Can be of any shape.  Each must contain serialized `Summary` protocol
-   *  buffers.
-   * @return a new instance of MergeSummary
-   */
-  public MergeSummary mergeSummary(Iterable<Operand<TString>> inputs) {
-    return MergeSummary.create(scope, inputs);
-  }
-
-  /**
-   * Outputs a `Summary` protocol buffer with scalar values.
-   *  <p>
-   *  The input `tags` and `values` must have the same shape.  The generated summary
-   *  has a summary value for each tag-value pair in `tags` and `values`.
-   *
-   * @param tags Tags for the summary.
-   * @param values Same shape as `tags.  Values for the summary.
-   * @return a new instance of ScalarSummary
-   */
-  public <T extends TNumber> ScalarSummary scalarSummary(Operand<TString> tags, Operand<T> values) {
-    return ScalarSummary.create(scope, tags, values);
-  }
-
-  /**
-   * Outputs a `Summary` protocol buffer with a tensor and per-plugin data.
-   *
-   * @param tag A string attached to this summary. Used for organization in TensorBoard.
-   * @param tensor A tensor to serialize.
-   * @param serializedSummaryMetadata A serialized SummaryMetadata proto. Contains plugin
-   *  data.
-   * @return a new instance of TensorSummary
-   */
-  public <T extends TType> TensorSummary tensorSummary(Operand<TString> tag, Operand<T> tensor,
-      Operand<TString> serializedSummaryMetadata) {
-    return TensorSummary.create(scope, tag, tensor, serializedSummaryMetadata);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/TrainOps.java b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/TrainOps.java
deleted file mode 100644
index 2c5d8752136..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/TrainOps.java
+++ /dev/null
@@ -1,1661 +0,0 @@
-// Copyright 2020 The TensorFlow Authors. All Rights Reserved.
-//
-// Licensed under the Apache License, Version 2.0 (the "License");
-// you may not use this file except in compliance with the License.
-// You may obtain a copy of the License at
-//
-//     http://www.apache.org/licenses/LICENSE-2.0
-//
-// Unless required by applicable law or agreed to in writing, software
-// distributed under the License is distributed on an "AS IS" BASIS,
-// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-// See the License for the specific language governing permissions and
-// limitations under the License.
-// ==============================================================================
-//
-// This class has been generated, DO NOT EDIT!
-//
-package org.tensorflow.op;
-
-import java.util.List;
-import org.tensorflow.DataType;
-import org.tensorflow.Operand;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.op.train.AccumulatorApplyGradient;
-import org.tensorflow.op.train.AccumulatorNumAccumulated;
-import org.tensorflow.op.train.AccumulatorSetGlobalStep;
-import org.tensorflow.op.train.AccumulatorTakeGradient;
-import org.tensorflow.op.train.ApplyAdadelta;
-import org.tensorflow.op.train.ApplyAdagrad;
-import org.tensorflow.op.train.ApplyAdagradDa;
-import org.tensorflow.op.train.ApplyAdam;
-import org.tensorflow.op.train.ApplyAddSign;
-import org.tensorflow.op.train.ApplyCenteredRmsProp;
-import org.tensorflow.op.train.ApplyFtrl;
-import org.tensorflow.op.train.ApplyGradientDescent;
-import org.tensorflow.op.train.ApplyMomentum;
-import org.tensorflow.op.train.ApplyPowerSign;
-import org.tensorflow.op.train.ApplyProximalAdagrad;
-import org.tensorflow.op.train.ApplyProximalGradientDescent;
-import org.tensorflow.op.train.ApplyRmsProp;
-import org.tensorflow.op.train.BatchMatMul;
-import org.tensorflow.op.train.ConditionalAccumulator;
-import org.tensorflow.op.train.GenerateVocabRemapping;
-import org.tensorflow.op.train.MergeV2Checkpoints;
-import org.tensorflow.op.train.NegTrain;
-import org.tensorflow.op.train.PreventGradient;
-import org.tensorflow.op.train.ResourceApplyAdadelta;
-import org.tensorflow.op.train.ResourceApplyAdagradDa;
-import org.tensorflow.op.train.ResourceApplyAdam;
-import org.tensorflow.op.train.ResourceApplyAdamWithAmsgrad;
-import org.tensorflow.op.train.ResourceApplyAddSign;
-import org.tensorflow.op.train.ResourceApplyCenteredRmsProp;
-import org.tensorflow.op.train.ResourceApplyFtrl;
-import org.tensorflow.op.train.ResourceApplyGradientDescent;
-import org.tensorflow.op.train.ResourceApplyKerasMomentum;
-import org.tensorflow.op.train.ResourceApplyMomentum;
-import org.tensorflow.op.train.ResourceApplyPowerSign;
-import org.tensorflow.op.train.ResourceApplyProximalAdagrad;
-import org.tensorflow.op.train.ResourceApplyProximalGradientDescent;
-import org.tensorflow.op.train.ResourceApplyRmsProp;
-import org.tensorflow.op.train.ResourceSparseApplyAdadelta;
-import org.tensorflow.op.train.ResourceSparseApplyAdagrad;
-import org.tensorflow.op.train.ResourceSparseApplyAdagradDa;
-import org.tensorflow.op.train.ResourceSparseApplyCenteredRmsProp;
-import org.tensorflow.op.train.ResourceSparseApplyFtrl;
-import org.tensorflow.op.train.ResourceSparseApplyKerasMomentum;
-import org.tensorflow.op.train.ResourceSparseApplyMomentum;
-import org.tensorflow.op.train.ResourceSparseApplyProximalAdagrad;
-import org.tensorflow.op.train.ResourceSparseApplyProximalGradientDescent;
-import org.tensorflow.op.train.ResourceSparseApplyRmsProp;
-import org.tensorflow.op.train.Restore;
-import org.tensorflow.op.train.RestoreSlice;
-import org.tensorflow.op.train.Save;
-import org.tensorflow.op.train.SaveSlices;
-import org.tensorflow.op.train.SdcaFprint;
-import org.tensorflow.op.train.SdcaShrinkL1;
-import org.tensorflow.op.train.SparseApplyAdadelta;
-import org.tensorflow.op.train.SparseApplyAdagradDa;
-import org.tensorflow.op.train.SparseApplyCenteredRmsProp;
-import org.tensorflow.op.train.SparseApplyFtrl;
-import org.tensorflow.op.train.SparseApplyMomentum;
-import org.tensorflow.op.train.SparseApplyProximalAdagrad;
-import org.tensorflow.op.train.SparseApplyProximalGradientDescent;
-import org.tensorflow.op.train.SparseApplyRmsProp;
-import org.tensorflow.op.train.TileGrad;
-import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.TString;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * An API for building {@code train} operations as {@link Op Op}s
- *
- * @see {@link Ops}
- */
-public final class TrainOps {
-  private final Scope scope;
-
-  TrainOps(Scope scope) {
-    this.scope = scope;
-  }
-
-  /**
-   * Applies a gradient to a given accumulator.
-   *  <p>
-   *  Does not add if local_step is lesser than the accumulator's global_step.
-   *
-   * @param handle The handle to a accumulator.
-   * @param localStep The local_step value at which the gradient was computed.
-   * @param gradient A tensor of the gradient to be accumulated.
-   * @return a new instance of AccumulatorApplyGradient
-   */
-  public <T extends TType> AccumulatorApplyGradient accumulatorApplyGradient(
-      Operand<TString> handle, Operand<TInt64> localStep, Operand<T> gradient) {
-    return AccumulatorApplyGradient.create(scope, handle, localStep, gradient);
-  }
-
-  /**
-   * Returns the number of gradients aggregated in the given accumulators.
-   *
-   * @param handle The handle to an accumulator.
-   * @return a new instance of AccumulatorNumAccumulated
-   */
-  public AccumulatorNumAccumulated accumulatorNumAccumulated(Operand<TString> handle) {
-    return AccumulatorNumAccumulated.create(scope, handle);
-  }
-
-  /**
-   * Updates the accumulator with a new value for global_step.
-   *  <p>
-   *  Logs warning if the accumulator's value is already higher than
-   *  new_global_step.
-   *
-   * @param handle The handle to an accumulator.
-   * @param newGlobalStep The new global_step value to set.
-   * @return a new instance of AccumulatorSetGlobalStep
-   */
-  public AccumulatorSetGlobalStep accumulatorSetGlobalStep(Operand<TString> handle,
-      Operand<TInt64> newGlobalStep) {
-    return AccumulatorSetGlobalStep.create(scope, handle, newGlobalStep);
-  }
-
-  /**
-   * Extracts the average gradient in the given ConditionalAccumulator.
-   *  <p>
-   *  The op blocks until sufficient (i.e., more than num_required)
-   *  gradients have been accumulated.  If the accumulator has already
-   *  aggregated more than num_required gradients, it returns the average of
-   *  the accumulated gradients.  Also automatically increments the recorded
-   *  global_step in the accumulator by 1, and resets the aggregate to 0.
-   *
-   * @param <T> data type for {@code average()} output
-   * @param handle The handle to an accumulator.
-   * @param numRequired Number of gradients required before we return an aggregate.
-   * @param dtype The data type of accumulated gradients. Needs to correspond to the type
-   *  of the accumulator.
-   * @return a new instance of AccumulatorTakeGradient
-   */
-  public <T extends TType> AccumulatorTakeGradient<T> accumulatorTakeGradient(
-      Operand<TString> handle, Operand<TInt32> numRequired, DataType<T> dtype) {
-    return AccumulatorTakeGradient.create(scope, handle, numRequired, dtype);
-  }
-
-  /**
-   * Update '*var' according to the adadelta scheme.
-   *  <p>
-   *  accum = rho() * accum + (1 - rho()) * grad.square();
-   *  update = (update_accum + epsilon).sqrt() * (accum + epsilon()).rsqrt() * grad;
-   *  update_accum = rho() * update_accum + (1 - rho()) * update.square();
-   *  var -= update;
-   *
-   * @param <T> data type for {@code out()} output
-   * @param var Should be from a Variable().
-   * @param accum Should be from a Variable().
-   * @param accumUpdate Should be from a Variable().
-   * @param lr Scaling factor. Must be a scalar.
-   * @param rho Decay factor. Must be a scalar.
-   * @param epsilon Constant factor. Must be a scalar.
-   * @param grad The gradient.
-   * @param options carries optional attributes values
-   * @return a new instance of ApplyAdadelta
-   */
-  public <T extends TType> ApplyAdadelta<T> applyAdadelta(Operand<T> var, Operand<T> accum,
-      Operand<T> accumUpdate, Operand<T> lr, Operand<T> rho, Operand<T> epsilon, Operand<T> grad,
-      ApplyAdadelta.Options... options) {
-    return ApplyAdadelta.create(scope, var, accum, accumUpdate, lr, rho, epsilon, grad, options);
-  }
-
-  /**
-   * Update '*var' according to the adagrad scheme.
-   *  <p>
-   *  accum += grad * grad
-   *  var -= lr * grad * (1 / sqrt(accum))
-   *
-   * @param <T> data type for {@code out()} output
-   * @param var Should be from a Variable().
-   * @param accum Should be from a Variable().
-   * @param lr Scaling factor. Must be a scalar.
-   * @param grad The gradient.
-   * @param options carries optional attributes values
-   * @return a new instance of ApplyAdagrad
-   */
-  public <T extends TType> ApplyAdagrad<T> applyAdagrad(Operand<T> var, Operand<T> accum,
-      Operand<T> lr, Operand<T> grad, ApplyAdagrad.Options... options) {
-    return ApplyAdagrad.create(scope, var, accum, lr, grad, options);
-  }
-
-  /**
-   * Update '*var' according to the proximal adagrad scheme.
-   *
-   * @param <T> data type for {@code out()} output
-   * @param var Should be from a Variable().
-   * @param gradientAccumulator Should be from a Variable().
-   * @param gradientSquaredAccumulator Should be from a Variable().
-   * @param grad The gradient.
-   * @param lr Scaling factor. Must be a scalar.
-   * @param l1 L1 regularization. Must be a scalar.
-   * @param l2 L2 regularization. Must be a scalar.
-   * @param globalStep Training step number. Must be a scalar.
-   * @param options carries optional attributes values
-   * @return a new instance of ApplyAdagradDa
-   */
-  public <T extends TType> ApplyAdagradDa<T> applyAdagradDa(Operand<T> var,
-      Operand<T> gradientAccumulator, Operand<T> gradientSquaredAccumulator, Operand<T> grad,
-      Operand<T> lr, Operand<T> l1, Operand<T> l2, Operand<TInt64> globalStep,
-      ApplyAdagradDa.Options... options) {
-    return ApplyAdagradDa.create(scope, var, gradientAccumulator, gradientSquaredAccumulator, grad, lr, l1, l2, globalStep, options);
-  }
-
-  /**
-   * Update '*var' according to the Adam algorithm.
-   *  <p>
-   *  $$lr_t := \text{learning\_rate} * \sqrt{1 - beta_2^t} / (1 - beta_1^t)$$
-   *  $$m_t := beta_1 * m_{t-1} + (1 - beta_1) * g$$
-   *  $$v_t := beta_2 * v_{t-1} + (1 - beta_2) * g * g$$
-   *  $$variable := variable - lr_t * m_t / (\sqrt{v_t} + \epsilon)$$
-   *
-   * @param <T> data type for {@code out()} output
-   * @param var Should be from a Variable().
-   * @param m Should be from a Variable().
-   * @param v Should be from a Variable().
-   * @param beta1Power Must be a scalar.
-   * @param beta2Power Must be a scalar.
-   * @param lr Scaling factor. Must be a scalar.
-   * @param beta1 Momentum factor. Must be a scalar.
-   * @param beta2 Momentum factor. Must be a scalar.
-   * @param epsilon Ridge term. Must be a scalar.
-   * @param grad The gradient.
-   * @param options carries optional attributes values
-   * @return a new instance of ApplyAdam
-   */
-  public <T extends TType> ApplyAdam<T> applyAdam(Operand<T> var, Operand<T> m, Operand<T> v,
-      Operand<T> beta1Power, Operand<T> beta2Power, Operand<T> lr, Operand<T> beta1,
-      Operand<T> beta2, Operand<T> epsilon, Operand<T> grad, ApplyAdam.Options... options) {
-    return ApplyAdam.create(scope, var, m, v, beta1Power, beta2Power, lr, beta1, beta2, epsilon, grad, options);
-  }
-
-  /**
-   * Update '*var' according to the AddSign update.
-   *  <p>
-   *  m_t <- beta1 * m_{t-1} + (1 - beta1) * g
-   *  update <- (alpha + sign_decay * sign(g) *sign(m)) * g
-   *  variable <- variable - lr_t * update
-   *
-   * @param <T> data type for {@code out()} output
-   * @param var Should be from a Variable().
-   * @param m Should be from a Variable().
-   * @param lr Scaling factor. Must be a scalar.
-   * @param alpha Must be a scalar.
-   * @param signDecay Must be a scalar.
-   * @param beta Must be a scalar.
-   * @param grad The gradient.
-   * @param options carries optional attributes values
-   * @return a new instance of ApplyAddSign
-   */
-  public <T extends TType> ApplyAddSign<T> applyAddSign(Operand<T> var, Operand<T> m, Operand<T> lr,
-      Operand<T> alpha, Operand<T> signDecay, Operand<T> beta, Operand<T> grad,
-      ApplyAddSign.Options... options) {
-    return ApplyAddSign.create(scope, var, m, lr, alpha, signDecay, beta, grad, options);
-  }
-
-  /**
-   * Update '*var' according to the centered RMSProp algorithm.
-   *  <p>
-   *  The centered RMSProp algorithm uses an estimate of the centered second moment
-   *  (i.e., the variance) for normalization, as opposed to regular RMSProp, which
-   *  uses the (uncentered) second moment. This often helps with training, but is
-   *  slightly more expensive in terms of computation and memory.
-   *  <p>
-   *  Note that in dense implementation of this algorithm, mg, ms, and mom will
-   *  update even if the grad is zero, but in this sparse implementation, mg, ms,
-   *  and mom will not update in iterations during which the grad is zero.
-   *  <p>
-   *  mean_square = decay * mean_square + (1-decay) * gradient ** 2
-   *  mean_grad = decay * mean_grad + (1-decay) * gradient
-   *  <p>
-   *  Delta = learning_rate * gradient / sqrt(mean_square + epsilon - mean_grad ** 2)
-   *  <p>
-   *  mg <- rho * mg_{t-1} + (1-rho) * grad
-   *  ms <- rho * ms_{t-1} + (1-rho) * grad * grad
-   *  mom <- momentum * mom_{t-1} + lr * grad / sqrt(ms - mg * mg + epsilon)
-   *  var <- var - mom
-   *
-   * @param <T> data type for {@code out()} output
-   * @param var Should be from a Variable().
-   * @param mg Should be from a Variable().
-   * @param ms Should be from a Variable().
-   * @param mom Should be from a Variable().
-   * @param lr Scaling factor. Must be a scalar.
-   * @param rho Decay rate. Must be a scalar.
-   * @param momentum
-   * @param epsilon Ridge term. Must be a scalar.
-   * @param grad The gradient.
-   * @param options carries optional attributes values
-   * @return a new instance of ApplyCenteredRmsProp
-   */
-  public <T extends TType> ApplyCenteredRmsProp<T> applyCenteredRmsProp(Operand<T> var,
-      Operand<T> mg, Operand<T> ms, Operand<T> mom, Operand<T> lr, Operand<T> rho,
-      Operand<T> momentum, Operand<T> epsilon, Operand<T> grad,
-      ApplyCenteredRmsProp.Options... options) {
-    return ApplyCenteredRmsProp.create(scope, var, mg, ms, mom, lr, rho, momentum, epsilon, grad, options);
-  }
-
-  /**
-   * Update '*var' according to the Ftrl-proximal scheme.
-   *  <p>
-   *  grad_with_shrinkage = grad + 2 * l2_shrinkage * var
-   *  accum_new = accum + grad * grad
-   *  linear += grad_with_shrinkage -
-   *      (accum_new^(-lr_power) - accum^(-lr_power)) / lr * var
-   *  quadratic = 1.0 / (accum_new^(lr_power) * lr) + 2 * l2
-   *  var = (sign(linear) * l1 - linear) / quadratic if |linear| > l1 else 0.0
-   *  accum = accum_new
-   *
-   * @param <T> data type for {@code out()} output
-   * @param var Should be from a Variable().
-   * @param accum Should be from a Variable().
-   * @param linear Should be from a Variable().
-   * @param grad The gradient.
-   * @param lr Scaling factor. Must be a scalar.
-   * @param l1 L1 regularization. Must be a scalar.
-   * @param l2 L2 shrinkage regularization. Must be a scalar.
-   * @param l2Shrinkage
-   * @param lrPower Scaling factor. Must be a scalar.
-   * @param options carries optional attributes values
-   * @return a new instance of ApplyFtrl
-   */
-  public <T extends TType> ApplyFtrl<T> applyFtrl(Operand<T> var, Operand<T> accum,
-      Operand<T> linear, Operand<T> grad, Operand<T> lr, Operand<T> l1, Operand<T> l2,
-      Operand<T> l2Shrinkage, Operand<T> lrPower, ApplyFtrl.Options... options) {
-    return ApplyFtrl.create(scope, var, accum, linear, grad, lr, l1, l2, l2Shrinkage, lrPower, options);
-  }
-
-  /**
-   * Update '*var' by subtracting 'alpha' * 'delta' from it.
-   *
-   * @param <T> data type for {@code out()} output
-   * @param var Should be from a Variable().
-   * @param alpha Scaling factor. Must be a scalar.
-   * @param delta The change.
-   * @param options carries optional attributes values
-   * @return a new instance of ApplyGradientDescent
-   */
-  public <T extends TType> ApplyGradientDescent<T> applyGradientDescent(Operand<T> var,
-      Operand<T> alpha, Operand<T> delta, ApplyGradientDescent.Options... options) {
-    return ApplyGradientDescent.create(scope, var, alpha, delta, options);
-  }
-
-  /**
-   * Update '*var' according to the momentum scheme.
-   *  <p>
-   *  Set use_nesterov = True if you want to use Nesterov momentum.
-   *  <p>
-   *  accum = accum * momentum + grad
-   *  var -= lr * accum
-   *
-   * @param <T> data type for {@code out()} output
-   * @param var Should be from a Variable().
-   * @param accum Should be from a Variable().
-   * @param lr Scaling factor. Must be a scalar.
-   * @param grad The gradient.
-   * @param momentum Momentum. Must be a scalar.
-   * @param options carries optional attributes values
-   * @return a new instance of ApplyMomentum
-   */
-  public <T extends TType> ApplyMomentum<T> applyMomentum(Operand<T> var, Operand<T> accum,
-      Operand<T> lr, Operand<T> grad, Operand<T> momentum, ApplyMomentum.Options... options) {
-    return ApplyMomentum.create(scope, var, accum, lr, grad, momentum, options);
-  }
-
-  /**
-   * Update '*var' according to the AddSign update.
-   *  <p>
-   *  m_t <- beta1 * m_{t-1} + (1 - beta1) * g
-   *  update <- exp(logbase * sign_decay * sign(g) * sign(m_t)) * g
-   *  variable <- variable - lr_t * update
-   *
-   * @param <T> data type for {@code out()} output
-   * @param var Should be from a Variable().
-   * @param m Should be from a Variable().
-   * @param lr Scaling factor. Must be a scalar.
-   * @param logbase Must be a scalar.
-   * @param signDecay Must be a scalar.
-   * @param beta Must be a scalar.
-   * @param grad The gradient.
-   * @param options carries optional attributes values
-   * @return a new instance of ApplyPowerSign
-   */
-  public <T extends TType> ApplyPowerSign<T> applyPowerSign(Operand<T> var, Operand<T> m,
-      Operand<T> lr, Operand<T> logbase, Operand<T> signDecay, Operand<T> beta, Operand<T> grad,
-      ApplyPowerSign.Options... options) {
-    return ApplyPowerSign.create(scope, var, m, lr, logbase, signDecay, beta, grad, options);
-  }
-
-  /**
-   * Update '*var' and '*accum' according to FOBOS with Adagrad learning rate.
-   *  <p>
-   *  accum += grad <i> grad
-   *  prox_v = var - lr </i> grad <i> (1 / sqrt(accum))
-   *  var = sign(prox_v)/(1+lr</i>l2) <i> max{|prox_v|-lr</i>l1,0}
-   *
-   * @param <T> data type for {@code out()} output
-   * @param var Should be from a Variable().
-   * @param accum Should be from a Variable().
-   * @param lr Scaling factor. Must be a scalar.
-   * @param l1 L1 regularization. Must be a scalar.
-   * @param l2 L2 regularization. Must be a scalar.
-   * @param grad The gradient.
-   * @param options carries optional attributes values
-   * @return a new instance of ApplyProximalAdagrad
-   */
-  public <T extends TType> ApplyProximalAdagrad<T> applyProximalAdagrad(Operand<T> var,
-      Operand<T> accum, Operand<T> lr, Operand<T> l1, Operand<T> l2, Operand<T> grad,
-      ApplyProximalAdagrad.Options... options) {
-    return ApplyProximalAdagrad.create(scope, var, accum, lr, l1, l2, grad, options);
-  }
-
-  /**
-   * Update '*var' as FOBOS algorithm with fixed learning rate.
-   *  <p>
-   *  prox_v = var - alpha <i> delta
-   *  var = sign(prox_v)/(1+alpha</i>l2) <i> max{|prox_v|-alpha</i>l1,0}
-   *
-   * @param <T> data type for {@code out()} output
-   * @param var Should be from a Variable().
-   * @param alpha Scaling factor. Must be a scalar.
-   * @param l1 L1 regularization. Must be a scalar.
-   * @param l2 L2 regularization. Must be a scalar.
-   * @param delta The change.
-   * @param options carries optional attributes values
-   * @return a new instance of ApplyProximalGradientDescent
-   */
-  public <T extends TType> ApplyProximalGradientDescent<T> applyProximalGradientDescent(
-      Operand<T> var, Operand<T> alpha, Operand<T> l1, Operand<T> l2, Operand<T> delta,
-      ApplyProximalGradientDescent.Options... options) {
-    return ApplyProximalGradientDescent.create(scope, var, alpha, l1, l2, delta, options);
-  }
-
-  /**
-   * Update '*var' according to the RMSProp algorithm.
-   *  <p>
-   *  Note that in dense implementation of this algorithm, ms and mom will
-   *  update even if the grad is zero, but in this sparse implementation, ms
-   *  and mom will not update in iterations during which the grad is zero.
-   *  <p>
-   *  mean_square = decay * mean_square + (1-decay) * gradient ** 2
-   *  Delta = learning_rate * gradient / sqrt(mean_square + epsilon)
-   *  <p>
-   *  ms <- rho * ms_{t-1} + (1-rho) * grad * grad
-   *  mom <- momentum * mom_{t-1} + lr * grad / sqrt(ms + epsilon)
-   *  var <- var - mom
-   *
-   * @param <T> data type for {@code out()} output
-   * @param var Should be from a Variable().
-   * @param ms Should be from a Variable().
-   * @param mom Should be from a Variable().
-   * @param lr Scaling factor. Must be a scalar.
-   * @param rho Decay rate. Must be a scalar.
-   * @param momentum
-   * @param epsilon Ridge term. Must be a scalar.
-   * @param grad The gradient.
-   * @param options carries optional attributes values
-   * @return a new instance of ApplyRmsProp
-   */
-  public <T extends TType> ApplyRmsProp<T> applyRmsProp(Operand<T> var, Operand<T> ms,
-      Operand<T> mom, Operand<T> lr, Operand<T> rho, Operand<T> momentum, Operand<T> epsilon,
-      Operand<T> grad, ApplyRmsProp.Options... options) {
-    return ApplyRmsProp.create(scope, var, ms, mom, lr, rho, momentum, epsilon, grad, options);
-  }
-
-  /**
-   * Multiplies slices of two tensors in batches.
-   *  <p>
-   *  Multiplies all slices of `Tensor` `x` and `y` (each slice can be
-   *  viewed as an element of a batch), and arranges the individual results
-   *  in a single output tensor of the same batch size. Each of the
-   *  individual slices can optionally be adjointed (to adjoint a matrix
-   *  means to transpose and conjugate it) before multiplication by setting
-   *  the `adj_x` or `adj_y` flag to `True`, which are by default `False`.
-   *  <p>
-   *  The input tensors `x` and `y` are 2-D or higher with shape `[..., r_x, c_x]`
-   *  and `[..., r_y, c_y]`.
-   *  <p>
-   *  The output tensor is 2-D or higher with shape `[..., r_o, c_o]`, where:
-   *  <p>
-   *      r_o = c_x if adj_x else r_x
-   *      c_o = r_y if adj_y else c_y
-   *  <p>
-   *  It is computed as:
-   *  <p>
-   *      output[..., :, :] = matrix(x[..., :, :]) * matrix(y[..., :, :])
-   *  <p>
-   *  <i>NOTE</i>: `train.BatchMatMul` supports broadcasting in the batch dimensions. More
-   *  about broadcasting
-   *  [here](http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html).
-   *
-   * @param <T> data type for {@code output()} output
-   * @param x 2-D or higher with shape `[..., r_x, c_x]`.
-   * @param y 2-D or higher with shape `[..., r_y, c_y]`.
-   * @param options carries optional attributes values
-   * @return a new instance of BatchMatMul
-   */
-  public <T extends TType> BatchMatMul<T> batchMatMul(Operand<T> x, Operand<T> y,
-      BatchMatMul.Options... options) {
-    return BatchMatMul.create(scope, x, y, options);
-  }
-
-  /**
-   * A conditional accumulator for aggregating gradients.
-   *  <p>
-   *  The accumulator accepts gradients marked with local_step greater or
-   *  equal to the most recent global_step known to the accumulator. The
-   *  average can be extracted from the accumulator, provided sufficient
-   *  gradients have been accumulated. Extracting the average automatically
-   *  resets the aggregate to 0, and increments the global_step recorded by
-   *  the accumulator.
-   *
-   * @param dtype The type of the value being accumulated.
-   * @param shape The shape of the values, can be [], in which case shape is unknown.
-   * @param options carries optional attributes values
-   * @return a new instance of ConditionalAccumulator
-   */
-  public <T extends TType> ConditionalAccumulator conditionalAccumulator(DataType<T> dtype,
-      Shape shape, ConditionalAccumulator.Options... options) {
-    return ConditionalAccumulator.create(scope, dtype, shape, options);
-  }
-
-  /**
-   * Given a path to new and old vocabulary files, returns a remapping Tensor of
-   *  <p>
-   *  length `num_new_vocab`, where `remapping[i]` contains the row number in the old
-   *  vocabulary that corresponds to row `i` in the new vocabulary (starting at line
-   *  `new_vocab_offset` and up to `num_new_vocab` entities), or `-1` if entry `i`
-   *  in the new vocabulary is not in the old vocabulary.  The old vocabulary is
-   *  constrained to the first `old_vocab_size` entries if `old_vocab_size` is not the
-   *  default value of -1.
-   *  <p>
-   *  `num_vocab_offset` enables
-   *  use in the partitioned variable case, and should generally be set through
-   *  examining partitioning info.  The format of the files should be a text file,
-   *  with each line containing a single entity within the vocabulary.
-   *  <p>
-   *  For example, with `new_vocab_file` a text file containing each of the following
-   *  elements on a single line: `[f0, f1, f2, f3]`, old_vocab_file = [f1, f0, f3],
-   *  `num_new_vocab = 3, new_vocab_offset = 1`, the returned remapping would be
-   *  `[0, -1, 2]`.
-   *  <p>
-   *  The op also returns a count of how many entries in the new vocabulary
-   *  were present in the old vocabulary, which is used to calculate the number of
-   *  values to initialize in a weight matrix remapping
-   *  <p>
-   *  This functionality can be used to remap both row vocabularies (typically,
-   *  features) and column vocabularies (typically, classes) from TensorFlow
-   *  checkpoints.  Note that the partitioning logic relies on contiguous vocabularies
-   *  corresponding to div-partitioned variables.  Moreover, the underlying remapping
-   *  uses an IndexTable (as opposed to an inexact CuckooTable), so client code should
-   *  use the corresponding index_table_from_file() as the FeatureColumn framework
-   *  does (as opposed to tf.feature_to_id(), which uses a CuckooTable).
-   *
-   * @param newVocabFile Path to the new vocab file.
-   * @param oldVocabFile Path to the old vocab file.
-   * @param newVocabOffset How many entries into the new vocab file to start reading.
-   * @param numNewVocab Number of entries in the new vocab file to remap.
-   * @param options carries optional attributes values
-   * @return a new instance of GenerateVocabRemapping
-   */
-  public GenerateVocabRemapping generateVocabRemapping(Operand<TString> newVocabFile,
-      Operand<TString> oldVocabFile, Long newVocabOffset, Long numNewVocab,
-      GenerateVocabRemapping.Options... options) {
-    return GenerateVocabRemapping.create(scope, newVocabFile, oldVocabFile, newVocabOffset, numNewVocab, options);
-  }
-
-  /**
-   * V2 format specific: merges the metadata files of sharded checkpoints.  The
-   *  <p>
-   *  result is one logical checkpoint, with one physical metadata file and renamed
-   *  data files.
-   *  <p>
-   *  Intended for "grouping" multiple checkpoints in a sharded checkpoint setup.
-   *  <p>
-   *  If delete_old_dirs is true, attempts to delete recursively the dirname of each
-   *  path in the input checkpoint_prefixes.  This is useful when those paths are non
-   *  user-facing temporary locations.
-   *
-   * @param checkpointPrefixes prefixes of V2 checkpoints to merge.
-   * @param destinationPrefix scalar.  The desired final prefix.  Allowed to be the same
-   *  as one of the checkpoint_prefixes.
-   * @param options carries optional attributes values
-   * @return a new instance of MergeV2Checkpoints
-   */
-  public MergeV2Checkpoints mergeV2Checkpoints(Operand<TString> checkpointPrefixes,
-      Operand<TString> destinationPrefix, MergeV2Checkpoints.Options... options) {
-    return MergeV2Checkpoints.create(scope, checkpointPrefixes, destinationPrefix, options);
-  }
-
-  /**
-   * Training via negative sampling.
-   *
-   * @param wIn input word embedding.
-   * @param wOut output word embedding.
-   * @param examples A vector of word ids.
-   * @param labels A vector of word ids.
-   * @param lr
-   * @param vocabCount Count of words in the vocabulary.
-   * @param numNegativeSamples Number of negative samples per example.
-   * @return a new instance of NegTrain
-   */
-  public NegTrain negTrain(Operand<TFloat32> wIn, Operand<TFloat32> wOut, Operand<TInt32> examples,
-      Operand<TInt32> labels, Operand<TFloat32> lr, List<Long> vocabCount,
-      Long numNegativeSamples) {
-    return NegTrain.create(scope, wIn, wOut, examples, labels, lr, vocabCount, numNegativeSamples);
-  }
-
-  /**
-   * An identity op that triggers an error if a gradient is requested.
-   *  <p>
-   *  When executed in a graph, this op outputs its input tensor as-is.
-   *  <p>
-   *  When building ops to compute gradients, the TensorFlow gradient system
-   *  will return an error when trying to lookup the gradient of this op,
-   *  because no gradient must ever be registered for this function.  This
-   *  op exists to prevent subtle bugs from silently returning unimplemented
-   *  gradients in some corner cases.
-   *
-   * @param <T> data type for {@code output()} output
-   * @param input any tensor.
-   * @param options carries optional attributes values
-   * @return a new instance of PreventGradient
-   */
-  public <T extends TType> PreventGradient<T> preventGradient(Operand<T> input,
-      PreventGradient.Options... options) {
-    return PreventGradient.create(scope, input, options);
-  }
-
-  /**
-   * Update '*var' according to the adadelta scheme.
-   *  <p>
-   *  accum = rho() * accum + (1 - rho()) * grad.square();
-   *  update = (update_accum + epsilon).sqrt() * (accum + epsilon()).rsqrt() * grad;
-   *  update_accum = rho() * update_accum + (1 - rho()) * update.square();
-   *  var -= update;
-   *
-   * @param var Should be from a Variable().
-   * @param accum Should be from a Variable().
-   * @param accumUpdate Should be from a Variable().
-   * @param lr Scaling factor. Must be a scalar.
-   * @param rho Decay factor. Must be a scalar.
-   * @param epsilon Constant factor. Must be a scalar.
-   * @param grad The gradient.
-   * @param options carries optional attributes values
-   * @return a new instance of ResourceApplyAdadelta
-   */
-  public <T extends TType> ResourceApplyAdadelta resourceApplyAdadelta(Operand<?> var,
-      Operand<?> accum, Operand<?> accumUpdate, Operand<T> lr, Operand<T> rho, Operand<T> epsilon,
-      Operand<T> grad, ResourceApplyAdadelta.Options... options) {
-    return ResourceApplyAdadelta.create(scope, var, accum, accumUpdate, lr, rho, epsilon, grad, options);
-  }
-
-  /**
-   * Update '*var' according to the proximal adagrad scheme.
-   *
-   * @param var Should be from a Variable().
-   * @param gradientAccumulator Should be from a Variable().
-   * @param gradientSquaredAccumulator Should be from a Variable().
-   * @param grad The gradient.
-   * @param lr Scaling factor. Must be a scalar.
-   * @param l1 L1 regularization. Must be a scalar.
-   * @param l2 L2 regularization. Must be a scalar.
-   * @param globalStep Training step number. Must be a scalar.
-   * @param options carries optional attributes values
-   * @return a new instance of ResourceApplyAdagradDa
-   */
-  public <T extends TType> ResourceApplyAdagradDa resourceApplyAdagradDa(Operand<?> var,
-      Operand<?> gradientAccumulator, Operand<?> gradientSquaredAccumulator, Operand<T> grad,
-      Operand<T> lr, Operand<T> l1, Operand<T> l2, Operand<TInt64> globalStep,
-      ResourceApplyAdagradDa.Options... options) {
-    return ResourceApplyAdagradDa.create(scope, var, gradientAccumulator, gradientSquaredAccumulator, grad, lr, l1, l2, globalStep, options);
-  }
-
-  /**
-   * Update '*var' according to the Adam algorithm.
-   *  <p>
-   *  $$\text{lr}_t := \mathrm{learning_rate} * \sqrt{1 - \beta_2^t} / (1 - \beta_1^t)$$
-   *  $$m_t := \beta_1 * m_{t-1} + (1 - \beta_1) * g$$
-   *  $$v_t := \beta_2 * v_{t-1} + (1 - \beta_2) * g * g$$
-   *  $$\text{variable} := \text{variable} - \text{lr}_t * m_t / (\sqrt{v_t} + \epsilon)$$
-   *
-   * @param var Should be from a Variable().
-   * @param m Should be from a Variable().
-   * @param v Should be from a Variable().
-   * @param beta1Power Must be a scalar.
-   * @param beta2Power Must be a scalar.
-   * @param lr Scaling factor. Must be a scalar.
-   * @param beta1 Momentum factor. Must be a scalar.
-   * @param beta2 Momentum factor. Must be a scalar.
-   * @param epsilon Ridge term. Must be a scalar.
-   * @param grad The gradient.
-   * @param options carries optional attributes values
-   * @return a new instance of ResourceApplyAdam
-   */
-  public <T extends TType> ResourceApplyAdam resourceApplyAdam(Operand<?> var, Operand<?> m,
-      Operand<?> v, Operand<T> beta1Power, Operand<T> beta2Power, Operand<T> lr, Operand<T> beta1,
-      Operand<T> beta2, Operand<T> epsilon, Operand<T> grad, ResourceApplyAdam.Options... options) {
-    return ResourceApplyAdam.create(scope, var, m, v, beta1Power, beta2Power, lr, beta1, beta2, epsilon, grad, options);
-  }
-
-  /**
-   * Update '*var' according to the Adam algorithm.
-   *  <p>
-   *  $$\text{lr}_t := \mathrm{learning_rate} * \sqrt{1 - \beta_2^t} / (1 - \beta_1^t)$$
-   *  $$m_t := \beta_1 * m_{t-1} + (1 - \beta_1) * g$$
-   *  $$v_t := \beta_2 * v_{t-1} + (1 - \beta_2) * g * g$$
-   *  $$\hat{v}_t := max{\hat{v}_{t-1}, v_t}$$
-   *  $$\text{variable} := \text{variable} - \text{lr}_t * m_t / (\sqrt{\hat{v}_t} + \epsilon)$$
-   *
-   * @param var Should be from a Variable().
-   * @param m Should be from a Variable().
-   * @param v Should be from a Variable().
-   * @param vhat Should be from a Variable().
-   * @param beta1Power Must be a scalar.
-   * @param beta2Power Must be a scalar.
-   * @param lr Scaling factor. Must be a scalar.
-   * @param beta1 Momentum factor. Must be a scalar.
-   * @param beta2 Momentum factor. Must be a scalar.
-   * @param epsilon Ridge term. Must be a scalar.
-   * @param grad The gradient.
-   * @param options carries optional attributes values
-   * @return a new instance of ResourceApplyAdamWithAmsgrad
-   */
-  public <T extends TType> ResourceApplyAdamWithAmsgrad resourceApplyAdamWithAmsgrad(Operand<?> var,
-      Operand<?> m, Operand<?> v, Operand<?> vhat, Operand<T> beta1Power, Operand<T> beta2Power,
-      Operand<T> lr, Operand<T> beta1, Operand<T> beta2, Operand<T> epsilon, Operand<T> grad,
-      ResourceApplyAdamWithAmsgrad.Options... options) {
-    return ResourceApplyAdamWithAmsgrad.create(scope, var, m, v, vhat, beta1Power, beta2Power, lr, beta1, beta2, epsilon, grad, options);
-  }
-
-  /**
-   * Update '*var' according to the AddSign update.
-   *  <p>
-   *  m_t <- beta1 * m_{t-1} + (1 - beta1) * g
-   *  update <- (alpha + sign_decay * sign(g) *sign(m)) * g
-   *  variable <- variable - lr_t * update
-   *
-   * @param var Should be from a Variable().
-   * @param m Should be from a Variable().
-   * @param lr Scaling factor. Must be a scalar.
-   * @param alpha Must be a scalar.
-   * @param signDecay Must be a scalar.
-   * @param beta Must be a scalar.
-   * @param grad The gradient.
-   * @param options carries optional attributes values
-   * @return a new instance of ResourceApplyAddSign
-   */
-  public <T extends TType> ResourceApplyAddSign resourceApplyAddSign(Operand<?> var, Operand<?> m,
-      Operand<T> lr, Operand<T> alpha, Operand<T> signDecay, Operand<T> beta, Operand<T> grad,
-      ResourceApplyAddSign.Options... options) {
-    return ResourceApplyAddSign.create(scope, var, m, lr, alpha, signDecay, beta, grad, options);
-  }
-
-  /**
-   * Update '*var' according to the centered RMSProp algorithm.
-   *  <p>
-   *  The centered RMSProp algorithm uses an estimate of the centered second moment
-   *  (i.e., the variance) for normalization, as opposed to regular RMSProp, which
-   *  uses the (uncentered) second moment. This often helps with training, but is
-   *  slightly more expensive in terms of computation and memory.
-   *  <p>
-   *  Note that in dense implementation of this algorithm, mg, ms, and mom will
-   *  update even if the grad is zero, but in this sparse implementation, mg, ms,
-   *  and mom will not update in iterations during which the grad is zero.
-   *  <p>
-   *  mean_square = decay * mean_square + (1-decay) * gradient ** 2
-   *  mean_grad = decay * mean_grad + (1-decay) * gradient
-   *  <p>
-   *  Delta = learning_rate * gradient / sqrt(mean_square + epsilon - mean_grad ** 2)
-   *  <p>
-   *  mg <- rho * mg_{t-1} + (1-rho) * grad
-   *  ms <- rho * ms_{t-1} + (1-rho) * grad * grad
-   *  mom <- momentum * mom_{t-1} + lr * grad / sqrt(ms - mg * mg + epsilon)
-   *  var <- var - mom
-   *
-   * @param var Should be from a Variable().
-   * @param mg Should be from a Variable().
-   * @param ms Should be from a Variable().
-   * @param mom Should be from a Variable().
-   * @param lr Scaling factor. Must be a scalar.
-   * @param rho Decay rate. Must be a scalar.
-   * @param momentum
-   * @param epsilon Ridge term. Must be a scalar.
-   * @param grad The gradient.
-   * @param options carries optional attributes values
-   * @return a new instance of ResourceApplyCenteredRmsProp
-   */
-  public <T extends TType> ResourceApplyCenteredRmsProp resourceApplyCenteredRmsProp(Operand<?> var,
-      Operand<?> mg, Operand<?> ms, Operand<?> mom, Operand<T> lr, Operand<T> rho,
-      Operand<T> momentum, Operand<T> epsilon, Operand<T> grad,
-      ResourceApplyCenteredRmsProp.Options... options) {
-    return ResourceApplyCenteredRmsProp.create(scope, var, mg, ms, mom, lr, rho, momentum, epsilon, grad, options);
-  }
-
-  /**
-   * Update '*var' according to the Ftrl-proximal scheme.
-   *  <p>
-   *  grad_with_shrinkage = grad + 2 * l2_shrinkage * var
-   *  accum_new = accum + grad_with_shrinkage * grad_with_shrinkage
-   *  linear += grad_with_shrinkage +
-   *      (accum_new^(-lr_power) - accum^(-lr_power)) / lr * var
-   *  quadratic = 1.0 / (accum_new^(lr_power) * lr) + 2 * l2
-   *  var = (sign(linear) * l1 - linear) / quadratic if |linear| > l1 else 0.0
-   *  accum = accum_new
-   *
-   * @param var Should be from a Variable().
-   * @param accum Should be from a Variable().
-   * @param linear Should be from a Variable().
-   * @param grad The gradient.
-   * @param lr Scaling factor. Must be a scalar.
-   * @param l1 L1 regularization. Must be a scalar.
-   * @param l2 L2 shrinkage regularization. Must be a scalar.
-   * @param l2Shrinkage
-   * @param lrPower Scaling factor. Must be a scalar.
-   * @param options carries optional attributes values
-   * @return a new instance of ResourceApplyFtrl
-   */
-  public <T extends TType> ResourceApplyFtrl resourceApplyFtrl(Operand<?> var, Operand<?> accum,
-      Operand<?> linear, Operand<T> grad, Operand<T> lr, Operand<T> l1, Operand<T> l2,
-      Operand<T> l2Shrinkage, Operand<T> lrPower, ResourceApplyFtrl.Options... options) {
-    return ResourceApplyFtrl.create(scope, var, accum, linear, grad, lr, l1, l2, l2Shrinkage, lrPower, options);
-  }
-
-  /**
-   * Update '*var' by subtracting 'alpha' * 'delta' from it.
-   *
-   * @param var Should be from a Variable().
-   * @param alpha Scaling factor. Must be a scalar.
-   * @param delta The change.
-   * @param options carries optional attributes values
-   * @return a new instance of ResourceApplyGradientDescent
-   */
-  public <T extends TType> ResourceApplyGradientDescent resourceApplyGradientDescent(Operand<?> var,
-      Operand<T> alpha, Operand<T> delta, ResourceApplyGradientDescent.Options... options) {
-    return ResourceApplyGradientDescent.create(scope, var, alpha, delta, options);
-  }
-
-  /**
-   * Update '*var' according to the momentum scheme.
-   *  <p>
-   *  Set use_nesterov = True if you want to use Nesterov momentum.
-   *  <p>
-   *  accum = accum * momentum - lr * grad
-   *  var += accum
-   *
-   * @param var Should be from a Variable().
-   * @param accum Should be from a Variable().
-   * @param lr Scaling factor. Must be a scalar.
-   * @param grad The gradient.
-   * @param momentum Momentum. Must be a scalar.
-   * @param options carries optional attributes values
-   * @return a new instance of ResourceApplyKerasMomentum
-   */
-  public <T extends TType> ResourceApplyKerasMomentum resourceApplyKerasMomentum(Operand<?> var,
-      Operand<?> accum, Operand<T> lr, Operand<T> grad, Operand<T> momentum,
-      ResourceApplyKerasMomentum.Options... options) {
-    return ResourceApplyKerasMomentum.create(scope, var, accum, lr, grad, momentum, options);
-  }
-
-  /**
-   * Update '*var' according to the momentum scheme.
-   *  <p>
-   *  Set use_nesterov = True if you want to use Nesterov momentum.
-   *  <p>
-   *  accum = accum * momentum + grad
-   *  var -= lr * accum
-   *
-   * @param var Should be from a Variable().
-   * @param accum Should be from a Variable().
-   * @param lr Scaling factor. Must be a scalar.
-   * @param grad The gradient.
-   * @param momentum Momentum. Must be a scalar.
-   * @param options carries optional attributes values
-   * @return a new instance of ResourceApplyMomentum
-   */
-  public <T extends TType> ResourceApplyMomentum resourceApplyMomentum(Operand<?> var,
-      Operand<?> accum, Operand<T> lr, Operand<T> grad, Operand<T> momentum,
-      ResourceApplyMomentum.Options... options) {
-    return ResourceApplyMomentum.create(scope, var, accum, lr, grad, momentum, options);
-  }
-
-  /**
-   * Update '*var' according to the AddSign update.
-   *  <p>
-   *  m_t <- beta1 * m_{t-1} + (1 - beta1) * g
-   *  update <- exp(logbase * sign_decay * sign(g) * sign(m_t)) * g
-   *  variable <- variable - lr_t * update
-   *
-   * @param var Should be from a Variable().
-   * @param m Should be from a Variable().
-   * @param lr Scaling factor. Must be a scalar.
-   * @param logbase Must be a scalar.
-   * @param signDecay Must be a scalar.
-   * @param beta Must be a scalar.
-   * @param grad The gradient.
-   * @param options carries optional attributes values
-   * @return a new instance of ResourceApplyPowerSign
-   */
-  public <T extends TType> ResourceApplyPowerSign resourceApplyPowerSign(Operand<?> var,
-      Operand<?> m, Operand<T> lr, Operand<T> logbase, Operand<T> signDecay, Operand<T> beta,
-      Operand<T> grad, ResourceApplyPowerSign.Options... options) {
-    return ResourceApplyPowerSign.create(scope, var, m, lr, logbase, signDecay, beta, grad, options);
-  }
-
-  /**
-   * Update '*var' and '*accum' according to FOBOS with Adagrad learning rate.
-   *  <p>
-   *  accum += grad <i> grad
-   *  prox_v = var - lr </i> grad <i> (1 / sqrt(accum))
-   *  var = sign(prox_v)/(1+lr</i>l2) <i> max{|prox_v|-lr</i>l1,0}
-   *
-   * @param var Should be from a Variable().
-   * @param accum Should be from a Variable().
-   * @param lr Scaling factor. Must be a scalar.
-   * @param l1 L1 regularization. Must be a scalar.
-   * @param l2 L2 regularization. Must be a scalar.
-   * @param grad The gradient.
-   * @param options carries optional attributes values
-   * @return a new instance of ResourceApplyProximalAdagrad
-   */
-  public <T extends TType> ResourceApplyProximalAdagrad resourceApplyProximalAdagrad(Operand<?> var,
-      Operand<?> accum, Operand<T> lr, Operand<T> l1, Operand<T> l2, Operand<T> grad,
-      ResourceApplyProximalAdagrad.Options... options) {
-    return ResourceApplyProximalAdagrad.create(scope, var, accum, lr, l1, l2, grad, options);
-  }
-
-  /**
-   * Update '*var' as FOBOS algorithm with fixed learning rate.
-   *  <p>
-   *  prox_v = var - alpha <i> delta
-   *  var = sign(prox_v)/(1+alpha</i>l2) <i> max{|prox_v|-alpha</i>l1,0}
-   *
-   * @param var Should be from a Variable().
-   * @param alpha Scaling factor. Must be a scalar.
-   * @param l1 L1 regularization. Must be a scalar.
-   * @param l2 L2 regularization. Must be a scalar.
-   * @param delta The change.
-   * @param options carries optional attributes values
-   * @return a new instance of ResourceApplyProximalGradientDescent
-   */
-  public <T extends TType> ResourceApplyProximalGradientDescent resourceApplyProximalGradientDescent(
-      Operand<?> var, Operand<T> alpha, Operand<T> l1, Operand<T> l2, Operand<T> delta,
-      ResourceApplyProximalGradientDescent.Options... options) {
-    return ResourceApplyProximalGradientDescent.create(scope, var, alpha, l1, l2, delta, options);
-  }
-
-  /**
-   * Update '*var' according to the RMSProp algorithm.
-   *  <p>
-   *  Note that in dense implementation of this algorithm, ms and mom will
-   *  update even if the grad is zero, but in this sparse implementation, ms
-   *  and mom will not update in iterations during which the grad is zero.
-   *  <p>
-   *  mean_square = decay * mean_square + (1-decay) * gradient ** 2
-   *  Delta = learning_rate * gradient / sqrt(mean_square + epsilon)
-   *  <p>
-   *  ms <- rho * ms_{t-1} + (1-rho) * grad * grad
-   *  mom <- momentum * mom_{t-1} + lr * grad / sqrt(ms + epsilon)
-   *  var <- var - mom
-   *
-   * @param var Should be from a Variable().
-   * @param ms Should be from a Variable().
-   * @param mom Should be from a Variable().
-   * @param lr Scaling factor. Must be a scalar.
-   * @param rho Decay rate. Must be a scalar.
-   * @param momentum
-   * @param epsilon Ridge term. Must be a scalar.
-   * @param grad The gradient.
-   * @param options carries optional attributes values
-   * @return a new instance of ResourceApplyRmsProp
-   */
-  public <T extends TType> ResourceApplyRmsProp resourceApplyRmsProp(Operand<?> var, Operand<?> ms,
-      Operand<?> mom, Operand<T> lr, Operand<T> rho, Operand<T> momentum, Operand<T> epsilon,
-      Operand<T> grad, ResourceApplyRmsProp.Options... options) {
-    return ResourceApplyRmsProp.create(scope, var, ms, mom, lr, rho, momentum, epsilon, grad, options);
-  }
-
-  /**
-   * var: Should be from a Variable().
-   *
-   * @param var
-   * @param accum Should be from a Variable().
-   * @param accumUpdate : Should be from a Variable().
-   * @param lr Learning rate. Must be a scalar.
-   * @param rho Decay factor. Must be a scalar.
-   * @param epsilon Constant factor. Must be a scalar.
-   * @param grad The gradient.
-   * @param indices A vector of indices into the first dimension of var and accum.
-   * @param options carries optional attributes values
-   * @return a new instance of ResourceSparseApplyAdadelta
-   */
-  public <T extends TType, U extends TNumber> ResourceSparseApplyAdadelta resourceSparseApplyAdadelta(
-      Operand<?> var, Operand<?> accum, Operand<?> accumUpdate, Operand<T> lr, Operand<T> rho,
-      Operand<T> epsilon, Operand<T> grad, Operand<U> indices,
-      ResourceSparseApplyAdadelta.Options... options) {
-    return ResourceSparseApplyAdadelta.create(scope, var, accum, accumUpdate, lr, rho, epsilon, grad, indices, options);
-  }
-
-  /**
-   * Update relevant entries in '*var' and '*accum' according to the adagrad scheme.
-   *  <p>
-   *  That is for rows we have grad for, we update var and accum as follows:
-   *  accum += grad * grad
-   *  var -= lr * grad * (1 / sqrt(accum))
-   *
-   * @param var Should be from a Variable().
-   * @param accum Should be from a Variable().
-   * @param lr Learning rate. Must be a scalar.
-   * @param grad The gradient.
-   * @param indices A vector of indices into the first dimension of var and accum.
-   * @param options carries optional attributes values
-   * @return a new instance of ResourceSparseApplyAdagrad
-   */
-  public <T extends TType, U extends TNumber> ResourceSparseApplyAdagrad resourceSparseApplyAdagrad(
-      Operand<?> var, Operand<?> accum, Operand<T> lr, Operand<T> grad, Operand<U> indices,
-      ResourceSparseApplyAdagrad.Options... options) {
-    return ResourceSparseApplyAdagrad.create(scope, var, accum, lr, grad, indices, options);
-  }
-
-  /**
-   * Update entries in '*var' and '*accum' according to the proximal adagrad scheme.
-   *
-   * @param var Should be from a Variable().
-   * @param gradientAccumulator Should be from a Variable().
-   * @param gradientSquaredAccumulator Should be from a Variable().
-   * @param grad The gradient.
-   * @param indices A vector of indices into the first dimension of var and accum.
-   * @param lr Learning rate. Must be a scalar.
-   * @param l1 L1 regularization. Must be a scalar.
-   * @param l2 L2 regularization. Must be a scalar.
-   * @param globalStep Training step number. Must be a scalar.
-   * @param options carries optional attributes values
-   * @return a new instance of ResourceSparseApplyAdagradDa
-   */
-  public <T extends TType, U extends TNumber> ResourceSparseApplyAdagradDa resourceSparseApplyAdagradDa(
-      Operand<?> var, Operand<?> gradientAccumulator, Operand<?> gradientSquaredAccumulator,
-      Operand<T> grad, Operand<U> indices, Operand<T> lr, Operand<T> l1, Operand<T> l2,
-      Operand<TInt64> globalStep, ResourceSparseApplyAdagradDa.Options... options) {
-    return ResourceSparseApplyAdagradDa.create(scope, var, gradientAccumulator, gradientSquaredAccumulator, grad, indices, lr, l1, l2, globalStep, options);
-  }
-
-  /**
-   * Update '*var' according to the centered RMSProp algorithm.
-   *  <p>
-   *  The centered RMSProp algorithm uses an estimate of the centered second moment
-   *  (i.e., the variance) for normalization, as opposed to regular RMSProp, which
-   *  uses the (uncentered) second moment. This often helps with training, but is
-   *  slightly more expensive in terms of computation and memory.
-   *  <p>
-   *  Note that in dense implementation of this algorithm, mg, ms, and mom will
-   *  update even if the grad is zero, but in this sparse implementation, mg, ms,
-   *  and mom will not update in iterations during which the grad is zero.
-   *  <p>
-   *  mean_square = decay * mean_square + (1-decay) * gradient ** 2
-   *  mean_grad = decay * mean_grad + (1-decay) * gradient
-   *  Delta = learning_rate * gradient / sqrt(mean_square + epsilon - mean_grad ** 2)
-   *  <p>
-   *  ms <- rho * ms_{t-1} + (1-rho) * grad * grad
-   *  mom <- momentum * mom_{t-1} + lr * grad / sqrt(ms + epsilon)
-   *  var <- var - mom
-   *
-   * @param var Should be from a Variable().
-   * @param mg Should be from a Variable().
-   * @param ms Should be from a Variable().
-   * @param mom Should be from a Variable().
-   * @param lr Scaling factor. Must be a scalar.
-   * @param rho Decay rate. Must be a scalar.
-   * @param momentum
-   * @param epsilon Ridge term. Must be a scalar.
-   * @param grad The gradient.
-   * @param indices A vector of indices into the first dimension of var, ms and mom.
-   * @param options carries optional attributes values
-   * @return a new instance of ResourceSparseApplyCenteredRmsProp
-   */
-  public <T extends TType, U extends TNumber> ResourceSparseApplyCenteredRmsProp resourceSparseApplyCenteredRmsProp(
-      Operand<?> var, Operand<?> mg, Operand<?> ms, Operand<?> mom, Operand<T> lr, Operand<T> rho,
-      Operand<T> momentum, Operand<T> epsilon, Operand<T> grad, Operand<U> indices,
-      ResourceSparseApplyCenteredRmsProp.Options... options) {
-    return ResourceSparseApplyCenteredRmsProp.create(scope, var, mg, ms, mom, lr, rho, momentum, epsilon, grad, indices, options);
-  }
-
-  /**
-   * Update relevant entries in '*var' according to the Ftrl-proximal scheme.
-   *  <p>
-   *  That is for rows we have grad for, we update var, accum and linear as follows:
-   *  grad_with_shrinkage = grad + 2 * l2_shrinkage * var
-   *  accum_new = accum + grad_with_shrinkage * grad_with_shrinkage
-   *  linear += grad_with_shrinkage +
-   *      (accum_new^(-lr_power) - accum^(-lr_power)) / lr * var
-   *  quadratic = 1.0 / (accum_new^(lr_power) * lr) + 2 * l2
-   *  var = (sign(linear) * l1 - linear) / quadratic if |linear| > l1 else 0.0
-   *  accum = accum_new
-   *
-   * @param var Should be from a Variable().
-   * @param accum Should be from a Variable().
-   * @param linear Should be from a Variable().
-   * @param grad The gradient.
-   * @param indices A vector of indices into the first dimension of var and accum.
-   * @param lr Scaling factor. Must be a scalar.
-   * @param l1 L1 regularization. Must be a scalar.
-   * @param l2 L2 shrinkage regularization. Must be a scalar.
-   * @param l2Shrinkage
-   * @param lrPower Scaling factor. Must be a scalar.
-   * @param options carries optional attributes values
-   * @return a new instance of ResourceSparseApplyFtrl
-   */
-  public <T extends TType, U extends TNumber> ResourceSparseApplyFtrl resourceSparseApplyFtrl(
-      Operand<?> var, Operand<?> accum, Operand<?> linear, Operand<T> grad, Operand<U> indices,
-      Operand<T> lr, Operand<T> l1, Operand<T> l2, Operand<T> l2Shrinkage, Operand<T> lrPower,
-      ResourceSparseApplyFtrl.Options... options) {
-    return ResourceSparseApplyFtrl.create(scope, var, accum, linear, grad, indices, lr, l1, l2, l2Shrinkage, lrPower, options);
-  }
-
-  /**
-   * Update relevant entries in '*var' and '*accum' according to the momentum scheme.
-   *  <p>
-   *  Set use_nesterov = True if you want to use Nesterov momentum.
-   *  <p>
-   *  That is for rows we have grad for, we update var and accum as follows:
-   *  <p>
-   *  accum = accum * momentum - lr * grad
-   *  var += accum
-   *
-   * @param var Should be from a Variable().
-   * @param accum Should be from a Variable().
-   * @param lr Learning rate. Must be a scalar.
-   * @param grad The gradient.
-   * @param indices A vector of indices into the first dimension of var and accum.
-   * @param momentum Momentum. Must be a scalar.
-   * @param options carries optional attributes values
-   * @return a new instance of ResourceSparseApplyKerasMomentum
-   */
-  public <T extends TType, U extends TNumber> ResourceSparseApplyKerasMomentum resourceSparseApplyKerasMomentum(
-      Operand<?> var, Operand<?> accum, Operand<T> lr, Operand<T> grad, Operand<U> indices,
-      Operand<T> momentum, ResourceSparseApplyKerasMomentum.Options... options) {
-    return ResourceSparseApplyKerasMomentum.create(scope, var, accum, lr, grad, indices, momentum, options);
-  }
-
-  /**
-   * Update relevant entries in '*var' and '*accum' according to the momentum scheme.
-   *  <p>
-   *  Set use_nesterov = True if you want to use Nesterov momentum.
-   *  <p>
-   *  That is for rows we have grad for, we update var and accum as follows:
-   *  <p>
-   *  accum = accum * momentum + grad
-   *  var -= lr * accum
-   *
-   * @param var Should be from a Variable().
-   * @param accum Should be from a Variable().
-   * @param lr Learning rate. Must be a scalar.
-   * @param grad The gradient.
-   * @param indices A vector of indices into the first dimension of var and accum.
-   * @param momentum Momentum. Must be a scalar.
-   * @param options carries optional attributes values
-   * @return a new instance of ResourceSparseApplyMomentum
-   */
-  public <T extends TType, U extends TNumber> ResourceSparseApplyMomentum resourceSparseApplyMomentum(
-      Operand<?> var, Operand<?> accum, Operand<T> lr, Operand<T> grad, Operand<U> indices,
-      Operand<T> momentum, ResourceSparseApplyMomentum.Options... options) {
-    return ResourceSparseApplyMomentum.create(scope, var, accum, lr, grad, indices, momentum, options);
-  }
-
-  /**
-   * Sparse update entries in '*var' and '*accum' according to FOBOS algorithm.
-   *  <p>
-   *  That is for rows we have grad for, we update var and accum as follows:
-   *  accum += grad <i> grad
-   *  prox_v = var
-   *  prox_v -= lr </i> grad <i> (1 / sqrt(accum))
-   *  var = sign(prox_v)/(1+lr</i>l2) <i> max{|prox_v|-lr</i>l1,0}
-   *
-   * @param var Should be from a Variable().
-   * @param accum Should be from a Variable().
-   * @param lr Learning rate. Must be a scalar.
-   * @param l1 L1 regularization. Must be a scalar.
-   * @param l2 L2 regularization. Must be a scalar.
-   * @param grad The gradient.
-   * @param indices A vector of indices into the first dimension of var and accum.
-   * @param options carries optional attributes values
-   * @return a new instance of ResourceSparseApplyProximalAdagrad
-   */
-  public <T extends TType, U extends TNumber> ResourceSparseApplyProximalAdagrad resourceSparseApplyProximalAdagrad(
-      Operand<?> var, Operand<?> accum, Operand<T> lr, Operand<T> l1, Operand<T> l2,
-      Operand<T> grad, Operand<U> indices, ResourceSparseApplyProximalAdagrad.Options... options) {
-    return ResourceSparseApplyProximalAdagrad.create(scope, var, accum, lr, l1, l2, grad, indices, options);
-  }
-
-  /**
-   * Sparse update '*var' as FOBOS algorithm with fixed learning rate.
-   *  <p>
-   *  That is for rows we have grad for, we update var as follows:
-   *  prox_v = var - alpha <i> grad
-   *  var = sign(prox_v)/(1+alpha</i>l2) <i> max{|prox_v|-alpha</i>l1,0}
-   *
-   * @param var Should be from a Variable().
-   * @param alpha Scaling factor. Must be a scalar.
-   * @param l1 L1 regularization. Must be a scalar.
-   * @param l2 L2 regularization. Must be a scalar.
-   * @param grad The gradient.
-   * @param indices A vector of indices into the first dimension of var and accum.
-   * @param options carries optional attributes values
-   * @return a new instance of ResourceSparseApplyProximalGradientDescent
-   */
-  public <T extends TType, U extends TNumber> ResourceSparseApplyProximalGradientDescent resourceSparseApplyProximalGradientDescent(
-      Operand<?> var, Operand<T> alpha, Operand<T> l1, Operand<T> l2, Operand<T> grad,
-      Operand<U> indices, ResourceSparseApplyProximalGradientDescent.Options... options) {
-    return ResourceSparseApplyProximalGradientDescent.create(scope, var, alpha, l1, l2, grad, indices, options);
-  }
-
-  /**
-   * Update '*var' according to the RMSProp algorithm.
-   *  <p>
-   *  Note that in dense implementation of this algorithm, ms and mom will
-   *  update even if the grad is zero, but in this sparse implementation, ms
-   *  and mom will not update in iterations during which the grad is zero.
-   *  <p>
-   *  mean_square = decay * mean_square + (1-decay) * gradient ** 2
-   *  Delta = learning_rate * gradient / sqrt(mean_square + epsilon)
-   *  <p>
-   *  ms <- rho * ms_{t-1} + (1-rho) * grad * grad
-   *  mom <- momentum * mom_{t-1} + lr * grad / sqrt(ms + epsilon)
-   *  var <- var - mom
-   *
-   * @param var Should be from a Variable().
-   * @param ms Should be from a Variable().
-   * @param mom Should be from a Variable().
-   * @param lr Scaling factor. Must be a scalar.
-   * @param rho Decay rate. Must be a scalar.
-   * @param momentum
-   * @param epsilon Ridge term. Must be a scalar.
-   * @param grad The gradient.
-   * @param indices A vector of indices into the first dimension of var, ms and mom.
-   * @param options carries optional attributes values
-   * @return a new instance of ResourceSparseApplyRmsProp
-   */
-  public <T extends TType, U extends TNumber> ResourceSparseApplyRmsProp resourceSparseApplyRmsProp(
-      Operand<?> var, Operand<?> ms, Operand<?> mom, Operand<T> lr, Operand<T> rho,
-      Operand<T> momentum, Operand<T> epsilon, Operand<T> grad, Operand<U> indices,
-      ResourceSparseApplyRmsProp.Options... options) {
-    return ResourceSparseApplyRmsProp.create(scope, var, ms, mom, lr, rho, momentum, epsilon, grad, indices, options);
-  }
-
-  /**
-   * Restores tensors from a V2 checkpoint.
-   *  <p>
-   *  For backward compatibility with the V1 format, this Op currently allows
-   *  restoring from a V1 checkpoint as well:
-   *    - This Op first attempts to find the V2 index file pointed to by "prefix", and
-   *      if found proceed to read it as a V2 checkpoint;
-   *    - Otherwise the V1 read path is invoked.
-   *  Relying on this behavior is not recommended, as the ability to fall back to read
-   *  V1 might be deprecated and eventually removed.
-   *  <p>
-   *  By default, restores the named tensors in full.  If the caller wishes to restore
-   *  specific slices of stored tensors, "shape_and_slices" should be non-empty
-   *  strings and correspondingly well-formed.
-   *  <p>
-   *  Callers must ensure all the named tensors are indeed stored in the checkpoint.
-   *
-   * @param prefix Must have a single element.  The prefix of a V2 checkpoint.
-   * @param tensorNames shape {N}.  The names of the tensors to be restored.
-   * @param shapeAndSlices shape {N}.  The slice specs of the tensors to be restored.
-   *  Empty strings indicate that they are non-partitioned tensors.
-   * @param dtypes shape {N}.  The list of expected dtype for the tensors.  Must match
-   *  those stored in the checkpoint.
-   * @return a new instance of Restore
-   */
-  public Restore restore(Operand<TString> prefix, Operand<TString> tensorNames,
-      Operand<TString> shapeAndSlices, List<DataType<?>> dtypes) {
-    return Restore.create(scope, prefix, tensorNames, shapeAndSlices, dtypes);
-  }
-
-  /**
-   * Restores a tensor from checkpoint files.
-   *  <p>
-   *  This is like `Restore` except that restored tensor can be listed as filling
-   *  only a slice of a larger tensor.  `shape_and_slice` specifies the shape of the
-   *  larger tensor and the slice that the restored tensor covers.
-   *  <p>
-   *  The `shape_and_slice` input has the same format as the
-   *  elements of the `shapes_and_slices` input of the `SaveSlices` op.
-   *
-   * @param <T> data type for {@code tensor()} output
-   * @param filePattern Must have a single element. The pattern of the files from
-   *  which we read the tensor.
-   * @param tensorName Must have a single element. The name of the tensor to be
-   *  restored.
-   * @param shapeAndSlice Scalar. The shapes and slice specifications to use when
-   *  restoring a tensors.
-   * @param dt The type of the tensor to be restored.
-   * @param options carries optional attributes values
-   * @return a new instance of RestoreSlice
-   */
-  public <T extends TType> RestoreSlice<T> restoreSlice(Operand<TString> filePattern,
-      Operand<TString> tensorName, Operand<TString> shapeAndSlice, DataType<T> dt,
-      RestoreSlice.Options... options) {
-    return RestoreSlice.create(scope, filePattern, tensorName, shapeAndSlice, dt, options);
-  }
-
-  /**
-   * Saves tensors in V2 checkpoint format.
-   *  <p>
-   *  By default, saves the named tensors in full.  If the caller wishes to save
-   *  specific slices of full tensors, "shape_and_slices" should be non-empty strings
-   *  and correspondingly well-formed.
-   *
-   * @param prefix Must have a single element. The prefix of the V2 checkpoint to which we
-   *  write the tensors.
-   * @param tensorNames shape {N}. The names of the tensors to be saved.
-   * @param shapeAndSlices shape {N}.  The slice specs of the tensors to be saved.
-   *  Empty strings indicate that they are non-partitioned tensors.
-   * @param tensors `N` tensors to save.
-   * @return a new instance of Save
-   */
-  public Save save(Operand<TString> prefix, Operand<TString> tensorNames,
-      Operand<TString> shapeAndSlices, Iterable<Operand<?>> tensors) {
-    return Save.create(scope, prefix, tensorNames, shapeAndSlices, tensors);
-  }
-
-  /**
-   * Saves input tensors slices to disk.
-   *  <p>
-   *  This is like `Save` except that tensors can be listed in the saved file as being
-   *  a slice of a larger tensor.  `shapes_and_slices` specifies the shape of the
-   *  larger tensor and the slice that this tensor covers. `shapes_and_slices` must
-   *  have as many elements as `tensor_names`.
-   *  <p>
-   *  Elements of the `shapes_and_slices` input must either be:
-   *  <ul>
-   *  <li>
-   *  The empty string, in which case the corresponding tensor is
-   *     saved normally.
-   *  </li>
-   *  <li>
-   *  A string of the form `dim0 dim1 ... dimN-1 slice-spec` where the
-   *     `dimI` are the dimensions of the larger tensor and `slice-spec`
-   *     specifies what part is covered by the tensor to save.
-   *  </li>
-   *  </ul>
-   *  `slice-spec` itself is a `:`-separated list: `slice0:slice1:...:sliceN-1`
-   *  where each `sliceI` is either:
-   *  <ul>
-   *  <li>
-   *  The string `-` meaning that the slice covers all indices of this dimension
-   *  </li>
-   *  <li>
-   *  `start,length` where `start` and `length` are integers.  In that
-   *     case the slice covers `length` indices starting at `start`.
-   *  </li>
-   *  </ul>
-   *  See also `Save`.
-   *
-   * @param filename Must have a single element. The name of the file to which we write the
-   *  tensor.
-   * @param tensorNames Shape `[N]`. The names of the tensors to be saved.
-   * @param shapesAndSlices Shape `[N]`.  The shapes and slice specifications to use when
-   *  saving the tensors.
-   * @param data `N` tensors to save.
-   * @return a new instance of SaveSlices
-   */
-  public SaveSlices saveSlices(Operand<TString> filename, Operand<TString> tensorNames,
-      Operand<TString> shapesAndSlices, Iterable<Operand<?>> data) {
-    return SaveSlices.create(scope, filename, tensorNames, shapesAndSlices, data);
-  }
-
-  /**
-   * Computes fingerprints of the input strings.
-   *
-   * @param input vector of strings to compute fingerprints on.
-   * @return a new instance of SdcaFprint
-   */
-  public SdcaFprint sdcaFprint(Operand<TString> input) {
-    return SdcaFprint.create(scope, input);
-  }
-
-  /**
-   * Applies L1 regularization shrink step on the parameters.
-   *
-   * @param weights a list of vectors where each value is the weight associated with a
-   *  feature group.
-   * @param l1 Symmetric l1 regularization strength.
-   * @param l2 Symmetric l2 regularization strength. Should be a positive float.
-   * @return a new instance of SdcaShrinkL1
-   */
-  public SdcaShrinkL1 sdcaShrinkL1(Iterable<Operand<TFloat32>> weights, Float l1, Float l2) {
-    return SdcaShrinkL1.create(scope, weights, l1, l2);
-  }
-
-  /**
-   * var: Should be from a Variable().
-   *
-   * @param <T> data type for {@code out()} output
-   * @param var
-   * @param accum Should be from a Variable().
-   * @param accumUpdate : Should be from a Variable().
-   * @param lr Learning rate. Must be a scalar.
-   * @param rho Decay factor. Must be a scalar.
-   * @param epsilon Constant factor. Must be a scalar.
-   * @param grad The gradient.
-   * @param indices A vector of indices into the first dimension of var and accum.
-   * @param options carries optional attributes values
-   * @return a new instance of SparseApplyAdadelta
-   */
-  public <T extends TType, U extends TNumber> SparseApplyAdadelta<T> sparseApplyAdadelta(
-      Operand<T> var, Operand<T> accum, Operand<T> accumUpdate, Operand<T> lr, Operand<T> rho,
-      Operand<T> epsilon, Operand<T> grad, Operand<U> indices,
-      SparseApplyAdadelta.Options... options) {
-    return SparseApplyAdadelta.create(scope, var, accum, accumUpdate, lr, rho, epsilon, grad, indices, options);
-  }
-
-  /**
-   * Update entries in '*var' and '*accum' according to the proximal adagrad scheme.
-   *
-   * @param <T> data type for {@code out()} output
-   * @param var Should be from a Variable().
-   * @param gradientAccumulator Should be from a Variable().
-   * @param gradientSquaredAccumulator Should be from a Variable().
-   * @param grad The gradient.
-   * @param indices A vector of indices into the first dimension of var and accum.
-   * @param lr Learning rate. Must be a scalar.
-   * @param l1 L1 regularization. Must be a scalar.
-   * @param l2 L2 regularization. Must be a scalar.
-   * @param globalStep Training step number. Must be a scalar.
-   * @param options carries optional attributes values
-   * @return a new instance of SparseApplyAdagradDa
-   */
-  public <T extends TType, U extends TNumber> SparseApplyAdagradDa<T> sparseApplyAdagradDa(
-      Operand<T> var, Operand<T> gradientAccumulator, Operand<T> gradientSquaredAccumulator,
-      Operand<T> grad, Operand<U> indices, Operand<T> lr, Operand<T> l1, Operand<T> l2,
-      Operand<TInt64> globalStep, SparseApplyAdagradDa.Options... options) {
-    return SparseApplyAdagradDa.create(scope, var, gradientAccumulator, gradientSquaredAccumulator, grad, indices, lr, l1, l2, globalStep, options);
-  }
-
-  /**
-   * Update '*var' according to the centered RMSProp algorithm.
-   *  <p>
-   *  The centered RMSProp algorithm uses an estimate of the centered second moment
-   *  (i.e., the variance) for normalization, as opposed to regular RMSProp, which
-   *  uses the (uncentered) second moment. This often helps with training, but is
-   *  slightly more expensive in terms of computation and memory.
-   *  <p>
-   *  Note that in dense implementation of this algorithm, mg, ms, and mom will
-   *  update even if the grad is zero, but in this sparse implementation, mg, ms,
-   *  and mom will not update in iterations during which the grad is zero.
-   *  <p>
-   *  mean_square = decay * mean_square + (1-decay) * gradient ** 2
-   *  mean_grad = decay * mean_grad + (1-decay) * gradient
-   *  Delta = learning_rate * gradient / sqrt(mean_square + epsilon - mean_grad ** 2)
-   *  <p>
-   *  $$ms <- rho * ms_{t-1} + (1-rho) * grad * grad$$
-   *  $$mom <- momentum * mom_{t-1} + lr * grad / sqrt(ms + epsilon)$$
-   *  $$var <- var - mom$$
-   *
-   * @param <T> data type for {@code out()} output
-   * @param var Should be from a Variable().
-   * @param mg Should be from a Variable().
-   * @param ms Should be from a Variable().
-   * @param mom Should be from a Variable().
-   * @param lr Scaling factor. Must be a scalar.
-   * @param rho Decay rate. Must be a scalar.
-   * @param momentum
-   * @param epsilon Ridge term. Must be a scalar.
-   * @param grad The gradient.
-   * @param indices A vector of indices into the first dimension of var, ms and mom.
-   * @param options carries optional attributes values
-   * @return a new instance of SparseApplyCenteredRmsProp
-   */
-  public <T extends TType, U extends TNumber> SparseApplyCenteredRmsProp<T> sparseApplyCenteredRmsProp(
-      Operand<T> var, Operand<T> mg, Operand<T> ms, Operand<T> mom, Operand<T> lr, Operand<T> rho,
-      Operand<T> momentum, Operand<T> epsilon, Operand<T> grad, Operand<U> indices,
-      SparseApplyCenteredRmsProp.Options... options) {
-    return SparseApplyCenteredRmsProp.create(scope, var, mg, ms, mom, lr, rho, momentum, epsilon, grad, indices, options);
-  }
-
-  /**
-   * Update relevant entries in '*var' according to the Ftrl-proximal scheme.
-   *  <p>
-   *  That is for rows we have grad for, we update var, accum and linear as follows:
-   *  grad_with_shrinkage = grad + 2 * l2_shrinkage * var
-   *  accum_new = accum + grad * grad
-   *  linear += grad_with_shrinkage -
-   *      (accum_new^(-lr_power) - accum^(-lr_power)) / lr * var
-   *  quadratic = 1.0 / (accum_new^(lr_power) * lr) + 2 * l2
-   *  var = (sign(linear) * l1 - linear) / quadratic if |linear| > l1 else 0.0
-   *  accum = accum_new
-   *
-   * @param <T> data type for {@code out()} output
-   * @param var Should be from a Variable().
-   * @param accum Should be from a Variable().
-   * @param linear Should be from a Variable().
-   * @param grad The gradient.
-   * @param indices A vector of indices into the first dimension of var and accum.
-   * @param lr Scaling factor. Must be a scalar.
-   * @param l1 L1 regularization. Must be a scalar.
-   * @param l2 L2 shrinkage regularization. Must be a scalar.
-   * @param l2Shrinkage
-   * @param lrPower Scaling factor. Must be a scalar.
-   * @param options carries optional attributes values
-   * @return a new instance of SparseApplyFtrl
-   */
-  public <T extends TType, U extends TNumber> SparseApplyFtrl<T> sparseApplyFtrl(Operand<T> var,
-      Operand<T> accum, Operand<T> linear, Operand<T> grad, Operand<U> indices, Operand<T> lr,
-      Operand<T> l1, Operand<T> l2, Operand<T> l2Shrinkage, Operand<T> lrPower,
-      SparseApplyFtrl.Options... options) {
-    return SparseApplyFtrl.create(scope, var, accum, linear, grad, indices, lr, l1, l2, l2Shrinkage, lrPower, options);
-  }
-
-  /**
-   * Update relevant entries in '*var' and '*accum' according to the momentum scheme.
-   *  <p>
-   *  Set use_nesterov = True if you want to use Nesterov momentum.
-   *  <p>
-   *  That is for rows we have grad for, we update var and accum as follows:
-   *  <p>
-   *  $$accum = accum * momentum + grad$$
-   *  $$var -= lr * accum$$
-   *
-   * @param <T> data type for {@code out()} output
-   * @param var Should be from a Variable().
-   * @param accum Should be from a Variable().
-   * @param lr Learning rate. Must be a scalar.
-   * @param grad The gradient.
-   * @param indices A vector of indices into the first dimension of var and accum.
-   * @param momentum Momentum. Must be a scalar.
-   * @param options carries optional attributes values
-   * @return a new instance of SparseApplyMomentum
-   */
-  public <T extends TType, U extends TNumber> SparseApplyMomentum<T> sparseApplyMomentum(
-      Operand<T> var, Operand<T> accum, Operand<T> lr, Operand<T> grad, Operand<U> indices,
-      Operand<T> momentum, SparseApplyMomentum.Options... options) {
-    return SparseApplyMomentum.create(scope, var, accum, lr, grad, indices, momentum, options);
-  }
-
-  /**
-   * Sparse update entries in '*var' and '*accum' according to FOBOS algorithm.
-   *  <p>
-   *  That is for rows we have grad for, we update var and accum as follows:
-   *  $$accum += grad <i> grad$$
-   *  $$prox_v = var$$
-   *  $$prox_v -= lr </i> grad <i> (1 / sqrt(accum))$$
-   *  $$var = sign(prox_v)/(1+lr</i>l2) <i> max{|prox_v|-lr</i>l1,0}$$
-   *
-   * @param <T> data type for {@code out()} output
-   * @param var Should be from a Variable().
-   * @param accum Should be from a Variable().
-   * @param lr Learning rate. Must be a scalar.
-   * @param l1 L1 regularization. Must be a scalar.
-   * @param l2 L2 regularization. Must be a scalar.
-   * @param grad The gradient.
-   * @param indices A vector of indices into the first dimension of var and accum.
-   * @param options carries optional attributes values
-   * @return a new instance of SparseApplyProximalAdagrad
-   */
-  public <T extends TType, U extends TNumber> SparseApplyProximalAdagrad<T> sparseApplyProximalAdagrad(
-      Operand<T> var, Operand<T> accum, Operand<T> lr, Operand<T> l1, Operand<T> l2,
-      Operand<T> grad, Operand<U> indices, SparseApplyProximalAdagrad.Options... options) {
-    return SparseApplyProximalAdagrad.create(scope, var, accum, lr, l1, l2, grad, indices, options);
-  }
-
-  /**
-   * Sparse update '*var' as FOBOS algorithm with fixed learning rate.
-   *  <p>
-   *  That is for rows we have grad for, we update var as follows:
-   *  $$prox_v = var - alpha <i> grad$$
-   *  $$var = sign(prox_v)/(1+alpha</i>l2) <i> max{|prox_v|-alpha</i>l1,0}$$
-   *
-   * @param <T> data type for {@code out()} output
-   * @param var Should be from a Variable().
-   * @param alpha Scaling factor. Must be a scalar.
-   * @param l1 L1 regularization. Must be a scalar.
-   * @param l2 L2 regularization. Must be a scalar.
-   * @param grad The gradient.
-   * @param indices A vector of indices into the first dimension of var and accum.
-   * @param options carries optional attributes values
-   * @return a new instance of SparseApplyProximalGradientDescent
-   */
-  public <T extends TType, U extends TNumber> SparseApplyProximalGradientDescent<T> sparseApplyProximalGradientDescent(
-      Operand<T> var, Operand<T> alpha, Operand<T> l1, Operand<T> l2, Operand<T> grad,
-      Operand<U> indices, SparseApplyProximalGradientDescent.Options... options) {
-    return SparseApplyProximalGradientDescent.create(scope, var, alpha, l1, l2, grad, indices, options);
-  }
-
-  /**
-   * Update '*var' according to the RMSProp algorithm.
-   *  <p>
-   *  Note that in dense implementation of this algorithm, ms and mom will
-   *  update even if the grad is zero, but in this sparse implementation, ms
-   *  and mom will not update in iterations during which the grad is zero.
-   *  <p>
-   *  mean_square = decay * mean_square + (1-decay) * gradient ** 2
-   *  Delta = learning_rate * gradient / sqrt(mean_square + epsilon)
-   *  <p>
-   *  $$ms <- rho * ms_{t-1} + (1-rho) * grad * grad$$
-   *  $$mom <- momentum * mom_{t-1} + lr * grad / sqrt(ms + epsilon)$$
-   *  $$var <- var - mom$$
-   *
-   * @param <T> data type for {@code out()} output
-   * @param var Should be from a Variable().
-   * @param ms Should be from a Variable().
-   * @param mom Should be from a Variable().
-   * @param lr Scaling factor. Must be a scalar.
-   * @param rho Decay rate. Must be a scalar.
-   * @param momentum
-   * @param epsilon Ridge term. Must be a scalar.
-   * @param grad The gradient.
-   * @param indices A vector of indices into the first dimension of var, ms and mom.
-   * @param options carries optional attributes values
-   * @return a new instance of SparseApplyRmsProp
-   */
-  public <T extends TType, U extends TNumber> SparseApplyRmsProp<T> sparseApplyRmsProp(
-      Operand<T> var, Operand<T> ms, Operand<T> mom, Operand<T> lr, Operand<T> rho,
-      Operand<T> momentum, Operand<T> epsilon, Operand<T> grad, Operand<U> indices,
-      SparseApplyRmsProp.Options... options) {
-    return SparseApplyRmsProp.create(scope, var, ms, mom, lr, rho, momentum, epsilon, grad, indices, options);
-  }
-
-  /**
-   * Returns the gradient of `Tile`.
-   *  <p>
-   *  Since `Tile` takes an input and repeats the input `multiples` times
-   *  along each dimension, `train.TileGrad` takes in `multiples` and aggregates
-   *  each repeated tile of `input` into `output`.
-   *
-   * @param <T> data type for {@code output()} output
-   * @param input
-   * @param multiples
-   * @return a new instance of TileGrad
-   */
-  public <T extends TType> TileGrad<T> tileGrad(Operand<T> input, Operand<TInt32> multiples) {
-    return TileGrad.create(scope, input, multiples);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/XlaOps.java b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/XlaOps.java
deleted file mode 100644
index 535972d4883..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/XlaOps.java
+++ /dev/null
@@ -1,382 +0,0 @@
-// Copyright 2020 The TensorFlow Authors. All Rights Reserved.
-//
-// Licensed under the Apache License, Version 2.0 (the "License");
-// you may not use this file except in compliance with the License.
-// You may obtain a copy of the License at
-//
-//     http://www.apache.org/licenses/LICENSE-2.0
-//
-// Unless required by applicable law or agreed to in writing, software
-// distributed under the License is distributed on an "AS IS" BASIS,
-// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-// See the License for the specific language governing permissions and
-// limitations under the License.
-// ==============================================================================
-//
-// This class has been generated, DO NOT EDIT!
-//
-package org.tensorflow.op;
-
-import org.tensorflow.DataType;
-import org.tensorflow.Operand;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.op.xla.BroadcastHelper;
-import org.tensorflow.op.xla.ClusterOutput;
-import org.tensorflow.op.xla.Conv;
-import org.tensorflow.op.xla.Dequantize;
-import org.tensorflow.op.xla.Dot;
-import org.tensorflow.op.xla.DynamicSlice;
-import org.tensorflow.op.xla.DynamicUpdateSlice;
-import org.tensorflow.op.xla.Einsum;
-import org.tensorflow.op.xla.Gather;
-import org.tensorflow.op.xla.KeyValueSort;
-import org.tensorflow.op.xla.Pad;
-import org.tensorflow.op.xla.Recv;
-import org.tensorflow.op.xla.ReplicaId;
-import org.tensorflow.op.xla.SelfAdjointEig;
-import org.tensorflow.op.xla.Send;
-import org.tensorflow.op.xla.Sharding;
-import org.tensorflow.op.xla.Sort;
-import org.tensorflow.op.xla.Svd;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * An API for building {@code xla} operations as {@link Op Op}s
- *
- * @see {@link Ops}
- */
-public final class XlaOps {
-  private final Scope scope;
-
-  XlaOps(Scope scope) {
-    this.scope = scope;
-  }
-
-  /**
-   * Helper operator for performing XLA-style broadcasts
-   *  <p>
-   *  Broadcasts `lhs` and `rhs` to the same rank, by adding size 1 dimensions to
-   *  whichever of `lhs` and `rhs` has the lower rank, using XLA's broadcasting rules
-   *  for binary operators.
-   *
-   * @param <T> data type for {@code lhsOutput()} output
-   * @param lhs the LHS input tensor
-   * @param rhs the RHS input tensor
-   * @param broadcastDims an XLA-style broadcast dimension specification
-   * @return a new instance of BroadcastHelper
-   */
-  public <T extends TType, U extends TNumber> BroadcastHelper<T> broadcastHelper(Operand<T> lhs,
-      Operand<T> rhs, Operand<U> broadcastDims) {
-    return BroadcastHelper.create(scope, lhs, rhs, broadcastDims);
-  }
-
-  /**
-   * Operator that connects the output of an XLA computation to other consumer graph nodes.
-   *
-   * @param <T> data type for {@code outputs()} output
-   * @param input
-   * @return a new instance of ClusterOutput
-   */
-  public <T extends TType> ClusterOutput<T> clusterOutput(Operand<T> input) {
-    return ClusterOutput.create(scope, input);
-  }
-
-  /**
-   * Wraps the XLA ConvGeneralDilated operator, documented at
-   *  <p>
-   *   https://www.tensorflow.org/performance/xla/operation_semantics#conv_convolution
-   *  .
-   *
-   * @param <T> data type for {@code output()} output
-   * @param lhs the input tensor
-   * @param rhs the kernel tensor
-   * @param windowStrides the inter-window strides
-   * @param padding the padding to apply at the start and end of each input dimensions
-   * @param lhsDilation dilation to apply between input elements
-   * @param rhsDilation dilation to apply between kernel elements
-   * @param featureGroupCount number of feature groups for grouped convolution.
-   * @param dimensionNumbers a serialized xla::ConvolutionDimensionNumbers proto.
-   * @param precisionConfig a serialized xla::PrecisionConfig proto.
-   * @return a new instance of Conv
-   */
-  public <T extends TType, U extends TNumber> Conv<T> conv(Operand<T> lhs, Operand<T> rhs,
-      Operand<U> windowStrides, Operand<U> padding, Operand<U> lhsDilation, Operand<U> rhsDilation,
-      Operand<U> featureGroupCount, String dimensionNumbers, String precisionConfig) {
-    return Conv.create(scope, lhs, rhs, windowStrides, padding, lhsDilation, rhsDilation, featureGroupCount, dimensionNumbers, precisionConfig);
-  }
-
-  /**
-   * Takes the packed uint32 input and unpacks the input to uint8 to do
-   *  <p>
-   *  Dequantization on device.
-   *
-   * @param input Input tensors whose types is uint32, shape is [d0, ..., dn].
-   * @param minRange The minimum scalar value possibly produced for the input.
-   * @param maxRange The maximum scalar value possibly produced for the input.
-   * @param mode String to determine the dequantize mode in {"MIN_COMBINED", "MIN_FIRST", "SCALED"}.
-   * @param transposeOutput Boolean to determine if output is transposed. transpose_output
-   *  is faster when input is large and rank of input is higher than 1.
-   * @return a new instance of Dequantize
-   */
-  public Dequantize dequantize(Operand<?> input, Float minRange, Float maxRange, String mode,
-      Boolean transposeOutput) {
-    return Dequantize.create(scope, input, minRange, maxRange, mode, transposeOutput);
-  }
-
-  /**
-   * Wraps the XLA DotGeneral operator, documented at
-   *  <p>
-   *   https://www.tensorflow.org/performance/xla/operation_semantics#dotgeneral
-   *  .
-   *
-   * @param <T> data type for {@code output()} output
-   * @param lhs the LHS tensor
-   * @param rhs the RHS tensor
-   * @param dimensionNumbers a serialized xla::DotDimensionNumbers proto.
-   * @param precisionConfig a serialized xla::PrecisionConfig proto.
-   * @return a new instance of Dot
-   */
-  public <T extends TType> Dot<T> dot(Operand<T> lhs, Operand<T> rhs, String dimensionNumbers,
-      String precisionConfig) {
-    return Dot.create(scope, lhs, rhs, dimensionNumbers, precisionConfig);
-  }
-
-  /**
-   * Wraps the XLA DynamicSlice operator, documented at
-   *  <p>
-   *   https://www.tensorflow.org/performance/xla/operation_semantics#dynamicslice
-   *  .
-   *  <p>
-   *  DynamicSlice extracts a sub-array from the input array at dynamic
-   *  start_indices. The size of the slice in each dimension is passed in
-   *  size_indices, which specify the end point of exclusive slice intervals in each
-   *  dimension -- [start, start + size). The shape of start_indices must have rank 1,
-   *  with dimension size equal to the rank of operand.
-   *
-   * @param <T> data type for {@code output()} output
-   * @param input A `Tensor` of type T.
-   * @param startIndices List of N integers containing the slice size for each
-   *  dimension. Each value must be strictly greater than zero, and start + size
-   *  must be less than or equal to the size of the dimension to avoid
-   *  implementation defined behavior.
-   * @param sizeIndices
-   * @return a new instance of DynamicSlice
-   */
-  public <T extends TType, U extends TNumber> DynamicSlice<T> dynamicSlice(Operand<T> input,
-      Operand<U> startIndices, Operand<U> sizeIndices) {
-    return DynamicSlice.create(scope, input, startIndices, sizeIndices);
-  }
-
-  /**
-   * Wraps the XLA DynamicUpdateSlice operator, documented at
-   *  <p>
-   *   https://www.tensorflow.org/performance/xla/operation_semantics#dynamicupdateslice
-   *  .
-   *  <p>
-   *  XlaDynamicUpdateSlice generates a result which is the value of the `input`
-   *  operand, with a slice update overwritten at `indices`. The shape of `update`
-   *  determines the shape of the sub-array of the result which is updated. The shape
-   *  of indices must be rank == 1, with dimension size equal to the rank of `input`.
-   *  <p>
-   *  Handling of out-of-bounds slice indices is implementation-defined.
-   *
-   * @param <T> data type for {@code output()} output
-   * @param input A `Tensor` of type T.
-   * @param update A `Tensor` of type T. Same rank as `input`.
-   * @param indices A vector of indices into `input`. Must have length equal to the rank of
-   *  `input`.
-   * @return a new instance of DynamicUpdateSlice
-   */
-  public <T extends TType, U extends TNumber> DynamicUpdateSlice<T> dynamicUpdateSlice(
-      Operand<T> input, Operand<T> update, Operand<U> indices) {
-    return DynamicUpdateSlice.create(scope, input, update, indices);
-  }
-
-  /**
-   * An op which supports basic einsum op with 2 inputs and 1 output.
-   *  <p>
-   *  This op has better TPU performance since it doesn't have explicitly reshape and
-   *  transpose operations as tf.einsum does.
-   *
-   * @param <T> data type for {@code product()} output
-   * @param a
-   * @param b
-   * @param equation
-   * @return a new instance of Einsum
-   */
-  public <T extends TType> Einsum<T> einsum(Operand<T> a, Operand<T> b, String equation) {
-    return Einsum.create(scope, a, b, equation);
-  }
-
-  /**
-   * Wraps the XLA Gather operator documented at
-   *  <p>
-   *    https://www.tensorflow.org/xla/operation_semantics#gather
-   *
-   * @param <T> data type for {@code output()} output
-   * @param operand The array we're gathering from.
-   * @param startIndices Array containing the starting indices of the slices we gather.
-   * @param sliceSizes slice_sizes[i] is the bounds for the slice on dimension i.
-   * @param dimensionNumbers A serialized xla::GatherDimensionNumbers proto.
-   * @param indicesAreSorted Boolean indicating if the indices are sorted.
-   * @return a new instance of Gather
-   */
-  public <T extends TType, U extends TNumber> Gather<T> gather(Operand<T> operand,
-      Operand<U> startIndices, Operand<U> sliceSizes, String dimensionNumbers,
-      Boolean indicesAreSorted) {
-    return Gather.create(scope, operand, startIndices, sliceSizes, dimensionNumbers, indicesAreSorted);
-  }
-
-  /**
-   * Wraps the XLA Sort operator, documented at
-   *  <p>
-   *   https://www.tensorflow.org/performance/xla/operation_semantics#sort
-   *  .
-   *  <p>
-   *  Sorts a tensor. Currently only sorts in ascending order are supported.
-   *
-   * @param <T> data type for {@code sortedKeys()} output
-   * @param <U> data type for {@code sortedValues()} output
-   * @param keys A `Tensor` of type K.
-   * @param values A `Tensor` of type V.
-   * @return a new instance of KeyValueSort
-   */
-  public <T extends TNumber, U extends TType> KeyValueSort<T, U> keyValueSort(Operand<T> keys,
-      Operand<U> values) {
-    return KeyValueSort.create(scope, keys, values);
-  }
-
-  /**
-   * Wraps the XLA Pad operator, documented at
-   *  <p>
-   *   https://www.tensorflow.org/performance/xla/operation_semantics#pad
-   *  .
-   *
-   * @param <T> data type for {@code output()} output
-   * @param input A `Tensor` of type T.
-   * @param paddingValue A scalar `Tensor` of type T.
-   * @param paddingLow the padding to apply at the start of each input dimensions
-   * @param paddingHigh the padding to apply at the end of each input dimension.
-   * @param paddingInterior the padding to apply between each input element.
-   * @return a new instance of Pad
-   */
-  public <T extends TType, U extends TNumber> Pad<T> pad(Operand<T> input, Operand<T> paddingValue,
-      Operand<U> paddingLow, Operand<U> paddingHigh, Operand<U> paddingInterior) {
-    return Pad.create(scope, input, paddingValue, paddingLow, paddingHigh, paddingInterior);
-  }
-
-  /**
-   * Receives the named tensor from another XLA computation. Wraps the XLA Recv
-   *  <p>
-   *  operator documented at
-   *   https://www.tensorflow.org/performance/xla/operation_semantics#recv .
-   *
-   * @param <T> data type for {@code tensor()} output
-   * @param dtype The type of the tensor.
-   * @param tensorName A string key that identifies the channel.
-   * @param shape The shape of the tensor.
-   * @return a new instance of Recv
-   */
-  public <T extends TType> Recv<T> recv(DataType<T> dtype, String tensorName, Shape shape) {
-    return Recv.create(scope, dtype, tensorName, shape);
-  }
-
-  /**
-   * Replica ID.
-   *
-   * @return a new instance of ReplicaId
-   */
-  public ReplicaId replicaId() {
-    return ReplicaId.create(scope);
-  }
-
-  /**
-   * Computes the eigen decomposition of a batch of self-adjoint matrices
-   *  <p>
-   *  (Note: Only real inputs are supported).
-   *  <p>
-   *  Computes the eigenvalues and eigenvectors of the innermost N-by-N matrices in
-   *  tensor such that tensor[...,:,:] * v[..., :,i] = e[..., i] * v[...,:,i], for
-   *  i=0...N-1.
-   *
-   * @param <T> data type for {@code w()} output
-   * @param a the input tensor.
-   * @param lower a boolean specifies whether the calculation is done with the lower
-   *  triangular part or the upper triangular part.
-   * @param maxIter maximum number of sweep update, i.e., the whole lower triangular
-   *  part or upper triangular part based on parameter lower. Heuristically, it has
-   *  been argued that approximately logN sweeps are needed in practice (Ref: Golub &
-   *  van Loan "Matrix Computation").
-   * @param epsilon the tolerance ratio.
-   * @return a new instance of SelfAdjointEig
-   */
-  public <T extends TType> SelfAdjointEig<T> selfAdjointEig(Operand<T> a, Boolean lower,
-      Long maxIter, Float epsilon) {
-    return SelfAdjointEig.create(scope, a, lower, maxIter, epsilon);
-  }
-
-  /**
-   * Sends the named tensor to another XLA computation. Wraps the XLA Send operator
-   *  <p>
-   *  documented at
-   *   https://www.tensorflow.org/performance/xla/operation_semantics#send .
-   *
-   * @param tensor The tensor to send.
-   * @param tensorName A string key that identifies the channel.
-   * @return a new instance of Send
-   */
-  public <T extends TType> Send send(Operand<T> tensor, String tensorName) {
-    return Send.create(scope, tensor, tensorName);
-  }
-
-  /**
-   * An op which shards the input based on the given sharding attribute.
-   *
-   * @param <T> data type for {@code output()} output
-   * @param input
-   * @return a new instance of Sharding
-   */
-  public <T extends TType> Sharding<T> sharding(Operand<T> input) {
-    return Sharding.create(scope, input);
-  }
-
-  /**
-   * Wraps the XLA Sort operator, documented at
-   *  <p>
-   *   https://www.tensorflow.org/performance/xla/operation_semantics#sort
-   *  .
-   *  <p>
-   *  Sorts a tensor. Currently only sorts in ascending order are supported.
-   *
-   * @param <T> data type for {@code output()} output
-   * @param input A `Tensor` of type T.
-   * @return a new instance of Sort
-   */
-  public <T extends TType> Sort<T> sort(Operand<T> input) {
-    return Sort.create(scope, input);
-  }
-
-  /**
-   * Computes the eigen decomposition of a batch of self-adjoint matrices
-   *  <p>
-   *  (Note: Only real inputs are supported).
-   *  <p>
-   *  Computes the eigenvalues and eigenvectors of the innermost M-by-N matrices in
-   *  tensor such that tensor[...,:,:] = u[..., :, :] * Diag(s[..., :]) * Transpose(v[...,:,:]).
-   *
-   * @param <T> data type for {@code s()} output
-   * @param a the input tensor.
-   * @param maxIter maximum number of sweep update, i.e., the whole lower triangular
-   *  part or upper triangular part based on parameter lower. Heuristically, it has
-   *  been argued that approximately log(min (M, N)) sweeps are needed in practice
-   *  (Ref: Golub & van Loan "Matrix Computation").
-   * @param epsilon the tolerance ratio.
-   * @param precisionConfig a serialized xla::PrecisionConfig proto.
-   * @return a new instance of Svd
-   */
-  public <T extends TType> Svd<T> svd(Operand<T> a, Long maxIter, Float epsilon,
-      String precisionConfig) {
-    return Svd.create(scope, a, maxIter, epsilon, precisionConfig);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/collective/BroadcastRecv.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/collective/BroadcastRecv.java
index ba1adab34f3..54e5416b33b 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/collective/BroadcastRecv.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/collective/BroadcastRecv.java
@@ -17,7 +17,6 @@
 
 package org.tensorflow.op.collective;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -77,7 +76,7 @@ private Options() {
    * @return a new instance of BroadcastRecv
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> BroadcastRecv<T> create(Scope scope, DataType<T> T, Long groupSize, Long groupKey, Long instanceKey, Shape shape, Options... options) {
+  public static <T extends TType> BroadcastRecv<T> create(Scope scope, Class<T> T, Long groupSize, Long groupKey, Long instanceKey, Shape shape, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("CollectiveBcastRecv", scope.makeOpName("BroadcastRecv"));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("T", T);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Barrier.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Barrier.java
index 4fcb8ef8f21..83e2d6a2261 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Barrier.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Barrier.java
@@ -18,7 +18,6 @@
 package org.tensorflow.op.core;
 
 import java.util.List;
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -105,10 +104,10 @@ private Options() {
    * @return a new instance of Barrier
    */
   @Endpoint(describeByClass = true)
-  public static Barrier create(Scope scope, List<DataType<?>> componentTypes, Options... options) {
+  public static Barrier create(Scope scope, List<Class<?>> componentTypes, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("Barrier", scope.makeOpName("Barrier"));
     opBuilder = scope.applyControlDependencies(opBuilder);
-    DataType[] componentTypesArray = new DataType[componentTypes.size()];
+    Class[] componentTypesArray = new Class[componentTypes.size()];
     for (int i = 0; i < componentTypesArray.length; ++i) {
       componentTypesArray[i] = componentTypes.get(i);
     }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BarrierTakeMany.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BarrierTakeMany.java
index c5a22ae6027..ed685428d90 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BarrierTakeMany.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BarrierTakeMany.java
@@ -19,7 +19,6 @@
 
 import java.util.Arrays;
 import java.util.List;
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -98,12 +97,12 @@ private Options() {
    * @return a new instance of BarrierTakeMany
    */
   @Endpoint(describeByClass = true)
-  public static BarrierTakeMany create(Scope scope, Operand<TString> handle, Operand<TInt32> numElements, List<DataType<?>> componentTypes, Options... options) {
+  public static BarrierTakeMany create(Scope scope, Operand<TString> handle, Operand<TInt32> numElements, List<Class<?>> componentTypes, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("BarrierTakeMany", scope.makeOpName("BarrierTakeMany"));
     opBuilder.addInput(handle.asOutput(scope));
     opBuilder.addInput(numElements.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
-    DataType[] componentTypesArray = new DataType[componentTypes.size()];
+    Class[] componentTypesArray = new Class[componentTypes.size()];
     for (int i = 0; i < componentTypesArray.length; ++i) {
       componentTypesArray[i] = componentTypes.get(i);
     }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Bitcast.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Bitcast.java
index 37391484e43..9dc11a8a61b 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Bitcast.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Bitcast.java
@@ -17,7 +17,6 @@
 
 package org.tensorflow.op.core;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -96,7 +95,7 @@ public final class Bitcast<U extends TType> extends RawOp implements Operand<U>
    * @return a new instance of Bitcast
    */
   @Endpoint(describeByClass = true)
-  public static <U extends TType, T extends TType> Bitcast<U> create(Scope scope, Operand<T> input, DataType<U> type) {
+  public static <U extends TType, T extends TType> Bitcast<U> create(Scope scope, Operand<T> input, Class<U> type) {
     OperationBuilder opBuilder = scope.env().opBuilder("Bitcast", scope.makeOpName("Bitcast"));
     opBuilder.addInput(input.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/DecodeProto.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/DecodeProto.java
index 5d01ad3ac00..7e8318e3b1c 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/DecodeProto.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/DecodeProto.java
@@ -19,7 +19,6 @@
 
 import java.util.Arrays;
 import java.util.List;
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -135,7 +134,7 @@ private Options() {
    * @return a new instance of DecodeProto
    */
   @Endpoint(describeByClass = true)
-  public static DecodeProto create(Scope scope, Operand<TString> bytes, String messageType, List<String> fieldNames, List<DataType<?>> outputTypes, Options... options) {
+  public static DecodeProto create(Scope scope, Operand<TString> bytes, String messageType, List<String> fieldNames, List<Class<?>> outputTypes, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("DecodeProtoV2", scope.makeOpName("DecodeProto"));
     opBuilder.addInput(bytes.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
@@ -145,7 +144,7 @@ public static DecodeProto create(Scope scope, Operand<TString> bytes, String mes
       fieldNamesArray[i] = fieldNames.get(i);
     }
     opBuilder.setAttr("field_names", fieldNamesArray);
-    DataType[] outputTypesArray = new DataType[outputTypes.size()];
+    Class[] outputTypesArray = new Class[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
       outputTypesArray[i] = outputTypes.get(i);
     }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/DeviceIndex.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/DeviceIndex.java
index f033d3fcc9d..bb76e3d5992 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/DeviceIndex.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/DeviceIndex.java
@@ -64,7 +64,7 @@ public Output<TInt32> index() {
   }
   
   @Override
-  public Output<TInt32> asOutput() {
+  public Output<TInt32> asOutput(Scope scope) {
     return index;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/DummySeedGenerator.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/DummySeedGenerator.java
deleted file mode 100644
index b9cf2c36d09..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/DummySeedGenerator.java
+++ /dev/null
@@ -1,66 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- */
-public final class DummySeedGenerator extends RawOp implements Operand<TType> {
-  
-  /**
-   * Factory method to create a class wrapping a new DummySeedGenerator operation.
-   * 
-   * @param scope current scope
-   * @return a new instance of DummySeedGenerator
-   */
-  @Endpoint(describeByClass = true)
-  public static DummySeedGenerator create(Scope scope) {
-    OperationBuilder opBuilder = scope.env().opBuilder("DummySeedGenerator", scope.makeOpName("DummySeedGenerator"));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new DummySeedGenerator(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<?> handle() {
-    return handle;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
-    return (Output<TType>) handle;
-  }
-  
-  private Output<?> handle;
-  
-  private DummySeedGenerator(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    handle = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Empty.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Empty.java
index b78e44b0709..dd2e60b3c6e 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Empty.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Empty.java
@@ -17,7 +17,6 @@
 
 package org.tensorflow.op.core;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -68,7 +67,7 @@ private Options() {
    * @return a new instance of Empty
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> Empty<T> create(Scope scope, Operand<TInt32> shape, DataType<T> dtype, Options... options) {
+  public static <T extends TType> Empty<T> create(Scope scope, Operand<TInt32> shape, Class<T> dtype, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("Empty", scope.makeOpName("Empty"));
     opBuilder.addInput(shape.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/EmptyTensorList.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/EmptyTensorList.java
index 68a46956ad6..d8f6813e542 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/EmptyTensorList.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/EmptyTensorList.java
@@ -17,7 +17,6 @@
 
 package org.tensorflow.op.core;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -53,7 +52,7 @@ public final class EmptyTensorList extends RawOp implements Operand<TType> {
    * @return a new instance of EmptyTensorList
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber, U extends TType> EmptyTensorList create(Scope scope, Operand<T> elementShape, Operand<TInt32> maxNumElements, DataType<U> elementDtype) {
+  public static <T extends TNumber, U extends TType> EmptyTensorList create(Scope scope, Operand<T> elementShape, Operand<TInt32> maxNumElements, Class<U> elementDtype) {
     OperationBuilder opBuilder = scope.env().opBuilder("EmptyTensorList", scope.makeOpName("EmptyTensorList"));
     opBuilder.addInput(elementShape.asOutput(scope));
     opBuilder.addInput(maxNumElements.asOutput(scope));
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/GetSessionTensor.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/GetSessionTensor.java
index a9abc7821c8..067a19cfb49 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/GetSessionTensor.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/GetSessionTensor.java
@@ -17,7 +17,6 @@
 
 package org.tensorflow.op.core;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -46,7 +45,7 @@ public final class GetSessionTensor<T extends TType> extends RawOp implements Op
    * @return a new instance of GetSessionTensor
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> GetSessionTensor<T> create(Scope scope, Operand<TString> handle, DataType<T> dtype) {
+  public static <T extends TType> GetSessionTensor<T> create(Scope scope, Operand<TString> handle, Class<T> dtype) {
     OperationBuilder opBuilder = scope.env().opBuilder("GetSessionTensor", scope.makeOpName("GetSessionTensor"));
     opBuilder.addInput(handle.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/HashTable.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/HashTable.java
index e191d086497..49205672c44 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/HashTable.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/HashTable.java
@@ -17,7 +17,6 @@
 
 package org.tensorflow.op.core;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -88,7 +87,7 @@ private Options() {
    * @return a new instance of HashTable
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TType> HashTable create(Scope scope, DataType<T> keyDtype, DataType<U> valueDtype, Options... options) {
+  public static <T extends TType, U extends TType> HashTable create(Scope scope, Class<T> keyDtype, Class<U> valueDtype, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("HashTableV2", scope.makeOpName("HashTable"));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("key_dtype", keyDtype);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/HistogramFixedWidth.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/HistogramFixedWidth.java
index e6f46c71057..a0a2a5d873d 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/HistogramFixedWidth.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/HistogramFixedWidth.java
@@ -17,7 +17,6 @@
 
 package org.tensorflow.op.core;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -66,7 +65,7 @@ public final class HistogramFixedWidth<U extends TNumber> extends RawOp implemen
    * @return a new instance of HistogramFixedWidth
    */
   @Endpoint(describeByClass = true)
-  public static <U extends TNumber, T extends TNumber> HistogramFixedWidth<U> create(Scope scope, Operand<T> values, Operand<T> valueRange, Operand<TInt32> nbins, DataType<U> dtype) {
+  public static <U extends TNumber, T extends TNumber> HistogramFixedWidth<U> create(Scope scope, Operand<T> values, Operand<T> valueRange, Operand<TInt32> nbins, Class<U> dtype) {
     OperationBuilder opBuilder = scope.env().opBuilder("HistogramFixedWidth", scope.makeOpName("HistogramFixedWidth"));
     opBuilder.addInput(values.asOutput(scope));
     opBuilder.addInput(valueRange.asOutput(scope));
@@ -89,7 +88,7 @@ public static <U extends TNumber, T extends TNumber> HistogramFixedWidth<U> crea
    */
   @Endpoint(describeByClass = true)
   public static <T extends TNumber> HistogramFixedWidth<TInt32> create(Scope scope, Operand<T> values, Operand<T> valueRange, Operand<TInt32> nbins) {
-    return create(scope, values, valueRange, nbins, TInt32.DTYPE);
+    return create(scope, values, valueRange, nbins, TInt32.class);
   }
   
   /**
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ImmutableConst.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ImmutableConst.java
index a6976162ff0..0038baef8d5 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ImmutableConst.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ImmutableConst.java
@@ -17,7 +17,6 @@
 
 package org.tensorflow.op.core;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -50,7 +49,7 @@ public final class ImmutableConst<T extends TType> extends RawOp implements Oper
    * @return a new instance of ImmutableConst
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> ImmutableConst<T> create(Scope scope, DataType<T> dtype, Shape shape, String memoryRegionName) {
+  public static <T extends TType> ImmutableConst<T> create(Scope scope, Class<T> dtype, Shape shape, String memoryRegionName) {
     OperationBuilder opBuilder = scope.env().opBuilder("ImmutableConst", scope.makeOpName("ImmutableConst"));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("dtype", dtype);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LinSpace.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LinSpace.java
index d2061ea6553..095d5bf6b9c 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LinSpace.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LinSpace.java
@@ -42,7 +42,6 @@
  * 
  * @param <T> data type for {@code output()} output
  */
-@Operator
 public final class LinSpace<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
@@ -76,6 +75,9 @@ public Output<T> asOutput(Scope scope) {
     return output;
   }
   
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "LinSpace";
+  
   private Output<T> output;
   
   private LinSpace(Operation operation) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LookupTableExport.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LookupTableExport.java
index 6d59e0404f8..598c93b5322 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LookupTableExport.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LookupTableExport.java
@@ -17,7 +17,6 @@
 
 package org.tensorflow.op.core;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -47,7 +46,7 @@ public final class LookupTableExport<T extends TType, U extends TType> extends R
    * @return a new instance of LookupTableExport
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TType> LookupTableExport<T, U> create(Scope scope, Operand<?> tableHandle, DataType<T> Tkeys, DataType<U> Tvalues) {
+  public static <T extends TType, U extends TType> LookupTableExport<T, U> create(Scope scope, Operand<?> tableHandle, Class<T> Tkeys, Class<U> Tvalues) {
     OperationBuilder opBuilder = scope.env().opBuilder("LookupTableExportV2", scope.makeOpName("LookupTableExport"));
     opBuilder.addInput(tableHandle.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LowerBound.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LowerBound.java
index 8d6341857be..64acd0bba13 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LowerBound.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LowerBound.java
@@ -17,7 +17,6 @@
 
 package org.tensorflow.op.core;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -66,7 +65,7 @@ public final class LowerBound<U extends TNumber> extends RawOp implements Operan
    * @return a new instance of LowerBound
    */
   @Endpoint(describeByClass = true)
-  public static <U extends TNumber, T extends TType> LowerBound<U> create(Scope scope, Operand<T> sortedInputs, Operand<T> values, DataType<U> outType) {
+  public static <U extends TNumber, T extends TType> LowerBound<U> create(Scope scope, Operand<T> sortedInputs, Operand<T> values, Class<U> outType) {
     OperationBuilder opBuilder = scope.env().opBuilder("LowerBound", scope.makeOpName("LowerBound"));
     opBuilder.addInput(sortedInputs.asOutput(scope));
     opBuilder.addInput(values.asOutput(scope));
@@ -86,7 +85,7 @@ public static <U extends TNumber, T extends TType> LowerBound<U> create(Scope sc
    */
   @Endpoint(describeByClass = true)
   public static <T extends TType> LowerBound<TInt32> create(Scope scope, Operand<T> sortedInputs, Operand<T> values) {
-    return create(scope, sortedInputs, values, TInt32.DTYPE);
+    return create(scope, sortedInputs, values, TInt32.class);
   }
   
   /**
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MapClear.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MapClear.java
index bad1e90554f..bd24fcfef5f 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MapClear.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MapClear.java
@@ -18,7 +18,6 @@
 package org.tensorflow.op.core;
 
 import java.util.List;
-import org.tensorflow.DataType;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.op.RawOp;
@@ -87,10 +86,10 @@ private Options() {
    * @return a new instance of MapClear
    */
   @Endpoint(describeByClass = true)
-  public static MapClear create(Scope scope, List<DataType<?>> dtypes, Options... options) {
+  public static MapClear create(Scope scope, List<Class<?>> dtypes, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("MapClear", scope.makeOpName("MapClear"));
     opBuilder = scope.applyControlDependencies(opBuilder);
-    DataType[] dtypesArray = new DataType[dtypes.size()];
+    Class[] dtypesArray = new Class[dtypes.size()];
     for (int i = 0; i < dtypesArray.length; ++i) {
       dtypesArray[i] = dtypes.get(i);
     }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MapIncompleteSize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MapIncompleteSize.java
index 18cc4e059b1..173abfcc3b0 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MapIncompleteSize.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MapIncompleteSize.java
@@ -18,7 +18,6 @@
 package org.tensorflow.op.core;
 
 import java.util.List;
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -90,10 +89,10 @@ private Options() {
    * @return a new instance of MapIncompleteSize
    */
   @Endpoint(describeByClass = true)
-  public static MapIncompleteSize create(Scope scope, List<DataType<?>> dtypes, Options... options) {
+  public static MapIncompleteSize create(Scope scope, List<Class<?>> dtypes, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("MapIncompleteSize", scope.makeOpName("MapIncompleteSize"));
     opBuilder = scope.applyControlDependencies(opBuilder);
-    DataType[] dtypesArray = new DataType[dtypes.size()];
+    Class[] dtypesArray = new Class[dtypes.size()];
     for (int i = 0; i < dtypesArray.length; ++i) {
       dtypesArray[i] = dtypes.get(i);
     }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MapPeek.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MapPeek.java
index 7408c06644e..f2df24b15b8 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MapPeek.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MapPeek.java
@@ -20,7 +20,6 @@
 import java.util.Arrays;
 import java.util.Iterator;
 import java.util.List;
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -99,12 +98,12 @@ private Options() {
    * @return a new instance of MapPeek
    */
   @Endpoint(describeByClass = true)
-  public static MapPeek create(Scope scope, Operand<TInt64> key, Operand<TInt32> indices, List<DataType<?>> dtypes, Options... options) {
+  public static MapPeek create(Scope scope, Operand<TInt64> key, Operand<TInt32> indices, List<Class<?>> dtypes, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("MapPeek", scope.makeOpName("MapPeek"));
     opBuilder.addInput(key.asOutput(scope));
     opBuilder.addInput(indices.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
-    DataType[] dtypesArray = new DataType[dtypes.size()];
+    Class[] dtypesArray = new Class[dtypes.size()];
     for (int i = 0; i < dtypesArray.length; ++i) {
       dtypesArray[i] = dtypes.get(i);
     }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MapSize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MapSize.java
index b6e467af2d3..58d13795e86 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MapSize.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MapSize.java
@@ -18,7 +18,6 @@
 package org.tensorflow.op.core;
 
 import java.util.List;
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -90,10 +89,10 @@ private Options() {
    * @return a new instance of MapSize
    */
   @Endpoint(describeByClass = true)
-  public static MapSize create(Scope scope, List<DataType<?>> dtypes, Options... options) {
+  public static MapSize create(Scope scope, List<Class<?>> dtypes, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("MapSize", scope.makeOpName("MapSize"));
     opBuilder = scope.applyControlDependencies(opBuilder);
-    DataType[] dtypesArray = new DataType[dtypes.size()];
+    Class[] dtypesArray = new Class[dtypes.size()];
     for (int i = 0; i < dtypesArray.length; ++i) {
       dtypesArray[i] = dtypes.get(i);
     }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MapStage.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MapStage.java
index ebbdd445b62..8a9741a417f 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MapStage.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MapStage.java
@@ -18,7 +18,6 @@
 package org.tensorflow.op.core;
 
 import java.util.List;
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -97,13 +96,13 @@ private Options() {
    * @return a new instance of MapStage
    */
   @Endpoint(describeByClass = true)
-  public static MapStage create(Scope scope, Operand<TInt64> key, Operand<TInt32> indices, Iterable<Operand<?>> values, List<DataType<?>> dtypes, Options... options) {
+  public static MapStage create(Scope scope, Operand<TInt64> key, Operand<TInt32> indices, Iterable<Operand<?>> values, List<Class<?>> dtypes, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("MapStage", scope.makeOpName("MapStage"));
     opBuilder.addInput(key.asOutput(scope));
     opBuilder.addInput(indices.asOutput(scope));
     opBuilder.addInputList(Operands.asOutputs(scope, values));
     opBuilder = scope.applyControlDependencies(opBuilder);
-    DataType[] dtypesArray = new DataType[dtypes.size()];
+    Class[] dtypesArray = new Class[dtypes.size()];
     for (int i = 0; i < dtypesArray.length; ++i) {
       dtypesArray[i] = dtypes.get(i);
     }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MapUnstage.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MapUnstage.java
index 0d8e9d854d7..07442b96b01 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MapUnstage.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MapUnstage.java
@@ -20,7 +20,6 @@
 import java.util.Arrays;
 import java.util.Iterator;
 import java.util.List;
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -99,12 +98,12 @@ private Options() {
    * @return a new instance of MapUnstage
    */
   @Endpoint(describeByClass = true)
-  public static MapUnstage create(Scope scope, Operand<TInt64> key, Operand<TInt32> indices, List<DataType<?>> dtypes, Options... options) {
+  public static MapUnstage create(Scope scope, Operand<TInt64> key, Operand<TInt32> indices, List<Class<?>> dtypes, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("MapUnstage", scope.makeOpName("MapUnstage"));
     opBuilder.addInput(key.asOutput(scope));
     opBuilder.addInput(indices.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
-    DataType[] dtypesArray = new DataType[dtypes.size()];
+    Class[] dtypesArray = new Class[dtypes.size()];
     for (int i = 0; i < dtypesArray.length; ++i) {
       dtypesArray[i] = dtypes.get(i);
     }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MapUnstageNoKey.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MapUnstageNoKey.java
index 58e49e3616f..6d4a767f0f9 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MapUnstageNoKey.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MapUnstageNoKey.java
@@ -19,7 +19,6 @@
 
 import java.util.Arrays;
 import java.util.List;
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -96,11 +95,11 @@ private Options() {
    * @return a new instance of MapUnstageNoKey
    */
   @Endpoint(describeByClass = true)
-  public static MapUnstageNoKey create(Scope scope, Operand<TInt32> indices, List<DataType<?>> dtypes, Options... options) {
+  public static MapUnstageNoKey create(Scope scope, Operand<TInt32> indices, List<Class<?>> dtypes, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("MapUnstageNoKey", scope.makeOpName("MapUnstageNoKey"));
     opBuilder.addInput(indices.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
-    DataType[] dtypesArray = new DataType[dtypes.size()];
+    Class[] dtypesArray = new Class[dtypes.size()];
     for (int i = 0; i < dtypesArray.length; ++i) {
       dtypesArray[i] = dtypes.get(i);
     }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MlirPassthroughOp.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MlirPassthroughOp.java
index 0dcd4a25d20..270d184c011 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MlirPassthroughOp.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MlirPassthroughOp.java
@@ -20,7 +20,6 @@
 import java.util.Arrays;
 import java.util.Iterator;
 import java.util.List;
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -78,12 +77,12 @@ public final class MlirPassthroughOp extends RawOp implements Iterable<Operand<T
    * @return a new instance of MlirPassthroughOp
    */
   @Endpoint(describeByClass = true)
-  public static MlirPassthroughOp create(Scope scope, Iterable<Operand<?>> inputs, String mlirModule, List<DataType<?>> Toutputs) {
+  public static MlirPassthroughOp create(Scope scope, Iterable<Operand<?>> inputs, String mlirModule, List<Class<?>> Toutputs) {
     OperationBuilder opBuilder = scope.env().opBuilder("MlirPassthroughOp", scope.makeOpName("MlirPassthroughOp"));
     opBuilder.addInputList(Operands.asOutputs(scope, inputs));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("mlir_module", mlirModule);
-    DataType[] ToutputsArray = new DataType[Toutputs.size()];
+    Class[] ToutputsArray = new Class[Toutputs.size()];
     for (int i = 0; i < ToutputsArray.length; ++i) {
       ToutputsArray[i] = Toutputs.get(i);
     }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MutableDenseHashTable.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MutableDenseHashTable.java
index b9fa39dc7c1..3a8988e8172 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MutableDenseHashTable.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MutableDenseHashTable.java
@@ -17,7 +17,6 @@
 
 package org.tensorflow.op.core;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -122,7 +121,7 @@ private Options() {
    * @return a new instance of MutableDenseHashTable
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TType> MutableDenseHashTable create(Scope scope, Operand<T> emptyKey, Operand<T> deletedKey, DataType<U> valueDtype, Options... options) {
+  public static <T extends TType, U extends TType> MutableDenseHashTable create(Scope scope, Operand<T> emptyKey, Operand<T> deletedKey, Class<U> valueDtype, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("MutableDenseHashTableV2", scope.makeOpName("MutableDenseHashTable"));
     opBuilder.addInput(emptyKey.asOutput(scope));
     opBuilder.addInput(deletedKey.asOutput(scope));
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MutableHashTable.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MutableHashTable.java
index b108d397980..fe0449157d8 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MutableHashTable.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MutableHashTable.java
@@ -17,7 +17,6 @@
 
 package org.tensorflow.op.core;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -88,7 +87,7 @@ private Options() {
    * @return a new instance of MutableHashTable
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TType> MutableHashTable create(Scope scope, DataType<T> keyDtype, DataType<U> valueDtype, Options... options) {
+  public static <T extends TType, U extends TType> MutableHashTable create(Scope scope, Class<T> keyDtype, Class<U> valueDtype, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("MutableHashTableV2", scope.makeOpName("MutableHashTable"));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("key_dtype", keyDtype);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MutableHashTableOfTensors.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MutableHashTableOfTensors.java
index 5ad501f5175..c61f37dacb0 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MutableHashTableOfTensors.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MutableHashTableOfTensors.java
@@ -17,7 +17,6 @@
 
 package org.tensorflow.op.core;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -97,7 +96,7 @@ private Options() {
    * @return a new instance of MutableHashTableOfTensors
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TType> MutableHashTableOfTensors create(Scope scope, DataType<T> keyDtype, DataType<U> valueDtype, Options... options) {
+  public static <T extends TType, U extends TType> MutableHashTableOfTensors create(Scope scope, Class<T> keyDtype, Class<U> valueDtype, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("MutableHashTableOfTensorsV2", scope.makeOpName("MutableHashTableOfTensors"));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("key_dtype", keyDtype);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OrderedMapClear.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OrderedMapClear.java
index 05a1b7ab984..e44d88210ca 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OrderedMapClear.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OrderedMapClear.java
@@ -18,7 +18,6 @@
 package org.tensorflow.op.core;
 
 import java.util.List;
-import org.tensorflow.DataType;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.op.RawOp;
@@ -87,10 +86,10 @@ private Options() {
    * @return a new instance of OrderedMapClear
    */
   @Endpoint(describeByClass = true)
-  public static OrderedMapClear create(Scope scope, List<DataType<?>> dtypes, Options... options) {
+  public static OrderedMapClear create(Scope scope, List<Class<?>> dtypes, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("OrderedMapClear", scope.makeOpName("OrderedMapClear"));
     opBuilder = scope.applyControlDependencies(opBuilder);
-    DataType[] dtypesArray = new DataType[dtypes.size()];
+    Class[] dtypesArray = new Class[dtypes.size()];
     for (int i = 0; i < dtypesArray.length; ++i) {
       dtypesArray[i] = dtypes.get(i);
     }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OrderedMapIncompleteSize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OrderedMapIncompleteSize.java
index b92e8cb4273..68fe5a7a610 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OrderedMapIncompleteSize.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OrderedMapIncompleteSize.java
@@ -18,7 +18,6 @@
 package org.tensorflow.op.core;
 
 import java.util.List;
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -90,10 +89,10 @@ private Options() {
    * @return a new instance of OrderedMapIncompleteSize
    */
   @Endpoint(describeByClass = true)
-  public static OrderedMapIncompleteSize create(Scope scope, List<DataType<?>> dtypes, Options... options) {
+  public static OrderedMapIncompleteSize create(Scope scope, List<Class<?>> dtypes, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("OrderedMapIncompleteSize", scope.makeOpName("OrderedMapIncompleteSize"));
     opBuilder = scope.applyControlDependencies(opBuilder);
-    DataType[] dtypesArray = new DataType[dtypes.size()];
+    Class[] dtypesArray = new Class[dtypes.size()];
     for (int i = 0; i < dtypesArray.length; ++i) {
       dtypesArray[i] = dtypes.get(i);
     }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OrderedMapPeek.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OrderedMapPeek.java
index 45ae4630076..f330950304d 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OrderedMapPeek.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OrderedMapPeek.java
@@ -20,7 +20,6 @@
 import java.util.Arrays;
 import java.util.Iterator;
 import java.util.List;
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -100,12 +99,12 @@ private Options() {
    * @return a new instance of OrderedMapPeek
    */
   @Endpoint(describeByClass = true)
-  public static OrderedMapPeek create(Scope scope, Operand<TInt64> key, Operand<TInt32> indices, List<DataType<?>> dtypes, Options... options) {
+  public static OrderedMapPeek create(Scope scope, Operand<TInt64> key, Operand<TInt32> indices, List<Class<?>> dtypes, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("OrderedMapPeek", scope.makeOpName("OrderedMapPeek"));
     opBuilder.addInput(key.asOutput(scope));
     opBuilder.addInput(indices.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
-    DataType[] dtypesArray = new DataType[dtypes.size()];
+    Class[] dtypesArray = new Class[dtypes.size()];
     for (int i = 0; i < dtypesArray.length; ++i) {
       dtypesArray[i] = dtypes.get(i);
     }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OrderedMapSize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OrderedMapSize.java
index d818185302d..60bed92f128 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OrderedMapSize.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OrderedMapSize.java
@@ -18,7 +18,6 @@
 package org.tensorflow.op.core;
 
 import java.util.List;
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -90,10 +89,10 @@ private Options() {
    * @return a new instance of OrderedMapSize
    */
   @Endpoint(describeByClass = true)
-  public static OrderedMapSize create(Scope scope, List<DataType<?>> dtypes, Options... options) {
+  public static OrderedMapSize create(Scope scope, List<Class<?>> dtypes, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("OrderedMapSize", scope.makeOpName("OrderedMapSize"));
     opBuilder = scope.applyControlDependencies(opBuilder);
-    DataType[] dtypesArray = new DataType[dtypes.size()];
+    Class[] dtypesArray = new Class[dtypes.size()];
     for (int i = 0; i < dtypesArray.length; ++i) {
       dtypesArray[i] = dtypes.get(i);
     }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OrderedMapStage.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OrderedMapStage.java
index 523fabafa6e..93f369860f0 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OrderedMapStage.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OrderedMapStage.java
@@ -18,7 +18,6 @@
 package org.tensorflow.op.core;
 
 import java.util.List;
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -99,13 +98,13 @@ private Options() {
    * @return a new instance of OrderedMapStage
    */
   @Endpoint(describeByClass = true)
-  public static OrderedMapStage create(Scope scope, Operand<TInt64> key, Operand<TInt32> indices, Iterable<Operand<?>> values, List<DataType<?>> dtypes, Options... options) {
+  public static OrderedMapStage create(Scope scope, Operand<TInt64> key, Operand<TInt32> indices, Iterable<Operand<?>> values, List<Class<?>> dtypes, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("OrderedMapStage", scope.makeOpName("OrderedMapStage"));
     opBuilder.addInput(key.asOutput(scope));
     opBuilder.addInput(indices.asOutput(scope));
     opBuilder.addInputList(Operands.asOutputs(scope, values));
     opBuilder = scope.applyControlDependencies(opBuilder);
-    DataType[] dtypesArray = new DataType[dtypes.size()];
+    Class[] dtypesArray = new Class[dtypes.size()];
     for (int i = 0; i < dtypesArray.length; ++i) {
       dtypesArray[i] = dtypes.get(i);
     }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OrderedMapUnstage.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OrderedMapUnstage.java
index b9289984e5f..28156e9c09c 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OrderedMapUnstage.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OrderedMapUnstage.java
@@ -20,7 +20,6 @@
 import java.util.Arrays;
 import java.util.Iterator;
 import java.util.List;
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -99,12 +98,12 @@ private Options() {
    * @return a new instance of OrderedMapUnstage
    */
   @Endpoint(describeByClass = true)
-  public static OrderedMapUnstage create(Scope scope, Operand<TInt64> key, Operand<TInt32> indices, List<DataType<?>> dtypes, Options... options) {
+  public static OrderedMapUnstage create(Scope scope, Operand<TInt64> key, Operand<TInt32> indices, List<Class<?>> dtypes, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("OrderedMapUnstage", scope.makeOpName("OrderedMapUnstage"));
     opBuilder.addInput(key.asOutput(scope));
     opBuilder.addInput(indices.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
-    DataType[] dtypesArray = new DataType[dtypes.size()];
+    Class[] dtypesArray = new Class[dtypes.size()];
     for (int i = 0; i < dtypesArray.length; ++i) {
       dtypesArray[i] = dtypes.get(i);
     }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OrderedMapUnstageNoKey.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OrderedMapUnstageNoKey.java
index c401e84c1c3..b8eeff0ee9c 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OrderedMapUnstageNoKey.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OrderedMapUnstageNoKey.java
@@ -19,7 +19,6 @@
 
 import java.util.Arrays;
 import java.util.List;
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -96,11 +95,11 @@ private Options() {
    * @return a new instance of OrderedMapUnstageNoKey
    */
   @Endpoint(describeByClass = true)
-  public static OrderedMapUnstageNoKey create(Scope scope, Operand<TInt32> indices, List<DataType<?>> dtypes, Options... options) {
+  public static OrderedMapUnstageNoKey create(Scope scope, Operand<TInt32> indices, List<Class<?>> dtypes, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("OrderedMapUnstageNoKey", scope.makeOpName("OrderedMapUnstageNoKey"));
     opBuilder.addInput(indices.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
-    DataType[] dtypesArray = new DataType[dtypes.size()];
+    Class[] dtypesArray = new Class[dtypes.size()];
     for (int i = 0; i < dtypesArray.length; ++i) {
       dtypesArray[i] = dtypes.get(i);
     }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Placeholder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Placeholder.java
index b8bbafd09b6..f38e589d90c 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Placeholder.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Placeholder.java
@@ -17,7 +17,6 @@
 
 package org.tensorflow.op.core;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -70,7 +69,7 @@ private Options() {
    * @return a new instance of Placeholder
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> Placeholder<T> create(Scope scope, DataType<T> dtype, Options... options) {
+  public static <T extends TType> Placeholder<T> create(Scope scope, Class<T> dtype, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("Placeholder", scope.makeOpName("Placeholder"));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("dtype", dtype);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ReadVariableOp.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ReadVariableOp.java
index 2f23e6aed41..9ca734e780e 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ReadVariableOp.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ReadVariableOp.java
@@ -17,7 +17,6 @@
 
 package org.tensorflow.op.core;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -52,7 +51,7 @@ public final class ReadVariableOp<T extends TType> extends RawOp implements Oper
    * @return a new instance of ReadVariableOp
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> ReadVariableOp<T> create(Scope scope, Operand<?> resource, DataType<T> dtype) {
+  public static <T extends TType> ReadVariableOp<T> create(Scope scope, Operand<?> resource, Class<T> dtype) {
     OperationBuilder opBuilder = scope.env().opBuilder("ReadVariableOp", scope.makeOpName("ReadVariableOp"));
     opBuilder.addInput(resource.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Recv.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Recv.java
index 1ebad338979..002eee646d6 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Recv.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Recv.java
@@ -17,7 +17,6 @@
 
 package org.tensorflow.op.core;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -70,7 +69,7 @@ private Options() {
    * @return a new instance of Recv
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> Recv<T> create(Scope scope, DataType<T> tensorType, String tensorName, String sendDevice, Long sendDeviceIncarnation, String recvDevice, Options... options) {
+  public static <T extends TType> Recv<T> create(Scope scope, Class<T> tensorType, String tensorName, String sendDevice, Long sendDeviceIncarnation, String recvDevice, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("Recv", scope.makeOpName("Recv"));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("tensor_type", tensorType);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/RemoteFusedGraphExecute.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/RemoteFusedGraphExecute.java
index 5462c762ac0..b92d9388f69 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/RemoteFusedGraphExecute.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/RemoteFusedGraphExecute.java
@@ -20,7 +20,6 @@
 import java.util.Arrays;
 import java.util.Iterator;
 import java.util.List;
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -57,11 +56,11 @@ public final class RemoteFusedGraphExecute extends RawOp implements Iterable<Ope
    * @return a new instance of RemoteFusedGraphExecute
    */
   @Endpoint(describeByClass = true)
-  public static RemoteFusedGraphExecute create(Scope scope, Iterable<Operand<?>> inputs, List<DataType<?>> Toutputs, String serializedRemoteFusedGraphExecuteInfo) {
+  public static RemoteFusedGraphExecute create(Scope scope, Iterable<Operand<?>> inputs, List<Class<?>> Toutputs, String serializedRemoteFusedGraphExecuteInfo) {
     OperationBuilder opBuilder = scope.env().opBuilder("RemoteFusedGraphExecute", scope.makeOpName("RemoteFusedGraphExecute"));
     opBuilder.addInputList(Operands.asOutputs(scope, inputs));
     opBuilder = scope.applyControlDependencies(opBuilder);
-    DataType[] ToutputsArray = new DataType[Toutputs.size()];
+    Class[] ToutputsArray = new Class[Toutputs.size()];
     for (int i = 0; i < ToutputsArray.length; ++i) {
       ToutputsArray[i] = Toutputs.get(i);
     }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceCountUpTo.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceCountUpTo.java
index 9070824ccc2..50c2af8aa7e 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceCountUpTo.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceCountUpTo.java
@@ -17,7 +17,6 @@
 
 package org.tensorflow.op.core;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -47,7 +46,7 @@ public final class ResourceCountUpTo<T extends TNumber> extends RawOp implements
    * @return a new instance of ResourceCountUpTo
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> ResourceCountUpTo<T> create(Scope scope, Operand<?> resource, Long limit, DataType<T> T) {
+  public static <T extends TNumber> ResourceCountUpTo<T> create(Scope scope, Operand<?> resource, Long limit, Class<T> T) {
     OperationBuilder opBuilder = scope.env().opBuilder("ResourceCountUpTo", scope.makeOpName("ResourceCountUpTo"));
     opBuilder.addInput(resource.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceGather.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceGather.java
index f112b891109..c9515a25932 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceGather.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceGather.java
@@ -17,7 +17,6 @@
 
 package org.tensorflow.op.core;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -90,7 +89,7 @@ private Options() {
    * @return a new instance of ResourceGather
    */
   @Endpoint(describeByClass = true)
-  public static <U extends TType, T extends TNumber> ResourceGather<U> create(Scope scope, Operand<?> resource, Operand<T> indices, DataType<U> dtype, Options... options) {
+  public static <U extends TType, T extends TNumber> ResourceGather<U> create(Scope scope, Operand<?> resource, Operand<T> indices, Class<U> dtype, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ResourceGather", scope.makeOpName("ResourceGather"));
     opBuilder.addInput(resource.asOutput(scope));
     opBuilder.addInput(indices.asOutput(scope));
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceGatherNd.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceGatherNd.java
index 46120fa6249..76c29afedd7 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceGatherNd.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceGatherNd.java
@@ -17,7 +17,6 @@
 
 package org.tensorflow.op.core;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -45,7 +44,7 @@ public final class ResourceGatherNd<U extends TType> extends RawOp implements Op
    * @return a new instance of ResourceGatherNd
    */
   @Endpoint(describeByClass = true)
-  public static <U extends TType, T extends TNumber> ResourceGatherNd<U> create(Scope scope, Operand<?> resource, Operand<T> indices, DataType<U> dtype) {
+  public static <U extends TType, T extends TNumber> ResourceGatherNd<U> create(Scope scope, Operand<?> resource, Operand<T> indices, Class<U> dtype) {
     OperationBuilder opBuilder = scope.env().opBuilder("ResourceGatherNd", scope.makeOpName("ResourceGatherNd"));
     opBuilder.addInput(resource.asOutput(scope));
     opBuilder.addInput(indices.asOutput(scope));
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterNdMax.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterNdMax.java
index 82c1f766308..40744ec4722 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterNdMax.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterNdMax.java
@@ -68,9 +68,9 @@ private Options() {
   @Endpoint(describeByClass = true)
   public static <T extends TNumber, U extends TType> ResourceScatterNdMax create(Scope scope, Operand<?> ref, Operand<T> indices, Operand<U> updates, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ResourceScatterNdMax", scope.makeOpName("ResourceScatterNdMax"));
-    opBuilder.addInput(ref.asOutput());
-    opBuilder.addInput(indices.asOutput());
-    opBuilder.addInput(updates.asOutput());
+    opBuilder.addInput(ref.asOutput(scope));
+    opBuilder.addInput(indices.asOutput(scope));
+    opBuilder.addInput(updates.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterNdMin.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterNdMin.java
index 88e107c65c7..e0645fca84e 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterNdMin.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterNdMin.java
@@ -68,9 +68,9 @@ private Options() {
   @Endpoint(describeByClass = true)
   public static <T extends TNumber, U extends TType> ResourceScatterNdMin create(Scope scope, Operand<?> ref, Operand<T> indices, Operand<U> updates, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ResourceScatterNdMin", scope.makeOpName("ResourceScatterNdMin"));
-    opBuilder.addInput(ref.asOutput());
-    opBuilder.addInput(indices.asOutput());
-    opBuilder.addInput(updates.asOutput());
+    opBuilder.addInput(ref.asOutput(scope));
+    opBuilder.addInput(indices.asOutput(scope));
+    opBuilder.addInput(updates.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterNdMax.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterNdMax.java
index da94c783cae..dcce88230ec 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterNdMax.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterNdMax.java
@@ -71,9 +71,9 @@ private Options() {
   @Endpoint(describeByClass = true)
   public static <T extends TType, U extends TNumber> ScatterNdMax<T> create(Scope scope, Operand<T> ref, Operand<U> indices, Operand<T> updates, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ScatterNdMax", scope.makeOpName("ScatterNdMax"));
-    opBuilder.addInput(ref.asOutput());
-    opBuilder.addInput(indices.asOutput());
-    opBuilder.addInput(updates.asOutput());
+    opBuilder.addInput(ref.asOutput(scope));
+    opBuilder.addInput(indices.asOutput(scope));
+    opBuilder.addInput(updates.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -103,7 +103,7 @@ public Output<T> outputRef() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return outputRef;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterNdMin.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterNdMin.java
index 5aea70bc929..73e74d4b50b 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterNdMin.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterNdMin.java
@@ -71,9 +71,9 @@ private Options() {
   @Endpoint(describeByClass = true)
   public static <T extends TType, U extends TNumber> ScatterNdMin<T> create(Scope scope, Operand<T> ref, Operand<U> indices, Operand<T> updates, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ScatterNdMin", scope.makeOpName("ScatterNdMin"));
-    opBuilder.addInput(ref.asOutput());
-    opBuilder.addInput(indices.asOutput());
-    opBuilder.addInput(updates.asOutput());
+    opBuilder.addInput(ref.asOutput(scope));
+    opBuilder.addInput(indices.asOutput(scope));
+    opBuilder.addInput(updates.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -103,7 +103,7 @@ public Output<T> outputRef() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return outputRef;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/SetDiff1d.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/SetDiff1d.java
index c0130ee43c3..ece12ee2a91 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/SetDiff1d.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/SetDiff1d.java
@@ -17,7 +17,6 @@
 
 package org.tensorflow.op.core;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -69,7 +68,7 @@ public final class SetDiff1d<T extends TType, U extends TNumber> extends RawOp {
    * @return a new instance of SetDiff1d
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> SetDiff1d<T, U> create(Scope scope, Operand<T> x, Operand<T> y, DataType<U> outIdx) {
+  public static <T extends TType, U extends TNumber> SetDiff1d<T, U> create(Scope scope, Operand<T> x, Operand<T> y, Class<U> outIdx) {
     OperationBuilder opBuilder = scope.env().opBuilder("ListDiff", scope.makeOpName("SetDiff1d"));
     opBuilder.addInput(x.asOutput(scope));
     opBuilder.addInput(y.asOutput(scope));
@@ -88,7 +87,7 @@ public static <T extends TType, U extends TNumber> SetDiff1d<T, U> create(Scope
    */
   @Endpoint(describeByClass = true)
   public static <T extends TType> SetDiff1d<T, TInt32> create(Scope scope, Operand<T> x, Operand<T> y) {
-    return create(scope, x, y, TInt32.DTYPE);
+    return create(scope, x, y, TInt32.class);
   }
   
   /**
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Shape.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Shape.java
index 3867a539be3..3b9779bb418 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Shape.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Shape.java
@@ -17,7 +17,6 @@
 
 package org.tensorflow.op.core;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -56,7 +55,7 @@ public final class Shape<U extends TNumber> extends RawOp implements Operand<U>
    * @return a new instance of Shape
    */
   @Endpoint(describeByClass = true)
-  public static <U extends TNumber, T extends TType> Shape<U> create(Scope scope, Operand<T> input, DataType<U> outType) {
+  public static <U extends TNumber, T extends TType> Shape<U> create(Scope scope, Operand<T> input, Class<U> outType) {
     OperationBuilder opBuilder = scope.env().opBuilder("Shape", scope.makeOpName("Shape"));
     opBuilder.addInput(input.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
@@ -73,7 +72,7 @@ public static <U extends TNumber, T extends TType> Shape<U> create(Scope scope,
    */
   @Endpoint(describeByClass = true)
   public static <T extends TType> Shape<TInt32> create(Scope scope, Operand<T> input) {
-    return create(scope, input, TInt32.DTYPE);
+    return create(scope, input, TInt32.class);
   }
   
   /**
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ShapeN.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ShapeN.java
index d390088de03..5d4ba217bb5 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ShapeN.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ShapeN.java
@@ -20,7 +20,6 @@
 import java.util.Arrays;
 import java.util.Iterator;
 import java.util.List;
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -53,7 +52,7 @@ public final class ShapeN<U extends TNumber> extends RawOp implements Iterable<O
    * @return a new instance of ShapeN
    */
   @Endpoint(describeByClass = true)
-  public static <U extends TNumber, T extends TType> ShapeN<U> create(Scope scope, Iterable<Operand<T>> input, DataType<U> outType) {
+  public static <U extends TNumber, T extends TType> ShapeN<U> create(Scope scope, Iterable<Operand<T>> input, Class<U> outType) {
     OperationBuilder opBuilder = scope.env().opBuilder("ShapeN", scope.makeOpName("ShapeN"));
     opBuilder.addInputList(Operands.asOutputs(scope, input));
     opBuilder = scope.applyControlDependencies(opBuilder);
@@ -70,7 +69,7 @@ public static <U extends TNumber, T extends TType> ShapeN<U> create(Scope scope,
    */
   @Endpoint(describeByClass = true)
   public static <T extends TType> ShapeN<TInt32> create(Scope scope, Iterable<Operand<T>> input) {
-    return create(scope, input, TInt32.DTYPE);
+    return create(scope, input, TInt32.class);
   }
   
   /**
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Size.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Size.java
index 18ccf934513..db1fc6180cf 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Size.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Size.java
@@ -17,7 +17,6 @@
 
 package org.tensorflow.op.core;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -57,7 +56,7 @@ public final class Size<U extends TNumber> extends RawOp implements Operand<U> {
    * @return a new instance of Size
    */
   @Endpoint(describeByClass = true)
-  public static <U extends TNumber, T extends TType> Size<U> create(Scope scope, Operand<T> input, DataType<U> outType) {
+  public static <U extends TNumber, T extends TType> Size<U> create(Scope scope, Operand<T> input, Class<U> outType) {
     OperationBuilder opBuilder = scope.env().opBuilder("Size", scope.makeOpName("Size"));
     opBuilder.addInput(input.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
@@ -74,7 +73,7 @@ public static <U extends TNumber, T extends TType> Size<U> create(Scope scope, O
    */
   @Endpoint(describeByClass = true)
   public static <T extends TType> Size<TInt32> create(Scope scope, Operand<T> input) {
-    return create(scope, input, TInt32.DTYPE);
+    return create(scope, input, TInt32.class);
   }
   
   /**
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/StageClear.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/StageClear.java
index 60e51559f74..11175fc9e4d 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/StageClear.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/StageClear.java
@@ -18,7 +18,6 @@
 package org.tensorflow.op.core;
 
 import java.util.List;
-import org.tensorflow.DataType;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.op.RawOp;
@@ -87,10 +86,10 @@ private Options() {
    * @return a new instance of StageClear
    */
   @Endpoint(describeByClass = true)
-  public static StageClear create(Scope scope, List<DataType<?>> dtypes, Options... options) {
+  public static StageClear create(Scope scope, List<Class<?>> dtypes, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("StageClear", scope.makeOpName("StageClear"));
     opBuilder = scope.applyControlDependencies(opBuilder);
-    DataType[] dtypesArray = new DataType[dtypes.size()];
+    Class[] dtypesArray = new Class[dtypes.size()];
     for (int i = 0; i < dtypesArray.length; ++i) {
       dtypesArray[i] = dtypes.get(i);
     }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/StagePeek.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/StagePeek.java
index 9256b9c9ef0..646d721a4e9 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/StagePeek.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/StagePeek.java
@@ -20,7 +20,6 @@
 import java.util.Arrays;
 import java.util.Iterator;
 import java.util.List;
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -98,11 +97,11 @@ private Options() {
    * @return a new instance of StagePeek
    */
   @Endpoint(describeByClass = true)
-  public static StagePeek create(Scope scope, Operand<TInt32> index, List<DataType<?>> dtypes, Options... options) {
+  public static StagePeek create(Scope scope, Operand<TInt32> index, List<Class<?>> dtypes, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("StagePeek", scope.makeOpName("StagePeek"));
     opBuilder.addInput(index.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
-    DataType[] dtypesArray = new DataType[dtypes.size()];
+    Class[] dtypesArray = new Class[dtypes.size()];
     for (int i = 0; i < dtypesArray.length; ++i) {
       dtypesArray[i] = dtypes.get(i);
     }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/StageSize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/StageSize.java
index fb489970ed1..b4d06a08bb4 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/StageSize.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/StageSize.java
@@ -18,7 +18,6 @@
 package org.tensorflow.op.core;
 
 import java.util.List;
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -90,10 +89,10 @@ private Options() {
    * @return a new instance of StageSize
    */
   @Endpoint(describeByClass = true)
-  public static StageSize create(Scope scope, List<DataType<?>> dtypes, Options... options) {
+  public static StageSize create(Scope scope, List<Class<?>> dtypes, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("StageSize", scope.makeOpName("StageSize"));
     opBuilder = scope.applyControlDependencies(opBuilder);
-    DataType[] dtypesArray = new DataType[dtypes.size()];
+    Class[] dtypesArray = new Class[dtypes.size()];
     for (int i = 0; i < dtypesArray.length; ++i) {
       dtypesArray[i] = dtypes.get(i);
     }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TemporaryVariable.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TemporaryVariable.java
index 85213671649..97b76053a4a 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TemporaryVariable.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TemporaryVariable.java
@@ -17,7 +17,6 @@
 
 package org.tensorflow.op.core;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -82,7 +81,7 @@ private Options() {
    * @return a new instance of TemporaryVariable
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> TemporaryVariable<T> create(Scope scope, Shape shape, DataType<T> dtype, Options... options) {
+  public static <T extends TType> TemporaryVariable<T> create(Scope scope, Shape shape, Class<T> dtype, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("TemporaryVariable", scope.makeOpName("TemporaryVariable"));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("shape", shape);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArray.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArray.java
index 41fea3a831a..41cff2dce10 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArray.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArray.java
@@ -17,7 +17,6 @@
 
 package org.tensorflow.op.core;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -116,7 +115,7 @@ private Options() {
    * @return a new instance of TensorArray
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> TensorArray create(Scope scope, Operand<TInt32> size, DataType<T> dtype, Options... options) {
+  public static <T extends TType> TensorArray create(Scope scope, Operand<TInt32> size, Class<T> dtype, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("TensorArrayV3", scope.makeOpName("TensorArray"));
     opBuilder.addInput(size.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayConcat.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayConcat.java
index 76b519c6ff5..3151585cf31 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayConcat.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayConcat.java
@@ -17,7 +17,6 @@
 
 package org.tensorflow.op.core;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -83,7 +82,7 @@ private Options() {
    * @return a new instance of TensorArrayConcat
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> TensorArrayConcat<T> create(Scope scope, Operand<?> handle, Operand<TFloat32> flowIn, DataType<T> dtype, Options... options) {
+  public static <T extends TType> TensorArrayConcat<T> create(Scope scope, Operand<?> handle, Operand<TFloat32> flowIn, Class<T> dtype, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("TensorArrayConcatV3", scope.makeOpName("TensorArrayConcat"));
     opBuilder.addInput(handle.asOutput(scope));
     opBuilder.addInput(flowIn.asOutput(scope));
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayGather.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayGather.java
index 4595687db01..69fdd98af0d 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayGather.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayGather.java
@@ -17,7 +17,6 @@
 
 package org.tensorflow.op.core;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -74,7 +73,7 @@ private Options() {
    * @return a new instance of TensorArrayGather
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> TensorArrayGather<T> create(Scope scope, Operand<?> handle, Operand<TInt32> indices, Operand<TFloat32> flowIn, DataType<T> dtype, Options... options) {
+  public static <T extends TType> TensorArrayGather<T> create(Scope scope, Operand<?> handle, Operand<TInt32> indices, Operand<TFloat32> flowIn, Class<T> dtype, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("TensorArrayGatherV3", scope.makeOpName("TensorArrayGather"));
     opBuilder.addInput(handle.asOutput(scope));
     opBuilder.addInput(indices.asOutput(scope));
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayPack.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayPack.java
index e2ad54ec5a3..dbed7f00a84 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayPack.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayPack.java
@@ -17,7 +17,6 @@
 
 package org.tensorflow.op.core;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -67,7 +66,7 @@ private Options() {
    * @return a new instance of TensorArrayPack
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> TensorArrayPack<T> create(Scope scope, Operand<TString> handle, Operand<TFloat32> flowIn, DataType<T> dtype, Options... options) {
+  public static <T extends TType> TensorArrayPack<T> create(Scope scope, Operand<TString> handle, Operand<TFloat32> flowIn, Class<T> dtype, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("TensorArrayPack", scope.makeOpName("TensorArrayPack"));
     opBuilder.addInput(handle.asOutput(scope));
     opBuilder.addInput(flowIn.asOutput(scope));
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayRead.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayRead.java
index 2634a6c0cf2..bbefe81a7bf 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayRead.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayRead.java
@@ -17,7 +17,6 @@
 
 package org.tensorflow.op.core;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -49,7 +48,7 @@ public final class TensorArrayRead<T extends TType> extends RawOp implements Ope
    * @return a new instance of TensorArrayRead
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> TensorArrayRead<T> create(Scope scope, Operand<?> handle, Operand<TInt32> index, Operand<TFloat32> flowIn, DataType<T> dtype) {
+  public static <T extends TType> TensorArrayRead<T> create(Scope scope, Operand<?> handle, Operand<TInt32> index, Operand<TFloat32> flowIn, Class<T> dtype) {
     OperationBuilder opBuilder = scope.env().opBuilder("TensorArrayReadV3", scope.makeOpName("TensorArrayRead"));
     opBuilder.addInput(handle.asOutput(scope));
     opBuilder.addInput(index.asOutput(scope));
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListConcat.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListConcat.java
index da39e92abbc..387cf6db10b 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListConcat.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListConcat.java
@@ -17,7 +17,6 @@
 
 package org.tensorflow.op.core;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -62,7 +61,7 @@ public final class TensorListConcat<U extends TType> extends RawOp {
    * @return a new instance of TensorListConcat
    */
   @Endpoint(describeByClass = true)
-  public static <U extends TType, T extends TNumber> TensorListConcat<U> create(Scope scope, Operand<?> inputHandle, Operand<T> elementShape, Operand<TInt64> leadingDims, DataType<U> elementDtype) {
+  public static <U extends TType, T extends TNumber> TensorListConcat<U> create(Scope scope, Operand<?> inputHandle, Operand<T> elementShape, Operand<TInt64> leadingDims, Class<U> elementDtype) {
     OperationBuilder opBuilder = scope.env().opBuilder("TensorListConcatV2", scope.makeOpName("TensorListConcat"));
     opBuilder.addInput(inputHandle.asOutput(scope));
     opBuilder.addInput(elementShape.asOutput(scope));
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListConcatLists.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListConcatLists.java
index ae959e28808..20a4ee32dff 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListConcatLists.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListConcatLists.java
@@ -17,7 +17,6 @@
 
 package org.tensorflow.op.core;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -43,7 +42,7 @@ public final class TensorListConcatLists extends RawOp implements Operand<TType>
    * @return a new instance of TensorListConcatLists
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> TensorListConcatLists create(Scope scope, Operand<?> inputA, Operand<?> inputB, DataType<T> elementDtype) {
+  public static <T extends TType> TensorListConcatLists create(Scope scope, Operand<?> inputA, Operand<?> inputB, Class<T> elementDtype) {
     OperationBuilder opBuilder = scope.env().opBuilder("TensorListConcatLists", scope.makeOpName("TensorListConcatLists"));
     opBuilder.addInput(inputA.asOutput(scope));
     opBuilder.addInput(inputB.asOutput(scope));
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListElementShape.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListElementShape.java
index f9b95898d6d..0099f14585c 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListElementShape.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListElementShape.java
@@ -17,7 +17,6 @@
 
 package org.tensorflow.op.core;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -48,7 +47,7 @@ public final class TensorListElementShape<T extends TNumber> extends RawOp imple
    * @return a new instance of TensorListElementShape
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> TensorListElementShape<T> create(Scope scope, Operand<?> inputHandle, DataType<T> shapeType) {
+  public static <T extends TNumber> TensorListElementShape<T> create(Scope scope, Operand<?> inputHandle, Class<T> shapeType) {
     OperationBuilder opBuilder = scope.env().opBuilder("TensorListElementShape", scope.makeOpName("TensorListElementShape"));
     opBuilder.addInput(inputHandle.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListGather.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListGather.java
index 224fe935e5b..1062c76ad17 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListGather.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListGather.java
@@ -17,7 +17,6 @@
 
 package org.tensorflow.op.core;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -55,7 +54,7 @@ public final class TensorListGather<T extends TType> extends RawOp implements Op
    * @return a new instance of TensorListGather
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> TensorListGather<T> create(Scope scope, Operand<?> inputHandle, Operand<TInt32> indices, Operand<TInt32> elementShape, DataType<T> elementDtype) {
+  public static <T extends TType> TensorListGather<T> create(Scope scope, Operand<?> inputHandle, Operand<TInt32> indices, Operand<TInt32> elementShape, Class<T> elementDtype) {
     OperationBuilder opBuilder = scope.env().opBuilder("TensorListGather", scope.makeOpName("TensorListGather"));
     opBuilder.addInput(inputHandle.asOutput(scope));
     opBuilder.addInput(indices.asOutput(scope));
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListGetItem.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListGetItem.java
index 7f435a21a27..eb2ab5423eb 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListGetItem.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListGetItem.java
@@ -17,7 +17,6 @@
 
 package org.tensorflow.op.core;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -46,7 +45,7 @@ public final class TensorListGetItem<T extends TType> extends RawOp implements O
    * @return a new instance of TensorListGetItem
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> TensorListGetItem<T> create(Scope scope, Operand<?> inputHandle, Operand<TInt32> index, Operand<TInt32> elementShape, DataType<T> elementDtype) {
+  public static <T extends TType> TensorListGetItem<T> create(Scope scope, Operand<?> inputHandle, Operand<TInt32> index, Operand<TInt32> elementShape, Class<T> elementDtype) {
     OperationBuilder opBuilder = scope.env().opBuilder("TensorListGetItem", scope.makeOpName("TensorListGetItem"));
     opBuilder.addInput(inputHandle.asOutput(scope));
     opBuilder.addInput(index.asOutput(scope));
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListPopBack.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListPopBack.java
index 4489f7ae733..d2df132000c 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListPopBack.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListPopBack.java
@@ -17,7 +17,6 @@
 
 package org.tensorflow.op.core;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -54,7 +53,7 @@ public final class TensorListPopBack<T extends TType> extends RawOp {
    * @return a new instance of TensorListPopBack
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> TensorListPopBack<T> create(Scope scope, Operand<?> inputHandle, Operand<TInt32> elementShape, DataType<T> elementDtype) {
+  public static <T extends TType> TensorListPopBack<T> create(Scope scope, Operand<?> inputHandle, Operand<TInt32> elementShape, Class<T> elementDtype) {
     OperationBuilder opBuilder = scope.env().opBuilder("TensorListPopBack", scope.makeOpName("TensorListPopBack"));
     opBuilder.addInput(inputHandle.asOutput(scope));
     opBuilder.addInput(elementShape.asOutput(scope));
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListReserve.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListReserve.java
index c5a2e63bff5..13f68b7492f 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListReserve.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListReserve.java
@@ -17,7 +17,6 @@
 
 package org.tensorflow.op.core;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -51,7 +50,7 @@ public final class TensorListReserve extends RawOp implements Operand<TType> {
    * @return a new instance of TensorListReserve
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber, U extends TType> TensorListReserve create(Scope scope, Operand<T> elementShape, Operand<TInt32> numElements, DataType<U> elementDtype) {
+  public static <T extends TNumber, U extends TType> TensorListReserve create(Scope scope, Operand<T> elementShape, Operand<TInt32> numElements, Class<U> elementDtype) {
     OperationBuilder opBuilder = scope.env().opBuilder("TensorListReserve", scope.makeOpName("TensorListReserve"));
     opBuilder.addInput(elementShape.asOutput(scope));
     opBuilder.addInput(numElements.asOutput(scope));
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListStack.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListStack.java
index 7919721b524..42dc682cda6 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListStack.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListStack.java
@@ -17,7 +17,6 @@
 
 package org.tensorflow.op.core;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -74,7 +73,7 @@ private Options() {
    * @return a new instance of TensorListStack
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> TensorListStack<T> create(Scope scope, Operand<?> inputHandle, Operand<TInt32> elementShape, DataType<T> elementDtype, Options... options) {
+  public static <T extends TType> TensorListStack<T> create(Scope scope, Operand<?> inputHandle, Operand<TInt32> elementShape, Class<T> elementDtype, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("TensorListStack", scope.makeOpName("TensorListStack"));
     opBuilder.addInput(inputHandle.asOutput(scope));
     opBuilder.addInput(elementShape.asOutput(scope));
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorScatterMax.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorScatterMax.java
deleted file mode 100644
index 1a3042926f6..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorScatterMax.java
+++ /dev/null
@@ -1,75 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * @param <T> data type for {@code output()} output
- */
-@Operator
-public final class TensorScatterMax<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new TensorScatterMax operation.
-   * 
-   * @param scope current scope
-   * @param tensor Tensor to update.
-   * @param indices Index tensor.
-   * @param updates Updates to scatter into output.
-   * @return a new instance of TensorScatterMax
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> TensorScatterMax<T> create(Scope scope, Operand<T> tensor, Operand<U> indices, Operand<T> updates) {
-    OperationBuilder opBuilder = scope.env().opBuilder("TensorScatterMax", scope.makeOpName("TensorScatterMax"));
-    opBuilder.addInput(tensor.asOutput());
-    opBuilder.addInput(indices.asOutput());
-    opBuilder.addInput(updates.asOutput());
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new TensorScatterMax<T>(opBuilder.build());
-  }
-  
-  /**
-   * A new tensor copied from tensor whose values are element-wise maximum between tensor and updates according to the indices.
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput() {
-    return output;
-  }
-  
-  private Output<T> output;
-  
-  private TensorScatterMax(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorScatterMin.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorScatterMin.java
deleted file mode 100644
index 5f7b0d0eb77..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorScatterMin.java
+++ /dev/null
@@ -1,75 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * @param <T> data type for {@code output()} output
- */
-@Operator
-public final class TensorScatterMin<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new TensorScatterMin operation.
-   * 
-   * @param scope current scope
-   * @param tensor Tensor to update.
-   * @param indices Index tensor.
-   * @param updates Updates to scatter into output.
-   * @return a new instance of TensorScatterMin
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> TensorScatterMin<T> create(Scope scope, Operand<T> tensor, Operand<U> indices, Operand<T> updates) {
-    OperationBuilder opBuilder = scope.env().opBuilder("TensorScatterMin", scope.makeOpName("TensorScatterMin"));
-    opBuilder.addInput(tensor.asOutput());
-    opBuilder.addInput(indices.asOutput());
-    opBuilder.addInput(updates.asOutput());
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new TensorScatterMin<T>(opBuilder.build());
-  }
-  
-  /**
-   * A new tensor copied from tensor whose values are element-wise minimum between tensor and updates according to the indices.
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput() {
-    return output;
-  }
-  
-  private Output<T> output;
-  
-  private TensorScatterMin(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorScatterNdMax.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorScatterNdMax.java
index a14b20195af..55d07a26a4e 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorScatterNdMax.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorScatterNdMax.java
@@ -46,9 +46,9 @@ public final class TensorScatterNdMax<T extends TType> extends RawOp implements
   @Endpoint(describeByClass = true)
   public static <T extends TType, U extends TNumber> TensorScatterNdMax<T> create(Scope scope, Operand<T> tensor, Operand<U> indices, Operand<T> updates) {
     OperationBuilder opBuilder = scope.env().opBuilder("TensorScatterMax", scope.makeOpName("TensorScatterNdMax"));
-    opBuilder.addInput(tensor.asOutput());
-    opBuilder.addInput(indices.asOutput());
-    opBuilder.addInput(updates.asOutput());
+    opBuilder.addInput(tensor.asOutput(scope));
+    opBuilder.addInput(indices.asOutput(scope));
+    opBuilder.addInput(updates.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new TensorScatterNdMax<T>(opBuilder.build());
   }
@@ -61,7 +61,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorScatterNdMin.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorScatterNdMin.java
index b202b72eebd..a021b1cfc03 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorScatterNdMin.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorScatterNdMin.java
@@ -46,9 +46,9 @@ public final class TensorScatterNdMin<T extends TType> extends RawOp implements
   @Endpoint(describeByClass = true)
   public static <T extends TType, U extends TNumber> TensorScatterNdMin<T> create(Scope scope, Operand<T> tensor, Operand<U> indices, Operand<T> updates) {
     OperationBuilder opBuilder = scope.env().opBuilder("TensorScatterMin", scope.makeOpName("TensorScatterNdMin"));
-    opBuilder.addInput(tensor.asOutput());
-    opBuilder.addInput(indices.asOutput());
-    opBuilder.addInput(updates.asOutput());
+    opBuilder.addInput(tensor.asOutput(scope));
+    opBuilder.addInput(indices.asOutput(scope));
+    opBuilder.addInput(updates.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new TensorScatterNdMin<T>(opBuilder.build());
   }
@@ -61,7 +61,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Unique.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Unique.java
index a3f64684cd7..c4d56b4e8b6 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Unique.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Unique.java
@@ -17,7 +17,6 @@
 
 package org.tensorflow.op.core;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -90,7 +89,7 @@ public final class Unique<T extends TType, V extends TNumber> extends RawOp {
    * @return a new instance of Unique
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType, V extends TNumber, U extends TNumber> Unique<T, V> create(Scope scope, Operand<T> x, Operand<U> axis, DataType<V> outIdx) {
+  public static <T extends TType, V extends TNumber, U extends TNumber> Unique<T, V> create(Scope scope, Operand<T> x, Operand<U> axis, Class<V> outIdx) {
     OperationBuilder opBuilder = scope.env().opBuilder("UniqueV2", scope.makeOpName("Unique"));
     opBuilder.addInput(x.asOutput(scope));
     opBuilder.addInput(axis.asOutput(scope));
@@ -110,7 +109,7 @@ public static <T extends TType, V extends TNumber, U extends TNumber> Unique<T,
    */
   @Endpoint(describeByClass = true)
   public static <T extends TType, U extends TNumber> Unique<T, TInt32> create(Scope scope, Operand<T> x, Operand<U> axis) {
-    return create(scope, x, axis, TInt32.DTYPE);
+    return create(scope, x, axis, TInt32.class);
   }
   
   /**
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/UniqueWithCounts.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/UniqueWithCounts.java
index 62804827da5..74046926294 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/UniqueWithCounts.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/UniqueWithCounts.java
@@ -17,7 +17,6 @@
 
 package org.tensorflow.op.core;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -94,7 +93,7 @@ public final class UniqueWithCounts<T extends TType, V extends TNumber> extends
    * @return a new instance of UniqueWithCounts
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType, V extends TNumber, U extends TNumber> UniqueWithCounts<T, V> create(Scope scope, Operand<T> x, Operand<U> axis, DataType<V> outIdx) {
+  public static <T extends TType, V extends TNumber, U extends TNumber> UniqueWithCounts<T, V> create(Scope scope, Operand<T> x, Operand<U> axis, Class<V> outIdx) {
     OperationBuilder opBuilder = scope.env().opBuilder("UniqueWithCountsV2", scope.makeOpName("UniqueWithCounts"));
     opBuilder.addInput(x.asOutput(scope));
     opBuilder.addInput(axis.asOutput(scope));
@@ -114,7 +113,7 @@ public static <T extends TType, V extends TNumber, U extends TNumber> UniqueWith
    */
   @Endpoint(describeByClass = true)
   public static <T extends TType, U extends TNumber> UniqueWithCounts<T, TInt32> create(Scope scope, Operand<T> x, Operand<U> axis) {
-    return create(scope, x, axis, TInt32.DTYPE);
+    return create(scope, x, axis, TInt32.class);
   }
   
   /**
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Unstage.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Unstage.java
index 1878ce6fb88..d92ec49bc2a 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Unstage.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Unstage.java
@@ -20,7 +20,6 @@
 import java.util.Arrays;
 import java.util.Iterator;
 import java.util.List;
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -95,10 +94,10 @@ private Options() {
    * @return a new instance of Unstage
    */
   @Endpoint(describeByClass = true)
-  public static Unstage create(Scope scope, List<DataType<?>> dtypes, Options... options) {
+  public static Unstage create(Scope scope, List<Class<?>> dtypes, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("Unstage", scope.makeOpName("Unstage"));
     opBuilder = scope.applyControlDependencies(opBuilder);
-    DataType[] dtypesArray = new DataType[dtypes.size()];
+    Class[] dtypesArray = new Class[dtypes.size()];
     for (int i = 0; i < dtypesArray.length; ++i) {
       dtypesArray[i] = dtypes.get(i);
     }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/UpperBound.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/UpperBound.java
index f7117601ec3..8e5b59b0fff 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/UpperBound.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/UpperBound.java
@@ -17,7 +17,6 @@
 
 package org.tensorflow.op.core;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -66,7 +65,7 @@ public final class UpperBound<U extends TNumber> extends RawOp implements Operan
    * @return a new instance of UpperBound
    */
   @Endpoint(describeByClass = true)
-  public static <U extends TNumber, T extends TType> UpperBound<U> create(Scope scope, Operand<T> sortedInputs, Operand<T> values, DataType<U> outType) {
+  public static <U extends TNumber, T extends TType> UpperBound<U> create(Scope scope, Operand<T> sortedInputs, Operand<T> values, Class<U> outType) {
     OperationBuilder opBuilder = scope.env().opBuilder("UpperBound", scope.makeOpName("UpperBound"));
     opBuilder.addInput(sortedInputs.asOutput(scope));
     opBuilder.addInput(values.asOutput(scope));
@@ -86,7 +85,7 @@ public static <U extends TNumber, T extends TType> UpperBound<U> create(Scope sc
    */
   @Endpoint(describeByClass = true)
   public static <T extends TType> UpperBound<TInt32> create(Scope scope, Operand<T> sortedInputs, Operand<T> values) {
-    return create(scope, sortedInputs, values, TInt32.DTYPE);
+    return create(scope, sortedInputs, values, TInt32.class);
   }
   
   /**
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/VarHandleOp.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/VarHandleOp.java
index 2bd97143dcb..65f8784c3ae 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/VarHandleOp.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/VarHandleOp.java
@@ -18,7 +18,6 @@
 package org.tensorflow.op.core;
 
 import java.util.List;
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -85,7 +84,7 @@ private Options() {
    * @return a new instance of VarHandleOp
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> VarHandleOp create(Scope scope, DataType<T> dtype, Shape shape, Options... options) {
+  public static <T extends TType> VarHandleOp create(Scope scope, Class<T> dtype, Shape shape, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("VarHandleOp", scope.makeOpName("VarHandleOp"));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("dtype", dtype);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Variable.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Variable.java
index 1c0c9151156..7c89a2518e2 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Variable.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Variable.java
@@ -17,7 +17,6 @@
 
 package org.tensorflow.op.core;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -81,7 +80,7 @@ private Options() {
    * @return a new instance of Variable
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> Variable<T> create(Scope scope, Shape shape, DataType<T> dtype, Options... options) {
+  public static <T extends TType> Variable<T> create(Scope scope, Shape shape, Class<T> dtype, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("VariableV2", scope.makeOpName("Variable"));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("shape", shape);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/VariableShape.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/VariableShape.java
index 97ae695bcc4..fa10b66ef00 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/VariableShape.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/VariableShape.java
@@ -17,7 +17,6 @@
 
 package org.tensorflow.op.core;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -55,7 +54,7 @@ public final class VariableShape<T extends TNumber> extends RawOp implements Ope
    * @return a new instance of VariableShape
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> VariableShape<T> create(Scope scope, Operand<?> input, DataType<T> outType) {
+  public static <T extends TNumber> VariableShape<T> create(Scope scope, Operand<?> input, Class<T> outType) {
     OperationBuilder opBuilder = scope.env().opBuilder("VariableShape", scope.makeOpName("VariableShape"));
     opBuilder.addInput(input.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
@@ -72,7 +71,7 @@ public static <T extends TNumber> VariableShape<T> create(Scope scope, Operand<?
    */
   @Endpoint(describeByClass = true)
   public static VariableShape<TInt32> create(Scope scope, Operand<?> input) {
-    return create(scope, input, TInt32.DTYPE);
+    return create(scope, input, TInt32.class);
   }
   
   /**
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/XlaSpmdFullToShardShape.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/XlaSpmdFullToShardShape.java
index 6615d2ef9f6..c28e44252b5 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/XlaSpmdFullToShardShape.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/XlaSpmdFullToShardShape.java
@@ -51,7 +51,7 @@ public final class XlaSpmdFullToShardShape<T extends TType> extends RawOp implem
   @Endpoint(describeByClass = true)
   public static <T extends TType> XlaSpmdFullToShardShape<T> create(Scope scope, Operand<T> input, String manualSharding) {
     OperationBuilder opBuilder = scope.env().opBuilder("XlaSpmdFullToShardShape", scope.makeOpName("XlaSpmdFullToShardShape"));
-    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("manual_sharding", manualSharding);
     return new XlaSpmdFullToShardShape<T>(opBuilder.build());
@@ -64,7 +64,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/XlaSpmdShardToFullShape.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/XlaSpmdShardToFullShape.java
index 75e31c7c317..ef9d7379d21 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/XlaSpmdShardToFullShape.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/XlaSpmdShardToFullShape.java
@@ -52,7 +52,7 @@ public final class XlaSpmdShardToFullShape<T extends TType> extends RawOp implem
   @Endpoint(describeByClass = true)
   public static <T extends TType> XlaSpmdShardToFullShape<T> create(Scope scope, Operand<T> input, String manualSharding, Shape fullShape) {
     OperationBuilder opBuilder = scope.env().opBuilder("XlaSpmdShardToFullShape", scope.makeOpName("XlaSpmdShardToFullShape"));
-    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("manual_sharding", manualSharding);
     opBuilder.setAttr("full_shape", fullShape);
@@ -66,7 +66,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/AnonymousIterator.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/AnonymousIterator.java
index bd3d1a400f5..ce6d0515fc2 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/AnonymousIterator.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/AnonymousIterator.java
@@ -18,7 +18,6 @@
 package org.tensorflow.op.data;
 
 import java.util.List;
-import org.tensorflow.DataType;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
@@ -43,10 +42,10 @@ public final class AnonymousIterator extends RawOp {
    * @return a new instance of AnonymousIterator
    */
   @Endpoint(describeByClass = true)
-  public static AnonymousIterator create(Scope scope, List<DataType<?>> outputTypes, List<Shape> outputShapes) {
+  public static AnonymousIterator create(Scope scope, List<Class<?>> outputTypes, List<Shape> outputShapes) {
     OperationBuilder opBuilder = scope.env().opBuilder("AnonymousIteratorV2", scope.makeOpName("AnonymousIterator"));
     opBuilder = scope.applyControlDependencies(opBuilder);
-    DataType[] outputTypesArray = new DataType[outputTypes.size()];
+    Class[] outputTypesArray = new Class[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
       outputTypesArray[i] = outputTypes.get(i);
     }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/AnonymousMultiDeviceIterator.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/AnonymousMultiDeviceIterator.java
index bff57b33c8f..8857254635f 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/AnonymousMultiDeviceIterator.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/AnonymousMultiDeviceIterator.java
@@ -18,7 +18,6 @@
 package org.tensorflow.op.data;
 
 import java.util.List;
-import org.tensorflow.DataType;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
 import org.tensorflow.Output;
@@ -43,7 +42,7 @@ public final class AnonymousMultiDeviceIterator extends RawOp {
    * @return a new instance of AnonymousMultiDeviceIterator
    */
   @Endpoint(describeByClass = true)
-  public static AnonymousMultiDeviceIterator create(Scope scope, List<String> devices, List<DataType<?>> outputTypes, List<Shape> outputShapes) {
+  public static AnonymousMultiDeviceIterator create(Scope scope, List<String> devices, List<Class<?>> outputTypes, List<Shape> outputShapes) {
     OperationBuilder opBuilder = scope.env().opBuilder("AnonymousMultiDeviceIterator", scope.makeOpName("AnonymousMultiDeviceIterator"));
     opBuilder = scope.applyControlDependencies(opBuilder);
     String[] devicesArray = new String[devices.size()];
@@ -51,7 +50,7 @@ public static AnonymousMultiDeviceIterator create(Scope scope, List<String> devi
       devicesArray[i] = devices.get(i);
     }
     opBuilder.setAttr("devices", devicesArray);
-    DataType[] outputTypesArray = new DataType[outputTypes.size()];
+    Class[] outputTypesArray = new Class[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
       outputTypesArray[i] = outputTypes.get(i);
     }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/AssertNextDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/AssertNextDataset.java
index ee9c8977d29..ba317ce9b11 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/AssertNextDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/AssertNextDataset.java
@@ -18,7 +18,6 @@
 package org.tensorflow.op.data;
 
 import java.util.List;
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -58,12 +57,12 @@ public final class AssertNextDataset extends RawOp implements Operand<TType> {
    * @return a new instance of AssertNextDataset
    */
   @Endpoint(describeByClass = true)
-  public static AssertNextDataset create(Scope scope, Operand<?> inputDataset, Operand<TString> transformations, List<DataType<?>> outputTypes, List<Shape> outputShapes) {
+  public static AssertNextDataset create(Scope scope, Operand<?> inputDataset, Operand<TString> transformations, List<Class<?>> outputTypes, List<Shape> outputShapes) {
     OperationBuilder opBuilder = scope.env().opBuilder("AssertNextDataset", scope.makeOpName("AssertNextDataset"));
     opBuilder.addInput(inputDataset.asOutput(scope));
     opBuilder.addInput(transformations.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
-    DataType[] outputTypesArray = new DataType[outputTypes.size()];
+    Class[] outputTypesArray = new Class[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
       outputTypesArray[i] = outputTypes.get(i);
     }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/AutoShardDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/AutoShardDataset.java
index 180ba4f5d71..818c94926cb 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/AutoShardDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/AutoShardDataset.java
@@ -18,7 +18,6 @@
 package org.tensorflow.op.data;
 
 import java.util.List;
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -76,13 +75,13 @@ private Options() {
    * @return a new instance of AutoShardDataset
    */
   @Endpoint(describeByClass = true)
-  public static AutoShardDataset create(Scope scope, Operand<?> inputDataset, Operand<TInt64> numWorkers, Operand<TInt64> index, List<DataType<?>> outputTypes, List<Shape> outputShapes, Options... options) {
+  public static AutoShardDataset create(Scope scope, Operand<?> inputDataset, Operand<TInt64> numWorkers, Operand<TInt64> index, List<Class<?>> outputTypes, List<Shape> outputShapes, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("AutoShardDataset", scope.makeOpName("AutoShardDataset"));
     opBuilder.addInput(inputDataset.asOutput(scope));
     opBuilder.addInput(numWorkers.asOutput(scope));
     opBuilder.addInput(index.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
-    DataType[] outputTypesArray = new DataType[outputTypes.size()];
+    Class[] outputTypesArray = new Class[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
       outputTypesArray[i] = outputTypes.get(i);
     }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/BatchDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/BatchDataset.java
index 7a827ad43eb..cb3a36abf99 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/BatchDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/BatchDataset.java
@@ -18,7 +18,6 @@
 package org.tensorflow.op.data;
 
 import java.util.List;
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -71,13 +70,13 @@ private Options() {
    * @return a new instance of BatchDataset
    */
   @Endpoint(describeByClass = true)
-  public static BatchDataset create(Scope scope, Operand<?> inputDataset, Operand<TInt64> batchSize, Operand<TBool> dropRemainder, List<DataType<?>> outputTypes, List<Shape> outputShapes, Options... options) {
+  public static BatchDataset create(Scope scope, Operand<?> inputDataset, Operand<TInt64> batchSize, Operand<TBool> dropRemainder, List<Class<?>> outputTypes, List<Shape> outputShapes, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("BatchDatasetV2", scope.makeOpName("BatchDataset"));
     opBuilder.addInput(inputDataset.asOutput(scope));
     opBuilder.addInput(batchSize.asOutput(scope));
     opBuilder.addInput(dropRemainder.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
-    DataType[] outputTypesArray = new DataType[outputTypes.size()];
+    Class[] outputTypesArray = new Class[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
       outputTypesArray[i] = outputTypes.get(i);
     }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/BytesProducedStatsDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/BytesProducedStatsDataset.java
index 7b73e98c293..d6c33dbd767 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/BytesProducedStatsDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/BytesProducedStatsDataset.java
@@ -18,7 +18,6 @@
 package org.tensorflow.op.data;
 
 import java.util.List;
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -47,12 +46,12 @@ public final class BytesProducedStatsDataset extends RawOp implements Operand<TT
    * @return a new instance of BytesProducedStatsDataset
    */
   @Endpoint(describeByClass = true)
-  public static BytesProducedStatsDataset create(Scope scope, Operand<?> inputDataset, Operand<TString> tag, List<DataType<?>> outputTypes, List<Shape> outputShapes) {
+  public static BytesProducedStatsDataset create(Scope scope, Operand<?> inputDataset, Operand<TString> tag, List<Class<?>> outputTypes, List<Shape> outputShapes) {
     OperationBuilder opBuilder = scope.env().opBuilder("BytesProducedStatsDataset", scope.makeOpName("BytesProducedStatsDataset"));
     opBuilder.addInput(inputDataset.asOutput(scope));
     opBuilder.addInput(tag.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
-    DataType[] outputTypesArray = new DataType[outputTypes.size()];
+    Class[] outputTypesArray = new Class[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
       outputTypesArray[i] = outputTypes.get(i);
     }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/CacheDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/CacheDataset.java
index 1dd8e604bc4..e2c3d4444e6 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/CacheDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/CacheDataset.java
@@ -18,7 +18,6 @@
 package org.tensorflow.op.data;
 
 import java.util.List;
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -53,12 +52,12 @@ public final class CacheDataset extends RawOp implements Operand<TType> {
    * @return a new instance of CacheDataset
    */
   @Endpoint(describeByClass = true)
-  public static CacheDataset create(Scope scope, Operand<?> inputDataset, Operand<TString> filename, List<DataType<?>> outputTypes, List<Shape> outputShapes) {
+  public static CacheDataset create(Scope scope, Operand<?> inputDataset, Operand<TString> filename, List<Class<?>> outputTypes, List<Shape> outputShapes) {
     OperationBuilder opBuilder = scope.env().opBuilder("CacheDataset", scope.makeOpName("CacheDataset"));
     opBuilder.addInput(inputDataset.asOutput(scope));
     opBuilder.addInput(filename.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
-    DataType[] outputTypesArray = new DataType[outputTypes.size()];
+    Class[] outputTypesArray = new Class[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
       outputTypesArray[i] = outputTypes.get(i);
     }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/CacheDatasetV2.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/CacheDatasetV2.java
index 7dac3483939..7cd69f5fc66 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/CacheDatasetV2.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/CacheDatasetV2.java
@@ -18,7 +18,6 @@
 package org.tensorflow.op.data;
 
 import java.util.List;
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -47,13 +46,13 @@ public final class CacheDatasetV2 extends RawOp implements Operand<TType> {
    * @return a new instance of CacheDatasetV2
    */
   @Endpoint(describeByClass = true)
-  public static CacheDatasetV2 create(Scope scope, Operand<?> inputDataset, Operand<TString> filename, Operand<?> cache, List<DataType<?>> outputTypes, List<Shape> outputShapes) {
+  public static CacheDatasetV2 create(Scope scope, Operand<?> inputDataset, Operand<TString> filename, Operand<?> cache, List<Class<?>> outputTypes, List<Shape> outputShapes) {
     OperationBuilder opBuilder = scope.env().opBuilder("CacheDatasetV2", scope.makeOpName("CacheDatasetV2"));
     opBuilder.addInput(inputDataset.asOutput(scope));
     opBuilder.addInput(filename.asOutput(scope));
     opBuilder.addInput(cache.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
-    DataType[] outputTypesArray = new DataType[outputTypes.size()];
+    Class[] outputTypesArray = new Class[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
       outputTypesArray[i] = outputTypes.get(i);
     }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ChooseFastestDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ChooseFastestDataset.java
index e1e2e718f21..d97ddf04970 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ChooseFastestDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ChooseFastestDataset.java
@@ -18,7 +18,6 @@
 package org.tensorflow.op.data;
 
 import java.util.List;
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -46,12 +45,12 @@ public final class ChooseFastestDataset extends RawOp implements Operand<TType>
    * @return a new instance of ChooseFastestDataset
    */
   @Endpoint(describeByClass = true)
-  public static ChooseFastestDataset create(Scope scope, Iterable<Operand<?>> inputDatasets, Long numExperiments, List<DataType<?>> outputTypes, List<Shape> outputShapes) {
+  public static ChooseFastestDataset create(Scope scope, Iterable<Operand<?>> inputDatasets, Long numExperiments, List<Class<?>> outputTypes, List<Shape> outputShapes) {
     OperationBuilder opBuilder = scope.env().opBuilder("ChooseFastestDataset", scope.makeOpName("ChooseFastestDataset"));
     opBuilder.addInputList(Operands.asOutputs(scope, inputDatasets));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("num_experiments", numExperiments);
-    DataType[] outputTypesArray = new DataType[outputTypes.size()];
+    Class[] outputTypesArray = new Class[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
       outputTypesArray[i] = outputTypes.get(i);
     }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ConcatenateDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ConcatenateDataset.java
index 7eb4581f462..5c566d50f0b 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ConcatenateDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ConcatenateDataset.java
@@ -18,7 +18,6 @@
 package org.tensorflow.op.data;
 
 import java.util.List;
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -47,12 +46,12 @@ public final class ConcatenateDataset extends RawOp implements Operand<TType> {
    * @return a new instance of ConcatenateDataset
    */
   @Endpoint(describeByClass = true)
-  public static ConcatenateDataset create(Scope scope, Operand<?> inputDataset, Operand<?> anotherDataset, List<DataType<?>> outputTypes, List<Shape> outputShapes) {
+  public static ConcatenateDataset create(Scope scope, Operand<?> inputDataset, Operand<?> anotherDataset, List<Class<?>> outputTypes, List<Shape> outputShapes) {
     OperationBuilder opBuilder = scope.env().opBuilder("ConcatenateDataset", scope.makeOpName("ConcatenateDataset"));
     opBuilder.addInput(inputDataset.asOutput(scope));
     opBuilder.addInput(anotherDataset.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
-    DataType[] outputTypesArray = new DataType[outputTypes.size()];
+    Class[] outputTypesArray = new Class[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
       outputTypesArray[i] = outputTypes.get(i);
     }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DatasetToSingleElement.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DatasetToSingleElement.java
index 69c4a717eb4..ed49fe284d2 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DatasetToSingleElement.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DatasetToSingleElement.java
@@ -20,7 +20,6 @@
 import java.util.Arrays;
 import java.util.Iterator;
 import java.util.List;
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -47,11 +46,11 @@ public final class DatasetToSingleElement extends RawOp implements Iterable<Oper
    * @return a new instance of DatasetToSingleElement
    */
   @Endpoint(describeByClass = true)
-  public static DatasetToSingleElement create(Scope scope, Operand<?> dataset, List<DataType<?>> outputTypes, List<Shape> outputShapes) {
+  public static DatasetToSingleElement create(Scope scope, Operand<?> dataset, List<Class<?>> outputTypes, List<Shape> outputShapes) {
     OperationBuilder opBuilder = scope.env().opBuilder("DatasetToSingleElement", scope.makeOpName("DatasetToSingleElement"));
     opBuilder.addInput(dataset.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
-    DataType[] outputTypesArray = new DataType[outputTypes.size()];
+    Class[] outputTypesArray = new Class[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
       outputTypesArray[i] = outputTypes.get(i);
     }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DenseToSparseBatchDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DenseToSparseBatchDataset.java
index 2f89979fa98..a8566b736d5 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DenseToSparseBatchDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DenseToSparseBatchDataset.java
@@ -18,7 +18,6 @@
 package org.tensorflow.op.data;
 
 import java.util.List;
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -51,13 +50,13 @@ public final class DenseToSparseBatchDataset extends RawOp implements Operand<TT
    * @return a new instance of DenseToSparseBatchDataset
    */
   @Endpoint(describeByClass = true)
-  public static DenseToSparseBatchDataset create(Scope scope, Operand<?> inputDataset, Operand<TInt64> batchSize, Operand<TInt64> rowShape, List<DataType<?>> outputTypes, List<Shape> outputShapes) {
+  public static DenseToSparseBatchDataset create(Scope scope, Operand<?> inputDataset, Operand<TInt64> batchSize, Operand<TInt64> rowShape, List<Class<?>> outputTypes, List<Shape> outputShapes) {
     OperationBuilder opBuilder = scope.env().opBuilder("DenseToSparseBatchDataset", scope.makeOpName("DenseToSparseBatchDataset"));
     opBuilder.addInput(inputDataset.asOutput(scope));
     opBuilder.addInput(batchSize.asOutput(scope));
     opBuilder.addInput(rowShape.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
-    DataType[] outputTypesArray = new DataType[outputTypes.size()];
+    Class[] outputTypesArray = new Class[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
       outputTypesArray[i] = outputTypes.get(i);
     }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DirectedInterleaveDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DirectedInterleaveDataset.java
index 9fff9e7e6e7..eb196480822 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DirectedInterleaveDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DirectedInterleaveDataset.java
@@ -18,7 +18,6 @@
 package org.tensorflow.op.data;
 
 import java.util.List;
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -49,12 +48,12 @@ public final class DirectedInterleaveDataset extends RawOp implements Operand<TT
    * @return a new instance of DirectedInterleaveDataset
    */
   @Endpoint(describeByClass = true)
-  public static DirectedInterleaveDataset create(Scope scope, Operand<?> selectorInputDataset, Iterable<Operand<?>> dataInputDatasets, List<DataType<?>> outputTypes, List<Shape> outputShapes) {
+  public static DirectedInterleaveDataset create(Scope scope, Operand<?> selectorInputDataset, Iterable<Operand<?>> dataInputDatasets, List<Class<?>> outputTypes, List<Shape> outputShapes) {
     OperationBuilder opBuilder = scope.env().opBuilder("DirectedInterleaveDataset", scope.makeOpName("DirectedInterleaveDataset"));
     opBuilder.addInput(selectorInputDataset.asOutput(scope));
     opBuilder.addInputList(Operands.asOutputs(scope, dataInputDatasets));
     opBuilder = scope.applyControlDependencies(opBuilder);
-    DataType[] outputTypesArray = new DataType[outputTypes.size()];
+    Class[] outputTypesArray = new Class[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
       outputTypesArray[i] = outputTypes.get(i);
     }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/FilterByLastComponentDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/FilterByLastComponentDataset.java
index 7d1421f46da..618c6b07d2e 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/FilterByLastComponentDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/FilterByLastComponentDataset.java
@@ -18,7 +18,6 @@
 package org.tensorflow.op.data;
 
 import java.util.List;
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -45,11 +44,11 @@ public final class FilterByLastComponentDataset extends RawOp implements Operand
    * @return a new instance of FilterByLastComponentDataset
    */
   @Endpoint(describeByClass = true)
-  public static FilterByLastComponentDataset create(Scope scope, Operand<?> inputDataset, List<DataType<?>> outputTypes, List<Shape> outputShapes) {
+  public static FilterByLastComponentDataset create(Scope scope, Operand<?> inputDataset, List<Class<?>> outputTypes, List<Shape> outputShapes) {
     OperationBuilder opBuilder = scope.env().opBuilder("FilterByLastComponentDataset", scope.makeOpName("FilterByLastComponentDataset"));
     opBuilder.addInput(inputDataset.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
-    DataType[] outputTypesArray = new DataType[outputTypes.size()];
+    Class[] outputTypesArray = new Class[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
       outputTypesArray[i] = outputTypes.get(i);
     }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/IgnoreErrorsDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/IgnoreErrorsDataset.java
index b58693e2031..dbc01878164 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/IgnoreErrorsDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/IgnoreErrorsDataset.java
@@ -18,7 +18,6 @@
 package org.tensorflow.op.data;
 
 import java.util.List;
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -45,11 +44,11 @@ public final class IgnoreErrorsDataset extends RawOp implements Operand<TType> {
    * @return a new instance of IgnoreErrorsDataset
    */
   @Endpoint(describeByClass = true)
-  public static IgnoreErrorsDataset create(Scope scope, Operand<?> inputDataset, List<DataType<?>> outputTypes, List<Shape> outputShapes) {
+  public static IgnoreErrorsDataset create(Scope scope, Operand<?> inputDataset, List<Class<?>> outputTypes, List<Shape> outputShapes) {
     OperationBuilder opBuilder = scope.env().opBuilder("IgnoreErrorsDataset", scope.makeOpName("IgnoreErrorsDataset"));
     opBuilder.addInput(inputDataset.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
-    DataType[] outputTypesArray = new DataType[outputTypes.size()];
+    Class[] outputTypesArray = new Class[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
       outputTypesArray[i] = outputTypes.get(i);
     }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/InitializeTableFromDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/InitializeTableFromDataset.java
index 05a263a4ec1..5a1d007d572 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/InitializeTableFromDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/InitializeTableFromDataset.java
@@ -40,8 +40,8 @@ public final class InitializeTableFromDataset extends RawOp {
   @Endpoint(describeByClass = true)
   public static InitializeTableFromDataset create(Scope scope, Operand<?> tableHandle, Operand<?> dataset) {
     OperationBuilder opBuilder = scope.env().opBuilder("InitializeTableFromDataset", scope.makeOpName("InitializeTableFromDataset"));
-    opBuilder.addInput(tableHandle.asOutput());
-    opBuilder.addInput(dataset.asOutput());
+    opBuilder.addInput(tableHandle.asOutput(scope));
+    opBuilder.addInput(dataset.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new InitializeTableFromDataset(opBuilder.build());
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/Iterator.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/Iterator.java
index d3c03cd40b0..427120ded4d 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/Iterator.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/Iterator.java
@@ -18,7 +18,6 @@
 package org.tensorflow.op.data;
 
 import java.util.List;
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -46,12 +45,12 @@ public final class Iterator extends RawOp implements Operand<TType> {
    * @return a new instance of Iterator
    */
   @Endpoint(describeByClass = true)
-  public static Iterator create(Scope scope, String sharedName, String container, List<DataType<?>> outputTypes, List<Shape> outputShapes) {
+  public static Iterator create(Scope scope, String sharedName, String container, List<Class<?>> outputTypes, List<Shape> outputShapes) {
     OperationBuilder opBuilder = scope.env().opBuilder("IteratorV2", scope.makeOpName("Iterator"));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("shared_name", sharedName);
     opBuilder.setAttr("container", container);
-    DataType[] outputTypesArray = new DataType[outputTypes.size()];
+    Class[] outputTypesArray = new Class[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
       outputTypesArray[i] = outputTypes.get(i);
     }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/IteratorFromStringHandle.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/IteratorFromStringHandle.java
index c06bcd607f1..c36d2ea804d 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/IteratorFromStringHandle.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/IteratorFromStringHandle.java
@@ -18,7 +18,6 @@
 package org.tensorflow.op.data;
 
 import java.util.List;
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -64,11 +63,11 @@ private Options() {
    * @return a new instance of IteratorFromStringHandle
    */
   @Endpoint(describeByClass = true)
-  public static IteratorFromStringHandle create(Scope scope, Operand<TString> stringHandle, List<DataType<?>> outputTypes, Options... options) {
+  public static IteratorFromStringHandle create(Scope scope, Operand<TString> stringHandle, List<Class<?>> outputTypes, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("IteratorFromStringHandleV2", scope.makeOpName("IteratorFromStringHandle"));
     opBuilder.addInput(stringHandle.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
-    DataType[] outputTypesArray = new DataType[outputTypes.size()];
+    Class[] outputTypesArray = new Class[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
       outputTypesArray[i] = outputTypes.get(i);
     }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/IteratorGetNext.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/IteratorGetNext.java
index 97c55d78184..fb71fc18c97 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/IteratorGetNext.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/IteratorGetNext.java
@@ -20,7 +20,6 @@
 import java.util.Arrays;
 import java.util.Iterator;
 import java.util.List;
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -48,11 +47,11 @@ public final class IteratorGetNext extends RawOp implements Iterable<Operand<TTy
    * @return a new instance of IteratorGetNext
    */
   @Endpoint(describeByClass = true)
-  public static IteratorGetNext create(Scope scope, Operand<?> iterator, List<DataType<?>> outputTypes, List<Shape> outputShapes) {
+  public static IteratorGetNext create(Scope scope, Operand<?> iterator, List<Class<?>> outputTypes, List<Shape> outputShapes) {
     OperationBuilder opBuilder = scope.env().opBuilder("IteratorGetNext", scope.makeOpName("IteratorGetNext"));
     opBuilder.addInput(iterator.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
-    DataType[] outputTypesArray = new DataType[outputTypes.size()];
+    Class[] outputTypesArray = new Class[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
       outputTypesArray[i] = outputTypes.get(i);
     }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/IteratorGetNextAsOptional.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/IteratorGetNextAsOptional.java
index 1243a53e435..70e0a91c8e2 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/IteratorGetNextAsOptional.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/IteratorGetNextAsOptional.java
@@ -18,7 +18,6 @@
 package org.tensorflow.op.data;
 
 import java.util.List;
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -46,11 +45,11 @@ public final class IteratorGetNextAsOptional extends RawOp implements Operand<TT
    * @return a new instance of IteratorGetNextAsOptional
    */
   @Endpoint(describeByClass = true)
-  public static IteratorGetNextAsOptional create(Scope scope, Operand<?> iterator, List<DataType<?>> outputTypes, List<Shape> outputShapes) {
+  public static IteratorGetNextAsOptional create(Scope scope, Operand<?> iterator, List<Class<?>> outputTypes, List<Shape> outputShapes) {
     OperationBuilder opBuilder = scope.env().opBuilder("IteratorGetNextAsOptional", scope.makeOpName("IteratorGetNextAsOptional"));
     opBuilder.addInput(iterator.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
-    DataType[] outputTypesArray = new DataType[outputTypes.size()];
+    Class[] outputTypesArray = new Class[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
       outputTypesArray[i] = outputTypes.get(i);
     }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/IteratorGetNextSync.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/IteratorGetNextSync.java
index 84670f3fc5b..b211a5b1d48 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/IteratorGetNextSync.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/IteratorGetNextSync.java
@@ -20,7 +20,6 @@
 import java.util.Arrays;
 import java.util.Iterator;
 import java.util.List;
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -53,11 +52,11 @@ public final class IteratorGetNextSync extends RawOp implements Iterable<Operand
    * @return a new instance of IteratorGetNextSync
    */
   @Endpoint(describeByClass = true)
-  public static IteratorGetNextSync create(Scope scope, Operand<?> iterator, List<DataType<?>> outputTypes, List<Shape> outputShapes) {
+  public static IteratorGetNextSync create(Scope scope, Operand<?> iterator, List<Class<?>> outputTypes, List<Shape> outputShapes) {
     OperationBuilder opBuilder = scope.env().opBuilder("IteratorGetNextSync", scope.makeOpName("IteratorGetNextSync"));
     opBuilder.addInput(iterator.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
-    DataType[] outputTypesArray = new DataType[outputTypes.size()];
+    Class[] outputTypesArray = new Class[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
       outputTypesArray[i] = outputTypes.get(i);
     }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/LMDBDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/LMDBDataset.java
index 208785fad35..51433bfe9cf 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/LMDBDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/LMDBDataset.java
@@ -18,7 +18,6 @@
 package org.tensorflow.op.data;
 
 import java.util.List;
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -58,11 +57,11 @@ public final class LMDBDataset extends RawOp implements Operand<TType> {
    * @return a new instance of LMDBDataset
    */
   @Endpoint(describeByClass = true)
-  public static LMDBDataset create(Scope scope, Operand<TString> filenames, List<DataType<?>> outputTypes, List<Shape> outputShapes) {
+  public static LMDBDataset create(Scope scope, Operand<TString> filenames, List<Class<?>> outputTypes, List<Shape> outputShapes) {
     OperationBuilder opBuilder = scope.env().opBuilder("LMDBDataset", scope.makeOpName("LMDBDataset"));
     opBuilder.addInput(filenames.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
-    DataType[] outputTypesArray = new DataType[outputTypes.size()];
+    Class[] outputTypesArray = new Class[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
       outputTypesArray[i] = outputTypes.get(i);
     }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/LatencyStatsDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/LatencyStatsDataset.java
index 1a4af719e32..2f4ec7d69d7 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/LatencyStatsDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/LatencyStatsDataset.java
@@ -18,7 +18,6 @@
 package org.tensorflow.op.data;
 
 import java.util.List;
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -47,12 +46,12 @@ public final class LatencyStatsDataset extends RawOp implements Operand<TType> {
    * @return a new instance of LatencyStatsDataset
    */
   @Endpoint(describeByClass = true)
-  public static LatencyStatsDataset create(Scope scope, Operand<?> inputDataset, Operand<TString> tag, List<DataType<?>> outputTypes, List<Shape> outputShapes) {
+  public static LatencyStatsDataset create(Scope scope, Operand<?> inputDataset, Operand<TString> tag, List<Class<?>> outputTypes, List<Shape> outputShapes) {
     OperationBuilder opBuilder = scope.env().opBuilder("LatencyStatsDataset", scope.makeOpName("LatencyStatsDataset"));
     opBuilder.addInput(inputDataset.asOutput(scope));
     opBuilder.addInput(tag.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
-    DataType[] outputTypesArray = new DataType[outputTypes.size()];
+    Class[] outputTypesArray = new Class[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
       outputTypesArray[i] = outputTypes.get(i);
     }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MaxIntraOpParallelismDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MaxIntraOpParallelismDataset.java
index 4d365c94ed3..71ae5c76dde 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MaxIntraOpParallelismDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MaxIntraOpParallelismDataset.java
@@ -18,7 +18,6 @@
 package org.tensorflow.op.data;
 
 import java.util.List;
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -47,12 +46,12 @@ public final class MaxIntraOpParallelismDataset extends RawOp implements Operand
    * @return a new instance of MaxIntraOpParallelismDataset
    */
   @Endpoint(describeByClass = true)
-  public static MaxIntraOpParallelismDataset create(Scope scope, Operand<?> inputDataset, Operand<TInt64> maxIntraOpParallelism, List<DataType<?>> outputTypes, List<Shape> outputShapes) {
+  public static MaxIntraOpParallelismDataset create(Scope scope, Operand<?> inputDataset, Operand<TInt64> maxIntraOpParallelism, List<Class<?>> outputTypes, List<Shape> outputShapes) {
     OperationBuilder opBuilder = scope.env().opBuilder("MaxIntraOpParallelismDataset", scope.makeOpName("MaxIntraOpParallelismDataset"));
     opBuilder.addInput(inputDataset.asOutput(scope));
     opBuilder.addInput(maxIntraOpParallelism.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
-    DataType[] outputTypesArray = new DataType[outputTypes.size()];
+    Class[] outputTypesArray = new Class[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
       outputTypesArray[i] = outputTypes.get(i);
     }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ModelDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ModelDataset.java
index 9fed6fcaa30..837a9ff4b46 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ModelDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ModelDataset.java
@@ -18,7 +18,6 @@
 package org.tensorflow.op.data;
 
 import java.util.List;
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -76,11 +75,11 @@ private Options() {
    * @return a new instance of ModelDataset
    */
   @Endpoint(describeByClass = true)
-  public static ModelDataset create(Scope scope, Operand<?> inputDataset, List<DataType<?>> outputTypes, List<Shape> outputShapes, Options... options) {
+  public static ModelDataset create(Scope scope, Operand<?> inputDataset, List<Class<?>> outputTypes, List<Shape> outputShapes, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ModelDataset", scope.makeOpName("ModelDataset"));
     opBuilder.addInput(inputDataset.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
-    DataType[] outputTypesArray = new DataType[outputTypes.size()];
+    Class[] outputTypesArray = new Class[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
       outputTypesArray[i] = outputTypes.get(i);
     }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MultiDeviceIterator.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MultiDeviceIterator.java
index 1ac9874ab6a..5937c97a388 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MultiDeviceIterator.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MultiDeviceIterator.java
@@ -18,7 +18,6 @@
 package org.tensorflow.op.data;
 
 import java.util.List;
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -49,7 +48,7 @@ public final class MultiDeviceIterator extends RawOp implements Operand<TType> {
    * @return a new instance of MultiDeviceIterator
    */
   @Endpoint(describeByClass = true)
-  public static MultiDeviceIterator create(Scope scope, List<String> devices, String sharedName, String container, List<DataType<?>> outputTypes, List<Shape> outputShapes) {
+  public static MultiDeviceIterator create(Scope scope, List<String> devices, String sharedName, String container, List<Class<?>> outputTypes, List<Shape> outputShapes) {
     OperationBuilder opBuilder = scope.env().opBuilder("MultiDeviceIterator", scope.makeOpName("MultiDeviceIterator"));
     opBuilder = scope.applyControlDependencies(opBuilder);
     String[] devicesArray = new String[devices.size()];
@@ -59,7 +58,7 @@ public static MultiDeviceIterator create(Scope scope, List<String> devices, Stri
     opBuilder.setAttr("devices", devicesArray);
     opBuilder.setAttr("shared_name", sharedName);
     opBuilder.setAttr("container", container);
-    DataType[] outputTypesArray = new DataType[outputTypes.size()];
+    Class[] outputTypesArray = new Class[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
       outputTypesArray[i] = outputTypes.get(i);
     }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MultiDeviceIteratorFromStringHandle.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MultiDeviceIteratorFromStringHandle.java
index 61723160f2b..0c1ecab01d1 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MultiDeviceIteratorFromStringHandle.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MultiDeviceIteratorFromStringHandle.java
@@ -18,7 +18,6 @@
 package org.tensorflow.op.data;
 
 import java.util.List;
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -65,11 +64,11 @@ private Options() {
    * @return a new instance of MultiDeviceIteratorFromStringHandle
    */
   @Endpoint(describeByClass = true)
-  public static MultiDeviceIteratorFromStringHandle create(Scope scope, Operand<TString> stringHandle, List<DataType<?>> outputTypes, Options... options) {
+  public static MultiDeviceIteratorFromStringHandle create(Scope scope, Operand<TString> stringHandle, List<Class<?>> outputTypes, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("MultiDeviceIteratorFromStringHandle", scope.makeOpName("MultiDeviceIteratorFromStringHandle"));
     opBuilder.addInput(stringHandle.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
-    DataType[] outputTypesArray = new DataType[outputTypes.size()];
+    Class[] outputTypesArray = new Class[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
       outputTypesArray[i] = outputTypes.get(i);
     }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MultiDeviceIteratorGetNextFromShard.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MultiDeviceIteratorGetNextFromShard.java
index d08b76a7305..2156bab1b17 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MultiDeviceIteratorGetNextFromShard.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MultiDeviceIteratorGetNextFromShard.java
@@ -20,7 +20,6 @@
 import java.util.Arrays;
 import java.util.Iterator;
 import java.util.List;
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -51,13 +50,13 @@ public final class MultiDeviceIteratorGetNextFromShard extends RawOp implements
    * @return a new instance of MultiDeviceIteratorGetNextFromShard
    */
   @Endpoint(describeByClass = true)
-  public static MultiDeviceIteratorGetNextFromShard create(Scope scope, Operand<?> multiDeviceIterator, Operand<TInt32> shardNum, Operand<TInt64> incarnationId, List<DataType<?>> outputTypes, List<Shape> outputShapes) {
+  public static MultiDeviceIteratorGetNextFromShard create(Scope scope, Operand<?> multiDeviceIterator, Operand<TInt32> shardNum, Operand<TInt64> incarnationId, List<Class<?>> outputTypes, List<Shape> outputShapes) {
     OperationBuilder opBuilder = scope.env().opBuilder("MultiDeviceIteratorGetNextFromShard", scope.makeOpName("MultiDeviceIteratorGetNextFromShard"));
     opBuilder.addInput(multiDeviceIterator.asOutput(scope));
     opBuilder.addInput(shardNum.asOutput(scope));
     opBuilder.addInput(incarnationId.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
-    DataType[] outputTypesArray = new DataType[outputTypes.size()];
+    Class[] outputTypesArray = new Class[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
       outputTypesArray[i] = outputTypes.get(i);
     }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/NonSerializableDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/NonSerializableDataset.java
index fb0c21e115e..f1be418af20 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/NonSerializableDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/NonSerializableDataset.java
@@ -18,7 +18,6 @@
 package org.tensorflow.op.data;
 
 import java.util.List;
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -44,11 +43,11 @@ public final class NonSerializableDataset extends RawOp implements Operand<TType
    * @return a new instance of NonSerializableDataset
    */
   @Endpoint(describeByClass = true)
-  public static NonSerializableDataset create(Scope scope, Operand<?> inputDataset, List<DataType<?>> outputTypes, List<Shape> outputShapes) {
+  public static NonSerializableDataset create(Scope scope, Operand<?> inputDataset, List<Class<?>> outputTypes, List<Shape> outputShapes) {
     OperationBuilder opBuilder = scope.env().opBuilder("NonSerializableDataset", scope.makeOpName("NonSerializableDataset"));
     opBuilder.addInput(inputDataset.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
-    DataType[] outputTypesArray = new DataType[outputTypes.size()];
+    Class[] outputTypesArray = new Class[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
       outputTypesArray[i] = outputTypes.get(i);
     }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/OptimizeDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/OptimizeDataset.java
index d82141b019b..41d6bf9d6c1 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/OptimizeDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/OptimizeDataset.java
@@ -18,7 +18,6 @@
 package org.tensorflow.op.data;
 
 import java.util.List;
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -69,12 +68,12 @@ private Options() {
    * @return a new instance of OptimizeDataset
    */
   @Endpoint(describeByClass = true)
-  public static OptimizeDataset create(Scope scope, Operand<?> inputDataset, Operand<TString> optimizations, List<DataType<?>> outputTypes, List<Shape> outputShapes, Options... options) {
+  public static OptimizeDataset create(Scope scope, Operand<?> inputDataset, Operand<TString> optimizations, List<Class<?>> outputTypes, List<Shape> outputShapes, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("OptimizeDataset", scope.makeOpName("OptimizeDataset"));
     opBuilder.addInput(inputDataset.asOutput(scope));
     opBuilder.addInput(optimizations.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
-    DataType[] outputTypesArray = new DataType[outputTypes.size()];
+    Class[] outputTypesArray = new Class[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
       outputTypesArray[i] = outputTypes.get(i);
     }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/OptionalGetValue.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/OptionalGetValue.java
index a6ec01b632c..f3499e4a099 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/OptionalGetValue.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/OptionalGetValue.java
@@ -20,7 +20,6 @@
 import java.util.Arrays;
 import java.util.Iterator;
 import java.util.List;
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -48,11 +47,11 @@ public final class OptionalGetValue extends RawOp implements Iterable<Operand<TT
    * @return a new instance of OptionalGetValue
    */
   @Endpoint(describeByClass = true)
-  public static OptionalGetValue create(Scope scope, Operand<?> optional, List<DataType<?>> outputTypes, List<Shape> outputShapes) {
+  public static OptionalGetValue create(Scope scope, Operand<?> optional, List<Class<?>> outputTypes, List<Shape> outputShapes) {
     OperationBuilder opBuilder = scope.env().opBuilder("OptionalGetValue", scope.makeOpName("OptionalGetValue"));
     opBuilder.addInput(optional.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
-    DataType[] outputTypesArray = new DataType[outputTypes.size()];
+    Class[] outputTypesArray = new Class[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
       outputTypesArray[i] = outputTypes.get(i);
     }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/PrefetchDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/PrefetchDataset.java
index fd8a0cc8260..c8d61ba16f3 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/PrefetchDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/PrefetchDataset.java
@@ -18,7 +18,6 @@
 package org.tensorflow.op.data;
 
 import java.util.List;
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -77,12 +76,12 @@ private Options() {
    * @return a new instance of PrefetchDataset
    */
   @Endpoint(describeByClass = true)
-  public static PrefetchDataset create(Scope scope, Operand<?> inputDataset, Operand<TInt64> bufferSize, List<DataType<?>> outputTypes, List<Shape> outputShapes, Options... options) {
+  public static PrefetchDataset create(Scope scope, Operand<?> inputDataset, Operand<TInt64> bufferSize, List<Class<?>> outputTypes, List<Shape> outputShapes, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("PrefetchDataset", scope.makeOpName("PrefetchDataset"));
     opBuilder.addInput(inputDataset.asOutput(scope));
     opBuilder.addInput(bufferSize.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
-    DataType[] outputTypesArray = new DataType[outputTypes.size()];
+    Class[] outputTypesArray = new Class[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
       outputTypesArray[i] = outputTypes.get(i);
     }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/PrivateThreadPoolDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/PrivateThreadPoolDataset.java
index 30a9fce5626..22d669b94b5 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/PrivateThreadPoolDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/PrivateThreadPoolDataset.java
@@ -18,7 +18,6 @@
 package org.tensorflow.op.data;
 
 import java.util.List;
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -47,12 +46,12 @@ public final class PrivateThreadPoolDataset extends RawOp implements Operand<TTy
    * @return a new instance of PrivateThreadPoolDataset
    */
   @Endpoint(describeByClass = true)
-  public static PrivateThreadPoolDataset create(Scope scope, Operand<?> inputDataset, Operand<TInt64> numThreads, List<DataType<?>> outputTypes, List<Shape> outputShapes) {
+  public static PrivateThreadPoolDataset create(Scope scope, Operand<?> inputDataset, Operand<TInt64> numThreads, List<Class<?>> outputTypes, List<Shape> outputShapes) {
     OperationBuilder opBuilder = scope.env().opBuilder("PrivateThreadPoolDataset", scope.makeOpName("PrivateThreadPoolDataset"));
     opBuilder.addInput(inputDataset.asOutput(scope));
     opBuilder.addInput(numThreads.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
-    DataType[] outputTypesArray = new DataType[outputTypes.size()];
+    Class[] outputTypesArray = new Class[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
       outputTypesArray[i] = outputTypes.get(i);
     }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/RandomDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/RandomDataset.java
index 749a517522e..0de67cb18f0 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/RandomDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/RandomDataset.java
@@ -18,7 +18,6 @@
 package org.tensorflow.op.data;
 
 import java.util.List;
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -60,12 +59,12 @@ public final class RandomDataset extends RawOp implements Operand<TType> {
    * @return a new instance of RandomDataset
    */
   @Endpoint(describeByClass = true)
-  public static RandomDataset create(Scope scope, Operand<TInt64> seed, Operand<TInt64> seed2, List<DataType<?>> outputTypes, List<Shape> outputShapes) {
+  public static RandomDataset create(Scope scope, Operand<TInt64> seed, Operand<TInt64> seed2, List<Class<?>> outputTypes, List<Shape> outputShapes) {
     OperationBuilder opBuilder = scope.env().opBuilder("RandomDataset", scope.makeOpName("RandomDataset"));
     opBuilder.addInput(seed.asOutput(scope));
     opBuilder.addInput(seed2.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
-    DataType[] outputTypesArray = new DataType[outputTypes.size()];
+    Class[] outputTypesArray = new Class[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
       outputTypesArray[i] = outputTypes.get(i);
     }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/RangeDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/RangeDataset.java
index 6c45680f1a5..6133f124725 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/RangeDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/RangeDataset.java
@@ -18,7 +18,6 @@
 package org.tensorflow.op.data;
 
 import java.util.List;
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -49,13 +48,13 @@ public final class RangeDataset extends RawOp implements Operand<TType> {
    * @return a new instance of RangeDataset
    */
   @Endpoint(describeByClass = true)
-  public static RangeDataset create(Scope scope, Operand<TInt64> start, Operand<TInt64> stop, Operand<TInt64> step, List<DataType<?>> outputTypes, List<Shape> outputShapes) {
+  public static RangeDataset create(Scope scope, Operand<TInt64> start, Operand<TInt64> stop, Operand<TInt64> step, List<Class<?>> outputTypes, List<Shape> outputShapes) {
     OperationBuilder opBuilder = scope.env().opBuilder("RangeDataset", scope.makeOpName("RangeDataset"));
     opBuilder.addInput(start.asOutput(scope));
     opBuilder.addInput(stop.asOutput(scope));
     opBuilder.addInput(step.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
-    DataType[] outputTypesArray = new DataType[outputTypes.size()];
+    Class[] outputTypesArray = new Class[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
       outputTypesArray[i] = outputTypes.get(i);
     }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/RebatchDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/RebatchDataset.java
index 2b741f37f7f..5340276fb24 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/RebatchDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/RebatchDataset.java
@@ -18,7 +18,6 @@
 package org.tensorflow.op.data;
 
 import java.util.List;
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -72,12 +71,12 @@ private Options() {
    * @return a new instance of RebatchDataset
    */
   @Endpoint(describeByClass = true)
-  public static RebatchDataset create(Scope scope, Operand<?> inputDataset, Operand<TInt64> numReplicas, List<DataType<?>> outputTypes, List<Shape> outputShapes, Options... options) {
+  public static RebatchDataset create(Scope scope, Operand<?> inputDataset, Operand<TInt64> numReplicas, List<Class<?>> outputTypes, List<Shape> outputShapes, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("RebatchDataset", scope.makeOpName("RebatchDataset"));
     opBuilder.addInput(inputDataset.asOutput(scope));
     opBuilder.addInput(numReplicas.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
-    DataType[] outputTypesArray = new DataType[outputTypes.size()];
+    Class[] outputTypesArray = new Class[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
       outputTypesArray[i] = outputTypes.get(i);
     }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/RegisterDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/RegisterDataset.java
index 5705413165c..fe07a1ef30c 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/RegisterDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/RegisterDataset.java
@@ -46,9 +46,9 @@ public final class RegisterDataset extends RawOp implements Operand<TInt64> {
   @Endpoint(describeByClass = true)
   public static RegisterDataset create(Scope scope, Operand<?> dataset, Operand<TString> address, Operand<TString> protocol, Long externalStatePolicy) {
     OperationBuilder opBuilder = scope.env().opBuilder("RegisterDataset", scope.makeOpName("RegisterDataset"));
-    opBuilder.addInput(dataset.asOutput());
-    opBuilder.addInput(address.asOutput());
-    opBuilder.addInput(protocol.asOutput());
+    opBuilder.addInput(dataset.asOutput(scope));
+    opBuilder.addInput(address.asOutput(scope));
+    opBuilder.addInput(protocol.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("external_state_policy", externalStatePolicy);
     return new RegisterDataset(opBuilder.build());
@@ -61,7 +61,7 @@ public Output<TInt64> datasetId() {
   }
   
   @Override
-  public Output<TInt64> asOutput() {
+  public Output<TInt64> asOutput(Scope scope) {
     return datasetId;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/RepeatDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/RepeatDataset.java
index e4e432f2a1b..4d5fffaf5fa 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/RepeatDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/RepeatDataset.java
@@ -18,7 +18,6 @@
 package org.tensorflow.op.data;
 
 import java.util.List;
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -49,12 +48,12 @@ public final class RepeatDataset extends RawOp implements Operand<TType> {
    * @return a new instance of RepeatDataset
    */
   @Endpoint(describeByClass = true)
-  public static RepeatDataset create(Scope scope, Operand<?> inputDataset, Operand<TInt64> count, List<DataType<?>> outputTypes, List<Shape> outputShapes) {
+  public static RepeatDataset create(Scope scope, Operand<?> inputDataset, Operand<TInt64> count, List<Class<?>> outputTypes, List<Shape> outputShapes) {
     OperationBuilder opBuilder = scope.env().opBuilder("RepeatDataset", scope.makeOpName("RepeatDataset"));
     opBuilder.addInput(inputDataset.asOutput(scope));
     opBuilder.addInput(count.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
-    DataType[] outputTypesArray = new DataType[outputTypes.size()];
+    Class[] outputTypesArray = new Class[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
       outputTypesArray[i] = outputTypes.get(i);
     }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SamplingDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SamplingDataset.java
index d535f6855b4..649645e6dd3 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SamplingDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SamplingDataset.java
@@ -18,7 +18,6 @@
 package org.tensorflow.op.data;
 
 import java.util.List;
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -57,14 +56,14 @@ public final class SamplingDataset extends RawOp implements Operand<TType> {
    * @return a new instance of SamplingDataset
    */
   @Endpoint(describeByClass = true)
-  public static SamplingDataset create(Scope scope, Operand<?> inputDataset, Operand<TFloat32> rate, Operand<TInt64> seed, Operand<TInt64> seed2, List<DataType<?>> outputTypes, List<Shape> outputShapes) {
+  public static SamplingDataset create(Scope scope, Operand<?> inputDataset, Operand<TFloat32> rate, Operand<TInt64> seed, Operand<TInt64> seed2, List<Class<?>> outputTypes, List<Shape> outputShapes) {
     OperationBuilder opBuilder = scope.env().opBuilder("SamplingDataset", scope.makeOpName("SamplingDataset"));
     opBuilder.addInput(inputDataset.asOutput(scope));
     opBuilder.addInput(rate.asOutput(scope));
     opBuilder.addInput(seed.asOutput(scope));
     opBuilder.addInput(seed2.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
-    DataType[] outputTypesArray = new DataType[outputTypes.size()];
+    Class[] outputTypesArray = new Class[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
       outputTypesArray[i] = outputTypes.get(i);
     }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SetStatsAggregatorDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SetStatsAggregatorDataset.java
index 6c123952bdc..69fd92189eb 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SetStatsAggregatorDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SetStatsAggregatorDataset.java
@@ -18,7 +18,6 @@
 package org.tensorflow.op.data;
 
 import java.util.List;
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -48,14 +47,14 @@ public final class SetStatsAggregatorDataset extends RawOp implements Operand<TT
    * @return a new instance of SetStatsAggregatorDataset
    */
   @Endpoint(describeByClass = true)
-  public static SetStatsAggregatorDataset create(Scope scope, Operand<?> inputDataset, Operand<?> statsAggregator, Operand<TString> tag, Operand<TString> counterPrefix, List<DataType<?>> outputTypes, List<Shape> outputShapes) {
+  public static SetStatsAggregatorDataset create(Scope scope, Operand<?> inputDataset, Operand<?> statsAggregator, Operand<TString> tag, Operand<TString> counterPrefix, List<Class<?>> outputTypes, List<Shape> outputShapes) {
     OperationBuilder opBuilder = scope.env().opBuilder("SetStatsAggregatorDataset", scope.makeOpName("SetStatsAggregatorDataset"));
     opBuilder.addInput(inputDataset.asOutput(scope));
     opBuilder.addInput(statsAggregator.asOutput(scope));
     opBuilder.addInput(tag.asOutput(scope));
     opBuilder.addInput(counterPrefix.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
-    DataType[] outputTypesArray = new DataType[outputTypes.size()];
+    Class[] outputTypesArray = new Class[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
       outputTypesArray[i] = outputTypes.get(i);
     }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ShardDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ShardDataset.java
index 7a77bd7eed7..6ac6547cc41 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ShardDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ShardDataset.java
@@ -18,7 +18,6 @@
 package org.tensorflow.op.data;
 
 import java.util.List;
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -68,13 +67,13 @@ private Options() {
    * @return a new instance of ShardDataset
    */
   @Endpoint(describeByClass = true)
-  public static ShardDataset create(Scope scope, Operand<?> inputDataset, Operand<TInt64> numShards, Operand<TInt64> index, List<DataType<?>> outputTypes, List<Shape> outputShapes, Options... options) {
+  public static ShardDataset create(Scope scope, Operand<?> inputDataset, Operand<TInt64> numShards, Operand<TInt64> index, List<Class<?>> outputTypes, List<Shape> outputShapes, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ShardDataset", scope.makeOpName("ShardDataset"));
     opBuilder.addInput(inputDataset.asOutput(scope));
     opBuilder.addInput(numShards.asOutput(scope));
     opBuilder.addInput(index.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
-    DataType[] outputTypesArray = new DataType[outputTypes.size()];
+    Class[] outputTypesArray = new Class[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
       outputTypesArray[i] = outputTypes.get(i);
     }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ShuffleAndRepeatDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ShuffleAndRepeatDataset.java
index 13a629bb3c8..2c9ab0f4218 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ShuffleAndRepeatDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ShuffleAndRepeatDataset.java
@@ -18,7 +18,6 @@
 package org.tensorflow.op.data;
 
 import java.util.List;
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -32,40 +31,54 @@
 import org.tensorflow.types.family.TType;
 
 /**
- * Creates a dataset that shuffles and repeats elements from `input_dataset`
- * <p>
- * pseudorandomly.
  */
 public final class ShuffleAndRepeatDataset extends RawOp implements Operand<TType> {
   
+  /**
+   * Optional attributes for {@link org.tensorflow.op.data.ShuffleAndRepeatDataset}
+   */
+  public static class Options {
+    
+    /**
+     * @param reshuffleEachIteration 
+     */
+    public Options reshuffleEachIteration(Boolean reshuffleEachIteration) {
+      this.reshuffleEachIteration = reshuffleEachIteration;
+      return this;
+    }
+    
+    private Boolean reshuffleEachIteration;
+    
+    private Options() {
+    }
+  }
+  
   /**
    * Factory method to create a class wrapping a new ShuffleAndRepeatDataset operation.
    * 
    * @param scope current scope
    * @param inputDataset 
-   * @param bufferSize The number of output elements to buffer in an iterator over
-   * this dataset. Compare with the `min_after_dequeue` attr when creating a
-   * `RandomShuffleQueue`.
-   * @param seed A scalar seed for the random number generator. If either `seed` or
-   * `seed2` is set to be non-zero, the random number generator is seeded
-   * by the given seed.  Otherwise, a random seed is used.
-   * @param seed2 A second scalar seed to avoid seed collision.
-   * @param count A scalar representing the number of times the underlying dataset
-   * should be repeated. The default is `-1`, which results in infinite repetition.
+   * @param bufferSize 
+   * @param seed 
+   * @param seed2 
+   * @param count 
+   * @param seedGenerator 
    * @param outputTypes 
    * @param outputShapes 
+   * @param options carries optional attributes values
    * @return a new instance of ShuffleAndRepeatDataset
    */
   @Endpoint(describeByClass = true)
-  public static ShuffleAndRepeatDataset create(Scope scope, Operand<?> inputDataset, Operand<TInt64> bufferSize, Operand<TInt64> seed, Operand<TInt64> seed2, Operand<TInt64> count, List<DataType<?>> outputTypes, List<Shape> outputShapes) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ShuffleAndRepeatDataset", scope.makeOpName("ShuffleAndRepeatDataset"));
+  public static ShuffleAndRepeatDataset create(Scope scope, Operand<?> inputDataset, Operand<TInt64> bufferSize, Operand<TInt64> seed, Operand<TInt64> seed2, Operand<TInt64> count, Operand<?> seedGenerator, List<Class<?>> outputTypes, List<Shape> outputShapes, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ShuffleAndRepeatDatasetV2", scope.makeOpName("ShuffleAndRepeatDataset"));
     opBuilder.addInput(inputDataset.asOutput(scope));
     opBuilder.addInput(bufferSize.asOutput(scope));
     opBuilder.addInput(seed.asOutput(scope));
     opBuilder.addInput(seed2.asOutput(scope));
     opBuilder.addInput(count.asOutput(scope));
+    opBuilder.addInput(seedGenerator.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
-    DataType[] outputTypesArray = new DataType[outputTypes.size()];
+    Class[] outputTypesArray = new Class[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
       outputTypesArray[i] = outputTypes.get(i);
     }
@@ -75,9 +88,23 @@ public static ShuffleAndRepeatDataset create(Scope scope, Operand<?> inputDatase
       outputShapesArray[i] = outputShapes.get(i);
     }
     opBuilder.setAttr("output_shapes", outputShapesArray);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.reshuffleEachIteration != null) {
+          opBuilder.setAttr("reshuffle_each_iteration", opts.reshuffleEachIteration);
+        }
+      }
+    }
     return new ShuffleAndRepeatDataset(opBuilder.build());
   }
   
+  /**
+   * @param reshuffleEachIteration 
+   */
+  public static Options reshuffleEachIteration(Boolean reshuffleEachIteration) {
+    return new Options().reshuffleEachIteration(reshuffleEachIteration);
+  }
+  
   /**
    */
   public Output<?> handle() {
@@ -90,6 +117,9 @@ public Output<TType> asOutput(Scope scope) {
     return (Output<TType>) handle;
   }
   
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ShuffleAndRepeatDatasetV2";
+  
   private Output<?> handle;
   
   private ShuffleAndRepeatDataset(Operation operation) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ShuffleDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ShuffleDataset.java
index 1401683e34f..018958cb2da 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ShuffleDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ShuffleDataset.java
@@ -18,7 +18,6 @@
 package org.tensorflow.op.data;
 
 import java.util.List;
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -35,25 +34,49 @@
  */
 public final class ShuffleDataset extends RawOp implements Operand<TType> {
   
+  /**
+   * Optional attributes for {@link org.tensorflow.op.data.ShuffleDataset}
+   */
+  public static class Options {
+    
+    /**
+     * @param reshuffleEachIteration 
+     */
+    public Options reshuffleEachIteration(Boolean reshuffleEachIteration) {
+      this.reshuffleEachIteration = reshuffleEachIteration;
+      return this;
+    }
+    
+    private Boolean reshuffleEachIteration;
+    
+    private Options() {
+    }
+  }
+  
   /**
    * Factory method to create a class wrapping a new ShuffleDataset operation.
    * 
    * @param scope current scope
    * @param inputDataset 
    * @param bufferSize 
+   * @param seed 
+   * @param seed2 
    * @param seedGenerator 
    * @param outputTypes 
    * @param outputShapes 
+   * @param options carries optional attributes values
    * @return a new instance of ShuffleDataset
    */
   @Endpoint(describeByClass = true)
-  public static ShuffleDataset create(Scope scope, Operand<?> inputDataset, Operand<TInt64> bufferSize, Operand<?> seedGenerator, List<DataType<?>> outputTypes, List<Shape> outputShapes) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ShuffleDatasetV2", scope.makeOpName("ShuffleDataset"));
+  public static ShuffleDataset create(Scope scope, Operand<?> inputDataset, Operand<TInt64> bufferSize, Operand<TInt64> seed, Operand<TInt64> seed2, Operand<?> seedGenerator, List<Class<?>> outputTypes, List<Shape> outputShapes, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ShuffleDatasetV3", scope.makeOpName("ShuffleDataset"));
     opBuilder.addInput(inputDataset.asOutput(scope));
     opBuilder.addInput(bufferSize.asOutput(scope));
+    opBuilder.addInput(seed.asOutput(scope));
+    opBuilder.addInput(seed2.asOutput(scope));
     opBuilder.addInput(seedGenerator.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
-    DataType[] outputTypesArray = new DataType[outputTypes.size()];
+    Class[] outputTypesArray = new Class[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
       outputTypesArray[i] = outputTypes.get(i);
     }
@@ -63,9 +86,23 @@ public static ShuffleDataset create(Scope scope, Operand<?> inputDataset, Operan
       outputShapesArray[i] = outputShapes.get(i);
     }
     opBuilder.setAttr("output_shapes", outputShapesArray);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.reshuffleEachIteration != null) {
+          opBuilder.setAttr("reshuffle_each_iteration", opts.reshuffleEachIteration);
+        }
+      }
+    }
     return new ShuffleDataset(opBuilder.build());
   }
   
+  /**
+   * @param reshuffleEachIteration 
+   */
+  public static Options reshuffleEachIteration(Boolean reshuffleEachIteration) {
+    return new Options().reshuffleEachIteration(reshuffleEachIteration);
+  }
+  
   /**
    */
   public Output<?> handle() {
@@ -78,6 +115,9 @@ public Output<TType> asOutput(Scope scope) {
     return (Output<TType>) handle;
   }
   
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ShuffleDatasetV3";
+  
   private Output<?> handle;
   
   private ShuffleDataset(Operation operation) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SkipDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SkipDataset.java
index 266bc4a9ce3..a106c523ec4 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SkipDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SkipDataset.java
@@ -18,7 +18,6 @@
 package org.tensorflow.op.data;
 
 import java.util.List;
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -49,12 +48,12 @@ public final class SkipDataset extends RawOp implements Operand<TType> {
    * @return a new instance of SkipDataset
    */
   @Endpoint(describeByClass = true)
-  public static SkipDataset create(Scope scope, Operand<?> inputDataset, Operand<TInt64> count, List<DataType<?>> outputTypes, List<Shape> outputShapes) {
+  public static SkipDataset create(Scope scope, Operand<?> inputDataset, Operand<TInt64> count, List<Class<?>> outputTypes, List<Shape> outputShapes) {
     OperationBuilder opBuilder = scope.env().opBuilder("SkipDataset", scope.makeOpName("SkipDataset"));
     opBuilder.addInput(inputDataset.asOutput(scope));
     opBuilder.addInput(count.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
-    DataType[] outputTypesArray = new DataType[outputTypes.size()];
+    Class[] outputTypesArray = new Class[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
       outputTypesArray[i] = outputTypes.get(i);
     }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SleepDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SleepDataset.java
index 965eece0409..3ced1cb7a24 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SleepDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SleepDataset.java
@@ -18,7 +18,6 @@
 package org.tensorflow.op.data;
 
 import java.util.List;
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -46,12 +45,12 @@ public final class SleepDataset extends RawOp implements Operand<TType> {
    * @return a new instance of SleepDataset
    */
   @Endpoint(describeByClass = true)
-  public static SleepDataset create(Scope scope, Operand<?> inputDataset, Operand<TInt64> sleepMicroseconds, List<DataType<?>> outputTypes, List<Shape> outputShapes) {
+  public static SleepDataset create(Scope scope, Operand<?> inputDataset, Operand<TInt64> sleepMicroseconds, List<Class<?>> outputTypes, List<Shape> outputShapes) {
     OperationBuilder opBuilder = scope.env().opBuilder("SleepDataset", scope.makeOpName("SleepDataset"));
     opBuilder.addInput(inputDataset.asOutput(scope));
     opBuilder.addInput(sleepMicroseconds.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
-    DataType[] outputTypesArray = new DataType[outputTypes.size()];
+    Class[] outputTypesArray = new Class[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
       outputTypesArray[i] = outputTypes.get(i);
     }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SlidingWindowDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SlidingWindowDataset.java
index 374166983b3..fd539ce0c11 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SlidingWindowDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SlidingWindowDataset.java
@@ -18,7 +18,6 @@
 package org.tensorflow.op.data;
 
 import java.util.List;
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -52,14 +51,14 @@ public final class SlidingWindowDataset extends RawOp implements Operand<TType>
    * @return a new instance of SlidingWindowDataset
    */
   @Endpoint(describeByClass = true)
-  public static SlidingWindowDataset create(Scope scope, Operand<?> inputDataset, Operand<TInt64> windowSize, Operand<TInt64> windowShift, Operand<TInt64> windowStride, List<DataType<?>> outputTypes, List<Shape> outputShapes) {
+  public static SlidingWindowDataset create(Scope scope, Operand<?> inputDataset, Operand<TInt64> windowSize, Operand<TInt64> windowShift, Operand<TInt64> windowStride, List<Class<?>> outputTypes, List<Shape> outputShapes) {
     OperationBuilder opBuilder = scope.env().opBuilder("SlidingWindowDataset", scope.makeOpName("SlidingWindowDataset"));
     opBuilder.addInput(inputDataset.asOutput(scope));
     opBuilder.addInput(windowSize.asOutput(scope));
     opBuilder.addInput(windowShift.asOutput(scope));
     opBuilder.addInput(windowStride.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
-    DataType[] outputTypesArray = new DataType[outputTypes.size()];
+    Class[] outputTypesArray = new Class[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
       outputTypesArray[i] = outputTypes.get(i);
     }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SnapshotDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SnapshotDataset.java
deleted file mode 100644
index 89b1246381b..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SnapshotDataset.java
+++ /dev/null
@@ -1,376 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.data;
-
-import java.util.List;
-import org.tensorflow.DataType;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TString;
-import org.tensorflow.types.family.TType;
-
-/**
- * Creates a dataset that will write to / read from a snapshot.
- * <p>
- * This dataset attempts to determine whether a valid snapshot exists at the
- * `snapshot_path`, and reads from the snapshot in lieu of using `input_dataset`.
- * If not, it will run the preprocessing pipeline as usual, and write out a
- * snapshot of the data processed for future use.
- */
-public final class SnapshotDataset extends RawOp implements Operand<TType> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.data.SnapshotDataset}
-   */
-  public static class Options {
-    
-    /**
-     * @param compression 
-     */
-    public Options compression(String compression) {
-      this.compression = compression;
-      return this;
-    }
-    
-    /**
-     * @param readerPathPrefix 
-     */
-    public Options readerPathPrefix(String readerPathPrefix) {
-      this.readerPathPrefix = readerPathPrefix;
-      return this;
-    }
-    
-    /**
-     * @param writerPathPrefix 
-     */
-    public Options writerPathPrefix(String writerPathPrefix) {
-      this.writerPathPrefix = writerPathPrefix;
-      return this;
-    }
-    
-    /**
-     * @param shardSizeBytes 
-     */
-    public Options shardSizeBytes(Long shardSizeBytes) {
-      this.shardSizeBytes = shardSizeBytes;
-      return this;
-    }
-    
-    /**
-     * @param pendingSnapshotExpirySeconds 
-     */
-    public Options pendingSnapshotExpirySeconds(Long pendingSnapshotExpirySeconds) {
-      this.pendingSnapshotExpirySeconds = pendingSnapshotExpirySeconds;
-      return this;
-    }
-    
-    /**
-     * @param numReaderThreads 
-     */
-    public Options numReaderThreads(Long numReaderThreads) {
-      this.numReaderThreads = numReaderThreads;
-      return this;
-    }
-    
-    /**
-     * @param readerBufferSize 
-     */
-    public Options readerBufferSize(Long readerBufferSize) {
-      this.readerBufferSize = readerBufferSize;
-      return this;
-    }
-    
-    /**
-     * @param numWriterThreads 
-     */
-    public Options numWriterThreads(Long numWriterThreads) {
-      this.numWriterThreads = numWriterThreads;
-      return this;
-    }
-    
-    /**
-     * @param writerBufferSize 
-     */
-    public Options writerBufferSize(Long writerBufferSize) {
-      this.writerBufferSize = writerBufferSize;
-      return this;
-    }
-    
-    /**
-     * @param shuffleOnRead 
-     */
-    public Options shuffleOnRead(Boolean shuffleOnRead) {
-      this.shuffleOnRead = shuffleOnRead;
-      return this;
-    }
-    
-    /**
-     * @param seed 
-     */
-    public Options seed(Long seed) {
-      this.seed = seed;
-      return this;
-    }
-    
-    /**
-     * @param seed2 
-     */
-    public Options seed2(Long seed2) {
-      this.seed2 = seed2;
-      return this;
-    }
-    
-    /**
-     * @param mode 
-     */
-    public Options mode(String mode) {
-      this.mode = mode;
-      return this;
-    }
-    
-    /**
-     * @param snapshotName 
-     */
-    public Options snapshotName(String snapshotName) {
-      this.snapshotName = snapshotName;
-      return this;
-    }
-    
-    private String compression;
-    private String readerPathPrefix;
-    private String writerPathPrefix;
-    private Long shardSizeBytes;
-    private Long pendingSnapshotExpirySeconds;
-    private Long numReaderThreads;
-    private Long readerBufferSize;
-    private Long numWriterThreads;
-    private Long writerBufferSize;
-    private Boolean shuffleOnRead;
-    private Long seed;
-    private Long seed2;
-    private String mode;
-    private String snapshotName;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new SnapshotDataset operation.
-   * 
-   * @param scope current scope
-   * @param inputDataset A variant tensor representing the input dataset.
-   * @param path The path we should write snapshots to / read snapshots from.
-   * @param outputTypes 
-   * @param outputShapes 
-   * @param options carries optional attributes values
-   * @return a new instance of SnapshotDataset
-   */
-  @Endpoint(describeByClass = true)
-  public static SnapshotDataset create(Scope scope, Operand<?> inputDataset, Operand<TString> path, List<DataType<?>> outputTypes, List<Shape> outputShapes, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("SnapshotDataset", scope.makeOpName("SnapshotDataset"));
-    opBuilder.addInput(inputDataset.asOutput(scope));
-    opBuilder.addInput(path.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    DataType[] outputTypesArray = new DataType[outputTypes.size()];
-    for (int i = 0; i < outputTypesArray.length; ++i) {
-      outputTypesArray[i] = outputTypes.get(i);
-    }
-    opBuilder.setAttr("output_types", outputTypesArray);
-    Shape[] outputShapesArray = new Shape[outputShapes.size()];
-    for (int i = 0; i < outputShapesArray.length; ++i) {
-      outputShapesArray[i] = outputShapes.get(i);
-    }
-    opBuilder.setAttr("output_shapes", outputShapesArray);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.compression != null) {
-          opBuilder.setAttr("compression", opts.compression);
-        }
-        if (opts.readerPathPrefix != null) {
-          opBuilder.setAttr("reader_path_prefix", opts.readerPathPrefix);
-        }
-        if (opts.writerPathPrefix != null) {
-          opBuilder.setAttr("writer_path_prefix", opts.writerPathPrefix);
-        }
-        if (opts.shardSizeBytes != null) {
-          opBuilder.setAttr("shard_size_bytes", opts.shardSizeBytes);
-        }
-        if (opts.pendingSnapshotExpirySeconds != null) {
-          opBuilder.setAttr("pending_snapshot_expiry_seconds", opts.pendingSnapshotExpirySeconds);
-        }
-        if (opts.numReaderThreads != null) {
-          opBuilder.setAttr("num_reader_threads", opts.numReaderThreads);
-        }
-        if (opts.readerBufferSize != null) {
-          opBuilder.setAttr("reader_buffer_size", opts.readerBufferSize);
-        }
-        if (opts.numWriterThreads != null) {
-          opBuilder.setAttr("num_writer_threads", opts.numWriterThreads);
-        }
-        if (opts.writerBufferSize != null) {
-          opBuilder.setAttr("writer_buffer_size", opts.writerBufferSize);
-        }
-        if (opts.shuffleOnRead != null) {
-          opBuilder.setAttr("shuffle_on_read", opts.shuffleOnRead);
-        }
-        if (opts.seed != null) {
-          opBuilder.setAttr("seed", opts.seed);
-        }
-        if (opts.seed2 != null) {
-          opBuilder.setAttr("seed2", opts.seed2);
-        }
-        if (opts.mode != null) {
-          opBuilder.setAttr("mode", opts.mode);
-        }
-        if (opts.snapshotName != null) {
-          opBuilder.setAttr("snapshot_name", opts.snapshotName);
-        }
-      }
-    }
-    return new SnapshotDataset(opBuilder.build());
-  }
-  
-  /**
-   * @param compression 
-   */
-  public static Options compression(String compression) {
-    return new Options().compression(compression);
-  }
-  
-  /**
-   * @param readerPathPrefix 
-   */
-  public static Options readerPathPrefix(String readerPathPrefix) {
-    return new Options().readerPathPrefix(readerPathPrefix);
-  }
-  
-  /**
-   * @param writerPathPrefix 
-   */
-  public static Options writerPathPrefix(String writerPathPrefix) {
-    return new Options().writerPathPrefix(writerPathPrefix);
-  }
-  
-  /**
-   * @param shardSizeBytes 
-   */
-  public static Options shardSizeBytes(Long shardSizeBytes) {
-    return new Options().shardSizeBytes(shardSizeBytes);
-  }
-  
-  /**
-   * @param pendingSnapshotExpirySeconds 
-   */
-  public static Options pendingSnapshotExpirySeconds(Long pendingSnapshotExpirySeconds) {
-    return new Options().pendingSnapshotExpirySeconds(pendingSnapshotExpirySeconds);
-  }
-  
-  /**
-   * @param numReaderThreads 
-   */
-  public static Options numReaderThreads(Long numReaderThreads) {
-    return new Options().numReaderThreads(numReaderThreads);
-  }
-  
-  /**
-   * @param readerBufferSize 
-   */
-  public static Options readerBufferSize(Long readerBufferSize) {
-    return new Options().readerBufferSize(readerBufferSize);
-  }
-  
-  /**
-   * @param numWriterThreads 
-   */
-  public static Options numWriterThreads(Long numWriterThreads) {
-    return new Options().numWriterThreads(numWriterThreads);
-  }
-  
-  /**
-   * @param writerBufferSize 
-   */
-  public static Options writerBufferSize(Long writerBufferSize) {
-    return new Options().writerBufferSize(writerBufferSize);
-  }
-  
-  /**
-   * @param shuffleOnRead 
-   */
-  public static Options shuffleOnRead(Boolean shuffleOnRead) {
-    return new Options().shuffleOnRead(shuffleOnRead);
-  }
-  
-  /**
-   * @param seed 
-   */
-  public static Options seed(Long seed) {
-    return new Options().seed(seed);
-  }
-  
-  /**
-   * @param seed2 
-   */
-  public static Options seed2(Long seed2) {
-    return new Options().seed2(seed2);
-  }
-  
-  /**
-   * @param mode 
-   */
-  public static Options mode(String mode) {
-    return new Options().mode(mode);
-  }
-  
-  /**
-   * @param snapshotName 
-   */
-  public static Options snapshotName(String snapshotName) {
-    return new Options().snapshotName(snapshotName);
-  }
-  
-  /**
-   */
-  public Output<?> handle() {
-    return handle;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) handle;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "SnapshotDataset";
-  
-  private Output<?> handle;
-  
-  private SnapshotDataset(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    handle = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SqlDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SqlDataset.java
index 68088e5353a..3817e9926f7 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SqlDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SqlDataset.java
@@ -18,7 +18,6 @@
 package org.tensorflow.op.data;
 
 import java.util.List;
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -48,13 +47,13 @@ public final class SqlDataset extends RawOp implements Operand<TType> {
    * @return a new instance of SqlDataset
    */
   @Endpoint(describeByClass = true)
-  public static SqlDataset create(Scope scope, Operand<TString> driverName, Operand<TString> dataSourceName, Operand<TString> query, List<DataType<?>> outputTypes, List<Shape> outputShapes) {
+  public static SqlDataset create(Scope scope, Operand<TString> driverName, Operand<TString> dataSourceName, Operand<TString> query, List<Class<?>> outputTypes, List<Shape> outputShapes) {
     OperationBuilder opBuilder = scope.env().opBuilder("SqlDataset", scope.makeOpName("SqlDataset"));
     opBuilder.addInput(driverName.asOutput(scope));
     opBuilder.addInput(dataSourceName.asOutput(scope));
     opBuilder.addInput(query.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
-    DataType[] outputTypesArray = new DataType[outputTypes.size()];
+    Class[] outputTypesArray = new Class[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
       outputTypesArray[i] = outputTypes.get(i);
     }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/TakeDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/TakeDataset.java
index 25199e03482..c0e8b37d4ce 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/TakeDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/TakeDataset.java
@@ -18,7 +18,6 @@
 package org.tensorflow.op.data;
 
 import java.util.List;
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -50,12 +49,12 @@ public final class TakeDataset extends RawOp implements Operand<TType> {
    * @return a new instance of TakeDataset
    */
   @Endpoint(describeByClass = true)
-  public static TakeDataset create(Scope scope, Operand<?> inputDataset, Operand<TInt64> count, List<DataType<?>> outputTypes, List<Shape> outputShapes) {
+  public static TakeDataset create(Scope scope, Operand<?> inputDataset, Operand<TInt64> count, List<Class<?>> outputTypes, List<Shape> outputShapes) {
     OperationBuilder opBuilder = scope.env().opBuilder("TakeDataset", scope.makeOpName("TakeDataset"));
     opBuilder.addInput(inputDataset.asOutput(scope));
     opBuilder.addInput(count.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
-    DataType[] outputTypesArray = new DataType[outputTypes.size()];
+    Class[] outputTypesArray = new Class[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
       outputTypesArray[i] = outputTypes.get(i);
     }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ThreadPoolDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ThreadPoolDataset.java
index 001f6e6ebe1..b42f03010a5 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ThreadPoolDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ThreadPoolDataset.java
@@ -18,7 +18,6 @@
 package org.tensorflow.op.data;
 
 import java.util.List;
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -46,12 +45,12 @@ public final class ThreadPoolDataset extends RawOp implements Operand<TType> {
    * @return a new instance of ThreadPoolDataset
    */
   @Endpoint(describeByClass = true)
-  public static ThreadPoolDataset create(Scope scope, Operand<?> inputDataset, Operand<?> threadPool, List<DataType<?>> outputTypes, List<Shape> outputShapes) {
+  public static ThreadPoolDataset create(Scope scope, Operand<?> inputDataset, Operand<?> threadPool, List<Class<?>> outputTypes, List<Shape> outputShapes) {
     OperationBuilder opBuilder = scope.env().opBuilder("ThreadPoolDataset", scope.makeOpName("ThreadPoolDataset"));
     opBuilder.addInput(inputDataset.asOutput(scope));
     opBuilder.addInput(threadPool.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
-    DataType[] outputTypesArray = new DataType[outputTypes.size()];
+    Class[] outputTypesArray = new Class[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
       outputTypesArray[i] = outputTypes.get(i);
     }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/UnbatchDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/UnbatchDataset.java
index 568a57e051f..a808972d5f5 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/UnbatchDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/UnbatchDataset.java
@@ -18,7 +18,6 @@
 package org.tensorflow.op.data;
 
 import java.util.List;
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -45,11 +44,11 @@ public final class UnbatchDataset extends RawOp implements Operand<TType> {
    * @return a new instance of UnbatchDataset
    */
   @Endpoint(describeByClass = true)
-  public static UnbatchDataset create(Scope scope, Operand<?> inputDataset, List<DataType<?>> outputTypes, List<Shape> outputShapes) {
+  public static UnbatchDataset create(Scope scope, Operand<?> inputDataset, List<Class<?>> outputTypes, List<Shape> outputShapes) {
     OperationBuilder opBuilder = scope.env().opBuilder("UnbatchDataset", scope.makeOpName("UnbatchDataset"));
     opBuilder.addInput(inputDataset.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
-    DataType[] outputTypesArray = new DataType[outputTypes.size()];
+    Class[] outputTypesArray = new Class[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
       outputTypesArray[i] = outputTypes.get(i);
     }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/UniqueDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/UniqueDataset.java
index 10cd6e1d5a3..e00906012c8 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/UniqueDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/UniqueDataset.java
@@ -18,7 +18,6 @@
 package org.tensorflow.op.data;
 
 import java.util.List;
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -45,11 +44,11 @@ public final class UniqueDataset extends RawOp implements Operand<TType> {
    * @return a new instance of UniqueDataset
    */
   @Endpoint(describeByClass = true)
-  public static UniqueDataset create(Scope scope, Operand<?> inputDataset, List<DataType<?>> outputTypes, List<Shape> outputShapes) {
+  public static UniqueDataset create(Scope scope, Operand<?> inputDataset, List<Class<?>> outputTypes, List<Shape> outputShapes) {
     OperationBuilder opBuilder = scope.env().opBuilder("UniqueDataset", scope.makeOpName("UniqueDataset"));
     opBuilder.addInput(inputDataset.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
-    DataType[] outputTypesArray = new DataType[outputTypes.size()];
+    Class[] outputTypesArray = new Class[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
       outputTypesArray[i] = outputTypes.get(i);
     }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/WindowDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/WindowDataset.java
index df6088f1346..33f679e2f31 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/WindowDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/WindowDataset.java
@@ -18,7 +18,6 @@
 package org.tensorflow.op.data;
 
 import java.util.List;
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -98,7 +97,7 @@ public final class WindowDataset extends RawOp implements Operand<TType> {
    * @return a new instance of WindowDataset
    */
   @Endpoint(describeByClass = true)
-  public static WindowDataset create(Scope scope, Operand<?> inputDataset, Operand<TInt64> size, Operand<TInt64> shift, Operand<TInt64> stride, Operand<TBool> dropRemainder, List<DataType<?>> outputTypes, List<Shape> outputShapes) {
+  public static WindowDataset create(Scope scope, Operand<?> inputDataset, Operand<TInt64> size, Operand<TInt64> shift, Operand<TInt64> stride, Operand<TBool> dropRemainder, List<Class<?>> outputTypes, List<Shape> outputShapes) {
     OperationBuilder opBuilder = scope.env().opBuilder("WindowDataset", scope.makeOpName("WindowDataset"));
     opBuilder.addInput(inputDataset.asOutput(scope));
     opBuilder.addInput(size.asOutput(scope));
@@ -106,7 +105,7 @@ public static WindowDataset create(Scope scope, Operand<?> inputDataset, Operand
     opBuilder.addInput(stride.asOutput(scope));
     opBuilder.addInput(dropRemainder.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
-    DataType[] outputTypesArray = new DataType[outputTypes.size()];
+    Class[] outputTypesArray = new Class[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
       outputTypesArray[i] = outputTypes.get(i);
     }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ZipDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ZipDataset.java
index 8cf4d8abc6a..f4497d1c109 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ZipDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ZipDataset.java
@@ -18,7 +18,6 @@
 package org.tensorflow.op.data;
 
 import java.util.List;
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -53,11 +52,11 @@ public final class ZipDataset extends RawOp implements Operand<TType> {
    * @return a new instance of ZipDataset
    */
   @Endpoint(describeByClass = true)
-  public static ZipDataset create(Scope scope, Iterable<Operand<?>> inputDatasets, List<DataType<?>> outputTypes, List<Shape> outputShapes) {
+  public static ZipDataset create(Scope scope, Iterable<Operand<?>> inputDatasets, List<Class<?>> outputTypes, List<Shape> outputShapes) {
     OperationBuilder opBuilder = scope.env().opBuilder("ZipDataset", scope.makeOpName("ZipDataset"));
     opBuilder.addInputList(Operands.asOutputs(scope, inputDatasets));
     opBuilder = scope.applyControlDependencies(opBuilder);
-    DataType[] outputTypesArray = new DataType[outputTypes.size()];
+    Class[] outputTypesArray = new Class[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
       outputTypesArray[i] = outputTypes.get(i);
     }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/AssertCardinalityDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/AssertCardinalityDataset.java
index 67ba6f37fba..69e7f0d769e 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/AssertCardinalityDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/AssertCardinalityDataset.java
@@ -18,7 +18,6 @@
 package org.tensorflow.op.data.experimental;
 
 import java.util.List;
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -46,12 +45,12 @@ public final class AssertCardinalityDataset extends RawOp implements Operand<TTy
    * @return a new instance of AssertCardinalityDataset
    */
   @Endpoint(describeByClass = true)
-  public static AssertCardinalityDataset create(Scope scope, Operand<?> inputDataset, Operand<TInt64> cardinality, List<DataType<?>> outputTypes, List<Shape> outputShapes) {
+  public static AssertCardinalityDataset create(Scope scope, Operand<?> inputDataset, Operand<TInt64> cardinality, List<Class<?>> outputTypes, List<Shape> outputShapes) {
     OperationBuilder opBuilder = scope.env().opBuilder("AssertCardinalityDataset", scope.makeOpName("AssertCardinalityDataset"));
     opBuilder.addInput(inputDataset.asOutput(scope));
     opBuilder.addInput(cardinality.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
-    DataType[] outputTypesArray = new DataType[outputTypes.size()];
+    Class[] outputTypesArray = new Class[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
       outputTypesArray[i] = outputTypes.get(i);
     }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/AssertNextDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/AssertNextDataset.java
index 45008edaf8e..213dd6428e2 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/AssertNextDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/AssertNextDataset.java
@@ -18,7 +18,6 @@
 package org.tensorflow.op.data.experimental;
 
 import java.util.List;
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -46,12 +45,12 @@ public final class AssertNextDataset extends RawOp implements Operand<TType> {
    * @return a new instance of AssertNextDataset
    */
   @Endpoint(describeByClass = true)
-  public static AssertNextDataset create(Scope scope, Operand<?> inputDataset, Operand<TString> transformations, List<DataType<?>> outputTypes, List<Shape> outputShapes) {
+  public static AssertNextDataset create(Scope scope, Operand<?> inputDataset, Operand<TString> transformations, List<Class<?>> outputTypes, List<Shape> outputShapes) {
     OperationBuilder opBuilder = scope.env().opBuilder("ExperimentalAssertNextDataset", scope.makeOpName("AssertNextDataset"));
     opBuilder.addInput(inputDataset.asOutput(scope));
     opBuilder.addInput(transformations.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
-    DataType[] outputTypesArray = new DataType[outputTypes.size()];
+    Class[] outputTypesArray = new Class[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
       outputTypesArray[i] = outputTypes.get(i);
     }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/AutoShardDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/AutoShardDataset.java
index 2eab290b2ab..5cc6afb9abc 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/AutoShardDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/AutoShardDataset.java
@@ -18,7 +18,6 @@
 package org.tensorflow.op.data.experimental;
 
 import java.util.List;
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -76,13 +75,13 @@ private Options() {
    * @return a new instance of AutoShardDataset
    */
   @Endpoint(describeByClass = true)
-  public static AutoShardDataset create(Scope scope, Operand<?> inputDataset, Operand<TInt64> numWorkers, Operand<TInt64> index, List<DataType<?>> outputTypes, List<Shape> outputShapes, Options... options) {
+  public static AutoShardDataset create(Scope scope, Operand<?> inputDataset, Operand<TInt64> numWorkers, Operand<TInt64> index, List<Class<?>> outputTypes, List<Shape> outputShapes, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ExperimentalAutoShardDataset", scope.makeOpName("AutoShardDataset"));
     opBuilder.addInput(inputDataset.asOutput(scope));
     opBuilder.addInput(numWorkers.asOutput(scope));
     opBuilder.addInput(index.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
-    DataType[] outputTypesArray = new DataType[outputTypes.size()];
+    Class[] outputTypesArray = new Class[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
       outputTypesArray[i] = outputTypes.get(i);
     }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/BytesProducedStatsDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/BytesProducedStatsDataset.java
index 3619b6f73ed..6c102aef09d 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/BytesProducedStatsDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/BytesProducedStatsDataset.java
@@ -18,7 +18,6 @@
 package org.tensorflow.op.data.experimental;
 
 import java.util.List;
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -47,12 +46,12 @@ public final class BytesProducedStatsDataset extends RawOp implements Operand<TT
    * @return a new instance of BytesProducedStatsDataset
    */
   @Endpoint(describeByClass = true)
-  public static BytesProducedStatsDataset create(Scope scope, Operand<?> inputDataset, Operand<TString> tag, List<DataType<?>> outputTypes, List<Shape> outputShapes) {
+  public static BytesProducedStatsDataset create(Scope scope, Operand<?> inputDataset, Operand<TString> tag, List<Class<?>> outputTypes, List<Shape> outputShapes) {
     OperationBuilder opBuilder = scope.env().opBuilder("ExperimentalBytesProducedStatsDataset", scope.makeOpName("BytesProducedStatsDataset"));
     opBuilder.addInput(inputDataset.asOutput(scope));
     opBuilder.addInput(tag.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
-    DataType[] outputTypesArray = new DataType[outputTypes.size()];
+    Class[] outputTypesArray = new Class[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
       outputTypesArray[i] = outputTypes.get(i);
     }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/ChooseFastestDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/ChooseFastestDataset.java
index 627a9e4784b..25781e4afa1 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/ChooseFastestDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/ChooseFastestDataset.java
@@ -18,7 +18,6 @@
 package org.tensorflow.op.data.experimental;
 
 import java.util.List;
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -46,12 +45,12 @@ public final class ChooseFastestDataset extends RawOp implements Operand<TType>
    * @return a new instance of ChooseFastestDataset
    */
   @Endpoint(describeByClass = true)
-  public static ChooseFastestDataset create(Scope scope, Iterable<Operand<?>> inputDatasets, Long numExperiments, List<DataType<?>> outputTypes, List<Shape> outputShapes) {
+  public static ChooseFastestDataset create(Scope scope, Iterable<Operand<?>> inputDatasets, Long numExperiments, List<Class<?>> outputTypes, List<Shape> outputShapes) {
     OperationBuilder opBuilder = scope.env().opBuilder("ExperimentalChooseFastestDataset", scope.makeOpName("ChooseFastestDataset"));
     opBuilder.addInputList(Operands.asOutputs(scope, inputDatasets));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("num_experiments", numExperiments);
-    DataType[] outputTypesArray = new DataType[outputTypes.size()];
+    Class[] outputTypesArray = new Class[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
       outputTypesArray[i] = outputTypes.get(i);
     }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/CompressElement.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/CompressElement.java
index 9e4bfb34a8b..0e853f754c3 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/CompressElement.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/CompressElement.java
@@ -43,7 +43,7 @@ public final class CompressElement extends RawOp implements Operand<TType> {
   @Endpoint(describeByClass = true)
   public static CompressElement create(Scope scope, Iterable<Operand<?>> components) {
     OperationBuilder opBuilder = scope.env().opBuilder("CompressElement", scope.makeOpName("CompressElement"));
-    opBuilder.addInputList(Operands.asOutputs(components));
+    opBuilder.addInputList(Operands.asOutputs(scope, components));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new CompressElement(opBuilder.build());
   }
@@ -56,7 +56,7 @@ public Output<?> compressed() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
+  public Output<TType> asOutput(Scope scope) {
     return (Output<TType>) compressed;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/DataServiceDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/DataServiceDataset.java
index b3e853f9de4..da71533bd2b 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/DataServiceDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/DataServiceDataset.java
@@ -18,7 +18,6 @@
 package org.tensorflow.op.data.experimental;
 
 import java.util.List;
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -73,17 +72,17 @@ private Options() {
    * @return a new instance of DataServiceDataset
    */
   @Endpoint(describeByClass = true)
-  public static DataServiceDataset create(Scope scope, Operand<TInt64> datasetId, Operand<TString> processingMode, Operand<TString> address, Operand<TString> protocol, Operand<TString> jobName, Operand<TInt64> maxOutstandingRequests, Operand<?> iterationCounter, List<DataType<?>> outputTypes, List<Shape> outputShapes, Options... options) {
+  public static DataServiceDataset create(Scope scope, Operand<TInt64> datasetId, Operand<TString> processingMode, Operand<TString> address, Operand<TString> protocol, Operand<TString> jobName, Operand<TInt64> maxOutstandingRequests, Operand<?> iterationCounter, List<Class<?>> outputTypes, List<Shape> outputShapes, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("DataServiceDataset", scope.makeOpName("DataServiceDataset"));
-    opBuilder.addInput(datasetId.asOutput());
-    opBuilder.addInput(processingMode.asOutput());
-    opBuilder.addInput(address.asOutput());
-    opBuilder.addInput(protocol.asOutput());
-    opBuilder.addInput(jobName.asOutput());
-    opBuilder.addInput(maxOutstandingRequests.asOutput());
-    opBuilder.addInput(iterationCounter.asOutput());
+    opBuilder.addInput(datasetId.asOutput(scope));
+    opBuilder.addInput(processingMode.asOutput(scope));
+    opBuilder.addInput(address.asOutput(scope));
+    opBuilder.addInput(protocol.asOutput(scope));
+    opBuilder.addInput(jobName.asOutput(scope));
+    opBuilder.addInput(maxOutstandingRequests.asOutput(scope));
+    opBuilder.addInput(iterationCounter.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
-    DataType[] outputTypesArray = new DataType[outputTypes.size()];
+    Class[] outputTypesArray = new Class[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
       outputTypesArray[i] = outputTypes.get(i);
     }
@@ -118,7 +117,7 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
+  public Output<TType> asOutput(Scope scope) {
     return (Output<TType>) handle;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/DenseToSparseBatchDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/DenseToSparseBatchDataset.java
index a1cae77fa4e..a2b5f8a6db6 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/DenseToSparseBatchDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/DenseToSparseBatchDataset.java
@@ -18,7 +18,6 @@
 package org.tensorflow.op.data.experimental;
 
 import java.util.List;
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -51,13 +50,13 @@ public final class DenseToSparseBatchDataset extends RawOp implements Operand<TT
    * @return a new instance of DenseToSparseBatchDataset
    */
   @Endpoint(describeByClass = true)
-  public static DenseToSparseBatchDataset create(Scope scope, Operand<?> inputDataset, Operand<TInt64> batchSize, Operand<TInt64> rowShape, List<DataType<?>> outputTypes, List<Shape> outputShapes) {
+  public static DenseToSparseBatchDataset create(Scope scope, Operand<?> inputDataset, Operand<TInt64> batchSize, Operand<TInt64> rowShape, List<Class<?>> outputTypes, List<Shape> outputShapes) {
     OperationBuilder opBuilder = scope.env().opBuilder("ExperimentalDenseToSparseBatchDataset", scope.makeOpName("DenseToSparseBatchDataset"));
     opBuilder.addInput(inputDataset.asOutput(scope));
     opBuilder.addInput(batchSize.asOutput(scope));
     opBuilder.addInput(rowShape.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
-    DataType[] outputTypesArray = new DataType[outputTypes.size()];
+    Class[] outputTypesArray = new Class[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
       outputTypesArray[i] = outputTypes.get(i);
     }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/DirectedInterleaveDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/DirectedInterleaveDataset.java
index 9fda314f40c..b19d75ce552 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/DirectedInterleaveDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/DirectedInterleaveDataset.java
@@ -18,7 +18,6 @@
 package org.tensorflow.op.data.experimental;
 
 import java.util.List;
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -49,12 +48,12 @@ public final class DirectedInterleaveDataset extends RawOp implements Operand<TT
    * @return a new instance of DirectedInterleaveDataset
    */
   @Endpoint(describeByClass = true)
-  public static DirectedInterleaveDataset create(Scope scope, Operand<?> selectorInputDataset, Iterable<Operand<?>> dataInputDatasets, List<DataType<?>> outputTypes, List<Shape> outputShapes) {
+  public static DirectedInterleaveDataset create(Scope scope, Operand<?> selectorInputDataset, Iterable<Operand<?>> dataInputDatasets, List<Class<?>> outputTypes, List<Shape> outputShapes) {
     OperationBuilder opBuilder = scope.env().opBuilder("ExperimentalDirectedInterleaveDataset", scope.makeOpName("DirectedInterleaveDataset"));
     opBuilder.addInput(selectorInputDataset.asOutput(scope));
     opBuilder.addInputList(Operands.asOutputs(scope, dataInputDatasets));
     opBuilder = scope.applyControlDependencies(opBuilder);
-    DataType[] outputTypesArray = new DataType[outputTypes.size()];
+    Class[] outputTypesArray = new Class[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
       outputTypesArray[i] = outputTypes.get(i);
     }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/DummyIterationCounter.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/DummyIterationCounter.java
index 72f83847285..9bc76fc0ba4 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/DummyIterationCounter.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/DummyIterationCounter.java
@@ -52,7 +52,7 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
+  public Output<TType> asOutput(Scope scope) {
     return (Output<TType>) handle;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/IgnoreErrorsDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/IgnoreErrorsDataset.java
index a14afc598a5..4ee758e24b8 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/IgnoreErrorsDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/IgnoreErrorsDataset.java
@@ -18,7 +18,6 @@
 package org.tensorflow.op.data.experimental;
 
 import java.util.List;
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -45,11 +44,11 @@ public final class IgnoreErrorsDataset extends RawOp implements Operand<TType> {
    * @return a new instance of IgnoreErrorsDataset
    */
   @Endpoint(describeByClass = true)
-  public static IgnoreErrorsDataset create(Scope scope, Operand<?> inputDataset, List<DataType<?>> outputTypes, List<Shape> outputShapes) {
+  public static IgnoreErrorsDataset create(Scope scope, Operand<?> inputDataset, List<Class<?>> outputTypes, List<Shape> outputShapes) {
     OperationBuilder opBuilder = scope.env().opBuilder("ExperimentalIgnoreErrorsDataset", scope.makeOpName("IgnoreErrorsDataset"));
     opBuilder.addInput(inputDataset.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
-    DataType[] outputTypesArray = new DataType[outputTypes.size()];
+    Class[] outputTypesArray = new Class[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
       outputTypesArray[i] = outputTypes.get(i);
     }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/LatencyStatsDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/LatencyStatsDataset.java
index 8073d397acd..335d3a44a5e 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/LatencyStatsDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/LatencyStatsDataset.java
@@ -18,7 +18,6 @@
 package org.tensorflow.op.data.experimental;
 
 import java.util.List;
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -47,12 +46,12 @@ public final class LatencyStatsDataset extends RawOp implements Operand<TType> {
    * @return a new instance of LatencyStatsDataset
    */
   @Endpoint(describeByClass = true)
-  public static LatencyStatsDataset create(Scope scope, Operand<?> inputDataset, Operand<TString> tag, List<DataType<?>> outputTypes, List<Shape> outputShapes) {
+  public static LatencyStatsDataset create(Scope scope, Operand<?> inputDataset, Operand<TString> tag, List<Class<?>> outputTypes, List<Shape> outputShapes) {
     OperationBuilder opBuilder = scope.env().opBuilder("ExperimentalLatencyStatsDataset", scope.makeOpName("LatencyStatsDataset"));
     opBuilder.addInput(inputDataset.asOutput(scope));
     opBuilder.addInput(tag.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
-    DataType[] outputTypesArray = new DataType[outputTypes.size()];
+    Class[] outputTypesArray = new Class[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
       outputTypesArray[i] = outputTypes.get(i);
     }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/LmdbDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/LmdbDataset.java
index 4c04fab718a..f0cba4ef849 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/LmdbDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/LmdbDataset.java
@@ -18,7 +18,6 @@
 package org.tensorflow.op.data.experimental;
 
 import java.util.List;
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -45,11 +44,11 @@ public final class LmdbDataset extends RawOp implements Operand<TType> {
    * @return a new instance of LmdbDataset
    */
   @Endpoint(describeByClass = true)
-  public static LmdbDataset create(Scope scope, Operand<TString> filenames, List<DataType<?>> outputTypes, List<Shape> outputShapes) {
+  public static LmdbDataset create(Scope scope, Operand<TString> filenames, List<Class<?>> outputTypes, List<Shape> outputShapes) {
     OperationBuilder opBuilder = scope.env().opBuilder("ExperimentalLMDBDataset", scope.makeOpName("LmdbDataset"));
     opBuilder.addInput(filenames.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
-    DataType[] outputTypesArray = new DataType[outputTypes.size()];
+    Class[] outputTypesArray = new Class[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
       outputTypesArray[i] = outputTypes.get(i);
     }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/MaxIntraOpParallelismDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/MaxIntraOpParallelismDataset.java
index 16e5d1a9d51..62ae49c957d 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/MaxIntraOpParallelismDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/MaxIntraOpParallelismDataset.java
@@ -18,7 +18,6 @@
 package org.tensorflow.op.data.experimental;
 
 import java.util.List;
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -47,12 +46,12 @@ public final class MaxIntraOpParallelismDataset extends RawOp implements Operand
    * @return a new instance of MaxIntraOpParallelismDataset
    */
   @Endpoint(describeByClass = true)
-  public static MaxIntraOpParallelismDataset create(Scope scope, Operand<?> inputDataset, Operand<TInt64> maxIntraOpParallelism, List<DataType<?>> outputTypes, List<Shape> outputShapes) {
+  public static MaxIntraOpParallelismDataset create(Scope scope, Operand<?> inputDataset, Operand<TInt64> maxIntraOpParallelism, List<Class<?>> outputTypes, List<Shape> outputShapes) {
     OperationBuilder opBuilder = scope.env().opBuilder("ExperimentalMaxIntraOpParallelismDataset", scope.makeOpName("MaxIntraOpParallelismDataset"));
     opBuilder.addInput(inputDataset.asOutput(scope));
     opBuilder.addInput(maxIntraOpParallelism.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
-    DataType[] outputTypesArray = new DataType[outputTypes.size()];
+    Class[] outputTypesArray = new Class[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
       outputTypesArray[i] = outputTypes.get(i);
     }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/NonSerializableDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/NonSerializableDataset.java
index 4ab5e682300..e1a99b84aa3 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/NonSerializableDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/NonSerializableDataset.java
@@ -18,7 +18,6 @@
 package org.tensorflow.op.data.experimental;
 
 import java.util.List;
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -44,11 +43,11 @@ public final class NonSerializableDataset extends RawOp implements Operand<TType
    * @return a new instance of NonSerializableDataset
    */
   @Endpoint(describeByClass = true)
-  public static NonSerializableDataset create(Scope scope, Operand<?> inputDataset, List<DataType<?>> outputTypes, List<Shape> outputShapes) {
+  public static NonSerializableDataset create(Scope scope, Operand<?> inputDataset, List<Class<?>> outputTypes, List<Shape> outputShapes) {
     OperationBuilder opBuilder = scope.env().opBuilder("ExperimentalNonSerializableDataset", scope.makeOpName("NonSerializableDataset"));
     opBuilder.addInput(inputDataset.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
-    DataType[] outputTypesArray = new DataType[outputTypes.size()];
+    Class[] outputTypesArray = new Class[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
       outputTypesArray[i] = outputTypes.get(i);
     }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/ParseExampleDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/ParseExampleDataset.java
index 104939a7cf9..f64b9741b13 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/ParseExampleDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/ParseExampleDataset.java
@@ -18,7 +18,6 @@
 package org.tensorflow.op.data.experimental;
 
 import java.util.List;
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -101,7 +100,7 @@ private Options() {
    * @return a new instance of ParseExampleDataset
    */
   @Endpoint(describeByClass = true)
-  public static ParseExampleDataset create(Scope scope, Operand<?> inputDataset, Operand<TInt64> numParallelCalls, Iterable<Operand<?>> denseDefaults, List<String> sparseKeys, List<String> denseKeys, List<DataType<?>> sparseTypes, List<Shape> denseShapes, List<DataType<?>> outputTypes, List<Shape> outputShapes, List<DataType<?>> raggedValueTypes, List<DataType<?>> raggedSplitTypes, Options... options) {
+  public static ParseExampleDataset create(Scope scope, Operand<?> inputDataset, Operand<TInt64> numParallelCalls, Iterable<Operand<?>> denseDefaults, List<String> sparseKeys, List<String> denseKeys, List<Class<?>> sparseTypes, List<Shape> denseShapes, List<Class<?>> outputTypes, List<Shape> outputShapes, List<Class<?>> raggedValueTypes, List<Class<?>> raggedSplitTypes, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ParseExampleDatasetV2", scope.makeOpName("ParseExampleDataset"));
     opBuilder.addInput(inputDataset.asOutput(scope));
     opBuilder.addInput(numParallelCalls.asOutput(scope));
@@ -117,7 +116,7 @@ public static ParseExampleDataset create(Scope scope, Operand<?> inputDataset, O
       denseKeysArray[i] = denseKeys.get(i);
     }
     opBuilder.setAttr("dense_keys", denseKeysArray);
-    DataType[] sparseTypesArray = new DataType[sparseTypes.size()];
+    Class[] sparseTypesArray = new Class[sparseTypes.size()];
     for (int i = 0; i < sparseTypesArray.length; ++i) {
       sparseTypesArray[i] = sparseTypes.get(i);
     }
@@ -127,7 +126,7 @@ public static ParseExampleDataset create(Scope scope, Operand<?> inputDataset, O
       denseShapesArray[i] = denseShapes.get(i);
     }
     opBuilder.setAttr("dense_shapes", denseShapesArray);
-    DataType[] outputTypesArray = new DataType[outputTypes.size()];
+    Class[] outputTypesArray = new Class[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
       outputTypesArray[i] = outputTypes.get(i);
     }
@@ -137,12 +136,12 @@ public static ParseExampleDataset create(Scope scope, Operand<?> inputDataset, O
       outputShapesArray[i] = outputShapes.get(i);
     }
     opBuilder.setAttr("output_shapes", outputShapesArray);
-    DataType[] raggedValueTypesArray = new DataType[raggedValueTypes.size()];
+    Class[] raggedValueTypesArray = new Class[raggedValueTypes.size()];
     for (int i = 0; i < raggedValueTypesArray.length; ++i) {
       raggedValueTypesArray[i] = raggedValueTypes.get(i);
     }
     opBuilder.setAttr("ragged_value_types", raggedValueTypesArray);
-    DataType[] raggedSplitTypesArray = new DataType[raggedSplitTypes.size()];
+    Class[] raggedSplitTypesArray = new Class[raggedSplitTypes.size()];
     for (int i = 0; i < raggedSplitTypesArray.length; ++i) {
       raggedSplitTypesArray[i] = raggedSplitTypes.get(i);
     }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/PrivateThreadPoolDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/PrivateThreadPoolDataset.java
index 2f31ef615da..0d972e9c9f4 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/PrivateThreadPoolDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/PrivateThreadPoolDataset.java
@@ -18,7 +18,6 @@
 package org.tensorflow.op.data.experimental;
 
 import java.util.List;
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -47,12 +46,12 @@ public final class PrivateThreadPoolDataset extends RawOp implements Operand<TTy
    * @return a new instance of PrivateThreadPoolDataset
    */
   @Endpoint(describeByClass = true)
-  public static PrivateThreadPoolDataset create(Scope scope, Operand<?> inputDataset, Operand<TInt64> numThreads, List<DataType<?>> outputTypes, List<Shape> outputShapes) {
+  public static PrivateThreadPoolDataset create(Scope scope, Operand<?> inputDataset, Operand<TInt64> numThreads, List<Class<?>> outputTypes, List<Shape> outputShapes) {
     OperationBuilder opBuilder = scope.env().opBuilder("ExperimentalPrivateThreadPoolDataset", scope.makeOpName("PrivateThreadPoolDataset"));
     opBuilder.addInput(inputDataset.asOutput(scope));
     opBuilder.addInput(numThreads.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
-    DataType[] outputTypesArray = new DataType[outputTypes.size()];
+    Class[] outputTypesArray = new Class[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
       outputTypesArray[i] = outputTypes.get(i);
     }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/RandomDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/RandomDataset.java
index 2ad127b91d6..a00a6be2fe4 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/RandomDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/RandomDataset.java
@@ -18,7 +18,6 @@
 package org.tensorflow.op.data.experimental;
 
 import java.util.List;
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -49,12 +48,12 @@ public final class RandomDataset extends RawOp implements Operand<TType> {
    * @return a new instance of RandomDataset
    */
   @Endpoint(describeByClass = true)
-  public static RandomDataset create(Scope scope, Operand<TInt64> seed, Operand<TInt64> seed2, List<DataType<?>> outputTypes, List<Shape> outputShapes) {
+  public static RandomDataset create(Scope scope, Operand<TInt64> seed, Operand<TInt64> seed2, List<Class<?>> outputTypes, List<Shape> outputShapes) {
     OperationBuilder opBuilder = scope.env().opBuilder("ExperimentalRandomDataset", scope.makeOpName("RandomDataset"));
     opBuilder.addInput(seed.asOutput(scope));
     opBuilder.addInput(seed2.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
-    DataType[] outputTypesArray = new DataType[outputTypes.size()];
+    Class[] outputTypesArray = new Class[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
       outputTypesArray[i] = outputTypes.get(i);
     }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/RebatchDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/RebatchDataset.java
index 9c0ee15dcce..f59e00f2d48 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/RebatchDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/RebatchDataset.java
@@ -18,7 +18,6 @@
 package org.tensorflow.op.data.experimental;
 
 import java.util.List;
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -72,12 +71,12 @@ private Options() {
    * @return a new instance of RebatchDataset
    */
   @Endpoint(describeByClass = true)
-  public static RebatchDataset create(Scope scope, Operand<?> inputDataset, Operand<TInt64> numReplicas, List<DataType<?>> outputTypes, List<Shape> outputShapes, Options... options) {
+  public static RebatchDataset create(Scope scope, Operand<?> inputDataset, Operand<TInt64> numReplicas, List<Class<?>> outputTypes, List<Shape> outputShapes, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ExperimentalRebatchDataset", scope.makeOpName("RebatchDataset"));
     opBuilder.addInput(inputDataset.asOutput(scope));
     opBuilder.addInput(numReplicas.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
-    DataType[] outputTypesArray = new DataType[outputTypes.size()];
+    Class[] outputTypesArray = new Class[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
       outputTypesArray[i] = outputTypes.get(i);
     }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/SetStatsAggregatorDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/SetStatsAggregatorDataset.java
index 9e80103945d..1de0c59d921 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/SetStatsAggregatorDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/SetStatsAggregatorDataset.java
@@ -18,7 +18,6 @@
 package org.tensorflow.op.data.experimental;
 
 import java.util.List;
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -48,14 +47,14 @@ public final class SetStatsAggregatorDataset extends RawOp implements Operand<TT
    * @return a new instance of SetStatsAggregatorDataset
    */
   @Endpoint(describeByClass = true)
-  public static SetStatsAggregatorDataset create(Scope scope, Operand<?> inputDataset, Operand<?> statsAggregator, Operand<TString> tag, Operand<TString> counterPrefix, List<DataType<?>> outputTypes, List<Shape> outputShapes) {
+  public static SetStatsAggregatorDataset create(Scope scope, Operand<?> inputDataset, Operand<?> statsAggregator, Operand<TString> tag, Operand<TString> counterPrefix, List<Class<?>> outputTypes, List<Shape> outputShapes) {
     OperationBuilder opBuilder = scope.env().opBuilder("ExperimentalSetStatsAggregatorDataset", scope.makeOpName("SetStatsAggregatorDataset"));
     opBuilder.addInput(inputDataset.asOutput(scope));
     opBuilder.addInput(statsAggregator.asOutput(scope));
     opBuilder.addInput(tag.asOutput(scope));
     opBuilder.addInput(counterPrefix.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
-    DataType[] outputTypesArray = new DataType[outputTypes.size()];
+    Class[] outputTypesArray = new Class[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
       outputTypesArray[i] = outputTypes.get(i);
     }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/SleepDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/SleepDataset.java
index bf8753e5210..e26f82bb21d 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/SleepDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/SleepDataset.java
@@ -18,7 +18,6 @@
 package org.tensorflow.op.data.experimental;
 
 import java.util.List;
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -46,12 +45,12 @@ public final class SleepDataset extends RawOp implements Operand<TType> {
    * @return a new instance of SleepDataset
    */
   @Endpoint(describeByClass = true)
-  public static SleepDataset create(Scope scope, Operand<?> inputDataset, Operand<TInt64> sleepMicroseconds, List<DataType<?>> outputTypes, List<Shape> outputShapes) {
+  public static SleepDataset create(Scope scope, Operand<?> inputDataset, Operand<TInt64> sleepMicroseconds, List<Class<?>> outputTypes, List<Shape> outputShapes) {
     OperationBuilder opBuilder = scope.env().opBuilder("ExperimentalSleepDataset", scope.makeOpName("SleepDataset"));
     opBuilder.addInput(inputDataset.asOutput(scope));
     opBuilder.addInput(sleepMicroseconds.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
-    DataType[] outputTypesArray = new DataType[outputTypes.size()];
+    Class[] outputTypesArray = new Class[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
       outputTypesArray[i] = outputTypes.get(i);
     }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/SlidingWindowDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/SlidingWindowDataset.java
index 6c583c31979..3ee0aaacd30 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/SlidingWindowDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/SlidingWindowDataset.java
@@ -18,7 +18,6 @@
 package org.tensorflow.op.data.experimental;
 
 import java.util.List;
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -52,14 +51,14 @@ public final class SlidingWindowDataset extends RawOp implements Operand<TType>
    * @return a new instance of SlidingWindowDataset
    */
   @Endpoint(describeByClass = true)
-  public static SlidingWindowDataset create(Scope scope, Operand<?> inputDataset, Operand<TInt64> windowSize, Operand<TInt64> windowShift, Operand<TInt64> windowStride, List<DataType<?>> outputTypes, List<Shape> outputShapes) {
+  public static SlidingWindowDataset create(Scope scope, Operand<?> inputDataset, Operand<TInt64> windowSize, Operand<TInt64> windowShift, Operand<TInt64> windowStride, List<Class<?>> outputTypes, List<Shape> outputShapes) {
     OperationBuilder opBuilder = scope.env().opBuilder("ExperimentalSlidingWindowDataset", scope.makeOpName("SlidingWindowDataset"));
     opBuilder.addInput(inputDataset.asOutput(scope));
     opBuilder.addInput(windowSize.asOutput(scope));
     opBuilder.addInput(windowShift.asOutput(scope));
     opBuilder.addInput(windowStride.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
-    DataType[] outputTypesArray = new DataType[outputTypes.size()];
+    Class[] outputTypesArray = new Class[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
       outputTypesArray[i] = outputTypes.get(i);
     }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/SqlDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/SqlDataset.java
index c0d2f90ed3b..9e5f7d40144 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/SqlDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/SqlDataset.java
@@ -18,7 +18,6 @@
 package org.tensorflow.op.data.experimental;
 
 import java.util.List;
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -48,13 +47,13 @@ public final class SqlDataset extends RawOp implements Operand<TType> {
    * @return a new instance of SqlDataset
    */
   @Endpoint(describeByClass = true)
-  public static SqlDataset create(Scope scope, Operand<TString> driverName, Operand<TString> dataSourceName, Operand<TString> query, List<DataType<?>> outputTypes, List<Shape> outputShapes) {
+  public static SqlDataset create(Scope scope, Operand<TString> driverName, Operand<TString> dataSourceName, Operand<TString> query, List<Class<?>> outputTypes, List<Shape> outputShapes) {
     OperationBuilder opBuilder = scope.env().opBuilder("ExperimentalSqlDataset", scope.makeOpName("SqlDataset"));
     opBuilder.addInput(driverName.asOutput(scope));
     opBuilder.addInput(dataSourceName.asOutput(scope));
     opBuilder.addInput(query.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
-    DataType[] outputTypesArray = new DataType[outputTypes.size()];
+    Class[] outputTypesArray = new Class[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
       outputTypesArray[i] = outputTypes.get(i);
     }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/ThreadPoolDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/ThreadPoolDataset.java
index 1e3c110f8c5..a6f66856cbc 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/ThreadPoolDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/ThreadPoolDataset.java
@@ -18,7 +18,6 @@
 package org.tensorflow.op.data.experimental;
 
 import java.util.List;
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -46,12 +45,12 @@ public final class ThreadPoolDataset extends RawOp implements Operand<TType> {
    * @return a new instance of ThreadPoolDataset
    */
   @Endpoint(describeByClass = true)
-  public static ThreadPoolDataset create(Scope scope, Operand<?> inputDataset, Operand<?> threadPool, List<DataType<?>> outputTypes, List<Shape> outputShapes) {
+  public static ThreadPoolDataset create(Scope scope, Operand<?> inputDataset, Operand<?> threadPool, List<Class<?>> outputTypes, List<Shape> outputShapes) {
     OperationBuilder opBuilder = scope.env().opBuilder("ExperimentalThreadPoolDataset", scope.makeOpName("ThreadPoolDataset"));
     opBuilder.addInput(inputDataset.asOutput(scope));
     opBuilder.addInput(threadPool.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
-    DataType[] outputTypesArray = new DataType[outputTypes.size()];
+    Class[] outputTypesArray = new Class[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
       outputTypesArray[i] = outputTypes.get(i);
     }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/UnbatchDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/UnbatchDataset.java
index 6a34aeef823..b44a2ce1e25 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/UnbatchDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/UnbatchDataset.java
@@ -18,7 +18,6 @@
 package org.tensorflow.op.data.experimental;
 
 import java.util.List;
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -45,11 +44,11 @@ public final class UnbatchDataset extends RawOp implements Operand<TType> {
    * @return a new instance of UnbatchDataset
    */
   @Endpoint(describeByClass = true)
-  public static UnbatchDataset create(Scope scope, Operand<?> inputDataset, List<DataType<?>> outputTypes, List<Shape> outputShapes) {
+  public static UnbatchDataset create(Scope scope, Operand<?> inputDataset, List<Class<?>> outputTypes, List<Shape> outputShapes) {
     OperationBuilder opBuilder = scope.env().opBuilder("ExperimentalUnbatchDataset", scope.makeOpName("UnbatchDataset"));
     opBuilder.addInput(inputDataset.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
-    DataType[] outputTypesArray = new DataType[outputTypes.size()];
+    Class[] outputTypesArray = new Class[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
       outputTypesArray[i] = outputTypes.get(i);
     }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/UncompressElement.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/UncompressElement.java
index c5732154e94..384ba0d46d4 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/UncompressElement.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/UncompressElement.java
@@ -20,7 +20,6 @@
 import java.util.Arrays;
 import java.util.Iterator;
 import java.util.List;
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -47,11 +46,11 @@ public final class UncompressElement extends RawOp implements Iterable<Operand<T
    * @return a new instance of UncompressElement
    */
   @Endpoint(describeByClass = true)
-  public static UncompressElement create(Scope scope, Operand<?> compressed, List<DataType<?>> outputTypes, List<Shape> outputShapes) {
+  public static UncompressElement create(Scope scope, Operand<?> compressed, List<Class<?>> outputTypes, List<Shape> outputShapes) {
     OperationBuilder opBuilder = scope.env().opBuilder("UncompressElement", scope.makeOpName("UncompressElement"));
-    opBuilder.addInput(compressed.asOutput());
+    opBuilder.addInput(compressed.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
-    DataType[] outputTypesArray = new DataType[outputTypes.size()];
+    Class[] outputTypesArray = new Class[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
       outputTypesArray[i] = outputTypes.get(i);
     }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/UniqueDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/UniqueDataset.java
index 88c2928b20e..d59ddf296fd 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/UniqueDataset.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/UniqueDataset.java
@@ -18,7 +18,6 @@
 package org.tensorflow.op.data.experimental;
 
 import java.util.List;
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -45,11 +44,11 @@ public final class UniqueDataset extends RawOp implements Operand<TType> {
    * @return a new instance of UniqueDataset
    */
   @Endpoint(describeByClass = true)
-  public static UniqueDataset create(Scope scope, Operand<?> inputDataset, List<DataType<?>> outputTypes, List<Shape> outputShapes) {
+  public static UniqueDataset create(Scope scope, Operand<?> inputDataset, List<Class<?>> outputTypes, List<Shape> outputShapes) {
     OperationBuilder opBuilder = scope.env().opBuilder("ExperimentalUniqueDataset", scope.makeOpName("UniqueDataset"));
     opBuilder.addInput(inputDataset.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
-    DataType[] outputTypesArray = new DataType[outputTypes.size()];
+    Class[] outputTypesArray = new Class[outputTypes.size()];
     for (int i = 0; i < outputTypesArray.length; ++i) {
       outputTypesArray[i] = outputTypes.get(i);
     }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/debugging/DebugNumericsSummary.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/debugging/DebugNumericsSummary.java
index e30b0390eb2..6e5cd1fb011 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/debugging/DebugNumericsSummary.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/debugging/DebugNumericsSummary.java
@@ -17,7 +17,6 @@
 
 package org.tensorflow.op.debugging;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -132,7 +131,7 @@ private Options() {
    * @return a new instance of DebugNumericsSummary
    */
   @Endpoint(describeByClass = true)
-  public static <U extends TNumber, T extends TType> DebugNumericsSummary<U> create(Scope scope, Operand<T> input, DataType<U> outputDtype, Options... options) {
+  public static <U extends TNumber, T extends TType> DebugNumericsSummary<U> create(Scope scope, Operand<T> input, Class<U> outputDtype, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("DebugNumericSummaryV2", scope.makeOpName("DebugNumericsSummary"));
     opBuilder.addInput(input.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
@@ -160,7 +159,7 @@ public static <U extends TNumber, T extends TType> DebugNumericsSummary<U> creat
    */
   @Endpoint(describeByClass = true)
   public static <T extends TType> DebugNumericsSummary<TFloat32> create(Scope scope, Operand<T> input, Options... options) {
-    return create(scope, input, TFloat32.DTYPE, options);
+    return create(scope, input, TFloat32.class, options);
   }
   
   /**
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/dtypes/Cast.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/dtypes/Cast.java
index 2f3400bfd3b..23fdd3d3a42 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/dtypes/Cast.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/dtypes/Cast.java
@@ -17,7 +17,6 @@
 
 package org.tensorflow.op.dtypes;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -65,7 +64,7 @@ private Options() {
    * @return a new instance of Cast
    */
   @Endpoint(describeByClass = true)
-  public static <U extends TType, T extends TType> Cast<U> create(Scope scope, Operand<T> x, DataType<U> DstT, Options... options) {
+  public static <U extends TType, T extends TType> Cast<U> create(Scope scope, Operand<T> x, Class<U> DstT, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("Cast", scope.makeOpName("Cast"));
     opBuilder.addInput(x.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/dtypes/Complex.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/dtypes/Complex.java
index a9711798846..17f10b67977 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/dtypes/Complex.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/dtypes/Complex.java
@@ -17,7 +17,6 @@
 
 package org.tensorflow.op.dtypes;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -62,7 +61,7 @@ public final class Complex<U extends TType> extends RawOp implements Operand<U>
    * @return a new instance of Complex
    */
   @Endpoint(describeByClass = true)
-  public static <U extends TType, T extends TNumber> Complex<U> create(Scope scope, Operand<T> real, Operand<T> imag, DataType<U> Tout) {
+  public static <U extends TType, T extends TNumber> Complex<U> create(Scope scope, Operand<T> real, Operand<T> imag, Class<U> Tout) {
     OperationBuilder opBuilder = scope.env().opBuilder("Complex", scope.makeOpName("Complex"));
     opBuilder.addInput(real.asOutput(scope));
     opBuilder.addInput(imag.asOutput(scope));
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/CropAndResizeGradImage.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/CropAndResizeGradImage.java
index e3a70effa94..b6a3830a3a6 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/CropAndResizeGradImage.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/CropAndResizeGradImage.java
@@ -17,7 +17,6 @@
 
 package org.tensorflow.op.image;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -83,7 +82,7 @@ private Options() {
    * @return a new instance of CropAndResizeGradImage
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> CropAndResizeGradImage<T> create(Scope scope, Operand<TFloat32> grads, Operand<TFloat32> boxes, Operand<TInt32> boxInd, Operand<TInt32> imageSize, DataType<T> T, Options... options) {
+  public static <T extends TNumber> CropAndResizeGradImage<T> create(Scope scope, Operand<TFloat32> grads, Operand<TFloat32> boxes, Operand<TInt32> boxInd, Operand<TInt32> imageSize, Class<T> T, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("CropAndResizeGradImage", scope.makeOpName("CropAndResizeGradImage"));
     opBuilder.addInput(grads.asOutput(scope));
     opBuilder.addInput(boxes.asOutput(scope));
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/DecodePng.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/DecodePng.java
index 0f0b995b868..7cb68f21721 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/DecodePng.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/DecodePng.java
@@ -17,7 +17,6 @@
 
 package org.tensorflow.op.image;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -91,7 +90,7 @@ private Options() {
    * @return a new instance of DecodePng
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> DecodePng<T> create(Scope scope, Operand<TString> contents, DataType<T> dtype, Options... options) {
+  public static <T extends TNumber> DecodePng<T> create(Scope scope, Operand<TString> contents, Class<T> dtype, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("DecodePng", scope.makeOpName("DecodePng"));
     opBuilder.addInput(contents.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
@@ -116,7 +115,7 @@ public static <T extends TNumber> DecodePng<T> create(Scope scope, Operand<TStri
    */
   @Endpoint(describeByClass = true)
   public static DecodePng<TUint8> create(Scope scope, Operand<TString> contents, Options... options) {
-    return create(scope, contents, TUint8.DTYPE, options);
+    return create(scope, contents, TUint8.class, options);
   }
   
   /**
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ExtractGlimpse.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ExtractGlimpse.java
index 2f41f806d48..59f46dd1947 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ExtractGlimpse.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ExtractGlimpse.java
@@ -58,7 +58,6 @@
  * If the coordinates are not normalized they are interpreted as
  *   numbers of pixels.
  */
-@Operator(group = "image")
 public final class ExtractGlimpse extends RawOp implements Operand<TFloat32> {
   
   /**
@@ -128,7 +127,7 @@ private Options() {
    */
   @Endpoint(describeByClass = true)
   public static ExtractGlimpse create(Scope scope, Operand<TFloat32> input, Operand<TInt32> size, Operand<TFloat32> offsets, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ExtractGlimpse", scope.makeOpName("ExtractGlimpse"));
+    OperationBuilder opBuilder = scope.env().opBuilder("ExtractGlimpseV2", scope.makeOpName("ExtractGlimpse"));
     opBuilder.addInput(input.asOutput(scope));
     opBuilder.addInput(size.asOutput(scope));
     opBuilder.addInput(offsets.asOutput(scope));
@@ -199,6 +198,9 @@ public Output<TFloat32> asOutput(Scope scope) {
     return glimpse;
   }
   
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ExtractGlimpseV2";
+  
   private Output<TFloat32> glimpse;
   
   private ExtractGlimpse(Operation operation) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ExtractJpegShape.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ExtractJpegShape.java
index d5932d9fe45..467fa8820ed 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ExtractJpegShape.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ExtractJpegShape.java
@@ -17,7 +17,6 @@
 
 package org.tensorflow.op.image;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -50,7 +49,7 @@ public final class ExtractJpegShape<T extends TNumber> extends RawOp implements
    * @return a new instance of ExtractJpegShape
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> ExtractJpegShape<T> create(Scope scope, Operand<TString> contents, DataType<T> outputType) {
+  public static <T extends TNumber> ExtractJpegShape<T> create(Scope scope, Operand<TString> contents, Class<T> outputType) {
     OperationBuilder opBuilder = scope.env().opBuilder("ExtractJpegShape", scope.makeOpName("ExtractJpegShape"));
     opBuilder.addInput(contents.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
@@ -67,7 +66,7 @@ public static <T extends TNumber> ExtractJpegShape<T> create(Scope scope, Operan
    */
   @Endpoint(describeByClass = true)
   public static ExtractJpegShape<TInt32> create(Scope scope, Operand<TString> contents) {
-    return create(scope, contents, TInt32.DTYPE);
+    return create(scope, contents, TInt32.class);
   }
   
   /**
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/DecodePaddedRaw.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/DecodePaddedRaw.java
index 6eb973f6e01..a7520073665 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/DecodePaddedRaw.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/DecodePaddedRaw.java
@@ -17,7 +17,6 @@
 
 package org.tensorflow.op.io;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -70,7 +69,7 @@ private Options() {
    * @return a new instance of DecodePaddedRaw
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> DecodePaddedRaw<T> create(Scope scope, Operand<TString> inputBytes, Operand<TInt32> fixedLength, DataType<T> outType, Options... options) {
+  public static <T extends TNumber> DecodePaddedRaw<T> create(Scope scope, Operand<TString> inputBytes, Operand<TInt32> fixedLength, Class<T> outType, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("DecodePaddedRaw", scope.makeOpName("DecodePaddedRaw"));
     opBuilder.addInput(inputBytes.asOutput(scope));
     opBuilder.addInput(fixedLength.asOutput(scope));
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/DecodeRaw.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/DecodeRaw.java
index ec062e780a3..97f5543c8d4 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/DecodeRaw.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/DecodeRaw.java
@@ -17,7 +17,6 @@
 
 package org.tensorflow.op.io;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -68,7 +67,7 @@ private Options() {
    * @return a new instance of DecodeRaw
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> DecodeRaw<T> create(Scope scope, Operand<TString> bytes, DataType<T> outType, Options... options) {
+  public static <T extends TType> DecodeRaw<T> create(Scope scope, Operand<TString> bytes, Class<T> outType, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("DecodeRaw", scope.makeOpName("DecodeRaw"));
     opBuilder.addInput(bytes.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/DeserializeManySparse.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/DeserializeManySparse.java
index d31ab4de4d0..43d30a48a90 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/DeserializeManySparse.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/DeserializeManySparse.java
@@ -17,7 +17,6 @@
 
 package org.tensorflow.op.io;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -90,7 +89,7 @@ public final class DeserializeManySparse<T extends TType> extends RawOp {
    * @return a new instance of DeserializeManySparse
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> DeserializeManySparse<T> create(Scope scope, Operand<TString> serializedSparse, DataType<T> dtype) {
+  public static <T extends TType> DeserializeManySparse<T> create(Scope scope, Operand<TString> serializedSparse, Class<T> dtype) {
     OperationBuilder opBuilder = scope.env().opBuilder("DeserializeManySparse", scope.makeOpName("DeserializeManySparse"));
     opBuilder.addInput(serializedSparse.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/FifoQueue.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/FifoQueue.java
index e7509130d50..ca239776907 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/FifoQueue.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/FifoQueue.java
@@ -18,7 +18,6 @@
 package org.tensorflow.op.io;
 
 import java.util.List;
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -97,10 +96,10 @@ private Options() {
    * @return a new instance of FifoQueue
    */
   @Endpoint(describeByClass = true)
-  public static FifoQueue create(Scope scope, List<DataType<?>> componentTypes, Options... options) {
+  public static FifoQueue create(Scope scope, List<Class<?>> componentTypes, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("FIFOQueueV2", scope.makeOpName("FifoQueue"));
     opBuilder = scope.applyControlDependencies(opBuilder);
-    DataType[] componentTypesArray = new DataType[componentTypes.size()];
+    Class[] componentTypesArray = new Class[componentTypes.size()];
     for (int i = 0; i < componentTypesArray.length; ++i) {
       componentTypesArray[i] = componentTypes.get(i);
     }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/PaddingFifoQueue.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/PaddingFifoQueue.java
index 5fb8d2c4134..ef993fa907d 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/PaddingFifoQueue.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/PaddingFifoQueue.java
@@ -18,7 +18,6 @@
 package org.tensorflow.op.io;
 
 import java.util.List;
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -105,10 +104,10 @@ private Options() {
    * @return a new instance of PaddingFifoQueue
    */
   @Endpoint(describeByClass = true)
-  public static PaddingFifoQueue create(Scope scope, List<DataType<?>> componentTypes, Options... options) {
+  public static PaddingFifoQueue create(Scope scope, List<Class<?>> componentTypes, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("PaddingFIFOQueueV2", scope.makeOpName("PaddingFifoQueue"));
     opBuilder = scope.applyControlDependencies(opBuilder);
-    DataType[] componentTypesArray = new DataType[componentTypes.size()];
+    Class[] componentTypesArray = new Class[componentTypes.size()];
     for (int i = 0; i < componentTypesArray.length; ++i) {
       componentTypesArray[i] = componentTypes.get(i);
     }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ParseExample.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ParseExample.java
index c6c5be5046e..dddc8502a2b 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ParseExample.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ParseExample.java
@@ -19,7 +19,6 @@
 
 import java.util.Arrays;
 import java.util.List;
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -96,7 +95,7 @@ public final class ParseExample extends RawOp {
    * @return a new instance of ParseExample
    */
   @Endpoint(describeByClass = true)
-  public static ParseExample create(Scope scope, Operand<TString> serialized, Operand<TString> names, Operand<TString> sparseKeys, Operand<TString> denseKeys, Operand<TString> raggedKeys, Iterable<Operand<?>> denseDefaults, Long numSparse, List<DataType<?>> sparseTypes, List<DataType<?>> raggedValueTypes, List<DataType<?>> raggedSplitTypes, List<Shape> denseShapes) {
+  public static ParseExample create(Scope scope, Operand<TString> serialized, Operand<TString> names, Operand<TString> sparseKeys, Operand<TString> denseKeys, Operand<TString> raggedKeys, Iterable<Operand<?>> denseDefaults, Long numSparse, List<Class<?>> sparseTypes, List<Class<?>> raggedValueTypes, List<Class<?>> raggedSplitTypes, List<Shape> denseShapes) {
     OperationBuilder opBuilder = scope.env().opBuilder("ParseExampleV2", scope.makeOpName("ParseExample"));
     opBuilder.addInput(serialized.asOutput(scope));
     opBuilder.addInput(names.asOutput(scope));
@@ -106,17 +105,17 @@ public static ParseExample create(Scope scope, Operand<TString> serialized, Oper
     opBuilder.addInputList(Operands.asOutputs(scope, denseDefaults));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("num_sparse", numSparse);
-    DataType[] sparseTypesArray = new DataType[sparseTypes.size()];
+    Class[] sparseTypesArray = new Class[sparseTypes.size()];
     for (int i = 0; i < sparseTypesArray.length; ++i) {
       sparseTypesArray[i] = sparseTypes.get(i);
     }
     opBuilder.setAttr("sparse_types", sparseTypesArray);
-    DataType[] raggedValueTypesArray = new DataType[raggedValueTypes.size()];
+    Class[] raggedValueTypesArray = new Class[raggedValueTypes.size()];
     for (int i = 0; i < raggedValueTypesArray.length; ++i) {
       raggedValueTypesArray[i] = raggedValueTypes.get(i);
     }
     opBuilder.setAttr("ragged_value_types", raggedValueTypesArray);
-    DataType[] raggedSplitTypesArray = new DataType[raggedSplitTypes.size()];
+    Class[] raggedSplitTypesArray = new Class[raggedSplitTypes.size()];
     for (int i = 0; i < raggedSplitTypesArray.length; ++i) {
       raggedSplitTypesArray[i] = raggedSplitTypes.get(i);
     }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ParseSequenceExample.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ParseSequenceExample.java
index d1a60de633f..fc3d57c9e49 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ParseSequenceExample.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ParseSequenceExample.java
@@ -19,7 +19,6 @@
 
 import java.util.Arrays;
 import java.util.List;
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -152,7 +151,7 @@ private Options() {
    * @return a new instance of ParseSequenceExample
    */
   @Endpoint(describeByClass = true)
-  public static ParseSequenceExample create(Scope scope, Operand<TString> serialized, Operand<TString> debugName, Operand<TString> contextSparseKeys, Operand<TString> contextDenseKeys, Operand<TString> contextRaggedKeys, Operand<TString> featureListSparseKeys, Operand<TString> featureListDenseKeys, Operand<TString> featureListRaggedKeys, Operand<TBool> featureListDenseMissingAssumedEmpty, Iterable<Operand<?>> contextDenseDefaults, List<DataType<?>> contextSparseTypes, List<DataType<?>> contextRaggedValueTypes, List<DataType<?>> contextRaggedSplitTypes, List<DataType<?>> featureListDenseTypes, List<DataType<?>> featureListSparseTypes, List<DataType<?>> featureListRaggedValueTypes, List<DataType<?>> featureListRaggedSplitTypes, Options... options) {
+  public static ParseSequenceExample create(Scope scope, Operand<TString> serialized, Operand<TString> debugName, Operand<TString> contextSparseKeys, Operand<TString> contextDenseKeys, Operand<TString> contextRaggedKeys, Operand<TString> featureListSparseKeys, Operand<TString> featureListDenseKeys, Operand<TString> featureListRaggedKeys, Operand<TBool> featureListDenseMissingAssumedEmpty, Iterable<Operand<?>> contextDenseDefaults, List<Class<?>> contextSparseTypes, List<Class<?>> contextRaggedValueTypes, List<Class<?>> contextRaggedSplitTypes, List<Class<?>> featureListDenseTypes, List<Class<?>> featureListSparseTypes, List<Class<?>> featureListRaggedValueTypes, List<Class<?>> featureListRaggedSplitTypes, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ParseSequenceExampleV2", scope.makeOpName("ParseSequenceExample"));
     opBuilder.addInput(serialized.asOutput(scope));
     opBuilder.addInput(debugName.asOutput(scope));
@@ -165,37 +164,37 @@ public static ParseSequenceExample create(Scope scope, Operand<TString> serializ
     opBuilder.addInput(featureListDenseMissingAssumedEmpty.asOutput(scope));
     opBuilder.addInputList(Operands.asOutputs(scope, contextDenseDefaults));
     opBuilder = scope.applyControlDependencies(opBuilder);
-    DataType[] contextSparseTypesArray = new DataType[contextSparseTypes.size()];
+    Class[] contextSparseTypesArray = new Class[contextSparseTypes.size()];
     for (int i = 0; i < contextSparseTypesArray.length; ++i) {
       contextSparseTypesArray[i] = contextSparseTypes.get(i);
     }
     opBuilder.setAttr("context_sparse_types", contextSparseTypesArray);
-    DataType[] contextRaggedValueTypesArray = new DataType[contextRaggedValueTypes.size()];
+    Class[] contextRaggedValueTypesArray = new Class[contextRaggedValueTypes.size()];
     for (int i = 0; i < contextRaggedValueTypesArray.length; ++i) {
       contextRaggedValueTypesArray[i] = contextRaggedValueTypes.get(i);
     }
     opBuilder.setAttr("context_ragged_value_types", contextRaggedValueTypesArray);
-    DataType[] contextRaggedSplitTypesArray = new DataType[contextRaggedSplitTypes.size()];
+    Class[] contextRaggedSplitTypesArray = new Class[contextRaggedSplitTypes.size()];
     for (int i = 0; i < contextRaggedSplitTypesArray.length; ++i) {
       contextRaggedSplitTypesArray[i] = contextRaggedSplitTypes.get(i);
     }
     opBuilder.setAttr("context_ragged_split_types", contextRaggedSplitTypesArray);
-    DataType[] featureListDenseTypesArray = new DataType[featureListDenseTypes.size()];
+    Class[] featureListDenseTypesArray = new Class[featureListDenseTypes.size()];
     for (int i = 0; i < featureListDenseTypesArray.length; ++i) {
       featureListDenseTypesArray[i] = featureListDenseTypes.get(i);
     }
     opBuilder.setAttr("feature_list_dense_types", featureListDenseTypesArray);
-    DataType[] featureListSparseTypesArray = new DataType[featureListSparseTypes.size()];
+    Class[] featureListSparseTypesArray = new Class[featureListSparseTypes.size()];
     for (int i = 0; i < featureListSparseTypesArray.length; ++i) {
       featureListSparseTypesArray[i] = featureListSparseTypes.get(i);
     }
     opBuilder.setAttr("feature_list_sparse_types", featureListSparseTypesArray);
-    DataType[] featureListRaggedValueTypesArray = new DataType[featureListRaggedValueTypes.size()];
+    Class[] featureListRaggedValueTypesArray = new Class[featureListRaggedValueTypes.size()];
     for (int i = 0; i < featureListRaggedValueTypesArray.length; ++i) {
       featureListRaggedValueTypesArray[i] = featureListRaggedValueTypes.get(i);
     }
     opBuilder.setAttr("feature_list_ragged_value_types", featureListRaggedValueTypesArray);
-    DataType[] featureListRaggedSplitTypesArray = new DataType[featureListRaggedSplitTypes.size()];
+    Class[] featureListRaggedSplitTypesArray = new Class[featureListRaggedSplitTypes.size()];
     for (int i = 0; i < featureListRaggedSplitTypesArray.length; ++i) {
       featureListRaggedSplitTypesArray[i] = featureListRaggedSplitTypes.get(i);
     }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ParseSingleExample.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ParseSingleExample.java
index c05bd4f21ce..a51abe0e89e 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ParseSingleExample.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ParseSingleExample.java
@@ -19,7 +19,6 @@
 
 import java.util.Arrays;
 import java.util.List;
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -76,7 +75,7 @@ public final class ParseSingleExample extends RawOp {
    * @return a new instance of ParseSingleExample
    */
   @Endpoint(describeByClass = true)
-  public static ParseSingleExample create(Scope scope, Operand<TString> serialized, Iterable<Operand<?>> denseDefaults, Long numSparse, List<String> sparseKeys, List<String> denseKeys, List<DataType<?>> sparseTypes, List<Shape> denseShapes) {
+  public static ParseSingleExample create(Scope scope, Operand<TString> serialized, Iterable<Operand<?>> denseDefaults, Long numSparse, List<String> sparseKeys, List<String> denseKeys, List<Class<?>> sparseTypes, List<Shape> denseShapes) {
     OperationBuilder opBuilder = scope.env().opBuilder("ParseSingleExample", scope.makeOpName("ParseSingleExample"));
     opBuilder.addInput(serialized.asOutput(scope));
     opBuilder.addInputList(Operands.asOutputs(scope, denseDefaults));
@@ -92,7 +91,7 @@ public static ParseSingleExample create(Scope scope, Operand<TString> serialized
       denseKeysArray[i] = denseKeys.get(i);
     }
     opBuilder.setAttr("dense_keys", denseKeysArray);
-    DataType[] sparseTypesArray = new DataType[sparseTypes.size()];
+    Class[] sparseTypesArray = new Class[sparseTypes.size()];
     for (int i = 0; i < sparseTypesArray.length; ++i) {
       sparseTypesArray[i] = sparseTypes.get(i);
     }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ParseSingleSequenceExample.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ParseSingleSequenceExample.java
index 96a8bd9c413..55fdbf10265 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ParseSingleSequenceExample.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ParseSingleSequenceExample.java
@@ -19,7 +19,6 @@
 
 import java.util.Arrays;
 import java.util.List;
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -122,7 +121,7 @@ private Options() {
    * @return a new instance of ParseSingleSequenceExample
    */
   @Endpoint(describeByClass = true)
-  public static ParseSingleSequenceExample create(Scope scope, Operand<TString> serialized, Operand<TString> featureListDenseMissingAssumedEmpty, Iterable<Operand<TString>> contextSparseKeys, Iterable<Operand<TString>> contextDenseKeys, Iterable<Operand<TString>> featureListSparseKeys, Iterable<Operand<TString>> featureListDenseKeys, Iterable<Operand<?>> contextDenseDefaults, Operand<TString> debugName, List<DataType<?>> contextSparseTypes, List<DataType<?>> featureListDenseTypes, List<DataType<?>> featureListSparseTypes, Options... options) {
+  public static ParseSingleSequenceExample create(Scope scope, Operand<TString> serialized, Operand<TString> featureListDenseMissingAssumedEmpty, Iterable<Operand<TString>> contextSparseKeys, Iterable<Operand<TString>> contextDenseKeys, Iterable<Operand<TString>> featureListSparseKeys, Iterable<Operand<TString>> featureListDenseKeys, Iterable<Operand<?>> contextDenseDefaults, Operand<TString> debugName, List<Class<?>> contextSparseTypes, List<Class<?>> featureListDenseTypes, List<Class<?>> featureListSparseTypes, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ParseSingleSequenceExample", scope.makeOpName("ParseSingleSequenceExample"));
     opBuilder.addInput(serialized.asOutput(scope));
     opBuilder.addInput(featureListDenseMissingAssumedEmpty.asOutput(scope));
@@ -133,17 +132,17 @@ public static ParseSingleSequenceExample create(Scope scope, Operand<TString> se
     opBuilder.addInputList(Operands.asOutputs(scope, contextDenseDefaults));
     opBuilder.addInput(debugName.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
-    DataType[] contextSparseTypesArray = new DataType[contextSparseTypes.size()];
+    Class[] contextSparseTypesArray = new Class[contextSparseTypes.size()];
     for (int i = 0; i < contextSparseTypesArray.length; ++i) {
       contextSparseTypesArray[i] = contextSparseTypes.get(i);
     }
     opBuilder.setAttr("context_sparse_types", contextSparseTypesArray);
-    DataType[] featureListDenseTypesArray = new DataType[featureListDenseTypes.size()];
+    Class[] featureListDenseTypesArray = new Class[featureListDenseTypes.size()];
     for (int i = 0; i < featureListDenseTypesArray.length; ++i) {
       featureListDenseTypesArray[i] = featureListDenseTypes.get(i);
     }
     opBuilder.setAttr("feature_list_dense_types", featureListDenseTypesArray);
-    DataType[] featureListSparseTypesArray = new DataType[featureListSparseTypes.size()];
+    Class[] featureListSparseTypesArray = new Class[featureListSparseTypes.size()];
     for (int i = 0; i < featureListSparseTypesArray.length; ++i) {
       featureListSparseTypesArray[i] = featureListSparseTypes.get(i);
     }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ParseTensor.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ParseTensor.java
index 1fe2d9cebaf..2a560908647 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ParseTensor.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ParseTensor.java
@@ -17,7 +17,6 @@
 
 package org.tensorflow.op.io;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -47,7 +46,7 @@ public final class ParseTensor<T extends TType> extends RawOp implements Operand
    * @return a new instance of ParseTensor
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> ParseTensor<T> create(Scope scope, Operand<TString> serialized, DataType<T> outType) {
+  public static <T extends TType> ParseTensor<T> create(Scope scope, Operand<TString> serialized, Class<T> outType) {
     OperationBuilder opBuilder = scope.env().opBuilder("ParseTensor", scope.makeOpName("ParseTensor"));
     opBuilder.addInput(serialized.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/PriorityQueue.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/PriorityQueue.java
index f3abb008e73..5e2bd63fec8 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/PriorityQueue.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/PriorityQueue.java
@@ -18,7 +18,6 @@
 package org.tensorflow.op.io;
 
 import java.util.List;
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -95,10 +94,10 @@ private Options() {
    * @return a new instance of PriorityQueue
    */
   @Endpoint(describeByClass = true)
-  public static PriorityQueue create(Scope scope, List<DataType<?>> componentTypes, List<Shape> shapes, Options... options) {
+  public static PriorityQueue create(Scope scope, List<Class<?>> componentTypes, List<Shape> shapes, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("PriorityQueueV2", scope.makeOpName("PriorityQueue"));
     opBuilder = scope.applyControlDependencies(opBuilder);
-    DataType[] componentTypesArray = new DataType[componentTypes.size()];
+    Class[] componentTypesArray = new Class[componentTypes.size()];
     for (int i = 0; i < componentTypesArray.length; ++i) {
       componentTypesArray[i] = componentTypes.get(i);
     }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/QueueDequeue.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/QueueDequeue.java
index 2da189694d7..59ea5a8af61 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/QueueDequeue.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/QueueDequeue.java
@@ -20,7 +20,6 @@
 import java.util.Arrays;
 import java.util.Iterator;
 import java.util.List;
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -75,11 +74,11 @@ private Options() {
    * @return a new instance of QueueDequeue
    */
   @Endpoint(describeByClass = true)
-  public static QueueDequeue create(Scope scope, Operand<?> handle, List<DataType<?>> componentTypes, Options... options) {
+  public static QueueDequeue create(Scope scope, Operand<?> handle, List<Class<?>> componentTypes, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("QueueDequeueV2", scope.makeOpName("QueueDequeue"));
     opBuilder.addInput(handle.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
-    DataType[] componentTypesArray = new DataType[componentTypes.size()];
+    Class[] componentTypesArray = new Class[componentTypes.size()];
     for (int i = 0; i < componentTypesArray.length; ++i) {
       componentTypesArray[i] = componentTypes.get(i);
     }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/QueueDequeueMany.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/QueueDequeueMany.java
index 18d79dfa2de..b40686b18e6 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/QueueDequeueMany.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/QueueDequeueMany.java
@@ -20,7 +20,6 @@
 import java.util.Arrays;
 import java.util.Iterator;
 import java.util.List;
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -84,12 +83,12 @@ private Options() {
    * @return a new instance of QueueDequeueMany
    */
   @Endpoint(describeByClass = true)
-  public static QueueDequeueMany create(Scope scope, Operand<?> handle, Operand<TInt32> n, List<DataType<?>> componentTypes, Options... options) {
+  public static QueueDequeueMany create(Scope scope, Operand<?> handle, Operand<TInt32> n, List<Class<?>> componentTypes, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("QueueDequeueManyV2", scope.makeOpName("QueueDequeueMany"));
     opBuilder.addInput(handle.asOutput(scope));
     opBuilder.addInput(n.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
-    DataType[] componentTypesArray = new DataType[componentTypes.size()];
+    Class[] componentTypesArray = new Class[componentTypes.size()];
     for (int i = 0; i < componentTypesArray.length; ++i) {
       componentTypesArray[i] = componentTypes.get(i);
     }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/QueueDequeueUpTo.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/QueueDequeueUpTo.java
index 4358a202318..82109f028d9 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/QueueDequeueUpTo.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/QueueDequeueUpTo.java
@@ -20,7 +20,6 @@
 import java.util.Arrays;
 import java.util.Iterator;
 import java.util.List;
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -88,12 +87,12 @@ private Options() {
    * @return a new instance of QueueDequeueUpTo
    */
   @Endpoint(describeByClass = true)
-  public static QueueDequeueUpTo create(Scope scope, Operand<?> handle, Operand<TInt32> n, List<DataType<?>> componentTypes, Options... options) {
+  public static QueueDequeueUpTo create(Scope scope, Operand<?> handle, Operand<TInt32> n, List<Class<?>> componentTypes, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("QueueDequeueUpToV2", scope.makeOpName("QueueDequeueUpTo"));
     opBuilder.addInput(handle.asOutput(scope));
     opBuilder.addInput(n.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
-    DataType[] componentTypesArray = new DataType[componentTypes.size()];
+    Class[] componentTypesArray = new Class[componentTypes.size()];
     for (int i = 0; i < componentTypesArray.length; ++i) {
       componentTypesArray[i] = componentTypes.get(i);
     }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/RandomShuffleQueue.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/RandomShuffleQueue.java
index 42bfd44f9ad..16dd63c9b92 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/RandomShuffleQueue.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/RandomShuffleQueue.java
@@ -18,7 +18,6 @@
 package org.tensorflow.op.io;
 
 import java.util.List;
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -127,10 +126,10 @@ private Options() {
    * @return a new instance of RandomShuffleQueue
    */
   @Endpoint(describeByClass = true)
-  public static RandomShuffleQueue create(Scope scope, List<DataType<?>> componentTypes, Options... options) {
+  public static RandomShuffleQueue create(Scope scope, List<Class<?>> componentTypes, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("RandomShuffleQueueV2", scope.makeOpName("RandomShuffleQueue"));
     opBuilder = scope.applyControlDependencies(opBuilder);
-    DataType[] componentTypesArray = new DataType[componentTypes.size()];
+    Class[] componentTypesArray = new Class[componentTypes.size()];
     for (int i = 0; i < componentTypesArray.length; ++i) {
       componentTypesArray[i] = componentTypes.get(i);
     }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/SerializeManySparse.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/SerializeManySparse.java
index a29e431d44c..2acc3cef171 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/SerializeManySparse.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/SerializeManySparse.java
@@ -17,7 +17,6 @@
 
 package org.tensorflow.op.io;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -58,7 +57,7 @@ public final class SerializeManySparse<U extends TType> extends RawOp implements
    * @return a new instance of SerializeManySparse
    */
   @Endpoint(describeByClass = true)
-  public static <U extends TType, T extends TType> SerializeManySparse<U> create(Scope scope, Operand<TInt64> sparseIndices, Operand<T> sparseValues, Operand<TInt64> sparseShape, DataType<U> outType) {
+  public static <U extends TType, T extends TType> SerializeManySparse<U> create(Scope scope, Operand<TInt64> sparseIndices, Operand<T> sparseValues, Operand<TInt64> sparseShape, Class<U> outType) {
     OperationBuilder opBuilder = scope.env().opBuilder("SerializeManySparse", scope.makeOpName("SerializeManySparse"));
     opBuilder.addInput(sparseIndices.asOutput(scope));
     opBuilder.addInput(sparseValues.asOutput(scope));
@@ -79,7 +78,7 @@ public static <U extends TType, T extends TType> SerializeManySparse<U> create(S
    */
   @Endpoint(describeByClass = true)
   public static <T extends TType> SerializeManySparse<TString> create(Scope scope, Operand<TInt64> sparseIndices, Operand<T> sparseValues, Operand<TInt64> sparseShape) {
-    return create(scope, sparseIndices, sparseValues, sparseShape, TString.DTYPE);
+    return create(scope, sparseIndices, sparseValues, sparseShape, TString.class);
   }
   
   /**
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/SerializeSparse.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/SerializeSparse.java
index c4bdb3bbb74..c6935770458 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/SerializeSparse.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/SerializeSparse.java
@@ -17,7 +17,6 @@
 
 package org.tensorflow.op.io;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -50,7 +49,7 @@ public final class SerializeSparse<U extends TType> extends RawOp implements Ope
    * @return a new instance of SerializeSparse
    */
   @Endpoint(describeByClass = true)
-  public static <U extends TType, T extends TType> SerializeSparse<U> create(Scope scope, Operand<TInt64> sparseIndices, Operand<T> sparseValues, Operand<TInt64> sparseShape, DataType<U> outType) {
+  public static <U extends TType, T extends TType> SerializeSparse<U> create(Scope scope, Operand<TInt64> sparseIndices, Operand<T> sparseValues, Operand<TInt64> sparseShape, Class<U> outType) {
     OperationBuilder opBuilder = scope.env().opBuilder("SerializeSparse", scope.makeOpName("SerializeSparse"));
     opBuilder.addInput(sparseIndices.asOutput(scope));
     opBuilder.addInput(sparseValues.asOutput(scope));
@@ -71,7 +70,7 @@ public static <U extends TType, T extends TType> SerializeSparse<U> create(Scope
    */
   @Endpoint(describeByClass = true)
   public static <T extends TType> SerializeSparse<TString> create(Scope scope, Operand<TInt64> sparseIndices, Operand<T> sparseValues, Operand<TInt64> sparseShape) {
-    return create(scope, sparseIndices, sparseValues, sparseShape, TString.DTYPE);
+    return create(scope, sparseIndices, sparseValues, sparseShape, TString.class);
   }
   
   /**
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BandedTriangularSolve.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BandedTriangularSolve.java
index 8c1e184e6c9..cba49b57fb0 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BandedTriangularSolve.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BandedTriangularSolve.java
@@ -72,8 +72,8 @@ private Options() {
   @Endpoint(describeByClass = true)
   public static <T extends TType> BandedTriangularSolve<T> create(Scope scope, Operand<T> matrix, Operand<T> rhs, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("BandedTriangularSolve", scope.makeOpName("BandedTriangularSolve"));
-    opBuilder.addInput(matrix.asOutput());
-    opBuilder.addInput(rhs.asOutput());
+    opBuilder.addInput(matrix.asOutput(scope));
+    opBuilder.addInput(rhs.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -109,7 +109,7 @@ public Output<T> output() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Eig.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Eig.java
index 327f9be5396..5e84611590b 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Eig.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Eig.java
@@ -17,7 +17,6 @@
 
 package org.tensorflow.op.linalg;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -78,7 +77,7 @@ private Options() {
    * @return a new instance of Eig
    */
   @Endpoint(describeByClass = true)
-  public static <U extends TType, T extends TType> Eig<U> create(Scope scope, Operand<T> input, DataType<U> Tout, Options... options) {
+  public static <U extends TType, T extends TType> Eig<U> create(Scope scope, Operand<T> input, Class<U> Tout, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("Eig", scope.makeOpName("Eig"));
     opBuilder.addInput(input.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Lu.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Lu.java
index dad690bff5e..abc37991725 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Lu.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Lu.java
@@ -17,7 +17,6 @@
 
 package org.tensorflow.op.linalg;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -67,7 +66,7 @@ public final class Lu<T extends TType, U extends TNumber> extends RawOp {
    * @return a new instance of Lu
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> Lu<T, U> create(Scope scope, Operand<T> input, DataType<U> outputIdxType) {
+  public static <T extends TType, U extends TNumber> Lu<T, U> create(Scope scope, Operand<T> input, Class<U> outputIdxType) {
     OperationBuilder opBuilder = scope.env().opBuilder("Lu", scope.makeOpName("Lu"));
     opBuilder.addInput(input.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
@@ -85,7 +84,7 @@ public static <T extends TType, U extends TNumber> Lu<T, U> create(Scope scope,
    */
   @Endpoint(describeByClass = true)
   public static <T extends TType> Lu<T, TInt32> create(Scope scope, Operand<T> input) {
-    return create(scope, input, TInt32.DTYPE);
+    return create(scope, input, TInt32.class);
   }
   
   /**
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/QuantizedMatMul.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/QuantizedMatMul.java
index eaa128d3d35..e8fd87c1fb3 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/QuantizedMatMul.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/QuantizedMatMul.java
@@ -17,7 +17,6 @@
 
 package org.tensorflow.op.linalg;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -87,7 +86,7 @@ private Options() {
    * @return a new instance of QuantizedMatMul
    */
   @Endpoint(describeByClass = true)
-  public static <V extends TType, T extends TType, U extends TType, W extends TType> QuantizedMatMul<V> create(Scope scope, Operand<T> a, Operand<U> b, Operand<TFloat32> minA, Operand<TFloat32> maxA, Operand<TFloat32> minB, Operand<TFloat32> maxB, DataType<V> Toutput, DataType<W> Tactivation, Options... options) {
+  public static <V extends TType, T extends TType, U extends TType, W extends TType> QuantizedMatMul<V> create(Scope scope, Operand<T> a, Operand<U> b, Operand<TFloat32> minA, Operand<TFloat32> maxA, Operand<TFloat32> minB, Operand<TFloat32> maxB, Class<V> Toutput, Class<W> Tactivation, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("QuantizedMatMul", scope.makeOpName("QuantizedMatMul"));
     opBuilder.addInput(a.asOutput(scope));
     opBuilder.addInput(b.asOutput(scope));
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/QuantizedMatMulWithBias.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/QuantizedMatMulWithBias.java
index 775046e4353..faecc84f8cc 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/QuantizedMatMulWithBias.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/QuantizedMatMulWithBias.java
@@ -17,7 +17,6 @@
 
 package org.tensorflow.op.linalg;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -97,7 +96,7 @@ private Options() {
    * @return a new instance of QuantizedMatMulWithBias
    */
   @Endpoint(describeByClass = true)
-  public static <W extends TType, T extends TType, U extends TType, V extends TType> QuantizedMatMulWithBias<W> create(Scope scope, Operand<T> a, Operand<U> b, Operand<V> bias, Operand<TFloat32> minA, Operand<TFloat32> maxA, Operand<TFloat32> minB, Operand<TFloat32> maxB, DataType<W> Toutput, Options... options) {
+  public static <W extends TType, T extends TType, U extends TType, V extends TType> QuantizedMatMulWithBias<W> create(Scope scope, Operand<T> a, Operand<U> b, Operand<V> bias, Operand<TFloat32> minA, Operand<TFloat32> maxA, Operand<TFloat32> minB, Operand<TFloat32> maxB, Class<W> Toutput, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("QuantizedMatMulWithBias", scope.makeOpName("QuantizedMatMulWithBias"));
     opBuilder.addInput(a.asOutput(scope));
     opBuilder.addInput(b.asOutput(scope));
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/QuantizedMatMulWithBiasAndRelu.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/QuantizedMatMulWithBiasAndRelu.java
index 4612d13dbb7..55dbb0ba73a 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/QuantizedMatMulWithBiasAndRelu.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/QuantizedMatMulWithBiasAndRelu.java
@@ -17,7 +17,6 @@
 
 package org.tensorflow.op.linalg;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -98,7 +97,7 @@ private Options() {
    * @return a new instance of QuantizedMatMulWithBiasAndRelu
    */
   @Endpoint(describeByClass = true)
-  public static <V extends TType, T extends TType, U extends TType> QuantizedMatMulWithBiasAndRelu<V> create(Scope scope, Operand<T> a, Operand<U> b, Operand<TFloat32> bias, Operand<TFloat32> minA, Operand<TFloat32> maxA, Operand<TFloat32> minB, Operand<TFloat32> maxB, DataType<V> Toutput, Options... options) {
+  public static <V extends TType, T extends TType, U extends TType> QuantizedMatMulWithBiasAndRelu<V> create(Scope scope, Operand<T> a, Operand<U> b, Operand<TFloat32> bias, Operand<TFloat32> minA, Operand<TFloat32> maxA, Operand<TFloat32> minB, Operand<TFloat32> maxB, Class<V> Toutput, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("QuantizedMatMulWithBiasAndRelu", scope.makeOpName("QuantizedMatMulWithBiasAndRelu"));
     opBuilder.addInput(a.asOutput(scope));
     opBuilder.addInput(b.asOutput(scope));
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/QuantizedMatMulWithBiasAndReluAndRequantize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/QuantizedMatMulWithBiasAndReluAndRequantize.java
index 48cbdd88f21..8d0f81359f3 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/QuantizedMatMulWithBiasAndReluAndRequantize.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/QuantizedMatMulWithBiasAndReluAndRequantize.java
@@ -17,7 +17,6 @@
 
 package org.tensorflow.op.linalg;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -101,7 +100,7 @@ private Options() {
    * @return a new instance of QuantizedMatMulWithBiasAndReluAndRequantize
    */
   @Endpoint(describeByClass = true)
-  public static <W extends TType, T extends TType, U extends TType, V extends TType> QuantizedMatMulWithBiasAndReluAndRequantize<W> create(Scope scope, Operand<T> a, Operand<U> b, Operand<V> bias, Operand<TFloat32> minA, Operand<TFloat32> maxA, Operand<TFloat32> minB, Operand<TFloat32> maxB, Operand<TFloat32> minFreezedOutput, Operand<TFloat32> maxFreezedOutput, DataType<W> Toutput, Options... options) {
+  public static <W extends TType, T extends TType, U extends TType, V extends TType> QuantizedMatMulWithBiasAndReluAndRequantize<W> create(Scope scope, Operand<T> a, Operand<U> b, Operand<V> bias, Operand<TFloat32> minA, Operand<TFloat32> maxA, Operand<TFloat32> minB, Operand<TFloat32> maxB, Operand<TFloat32> minFreezedOutput, Operand<TFloat32> maxFreezedOutput, Class<W> Toutput, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("QuantizedMatMulWithBiasAndReluAndRequantize", scope.makeOpName("QuantizedMatMulWithBiasAndReluAndRequantize"));
     opBuilder.addInput(a.asOutput(scope));
     opBuilder.addInput(b.asOutput(scope));
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/CSRSparseMatrixComponents.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/CSRSparseMatrixComponents.java
index bb2dbb735ff..732a5641ab0 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/CSRSparseMatrixComponents.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/CSRSparseMatrixComponents.java
@@ -17,7 +17,6 @@
 
 package org.tensorflow.op.linalg.sparse;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -49,7 +48,7 @@ public final class CSRSparseMatrixComponents<T extends TType> extends RawOp {
    * @return a new instance of CSRSparseMatrixComponents
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> CSRSparseMatrixComponents<T> create(Scope scope, Operand<?> csrSparseMatrix, Operand<TInt32> index, DataType<T> type) {
+  public static <T extends TType> CSRSparseMatrixComponents<T> create(Scope scope, Operand<?> csrSparseMatrix, Operand<TInt32> index, Class<T> type) {
     OperationBuilder opBuilder = scope.env().opBuilder("CSRSparseMatrixComponents", scope.makeOpName("CSRSparseMatrixComponents"));
     opBuilder.addInput(csrSparseMatrix.asOutput(scope));
     opBuilder.addInput(index.asOutput(scope));
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/CSRSparseMatrixToDense.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/CSRSparseMatrixToDense.java
index 3c5caaeeb1b..b0e0275d49c 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/CSRSparseMatrixToDense.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/CSRSparseMatrixToDense.java
@@ -17,7 +17,6 @@
 
 package org.tensorflow.op.linalg.sparse;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -44,7 +43,7 @@ public final class CSRSparseMatrixToDense<T extends TType> extends RawOp impleme
    * @return a new instance of CSRSparseMatrixToDense
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> CSRSparseMatrixToDense<T> create(Scope scope, Operand<?> sparseInput, DataType<T> type) {
+  public static <T extends TType> CSRSparseMatrixToDense<T> create(Scope scope, Operand<?> sparseInput, Class<T> type) {
     OperationBuilder opBuilder = scope.env().opBuilder("CSRSparseMatrixToDense", scope.makeOpName("CSRSparseMatrixToDense"));
     opBuilder.addInput(sparseInput.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/CSRSparseMatrixToSparseTensor.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/CSRSparseMatrixToSparseTensor.java
index 365ef176926..5b89bc86a70 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/CSRSparseMatrixToSparseTensor.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/CSRSparseMatrixToSparseTensor.java
@@ -17,7 +17,6 @@
 
 package org.tensorflow.op.linalg.sparse;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -45,7 +44,7 @@ public final class CSRSparseMatrixToSparseTensor<T extends TType> extends RawOp
    * @return a new instance of CSRSparseMatrixToSparseTensor
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> CSRSparseMatrixToSparseTensor<T> create(Scope scope, Operand<?> sparseMatrix, DataType<T> type) {
+  public static <T extends TType> CSRSparseMatrixToSparseTensor<T> create(Scope scope, Operand<?> sparseMatrix, Class<T> type) {
     OperationBuilder opBuilder = scope.env().opBuilder("CSRSparseMatrixToSparseTensor", scope.makeOpName("CSRSparseMatrixToSparseTensor"));
     opBuilder.addInput(sparseMatrix.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixSoftmax.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixSoftmax.java
index bf04b321e7c..610706e5fa3 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixSoftmax.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixSoftmax.java
@@ -17,7 +17,6 @@
 
 package org.tensorflow.op.linalg.sparse;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -49,7 +48,7 @@ public final class SparseMatrixSoftmax extends RawOp implements Operand<TType> {
    * @return a new instance of SparseMatrixSoftmax
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> SparseMatrixSoftmax create(Scope scope, Operand<?> logits, DataType<T> type) {
+  public static <T extends TNumber> SparseMatrixSoftmax create(Scope scope, Operand<?> logits, Class<T> type) {
     OperationBuilder opBuilder = scope.env().opBuilder("SparseMatrixSoftmax", scope.makeOpName("SparseMatrixSoftmax"));
     opBuilder.addInput(logits.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixSoftmaxGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixSoftmaxGrad.java
index bdc65e0f1e7..4cfe45ff40d 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixSoftmaxGrad.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixSoftmaxGrad.java
@@ -17,7 +17,6 @@
 
 package org.tensorflow.op.linalg.sparse;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -44,7 +43,7 @@ public final class SparseMatrixSoftmaxGrad extends RawOp implements Operand<TTyp
    * @return a new instance of SparseMatrixSoftmaxGrad
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> SparseMatrixSoftmaxGrad create(Scope scope, Operand<?> softmax, Operand<?> gradSoftmax, DataType<T> type) {
+  public static <T extends TNumber> SparseMatrixSoftmaxGrad create(Scope scope, Operand<?> softmax, Operand<?> gradSoftmax, Class<T> type) {
     OperationBuilder opBuilder = scope.env().opBuilder("SparseMatrixSoftmaxGrad", scope.makeOpName("SparseMatrixSoftmaxGrad"));
     opBuilder.addInput(softmax.asOutput(scope));
     opBuilder.addInput(gradSoftmax.asOutput(scope));
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixSparseCholesky.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixSparseCholesky.java
index c92f753bdad..173275ccf57 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixSparseCholesky.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixSparseCholesky.java
@@ -17,7 +17,6 @@
 
 package org.tensorflow.op.linalg.sparse;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -115,7 +114,7 @@ public final class SparseMatrixSparseCholesky extends RawOp implements Operand<T
    * @return a new instance of SparseMatrixSparseCholesky
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> SparseMatrixSparseCholesky create(Scope scope, Operand<?> input, Operand<TInt32> permutation, DataType<T> type) {
+  public static <T extends TType> SparseMatrixSparseCholesky create(Scope scope, Operand<?> input, Operand<TInt32> permutation, Class<T> type) {
     OperationBuilder opBuilder = scope.env().opBuilder("SparseMatrixSparseCholesky", scope.makeOpName("SparseMatrixSparseCholesky"));
     opBuilder.addInput(input.asOutput(scope));
     opBuilder.addInput(permutation.asOutput(scope));
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixSparseMatMul.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixSparseMatMul.java
index dbd4c82fb98..ec46546119d 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixSparseMatMul.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixSparseMatMul.java
@@ -17,7 +17,6 @@
 
 package org.tensorflow.op.linalg.sparse;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -163,7 +162,7 @@ private Options() {
    * @return a new instance of SparseMatrixSparseMatMul
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> SparseMatrixSparseMatMul create(Scope scope, Operand<?> a, Operand<?> b, DataType<T> type, Options... options) {
+  public static <T extends TType> SparseMatrixSparseMatMul create(Scope scope, Operand<?> a, Operand<?> b, Class<T> type, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("SparseMatrixSparseMatMul", scope.makeOpName("SparseMatrixSparseMatMul"));
     opBuilder.addInput(a.asOutput(scope));
     opBuilder.addInput(b.asOutput(scope));
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixTranspose.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixTranspose.java
index c858a37383a..554c506be2a 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixTranspose.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixTranspose.java
@@ -17,7 +17,6 @@
 
 package org.tensorflow.op.linalg.sparse;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -65,7 +64,7 @@ private Options() {
    * @return a new instance of SparseMatrixTranspose
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> SparseMatrixTranspose create(Scope scope, Operand<?> input, DataType<T> type, Options... options) {
+  public static <T extends TType> SparseMatrixTranspose create(Scope scope, Operand<?> input, Class<T> type, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("SparseMatrixTranspose", scope.makeOpName("SparseMatrixTranspose"));
     opBuilder.addInput(input.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixZeros.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixZeros.java
index bdb0ac426d7..33cac876a73 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixZeros.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixZeros.java
@@ -17,7 +17,6 @@
 
 package org.tensorflow.op.linalg.sparse;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -43,7 +42,7 @@ public final class SparseMatrixZeros extends RawOp implements Operand<TType> {
    * @return a new instance of SparseMatrixZeros
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> SparseMatrixZeros create(Scope scope, Operand<TInt64> denseShape, DataType<T> type) {
+  public static <T extends TType> SparseMatrixZeros create(Scope scope, Operand<TInt64> denseShape, Class<T> type) {
     OperationBuilder opBuilder = scope.env().opBuilder("SparseMatrixZeros", scope.makeOpName("SparseMatrixZeros"));
     opBuilder.addInput(denseShape.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Angle.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Angle.java
index 5c1d9b6919b..4d4d662f614 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Angle.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Angle.java
@@ -17,7 +17,6 @@
 
 package org.tensorflow.op.math;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -63,7 +62,7 @@ public final class Angle<U extends TNumber> extends RawOp implements Operand<U>
    * @return a new instance of Angle
    */
   @Endpoint(describeByClass = true)
-  public static <U extends TNumber, T extends TType> Angle<U> create(Scope scope, Operand<T> input, DataType<U> Tout) {
+  public static <U extends TNumber, T extends TType> Angle<U> create(Scope scope, Operand<T> input, Class<U> Tout) {
     OperationBuilder opBuilder = scope.env().opBuilder("Angle", scope.makeOpName("Angle"));
     opBuilder.addInput(input.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
@@ -80,7 +79,7 @@ public static <U extends TNumber, T extends TType> Angle<U> create(Scope scope,
    */
   @Endpoint(describeByClass = true)
   public static <T extends TType> Angle<TFloat32> create(Scope scope, Operand<T> input) {
-    return create(scope, input, TFloat32.DTYPE);
+    return create(scope, input, TFloat32.class);
   }
   
   /**
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/ArgMax.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/ArgMax.java
index 955d47f8ba3..f5348fd11d3 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/ArgMax.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/ArgMax.java
@@ -17,7 +17,6 @@
 
 package org.tensorflow.op.math;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -63,7 +62,7 @@ public final class ArgMax<V extends TNumber> extends RawOp implements Operand<V>
    * @return a new instance of ArgMax
    */
   @Endpoint(describeByClass = true)
-  public static <V extends TNumber, T extends TType, U extends TNumber> ArgMax<V> create(Scope scope, Operand<T> input, Operand<U> dimension, DataType<V> outputType) {
+  public static <V extends TNumber, T extends TType, U extends TNumber> ArgMax<V> create(Scope scope, Operand<T> input, Operand<U> dimension, Class<V> outputType) {
     OperationBuilder opBuilder = scope.env().opBuilder("ArgMax", scope.makeOpName("ArgMax"));
     opBuilder.addInput(input.asOutput(scope));
     opBuilder.addInput(dimension.asOutput(scope));
@@ -84,7 +83,7 @@ public static <V extends TNumber, T extends TType, U extends TNumber> ArgMax<V>
    */
   @Endpoint(describeByClass = true)
   public static <T extends TType, U extends TNumber> ArgMax<TInt64> create(Scope scope, Operand<T> input, Operand<U> dimension) {
-    return create(scope, input, dimension, TInt64.DTYPE);
+    return create(scope, input, dimension, TInt64.class);
   }
   
   /**
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/ArgMin.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/ArgMin.java
index 73281d654b1..5bb45ff03b5 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/ArgMin.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/ArgMin.java
@@ -17,7 +17,6 @@
 
 package org.tensorflow.op.math;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -63,7 +62,7 @@ public final class ArgMin<V extends TNumber> extends RawOp implements Operand<V>
    * @return a new instance of ArgMin
    */
   @Endpoint(describeByClass = true)
-  public static <V extends TNumber, T extends TType, U extends TNumber> ArgMin<V> create(Scope scope, Operand<T> input, Operand<U> dimension, DataType<V> outputType) {
+  public static <V extends TNumber, T extends TType, U extends TNumber> ArgMin<V> create(Scope scope, Operand<T> input, Operand<U> dimension, Class<V> outputType) {
     OperationBuilder opBuilder = scope.env().opBuilder("ArgMin", scope.makeOpName("ArgMin"));
     opBuilder.addInput(input.asOutput(scope));
     opBuilder.addInput(dimension.asOutput(scope));
@@ -84,7 +83,7 @@ public static <V extends TNumber, T extends TType, U extends TNumber> ArgMin<V>
    */
   @Endpoint(describeByClass = true)
   public static <T extends TType, U extends TNumber> ArgMin<TInt64> create(Scope scope, Operand<T> input, Operand<U> dimension) {
-    return create(scope, input, dimension, TInt64.DTYPE);
+    return create(scope, input, dimension, TInt64.class);
   }
   
   /**
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/BesselI0.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/BesselI0.java
index 45dcd2b8e4c..ad238cc6aba 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/BesselI0.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/BesselI0.java
@@ -26,7 +26,6 @@
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * @param <T> data type for {@code y()} output
@@ -43,7 +42,7 @@ public final class BesselI0<T extends TNumber> extends RawOp implements Operand<
   @Endpoint(describeByClass = true)
   public static <T extends TNumber> BesselI0<T> create(Scope scope, Operand<T> x) {
     OperationBuilder opBuilder = scope.env().opBuilder("BesselI0", scope.makeOpName("BesselI0"));
-    opBuilder.addInput(x.asOutput());
+    opBuilder.addInput(x.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new BesselI0<T>(opBuilder.build());
   }
@@ -55,7 +54,7 @@ public Output<T> y() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return y;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/BesselI0e.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/BesselI0e.java
index fec91e77944..7bb31bfb405 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/BesselI0e.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/BesselI0e.java
@@ -28,16 +28,8 @@
 import org.tensorflow.types.family.TNumber;
 
 /**
- * Computes the Bessel i0e function of `x` element-wise.
- * <p>
- * Exponentially scaled modified Bessel function of order 0 defined as
- * `bessel_i0e(x) = exp(-abs(x)) bessel_i0(x)`.
- * <p>
- * This function is faster and numerically stabler than `bessel_i0(x)`.
- * 
  * @param <T> data type for {@code y()} output
  */
-@Operator(group = "math")
 public final class BesselI0e<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
@@ -66,6 +58,9 @@ public Output<T> asOutput(Scope scope) {
     return y;
   }
   
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "BesselI0e";
+  
   private Output<T> y;
   
   private BesselI0e(Operation operation) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/BesselI1.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/BesselI1.java
index 148758aa5a4..250d09d9d57 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/BesselI1.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/BesselI1.java
@@ -26,7 +26,6 @@
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * @param <T> data type for {@code y()} output
@@ -43,7 +42,7 @@ public final class BesselI1<T extends TNumber> extends RawOp implements Operand<
   @Endpoint(describeByClass = true)
   public static <T extends TNumber> BesselI1<T> create(Scope scope, Operand<T> x) {
     OperationBuilder opBuilder = scope.env().opBuilder("BesselI1", scope.makeOpName("BesselI1"));
-    opBuilder.addInput(x.asOutput());
+    opBuilder.addInput(x.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new BesselI1<T>(opBuilder.build());
   }
@@ -55,7 +54,7 @@ public Output<T> y() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return y;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/BesselI1e.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/BesselI1e.java
index 8d27f3b7bd1..7755807f940 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/BesselI1e.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/BesselI1e.java
@@ -28,16 +28,8 @@
 import org.tensorflow.types.family.TNumber;
 
 /**
- * Computes the Bessel i1e function of `x` element-wise.
- * <p>
- * Exponentially scaled modified Bessel function of order 0 defined as
- * `bessel_i1e(x) = exp(-abs(x)) bessel_i1(x)`.
- * <p>
- * This function is faster and numerically stabler than `bessel_i1(x)`.
- * 
  * @param <T> data type for {@code y()} output
  */
-@Operator(group = "math")
 public final class BesselI1e<T extends TNumber> extends RawOp implements Operand<T> {
   
   /**
@@ -66,6 +58,9 @@ public Output<T> asOutput(Scope scope) {
     return y;
   }
   
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "BesselI1e";
+  
   private Output<T> y;
   
   private BesselI1e(Operation operation) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/ComplexAbs.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/ComplexAbs.java
index a384396d001..be130a37a7c 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/ComplexAbs.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/ComplexAbs.java
@@ -17,7 +17,6 @@
 
 package org.tensorflow.op.math;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -52,7 +51,7 @@ public final class ComplexAbs<U extends TNumber> extends RawOp implements Operan
    * @return a new instance of ComplexAbs
    */
   @Endpoint(describeByClass = true)
-  public static <U extends TNumber, T extends TType> ComplexAbs<U> create(Scope scope, Operand<T> x, DataType<U> Tout) {
+  public static <U extends TNumber, T extends TType> ComplexAbs<U> create(Scope scope, Operand<T> x, Class<U> Tout) {
     OperationBuilder opBuilder = scope.env().opBuilder("ComplexAbs", scope.makeOpName("ComplexAbs"));
     opBuilder.addInput(x.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
@@ -69,7 +68,7 @@ public static <U extends TNumber, T extends TType> ComplexAbs<U> create(Scope sc
    */
   @Endpoint(describeByClass = true)
   public static <T extends TType> ComplexAbs<TFloat32> create(Scope scope, Operand<T> x) {
-    return create(scope, x, TFloat32.DTYPE);
+    return create(scope, x, TFloat32.class);
   }
   
   /**
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/DenseBincount.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/DenseBincount.java
index e38d559f4ae..9cf288414a5 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/DenseBincount.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/DenseBincount.java
@@ -26,7 +26,6 @@
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Counts the number of occurrences of each value in an integer array.
@@ -78,9 +77,9 @@ private Options() {
   @Endpoint(describeByClass = true)
   public static <U extends TNumber, T extends TNumber> DenseBincount<U> create(Scope scope, Operand<T> input, Operand<T> size, Operand<U> weights, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("DenseBincount", scope.makeOpName("DenseBincount"));
-    opBuilder.addInput(input.asOutput());
-    opBuilder.addInput(size.asOutput());
-    opBuilder.addInput(weights.asOutput());
+    opBuilder.addInput(input.asOutput(scope));
+    opBuilder.addInput(size.asOutput(scope));
+    opBuilder.addInput(weights.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -108,7 +107,7 @@ public Output<U> output() {
   }
   
   @Override
-  public Output<U> asOutput() {
+  public Output<U> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Imag.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Imag.java
index 3dd31bcc850..ce18da2105d 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Imag.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Imag.java
@@ -17,7 +17,6 @@
 
 package org.tensorflow.op.math;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -59,7 +58,7 @@ public final class Imag<U extends TNumber> extends RawOp implements Operand<U> {
    * @return a new instance of Imag
    */
   @Endpoint(describeByClass = true)
-  public static <U extends TNumber, T extends TType> Imag<U> create(Scope scope, Operand<T> input, DataType<U> Tout) {
+  public static <U extends TNumber, T extends TType> Imag<U> create(Scope scope, Operand<T> input, Class<U> Tout) {
     OperationBuilder opBuilder = scope.env().opBuilder("Imag", scope.makeOpName("Imag"));
     opBuilder.addInput(input.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
@@ -76,7 +75,7 @@ public static <U extends TNumber, T extends TType> Imag<U> create(Scope scope, O
    */
   @Endpoint(describeByClass = true)
   public static <T extends TType> Imag<TFloat32> create(Scope scope, Operand<T> input) {
-    return create(scope, input, TFloat32.DTYPE);
+    return create(scope, input, TFloat32.class);
   }
   
   /**
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/QuantizedAdd.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/QuantizedAdd.java
index a35716ae88c..cc36c2bfdd6 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/QuantizedAdd.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/QuantizedAdd.java
@@ -17,7 +17,6 @@
 
 package org.tensorflow.op.math;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -51,7 +50,7 @@ public final class QuantizedAdd<V extends TType> extends RawOp {
    * @return a new instance of QuantizedAdd
    */
   @Endpoint(describeByClass = true)
-  public static <V extends TType, T extends TType, U extends TType> QuantizedAdd<V> create(Scope scope, Operand<T> x, Operand<U> y, Operand<TFloat32> minX, Operand<TFloat32> maxX, Operand<TFloat32> minY, Operand<TFloat32> maxY, DataType<V> Toutput) {
+  public static <V extends TType, T extends TType, U extends TType> QuantizedAdd<V> create(Scope scope, Operand<T> x, Operand<U> y, Operand<TFloat32> minX, Operand<TFloat32> maxX, Operand<TFloat32> minY, Operand<TFloat32> maxY, Class<V> Toutput) {
     OperationBuilder opBuilder = scope.env().opBuilder("QuantizedAdd", scope.makeOpName("QuantizedAdd"));
     opBuilder.addInput(x.asOutput(scope));
     opBuilder.addInput(y.asOutput(scope));
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/QuantizedMul.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/QuantizedMul.java
index 89dd9d4fbb0..907e9721316 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/QuantizedMul.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/QuantizedMul.java
@@ -17,7 +17,6 @@
 
 package org.tensorflow.op.math;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -51,7 +50,7 @@ public final class QuantizedMul<V extends TType> extends RawOp {
    * @return a new instance of QuantizedMul
    */
   @Endpoint(describeByClass = true)
-  public static <V extends TType, T extends TType, U extends TType> QuantizedMul<V> create(Scope scope, Operand<T> x, Operand<U> y, Operand<TFloat32> minX, Operand<TFloat32> maxX, Operand<TFloat32> minY, Operand<TFloat32> maxY, DataType<V> Toutput) {
+  public static <V extends TType, T extends TType, U extends TType> QuantizedMul<V> create(Scope scope, Operand<T> x, Operand<U> y, Operand<TFloat32> minX, Operand<TFloat32> maxX, Operand<TFloat32> minY, Operand<TFloat32> maxY, Class<V> Toutput) {
     OperationBuilder opBuilder = scope.env().opBuilder("QuantizedMul", scope.makeOpName("QuantizedMul"));
     opBuilder.addInput(x.asOutput(scope));
     opBuilder.addInput(y.asOutput(scope));
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Real.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Real.java
index 651f4c41b83..72004717894 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Real.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Real.java
@@ -17,7 +17,6 @@
 
 package org.tensorflow.op.math;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -59,7 +58,7 @@ public final class Real<U extends TNumber> extends RawOp implements Operand<U> {
    * @return a new instance of Real
    */
   @Endpoint(describeByClass = true)
-  public static <U extends TNumber, T extends TType> Real<U> create(Scope scope, Operand<T> input, DataType<U> Tout) {
+  public static <U extends TNumber, T extends TType> Real<U> create(Scope scope, Operand<T> input, Class<U> Tout) {
     OperationBuilder opBuilder = scope.env().opBuilder("Real", scope.makeOpName("Real"));
     opBuilder.addInput(input.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
@@ -76,7 +75,7 @@ public static <U extends TNumber, T extends TType> Real<U> create(Scope scope, O
    */
   @Endpoint(describeByClass = true)
   public static <T extends TType> Real<TFloat32> create(Scope scope, Operand<T> input) {
-    return create(scope, input, TFloat32.DTYPE);
+    return create(scope, input, TFloat32.class);
   }
   
   /**
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/RequantizePerChannel.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/RequantizePerChannel.java
index bc1d250245c..81b5c105c5d 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/RequantizePerChannel.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/RequantizePerChannel.java
@@ -17,7 +17,6 @@
 
 package org.tensorflow.op.math;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -49,7 +48,7 @@ public final class RequantizePerChannel<U extends TType> extends RawOp {
    * @return a new instance of RequantizePerChannel
    */
   @Endpoint(describeByClass = true)
-  public static <U extends TType, T extends TType> RequantizePerChannel<U> create(Scope scope, Operand<T> input, Operand<TFloat32> inputMin, Operand<TFloat32> inputMax, Operand<TFloat32> requestedOutputMin, Operand<TFloat32> requestedOutputMax, DataType<U> outType) {
+  public static <U extends TType, T extends TType> RequantizePerChannel<U> create(Scope scope, Operand<T> input, Operand<TFloat32> inputMin, Operand<TFloat32> inputMax, Operand<TFloat32> requestedOutputMin, Operand<TFloat32> requestedOutputMax, Class<U> outType) {
     OperationBuilder opBuilder = scope.env().opBuilder("RequantizePerChannel", scope.makeOpName("RequantizePerChannel"));
     opBuilder.addInput(input.asOutput(scope));
     opBuilder.addInput(inputMin.asOutput(scope));
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SobolSample.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SobolSample.java
index bda81e6b304..f67da4dedf9 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SobolSample.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SobolSample.java
@@ -17,7 +17,6 @@
 
 package org.tensorflow.op.math;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -53,7 +52,7 @@ public final class SobolSample<T extends TNumber> extends RawOp implements Opera
    * @return a new instance of SobolSample
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> SobolSample<T> create(Scope scope, Operand<TInt32> dim, Operand<TInt32> numResults, Operand<TInt32> skip, DataType<T> dtype) {
+  public static <T extends TNumber> SobolSample<T> create(Scope scope, Operand<TInt32> dim, Operand<TInt32> numResults, Operand<TInt32> skip, Class<T> dtype) {
     OperationBuilder opBuilder = scope.env().opBuilder("SobolSample", scope.makeOpName("SobolSample"));
     opBuilder.addInput(dim.asOutput(scope));
     opBuilder.addInput(numResults.asOutput(scope));
@@ -76,7 +75,7 @@ public static <T extends TNumber> SobolSample<T> create(Scope scope, Operand<TIn
    */
   @Endpoint(describeByClass = true)
   public static SobolSample<TFloat32> create(Scope scope, Operand<TInt32> dim, Operand<TInt32> numResults, Operand<TInt32> skip) {
-    return create(scope, dim, numResults, skip, TFloat32.DTYPE);
+    return create(scope, dim, numResults, skip, TFloat32.class);
   }
   
   /**
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/BesselJ0.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/BesselJ0.java
index 8d2184a49cb..7f532e14f32 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/BesselJ0.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/BesselJ0.java
@@ -26,7 +26,6 @@
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * @param <T> data type for {@code y()} output
@@ -43,7 +42,7 @@ public final class BesselJ0<T extends TNumber> extends RawOp implements Operand<
   @Endpoint(describeByClass = true)
   public static <T extends TNumber> BesselJ0<T> create(Scope scope, Operand<T> x) {
     OperationBuilder opBuilder = scope.env().opBuilder("BesselJ0", scope.makeOpName("BesselJ0"));
-    opBuilder.addInput(x.asOutput());
+    opBuilder.addInput(x.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new BesselJ0<T>(opBuilder.build());
   }
@@ -55,7 +54,7 @@ public Output<T> y() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return y;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/BesselJ1.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/BesselJ1.java
index d8f9621a36c..d43d61e2a7b 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/BesselJ1.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/BesselJ1.java
@@ -26,7 +26,6 @@
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * @param <T> data type for {@code y()} output
@@ -43,7 +42,7 @@ public final class BesselJ1<T extends TNumber> extends RawOp implements Operand<
   @Endpoint(describeByClass = true)
   public static <T extends TNumber> BesselJ1<T> create(Scope scope, Operand<T> x) {
     OperationBuilder opBuilder = scope.env().opBuilder("BesselJ1", scope.makeOpName("BesselJ1"));
-    opBuilder.addInput(x.asOutput());
+    opBuilder.addInput(x.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new BesselJ1<T>(opBuilder.build());
   }
@@ -55,7 +54,7 @@ public Output<T> y() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return y;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/BesselK0.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/BesselK0.java
index eaae243f83f..23c71bde19d 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/BesselK0.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/BesselK0.java
@@ -26,7 +26,6 @@
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * @param <T> data type for {@code y()} output
@@ -43,7 +42,7 @@ public final class BesselK0<T extends TNumber> extends RawOp implements Operand<
   @Endpoint(describeByClass = true)
   public static <T extends TNumber> BesselK0<T> create(Scope scope, Operand<T> x) {
     OperationBuilder opBuilder = scope.env().opBuilder("BesselK0", scope.makeOpName("BesselK0"));
-    opBuilder.addInput(x.asOutput());
+    opBuilder.addInput(x.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new BesselK0<T>(opBuilder.build());
   }
@@ -55,7 +54,7 @@ public Output<T> y() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return y;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/BesselK0e.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/BesselK0e.java
index c57ae64e233..854c4c11c7d 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/BesselK0e.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/BesselK0e.java
@@ -26,7 +26,6 @@
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * @param <T> data type for {@code y()} output
@@ -43,7 +42,7 @@ public final class BesselK0e<T extends TNumber> extends RawOp implements Operand
   @Endpoint(describeByClass = true)
   public static <T extends TNumber> BesselK0e<T> create(Scope scope, Operand<T> x) {
     OperationBuilder opBuilder = scope.env().opBuilder("BesselK0e", scope.makeOpName("BesselK0e"));
-    opBuilder.addInput(x.asOutput());
+    opBuilder.addInput(x.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new BesselK0e<T>(opBuilder.build());
   }
@@ -55,7 +54,7 @@ public Output<T> y() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return y;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/BesselK1.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/BesselK1.java
index 1858d25fe3d..7e4a9b3339a 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/BesselK1.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/BesselK1.java
@@ -26,7 +26,6 @@
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * @param <T> data type for {@code y()} output
@@ -43,7 +42,7 @@ public final class BesselK1<T extends TNumber> extends RawOp implements Operand<
   @Endpoint(describeByClass = true)
   public static <T extends TNumber> BesselK1<T> create(Scope scope, Operand<T> x) {
     OperationBuilder opBuilder = scope.env().opBuilder("BesselK1", scope.makeOpName("BesselK1"));
-    opBuilder.addInput(x.asOutput());
+    opBuilder.addInput(x.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new BesselK1<T>(opBuilder.build());
   }
@@ -55,7 +54,7 @@ public Output<T> y() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return y;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/BesselK1e.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/BesselK1e.java
index e4a5cc23efd..0e012259687 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/BesselK1e.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/BesselK1e.java
@@ -26,7 +26,6 @@
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * @param <T> data type for {@code y()} output
@@ -43,7 +42,7 @@ public final class BesselK1e<T extends TNumber> extends RawOp implements Operand
   @Endpoint(describeByClass = true)
   public static <T extends TNumber> BesselK1e<T> create(Scope scope, Operand<T> x) {
     OperationBuilder opBuilder = scope.env().opBuilder("BesselK1e", scope.makeOpName("BesselK1e"));
-    opBuilder.addInput(x.asOutput());
+    opBuilder.addInput(x.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new BesselK1e<T>(opBuilder.build());
   }
@@ -55,7 +54,7 @@ public Output<T> y() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return y;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/BesselY0.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/BesselY0.java
index 9228d1b6145..318fd9d490a 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/BesselY0.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/BesselY0.java
@@ -26,7 +26,6 @@
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * @param <T> data type for {@code y()} output
@@ -43,7 +42,7 @@ public final class BesselY0<T extends TNumber> extends RawOp implements Operand<
   @Endpoint(describeByClass = true)
   public static <T extends TNumber> BesselY0<T> create(Scope scope, Operand<T> x) {
     OperationBuilder opBuilder = scope.env().opBuilder("BesselY0", scope.makeOpName("BesselY0"));
-    opBuilder.addInput(x.asOutput());
+    opBuilder.addInput(x.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new BesselY0<T>(opBuilder.build());
   }
@@ -55,7 +54,7 @@ public Output<T> y() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return y;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/BesselY1.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/BesselY1.java
index 0461416b808..954bc8a7768 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/BesselY1.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/BesselY1.java
@@ -26,7 +26,6 @@
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * @param <T> data type for {@code y()} output
@@ -43,7 +42,7 @@ public final class BesselY1<T extends TNumber> extends RawOp implements Operand<
   @Endpoint(describeByClass = true)
   public static <T extends TNumber> BesselY1<T> create(Scope scope, Operand<T> x) {
     OperationBuilder opBuilder = scope.env().opBuilder("BesselY1", scope.makeOpName("BesselY1"));
-    opBuilder.addInput(x.asOutput());
+    opBuilder.addInput(x.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new BesselY1<T>(opBuilder.build());
   }
@@ -55,7 +54,7 @@ public Output<T> y() {
   }
   
   @Override
-  public Output<T> asOutput() {
+  public Output<T> asOutput(Scope scope) {
     return y;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CudnnRnnParamsSize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CudnnRnnParamsSize.java
index cdffe933e65..3372899ee2f 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CudnnRnnParamsSize.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CudnnRnnParamsSize.java
@@ -17,7 +17,6 @@
 
 package org.tensorflow.op.nn;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -145,7 +144,7 @@ private Options() {
    * @return a new instance of CudnnRnnParamsSize
    */
   @Endpoint(describeByClass = true)
-  public static <U extends TNumber, T extends TNumber> CudnnRnnParamsSize<U> create(Scope scope, Operand<TInt32> numLayers, Operand<TInt32> numUnits, Operand<TInt32> inputSize, DataType<T> T, DataType<U> S, Options... options) {
+  public static <U extends TNumber, T extends TNumber> CudnnRnnParamsSize<U> create(Scope scope, Operand<TInt32> numLayers, Operand<TInt32> numUnits, Operand<TInt32> inputSize, Class<T> T, Class<U> S, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("CudnnRNNParamsSize", scope.makeOpName("CudnnRnnParamsSize"));
     opBuilder.addInput(numLayers.asOutput(scope));
     opBuilder.addInput(numUnits.asOutput(scope));
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPoolWithArgmax.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPoolWithArgmax.java
index 8e5fbba6a69..870e56fd600 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPoolWithArgmax.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPoolWithArgmax.java
@@ -18,7 +18,6 @@
 package org.tensorflow.op.nn;
 
 import java.util.List;
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -82,7 +81,7 @@ private Options() {
    * @return a new instance of MaxPoolWithArgmax
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber, U extends TNumber> MaxPoolWithArgmax<T, U> create(Scope scope, Operand<T> input, List<Long> ksize, List<Long> strides, DataType<U> Targmax, String padding, Options... options) {
+  public static <T extends TNumber, U extends TNumber> MaxPoolWithArgmax<T, U> create(Scope scope, Operand<T> input, List<Long> ksize, List<Long> strides, Class<U> Targmax, String padding, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("MaxPoolWithArgmax", scope.makeOpName("MaxPoolWithArgmax"));
     opBuilder.addInput(input.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
@@ -122,7 +121,7 @@ public static <T extends TNumber, U extends TNumber> MaxPoolWithArgmax<T, U> cre
    */
   @Endpoint(describeByClass = true)
   public static <T extends TNumber> MaxPoolWithArgmax<T, TInt64> create(Scope scope, Operand<T> input, List<Long> ksize, List<Long> strides, String padding, Options... options) {
-    return create(scope, input, ksize, strides, TInt64.DTYPE, padding, options);
+    return create(scope, input, ksize, strides, TInt64.class, padding, options);
   }
   
   /**
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedBatchNormWithGlobalNormalization.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedBatchNormWithGlobalNormalization.java
index 9c9e29fe78f..07caa55bf86 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedBatchNormWithGlobalNormalization.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedBatchNormWithGlobalNormalization.java
@@ -17,7 +17,6 @@
 
 package org.tensorflow.op.nn;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -73,7 +72,7 @@ public final class QuantizedBatchNormWithGlobalNormalization<U extends TType> ex
    * @return a new instance of QuantizedBatchNormWithGlobalNormalization
    */
   @Endpoint(describeByClass = true)
-  public static <U extends TType, T extends TType> QuantizedBatchNormWithGlobalNormalization<U> create(Scope scope, Operand<T> t, Operand<TFloat32> tMin, Operand<TFloat32> tMax, Operand<T> m, Operand<TFloat32> mMin, Operand<TFloat32> mMax, Operand<T> v, Operand<TFloat32> vMin, Operand<TFloat32> vMax, Operand<T> beta, Operand<TFloat32> betaMin, Operand<TFloat32> betaMax, Operand<T> gamma, Operand<TFloat32> gammaMin, Operand<TFloat32> gammaMax, DataType<U> outType, Float varianceEpsilon, Boolean scaleAfterNormalization) {
+  public static <U extends TType, T extends TType> QuantizedBatchNormWithGlobalNormalization<U> create(Scope scope, Operand<T> t, Operand<TFloat32> tMin, Operand<TFloat32> tMax, Operand<T> m, Operand<TFloat32> mMin, Operand<TFloat32> mMax, Operand<T> v, Operand<TFloat32> vMin, Operand<TFloat32> vMax, Operand<T> beta, Operand<TFloat32> betaMin, Operand<TFloat32> betaMax, Operand<T> gamma, Operand<TFloat32> gammaMin, Operand<TFloat32> gammaMax, Class<U> outType, Float varianceEpsilon, Boolean scaleAfterNormalization) {
     OperationBuilder opBuilder = scope.env().opBuilder("QuantizedBatchNormWithGlobalNormalization", scope.makeOpName("QuantizedBatchNormWithGlobalNormalization"));
     opBuilder.addInput(t.asOutput(scope));
     opBuilder.addInput(tMin.asOutput(scope));
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedBiasAdd.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedBiasAdd.java
index a1e464190bb..fddae6e6272 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedBiasAdd.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedBiasAdd.java
@@ -17,7 +17,6 @@
 
 package org.tensorflow.op.nn;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -53,7 +52,7 @@ public final class QuantizedBiasAdd<V extends TType> extends RawOp {
    * @return a new instance of QuantizedBiasAdd
    */
   @Endpoint(describeByClass = true)
-  public static <V extends TType, T extends TType, U extends TType> QuantizedBiasAdd<V> create(Scope scope, Operand<T> input, Operand<U> bias, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, Operand<TFloat32> minBias, Operand<TFloat32> maxBias, DataType<V> outType) {
+  public static <V extends TType, T extends TType, U extends TType> QuantizedBiasAdd<V> create(Scope scope, Operand<T> input, Operand<U> bias, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, Operand<TFloat32> minBias, Operand<TFloat32> maxBias, Class<V> outType) {
     OperationBuilder opBuilder = scope.env().opBuilder("QuantizedBiasAdd", scope.makeOpName("QuantizedBiasAdd"));
     opBuilder.addInput(input.asOutput(scope));
     opBuilder.addInput(bias.asOutput(scope));
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DAndRelu.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DAndRelu.java
index 12f565c6a8a..071f2b0269e 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DAndRelu.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DAndRelu.java
@@ -18,7 +18,6 @@
 package org.tensorflow.op.nn;
 
 import java.util.List;
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -80,7 +79,7 @@ private Options() {
    * @return a new instance of QuantizedConv2DAndRelu
    */
   @Endpoint(describeByClass = true)
-  public static <V extends TType, T extends TType, U extends TType> QuantizedConv2DAndRelu<V> create(Scope scope, Operand<T> input, Operand<U> filter, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, Operand<TFloat32> minFilter, Operand<TFloat32> maxFilter, DataType<V> outType, List<Long> strides, String padding, Options... options) {
+  public static <V extends TType, T extends TType, U extends TType> QuantizedConv2DAndRelu<V> create(Scope scope, Operand<T> input, Operand<U> filter, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, Operand<TFloat32> minFilter, Operand<TFloat32> maxFilter, Class<V> outType, List<Long> strides, String padding, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("QuantizedConv2DAndRelu", scope.makeOpName("QuantizedConv2DAndRelu"));
     opBuilder.addInput(input.asOutput(scope));
     opBuilder.addInput(filter.asOutput(scope));
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DAndReluAndRequantize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DAndReluAndRequantize.java
index a7fe326d647..e1909629558 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DAndReluAndRequantize.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DAndReluAndRequantize.java
@@ -18,7 +18,6 @@
 package org.tensorflow.op.nn;
 
 import java.util.List;
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -82,7 +81,7 @@ private Options() {
    * @return a new instance of QuantizedConv2DAndReluAndRequantize
    */
   @Endpoint(describeByClass = true)
-  public static <V extends TType, T extends TType, U extends TType> QuantizedConv2DAndReluAndRequantize<V> create(Scope scope, Operand<T> input, Operand<U> filter, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, Operand<TFloat32> minFilter, Operand<TFloat32> maxFilter, Operand<TFloat32> minFreezedOutput, Operand<TFloat32> maxFreezedOutput, DataType<V> outType, List<Long> strides, String padding, Options... options) {
+  public static <V extends TType, T extends TType, U extends TType> QuantizedConv2DAndReluAndRequantize<V> create(Scope scope, Operand<T> input, Operand<U> filter, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, Operand<TFloat32> minFilter, Operand<TFloat32> maxFilter, Operand<TFloat32> minFreezedOutput, Operand<TFloat32> maxFreezedOutput, Class<V> outType, List<Long> strides, String padding, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("QuantizedConv2DAndReluAndRequantize", scope.makeOpName("QuantizedConv2DAndReluAndRequantize"));
     opBuilder.addInput(input.asOutput(scope));
     opBuilder.addInput(filter.asOutput(scope));
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DAndRequantize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DAndRequantize.java
index 6ec1e000d71..0964c0e98fa 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DAndRequantize.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DAndRequantize.java
@@ -18,7 +18,6 @@
 package org.tensorflow.op.nn;
 
 import java.util.List;
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -82,7 +81,7 @@ private Options() {
    * @return a new instance of QuantizedConv2DAndRequantize
    */
   @Endpoint(describeByClass = true)
-  public static <V extends TType, T extends TType, U extends TType> QuantizedConv2DAndRequantize<V> create(Scope scope, Operand<T> input, Operand<U> filter, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, Operand<TFloat32> minFilter, Operand<TFloat32> maxFilter, Operand<TFloat32> minFreezedOutput, Operand<TFloat32> maxFreezedOutput, DataType<V> outType, List<Long> strides, String padding, Options... options) {
+  public static <V extends TType, T extends TType, U extends TType> QuantizedConv2DAndRequantize<V> create(Scope scope, Operand<T> input, Operand<U> filter, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, Operand<TFloat32> minFilter, Operand<TFloat32> maxFilter, Operand<TFloat32> minFreezedOutput, Operand<TFloat32> maxFreezedOutput, Class<V> outType, List<Long> strides, String padding, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("QuantizedConv2DAndRequantize", scope.makeOpName("QuantizedConv2DAndRequantize"));
     opBuilder.addInput(input.asOutput(scope));
     opBuilder.addInput(filter.asOutput(scope));
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DPerChannel.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DPerChannel.java
index 12a8159a1c4..51799cff652 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DPerChannel.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DPerChannel.java
@@ -18,7 +18,6 @@
 package org.tensorflow.op.nn;
 
 import java.util.List;
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -73,7 +72,7 @@ private Options() {
    * @return a new instance of QuantizedConv2DPerChannel
    */
   @Endpoint(describeByClass = true)
-  public static <V extends TType, T extends TType, U extends TType> QuantizedConv2DPerChannel<V> create(Scope scope, Operand<T> input, Operand<U> filter, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, Operand<TFloat32> minFilter, Operand<TFloat32> maxFilter, DataType<V> outType, List<Long> strides, String padding, Options... options) {
+  public static <V extends TType, T extends TType, U extends TType> QuantizedConv2DPerChannel<V> create(Scope scope, Operand<T> input, Operand<U> filter, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, Operand<TFloat32> minFilter, Operand<TFloat32> maxFilter, Class<V> outType, List<Long> strides, String padding, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("QuantizedConv2DPerChannel", scope.makeOpName("QuantizedConv2DPerChannel"));
     opBuilder.addInput(input.asOutput(scope));
     opBuilder.addInput(filter.asOutput(scope));
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DWithBias.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DWithBias.java
index 0e0e229f266..561c1854190 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DWithBias.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DWithBias.java
@@ -18,7 +18,6 @@
 package org.tensorflow.op.nn;
 
 import java.util.List;
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -81,7 +80,7 @@ private Options() {
    * @return a new instance of QuantizedConv2DWithBias
    */
   @Endpoint(describeByClass = true)
-  public static <V extends TType, T extends TType, U extends TType> QuantizedConv2DWithBias<V> create(Scope scope, Operand<T> input, Operand<U> filter, Operand<TFloat32> bias, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, Operand<TFloat32> minFilter, Operand<TFloat32> maxFilter, DataType<V> outType, List<Long> strides, String padding, Options... options) {
+  public static <V extends TType, T extends TType, U extends TType> QuantizedConv2DWithBias<V> create(Scope scope, Operand<T> input, Operand<U> filter, Operand<TFloat32> bias, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, Operand<TFloat32> minFilter, Operand<TFloat32> maxFilter, Class<V> outType, List<Long> strides, String padding, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("QuantizedConv2DWithBias", scope.makeOpName("QuantizedConv2DWithBias"));
     opBuilder.addInput(input.asOutput(scope));
     opBuilder.addInput(filter.asOutput(scope));
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DWithBiasAndRelu.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DWithBiasAndRelu.java
index fe01df2a4d8..6f39f8d6291 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DWithBiasAndRelu.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DWithBiasAndRelu.java
@@ -18,7 +18,6 @@
 package org.tensorflow.op.nn;
 
 import java.util.List;
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -81,7 +80,7 @@ private Options() {
    * @return a new instance of QuantizedConv2DWithBiasAndRelu
    */
   @Endpoint(describeByClass = true)
-  public static <V extends TType, T extends TType, U extends TType> QuantizedConv2DWithBiasAndRelu<V> create(Scope scope, Operand<T> input, Operand<U> filter, Operand<TFloat32> bias, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, Operand<TFloat32> minFilter, Operand<TFloat32> maxFilter, DataType<V> outType, List<Long> strides, String padding, Options... options) {
+  public static <V extends TType, T extends TType, U extends TType> QuantizedConv2DWithBiasAndRelu<V> create(Scope scope, Operand<T> input, Operand<U> filter, Operand<TFloat32> bias, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, Operand<TFloat32> minFilter, Operand<TFloat32> maxFilter, Class<V> outType, List<Long> strides, String padding, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("QuantizedConv2DWithBiasAndRelu", scope.makeOpName("QuantizedConv2DWithBiasAndRelu"));
     opBuilder.addInput(input.asOutput(scope));
     opBuilder.addInput(filter.asOutput(scope));
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DWithBiasAndReluAndRequantize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DWithBiasAndReluAndRequantize.java
index 207ac36f953..3b95469edb7 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DWithBiasAndReluAndRequantize.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DWithBiasAndReluAndRequantize.java
@@ -18,7 +18,6 @@
 package org.tensorflow.op.nn;
 
 import java.util.List;
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -83,7 +82,7 @@ private Options() {
    * @return a new instance of QuantizedConv2DWithBiasAndReluAndRequantize
    */
   @Endpoint(describeByClass = true)
-  public static <W extends TType, T extends TType, U extends TType, V extends TType> QuantizedConv2DWithBiasAndReluAndRequantize<W> create(Scope scope, Operand<T> input, Operand<U> filter, Operand<V> bias, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, Operand<TFloat32> minFilter, Operand<TFloat32> maxFilter, Operand<TFloat32> minFreezedOutput, Operand<TFloat32> maxFreezedOutput, DataType<W> outType, List<Long> strides, String padding, Options... options) {
+  public static <W extends TType, T extends TType, U extends TType, V extends TType> QuantizedConv2DWithBiasAndReluAndRequantize<W> create(Scope scope, Operand<T> input, Operand<U> filter, Operand<V> bias, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, Operand<TFloat32> minFilter, Operand<TFloat32> maxFilter, Operand<TFloat32> minFreezedOutput, Operand<TFloat32> maxFreezedOutput, Class<W> outType, List<Long> strides, String padding, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("QuantizedConv2DWithBiasAndReluAndRequantize", scope.makeOpName("QuantizedConv2DWithBiasAndReluAndRequantize"));
     opBuilder.addInput(input.asOutput(scope));
     opBuilder.addInput(filter.asOutput(scope));
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DWithBiasAndRequantize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DWithBiasAndRequantize.java
index 5ff329393b7..ac5d066e2d7 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DWithBiasAndRequantize.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DWithBiasAndRequantize.java
@@ -18,7 +18,6 @@
 package org.tensorflow.op.nn;
 
 import java.util.List;
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -83,7 +82,7 @@ private Options() {
    * @return a new instance of QuantizedConv2DWithBiasAndRequantize
    */
   @Endpoint(describeByClass = true)
-  public static <W extends TType, T extends TType, U extends TType, V extends TType> QuantizedConv2DWithBiasAndRequantize<W> create(Scope scope, Operand<T> input, Operand<U> filter, Operand<V> bias, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, Operand<TFloat32> minFilter, Operand<TFloat32> maxFilter, Operand<TFloat32> minFreezedOutput, Operand<TFloat32> maxFreezedOutput, DataType<W> outType, List<Long> strides, String padding, Options... options) {
+  public static <W extends TType, T extends TType, U extends TType, V extends TType> QuantizedConv2DWithBiasAndRequantize<W> create(Scope scope, Operand<T> input, Operand<U> filter, Operand<V> bias, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, Operand<TFloat32> minFilter, Operand<TFloat32> maxFilter, Operand<TFloat32> minFreezedOutput, Operand<TFloat32> maxFreezedOutput, Class<W> outType, List<Long> strides, String padding, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("QuantizedConv2DWithBiasAndRequantize", scope.makeOpName("QuantizedConv2DWithBiasAndRequantize"));
     opBuilder.addInput(input.asOutput(scope));
     opBuilder.addInput(filter.asOutput(scope));
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DWithBiasSignedSumAndReluAndRequantize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DWithBiasSignedSumAndReluAndRequantize.java
index 13f86998835..42aef6a8258 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DWithBiasSignedSumAndReluAndRequantize.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DWithBiasSignedSumAndReluAndRequantize.java
@@ -18,7 +18,6 @@
 package org.tensorflow.op.nn;
 
 import java.util.List;
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -86,7 +85,7 @@ private Options() {
    * @return a new instance of QuantizedConv2DWithBiasSignedSumAndReluAndRequantize
    */
   @Endpoint(describeByClass = true)
-  public static <X extends TType, T extends TType, U extends TType, V extends TType, W extends TType> QuantizedConv2DWithBiasSignedSumAndReluAndRequantize<X> create(Scope scope, Operand<T> input, Operand<U> filter, Operand<V> bias, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, Operand<TFloat32> minFilter, Operand<TFloat32> maxFilter, Operand<TFloat32> minFreezedOutput, Operand<TFloat32> maxFreezedOutput, Operand<W> summand, Operand<TFloat32> minSummand, Operand<TFloat32> maxSummand, DataType<X> outType, List<Long> strides, String padding, Options... options) {
+  public static <X extends TType, T extends TType, U extends TType, V extends TType, W extends TType> QuantizedConv2DWithBiasSignedSumAndReluAndRequantize<X> create(Scope scope, Operand<T> input, Operand<U> filter, Operand<V> bias, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, Operand<TFloat32> minFilter, Operand<TFloat32> maxFilter, Operand<TFloat32> minFreezedOutput, Operand<TFloat32> maxFreezedOutput, Operand<W> summand, Operand<TFloat32> minSummand, Operand<TFloat32> maxSummand, Class<X> outType, List<Long> strides, String padding, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("QuantizedConv2DWithBiasSignedSumAndReluAndRequantize", scope.makeOpName("QuantizedConv2DWithBiasSignedSumAndReluAndRequantize"));
     opBuilder.addInput(input.asOutput(scope));
     opBuilder.addInput(filter.asOutput(scope));
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DWithBiasSumAndRelu.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DWithBiasSumAndRelu.java
index bed47e5ac80..d2e9508074c 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DWithBiasSumAndRelu.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DWithBiasSumAndRelu.java
@@ -18,7 +18,6 @@
 package org.tensorflow.op.nn;
 
 import java.util.List;
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -82,7 +81,7 @@ private Options() {
    * @return a new instance of QuantizedConv2DWithBiasSumAndRelu
    */
   @Endpoint(describeByClass = true)
-  public static <V extends TType, T extends TType, U extends TType> QuantizedConv2DWithBiasSumAndRelu<V> create(Scope scope, Operand<T> input, Operand<U> filter, Operand<TFloat32> bias, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, Operand<TFloat32> minFilter, Operand<TFloat32> maxFilter, Operand<TFloat32> summand, DataType<V> outType, List<Long> strides, String padding, Options... options) {
+  public static <V extends TType, T extends TType, U extends TType> QuantizedConv2DWithBiasSumAndRelu<V> create(Scope scope, Operand<T> input, Operand<U> filter, Operand<TFloat32> bias, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, Operand<TFloat32> minFilter, Operand<TFloat32> maxFilter, Operand<TFloat32> summand, Class<V> outType, List<Long> strides, String padding, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("QuantizedConv2DWithBiasSumAndRelu", scope.makeOpName("QuantizedConv2DWithBiasSumAndRelu"));
     opBuilder.addInput(input.asOutput(scope));
     opBuilder.addInput(filter.asOutput(scope));
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DWithBiasSumAndReluAndRequantize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DWithBiasSumAndReluAndRequantize.java
index 72279500957..cfd4aff55cd 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DWithBiasSumAndReluAndRequantize.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DWithBiasSumAndReluAndRequantize.java
@@ -18,7 +18,6 @@
 package org.tensorflow.op.nn;
 
 import java.util.List;
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -86,7 +85,7 @@ private Options() {
    * @return a new instance of QuantizedConv2DWithBiasSumAndReluAndRequantize
    */
   @Endpoint(describeByClass = true)
-  public static <X extends TType, T extends TType, U extends TType, V extends TType, W extends TType> QuantizedConv2DWithBiasSumAndReluAndRequantize<X> create(Scope scope, Operand<T> input, Operand<U> filter, Operand<V> bias, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, Operand<TFloat32> minFilter, Operand<TFloat32> maxFilter, Operand<TFloat32> minFreezedOutput, Operand<TFloat32> maxFreezedOutput, Operand<W> summand, Operand<TFloat32> minSummand, Operand<TFloat32> maxSummand, DataType<X> outType, List<Long> strides, String padding, Options... options) {
+  public static <X extends TType, T extends TType, U extends TType, V extends TType, W extends TType> QuantizedConv2DWithBiasSumAndReluAndRequantize<X> create(Scope scope, Operand<T> input, Operand<U> filter, Operand<V> bias, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, Operand<TFloat32> minFilter, Operand<TFloat32> maxFilter, Operand<TFloat32> minFreezedOutput, Operand<TFloat32> maxFreezedOutput, Operand<W> summand, Operand<TFloat32> minSummand, Operand<TFloat32> maxSummand, Class<X> outType, List<Long> strides, String padding, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("QuantizedConv2DWithBiasSumAndReluAndRequantize", scope.makeOpName("QuantizedConv2DWithBiasSumAndReluAndRequantize"));
     opBuilder.addInput(input.asOutput(scope));
     opBuilder.addInput(filter.asOutput(scope));
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2d.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2d.java
index b4ad6e2485d..7ed9b36502d 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2d.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2d.java
@@ -18,7 +18,6 @@
 package org.tensorflow.op.nn;
 
 import java.util.List;
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -84,7 +83,7 @@ private Options() {
    * @return a new instance of QuantizedConv2d
    */
   @Endpoint(describeByClass = true)
-  public static <V extends TType, T extends TType, U extends TType> QuantizedConv2d<V> create(Scope scope, Operand<T> input, Operand<U> filter, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, Operand<TFloat32> minFilter, Operand<TFloat32> maxFilter, DataType<V> outType, List<Long> strides, String padding, Options... options) {
+  public static <V extends TType, T extends TType, U extends TType> QuantizedConv2d<V> create(Scope scope, Operand<T> input, Operand<U> filter, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, Operand<TFloat32> minFilter, Operand<TFloat32> maxFilter, Class<V> outType, List<Long> strides, String padding, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("QuantizedConv2D", scope.makeOpName("QuantizedConv2d"));
     opBuilder.addInput(input.asOutput(scope));
     opBuilder.addInput(filter.asOutput(scope));
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedDepthwiseConv2D.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedDepthwiseConv2D.java
index 715fba4d7af..a480adc28f5 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedDepthwiseConv2D.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedDepthwiseConv2D.java
@@ -18,7 +18,6 @@
 package org.tensorflow.op.nn;
 
 import java.util.List;
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -73,7 +72,7 @@ private Options() {
    * @return a new instance of QuantizedDepthwiseConv2D
    */
   @Endpoint(describeByClass = true)
-  public static <V extends TType, T extends TType, U extends TType> QuantizedDepthwiseConv2D<V> create(Scope scope, Operand<T> input, Operand<U> filter, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, Operand<TFloat32> minFilter, Operand<TFloat32> maxFilter, DataType<V> outType, List<Long> strides, String padding, Options... options) {
+  public static <V extends TType, T extends TType, U extends TType> QuantizedDepthwiseConv2D<V> create(Scope scope, Operand<T> input, Operand<U> filter, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, Operand<TFloat32> minFilter, Operand<TFloat32> maxFilter, Class<V> outType, List<Long> strides, String padding, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("QuantizedDepthwiseConv2D", scope.makeOpName("QuantizedDepthwiseConv2D"));
     opBuilder.addInput(input.asOutput(scope));
     opBuilder.addInput(filter.asOutput(scope));
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedDepthwiseConv2DWithBias.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedDepthwiseConv2DWithBias.java
index 7e5e703b462..4b3326a0f18 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedDepthwiseConv2DWithBias.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedDepthwiseConv2DWithBias.java
@@ -18,7 +18,6 @@
 package org.tensorflow.op.nn;
 
 import java.util.List;
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -74,7 +73,7 @@ private Options() {
    * @return a new instance of QuantizedDepthwiseConv2DWithBias
    */
   @Endpoint(describeByClass = true)
-  public static <V extends TType, T extends TType, U extends TType> QuantizedDepthwiseConv2DWithBias<V> create(Scope scope, Operand<T> input, Operand<U> filter, Operand<TFloat32> bias, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, Operand<TFloat32> minFilter, Operand<TFloat32> maxFilter, DataType<V> outType, List<Long> strides, String padding, Options... options) {
+  public static <V extends TType, T extends TType, U extends TType> QuantizedDepthwiseConv2DWithBias<V> create(Scope scope, Operand<T> input, Operand<U> filter, Operand<TFloat32> bias, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, Operand<TFloat32> minFilter, Operand<TFloat32> maxFilter, Class<V> outType, List<Long> strides, String padding, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("QuantizedDepthwiseConv2DWithBias", scope.makeOpName("QuantizedDepthwiseConv2DWithBias"));
     opBuilder.addInput(input.asOutput(scope));
     opBuilder.addInput(filter.asOutput(scope));
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedDepthwiseConv2DWithBiasAndRelu.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedDepthwiseConv2DWithBiasAndRelu.java
index bdf1762476f..b7c82b9ca57 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedDepthwiseConv2DWithBiasAndRelu.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedDepthwiseConv2DWithBiasAndRelu.java
@@ -18,7 +18,6 @@
 package org.tensorflow.op.nn;
 
 import java.util.List;
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -83,7 +82,7 @@ private Options() {
    * @return a new instance of QuantizedDepthwiseConv2DWithBiasAndRelu
    */
   @Endpoint(describeByClass = true)
-  public static <V extends TType, T extends TType, U extends TType> QuantizedDepthwiseConv2DWithBiasAndRelu<V> create(Scope scope, Operand<T> input, Operand<U> filter, Operand<TFloat32> bias, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, Operand<TFloat32> minFilter, Operand<TFloat32> maxFilter, DataType<V> outType, List<Long> strides, String padding, Options... options) {
+  public static <V extends TType, T extends TType, U extends TType> QuantizedDepthwiseConv2DWithBiasAndRelu<V> create(Scope scope, Operand<T> input, Operand<U> filter, Operand<TFloat32> bias, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, Operand<TFloat32> minFilter, Operand<TFloat32> maxFilter, Class<V> outType, List<Long> strides, String padding, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("QuantizedDepthwiseConv2DWithBiasAndRelu", scope.makeOpName("QuantizedDepthwiseConv2DWithBiasAndRelu"));
     opBuilder.addInput(input.asOutput(scope));
     opBuilder.addInput(filter.asOutput(scope));
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedDepthwiseConv2DWithBiasAndReluAndRequantize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedDepthwiseConv2DWithBiasAndReluAndRequantize.java
index 11193af5bfe..0046acce74b 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedDepthwiseConv2DWithBiasAndReluAndRequantize.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedDepthwiseConv2DWithBiasAndReluAndRequantize.java
@@ -18,7 +18,6 @@
 package org.tensorflow.op.nn;
 
 import java.util.List;
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -85,7 +84,7 @@ private Options() {
    * @return a new instance of QuantizedDepthwiseConv2DWithBiasAndReluAndRequantize
    */
   @Endpoint(describeByClass = true)
-  public static <W extends TType, T extends TType, U extends TType, V extends TType> QuantizedDepthwiseConv2DWithBiasAndReluAndRequantize<W> create(Scope scope, Operand<T> input, Operand<U> filter, Operand<V> bias, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, Operand<TFloat32> minFilter, Operand<TFloat32> maxFilter, Operand<TFloat32> minFreezedOutput, Operand<TFloat32> maxFreezedOutput, DataType<W> outType, List<Long> strides, String padding, Options... options) {
+  public static <W extends TType, T extends TType, U extends TType, V extends TType> QuantizedDepthwiseConv2DWithBiasAndReluAndRequantize<W> create(Scope scope, Operand<T> input, Operand<U> filter, Operand<V> bias, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, Operand<TFloat32> minFilter, Operand<TFloat32> maxFilter, Operand<TFloat32> minFreezedOutput, Operand<TFloat32> maxFreezedOutput, Class<W> outType, List<Long> strides, String padding, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("QuantizedDepthwiseConv2DWithBiasAndReluAndRequantize", scope.makeOpName("QuantizedDepthwiseConv2DWithBiasAndReluAndRequantize"));
     opBuilder.addInput(input.asOutput(scope));
     opBuilder.addInput(filter.asOutput(scope));
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedRelu.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedRelu.java
index 69e5f6e5a46..60115bdf49e 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedRelu.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedRelu.java
@@ -17,7 +17,6 @@
 
 package org.tensorflow.op.nn;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -48,7 +47,7 @@ public final class QuantizedRelu<U extends TType> extends RawOp {
    * @return a new instance of QuantizedRelu
    */
   @Endpoint(describeByClass = true)
-  public static <U extends TType, T extends TType> QuantizedRelu<U> create(Scope scope, Operand<T> features, Operand<TFloat32> minFeatures, Operand<TFloat32> maxFeatures, DataType<U> outType) {
+  public static <U extends TType, T extends TType> QuantizedRelu<U> create(Scope scope, Operand<T> features, Operand<TFloat32> minFeatures, Operand<TFloat32> maxFeatures, Class<U> outType) {
     OperationBuilder opBuilder = scope.env().opBuilder("QuantizedRelu", scope.makeOpName("QuantizedRelu"));
     opBuilder.addInput(features.asOutput(scope));
     opBuilder.addInput(minFeatures.asOutput(scope));
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedRelu6.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedRelu6.java
index 365fbf97003..3cf1ee8f7c5 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedRelu6.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedRelu6.java
@@ -17,7 +17,6 @@
 
 package org.tensorflow.op.nn;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -48,7 +47,7 @@ public final class QuantizedRelu6<U extends TType> extends RawOp {
    * @return a new instance of QuantizedRelu6
    */
   @Endpoint(describeByClass = true)
-  public static <U extends TType, T extends TType> QuantizedRelu6<U> create(Scope scope, Operand<T> features, Operand<TFloat32> minFeatures, Operand<TFloat32> maxFeatures, DataType<U> outType) {
+  public static <U extends TType, T extends TType> QuantizedRelu6<U> create(Scope scope, Operand<T> features, Operand<TFloat32> minFeatures, Operand<TFloat32> maxFeatures, Class<U> outType) {
     OperationBuilder opBuilder = scope.env().opBuilder("QuantizedRelu6", scope.makeOpName("QuantizedRelu6"));
     opBuilder.addInput(features.asOutput(scope));
     opBuilder.addInput(minFeatures.asOutput(scope));
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedReluX.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedReluX.java
index 73826f85087..23e1092e20d 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedReluX.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedReluX.java
@@ -17,7 +17,6 @@
 
 package org.tensorflow.op.nn;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -49,7 +48,7 @@ public final class QuantizedReluX<U extends TType> extends RawOp {
    * @return a new instance of QuantizedReluX
    */
   @Endpoint(describeByClass = true)
-  public static <U extends TType, T extends TType> QuantizedReluX<U> create(Scope scope, Operand<T> features, Operand<TFloat32> maxValue, Operand<TFloat32> minFeatures, Operand<TFloat32> maxFeatures, DataType<U> outType) {
+  public static <U extends TType, T extends TType> QuantizedReluX<U> create(Scope scope, Operand<T> features, Operand<TFloat32> maxValue, Operand<TFloat32> minFeatures, Operand<TFloat32> maxFeatures, Class<U> outType) {
     OperationBuilder opBuilder = scope.env().opBuilder("QuantizedReluX", scope.makeOpName("QuantizedReluX"));
     opBuilder.addInput(features.asOutput(scope));
     opBuilder.addInput(maxValue.asOutput(scope));
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/raw/SoftmaxCrossEntropyWithLogits.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/raw/SoftmaxCrossEntropyWithLogits.java
index 32e55c4f4ac..7f400f6650c 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/raw/SoftmaxCrossEntropyWithLogits.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/raw/SoftmaxCrossEntropyWithLogits.java
@@ -15,7 +15,7 @@
 
 // This class has been generated, DO NOT EDIT!
 
-package org.tensorflow.op.nn;
+package org.tensorflow.op.nn.raw;
 
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
@@ -34,7 +34,7 @@
  * 
  * @param <T> data type for {@code loss()} output
  */
-@Operator(group = "nn")
+@Operator(group = "nn.raw")
 public final class SoftmaxCrossEntropyWithLogits<T extends TNumber> extends RawOp {
   
   /**
@@ -70,6 +70,9 @@ public Output<T> backprop() {
     return backprop;
   }
   
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "SoftmaxCrossEntropyWithLogits";
+  
   private Output<T> loss;
   private Output<T> backprop;
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/raw/SparseSoftmaxCrossEntropyWithLogits.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/raw/SparseSoftmaxCrossEntropyWithLogits.java
index 0c62793e97e..2a2c4de517e 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/raw/SparseSoftmaxCrossEntropyWithLogits.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/raw/SparseSoftmaxCrossEntropyWithLogits.java
@@ -15,7 +15,7 @@
 
 // This class has been generated, DO NOT EDIT!
 
-package org.tensorflow.op.nn;
+package org.tensorflow.op.nn.raw;
 
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
@@ -39,7 +39,7 @@
  * 
  * @param <T> data type for {@code loss()} output
  */
-@Operator(group = "nn")
+@Operator(group = "nn.raw")
 public final class SparseSoftmaxCrossEntropyWithLogits<T extends TNumber> extends RawOp {
   
   /**
@@ -74,6 +74,9 @@ public Output<T> backprop() {
     return backprop;
   }
   
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "SparseSoftmaxCrossEntropyWithLogits";
+  
   private Output<T> loss;
   private Output<T> backprop;
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/Dequantize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/Dequantize.java
index b0bb05757cb..ce4071e349c 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/Dequantize.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/Dequantize.java
@@ -17,7 +17,6 @@
 
 package org.tensorflow.op.quantization;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -137,7 +136,7 @@ private Options() {
    * @return a new instance of Dequantize
    */
   @Endpoint(describeByClass = true)
-  public static <U extends TNumber, T extends TType> Dequantize<U> create(Scope scope, Operand<T> input, Operand<TFloat32> minRange, Operand<TFloat32> maxRange, DataType<U> dtype, Options... options) {
+  public static <U extends TNumber, T extends TType> Dequantize<U> create(Scope scope, Operand<T> input, Operand<TFloat32> minRange, Operand<TFloat32> maxRange, Class<U> dtype, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("Dequantize", scope.makeOpName("Dequantize"));
     opBuilder.addInput(input.asOutput(scope));
     opBuilder.addInput(minRange.asOutput(scope));
@@ -172,7 +171,7 @@ public static <U extends TNumber, T extends TType> Dequantize<U> create(Scope sc
    */
   @Endpoint(describeByClass = true)
   public static <T extends TType> Dequantize<TFloat32> create(Scope scope, Operand<T> input, Operand<TFloat32> minRange, Operand<TFloat32> maxRange, Options... options) {
-    return create(scope, input, minRange, maxRange, TFloat32.DTYPE, options);
+    return create(scope, input, minRange, maxRange, TFloat32.class, options);
   }
   
   /**
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/Quantize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/Quantize.java
index 5e9e0e5ca83..30c609e5b6d 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/Quantize.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/Quantize.java
@@ -17,7 +17,6 @@
 
 package org.tensorflow.op.quantization;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -220,7 +219,7 @@ private Options() {
    * @return a new instance of Quantize
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> Quantize<T> create(Scope scope, Operand<TFloat32> input, Operand<TFloat32> minRange, Operand<TFloat32> maxRange, DataType<T> T, Options... options) {
+  public static <T extends TType> Quantize<T> create(Scope scope, Operand<TFloat32> input, Operand<TFloat32> minRange, Operand<TFloat32> maxRange, Class<T> T, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("QuantizeV2", scope.makeOpName("Quantize"));
     opBuilder.addInput(input.asOutput(scope));
     opBuilder.addInput(minRange.asOutput(scope));
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/QuantizeDownAndShrinkRange.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/QuantizeDownAndShrinkRange.java
index 849191286c3..15497432bed 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/QuantizeDownAndShrinkRange.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/QuantizeDownAndShrinkRange.java
@@ -17,7 +17,6 @@
 
 package org.tensorflow.op.quantization;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -71,7 +70,7 @@ public final class QuantizeDownAndShrinkRange<U extends TType> extends RawOp {
    * @return a new instance of QuantizeDownAndShrinkRange
    */
   @Endpoint(describeByClass = true)
-  public static <U extends TType, T extends TType> QuantizeDownAndShrinkRange<U> create(Scope scope, Operand<T> input, Operand<TFloat32> inputMin, Operand<TFloat32> inputMax, DataType<U> outType) {
+  public static <U extends TType, T extends TType> QuantizeDownAndShrinkRange<U> create(Scope scope, Operand<T> input, Operand<TFloat32> inputMin, Operand<TFloat32> inputMax, Class<U> outType) {
     OperationBuilder opBuilder = scope.env().opBuilder("QuantizeDownAndShrinkRange", scope.makeOpName("QuantizeDownAndShrinkRange"));
     opBuilder.addInput(input.asOutput(scope));
     opBuilder.addInput(inputMin.asOutput(scope));
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/QuantizedMatMulWithBiasAndDequantize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/QuantizedMatMulWithBiasAndDequantize.java
index ef25c6d0b00..0ab6bd11b79 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/QuantizedMatMulWithBiasAndDequantize.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/QuantizedMatMulWithBiasAndDequantize.java
@@ -17,7 +17,6 @@
 
 package org.tensorflow.op.quantization;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -90,7 +89,7 @@ private Options() {
    * @return a new instance of QuantizedMatMulWithBiasAndDequantize
    */
   @Endpoint(describeByClass = true)
-  public static <W extends TNumber, T extends TType, U extends TType, V extends TType> QuantizedMatMulWithBiasAndDequantize<W> create(Scope scope, Operand<T> a, Operand<U> b, Operand<V> bias, Operand<TFloat32> minA, Operand<TFloat32> maxA, Operand<TFloat32> minB, Operand<TFloat32> maxB, Operand<TFloat32> minFreezedOutput, Operand<TFloat32> maxFreezedOutput, DataType<W> Toutput, Options... options) {
+  public static <W extends TNumber, T extends TType, U extends TType, V extends TType> QuantizedMatMulWithBiasAndDequantize<W> create(Scope scope, Operand<T> a, Operand<U> b, Operand<V> bias, Operand<TFloat32> minA, Operand<TFloat32> maxA, Operand<TFloat32> minB, Operand<TFloat32> maxB, Operand<TFloat32> minFreezedOutput, Operand<TFloat32> maxFreezedOutput, Class<W> Toutput, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("QuantizedMatMulWithBiasAndDequantize", scope.makeOpName("QuantizedMatMulWithBiasAndDequantize"));
     opBuilder.addInput(a.asOutput(scope));
     opBuilder.addInput(b.asOutput(scope));
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/QuantizedMatMulWithBiasAndRequantize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/QuantizedMatMulWithBiasAndRequantize.java
index 54ef67fd237..51341159d76 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/QuantizedMatMulWithBiasAndRequantize.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/QuantizedMatMulWithBiasAndRequantize.java
@@ -17,7 +17,6 @@
 
 package org.tensorflow.op.quantization;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -89,7 +88,7 @@ private Options() {
    * @return a new instance of QuantizedMatMulWithBiasAndRequantize
    */
   @Endpoint(describeByClass = true)
-  public static <W extends TType, T extends TType, U extends TType, V extends TType> QuantizedMatMulWithBiasAndRequantize<W> create(Scope scope, Operand<T> a, Operand<U> b, Operand<V> bias, Operand<TFloat32> minA, Operand<TFloat32> maxA, Operand<TFloat32> minB, Operand<TFloat32> maxB, Operand<TFloat32> minFreezedOutput, Operand<TFloat32> maxFreezedOutput, DataType<W> Toutput, Options... options) {
+  public static <W extends TType, T extends TType, U extends TType, V extends TType> QuantizedMatMulWithBiasAndRequantize<W> create(Scope scope, Operand<T> a, Operand<U> b, Operand<V> bias, Operand<TFloat32> minA, Operand<TFloat32> maxA, Operand<TFloat32> minB, Operand<TFloat32> maxB, Operand<TFloat32> minFreezedOutput, Operand<TFloat32> maxFreezedOutput, Class<W> Toutput, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("QuantizedMatMulWithBiasAndRequantize", scope.makeOpName("QuantizedMatMulWithBiasAndRequantize"));
     opBuilder.addInput(a.asOutput(scope));
     opBuilder.addInput(b.asOutput(scope));
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/Requantize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/Requantize.java
index f0a8541a854..1de8c064658 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/Requantize.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/Requantize.java
@@ -17,7 +17,6 @@
 
 package org.tensorflow.op.quantization;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -58,7 +57,7 @@ public final class Requantize<U extends TType> extends RawOp {
    * @return a new instance of Requantize
    */
   @Endpoint(describeByClass = true)
-  public static <U extends TType, T extends TType> Requantize<U> create(Scope scope, Operand<T> input, Operand<TFloat32> inputMin, Operand<TFloat32> inputMax, Operand<TFloat32> requestedOutputMin, Operand<TFloat32> requestedOutputMax, DataType<U> outType) {
+  public static <U extends TType, T extends TType> Requantize<U> create(Scope scope, Operand<T> input, Operand<TFloat32> inputMin, Operand<TFloat32> inputMax, Operand<TFloat32> requestedOutputMin, Operand<TFloat32> requestedOutputMax, Class<U> outType) {
     OperationBuilder opBuilder = scope.env().opBuilder("Requantize", scope.makeOpName("Requantize"));
     opBuilder.addInput(input.asOutput(scope));
     opBuilder.addInput(inputMin.asOutput(scope));
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/ragged/RaggedBincount.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/ragged/RaggedBincount.java
index 1e0224aa9ef..40aceaed5c8 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/ragged/RaggedBincount.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/ragged/RaggedBincount.java
@@ -27,7 +27,6 @@
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Counts the number of occurrences of each value in an integer array.
@@ -80,10 +79,10 @@ private Options() {
   @Endpoint(describeByClass = true)
   public static <U extends TNumber, T extends TNumber> RaggedBincount<U> create(Scope scope, Operand<TInt64> splits, Operand<T> values, Operand<T> size, Operand<U> weights, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("RaggedBincount", scope.makeOpName("RaggedBincount"));
-    opBuilder.addInput(splits.asOutput());
-    opBuilder.addInput(values.asOutput());
-    opBuilder.addInput(size.asOutput());
-    opBuilder.addInput(weights.asOutput());
+    opBuilder.addInput(splits.asOutput(scope));
+    opBuilder.addInput(values.asOutput(scope));
+    opBuilder.addInput(size.asOutput(scope));
+    opBuilder.addInput(weights.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -111,7 +110,7 @@ public Output<U> output() {
   }
   
   @Override
-  public Output<U> asOutput() {
+  public Output<U> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/ragged/RaggedCountSparseOutput.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/ragged/RaggedCountSparseOutput.java
index 4829e49488b..b747211fa5b 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/ragged/RaggedCountSparseOutput.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/ragged/RaggedCountSparseOutput.java
@@ -27,7 +27,6 @@
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Performs sparse-output bin counting for a ragged tensor input.
@@ -81,9 +80,9 @@ private Options() {
   @Endpoint(describeByClass = true)
   public static <U extends TNumber, T extends TNumber> RaggedCountSparseOutput<U> create(Scope scope, Operand<TInt64> splits, Operand<T> values, Operand<U> weights, Boolean binaryOutput, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("RaggedCountSparseOutput", scope.makeOpName("RaggedCountSparseOutput"));
-    opBuilder.addInput(splits.asOutput());
-    opBuilder.addInput(values.asOutput());
-    opBuilder.addInput(weights.asOutput());
+    opBuilder.addInput(splits.asOutput(scope));
+    opBuilder.addInput(values.asOutput(scope));
+    opBuilder.addInput(weights.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("binary_output", binaryOutput);
     if (options != null) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/ragged/RaggedCross.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/ragged/RaggedCross.java
index 9ea32878257..2d13027a88a 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/ragged/RaggedCross.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/ragged/RaggedCross.java
@@ -17,7 +17,6 @@
 
 package org.tensorflow.op.ragged;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -62,14 +61,14 @@ public final class RaggedCross<T extends TType, U extends TNumber> extends RawOp
    * @return a new instance of RaggedCross
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> RaggedCross<T, U> create(Scope scope, Iterable<Operand<?>> raggedValues, Iterable<Operand<?>> raggedRowSplits, Iterable<Operand<TInt64>> sparseIndices, Iterable<Operand<?>> sparseValues, Iterable<Operand<TInt64>> sparseShape, Iterable<Operand<?>> denseInputs, String inputOrder, Boolean hashedOutput, Long numBuckets, Long hashKey, DataType<T> outValuesType, DataType<U> outRowSplitsType) {
+  public static <T extends TType, U extends TNumber> RaggedCross<T, U> create(Scope scope, Iterable<Operand<?>> raggedValues, Iterable<Operand<?>> raggedRowSplits, Iterable<Operand<TInt64>> sparseIndices, Iterable<Operand<?>> sparseValues, Iterable<Operand<TInt64>> sparseShape, Iterable<Operand<?>> denseInputs, String inputOrder, Boolean hashedOutput, Long numBuckets, Long hashKey, Class<T> outValuesType, Class<U> outRowSplitsType) {
     OperationBuilder opBuilder = scope.env().opBuilder("RaggedCross", scope.makeOpName("RaggedCross"));
-    opBuilder.addInputList(Operands.asOutputs(raggedValues));
-    opBuilder.addInputList(Operands.asOutputs(raggedRowSplits));
-    opBuilder.addInputList(Operands.asOutputs(sparseIndices));
-    opBuilder.addInputList(Operands.asOutputs(sparseValues));
-    opBuilder.addInputList(Operands.asOutputs(sparseShape));
-    opBuilder.addInputList(Operands.asOutputs(denseInputs));
+    opBuilder.addInputList(Operands.asOutputs(scope, raggedValues));
+    opBuilder.addInputList(Operands.asOutputs(scope, raggedRowSplits));
+    opBuilder.addInputList(Operands.asOutputs(scope, sparseIndices));
+    opBuilder.addInputList(Operands.asOutputs(scope, sparseValues));
+    opBuilder.addInputList(Operands.asOutputs(scope, sparseShape));
+    opBuilder.addInputList(Operands.asOutputs(scope, denseInputs));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("input_order", inputOrder);
     opBuilder.setAttr("hashed_output", hashedOutput);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/ragged/RaggedRange.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/ragged/RaggedRange.java
index 69e86f2426f..327b8a3c1d8 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/ragged/RaggedRange.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/ragged/RaggedRange.java
@@ -17,7 +17,6 @@
 
 package org.tensorflow.op.ragged;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -63,7 +62,7 @@ public final class RaggedRange<U extends TNumber, T extends TNumber> extends Raw
    * @return a new instance of RaggedRange
    */
   @Endpoint(describeByClass = true)
-  public static <U extends TNumber, T extends TNumber> RaggedRange<U, T> create(Scope scope, Operand<T> starts, Operand<T> limits, Operand<T> deltas, DataType<U> Tsplits) {
+  public static <U extends TNumber, T extends TNumber> RaggedRange<U, T> create(Scope scope, Operand<T> starts, Operand<T> limits, Operand<T> deltas, Class<U> Tsplits) {
     OperationBuilder opBuilder = scope.env().opBuilder("RaggedRange", scope.makeOpName("RaggedRange"));
     opBuilder.addInput(starts.asOutput(scope));
     opBuilder.addInput(limits.asOutput(scope));
@@ -84,7 +83,7 @@ public static <U extends TNumber, T extends TNumber> RaggedRange<U, T> create(Sc
    */
   @Endpoint(describeByClass = true)
   public static <T extends TNumber> RaggedRange<TInt64, T> create(Scope scope, Operand<T> starts, Operand<T> limits, Operand<T> deltas) {
-    return create(scope, starts, limits, deltas, TInt64.DTYPE);
+    return create(scope, starts, limits, deltas, TInt64.class);
   }
   
   /**
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/ragged/RaggedTensorFromVariant.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/ragged/RaggedTensorFromVariant.java
index e2170e1e751..736b6c47b21 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/ragged/RaggedTensorFromVariant.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/ragged/RaggedTensorFromVariant.java
@@ -19,7 +19,6 @@
 
 import java.util.Arrays;
 import java.util.List;
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -67,7 +66,7 @@ public final class RaggedTensorFromVariant<U extends TNumber, T extends TType> e
    * @return a new instance of RaggedTensorFromVariant
    */
   @Endpoint(describeByClass = true)
-  public static <U extends TNumber, T extends TType> RaggedTensorFromVariant<U, T> create(Scope scope, Operand<?> encodedRagged, Long inputRaggedRank, Long outputRaggedRank, DataType<T> Tvalues, DataType<U> Tsplits) {
+  public static <U extends TNumber, T extends TType> RaggedTensorFromVariant<U, T> create(Scope scope, Operand<?> encodedRagged, Long inputRaggedRank, Long outputRaggedRank, Class<T> Tvalues, Class<U> Tsplits) {
     OperationBuilder opBuilder = scope.env().opBuilder("RaggedTensorFromVariant", scope.makeOpName("RaggedTensorFromVariant"));
     opBuilder.addInput(encodedRagged.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
@@ -91,8 +90,8 @@ public static <U extends TNumber, T extends TType> RaggedTensorFromVariant<U, T>
    * @return a new instance of RaggedTensorFromVariant
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> RaggedTensorFromVariant<TInt64, T> create(Scope scope, Operand<?> encodedRagged, Long inputRaggedRank, Long outputRaggedRank, DataType<T> Tvalues) {
-    return create(scope, encodedRagged, inputRaggedRank, outputRaggedRank, Tvalues, TInt64.DTYPE);
+  public static <T extends TType> RaggedTensorFromVariant<TInt64, T> create(Scope scope, Operand<?> encodedRagged, Long inputRaggedRank, Long outputRaggedRank, Class<T> Tvalues) {
+    return create(scope, encodedRagged, inputRaggedRank, outputRaggedRank, Tvalues, TInt64.class);
   }
   
   /**
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/AnonymousSeedGenerator.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/AnonymousSeedGenerator.java
index c724bb6d110..2929a630525 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/AnonymousSeedGenerator.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/AnonymousSeedGenerator.java
@@ -44,9 +44,9 @@ public final class AnonymousSeedGenerator extends RawOp {
   @Endpoint(describeByClass = true)
   public static AnonymousSeedGenerator create(Scope scope, Operand<TInt64> seed, Operand<TInt64> seed2, Operand<TBool> reshuffle) {
     OperationBuilder opBuilder = scope.env().opBuilder("AnonymousSeedGenerator", scope.makeOpName("AnonymousSeedGenerator"));
-    opBuilder.addInput(seed.asOutput());
-    opBuilder.addInput(seed2.asOutput());
-    opBuilder.addInput(reshuffle.asOutput());
+    opBuilder.addInput(seed.asOutput(scope));
+    opBuilder.addInput(seed2.asOutput(scope));
+    opBuilder.addInput(reshuffle.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new AnonymousSeedGenerator(opBuilder.build());
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/DeleteSeedGenerator.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/DeleteSeedGenerator.java
index 16982946d1f..5bc2c0dd1b7 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/DeleteSeedGenerator.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/DeleteSeedGenerator.java
@@ -40,8 +40,8 @@ public final class DeleteSeedGenerator extends RawOp {
   @Endpoint(describeByClass = true)
   public static DeleteSeedGenerator create(Scope scope, Operand<?> handle, Operand<?> deleter) {
     OperationBuilder opBuilder = scope.env().opBuilder("DeleteSeedGenerator", scope.makeOpName("DeleteSeedGenerator"));
-    opBuilder.addInput(handle.asOutput());
-    opBuilder.addInput(deleter.asOutput());
+    opBuilder.addInput(handle.asOutput(scope));
+    opBuilder.addInput(deleter.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new DeleteSeedGenerator(opBuilder.build());
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/Multinomial.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/Multinomial.java
index 3236ccf8bd7..bdff3925130 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/Multinomial.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/Multinomial.java
@@ -17,7 +17,6 @@
 
 package org.tensorflow.op.random;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -79,7 +78,7 @@ private Options() {
    * @return a new instance of Multinomial
    */
   @Endpoint(describeByClass = true)
-  public static <U extends TNumber, T extends TNumber> Multinomial<U> create(Scope scope, Operand<T> logits, Operand<TInt32> numSamples, DataType<U> outputDtype, Options... options) {
+  public static <U extends TNumber, T extends TNumber> Multinomial<U> create(Scope scope, Operand<T> logits, Operand<TInt32> numSamples, Class<U> outputDtype, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("Multinomial", scope.makeOpName("Multinomial"));
     opBuilder.addInput(logits.asOutput(scope));
     opBuilder.addInput(numSamples.asOutput(scope));
@@ -110,7 +109,7 @@ public static <U extends TNumber, T extends TNumber> Multinomial<U> create(Scope
    */
   @Endpoint(describeByClass = true)
   public static <T extends TNumber> Multinomial<TInt64> create(Scope scope, Operand<T> logits, Operand<TInt32> numSamples, Options... options) {
-    return create(scope, logits, numSamples, TInt64.DTYPE, options);
+    return create(scope, logits, numSamples, TInt64.class, options);
   }
   
   /**
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/NonDeterministicInts.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/NonDeterministicInts.java
index ea618bf846a..21b3ab97a77 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/NonDeterministicInts.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/NonDeterministicInts.java
@@ -17,7 +17,6 @@
 
 package org.tensorflow.op.random;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -47,7 +46,7 @@ public final class NonDeterministicInts<U extends TType> extends RawOp implement
    * @return a new instance of NonDeterministicInts
    */
   @Endpoint(describeByClass = true)
-  public static <U extends TType, T extends TType> NonDeterministicInts<U> create(Scope scope, Operand<T> shape, DataType<U> dtype) {
+  public static <U extends TType, T extends TType> NonDeterministicInts<U> create(Scope scope, Operand<T> shape, Class<U> dtype) {
     OperationBuilder opBuilder = scope.env().opBuilder("NonDeterministicInts", scope.makeOpName("NonDeterministicInts"));
     opBuilder.addInput(shape.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
@@ -64,7 +63,7 @@ public static <U extends TType, T extends TType> NonDeterministicInts<U> create(
    */
   @Endpoint(describeByClass = true)
   public static <T extends TType> NonDeterministicInts<TInt64> create(Scope scope, Operand<T> shape) {
-    return create(scope, shape, TInt64.DTYPE);
+    return create(scope, shape, TInt64.class);
   }
   
   /**
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RandomPoisson.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RandomPoisson.java
index f3934e43c6b..d0b62eae694 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RandomPoisson.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RandomPoisson.java
@@ -17,7 +17,6 @@
 
 package org.tensorflow.op.random;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -90,7 +89,7 @@ private Options() {
    * @return a new instance of RandomPoisson
    */
   @Endpoint(describeByClass = true)
-  public static <V extends TNumber, T extends TNumber, U extends TNumber> RandomPoisson<V> create(Scope scope, Operand<T> shape, Operand<U> rate, DataType<V> dtype, Options... options) {
+  public static <V extends TNumber, T extends TNumber, U extends TNumber> RandomPoisson<V> create(Scope scope, Operand<T> shape, Operand<U> rate, Class<V> dtype, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("RandomPoissonV2", scope.makeOpName("RandomPoisson"));
     opBuilder.addInput(shape.asOutput(scope));
     opBuilder.addInput(rate.asOutput(scope));
@@ -122,7 +121,7 @@ public static <V extends TNumber, T extends TNumber, U extends TNumber> RandomPo
    */
   @Endpoint(describeByClass = true)
   public static <T extends TNumber, U extends TNumber> RandomPoisson<TInt64> create(Scope scope, Operand<T> shape, Operand<U> rate, Options... options) {
-    return create(scope, shape, rate, TInt64.DTYPE, options);
+    return create(scope, shape, rate, TInt64.class, options);
   }
   
   /**
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RandomStandardNormal.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RandomStandardNormal.java
index a6f61546deb..0743f6860df 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RandomStandardNormal.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RandomStandardNormal.java
@@ -17,7 +17,6 @@
 
 package org.tensorflow.op.random;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -78,7 +77,7 @@ private Options() {
    * @return a new instance of RandomStandardNormal
    */
   @Endpoint(describeByClass = true)
-  public static <U extends TNumber, T extends TNumber> RandomStandardNormal<U> create(Scope scope, Operand<T> shape, DataType<U> dtype, Options... options) {
+  public static <U extends TNumber, T extends TNumber> RandomStandardNormal<U> create(Scope scope, Operand<T> shape, Class<U> dtype, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("RandomStandardNormal", scope.makeOpName("RandomStandardNormal"));
     opBuilder.addInput(shape.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RandomUniform.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RandomUniform.java
index 81ceb5de601..fcd57fc4158 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RandomUniform.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RandomUniform.java
@@ -17,7 +17,6 @@
 
 package org.tensorflow.op.random;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -79,7 +78,7 @@ private Options() {
    * @return a new instance of RandomUniform
    */
   @Endpoint(describeByClass = true)
-  public static <U extends TNumber, T extends TNumber> RandomUniform<U> create(Scope scope, Operand<T> shape, DataType<U> dtype, Options... options) {
+  public static <U extends TNumber, T extends TNumber> RandomUniform<U> create(Scope scope, Operand<T> shape, Class<U> dtype, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("RandomUniform", scope.makeOpName("RandomUniform"));
     opBuilder.addInput(shape.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatefulRandomBinomial.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatefulRandomBinomial.java
index 4c82bce1aac..cb2f012ca95 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatefulRandomBinomial.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatefulRandomBinomial.java
@@ -17,7 +17,6 @@
 
 package org.tensorflow.op.random;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -48,7 +47,7 @@ public final class StatefulRandomBinomial<V extends TNumber> extends RawOp imple
    * @return a new instance of StatefulRandomBinomial
    */
   @Endpoint(describeByClass = true)
-  public static <V extends TNumber, T extends TNumber, U extends TNumber> StatefulRandomBinomial<V> create(Scope scope, Operand<?> resource, Operand<TInt64> algorithm, Operand<T> shape, Operand<U> counts, Operand<U> probs, DataType<V> dtype) {
+  public static <V extends TNumber, T extends TNumber, U extends TNumber> StatefulRandomBinomial<V> create(Scope scope, Operand<?> resource, Operand<TInt64> algorithm, Operand<T> shape, Operand<U> counts, Operand<U> probs, Class<V> dtype) {
     OperationBuilder opBuilder = scope.env().opBuilder("StatefulRandomBinomial", scope.makeOpName("StatefulRandomBinomial"));
     opBuilder.addInput(resource.asOutput(scope));
     opBuilder.addInput(algorithm.asOutput(scope));
@@ -73,7 +72,7 @@ public static <V extends TNumber, T extends TNumber, U extends TNumber> Stateful
    */
   @Endpoint(describeByClass = true)
   public static <T extends TNumber, U extends TNumber> StatefulRandomBinomial<TInt64> create(Scope scope, Operand<?> resource, Operand<TInt64> algorithm, Operand<T> shape, Operand<U> counts, Operand<U> probs) {
-    return create(scope, resource, algorithm, shape, counts, probs, TInt64.DTYPE);
+    return create(scope, resource, algorithm, shape, counts, probs, TInt64.class);
   }
   
   /**
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatefulStandardNormal.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatefulStandardNormal.java
index 3de3c5211e3..779bd1b55b8 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatefulStandardNormal.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatefulStandardNormal.java
@@ -17,7 +17,6 @@
 
 package org.tensorflow.op.random;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -51,7 +50,7 @@ public final class StatefulStandardNormal<U extends TType> extends RawOp impleme
    * @return a new instance of StatefulStandardNormal
    */
   @Endpoint(describeByClass = true)
-  public static <U extends TType, T extends TType> StatefulStandardNormal<U> create(Scope scope, Operand<?> resource, Operand<TInt64> algorithm, Operand<T> shape, DataType<U> dtype) {
+  public static <U extends TType, T extends TType> StatefulStandardNormal<U> create(Scope scope, Operand<?> resource, Operand<TInt64> algorithm, Operand<T> shape, Class<U> dtype) {
     OperationBuilder opBuilder = scope.env().opBuilder("StatefulStandardNormalV2", scope.makeOpName("StatefulStandardNormal"));
     opBuilder.addInput(resource.asOutput(scope));
     opBuilder.addInput(algorithm.asOutput(scope));
@@ -72,7 +71,7 @@ public static <U extends TType, T extends TType> StatefulStandardNormal<U> creat
    */
   @Endpoint(describeByClass = true)
   public static <T extends TType> StatefulStandardNormal<TFloat32> create(Scope scope, Operand<?> resource, Operand<TInt64> algorithm, Operand<T> shape) {
-    return create(scope, resource, algorithm, shape, TFloat32.DTYPE);
+    return create(scope, resource, algorithm, shape, TFloat32.class);
   }
   
   /**
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatefulTruncatedNormal.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatefulTruncatedNormal.java
index 391462d9bbf..807c6ed335c 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatefulTruncatedNormal.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatefulTruncatedNormal.java
@@ -17,7 +17,6 @@
 
 package org.tensorflow.op.random;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -52,7 +51,7 @@ public final class StatefulTruncatedNormal<U extends TType> extends RawOp implem
    * @return a new instance of StatefulTruncatedNormal
    */
   @Endpoint(describeByClass = true)
-  public static <U extends TType, T extends TType> StatefulTruncatedNormal<U> create(Scope scope, Operand<?> resource, Operand<TInt64> algorithm, Operand<T> shape, DataType<U> dtype) {
+  public static <U extends TType, T extends TType> StatefulTruncatedNormal<U> create(Scope scope, Operand<?> resource, Operand<TInt64> algorithm, Operand<T> shape, Class<U> dtype) {
     OperationBuilder opBuilder = scope.env().opBuilder("StatefulTruncatedNormal", scope.makeOpName("StatefulTruncatedNormal"));
     opBuilder.addInput(resource.asOutput(scope));
     opBuilder.addInput(algorithm.asOutput(scope));
@@ -73,7 +72,7 @@ public static <U extends TType, T extends TType> StatefulTruncatedNormal<U> crea
    */
   @Endpoint(describeByClass = true)
   public static <T extends TType> StatefulTruncatedNormal<TFloat32> create(Scope scope, Operand<?> resource, Operand<TInt64> algorithm, Operand<T> shape) {
-    return create(scope, resource, algorithm, shape, TFloat32.DTYPE);
+    return create(scope, resource, algorithm, shape, TFloat32.class);
   }
   
   /**
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatefulUniform.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatefulUniform.java
index f82afca0163..bd8d035be1a 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatefulUniform.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatefulUniform.java
@@ -17,7 +17,6 @@
 
 package org.tensorflow.op.random;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -51,7 +50,7 @@ public final class StatefulUniform<U extends TType> extends RawOp implements Ope
    * @return a new instance of StatefulUniform
    */
   @Endpoint(describeByClass = true)
-  public static <U extends TType, T extends TType> StatefulUniform<U> create(Scope scope, Operand<?> resource, Operand<TInt64> algorithm, Operand<T> shape, DataType<U> dtype) {
+  public static <U extends TType, T extends TType> StatefulUniform<U> create(Scope scope, Operand<?> resource, Operand<TInt64> algorithm, Operand<T> shape, Class<U> dtype) {
     OperationBuilder opBuilder = scope.env().opBuilder("StatefulUniform", scope.makeOpName("StatefulUniform"));
     opBuilder.addInput(resource.asOutput(scope));
     opBuilder.addInput(algorithm.asOutput(scope));
@@ -72,7 +71,7 @@ public static <U extends TType, T extends TType> StatefulUniform<U> create(Scope
    */
   @Endpoint(describeByClass = true)
   public static <T extends TType> StatefulUniform<TFloat32> create(Scope scope, Operand<?> resource, Operand<TInt64> algorithm, Operand<T> shape) {
-    return create(scope, resource, algorithm, shape, TFloat32.DTYPE);
+    return create(scope, resource, algorithm, shape, TFloat32.class);
   }
   
   /**
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatefulUniformFullInt.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatefulUniformFullInt.java
index 9e59c5f6e15..722764bfb7c 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatefulUniformFullInt.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatefulUniformFullInt.java
@@ -17,7 +17,6 @@
 
 package org.tensorflow.op.random;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -49,7 +48,7 @@ public final class StatefulUniformFullInt<U extends TType> extends RawOp impleme
    * @return a new instance of StatefulUniformFullInt
    */
   @Endpoint(describeByClass = true)
-  public static <U extends TType, T extends TType> StatefulUniformFullInt<U> create(Scope scope, Operand<?> resource, Operand<TInt64> algorithm, Operand<T> shape, DataType<U> dtype) {
+  public static <U extends TType, T extends TType> StatefulUniformFullInt<U> create(Scope scope, Operand<?> resource, Operand<TInt64> algorithm, Operand<T> shape, Class<U> dtype) {
     OperationBuilder opBuilder = scope.env().opBuilder("StatefulUniformFullInt", scope.makeOpName("StatefulUniformFullInt"));
     opBuilder.addInput(resource.asOutput(scope));
     opBuilder.addInput(algorithm.asOutput(scope));
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessMultinomial.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessMultinomial.java
index c9ee3523d5a..442d7b9db0e 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessMultinomial.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessMultinomial.java
@@ -17,7 +17,6 @@
 
 package org.tensorflow.op.random;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -50,7 +49,7 @@ public final class StatelessMultinomial<V extends TNumber> extends RawOp impleme
    * @return a new instance of StatelessMultinomial
    */
   @Endpoint(describeByClass = true)
-  public static <V extends TNumber, T extends TNumber, U extends TNumber> StatelessMultinomial<V> create(Scope scope, Operand<T> logits, Operand<TInt32> numSamples, Operand<U> seed, DataType<V> outputDtype) {
+  public static <V extends TNumber, T extends TNumber, U extends TNumber> StatelessMultinomial<V> create(Scope scope, Operand<T> logits, Operand<TInt32> numSamples, Operand<U> seed, Class<V> outputDtype) {
     OperationBuilder opBuilder = scope.env().opBuilder("StatelessMultinomial", scope.makeOpName("StatelessMultinomial"));
     opBuilder.addInput(logits.asOutput(scope));
     opBuilder.addInput(numSamples.asOutput(scope));
@@ -72,7 +71,7 @@ public static <V extends TNumber, T extends TNumber, U extends TNumber> Stateles
    */
   @Endpoint(describeByClass = true)
   public static <T extends TNumber, U extends TNumber> StatelessMultinomial<TInt64> create(Scope scope, Operand<T> logits, Operand<TInt32> numSamples, Operand<U> seed) {
-    return create(scope, logits, numSamples, seed, TInt64.DTYPE);
+    return create(scope, logits, numSamples, seed, TInt64.class);
   }
   
   /**
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessParameterizedTruncatedNormal.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessParameterizedTruncatedNormal.java
index 179160463c7..6ad43c17fe2 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessParameterizedTruncatedNormal.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessParameterizedTruncatedNormal.java
@@ -26,7 +26,6 @@
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * @param <V> data type for {@code output()} output
@@ -49,12 +48,12 @@ public final class StatelessParameterizedTruncatedNormal<V extends TNumber> exte
   @Endpoint(describeByClass = true)
   public static <V extends TNumber, T extends TNumber, U extends TNumber> StatelessParameterizedTruncatedNormal<V> create(Scope scope, Operand<T> shape, Operand<U> seed, Operand<V> means, Operand<V> stddevs, Operand<V> minvals, Operand<V> maxvals) {
     OperationBuilder opBuilder = scope.env().opBuilder("StatelessParameterizedTruncatedNormal", scope.makeOpName("StatelessParameterizedTruncatedNormal"));
-    opBuilder.addInput(shape.asOutput());
-    opBuilder.addInput(seed.asOutput());
-    opBuilder.addInput(means.asOutput());
-    opBuilder.addInput(stddevs.asOutput());
-    opBuilder.addInput(minvals.asOutput());
-    opBuilder.addInput(maxvals.asOutput());
+    opBuilder.addInput(shape.asOutput(scope));
+    opBuilder.addInput(seed.asOutput(scope));
+    opBuilder.addInput(means.asOutput(scope));
+    opBuilder.addInput(stddevs.asOutput(scope));
+    opBuilder.addInput(minvals.asOutput(scope));
+    opBuilder.addInput(maxvals.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new StatelessParameterizedTruncatedNormal<V>(opBuilder.build());
   }
@@ -68,7 +67,7 @@ public Output<V> output() {
   }
   
   @Override
-  public Output<V> asOutput() {
+  public Output<V> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessRandomBinomial.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessRandomBinomial.java
index d96613dee92..870f995c6f0 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessRandomBinomial.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessRandomBinomial.java
@@ -17,7 +17,6 @@
 
 package org.tensorflow.op.random;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -54,7 +53,7 @@ public final class StatelessRandomBinomial<W extends TNumber> extends RawOp impl
    * @return a new instance of StatelessRandomBinomial
    */
   @Endpoint(describeByClass = true)
-  public static <W extends TNumber, T extends TNumber, U extends TNumber, V extends TNumber> StatelessRandomBinomial<W> create(Scope scope, Operand<T> shape, Operand<U> seed, Operand<V> counts, Operand<V> probs, DataType<W> dtype) {
+  public static <W extends TNumber, T extends TNumber, U extends TNumber, V extends TNumber> StatelessRandomBinomial<W> create(Scope scope, Operand<T> shape, Operand<U> seed, Operand<V> counts, Operand<V> probs, Class<W> dtype) {
     OperationBuilder opBuilder = scope.env().opBuilder("StatelessRandomBinomial", scope.makeOpName("StatelessRandomBinomial"));
     opBuilder.addInput(shape.asOutput(scope));
     opBuilder.addInput(seed.asOutput(scope));
@@ -79,7 +78,7 @@ public static <W extends TNumber, T extends TNumber, U extends TNumber, V extend
    */
   @Endpoint(describeByClass = true)
   public static <T extends TNumber, U extends TNumber, V extends TNumber> StatelessRandomBinomial<TInt64> create(Scope scope, Operand<T> shape, Operand<U> seed, Operand<V> counts, Operand<V> probs) {
-    return create(scope, shape, seed, counts, probs, TInt64.DTYPE);
+    return create(scope, shape, seed, counts, probs, TInt64.class);
   }
   
   /**
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessRandomNormal.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessRandomNormal.java
index 76c3aa9406a..bd7eb1e5fe8 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessRandomNormal.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessRandomNormal.java
@@ -17,7 +17,6 @@
 
 package org.tensorflow.op.random;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -51,7 +50,7 @@ public final class StatelessRandomNormal<V extends TNumber> extends RawOp implem
    * @return a new instance of StatelessRandomNormal
    */
   @Endpoint(describeByClass = true)
-  public static <V extends TNumber, T extends TNumber, U extends TNumber> StatelessRandomNormal<V> create(Scope scope, Operand<T> shape, Operand<U> seed, DataType<V> dtype) {
+  public static <V extends TNumber, T extends TNumber, U extends TNumber> StatelessRandomNormal<V> create(Scope scope, Operand<T> shape, Operand<U> seed, Class<V> dtype) {
     OperationBuilder opBuilder = scope.env().opBuilder("StatelessRandomNormal", scope.makeOpName("StatelessRandomNormal"));
     opBuilder.addInput(shape.asOutput(scope));
     opBuilder.addInput(seed.asOutput(scope));
@@ -70,7 +69,7 @@ public static <V extends TNumber, T extends TNumber, U extends TNumber> Stateles
    */
   @Endpoint(describeByClass = true)
   public static <T extends TNumber, U extends TNumber> StatelessRandomNormal<TFloat32> create(Scope scope, Operand<T> shape, Operand<U> seed) {
-    return create(scope, shape, seed, TFloat32.DTYPE);
+    return create(scope, shape, seed, TFloat32.class);
   }
   
   /**
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessRandomPoisson.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessRandomPoisson.java
index 10dc4f9bfab..13cb97ada3c 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessRandomPoisson.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessRandomPoisson.java
@@ -17,7 +17,6 @@
 
 package org.tensorflow.op.random;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -51,7 +50,7 @@ public final class StatelessRandomPoisson<W extends TNumber> extends RawOp imple
    * @return a new instance of StatelessRandomPoisson
    */
   @Endpoint(describeByClass = true)
-  public static <W extends TNumber, T extends TNumber, U extends TNumber, V extends TNumber> StatelessRandomPoisson<W> create(Scope scope, Operand<T> shape, Operand<U> seed, Operand<V> lam, DataType<W> dtype) {
+  public static <W extends TNumber, T extends TNumber, U extends TNumber, V extends TNumber> StatelessRandomPoisson<W> create(Scope scope, Operand<T> shape, Operand<U> seed, Operand<V> lam, Class<W> dtype) {
     OperationBuilder opBuilder = scope.env().opBuilder("StatelessRandomPoisson", scope.makeOpName("StatelessRandomPoisson"));
     opBuilder.addInput(shape.asOutput(scope));
     opBuilder.addInput(seed.asOutput(scope));
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessRandomUniform.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessRandomUniform.java
index d6718eeeca4..797f660371f 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessRandomUniform.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessRandomUniform.java
@@ -17,7 +17,6 @@
 
 package org.tensorflow.op.random;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -52,7 +51,7 @@ public final class StatelessRandomUniform<V extends TNumber> extends RawOp imple
    * @return a new instance of StatelessRandomUniform
    */
   @Endpoint(describeByClass = true)
-  public static <V extends TNumber, T extends TNumber, U extends TNumber> StatelessRandomUniform<V> create(Scope scope, Operand<T> shape, Operand<U> seed, DataType<V> dtype) {
+  public static <V extends TNumber, T extends TNumber, U extends TNumber> StatelessRandomUniform<V> create(Scope scope, Operand<T> shape, Operand<U> seed, Class<V> dtype) {
     OperationBuilder opBuilder = scope.env().opBuilder("StatelessRandomUniform", scope.makeOpName("StatelessRandomUniform"));
     opBuilder.addInput(shape.asOutput(scope));
     opBuilder.addInput(seed.asOutput(scope));
@@ -71,7 +70,7 @@ public static <V extends TNumber, T extends TNumber, U extends TNumber> Stateles
    */
   @Endpoint(describeByClass = true)
   public static <T extends TNumber, U extends TNumber> StatelessRandomUniform<TFloat32> create(Scope scope, Operand<T> shape, Operand<U> seed) {
-    return create(scope, shape, seed, TFloat32.DTYPE);
+    return create(scope, shape, seed, TFloat32.class);
   }
   
   /**
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessRandomUniformFullInt.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessRandomUniformFullInt.java
index a1bcc25573f..5132de53d4d 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessRandomUniformFullInt.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessRandomUniformFullInt.java
@@ -17,7 +17,6 @@
 
 package org.tensorflow.op.random;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -49,7 +48,7 @@ public final class StatelessRandomUniformFullInt<V extends TNumber> extends RawO
    * @return a new instance of StatelessRandomUniformFullInt
    */
   @Endpoint(describeByClass = true)
-  public static <V extends TNumber, T extends TNumber, U extends TNumber> StatelessRandomUniformFullInt<V> create(Scope scope, Operand<T> shape, Operand<U> seed, DataType<V> dtype) {
+  public static <V extends TNumber, T extends TNumber, U extends TNumber> StatelessRandomUniformFullInt<V> create(Scope scope, Operand<T> shape, Operand<U> seed, Class<V> dtype) {
     OperationBuilder opBuilder = scope.env().opBuilder("StatelessRandomUniformFullInt", scope.makeOpName("StatelessRandomUniformFullInt"));
     opBuilder.addInput(shape.asOutput(scope));
     opBuilder.addInput(seed.asOutput(scope));
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessTruncatedNormal.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessTruncatedNormal.java
index ca1ee733f62..81dfc469250 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessTruncatedNormal.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessTruncatedNormal.java
@@ -17,7 +17,6 @@
 
 package org.tensorflow.op.random;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -53,7 +52,7 @@ public final class StatelessTruncatedNormal<V extends TNumber> extends RawOp imp
    * @return a new instance of StatelessTruncatedNormal
    */
   @Endpoint(describeByClass = true)
-  public static <V extends TNumber, T extends TNumber, U extends TNumber> StatelessTruncatedNormal<V> create(Scope scope, Operand<T> shape, Operand<U> seed, DataType<V> dtype) {
+  public static <V extends TNumber, T extends TNumber, U extends TNumber> StatelessTruncatedNormal<V> create(Scope scope, Operand<T> shape, Operand<U> seed, Class<V> dtype) {
     OperationBuilder opBuilder = scope.env().opBuilder("StatelessTruncatedNormal", scope.makeOpName("StatelessTruncatedNormal"));
     opBuilder.addInput(shape.asOutput(scope));
     opBuilder.addInput(seed.asOutput(scope));
@@ -72,7 +71,7 @@ public static <V extends TNumber, T extends TNumber, U extends TNumber> Stateles
    */
   @Endpoint(describeByClass = true)
   public static <T extends TNumber, U extends TNumber> StatelessTruncatedNormal<TFloat32> create(Scope scope, Operand<T> shape, Operand<U> seed) {
-    return create(scope, shape, seed, TFloat32.DTYPE);
+    return create(scope, shape, seed, TFloat32.class);
   }
   
   /**
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/TruncatedNormal.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/TruncatedNormal.java
index e9588a4ab14..32be4a817a4 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/TruncatedNormal.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/TruncatedNormal.java
@@ -17,7 +17,6 @@
 
 package org.tensorflow.op.random;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -80,7 +79,7 @@ private Options() {
    * @return a new instance of TruncatedNormal
    */
   @Endpoint(describeByClass = true)
-  public static <U extends TNumber, T extends TNumber> TruncatedNormal<U> create(Scope scope, Operand<T> shape, DataType<U> dtype, Options... options) {
+  public static <U extends TNumber, T extends TNumber> TruncatedNormal<U> create(Scope scope, Operand<T> shape, Class<U> dtype, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("TruncatedNormal", scope.makeOpName("TruncatedNormal"));
     opBuilder.addInput(shape.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/experimental/DummySeedGenerator.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/experimental/DummySeedGenerator.java
index dd537fa2d68..dff26846e0f 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/experimental/DummySeedGenerator.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/experimental/DummySeedGenerator.java
@@ -52,7 +52,7 @@ public Output<?> handle() {
   
   @Override
   @SuppressWarnings("unchecked")
-  public Output<TType> asOutput() {
+  public Output<TType> asOutput(Scope scope) {
     return (Output<TType>) handle;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Irfft.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Irfft.java
index 3091bcdcff9..712e3b714e2 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Irfft.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Irfft.java
@@ -17,7 +17,6 @@
 
 package org.tensorflow.op.signal;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -63,7 +62,7 @@ public final class Irfft<U extends TNumber> extends RawOp implements Operand<U>
    * @return a new instance of Irfft
    */
   @Endpoint(describeByClass = true)
-  public static <U extends TNumber, T extends TType> Irfft<U> create(Scope scope, Operand<T> input, Operand<TInt32> fftLength, DataType<U> Treal) {
+  public static <U extends TNumber, T extends TType> Irfft<U> create(Scope scope, Operand<T> input, Operand<TInt32> fftLength, Class<U> Treal) {
     OperationBuilder opBuilder = scope.env().opBuilder("IRFFT", scope.makeOpName("Irfft"));
     opBuilder.addInput(input.asOutput(scope));
     opBuilder.addInput(fftLength.asOutput(scope));
@@ -82,7 +81,7 @@ public static <U extends TNumber, T extends TType> Irfft<U> create(Scope scope,
    */
   @Endpoint(describeByClass = true)
   public static <T extends TType> Irfft<TFloat32> create(Scope scope, Operand<T> input, Operand<TInt32> fftLength) {
-    return create(scope, input, fftLength, TFloat32.DTYPE);
+    return create(scope, input, fftLength, TFloat32.class);
   }
   
   /**
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Irfft2d.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Irfft2d.java
index 27af3c16aef..bc3a340d210 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Irfft2d.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Irfft2d.java
@@ -17,7 +17,6 @@
 
 package org.tensorflow.op.signal;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -64,7 +63,7 @@ public final class Irfft2d<U extends TNumber> extends RawOp implements Operand<U
    * @return a new instance of Irfft2d
    */
   @Endpoint(describeByClass = true)
-  public static <U extends TNumber, T extends TType> Irfft2d<U> create(Scope scope, Operand<T> input, Operand<TInt32> fftLength, DataType<U> Treal) {
+  public static <U extends TNumber, T extends TType> Irfft2d<U> create(Scope scope, Operand<T> input, Operand<TInt32> fftLength, Class<U> Treal) {
     OperationBuilder opBuilder = scope.env().opBuilder("IRFFT2D", scope.makeOpName("Irfft2d"));
     opBuilder.addInput(input.asOutput(scope));
     opBuilder.addInput(fftLength.asOutput(scope));
@@ -83,7 +82,7 @@ public static <U extends TNumber, T extends TType> Irfft2d<U> create(Scope scope
    */
   @Endpoint(describeByClass = true)
   public static <T extends TType> Irfft2d<TFloat32> create(Scope scope, Operand<T> input, Operand<TInt32> fftLength) {
-    return create(scope, input, fftLength, TFloat32.DTYPE);
+    return create(scope, input, fftLength, TFloat32.class);
   }
   
   /**
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Irfft3d.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Irfft3d.java
index 25a5ade8b6f..d4306fe1c2a 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Irfft3d.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Irfft3d.java
@@ -17,7 +17,6 @@
 
 package org.tensorflow.op.signal;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -64,7 +63,7 @@ public final class Irfft3d<U extends TNumber> extends RawOp implements Operand<U
    * @return a new instance of Irfft3d
    */
   @Endpoint(describeByClass = true)
-  public static <U extends TNumber, T extends TType> Irfft3d<U> create(Scope scope, Operand<T> input, Operand<TInt32> fftLength, DataType<U> Treal) {
+  public static <U extends TNumber, T extends TType> Irfft3d<U> create(Scope scope, Operand<T> input, Operand<TInt32> fftLength, Class<U> Treal) {
     OperationBuilder opBuilder = scope.env().opBuilder("IRFFT3D", scope.makeOpName("Irfft3d"));
     opBuilder.addInput(input.asOutput(scope));
     opBuilder.addInput(fftLength.asOutput(scope));
@@ -83,7 +82,7 @@ public static <U extends TNumber, T extends TType> Irfft3d<U> create(Scope scope
    */
   @Endpoint(describeByClass = true)
   public static <T extends TType> Irfft3d<TFloat32> create(Scope scope, Operand<T> input, Operand<TInt32> fftLength) {
-    return create(scope, input, fftLength, TFloat32.DTYPE);
+    return create(scope, input, fftLength, TFloat32.class);
   }
   
   /**
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Rfft.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Rfft.java
index bbd122f47fc..c4a47a60b8d 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Rfft.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Rfft.java
@@ -17,7 +17,6 @@
 
 package org.tensorflow.op.signal;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -59,7 +58,7 @@ public final class Rfft<U extends TType> extends RawOp implements Operand<U> {
    * @return a new instance of Rfft
    */
   @Endpoint(describeByClass = true)
-  public static <U extends TType, T extends TNumber> Rfft<U> create(Scope scope, Operand<T> input, Operand<TInt32> fftLength, DataType<U> Tcomplex) {
+  public static <U extends TType, T extends TNumber> Rfft<U> create(Scope scope, Operand<T> input, Operand<TInt32> fftLength, Class<U> Tcomplex) {
     OperationBuilder opBuilder = scope.env().opBuilder("RFFT", scope.makeOpName("Rfft"));
     opBuilder.addInput(input.asOutput(scope));
     opBuilder.addInput(fftLength.asOutput(scope));
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Rfft2d.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Rfft2d.java
index c206590bf30..173b09c53fd 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Rfft2d.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Rfft2d.java
@@ -17,7 +17,6 @@
 
 package org.tensorflow.op.signal;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -60,7 +59,7 @@ public final class Rfft2d<U extends TType> extends RawOp implements Operand<U> {
    * @return a new instance of Rfft2d
    */
   @Endpoint(describeByClass = true)
-  public static <U extends TType, T extends TNumber> Rfft2d<U> create(Scope scope, Operand<T> input, Operand<TInt32> fftLength, DataType<U> Tcomplex) {
+  public static <U extends TType, T extends TNumber> Rfft2d<U> create(Scope scope, Operand<T> input, Operand<TInt32> fftLength, Class<U> Tcomplex) {
     OperationBuilder opBuilder = scope.env().opBuilder("RFFT2D", scope.makeOpName("Rfft2d"));
     opBuilder.addInput(input.asOutput(scope));
     opBuilder.addInput(fftLength.asOutput(scope));
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Rfft3d.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Rfft3d.java
index 36d3facf770..48620690a7e 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Rfft3d.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Rfft3d.java
@@ -17,7 +17,6 @@
 
 package org.tensorflow.op.signal;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -60,7 +59,7 @@ public final class Rfft3d<U extends TType> extends RawOp implements Operand<U> {
    * @return a new instance of Rfft3d
    */
   @Endpoint(describeByClass = true)
-  public static <U extends TType, T extends TNumber> Rfft3d<U> create(Scope scope, Operand<T> input, Operand<TInt32> fftLength, DataType<U> Tcomplex) {
+  public static <U extends TType, T extends TNumber> Rfft3d<U> create(Scope scope, Operand<T> input, Operand<TInt32> fftLength, Class<U> Tcomplex) {
     OperationBuilder opBuilder = scope.env().opBuilder("RFFT3D", scope.makeOpName("Rfft3d"));
     opBuilder.addInput(input.asOutput(scope));
     opBuilder.addInput(fftLength.asOutput(scope));
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/DenseCountSparseOutput.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/DenseCountSparseOutput.java
index ed390a7ba47..e59f5dbe644 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/DenseCountSparseOutput.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/DenseCountSparseOutput.java
@@ -27,7 +27,6 @@
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Performs sparse-output bin counting for a tf.tensor input.
@@ -80,8 +79,8 @@ private Options() {
   @Endpoint(describeByClass = true)
   public static <U extends TNumber, T extends TNumber> DenseCountSparseOutput<U> create(Scope scope, Operand<T> values, Operand<U> weights, Boolean binaryOutput, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("DenseCountSparseOutput", scope.makeOpName("DenseCountSparseOutput"));
-    opBuilder.addInput(values.asOutput());
-    opBuilder.addInput(weights.asOutput());
+    opBuilder.addInput(values.asOutput(scope));
+    opBuilder.addInput(weights.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("binary_output", binaryOutput);
     if (options != null) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/DeserializeSparse.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/DeserializeSparse.java
index 76be1dcb2d8..b079981f9b1 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/DeserializeSparse.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/DeserializeSparse.java
@@ -17,7 +17,6 @@
 
 package org.tensorflow.op.sparse;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -89,7 +88,7 @@ public final class DeserializeSparse<U extends TType> extends RawOp {
    * @return a new instance of DeserializeSparse
    */
   @Endpoint(describeByClass = true)
-  public static <U extends TType, T extends TType> DeserializeSparse<U> create(Scope scope, Operand<T> serializedSparse, DataType<U> dtype) {
+  public static <U extends TType, T extends TType> DeserializeSparse<U> create(Scope scope, Operand<T> serializedSparse, Class<U> dtype) {
     OperationBuilder opBuilder = scope.env().opBuilder("DeserializeSparse", scope.makeOpName("DeserializeSparse"));
     opBuilder.addInput(serializedSparse.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseAccumulatorTakeGradient.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseAccumulatorTakeGradient.java
index 50836e4c0a4..75f34a7155c 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseAccumulatorTakeGradient.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseAccumulatorTakeGradient.java
@@ -17,7 +17,6 @@
 
 package org.tensorflow.op.sparse;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -57,7 +56,7 @@ public final class SparseAccumulatorTakeGradient<T extends TType> extends RawOp
    * @return a new instance of SparseAccumulatorTakeGradient
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> SparseAccumulatorTakeGradient<T> create(Scope scope, Operand<TString> handle, Operand<TInt32> numRequired, DataType<T> dtype) {
+  public static <T extends TType> SparseAccumulatorTakeGradient<T> create(Scope scope, Operand<TString> handle, Operand<TInt32> numRequired, Class<T> dtype) {
     OperationBuilder opBuilder = scope.env().opBuilder("SparseAccumulatorTakeGradient", scope.makeOpName("SparseAccumulatorTakeGradient"));
     opBuilder.addInput(handle.asOutput(scope));
     opBuilder.addInput(numRequired.asOutput(scope));
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseBincount.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseBincount.java
index 344e27f1346..3f41ac7bb37 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseBincount.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseBincount.java
@@ -27,7 +27,6 @@
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Counts the number of occurrences of each value in an integer array.
@@ -81,11 +80,11 @@ private Options() {
   @Endpoint(describeByClass = true)
   public static <U extends TNumber, T extends TNumber> SparseBincount<U> create(Scope scope, Operand<TInt64> indices, Operand<T> values, Operand<TInt64> denseShape, Operand<T> size, Operand<U> weights, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("SparseBincount", scope.makeOpName("SparseBincount"));
-    opBuilder.addInput(indices.asOutput());
-    opBuilder.addInput(values.asOutput());
-    opBuilder.addInput(denseShape.asOutput());
-    opBuilder.addInput(size.asOutput());
-    opBuilder.addInput(weights.asOutput());
+    opBuilder.addInput(indices.asOutput(scope));
+    opBuilder.addInput(values.asOutput(scope));
+    opBuilder.addInput(denseShape.asOutput(scope));
+    opBuilder.addInput(size.asOutput(scope));
+    opBuilder.addInput(weights.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -113,7 +112,7 @@ public Output<U> output() {
   }
   
   @Override
-  public Output<U> asOutput() {
+  public Output<U> asOutput(Scope scope) {
     return output;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseConditionalAccumulator.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseConditionalAccumulator.java
index 7be9dee9733..045169e2f9a 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseConditionalAccumulator.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseConditionalAccumulator.java
@@ -17,7 +17,6 @@
 
 package org.tensorflow.op.sparse;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -92,7 +91,7 @@ private Options() {
    * @return a new instance of SparseConditionalAccumulator
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> SparseConditionalAccumulator create(Scope scope, DataType<T> dtype, Shape shape, Options... options) {
+  public static <T extends TType> SparseConditionalAccumulator create(Scope scope, Class<T> dtype, Shape shape, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("SparseConditionalAccumulator", scope.makeOpName("SparseConditionalAccumulator"));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("dtype", dtype);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseCountSparseOutput.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseCountSparseOutput.java
index 5e5566db5ec..452a1830055 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseCountSparseOutput.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseCountSparseOutput.java
@@ -27,7 +27,6 @@
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Performs sparse-output bin counting for a sparse tensor input.
@@ -82,10 +81,10 @@ private Options() {
   @Endpoint(describeByClass = true)
   public static <U extends TNumber, T extends TNumber> SparseCountSparseOutput<U> create(Scope scope, Operand<TInt64> indices, Operand<T> values, Operand<TInt64> denseShape, Operand<U> weights, Boolean binaryOutput, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("SparseCountSparseOutput", scope.makeOpName("SparseCountSparseOutput"));
-    opBuilder.addInput(indices.asOutput());
-    opBuilder.addInput(values.asOutput());
-    opBuilder.addInput(denseShape.asOutput());
-    opBuilder.addInput(weights.asOutput());
+    opBuilder.addInput(indices.asOutput(scope));
+    opBuilder.addInput(values.asOutput(scope));
+    opBuilder.addInput(denseShape.asOutput(scope));
+    opBuilder.addInput(weights.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("binary_output", binaryOutput);
     if (options != null) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseCross.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseCross.java
index 0c63e102925..0b3a25c3eb7 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseCross.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseCross.java
@@ -17,7 +17,6 @@
 
 package org.tensorflow.op.sparse;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -28,7 +27,7 @@
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
+import org.tensorflow.types.TString;
 
 /**
  * Generates sparse cross from a list of sparse and dense tensors.
@@ -69,11 +68,9 @@
  *     [1, 1]: FingerprintCat64(
  *                 Fingerprint64("g"), FingerprintCat64(
  *                     Fingerprint64("e"), Fingerprint64("c")))
- * 
- * @param <T> data type for {@code outputValues()} output
  */
 @Operator(group = "sparse")
-public final class SparseCross<T extends TType> extends RawOp {
+public final class SparseCross extends RawOp {
   
   /**
    * Factory method to create a class wrapping a new SparseCross operation.
@@ -83,30 +80,19 @@ public final class SparseCross<T extends TType> extends RawOp {
    * @param values 1-D.   values of each `SparseTensor`.
    * @param shapes 1-D.   Shapes of each `SparseTensor`.
    * @param denseInputs 2-D.    Columns represented by dense `Tensor`.
-   * @param hashedOutput If true, returns the hash of the cross instead of the string.
-   * This will allow us avoiding string manipulations.
-   * @param numBuckets It is used if hashed_output is true.
-   * output = hashed_value%num_buckets if num_buckets > 0 else hashed_value.
-   * @param hashKey Specify the hash_key that will be used by the `FingerprintCat64`
-   * function to combine the crosses fingerprints.
-   * @param outType 
-   * @param internalType 
+   * @param sep string used when joining a list of string inputs, can be used as separator later.
    * @return a new instance of SparseCross
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TType> SparseCross<T> create(Scope scope, Iterable<Operand<TInt64>> indices, Iterable<Operand<?>> values, Iterable<Operand<TInt64>> shapes, Iterable<Operand<?>> denseInputs, Boolean hashedOutput, Long numBuckets, Long hashKey, DataType<T> outType, DataType<U> internalType) {
-    OperationBuilder opBuilder = scope.env().opBuilder("SparseCross", scope.makeOpName("SparseCross"));
+  public static SparseCross create(Scope scope, Iterable<Operand<TInt64>> indices, Iterable<Operand<?>> values, Iterable<Operand<TInt64>> shapes, Iterable<Operand<?>> denseInputs, Operand<TString> sep) {
+    OperationBuilder opBuilder = scope.env().opBuilder("SparseCrossV2", scope.makeOpName("SparseCross"));
     opBuilder.addInputList(Operands.asOutputs(scope, indices));
     opBuilder.addInputList(Operands.asOutputs(scope, values));
     opBuilder.addInputList(Operands.asOutputs(scope, shapes));
     opBuilder.addInputList(Operands.asOutputs(scope, denseInputs));
+    opBuilder.addInput(sep.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("hashed_output", hashedOutput);
-    opBuilder.setAttr("num_buckets", numBuckets);
-    opBuilder.setAttr("hash_key", hashKey);
-    opBuilder.setAttr("out_type", outType);
-    opBuilder.setAttr("internal_type", internalType);
-    return new SparseCross<T>(opBuilder.build());
+    return new SparseCross(opBuilder.build());
   }
   
   /**
@@ -120,7 +106,7 @@ public Output<TInt64> outputIndices() {
    * 1-D.  Non-empty values of the concatenated or hashed
    * `SparseTensor`.
    */
-  public Output<T> outputValues() {
+  public Output<TString> outputValues() {
     return outputValues;
   }
   
@@ -131,8 +117,11 @@ public Output<TInt64> outputShape() {
     return outputShape;
   }
   
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "SparseCrossV2";
+  
   private Output<TInt64> outputIndices;
-  private Output<T> outputValues;
+  private Output<TString> outputValues;
   private Output<TInt64> outputShape;
   
   private SparseCross(Operation operation) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseCrossHashed.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseCrossHashed.java
index 2fc6976079e..9b65e55d0b4 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseCrossHashed.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseCrossHashed.java
@@ -89,13 +89,13 @@ public final class SparseCrossHashed extends RawOp {
   @Endpoint(describeByClass = true)
   public static SparseCrossHashed create(Scope scope, Iterable<Operand<TInt64>> indices, Iterable<Operand<?>> values, Iterable<Operand<TInt64>> shapes, Iterable<Operand<?>> denseInputs, Operand<TInt64> numBuckets, Operand<TBool> strongHash, Operand<TInt64> salt) {
     OperationBuilder opBuilder = scope.env().opBuilder("SparseCrossHashed", scope.makeOpName("SparseCrossHashed"));
-    opBuilder.addInputList(Operands.asOutputs(indices));
-    opBuilder.addInputList(Operands.asOutputs(values));
-    opBuilder.addInputList(Operands.asOutputs(shapes));
-    opBuilder.addInputList(Operands.asOutputs(denseInputs));
-    opBuilder.addInput(numBuckets.asOutput());
-    opBuilder.addInput(strongHash.asOutput());
-    opBuilder.addInput(salt.asOutput());
+    opBuilder.addInputList(Operands.asOutputs(scope, indices));
+    opBuilder.addInputList(Operands.asOutputs(scope, values));
+    opBuilder.addInputList(Operands.asOutputs(scope, shapes));
+    opBuilder.addInputList(Operands.asOutputs(scope, denseInputs));
+    opBuilder.addInput(numBuckets.asOutput(scope));
+    opBuilder.addInput(strongHash.asOutput(scope));
+    opBuilder.addInput(salt.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     return new SparseCrossHashed(opBuilder.build());
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentMean.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentMean.java
index 8cdeb61508f..9287fb2688a 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentMean.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentMean.java
@@ -25,7 +25,6 @@
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
 import org.tensorflow.types.family.TNumber;
 
 /**
@@ -51,7 +50,7 @@ public final class SparseSegmentMean<T extends TNumber> extends RawOp implements
    * @return a new instance of SparseSegmentMean
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber, U extends TNumber> SparseSegmentMean<T> create(Scope scope, Operand<T> data, Operand<U> indices, Operand<TInt32> segmentIds) {
+  public static <T extends TNumber, U extends TNumber, V extends TNumber> SparseSegmentMean<T> create(Scope scope, Operand<T> data, Operand<U> indices, Operand<V> segmentIds) {
     OperationBuilder opBuilder = scope.env().opBuilder("SparseSegmentMean", scope.makeOpName("SparseSegmentMean"));
     opBuilder.addInput(data.asOutput(scope));
     opBuilder.addInput(indices.asOutput(scope));
@@ -73,6 +72,9 @@ public Output<T> asOutput(Scope scope) {
     return output;
   }
   
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "SparseSegmentMean";
+  
   private Output<T> output;
   
   private SparseSegmentMean(Operation operation) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentMeanGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentMeanGrad.java
index f12deda58e8..4aa3800c0d4 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentMeanGrad.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentMeanGrad.java
@@ -50,7 +50,7 @@ public final class SparseSegmentMeanGrad<T extends TNumber> extends RawOp implem
    * @return a new instance of SparseSegmentMeanGrad
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber, U extends TNumber> SparseSegmentMeanGrad<T> create(Scope scope, Operand<T> grad, Operand<U> indices, Operand<TInt32> segmentIds, Operand<TInt32> outputDim0) {
+  public static <T extends TNumber, U extends TNumber, V extends TNumber> SparseSegmentMeanGrad<T> create(Scope scope, Operand<T> grad, Operand<U> indices, Operand<V> segmentIds, Operand<TInt32> outputDim0) {
     OperationBuilder opBuilder = scope.env().opBuilder("SparseSegmentMeanGrad", scope.makeOpName("SparseSegmentMeanGrad"));
     opBuilder.addInput(grad.asOutput(scope));
     opBuilder.addInput(indices.asOutput(scope));
@@ -71,6 +71,9 @@ public Output<T> asOutput(Scope scope) {
     return output;
   }
   
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "SparseSegmentMeanGrad";
+  
   private Output<T> output;
   
   private SparseSegmentMeanGrad(Operation operation) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentMeanWithNumSegments.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentMeanWithNumSegments.java
index 2442e767106..14c069899dc 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentMeanWithNumSegments.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentMeanWithNumSegments.java
@@ -25,14 +25,13 @@
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
 import org.tensorflow.types.family.TNumber;
 
 /**
  * Computes the mean along sparse segments of a tensor.
  * <p>
  * Like `SparseSegmentMean`, but allows missing ids in `segment_ids`. If an id is
- * misisng, the `output` tensor at that position will be zeroed.
+ * missing, the `output` tensor at that position will be zeroed.
  * <p>
  * Read
  * [the section on segmentation](https://tensorflow.org/api_docs/python/tf/math#Segmentation)
@@ -54,7 +53,7 @@ public final class SparseSegmentMeanWithNumSegments<T extends TNumber> extends R
    * @return a new instance of SparseSegmentMeanWithNumSegments
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber, U extends TNumber, V extends TNumber> SparseSegmentMeanWithNumSegments<T> create(Scope scope, Operand<T> data, Operand<U> indices, Operand<TInt32> segmentIds, Operand<V> numSegments) {
+  public static <T extends TNumber, U extends TNumber, V extends TNumber, W extends TNumber> SparseSegmentMeanWithNumSegments<T> create(Scope scope, Operand<T> data, Operand<U> indices, Operand<V> segmentIds, Operand<W> numSegments) {
     OperationBuilder opBuilder = scope.env().opBuilder("SparseSegmentMeanWithNumSegments", scope.makeOpName("SparseSegmentMeanWithNumSegments"));
     opBuilder.addInput(data.asOutput(scope));
     opBuilder.addInput(indices.asOutput(scope));
@@ -77,6 +76,9 @@ public Output<T> asOutput(Scope scope) {
     return output;
   }
   
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "SparseSegmentMeanWithNumSegments";
+  
   private Output<T> output;
   
   private SparseSegmentMeanWithNumSegments(Operation operation) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentSqrtN.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentSqrtN.java
index e2e2b56e4f7..93851b0abf1 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentSqrtN.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentSqrtN.java
@@ -25,7 +25,6 @@
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
 import org.tensorflow.types.family.TNumber;
 
 /**
@@ -51,7 +50,7 @@ public final class SparseSegmentSqrtN<T extends TNumber> extends RawOp implement
    * @return a new instance of SparseSegmentSqrtN
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber, U extends TNumber> SparseSegmentSqrtN<T> create(Scope scope, Operand<T> data, Operand<U> indices, Operand<TInt32> segmentIds) {
+  public static <T extends TNumber, U extends TNumber, V extends TNumber> SparseSegmentSqrtN<T> create(Scope scope, Operand<T> data, Operand<U> indices, Operand<V> segmentIds) {
     OperationBuilder opBuilder = scope.env().opBuilder("SparseSegmentSqrtN", scope.makeOpName("SparseSegmentSqrtN"));
     opBuilder.addInput(data.asOutput(scope));
     opBuilder.addInput(indices.asOutput(scope));
@@ -73,6 +72,9 @@ public Output<T> asOutput(Scope scope) {
     return output;
   }
   
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "SparseSegmentSqrtN";
+  
   private Output<T> output;
   
   private SparseSegmentSqrtN(Operation operation) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentSqrtNGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentSqrtNGrad.java
index 93b737c96a8..22a6610f717 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentSqrtNGrad.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentSqrtNGrad.java
@@ -50,7 +50,7 @@ public final class SparseSegmentSqrtNGrad<T extends TNumber> extends RawOp imple
    * @return a new instance of SparseSegmentSqrtNGrad
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber, U extends TNumber> SparseSegmentSqrtNGrad<T> create(Scope scope, Operand<T> grad, Operand<U> indices, Operand<TInt32> segmentIds, Operand<TInt32> outputDim0) {
+  public static <T extends TNumber, U extends TNumber, V extends TNumber> SparseSegmentSqrtNGrad<T> create(Scope scope, Operand<T> grad, Operand<U> indices, Operand<V> segmentIds, Operand<TInt32> outputDim0) {
     OperationBuilder opBuilder = scope.env().opBuilder("SparseSegmentSqrtNGrad", scope.makeOpName("SparseSegmentSqrtNGrad"));
     opBuilder.addInput(grad.asOutput(scope));
     opBuilder.addInput(indices.asOutput(scope));
@@ -71,6 +71,9 @@ public Output<T> asOutput(Scope scope) {
     return output;
   }
   
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "SparseSegmentSqrtNGrad";
+  
   private Output<T> output;
   
   private SparseSegmentSqrtNGrad(Operation operation) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentSqrtNWithNumSegments.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentSqrtNWithNumSegments.java
index a0b30e3fca1..8b87a35353a 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentSqrtNWithNumSegments.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentSqrtNWithNumSegments.java
@@ -25,7 +25,6 @@
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
 import org.tensorflow.types.family.TNumber;
 
 /**
@@ -34,7 +33,7 @@
  * N is the size of the segment being reduced.
  * <p>
  * Like `SparseSegmentSqrtN`, but allows missing ids in `segment_ids`. If an id is
- * misisng, the `output` tensor at that position will be zeroed.
+ * missing, the `output` tensor at that position will be zeroed.
  * <p>
  * Read
  * [the section on segmentation](https://tensorflow.org/api_docs/python/tf/math#Segmentation)
@@ -56,7 +55,7 @@ public final class SparseSegmentSqrtNWithNumSegments<T extends TNumber> extends
    * @return a new instance of SparseSegmentSqrtNWithNumSegments
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber, U extends TNumber, V extends TNumber> SparseSegmentSqrtNWithNumSegments<T> create(Scope scope, Operand<T> data, Operand<U> indices, Operand<TInt32> segmentIds, Operand<V> numSegments) {
+  public static <T extends TNumber, U extends TNumber, V extends TNumber, W extends TNumber> SparseSegmentSqrtNWithNumSegments<T> create(Scope scope, Operand<T> data, Operand<U> indices, Operand<V> segmentIds, Operand<W> numSegments) {
     OperationBuilder opBuilder = scope.env().opBuilder("SparseSegmentSqrtNWithNumSegments", scope.makeOpName("SparseSegmentSqrtNWithNumSegments"));
     opBuilder.addInput(data.asOutput(scope));
     opBuilder.addInput(indices.asOutput(scope));
@@ -79,6 +78,9 @@ public Output<T> asOutput(Scope scope) {
     return output;
   }
   
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "SparseSegmentSqrtNWithNumSegments";
+  
   private Output<T> output;
   
   private SparseSegmentSqrtNWithNumSegments(Operation operation) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentSum.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentSum.java
index f539b93ab86..b393d51e753 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentSum.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentSum.java
@@ -25,7 +25,6 @@
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
 import org.tensorflow.types.family.TNumber;
 
 /**
@@ -76,7 +75,7 @@ public final class SparseSegmentSum<T extends TNumber> extends RawOp implements
    * @return a new instance of SparseSegmentSum
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber, U extends TNumber> SparseSegmentSum<T> create(Scope scope, Operand<T> data, Operand<U> indices, Operand<TInt32> segmentIds) {
+  public static <T extends TNumber, U extends TNumber, V extends TNumber> SparseSegmentSum<T> create(Scope scope, Operand<T> data, Operand<U> indices, Operand<V> segmentIds) {
     OperationBuilder opBuilder = scope.env().opBuilder("SparseSegmentSum", scope.makeOpName("SparseSegmentSum"));
     opBuilder.addInput(data.asOutput(scope));
     opBuilder.addInput(indices.asOutput(scope));
@@ -98,6 +97,9 @@ public Output<T> asOutput(Scope scope) {
     return output;
   }
   
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "SparseSegmentSum";
+  
   private Output<T> output;
   
   private SparseSegmentSum(Operation operation) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentSumWithNumSegments.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentSumWithNumSegments.java
index ac8c400fa15..1cf4df85132 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentSumWithNumSegments.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentSumWithNumSegments.java
@@ -25,14 +25,13 @@
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
 import org.tensorflow.types.family.TNumber;
 
 /**
  * Computes the sum along sparse segments of a tensor.
  * <p>
  * Like `SparseSegmentSum`, but allows missing ids in `segment_ids`. If an id is
- * misisng, the `output` tensor at that position will be zeroed.
+ * missing, the `output` tensor at that position will be zeroed.
  * <p>
  * Read
  * [the section on segmentation](https://tensorflow.org/api_docs/python/tf/sparse#Segmentation)
@@ -75,7 +74,7 @@ public final class SparseSegmentSumWithNumSegments<T extends TNumber> extends Ra
    * @return a new instance of SparseSegmentSumWithNumSegments
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber, U extends TNumber, V extends TNumber> SparseSegmentSumWithNumSegments<T> create(Scope scope, Operand<T> data, Operand<U> indices, Operand<TInt32> segmentIds, Operand<V> numSegments) {
+  public static <T extends TNumber, U extends TNumber, V extends TNumber, W extends TNumber> SparseSegmentSumWithNumSegments<T> create(Scope scope, Operand<T> data, Operand<U> indices, Operand<V> segmentIds, Operand<W> numSegments) {
     OperationBuilder opBuilder = scope.env().opBuilder("SparseSegmentSumWithNumSegments", scope.makeOpName("SparseSegmentSumWithNumSegments"));
     opBuilder.addInput(data.asOutput(scope));
     opBuilder.addInput(indices.asOutput(scope));
@@ -98,6 +97,9 @@ public Output<T> asOutput(Scope scope) {
     return output;
   }
   
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "SparseSegmentSumWithNumSegments";
+  
   private Output<T> output;
   
   private SparseSegmentSumWithNumSegments(Operation operation) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/TakeManySparseFromTensorsMap.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/TakeManySparseFromTensorsMap.java
index c8af50dc2bc..7cf6141873f 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/TakeManySparseFromTensorsMap.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/TakeManySparseFromTensorsMap.java
@@ -17,7 +17,6 @@
 
 package org.tensorflow.op.sparse;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -124,7 +123,7 @@ private Options() {
    * @return a new instance of TakeManySparseFromTensorsMap
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> TakeManySparseFromTensorsMap<T> create(Scope scope, Operand<TInt64> sparseHandles, DataType<T> dtype, Options... options) {
+  public static <T extends TType> TakeManySparseFromTensorsMap<T> create(Scope scope, Operand<TInt64> sparseHandles, Class<T> dtype, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("TakeManySparseFromTensorsMap", scope.makeOpName("TakeManySparseFromTensorsMap"));
     opBuilder.addInput(sparseHandles.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/ToNumber.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/ToNumber.java
index 604b8161cc6..9c70d36f10b 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/ToNumber.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/ToNumber.java
@@ -17,7 +17,6 @@
 
 package org.tensorflow.op.strings;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -57,7 +56,7 @@ public final class ToNumber<T extends TNumber> extends RawOp implements Operand<
    * @return a new instance of ToNumber
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> ToNumber<T> create(Scope scope, Operand<TString> stringTensor, DataType<T> outType) {
+  public static <T extends TNumber> ToNumber<T> create(Scope scope, Operand<TString> stringTensor, Class<T> outType) {
     OperationBuilder opBuilder = scope.env().opBuilder("StringToNumber", scope.makeOpName("ToNumber"));
     opBuilder.addInput(stringTensor.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
@@ -74,7 +73,7 @@ public static <T extends TNumber> ToNumber<T> create(Scope scope, Operand<TStrin
    */
   @Endpoint(describeByClass = true)
   public static ToNumber<TFloat32> create(Scope scope, Operand<TString> stringTensor) {
-    return create(scope, stringTensor, TFloat32.DTYPE);
+    return create(scope, stringTensor, TFloat32.class);
   }
   
   /**
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/UnicodeDecode.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/UnicodeDecode.java
index 7e04b21ed62..2b573f03eba 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/UnicodeDecode.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/UnicodeDecode.java
@@ -17,7 +17,6 @@
 
 package org.tensorflow.op.strings;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -115,7 +114,7 @@ private Options() {
    * @return a new instance of UnicodeDecode
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> UnicodeDecode<T> create(Scope scope, Operand<TString> input, String inputEncoding, DataType<T> Tsplits, Options... options) {
+  public static <T extends TNumber> UnicodeDecode<T> create(Scope scope, Operand<TString> input, String inputEncoding, Class<T> Tsplits, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("UnicodeDecode", scope.makeOpName("UnicodeDecode"));
     opBuilder.addInput(input.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
@@ -150,7 +149,7 @@ public static <T extends TNumber> UnicodeDecode<T> create(Scope scope, Operand<T
    */
   @Endpoint(describeByClass = true)
   public static UnicodeDecode<TInt64> create(Scope scope, Operand<TString> input, String inputEncoding, Options... options) {
-    return create(scope, input, inputEncoding, TInt64.DTYPE, options);
+    return create(scope, input, inputEncoding, TInt64.class, options);
   }
   
   /**
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/UnicodeDecodeWithOffsets.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/UnicodeDecodeWithOffsets.java
index 987c0d6834c..a13775dba39 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/UnicodeDecodeWithOffsets.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/UnicodeDecodeWithOffsets.java
@@ -17,7 +17,6 @@
 
 package org.tensorflow.op.strings;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -121,7 +120,7 @@ private Options() {
    * @return a new instance of UnicodeDecodeWithOffsets
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TNumber> UnicodeDecodeWithOffsets<T> create(Scope scope, Operand<TString> input, String inputEncoding, DataType<T> Tsplits, Options... options) {
+  public static <T extends TNumber> UnicodeDecodeWithOffsets<T> create(Scope scope, Operand<TString> input, String inputEncoding, Class<T> Tsplits, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("UnicodeDecodeWithOffsets", scope.makeOpName("UnicodeDecodeWithOffsets"));
     opBuilder.addInput(input.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
@@ -156,7 +155,7 @@ public static <T extends TNumber> UnicodeDecodeWithOffsets<T> create(Scope scope
    */
   @Endpoint(describeByClass = true)
   public static UnicodeDecodeWithOffsets<TInt64> create(Scope scope, Operand<TString> input, String inputEncoding, Options... options) {
-    return create(scope, input, inputEncoding, TInt64.DTYPE, options);
+    return create(scope, input, inputEncoding, TInt64.class, options);
   }
   
   /**
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/EnqueueTPUEmbeddingRaggedTensorBatch.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/EnqueueTPUEmbeddingRaggedTensorBatch.java
index bf4da86d05d..c4d1286e441 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/EnqueueTPUEmbeddingRaggedTensorBatch.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/EnqueueTPUEmbeddingRaggedTensorBatch.java
@@ -28,7 +28,6 @@
 import org.tensorflow.op.annotation.Operator;
 import org.tensorflow.types.TString;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
 
 /**
  * Eases the porting of code that uses tf.nn.embedding_lookup().
@@ -116,10 +115,10 @@ private Options() {
   @Endpoint(describeByClass = true)
   public static <T extends TNumber, U extends TNumber, V extends TNumber> EnqueueTPUEmbeddingRaggedTensorBatch create(Scope scope, Iterable<Operand<T>> sampleSplits, Iterable<Operand<U>> embeddingIndices, Iterable<Operand<V>> aggregationWeights, Operand<TString> modeOverride, List<Long> tableIds, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("EnqueueTPUEmbeddingRaggedTensorBatch", scope.makeOpName("EnqueueTPUEmbeddingRaggedTensorBatch"));
-    opBuilder.addInputList(Operands.asOutputs(sampleSplits));
-    opBuilder.addInputList(Operands.asOutputs(embeddingIndices));
-    opBuilder.addInputList(Operands.asOutputs(aggregationWeights));
-    opBuilder.addInput(modeOverride.asOutput());
+    opBuilder.addInputList(Operands.asOutputs(scope, sampleSplits));
+    opBuilder.addInputList(Operands.asOutputs(scope, embeddingIndices));
+    opBuilder.addInputList(Operands.asOutputs(scope, aggregationWeights));
+    opBuilder.addInput(modeOverride.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     long[] tableIdsArray = new long[tableIds.size()];
     for (int i = 0; i < tableIdsArray.length; ++i) {
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/InfeedDequeue.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/InfeedDequeue.java
index db236d3ccca..ae5e04f0276 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/InfeedDequeue.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/InfeedDequeue.java
@@ -17,7 +17,6 @@
 
 package org.tensorflow.op.tpu;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -45,7 +44,7 @@ public final class InfeedDequeue<T extends TType> extends RawOp implements Opera
    * @return a new instance of InfeedDequeue
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> InfeedDequeue<T> create(Scope scope, DataType<T> dtype, Shape shape) {
+  public static <T extends TType> InfeedDequeue<T> create(Scope scope, Class<T> dtype, Shape shape) {
     OperationBuilder opBuilder = scope.env().opBuilder("InfeedDequeue", scope.makeOpName("InfeedDequeue"));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("dtype", dtype);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/InfeedDequeueTuple.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/InfeedDequeueTuple.java
index f471da5b5d7..95f6b023f59 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/InfeedDequeueTuple.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/InfeedDequeueTuple.java
@@ -20,7 +20,6 @@
 import java.util.Arrays;
 import java.util.Iterator;
 import java.util.List;
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -46,10 +45,10 @@ public final class InfeedDequeueTuple extends RawOp implements Iterable<Operand<
    * @return a new instance of InfeedDequeueTuple
    */
   @Endpoint(describeByClass = true)
-  public static InfeedDequeueTuple create(Scope scope, List<DataType<?>> dtypes, List<Shape> shapes) {
+  public static InfeedDequeueTuple create(Scope scope, List<Class<?>> dtypes, List<Shape> shapes) {
     OperationBuilder opBuilder = scope.env().opBuilder("InfeedDequeueTuple", scope.makeOpName("InfeedDequeueTuple"));
     opBuilder = scope.applyControlDependencies(opBuilder);
-    DataType[] dtypesArray = new DataType[dtypes.size()];
+    Class[] dtypesArray = new Class[dtypes.size()];
     for (int i = 0; i < dtypesArray.length; ++i) {
       dtypesArray[i] = dtypes.get(i);
     }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingProximalYogiParameters.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingProximalYogiParameters.java
index 2a96916c4f5..233c64be04b 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingProximalYogiParameters.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingProximalYogiParameters.java
@@ -82,9 +82,9 @@ private Options() {
   @Endpoint(describeByClass = true)
   public static LoadTPUEmbeddingProximalYogiParameters create(Scope scope, Operand<TFloat32> parameters, Operand<TFloat32> v, Operand<TFloat32> m, Long numShards, Long shardId, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("LoadTPUEmbeddingProximalYogiParameters", scope.makeOpName("LoadTPUEmbeddingProximalYogiParameters"));
-    opBuilder.addInput(parameters.asOutput());
-    opBuilder.addInput(v.asOutput());
-    opBuilder.addInput(m.asOutput());
+    opBuilder.addInput(parameters.asOutput(scope));
+    opBuilder.addInput(v.asOutput(scope));
+    opBuilder.addInput(m.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("num_shards", numShards);
     opBuilder.setAttr("shard_id", shardId);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingProximalYogiParametersGradAccumDebug.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingProximalYogiParametersGradAccumDebug.java
index e863dc554d6..57160686c95 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingProximalYogiParametersGradAccumDebug.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingProximalYogiParametersGradAccumDebug.java
@@ -83,10 +83,10 @@ private Options() {
   @Endpoint(describeByClass = true)
   public static LoadTPUEmbeddingProximalYogiParametersGradAccumDebug create(Scope scope, Operand<TFloat32> parameters, Operand<TFloat32> v, Operand<TFloat32> m, Operand<TFloat32> gradientAccumulators, Long numShards, Long shardId, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("LoadTPUEmbeddingProximalYogiParametersGradAccumDebug", scope.makeOpName("LoadTPUEmbeddingProximalYogiParametersGradAccumDebug"));
-    opBuilder.addInput(parameters.asOutput());
-    opBuilder.addInput(v.asOutput());
-    opBuilder.addInput(m.asOutput());
-    opBuilder.addInput(gradientAccumulators.asOutput());
+    opBuilder.addInput(parameters.asOutput(scope));
+    opBuilder.addInput(v.asOutput(scope));
+    opBuilder.addInput(m.asOutput(scope));
+    opBuilder.addInput(gradientAccumulators.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("num_shards", numShards);
     opBuilder.setAttr("shard_id", shardId);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingStochasticGradientDescentParametersGradAccumDebug.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingStochasticGradientDescentParametersGradAccumDebug.java
index e408844e484..4f22a6b9996 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingStochasticGradientDescentParametersGradAccumDebug.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingStochasticGradientDescentParametersGradAccumDebug.java
@@ -88,8 +88,8 @@ private Options() {
   @Endpoint(describeByClass = true)
   public static LoadTPUEmbeddingStochasticGradientDescentParametersGradAccumDebug create(Scope scope, Operand<TFloat32> parameters, Operand<TFloat32> gradientAccumulators, Long numShards, Long shardId, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("LoadTPUEmbeddingStochasticGradientDescentParametersGradAccumDebug", scope.makeOpName("LoadTPUEmbeddingStochasticGradientDescentParametersGradAccumDebug"));
-    opBuilder.addInput(parameters.asOutput());
-    opBuilder.addInput(gradientAccumulators.asOutput());
+    opBuilder.addInput(parameters.asOutput(scope));
+    opBuilder.addInput(gradientAccumulators.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("num_shards", numShards);
     opBuilder.setAttr("shard_id", shardId);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/OutfeedDequeue.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/OutfeedDequeue.java
index fc4e30162d4..c61ed2d5bad 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/OutfeedDequeue.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/OutfeedDequeue.java
@@ -17,7 +17,6 @@
 
 package org.tensorflow.op.tpu;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -69,7 +68,7 @@ private Options() {
    * @return a new instance of OutfeedDequeue
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> OutfeedDequeue<T> create(Scope scope, DataType<T> dtype, Shape shape, Options... options) {
+  public static <T extends TType> OutfeedDequeue<T> create(Scope scope, Class<T> dtype, Shape shape, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("OutfeedDequeue", scope.makeOpName("OutfeedDequeue"));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("dtype", dtype);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/OutfeedDequeueTuple.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/OutfeedDequeueTuple.java
index 6b9110232d8..068423a6f80 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/OutfeedDequeueTuple.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/OutfeedDequeueTuple.java
@@ -20,7 +20,6 @@
 import java.util.Arrays;
 import java.util.Iterator;
 import java.util.List;
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -71,10 +70,10 @@ private Options() {
    * @return a new instance of OutfeedDequeueTuple
    */
   @Endpoint(describeByClass = true)
-  public static OutfeedDequeueTuple create(Scope scope, List<DataType<?>> dtypes, List<Shape> shapes, Options... options) {
+  public static OutfeedDequeueTuple create(Scope scope, List<Class<?>> dtypes, List<Shape> shapes, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("OutfeedDequeueTuple", scope.makeOpName("OutfeedDequeueTuple"));
     opBuilder = scope.applyControlDependencies(opBuilder);
-    DataType[] dtypesArray = new DataType[dtypes.size()];
+    Class[] dtypesArray = new Class[dtypes.size()];
     for (int i = 0; i < dtypesArray.length; ++i) {
       dtypesArray[i] = dtypes.get(i);
     }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/AccumulatorTakeGradient.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/AccumulatorTakeGradient.java
index e683c7c6496..70c01e2ada6 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/AccumulatorTakeGradient.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/AccumulatorTakeGradient.java
@@ -17,7 +17,6 @@
 
 package org.tensorflow.op.train;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -55,7 +54,7 @@ public final class AccumulatorTakeGradient<T extends TType> extends RawOp implem
    * @return a new instance of AccumulatorTakeGradient
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> AccumulatorTakeGradient<T> create(Scope scope, Operand<TString> handle, Operand<TInt32> numRequired, DataType<T> dtype) {
+  public static <T extends TType> AccumulatorTakeGradient<T> create(Scope scope, Operand<TString> handle, Operand<TInt32> numRequired, Class<T> dtype) {
     OperationBuilder opBuilder = scope.env().opBuilder("AccumulatorTakeGradient", scope.makeOpName("AccumulatorTakeGradient"));
     opBuilder.addInput(handle.asOutput(scope));
     opBuilder.addInput(numRequired.asOutput(scope));
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ConditionalAccumulator.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ConditionalAccumulator.java
index 9bd06537e1f..a3eeb3d0606 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ConditionalAccumulator.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ConditionalAccumulator.java
@@ -17,7 +17,6 @@
 
 package org.tensorflow.op.train;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -92,7 +91,7 @@ private Options() {
    * @return a new instance of ConditionalAccumulator
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> ConditionalAccumulator create(Scope scope, DataType<T> dtype, Shape shape, Options... options) {
+  public static <T extends TType> ConditionalAccumulator create(Scope scope, Class<T> dtype, Shape shape, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ConditionalAccumulator", scope.makeOpName("ConditionalAccumulator"));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("dtype", dtype);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceAccumulatorTakeGradient.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceAccumulatorTakeGradient.java
index 91314a0f00a..5a8f85ee16a 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceAccumulatorTakeGradient.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceAccumulatorTakeGradient.java
@@ -17,7 +17,6 @@
 
 package org.tensorflow.op.train;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -53,7 +52,7 @@ public final class ResourceAccumulatorTakeGradient<T extends TType> extends RawO
    * @return a new instance of ResourceAccumulatorTakeGradient
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> ResourceAccumulatorTakeGradient<T> create(Scope scope, Operand<?> handle, Operand<TInt32> numRequired, DataType<T> dtype) {
+  public static <T extends TType> ResourceAccumulatorTakeGradient<T> create(Scope scope, Operand<?> handle, Operand<TInt32> numRequired, Class<T> dtype) {
     OperationBuilder opBuilder = scope.env().opBuilder("ResourceAccumulatorTakeGradient", scope.makeOpName("ResourceAccumulatorTakeGradient"));
     opBuilder.addInput(handle.asOutput(scope));
     opBuilder.addInput(numRequired.asOutput(scope));
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceConditionalAccumulator.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceConditionalAccumulator.java
index 47bb1bb3027..a23d0c850e1 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceConditionalAccumulator.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceConditionalAccumulator.java
@@ -17,7 +17,6 @@
 
 package org.tensorflow.op.train;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -92,7 +91,7 @@ private Options() {
    * @return a new instance of ResourceConditionalAccumulator
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> ResourceConditionalAccumulator create(Scope scope, DataType<T> dtype, Shape shape, Options... options) {
+  public static <T extends TType> ResourceConditionalAccumulator create(Scope scope, Class<T> dtype, Shape shape, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("ResourceConditionalAccumulator", scope.makeOpName("ResourceConditionalAccumulator"));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("dtype", dtype);
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/Restore.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/Restore.java
index 6d613546764..b9d131d107b 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/Restore.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/Restore.java
@@ -20,7 +20,6 @@
 import java.util.Arrays;
 import java.util.Iterator;
 import java.util.List;
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -65,13 +64,13 @@ public final class Restore extends RawOp implements Iterable<Operand<TType>> {
    * @return a new instance of Restore
    */
   @Endpoint(describeByClass = true)
-  public static Restore create(Scope scope, Operand<TString> prefix, Operand<TString> tensorNames, Operand<TString> shapeAndSlices, List<DataType<?>> dtypes) {
+  public static Restore create(Scope scope, Operand<TString> prefix, Operand<TString> tensorNames, Operand<TString> shapeAndSlices, List<Class<?>> dtypes) {
     OperationBuilder opBuilder = scope.env().opBuilder("RestoreV2", scope.makeOpName("Restore"));
     opBuilder.addInput(prefix.asOutput(scope));
     opBuilder.addInput(tensorNames.asOutput(scope));
     opBuilder.addInput(shapeAndSlices.asOutput(scope));
     opBuilder = scope.applyControlDependencies(opBuilder);
-    DataType[] dtypesArray = new DataType[dtypes.size()];
+    Class[] dtypesArray = new Class[dtypes.size()];
     for (int i = 0; i < dtypesArray.length; ++i) {
       dtypesArray[i] = dtypes.get(i);
     }
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/RestoreSlice.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/RestoreSlice.java
index 4e1cfc35405..11614978792 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/RestoreSlice.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/RestoreSlice.java
@@ -17,7 +17,6 @@
 
 package org.tensorflow.op.train;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -79,7 +78,7 @@ private Options() {
    * @return a new instance of RestoreSlice
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> RestoreSlice<T> create(Scope scope, Operand<TString> filePattern, Operand<TString> tensorName, Operand<TString> shapeAndSlice, DataType<T> dt, Options... options) {
+  public static <T extends TType> RestoreSlice<T> create(Scope scope, Operand<TString> filePattern, Operand<TString> tensorName, Operand<TString> shapeAndSlice, Class<T> dt, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("RestoreSlice", scope.makeOpName("RestoreSlice"));
     opBuilder.addInput(filePattern.asOutput(scope));
     opBuilder.addInput(tensorName.asOutput(scope));
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Recv.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Recv.java
index 602e85a5b90..eaedb5aaeef 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Recv.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Recv.java
@@ -17,7 +17,6 @@
 
 package org.tensorflow.op.xla;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.OperationBuilder;
@@ -50,7 +49,7 @@ public final class Recv<T extends TType> extends RawOp implements Operand<T> {
    * @return a new instance of Recv
    */
   @Endpoint(describeByClass = true)
-  public static <T extends TType> Recv<T> create(Scope scope, DataType<T> dtype, String tensorName, Shape shape) {
+  public static <T extends TType> Recv<T> create(Scope scope, Class<T> dtype, String tensorName, Shape shape) {
     OperationBuilder opBuilder = scope.env().opBuilder("XlaRecv", scope.makeOpName("Recv"));
     opBuilder = scope.applyControlDependencies(opBuilder);
     opBuilder.setAttr("dtype", dtype);
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/AbstractOperation.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/AbstractOperation.java
index 96da6bc5ff4..bb83f21fe76 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/AbstractOperation.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/AbstractOperation.java
@@ -74,9 +74,9 @@ public String toString() {
    * Returns the datatype of the tensor of the {@code outputIdx}th output of this operation.
    *
    * @param outputIdx index of the output of this operation
-   * @return output tensor datatype
+   * @return datatype native code
    */
-  abstract DataType<?> dtype(int outputIdx);
+  abstract int dtype(int outputIdx);
 
   /**
    * Returns the tensor of the {@code outputIdx}th output of this operation.
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/AbstractTensor.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/AbstractTensor.java
index 71fd952bd6a..9c2fff4e2e3 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/AbstractTensor.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/AbstractTensor.java
@@ -15,17 +15,11 @@
 
 package org.tensorflow;
 
-import static org.tensorflow.internal.c_api.global.tensorflow.TF_Dim;
-import static org.tensorflow.internal.c_api.global.tensorflow.TF_NumDims;
 import static org.tensorflow.internal.c_api.global.tensorflow.TF_TensorByteSize;
-import static org.tensorflow.internal.c_api.global.tensorflow.TF_TensorType;
 
-import java.util.function.Consumer;
-import org.bytedeco.javacpp.PointerScope;
 import org.tensorflow.internal.c_api.TF_Tensor;
 import org.tensorflow.internal.tensor.buffer.TensorBuffers;
 import org.tensorflow.ndarray.NdArray;
-import org.tensorflow.ndarray.Shape;
 import org.tensorflow.ndarray.buffer.ByteDataBuffer;
 import org.tensorflow.types.family.TType;
 
@@ -37,28 +31,13 @@
  *
  * <p>Usage of this class is meant to be kept internal.
  */
-public abstract class AbstractDenseTensor<T> implements Tensor<T> {
-
-  @Override
-  @SuppressWarnings("unchecked")
-  public <U extends TType> U expect(DataType<U> dt) {
-    if (!dt.equals(this.dtype)) {
-      throw new IllegalArgumentException(
-          "Cannot cast from tensor of " + dtype + " to tensor of " + dt);
-    }
-    return (U)this;
-  }
+public abstract class AbstractTensor<T> implements Tensor<T> {
 
   @Override
   public void close() {
     nativeHandle().close();
   }
 
-  @Override
-  public DataType<?> dataType() {
-    return dtype;
-  }
-
   @Override
   public long numBytes() {
     return TF_TensorByteSize(nativeHandle());
@@ -71,12 +50,11 @@ public ByteDataBuffer rawData() {
 
   @Override
   public String toString() {
-    return String.format("%s tensor with shape %s", dtype.toString(), shape());
+    return String.format("%s tensor with shape %s", TensorTypes.find(type()).dataType(), shape());
   }
 
-  protected AbstractDenseTensor(TF_Tensor nativeHandle, DataType<?> dtype) {
+  protected AbstractTensor(TF_Tensor nativeHandle) {
     this.nativeHandle = nativeHandle;
-    this.dtype = dtype;
     nativeHandle.retainReference();
   }
 
@@ -100,6 +78,5 @@ protected static TF_Tensor requireHandle(TF_Tensor handle) {
     return handle;
   }
 
-  private final TF_Tensor nativeHandle;
-  private final DataType<?> dtype;
+  protected final TF_Tensor nativeHandle;
 }
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/DataType.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/DataType.java
index 70edafa391a..fe71905255e 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/DataType.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/DataType.java
@@ -15,90 +15,23 @@
 
 package org.tensorflow;
 
-import org.tensorflow.internal.c_api.TF_Tensor;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.types.family.TType;
-
 /** Represents a type of elements in a {@link Tensor} */
-public final class DataType<T extends TType> {
-
-  @FunctionalInterface
-  public interface TensorInstantiator<T> {
-
-    /**
-     * Maps tensor memory to a data structure for manipulating elements of this type.
-     *
-     * @param nativeTensor pointer to the native tensor
-     * @param shape the shape of the tensor
-     * @return data structure of elements of this type
-     */
-    T apply(TF_Tensor nativeTensor, Shape shape);
-  }
-
-  /**
-   * Creates a new datatype
-   *
-   * @param name readable-name for this type
-   * @param value must match the corresponding TF_* value in the TensorFlow C API.
-   * @param byteSize size of an element of this type, in bytes, -1 if unknown
-   * @param tensorMapper method for instantiating tensor from a native reference
-   */
-  public static <T extends TType> DataType<T> create(String name, int value, int byteSize, TensorInstantiator<T> instantiator) {
-    return new DataType<>(name, value, byteSize, instantiator);
-  }
-
-  /**
-   * Returns the size of an element of this type, in bytes, or -1 if element size is variable.
-   */
-  public int byteSize() {
-    return byteSize;
-  }
-
-  /**
-   * Returns true if this datatype has elements of variable length
-   */
-  public boolean isVariableLength() {
-    return byteSize == -1;
-  }
-
-  /**
-   * Returns a readable name for this type
-   */
-  public String name() {
-    return name;
-  }
-
-  @Override
-  public String toString() {
-    return name + " (" + nativeCode + ")";
-  }
-
-  /**
-   * Returns the numeric code for this datatype, as recognized by the native library (C API)
-   */
-  public int nativeCode() {
-    return nativeCode;
-  }
-
-  /**
-   * Instantiate a tensor of this datatype from the provided native handle
-   *
-   * @param handle tensor native handle
-   * @return a tensor of this datatype
-   */
-  T instantiateTensor(TF_Tensor handle, Shape shape) {
-    return tensorInstantiator.apply(handle, shape);
-  }
-
-  private final int nativeCode;
-  private final int byteSize;
-  private final String name;
-  private final TensorInstantiator<T> tensorInstantiator;
-
-  private DataType(String name, int nativeCode, int byteSize, TensorInstantiator<T> tensorInstantiator) {
-    this.name = name;
-    this.nativeCode = nativeCode;
+public enum DataType {
+  FLOAT(1, 4),
+  DOUBLE(2, 8),
+  INT32(3, 4),
+  UINT8(4, 1),
+  STRING(7, -1),
+  INT64(9, 8),
+  BOOL(10, 1),
+  BFLOAT16(14, 2),
+  HALF(19, 2);
+
+  final int number;
+  final int byteSize;
+
+  private DataType(int number, int byteSize) {
+    this.number = number;
     this.byteSize = byteSize;
-    this.tensorInstantiator = tensorInstantiator;
   }
 }
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/DataTypes.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/DataTypes.java
deleted file mode 100644
index 03d613475b5..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/DataTypes.java
+++ /dev/null
@@ -1,75 +0,0 @@
-/*
- *  Copyright 2019 The TensorFlow Authors. All Rights Reserved.
- *
- *  Licensed under the Apache License, Version 2.0 (the "License");
- *  you may not use this file except in compliance with the License.
- *  You may obtain a copy of the License at
- *
- *      http://www.apache.org/licenses/LICENSE-2.0
- *
- *  Unless required by applicable law or agreed to in writing, software
- *  distributed under the License is distributed on an "AS IS" BASIS,
- *  WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
- *  See the License for the specific language governing permissions and
- *  limitations under the License.
- *  =======================================================================
- */
-
-package org.tensorflow;
-
-import java.util.HashMap;
-import java.util.Map;
-import org.tensorflow.types.TBfloat16;
-import org.tensorflow.types.TBool;
-import org.tensorflow.types.TFloat16;
-import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.TFloat64;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.TString;
-import org.tensorflow.types.TUint8;
-
-/**
- * Utility class for working with {@link DataType} objects.
- */
-final class DataTypes {
-
-  /**
-   * Find a data type from the type code returned by the native layer (C API).
-   *
-   * <p>Only data types registered via {@link #register(DataType)} can be resolved.
-   *
-   * @param nativeCode native code
-   * @return data type for this code
-   * @throws IllegalArgumentException if the code matches no registered data type
-   */
-  static DataType<? extends Tensor> fromNativeCode(int nativeCode) {
-    DataType<? extends Tensor> dataType = DATA_TYPE_REGISTRY.get(nativeCode);
-    if (dataType == null) {
-      throw new IllegalArgumentException(
-          "DataType " + nativeCode + " is not recognized in Java (version " + TensorFlow.version() + ")");
-    }
-    return dataType;
-  }
-
-  private static final Map<Integer, DataType<? extends Tensor>> DATA_TYPE_REGISTRY = new HashMap<>();
-
-  static {
-    register(TBool.DTYPE);
-    register(TFloat64.DTYPE);
-    register(TFloat32.DTYPE);
-    register(TFloat16.DTYPE);
-    register(TInt32.DTYPE);
-    register(TInt64.DTYPE);
-    register(TString.DTYPE);
-    register(TUint8.DTYPE);
-    register(TBfloat16.DTYPE);
-  }
-
-  // TODO (karllessard): Right now this method is private but we might want to expose it
-  //      to allow user to register custom data types?
-  private static void register(DataType<? extends Tensor> dataType) {
-    DATA_TYPE_REGISTRY.put(dataType.nativeCode(), dataType);
-    DATA_TYPE_REGISTRY.put(dataType.nativeCode() + 100, dataType);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/EagerOperation.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/EagerOperation.java
index cc0d93d6bac..65f7ad443f1 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/EagerOperation.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/EagerOperation.java
@@ -29,6 +29,7 @@
 import org.tensorflow.internal.c_api.TF_Status;
 import org.tensorflow.internal.c_api.TF_Tensor;
 import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.Ops;
 
 /**
  * Implementation of an {@link Operation} executed eagerly.
@@ -104,15 +105,9 @@ public Shape shape(int outputIndex) {
   }
 
   @Override
-  public DataType<?> dtype(int outputIndex) {
-    // If the tensor of this output has already been resolved, return its datatype.
-    // Otherwise, retrieve the tensor datatype from the native library.
-    Tensor<?> tensor = outputTensors.get(outputIndex);
-    if (tensor != null) {
-      return tensor.dataType();
-    }
+  public int dtype(int outputIndex) {
     TFE_TensorHandle outputNativeHandle = getUnsafeNativeHandle(outputIndex);
-    return DataTypes.fromNativeCode(dataType(outputNativeHandle));
+    return dataType(outputNativeHandle);
   }
 
   @Override
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/EagerOperationBuilder.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/EagerOperationBuilder.java
index 8bebb554e57..807585c4d17 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/EagerOperationBuilder.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/EagerOperationBuilder.java
@@ -48,6 +48,7 @@
 import org.tensorflow.internal.c_api.TF_Status;
 import org.tensorflow.internal.c_api.TF_Tensor;
 import org.tensorflow.ndarray.Shape;
+import org.tensorflow.types.family.TType;
 
 /**
  * An {@link OperationBuilder} for building {@link Operation Operations} that are executed eagerly.
@@ -159,16 +160,16 @@ public EagerOperationBuilder setAttr(String name, boolean[] values) {
   }
 
   @Override
-  public EagerOperationBuilder setAttr(String name, DataType<?> value) {
-    setAttrType(opHandle, name, value.nativeCode());
+  public EagerOperationBuilder setAttr(String name, Class<? extends TType> type) {
+    setAttrType(opHandle, name, TensorTypes.numberOf(type));
     return this;
   }
 
   @Override
-  public EagerOperationBuilder setAttr(String name, DataType<?>[] values) {
-    int[] c = new int[values.length];
-    for (int i = 0; i < values.length; ++i) {
-      c[i] = values[i].nativeCode();
+  public EagerOperationBuilder setAttr(String name, Class<? extends TType>[] types) {
+    int[] c = new int[types.length];
+    for (int i = 0; i < types.length; ++i) {
+      c[i] = TensorTypes.numberOf(types[i]);
     }
     setAttrTypeList(opHandle, name, c);
     return this;
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Graph.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Graph.java
index 142d481c04f..e9551f6c9b6 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Graph.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Graph.java
@@ -767,14 +767,14 @@ private static SaverDef addVariableSaver(Graph graph) {
 
     List<String> varNames = new ArrayList<>();
     List<Operand<?>> varOutputs = new ArrayList<>();
-    List<DataType<?>> varTypes = new ArrayList<>();
+    List<Class<?>> varTypes = new ArrayList<>();
 
     for (Iterator<Operation> iter = graph.operations(); iter.hasNext();) {
       Operation op = iter.next();
       if (op.type().equals("VariableV2")) {
         varNames.add(op.name());
         varOutputs.add(op.output(0));
-        varTypes.add(op.output(0).dataType());
+        varTypes.add(op.output(0).type());
       }
     }
 
@@ -783,7 +783,7 @@ private static SaverDef addVariableSaver(Graph graph) {
     Constant<TString> varNamesTensor = tf.constant(StdArrays.ndCopyOf(varNames.toArray(tmp)));
     Operand<TString> varSlices = tf.zerosLike(varNamesTensor);
 
-    Placeholder<TString> saveFilename = tf.placeholder(TString.DTYPE);
+    Placeholder<TString> saveFilename = tf.placeholder(TString.class);
     Save saveVariables = tf.train.save(
         saveFilename,
         varNamesTensor,
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/GraphOperation.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/GraphOperation.java
index 70cd31366ce..8757bec651b 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/GraphOperation.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/GraphOperation.java
@@ -148,10 +148,10 @@ Shape shape(int outputIdx) {
   }
 
   @Override
-  DataType<?> dtype(int outputIdx) {
+  int dtype(int outputIdx) {
     Graph.Reference r = graph.ref();
     try {
-      return DataTypes.fromNativeCode(dtype(r.nativeHandle(), getUnsafeNativeHandle(), outputIdx));
+      return dtype(r.nativeHandle(), getUnsafeNativeHandle(), outputIdx);
     } finally {
       r.close();
     }
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/GraphOperationBuilder.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/GraphOperationBuilder.java
index e57c997b4b0..ab68d809a4e 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/GraphOperationBuilder.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/GraphOperationBuilder.java
@@ -52,6 +52,7 @@
 import org.tensorflow.internal.c_api.TF_Status;
 import org.tensorflow.internal.c_api.TF_Tensor;
 import org.tensorflow.ndarray.Shape;
+import org.tensorflow.types.family.TType;
 
 /** An {@link OperationBuilder} for adding {@link GraphOperation}s to a {@link Graph}. */
 public final class GraphOperationBuilder implements OperationBuilder {
@@ -221,10 +222,10 @@ public GraphOperationBuilder setAttr(String name, boolean[] value) {
   }
 
   @Override
-  public GraphOperationBuilder setAttr(String name, DataType value) {
+  public GraphOperationBuilder setAttr(String name, Class<? extends TType> type) {
     Graph.Reference r = graph.ref();
     try {
-      setAttrType(unsafeNativeHandle, name, value.nativeCode());
+      setAttrType(unsafeNativeHandle, name, TensorTypes.numberOf(type));
     } finally {
       r.close();
     }
@@ -232,10 +233,10 @@ public GraphOperationBuilder setAttr(String name, DataType value) {
   }
 
   @Override
-  public GraphOperationBuilder setAttr(String name, DataType[] value) {
-    int[] ctypes = new int[value.length];
-    for (int i = 0; i < value.length; ++i) {
-      ctypes[i] = value[i].nativeCode();
+  public GraphOperationBuilder setAttr(String name, Class<? extends TType>[] types) {
+    int[] ctypes = new int[types.length];
+    for (int i = 0; i < types.length; ++i) {
+      ctypes[i] = TensorTypes.numberOf(types[i]);
     }
     Graph.Reference r = graph.ref();
     try {
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Operand.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Operand.java
index d80ce85dab2..18a2385f492 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Operand.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Operand.java
@@ -81,18 +81,4 @@ default Output<T> asOutput() {
   default T asTensor() {
     return asOutput().tensor();
   }
-
-  /**
-   * Returns this operand as a tensor of the given type.
-   *
-   * <i>Only works when running in an eager execution</i>
-   * <p>This helper method is equivalent to {@code asOutput().tensor()}
-   *
-   * @return the tensor
-   * @throws IllegalStateException if this is an operand of a graph
-   * @throws IllegalArgumentException if the {@code dataType} is incompatible with tensor type
-   */
-  default <U extends TType> U asTensor(DataType<U> dataType) {
-    return (U)asOutput().tensor().expect(dataType);
-  }
 }
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/OperationBuilder.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/OperationBuilder.java
index af1b8cc9130..89a18741846 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/OperationBuilder.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/OperationBuilder.java
@@ -16,6 +16,7 @@
 package org.tensorflow;
 
 import org.tensorflow.ndarray.Shape;
+import org.tensorflow.types.family.TType;
 
 /**
  * A builder for {@link Operation}s.
@@ -177,7 +178,7 @@ public interface OperationBuilder {
    * @param value attribute value
    * @return the OperationBuilder instance for chaining.
    */
-   OperationBuilder setAttr(String name, DataType<?> value);
+   OperationBuilder setAttr(String name, Class<? extends TType> value);
 
   /**
    * Set the type values of an attribute of the operation being built.
@@ -186,7 +187,7 @@ public interface OperationBuilder {
    * @param value attribute values
    * @return the OperationBuilder instance for chaining.
    */
-   OperationBuilder setAttr(String name, DataType<?>[] value);
+   OperationBuilder setAttr(String name, Class<? extends TType>[] value);
 
   /**
    * Set the tensor value of an attribute of the operation being built.
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Output.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Output.java
index 55979a5f6da..5c1c24bc0f4 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Output.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Output.java
@@ -43,24 +43,23 @@ public Shape shape() {
   }
 
   /** Returns the DataType of the tensor referred to by this Output. */
-  @SuppressWarnings("unchecked")
-  public DataType<T> dataType() {
-    return (DataType<T>)operation.dtype(index);
+  public Class<T> type() {
+    return (Class)TensorTypes.find(operation.dtype(index)).typeClass();
   }
 
   /**
    * Returns this Output object with the type {@code Output<U>}. This method is useful when given a
    * value of type {@code Output<?>}.
    *
-   * @param dt any supported tensor data type
+   * @param tensorType type of tensor at this output
    * @throws IllegalArgumentException if the actual data type of this object does not match the type
    *     {@code U}.
    */
   @SuppressWarnings("unchecked")
-  public <U extends TType> Output<U> expect(DataType<U> dt) {
-    if (!dt.equals(this.dataType())) {
+  public <U extends TType> Output<U> expect(Class<U> tensorType) {
+    if (tensorType == type()) {
       throw new IllegalArgumentException(
-          "Cannot cast from output of " + this.dataType() + " to output of " + dt);
+          "Cannot cast from output of " + type().getSimpleName() + " to output of type " + tensorType.getSimpleName());
     }
     return ((Output<U>) this);
   }
@@ -116,7 +115,7 @@ public boolean equals(Object o) {
   public String toString() {
     return String.format(
         "<%s '%s:%d' shape=%s dtype=%s>",
-        operation.type(), operation.name(), index, shape().toString(), dataType());
+        operation.type(), operation.name(), index, shape().toString(), type().getSimpleName());
   }
 
   /** Handle to the idx-th output of the Operation {@code op}. */
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Signature.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Signature.java
index 376dc9039fc..66bccbb049e 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Signature.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Signature.java
@@ -113,7 +113,7 @@ private static TensorInfo toTensorInfo(Output<?> operand) {
         tensorShapeBuilder.addDim(Dim.newBuilder().setSize(shape.size(i)));
       }
       return TensorInfo.newBuilder()
-          .setDtype(DataType.forNumber(operand.dataType().nativeCode()))
+          .setDtype(DataType.forNumber(TensorTypes.numberOf(operand.type())))
           .setTensorShape(tensorShapeBuilder)
           .setName(operand.op().name() + ":" + operand.index())
           .build();
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Tensor.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Tensor.java
index feb308a513e..1344ec36db7 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Tensor.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Tensor.java
@@ -15,7 +15,6 @@
 
 package org.tensorflow;
 
-import org.checkerframework.checker.units.qual.C;
 import org.tensorflow.ndarray.NdArray;
 import org.tensorflow.ndarray.buffer.ByteDataBuffer;
 import org.tensorflow.ndarray.buffer.DataBuffer;
@@ -38,18 +37,6 @@
  */
 public interface Tensor<T> extends NdArray<T>, AutoCloseable {
 
-  /**
-   * Returns this Tensor object with the type {@code Tensor<U>}. This method is useful when given a
-   * value of type {@code Tensor<?>}.
-   *
-   * @param dt any supported tensor data type
-   * @param <U> a tensor type
-   * @return a tensor of the requested data type
-   * @throws IllegalArgumentException if the actual data type of this object does not match the type
-   *     {@code U}.
-   */
-  <U extends TType> U expect(DataType<U> dt);
-
   /**
    * Release resources associated with the Tensor.
    *
@@ -61,9 +48,6 @@ public interface Tensor<T> extends NdArray<T>, AutoCloseable {
   @Override
   void close();
 
-  /** Returns the {@link DataType} of elements stored in the Tensor. */
-  DataType<?> dataType();
-
   /** Returns the size, in bytes, of the tensor data. */
   long numBytes();
 
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/TensorType.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/TensorType.java
new file mode 100644
index 00000000000..c8e270cb2c8
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/TensorType.java
@@ -0,0 +1,40 @@
+package org.tensorflow;
+
+import java.lang.reflect.Constructor;
+import org.tensorflow.exceptions.TensorFlowException;
+import org.tensorflow.internal.c_api.TF_Tensor;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.types.family.TType;
+
+class TensorType {
+
+  Class<?> typeClass() {
+    return typeClass;
+  }
+
+  DataType dataType() {
+    return dataType;
+  }
+
+  boolean isVariableLength() {
+    return dataType.byteSize < 0;
+  }
+
+  <T extends TType> T newInstance(TF_Tensor nativeHandle, Shape shape) {
+    try {
+      return (T) implConstructor.newInstance(nativeHandle, shape);
+    } catch (ReflectiveOperationException e) {
+      throw new TensorFlowException("Fail to instantiate tensor: " + e.getMessage(), e);
+    }
+  }
+
+  TensorType(Class<?> typeClass, DataType dataType, Constructor<?> implConstructor) {
+    this.typeClass = typeClass;
+    this.dataType = dataType;
+    this.implConstructor = implConstructor;
+  }
+
+  private final Class<?> typeClass;
+  private final DataType dataType;
+  private final Constructor<?> implConstructor;
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/TensorTypes.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/TensorTypes.java
new file mode 100644
index 00000000000..2d0acb16a50
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/TensorTypes.java
@@ -0,0 +1,114 @@
+/*
+ *  Copyright 2019 The TensorFlow Authors. All Rights Reserved.
+ *
+ *  Licensed under the Apache License, Version 2.0 (the "License");
+ *  you may not use this file except in compliance with the License.
+ *  You may obtain a copy of the License at
+ *
+ *      http://www.apache.org/licenses/LICENSE-2.0
+ *
+ *  Unless required by applicable law or agreed to in writing, software
+ *  distributed under the License is distributed on an "AS IS" BASIS,
+ *  WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ *  See the License for the specific language governing permissions and
+ *  limitations under the License.
+ *  =======================================================================
+ */
+
+package org.tensorflow;
+
+import java.lang.reflect.Constructor;
+import java.util.HashMap;
+import java.util.Map;
+import org.tensorflow.internal.c_api.TF_Tensor;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.types.TBfloat16;
+import org.tensorflow.types.TBool;
+import org.tensorflow.types.TFloat16;
+import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.TFloat64;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.TString;
+import org.tensorflow.types.TUint8;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Utility class for working with {@link DataType} objects.
+ */
+final class TensorTypes {
+
+  /**
+   * Find a type registration from the code returned by the native layer (C API).
+   *
+   * @param nativeCode native type code
+   * @return type registered to this code
+   * @throws IllegalArgumentException if the code matches no registered data type
+   */
+  static TensorType find(int nativeCode) {
+    TensorType entry = TYPES_BY_CODE.get(nativeCode);
+    if (entry == null) {
+      throw new IllegalArgumentException(
+          "DataType " + nativeCode + " is not recognized in Java (version " + TensorFlow.version() + ")");
+    }
+    return entry;
+  }
+
+  /**
+   * Find a type registration.
+   *
+   * @param typeClass class implementing {@link TType}
+   * @return type registration
+   * @throws IllegalArgumentException if the code matches no registered data type
+   */
+  static TensorType find(Class<? extends TType> typeClass) {
+    TensorType entry = TYPES_BY_CLASS.get(typeClass);
+    if (entry == null) {
+      throw new IllegalArgumentException("Class \"" + typeClass.getName() + "\" is not a valid datatype class");
+    }
+    return entry;
+  }
+
+  static int numberOf(Class<? extends TType> typeClass) {
+    return find(typeClass).dataType().number;
+  }
+
+  private static final Map<Integer, TensorType> TYPES_BY_CODE = new HashMap<>();
+  private static final Map<Class<? extends TType>, TensorType> TYPES_BY_CLASS = new HashMap<>();
+
+  private static void register(Class<? extends TType> typeClass) {
+    org.tensorflow.types.annotation.TensorType typeAnnot =
+        typeClass.getDeclaredAnnotation(org.tensorflow.types.annotation.TensorType.class);
+    if (typeAnnot == null) {
+      throw new IllegalArgumentException("Class \"" + typeClass.getName() + "\" must be annotated "
+          + "with @TensorType to be registered as a tensor type");
+    }
+    Constructor<?> implConstructor;
+    try {
+      implConstructor = typeAnnot.impl().getConstructor(TF_Tensor.class, Shape.class);
+    } catch (NoSuchMethodException e) {
+      throw new IllegalArgumentException("Class \"" + typeClass.getName() + "\" must have a constructor "
+          + "accepting a native `TF_Tensor` handle and a `Shape` to be implement as a tensor type");
+    }
+    TensorType type = new TensorType(typeClass, typeAnnot.dataType(), implConstructor);
+    TYPES_BY_CLASS.put(typeClass, type);
+
+    // If more than one tensor type is mapped to a given native code, the last registered will
+    // have priority. This way, we can allow user to register their own classes to map tensors
+    // of a given data type.
+    TYPES_BY_CODE.put(type.dataType().number, type);
+    TYPES_BY_CODE.put(type.dataType().number + 100, type);
+  }
+
+  static {
+    register(TBool.class);
+    register(TFloat64.class);
+    register(TFloat32.class);
+    register(TFloat16.class);
+    register(TInt32.class);
+    register(TInt64.class);
+    register(TString.class);
+    register(TUint8.class);
+    register(TBfloat16.class);
+  }
+}
\ No newline at end of file
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Tensors.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Tensors.java
index 3df1df81ddf..bc7468720c6 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Tensors.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Tensors.java
@@ -31,13 +31,13 @@ public final class Tensors {
    * }</pre>
    *
    * @param <T> the tensor element type
-   * @param dtype datatype of the tensor
+   * @param type tensor type
    * @param shape shape of the tensor
    * @return an allocated but uninitialized tensor
    * @throws IllegalStateException if tensor failed to be allocated
    */
-  public static <T extends TType> T of(DataType<T> dtype, Shape shape) {
-    return of(dtype, shape, shape.size() * dtype.byteSize());
+  public static <T extends TType> T of(Class<T> type, Shape shape) {
+    return of(type, shape, -1);
   }
 
   /**
@@ -50,7 +50,7 @@ public static <T extends TType> T of(DataType<T> dtype, Shape shape) {
    * like {@link org.tensorflow.types.TString TString}.
    *
    * @param <T> the tensor element type
-   * @param dtype datatype of the tensor
+   * @param type tensor type
    * @param shape shape of the tensor
    * @param size size, in bytes, of the tensor
    * @return an allocated but uninitialized tensor
@@ -59,8 +59,8 @@ public static <T extends TType> T of(DataType<T> dtype, Shape shape) {
    *                                  store the tensor data
    * @throws IllegalStateException if tensor failed to be allocated
    */
-  public static <T extends TType> T of(DataType<T> dtype, Shape shape, long size) {
-    return Tensors.allocate(dtype, shape, size);
+  public static <T extends TType> T of(Class<T> type, Shape shape, long size) {
+    return allocate(type, shape, size);
   }
 
   /**
@@ -81,14 +81,14 @@ public static <T extends TType> T of(DataType<T> dtype, Shape shape, long size)
    * automatically released before rethrowing the same exception.
    *
    * @param <T> the tensor element type
-   * @param dtype datatype of the tensor
+   * @param type tensor type
    * @param shape shape of the tensor
    * @param dataInitializer method receiving accessor to the allocated tensor data for initialization
    * @return an allocated and initialized tensor
    * @throws IllegalStateException if tensor failed to be allocated
    */
-  public static <T extends TType> T of(DataType<T> dtype, Shape shape, Consumer<T> dataInitializer) {
-    return of(dtype, shape, shape.size() * dtype.byteSize(), dataInitializer);
+  public static <T extends TType> T of(Class<T> type, Shape shape, Consumer<T> dataInitializer) {
+    return of(type, shape, -1, dataInitializer);
   }
 
   /**
@@ -101,20 +101,20 @@ public static <T extends TType> T of(DataType<T> dtype, Shape shape, Consumer<T>
    * such as {@link org.tensorflow.types.TString TString}.
    *
    * @param <T> the tensor element type
-   * @param dtype datatype of the tensor
+   * @param type tensor type
    * @param shape shape of the tensor
    * @param size size, in bytes, of the tensor
-   * @param tensorInit method receiving accessor to the allocated tensor data for initialization
+   * @param dataInitializer method receiving accessor to the allocated tensor data for initialization
    * @return an allocated and initialized tensor
    * @see #of(DataType, Shape, long, Consumer)
    * @throws IllegalArgumentException if {@code size} is smaller than the minimum space required to
    *                                  store the tensor data
    * @throws IllegalStateException if tensor failed to be allocated
    */
-  public static <T extends TType> T of(DataType<T> dtype, Shape shape, long size, Consumer<T> tensorInit) {
-    T tensor = of(dtype, shape, size);
+  public static <T extends TType> T of(Class<T> type, Shape shape, long size, Consumer<T> dataInitializer) {
+    T tensor = of(type, shape, size);
     try {
-      tensorInit.accept(tensor);
+      dataInitializer.accept(tensor);
       return tensor;
     } catch (Throwable t) {
       tensor.close();
@@ -129,40 +129,48 @@ public static <T extends TType> T of(DataType<T> dtype, Shape shape, long size,
    * href="https://www.tensorflow.org/code/tensorflow/c/c_api.h">C API</a>.
    *
    * @param <T> the tensor element type
+   * @param type tensor type
    * @param dtype the tensor element data type
    * @param shape the tensor shape.
    * @param rawData a buffer containing the tensor raw data.
    * @throws IllegalArgumentException if {@code rawData} is not large enough to contain the tensor data
    * @throws IllegalStateException if tensor failed to be allocated with the given parameters
    */
-  public static <T extends TType> T of(DataType<T> dtype, Shape shape, ByteDataBuffer rawData) {
-    T tensor = of(dtype, shape, rawData.size());
+  public static <T extends TType> T of(Class<T> type, Shape shape, ByteDataBuffer rawData) {
+    T tensor = of(type, shape, rawData.size());
     rawData.copyTo(TensorBuffers.toBytes(((AbstractTensor)tensor).nativeHandle()), rawData.size());
     return tensor;
   }
 
-  static <T extends TType> T allocate(DataType<T> dataType, Shape shape, long size) {
-    // Minimum requirements for datatypes of variable length cannot be verified in a relevant way so
-    // we only validate them for fixed length datatypes
-    if (!dataType.isVariableLength() && shape.size() * dataType.byteSize() > size) {
-      throw new IllegalArgumentException("Tensor size is not large enough to contain all scalar values");
-    }
-    TF_Tensor nativeHandle = allocate(dataType.nativeCode(), shape.asArray(), size);
-    try (PointerScope scope = new PointerScope()) {
-      scope.attach(nativeHandle);
-      return dataType.instantiateTensor(nativeHandle, shape);
-    }
-  }
-
   /**
    * Create a Tensor object from a handle to the C TF_Tensor object.
    *
    * <p>Takes ownership of the handle.
    */
   static Tensor<?> fromHandle(TF_Tensor handle) {
-    DataType<?> dataType = DataTypes.fromNativeCode(dtype(handle));
+    TensorType type = TensorTypes.find(dtype(handle));
     Shape shape = Shape.of(shape(handle));
-    return dataType.instantiateTensor(handle, shape);
+    return type.newInstance(handle, shape);
+  }
+
+  private static <T extends TType> T allocate(Class<T> type, Shape shape, long size) {
+    TensorType tensorType = TensorTypes.find(type);
+    DataType dataType = tensorType.dataType();
+    long effectiveSize = size;
+    if (effectiveSize < 0) {
+      // Size of the tensor is by default the sum of the size of all its element
+      effectiveSize = shape.size() * dataType.byteSize;
+
+    } else if (dataType.byteSize > 0 && shape.size() * dataType.byteSize > effectiveSize) {
+      // Minimum requirements for datatypes of variable length cannot be verified in a relevant way
+      // so we only validate them for fixed length datatypes
+      throw new IllegalArgumentException("Tensor size is not large enough to contain all scalar values");
+    }
+    TF_Tensor nativeHandle = allocate(dataType.number, shape.asArray(), effectiveSize);
+    try (PointerScope scope = new PointerScope()) {
+      scope.attach(nativeHandle);
+      return tensorType.newInstance(nativeHandle, shape);
+    }
   }
 
   private static TF_Tensor requireHandle(TF_Tensor handle) {
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/BooleanTensorImpl.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/BooleanTensorImpl.java
index 9c7c059691d..dd1eb17fe57 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/BooleanTensorImpl.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/BooleanTensorImpl.java
@@ -12,15 +12,14 @@
 import org.tensorflow.ndarray.buffer.DataBuffer;
 import org.tensorflow.ndarray.buffer.BooleanDataBuffer;
 import org.tensorflow.ndarray.index.Index;
-import org.tensorflow.tensor.BooleanTensor;
+import org.tensorflow.types.tensor.BooleanTensor;
 
 public class BooleanTensorImpl extends AbstractTensor<Boolean> implements BooleanTensor {
 
   private final BooleanNdArray data;
 
-  public BooleanTensorImpl(TF_Tensor nativeHandle, DataType<?> dataType, Shape shape,
-      BooleanDataBuffer buffer) {
-    super(nativeHandle, dataType);
+  public BooleanTensorImpl(TF_Tensor nativeHandle, Shape shape, BooleanDataBuffer buffer) {
+    super(nativeHandle);
     data = NdArrays.wrap(shape, buffer);
   }
 
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/ByteTensorImpl.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/ByteTensorImpl.java
index 18d993219c6..03464538937 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/ByteTensorImpl.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/ByteTensorImpl.java
@@ -12,15 +12,14 @@
 import org.tensorflow.ndarray.buffer.DataBuffer;
 import org.tensorflow.ndarray.buffer.ByteDataBuffer;
 import org.tensorflow.ndarray.index.Index;
-import org.tensorflow.tensor.ByteTensor;
+import org.tensorflow.types.tensor.ByteTensor;
 
 public class ByteTensorImpl extends AbstractTensor<Byte> implements ByteTensor {
 
   private final ByteNdArray data;
 
-  public ByteTensorImpl(TF_Tensor nativeHandle, DataType<?> dataType, Shape shape,
-      ByteDataBuffer buffer) {
-    super(nativeHandle, dataType);
+  public ByteTensorImpl(TF_Tensor nativeHandle, Shape shape, ByteDataBuffer buffer) {
+    super(nativeHandle);
     data = NdArrays.wrap(shape, buffer);
   }
 
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/DoubleTensorImpl.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/DoubleTensorImpl.java
index 7f85d86c372..d6f95e7e420 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/DoubleTensorImpl.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/DoubleTensorImpl.java
@@ -12,15 +12,14 @@
 import org.tensorflow.ndarray.buffer.DataBuffer;
 import org.tensorflow.ndarray.buffer.DoubleDataBuffer;
 import org.tensorflow.ndarray.index.Index;
-import org.tensorflow.tensor.DoubleTensor;
+import org.tensorflow.types.tensor.DoubleTensor;
 
 public class DoubleTensorImpl extends AbstractTensor<Double> implements DoubleTensor {
 
   private final DoubleNdArray data;
 
-  public DoubleTensorImpl(TF_Tensor nativeHandle, DataType<?> dataType, Shape shape,
-      DoubleDataBuffer buffer) {
-    super(nativeHandle, dataType);
+  public DoubleTensorImpl(TF_Tensor nativeHandle, Shape shape, DoubleDataBuffer buffer) {
+    super(nativeHandle);
     data = NdArrays.wrap(shape, buffer);
   }
 
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/FloatTensorImpl.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/FloatTensorImpl.java
index e2d42bba37c..0e98fd74ec6 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/FloatTensorImpl.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/FloatTensorImpl.java
@@ -11,15 +11,14 @@
 import org.tensorflow.ndarray.buffer.DataBuffer;
 import org.tensorflow.ndarray.buffer.FloatDataBuffer;
 import org.tensorflow.ndarray.index.Index;
-import org.tensorflow.tensor.FloatTensor;
+import org.tensorflow.types.tensor.FloatTensor;
 
 public class FloatTensorImpl extends AbstractTensor<Float> implements FloatTensor {
 
   private final FloatNdArray data;
 
-  public FloatTensorImpl(TF_Tensor nativeHandle, DataType<?> dataType, Shape shape,
-      FloatDataBuffer buffer) {
-    super(nativeHandle, dataType);
+  public FloatTensorImpl(TF_Tensor nativeHandle, Shape shape, FloatDataBuffer buffer) {
+    super(nativeHandle);
     data = NdArrays.wrap(shape, buffer);
   }
 
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/IntTensorImpl.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/IntTensorImpl.java
index bff637e38ab..c4f93ec07db 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/IntTensorImpl.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/IntTensorImpl.java
@@ -12,15 +12,14 @@
 import org.tensorflow.ndarray.buffer.DataBuffer;
 import org.tensorflow.ndarray.buffer.IntDataBuffer;
 import org.tensorflow.ndarray.index.Index;
-import org.tensorflow.tensor.IntTensor;
+import org.tensorflow.types.tensor.IntTensor;
 
 public class IntTensorImpl extends AbstractTensor<Integer> implements IntTensor {
 
   private final IntNdArray data;
 
-  public IntTensorImpl(TF_Tensor nativeHandle, DataType<?> dataType, Shape shape,
-      IntDataBuffer buffer) {
-    super(nativeHandle, dataType);
+  public IntTensorImpl(TF_Tensor nativeHandle, Shape shape, IntDataBuffer buffer) {
+    super(nativeHandle);
     data = NdArrays.wrap(shape, buffer);
   }
 
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/LongTensorImpl.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/LongTensorImpl.java
index 8b3436b4e77..b91435eeef3 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/LongTensorImpl.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/LongTensorImpl.java
@@ -12,15 +12,14 @@
 import org.tensorflow.ndarray.buffer.DataBuffer;
 import org.tensorflow.ndarray.buffer.LongDataBuffer;
 import org.tensorflow.ndarray.index.Index;
-import org.tensorflow.tensor.LongTensor;
+import org.tensorflow.types.tensor.LongTensor;
 
 public class LongTensorImpl extends AbstractTensor<Long> implements LongTensor {
 
   private final LongNdArray data;
 
-  public LongTensorImpl(TF_Tensor nativeHandle, DataType<?> dataType, Shape shape,
-      LongDataBuffer buffer) {
-    super(nativeHandle, dataType);
+  public LongTensorImpl(TF_Tensor nativeHandle, Shape shape, LongDataBuffer buffer) {
+    super(nativeHandle);
     data = NdArrays.wrap(shape, buffer);
   }
 
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/StringTensorImpl.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/StringTensorImpl.java
index 4a1cd3333e5..62edee7cf85 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/StringTensorImpl.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/StringTensorImpl.java
@@ -12,7 +12,7 @@
 import org.tensorflow.ndarray.buffer.DataBuffer;
 import org.tensorflow.ndarray.buffer.layout.DataLayout;
 import org.tensorflow.ndarray.index.Index;
-import org.tensorflow.tensor.StringTensor;
+import org.tensorflow.types.tensor.StringTensor;
 
 public class StringTensorImpl extends AbstractTensor<String> implements StringTensor {
 
@@ -21,12 +21,11 @@ public class StringTensorImpl extends AbstractTensor<String> implements StringTe
 
   public StringTensorImpl(
       TF_Tensor nativeHandle,
-      DataType<?> dataType,
       Shape shape,
       DataLayout<DataBuffer<byte[]>, String> layout,
       ByteSequenceTensorBuffer buffer
   ) {
-    super(nativeHandle, dataType);
+    super(nativeHandle);
     this.rawBuffer = buffer;
     data = NdArrays.wrap(shape, layout.applyTo(buffer));
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/op/core/Constant.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/op/core/Constant.java
index e7ce5158ff0..176c57cbe6b 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/op/core/Constant.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/op/core/Constant.java
@@ -43,7 +43,6 @@
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.tensor.BooleanTensor;
 import org.tensorflow.types.TBool;
 import org.tensorflow.types.TFloat32;
 import org.tensorflow.types.TFloat64;
@@ -998,7 +997,7 @@ public static Constant<TUint8> tensorOf(Scope scope, Shape shape, ByteDataBuffer
    * Create a constant with data from the given buffer.
    *
    * @param scope is a scope used to add the underlying operation.
-   * @param type the tensor datatype.
+   * @param type the tensor type.
    * @param shape the tensor shape.
    * @param data a buffer containing the tensor data.
    * @return a constant of type `type`
@@ -1006,7 +1005,7 @@ public static Constant<TUint8> tensorOf(Scope scope, Shape shape, ByteDataBuffer
    *     buffer
    */
   @Endpoint
-  public static <T extends TType> Constant<T> tensorOf(Scope scope, DataType<T> type, Shape shape,
+  public static <T extends TType> Constant<T> tensorOf(Scope scope, Class<T> type, Shape shape,
       ByteDataBuffer data) {
     try (T value = Tensors.of(type, shape, data)) {
       return create(scope, value);
@@ -1272,7 +1271,7 @@ public static <T extends TType> Constant<T> create(Scope scope, T tensor) {
             .env()
             .opBuilder("Const", scope.makeOpName("Const"))
             .setAttr("value", tensor)
-            .setAttr("dtype", tensor.dataType())
+            .setAttr("dtype", tensor.type())
             .build());
   }
 
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/op/core/Helpers.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/op/core/Helpers.java
index f9ce837fe60..6a3c3521997 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/op/core/Helpers.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/op/core/Helpers.java
@@ -47,7 +47,7 @@ private Helpers() {}
     @Endpoint(name = "variable")
     public static <T extends TType> Variable<T> createVariableWithInit(Scope scope, Operand<T> init, Variable.Options... options) {
         Output<T> initOutput = init.asOutput();
-        Variable<T> newVar = Variable.create(scope,initOutput.shape(), initOutput.dataType(), options);
+        Variable<T> newVar = Variable.create(scope,initOutput.shape(), initOutput.type(), options);
         Assign<T> assignOp = Assign.create(scope, newVar, init);
         Init.add(scope, assignOp);
         return newVar;
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/op/core/Shapes.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/op/core/Shapes.java
index a2dc0cc6b5d..46731e09ea1 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/op/core/Shapes.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/op/core/Shapes.java
@@ -15,7 +15,6 @@
 package org.tensorflow.op.core;
 
 import java.util.Arrays;
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Tensor;
 import org.tensorflow.op.Scope;
@@ -49,8 +48,8 @@
  * Operand<TInt32> numPred = tf.shape.size(predShape, tf.constant(0));
  * Operand<TFloat32> predFlat = tf.shape.flatten(yPred);
  *
- * Shape<TInt64> predShape64 = tf.shape(yPred, TInt64.DTYPE);
- * Operand<TInt64> predSqueezed = tf.shape.squeeze(predShape64, TInt64.DTYPE);
+ * Shape<TInt64> predShape64 = tf.shape(yPred, TInt64.class);
+ * Operand<TInt64> predSqueezed = tf.shape.squeeze(predShape64, TInt64.class);
  * }</pre>
  */
 @Operator(group = "shape")
@@ -66,7 +65,7 @@ public abstract class Shapes {
    */
   @Endpoint(name = "flatten")
   public static <T extends TType> Operand<T> flatten(Scope scope, Operand<T> operand) {
-    return flatten(scope, operand, TInt32.DTYPE);
+    return flatten(scope, operand, TInt32.class);
   }
 
   /**
@@ -81,7 +80,7 @@ public static <T extends TType> Operand<T> flatten(Scope scope, Operand<T> opera
    */
   @Endpoint(name = "flatten")
   public static <T extends TType, U extends TNumber> Operand<T> flatten(
-      Scope scope, Operand<T> operand, DataType<U> dType) {
+      Scope scope, Operand<T> operand, Class<U> dType) {
     Operand<U> flatShape = flatten(scope, Shape.create(scope, operand, dType), dType);
     return Reshape.create(scope, operand, flatShape);
   }
@@ -95,7 +94,7 @@ public static <T extends TType, U extends TNumber> Operand<T> flatten(
    */
   @Endpoint(name = "flatten")
   public static Operand<TInt32> flatten(Scope scope, Shape<TInt32> shape) {
-    return flatten(scope, shape, TInt32.DTYPE);
+    return flatten(scope, shape, TInt32.class);
   }
 
   /**
@@ -109,11 +108,11 @@ public static Operand<TInt32> flatten(Scope scope, Shape<TInt32> shape) {
    */
   @Endpoint(name = "flatten")
   public static <U extends TNumber> Operand<U> flatten(
-      Scope scope, Shape<U> shape, DataType<U> dType) {
+      Scope scope, Shape<U> shape, Class<U> dType) {
     return ExpandDims.create(
         scope,
         size(scope, shape, dType),
-        Cast.create(scope, Constant.scalarOf(scope, -1), TInt32.DTYPE));
+        Cast.create(scope, Constant.scalarOf(scope, -1), TInt32.class));
   }
 
   /**
@@ -125,7 +124,7 @@ public static <U extends TNumber> Operand<U> flatten(
    */
   @Endpoint(name = "size")
   public static Operand<TInt32> size(Scope scope, Shape<TInt32> shape) {
-    return size(scope, shape, TInt32.DTYPE);
+    return size(scope, shape, TInt32.class);
   }
 
   /**
@@ -139,7 +138,7 @@ public static Operand<TInt32> size(Scope scope, Shape<TInt32> shape) {
    */
   @Endpoint(name = "size")
   public static <U extends TNumber> Operand<U> size(
-      Scope scope, Shape<U> shape, DataType<U> dType) {
+      Scope scope, Shape<U> shape, Class<U> dType) {
     Slice<U> dims =
         Slice.create(
             scope,
@@ -162,7 +161,7 @@ public static <U extends TNumber> Operand<U> size(
    */
   @Endpoint(name = "size")
   public static Operand<TInt32> size(Scope scope, Shape<TInt32> shape, Operand<TInt32> dim) {
-    return size(scope, shape, dim, TInt32.DTYPE);
+    return size(scope, shape, dim, TInt32.class);
   }
 
   /**
@@ -177,7 +176,7 @@ public static Operand<TInt32> size(Scope scope, Shape<TInt32> shape, Operand<TIn
    */
   @Endpoint(name = "size")
   public static <U extends TNumber> Operand<U> size(
-      Scope scope, Shape<U> shape, Operand<U> dim, DataType<U> dType) {
+      Scope scope, Shape<U> shape, Operand<U> dim, Class<U> dType) {
     return Slice.create(
         scope,
         shape,
@@ -199,7 +198,7 @@ public static <U extends TNumber> Operand<U> size(
   @Endpoint(name = "size")
   public static <T extends TType> Operand<TInt32> size(
       Scope scope, Operand<T> input, Operand<TInt32> dim) {
-    return size(scope, input, dim, TInt32.DTYPE);
+    return size(scope, input, dim, TInt32.class);
   }
 
   /**
@@ -214,7 +213,7 @@ public static <T extends TType> Operand<TInt32> size(
    */
   @Endpoint(name = "size")
   public static <T extends TType, U extends TNumber> Operand<U> size(
-      Scope scope, Operand<T> input, Operand<U> dim, DataType<U> dType) {
+      Scope scope, Operand<T> input, Operand<U> dim, Class<U> dType) {
     return size(scope, Shape.create(scope, input, dType), dim, dType);
   }
 
@@ -227,7 +226,7 @@ public static <T extends TType, U extends TNumber> Operand<U> size(
    */
   @Endpoint(name = "numDimensions")
   public static Operand<TInt32> numDimensions(Scope scope, Shape<TInt32> shape) {
-    return Size.create(scope, shape, TInt32.DTYPE);
+    return Size.create(scope, shape, TInt32.class);
   }
 
   /**
@@ -241,7 +240,7 @@ public static Operand<TInt32> numDimensions(Scope scope, Shape<TInt32> shape) {
    */
   @Endpoint(name = "numDimensions")
   public static <U extends TNumber> Operand<U> numDimensions(
-      Scope scope, Shape<U> shape, DataType<U> dType) {
+      Scope scope, Shape<U> shape, Class<U> dType) {
     return Size.create(scope, shape, dType);
   }
 
@@ -257,7 +256,7 @@ public static <U extends TNumber> Operand<U> numDimensions(
   @Endpoint(name = "reduceDims")
   public static <T extends TType> Operand<T> reduceDims(
       Scope scope, Operand<T> operand, Operand<TInt32> axis) {
-    return reduceDims(scope, operand, axis, TInt32.DTYPE);
+    return reduceDims(scope, operand, axis, TInt32.class);
   }
 
   /**
@@ -273,7 +272,7 @@ public static <T extends TType> Operand<T> reduceDims(
    */
   @Endpoint(name = "reduceDims")
   public static <T extends TType, U extends TNumber> Operand<T> reduceDims(
-      Scope scope, Operand<T> operand, Operand<U> axis, DataType<U> dType) {
+      Scope scope, Operand<T> operand, Operand<U> axis, Class<U> dType) {
     Shape<U> newShape = Shape.create(scope, operand, dType);
     return Reshape.create(scope, operand, reduceDims(scope, newShape, axis, dType));
   }
@@ -288,7 +287,7 @@ public static <T extends TType, U extends TNumber> Operand<T> reduceDims(
    */
   @Endpoint(name = "reduceDims")
   public static Operand<TInt32> reduceDims(Scope scope, Shape<TInt32> shape, Operand<TInt32> axis) {
-    return reduceDims(scope, shape, axis, TInt32.DTYPE);
+    return reduceDims(scope, shape, axis, TInt32.class);
   }
 
   /**
@@ -303,7 +302,7 @@ public static Operand<TInt32> reduceDims(Scope scope, Shape<TInt32> shape, Opera
    */
   @Endpoint(name = "reduceDims")
   public static <U extends TNumber> Operand<U> reduceDims(
-      Scope scope, Shape<U> shape, Operand<U> axis, DataType<U> dType) {
+      Scope scope, Shape<U> shape, Operand<U> axis, Class<U> dType) {
     Size<U> rank = Size.create(scope, shape, dType);
     axis = FloorMod.create(scope, axis, rank);
     Sub<U> remainder = Sub.create(scope, rank, axis);
@@ -341,7 +340,7 @@ public static <U extends TNumber> Operand<U> reduceDims(
    */
   @Endpoint(name = "squeeze")
   public static Operand<TInt32> squeeze(Scope scope, Shape<TInt32> shape) {
-    return squeeze(scope, shape, TInt32.DTYPE);
+    return squeeze(scope, shape, TInt32.class);
   }
 
   /**
@@ -355,7 +354,7 @@ public static Operand<TInt32> squeeze(Scope scope, Shape<TInt32> shape) {
    */
   @Endpoint(name = "squeeze")
   public static <U extends TNumber> Operand<U> squeeze(
-      Scope scope, Shape<U> shape, DataType<U> dType) {
+      Scope scope, Shape<U> shape, Class<U> dType) {
     Operand<TBool> mask =
         NotEqual.create(scope, shape, Cast.create(scope, OnesLike.create(scope, shape), dType));
 
@@ -371,7 +370,7 @@ public static <U extends TNumber> Operand<U> squeeze(
    */
   @Endpoint(name = "head")
   public static Operand<TInt32> head(Scope scope, Shape<TInt32> shape) {
-    return head(scope, shape, TInt32.DTYPE);
+    return head(scope, shape, TInt32.class);
   }
 
   /**
@@ -385,7 +384,7 @@ public static Operand<TInt32> head(Scope scope, Shape<TInt32> shape) {
    */
   @Endpoint(name = "head")
   public static <U extends TNumber> Operand<U> head(
-      Scope scope, Shape<U> shape, DataType<U> dType) {
+      Scope scope, Shape<U> shape, Class<U> dType) {
     return take(scope, shape, Cast.create(scope, Constant.scalarOf(scope, 1), dType), dType);
   }
 
@@ -401,7 +400,7 @@ public static <U extends TNumber> Operand<U> head(
    */
   @Endpoint(name = "take")
   public static Operand<TInt32> take(Scope scope, Shape<TInt32> shape, Operand<TInt32> n) {
-    return take(scope, shape, n, TInt32.DTYPE);
+    return take(scope, shape, n, TInt32.class);
   }
 
   /**
@@ -418,7 +417,7 @@ public static Operand<TInt32> take(Scope scope, Shape<TInt32> shape, Operand<TIn
    */
   @Endpoint(name = "take")
   public static <U extends TNumber> Operand<U> take(
-      Scope scope, Shape<U> shape, Operand<U> n, DataType<U> dType) {
+      Scope scope, Shape<U> shape, Operand<U> n, Class<U> dType) {
     return Slice.create(
         scope,
         shape,
@@ -437,7 +436,7 @@ public static <U extends TNumber> Operand<U> take(
    */
   @Endpoint(name = "tail")
   public static Operand<TInt32> tail(Scope scope, Shape<TInt32> shape) {
-    return tail(scope, shape, TInt32.DTYPE);
+    return tail(scope, shape, TInt32.class);
   }
 
   /**
@@ -453,7 +452,7 @@ public static Operand<TInt32> tail(Scope scope, Shape<TInt32> shape) {
    */
   @Endpoint(name = "tail")
   public static <U extends TNumber> Operand<U> tail(
-      Scope scope, Shape<U> shape, DataType<U> dType) {
+      Scope scope, Shape<U> shape, Class<U> dType) {
     return takeLast(scope, shape, Cast.create(scope, Constant.scalarOf(scope, 1), dType), dType);
   }
 
@@ -470,7 +469,7 @@ public static <U extends TNumber> Operand<U> tail(
   @Endpoint(name = "takeLast")
   public static <U extends TNumber> Operand<TInt32> takeLast(
       Scope scope, Shape<TInt32> shape, Operand<TInt32> n) {
-    return takeLast(scope, shape, n, TInt32.DTYPE);
+    return takeLast(scope, shape, n, TInt32.class);
   }
 
   /**
@@ -487,7 +486,7 @@ public static <U extends TNumber> Operand<TInt32> takeLast(
    */
   @Endpoint(name = "takeLast")
   public static <U extends TNumber> Operand<U> takeLast(
-      Scope scope, Shape<U> shape, Operand<U> n, DataType<U> dType) {
+      Scope scope, Shape<U> shape, Operand<U> n, Class<U> dType) {
 
     Size<U> rank = Size.create(scope, shape, dType);
     Sub<U> start = Sub.create(scope, rank, n);
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/op/core/Zeros.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/op/core/Zeros.java
index db8fb8b9671..09fe4a4f7a5 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/op/core/Zeros.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/op/core/Zeros.java
@@ -51,10 +51,10 @@ public final class Zeros<T extends TType> implements Op, Operand<T> {
    */
   @Endpoint
   @SuppressWarnings("unchecked")
-  public static <T extends TType, U extends TNumber> Zeros<T> create(Scope scope, Operand<U> dims, DataType<T> type) {
+  public static <T extends TType, U extends TNumber> Zeros<T> create(Scope scope, Operand<U> dims, Class<T> type) {
     Scope zerosScope = scope.withSubScope("Zeros");
     Operand<T> zero;
-    if (type == TString.DTYPE) {
+    if (type == TString.class) {
       zero = (Operand<T>)Constant.scalarOf(zerosScope.withName("Zero"), "");
     } else {
       zero = Cast.create(zerosScope.withName("Zero"), Constant.scalarOf(zerosScope, 0), type);
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/op/nn/SigmoidCrossEntropyWithLogits.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/op/nn/SigmoidCrossEntropyWithLogits.java
index 4f3e9569103..cecad2e6fa3 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/op/nn/SigmoidCrossEntropyWithLogits.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/op/nn/SigmoidCrossEntropyWithLogits.java
@@ -72,7 +72,7 @@ public static <T extends TNumber> Operand<T> sigmoidCrossEntropyWithLogits(
     scope = scope.withSubScope("SigmoidCrossEntropyWithLogits");
 
     Operand<T> zeros =
-        Cast.create(scope, ZerosLike.create(scope, logits), logits.asOutput().dataType());
+        Cast.create(scope, ZerosLike.create(scope, logits), logits.asOutput().type());
     Operand<TBool> cond = GreaterEqual.create(scope, logits, zeros);
 
     Operand<T> reluLogits = Select.create(scope, cond, logits, zeros);
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/op/nn/SoftmaxCrossEntropyWithLogits.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/op/nn/SoftmaxCrossEntropyWithLogits.java
index 0c8bac697ed..5dae45ff17b 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/op/nn/SoftmaxCrossEntropyWithLogits.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/op/nn/SoftmaxCrossEntropyWithLogits.java
@@ -78,24 +78,20 @@ public static <T extends TNumber, U extends TNumber> Operand<T> softmaxCrossEntr
       axis += logits.asOutput().shape().numDimensions();
     }
 
-
-    boolean convertToFloat32 =
-        logits.asOutput().dataType() == TFloat16.DTYPE
-            || logits.asOutput().dataType() == TBfloat16.DTYPE;
-    if (convertToFloat32) {
+    if (logits.asOutput().type() == TFloat16.class || logits.asOutput().type() == TBfloat16.class) {
       Operand<TFloat32> result =  softmaxCrossEntropyWithLogits(scope,
-              Cast.create(scope, labels, TFloat32.DTYPE),
-              Cast.create(scope, logits, TFloat32.DTYPE),
+              Cast.create(scope, labels, TFloat32.class),
+              Cast.create(scope, logits, TFloat32.class),
               axis);
-      return Cast.create(scope, result, logits.asOutput().dataType());
-    } else if(!logits.asOutput().dataType().equals(labels.asOutput().dataType())) {
+      return Cast.create(scope, result, logits.asOutput().type());
+    } else if(!logits.asOutput().type().equals(labels.asOutput().type())) {
       return softmaxCrossEntropyWithLogits(scope,
-              Cast.create(scope, labels, logits.asOutput().dataType()),
+              Cast.create(scope, labels, logits.asOutput().type()),
               logits,
               axis);
     }
 
-    Operand<TInt64> inputRank = Cast.create(scope, Rank.create(scope, logits), TInt64.DTYPE);
+    Operand<TInt64> inputRank = Cast.create(scope, Rank.create(scope, logits), TInt64.class);
     Shape shape = logits.asOutput().shape();
 
     // Move the dim to the end if dim is not the last dimension.
@@ -167,13 +163,13 @@ private static <T extends TNumber> Operand<T> flattenOuterDims(Scope scope, Oper
       }
     }
 
-    Operand<TInt64> rank = Cast.create(scope, Rank.create(scope, logits), TInt64.DTYPE);
+    Operand<TInt64> rank = Cast.create(scope, Rank.create(scope, logits), TInt64.class);
     Operand<TInt64> rankMinusOne = Sub.create(scope, rank, one);
 
     Operand<TInt64> lastDimSize =
         Slice.create(
             scope,
-            org.tensorflow.op.core.Shape.create(scope, logits, TInt64.DTYPE),
+            org.tensorflow.op.core.Shape.create(scope, logits, TInt64.class),
             rankMinusOne,
             one);
     Operand<TInt64> concat =
@@ -197,7 +193,7 @@ private static <T extends TNumber> Operand<T> flattenOuterDims(Scope scope, Oper
    */
   private static <T extends TNumber, U extends TNumber> Operand<T> moveDimToEnd(
       Scope scope, Operand<T> input, int dimIndex, Operand<U> rank) {
-    DataType<? extends TNumber> rankDType = rank.asOutput().dataType();
+    Class<? extends TNumber> rankDType = rank.asOutput().type();
     Operand one = Cast.create(scope, Constant.scalarOf(scope, 1), rankDType);
     List<Operand<U>> concatList =
         Arrays.asList(
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/op/nn/SparseSoftmaxCrossEntropyWithLogits.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/op/nn/SparseSoftmaxCrossEntropyWithLogits.java
index ebd6f74e7d8..d99e2c1c00b 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/op/nn/SparseSoftmaxCrossEntropyWithLogits.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/op/nn/SparseSoftmaxCrossEntropyWithLogits.java
@@ -74,11 +74,8 @@ public static <T extends TNumber, U extends TNumber> Operand sparseSoftmaxCrossE
     scope = scope.withSubScope("SparseSoftmaxCrossEntropyWithLogits");
     /** cannot use generics on preciseLogits as it may be recast later */
     Operand preciseLogits = logits;
-    boolean convertToFloat32 =
-        logits.asOutput().dataType() == TFloat16.DTYPE
-            || logits.asOutput().dataType() == TBfloat16.DTYPE;
-    if (convertToFloat32) {
-      preciseLogits = Cast.create(scope, logits, TFloat32.DTYPE);
+    if (logits.asOutput().type() == TFloat16.class || logits.asOutput().type() == TBfloat16.class) {
+      preciseLogits = Cast.create(scope, logits, TFloat32.class);
     }
     Shape labelsStaticShape = labels.asOutput().shape();
     org.tensorflow.op.core.Shape<TInt32> labelsShape =
@@ -115,8 +112,8 @@ public static <T extends TNumber, U extends TNumber> Operand sparseSoftmaxCrossE
           org.tensorflow.op.nn.raw.SparseSoftmaxCrossEntropyWithLogits.create(
               scope, preciseLogits, labels);
       Operand loss = smax.loss();
-      if (logits.asOutput().dataType() == TFloat16.DTYPE) {
-        loss = Cast.create(scope, loss, TFloat16.DTYPE);
+      if (logits.asOutput().type() == TFloat16.class) {
+        loss = Cast.create(scope, loss, TFloat16.class);
       }
       return loss;
     }
@@ -153,8 +150,8 @@ public static <T extends TNumber, U extends TNumber> Operand sparseSoftmaxCrossE
             scope, preciseLogits, labels);
     Operand cost = smax.loss();
     cost = Reshape.create(scope, cost, labelsShape);
-    if (logits.asOutput().dataType() == TFloat16.DTYPE) {
-      cost = Cast.create(scope, cost, TFloat16.DTYPE);
+    if (logits.asOutput().type() == TFloat16.class) {
+      cost = Cast.create(scope, cost, TFloat16.class);
     }
     return cost;
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TBfloat16.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TBfloat16.java
index 402e7d61747..2b4d2cef13b 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TBfloat16.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TBfloat16.java
@@ -29,7 +29,8 @@
 import org.tensorflow.ndarray.StdArrays;
 import org.tensorflow.ndarray.buffer.FloatDataBuffer;
 import org.tensorflow.ndarray.buffer.layout.DataLayouts;
-import org.tensorflow.tensor.FloatTensor;
+import org.tensorflow.types.annotation.TensorType;
+import org.tensorflow.types.tensor.FloatTensor;
 import org.tensorflow.types.family.TFloating;
 
 /**
@@ -48,14 +49,9 @@
  * <p>Note that some CPUs support the bfloat16 format natively, which can result in faster
  * computation compared to {@link TFloat16} when GPUs are not used.
  */
+@TensorType(dataType = DataType.BFLOAT16, impl = TBfloat16Impl.class)
 public interface TBfloat16 extends FloatTensor, TFloating<TBfloat16, Float> {
 
-  /** readable-name for the data type */
-  static final String NAME = "BFLOAT16";
-
-  /** Type metadata */
-  DataType<TBfloat16> DTYPE = DataType.create(NAME, 14, 2, TBfloat16Impl::new);
-
   /**
    * Allocates a new tensor for storing a single float value.
    *
@@ -63,7 +59,7 @@ public interface TBfloat16 extends FloatTensor, TFloating<TBfloat16, Float> {
    * @return the new tensor
    */
   static TBfloat16 scalarOf(float value) {
-    return Tensors.of(DTYPE, Shape.scalar(), t -> t.setFloat(value));
+    return Tensors.of(TBfloat16.class, Shape.scalar(), t -> t.setFloat(value));
   }
 
   /**
@@ -76,7 +72,7 @@ static TBfloat16 vectorOf(float... values) {
     if (values == null) {
       throw new IllegalArgumentException();
     }
-    return Tensors.of(DTYPE, Shape.of(values.length), t -> StdArrays.copyTo(values, t));
+    return Tensors.of(TBfloat16.class, Shape.of(values.length), t -> StdArrays.copyTo(values, t));
   }
 
   /**
@@ -88,7 +84,7 @@ static TBfloat16 vectorOf(float... values) {
    * @return the new tensor
    */
   static TBfloat16 tensorOf(NdArray<Float> src) {
-    return Tensors.of(DTYPE, src.shape(), src::copyTo);
+    return Tensors.of(TBfloat16.class, src.shape(), src::copyTo);
   }
 
   /**
@@ -98,7 +94,7 @@ static TBfloat16 tensorOf(NdArray<Float> src) {
    * @return the new tensor
    */
   static TBfloat16 tensorOf(Shape shape) {
-    return Tensors.of(DTYPE, shape);
+    return Tensors.of(TBfloat16.class, shape);
   }
 
   /**
@@ -109,7 +105,7 @@ static TBfloat16 tensorOf(Shape shape) {
    * @return the new tensor
    */
   static TBfloat16 tensorOf(Shape shape, FloatDataBuffer data) {
-    return Tensors.of(DTYPE, shape, t -> t.write(data));
+    return Tensors.of(TBfloat16.class, shape, t -> t.write(data));
   }
 
   /**
@@ -121,7 +117,7 @@ static TBfloat16 tensorOf(Shape shape, FloatDataBuffer data) {
    * @throws TensorFlowException if the tensor cannot be allocated or initialized
    */
   static TBfloat16 tensorOf(Shape shape, Consumer<TBfloat16> tensorInit) {
-    return Tensors.of(DTYPE, shape, tensorInit);
+    return Tensors.of(TBfloat16.class, shape, tensorInit);
   }
 }
 
@@ -131,11 +127,16 @@ static TBfloat16 tensorOf(Shape shape, Consumer<TBfloat16> tensorInit) {
 class TBfloat16Impl extends FloatTensorImpl implements TBfloat16 {
 
   TBfloat16Impl(TF_Tensor nativeTensorHandle, Shape shape) {
-    super(nativeTensorHandle, DTYPE, shape, mapMemory(nativeTensorHandle));
+    super(nativeTensorHandle, shape, mapMemory(nativeTensorHandle));
   }
 
   private static FloatDataBuffer mapMemory(TF_Tensor nativeTensorHandle) {
     return DataLayouts.BFLOAT16.applyTo(TensorBuffers.toShorts(nativeTensorHandle));
   }
+
+  @Override
+  public Class<TBfloat16> type() {
+    return TBfloat16.class;
+  }
 }
 
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TBool.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TBool.java
index 3d95b593d3b..54c4ed5262a 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TBool.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TBool.java
@@ -29,7 +29,8 @@
 import org.tensorflow.ndarray.StdArrays;
 import org.tensorflow.ndarray.buffer.BooleanDataBuffer;
 import org.tensorflow.ndarray.buffer.layout.DataLayouts;
-import org.tensorflow.tensor.BooleanTensor;
+import org.tensorflow.types.annotation.TensorType;
+import org.tensorflow.types.tensor.BooleanTensor;
 import org.tensorflow.types.family.TType;
 
 /**
@@ -39,14 +40,9 @@
  * explicit mapping between Java boolean values and byte buffers using the {@link DataLayouts#BOOL
  * BOOL} layout, which may impact I/O performances.
  */
+@TensorType(dataType = DataType.BOOL, impl = TBoolImpl.class)
 public interface TBool extends BooleanTensor, TType<TBool, Boolean> {
 
-  /** readable-name for the data type */
-  static final String NAME = "BOOL";
-
-  /** Type metadata */
-  DataType<TBool> DTYPE = DataType.create(NAME, 10, 1, TBoolImpl::new);
-
   /**
    * Allocates a new tensor for storing a single boolean value.
    *
@@ -54,7 +50,7 @@ public interface TBool extends BooleanTensor, TType<TBool, Boolean> {
    * @return the new tensor
    */
   static TBool scalarOf(boolean value) {
-    return Tensors.of(DTYPE, Shape.scalar(), t -> t.setBoolean(value));
+    return Tensors.of(TBool.class, Shape.scalar(), t -> t.setBoolean(value));
   }
 
   /**
@@ -67,7 +63,7 @@ static TBool vectorOf(boolean... values) {
     if (values == null) {
       throw new IllegalArgumentException();
     }
-    return Tensors.of(DTYPE, Shape.of(values.length), t -> StdArrays.copyTo(values, t));
+    return Tensors.of(TBool.class, Shape.of(values.length), t -> StdArrays.copyTo(values, t));
   }
 
   /**
@@ -79,7 +75,7 @@ static TBool vectorOf(boolean... values) {
    * @return the new tensor
    */
   static TBool tensorOf(NdArray<Boolean> src) {
-    return Tensors.of(DTYPE, src.shape(), src::copyTo);
+    return Tensors.of(TBool.class, src.shape(), src::copyTo);
   }
 
   /**
@@ -89,7 +85,7 @@ static TBool tensorOf(NdArray<Boolean> src) {
    * @return the new tensor
    */
   static TBool tensorOf(Shape shape) {
-    return Tensors.of(DTYPE, shape);
+    return Tensors.of(TBool.class, shape);
   }
 
   /**
@@ -100,7 +96,7 @@ static TBool tensorOf(Shape shape) {
    * @return the new tensor
    */
   static TBool tensorOf(Shape shape, BooleanDataBuffer data) {
-    return Tensors.of(DTYPE, shape, d -> d.write(data));
+    return Tensors.of(TBool.class, shape, d -> d.write(data));
   }
 
   /**
@@ -112,7 +108,7 @@ static TBool tensorOf(Shape shape, BooleanDataBuffer data) {
    * @throws TensorFlowException if the tensor cannot be allocated or initialized
    */
   static TBool tensorOf(Shape shape, Consumer<TBool> tensorInit) {
-    return Tensors.of(DTYPE, shape, tensorInit);
+    return Tensors.of(TBool.class, shape, tensorInit);
   }
 }
 
@@ -122,6 +118,11 @@ static TBool tensorOf(Shape shape, Consumer<TBool> tensorInit) {
 class TBoolImpl extends BooleanTensorImpl implements TBool {
 
   TBoolImpl(TF_Tensor nativeTensor, Shape shape) {
-    super(nativeTensor, DTYPE, shape, TensorBuffers.toBooleans(nativeTensor));
+    super(nativeTensor, shape, TensorBuffers.toBooleans(nativeTensor));
+  }
+
+  @Override
+  public Class<TBool> type() {
+    return TBool.class;
   }
 }
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TFloat16.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TFloat16.java
index 1e8a3d2a1e1..a4043c2d8e3 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TFloat16.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TFloat16.java
@@ -29,7 +29,8 @@
 import org.tensorflow.ndarray.StdArrays;
 import org.tensorflow.ndarray.buffer.FloatDataBuffer;
 import org.tensorflow.ndarray.buffer.layout.DataLayouts;
-import org.tensorflow.tensor.FloatTensor;
+import org.tensorflow.types.annotation.TensorType;
+import org.tensorflow.types.tensor.FloatTensor;
 import org.tensorflow.types.family.TFloating;
 
 /**
@@ -45,14 +46,9 @@
  * most CPUs do not support this format natively. For CPU computation on 16-bit floats, the {@link
  * TBfloat16} tensor type might be a better option.
  */
+@TensorType(dataType = DataType.HALF, impl = TFloat16Impl.class)
 public interface TFloat16 extends FloatTensor, TFloating<TFloat16, Float> {
 
-  /** readable-name for the data type */
-  static final String NAME = "FLOAT16";
-
-  /** Type metadata */
-  DataType<TFloat16> DTYPE = DataType.create(NAME, 19, 2, TFloat16Impl::new);
-
   /**
    * Allocates a new tensor for storing a single float value.
    *
@@ -60,7 +56,7 @@ public interface TFloat16 extends FloatTensor, TFloating<TFloat16, Float> {
    * @return the new tensor
    */
   static TFloat16 scalarOf(float value) {
-    return Tensors.of(DTYPE, Shape.scalar(), t -> t.setFloat(value));
+    return Tensors.of(TFloat16.class, Shape.scalar(), t -> t.setFloat(value));
   }
 
   /**
@@ -73,7 +69,7 @@ static TFloat16 vectorOf(float... values) {
     if (values == null) {
       throw new IllegalArgumentException();
     }
-    return Tensors.of(DTYPE, Shape.of(values.length), t -> StdArrays.copyTo(values, t));
+    return Tensors.of(TFloat16.class, Shape.of(values.length), t -> StdArrays.copyTo(values, t));
   }
 
   /**
@@ -85,7 +81,7 @@ static TFloat16 vectorOf(float... values) {
    * @return the new tensor
    */
   static TFloat16 tensorOf(NdArray<Float> src) {
-    return Tensors.of(DTYPE, src.shape(), src::copyTo);
+    return Tensors.of(TFloat16.class, src.shape(), src::copyTo);
   }
 
   /**
@@ -95,7 +91,7 @@ static TFloat16 tensorOf(NdArray<Float> src) {
    * @return the new tensor
    */
   static TFloat16 tensorOf(Shape shape) {
-    return Tensors.of(DTYPE, shape);
+    return Tensors.of(TFloat16.class, shape);
   }
 
   /**
@@ -106,7 +102,7 @@ static TFloat16 tensorOf(Shape shape) {
    * @return the new tensor
    */
   static TFloat16 tensorOf(Shape shape, FloatDataBuffer data) {
-    return Tensors.of(DTYPE, shape, t -> t.write(data));
+    return Tensors.of(TFloat16.class, shape, t -> t.write(data));
   }
 
   /**
@@ -118,7 +114,7 @@ static TFloat16 tensorOf(Shape shape, FloatDataBuffer data) {
    * @throws TensorFlowException if the tensor cannot be allocated or initialized
    */
   static TFloat16 tensorOf(Shape shape, Consumer<TFloat16> tensorInit) {
-    return Tensors.of(DTYPE, shape, tensorInit);
+    return Tensors.of(TFloat16.class, shape, tensorInit);
   }
 }
 
@@ -128,11 +124,16 @@ static TFloat16 tensorOf(Shape shape, Consumer<TFloat16> tensorInit) {
 class TFloat16Impl extends FloatTensorImpl implements TFloat16 {
 
   TFloat16Impl(TF_Tensor nativeTensor, Shape shape) {
-    super(nativeTensor, DTYPE, shape, mapMemory(nativeTensor));
+    super(nativeTensor, shape, mapMemory(nativeTensor));
   }
 
   private static FloatDataBuffer mapMemory(TF_Tensor nativeTensor) {
     return DataLayouts.FLOAT16.applyTo(TensorBuffers.toShorts(nativeTensor));
   }
+
+  @Override
+  public Class<TFloat16> type() {
+    return TFloat16.class;
+  }
 }
 
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TFloat32.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TFloat32.java
index 378ed62d2a4..d68ec86af53 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TFloat32.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TFloat32.java
@@ -28,20 +28,16 @@
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.ndarray.StdArrays;
 import org.tensorflow.ndarray.buffer.FloatDataBuffer;
-import org.tensorflow.tensor.FloatTensor;
+import org.tensorflow.types.annotation.TensorType;
+import org.tensorflow.types.tensor.FloatTensor;
 import org.tensorflow.types.family.TFloating;
 
 /**
  * IEEE-754 single-precision 32-bit float tensor type.
  */
+@TensorType(dataType = DataType.FLOAT, impl = TFloat32Impl.class)
 public interface TFloat32 extends FloatTensor, TFloating<TFloat32, Float> {
 
-  /** readable-name for the data type */
-  static final String NAME = "FLOAT";
-
-  /** Type metadata */
-  DataType<TFloat32> DTYPE = DataType.create(NAME, 1, 4, TFloat32Impl::new);
-
   /**
    * Allocates a new tensor for storing a single float value.
    *
@@ -49,7 +45,7 @@ public interface TFloat32 extends FloatTensor, TFloating<TFloat32, Float> {
    * @return the new tensor
    */
   static TFloat32 scalarOf(float value) {
-    return Tensors.of(DTYPE, Shape.scalar(), t -> t.setFloat(value));
+    return Tensors.of(TFloat32.class, Shape.scalar(), t -> t.setFloat(value));
   }
 
   /**
@@ -62,7 +58,7 @@ static TFloat32 vectorOf(float... values) {
     if (values == null) {
       throw new IllegalArgumentException();
     }
-    return Tensors.of(DTYPE, Shape.of(values.length), t -> StdArrays.copyTo(values, t));
+    return Tensors.of(TFloat32.class, Shape.of(values.length), t -> StdArrays.copyTo(values, t));
   }
 
   /**
@@ -74,7 +70,7 @@ static TFloat32 vectorOf(float... values) {
    * @return the new tensor
    */
   static TFloat32 tensorOf(NdArray<Float> src) {
-    return Tensors.of(DTYPE, src.shape(), src::copyTo);
+    return Tensors.of(TFloat32.class, src.shape(), src::copyTo);
   }
 
   /**
@@ -84,7 +80,7 @@ static TFloat32 tensorOf(NdArray<Float> src) {
    * @return the new tensor
    */
   static TFloat32 tensorOf(Shape shape) {
-    return Tensors.of(DTYPE, shape);
+    return Tensors.of(TFloat32.class, shape);
   }
 
   /**
@@ -95,7 +91,7 @@ static TFloat32 tensorOf(Shape shape) {
    * @return the new tensor
    */
   static TFloat32 tensorOf(Shape shape, FloatDataBuffer data) {
-    return Tensors.of(DTYPE, shape, t -> t.write(data));
+    return Tensors.of(TFloat32.class, shape, t -> t.write(data));
   }
 
   /**
@@ -107,7 +103,7 @@ static TFloat32 tensorOf(Shape shape, FloatDataBuffer data) {
    * @throws TensorFlowException if the tensor cannot be allocated or initialized
    */
   static TFloat32 tensorOf(Shape shape, Consumer<TFloat32> tensorInit) {
-    return Tensors.of(DTYPE, shape, tensorInit);
+    return Tensors.of(TFloat32.class, shape, tensorInit);
   }
 }
 
@@ -117,7 +113,12 @@ static TFloat32 tensorOf(Shape shape, Consumer<TFloat32> tensorInit) {
 class TFloat32Impl extends FloatTensorImpl implements TFloat32 {
 
   TFloat32Impl(TF_Tensor nativeTensor, Shape shape) {
-    super(nativeTensor, DTYPE, shape, TensorBuffers.toFloats(nativeTensor));
+    super(nativeTensor, shape, TensorBuffers.toFloats(nativeTensor));
+  }
+
+  @Override
+  public Class<TFloat32> type() {
+    return TFloat32.class;
   }
 }
 
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TFloat64.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TFloat64.java
index 3c55d64fbaf..73eb9632859 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TFloat64.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TFloat64.java
@@ -28,20 +28,16 @@
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.ndarray.StdArrays;
 import org.tensorflow.ndarray.buffer.DoubleDataBuffer;
-import org.tensorflow.tensor.DoubleTensor;
+import org.tensorflow.types.annotation.TensorType;
+import org.tensorflow.types.tensor.DoubleTensor;
 import org.tensorflow.types.family.TFloating;
 
 /**
  * IEEE-754 double-precision 64-bit float tensor type.
  */
+@TensorType(dataType = DataType.DOUBLE, impl = TFloat64Impl.class)
 public interface TFloat64 extends DoubleTensor, TFloating<TFloat64, Double> {
 
-  /** readable-name for the data type */
-  static final String NAME = "DOUBLE";
-
-  /** Type metadata */
-  DataType<TFloat64> DTYPE = DataType.create(NAME, 2, 8, TFloat64Impl::new);
-
   /**
    * Allocates a new tensor for storing a single double value.
    *
@@ -49,7 +45,7 @@ public interface TFloat64 extends DoubleTensor, TFloating<TFloat64, Double> {
    * @return the new tensor
    */
   static TFloat64 scalarOf(double value) {
-    return Tensors.of(DTYPE, Shape.scalar(), t -> t.setDouble(value));
+    return Tensors.of(TFloat64.class, Shape.scalar(), t -> t.setDouble(value));
   }
 
   /**
@@ -62,7 +58,7 @@ static TFloat64 vectorOf(double... values) {
     if (values == null) {
       throw new IllegalArgumentException();
     }
-    return Tensors.of(DTYPE, Shape.of(values.length), t -> StdArrays.copyTo(values, t));
+    return Tensors.of(TFloat64.class, Shape.of(values.length), t -> StdArrays.copyTo(values, t));
   }
 
   /**
@@ -74,7 +70,7 @@ static TFloat64 vectorOf(double... values) {
    * @return the new tensor
    */
   static TFloat64 tensorOf(NdArray<Double> src) {
-    return Tensors.of(DTYPE, src.shape(), src::copyTo);
+    return Tensors.of(TFloat64.class, src.shape(), src::copyTo);
   }
 
   /**
@@ -84,7 +80,7 @@ static TFloat64 tensorOf(NdArray<Double> src) {
    * @return the new tensor
    */
   static TFloat64 tensorOf(Shape shape) {
-    return Tensors.of(DTYPE, shape);
+    return Tensors.of(TFloat64.class, shape);
   }
 
   /**
@@ -95,7 +91,7 @@ static TFloat64 tensorOf(Shape shape) {
    * @return the new tensor
    */
   static TFloat64 tensorOf(Shape shape, DoubleDataBuffer data) {
-    return Tensors.of(DTYPE, shape, t -> t.write(data));
+    return Tensors.of(TFloat64.class, shape, t -> t.write(data));
   }
 
   /**
@@ -107,7 +103,7 @@ static TFloat64 tensorOf(Shape shape, DoubleDataBuffer data) {
    * @throws TensorFlowException if the tensor cannot be allocated or initialized
    */
   static TFloat64 tensorOf(Shape shape, Consumer<TFloat64> tensorInit) {
-    return Tensors.of(DTYPE, shape, tensorInit);
+    return Tensors.of(TFloat64.class, shape, tensorInit);
   }
 }
 
@@ -117,6 +113,11 @@ static TFloat64 tensorOf(Shape shape, Consumer<TFloat64> tensorInit) {
 class TFloat64Impl extends DoubleTensorImpl implements TFloat64 {
 
   TFloat64Impl(TF_Tensor nativeTensor, Shape shape) {
-    super(nativeTensor, DTYPE, shape, TensorBuffers.toDoubles(nativeTensor));
+    super(nativeTensor, shape, TensorBuffers.toDoubles(nativeTensor));
+  }
+
+  @Override
+  public Class<TFloat64> type() {
+    return TFloat64.class;
   }
 }
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TInt32.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TInt32.java
index 5ae4b9a388e..ed68aaaa9db 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TInt32.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TInt32.java
@@ -27,20 +27,16 @@
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.ndarray.StdArrays;
 import org.tensorflow.ndarray.buffer.IntDataBuffer;
-import org.tensorflow.tensor.IntTensor;
+import org.tensorflow.types.annotation.TensorType;
+import org.tensorflow.types.tensor.IntTensor;
 import org.tensorflow.types.family.TNumber;
 
 /**
  * 32-bit signed integer tensor type.
  */
+@TensorType(dataType = DataType.INT32, impl = TInt32Impl.class)
 public interface TInt32 extends IntTensor, TNumber<TInt32, Integer> {
 
-  /** readable-name for the data type */
-  static final String NAME = "INT32";
-
-  /** Type metadata */
-  DataType<TInt32> DTYPE = DataType.create(NAME, 3, 4, TInt32Impl::new);
-
   /**
    * Allocates a new tensor for storing a single int value.
    *
@@ -48,7 +44,7 @@ public interface TInt32 extends IntTensor, TNumber<TInt32, Integer> {
    * @return the new tensor
    */
   static TInt32 scalarOf(int value) {
-    return Tensors.of(DTYPE, Shape.scalar(), t -> t.setInt(value));
+    return Tensors.of(TInt32.class, Shape.scalar(), t -> t.setInt(value));
   }
 
   /**
@@ -62,7 +58,7 @@ static TInt32 vectorOf(int... values) {
     if (values == null) {
       throw new IllegalArgumentException();
     }
-    return Tensors.of(DTYPE, Shape.of(values.length), t -> StdArrays.copyTo(values, t));
+    return Tensors.of(TInt32.class, Shape.of(values.length), t -> StdArrays.copyTo(values, t));
   }
 
   /**
@@ -74,7 +70,7 @@ static TInt32 vectorOf(int... values) {
    * @return the new tensor
    */
   static TInt32 tensorOf(NdArray<Integer> src) {
-    return Tensors.of(DTYPE, src.shape(), src::copyTo);
+    return Tensors.of(TInt32.class, src.shape(), src::copyTo);
   }
 
   /**
@@ -84,7 +80,7 @@ static TInt32 tensorOf(NdArray<Integer> src) {
    * @return the new tensor
    */
   static TInt32 tensorOf(Shape shape) {
-    return Tensors.of(DTYPE, shape);
+    return Tensors.of(TInt32.class, shape);
   }
 
   /**
@@ -95,7 +91,7 @@ static TInt32 tensorOf(Shape shape) {
    * @return the new tensor
    */
   static TInt32 tensorOf(Shape shape, IntDataBuffer data) {
-    return Tensors.of(DTYPE, shape, t -> t.write(data));
+    return Tensors.of(TInt32.class, shape, t -> t.write(data));
   }
 
   /**
@@ -106,7 +102,7 @@ static TInt32 tensorOf(Shape shape, IntDataBuffer data) {
    * @return the new tensor
    */
   static TInt32 tensorOf(Shape shape, Consumer<TInt32> tensorInit) {
-    return Tensors.of(DTYPE, shape, tensorInit);
+    return Tensors.of(TInt32.class, shape, tensorInit);
   }
 }
 
@@ -116,6 +112,11 @@ static TInt32 tensorOf(Shape shape, Consumer<TInt32> tensorInit) {
 class TInt32Impl extends IntTensorImpl implements TInt32 {
 
   TInt32Impl(TF_Tensor nativeTensor, Shape shape) {
-    super(nativeTensor, DTYPE, shape, TensorBuffers.toInts(nativeTensor));
+    super(nativeTensor, shape, TensorBuffers.toInts(nativeTensor));
+  }
+
+  @Override
+  public Class<TInt32> type() {
+    return TInt32.class;
   }
 }
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TInt64.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TInt64.java
index e91c8a7703e..97e83c34a6d 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TInt64.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TInt64.java
@@ -28,20 +28,16 @@
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.ndarray.StdArrays;
 import org.tensorflow.ndarray.buffer.LongDataBuffer;
-import org.tensorflow.tensor.LongTensor;
+import org.tensorflow.types.annotation.TensorType;
+import org.tensorflow.types.tensor.LongTensor;
 import org.tensorflow.types.family.TNumber;
 
 /**
  * 64-bit signed integer tensor type.
  */
+@TensorType(dataType = DataType.INT64, impl = TInt64Impl.class)
 public interface TInt64 extends LongTensor, TNumber<TInt64, Long> {
 
-  /** readable-name for the data type */
-  static final String NAME = "INT64";
-
-  /** Type metadata */
-  DataType<TInt64> DTYPE = DataType.create(NAME, 9, 8, TInt64Impl::new);
-
   /**
    * Allocates a new tensor for storing a single long value.
    *
@@ -49,7 +45,7 @@ public interface TInt64 extends LongTensor, TNumber<TInt64, Long> {
    * @return the new tensor
    */
   static TInt64 scalarOf(long value) {
-    return Tensors.of(DTYPE, Shape.scalar(), t -> t.setLong(value));
+    return Tensors.of(TInt64.class, Shape.scalar(), t -> t.setLong(value));
   }
 
   /**
@@ -62,7 +58,7 @@ static TInt64 vectorOf(long... values) {
     if (values == null) {
       throw new IllegalArgumentException();
     }
-    return Tensors.of(DTYPE, Shape.of(values.length), t -> StdArrays.copyTo(values, t));
+    return Tensors.of(TInt64.class, Shape.of(values.length), t -> StdArrays.copyTo(values, t));
   }
 
   /**
@@ -74,7 +70,7 @@ static TInt64 vectorOf(long... values) {
    * @return the new tensor
    */
   static TInt64 tensorOf(NdArray<Long> src) {
-    return Tensors.of(DTYPE, src.shape(), src::copyTo);
+    return Tensors.of(TInt64.class, src.shape(), src::copyTo);
   }
 
   /**
@@ -84,7 +80,7 @@ static TInt64 tensorOf(NdArray<Long> src) {
    * @return the new tensor
    */
   static TInt64 tensorOf(Shape shape) {
-    return Tensors.of(DTYPE, shape);
+    return Tensors.of(TInt64.class, shape);
   }
 
   /**
@@ -95,7 +91,7 @@ static TInt64 tensorOf(Shape shape) {
    * @return the new tensor
    */
   static TInt64 tensorOf(Shape shape, LongDataBuffer data) {
-    return Tensors.of(DTYPE, shape, t -> t.write(data));
+    return Tensors.of(TInt64.class, shape, t -> t.write(data));
   }
 
   /**
@@ -107,7 +103,7 @@ static TInt64 tensorOf(Shape shape, LongDataBuffer data) {
    * @throws TensorFlowException if the tensor cannot be allocated or initialized
    */
   static TInt64 tensorOf(Shape shape, Consumer<TInt64> tensorInit) {
-    return Tensors.of(DTYPE, shape, tensorInit);
+    return Tensors.of(TInt64.class, shape, tensorInit);
   }
 }
 
@@ -117,6 +113,11 @@ static TInt64 tensorOf(Shape shape, Consumer<TInt64> tensorInit) {
 class TInt64Impl extends LongTensorImpl implements TInt64 {
 
   TInt64Impl(TF_Tensor nativeTensor, Shape shape) {
-    super(nativeTensor, DTYPE, shape, TensorBuffers.toLongs(nativeTensor));
+    super(nativeTensor, shape, TensorBuffers.toLongs(nativeTensor));
+  }
+
+  @Override
+  public Class<TInt64> type() {
+    return TInt64.class;
   }
 }
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TString.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TString.java
index 69dc26806c5..948833a5b71 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TString.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TString.java
@@ -32,7 +32,8 @@
 import org.tensorflow.ndarray.buffer.DataBuffer;
 import org.tensorflow.ndarray.buffer.layout.DataLayout;
 import org.tensorflow.ndarray.buffer.layout.DataLayouts;
-import org.tensorflow.tensor.StringTensor;
+import org.tensorflow.types.annotation.TensorType;
+import org.tensorflow.types.tensor.StringTensor;
 import org.tensorflow.types.family.TType;
 
 /**
@@ -44,16 +45,9 @@
  * its values initially, so TensorFlow can compute and allocate the right amount of memory. Then the
  * data in the tensor is initialized once and cannot be modified afterwards.
  */
+@TensorType(dataType = DataType.STRING, impl = TStringImpl.class)
 public interface TString extends StringTensor, TType<TString, String> {
 
-  /** readable-name for the data type */
-  static final String NAME = "STRING";
-
-  /**
-   * Type metadata
-   */
-  DataType<TString> DTYPE = DataType.create(NAME, 7, -1, TStringImpl::new);
-
   /**
    * Allocates a new tensor for storing a string scalar.
    *
@@ -230,7 +224,7 @@ public TString using(Charset charset) {
 
   static <T> TString createTensor(NdArray<T> src, Function<T, byte[]> getBytes) {
     long size = ByteSequenceTensorBuffer.computeSize(src, getBytes);
-    return Tensors.of(TString.DTYPE, src.shape(), size, t ->
+    return Tensors.of(TString.class, src.shape(), size, t ->
         ((TStringImpl)t).rawBuffer().init(src, getBytes)
     );
   }
@@ -242,7 +236,12 @@ static <T> TString createTensor(NdArray<T> src, Function<T, byte[]> getBytes) {
   private static final DataLayout<DataBuffer<byte[]>, String> UTF_8_LAYOUT = DataLayouts.ofStrings(StandardCharsets.UTF_8);
 
   private TStringImpl(TF_Tensor nativeTensor, Shape shape, DataLayout<DataBuffer<byte[]>, String> layout, ByteSequenceTensorBuffer rawBuffer) {
-    super(nativeTensor, TString.DTYPE, shape, layout, rawBuffer);
+    super(nativeTensor, shape, layout, rawBuffer);
+  }
+
+  @Override
+  public Class<TString> type() {
+    return TString.class;
   }
 }
 
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TUint8.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TUint8.java
index e5780a18ab8..048826d6fc3 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TUint8.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TUint8.java
@@ -28,20 +28,16 @@
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.ndarray.StdArrays;
 import org.tensorflow.ndarray.buffer.ByteDataBuffer;
-import org.tensorflow.tensor.ByteTensor;
+import org.tensorflow.types.annotation.TensorType;
+import org.tensorflow.types.tensor.ByteTensor;
 import org.tensorflow.types.family.TNumber;
 
 /**
  * 8-bit unsigned integer tensor type.
  */
+@TensorType(dataType = DataType.UINT8, impl = TUint8Impl.class)
 public interface TUint8 extends ByteTensor, TNumber<TUint8, Byte> {
 
-  /** readable-name for the data type */
-  static final String NAME = "UINT8";
-
-  /** Type metadata */
-  DataType<TUint8> DTYPE = DataType.create(NAME, 4, 1, TUint8Impl::new);
-
   /**
    * Allocates a new tensor for storing a single byte value.
    *
@@ -49,7 +45,7 @@ public interface TUint8 extends ByteTensor, TNumber<TUint8, Byte> {
    * @return the new tensor
    */
   static TUint8 scalarOf(byte value) {
-    return Tensors.of(DTYPE, Shape.scalar(), t -> t.setByte(value));
+    return Tensors.of(TUint8.class, Shape.scalar(), t -> t.setByte(value));
   }
 
   /**
@@ -62,7 +58,7 @@ static TUint8 vectorOf(byte... values) {
     if (values == null) {
       throw new IllegalArgumentException();
     }
-    return Tensors.of(DTYPE, Shape.of(values.length), t -> StdArrays.copyTo(values, t));
+    return Tensors.of(TUint8.class, Shape.of(values.length), t -> StdArrays.copyTo(values, t));
   }
 
   /**
@@ -74,7 +70,7 @@ static TUint8 vectorOf(byte... values) {
    * @return the new tensor
    */
   static TUint8 tensorOf(NdArray<Byte> src) {
-    return Tensors.of(DTYPE, src.shape(), src::copyTo);
+    return Tensors.of(TUint8.class, src.shape(), src::copyTo);
   }
 
   /**
@@ -84,7 +80,7 @@ static TUint8 tensorOf(NdArray<Byte> src) {
    * @return the new tensor
    */
   static TUint8 tensorOf(Shape shape) {
-    return Tensors.of(DTYPE, shape);
+    return Tensors.of(TUint8.class, shape);
   }
 
   /**
@@ -95,7 +91,7 @@ static TUint8 tensorOf(Shape shape) {
    * @return the new tensor
    */
   static TUint8 tensorOf(Shape shape, ByteDataBuffer data) {
-    return Tensors.of(DTYPE, shape, d -> d.write(data));
+    return Tensors.of(TUint8.class, shape, d -> d.write(data));
   }
 
   /**
@@ -107,7 +103,7 @@ static TUint8 tensorOf(Shape shape, ByteDataBuffer data) {
    * @throws TensorFlowException if the tensor cannot be allocated or initialized
    */
   static TUint8 tensorOf(Shape shape, Consumer<TUint8> tensorInit) {
-    return Tensors.of(DTYPE, shape, tensorInit);
+    return Tensors.of(TUint8.class, shape, tensorInit);
   }
 }
 
@@ -117,6 +113,11 @@ static TUint8 tensorOf(Shape shape, Consumer<TUint8> tensorInit) {
 class TUint8Impl extends ByteTensorImpl implements TUint8 {
 
   TUint8Impl(TF_Tensor nativeTensor, Shape shape) {
-    super(nativeTensor, DTYPE, shape, TensorBuffers.toBytes(nativeTensor));
+    super(nativeTensor, shape, TensorBuffers.toBytes(nativeTensor));
+  }
+
+  @Override
+  public Class<TUint8> type() {
+    return TUint8.class;
   }
 }
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/annotation/TensorType.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/annotation/TensorType.java
new file mode 100644
index 00000000000..c338cd0ec6d
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/annotation/TensorType.java
@@ -0,0 +1,30 @@
+/* Copyright 2016 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+==============================================================================*/
+
+package org.tensorflow.types.annotation;
+
+import org.tensorflow.DataType;
+import org.tensorflow.Tensor;
+
+/** Represents a type of elements in a {@link Tensor} */
+public @interface TensorType {
+
+  DataType dataType();
+
+  /**
+   * The class implementing this tensor type
+   */
+  Class<?> impl();
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/family/TType.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/family/TType.java
index d8a8063eeb9..d4adb8a7b3b 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/family/TType.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/family/TType.java
@@ -60,4 +60,6 @@ default Output<T> asOutput(Scope scope) {
   default Operation op() {
     throw new IllegalStateException("A tensor is not attached to a specific operation");
   }
+
+  Class<T> type();
 }
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/tensor/BooleanTensor.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/tensor/BooleanTensor.java
similarity index 100%
rename from tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/tensor/BooleanTensor.java
rename to tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/tensor/BooleanTensor.java
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/tensor/ByteTensor.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/tensor/ByteTensor.java
similarity index 100%
rename from tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/tensor/ByteTensor.java
rename to tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/tensor/ByteTensor.java
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/tensor/DoubleTensor.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/tensor/DoubleTensor.java
similarity index 100%
rename from tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/tensor/DoubleTensor.java
rename to tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/tensor/DoubleTensor.java
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/tensor/FloatTensor.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/tensor/FloatTensor.java
similarity index 100%
rename from tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/tensor/FloatTensor.java
rename to tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/tensor/FloatTensor.java
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/tensor/IntTensor.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/tensor/IntTensor.java
similarity index 100%
rename from tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/tensor/IntTensor.java
rename to tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/tensor/IntTensor.java
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/tensor/LongTensor.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/tensor/LongTensor.java
similarity index 100%
rename from tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/tensor/LongTensor.java
rename to tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/tensor/LongTensor.java
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/tensor/StringTensor.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/tensor/StringTensor.java
similarity index 100%
rename from tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/tensor/StringTensor.java
rename to tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/tensor/StringTensor.java

From f4d9b38ab5b4b38a388863518f4cb1387bf9ecf6 Mon Sep 17 00:00:00 2001
From: klessard <klessard@expedia.com>
Date: Mon, 26 Oct 2020 09:07:10 -0400
Subject: [PATCH 06/11] Remove our copy of DataType

---
 .../main/java/org/tensorflow/DataType.java    | 37 -------------------
 .../main/java/org/tensorflow/TensorType.java  | 11 +++++-
 .../main/java/org/tensorflow/TensorTypes.java |  8 ++--
 .../src/main/java/org/tensorflow/Tensors.java |  7 ++--
 .../internal/tensor/BooleanTensorImpl.java    |  1 -
 .../internal/tensor/ByteTensorImpl.java       |  1 -
 .../internal/tensor/DoubleTensorImpl.java     |  1 -
 .../internal/tensor/FloatTensorImpl.java      |  1 -
 .../internal/tensor/IntTensorImpl.java        |  1 -
 .../internal/tensor/LongTensorImpl.java       |  1 -
 .../internal/tensor/StringTensorImpl.java     |  1 -
 .../java/org/tensorflow/op/core/Constant.java |  1 -
 .../java/org/tensorflow/op/core/Zeros.java    |  1 -
 .../op/nn/SoftmaxCrossEntropyWithLogits.java  |  1 -
 .../java/org/tensorflow/types/TBfloat16.java  |  4 +-
 .../main/java/org/tensorflow/types/TBool.java |  4 +-
 .../java/org/tensorflow/types/TFloat16.java   |  4 +-
 .../java/org/tensorflow/types/TFloat32.java   |  4 +-
 .../java/org/tensorflow/types/TFloat64.java   |  4 +-
 .../java/org/tensorflow/types/TInt32.java     |  4 +-
 .../java/org/tensorflow/types/TInt64.java     |  4 +-
 .../java/org/tensorflow/types/TString.java    |  4 +-
 .../java/org/tensorflow/types/TUint8.java     |  4 +-
 .../types/annotation/TensorType.java          |  4 +-
 24 files changed, 37 insertions(+), 76 deletions(-)
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/DataType.java

diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/DataType.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/DataType.java
deleted file mode 100644
index fe71905255e..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/DataType.java
+++ /dev/null
@@ -1,37 +0,0 @@
-/* Copyright 2016 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-==============================================================================*/
-
-package org.tensorflow;
-
-/** Represents a type of elements in a {@link Tensor} */
-public enum DataType {
-  FLOAT(1, 4),
-  DOUBLE(2, 8),
-  INT32(3, 4),
-  UINT8(4, 1),
-  STRING(7, -1),
-  INT64(9, 8),
-  BOOL(10, 1),
-  BFLOAT16(14, 2),
-  HALF(19, 2);
-
-  final int number;
-  final int byteSize;
-
-  private DataType(int number, int byteSize) {
-    this.number = number;
-    this.byteSize = byteSize;
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/TensorType.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/TensorType.java
index c8e270cb2c8..309d0f47225 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/TensorType.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/TensorType.java
@@ -4,6 +4,7 @@
 import org.tensorflow.exceptions.TensorFlowException;
 import org.tensorflow.internal.c_api.TF_Tensor;
 import org.tensorflow.ndarray.Shape;
+import org.tensorflow.proto.framework.DataType;
 import org.tensorflow.types.family.TType;
 
 class TensorType {
@@ -16,8 +17,12 @@ DataType dataType() {
     return dataType;
   }
 
+  int byteSize() {
+    return byteSize;
+  }
+
   boolean isVariableLength() {
-    return dataType.byteSize < 0;
+    return byteSize < 0;
   }
 
   <T extends TType> T newInstance(TF_Tensor nativeHandle, Shape shape) {
@@ -28,13 +33,15 @@ <T extends TType> T newInstance(TF_Tensor nativeHandle, Shape shape) {
     }
   }
 
-  TensorType(Class<?> typeClass, DataType dataType, Constructor<?> implConstructor) {
+  TensorType(Class<?> typeClass, DataType dataType, int byteSize, Constructor<?> implConstructor) {
     this.typeClass = typeClass;
     this.dataType = dataType;
+    this.byteSize = byteSize;
     this.implConstructor = implConstructor;
   }
 
   private final Class<?> typeClass;
   private final DataType dataType;
+  private final int byteSize;
   private final Constructor<?> implConstructor;
 }
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/TensorTypes.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/TensorTypes.java
index 2d0acb16a50..20345aadd0f 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/TensorTypes.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/TensorTypes.java
@@ -70,7 +70,7 @@ static TensorType find(Class<? extends TType> typeClass) {
   }
 
   static int numberOf(Class<? extends TType> typeClass) {
-    return find(typeClass).dataType().number;
+    return find(typeClass).dataType().getNumber();
   }
 
   private static final Map<Integer, TensorType> TYPES_BY_CODE = new HashMap<>();
@@ -90,14 +90,14 @@ private static void register(Class<? extends TType> typeClass) {
       throw new IllegalArgumentException("Class \"" + typeClass.getName() + "\" must have a constructor "
           + "accepting a native `TF_Tensor` handle and a `Shape` to be implement as a tensor type");
     }
-    TensorType type = new TensorType(typeClass, typeAnnot.dataType(), implConstructor);
+    TensorType type = new TensorType(typeClass, typeAnnot.dataType(), typeAnnot.byteSize(), implConstructor);
     TYPES_BY_CLASS.put(typeClass, type);
 
     // If more than one tensor type is mapped to a given native code, the last registered will
     // have priority. This way, we can allow user to register their own classes to map tensors
     // of a given data type.
-    TYPES_BY_CODE.put(type.dataType().number, type);
-    TYPES_BY_CODE.put(type.dataType().number + 100, type);
+    TYPES_BY_CODE.put(type.dataType().getNumber(), type);
+    TYPES_BY_CODE.put(type.dataType().getNumber() + 100, type);
   }
 
   static {
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Tensors.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Tensors.java
index bc7468720c6..a19d39429d4 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Tensors.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Tensors.java
@@ -155,18 +155,17 @@ static Tensor<?> fromHandle(TF_Tensor handle) {
 
   private static <T extends TType> T allocate(Class<T> type, Shape shape, long size) {
     TensorType tensorType = TensorTypes.find(type);
-    DataType dataType = tensorType.dataType();
     long effectiveSize = size;
     if (effectiveSize < 0) {
       // Size of the tensor is by default the sum of the size of all its element
-      effectiveSize = shape.size() * dataType.byteSize;
+      effectiveSize = shape.size() * tensorType.byteSize();
 
-    } else if (dataType.byteSize > 0 && shape.size() * dataType.byteSize > effectiveSize) {
+    } else if (!tensorType.isVariableLength() && shape.size() * tensorType.byteSize() > effectiveSize) {
       // Minimum requirements for datatypes of variable length cannot be verified in a relevant way
       // so we only validate them for fixed length datatypes
       throw new IllegalArgumentException("Tensor size is not large enough to contain all scalar values");
     }
-    TF_Tensor nativeHandle = allocate(dataType.number, shape.asArray(), effectiveSize);
+    TF_Tensor nativeHandle = allocate(tensorType.dataType().getNumber(), shape.asArray(), effectiveSize);
     try (PointerScope scope = new PointerScope()) {
       scope.attach(nativeHandle);
       return tensorType.newInstance(nativeHandle, shape);
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/BooleanTensorImpl.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/BooleanTensorImpl.java
index dd1eb17fe57..1f9a6b37e6d 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/BooleanTensorImpl.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/BooleanTensorImpl.java
@@ -1,7 +1,6 @@
 package org.tensorflow.internal.tensor;
 
 import org.tensorflow.AbstractTensor;
-import org.tensorflow.DataType;
 import org.tensorflow.Tensor;
 import org.tensorflow.internal.c_api.TF_Tensor;
 import org.tensorflow.ndarray.BooleanNdArray;
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/ByteTensorImpl.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/ByteTensorImpl.java
index 03464538937..f479b8e0bbf 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/ByteTensorImpl.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/ByteTensorImpl.java
@@ -1,7 +1,6 @@
 package org.tensorflow.internal.tensor;
 
 import org.tensorflow.AbstractTensor;
-import org.tensorflow.DataType;
 import org.tensorflow.Tensor;
 import org.tensorflow.internal.c_api.TF_Tensor;
 import org.tensorflow.ndarray.ByteNdArray;
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/DoubleTensorImpl.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/DoubleTensorImpl.java
index d6f95e7e420..222ca373645 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/DoubleTensorImpl.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/DoubleTensorImpl.java
@@ -1,7 +1,6 @@
 package org.tensorflow.internal.tensor;
 
 import org.tensorflow.AbstractTensor;
-import org.tensorflow.DataType;
 import org.tensorflow.Tensor;
 import org.tensorflow.internal.c_api.TF_Tensor;
 import org.tensorflow.ndarray.DoubleNdArray;
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/FloatTensorImpl.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/FloatTensorImpl.java
index 0e98fd74ec6..5d9d942717e 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/FloatTensorImpl.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/FloatTensorImpl.java
@@ -1,7 +1,6 @@
 package org.tensorflow.internal.tensor;
 
 import org.tensorflow.AbstractTensor;
-import org.tensorflow.DataType;
 import org.tensorflow.internal.c_api.TF_Tensor;
 import org.tensorflow.ndarray.FloatNdArray;
 import org.tensorflow.ndarray.NdArray;
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/IntTensorImpl.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/IntTensorImpl.java
index c4f93ec07db..63b507676a2 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/IntTensorImpl.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/IntTensorImpl.java
@@ -1,7 +1,6 @@
 package org.tensorflow.internal.tensor;
 
 import org.tensorflow.AbstractTensor;
-import org.tensorflow.DataType;
 import org.tensorflow.Tensor;
 import org.tensorflow.internal.c_api.TF_Tensor;
 import org.tensorflow.ndarray.IntNdArray;
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/LongTensorImpl.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/LongTensorImpl.java
index b91435eeef3..cc235cf6e19 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/LongTensorImpl.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/LongTensorImpl.java
@@ -1,7 +1,6 @@
 package org.tensorflow.internal.tensor;
 
 import org.tensorflow.AbstractTensor;
-import org.tensorflow.DataType;
 import org.tensorflow.Tensor;
 import org.tensorflow.internal.c_api.TF_Tensor;
 import org.tensorflow.ndarray.LongNdArray;
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/StringTensorImpl.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/StringTensorImpl.java
index 62edee7cf85..306d29f3cbe 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/StringTensorImpl.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/StringTensorImpl.java
@@ -1,7 +1,6 @@
 package org.tensorflow.internal.tensor;
 
 import org.tensorflow.AbstractTensor;
-import org.tensorflow.DataType;
 import org.tensorflow.Tensor;
 import org.tensorflow.internal.c_api.TF_Tensor;
 import org.tensorflow.internal.tensor.buffer.ByteSequenceTensorBuffer;
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/op/core/Constant.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/op/core/Constant.java
index 176c57cbe6b..67b08107b95 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/op/core/Constant.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/op/core/Constant.java
@@ -16,7 +16,6 @@
 package org.tensorflow.op.core;
 
 import java.nio.charset.Charset;
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.Output;
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/op/core/Zeros.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/op/core/Zeros.java
index 09fe4a4f7a5..7a34d14d541 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/op/core/Zeros.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/op/core/Zeros.java
@@ -14,7 +14,6 @@
 ==============================================================================*/
 package org.tensorflow.op.core;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.Output;
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/op/nn/SoftmaxCrossEntropyWithLogits.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/op/nn/SoftmaxCrossEntropyWithLogits.java
index 5dae45ff17b..7d5f80d8ba4 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/op/nn/SoftmaxCrossEntropyWithLogits.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/op/nn/SoftmaxCrossEntropyWithLogits.java
@@ -1,6 +1,5 @@
 package org.tensorflow.op.nn;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.Scope;
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TBfloat16.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TBfloat16.java
index 2b4d2cef13b..bdee8e2faaf 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TBfloat16.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TBfloat16.java
@@ -18,7 +18,6 @@
 package org.tensorflow.types;
 
 import java.util.function.Consumer;
-import org.tensorflow.DataType;
 import org.tensorflow.Tensors;
 import org.tensorflow.exceptions.TensorFlowException;
 import org.tensorflow.internal.c_api.TF_Tensor;
@@ -29,6 +28,7 @@
 import org.tensorflow.ndarray.StdArrays;
 import org.tensorflow.ndarray.buffer.FloatDataBuffer;
 import org.tensorflow.ndarray.buffer.layout.DataLayouts;
+import org.tensorflow.proto.framework.DataType;
 import org.tensorflow.types.annotation.TensorType;
 import org.tensorflow.types.tensor.FloatTensor;
 import org.tensorflow.types.family.TFloating;
@@ -49,7 +49,7 @@
  * <p>Note that some CPUs support the bfloat16 format natively, which can result in faster
  * computation compared to {@link TFloat16} when GPUs are not used.
  */
-@TensorType(dataType = DataType.BFLOAT16, impl = TBfloat16Impl.class)
+@TensorType(dataType = DataType.DT_BFLOAT16, byteSize = 2, impl = TBfloat16Impl.class)
 public interface TBfloat16 extends FloatTensor, TFloating<TBfloat16, Float> {
 
   /**
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TBool.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TBool.java
index 54c4ed5262a..a15719f713b 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TBool.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TBool.java
@@ -18,7 +18,6 @@
 package org.tensorflow.types;
 
 import java.util.function.Consumer;
-import org.tensorflow.DataType;
 import org.tensorflow.Tensors;
 import org.tensorflow.exceptions.TensorFlowException;
 import org.tensorflow.internal.c_api.TF_Tensor;
@@ -29,6 +28,7 @@
 import org.tensorflow.ndarray.StdArrays;
 import org.tensorflow.ndarray.buffer.BooleanDataBuffer;
 import org.tensorflow.ndarray.buffer.layout.DataLayouts;
+import org.tensorflow.proto.framework.DataType;
 import org.tensorflow.types.annotation.TensorType;
 import org.tensorflow.types.tensor.BooleanTensor;
 import org.tensorflow.types.family.TType;
@@ -40,7 +40,7 @@
  * explicit mapping between Java boolean values and byte buffers using the {@link DataLayouts#BOOL
  * BOOL} layout, which may impact I/O performances.
  */
-@TensorType(dataType = DataType.BOOL, impl = TBoolImpl.class)
+@TensorType(dataType = DataType.DT_BOOL, byteSize = 1, impl = TBoolImpl.class)
 public interface TBool extends BooleanTensor, TType<TBool, Boolean> {
 
   /**
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TFloat16.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TFloat16.java
index a4043c2d8e3..0cc483e4724 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TFloat16.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TFloat16.java
@@ -18,7 +18,6 @@
 package org.tensorflow.types;
 
 import java.util.function.Consumer;
-import org.tensorflow.DataType;
 import org.tensorflow.Tensors;
 import org.tensorflow.exceptions.TensorFlowException;
 import org.tensorflow.internal.c_api.TF_Tensor;
@@ -29,6 +28,7 @@
 import org.tensorflow.ndarray.StdArrays;
 import org.tensorflow.ndarray.buffer.FloatDataBuffer;
 import org.tensorflow.ndarray.buffer.layout.DataLayouts;
+import org.tensorflow.proto.framework.DataType;
 import org.tensorflow.types.annotation.TensorType;
 import org.tensorflow.types.tensor.FloatTensor;
 import org.tensorflow.types.family.TFloating;
@@ -46,7 +46,7 @@
  * most CPUs do not support this format natively. For CPU computation on 16-bit floats, the {@link
  * TBfloat16} tensor type might be a better option.
  */
-@TensorType(dataType = DataType.HALF, impl = TFloat16Impl.class)
+@TensorType(dataType = DataType.DT_HALF, byteSize = 2, impl = TFloat16Impl.class)
 public interface TFloat16 extends FloatTensor, TFloating<TFloat16, Float> {
 
   /**
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TFloat32.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TFloat32.java
index d68ec86af53..da31526b71d 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TFloat32.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TFloat32.java
@@ -18,7 +18,6 @@
 package org.tensorflow.types;
 
 import java.util.function.Consumer;
-import org.tensorflow.DataType;
 import org.tensorflow.Tensors;
 import org.tensorflow.exceptions.TensorFlowException;
 import org.tensorflow.internal.c_api.TF_Tensor;
@@ -28,6 +27,7 @@
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.ndarray.StdArrays;
 import org.tensorflow.ndarray.buffer.FloatDataBuffer;
+import org.tensorflow.proto.framework.DataType;
 import org.tensorflow.types.annotation.TensorType;
 import org.tensorflow.types.tensor.FloatTensor;
 import org.tensorflow.types.family.TFloating;
@@ -35,7 +35,7 @@
 /**
  * IEEE-754 single-precision 32-bit float tensor type.
  */
-@TensorType(dataType = DataType.FLOAT, impl = TFloat32Impl.class)
+@TensorType(dataType = DataType.DT_FLOAT, byteSize = 4, impl = TFloat32Impl.class)
 public interface TFloat32 extends FloatTensor, TFloating<TFloat32, Float> {
 
   /**
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TFloat64.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TFloat64.java
index 73eb9632859..8c53ca2dfbf 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TFloat64.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TFloat64.java
@@ -18,7 +18,6 @@
 package org.tensorflow.types;
 
 import java.util.function.Consumer;
-import org.tensorflow.DataType;
 import org.tensorflow.Tensors;
 import org.tensorflow.exceptions.TensorFlowException;
 import org.tensorflow.internal.c_api.TF_Tensor;
@@ -28,6 +27,7 @@
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.ndarray.StdArrays;
 import org.tensorflow.ndarray.buffer.DoubleDataBuffer;
+import org.tensorflow.proto.framework.DataType;
 import org.tensorflow.types.annotation.TensorType;
 import org.tensorflow.types.tensor.DoubleTensor;
 import org.tensorflow.types.family.TFloating;
@@ -35,7 +35,7 @@
 /**
  * IEEE-754 double-precision 64-bit float tensor type.
  */
-@TensorType(dataType = DataType.DOUBLE, impl = TFloat64Impl.class)
+@TensorType(dataType = DataType.DT_DOUBLE, byteSize = 8, impl = TFloat64Impl.class)
 public interface TFloat64 extends DoubleTensor, TFloating<TFloat64, Double> {
 
   /**
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TInt32.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TInt32.java
index ed68aaaa9db..657cc134c81 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TInt32.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TInt32.java
@@ -18,7 +18,6 @@
 package org.tensorflow.types;
 
 import java.util.function.Consumer;
-import org.tensorflow.DataType;
 import org.tensorflow.Tensors;
 import org.tensorflow.internal.c_api.TF_Tensor;
 import org.tensorflow.internal.tensor.IntTensorImpl;
@@ -27,6 +26,7 @@
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.ndarray.StdArrays;
 import org.tensorflow.ndarray.buffer.IntDataBuffer;
+import org.tensorflow.proto.framework.DataType;
 import org.tensorflow.types.annotation.TensorType;
 import org.tensorflow.types.tensor.IntTensor;
 import org.tensorflow.types.family.TNumber;
@@ -34,7 +34,7 @@
 /**
  * 32-bit signed integer tensor type.
  */
-@TensorType(dataType = DataType.INT32, impl = TInt32Impl.class)
+@TensorType(dataType = DataType.DT_INT32, byteSize = 4, impl = TInt32Impl.class)
 public interface TInt32 extends IntTensor, TNumber<TInt32, Integer> {
 
   /**
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TInt64.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TInt64.java
index 97e83c34a6d..5758eb7f4df 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TInt64.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TInt64.java
@@ -18,7 +18,6 @@
 package org.tensorflow.types;
 
 import java.util.function.Consumer;
-import org.tensorflow.DataType;
 import org.tensorflow.Tensors;
 import org.tensorflow.exceptions.TensorFlowException;
 import org.tensorflow.internal.c_api.TF_Tensor;
@@ -28,6 +27,7 @@
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.ndarray.StdArrays;
 import org.tensorflow.ndarray.buffer.LongDataBuffer;
+import org.tensorflow.proto.framework.DataType;
 import org.tensorflow.types.annotation.TensorType;
 import org.tensorflow.types.tensor.LongTensor;
 import org.tensorflow.types.family.TNumber;
@@ -35,7 +35,7 @@
 /**
  * 64-bit signed integer tensor type.
  */
-@TensorType(dataType = DataType.INT64, impl = TInt64Impl.class)
+@TensorType(dataType = DataType.DT_INT64, byteSize = 8, impl = TInt64Impl.class)
 public interface TInt64 extends LongTensor, TNumber<TInt64, Long> {
 
   /**
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TString.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TString.java
index 948833a5b71..44cce1363de 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TString.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TString.java
@@ -20,7 +20,6 @@
 import java.nio.charset.Charset;
 import java.nio.charset.StandardCharsets;
 import java.util.function.Function;
-import org.tensorflow.DataType;
 import org.tensorflow.Tensors;
 import org.tensorflow.internal.c_api.TF_Tensor;
 import org.tensorflow.internal.tensor.StringTensorImpl;
@@ -32,6 +31,7 @@
 import org.tensorflow.ndarray.buffer.DataBuffer;
 import org.tensorflow.ndarray.buffer.layout.DataLayout;
 import org.tensorflow.ndarray.buffer.layout.DataLayouts;
+import org.tensorflow.proto.framework.DataType;
 import org.tensorflow.types.annotation.TensorType;
 import org.tensorflow.types.tensor.StringTensor;
 import org.tensorflow.types.family.TType;
@@ -45,7 +45,7 @@
  * its values initially, so TensorFlow can compute and allocate the right amount of memory. Then the
  * data in the tensor is initialized once and cannot be modified afterwards.
  */
-@TensorType(dataType = DataType.STRING, impl = TStringImpl.class)
+@TensorType(dataType = DataType.DT_STRING, byteSize = -1, impl = TStringImpl.class)
 public interface TString extends StringTensor, TType<TString, String> {
 
   /**
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TUint8.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TUint8.java
index 048826d6fc3..145f66de745 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TUint8.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TUint8.java
@@ -18,7 +18,6 @@
 package org.tensorflow.types;
 
 import java.util.function.Consumer;
-import org.tensorflow.DataType;
 import org.tensorflow.Tensors;
 import org.tensorflow.exceptions.TensorFlowException;
 import org.tensorflow.internal.c_api.TF_Tensor;
@@ -28,6 +27,7 @@
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.ndarray.StdArrays;
 import org.tensorflow.ndarray.buffer.ByteDataBuffer;
+import org.tensorflow.proto.framework.DataType;
 import org.tensorflow.types.annotation.TensorType;
 import org.tensorflow.types.tensor.ByteTensor;
 import org.tensorflow.types.family.TNumber;
@@ -35,7 +35,7 @@
 /**
  * 8-bit unsigned integer tensor type.
  */
-@TensorType(dataType = DataType.UINT8, impl = TUint8Impl.class)
+@TensorType(dataType = DataType.DT_UINT8, byteSize = 1, impl = TUint8Impl.class)
 public interface TUint8 extends ByteTensor, TNumber<TUint8, Byte> {
 
   /**
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/annotation/TensorType.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/annotation/TensorType.java
index c338cd0ec6d..18130ae7377 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/annotation/TensorType.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/annotation/TensorType.java
@@ -15,14 +15,16 @@
 
 package org.tensorflow.types.annotation;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Tensor;
+import org.tensorflow.proto.framework.DataType;
 
 /** Represents a type of elements in a {@link Tensor} */
 public @interface TensorType {
 
   DataType dataType();
 
+  int byteSize();
+
   /**
    * The class implementing this tensor type
    */

From 811ffbe0b0de017f89d9d322b6ec808d97232af1 Mon Sep 17 00:00:00 2001
From: klessard <klessard@expedia.com>
Date: Tue, 27 Oct 2020 21:52:33 -0400
Subject: [PATCH 07/11] Create type factories

---
 .../src/bazel/op_generator/op_generator.cc    |    9 +-
 .../src/bazel/op_generator/op_specs.cc        |   32 +-
 ...mpute_func_Pointer_TF_OpKernelContext.java |   19 -
 .../Create_func_TF_OpKernelConstruction.java  |   48 -
 .../Deallocator_Pointer_long_Pointer.java     |   30 -
 .../internal/c_api/Delete_func_Pointer.java   |   19 -
 .../internal/c_api/Listener_BytePointer.java  |   25 -
 .../internal/c_api/Listener_String.java       |   19 -
 ...nc_TF_ShapeInferenceContext_TF_Status.java |   22 -
 .../internal/c_api/TFE_Context.java           |   25 -
 .../internal/c_api/TFE_ContextOptions.java    |   19 -
 .../org/tensorflow/internal/c_api/TFE_Op.java |   29 -
 .../internal/c_api/TFE_TensorDebugInfo.java   |   22 -
 .../internal/c_api/TFE_TensorHandle.java      |   23 -
 .../internal/c_api/TF_ApiDefMap.java          |   26 -
 .../internal/c_api/TF_AttrMetadata.java       |   58 -
 .../tensorflow/internal/c_api/TF_Buffer.java  |   49 -
 .../internal/c_api/TF_DeprecatedSession.java  |   23 -
 .../internal/c_api/TF_DeviceList.java         |   18 -
 .../internal/c_api/TF_DimensionHandle.java    |   19 -
 .../internal/c_api/TF_Function.java           |   21 -
 .../internal/c_api/TF_FunctionOptions.java    |   19 -
 .../tensorflow/internal/c_api/TF_Graph.java   |   26 -
 .../c_api/TF_ImportGraphDefOptions.java       |   20 -
 .../c_api/TF_ImportGraphDefResults.java       |   20 -
 .../tensorflow/internal/c_api/TF_Input.java   |   33 -
 .../internal/c_api/TF_KernelBuilder.java      |   31 -
 .../tensorflow/internal/c_api/TF_Library.java |   22 -
 .../c_api/TF_OpDefinitionBuilder.java         |   17 -
 .../c_api/TF_OpKernelConstruction.java        |   17 -
 .../internal/c_api/TF_OpKernelContext.java    |   17 -
 .../internal/c_api/TF_Operation.java          |   21 -
 .../c_api/TF_OperationDescription.java        |   19 -
 .../tensorflow/internal/c_api/TF_Output.java  |   33 -
 .../tensorflow/internal/c_api/TF_Server.java  |   27 -
 .../tensorflow/internal/c_api/TF_Session.java |   24 -
 .../internal/c_api/TF_SessionOptions.java     |   20 -
 .../internal/c_api/TF_ShapeHandle.java        |   17 -
 .../c_api/TF_ShapeInferenceContext.java       |   17 -
 .../tensorflow/internal/c_api/TF_Status.java  |   19 -
 .../tensorflow/internal/c_api/TF_Tensor.java  |   36 -
 .../internal/c_api/TF_WhileParams.java        |   37 -
 .../org/tensorflow/internal/c_api/Tensor.java |   25 -
 .../internal/c_api/global/tensorflow.java     | 4021 -----------
 .../tensorflow/op/audio/AudioSpectrogram.java |  140 -
 .../org/tensorflow/op/audio/DecodeWav.java    |  147 -
 .../org/tensorflow/op/audio/EncodeWav.java    |   85 -
 .../java/org/tensorflow/op/audio/Mfcc.java    |  178 -
 .../org/tensorflow/op/bitwise/BitwiseAnd.java |   96 -
 .../org/tensorflow/op/bitwise/BitwiseOr.java  |   96 -
 .../org/tensorflow/op/bitwise/BitwiseXor.java |   96 -
 .../org/tensorflow/op/bitwise/Invert.java     |  115 -
 .../org/tensorflow/op/bitwise/LeftShift.java  |  107 -
 .../org/tensorflow/op/bitwise/RightShift.java |  110 -
 .../op/cluster/KMC2ChainInitialization.java   |   81 -
 .../cluster/KmeansPlusPlusInitialization.java |   87 -
 .../tensorflow/op/collective/AllReduce.java   |  166 -
 .../op/collective/BroadcastRecv.java          |  135 -
 .../op/collective/BroadcastSend.java          |  135 -
 .../java/org/tensorflow/op/core/Abort.java    |  110 -
 .../gen/java/org/tensorflow/op/core/All.java  |  116 -
 .../gen/java/org/tensorflow/op/core/Any.java  |  116 -
 .../org/tensorflow/op/core/AssertThat.java    |   96 -
 .../java/org/tensorflow/op/core/Assign.java   |  140 -
 .../org/tensorflow/op/core/AssignAdd.java     |  117 -
 .../op/core/AssignAddVariableOp.java          |   61 -
 .../org/tensorflow/op/core/AssignSub.java     |  117 -
 .../op/core/AssignSubVariableOp.java          |   61 -
 .../tensorflow/op/core/AssignVariableOp.java  |   61 -
 .../java/org/tensorflow/op/core/Barrier.java  |  193 -
 .../org/tensorflow/op/core/BarrierClose.java  |  101 -
 .../op/core/BarrierIncompleteSize.java        |   75 -
 .../tensorflow/op/core/BarrierInsertMany.java |   69 -
 .../tensorflow/op/core/BarrierReadySize.java  |   75 -
 .../tensorflow/op/core/BarrierTakeMany.java   |  190 -
 .../java/org/tensorflow/op/core/Batch.java    |  247 -
 .../org/tensorflow/op/core/BatchToSpace.java  |  147 -
 .../tensorflow/op/core/BatchToSpaceNd.java    |  181 -
 .../java/org/tensorflow/op/core/Bitcast.java  |  127 -
 .../op/core/BroadcastDynamicShape.java        |   79 -
 .../op/core/BroadcastGradientArgs.java        |   80 -
 .../org/tensorflow/op/core/BroadcastTo.java   |  106 -
 .../org/tensorflow/op/core/Bucketize.java     |   96 -
 .../org/tensorflow/op/core/ClipByValue.java   |   86 -
 .../tensorflow/op/core/CollectiveGather.java  |  135 -
 .../java/org/tensorflow/op/core/Concat.java   |   83 -
 .../tensorflow/op/core/ConsumeMutexLock.java  |   63 -
 .../tensorflow/op/core/ControlTrigger.java    |   54 -
 .../gen/java/org/tensorflow/op/core/Copy.java |  148 -
 .../java/org/tensorflow/op/core/CopyHost.java |  146 -
 .../org/tensorflow/op/core/CountUpTo.java     |   79 -
 .../org/tensorflow/op/core/DecodeProto.java   |  222 -
 .../java/org/tensorflow/op/core/DeepCopy.java |   76 -
 .../op/core/DeleteSessionTensor.java          |   56 -
 .../tensorflow/op/core/DestroyResourceOp.java |   94 -
 .../op/core/DestroyTemporaryVariable.java     |   85 -
 .../org/tensorflow/op/core/DeviceIndex.java   |   81 -
 .../tensorflow/op/core/DummyMemoryCache.java  |   69 -
 .../tensorflow/op/core/DynamicPartition.java  |  120 -
 .../org/tensorflow/op/core/DynamicStitch.java |  133 -
 .../org/tensorflow/op/core/EditDistance.java  |  164 -
 .../java/org/tensorflow/op/core/Empty.java    |  114 -
 .../tensorflow/op/core/EmptyTensorList.java   |   86 -
 .../org/tensorflow/op/core/EncodeProto.java   |  160 -
 .../org/tensorflow/op/core/EnsureShape.java   |   81 -
 .../java/org/tensorflow/op/core/Enter.java    |  135 -
 .../gen/java/org/tensorflow/op/core/Exit.java |   76 -
 .../org/tensorflow/op/core/ExpandDims.java    |  110 -
 .../op/core/ExtractVolumePatches.java         |  102 -
 .../gen/java/org/tensorflow/op/core/Fill.java |  104 -
 .../org/tensorflow/op/core/Fingerprint.java   |  109 -
 .../java/org/tensorflow/op/core/Gather.java   |  143 -
 .../java/org/tensorflow/op/core/GatherNd.java |  172 -
 .../tensorflow/op/core/GetSessionHandle.java  |   75 -
 .../tensorflow/op/core/GetSessionTensor.java  |   78 -
 .../tensorflow/op/core/GuaranteeConst.java    |   81 -
 .../org/tensorflow/op/core/HashTable.java     |  158 -
 .../op/core/HistogramFixedWidth.java          |  116 -
 .../java/org/tensorflow/op/core/Identity.java |   74 -
 .../org/tensorflow/op/core/IdentityN.java     |   95 -
 .../tensorflow/op/core/ImmutableConst.java    |   82 -
 .../tensorflow/op/core/InitializeTable.java   |   60 -
 .../op/core/InitializeTableFromTextFile.java  |  127 -
 .../org/tensorflow/op/core/InplaceAdd.java    |   82 -
 .../org/tensorflow/op/core/InplaceSub.java    |   82 -
 .../org/tensorflow/op/core/InplaceUpdate.java |   85 -
 .../op/core/IsVariableInitialized.java        |   75 -
 .../java/org/tensorflow/op/core/LinSpace.java |   88 -
 .../tensorflow/op/core/LookupTableExport.java |   84 -
 .../tensorflow/op/core/LookupTableFind.java   |   86 -
 .../tensorflow/op/core/LookupTableImport.java |   63 -
 .../tensorflow/op/core/LookupTableInsert.java |   63 -
 .../tensorflow/op/core/LookupTableRemove.java |   60 -
 .../tensorflow/op/core/LookupTableSize.java   |   73 -
 .../java/org/tensorflow/op/core/LoopCond.java |   76 -
 .../org/tensorflow/op/core/LowerBound.java    |  115 -
 .../java/org/tensorflow/op/core/MapClear.java |  150 -
 .../tensorflow/op/core/MapIncompleteSize.java |  168 -
 .../java/org/tensorflow/op/core/MapPeek.java  |  182 -
 .../java/org/tensorflow/op/core/MapSize.java  |  168 -
 .../java/org/tensorflow/op/core/MapStage.java |  165 -
 .../org/tensorflow/op/core/MapUnstage.java    |  182 -
 .../tensorflow/op/core/MapUnstageNoKey.java   |  180 -
 .../gen/java/org/tensorflow/op/core/Max.java  |  118 -
 .../java/org/tensorflow/op/core/Merge.java    |   87 -
 .../gen/java/org/tensorflow/op/core/Min.java  |  118 -
 .../org/tensorflow/op/core/MirrorPad.java     |  111 -
 .../org/tensorflow/op/core/MirrorPadGrad.java |   99 -
 .../tensorflow/op/core/MlirPassthroughOp.java |  117 -
 .../op/core/MutableDenseHashTable.java        |  224 -
 .../tensorflow/op/core/MutableHashTable.java  |  158 -
 .../op/core/MutableHashTableOfTensors.java    |  176 -
 .../java/org/tensorflow/op/core/Mutex.java    |  129 -
 .../org/tensorflow/op/core/MutexLock.java     |  112 -
 .../org/tensorflow/op/core/NcclAllReduce.java |   92 -
 .../org/tensorflow/op/core/NcclBroadcast.java |   86 -
 .../org/tensorflow/op/core/NcclReduce.java    |   85 -
 .../org/tensorflow/op/core/NextIteration.java |   75 -
 .../gen/java/org/tensorflow/op/core/NoOp.java |   52 -
 .../java/org/tensorflow/op/core/OneHot.java   |  198 -
 .../java/org/tensorflow/op/core/OnesLike.java |   75 -
 .../tensorflow/op/core/OrderedMapClear.java   |  150 -
 .../op/core/OrderedMapIncompleteSize.java     |  168 -
 .../tensorflow/op/core/OrderedMapPeek.java    |  183 -
 .../tensorflow/op/core/OrderedMapSize.java    |  168 -
 .../tensorflow/op/core/OrderedMapStage.java   |  167 -
 .../tensorflow/op/core/OrderedMapUnstage.java |  182 -
 .../op/core/OrderedMapUnstageNoKey.java       |  180 -
 .../gen/java/org/tensorflow/op/core/Pad.java  |  104 -
 .../tensorflow/op/core/ParallelConcat.java    |   96 -
 .../op/core/ParallelDynamicStitch.java        |  132 -
 .../org/tensorflow/op/core/Placeholder.java   |  116 -
 .../op/core/PlaceholderWithDefault.java       |   78 -
 .../java/org/tensorflow/op/core/Print.java    |  111 -
 .../gen/java/org/tensorflow/op/core/Prod.java |  118 -
 .../tensorflow/op/core/QuantizedReshape.java  |   97 -
 .../java/org/tensorflow/op/core/Range.java    |   91 -
 .../gen/java/org/tensorflow/op/core/Rank.java |   85 -
 .../tensorflow/op/core/ReadVariableOp.java    |   83 -
 .../gen/java/org/tensorflow/op/core/Recv.java |  122 -
 .../org/tensorflow/op/core/ReduceAll.java     |  116 -
 .../org/tensorflow/op/core/ReduceAny.java     |  116 -
 .../org/tensorflow/op/core/ReduceMax.java     |  118 -
 .../org/tensorflow/op/core/ReduceMin.java     |  118 -
 .../org/tensorflow/op/core/ReduceProd.java    |  118 -
 .../org/tensorflow/op/core/ReduceSum.java     |  118 -
 .../java/org/tensorflow/op/core/RefEnter.java |  134 -
 .../java/org/tensorflow/op/core/RefExit.java  |   76 -
 .../org/tensorflow/op/core/RefIdentity.java   |   73 -
 .../java/org/tensorflow/op/core/RefMerge.java |   86 -
 .../tensorflow/op/core/RefNextIteration.java  |   75 -
 .../org/tensorflow/op/core/RefSelect.java     |   79 -
 .../org/tensorflow/op/core/RefSwitch.java     |   87 -
 .../op/core/RemoteFusedGraphExecute.java      |   97 -
 .../java/org/tensorflow/op/core/Reshape.java  |  137 -
 .../tensorflow/op/core/ResourceCountUpTo.java |   81 -
 .../tensorflow/op/core/ResourceGather.java    |  146 -
 .../tensorflow/op/core/ResourceGatherNd.java  |   77 -
 .../op/core/ResourceScatterAdd.java           |   81 -
 .../op/core/ResourceScatterDiv.java           |   81 -
 .../op/core/ResourceScatterMax.java           |   81 -
 .../op/core/ResourceScatterMin.java           |   81 -
 .../op/core/ResourceScatterMul.java           |   81 -
 .../op/core/ResourceScatterNdAdd.java         |  131 -
 .../op/core/ResourceScatterNdMax.java         |  100 -
 .../op/core/ResourceScatterNdMin.java         |  100 -
 .../op/core/ResourceScatterNdSub.java         |  131 -
 .../op/core/ResourceScatterNdUpdate.java      |  133 -
 .../op/core/ResourceScatterSub.java           |   81 -
 .../op/core/ResourceScatterUpdate.java        |   72 -
 .../op/core/ResourceStridedSliceAssign.java   |  182 -
 .../java/org/tensorflow/op/core/Reverse.java  |  126 -
 .../tensorflow/op/core/ReverseSequence.java   |  168 -
 .../gen/java/org/tensorflow/op/core/Roll.java |  109 -
 .../gen/java/org/tensorflow/op/core/Rpc.java  |  209 -
 .../org/tensorflow/op/core/ScatterAdd.java    |  140 -
 .../org/tensorflow/op/core/ScatterDiv.java    |  136 -
 .../org/tensorflow/op/core/ScatterMax.java    |  139 -
 .../org/tensorflow/op/core/ScatterMin.java    |  139 -
 .../org/tensorflow/op/core/ScatterMul.java    |  136 -
 .../org/tensorflow/op/core/ScatterNd.java     |  158 -
 .../org/tensorflow/op/core/ScatterNdAdd.java  |  151 -
 .../org/tensorflow/op/core/ScatterNdMax.java  |  120 -
 .../org/tensorflow/op/core/ScatterNdMin.java  |  120 -
 .../op/core/ScatterNdNonAliasingAdd.java      |  117 -
 .../org/tensorflow/op/core/ScatterNdSub.java  |  153 -
 .../tensorflow/op/core/ScatterNdUpdate.java   |  155 -
 .../org/tensorflow/op/core/ScatterSub.java    |  139 -
 .../org/tensorflow/op/core/ScatterUpdate.java |  143 -
 .../java/org/tensorflow/op/core/Select.java   |   77 -
 .../gen/java/org/tensorflow/op/core/Send.java |  103 -
 .../org/tensorflow/op/core/SetDiff1d.java     |  119 -
 .../java/org/tensorflow/op/core/SetSize.java  |  122 -
 .../java/org/tensorflow/op/core/Shape.java    |   99 -
 .../java/org/tensorflow/op/core/ShapeN.java   |  100 -
 .../gen/java/org/tensorflow/op/core/Size.java |  100 -
 .../java/org/tensorflow/op/core/Skipgram.java |  201 -
 .../java/org/tensorflow/op/core/Slice.java    |   90 -
 .../java/org/tensorflow/op/core/Snapshot.java |   74 -
 .../tensorflow/op/core/SpaceToBatchNd.java    |  180 -
 .../java/org/tensorflow/op/core/Split.java    |   91 -
 .../java/org/tensorflow/op/core/SplitV.java   |   95 -
 .../java/org/tensorflow/op/core/Squeeze.java  |  136 -
 .../java/org/tensorflow/op/core/Stack.java    |  130 -
 .../java/org/tensorflow/op/core/Stage.java    |  157 -
 .../org/tensorflow/op/core/StageClear.java    |  150 -
 .../org/tensorflow/op/core/StagePeek.java     |  180 -
 .../org/tensorflow/op/core/StageSize.java     |  168 -
 .../org/tensorflow/op/core/StopGradient.java  |   99 -
 .../org/tensorflow/op/core/StridedSlice.java  |  317 -
 .../op/core/StridedSliceAssign.java           |  200 -
 .../tensorflow/op/core/StridedSliceGrad.java  |  202 -
 .../gen/java/org/tensorflow/op/core/Sum.java  |  118 -
 .../org/tensorflow/op/core/SwitchCond.java    |   87 -
 .../tensorflow/op/core/TemporaryVariable.java |  129 -
 .../org/tensorflow/op/core/TensorArray.java   |  218 -
 .../tensorflow/op/core/TensorArrayClose.java  |   58 -
 .../tensorflow/op/core/TensorArrayConcat.java |  140 -
 .../tensorflow/op/core/TensorArrayGather.java |  125 -
 .../tensorflow/op/core/TensorArrayGrad.java   |  117 -
 .../op/core/TensorArrayGradWithShape.java     |   90 -
 .../tensorflow/op/core/TensorArrayPack.java   |  113 -
 .../tensorflow/op/core/TensorArrayRead.java   |   83 -
 .../op/core/TensorArrayScatter.java           |   83 -
 .../tensorflow/op/core/TensorArraySize.java   |   76 -
 .../tensorflow/op/core/TensorArraySplit.java  |  104 -
 .../tensorflow/op/core/TensorArrayUnpack.java |   77 -
 .../tensorflow/op/core/TensorArrayWrite.java  |   81 -
 .../core/TensorForestCreateTreeVariable.java  |   57 -
 .../op/core/TensorForestTreeDeserialize.java  |   57 -
 .../core/TensorForestTreeIsInitializedOp.java |   72 -
 .../op/core/TensorForestTreePredict.java      |   76 -
 .../TensorForestTreeResourceHandleOp.java     |  123 -
 .../op/core/TensorForestTreeSerialize.java    |   72 -
 .../op/core/TensorForestTreeSize.java         |   72 -
 .../tensorflow/op/core/TensorListConcat.java  |   98 -
 .../op/core/TensorListConcatLists.java        |   76 -
 .../op/core/TensorListElementShape.java       |   79 -
 .../op/core/TensorListFromTensor.java         |   81 -
 .../tensorflow/op/core/TensorListGather.java  |   88 -
 .../tensorflow/op/core/TensorListGetItem.java |   79 -
 .../tensorflow/op/core/TensorListLength.java  |   75 -
 .../tensorflow/op/core/TensorListPopBack.java |   89 -
 .../op/core/TensorListPushBack.java           |   81 -
 .../op/core/TensorListPushBackBatch.java      |   74 -
 .../tensorflow/op/core/TensorListReserve.java |   84 -
 .../tensorflow/op/core/TensorListResize.java  |   81 -
 .../tensorflow/op/core/TensorListScatter.java |   93 -
 .../TensorListScatterIntoExistingList.java    |   86 -
 .../tensorflow/op/core/TensorListSetItem.java |   77 -
 .../tensorflow/op/core/TensorListSplit.java   |   87 -
 .../tensorflow/op/core/TensorListStack.java   |  120 -
 .../op/core/TensorScatterNdAdd.java           |  140 -
 .../op/core/TensorScatterNdMax.java           |   78 -
 .../op/core/TensorScatterNdMin.java           |   78 -
 .../op/core/TensorScatterNdSub.java           |  139 -
 .../op/core/TensorScatterNdUpdate.java        |  155 -
 .../op/core/TensorStridedSliceUpdate.java     |  200 -
 .../gen/java/org/tensorflow/op/core/Tile.java |  104 -
 .../org/tensorflow/op/core/Timestamp.java     |   75 -
 .../java/org/tensorflow/op/core/TryRpc.java   |  226 -
 .../java/org/tensorflow/op/core/Unbatch.java  |  153 -
 .../org/tensorflow/op/core/UnbatchGrad.java   |  149 -
 .../java/org/tensorflow/op/core/Unique.java   |  142 -
 .../tensorflow/op/core/UniqueWithCounts.java  |  155 -
 .../org/tensorflow/op/core/UnravelIndex.java  |   99 -
 .../java/org/tensorflow/op/core/Unstack.java  |  133 -
 .../java/org/tensorflow/op/core/Unstage.java  |  176 -
 .../org/tensorflow/op/core/UpperBound.java    |  115 -
 .../org/tensorflow/op/core/VarHandleOp.java   |  156 -
 .../op/core/VarIsInitializedOp.java           |   74 -
 .../java/org/tensorflow/op/core/Variable.java |  139 -
 .../org/tensorflow/op/core/VariableShape.java |   98 -
 .../java/org/tensorflow/op/core/Where.java    |  133 -
 .../op/core/XlaSpmdFullToShardShape.java      |   81 -
 .../op/core/XlaSpmdShardToFullShape.java      |   83 -
 .../org/tensorflow/op/core/ZerosLike.java     |   75 -
 .../tensorflow/op/data/AnonymousIterator.java |   90 -
 .../op/data/AnonymousMemoryCache.java         |   69 -
 .../op/data/AnonymousMultiDeviceIterator.java |   95 -
 .../tensorflow/op/data/AssertNextDataset.java |  100 -
 .../tensorflow/op/data/AutoShardDataset.java  |  133 -
 .../org/tensorflow/op/data/BatchDataset.java  |  128 -
 .../op/data/BytesProducedStatsDataset.java    |   89 -
 .../org/tensorflow/op/data/CSVDataset.java    |  100 -
 .../org/tensorflow/op/data/CacheDataset.java  |   95 -
 .../tensorflow/op/data/CacheDatasetV2.java    |   90 -
 .../op/data/ChooseFastestDataset.java         |   88 -
 .../op/data/ConcatenateDataset.java           |   89 -
 .../op/data/DatasetCardinality.java           |   75 -
 .../tensorflow/op/data/DatasetFromGraph.java  |   76 -
 .../tensorflow/op/data/DatasetToGraph.java    |  127 -
 .../op/data/DatasetToSingleElement.java       |   91 -
 .../tensorflow/op/data/DatasetToTfRecord.java |   60 -
 .../tensorflow/op/data/DeleteIterator.java    |   57 -
 .../tensorflow/op/data/DeleteMemoryCache.java |   55 -
 .../op/data/DeleteMultiDeviceIterator.java    |   59 -
 .../op/data/DenseToSparseBatchDataset.java    |   94 -
 .../op/data/DeserializeIterator.java          |   58 -
 .../op/data/DirectedInterleaveDataset.java    |   91 -
 .../op/data/FilterByLastComponentDataset.java |   86 -
 .../op/data/FixedLengthRecordDataset.java     |   83 -
 .../op/data/IgnoreErrorsDataset.java          |   86 -
 .../op/data/InitializeTableFromDataset.java   |   55 -
 .../java/org/tensorflow/op/data/Iterator.java |   88 -
 .../op/data/IteratorFromStringHandle.java     |  118 -
 .../tensorflow/op/data/IteratorGetDevice.java |   71 -
 .../tensorflow/op/data/IteratorGetNext.java   |   91 -
 .../op/data/IteratorGetNextAsOptional.java    |   87 -
 .../op/data/IteratorGetNextSync.java          |   96 -
 .../op/data/IteratorToStringHandle.java       |   73 -
 .../org/tensorflow/op/data/LMDBDataset.java   |   99 -
 .../op/data/LatencyStatsDataset.java          |   89 -
 .../org/tensorflow/op/data/LeakyReluGrad.java |  111 -
 .../org/tensorflow/op/data/MakeIterator.java  |   60 -
 .../op/data/MatchingFilesDataset.java         |   72 -
 .../op/data/MaxIntraOpParallelismDataset.java |   89 -
 .../org/tensorflow/op/data/ModelDataset.java  |  141 -
 .../op/data/MultiDeviceIterator.java          |   97 -
 .../MultiDeviceIteratorFromStringHandle.java  |  120 -
 .../MultiDeviceIteratorGetNextFromShard.java  |   97 -
 .../op/data/MultiDeviceIteratorInit.java      |   77 -
 .../MultiDeviceIteratorToStringHandle.java    |   72 -
 .../op/data/NonSerializableDataset.java       |   85 -
 .../tensorflow/op/data/OptimizeDataset.java   |  129 -
 .../tensorflow/op/data/OptionalFromValue.java |   74 -
 .../tensorflow/op/data/OptionalGetValue.java  |   91 -
 .../tensorflow/op/data/OptionalHasValue.java  |   72 -
 .../org/tensorflow/op/data/OptionalNone.java  |   71 -
 .../op/data/PaddedBatchDataset.java           |  131 -
 .../tensorflow/op/data/PrefetchDataset.java   |  143 -
 .../op/data/PrivateThreadPoolDataset.java     |   89 -
 .../org/tensorflow/op/data/RandomDataset.java |  102 -
 .../org/tensorflow/op/data/RangeDataset.java  |   92 -
 .../tensorflow/op/data/RebatchDataset.java    |  128 -
 .../tensorflow/op/data/RegisterDataset.java   |   78 -
 .../org/tensorflow/op/data/RepeatDataset.java |   91 -
 .../tensorflow/op/data/SamplingDataset.java   |  101 -
 .../tensorflow/op/data/SerializeIterator.java |  109 -
 .../op/data/SetStatsAggregatorDataset.java    |   92 -
 .../org/tensorflow/op/data/ShardDataset.java  |  125 -
 .../op/data/ShuffleAndRepeatDataset.java      |  130 -
 .../tensorflow/op/data/ShuffleDataset.java    |  128 -
 .../org/tensorflow/op/data/SkipDataset.java   |   91 -
 .../org/tensorflow/op/data/SleepDataset.java  |   88 -
 .../op/data/SlidingWindowDataset.java         |   96 -
 .../op/data/SparseTensorSliceDataset.java     |   77 -
 .../org/tensorflow/op/data/SqlDataset.java    |   91 -
 .../op/data/StatsAggregatorHandle.java        |  123 -
 .../org/tensorflow/op/data/TakeDataset.java   |   92 -
 .../org/tensorflow/op/data/TensorDataset.java |   81 -
 .../op/data/TensorSliceDataset.java           |   82 -
 .../tensorflow/op/data/TextLineDataset.java   |   81 -
 .../tensorflow/op/data/TfRecordDataset.java   |   82 -
 .../tensorflow/op/data/ThreadPoolDataset.java |   88 -
 .../tensorflow/op/data/ThreadPoolHandle.java  |  152 -
 .../tensorflow/op/data/UnbatchDataset.java    |   86 -
 .../org/tensorflow/op/data/UniqueDataset.java |   86 -
 .../op/data/UnwrapDatasetVariant.java         |   71 -
 .../org/tensorflow/op/data/WindowDataset.java |  143 -
 .../op/data/WrapDatasetVariant.java           |   71 -
 .../org/tensorflow/op/data/ZipDataset.java    |   94 -
 .../AssertCardinalityDataset.java             |   88 -
 .../data/experimental/AssertNextDataset.java  |   88 -
 .../data/experimental/AutoShardDataset.java   |  133 -
 .../BytesProducedStatsDataset.java            |   89 -
 .../op/data/experimental/CSVDataset.java      |   99 -
 .../experimental/ChooseFastestDataset.java    |   88 -
 .../op/data/experimental/CompressElement.java |   73 -
 .../data/experimental/DataServiceDataset.java |  134 -
 .../data/experimental/DatasetCardinality.java |   75 -
 .../data/experimental/DatasetToTFRecord.java  |   60 -
 .../DenseToSparseBatchDataset.java            |   94 -
 .../DirectedInterleaveDataset.java            |   91 -
 .../experimental/DummyIterationCounter.java   |   69 -
 .../experimental/IgnoreErrorsDataset.java     |   86 -
 .../data/experimental/IteratorGetDevice.java  |   71 -
 .../experimental/LatencyStatsDataset.java     |   89 -
 .../op/data/experimental/LmdbDataset.java     |   86 -
 .../experimental/MatchingFilesDataset.java    |   72 -
 .../MaxIntraOpParallelismDataset.java         |   89 -
 .../experimental/NonSerializableDataset.java  |   85 -
 .../experimental/ParseExampleDataset.java     |  206 -
 .../PrivateThreadPoolDataset.java             |   89 -
 .../op/data/experimental/RandomDataset.java   |   91 -
 .../op/data/experimental/RebatchDataset.java  |  128 -
 .../SetStatsAggregatorDataset.java            |   92 -
 .../op/data/experimental/SleepDataset.java    |   88 -
 .../experimental/SlidingWindowDataset.java    |   96 -
 .../op/data/experimental/SqlDataset.java      |   91 -
 .../experimental/StatsAggregatorHandle.java   |  122 -
 .../StatsAggregatorSetSummaryWriter.java      |   56 -
 .../experimental/StatsAggregatorSummary.java  |   71 -
 .../data/experimental/ThreadPoolDataset.java  |   88 -
 .../data/experimental/ThreadPoolHandle.java   |  152 -
 .../op/data/experimental/UnbatchDataset.java  |   86 -
 .../data/experimental/UncompressElement.java  |   90 -
 .../op/data/experimental/UniqueDataset.java   |   86 -
 .../op/debugging/CheckNumerics.java           |   80 -
 .../op/debugging/DebugGradientIdentity.java   |   77 -
 .../debugging/DebugGradientRefIdentity.java   |   77 -
 .../op/debugging/DebugIdentity.java           |  242 -
 .../op/debugging/DebugNanCount.java           |  182 -
 .../op/debugging/DebugNumericsSummary.java    |  253 -
 .../org/tensorflow/op/dtypes/AsString.java    |  206 -
 .../java/org/tensorflow/op/dtypes/Cast.java   |  110 -
 .../org/tensorflow/op/dtypes/Complex.java     |   94 -
 .../java/org/tensorflow/op/dtypes/ToBool.java |   85 -
 .../estimator/BoostedTreesAggregateStats.java |   86 -
 .../op/estimator/BoostedTreesBucketize.java   |   87 -
 ...BoostedTreesCalculateBestFeatureSplit.java |  179 -
 ...ostedTreesCalculateBestFeatureSplitV2.java |  159 -
 ...stedTreesCalculateBestGainsPerFeature.java |  139 -
 .../op/estimator/BoostedTreesCenterBias.java  |   81 -
 .../estimator/BoostedTreesCreateEnsemble.java |   60 -
 ...stedTreesCreateQuantileStreamResource.java |   94 -
 .../BoostedTreesDeserializeEnsemble.java      |   62 -
 .../BoostedTreesEnsembleResourceHandleOp.java |  123 -
 .../BoostedTreesExampleDebugOutputs.java      |   84 -
 .../BoostedTreesFlushQuantileSummaries.java   |   84 -
 .../BoostedTreesGetEnsembleStates.java        |  105 -
 .../BoostedTreesMakeQuantileSummaries.java    |   88 -
 .../BoostedTreesMakeStatsSummary.java         |   86 -
 .../op/estimator/BoostedTreesPredict.java     |   83 -
 ...eesQuantileStreamResourceAddSummaries.java |   62 -
 ...reesQuantileStreamResourceDeserialize.java |   60 -
 ...ostedTreesQuantileStreamResourceFlush.java |  103 -
 ...tileStreamResourceGetBucketBoundaries.java |   84 -
 ...edTreesQuantileStreamResourceHandleOp.java |  123 -
 .../BoostedTreesSerializeEnsemble.java        |   77 -
 .../BoostedTreesSparseAggregateStats.java     |  115 -
 ...dTreesSparseCalculateBestFeatureSplit.java |  184 -
 .../BoostedTreesTrainingPredict.java          |  104 -
 .../estimator/BoostedTreesUpdateEnsemble.java |   85 -
 .../BoostedTreesUpdateEnsembleV2.java         |  124 -
 .../IsBoostedTreesEnsembleInitialized.java    |   72 -
 ...reesQuantileStreamResourceInitialized.java |   74 -
 .../tensorflow/op/image/AdjustContrast.java   |   88 -
 .../org/tensorflow/op/image/AdjustHue.java    |   85 -
 .../tensorflow/op/image/AdjustSaturation.java |   85 -
 .../op/image/CombinedNonMaxSuppression.java   |  197 -
 .../tensorflow/op/image/CropAndResize.java    |  168 -
 .../op/image/CropAndResizeGradBoxes.java      |  128 -
 .../op/image/CropAndResizeGradImage.java      |  133 -
 .../op/image/DecodeAndCropJpeg.java           |  245 -
 .../org/tensorflow/op/image/DecodeBmp.java    |  122 -
 .../org/tensorflow/op/image/DecodeGif.java    |   83 -
 .../org/tensorflow/op/image/DecodeJpeg.java   |  242 -
 .../org/tensorflow/op/image/DecodePng.java    |  150 -
 .../op/image/DrawBoundingBoxes.java           |   94 -
 .../org/tensorflow/op/image/EncodeJpeg.java   |  288 -
 .../op/image/EncodeJpegVariableQuality.java   |   81 -
 .../org/tensorflow/op/image/EncodePng.java    |  128 -
 .../tensorflow/op/image/ExtractGlimpse.java   |  211 -
 .../op/image/ExtractImagePatches.java         |  105 -
 .../tensorflow/op/image/ExtractJpegShape.java |   94 -
 .../image/GenerateBoundingBoxProposals.java   |  138 -
 .../org/tensorflow/op/image/HsvToRgb.java     |   81 -
 .../op/image/ImageProjectiveTransformV2.java  |  125 -
 .../tensorflow/op/image/NearestNeighbors.java |   88 -
 .../op/image/NonMaxSuppression.java           |  169 -
 .../image/NonMaxSuppressionWithOverlaps.java  |  103 -
 .../op/image/QuantizedResizeBilinear.java     |  153 -
 .../org/tensorflow/op/image/RandomCrop.java   |  142 -
 .../org/tensorflow/op/image/ResizeArea.java   |  127 -
 .../tensorflow/op/image/ResizeBicubic.java    |  136 -
 .../op/image/ResizeBicubicGrad.java           |  135 -
 .../tensorflow/op/image/ResizeBilinear.java   |  136 -
 .../op/image/ResizeBilinearGrad.java          |  135 -
 .../op/image/ResizeNearestNeighbor.java       |  135 -
 .../op/image/ResizeNearestNeighborGrad.java   |  134 -
 .../org/tensorflow/op/image/RgbToHsv.java     |   95 -
 .../op/image/SampleDistortedBoundingBox.java  |  291 -
 .../op/image/ScaleAndTranslate.java           |  132 -
 .../op/image/ScaleAndTranslateGrad.java       |  131 -
 .../org/tensorflow/op/io/DecodeBase64.java    |   76 -
 .../tensorflow/op/io/DecodeCompressed.java    |  117 -
 .../java/org/tensorflow/op/io/DecodeCsv.java  |  189 -
 .../tensorflow/op/io/DecodeJsonExample.java   |   82 -
 .../org/tensorflow/op/io/DecodePaddedRaw.java |  120 -
 .../java/org/tensorflow/op/io/DecodeRaw.java  |  118 -
 .../op/io/DeserializeManySparse.java          |  132 -
 .../org/tensorflow/op/io/EncodeBase64.java    |  114 -
 .../java/org/tensorflow/op/io/FifoQueue.java  |  187 -
 .../op/io/FixedLengthRecordReader.java        |  211 -
 .../org/tensorflow/op/io/IdentityReader.java  |  132 -
 .../java/org/tensorflow/op/io/LmdbReader.java |  128 -
 .../org/tensorflow/op/io/MatchingFiles.java   |   77 -
 .../tensorflow/op/io/PaddingFifoQueue.java    |  199 -
 .../org/tensorflow/op/io/ParseExample.java    |  200 -
 .../op/io/ParseSequenceExample.java           |  424 --
 .../tensorflow/op/io/ParseSingleExample.java  |  156 -
 .../op/io/ParseSingleSequenceExample.java     |  282 -
 .../org/tensorflow/op/io/ParseTensor.java     |   79 -
 .../org/tensorflow/op/io/PriorityQueue.java   |  173 -
 .../java/org/tensorflow/op/io/QueueClose.java |   97 -
 .../org/tensorflow/op/io/QueueDequeue.java    |  130 -
 .../tensorflow/op/io/QueueDequeueMany.java    |  140 -
 .../tensorflow/op/io/QueueDequeueUpTo.java    |  144 -
 .../org/tensorflow/op/io/QueueEnqueue.java    |  102 -
 .../tensorflow/op/io/QueueEnqueueMany.java    |  107 -
 .../org/tensorflow/op/io/QueueIsClosed.java   |   75 -
 .../java/org/tensorflow/op/io/QueueSize.java  |   73 -
 .../tensorflow/op/io/RandomShuffleQueue.java  |  250 -
 .../java/org/tensorflow/op/io/ReadFile.java   |   72 -
 .../op/io/ReaderNumRecordsProduced.java       |   75 -
 .../op/io/ReaderNumWorkUnitsCompleted.java    |   72 -
 .../java/org/tensorflow/op/io/ReaderRead.java |   83 -
 .../org/tensorflow/op/io/ReaderReadUpTo.java  |   87 -
 .../org/tensorflow/op/io/ReaderReset.java     |   55 -
 .../tensorflow/op/io/ReaderRestoreState.java  |   62 -
 .../op/io/ReaderSerializeState.java           |   75 -
 .../tensorflow/op/io/SerializeManySparse.java |  105 -
 .../org/tensorflow/op/io/SerializeSparse.java |   97 -
 .../org/tensorflow/op/io/SerializeTensor.java |   74 -
 .../org/tensorflow/op/io/ShardedFilename.java |   79 -
 .../org/tensorflow/op/io/ShardedFilespec.java |   75 -
 .../org/tensorflow/op/io/TextLineReader.java  |  148 -
 .../org/tensorflow/op/io/TfRecordReader.java  |  148 -
 .../org/tensorflow/op/io/WholeFileReader.java |  132 -
 .../java/org/tensorflow/op/io/WriteFile.java  |   60 -
 .../org/tensorflow/op/linalg/BandPart.java    |  118 -
 .../op/linalg/BandedTriangularSolve.java      |  126 -
 .../tensorflow/op/linalg/BatchCholesky.java   |   72 -
 .../op/linalg/BatchCholeskyGrad.java          |   74 -
 .../op/linalg/BatchMatrixBandPart.java        |   77 -
 .../op/linalg/BatchMatrixDeterminant.java     |   72 -
 .../tensorflow/op/linalg/BatchMatrixDiag.java |   72 -
 .../op/linalg/BatchMatrixDiagPart.java        |   72 -
 .../op/linalg/BatchMatrixInverse.java         |  106 -
 .../op/linalg/BatchMatrixSetDiag.java         |   74 -
 .../op/linalg/BatchMatrixSolve.java           |  108 -
 .../op/linalg/BatchMatrixSolveLs.java         |  111 -
 .../op/linalg/BatchMatrixTriangularSolve.java |  127 -
 .../op/linalg/BatchSelfAdjointEig.java        |  109 -
 .../org/tensorflow/op/linalg/BatchSvd.java    |  136 -
 .../org/tensorflow/op/linalg/Cholesky.java    |   89 -
 .../tensorflow/op/linalg/CholeskyGrad.java    |   84 -
 .../op/linalg/ConjugateTranspose.java         |   81 -
 .../java/org/tensorflow/op/linalg/Cross.java  |   81 -
 .../java/org/tensorflow/op/linalg/Det.java    |   79 -
 .../java/org/tensorflow/op/linalg/Eig.java    |  129 -
 .../java/org/tensorflow/op/linalg/Einsum.java |  155 -
 .../tensorflow/op/linalg/EuclideanNorm.java   |  118 -
 .../java/org/tensorflow/op/linalg/Inv.java    |  125 -
 .../op/linalg/LoadAndRemapMatrix.java         |  175 -
 .../op/linalg/LogMatrixDeterminant.java       |   90 -
 .../gen/java/org/tensorflow/op/linalg/Lu.java |  125 -
 .../java/org/tensorflow/op/linalg/MatMul.java |  137 -
 .../org/tensorflow/op/linalg/MatrixDiag.java  |  177 -
 .../tensorflow/op/linalg/MatrixDiagPart.java  |  151 -
 .../op/linalg/MatrixDiagPartV3.java           |  228 -
 .../tensorflow/op/linalg/MatrixDiagV3.java    |  252 -
 .../tensorflow/op/linalg/MatrixLogarithm.java |   94 -
 .../tensorflow/op/linalg/MatrixSetDiag.java   |  232 -
 .../tensorflow/op/linalg/MatrixSolveLs.java   |  153 -
 .../gen/java/org/tensorflow/op/linalg/Qr.java |  130 -
 .../tensorflow/op/linalg/QuantizedMatMul.java |  161 -
 .../op/linalg/QuantizedMatMulWithBias.java    |  181 -
 .../QuantizedMatMulWithBiasAndRelu.java       |  182 -
 ...zedMatMulWithBiasAndReluAndRequantize.java |  187 -
 .../tensorflow/op/linalg/SelfAdjointEig.java  |  127 -
 .../java/org/tensorflow/op/linalg/Solve.java  |  120 -
 .../java/org/tensorflow/op/linalg/Sqrtm.java  |   95 -
 .../java/org/tensorflow/op/linalg/Svd.java    |  166 -
 .../org/tensorflow/op/linalg/TensorDiag.java  |   92 -
 .../tensorflow/op/linalg/TensorDiagPart.java  |   94 -
 .../org/tensorflow/op/linalg/Transpose.java   |   80 -
 .../tensorflow/op/linalg/TriangularSolve.java |  189 -
 .../op/linalg/TridiagonalMatMul.java          |   86 -
 .../op/linalg/TridiagonalSolve.java           |  124 -
 .../sparse/CSRSparseMatrixComponents.java     |   95 -
 .../linalg/sparse/CSRSparseMatrixToDense.java |   76 -
 .../sparse/CSRSparseMatrixToSparseTensor.java |   90 -
 .../linalg/sparse/DenseToCSRSparseMatrix.java |   76 -
 .../op/linalg/sparse/SparseMatrixAdd.java     |   82 -
 .../op/linalg/sparse/SparseMatrixMatMul.java  |  230 -
 .../op/linalg/sparse/SparseMatrixMul.java     |   84 -
 .../op/linalg/sparse/SparseMatrixNNZ.java     |   72 -
 .../sparse/SparseMatrixOrderingAMD.java       |  120 -
 .../op/linalg/sparse/SparseMatrixSoftmax.java |   82 -
 .../sparse/SparseMatrixSoftmaxGrad.java       |   78 -
 .../sparse/SparseMatrixSparseCholesky.java    |  149 -
 .../sparse/SparseMatrixSparseMatMul.java      |  241 -
 .../linalg/sparse/SparseMatrixTranspose.java  |  112 -
 .../op/linalg/sparse/SparseMatrixZeros.java   |   76 -
 .../sparse/SparseTensorToCSRSparseMatrix.java |   78 -
 .../gen/java/org/tensorflow/op/math/Abs.java  |   78 -
 .../org/tensorflow/op/math/AccumulateN.java   |   87 -
 .../gen/java/org/tensorflow/op/math/Acos.java |   74 -
 .../java/org/tensorflow/op/math/Acosh.java    |   82 -
 .../gen/java/org/tensorflow/op/math/Add.java  |   79 -
 .../gen/java/org/tensorflow/op/math/AddN.java |   83 -
 .../java/org/tensorflow/op/math/Angle.java    |  106 -
 .../tensorflow/op/math/ApproximateEqual.java  |  109 -
 .../java/org/tensorflow/op/math/ArgMax.java   |  110 -
 .../java/org/tensorflow/op/math/ArgMin.java   |  110 -
 .../gen/java/org/tensorflow/op/math/Asin.java |   90 -
 .../java/org/tensorflow/op/math/Asinh.java    |   84 -
 .../gen/java/org/tensorflow/op/math/Atan.java |   90 -
 .../java/org/tensorflow/op/math/Atan2.java    |   82 -
 .../java/org/tensorflow/op/math/Atanh.java    |   86 -
 .../java/org/tensorflow/op/math/BesselI0.java |   71 -
 .../org/tensorflow/op/math/BesselI0e.java     |   71 -
 .../java/org/tensorflow/op/math/BesselI1.java |   71 -
 .../org/tensorflow/op/math/BesselI1e.java     |   71 -
 .../java/org/tensorflow/op/math/Betainc.java  |   89 -
 .../java/org/tensorflow/op/math/Bincount.java |   91 -
 .../gen/java/org/tensorflow/op/math/Ceil.java |   74 -
 .../tensorflow/op/math/CompareAndBitpack.java |   98 -
 .../org/tensorflow/op/math/ComplexAbs.java    |   95 -
 .../gen/java/org/tensorflow/op/math/Conj.java |   88 -
 .../gen/java/org/tensorflow/op/math/Cos.java  |   85 -
 .../gen/java/org/tensorflow/op/math/Cosh.java |   84 -
 .../java/org/tensorflow/op/math/Cumprod.java  |  156 -
 .../java/org/tensorflow/op/math/Cumsum.java   |  156 -
 .../op/math/CumulativeLogsumexp.java          |  147 -
 .../org/tensorflow/op/math/DenseBincount.java |  124 -
 .../java/org/tensorflow/op/math/Digamma.java  |   76 -
 .../gen/java/org/tensorflow/op/math/Div.java  |   79 -
 .../java/org/tensorflow/op/math/DivNoNan.java |   80 -
 .../java/org/tensorflow/op/math/Equal.java    |  122 -
 .../gen/java/org/tensorflow/op/math/Erf.java  |   74 -
 .../gen/java/org/tensorflow/op/math/Erfc.java |   74 -
 .../gen/java/org/tensorflow/op/math/Exp.java  |  100 -
 .../java/org/tensorflow/op/math/Expm1.java    |   89 -
 .../gen/java/org/tensorflow/op/math/Fact.java |   70 -
 .../java/org/tensorflow/op/math/Floor.java    |   74 -
 .../java/org/tensorflow/op/math/FloorDiv.java |   79 -
 .../java/org/tensorflow/op/math/FloorMod.java |   82 -
 .../java/org/tensorflow/op/math/Greater.java  |   90 -
 .../org/tensorflow/op/math/GreaterEqual.java  |   90 -
 .../java/org/tensorflow/op/math/Igamma.java   |   89 -
 .../org/tensorflow/op/math/IgammaGradA.java   |   75 -
 .../java/org/tensorflow/op/math/Igammac.java  |   89 -
 .../gen/java/org/tensorflow/op/math/Imag.java |  102 -
 .../tensorflow/op/math/InvertPermutation.java |   91 -
 .../java/org/tensorflow/op/math/IsFinite.java |   84 -
 .../java/org/tensorflow/op/math/IsInf.java    |   84 -
 .../java/org/tensorflow/op/math/IsNan.java    |   84 -
 .../gen/java/org/tensorflow/op/math/Less.java |   90 -
 .../org/tensorflow/op/math/LessEqual.java     |   90 -
 .../java/org/tensorflow/op/math/Lgamma.java   |   84 -
 .../gen/java/org/tensorflow/op/math/Log.java  |   83 -
 .../java/org/tensorflow/op/math/Log1p.java    |   83 -
 .../org/tensorflow/op/math/LogicalAnd.java    |   77 -
 .../org/tensorflow/op/math/LogicalNot.java    |   73 -
 .../org/tensorflow/op/math/LogicalOr.java     |   77 -
 .../java/org/tensorflow/op/math/Maximum.java  |   79 -
 .../gen/java/org/tensorflow/op/math/Mean.java |  118 -
 .../java/org/tensorflow/op/math/Minimum.java  |   79 -
 .../gen/java/org/tensorflow/op/math/Mod.java  |   82 -
 .../gen/java/org/tensorflow/op/math/Mul.java  |   79 -
 .../java/org/tensorflow/op/math/MulNoNan.java |   79 -
 .../java/org/tensorflow/op/math/Ndtri.java    |   72 -
 .../gen/java/org/tensorflow/op/math/Neg.java  |   76 -
 .../org/tensorflow/op/math/NextAfter.java     |   84 -
 .../java/org/tensorflow/op/math/NotEqual.java |  112 -
 .../org/tensorflow/op/math/Polygamma.java     |   83 -
 .../tensorflow/op/math/PopulationCount.java   |   80 -
 .../gen/java/org/tensorflow/op/math/Pow.java  |   85 -
 .../org/tensorflow/op/math/QuantizedAdd.java  |  103 -
 .../org/tensorflow/op/math/QuantizedMul.java  |  103 -
 .../gen/java/org/tensorflow/op/math/Real.java |  102 -
 .../java/org/tensorflow/op/math/RealDiv.java  |   81 -
 .../org/tensorflow/op/math/Reciprocal.java    |   76 -
 .../tensorflow/op/math/ReciprocalGrad.java    |   78 -
 .../math/RequantizationRangePerChannel.java   |   84 -
 .../op/math/RequantizePerChannel.java         |   98 -
 .../gen/java/org/tensorflow/op/math/Rint.java |   84 -
 .../java/org/tensorflow/op/math/Round.java    |   77 -
 .../java/org/tensorflow/op/math/Rsqrt.java    |   76 -
 .../org/tensorflow/op/math/RsqrtGrad.java     |   78 -
 .../org/tensorflow/op/math/SegmentMax.java    |  102 -
 .../org/tensorflow/op/math/SegmentMean.java   |  104 -
 .../org/tensorflow/op/math/SegmentMin.java    |  102 -
 .../org/tensorflow/op/math/SegmentProd.java   |  103 -
 .../org/tensorflow/op/math/SegmentSum.java    |  103 -
 .../java/org/tensorflow/op/math/Sigmoid.java  |   76 -
 .../org/tensorflow/op/math/SigmoidGrad.java   |   78 -
 .../gen/java/org/tensorflow/op/math/Sign.java |   82 -
 .../gen/java/org/tensorflow/op/math/Sin.java  |   84 -
 .../gen/java/org/tensorflow/op/math/Sinh.java |   84 -
 .../org/tensorflow/op/math/SobolSample.java   |  103 -
 .../java/org/tensorflow/op/math/Softplus.java |   74 -
 .../org/tensorflow/op/math/SoftplusGrad.java  |   76 -
 .../gen/java/org/tensorflow/op/math/Sqrt.java |   76 -
 .../java/org/tensorflow/op/math/SqrtGrad.java |   78 -
 .../java/org/tensorflow/op/math/Square.java   |   76 -
 .../tensorflow/op/math/SquaredDifference.java |   79 -
 .../gen/java/org/tensorflow/op/math/Sub.java  |   79 -
 .../gen/java/org/tensorflow/op/math/Tan.java  |   85 -
 .../gen/java/org/tensorflow/op/math/Tanh.java |   84 -
 .../java/org/tensorflow/op/math/TanhGrad.java |   78 -
 .../org/tensorflow/op/math/TruncateDiv.java   |   84 -
 .../org/tensorflow/op/math/TruncateMod.java   |   82 -
 .../op/math/UnsortedSegmentMax.java           |  112 -
 .../op/math/UnsortedSegmentMin.java           |  106 -
 .../op/math/UnsortedSegmentProd.java          |  106 -
 .../op/math/UnsortedSegmentSum.java           |  109 -
 .../java/org/tensorflow/op/math/Xdivy.java    |   76 -
 .../java/org/tensorflow/op/math/Xlog1py.java  |   76 -
 .../java/org/tensorflow/op/math/Xlogy.java    |   76 -
 .../gen/java/org/tensorflow/op/math/Zeta.java |   80 -
 .../java/org/tensorflow/op/math/erfinv.java   |   72 -
 .../tensorflow/op/math/special/BesselJ0.java  |   71 -
 .../tensorflow/op/math/special/BesselJ1.java  |   71 -
 .../tensorflow/op/math/special/BesselK0.java  |   71 -
 .../tensorflow/op/math/special/BesselK0e.java |   71 -
 .../tensorflow/op/math/special/BesselK1.java  |   71 -
 .../tensorflow/op/math/special/BesselK1e.java |   71 -
 .../tensorflow/op/math/special/BesselY0.java  |   71 -
 .../tensorflow/op/math/special/BesselY1.java  |   71 -
 .../org/tensorflow/op/math/special/Dawsn.java |   71 -
 .../tensorflow/op/math/special/Expint.java    |   71 -
 .../op/math/special/FresnelCos.java           |   71 -
 .../op/math/special/FresnelSin.java           |   71 -
 .../tensorflow/op/math/special/Spence.java    |   71 -
 .../java/org/tensorflow/op/nn/AvgPool.java    |  135 -
 .../java/org/tensorflow/op/nn/AvgPool3d.java  |  134 -
 .../org/tensorflow/op/nn/AvgPool3dGrad.java   |  137 -
 .../org/tensorflow/op/nn/AvgPoolGrad.java     |  135 -
 .../nn/BatchNormWithGlobalNormalization.java  |   96 -
 .../BatchNormWithGlobalNormalizationGrad.java |  127 -
 .../java/org/tensorflow/op/nn/BiasAdd.java    |  126 -
 .../org/tensorflow/op/nn/BiasAddGrad.java     |  125 -
 .../java/org/tensorflow/op/nn/BlockLSTM.java  |  216 -
 .../org/tensorflow/op/nn/BlockLSTMGrad.java   |  172 -
 .../java/org/tensorflow/op/nn/CTCLossV2.java  |  173 -
 .../op/nn/ComputeAccidentalHits.java          |  156 -
 .../gen/java/org/tensorflow/op/nn/Conv2d.java |  234 -
 .../op/nn/Conv2dBackpropFilter.java           |  217 -
 .../tensorflow/op/nn/Conv2dBackpropInput.java |  216 -
 .../gen/java/org/tensorflow/op/nn/Conv3d.java |  166 -
 .../op/nn/Conv3dBackpropFilter.java           |  166 -
 .../tensorflow/op/nn/Conv3dBackpropInput.java |  165 -
 .../op/nn/CtcBeamSearchDecoder.java           |  161 -
 .../tensorflow/op/nn/CtcGreedyDecoder.java    |  149 -
 .../java/org/tensorflow/op/nn/CtcLoss.java    |  175 -
 .../java/org/tensorflow/op/nn/CudnnRNN.java   |  333 -
 .../tensorflow/op/nn/CudnnRNNBackprop.java    |  332 -
 .../op/nn/CudnnRNNCanonicalToParams.java      |  263 -
 .../op/nn/CudnnRNNParamsToCanonical.java      |  274 -
 .../tensorflow/op/nn/CudnnRnnParamsSize.java  |  253 -
 .../tensorflow/op/nn/DataFormatDimMap.java    |  131 -
 .../op/nn/DataFormatVecPermute.java           |  130 -
 .../org/tensorflow/op/nn/DepthToSpace.java    |  190 -
 .../op/nn/DepthwiseConv2dNative.java          |  199 -
 .../DepthwiseConv2dNativeBackpropFilter.java  |  195 -
 .../DepthwiseConv2dNativeBackpropInput.java   |  195 -
 .../java/org/tensorflow/op/nn/Dilation2d.java |  118 -
 .../op/nn/Dilation2dBackpropFilter.java       |   96 -
 .../op/nn/Dilation2dBackpropInput.java        |   96 -
 .../gen/java/org/tensorflow/op/nn/Elu.java    |   77 -
 .../java/org/tensorflow/op/nn/EluGrad.java    |   77 -
 .../op/nn/FixedUnigramCandidateSampler.java   |  329 -
 .../tensorflow/op/nn/FractionalAvgPool.java   |  246 -
 .../op/nn/FractionalAvgPoolGrad.java          |  140 -
 .../tensorflow/op/nn/FractionalMaxPool.java   |  270 -
 .../op/nn/FractionalMaxPoolGrad.java          |  136 -
 .../org/tensorflow/op/nn/FusedBatchNorm.java  |  227 -
 .../tensorflow/op/nn/FusedBatchNormGrad.java  |  207 -
 .../org/tensorflow/op/nn/FusedPadConv2d.java  |  105 -
 .../op/nn/FusedResizeAndPadConv2d.java        |  143 -
 .../org/tensorflow/op/nn/GRUBlockCell.java    |  147 -
 .../tensorflow/op/nn/GRUBlockCellGrad.java    |  191 -
 .../gen/java/org/tensorflow/op/nn/InTopK.java |   94 -
 .../java/org/tensorflow/op/nn/InvGrad.java    |   78 -
 .../gen/java/org/tensorflow/op/nn/L2Loss.java |   79 -
 .../org/tensorflow/op/nn/LSTMBlockCell.java   |  234 -
 .../tensorflow/op/nn/LSTMBlockCellGrad.java   |  139 -
 .../op/nn/LearnedUnigramCandidateSampler.java |  171 -
 .../op/nn/LocalResponseNormalization.java     |  177 -
 .../op/nn/LocalResponseNormalizationGrad.java |  169 -
 .../java/org/tensorflow/op/nn/LogSoftmax.java |   79 -
 .../java/org/tensorflow/op/nn/MaxPool.java    |  125 -
 .../java/org/tensorflow/op/nn/MaxPool3d.java  |  134 -
 .../org/tensorflow/op/nn/MaxPool3dGrad.java   |  137 -
 .../tensorflow/op/nn/MaxPool3dGradGrad.java   |  138 -
 .../org/tensorflow/op/nn/MaxPoolGrad.java     |  129 -
 .../org/tensorflow/op/nn/MaxPoolGradGrad.java |  129 -
 .../op/nn/MaxPoolGradGradWithArgmax.java      |  130 -
 .../op/nn/MaxPoolGradWithArgmax.java          |  129 -
 .../tensorflow/op/nn/MaxPoolWithArgmax.java   |  160 -
 .../java/org/tensorflow/op/nn/NthElement.java |  123 -
 .../tensorflow/op/nn/QuantizedAvgPool.java    |  109 -
 ...tizedBatchNormWithGlobalNormalization.java |  131 -
 .../tensorflow/op/nn/QuantizedBiasAdd.java    |  102 -
 .../op/nn/QuantizedConv2DAndRelu.java         |  165 -
 .../QuantizedConv2DAndReluAndRequantize.java  |  169 -
 .../op/nn/QuantizedConv2DAndRequantize.java   |  169 -
 .../op/nn/QuantizedConv2DPerChannel.java      |  147 -
 .../op/nn/QuantizedConv2DWithBias.java        |  167 -
 .../op/nn/QuantizedConv2DWithBiasAndRelu.java |  167 -
 ...zedConv2DWithBiasAndReluAndRequantize.java |  171 -
 .../QuantizedConv2DWithBiasAndRequantize.java |  171 -
 ...WithBiasSignedSumAndReluAndRequantize.java |  177 -
 .../nn/QuantizedConv2DWithBiasSumAndRelu.java |  169 -
 ...Conv2DWithBiasSumAndReluAndRequantize.java |  177 -
 .../org/tensorflow/op/nn/QuantizedConv2d.java |  161 -
 .../op/nn/QuantizedDepthwiseConv2D.java       |  147 -
 .../nn/QuantizedDepthwiseConv2DWithBias.java  |  149 -
 ...antizedDepthwiseConv2DWithBiasAndRelu.java |  172 -
 ...iseConv2DWithBiasAndReluAndRequantize.java |  176 -
 .../op/nn/QuantizedInstanceNorm.java          |  207 -
 .../tensorflow/op/nn/QuantizedMaxPool.java    |  109 -
 .../org/tensorflow/op/nn/QuantizedRelu.java   |   95 -
 .../org/tensorflow/op/nn/QuantizedRelu6.java  |   95 -
 .../org/tensorflow/op/nn/QuantizedReluX.java  |   97 -
 .../gen/java/org/tensorflow/op/nn/Relu.java   |   79 -
 .../gen/java/org/tensorflow/op/nn/Relu6.java  |   74 -
 .../java/org/tensorflow/op/nn/Relu6Grad.java  |   78 -
 .../java/org/tensorflow/op/nn/ReluGrad.java   |   77 -
 .../gen/java/org/tensorflow/op/nn/Selu.java   |   82 -
 .../java/org/tensorflow/op/nn/SeluGrad.java   |   77 -
 .../java/org/tensorflow/op/nn/Softmax.java    |   79 -
 .../java/org/tensorflow/op/nn/Softsign.java   |   74 -
 .../org/tensorflow/op/nn/SoftsignGrad.java    |   76 -
 .../org/tensorflow/op/nn/SpaceToBatch.java    |  155 -
 .../org/tensorflow/op/nn/SpaceToDepth.java    |  184 -
 .../gen/java/org/tensorflow/op/nn/TopK.java   |  130 -
 .../nn/raw/SoftmaxCrossEntropyWithLogits.java |   85 -
 .../SparseSoftmaxCrossEntropyWithLogits.java  |   89 -
 .../op/quantization/Dequantize.java           |  219 -
 .../quantization/FakeQuantWithMinMaxArgs.java |  196 -
 .../FakeQuantWithMinMaxArgsGradient.java      |  167 -
 .../quantization/FakeQuantWithMinMaxVars.java |  166 -
 .../FakeQuantWithMinMaxVarsGradient.java      |  149 -
 .../FakeQuantWithMinMaxVarsPerChannel.java    |  167 -
 ...QuantWithMinMaxVarsPerChannelGradient.java |  152 -
 .../tensorflow/op/quantization/Quantize.java  |  327 -
 .../quantization/QuantizeAndDequantize.java   |  175 -
 .../QuantizeDownAndShrinkRange.java           |  117 -
 .../op/quantization/QuantizedConcat.java      |  101 -
 .../QuantizedMatMulWithBiasAndDequantize.java |  163 -
 .../QuantizedMatMulWithBiasAndRequantize.java |  173 -
 .../op/quantization/RequantizationRange.java  |   87 -
 .../op/quantization/Requantize.java           |  106 -
 .../tensorflow/op/ragged/RaggedBincount.java  |  127 -
 .../op/ragged/RaggedCountSparseOutput.java    |  156 -
 .../org/tensorflow/op/ragged/RaggedCross.java |  108 -
 .../tensorflow/op/ragged/RaggedGather.java    |  127 -
 .../org/tensorflow/op/ragged/RaggedRange.java |  115 -
 .../op/ragged/RaggedTensorFromVariant.java    |  127 -
 .../op/ragged/RaggedTensorToSparse.java       |   96 -
 .../op/ragged/RaggedTensorToTensor.java       |  146 -
 .../op/ragged/RaggedTensorToVariant.java      |   93 -
 .../op/random/AllCandidateSampler.java        |  169 -
 .../random/AnonymousRandomSeedGenerator.java  |   75 -
 .../op/random/AnonymousSeedGenerator.java     |   78 -
 .../op/random/DeleteRandomSeedGenerator.java  |   55 -
 .../op/random/DeleteSeedGenerator.java        |   55 -
 .../op/random/LogUniformCandidateSampler.java |  171 -
 .../org/tensorflow/op/random/Multinomial.java |  153 -
 .../op/random/NonDeterministicInts.java       |   91 -
 .../random/ParameterizedTruncatedNormal.java  |  145 -
 .../org/tensorflow/op/random/RandomGamma.java |  142 -
 .../tensorflow/op/random/RandomGammaGrad.java |   75 -
 .../tensorflow/op/random/RandomPoisson.java   |  167 -
 .../tensorflow/op/random/RandomShuffle.java   |  143 -
 .../op/random/RandomStandardNormal.java       |  136 -
 .../tensorflow/op/random/RandomUniform.java   |  137 -
 .../op/random/RandomUniformInt.java           |  144 -
 .../org/tensorflow/op/random/RecordInput.java |  206 -
 .../org/tensorflow/op/random/RngSkip.java     |   64 -
 .../op/random/StatefulRandomBinomial.java     |   99 -
 .../op/random/StatefulStandardNormal.java     |   99 -
 .../op/random/StatefulTruncatedNormal.java    |  100 -
 .../tensorflow/op/random/StatefulUniform.java |   99 -
 .../op/random/StatefulUniformFullInt.java     |   83 -
 .../op/random/StatefulUniformInt.java         |   91 -
 .../op/random/StatelessMultinomial.java       |  100 -
 ...StatelessParameterizedTruncatedNormal.java |   84 -
 .../op/random/StatelessRandomBinomial.java    |  106 -
 .../op/random/StatelessRandomGamma.java       |   83 -
 .../op/random/StatelessRandomNormal.java      |   97 -
 .../op/random/StatelessRandomPoisson.java     |   85 -
 .../op/random/StatelessRandomUniform.java     |   98 -
 .../random/StatelessRandomUniformFullInt.java |   82 -
 .../op/random/StatelessRandomUniformInt.java  |   84 -
 .../op/random/StatelessTruncatedNormal.java   |   99 -
 .../tensorflow/op/random/TruncatedNormal.java |  139 -
 .../op/random/UniformCandidateSampler.java    |  171 -
 .../experimental/DummySeedGenerator.java      |   69 -
 .../org/tensorflow/op/signal/BatchFft.java    |   72 -
 .../org/tensorflow/op/signal/BatchFft2d.java  |   72 -
 .../org/tensorflow/op/signal/BatchFft3d.java  |   72 -
 .../org/tensorflow/op/signal/BatchIfft.java   |   72 -
 .../org/tensorflow/op/signal/BatchIfft2d.java |   72 -
 .../org/tensorflow/op/signal/BatchIfft3d.java |   72 -
 .../java/org/tensorflow/op/signal/Fft.java    |   83 -
 .../java/org/tensorflow/op/signal/Fft2d.java  |   83 -
 .../java/org/tensorflow/op/signal/Fft3d.java  |   83 -
 .../java/org/tensorflow/op/signal/Ifft.java   |   83 -
 .../java/org/tensorflow/op/signal/Ifft2d.java |   83 -
 .../java/org/tensorflow/op/signal/Ifft3d.java |   83 -
 .../java/org/tensorflow/op/signal/Irfft.java  |  115 -
 .../org/tensorflow/op/signal/Irfft2d.java     |  116 -
 .../org/tensorflow/op/signal/Irfft3d.java     |  116 -
 .../java/org/tensorflow/op/signal/Rfft.java   |   98 -
 .../java/org/tensorflow/op/signal/Rfft2d.java |  100 -
 .../java/org/tensorflow/op/signal/Rfft3d.java |  100 -
 .../op/sparse/AddManySparseToTensorsMap.java  |  158 -
 .../op/sparse/AddSparseToTensorsMap.java      |  149 -
 .../op/sparse/DenseCountSparseOutput.java     |  148 -
 .../op/sparse/DenseToDenseSetOperation.java   |  139 -
 .../op/sparse/DenseToSparseSetOperation.java  |  154 -
 .../op/sparse/DeserializeSparse.java          |  131 -
 .../SparseAccumulatorApplyGradient.java       |   75 -
 .../sparse/SparseAccumulatorTakeGradient.java |  103 -
 .../org/tensorflow/op/sparse/SparseAdd.java   |  114 -
 .../tensorflow/op/sparse/SparseAddGrad.java   |   95 -
 .../tensorflow/op/sparse/SparseBincount.java  |  129 -
 .../tensorflow/op/sparse/SparseConcat.java    |  139 -
 .../sparse/SparseConditionalAccumulator.java  |  160 -
 .../op/sparse/SparseCountSparseOutput.java    |  152 -
 .../org/tensorflow/op/sparse/SparseCross.java |  134 -
 .../op/sparse/SparseCrossHashed.java          |  139 -
 .../op/sparse/SparseDenseCwiseAdd.java        |   92 -
 .../op/sparse/SparseDenseCwiseDiv.java        |   86 -
 .../op/sparse/SparseDenseCwiseMul.java        |   90 -
 .../op/sparse/SparseFillEmptyRows.java        |  144 -
 .../op/sparse/SparseFillEmptyRowsGrad.java    |   91 -
 .../tensorflow/op/sparse/SparseMatMul.java    |  176 -
 .../tensorflow/op/sparse/SparseReduceMax.java |  131 -
 .../op/sparse/SparseReduceMaxSparse.java      |  141 -
 .../tensorflow/op/sparse/SparseReduceSum.java |  131 -
 .../op/sparse/SparseReduceSumSparse.java      |  141 -
 .../tensorflow/op/sparse/SparseReorder.java   |   95 -
 .../tensorflow/op/sparse/SparseReshape.java   |  101 -
 .../op/sparse/SparseSegmentMean.java          |   85 -
 .../op/sparse/SparseSegmentMeanGrad.java      |   84 -
 .../SparseSegmentMeanWithNumSegments.java     |   89 -
 .../op/sparse/SparseSegmentSqrtN.java         |   85 -
 .../op/sparse/SparseSegmentSqrtNGrad.java     |   84 -
 .../SparseSegmentSqrtNWithNumSegments.java    |   91 -
 .../op/sparse/SparseSegmentSum.java           |  110 -
 .../SparseSegmentSumWithNumSegments.java      |  110 -
 .../org/tensorflow/op/sparse/SparseSlice.java |  116 -
 .../tensorflow/op/sparse/SparseSliceGrad.java |   87 -
 .../tensorflow/op/sparse/SparseSoftmax.java   |   97 -
 .../op/sparse/SparseSparseMaximum.java        |   93 -
 .../op/sparse/SparseSparseMinimum.java        |   93 -
 .../org/tensorflow/op/sparse/SparseSplit.java |  128 -
 .../op/sparse/SparseTensorDenseAdd.java       |   83 -
 .../op/sparse/SparseTensorDenseMatMul.java    |  149 -
 .../tensorflow/op/sparse/SparseToDense.java   |  139 -
 .../op/sparse/SparseToSparseSetOperation.java |  169 -
 .../sparse/TakeManySparseFromTensorsMap.java  |  195 -
 .../java/org/tensorflow/op/strings/Join.java  |  117 -
 .../java/org/tensorflow/op/strings/Lower.java |  112 -
 .../org/tensorflow/op/strings/ReduceJoin.java |  156 -
 .../tensorflow/op/strings/RegexFullMatch.java |   90 -
 .../tensorflow/op/strings/RegexReplace.java   |  119 -
 .../op/strings/StaticRegexFullMatch.java      |   82 -
 .../op/strings/StaticRegexReplace.java        |  114 -
 .../tensorflow/op/strings/StringFormat.java   |  148 -
 .../tensorflow/op/strings/StringLength.java   |  126 -
 .../tensorflow/op/strings/StringNGrams.java   |  109 -
 .../tensorflow/op/strings/StringSplit.java    |  144 -
 .../java/org/tensorflow/op/strings/Strip.java |   78 -
 .../org/tensorflow/op/strings/Substr.java     |  196 -
 .../tensorflow/op/strings/ToHashBucket.java   |   83 -
 .../op/strings/ToHashBucketFast.java          |   88 -
 .../op/strings/ToHashBucketStrong.java        |  103 -
 .../org/tensorflow/op/strings/ToNumber.java   |  101 -
 .../tensorflow/op/strings/UnicodeDecode.java  |  212 -
 .../op/strings/UnicodeDecodeWithOffsets.java  |  228 -
 .../tensorflow/op/strings/UnicodeEncode.java  |  168 -
 .../tensorflow/op/strings/UnicodeScript.java  |   84 -
 .../op/strings/UnicodeTranscode.java          |  216 -
 .../op/strings/UnsortedSegmentJoin.java       |  139 -
 .../java/org/tensorflow/op/strings/Upper.java |  112 -
 .../tensorflow/op/summary/AudioSummary.java   |  127 -
 .../op/summary/CloseSummaryWriter.java        |   53 -
 .../op/summary/CreateSummaryDbWriter.java     |   62 -
 .../op/summary/CreateSummaryFileWriter.java   |   63 -
 .../op/summary/FlushSummaryWriter.java        |   53 -
 .../op/summary/HistogramSummary.java          |   82 -
 .../tensorflow/op/summary/ImageSummary.java   |  178 -
 .../tensorflow/op/summary/ImportEvent.java    |   56 -
 .../tensorflow/op/summary/MergeSummary.java   |   83 -
 .../tensorflow/op/summary/ScalarSummary.java  |   79 -
 .../op/summary/StatsAggregatorSummary.java    |   71 -
 .../tensorflow/op/summary/SummaryWriter.java  |  122 -
 .../tensorflow/op/summary/TensorSummary.java  |   78 -
 .../op/summary/WriteAudioSummary.java         |   98 -
 .../op/summary/WriteGraphSummary.java         |   59 -
 .../op/summary/WriteHistogramSummary.java     |   62 -
 .../op/summary/WriteImageSummary.java         |   99 -
 .../op/summary/WriteRawProtoSummary.java      |   59 -
 .../op/summary/WriteScalarSummary.java        |   62 -
 .../tensorflow/op/summary/WriteSummary.java   |   64 -
 .../java/org/tensorflow/op/tpu/AllToAll.java  |  103 -
 .../tensorflow/op/tpu/CollectivePermute.java  |   84 -
 .../op/tpu/ConfigureDistributedTPU.java       |  183 -
 .../op/tpu/ConfigureTPUEmbedding.java         |   54 -
 .../tensorflow/op/tpu/CrossReplicaSum.java    |   86 -
 .../tpu/EnqueueTPUEmbeddingIntegerBatch.java  |   99 -
 .../EnqueueTPUEmbeddingRaggedTensorBatch.java |  185 -
 .../tpu/EnqueueTPUEmbeddingSparseBatch.java   |  151 -
 .../EnqueueTPUEmbeddingSparseTensorBatch.java |  183 -
 .../org/tensorflow/op/tpu/InfeedDequeue.java  |   77 -
 .../tensorflow/op/tpu/InfeedDequeueTuple.java |   89 -
 .../org/tensorflow/op/tpu/InfeedEnqueue.java  |  141 -
 .../tpu/InfeedEnqueuePrelinearizedBuffer.java |   90 -
 .../tensorflow/op/tpu/InfeedEnqueueTuple.java |  130 -
 .../tpu/LoadTPUEmbeddingADAMParameters.java   |  141 -
 ...EmbeddingADAMParametersGradAccumDebug.java |  143 -
 .../LoadTPUEmbeddingAdadeltaParameters.java   |  141 -
 ...ddingAdadeltaParametersGradAccumDebug.java |  143 -
 .../LoadTPUEmbeddingAdagradParameters.java    |  139 -
 ...eddingAdagradParametersGradAccumDebug.java |  141 -
 ...TPUEmbeddingCenteredRMSPropParameters.java |  143 -
 .../tpu/LoadTPUEmbeddingFTRLParameters.java   |  141 -
 ...EmbeddingFTRLParametersGradAccumDebug.java |  143 -
 ...TPUEmbeddingMDLAdagradLightParameters.java |  143 -
 .../LoadTPUEmbeddingMomentumParameters.java   |  139 -
 ...ddingMomentumParametersGradAccumDebug.java |  141 -
 ...TPUEmbeddingProximalAdagradParameters.java |  139 -
 ...oximalAdagradParametersGradAccumDebug.java |  141 -
 ...oadTPUEmbeddingProximalYogiParameters.java |  134 -
 ...gProximalYogiParametersGradAccumDebug.java |  136 -
 .../LoadTPUEmbeddingRMSPropParameters.java    |  141 -
 ...eddingRMSPropParametersGradAccumDebug.java |  143 -
 ...ngStochasticGradientDescentParameters.java |  137 -
 ...adientDescentParametersGradAccumDebug.java |  139 -
 .../org/tensorflow/op/tpu/OutfeedDequeue.java |  117 -
 .../op/tpu/OutfeedDequeueTuple.java           |  130 -
 .../org/tensorflow/op/tpu/OutfeedEnqueue.java |   55 -
 .../op/tpu/OutfeedEnqueueTuple.java           |   56 -
 .../org/tensorflow/op/tpu/Prelinearize.java   |  135 -
 .../tensorflow/op/tpu/PrelinearizeTuple.java  |  125 -
 .../op/tpu/RecvTPUEmbeddingActivations.java   |   90 -
 .../RetrieveTPUEmbeddingADAMParameters.java   |  163 -
 ...EmbeddingADAMParametersGradAccumDebug.java |  172 -
 ...etrieveTPUEmbeddingAdadeltaParameters.java |  163 -
 ...ddingAdadeltaParametersGradAccumDebug.java |  172 -
 ...RetrieveTPUEmbeddingAdagradParameters.java |  154 -
 ...eddingAdagradParametersGradAccumDebug.java |  163 -
 ...TPUEmbeddingCenteredRMSPropParameters.java |  172 -
 .../RetrieveTPUEmbeddingFTRLParameters.java   |  163 -
 ...EmbeddingFTRLParametersGradAccumDebug.java |  172 -
 ...TPUEmbeddingMDLAdagradLightParameters.java |  172 -
 ...etrieveTPUEmbeddingMomentumParameters.java |  154 -
 ...ddingMomentumParametersGradAccumDebug.java |  163 -
 ...TPUEmbeddingProximalAdagradParameters.java |  154 -
 ...oximalAdagradParametersGradAccumDebug.java |  163 -
 ...eveTPUEmbeddingProximalYogiParameters.java |  154 -
 ...gProximalYogiParametersGradAccumDebug.java |  162 -
 ...RetrieveTPUEmbeddingRMSPropParameters.java |  163 -
 ...eddingRMSPropParametersGradAccumDebug.java |  172 -
 ...ngStochasticGradientDescentParameters.java |  151 -
 ...adientDescentParametersGradAccumDebug.java |  154 -
 .../op/tpu/SendTPUEmbeddingGradients.java     |   70 -
 .../op/tpu/ShutdownDistributedTPU.java        |   53 -
 .../op/tpu/TPUCompilationResult.java          |   73 -
 .../op/tpu/TPUEmbeddingActivations.java       |   85 -
 .../tensorflow/op/tpu/TPUOrdinalSelector.java |   74 -
 .../op/tpu/TPUReplicateMetadata.java          |  258 -
 .../tensorflow/op/tpu/TPUReplicatedInput.java |  158 -
 .../op/tpu/TPUReplicatedOutput.java           |   92 -
 .../tensorflow/op/tpu/WorkerHeartbeat.java    |   75 -
 .../op/train/AccumulatorApplyGradient.java    |   64 -
 .../op/train/AccumulatorNumAccumulated.java   |   74 -
 .../op/train/AccumulatorSetGlobalStep.java    |   62 -
 .../op/train/AccumulatorTakeGradient.java     |   88 -
 .../org/tensorflow/op/train/ApplyAdaMax.java  |  132 -
 .../tensorflow/op/train/ApplyAdadelta.java    |  128 -
 .../org/tensorflow/op/train/ApplyAdagrad.java |  141 -
 .../tensorflow/op/train/ApplyAdagradDa.java   |  126 -
 .../tensorflow/op/train/ApplyAdagradV2.java   |  142 -
 .../org/tensorflow/op/train/ApplyAdam.java    |  155 -
 .../org/tensorflow/op/train/ApplyAddSign.java |  129 -
 .../op/train/ApplyCenteredRmsProp.java        |  148 -
 .../org/tensorflow/op/train/ApplyFtrl.java    |  156 -
 .../op/train/ApplyGradientDescent.java        |  115 -
 .../tensorflow/op/train/ApplyMomentum.java    |  149 -
 .../tensorflow/op/train/ApplyPowerSign.java   |  129 -
 .../op/train/ApplyProximalAdagrad.java        |  125 -
 .../train/ApplyProximalGradientDescent.java   |  122 -
 .../org/tensorflow/op/train/ApplyRmsProp.java |  138 -
 .../org/tensorflow/op/train/BatchMatMul.java  |  154 -
 .../op/train/ConditionalAccumulator.java      |  160 -
 .../op/train/GenerateVocabRemapping.java      |  152 -
 .../op/train/MergeV2Checkpoints.java          |  102 -
 .../org/tensorflow/op/train/NegTrain.java     |   74 -
 .../tensorflow/op/train/PreventGradient.java  |  119 -
 .../ResourceAccumulatorApplyGradient.java     |   62 -
 .../ResourceAccumulatorNumAccumulated.java    |   72 -
 .../ResourceAccumulatorSetGlobalStep.java     |   60 -
 .../ResourceAccumulatorTakeGradient.java      |   86 -
 .../op/train/ResourceApplyAdaMax.java         |  113 -
 .../op/train/ResourceApplyAdadelta.java       |  109 -
 .../op/train/ResourceApplyAdagrad.java        |  123 -
 .../op/train/ResourceApplyAdagradDa.java      |  107 -
 .../op/train/ResourceApplyAdam.java           |  136 -
 .../train/ResourceApplyAdamWithAmsgrad.java   |  120 -
 .../op/train/ResourceApplyAddSign.java        |  110 -
 .../train/ResourceApplyCenteredRmsProp.java   |  129 -
 .../op/train/ResourceApplyFtrl.java           |  137 -
 .../train/ResourceApplyGradientDescent.java   |   96 -
 .../op/train/ResourceApplyKerasMomentum.java  |  130 -
 .../op/train/ResourceApplyMomentum.java       |  130 -
 .../op/train/ResourceApplyPowerSign.java      |  110 -
 .../train/ResourceApplyProximalAdagrad.java   |  106 -
 .../ResourceApplyProximalGradientDescent.java |  103 -
 .../op/train/ResourceApplyRmsProp.java        |  119 -
 .../train/ResourceConditionalAccumulator.java |  161 -
 .../op/train/ResourceSparseApplyAdadelta.java |  107 -
 .../op/train/ResourceSparseApplyAdagrad.java  |  126 -
 .../train/ResourceSparseApplyAdagradDa.java   |  110 -
 .../train/ResourceSparseApplyAdagradV2.java   |  127 -
 .../ResourceSparseApplyCenteredRmsProp.java   |  130 -
 .../op/train/ResourceSparseApplyFtrl.java     |  141 -
 .../ResourceSparseApplyKerasMomentum.java     |  135 -
 .../op/train/ResourceSparseApplyMomentum.java |  135 -
 .../ResourceSparseApplyProximalAdagrad.java   |  111 -
 ...rceSparseApplyProximalGradientDescent.java |  107 -
 .../op/train/ResourceSparseApplyRmsProp.java  |  122 -
 .../java/org/tensorflow/op/train/Restore.java |  107 -
 .../org/tensorflow/op/train/RestoreSlice.java |  128 -
 .../java/org/tensorflow/op/train/Save.java    |   69 -
 .../org/tensorflow/op/train/SaveSlices.java   |   95 -
 .../org/tensorflow/op/train/SdcaFprint.java   |   75 -
 .../tensorflow/op/train/SdcaOptimizer.java    |  185 -
 .../org/tensorflow/op/train/SdcaShrinkL1.java |   62 -
 .../op/train/SparseApplyAdadelta.java         |  126 -
 .../op/train/SparseApplyAdagrad.java          |  146 -
 .../op/train/SparseApplyAdagradDa.java        |  129 -
 .../op/train/SparseApplyCenteredRmsProp.java  |  149 -
 .../tensorflow/op/train/SparseApplyFtrl.java  |  160 -
 .../op/train/SparseApplyMomentum.java         |  154 -
 .../op/train/SparseApplyProximalAdagrad.java  |  130 -
 .../SparseApplyProximalGradientDescent.java   |  126 -
 .../op/train/SparseApplyRmsProp.java          |  141 -
 .../org/tensorflow/op/train/TileGrad.java     |   81 -
 .../tensorflow/op/xla/BroadcastHelper.java    |   88 -
 .../org/tensorflow/op/xla/ClusterOutput.java  |   74 -
 .../gen/java/org/tensorflow/op/xla/Conv.java  |   94 -
 .../org/tensorflow/op/xla/Dequantize.java     |   86 -
 .../gen/java/org/tensorflow/op/xla/Dot.java   |   83 -
 .../org/tensorflow/op/xla/DynamicSlice.java   |   91 -
 .../tensorflow/op/xla/DynamicUpdateSlice.java |   91 -
 .../java/org/tensorflow/op/xla/Einsum.java    |   81 -
 .../java/org/tensorflow/op/xla/Gather.java    |   85 -
 .../org/tensorflow/op/xla/KeyValueSort.java   |   88 -
 .../gen/java/org/tensorflow/op/xla/Pad.java   |   87 -
 .../gen/java/org/tensorflow/op/xla/Recv.java  |   83 -
 .../java/org/tensorflow/op/xla/ReplicaId.java |   70 -
 .../org/tensorflow/op/xla/SelfAdjointEig.java |   97 -
 .../gen/java/org/tensorflow/op/xla/Send.java  |   61 -
 .../java/org/tensorflow/op/xla/Sharding.java  |   74 -
 .../gen/java/org/tensorflow/op/xla/Sort.java  |   80 -
 .../gen/java/org/tensorflow/op/xla/Svd.java   |  103 -
 .../proto/distruntime/ClusterDef.java         |  865 ---
 .../distruntime/ClusterDefOrBuilder.java      |   53 -
 .../distruntime/ClusterDeviceFilters.java     |  773 --
 .../ClusterDeviceFiltersOrBuilder.java        |   33 -
 .../proto/distruntime/ClusterProtos.java      |   76 -
 .../distruntime/DeviceFiltersProtos.java      |   91 -
 .../tensorflow/proto/distruntime/JobDef.java  |  935 ---
 .../proto/distruntime/JobDefOrBuilder.java    |   96 -
 .../proto/distruntime/JobDeviceFilters.java   |  902 ---
 .../JobDeviceFiltersOrBuilder.java            |   81 -
 .../proto/distruntime/ServerDef.java          | 1611 -----
 .../proto/distruntime/ServerDefOrBuilder.java |  149 -
 .../proto/distruntime/ServerProtos.java       |   66 -
 .../proto/distruntime/TaskDeviceFilters.java  |  600 --
 .../TaskDeviceFiltersOrBuilder.java           |   28 -
 .../proto/distruntime/TransportOptions.java   |  635 --
 .../tensorflow/proto/example/BytesList.java   |  572 --
 .../proto/example/BytesListOrBuilder.java     |   22 -
 .../org/tensorflow/proto/example/Example.java |  600 --
 .../proto/example/ExampleOrBuilder.java       |   22 -
 .../example/ExampleParserConfiguration.java   |  695 --
 .../ExampleParserConfigurationOrBuilder.java  |   43 -
 .../ExampleParserConfigurationProtos.java     |  123 -
 .../proto/example/ExampleProtos.java          |   68 -
 .../org/tensorflow/proto/example/Feature.java | 1105 ---
 .../proto/example/FeatureConfiguration.java   |  893 ---
 .../FeatureConfigurationOrBuilder.java        |   37 -
 .../tensorflow/proto/example/FeatureList.java |  781 --
 .../proto/example/FeatureListOrBuilder.java   |   33 -
 .../proto/example/FeatureLists.java           |  739 --
 .../proto/example/FeatureListsOrBuilder.java  |   63 -
 .../proto/example/FeatureOrBuilder.java       |   50 -
 .../proto/example/FeatureProtos.java          |  153 -
 .../tensorflow/proto/example/Features.java    |  739 --
 .../proto/example/FeaturesOrBuilder.java      |   63 -
 .../proto/example/FixedLenFeatureProto.java   |  993 ---
 .../FixedLenFeatureProtoOrBuilder.java        |   54 -
 .../tensorflow/proto/example/FloatList.java   |  579 --
 .../proto/example/FloatListOrBuilder.java     |   22 -
 .../tensorflow/proto/example/Int64List.java   |  582 --
 .../proto/example/Int64ListOrBuilder.java     |   22 -
 .../proto/example/SequenceExample.java        |  781 --
 .../example/SequenceExampleOrBuilder.java     |   35 -
 .../proto/example/VarLenFeatureProto.java     |  885 ---
 .../example/VarLenFeatureProtoOrBuilder.java  |   48 -
 .../framework/AllocationDescription.java      |  944 ---
 .../AllocationDescriptionOrBuilder.java       |   72 -
 .../AllocationDescriptionProtos.java          |   56 -
 .../proto/framework/AllocationRecord.java     |  575 --
 .../framework/AllocationRecordOrBuilder.java  |   27 -
 .../proto/framework/AllocatorMemoryUsed.java  | 1268 ----
 .../AllocatorMemoryUsedOrBuilder.java         |   96 -
 .../tensorflow/proto/framework/ApiDef.java    | 6303 -----------------
 .../proto/framework/ApiDefOrBuilder.java      |  274 -
 .../proto/framework/ApiDefProtos.java         |  117 -
 .../tensorflow/proto/framework/ApiDefs.java   |  765 --
 .../proto/framework/ApiDefsOrBuilder.java     |   33 -
 .../proto/framework/AssetFileDef.java         |  827 ---
 .../framework/AssetFileDefOrBuilder.java      |   56 -
 .../tensorflow/proto/framework/AttrValue.java | 5341 --------------
 .../proto/framework/AttrValueOrBuilder.java   |  203 -
 .../proto/framework/AttrValueProtos.java      |  110 -
 .../proto/framework/AutoParallelOptions.java  |  534 --
 .../AutoParallelOptionsOrBuilder.java         |   19 -
 .../framework/BoundedTensorSpecProto.java     | 1182 ----
 .../BoundedTensorSpecProtoOrBuilder.java      |   67 -
 .../proto/framework/CallableOptions.java      | 2902 --------
 .../framework/CallableOptionsOrBuilder.java   |  485 --
 .../org/tensorflow/proto/framework/Code.java  |  536 --
 .../proto/framework/CollectionDef.java        | 4858 -------------
 .../framework/CollectionDefOrBuilder.java     |   76 -
 .../proto/framework/CondContextDef.java       | 1632 -----
 .../framework/CondContextDefOrBuilder.java    |  141 -
 .../proto/framework/ConfigProto.java          | 6064 ----------------
 .../proto/framework/ConfigProtoOrBuilder.java |  477 --
 .../proto/framework/ConfigProtos.java         |  393 -
 .../framework/ControlFlowContextDef.java      |  903 ---
 .../ControlFlowContextDefOrBuilder.java       |   37 -
 .../proto/framework/ControlFlowProtos.java    |  116 -
 .../proto/framework/CostGraphDef.java         | 5817 ---------------
 .../framework/CostGraphDefOrBuilder.java      |   57 -
 .../proto/framework/CostGraphProtos.java      |  122 -
 .../tensorflow/proto/framework/DataClass.java |  167 -
 .../tensorflow/proto/framework/DataType.java  |  615 --
 .../proto/framework/DebugOptions.java         | 1033 ---
 .../framework/DebugOptionsOrBuilder.java      |   76 -
 .../proto/framework/DebugProtos.java          |   95 -
 .../proto/framework/DebugTensorWatch.java     | 1444 ----
 .../framework/DebugTensorWatchOrBuilder.java  |  196 -
 .../proto/framework/DebuggedSourceFile.java   | 1102 ---
 .../DebuggedSourceFileOrBuilder.java          |   98 -
 .../proto/framework/DebuggedSourceFiles.java  |  857 ---
 .../DebuggedSourceFilesOrBuilder.java         |   53 -
 .../tensorflow/proto/framework/Device.java    |  977 ---
 .../proto/framework/DeviceAttributes.java     | 1277 ----
 .../framework/DeviceAttributesOrBuilder.java  |  110 -
 .../framework/DeviceAttributesProtos.java     |   94 -
 .../proto/framework/DeviceLocality.java       |  798 ---
 .../framework/DeviceLocalityOrBuilder.java    |   53 -
 .../proto/framework/DeviceOrBuilder.java      |   90 -
 .../proto/framework/DeviceProperties.java     | 1885 -----
 .../framework/DevicePropertiesOrBuilder.java  |  204 -
 .../framework/DevicePropertiesProtos.java     |   85 -
 .../proto/framework/DeviceStepStats.java      | 1209 ----
 .../framework/DeviceStepStatsOrBuilder.java   |   97 -
 .../tensorflow/proto/framework/DictValue.java |  705 --
 .../proto/framework/DictValueOrBuilder.java   |   43 -
 .../proto/framework/ErrorCodes.java           |   40 -
 .../proto/framework/ErrorCodesProtos.java     |   49 -
 .../proto/framework/FunctionDef.java          | 3415 ---------
 .../proto/framework/FunctionDefLibrary.java   | 1116 ---
 .../FunctionDefLibraryOrBuilder.java          |   57 -
 .../proto/framework/FunctionDefOrBuilder.java |  381 -
 .../proto/framework/FunctionProtos.java       |  184 -
 .../proto/framework/FunctionSpec.java         |  961 ---
 .../framework/FunctionSpecOrBuilder.java      |   68 -
 .../proto/framework/GPUOptions.java           | 5061 -------------
 .../proto/framework/GPUOptionsOrBuilder.java  |  206 -
 .../proto/framework/GradientDef.java          |  765 --
 .../proto/framework/GradientDefOrBuilder.java |   45 -
 .../proto/framework/GraphDebugInfo.java       | 3004 --------
 .../framework/GraphDebugInfoOrBuilder.java    |  147 -
 .../proto/framework/GraphDebugInfoProtos.java |   92 -
 .../tensorflow/proto/framework/GraphDef.java  | 1588 -----
 .../proto/framework/GraphDefOrBuilder.java    |  163 -
 .../proto/framework/GraphOptions.java         | 1462 ----
 .../framework/GraphOptionsOrBuilder.java      |  139 -
 .../proto/framework/GraphProtos.java          |   63 -
 .../framework/GraphTransferConstNodeInfo.java |  912 ---
 .../GraphTransferConstNodeInfoOrBuilder.java  |   51 -
 .../GraphTransferGraphInputNodeInfo.java      |  794 ---
 ...phTransferGraphInputNodeInfoOrBuilder.java |   41 -
 .../GraphTransferGraphOutputNodeInfo.java     |  794 ---
 ...hTransferGraphOutputNodeInfoOrBuilder.java |   41 -
 .../proto/framework/GraphTransferInfo.java    | 2795 --------
 .../framework/GraphTransferInfoOrBuilder.java |  190 -
 .../framework/GraphTransferInfoProto.java     |  163 -
 .../framework/GraphTransferNodeInfo.java      |  958 ---
 .../GraphTransferNodeInfoOrBuilder.java       |   54 -
 .../framework/GraphTransferNodeInput.java     |  533 --
 .../framework/GraphTransferNodeInputInfo.java |  822 ---
 .../GraphTransferNodeInputInfoOrBuilder.java  |   38 -
 .../GraphTransferNodeInputOrBuilder.java      |   19 -
 .../GraphTransferNodeOutputInfo.java          |  639 --
 .../GraphTransferNodeOutputInfoOrBuilder.java |   27 -
 .../proto/framework/HistogramProto.java       | 1120 ---
 .../framework/HistogramProtoOrBuilder.java    |   84 -
 .../proto/framework/InterconnectLink.java     |  660 --
 .../framework/InterconnectLinkOrBuilder.java  |   29 -
 .../tensorflow/proto/framework/KernelDef.java | 2420 -------
 .../proto/framework/KernelDefOrBuilder.java   |  141 -
 .../proto/framework/KernelDefProtos.java      |   83 -
 .../proto/framework/KernelList.java           |  773 --
 .../proto/framework/KernelListOrBuilder.java  |   33 -
 .../tensorflow/proto/framework/ListValue.java |  773 --
 .../proto/framework/ListValueOrBuilder.java   |   33 -
 .../proto/framework/LocalLinks.java           |  765 --
 .../proto/framework/LocalLinksOrBuilder.java  |   33 -
 .../proto/framework/LogMemoryProtos.java      |  125 -
 .../framework/MemoryLogRawAllocation.java     | 1029 ---
 .../MemoryLogRawAllocationOrBuilder.java      |   82 -
 .../framework/MemoryLogRawDeallocation.java   |  959 ---
 .../MemoryLogRawDeallocationOrBuilder.java    |   74 -
 .../proto/framework/MemoryLogStep.java        |  648 --
 .../framework/MemoryLogStepOrBuilder.java     |   36 -
 .../framework/MemoryLogTensorAllocation.java  |  884 ---
 .../MemoryLogTensorAllocationOrBuilder.java   |   63 -
 .../MemoryLogTensorDeallocation.java          |  652 --
 .../MemoryLogTensorDeallocationOrBuilder.java |   37 -
 .../framework/MemoryLogTensorOutput.java      |  957 ---
 .../MemoryLogTensorOutputOrBuilder.java       |   72 -
 .../proto/framework/MemoryStats.java          |  981 ---
 .../proto/framework/MemoryStatsOrBuilder.java |   55 -
 .../proto/framework/MetaGraphDef.java         | 4702 ------------
 .../framework/MetaGraphDefOrBuilder.java      |  259 -
 .../proto/framework/MetaGraphProtos.java      |  318 -
 .../proto/framework/NameAttrList.java         |  832 ---
 .../framework/NameAttrListOrBuilder.java      |   53 -
 .../proto/framework/NamedDevice.java          |  727 --
 .../proto/framework/NamedDeviceOrBuilder.java |   32 -
 .../proto/framework/NamedTensorProto.java     |  859 ---
 .../framework/NamedTensorProtoOrBuilder.java  |   64 -
 .../proto/framework/NamedTensorProtos.java    |   55 -
 .../proto/framework/NamedTupleValue.java      |  900 ---
 .../framework/NamedTupleValueOrBuilder.java   |   43 -
 .../tensorflow/proto/framework/NodeDef.java   | 3076 --------
 .../proto/framework/NodeDefOrBuilder.java     |  284 -
 .../proto/framework/NodeExecStats.java        | 2580 -------
 .../framework/NodeExecStatsOrBuilder.java     |  183 -
 .../proto/framework/NodeOutput.java           |  665 --
 .../proto/framework/NodeOutputOrBuilder.java  |   27 -
 .../tensorflow/proto/framework/NodeProto.java |   84 -
 .../tensorflow/proto/framework/NoneValue.java |  427 --
 .../proto/framework/NoneValueOrBuilder.java   |    9 -
 .../org/tensorflow/proto/framework/OpDef.java | 6284 ----------------
 .../proto/framework/OpDefOrBuilder.java       |  296 -
 .../proto/framework/OpDefProtos.java          |  121 -
 .../proto/framework/OpDeprecation.java        |  655 --
 .../framework/OpDeprecationOrBuilder.java     |   36 -
 .../tensorflow/proto/framework/OpList.java    |  773 --
 .../proto/framework/OpListOrBuilder.java      |   33 -
 .../proto/framework/OptimizerOptions.java     | 1210 ----
 .../framework/OptimizerOptionsOrBuilder.java  |   77 -
 .../tensorflow/proto/framework/PairValue.java |  735 --
 .../proto/framework/PairValueOrBuilder.java   |   32 -
 .../proto/framework/QueueRunnerDef.java       | 1435 ----
 .../framework/QueueRunnerDefOrBuilder.java    |  145 -
 .../proto/framework/QueueRunnerProtos.java    |   58 -
 .../proto/framework/RPCOptions.java           |  905 ---
 .../proto/framework/RPCOptionsOrBuilder.java  |   72 -
 .../proto/framework/ReaderBaseProtos.java     |   54 -
 .../proto/framework/ReaderBaseState.java      |  664 --
 .../framework/ReaderBaseStateOrBuilder.java   |   29 -
 .../RemoteFusedGraphExecuteInfo.java          | 3007 --------
 .../RemoteFusedGraphExecuteInfoOrBuilder.java |  239 -
 .../RemoteFusedGraphExecuteInfoProto.java     |   85 -
 .../proto/framework/RemoteTensorHandle.java   | 1441 ----
 .../RemoteTensorHandleOrBuilder.java          |  128 -
 .../framework/RemoteTensorHandleProtos.java   |   76 -
 .../tensorflow/proto/framework/Resource.java  |  659 --
 .../framework/ResourceDtypeAndShape.java      |  685 --
 .../ResourceDtypeAndShapeOrBuilder.java       |   31 -
 .../proto/framework/ResourceHandle.java       |   75 -
 .../proto/framework/ResourceHandleProto.java  | 2288 ------
 .../ResourceHandleProtoOrBuilder.java         |  137 -
 .../proto/framework/ResourceOrBuilder.java    |   36 -
 .../proto/framework/RewriterConfig.java       | 6064 ----------------
 .../framework/RewriterConfigOrBuilder.java    |  627 --
 .../proto/framework/RewriterConfigProtos.java |  159 -
 .../proto/framework/RunMetadata.java          | 3277 ---------
 .../proto/framework/RunMetadataOrBuilder.java |  198 -
 .../proto/framework/RunOptions.java           | 2708 -------
 .../proto/framework/RunOptionsOrBuilder.java  |  101 -
 .../proto/framework/SaveSliceInfoDef.java     | 1175 ---
 .../framework/SaveSliceInfoDefOrBuilder.java  |  102 -
 .../proto/framework/SaveableObject.java       |  549 --
 .../framework/SaveableObjectOrBuilder.java    |   23 -
 .../proto/framework/SavedAsset.java           |  514 --
 .../proto/framework/SavedAssetOrBuilder.java  |   20 -
 .../framework/SavedBareConcreteFunction.java  |  873 ---
 .../SavedBareConcreteFunctionOrBuilder.java   |   71 -
 .../framework/SavedConcreteFunction.java      | 1156 ---
 .../SavedConcreteFunctionOrBuilder.java       |  102 -
 .../proto/framework/SavedConstant.java        |  574 --
 .../framework/SavedConstantOrBuilder.java     |   27 -
 .../proto/framework/SavedFunction.java        |  781 --
 .../framework/SavedFunctionOrBuilder.java     |   41 -
 .../proto/framework/SavedModel.java           |  949 ---
 .../proto/framework/SavedModelOrBuilder.java  |   64 -
 .../proto/framework/SavedModelProtos.java     |   56 -
 .../proto/framework/SavedObject.java          | 3174 ---------
 .../proto/framework/SavedObjectGraph.java     | 1231 ----
 .../framework/SavedObjectGraphOrBuilder.java  |  122 -
 .../framework/SavedObjectGraphProtos.java     |  263 -
 .../proto/framework/SavedObjectOrBuilder.java |  249 -
 .../proto/framework/SavedResource.java        |  600 --
 .../framework/SavedResourceOrBuilder.java     |   31 -
 .../proto/framework/SavedUserObject.java      |  974 ---
 .../framework/SavedUserObjectOrBuilder.java   |   70 -
 .../proto/framework/SavedVariable.java        | 1050 ---
 .../framework/SavedVariableOrBuilder.java     |   64 -
 .../framework/ScopedAllocatorOptions.java     |  644 --
 .../ScopedAllocatorOptionsOrBuilder.java      |   44 -
 .../proto/framework/SessionMetadata.java      |  636 --
 .../framework/SessionMetadataOrBuilder.java   |   28 -
 .../proto/framework/SignatureDef.java         | 1339 ----
 .../framework/SignatureDefOrBuilder.java      |  147 -
 .../tensorflow/proto/framework/StepStats.java |  765 --
 .../proto/framework/StepStatsOrBuilder.java   |   33 -
 .../proto/framework/StepStatsProtos.java      |  169 -
 .../proto/framework/StructProtos.java         |  215 -
 .../proto/framework/StructuredValue.java      | 3423 ---------
 .../framework/StructuredValueOrBuilder.java   |  309 -
 .../tensorflow/proto/framework/Summary.java   | 4725 ------------
 .../proto/framework/SummaryDescription.java   |  589 --
 .../SummaryDescriptionOrBuilder.java          |   29 -
 .../proto/framework/SummaryMetadata.java      | 1784 -----
 .../framework/SummaryMetadataOrBuilder.java   |   93 -
 .../proto/framework/SummaryOrBuilder.java     |   53 -
 .../proto/framework/SummaryProtos.java        |  159 -
 .../proto/framework/TensorConnection.java     |  751 --
 .../framework/TensorConnectionOrBuilder.java  |   49 -
 .../proto/framework/TensorDescription.java    |  990 ---
 .../framework/TensorDescriptionOrBuilder.java |   76 -
 .../framework/TensorDescriptionProtos.java    |   65 -
 .../proto/framework/TensorInfo.java           | 3680 ----------
 .../proto/framework/TensorInfoOrBuilder.java  |  128 -
 .../proto/framework/TensorProto.java          | 3980 -----------
 .../proto/framework/TensorProtoOrBuilder.java |  440 --
 .../proto/framework/TensorProtos.java         |   86 -
 .../proto/framework/TensorShapeProto.java     | 1852 -----
 .../framework/TensorShapeProtoOrBuilder.java  |  108 -
 .../proto/framework/TensorShapeProtos.java    |   65 -
 .../proto/framework/TensorSliceProto.java     | 1589 -----
 .../framework/TensorSliceProtoOrBuilder.java  |   68 -
 .../proto/framework/TensorSliceProtos.java    |   65 -
 .../proto/framework/TensorSpecProto.java      |  820 ---
 .../framework/TensorSpecProtoOrBuilder.java   |   41 -
 .../framework/ThreadPoolOptionProto.java      |  739 --
 .../ThreadPoolOptionProtoOrBuilder.java       |   62 -
 .../org/tensorflow/proto/framework/Trace.java | 1193 ----
 .../proto/framework/TraceEvent.java           | 1208 ----
 .../proto/framework/TraceEventOrBuilder.java  |  122 -
 .../proto/framework/TraceEventsProtos.java    |  133 -
 .../proto/framework/TraceOrBuilder.java       |  112 -
 .../proto/framework/TrackableObjectGraph.java | 5141 --------------
 .../TrackableObjectGraphOrBuilder.java        |   33 -
 .../framework/TrackableObjectGraphProtos.java |  111 -
 .../proto/framework/TupleValue.java           |  773 --
 .../proto/framework/TupleValueOrBuilder.java  |   33 -
 .../proto/framework/TypeSpecProto.java        | 1176 ---
 .../framework/TypeSpecProtoOrBuilder.java     |   67 -
 .../proto/framework/TypesProtos.java          |   60 -
 .../tensorflow/proto/framework/ValuesDef.java |  969 ---
 .../proto/framework/ValuesDefOrBuilder.java   |   98 -
 .../proto/framework/VariableAggregation.java  |  165 -
 .../proto/framework/VariableDef.java          | 1650 -----
 .../proto/framework/VariableDefOrBuilder.java |  158 -
 .../proto/framework/VariableProtos.java       |   81 -
 .../framework/VariableSynchronization.java    |  169 -
 .../framework/VariantTensorDataProto.java     | 1097 ---
 .../VariantTensorDataProtoOrBuilder.java      |   80 -
 .../proto/framework/VerifierConfig.java       |  725 --
 .../framework/VerifierConfigOrBuilder.java    |   36 -
 .../proto/framework/VerifierConfigProtos.java |   55 -
 .../proto/framework/VersionDef.java           |  792 ---
 .../proto/framework/VersionDefOrBuilder.java  |   52 -
 .../proto/framework/VersionsProtos.java       |   52 -
 .../proto/framework/WhileContextDef.java      | 2534 -------
 .../framework/WhileContextDefOrBuilder.java   |  265 -
 .../proto/profiler/ProfileOptions.java        | 1176 ---
 .../profiler/ProfileOptionsOrBuilder.java     |  103 -
 .../proto/profiler/ProfilerOptionsProtos.java |   56 -
 .../org/tensorflow/proto/profiler/XEvent.java | 1263 ----
 .../proto/profiler/XEventMetadata.java        |  890 ---
 .../profiler/XEventMetadataOrBuilder.java     |   63 -
 .../proto/profiler/XEventOrBuilder.java       |   92 -
 .../org/tensorflow/proto/profiler/XLine.java  | 1508 ----
 .../proto/profiler/XLineOrBuilder.java        |  133 -
 .../org/tensorflow/proto/profiler/XPlane.java | 2168 ------
 .../proto/profiler/XPlaneOrBuilder.java       |  243 -
 .../proto/profiler/XPlaneProtos.java          |  167 -
 .../org/tensorflow/proto/profiler/XSpace.java | 1219 ----
 .../proto/profiler/XSpaceOrBuilder.java       |  103 -
 .../org/tensorflow/proto/profiler/XStat.java  | 1111 ---
 .../proto/profiler/XStatMetadata.java         |  813 ---
 .../profiler/XStatMetadataOrBuilder.java      |   54 -
 .../proto/profiler/XStatOrBuilder.java        |   59 -
 .../proto/util/BfcMemoryMapProtos.java        |  112 -
 .../org/tensorflow/proto/util/BinSummary.java |  708 --
 .../proto/util/BinSummaryOrBuilder.java       |   34 -
 .../proto/util/BundleEntryProto.java          | 1540 ----
 .../proto/util/BundleEntryProtoOrBuilder.java |  137 -
 .../proto/util/BundleHeaderProto.java         |  929 ---
 .../util/BundleHeaderProtoOrBuilder.java      |   52 -
 .../tensorflow/proto/util/CodeLocation.java   |  837 ---
 .../proto/util/CodeLocationOrBuilder.java     |   70 -
 .../org/tensorflow/proto/util/DebugEvent.java | 2883 --------
 .../proto/util/DebugEventOrBuilder.java       |  258 -
 .../proto/util/DebugEventProtos.java          |  205 -
 .../tensorflow/proto/util/DebugMetadata.java  |  920 ---
 .../proto/util/DebugMetadataOrBuilder.java    |   71 -
 .../tensorflow/proto/util/DebuggedDevice.java |  667 --
 .../proto/util/DebuggedDeviceOrBuilder.java   |   39 -
 .../tensorflow/proto/util/DebuggedGraph.java  | 1295 ----
 .../proto/util/DebuggedGraphOrBuilder.java    |  123 -
 .../java/org/tensorflow/proto/util/Event.java | 2049 ------
 .../tensorflow/proto/util/EventOrBuilder.java |  174 -
 .../tensorflow/proto/util/EventProtos.java    |  161 -
 .../org/tensorflow/proto/util/Execution.java  | 2259 ------
 .../proto/util/ExecutionOrBuilder.java        |  233 -
 .../proto/util/GraphExecutionTrace.java       | 1359 ----
 .../util/GraphExecutionTraceOrBuilder.java    |  125 -
 .../proto/util/GraphOpCreation.java           | 1936 -----
 .../proto/util/GraphOpCreationOrBuilder.java  |  195 -
 .../org/tensorflow/proto/util/LogMessage.java |  787 --
 .../proto/util/LogMessageOrBuilder.java       |   28 -
 .../proto/util/MemAllocatorStats.java         |  718 --
 .../util/MemAllocatorStatsOrBuilder.java      |   34 -
 .../org/tensorflow/proto/util/MemChunk.java   | 1009 ---
 .../proto/util/MemChunkOrBuilder.java         |   59 -
 .../util/MemmappedFileSystemDirectory.java    |  773 --
 .../MemmappedFileSystemDirectoryElement.java  |  670 --
 ...edFileSystemDirectoryElementOrBuilder.java |   29 -
 ...MemmappedFileSystemDirectoryOrBuilder.java |   33 -
 .../proto/util/MemmappedFileSystemProtos.java |   64 -
 .../org/tensorflow/proto/util/MemoryDump.java | 1759 -----
 .../proto/util/MemoryDumpOrBuilder.java       |  104 -
 .../proto/util/RequestedExitCode.java         |  476 --
 .../util/RequestedExitCodeOrBuilder.java      |   14 -
 .../org/tensorflow/proto/util/SavedSlice.java | 1073 ---
 .../tensorflow/proto/util/SavedSliceMeta.java | 1364 ----
 .../proto/util/SavedSliceMetaOrBuilder.java   |  113 -
 .../proto/util/SavedSliceOrBuilder.java       |   85 -
 .../proto/util/SavedTensorSliceMeta.java      | 1108 ---
 .../util/SavedTensorSliceMetaOrBuilder.java   |   81 -
 .../proto/util/SavedTensorSliceProtos.java    |  109 -
 .../proto/util/SavedTensorSlices.java         |  899 ---
 .../util/SavedTensorSlicesOrBuilder.java      |   62 -
 .../org/tensorflow/proto/util/SaverDef.java   | 1356 ----
 .../proto/util/SaverDefOrBuilder.java         |  104 -
 .../tensorflow/proto/util/SaverProtos.java    |   57 -
 .../org/tensorflow/proto/util/SessionLog.java |  910 ---
 .../proto/util/SessionLogOrBuilder.java       |   46 -
 .../org/tensorflow/proto/util/SnapShot.java   |  535 --
 .../proto/util/SnapShotOrBuilder.java         |   19 -
 .../proto/util/SnapshotMetadataRecord.java    | 1328 ----
 .../util/SnapshotMetadataRecordOrBuilder.java |  121 -
 .../tensorflow/proto/util/SnapshotProtos.java |  102 -
 .../tensorflow/proto/util/SnapshotRecord.java |  777 --
 .../proto/util/SnapshotRecordOrBuilder.java   |   33 -
 .../proto/util/SnapshotTensorMetadata.java    |  773 --
 .../util/SnapshotTensorMetadataOrBuilder.java |   33 -
 .../org/tensorflow/proto/util/SourceFile.java |  964 ---
 .../proto/util/SourceFileOrBuilder.java       |   80 -
 .../proto/util/StackFrameWithId.java          |  835 ---
 .../proto/util/StackFrameWithIdOrBuilder.java |   58 -
 .../proto/util/TaggedRunMetadata.java         |  663 --
 .../util/TaggedRunMetadataOrBuilder.java      |   37 -
 .../proto/util/TensorBundleProtos.java        |   83 -
 .../proto/util/TensorDebugMode.java           |  282 -
 .../tensorflow/proto/util/TensorMetadata.java |  682 --
 .../proto/util/TensorMetadataOrBuilder.java   |   31 -
 .../tensorflow/proto/util/WatchdogConfig.java |  477 --
 .../proto/util/WatchdogConfigOrBuilder.java   |   14 -
 .../tensorflow/proto/util/WorkerHealth.java   |  141 -
 .../proto/util/WorkerHeartbeatRequest.java    |  866 ---
 .../util/WorkerHeartbeatRequestOrBuilder.java |   44 -
 .../proto/util/WorkerHeartbeatResponse.java   |  977 ---
 .../WorkerHeartbeatResponseOrBuilder.java     |   52 -
 .../proto/util/WorkerShutdownMode.java        |  126 -
 .../util/testlog/AvailableDeviceInfo.java     |  966 ---
 .../testlog/AvailableDeviceInfoOrBuilder.java |   72 -
 .../proto/util/testlog/BenchmarkEntries.java  |  765 --
 .../testlog/BenchmarkEntriesOrBuilder.java    |   33 -
 .../proto/util/testlog/BenchmarkEntry.java    | 1676 -----
 .../util/testlog/BenchmarkEntryOrBuilder.java |  168 -
 .../util/testlog/BuildConfiguration.java      | 1021 ---
 .../testlog/BuildConfigurationOrBuilder.java  |   97 -
 .../proto/util/testlog/CPUInfo.java           | 1254 ----
 .../proto/util/testlog/CPUInfoOrBuilder.java  |  122 -
 .../proto/util/testlog/CommitId.java          |  964 ---
 .../proto/util/testlog/CommitIdOrBuilder.java |   59 -
 .../proto/util/testlog/EntryValue.java        |  713 --
 .../util/testlog/EntryValueOrBuilder.java     |   26 -
 .../proto/util/testlog/GPUInfo.java           |  884 ---
 .../proto/util/testlog/GPUInfoOrBuilder.java  |   63 -
 .../util/testlog/MachineConfiguration.java    | 2241 ------
 .../MachineConfigurationOrBuilder.java        |  196 -
 .../proto/util/testlog/MemoryInfo.java        |  567 --
 .../util/testlog/MemoryInfoOrBuilder.java     |   27 -
 .../proto/util/testlog/MetricEntry.java       | 1107 ---
 .../util/testlog/MetricEntryOrBuilder.java    |   86 -
 .../proto/util/testlog/PlatformInfo.java      | 1349 ----
 .../util/testlog/PlatformInfoOrBuilder.java   |  117 -
 .../proto/util/testlog/RunConfiguration.java  |  917 ---
 .../testlog/RunConfigurationOrBuilder.java    |   82 -
 .../proto/util/testlog/TestLogProtos.java     |  288 -
 .../proto/util/testlog/TestResults.java       | 2624 -------
 .../util/testlog/TestResultsOrBuilder.java    |  245 -
 .../org/tensorflow/AbstractOperation.java     |    5 +-
 .../java/org/tensorflow/ConcreteFunction.java |   32 +-
 .../java/org/tensorflow/EagerOperation.java   |   30 +-
 .../org/tensorflow/EagerOperationBuilder.java |   11 +-
 .../java/org/tensorflow/GraphOperation.java   |    8 +-
 .../org/tensorflow/GraphOperationBuilder.java |   15 +-
 .../java/org/tensorflow/OperationBuilder.java |    4 +-
 .../src/main/java/org/tensorflow/Output.java  |    5 +-
 .../java/org/tensorflow/SavedModelBundle.java |    4 +-
 .../src/main/java/org/tensorflow/Session.java |   47 +-
 .../main/java/org/tensorflow/Signature.java   |    4 +-
 .../src/main/java/org/tensorflow/Tensor.java  |   25 +-
 ...{AbstractTensor.java => TensorHandle.java} |   49 +-
 .../main/java/org/tensorflow/TensorType.java  |   47 -
 .../src/main/java/org/tensorflow/Tensors.java |   36 +-
 .../buffer/ByteSequenceTensorBuffer.java      |    2 +-
 .../{tensor => }/buffer/TensorBuffers.java    |    2 +-
 .../buffer/TensorRawDataBufferFactory.java    |    2 +-
 .../internal/tensor/BooleanTensorImpl.java    |  141 -
 .../internal/tensor/ByteTensorImpl.java       |  141 -
 .../internal/tensor/DoubleTensorImpl.java     |  141 -
 .../internal/tensor/FloatTensorImpl.java      |  140 -
 .../internal/tensor/IntTensorImpl.java        |  141 -
 .../internal/tensor/LongTensorImpl.java       |  141 -
 .../internal/tensor/StringTensorImpl.java     |  130 -
 .../internal/types/TBfloat16Factory.java      |   42 +
 .../internal/types/TBoolFactory.java          |   41 +
 .../internal/types/TFloat16Factory.java       |   42 +
 .../internal/types/TFloat32Factory.java       |   41 +
 .../internal/types/TFloat64Factory.java       |   41 +
 .../internal/types/TInt32Factory.java         |   41 +
 .../internal/types/TInt64Factory.java         |   41 +
 .../internal/types/TStringFactory.java        |   78 +
 .../internal/types/TUint8Factory.java         |   42 +
 .../java/org/tensorflow/op/core/Constant.java |    1 -
 .../org/tensorflow/op/core/Gradients.java     |    1 -
 .../java/org/tensorflow/op/core/Shapes.java   |    1 -
 .../op/nn/SigmoidCrossEntropyWithLogits.java  |    8 +-
 .../op/nn/SoftmaxCrossEntropyWithLogits.java  |   15 +-
 .../SparseSoftmaxCrossEntropyWithLogits.java  |    7 +-
 .../java/org/tensorflow/types/TBfloat16.java  |   30 +-
 .../main/java/org/tensorflow/types/TBool.java |   24 +-
 .../java/org/tensorflow/types/TFloat16.java   |   30 +-
 .../java/org/tensorflow/types/TFloat32.java   |   25 +-
 .../java/org/tensorflow/types/TFloat64.java   |   25 +-
 .../java/org/tensorflow/types/TInt32.java     |   25 +-
 .../java/org/tensorflow/types/TInt64.java     |   25 +-
 .../java/org/tensorflow/types/TString.java    |   55 +-
 .../java/org/tensorflow/types/TUint8.java     |   25 +-
 .../main/java/org/tensorflow/types/Type.java  |   39 +
 .../org/tensorflow/types/TypeFactory.java     |   10 +
 .../TypeRegistry.java}                        |   62 +-
 .../types/annotation/TensorType.java          |   12 +-
 .../tensorflow/types/family/TFloating.java    |    2 +-
 .../org/tensorflow/types/family/TNumber.java  |    2 +-
 .../org/tensorflow/types/family/TType.java    |   71 +-
 .../org/tensorflow/types/package-info.java    |    4 +-
 .../types/tensor/BooleanTensor.java           |   37 -
 .../tensorflow/types/tensor/ByteTensor.java   |   37 -
 .../tensorflow/types/tensor/DoubleTensor.java |   37 -
 .../tensorflow/types/tensor/FloatTensor.java  |   37 -
 .../tensorflow/types/tensor/IntTensor.java    |   37 -
 .../tensorflow/types/tensor/LongTensor.java   |   37 -
 .../tensorflow/types/tensor/StringTensor.java |   31 -
 .../java/org/tensorflow/util/TensorList.java  |   60 +
 .../java/org/tensorflow/util/TensorMap.java   |   73 +
 .../org/tensorflow/AutoCloseableList.java     |   27 -
 .../org/tensorflow/ConcreteFunctionTest.java  |   21 +-
 .../tensorflow/EagerOperationBuilderTest.java |    8 +-
 .../org/tensorflow/EagerOperationTest.java    |   11 +-
 .../tensorflow/GraphOperationBuilderTest.java |   22 +-
 .../test/java/org/tensorflow/GraphTest.java   |   69 +-
 .../org/tensorflow/SavedModelBundleTest.java  |   57 +-
 .../test/java/org/tensorflow/SessionTest.java |   43 +-
 .../test/java/org/tensorflow/TensorTest.java  |   55 +-
 .../java/org/tensorflow/op/ScopeTest.java     |   16 +-
 .../org/tensorflow/op/core/ConstantTest.java  |   38 +-
 .../op/core/GeneratedOperationsTest.java      |    9 +-
 .../org/tensorflow/op/core/GradientsTest.java |   33 +-
 .../org/tensorflow/op/core/ShapesTest.java    |   79 +-
 .../org/tensorflow/op/core/ZerosTest.java     |   37 +-
 .../types/NumericTypesTestBase.java           |    7 +-
 .../org/tensorflow/types/TBfloat16Test.java   |    1 -
 .../org/tensorflow/types/TFloat16Test.java    |    1 -
 .../org/tensorflow/types/TFloat32Test.java    |    1 -
 .../org/tensorflow/types/TFloat64Test.java    |    1 -
 .../java/org/tensorflow/types/TInt32Test.java |    8 -
 .../java/org/tensorflow/types/TInt64Test.java |    1 -
 .../org/tensorflow/types/TStringTest.java     |    1 -
 .../java/org/tensorflow/types/TUint8Test.java |    1 -
 .../framework/utils/ShapeUtils.java           |    1 +
 .../framework/data/BatchDatasetTest.java      |    1 -
 .../framework/data/DatasetIteratorTest.java   |    2 +-
 .../framework/data/MapDatasetTest.java        |    2 +-
 .../framework/data/SkipDatasetTest.java       |    1 -
 .../framework/data/TakeDatasetTest.java       |    1 -
 .../framework/optimizers/AdamTest.java        |    2 +-
 .../framework/optimizers/AdamaxTest.java      |    2 +-
 .../framework/optimizers/NadamTest.java       |    2 +-
 .../framework/utils/GraphTestSession.java     |    1 +
 1755 files changed, 1097 insertions(+), 445864 deletions(-)
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/Compute_func_Pointer_TF_OpKernelContext.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/Create_func_TF_OpKernelConstruction.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/Deallocator_Pointer_long_Pointer.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/Delete_func_Pointer.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/Listener_BytePointer.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/Listener_String.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/Shape_inference_func_TF_ShapeInferenceContext_TF_Status.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TFE_Context.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TFE_ContextOptions.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TFE_Op.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TFE_TensorDebugInfo.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TFE_TensorHandle.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_ApiDefMap.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_AttrMetadata.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_Buffer.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_DeprecatedSession.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_DeviceList.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_DimensionHandle.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_Function.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_FunctionOptions.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_Graph.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_ImportGraphDefOptions.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_ImportGraphDefResults.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_Input.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_KernelBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_Library.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_OpDefinitionBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_OpKernelConstruction.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_OpKernelContext.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_Operation.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_OperationDescription.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_Output.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_Server.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_Session.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_SessionOptions.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_ShapeHandle.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_ShapeInferenceContext.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_Status.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_Tensor.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_WhileParams.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/Tensor.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/global/tensorflow.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/audio/AudioSpectrogram.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/audio/DecodeWav.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/audio/EncodeWav.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/audio/Mfcc.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/bitwise/BitwiseAnd.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/bitwise/BitwiseOr.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/bitwise/BitwiseXor.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/bitwise/Invert.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/bitwise/LeftShift.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/bitwise/RightShift.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/cluster/KMC2ChainInitialization.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/cluster/KmeansPlusPlusInitialization.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/collective/AllReduce.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/collective/BroadcastRecv.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/collective/BroadcastSend.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Abort.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/All.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Any.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/AssertThat.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Assign.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/AssignAdd.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/AssignAddVariableOp.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/AssignSub.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/AssignSubVariableOp.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/AssignVariableOp.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Barrier.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BarrierClose.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BarrierIncompleteSize.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BarrierInsertMany.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BarrierReadySize.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BarrierTakeMany.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Batch.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BatchToSpace.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BatchToSpaceNd.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Bitcast.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BroadcastDynamicShape.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BroadcastGradientArgs.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BroadcastTo.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Bucketize.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ClipByValue.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/CollectiveGather.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Concat.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ConsumeMutexLock.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ControlTrigger.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Copy.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/CopyHost.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/CountUpTo.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/DecodeProto.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/DeepCopy.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/DeleteSessionTensor.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/DestroyResourceOp.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/DestroyTemporaryVariable.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/DeviceIndex.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/DummyMemoryCache.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/DynamicPartition.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/DynamicStitch.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/EditDistance.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Empty.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/EmptyTensorList.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/EncodeProto.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/EnsureShape.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Enter.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Exit.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ExpandDims.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ExtractVolumePatches.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Fill.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Fingerprint.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Gather.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/GatherNd.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/GetSessionHandle.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/GetSessionTensor.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/GuaranteeConst.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/HashTable.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/HistogramFixedWidth.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Identity.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/IdentityN.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ImmutableConst.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/InitializeTable.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/InitializeTableFromTextFile.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/InplaceAdd.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/InplaceSub.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/InplaceUpdate.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/IsVariableInitialized.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LinSpace.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LookupTableExport.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LookupTableFind.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LookupTableImport.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LookupTableInsert.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LookupTableRemove.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LookupTableSize.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LoopCond.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LowerBound.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MapClear.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MapIncompleteSize.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MapPeek.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MapSize.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MapStage.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MapUnstage.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MapUnstageNoKey.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Max.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Merge.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Min.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MirrorPad.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MirrorPadGrad.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MlirPassthroughOp.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MutableDenseHashTable.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MutableHashTable.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MutableHashTableOfTensors.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Mutex.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MutexLock.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/NcclAllReduce.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/NcclBroadcast.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/NcclReduce.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/NextIteration.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/NoOp.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OneHot.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OnesLike.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OrderedMapClear.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OrderedMapIncompleteSize.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OrderedMapPeek.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OrderedMapSize.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OrderedMapStage.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OrderedMapUnstage.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OrderedMapUnstageNoKey.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Pad.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ParallelConcat.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ParallelDynamicStitch.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Placeholder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/PlaceholderWithDefault.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Print.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Prod.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/QuantizedReshape.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Range.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Rank.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ReadVariableOp.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Recv.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ReduceAll.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ReduceAny.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ReduceMax.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ReduceMin.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ReduceProd.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ReduceSum.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/RefEnter.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/RefExit.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/RefIdentity.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/RefMerge.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/RefNextIteration.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/RefSelect.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/RefSwitch.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/RemoteFusedGraphExecute.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Reshape.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceCountUpTo.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceGather.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceGatherNd.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterAdd.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterDiv.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterMax.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterMin.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterMul.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterNdAdd.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterNdMax.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterNdMin.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterNdSub.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterNdUpdate.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterSub.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterUpdate.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceStridedSliceAssign.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Reverse.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ReverseSequence.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Roll.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Rpc.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterAdd.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterDiv.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterMax.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterMin.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterMul.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterNd.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterNdAdd.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterNdMax.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterNdMin.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterNdNonAliasingAdd.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterNdSub.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterNdUpdate.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterSub.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterUpdate.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Select.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Send.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/SetDiff1d.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/SetSize.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Shape.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ShapeN.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Size.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Skipgram.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Slice.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Snapshot.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/SpaceToBatchNd.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Split.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/SplitV.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Squeeze.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Stack.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Stage.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/StageClear.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/StagePeek.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/StageSize.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/StopGradient.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/StridedSlice.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/StridedSliceAssign.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/StridedSliceGrad.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Sum.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/SwitchCond.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TemporaryVariable.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArray.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayClose.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayConcat.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayGather.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayGrad.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayGradWithShape.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayPack.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayRead.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayScatter.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArraySize.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArraySplit.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayUnpack.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayWrite.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorForestCreateTreeVariable.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorForestTreeDeserialize.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorForestTreeIsInitializedOp.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorForestTreePredict.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorForestTreeResourceHandleOp.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorForestTreeSerialize.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorForestTreeSize.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListConcat.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListConcatLists.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListElementShape.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListFromTensor.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListGather.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListGetItem.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListLength.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListPopBack.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListPushBack.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListPushBackBatch.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListReserve.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListResize.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListScatter.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListScatterIntoExistingList.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListSetItem.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListSplit.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListStack.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorScatterNdAdd.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorScatterNdMax.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorScatterNdMin.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorScatterNdSub.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorScatterNdUpdate.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorStridedSliceUpdate.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Tile.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Timestamp.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TryRpc.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Unbatch.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/UnbatchGrad.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Unique.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/UniqueWithCounts.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/UnravelIndex.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Unstack.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Unstage.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/UpperBound.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/VarHandleOp.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/VarIsInitializedOp.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Variable.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/VariableShape.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Where.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/XlaSpmdFullToShardShape.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/XlaSpmdShardToFullShape.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ZerosLike.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/AnonymousIterator.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/AnonymousMemoryCache.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/AnonymousMultiDeviceIterator.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/AssertNextDataset.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/AutoShardDataset.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/BatchDataset.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/BytesProducedStatsDataset.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/CSVDataset.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/CacheDataset.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/CacheDatasetV2.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ChooseFastestDataset.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ConcatenateDataset.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DatasetCardinality.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DatasetFromGraph.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DatasetToGraph.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DatasetToSingleElement.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DatasetToTfRecord.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DeleteIterator.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DeleteMemoryCache.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DeleteMultiDeviceIterator.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DenseToSparseBatchDataset.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DeserializeIterator.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DirectedInterleaveDataset.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/FilterByLastComponentDataset.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/FixedLengthRecordDataset.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/IgnoreErrorsDataset.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/InitializeTableFromDataset.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/Iterator.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/IteratorFromStringHandle.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/IteratorGetDevice.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/IteratorGetNext.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/IteratorGetNextAsOptional.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/IteratorGetNextSync.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/IteratorToStringHandle.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/LMDBDataset.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/LatencyStatsDataset.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/LeakyReluGrad.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MakeIterator.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MatchingFilesDataset.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MaxIntraOpParallelismDataset.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ModelDataset.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MultiDeviceIterator.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MultiDeviceIteratorFromStringHandle.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MultiDeviceIteratorGetNextFromShard.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MultiDeviceIteratorInit.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MultiDeviceIteratorToStringHandle.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/NonSerializableDataset.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/OptimizeDataset.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/OptionalFromValue.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/OptionalGetValue.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/OptionalHasValue.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/OptionalNone.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/PaddedBatchDataset.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/PrefetchDataset.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/PrivateThreadPoolDataset.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/RandomDataset.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/RangeDataset.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/RebatchDataset.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/RegisterDataset.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/RepeatDataset.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SamplingDataset.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SerializeIterator.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SetStatsAggregatorDataset.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ShardDataset.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ShuffleAndRepeatDataset.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ShuffleDataset.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SkipDataset.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SleepDataset.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SlidingWindowDataset.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SparseTensorSliceDataset.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SqlDataset.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/StatsAggregatorHandle.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/TakeDataset.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/TensorDataset.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/TensorSliceDataset.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/TextLineDataset.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/TfRecordDataset.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ThreadPoolDataset.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ThreadPoolHandle.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/UnbatchDataset.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/UniqueDataset.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/UnwrapDatasetVariant.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/WindowDataset.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/WrapDatasetVariant.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ZipDataset.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/AssertCardinalityDataset.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/AssertNextDataset.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/AutoShardDataset.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/BytesProducedStatsDataset.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/CSVDataset.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/ChooseFastestDataset.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/CompressElement.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/DataServiceDataset.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/DatasetCardinality.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/DatasetToTFRecord.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/DenseToSparseBatchDataset.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/DirectedInterleaveDataset.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/DummyIterationCounter.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/IgnoreErrorsDataset.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/IteratorGetDevice.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/LatencyStatsDataset.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/LmdbDataset.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/MatchingFilesDataset.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/MaxIntraOpParallelismDataset.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/NonSerializableDataset.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/ParseExampleDataset.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/PrivateThreadPoolDataset.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/RandomDataset.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/RebatchDataset.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/SetStatsAggregatorDataset.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/SleepDataset.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/SlidingWindowDataset.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/SqlDataset.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/StatsAggregatorHandle.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/StatsAggregatorSetSummaryWriter.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/StatsAggregatorSummary.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/ThreadPoolDataset.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/ThreadPoolHandle.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/UnbatchDataset.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/UncompressElement.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/UniqueDataset.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/debugging/CheckNumerics.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/debugging/DebugGradientIdentity.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/debugging/DebugGradientRefIdentity.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/debugging/DebugIdentity.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/debugging/DebugNanCount.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/debugging/DebugNumericsSummary.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/dtypes/AsString.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/dtypes/Cast.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/dtypes/Complex.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/dtypes/ToBool.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesAggregateStats.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesBucketize.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesCalculateBestFeatureSplit.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesCalculateBestFeatureSplitV2.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesCalculateBestGainsPerFeature.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesCenterBias.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesCreateEnsemble.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesCreateQuantileStreamResource.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesDeserializeEnsemble.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesEnsembleResourceHandleOp.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesExampleDebugOutputs.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesFlushQuantileSummaries.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesGetEnsembleStates.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesMakeQuantileSummaries.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesMakeStatsSummary.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesPredict.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesQuantileStreamResourceAddSummaries.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesQuantileStreamResourceDeserialize.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesQuantileStreamResourceFlush.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesQuantileStreamResourceGetBucketBoundaries.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesQuantileStreamResourceHandleOp.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesSerializeEnsemble.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesSparseAggregateStats.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesSparseCalculateBestFeatureSplit.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesTrainingPredict.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesUpdateEnsemble.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesUpdateEnsembleV2.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/IsBoostedTreesEnsembleInitialized.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/IsBoostedTreesQuantileStreamResourceInitialized.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/AdjustContrast.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/AdjustHue.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/AdjustSaturation.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/CombinedNonMaxSuppression.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/CropAndResize.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/CropAndResizeGradBoxes.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/CropAndResizeGradImage.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/DecodeAndCropJpeg.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/DecodeBmp.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/DecodeGif.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/DecodeJpeg.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/DecodePng.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/DrawBoundingBoxes.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/EncodeJpeg.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/EncodeJpegVariableQuality.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/EncodePng.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ExtractGlimpse.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ExtractImagePatches.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ExtractJpegShape.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/GenerateBoundingBoxProposals.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/HsvToRgb.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ImageProjectiveTransformV2.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/NearestNeighbors.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/NonMaxSuppression.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/NonMaxSuppressionWithOverlaps.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/QuantizedResizeBilinear.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/RandomCrop.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ResizeArea.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ResizeBicubic.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ResizeBicubicGrad.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ResizeBilinear.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ResizeBilinearGrad.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ResizeNearestNeighbor.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ResizeNearestNeighborGrad.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/RgbToHsv.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/SampleDistortedBoundingBox.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ScaleAndTranslate.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ScaleAndTranslateGrad.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/DecodeBase64.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/DecodeCompressed.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/DecodeCsv.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/DecodeJsonExample.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/DecodePaddedRaw.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/DecodeRaw.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/DeserializeManySparse.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/EncodeBase64.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/FifoQueue.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/FixedLengthRecordReader.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/IdentityReader.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/LmdbReader.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/MatchingFiles.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/PaddingFifoQueue.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ParseExample.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ParseSequenceExample.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ParseSingleExample.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ParseSingleSequenceExample.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ParseTensor.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/PriorityQueue.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/QueueClose.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/QueueDequeue.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/QueueDequeueMany.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/QueueDequeueUpTo.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/QueueEnqueue.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/QueueEnqueueMany.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/QueueIsClosed.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/QueueSize.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/RandomShuffleQueue.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ReadFile.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ReaderNumRecordsProduced.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ReaderNumWorkUnitsCompleted.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ReaderRead.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ReaderReadUpTo.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ReaderReset.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ReaderRestoreState.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ReaderSerializeState.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/SerializeManySparse.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/SerializeSparse.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/SerializeTensor.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ShardedFilename.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ShardedFilespec.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/TextLineReader.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/TfRecordReader.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/WholeFileReader.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/WriteFile.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BandPart.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BandedTriangularSolve.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchCholesky.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchCholeskyGrad.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixBandPart.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixDeterminant.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixDiag.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixDiagPart.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixInverse.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixSetDiag.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixSolve.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixSolveLs.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixTriangularSolve.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchSelfAdjointEig.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchSvd.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Cholesky.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/CholeskyGrad.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/ConjugateTranspose.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Cross.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Det.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Eig.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Einsum.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/EuclideanNorm.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Inv.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/LoadAndRemapMatrix.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/LogMatrixDeterminant.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Lu.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/MatMul.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/MatrixDiag.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/MatrixDiagPart.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/MatrixDiagPartV3.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/MatrixDiagV3.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/MatrixLogarithm.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/MatrixSetDiag.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/MatrixSolveLs.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Qr.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/QuantizedMatMul.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/QuantizedMatMulWithBias.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/QuantizedMatMulWithBiasAndRelu.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/QuantizedMatMulWithBiasAndReluAndRequantize.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/SelfAdjointEig.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Solve.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Sqrtm.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Svd.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/TensorDiag.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/TensorDiagPart.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Transpose.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/TriangularSolve.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/TridiagonalMatMul.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/TridiagonalSolve.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/CSRSparseMatrixComponents.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/CSRSparseMatrixToDense.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/CSRSparseMatrixToSparseTensor.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/DenseToCSRSparseMatrix.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixAdd.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixMatMul.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixMul.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixNNZ.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixOrderingAMD.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixSoftmax.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixSoftmaxGrad.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixSparseCholesky.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixSparseMatMul.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixTranspose.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixZeros.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseTensorToCSRSparseMatrix.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Abs.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/AccumulateN.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Acos.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Acosh.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Add.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/AddN.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Angle.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/ApproximateEqual.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/ArgMax.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/ArgMin.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Asin.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Asinh.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Atan.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Atan2.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Atanh.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/BesselI0.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/BesselI0e.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/BesselI1.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/BesselI1e.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Betainc.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Bincount.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Ceil.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/CompareAndBitpack.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/ComplexAbs.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Conj.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Cos.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Cosh.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Cumprod.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Cumsum.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/CumulativeLogsumexp.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/DenseBincount.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Digamma.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Div.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/DivNoNan.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Equal.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Erf.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Erfc.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Exp.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Expm1.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Fact.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Floor.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/FloorDiv.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/FloorMod.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Greater.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/GreaterEqual.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Igamma.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/IgammaGradA.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Igammac.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Imag.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/InvertPermutation.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/IsFinite.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/IsInf.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/IsNan.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Less.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/LessEqual.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Lgamma.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Log.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Log1p.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/LogicalAnd.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/LogicalNot.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/LogicalOr.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Maximum.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Mean.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Minimum.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Mod.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Mul.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/MulNoNan.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Ndtri.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Neg.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/NextAfter.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/NotEqual.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Polygamma.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/PopulationCount.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Pow.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/QuantizedAdd.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/QuantizedMul.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Real.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/RealDiv.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Reciprocal.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/ReciprocalGrad.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/RequantizationRangePerChannel.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/RequantizePerChannel.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Rint.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Round.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Rsqrt.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/RsqrtGrad.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SegmentMax.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SegmentMean.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SegmentMin.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SegmentProd.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SegmentSum.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Sigmoid.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SigmoidGrad.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Sign.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Sin.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Sinh.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SobolSample.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Softplus.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SoftplusGrad.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Sqrt.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SqrtGrad.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Square.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SquaredDifference.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Sub.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Tan.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Tanh.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/TanhGrad.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/TruncateDiv.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/TruncateMod.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/UnsortedSegmentMax.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/UnsortedSegmentMin.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/UnsortedSegmentProd.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/UnsortedSegmentSum.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Xdivy.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Xlog1py.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Xlogy.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Zeta.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/erfinv.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/BesselJ0.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/BesselJ1.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/BesselK0.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/BesselK0e.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/BesselK1.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/BesselK1e.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/BesselY0.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/BesselY1.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/Dawsn.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/Expint.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/FresnelCos.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/FresnelSin.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/Spence.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/AvgPool.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/AvgPool3d.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/AvgPool3dGrad.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/AvgPoolGrad.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/BatchNormWithGlobalNormalization.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/BatchNormWithGlobalNormalizationGrad.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/BiasAdd.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/BiasAddGrad.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/BlockLSTM.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/BlockLSTMGrad.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CTCLossV2.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/ComputeAccidentalHits.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Conv2d.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Conv2dBackpropFilter.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Conv2dBackpropInput.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Conv3d.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Conv3dBackpropFilter.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Conv3dBackpropInput.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CtcBeamSearchDecoder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CtcGreedyDecoder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CtcLoss.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CudnnRNN.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CudnnRNNBackprop.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CudnnRNNCanonicalToParams.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CudnnRNNParamsToCanonical.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CudnnRnnParamsSize.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/DataFormatDimMap.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/DataFormatVecPermute.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/DepthToSpace.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/DepthwiseConv2dNative.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/DepthwiseConv2dNativeBackpropFilter.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/DepthwiseConv2dNativeBackpropInput.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Dilation2d.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Dilation2dBackpropFilter.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Dilation2dBackpropInput.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Elu.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/EluGrad.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/FixedUnigramCandidateSampler.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/FractionalAvgPool.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/FractionalAvgPoolGrad.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/FractionalMaxPool.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/FractionalMaxPoolGrad.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/FusedBatchNorm.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/FusedBatchNormGrad.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/FusedPadConv2d.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/FusedResizeAndPadConv2d.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/GRUBlockCell.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/GRUBlockCellGrad.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/InTopK.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/InvGrad.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/L2Loss.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/LSTMBlockCell.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/LSTMBlockCellGrad.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/LearnedUnigramCandidateSampler.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/LocalResponseNormalization.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/LocalResponseNormalizationGrad.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/LogSoftmax.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPool.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPool3d.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPool3dGrad.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPool3dGradGrad.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPoolGrad.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPoolGradGrad.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPoolGradGradWithArgmax.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPoolGradWithArgmax.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPoolWithArgmax.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/NthElement.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedAvgPool.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedBatchNormWithGlobalNormalization.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedBiasAdd.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DAndRelu.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DAndReluAndRequantize.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DAndRequantize.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DPerChannel.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DWithBias.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DWithBiasAndRelu.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DWithBiasAndReluAndRequantize.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DWithBiasAndRequantize.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DWithBiasSignedSumAndReluAndRequantize.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DWithBiasSumAndRelu.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DWithBiasSumAndReluAndRequantize.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2d.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedDepthwiseConv2D.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedDepthwiseConv2DWithBias.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedDepthwiseConv2DWithBiasAndRelu.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedDepthwiseConv2DWithBiasAndReluAndRequantize.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedInstanceNorm.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedMaxPool.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedRelu.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedRelu6.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedReluX.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Relu.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Relu6.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Relu6Grad.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/ReluGrad.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Selu.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/SeluGrad.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Softmax.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Softsign.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/SoftsignGrad.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/SpaceToBatch.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/SpaceToDepth.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/TopK.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/raw/SoftmaxCrossEntropyWithLogits.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/raw/SparseSoftmaxCrossEntropyWithLogits.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/Dequantize.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/FakeQuantWithMinMaxArgs.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/FakeQuantWithMinMaxArgsGradient.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/FakeQuantWithMinMaxVars.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/FakeQuantWithMinMaxVarsGradient.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/FakeQuantWithMinMaxVarsPerChannel.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/FakeQuantWithMinMaxVarsPerChannelGradient.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/Quantize.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/QuantizeAndDequantize.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/QuantizeDownAndShrinkRange.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/QuantizedConcat.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/QuantizedMatMulWithBiasAndDequantize.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/QuantizedMatMulWithBiasAndRequantize.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/RequantizationRange.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/Requantize.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/ragged/RaggedBincount.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/ragged/RaggedCountSparseOutput.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/ragged/RaggedCross.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/ragged/RaggedGather.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/ragged/RaggedRange.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/ragged/RaggedTensorFromVariant.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/ragged/RaggedTensorToSparse.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/ragged/RaggedTensorToTensor.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/ragged/RaggedTensorToVariant.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/AllCandidateSampler.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/AnonymousRandomSeedGenerator.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/AnonymousSeedGenerator.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/DeleteRandomSeedGenerator.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/DeleteSeedGenerator.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/LogUniformCandidateSampler.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/Multinomial.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/NonDeterministicInts.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/ParameterizedTruncatedNormal.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RandomGamma.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RandomGammaGrad.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RandomPoisson.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RandomShuffle.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RandomStandardNormal.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RandomUniform.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RandomUniformInt.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RecordInput.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RngSkip.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatefulRandomBinomial.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatefulStandardNormal.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatefulTruncatedNormal.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatefulUniform.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatefulUniformFullInt.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatefulUniformInt.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessMultinomial.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessParameterizedTruncatedNormal.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessRandomBinomial.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessRandomGamma.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessRandomNormal.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessRandomPoisson.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessRandomUniform.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessRandomUniformFullInt.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessRandomUniformInt.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessTruncatedNormal.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/TruncatedNormal.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/UniformCandidateSampler.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/experimental/DummySeedGenerator.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/BatchFft.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/BatchFft2d.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/BatchFft3d.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/BatchIfft.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/BatchIfft2d.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/BatchIfft3d.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Fft.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Fft2d.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Fft3d.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Ifft.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Ifft2d.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Ifft3d.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Irfft.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Irfft2d.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Irfft3d.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Rfft.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Rfft2d.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Rfft3d.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/AddManySparseToTensorsMap.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/AddSparseToTensorsMap.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/DenseCountSparseOutput.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/DenseToDenseSetOperation.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/DenseToSparseSetOperation.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/DeserializeSparse.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseAccumulatorApplyGradient.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseAccumulatorTakeGradient.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseAdd.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseAddGrad.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseBincount.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseConcat.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseConditionalAccumulator.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseCountSparseOutput.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseCross.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseCrossHashed.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseDenseCwiseAdd.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseDenseCwiseDiv.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseDenseCwiseMul.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseFillEmptyRows.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseFillEmptyRowsGrad.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseMatMul.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseReduceMax.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseReduceMaxSparse.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseReduceSum.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseReduceSumSparse.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseReorder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseReshape.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentMean.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentMeanGrad.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentMeanWithNumSegments.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentSqrtN.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentSqrtNGrad.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentSqrtNWithNumSegments.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentSum.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentSumWithNumSegments.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSlice.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSliceGrad.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSoftmax.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSparseMaximum.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSparseMinimum.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSplit.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseTensorDenseAdd.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseTensorDenseMatMul.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseToDense.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseToSparseSetOperation.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/TakeManySparseFromTensorsMap.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/Join.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/Lower.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/ReduceJoin.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/RegexFullMatch.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/RegexReplace.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/StaticRegexFullMatch.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/StaticRegexReplace.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/StringFormat.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/StringLength.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/StringNGrams.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/StringSplit.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/Strip.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/Substr.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/ToHashBucket.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/ToHashBucketFast.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/ToHashBucketStrong.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/ToNumber.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/UnicodeDecode.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/UnicodeDecodeWithOffsets.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/UnicodeEncode.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/UnicodeScript.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/UnicodeTranscode.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/UnsortedSegmentJoin.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/Upper.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/AudioSummary.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/CloseSummaryWriter.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/CreateSummaryDbWriter.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/CreateSummaryFileWriter.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/FlushSummaryWriter.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/HistogramSummary.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/ImageSummary.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/ImportEvent.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/MergeSummary.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/ScalarSummary.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/StatsAggregatorSummary.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/SummaryWriter.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/TensorSummary.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/WriteAudioSummary.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/WriteGraphSummary.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/WriteHistogramSummary.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/WriteImageSummary.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/WriteRawProtoSummary.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/WriteScalarSummary.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/WriteSummary.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/AllToAll.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/CollectivePermute.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/ConfigureDistributedTPU.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/ConfigureTPUEmbedding.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/CrossReplicaSum.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/EnqueueTPUEmbeddingIntegerBatch.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/EnqueueTPUEmbeddingRaggedTensorBatch.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/EnqueueTPUEmbeddingSparseBatch.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/EnqueueTPUEmbeddingSparseTensorBatch.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/InfeedDequeue.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/InfeedDequeueTuple.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/InfeedEnqueue.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/InfeedEnqueuePrelinearizedBuffer.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/InfeedEnqueueTuple.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingADAMParameters.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingADAMParametersGradAccumDebug.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingAdadeltaParameters.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingAdadeltaParametersGradAccumDebug.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingAdagradParameters.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingAdagradParametersGradAccumDebug.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingCenteredRMSPropParameters.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingFTRLParameters.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingFTRLParametersGradAccumDebug.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingMDLAdagradLightParameters.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingMomentumParameters.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingMomentumParametersGradAccumDebug.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingProximalAdagradParameters.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingProximalAdagradParametersGradAccumDebug.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingProximalYogiParameters.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingProximalYogiParametersGradAccumDebug.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingRMSPropParameters.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingRMSPropParametersGradAccumDebug.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingStochasticGradientDescentParameters.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingStochasticGradientDescentParametersGradAccumDebug.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/OutfeedDequeue.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/OutfeedDequeueTuple.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/OutfeedEnqueue.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/OutfeedEnqueueTuple.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/Prelinearize.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/PrelinearizeTuple.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RecvTPUEmbeddingActivations.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingADAMParameters.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingADAMParametersGradAccumDebug.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingAdadeltaParameters.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingAdadeltaParametersGradAccumDebug.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingAdagradParameters.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingAdagradParametersGradAccumDebug.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingCenteredRMSPropParameters.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingFTRLParameters.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingFTRLParametersGradAccumDebug.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingMDLAdagradLightParameters.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingMomentumParameters.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingMomentumParametersGradAccumDebug.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingProximalAdagradParameters.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingProximalAdagradParametersGradAccumDebug.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingProximalYogiParameters.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingProximalYogiParametersGradAccumDebug.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingRMSPropParameters.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingRMSPropParametersGradAccumDebug.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingStochasticGradientDescentParameters.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingStochasticGradientDescentParametersGradAccumDebug.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/SendTPUEmbeddingGradients.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/ShutdownDistributedTPU.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/TPUCompilationResult.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/TPUEmbeddingActivations.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/TPUOrdinalSelector.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/TPUReplicateMetadata.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/TPUReplicatedInput.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/TPUReplicatedOutput.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/WorkerHeartbeat.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/AccumulatorApplyGradient.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/AccumulatorNumAccumulated.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/AccumulatorSetGlobalStep.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/AccumulatorTakeGradient.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyAdaMax.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyAdadelta.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyAdagrad.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyAdagradDa.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyAdagradV2.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyAdam.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyAddSign.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyCenteredRmsProp.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyFtrl.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyGradientDescent.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyMomentum.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyPowerSign.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyProximalAdagrad.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyProximalGradientDescent.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyRmsProp.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/BatchMatMul.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ConditionalAccumulator.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/GenerateVocabRemapping.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/MergeV2Checkpoints.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/NegTrain.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/PreventGradient.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceAccumulatorApplyGradient.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceAccumulatorNumAccumulated.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceAccumulatorSetGlobalStep.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceAccumulatorTakeGradient.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyAdaMax.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyAdadelta.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyAdagrad.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyAdagradDa.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyAdam.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyAdamWithAmsgrad.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyAddSign.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyCenteredRmsProp.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyFtrl.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyGradientDescent.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyKerasMomentum.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyMomentum.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyPowerSign.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyProximalAdagrad.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyProximalGradientDescent.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyRmsProp.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceConditionalAccumulator.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyAdadelta.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyAdagrad.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyAdagradDa.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyAdagradV2.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyCenteredRmsProp.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyFtrl.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyKerasMomentum.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyMomentum.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyProximalAdagrad.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyProximalGradientDescent.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyRmsProp.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/Restore.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/RestoreSlice.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/Save.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SaveSlices.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SdcaFprint.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SdcaOptimizer.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SdcaShrinkL1.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyAdadelta.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyAdagrad.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyAdagradDa.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyCenteredRmsProp.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyFtrl.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyMomentum.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyProximalAdagrad.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyProximalGradientDescent.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyRmsProp.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/TileGrad.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/BroadcastHelper.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/ClusterOutput.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Conv.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Dequantize.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Dot.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/DynamicSlice.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/DynamicUpdateSlice.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Einsum.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Gather.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/KeyValueSort.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Pad.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Recv.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/ReplicaId.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/SelfAdjointEig.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Send.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Sharding.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Sort.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Svd.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/distruntime/ClusterDef.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/distruntime/ClusterDefOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/distruntime/ClusterDeviceFilters.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/distruntime/ClusterDeviceFiltersOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/distruntime/ClusterProtos.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/distruntime/DeviceFiltersProtos.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/distruntime/JobDef.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/distruntime/JobDefOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/distruntime/JobDeviceFilters.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/distruntime/JobDeviceFiltersOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/distruntime/ServerDef.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/distruntime/ServerDefOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/distruntime/ServerProtos.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/distruntime/TaskDeviceFilters.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/distruntime/TaskDeviceFiltersOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/distruntime/TransportOptions.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/BytesList.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/BytesListOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/Example.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/ExampleOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/ExampleParserConfiguration.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/ExampleParserConfigurationOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/ExampleParserConfigurationProtos.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/ExampleProtos.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/Feature.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/FeatureConfiguration.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/FeatureConfigurationOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/FeatureList.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/FeatureListOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/FeatureLists.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/FeatureListsOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/FeatureOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/FeatureProtos.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/Features.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/FeaturesOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/FixedLenFeatureProto.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/FixedLenFeatureProtoOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/FloatList.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/FloatListOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/Int64List.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/Int64ListOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/SequenceExample.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/SequenceExampleOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/VarLenFeatureProto.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/VarLenFeatureProtoOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/AllocationDescription.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/AllocationDescriptionOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/AllocationDescriptionProtos.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/AllocationRecord.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/AllocationRecordOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/AllocatorMemoryUsed.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/AllocatorMemoryUsedOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ApiDef.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ApiDefOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ApiDefProtos.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ApiDefs.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ApiDefsOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/AssetFileDef.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/AssetFileDefOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/AttrValue.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/AttrValueOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/AttrValueProtos.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/AutoParallelOptions.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/AutoParallelOptionsOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/BoundedTensorSpecProto.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/BoundedTensorSpecProtoOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/CallableOptions.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/CallableOptionsOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/Code.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/CollectionDef.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/CollectionDefOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/CondContextDef.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/CondContextDefOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ConfigProto.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ConfigProtoOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ConfigProtos.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ControlFlowContextDef.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ControlFlowContextDefOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ControlFlowProtos.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/CostGraphDef.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/CostGraphDefOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/CostGraphProtos.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DataClass.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DataType.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DebugOptions.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DebugOptionsOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DebugProtos.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DebugTensorWatch.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DebugTensorWatchOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DebuggedSourceFile.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DebuggedSourceFileOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DebuggedSourceFiles.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DebuggedSourceFilesOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/Device.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DeviceAttributes.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DeviceAttributesOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DeviceAttributesProtos.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DeviceLocality.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DeviceLocalityOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DeviceOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DeviceProperties.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DevicePropertiesOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DevicePropertiesProtos.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DeviceStepStats.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DeviceStepStatsOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DictValue.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DictValueOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ErrorCodes.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ErrorCodesProtos.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/FunctionDef.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/FunctionDefLibrary.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/FunctionDefLibraryOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/FunctionDefOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/FunctionProtos.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/FunctionSpec.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/FunctionSpecOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GPUOptions.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GPUOptionsOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GradientDef.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GradientDefOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphDebugInfo.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphDebugInfoOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphDebugInfoProtos.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphDef.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphDefOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphOptions.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphOptionsOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphProtos.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphTransferConstNodeInfo.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphTransferConstNodeInfoOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphTransferGraphInputNodeInfo.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphTransferGraphInputNodeInfoOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphTransferGraphOutputNodeInfo.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphTransferGraphOutputNodeInfoOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphTransferInfo.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphTransferInfoOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphTransferInfoProto.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphTransferNodeInfo.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphTransferNodeInfoOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphTransferNodeInput.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphTransferNodeInputInfo.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphTransferNodeInputInfoOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphTransferNodeInputOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphTransferNodeOutputInfo.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphTransferNodeOutputInfoOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/HistogramProto.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/HistogramProtoOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/InterconnectLink.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/InterconnectLinkOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/KernelDef.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/KernelDefOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/KernelDefProtos.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/KernelList.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/KernelListOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ListValue.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ListValueOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/LocalLinks.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/LocalLinksOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/LogMemoryProtos.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/MemoryLogRawAllocation.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/MemoryLogRawAllocationOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/MemoryLogRawDeallocation.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/MemoryLogRawDeallocationOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/MemoryLogStep.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/MemoryLogStepOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/MemoryLogTensorAllocation.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/MemoryLogTensorAllocationOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/MemoryLogTensorDeallocation.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/MemoryLogTensorDeallocationOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/MemoryLogTensorOutput.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/MemoryLogTensorOutputOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/MemoryStats.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/MemoryStatsOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/MetaGraphDef.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/MetaGraphDefOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/MetaGraphProtos.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NameAttrList.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NameAttrListOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NamedDevice.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NamedDeviceOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NamedTensorProto.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NamedTensorProtoOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NamedTensorProtos.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NamedTupleValue.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NamedTupleValueOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NodeDef.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NodeDefOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NodeExecStats.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NodeExecStatsOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NodeOutput.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NodeOutputOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NodeProto.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NoneValue.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NoneValueOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/OpDef.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/OpDefOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/OpDefProtos.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/OpDeprecation.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/OpDeprecationOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/OpList.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/OpListOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/OptimizerOptions.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/OptimizerOptionsOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/PairValue.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/PairValueOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/QueueRunnerDef.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/QueueRunnerDefOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/QueueRunnerProtos.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/RPCOptions.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/RPCOptionsOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ReaderBaseProtos.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ReaderBaseState.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ReaderBaseStateOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/RemoteFusedGraphExecuteInfo.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/RemoteFusedGraphExecuteInfoOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/RemoteFusedGraphExecuteInfoProto.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/RemoteTensorHandle.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/RemoteTensorHandleOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/RemoteTensorHandleProtos.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/Resource.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ResourceDtypeAndShape.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ResourceDtypeAndShapeOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ResourceHandle.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ResourceHandleProto.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ResourceHandleProtoOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ResourceOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/RewriterConfig.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/RewriterConfigOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/RewriterConfigProtos.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/RunMetadata.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/RunMetadataOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/RunOptions.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/RunOptionsOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SaveSliceInfoDef.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SaveSliceInfoDefOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SaveableObject.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SaveableObjectOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedAsset.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedAssetOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedBareConcreteFunction.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedBareConcreteFunctionOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedConcreteFunction.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedConcreteFunctionOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedConstant.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedConstantOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedFunction.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedFunctionOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedModel.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedModelOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedModelProtos.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedObject.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedObjectGraph.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedObjectGraphOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedObjectGraphProtos.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedObjectOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedResource.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedResourceOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedUserObject.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedUserObjectOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedVariable.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedVariableOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ScopedAllocatorOptions.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ScopedAllocatorOptionsOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SessionMetadata.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SessionMetadataOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SignatureDef.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SignatureDefOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/StepStats.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/StepStatsOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/StepStatsProtos.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/StructProtos.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/StructuredValue.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/StructuredValueOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/Summary.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SummaryDescription.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SummaryDescriptionOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SummaryMetadata.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SummaryMetadataOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SummaryOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SummaryProtos.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorConnection.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorConnectionOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorDescription.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorDescriptionOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorDescriptionProtos.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorInfo.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorInfoOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorProto.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorProtoOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorProtos.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorShapeProto.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorShapeProtoOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorShapeProtos.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorSliceProto.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorSliceProtoOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorSliceProtos.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorSpecProto.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorSpecProtoOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ThreadPoolOptionProto.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ThreadPoolOptionProtoOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/Trace.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TraceEvent.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TraceEventOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TraceEventsProtos.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TraceOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TrackableObjectGraph.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TrackableObjectGraphOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TrackableObjectGraphProtos.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TupleValue.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TupleValueOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TypeSpecProto.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TypeSpecProtoOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TypesProtos.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ValuesDef.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ValuesDefOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/VariableAggregation.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/VariableDef.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/VariableDefOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/VariableProtos.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/VariableSynchronization.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/VariantTensorDataProto.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/VariantTensorDataProtoOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/VerifierConfig.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/VerifierConfigOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/VerifierConfigProtos.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/VersionDef.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/VersionDefOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/VersionsProtos.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/WhileContextDef.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/WhileContextDefOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/ProfileOptions.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/ProfileOptionsOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/ProfilerOptionsProtos.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/XEvent.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/XEventMetadata.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/XEventMetadataOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/XEventOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/XLine.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/XLineOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/XPlane.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/XPlaneOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/XPlaneProtos.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/XSpace.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/XSpaceOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/XStat.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/XStatMetadata.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/XStatMetadataOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/XStatOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/BfcMemoryMapProtos.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/BinSummary.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/BinSummaryOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/BundleEntryProto.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/BundleEntryProtoOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/BundleHeaderProto.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/BundleHeaderProtoOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/CodeLocation.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/CodeLocationOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/DebugEvent.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/DebugEventOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/DebugEventProtos.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/DebugMetadata.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/DebugMetadataOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/DebuggedDevice.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/DebuggedDeviceOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/DebuggedGraph.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/DebuggedGraphOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/Event.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/EventOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/EventProtos.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/Execution.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/ExecutionOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/GraphExecutionTrace.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/GraphExecutionTraceOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/GraphOpCreation.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/GraphOpCreationOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/LogMessage.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/LogMessageOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/MemAllocatorStats.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/MemAllocatorStatsOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/MemChunk.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/MemChunkOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/MemmappedFileSystemDirectory.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/MemmappedFileSystemDirectoryElement.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/MemmappedFileSystemDirectoryElementOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/MemmappedFileSystemDirectoryOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/MemmappedFileSystemProtos.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/MemoryDump.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/MemoryDumpOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/RequestedExitCode.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/RequestedExitCodeOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SavedSlice.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SavedSliceMeta.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SavedSliceMetaOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SavedSliceOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SavedTensorSliceMeta.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SavedTensorSliceMetaOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SavedTensorSliceProtos.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SavedTensorSlices.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SavedTensorSlicesOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SaverDef.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SaverDefOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SaverProtos.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SessionLog.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SessionLogOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SnapShot.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SnapShotOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SnapshotMetadataRecord.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SnapshotMetadataRecordOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SnapshotProtos.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SnapshotRecord.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SnapshotRecordOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SnapshotTensorMetadata.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SnapshotTensorMetadataOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SourceFile.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SourceFileOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/StackFrameWithId.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/StackFrameWithIdOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/TaggedRunMetadata.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/TaggedRunMetadataOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/TensorBundleProtos.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/TensorDebugMode.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/TensorMetadata.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/TensorMetadataOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/WatchdogConfig.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/WatchdogConfigOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/WorkerHealth.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/WorkerHeartbeatRequest.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/WorkerHeartbeatRequestOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/WorkerHeartbeatResponse.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/WorkerHeartbeatResponseOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/WorkerShutdownMode.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/AvailableDeviceInfo.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/AvailableDeviceInfoOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/BenchmarkEntries.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/BenchmarkEntriesOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/BenchmarkEntry.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/BenchmarkEntryOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/BuildConfiguration.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/BuildConfigurationOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/CPUInfo.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/CPUInfoOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/CommitId.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/CommitIdOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/EntryValue.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/EntryValueOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/GPUInfo.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/GPUInfoOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/MachineConfiguration.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/MachineConfigurationOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/MemoryInfo.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/MemoryInfoOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/MetricEntry.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/MetricEntryOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/PlatformInfo.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/PlatformInfoOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/RunConfiguration.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/RunConfigurationOrBuilder.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/TestLogProtos.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/TestResults.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/TestResultsOrBuilder.java
 rename tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/{AbstractTensor.java => TensorHandle.java} (66%)
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/TensorType.java
 rename tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/{tensor => }/buffer/ByteSequenceTensorBuffer.java (99%)
 rename tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/{tensor => }/buffer/TensorBuffers.java (99%)
 rename tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/{tensor => }/buffer/TensorRawDataBufferFactory.java (98%)
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/BooleanTensorImpl.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/ByteTensorImpl.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/DoubleTensorImpl.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/FloatTensorImpl.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/IntTensorImpl.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/LongTensorImpl.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/StringTensorImpl.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/types/TBfloat16Factory.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/types/TBoolFactory.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/types/TFloat16Factory.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/types/TFloat32Factory.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/types/TFloat64Factory.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/types/TInt32Factory.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/types/TInt64Factory.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/types/TStringFactory.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/types/TUint8Factory.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/Type.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TypeFactory.java
 rename tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/{TensorTypes.java => types/TypeRegistry.java} (53%)
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/tensor/BooleanTensor.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/tensor/ByteTensor.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/tensor/DoubleTensor.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/tensor/FloatTensor.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/tensor/IntTensor.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/tensor/LongTensor.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/tensor/StringTensor.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/util/TensorList.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/util/TensorMap.java
 delete mode 100644 tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/AutoCloseableList.java

diff --git a/tensorflow-core/tensorflow-core-api/src/bazel/op_generator/op_generator.cc b/tensorflow-core/tensorflow-core-api/src/bazel/op_generator/op_generator.cc
index 4ad39a07a2a..a3b76b71e70 100644
--- a/tensorflow-core/tensorflow-core-api/src/bazel/op_generator/op_generator.cc
+++ b/tensorflow-core/tensorflow-core-api/src/bazel/op_generator/op_generator.cc
@@ -348,7 +348,7 @@ void RenderInterfaceImpl(const OpSpec& op, RenderMode mode,
   if (mode == OPERAND) {
     bool cast2obj = output.type().wildcard();
     Type return_type = Type::Class("Output", "org.tensorflow")
-        .add_parameter(cast2obj ? Type::Interface("TType", "org.tensorflow.types.family") : output.type());
+        .add_parameter(cast2obj ? Type::Class("TType", "org.tensorflow.types.family") : output.type());
     Method as_output = Method::Create("asOutput", return_type)
         .add_argument(Variable::Create("scope", Type::Class("Scope", "org.tensorflow.op")))
         .add_annotation(Annotation::Create("Override"));
@@ -367,7 +367,7 @@ void RenderInterfaceImpl(const OpSpec& op, RenderMode mode,
   } else if (mode == LIST_OPERAND) {
     Type operand = Type::Interface("Operand", "org.tensorflow");
     if (output.type().wildcard()) {
-      operand.add_parameter(Type::Interface("TType", "org.tensorflow.types.family"));
+      operand.add_parameter(Type::Class("TType", "org.tensorflow.types.family"));
     } else {
       operand.add_parameter(output.type());
     }
@@ -431,9 +431,10 @@ void GenerateOp(const OpSpec& op, const EndpointSpec& endpoint,
   RenderMode mode = DEFAULT;
   if (op.outputs().size() == 1) {
     const ArgumentSpec& output = op.outputs().front();
-    Type operand_type(output.type().wildcard() ? Type::Interface("TType", "org.tensorflow.types.family")
+    Type operand_type(output.type().wildcard() ? Type::Class("TType", "org.tensorflow.types.family")
                                                : output.type());
-    Type operand_inf(Type::Interface("Operand", "org.tensorflow").add_parameter(operand_type));
+    Type operand_inf(Type::Interface("Operand", "org.tensorflow")
+                         .add_parameter(operand_type));
     if (output.iterable()) {
       mode = LIST_OPERAND;
       op_class.add_supertype(Type::IterableOf(operand_inf));
diff --git a/tensorflow-core/tensorflow-core-api/src/bazel/op_generator/op_specs.cc b/tensorflow-core/tensorflow-core-api/src/bazel/op_generator/op_specs.cc
index b7452c8f5b6..958e641610e 100644
--- a/tensorflow-core/tensorflow-core-api/src/bazel/op_generator/op_specs.cc
+++ b/tensorflow-core/tensorflow-core-api/src/bazel/op_generator/op_specs.cc
@@ -65,13 +65,6 @@ class TypeResolver {
   std::pair<Type, Type> TypesOf(const OpDef_AttrDef& attr_def,
                                 bool* iterable_out);
 
-  // Returns the highest type family this attribute is part of
-  //
-  // For example, if the attribute is of type 'bool', the base 'TType' family
-  // returned. But if it represents a number, like a float or an integer,
-  // then 'TNumber' (which supersedes 'TType') is returned.
-  Type FamilyOf(const OpDef_AttrDef& attr_def);
-
   // Returns true if the type of this attribute has already been resolved
   bool IsAttributeVisited(const string& attr_name) {
     return visited_attrs_.find(attr_name) != visited_attrs_.cend();
@@ -88,12 +81,19 @@ class TypeResolver {
   std::pair<Type, Type> MakeTypePair(const Type& type) {
     return std::make_pair(type, type);
   }
-  Type NextGeneric(const Type& typeFamily) {
+  Type NextGeneric(const OpDef_AttrDef& attr_def) {
     char generic_letter = next_generic_letter_++;
     if (next_generic_letter_ > 'Z') {
       next_generic_letter_ = 'A';
     }
-    return Type::Generic(string(1, generic_letter)).add_supertype(typeFamily);
+    Type generic_type = Type::Generic(string(1, generic_letter));
+    // TODO(karllessard) support more type families
+    if (IsRealNumbers(attr_def.allowed_values())) {
+      generic_type.add_supertype(Type::Interface("TNumber", "org.tensorflow.types.family"));
+    } else {
+      generic_type.add_supertype(Type::Interface("TType", "org.tensorflow.types.family"));
+    }
+    return generic_type;
   }
 };
 
@@ -158,11 +158,11 @@ std::pair<Type, Type> TypeResolver::TypesOf(const OpDef_AttrDef& attr_def,
     types = MakeTypePair(Type::Class("Shape", "org.tensorflow.ndarray"));
 
   } else if (attr_type == "tensor") {
-    types = MakeTypePair(Type::Class("Tensor", "org.tensorflow")
-                             .add_parameter(Type::Wildcard()));
+    types = MakeTypePair(Type::Class("TType", "org.tensorflow.types.family")
+        .add_parameter(Type::Wildcard()));
 
   } else if (attr_type == "type") {
-    Type type = *iterable_out ? Type::Wildcard() : NextGeneric(FamilyOf(attr_def));
+    Type type = *iterable_out ? Type::Wildcard() : NextGeneric(attr_def);
     types = MakeTypePair(type, Type::Class("Class"));
 
   } else {
@@ -173,14 +173,6 @@ std::pair<Type, Type> TypeResolver::TypesOf(const OpDef_AttrDef& attr_def,
   return types;
 }
 
-Type TypeResolver::FamilyOf(const OpDef_AttrDef& attr_def) {
-  // TODO (karlllessard): add more type families
-  if (IsRealNumbers(attr_def.allowed_values())) {
-    return Type::Interface("TNumber", "org.tensorflow.types.family");
-  }
-  return Type::Interface("TType", "org.tensorflow.types.family");
-}
-
 string SnakeToCamelCase(const string& str, bool upper = false) {
   string result;
   bool cap = upper;
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/Compute_func_Pointer_TF_OpKernelContext.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/Compute_func_Pointer_TF_OpKernelContext.java
deleted file mode 100644
index fd70471e100..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/Compute_func_Pointer_TF_OpKernelContext.java
+++ /dev/null
@@ -1,19 +0,0 @@
-// Targeted by JavaCPP version 1.5.4: DO NOT EDIT THIS FILE
-
-package org.tensorflow.internal.c_api;
-
-import java.nio.*;
-import org.bytedeco.javacpp.*;
-import org.bytedeco.javacpp.annotation.*;
-
-import static org.tensorflow.internal.c_api.global.tensorflow.*;
-
-@Properties(inherit = org.tensorflow.internal.c_api.presets.tensorflow.class)
-public class Compute_func_Pointer_TF_OpKernelContext extends FunctionPointer {
-    static { Loader.load(); }
-    /** Pointer cast constructor. Invokes {@link Pointer#Pointer(Pointer)}. */
-    public    Compute_func_Pointer_TF_OpKernelContext(Pointer p) { super(p); }
-    protected Compute_func_Pointer_TF_OpKernelContext() { allocate(); }
-    private native void allocate();
-    public native void call(Pointer arg0, TF_OpKernelContext arg1);
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/Create_func_TF_OpKernelConstruction.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/Create_func_TF_OpKernelConstruction.java
deleted file mode 100644
index 8f951ea6a73..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/Create_func_TF_OpKernelConstruction.java
+++ /dev/null
@@ -1,48 +0,0 @@
-// Targeted by JavaCPP version 1.5.4: DO NOT EDIT THIS FILE
-
-package org.tensorflow.internal.c_api;
-
-import java.nio.*;
-import org.bytedeco.javacpp.*;
-import org.bytedeco.javacpp.annotation.*;
-
-import static org.tensorflow.internal.c_api.global.tensorflow.*;
-
-
-// Allocates a new kernel builder and returns a pointer to it.
-//
-// If non-null, TensorFlow will call create_func when it needs to instantiate
-// the kernel. The pointer returned by create_func will be passed to
-// compute_func and delete_func, thereby functioning as a "this" pointer for
-// referring to kernel instances.
-//
-// The TF_OpKernelConstruction pointer passed to create_func is owned by
-// TensorFlow and will be deleted once create_func returns. It must not be used
-// after this.
-//
-// When TensorFlow needs to perform a computation with this kernel, it will
-// call compute_func. This function will receive the pointer returned by
-// create_func (or null if no create_func was provided), along with the inputs
-// to the computation.
-//
-// The TF_OpKernelContext pointer received by compute_func is owned by
-// TensorFlow and will be deleted once compute_func returns. It must not be used
-// after this.
-//
-// Finally, when TensorFlow no longer needs the kernel, it will call
-// delete_func if one is provided. This function will receive the pointer
-// returned in `create_func` or nullptr if no `create_func` was provided.
-//
-// The caller should pass the result of this function to
-// TF_RegisterKernelBuilder, which will take ownership of the pointer. If, for
-// some reason, the kernel builder will not be registered, the caller should
-// delete it with TF_DeleteKernelBuilder.
-@Properties(inherit = org.tensorflow.internal.c_api.presets.tensorflow.class)
-public class Create_func_TF_OpKernelConstruction extends FunctionPointer {
-    static { Loader.load(); }
-    /** Pointer cast constructor. Invokes {@link Pointer#Pointer(Pointer)}. */
-    public    Create_func_TF_OpKernelConstruction(Pointer p) { super(p); }
-    protected Create_func_TF_OpKernelConstruction() { allocate(); }
-    private native void allocate();
-    public native Pointer call(TF_OpKernelConstruction arg0);
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/Deallocator_Pointer_long_Pointer.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/Deallocator_Pointer_long_Pointer.java
deleted file mode 100644
index 168135a5b14..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/Deallocator_Pointer_long_Pointer.java
+++ /dev/null
@@ -1,30 +0,0 @@
-// Targeted by JavaCPP version 1.5.4: DO NOT EDIT THIS FILE
-
-package org.tensorflow.internal.c_api;
-
-import java.nio.*;
-import org.bytedeco.javacpp.*;
-import org.bytedeco.javacpp.annotation.*;
-
-import static org.tensorflow.internal.c_api.global.tensorflow.*;
-
-
-// Return a new tensor that holds the bytes data[0,len-1].
-//
-// The data will be deallocated by a subsequent call to TF_DeleteTensor via:
-//      (*deallocator)(data, len, deallocator_arg)
-// Clients must provide a custom deallocator function so they can pass in
-// memory managed by something like numpy.
-//
-// May return NULL (and invoke the deallocator) if the provided data buffer
-// (data, len) is inconsistent with a tensor of the given TF_DataType
-// and the shape specified by (dima, num_dims).
-@Properties(inherit = org.tensorflow.internal.c_api.presets.tensorflow.class)
-public class Deallocator_Pointer_long_Pointer extends FunctionPointer {
-    static { Loader.load(); }
-    /** Pointer cast constructor. Invokes {@link Pointer#Pointer(Pointer)}. */
-    public    Deallocator_Pointer_long_Pointer(Pointer p) { super(p); }
-    protected Deallocator_Pointer_long_Pointer() { allocate(); }
-    private native void allocate();
-    public native void call(Pointer data, @Cast("size_t") long len, Pointer arg);
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/Delete_func_Pointer.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/Delete_func_Pointer.java
deleted file mode 100644
index 734040f20f4..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/Delete_func_Pointer.java
+++ /dev/null
@@ -1,19 +0,0 @@
-// Targeted by JavaCPP version 1.5.4: DO NOT EDIT THIS FILE
-
-package org.tensorflow.internal.c_api;
-
-import java.nio.*;
-import org.bytedeco.javacpp.*;
-import org.bytedeco.javacpp.annotation.*;
-
-import static org.tensorflow.internal.c_api.global.tensorflow.*;
-
-@Properties(inherit = org.tensorflow.internal.c_api.presets.tensorflow.class)
-public class Delete_func_Pointer extends FunctionPointer {
-    static { Loader.load(); }
-    /** Pointer cast constructor. Invokes {@link Pointer#Pointer(Pointer)}. */
-    public    Delete_func_Pointer(Pointer p) { super(p); }
-    protected Delete_func_Pointer() { allocate(); }
-    private native void allocate();
-    public native void call(Pointer arg0);
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/Listener_BytePointer.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/Listener_BytePointer.java
deleted file mode 100644
index f1775998256..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/Listener_BytePointer.java
+++ /dev/null
@@ -1,25 +0,0 @@
-// Targeted by JavaCPP version 1.5.4: DO NOT EDIT THIS FILE
-
-package org.tensorflow.internal.c_api;
-
-import java.nio.*;
-import org.bytedeco.javacpp.*;
-import org.bytedeco.javacpp.annotation.*;
-
-import static org.tensorflow.internal.c_api.global.tensorflow.*;
-
-
-// Register a listener method that processes printed messages.
-//
-// If any listeners are registered, the print operator will call all listeners
-// with the printed messages and immediately return without writing to the
-// logs.
-@Properties(inherit = org.tensorflow.internal.c_api.presets.tensorflow.class)
-public class Listener_BytePointer extends FunctionPointer {
-    static { Loader.load(); }
-    /** Pointer cast constructor. Invokes {@link Pointer#Pointer(Pointer)}. */
-    public    Listener_BytePointer(Pointer p) { super(p); }
-    protected Listener_BytePointer() { allocate(); }
-    private native void allocate();
-    public native void call(@Cast("const char*") BytePointer arg0);
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/Listener_String.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/Listener_String.java
deleted file mode 100644
index a4114f23dbe..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/Listener_String.java
+++ /dev/null
@@ -1,19 +0,0 @@
-// Targeted by JavaCPP version 1.5.4: DO NOT EDIT THIS FILE
-
-package org.tensorflow.internal.c_api;
-
-import java.nio.*;
-import org.bytedeco.javacpp.*;
-import org.bytedeco.javacpp.annotation.*;
-
-import static org.tensorflow.internal.c_api.global.tensorflow.*;
-
-@Properties(inherit = org.tensorflow.internal.c_api.presets.tensorflow.class)
-public class Listener_String extends FunctionPointer {
-    static { Loader.load(); }
-    /** Pointer cast constructor. Invokes {@link Pointer#Pointer(Pointer)}. */
-    public    Listener_String(Pointer p) { super(p); }
-    protected Listener_String() { allocate(); }
-    private native void allocate();
-    public native void call(String arg0);
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/Shape_inference_func_TF_ShapeInferenceContext_TF_Status.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/Shape_inference_func_TF_ShapeInferenceContext_TF_Status.java
deleted file mode 100644
index 56e99923a0b..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/Shape_inference_func_TF_ShapeInferenceContext_TF_Status.java
+++ /dev/null
@@ -1,22 +0,0 @@
-// Targeted by JavaCPP version 1.5.4: DO NOT EDIT THIS FILE
-
-package org.tensorflow.internal.c_api;
-
-import java.nio.*;
-import org.bytedeco.javacpp.*;
-import org.bytedeco.javacpp.annotation.*;
-
-import static org.tensorflow.internal.c_api.global.tensorflow.*;
-
-
-// Sets the shape inference function for the op.
-@Properties(inherit = org.tensorflow.internal.c_api.presets.tensorflow.class)
-public class Shape_inference_func_TF_ShapeInferenceContext_TF_Status extends FunctionPointer {
-    static { Loader.load(); }
-    /** Pointer cast constructor. Invokes {@link Pointer#Pointer(Pointer)}. */
-    public    Shape_inference_func_TF_ShapeInferenceContext_TF_Status(Pointer p) { super(p); }
-    protected Shape_inference_func_TF_ShapeInferenceContext_TF_Status() { allocate(); }
-    private native void allocate();
-    public native void call(TF_ShapeInferenceContext ctx,
-                                 TF_Status status);
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TFE_Context.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TFE_Context.java
deleted file mode 100644
index 886782d2e0b..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TFE_Context.java
+++ /dev/null
@@ -1,25 +0,0 @@
-// Targeted by JavaCPP version 1.5.4: DO NOT EDIT THIS FILE
-
-package org.tensorflow.internal.c_api;
-
-import java.nio.*;
-import org.bytedeco.javacpp.*;
-import org.bytedeco.javacpp.annotation.*;
-
-import static org.tensorflow.internal.c_api.global.tensorflow.*;
-
-
-// "Context" under which operations/functions are executed. It encapsulates
-// things like the available devices, resource manager etc.
-// TFE_Context must outlive all tensor handles created using it. In other
-// words, TFE_DeleteContext() must be called after all tensor handles have
-// been deleted (with TFE_DeleteTensorHandle).
-//
-// TODO(ashankar): Merge with TF_Session?
-@Opaque @Properties(inherit = org.tensorflow.internal.c_api.presets.tensorflow.class)
-public class TFE_Context extends org.tensorflow.internal.c_api.AbstractTFE_Context {
-    /** Empty constructor. Calls {@code super((Pointer)null)}. */
-    public TFE_Context() { super((Pointer)null); }
-    /** Pointer cast constructor. Invokes {@link Pointer#Pointer(Pointer)}. */
-    public TFE_Context(Pointer p) { super(p); }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TFE_ContextOptions.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TFE_ContextOptions.java
deleted file mode 100644
index 7d986372863..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TFE_ContextOptions.java
+++ /dev/null
@@ -1,19 +0,0 @@
-// Targeted by JavaCPP version 1.5.4: DO NOT EDIT THIS FILE
-
-package org.tensorflow.internal.c_api;
-
-import java.nio.*;
-import org.bytedeco.javacpp.*;
-import org.bytedeco.javacpp.annotation.*;
-
-import static org.tensorflow.internal.c_api.global.tensorflow.*;
-
-// #endif
-
-@Opaque @Properties(inherit = org.tensorflow.internal.c_api.presets.tensorflow.class)
-public class TFE_ContextOptions extends org.tensorflow.internal.c_api.AbstractTFE_ContextOptions {
-    /** Empty constructor. Calls {@code super((Pointer)null)}. */
-    public TFE_ContextOptions() { super((Pointer)null); }
-    /** Pointer cast constructor. Invokes {@link Pointer#Pointer(Pointer)}. */
-    public TFE_ContextOptions(Pointer p) { super(p); }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TFE_Op.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TFE_Op.java
deleted file mode 100644
index b960617ab40..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TFE_Op.java
+++ /dev/null
@@ -1,29 +0,0 @@
-// Targeted by JavaCPP version 1.5.4: DO NOT EDIT THIS FILE
-
-package org.tensorflow.internal.c_api;
-
-import java.nio.*;
-import org.bytedeco.javacpp.*;
-import org.bytedeco.javacpp.annotation.*;
-
-import static org.tensorflow.internal.c_api.global.tensorflow.*;
-
-
-// Description of the TensorFlow op to execute.
-//
-// Assumes that the provided 'ctx' outlives the returned TFE_Op, i.e.,
-// TFE_DeleteOp() is called before TFE_DeleteContext().
-//
-// Very similar to TF_OperationDescription with some differences:
-// (1) TF_Output or TFE_TensorHandle* as arguments to TF_AddInput,
-//     TF_AddInputList
-// (2) TF_ColocateWith, TF_AddControlInput etc. do not make sense.
-// (3) Implementation detail: Avoid use of NodeBuilder/NodeDefBuilder since
-//     the additional sanity checks there seem unnecessary;
-@Opaque @Properties(inherit = org.tensorflow.internal.c_api.presets.tensorflow.class)
-public class TFE_Op extends org.tensorflow.internal.c_api.AbstractTFE_Op {
-    /** Empty constructor. Calls {@code super((Pointer)null)}. */
-    public TFE_Op() { super((Pointer)null); }
-    /** Pointer cast constructor. Invokes {@link Pointer#Pointer(Pointer)}. */
-    public TFE_Op(Pointer p) { super(p); }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TFE_TensorDebugInfo.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TFE_TensorDebugInfo.java
deleted file mode 100644
index e42ebf581b8..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TFE_TensorDebugInfo.java
+++ /dev/null
@@ -1,22 +0,0 @@
-// Targeted by JavaCPP version 1.5.4: DO NOT EDIT THIS FILE
-
-package org.tensorflow.internal.c_api;
-
-import java.nio.*;
-import org.bytedeco.javacpp.*;
-import org.bytedeco.javacpp.annotation.*;
-
-import static org.tensorflow.internal.c_api.global.tensorflow.*;
-
-
-// Debugging/Profiling information for TFE_TensorHandle
-//
-// TFE_TensorDebugInfo contains information useful for debugging and
-// profiling tensors.
-@Opaque @Properties(inherit = org.tensorflow.internal.c_api.presets.tensorflow.class)
-public class TFE_TensorDebugInfo extends Pointer {
-    /** Empty constructor. Calls {@code super((Pointer)null)}. */
-    public TFE_TensorDebugInfo() { super((Pointer)null); }
-    /** Pointer cast constructor. Invokes {@link Pointer#Pointer(Pointer)}. */
-    public TFE_TensorDebugInfo(Pointer p) { super(p); }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TFE_TensorHandle.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TFE_TensorHandle.java
deleted file mode 100644
index 00fab08ff54..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TFE_TensorHandle.java
+++ /dev/null
@@ -1,23 +0,0 @@
-// Targeted by JavaCPP version 1.5.4: DO NOT EDIT THIS FILE
-
-package org.tensorflow.internal.c_api;
-
-import java.nio.*;
-import org.bytedeco.javacpp.*;
-import org.bytedeco.javacpp.annotation.*;
-
-import static org.tensorflow.internal.c_api.global.tensorflow.*;
-
-
-// A handle to a tensor on a device.
-//
-// Like a TF_Tensor, a TFE_TensorHandle refers to a tensor with a value, shape,
-// type etc. Unlike a TF_Tensor, a TFE_TensorHandle may refer to such tensors
-// placed in memory of different devices or remote address spaces.
-@Opaque @Properties(inherit = org.tensorflow.internal.c_api.presets.tensorflow.class)
-public class TFE_TensorHandle extends org.tensorflow.internal.c_api.AbstractTFE_TensorHandle {
-    /** Empty constructor. Calls {@code super((Pointer)null)}. */
-    public TFE_TensorHandle() { super((Pointer)null); }
-    /** Pointer cast constructor. Invokes {@link Pointer#Pointer(Pointer)}. */
-    public TFE_TensorHandle(Pointer p) { super(p); }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_ApiDefMap.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_ApiDefMap.java
deleted file mode 100644
index cfe167e125a..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_ApiDefMap.java
+++ /dev/null
@@ -1,26 +0,0 @@
-// Targeted by JavaCPP version 1.5.4: DO NOT EDIT THIS FILE
-
-package org.tensorflow.internal.c_api;
-
-import java.nio.*;
-import org.bytedeco.javacpp.*;
-import org.bytedeco.javacpp.annotation.*;
-
-import static org.tensorflow.internal.c_api.global.tensorflow.*;
-
-
-// TF_ApiDefMap encapsulates a collection of API definitions for an operation.
-//
-// This object maps the name of a TensorFlow operation to a description of the
-// API to generate for it, as defined by the ApiDef protocol buffer (
-// https://www.tensorflow.org/code/tensorflow/core/framework/api_def.proto)
-//
-// The ApiDef messages are typically used to generate convenience wrapper
-// functions for TensorFlow operations in various language bindings.
-@Opaque @Properties(inherit = org.tensorflow.internal.c_api.presets.tensorflow.class)
-public class TF_ApiDefMap extends Pointer {
-    /** Empty constructor. Calls {@code super((Pointer)null)}. */
-    public TF_ApiDefMap() { super((Pointer)null); }
-    /** Pointer cast constructor. Invokes {@link Pointer#Pointer(Pointer)}. */
-    public TF_ApiDefMap(Pointer p) { super(p); }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_AttrMetadata.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_AttrMetadata.java
deleted file mode 100644
index 50d82309003..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_AttrMetadata.java
+++ /dev/null
@@ -1,58 +0,0 @@
-// Targeted by JavaCPP version 1.5.4: DO NOT EDIT THIS FILE
-
-package org.tensorflow.internal.c_api;
-
-import java.nio.*;
-import org.bytedeco.javacpp.*;
-import org.bytedeco.javacpp.annotation.*;
-
-import static org.tensorflow.internal.c_api.global.tensorflow.*;
-
-
-// TF_AttrMetadata describes the value of an attribute on an operation.
-@Properties(inherit = org.tensorflow.internal.c_api.presets.tensorflow.class)
-public class TF_AttrMetadata extends Pointer {
-    static { Loader.load(); }
-    /** Default native constructor. */
-    public TF_AttrMetadata() { super((Pointer)null); allocate(); }
-    /** Native array allocator. Access with {@link Pointer#position(long)}. */
-    public TF_AttrMetadata(long size) { super((Pointer)null); allocateArray(size); }
-    /** Pointer cast constructor. Invokes {@link Pointer#Pointer(Pointer)}. */
-    public TF_AttrMetadata(Pointer p) { super(p); }
-    private native void allocate();
-    private native void allocateArray(long size);
-    @Override public TF_AttrMetadata position(long position) {
-        return (TF_AttrMetadata)super.position(position);
-    }
-    @Override public TF_AttrMetadata getPointer(long i) {
-        return new TF_AttrMetadata(this).position(position + i);
-    }
-
-  // A boolean: 1 if the attribute value is a list, 0 otherwise.
-  public native @Cast("unsigned char") byte is_list(); public native TF_AttrMetadata is_list(byte setter);
-
-  // Length of the list if is_list is true. Undefined otherwise.
-  public native @Cast("int64_t") long list_size(); public native TF_AttrMetadata list_size(long setter);
-
-  // Type of elements of the list if is_list != 0.
-  // Type of the single value stored in the attribute if is_list == 0.
-  public native @Cast("TF_AttrType") int type(); public native TF_AttrMetadata type(int setter);
-
-  // Total size the attribute value.
-  // The units of total_size depend on is_list and type.
-  // (1) If type == TF_ATTR_STRING and is_list == 0
-  //     then total_size is the byte size of the string
-  //     valued attribute.
-  // (2) If type == TF_ATTR_STRING and is_list == 1
-  //     then total_size is the cumulative byte size
-  //     of all the strings in the list.
-  // (3) If type == TF_ATTR_SHAPE and is_list == 0
-  //     then total_size is the number of dimensions
-  //     of the shape valued attribute, or -1
-  //     if its rank is unknown.
-  // (4) If type == TF_ATTR_SHAPE and is_list == 1
-  //     then total_size is the cumulative number
-  //     of dimensions of all shapes in the list.
-  // (5) Otherwise, total_size is undefined.
-  public native @Cast("int64_t") long total_size(); public native TF_AttrMetadata total_size(long setter);
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_Buffer.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_Buffer.java
deleted file mode 100644
index a3d80f3b50e..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_Buffer.java
+++ /dev/null
@@ -1,49 +0,0 @@
-// Targeted by JavaCPP version 1.5.4: DO NOT EDIT THIS FILE
-
-package org.tensorflow.internal.c_api;
-
-import java.nio.*;
-import org.bytedeco.javacpp.*;
-import org.bytedeco.javacpp.annotation.*;
-
-import static org.tensorflow.internal.c_api.global.tensorflow.*;
-
-
-// --------------------------------------------------------------------------
-// TF_Buffer holds a pointer to a block of data and its associated length.
-// Typically, the data consists of a serialized protocol buffer, but other data
-// may also be held in a buffer.
-//
-// By default, TF_Buffer itself does not do any memory management of the
-// pointed-to block.  If need be, users of this struct should specify how to
-// deallocate the block by setting the `data_deallocator` function pointer.
-@Properties(inherit = org.tensorflow.internal.c_api.presets.tensorflow.class)
-public class TF_Buffer extends org.tensorflow.internal.c_api.AbstractTF_Buffer {
-    static { Loader.load(); }
-    /** Default native constructor. */
-    public TF_Buffer() { super((Pointer)null); allocate(); }
-    /** Native array allocator. Access with {@link Pointer#position(long)}. */
-    public TF_Buffer(long size) { super((Pointer)null); allocateArray(size); }
-    /** Pointer cast constructor. Invokes {@link Pointer#Pointer(Pointer)}. */
-    public TF_Buffer(Pointer p) { super(p); }
-    private native void allocate();
-    private native void allocateArray(long size);
-    @Override public TF_Buffer position(long position) {
-        return (TF_Buffer)super.position(position);
-    }
-    @Override public TF_Buffer getPointer(long i) {
-        return new TF_Buffer(this).position(position + i);
-    }
-
-  public native @Const Pointer data(); public native TF_Buffer data(Pointer data);
-  public native @Cast("size_t") long length(); public native TF_Buffer length(long setter);
-  public static class Data_deallocator_Pointer_long extends FunctionPointer {
-      static { Loader.load(); }
-      /** Pointer cast constructor. Invokes {@link Pointer#Pointer(Pointer)}. */
-      public    Data_deallocator_Pointer_long(Pointer p) { super(p); }
-      protected Data_deallocator_Pointer_long() { allocate(); }
-      private native void allocate();
-      public native void call(Pointer data, @Cast("size_t") long length);
-  }
-  public native Data_deallocator_Pointer_long data_deallocator(); public native TF_Buffer data_deallocator(Data_deallocator_Pointer_long setter);
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_DeprecatedSession.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_DeprecatedSession.java
deleted file mode 100644
index fdaab8327a9..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_DeprecatedSession.java
+++ /dev/null
@@ -1,23 +0,0 @@
-// Targeted by JavaCPP version 1.5.4: DO NOT EDIT THIS FILE
-
-package org.tensorflow.internal.c_api;
-
-import java.nio.*;
-import org.bytedeco.javacpp.*;
-import org.bytedeco.javacpp.annotation.*;
-
-import static org.tensorflow.internal.c_api.global.tensorflow.*;
-
-
-// --------------------------------------------------------------------------
-// The deprecated session API.  Please switch to the above instead of
-// TF_ExtendGraph(). This deprecated API can be removed at any time without
-// notice.
-
-@Opaque @Properties(inherit = org.tensorflow.internal.c_api.presets.tensorflow.class)
-public class TF_DeprecatedSession extends Pointer {
-    /** Empty constructor. Calls {@code super((Pointer)null)}. */
-    public TF_DeprecatedSession() { super((Pointer)null); }
-    /** Pointer cast constructor. Invokes {@link Pointer#Pointer(Pointer)}. */
-    public TF_DeprecatedSession(Pointer p) { super(p); }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_DeviceList.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_DeviceList.java
deleted file mode 100644
index e56d93340c1..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_DeviceList.java
+++ /dev/null
@@ -1,18 +0,0 @@
-// Targeted by JavaCPP version 1.5.4: DO NOT EDIT THIS FILE
-
-package org.tensorflow.internal.c_api;
-
-import java.nio.*;
-import org.bytedeco.javacpp.*;
-import org.bytedeco.javacpp.annotation.*;
-
-import static org.tensorflow.internal.c_api.global.tensorflow.*;
-
-
-@Opaque @Properties(inherit = org.tensorflow.internal.c_api.presets.tensorflow.class)
-public class TF_DeviceList extends Pointer {
-    /** Empty constructor. Calls {@code super((Pointer)null)}. */
-    public TF_DeviceList() { super((Pointer)null); }
-    /** Pointer cast constructor. Invokes {@link Pointer#Pointer(Pointer)}. */
-    public TF_DeviceList(Pointer p) { super(p); }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_DimensionHandle.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_DimensionHandle.java
deleted file mode 100644
index 2eb78c52b75..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_DimensionHandle.java
+++ /dev/null
@@ -1,19 +0,0 @@
-// Targeted by JavaCPP version 1.5.4: DO NOT EDIT THIS FILE
-
-package org.tensorflow.internal.c_api;
-
-import java.nio.*;
-import org.bytedeco.javacpp.*;
-import org.bytedeco.javacpp.annotation.*;
-
-import static org.tensorflow.internal.c_api.global.tensorflow.*;
-
-// #endif
-
-@Opaque @Properties(inherit = org.tensorflow.internal.c_api.presets.tensorflow.class)
-public class TF_DimensionHandle extends Pointer {
-    /** Empty constructor. Calls {@code super((Pointer)null)}. */
-    public TF_DimensionHandle() { super((Pointer)null); }
-    /** Pointer cast constructor. Invokes {@link Pointer#Pointer(Pointer)}. */
-    public TF_DimensionHandle(Pointer p) { super(p); }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_Function.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_Function.java
deleted file mode 100644
index e370b2f9f08..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_Function.java
+++ /dev/null
@@ -1,21 +0,0 @@
-// Targeted by JavaCPP version 1.5.4: DO NOT EDIT THIS FILE
-
-package org.tensorflow.internal.c_api;
-
-import java.nio.*;
-import org.bytedeco.javacpp.*;
-import org.bytedeco.javacpp.annotation.*;
-
-import static org.tensorflow.internal.c_api.global.tensorflow.*;
-
-
-// TF_Function is a grouping of operations with defined inputs and outputs.
-// Once created and added to graphs, functions can be invoked by creating an
-// operation whose operation type matches the function name.
-@Opaque @Properties(inherit = org.tensorflow.internal.c_api.presets.tensorflow.class)
-public class TF_Function extends Pointer {
-    /** Empty constructor. Calls {@code super((Pointer)null)}. */
-    public TF_Function() { super((Pointer)null); }
-    /** Pointer cast constructor. Invokes {@link Pointer#Pointer(Pointer)}. */
-    public TF_Function(Pointer p) { super(p); }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_FunctionOptions.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_FunctionOptions.java
deleted file mode 100644
index 5610e784a6f..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_FunctionOptions.java
+++ /dev/null
@@ -1,19 +0,0 @@
-// Targeted by JavaCPP version 1.5.4: DO NOT EDIT THIS FILE
-
-package org.tensorflow.internal.c_api;
-
-import java.nio.*;
-import org.bytedeco.javacpp.*;
-import org.bytedeco.javacpp.annotation.*;
-
-import static org.tensorflow.internal.c_api.global.tensorflow.*;
-
-
-// Function definition options. TODO(iga): Define and implement
-@Opaque @Properties(inherit = org.tensorflow.internal.c_api.presets.tensorflow.class)
-public class TF_FunctionOptions extends Pointer {
-    /** Empty constructor. Calls {@code super((Pointer)null)}. */
-    public TF_FunctionOptions() { super((Pointer)null); }
-    /** Pointer cast constructor. Invokes {@link Pointer#Pointer(Pointer)}. */
-    public TF_FunctionOptions(Pointer p) { super(p); }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_Graph.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_Graph.java
deleted file mode 100644
index 0a287cd5642..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_Graph.java
+++ /dev/null
@@ -1,26 +0,0 @@
-// Targeted by JavaCPP version 1.5.4: DO NOT EDIT THIS FILE
-
-package org.tensorflow.internal.c_api;
-
-import java.nio.*;
-import org.bytedeco.javacpp.*;
-import org.bytedeco.javacpp.annotation.*;
-
-import static org.tensorflow.internal.c_api.global.tensorflow.*;
-
-
-// TODO(jeff,sanjay):
-// - export functions to set Config fields
-
-// --------------------------------------------------------------------------
-// The new graph construction API, still under development.
-
-// Represents a computation graph.  Graphs may be shared between sessions.
-// Graphs are thread-safe when used as directed below.
-@Opaque @Properties(inherit = org.tensorflow.internal.c_api.presets.tensorflow.class)
-public class TF_Graph extends org.tensorflow.internal.c_api.AbstractTF_Graph {
-    /** Empty constructor. Calls {@code super((Pointer)null)}. */
-    public TF_Graph() { super((Pointer)null); }
-    /** Pointer cast constructor. Invokes {@link Pointer#Pointer(Pointer)}. */
-    public TF_Graph(Pointer p) { super(p); }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_ImportGraphDefOptions.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_ImportGraphDefOptions.java
deleted file mode 100644
index 442488d561d..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_ImportGraphDefOptions.java
+++ /dev/null
@@ -1,20 +0,0 @@
-// Targeted by JavaCPP version 1.5.4: DO NOT EDIT THIS FILE
-
-package org.tensorflow.internal.c_api;
-
-import java.nio.*;
-import org.bytedeco.javacpp.*;
-import org.bytedeco.javacpp.annotation.*;
-
-import static org.tensorflow.internal.c_api.global.tensorflow.*;
-
-
-// TF_ImportGraphDefOptions holds options that can be passed to
-// TF_GraphImportGraphDef.
-@Opaque @Properties(inherit = org.tensorflow.internal.c_api.presets.tensorflow.class)
-public class TF_ImportGraphDefOptions extends org.tensorflow.internal.c_api.AbstractTF_ImportGraphDefOptions {
-    /** Empty constructor. Calls {@code super((Pointer)null)}. */
-    public TF_ImportGraphDefOptions() { super((Pointer)null); }
-    /** Pointer cast constructor. Invokes {@link Pointer#Pointer(Pointer)}. */
-    public TF_ImportGraphDefOptions(Pointer p) { super(p); }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_ImportGraphDefResults.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_ImportGraphDefResults.java
deleted file mode 100644
index ac800f534e1..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_ImportGraphDefResults.java
+++ /dev/null
@@ -1,20 +0,0 @@
-// Targeted by JavaCPP version 1.5.4: DO NOT EDIT THIS FILE
-
-package org.tensorflow.internal.c_api;
-
-import java.nio.*;
-import org.bytedeco.javacpp.*;
-import org.bytedeco.javacpp.annotation.*;
-
-import static org.tensorflow.internal.c_api.global.tensorflow.*;
-
-
-// TF_ImportGraphDefResults holds results that are generated by
-// TF_GraphImportGraphDefWithResults().
-@Opaque @Properties(inherit = org.tensorflow.internal.c_api.presets.tensorflow.class)
-public class TF_ImportGraphDefResults extends Pointer {
-    /** Empty constructor. Calls {@code super((Pointer)null)}. */
-    public TF_ImportGraphDefResults() { super((Pointer)null); }
-    /** Pointer cast constructor. Invokes {@link Pointer#Pointer(Pointer)}. */
-    public TF_ImportGraphDefResults(Pointer p) { super(p); }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_Input.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_Input.java
deleted file mode 100644
index aa03e503d3a..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_Input.java
+++ /dev/null
@@ -1,33 +0,0 @@
-// Targeted by JavaCPP version 1.5.4: DO NOT EDIT THIS FILE
-
-package org.tensorflow.internal.c_api;
-
-import java.nio.*;
-import org.bytedeco.javacpp.*;
-import org.bytedeco.javacpp.annotation.*;
-
-import static org.tensorflow.internal.c_api.global.tensorflow.*;
-
-
-// Represents a specific input of an operation.
-@Properties(inherit = org.tensorflow.internal.c_api.presets.tensorflow.class)
-public class TF_Input extends Pointer {
-    static { Loader.load(); }
-    /** Default native constructor. */
-    public TF_Input() { super((Pointer)null); allocate(); }
-    /** Native array allocator. Access with {@link Pointer#position(long)}. */
-    public TF_Input(long size) { super((Pointer)null); allocateArray(size); }
-    /** Pointer cast constructor. Invokes {@link Pointer#Pointer(Pointer)}. */
-    public TF_Input(Pointer p) { super(p); }
-    private native void allocate();
-    private native void allocateArray(long size);
-    @Override public TF_Input position(long position) {
-        return (TF_Input)super.position(position);
-    }
-    @Override public TF_Input getPointer(long i) {
-        return new TF_Input(this).position(position + i);
-    }
-
-  public native TF_Operation oper(); public native TF_Input oper(TF_Operation setter);
-  public native int index(); public native TF_Input index(int setter);  // The index of the input within oper.
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_KernelBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_KernelBuilder.java
deleted file mode 100644
index 796bfa5aefc..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_KernelBuilder.java
+++ /dev/null
@@ -1,31 +0,0 @@
-// Targeted by JavaCPP version 1.5.4: DO NOT EDIT THIS FILE
-
-package org.tensorflow.internal.c_api;
-
-import java.nio.*;
-import org.bytedeco.javacpp.*;
-import org.bytedeco.javacpp.annotation.*;
-
-import static org.tensorflow.internal.c_api.global.tensorflow.*;
-
-
-// --------------------------------------------------------------------------
-// C API for TensorFlow Kernels.
-//
-// This API allows developers to register custom kernel implementations for
-// TensorFlow.
-//
-// See c_api.h header comments for a discussion about API conventions.
-//
-// Users wishing to extend TensorFlow with new kernels will call
-// `TF_NewKernelBuilder`. The resulting kernel builder can be registered with
-// `TF_RegisterKernelBuilder`, which will allow TF to construct user-provided
-// kernels when necessary.
-
-@Opaque @Properties(inherit = org.tensorflow.internal.c_api.presets.tensorflow.class)
-public class TF_KernelBuilder extends Pointer {
-    /** Empty constructor. Calls {@code super((Pointer)null)}. */
-    public TF_KernelBuilder() { super((Pointer)null); }
-    /** Pointer cast constructor. Invokes {@link Pointer#Pointer(Pointer)}. */
-    public TF_KernelBuilder(Pointer p) { super(p); }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_Library.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_Library.java
deleted file mode 100644
index 128efd99025..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_Library.java
+++ /dev/null
@@ -1,22 +0,0 @@
-// Targeted by JavaCPP version 1.5.4: DO NOT EDIT THIS FILE
-
-package org.tensorflow.internal.c_api;
-
-import java.nio.*;
-import org.bytedeco.javacpp.*;
-import org.bytedeco.javacpp.annotation.*;
-
-import static org.tensorflow.internal.c_api.global.tensorflow.*;
-
-
-// --------------------------------------------------------------------------
-// Load plugins containing custom ops and kernels
-
-// TF_Library holds information about dynamically loaded TensorFlow plugins.
-@Opaque @Properties(inherit = org.tensorflow.internal.c_api.presets.tensorflow.class)
-public class TF_Library extends Pointer {
-    /** Empty constructor. Calls {@code super((Pointer)null)}. */
-    public TF_Library() { super((Pointer)null); }
-    /** Pointer cast constructor. Invokes {@link Pointer#Pointer(Pointer)}. */
-    public TF_Library(Pointer p) { super(p); }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_OpDefinitionBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_OpDefinitionBuilder.java
deleted file mode 100644
index ef8a6b7d7f1..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_OpDefinitionBuilder.java
+++ /dev/null
@@ -1,17 +0,0 @@
-// Targeted by JavaCPP version 1.5.4: DO NOT EDIT THIS FILE
-
-package org.tensorflow.internal.c_api;
-
-import java.nio.*;
-import org.bytedeco.javacpp.*;
-import org.bytedeco.javacpp.annotation.*;
-
-import static org.tensorflow.internal.c_api.global.tensorflow.*;
-
-@Opaque @Properties(inherit = org.tensorflow.internal.c_api.presets.tensorflow.class)
-public class TF_OpDefinitionBuilder extends Pointer {
-    /** Empty constructor. Calls {@code super((Pointer)null)}. */
-    public TF_OpDefinitionBuilder() { super((Pointer)null); }
-    /** Pointer cast constructor. Invokes {@link Pointer#Pointer(Pointer)}. */
-    public TF_OpDefinitionBuilder(Pointer p) { super(p); }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_OpKernelConstruction.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_OpKernelConstruction.java
deleted file mode 100644
index 6a984f2a25d..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_OpKernelConstruction.java
+++ /dev/null
@@ -1,17 +0,0 @@
-// Targeted by JavaCPP version 1.5.4: DO NOT EDIT THIS FILE
-
-package org.tensorflow.internal.c_api;
-
-import java.nio.*;
-import org.bytedeco.javacpp.*;
-import org.bytedeco.javacpp.annotation.*;
-
-import static org.tensorflow.internal.c_api.global.tensorflow.*;
-
-@Opaque @Properties(inherit = org.tensorflow.internal.c_api.presets.tensorflow.class)
-public class TF_OpKernelConstruction extends Pointer {
-    /** Empty constructor. Calls {@code super((Pointer)null)}. */
-    public TF_OpKernelConstruction() { super((Pointer)null); }
-    /** Pointer cast constructor. Invokes {@link Pointer#Pointer(Pointer)}. */
-    public TF_OpKernelConstruction(Pointer p) { super(p); }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_OpKernelContext.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_OpKernelContext.java
deleted file mode 100644
index 9c145e89bdf..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_OpKernelContext.java
+++ /dev/null
@@ -1,17 +0,0 @@
-// Targeted by JavaCPP version 1.5.4: DO NOT EDIT THIS FILE
-
-package org.tensorflow.internal.c_api;
-
-import java.nio.*;
-import org.bytedeco.javacpp.*;
-import org.bytedeco.javacpp.annotation.*;
-
-import static org.tensorflow.internal.c_api.global.tensorflow.*;
-
-@Opaque @Properties(inherit = org.tensorflow.internal.c_api.presets.tensorflow.class)
-public class TF_OpKernelContext extends Pointer {
-    /** Empty constructor. Calls {@code super((Pointer)null)}. */
-    public TF_OpKernelContext() { super((Pointer)null); }
-    /** Pointer cast constructor. Invokes {@link Pointer#Pointer(Pointer)}. */
-    public TF_OpKernelContext(Pointer p) { super(p); }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_Operation.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_Operation.java
deleted file mode 100644
index 0cd1c90ecae..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_Operation.java
+++ /dev/null
@@ -1,21 +0,0 @@
-// Targeted by JavaCPP version 1.5.4: DO NOT EDIT THIS FILE
-
-package org.tensorflow.internal.c_api;
-
-import java.nio.*;
-import org.bytedeco.javacpp.*;
-import org.bytedeco.javacpp.annotation.*;
-
-import static org.tensorflow.internal.c_api.global.tensorflow.*;
-
-
-// Operation that has been added to the graph. Valid until the graph is
-// deleted -- in particular adding a new operation to the graph does not
-// invalidate old TF_Operation* pointers.
-@Opaque @Properties(inherit = org.tensorflow.internal.c_api.presets.tensorflow.class)
-public class TF_Operation extends Pointer {
-    /** Empty constructor. Calls {@code super((Pointer)null)}. */
-    public TF_Operation() { super((Pointer)null); }
-    /** Pointer cast constructor. Invokes {@link Pointer#Pointer(Pointer)}. */
-    public TF_Operation(Pointer p) { super(p); }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_OperationDescription.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_OperationDescription.java
deleted file mode 100644
index 64a577c28de..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_OperationDescription.java
+++ /dev/null
@@ -1,19 +0,0 @@
-// Targeted by JavaCPP version 1.5.4: DO NOT EDIT THIS FILE
-
-package org.tensorflow.internal.c_api;
-
-import java.nio.*;
-import org.bytedeco.javacpp.*;
-import org.bytedeco.javacpp.annotation.*;
-
-import static org.tensorflow.internal.c_api.global.tensorflow.*;
-
-
-// Operation being built. The underlying graph must outlive this.
-@Opaque @Properties(inherit = org.tensorflow.internal.c_api.presets.tensorflow.class)
-public class TF_OperationDescription extends Pointer {
-    /** Empty constructor. Calls {@code super((Pointer)null)}. */
-    public TF_OperationDescription() { super((Pointer)null); }
-    /** Pointer cast constructor. Invokes {@link Pointer#Pointer(Pointer)}. */
-    public TF_OperationDescription(Pointer p) { super(p); }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_Output.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_Output.java
deleted file mode 100644
index b990302d373..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_Output.java
+++ /dev/null
@@ -1,33 +0,0 @@
-// Targeted by JavaCPP version 1.5.4: DO NOT EDIT THIS FILE
-
-package org.tensorflow.internal.c_api;
-
-import java.nio.*;
-import org.bytedeco.javacpp.*;
-import org.bytedeco.javacpp.annotation.*;
-
-import static org.tensorflow.internal.c_api.global.tensorflow.*;
-
-
-// Represents a specific output of an operation.
-@Properties(inherit = org.tensorflow.internal.c_api.presets.tensorflow.class)
-public class TF_Output extends Pointer {
-    static { Loader.load(); }
-    /** Default native constructor. */
-    public TF_Output() { super((Pointer)null); allocate(); }
-    /** Native array allocator. Access with {@link Pointer#position(long)}. */
-    public TF_Output(long size) { super((Pointer)null); allocateArray(size); }
-    /** Pointer cast constructor. Invokes {@link Pointer#Pointer(Pointer)}. */
-    public TF_Output(Pointer p) { super(p); }
-    private native void allocate();
-    private native void allocateArray(long size);
-    @Override public TF_Output position(long position) {
-        return (TF_Output)super.position(position);
-    }
-    @Override public TF_Output getPointer(long i) {
-        return new TF_Output(this).position(position + i);
-    }
-
-  public native TF_Operation oper(); public native TF_Output oper(TF_Operation setter);
-  public native int index(); public native TF_Output index(int setter);  // The index of the output within oper.
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_Server.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_Server.java
deleted file mode 100644
index c34d6dd2eab..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_Server.java
+++ /dev/null
@@ -1,27 +0,0 @@
-// Targeted by JavaCPP version 1.5.4: DO NOT EDIT THIS FILE
-
-package org.tensorflow.internal.c_api;
-
-import java.nio.*;
-import org.bytedeco.javacpp.*;
-import org.bytedeco.javacpp.annotation.*;
-
-import static org.tensorflow.internal.c_api.global.tensorflow.*;
-
-
-// --------------------------------------------------------------------------
-// In-process TensorFlow server functionality, for use in distributed training.
-// A Server instance encapsulates a set of devices and a Session target that
-// can participate in distributed training. A server belongs to a cluster
-// (specified by a ClusterSpec), and corresponds to a particular task in a
-// named job. The server can communicate with any other server in the same
-// cluster.
-
-// In-process TensorFlow server.
-@Opaque @Properties(inherit = org.tensorflow.internal.c_api.presets.tensorflow.class)
-public class TF_Server extends Pointer {
-    /** Empty constructor. Calls {@code super((Pointer)null)}. */
-    public TF_Server() { super((Pointer)null); }
-    /** Pointer cast constructor. Invokes {@link Pointer#Pointer(Pointer)}. */
-    public TF_Server(Pointer p) { super(p); }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_Session.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_Session.java
deleted file mode 100644
index 10818f6b59f..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_Session.java
+++ /dev/null
@@ -1,24 +0,0 @@
-// Targeted by JavaCPP version 1.5.4: DO NOT EDIT THIS FILE
-
-package org.tensorflow.internal.c_api;
-
-import java.nio.*;
-import org.bytedeco.javacpp.*;
-import org.bytedeco.javacpp.annotation.*;
-
-import static org.tensorflow.internal.c_api.global.tensorflow.*;
-
-
-// TODO(josh11b): Register OpDef, available to all operations added
-// to this graph.
-
-// --------------------------------------------------------------------------
-// API for driving Graph execution.
-
-@Opaque @Properties(inherit = org.tensorflow.internal.c_api.presets.tensorflow.class)
-public class TF_Session extends org.tensorflow.internal.c_api.AbstractTF_Session {
-    /** Empty constructor. Calls {@code super((Pointer)null)}. */
-    public TF_Session() { super((Pointer)null); }
-    /** Pointer cast constructor. Invokes {@link Pointer#Pointer(Pointer)}. */
-    public TF_Session(Pointer p) { super(p); }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_SessionOptions.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_SessionOptions.java
deleted file mode 100644
index c96017ab554..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_SessionOptions.java
+++ /dev/null
@@ -1,20 +0,0 @@
-// Targeted by JavaCPP version 1.5.4: DO NOT EDIT THIS FILE
-
-package org.tensorflow.internal.c_api;
-
-import java.nio.*;
-import org.bytedeco.javacpp.*;
-import org.bytedeco.javacpp.annotation.*;
-
-import static org.tensorflow.internal.c_api.global.tensorflow.*;
-
-
-// --------------------------------------------------------------------------
-// TF_SessionOptions holds options that can be passed during session creation.
-@Opaque @Properties(inherit = org.tensorflow.internal.c_api.presets.tensorflow.class)
-public class TF_SessionOptions extends org.tensorflow.internal.c_api.AbstractTF_SessionOptions {
-    /** Empty constructor. Calls {@code super((Pointer)null)}. */
-    public TF_SessionOptions() { super((Pointer)null); }
-    /** Pointer cast constructor. Invokes {@link Pointer#Pointer(Pointer)}. */
-    public TF_SessionOptions(Pointer p) { super(p); }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_ShapeHandle.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_ShapeHandle.java
deleted file mode 100644
index 7fc4dd37276..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_ShapeHandle.java
+++ /dev/null
@@ -1,17 +0,0 @@
-// Targeted by JavaCPP version 1.5.4: DO NOT EDIT THIS FILE
-
-package org.tensorflow.internal.c_api;
-
-import java.nio.*;
-import org.bytedeco.javacpp.*;
-import org.bytedeco.javacpp.annotation.*;
-
-import static org.tensorflow.internal.c_api.global.tensorflow.*;
-
-@Opaque @Properties(inherit = org.tensorflow.internal.c_api.presets.tensorflow.class)
-public class TF_ShapeHandle extends Pointer {
-    /** Empty constructor. Calls {@code super((Pointer)null)}. */
-    public TF_ShapeHandle() { super((Pointer)null); }
-    /** Pointer cast constructor. Invokes {@link Pointer#Pointer(Pointer)}. */
-    public TF_ShapeHandle(Pointer p) { super(p); }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_ShapeInferenceContext.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_ShapeInferenceContext.java
deleted file mode 100644
index 4d2bd0c7441..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_ShapeInferenceContext.java
+++ /dev/null
@@ -1,17 +0,0 @@
-// Targeted by JavaCPP version 1.5.4: DO NOT EDIT THIS FILE
-
-package org.tensorflow.internal.c_api;
-
-import java.nio.*;
-import org.bytedeco.javacpp.*;
-import org.bytedeco.javacpp.annotation.*;
-
-import static org.tensorflow.internal.c_api.global.tensorflow.*;
-
-@Opaque @Properties(inherit = org.tensorflow.internal.c_api.presets.tensorflow.class)
-public class TF_ShapeInferenceContext extends Pointer {
-    /** Empty constructor. Calls {@code super((Pointer)null)}. */
-    public TF_ShapeInferenceContext() { super((Pointer)null); }
-    /** Pointer cast constructor. Invokes {@link Pointer#Pointer(Pointer)}. */
-    public TF_ShapeInferenceContext(Pointer p) { super(p); }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_Status.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_Status.java
deleted file mode 100644
index 0ea4e3415c3..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_Status.java
+++ /dev/null
@@ -1,19 +0,0 @@
-// Targeted by JavaCPP version 1.5.4: DO NOT EDIT THIS FILE
-
-package org.tensorflow.internal.c_api;
-
-import java.nio.*;
-import org.bytedeco.javacpp.*;
-import org.bytedeco.javacpp.annotation.*;
-
-import static org.tensorflow.internal.c_api.global.tensorflow.*;
-
-// #endif
-
-@Opaque @Properties(inherit = org.tensorflow.internal.c_api.presets.tensorflow.class)
-public class TF_Status extends org.tensorflow.internal.c_api.AbstractTF_Status {
-    /** Empty constructor. Calls {@code super((Pointer)null)}. */
-    public TF_Status() { super((Pointer)null); }
-    /** Pointer cast constructor. Invokes {@link Pointer#Pointer(Pointer)}. */
-    public TF_Status(Pointer p) { super(p); }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_Tensor.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_Tensor.java
deleted file mode 100644
index b1839dadd2d..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_Tensor.java
+++ /dev/null
@@ -1,36 +0,0 @@
-// Targeted by JavaCPP version 1.5.4: DO NOT EDIT THIS FILE
-
-package org.tensorflow.internal.c_api;
-
-import java.nio.*;
-import org.bytedeco.javacpp.*;
-import org.bytedeco.javacpp.annotation.*;
-
-import static org.tensorflow.internal.c_api.global.tensorflow.*;
-
-// #endif
-
-// --------------------------------------------------------------------------
-// TF_Tensor holds a multi-dimensional array of elements of a single data type.
-// For all types other than TF_STRING, the data buffer stores elements
-// in row major order.  E.g. if data is treated as a vector of TF_DataType:
-//
-//   element 0:   index (0, ..., 0)
-//   element 1:   index (0, ..., 1)
-//   ...
-//
-// The format for TF_STRING tensors is:
-//   start_offset: array[uint64]
-//   data:         byte[...]
-//
-//   The string length (as a varint, start_offset[i + 1] - start_offset[i]),
-//   followed by the contents of the string is encoded at data[start_offset[i]].
-//   TF_StringEncode and TF_StringDecode facilitate this encoding.
-
-@Opaque @Properties(inherit = org.tensorflow.internal.c_api.presets.tensorflow.class)
-public class TF_Tensor extends org.tensorflow.internal.c_api.AbstractTF_Tensor {
-    /** Empty constructor. Calls {@code super((Pointer)null)}. */
-    public TF_Tensor() { super((Pointer)null); }
-    /** Pointer cast constructor. Invokes {@link Pointer#Pointer(Pointer)}. */
-    public TF_Tensor(Pointer p) { super(p); }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_WhileParams.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_WhileParams.java
deleted file mode 100644
index 4c6dc486d52..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_WhileParams.java
+++ /dev/null
@@ -1,37 +0,0 @@
-// Targeted by JavaCPP version 1.5.4: DO NOT EDIT THIS FILE
-
-package org.tensorflow.internal.c_api;
-
-import java.nio.*;
-import org.bytedeco.javacpp.*;
-import org.bytedeco.javacpp.annotation.*;
-
-import static org.tensorflow.internal.c_api.global.tensorflow.*;
-
-
-@Properties(inherit = org.tensorflow.internal.c_api.presets.tensorflow.class)
-public class TF_WhileParams extends Pointer {
-    static { Loader.load(); }
-    /** Pointer cast constructor. Invokes {@link Pointer#Pointer(Pointer)}. */
-    public TF_WhileParams(Pointer p) { super(p); }
-
-  // The number of inputs to the while loop, i.e. the number of loop variables.
-  // This is the size of cond_inputs, body_inputs, and body_outputs.
-  @MemberGetter public native int ninputs();
-
-  // The while condition graph. The inputs are the current values of the loop
-  // variables. The output should be a scalar boolean.
-  @MemberGetter public native TF_Graph cond_graph();
-  @MemberGetter public native @Const TF_Output cond_inputs();
-  public native @ByRef TF_Output cond_output(); public native TF_WhileParams cond_output(TF_Output setter);
-
-  // The loop body graph. The inputs are the current values of the loop
-  // variables. The outputs are the updated values of the loop variables.
-  @MemberGetter public native TF_Graph body_graph();
-  @MemberGetter public native @Const TF_Output body_inputs();
-  @MemberGetter public native TF_Output body_outputs();
-
-  // Unique null-terminated name for this while loop. This is used as a prefix
-  // for created operations.
-  public native @Cast("const char*") BytePointer name(); public native TF_WhileParams name(BytePointer setter);
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/Tensor.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/Tensor.java
deleted file mode 100644
index 0a9efa65762..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/Tensor.java
+++ /dev/null
@@ -1,25 +0,0 @@
-// Targeted by JavaCPP version 1.5.4: DO NOT EDIT THIS FILE
-
-package org.tensorflow.internal.c_api;
-
-import java.nio.*;
-import org.bytedeco.javacpp.*;
-import org.bytedeco.javacpp.annotation.*;
-
-import static org.tensorflow.internal.c_api.global.tensorflow.*;
- /* end extern "C" */
-// #endif
-
-// #ifdef __cplusplus
-// A workaround to ease conversion to and from numpy objects and
-// TFE_TensorHandle's.
-//
-// TODO(ashankar): Figure out an alternative scheme that precludes the need for
-// these API-boundary breaking methods.
-@Namespace("tensorflow") @Opaque @Properties(inherit = org.tensorflow.internal.c_api.presets.tensorflow.class)
-public class Tensor extends Pointer {
-    /** Empty constructor. Calls {@code super((Pointer)null)}. */
-    public Tensor() { super((Pointer)null); }
-    /** Pointer cast constructor. Invokes {@link Pointer#Pointer(Pointer)}. */
-    public Tensor(Pointer p) { super(p); }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/global/tensorflow.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/global/tensorflow.java
deleted file mode 100644
index bdd9cc618ac..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/global/tensorflow.java
+++ /dev/null
@@ -1,4021 +0,0 @@
-// Targeted by JavaCPP version 1.5.4: DO NOT EDIT THIS FILE
-
-package org.tensorflow.internal.c_api.global;
-
-import org.tensorflow.internal.c_api.*;
-
-import java.nio.*;
-import org.bytedeco.javacpp.*;
-import org.bytedeco.javacpp.annotation.*;
-
-public class tensorflow extends org.tensorflow.internal.c_api.presets.tensorflow {
-    static { Loader.load(); }
-
-// Parsed from tensorflow/core/util/port.h
-
-/* Copyright 2015 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-==============================================================================*/
-
-// #ifndef TENSORFLOW_CORE_UTIL_PORT_H_
-// #define TENSORFLOW_CORE_UTIL_PORT_H_
-
-// Returns true if GOOGLE_CUDA is defined.
-@Namespace("tensorflow") public static native @Cast("bool") boolean IsGoogleCudaEnabled();
-
-// Returns true if TENSORFLOW_USE_ROCM is defined. (i.e. TF is built with ROCm)
-@Namespace("tensorflow") public static native @Cast("bool") boolean IsBuiltWithROCm();
-
-// Returns true if TENSORFLOW_USE_XLA is defined. (i.e. TF is built with XLA)
-@Namespace("tensorflow") public static native @Cast("bool") boolean IsBuiltWithXLA();
-
-// Returns true if TENSORFLOW_USE_NVCC is defined. (i.e. TF is built with nvcc)
-@Namespace("tensorflow") public static native @Cast("bool") boolean IsBuiltWithNvcc();
-
-// Returns true if either
-//
-//   GOOGLE_CUDA is defined, and the given CUDA version supports
-//   half-precision matrix multiplications and convolution operations.
-//
-//     OR
-//
-//   TENSORFLOW_USE_ROCM is defined
-//
-@Namespace("tensorflow") public static native @Cast("bool") boolean GpuSupportsHalfMatMulAndConv();
-
-// Returns true if INTEL_MKL is defined
-@Namespace("tensorflow") public static native @Cast("bool") boolean IsMklEnabled();
-
-  // end namespace tensorflow
-
-// #endif  // TENSORFLOW_CORE_UTIL_PORT_H_
-
-
-// Parsed from tensorflow/c/tf_attrtype.h
-
-/* Copyright 2019 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-==============================================================================*/
-// #ifndef TENSORFLOW_C_TF_ATTRTYPE_H_
-// #define TENSORFLOW_C_TF_ATTRTYPE_H_
-
-// #ifdef __cplusplus
-// #endif
-
-// TF_AttrType describes the type of the value of an attribute on an operation.
-/** enum TF_AttrType */
-public static final int
-  TF_ATTR_STRING = 0,
-  TF_ATTR_INT = 1,
-  TF_ATTR_FLOAT = 2,
-  TF_ATTR_BOOL = 3,
-  TF_ATTR_TYPE = 4,
-  TF_ATTR_SHAPE = 5,
-  TF_ATTR_TENSOR = 6,
-  TF_ATTR_PLACEHOLDER = 7,
-  TF_ATTR_FUNC = 8;
-
-// #ifdef __cplusplus /* end extern "C" */
-// #endif
-
-// #endif  // TENSORFLOW_C_TF_ATTRTYPE_H_
-
-
-// Parsed from tensorflow/c/tf_datatype.h
-
-/* Copyright 2019 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-==============================================================================*/
-
-// #ifndef TENSORFLOW_C_TF_DATATYPE_H_
-// #define TENSORFLOW_C_TF_DATATYPE_H_
-
-// #include <stddef.h>
-
-// Macro to control visibility of exported symbols in the shared library (.so,
-// .dylib, .dll).
-// This duplicates the TF_EXPORT macro definition in
-// tensorflow/core/platform/macros.h in order to keep this .h file independent
-// of any other includes.
-// #ifdef SWIG
-// #define TF_CAPI_EXPORT
-// #else
-// #if defined(_WIN32)
-// #ifdef TF_COMPILE_LIBRARY
-// #define TF_CAPI_EXPORT __declspec(dllexport)
-// #else
-// #define TF_CAPI_EXPORT __declspec(dllimport)
-// #endif  // TF_COMPILE_LIBRARY
-// #else
-// #define TF_CAPI_EXPORT __attribute__((visibility("default")))
-// #endif  // _WIN32
-// #endif  // SWIG
-
-// #ifdef __cplusplus
-// #endif
-
-// --------------------------------------------------------------------------
-// TF_DataType holds the type for a scalar value.  E.g., one slot in a tensor.
-// The enum values here are identical to corresponding values in types.proto.
-/** enum TF_DataType */
-public static final int
-  TF_FLOAT = 1,
-  TF_DOUBLE = 2,
-  TF_INT32 = 3,  // Int32 tensors are always in 'host' memory.
-  TF_UINT8 = 4,
-  TF_INT16 = 5,
-  TF_INT8 = 6,
-  TF_STRING = 7,
-  TF_COMPLEX64 = 8,  // Single-precision complex
-  TF_COMPLEX = 8,    // Old identifier kept for API backwards compatibility
-  TF_INT64 = 9,
-  TF_BOOL = 10,
-  TF_QINT8 = 11,     // Quantized int8
-  TF_QUINT8 = 12,    // Quantized uint8
-  TF_QINT32 = 13,    // Quantized int32
-  TF_BFLOAT16 = 14,  // Float32 truncated to 16 bits.  Only for cast ops.
-  TF_QINT16 = 15,    // Quantized int16
-  TF_QUINT16 = 16,   // Quantized uint16
-  TF_UINT16 = 17,
-  TF_COMPLEX128 = 18,  // Double-precision complex
-  TF_HALF = 19,
-  TF_RESOURCE = 20,
-  TF_VARIANT = 21,
-  TF_UINT32 = 22,
-  TF_UINT64 = 23;
-
-// TF_DataTypeSize returns the sizeof() for the underlying type corresponding
-// to the given TF_DataType enum value. Returns 0 for variable length types
-// (eg. TF_STRING) or on failure.
-public static native @Cast("size_t") long TF_DataTypeSize(@Cast("TF_DataType") int dt);
-
-// #ifdef __cplusplus /* end extern "C" */
-// #endif
-
-// #endif  // TENSORFLOW_C_TF_DATATYPE_H_
-
-
-// Parsed from tensorflow/c/tf_status.h
-
-/* Copyright 2019 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-==============================================================================*/
-
-// #ifndef TENSORFLOW_C_TF_STATUS_H_
-// #define TENSORFLOW_C_TF_STATUS_H_
-
-// #ifdef SWIG
-// #define TF_CAPI_EXPORT
-// #else
-// #if defined(_WIN32)
-// #ifdef TF_COMPILE_LIBRARY
-// #define TF_CAPI_EXPORT __declspec(dllexport)
-// #else
-// #define TF_CAPI_EXPORT __declspec(dllimport)
-// #endif  // TF_COMPILE_LIBRARY
-// #else
-// #define TF_CAPI_EXPORT __attribute__((visibility("default")))
-// #endif  // _WIN32
-// #endif  // SWIG
-
-// #ifdef __cplusplus
-// Targeting ../TF_Status.java
-
-
-
-// --------------------------------------------------------------------------
-// TF_Code holds an error code.  The enum values here are identical to
-// corresponding values in error_codes.proto.
-/** enum TF_Code */
-public static final int
-  TF_OK = 0,
-  TF_CANCELLED = 1,
-  TF_UNKNOWN = 2,
-  TF_INVALID_ARGUMENT = 3,
-  TF_DEADLINE_EXCEEDED = 4,
-  TF_NOT_FOUND = 5,
-  TF_ALREADY_EXISTS = 6,
-  TF_PERMISSION_DENIED = 7,
-  TF_UNAUTHENTICATED = 16,
-  TF_RESOURCE_EXHAUSTED = 8,
-  TF_FAILED_PRECONDITION = 9,
-  TF_ABORTED = 10,
-  TF_OUT_OF_RANGE = 11,
-  TF_UNIMPLEMENTED = 12,
-  TF_INTERNAL = 13,
-  TF_UNAVAILABLE = 14,
-  TF_DATA_LOSS = 15;
-
-// --------------------------------------------------------------------------
-
-// Return a new status object.
-public static native TF_Status TF_NewStatus();
-
-// Delete a previously created status object.
-public static native void TF_DeleteStatus(TF_Status arg0);
-
-// Record <code, msg> in *s.  Any previous information is lost.
-// A common use is to clear a status: TF_SetStatus(s, TF_OK, "");
-public static native void TF_SetStatus(TF_Status s, @Cast("TF_Code") int code,
-                                        @Cast("const char*") BytePointer msg);
-public static native void TF_SetStatus(TF_Status s, @Cast("TF_Code") int code,
-                                        String msg);
-
-// Convert from an I/O error code (e.g., errno) to a TF_Status value.
-// Any previous information is lost. Prefer to use this instead of TF_SetStatus
-// when the error comes from I/O operations.
-public static native void TF_SetStatusFromIOError(TF_Status s, int error_code,
-                                                   @Cast("const char*") BytePointer context);
-public static native void TF_SetStatusFromIOError(TF_Status s, int error_code,
-                                                   String context);
-
-// Return the code record in *s.
-public static native @Cast("TF_Code") int TF_GetCode(@Const TF_Status s);
-
-// Return a pointer to the (null-terminated) error message in *s.  The
-// return value points to memory that is only usable until the next
-// mutation to *s.  Always returns an empty string if TF_GetCode(s) is
-// TF_OK.
-public static native @Cast("const char*") BytePointer TF_Message(@Const TF_Status s);
-
-// #ifdef __cplusplus /* end extern "C" */
-// #endif
-
-// #endif  // TENSORFLOW_C_TF_STATUS_H_
-
-
-// Parsed from tensorflow/c/tf_tensor.h
-
-/* Copyright 2019 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-==============================================================================*/
-
-// #ifndef TENSORFLOW_C_TF_TENSOR_H_
-// #define TENSORFLOW_C_TF_TENSOR_H_
-
-// #include <stdbool.h>
-// #include <stdint.h>
-
-// #include "tensorflow/c/tf_datatype.h"
-// #include "tensorflow/c/tf_status.h"
-
-// Macro to control visibility of exported symbols in the shared library (.so,
-// .dylib, .dll).
-// This duplicates the TF_EXPORT macro definition in
-// tensorflow/core/platform/macros.h in order to keep this .h file independent
-// of any other includes.
-// #ifdef SWIG
-// #define TF_CAPI_EXPORT
-// #else
-// #if defined(_WIN32)
-// #ifdef TF_COMPILE_LIBRARY
-// #define TF_CAPI_EXPORT __declspec(dllexport)
-// #else
-// #define TF_CAPI_EXPORT __declspec(dllimport)
-// #endif  // TF_COMPILE_LIBRARY
-// #else
-// #define TF_CAPI_EXPORT __attribute__((visibility("default")))
-// #endif  // _WIN32
-// #endif  // SWIG
-
-// #ifdef __cplusplus
-// Targeting ../TF_Tensor.java
-
-
-// Targeting ../Deallocator_Pointer_long_Pointer.java
-
-
-public static native TF_Tensor TF_NewTensor(
-    @Cast("TF_DataType") int arg0, @Cast("const int64_t*") LongPointer dims, int num_dims, Pointer data, @Cast("size_t") long len,
-    Deallocator_Pointer_long_Pointer deallocator,
-    Pointer deallocator_arg);
-public static native TF_Tensor TF_NewTensor(
-    @Cast("TF_DataType") int arg0, @Cast("const int64_t*") LongBuffer dims, int num_dims, Pointer data, @Cast("size_t") long len,
-    Deallocator_Pointer_long_Pointer deallocator,
-    Pointer deallocator_arg);
-public static native TF_Tensor TF_NewTensor(
-    @Cast("TF_DataType") int arg0, @Cast("const int64_t*") long[] dims, int num_dims, Pointer data, @Cast("size_t") long len,
-    Deallocator_Pointer_long_Pointer deallocator,
-    Pointer deallocator_arg);
-
-// Allocate and return a new Tensor.
-//
-// This function is an alternative to TF_NewTensor and should be used when
-// memory is allocated to pass the Tensor to the C API. The allocated memory
-// satisfies TensorFlow's memory alignment preferences and should be preferred
-// over calling malloc and free.
-//
-// The caller must set the Tensor values by writing them to the pointer returned
-// by TF_TensorData with length TF_TensorByteSize.
-public static native TF_Tensor TF_AllocateTensor(@Cast("TF_DataType") int arg0,
-                                                   @Cast("const int64_t*") LongPointer dims,
-                                                   int num_dims, @Cast("size_t") long len);
-public static native TF_Tensor TF_AllocateTensor(@Cast("TF_DataType") int arg0,
-                                                   @Cast("const int64_t*") LongBuffer dims,
-                                                   int num_dims, @Cast("size_t") long len);
-public static native TF_Tensor TF_AllocateTensor(@Cast("TF_DataType") int arg0,
-                                                   @Cast("const int64_t*") long[] dims,
-                                                   int num_dims, @Cast("size_t") long len);
-
-// Deletes `tensor` and returns a new TF_Tensor with the same content if
-// possible. Returns nullptr and leaves `tensor` untouched if not.
-public static native TF_Tensor TF_TensorMaybeMove(TF_Tensor tensor);
-
-// Destroy a tensor.
-public static native void TF_DeleteTensor(TF_Tensor arg0);
-
-// Return the type of a tensor element.
-public static native @Cast("TF_DataType") int TF_TensorType(@Const TF_Tensor arg0);
-
-// Return the number of dimensions that the tensor has.
-public static native int TF_NumDims(@Const TF_Tensor arg0);
-
-// Return the length of the tensor in the "dim_index" dimension.
-// REQUIRES: 0 <= dim_index < TF_NumDims(tensor)
-public static native @Cast("int64_t") long TF_Dim(@Const TF_Tensor tensor, int dim_index);
-
-// Return the size of the underlying data in bytes.
-public static native @Cast("size_t") long TF_TensorByteSize(@Const TF_Tensor arg0);
-
-// Return a pointer to the underlying data buffer.
-public static native Pointer TF_TensorData(@Const TF_Tensor arg0);
-
-// Returns the number of elements in the tensor.
-public static native @Cast("int64_t") long TF_TensorElementCount(@Const TF_Tensor tensor);
-
-// Copy the internal data representation of `from` to `to`. `new_dims` and
-// `num_new_dims` specify the new shape of the `to` tensor, `type` specifies its
-// data type. On success, *status is set to TF_OK and the two tensors share the
-// same data buffer.
-//
-// This call requires that the `from` tensor and the given type and shape (dims
-// and num_dims) are "compatible" (i.e. they occupy the same number of bytes).
-// Specifically, given from_type_size = TF_DataTypeSize(TF_TensorType(from)):
-//
-// ShapeElementCount(dims, num_dims) * TF_DataTypeSize(type)
-//
-// must equal
-//
-// TF_TensorElementCount(from) * from_type_size
-//
-// where TF_ShapeElementCount would be the number of elements in a tensor with
-// the given shape.
-//
-// In addition, this function requires:
-//   * TF_DataTypeSize(TF_TensorType(from)) != 0
-//   * TF_DataTypeSize(type) != 0
-//
-// If any of the requirements are not met, *status is set to
-// TF_INVALID_ARGUMENT.
-public static native void TF_TensorBitcastFrom(@Const TF_Tensor from,
-                                                @Cast("TF_DataType") int type, TF_Tensor to,
-                                                @Cast("const int64_t*") LongPointer new_dims,
-                                                int num_new_dims,
-                                                TF_Status status);
-public static native void TF_TensorBitcastFrom(@Const TF_Tensor from,
-                                                @Cast("TF_DataType") int type, TF_Tensor to,
-                                                @Cast("const int64_t*") LongBuffer new_dims,
-                                                int num_new_dims,
-                                                TF_Status status);
-public static native void TF_TensorBitcastFrom(@Const TF_Tensor from,
-                                                @Cast("TF_DataType") int type, TF_Tensor to,
-                                                @Cast("const int64_t*") long[] new_dims,
-                                                int num_new_dims,
-                                                TF_Status status);
-
-// --------------------------------------------------------------------------
-// Encode the string `src` (`src_len` bytes long) into `dst` in the format
-// required by TF_STRING tensors. Does not write to memory more than `dst_len`
-// bytes beyond `*dst`. `dst_len` should be at least
-// TF_StringEncodedSize(src_len).
-//
-// On success returns the size in bytes of the encoded string.
-// Returns an error into `status` otherwise.
-public static native @Cast("size_t") long TF_StringEncode(@Cast("const char*") BytePointer src, @Cast("size_t") long src_len,
-                                             @Cast("char*") BytePointer dst, @Cast("size_t") long dst_len,
-                                             TF_Status status);
-public static native @Cast("size_t") long TF_StringEncode(String src, @Cast("size_t") long src_len,
-                                             @Cast("char*") ByteBuffer dst, @Cast("size_t") long dst_len,
-                                             TF_Status status);
-public static native @Cast("size_t") long TF_StringEncode(@Cast("const char*") BytePointer src, @Cast("size_t") long src_len,
-                                             @Cast("char*") byte[] dst, @Cast("size_t") long dst_len,
-                                             TF_Status status);
-public static native @Cast("size_t") long TF_StringEncode(String src, @Cast("size_t") long src_len,
-                                             @Cast("char*") BytePointer dst, @Cast("size_t") long dst_len,
-                                             TF_Status status);
-public static native @Cast("size_t") long TF_StringEncode(@Cast("const char*") BytePointer src, @Cast("size_t") long src_len,
-                                             @Cast("char*") ByteBuffer dst, @Cast("size_t") long dst_len,
-                                             TF_Status status);
-public static native @Cast("size_t") long TF_StringEncode(String src, @Cast("size_t") long src_len,
-                                             @Cast("char*") byte[] dst, @Cast("size_t") long dst_len,
-                                             TF_Status status);
-
-// Decode a string encoded using TF_StringEncode.
-//
-// On success, sets `*dst` to the start of the decoded string and `*dst_len` to
-// its length. Returns the number of bytes starting at `src` consumed while
-// decoding. `*dst` points to memory within the encoded buffer.  On failure,
-// `*dst` and `*dst_len` are undefined and an error is set in `status`.
-//
-// Does not read memory more than `src_len` bytes beyond `src`.
-public static native @Cast("size_t") long TF_StringDecode(@Cast("const char*") BytePointer src, @Cast("size_t") long src_len,
-                                             @Cast("const char**") PointerPointer dst, @Cast("size_t*") SizeTPointer dst_len,
-                                             TF_Status status);
-public static native @Cast("size_t") long TF_StringDecode(@Cast("const char*") BytePointer src, @Cast("size_t") long src_len,
-                                             @Cast("const char**") @ByPtrPtr BytePointer dst, @Cast("size_t*") SizeTPointer dst_len,
-                                             TF_Status status);
-public static native @Cast("size_t") long TF_StringDecode(String src, @Cast("size_t") long src_len,
-                                             @Cast("const char**") @ByPtrPtr ByteBuffer dst, @Cast("size_t*") SizeTPointer dst_len,
-                                             TF_Status status);
-public static native @Cast("size_t") long TF_StringDecode(@Cast("const char*") BytePointer src, @Cast("size_t") long src_len,
-                                             @Cast("const char**") @ByPtrPtr byte[] dst, @Cast("size_t*") SizeTPointer dst_len,
-                                             TF_Status status);
-public static native @Cast("size_t") long TF_StringDecode(String src, @Cast("size_t") long src_len,
-                                             @Cast("const char**") @ByPtrPtr BytePointer dst, @Cast("size_t*") SizeTPointer dst_len,
-                                             TF_Status status);
-public static native @Cast("size_t") long TF_StringDecode(@Cast("const char*") BytePointer src, @Cast("size_t") long src_len,
-                                             @Cast("const char**") @ByPtrPtr ByteBuffer dst, @Cast("size_t*") SizeTPointer dst_len,
-                                             TF_Status status);
-public static native @Cast("size_t") long TF_StringDecode(String src, @Cast("size_t") long src_len,
-                                             @Cast("const char**") @ByPtrPtr byte[] dst, @Cast("size_t*") SizeTPointer dst_len,
-                                             TF_Status status);
-
-// Return the size in bytes required to encode a string `len` bytes long into a
-// TF_STRING tensor.
-public static native @Cast("size_t") long TF_StringEncodedSize(@Cast("size_t") long len);
-
-// Returns bool iff this tensor is aligned.
-public static native @Cast("bool") boolean TF_TensorIsAligned(@Const TF_Tensor arg0);
-
-// #ifdef __cplusplus /* end extern "C" */
-// #endif
-
-// #endif  // TENSORFLOW_C_TF_TENSOR_H_
-
-
-// Parsed from tensorflow/c/c_api.h
-
-/* Copyright 2015 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-==============================================================================*/
-
-// #ifndef TENSORFLOW_C_C_API_H_
-// #define TENSORFLOW_C_C_API_H_
-
-// #include <stddef.h>
-// #include <stdint.h>
-
-// #include "tensorflow/c/tf_attrtype.h"
-// #include "tensorflow/c/tf_datatype.h"
-// #include "tensorflow/c/tf_status.h"
-// #include "tensorflow/c/tf_tensor.h"
-
-// --------------------------------------------------------------------------
-// C API for TensorFlow.
-//
-// The API leans towards simplicity and uniformity instead of convenience
-// since most usage will be by language specific wrappers.
-//
-// Conventions:
-// * We use the prefix TF_ for everything in the API.
-// * Objects are always passed around as pointers to opaque structs
-//   and these structs are allocated/deallocated via the API.
-// * TF_Status holds error information.  It is an object type
-//   and therefore is passed around as a pointer to an opaque
-//   struct as mentioned above.
-// * Every call that has a TF_Status* argument clears it on success
-//   and fills it with error info on failure.
-// * unsigned char is used for booleans (instead of the 'bool' type).
-//   In C++ bool is a keyword while in C99 bool is a macro defined
-//   in stdbool.h. It is possible for the two to be inconsistent.
-//   For example, neither the C99 nor the C++11 standard force a byte
-//   size on the bool type, so the macro defined in stdbool.h could
-//   be inconsistent with the bool keyword in C++. Thus, the use
-//   of stdbool.h is avoided and unsigned char is used instead.
-// * size_t is used to represent byte sizes of objects that are
-//   materialized in the address space of the calling process.
-// * int is used as an index into arrays.
-// * Deletion functions are safe to call on nullptr.
-//
-// Questions left to address:
-// * Might at some point need a way for callers to provide their own Env.
-// * Maybe add TF_TensorShape that encapsulates dimension info.
-//
-// Design decisions made:
-// * Backing store for tensor memory has an associated deallocation
-//   function.  This deallocation function will point to client code
-//   for tensors populated by the client.  So the client can do things
-//   like shadowing a numpy array.
-// * We do not provide TF_OK since it is not strictly necessary and we
-//   are not optimizing for convenience.
-// * We make assumption that one session has one graph.  This should be
-//   fine since we have the ability to run sub-graphs.
-// * We could allow NULL for some arguments (e.g., NULL options arg).
-//   However since convenience is not a primary goal, we don't do this.
-// * Devices are not in this API.  Instead, they are created/used internally
-//   and the API just provides high level controls over the number of
-//   devices of each type.
-
-// Macro to control visibility of exported symbols in the shared library (.so,
-// .dylib, .dll).
-// This duplicates the TF_EXPORT macro definition in
-// tensorflow/core/platform/macros.h in order to keep this .h file independent
-// of any other includes.
-// #ifdef SWIG
-// #define TF_CAPI_EXPORT
-// #else
-// #if defined(_WIN32)
-// #ifdef TF_COMPILE_LIBRARY
-// #define TF_CAPI_EXPORT __declspec(dllexport)
-// #else
-// #define TF_CAPI_EXPORT __declspec(dllimport)
-// #endif  // TF_COMPILE_LIBRARY
-// #else
-// #define TF_CAPI_EXPORT __attribute__((visibility("default")))
-// #endif  // _WIN32
-// #endif  // SWIG
-
-// #ifdef __cplusplus
-// #endif
-
-// --------------------------------------------------------------------------
-// TF_Version returns a string describing version information of the
-// TensorFlow library. TensorFlow using semantic versioning.
-public static native @Cast("const char*") BytePointer TF_Version();
-// Targeting ../TF_Buffer.java
-
-
-
-// Makes a copy of the input and sets an appropriate deallocator.  Useful for
-// passing in read-only, input protobufs.
-public static native TF_Buffer TF_NewBufferFromString(@Const Pointer proto,
-                                                        @Cast("size_t") long proto_len);
-
-// Useful for passing *out* a protobuf.
-public static native TF_Buffer TF_NewBuffer();
-
-public static native void TF_DeleteBuffer(TF_Buffer arg0);
-
-public static native @ByVal TF_Buffer TF_GetBuffer(TF_Buffer buffer);
-// Targeting ../TF_SessionOptions.java
-
-
-
-// Return a new options object.
-public static native TF_SessionOptions TF_NewSessionOptions();
-
-// Set the target in TF_SessionOptions.options.
-// target can be empty, a single entry, or a comma separated list of entries.
-// Each entry is in one of the following formats :
-// "local"
-// ip:port
-// host:port
-public static native void TF_SetTarget(TF_SessionOptions options,
-                                        @Cast("const char*") BytePointer target);
-public static native void TF_SetTarget(TF_SessionOptions options,
-                                        String target);
-
-// Set the config in TF_SessionOptions.options.
-// config should be a serialized tensorflow.ConfigProto proto.
-// If config was not parsed successfully as a ConfigProto, record the
-// error information in *status.
-public static native void TF_SetConfig(TF_SessionOptions options,
-                                        @Const Pointer proto, @Cast("size_t") long proto_len,
-                                        TF_Status status);
-
-// Destroy an options object.
-public static native void TF_DeleteSessionOptions(TF_SessionOptions arg0);
-// Targeting ../TF_Graph.java
-
-
-
-// Return a new graph object.
-public static native TF_Graph TF_NewGraph();
-
-// Destroy an options object.  Graph will be deleted once no more
-// TFSession's are referencing it.
-public static native void TF_DeleteGraph(TF_Graph arg0);
-// Targeting ../TF_OperationDescription.java
-
-
-// Targeting ../TF_Operation.java
-
-
-// Targeting ../TF_Input.java
-
-
-// Targeting ../TF_Output.java
-
-
-// Targeting ../TF_Function.java
-
-
-// Targeting ../TF_FunctionOptions.java
-
-
-
-// Sets the shape of the Tensor referenced by `output` in `graph` to
-// the shape described by `dims` and `num_dims`.
-//
-// If the number of dimensions is unknown, `num_dims` must be set to
-// -1 and `dims` can be null. If a dimension is unknown, the
-// corresponding entry in the `dims` array must be -1.
-//
-// This does not overwrite the existing shape associated with `output`,
-// but merges the input shape with the existing shape.  For example,
-// setting a shape of [-1, 2] with an existing shape [2, -1] would set
-// a final shape of [2, 2] based on shape merging semantics.
-//
-// Returns an error into `status` if:
-//   * `output` is not in `graph`.
-//   * An invalid shape is being set (e.g., the shape being set
-//     is incompatible with the existing shape).
-public static native void TF_GraphSetTensorShape(TF_Graph graph,
-                                                  @ByVal TF_Output output,
-                                                  @Cast("const int64_t*") LongPointer dims,
-                                                  int num_dims,
-                                                  TF_Status status);
-public static native void TF_GraphSetTensorShape(TF_Graph graph,
-                                                  @ByVal TF_Output output,
-                                                  @Cast("const int64_t*") LongBuffer dims,
-                                                  int num_dims,
-                                                  TF_Status status);
-public static native void TF_GraphSetTensorShape(TF_Graph graph,
-                                                  @ByVal TF_Output output,
-                                                  @Cast("const int64_t*") long[] dims,
-                                                  int num_dims,
-                                                  TF_Status status);
-
-// Returns the number of dimensions of the Tensor referenced by `output`
-// in `graph`.
-//
-// If the number of dimensions in the shape is unknown, returns -1.
-//
-// Returns an error into `status` if:
-//   * `output` is not in `graph`.
-public static native int TF_GraphGetTensorNumDims(TF_Graph graph,
-                                                   @ByVal TF_Output output,
-                                                   TF_Status status);
-
-// Returns the shape of the Tensor referenced by `output` in `graph`
-// into `dims`. `dims` must be an array large enough to hold `num_dims`
-// entries (e.g., the return value of TF_GraphGetTensorNumDims).
-//
-// If the number of dimensions in the shape is unknown or the shape is
-// a scalar, `dims` will remain untouched. Otherwise, each element of
-// `dims` will be set corresponding to the size of the dimension. An
-// unknown dimension is represented by `-1`.
-//
-// Returns an error into `status` if:
-//   * `output` is not in `graph`.
-//   * `num_dims` does not match the actual number of dimensions.
-public static native void TF_GraphGetTensorShape(TF_Graph graph,
-                                                  @ByVal TF_Output output,
-                                                  @Cast("int64_t*") LongPointer dims, int num_dims,
-                                                  TF_Status status);
-public static native void TF_GraphGetTensorShape(TF_Graph graph,
-                                                  @ByVal TF_Output output,
-                                                  @Cast("int64_t*") LongBuffer dims, int num_dims,
-                                                  TF_Status status);
-public static native void TF_GraphGetTensorShape(TF_Graph graph,
-                                                  @ByVal TF_Output output,
-                                                  @Cast("int64_t*") long[] dims, int num_dims,
-                                                  TF_Status status);
-
-// Operation will only be added to *graph when TF_FinishOperation() is
-// called (assuming TF_FinishOperation() does not return an error).
-// *graph must not be deleted until after TF_FinishOperation() is
-// called.
-public static native TF_OperationDescription TF_NewOperation(
-    TF_Graph graph, @Cast("const char*") BytePointer op_type, @Cast("const char*") BytePointer oper_name);
-public static native TF_OperationDescription TF_NewOperation(
-    TF_Graph graph, String op_type, String oper_name);
-
-// Specify the device for `desc`.  Defaults to empty, meaning unconstrained.
-public static native void TF_SetDevice(TF_OperationDescription desc,
-                                        @Cast("const char*") BytePointer device);
-public static native void TF_SetDevice(TF_OperationDescription desc,
-                                        String device);
-
-// The calls to TF_AddInput and TF_AddInputList must match (in number,
-// order, and type) the op declaration.  For example, the "Concat" op
-// has registration:
-//   REGISTER_OP("Concat")
-//       .Input("concat_dim: int32")
-//       .Input("values: N * T")
-//       .Output("output: T")
-//       .Attr("N: int >= 2")
-//       .Attr("T: type");
-// that defines two inputs, "concat_dim" and "values" (in that order).
-// You must use TF_AddInput() for the first input (since it takes a
-// single tensor), and TF_AddInputList() for the second input (since
-// it takes a list, even if you were to pass a list with a single
-// tensor), as in:
-//   TF_OperationDescription* desc = TF_NewOperation(graph, "Concat", "c");
-//   TF_Output concat_dim_input = {...};
-//   TF_AddInput(desc, concat_dim_input);
-//   TF_Output values_inputs[5] = {{...}, ..., {...}};
-//   TF_AddInputList(desc, values_inputs, 5);
-
-// For inputs that take a single tensor.
-public static native void TF_AddInput(TF_OperationDescription desc,
-                                       @ByVal TF_Output input);
-
-// For inputs that take a list of tensors.
-// inputs must point to TF_Output[num_inputs].
-public static native void TF_AddInputList(TF_OperationDescription desc,
-                                           @Const TF_Output inputs,
-                                           int num_inputs);
-
-// Call once per control input to `desc`.
-public static native void TF_AddControlInput(TF_OperationDescription desc,
-                                              TF_Operation input);
-
-// Request that `desc` be co-located on the device where `op`
-// is placed.
-//
-// Use of this is discouraged since the implementation of device placement is
-// subject to change. Primarily intended for internal libraries
-public static native void TF_ColocateWith(TF_OperationDescription desc,
-                                           TF_Operation op);
-
-// Call some TF_SetAttr*() function for every attr that is not
-// inferred from an input and doesn't have a default value you wish to
-// keep.
-
-// `value` must point to a string of length `length` bytes.
-public static native void TF_SetAttrString(TF_OperationDescription desc,
-                                            @Cast("const char*") BytePointer attr_name,
-                                            @Const Pointer value, @Cast("size_t") long length);
-public static native void TF_SetAttrString(TF_OperationDescription desc,
-                                            String attr_name,
-                                            @Const Pointer value, @Cast("size_t") long length);
-// `values` and `lengths` each must have lengths `num_values`.
-// `values[i]` must point to a string of length `lengths[i]` bytes.
-public static native void TF_SetAttrStringList(TF_OperationDescription desc,
-                                                @Cast("const char*") BytePointer attr_name,
-                                                @Cast("const void*const*") PointerPointer values,
-                                                @Cast("const size_t*") SizeTPointer lengths,
-                                                int num_values);
-public static native void TF_SetAttrStringList(TF_OperationDescription desc,
-                                                @Cast("const char*") BytePointer attr_name,
-                                                @Cast("const void*const*") @ByPtrPtr Pointer values,
-                                                @Cast("const size_t*") SizeTPointer lengths,
-                                                int num_values);
-public static native void TF_SetAttrStringList(TF_OperationDescription desc,
-                                                String attr_name,
-                                                @Cast("const void*const*") @ByPtrPtr Pointer values,
-                                                @Cast("const size_t*") SizeTPointer lengths,
-                                                int num_values);
-public static native void TF_SetAttrInt(TF_OperationDescription desc,
-                                         @Cast("const char*") BytePointer attr_name, @Cast("int64_t") long value);
-public static native void TF_SetAttrInt(TF_OperationDescription desc,
-                                         String attr_name, @Cast("int64_t") long value);
-public static native void TF_SetAttrIntList(TF_OperationDescription desc,
-                                             @Cast("const char*") BytePointer attr_name,
-                                             @Cast("const int64_t*") LongPointer values,
-                                             int num_values);
-public static native void TF_SetAttrIntList(TF_OperationDescription desc,
-                                             String attr_name,
-                                             @Cast("const int64_t*") LongBuffer values,
-                                             int num_values);
-public static native void TF_SetAttrIntList(TF_OperationDescription desc,
-                                             @Cast("const char*") BytePointer attr_name,
-                                             @Cast("const int64_t*") long[] values,
-                                             int num_values);
-public static native void TF_SetAttrIntList(TF_OperationDescription desc,
-                                             String attr_name,
-                                             @Cast("const int64_t*") LongPointer values,
-                                             int num_values);
-public static native void TF_SetAttrIntList(TF_OperationDescription desc,
-                                             @Cast("const char*") BytePointer attr_name,
-                                             @Cast("const int64_t*") LongBuffer values,
-                                             int num_values);
-public static native void TF_SetAttrIntList(TF_OperationDescription desc,
-                                             String attr_name,
-                                             @Cast("const int64_t*") long[] values,
-                                             int num_values);
-public static native void TF_SetAttrFloat(TF_OperationDescription desc,
-                                           @Cast("const char*") BytePointer attr_name, float value);
-public static native void TF_SetAttrFloat(TF_OperationDescription desc,
-                                           String attr_name, float value);
-public static native void TF_SetAttrFloatList(TF_OperationDescription desc,
-                                               @Cast("const char*") BytePointer attr_name,
-                                               @Const FloatPointer values,
-                                               int num_values);
-public static native void TF_SetAttrFloatList(TF_OperationDescription desc,
-                                               String attr_name,
-                                               @Const FloatBuffer values,
-                                               int num_values);
-public static native void TF_SetAttrFloatList(TF_OperationDescription desc,
-                                               @Cast("const char*") BytePointer attr_name,
-                                               @Const float[] values,
-                                               int num_values);
-public static native void TF_SetAttrFloatList(TF_OperationDescription desc,
-                                               String attr_name,
-                                               @Const FloatPointer values,
-                                               int num_values);
-public static native void TF_SetAttrFloatList(TF_OperationDescription desc,
-                                               @Cast("const char*") BytePointer attr_name,
-                                               @Const FloatBuffer values,
-                                               int num_values);
-public static native void TF_SetAttrFloatList(TF_OperationDescription desc,
-                                               String attr_name,
-                                               @Const float[] values,
-                                               int num_values);
-public static native void TF_SetAttrBool(TF_OperationDescription desc,
-                                          @Cast("const char*") BytePointer attr_name,
-                                          @Cast("unsigned char") byte value);
-public static native void TF_SetAttrBool(TF_OperationDescription desc,
-                                          String attr_name,
-                                          @Cast("unsigned char") byte value);
-public static native void TF_SetAttrBoolList(TF_OperationDescription desc,
-                                              @Cast("const char*") BytePointer attr_name,
-                                              @Cast("const unsigned char*") BytePointer values,
-                                              int num_values);
-public static native void TF_SetAttrBoolList(TF_OperationDescription desc,
-                                              String attr_name,
-                                              @Cast("const unsigned char*") ByteBuffer values,
-                                              int num_values);
-public static native void TF_SetAttrBoolList(TF_OperationDescription desc,
-                                              @Cast("const char*") BytePointer attr_name,
-                                              @Cast("const unsigned char*") byte[] values,
-                                              int num_values);
-public static native void TF_SetAttrBoolList(TF_OperationDescription desc,
-                                              String attr_name,
-                                              @Cast("const unsigned char*") BytePointer values,
-                                              int num_values);
-public static native void TF_SetAttrBoolList(TF_OperationDescription desc,
-                                              @Cast("const char*") BytePointer attr_name,
-                                              @Cast("const unsigned char*") ByteBuffer values,
-                                              int num_values);
-public static native void TF_SetAttrBoolList(TF_OperationDescription desc,
-                                              String attr_name,
-                                              @Cast("const unsigned char*") byte[] values,
-                                              int num_values);
-public static native void TF_SetAttrType(TF_OperationDescription desc,
-                                          @Cast("const char*") BytePointer attr_name,
-                                          @Cast("TF_DataType") int value);
-public static native void TF_SetAttrType(TF_OperationDescription desc,
-                                          String attr_name,
-                                          @Cast("TF_DataType") int value);
-public static native void TF_SetAttrTypeList(TF_OperationDescription desc,
-                                              @Cast("const char*") BytePointer attr_name,
-                                              @Cast("const TF_DataType*") IntPointer values,
-                                              int num_values);
-public static native void TF_SetAttrTypeList(TF_OperationDescription desc,
-                                              String attr_name,
-                                              @Cast("const TF_DataType*") IntBuffer values,
-                                              int num_values);
-public static native void TF_SetAttrTypeList(TF_OperationDescription desc,
-                                              @Cast("const char*") BytePointer attr_name,
-                                              @Cast("const TF_DataType*") int[] values,
-                                              int num_values);
-public static native void TF_SetAttrTypeList(TF_OperationDescription desc,
-                                              String attr_name,
-                                              @Cast("const TF_DataType*") IntPointer values,
-                                              int num_values);
-public static native void TF_SetAttrTypeList(TF_OperationDescription desc,
-                                              @Cast("const char*") BytePointer attr_name,
-                                              @Cast("const TF_DataType*") IntBuffer values,
-                                              int num_values);
-public static native void TF_SetAttrTypeList(TF_OperationDescription desc,
-                                              String attr_name,
-                                              @Cast("const TF_DataType*") int[] values,
-                                              int num_values);
-public static native void TF_SetAttrPlaceholder(TF_OperationDescription desc,
-                                                 @Cast("const char*") BytePointer attr_name,
-                                                 @Cast("const char*") BytePointer placeholder);
-public static native void TF_SetAttrPlaceholder(TF_OperationDescription desc,
-                                                 String attr_name,
-                                                 String placeholder);
-
-// Set a 'func' attribute to the specified name.
-// `value` must point to a string of length `length` bytes.
-public static native void TF_SetAttrFuncName(TF_OperationDescription desc,
-                                              @Cast("const char*") BytePointer attr_name,
-                                              @Cast("const char*") BytePointer value, @Cast("size_t") long length);
-public static native void TF_SetAttrFuncName(TF_OperationDescription desc,
-                                              String attr_name,
-                                              String value, @Cast("size_t") long length);
-
-// Set `num_dims` to -1 to represent "unknown rank".  Otherwise,
-// `dims` points to an array of length `num_dims`.  `dims[i]` must be
-// >= -1, with -1 meaning "unknown dimension".
-public static native void TF_SetAttrShape(TF_OperationDescription desc,
-                                           @Cast("const char*") BytePointer attr_name,
-                                           @Cast("const int64_t*") LongPointer dims, int num_dims);
-public static native void TF_SetAttrShape(TF_OperationDescription desc,
-                                           String attr_name,
-                                           @Cast("const int64_t*") LongBuffer dims, int num_dims);
-public static native void TF_SetAttrShape(TF_OperationDescription desc,
-                                           @Cast("const char*") BytePointer attr_name,
-                                           @Cast("const int64_t*") long[] dims, int num_dims);
-public static native void TF_SetAttrShape(TF_OperationDescription desc,
-                                           String attr_name,
-                                           @Cast("const int64_t*") LongPointer dims, int num_dims);
-public static native void TF_SetAttrShape(TF_OperationDescription desc,
-                                           @Cast("const char*") BytePointer attr_name,
-                                           @Cast("const int64_t*") LongBuffer dims, int num_dims);
-public static native void TF_SetAttrShape(TF_OperationDescription desc,
-                                           String attr_name,
-                                           @Cast("const int64_t*") long[] dims, int num_dims);
-// `dims` and `num_dims` must point to arrays of length `num_shapes`.
-// Set `num_dims[i]` to -1 to represent "unknown rank".  Otherwise,
-// `dims[i]` points to an array of length `num_dims[i]`.  `dims[i][j]`
-// must be >= -1, with -1 meaning "unknown dimension".
-public static native void TF_SetAttrShapeList(TF_OperationDescription desc,
-                                               @Cast("const char*") BytePointer attr_name,
-                                               @Cast("const int64_t*const*") PointerPointer dims,
-                                               @Const IntPointer num_dims,
-                                               int num_shapes);
-public static native void TF_SetAttrShapeList(TF_OperationDescription desc,
-                                               @Cast("const char*") BytePointer attr_name,
-                                               @Cast("const int64_t*const*") @ByPtrPtr LongPointer dims,
-                                               @Const IntPointer num_dims,
-                                               int num_shapes);
-public static native void TF_SetAttrShapeList(TF_OperationDescription desc,
-                                               String attr_name,
-                                               @Cast("const int64_t*const*") @ByPtrPtr LongBuffer dims,
-                                               @Const IntBuffer num_dims,
-                                               int num_shapes);
-public static native void TF_SetAttrShapeList(TF_OperationDescription desc,
-                                               @Cast("const char*") BytePointer attr_name,
-                                               @Cast("const int64_t*const*") @ByPtrPtr long[] dims,
-                                               @Const int[] num_dims,
-                                               int num_shapes);
-public static native void TF_SetAttrShapeList(TF_OperationDescription desc,
-                                               String attr_name,
-                                               @Cast("const int64_t*const*") @ByPtrPtr LongPointer dims,
-                                               @Const IntPointer num_dims,
-                                               int num_shapes);
-public static native void TF_SetAttrShapeList(TF_OperationDescription desc,
-                                               @Cast("const char*") BytePointer attr_name,
-                                               @Cast("const int64_t*const*") @ByPtrPtr LongBuffer dims,
-                                               @Const IntBuffer num_dims,
-                                               int num_shapes);
-public static native void TF_SetAttrShapeList(TF_OperationDescription desc,
-                                               String attr_name,
-                                               @Cast("const int64_t*const*") @ByPtrPtr long[] dims,
-                                               @Const int[] num_dims,
-                                               int num_shapes);
-// `proto` must point to an array of `proto_len` bytes representing a
-// binary-serialized TensorShapeProto.
-public static native void TF_SetAttrTensorShapeProto(
-    TF_OperationDescription desc, @Cast("const char*") BytePointer attr_name, @Const Pointer proto,
-    @Cast("size_t") long proto_len, TF_Status status);
-public static native void TF_SetAttrTensorShapeProto(
-    TF_OperationDescription desc, String attr_name, @Const Pointer proto,
-    @Cast("size_t") long proto_len, TF_Status status);
-// `protos` and `proto_lens` must point to arrays of length `num_shapes`.
-// `protos[i]` must point to an array of `proto_lens[i]` bytes
-// representing a binary-serialized TensorShapeProto.
-public static native void TF_SetAttrTensorShapeProtoList(
-    TF_OperationDescription desc, @Cast("const char*") BytePointer attr_name,
-    @Cast("const void*const*") PointerPointer protos, @Cast("const size_t*") SizeTPointer proto_lens, int num_shapes,
-    TF_Status status);
-public static native void TF_SetAttrTensorShapeProtoList(
-    TF_OperationDescription desc, @Cast("const char*") BytePointer attr_name,
-    @Cast("const void*const*") @ByPtrPtr Pointer protos, @Cast("const size_t*") SizeTPointer proto_lens, int num_shapes,
-    TF_Status status);
-public static native void TF_SetAttrTensorShapeProtoList(
-    TF_OperationDescription desc, String attr_name,
-    @Cast("const void*const*") @ByPtrPtr Pointer protos, @Cast("const size_t*") SizeTPointer proto_lens, int num_shapes,
-    TF_Status status);
-
-public static native void TF_SetAttrTensor(TF_OperationDescription desc,
-                                            @Cast("const char*") BytePointer attr_name,
-                                            TF_Tensor value,
-                                            TF_Status status);
-public static native void TF_SetAttrTensor(TF_OperationDescription desc,
-                                            String attr_name,
-                                            TF_Tensor value,
-                                            TF_Status status);
-public static native void TF_SetAttrTensorList(TF_OperationDescription desc,
-                                                @Cast("const char*") BytePointer attr_name,
-                                                @Cast("TF_Tensor*const*") PointerPointer values,
-                                                int num_values,
-                                                TF_Status status);
-public static native void TF_SetAttrTensorList(TF_OperationDescription desc,
-                                                @Cast("const char*") BytePointer attr_name,
-                                                @ByPtrPtr TF_Tensor values,
-                                                int num_values,
-                                                TF_Status status);
-public static native void TF_SetAttrTensorList(TF_OperationDescription desc,
-                                                String attr_name,
-                                                @ByPtrPtr TF_Tensor values,
-                                                int num_values,
-                                                TF_Status status);
-
-// `proto` should point to a sequence of bytes of length `proto_len`
-// representing a binary serialization of an AttrValue protocol
-// buffer.
-public static native void TF_SetAttrValueProto(TF_OperationDescription desc,
-                                                @Cast("const char*") BytePointer attr_name,
-                                                @Const Pointer proto,
-                                                @Cast("size_t") long proto_len,
-                                                TF_Status status);
-public static native void TF_SetAttrValueProto(TF_OperationDescription desc,
-                                                String attr_name,
-                                                @Const Pointer proto,
-                                                @Cast("size_t") long proto_len,
-                                                TF_Status status);
-
-// If this function succeeds:
-//   * *status is set to an OK value,
-//   * a TF_Operation is added to the graph,
-//   * a non-null value pointing to the added operation is returned --
-//     this value is valid until the underlying graph is deleted.
-// Otherwise:
-//   * *status is set to a non-OK value,
-//   * the graph is not modified,
-//   * a null value is returned.
-// In either case, it deletes `desc`.
-public static native TF_Operation TF_FinishOperation(
-    TF_OperationDescription desc, TF_Status status);
-
-// TF_Operation functions.  Operations are immutable once created, so
-// these are all query functions.
-
-public static native @Cast("const char*") BytePointer TF_OperationName(TF_Operation oper);
-public static native @Cast("const char*") BytePointer TF_OperationOpType(TF_Operation oper);
-public static native @Cast("const char*") BytePointer TF_OperationDevice(TF_Operation oper);
-
-public static native int TF_OperationNumOutputs(TF_Operation oper);
-public static native @Cast("TF_DataType") int TF_OperationOutputType(@ByVal TF_Output oper_out);
-public static native int TF_OperationOutputListLength(TF_Operation oper,
-                                                       @Cast("const char*") BytePointer arg_name,
-                                                       TF_Status status);
-public static native int TF_OperationOutputListLength(TF_Operation oper,
-                                                       String arg_name,
-                                                       TF_Status status);
-
-public static native int TF_OperationNumInputs(TF_Operation oper);
-public static native @Cast("TF_DataType") int TF_OperationInputType(@ByVal TF_Input oper_in);
-public static native int TF_OperationInputListLength(TF_Operation oper,
-                                                      @Cast("const char*") BytePointer arg_name,
-                                                      TF_Status status);
-public static native int TF_OperationInputListLength(TF_Operation oper,
-                                                      String arg_name,
-                                                      TF_Status status);
-
-// In this code:
-//   TF_Output producer = TF_OperationInput(consumer);
-// There is an edge from producer.oper's output (given by
-// producer.index) to consumer.oper's input (given by consumer.index).
-public static native @ByVal TF_Output TF_OperationInput(@ByVal TF_Input oper_in);
-
-// Get list of all inputs of a specific operation.  `inputs` must point to
-// an array of length at least `max_inputs` (ideally set to
-// TF_OperationNumInputs(oper)).  Beware that a concurrent
-// modification of the graph can increase the number of inputs of
-// an operation.
-public static native void TF_OperationAllInputs(TF_Operation oper,
-                                                 TF_Output inputs,
-                                                 int max_inputs);
-
-// Get the number of current consumers of a specific output of an
-// operation.  Note that this number can change when new operations
-// are added to the graph.
-public static native int TF_OperationOutputNumConsumers(@ByVal TF_Output oper_out);
-
-// Get list of all current consumers of a specific output of an
-// operation.  `consumers` must point to an array of length at least
-// `max_consumers` (ideally set to
-// TF_OperationOutputNumConsumers(oper_out)).  Beware that a concurrent
-// modification of the graph can increase the number of consumers of
-// an operation.  Returns the number of output consumers (should match
-// TF_OperationOutputNumConsumers(oper_out)).
-public static native int TF_OperationOutputConsumers(@ByVal TF_Output oper_out,
-                                                      TF_Input consumers,
-                                                      int max_consumers);
-
-// Get the number of control inputs to an operation.
-public static native int TF_OperationNumControlInputs(TF_Operation oper);
-
-// Get list of all control inputs to an operation.  `control_inputs` must
-// point to an array of length `max_control_inputs` (ideally set to
-// TF_OperationNumControlInputs(oper)).  Returns the number of control
-// inputs (should match TF_OperationNumControlInputs(oper)).
-public static native int TF_OperationGetControlInputs(
-    TF_Operation oper, @Cast("TF_Operation**") PointerPointer control_inputs, int max_control_inputs);
-public static native int TF_OperationGetControlInputs(
-    TF_Operation oper, @ByPtrPtr TF_Operation control_inputs, int max_control_inputs);
-
-// Get the number of operations that have `*oper` as a control input.
-// Note that this number can change when new operations are added to
-// the graph.
-public static native int TF_OperationNumControlOutputs(TF_Operation oper);
-
-// Get the list of operations that have `*oper` as a control input.
-// `control_outputs` must point to an array of length at least
-// `max_control_outputs` (ideally set to
-// TF_OperationNumControlOutputs(oper)). Beware that a concurrent
-// modification of the graph can increase the number of control
-// outputs.  Returns the number of control outputs (should match
-// TF_OperationNumControlOutputs(oper)).
-public static native int TF_OperationGetControlOutputs(
-    TF_Operation oper, @Cast("TF_Operation**") PointerPointer control_outputs,
-    int max_control_outputs);
-public static native int TF_OperationGetControlOutputs(
-    TF_Operation oper, @ByPtrPtr TF_Operation control_outputs,
-    int max_control_outputs);
-// Targeting ../TF_AttrMetadata.java
-
-
-
-// Returns metadata about the value of the attribute `attr_name` of `oper`.
-public static native @ByVal TF_AttrMetadata TF_OperationGetAttrMetadata(
-    TF_Operation oper, @Cast("const char*") BytePointer attr_name, TF_Status status);
-public static native @ByVal TF_AttrMetadata TF_OperationGetAttrMetadata(
-    TF_Operation oper, String attr_name, TF_Status status);
-
-// Fills in `value` with the value of the attribute `attr_name`.  `value` must
-// point to an array of length at least `max_length` (ideally set to
-// TF_AttrMetadata.total_size from TF_OperationGetAttrMetadata(oper,
-// attr_name)).
-public static native void TF_OperationGetAttrString(TF_Operation oper,
-                                                     @Cast("const char*") BytePointer attr_name,
-                                                     Pointer value,
-                                                     @Cast("size_t") long max_length,
-                                                     TF_Status status);
-public static native void TF_OperationGetAttrString(TF_Operation oper,
-                                                     String attr_name,
-                                                     Pointer value,
-                                                     @Cast("size_t") long max_length,
-                                                     TF_Status status);
-
-// Get the list of strings in the value of the attribute `attr_name`.  Fills in
-// `values` and `lengths`, each of which must point to an array of length at
-// least `max_values`.
-//
-// The elements of values will point to addresses in `storage` which must be at
-// least `storage_size` bytes in length.  Ideally, max_values would be set to
-// TF_AttrMetadata.list_size and `storage` would be at least
-// TF_AttrMetadata.total_size, obtained from TF_OperationGetAttrMetadata(oper,
-// attr_name).
-//
-// Fails if storage_size is too small to hold the requested number of strings.
-public static native void TF_OperationGetAttrStringList(
-    TF_Operation oper, @Cast("const char*") BytePointer attr_name, @Cast("void**") PointerPointer values, @Cast("size_t*") SizeTPointer lengths,
-    int max_values, Pointer storage, @Cast("size_t") long storage_size, TF_Status status);
-public static native void TF_OperationGetAttrStringList(
-    TF_Operation oper, @Cast("const char*") BytePointer attr_name, @Cast("void**") @ByPtrPtr Pointer values, @Cast("size_t*") SizeTPointer lengths,
-    int max_values, Pointer storage, @Cast("size_t") long storage_size, TF_Status status);
-public static native void TF_OperationGetAttrStringList(
-    TF_Operation oper, String attr_name, @Cast("void**") @ByPtrPtr Pointer values, @Cast("size_t*") SizeTPointer lengths,
-    int max_values, Pointer storage, @Cast("size_t") long storage_size, TF_Status status);
-
-public static native void TF_OperationGetAttrInt(TF_Operation oper,
-                                                  @Cast("const char*") BytePointer attr_name,
-                                                  @Cast("int64_t*") LongPointer value,
-                                                  TF_Status status);
-public static native void TF_OperationGetAttrInt(TF_Operation oper,
-                                                  String attr_name,
-                                                  @Cast("int64_t*") LongBuffer value,
-                                                  TF_Status status);
-public static native void TF_OperationGetAttrInt(TF_Operation oper,
-                                                  @Cast("const char*") BytePointer attr_name,
-                                                  @Cast("int64_t*") long[] value,
-                                                  TF_Status status);
-public static native void TF_OperationGetAttrInt(TF_Operation oper,
-                                                  String attr_name,
-                                                  @Cast("int64_t*") LongPointer value,
-                                                  TF_Status status);
-public static native void TF_OperationGetAttrInt(TF_Operation oper,
-                                                  @Cast("const char*") BytePointer attr_name,
-                                                  @Cast("int64_t*") LongBuffer value,
-                                                  TF_Status status);
-public static native void TF_OperationGetAttrInt(TF_Operation oper,
-                                                  String attr_name,
-                                                  @Cast("int64_t*") long[] value,
-                                                  TF_Status status);
-
-// Fills in `values` with the value of the attribute `attr_name` of `oper`.
-// `values` must point to an array of length at least `max_values` (ideally set
-// TF_AttrMetadata.list_size from TF_OperationGetAttrMetadata(oper,
-// attr_name)).
-public static native void TF_OperationGetAttrIntList(TF_Operation oper,
-                                                      @Cast("const char*") BytePointer attr_name,
-                                                      @Cast("int64_t*") LongPointer values,
-                                                      int max_values,
-                                                      TF_Status status);
-public static native void TF_OperationGetAttrIntList(TF_Operation oper,
-                                                      String attr_name,
-                                                      @Cast("int64_t*") LongBuffer values,
-                                                      int max_values,
-                                                      TF_Status status);
-public static native void TF_OperationGetAttrIntList(TF_Operation oper,
-                                                      @Cast("const char*") BytePointer attr_name,
-                                                      @Cast("int64_t*") long[] values,
-                                                      int max_values,
-                                                      TF_Status status);
-public static native void TF_OperationGetAttrIntList(TF_Operation oper,
-                                                      String attr_name,
-                                                      @Cast("int64_t*") LongPointer values,
-                                                      int max_values,
-                                                      TF_Status status);
-public static native void TF_OperationGetAttrIntList(TF_Operation oper,
-                                                      @Cast("const char*") BytePointer attr_name,
-                                                      @Cast("int64_t*") LongBuffer values,
-                                                      int max_values,
-                                                      TF_Status status);
-public static native void TF_OperationGetAttrIntList(TF_Operation oper,
-                                                      String attr_name,
-                                                      @Cast("int64_t*") long[] values,
-                                                      int max_values,
-                                                      TF_Status status);
-
-public static native void TF_OperationGetAttrFloat(TF_Operation oper,
-                                                    @Cast("const char*") BytePointer attr_name,
-                                                    FloatPointer value,
-                                                    TF_Status status);
-public static native void TF_OperationGetAttrFloat(TF_Operation oper,
-                                                    String attr_name,
-                                                    FloatBuffer value,
-                                                    TF_Status status);
-public static native void TF_OperationGetAttrFloat(TF_Operation oper,
-                                                    @Cast("const char*") BytePointer attr_name,
-                                                    float[] value,
-                                                    TF_Status status);
-public static native void TF_OperationGetAttrFloat(TF_Operation oper,
-                                                    String attr_name,
-                                                    FloatPointer value,
-                                                    TF_Status status);
-public static native void TF_OperationGetAttrFloat(TF_Operation oper,
-                                                    @Cast("const char*") BytePointer attr_name,
-                                                    FloatBuffer value,
-                                                    TF_Status status);
-public static native void TF_OperationGetAttrFloat(TF_Operation oper,
-                                                    String attr_name,
-                                                    float[] value,
-                                                    TF_Status status);
-
-// Fills in `values` with the value of the attribute `attr_name` of `oper`.
-// `values` must point to an array of length at least `max_values` (ideally set
-// to TF_AttrMetadata.list_size from TF_OperationGetAttrMetadata(oper,
-// attr_name)).
-public static native void TF_OperationGetAttrFloatList(TF_Operation oper,
-                                                        @Cast("const char*") BytePointer attr_name,
-                                                        FloatPointer values,
-                                                        int max_values,
-                                                        TF_Status status);
-public static native void TF_OperationGetAttrFloatList(TF_Operation oper,
-                                                        String attr_name,
-                                                        FloatBuffer values,
-                                                        int max_values,
-                                                        TF_Status status);
-public static native void TF_OperationGetAttrFloatList(TF_Operation oper,
-                                                        @Cast("const char*") BytePointer attr_name,
-                                                        float[] values,
-                                                        int max_values,
-                                                        TF_Status status);
-public static native void TF_OperationGetAttrFloatList(TF_Operation oper,
-                                                        String attr_name,
-                                                        FloatPointer values,
-                                                        int max_values,
-                                                        TF_Status status);
-public static native void TF_OperationGetAttrFloatList(TF_Operation oper,
-                                                        @Cast("const char*") BytePointer attr_name,
-                                                        FloatBuffer values,
-                                                        int max_values,
-                                                        TF_Status status);
-public static native void TF_OperationGetAttrFloatList(TF_Operation oper,
-                                                        String attr_name,
-                                                        float[] values,
-                                                        int max_values,
-                                                        TF_Status status);
-
-public static native void TF_OperationGetAttrBool(TF_Operation oper,
-                                                   @Cast("const char*") BytePointer attr_name,
-                                                   @Cast("unsigned char*") BytePointer value,
-                                                   TF_Status status);
-public static native void TF_OperationGetAttrBool(TF_Operation oper,
-                                                   String attr_name,
-                                                   @Cast("unsigned char*") ByteBuffer value,
-                                                   TF_Status status);
-public static native void TF_OperationGetAttrBool(TF_Operation oper,
-                                                   @Cast("const char*") BytePointer attr_name,
-                                                   @Cast("unsigned char*") byte[] value,
-                                                   TF_Status status);
-public static native void TF_OperationGetAttrBool(TF_Operation oper,
-                                                   String attr_name,
-                                                   @Cast("unsigned char*") BytePointer value,
-                                                   TF_Status status);
-public static native void TF_OperationGetAttrBool(TF_Operation oper,
-                                                   @Cast("const char*") BytePointer attr_name,
-                                                   @Cast("unsigned char*") ByteBuffer value,
-                                                   TF_Status status);
-public static native void TF_OperationGetAttrBool(TF_Operation oper,
-                                                   String attr_name,
-                                                   @Cast("unsigned char*") byte[] value,
-                                                   TF_Status status);
-
-// Fills in `values` with the value of the attribute `attr_name` of `oper`.
-// `values` must point to an array of length at least `max_values` (ideally set
-// to TF_AttrMetadata.list_size from TF_OperationGetAttrMetadata(oper,
-// attr_name)).
-public static native void TF_OperationGetAttrBoolList(TF_Operation oper,
-                                                       @Cast("const char*") BytePointer attr_name,
-                                                       @Cast("unsigned char*") BytePointer values,
-                                                       int max_values,
-                                                       TF_Status status);
-public static native void TF_OperationGetAttrBoolList(TF_Operation oper,
-                                                       String attr_name,
-                                                       @Cast("unsigned char*") ByteBuffer values,
-                                                       int max_values,
-                                                       TF_Status status);
-public static native void TF_OperationGetAttrBoolList(TF_Operation oper,
-                                                       @Cast("const char*") BytePointer attr_name,
-                                                       @Cast("unsigned char*") byte[] values,
-                                                       int max_values,
-                                                       TF_Status status);
-public static native void TF_OperationGetAttrBoolList(TF_Operation oper,
-                                                       String attr_name,
-                                                       @Cast("unsigned char*") BytePointer values,
-                                                       int max_values,
-                                                       TF_Status status);
-public static native void TF_OperationGetAttrBoolList(TF_Operation oper,
-                                                       @Cast("const char*") BytePointer attr_name,
-                                                       @Cast("unsigned char*") ByteBuffer values,
-                                                       int max_values,
-                                                       TF_Status status);
-public static native void TF_OperationGetAttrBoolList(TF_Operation oper,
-                                                       String attr_name,
-                                                       @Cast("unsigned char*") byte[] values,
-                                                       int max_values,
-                                                       TF_Status status);
-
-public static native void TF_OperationGetAttrType(TF_Operation oper,
-                                                   @Cast("const char*") BytePointer attr_name,
-                                                   @Cast("TF_DataType*") IntPointer value,
-                                                   TF_Status status);
-public static native void TF_OperationGetAttrType(TF_Operation oper,
-                                                   String attr_name,
-                                                   @Cast("TF_DataType*") IntBuffer value,
-                                                   TF_Status status);
-public static native void TF_OperationGetAttrType(TF_Operation oper,
-                                                   @Cast("const char*") BytePointer attr_name,
-                                                   @Cast("TF_DataType*") int[] value,
-                                                   TF_Status status);
-public static native void TF_OperationGetAttrType(TF_Operation oper,
-                                                   String attr_name,
-                                                   @Cast("TF_DataType*") IntPointer value,
-                                                   TF_Status status);
-public static native void TF_OperationGetAttrType(TF_Operation oper,
-                                                   @Cast("const char*") BytePointer attr_name,
-                                                   @Cast("TF_DataType*") IntBuffer value,
-                                                   TF_Status status);
-public static native void TF_OperationGetAttrType(TF_Operation oper,
-                                                   String attr_name,
-                                                   @Cast("TF_DataType*") int[] value,
-                                                   TF_Status status);
-
-// Fills in `values` with the value of the attribute `attr_name` of `oper`.
-// `values` must point to an array of length at least `max_values` (ideally set
-// to TF_AttrMetadata.list_size from TF_OperationGetAttrMetadata(oper,
-// attr_name)).
-public static native void TF_OperationGetAttrTypeList(TF_Operation oper,
-                                                       @Cast("const char*") BytePointer attr_name,
-                                                       @Cast("TF_DataType*") IntPointer values,
-                                                       int max_values,
-                                                       TF_Status status);
-public static native void TF_OperationGetAttrTypeList(TF_Operation oper,
-                                                       String attr_name,
-                                                       @Cast("TF_DataType*") IntBuffer values,
-                                                       int max_values,
-                                                       TF_Status status);
-public static native void TF_OperationGetAttrTypeList(TF_Operation oper,
-                                                       @Cast("const char*") BytePointer attr_name,
-                                                       @Cast("TF_DataType*") int[] values,
-                                                       int max_values,
-                                                       TF_Status status);
-public static native void TF_OperationGetAttrTypeList(TF_Operation oper,
-                                                       String attr_name,
-                                                       @Cast("TF_DataType*") IntPointer values,
-                                                       int max_values,
-                                                       TF_Status status);
-public static native void TF_OperationGetAttrTypeList(TF_Operation oper,
-                                                       @Cast("const char*") BytePointer attr_name,
-                                                       @Cast("TF_DataType*") IntBuffer values,
-                                                       int max_values,
-                                                       TF_Status status);
-public static native void TF_OperationGetAttrTypeList(TF_Operation oper,
-                                                       String attr_name,
-                                                       @Cast("TF_DataType*") int[] values,
-                                                       int max_values,
-                                                       TF_Status status);
-
-// Fills in `value` with the value of the attribute `attr_name` of `oper`.
-// `values` must point to an array of length at least `num_dims` (ideally set to
-// TF_Attr_Meta.size from TF_OperationGetAttrMetadata(oper, attr_name)).
-public static native void TF_OperationGetAttrShape(TF_Operation oper,
-                                                    @Cast("const char*") BytePointer attr_name,
-                                                    @Cast("int64_t*") LongPointer value,
-                                                    int num_dims,
-                                                    TF_Status status);
-public static native void TF_OperationGetAttrShape(TF_Operation oper,
-                                                    String attr_name,
-                                                    @Cast("int64_t*") LongBuffer value,
-                                                    int num_dims,
-                                                    TF_Status status);
-public static native void TF_OperationGetAttrShape(TF_Operation oper,
-                                                    @Cast("const char*") BytePointer attr_name,
-                                                    @Cast("int64_t*") long[] value,
-                                                    int num_dims,
-                                                    TF_Status status);
-public static native void TF_OperationGetAttrShape(TF_Operation oper,
-                                                    String attr_name,
-                                                    @Cast("int64_t*") LongPointer value,
-                                                    int num_dims,
-                                                    TF_Status status);
-public static native void TF_OperationGetAttrShape(TF_Operation oper,
-                                                    @Cast("const char*") BytePointer attr_name,
-                                                    @Cast("int64_t*") LongBuffer value,
-                                                    int num_dims,
-                                                    TF_Status status);
-public static native void TF_OperationGetAttrShape(TF_Operation oper,
-                                                    String attr_name,
-                                                    @Cast("int64_t*") long[] value,
-                                                    int num_dims,
-                                                    TF_Status status);
-
-// Fills in `dims` with the list of shapes in the attribute `attr_name` of
-// `oper` and `num_dims` with the corresponding number of dimensions. On return,
-// for every i where `num_dims[i]` > 0, `dims[i]` will be an array of
-// `num_dims[i]` elements. A value of -1 for `num_dims[i]` indicates that the
-// i-th shape in the list is unknown.
-//
-// The elements of `dims` will point to addresses in `storage` which must be
-// large enough to hold at least `storage_size` int64_ts.  Ideally, `num_shapes`
-// would be set to TF_AttrMetadata.list_size and `storage_size` would be set to
-// TF_AttrMetadata.total_size from TF_OperationGetAttrMetadata(oper,
-// attr_name).
-//
-// Fails if storage_size is insufficient to hold the requested shapes.
-public static native void TF_OperationGetAttrShapeList(
-    TF_Operation oper, @Cast("const char*") BytePointer attr_name, @Cast("int64_t**") PointerPointer dims, IntPointer num_dims,
-    int num_shapes, @Cast("int64_t*") LongPointer storage, int storage_size, TF_Status status);
-public static native void TF_OperationGetAttrShapeList(
-    TF_Operation oper, @Cast("const char*") BytePointer attr_name, @Cast("int64_t**") @ByPtrPtr LongPointer dims, IntPointer num_dims,
-    int num_shapes, @Cast("int64_t*") LongPointer storage, int storage_size, TF_Status status);
-public static native void TF_OperationGetAttrShapeList(
-    TF_Operation oper, String attr_name, @Cast("int64_t**") @ByPtrPtr LongBuffer dims, IntBuffer num_dims,
-    int num_shapes, @Cast("int64_t*") LongBuffer storage, int storage_size, TF_Status status);
-public static native void TF_OperationGetAttrShapeList(
-    TF_Operation oper, @Cast("const char*") BytePointer attr_name, @Cast("int64_t**") @ByPtrPtr long[] dims, int[] num_dims,
-    int num_shapes, @Cast("int64_t*") long[] storage, int storage_size, TF_Status status);
-public static native void TF_OperationGetAttrShapeList(
-    TF_Operation oper, String attr_name, @Cast("int64_t**") @ByPtrPtr LongPointer dims, IntPointer num_dims,
-    int num_shapes, @Cast("int64_t*") LongPointer storage, int storage_size, TF_Status status);
-public static native void TF_OperationGetAttrShapeList(
-    TF_Operation oper, @Cast("const char*") BytePointer attr_name, @Cast("int64_t**") @ByPtrPtr LongBuffer dims, IntBuffer num_dims,
-    int num_shapes, @Cast("int64_t*") LongBuffer storage, int storage_size, TF_Status status);
-public static native void TF_OperationGetAttrShapeList(
-    TF_Operation oper, String attr_name, @Cast("int64_t**") @ByPtrPtr long[] dims, int[] num_dims,
-    int num_shapes, @Cast("int64_t*") long[] storage, int storage_size, TF_Status status);
-
-// Sets `value` to the binary-serialized TensorShapeProto of the value of
-// `attr_name` attribute of `oper`'.
-public static native void TF_OperationGetAttrTensorShapeProto(
-    TF_Operation oper, @Cast("const char*") BytePointer attr_name, TF_Buffer value,
-    TF_Status status);
-public static native void TF_OperationGetAttrTensorShapeProto(
-    TF_Operation oper, String attr_name, TF_Buffer value,
-    TF_Status status);
-
-// Fills in `values` with binary-serialized TensorShapeProto values of the
-// attribute `attr_name` of `oper`. `values` must point to an array of length at
-// least `num_values` (ideally set to TF_AttrMetadata.list_size from
-// TF_OperationGetAttrMetadata(oper, attr_name)).
-public static native void TF_OperationGetAttrTensorShapeProtoList(
-    TF_Operation oper, @Cast("const char*") BytePointer attr_name, @Cast("TF_Buffer**") PointerPointer values,
-    int max_values, TF_Status status);
-public static native void TF_OperationGetAttrTensorShapeProtoList(
-    TF_Operation oper, @Cast("const char*") BytePointer attr_name, @ByPtrPtr TF_Buffer values,
-    int max_values, TF_Status status);
-public static native void TF_OperationGetAttrTensorShapeProtoList(
-    TF_Operation oper, String attr_name, @ByPtrPtr TF_Buffer values,
-    int max_values, TF_Status status);
-
-// Gets the TF_Tensor valued attribute of `attr_name` of `oper`.
-//
-// Allocates a new TF_Tensor which the caller is expected to take
-// ownership of (and can deallocate using TF_DeleteTensor).
-public static native void TF_OperationGetAttrTensor(TF_Operation oper,
-                                                     @Cast("const char*") BytePointer attr_name,
-                                                     @Cast("TF_Tensor**") PointerPointer value,
-                                                     TF_Status status);
-public static native void TF_OperationGetAttrTensor(TF_Operation oper,
-                                                     @Cast("const char*") BytePointer attr_name,
-                                                     @ByPtrPtr TF_Tensor value,
-                                                     TF_Status status);
-public static native void TF_OperationGetAttrTensor(TF_Operation oper,
-                                                     String attr_name,
-                                                     @ByPtrPtr TF_Tensor value,
-                                                     TF_Status status);
-
-// Fills in `values` with the TF_Tensor values of the attribute `attr_name` of
-// `oper`. `values` must point to an array of TF_Tensor* of length at least
-// `max_values` (ideally set to TF_AttrMetadata.list_size from
-// TF_OperationGetAttrMetadata(oper, attr_name)).
-//
-// The caller takes ownership of all the non-null TF_Tensor* entries in `values`
-// (which can be deleted using TF_DeleteTensor(values[i])).
-public static native void TF_OperationGetAttrTensorList(TF_Operation oper,
-                                                         @Cast("const char*") BytePointer attr_name,
-                                                         @Cast("TF_Tensor**") PointerPointer values,
-                                                         int max_values,
-                                                         TF_Status status);
-public static native void TF_OperationGetAttrTensorList(TF_Operation oper,
-                                                         @Cast("const char*") BytePointer attr_name,
-                                                         @ByPtrPtr TF_Tensor values,
-                                                         int max_values,
-                                                         TF_Status status);
-public static native void TF_OperationGetAttrTensorList(TF_Operation oper,
-                                                         String attr_name,
-                                                         @ByPtrPtr TF_Tensor values,
-                                                         int max_values,
-                                                         TF_Status status);
-
-// Sets `output_attr_value` to the binary-serialized AttrValue proto
-// representation of the value of the `attr_name` attr of `oper`.
-public static native void TF_OperationGetAttrValueProto(
-    TF_Operation oper, @Cast("const char*") BytePointer attr_name, TF_Buffer output_attr_value,
-    TF_Status status);
-public static native void TF_OperationGetAttrValueProto(
-    TF_Operation oper, String attr_name, TF_Buffer output_attr_value,
-    TF_Status status);
-
-// Returns the operation in the graph with `oper_name`. Returns nullptr if
-// no operation found.
-public static native TF_Operation TF_GraphOperationByName(
-    TF_Graph graph, @Cast("const char*") BytePointer oper_name);
-public static native TF_Operation TF_GraphOperationByName(
-    TF_Graph graph, String oper_name);
-
-// Iterate through the operations of a graph.  To use:
-// size_t pos = 0;
-// TF_Operation* oper;
-// while ((oper = TF_GraphNextOperation(graph, &pos)) != nullptr) {
-//   DoSomethingWithOperation(oper);
-// }
-public static native TF_Operation TF_GraphNextOperation(TF_Graph graph,
-                                                          @Cast("size_t*") SizeTPointer pos);
-
-// Write out a serialized representation of `graph` (as a GraphDef protocol
-// message) to `output_graph_def` (allocated by TF_NewBuffer()).
-// `output_graph_def`'s underlying buffer will be freed when TF_DeleteBuffer()
-// is called.
-//
-// May fail on very large graphs in the future.
-public static native void TF_GraphToGraphDef(TF_Graph graph,
-                                              TF_Buffer output_graph_def,
-                                              TF_Status status);
-
-// Returns the serialized OpDef proto with name `op_name`, or a bad status if no
-// such op exists. This can return OpDefs of functions copied into the graph.
-public static native void TF_GraphGetOpDef(TF_Graph graph,
-                                            @Cast("const char*") BytePointer op_name,
-                                            TF_Buffer output_op_def,
-                                            TF_Status status);
-public static native void TF_GraphGetOpDef(TF_Graph graph,
-                                            String op_name,
-                                            TF_Buffer output_op_def,
-                                            TF_Status status);
-
-// Returns the serialized VersionDef proto for this graph.
-public static native void TF_GraphVersions(TF_Graph graph,
-                                            TF_Buffer output_version_def,
-                                            TF_Status status);
-// Targeting ../TF_ImportGraphDefOptions.java
-
-
-
-public static native TF_ImportGraphDefOptions TF_NewImportGraphDefOptions();
-public static native void TF_DeleteImportGraphDefOptions(
-    TF_ImportGraphDefOptions opts);
-
-// Set the prefix to be prepended to the names of nodes in `graph_def` that will
-// be imported into `graph`. `prefix` is copied and has no lifetime
-// requirements.
-public static native void TF_ImportGraphDefOptionsSetPrefix(
-    TF_ImportGraphDefOptions opts, @Cast("const char*") BytePointer prefix);
-public static native void TF_ImportGraphDefOptionsSetPrefix(
-    TF_ImportGraphDefOptions opts, String prefix);
-
-// Set the execution device for nodes in `graph_def`.
-// Only applies to nodes where a device was not already explicitly specified.
-// `device` is copied and has no lifetime requirements.
-public static native void TF_ImportGraphDefOptionsSetDefaultDevice(
-    TF_ImportGraphDefOptions opts, @Cast("const char*") BytePointer device);
-public static native void TF_ImportGraphDefOptionsSetDefaultDevice(
-    TF_ImportGraphDefOptions opts, String device);
-
-// Set whether to uniquify imported operation names. If true, imported operation
-// names will be modified if their name already exists in the graph. If false,
-// conflicting names will be treated as an error. Note that this option has no
-// effect if a prefix is set, since the prefix will guarantee all names are
-// unique. Defaults to false.
-public static native void TF_ImportGraphDefOptionsSetUniquifyNames(
-    TF_ImportGraphDefOptions opts, @Cast("unsigned char") byte uniquify_names);
-
-// If true, the specified prefix will be modified if it already exists as an
-// operation name or prefix in the graph. If false, a conflicting prefix will be
-// treated as an error. This option has no effect if no prefix is specified.
-public static native void TF_ImportGraphDefOptionsSetUniquifyPrefix(
-    TF_ImportGraphDefOptions opts, @Cast("unsigned char") byte uniquify_prefix);
-
-// Set any imported nodes with input `src_name:src_index` to have that input
-// replaced with `dst`. `src_name` refers to a node in the graph to be imported,
-// `dst` references a node already existing in the graph being imported into.
-// `src_name` is copied and has no lifetime requirements.
-public static native void TF_ImportGraphDefOptionsAddInputMapping(
-    TF_ImportGraphDefOptions opts, @Cast("const char*") BytePointer src_name, int src_index,
-    @ByVal TF_Output dst);
-public static native void TF_ImportGraphDefOptionsAddInputMapping(
-    TF_ImportGraphDefOptions opts, String src_name, int src_index,
-    @ByVal TF_Output dst);
-
-// Set any imported nodes with control input `src_name` to have that input
-// replaced with `dst`. `src_name` refers to a node in the graph to be imported,
-// `dst` references an operation already existing in the graph being imported
-// into. `src_name` is copied and has no lifetime requirements.
-public static native void TF_ImportGraphDefOptionsRemapControlDependency(
-    TF_ImportGraphDefOptions opts, @Cast("const char*") BytePointer src_name, TF_Operation dst);
-public static native void TF_ImportGraphDefOptionsRemapControlDependency(
-    TF_ImportGraphDefOptions opts, String src_name, TF_Operation dst);
-
-// Cause the imported graph to have a control dependency on `oper`. `oper`
-// should exist in the graph being imported into.
-public static native void TF_ImportGraphDefOptionsAddControlDependency(
-    TF_ImportGraphDefOptions opts, TF_Operation oper);
-
-// Add an output in `graph_def` to be returned via the `return_outputs` output
-// parameter of TF_GraphImportGraphDef(). If the output is remapped via an input
-// mapping, the corresponding existing tensor in `graph` will be returned.
-// `oper_name` is copied and has no lifetime requirements.
-public static native void TF_ImportGraphDefOptionsAddReturnOutput(
-    TF_ImportGraphDefOptions opts, @Cast("const char*") BytePointer oper_name, int index);
-public static native void TF_ImportGraphDefOptionsAddReturnOutput(
-    TF_ImportGraphDefOptions opts, String oper_name, int index);
-
-// Returns the number of return outputs added via
-// TF_ImportGraphDefOptionsAddReturnOutput().
-public static native int TF_ImportGraphDefOptionsNumReturnOutputs(
-    @Const TF_ImportGraphDefOptions opts);
-
-// Add an operation in `graph_def` to be returned via the `return_opers` output
-// parameter of TF_GraphImportGraphDef(). `oper_name` is copied and has no
-// lifetime requirements.
-public static native void TF_ImportGraphDefOptionsAddReturnOperation(
-    TF_ImportGraphDefOptions opts, @Cast("const char*") BytePointer oper_name);
-public static native void TF_ImportGraphDefOptionsAddReturnOperation(
-    TF_ImportGraphDefOptions opts, String oper_name);
-
-// Returns the number of return operations added via
-// TF_ImportGraphDefOptionsAddReturnOperation().
-public static native int TF_ImportGraphDefOptionsNumReturnOperations(
-    @Const TF_ImportGraphDefOptions opts);
-// Targeting ../TF_ImportGraphDefResults.java
-
-
-
-// Fetches the return outputs requested via
-// TF_ImportGraphDefOptionsAddReturnOutput(). The number of fetched outputs is
-// returned in `num_outputs`. The array of return outputs is returned in
-// `outputs`. `*outputs` is owned by and has the lifetime of `results`.
-public static native void TF_ImportGraphDefResultsReturnOutputs(
-    TF_ImportGraphDefResults results, IntPointer num_outputs, @Cast("TF_Output**") PointerPointer outputs);
-public static native void TF_ImportGraphDefResultsReturnOutputs(
-    TF_ImportGraphDefResults results, IntPointer num_outputs, @ByPtrPtr TF_Output outputs);
-public static native void TF_ImportGraphDefResultsReturnOutputs(
-    TF_ImportGraphDefResults results, IntBuffer num_outputs, @ByPtrPtr TF_Output outputs);
-public static native void TF_ImportGraphDefResultsReturnOutputs(
-    TF_ImportGraphDefResults results, int[] num_outputs, @ByPtrPtr TF_Output outputs);
-
-// Fetches the return operations requested via
-// TF_ImportGraphDefOptionsAddReturnOperation(). The number of fetched
-// operations is returned in `num_opers`. The array of return operations is
-// returned in `opers`. `*opers` is owned by and has the lifetime of `results`.
-public static native void TF_ImportGraphDefResultsReturnOperations(
-    TF_ImportGraphDefResults results, IntPointer num_opers, @Cast("TF_Operation***") @ByPtrPtr PointerPointer opers);
-public static native void TF_ImportGraphDefResultsReturnOperations(
-    TF_ImportGraphDefResults results, IntBuffer num_opers, @Cast("TF_Operation***") @ByPtrPtr PointerPointer opers);
-public static native void TF_ImportGraphDefResultsReturnOperations(
-    TF_ImportGraphDefResults results, int[] num_opers, @Cast("TF_Operation***") @ByPtrPtr PointerPointer opers);
-
-// Fetches any input mappings requested via
-// TF_ImportGraphDefOptionsAddInputMapping() that didn't appear in the GraphDef
-// and weren't used as input to any node in the imported graph def. The number
-// of fetched mappings is returned in `num_missing_unused_input_mappings`. The
-// array of each mapping's source node name is returned in `src_names`, and the
-// array of each mapping's source index is returned in `src_indexes`.
-//
-// `*src_names`, `*src_indexes`, and the memory backing each string in
-// `src_names` are owned by and have the lifetime of `results`.
-public static native void TF_ImportGraphDefResultsMissingUnusedInputMappings(
-    TF_ImportGraphDefResults results, IntPointer num_missing_unused_input_mappings,
-    @Cast("const char***") @ByPtrPtr PointerPointer src_names, @Cast("int**") PointerPointer src_indexes);
-public static native void TF_ImportGraphDefResultsMissingUnusedInputMappings(
-    TF_ImportGraphDefResults results, IntPointer num_missing_unused_input_mappings,
-    @Cast("const char***") @ByPtrPtr PointerPointer src_names, @ByPtrPtr IntPointer src_indexes);
-public static native void TF_ImportGraphDefResultsMissingUnusedInputMappings(
-    TF_ImportGraphDefResults results, IntBuffer num_missing_unused_input_mappings,
-    @Cast("const char***") @ByPtrPtr PointerPointer src_names, @ByPtrPtr IntBuffer src_indexes);
-public static native void TF_ImportGraphDefResultsMissingUnusedInputMappings(
-    TF_ImportGraphDefResults results, int[] num_missing_unused_input_mappings,
-    @Cast("const char***") @ByPtrPtr PointerPointer src_names, @ByPtrPtr int[] src_indexes);
-
-// Deletes a results object returned by TF_GraphImportGraphDefWithResults().
-public static native void TF_DeleteImportGraphDefResults(
-    TF_ImportGraphDefResults results);
-
-// Import the graph serialized in `graph_def` into `graph`.  Returns nullptr and
-// a bad status on error. Otherwise, returns a populated
-// TF_ImportGraphDefResults instance. The returned instance must be deleted via
-// TF_DeleteImportGraphDefResults().
-public static native TF_ImportGraphDefResults TF_GraphImportGraphDefWithResults(TF_Graph graph, @Const TF_Buffer graph_def,
-                                  @Const TF_ImportGraphDefOptions options,
-                                  TF_Status status);
-
-// Import the graph serialized in `graph_def` into `graph`.
-// Convenience function for when only return outputs are needed.
-//
-// `num_return_outputs` must be the number of return outputs added (i.e. the
-// result of TF_ImportGraphDefOptionsNumReturnOutputs()).  If
-// `num_return_outputs` is non-zero, `return_outputs` must be of length
-// `num_return_outputs`. Otherwise it can be null.
-public static native void TF_GraphImportGraphDefWithReturnOutputs(
-    TF_Graph graph, @Const TF_Buffer graph_def,
-    @Const TF_ImportGraphDefOptions options, TF_Output return_outputs,
-    int num_return_outputs, TF_Status status);
-
-// Import the graph serialized in `graph_def` into `graph`.
-// Convenience function for when no results are needed.
-public static native void TF_GraphImportGraphDef(
-    TF_Graph graph, @Const TF_Buffer graph_def,
-    @Const TF_ImportGraphDefOptions options, TF_Status status);
-
-// Adds a copy of function `func` and optionally its gradient function `grad`
-// to `g`. Once `func`/`grad` is added to `g`, it can be called by creating
-// an operation using the function's name.
-// Any changes to `func`/`grad` (including deleting it) done after this method
-// returns, won't affect the copy of `func`/`grad` in `g`.
-// If `func` or `grad` are already in `g`, TF_GraphCopyFunction has no
-// effect on them, but can establish the function->gradient relationship
-// between them if `func` does not already have a gradient. If `func` already
-// has a gradient different from `grad`, an error is returned.
-//
-// `func` must not be null.
-// If `grad` is null and `func` is not in `g`, `func` is added without a
-// gradient.
-// If `grad` is null and `func` is in `g`, TF_GraphCopyFunction is a noop.
-// `grad` must have appropriate signature as described in the doc of
-// GradientDef in tensorflow/core/framework/function.proto.
-//
-// If successful, status is set to OK and `func` and `grad` are added to `g`.
-// Otherwise, status is set to the encountered error and `g` is unmodified.
-public static native void TF_GraphCopyFunction(TF_Graph g,
-                                                @Const TF_Function func,
-                                                @Const TF_Function grad,
-                                                TF_Status status);
-
-// Returns the number of TF_Functions registered in `g`.
-public static native int TF_GraphNumFunctions(TF_Graph g);
-
-// Fills in `funcs` with the TF_Function* registered in `g`.
-// `funcs` must point to an array of TF_Function* of length at least
-// `max_func`. In usual usage, max_func should be set to the result of
-// TF_GraphNumFunctions(g). In this case, all the functions registered in
-// `g` will be returned. Else, an unspecified subset.
-//
-// If successful, returns the number of TF_Function* successfully set in
-// `funcs` and sets status to OK. The caller takes ownership of
-// all the returned TF_Functions. They must be deleted with TF_DeleteFunction.
-// On error, returns 0, sets status to the encountered error, and the contents
-// of funcs will be undefined.
-public static native int TF_GraphGetFunctions(TF_Graph g, @Cast("TF_Function**") PointerPointer funcs,
-                                               int max_func, TF_Status status);
-public static native int TF_GraphGetFunctions(TF_Graph g, @ByPtrPtr TF_Function funcs,
-                                               int max_func, TF_Status status);
-
-// Note: The following function may fail on very large protos in the future.
-
-public static native void TF_OperationToNodeDef(TF_Operation oper,
-                                                 TF_Buffer output_node_def,
-                                                 TF_Status status);
-// Targeting ../TF_WhileParams.java
-
-
-
-// Creates a TF_WhileParams for creating a while loop in `g`. `inputs` are
-// outputs that already exist in `g` used as initial values for the loop
-// variables.
-//
-// The returned TF_WhileParams will have all fields initialized except
-// `cond_output`, `body_outputs`, and `name`. The `body_outputs` buffer will be
-// allocated to size `ninputs`. The caller should build `cond_graph` and
-// `body_graph` starting from the inputs, and store the final outputs in
-// `cond_output` and `body_outputs`.
-//
-// If `status` is OK, the caller must call either TF_FinishWhile or
-// TF_AbortWhile on the returned TF_WhileParams. If `status` isn't OK, the
-// returned TF_WhileParams is not valid, and the caller should not call
-// TF_FinishWhile() or TF_AbortWhile().
-//
-// Missing functionality (TODO):
-// - Gradients
-// - Reference-type inputs
-// - Directly referencing external tensors from the cond/body graphs (this is
-//   possible in the Python API)
-public static native @ByVal TF_WhileParams TF_NewWhile(TF_Graph g, TF_Output inputs,
-                                                 int ninputs,
-                                                 TF_Status status);
-
-// Builds the while loop specified by `params` and returns the output tensors of
-// the while loop in `outputs`. `outputs` should be allocated to size
-// `params.ninputs`.
-//
-// `params` is no longer valid once this returns.
-//
-// Either this or TF_AbortWhile() must be called after a successful
-// TF_NewWhile() call.
-public static native void TF_FinishWhile(@Const TF_WhileParams params,
-                                          TF_Status status,
-                                          TF_Output outputs);
-
-// Frees `params`s resources without building a while loop. `params` is no
-// longer valid after this returns. Either this or TF_FinishWhile() must be
-// called after a successful TF_NewWhile() call.
-public static native void TF_AbortWhile(@Const TF_WhileParams params);
-
-// Adds operations to compute the partial derivatives of sum of `y`s w.r.t `x`s,
-// i.e., d(y_1 + y_2 + ...)/dx_1, d(y_1 + y_2 + ...)/dx_2...
-//
-// `dx` are used as initial gradients (which represent the symbolic partial
-// derivatives of some loss function `L` w.r.t. `y`).
-// `dx` must be nullptr or have size `ny`.
-// If `dx` is nullptr, the implementation will use dx of `OnesLike` for all
-// shapes in `y`.
-// The partial derivatives are returned in `dy`. `dy` should be allocated to
-// size `nx`.
-//
-// Gradient nodes are automatically named under the "gradients/" prefix. To
-// guarantee name uniqueness, subsequent calls to the same graph will
-// append an incremental tag to the prefix: "gradients_1/", "gradients_2/", ...
-// See TF_AddGradientsWithPrefix, which provides a means to specify a custom
-// name prefix for operations added to a graph to compute the gradients.
-//
-// WARNING: This function does not yet support all the gradients that python
-// supports. See
-// https://www.tensorflow.org/code/tensorflow/cc/gradients/README.md
-// for instructions on how to add C++ more gradients.
-public static native void TF_AddGradients(TF_Graph g, TF_Output y, int ny,
-                                    TF_Output x, int nx, TF_Output dx,
-                                    TF_Status status, TF_Output dy);
-
-// Adds operations to compute the partial derivatives of sum of `y`s w.r.t `x`s,
-// i.e., d(y_1 + y_2 + ...)/dx_1, d(y_1 + y_2 + ...)/dx_2...
-// This is a variant of TF_AddGradients that allows to caller to pass a custom
-// name prefix to the operations added to a graph to compute the gradients.
-//
-// `dx` are used as initial gradients (which represent the symbolic partial
-// derivatives of some loss function `L` w.r.t. `y`).
-// `dx` must be nullptr or have size `ny`.
-// If `dx` is nullptr, the implementation will use dx of `OnesLike` for all
-// shapes in `y`.
-// The partial derivatives are returned in `dy`. `dy` should be allocated to
-// size `nx`.
-// `prefix` names the scope into which all gradients operations are being added.
-// `prefix` must be unique within the provided graph otherwise this operation
-// will fail. If `prefix` is nullptr, the default prefixing behaviour takes
-// place, see TF_AddGradients for more details.
-//
-// WARNING: This function does not yet support all the gradients that python
-// supports. See
-// https://www.tensorflow.org/code/tensorflow/cc/gradients/README.md
-// for instructions on how to add C++ more gradients.
-public static native void TF_AddGradientsWithPrefix(TF_Graph g, @Cast("const char*") BytePointer prefix,
-                                              TF_Output y, int ny,
-                                              TF_Output x, int nx,
-                                              TF_Output dx, TF_Status status,
-                                              TF_Output dy);
-public static native void TF_AddGradientsWithPrefix(TF_Graph g, String prefix,
-                                              TF_Output y, int ny,
-                                              TF_Output x, int nx,
-                                              TF_Output dx, TF_Status status,
-                                              TF_Output dy);
-
-// Create a TF_Function from a TF_Graph
-//
-// Params:
-//  fn_body - the graph whose operations (or subset of whose operations) will be
-//            converted to TF_Function.
-//  fn_name - the name of the new TF_Function. Should match the operation
-//            name (OpDef.name) regexp [A-Z][A-Za-z0-9_.\\-/]*.
-//            If `append_hash_to_fn_name` is false, `fn_name` must be distinct
-//            from other function and operation names (at least those
-//            registered in graphs where this function will be used).
-//  append_hash_to_fn_name - Must be 0 or 1. If set to 1, the actual name
-//                           of the function will be `fn_name` appended with
-//                           '_<hash_of_this_function's_definition>'.
-//                           If set to 0, the function's name will be `fn_name`.
-//  num_opers - `num_opers` contains the number of elements in the `opers` array
-//              or a special value of -1 meaning that no array is given.
-//              The distinction between an empty array of operations and no
-//              array of operations is necessary to distinguish the case of
-//              creating a function with no body (e.g. identity or permutation)
-//              and the case of creating a function whose body contains all
-//              the nodes in the graph (except for the automatic skipping, see
-//              below).
-//  opers - Array of operations to become the body of the function or null.
-//          - If no array is given (`num_opers`  = -1), all the
-//          operations in `fn_body` will become part of the function
-//          except operations referenced in `inputs`. These operations
-//          must have a single output (these operations are typically
-//          placeholders created for the sole purpose of representing
-//          an input. We can relax this constraint if there are
-//          compelling use cases).
-//          - If an array is given (`num_opers` >= 0), all operations
-//          in it will become part of the function. In particular, no
-//          automatic skipping of dummy input operations is performed.
-//  ninputs - number of elements in `inputs` array
-//  inputs - array of TF_Outputs that specify the inputs to the function.
-//           If `ninputs` is zero (the function takes no inputs), `inputs`
-//           can be null. The names used for function inputs are normalized
-//           names of the operations (usually placeholders) pointed to by
-//           `inputs`. These operation names should start with a letter.
-//           Normalization will convert all letters to lowercase and
-//           non-alphanumeric characters to '_' to make resulting names match
-//           the "[a-z][a-z0-9_]*" pattern for operation argument names.
-//           `inputs` cannot contain the same tensor twice.
-//  noutputs - number of elements in `outputs` array
-//  outputs - array of TF_Outputs that specify the outputs of the function.
-//            If `noutputs` is zero (the function returns no outputs), `outputs`
-//            can be null. `outputs` can contain the same tensor more than once.
-//  output_names - The names of the function's outputs. `output_names` array
-//                 must either have the same length as `outputs`
-//                 (i.e. `noutputs`) or be null. In the former case,
-//                 the names should match the regular expression for ArgDef
-//                 names - "[a-z][a-z0-9_]*". In the latter case,
-//                 names for outputs will be generated automatically.
-//  opts - various options for the function, e.g. XLA's inlining control.
-//  description - optional human-readable description of this function.
-//  status - Set to OK on success and an appropriate error on failure.
-//
-// Note that when the same TF_Output is listed as both an input and an output,
-// the corresponding function's output will equal to this input,
-// instead of the original node's output.
-//
-// Callers must also satisfy the following constraints:
-// - `inputs` cannot refer to TF_Outputs within a control flow context. For
-//   example, one cannot use the output of "switch" node as input.
-// - `inputs` and `outputs` cannot have reference types. Reference types are
-//   not exposed through C API and are being replaced with Resources. We support
-//   reference types inside function's body to support legacy code. Do not
-//   use them in new code.
-// - Every node in the function's body must have all of its inputs (including
-//   control inputs). In other words, for every node in the body, each input
-//   must be either listed in `inputs` or must come from another node in
-//   the body. In particular, it is an error to have a control edge going from
-//   a node outside of the body into a node in the body. This applies to control
-//   edges going from nodes referenced in `inputs` to nodes in the body when
-//   the former nodes are not in the body (automatically skipped or not
-//   included in explicitly specified body).
-//
-// Returns:
-//  On success, a newly created TF_Function instance. It must be deleted by
-//  calling TF_DeleteFunction.
-//
-//  On failure, null.
-public static native TF_Function TF_GraphToFunction(
-    @Const TF_Graph fn_body, @Cast("const char*") BytePointer fn_name,
-    @Cast("unsigned char") byte append_hash_to_fn_name, int num_opers,
-    @Cast("const TF_Operation*const*") PointerPointer opers, int ninputs, @Const TF_Output inputs,
-    int noutputs, @Const TF_Output outputs, @Cast("const char*const*") PointerPointer output_names,
-    @Const TF_FunctionOptions opts, @Cast("const char*") BytePointer description, TF_Status status);
-public static native TF_Function TF_GraphToFunction(
-    @Const TF_Graph fn_body, @Cast("const char*") BytePointer fn_name,
-    @Cast("unsigned char") byte append_hash_to_fn_name, int num_opers,
-    @Const @ByPtrPtr TF_Operation opers, int ninputs, @Const TF_Output inputs,
-    int noutputs, @Const TF_Output outputs, @Cast("const char*const*") @ByPtrPtr BytePointer output_names,
-    @Const TF_FunctionOptions opts, @Cast("const char*") BytePointer description, TF_Status status);
-public static native TF_Function TF_GraphToFunction(
-    @Const TF_Graph fn_body, String fn_name,
-    @Cast("unsigned char") byte append_hash_to_fn_name, int num_opers,
-    @Const @ByPtrPtr TF_Operation opers, int ninputs, @Const TF_Output inputs,
-    int noutputs, @Const TF_Output outputs, @Cast("const char*const*") @ByPtrPtr ByteBuffer output_names,
-    @Const TF_FunctionOptions opts, String description, TF_Status status);
-public static native TF_Function TF_GraphToFunction(
-    @Const TF_Graph fn_body, @Cast("const char*") BytePointer fn_name,
-    @Cast("unsigned char") byte append_hash_to_fn_name, int num_opers,
-    @Const @ByPtrPtr TF_Operation opers, int ninputs, @Const TF_Output inputs,
-    int noutputs, @Const TF_Output outputs, @Cast("const char*const*") @ByPtrPtr byte[] output_names,
-    @Const TF_FunctionOptions opts, @Cast("const char*") BytePointer description, TF_Status status);
-public static native TF_Function TF_GraphToFunction(
-    @Const TF_Graph fn_body, String fn_name,
-    @Cast("unsigned char") byte append_hash_to_fn_name, int num_opers,
-    @Const @ByPtrPtr TF_Operation opers, int ninputs, @Const TF_Output inputs,
-    int noutputs, @Const TF_Output outputs, @Cast("const char*const*") @ByPtrPtr BytePointer output_names,
-    @Const TF_FunctionOptions opts, String description, TF_Status status);
-public static native TF_Function TF_GraphToFunction(
-    @Const TF_Graph fn_body, @Cast("const char*") BytePointer fn_name,
-    @Cast("unsigned char") byte append_hash_to_fn_name, int num_opers,
-    @Const @ByPtrPtr TF_Operation opers, int ninputs, @Const TF_Output inputs,
-    int noutputs, @Const TF_Output outputs, @Cast("const char*const*") @ByPtrPtr ByteBuffer output_names,
-    @Const TF_FunctionOptions opts, @Cast("const char*") BytePointer description, TF_Status status);
-public static native TF_Function TF_GraphToFunction(
-    @Const TF_Graph fn_body, String fn_name,
-    @Cast("unsigned char") byte append_hash_to_fn_name, int num_opers,
-    @Const @ByPtrPtr TF_Operation opers, int ninputs, @Const TF_Output inputs,
-    int noutputs, @Const TF_Output outputs, @Cast("const char*const*") @ByPtrPtr byte[] output_names,
-    @Const TF_FunctionOptions opts, String description, TF_Status status);
-
-// Similar to TF_GraphToFunction but allows specifying control outputs of the
-// function.
-//
-//  The arguments of TF_GraphToFunction have the same meaning, but the new
-//  arguments are as follows:
-//
-//    ncontrol_outputs: Number of control outputs of the function.
-//    control_outputs: vector of TF_Operation objects to be marked as control
-//      outputs of the function. Operations marked as control outputs are
-//      guaranteed to execute.
-//    control_output_names: Optional. If not nullptr, vector of strings, one
-//      per control output, with their names to be added to the function's
-//      OpDef.
-public static native TF_Function TF_GraphToFunctionWithControlOutputs(
-    @Const TF_Graph fn_body, @Cast("const char*") BytePointer fn_name,
-    @Cast("unsigned char") byte append_hash_to_fn_name, int num_opers,
-    @Cast("const TF_Operation*const*") PointerPointer opers, int ninputs, @Const TF_Output inputs,
-    int noutputs, @Const TF_Output outputs, @Cast("const char*const*") PointerPointer output_names,
-    int ncontrol_outputs, @Cast("const TF_Operation*const*") PointerPointer control_outputs,
-    @Cast("const char*const*") PointerPointer control_output_names, @Const TF_FunctionOptions opts,
-    @Cast("const char*") BytePointer description, TF_Status status);
-public static native TF_Function TF_GraphToFunctionWithControlOutputs(
-    @Const TF_Graph fn_body, @Cast("const char*") BytePointer fn_name,
-    @Cast("unsigned char") byte append_hash_to_fn_name, int num_opers,
-    @Const @ByPtrPtr TF_Operation opers, int ninputs, @Const TF_Output inputs,
-    int noutputs, @Const TF_Output outputs, @Cast("const char*const*") @ByPtrPtr BytePointer output_names,
-    int ncontrol_outputs, @Const @ByPtrPtr TF_Operation control_outputs,
-    @Cast("const char*const*") @ByPtrPtr BytePointer control_output_names, @Const TF_FunctionOptions opts,
-    @Cast("const char*") BytePointer description, TF_Status status);
-public static native TF_Function TF_GraphToFunctionWithControlOutputs(
-    @Const TF_Graph fn_body, String fn_name,
-    @Cast("unsigned char") byte append_hash_to_fn_name, int num_opers,
-    @Const @ByPtrPtr TF_Operation opers, int ninputs, @Const TF_Output inputs,
-    int noutputs, @Const TF_Output outputs, @Cast("const char*const*") @ByPtrPtr ByteBuffer output_names,
-    int ncontrol_outputs, @Const @ByPtrPtr TF_Operation control_outputs,
-    @Cast("const char*const*") @ByPtrPtr ByteBuffer control_output_names, @Const TF_FunctionOptions opts,
-    String description, TF_Status status);
-public static native TF_Function TF_GraphToFunctionWithControlOutputs(
-    @Const TF_Graph fn_body, @Cast("const char*") BytePointer fn_name,
-    @Cast("unsigned char") byte append_hash_to_fn_name, int num_opers,
-    @Const @ByPtrPtr TF_Operation opers, int ninputs, @Const TF_Output inputs,
-    int noutputs, @Const TF_Output outputs, @Cast("const char*const*") @ByPtrPtr byte[] output_names,
-    int ncontrol_outputs, @Const @ByPtrPtr TF_Operation control_outputs,
-    @Cast("const char*const*") @ByPtrPtr byte[] control_output_names, @Const TF_FunctionOptions opts,
-    @Cast("const char*") BytePointer description, TF_Status status);
-public static native TF_Function TF_GraphToFunctionWithControlOutputs(
-    @Const TF_Graph fn_body, String fn_name,
-    @Cast("unsigned char") byte append_hash_to_fn_name, int num_opers,
-    @Const @ByPtrPtr TF_Operation opers, int ninputs, @Const TF_Output inputs,
-    int noutputs, @Const TF_Output outputs, @Cast("const char*const*") @ByPtrPtr BytePointer output_names,
-    int ncontrol_outputs, @Const @ByPtrPtr TF_Operation control_outputs,
-    @Cast("const char*const*") @ByPtrPtr BytePointer control_output_names, @Const TF_FunctionOptions opts,
-    String description, TF_Status status);
-public static native TF_Function TF_GraphToFunctionWithControlOutputs(
-    @Const TF_Graph fn_body, @Cast("const char*") BytePointer fn_name,
-    @Cast("unsigned char") byte append_hash_to_fn_name, int num_opers,
-    @Const @ByPtrPtr TF_Operation opers, int ninputs, @Const TF_Output inputs,
-    int noutputs, @Const TF_Output outputs, @Cast("const char*const*") @ByPtrPtr ByteBuffer output_names,
-    int ncontrol_outputs, @Const @ByPtrPtr TF_Operation control_outputs,
-    @Cast("const char*const*") @ByPtrPtr ByteBuffer control_output_names, @Const TF_FunctionOptions opts,
-    @Cast("const char*") BytePointer description, TF_Status status);
-public static native TF_Function TF_GraphToFunctionWithControlOutputs(
-    @Const TF_Graph fn_body, String fn_name,
-    @Cast("unsigned char") byte append_hash_to_fn_name, int num_opers,
-    @Const @ByPtrPtr TF_Operation opers, int ninputs, @Const TF_Output inputs,
-    int noutputs, @Const TF_Output outputs, @Cast("const char*const*") @ByPtrPtr byte[] output_names,
-    int ncontrol_outputs, @Const @ByPtrPtr TF_Operation control_outputs,
-    @Cast("const char*const*") @ByPtrPtr byte[] control_output_names, @Const TF_FunctionOptions opts,
-    String description, TF_Status status);
-
-// Returns the name of the graph function.
-// The return value points to memory that is only usable until the next
-// mutation to *func.
-public static native @Cast("const char*") BytePointer TF_FunctionName(TF_Function func);
-
-// Write out a serialized representation of `func` (as a FunctionDef protocol
-// message) to `output_func_def` (allocated by TF_NewBuffer()).
-// `output_func_def`'s underlying buffer will be freed when TF_DeleteBuffer()
-// is called.
-//
-// May fail on very large graphs in the future.
-public static native void TF_FunctionToFunctionDef(TF_Function func,
-                                                    TF_Buffer output_func_def,
-                                                    TF_Status status);
-
-// Construct and return the function whose FunctionDef representation is
-// serialized in `proto`. `proto_len` must equal the number of bytes
-// pointed to by `proto`.
-// Returns:
-//  On success, a newly created TF_Function instance. It must be deleted by
-//  calling TF_DeleteFunction.
-//
-//  On failure, null.
-public static native TF_Function TF_FunctionImportFunctionDef(
-    @Const Pointer proto, @Cast("size_t") long proto_len, TF_Status status);
-
-// Sets function attribute named `attr_name` to value stored in `proto`.
-// If this attribute is already set to another value, it is overridden.
-// `proto` should point to a sequence of bytes of length `proto_len`
-// representing a binary serialization of an AttrValue protocol
-// buffer.
-public static native void TF_FunctionSetAttrValueProto(TF_Function func,
-                                                        @Cast("const char*") BytePointer attr_name,
-                                                        @Const Pointer proto,
-                                                        @Cast("size_t") long proto_len,
-                                                        TF_Status status);
-public static native void TF_FunctionSetAttrValueProto(TF_Function func,
-                                                        String attr_name,
-                                                        @Const Pointer proto,
-                                                        @Cast("size_t") long proto_len,
-                                                        TF_Status status);
-
-// Sets `output_attr_value` to the binary-serialized AttrValue proto
-// representation of the value of the `attr_name` attr of `func`.
-// If `attr_name` attribute is not present, status is set to an error.
-public static native void TF_FunctionGetAttrValueProto(
-    TF_Function func, @Cast("const char*") BytePointer attr_name, TF_Buffer output_attr_value,
-    TF_Status status);
-public static native void TF_FunctionGetAttrValueProto(
-    TF_Function func, String attr_name, TF_Buffer output_attr_value,
-    TF_Status status);
-
-// Frees the memory used by the `func` struct.
-// TF_DeleteFunction is a noop if `func` is null.
-// Deleting a function does not remove it from any graphs it was copied to.
-public static native void TF_DeleteFunction(TF_Function func);
-
-// Attempts to evaluate `output`. This will only be possible if `output` doesn't
-// depend on any graph inputs (this function is safe to call if this isn't the
-// case though).
-//
-// If the evaluation is successful, this function returns true and `output`s
-// value is returned in `result`. Otherwise returns false. An error status is
-// returned if something is wrong with the graph or input. Note that this may
-// return false even if no error status is set.
-public static native @Cast("unsigned char") byte TF_TryEvaluateConstant(TF_Graph graph,
-                                                           @ByVal TF_Output output,
-                                                           @Cast("TF_Tensor**") PointerPointer result,
-                                                           TF_Status status);
-public static native @Cast("unsigned char") byte TF_TryEvaluateConstant(TF_Graph graph,
-                                                           @ByVal TF_Output output,
-                                                           @ByPtrPtr TF_Tensor result,
-                                                           TF_Status status);
-// Targeting ../TF_Session.java
-
-
-
-// Return a new execution session with the associated graph, or NULL on
-// error. Does not take ownership of any input parameters.
-//
-// *`graph` must be a valid graph (not deleted or nullptr). `graph` will be be
-// kept alive for the lifetime of the returned TF_Session. New nodes can still
-// be added to `graph` after this call.
-public static native TF_Session TF_NewSession(TF_Graph graph,
-                                                @Const TF_SessionOptions opts,
-                                                TF_Status status);
-
-// This function creates a new TF_Session (which is created on success) using
-// `session_options`, and then initializes state (restoring tensors and other
-// assets) using `run_options`.
-//
-// Any NULL and non-NULL value combinations for (`run_options, `meta_graph_def`)
-// are valid.
-//
-// - `export_dir` must be set to the path of the exported SavedModel.
-// - `tags` must include the set of tags used to identify one MetaGraphDef in
-//    the SavedModel.
-// - `graph` must be a graph newly allocated with TF_NewGraph().
-//
-// If successful, populates `graph` with the contents of the Graph and
-// `meta_graph_def` with the MetaGraphDef of the loaded model.
-public static native TF_Session TF_LoadSessionFromSavedModel(
-    @Const TF_SessionOptions session_options, @Const TF_Buffer run_options,
-    @Cast("const char*") BytePointer export_dir, @Cast("const char*const*") PointerPointer tags, int tags_len,
-    TF_Graph graph, TF_Buffer meta_graph_def, TF_Status status);
-public static native TF_Session TF_LoadSessionFromSavedModel(
-    @Const TF_SessionOptions session_options, @Const TF_Buffer run_options,
-    @Cast("const char*") BytePointer export_dir, @Cast("const char*const*") @ByPtrPtr BytePointer tags, int tags_len,
-    TF_Graph graph, TF_Buffer meta_graph_def, TF_Status status);
-public static native TF_Session TF_LoadSessionFromSavedModel(
-    @Const TF_SessionOptions session_options, @Const TF_Buffer run_options,
-    String export_dir, @Cast("const char*const*") @ByPtrPtr ByteBuffer tags, int tags_len,
-    TF_Graph graph, TF_Buffer meta_graph_def, TF_Status status);
-public static native TF_Session TF_LoadSessionFromSavedModel(
-    @Const TF_SessionOptions session_options, @Const TF_Buffer run_options,
-    @Cast("const char*") BytePointer export_dir, @Cast("const char*const*") @ByPtrPtr byte[] tags, int tags_len,
-    TF_Graph graph, TF_Buffer meta_graph_def, TF_Status status);
-public static native TF_Session TF_LoadSessionFromSavedModel(
-    @Const TF_SessionOptions session_options, @Const TF_Buffer run_options,
-    String export_dir, @Cast("const char*const*") @ByPtrPtr BytePointer tags, int tags_len,
-    TF_Graph graph, TF_Buffer meta_graph_def, TF_Status status);
-public static native TF_Session TF_LoadSessionFromSavedModel(
-    @Const TF_SessionOptions session_options, @Const TF_Buffer run_options,
-    @Cast("const char*") BytePointer export_dir, @Cast("const char*const*") @ByPtrPtr ByteBuffer tags, int tags_len,
-    TF_Graph graph, TF_Buffer meta_graph_def, TF_Status status);
-public static native TF_Session TF_LoadSessionFromSavedModel(
-    @Const TF_SessionOptions session_options, @Const TF_Buffer run_options,
-    String export_dir, @Cast("const char*const*") @ByPtrPtr byte[] tags, int tags_len,
-    TF_Graph graph, TF_Buffer meta_graph_def, TF_Status status);
-
-// Close a session.
-//
-// Contacts any other processes associated with the session, if applicable.
-// May not be called after TF_DeleteSession().
-public static native void TF_CloseSession(TF_Session arg0, TF_Status status);
-
-// Destroy a session object.
-//
-// Even if error information is recorded in *status, this call discards all
-// local resources associated with the session.  The session may not be used
-// during or after this call (and the session drops its reference to the
-// corresponding graph).
-public static native void TF_DeleteSession(TF_Session arg0, TF_Status status);
-
-// Run the graph associated with the session starting with the supplied inputs
-// (inputs[0,ninputs-1] with corresponding values in input_values[0,ninputs-1]).
-//
-// Any NULL and non-NULL value combinations for (`run_options`,
-// `run_metadata`) are valid.
-//
-//    - `run_options` may be NULL, in which case it will be ignored; or
-//      non-NULL, in which case it must point to a `TF_Buffer` containing the
-//      serialized representation of a `RunOptions` protocol buffer.
-//    - `run_metadata` may be NULL, in which case it will be ignored; or
-//      non-NULL, in which case it must point to an empty, freshly allocated
-//      `TF_Buffer` that may be updated to contain the serialized representation
-//      of a `RunMetadata` protocol buffer.
-//
-// The caller retains ownership of `input_values` (which can be deleted using
-// TF_DeleteTensor). The caller also retains ownership of `run_options` and/or
-// `run_metadata` (when not NULL) and should manually call TF_DeleteBuffer on
-// them.
-//
-// On success, the tensors corresponding to outputs[0,noutputs-1] are placed in
-// output_values[]. Ownership of the elements of output_values[] is transferred
-// to the caller, which must eventually call TF_DeleteTensor on them.
-//
-// On failure, output_values[] contains NULLs.
-public static native void TF_SessionRun(
-    TF_Session session,
-    @Const TF_Buffer run_options,
-    @Const TF_Output inputs, @Cast("TF_Tensor*const*") PointerPointer input_values, int ninputs,
-    @Const TF_Output outputs, @Cast("TF_Tensor**") PointerPointer output_values, int noutputs,
-    @Cast("const TF_Operation*const*") PointerPointer target_opers, int ntargets,
-    TF_Buffer run_metadata,
-    TF_Status arg11);
-public static native void TF_SessionRun(
-    TF_Session session,
-    @Const TF_Buffer run_options,
-    @Const TF_Output inputs, @ByPtrPtr TF_Tensor input_values, int ninputs,
-    @Const TF_Output outputs, @ByPtrPtr TF_Tensor output_values, int noutputs,
-    @Const @ByPtrPtr TF_Operation target_opers, int ntargets,
-    TF_Buffer run_metadata,
-    TF_Status arg11);
-
-// Set up the graph with the intended feeds (inputs) and fetches (outputs) for a
-// sequence of partial run calls.
-//
-// On success, returns a handle that is used for subsequent PRun calls. The
-// handle should be deleted with TF_DeletePRunHandle when it is no longer
-// needed.
-//
-// On failure, out_status contains a tensorflow::Status with an error
-// message. *handle is set to nullptr.
-public static native void TF_SessionPRunSetup(
-    TF_Session arg0,
-    @Const TF_Output inputs, int ninputs,
-    @Const TF_Output outputs, int noutputs,
-    @Cast("const TF_Operation*const*") PointerPointer target_opers, int ntargets,
-    @Cast("const char**") PointerPointer handle,
-    TF_Status arg8);
-public static native void TF_SessionPRunSetup(
-    TF_Session arg0,
-    @Const TF_Output inputs, int ninputs,
-    @Const TF_Output outputs, int noutputs,
-    @Const @ByPtrPtr TF_Operation target_opers, int ntargets,
-    @Cast("const char**") @ByPtrPtr BytePointer handle,
-    TF_Status arg8);
-public static native void TF_SessionPRunSetup(
-    TF_Session arg0,
-    @Const TF_Output inputs, int ninputs,
-    @Const TF_Output outputs, int noutputs,
-    @Const @ByPtrPtr TF_Operation target_opers, int ntargets,
-    @Cast("const char**") @ByPtrPtr ByteBuffer handle,
-    TF_Status arg8);
-public static native void TF_SessionPRunSetup(
-    TF_Session arg0,
-    @Const TF_Output inputs, int ninputs,
-    @Const TF_Output outputs, int noutputs,
-    @Const @ByPtrPtr TF_Operation target_opers, int ntargets,
-    @Cast("const char**") @ByPtrPtr byte[] handle,
-    TF_Status arg8);
-
-// Continue to run the graph with additional feeds and fetches. The
-// execution state is uniquely identified by the handle.
-public static native void TF_SessionPRun(
-    TF_Session arg0, @Cast("const char*") BytePointer handle,
-    @Const TF_Output inputs, @Cast("TF_Tensor*const*") PointerPointer input_values, int ninputs,
-    @Const TF_Output outputs, @Cast("TF_Tensor**") PointerPointer output_values, int noutputs,
-    @Cast("const TF_Operation*const*") PointerPointer target_opers, int ntargets,
-    TF_Status arg10);
-public static native void TF_SessionPRun(
-    TF_Session arg0, @Cast("const char*") BytePointer handle,
-    @Const TF_Output inputs, @ByPtrPtr TF_Tensor input_values, int ninputs,
-    @Const TF_Output outputs, @ByPtrPtr TF_Tensor output_values, int noutputs,
-    @Const @ByPtrPtr TF_Operation target_opers, int ntargets,
-    TF_Status arg10);
-public static native void TF_SessionPRun(
-    TF_Session arg0, String handle,
-    @Const TF_Output inputs, @ByPtrPtr TF_Tensor input_values, int ninputs,
-    @Const TF_Output outputs, @ByPtrPtr TF_Tensor output_values, int noutputs,
-    @Const @ByPtrPtr TF_Operation target_opers, int ntargets,
-    TF_Status arg10);
-
-// Deletes a handle allocated by TF_SessionPRunSetup.
-// Once called, no more calls to TF_SessionPRun should be made.
-public static native void TF_DeletePRunHandle(@Cast("const char*") BytePointer handle);
-public static native void TF_DeletePRunHandle(String handle);
-// Targeting ../TF_DeprecatedSession.java
-
-
-
-public static native TF_DeprecatedSession TF_NewDeprecatedSession(
-    @Const TF_SessionOptions arg0, TF_Status status);
-public static native void TF_CloseDeprecatedSession(TF_DeprecatedSession arg0,
-                                                     TF_Status status);
-public static native void TF_DeleteDeprecatedSession(TF_DeprecatedSession arg0,
-                                                      TF_Status status);
-public static native void TF_Reset(@Const TF_SessionOptions opt,
-                                    @Cast("const char**") PointerPointer containers, int ncontainers,
-                                    TF_Status status);
-public static native void TF_Reset(@Const TF_SessionOptions opt,
-                                    @Cast("const char**") @ByPtrPtr BytePointer containers, int ncontainers,
-                                    TF_Status status);
-public static native void TF_Reset(@Const TF_SessionOptions opt,
-                                    @Cast("const char**") @ByPtrPtr ByteBuffer containers, int ncontainers,
-                                    TF_Status status);
-public static native void TF_Reset(@Const TF_SessionOptions opt,
-                                    @Cast("const char**") @ByPtrPtr byte[] containers, int ncontainers,
-                                    TF_Status status);
-// Treat the bytes proto[0,proto_len-1] as a serialized GraphDef and
-// add the nodes in that GraphDef to the graph for the session.
-//
-// Prefer use of TF_Session and TF_GraphImportGraphDef over this.
-public static native void TF_ExtendGraph(TF_DeprecatedSession arg0,
-                                          @Const Pointer proto, @Cast("size_t") long proto_len,
-                                          TF_Status arg3);
-
-// See TF_SessionRun() above.
-public static native void TF_Run(TF_DeprecatedSession arg0,
-                                  @Const TF_Buffer run_options,
-                                  @Cast("const char**") PointerPointer input_names, @Cast("TF_Tensor**") PointerPointer inputs,
-                                  int ninputs, @Cast("const char**") PointerPointer output_names,
-                                  @Cast("TF_Tensor**") PointerPointer outputs, int noutputs,
-                                  @Cast("const char**") PointerPointer target_oper_names, int ntargets,
-                                  TF_Buffer run_metadata, TF_Status arg11);
-public static native void TF_Run(TF_DeprecatedSession arg0,
-                                  @Const TF_Buffer run_options,
-                                  @Cast("const char**") @ByPtrPtr BytePointer input_names, @ByPtrPtr TF_Tensor inputs,
-                                  int ninputs, @Cast("const char**") @ByPtrPtr BytePointer output_names,
-                                  @ByPtrPtr TF_Tensor outputs, int noutputs,
-                                  @Cast("const char**") @ByPtrPtr BytePointer target_oper_names, int ntargets,
-                                  TF_Buffer run_metadata, TF_Status arg11);
-public static native void TF_Run(TF_DeprecatedSession arg0,
-                                  @Const TF_Buffer run_options,
-                                  @Cast("const char**") @ByPtrPtr ByteBuffer input_names, @ByPtrPtr TF_Tensor inputs,
-                                  int ninputs, @Cast("const char**") @ByPtrPtr ByteBuffer output_names,
-                                  @ByPtrPtr TF_Tensor outputs, int noutputs,
-                                  @Cast("const char**") @ByPtrPtr ByteBuffer target_oper_names, int ntargets,
-                                  TF_Buffer run_metadata, TF_Status arg11);
-public static native void TF_Run(TF_DeprecatedSession arg0,
-                                  @Const TF_Buffer run_options,
-                                  @Cast("const char**") @ByPtrPtr byte[] input_names, @ByPtrPtr TF_Tensor inputs,
-                                  int ninputs, @Cast("const char**") @ByPtrPtr byte[] output_names,
-                                  @ByPtrPtr TF_Tensor outputs, int noutputs,
-                                  @Cast("const char**") @ByPtrPtr byte[] target_oper_names, int ntargets,
-                                  TF_Buffer run_metadata, TF_Status arg11);
-
-// See TF_SessionPRunSetup() above.
-public static native void TF_PRunSetup(TF_DeprecatedSession arg0,
-                                        @Cast("const char**") PointerPointer input_names, int ninputs,
-                                        @Cast("const char**") PointerPointer output_names, int noutputs,
-                                        @Cast("const char**") PointerPointer target_oper_names,
-                                        int ntargets, @Cast("const char**") PointerPointer handle,
-                                        TF_Status arg8);
-public static native void TF_PRunSetup(TF_DeprecatedSession arg0,
-                                        @Cast("const char**") @ByPtrPtr BytePointer input_names, int ninputs,
-                                        @Cast("const char**") @ByPtrPtr BytePointer output_names, int noutputs,
-                                        @Cast("const char**") @ByPtrPtr BytePointer target_oper_names,
-                                        int ntargets, @Cast("const char**") @ByPtrPtr BytePointer handle,
-                                        TF_Status arg8);
-public static native void TF_PRunSetup(TF_DeprecatedSession arg0,
-                                        @Cast("const char**") @ByPtrPtr ByteBuffer input_names, int ninputs,
-                                        @Cast("const char**") @ByPtrPtr ByteBuffer output_names, int noutputs,
-                                        @Cast("const char**") @ByPtrPtr ByteBuffer target_oper_names,
-                                        int ntargets, @Cast("const char**") @ByPtrPtr ByteBuffer handle,
-                                        TF_Status arg8);
-public static native void TF_PRunSetup(TF_DeprecatedSession arg0,
-                                        @Cast("const char**") @ByPtrPtr byte[] input_names, int ninputs,
-                                        @Cast("const char**") @ByPtrPtr byte[] output_names, int noutputs,
-                                        @Cast("const char**") @ByPtrPtr byte[] target_oper_names,
-                                        int ntargets, @Cast("const char**") @ByPtrPtr byte[] handle,
-                                        TF_Status arg8);
-
-// See TF_SessionPRun above.
-public static native void TF_PRun(TF_DeprecatedSession arg0, @Cast("const char*") BytePointer handle,
-                                   @Cast("const char**") PointerPointer input_names, @Cast("TF_Tensor**") PointerPointer inputs,
-                                   int ninputs, @Cast("const char**") PointerPointer output_names,
-                                   @Cast("TF_Tensor**") PointerPointer outputs, int noutputs,
-                                   @Cast("const char**") PointerPointer target_oper_names, int ntargets,
-                                   TF_Status arg10);
-public static native void TF_PRun(TF_DeprecatedSession arg0, @Cast("const char*") BytePointer handle,
-                                   @Cast("const char**") @ByPtrPtr BytePointer input_names, @ByPtrPtr TF_Tensor inputs,
-                                   int ninputs, @Cast("const char**") @ByPtrPtr BytePointer output_names,
-                                   @ByPtrPtr TF_Tensor outputs, int noutputs,
-                                   @Cast("const char**") @ByPtrPtr BytePointer target_oper_names, int ntargets,
-                                   TF_Status arg10);
-public static native void TF_PRun(TF_DeprecatedSession arg0, String handle,
-                                   @Cast("const char**") @ByPtrPtr ByteBuffer input_names, @ByPtrPtr TF_Tensor inputs,
-                                   int ninputs, @Cast("const char**") @ByPtrPtr ByteBuffer output_names,
-                                   @ByPtrPtr TF_Tensor outputs, int noutputs,
-                                   @Cast("const char**") @ByPtrPtr ByteBuffer target_oper_names, int ntargets,
-                                   TF_Status arg10);
-public static native void TF_PRun(TF_DeprecatedSession arg0, @Cast("const char*") BytePointer handle,
-                                   @Cast("const char**") @ByPtrPtr byte[] input_names, @ByPtrPtr TF_Tensor inputs,
-                                   int ninputs, @Cast("const char**") @ByPtrPtr byte[] output_names,
-                                   @ByPtrPtr TF_Tensor outputs, int noutputs,
-                                   @Cast("const char**") @ByPtrPtr byte[] target_oper_names, int ntargets,
-                                   TF_Status arg10);
-public static native void TF_PRun(TF_DeprecatedSession arg0, String handle,
-                                   @Cast("const char**") @ByPtrPtr BytePointer input_names, @ByPtrPtr TF_Tensor inputs,
-                                   int ninputs, @Cast("const char**") @ByPtrPtr BytePointer output_names,
-                                   @ByPtrPtr TF_Tensor outputs, int noutputs,
-                                   @Cast("const char**") @ByPtrPtr BytePointer target_oper_names, int ntargets,
-                                   TF_Status arg10);
-public static native void TF_PRun(TF_DeprecatedSession arg0, @Cast("const char*") BytePointer handle,
-                                   @Cast("const char**") @ByPtrPtr ByteBuffer input_names, @ByPtrPtr TF_Tensor inputs,
-                                   int ninputs, @Cast("const char**") @ByPtrPtr ByteBuffer output_names,
-                                   @ByPtrPtr TF_Tensor outputs, int noutputs,
-                                   @Cast("const char**") @ByPtrPtr ByteBuffer target_oper_names, int ntargets,
-                                   TF_Status arg10);
-public static native void TF_PRun(TF_DeprecatedSession arg0, String handle,
-                                   @Cast("const char**") @ByPtrPtr byte[] input_names, @ByPtrPtr TF_Tensor inputs,
-                                   int ninputs, @Cast("const char**") @ByPtrPtr byte[] output_names,
-                                   @ByPtrPtr TF_Tensor outputs, int noutputs,
-                                   @Cast("const char**") @ByPtrPtr byte[] target_oper_names, int ntargets,
-                                   TF_Status arg10);
-// Targeting ../TF_DeviceList.java
-
-
-
-// Lists all devices in a TF_Session.
-//
-// Caller takes ownership of the returned TF_DeviceList* which must eventually
-// be freed with a call to TF_DeleteDeviceList.
-public static native TF_DeviceList TF_SessionListDevices(TF_Session session,
-                                                           TF_Status status);
-
-// Lists all devices in a TF_Session.
-//
-// Caller takes ownership of the returned TF_DeviceList* which must eventually
-// be freed with a call to TF_DeleteDeviceList.
-public static native TF_DeviceList TF_DeprecatedSessionListDevices(
-    TF_DeprecatedSession session, TF_Status status);
-
-// Deallocates the device list.
-public static native void TF_DeleteDeviceList(TF_DeviceList list);
-
-// Counts the number of elements in the device list.
-public static native int TF_DeviceListCount(@Const TF_DeviceList list);
-
-// Retrieves the full name of the device (e.g. /job:worker/replica:0/...)
-// The return value will be a pointer to a null terminated string. The caller
-// must not modify or delete the string. It will be deallocated upon a call to
-// TF_DeleteDeviceList.
-//
-// If index is out of bounds, an error code will be set in the status object,
-// and a null pointer will be returned.
-public static native @Cast("const char*") BytePointer TF_DeviceListName(@Const TF_DeviceList list,
-                                                    int index,
-                                                    TF_Status status);
-
-// Retrieves the type of the device at the given index.
-//
-// The caller must not modify or delete the string. It will be deallocated upon
-// a call to TF_DeleteDeviceList.
-//
-// If index is out of bounds, an error code will be set in the status object,
-// and a null pointer will be returned.
-public static native @Cast("const char*") BytePointer TF_DeviceListType(@Const TF_DeviceList list,
-                                                    int index,
-                                                    TF_Status status);
-
-// Retrieve the amount of memory associated with a given device.
-//
-// If index is out of bounds, an error code will be set in the status object,
-// and -1 will be returned.
-public static native @Cast("int64_t") long TF_DeviceListMemoryBytes(
-    @Const TF_DeviceList list, int index, TF_Status status);
-
-// Retrieve the incarnation number of a given device.
-//
-// If index is out of bounds, an error code will be set in the status object,
-// and 0 will be returned.
-public static native @Cast("uint64_t") long TF_DeviceListIncarnation(
-    @Const TF_DeviceList list, int index, TF_Status status);
-// Targeting ../TF_Library.java
-
-
-
-// Load the library specified by library_filename and register the ops and
-// kernels present in that library.
-//
-// Pass "library_filename" to a platform-specific mechanism for dynamically
-// loading a library. The rules for determining the exact location of the
-// library are platform-specific and are not documented here.
-//
-// On success, place OK in status and return the newly created library handle.
-// The caller owns the library handle.
-//
-// On failure, place an error status in status and return NULL.
-public static native TF_Library TF_LoadLibrary(@Cast("const char*") BytePointer library_filename,
-                                                 TF_Status status);
-public static native TF_Library TF_LoadLibrary(String library_filename,
-                                                 TF_Status status);
-
-// Get the OpList of OpDefs defined in the library pointed by lib_handle.
-//
-// Returns a TF_Buffer. The memory pointed to by the result is owned by
-// lib_handle. The data in the buffer will be the serialized OpList proto for
-// ops defined in the library.
-public static native @ByVal TF_Buffer TF_GetOpList(TF_Library lib_handle);
-
-// Frees the memory associated with the library handle.
-// Does NOT unload the library.
-public static native void TF_DeleteLibraryHandle(TF_Library lib_handle);
-
-// Get the OpList of all OpDefs defined in this address space.
-// Returns a TF_Buffer, ownership of which is transferred to the caller
-// (and can be freed using TF_DeleteBuffer).
-//
-// The data in the buffer will be the serialized OpList proto for ops registered
-// in this address space.
-public static native TF_Buffer TF_GetAllOpList();
-// Targeting ../TF_ApiDefMap.java
-
-
-
-// Creates a new TF_ApiDefMap instance.
-//
-// Params:
-//  op_list_buffer - TF_Buffer instance containing serialized OpList
-//    protocol buffer. (See
-//    https://www.tensorflow.org/code/tensorflow/core/framework/op_def.proto
-//    for the OpList proto definition).
-//  status - Set to OK on success and an appropriate error on failure.
-public static native TF_ApiDefMap TF_NewApiDefMap(TF_Buffer op_list_buffer,
-                                                    TF_Status status);
-
-// Deallocates a TF_ApiDefMap.
-public static native void TF_DeleteApiDefMap(TF_ApiDefMap apimap);
-
-// Add ApiDefs to the map.
-//
-// `text` corresponds to a text representation of an ApiDefs protocol message.
-// (https://www.tensorflow.org/code/tensorflow/core/framework/api_def.proto).
-//
-// The provided ApiDefs will be merged with existing ones in the map, with
-// precedence given to the newly added version in case of conflicts with
-// previous calls to TF_ApiDefMapPut.
-public static native void TF_ApiDefMapPut(TF_ApiDefMap api_def_map,
-                                           @Cast("const char*") BytePointer text, @Cast("size_t") long text_len,
-                                           TF_Status status);
-public static native void TF_ApiDefMapPut(TF_ApiDefMap api_def_map,
-                                           String text, @Cast("size_t") long text_len,
-                                           TF_Status status);
-
-// Returns a serialized ApiDef protocol buffer for the TensorFlow operation
-// named `name`.
-public static native TF_Buffer TF_ApiDefMapGet(TF_ApiDefMap api_def_map,
-                                                 @Cast("const char*") BytePointer name,
-                                                 @Cast("size_t") long name_len,
-                                                 TF_Status status);
-public static native TF_Buffer TF_ApiDefMapGet(TF_ApiDefMap api_def_map,
-                                                 String name,
-                                                 @Cast("size_t") long name_len,
-                                                 TF_Status status);
-
-// --------------------------------------------------------------------------
-// Kernel definition information.
-
-// Returns a serialized KernelList protocol buffer containing KernelDefs for all
-// registered kernels.
-public static native TF_Buffer TF_GetAllRegisteredKernels(TF_Status status);
-
-// Returns a serialized KernelList protocol buffer containing KernelDefs for all
-// kernels registered for the operation named `name`.
-public static native TF_Buffer TF_GetRegisteredKernelsForOp(
-    @Cast("const char*") BytePointer name, TF_Status status);
-public static native TF_Buffer TF_GetRegisteredKernelsForOp(
-    String name, TF_Status status);
-// Targeting ../TF_Server.java
-
-
-
-// Creates a new in-process TensorFlow server configured using a serialized
-// ServerDef protocol buffer provided via `proto` and `proto_len`.
-//
-// The server will not serve any requests until TF_ServerStart is invoked.
-// The server will stop serving requests once TF_ServerStop or
-// TF_DeleteServer is invoked.
-public static native TF_Server TF_NewServer(@Const Pointer proto,
-                                              @Cast("size_t") long proto_len,
-                                              TF_Status status);
-
-// Starts an in-process TensorFlow server.
-public static native void TF_ServerStart(TF_Server server, TF_Status status);
-
-// Stops an in-process TensorFlow server.
-public static native void TF_ServerStop(TF_Server server, TF_Status status);
-
-// Blocks until the server has been successfully stopped (via TF_ServerStop or
-// TF_ServerClose).
-public static native void TF_ServerJoin(TF_Server server, TF_Status status);
-
-// Returns the target string that can be provided to TF_SetTarget() to connect
-// a TF_Session to `server`.
-//
-// The returned string is valid only until TF_DeleteServer is invoked.
-public static native @Cast("const char*") BytePointer TF_ServerTarget(TF_Server server);
-
-// Destroy an in-process TensorFlow server, frees memory. If server is running
-// it will be stopped and joined.
-public static native void TF_DeleteServer(TF_Server server);
-// Targeting ../Listener_BytePointer.java
-
-
-public static native void TF_RegisterLogListener(
-    Listener_BytePointer listener);
-// Targeting ../Listener_String.java
-
-
-public static native void TF_RegisterLogListener(
-    Listener_String listener);
-
-// #ifdef __cplusplus /* end extern "C" */
-// #endif
-
-// #endif  // TENSORFLOW_C_C_API_H_
-
-
-// Parsed from tensorflow/c/kernels.h
-
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-==============================================================================*/
-
-// #ifndef TENSORFLOW_C_KERNELS_H_
-// #define TENSORFLOW_C_KERNELS_H_
-
-// #include <stdint.h>
-
-// #include "tensorflow/c/tf_datatype.h"
-// #include "tensorflow/c/tf_status.h"
-
-// Macro to control visibility of exported symbols in the shared library (.so,
-// .dylib, .dll).
-// This duplicates the TF_EXPORT macro definition in
-// tensorflow/core/platform/macros.h in order to keep this .h file independent
-// of any other includes.
-// #ifdef SWIG
-// #define TF_CAPI_EXPORT
-// #else
-// #if defined(_WIN32)
-// #ifdef TF_COMPILE_LIBRARY
-// #define TF_CAPI_EXPORT __declspec(dllexport)
-// #else
-// #define TF_CAPI_EXPORT __declspec(dllimport)
-// #endif  // TF_COMPILE_LIBRARY
-// #else
-// #define TF_CAPI_EXPORT __attribute__((visibility("default")))
-// #endif  // _WIN32
-// #endif  // SWIG
-
-// #ifdef __cplusplus
-// #endif
-// Targeting ../TF_KernelBuilder.java
-
-
-// Targeting ../TF_OpKernelConstruction.java
-
-
-// Targeting ../TF_OpKernelContext.java
-
-
-// Targeting ../Create_func_TF_OpKernelConstruction.java
-
-
-// Targeting ../Compute_func_Pointer_TF_OpKernelContext.java
-
-
-// Targeting ../Delete_func_Pointer.java
-
-
-public static native TF_KernelBuilder TF_NewKernelBuilder(
-    @Cast("const char*") BytePointer op_name, @Cast("const char*") BytePointer device_name,
-    Create_func_TF_OpKernelConstruction create_func,
-    Compute_func_Pointer_TF_OpKernelContext compute_func,
-    Delete_func_Pointer delete_func);
-public static native TF_KernelBuilder TF_NewKernelBuilder(
-    String op_name, String device_name,
-    Create_func_TF_OpKernelConstruction create_func,
-    Compute_func_Pointer_TF_OpKernelContext compute_func,
-    Delete_func_Pointer delete_func);
-
-// Specifies that this kernel's attribute only supports the given type.
-public static native void TF_KernelBuilder_TypeConstraint(
-    TF_KernelBuilder kernel_builder, @Cast("const char*") BytePointer attr_name,
-    @Cast("const TF_DataType") int type, TF_Status status);
-public static native void TF_KernelBuilder_TypeConstraint(
-    TF_KernelBuilder kernel_builder, String attr_name,
-    @Cast("const TF_DataType") int type, TF_Status status);
-
-// Specify that this kernel requires/provides an input/output arg
-// in host memory (instead of the default, device memory).
-public static native void TF_KernelBuilder_HostMemory(
-    TF_KernelBuilder kernel_builder, @Cast("const char*") BytePointer arg_name);
-public static native void TF_KernelBuilder_HostMemory(
-    TF_KernelBuilder kernel_builder, String arg_name);
-
-// Register the given kernel builder with the TensorFlow runtime. If
-// registration fails, the given status will be populated.
-//
-// This call takes ownership of the `builder` pointer.
-public static native void TF_RegisterKernelBuilder(@Cast("const char*") BytePointer kernel_name,
-                                                    TF_KernelBuilder builder,
-                                                    TF_Status status);
-public static native void TF_RegisterKernelBuilder(String kernel_name,
-                                                    TF_KernelBuilder builder,
-                                                    TF_Status status);
-
-// Deletes the given TF_KernelBuilder. This should be called only if the kernel
-// builder is not registered with TensorFlow via TF_RegisterKernelBuilder.
-public static native void TF_DeleteKernelBuilder(TF_KernelBuilder builder);
-
-// --------------------------------------------------------------------------
-// OpKernelContext routines
-
-// TF_NumInputs returns the number of inputs available in ctx.
-public static native int TF_NumInputs(TF_OpKernelContext ctx);
-
-// TF_NumOutputs returns the number of outputs to be placed in *ctx by the
-// kernel.
-public static native int TF_NumOutputs(TF_OpKernelContext ctx);
-
-// Retrieves the ith input from ctx. If TF_GetCode(status) is TF_OK, *tensor is
-// populated and its ownership is passed to the caller. In any other case,
-// *tensor is not modified.
-//
-// If i < 0 or i >= TF_NumInputs(ctx), *status is set to TF_OUT_OF_RANGE.
-public static native void TF_GetInput(TF_OpKernelContext ctx, int i,
-                                       @Cast("TF_Tensor**") PointerPointer tensor, TF_Status status);
-public static native void TF_GetInput(TF_OpKernelContext ctx, int i,
-                                       @ByPtrPtr TF_Tensor tensor, TF_Status status);
-
-// Sets the ith output of ctx to tensor. If TF_GetCode(status) is anything but
-// TF_OK, ctx is left unmodified.
-//
-// If i < 0 or i >= TF_NumOutputs(ctx), *status is set to TF_OUT_OF_RANGE.
-public static native void TF_SetOutput(TF_OpKernelContext ctx, int i,
-                                        @Const TF_Tensor tensor,
-                                        TF_Status status);
-
-// Notifies the given OpKernelConstruction that kernel construction has failed.
-public static native void TF_OpKernelConstruction_Failure(
-    TF_OpKernelConstruction ctx, TF_Status status);
-
-// Notifies the given OpKernelContext that the kernel's compute function has
-// failed.
-public static native void TF_OpKernelContext_Failure(TF_OpKernelContext ctx,
-                                                      TF_Status status);
-
-// Returns the expected output data type of the ith output. If i < 0 or
-// i >= TF_NumOutputs(ctx), the program aborts.
-public static native @Cast("TF_DataType") int TF_ExpectedOutputDataType(
-    TF_OpKernelContext ctx, int i);
-
-// Returns the step ID of the given context.
-public static native @Cast("int64_t") long TF_StepId(TF_OpKernelContext ctx);
-
-// Interprets the named kernel construction attribute as a TF_DataType and
-// places it into *val. *status is set to TF_OK.
-//
-// If the attribute could not be found or could not be interpreted as
-// TF_DataType, *status is populated with an error.
-public static native void TF_OpKernelConstruction_GetAttrType(
-    TF_OpKernelConstruction ctx, @Cast("const char*") BytePointer attr_name, @Cast("TF_DataType*") IntPointer val,
-    TF_Status status);
-public static native void TF_OpKernelConstruction_GetAttrType(
-    TF_OpKernelConstruction ctx, String attr_name, @Cast("TF_DataType*") IntBuffer val,
-    TF_Status status);
-public static native void TF_OpKernelConstruction_GetAttrType(
-    TF_OpKernelConstruction ctx, @Cast("const char*") BytePointer attr_name, @Cast("TF_DataType*") int[] val,
-    TF_Status status);
-public static native void TF_OpKernelConstruction_GetAttrType(
-    TF_OpKernelConstruction ctx, String attr_name, @Cast("TF_DataType*") IntPointer val,
-    TF_Status status);
-public static native void TF_OpKernelConstruction_GetAttrType(
-    TF_OpKernelConstruction ctx, @Cast("const char*") BytePointer attr_name, @Cast("TF_DataType*") IntBuffer val,
-    TF_Status status);
-public static native void TF_OpKernelConstruction_GetAttrType(
-    TF_OpKernelConstruction ctx, String attr_name, @Cast("TF_DataType*") int[] val,
-    TF_Status status);
-
-// Interprets the named kernel construction attribute as int32_t and
-// places it into *val. *status is set to TF_OK.
-//
-// If the attribute could not be found or could not be interpreted as
-// int32, *status is populated with an error.
-public static native void TF_OpKernelConstruction_GetAttrInt32(
-    TF_OpKernelConstruction ctx, @Cast("const char*") BytePointer attr_name, IntPointer val,
-    TF_Status status);
-public static native void TF_OpKernelConstruction_GetAttrInt32(
-    TF_OpKernelConstruction ctx, String attr_name, IntBuffer val,
-    TF_Status status);
-public static native void TF_OpKernelConstruction_GetAttrInt32(
-    TF_OpKernelConstruction ctx, @Cast("const char*") BytePointer attr_name, int[] val,
-    TF_Status status);
-public static native void TF_OpKernelConstruction_GetAttrInt32(
-    TF_OpKernelConstruction ctx, String attr_name, IntPointer val,
-    TF_Status status);
-public static native void TF_OpKernelConstruction_GetAttrInt32(
-    TF_OpKernelConstruction ctx, @Cast("const char*") BytePointer attr_name, IntBuffer val,
-    TF_Status status);
-public static native void TF_OpKernelConstruction_GetAttrInt32(
-    TF_OpKernelConstruction ctx, String attr_name, int[] val,
-    TF_Status status);
-
-// Allocates Tensor for output at given index. Caller takes ownership of
-// returned TF_Tensor and should deallocate it using TF_DeleteTensor(tensor).
-//
-// This function should be used to allocate outputs inside kernel
-// compute function.
-public static native TF_Tensor TF_AllocateOutput(TF_OpKernelContext context,
-                                            int index, @Cast("TF_DataType") int dtype,
-                                            @Cast("int64_t*") LongPointer dims, int num_dims,
-                                            @Cast("size_t") long len, TF_Status status);
-public static native TF_Tensor TF_AllocateOutput(TF_OpKernelContext context,
-                                            int index, @Cast("TF_DataType") int dtype,
-                                            @Cast("int64_t*") LongBuffer dims, int num_dims,
-                                            @Cast("size_t") long len, TF_Status status);
-public static native TF_Tensor TF_AllocateOutput(TF_OpKernelContext context,
-                                            int index, @Cast("TF_DataType") int dtype,
-                                            @Cast("int64_t*") long[] dims, int num_dims,
-                                            @Cast("size_t") long len, TF_Status status);
-
-// #ifdef __cplusplus /* end extern "C" */
-// #endif
-
-// #endif  // TENSORFLOW_C_KERNELS_H_
-
-
-// Parsed from tensorflow/c/ops.h
-
-/* Copyright 2019 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-==============================================================================*/
-
-// Routines for registering new ops and for implementing op shape inference
-// functions.
-//
-// This API is alpha software and is subject to change.
-//
-// REGISTRATION
-// ------------
-//
-// In order to register a new op, create a new TF_OpDefinitionBuilder:
-//
-// TF_OpDefinitionBuilder* builder = TF_NewOpDefinitionBuilder("OpName");
-//
-// Inputs, outputs and attributes can be added to the builder with the
-// corresponding functions, e.g.
-//
-// TF_OpDefinitionBuilderAddInput(builder, "input1: int32");
-// TF_OpDefinitionBuilderAddOutput(builder, "output1: int64");
-// TF_OpDefinitionBuilderAddAttr(builder, "attr: int32");
-//
-// The builder may then be registered with TensorFlow using the
-// TF_RegisterOpDefinition function. E.g.
-//
-// TF_Status* status = TF_NewStatus();
-// TF_RegisterOpDefinition(builder, &status);
-// if (TF_GetCode(status) != TF_OK) {
-//   // handle error
-// }
-//
-// SHAPE INFERENCE
-// ---------------
-//
-// You can provide a shape inference function that TensorFlow will call when it
-// wants to understand the shape of outputs that the op will produce. Use the
-// TF_OpDefinitionBuilderSetShapeInferenceFunction function to register a shape
-// inference function pointer with TensorFlow. The following is an example of a
-// very simple shape inference function:
-//
-// void identity_shape_fn(TF_ShapeInferenceContext* ctx, TF_Status* status) {
-//   TF_ShapeHandle* input = TF_NewShapeHandle();
-//   TF_ShapeInferenceContextGetInput(ctx, 0, input, status);
-//   if (TF_GetCode(status) == TF_OK) {
-//     TF_ShapeInferenceContextSetOutput(ctx, 0, input, status);
-//   }
-//   TF_DeleteShapeHandle(input);
-// }
-//
-// The following code registers the inference function with TensorFlow:
-//
-// TF_OpDefinitionBuilderSetShapeInferenceFunction(builder, &identity_shape_fn);
-//
-// For more details about shape inference, see the documentation for
-// TF_OpDefinitionBuilderSetShapeInferenceFunction.
-
-// #ifndef TENSORFLOW_C_OPS_H_
-// #define TENSORFLOW_C_OPS_H_
-
-// #include <stdbool.h>
-// #include <stdint.h>
-// #include <stdlib.h>
-
-// #include "tensorflow/c/tf_datatype.h"
-// #include "tensorflow/c/tf_status.h"
-
-// #ifdef SWIG
-// #define TF_CAPI_EXPORT
-// #else
-// #if defined(_WIN32)
-// #ifdef TF_COMPILE_LIBRARY
-// #define TF_CAPI_EXPORT __declspec(dllexport)
-// #else
-// #define TF_CAPI_EXPORT __declspec(dllimport)
-// #endif  // TF_COMPILE_LIBRARY
-// #else
-// #define TF_CAPI_EXPORT __attribute__((visibility("default")))
-// #endif  // _WIN32
-// #endif  // SWIG
-
-// #ifdef __cplusplus
-// Targeting ../TF_DimensionHandle.java
-
-
-// Targeting ../TF_OpDefinitionBuilder.java
-
-
-// Targeting ../TF_ShapeHandle.java
-
-
-// Targeting ../TF_ShapeInferenceContext.java
-
-
-
-// Returns a newly allocated op definition builder for the given op name. The
-// returned builder may be customized with the `TF_OpDefinitionBuilder...`
-// functions and then registered with TensorFlow with TF_RegisterOpDefinition.
-//
-// The returned pointer is either freed by a call to TF_RegisterOpDefinition, or
-// can be manually deleted by TF_DeleteOpDefinitionBuilder if it is never
-// registered.
-public static native TF_OpDefinitionBuilder TF_NewOpDefinitionBuilder(
-    @Cast("const char*") BytePointer op_name);
-public static native TF_OpDefinitionBuilder TF_NewOpDefinitionBuilder(
-    String op_name);
-
-// Registers the given op builder with TensorFlow. Indicates success or
-// otherwise in the given status.
-//
-// `builder` is freed whether the op was successfully registered or not. You
-// must call either this function or TF_DeleteOpDefinitionBuilder to free the
-// builder, but never both.
-public static native void TF_RegisterOpDefinition(
-    TF_OpDefinitionBuilder builder, TF_Status status);
-
-// Frees the given op definition builder. You must call either this function or
-// TF_RegisterOpDefinition to free the builder, but never both.
-public static native void TF_DeleteOpDefinitionBuilder(
-    TF_OpDefinitionBuilder builder);
-
-//----------------------------------------------------
-// Attribute functions.
-
-// Adds an attr to the given TF_OpDefinitionBuilder. The spec has
-// format "<name>:<type>" or "<name>:<type>=<default>"
-// where <name> matches regexp [a-zA-Z][a-zA-Z0-9_]*.
-// By convention, names containing only capital letters are reserved for
-// attributes whose values can be inferred by the operator implementation if not
-// supplied by the user. If the attribute name contains characters other than
-// capital letters, the operator expects the user to provide the attribute value
-// at operation runtime.
-//
-// <type> can be:
-//   "string", "int", "float", "bool", "type", "shape", or "tensor"
-//   "numbertype", "realnumbertype", "quantizedtype"
-//       (meaning "type" with a restriction on valid values)
-//   "{int32,int64}" or {realnumbertype,quantizedtype,string}"
-//       (meaning "type" with a restriction containing unions of value types)
-//   "{\"foo\", \"bar\n baz\"}", or "{'foo', 'bar\n baz'}"
-//       (meaning "string" with a restriction on valid values)
-//   "list(string)", ..., "list(tensor)", "list(numbertype)", ...
-//       (meaning lists of the above types)
-//   "int >= 2" (meaning "int" with a restriction on valid values)
-//   "list(string) >= 2", "list(int) >= 2"
-//       (meaning "list(string)" / "list(int)" with length at least 2)
-// <default>, if included, should use the Proto text format
-// of <type>.  For lists use [a, b, c] format.
-//
-// Note that any attr specifying the length of an input or output will
-// get a default minimum of 1 unless the >= # syntax is used.
-public static native void TF_OpDefinitionBuilderAddAttr(
-    TF_OpDefinitionBuilder builder, @Cast("const char*") BytePointer attr_spec);
-public static native void TF_OpDefinitionBuilderAddAttr(
-    TF_OpDefinitionBuilder builder, String attr_spec);
-
-// Adds an input to this TF_OpDefinitionBuilder.
-// The spec has form "<name>:<type-expr>" or "<name>:Ref(<type-expr>)"
-// where <name> matches regexp [a-z][a-z0-9_]* and <type-expr> can be:
-// * For a single tensor: <type>
-// * For a sequence of tensors with the same type: <number>*<type>
-// * For a sequence of tensors with different types: <type-list>
-// Where:
-//   <type> is either one of "float", "int32", "string", ...
-//          or the name of an attr (see TF_OpDefinitionBuilderAddAttr)
-//          with type "type".
-//   <number> is the name of an attr with type "int".
-//   <type-list> is the name of an attr with type "list(type)".
-public static native void TF_OpDefinitionBuilderAddInput(
-    TF_OpDefinitionBuilder builder, @Cast("const char*") BytePointer input_spec);
-public static native void TF_OpDefinitionBuilderAddInput(
-    TF_OpDefinitionBuilder builder, String input_spec);
-
-// Adds an output to this TF_OpDefinitionBuilder.
-// The spec has form "<name>:<type-expr>" or "<name>:Ref(<type-expr>)"
-// where <name> matches regexp [a-z][a-z0-9_]* and <type-expr> can be:
-// * For a single tensor: <type>
-// * For a sequence of tensors with the same type: <number>*<type>
-// * For a sequence of tensors with different types: <type-list>
-// Where:
-//   <type> is either one of "float", "int32", "string", ...
-//          or the name of an attr (see TF_OpDefinitionBuilderAddAttr)
-//          with type "type".
-//   <number> is the name of an attr with type "int".
-//   <type-list> is the name of an attr with type "list(type)".
-public static native void TF_OpDefinitionBuilderAddOutput(
-    TF_OpDefinitionBuilder builder, @Cast("const char*") BytePointer output_spec);
-public static native void TF_OpDefinitionBuilderAddOutput(
-    TF_OpDefinitionBuilder builder, String output_spec);
-
-// Sets the commutative property for the op built by the given builder.
-public static native void TF_OpDefinitionBuilderSetIsCommutative(
-    TF_OpDefinitionBuilder builder, @Cast("bool") boolean is_commutative);
-
-// Sets the is_aggregate property of the builder to the given value.
-//
-// If is_aggregate is true, then the operation produced by this builder accepts
-// N >= 2 inputs and produces 1 output all of the same type. Should be
-// associative and commutative, and produce output with the same shape as the
-// input. The optimizer may replace an aggregate op taking input from multiple
-// devices with a tree of aggregate ops that aggregate locally within each
-// device (and possibly within groups of nearby devices) before communicating.
-public static native void TF_OpDefinitionBuilderSetIsAggregate(
-    TF_OpDefinitionBuilder builder, @Cast("bool") boolean is_aggregate);
-
-// Sets the is_stateful property of the builder to the given value.
-//
-// The op built by this builder is stateful if its behavior depends on some
-// state beyond its input tensors (e.g. variable reading op) or if it has a
-// side-effect (e.g. printing or asserting ops). Equivalently, stateless ops
-// must always produce the same output for the same input and have no
-// side-effects.
-//
-// By default Ops may be moved between devices. Stateful ops should either not
-// be moved, or should only be moved if that state can also be moved (e.g. via
-// some sort of save / restore). Stateful ops are guaranteed to never be
-// optimized away by Common Subexpression Elimination (CSE).
-public static native void TF_OpDefinitionBuilderSetIsStateful(
-    TF_OpDefinitionBuilder builder, @Cast("bool") boolean is_stateful);
-
-// Sets the allows_uninitialized_input property of the operation built by this
-// builder.
-//
-// By default, all inputs to an Op must be initialized Tensors. Ops that may
-// initialize tensors for the first time should set this field to true, to allow
-// the Op to take an uninitialized Tensor as input.
-public static native void TF_OpDefinitionBuilderSetAllowsUninitializedInput(
-    TF_OpDefinitionBuilder builder, @Cast("bool") boolean allows_uninitialized_input);
-
-// Adds a deprecation warning for the given op. This indicates to the user that
-// `version` is the first TensorFlow GraphDef version for which the operation is
-// deprecated. `explanation` should contain the reason for the deprecation and
-// what to use instead.
-//
-// This function is only an indicator that the operation may disappear in a
-// version of TensorFlow after `version`. It does not affect op registration.
-public static native void TF_OpDefinitionBuilderDeprecated(
-    TF_OpDefinitionBuilder builder, int version, @Cast("const char*") BytePointer explanation);
-public static native void TF_OpDefinitionBuilderDeprecated(
-    TF_OpDefinitionBuilder builder, int version, String explanation);
-// Targeting ../Shape_inference_func_TF_ShapeInferenceContext_TF_Status.java
-
-
-public static native void TF_OpDefinitionBuilderSetShapeInferenceFunction(
-    TF_OpDefinitionBuilder builder,
-    Shape_inference_func_TF_ShapeInferenceContext_TF_Status shape_inference_func);
-
-//----------------------------------------------------
-// Functions for TF_ShapeInferenceContext.
-//
-// Functions for implementing shape inference functions. TensorFlow uses these
-// functions to determine the shape of tensors produced by an operation without
-// having to actually run the operation. If an operation chooses to provide a
-// shape inference function, it will be invoked by TensorFlow as needed.
-//
-// When invoked by TensorFlow, the shape inference function is provided with a
-// TF_ShapeInferenceContext pointer. The function's implementation will use the
-// accessor and mutator functions with names beginning with
-// TF_ShapeInferenceContext to examine the input state and determine the output
-// shape.
-
-// Returns the number of inputs in the given shape inference context.
-public static native @Cast("int64_t") long TF_ShapeInferenceContextNumInputs(
-    TF_ShapeInferenceContext ctx);
-
-// Returns a newly allocated shape handle. The shapes represented by these
-// handles may be queried or mutated with the corresponding
-// TF_ShapeInferenceContext...  functions.
-public static native TF_ShapeHandle TF_NewShapeHandle();
-
-// Places the ith input of the given shape inference context into the given
-// shape handle, or returns a status other than TF_OK indicating why the input
-// could not be retrieved
-// (for example, if i < 0 || i >= TF_ShapeInferenceContextNumInputs(ctx)).
-public static native void TF_ShapeInferenceContextGetInput(
-    TF_ShapeInferenceContext ctx, int i, TF_ShapeHandle handle,
-    TF_Status status);
-
-// Places the given shape handle into the `i`th output position of the given
-// context. Internally, the shape handle is copied; the caller may subsequently
-// delete `handle`.
-public static native void TF_ShapeInferenceContextSetOutput(TF_ShapeInferenceContext ctx,
-                                              int i, TF_ShapeHandle handle,
-                                              TF_Status status);
-
-// Returns a newly-allocate shape handle representing a vector of the given
-// size. The returned handle should be freed with TF_DeleteShapeHandle.
-public static native TF_ShapeHandle TF_ShapeInferenceContextVectorFromSize(
-    TF_ShapeInferenceContext ctx, @Cast("size_t") long size);
-
-// Returns a newly allocated dimension handle. It must be freed with
-// TF_DeleteDimensionHandle.
-public static native TF_DimensionHandle TF_NewDimensionHandle();
-
-// Interprets the named shape inference context attribute as a TF_DataType and
-// places it into *val. *status is set to TF_OK.
-//
-// If the attribute could not be found or could not be interpreted as
-// TF_DataType, *status is populated with an error.
-public static native void TF_ShapeInferenceContext_GetAttrType(
-    TF_ShapeInferenceContext ctx, @Cast("const char*") BytePointer attr_name, @Cast("TF_DataType*") IntPointer val,
-    TF_Status status);
-public static native void TF_ShapeInferenceContext_GetAttrType(
-    TF_ShapeInferenceContext ctx, String attr_name, @Cast("TF_DataType*") IntBuffer val,
-    TF_Status status);
-public static native void TF_ShapeInferenceContext_GetAttrType(
-    TF_ShapeInferenceContext ctx, @Cast("const char*") BytePointer attr_name, @Cast("TF_DataType*") int[] val,
-    TF_Status status);
-public static native void TF_ShapeInferenceContext_GetAttrType(
-    TF_ShapeInferenceContext ctx, String attr_name, @Cast("TF_DataType*") IntPointer val,
-    TF_Status status);
-public static native void TF_ShapeInferenceContext_GetAttrType(
-    TF_ShapeInferenceContext ctx, @Cast("const char*") BytePointer attr_name, @Cast("TF_DataType*") IntBuffer val,
-    TF_Status status);
-public static native void TF_ShapeInferenceContext_GetAttrType(
-    TF_ShapeInferenceContext ctx, String attr_name, @Cast("TF_DataType*") int[] val,
-    TF_Status status);
-
-// Returns the rank of the shape represented by the given handle.
-public static native @Cast("int64_t") long TF_ShapeInferenceContextRank(
-    TF_ShapeInferenceContext ctx, TF_ShapeHandle handle);
-
-// Returns 1 if `handle` has a known rank, 0 otherwise.
-public static native int TF_ShapeInferenceContextRankKnown(
-    TF_ShapeInferenceContext ctx, TF_ShapeHandle handle);
-
-// If <handle> has rank <rank>, or its rank is unknown, return OK and return the
-// shape with asserted rank in <*result>. Otherwise an error is placed into
-// `status`.
-public static native void TF_ShapeInferenceContextWithRank(
-    TF_ShapeInferenceContext ctx, TF_ShapeHandle handle, @Cast("int64_t") long rank,
-    TF_ShapeHandle result, TF_Status status);
-
-// If <handle> has rank at least <rank>, or its rank is unknown, return OK and
-// return the shape with asserted rank in <*result>. Otherwise an error is
-// placed into `status`.
-public static native void TF_ShapeInferenceContextWithRankAtLeast(
-    TF_ShapeInferenceContext ctx, TF_ShapeHandle handle, @Cast("int64_t") long rank,
-    TF_ShapeHandle result, TF_Status status);
-
-// If <handle> has rank at most <rank>, or its rank is unknown, return OK and
-// return the shape with asserted rank in <*result>. Otherwise an error is
-// placed into `status`.
-public static native void TF_ShapeInferenceContextWithRankAtMost(
-    TF_ShapeInferenceContext ctx, TF_ShapeHandle handle, @Cast("int64_t") long rank,
-    TF_ShapeHandle result, TF_Status status);
-
-// Places a handle to the ith dimension of the given shape into *result.
-public static native void TF_ShapeInferenceContextDim(
-    TF_ShapeInferenceContext ctx, TF_ShapeHandle shape_handle, @Cast("int64_t") long i,
-    TF_DimensionHandle result);
-
-// Returns in <*result> a sub-shape of <shape_handle>, with dimensions
-// [start:end]. <start> and <end> can be negative, to index from the end of the
-// shape. <start> and <end> are set to the rank of <shape_handle> if > rank of
-// <shape_handle>.
-public static native void TF_ShapeInferenceContextSubshape(
-    TF_ShapeInferenceContext ctx, TF_ShapeHandle shape_handle, @Cast("int64_t") long start,
-    @Cast("int64_t") long end, TF_ShapeHandle result, TF_Status status);
-
-// Places an unknown shape in all outputs for the given inference context. Used
-// for shape inference functions with ops whose output shapes are unknown.
-public static native void TF_ShapeInferenceContextSetUnknownShape(
-    TF_ShapeInferenceContext ctx, TF_Status status);
-
-// Returns whether the given handle represents a known dimension.
-public static native int TF_DimensionHandleValueKnown(
-    TF_DimensionHandle dim_handle);
-
-// Returns the value of the given dimension.
-public static native @Cast("int64_t") long TF_DimensionHandleValue(
-    TF_DimensionHandle dim_handle);
-
-// Returns in <*result> the result of appending the dimensions of <second> to
-// those of <first>.
-public static native void TF_ShapeInferenceContextConcatenateShapes(
-    TF_ShapeInferenceContext ctx, TF_ShapeHandle first,
-    TF_ShapeHandle second, TF_ShapeHandle result, TF_Status status);
-
-// Frees the given shape handle.
-public static native void TF_DeleteShapeHandle(TF_ShapeHandle handle);
-
-// Frees the given dimension handle.
-public static native void TF_DeleteDimensionHandle(TF_DimensionHandle handle);
-
-// #ifdef __cplusplus /* end extern "C" */
-// #endif
-
-// #endif  // TENSORFLOW_C_OPS_H_
-
-
-// Parsed from tensorflow/c/eager/c_api.h
-
-/* Copyright 2017 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-==============================================================================*/
-
-// #ifndef TENSORFLOW_C_EAGER_C_API_H_
-// #define TENSORFLOW_C_EAGER_C_API_H_
-
-// C API extensions to experiment with eager execution of kernels.
-// WARNING: Unlike tensorflow/c/c_api.h, the API here is not guaranteed to be
-// stable and can change without notice.
-
-// #include "tensorflow/c/c_api.h"
-
-// Macro to control visibility of exported symbols in the shared library (.so,
-// .dylib, .dll).
-// This duplicates the TF_EXPORT macro definition in
-// tensorflow/core/platform/macros.h in order to keep this .h file independent
-// of any other includes.$a
-// #ifdef SWIG
-// #define TF_CAPI_EXPORT
-// #else
-// #if defined(_WIN32)
-// #ifdef TF_COMPILE_LIBRARY
-// #define TF_CAPI_EXPORT __declspec(dllexport)
-// #else
-// #define TF_CAPI_EXPORT __declspec(dllimport)
-// #endif  // TF_COMPILE_LIBRARY
-// #else
-// #define TF_CAPI_EXPORT __attribute__((visibility("default")))
-// #endif  // _WIN32
-// #endif  // SWIG
-
-// #ifdef __cplusplus
-// Targeting ../TFE_ContextOptions.java
-
-
-
-// Return a new options object.
-public static native TFE_ContextOptions TFE_NewContextOptions();
-
-// Set the config in TF_ContextOptions.options.
-// config should be a serialized tensorflow.ConfigProto proto.
-// If config was not parsed successfully as a ConfigProto, record the
-// error information in *status.
-public static native void TFE_ContextOptionsSetConfig(
-    TFE_ContextOptions options, @Const Pointer proto, @Cast("size_t") long proto_len,
-    TF_Status status);
-
-// Controls how to act when we try to run an operation on a given device but
-// some input tensors are not on that device.
-// LINT.IfChange
-// Note: Keep in sync with internal copy of enum in eager/context.h.
-/** enum TFE_ContextDevicePlacementPolicy */
-public static final int
-  // Running operations with input tensors on the wrong device will fail.
-  TFE_DEVICE_PLACEMENT_EXPLICIT = 0,
-  // Copy the tensor to the right device but log a warning.
-  TFE_DEVICE_PLACEMENT_WARN = 1,
-  // Silently copy the tensor, which has a performance cost since the operation
-  // will be blocked till the copy completes. This is the default placement
-  // policy.
-  TFE_DEVICE_PLACEMENT_SILENT = 2,
-  // Placement policy which silently copies int32 tensors but not other dtypes.
-  TFE_DEVICE_PLACEMENT_SILENT_FOR_INT32 = 3;
-// LINT.ThenChange(//tensorflow/core/common_runtime/eager/context.h)
-
-// Sets the default execution mode (sync/async). Note that this can be
-// overridden per thread using TFE_ContextSetExecutorForThread.
-public static native void TFE_ContextOptionsSetAsync(TFE_ContextOptions arg0,
-                                                      @Cast("unsigned char") byte enable);
-
-public static native void TFE_ContextOptionsSetDevicePlacementPolicy(
-    TFE_ContextOptions arg0, @Cast("TFE_ContextDevicePlacementPolicy") int arg1);
-
-// Destroy an options object.
-public static native void TFE_DeleteContextOptions(TFE_ContextOptions arg0);
-// Targeting ../TFE_Context.java
-
-
-
-public static native TFE_Context TFE_NewContext(
-    @Const TFE_ContextOptions opts, TF_Status status);
-public static native void TFE_DeleteContext(TFE_Context ctx);
-public static native TF_DeviceList TFE_ContextListDevices(TFE_Context ctx,
-                                                            TF_Status status);
-
-// Clears the internal caches in the TFE context. Useful when reseeding random
-// ops.
-public static native void TFE_ContextClearCaches(TFE_Context ctx);
-
-// Sets a thread-local device placement policy. After this call, other calls to
-// TFE_Execute in the same thread will use the device policy specified here
-// instead of the device policy used to construct the context. This has no
-// effect on the device policy used by other program threads.
-public static native void TFE_ContextSetThreadLocalDevicePlacementPolicy(
-    TFE_Context ctx, @Cast("TFE_ContextDevicePlacementPolicy") int policy);
-
-// Returns the device placement policy to be used by this context in the current
-// thread.
-public static native @Cast("TFE_ContextDevicePlacementPolicy") int TFE_ContextGetDevicePlacementPolicy(TFE_Context ctx);
-
-// A tensorflow.ServerDef specifies remote workers (in addition to the current
-// workers name). Operations created on this context can then be executed on
-// any of these remote workers by setting an appropriate device.
-//
-// If the following is set, all servers identified by the
-// ServerDef must be up when the context is created.
-public static native void TFE_ContextSetServerDef(TFE_Context ctx,
-                                                   int keep_alive_secs,
-                                                   @Const Pointer proto,
-                                                   @Cast("size_t") long proto_len,
-                                                   TF_Status status);
-// Targeting ../TFE_TensorHandle.java
-
-
-
-public static native TFE_TensorHandle TFE_NewTensorHandle(@Const TF_Tensor t,
-                                                            TF_Status status);
-// Indicates that the caller will not be using `h` any more.
-public static native void TFE_DeleteTensorHandle(TFE_TensorHandle h);
-public static native @Cast("TF_DataType") int TFE_TensorHandleDataType(TFE_TensorHandle h);
-// This function will block till the operation that produces `h` has completed.
-public static native int TFE_TensorHandleNumDims(TFE_TensorHandle h,
-                                                  TF_Status status);
-public static native @Cast("int64_t") long TFE_TensorHandleNumElements(TFE_TensorHandle h,
-                                                          TF_Status status);
-// This function will block till the operation that produces `h` has completed.
-public static native @Cast("int64_t") long TFE_TensorHandleDim(TFE_TensorHandle h,
-                                                  int dim_index,
-                                                  TF_Status status);
-
-// Returns the device of the operation that produced `h`. If `h` was produced by
-// a copy, returns the destination device of the copy. Note that the returned
-// device name is not always the device holding the tensor handle's memory. If
-// you want the latter, use TFE_TensorHandleBackingDeviceName. This function
-// will block till the operation that produces `h` has completed.
-public static native @Cast("const char*") BytePointer TFE_TensorHandleDeviceName(
-    TFE_TensorHandle h, TF_Status status);
-
-// Returns the name of the device in whose memory `h` resides.
-//
-// This function will block till the operation that produces `h` has completed.
-public static native @Cast("const char*") BytePointer TFE_TensorHandleBackingDeviceName(
-    TFE_TensorHandle h, TF_Status status);
-
-// Return a pointer to a new TFE_TensorHandle that shares the underlying tensor
-// with `h`. On success, `status` is set to OK. On failure, `status` reflects
-// the error and a nullptr is returned.
-public static native TFE_TensorHandle TFE_TensorHandleCopySharingTensor(
-    TFE_TensorHandle h, TF_Status status);
-
-// This function will block till the operation that produces `h` has
-// completed. The memory returned might alias the internal memory used by
-// TensorFlow. Hence, callers should not mutate this memory (for example by
-// modifying the memory region pointed to by TF_TensorData() on the returned
-// TF_Tensor).
-public static native TF_Tensor TFE_TensorHandleResolve(TFE_TensorHandle h,
-                                                         TF_Status status);
-
-// Create a new TFE_TensorHandle with the same contents as 'h' but placed
-// in the memory of the device name 'device_name'.
-// If source and destination are the same device, then this creates a new handle
-// that shares the underlying buffer. Otherwise, it currently requires at least
-// one of the source or destination devices to be CPU (i.e., for the source or
-// destination tensor to be placed in host memory).
-// If async execution is enabled, the copy may be enqueued and the call will
-// return "non-ready" handle. Else, this function returns after the copy has
-// been done.
-public static native TFE_TensorHandle TFE_TensorHandleCopyToDevice(
-    TFE_TensorHandle h, TFE_Context ctx, @Cast("const char*") BytePointer device_name,
-    TF_Status status);
-public static native TFE_TensorHandle TFE_TensorHandleCopyToDevice(
-    TFE_TensorHandle h, TFE_Context ctx, String device_name,
-    TF_Status status);
-// Targeting ../TFE_TensorDebugInfo.java
-
-
-
-// Retrieves TFE_TensorDebugInfo for `handle`.
-// If TFE_TensorHandleTensorDebugInfo succeeds, `status` is set to OK and caller
-// is responsible for deleting returned TFE_TensorDebugInfo.
-// If TFE_TensorHandleTensorDebugInfo fails, `status` is set to appropriate
-// error and nullptr is returned. This function can block till the operation
-// that produces `handle` has completed.
-public static native TFE_TensorDebugInfo TFE_TensorHandleTensorDebugInfo(
-    TFE_TensorHandle h, TF_Status status);
-
-// Deletes `debug_info`.
-public static native void TFE_DeleteTensorDebugInfo(
-    TFE_TensorDebugInfo debug_info);
-
-// Returns the number of dimensions used to represent the tensor on its device.
-// The number of dimensions used to represent the tensor on device can be
-// different from the number returned by TFE_TensorHandleNumDims.
-// The return value was current at the time of TFE_TensorDebugInfo creation.
-public static native int TFE_TensorDebugInfoOnDeviceNumDims(
-    TFE_TensorDebugInfo debug_info);
-
-// Returns the number of elements in dimension `dim_index`.
-// Tensor representation on device can be transposed from its representation
-// on host. The data contained in dimension `dim_index` on device
-// can correspond to the data contained in another dimension in on-host
-// representation. The dimensions are indexed using the standard TensorFlow
-// major-to-minor order (slowest varying dimension first),
-// not the XLA's minor-to-major order.
-// On-device dimensions can be padded. TFE_TensorDebugInfoOnDeviceDim returns
-// the number of elements in a dimension after padding.
-// The return value was current at the time of TFE_TensorDebugInfo creation.
-public static native @Cast("int64_t") long TFE_TensorDebugInfoOnDeviceDim(
-    TFE_TensorDebugInfo debug_info, int dim_index);
-// Targeting ../TFE_Op.java
-
-
-
-public static native TFE_Op TFE_NewOp(TFE_Context ctx,
-                                        @Cast("const char*") BytePointer op_or_function_name,
-                                        TF_Status status);
-public static native TFE_Op TFE_NewOp(TFE_Context ctx,
-                                        String op_or_function_name,
-                                        TF_Status status);
-
-public static native void TFE_DeleteOp(TFE_Op op);
-
-public static native void TFE_OpSetDevice(TFE_Op op, @Cast("const char*") BytePointer device_name,
-                                           TF_Status status);
-public static native void TFE_OpSetDevice(TFE_Op op, String device_name,
-                                           TF_Status status);
-// The returned string remains valid throughout the lifetime of 'op'.
-public static native @Cast("const char*") BytePointer TFE_OpGetDevice(TFE_Op op,
-                                                  TF_Status status);
-
-// When 'enable' is set to 1, and if TensorFlow library is built with XLA
-// support, a subsequent TFE_Execute() call on `op` will run the op via XLA.
-//
-// If the library is not built with XLA support, this call would be a no-op.
-public static native void TFE_OpSetXLACompilation(TFE_Op op,
-                                                   @Cast("unsigned char") byte enable);
-
-public static native void TFE_OpAddInput(TFE_Op op, TFE_TensorHandle input,
-                                          TF_Status status);
-
-public static native void TFE_OpAddInputList(TFE_Op op,
-                                              @Cast("TFE_TensorHandle**") PointerPointer inputs,
-                                              int num_inputs,
-                                              TF_Status status);
-public static native void TFE_OpAddInputList(TFE_Op op,
-                                              @ByPtrPtr TFE_TensorHandle inputs,
-                                              int num_inputs,
-                                              TF_Status status);
-
-public static native @Cast("TF_AttrType") int TFE_OpGetAttrType(TFE_Op op,
-                                                    @Cast("const char*") BytePointer attr_name,
-                                                    @Cast("unsigned char*") BytePointer is_list,
-                                                    TF_Status status);
-public static native @Cast("TF_AttrType") int TFE_OpGetAttrType(TFE_Op op,
-                                                    String attr_name,
-                                                    @Cast("unsigned char*") ByteBuffer is_list,
-                                                    TF_Status status);
-public static native @Cast("TF_AttrType") int TFE_OpGetAttrType(TFE_Op op,
-                                                    @Cast("const char*") BytePointer attr_name,
-                                                    @Cast("unsigned char*") byte[] is_list,
-                                                    TF_Status status);
-public static native @Cast("TF_AttrType") int TFE_OpGetAttrType(TFE_Op op,
-                                                    String attr_name,
-                                                    @Cast("unsigned char*") BytePointer is_list,
-                                                    TF_Status status);
-public static native @Cast("TF_AttrType") int TFE_OpGetAttrType(TFE_Op op,
-                                                    @Cast("const char*") BytePointer attr_name,
-                                                    @Cast("unsigned char*") ByteBuffer is_list,
-                                                    TF_Status status);
-public static native @Cast("TF_AttrType") int TFE_OpGetAttrType(TFE_Op op,
-                                                    String attr_name,
-                                                    @Cast("unsigned char*") byte[] is_list,
-                                                    TF_Status status);
-// Get an attribute type given an op name; a fusion of TFE_NewOp and
-// TFE_OpGetAttrType for use from Python without the overhead of the individual
-// calls and memory management of TFE_Op.
-public static native @Cast("TF_AttrType") int TFE_OpNameGetAttrType(
-    TFE_Context ctx, @Cast("const char*") BytePointer op_or_function_name, @Cast("const char*") BytePointer attr_name,
-    @Cast("unsigned char*") BytePointer is_list, TF_Status status);
-public static native @Cast("TF_AttrType") int TFE_OpNameGetAttrType(
-    TFE_Context ctx, String op_or_function_name, String attr_name,
-    @Cast("unsigned char*") ByteBuffer is_list, TF_Status status);
-public static native @Cast("TF_AttrType") int TFE_OpNameGetAttrType(
-    TFE_Context ctx, @Cast("const char*") BytePointer op_or_function_name, @Cast("const char*") BytePointer attr_name,
-    @Cast("unsigned char*") byte[] is_list, TF_Status status);
-public static native @Cast("TF_AttrType") int TFE_OpNameGetAttrType(
-    TFE_Context ctx, String op_or_function_name, String attr_name,
-    @Cast("unsigned char*") BytePointer is_list, TF_Status status);
-public static native @Cast("TF_AttrType") int TFE_OpNameGetAttrType(
-    TFE_Context ctx, @Cast("const char*") BytePointer op_or_function_name, @Cast("const char*") BytePointer attr_name,
-    @Cast("unsigned char*") ByteBuffer is_list, TF_Status status);
-public static native @Cast("TF_AttrType") int TFE_OpNameGetAttrType(
-    TFE_Context ctx, String op_or_function_name, String attr_name,
-    @Cast("unsigned char*") byte[] is_list, TF_Status status);
-
-public static native void TFE_OpSetAttrString(TFE_Op op,
-                                               @Cast("const char*") BytePointer attr_name,
-                                               @Const Pointer value,
-                                               @Cast("size_t") long length);
-public static native void TFE_OpSetAttrString(TFE_Op op,
-                                               String attr_name,
-                                               @Const Pointer value,
-                                               @Cast("size_t") long length);
-public static native void TFE_OpSetAttrInt(TFE_Op op, @Cast("const char*") BytePointer attr_name,
-                                            @Cast("int64_t") long value);
-public static native void TFE_OpSetAttrInt(TFE_Op op, String attr_name,
-                                            @Cast("int64_t") long value);
-public static native void TFE_OpSetAttrFloat(TFE_Op op, @Cast("const char*") BytePointer attr_name,
-                                              float value);
-public static native void TFE_OpSetAttrFloat(TFE_Op op, String attr_name,
-                                              float value);
-public static native void TFE_OpSetAttrBool(TFE_Op op, @Cast("const char*") BytePointer attr_name,
-                                             @Cast("unsigned char") byte value);
-public static native void TFE_OpSetAttrBool(TFE_Op op, String attr_name,
-                                             @Cast("unsigned char") byte value);
-public static native void TFE_OpSetAttrType(TFE_Op op, @Cast("const char*") BytePointer attr_name,
-                                             @Cast("TF_DataType") int value);
-public static native void TFE_OpSetAttrType(TFE_Op op, String attr_name,
-                                             @Cast("TF_DataType") int value);
-// If the number of dimensions is unknown, `num_dims` must be set to
-// -1 and `dims` can be null.  If a dimension is unknown, the
-// corresponding entry in the `dims` array must be -1.
-public static native void TFE_OpSetAttrShape(TFE_Op op, @Cast("const char*") BytePointer attr_name,
-                                              @Cast("const int64_t*") LongPointer dims,
-                                              int num_dims,
-                                              TF_Status out_status);
-public static native void TFE_OpSetAttrShape(TFE_Op op, String attr_name,
-                                              @Cast("const int64_t*") LongBuffer dims,
-                                              int num_dims,
-                                              TF_Status out_status);
-public static native void TFE_OpSetAttrShape(TFE_Op op, @Cast("const char*") BytePointer attr_name,
-                                              @Cast("const int64_t*") long[] dims,
-                                              int num_dims,
-                                              TF_Status out_status);
-public static native void TFE_OpSetAttrShape(TFE_Op op, String attr_name,
-                                              @Cast("const int64_t*") LongPointer dims,
-                                              int num_dims,
-                                              TF_Status out_status);
-public static native void TFE_OpSetAttrShape(TFE_Op op, @Cast("const char*") BytePointer attr_name,
-                                              @Cast("const int64_t*") LongBuffer dims,
-                                              int num_dims,
-                                              TF_Status out_status);
-public static native void TFE_OpSetAttrShape(TFE_Op op, String attr_name,
-                                              @Cast("const int64_t*") long[] dims,
-                                              int num_dims,
-                                              TF_Status out_status);
-
-// Sets the attribute attr_name to be a function specified by 'function'.
-//
-// TODO(ashankar,iga): Add this functionality to the C API for graph
-// construction. Perhaps we want an AttrValueMap equivalent in the C API?
-public static native void TFE_OpSetAttrFunction(TFE_Op op,
-                                                 @Cast("const char*") BytePointer attr_name,
-                                                 @Const TFE_Op value);
-public static native void TFE_OpSetAttrFunction(TFE_Op op,
-                                                 String attr_name,
-                                                 @Const TFE_Op value);
-
-public static native void TFE_OpSetAttrFunctionName(TFE_Op op, @Cast("const char*") BytePointer attr_name,
-                                              @Cast("const char*") BytePointer data, @Cast("size_t") long length);
-public static native void TFE_OpSetAttrFunctionName(TFE_Op op, String attr_name,
-                                              String data, @Cast("size_t") long length);
-
-public static native void TFE_OpSetAttrTensor(TFE_Op op,
-                                               @Cast("const char*") BytePointer attr_name,
-                                               TF_Tensor tensor,
-                                               TF_Status status);
-public static native void TFE_OpSetAttrTensor(TFE_Op op,
-                                               String attr_name,
-                                               TF_Tensor tensor,
-                                               TF_Status status);
-
-public static native void TFE_OpSetAttrStringList(TFE_Op op,
-                                                   @Cast("const char*") BytePointer attr_name,
-                                                   @Cast("const void*const*") PointerPointer values,
-                                                   @Cast("const size_t*") SizeTPointer lengths,
-                                                   int num_values);
-public static native void TFE_OpSetAttrStringList(TFE_Op op,
-                                                   @Cast("const char*") BytePointer attr_name,
-                                                   @Cast("const void*const*") @ByPtrPtr Pointer values,
-                                                   @Cast("const size_t*") SizeTPointer lengths,
-                                                   int num_values);
-public static native void TFE_OpSetAttrStringList(TFE_Op op,
-                                                   String attr_name,
-                                                   @Cast("const void*const*") @ByPtrPtr Pointer values,
-                                                   @Cast("const size_t*") SizeTPointer lengths,
-                                                   int num_values);
-public static native void TFE_OpSetAttrIntList(TFE_Op op,
-                                                @Cast("const char*") BytePointer attr_name,
-                                                @Cast("const int64_t*") LongPointer values,
-                                                int num_values);
-public static native void TFE_OpSetAttrIntList(TFE_Op op,
-                                                String attr_name,
-                                                @Cast("const int64_t*") LongBuffer values,
-                                                int num_values);
-public static native void TFE_OpSetAttrIntList(TFE_Op op,
-                                                @Cast("const char*") BytePointer attr_name,
-                                                @Cast("const int64_t*") long[] values,
-                                                int num_values);
-public static native void TFE_OpSetAttrIntList(TFE_Op op,
-                                                String attr_name,
-                                                @Cast("const int64_t*") LongPointer values,
-                                                int num_values);
-public static native void TFE_OpSetAttrIntList(TFE_Op op,
-                                                @Cast("const char*") BytePointer attr_name,
-                                                @Cast("const int64_t*") LongBuffer values,
-                                                int num_values);
-public static native void TFE_OpSetAttrIntList(TFE_Op op,
-                                                String attr_name,
-                                                @Cast("const int64_t*") long[] values,
-                                                int num_values);
-public static native void TFE_OpSetAttrFloatList(TFE_Op op,
-                                                  @Cast("const char*") BytePointer attr_name,
-                                                  @Const FloatPointer values,
-                                                  int num_values);
-public static native void TFE_OpSetAttrFloatList(TFE_Op op,
-                                                  String attr_name,
-                                                  @Const FloatBuffer values,
-                                                  int num_values);
-public static native void TFE_OpSetAttrFloatList(TFE_Op op,
-                                                  @Cast("const char*") BytePointer attr_name,
-                                                  @Const float[] values,
-                                                  int num_values);
-public static native void TFE_OpSetAttrFloatList(TFE_Op op,
-                                                  String attr_name,
-                                                  @Const FloatPointer values,
-                                                  int num_values);
-public static native void TFE_OpSetAttrFloatList(TFE_Op op,
-                                                  @Cast("const char*") BytePointer attr_name,
-                                                  @Const FloatBuffer values,
-                                                  int num_values);
-public static native void TFE_OpSetAttrFloatList(TFE_Op op,
-                                                  String attr_name,
-                                                  @Const float[] values,
-                                                  int num_values);
-public static native void TFE_OpSetAttrBoolList(TFE_Op op,
-                                                 @Cast("const char*") BytePointer attr_name,
-                                                 @Cast("const unsigned char*") BytePointer values,
-                                                 int num_values);
-public static native void TFE_OpSetAttrBoolList(TFE_Op op,
-                                                 String attr_name,
-                                                 @Cast("const unsigned char*") ByteBuffer values,
-                                                 int num_values);
-public static native void TFE_OpSetAttrBoolList(TFE_Op op,
-                                                 @Cast("const char*") BytePointer attr_name,
-                                                 @Cast("const unsigned char*") byte[] values,
-                                                 int num_values);
-public static native void TFE_OpSetAttrBoolList(TFE_Op op,
-                                                 String attr_name,
-                                                 @Cast("const unsigned char*") BytePointer values,
-                                                 int num_values);
-public static native void TFE_OpSetAttrBoolList(TFE_Op op,
-                                                 @Cast("const char*") BytePointer attr_name,
-                                                 @Cast("const unsigned char*") ByteBuffer values,
-                                                 int num_values);
-public static native void TFE_OpSetAttrBoolList(TFE_Op op,
-                                                 String attr_name,
-                                                 @Cast("const unsigned char*") byte[] values,
-                                                 int num_values);
-public static native void TFE_OpSetAttrTypeList(TFE_Op op,
-                                                 @Cast("const char*") BytePointer attr_name,
-                                                 @Cast("const TF_DataType*") IntPointer values,
-                                                 int num_values);
-public static native void TFE_OpSetAttrTypeList(TFE_Op op,
-                                                 String attr_name,
-                                                 @Cast("const TF_DataType*") IntBuffer values,
-                                                 int num_values);
-public static native void TFE_OpSetAttrTypeList(TFE_Op op,
-                                                 @Cast("const char*") BytePointer attr_name,
-                                                 @Cast("const TF_DataType*") int[] values,
-                                                 int num_values);
-public static native void TFE_OpSetAttrTypeList(TFE_Op op,
-                                                 String attr_name,
-                                                 @Cast("const TF_DataType*") IntPointer values,
-                                                 int num_values);
-public static native void TFE_OpSetAttrTypeList(TFE_Op op,
-                                                 @Cast("const char*") BytePointer attr_name,
-                                                 @Cast("const TF_DataType*") IntBuffer values,
-                                                 int num_values);
-public static native void TFE_OpSetAttrTypeList(TFE_Op op,
-                                                 String attr_name,
-                                                 @Cast("const TF_DataType*") int[] values,
-                                                 int num_values);
-public static native void TFE_OpSetAttrShapeList(
-    TFE_Op op, @Cast("const char*") BytePointer attr_name, @Cast("const int64_t**") PointerPointer dims,
-    @Const IntPointer num_dims, int num_values, TF_Status out_status);
-public static native void TFE_OpSetAttrShapeList(
-    TFE_Op op, @Cast("const char*") BytePointer attr_name, @Cast("const int64_t**") @ByPtrPtr LongPointer dims,
-    @Const IntPointer num_dims, int num_values, TF_Status out_status);
-public static native void TFE_OpSetAttrShapeList(
-    TFE_Op op, String attr_name, @Cast("const int64_t**") @ByPtrPtr LongBuffer dims,
-    @Const IntBuffer num_dims, int num_values, TF_Status out_status);
-public static native void TFE_OpSetAttrShapeList(
-    TFE_Op op, @Cast("const char*") BytePointer attr_name, @Cast("const int64_t**") @ByPtrPtr long[] dims,
-    @Const int[] num_dims, int num_values, TF_Status out_status);
-public static native void TFE_OpSetAttrShapeList(
-    TFE_Op op, String attr_name, @Cast("const int64_t**") @ByPtrPtr LongPointer dims,
-    @Const IntPointer num_dims, int num_values, TF_Status out_status);
-public static native void TFE_OpSetAttrShapeList(
-    TFE_Op op, @Cast("const char*") BytePointer attr_name, @Cast("const int64_t**") @ByPtrPtr LongBuffer dims,
-    @Const IntBuffer num_dims, int num_values, TF_Status out_status);
-public static native void TFE_OpSetAttrShapeList(
-    TFE_Op op, String attr_name, @Cast("const int64_t**") @ByPtrPtr long[] dims,
-    @Const int[] num_dims, int num_values, TF_Status out_status);
-public static native void TFE_OpSetAttrFunctionList(TFE_Op op,
-                                                     @Cast("const char*") BytePointer attr_name,
-                                                     @Cast("const TFE_Op**") PointerPointer value,
-                                                     int num_values);
-public static native void TFE_OpSetAttrFunctionList(TFE_Op op,
-                                                     @Cast("const char*") BytePointer attr_name,
-                                                     @Const @ByPtrPtr TFE_Op value,
-                                                     int num_values);
-public static native void TFE_OpSetAttrFunctionList(TFE_Op op,
-                                                     String attr_name,
-                                                     @Const @ByPtrPtr TFE_Op value,
-                                                     int num_values);
-
-// Returns the length (number of tensors) of the input argument `input_name`
-// found in the provided `op`.
-public static native int TFE_OpGetInputLength(TFE_Op op,
-                                               @Cast("const char*") BytePointer input_name,
-                                               TF_Status status);
-public static native int TFE_OpGetInputLength(TFE_Op op,
-                                               String input_name,
-                                               TF_Status status);
-
-// Returns the length (number of tensors) of the output argument `output_name`
-// found in the provided `op`.
-public static native int TFE_OpGetOutputLength(TFE_Op op,
-                                                @Cast("const char*") BytePointer output_name,
-                                                TF_Status status);
-public static native int TFE_OpGetOutputLength(TFE_Op op,
-                                                String output_name,
-                                                TF_Status status);
-
-// Execute the operation defined by 'op' and return handles to computed
-// tensors in `retvals`.
-//
-// 'retvals' must point to a pre-allocated array of TFE_TensorHandle* and
-// '*num_retvals' should be set to the size of this array. It is an error if
-// the size of 'retvals' is less than the number of outputs. This call sets
-// *num_retvals to the number of outputs.
-//
-// If async execution is enabled, the call may simply enqueue the execution
-// and return "non-ready" handles in `retvals`. Note that any handles contained
-// in 'op' should not be mutated till the kernel execution actually finishes.
-//
-// For sync execution, if any of the inputs to `op` are not ready, this call
-// will block till they become ready and then return when the kernel execution
-// is done.
-// TODO(agarwal): change num_retvals to int from int*.
-public static native void TFE_Execute(TFE_Op op, @Cast("TFE_TensorHandle**") PointerPointer retvals,
-                                       IntPointer num_retvals, TF_Status status);
-public static native void TFE_Execute(TFE_Op op, @ByPtrPtr TFE_TensorHandle retvals,
-                                       IntPointer num_retvals, TF_Status status);
-public static native void TFE_Execute(TFE_Op op, @ByPtrPtr TFE_TensorHandle retvals,
-                                       IntBuffer num_retvals, TF_Status status);
-public static native void TFE_Execute(TFE_Op op, @ByPtrPtr TFE_TensorHandle retvals,
-                                       int[] num_retvals, TF_Status status);
-
-// Add a function (serialized FunctionDef protocol buffer) to ctx so
-// that it can be invoked using TFE_Execute.
-public static native void TFE_ContextAddFunctionDef(
-    TFE_Context ctx, @Cast("const char*") BytePointer serialized_function_def, @Cast("size_t") long size,
-    TF_Status status);
-public static native void TFE_ContextAddFunctionDef(
-    TFE_Context ctx, String serialized_function_def, @Cast("size_t") long size,
-    TF_Status status);
-
-// Adds a function (created from TF_GraphToFunction or
-// TF_FunctionImportFunctionDef) to the context, allowing it to be executed with
-// TFE_Execute by creating an op with the same name as the function.
-public static native void TFE_ContextAddFunction(TFE_Context ctx,
-                                                  TF_Function function,
-                                                  TF_Status status);
-
-// Removes a function from the context. Once removed, you can no longer
-// TFE_Execute it or TFE_Execute any TFE_Op which has it as an attribute or any
-// other function which calls it as an attribute.
-public static native void TFE_ContextRemoveFunction(TFE_Context ctx,
-                                                     @Cast("const char*") BytePointer name,
-                                                     TF_Status status);
-public static native void TFE_ContextRemoveFunction(TFE_Context ctx,
-                                                     String name,
-                                                     TF_Status status);
-
-// Checks whether a function is registered under `name`.
-public static native @Cast("unsigned char") byte TFE_ContextHasFunction(TFE_Context ctx,
-                                                    @Cast("const char*") BytePointer name);
-public static native @Cast("unsigned char") byte TFE_ContextHasFunction(TFE_Context ctx,
-                                                    String name);
-
-// Enables tracing of RunMetadata on the ops executed from this context.
-public static native void TFE_ContextEnableRunMetadata(TFE_Context ctx);
-
-// Disables tracing of RunMetadata on the ops executed from this context.
-public static native void TFE_ContextDisableRunMetadata(TFE_Context ctx);
-
-// Populates the passed-in buffer with a serialized RunMetadata protocol buffer
-// containing any run metadata information accumulated so far and clears this
-// information.
-// If async mode is enabled, this call blocks till all currently pending ops are
-// done.
-public static native void TFE_ContextExportRunMetadata(TFE_Context ctx,
-                                                        TF_Buffer buf,
-                                                        TF_Status status);
-
-// Some TF ops need a step container to be set to limit the lifetime of some
-// resources (mostly TensorArray and Stack, used in while loop gradients in
-// graph mode). Calling this on a context tells it to start a step.
-public static native void TFE_ContextStartStep(TFE_Context ctx);
-
-// Ends a step. When there is no active step (that is, every started step has
-// been ended) step containers will be cleared. Note: it is not safe to call
-// TFE_ContextEndStep while ops which rely on the step container may be running.
-public static native void TFE_ContextEndStep(TFE_Context ctx);
-
-// #ifdef __cplusplus
-// Targeting ../Tensor.java
-
-
-  // namespace tensorflow
-
-
-// #endif
-
-// #endif  // TENSORFLOW_C_EAGER_C_API_H_
-
-
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/audio/AudioSpectrogram.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/audio/AudioSpectrogram.java
deleted file mode 100644
index 1e88316cdd1..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/audio/AudioSpectrogram.java
+++ /dev/null
@@ -1,140 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.audio;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-
-/**
- * Produces a visualization of audio data over time.
- * <p>
- * Spectrograms are a standard way of representing audio information as a series of
- * slices of frequency information, one slice for each window of time. By joining
- * these together into a sequence, they form a distinctive fingerprint of the sound
- * over time.
- * <p>
- * This op expects to receive audio data as an input, stored as floats in the range
- * -1 to 1, together with a window width in samples, and a stride specifying how
- * far to move the window between slices. From this it generates a three
- * dimensional output. The first dimension is for the channels in the input, so a
- * stereo audio input would have two here for example. The second dimension is time,
- * with successive frequency slices. The third dimension has an amplitude value for
- * each frequency during that time slice.
- * <p>
- * This means the layout when converted and saved as an image is rotated 90 degrees
- * clockwise from a typical spectrogram. Time is descending down the Y axis, and
- * the frequency decreases from left to right.
- * <p>
- * Each value in the result represents the square root of the sum of the real and
- * imaginary parts of an FFT on the current window of samples. In this way, the
- * lowest dimension represents the power of each frequency in the current window,
- * and adjacent windows are concatenated in the next dimension.
- * <p>
- * To get a more intuitive and visual look at what this operation does, you can run
- * tensorflow/examples/wav_to_spectrogram to read in an audio file and save out the
- * resulting spectrogram as a PNG image.
- */
-@Operator(group = "audio")
-public final class AudioSpectrogram extends RawOp implements Operand<TFloat32> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.audio.AudioSpectrogram}
-   */
-  public static class Options {
-    
-    /**
-     * @param magnitudeSquared Whether to return the squared magnitude or just the
-     * magnitude. Using squared magnitude can avoid extra calculations.
-     */
-    public Options magnitudeSquared(Boolean magnitudeSquared) {
-      this.magnitudeSquared = magnitudeSquared;
-      return this;
-    }
-    
-    private Boolean magnitudeSquared;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new AudioSpectrogram operation.
-   * 
-   * @param scope current scope
-   * @param input Float representation of audio data.
-   * @param windowSize How wide the input window is in samples. For the highest efficiency
-   * this should be a power of two, but other values are accepted.
-   * @param stride How widely apart the center of adjacent sample windows should be.
-   * @param options carries optional attributes values
-   * @return a new instance of AudioSpectrogram
-   */
-  @Endpoint(describeByClass = true)
-  public static AudioSpectrogram create(Scope scope, Operand<TFloat32> input, Long windowSize, Long stride, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("AudioSpectrogram", scope.makeOpName("AudioSpectrogram"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("window_size", windowSize);
-    opBuilder.setAttr("stride", stride);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.magnitudeSquared != null) {
-          opBuilder.setAttr("magnitude_squared", opts.magnitudeSquared);
-        }
-      }
-    }
-    return new AudioSpectrogram(opBuilder.build());
-  }
-  
-  /**
-   * @param magnitudeSquared Whether to return the squared magnitude or just the
-   * magnitude. Using squared magnitude can avoid extra calculations.
-   */
-  public static Options magnitudeSquared(Boolean magnitudeSquared) {
-    return new Options().magnitudeSquared(magnitudeSquared);
-  }
-  
-  /**
-   * 3D representation of the audio frequencies as an image.
-   */
-  public Output<TFloat32> spectrogram() {
-    return spectrogram;
-  }
-  
-  @Override
-  public Output<TFloat32> asOutput(Scope scope) {
-    return spectrogram;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "AudioSpectrogram";
-  
-  private Output<TFloat32> spectrogram;
-  
-  private AudioSpectrogram(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    spectrogram = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/audio/DecodeWav.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/audio/DecodeWav.java
deleted file mode 100644
index d4ed42012af..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/audio/DecodeWav.java
+++ /dev/null
@@ -1,147 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.audio;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.TString;
-
-/**
- * Decode a 16-bit PCM WAV file to a float tensor.
- * <p>
- * The -32768 to 32767 signed 16-bit values will be scaled to -1.0 to 1.0 in float.
- * <p>
- * When desired_channels is set, if the input contains fewer channels than this
- * then the last channel will be duplicated to give the requested number, else if
- * the input has more channels than requested then the additional channels will be
- * ignored.
- * <p>
- * If desired_samples is set, then the audio will be cropped or padded with zeroes
- * to the requested length.
- * <p>
- * The first output contains a Tensor with the content of the audio samples. The
- * lowest dimension will be the number of channels, and the second will be the
- * number of samples. For example, a ten-sample-long stereo WAV file should give an
- * output shape of [10, 2].
- */
-@Operator(group = "audio")
-public final class DecodeWav extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.audio.DecodeWav}
-   */
-  public static class Options {
-    
-    /**
-     * @param desiredChannels Number of sample channels wanted.
-     */
-    public Options desiredChannels(Long desiredChannels) {
-      this.desiredChannels = desiredChannels;
-      return this;
-    }
-    
-    /**
-     * @param desiredSamples Length of audio requested.
-     */
-    public Options desiredSamples(Long desiredSamples) {
-      this.desiredSamples = desiredSamples;
-      return this;
-    }
-    
-    private Long desiredChannels;
-    private Long desiredSamples;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new DecodeWav operation.
-   * 
-   * @param scope current scope
-   * @param contents The WAV-encoded audio, usually from a file.
-   * @param options carries optional attributes values
-   * @return a new instance of DecodeWav
-   */
-  @Endpoint(describeByClass = true)
-  public static DecodeWav create(Scope scope, Operand<TString> contents, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("DecodeWav", scope.makeOpName("DecodeWav"));
-    opBuilder.addInput(contents.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.desiredChannels != null) {
-          opBuilder.setAttr("desired_channels", opts.desiredChannels);
-        }
-        if (opts.desiredSamples != null) {
-          opBuilder.setAttr("desired_samples", opts.desiredSamples);
-        }
-      }
-    }
-    return new DecodeWav(opBuilder.build());
-  }
-  
-  /**
-   * @param desiredChannels Number of sample channels wanted.
-   */
-  public static Options desiredChannels(Long desiredChannels) {
-    return new Options().desiredChannels(desiredChannels);
-  }
-  
-  /**
-   * @param desiredSamples Length of audio requested.
-   */
-  public static Options desiredSamples(Long desiredSamples) {
-    return new Options().desiredSamples(desiredSamples);
-  }
-  
-  /**
-   * 2-D with shape `[length, channels]`.
-   */
-  public Output<TFloat32> audio() {
-    return audio;
-  }
-  
-  /**
-   * Scalar holding the sample rate found in the WAV header.
-   */
-  public Output<TInt32> sampleRate() {
-    return sampleRate;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "DecodeWav";
-  
-  private Output<TFloat32> audio;
-  private Output<TInt32> sampleRate;
-  
-  private DecodeWav(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    audio = operation.output(outputIdx++);
-    sampleRate = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/audio/EncodeWav.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/audio/EncodeWav.java
deleted file mode 100644
index 3a7fc674e51..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/audio/EncodeWav.java
+++ /dev/null
@@ -1,85 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.audio;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.TString;
-
-/**
- * Encode audio data using the WAV file format.
- * <p>
- * This operation will generate a string suitable to be saved out to create a .wav
- * audio file. It will be encoded in the 16-bit PCM format. It takes in float
- * values in the range -1.0f to 1.0f, and any outside that value will be clamped to
- * that range.
- * <p>
- * `audio` is a 2-D float Tensor of shape `[length, channels]`.
- * `sample_rate` is a scalar Tensor holding the rate to use (e.g. 44100).
- */
-@Operator(group = "audio")
-public final class EncodeWav extends RawOp implements Operand<TString> {
-  
-  /**
-   * Factory method to create a class wrapping a new EncodeWav operation.
-   * 
-   * @param scope current scope
-   * @param audio 2-D with shape `[length, channels]`.
-   * @param sampleRate Scalar containing the sample frequency.
-   * @return a new instance of EncodeWav
-   */
-  @Endpoint(describeByClass = true)
-  public static EncodeWav create(Scope scope, Operand<TFloat32> audio, Operand<TInt32> sampleRate) {
-    OperationBuilder opBuilder = scope.env().opBuilder("EncodeWav", scope.makeOpName("EncodeWav"));
-    opBuilder.addInput(audio.asOutput(scope));
-    opBuilder.addInput(sampleRate.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new EncodeWav(opBuilder.build());
-  }
-  
-  /**
-   * 0-D. WAV-encoded file contents.
-   */
-  public Output<TString> contents() {
-    return contents;
-  }
-  
-  @Override
-  public Output<TString> asOutput(Scope scope) {
-    return contents;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "EncodeWav";
-  
-  private Output<TString> contents;
-  
-  private EncodeWav(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    contents = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/audio/Mfcc.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/audio/Mfcc.java
deleted file mode 100644
index 1bf6eaca34d..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/audio/Mfcc.java
+++ /dev/null
@@ -1,178 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.audio;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.TInt32;
-
-/**
- * Transforms a spectrogram into a form that's useful for speech recognition.
- * <p>
- * Mel Frequency Cepstral Coefficients are a way of representing audio data that's
- * been effective as an input feature for machine learning. They are created by
- * taking the spectrum of a spectrogram (a 'cepstrum'), and discarding some of the
- * higher frequencies that are less significant to the human ear. They have a long
- * history in the speech recognition world, and https://en.wikipedia.org/wiki/Mel-frequency_cepstrum
- * is a good resource to learn more.
- */
-@Operator(group = "audio")
-public final class Mfcc extends RawOp implements Operand<TFloat32> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.audio.Mfcc}
-   */
-  public static class Options {
-    
-    /**
-     * @param upperFrequencyLimit The highest frequency to use when calculating the
-     * ceptstrum.
-     */
-    public Options upperFrequencyLimit(Float upperFrequencyLimit) {
-      this.upperFrequencyLimit = upperFrequencyLimit;
-      return this;
-    }
-    
-    /**
-     * @param lowerFrequencyLimit The lowest frequency to use when calculating the
-     * ceptstrum.
-     */
-    public Options lowerFrequencyLimit(Float lowerFrequencyLimit) {
-      this.lowerFrequencyLimit = lowerFrequencyLimit;
-      return this;
-    }
-    
-    /**
-     * @param filterbankChannelCount Resolution of the Mel bank used internally.
-     */
-    public Options filterbankChannelCount(Long filterbankChannelCount) {
-      this.filterbankChannelCount = filterbankChannelCount;
-      return this;
-    }
-    
-    /**
-     * @param dctCoefficientCount How many output channels to produce per time slice.
-     */
-    public Options dctCoefficientCount(Long dctCoefficientCount) {
-      this.dctCoefficientCount = dctCoefficientCount;
-      return this;
-    }
-    
-    private Float upperFrequencyLimit;
-    private Float lowerFrequencyLimit;
-    private Long filterbankChannelCount;
-    private Long dctCoefficientCount;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new Mfcc operation.
-   * 
-   * @param scope current scope
-   * @param spectrogram Typically produced by the Spectrogram op, with magnitude_squared
-   * set to true.
-   * @param sampleRate How many samples per second the source audio used.
-   * @param options carries optional attributes values
-   * @return a new instance of Mfcc
-   */
-  @Endpoint(describeByClass = true)
-  public static Mfcc create(Scope scope, Operand<TFloat32> spectrogram, Operand<TInt32> sampleRate, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Mfcc", scope.makeOpName("Mfcc"));
-    opBuilder.addInput(spectrogram.asOutput(scope));
-    opBuilder.addInput(sampleRate.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.upperFrequencyLimit != null) {
-          opBuilder.setAttr("upper_frequency_limit", opts.upperFrequencyLimit);
-        }
-        if (opts.lowerFrequencyLimit != null) {
-          opBuilder.setAttr("lower_frequency_limit", opts.lowerFrequencyLimit);
-        }
-        if (opts.filterbankChannelCount != null) {
-          opBuilder.setAttr("filterbank_channel_count", opts.filterbankChannelCount);
-        }
-        if (opts.dctCoefficientCount != null) {
-          opBuilder.setAttr("dct_coefficient_count", opts.dctCoefficientCount);
-        }
-      }
-    }
-    return new Mfcc(opBuilder.build());
-  }
-  
-  /**
-   * @param upperFrequencyLimit The highest frequency to use when calculating the
-   * ceptstrum.
-   */
-  public static Options upperFrequencyLimit(Float upperFrequencyLimit) {
-    return new Options().upperFrequencyLimit(upperFrequencyLimit);
-  }
-  
-  /**
-   * @param lowerFrequencyLimit The lowest frequency to use when calculating the
-   * ceptstrum.
-   */
-  public static Options lowerFrequencyLimit(Float lowerFrequencyLimit) {
-    return new Options().lowerFrequencyLimit(lowerFrequencyLimit);
-  }
-  
-  /**
-   * @param filterbankChannelCount Resolution of the Mel bank used internally.
-   */
-  public static Options filterbankChannelCount(Long filterbankChannelCount) {
-    return new Options().filterbankChannelCount(filterbankChannelCount);
-  }
-  
-  /**
-   * @param dctCoefficientCount How many output channels to produce per time slice.
-   */
-  public static Options dctCoefficientCount(Long dctCoefficientCount) {
-    return new Options().dctCoefficientCount(dctCoefficientCount);
-  }
-  
-  /**
-   */
-  public Output<TFloat32> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<TFloat32> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Mfcc";
-  
-  private Output<TFloat32> output;
-  
-  private Mfcc(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/bitwise/BitwiseAnd.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/bitwise/BitwiseAnd.java
deleted file mode 100644
index 99ef38701b5..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/bitwise/BitwiseAnd.java
+++ /dev/null
@@ -1,96 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.bitwise;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Elementwise computes the bitwise AND of `x` and `y`.
- * <p>
- * The result will have those bits set, that are set in both `x` and `y`. The
- * computation is performed on the underlying representations of `x` and `y`.
- * <p>
- * For example:
- * <pre>{@code
- * import tensorflow as tf
- * from tensorflow.python.ops import bitwise_ops
- * dtype_list = [tf.int8, tf.int16, tf.int32, tf.int64,
- *               tf.uint8, tf.uint16, tf.uint32, tf.uint64]
- * 
- * for dtype in dtype_list:
- *   lhs = tf.constant([0, 5, 3, 14], dtype=dtype)
- *   rhs = tf.constant([5, 0, 7, 11], dtype=dtype)
- *   exp = tf.constant([0, 0, 3, 10], dtype=tf.float32)
- * 
- *   res = bitwise_ops.bitwise_and(lhs, rhs)
- *   tf.assert_equal(tf.cast(res, tf.float32), exp) # TRUE
- * }</pre>
- * 
- * 
- * @param <T> data type for {@code z()} output
- */
-@Operator(group = "bitwise")
-public final class BitwiseAnd<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new BitwiseAnd operation.
-   * 
-   * @param scope current scope
-   * @param x 
-   * @param y 
-   * @return a new instance of BitwiseAnd
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> BitwiseAnd<T> create(Scope scope, Operand<T> x, Operand<T> y) {
-    OperationBuilder opBuilder = scope.env().opBuilder("BitwiseAnd", scope.makeOpName("BitwiseAnd"));
-    opBuilder.addInput(x.asOutput(scope));
-    opBuilder.addInput(y.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new BitwiseAnd<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> z() {
-    return z;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return z;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "BitwiseAnd";
-  
-  private Output<T> z;
-  
-  private BitwiseAnd(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    z = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/bitwise/BitwiseOr.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/bitwise/BitwiseOr.java
deleted file mode 100644
index 5532fc43f65..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/bitwise/BitwiseOr.java
+++ /dev/null
@@ -1,96 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.bitwise;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Elementwise computes the bitwise OR of `x` and `y`.
- * <p>
- * The result will have those bits set, that are set in `x`, `y` or both. The
- * computation is performed on the underlying representations of `x` and `y`.
- * <p>
- * For example:
- * <pre>{@code
- * import tensorflow as tf
- * from tensorflow.python.ops import bitwise_ops
- * dtype_list = [tf.int8, tf.int16, tf.int32, tf.int64,
- *               tf.uint8, tf.uint16, tf.uint32, tf.uint64]
- * 
- * for dtype in dtype_list:
- *   lhs = tf.constant([0, 5, 3, 14], dtype=dtype)
- *   rhs = tf.constant([5, 0, 7, 11], dtype=dtype)
- *   exp = tf.constant([5, 5, 7, 15], dtype=tf.float32)
- * 
- *   res = bitwise_ops.bitwise_or(lhs, rhs)
- *   tf.assert_equal(tf.cast(res,  tf.float32), exp)  # TRUE
- * }</pre>
- * 
- * 
- * @param <T> data type for {@code z()} output
- */
-@Operator(group = "bitwise")
-public final class BitwiseOr<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new BitwiseOr operation.
-   * 
-   * @param scope current scope
-   * @param x 
-   * @param y 
-   * @return a new instance of BitwiseOr
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> BitwiseOr<T> create(Scope scope, Operand<T> x, Operand<T> y) {
-    OperationBuilder opBuilder = scope.env().opBuilder("BitwiseOr", scope.makeOpName("BitwiseOr"));
-    opBuilder.addInput(x.asOutput(scope));
-    opBuilder.addInput(y.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new BitwiseOr<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> z() {
-    return z;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return z;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "BitwiseOr";
-  
-  private Output<T> z;
-  
-  private BitwiseOr(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    z = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/bitwise/BitwiseXor.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/bitwise/BitwiseXor.java
deleted file mode 100644
index 624cd8a1df3..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/bitwise/BitwiseXor.java
+++ /dev/null
@@ -1,96 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.bitwise;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Elementwise computes the bitwise XOR of `x` and `y`.
- * <p>
- * The result will have those bits set, that are different in `x` and `y`. The
- * computation is performed on the underlying representations of `x` and `y`.
- * <p>
- * For example:
- * <pre>{@code
- * import tensorflow as tf
- * from tensorflow.python.ops import bitwise_ops
- * dtype_list = [tf.int8, tf.int16, tf.int32, tf.int64,
- *               tf.uint8, tf.uint16, tf.uint32, tf.uint64]
- * 
- * for dtype in dtype_list:
- *   lhs = tf.constant([0, 5, 3, 14], dtype=dtype)
- *   rhs = tf.constant([5, 0, 7, 11], dtype=dtype)
- *   exp = tf.constant([5, 5, 4, 5],  dtype=tf.float32)
- * 
- *   res = bitwise_ops.bitwise_xor(lhs, rhs)
- *   tf.assert_equal(tf.cast(res, tf.float32), exp) # TRUE
- * }</pre>
- * 
- * 
- * @param <T> data type for {@code z()} output
- */
-@Operator(group = "bitwise")
-public final class BitwiseXor<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new BitwiseXor operation.
-   * 
-   * @param scope current scope
-   * @param x 
-   * @param y 
-   * @return a new instance of BitwiseXor
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> BitwiseXor<T> create(Scope scope, Operand<T> x, Operand<T> y) {
-    OperationBuilder opBuilder = scope.env().opBuilder("BitwiseXor", scope.makeOpName("BitwiseXor"));
-    opBuilder.addInput(x.asOutput(scope));
-    opBuilder.addInput(y.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new BitwiseXor<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> z() {
-    return z;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return z;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "BitwiseXor";
-  
-  private Output<T> z;
-  
-  private BitwiseXor(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    z = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/bitwise/Invert.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/bitwise/Invert.java
deleted file mode 100644
index c0f882d0791..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/bitwise/Invert.java
+++ /dev/null
@@ -1,115 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.bitwise;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Invert (flip) each bit of supported types; for example, type `uint8` value 01010101 becomes 10101010.
- * <p>
- * Flip each bit of supported types.  For example, type `int8` (decimal 2) binary 00000010 becomes (decimal -3) binary 11111101.
- * This operation is performed on each element of the tensor argument `x`.
- * <p>
- * Example:
- * <pre>{@code
- * import tensorflow as tf
- * from tensorflow.python.ops import bitwise_ops
- * 
- * # flip 2 (00000010) to -3 (11111101)
- * tf.assert_equal(-3, bitwise_ops.invert(2))
- * 
- * dtype_list = [dtypes.int8, dtypes.int16, dtypes.int32, dtypes.int64,
- *               dtypes.uint8, dtypes.uint16, dtypes.uint32, dtypes.uint64]
- * 
- * inputs = [0, 5, 3, 14]
- * for dtype in dtype_list:
- *   # Because of issues with negative numbers, let's test this indirectly.
- *   # 1. invert(a) and a = 0
- *   # 2. invert(a) or a = invert(0)
- *   input_tensor = tf.constant([0, 5, 3, 14], dtype=dtype)
- *   not_a_and_a, not_a_or_a, not_0 = [bitwise_ops.bitwise_and(
- *                                       input_tensor, bitwise_ops.invert(input_tensor)),
- *                                     bitwise_ops.bitwise_or(
- *                                       input_tensor, bitwise_ops.invert(input_tensor)),
- *                                     bitwise_ops.invert(
- *                                       tf.constant(0, dtype=dtype))]
- * 
- *   expected = tf.constant([0, 0, 0, 0], dtype=tf.float32)
- *   tf.assert_equal(tf.cast(not_a_and_a, tf.float32), expected)
- * 
- *   expected = tf.cast([not_0] * 4, tf.float32)
- *   tf.assert_equal(tf.cast(not_a_or_a, tf.float32), expected)
- * 
- *   # For unsigned dtypes let's also check the result directly.
- *   if dtype.is_unsigned:
- *     inverted = bitwise_ops.invert(input_tensor)
- *     expected = tf.constant([dtype.max - x for x in inputs], dtype=tf.float32)
- *     tf.assert_equal(tf.cast(inverted, tf.float32), tf.cast(expected, tf.float32))
- * }</pre>
- * 
- * 
- * @param <T> data type for {@code y()} output
- */
-@Operator(group = "bitwise")
-public final class Invert<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new Invert operation.
-   * 
-   * @param scope current scope
-   * @param x 
-   * @return a new instance of Invert
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> Invert<T> create(Scope scope, Operand<T> x) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Invert", scope.makeOpName("Invert"));
-    opBuilder.addInput(x.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new Invert<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> y() {
-    return y;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return y;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Invert";
-  
-  private Output<T> y;
-  
-  private Invert(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    y = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/bitwise/LeftShift.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/bitwise/LeftShift.java
deleted file mode 100644
index 3de00d2b037..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/bitwise/LeftShift.java
+++ /dev/null
@@ -1,107 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.bitwise;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Elementwise computes the bitwise left-shift of `x` and `y`.
- * <p>
- * If `y` is negative, or greater than or equal to the width of `x` in bits the
- * result is implementation defined.
- * <p>
- * Example:
- * <pre>{@code
- * import tensorflow as tf
- * from tensorflow.python.ops import bitwise_ops
- * import numpy as np
- * dtype_list = [tf.int8, tf.int16, tf.int32, tf.int64]
- * 
- * for dtype in dtype_list:
- *   lhs = tf.constant([-1, -5, -3, -14], dtype=dtype)
- *   rhs = tf.constant([5, 0, 7, 11], dtype=dtype)
- * 
- *   left_shift_result = bitwise_ops.left_shift(lhs, rhs)
- * 
- *   print(left_shift_result)
- * 
- * # This will print:
- * # tf.Tensor([ -32   -5 -128    0], shape=(4,), dtype=int8)
- * # tf.Tensor([   -32     -5   -384 -28672], shape=(4,), dtype=int16)
- * # tf.Tensor([   -32     -5   -384 -28672], shape=(4,), dtype=int32)
- * # tf.Tensor([   -32     -5   -384 -28672], shape=(4,), dtype=int64)
- * 
- * lhs = np.array([-2, 64, 101, 32], dtype=np.int8)
- * rhs = np.array([-1, -5, -3, -14], dtype=np.int8)
- * bitwise_ops.left_shift(lhs, rhs)
- * # <tf.Tensor: shape=(4,), dtype=int8, numpy=array([ -2,  64, 101,  32], dtype=int8)>
- * }</pre>
- * 
- * 
- * @param <T> data type for {@code z()} output
- */
-@Operator(group = "bitwise")
-public final class LeftShift<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new LeftShift operation.
-   * 
-   * @param scope current scope
-   * @param x 
-   * @param y 
-   * @return a new instance of LeftShift
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> LeftShift<T> create(Scope scope, Operand<T> x, Operand<T> y) {
-    OperationBuilder opBuilder = scope.env().opBuilder("LeftShift", scope.makeOpName("LeftShift"));
-    opBuilder.addInput(x.asOutput(scope));
-    opBuilder.addInput(y.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new LeftShift<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> z() {
-    return z;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return z;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "LeftShift";
-  
-  private Output<T> z;
-  
-  private LeftShift(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    z = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/bitwise/RightShift.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/bitwise/RightShift.java
deleted file mode 100644
index 00117180819..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/bitwise/RightShift.java
+++ /dev/null
@@ -1,110 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.bitwise;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Elementwise computes the bitwise right-shift of `x` and `y`.
- * <p>
- * Performs a logical shift for unsigned integer types, and an arithmetic shift
- * for signed integer types.
- * <p>
- * If `y` is negative, or greater than or equal to than the width of `x` in bits
- * the result is implementation defined.
- * <p>
- * Example:
- * <pre>{@code
- * import tensorflow as tf
- * from tensorflow.python.ops import bitwise_ops
- * import numpy as np
- * dtype_list = [tf.int8, tf.int16, tf.int32, tf.int64]
- * 
- * for dtype in dtype_list:
- *   lhs = tf.constant([-1, -5, -3, -14], dtype=dtype)
- *   rhs = tf.constant([5, 0, 7, 11], dtype=dtype)
- * 
- *   right_shift_result = bitwise_ops.right_shift(lhs, rhs)
- * 
- *   print(right_shift_result)
- * 
- * # This will print:
- * # tf.Tensor([-1 -5 -1 -1], shape=(4,), dtype=int8)
- * # tf.Tensor([-1 -5 -1 -1], shape=(4,), dtype=int16)
- * # tf.Tensor([-1 -5 -1 -1], shape=(4,), dtype=int32)
- * # tf.Tensor([-1 -5 -1 -1], shape=(4,), dtype=int64)
- * 
- * lhs = np.array([-2, 64, 101, 32], dtype=np.int8)
- * rhs = np.array([-1, -5, -3, -14], dtype=np.int8)
- * bitwise_ops.right_shift(lhs, rhs)
- * # <tf.Tensor: shape=(4,), dtype=int8, numpy=array([ -2,  64, 101,  32], dtype=int8)>
- * }</pre>
- * 
- * 
- * @param <T> data type for {@code z()} output
- */
-@Operator(group = "bitwise")
-public final class RightShift<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new RightShift operation.
-   * 
-   * @param scope current scope
-   * @param x 
-   * @param y 
-   * @return a new instance of RightShift
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> RightShift<T> create(Scope scope, Operand<T> x, Operand<T> y) {
-    OperationBuilder opBuilder = scope.env().opBuilder("RightShift", scope.makeOpName("RightShift"));
-    opBuilder.addInput(x.asOutput(scope));
-    opBuilder.addInput(y.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new RightShift<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> z() {
-    return z;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return z;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "RightShift";
-  
-  private Output<T> z;
-  
-  private RightShift(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    z = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/cluster/KMC2ChainInitialization.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/cluster/KMC2ChainInitialization.java
deleted file mode 100644
index 4b8a4723e1a..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/cluster/KMC2ChainInitialization.java
+++ /dev/null
@@ -1,81 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.cluster;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.TInt64;
-
-/**
- * Returns the index of a data point that should be added to the seed set.
- * <p>
- * Entries in distances are assumed to be squared distances of candidate points to
- * the already sampled centers in the seed set. The op constructs one Markov chain
- * of the k-MC^2 algorithm and returns the index of one candidate point to be added
- * as an additional cluster center.
- */
-public final class KMC2ChainInitialization extends RawOp implements Operand<TInt64> {
-  
-  /**
-   * Factory method to create a class wrapping a new KMC2ChainInitialization operation.
-   * 
-   * @param scope current scope
-   * @param distances Vector with squared distances to the closest previously sampled cluster center
-   * for each candidate point.
-   * @param seed Scalar. Seed for initializing the random number generator.
-   * @return a new instance of KMC2ChainInitialization
-   */
-  @Endpoint(describeByClass = true)
-  public static KMC2ChainInitialization create(Scope scope, Operand<TFloat32> distances, Operand<TInt64> seed) {
-    OperationBuilder opBuilder = scope.env().opBuilder("KMC2ChainInitialization", scope.makeOpName("KMC2ChainInitialization"));
-    opBuilder.addInput(distances.asOutput(scope));
-    opBuilder.addInput(seed.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new KMC2ChainInitialization(opBuilder.build());
-  }
-  
-  /**
-   * Scalar with the index of the sampled point.
-   */
-  public Output<TInt64> index() {
-    return index;
-  }
-  
-  @Override
-  public Output<TInt64> asOutput(Scope scope) {
-    return index;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "KMC2ChainInitialization";
-  
-  private Output<TInt64> index;
-  
-  private KMC2ChainInitialization(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    index = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/cluster/KmeansPlusPlusInitialization.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/cluster/KmeansPlusPlusInitialization.java
deleted file mode 100644
index 7f142468754..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/cluster/KmeansPlusPlusInitialization.java
+++ /dev/null
@@ -1,87 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.cluster;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.TInt64;
-
-/**
- * Selects num_to_sample rows of input using the KMeans++ criterion.
- * <p>
- * Rows of points are assumed to be input points. One row is selected at random.
- * Subsequent rows are sampled with probability proportional to the squared L2
- * distance from the nearest row selected thus far till num_to_sample rows have
- * been sampled.
- */
-public final class KmeansPlusPlusInitialization extends RawOp implements Operand<TFloat32> {
-  
-  /**
-   * Factory method to create a class wrapping a new KmeansPlusPlusInitialization operation.
-   * 
-   * @param scope current scope
-   * @param points Matrix of shape (n, d). Rows are assumed to be input points.
-   * @param numToSample Scalar. The number of rows to sample. This value must not be larger than n.
-   * @param seed Scalar. Seed for initializing the random number generator.
-   * @param numRetriesPerSample Scalar. For each row that is sampled, this parameter
-   * specifies the number of additional points to draw from the current
-   * distribution before selecting the best. If a negative value is specified, a
-   * heuristic is used to sample O(log(num_to_sample)) additional points.
-   * @return a new instance of KmeansPlusPlusInitialization
-   */
-  @Endpoint(describeByClass = true)
-  public static KmeansPlusPlusInitialization create(Scope scope, Operand<TFloat32> points, Operand<TInt64> numToSample, Operand<TInt64> seed, Operand<TInt64> numRetriesPerSample) {
-    OperationBuilder opBuilder = scope.env().opBuilder("KmeansPlusPlusInitialization", scope.makeOpName("KmeansPlusPlusInitialization"));
-    opBuilder.addInput(points.asOutput(scope));
-    opBuilder.addInput(numToSample.asOutput(scope));
-    opBuilder.addInput(seed.asOutput(scope));
-    opBuilder.addInput(numRetriesPerSample.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new KmeansPlusPlusInitialization(opBuilder.build());
-  }
-  
-  /**
-   * Matrix of shape (num_to_sample, d). The sampled rows.
-   */
-  public Output<TFloat32> samples() {
-    return samples;
-  }
-  
-  @Override
-  public Output<TFloat32> asOutput(Scope scope) {
-    return samples;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "KmeansPlusPlusInitialization";
-  
-  private Output<TFloat32> samples;
-  
-  private KmeansPlusPlusInitialization(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    samples = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/collective/AllReduce.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/collective/AllReduce.java
deleted file mode 100644
index 6c832043526..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/collective/AllReduce.java
+++ /dev/null
@@ -1,166 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.collective;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Mutually reduces multiple tensors of identical type and shape.
- * 
- * @param <T> data type for {@code output()} output
- */
-public final class AllReduce<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.collective.AllReduce}
-   */
-  public static class Options {
-    
-    /**
-     * @param waitFor 
-     */
-    public Options waitFor(List<Long> waitFor) {
-      this.waitFor = waitFor;
-      return this;
-    }
-    
-    /**
-     * @param communicationHint 
-     */
-    public Options communicationHint(String communicationHint) {
-      this.communicationHint = communicationHint;
-      return this;
-    }
-    
-    /**
-     * @param timeoutSeconds 
-     */
-    public Options timeoutSeconds(Float timeoutSeconds) {
-      this.timeoutSeconds = timeoutSeconds;
-      return this;
-    }
-    
-    private List<Long> waitFor;
-    private String communicationHint;
-    private Float timeoutSeconds;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new AllReduce operation.
-   * 
-   * @param scope current scope
-   * @param input 
-   * @param groupSize 
-   * @param groupKey 
-   * @param instanceKey 
-   * @param mergeOp 
-   * @param finalOp 
-   * @param subdivOffsets 
-   * @param options carries optional attributes values
-   * @return a new instance of AllReduce
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> AllReduce<T> create(Scope scope, Operand<T> input, Long groupSize, Long groupKey, Long instanceKey, String mergeOp, String finalOp, List<Long> subdivOffsets, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("CollectiveReduce", scope.makeOpName("AllReduce"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("group_size", groupSize);
-    opBuilder.setAttr("group_key", groupKey);
-    opBuilder.setAttr("instance_key", instanceKey);
-    opBuilder.setAttr("merge_op", mergeOp);
-    opBuilder.setAttr("final_op", finalOp);
-    long[] subdivOffsetsArray = new long[subdivOffsets.size()];
-    for (int i = 0; i < subdivOffsetsArray.length; ++i) {
-      subdivOffsetsArray[i] = subdivOffsets.get(i);
-    }
-    opBuilder.setAttr("subdiv_offsets", subdivOffsetsArray);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.waitFor != null) {
-          long[] waitForArray = new long[opts.waitFor.size()];
-          for (int i = 0; i < waitForArray.length; ++i) {
-            waitForArray[i] = opts.waitFor.get(i);
-          }
-          opBuilder.setAttr("wait_for", waitForArray);
-        }
-        if (opts.communicationHint != null) {
-          opBuilder.setAttr("communication_hint", opts.communicationHint);
-        }
-        if (opts.timeoutSeconds != null) {
-          opBuilder.setAttr("timeout_seconds", opts.timeoutSeconds);
-        }
-      }
-    }
-    return new AllReduce<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param waitFor 
-   */
-  public static Options waitFor(List<Long> waitFor) {
-    return new Options().waitFor(waitFor);
-  }
-  
-  /**
-   * @param communicationHint 
-   */
-  public static Options communicationHint(String communicationHint) {
-    return new Options().communicationHint(communicationHint);
-  }
-  
-  /**
-   * @param timeoutSeconds 
-   */
-  public static Options timeoutSeconds(Float timeoutSeconds) {
-    return new Options().timeoutSeconds(timeoutSeconds);
-  }
-  
-  /**
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "CollectiveReduce";
-  
-  private Output<T> output;
-  
-  private AllReduce(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/collective/BroadcastRecv.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/collective/BroadcastRecv.java
deleted file mode 100644
index 54e5416b33b..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/collective/BroadcastRecv.java
+++ /dev/null
@@ -1,135 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.collective;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Receives a tensor value broadcast from another device.
- * 
- * @param <T> data type for {@code output()} output
- */
-public final class BroadcastRecv<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.collective.BroadcastRecv}
-   */
-  public static class Options {
-    
-    /**
-     * @param communicationHint 
-     */
-    public Options communicationHint(String communicationHint) {
-      this.communicationHint = communicationHint;
-      return this;
-    }
-    
-    /**
-     * @param timeoutSeconds 
-     */
-    public Options timeoutSeconds(Float timeoutSeconds) {
-      this.timeoutSeconds = timeoutSeconds;
-      return this;
-    }
-    
-    private String communicationHint;
-    private Float timeoutSeconds;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new BroadcastRecv operation.
-   * 
-   * @param scope current scope
-   * @param T 
-   * @param groupSize 
-   * @param groupKey 
-   * @param instanceKey 
-   * @param shape 
-   * @param options carries optional attributes values
-   * @return a new instance of BroadcastRecv
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> BroadcastRecv<T> create(Scope scope, Class<T> T, Long groupSize, Long groupKey, Long instanceKey, Shape shape, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("CollectiveBcastRecv", scope.makeOpName("BroadcastRecv"));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("T", T);
-    opBuilder.setAttr("group_size", groupSize);
-    opBuilder.setAttr("group_key", groupKey);
-    opBuilder.setAttr("instance_key", instanceKey);
-    opBuilder.setAttr("shape", shape);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.communicationHint != null) {
-          opBuilder.setAttr("communication_hint", opts.communicationHint);
-        }
-        if (opts.timeoutSeconds != null) {
-          opBuilder.setAttr("timeout_seconds", opts.timeoutSeconds);
-        }
-      }
-    }
-    return new BroadcastRecv<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param communicationHint 
-   */
-  public static Options communicationHint(String communicationHint) {
-    return new Options().communicationHint(communicationHint);
-  }
-  
-  /**
-   * @param timeoutSeconds 
-   */
-  public static Options timeoutSeconds(Float timeoutSeconds) {
-    return new Options().timeoutSeconds(timeoutSeconds);
-  }
-  
-  /**
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "CollectiveBcastRecv";
-  
-  private Output<T> output;
-  
-  private BroadcastRecv(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/collective/BroadcastSend.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/collective/BroadcastSend.java
deleted file mode 100644
index ad95ac7a58b..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/collective/BroadcastSend.java
+++ /dev/null
@@ -1,135 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.collective;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Broadcasts a tensor value to one or more other devices.
- * 
- * @param <T> data type for {@code output()} output
- */
-public final class BroadcastSend<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.collective.BroadcastSend}
-   */
-  public static class Options {
-    
-    /**
-     * @param communicationHint 
-     */
-    public Options communicationHint(String communicationHint) {
-      this.communicationHint = communicationHint;
-      return this;
-    }
-    
-    /**
-     * @param timeoutSeconds 
-     */
-    public Options timeoutSeconds(Float timeoutSeconds) {
-      this.timeoutSeconds = timeoutSeconds;
-      return this;
-    }
-    
-    private String communicationHint;
-    private Float timeoutSeconds;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new BroadcastSend operation.
-   * 
-   * @param scope current scope
-   * @param input 
-   * @param groupSize 
-   * @param groupKey 
-   * @param instanceKey 
-   * @param shape 
-   * @param options carries optional attributes values
-   * @return a new instance of BroadcastSend
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> BroadcastSend<T> create(Scope scope, Operand<T> input, Long groupSize, Long groupKey, Long instanceKey, Shape shape, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("CollectiveBcastSend", scope.makeOpName("BroadcastSend"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("group_size", groupSize);
-    opBuilder.setAttr("group_key", groupKey);
-    opBuilder.setAttr("instance_key", instanceKey);
-    opBuilder.setAttr("shape", shape);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.communicationHint != null) {
-          opBuilder.setAttr("communication_hint", opts.communicationHint);
-        }
-        if (opts.timeoutSeconds != null) {
-          opBuilder.setAttr("timeout_seconds", opts.timeoutSeconds);
-        }
-      }
-    }
-    return new BroadcastSend<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param communicationHint 
-   */
-  public static Options communicationHint(String communicationHint) {
-    return new Options().communicationHint(communicationHint);
-  }
-  
-  /**
-   * @param timeoutSeconds 
-   */
-  public static Options timeoutSeconds(Float timeoutSeconds) {
-    return new Options().timeoutSeconds(timeoutSeconds);
-  }
-  
-  /**
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "CollectiveBcastSend";
-  
-  private Output<T> output;
-  
-  private BroadcastSend(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Abort.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Abort.java
deleted file mode 100644
index a84f2405b19..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Abort.java
+++ /dev/null
@@ -1,110 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-
-/**
- * Raise a exception to abort the process when called.
- * <p>
- * If exit_without_error is true, the process will exit normally,
- * otherwise it will exit with a SIGABORT signal.
- * <p>
- * Returns nothing but an exception.
- */
-@Operator
-public final class Abort extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.core.Abort}
-   */
-  public static class Options {
-    
-    /**
-     * @param errorMsg A string which is the message associated with the exception.
-     */
-    public Options errorMsg(String errorMsg) {
-      this.errorMsg = errorMsg;
-      return this;
-    }
-    
-    /**
-     * @param exitWithoutError 
-     */
-    public Options exitWithoutError(Boolean exitWithoutError) {
-      this.exitWithoutError = exitWithoutError;
-      return this;
-    }
-    
-    private String errorMsg;
-    private Boolean exitWithoutError;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new Abort operation.
-   * 
-   * @param scope current scope
-   * @param options carries optional attributes values
-   * @return a new instance of Abort
-   */
-  @Endpoint(describeByClass = true)
-  public static Abort create(Scope scope, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Abort", scope.makeOpName("Abort"));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.errorMsg != null) {
-          opBuilder.setAttr("error_msg", opts.errorMsg);
-        }
-        if (opts.exitWithoutError != null) {
-          opBuilder.setAttr("exit_without_error", opts.exitWithoutError);
-        }
-      }
-    }
-    return new Abort(opBuilder.build());
-  }
-  
-  /**
-   * @param errorMsg A string which is the message associated with the exception.
-   */
-  public static Options errorMsg(String errorMsg) {
-    return new Options().errorMsg(errorMsg);
-  }
-  
-  /**
-   * @param exitWithoutError 
-   */
-  public static Options exitWithoutError(Boolean exitWithoutError) {
-    return new Options().exitWithoutError(exitWithoutError);
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Abort";
-  
-  private Abort(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/All.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/All.java
deleted file mode 100644
index ae82b78ad75..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/All.java
+++ /dev/null
@@ -1,116 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TBool;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Computes the "logical and" of elements across dimensions of a tensor.
- * <p>
- * Reduces `input` along the dimensions given in `axis`. Unless
- * `keep_dims` is true, the rank of the tensor is reduced by 1 for each entry in
- * `axis`. If `keep_dims` is true, the reduced dimensions are
- * retained with length 1.
- */
-@Operator
-public final class All extends RawOp implements Operand<TBool> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.core.All}
-   */
-  public static class Options {
-    
-    /**
-     * @param keepDims If true, retain reduced dimensions with length 1.
-     */
-    public Options keepDims(Boolean keepDims) {
-      this.keepDims = keepDims;
-      return this;
-    }
-    
-    private Boolean keepDims;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new All operation.
-   * 
-   * @param scope current scope
-   * @param input The tensor to reduce.
-   * @param axis The dimensions to reduce. Must be in the range
-   * `[-rank(input), rank(input))`.
-   * @param options carries optional attributes values
-   * @return a new instance of All
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> All create(Scope scope, Operand<TBool> input, Operand<T> axis, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("All", scope.makeOpName("All"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder.addInput(axis.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.keepDims != null) {
-          opBuilder.setAttr("keep_dims", opts.keepDims);
-        }
-      }
-    }
-    return new All(opBuilder.build());
-  }
-  
-  /**
-   * @param keepDims If true, retain reduced dimensions with length 1.
-   */
-  public static Options keepDims(Boolean keepDims) {
-    return new Options().keepDims(keepDims);
-  }
-  
-  /**
-   * The reduced tensor.
-   */
-  public Output<TBool> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<TBool> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "All";
-  
-  private Output<TBool> output;
-  
-  private All(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Any.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Any.java
deleted file mode 100644
index 1be811c16f0..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Any.java
+++ /dev/null
@@ -1,116 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TBool;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Computes the "logical or" of elements across dimensions of a tensor.
- * <p>
- * Reduces `input` along the dimensions given in `axis`. Unless
- * `keep_dims` is true, the rank of the tensor is reduced by 1 for each entry in
- * `axis`. If `keep_dims` is true, the reduced dimensions are
- * retained with length 1.
- */
-@Operator
-public final class Any extends RawOp implements Operand<TBool> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.core.Any}
-   */
-  public static class Options {
-    
-    /**
-     * @param keepDims If true, retain reduced dimensions with length 1.
-     */
-    public Options keepDims(Boolean keepDims) {
-      this.keepDims = keepDims;
-      return this;
-    }
-    
-    private Boolean keepDims;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new Any operation.
-   * 
-   * @param scope current scope
-   * @param input The tensor to reduce.
-   * @param axis The dimensions to reduce. Must be in the range
-   * `[-rank(input), rank(input))`.
-   * @param options carries optional attributes values
-   * @return a new instance of Any
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> Any create(Scope scope, Operand<TBool> input, Operand<T> axis, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Any", scope.makeOpName("Any"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder.addInput(axis.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.keepDims != null) {
-          opBuilder.setAttr("keep_dims", opts.keepDims);
-        }
-      }
-    }
-    return new Any(opBuilder.build());
-  }
-  
-  /**
-   * @param keepDims If true, retain reduced dimensions with length 1.
-   */
-  public static Options keepDims(Boolean keepDims) {
-    return new Options().keepDims(keepDims);
-  }
-  
-  /**
-   * The reduced tensor.
-   */
-  public Output<TBool> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<TBool> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Any";
-  
-  private Output<TBool> output;
-  
-  private Any(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/AssertThat.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/AssertThat.java
deleted file mode 100644
index 19bb1cda9d8..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/AssertThat.java
+++ /dev/null
@@ -1,96 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.Operands;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TBool;
-
-/**
- * Asserts that the given condition is true.
- * <p>
- * If `condition` evaluates to false, print the list of tensors in `data`.
- * `summarize` determines how many entries of the tensors to print.
- */
-@Operator
-public final class AssertThat extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.core.AssertThat}
-   */
-  public static class Options {
-    
-    /**
-     * @param summarize Print this many entries of each tensor.
-     */
-    public Options summarize(Long summarize) {
-      this.summarize = summarize;
-      return this;
-    }
-    
-    private Long summarize;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new AssertThat operation.
-   * 
-   * @param scope current scope
-   * @param condition The condition to evaluate.
-   * @param data The tensors to print out when condition is false.
-   * @param options carries optional attributes values
-   * @return a new instance of AssertThat
-   */
-  @Endpoint(describeByClass = true)
-  public static AssertThat create(Scope scope, Operand<TBool> condition, Iterable<Operand<?>> data, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Assert", scope.makeOpName("AssertThat"));
-    opBuilder.addInput(condition.asOutput(scope));
-    opBuilder.addInputList(Operands.asOutputs(scope, data));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.summarize != null) {
-          opBuilder.setAttr("summarize", opts.summarize);
-        }
-      }
-    }
-    return new AssertThat(opBuilder.build());
-  }
-  
-  /**
-   * @param summarize Print this many entries of each tensor.
-   */
-  public static Options summarize(Long summarize) {
-    return new Options().summarize(summarize);
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Assert";
-  
-  private AssertThat(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Assign.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Assign.java
deleted file mode 100644
index f0b5b6174c3..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Assign.java
+++ /dev/null
@@ -1,140 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Update 'ref' by assigning 'value' to it.
- * <p>
- * This operation outputs "ref" after the assignment is done.
- * This makes it easier to chain operations that need to use the reset value.
- * 
- * @param <T> data type for {@code outputRef()} output
- */
-@Operator
-public final class Assign<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.core.Assign}
-   */
-  public static class Options {
-    
-    /**
-     * @param validateShape If true, the operation will validate that the shape
-     * of 'value' matches the shape of the Tensor being assigned to.  If false,
-     * 'ref' will take on the shape of 'value'.
-     */
-    public Options validateShape(Boolean validateShape) {
-      this.validateShape = validateShape;
-      return this;
-    }
-    
-    /**
-     * @param useLocking If True, the assignment will be protected by a lock;
-     * otherwise the behavior is undefined, but may exhibit less contention.
-     */
-    public Options useLocking(Boolean useLocking) {
-      this.useLocking = useLocking;
-      return this;
-    }
-    
-    private Boolean validateShape;
-    private Boolean useLocking;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new Assign operation.
-   * 
-   * @param scope current scope
-   * @param ref Should be from a `Variable` node. May be uninitialized.
-   * @param value The value to be assigned to the variable.
-   * @param options carries optional attributes values
-   * @return a new instance of Assign
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> Assign<T> create(Scope scope, Operand<T> ref, Operand<T> value, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Assign", scope.makeOpName("Assign"));
-    opBuilder.addInput(ref.asOutput(scope));
-    opBuilder.addInput(value.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.validateShape != null) {
-          opBuilder.setAttr("validate_shape", opts.validateShape);
-        }
-        if (opts.useLocking != null) {
-          opBuilder.setAttr("use_locking", opts.useLocking);
-        }
-      }
-    }
-    return new Assign<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param validateShape If true, the operation will validate that the shape
-   * of 'value' matches the shape of the Tensor being assigned to.  If false,
-   * 'ref' will take on the shape of 'value'.
-   */
-  public static Options validateShape(Boolean validateShape) {
-    return new Options().validateShape(validateShape);
-  }
-  
-  /**
-   * @param useLocking If True, the assignment will be protected by a lock;
-   * otherwise the behavior is undefined, but may exhibit less contention.
-   */
-  public static Options useLocking(Boolean useLocking) {
-    return new Options().useLocking(useLocking);
-  }
-  
-  /**
-   * = Same as "ref".  Returned as a convenience for operations that want
-   * to use the new value after the variable has been reset.
-   */
-  public Output<T> outputRef() {
-    return outputRef;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return outputRef;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Assign";
-  
-  private Output<T> outputRef;
-  
-  private Assign(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    outputRef = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/AssignAdd.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/AssignAdd.java
deleted file mode 100644
index 8c9052a4672..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/AssignAdd.java
+++ /dev/null
@@ -1,117 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Update 'ref' by adding 'value' to it.
- * <p>
- * This operation outputs "ref" after the update is done.
- * This makes it easier to chain operations that need to use the reset value.
- * 
- * @param <T> data type for {@code outputRef()} output
- */
-@Operator
-public final class AssignAdd<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.core.AssignAdd}
-   */
-  public static class Options {
-    
-    /**
-     * @param useLocking If True, the addition will be protected by a lock;
-     * otherwise the behavior is undefined, but may exhibit less contention.
-     */
-    public Options useLocking(Boolean useLocking) {
-      this.useLocking = useLocking;
-      return this;
-    }
-    
-    private Boolean useLocking;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new AssignAdd operation.
-   * 
-   * @param scope current scope
-   * @param ref Should be from a `Variable` node.
-   * @param value The value to be added to the variable.
-   * @param options carries optional attributes values
-   * @return a new instance of AssignAdd
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> AssignAdd<T> create(Scope scope, Operand<T> ref, Operand<T> value, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("AssignAdd", scope.makeOpName("AssignAdd"));
-    opBuilder.addInput(ref.asOutput(scope));
-    opBuilder.addInput(value.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.useLocking != null) {
-          opBuilder.setAttr("use_locking", opts.useLocking);
-        }
-      }
-    }
-    return new AssignAdd<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param useLocking If True, the addition will be protected by a lock;
-   * otherwise the behavior is undefined, but may exhibit less contention.
-   */
-  public static Options useLocking(Boolean useLocking) {
-    return new Options().useLocking(useLocking);
-  }
-  
-  /**
-   * = Same as "ref".  Returned as a convenience for operations that want
-   * to use the new value after the variable has been updated.
-   */
-  public Output<T> outputRef() {
-    return outputRef;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return outputRef;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "AssignAdd";
-  
-  private Output<T> outputRef;
-  
-  private AssignAdd(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    outputRef = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/AssignAddVariableOp.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/AssignAddVariableOp.java
deleted file mode 100644
index 641f8a435eb..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/AssignAddVariableOp.java
+++ /dev/null
@@ -1,61 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Adds a value to the current value of a variable.
- * <p>
- * Any ReadVariableOp with a control dependency on this op is guaranteed to
- * see the incremented value or a subsequent newer one.
- */
-@Operator
-public final class AssignAddVariableOp extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new AssignAddVariableOp operation.
-   * 
-   * @param scope current scope
-   * @param resource handle to the resource in which to store the variable.
-   * @param value the value by which the variable will be incremented.
-   * @return a new instance of AssignAddVariableOp
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> AssignAddVariableOp create(Scope scope, Operand<?> resource, Operand<T> value) {
-    OperationBuilder opBuilder = scope.env().opBuilder("AssignAddVariableOp", scope.makeOpName("AssignAddVariableOp"));
-    opBuilder.addInput(resource.asOutput(scope));
-    opBuilder.addInput(value.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new AssignAddVariableOp(opBuilder.build());
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "AssignAddVariableOp";
-  
-  private AssignAddVariableOp(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/AssignSub.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/AssignSub.java
deleted file mode 100644
index 0aec7a7afac..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/AssignSub.java
+++ /dev/null
@@ -1,117 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Update 'ref' by subtracting 'value' from it.
- * <p>
- * This operation outputs "ref" after the update is done.
- * This makes it easier to chain operations that need to use the reset value.
- * 
- * @param <T> data type for {@code outputRef()} output
- */
-@Operator
-public final class AssignSub<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.core.AssignSub}
-   */
-  public static class Options {
-    
-    /**
-     * @param useLocking If True, the subtraction will be protected by a lock;
-     * otherwise the behavior is undefined, but may exhibit less contention.
-     */
-    public Options useLocking(Boolean useLocking) {
-      this.useLocking = useLocking;
-      return this;
-    }
-    
-    private Boolean useLocking;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new AssignSub operation.
-   * 
-   * @param scope current scope
-   * @param ref Should be from a `Variable` node.
-   * @param value The value to be subtracted to the variable.
-   * @param options carries optional attributes values
-   * @return a new instance of AssignSub
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> AssignSub<T> create(Scope scope, Operand<T> ref, Operand<T> value, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("AssignSub", scope.makeOpName("AssignSub"));
-    opBuilder.addInput(ref.asOutput(scope));
-    opBuilder.addInput(value.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.useLocking != null) {
-          opBuilder.setAttr("use_locking", opts.useLocking);
-        }
-      }
-    }
-    return new AssignSub<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param useLocking If True, the subtraction will be protected by a lock;
-   * otherwise the behavior is undefined, but may exhibit less contention.
-   */
-  public static Options useLocking(Boolean useLocking) {
-    return new Options().useLocking(useLocking);
-  }
-  
-  /**
-   * = Same as "ref".  Returned as a convenience for operations that want
-   * to use the new value after the variable has been updated.
-   */
-  public Output<T> outputRef() {
-    return outputRef;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return outputRef;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "AssignSub";
-  
-  private Output<T> outputRef;
-  
-  private AssignSub(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    outputRef = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/AssignSubVariableOp.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/AssignSubVariableOp.java
deleted file mode 100644
index 44ef8585450..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/AssignSubVariableOp.java
+++ /dev/null
@@ -1,61 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Subtracts a value from the current value of a variable.
- * <p>
- * Any ReadVariableOp with a control dependency on this op is guaranteed to
- * see the decremented value or a subsequent newer one.
- */
-@Operator
-public final class AssignSubVariableOp extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new AssignSubVariableOp operation.
-   * 
-   * @param scope current scope
-   * @param resource handle to the resource in which to store the variable.
-   * @param value the value by which the variable will be incremented.
-   * @return a new instance of AssignSubVariableOp
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> AssignSubVariableOp create(Scope scope, Operand<?> resource, Operand<T> value) {
-    OperationBuilder opBuilder = scope.env().opBuilder("AssignSubVariableOp", scope.makeOpName("AssignSubVariableOp"));
-    opBuilder.addInput(resource.asOutput(scope));
-    opBuilder.addInput(value.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new AssignSubVariableOp(opBuilder.build());
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "AssignSubVariableOp";
-  
-  private AssignSubVariableOp(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/AssignVariableOp.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/AssignVariableOp.java
deleted file mode 100644
index 581f980f517..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/AssignVariableOp.java
+++ /dev/null
@@ -1,61 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Assigns a new value to a variable.
- * <p>
- * Any ReadVariableOp with a control dependency on this op is guaranteed to return
- * this value or a subsequent newer value of the variable.
- */
-@Operator
-public final class AssignVariableOp extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new AssignVariableOp operation.
-   * 
-   * @param scope current scope
-   * @param resource handle to the resource in which to store the variable.
-   * @param value the value to set the new tensor to use.
-   * @return a new instance of AssignVariableOp
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> AssignVariableOp create(Scope scope, Operand<?> resource, Operand<T> value) {
-    OperationBuilder opBuilder = scope.env().opBuilder("AssignVariableOp", scope.makeOpName("AssignVariableOp"));
-    opBuilder.addInput(resource.asOutput(scope));
-    opBuilder.addInput(value.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new AssignVariableOp(opBuilder.build());
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "AssignVariableOp";
-  
-  private AssignVariableOp(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Barrier.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Barrier.java
deleted file mode 100644
index 83e2d6a2261..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Barrier.java
+++ /dev/null
@@ -1,193 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TString;
-
-/**
- * Defines a barrier that persists across different graph executions.
- * <p>
- * A barrier represents a key-value map, where each key is a string, and
- * each value is a tuple of tensors.
- * <p>
- * At runtime, the barrier contains 'complete' and 'incomplete'
- * elements. A complete element has defined tensors for all components of
- * its value tuple, and may be accessed using BarrierTakeMany. An
- * incomplete element has some undefined components in its value tuple,
- * and may be updated using BarrierInsertMany.
- */
-@Operator
-public final class Barrier extends RawOp implements Operand<TString> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.core.Barrier}
-   */
-  public static class Options {
-    
-    /**
-     * @param shapes The shape of each component in a value. Each shape must be 1 in the
-     * first dimension. The length of this attr must be the same as the length of
-     * component_types.
-     */
-    public Options shapes(List<Shape> shapes) {
-      this.shapes = shapes;
-      return this;
-    }
-    
-    /**
-     * @param capacity The capacity of the barrier.  The default capacity is MAX_INT32,
-     * which is the largest capacity of the underlying queue.
-     */
-    public Options capacity(Long capacity) {
-      this.capacity = capacity;
-      return this;
-    }
-    
-    /**
-     * @param container If non-empty, this barrier is placed in the given container.
-     * Otherwise, a default container is used.
-     */
-    public Options container(String container) {
-      this.container = container;
-      return this;
-    }
-    
-    /**
-     * @param sharedName If non-empty, this barrier will be shared under the given name
-     * across multiple sessions.
-     */
-    public Options sharedName(String sharedName) {
-      this.sharedName = sharedName;
-      return this;
-    }
-    
-    private List<Shape> shapes;
-    private Long capacity;
-    private String container;
-    private String sharedName;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new Barrier operation.
-   * 
-   * @param scope current scope
-   * @param componentTypes The type of each component in a value.
-   * @param options carries optional attributes values
-   * @return a new instance of Barrier
-   */
-  @Endpoint(describeByClass = true)
-  public static Barrier create(Scope scope, List<Class<?>> componentTypes, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Barrier", scope.makeOpName("Barrier"));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    Class[] componentTypesArray = new Class[componentTypes.size()];
-    for (int i = 0; i < componentTypesArray.length; ++i) {
-      componentTypesArray[i] = componentTypes.get(i);
-    }
-    opBuilder.setAttr("component_types", componentTypesArray);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.shapes != null) {
-          Shape[] shapesArray = new Shape[opts.shapes.size()];
-          for (int i = 0; i < shapesArray.length; ++i) {
-            shapesArray[i] = opts.shapes.get(i);
-          }
-          opBuilder.setAttr("shapes", shapesArray);
-        }
-        if (opts.capacity != null) {
-          opBuilder.setAttr("capacity", opts.capacity);
-        }
-        if (opts.container != null) {
-          opBuilder.setAttr("container", opts.container);
-        }
-        if (opts.sharedName != null) {
-          opBuilder.setAttr("shared_name", opts.sharedName);
-        }
-      }
-    }
-    return new Barrier(opBuilder.build());
-  }
-  
-  /**
-   * @param shapes The shape of each component in a value. Each shape must be 1 in the
-   * first dimension. The length of this attr must be the same as the length of
-   * component_types.
-   */
-  public static Options shapes(List<Shape> shapes) {
-    return new Options().shapes(shapes);
-  }
-  
-  /**
-   * @param capacity The capacity of the barrier.  The default capacity is MAX_INT32,
-   * which is the largest capacity of the underlying queue.
-   */
-  public static Options capacity(Long capacity) {
-    return new Options().capacity(capacity);
-  }
-  
-  /**
-   * @param container If non-empty, this barrier is placed in the given container.
-   * Otherwise, a default container is used.
-   */
-  public static Options container(String container) {
-    return new Options().container(container);
-  }
-  
-  /**
-   * @param sharedName If non-empty, this barrier will be shared under the given name
-   * across multiple sessions.
-   */
-  public static Options sharedName(String sharedName) {
-    return new Options().sharedName(sharedName);
-  }
-  
-  /**
-   * The handle to the barrier.
-   */
-  public Output<TString> handle() {
-    return handle;
-  }
-  
-  @Override
-  public Output<TString> asOutput(Scope scope) {
-    return handle;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Barrier";
-  
-  private Output<TString> handle;
-  
-  private Barrier(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    handle = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BarrierClose.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BarrierClose.java
deleted file mode 100644
index e4ec2900d0a..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BarrierClose.java
+++ /dev/null
@@ -1,101 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TString;
-
-/**
- * Closes the given barrier.
- * <p>
- * This operation signals that no more new elements will be inserted in the
- * given barrier. Subsequent InsertMany that try to introduce a new key will fail.
- * Subsequent InsertMany operations that just add missing components to already
- * existing elements will continue to succeed. Subsequent TakeMany operations will
- * continue to succeed if sufficient completed elements remain in the barrier.
- * Subsequent TakeMany operations that would block will fail immediately.
- */
-@Operator
-public final class BarrierClose extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.core.BarrierClose}
-   */
-  public static class Options {
-    
-    /**
-     * @param cancelPendingEnqueues If true, all pending enqueue requests that are
-     * blocked on the barrier's queue will be canceled. InsertMany will fail, even
-     * if no new key is introduced.
-     */
-    public Options cancelPendingEnqueues(Boolean cancelPendingEnqueues) {
-      this.cancelPendingEnqueues = cancelPendingEnqueues;
-      return this;
-    }
-    
-    private Boolean cancelPendingEnqueues;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new BarrierClose operation.
-   * 
-   * @param scope current scope
-   * @param handle The handle to a barrier.
-   * @param options carries optional attributes values
-   * @return a new instance of BarrierClose
-   */
-  @Endpoint(describeByClass = true)
-  public static BarrierClose create(Scope scope, Operand<TString> handle, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("BarrierClose", scope.makeOpName("BarrierClose"));
-    opBuilder.addInput(handle.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.cancelPendingEnqueues != null) {
-          opBuilder.setAttr("cancel_pending_enqueues", opts.cancelPendingEnqueues);
-        }
-      }
-    }
-    return new BarrierClose(opBuilder.build());
-  }
-  
-  /**
-   * @param cancelPendingEnqueues If true, all pending enqueue requests that are
-   * blocked on the barrier's queue will be canceled. InsertMany will fail, even
-   * if no new key is introduced.
-   */
-  public static Options cancelPendingEnqueues(Boolean cancelPendingEnqueues) {
-    return new Options().cancelPendingEnqueues(cancelPendingEnqueues);
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "BarrierClose";
-  
-  private BarrierClose(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BarrierIncompleteSize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BarrierIncompleteSize.java
deleted file mode 100644
index f4c404030d9..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BarrierIncompleteSize.java
+++ /dev/null
@@ -1,75 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.TString;
-
-/**
- * Computes the number of incomplete elements in the given barrier.
- */
-@Operator
-public final class BarrierIncompleteSize extends RawOp implements Operand<TInt32> {
-  
-  /**
-   * Factory method to create a class wrapping a new BarrierIncompleteSize operation.
-   * 
-   * @param scope current scope
-   * @param handle The handle to a barrier.
-   * @return a new instance of BarrierIncompleteSize
-   */
-  @Endpoint(describeByClass = true)
-  public static BarrierIncompleteSize create(Scope scope, Operand<TString> handle) {
-    OperationBuilder opBuilder = scope.env().opBuilder("BarrierIncompleteSize", scope.makeOpName("BarrierIncompleteSize"));
-    opBuilder.addInput(handle.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new BarrierIncompleteSize(opBuilder.build());
-  }
-  
-  /**
-   * The number of incomplete elements (i.e. those with some of their value
-   * components not set) in the barrier.
-   */
-  public Output<TInt32> size() {
-    return size;
-  }
-  
-  @Override
-  public Output<TInt32> asOutput(Scope scope) {
-    return size;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "BarrierIncompleteSize";
-  
-  private Output<TInt32> size;
-  
-  private BarrierIncompleteSize(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    size = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BarrierInsertMany.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BarrierInsertMany.java
deleted file mode 100644
index 85ee4c294cf..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BarrierInsertMany.java
+++ /dev/null
@@ -1,69 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TString;
-import org.tensorflow.types.family.TType;
-
-/**
- * For each key, assigns the respective value to the specified component.
- * <p>
- * If a key is not found in the barrier, this operation will create a new
- * incomplete element. If a key is found in the barrier, and the element
- * already has a value at component_index, this operation will fail with
- * INVALID_ARGUMENT, and leave the barrier in an undefined state.
- */
-@Operator
-public final class BarrierInsertMany extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new BarrierInsertMany operation.
-   * 
-   * @param scope current scope
-   * @param handle The handle to a barrier.
-   * @param keys A one-dimensional tensor of keys, with length n.
-   * @param values An any-dimensional tensor of values, which are associated with the
-   * respective keys. The 0th dimension must have length n.
-   * @param componentIndex The component of the barrier elements that is being assigned.
-   * @return a new instance of BarrierInsertMany
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> BarrierInsertMany create(Scope scope, Operand<TString> handle, Operand<TString> keys, Operand<T> values, Long componentIndex) {
-    OperationBuilder opBuilder = scope.env().opBuilder("BarrierInsertMany", scope.makeOpName("BarrierInsertMany"));
-    opBuilder.addInput(handle.asOutput(scope));
-    opBuilder.addInput(keys.asOutput(scope));
-    opBuilder.addInput(values.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("component_index", componentIndex);
-    return new BarrierInsertMany(opBuilder.build());
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "BarrierInsertMany";
-  
-  private BarrierInsertMany(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BarrierReadySize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BarrierReadySize.java
deleted file mode 100644
index 0eda2364bca..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BarrierReadySize.java
+++ /dev/null
@@ -1,75 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.TString;
-
-/**
- * Computes the number of complete elements in the given barrier.
- */
-@Operator
-public final class BarrierReadySize extends RawOp implements Operand<TInt32> {
-  
-  /**
-   * Factory method to create a class wrapping a new BarrierReadySize operation.
-   * 
-   * @param scope current scope
-   * @param handle The handle to a barrier.
-   * @return a new instance of BarrierReadySize
-   */
-  @Endpoint(describeByClass = true)
-  public static BarrierReadySize create(Scope scope, Operand<TString> handle) {
-    OperationBuilder opBuilder = scope.env().opBuilder("BarrierReadySize", scope.makeOpName("BarrierReadySize"));
-    opBuilder.addInput(handle.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new BarrierReadySize(opBuilder.build());
-  }
-  
-  /**
-   * The number of complete elements (i.e. those with all of their value
-   * components set) in the barrier.
-   */
-  public Output<TInt32> size() {
-    return size;
-  }
-  
-  @Override
-  public Output<TInt32> asOutput(Scope scope) {
-    return size;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "BarrierReadySize";
-  
-  private Output<TInt32> size;
-  
-  private BarrierReadySize(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    size = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BarrierTakeMany.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BarrierTakeMany.java
deleted file mode 100644
index ed685428d90..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BarrierTakeMany.java
+++ /dev/null
@@ -1,190 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import java.util.Arrays;
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.TString;
-
-/**
- * Takes the given number of completed elements from a barrier.
- * <p>
- * This operation concatenates completed-element component tensors along
- * the 0th dimension to make a single component tensor.
- * <p>
- * Elements come out of the barrier when they are complete, and in the order
- * in which they were placed into the barrier.  The indices output provides
- * information about the batch in which each element was originally inserted
- * into the barrier.
- */
-@Operator
-public final class BarrierTakeMany extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.core.BarrierTakeMany}
-   */
-  public static class Options {
-    
-    /**
-     * @param allowSmallBatch Allow to return less than num_elements items if barrier is
-     * already closed.
-     */
-    public Options allowSmallBatch(Boolean allowSmallBatch) {
-      this.allowSmallBatch = allowSmallBatch;
-      return this;
-    }
-    
-    /**
-     * @param waitForIncomplete 
-     */
-    public Options waitForIncomplete(Boolean waitForIncomplete) {
-      this.waitForIncomplete = waitForIncomplete;
-      return this;
-    }
-    
-    /**
-     * @param timeoutMs If the queue is empty, this operation will block for up to
-     * timeout_ms milliseconds.
-     * Note: This option is not supported yet.
-     */
-    public Options timeoutMs(Long timeoutMs) {
-      this.timeoutMs = timeoutMs;
-      return this;
-    }
-    
-    private Boolean allowSmallBatch;
-    private Boolean waitForIncomplete;
-    private Long timeoutMs;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new BarrierTakeMany operation.
-   * 
-   * @param scope current scope
-   * @param handle The handle to a barrier.
-   * @param numElements A single-element tensor containing the number of elements to
-   * take.
-   * @param componentTypes The type of each component in a value.
-   * @param options carries optional attributes values
-   * @return a new instance of BarrierTakeMany
-   */
-  @Endpoint(describeByClass = true)
-  public static BarrierTakeMany create(Scope scope, Operand<TString> handle, Operand<TInt32> numElements, List<Class<?>> componentTypes, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("BarrierTakeMany", scope.makeOpName("BarrierTakeMany"));
-    opBuilder.addInput(handle.asOutput(scope));
-    opBuilder.addInput(numElements.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    Class[] componentTypesArray = new Class[componentTypes.size()];
-    for (int i = 0; i < componentTypesArray.length; ++i) {
-      componentTypesArray[i] = componentTypes.get(i);
-    }
-    opBuilder.setAttr("component_types", componentTypesArray);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.allowSmallBatch != null) {
-          opBuilder.setAttr("allow_small_batch", opts.allowSmallBatch);
-        }
-        if (opts.waitForIncomplete != null) {
-          opBuilder.setAttr("wait_for_incomplete", opts.waitForIncomplete);
-        }
-        if (opts.timeoutMs != null) {
-          opBuilder.setAttr("timeout_ms", opts.timeoutMs);
-        }
-      }
-    }
-    return new BarrierTakeMany(opBuilder.build());
-  }
-  
-  /**
-   * @param allowSmallBatch Allow to return less than num_elements items if barrier is
-   * already closed.
-   */
-  public static Options allowSmallBatch(Boolean allowSmallBatch) {
-    return new Options().allowSmallBatch(allowSmallBatch);
-  }
-  
-  /**
-   * @param waitForIncomplete 
-   */
-  public static Options waitForIncomplete(Boolean waitForIncomplete) {
-    return new Options().waitForIncomplete(waitForIncomplete);
-  }
-  
-  /**
-   * @param timeoutMs If the queue is empty, this operation will block for up to
-   * timeout_ms milliseconds.
-   * Note: This option is not supported yet.
-   */
-  public static Options timeoutMs(Long timeoutMs) {
-    return new Options().timeoutMs(timeoutMs);
-  }
-  
-  /**
-   * A one-dimensional tensor of indices, with length num_elems.
-   * These indices refer to the batch in which the values were placed into the
-   * barrier (starting with MIN_LONG and increasing with each BarrierInsertMany).
-   */
-  public Output<TInt64> indices() {
-    return indices;
-  }
-  
-  /**
-   * A one-dimensional tensor of keys, with length num_elements.
-   */
-  public Output<TString> keys() {
-    return keys;
-  }
-  
-  /**
-   * One any-dimensional tensor per component in a barrier element. All
-   * values have length num_elements in the 0th dimension.
-   */
-  public List<Output<?>> values() {
-    return values;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "BarrierTakeMany";
-  
-  private Output<TInt64> indices;
-  private Output<TString> keys;
-  private List<Output<?>> values;
-  
-  private BarrierTakeMany(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    indices = operation.output(outputIdx++);
-    keys = operation.output(outputIdx++);
-    int valuesLength = operation.outputListLength("values");
-    values = Arrays.asList(operation.outputList(outputIdx, valuesLength));
-    outputIdx += valuesLength;
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Batch.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Batch.java
deleted file mode 100644
index b8b418f978b..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Batch.java
+++ /dev/null
@@ -1,247 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import java.util.Arrays;
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.Operands;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt64;
-
-/**
- * Batches all input tensors nondeterministically.
- * <p>
- * When many instances of this Op are being run concurrently with the same
- * container/shared_name in the same device, some will output zero-shaped Tensors
- * and others will output Tensors of size up to max_batch_size.
- * <p>
- * All Tensors in in_tensors are batched together (so, for example, labels and
- * features should be batched with a single instance of this operation.
- * <p>
- * Each invocation of batch emits an `id` scalar which will be used to identify
- * this particular invocation when doing unbatch or its gradient.
- * <p>
- * Each op which emits a non-empty batch will also emit a non-empty batch_index
- * Tensor, which, is a [K, 3] matrix where each row contains the invocation's id,
- * start, and length of elements of each set of Tensors present in batched_tensors.
- * <p>
- * Batched tensors are concatenated along the first dimension, and all tensors in
- * in_tensors must have the first dimension of the same size.
- * <p>
- * in_tensors: The tensors to be batched.
- * num_batch_threads: Number of scheduling threads for processing batches of work.
- *  Determines the number of batches processed in parallel.
- * max_batch_size: Batch sizes will never be bigger than this.
- * batch_timeout_micros: Maximum number of microseconds to wait before outputting
- *  an incomplete batch.
- * allowed_batch_sizes: Optional list of allowed batch sizes. If left empty, does
- *  nothing. Otherwise, supplies a list of batch sizes, causing the op to pad
- *  batches up to one of those sizes. The entries must increase monotonically, and
- *  the final entry must equal max_batch_size.
- * grad_timeout_micros: The timeout to use for the gradient. See Unbatch.
- * batched_tensors: Either empty tensors or a batch of concatenated Tensors.
- * batch_index: If out_tensors is non-empty, has information to invert it.
- * container: Controls the scope of sharing of this batch.
- * id: always contains a scalar with a unique ID for this invocation of Batch.
- * shared_name: Concurrently running instances of batch in the same device with the
- *  same container and shared_name will batch their elements together. If left
- *  empty, the op name will be used as the shared name.
- * T: the types of tensors to be batched.
- */
-@Operator
-public final class Batch extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.core.Batch}
-   */
-  public static class Options {
-    
-    /**
-     * @param maxEnqueuedBatches 
-     */
-    public Options maxEnqueuedBatches(Long maxEnqueuedBatches) {
-      this.maxEnqueuedBatches = maxEnqueuedBatches;
-      return this;
-    }
-    
-    /**
-     * @param allowedBatchSizes 
-     */
-    public Options allowedBatchSizes(List<Long> allowedBatchSizes) {
-      this.allowedBatchSizes = allowedBatchSizes;
-      return this;
-    }
-    
-    /**
-     * @param container 
-     */
-    public Options container(String container) {
-      this.container = container;
-      return this;
-    }
-    
-    /**
-     * @param sharedName 
-     */
-    public Options sharedName(String sharedName) {
-      this.sharedName = sharedName;
-      return this;
-    }
-    
-    /**
-     * @param batchingQueue 
-     */
-    public Options batchingQueue(String batchingQueue) {
-      this.batchingQueue = batchingQueue;
-      return this;
-    }
-    
-    private Long maxEnqueuedBatches;
-    private List<Long> allowedBatchSizes;
-    private String container;
-    private String sharedName;
-    private String batchingQueue;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new Batch operation.
-   * 
-   * @param scope current scope
-   * @param inTensors 
-   * @param numBatchThreads 
-   * @param maxBatchSize 
-   * @param batchTimeoutMicros 
-   * @param gradTimeoutMicros 
-   * @param options carries optional attributes values
-   * @return a new instance of Batch
-   */
-  @Endpoint(describeByClass = true)
-  public static Batch create(Scope scope, Iterable<Operand<?>> inTensors, Long numBatchThreads, Long maxBatchSize, Long batchTimeoutMicros, Long gradTimeoutMicros, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Batch", scope.makeOpName("Batch"));
-    opBuilder.addInputList(Operands.asOutputs(scope, inTensors));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("num_batch_threads", numBatchThreads);
-    opBuilder.setAttr("max_batch_size", maxBatchSize);
-    opBuilder.setAttr("batch_timeout_micros", batchTimeoutMicros);
-    opBuilder.setAttr("grad_timeout_micros", gradTimeoutMicros);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.maxEnqueuedBatches != null) {
-          opBuilder.setAttr("max_enqueued_batches", opts.maxEnqueuedBatches);
-        }
-        if (opts.allowedBatchSizes != null) {
-          long[] allowedBatchSizesArray = new long[opts.allowedBatchSizes.size()];
-          for (int i = 0; i < allowedBatchSizesArray.length; ++i) {
-            allowedBatchSizesArray[i] = opts.allowedBatchSizes.get(i);
-          }
-          opBuilder.setAttr("allowed_batch_sizes", allowedBatchSizesArray);
-        }
-        if (opts.container != null) {
-          opBuilder.setAttr("container", opts.container);
-        }
-        if (opts.sharedName != null) {
-          opBuilder.setAttr("shared_name", opts.sharedName);
-        }
-        if (opts.batchingQueue != null) {
-          opBuilder.setAttr("batching_queue", opts.batchingQueue);
-        }
-      }
-    }
-    return new Batch(opBuilder.build());
-  }
-  
-  /**
-   * @param maxEnqueuedBatches 
-   */
-  public static Options maxEnqueuedBatches(Long maxEnqueuedBatches) {
-    return new Options().maxEnqueuedBatches(maxEnqueuedBatches);
-  }
-  
-  /**
-   * @param allowedBatchSizes 
-   */
-  public static Options allowedBatchSizes(List<Long> allowedBatchSizes) {
-    return new Options().allowedBatchSizes(allowedBatchSizes);
-  }
-  
-  /**
-   * @param container 
-   */
-  public static Options container(String container) {
-    return new Options().container(container);
-  }
-  
-  /**
-   * @param sharedName 
-   */
-  public static Options sharedName(String sharedName) {
-    return new Options().sharedName(sharedName);
-  }
-  
-  /**
-   * @param batchingQueue 
-   */
-  public static Options batchingQueue(String batchingQueue) {
-    return new Options().batchingQueue(batchingQueue);
-  }
-  
-  /**
-   */
-  public List<Output<?>> batchedTensors() {
-    return batchedTensors;
-  }
-  
-  /**
-   */
-  public Output<TInt64> batchIndex() {
-    return batchIndex;
-  }
-  
-  /**
-   */
-  public Output<TInt64> id() {
-    return id;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Batch";
-  
-  private List<Output<?>> batchedTensors;
-  private Output<TInt64> batchIndex;
-  private Output<TInt64> id;
-  
-  private Batch(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    int batchedTensorsLength = operation.outputListLength("batched_tensors");
-    batchedTensors = Arrays.asList(operation.outputList(outputIdx, batchedTensorsLength));
-    outputIdx += batchedTensorsLength;
-    batchIndex = operation.output(outputIdx++);
-    id = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BatchToSpace.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BatchToSpace.java
deleted file mode 100644
index 5bd9366d82d..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BatchToSpace.java
+++ /dev/null
@@ -1,147 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * BatchToSpace for 4-D tensors of type T.
- * <p>
- * This is a legacy version of the more general BatchToSpaceND.
- * <p>
- * Rearranges (permutes) data from batch into blocks of spatial data, followed by
- * cropping. This is the reverse transformation of SpaceToBatch. More specifically,
- * this op outputs a copy of the input tensor where values from the `batch`
- * dimension are moved in spatial blocks to the `height` and `width` dimensions,
- * followed by cropping along the `height` and `width` dimensions.
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator
-public final class BatchToSpace<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new BatchToSpace operation.
-   * 
-   * @param scope current scope
-   * @param input 4-D tensor with shape
-   * `[batch<i>block_size</i>block_size, height_pad/block_size, width_pad/block_size,
-   *   depth]`. Note that the batch size of the input tensor must be divisible by
-   * `block_size * block_size`.
-   * @param crops 2-D tensor of non-negative integers with shape `[2, 2]`. It specifies
-   * how many elements to crop from the intermediate result across the spatial
-   * dimensions as follows:
-   * <p>
-   *     crops = [[crop_top, crop_bottom], [crop_left, crop_right]]
-   * @param blockSize 
-   * @return a new instance of BatchToSpace
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> BatchToSpace<T> create(Scope scope, Operand<T> input, Operand<U> crops, Long blockSize) {
-    OperationBuilder opBuilder = scope.env().opBuilder("BatchToSpace", scope.makeOpName("BatchToSpace"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder.addInput(crops.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("block_size", blockSize);
-    return new BatchToSpace<T>(opBuilder.build());
-  }
-  
-  /**
-   * 4-D with shape `[batch, height, width, depth]`, where:
-   * <p>
-   *       height = height_pad - crop_top - crop_bottom
-   *       width = width_pad - crop_left - crop_right
-   * <p>
-   * The attr `block_size` must be greater than one. It indicates the block size.
-   * <p>
-   * Some examples:
-   * <p>
-   * (1) For the following input of shape `[4, 1, 1, 1]` and block_size of 2:
-   * <pre>{@code
-   * [[[[1]]], [[[2]]], [[[3]]], [[[4]]]]
-   * }</pre>
-   * The output tensor has shape `[1, 2, 2, 1]` and value:
-   * <pre>{@code
-   * x = [[[[1], [2]], [[3], [4]]]]
-   * }</pre>
-   * (2) For the following input of shape `[4, 1, 1, 3]` and block_size of 2:
-   * <pre>{@code
-   * [[[[1, 2, 3]]], [[[4, 5, 6]]], [[[7, 8, 9]]], [[[10, 11, 12]]]]
-   * }</pre>
-   * The output tensor has shape `[1, 2, 2, 3]` and value:
-   * <pre>{@code
-   * x = [[[[1, 2, 3], [4, 5, 6]],
-   *       [[7, 8, 9], [10, 11, 12]]]]
-   * }</pre>
-   * (3) For the following input of shape `[4, 2, 2, 1]` and block_size of 2:
-   * <pre>{@code
-   * x = [[[[1], [3]], [[9], [11]]],
-   *      [[[2], [4]], [[10], [12]]],
-   *      [[[5], [7]], [[13], [15]]],
-   *      [[[6], [8]], [[14], [16]]]]
-   * }</pre>
-   * The output tensor has shape `[1, 4, 4, 1]` and value:
-   * <pre>{@code
-   * x = [[[[1],   [2],  [3],  [4]],
-   *      [[5],   [6],  [7],  [8]],
-   *      [[9],  [10], [11],  [12]],
-   *      [[13], [14], [15],  [16]]]]
-   * }</pre>
-   * (4) For the following input of shape `[8, 1, 2, 1]` and block_size of 2:
-   * <pre>{@code
-   * x = [[[[1], [3]]], [[[9], [11]]], [[[2], [4]]], [[[10], [12]]],
-   *      [[[5], [7]]], [[[13], [15]]], [[[6], [8]]], [[[14], [16]]]]
-   * }</pre>
-   * The output tensor has shape `[2, 2, 4, 1]` and value:
-   * <pre>{@code
-   * x = [[[[1], [3]], [[5], [7]]],
-   *      [[[2], [4]], [[10], [12]]],
-   *      [[[5], [7]], [[13], [15]]],
-   *      [[[6], [8]], [[14], [16]]]]
-   * }</pre>
-   * 
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "BatchToSpace";
-  
-  private Output<T> output;
-  
-  private BatchToSpace(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BatchToSpaceNd.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BatchToSpaceNd.java
deleted file mode 100644
index ae991e8bf52..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BatchToSpaceNd.java
+++ /dev/null
@@ -1,181 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * BatchToSpace for N-D tensors of type T.
- * <p>
- * This operation reshapes the "batch" dimension 0 into `M + 1` dimensions of shape
- * `block_shape + [batch]`, interleaves these blocks back into the grid defined by
- * the spatial dimensions `[1, ..., M]`, to obtain a result with the same rank as
- * the input.  The spatial dimensions of this intermediate result are then
- * optionally cropped according to `crops` to produce the output.  This is the
- * reverse of SpaceToBatch.  See below for a precise description.
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator
-public final class BatchToSpaceNd<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new BatchToSpaceNd operation.
-   * 
-   * @param scope current scope
-   * @param input N-D with shape `input_shape = [batch] + spatial_shape + remaining_shape`,
-   * where spatial_shape has M dimensions.
-   * @param blockShape 1-D with shape `[M]`, all values must be >= 1.
-   * @param crops 2-D with shape `[M, 2]`, all values must be >= 0.
-   *   `crops[i] = [crop_start, crop_end]` specifies the amount to crop from input
-   *   dimension `i + 1`, which corresponds to spatial dimension `i`.  It is
-   *   required that
-   *   `crop_start[i] + crop_end[i] <= block_shape[i] * input_shape[i + 1]`.
-   * <p>
-   * This operation is equivalent to the following steps:
-   * <p>
-   * 1. Reshape `input` to `reshaped` of shape:
-   *      [block_shape[0], ..., block_shape[M-1],
-   *       batch / prod(block_shape),
-   *       input_shape[1], ..., input_shape[N-1]]
-   * <p>
-   * 2. Permute dimensions of `reshaped` to produce `permuted` of shape
-   *      [batch / prod(block_shape),
-   * <p>
-   *       input_shape[1], block_shape[0],
-   *       ...,
-   *       input_shape[M], block_shape[M-1],
-   * <p>
-   *       input_shape[M+1], ..., input_shape[N-1]]
-   * <p>
-   * 3. Reshape `permuted` to produce `reshaped_permuted` of shape
-   *      [batch / prod(block_shape),
-   * <p>
-   *       input_shape[1] * block_shape[0],
-   *       ...,
-   *       input_shape[M] * block_shape[M-1],
-   * <p>
-   *       input_shape[M+1],
-   *       ...,
-   *       input_shape[N-1]]
-   * <p>
-   * 4. Crop the start and end of dimensions `[1, ..., M]` of
-   *    `reshaped_permuted` according to `crops` to produce the output of shape:
-   *      [batch / prod(block_shape),
-   * <p>
-   *       input_shape[1] * block_shape[0] - crops[0,0] - crops[0,1],
-   *       ...,
-   *       input_shape[M] * block_shape[M-1] - crops[M-1,0] - crops[M-1,1],
-   * <p>
-   *       input_shape[M+1], ..., input_shape[N-1]]
-   * <p>
-   * Some examples:
-   * <p>
-   * (1) For the following input of shape `[4, 1, 1, 1]`, `block_shape = [2, 2]`, and
-   *     `crops = [[0, 0], [0, 0]]`:
-   * <pre>{@code
-   * [[[[1]]], [[[2]]], [[[3]]], [[[4]]]]
-   * }</pre>
-   * The output tensor has shape `[1, 2, 2, 1]` and value:
-   * <pre>{@code
-   * x = [[[[1], [2]], [[3], [4]]]]
-   * }</pre>
-   * (2) For the following input of shape `[4, 1, 1, 3]`, `block_shape = [2, 2]`, and
-   *     `crops = [[0, 0], [0, 0]]`:
-   * <pre>{@code
-   * [[[[1, 2, 3]]], [[[4, 5, 6]]], [[[7, 8, 9]]], [[[10, 11, 12]]]]
-   * }</pre>
-   * The output tensor has shape `[1, 2, 2, 3]` and value:
-   * <pre>{@code
-   * x = [[[[1, 2, 3], [4, 5, 6]],
-   *       [[7, 8, 9], [10, 11, 12]]]]
-   * }</pre>
-   * (3) For the following input of shape `[4, 2, 2, 1]`, `block_shape = [2, 2]`, and
-   *     `crops = [[0, 0], [0, 0]]`:
-   * <pre>{@code
-   * x = [[[[1], [3]], [[9], [11]]],
-   *      [[[2], [4]], [[10], [12]]],
-   *      [[[5], [7]], [[13], [15]]],
-   *      [[[6], [8]], [[14], [16]]]]
-   * }</pre>
-   * The output tensor has shape `[1, 4, 4, 1]` and value:
-   * <pre>{@code
-   * x = [[[[1],   [2],  [3],  [4]],
-   *      [[5],   [6],  [7],  [8]],
-   *      [[9],  [10], [11],  [12]],
-   *      [[13], [14], [15],  [16]]]]
-   * }</pre>
-   * (4) For the following input of shape `[8, 1, 3, 1]`, `block_shape = [2, 2]`, and
-   *     `crops = [[0, 0], [2, 0]]`:
-   * <pre>{@code
-   * x = [[[[0], [1], [3]]], [[[0], [9], [11]]],
-   *      [[[0], [2], [4]]], [[[0], [10], [12]]],
-   *      [[[0], [5], [7]]], [[[0], [13], [15]]],
-   *      [[[0], [6], [8]]], [[[0], [14], [16]]]]
-   * }</pre>
-   * The output tensor has shape `[2, 2, 4, 1]` and value:
-   * <pre>{@code
-   * x = [[[[1],   [2],  [3],  [4]],
-   *       [[5],   [6],  [7],  [8]]],
-   *      [[[9],  [10], [11],  [12]],
-   *       [[13], [14], [15],  [16]]]]
-   * }</pre>
-   * 
-   * @return a new instance of BatchToSpaceNd
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber, V extends TNumber> BatchToSpaceNd<T> create(Scope scope, Operand<T> input, Operand<U> blockShape, Operand<V> crops) {
-    OperationBuilder opBuilder = scope.env().opBuilder("BatchToSpaceND", scope.makeOpName("BatchToSpaceNd"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder.addInput(blockShape.asOutput(scope));
-    opBuilder.addInput(crops.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new BatchToSpaceNd<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "BatchToSpaceND";
-  
-  private Output<T> output;
-  
-  private BatchToSpaceNd(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Bitcast.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Bitcast.java
deleted file mode 100644
index 9dc11a8a61b..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Bitcast.java
+++ /dev/null
@@ -1,127 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Bitcasts a tensor from one type to another without copying data.
- * <p>
- * Given a tensor `input`, this operation returns a tensor that has the same buffer
- * data as `input` with datatype `type`.
- * <p>
- * If the input datatype `T` is larger than the output datatype `type` then the
- * shape changes from [...] to [..., sizeof(`T`)/sizeof(`type`)].
- * <p>
- * If `T` is smaller than `type`, the operator requires that the rightmost
- * dimension be equal to sizeof(`type`)/sizeof(`T`). The shape then goes from
- * [..., sizeof(`type`)/sizeof(`T`)] to [...].
- * <p>
- * tf.bitcast() and tf.cast() work differently when real dtype is casted as a complex dtype
- * (e.g. tf.complex64 or tf.complex128) as tf.cast() make imaginary part 0 while tf.bitcast()
- * gives module error.
- * For example,
- * <p>
- * Example 1:
- * <p>
- * >>> a = [1., 2., 3.]
- * >>> equality_bitcast = tf.bitcast(a, tf.complex128)
- * Traceback (most recent call last):
- * ...
- * InvalidArgumentError: Cannot bitcast from 1 to 18 [Op:Bitcast]
- * >>> equality_cast = tf.cast(a, tf.complex128)
- * >>> print(equality_cast)
- * tf.Tensor([1.+0.j 2.+0.j 3.+0.j], shape=(3,), dtype=complex128)
- * <p>
- * Example 2:
- * <p>
- * >>> tf.bitcast(tf.constant(0xffffffff, dtype=tf.uint32), tf.uint8)
- * <tf.Tensor: shape=(4,), dtype=uint8, numpy=array([255, 255, 255, 255], dtype=uint8)>
- * <p>
- * Example 3:
- * <p>
- * >>> x = [1., 2., 3.]
- * >>> y = [0., 2., 3.]
- * >>> equality= tf.equal(x,y)
- * >>> equality_cast = tf.cast(equality,tf.float32)
- * >>> equality_bitcast = tf.bitcast(equality_cast,tf.uint8)
- * >>> print(equality)
- * tf.Tensor([False True True], shape=(3,), dtype=bool)
- * >>> print(equality_cast)
- * tf.Tensor([0. 1. 1.], shape=(3,), dtype=float32)
- * >>> print(equality_bitcast)
- * tf.Tensor(
- *     [[  0   0   0   0]
- *      [  0   0 128  63]
- *      [  0   0 128  63]], shape=(3, 4), dtype=uint8)
- * <p>
- * <i>NOTE</i>: Bitcast is implemented as a low-level cast, so machines with different
- * endian orderings will give different results.
- * 
- * @param <U> data type for {@code output()} output
- */
-@Operator
-public final class Bitcast<U extends TType> extends RawOp implements Operand<U> {
-  
-  /**
-   * Factory method to create a class wrapping a new Bitcast operation.
-   * 
-   * @param scope current scope
-   * @param input 
-   * @param type 
-   * @return a new instance of Bitcast
-   */
-  @Endpoint(describeByClass = true)
-  public static <U extends TType, T extends TType> Bitcast<U> create(Scope scope, Operand<T> input, Class<U> type) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Bitcast", scope.makeOpName("Bitcast"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("type", type);
-    return new Bitcast<U>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<U> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<U> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Bitcast";
-  
-  private Output<U> output;
-  
-  private Bitcast(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BroadcastDynamicShape.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BroadcastDynamicShape.java
deleted file mode 100644
index 51e8295f2b3..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BroadcastDynamicShape.java
+++ /dev/null
@@ -1,79 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Return the shape of s0 op s1 with broadcast.
- * <p>
- * Given `s0` and `s1`, tensors that represent shapes, compute `r0`, the
- * broadcasted shape. `s0`, `s1` and `r0` are all integer vectors.
- * 
- * @param <T> data type for {@code r0()} output
- */
-@Operator
-public final class BroadcastDynamicShape<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new BroadcastDynamicShape operation.
-   * 
-   * @param scope current scope
-   * @param s0 
-   * @param s1 
-   * @return a new instance of BroadcastDynamicShape
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> BroadcastDynamicShape<T> create(Scope scope, Operand<T> s0, Operand<T> s1) {
-    OperationBuilder opBuilder = scope.env().opBuilder("BroadcastArgs", scope.makeOpName("BroadcastDynamicShape"));
-    opBuilder.addInput(s0.asOutput(scope));
-    opBuilder.addInput(s1.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new BroadcastDynamicShape<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> r0() {
-    return r0;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return r0;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "BroadcastArgs";
-  
-  private Output<T> r0;
-  
-  private BroadcastDynamicShape(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    r0 = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BroadcastGradientArgs.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BroadcastGradientArgs.java
deleted file mode 100644
index d5bbded593e..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BroadcastGradientArgs.java
+++ /dev/null
@@ -1,80 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Return the reduction indices for computing gradients of s0 op s1 with broadcast.
- * <p>
- * This is typically used by gradient computations for a broadcasting operation.
- * 
- * @param <T> data type for {@code r0()} output
- */
-public final class BroadcastGradientArgs<T extends TNumber> extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new BroadcastGradientArgs operation.
-   * 
-   * @param scope current scope
-   * @param s0 
-   * @param s1 
-   * @return a new instance of BroadcastGradientArgs
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> BroadcastGradientArgs<T> create(Scope scope, Operand<T> s0, Operand<T> s1) {
-    OperationBuilder opBuilder = scope.env().opBuilder("BroadcastGradientArgs", scope.makeOpName("BroadcastGradientArgs"));
-    opBuilder.addInput(s0.asOutput(scope));
-    opBuilder.addInput(s1.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new BroadcastGradientArgs<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> r0() {
-    return r0;
-  }
-  
-  /**
-   */
-  public Output<T> r1() {
-    return r1;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "BroadcastGradientArgs";
-  
-  private Output<T> r0;
-  private Output<T> r1;
-  
-  private BroadcastGradientArgs(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    r0 = operation.output(outputIdx++);
-    r1 = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BroadcastTo.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BroadcastTo.java
deleted file mode 100644
index b99bbb51cf5..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BroadcastTo.java
+++ /dev/null
@@ -1,106 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * Broadcast an array for a compatible shape.
- * <p>
- * Broadcasting is the process of making arrays to have compatible shapes
- * for arithmetic operations. Two shapes are compatible if for each
- * dimension pair they are either equal or one of them is one. When trying
- * to broadcast a Tensor to a shape, it starts with the trailing dimensions,
- * and works its way forward.
- * <p>
- * For example,
- * <p>
- * >>> x = tf.constant([1, 2, 3])
- * >>> y = tf.broadcast_to(x, [3, 3])
- * >>> print(y)
- * tf.Tensor(
- *     [[1 2 3]
- *      [1 2 3]
- *      [1 2 3]], shape=(3, 3), dtype=int32)
- * <p>
- * In the above example, the input Tensor with the shape of `[1, 3]`
- * is broadcasted to output Tensor with shape of `[3, 3]`.
- * <p>
- * When doing broadcasted operations such as multiplying a tensor
- * by a scalar, broadcasting (usually) confers some time or space
- * benefit, as the broadcasted tensor is never materialized.
- * <p>
- * However, `broadcast_to` does not carry with it any such benefits.
- * The newly-created tensor takes the full memory of the broadcasted
- * shape. (In a graph context, `broadcast_to` might be fused to
- * subsequent operation and then be optimized away, however.)
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator
-public final class BroadcastTo<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new BroadcastTo operation.
-   * 
-   * @param scope current scope
-   * @param input A Tensor to broadcast.
-   * @param shape An 1-D `int` Tensor. The shape of the desired output.
-   * @return a new instance of BroadcastTo
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> BroadcastTo<T> create(Scope scope, Operand<T> input, Operand<U> shape) {
-    OperationBuilder opBuilder = scope.env().opBuilder("BroadcastTo", scope.makeOpName("BroadcastTo"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder.addInput(shape.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new BroadcastTo<T>(opBuilder.build());
-  }
-  
-  /**
-   * A Tensor.
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "BroadcastTo";
-  
-  private Output<T> output;
-  
-  private BroadcastTo(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Bucketize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Bucketize.java
deleted file mode 100644
index ee6a3660c37..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Bucketize.java
+++ /dev/null
@@ -1,96 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Bucketizes 'input' based on 'boundaries'.
- * <p>
- * For example, if the inputs are
- *     boundaries = [0, 10, 100]
- *     input = [[-5, 10000]
- *              [150,   10]
- *              [5,    100]]
- * <p>
- * then the output will be
- *     output = [[0, 3]
- *               [3, 2]
- *               [1, 3]]
- */
-@Operator
-public final class Bucketize extends RawOp implements Operand<TInt32> {
-  
-  /**
-   * Factory method to create a class wrapping a new Bucketize operation.
-   * 
-   * @param scope current scope
-   * @param input Any shape of Tensor contains with int or float type.
-   * @param boundaries A sorted list of floats gives the boundary of the buckets.
-   * @return a new instance of Bucketize
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> Bucketize create(Scope scope, Operand<T> input, List<Float> boundaries) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Bucketize", scope.makeOpName("Bucketize"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    float[] boundariesArray = new float[boundaries.size()];
-    for (int i = 0; i < boundariesArray.length; ++i) {
-      boundariesArray[i] = boundaries.get(i);
-    }
-    opBuilder.setAttr("boundaries", boundariesArray);
-    return new Bucketize(opBuilder.build());
-  }
-  
-  /**
-   * Same shape with 'input', each value of input replaced with bucket index.
-   * <p>
-   * @compatibility(numpy)
-   * Equivalent to np.digitize.
-   * @end_compatibility
-   */
-  public Output<TInt32> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<TInt32> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Bucketize";
-  
-  private Output<TInt32> output;
-  
-  private Bucketize(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ClipByValue.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ClipByValue.java
deleted file mode 100644
index 7d4859664ab..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ClipByValue.java
+++ /dev/null
@@ -1,86 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Clips tensor values to a specified min and max.
- * <p>
- * Given a tensor `t`, this operation returns a tensor of the same type and
- * shape as `t` with its values clipped to `clip_value_min` and `clip_value_max`.
- * Any values less than `clip_value_min` are set to `clip_value_min`. Any values
- * greater than `clip_value_max` are set to `clip_value_max`.
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator
-public final class ClipByValue<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new ClipByValue operation.
-   * 
-   * @param scope current scope
-   * @param t A `Tensor`.
-   * @param clipValueMin A 0-D (scalar) `Tensor`, or a `Tensor` with the same shape
-   * as `t`. The minimum value to clip by.
-   * @param clipValueMax A 0-D (scalar) `Tensor`, or a `Tensor` with the same shape
-   * as `t`. The maximum value to clip by.
-   * @return a new instance of ClipByValue
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> ClipByValue<T> create(Scope scope, Operand<T> t, Operand<T> clipValueMin, Operand<T> clipValueMax) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ClipByValue", scope.makeOpName("ClipByValue"));
-    opBuilder.addInput(t.asOutput(scope));
-    opBuilder.addInput(clipValueMin.asOutput(scope));
-    opBuilder.addInput(clipValueMax.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new ClipByValue<T>(opBuilder.build());
-  }
-  
-  /**
-   * A clipped `Tensor` with the same shape as input 't'.
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ClipByValue";
-  
-  private Output<T> output;
-  
-  private ClipByValue(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/CollectiveGather.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/CollectiveGather.java
deleted file mode 100644
index 2b504684e85..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/CollectiveGather.java
+++ /dev/null
@@ -1,135 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Mutually accumulates multiple tensors of identical type and shape.
- * 
- * @param <T> data type for {@code output()} output
- */
-public final class CollectiveGather<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.core.CollectiveGather}
-   */
-  public static class Options {
-    
-    /**
-     * @param communicationHint 
-     */
-    public Options communicationHint(String communicationHint) {
-      this.communicationHint = communicationHint;
-      return this;
-    }
-    
-    /**
-     * @param timeoutSeconds 
-     */
-    public Options timeoutSeconds(Float timeoutSeconds) {
-      this.timeoutSeconds = timeoutSeconds;
-      return this;
-    }
-    
-    private String communicationHint;
-    private Float timeoutSeconds;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new CollectiveGather operation.
-   * 
-   * @param scope current scope
-   * @param input 
-   * @param groupSize 
-   * @param groupKey 
-   * @param instanceKey 
-   * @param shape 
-   * @param options carries optional attributes values
-   * @return a new instance of CollectiveGather
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> CollectiveGather<T> create(Scope scope, Operand<T> input, Long groupSize, Long groupKey, Long instanceKey, Shape shape, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("CollectiveGather", scope.makeOpName("CollectiveGather"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("group_size", groupSize);
-    opBuilder.setAttr("group_key", groupKey);
-    opBuilder.setAttr("instance_key", instanceKey);
-    opBuilder.setAttr("shape", shape);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.communicationHint != null) {
-          opBuilder.setAttr("communication_hint", opts.communicationHint);
-        }
-        if (opts.timeoutSeconds != null) {
-          opBuilder.setAttr("timeout_seconds", opts.timeoutSeconds);
-        }
-      }
-    }
-    return new CollectiveGather<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param communicationHint 
-   */
-  public static Options communicationHint(String communicationHint) {
-    return new Options().communicationHint(communicationHint);
-  }
-  
-  /**
-   * @param timeoutSeconds 
-   */
-  public static Options timeoutSeconds(Float timeoutSeconds) {
-    return new Options().timeoutSeconds(timeoutSeconds);
-  }
-  
-  /**
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "CollectiveGather";
-  
-  private Output<T> output;
-  
-  private CollectiveGather(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Concat.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Concat.java
deleted file mode 100644
index 116cea2d4aa..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Concat.java
+++ /dev/null
@@ -1,83 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.Operands;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * Concatenates tensors along one dimension.
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator
-public final class Concat<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new Concat operation.
-   * 
-   * @param scope current scope
-   * @param values List of `N` Tensors to concatenate. Their ranks and types must match,
-   * and their sizes must match in all dimensions except `concat_dim`.
-   * @param axis 0-D.  The dimension along which to concatenate.  Must be in the
-   * range [-rank(values), rank(values)).
-   * @return a new instance of Concat
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> Concat<T> create(Scope scope, Iterable<Operand<T>> values, Operand<U> axis) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ConcatV2", scope.makeOpName("Concat"));
-    opBuilder.addInputList(Operands.asOutputs(scope, values));
-    opBuilder.addInput(axis.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new Concat<T>(opBuilder.build());
-  }
-  
-  /**
-   * A `Tensor` with the concatenation of values stacked along the
-   * `concat_dim` dimension.  This tensor's shape matches that of `values` except
-   * in `concat_dim` where it has the sum of the sizes.
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ConcatV2";
-  
-  private Output<T> output;
-  
-  private Concat(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ConsumeMutexLock.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ConsumeMutexLock.java
deleted file mode 100644
index 2c88d65429f..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ConsumeMutexLock.java
+++ /dev/null
@@ -1,63 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-
-/**
- * This op consumes a lock created by `MutexLock`.
- * <p>
- * This op exists to consume a tensor created by `MutexLock` (other than
- * direct control dependencies).  It should be the only that consumes the tensor,
- * and will raise an error if it is not.  Its only purpose is to keep the
- * mutex lock tensor alive until it is consumed by this op.
- * <p>
- * <b>NOTE</b>: This operation must run on the same device as its input.  This may
- * be enforced via the `colocate_with` mechanism.
- */
-@Operator
-public final class ConsumeMutexLock extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new ConsumeMutexLock operation.
-   * 
-   * @param scope current scope
-   * @param mutexLock A tensor returned by `MutexLock`.
-   * @return a new instance of ConsumeMutexLock
-   */
-  @Endpoint(describeByClass = true)
-  public static ConsumeMutexLock create(Scope scope, Operand<?> mutexLock) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ConsumeMutexLock", scope.makeOpName("ConsumeMutexLock"));
-    opBuilder.addInput(mutexLock.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new ConsumeMutexLock(opBuilder.build());
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ConsumeMutexLock";
-  
-  private ConsumeMutexLock(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ControlTrigger.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ControlTrigger.java
deleted file mode 100644
index e40715c9f2c..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ControlTrigger.java
+++ /dev/null
@@ -1,54 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-
-/**
- * Does nothing. Serves as a control trigger for scheduling.
- * <p>
- * Only useful as a placeholder for control edges.
- */
-@Operator
-public final class ControlTrigger extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new ControlTrigger operation.
-   * 
-   * @param scope current scope
-   * @return a new instance of ControlTrigger
-   */
-  @Endpoint(describeByClass = true)
-  public static ControlTrigger create(Scope scope) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ControlTrigger", scope.makeOpName("ControlTrigger"));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new ControlTrigger(opBuilder.build());
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ControlTrigger";
-  
-  private ControlTrigger(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Copy.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Copy.java
deleted file mode 100644
index 13a00037a10..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Copy.java
+++ /dev/null
@@ -1,148 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Copy a tensor from CPU-to-CPU or GPU-to-GPU.
- * <p>
- * Performs CPU-to-CPU or GPU-to-GPU deep-copying of tensor, depending on the
- * device on which the tensor is allocated.
- * N.B.: If the all downstream attached debug ops are disabled given the current
- * gRPC gating status, the output will simply forward the input tensor without
- * deep-copying. See the documentation of Debug* ops for more details.
- * <p>
- * Unlike the CopyHost Op, this op does not have HostMemory constraint on its
- * input or output.
- * 
- * @param <T> data type for {@code output()} output
- */
-public final class Copy<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.core.Copy}
-   */
-  public static class Options {
-    
-    /**
-     * @param tensorName The name of the input tensor.
-     */
-    public Options tensorName(String tensorName) {
-      this.tensorName = tensorName;
-      return this;
-    }
-    
-    /**
-     * @param debugOpsSpec A list of debug op spec (op, url, gated_grpc) for attached debug
-     * ops. Each element of the list has the format
-     * <debug_op>;<grpc_url>;<gated_grpc>, wherein gated_grpc is boolean represented
-     * as 0/1. E.g., "DebugIdentity;grpc://foo:3333;1",
-     * "DebugIdentity;file:///tmp/tfdbg_1;0".
-     */
-    public Options debugOpsSpec(List<String> debugOpsSpec) {
-      this.debugOpsSpec = debugOpsSpec;
-      return this;
-    }
-    
-    private String tensorName;
-    private List<String> debugOpsSpec;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new Copy operation.
-   * 
-   * @param scope current scope
-   * @param input Input tensor.
-   * @param options carries optional attributes values
-   * @return a new instance of Copy
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> Copy<T> create(Scope scope, Operand<T> input, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Copy", scope.makeOpName("Copy"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.tensorName != null) {
-          opBuilder.setAttr("tensor_name", opts.tensorName);
-        }
-        if (opts.debugOpsSpec != null) {
-          String[] debugOpsSpecArray = new String[opts.debugOpsSpec.size()];
-          for (int i = 0; i < debugOpsSpecArray.length; ++i) {
-            debugOpsSpecArray[i] = opts.debugOpsSpec.get(i);
-          }
-          opBuilder.setAttr("debug_ops_spec", debugOpsSpecArray);
-        }
-      }
-    }
-    return new Copy<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param tensorName The name of the input tensor.
-   */
-  public static Options tensorName(String tensorName) {
-    return new Options().tensorName(tensorName);
-  }
-  
-  /**
-   * @param debugOpsSpec A list of debug op spec (op, url, gated_grpc) for attached debug
-   * ops. Each element of the list has the format
-   * <debug_op>;<grpc_url>;<gated_grpc>, wherein gated_grpc is boolean represented
-   * as 0/1. E.g., "DebugIdentity;grpc://foo:3333;1",
-   * "DebugIdentity;file:///tmp/tfdbg_1;0".
-   */
-  public static Options debugOpsSpec(List<String> debugOpsSpec) {
-    return new Options().debugOpsSpec(debugOpsSpec);
-  }
-  
-  /**
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Copy";
-  
-  private Output<T> output;
-  
-  private Copy(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/CopyHost.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/CopyHost.java
deleted file mode 100644
index d18b72eaaba..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/CopyHost.java
+++ /dev/null
@@ -1,146 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Copy a tensor to host.
- * <p>
- * Performs CPU-to-CPU deep-copying of tensor.
- * N.B.: If the all downstream attached debug ops are disabled given the current
- * gRPC gating status, the output will simply forward the input tensor without
- * deep-copying. See the documentation of Debug* ops for more details.
- * <p>
- * Unlike the Copy Op, this op has HostMemory constraint on its input or output.
- * 
- * @param <T> data type for {@code output()} output
- */
-public final class CopyHost<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.core.CopyHost}
-   */
-  public static class Options {
-    
-    /**
-     * @param tensorName The name of the input tensor.
-     */
-    public Options tensorName(String tensorName) {
-      this.tensorName = tensorName;
-      return this;
-    }
-    
-    /**
-     * @param debugOpsSpec A list of debug op spec (op, url, gated_grpc) for attached debug
-     * ops. Each element of the list has the format
-     * <debug_op>;<grpc_url>;<gated_grpc>, wherein gated_grpc is boolean represented
-     * as 0/1. E.g., "DebugIdentity;grpc://foo:3333;1",
-     * "DebugIdentity;file:///tmp/tfdbg_1;0".
-     */
-    public Options debugOpsSpec(List<String> debugOpsSpec) {
-      this.debugOpsSpec = debugOpsSpec;
-      return this;
-    }
-    
-    private String tensorName;
-    private List<String> debugOpsSpec;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new CopyHost operation.
-   * 
-   * @param scope current scope
-   * @param input Input tensor.
-   * @param options carries optional attributes values
-   * @return a new instance of CopyHost
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> CopyHost<T> create(Scope scope, Operand<T> input, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("CopyHost", scope.makeOpName("CopyHost"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.tensorName != null) {
-          opBuilder.setAttr("tensor_name", opts.tensorName);
-        }
-        if (opts.debugOpsSpec != null) {
-          String[] debugOpsSpecArray = new String[opts.debugOpsSpec.size()];
-          for (int i = 0; i < debugOpsSpecArray.length; ++i) {
-            debugOpsSpecArray[i] = opts.debugOpsSpec.get(i);
-          }
-          opBuilder.setAttr("debug_ops_spec", debugOpsSpecArray);
-        }
-      }
-    }
-    return new CopyHost<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param tensorName The name of the input tensor.
-   */
-  public static Options tensorName(String tensorName) {
-    return new Options().tensorName(tensorName);
-  }
-  
-  /**
-   * @param debugOpsSpec A list of debug op spec (op, url, gated_grpc) for attached debug
-   * ops. Each element of the list has the format
-   * <debug_op>;<grpc_url>;<gated_grpc>, wherein gated_grpc is boolean represented
-   * as 0/1. E.g., "DebugIdentity;grpc://foo:3333;1",
-   * "DebugIdentity;file:///tmp/tfdbg_1;0".
-   */
-  public static Options debugOpsSpec(List<String> debugOpsSpec) {
-    return new Options().debugOpsSpec(debugOpsSpec);
-  }
-  
-  /**
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "CopyHost";
-  
-  private Output<T> output;
-  
-  private CopyHost(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/CountUpTo.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/CountUpTo.java
deleted file mode 100644
index 47aad076465..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/CountUpTo.java
+++ /dev/null
@@ -1,79 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Increments 'ref' until it reaches 'limit'.
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator
-public final class CountUpTo<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new CountUpTo operation.
-   * 
-   * @param scope current scope
-   * @param ref Should be from a scalar `Variable` node.
-   * @param limit If incrementing ref would bring it above limit, instead generates an
-   * 'OutOfRange' error.
-   * @return a new instance of CountUpTo
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> CountUpTo<T> create(Scope scope, Operand<T> ref, Long limit) {
-    OperationBuilder opBuilder = scope.env().opBuilder("CountUpTo", scope.makeOpName("CountUpTo"));
-    opBuilder.addInput(ref.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("limit", limit);
-    return new CountUpTo<T>(opBuilder.build());
-  }
-  
-  /**
-   * A copy of the input before increment. If nothing else modifies the
-   * input, the values produced will all be distinct.
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "CountUpTo";
-  
-  private Output<T> output;
-  
-  private CountUpTo(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/DecodeProto.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/DecodeProto.java
deleted file mode 100644
index 7e8318e3b1c..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/DecodeProto.java
+++ /dev/null
@@ -1,222 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import java.util.Arrays;
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.TString;
-
-/**
- * The op extracts fields from a serialized protocol buffers message into tensors.
- * <p>
- * The `decode_proto` op extracts fields from a serialized protocol buffers
- * message into tensors.  The fields in `field_names` are decoded and converted
- * to the corresponding `output_types` if possible.
- * <p>
- * A `message_type` name must be provided to give context for the field names.
- * The actual message descriptor can be looked up either in the linked-in
- * descriptor pool or a filename provided by the caller using the
- * `descriptor_source` attribute.
- * <p>
- * Each output tensor is a dense tensor. This means that it is padded to hold
- * the largest number of repeated elements seen in the input minibatch. (The
- * shape is also padded by one to prevent zero-sized dimensions). The actual
- * repeat counts for each example in the minibatch can be found in the `sizes`
- * output. In many cases the output of `decode_proto` is fed immediately into
- * tf.squeeze if missing values are not a concern. When using tf.squeeze, always
- * pass the squeeze dimension explicitly to avoid surprises.
- * <p>
- * For the most part, the mapping between Proto field types and TensorFlow dtypes
- * is straightforward. However, there are a few special cases:
- * <p>
- * - A proto field that contains a submessage or group can only be converted
- * to `DT_STRING` (the serialized submessage). This is to reduce the complexity
- * of the API. The resulting string can be used as input to another instance of
- * the decode_proto op.
- * <p>
- * - TensorFlow lacks support for unsigned integers. The ops represent uint64
- * types as a `DT_INT64` with the same twos-complement bit pattern (the obvious
- * way). Unsigned int32 values can be represented exactly by specifying type
- * `DT_INT64`, or using twos-complement if the caller specifies `DT_INT32` in
- * the `output_types` attribute.
- * <p>
- * Both binary and text proto serializations are supported, and can be
- * chosen using the `format` attribute.
- * <p>
- * The `descriptor_source` attribute selects the source of protocol
- * descriptors to consult when looking up `message_type`. This may be:
- * <p>
- * - An empty string  or "local://", in which case protocol descriptors are
- * created for C++ (not Python) proto definitions linked to the binary.
- * <p>
- * - A file, in which case protocol descriptors are created from the file,
- * which is expected to contain a `FileDescriptorSet` serialized as a string.
- * NOTE: You can build a `descriptor_source` file using the `--descriptor_set_out`
- * and `--include_imports` options to the protocol compiler `protoc`.
- * <p>
- * - A "bytes://<bytes>", in which protocol descriptors are created from `<bytes>`,
- * which is expected to be a `FileDescriptorSet` serialized as a string.
- */
-public final class DecodeProto extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.core.DecodeProto}
-   */
-  public static class Options {
-    
-    /**
-     * @param descriptorSource Either the special value `local://` or a path to a file containing
-     * a serialized `FileDescriptorSet`.
-     */
-    public Options descriptorSource(String descriptorSource) {
-      this.descriptorSource = descriptorSource;
-      return this;
-    }
-    
-    /**
-     * @param messageFormat Either `binary` or `text`.
-     */
-    public Options messageFormat(String messageFormat) {
-      this.messageFormat = messageFormat;
-      return this;
-    }
-    
-    /**
-     * @param sanitize Whether to sanitize the result or not.
-     */
-    public Options sanitize(Boolean sanitize) {
-      this.sanitize = sanitize;
-      return this;
-    }
-    
-    private String descriptorSource;
-    private String messageFormat;
-    private Boolean sanitize;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new DecodeProto operation.
-   * 
-   * @param scope current scope
-   * @param bytes Tensor of serialized protos with shape `batch_shape`.
-   * @param messageType Name of the proto message type to decode.
-   * @param fieldNames List of strings containing proto field names. An extension field can be decoded
-   * by using its full name, e.g. EXT_PACKAGE.EXT_FIELD_NAME.
-   * @param outputTypes List of TF types to use for the respective field in field_names.
-   * @param options carries optional attributes values
-   * @return a new instance of DecodeProto
-   */
-  @Endpoint(describeByClass = true)
-  public static DecodeProto create(Scope scope, Operand<TString> bytes, String messageType, List<String> fieldNames, List<Class<?>> outputTypes, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("DecodeProtoV2", scope.makeOpName("DecodeProto"));
-    opBuilder.addInput(bytes.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("message_type", messageType);
-    String[] fieldNamesArray = new String[fieldNames.size()];
-    for (int i = 0; i < fieldNamesArray.length; ++i) {
-      fieldNamesArray[i] = fieldNames.get(i);
-    }
-    opBuilder.setAttr("field_names", fieldNamesArray);
-    Class[] outputTypesArray = new Class[outputTypes.size()];
-    for (int i = 0; i < outputTypesArray.length; ++i) {
-      outputTypesArray[i] = outputTypes.get(i);
-    }
-    opBuilder.setAttr("output_types", outputTypesArray);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.descriptorSource != null) {
-          opBuilder.setAttr("descriptor_source", opts.descriptorSource);
-        }
-        if (opts.messageFormat != null) {
-          opBuilder.setAttr("message_format", opts.messageFormat);
-        }
-        if (opts.sanitize != null) {
-          opBuilder.setAttr("sanitize", opts.sanitize);
-        }
-      }
-    }
-    return new DecodeProto(opBuilder.build());
-  }
-  
-  /**
-   * @param descriptorSource Either the special value `local://` or a path to a file containing
-   * a serialized `FileDescriptorSet`.
-   */
-  public static Options descriptorSource(String descriptorSource) {
-    return new Options().descriptorSource(descriptorSource);
-  }
-  
-  /**
-   * @param messageFormat Either `binary` or `text`.
-   */
-  public static Options messageFormat(String messageFormat) {
-    return new Options().messageFormat(messageFormat);
-  }
-  
-  /**
-   * @param sanitize Whether to sanitize the result or not.
-   */
-  public static Options sanitize(Boolean sanitize) {
-    return new Options().sanitize(sanitize);
-  }
-  
-  /**
-   * Tensor of int32 with shape `[batch_shape, len(field_names)]`.
-   * Each entry is the number of values found for the corresponding field.
-   * Optional fields may have 0 or 1 values.
-   */
-  public Output<TInt32> sizes() {
-    return sizes;
-  }
-  
-  /**
-   * List of tensors containing values for the corresponding field.
-   * `values[i]` has datatype `output_types[i]`
-   * and shape `[batch_shape, max(sizes[...,i])]`.
-   */
-  public List<Output<?>> values() {
-    return values;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "DecodeProtoV2";
-  
-  private Output<TInt32> sizes;
-  private List<Output<?>> values;
-  
-  private DecodeProto(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    sizes = operation.output(outputIdx++);
-    int valuesLength = operation.outputListLength("values");
-    values = Arrays.asList(operation.outputList(outputIdx, valuesLength));
-    outputIdx += valuesLength;
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/DeepCopy.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/DeepCopy.java
deleted file mode 100644
index e6bcb413ff8..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/DeepCopy.java
+++ /dev/null
@@ -1,76 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Makes a copy of `x`.
- * 
- * @param <T> data type for {@code y()} output
- */
-@Operator
-public final class DeepCopy<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new DeepCopy operation.
-   * 
-   * @param scope current scope
-   * @param x The source tensor of type `T`.
-   * @return a new instance of DeepCopy
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> DeepCopy<T> create(Scope scope, Operand<T> x) {
-    OperationBuilder opBuilder = scope.env().opBuilder("DeepCopy", scope.makeOpName("DeepCopy"));
-    opBuilder.addInput(x.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new DeepCopy<T>(opBuilder.build());
-  }
-  
-  /**
-   *     y: A `Tensor` of type `T`. A copy of `x`. Guaranteed that `y`
-   *       is not an alias of `x`.
-   */
-  public Output<T> y() {
-    return y;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return y;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "DeepCopy";
-  
-  private Output<T> y;
-  
-  private DeepCopy(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    y = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/DeleteSessionTensor.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/DeleteSessionTensor.java
deleted file mode 100644
index 30f29c07e20..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/DeleteSessionTensor.java
+++ /dev/null
@@ -1,56 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TString;
-
-/**
- * Delete the tensor specified by its handle in the session.
- */
-@Operator
-public final class DeleteSessionTensor extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new DeleteSessionTensor operation.
-   * 
-   * @param scope current scope
-   * @param handle The handle for a tensor stored in the session state.
-   * @return a new instance of DeleteSessionTensor
-   */
-  @Endpoint(describeByClass = true)
-  public static DeleteSessionTensor create(Scope scope, Operand<TString> handle) {
-    OperationBuilder opBuilder = scope.env().opBuilder("DeleteSessionTensor", scope.makeOpName("DeleteSessionTensor"));
-    opBuilder.addInput(handle.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new DeleteSessionTensor(opBuilder.build());
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "DeleteSessionTensor";
-  
-  private DeleteSessionTensor(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/DestroyResourceOp.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/DestroyResourceOp.java
deleted file mode 100644
index b4dd2ba4acd..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/DestroyResourceOp.java
+++ /dev/null
@@ -1,94 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-
-/**
- * Deletes the resource specified by the handle.
- * <p>
- * All subsequent operations using the resource will result in a NotFound
- * error status.
- */
-@Operator
-public final class DestroyResourceOp extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.core.DestroyResourceOp}
-   */
-  public static class Options {
-    
-    /**
-     * @param ignoreLookupError whether to ignore the error when the resource
-     * doesn't exist.
-     */
-    public Options ignoreLookupError(Boolean ignoreLookupError) {
-      this.ignoreLookupError = ignoreLookupError;
-      return this;
-    }
-    
-    private Boolean ignoreLookupError;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new DestroyResourceOp operation.
-   * 
-   * @param scope current scope
-   * @param resource handle to the resource to delete.
-   * @param options carries optional attributes values
-   * @return a new instance of DestroyResourceOp
-   */
-  @Endpoint(describeByClass = true)
-  public static DestroyResourceOp create(Scope scope, Operand<?> resource, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("DestroyResourceOp", scope.makeOpName("DestroyResourceOp"));
-    opBuilder.addInput(resource.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.ignoreLookupError != null) {
-          opBuilder.setAttr("ignore_lookup_error", opts.ignoreLookupError);
-        }
-      }
-    }
-    return new DestroyResourceOp(opBuilder.build());
-  }
-  
-  /**
-   * @param ignoreLookupError whether to ignore the error when the resource
-   * doesn't exist.
-   */
-  public static Options ignoreLookupError(Boolean ignoreLookupError) {
-    return new Options().ignoreLookupError(ignoreLookupError);
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "DestroyResourceOp";
-  
-  private DestroyResourceOp(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/DestroyTemporaryVariable.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/DestroyTemporaryVariable.java
deleted file mode 100644
index 6dd40821a47..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/DestroyTemporaryVariable.java
+++ /dev/null
@@ -1,85 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Destroys the temporary variable and returns its final value.
- * <p>
- * Sets output to the value of the Tensor pointed to by 'ref', then destroys
- * the temporary variable called 'var_name'.
- * All other uses of 'ref' <i>must</i> have executed before this op.
- * This is typically achieved by chaining the ref through each assign op, or by
- * using control dependencies.
- * <p>
- * Outputs the final value of the tensor pointed to by 'ref'.
- * 
- * @param <T> data type for {@code value()} output
- */
-@Operator
-public final class DestroyTemporaryVariable<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new DestroyTemporaryVariable operation.
-   * 
-   * @param scope current scope
-   * @param ref A reference to the temporary variable tensor.
-   * @param varName Name of the temporary variable, usually the name of the matching
-   * 'TemporaryVariable' op.
-   * @return a new instance of DestroyTemporaryVariable
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> DestroyTemporaryVariable<T> create(Scope scope, Operand<T> ref, String varName) {
-    OperationBuilder opBuilder = scope.env().opBuilder("DestroyTemporaryVariable", scope.makeOpName("DestroyTemporaryVariable"));
-    opBuilder.addInput(ref.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("var_name", varName);
-    return new DestroyTemporaryVariable<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> value() {
-    return value;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return value;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "DestroyTemporaryVariable";
-  
-  private Output<T> value;
-  
-  private DestroyTemporaryVariable(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    value = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/DeviceIndex.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/DeviceIndex.java
deleted file mode 100644
index bb76e3d5992..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/DeviceIndex.java
+++ /dev/null
@@ -1,81 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
-
-/**
- * Return the index of device the op runs.
- * <p>
- * Given a list of device names, this operation returns the index of the device
- * this op runs. The length of the list is returned in two cases:
- * (1) Device does not exist in the given device list.
- * (2) It is in XLA compilation.
- */
-public final class DeviceIndex extends RawOp implements Operand<TInt32> {
-  
-  /**
-   * Factory method to create a class wrapping a new DeviceIndex operation.
-   * 
-   * @param scope current scope
-   * @param deviceNames 
-   * @return a new instance of DeviceIndex
-   */
-  @Endpoint(describeByClass = true)
-  public static DeviceIndex create(Scope scope, List<String> deviceNames) {
-    OperationBuilder opBuilder = scope.env().opBuilder("DeviceIndex", scope.makeOpName("DeviceIndex"));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    String[] deviceNamesArray = new String[deviceNames.size()];
-    for (int i = 0; i < deviceNamesArray.length; ++i) {
-      deviceNamesArray[i] = deviceNames.get(i);
-    }
-    opBuilder.setAttr("device_names", deviceNamesArray);
-    return new DeviceIndex(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<TInt32> index() {
-    return index;
-  }
-  
-  @Override
-  public Output<TInt32> asOutput(Scope scope) {
-    return index;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "DeviceIndex";
-  
-  private Output<TInt32> index;
-  
-  private DeviceIndex(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    index = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/DummyMemoryCache.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/DummyMemoryCache.java
deleted file mode 100644
index efc6061d293..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/DummyMemoryCache.java
+++ /dev/null
@@ -1,69 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- */
-public final class DummyMemoryCache extends RawOp implements Operand<TType> {
-  
-  /**
-   * Factory method to create a class wrapping a new DummyMemoryCache operation.
-   * 
-   * @param scope current scope
-   * @return a new instance of DummyMemoryCache
-   */
-  @Endpoint(describeByClass = true)
-  public static DummyMemoryCache create(Scope scope) {
-    OperationBuilder opBuilder = scope.env().opBuilder("DummyMemoryCache", scope.makeOpName("DummyMemoryCache"));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new DummyMemoryCache(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<?> handle() {
-    return handle;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) handle;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "DummyMemoryCache";
-  
-  private Output<?> handle;
-  
-  private DummyMemoryCache(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    handle = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/DynamicPartition.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/DynamicPartition.java
deleted file mode 100644
index 3561ff823d5..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/DynamicPartition.java
+++ /dev/null
@@ -1,120 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import java.util.Arrays;
-import java.util.Iterator;
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.family.TType;
-
-/**
- * Partitions `data` into `num_partitions` tensors using indices from `partitions`.
- * <p>
- * For each index tuple `js` of size `partitions.ndim`, the slice `data[js, ...]`
- * becomes part of `outputs[partitions[js]]`.  The slices with `partitions[js] = i`
- * are placed in `outputs[i]` in lexicographic order of `js`, and the first
- * dimension of `outputs[i]` is the number of entries in `partitions` equal to `i`.
- * In detail,
- * <pre>{@code
- *     outputs[i].shape = [sum(partitions == i)] + data.shape[partitions.ndim:]
- * 
- *     outputs[i] = pack([data[js, ...] for js if partitions[js] == i])
- * }</pre>
- * `data.shape` must start with `partitions.shape`.
- * <p>
- * For example:
- * <pre>{@code
- *     # Scalar partitions.
- *     partitions = 1
- *     num_partitions = 2
- *     data = [10, 20]
- *     outputs[0] = []  # Empty with shape [0, 2]
- *     outputs[1] = [[10, 20]]
- * 
- *     # Vector partitions.
- *     partitions = [0, 0, 1, 1, 0]
- *     num_partitions = 2
- *     data = [10, 20, 30, 40, 50]
- *     outputs[0] = [10, 20, 50]
- *     outputs[1] = [30, 40]
- * }</pre>
- * See `dynamic_stitch` for an example on how to merge partitions back.
- * <p>
- * <div style="width:70%; margin:auto; margin-bottom:10px; margin-top:20px;">
- * <img style="width:100%" src="https://www.tensorflow.org/images/DynamicPartition.png" alt>
- * </div>
- * 
- * @param <T> data type for {@code outputs()} output
- */
-@Operator
-public final class DynamicPartition<T extends TType> extends RawOp implements Iterable<Operand<T>> {
-  
-  /**
-   * Factory method to create a class wrapping a new DynamicPartition operation.
-   * 
-   * @param scope current scope
-   * @param data 
-   * @param partitions Any shape.  Indices in the range `[0, num_partitions)`.
-   * @param numPartitions The number of partitions to output.
-   * @return a new instance of DynamicPartition
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> DynamicPartition<T> create(Scope scope, Operand<T> data, Operand<TInt32> partitions, Long numPartitions) {
-    OperationBuilder opBuilder = scope.env().opBuilder("DynamicPartition", scope.makeOpName("DynamicPartition"));
-    opBuilder.addInput(data.asOutput(scope));
-    opBuilder.addInput(partitions.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("num_partitions", numPartitions);
-    return new DynamicPartition<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public List<Output<T>> outputs() {
-    return outputs;
-  }
-  
-  @Override
-  @SuppressWarnings({"rawtypes", "unchecked"})
-  public Iterator<Operand<T>> iterator() {
-    return (Iterator) outputs.iterator();
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "DynamicPartition";
-  
-  private List<Output<T>> outputs;
-  
-  @SuppressWarnings("unchecked")
-  private DynamicPartition(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    int outputsLength = operation.outputListLength("outputs");
-    outputs = Arrays.asList((Output<T>[])operation.outputList(outputIdx, outputsLength));
-    outputIdx += outputsLength;
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/DynamicStitch.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/DynamicStitch.java
deleted file mode 100644
index c2f3f9e06c6..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/DynamicStitch.java
+++ /dev/null
@@ -1,133 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.Operands;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.family.TType;
-
-/**
- * Interleave the values from the `data` tensors into a single tensor.
- * <p>
- * Builds a merged tensor such that
- * <pre>{@code
- *     merged[indices[m][i, ..., j], ...] = data[m][i, ..., j, ...]
- * }</pre>
- * For example, if each `indices[m]` is scalar or vector, we have
- * <pre>{@code
- *     # Scalar indices:
- *     merged[indices[m], ...] = data[m][...]
- * 
- *     # Vector indices:
- *     merged[indices[m][i], ...] = data[m][i, ...]
- * }</pre>
- * Each `data[i].shape` must start with the corresponding `indices[i].shape`,
- * and the rest of `data[i].shape` must be constant w.r.t. `i`.  That is, we
- * must have `data[i].shape = indices[i].shape + constant`.  In terms of this
- * `constant`, the output shape is
- * <p>
- *     merged.shape = [max(indices)] + constant
- * <p>
- * Values are merged in order, so if an index appears in both `indices[m][i]` and
- * `indices[n][j]` for `(m,i) < (n,j)` the slice `data[n][j]` will appear in the
- * merged result. If you do not need this guarantee, ParallelDynamicStitch might
- * perform better on some devices.
- * <p>
- * For example:
- * <pre>{@code
- *     indices[0] = 6
- *     indices[1] = [4, 1]
- *     indices[2] = [[5, 2], [0, 3]]
- *     data[0] = [61, 62]
- *     data[1] = [[41, 42], [11, 12]]
- *     data[2] = [[[51, 52], [21, 22]], [[1, 2], [31, 32]]]
- *     merged = [[1, 2], [11, 12], [21, 22], [31, 32], [41, 42],
- *               [51, 52], [61, 62]]
- * }</pre>
- * This method can be used to merge partitions created by `dynamic_partition`
- * as illustrated on the following example:
- * <pre>{@code
- *     # Apply function (increments x_i) on elements for which a certain condition
- *     # apply (x_i != -1 in this example).
- *     x=tf.constant([0.1, -1., 5.2, 4.3, -1., 7.4])
- *     condition_mask=tf.not_equal(x,tf.constant(-1.))
- *     partitioned_data = tf.dynamic_partition(
- *         x, tf.cast(condition_mask, tf.int32) , 2)
- *     partitioned_data[1] = partitioned_data[1] + 1.0
- *     condition_indices = tf.dynamic_partition(
- *         tf.range(tf.shape(x)[0]), tf.cast(condition_mask, tf.int32) , 2)
- *     x = tf.dynamic_stitch(condition_indices, partitioned_data)
- *     # Here x=[1.1, -1., 6.2, 5.3, -1, 8.4], the -1. values remain
- *     # unchanged.
- * }</pre>
- * <div style="width:70%; margin:auto; margin-bottom:10px; margin-top:20px;">
- * <img style="width:100%" src="https://www.tensorflow.org/images/DynamicStitch.png" alt>
- * </div>
- * 
- * @param <T> data type for {@code merged()} output
- */
-@Operator
-public final class DynamicStitch<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new DynamicStitch operation.
-   * 
-   * @param scope current scope
-   * @param indices 
-   * @param data 
-   * @return a new instance of DynamicStitch
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> DynamicStitch<T> create(Scope scope, Iterable<Operand<TInt32>> indices, Iterable<Operand<T>> data) {
-    OperationBuilder opBuilder = scope.env().opBuilder("DynamicStitch", scope.makeOpName("DynamicStitch"));
-    opBuilder.addInputList(Operands.asOutputs(scope, indices));
-    opBuilder.addInputList(Operands.asOutputs(scope, data));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new DynamicStitch<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> merged() {
-    return merged;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return merged;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "DynamicStitch";
-  
-  private Output<T> merged;
-  
-  private DynamicStitch(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    merged = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/EditDistance.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/EditDistance.java
deleted file mode 100644
index f914700e9f4..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/EditDistance.java
+++ /dev/null
@@ -1,164 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
-
-/**
- * Computes the (possibly normalized) Levenshtein Edit Distance.
- * <p>
- * The inputs are variable-length sequences provided by SparseTensors
- *   (hypothesis_indices, hypothesis_values, hypothesis_shape)
- * and
- *   (truth_indices, truth_values, truth_shape).
- * <p>
- * The inputs are:
- */
-@Operator
-public final class EditDistance extends RawOp implements Operand<TFloat32> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.core.EditDistance}
-   */
-  public static class Options {
-    
-    /**
-     * @param normalize boolean (if true, edit distances are normalized by length of truth).
-     * <p>
-     * The output is:
-     */
-    public Options normalize(Boolean normalize) {
-      this.normalize = normalize;
-      return this;
-    }
-    
-    private Boolean normalize;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new EditDistance operation.
-   * 
-   * @param scope current scope
-   * @param hypothesisIndices The indices of the hypothesis list SparseTensor.
-   * This is an N x R int64 matrix.
-   * @param hypothesisValues The values of the hypothesis list SparseTensor.
-   * This is an N-length vector.
-   * @param hypothesisShape The shape of the hypothesis list SparseTensor.
-   * This is an R-length vector.
-   * @param truthIndices The indices of the truth list SparseTensor.
-   * This is an M x R int64 matrix.
-   * @param truthValues The values of the truth list SparseTensor.
-   * This is an M-length vector.
-   * @param truthShape truth indices, vector.
-   * @param options carries optional attributes values
-   * @return a new instance of EditDistance
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> EditDistance create(Scope scope, Operand<TInt64> hypothesisIndices, Operand<T> hypothesisValues, Operand<TInt64> hypothesisShape, Operand<TInt64> truthIndices, Operand<T> truthValues, Operand<TInt64> truthShape, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("EditDistance", scope.makeOpName("EditDistance"));
-    opBuilder.addInput(hypothesisIndices.asOutput(scope));
-    opBuilder.addInput(hypothesisValues.asOutput(scope));
-    opBuilder.addInput(hypothesisShape.asOutput(scope));
-    opBuilder.addInput(truthIndices.asOutput(scope));
-    opBuilder.addInput(truthValues.asOutput(scope));
-    opBuilder.addInput(truthShape.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.normalize != null) {
-          opBuilder.setAttr("normalize", opts.normalize);
-        }
-      }
-    }
-    return new EditDistance(opBuilder.build());
-  }
-  
-  /**
-   * @param normalize boolean (if true, edit distances are normalized by length of truth).
-   * <p>
-   * The output is:
-   */
-  public static Options normalize(Boolean normalize) {
-    return new Options().normalize(normalize);
-  }
-  
-  /**
-   * A dense float tensor with rank R - 1.
-   * <p>
-   * For the example input:
-   * <p>
-   *     // hypothesis represents a 2x1 matrix with variable-length values:
-   *     //   (0,0) = ["a"]
-   *     //   (1,0) = ["b"]
-   *     hypothesis_indices = [[0, 0, 0],
-   *                           [1, 0, 0]]
-   *     hypothesis_values = ["a", "b"]
-   *     hypothesis_shape = [2, 1, 1]
-   * <p>
-   *     // truth represents a 2x2 matrix with variable-length values:
-   *     //   (0,0) = []
-   *     //   (0,1) = ["a"]
-   *     //   (1,0) = ["b", "c"]
-   *     //   (1,1) = ["a"]
-   *     truth_indices = [[0, 1, 0],
-   *                      [1, 0, 0],
-   *                      [1, 0, 1],
-   *                      [1, 1, 0]]
-   *     truth_values = ["a", "b", "c", "a"]
-   *     truth_shape = [2, 2, 2]
-   *     normalize = true
-   * <p>
-   * The output will be:
-   * <p>
-   *     // output is a 2x2 matrix with edit distances normalized by truth lengths.
-   *     output = [[inf, 1.0],  // (0,0): no truth, (0,1): no hypothesis
-   *               [0.5, 1.0]]  // (1,0): addition, (1,1): no hypothesis
-   */
-  public Output<TFloat32> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<TFloat32> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "EditDistance";
-  
-  private Output<TFloat32> output;
-  
-  private EditDistance(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Empty.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Empty.java
deleted file mode 100644
index dd2e60b3c6e..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Empty.java
+++ /dev/null
@@ -1,114 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.family.TType;
-
-/**
- * Creates a tensor with the given shape.
- * <p>
- * This operation creates a tensor of `shape` and `dtype`.
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator
-public final class Empty<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.core.Empty}
-   */
-  public static class Options {
-    
-    /**
-     * @param init If True, initialize the returned tensor with the default value of dtype.  Otherwise, the implementation is free not to initializethe tensor's content.
-     */
-    public Options init(Boolean init) {
-      this.init = init;
-      return this;
-    }
-    
-    private Boolean init;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new Empty operation.
-   * 
-   * @param scope current scope
-   * @param shape 1-D. Represents the shape of the output tensor.
-   * @param dtype 
-   * @param options carries optional attributes values
-   * @return a new instance of Empty
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> Empty<T> create(Scope scope, Operand<TInt32> shape, Class<T> dtype, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Empty", scope.makeOpName("Empty"));
-    opBuilder.addInput(shape.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("dtype", dtype);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.init != null) {
-          opBuilder.setAttr("init", opts.init);
-        }
-      }
-    }
-    return new Empty<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param init If True, initialize the returned tensor with the default value of dtype.  Otherwise, the implementation is free not to initializethe tensor's content.
-   */
-  public static Options init(Boolean init) {
-    return new Options().init(init);
-  }
-  
-  /**
-   * A `Tensor` of type `T`.
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Empty";
-  
-  private Output<T> output;
-  
-  private Empty(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/EmptyTensorList.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/EmptyTensorList.java
deleted file mode 100644
index d8f6813e542..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/EmptyTensorList.java
+++ /dev/null
@@ -1,86 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * Creates and returns an empty tensor list.
- * <p>
- * All list elements must be tensors of dtype element_dtype and shape compatible
- * with element_shape.
- * <p>
- * handle: an empty tensor list.
- * element_dtype: the type of elements in the list.
- * element_shape: a shape compatible with that of elements in the list.
- */
-@Operator
-public final class EmptyTensorList extends RawOp implements Operand<TType> {
-  
-  /**
-   * Factory method to create a class wrapping a new EmptyTensorList operation.
-   * 
-   * @param scope current scope
-   * @param elementShape 
-   * @param maxNumElements 
-   * @param elementDtype 
-   * @return a new instance of EmptyTensorList
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber, U extends TType> EmptyTensorList create(Scope scope, Operand<T> elementShape, Operand<TInt32> maxNumElements, Class<U> elementDtype) {
-    OperationBuilder opBuilder = scope.env().opBuilder("EmptyTensorList", scope.makeOpName("EmptyTensorList"));
-    opBuilder.addInput(elementShape.asOutput(scope));
-    opBuilder.addInput(maxNumElements.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("element_dtype", elementDtype);
-    return new EmptyTensorList(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<?> handle() {
-    return handle;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) handle;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "EmptyTensorList";
-  
-  private Output<?> handle;
-  
-  private EmptyTensorList(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    handle = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/EncodeProto.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/EncodeProto.java
deleted file mode 100644
index 9e7a5076a37..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/EncodeProto.java
+++ /dev/null
@@ -1,160 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.Operands;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.TString;
-
-/**
- * The op serializes protobuf messages provided in the input tensors.
- * <p>
- * The types of the tensors in `values` must match the schema for the fields
- * specified in `field_names`. All the tensors in `values` must have a common
- * shape prefix, <i>batch_shape</i>.
- * <p>
- * The `sizes` tensor specifies repeat counts for each field.  The repeat count
- * (last dimension) of a each tensor in `values` must be greater than or equal
- * to corresponding repeat count in `sizes`.
- * <p>
- * A `message_type` name must be provided to give context for the field names.
- * The actual message descriptor can be looked up either in the linked-in
- * descriptor pool or a filename provided by the caller using the
- * `descriptor_source` attribute.
- * <p>
- * For the most part, the mapping between Proto field types and TensorFlow dtypes
- * is straightforward. However, there are a few special cases:
- * <p>
- * - A proto field that contains a submessage or group can only be converted
- * to `DT_STRING` (the serialized submessage). This is to reduce the complexity
- * of the API. The resulting string can be used as input to another instance of
- * the decode_proto op.
- * <p>
- * - TensorFlow lacks support for unsigned integers. The ops represent uint64
- * types as a `DT_INT64` with the same twos-complement bit pattern (the obvious
- * way). Unsigned int32 values can be represented exactly by specifying type
- * `DT_INT64`, or using twos-complement if the caller specifies `DT_INT32` in
- * the `output_types` attribute.
- * <p>
- * The `descriptor_source` attribute selects the source of protocol
- * descriptors to consult when looking up `message_type`. This may be:
- * <p>
- * - An empty string  or "local://", in which case protocol descriptors are
- * created for C++ (not Python) proto definitions linked to the binary.
- * <p>
- * - A file, in which case protocol descriptors are created from the file,
- * which is expected to contain a `FileDescriptorSet` serialized as a string.
- * NOTE: You can build a `descriptor_source` file using the `--descriptor_set_out`
- * and `--include_imports` options to the protocol compiler `protoc`.
- * <p>
- * - A "bytes://<bytes>", in which protocol descriptors are created from `<bytes>`,
- * which is expected to be a `FileDescriptorSet` serialized as a string.
- */
-public final class EncodeProto extends RawOp implements Operand<TString> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.core.EncodeProto}
-   */
-  public static class Options {
-    
-    /**
-     * @param descriptorSource 
-     */
-    public Options descriptorSource(String descriptorSource) {
-      this.descriptorSource = descriptorSource;
-      return this;
-    }
-    
-    private String descriptorSource;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new EncodeProto operation.
-   * 
-   * @param scope current scope
-   * @param sizes Tensor of int32 with shape `[batch_shape, len(field_names)]`.
-   * @param values List of tensors containing values for the corresponding field.
-   * @param fieldNames List of strings containing proto field names.
-   * @param messageType Name of the proto message type to decode.
-   * @param options carries optional attributes values
-   * @return a new instance of EncodeProto
-   */
-  @Endpoint(describeByClass = true)
-  public static EncodeProto create(Scope scope, Operand<TInt32> sizes, Iterable<Operand<?>> values, List<String> fieldNames, String messageType, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("EncodeProto", scope.makeOpName("EncodeProto"));
-    opBuilder.addInput(sizes.asOutput(scope));
-    opBuilder.addInputList(Operands.asOutputs(scope, values));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    String[] fieldNamesArray = new String[fieldNames.size()];
-    for (int i = 0; i < fieldNamesArray.length; ++i) {
-      fieldNamesArray[i] = fieldNames.get(i);
-    }
-    opBuilder.setAttr("field_names", fieldNamesArray);
-    opBuilder.setAttr("message_type", messageType);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.descriptorSource != null) {
-          opBuilder.setAttr("descriptor_source", opts.descriptorSource);
-        }
-      }
-    }
-    return new EncodeProto(opBuilder.build());
-  }
-  
-  /**
-   * @param descriptorSource 
-   */
-  public static Options descriptorSource(String descriptorSource) {
-    return new Options().descriptorSource(descriptorSource);
-  }
-  
-  /**
-   * Tensor of serialized protos with shape `batch_shape`.
-   */
-  public Output<TString> bytes() {
-    return bytes;
-  }
-  
-  @Override
-  public Output<TString> asOutput(Scope scope) {
-    return bytes;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "EncodeProto";
-  
-  private Output<TString> bytes;
-  
-  private EncodeProto(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    bytes = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/EnsureShape.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/EnsureShape.java
deleted file mode 100644
index 0babbe5c112..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/EnsureShape.java
+++ /dev/null
@@ -1,81 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Ensures that the tensor's shape matches the expected shape.
- * <p>
- * Raises an error if the input tensor's shape does not match the specified shape.
- * Returns the input tensor otherwise.
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator
-public final class EnsureShape<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new EnsureShape operation.
-   * 
-   * @param scope current scope
-   * @param input A tensor, whose shape is to be validated.
-   * @param shape The expected (possibly partially specified) shape of the input tensor.
-   * @return a new instance of EnsureShape
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> EnsureShape<T> create(Scope scope, Operand<T> input, Shape shape) {
-    OperationBuilder opBuilder = scope.env().opBuilder("EnsureShape", scope.makeOpName("EnsureShape"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("shape", shape);
-    return new EnsureShape<T>(opBuilder.build());
-  }
-  
-  /**
-   * A tensor with the same shape and contents as the input tensor or value.
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "EnsureShape";
-  
-  private Output<T> output;
-  
-  private EnsureShape(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Enter.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Enter.java
deleted file mode 100644
index 9fb174a1942..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Enter.java
+++ /dev/null
@@ -1,135 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Creates or finds a child frame, and makes `data` available to the child frame.
- * <p>
- * This op is used together with `Exit` to create loops in the graph.
- * The unique `frame_name` is used by the `Executor` to identify frames. If
- * `is_constant` is true, `output` is a constant in the child frame; otherwise
- * it may be changed in the child frame. At most `parallel_iterations` iterations
- * are run in parallel in the child frame.
- * 
- * @param <T> data type for {@code output()} output
- */
-public final class Enter<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.core.Enter}
-   */
-  public static class Options {
-    
-    /**
-     * @param isConstant If true, the output is constant within the child frame.
-     */
-    public Options isConstant(Boolean isConstant) {
-      this.isConstant = isConstant;
-      return this;
-    }
-    
-    /**
-     * @param parallelIterations The number of iterations allowed to run in parallel.
-     */
-    public Options parallelIterations(Long parallelIterations) {
-      this.parallelIterations = parallelIterations;
-      return this;
-    }
-    
-    private Boolean isConstant;
-    private Long parallelIterations;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new Enter operation.
-   * 
-   * @param scope current scope
-   * @param data The tensor to be made available to the child frame.
-   * @param frameName The name of the child frame.
-   * @param options carries optional attributes values
-   * @return a new instance of Enter
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> Enter<T> create(Scope scope, Operand<T> data, String frameName, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Enter", scope.makeOpName("Enter"));
-    opBuilder.addInput(data.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("frame_name", frameName);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.isConstant != null) {
-          opBuilder.setAttr("is_constant", opts.isConstant);
-        }
-        if (opts.parallelIterations != null) {
-          opBuilder.setAttr("parallel_iterations", opts.parallelIterations);
-        }
-      }
-    }
-    return new Enter<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param isConstant If true, the output is constant within the child frame.
-   */
-  public static Options isConstant(Boolean isConstant) {
-    return new Options().isConstant(isConstant);
-  }
-  
-  /**
-   * @param parallelIterations The number of iterations allowed to run in parallel.
-   */
-  public static Options parallelIterations(Long parallelIterations) {
-    return new Options().parallelIterations(parallelIterations);
-  }
-  
-  /**
-   * The same tensor as `data`.
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Enter";
-  
-  private Output<T> output;
-  
-  private Enter(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Exit.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Exit.java
deleted file mode 100644
index 076464de2cc..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Exit.java
+++ /dev/null
@@ -1,76 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Exits the current frame to its parent frame.
- * <p>
- * Exit makes its input `data` available to the parent frame.
- * 
- * @param <T> data type for {@code output()} output
- */
-public final class Exit<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new Exit operation.
-   * 
-   * @param scope current scope
-   * @param data The tensor to be made available to the parent frame.
-   * @return a new instance of Exit
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> Exit<T> create(Scope scope, Operand<T> data) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Exit", scope.makeOpName("Exit"));
-    opBuilder.addInput(data.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new Exit<T>(opBuilder.build());
-  }
-  
-  /**
-   * The same tensor as `data`.
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Exit";
-  
-  private Output<T> output;
-  
-  private Exit(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ExpandDims.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ExpandDims.java
deleted file mode 100644
index a961f9abc77..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ExpandDims.java
+++ /dev/null
@@ -1,110 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * Inserts a dimension of 1 into a tensor's shape.
- * <p>
- * Given a tensor `input`, this operation inserts a dimension of 1 at the
- * dimension index `axis` of `input`'s shape. The dimension index `axis` starts at
- * zero; if you specify a negative number for `axis` it is counted backward from
- * the end.
- * <p>
- * This operation is useful if you want to add a batch dimension to a single
- * element. For example, if you have a single image of shape `[height, width,
- * channels]`, you can make it a batch of 1 image with `expand_dims(image, 0)`,
- * which will make the shape `[1, height, width, channels]`.
- * <p>
- * Other examples:
- * <pre>{@code
- * # 't' is a tensor of shape [2]
- * shape(expand_dims(t, 0)) ==> [1, 2]
- * shape(expand_dims(t, 1)) ==> [2, 1]
- * shape(expand_dims(t, -1)) ==> [2, 1]
- * 
- * # 't2' is a tensor of shape [2, 3, 5]
- * shape(expand_dims(t2, 0)) ==> [1, 2, 3, 5]
- * shape(expand_dims(t2, 2)) ==> [2, 3, 1, 5]
- * shape(expand_dims(t2, 3)) ==> [2, 3, 5, 1]
- * }</pre>
- * This operation requires that:
- * <p>
- * `-1-input.dims() <= dim <= input.dims()`
- * <p>
- * This operation is related to `squeeze()`, which removes dimensions of
- * size 1.
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator
-public final class ExpandDims<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new ExpandDims operation.
-   * 
-   * @param scope current scope
-   * @param input 
-   * @param axis 0-D (scalar). Specifies the dimension index at which to
-   * expand the shape of `input`. Must be in the range
-   * `[-rank(input) - 1, rank(input)]`.
-   * @return a new instance of ExpandDims
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> ExpandDims<T> create(Scope scope, Operand<T> input, Operand<U> axis) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ExpandDims", scope.makeOpName("ExpandDims"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder.addInput(axis.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new ExpandDims<T>(opBuilder.build());
-  }
-  
-  /**
-   * Contains the same data as `input`, but its shape has an additional
-   * dimension of size 1 added.
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ExpandDims";
-  
-  private Output<T> output;
-  
-  private ExpandDims(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ExtractVolumePatches.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ExtractVolumePatches.java
deleted file mode 100644
index b68e98b22cd..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ExtractVolumePatches.java
+++ /dev/null
@@ -1,102 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Extract `patches` from `input` and put them in the "depth" output dimension. 3D extension of `extract_image_patches`.
- * 
- * @param <T> data type for {@code patches()} output
- */
-@Operator
-public final class ExtractVolumePatches<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new ExtractVolumePatches operation.
-   * 
-   * @param scope current scope
-   * @param input 5-D Tensor with shape `[batch, in_planes, in_rows, in_cols, depth]`.
-   * @param ksizes The size of the sliding window for each dimension of `input`.
-   * @param strides 1-D of length 5. How far the centers of two consecutive patches are in
-   * `input`. Must be: `[1, stride_planes, stride_rows, stride_cols, 1]`.
-   * @param padding The type of padding algorithm to use.
-   * <p>
-   * We specify the size-related attributes as:
-   * <pre>{@code
-   *       ksizes = [1, ksize_planes, ksize_rows, ksize_cols, 1]
-   *       strides = [1, stride_planes, strides_rows, strides_cols, 1]
-   * }</pre>
-   * 
-   * @return a new instance of ExtractVolumePatches
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> ExtractVolumePatches<T> create(Scope scope, Operand<T> input, List<Long> ksizes, List<Long> strides, String padding) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ExtractVolumePatches", scope.makeOpName("ExtractVolumePatches"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    long[] ksizesArray = new long[ksizes.size()];
-    for (int i = 0; i < ksizesArray.length; ++i) {
-      ksizesArray[i] = ksizes.get(i);
-    }
-    opBuilder.setAttr("ksizes", ksizesArray);
-    long[] stridesArray = new long[strides.size()];
-    for (int i = 0; i < stridesArray.length; ++i) {
-      stridesArray[i] = strides.get(i);
-    }
-    opBuilder.setAttr("strides", stridesArray);
-    opBuilder.setAttr("padding", padding);
-    return new ExtractVolumePatches<T>(opBuilder.build());
-  }
-  
-  /**
-   * 5-D Tensor with shape `[batch, out_planes, out_rows, out_cols,
-   * ksize_planes * ksize_rows * ksize_cols * depth]` containing patches
-   * with size `ksize_planes x ksize_rows x ksize_cols x depth` vectorized
-   * in the "depth" dimension. Note `out_planes`, `out_rows` and `out_cols`
-   * are the dimensions of the output patches.
-   */
-  public Output<T> patches() {
-    return patches;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return patches;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ExtractVolumePatches";
-  
-  private Output<T> patches;
-  
-  private ExtractVolumePatches(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    patches = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Fill.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Fill.java
deleted file mode 100644
index 5f45e136e63..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Fill.java
+++ /dev/null
@@ -1,104 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * Creates a tensor filled with a scalar value.
- * <p>
- * This operation creates a tensor of shape `dims` and fills it with `value`.
- * <p>
- * For example:
- * <pre>{@code
- * # Output tensor has shape [2, 3].
- * fill([2, 3], 9) ==> [[9, 9, 9]
- *                      [9, 9, 9]]
- * }</pre>
- * `tf.fill` differs from `tf.constant` in a few ways:
- * <ul>
- * <li>
- * `tf.fill` only supports scalar contents, whereas `tf.constant` supports
- *     Tensor values.
- * </li>
- * <li>
- * `tf.fill` creates an Op in the computation graph that constructs the actual
- *     Tensor value at runtime. This is in contrast to `tf.constant` which embeds
- *     the entire Tensor into the graph with a `Const` node.
- * </li>
- * <li>
- * Because `tf.fill` evaluates at graph runtime, it supports dynamic shapes
- *     based on other runtime Tensors, unlike `tf.constant`.
- * 
- * @param <U> data type for {@code output()} output
- */
-@Operator
-public final class Fill<U extends TType> extends RawOp implements Operand<U> {
-  
-  /**
-   * Factory method to create a class wrapping a new Fill operation.
-   * 
-   * @param scope current scope
-   * @param dims 1-D. Represents the shape of the output tensor.
-   * @param value 0-D (scalar). Value to fill the returned tensor.
-   * <p>
-   * @compatibility(numpy)
-   * Equivalent to np.full
-   * @end_compatibility
-   * @return a new instance of Fill
-   */
-  @Endpoint(describeByClass = true)
-  public static <U extends TType, T extends TNumber> Fill<U> create(Scope scope, Operand<T> dims, Operand<U> value) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Fill", scope.makeOpName("Fill"));
-    opBuilder.addInput(dims.asOutput(scope));
-    opBuilder.addInput(value.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new Fill<U>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<U> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<U> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Fill";
-  
-  private Output<U> output;
-  
-  private Fill(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Fingerprint.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Fingerprint.java
deleted file mode 100644
index e4c2874e8c3..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Fingerprint.java
+++ /dev/null
@@ -1,109 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TString;
-import org.tensorflow.types.TUint8;
-import org.tensorflow.types.family.TType;
-
-/**
- * Generates fingerprint values.
- * <p>
- * Generates fingerprint values of `data`.
- * <p>
- * Fingerprint op considers the first dimension of `data` as the batch dimension,
- * and `output[i]` contains the fingerprint value generated from contents in
- * `data[i, ...]` for all `i`.
- * <p>
- * Fingerprint op writes fingerprint values as byte arrays. For example, the
- * default method `farmhash64` generates a 64-bit fingerprint value at a time.
- * This 8-byte value is written out as an `uint8` array of size 8, in little-endian
- * order.
- * <p>
- * For example, suppose that `data` has data type `DT_INT32` and shape (2, 3, 4),
- * and that the fingerprint method is `farmhash64`. In this case, the output shape
- * is (2, 8), where 2 is the batch dimension size of `data`, and 8 is the size of
- * each fingerprint value in bytes. `output[0, :]` is generated from 12 integers in
- * `data[0, :, :]` and similarly `output[1, :]` is generated from other 12 integers
- * in `data[1, :, :]`.
- * <p>
- * Note that this op fingerprints the raw underlying buffer, and it does not
- * fingerprint Tensor's metadata such as data type and/or shape. For example, the
- * fingerprint values are invariant under reshapes and bitcasts as long as the
- * batch dimension remain the same:
- * <pre>{@code
- * Fingerprint(data) == Fingerprint(Reshape(data, ...))
- * Fingerprint(data) == Fingerprint(Bitcast(data, ...))
- * }</pre>
- * For string data, one should expect `Fingerprint(data) !=
- * Fingerprint(ReduceJoin(data))` in general.
- */
-@Operator
-public final class Fingerprint extends RawOp implements Operand<TUint8> {
-  
-  /**
-   * Factory method to create a class wrapping a new Fingerprint operation.
-   * 
-   * @param scope current scope
-   * @param data Must have rank 1 or higher.
-   * @param method Fingerprint method used by this op. Currently available method is
-   * `farmhash::fingerprint64`.
-   * @return a new instance of Fingerprint
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> Fingerprint create(Scope scope, Operand<T> data, Operand<TString> method) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Fingerprint", scope.makeOpName("Fingerprint"));
-    opBuilder.addInput(data.asOutput(scope));
-    opBuilder.addInput(method.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new Fingerprint(opBuilder.build());
-  }
-  
-  /**
-   * A two-dimensional `Tensor` of type `tf.uint8`. The first dimension equals to
-   * `data`'s first dimension, and the second dimension size depends on the
-   * fingerprint algorithm.
-   */
-  public Output<TUint8> fingerprint() {
-    return fingerprint;
-  }
-  
-  @Override
-  public Output<TUint8> asOutput(Scope scope) {
-    return fingerprint;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Fingerprint";
-  
-  private Output<TUint8> fingerprint;
-  
-  private Fingerprint(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    fingerprint = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Gather.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Gather.java
deleted file mode 100644
index cf2f3f99ec4..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Gather.java
+++ /dev/null
@@ -1,143 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * Gather slices from `params` axis `axis` according to `indices`.
- * <p>
- * `indices` must be an integer tensor of any dimension (usually 0-D or 1-D).
- * Produces an output tensor with shape `params.shape[:axis] +
- * indices.shape[batch_dims:] + params.shape[axis + 1:]` where:
- * <pre>{@code
- *     # Scalar indices (output is rank(params) - 1).
- *     output[a_0, ..., a_n, b_0, ..., b_n] =
- *       params[a_0, ..., a_n, indices, b_0, ..., b_n]
- * 
- *     # Vector indices (output is rank(params)).
- *     output[a_0, ..., a_n, i, b_0, ..., b_n] =
- *       params[a_0, ..., a_n, indices[i], b_0, ..., b_n]
- * 
- *     # Higher rank indices (output is rank(params) + rank(indices) - 1).
- *     output[a_0, ..., a_n, i, ..., j, b_0, ... b_n] =
- *       params[a_0, ..., a_n, indices[i, ..., j], b_0, ..., b_n]
- * }</pre>
- * <div style="width:70%; margin:auto; margin-bottom:10px; margin-top:20px;">
- * <img style="width:100%" src="https://www.tensorflow.org/images/Gather.png" alt>
- * </div>
- * <p>
- * Note that on CPU, if an out of bound index is found, an error is returned.
- * On GPU, if an out of bound index is found, a 0 is stored in the
- * corresponding output value.
- * <p>
- * See also `tf.batch_gather` and `tf.gather_nd`.
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator
-public final class Gather<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.core.Gather}
-   */
-  public static class Options {
-    
-    /**
-     * @param batchDims 
-     */
-    public Options batchDims(Long batchDims) {
-      this.batchDims = batchDims;
-      return this;
-    }
-    
-    private Long batchDims;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new Gather operation.
-   * 
-   * @param scope current scope
-   * @param params The tensor from which to gather values. Must be at least rank
-   * `axis + 1`.
-   * @param indices Index tensor. Must be in range `[0, params.shape[axis])`.
-   * @param axis The axis in `params` to gather `indices` from. Defaults to the first
-   * dimension. Supports negative indexes.
-   * @param options carries optional attributes values
-   * @return a new instance of Gather
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber, V extends TNumber> Gather<T> create(Scope scope, Operand<T> params, Operand<U> indices, Operand<V> axis, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("GatherV2", scope.makeOpName("Gather"));
-    opBuilder.addInput(params.asOutput(scope));
-    opBuilder.addInput(indices.asOutput(scope));
-    opBuilder.addInput(axis.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.batchDims != null) {
-          opBuilder.setAttr("batch_dims", opts.batchDims);
-        }
-      }
-    }
-    return new Gather<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param batchDims 
-   */
-  public static Options batchDims(Long batchDims) {
-    return new Options().batchDims(batchDims);
-  }
-  
-  /**
-   * Values from `params` gathered from indices given by `indices`, with
-   * shape `params.shape[:axis] + indices.shape + params.shape[axis + 1:]`.
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "GatherV2";
-  
-  private Output<T> output;
-  
-  private Gather(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/GatherNd.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/GatherNd.java
deleted file mode 100644
index 4a10f694208..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/GatherNd.java
+++ /dev/null
@@ -1,172 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * Gather slices from `params` into a Tensor with shape specified by `indices`.
- * <p>
- * `indices` is a K-dimensional integer tensor, best thought of as a
- * (K-1)-dimensional tensor of indices into `params`, where each element defines a
- * slice of `params`:
- * <p>
- *     output[\\(i_0, ..., i_{K-2}\\)] = params[indices[\\(i_0, ..., i_{K-2}\\)]]
- * <p>
- * Whereas in `tf.gather` `indices` defines slices into the `axis`
- * dimension of `params`, in `tf.gather_nd`, `indices` defines slices into the
- * first `N` dimensions of `params`, where `N = indices.shape[-1]`.
- * <p>
- * The last dimension of `indices` can be at most the rank of
- * `params`:
- * <p>
- *     indices.shape[-1] <= params.rank
- * <p>
- * The last dimension of `indices` corresponds to elements
- * (if `indices.shape[-1] == params.rank`) or slices
- * (if `indices.shape[-1] < params.rank`) along dimension `indices.shape[-1]`
- * of `params`.  The output tensor has shape
- * <p>
- *     indices.shape[:-1] + params.shape[indices.shape[-1]:]
- * <p>
- * Note that on CPU, if an out of bound index is found, an error is returned.
- * On GPU, if an out of bound index is found, a 0 is stored in the
- * corresponding output value.
- * <p>
- * Some examples below.
- * <p>
- * Simple indexing into a matrix:
- * <pre>{@code
- *     indices = [[0, 0], [1, 1]]
- *     params = [['a', 'b'], ['c', 'd']]
- *     output = ['a', 'd']
- * }</pre>
- * Slice indexing into a matrix:
- * <pre>{@code
- *     indices = [[1], [0]]
- *     params = [['a', 'b'], ['c', 'd']]
- *     output = [['c', 'd'], ['a', 'b']]
- * }</pre>
- * Indexing into a 3-tensor:
- * <pre>{@code
- *     indices = [[1]]
- *     params = [[['a0', 'b0'], ['c0', 'd0']],
- *               [['a1', 'b1'], ['c1', 'd1']]]
- *     output = [[['a1', 'b1'], ['c1', 'd1']]]
- * 
- * 
- *     indices = [[0, 1], [1, 0]]
- *     params = [[['a0', 'b0'], ['c0', 'd0']],
- *               [['a1', 'b1'], ['c1', 'd1']]]
- *     output = [['c0', 'd0'], ['a1', 'b1']]
- * 
- * 
- *     indices = [[0, 0, 1], [1, 0, 1]]
- *     params = [[['a0', 'b0'], ['c0', 'd0']],
- *               [['a1', 'b1'], ['c1', 'd1']]]
- *     output = ['b0', 'b1']
- * }</pre>
- * Batched indexing into a matrix:
- * <pre>{@code
- *     indices = [[[0, 0]], [[0, 1]]]
- *     params = [['a', 'b'], ['c', 'd']]
- *     output = [['a'], ['b']]
- * }</pre>
- * Batched slice indexing into a matrix:
- * <pre>{@code
- *     indices = [[[1]], [[0]]]
- *     params = [['a', 'b'], ['c', 'd']]
- *     output = [[['c', 'd']], [['a', 'b']]]
- * }</pre>
- * Batched indexing into a 3-tensor:
- * <pre>{@code
- *     indices = [[[1]], [[0]]]
- *     params = [[['a0', 'b0'], ['c0', 'd0']],
- *               [['a1', 'b1'], ['c1', 'd1']]]
- *     output = [[[['a1', 'b1'], ['c1', 'd1']]],
- *               [[['a0', 'b0'], ['c0', 'd0']]]]
- * 
- *     indices = [[[0, 1], [1, 0]], [[0, 0], [1, 1]]]
- *     params = [[['a0', 'b0'], ['c0', 'd0']],
- *               [['a1', 'b1'], ['c1', 'd1']]]
- *     output = [[['c0', 'd0'], ['a1', 'b1']],
- *               [['a0', 'b0'], ['c1', 'd1']]]
- * 
- * 
- *     indices = [[[0, 0, 1], [1, 0, 1]], [[0, 1, 1], [1, 1, 0]]]
- *     params = [[['a0', 'b0'], ['c0', 'd0']],
- *               [['a1', 'b1'], ['c1', 'd1']]]
- *     output = [['b0', 'b1'], ['d0', 'c1']]
- * }</pre>
- * See also `tf.gather` and `tf.batch_gather`.
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator
-public final class GatherNd<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new GatherNd operation.
-   * 
-   * @param scope current scope
-   * @param params The tensor from which to gather values.
-   * @param indices Index tensor.
-   * @return a new instance of GatherNd
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> GatherNd<T> create(Scope scope, Operand<T> params, Operand<U> indices) {
-    OperationBuilder opBuilder = scope.env().opBuilder("GatherNd", scope.makeOpName("GatherNd"));
-    opBuilder.addInput(params.asOutput(scope));
-    opBuilder.addInput(indices.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new GatherNd<T>(opBuilder.build());
-  }
-  
-  /**
-   * Values from `params` gathered from indices given by `indices`, with
-   * shape `indices.shape[:-1] + params.shape[indices.shape[-1]:]`.
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "GatherNd";
-  
-  private Output<T> output;
-  
-  private GatherNd(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/GetSessionHandle.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/GetSessionHandle.java
deleted file mode 100644
index efde73b2a96..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/GetSessionHandle.java
+++ /dev/null
@@ -1,75 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Store the input tensor in the state of the current session.
- */
-@Operator
-public final class GetSessionHandle extends RawOp implements Operand<TType> {
-  
-  /**
-   * Factory method to create a class wrapping a new GetSessionHandle operation.
-   * 
-   * @param scope current scope
-   * @param value The tensor to be stored.
-   * @return a new instance of GetSessionHandle
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> GetSessionHandle create(Scope scope, Operand<T> value) {
-    OperationBuilder opBuilder = scope.env().opBuilder("GetSessionHandleV2", scope.makeOpName("GetSessionHandle"));
-    opBuilder.addInput(value.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new GetSessionHandle(opBuilder.build());
-  }
-  
-  /**
-   * The handle for the tensor stored in the session state, represented
-   * as a ResourceHandle object.
-   */
-  public Output<?> handle() {
-    return handle;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) handle;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "GetSessionHandleV2";
-  
-  private Output<?> handle;
-  
-  private GetSessionHandle(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    handle = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/GetSessionTensor.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/GetSessionTensor.java
deleted file mode 100644
index 067a19cfb49..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/GetSessionTensor.java
+++ /dev/null
@@ -1,78 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TString;
-import org.tensorflow.types.family.TType;
-
-/**
- * Get the value of the tensor specified by its handle.
- * 
- * @param <T> data type for {@code value()} output
- */
-@Operator
-public final class GetSessionTensor<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new GetSessionTensor operation.
-   * 
-   * @param scope current scope
-   * @param handle The handle for a tensor stored in the session state.
-   * @param dtype The type of the output value.
-   * @return a new instance of GetSessionTensor
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> GetSessionTensor<T> create(Scope scope, Operand<TString> handle, Class<T> dtype) {
-    OperationBuilder opBuilder = scope.env().opBuilder("GetSessionTensor", scope.makeOpName("GetSessionTensor"));
-    opBuilder.addInput(handle.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("dtype", dtype);
-    return new GetSessionTensor<T>(opBuilder.build());
-  }
-  
-  /**
-   * The tensor for the given handle.
-   */
-  public Output<T> value() {
-    return value;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return value;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "GetSessionTensor";
-  
-  private Output<T> value;
-  
-  private GetSessionTensor(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    value = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/GuaranteeConst.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/GuaranteeConst.java
deleted file mode 100644
index e3f4468fa28..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/GuaranteeConst.java
+++ /dev/null
@@ -1,81 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Gives a guarantee to the TF runtime that the input tensor is a constant.
- * <p>
- * The runtime is then free to make optimizations based on this.
- * <p>
- * Only accepts value typed tensors as inputs and rejects resource variable handles
- * as input.
- * <p>
- * Returns the input tensor without modification.
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator
-public final class GuaranteeConst<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new GuaranteeConst operation.
-   * 
-   * @param scope current scope
-   * @param input 
-   * @return a new instance of GuaranteeConst
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> GuaranteeConst<T> create(Scope scope, Operand<T> input) {
-    OperationBuilder opBuilder = scope.env().opBuilder("GuaranteeConst", scope.makeOpName("GuaranteeConst"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new GuaranteeConst<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "GuaranteeConst";
-  
-  private Output<T> output;
-  
-  private GuaranteeConst(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/HashTable.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/HashTable.java
deleted file mode 100644
index 49205672c44..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/HashTable.java
+++ /dev/null
@@ -1,158 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Creates a non-initialized hash table.
- * <p>
- * This op creates a hash table, specifying the type of its keys and values.
- * Before using the table you will have to initialize it.  After initialization the
- * table will be immutable.
- */
-@Operator
-public final class HashTable extends RawOp implements Operand<TType> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.core.HashTable}
-   */
-  public static class Options {
-    
-    /**
-     * @param container If non-empty, this table is placed in the given container.
-     * Otherwise, a default container is used.
-     */
-    public Options container(String container) {
-      this.container = container;
-      return this;
-    }
-    
-    /**
-     * @param sharedName If non-empty, this table is shared under the given name across
-     * multiple sessions.
-     */
-    public Options sharedName(String sharedName) {
-      this.sharedName = sharedName;
-      return this;
-    }
-    
-    /**
-     * @param useNodeNameSharing If true and shared_name is empty, the table is shared
-     * using the node name.
-     */
-    public Options useNodeNameSharing(Boolean useNodeNameSharing) {
-      this.useNodeNameSharing = useNodeNameSharing;
-      return this;
-    }
-    
-    private String container;
-    private String sharedName;
-    private Boolean useNodeNameSharing;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new HashTable operation.
-   * 
-   * @param scope current scope
-   * @param keyDtype Type of the table keys.
-   * @param valueDtype Type of the table values.
-   * @param options carries optional attributes values
-   * @return a new instance of HashTable
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TType> HashTable create(Scope scope, Class<T> keyDtype, Class<U> valueDtype, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("HashTableV2", scope.makeOpName("HashTable"));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("key_dtype", keyDtype);
-    opBuilder.setAttr("value_dtype", valueDtype);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.container != null) {
-          opBuilder.setAttr("container", opts.container);
-        }
-        if (opts.sharedName != null) {
-          opBuilder.setAttr("shared_name", opts.sharedName);
-        }
-        if (opts.useNodeNameSharing != null) {
-          opBuilder.setAttr("use_node_name_sharing", opts.useNodeNameSharing);
-        }
-      }
-    }
-    return new HashTable(opBuilder.build());
-  }
-  
-  /**
-   * @param container If non-empty, this table is placed in the given container.
-   * Otherwise, a default container is used.
-   */
-  public static Options container(String container) {
-    return new Options().container(container);
-  }
-  
-  /**
-   * @param sharedName If non-empty, this table is shared under the given name across
-   * multiple sessions.
-   */
-  public static Options sharedName(String sharedName) {
-    return new Options().sharedName(sharedName);
-  }
-  
-  /**
-   * @param useNodeNameSharing If true and shared_name is empty, the table is shared
-   * using the node name.
-   */
-  public static Options useNodeNameSharing(Boolean useNodeNameSharing) {
-    return new Options().useNodeNameSharing(useNodeNameSharing);
-  }
-  
-  /**
-   * Handle to a table.
-   */
-  public Output<?> tableHandle() {
-    return tableHandle;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) tableHandle;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "HashTableV2";
-  
-  private Output<?> tableHandle;
-  
-  private HashTable(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    tableHandle = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/HistogramFixedWidth.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/HistogramFixedWidth.java
deleted file mode 100644
index a0a2a5d873d..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/HistogramFixedWidth.java
+++ /dev/null
@@ -1,116 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Return histogram of values.
- * <p>
- * Given the tensor `values`, this operation returns a rank 1 histogram counting
- * the number of entries in `values` that fall into every bin.  The bins are
- * equal width and determined by the arguments `value_range` and `nbins`.
- * <pre>{@code
- * # Bins will be:  (-inf, 1), [1, 2), [2, 3), [3, 4), [4, inf)
- * nbins = 5
- * value_range = [0.0, 5.0]
- * new_values = [-1.0, 0.0, 1.5, 2.0, 5.0, 15]
- * 
- * with tf.get_default_session() as sess:
- *   hist = tf.histogram_fixed_width(new_values, value_range, nbins=5)
- *   variables.global_variables_initializer().run()
- *   sess.run(hist) => [2, 1, 1, 0, 2]
- * }</pre>
- * 
- * 
- * @param <U> data type for {@code out()} output
- */
-@Operator
-public final class HistogramFixedWidth<U extends TNumber> extends RawOp implements Operand<U> {
-  
-  /**
-   * Factory method to create a class wrapping a new HistogramFixedWidth operation.
-   * 
-   * @param scope current scope
-   * @param values Numeric `Tensor`.
-   * @param valueRange Shape [2] `Tensor` of same `dtype` as `values`.
-   * values <= value_range[0] will be mapped to hist[0],
-   * values >= value_range[1] will be mapped to hist[-1].
-   * @param nbins Scalar `int32 Tensor`.  Number of histogram bins.
-   * @param dtype 
-   * @return a new instance of HistogramFixedWidth
-   */
-  @Endpoint(describeByClass = true)
-  public static <U extends TNumber, T extends TNumber> HistogramFixedWidth<U> create(Scope scope, Operand<T> values, Operand<T> valueRange, Operand<TInt32> nbins, Class<U> dtype) {
-    OperationBuilder opBuilder = scope.env().opBuilder("HistogramFixedWidth", scope.makeOpName("HistogramFixedWidth"));
-    opBuilder.addInput(values.asOutput(scope));
-    opBuilder.addInput(valueRange.asOutput(scope));
-    opBuilder.addInput(nbins.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("dtype", dtype);
-    return new HistogramFixedWidth<U>(opBuilder.build());
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new HistogramFixedWidth operation using default output types.
-   * 
-   * @param scope current scope
-   * @param values Numeric `Tensor`.
-   * @param valueRange Shape [2] `Tensor` of same `dtype` as `values`.
-   * values <= value_range[0] will be mapped to hist[0],
-   * values >= value_range[1] will be mapped to hist[-1].
-   * @param nbins Scalar `int32 Tensor`.  Number of histogram bins.
-   * @return a new instance of HistogramFixedWidth
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> HistogramFixedWidth<TInt32> create(Scope scope, Operand<T> values, Operand<T> valueRange, Operand<TInt32> nbins) {
-    return create(scope, values, valueRange, nbins, TInt32.class);
-  }
-  
-  /**
-   * A 1-D `Tensor` holding histogram of values.
-   */
-  public Output<U> out() {
-    return out;
-  }
-  
-  @Override
-  public Output<U> asOutput(Scope scope) {
-    return out;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "HistogramFixedWidth";
-  
-  private Output<U> out;
-  
-  private HistogramFixedWidth(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    out = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Identity.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Identity.java
deleted file mode 100644
index c0731947bbb..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Identity.java
+++ /dev/null
@@ -1,74 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Return a tensor with the same shape and contents as the input tensor or value.
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator
-public final class Identity<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new Identity operation.
-   * 
-   * @param scope current scope
-   * @param input 
-   * @return a new instance of Identity
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> Identity<T> create(Scope scope, Operand<T> input) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Identity", scope.makeOpName("Identity"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new Identity<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Identity";
-  
-  private Output<T> output;
-  
-  private Identity(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/IdentityN.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/IdentityN.java
deleted file mode 100644
index 43364cf6d91..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/IdentityN.java
+++ /dev/null
@@ -1,95 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import java.util.Arrays;
-import java.util.Iterator;
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.Operands;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Returns a list of tensors with the same shapes and contents as the input
- * <p>
- * tensors.
- * <p>
- * This op can be used to override the gradient for complicated functions. For
- * example, suppose y = f(x) and we wish to apply a custom function g for backprop
- * such that dx = g(dy). In Python,
- * <pre>{@code
- * with tf.get_default_graph().gradient_override_map(
- *     {'IdentityN': 'OverrideGradientWithG'}):
- *   y, _ = identity_n([f(x), x])
- * 
- * @tf.RegisterGradient('OverrideGradientWithG')
- * def ApplyG(op, dy, _):
- *   return [None, g(dy)]  # Do not backprop to f(x).
- * }</pre>
- * 
- */
-@Operator
-public final class IdentityN extends RawOp implements Iterable<Operand<TType>> {
-  
-  /**
-   * Factory method to create a class wrapping a new IdentityN operation.
-   * 
-   * @param scope current scope
-   * @param input 
-   * @return a new instance of IdentityN
-   */
-  @Endpoint(describeByClass = true)
-  public static IdentityN create(Scope scope, Iterable<Operand<?>> input) {
-    OperationBuilder opBuilder = scope.env().opBuilder("IdentityN", scope.makeOpName("IdentityN"));
-    opBuilder.addInputList(Operands.asOutputs(scope, input));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new IdentityN(opBuilder.build());
-  }
-  
-  /**
-   */
-  public List<Output<?>> output() {
-    return output;
-  }
-  
-  @Override
-  @SuppressWarnings({"rawtypes", "unchecked"})
-  public Iterator<Operand<TType>> iterator() {
-    return (Iterator) output.iterator();
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "IdentityN";
-  
-  private List<Output<?>> output;
-  
-  private IdentityN(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    int outputLength = operation.outputListLength("output");
-    output = Arrays.asList(operation.outputList(outputIdx, outputLength));
-    outputIdx += outputLength;
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ImmutableConst.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ImmutableConst.java
deleted file mode 100644
index 0038baef8d5..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ImmutableConst.java
+++ /dev/null
@@ -1,82 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Returns immutable tensor from memory region.
- * <p>
- * The current implementation memmaps the tensor from a file.
- * 
- * @param <T> data type for {@code tensor()} output
- */
-@Operator
-public final class ImmutableConst<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new ImmutableConst operation.
-   * 
-   * @param scope current scope
-   * @param dtype Type of the returned tensor.
-   * @param shape Shape of the returned tensor.
-   * @param memoryRegionName Name of readonly memory region used by the tensor, see
-   * NewReadOnlyMemoryRegionFromFile in tensorflow::Env.
-   * @return a new instance of ImmutableConst
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> ImmutableConst<T> create(Scope scope, Class<T> dtype, Shape shape, String memoryRegionName) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ImmutableConst", scope.makeOpName("ImmutableConst"));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("dtype", dtype);
-    opBuilder.setAttr("shape", shape);
-    opBuilder.setAttr("memory_region_name", memoryRegionName);
-    return new ImmutableConst<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> tensor() {
-    return tensor;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return tensor;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ImmutableConst";
-  
-  private Output<T> tensor;
-  
-  private ImmutableConst(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    tensor = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/InitializeTable.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/InitializeTable.java
deleted file mode 100644
index e6e09c248b0..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/InitializeTable.java
+++ /dev/null
@@ -1,60 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Table initializer that takes two tensors for keys and values respectively.
- */
-@Operator
-public final class InitializeTable extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new InitializeTable operation.
-   * 
-   * @param scope current scope
-   * @param tableHandle Handle to a table which will be initialized.
-   * @param keys Keys of type Tkey.
-   * @param values Values of type Tval.
-   * @return a new instance of InitializeTable
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TType> InitializeTable create(Scope scope, Operand<?> tableHandle, Operand<T> keys, Operand<U> values) {
-    OperationBuilder opBuilder = scope.env().opBuilder("InitializeTableV2", scope.makeOpName("InitializeTable"));
-    opBuilder.addInput(tableHandle.asOutput(scope));
-    opBuilder.addInput(keys.asOutput(scope));
-    opBuilder.addInput(values.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new InitializeTable(opBuilder.build());
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "InitializeTableV2";
-  
-  private InitializeTable(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/InitializeTableFromTextFile.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/InitializeTableFromTextFile.java
deleted file mode 100644
index f830a07e387..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/InitializeTableFromTextFile.java
+++ /dev/null
@@ -1,127 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TString;
-
-/**
- * Initializes a table from a text file.
- * <p>
- * It inserts one key-value pair into the table for each line of the file.
- * The key and value is extracted from the whole line content, elements from the
- * split line based on `delimiter` or the line number (starting from zero).
- * Where to extract the key and value from a line is specified by `key_index` and
- * `value_index`.
- * <p>
- * - A value of -1 means use the line number(starting from zero), expects `int64`.
- * - A value of -2 means use the whole line content, expects `string`.
- * - A value >= 0 means use the index (starting at zero) of the split line based
- *   on `delimiter`.
- */
-@Operator
-public final class InitializeTableFromTextFile extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.core.InitializeTableFromTextFile}
-   */
-  public static class Options {
-    
-    /**
-     * @param vocabSize Number of elements of the file, use -1 if unknown.
-     */
-    public Options vocabSize(Long vocabSize) {
-      this.vocabSize = vocabSize;
-      return this;
-    }
-    
-    /**
-     * @param delimiter Delimiter to separate fields in a line.
-     */
-    public Options delimiter(String delimiter) {
-      this.delimiter = delimiter;
-      return this;
-    }
-    
-    private Long vocabSize;
-    private String delimiter;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new InitializeTableFromTextFile operation.
-   * 
-   * @param scope current scope
-   * @param tableHandle Handle to a table which will be initialized.
-   * @param filename Filename of a vocabulary text file.
-   * @param keyIndex Column index in a line to get the table `key` values from.
-   * @param valueIndex Column index that represents information of a line to get the table
-   * `value` values from.
-   * @param options carries optional attributes values
-   * @return a new instance of InitializeTableFromTextFile
-   */
-  @Endpoint(describeByClass = true)
-  public static InitializeTableFromTextFile create(Scope scope, Operand<?> tableHandle, Operand<TString> filename, Long keyIndex, Long valueIndex, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("InitializeTableFromTextFileV2", scope.makeOpName("InitializeTableFromTextFile"));
-    opBuilder.addInput(tableHandle.asOutput(scope));
-    opBuilder.addInput(filename.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("key_index", keyIndex);
-    opBuilder.setAttr("value_index", valueIndex);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.vocabSize != null) {
-          opBuilder.setAttr("vocab_size", opts.vocabSize);
-        }
-        if (opts.delimiter != null) {
-          opBuilder.setAttr("delimiter", opts.delimiter);
-        }
-      }
-    }
-    return new InitializeTableFromTextFile(opBuilder.build());
-  }
-  
-  /**
-   * @param vocabSize Number of elements of the file, use -1 if unknown.
-   */
-  public static Options vocabSize(Long vocabSize) {
-    return new Options().vocabSize(vocabSize);
-  }
-  
-  /**
-   * @param delimiter Delimiter to separate fields in a line.
-   */
-  public static Options delimiter(String delimiter) {
-    return new Options().delimiter(delimiter);
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "InitializeTableFromTextFileV2";
-  
-  private InitializeTableFromTextFile(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/InplaceAdd.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/InplaceAdd.java
deleted file mode 100644
index 0a84f9d71eb..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/InplaceAdd.java
+++ /dev/null
@@ -1,82 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.family.TType;
-
-/**
- *     Adds v into specified rows of x.
- * <p>
- *     Computes y = x; y[i, :] += v; return y.
- * 
- * @param <T> data type for {@code y()} output
- */
-@Operator
-public final class InplaceAdd<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new InplaceAdd operation.
-   * 
-   * @param scope current scope
-   * @param x A `Tensor` of type T.
-   * @param i A vector. Indices into the left-most dimension of `x`.
-   * @param v A `Tensor` of type T. Same dimension sizes as x except the first dimension, which must be the same as i's size.
-   * @return a new instance of InplaceAdd
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> InplaceAdd<T> create(Scope scope, Operand<T> x, Operand<TInt32> i, Operand<T> v) {
-    OperationBuilder opBuilder = scope.env().opBuilder("InplaceAdd", scope.makeOpName("InplaceAdd"));
-    opBuilder.addInput(x.asOutput(scope));
-    opBuilder.addInput(i.asOutput(scope));
-    opBuilder.addInput(v.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new InplaceAdd<T>(opBuilder.build());
-  }
-  
-  /**
-   * A `Tensor` of type T. An alias of `x`. The content of `y` is undefined if there are duplicates in `i`.
-   */
-  public Output<T> y() {
-    return y;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return y;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "InplaceAdd";
-  
-  private Output<T> y;
-  
-  private InplaceAdd(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    y = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/InplaceSub.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/InplaceSub.java
deleted file mode 100644
index ea18d37f3c0..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/InplaceSub.java
+++ /dev/null
@@ -1,82 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.family.TType;
-
-/**
- *     Subtracts `v` into specified rows of `x`.
- * <p>
- *     Computes y = x; y[i, :] -= v; return y.
- * 
- * @param <T> data type for {@code y()} output
- */
-@Operator
-public final class InplaceSub<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new InplaceSub operation.
-   * 
-   * @param scope current scope
-   * @param x A `Tensor` of type T.
-   * @param i A vector. Indices into the left-most dimension of `x`.
-   * @param v A `Tensor` of type T. Same dimension sizes as x except the first dimension, which must be the same as i's size.
-   * @return a new instance of InplaceSub
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> InplaceSub<T> create(Scope scope, Operand<T> x, Operand<TInt32> i, Operand<T> v) {
-    OperationBuilder opBuilder = scope.env().opBuilder("InplaceSub", scope.makeOpName("InplaceSub"));
-    opBuilder.addInput(x.asOutput(scope));
-    opBuilder.addInput(i.asOutput(scope));
-    opBuilder.addInput(v.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new InplaceSub<T>(opBuilder.build());
-  }
-  
-  /**
-   * A `Tensor` of type T. An alias of `x`. The content of `y` is undefined if there are duplicates in `i`.
-   */
-  public Output<T> y() {
-    return y;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return y;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "InplaceSub";
-  
-  private Output<T> y;
-  
-  private InplaceSub(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    y = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/InplaceUpdate.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/InplaceUpdate.java
deleted file mode 100644
index 691c8a39f1d..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/InplaceUpdate.java
+++ /dev/null
@@ -1,85 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.family.TType;
-
-/**
- * Updates specified rows 'i' with values 'v'.
- * <p>
- * Computes `x[i, :] = v; return x`.
- * <p>
- * Originally this function is mutative however for compilation we make this
- * operation create / operate on a copy of `x`.
- * 
- * @param <T> data type for {@code y()} output
- */
-@Operator
-public final class InplaceUpdate<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new InplaceUpdate operation.
-   * 
-   * @param scope current scope
-   * @param x A tensor of type `T`.
-   * @param i A vector. Indices into the left-most dimension of `x`.
-   * @param v A `Tensor` of type T. Same dimension sizes as x except the first dimension, which must be the same as i's size.
-   * @return a new instance of InplaceUpdate
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> InplaceUpdate<T> create(Scope scope, Operand<T> x, Operand<TInt32> i, Operand<T> v) {
-    OperationBuilder opBuilder = scope.env().opBuilder("InplaceUpdate", scope.makeOpName("InplaceUpdate"));
-    opBuilder.addInput(x.asOutput(scope));
-    opBuilder.addInput(i.asOutput(scope));
-    opBuilder.addInput(v.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new InplaceUpdate<T>(opBuilder.build());
-  }
-  
-  /**
-   * A `Tensor` of type T. An alias of `x`. The content of `y` is undefined if there are duplicates in `i`.
-   */
-  public Output<T> y() {
-    return y;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return y;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "InplaceUpdate";
-  
-  private Output<T> y;
-  
-  private InplaceUpdate(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    y = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/IsVariableInitialized.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/IsVariableInitialized.java
deleted file mode 100644
index 28aae03928c..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/IsVariableInitialized.java
+++ /dev/null
@@ -1,75 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TBool;
-import org.tensorflow.types.family.TType;
-
-/**
- * Checks whether a tensor has been initialized.
- * <p>
- * Outputs boolean scalar indicating whether the tensor has been initialized.
- */
-@Operator
-public final class IsVariableInitialized extends RawOp implements Operand<TBool> {
-  
-  /**
-   * Factory method to create a class wrapping a new IsVariableInitialized operation.
-   * 
-   * @param scope current scope
-   * @param ref Should be from a `Variable` node. May be uninitialized.
-   * @return a new instance of IsVariableInitialized
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> IsVariableInitialized create(Scope scope, Operand<T> ref) {
-    OperationBuilder opBuilder = scope.env().opBuilder("IsVariableInitialized", scope.makeOpName("IsVariableInitialized"));
-    opBuilder.addInput(ref.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new IsVariableInitialized(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<TBool> isInitialized() {
-    return isInitialized;
-  }
-  
-  @Override
-  public Output<TBool> asOutput(Scope scope) {
-    return isInitialized;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "IsVariableInitialized";
-  
-  private Output<TBool> isInitialized;
-  
-  private IsVariableInitialized(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    isInitialized = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LinSpace.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LinSpace.java
deleted file mode 100644
index 095d5bf6b9c..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LinSpace.java
+++ /dev/null
@@ -1,88 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Generates values in an interval.
- * <p>
- * A sequence of `num` evenly-spaced values are generated beginning at `start`.
- * If `num > 1`, the values in the sequence increase by `stop - start / num - 1`,
- * so that the last one is exactly `stop`.
- * <p>
- * For example:
- * <pre>{@code
- * tf.linspace(10.0, 12.0, 3, name="linspace") => [ 10.0  11.0  12.0]
- * }</pre>
- * 
- * 
- * @param <T> data type for {@code output()} output
- */
-public final class LinSpace<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new LinSpace operation.
-   * 
-   * @param scope current scope
-   * @param start 0-D tensor. First entry in the range.
-   * @param stop 0-D tensor. Last entry in the range.
-   * @param num 0-D tensor. Number of values to generate.
-   * @return a new instance of LinSpace
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber, U extends TNumber> LinSpace<T> create(Scope scope, Operand<T> start, Operand<T> stop, Operand<U> num) {
-    OperationBuilder opBuilder = scope.env().opBuilder("LinSpace", scope.makeOpName("LinSpace"));
-    opBuilder.addInput(start.asOutput(scope));
-    opBuilder.addInput(stop.asOutput(scope));
-    opBuilder.addInput(num.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new LinSpace<T>(opBuilder.build());
-  }
-  
-  /**
-   * 1-D. The generated values.
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "LinSpace";
-  
-  private Output<T> output;
-  
-  private LinSpace(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LookupTableExport.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LookupTableExport.java
deleted file mode 100644
index 598c93b5322..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LookupTableExport.java
+++ /dev/null
@@ -1,84 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Outputs all keys and values in the table.
- * 
- * @param <T> data type for {@code keys()} output
- * @param <U> data type for {@code values()} output
- */
-@Operator
-public final class LookupTableExport<T extends TType, U extends TType> extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new LookupTableExport operation.
-   * 
-   * @param scope current scope
-   * @param tableHandle Handle to the table.
-   * @param Tkeys 
-   * @param Tvalues 
-   * @return a new instance of LookupTableExport
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TType> LookupTableExport<T, U> create(Scope scope, Operand<?> tableHandle, Class<T> Tkeys, Class<U> Tvalues) {
-    OperationBuilder opBuilder = scope.env().opBuilder("LookupTableExportV2", scope.makeOpName("LookupTableExport"));
-    opBuilder.addInput(tableHandle.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("Tkeys", Tkeys);
-    opBuilder.setAttr("Tvalues", Tvalues);
-    return new LookupTableExport<T, U>(opBuilder.build());
-  }
-  
-  /**
-   * Vector of all keys present in the table.
-   */
-  public Output<T> keys() {
-    return keys;
-  }
-  
-  /**
-   * Tensor of all values in the table. Indexed in parallel with `keys`.
-   */
-  public Output<U> values() {
-    return values;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "LookupTableExportV2";
-  
-  private Output<T> keys;
-  private Output<U> values;
-  
-  private LookupTableExport(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    keys = operation.output(outputIdx++);
-    values = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LookupTableFind.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LookupTableFind.java
deleted file mode 100644
index 381907fabce..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LookupTableFind.java
+++ /dev/null
@@ -1,86 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Looks up keys in a table, outputs the corresponding values.
- * <p>
- * The tensor `keys` must of the same type as the keys of the table.
- * The output `values` is of the type of the table values.
- * <p>
- * The scalar `default_value` is the value output for keys not present in the
- * table. It must also be of the same type as the table values.
- * 
- * @param <U> data type for {@code values()} output
- */
-@Operator
-public final class LookupTableFind<U extends TType> extends RawOp implements Operand<U> {
-  
-  /**
-   * Factory method to create a class wrapping a new LookupTableFind operation.
-   * 
-   * @param scope current scope
-   * @param tableHandle Handle to the table.
-   * @param keys Any shape.  Keys to look up.
-   * @param defaultValue 
-   * @return a new instance of LookupTableFind
-   */
-  @Endpoint(describeByClass = true)
-  public static <U extends TType, T extends TType> LookupTableFind<U> create(Scope scope, Operand<?> tableHandle, Operand<T> keys, Operand<U> defaultValue) {
-    OperationBuilder opBuilder = scope.env().opBuilder("LookupTableFindV2", scope.makeOpName("LookupTableFind"));
-    opBuilder.addInput(tableHandle.asOutput(scope));
-    opBuilder.addInput(keys.asOutput(scope));
-    opBuilder.addInput(defaultValue.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new LookupTableFind<U>(opBuilder.build());
-  }
-  
-  /**
-   * Same shape as `keys`.  Values found in the table, or `default_values`
-   * for missing keys.
-   */
-  public Output<U> values() {
-    return values;
-  }
-  
-  @Override
-  public Output<U> asOutput(Scope scope) {
-    return values;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "LookupTableFindV2";
-  
-  private Output<U> values;
-  
-  private LookupTableFind(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    values = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LookupTableImport.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LookupTableImport.java
deleted file mode 100644
index 4b5121b7911..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LookupTableImport.java
+++ /dev/null
@@ -1,63 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Replaces the contents of the table with the specified keys and values.
- * <p>
- * The tensor `keys` must be of the same type as the keys of the table.
- * The tensor `values` must be of the type of the table values.
- */
-@Operator
-public final class LookupTableImport extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new LookupTableImport operation.
-   * 
-   * @param scope current scope
-   * @param tableHandle Handle to the table.
-   * @param keys Any shape.  Keys to look up.
-   * @param values Values to associate with keys.
-   * @return a new instance of LookupTableImport
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TType> LookupTableImport create(Scope scope, Operand<?> tableHandle, Operand<T> keys, Operand<U> values) {
-    OperationBuilder opBuilder = scope.env().opBuilder("LookupTableImportV2", scope.makeOpName("LookupTableImport"));
-    opBuilder.addInput(tableHandle.asOutput(scope));
-    opBuilder.addInput(keys.asOutput(scope));
-    opBuilder.addInput(values.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new LookupTableImport(opBuilder.build());
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "LookupTableImportV2";
-  
-  private LookupTableImport(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LookupTableInsert.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LookupTableInsert.java
deleted file mode 100644
index 6276b435130..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LookupTableInsert.java
+++ /dev/null
@@ -1,63 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Updates the table to associates keys with values.
- * <p>
- * The tensor `keys` must be of the same type as the keys of the table.
- * The tensor `values` must be of the type of the table values.
- */
-@Operator
-public final class LookupTableInsert extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new LookupTableInsert operation.
-   * 
-   * @param scope current scope
-   * @param tableHandle Handle to the table.
-   * @param keys Any shape.  Keys to look up.
-   * @param values Values to associate with keys.
-   * @return a new instance of LookupTableInsert
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TType> LookupTableInsert create(Scope scope, Operand<?> tableHandle, Operand<T> keys, Operand<U> values) {
-    OperationBuilder opBuilder = scope.env().opBuilder("LookupTableInsertV2", scope.makeOpName("LookupTableInsert"));
-    opBuilder.addInput(tableHandle.asOutput(scope));
-    opBuilder.addInput(keys.asOutput(scope));
-    opBuilder.addInput(values.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new LookupTableInsert(opBuilder.build());
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "LookupTableInsertV2";
-  
-  private LookupTableInsert(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LookupTableRemove.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LookupTableRemove.java
deleted file mode 100644
index ee86784c34f..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LookupTableRemove.java
+++ /dev/null
@@ -1,60 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Removes keys and its associated values from a table.
- * <p>
- * The tensor `keys` must of the same type as the keys of the table. Keys not
- * already in the table are silently ignored.
- */
-public final class LookupTableRemove extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new LookupTableRemove operation.
-   * 
-   * @param scope current scope
-   * @param tableHandle Handle to the table.
-   * @param keys Any shape.  Keys of the elements to remove.
-   * @return a new instance of LookupTableRemove
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> LookupTableRemove create(Scope scope, Operand<?> tableHandle, Operand<T> keys) {
-    OperationBuilder opBuilder = scope.env().opBuilder("LookupTableRemoveV2", scope.makeOpName("LookupTableRemove"));
-    opBuilder.addInput(tableHandle.asOutput(scope));
-    opBuilder.addInput(keys.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new LookupTableRemove(opBuilder.build());
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "LookupTableRemoveV2";
-  
-  private LookupTableRemove(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LookupTableSize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LookupTableSize.java
deleted file mode 100644
index a362eb3c54c..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LookupTableSize.java
+++ /dev/null
@@ -1,73 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt64;
-
-/**
- * Computes the number of elements in the given table.
- */
-@Operator
-public final class LookupTableSize extends RawOp implements Operand<TInt64> {
-  
-  /**
-   * Factory method to create a class wrapping a new LookupTableSize operation.
-   * 
-   * @param scope current scope
-   * @param tableHandle Handle to the table.
-   * @return a new instance of LookupTableSize
-   */
-  @Endpoint(describeByClass = true)
-  public static LookupTableSize create(Scope scope, Operand<?> tableHandle) {
-    OperationBuilder opBuilder = scope.env().opBuilder("LookupTableSizeV2", scope.makeOpName("LookupTableSize"));
-    opBuilder.addInput(tableHandle.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new LookupTableSize(opBuilder.build());
-  }
-  
-  /**
-   * Scalar that contains number of elements in the table.
-   */
-  public Output<TInt64> size() {
-    return size;
-  }
-  
-  @Override
-  public Output<TInt64> asOutput(Scope scope) {
-    return size;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "LookupTableSizeV2";
-  
-  private Output<TInt64> size;
-  
-  private LookupTableSize(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    size = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LoopCond.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LoopCond.java
deleted file mode 100644
index 59ac003a6f7..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LoopCond.java
+++ /dev/null
@@ -1,76 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TBool;
-
-/**
- * Forwards the input to the output.
- * <p>
- * This operator represents the loop termination condition used by the
- * "pivot" switches of a loop.
- */
-@Operator
-public final class LoopCond extends RawOp implements Operand<TBool> {
-  
-  /**
-   * Factory method to create a class wrapping a new LoopCond operation.
-   * 
-   * @param scope current scope
-   * @param input A boolean scalar, representing the branch predicate of the Switch op.
-   * @return a new instance of LoopCond
-   */
-  @Endpoint(describeByClass = true)
-  public static LoopCond create(Scope scope, Operand<TBool> input) {
-    OperationBuilder opBuilder = scope.env().opBuilder("LoopCond", scope.makeOpName("LoopCond"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new LoopCond(opBuilder.build());
-  }
-  
-  /**
-   * The same tensor as `input`.
-   */
-  public Output<TBool> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<TBool> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "LoopCond";
-  
-  private Output<TBool> output;
-  
-  private LoopCond(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LowerBound.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LowerBound.java
deleted file mode 100644
index 64acd0bba13..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LowerBound.java
+++ /dev/null
@@ -1,115 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * Applies lower_bound(sorted_search_values, values) along each row.
- * <p>
- * Each set of rows with the same index in (sorted_inputs, values) is treated
- * independently.  The resulting row is the equivalent of calling
- * `np.searchsorted(sorted_inputs, values, side='left')`.
- * <p>
- * The result is not a global index to the entire
- * `Tensor`, but rather just the index in the last dimension.
- * <p>
- * A 2-D example:
- *   sorted_sequence = [[0, 3, 9, 9, 10],
- *                      [1, 2, 3, 4, 5]]
- *   values = [[2, 4, 9],
- *             [0, 2, 6]]
- * <p>
- *   result = LowerBound(sorted_sequence, values)
- * <p>
- *   result == [[1, 2, 2],
- *              [0, 1, 5]]
- * 
- * @param <U> data type for {@code output()} output
- */
-public final class LowerBound<U extends TNumber> extends RawOp implements Operand<U> {
-  
-  /**
-   * Factory method to create a class wrapping a new LowerBound operation.
-   * 
-   * @param scope current scope
-   * @param sortedInputs 2-D Tensor where each row is ordered.
-   * @param values 2-D Tensor with the same numbers of rows as `sorted_search_values`. Contains
-   * the values that will be searched for in `sorted_search_values`.
-   * @param outType 
-   * @return a new instance of LowerBound
-   */
-  @Endpoint(describeByClass = true)
-  public static <U extends TNumber, T extends TType> LowerBound<U> create(Scope scope, Operand<T> sortedInputs, Operand<T> values, Class<U> outType) {
-    OperationBuilder opBuilder = scope.env().opBuilder("LowerBound", scope.makeOpName("LowerBound"));
-    opBuilder.addInput(sortedInputs.asOutput(scope));
-    opBuilder.addInput(values.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("out_type", outType);
-    return new LowerBound<U>(opBuilder.build());
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new LowerBound operation using default output types.
-   * 
-   * @param scope current scope
-   * @param sortedInputs 2-D Tensor where each row is ordered.
-   * @param values 2-D Tensor with the same numbers of rows as `sorted_search_values`. Contains
-   * the values that will be searched for in `sorted_search_values`.
-   * @return a new instance of LowerBound
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> LowerBound<TInt32> create(Scope scope, Operand<T> sortedInputs, Operand<T> values) {
-    return create(scope, sortedInputs, values, TInt32.class);
-  }
-  
-  /**
-   * A `Tensor` with the same shape as `values`.  It contains the first scalar index
-   * into the last dimension where values can be inserted without changing the
-   * ordered property.
-   */
-  public Output<U> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<U> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "LowerBound";
-  
-  private Output<U> output;
-  
-  private LowerBound(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MapClear.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MapClear.java
deleted file mode 100644
index bd24fcfef5f..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MapClear.java
+++ /dev/null
@@ -1,150 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import java.util.List;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-
-/**
- * Op removes all elements in the underlying container.
- */
-@Operator
-public final class MapClear extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.core.MapClear}
-   */
-  public static class Options {
-    
-    /**
-     * @param capacity 
-     */
-    public Options capacity(Long capacity) {
-      this.capacity = capacity;
-      return this;
-    }
-    
-    /**
-     * @param memoryLimit 
-     */
-    public Options memoryLimit(Long memoryLimit) {
-      this.memoryLimit = memoryLimit;
-      return this;
-    }
-    
-    /**
-     * @param container 
-     */
-    public Options container(String container) {
-      this.container = container;
-      return this;
-    }
-    
-    /**
-     * @param sharedName 
-     */
-    public Options sharedName(String sharedName) {
-      this.sharedName = sharedName;
-      return this;
-    }
-    
-    private Long capacity;
-    private Long memoryLimit;
-    private String container;
-    private String sharedName;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new MapClear operation.
-   * 
-   * @param scope current scope
-   * @param dtypes 
-   * @param options carries optional attributes values
-   * @return a new instance of MapClear
-   */
-  @Endpoint(describeByClass = true)
-  public static MapClear create(Scope scope, List<Class<?>> dtypes, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("MapClear", scope.makeOpName("MapClear"));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    Class[] dtypesArray = new Class[dtypes.size()];
-    for (int i = 0; i < dtypesArray.length; ++i) {
-      dtypesArray[i] = dtypes.get(i);
-    }
-    opBuilder.setAttr("dtypes", dtypesArray);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.capacity != null) {
-          opBuilder.setAttr("capacity", opts.capacity);
-        }
-        if (opts.memoryLimit != null) {
-          opBuilder.setAttr("memory_limit", opts.memoryLimit);
-        }
-        if (opts.container != null) {
-          opBuilder.setAttr("container", opts.container);
-        }
-        if (opts.sharedName != null) {
-          opBuilder.setAttr("shared_name", opts.sharedName);
-        }
-      }
-    }
-    return new MapClear(opBuilder.build());
-  }
-  
-  /**
-   * @param capacity 
-   */
-  public static Options capacity(Long capacity) {
-    return new Options().capacity(capacity);
-  }
-  
-  /**
-   * @param memoryLimit 
-   */
-  public static Options memoryLimit(Long memoryLimit) {
-    return new Options().memoryLimit(memoryLimit);
-  }
-  
-  /**
-   * @param container 
-   */
-  public static Options container(String container) {
-    return new Options().container(container);
-  }
-  
-  /**
-   * @param sharedName 
-   */
-  public static Options sharedName(String sharedName) {
-    return new Options().sharedName(sharedName);
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "MapClear";
-  
-  private MapClear(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MapIncompleteSize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MapIncompleteSize.java
deleted file mode 100644
index 173abfcc3b0..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MapIncompleteSize.java
+++ /dev/null
@@ -1,168 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
-
-/**
- * Op returns the number of incomplete elements in the underlying container.
- */
-@Operator
-public final class MapIncompleteSize extends RawOp implements Operand<TInt32> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.core.MapIncompleteSize}
-   */
-  public static class Options {
-    
-    /**
-     * @param capacity 
-     */
-    public Options capacity(Long capacity) {
-      this.capacity = capacity;
-      return this;
-    }
-    
-    /**
-     * @param memoryLimit 
-     */
-    public Options memoryLimit(Long memoryLimit) {
-      this.memoryLimit = memoryLimit;
-      return this;
-    }
-    
-    /**
-     * @param container 
-     */
-    public Options container(String container) {
-      this.container = container;
-      return this;
-    }
-    
-    /**
-     * @param sharedName 
-     */
-    public Options sharedName(String sharedName) {
-      this.sharedName = sharedName;
-      return this;
-    }
-    
-    private Long capacity;
-    private Long memoryLimit;
-    private String container;
-    private String sharedName;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new MapIncompleteSize operation.
-   * 
-   * @param scope current scope
-   * @param dtypes 
-   * @param options carries optional attributes values
-   * @return a new instance of MapIncompleteSize
-   */
-  @Endpoint(describeByClass = true)
-  public static MapIncompleteSize create(Scope scope, List<Class<?>> dtypes, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("MapIncompleteSize", scope.makeOpName("MapIncompleteSize"));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    Class[] dtypesArray = new Class[dtypes.size()];
-    for (int i = 0; i < dtypesArray.length; ++i) {
-      dtypesArray[i] = dtypes.get(i);
-    }
-    opBuilder.setAttr("dtypes", dtypesArray);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.capacity != null) {
-          opBuilder.setAttr("capacity", opts.capacity);
-        }
-        if (opts.memoryLimit != null) {
-          opBuilder.setAttr("memory_limit", opts.memoryLimit);
-        }
-        if (opts.container != null) {
-          opBuilder.setAttr("container", opts.container);
-        }
-        if (opts.sharedName != null) {
-          opBuilder.setAttr("shared_name", opts.sharedName);
-        }
-      }
-    }
-    return new MapIncompleteSize(opBuilder.build());
-  }
-  
-  /**
-   * @param capacity 
-   */
-  public static Options capacity(Long capacity) {
-    return new Options().capacity(capacity);
-  }
-  
-  /**
-   * @param memoryLimit 
-   */
-  public static Options memoryLimit(Long memoryLimit) {
-    return new Options().memoryLimit(memoryLimit);
-  }
-  
-  /**
-   * @param container 
-   */
-  public static Options container(String container) {
-    return new Options().container(container);
-  }
-  
-  /**
-   * @param sharedName 
-   */
-  public static Options sharedName(String sharedName) {
-    return new Options().sharedName(sharedName);
-  }
-  
-  /**
-   */
-  public Output<TInt32> size() {
-    return size;
-  }
-  
-  @Override
-  public Output<TInt32> asOutput(Scope scope) {
-    return size;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "MapIncompleteSize";
-  
-  private Output<TInt32> size;
-  
-  private MapIncompleteSize(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    size = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MapPeek.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MapPeek.java
deleted file mode 100644
index f2df24b15b8..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MapPeek.java
+++ /dev/null
@@ -1,182 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import java.util.Arrays;
-import java.util.Iterator;
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
-
-/**
- * Op peeks at the values at the specified key.  If the
- * <p>
- * underlying container does not contain this key
- * this op will block until it does.
- */
-@Operator
-public final class MapPeek extends RawOp implements Iterable<Operand<TType>> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.core.MapPeek}
-   */
-  public static class Options {
-    
-    /**
-     * @param capacity 
-     */
-    public Options capacity(Long capacity) {
-      this.capacity = capacity;
-      return this;
-    }
-    
-    /**
-     * @param memoryLimit 
-     */
-    public Options memoryLimit(Long memoryLimit) {
-      this.memoryLimit = memoryLimit;
-      return this;
-    }
-    
-    /**
-     * @param container 
-     */
-    public Options container(String container) {
-      this.container = container;
-      return this;
-    }
-    
-    /**
-     * @param sharedName 
-     */
-    public Options sharedName(String sharedName) {
-      this.sharedName = sharedName;
-      return this;
-    }
-    
-    private Long capacity;
-    private Long memoryLimit;
-    private String container;
-    private String sharedName;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new MapPeek operation.
-   * 
-   * @param scope current scope
-   * @param key 
-   * @param indices 
-   * @param dtypes 
-   * @param options carries optional attributes values
-   * @return a new instance of MapPeek
-   */
-  @Endpoint(describeByClass = true)
-  public static MapPeek create(Scope scope, Operand<TInt64> key, Operand<TInt32> indices, List<Class<?>> dtypes, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("MapPeek", scope.makeOpName("MapPeek"));
-    opBuilder.addInput(key.asOutput(scope));
-    opBuilder.addInput(indices.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    Class[] dtypesArray = new Class[dtypes.size()];
-    for (int i = 0; i < dtypesArray.length; ++i) {
-      dtypesArray[i] = dtypes.get(i);
-    }
-    opBuilder.setAttr("dtypes", dtypesArray);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.capacity != null) {
-          opBuilder.setAttr("capacity", opts.capacity);
-        }
-        if (opts.memoryLimit != null) {
-          opBuilder.setAttr("memory_limit", opts.memoryLimit);
-        }
-        if (opts.container != null) {
-          opBuilder.setAttr("container", opts.container);
-        }
-        if (opts.sharedName != null) {
-          opBuilder.setAttr("shared_name", opts.sharedName);
-        }
-      }
-    }
-    return new MapPeek(opBuilder.build());
-  }
-  
-  /**
-   * @param capacity 
-   */
-  public static Options capacity(Long capacity) {
-    return new Options().capacity(capacity);
-  }
-  
-  /**
-   * @param memoryLimit 
-   */
-  public static Options memoryLimit(Long memoryLimit) {
-    return new Options().memoryLimit(memoryLimit);
-  }
-  
-  /**
-   * @param container 
-   */
-  public static Options container(String container) {
-    return new Options().container(container);
-  }
-  
-  /**
-   * @param sharedName 
-   */
-  public static Options sharedName(String sharedName) {
-    return new Options().sharedName(sharedName);
-  }
-  
-  /**
-   */
-  public List<Output<?>> values() {
-    return values;
-  }
-  
-  @Override
-  @SuppressWarnings({"rawtypes", "unchecked"})
-  public Iterator<Operand<TType>> iterator() {
-    return (Iterator) values.iterator();
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "MapPeek";
-  
-  private List<Output<?>> values;
-  
-  private MapPeek(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    int valuesLength = operation.outputListLength("values");
-    values = Arrays.asList(operation.outputList(outputIdx, valuesLength));
-    outputIdx += valuesLength;
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MapSize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MapSize.java
deleted file mode 100644
index 58d13795e86..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MapSize.java
+++ /dev/null
@@ -1,168 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
-
-/**
- * Op returns the number of elements in the underlying container.
- */
-@Operator
-public final class MapSize extends RawOp implements Operand<TInt32> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.core.MapSize}
-   */
-  public static class Options {
-    
-    /**
-     * @param capacity 
-     */
-    public Options capacity(Long capacity) {
-      this.capacity = capacity;
-      return this;
-    }
-    
-    /**
-     * @param memoryLimit 
-     */
-    public Options memoryLimit(Long memoryLimit) {
-      this.memoryLimit = memoryLimit;
-      return this;
-    }
-    
-    /**
-     * @param container 
-     */
-    public Options container(String container) {
-      this.container = container;
-      return this;
-    }
-    
-    /**
-     * @param sharedName 
-     */
-    public Options sharedName(String sharedName) {
-      this.sharedName = sharedName;
-      return this;
-    }
-    
-    private Long capacity;
-    private Long memoryLimit;
-    private String container;
-    private String sharedName;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new MapSize operation.
-   * 
-   * @param scope current scope
-   * @param dtypes 
-   * @param options carries optional attributes values
-   * @return a new instance of MapSize
-   */
-  @Endpoint(describeByClass = true)
-  public static MapSize create(Scope scope, List<Class<?>> dtypes, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("MapSize", scope.makeOpName("MapSize"));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    Class[] dtypesArray = new Class[dtypes.size()];
-    for (int i = 0; i < dtypesArray.length; ++i) {
-      dtypesArray[i] = dtypes.get(i);
-    }
-    opBuilder.setAttr("dtypes", dtypesArray);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.capacity != null) {
-          opBuilder.setAttr("capacity", opts.capacity);
-        }
-        if (opts.memoryLimit != null) {
-          opBuilder.setAttr("memory_limit", opts.memoryLimit);
-        }
-        if (opts.container != null) {
-          opBuilder.setAttr("container", opts.container);
-        }
-        if (opts.sharedName != null) {
-          opBuilder.setAttr("shared_name", opts.sharedName);
-        }
-      }
-    }
-    return new MapSize(opBuilder.build());
-  }
-  
-  /**
-   * @param capacity 
-   */
-  public static Options capacity(Long capacity) {
-    return new Options().capacity(capacity);
-  }
-  
-  /**
-   * @param memoryLimit 
-   */
-  public static Options memoryLimit(Long memoryLimit) {
-    return new Options().memoryLimit(memoryLimit);
-  }
-  
-  /**
-   * @param container 
-   */
-  public static Options container(String container) {
-    return new Options().container(container);
-  }
-  
-  /**
-   * @param sharedName 
-   */
-  public static Options sharedName(String sharedName) {
-    return new Options().sharedName(sharedName);
-  }
-  
-  /**
-   */
-  public Output<TInt32> size() {
-    return size;
-  }
-  
-  @Override
-  public Output<TInt32> asOutput(Scope scope) {
-    return size;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "MapSize";
-  
-  private Output<TInt32> size;
-  
-  private MapSize(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    size = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MapStage.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MapStage.java
deleted file mode 100644
index 8a9741a417f..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MapStage.java
+++ /dev/null
@@ -1,165 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.Operands;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.TInt64;
-
-/**
- * Stage (key, values) in the underlying container which behaves like a hashtable.
- */
-@Operator
-public final class MapStage extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.core.MapStage}
-   */
-  public static class Options {
-    
-    /**
-     * @param capacity Maximum number of elements in the Staging Area. If > 0, inserts
-     * on the container will block when the capacity is reached.
-     */
-    public Options capacity(Long capacity) {
-      this.capacity = capacity;
-      return this;
-    }
-    
-    /**
-     * @param memoryLimit 
-     */
-    public Options memoryLimit(Long memoryLimit) {
-      this.memoryLimit = memoryLimit;
-      return this;
-    }
-    
-    /**
-     * @param container If non-empty, this queue is placed in the given container. Otherwise,
-     * a default container is used.
-     */
-    public Options container(String container) {
-      this.container = container;
-      return this;
-    }
-    
-    /**
-     * @param sharedName It is necessary to match this name to the matching Unstage Op.
-     */
-    public Options sharedName(String sharedName) {
-      this.sharedName = sharedName;
-      return this;
-    }
-    
-    private Long capacity;
-    private Long memoryLimit;
-    private String container;
-    private String sharedName;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new MapStage operation.
-   * 
-   * @param scope current scope
-   * @param key int64
-   * @param indices 
-   * @param values a list of tensors
-   * dtypes A list of data types that inserted values should adhere to.
-   * @param dtypes 
-   * @param options carries optional attributes values
-   * @return a new instance of MapStage
-   */
-  @Endpoint(describeByClass = true)
-  public static MapStage create(Scope scope, Operand<TInt64> key, Operand<TInt32> indices, Iterable<Operand<?>> values, List<Class<?>> dtypes, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("MapStage", scope.makeOpName("MapStage"));
-    opBuilder.addInput(key.asOutput(scope));
-    opBuilder.addInput(indices.asOutput(scope));
-    opBuilder.addInputList(Operands.asOutputs(scope, values));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    Class[] dtypesArray = new Class[dtypes.size()];
-    for (int i = 0; i < dtypesArray.length; ++i) {
-      dtypesArray[i] = dtypes.get(i);
-    }
-    opBuilder.setAttr("dtypes", dtypesArray);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.capacity != null) {
-          opBuilder.setAttr("capacity", opts.capacity);
-        }
-        if (opts.memoryLimit != null) {
-          opBuilder.setAttr("memory_limit", opts.memoryLimit);
-        }
-        if (opts.container != null) {
-          opBuilder.setAttr("container", opts.container);
-        }
-        if (opts.sharedName != null) {
-          opBuilder.setAttr("shared_name", opts.sharedName);
-        }
-      }
-    }
-    return new MapStage(opBuilder.build());
-  }
-  
-  /**
-   * @param capacity Maximum number of elements in the Staging Area. If > 0, inserts
-   * on the container will block when the capacity is reached.
-   */
-  public static Options capacity(Long capacity) {
-    return new Options().capacity(capacity);
-  }
-  
-  /**
-   * @param memoryLimit 
-   */
-  public static Options memoryLimit(Long memoryLimit) {
-    return new Options().memoryLimit(memoryLimit);
-  }
-  
-  /**
-   * @param container If non-empty, this queue is placed in the given container. Otherwise,
-   * a default container is used.
-   */
-  public static Options container(String container) {
-    return new Options().container(container);
-  }
-  
-  /**
-   * @param sharedName It is necessary to match this name to the matching Unstage Op.
-   */
-  public static Options sharedName(String sharedName) {
-    return new Options().sharedName(sharedName);
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "MapStage";
-  
-  private MapStage(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MapUnstage.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MapUnstage.java
deleted file mode 100644
index 07442b96b01..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MapUnstage.java
+++ /dev/null
@@ -1,182 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import java.util.Arrays;
-import java.util.Iterator;
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
-
-/**
- * Op removes and returns the values associated with the key
- * <p>
- * from the underlying container.   If the underlying container
- * does not contain this key, the op will block until it does.
- */
-@Operator
-public final class MapUnstage extends RawOp implements Iterable<Operand<TType>> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.core.MapUnstage}
-   */
-  public static class Options {
-    
-    /**
-     * @param capacity 
-     */
-    public Options capacity(Long capacity) {
-      this.capacity = capacity;
-      return this;
-    }
-    
-    /**
-     * @param memoryLimit 
-     */
-    public Options memoryLimit(Long memoryLimit) {
-      this.memoryLimit = memoryLimit;
-      return this;
-    }
-    
-    /**
-     * @param container 
-     */
-    public Options container(String container) {
-      this.container = container;
-      return this;
-    }
-    
-    /**
-     * @param sharedName 
-     */
-    public Options sharedName(String sharedName) {
-      this.sharedName = sharedName;
-      return this;
-    }
-    
-    private Long capacity;
-    private Long memoryLimit;
-    private String container;
-    private String sharedName;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new MapUnstage operation.
-   * 
-   * @param scope current scope
-   * @param key 
-   * @param indices 
-   * @param dtypes 
-   * @param options carries optional attributes values
-   * @return a new instance of MapUnstage
-   */
-  @Endpoint(describeByClass = true)
-  public static MapUnstage create(Scope scope, Operand<TInt64> key, Operand<TInt32> indices, List<Class<?>> dtypes, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("MapUnstage", scope.makeOpName("MapUnstage"));
-    opBuilder.addInput(key.asOutput(scope));
-    opBuilder.addInput(indices.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    Class[] dtypesArray = new Class[dtypes.size()];
-    for (int i = 0; i < dtypesArray.length; ++i) {
-      dtypesArray[i] = dtypes.get(i);
-    }
-    opBuilder.setAttr("dtypes", dtypesArray);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.capacity != null) {
-          opBuilder.setAttr("capacity", opts.capacity);
-        }
-        if (opts.memoryLimit != null) {
-          opBuilder.setAttr("memory_limit", opts.memoryLimit);
-        }
-        if (opts.container != null) {
-          opBuilder.setAttr("container", opts.container);
-        }
-        if (opts.sharedName != null) {
-          opBuilder.setAttr("shared_name", opts.sharedName);
-        }
-      }
-    }
-    return new MapUnstage(opBuilder.build());
-  }
-  
-  /**
-   * @param capacity 
-   */
-  public static Options capacity(Long capacity) {
-    return new Options().capacity(capacity);
-  }
-  
-  /**
-   * @param memoryLimit 
-   */
-  public static Options memoryLimit(Long memoryLimit) {
-    return new Options().memoryLimit(memoryLimit);
-  }
-  
-  /**
-   * @param container 
-   */
-  public static Options container(String container) {
-    return new Options().container(container);
-  }
-  
-  /**
-   * @param sharedName 
-   */
-  public static Options sharedName(String sharedName) {
-    return new Options().sharedName(sharedName);
-  }
-  
-  /**
-   */
-  public List<Output<?>> values() {
-    return values;
-  }
-  
-  @Override
-  @SuppressWarnings({"rawtypes", "unchecked"})
-  public Iterator<Operand<TType>> iterator() {
-    return (Iterator) values.iterator();
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "MapUnstage";
-  
-  private List<Output<?>> values;
-  
-  private MapUnstage(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    int valuesLength = operation.outputListLength("values");
-    values = Arrays.asList(operation.outputList(outputIdx, valuesLength));
-    outputIdx += valuesLength;
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MapUnstageNoKey.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MapUnstageNoKey.java
deleted file mode 100644
index 6d4a767f0f9..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MapUnstageNoKey.java
+++ /dev/null
@@ -1,180 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import java.util.Arrays;
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.TInt64;
-
-/**
- * Op removes and returns a random (key, value)
- * <p>
- * from the underlying container.   If the underlying container
- * does not contain elements, the op will block until it does.
- */
-@Operator
-public final class MapUnstageNoKey extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.core.MapUnstageNoKey}
-   */
-  public static class Options {
-    
-    /**
-     * @param capacity 
-     */
-    public Options capacity(Long capacity) {
-      this.capacity = capacity;
-      return this;
-    }
-    
-    /**
-     * @param memoryLimit 
-     */
-    public Options memoryLimit(Long memoryLimit) {
-      this.memoryLimit = memoryLimit;
-      return this;
-    }
-    
-    /**
-     * @param container 
-     */
-    public Options container(String container) {
-      this.container = container;
-      return this;
-    }
-    
-    /**
-     * @param sharedName 
-     */
-    public Options sharedName(String sharedName) {
-      this.sharedName = sharedName;
-      return this;
-    }
-    
-    private Long capacity;
-    private Long memoryLimit;
-    private String container;
-    private String sharedName;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new MapUnstageNoKey operation.
-   * 
-   * @param scope current scope
-   * @param indices 
-   * @param dtypes 
-   * @param options carries optional attributes values
-   * @return a new instance of MapUnstageNoKey
-   */
-  @Endpoint(describeByClass = true)
-  public static MapUnstageNoKey create(Scope scope, Operand<TInt32> indices, List<Class<?>> dtypes, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("MapUnstageNoKey", scope.makeOpName("MapUnstageNoKey"));
-    opBuilder.addInput(indices.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    Class[] dtypesArray = new Class[dtypes.size()];
-    for (int i = 0; i < dtypesArray.length; ++i) {
-      dtypesArray[i] = dtypes.get(i);
-    }
-    opBuilder.setAttr("dtypes", dtypesArray);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.capacity != null) {
-          opBuilder.setAttr("capacity", opts.capacity);
-        }
-        if (opts.memoryLimit != null) {
-          opBuilder.setAttr("memory_limit", opts.memoryLimit);
-        }
-        if (opts.container != null) {
-          opBuilder.setAttr("container", opts.container);
-        }
-        if (opts.sharedName != null) {
-          opBuilder.setAttr("shared_name", opts.sharedName);
-        }
-      }
-    }
-    return new MapUnstageNoKey(opBuilder.build());
-  }
-  
-  /**
-   * @param capacity 
-   */
-  public static Options capacity(Long capacity) {
-    return new Options().capacity(capacity);
-  }
-  
-  /**
-   * @param memoryLimit 
-   */
-  public static Options memoryLimit(Long memoryLimit) {
-    return new Options().memoryLimit(memoryLimit);
-  }
-  
-  /**
-   * @param container 
-   */
-  public static Options container(String container) {
-    return new Options().container(container);
-  }
-  
-  /**
-   * @param sharedName 
-   */
-  public static Options sharedName(String sharedName) {
-    return new Options().sharedName(sharedName);
-  }
-  
-  /**
-   */
-  public Output<TInt64> key() {
-    return key;
-  }
-  
-  /**
-   */
-  public List<Output<?>> values() {
-    return values;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "MapUnstageNoKey";
-  
-  private Output<TInt64> key;
-  private List<Output<?>> values;
-  
-  private MapUnstageNoKey(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    key = operation.output(outputIdx++);
-    int valuesLength = operation.outputListLength("values");
-    values = Arrays.asList(operation.outputList(outputIdx, valuesLength));
-    outputIdx += valuesLength;
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Max.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Max.java
deleted file mode 100644
index 84a862f4435..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Max.java
+++ /dev/null
@@ -1,118 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * Computes the maximum of elements across dimensions of a tensor.
- * <p>
- * Reduces `input` along the dimensions given in `axis`. Unless
- * `keep_dims` is true, the rank of the tensor is reduced by 1 for each entry in
- * `axis`. If `keep_dims` is true, the reduced dimensions are
- * retained with length 1.
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator
-public final class Max<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.core.Max}
-   */
-  public static class Options {
-    
-    /**
-     * @param keepDims If true, retain reduced dimensions with length 1.
-     */
-    public Options keepDims(Boolean keepDims) {
-      this.keepDims = keepDims;
-      return this;
-    }
-    
-    private Boolean keepDims;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new Max operation.
-   * 
-   * @param scope current scope
-   * @param input The tensor to reduce.
-   * @param axis The dimensions to reduce. Must be in the range
-   * `[-rank(input), rank(input))`.
-   * @param options carries optional attributes values
-   * @return a new instance of Max
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> Max<T> create(Scope scope, Operand<T> input, Operand<U> axis, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Max", scope.makeOpName("Max"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder.addInput(axis.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.keepDims != null) {
-          opBuilder.setAttr("keep_dims", opts.keepDims);
-        }
-      }
-    }
-    return new Max<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param keepDims If true, retain reduced dimensions with length 1.
-   */
-  public static Options keepDims(Boolean keepDims) {
-    return new Options().keepDims(keepDims);
-  }
-  
-  /**
-   * The reduced tensor.
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Max";
-  
-  private Output<T> output;
-  
-  private Max(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Merge.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Merge.java
deleted file mode 100644
index 7b6247f50f9..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Merge.java
+++ /dev/null
@@ -1,87 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.Operands;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.family.TType;
-
-/**
- * Forwards the value of an available tensor from `inputs` to `output`.
- * <p>
- * `Merge` waits for at least one of the tensors in `inputs` to become available.
- * It is usually combined with `Switch` to implement branching.
- * <p>
- * `Merge` forwards the first tensor to become available to `output`, and sets
- * `value_index` to its index in `inputs`.
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator
-public final class Merge<T extends TType> extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new Merge operation.
-   * 
-   * @param scope current scope
-   * @param inputs The input tensors, exactly one of which will become available.
-   * @return a new instance of Merge
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> Merge<T> create(Scope scope, Iterable<Operand<T>> inputs) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Merge", scope.makeOpName("Merge"));
-    opBuilder.addInputList(Operands.asOutputs(scope, inputs));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new Merge<T>(opBuilder.build());
-  }
-  
-  /**
-   * Will be set to the available input tensor.
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  /**
-   * The index of the chosen input tensor in `inputs`.
-   */
-  public Output<TInt32> valueIndex() {
-    return valueIndex;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Merge";
-  
-  private Output<T> output;
-  private Output<TInt32> valueIndex;
-  
-  private Merge(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-    valueIndex = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Min.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Min.java
deleted file mode 100644
index a6772fd009c..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Min.java
+++ /dev/null
@@ -1,118 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * Computes the minimum of elements across dimensions of a tensor.
- * <p>
- * Reduces `input` along the dimensions given in `axis`. Unless
- * `keep_dims` is true, the rank of the tensor is reduced by 1 for each entry in
- * `axis`. If `keep_dims` is true, the reduced dimensions are
- * retained with length 1.
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator
-public final class Min<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.core.Min}
-   */
-  public static class Options {
-    
-    /**
-     * @param keepDims If true, retain reduced dimensions with length 1.
-     */
-    public Options keepDims(Boolean keepDims) {
-      this.keepDims = keepDims;
-      return this;
-    }
-    
-    private Boolean keepDims;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new Min operation.
-   * 
-   * @param scope current scope
-   * @param input The tensor to reduce.
-   * @param axis The dimensions to reduce. Must be in the range
-   * `[-rank(input), rank(input))`.
-   * @param options carries optional attributes values
-   * @return a new instance of Min
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> Min<T> create(Scope scope, Operand<T> input, Operand<U> axis, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Min", scope.makeOpName("Min"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder.addInput(axis.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.keepDims != null) {
-          opBuilder.setAttr("keep_dims", opts.keepDims);
-        }
-      }
-    }
-    return new Min<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param keepDims If true, retain reduced dimensions with length 1.
-   */
-  public static Options keepDims(Boolean keepDims) {
-    return new Options().keepDims(keepDims);
-  }
-  
-  /**
-   * The reduced tensor.
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Min";
-  
-  private Output<T> output;
-  
-  private Min(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MirrorPad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MirrorPad.java
deleted file mode 100644
index 9a10a23b155..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MirrorPad.java
+++ /dev/null
@@ -1,111 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * Pads a tensor with mirrored values.
- * <p>
- * This operation pads a `input` with mirrored values according to the `paddings`
- * you specify. `paddings` is an integer tensor with shape `[n, 2]`, where n is
- * the rank of `input`. For each dimension D of `input`, `paddings[D, 0]` indicates
- * how many values to add before the contents of `input` in that dimension, and
- * `paddings[D, 1]` indicates how many values to add after the contents of `input`
- * in that dimension. Both `paddings[D, 0]` and `paddings[D, 1]` must be no greater
- * than `input.dim_size(D)` (or `input.dim_size(D) - 1`) if `copy_border` is true
- * (if false, respectively).
- * <p>
- * The padded size of each dimension D of the output is:
- * <p>
- * `paddings(D, 0) + input.dim_size(D) + paddings(D, 1)`
- * <p>
- * For example:
- * <pre>{@code
- * # 't' is [[1, 2, 3], [4, 5, 6]].
- * # 'paddings' is [[1, 1]], [2, 2]].
- * # 'mode' is SYMMETRIC.
- * # rank of 't' is 2.
- * pad(t, paddings) ==> [[2, 1, 1, 2, 3, 3, 2]
- *                       [2, 1, 1, 2, 3, 3, 2]
- *                       [5, 4, 4, 5, 6, 6, 5]
- *                       [5, 4, 4, 5, 6, 6, 5]]
- * }</pre>
- * 
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator
-public final class MirrorPad<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new MirrorPad operation.
-   * 
-   * @param scope current scope
-   * @param input The input tensor to be padded.
-   * @param paddings A two-column matrix specifying the padding sizes. The number of
-   * rows must be the same as the rank of `input`.
-   * @param mode Either `REFLECT` or `SYMMETRIC`. In reflect mode the padded regions
-   * do not include the borders, while in symmetric mode the padded regions
-   * do include the borders. For example, if `input` is `[1, 2, 3]` and `paddings`
-   * is `[0, 2]`, then the output is `[1, 2, 3, 2, 1]` in reflect mode, and
-   * it is `[1, 2, 3, 3, 2]` in symmetric mode.
-   * @return a new instance of MirrorPad
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> MirrorPad<T> create(Scope scope, Operand<T> input, Operand<U> paddings, String mode) {
-    OperationBuilder opBuilder = scope.env().opBuilder("MirrorPad", scope.makeOpName("MirrorPad"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder.addInput(paddings.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("mode", mode);
-    return new MirrorPad<T>(opBuilder.build());
-  }
-  
-  /**
-   * The padded tensor.
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "MirrorPad";
-  
-  private Output<T> output;
-  
-  private MirrorPad(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MirrorPadGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MirrorPadGrad.java
deleted file mode 100644
index 71a93147bd0..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MirrorPadGrad.java
+++ /dev/null
@@ -1,99 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * Gradient op for `MirrorPad` op. This op folds a mirror-padded tensor.
- * <p>
- * This operation folds the padded areas of `input` by `MirrorPad` according to the
- * `paddings` you specify. `paddings` must be the same as `paddings` argument
- * given to the corresponding `MirrorPad` op.
- * <p>
- * The folded size of each dimension D of the output is:
- * <p>
- * `input.dim_size(D) - paddings(D, 0) - paddings(D, 1)`
- * <p>
- * For example:
- * <pre>{@code
- * # 't' is [[1, 2, 3], [4, 5, 6], [7, 8, 9]].
- * # 'paddings' is [[0, 1]], [0, 1]].
- * # 'mode' is SYMMETRIC.
- * # rank of 't' is 2.
- * pad(t, paddings) ==> [[ 1,  5]
- *                       [11, 28]]
- * }</pre>
- * 
- * 
- * @param <T> data type for {@code output()} output
- */
-public final class MirrorPadGrad<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new MirrorPadGrad operation.
-   * 
-   * @param scope current scope
-   * @param input The input tensor to be folded.
-   * @param paddings A two-column matrix specifying the padding sizes. The number of
-   * rows must be the same as the rank of `input`.
-   * @param mode The mode used in the `MirrorPad` op.
-   * @return a new instance of MirrorPadGrad
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> MirrorPadGrad<T> create(Scope scope, Operand<T> input, Operand<U> paddings, String mode) {
-    OperationBuilder opBuilder = scope.env().opBuilder("MirrorPadGrad", scope.makeOpName("MirrorPadGrad"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder.addInput(paddings.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("mode", mode);
-    return new MirrorPadGrad<T>(opBuilder.build());
-  }
-  
-  /**
-   * The folded tensor.
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "MirrorPadGrad";
-  
-  private Output<T> output;
-  
-  private MirrorPadGrad(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MlirPassthroughOp.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MlirPassthroughOp.java
deleted file mode 100644
index 270d184c011..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MlirPassthroughOp.java
+++ /dev/null
@@ -1,117 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import java.util.Arrays;
-import java.util.Iterator;
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.Operands;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Wraps an arbitrary MLIR computation expressed as a module with a main() function.
- * <p>
- * This operation does not have an associated kernel and is not intended to be
- * executed in a regular TensorFlow session. Instead it is intended to be used for
- * testing or for special case where a user intends to pass custom MLIR computation
- * through a TensorFlow graph with the intent of having custom tooling processing
- * it downstream (when targeting a different environment, like TensorFlow lite for
- * example).
- * The MLIR module is expected to have a main() function that will be used as an
- * entry point. The inputs to the operations will be passed as argument to the
- * main() function and the returned values of the main function mapped to the
- * outputs.
- * Example usage:
- * <pre>{@code
- * import tensorflow as tf
- * from tensorflow.compiler.mlir.tensorflow.gen_mlir_passthrough_op import mlir_passthrough_op
- * 
- * mlir_module = '''python
- * func @main(%arg0 : tensor<10xf32>, %arg1 : tensor<10xf32>) -> tensor<10x10xf32> {
- *    %add = "magic.op"(%arg0, %arg1) : (tensor<10xf32>, tensor<10xf32>) -> tensor<10x10xf32>
- *    return %ret : tensor<10x10xf32>
- * }
- * '''
- * 
- * @tf.function
- * def foo(x, y):
- *   return mlir_passthrough_op([x, y], mlir_module, Toutputs=[tf.float32])
- * 
- * graph_def = foo.get_concrete_function(tf.TensorSpec([10], tf.float32), tf.TensorSpec([10], tf.float32)).graph.as_graph_def()
- * }</pre>
- * 
- */
-@Operator
-public final class MlirPassthroughOp extends RawOp implements Iterable<Operand<TType>> {
-  
-  /**
-   * Factory method to create a class wrapping a new MlirPassthroughOp operation.
-   * 
-   * @param scope current scope
-   * @param inputs 
-   * @param mlirModule 
-   * @param Toutputs 
-   * @return a new instance of MlirPassthroughOp
-   */
-  @Endpoint(describeByClass = true)
-  public static MlirPassthroughOp create(Scope scope, Iterable<Operand<?>> inputs, String mlirModule, List<Class<?>> Toutputs) {
-    OperationBuilder opBuilder = scope.env().opBuilder("MlirPassthroughOp", scope.makeOpName("MlirPassthroughOp"));
-    opBuilder.addInputList(Operands.asOutputs(scope, inputs));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("mlir_module", mlirModule);
-    Class[] ToutputsArray = new Class[Toutputs.size()];
-    for (int i = 0; i < ToutputsArray.length; ++i) {
-      ToutputsArray[i] = Toutputs.get(i);
-    }
-    opBuilder.setAttr("Toutputs", ToutputsArray);
-    return new MlirPassthroughOp(opBuilder.build());
-  }
-  
-  /**
-   */
-  public List<Output<?>> outputs() {
-    return outputs;
-  }
-  
-  @Override
-  @SuppressWarnings({"rawtypes", "unchecked"})
-  public Iterator<Operand<TType>> iterator() {
-    return (Iterator) outputs.iterator();
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "MlirPassthroughOp";
-  
-  private List<Output<?>> outputs;
-  
-  private MlirPassthroughOp(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    int outputsLength = operation.outputListLength("outputs");
-    outputs = Arrays.asList(operation.outputList(outputIdx, outputsLength));
-    outputIdx += outputsLength;
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MutableDenseHashTable.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MutableDenseHashTable.java
deleted file mode 100644
index 3a8988e8172..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MutableDenseHashTable.java
+++ /dev/null
@@ -1,224 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Creates an empty hash table that uses tensors as the backing store.
- * <p>
- * It uses "open addressing" with quadratic reprobing to resolve
- * collisions.
- * <p>
- * This op creates a mutable hash table, specifying the type of its keys and
- * values. Each value must be a scalar. Data can be inserted into the table using
- * the insert operations. It does not support the initialization operation.
- */
-@Operator
-public final class MutableDenseHashTable extends RawOp implements Operand<TType> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.core.MutableDenseHashTable}
-   */
-  public static class Options {
-    
-    /**
-     * @param container If non-empty, this table is placed in the given container.
-     * Otherwise, a default container is used.
-     */
-    public Options container(String container) {
-      this.container = container;
-      return this;
-    }
-    
-    /**
-     * @param sharedName If non-empty, this table is shared under the given name across
-     * multiple sessions.
-     */
-    public Options sharedName(String sharedName) {
-      this.sharedName = sharedName;
-      return this;
-    }
-    
-    /**
-     * @param useNodeNameSharing 
-     */
-    public Options useNodeNameSharing(Boolean useNodeNameSharing) {
-      this.useNodeNameSharing = useNodeNameSharing;
-      return this;
-    }
-    
-    /**
-     * @param valueShape The shape of each value.
-     */
-    public Options valueShape(Shape valueShape) {
-      this.valueShape = valueShape;
-      return this;
-    }
-    
-    /**
-     * @param initialNumBuckets The initial number of hash table buckets. Must be a power
-     * to 2.
-     */
-    public Options initialNumBuckets(Long initialNumBuckets) {
-      this.initialNumBuckets = initialNumBuckets;
-      return this;
-    }
-    
-    /**
-     * @param maxLoadFactor The maximum ratio between number of entries and number of
-     * buckets before growing the table. Must be between 0 and 1.
-     */
-    public Options maxLoadFactor(Float maxLoadFactor) {
-      this.maxLoadFactor = maxLoadFactor;
-      return this;
-    }
-    
-    private String container;
-    private String sharedName;
-    private Boolean useNodeNameSharing;
-    private Shape valueShape;
-    private Long initialNumBuckets;
-    private Float maxLoadFactor;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new MutableDenseHashTable operation.
-   * 
-   * @param scope current scope
-   * @param emptyKey The key used to represent empty key buckets internally. Must not
-   * be used in insert or lookup operations.
-   * @param deletedKey 
-   * @param valueDtype Type of the table values.
-   * @param options carries optional attributes values
-   * @return a new instance of MutableDenseHashTable
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TType> MutableDenseHashTable create(Scope scope, Operand<T> emptyKey, Operand<T> deletedKey, Class<U> valueDtype, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("MutableDenseHashTableV2", scope.makeOpName("MutableDenseHashTable"));
-    opBuilder.addInput(emptyKey.asOutput(scope));
-    opBuilder.addInput(deletedKey.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("value_dtype", valueDtype);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.container != null) {
-          opBuilder.setAttr("container", opts.container);
-        }
-        if (opts.sharedName != null) {
-          opBuilder.setAttr("shared_name", opts.sharedName);
-        }
-        if (opts.useNodeNameSharing != null) {
-          opBuilder.setAttr("use_node_name_sharing", opts.useNodeNameSharing);
-        }
-        if (opts.valueShape != null) {
-          opBuilder.setAttr("value_shape", opts.valueShape);
-        }
-        if (opts.initialNumBuckets != null) {
-          opBuilder.setAttr("initial_num_buckets", opts.initialNumBuckets);
-        }
-        if (opts.maxLoadFactor != null) {
-          opBuilder.setAttr("max_load_factor", opts.maxLoadFactor);
-        }
-      }
-    }
-    return new MutableDenseHashTable(opBuilder.build());
-  }
-  
-  /**
-   * @param container If non-empty, this table is placed in the given container.
-   * Otherwise, a default container is used.
-   */
-  public static Options container(String container) {
-    return new Options().container(container);
-  }
-  
-  /**
-   * @param sharedName If non-empty, this table is shared under the given name across
-   * multiple sessions.
-   */
-  public static Options sharedName(String sharedName) {
-    return new Options().sharedName(sharedName);
-  }
-  
-  /**
-   * @param useNodeNameSharing 
-   */
-  public static Options useNodeNameSharing(Boolean useNodeNameSharing) {
-    return new Options().useNodeNameSharing(useNodeNameSharing);
-  }
-  
-  /**
-   * @param valueShape The shape of each value.
-   */
-  public static Options valueShape(Shape valueShape) {
-    return new Options().valueShape(valueShape);
-  }
-  
-  /**
-   * @param initialNumBuckets The initial number of hash table buckets. Must be a power
-   * to 2.
-   */
-  public static Options initialNumBuckets(Long initialNumBuckets) {
-    return new Options().initialNumBuckets(initialNumBuckets);
-  }
-  
-  /**
-   * @param maxLoadFactor The maximum ratio between number of entries and number of
-   * buckets before growing the table. Must be between 0 and 1.
-   */
-  public static Options maxLoadFactor(Float maxLoadFactor) {
-    return new Options().maxLoadFactor(maxLoadFactor);
-  }
-  
-  /**
-   * Handle to a table.
-   */
-  public Output<?> tableHandle() {
-    return tableHandle;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) tableHandle;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "MutableDenseHashTableV2";
-  
-  private Output<?> tableHandle;
-  
-  private MutableDenseHashTable(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    tableHandle = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MutableHashTable.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MutableHashTable.java
deleted file mode 100644
index fe0449157d8..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MutableHashTable.java
+++ /dev/null
@@ -1,158 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Creates an empty hash table.
- * <p>
- * This op creates a mutable hash table, specifying the type of its keys and
- * values. Each value must be a scalar. Data can be inserted into the table using
- * the insert operations. It does not support the initialization operation.
- */
-@Operator
-public final class MutableHashTable extends RawOp implements Operand<TType> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.core.MutableHashTable}
-   */
-  public static class Options {
-    
-    /**
-     * @param container If non-empty, this table is placed in the given container.
-     * Otherwise, a default container is used.
-     */
-    public Options container(String container) {
-      this.container = container;
-      return this;
-    }
-    
-    /**
-     * @param sharedName If non-empty, this table is shared under the given name across
-     * multiple sessions.
-     */
-    public Options sharedName(String sharedName) {
-      this.sharedName = sharedName;
-      return this;
-    }
-    
-    /**
-     * @param useNodeNameSharing If true and shared_name is empty, the table is shared
-     * using the node name.
-     */
-    public Options useNodeNameSharing(Boolean useNodeNameSharing) {
-      this.useNodeNameSharing = useNodeNameSharing;
-      return this;
-    }
-    
-    private String container;
-    private String sharedName;
-    private Boolean useNodeNameSharing;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new MutableHashTable operation.
-   * 
-   * @param scope current scope
-   * @param keyDtype Type of the table keys.
-   * @param valueDtype Type of the table values.
-   * @param options carries optional attributes values
-   * @return a new instance of MutableHashTable
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TType> MutableHashTable create(Scope scope, Class<T> keyDtype, Class<U> valueDtype, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("MutableHashTableV2", scope.makeOpName("MutableHashTable"));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("key_dtype", keyDtype);
-    opBuilder.setAttr("value_dtype", valueDtype);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.container != null) {
-          opBuilder.setAttr("container", opts.container);
-        }
-        if (opts.sharedName != null) {
-          opBuilder.setAttr("shared_name", opts.sharedName);
-        }
-        if (opts.useNodeNameSharing != null) {
-          opBuilder.setAttr("use_node_name_sharing", opts.useNodeNameSharing);
-        }
-      }
-    }
-    return new MutableHashTable(opBuilder.build());
-  }
-  
-  /**
-   * @param container If non-empty, this table is placed in the given container.
-   * Otherwise, a default container is used.
-   */
-  public static Options container(String container) {
-    return new Options().container(container);
-  }
-  
-  /**
-   * @param sharedName If non-empty, this table is shared under the given name across
-   * multiple sessions.
-   */
-  public static Options sharedName(String sharedName) {
-    return new Options().sharedName(sharedName);
-  }
-  
-  /**
-   * @param useNodeNameSharing If true and shared_name is empty, the table is shared
-   * using the node name.
-   */
-  public static Options useNodeNameSharing(Boolean useNodeNameSharing) {
-    return new Options().useNodeNameSharing(useNodeNameSharing);
-  }
-  
-  /**
-   * Handle to a table.
-   */
-  public Output<?> tableHandle() {
-    return tableHandle;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) tableHandle;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "MutableHashTableV2";
-  
-  private Output<?> tableHandle;
-  
-  private MutableHashTable(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    tableHandle = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MutableHashTableOfTensors.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MutableHashTableOfTensors.java
deleted file mode 100644
index c61f37dacb0..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MutableHashTableOfTensors.java
+++ /dev/null
@@ -1,176 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Creates an empty hash table.
- * <p>
- * This op creates a mutable hash table, specifying the type of its keys and
- * values. Each value must be a vector. Data can be inserted into the table using
- * the insert operations. It does not support the initialization operation.
- */
-@Operator
-public final class MutableHashTableOfTensors extends RawOp implements Operand<TType> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.core.MutableHashTableOfTensors}
-   */
-  public static class Options {
-    
-    /**
-     * @param container If non-empty, this table is placed in the given container.
-     * Otherwise, a default container is used.
-     */
-    public Options container(String container) {
-      this.container = container;
-      return this;
-    }
-    
-    /**
-     * @param sharedName If non-empty, this table is shared under the given name across
-     * multiple sessions.
-     */
-    public Options sharedName(String sharedName) {
-      this.sharedName = sharedName;
-      return this;
-    }
-    
-    /**
-     * @param useNodeNameSharing 
-     */
-    public Options useNodeNameSharing(Boolean useNodeNameSharing) {
-      this.useNodeNameSharing = useNodeNameSharing;
-      return this;
-    }
-    
-    /**
-     * @param valueShape 
-     */
-    public Options valueShape(Shape valueShape) {
-      this.valueShape = valueShape;
-      return this;
-    }
-    
-    private String container;
-    private String sharedName;
-    private Boolean useNodeNameSharing;
-    private Shape valueShape;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new MutableHashTableOfTensors operation.
-   * 
-   * @param scope current scope
-   * @param keyDtype Type of the table keys.
-   * @param valueDtype Type of the table values.
-   * @param options carries optional attributes values
-   * @return a new instance of MutableHashTableOfTensors
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TType> MutableHashTableOfTensors create(Scope scope, Class<T> keyDtype, Class<U> valueDtype, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("MutableHashTableOfTensorsV2", scope.makeOpName("MutableHashTableOfTensors"));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("key_dtype", keyDtype);
-    opBuilder.setAttr("value_dtype", valueDtype);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.container != null) {
-          opBuilder.setAttr("container", opts.container);
-        }
-        if (opts.sharedName != null) {
-          opBuilder.setAttr("shared_name", opts.sharedName);
-        }
-        if (opts.useNodeNameSharing != null) {
-          opBuilder.setAttr("use_node_name_sharing", opts.useNodeNameSharing);
-        }
-        if (opts.valueShape != null) {
-          opBuilder.setAttr("value_shape", opts.valueShape);
-        }
-      }
-    }
-    return new MutableHashTableOfTensors(opBuilder.build());
-  }
-  
-  /**
-   * @param container If non-empty, this table is placed in the given container.
-   * Otherwise, a default container is used.
-   */
-  public static Options container(String container) {
-    return new Options().container(container);
-  }
-  
-  /**
-   * @param sharedName If non-empty, this table is shared under the given name across
-   * multiple sessions.
-   */
-  public static Options sharedName(String sharedName) {
-    return new Options().sharedName(sharedName);
-  }
-  
-  /**
-   * @param useNodeNameSharing 
-   */
-  public static Options useNodeNameSharing(Boolean useNodeNameSharing) {
-    return new Options().useNodeNameSharing(useNodeNameSharing);
-  }
-  
-  /**
-   * @param valueShape 
-   */
-  public static Options valueShape(Shape valueShape) {
-    return new Options().valueShape(valueShape);
-  }
-  
-  /**
-   * Handle to a table.
-   */
-  public Output<?> tableHandle() {
-    return tableHandle;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) tableHandle;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "MutableHashTableOfTensorsV2";
-  
-  private Output<?> tableHandle;
-  
-  private MutableHashTableOfTensors(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    tableHandle = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Mutex.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Mutex.java
deleted file mode 100644
index 62c1434e39b..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Mutex.java
+++ /dev/null
@@ -1,129 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Creates a Mutex resource that can be locked by `MutexLock`.
- */
-@Operator
-public final class Mutex extends RawOp implements Operand<TType> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.core.Mutex}
-   */
-  public static class Options {
-    
-    /**
-     * @param container If non-empty, this variable is placed in the given container.
-     * Otherwise, a default container is used.
-     */
-    public Options container(String container) {
-      this.container = container;
-      return this;
-    }
-    
-    /**
-     * @param sharedName If non-empty, this variable is named in the given bucket
-     * with this shared_name. Otherwise, the node name is used instead.
-     */
-    public Options sharedName(String sharedName) {
-      this.sharedName = sharedName;
-      return this;
-    }
-    
-    private String container;
-    private String sharedName;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new Mutex operation.
-   * 
-   * @param scope current scope
-   * @param options carries optional attributes values
-   * @return a new instance of Mutex
-   */
-  @Endpoint(describeByClass = true)
-  public static Mutex create(Scope scope, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("MutexV2", scope.makeOpName("Mutex"));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.container != null) {
-          opBuilder.setAttr("container", opts.container);
-        }
-        if (opts.sharedName != null) {
-          opBuilder.setAttr("shared_name", opts.sharedName);
-        }
-      }
-    }
-    return new Mutex(opBuilder.build());
-  }
-  
-  /**
-   * @param container If non-empty, this variable is placed in the given container.
-   * Otherwise, a default container is used.
-   */
-  public static Options container(String container) {
-    return new Options().container(container);
-  }
-  
-  /**
-   * @param sharedName If non-empty, this variable is named in the given bucket
-   * with this shared_name. Otherwise, the node name is used instead.
-   */
-  public static Options sharedName(String sharedName) {
-    return new Options().sharedName(sharedName);
-  }
-  
-  /**
-   * The mutex resource.
-   */
-  public Output<?> resource() {
-    return resource;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) resource;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "MutexV2";
-  
-  private Output<?> resource;
-  
-  private Mutex(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    resource = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MutexLock.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MutexLock.java
deleted file mode 100644
index ac2927b03c9..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MutexLock.java
+++ /dev/null
@@ -1,112 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Locks a mutex resource.  The output is the lock.  So long as the lock tensor
- * <p>
- * is alive, any other request to use `MutexLock` with this mutex will wait.
- * <p>
- * This is particularly useful for creating a critical section when used in
- * conjunction with `MutexLockIdentity`:
- * <pre>{@code
- * mutex = mutex_v2(
- *   shared_name=handle_name, container=container, name=name)
- * 
- * def execute_in_critical_section(fn, *args, **kwargs):
- *   lock = gen_resource_variable_ops.mutex_lock(mutex)
- * 
- *   with ops.control_dependencies([lock]):
- *     r = fn(*args, **kwargs)
- * 
- *   with ops.control_dependencies(nest.flatten(r)):
- *     with ops.colocate_with(mutex):
- *       ensure_lock_exists = mutex_lock_identity(lock)
- * 
- *     # Make sure that if any element of r is accessed, all of
- *     # them are executed together.
- *     r = nest.map_structure(tf.identity, r)
- * 
- *   with ops.control_dependencies([ensure_lock_exists]):
- *     return nest.map_structure(tf.identity, r)
- * }</pre>
- * While `fn` is running in the critical section, no other functions which wish to
- * use this critical section may run.
- * <p>
- * Often the use case is that two executions of the same graph, in parallel,
- * wish to run `fn`; and we wish to ensure that only one of them executes
- * at a time.  This is especially important if `fn` modifies one or more
- * variables at a time.
- * <p>
- * It is also useful if two separate functions must share a resource, but we
- * wish to ensure the usage is exclusive.
- */
-@Operator
-public final class MutexLock extends RawOp implements Operand<TType> {
-  
-  /**
-   * Factory method to create a class wrapping a new MutexLock operation.
-   * 
-   * @param scope current scope
-   * @param mutex The mutex resource to lock.
-   * @return a new instance of MutexLock
-   */
-  @Endpoint(describeByClass = true)
-  public static MutexLock create(Scope scope, Operand<?> mutex) {
-    OperationBuilder opBuilder = scope.env().opBuilder("MutexLock", scope.makeOpName("MutexLock"));
-    opBuilder.addInput(mutex.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new MutexLock(opBuilder.build());
-  }
-  
-  /**
-   * A tensor that keeps a shared pointer to a lock on the mutex;
-   * when the Tensor is destroyed, the use count on the shared pointer is decreased
-   * by 1.  When it reaches 0, the lock is released.
-   */
-  public Output<?> mutexLock() {
-    return mutexLock;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) mutexLock;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "MutexLock";
-  
-  private Output<?> mutexLock;
-  
-  private MutexLock(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    mutexLock = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/NcclAllReduce.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/NcclAllReduce.java
deleted file mode 100644
index a13de0e82ab..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/NcclAllReduce.java
+++ /dev/null
@@ -1,92 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Outputs a tensor containing the reduction across all input tensors.
- * <p>
- * Outputs a tensor containing the reduction across all input tensors passed to ops
- * within the same `shared_name.
- * <p>
- * The graph should be constructed so if one op runs with shared_name value `c`,
- * then `num_devices` ops will run with shared_name value `c`.  Failure to do so
- * will cause the graph execution to fail to complete.
- * <p>
- * input: the input to the reduction
- * data: the value of the reduction across all `num_devices` devices.
- * reduction: the reduction operation to perform.
- * num_devices: The number of devices participating in this reduction.
- * shared_name: Identifier that shared between ops of the same reduction.
- * 
- * @param <T> data type for {@code output()} output
- */
-public final class NcclAllReduce<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new NcclAllReduce operation.
-   * 
-   * @param scope current scope
-   * @param input 
-   * @param reduction 
-   * @param numDevices 
-   * @param sharedName 
-   * @return a new instance of NcclAllReduce
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> NcclAllReduce<T> create(Scope scope, Operand<T> input, String reduction, Long numDevices, String sharedName) {
-    OperationBuilder opBuilder = scope.env().opBuilder("NcclAllReduce", scope.makeOpName("NcclAllReduce"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("reduction", reduction);
-    opBuilder.setAttr("num_devices", numDevices);
-    opBuilder.setAttr("shared_name", sharedName);
-    return new NcclAllReduce<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "NcclAllReduce";
-  
-  private Output<T> output;
-  
-  private NcclAllReduce(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/NcclBroadcast.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/NcclBroadcast.java
deleted file mode 100644
index c4b727efa7e..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/NcclBroadcast.java
+++ /dev/null
@@ -1,86 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Sends `input` to all devices that are connected to the output.
- * <p>
- * Sends `input` to all devices that are connected to the output.
- * <p>
- * The graph should be constructed so that all ops connected to the output have a
- * valid device assignment, and the op itself is assigned one of these devices.
- * <p>
- * input: The input to the broadcast.
- * output: The same as input.
- * shape: The shape of the input tensor.
- * 
- * 
- * @param <T> data type for {@code output()} output
- */
-public final class NcclBroadcast<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new NcclBroadcast operation.
-   * 
-   * @param scope current scope
-   * @param input 
-   * @param shape 
-   * @return a new instance of NcclBroadcast
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> NcclBroadcast<T> create(Scope scope, Operand<T> input, Shape shape) {
-    OperationBuilder opBuilder = scope.env().opBuilder("NcclBroadcast", scope.makeOpName("NcclBroadcast"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("shape", shape);
-    return new NcclBroadcast<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "NcclBroadcast";
-  
-  private Output<T> output;
-  
-  private NcclBroadcast(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/NcclReduce.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/NcclReduce.java
deleted file mode 100644
index fecac09710d..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/NcclReduce.java
+++ /dev/null
@@ -1,85 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.Operands;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Reduces `input` from `num_devices` using `reduction` to a single device.
- * <p>
- * Reduces `input` from `num_devices` using `reduction` to a single device.
- * <p>
- * The graph should be constructed so that all inputs have a valid device
- * assignment, and the op itself is assigned one of these devices.
- * <p>
- * input: The input to the reduction.
- * data: the value of the reduction across all `num_devices` devices.
- * reduction: the reduction operation to perform.
- * 
- * @param <T> data type for {@code output()} output
- */
-public final class NcclReduce<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new NcclReduce operation.
-   * 
-   * @param scope current scope
-   * @param input 
-   * @param reduction 
-   * @return a new instance of NcclReduce
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> NcclReduce<T> create(Scope scope, Iterable<Operand<T>> input, String reduction) {
-    OperationBuilder opBuilder = scope.env().opBuilder("NcclReduce", scope.makeOpName("NcclReduce"));
-    opBuilder.addInputList(Operands.asOutputs(scope, input));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("reduction", reduction);
-    return new NcclReduce<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "NcclReduce";
-  
-  private Output<T> output;
-  
-  private NcclReduce(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/NextIteration.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/NextIteration.java
deleted file mode 100644
index d246bdfd764..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/NextIteration.java
+++ /dev/null
@@ -1,75 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Makes its input available to the next iteration.
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator
-public final class NextIteration<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new NextIteration operation.
-   * 
-   * @param scope current scope
-   * @param data The tensor to be made available to the next iteration.
-   * @return a new instance of NextIteration
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> NextIteration<T> create(Scope scope, Operand<T> data) {
-    OperationBuilder opBuilder = scope.env().opBuilder("NextIteration", scope.makeOpName("NextIteration"));
-    opBuilder.addInput(data.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new NextIteration<T>(opBuilder.build());
-  }
-  
-  /**
-   * The same tensor as `data`.
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "NextIteration";
-  
-  private Output<T> output;
-  
-  private NextIteration(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/NoOp.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/NoOp.java
deleted file mode 100644
index 922b5d55ce3..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/NoOp.java
+++ /dev/null
@@ -1,52 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-
-/**
- * Does nothing. Only useful as a placeholder for control edges.
- */
-@Operator
-public final class NoOp extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new NoOp operation.
-   * 
-   * @param scope current scope
-   * @return a new instance of NoOp
-   */
-  @Endpoint(describeByClass = true)
-  public static NoOp create(Scope scope) {
-    OperationBuilder opBuilder = scope.env().opBuilder("NoOp", scope.makeOpName("NoOp"));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new NoOp(opBuilder.build());
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "NoOp";
-  
-  private NoOp(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OneHot.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OneHot.java
deleted file mode 100644
index 7969d18bd4d..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OneHot.java
+++ /dev/null
@@ -1,198 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * Returns a one-hot tensor.
- * <p>
- * The locations represented by indices in `indices` take value `on_value`,
- * while all other locations take value `off_value`.
- * <p>
- * If the input `indices` is rank `N`, the output will have rank `N+1`,
- * The new axis is created at dimension `axis` (default: the new axis is
- * appended at the end).
- * <p>
- * If `indices` is a scalar the output shape will be a vector of length `depth`.
- * <p>
- * If `indices` is a vector of length `features`, the output shape will be:
- * <pre>{@code
- *   features x depth if axis == -1
- *   depth x features if axis == 0
- * }</pre>
- * If `indices` is a matrix (batch) with shape `[batch, features]`,
- * the output shape will be:
- * <pre>{@code
- *   batch x features x depth if axis == -1
- *   batch x depth x features if axis == 1
- *   depth x batch x features if axis == 0
- * }</pre>
- * Examples
- * =========
- * <p>
- * Suppose that
- * <pre>{@code
- *   indices = [0, 2, -1, 1]
- *   depth = 3
- *   on_value = 5.0
- *   off_value = 0.0
- *   axis = -1
- * }</pre>
- * Then output is `[4 x 3]`:
- * <pre>{@code
- * output =
- *   [5.0 0.0 0.0]  // one_hot(0)
- *   [0.0 0.0 5.0]  // one_hot(2)
- *   [0.0 0.0 0.0]  // one_hot(-1)
- *   [0.0 5.0 0.0]  // one_hot(1)
- * }</pre>
- * Suppose that
- * <pre>{@code
- *   indices = [0, 2, -1, 1]
- *   depth = 3
- *   on_value = 0.0
- *   off_value = 3.0
- *   axis = 0
- * }</pre>
- * Then output is `[3 x 4]`:
- * <pre>{@code
- * output =
- *   [0.0 3.0 3.0 3.0]
- *   [3.0 3.0 3.0 0.0]
- *   [3.0 3.0 3.0 3.0]
- *   [3.0 0.0 3.0 3.0]
- * //  ^                one_hot(0)
- * //      ^            one_hot(2)
- * //          ^        one_hot(-1)
- * //              ^    one_hot(1)
- * }</pre>
- * Suppose that
- * <pre>{@code
- *   indices = [[0, 2], [1, -1]]
- *   depth = 3
- *   on_value = 1.0
- *   off_value = 0.0
- *   axis = -1
- * }</pre>
- * Then output is `[2 x 2 x 3]`:
- * <pre>{@code
- * output =
- *   [
- *     [1.0, 0.0, 0.0]  // one_hot(0)
- *     [0.0, 0.0, 1.0]  // one_hot(2)
- *   ][
- *     [0.0, 1.0, 0.0]  // one_hot(1)
- *     [0.0, 0.0, 0.0]  // one_hot(-1)
- *   ]
- * }</pre>
- * 
- * 
- * @param <U> data type for {@code output()} output
- */
-@Operator
-public final class OneHot<U extends TType> extends RawOp implements Operand<U> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.core.OneHot}
-   */
-  public static class Options {
-    
-    /**
-     * @param axis The axis to fill (default: -1, a new inner-most axis).
-     */
-    public Options axis(Long axis) {
-      this.axis = axis;
-      return this;
-    }
-    
-    private Long axis;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new OneHot operation.
-   * 
-   * @param scope current scope
-   * @param indices A tensor of indices.
-   * @param depth A scalar defining the depth of the one hot dimension.
-   * @param onValue A scalar defining the value to fill in output when `indices[j] = i`.
-   * @param offValue A scalar defining the value to fill in output when `indices[j] != i`.
-   * @param options carries optional attributes values
-   * @return a new instance of OneHot
-   */
-  @Endpoint(describeByClass = true)
-  public static <U extends TType, T extends TNumber> OneHot<U> create(Scope scope, Operand<T> indices, Operand<TInt32> depth, Operand<U> onValue, Operand<U> offValue, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("OneHot", scope.makeOpName("OneHot"));
-    opBuilder.addInput(indices.asOutput(scope));
-    opBuilder.addInput(depth.asOutput(scope));
-    opBuilder.addInput(onValue.asOutput(scope));
-    opBuilder.addInput(offValue.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.axis != null) {
-          opBuilder.setAttr("axis", opts.axis);
-        }
-      }
-    }
-    return new OneHot<U>(opBuilder.build());
-  }
-  
-  /**
-   * @param axis The axis to fill (default: -1, a new inner-most axis).
-   */
-  public static Options axis(Long axis) {
-    return new Options().axis(axis);
-  }
-  
-  /**
-   * The one-hot tensor.
-   */
-  public Output<U> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<U> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "OneHot";
-  
-  private Output<U> output;
-  
-  private OneHot(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OnesLike.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OnesLike.java
deleted file mode 100644
index 53255e2fc85..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OnesLike.java
+++ /dev/null
@@ -1,75 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Returns a tensor of ones with the same shape and type as x.
- * 
- * @param <T> data type for {@code y()} output
- */
-@Operator
-public final class OnesLike<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new OnesLike operation.
-   * 
-   * @param scope current scope
-   * @param x a tensor of type T.
-   * @return a new instance of OnesLike
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> OnesLike<T> create(Scope scope, Operand<T> x) {
-    OperationBuilder opBuilder = scope.env().opBuilder("OnesLike", scope.makeOpName("OnesLike"));
-    opBuilder.addInput(x.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new OnesLike<T>(opBuilder.build());
-  }
-  
-  /**
-   * a tensor of the same shape and type as x but filled with ones.
-   */
-  public Output<T> y() {
-    return y;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return y;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "OnesLike";
-  
-  private Output<T> y;
-  
-  private OnesLike(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    y = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OrderedMapClear.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OrderedMapClear.java
deleted file mode 100644
index e44d88210ca..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OrderedMapClear.java
+++ /dev/null
@@ -1,150 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import java.util.List;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-
-/**
- * Op removes all elements in the underlying container.
- */
-@Operator
-public final class OrderedMapClear extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.core.OrderedMapClear}
-   */
-  public static class Options {
-    
-    /**
-     * @param capacity 
-     */
-    public Options capacity(Long capacity) {
-      this.capacity = capacity;
-      return this;
-    }
-    
-    /**
-     * @param memoryLimit 
-     */
-    public Options memoryLimit(Long memoryLimit) {
-      this.memoryLimit = memoryLimit;
-      return this;
-    }
-    
-    /**
-     * @param container 
-     */
-    public Options container(String container) {
-      this.container = container;
-      return this;
-    }
-    
-    /**
-     * @param sharedName 
-     */
-    public Options sharedName(String sharedName) {
-      this.sharedName = sharedName;
-      return this;
-    }
-    
-    private Long capacity;
-    private Long memoryLimit;
-    private String container;
-    private String sharedName;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new OrderedMapClear operation.
-   * 
-   * @param scope current scope
-   * @param dtypes 
-   * @param options carries optional attributes values
-   * @return a new instance of OrderedMapClear
-   */
-  @Endpoint(describeByClass = true)
-  public static OrderedMapClear create(Scope scope, List<Class<?>> dtypes, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("OrderedMapClear", scope.makeOpName("OrderedMapClear"));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    Class[] dtypesArray = new Class[dtypes.size()];
-    for (int i = 0; i < dtypesArray.length; ++i) {
-      dtypesArray[i] = dtypes.get(i);
-    }
-    opBuilder.setAttr("dtypes", dtypesArray);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.capacity != null) {
-          opBuilder.setAttr("capacity", opts.capacity);
-        }
-        if (opts.memoryLimit != null) {
-          opBuilder.setAttr("memory_limit", opts.memoryLimit);
-        }
-        if (opts.container != null) {
-          opBuilder.setAttr("container", opts.container);
-        }
-        if (opts.sharedName != null) {
-          opBuilder.setAttr("shared_name", opts.sharedName);
-        }
-      }
-    }
-    return new OrderedMapClear(opBuilder.build());
-  }
-  
-  /**
-   * @param capacity 
-   */
-  public static Options capacity(Long capacity) {
-    return new Options().capacity(capacity);
-  }
-  
-  /**
-   * @param memoryLimit 
-   */
-  public static Options memoryLimit(Long memoryLimit) {
-    return new Options().memoryLimit(memoryLimit);
-  }
-  
-  /**
-   * @param container 
-   */
-  public static Options container(String container) {
-    return new Options().container(container);
-  }
-  
-  /**
-   * @param sharedName 
-   */
-  public static Options sharedName(String sharedName) {
-    return new Options().sharedName(sharedName);
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "OrderedMapClear";
-  
-  private OrderedMapClear(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OrderedMapIncompleteSize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OrderedMapIncompleteSize.java
deleted file mode 100644
index 68fe5a7a610..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OrderedMapIncompleteSize.java
+++ /dev/null
@@ -1,168 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
-
-/**
- * Op returns the number of incomplete elements in the underlying container.
- */
-@Operator
-public final class OrderedMapIncompleteSize extends RawOp implements Operand<TInt32> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.core.OrderedMapIncompleteSize}
-   */
-  public static class Options {
-    
-    /**
-     * @param capacity 
-     */
-    public Options capacity(Long capacity) {
-      this.capacity = capacity;
-      return this;
-    }
-    
-    /**
-     * @param memoryLimit 
-     */
-    public Options memoryLimit(Long memoryLimit) {
-      this.memoryLimit = memoryLimit;
-      return this;
-    }
-    
-    /**
-     * @param container 
-     */
-    public Options container(String container) {
-      this.container = container;
-      return this;
-    }
-    
-    /**
-     * @param sharedName 
-     */
-    public Options sharedName(String sharedName) {
-      this.sharedName = sharedName;
-      return this;
-    }
-    
-    private Long capacity;
-    private Long memoryLimit;
-    private String container;
-    private String sharedName;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new OrderedMapIncompleteSize operation.
-   * 
-   * @param scope current scope
-   * @param dtypes 
-   * @param options carries optional attributes values
-   * @return a new instance of OrderedMapIncompleteSize
-   */
-  @Endpoint(describeByClass = true)
-  public static OrderedMapIncompleteSize create(Scope scope, List<Class<?>> dtypes, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("OrderedMapIncompleteSize", scope.makeOpName("OrderedMapIncompleteSize"));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    Class[] dtypesArray = new Class[dtypes.size()];
-    for (int i = 0; i < dtypesArray.length; ++i) {
-      dtypesArray[i] = dtypes.get(i);
-    }
-    opBuilder.setAttr("dtypes", dtypesArray);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.capacity != null) {
-          opBuilder.setAttr("capacity", opts.capacity);
-        }
-        if (opts.memoryLimit != null) {
-          opBuilder.setAttr("memory_limit", opts.memoryLimit);
-        }
-        if (opts.container != null) {
-          opBuilder.setAttr("container", opts.container);
-        }
-        if (opts.sharedName != null) {
-          opBuilder.setAttr("shared_name", opts.sharedName);
-        }
-      }
-    }
-    return new OrderedMapIncompleteSize(opBuilder.build());
-  }
-  
-  /**
-   * @param capacity 
-   */
-  public static Options capacity(Long capacity) {
-    return new Options().capacity(capacity);
-  }
-  
-  /**
-   * @param memoryLimit 
-   */
-  public static Options memoryLimit(Long memoryLimit) {
-    return new Options().memoryLimit(memoryLimit);
-  }
-  
-  /**
-   * @param container 
-   */
-  public static Options container(String container) {
-    return new Options().container(container);
-  }
-  
-  /**
-   * @param sharedName 
-   */
-  public static Options sharedName(String sharedName) {
-    return new Options().sharedName(sharedName);
-  }
-  
-  /**
-   */
-  public Output<TInt32> size() {
-    return size;
-  }
-  
-  @Override
-  public Output<TInt32> asOutput(Scope scope) {
-    return size;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "OrderedMapIncompleteSize";
-  
-  private Output<TInt32> size;
-  
-  private OrderedMapIncompleteSize(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    size = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OrderedMapPeek.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OrderedMapPeek.java
deleted file mode 100644
index f330950304d..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OrderedMapPeek.java
+++ /dev/null
@@ -1,183 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import java.util.Arrays;
-import java.util.Iterator;
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
-
-/**
- * Op peeks at the values at the specified key.  If the
- * <p>
- * underlying container does not contain this key
- * this op will block until it does.   This Op is optimized for
- * performance.
- */
-@Operator
-public final class OrderedMapPeek extends RawOp implements Iterable<Operand<TType>> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.core.OrderedMapPeek}
-   */
-  public static class Options {
-    
-    /**
-     * @param capacity 
-     */
-    public Options capacity(Long capacity) {
-      this.capacity = capacity;
-      return this;
-    }
-    
-    /**
-     * @param memoryLimit 
-     */
-    public Options memoryLimit(Long memoryLimit) {
-      this.memoryLimit = memoryLimit;
-      return this;
-    }
-    
-    /**
-     * @param container 
-     */
-    public Options container(String container) {
-      this.container = container;
-      return this;
-    }
-    
-    /**
-     * @param sharedName 
-     */
-    public Options sharedName(String sharedName) {
-      this.sharedName = sharedName;
-      return this;
-    }
-    
-    private Long capacity;
-    private Long memoryLimit;
-    private String container;
-    private String sharedName;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new OrderedMapPeek operation.
-   * 
-   * @param scope current scope
-   * @param key 
-   * @param indices 
-   * @param dtypes 
-   * @param options carries optional attributes values
-   * @return a new instance of OrderedMapPeek
-   */
-  @Endpoint(describeByClass = true)
-  public static OrderedMapPeek create(Scope scope, Operand<TInt64> key, Operand<TInt32> indices, List<Class<?>> dtypes, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("OrderedMapPeek", scope.makeOpName("OrderedMapPeek"));
-    opBuilder.addInput(key.asOutput(scope));
-    opBuilder.addInput(indices.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    Class[] dtypesArray = new Class[dtypes.size()];
-    for (int i = 0; i < dtypesArray.length; ++i) {
-      dtypesArray[i] = dtypes.get(i);
-    }
-    opBuilder.setAttr("dtypes", dtypesArray);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.capacity != null) {
-          opBuilder.setAttr("capacity", opts.capacity);
-        }
-        if (opts.memoryLimit != null) {
-          opBuilder.setAttr("memory_limit", opts.memoryLimit);
-        }
-        if (opts.container != null) {
-          opBuilder.setAttr("container", opts.container);
-        }
-        if (opts.sharedName != null) {
-          opBuilder.setAttr("shared_name", opts.sharedName);
-        }
-      }
-    }
-    return new OrderedMapPeek(opBuilder.build());
-  }
-  
-  /**
-   * @param capacity 
-   */
-  public static Options capacity(Long capacity) {
-    return new Options().capacity(capacity);
-  }
-  
-  /**
-   * @param memoryLimit 
-   */
-  public static Options memoryLimit(Long memoryLimit) {
-    return new Options().memoryLimit(memoryLimit);
-  }
-  
-  /**
-   * @param container 
-   */
-  public static Options container(String container) {
-    return new Options().container(container);
-  }
-  
-  /**
-   * @param sharedName 
-   */
-  public static Options sharedName(String sharedName) {
-    return new Options().sharedName(sharedName);
-  }
-  
-  /**
-   */
-  public List<Output<?>> values() {
-    return values;
-  }
-  
-  @Override
-  @SuppressWarnings({"rawtypes", "unchecked"})
-  public Iterator<Operand<TType>> iterator() {
-    return (Iterator) values.iterator();
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "OrderedMapPeek";
-  
-  private List<Output<?>> values;
-  
-  private OrderedMapPeek(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    int valuesLength = operation.outputListLength("values");
-    values = Arrays.asList(operation.outputList(outputIdx, valuesLength));
-    outputIdx += valuesLength;
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OrderedMapSize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OrderedMapSize.java
deleted file mode 100644
index 60bed92f128..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OrderedMapSize.java
+++ /dev/null
@@ -1,168 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
-
-/**
- * Op returns the number of elements in the underlying container.
- */
-@Operator
-public final class OrderedMapSize extends RawOp implements Operand<TInt32> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.core.OrderedMapSize}
-   */
-  public static class Options {
-    
-    /**
-     * @param capacity 
-     */
-    public Options capacity(Long capacity) {
-      this.capacity = capacity;
-      return this;
-    }
-    
-    /**
-     * @param memoryLimit 
-     */
-    public Options memoryLimit(Long memoryLimit) {
-      this.memoryLimit = memoryLimit;
-      return this;
-    }
-    
-    /**
-     * @param container 
-     */
-    public Options container(String container) {
-      this.container = container;
-      return this;
-    }
-    
-    /**
-     * @param sharedName 
-     */
-    public Options sharedName(String sharedName) {
-      this.sharedName = sharedName;
-      return this;
-    }
-    
-    private Long capacity;
-    private Long memoryLimit;
-    private String container;
-    private String sharedName;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new OrderedMapSize operation.
-   * 
-   * @param scope current scope
-   * @param dtypes 
-   * @param options carries optional attributes values
-   * @return a new instance of OrderedMapSize
-   */
-  @Endpoint(describeByClass = true)
-  public static OrderedMapSize create(Scope scope, List<Class<?>> dtypes, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("OrderedMapSize", scope.makeOpName("OrderedMapSize"));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    Class[] dtypesArray = new Class[dtypes.size()];
-    for (int i = 0; i < dtypesArray.length; ++i) {
-      dtypesArray[i] = dtypes.get(i);
-    }
-    opBuilder.setAttr("dtypes", dtypesArray);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.capacity != null) {
-          opBuilder.setAttr("capacity", opts.capacity);
-        }
-        if (opts.memoryLimit != null) {
-          opBuilder.setAttr("memory_limit", opts.memoryLimit);
-        }
-        if (opts.container != null) {
-          opBuilder.setAttr("container", opts.container);
-        }
-        if (opts.sharedName != null) {
-          opBuilder.setAttr("shared_name", opts.sharedName);
-        }
-      }
-    }
-    return new OrderedMapSize(opBuilder.build());
-  }
-  
-  /**
-   * @param capacity 
-   */
-  public static Options capacity(Long capacity) {
-    return new Options().capacity(capacity);
-  }
-  
-  /**
-   * @param memoryLimit 
-   */
-  public static Options memoryLimit(Long memoryLimit) {
-    return new Options().memoryLimit(memoryLimit);
-  }
-  
-  /**
-   * @param container 
-   */
-  public static Options container(String container) {
-    return new Options().container(container);
-  }
-  
-  /**
-   * @param sharedName 
-   */
-  public static Options sharedName(String sharedName) {
-    return new Options().sharedName(sharedName);
-  }
-  
-  /**
-   */
-  public Output<TInt32> size() {
-    return size;
-  }
-  
-  @Override
-  public Output<TInt32> asOutput(Scope scope) {
-    return size;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "OrderedMapSize";
-  
-  private Output<TInt32> size;
-  
-  private OrderedMapSize(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    size = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OrderedMapStage.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OrderedMapStage.java
deleted file mode 100644
index 93f369860f0..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OrderedMapStage.java
+++ /dev/null
@@ -1,167 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.Operands;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.TInt64;
-
-/**
- * Stage (key, values) in the underlying container which behaves like a ordered
- * <p>
- * associative container.   Elements are ordered by key.
- */
-@Operator
-public final class OrderedMapStage extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.core.OrderedMapStage}
-   */
-  public static class Options {
-    
-    /**
-     * @param capacity Maximum number of elements in the Staging Area. If > 0, inserts
-     * on the container will block when the capacity is reached.
-     */
-    public Options capacity(Long capacity) {
-      this.capacity = capacity;
-      return this;
-    }
-    
-    /**
-     * @param memoryLimit 
-     */
-    public Options memoryLimit(Long memoryLimit) {
-      this.memoryLimit = memoryLimit;
-      return this;
-    }
-    
-    /**
-     * @param container If non-empty, this queue is placed in the given container. Otherwise,
-     * a default container is used.
-     */
-    public Options container(String container) {
-      this.container = container;
-      return this;
-    }
-    
-    /**
-     * @param sharedName It is necessary to match this name to the matching Unstage Op.
-     */
-    public Options sharedName(String sharedName) {
-      this.sharedName = sharedName;
-      return this;
-    }
-    
-    private Long capacity;
-    private Long memoryLimit;
-    private String container;
-    private String sharedName;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new OrderedMapStage operation.
-   * 
-   * @param scope current scope
-   * @param key int64
-   * @param indices 
-   * @param values a list of tensors
-   * dtypes A list of data types that inserted values should adhere to.
-   * @param dtypes 
-   * @param options carries optional attributes values
-   * @return a new instance of OrderedMapStage
-   */
-  @Endpoint(describeByClass = true)
-  public static OrderedMapStage create(Scope scope, Operand<TInt64> key, Operand<TInt32> indices, Iterable<Operand<?>> values, List<Class<?>> dtypes, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("OrderedMapStage", scope.makeOpName("OrderedMapStage"));
-    opBuilder.addInput(key.asOutput(scope));
-    opBuilder.addInput(indices.asOutput(scope));
-    opBuilder.addInputList(Operands.asOutputs(scope, values));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    Class[] dtypesArray = new Class[dtypes.size()];
-    for (int i = 0; i < dtypesArray.length; ++i) {
-      dtypesArray[i] = dtypes.get(i);
-    }
-    opBuilder.setAttr("dtypes", dtypesArray);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.capacity != null) {
-          opBuilder.setAttr("capacity", opts.capacity);
-        }
-        if (opts.memoryLimit != null) {
-          opBuilder.setAttr("memory_limit", opts.memoryLimit);
-        }
-        if (opts.container != null) {
-          opBuilder.setAttr("container", opts.container);
-        }
-        if (opts.sharedName != null) {
-          opBuilder.setAttr("shared_name", opts.sharedName);
-        }
-      }
-    }
-    return new OrderedMapStage(opBuilder.build());
-  }
-  
-  /**
-   * @param capacity Maximum number of elements in the Staging Area. If > 0, inserts
-   * on the container will block when the capacity is reached.
-   */
-  public static Options capacity(Long capacity) {
-    return new Options().capacity(capacity);
-  }
-  
-  /**
-   * @param memoryLimit 
-   */
-  public static Options memoryLimit(Long memoryLimit) {
-    return new Options().memoryLimit(memoryLimit);
-  }
-  
-  /**
-   * @param container If non-empty, this queue is placed in the given container. Otherwise,
-   * a default container is used.
-   */
-  public static Options container(String container) {
-    return new Options().container(container);
-  }
-  
-  /**
-   * @param sharedName It is necessary to match this name to the matching Unstage Op.
-   */
-  public static Options sharedName(String sharedName) {
-    return new Options().sharedName(sharedName);
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "OrderedMapStage";
-  
-  private OrderedMapStage(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OrderedMapUnstage.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OrderedMapUnstage.java
deleted file mode 100644
index 28156e9c09c..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OrderedMapUnstage.java
+++ /dev/null
@@ -1,182 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import java.util.Arrays;
-import java.util.Iterator;
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
-
-/**
- * Op removes and returns the values associated with the key
- * <p>
- * from the underlying container.   If the underlying container
- * does not contain this key, the op will block until it does.
- */
-@Operator
-public final class OrderedMapUnstage extends RawOp implements Iterable<Operand<TType>> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.core.OrderedMapUnstage}
-   */
-  public static class Options {
-    
-    /**
-     * @param capacity 
-     */
-    public Options capacity(Long capacity) {
-      this.capacity = capacity;
-      return this;
-    }
-    
-    /**
-     * @param memoryLimit 
-     */
-    public Options memoryLimit(Long memoryLimit) {
-      this.memoryLimit = memoryLimit;
-      return this;
-    }
-    
-    /**
-     * @param container 
-     */
-    public Options container(String container) {
-      this.container = container;
-      return this;
-    }
-    
-    /**
-     * @param sharedName 
-     */
-    public Options sharedName(String sharedName) {
-      this.sharedName = sharedName;
-      return this;
-    }
-    
-    private Long capacity;
-    private Long memoryLimit;
-    private String container;
-    private String sharedName;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new OrderedMapUnstage operation.
-   * 
-   * @param scope current scope
-   * @param key 
-   * @param indices 
-   * @param dtypes 
-   * @param options carries optional attributes values
-   * @return a new instance of OrderedMapUnstage
-   */
-  @Endpoint(describeByClass = true)
-  public static OrderedMapUnstage create(Scope scope, Operand<TInt64> key, Operand<TInt32> indices, List<Class<?>> dtypes, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("OrderedMapUnstage", scope.makeOpName("OrderedMapUnstage"));
-    opBuilder.addInput(key.asOutput(scope));
-    opBuilder.addInput(indices.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    Class[] dtypesArray = new Class[dtypes.size()];
-    for (int i = 0; i < dtypesArray.length; ++i) {
-      dtypesArray[i] = dtypes.get(i);
-    }
-    opBuilder.setAttr("dtypes", dtypesArray);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.capacity != null) {
-          opBuilder.setAttr("capacity", opts.capacity);
-        }
-        if (opts.memoryLimit != null) {
-          opBuilder.setAttr("memory_limit", opts.memoryLimit);
-        }
-        if (opts.container != null) {
-          opBuilder.setAttr("container", opts.container);
-        }
-        if (opts.sharedName != null) {
-          opBuilder.setAttr("shared_name", opts.sharedName);
-        }
-      }
-    }
-    return new OrderedMapUnstage(opBuilder.build());
-  }
-  
-  /**
-   * @param capacity 
-   */
-  public static Options capacity(Long capacity) {
-    return new Options().capacity(capacity);
-  }
-  
-  /**
-   * @param memoryLimit 
-   */
-  public static Options memoryLimit(Long memoryLimit) {
-    return new Options().memoryLimit(memoryLimit);
-  }
-  
-  /**
-   * @param container 
-   */
-  public static Options container(String container) {
-    return new Options().container(container);
-  }
-  
-  /**
-   * @param sharedName 
-   */
-  public static Options sharedName(String sharedName) {
-    return new Options().sharedName(sharedName);
-  }
-  
-  /**
-   */
-  public List<Output<?>> values() {
-    return values;
-  }
-  
-  @Override
-  @SuppressWarnings({"rawtypes", "unchecked"})
-  public Iterator<Operand<TType>> iterator() {
-    return (Iterator) values.iterator();
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "OrderedMapUnstage";
-  
-  private List<Output<?>> values;
-  
-  private OrderedMapUnstage(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    int valuesLength = operation.outputListLength("values");
-    values = Arrays.asList(operation.outputList(outputIdx, valuesLength));
-    outputIdx += valuesLength;
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OrderedMapUnstageNoKey.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OrderedMapUnstageNoKey.java
deleted file mode 100644
index b8eeff0ee9c..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OrderedMapUnstageNoKey.java
+++ /dev/null
@@ -1,180 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import java.util.Arrays;
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.TInt64;
-
-/**
- * Op removes and returns the (key, value) element with the smallest
- * <p>
- * key from the underlying container.   If the underlying container
- * does not contain elements, the op will block until it does.
- */
-@Operator
-public final class OrderedMapUnstageNoKey extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.core.OrderedMapUnstageNoKey}
-   */
-  public static class Options {
-    
-    /**
-     * @param capacity 
-     */
-    public Options capacity(Long capacity) {
-      this.capacity = capacity;
-      return this;
-    }
-    
-    /**
-     * @param memoryLimit 
-     */
-    public Options memoryLimit(Long memoryLimit) {
-      this.memoryLimit = memoryLimit;
-      return this;
-    }
-    
-    /**
-     * @param container 
-     */
-    public Options container(String container) {
-      this.container = container;
-      return this;
-    }
-    
-    /**
-     * @param sharedName 
-     */
-    public Options sharedName(String sharedName) {
-      this.sharedName = sharedName;
-      return this;
-    }
-    
-    private Long capacity;
-    private Long memoryLimit;
-    private String container;
-    private String sharedName;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new OrderedMapUnstageNoKey operation.
-   * 
-   * @param scope current scope
-   * @param indices 
-   * @param dtypes 
-   * @param options carries optional attributes values
-   * @return a new instance of OrderedMapUnstageNoKey
-   */
-  @Endpoint(describeByClass = true)
-  public static OrderedMapUnstageNoKey create(Scope scope, Operand<TInt32> indices, List<Class<?>> dtypes, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("OrderedMapUnstageNoKey", scope.makeOpName("OrderedMapUnstageNoKey"));
-    opBuilder.addInput(indices.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    Class[] dtypesArray = new Class[dtypes.size()];
-    for (int i = 0; i < dtypesArray.length; ++i) {
-      dtypesArray[i] = dtypes.get(i);
-    }
-    opBuilder.setAttr("dtypes", dtypesArray);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.capacity != null) {
-          opBuilder.setAttr("capacity", opts.capacity);
-        }
-        if (opts.memoryLimit != null) {
-          opBuilder.setAttr("memory_limit", opts.memoryLimit);
-        }
-        if (opts.container != null) {
-          opBuilder.setAttr("container", opts.container);
-        }
-        if (opts.sharedName != null) {
-          opBuilder.setAttr("shared_name", opts.sharedName);
-        }
-      }
-    }
-    return new OrderedMapUnstageNoKey(opBuilder.build());
-  }
-  
-  /**
-   * @param capacity 
-   */
-  public static Options capacity(Long capacity) {
-    return new Options().capacity(capacity);
-  }
-  
-  /**
-   * @param memoryLimit 
-   */
-  public static Options memoryLimit(Long memoryLimit) {
-    return new Options().memoryLimit(memoryLimit);
-  }
-  
-  /**
-   * @param container 
-   */
-  public static Options container(String container) {
-    return new Options().container(container);
-  }
-  
-  /**
-   * @param sharedName 
-   */
-  public static Options sharedName(String sharedName) {
-    return new Options().sharedName(sharedName);
-  }
-  
-  /**
-   */
-  public Output<TInt64> key() {
-    return key;
-  }
-  
-  /**
-   */
-  public List<Output<?>> values() {
-    return values;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "OrderedMapUnstageNoKey";
-  
-  private Output<TInt64> key;
-  private List<Output<?>> values;
-  
-  private OrderedMapUnstageNoKey(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    key = operation.output(outputIdx++);
-    int valuesLength = operation.outputListLength("values");
-    values = Arrays.asList(operation.outputList(outputIdx, valuesLength));
-    outputIdx += valuesLength;
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Pad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Pad.java
deleted file mode 100644
index bcb29dfb937..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Pad.java
+++ /dev/null
@@ -1,104 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * Pads a tensor.
- * <p>
- * This operation pads `input` according to the `paddings` and `constant_values`
- * you specify. `paddings` is an integer tensor with shape `[Dn, 2]`, where n is
- * the rank of `input`. For each dimension D of `input`, `paddings[D, 0]` indicates
- * how many padding values to add before the contents of `input` in that dimension,
- * and `paddings[D, 1]` indicates how many padding values to add after the contents
- * of `input` in that dimension. `constant_values` is a scalar tensor of the same
- * type as `input` that indicates the value to use for padding `input`.
- * <p>
- * The padded size of each dimension D of the output is:
- * <p>
- * `paddings(D, 0) + input.dim_size(D) + paddings(D, 1)`
- * <p>
- * For example:
- * <pre>{@code
- * # 't' is [[1, 1], [2, 2]]
- * # 'paddings' is [[1, 1], [2, 2]]
- * # 'constant_values' is 0
- * # rank of 't' is 2
- * pad(t, paddings) ==> [[0, 0, 0, 0, 0, 0]
- *                       [0, 0, 1, 1, 0, 0]
- *                       [0, 0, 2, 2, 0, 0]
- *                       [0, 0, 0, 0, 0, 0]]
- * }</pre>
- * 
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator
-public final class Pad<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new Pad operation.
-   * 
-   * @param scope current scope
-   * @param input 
-   * @param paddings 
-   * @param constantValues 
-   * @return a new instance of Pad
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> Pad<T> create(Scope scope, Operand<T> input, Operand<U> paddings, Operand<T> constantValues) {
-    OperationBuilder opBuilder = scope.env().opBuilder("PadV2", scope.makeOpName("Pad"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder.addInput(paddings.asOutput(scope));
-    opBuilder.addInput(constantValues.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new Pad<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "PadV2";
-  
-  private Output<T> output;
-  
-  private Pad(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ParallelConcat.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ParallelConcat.java
deleted file mode 100644
index fdc0b3a6763..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ParallelConcat.java
+++ /dev/null
@@ -1,96 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.op.Operands;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Concatenates a list of `N` tensors along the first dimension.
- * <p>
- * The input tensors are all required to have size 1 in the first dimension.
- * <p>
- * For example:
- * <pre>{@code
- * # 'x' is [[1, 4]]
- * # 'y' is [[2, 5]]
- * # 'z' is [[3, 6]]
- * parallel_concat([x, y, z]) => [[1, 4], [2, 5], [3, 6]]  # Pack along first dim.
- * }</pre>
- * The difference between concat and parallel_concat is that concat requires all
- * of the inputs be computed before the operation will begin but doesn't require
- * that the input shapes be known during graph construction.  Parallel concat
- * will copy pieces of the input into the output as they become available, in
- * some situations this can provide a performance benefit.
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator
-public final class ParallelConcat<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new ParallelConcat operation.
-   * 
-   * @param scope current scope
-   * @param values Tensors to be concatenated. All must have size 1 in the first dimension
-   * and same shape.
-   * @param shape the final shape of the result; should be equal to the shapes of any input
-   * but with the number of input values in the first dimension.
-   * @return a new instance of ParallelConcat
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> ParallelConcat<T> create(Scope scope, Iterable<Operand<T>> values, Shape shape) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ParallelConcat", scope.makeOpName("ParallelConcat"));
-    opBuilder.addInputList(Operands.asOutputs(scope, values));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("shape", shape);
-    return new ParallelConcat<T>(opBuilder.build());
-  }
-  
-  /**
-   * The concatenated tensor.
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ParallelConcat";
-  
-  private Output<T> output;
-  
-  private ParallelConcat(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ParallelDynamicStitch.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ParallelDynamicStitch.java
deleted file mode 100644
index f6a4e709687..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ParallelDynamicStitch.java
+++ /dev/null
@@ -1,132 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.Operands;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.family.TType;
-
-/**
- * Interleave the values from the `data` tensors into a single tensor.
- * <p>
- * Builds a merged tensor such that
- * <pre>{@code
- *     merged[indices[m][i, ..., j], ...] = data[m][i, ..., j, ...]
- * }</pre>
- * For example, if each `indices[m]` is scalar or vector, we have
- * <pre>{@code
- *     # Scalar indices:
- *     merged[indices[m], ...] = data[m][...]
- * 
- *     # Vector indices:
- *     merged[indices[m][i], ...] = data[m][i, ...]
- * }</pre>
- * Each `data[i].shape` must start with the corresponding `indices[i].shape`,
- * and the rest of `data[i].shape` must be constant w.r.t. `i`.  That is, we
- * must have `data[i].shape = indices[i].shape + constant`.  In terms of this
- * `constant`, the output shape is
- * <p>
- *     merged.shape = [max(indices)] + constant
- * <p>
- * Values may be merged in parallel, so if an index appears in both `indices[m][i]`
- * and `indices[n][j]`, the result may be invalid. This differs from the normal
- * DynamicStitch operator that defines the behavior in that case.
- * <p>
- * For example:
- * <pre>{@code
- *     indices[0] = 6
- *     indices[1] = [4, 1]
- *     indices[2] = [[5, 2], [0, 3]]
- *     data[0] = [61, 62]
- *     data[1] = [[41, 42], [11, 12]]
- *     data[2] = [[[51, 52], [21, 22]], [[1, 2], [31, 32]]]
- *     merged = [[1, 2], [11, 12], [21, 22], [31, 32], [41, 42],
- *               [51, 52], [61, 62]]
- * }</pre>
- * This method can be used to merge partitions created by `dynamic_partition`
- * as illustrated on the following example:
- * <pre>{@code
- *     # Apply function (increments x_i) on elements for which a certain condition
- *     # apply (x_i != -1 in this example).
- *     x=tf.constant([0.1, -1., 5.2, 4.3, -1., 7.4])
- *     condition_mask=tf.not_equal(x,tf.constant(-1.))
- *     partitioned_data = tf.dynamic_partition(
- *         x, tf.cast(condition_mask, tf.int32) , 2)
- *     partitioned_data[1] = partitioned_data[1] + 1.0
- *     condition_indices = tf.dynamic_partition(
- *         tf.range(tf.shape(x)[0]), tf.cast(condition_mask, tf.int32) , 2)
- *     x = tf.dynamic_stitch(condition_indices, partitioned_data)
- *     # Here x=[1.1, -1., 6.2, 5.3, -1, 8.4], the -1. values remain
- *     # unchanged.
- * }</pre>
- * <div style="width:70%; margin:auto; margin-bottom:10px; margin-top:20px;">
- * <img style="width:100%" src="https://www.tensorflow.org/images/DynamicStitch.png" alt>
- * </div>
- * 
- * @param <T> data type for {@code merged()} output
- */
-@Operator
-public final class ParallelDynamicStitch<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new ParallelDynamicStitch operation.
-   * 
-   * @param scope current scope
-   * @param indices 
-   * @param data 
-   * @return a new instance of ParallelDynamicStitch
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> ParallelDynamicStitch<T> create(Scope scope, Iterable<Operand<TInt32>> indices, Iterable<Operand<T>> data) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ParallelDynamicStitch", scope.makeOpName("ParallelDynamicStitch"));
-    opBuilder.addInputList(Operands.asOutputs(scope, indices));
-    opBuilder.addInputList(Operands.asOutputs(scope, data));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new ParallelDynamicStitch<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> merged() {
-    return merged;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return merged;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ParallelDynamicStitch";
-  
-  private Output<T> merged;
-  
-  private ParallelDynamicStitch(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    merged = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Placeholder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Placeholder.java
deleted file mode 100644
index f38e589d90c..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Placeholder.java
+++ /dev/null
@@ -1,116 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * A placeholder op for a value that will be fed into the computation.
- * <p>
- * N.B. This operation will fail with an error if it is executed. It is
- * intended as a way to represent a value that will always be fed, and to
- * provide attrs that enable the fed value to be checked at runtime.
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator
-public final class Placeholder<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.core.Placeholder}
-   */
-  public static class Options {
-    
-    /**
-     * @param shape (Optional) The shape of the tensor. If the shape has 0 dimensions, the
-     * shape is unconstrained.
-     */
-    public Options shape(Shape shape) {
-      this.shape = shape;
-      return this;
-    }
-    
-    private Shape shape;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new Placeholder operation.
-   * 
-   * @param scope current scope
-   * @param dtype The type of elements in the tensor.
-   * @param options carries optional attributes values
-   * @return a new instance of Placeholder
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> Placeholder<T> create(Scope scope, Class<T> dtype, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Placeholder", scope.makeOpName("Placeholder"));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("dtype", dtype);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.shape != null) {
-          opBuilder.setAttr("shape", opts.shape);
-        }
-      }
-    }
-    return new Placeholder<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param shape (Optional) The shape of the tensor. If the shape has 0 dimensions, the
-   * shape is unconstrained.
-   */
-  public static Options shape(Shape shape) {
-    return new Options().shape(shape);
-  }
-  
-  /**
-   * A placeholder tensor that must be replaced using the feed mechanism.
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Placeholder";
-  
-  private Output<T> output;
-  
-  private Placeholder(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/PlaceholderWithDefault.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/PlaceholderWithDefault.java
deleted file mode 100644
index 87d6691333a..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/PlaceholderWithDefault.java
+++ /dev/null
@@ -1,78 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * A placeholder op that passes through `input` when its output is not fed.
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator
-public final class PlaceholderWithDefault<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new PlaceholderWithDefault operation.
-   * 
-   * @param scope current scope
-   * @param input The default value to produce when `output` is not fed.
-   * @param shape The (possibly partial) shape of the tensor.
-   * @return a new instance of PlaceholderWithDefault
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> PlaceholderWithDefault<T> create(Scope scope, Operand<T> input, Shape shape) {
-    OperationBuilder opBuilder = scope.env().opBuilder("PlaceholderWithDefault", scope.makeOpName("PlaceholderWithDefault"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("shape", shape);
-    return new PlaceholderWithDefault<T>(opBuilder.build());
-  }
-  
-  /**
-   * A placeholder tensor that defaults to `input` if it is not fed.
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "PlaceholderWithDefault";
-  
-  private Output<T> output;
-  
-  private PlaceholderWithDefault(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Print.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Print.java
deleted file mode 100644
index 9406d259ee3..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Print.java
+++ /dev/null
@@ -1,111 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TString;
-
-/**
- * Prints a string scalar.
- * <p>
- * Prints a string scalar to the desired output_stream.
- */
-@Operator
-public final class Print extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.core.Print}
-   */
-  public static class Options {
-    
-    /**
-     * @param outputStream A string specifying the output stream or logging level to print to.
-     */
-    public Options outputStream(String outputStream) {
-      this.outputStream = outputStream;
-      return this;
-    }
-    
-    /**
-     * @param end 
-     */
-    public Options end(String end) {
-      this.end = end;
-      return this;
-    }
-    
-    private String outputStream;
-    private String end;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new Print operation.
-   * 
-   * @param scope current scope
-   * @param input The string scalar to print.
-   * @param options carries optional attributes values
-   * @return a new instance of Print
-   */
-  @Endpoint(describeByClass = true)
-  public static Print create(Scope scope, Operand<TString> input, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("PrintV2", scope.makeOpName("Print"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.outputStream != null) {
-          opBuilder.setAttr("output_stream", opts.outputStream);
-        }
-        if (opts.end != null) {
-          opBuilder.setAttr("end", opts.end);
-        }
-      }
-    }
-    return new Print(opBuilder.build());
-  }
-  
-  /**
-   * @param outputStream A string specifying the output stream or logging level to print to.
-   */
-  public static Options outputStream(String outputStream) {
-    return new Options().outputStream(outputStream);
-  }
-  
-  /**
-   * @param end 
-   */
-  public static Options end(String end) {
-    return new Options().end(end);
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "PrintV2";
-  
-  private Print(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Prod.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Prod.java
deleted file mode 100644
index 16e401cb760..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Prod.java
+++ /dev/null
@@ -1,118 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * Computes the product of elements across dimensions of a tensor.
- * <p>
- * Reduces `input` along the dimensions given in `axis`. Unless
- * `keep_dims` is true, the rank of the tensor is reduced by 1 for each entry in
- * `axis`. If `keep_dims` is true, the reduced dimensions are
- * retained with length 1.
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator
-public final class Prod<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.core.Prod}
-   */
-  public static class Options {
-    
-    /**
-     * @param keepDims If true, retain reduced dimensions with length 1.
-     */
-    public Options keepDims(Boolean keepDims) {
-      this.keepDims = keepDims;
-      return this;
-    }
-    
-    private Boolean keepDims;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new Prod operation.
-   * 
-   * @param scope current scope
-   * @param input The tensor to reduce.
-   * @param axis The dimensions to reduce. Must be in the range
-   * `[-rank(input), rank(input))`.
-   * @param options carries optional attributes values
-   * @return a new instance of Prod
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> Prod<T> create(Scope scope, Operand<T> input, Operand<U> axis, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Prod", scope.makeOpName("Prod"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder.addInput(axis.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.keepDims != null) {
-          opBuilder.setAttr("keep_dims", opts.keepDims);
-        }
-      }
-    }
-    return new Prod<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param keepDims If true, retain reduced dimensions with length 1.
-   */
-  public static Options keepDims(Boolean keepDims) {
-    return new Options().keepDims(keepDims);
-  }
-  
-  /**
-   * The reduced tensor.
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Prod";
-  
-  private Output<T> output;
-  
-  private Prod(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/QuantizedReshape.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/QuantizedReshape.java
deleted file mode 100644
index b2de966aa32..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/QuantizedReshape.java
+++ /dev/null
@@ -1,97 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * Reshapes a quantized tensor as per the Reshape op.
- * <p>
- * ```
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator
-public final class QuantizedReshape<T extends TType> extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new QuantizedReshape operation.
-   * 
-   * @param scope current scope
-   * @param tensor 
-   * @param shape Defines the shape of the output tensor.
-   * @param inputMin The minimum value of the input.
-   * @param inputMax The maximum value of the input.
-   * @return a new instance of QuantizedReshape
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> QuantizedReshape<T> create(Scope scope, Operand<T> tensor, Operand<U> shape, Operand<TFloat32> inputMin, Operand<TFloat32> inputMax) {
-    OperationBuilder opBuilder = scope.env().opBuilder("QuantizedReshape", scope.makeOpName("QuantizedReshape"));
-    opBuilder.addInput(tensor.asOutput(scope));
-    opBuilder.addInput(shape.asOutput(scope));
-    opBuilder.addInput(inputMin.asOutput(scope));
-    opBuilder.addInput(inputMax.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new QuantizedReshape<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  /**
-   * This value is copied from input_min.
-   */
-  public Output<TFloat32> outputMin() {
-    return outputMin;
-  }
-  
-  /**
-   * This value is copied from input_max.
-   */
-  public Output<TFloat32> outputMax() {
-    return outputMax;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "QuantizedReshape";
-  
-  private Output<T> output;
-  private Output<TFloat32> outputMin;
-  private Output<TFloat32> outputMax;
-  
-  private QuantizedReshape(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-    outputMin = operation.output(outputIdx++);
-    outputMax = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Range.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Range.java
deleted file mode 100644
index 528ea8017c5..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Range.java
+++ /dev/null
@@ -1,91 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Creates a sequence of numbers.
- * <p>
- * This operation creates a sequence of numbers that begins at `start` and
- * extends by increments of `delta` up to but not including `limit`.
- * <p>
- * For example:
- * <pre>{@code
- * # 'start' is 3
- * # 'limit' is 18
- * # 'delta' is 3
- * tf.range(start, limit, delta) ==> [3, 6, 9, 12, 15]
- * }</pre>
- * 
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator
-public final class Range<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new Range operation.
-   * 
-   * @param scope current scope
-   * @param start 0-D (scalar). First entry in the sequence.
-   * @param limit 0-D (scalar). Upper limit of sequence, exclusive.
-   * @param delta 0-D (scalar). Optional. Default is 1. Number that increments `start`.
-   * @return a new instance of Range
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> Range<T> create(Scope scope, Operand<T> start, Operand<T> limit, Operand<T> delta) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Range", scope.makeOpName("Range"));
-    opBuilder.addInput(start.asOutput(scope));
-    opBuilder.addInput(limit.asOutput(scope));
-    opBuilder.addInput(delta.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new Range<T>(opBuilder.build());
-  }
-  
-  /**
-   * 1-D.
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Range";
-  
-  private Output<T> output;
-  
-  private Range(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Rank.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Rank.java
deleted file mode 100644
index 84fd0a9c755..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Rank.java
+++ /dev/null
@@ -1,85 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.family.TType;
-
-/**
- * Returns the rank of a tensor.
- * <p>
- * This operation returns an integer representing the rank of `input`.
- * <p>
- * For example:
- * <pre>{@code
- * # 't' is [[[1, 1, 1], [2, 2, 2]], [[3, 3, 3], [4, 4, 4]]]
- * # shape of tensor 't' is [2, 2, 3]
- * rank(t) ==> 3
- * }</pre>
- * <b>Note</b>: The rank of a tensor is not the same as the rank of a matrix. The rank
- * of a tensor is the number of indices required to uniquely select each element
- * of the tensor. Rank is also known as "order", "degree", or "ndims."
- */
-@Operator
-public final class Rank extends RawOp implements Operand<TInt32> {
-  
-  /**
-   * Factory method to create a class wrapping a new Rank operation.
-   * 
-   * @param scope current scope
-   * @param input 
-   * @return a new instance of Rank
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> Rank create(Scope scope, Operand<T> input) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Rank", scope.makeOpName("Rank"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new Rank(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<TInt32> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<TInt32> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Rank";
-  
-  private Output<TInt32> output;
-  
-  private Rank(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ReadVariableOp.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ReadVariableOp.java
deleted file mode 100644
index 9ca734e780e..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ReadVariableOp.java
+++ /dev/null
@@ -1,83 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Reads the value of a variable.
- * <p>
- * The tensor returned by this operation is immutable.
- * <p>
- * The value returned by this operation is guaranteed to be influenced by all the
- * writes on which this operation depends directly or indirectly, and to not be
- * influenced by any of the writes which depend directly or indirectly on this
- * operation.
- * 
- * @param <T> data type for {@code value()} output
- */
-@Operator
-public final class ReadVariableOp<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new ReadVariableOp operation.
-   * 
-   * @param scope current scope
-   * @param resource handle to the resource in which to store the variable.
-   * @param dtype the dtype of the value.
-   * @return a new instance of ReadVariableOp
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> ReadVariableOp<T> create(Scope scope, Operand<?> resource, Class<T> dtype) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ReadVariableOp", scope.makeOpName("ReadVariableOp"));
-    opBuilder.addInput(resource.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("dtype", dtype);
-    return new ReadVariableOp<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> value() {
-    return value;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return value;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ReadVariableOp";
-  
-  private Output<T> value;
-  
-  private ReadVariableOp(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    value = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Recv.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Recv.java
deleted file mode 100644
index 002eee646d6..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Recv.java
+++ /dev/null
@@ -1,122 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Receives the named tensor from send_device on recv_device.
- * 
- * @param <T> data type for {@code tensor()} output
- */
-public final class Recv<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.core.Recv}
-   */
-  public static class Options {
-    
-    /**
-     * @param clientTerminated If set to true, this indicates that the node was added
-     * to the graph as a result of a client-side feed or fetch of Tensor data,
-     * in which case the corresponding send or recv is expected to be managed
-     * locally by the caller.
-     */
-    public Options clientTerminated(Boolean clientTerminated) {
-      this.clientTerminated = clientTerminated;
-      return this;
-    }
-    
-    private Boolean clientTerminated;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new Recv operation.
-   * 
-   * @param scope current scope
-   * @param tensorType 
-   * @param tensorName The name of the tensor to receive.
-   * @param sendDevice The name of the device sending the tensor.
-   * @param sendDeviceIncarnation The current incarnation of send_device.
-   * @param recvDevice The name of the device receiving the tensor.
-   * @param options carries optional attributes values
-   * @return a new instance of Recv
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> Recv<T> create(Scope scope, Class<T> tensorType, String tensorName, String sendDevice, Long sendDeviceIncarnation, String recvDevice, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Recv", scope.makeOpName("Recv"));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("tensor_type", tensorType);
-    opBuilder.setAttr("tensor_name", tensorName);
-    opBuilder.setAttr("send_device", sendDevice);
-    opBuilder.setAttr("send_device_incarnation", sendDeviceIncarnation);
-    opBuilder.setAttr("recv_device", recvDevice);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.clientTerminated != null) {
-          opBuilder.setAttr("client_terminated", opts.clientTerminated);
-        }
-      }
-    }
-    return new Recv<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param clientTerminated If set to true, this indicates that the node was added
-   * to the graph as a result of a client-side feed or fetch of Tensor data,
-   * in which case the corresponding send or recv is expected to be managed
-   * locally by the caller.
-   */
-  public static Options clientTerminated(Boolean clientTerminated) {
-    return new Options().clientTerminated(clientTerminated);
-  }
-  
-  /**
-   * The tensor to receive.
-   */
-  public Output<T> tensor() {
-    return tensor;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return tensor;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Recv";
-  
-  private Output<T> tensor;
-  
-  private Recv(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    tensor = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ReduceAll.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ReduceAll.java
deleted file mode 100644
index 0bdb46ee38c..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ReduceAll.java
+++ /dev/null
@@ -1,116 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TBool;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Computes the "logical and" of elements across dimensions of a tensor.
- * <p>
- * Reduces `input` along the dimensions given in `axis`. Unless
- * `keep_dims` is true, the rank of the tensor is reduced by 1 for each entry in
- * `axis`. If `keep_dims` is true, the reduced dimensions are
- * retained with length 1.
- */
-@Operator
-public final class ReduceAll extends RawOp implements Operand<TBool> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.core.ReduceAll}
-   */
-  public static class Options {
-    
-    /**
-     * @param keepDims If true, retain reduced dimensions with length 1.
-     */
-    public Options keepDims(Boolean keepDims) {
-      this.keepDims = keepDims;
-      return this;
-    }
-    
-    private Boolean keepDims;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new ReduceAll operation.
-   * 
-   * @param scope current scope
-   * @param input The tensor to reduce.
-   * @param axis The dimensions to reduce. Must be in the range
-   * `[-rank(input), rank(input))`.
-   * @param options carries optional attributes values
-   * @return a new instance of ReduceAll
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> ReduceAll create(Scope scope, Operand<TBool> input, Operand<T> axis, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("All", scope.makeOpName("ReduceAll"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder.addInput(axis.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.keepDims != null) {
-          opBuilder.setAttr("keep_dims", opts.keepDims);
-        }
-      }
-    }
-    return new ReduceAll(opBuilder.build());
-  }
-  
-  /**
-   * @param keepDims If true, retain reduced dimensions with length 1.
-   */
-  public static Options keepDims(Boolean keepDims) {
-    return new Options().keepDims(keepDims);
-  }
-  
-  /**
-   * The reduced tensor.
-   */
-  public Output<TBool> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<TBool> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "All";
-  
-  private Output<TBool> output;
-  
-  private ReduceAll(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ReduceAny.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ReduceAny.java
deleted file mode 100644
index b54d2a5c18d..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ReduceAny.java
+++ /dev/null
@@ -1,116 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TBool;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Computes the "logical or" of elements across dimensions of a tensor.
- * <p>
- * Reduces `input` along the dimensions given in `axis`. Unless
- * `keep_dims` is true, the rank of the tensor is reduced by 1 for each entry in
- * `axis`. If `keep_dims` is true, the reduced dimensions are
- * retained with length 1.
- */
-@Operator
-public final class ReduceAny extends RawOp implements Operand<TBool> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.core.ReduceAny}
-   */
-  public static class Options {
-    
-    /**
-     * @param keepDims If true, retain reduced dimensions with length 1.
-     */
-    public Options keepDims(Boolean keepDims) {
-      this.keepDims = keepDims;
-      return this;
-    }
-    
-    private Boolean keepDims;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new ReduceAny operation.
-   * 
-   * @param scope current scope
-   * @param input The tensor to reduce.
-   * @param axis The dimensions to reduce. Must be in the range
-   * `[-rank(input), rank(input))`.
-   * @param options carries optional attributes values
-   * @return a new instance of ReduceAny
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> ReduceAny create(Scope scope, Operand<TBool> input, Operand<T> axis, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Any", scope.makeOpName("ReduceAny"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder.addInput(axis.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.keepDims != null) {
-          opBuilder.setAttr("keep_dims", opts.keepDims);
-        }
-      }
-    }
-    return new ReduceAny(opBuilder.build());
-  }
-  
-  /**
-   * @param keepDims If true, retain reduced dimensions with length 1.
-   */
-  public static Options keepDims(Boolean keepDims) {
-    return new Options().keepDims(keepDims);
-  }
-  
-  /**
-   * The reduced tensor.
-   */
-  public Output<TBool> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<TBool> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Any";
-  
-  private Output<TBool> output;
-  
-  private ReduceAny(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ReduceMax.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ReduceMax.java
deleted file mode 100644
index 355390df9c5..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ReduceMax.java
+++ /dev/null
@@ -1,118 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * Computes the maximum of elements across dimensions of a tensor.
- * <p>
- * Reduces `input` along the dimensions given in `axis`. Unless
- * `keep_dims` is true, the rank of the tensor is reduced by 1 for each entry in
- * `axis`. If `keep_dims` is true, the reduced dimensions are
- * retained with length 1.
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator
-public final class ReduceMax<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.core.ReduceMax}
-   */
-  public static class Options {
-    
-    /**
-     * @param keepDims If true, retain reduced dimensions with length 1.
-     */
-    public Options keepDims(Boolean keepDims) {
-      this.keepDims = keepDims;
-      return this;
-    }
-    
-    private Boolean keepDims;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new ReduceMax operation.
-   * 
-   * @param scope current scope
-   * @param input The tensor to reduce.
-   * @param axis The dimensions to reduce. Must be in the range
-   * `[-rank(input), rank(input))`.
-   * @param options carries optional attributes values
-   * @return a new instance of ReduceMax
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> ReduceMax<T> create(Scope scope, Operand<T> input, Operand<U> axis, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Max", scope.makeOpName("ReduceMax"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder.addInput(axis.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.keepDims != null) {
-          opBuilder.setAttr("keep_dims", opts.keepDims);
-        }
-      }
-    }
-    return new ReduceMax<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param keepDims If true, retain reduced dimensions with length 1.
-   */
-  public static Options keepDims(Boolean keepDims) {
-    return new Options().keepDims(keepDims);
-  }
-  
-  /**
-   * The reduced tensor.
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Max";
-  
-  private Output<T> output;
-  
-  private ReduceMax(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ReduceMin.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ReduceMin.java
deleted file mode 100644
index 0e5b82c5267..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ReduceMin.java
+++ /dev/null
@@ -1,118 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * Computes the minimum of elements across dimensions of a tensor.
- * <p>
- * Reduces `input` along the dimensions given in `axis`. Unless
- * `keep_dims` is true, the rank of the tensor is reduced by 1 for each entry in
- * `axis`. If `keep_dims` is true, the reduced dimensions are
- * retained with length 1.
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator
-public final class ReduceMin<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.core.ReduceMin}
-   */
-  public static class Options {
-    
-    /**
-     * @param keepDims If true, retain reduced dimensions with length 1.
-     */
-    public Options keepDims(Boolean keepDims) {
-      this.keepDims = keepDims;
-      return this;
-    }
-    
-    private Boolean keepDims;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new ReduceMin operation.
-   * 
-   * @param scope current scope
-   * @param input The tensor to reduce.
-   * @param axis The dimensions to reduce. Must be in the range
-   * `[-rank(input), rank(input))`.
-   * @param options carries optional attributes values
-   * @return a new instance of ReduceMin
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> ReduceMin<T> create(Scope scope, Operand<T> input, Operand<U> axis, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Min", scope.makeOpName("ReduceMin"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder.addInput(axis.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.keepDims != null) {
-          opBuilder.setAttr("keep_dims", opts.keepDims);
-        }
-      }
-    }
-    return new ReduceMin<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param keepDims If true, retain reduced dimensions with length 1.
-   */
-  public static Options keepDims(Boolean keepDims) {
-    return new Options().keepDims(keepDims);
-  }
-  
-  /**
-   * The reduced tensor.
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Min";
-  
-  private Output<T> output;
-  
-  private ReduceMin(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ReduceProd.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ReduceProd.java
deleted file mode 100644
index 779372c1079..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ReduceProd.java
+++ /dev/null
@@ -1,118 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * Computes the product of elements across dimensions of a tensor.
- * <p>
- * Reduces `input` along the dimensions given in `axis`. Unless
- * `keep_dims` is true, the rank of the tensor is reduced by 1 for each entry in
- * `axis`. If `keep_dims` is true, the reduced dimensions are
- * retained with length 1.
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator
-public final class ReduceProd<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.core.ReduceProd}
-   */
-  public static class Options {
-    
-    /**
-     * @param keepDims If true, retain reduced dimensions with length 1.
-     */
-    public Options keepDims(Boolean keepDims) {
-      this.keepDims = keepDims;
-      return this;
-    }
-    
-    private Boolean keepDims;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new ReduceProd operation.
-   * 
-   * @param scope current scope
-   * @param input The tensor to reduce.
-   * @param axis The dimensions to reduce. Must be in the range
-   * `[-rank(input), rank(input))`.
-   * @param options carries optional attributes values
-   * @return a new instance of ReduceProd
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> ReduceProd<T> create(Scope scope, Operand<T> input, Operand<U> axis, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Prod", scope.makeOpName("ReduceProd"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder.addInput(axis.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.keepDims != null) {
-          opBuilder.setAttr("keep_dims", opts.keepDims);
-        }
-      }
-    }
-    return new ReduceProd<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param keepDims If true, retain reduced dimensions with length 1.
-   */
-  public static Options keepDims(Boolean keepDims) {
-    return new Options().keepDims(keepDims);
-  }
-  
-  /**
-   * The reduced tensor.
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Prod";
-  
-  private Output<T> output;
-  
-  private ReduceProd(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ReduceSum.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ReduceSum.java
deleted file mode 100644
index 5cbce3b033c..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ReduceSum.java
+++ /dev/null
@@ -1,118 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * Computes the sum of elements across dimensions of a tensor.
- * <p>
- * Reduces `input` along the dimensions given in `axis`. Unless
- * `keep_dims` is true, the rank of the tensor is reduced by 1 for each entry in
- * `axis`. If `keep_dims` is true, the reduced dimensions are
- * retained with length 1.
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator
-public final class ReduceSum<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.core.ReduceSum}
-   */
-  public static class Options {
-    
-    /**
-     * @param keepDims If true, retain reduced dimensions with length 1.
-     */
-    public Options keepDims(Boolean keepDims) {
-      this.keepDims = keepDims;
-      return this;
-    }
-    
-    private Boolean keepDims;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new ReduceSum operation.
-   * 
-   * @param scope current scope
-   * @param input The tensor to reduce.
-   * @param axis The dimensions to reduce. Must be in the range
-   * `[-rank(input), rank(input))`.
-   * @param options carries optional attributes values
-   * @return a new instance of ReduceSum
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> ReduceSum<T> create(Scope scope, Operand<T> input, Operand<U> axis, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Sum", scope.makeOpName("ReduceSum"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder.addInput(axis.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.keepDims != null) {
-          opBuilder.setAttr("keep_dims", opts.keepDims);
-        }
-      }
-    }
-    return new ReduceSum<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param keepDims If true, retain reduced dimensions with length 1.
-   */
-  public static Options keepDims(Boolean keepDims) {
-    return new Options().keepDims(keepDims);
-  }
-  
-  /**
-   * The reduced tensor.
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Sum";
-  
-  private Output<T> output;
-  
-  private ReduceSum(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/RefEnter.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/RefEnter.java
deleted file mode 100644
index f36104fc2dd..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/RefEnter.java
+++ /dev/null
@@ -1,134 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Creates or finds a child frame, and makes `data` available to the child frame.
- * <p>
- * The unique `frame_name` is used by the `Executor` to identify frames. If
- * `is_constant` is true, `output` is a constant in the child frame; otherwise
- * it may be changed in the child frame. At most `parallel_iterations` iterations
- * are run in parallel in the child frame.
- * 
- * @param <T> data type for {@code output()} output
- */
-public final class RefEnter<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.core.RefEnter}
-   */
-  public static class Options {
-    
-    /**
-     * @param isConstant If true, the output is constant within the child frame.
-     */
-    public Options isConstant(Boolean isConstant) {
-      this.isConstant = isConstant;
-      return this;
-    }
-    
-    /**
-     * @param parallelIterations The number of iterations allowed to run in parallel.
-     */
-    public Options parallelIterations(Long parallelIterations) {
-      this.parallelIterations = parallelIterations;
-      return this;
-    }
-    
-    private Boolean isConstant;
-    private Long parallelIterations;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new RefEnter operation.
-   * 
-   * @param scope current scope
-   * @param data The tensor to be made available to the child frame.
-   * @param frameName The name of the child frame.
-   * @param options carries optional attributes values
-   * @return a new instance of RefEnter
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> RefEnter<T> create(Scope scope, Operand<T> data, String frameName, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("RefEnter", scope.makeOpName("RefEnter"));
-    opBuilder.addInput(data.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("frame_name", frameName);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.isConstant != null) {
-          opBuilder.setAttr("is_constant", opts.isConstant);
-        }
-        if (opts.parallelIterations != null) {
-          opBuilder.setAttr("parallel_iterations", opts.parallelIterations);
-        }
-      }
-    }
-    return new RefEnter<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param isConstant If true, the output is constant within the child frame.
-   */
-  public static Options isConstant(Boolean isConstant) {
-    return new Options().isConstant(isConstant);
-  }
-  
-  /**
-   * @param parallelIterations The number of iterations allowed to run in parallel.
-   */
-  public static Options parallelIterations(Long parallelIterations) {
-    return new Options().parallelIterations(parallelIterations);
-  }
-  
-  /**
-   * The same tensor as `data`.
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "RefEnter";
-  
-  private Output<T> output;
-  
-  private RefEnter(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/RefExit.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/RefExit.java
deleted file mode 100644
index 0438696ed02..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/RefExit.java
+++ /dev/null
@@ -1,76 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Exits the current frame to its parent frame.
- * <p>
- * Exit makes its input `data` available to the parent frame.
- * 
- * @param <T> data type for {@code output()} output
- */
-public final class RefExit<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new RefExit operation.
-   * 
-   * @param scope current scope
-   * @param data The tensor to be made available to the parent frame.
-   * @return a new instance of RefExit
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> RefExit<T> create(Scope scope, Operand<T> data) {
-    OperationBuilder opBuilder = scope.env().opBuilder("RefExit", scope.makeOpName("RefExit"));
-    opBuilder.addInput(data.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new RefExit<T>(opBuilder.build());
-  }
-  
-  /**
-   * The same tensor as `data`.
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "RefExit";
-  
-  private Output<T> output;
-  
-  private RefExit(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/RefIdentity.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/RefIdentity.java
deleted file mode 100644
index e3da108a249..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/RefIdentity.java
+++ /dev/null
@@ -1,73 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Return the same ref tensor as the input ref tensor.
- * 
- * @param <T> data type for {@code output()} output
- */
-public final class RefIdentity<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new RefIdentity operation.
-   * 
-   * @param scope current scope
-   * @param input 
-   * @return a new instance of RefIdentity
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> RefIdentity<T> create(Scope scope, Operand<T> input) {
-    OperationBuilder opBuilder = scope.env().opBuilder("RefIdentity", scope.makeOpName("RefIdentity"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new RefIdentity<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "RefIdentity";
-  
-  private Output<T> output;
-  
-  private RefIdentity(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/RefMerge.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/RefMerge.java
deleted file mode 100644
index 04ee46da33e..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/RefMerge.java
+++ /dev/null
@@ -1,86 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.Operands;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.family.TType;
-
-/**
- * Forwards the value of an available tensor from `inputs` to `output`.
- * <p>
- * `Merge` waits for at least one of the tensors in `inputs` to become available.
- * It is usually combined with `Switch` to implement branching.
- * <p>
- * `Merge` forwards the first tensor for become available to `output`, and sets
- * `value_index` to its index in `inputs`.
- * 
- * @param <T> data type for {@code output()} output
- */
-public final class RefMerge<T extends TType> extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new RefMerge operation.
-   * 
-   * @param scope current scope
-   * @param inputs The input tensors, exactly one of which will become available.
-   * @return a new instance of RefMerge
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> RefMerge<T> create(Scope scope, Iterable<Operand<T>> inputs) {
-    OperationBuilder opBuilder = scope.env().opBuilder("RefMerge", scope.makeOpName("RefMerge"));
-    opBuilder.addInputList(Operands.asOutputs(scope, inputs));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new RefMerge<T>(opBuilder.build());
-  }
-  
-  /**
-   * Will be set to the available input tensor.
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  /**
-   * The index of the chosen input tensor in `inputs`.
-   */
-  public Output<TInt32> valueIndex() {
-    return valueIndex;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "RefMerge";
-  
-  private Output<T> output;
-  private Output<TInt32> valueIndex;
-  
-  private RefMerge(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-    valueIndex = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/RefNextIteration.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/RefNextIteration.java
deleted file mode 100644
index 935207485a4..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/RefNextIteration.java
+++ /dev/null
@@ -1,75 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Makes its input available to the next iteration.
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator
-public final class RefNextIteration<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new RefNextIteration operation.
-   * 
-   * @param scope current scope
-   * @param data The tensor to be made available to the next iteration.
-   * @return a new instance of RefNextIteration
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> RefNextIteration<T> create(Scope scope, Operand<T> data) {
-    OperationBuilder opBuilder = scope.env().opBuilder("RefNextIteration", scope.makeOpName("RefNextIteration"));
-    opBuilder.addInput(data.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new RefNextIteration<T>(opBuilder.build());
-  }
-  
-  /**
-   * The same tensor as `data`.
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "RefNextIteration";
-  
-  private Output<T> output;
-  
-  private RefNextIteration(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/RefSelect.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/RefSelect.java
deleted file mode 100644
index 1678f30289f..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/RefSelect.java
+++ /dev/null
@@ -1,79 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.Operands;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.family.TType;
-
-/**
- * Forwards the `index`th element of `inputs` to `output`.
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator
-public final class RefSelect<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new RefSelect operation.
-   * 
-   * @param scope current scope
-   * @param index A scalar that determines the input that gets selected.
-   * @param inputs A list of ref tensors, one of which will be forwarded to `output`.
-   * @return a new instance of RefSelect
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> RefSelect<T> create(Scope scope, Operand<TInt32> index, Iterable<Operand<T>> inputs) {
-    OperationBuilder opBuilder = scope.env().opBuilder("RefSelect", scope.makeOpName("RefSelect"));
-    opBuilder.addInput(index.asOutput(scope));
-    opBuilder.addInputList(Operands.asOutputs(scope, inputs));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new RefSelect<T>(opBuilder.build());
-  }
-  
-  /**
-   * The forwarded tensor.
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "RefSelect";
-  
-  private Output<T> output;
-  
-  private RefSelect(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/RefSwitch.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/RefSwitch.java
deleted file mode 100644
index be17ff3fb60..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/RefSwitch.java
+++ /dev/null
@@ -1,87 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TBool;
-import org.tensorflow.types.family.TType;
-
-/**
- * Forwards the ref tensor `data` to the output port determined by `pred`.
- * <p>
- * If `pred` is true, the `data` input is forwarded to `output_true`. Otherwise,
- * the data goes to `output_false`.
- * <p>
- * See also `Switch` and `Merge`.
- * 
- * @param <T> data type for {@code outputFalse()} output
- */
-@Operator
-public final class RefSwitch<T extends TType> extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new RefSwitch operation.
-   * 
-   * @param scope current scope
-   * @param data The ref tensor to be forwarded to the appropriate output.
-   * @param pred A scalar that specifies which output port will receive data.
-   * @return a new instance of RefSwitch
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> RefSwitch<T> create(Scope scope, Operand<T> data, Operand<TBool> pred) {
-    OperationBuilder opBuilder = scope.env().opBuilder("RefSwitch", scope.makeOpName("RefSwitch"));
-    opBuilder.addInput(data.asOutput(scope));
-    opBuilder.addInput(pred.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new RefSwitch<T>(opBuilder.build());
-  }
-  
-  /**
-   * If `pred` is false, data will be forwarded to this output.
-   */
-  public Output<T> outputFalse() {
-    return outputFalse;
-  }
-  
-  /**
-   * If `pred` is true, data will be forwarded to this output.
-   */
-  public Output<T> outputTrue() {
-    return outputTrue;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "RefSwitch";
-  
-  private Output<T> outputFalse;
-  private Output<T> outputTrue;
-  
-  private RefSwitch(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    outputFalse = operation.output(outputIdx++);
-    outputTrue = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/RemoteFusedGraphExecute.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/RemoteFusedGraphExecute.java
deleted file mode 100644
index b92d9388f69..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/RemoteFusedGraphExecute.java
+++ /dev/null
@@ -1,97 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import java.util.Arrays;
-import java.util.Iterator;
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.Operands;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Execute a sub graph on a remote processor.
- * <p>
- * The graph specifications(such as graph itself, input tensors and output names)
- * are stored as a serialized protocol buffer of RemoteFusedGraphExecuteInfo
- * as serialized_remote_fused_graph_execute_info.
- * The specifications will be passed to a dedicated registered
- * remote fused graph executor.  The executor will send the graph specifications
- * to a remote processor and execute that graph.  The execution results
- * will be passed to consumer nodes as outputs of this node.
- */
-@Operator
-public final class RemoteFusedGraphExecute extends RawOp implements Iterable<Operand<TType>> {
-  
-  /**
-   * Factory method to create a class wrapping a new RemoteFusedGraphExecute operation.
-   * 
-   * @param scope current scope
-   * @param inputs Arbitrary number of tensors with arbitrary data types
-   * @param Toutputs 
-   * @param serializedRemoteFusedGraphExecuteInfo Serialized protocol buffer
-   * of RemoteFusedGraphExecuteInfo which contains graph specifications.
-   * @return a new instance of RemoteFusedGraphExecute
-   */
-  @Endpoint(describeByClass = true)
-  public static RemoteFusedGraphExecute create(Scope scope, Iterable<Operand<?>> inputs, List<Class<?>> Toutputs, String serializedRemoteFusedGraphExecuteInfo) {
-    OperationBuilder opBuilder = scope.env().opBuilder("RemoteFusedGraphExecute", scope.makeOpName("RemoteFusedGraphExecute"));
-    opBuilder.addInputList(Operands.asOutputs(scope, inputs));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    Class[] ToutputsArray = new Class[Toutputs.size()];
-    for (int i = 0; i < ToutputsArray.length; ++i) {
-      ToutputsArray[i] = Toutputs.get(i);
-    }
-    opBuilder.setAttr("Toutputs", ToutputsArray);
-    opBuilder.setAttr("serialized_remote_fused_graph_execute_info", serializedRemoteFusedGraphExecuteInfo);
-    return new RemoteFusedGraphExecute(opBuilder.build());
-  }
-  
-  /**
-   * Arbitrary number of tensors with arbitrary data types
-   */
-  public List<Output<?>> outputs() {
-    return outputs;
-  }
-  
-  @Override
-  @SuppressWarnings({"rawtypes", "unchecked"})
-  public Iterator<Operand<TType>> iterator() {
-    return (Iterator) outputs.iterator();
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "RemoteFusedGraphExecute";
-  
-  private List<Output<?>> outputs;
-  
-  private RemoteFusedGraphExecute(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    int outputsLength = operation.outputListLength("outputs");
-    outputs = Arrays.asList(operation.outputList(outputIdx, outputsLength));
-    outputIdx += outputsLength;
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Reshape.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Reshape.java
deleted file mode 100644
index 4e88d4d7f03..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Reshape.java
+++ /dev/null
@@ -1,137 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * Reshapes a tensor.
- * <p>
- * Given `tensor`, this operation returns a tensor that has the same values
- * as `tensor` with shape `shape`.
- * <p>
- * If one component of 1-D tensor `shape` is the special value -1, the size of that
- * dimension is computed so that the total size remains constant.  In particular, a
- * `shape` of `[-1]` flattens into 1-D.  At most one component of `shape` may be
- * unknown.
- * <p>
- * The `shape` must be 1-D and the operation returns a tensor with shape
- * `shape` filled with the values of `tensor`. In this case, the number of elements
- * implied by `shape` must be the same as the number of elements in `tensor`.
- * <p>
- * It is an error if `shape` is not 1-D.
- * <p>
- * For example:
- * <pre>{@code
- * # tensor 't' is [1, 2, 3, 4, 5, 6, 7, 8, 9]
- * # tensor 't' has shape [9]
- * reshape(t, [3, 3]) ==> [[1, 2, 3],
- *                         [4, 5, 6],
- *                         [7, 8, 9]]
- * 
- * # tensor 't' is [[[1, 1], [2, 2]],
- * #                [[3, 3], [4, 4]]]
- * # tensor 't' has shape [2, 2, 2]
- * reshape(t, [2, 4]) ==> [[1, 1, 2, 2],
- *                         [3, 3, 4, 4]]
- * 
- * # tensor 't' is [[[1, 1, 1],
- * #                 [2, 2, 2]],
- * #                [[3, 3, 3],
- * #                 [4, 4, 4]],
- * #                [[5, 5, 5],
- * #                 [6, 6, 6]]]
- * # tensor 't' has shape [3, 2, 3]
- * # pass '[-1]' to flatten 't'
- * reshape(t, [-1]) ==> [1, 1, 1, 2, 2, 2, 3, 3, 3, 4, 4, 4, 5, 5, 5, 6, 6, 6]
- * 
- * # -1 can also be used to infer the shape
- * 
- * # -1 is inferred to be 9:
- * reshape(t, [2, -1]) ==> [[1, 1, 1, 2, 2, 2, 3, 3, 3],
- *                          [4, 4, 4, 5, 5, 5, 6, 6, 6]]
- * # -1 is inferred to be 2:
- * reshape(t, [-1, 9]) ==> [[1, 1, 1, 2, 2, 2, 3, 3, 3],
- *                          [4, 4, 4, 5, 5, 5, 6, 6, 6]]
- * # -1 is inferred to be 3:
- * reshape(t, [ 2, -1, 3]) ==> [[[1, 1, 1],
- *                               [2, 2, 2],
- *                               [3, 3, 3]],
- *                              [[4, 4, 4],
- *                               [5, 5, 5],
- *                               [6, 6, 6]]]
- * 
- * # tensor 't' is [7]
- * # shape `[]` reshapes to a scalar
- * reshape(t, []) ==> 7
- * }</pre>
- * 
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator
-public final class Reshape<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new Reshape operation.
-   * 
-   * @param scope current scope
-   * @param tensor 
-   * @param shape Defines the shape of the output tensor.
-   * @return a new instance of Reshape
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> Reshape<T> create(Scope scope, Operand<T> tensor, Operand<U> shape) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Reshape", scope.makeOpName("Reshape"));
-    opBuilder.addInput(tensor.asOutput(scope));
-    opBuilder.addInput(shape.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new Reshape<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Reshape";
-  
-  private Output<T> output;
-  
-  private Reshape(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceCountUpTo.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceCountUpTo.java
deleted file mode 100644
index 50c2af8aa7e..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceCountUpTo.java
+++ /dev/null
@@ -1,81 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Increments variable pointed to by 'resource' until it reaches 'limit'.
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator
-public final class ResourceCountUpTo<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new ResourceCountUpTo operation.
-   * 
-   * @param scope current scope
-   * @param resource Should be from a scalar `Variable` node.
-   * @param limit If incrementing ref would bring it above limit, instead generates an
-   * 'OutOfRange' error.
-   * @param T 
-   * @return a new instance of ResourceCountUpTo
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> ResourceCountUpTo<T> create(Scope scope, Operand<?> resource, Long limit, Class<T> T) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ResourceCountUpTo", scope.makeOpName("ResourceCountUpTo"));
-    opBuilder.addInput(resource.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("limit", limit);
-    opBuilder.setAttr("T", T);
-    return new ResourceCountUpTo<T>(opBuilder.build());
-  }
-  
-  /**
-   * A copy of the input before increment. If nothing else modifies the
-   * input, the values produced will all be distinct.
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ResourceCountUpTo";
-  
-  private Output<T> output;
-  
-  private ResourceCountUpTo(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceGather.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceGather.java
deleted file mode 100644
index c9515a25932..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceGather.java
+++ /dev/null
@@ -1,146 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * Gather slices from the variable pointed to by `resource` according to `indices`.
- * <p>
- * `indices` must be an integer tensor of any dimension (usually 0-D or 1-D).
- * Produces an output tensor with shape `indices.shape + params.shape[1:]` where:
- * <pre>{@code
- *     # Scalar indices
- *     output[:, ..., :] = params[indices, :, ... :]
- * 
- *     # Vector indices
- *     output[i, :, ..., :] = params[indices[i], :, ... :]
- * 
- *     # Higher rank indices
- *     output[i, ..., j, :, ... :] = params[indices[i, ..., j], :, ..., :]
- * }</pre>
- * 
- * 
- * @param <U> data type for {@code output()} output
- */
-@Operator
-public final class ResourceGather<U extends TType> extends RawOp implements Operand<U> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.core.ResourceGather}
-   */
-  public static class Options {
-    
-    /**
-     * @param batchDims 
-     */
-    public Options batchDims(Long batchDims) {
-      this.batchDims = batchDims;
-      return this;
-    }
-    
-    /**
-     * @param validateIndices 
-     */
-    public Options validateIndices(Boolean validateIndices) {
-      this.validateIndices = validateIndices;
-      return this;
-    }
-    
-    private Long batchDims;
-    private Boolean validateIndices;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new ResourceGather operation.
-   * 
-   * @param scope current scope
-   * @param resource 
-   * @param indices 
-   * @param dtype 
-   * @param options carries optional attributes values
-   * @return a new instance of ResourceGather
-   */
-  @Endpoint(describeByClass = true)
-  public static <U extends TType, T extends TNumber> ResourceGather<U> create(Scope scope, Operand<?> resource, Operand<T> indices, Class<U> dtype, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ResourceGather", scope.makeOpName("ResourceGather"));
-    opBuilder.addInput(resource.asOutput(scope));
-    opBuilder.addInput(indices.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("dtype", dtype);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.batchDims != null) {
-          opBuilder.setAttr("batch_dims", opts.batchDims);
-        }
-        if (opts.validateIndices != null) {
-          opBuilder.setAttr("validate_indices", opts.validateIndices);
-        }
-      }
-    }
-    return new ResourceGather<U>(opBuilder.build());
-  }
-  
-  /**
-   * @param batchDims 
-   */
-  public static Options batchDims(Long batchDims) {
-    return new Options().batchDims(batchDims);
-  }
-  
-  /**
-   * @param validateIndices 
-   */
-  public static Options validateIndices(Boolean validateIndices) {
-    return new Options().validateIndices(validateIndices);
-  }
-  
-  /**
-   */
-  public Output<U> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<U> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ResourceGather";
-  
-  private Output<U> output;
-  
-  private ResourceGather(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceGatherNd.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceGatherNd.java
deleted file mode 100644
index 76c29afedd7..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceGatherNd.java
+++ /dev/null
@@ -1,77 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * @param <U> data type for {@code output()} output
- */
-@Operator
-public final class ResourceGatherNd<U extends TType> extends RawOp implements Operand<U> {
-  
-  /**
-   * Factory method to create a class wrapping a new ResourceGatherNd operation.
-   * 
-   * @param scope current scope
-   * @param resource 
-   * @param indices 
-   * @param dtype 
-   * @return a new instance of ResourceGatherNd
-   */
-  @Endpoint(describeByClass = true)
-  public static <U extends TType, T extends TNumber> ResourceGatherNd<U> create(Scope scope, Operand<?> resource, Operand<T> indices, Class<U> dtype) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ResourceGatherNd", scope.makeOpName("ResourceGatherNd"));
-    opBuilder.addInput(resource.asOutput(scope));
-    opBuilder.addInput(indices.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("dtype", dtype);
-    return new ResourceGatherNd<U>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<U> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<U> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ResourceGatherNd";
-  
-  private Output<U> output;
-  
-  private ResourceGatherNd(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterAdd.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterAdd.java
deleted file mode 100644
index 9091ceaf189..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterAdd.java
+++ /dev/null
@@ -1,81 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * Adds sparse updates to the variable referenced by `resource`.
- * <p>
- * This operation computes
- * <p>
- *     # Scalar indices
- *     ref[indices, ...] += updates[...]
- * <p>
- *     # Vector indices (for each i)
- *     ref[indices[i], ...] += updates[i, ...]
- * <p>
- *     # High rank indices (for each i, ..., j)
- *     ref[indices[i, ..., j], ...] += updates[i, ..., j, ...]
- * <p>
- * Duplicate entries are handled correctly: if multiple `indices` reference
- * the same location, their contributions add.
- * <p>
- * Requires `updates.shape = indices.shape + ref.shape[1:]` or `updates.shape = []`.
- * <p>
- * <div style="width:70%; margin:auto; margin-bottom:10px; margin-top:20px;">
- * <img style="width:100%" src='https://www.tensorflow.org/images/ScatterAdd.png' alt>
- * </div>
- */
-@Operator
-public final class ResourceScatterAdd extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new ResourceScatterAdd operation.
-   * 
-   * @param scope current scope
-   * @param resource Should be from a `Variable` node.
-   * @param indices A tensor of indices into the first dimension of `ref`.
-   * @param updates A tensor of updated values to add to `ref`.
-   * @return a new instance of ResourceScatterAdd
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber, U extends TType> ResourceScatterAdd create(Scope scope, Operand<?> resource, Operand<T> indices, Operand<U> updates) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ResourceScatterAdd", scope.makeOpName("ResourceScatterAdd"));
-    opBuilder.addInput(resource.asOutput(scope));
-    opBuilder.addInput(indices.asOutput(scope));
-    opBuilder.addInput(updates.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new ResourceScatterAdd(opBuilder.build());
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ResourceScatterAdd";
-  
-  private ResourceScatterAdd(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterDiv.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterDiv.java
deleted file mode 100644
index 3125cde19b0..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterDiv.java
+++ /dev/null
@@ -1,81 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * Divides sparse updates into the variable referenced by `resource`.
- * <p>
- * This operation computes
- * <p>
- *     # Scalar indices
- *     ref[indices, ...] /= updates[...]
- * <p>
- *     # Vector indices (for each i)
- *     ref[indices[i], ...] /= updates[i, ...]
- * <p>
- *     # High rank indices (for each i, ..., j)
- *     ref[indices[i, ..., j], ...] /= updates[i, ..., j, ...]
- * <p>
- * Duplicate entries are handled correctly: if multiple `indices` reference
- * the same location, their contributions multiply.
- * <p>
- * Requires `updates.shape = indices.shape + ref.shape[1:]` or `updates.shape = []`.
- * <p>
- * <div style="width:70%; margin:auto; margin-bottom:10px; margin-top:20px;">
- * <img style="width:100%" src='https://www.tensorflow.org/images/ScatterAdd.png' alt>
- * </div>
- */
-@Operator
-public final class ResourceScatterDiv extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new ResourceScatterDiv operation.
-   * 
-   * @param scope current scope
-   * @param resource Should be from a `Variable` node.
-   * @param indices A tensor of indices into the first dimension of `ref`.
-   * @param updates A tensor of updated values to add to `ref`.
-   * @return a new instance of ResourceScatterDiv
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber, U extends TType> ResourceScatterDiv create(Scope scope, Operand<?> resource, Operand<T> indices, Operand<U> updates) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ResourceScatterDiv", scope.makeOpName("ResourceScatterDiv"));
-    opBuilder.addInput(resource.asOutput(scope));
-    opBuilder.addInput(indices.asOutput(scope));
-    opBuilder.addInput(updates.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new ResourceScatterDiv(opBuilder.build());
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ResourceScatterDiv";
-  
-  private ResourceScatterDiv(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterMax.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterMax.java
deleted file mode 100644
index f06e1862bfd..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterMax.java
+++ /dev/null
@@ -1,81 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * Reduces sparse updates into the variable referenced by `resource` using the `max` operation.
- * <p>
- * This operation computes
- * <p>
- *     # Scalar indices
- *     ref[indices, ...] = max(ref[indices, ...], updates[...])
- * <p>
- *     # Vector indices (for each i)
- *     ref[indices[i], ...] = max(ref[indices[i], ...], updates[i, ...])
- * <p>
- *     # High rank indices (for each i, ..., j)
- *     ref[indices[i, ..., j], ...] = max(ref[indices[i, ..., j], ...], updates[i, ..., j, ...])
- * <p>
- * Duplicate entries are handled correctly: if multiple `indices` reference
- * the same location, their contributions are combined.
- * <p>
- * Requires `updates.shape = indices.shape + ref.shape[1:]` or `updates.shape = []`.
- * <p>
- * <div style="width:70%; margin:auto; margin-bottom:10px; margin-top:20px;">
- * <img style="width:100%" src='https://www.tensorflow.org/images/ScatterAdd.png' alt>
- * </div>
- */
-@Operator
-public final class ResourceScatterMax extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new ResourceScatterMax operation.
-   * 
-   * @param scope current scope
-   * @param resource Should be from a `Variable` node.
-   * @param indices A tensor of indices into the first dimension of `ref`.
-   * @param updates A tensor of updated values to add to `ref`.
-   * @return a new instance of ResourceScatterMax
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber, U extends TType> ResourceScatterMax create(Scope scope, Operand<?> resource, Operand<T> indices, Operand<U> updates) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ResourceScatterMax", scope.makeOpName("ResourceScatterMax"));
-    opBuilder.addInput(resource.asOutput(scope));
-    opBuilder.addInput(indices.asOutput(scope));
-    opBuilder.addInput(updates.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new ResourceScatterMax(opBuilder.build());
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ResourceScatterMax";
-  
-  private ResourceScatterMax(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterMin.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterMin.java
deleted file mode 100644
index b7f4ff1e51c..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterMin.java
+++ /dev/null
@@ -1,81 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * Reduces sparse updates into the variable referenced by `resource` using the `min` operation.
- * <p>
- * This operation computes
- * <p>
- *     # Scalar indices
- *     ref[indices, ...] = min(ref[indices, ...], updates[...])
- * <p>
- *     # Vector indices (for each i)
- *     ref[indices[i], ...] = min(ref[indices[i], ...], updates[i, ...])
- * <p>
- *     # High rank indices (for each i, ..., j)
- *     ref[indices[i, ..., j], ...] = min(ref[indices[i, ..., j], ...], updates[i, ..., j, ...])
- * <p>
- * Duplicate entries are handled correctly: if multiple `indices` reference
- * the same location, their contributions are combined.
- * <p>
- * Requires `updates.shape = indices.shape + ref.shape[1:]` or `updates.shape = []`.
- * <p>
- * <div style="width:70%; margin:auto; margin-bottom:10px; margin-top:20px;">
- * <img style="width:100%" src='https://www.tensorflow.org/images/ScatterAdd.png' alt>
- * </div>
- */
-@Operator
-public final class ResourceScatterMin extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new ResourceScatterMin operation.
-   * 
-   * @param scope current scope
-   * @param resource Should be from a `Variable` node.
-   * @param indices A tensor of indices into the first dimension of `ref`.
-   * @param updates A tensor of updated values to add to `ref`.
-   * @return a new instance of ResourceScatterMin
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber, U extends TType> ResourceScatterMin create(Scope scope, Operand<?> resource, Operand<T> indices, Operand<U> updates) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ResourceScatterMin", scope.makeOpName("ResourceScatterMin"));
-    opBuilder.addInput(resource.asOutput(scope));
-    opBuilder.addInput(indices.asOutput(scope));
-    opBuilder.addInput(updates.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new ResourceScatterMin(opBuilder.build());
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ResourceScatterMin";
-  
-  private ResourceScatterMin(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterMul.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterMul.java
deleted file mode 100644
index 181b4c372b8..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterMul.java
+++ /dev/null
@@ -1,81 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * Multiplies sparse updates into the variable referenced by `resource`.
- * <p>
- * This operation computes
- * <p>
- *     # Scalar indices
- *     ref[indices, ...] *= updates[...]
- * <p>
- *     # Vector indices (for each i)
- *     ref[indices[i], ...] *= updates[i, ...]
- * <p>
- *     # High rank indices (for each i, ..., j)
- *     ref[indices[i, ..., j], ...] *= updates[i, ..., j, ...]
- * <p>
- * Duplicate entries are handled correctly: if multiple `indices` reference
- * the same location, their contributions multiply.
- * <p>
- * Requires `updates.shape = indices.shape + ref.shape[1:]` or `updates.shape = []`.
- * <p>
- * <div style="width:70%; margin:auto; margin-bottom:10px; margin-top:20px;">
- * <img style="width:100%" src='https://www.tensorflow.org/images/ScatterAdd.png' alt>
- * </div>
- */
-@Operator
-public final class ResourceScatterMul extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new ResourceScatterMul operation.
-   * 
-   * @param scope current scope
-   * @param resource Should be from a `Variable` node.
-   * @param indices A tensor of indices into the first dimension of `ref`.
-   * @param updates A tensor of updated values to add to `ref`.
-   * @return a new instance of ResourceScatterMul
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber, U extends TType> ResourceScatterMul create(Scope scope, Operand<?> resource, Operand<T> indices, Operand<U> updates) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ResourceScatterMul", scope.makeOpName("ResourceScatterMul"));
-    opBuilder.addInput(resource.asOutput(scope));
-    opBuilder.addInput(indices.asOutput(scope));
-    opBuilder.addInput(updates.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new ResourceScatterMul(opBuilder.build());
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ResourceScatterMul";
-  
-  private ResourceScatterMul(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterNdAdd.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterNdAdd.java
deleted file mode 100644
index 83c9ddc52ed..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterNdAdd.java
+++ /dev/null
@@ -1,131 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * Applies sparse addition to individual values or slices in a Variable.
- * <p>
- * `ref` is a `Tensor` with rank `P` and `indices` is a `Tensor` of rank `Q`.
- * <p>
- * `indices` must be integer tensor, containing indices into `ref`.
- * It must be shape `[d_0, ..., d_{Q-2}, K]` where `0 < K <= P`.
- * <p>
- * The innermost dimension of `indices` (with length `K`) corresponds to
- * indices into elements (if `K = P`) or slices (if `K < P`) along the `K`th
- * dimension of `ref`.
- * <p>
- * `updates` is `Tensor` of rank `Q-1+P-K` with shape:
- * <pre>{@code
- * [d_0, ..., d_{Q-2}, ref.shape[K], ..., ref.shape[P-1]]
- * }</pre>
- * For example, say we want to add 4 scattered elements to a rank-1 tensor to
- * 8 elements. In Python, that addition would look like this:
- * <pre>{@code
- * ref = tf.Variable([1, 2, 3, 4, 5, 6, 7, 8], use_resource=True)
- * indices = tf.constant([[4], [3], [1], [7]])
- * updates = tf.constant([9, 10, 11, 12])
- * add = tf.scatter_nd_add(ref, indices, updates)
- * with tf.Session() as sess:
- *   print sess.run(add)
- * }</pre>
- * The resulting update to ref would look like this:
- * <p>
- *     [1, 13, 3, 14, 14, 6, 7, 20]
- * <p>
- * See `tf.scatter_nd` for more details about how to make updates to
- * slices.
- */
-@Operator
-public final class ResourceScatterNdAdd extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.core.ResourceScatterNdAdd}
-   */
-  public static class Options {
-    
-    /**
-     * @param useLocking An optional bool. Defaults to True. If True, the assignment will
-     * be protected by a lock; otherwise the behavior is undefined,
-     * but may exhibit less contention.
-     */
-    public Options useLocking(Boolean useLocking) {
-      this.useLocking = useLocking;
-      return this;
-    }
-    
-    private Boolean useLocking;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new ResourceScatterNdAdd operation.
-   * 
-   * @param scope current scope
-   * @param ref A resource handle. Must be from a VarHandleOp.
-   * @param indices A Tensor. Must be one of the following types: int32, int64.
-   * A tensor of indices into ref.
-   * @param updates A Tensor. Must have the same type as ref. A tensor of
-   * values to add to ref.
-   * @param options carries optional attributes values
-   * @return a new instance of ResourceScatterNdAdd
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber, U extends TType> ResourceScatterNdAdd create(Scope scope, Operand<?> ref, Operand<T> indices, Operand<U> updates, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ResourceScatterNdAdd", scope.makeOpName("ResourceScatterNdAdd"));
-    opBuilder.addInput(ref.asOutput(scope));
-    opBuilder.addInput(indices.asOutput(scope));
-    opBuilder.addInput(updates.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.useLocking != null) {
-          opBuilder.setAttr("use_locking", opts.useLocking);
-        }
-      }
-    }
-    return new ResourceScatterNdAdd(opBuilder.build());
-  }
-  
-  /**
-   * @param useLocking An optional bool. Defaults to True. If True, the assignment will
-   * be protected by a lock; otherwise the behavior is undefined,
-   * but may exhibit less contention.
-   */
-  public static Options useLocking(Boolean useLocking) {
-    return new Options().useLocking(useLocking);
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ResourceScatterNdAdd";
-  
-  private ResourceScatterNdAdd(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterNdMax.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterNdMax.java
deleted file mode 100644
index 40744ec4722..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterNdMax.java
+++ /dev/null
@@ -1,100 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- */
-@Operator
-public final class ResourceScatterNdMax extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.core.ResourceScatterNdMax}
-   */
-  public static class Options {
-    
-    /**
-     * @param useLocking An optional bool. Defaults to True. If True, the assignment will
-     * be protected by a lock; otherwise the behavior is undefined,
-     * but may exhibit less contention.
-     */
-    public Options useLocking(Boolean useLocking) {
-      this.useLocking = useLocking;
-      return this;
-    }
-    
-    private Boolean useLocking;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new ResourceScatterNdMax operation.
-   * 
-   * @param scope current scope
-   * @param ref A resource handle. Must be from a VarHandleOp.
-   * @param indices A Tensor. Must be one of the following types: int32, int64.
-   * A tensor of indices into ref.
-   * @param updates A Tensor. Must have the same type as ref. A tensor of
-   * values whose element wise max is taken with ref
-   * @param options carries optional attributes values
-   * @return a new instance of ResourceScatterNdMax
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber, U extends TType> ResourceScatterNdMax create(Scope scope, Operand<?> ref, Operand<T> indices, Operand<U> updates, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ResourceScatterNdMax", scope.makeOpName("ResourceScatterNdMax"));
-    opBuilder.addInput(ref.asOutput(scope));
-    opBuilder.addInput(indices.asOutput(scope));
-    opBuilder.addInput(updates.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.useLocking != null) {
-          opBuilder.setAttr("use_locking", opts.useLocking);
-        }
-      }
-    }
-    return new ResourceScatterNdMax(opBuilder.build());
-  }
-  
-  /**
-   * @param useLocking An optional bool. Defaults to True. If True, the assignment will
-   * be protected by a lock; otherwise the behavior is undefined,
-   * but may exhibit less contention.
-   */
-  public static Options useLocking(Boolean useLocking) {
-    return new Options().useLocking(useLocking);
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ResourceScatterNdMax";
-  
-  private ResourceScatterNdMax(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterNdMin.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterNdMin.java
deleted file mode 100644
index e0645fca84e..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterNdMin.java
+++ /dev/null
@@ -1,100 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- */
-@Operator
-public final class ResourceScatterNdMin extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.core.ResourceScatterNdMin}
-   */
-  public static class Options {
-    
-    /**
-     * @param useLocking An optional bool. Defaults to True. If True, the assignment will
-     * be protected by a lock; otherwise the behavior is undefined,
-     * but may exhibit less contention.
-     */
-    public Options useLocking(Boolean useLocking) {
-      this.useLocking = useLocking;
-      return this;
-    }
-    
-    private Boolean useLocking;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new ResourceScatterNdMin operation.
-   * 
-   * @param scope current scope
-   * @param ref A resource handle. Must be from a VarHandleOp.
-   * @param indices A Tensor. Must be one of the following types: int32, int64.
-   * A tensor of indices into ref.
-   * @param updates A Tensor. Must have the same type as ref. A tensor of
-   * values whose element wise min is taken with ref.
-   * @param options carries optional attributes values
-   * @return a new instance of ResourceScatterNdMin
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber, U extends TType> ResourceScatterNdMin create(Scope scope, Operand<?> ref, Operand<T> indices, Operand<U> updates, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ResourceScatterNdMin", scope.makeOpName("ResourceScatterNdMin"));
-    opBuilder.addInput(ref.asOutput(scope));
-    opBuilder.addInput(indices.asOutput(scope));
-    opBuilder.addInput(updates.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.useLocking != null) {
-          opBuilder.setAttr("use_locking", opts.useLocking);
-        }
-      }
-    }
-    return new ResourceScatterNdMin(opBuilder.build());
-  }
-  
-  /**
-   * @param useLocking An optional bool. Defaults to True. If True, the assignment will
-   * be protected by a lock; otherwise the behavior is undefined,
-   * but may exhibit less contention.
-   */
-  public static Options useLocking(Boolean useLocking) {
-    return new Options().useLocking(useLocking);
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ResourceScatterNdMin";
-  
-  private ResourceScatterNdMin(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterNdSub.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterNdSub.java
deleted file mode 100644
index 4bf0dd018ce..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterNdSub.java
+++ /dev/null
@@ -1,131 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * Applies sparse subtraction to individual values or slices in a Variable.
- * <p>
- * `ref` is a `Tensor` with rank `P` and `indices` is a `Tensor` of rank `Q`.
- * <p>
- * `indices` must be integer tensor, containing indices into `ref`.
- * It must be shape `[d_0, ..., d_{Q-2}, K]` where `0 < K <= P`.
- * <p>
- * The innermost dimension of `indices` (with length `K`) corresponds to
- * indices into elements (if `K = P`) or slices (if `K < P`) along the `K`th
- * dimension of `ref`.
- * <p>
- * `updates` is `Tensor` of rank `Q-1+P-K` with shape:
- * <pre>{@code
- * [d_0, ..., d_{Q-2}, ref.shape[K], ..., ref.shape[P-1]]
- * }</pre>
- * For example, say we want to subtract 4 scattered elements from a rank-1 tensor
- * with 8 elements. In Python, that subtraction would look like this:
- * <pre>{@code
- * ref = tf.Variable([1, 2, 3, 4, 5, 6, 7, 8], use_resource=True)
- * indices = tf.constant([[4], [3], [1], [7]])
- * updates = tf.constant([9, 10, 11, 12])
- * sub = tf.scatter_nd_sub(ref, indices, updates)
- * with tf.Session() as sess:
- *   print sess.run(sub)
- * }</pre>
- * The resulting update to ref would look like this:
- * <p>
- *     [1, -9, 3, -6, -4, 6, 7, -4]
- * <p>
- * See `tf.scatter_nd` for more details about how to make updates to
- * slices.
- */
-@Operator
-public final class ResourceScatterNdSub extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.core.ResourceScatterNdSub}
-   */
-  public static class Options {
-    
-    /**
-     * @param useLocking An optional bool. Defaults to True. If True, the assignment will
-     * be protected by a lock; otherwise the behavior is undefined,
-     * but may exhibit less contention.
-     */
-    public Options useLocking(Boolean useLocking) {
-      this.useLocking = useLocking;
-      return this;
-    }
-    
-    private Boolean useLocking;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new ResourceScatterNdSub operation.
-   * 
-   * @param scope current scope
-   * @param ref A resource handle. Must be from a VarHandleOp.
-   * @param indices A Tensor. Must be one of the following types: int32, int64.
-   * A tensor of indices into ref.
-   * @param updates A Tensor. Must have the same type as ref. A tensor of
-   * values to add to ref.
-   * @param options carries optional attributes values
-   * @return a new instance of ResourceScatterNdSub
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber, U extends TType> ResourceScatterNdSub create(Scope scope, Operand<?> ref, Operand<T> indices, Operand<U> updates, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ResourceScatterNdSub", scope.makeOpName("ResourceScatterNdSub"));
-    opBuilder.addInput(ref.asOutput(scope));
-    opBuilder.addInput(indices.asOutput(scope));
-    opBuilder.addInput(updates.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.useLocking != null) {
-          opBuilder.setAttr("use_locking", opts.useLocking);
-        }
-      }
-    }
-    return new ResourceScatterNdSub(opBuilder.build());
-  }
-  
-  /**
-   * @param useLocking An optional bool. Defaults to True. If True, the assignment will
-   * be protected by a lock; otherwise the behavior is undefined,
-   * but may exhibit less contention.
-   */
-  public static Options useLocking(Boolean useLocking) {
-    return new Options().useLocking(useLocking);
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ResourceScatterNdSub";
-  
-  private ResourceScatterNdSub(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterNdUpdate.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterNdUpdate.java
deleted file mode 100644
index aec3c3dd214..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterNdUpdate.java
+++ /dev/null
@@ -1,133 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * Applies sparse `updates` to individual values or slices within a given
- * <p>
- * variable according to `indices`.
- * <p>
- * `ref` is a `Tensor` with rank `P` and `indices` is a `Tensor` of rank `Q`.
- * <p>
- * `indices` must be integer tensor, containing indices into `ref`.
- * It must be shape `[d_0, ..., d_{Q-2}, K]` where `0 < K <= P`.
- * <p>
- * The innermost dimension of `indices` (with length `K`) corresponds to
- * indices into elements (if `K = P`) or slices (if `K < P`) along the `K`th
- * dimension of `ref`.
- * <p>
- * `updates` is `Tensor` of rank `Q-1+P-K` with shape:
- * <pre>{@code
- * [d_0, ..., d_{Q-2}, ref.shape[K], ..., ref.shape[P-1]].
- * }</pre>
- * For example, say we want to update 4 scattered elements to a rank-1 tensor to
- * 8 elements. In Python, that update would look like this:
- * <pre>{@code
- *     ref = tf.Variable([1, 2, 3, 4, 5, 6, 7, 8])
- *     indices = tf.constant([[4], [3], [1] ,[7]])
- *     updates = tf.constant([9, 10, 11, 12])
- *     update = tf.scatter_nd_update(ref, indices, updates)
- *     with tf.Session() as sess:
- *       print sess.run(update)
- * }</pre>
- * The resulting update to ref would look like this:
- * <p>
- *     [1, 11, 3, 10, 9, 6, 7, 12]
- * <p>
- * See `tf.scatter_nd` for more details about how to make updates to
- * slices.
- */
-@Operator
-public final class ResourceScatterNdUpdate extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.core.ResourceScatterNdUpdate}
-   */
-  public static class Options {
-    
-    /**
-     * @param useLocking An optional bool. Defaults to True. If True, the assignment will
-     * be protected by a lock; otherwise the behavior is undefined,
-     * but may exhibit less contention.
-     */
-    public Options useLocking(Boolean useLocking) {
-      this.useLocking = useLocking;
-      return this;
-    }
-    
-    private Boolean useLocking;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new ResourceScatterNdUpdate operation.
-   * 
-   * @param scope current scope
-   * @param ref A resource handle. Must be from a VarHandleOp.
-   * @param indices A Tensor. Must be one of the following types: int32, int64.
-   * A tensor of indices into ref.
-   * @param updates A Tensor. Must have the same type as ref. A tensor of updated
-   * values to add to ref.
-   * @param options carries optional attributes values
-   * @return a new instance of ResourceScatterNdUpdate
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber, U extends TType> ResourceScatterNdUpdate create(Scope scope, Operand<?> ref, Operand<T> indices, Operand<U> updates, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ResourceScatterNdUpdate", scope.makeOpName("ResourceScatterNdUpdate"));
-    opBuilder.addInput(ref.asOutput(scope));
-    opBuilder.addInput(indices.asOutput(scope));
-    opBuilder.addInput(updates.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.useLocking != null) {
-          opBuilder.setAttr("use_locking", opts.useLocking);
-        }
-      }
-    }
-    return new ResourceScatterNdUpdate(opBuilder.build());
-  }
-  
-  /**
-   * @param useLocking An optional bool. Defaults to True. If True, the assignment will
-   * be protected by a lock; otherwise the behavior is undefined,
-   * but may exhibit less contention.
-   */
-  public static Options useLocking(Boolean useLocking) {
-    return new Options().useLocking(useLocking);
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ResourceScatterNdUpdate";
-  
-  private ResourceScatterNdUpdate(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterSub.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterSub.java
deleted file mode 100644
index 306cda55863..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterSub.java
+++ /dev/null
@@ -1,81 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * Subtracts sparse updates from the variable referenced by `resource`.
- * <p>
- * This operation computes
- * <p>
- *     # Scalar indices
- *     ref[indices, ...] -= updates[...]
- * <p>
- *     # Vector indices (for each i)
- *     ref[indices[i], ...] -= updates[i, ...]
- * <p>
- *     # High rank indices (for each i, ..., j)
- *     ref[indices[i, ..., j], ...] -= updates[i, ..., j, ...]
- * <p>
- * Duplicate entries are handled correctly: if multiple `indices` reference
- * the same location, their contributions add.
- * <p>
- * Requires `updates.shape = indices.shape + ref.shape[1:]` or `updates.shape = []`.
- * <p>
- * <div style="width:70%; margin:auto; margin-bottom:10px; margin-top:20px;">
- * <img style="width:100%" src='https://www.tensorflow.org/images/ScatterAdd.png' alt>
- * </div>
- */
-@Operator
-public final class ResourceScatterSub extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new ResourceScatterSub operation.
-   * 
-   * @param scope current scope
-   * @param resource Should be from a `Variable` node.
-   * @param indices A tensor of indices into the first dimension of `ref`.
-   * @param updates A tensor of updated values to add to `ref`.
-   * @return a new instance of ResourceScatterSub
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber, U extends TType> ResourceScatterSub create(Scope scope, Operand<?> resource, Operand<T> indices, Operand<U> updates) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ResourceScatterSub", scope.makeOpName("ResourceScatterSub"));
-    opBuilder.addInput(resource.asOutput(scope));
-    opBuilder.addInput(indices.asOutput(scope));
-    opBuilder.addInput(updates.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new ResourceScatterSub(opBuilder.build());
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ResourceScatterSub";
-  
-  private ResourceScatterSub(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterUpdate.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterUpdate.java
deleted file mode 100644
index ad18af339c1..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterUpdate.java
+++ /dev/null
@@ -1,72 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * Assigns sparse updates to the variable referenced by `resource`.
- * <p>
- * This operation computes
- * <p>
- *     # Scalar indices
- *     ref[indices, ...] = updates[...]
- * <p>
- *     # Vector indices (for each i)
- *     ref[indices[i], ...] = updates[i, ...]
- * <p>
- *     # High rank indices (for each i, ..., j)
- *     ref[indices[i, ..., j], ...] = updates[i, ..., j, ...]
- */
-@Operator
-public final class ResourceScatterUpdate extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new ResourceScatterUpdate operation.
-   * 
-   * @param scope current scope
-   * @param resource Should be from a `Variable` node.
-   * @param indices A tensor of indices into the first dimension of `ref`.
-   * @param updates A tensor of updated values to add to `ref`.
-   * @return a new instance of ResourceScatterUpdate
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber, U extends TType> ResourceScatterUpdate create(Scope scope, Operand<?> resource, Operand<T> indices, Operand<U> updates) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ResourceScatterUpdate", scope.makeOpName("ResourceScatterUpdate"));
-    opBuilder.addInput(resource.asOutput(scope));
-    opBuilder.addInput(indices.asOutput(scope));
-    opBuilder.addInput(updates.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new ResourceScatterUpdate(opBuilder.build());
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ResourceScatterUpdate";
-  
-  private ResourceScatterUpdate(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceStridedSliceAssign.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceStridedSliceAssign.java
deleted file mode 100644
index ce676c8c2b6..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceStridedSliceAssign.java
+++ /dev/null
@@ -1,182 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * Assign `value` to the sliced l-value reference of `ref`.
- * <p>
- * The values of `value` are assigned to the positions in the variable
- * `ref` that are selected by the slice parameters. The slice parameters
- * `begin, `end`, `strides`, etc. work exactly as in `StridedSlice`.
- * <p>
- * NOTE this op currently does not support broadcasting and so `value`'s
- * shape must be exactly the shape produced by the slice of `ref`.
- */
-@Operator
-public final class ResourceStridedSliceAssign extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.core.ResourceStridedSliceAssign}
-   */
-  public static class Options {
-    
-    /**
-     * @param beginMask 
-     */
-    public Options beginMask(Long beginMask) {
-      this.beginMask = beginMask;
-      return this;
-    }
-    
-    /**
-     * @param endMask 
-     */
-    public Options endMask(Long endMask) {
-      this.endMask = endMask;
-      return this;
-    }
-    
-    /**
-     * @param ellipsisMask 
-     */
-    public Options ellipsisMask(Long ellipsisMask) {
-      this.ellipsisMask = ellipsisMask;
-      return this;
-    }
-    
-    /**
-     * @param newAxisMask 
-     */
-    public Options newAxisMask(Long newAxisMask) {
-      this.newAxisMask = newAxisMask;
-      return this;
-    }
-    
-    /**
-     * @param shrinkAxisMask 
-     */
-    public Options shrinkAxisMask(Long shrinkAxisMask) {
-      this.shrinkAxisMask = shrinkAxisMask;
-      return this;
-    }
-    
-    private Long beginMask;
-    private Long endMask;
-    private Long ellipsisMask;
-    private Long newAxisMask;
-    private Long shrinkAxisMask;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new ResourceStridedSliceAssign operation.
-   * 
-   * @param scope current scope
-   * @param ref 
-   * @param begin 
-   * @param end 
-   * @param strides 
-   * @param value 
-   * @param options carries optional attributes values
-   * @return a new instance of ResourceStridedSliceAssign
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber, U extends TType> ResourceStridedSliceAssign create(Scope scope, Operand<?> ref, Operand<T> begin, Operand<T> end, Operand<T> strides, Operand<U> value, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ResourceStridedSliceAssign", scope.makeOpName("ResourceStridedSliceAssign"));
-    opBuilder.addInput(ref.asOutput(scope));
-    opBuilder.addInput(begin.asOutput(scope));
-    opBuilder.addInput(end.asOutput(scope));
-    opBuilder.addInput(strides.asOutput(scope));
-    opBuilder.addInput(value.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.beginMask != null) {
-          opBuilder.setAttr("begin_mask", opts.beginMask);
-        }
-        if (opts.endMask != null) {
-          opBuilder.setAttr("end_mask", opts.endMask);
-        }
-        if (opts.ellipsisMask != null) {
-          opBuilder.setAttr("ellipsis_mask", opts.ellipsisMask);
-        }
-        if (opts.newAxisMask != null) {
-          opBuilder.setAttr("new_axis_mask", opts.newAxisMask);
-        }
-        if (opts.shrinkAxisMask != null) {
-          opBuilder.setAttr("shrink_axis_mask", opts.shrinkAxisMask);
-        }
-      }
-    }
-    return new ResourceStridedSliceAssign(opBuilder.build());
-  }
-  
-  /**
-   * @param beginMask 
-   */
-  public static Options beginMask(Long beginMask) {
-    return new Options().beginMask(beginMask);
-  }
-  
-  /**
-   * @param endMask 
-   */
-  public static Options endMask(Long endMask) {
-    return new Options().endMask(endMask);
-  }
-  
-  /**
-   * @param ellipsisMask 
-   */
-  public static Options ellipsisMask(Long ellipsisMask) {
-    return new Options().ellipsisMask(ellipsisMask);
-  }
-  
-  /**
-   * @param newAxisMask 
-   */
-  public static Options newAxisMask(Long newAxisMask) {
-    return new Options().newAxisMask(newAxisMask);
-  }
-  
-  /**
-   * @param shrinkAxisMask 
-   */
-  public static Options shrinkAxisMask(Long shrinkAxisMask) {
-    return new Options().shrinkAxisMask(shrinkAxisMask);
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ResourceStridedSliceAssign";
-  
-  private ResourceStridedSliceAssign(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Reverse.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Reverse.java
deleted file mode 100644
index 3db73f986f2..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Reverse.java
+++ /dev/null
@@ -1,126 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * Reverses specific dimensions of a tensor.
- * <p>
- * NOTE `tf.reverse` has now changed behavior in preparation for 1.0.
- * `tf.reverse_v2` is currently an alias that will be deprecated before TF 1.0.
- * <p>
- * Given a `tensor`, and a `int32` tensor `axis` representing the set of
- * dimensions of `tensor` to reverse. This operation reverses each dimension
- * `i` for which there exists `j` s.t. `axis[j] == i`.
- * <p>
- * `tensor` can have up to 8 dimensions. The number of dimensions specified
- * in `axis` may be 0 or more entries. If an index is specified more than
- * once, a InvalidArgument error is raised.
- * <p>
- * For example:
- * <pre>{@code
- * # tensor 't' is [[[[ 0,  1,  2,  3],
- * #                  [ 4,  5,  6,  7],
- * #                  [ 8,  9, 10, 11]],
- * #                 [[12, 13, 14, 15],
- * #                  [16, 17, 18, 19],
- * #                  [20, 21, 22, 23]]]]
- * # tensor 't' shape is [1, 2, 3, 4]
- * 
- * # 'dims' is [3] or 'dims' is [-1]
- * reverse(t, dims) ==> [[[[ 3,  2,  1,  0],
- *                         [ 7,  6,  5,  4],
- *                         [ 11, 10, 9, 8]],
- *                        [[15, 14, 13, 12],
- *                         [19, 18, 17, 16],
- *                         [23, 22, 21, 20]]]]
- * 
- * # 'dims' is '[1]' (or 'dims' is '[-3]')
- * reverse(t, dims) ==> [[[[12, 13, 14, 15],
- *                         [16, 17, 18, 19],
- *                         [20, 21, 22, 23]
- *                        [[ 0,  1,  2,  3],
- *                         [ 4,  5,  6,  7],
- *                         [ 8,  9, 10, 11]]]]
- * 
- * # 'dims' is '[2]' (or 'dims' is '[-2]')
- * reverse(t, dims) ==> [[[[8, 9, 10, 11],
- *                         [4, 5, 6, 7],
- *                         [0, 1, 2, 3]]
- *                        [[20, 21, 22, 23],
- *                         [16, 17, 18, 19],
- *                         [12, 13, 14, 15]]]]
- * }</pre>
- * 
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator
-public final class Reverse<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new Reverse operation.
-   * 
-   * @param scope current scope
-   * @param tensor Up to 8-D.
-   * @param axis 1-D. The indices of the dimensions to reverse. Must be in the range
-   * `[-rank(tensor), rank(tensor))`.
-   * @return a new instance of Reverse
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> Reverse<T> create(Scope scope, Operand<T> tensor, Operand<U> axis) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ReverseV2", scope.makeOpName("Reverse"));
-    opBuilder.addInput(tensor.asOutput(scope));
-    opBuilder.addInput(axis.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new Reverse<T>(opBuilder.build());
-  }
-  
-  /**
-   * The same shape as `tensor`.
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ReverseV2";
-  
-  private Output<T> output;
-  
-  private Reverse(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ReverseSequence.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ReverseSequence.java
deleted file mode 100644
index aed27ec03df..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ReverseSequence.java
+++ /dev/null
@@ -1,168 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * Reverses variable length slices.
- * <p>
- * This op first slices `input` along the dimension `batch_dim`, and for each
- * slice `i`, reverses the first `seq_lengths[i]` elements along
- * the dimension `seq_dim`.
- * <p>
- * The elements of `seq_lengths` must obey `seq_lengths[i] <= input.dims[seq_dim]`,
- * and `seq_lengths` must be a vector of length `input.dims[batch_dim]`.
- * <p>
- * The output slice `i` along dimension `batch_dim` is then given by input
- * slice `i`, with the first `seq_lengths[i]` slices along dimension
- * `seq_dim` reversed.
- * <p>
- * For example:
- * <pre>{@code
- * # Given this:
- * batch_dim = 0
- * seq_dim = 1
- * input.dims = (4, 8, ...)
- * seq_lengths = [7, 2, 3, 5]
- * 
- * # then slices of input are reversed on seq_dim, but only up to seq_lengths:
- * output[0, 0:7, :, ...] = input[0, 7:0:-1, :, ...]
- * output[1, 0:2, :, ...] = input[1, 2:0:-1, :, ...]
- * output[2, 0:3, :, ...] = input[2, 3:0:-1, :, ...]
- * output[3, 0:5, :, ...] = input[3, 5:0:-1, :, ...]
- * 
- * # while entries past seq_lens are copied through:
- * output[0, 7:, :, ...] = input[0, 7:, :, ...]
- * output[1, 2:, :, ...] = input[1, 2:, :, ...]
- * output[2, 3:, :, ...] = input[2, 3:, :, ...]
- * output[3, 2:, :, ...] = input[3, 2:, :, ...]
- * }</pre>
- * In contrast, if:
- * <pre>{@code
- * # Given this:
- * batch_dim = 2
- * seq_dim = 0
- * input.dims = (8, ?, 4, ...)
- * seq_lengths = [7, 2, 3, 5]
- * 
- * # then slices of input are reversed on seq_dim, but only up to seq_lengths:
- * output[0:7, :, 0, :, ...] = input[7:0:-1, :, 0, :, ...]
- * output[0:2, :, 1, :, ...] = input[2:0:-1, :, 1, :, ...]
- * output[0:3, :, 2, :, ...] = input[3:0:-1, :, 2, :, ...]
- * output[0:5, :, 3, :, ...] = input[5:0:-1, :, 3, :, ...]
- * 
- * # while entries past seq_lens are copied through:
- * output[7:, :, 0, :, ...] = input[7:, :, 0, :, ...]
- * output[2:, :, 1, :, ...] = input[2:, :, 1, :, ...]
- * output[3:, :, 2, :, ...] = input[3:, :, 2, :, ...]
- * output[2:, :, 3, :, ...] = input[2:, :, 3, :, ...]
- * }</pre>
- * 
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator
-public final class ReverseSequence<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.core.ReverseSequence}
-   */
-  public static class Options {
-    
-    /**
-     * @param batchDim The dimension along which reversal is performed.
-     */
-    public Options batchDim(Long batchDim) {
-      this.batchDim = batchDim;
-      return this;
-    }
-    
-    private Long batchDim;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new ReverseSequence operation.
-   * 
-   * @param scope current scope
-   * @param input The input to reverse.
-   * @param seqLengths 1-D with length `input.dims(batch_dim)` and
-   * `max(seq_lengths) <= input.dims(seq_dim)`
-   * @param seqDim The dimension which is partially reversed.
-   * @param options carries optional attributes values
-   * @return a new instance of ReverseSequence
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> ReverseSequence<T> create(Scope scope, Operand<T> input, Operand<U> seqLengths, Long seqDim, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ReverseSequence", scope.makeOpName("ReverseSequence"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder.addInput(seqLengths.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("seq_dim", seqDim);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.batchDim != null) {
-          opBuilder.setAttr("batch_dim", opts.batchDim);
-        }
-      }
-    }
-    return new ReverseSequence<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param batchDim The dimension along which reversal is performed.
-   */
-  public static Options batchDim(Long batchDim) {
-    return new Options().batchDim(batchDim);
-  }
-  
-  /**
-   * The partially reversed input. It has the same shape as `input`.
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ReverseSequence";
-  
-  private Output<T> output;
-  
-  private ReverseSequence(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Roll.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Roll.java
deleted file mode 100644
index e3baaa91241..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Roll.java
+++ /dev/null
@@ -1,109 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * Rolls the elements of a tensor along an axis.
- * <p>
- * The elements are shifted positively (towards larger indices) by the offset of
- * `shift` along the dimension of `axis`. Negative `shift` values will shift
- * elements in the opposite direction. Elements that roll passed the last position
- * will wrap around to the first and vice versa. Multiple shifts along multiple
- * axes may be specified.
- * <p>
- * For example:
- * <pre>{@code
- * # 't' is [0, 1, 2, 3, 4]
- * roll(t, shift=2, axis=0) ==> [3, 4, 0, 1, 2]
- * 
- * # shifting along multiple dimensions
- * # 't' is [[0, 1, 2, 3, 4], [5, 6, 7, 8, 9]]
- * roll(t, shift=[1, -2], axis=[0, 1]) ==> [[7, 8, 9, 5, 6], [2, 3, 4, 0, 1]]
- * 
- * # shifting along the same axis multiple times
- * # 't' is [[0, 1, 2, 3, 4], [5, 6, 7, 8, 9]]
- * roll(t, shift=[2, -3], axis=[1, 1]) ==> [[1, 2, 3, 4, 0], [6, 7, 8, 9, 5]]
- * }</pre>
- * 
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator
-public final class Roll<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new Roll operation.
-   * 
-   * @param scope current scope
-   * @param input 
-   * @param shift Dimension must be 0-D or 1-D. `shift[i]` specifies the number of places by which
-   * elements are shifted positively (towards larger indices) along the dimension
-   * specified by `axis[i]`. Negative shifts will roll the elements in the opposite
-   * direction.
-   * @param axis Dimension must be 0-D or 1-D. `axis[i]` specifies the dimension that the shift
-   * `shift[i]` should occur. If the same axis is referenced more than once, the
-   * total shift for that axis will be the sum of all the shifts that belong to that
-   * axis.
-   * @return a new instance of Roll
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber, V extends TNumber> Roll<T> create(Scope scope, Operand<T> input, Operand<U> shift, Operand<V> axis) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Roll", scope.makeOpName("Roll"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder.addInput(shift.asOutput(scope));
-    opBuilder.addInput(axis.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new Roll<T>(opBuilder.build());
-  }
-  
-  /**
-   * Has the same shape and size as the input. The elements are shifted
-   * positively (towards larger indices) by the offsets of `shift` along the
-   * dimensions of `axis`.
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Roll";
-  
-  private Output<T> output;
-  
-  private Roll(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Rpc.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Rpc.java
deleted file mode 100644
index ae82347d5d4..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Rpc.java
+++ /dev/null
@@ -1,209 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TString;
-
-/**
- * Perform batches of RPC requests.
- * <p>
- * This op asynchronously performs either a single RPC request, or a batch
- * of requests.  RPC requests are defined by three main parameters:
- * <p>
- *   - `address` (the host+port or BNS address of the request)
- *   - `method` (the RPC method name for the request)
- *   - `request` (the serialized proto string, or vector of strings,
- *      of the RPC request argument).
- * <p>
- * For example, if you have an RPC service running on port localhost:2345,
- * and its interface is configured with the following proto declaration:
- * <pre>{@code
- * service MyService {
- *   rpc MyMethod(MyRequestProto) returns (MyResponseProto) {
- *   }
- * };
- * }</pre>
- * then call this op with arguments:
- * <pre>{@code
- * address = "localhost:2345"
- * method = "MyService/MyMethod"
- * }</pre>
- * The `request` tensor is a string tensor representing serialized `MyRequestProto`
- * strings; and the output string tensor `response` will have the same shape
- * and contain (upon successful completion) corresponding serialized
- * `MyResponseProto` strings.
- * <p>
- * For example, to send a single, empty, `MyRequestProto`, call
- * this op with `request = ""`.  To send 5 <b>parallel</b> empty requests,
- * call this op with `request = ["", "", "", "", ""]`.
- * <p>
- * More generally, one can create a batch of `MyRequestProto` serialized protos
- * from regular batched tensors using the `encode_proto` op, and convert
- * the response `MyResponseProto` serialized protos to batched tensors
- * using the `decode_proto` op.
- * <p>
- * <b>NOTE</b> Working with serialized proto strings is faster than instantiating
- * actual proto objects in memory, so no performance degradation is expected
- * compared to writing custom kernels for this workflow.
- * <p>
- * If the connection fails or the remote worker returns an error
- * status, the op reraises this exception locally.
- * <p>
- * See the `TryRpc` op if you prefer to handle RPC failures manually in the graph.
- */
-@Operator
-public final class Rpc extends RawOp implements Operand<TString> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.core.Rpc}
-   */
-  public static class Options {
-    
-    /**
-     * @param protocol RPC protocol to use.  Empty string means use the default protocol.
-     * Options include 'grpc'.
-     */
-    public Options protocol(String protocol) {
-      this.protocol = protocol;
-      return this;
-    }
-    
-    /**
-     * @param failFast `boolean`. If `true` (default), then failures to connect
-     * (i.e., the server does not immediately respond) cause an RPC failure.
-     */
-    public Options failFast(Boolean failFast) {
-      this.failFast = failFast;
-      return this;
-    }
-    
-    /**
-     * @param timeoutInMs `int`. If `0` (default), then the kernel will run the RPC
-     * request and only time out if the RPC deadline passes or the session times out.
-     * If this value is greater than `0`, then the op will raise an exception if
-     * the RPC takes longer than `timeout_in_ms`.
-     */
-    public Options timeoutInMs(Long timeoutInMs) {
-      this.timeoutInMs = timeoutInMs;
-      return this;
-    }
-    
-    private String protocol;
-    private Boolean failFast;
-    private Long timeoutInMs;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new Rpc operation.
-   * 
-   * @param scope current scope
-   * @param address `0-D` or `1-D`.  The address (i.e. host_name:port) of the RPC server.
-   * If this tensor has more than 1 element, then multiple parallel rpc requests
-   * are sent.  This argument broadcasts with `method` and `request`.
-   * @param method `0-D` or `1-D`.  The method address on the RPC server.
-   * If this tensor has more than 1 element, then multiple parallel rpc requests
-   * are sent.  This argument broadcasts with `address` and `request`.
-   * @param request `0-D` or `1-D`.  Serialized proto strings: the rpc request argument.
-   * If this tensor has more than 1 element, then multiple parallel rpc requests
-   * are sent.  This argument broadcasts with `address` and `method`.
-   * @param options carries optional attributes values
-   * @return a new instance of Rpc
-   */
-  @Endpoint(describeByClass = true)
-  public static Rpc create(Scope scope, Operand<TString> address, Operand<TString> method, Operand<TString> request, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Rpc", scope.makeOpName("Rpc"));
-    opBuilder.addInput(address.asOutput(scope));
-    opBuilder.addInput(method.asOutput(scope));
-    opBuilder.addInput(request.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.protocol != null) {
-          opBuilder.setAttr("protocol", opts.protocol);
-        }
-        if (opts.failFast != null) {
-          opBuilder.setAttr("fail_fast", opts.failFast);
-        }
-        if (opts.timeoutInMs != null) {
-          opBuilder.setAttr("timeout_in_ms", opts.timeoutInMs);
-        }
-      }
-    }
-    return new Rpc(opBuilder.build());
-  }
-  
-  /**
-   * @param protocol RPC protocol to use.  Empty string means use the default protocol.
-   * Options include 'grpc'.
-   */
-  public static Options protocol(String protocol) {
-    return new Options().protocol(protocol);
-  }
-  
-  /**
-   * @param failFast `boolean`. If `true` (default), then failures to connect
-   * (i.e., the server does not immediately respond) cause an RPC failure.
-   */
-  public static Options failFast(Boolean failFast) {
-    return new Options().failFast(failFast);
-  }
-  
-  /**
-   * @param timeoutInMs `int`. If `0` (default), then the kernel will run the RPC
-   * request and only time out if the RPC deadline passes or the session times out.
-   * If this value is greater than `0`, then the op will raise an exception if
-   * the RPC takes longer than `timeout_in_ms`.
-   */
-  public static Options timeoutInMs(Long timeoutInMs) {
-    return new Options().timeoutInMs(timeoutInMs);
-  }
-  
-  /**
-   * Same shape as `request`. Serialized proto strings: the rpc responses.
-   */
-  public Output<TString> response() {
-    return response;
-  }
-  
-  @Override
-  public Output<TString> asOutput(Scope scope) {
-    return response;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Rpc";
-  
-  private Output<TString> response;
-  
-  private Rpc(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    response = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterAdd.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterAdd.java
deleted file mode 100644
index 88e107f365c..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterAdd.java
+++ /dev/null
@@ -1,140 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * Adds sparse updates to a variable reference.
- * <p>
- * This operation computes
- * <p>
- *     # Scalar indices
- *     ref[indices, ...] += updates[...]
- * <p>
- *     # Vector indices (for each i)
- *     ref[indices[i], ...] += updates[i, ...]
- * <p>
- *     # High rank indices (for each i, ..., j)
- *     ref[indices[i, ..., j], ...] += updates[i, ..., j, ...]
- * <p>
- * This operation outputs `ref` after the update is done.
- * This makes it easier to chain operations that need to use the reset value.
- * <p>
- * Duplicate entries are handled correctly: if multiple `indices` reference
- * the same location, their contributions add.
- * <p>
- * Requires `updates.shape = indices.shape + ref.shape[1:]` or `updates.shape = []`.
- * <p>
- * <div style="width:70%; margin:auto; margin-bottom:10px; margin-top:20px;">
- * <img style="width:100%" src="https://www.tensorflow.org/images/ScatterAdd.png" alt>
- * </div>
- * 
- * @param <T> data type for {@code outputRef()} output
- */
-@Operator
-public final class ScatterAdd<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.core.ScatterAdd}
-   */
-  public static class Options {
-    
-    /**
-     * @param useLocking If True, the addition will be protected by a lock;
-     * otherwise the behavior is undefined, but may exhibit less contention.
-     */
-    public Options useLocking(Boolean useLocking) {
-      this.useLocking = useLocking;
-      return this;
-    }
-    
-    private Boolean useLocking;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new ScatterAdd operation.
-   * 
-   * @param scope current scope
-   * @param ref Should be from a `Variable` node.
-   * @param indices A tensor of indices into the first dimension of `ref`.
-   * @param updates A tensor of updated values to add to `ref`.
-   * @param options carries optional attributes values
-   * @return a new instance of ScatterAdd
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> ScatterAdd<T> create(Scope scope, Operand<T> ref, Operand<U> indices, Operand<T> updates, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ScatterAdd", scope.makeOpName("ScatterAdd"));
-    opBuilder.addInput(ref.asOutput(scope));
-    opBuilder.addInput(indices.asOutput(scope));
-    opBuilder.addInput(updates.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.useLocking != null) {
-          opBuilder.setAttr("use_locking", opts.useLocking);
-        }
-      }
-    }
-    return new ScatterAdd<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param useLocking If True, the addition will be protected by a lock;
-   * otherwise the behavior is undefined, but may exhibit less contention.
-   */
-  public static Options useLocking(Boolean useLocking) {
-    return new Options().useLocking(useLocking);
-  }
-  
-  /**
-   * = Same as `ref`.  Returned as a convenience for operations that want
-   * to use the updated values after the update is done.
-   */
-  public Output<T> outputRef() {
-    return outputRef;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return outputRef;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ScatterAdd";
-  
-  private Output<T> outputRef;
-  
-  private ScatterAdd(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    outputRef = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterDiv.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterDiv.java
deleted file mode 100644
index c9c1e2e0772..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterDiv.java
+++ /dev/null
@@ -1,136 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * Divides a variable reference by sparse updates.
- * <p>
- * This operation computes
- * <pre>{@code
- *     # Scalar indices
- *     ref[indices, ...] /= updates[...]
- * 
- *     # Vector indices (for each i)
- *     ref[indices[i], ...] /= updates[i, ...]
- * 
- *     # High rank indices (for each i, ..., j)
- *     ref[indices[i, ..., j], ...] /= updates[i, ..., j, ...]
- * }</pre>
- * This operation outputs `ref` after the update is done.
- * This makes it easier to chain operations that need to use the reset value.
- * <p>
- * Duplicate entries are handled correctly: if multiple `indices` reference
- * the same location, their contributions divide.
- * <p>
- * Requires `updates.shape = indices.shape + ref.shape[1:]` or `updates.shape = []`.
- * 
- * @param <T> data type for {@code outputRef()} output
- */
-@Operator
-public final class ScatterDiv<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.core.ScatterDiv}
-   */
-  public static class Options {
-    
-    /**
-     * @param useLocking If True, the operation will be protected by a lock;
-     * otherwise the behavior is undefined, but may exhibit less contention.
-     */
-    public Options useLocking(Boolean useLocking) {
-      this.useLocking = useLocking;
-      return this;
-    }
-    
-    private Boolean useLocking;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new ScatterDiv operation.
-   * 
-   * @param scope current scope
-   * @param ref Should be from a `Variable` node.
-   * @param indices A tensor of indices into the first dimension of `ref`.
-   * @param updates A tensor of values that `ref` is divided by.
-   * @param options carries optional attributes values
-   * @return a new instance of ScatterDiv
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> ScatterDiv<T> create(Scope scope, Operand<T> ref, Operand<U> indices, Operand<T> updates, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ScatterDiv", scope.makeOpName("ScatterDiv"));
-    opBuilder.addInput(ref.asOutput(scope));
-    opBuilder.addInput(indices.asOutput(scope));
-    opBuilder.addInput(updates.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.useLocking != null) {
-          opBuilder.setAttr("use_locking", opts.useLocking);
-        }
-      }
-    }
-    return new ScatterDiv<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param useLocking If True, the operation will be protected by a lock;
-   * otherwise the behavior is undefined, but may exhibit less contention.
-   */
-  public static Options useLocking(Boolean useLocking) {
-    return new Options().useLocking(useLocking);
-  }
-  
-  /**
-   * = Same as `ref`.  Returned as a convenience for operations that want
-   * to use the updated values after the update is done.
-   */
-  public Output<T> outputRef() {
-    return outputRef;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return outputRef;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ScatterDiv";
-  
-  private Output<T> outputRef;
-  
-  private ScatterDiv(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    outputRef = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterMax.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterMax.java
deleted file mode 100644
index 75ef0b8455c..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterMax.java
+++ /dev/null
@@ -1,139 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Reduces sparse updates into a variable reference using the `max` operation.
- * <p>
- * This operation computes
- * <p>
- *     # Scalar indices
- *     ref[indices, ...] = max(ref[indices, ...], updates[...])
- * <p>
- *     # Vector indices (for each i)
- *     ref[indices[i], ...] = max(ref[indices[i], ...], updates[i, ...])
- * <p>
- *     # High rank indices (for each i, ..., j)
- *     ref[indices[i, ..., j], ...] = max(ref[indices[i, ..., j], ...], updates[i, ..., j, ...])
- * <p>
- * This operation outputs `ref` after the update is done.
- * This makes it easier to chain operations that need to use the reset value.
- * <p>
- * Duplicate entries are handled correctly: if multiple `indices` reference
- * the same location, their contributions combine.
- * <p>
- * Requires `updates.shape = indices.shape + ref.shape[1:]` or `updates.shape = []`.
- * <p>
- * <div style="width:70%; margin:auto; margin-bottom:10px; margin-top:20px;">
- * <img style="width:100%" src="https://www.tensorflow.org/images/ScatterAdd.png" alt>
- * </div>
- * 
- * @param <T> data type for {@code outputRef()} output
- */
-@Operator
-public final class ScatterMax<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.core.ScatterMax}
-   */
-  public static class Options {
-    
-    /**
-     * @param useLocking If True, the update will be protected by a lock;
-     * otherwise the behavior is undefined, but may exhibit less contention.
-     */
-    public Options useLocking(Boolean useLocking) {
-      this.useLocking = useLocking;
-      return this;
-    }
-    
-    private Boolean useLocking;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new ScatterMax operation.
-   * 
-   * @param scope current scope
-   * @param ref Should be from a `Variable` node.
-   * @param indices A tensor of indices into the first dimension of `ref`.
-   * @param updates A tensor of updated values to reduce into `ref`.
-   * @param options carries optional attributes values
-   * @return a new instance of ScatterMax
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber, U extends TNumber> ScatterMax<T> create(Scope scope, Operand<T> ref, Operand<U> indices, Operand<T> updates, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ScatterMax", scope.makeOpName("ScatterMax"));
-    opBuilder.addInput(ref.asOutput(scope));
-    opBuilder.addInput(indices.asOutput(scope));
-    opBuilder.addInput(updates.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.useLocking != null) {
-          opBuilder.setAttr("use_locking", opts.useLocking);
-        }
-      }
-    }
-    return new ScatterMax<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param useLocking If True, the update will be protected by a lock;
-   * otherwise the behavior is undefined, but may exhibit less contention.
-   */
-  public static Options useLocking(Boolean useLocking) {
-    return new Options().useLocking(useLocking);
-  }
-  
-  /**
-   * = Same as `ref`.  Returned as a convenience for operations that want
-   * to use the updated values after the update is done.
-   */
-  public Output<T> outputRef() {
-    return outputRef;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return outputRef;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ScatterMax";
-  
-  private Output<T> outputRef;
-  
-  private ScatterMax(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    outputRef = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterMin.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterMin.java
deleted file mode 100644
index b14c0f3f84e..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterMin.java
+++ /dev/null
@@ -1,139 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Reduces sparse updates into a variable reference using the `min` operation.
- * <p>
- * This operation computes
- * <p>
- *     # Scalar indices
- *     ref[indices, ...] = min(ref[indices, ...], updates[...])
- * <p>
- *     # Vector indices (for each i)
- *     ref[indices[i], ...] = min(ref[indices[i], ...], updates[i, ...])
- * <p>
- *     # High rank indices (for each i, ..., j)
- *     ref[indices[i, ..., j], ...] = min(ref[indices[i, ..., j], ...], updates[i, ..., j, ...])
- * <p>
- * This operation outputs `ref` after the update is done.
- * This makes it easier to chain operations that need to use the reset value.
- * <p>
- * Duplicate entries are handled correctly: if multiple `indices` reference
- * the same location, their contributions combine.
- * <p>
- * Requires `updates.shape = indices.shape + ref.shape[1:]` or `updates.shape = []`.
- * <p>
- * <div style="width:70%; margin:auto; margin-bottom:10px; margin-top:20px;">
- * <img style="width:100%" src="https://www.tensorflow.org/images/ScatterAdd.png" alt>
- * </div>
- * 
- * @param <T> data type for {@code outputRef()} output
- */
-@Operator
-public final class ScatterMin<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.core.ScatterMin}
-   */
-  public static class Options {
-    
-    /**
-     * @param useLocking If True, the update will be protected by a lock;
-     * otherwise the behavior is undefined, but may exhibit less contention.
-     */
-    public Options useLocking(Boolean useLocking) {
-      this.useLocking = useLocking;
-      return this;
-    }
-    
-    private Boolean useLocking;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new ScatterMin operation.
-   * 
-   * @param scope current scope
-   * @param ref Should be from a `Variable` node.
-   * @param indices A tensor of indices into the first dimension of `ref`.
-   * @param updates A tensor of updated values to reduce into `ref`.
-   * @param options carries optional attributes values
-   * @return a new instance of ScatterMin
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber, U extends TNumber> ScatterMin<T> create(Scope scope, Operand<T> ref, Operand<U> indices, Operand<T> updates, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ScatterMin", scope.makeOpName("ScatterMin"));
-    opBuilder.addInput(ref.asOutput(scope));
-    opBuilder.addInput(indices.asOutput(scope));
-    opBuilder.addInput(updates.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.useLocking != null) {
-          opBuilder.setAttr("use_locking", opts.useLocking);
-        }
-      }
-    }
-    return new ScatterMin<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param useLocking If True, the update will be protected by a lock;
-   * otherwise the behavior is undefined, but may exhibit less contention.
-   */
-  public static Options useLocking(Boolean useLocking) {
-    return new Options().useLocking(useLocking);
-  }
-  
-  /**
-   * = Same as `ref`.  Returned as a convenience for operations that want
-   * to use the updated values after the update is done.
-   */
-  public Output<T> outputRef() {
-    return outputRef;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return outputRef;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ScatterMin";
-  
-  private Output<T> outputRef;
-  
-  private ScatterMin(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    outputRef = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterMul.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterMul.java
deleted file mode 100644
index c2497cb69ad..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterMul.java
+++ /dev/null
@@ -1,136 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * Multiplies sparse updates into a variable reference.
- * <p>
- * This operation computes
- * <pre>{@code
- *     # Scalar indices
- *     ref[indices, ...] *= updates[...]
- * 
- *     # Vector indices (for each i)
- *     ref[indices[i], ...] *= updates[i, ...]
- * 
- *     # High rank indices (for each i, ..., j)
- *     ref[indices[i, ..., j], ...] *= updates[i, ..., j, ...]
- * }</pre>
- * This operation outputs `ref` after the update is done.
- * This makes it easier to chain operations that need to use the reset value.
- * <p>
- * Duplicate entries are handled correctly: if multiple `indices` reference
- * the same location, their contributions multiply.
- * <p>
- * Requires `updates.shape = indices.shape + ref.shape[1:]` or `updates.shape = []`.
- * 
- * @param <T> data type for {@code outputRef()} output
- */
-@Operator
-public final class ScatterMul<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.core.ScatterMul}
-   */
-  public static class Options {
-    
-    /**
-     * @param useLocking If True, the operation will be protected by a lock;
-     * otherwise the behavior is undefined, but may exhibit less contention.
-     */
-    public Options useLocking(Boolean useLocking) {
-      this.useLocking = useLocking;
-      return this;
-    }
-    
-    private Boolean useLocking;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new ScatterMul operation.
-   * 
-   * @param scope current scope
-   * @param ref Should be from a `Variable` node.
-   * @param indices A tensor of indices into the first dimension of `ref`.
-   * @param updates A tensor of updated values to multiply to `ref`.
-   * @param options carries optional attributes values
-   * @return a new instance of ScatterMul
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> ScatterMul<T> create(Scope scope, Operand<T> ref, Operand<U> indices, Operand<T> updates, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ScatterMul", scope.makeOpName("ScatterMul"));
-    opBuilder.addInput(ref.asOutput(scope));
-    opBuilder.addInput(indices.asOutput(scope));
-    opBuilder.addInput(updates.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.useLocking != null) {
-          opBuilder.setAttr("use_locking", opts.useLocking);
-        }
-      }
-    }
-    return new ScatterMul<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param useLocking If True, the operation will be protected by a lock;
-   * otherwise the behavior is undefined, but may exhibit less contention.
-   */
-  public static Options useLocking(Boolean useLocking) {
-    return new Options().useLocking(useLocking);
-  }
-  
-  /**
-   * = Same as `ref`.  Returned as a convenience for operations that want
-   * to use the updated values after the update is done.
-   */
-  public Output<T> outputRef() {
-    return outputRef;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return outputRef;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ScatterMul";
-  
-  private Output<T> outputRef;
-  
-  private ScatterMul(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    outputRef = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterNd.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterNd.java
deleted file mode 100644
index efba30f95a1..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterNd.java
+++ /dev/null
@@ -1,158 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * Scatter `updates` into a new tensor according to `indices`.
- * <p>
- * Creates a new tensor by applying sparse `updates` to individual values or
- * slices within a tensor (initially zero for numeric, empty for string) of
- * the given `shape` according to indices.  This operator is the inverse of the
- * `tf.gather_nd` operator which extracts values or slices from a given tensor.
- * <p>
- * This operation is similar to tensor_scatter_add, except that the tensor is
- * zero-initialized. Calling `tf.scatter_nd(indices, values, shape)` is identical
- * to `tensor_scatter_add(tf.zeros(shape, values.dtype), indices, values)`
- * <p>
- * If `indices` contains duplicates, then their updates are accumulated (summed).
- * <p>
- * <b>WARNING</b>: The order in which updates are applied is nondeterministic, so the
- * output will be nondeterministic if `indices` contains duplicates -- because
- * of some numerical approximation issues, numbers summed in different order
- * may yield different results.
- * <p>
- * `indices` is an integer tensor containing indices into a new tensor of shape
- * `shape`.  The last dimension of `indices` can be at most the rank of `shape`:
- * <p>
- *     indices.shape[-1] <= shape.rank
- * <p>
- * The last dimension of `indices` corresponds to indices into elements
- * (if `indices.shape[-1] = shape.rank`) or slices
- * (if `indices.shape[-1] < shape.rank`) along dimension `indices.shape[-1]` of
- * `shape`.  `updates` is a tensor with shape
- * <p>
- *     indices.shape[:-1] + shape[indices.shape[-1]:]
- * <p>
- * The simplest form of scatter is to insert individual elements in a tensor by
- * index. For example, say we want to insert 4 scattered elements in a rank-1
- * tensor with 8 elements.
- * <p>
- * <div style="width:70%; margin:auto; margin-bottom:10px; margin-top:20px;">
- * <img style="width:100%" src="https://www.tensorflow.org/images/ScatterNd1.png" alt>
- * </div>
- * <p>
- * In Python, this scatter operation would look like this:
- * <pre>{@code
- *     indices = tf.constant([[4], [3], [1], [7]])
- *     updates = tf.constant([9, 10, 11, 12])
- *     shape = tf.constant([8])
- *     scatter = tf.scatter_nd(indices, updates, shape)
- *     print(scatter)
- * }</pre>
- * The resulting tensor would look like this:
- * <p>
- *     [0, 11, 0, 10, 9, 0, 0, 12]
- * <p>
- * We can also, insert entire slices of a higher rank tensor all at once. For
- * example, if we wanted to insert two slices in the first dimension of a
- * rank-3 tensor with two matrices of new values.
- * <p>
- * <div style="width:70%; margin:auto; margin-bottom:10px; margin-top:20px;">
- * <img style="width:100%" src="https://www.tensorflow.org/images/ScatterNd2.png" alt>
- * </div>
- * <p>
- * In Python, this scatter operation would look like this:
- * <pre>{@code
- *     indices = tf.constant([[0], [2]])
- *     updates = tf.constant([[[5, 5, 5, 5], [6, 6, 6, 6],
- *                             [7, 7, 7, 7], [8, 8, 8, 8]],
- *                            [[5, 5, 5, 5], [6, 6, 6, 6],
- *                             [7, 7, 7, 7], [8, 8, 8, 8]]])
- *     shape = tf.constant([4, 4, 4])
- *     scatter = tf.scatter_nd(indices, updates, shape)
- *     print(scatter)
- * }</pre>
- * The resulting tensor would look like this:
- * <p>
- *     [[[5, 5, 5, 5], [6, 6, 6, 6], [7, 7, 7, 7], [8, 8, 8, 8]],
- *      [[0, 0, 0, 0], [0, 0, 0, 0], [0, 0, 0, 0], [0, 0, 0, 0]],
- *      [[5, 5, 5, 5], [6, 6, 6, 6], [7, 7, 7, 7], [8, 8, 8, 8]],
- *      [[0, 0, 0, 0], [0, 0, 0, 0], [0, 0, 0, 0], [0, 0, 0, 0]]]
- * <p>
- * Note that on CPU, if an out of bound index is found, an error is returned.
- * On GPU, if an out of bound index is found, the index is ignored.
- * 
- * @param <U> data type for {@code output()} output
- */
-@Operator
-public final class ScatterNd<U extends TType> extends RawOp implements Operand<U> {
-  
-  /**
-   * Factory method to create a class wrapping a new ScatterNd operation.
-   * 
-   * @param scope current scope
-   * @param indices Index tensor.
-   * @param updates Updates to scatter into output.
-   * @param shape 1-D. The shape of the resulting tensor.
-   * @return a new instance of ScatterNd
-   */
-  @Endpoint(describeByClass = true)
-  public static <U extends TType, T extends TNumber> ScatterNd<U> create(Scope scope, Operand<T> indices, Operand<U> updates, Operand<T> shape) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ScatterNd", scope.makeOpName("ScatterNd"));
-    opBuilder.addInput(indices.asOutput(scope));
-    opBuilder.addInput(updates.asOutput(scope));
-    opBuilder.addInput(shape.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new ScatterNd<U>(opBuilder.build());
-  }
-  
-  /**
-   * A new tensor with the given shape and updates applied according
-   * to the indices.
-   */
-  public Output<U> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<U> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ScatterNd";
-  
-  private Output<U> output;
-  
-  private ScatterNd(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterNdAdd.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterNdAdd.java
deleted file mode 100644
index 92d40566bc2..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterNdAdd.java
+++ /dev/null
@@ -1,151 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * Applies sparse addition to individual values or slices in a Variable.
- * <p>
- * `ref` is a `Tensor` with rank `P` and `indices` is a `Tensor` of rank `Q`.
- * <p>
- * `indices` must be integer tensor, containing indices into `ref`.
- * It must be shape `[d_0, ..., d_{Q-2}, K]` where `0 < K <= P`.
- * <p>
- * The innermost dimension of `indices` (with length `K`) corresponds to
- * indices into elements (if `K = P`) or slices (if `K < P`) along the `K`th
- * dimension of `ref`.
- * <p>
- * `updates` is `Tensor` of rank `Q-1+P-K` with shape:
- * <pre>{@code
- * [d_0, ..., d_{Q-2}, ref.shape[K], ..., ref.shape[P-1]]
- * }</pre>
- * For example, say we want to add 4 scattered elements to a rank-1 tensor to
- * 8 elements. In Python, that addition would look like this:
- * <pre>{@code
- * ref = tf.Variable([1, 2, 3, 4, 5, 6, 7, 8])
- * indices = tf.constant([[4], [3], [1], [7]])
- * updates = tf.constant([9, 10, 11, 12])
- * add = tf.scatter_nd_add(ref, indices, updates)
- * with tf.Session() as sess:
- *   print sess.run(add)
- * }</pre>
- * The resulting update to ref would look like this:
- * <p>
- *     [1, 13, 3, 14, 14, 6, 7, 20]
- * <p>
- * See `tf.scatter_nd` for more details about how to make updates to
- * slices.
- * 
- * @param <T> data type for {@code outputRef()} output
- */
-@Operator
-public final class ScatterNdAdd<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.core.ScatterNdAdd}
-   */
-  public static class Options {
-    
-    /**
-     * @param useLocking An optional bool. Defaults to True. If True, the assignment will
-     * be protected by a lock; otherwise the behavior is undefined,
-     * but may exhibit less contention.
-     */
-    public Options useLocking(Boolean useLocking) {
-      this.useLocking = useLocking;
-      return this;
-    }
-    
-    private Boolean useLocking;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new ScatterNdAdd operation.
-   * 
-   * @param scope current scope
-   * @param ref A mutable Tensor. Should be from a Variable node.
-   * @param indices A Tensor. Must be one of the following types: int32, int64.
-   * A tensor of indices into ref.
-   * @param updates A Tensor. Must have the same type as ref. A tensor of updated values
-   * to add to ref.
-   * @param options carries optional attributes values
-   * @return a new instance of ScatterNdAdd
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> ScatterNdAdd<T> create(Scope scope, Operand<T> ref, Operand<U> indices, Operand<T> updates, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ScatterNdAdd", scope.makeOpName("ScatterNdAdd"));
-    opBuilder.addInput(ref.asOutput(scope));
-    opBuilder.addInput(indices.asOutput(scope));
-    opBuilder.addInput(updates.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.useLocking != null) {
-          opBuilder.setAttr("use_locking", opts.useLocking);
-        }
-      }
-    }
-    return new ScatterNdAdd<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param useLocking An optional bool. Defaults to True. If True, the assignment will
-   * be protected by a lock; otherwise the behavior is undefined,
-   * but may exhibit less contention.
-   */
-  public static Options useLocking(Boolean useLocking) {
-    return new Options().useLocking(useLocking);
-  }
-  
-  /**
-   * Same as ref. Returned as a convenience for operations that want
-   * to use the updated values after the update is done.
-   */
-  public Output<T> outputRef() {
-    return outputRef;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return outputRef;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ScatterNdAdd";
-  
-  private Output<T> outputRef;
-  
-  private ScatterNdAdd(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    outputRef = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterNdMax.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterNdMax.java
deleted file mode 100644
index dcce88230ec..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterNdMax.java
+++ /dev/null
@@ -1,120 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * Computes element-wise maximum.
- * 
- * @param <T> data type for {@code outputRef()} output
- */
-public final class ScatterNdMax<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.core.ScatterNdMax}
-   */
-  public static class Options {
-    
-    /**
-     * @param useLocking An optional bool. Defaults to True. If True, the assignment will
-     * be protected by a lock; otherwise the behavior is undefined,
-     * but may exhibit less contention.
-     */
-    public Options useLocking(Boolean useLocking) {
-      this.useLocking = useLocking;
-      return this;
-    }
-    
-    private Boolean useLocking;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new ScatterNdMax operation.
-   * 
-   * @param scope current scope
-   * @param ref A mutable Tensor. Should be from a Variable node.
-   * @param indices A Tensor. Must be one of the following types: int32, int64.
-   * A tensor of indices into ref.
-   * @param updates A Tensor. Must have the same type as ref. A tensor of updated values
-   * to add to ref.
-   * @param options carries optional attributes values
-   * @return a new instance of ScatterNdMax
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> ScatterNdMax<T> create(Scope scope, Operand<T> ref, Operand<U> indices, Operand<T> updates, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ScatterNdMax", scope.makeOpName("ScatterNdMax"));
-    opBuilder.addInput(ref.asOutput(scope));
-    opBuilder.addInput(indices.asOutput(scope));
-    opBuilder.addInput(updates.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.useLocking != null) {
-          opBuilder.setAttr("use_locking", opts.useLocking);
-        }
-      }
-    }
-    return new ScatterNdMax<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param useLocking An optional bool. Defaults to True. If True, the assignment will
-   * be protected by a lock; otherwise the behavior is undefined,
-   * but may exhibit less contention.
-   */
-  public static Options useLocking(Boolean useLocking) {
-    return new Options().useLocking(useLocking);
-  }
-  
-  /**
-   * Same as ref. Returned as a convenience for operations that want
-   * to use the updated values after the update is done.
-   */
-  public Output<T> outputRef() {
-    return outputRef;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return outputRef;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ScatterNdMax";
-  
-  private Output<T> outputRef;
-  
-  private ScatterNdMax(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    outputRef = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterNdMin.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterNdMin.java
deleted file mode 100644
index 73e74d4b50b..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterNdMin.java
+++ /dev/null
@@ -1,120 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * Computes element-wise minimum.
- * 
- * @param <T> data type for {@code outputRef()} output
- */
-public final class ScatterNdMin<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.core.ScatterNdMin}
-   */
-  public static class Options {
-    
-    /**
-     * @param useLocking An optional bool. Defaults to True. If True, the assignment will
-     * be protected by a lock; otherwise the behavior is undefined,
-     * but may exhibit less contention.
-     */
-    public Options useLocking(Boolean useLocking) {
-      this.useLocking = useLocking;
-      return this;
-    }
-    
-    private Boolean useLocking;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new ScatterNdMin operation.
-   * 
-   * @param scope current scope
-   * @param ref A mutable Tensor. Should be from a Variable node.
-   * @param indices A Tensor. Must be one of the following types: int32, int64.
-   * A tensor of indices into ref.
-   * @param updates A Tensor. Must have the same type as ref. A tensor of updated values
-   * to add to ref.
-   * @param options carries optional attributes values
-   * @return a new instance of ScatterNdMin
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> ScatterNdMin<T> create(Scope scope, Operand<T> ref, Operand<U> indices, Operand<T> updates, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ScatterNdMin", scope.makeOpName("ScatterNdMin"));
-    opBuilder.addInput(ref.asOutput(scope));
-    opBuilder.addInput(indices.asOutput(scope));
-    opBuilder.addInput(updates.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.useLocking != null) {
-          opBuilder.setAttr("use_locking", opts.useLocking);
-        }
-      }
-    }
-    return new ScatterNdMin<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param useLocking An optional bool. Defaults to True. If True, the assignment will
-   * be protected by a lock; otherwise the behavior is undefined,
-   * but may exhibit less contention.
-   */
-  public static Options useLocking(Boolean useLocking) {
-    return new Options().useLocking(useLocking);
-  }
-  
-  /**
-   * Same as ref. Returned as a convenience for operations that want
-   * to use the updated values after the update is done.
-   */
-  public Output<T> outputRef() {
-    return outputRef;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return outputRef;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ScatterNdMin";
-  
-  private Output<T> outputRef;
-  
-  private ScatterNdMin(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    outputRef = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterNdNonAliasingAdd.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterNdNonAliasingAdd.java
deleted file mode 100644
index a76b06dd47c..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterNdNonAliasingAdd.java
+++ /dev/null
@@ -1,117 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * Applies sparse addition to `input` using individual values or slices
- * <p>
- * from `updates` according to indices `indices`.  The updates are non-aliasing:
- * `input` is only modified in-place if no other operations will use it.
- * Otherwise, a copy of `input` is made.  This operation has a gradient with
- * respect to both `input` and `updates`.
- * <p>
- * `input` is a `Tensor` with rank `P` and `indices` is a `Tensor` of rank `Q`.
- * <p>
- * `indices` must be integer tensor, containing indices into `input`.
- * It must be shape \\([d_0, ..., d_{Q-2}, K]\\) where `0 < K <= P`.
- * <p>
- * The innermost dimension of `indices` (with length `K`) corresponds to
- * indices into elements (if `K = P`) or `(P-K)`-dimensional slices
- * (if `K < P`) along the `K`th dimension of `input`.
- * <p>
- * `updates` is `Tensor` of rank `Q-1+P-K` with shape:
- * <p>
- * $$[d_0, ..., d_{Q-2}, input.shape[K], ..., input.shape[P-1]].$$
- * <p>
- * For example, say we want to add 4 scattered elements to a rank-1 tensor to 8
- * elements. In Python, that addition would look like this:
- * <p>
- *     input = tf.constant([1, 2, 3, 4, 5, 6, 7, 8])
- *     indices = tf.constant([[4], [3], [1], [7]])
- *     updates = tf.constant([9, 10, 11, 12])
- *     output = tf.scatter_nd_non_aliasing_add(input, indices, updates)
- *     with tf.Session() as sess:
- *       print(sess.run(output))
- * <p>
- * The resulting value `output` would look like this:
- * <p>
- *     [1, 13, 3, 14, 14, 6, 7, 20]
- * <p>
- * See `tf.scatter_nd` for more details about how to make updates to slices.
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator
-public final class ScatterNdNonAliasingAdd<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new ScatterNdNonAliasingAdd operation.
-   * 
-   * @param scope current scope
-   * @param input A Tensor.
-   * @param indices A Tensor. Must be one of the following types: `int32`, `int64`.
-   * A tensor of indices into `input`.
-   * @param updates A Tensor. Must have the same type as ref. A tensor of updated values
-   * to add to `input`.
-   * @return a new instance of ScatterNdNonAliasingAdd
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> ScatterNdNonAliasingAdd<T> create(Scope scope, Operand<T> input, Operand<U> indices, Operand<T> updates) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ScatterNdNonAliasingAdd", scope.makeOpName("ScatterNdNonAliasingAdd"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder.addInput(indices.asOutput(scope));
-    opBuilder.addInput(updates.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new ScatterNdNonAliasingAdd<T>(opBuilder.build());
-  }
-  
-  /**
-   * A `Tensor` with the same shape as `input`, containing values of `input`
-   * updated with `updates`.
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ScatterNdNonAliasingAdd";
-  
-  private Output<T> output;
-  
-  private ScatterNdNonAliasingAdd(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterNdSub.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterNdSub.java
deleted file mode 100644
index 402367ab5d1..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterNdSub.java
+++ /dev/null
@@ -1,153 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * Applies sparse subtraction to individual values or slices in a Variable.
- * <p>
- * within a given variable according to `indices`.
- * <p>
- * `ref` is a `Tensor` with rank `P` and `indices` is a `Tensor` of rank `Q`.
- * <p>
- * `indices` must be integer tensor, containing indices into `ref`.
- * It must be shape `[d_0, ..., d_{Q-2}, K]` where `0 < K <= P`.
- * <p>
- * The innermost dimension of `indices` (with length `K`) corresponds to
- * indices into elements (if `K = P`) or slices (if `K < P`) along the `K`th
- * dimension of `ref`.
- * <p>
- * `updates` is `Tensor` of rank `Q-1+P-K` with shape:
- * <pre>{@code
- * [d_0, ..., d_{Q-2}, ref.shape[K], ..., ref.shape[P-1]]
- * }</pre>
- * For example, say we want to subtract 4 scattered elements from a rank-1 tensor
- * with 8 elements. In Python, that subtraction would look like this:
- * <pre>{@code
- * ref = tf.Variable([1, 2, 3, 4, 5, 6, 7, 8])
- * indices = tf.constant([[4], [3], [1], [7]])
- * updates = tf.constant([9, 10, 11, 12])
- * sub = tf.scatter_nd_sub(ref, indices, updates)
- * with tf.Session() as sess:
- *   print sess.run(sub)
- * }</pre>
- * The resulting update to ref would look like this:
- * <p>
- *     [1, -9, 3, -6, -4, 6, 7, -4]
- * <p>
- * See `tf.scatter_nd` for more details about how to make updates to
- * slices.
- * 
- * @param <T> data type for {@code outputRef()} output
- */
-@Operator
-public final class ScatterNdSub<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.core.ScatterNdSub}
-   */
-  public static class Options {
-    
-    /**
-     * @param useLocking An optional bool. Defaults to True. If True, the assignment will
-     * be protected by a lock; otherwise the behavior is undefined,
-     * but may exhibit less contention.
-     */
-    public Options useLocking(Boolean useLocking) {
-      this.useLocking = useLocking;
-      return this;
-    }
-    
-    private Boolean useLocking;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new ScatterNdSub operation.
-   * 
-   * @param scope current scope
-   * @param ref A mutable Tensor. Should be from a Variable node.
-   * @param indices A Tensor. Must be one of the following types: int32, int64.
-   * A tensor of indices into ref.
-   * @param updates A Tensor. Must have the same type as ref. A tensor of updated values
-   * to subtract from ref.
-   * @param options carries optional attributes values
-   * @return a new instance of ScatterNdSub
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> ScatterNdSub<T> create(Scope scope, Operand<T> ref, Operand<U> indices, Operand<T> updates, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ScatterNdSub", scope.makeOpName("ScatterNdSub"));
-    opBuilder.addInput(ref.asOutput(scope));
-    opBuilder.addInput(indices.asOutput(scope));
-    opBuilder.addInput(updates.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.useLocking != null) {
-          opBuilder.setAttr("use_locking", opts.useLocking);
-        }
-      }
-    }
-    return new ScatterNdSub<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param useLocking An optional bool. Defaults to True. If True, the assignment will
-   * be protected by a lock; otherwise the behavior is undefined,
-   * but may exhibit less contention.
-   */
-  public static Options useLocking(Boolean useLocking) {
-    return new Options().useLocking(useLocking);
-  }
-  
-  /**
-   * Same as ref. Returned as a convenience for operations that want
-   * to use the updated values after the update is done.
-   */
-  public Output<T> outputRef() {
-    return outputRef;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return outputRef;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ScatterNdSub";
-  
-  private Output<T> outputRef;
-  
-  private ScatterNdSub(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    outputRef = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterNdUpdate.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterNdUpdate.java
deleted file mode 100644
index 62e7d3461db..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterNdUpdate.java
+++ /dev/null
@@ -1,155 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * Applies sparse `updates` to individual values or slices within a given
- * <p>
- * variable according to `indices`.
- * <p>
- * `ref` is a `Tensor` with rank `P` and `indices` is a `Tensor` of rank `Q`.
- * <p>
- * `indices` must be integer tensor, containing indices into `ref`.
- * It must be shape \\([d_0, ..., d_{Q-2}, K]\\) where `0 < K <= P`.
- * <p>
- * The innermost dimension of `indices` (with length `K`) corresponds to
- * indices into elements (if `K = P`) or slices (if `K < P`) along the `K`th
- * dimension of `ref`.
- * <p>
- * `updates` is `Tensor` of rank `Q-1+P-K` with shape:
- * <p>
- * $$[d_0, ..., d_{Q-2}, ref.shape[K], ..., ref.shape[P-1]].$$
- * <p>
- * For example, say we want to update 4 scattered elements to a rank-1 tensor to
- * 8 elements. In Python, that update would look like this:
- * <pre>{@code
- *     ref = tf.Variable([1, 2, 3, 4, 5, 6, 7, 8])
- *     indices = tf.constant([[4], [3], [1] ,[7]])
- *     updates = tf.constant([9, 10, 11, 12])
- *     update = tf.scatter_nd_update(ref, indices, updates)
- *     with tf.Session() as sess:
- *       print sess.run(update)
- * }</pre>
- * The resulting update to ref would look like this:
- * <p>
- *     [1, 11, 3, 10, 9, 6, 7, 12]
- * <p>
- * See `tf.scatter_nd` for more details about how to make updates to
- * slices.
- * <p>
- * See also `tf.scatter_update` and `tf.batch_scatter_update`.
- * 
- * @param <T> data type for {@code outputRef()} output
- */
-@Operator
-public final class ScatterNdUpdate<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.core.ScatterNdUpdate}
-   */
-  public static class Options {
-    
-    /**
-     * @param useLocking An optional bool. Defaults to True. If True, the assignment will
-     * be protected by a lock; otherwise the behavior is undefined,
-     * but may exhibit less contention.
-     */
-    public Options useLocking(Boolean useLocking) {
-      this.useLocking = useLocking;
-      return this;
-    }
-    
-    private Boolean useLocking;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new ScatterNdUpdate operation.
-   * 
-   * @param scope current scope
-   * @param ref A mutable Tensor. Should be from a Variable node.
-   * @param indices A Tensor. Must be one of the following types: int32, int64.
-   * A tensor of indices into ref.
-   * @param updates A Tensor. Must have the same type as ref. A tensor of updated
-   * values to add to ref.
-   * @param options carries optional attributes values
-   * @return a new instance of ScatterNdUpdate
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> ScatterNdUpdate<T> create(Scope scope, Operand<T> ref, Operand<U> indices, Operand<T> updates, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ScatterNdUpdate", scope.makeOpName("ScatterNdUpdate"));
-    opBuilder.addInput(ref.asOutput(scope));
-    opBuilder.addInput(indices.asOutput(scope));
-    opBuilder.addInput(updates.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.useLocking != null) {
-          opBuilder.setAttr("use_locking", opts.useLocking);
-        }
-      }
-    }
-    return new ScatterNdUpdate<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param useLocking An optional bool. Defaults to True. If True, the assignment will
-   * be protected by a lock; otherwise the behavior is undefined,
-   * but may exhibit less contention.
-   */
-  public static Options useLocking(Boolean useLocking) {
-    return new Options().useLocking(useLocking);
-  }
-  
-  /**
-   * Same as ref. Returned as a convenience for operations that want to
-   * use the updated values after the update is done.
-   */
-  public Output<T> outputRef() {
-    return outputRef;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return outputRef;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ScatterNdUpdate";
-  
-  private Output<T> outputRef;
-  
-  private ScatterNdUpdate(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    outputRef = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterSub.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterSub.java
deleted file mode 100644
index ae5a81ffd50..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterSub.java
+++ /dev/null
@@ -1,139 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * Subtracts sparse updates to a variable reference.
- * <p>
- * <pre>{@code
- *     # Scalar indices
- *     ref[indices, ...] -= updates[...]
- * 
- *     # Vector indices (for each i)
- *     ref[indices[i], ...] -= updates[i, ...]
- * 
- *     # High rank indices (for each i, ..., j)
- *     ref[indices[i, ..., j], ...] -= updates[i, ..., j, ...]
- * }</pre>
- * This operation outputs `ref` after the update is done.
- * This makes it easier to chain operations that need to use the reset value.
- * <p>
- * Duplicate entries are handled correctly: if multiple `indices` reference
- * the same location, their (negated) contributions add.
- * <p>
- * Requires `updates.shape = indices.shape + ref.shape[1:]` or `updates.shape = []`.
- * <p>
- * <div style="width:70%; margin:auto; margin-bottom:10px; margin-top:20px;">
- * <img style="width:100%" src="https://www.tensorflow.org/images/ScatterSub.png" alt>
- * </div>
- * 
- * @param <T> data type for {@code outputRef()} output
- */
-@Operator
-public final class ScatterSub<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.core.ScatterSub}
-   */
-  public static class Options {
-    
-    /**
-     * @param useLocking If True, the subtraction will be protected by a lock;
-     * otherwise the behavior is undefined, but may exhibit less contention.
-     */
-    public Options useLocking(Boolean useLocking) {
-      this.useLocking = useLocking;
-      return this;
-    }
-    
-    private Boolean useLocking;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new ScatterSub operation.
-   * 
-   * @param scope current scope
-   * @param ref Should be from a `Variable` node.
-   * @param indices A tensor of indices into the first dimension of `ref`.
-   * @param updates A tensor of updated values to subtract from `ref`.
-   * @param options carries optional attributes values
-   * @return a new instance of ScatterSub
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> ScatterSub<T> create(Scope scope, Operand<T> ref, Operand<U> indices, Operand<T> updates, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ScatterSub", scope.makeOpName("ScatterSub"));
-    opBuilder.addInput(ref.asOutput(scope));
-    opBuilder.addInput(indices.asOutput(scope));
-    opBuilder.addInput(updates.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.useLocking != null) {
-          opBuilder.setAttr("use_locking", opts.useLocking);
-        }
-      }
-    }
-    return new ScatterSub<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param useLocking If True, the subtraction will be protected by a lock;
-   * otherwise the behavior is undefined, but may exhibit less contention.
-   */
-  public static Options useLocking(Boolean useLocking) {
-    return new Options().useLocking(useLocking);
-  }
-  
-  /**
-   * = Same as `ref`.  Returned as a convenience for operations that want
-   * to use the updated values after the update is done.
-   */
-  public Output<T> outputRef() {
-    return outputRef;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return outputRef;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ScatterSub";
-  
-  private Output<T> outputRef;
-  
-  private ScatterSub(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    outputRef = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterUpdate.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterUpdate.java
deleted file mode 100644
index 5768a097438..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterUpdate.java
+++ /dev/null
@@ -1,143 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * Applies sparse updates to a variable reference.
- * <p>
- * This operation computes
- * <pre>{@code
- *     # Scalar indices
- *     ref[indices, ...] = updates[...]
- * 
- *     # Vector indices (for each i)
- *     ref[indices[i], ...] = updates[i, ...]
- * 
- *     # High rank indices (for each i, ..., j)
- *     ref[indices[i, ..., j], ...] = updates[i, ..., j, ...]
- * }</pre>
- * This operation outputs `ref` after the update is done.
- * This makes it easier to chain operations that need to use the reset value.
- * <p>
- * If values in `ref` is to be updated more than once, because there are
- * duplicate entries in `indices`, the order at which the updates happen
- * for each value is undefined.
- * <p>
- * Requires `updates.shape = indices.shape + ref.shape[1:]` or `updates.shape = []`.
- * <p>
- * <div style="width:70%; margin:auto; margin-bottom:10px; margin-top:20px;">
- * <img style="width:100%" src="https://www.tensorflow.org/images/ScatterUpdate.png" alt>
- * </div>
- * <p>
- * See also `tf.batch_scatter_update` and `tf.scatter_nd_update`.
- * 
- * @param <T> data type for {@code outputRef()} output
- */
-@Operator
-public final class ScatterUpdate<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.core.ScatterUpdate}
-   */
-  public static class Options {
-    
-    /**
-     * @param useLocking If True, the assignment will be protected by a lock;
-     * otherwise the behavior is undefined, but may exhibit less contention.
-     */
-    public Options useLocking(Boolean useLocking) {
-      this.useLocking = useLocking;
-      return this;
-    }
-    
-    private Boolean useLocking;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new ScatterUpdate operation.
-   * 
-   * @param scope current scope
-   * @param ref Should be from a `Variable` node.
-   * @param indices A tensor of indices into the first dimension of `ref`.
-   * @param updates A tensor of updated values to store in `ref`.
-   * @param options carries optional attributes values
-   * @return a new instance of ScatterUpdate
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> ScatterUpdate<T> create(Scope scope, Operand<T> ref, Operand<U> indices, Operand<T> updates, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ScatterUpdate", scope.makeOpName("ScatterUpdate"));
-    opBuilder.addInput(ref.asOutput(scope));
-    opBuilder.addInput(indices.asOutput(scope));
-    opBuilder.addInput(updates.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.useLocking != null) {
-          opBuilder.setAttr("use_locking", opts.useLocking);
-        }
-      }
-    }
-    return new ScatterUpdate<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param useLocking If True, the assignment will be protected by a lock;
-   * otherwise the behavior is undefined, but may exhibit less contention.
-   */
-  public static Options useLocking(Boolean useLocking) {
-    return new Options().useLocking(useLocking);
-  }
-  
-  /**
-   * = Same as `ref`.  Returned as a convenience for operations that want
-   * to use the updated values after the update is done.
-   */
-  public Output<T> outputRef() {
-    return outputRef;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return outputRef;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ScatterUpdate";
-  
-  private Output<T> outputRef;
-  
-  private ScatterUpdate(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    outputRef = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Select.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Select.java
deleted file mode 100644
index caa85e1ea44..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Select.java
+++ /dev/null
@@ -1,77 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TBool;
-import org.tensorflow.types.family.TType;
-
-/**
- * @param <T> data type for {@code output()} output
- */
-@Operator
-public final class Select<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new Select operation.
-   * 
-   * @param scope current scope
-   * @param condition 
-   * @param t 
-   * @param e 
-   * @return a new instance of Select
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> Select<T> create(Scope scope, Operand<TBool> condition, Operand<T> t, Operand<T> e) {
-    OperationBuilder opBuilder = scope.env().opBuilder("SelectV2", scope.makeOpName("Select"));
-    opBuilder.addInput(condition.asOutput(scope));
-    opBuilder.addInput(t.asOutput(scope));
-    opBuilder.addInput(e.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new Select<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "SelectV2";
-  
-  private Output<T> output;
-  
-  private Select(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Send.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Send.java
deleted file mode 100644
index c3a5d086ad1..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Send.java
+++ /dev/null
@@ -1,103 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Sends the named tensor from send_device to recv_device.
- */
-public final class Send extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.core.Send}
-   */
-  public static class Options {
-    
-    /**
-     * @param clientTerminated If set to true, this indicates that the node was added
-     * to the graph as a result of a client-side feed or fetch of Tensor data,
-     * in which case the corresponding send or recv is expected to be managed
-     * locally by the caller.
-     */
-    public Options clientTerminated(Boolean clientTerminated) {
-      this.clientTerminated = clientTerminated;
-      return this;
-    }
-    
-    private Boolean clientTerminated;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new Send operation.
-   * 
-   * @param scope current scope
-   * @param tensor The tensor to send.
-   * @param tensorName The name of the tensor to send.
-   * @param sendDevice The name of the device sending the tensor.
-   * @param sendDeviceIncarnation The current incarnation of send_device.
-   * @param recvDevice The name of the device receiving the tensor.
-   * @param options carries optional attributes values
-   * @return a new instance of Send
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> Send create(Scope scope, Operand<T> tensor, String tensorName, String sendDevice, Long sendDeviceIncarnation, String recvDevice, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Send", scope.makeOpName("Send"));
-    opBuilder.addInput(tensor.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("tensor_name", tensorName);
-    opBuilder.setAttr("send_device", sendDevice);
-    opBuilder.setAttr("send_device_incarnation", sendDeviceIncarnation);
-    opBuilder.setAttr("recv_device", recvDevice);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.clientTerminated != null) {
-          opBuilder.setAttr("client_terminated", opts.clientTerminated);
-        }
-      }
-    }
-    return new Send(opBuilder.build());
-  }
-  
-  /**
-   * @param clientTerminated If set to true, this indicates that the node was added
-   * to the graph as a result of a client-side feed or fetch of Tensor data,
-   * in which case the corresponding send or recv is expected to be managed
-   * locally by the caller.
-   */
-  public static Options clientTerminated(Boolean clientTerminated) {
-    return new Options().clientTerminated(clientTerminated);
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Send";
-  
-  private Send(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/SetDiff1d.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/SetDiff1d.java
deleted file mode 100644
index ece12ee2a91..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/SetDiff1d.java
+++ /dev/null
@@ -1,119 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * Computes the difference between two lists of numbers or strings.
- * <p>
- * Given a list `x` and a list `y`, this operation returns a list `out` that
- * represents all values that are in `x` but not in `y`. The returned list `out`
- * is sorted in the same order that the numbers appear in `x` (duplicates are
- * preserved). This operation also returns a list `idx` that represents the
- * position of each `out` element in `x`. In other words:
- * <p>
- * `out[i] = x[idx[i]] for i in [0, 1, ..., len(out) - 1]`
- * <p>
- * For example, given this input:
- * <pre>{@code
- * x = [1, 2, 3, 4, 5, 6]
- * y = [1, 3, 5]
- * }</pre>
- * This operation would return:
- * <pre>{@code
- * out ==> [2, 4, 6]
- * idx ==> [1, 3, 5]
- * }</pre>
- * 
- * 
- * @param <T> data type for {@code out()} output
- * @param <U> data type for {@code idx()} output
- */
-@Operator
-public final class SetDiff1d<T extends TType, U extends TNumber> extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new SetDiff1d operation.
-   * 
-   * @param scope current scope
-   * @param x 1-D. Values to keep.
-   * @param y 1-D. Values to remove.
-   * @param outIdx 
-   * @return a new instance of SetDiff1d
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> SetDiff1d<T, U> create(Scope scope, Operand<T> x, Operand<T> y, Class<U> outIdx) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ListDiff", scope.makeOpName("SetDiff1d"));
-    opBuilder.addInput(x.asOutput(scope));
-    opBuilder.addInput(y.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("out_idx", outIdx);
-    return new SetDiff1d<T, U>(opBuilder.build());
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new SetDiff1d operation using default output types.
-   * 
-   * @param scope current scope
-   * @param x 1-D. Values to keep.
-   * @param y 1-D. Values to remove.
-   * @return a new instance of SetDiff1d
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> SetDiff1d<T, TInt32> create(Scope scope, Operand<T> x, Operand<T> y) {
-    return create(scope, x, y, TInt32.class);
-  }
-  
-  /**
-   * 1-D. Values present in `x` but not in `y`.
-   */
-  public Output<T> out() {
-    return out;
-  }
-  
-  /**
-   * 1-D. Positions of `x` values preserved in `out`.
-   */
-  public Output<U> idx() {
-    return idx;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ListDiff";
-  
-  private Output<T> out;
-  private Output<U> idx;
-  
-  private SetDiff1d(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    out = operation.output(outputIdx++);
-    idx = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/SetSize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/SetSize.java
deleted file mode 100644
index e5f73a23881..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/SetSize.java
+++ /dev/null
@@ -1,122 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
-
-/**
- * Number of unique elements along last dimension of input `set`.
- * <p>
- * Input `set` is a `SparseTensor` represented by `set_indices`, `set_values`,
- * and `set_shape`. The last dimension contains values in a set, duplicates are
- * allowed but ignored.
- * <p>
- * If `validate_indices` is `True`, this op validates the order and range of `set`
- * indices.
- */
-@Operator
-public final class SetSize extends RawOp implements Operand<TInt32> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.core.SetSize}
-   */
-  public static class Options {
-    
-    /**
-     * @param validateIndices 
-     */
-    public Options validateIndices(Boolean validateIndices) {
-      this.validateIndices = validateIndices;
-      return this;
-    }
-    
-    private Boolean validateIndices;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new SetSize operation.
-   * 
-   * @param scope current scope
-   * @param setIndices 2D `Tensor`, indices of a `SparseTensor`.
-   * @param setValues 1D `Tensor`, values of a `SparseTensor`.
-   * @param setShape 1D `Tensor`, shape of a `SparseTensor`.
-   * @param options carries optional attributes values
-   * @return a new instance of SetSize
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> SetSize create(Scope scope, Operand<TInt64> setIndices, Operand<T> setValues, Operand<TInt64> setShape, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("SetSize", scope.makeOpName("SetSize"));
-    opBuilder.addInput(setIndices.asOutput(scope));
-    opBuilder.addInput(setValues.asOutput(scope));
-    opBuilder.addInput(setShape.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.validateIndices != null) {
-          opBuilder.setAttr("validate_indices", opts.validateIndices);
-        }
-      }
-    }
-    return new SetSize(opBuilder.build());
-  }
-  
-  /**
-   * @param validateIndices 
-   */
-  public static Options validateIndices(Boolean validateIndices) {
-    return new Options().validateIndices(validateIndices);
-  }
-  
-  /**
-   * For `set` ranked `n`, this is a `Tensor` with rank `n-1`, and the same 1st
-   * `n-1` dimensions as `set`. Each value is the number of unique elements in
-   * the corresponding `[0...n-1]` dimension of `set`.
-   */
-  public Output<TInt32> size() {
-    return size;
-  }
-  
-  @Override
-  public Output<TInt32> asOutput(Scope scope) {
-    return size;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "SetSize";
-  
-  private Output<TInt32> size;
-  
-  private SetSize(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    size = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Shape.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Shape.java
deleted file mode 100644
index 3b9779bb418..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Shape.java
+++ /dev/null
@@ -1,99 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * Returns the shape of a tensor.
- * <p>
- * This operation returns a 1-D integer tensor representing the shape of `input`.
- * <p>
- * For example:
- * <pre>{@code
- * # 't' is [[[1, 1, 1], [2, 2, 2]], [[3, 3, 3], [4, 4, 4]]]
- * shape(t) ==> [2, 2, 3]
- * }</pre>
- * 
- * 
- * @param <U> data type for {@code output()} output
- */
-@Operator
-public final class Shape<U extends TNumber> extends RawOp implements Operand<U> {
-  
-  /**
-   * Factory method to create a class wrapping a new Shape operation.
-   * 
-   * @param scope current scope
-   * @param input 
-   * @param outType 
-   * @return a new instance of Shape
-   */
-  @Endpoint(describeByClass = true)
-  public static <U extends TNumber, T extends TType> Shape<U> create(Scope scope, Operand<T> input, Class<U> outType) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Shape", scope.makeOpName("Shape"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("out_type", outType);
-    return new Shape<U>(opBuilder.build());
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new Shape operation using default output types.
-   * 
-   * @param scope current scope
-   * @param input 
-   * @return a new instance of Shape
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> Shape<TInt32> create(Scope scope, Operand<T> input) {
-    return create(scope, input, TInt32.class);
-  }
-  
-  /**
-   */
-  public Output<U> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<U> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Shape";
-  
-  private Output<U> output;
-  
-  private Shape(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ShapeN.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ShapeN.java
deleted file mode 100644
index 5d4ba217bb5..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ShapeN.java
+++ /dev/null
@@ -1,100 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import java.util.Arrays;
-import java.util.Iterator;
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.Operands;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * Returns shape of tensors.
- * <p>
- * This operation returns N 1-D integer tensors representing shape of `input[i]s`.
- * 
- * @param <U> data type for {@code output()} output
- */
-@Operator
-public final class ShapeN<U extends TNumber> extends RawOp implements Iterable<Operand<U>> {
-  
-  /**
-   * Factory method to create a class wrapping a new ShapeN operation.
-   * 
-   * @param scope current scope
-   * @param input 
-   * @param outType 
-   * @return a new instance of ShapeN
-   */
-  @Endpoint(describeByClass = true)
-  public static <U extends TNumber, T extends TType> ShapeN<U> create(Scope scope, Iterable<Operand<T>> input, Class<U> outType) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ShapeN", scope.makeOpName("ShapeN"));
-    opBuilder.addInputList(Operands.asOutputs(scope, input));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("out_type", outType);
-    return new ShapeN<U>(opBuilder.build());
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new ShapeN operation using default output types.
-   * 
-   * @param scope current scope
-   * @param input 
-   * @return a new instance of ShapeN
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> ShapeN<TInt32> create(Scope scope, Iterable<Operand<T>> input) {
-    return create(scope, input, TInt32.class);
-  }
-  
-  /**
-   */
-  public List<Output<U>> output() {
-    return output;
-  }
-  
-  @Override
-  @SuppressWarnings({"rawtypes", "unchecked"})
-  public Iterator<Operand<U>> iterator() {
-    return (Iterator) output.iterator();
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ShapeN";
-  
-  private List<Output<U>> output;
-  
-  @SuppressWarnings("unchecked")
-  private ShapeN(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    int outputLength = operation.outputListLength("output");
-    output = Arrays.asList((Output<U>[])operation.outputList(outputIdx, outputLength));
-    outputIdx += outputLength;
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Size.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Size.java
deleted file mode 100644
index db1fc6180cf..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Size.java
+++ /dev/null
@@ -1,100 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * Returns the size of a tensor.
- * <p>
- * This operation returns an integer representing the number of elements in
- * `input`.
- * <p>
- * For example:
- * <pre>{@code
- * # 't' is [[[1, 1,, 1], [2, 2, 2]], [[3, 3, 3], [4, 4, 4]]]]
- * size(t) ==> 12
- * }</pre>
- * 
- * 
- * @param <U> data type for {@code output()} output
- */
-@Operator
-public final class Size<U extends TNumber> extends RawOp implements Operand<U> {
-  
-  /**
-   * Factory method to create a class wrapping a new Size operation.
-   * 
-   * @param scope current scope
-   * @param input 
-   * @param outType 
-   * @return a new instance of Size
-   */
-  @Endpoint(describeByClass = true)
-  public static <U extends TNumber, T extends TType> Size<U> create(Scope scope, Operand<T> input, Class<U> outType) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Size", scope.makeOpName("Size"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("out_type", outType);
-    return new Size<U>(opBuilder.build());
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new Size operation using default output types.
-   * 
-   * @param scope current scope
-   * @param input 
-   * @return a new instance of Size
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> Size<TInt32> create(Scope scope, Operand<T> input) {
-    return create(scope, input, TInt32.class);
-  }
-  
-  /**
-   */
-  public Output<U> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<U> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Size";
-  
-  private Output<U> output;
-  
-  private Size(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Skipgram.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Skipgram.java
deleted file mode 100644
index 9c63d317f98..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Skipgram.java
+++ /dev/null
@@ -1,201 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.TString;
-
-/**
- * Parses a text file and creates a batch of examples.
- */
-@Operator
-public final class Skipgram extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.core.Skipgram}
-   */
-  public static class Options {
-    
-    /**
-     * @param windowSize The number of words to predict to the left and right of the target.
-     */
-    public Options windowSize(Long windowSize) {
-      this.windowSize = windowSize;
-      return this;
-    }
-    
-    /**
-     * @param minCount The minimum number of word occurrences for it to be included in the
-     * vocabulary.
-     */
-    public Options minCount(Long minCount) {
-      this.minCount = minCount;
-      return this;
-    }
-    
-    /**
-     * @param subsample Threshold for word occurrence. Words that appear with higher
-     * frequency will be randomly down-sampled. Set to 0 to disable.
-     */
-    public Options subsample(Float subsample) {
-      this.subsample = subsample;
-      return this;
-    }
-    
-    private Long windowSize;
-    private Long minCount;
-    private Float subsample;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new Skipgram operation.
-   * 
-   * @param scope current scope
-   * @param filename The corpus's text file name.
-   * @param batchSize The size of produced batch.
-   * @param options carries optional attributes values
-   * @return a new instance of Skipgram
-   */
-  @Endpoint(describeByClass = true)
-  public static Skipgram create(Scope scope, String filename, Long batchSize, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Skipgram", scope.makeOpName("Skipgram"));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("filename", filename);
-    opBuilder.setAttr("batch_size", batchSize);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.windowSize != null) {
-          opBuilder.setAttr("window_size", opts.windowSize);
-        }
-        if (opts.minCount != null) {
-          opBuilder.setAttr("min_count", opts.minCount);
-        }
-        if (opts.subsample != null) {
-          opBuilder.setAttr("subsample", opts.subsample);
-        }
-      }
-    }
-    return new Skipgram(opBuilder.build());
-  }
-  
-  /**
-   * @param windowSize The number of words to predict to the left and right of the target.
-   */
-  public static Options windowSize(Long windowSize) {
-    return new Options().windowSize(windowSize);
-  }
-  
-  /**
-   * @param minCount The minimum number of word occurrences for it to be included in the
-   * vocabulary.
-   */
-  public static Options minCount(Long minCount) {
-    return new Options().minCount(minCount);
-  }
-  
-  /**
-   * @param subsample Threshold for word occurrence. Words that appear with higher
-   * frequency will be randomly down-sampled. Set to 0 to disable.
-   */
-  public static Options subsample(Float subsample) {
-    return new Options().subsample(subsample);
-  }
-  
-  /**
-   * A vector of words in the corpus.
-   */
-  public Output<TString> vocabWord() {
-    return vocabWord;
-  }
-  
-  /**
-   * Frequencies of words. Sorted in the non-ascending order.
-   */
-  public Output<TInt32> vocabFreq() {
-    return vocabFreq;
-  }
-  
-  /**
-   * Number of words per epoch in the data file.
-   */
-  public Output<TInt64> wordsPerEpoch() {
-    return wordsPerEpoch;
-  }
-  
-  /**
-   * The current epoch number.
-   */
-  public Output<TInt32> currentEpoch() {
-    return currentEpoch;
-  }
-  
-  /**
-   * The total number of words processed so far.
-   */
-  public Output<TInt64> totalWordsProcessed() {
-    return totalWordsProcessed;
-  }
-  
-  /**
-   * A vector of word ids.
-   */
-  public Output<TInt32> examples() {
-    return examples;
-  }
-  
-  /**
-   * A vector of word ids.
-   */
-  public Output<TInt32> labels() {
-    return labels;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Skipgram";
-  
-  private Output<TString> vocabWord;
-  private Output<TInt32> vocabFreq;
-  private Output<TInt64> wordsPerEpoch;
-  private Output<TInt32> currentEpoch;
-  private Output<TInt64> totalWordsProcessed;
-  private Output<TInt32> examples;
-  private Output<TInt32> labels;
-  
-  private Skipgram(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    vocabWord = operation.output(outputIdx++);
-    vocabFreq = operation.output(outputIdx++);
-    wordsPerEpoch = operation.output(outputIdx++);
-    currentEpoch = operation.output(outputIdx++);
-    totalWordsProcessed = operation.output(outputIdx++);
-    examples = operation.output(outputIdx++);
-    labels = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Slice.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Slice.java
deleted file mode 100644
index 5b93a65307e..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Slice.java
+++ /dev/null
@@ -1,90 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * Return a slice from 'input'.
- * <p>
- * The output tensor is a tensor with dimensions described by 'size'
- * whose values are extracted from 'input' starting at the offsets in
- * 'begin'.
- * <p>
- * <i>Requirements</i>:
- *   0 <= begin[i] <= begin[i] + size[i] <= Di  for i in [0, n)
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator
-public final class Slice<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new Slice operation.
-   * 
-   * @param scope current scope
-   * @param input 
-   * @param begin begin[i] specifies the offset into the 'i'th dimension of
-   * 'input' to slice from.
-   * @param size size[i] specifies the number of elements of the 'i'th dimension
-   * of 'input' to slice. If size[i] is -1, all remaining elements in dimension
-   * i are included in the slice (i.e. this is equivalent to setting
-   * size[i] = input.dim_size(i) - begin[i]).
-   * @return a new instance of Slice
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> Slice<T> create(Scope scope, Operand<T> input, Operand<U> begin, Operand<U> size) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Slice", scope.makeOpName("Slice"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder.addInput(begin.asOutput(scope));
-    opBuilder.addInput(size.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new Slice<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Slice";
-  
-  private Output<T> output;
-  
-  private Slice(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Snapshot.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Snapshot.java
deleted file mode 100644
index 9e654bb65da..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Snapshot.java
+++ /dev/null
@@ -1,74 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Returns a copy of the input tensor.
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator
-public final class Snapshot<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new Snapshot operation.
-   * 
-   * @param scope current scope
-   * @param input 
-   * @return a new instance of Snapshot
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> Snapshot<T> create(Scope scope, Operand<T> input) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Snapshot", scope.makeOpName("Snapshot"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new Snapshot<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Snapshot";
-  
-  private Output<T> output;
-  
-  private Snapshot(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/SpaceToBatchNd.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/SpaceToBatchNd.java
deleted file mode 100644
index 11d64395abf..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/SpaceToBatchNd.java
+++ /dev/null
@@ -1,180 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * SpaceToBatch for N-D tensors of type T.
- * <p>
- * This operation divides "spatial" dimensions `[1, ..., M]` of the input into a
- * grid of blocks of shape `block_shape`, and interleaves these blocks with the
- * "batch" dimension (0) such that in the output, the spatial dimensions
- * `[1, ..., M]` correspond to the position within the grid, and the batch
- * dimension combines both the position within a spatial block and the original
- * batch position.  Prior to division into blocks, the spatial dimensions of the
- * input are optionally zero padded according to `paddings`.  See below for a
- * precise description.
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator
-public final class SpaceToBatchNd<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new SpaceToBatchNd operation.
-   * 
-   * @param scope current scope
-   * @param input N-D with shape `input_shape = [batch] + spatial_shape + remaining_shape`,
-   * where spatial_shape has `M` dimensions.
-   * @param blockShape 1-D with shape `[M]`, all values must be >= 1.
-   * @param paddings 2-D with shape `[M, 2]`, all values must be >= 0.
-   *   `paddings[i] = [pad_start, pad_end]` specifies the padding for input dimension
-   *   `i + 1`, which corresponds to spatial dimension `i`.  It is required that
-   *   `block_shape[i]` divides `input_shape[i + 1] + pad_start + pad_end`.
-   * <p>
-   * This operation is equivalent to the following steps:
-   * <p>
-   * 1. Zero-pad the start and end of dimensions `[1, ..., M]` of the
-   *    input according to `paddings` to produce `padded` of shape `padded_shape`.
-   * <p>
-   * 2. Reshape `padded` to `reshaped_padded` of shape:
-   * <p>
-   *      [batch] +
-   *      [padded_shape[1] / block_shape[0],
-   *        block_shape[0],
-   *       ...,
-   *       padded_shape[M] / block_shape[M-1],
-   *       block_shape[M-1]] +
-   *      remaining_shape
-   * <p>
-   * 3. Permute dimensions of `reshaped_padded` to produce
-   *    `permuted_reshaped_padded` of shape:
-   * <p>
-   *      block_shape +
-   *      [batch] +
-   *      [padded_shape[1] / block_shape[0],
-   *       ...,
-   *       padded_shape[M] / block_shape[M-1]] +
-   *      remaining_shape
-   * <p>
-   * 4. Reshape `permuted_reshaped_padded` to flatten `block_shape` into the batch
-   *    dimension, producing an output tensor of shape:
-   * <p>
-   *      [batch * prod(block_shape)] +
-   *      [padded_shape[1] / block_shape[0],
-   *       ...,
-   *       padded_shape[M] / block_shape[M-1]] +
-   *      remaining_shape
-   * <p>
-   * Some examples:
-   * <p>
-   * (1) For the following input of shape `[1, 2, 2, 1]`, `block_shape = [2, 2]`, and
-   *     `paddings = [[0, 0], [0, 0]]`:
-   * <pre>{@code
-   * x = [[[[1], [2]], [[3], [4]]]]
-   * }</pre>
-   * The output tensor has shape `[4, 1, 1, 1]` and value:
-   * <pre>{@code
-   * [[[[1]]], [[[2]]], [[[3]]], [[[4]]]]
-   * }</pre>
-   * (2) For the following input of shape `[1, 2, 2, 3]`, `block_shape = [2, 2]`, and
-   *     `paddings = [[0, 0], [0, 0]]`:
-   * <pre>{@code
-   * x = [[[[1, 2, 3], [4, 5, 6]],
-   *       [[7, 8, 9], [10, 11, 12]]]]
-   * }</pre>
-   * The output tensor has shape `[4, 1, 1, 3]` and value:
-   * <pre>{@code
-   * [[[[1, 2, 3]]], [[[4, 5, 6]]], [[[7, 8, 9]]], [[[10, 11, 12]]]]
-   * }</pre>
-   * (3) For the following input of shape `[1, 4, 4, 1]`, `block_shape = [2, 2]`, and
-   *     `paddings = [[0, 0], [0, 0]]`:
-   * <pre>{@code
-   * x = [[[[1],   [2],  [3],  [4]],
-   *       [[5],   [6],  [7],  [8]],
-   *       [[9],  [10], [11],  [12]],
-   *       [[13], [14], [15],  [16]]]]
-   * }</pre>
-   * The output tensor has shape `[4, 2, 2, 1]` and value:
-   * <pre>{@code
-   * x = [[[[1], [3]], [[9], [11]]],
-   *      [[[2], [4]], [[10], [12]]],
-   *      [[[5], [7]], [[13], [15]]],
-   *      [[[6], [8]], [[14], [16]]]]
-   * }</pre>
-   * (4) For the following input of shape `[2, 2, 4, 1]`, block_shape = `[2, 2]`, and
-   *     paddings = `[[0, 0], [2, 0]]`:
-   * <pre>{@code
-   * x = [[[[1],   [2],  [3],  [4]],
-   *       [[5],   [6],  [7],  [8]]],
-   *      [[[9],  [10], [11],  [12]],
-   *       [[13], [14], [15],  [16]]]]
-   * }</pre>
-   * The output tensor has shape `[8, 1, 3, 1]` and value:
-   * <pre>{@code
-   * x = [[[[0], [1], [3]]], [[[0], [9], [11]]],
-   *      [[[0], [2], [4]]], [[[0], [10], [12]]],
-   *      [[[0], [5], [7]]], [[[0], [13], [15]]],
-   *      [[[0], [6], [8]]], [[[0], [14], [16]]]]
-   * }</pre>
-   * Among others, this operation is useful for reducing atrous convolution into
-   * regular convolution.
-   * @return a new instance of SpaceToBatchNd
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber, V extends TNumber> SpaceToBatchNd<T> create(Scope scope, Operand<T> input, Operand<U> blockShape, Operand<V> paddings) {
-    OperationBuilder opBuilder = scope.env().opBuilder("SpaceToBatchND", scope.makeOpName("SpaceToBatchNd"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder.addInput(blockShape.asOutput(scope));
-    opBuilder.addInput(paddings.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new SpaceToBatchNd<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "SpaceToBatchND";
-  
-  private Output<T> output;
-  
-  private SpaceToBatchNd(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Split.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Split.java
deleted file mode 100644
index 55fbae5d9cd..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Split.java
+++ /dev/null
@@ -1,91 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import java.util.Arrays;
-import java.util.Iterator;
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.family.TType;
-
-/**
- * Splits a tensor into `num_split` tensors along one dimension.
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator
-public final class Split<T extends TType> extends RawOp implements Iterable<Operand<T>> {
-  
-  /**
-   * Factory method to create a class wrapping a new Split operation.
-   * 
-   * @param scope current scope
-   * @param axis 0-D.  The dimension along which to split.  Must be in the range
-   * `[-rank(value), rank(value))`.
-   * @param value The tensor to split.
-   * @param numSplit The number of ways to split.  Must evenly divide
-   * `value.shape[split_dim]`.
-   * @return a new instance of Split
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> Split<T> create(Scope scope, Operand<TInt32> axis, Operand<T> value, Long numSplit) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Split", scope.makeOpName("Split"));
-    opBuilder.addInput(axis.asOutput(scope));
-    opBuilder.addInput(value.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("num_split", numSplit);
-    return new Split<T>(opBuilder.build());
-  }
-  
-  /**
-   * They are identically shaped tensors, whose shape matches that of `value`
-   * except along `axis`, where their sizes are
-   * `values.shape[split_dim] / num_split`.
-   */
-  public List<Output<T>> output() {
-    return output;
-  }
-  
-  @Override
-  @SuppressWarnings({"rawtypes", "unchecked"})
-  public Iterator<Operand<T>> iterator() {
-    return (Iterator) output.iterator();
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Split";
-  
-  private List<Output<T>> output;
-  
-  @SuppressWarnings("unchecked")
-  private Split(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    int outputLength = operation.outputListLength("output");
-    output = Arrays.asList((Output<T>[])operation.outputList(outputIdx, outputLength));
-    outputIdx += outputLength;
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/SplitV.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/SplitV.java
deleted file mode 100644
index 7972f7f523f..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/SplitV.java
+++ /dev/null
@@ -1,95 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import java.util.Arrays;
-import java.util.Iterator;
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * Splits a tensor into `num_split` tensors along one dimension.
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator
-public final class SplitV<T extends TType> extends RawOp implements Iterable<Operand<T>> {
-  
-  /**
-   * Factory method to create a class wrapping a new SplitV operation.
-   * 
-   * @param scope current scope
-   * @param value The tensor to split.
-   * @param sizeSplits list containing the sizes of each output tensor along the split
-   * dimension. Must sum to the dimension of value along split_dim.
-   * Can contain one -1 indicating that dimension is to be inferred.
-   * @param axis 0-D.  The dimension along which to split.  Must be in the range
-   * `[-rank(value), rank(value))`.
-   * @param numSplit 
-   * @return a new instance of SplitV
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> SplitV<T> create(Scope scope, Operand<T> value, Operand<U> sizeSplits, Operand<TInt32> axis, Long numSplit) {
-    OperationBuilder opBuilder = scope.env().opBuilder("SplitV", scope.makeOpName("SplitV"));
-    opBuilder.addInput(value.asOutput(scope));
-    opBuilder.addInput(sizeSplits.asOutput(scope));
-    opBuilder.addInput(axis.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("num_split", numSplit);
-    return new SplitV<T>(opBuilder.build());
-  }
-  
-  /**
-   * Tensors whose shape matches that of `value`
-   * except along `axis`, where their sizes are
-   * `size_splits[i]`.
-   */
-  public List<Output<T>> output() {
-    return output;
-  }
-  
-  @Override
-  @SuppressWarnings({"rawtypes", "unchecked"})
-  public Iterator<Operand<T>> iterator() {
-    return (Iterator) output.iterator();
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "SplitV";
-  
-  private List<Output<T>> output;
-  
-  @SuppressWarnings("unchecked")
-  private SplitV(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    int outputLength = operation.outputListLength("output");
-    output = Arrays.asList((Output<T>[])operation.outputList(outputIdx, outputLength));
-    outputIdx += outputLength;
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Squeeze.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Squeeze.java
deleted file mode 100644
index 4eb9838a298..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Squeeze.java
+++ /dev/null
@@ -1,136 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Removes dimensions of size 1 from the shape of a tensor.
- * <p>
- * Given a tensor `input`, this operation returns a tensor of the same type with
- * all dimensions of size 1 removed. If you don't want to remove all size 1
- * dimensions, you can remove specific size 1 dimensions by specifying
- * `axis`.
- * <p>
- * For example:
- * <pre>{@code
- * # 't' is a tensor of shape [1, 2, 1, 3, 1, 1]
- * shape(squeeze(t)) ==> [2, 3]
- * }</pre>
- * Or, to remove specific size 1 dimensions:
- * <pre>{@code
- * # 't' is a tensor of shape [1, 2, 1, 3, 1, 1]
- * shape(squeeze(t, [2, 4])) ==> [1, 2, 3, 1]
- * }</pre>
- * 
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator
-public final class Squeeze<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.core.Squeeze}
-   */
-  public static class Options {
-    
-    /**
-     * @param axis If specified, only squeezes the dimensions listed. The dimension
-     * index starts at 0. It is an error to squeeze a dimension that is not 1. Must
-     * be in the range `[-rank(input), rank(input))`.
-     */
-    public Options axis(List<Long> axis) {
-      this.axis = axis;
-      return this;
-    }
-    
-    private List<Long> axis;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new Squeeze operation.
-   * 
-   * @param scope current scope
-   * @param input The `input` to squeeze.
-   * @param options carries optional attributes values
-   * @return a new instance of Squeeze
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> Squeeze<T> create(Scope scope, Operand<T> input, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Squeeze", scope.makeOpName("Squeeze"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.axis != null) {
-          long[] axisArray = new long[opts.axis.size()];
-          for (int i = 0; i < axisArray.length; ++i) {
-            axisArray[i] = opts.axis.get(i);
-          }
-          opBuilder.setAttr("squeeze_dims", axisArray);
-        }
-      }
-    }
-    return new Squeeze<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param axis If specified, only squeezes the dimensions listed. The dimension
-   * index starts at 0. It is an error to squeeze a dimension that is not 1. Must
-   * be in the range `[-rank(input), rank(input))`.
-   */
-  public static Options axis(List<Long> axis) {
-    return new Options().axis(axis);
-  }
-  
-  /**
-   * Contains the same data as `input`, but has one or more dimensions of
-   * size 1 removed.
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Squeeze";
-  
-  private Output<T> output;
-  
-  private Squeeze(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Stack.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Stack.java
deleted file mode 100644
index 0b68d56ff7e..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Stack.java
+++ /dev/null
@@ -1,130 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.Operands;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Packs a list of `N` rank-`R` tensors into one rank-`(R+1)` tensor.
- * <p>
- * Packs the `N` tensors in `values` into a tensor with rank one higher than each
- * tensor in `values`, by packing them along the `axis` dimension.
- * Given a list of tensors of shape `(A, B, C)`;
- * <p>
- * if `axis == 0` then the `output` tensor will have the shape `(N, A, B, C)`.
- * if `axis == 1` then the `output` tensor will have the shape `(A, N, B, C)`.
- * Etc.
- * <p>
- * For example:
- * <pre>{@code
- * # 'x' is [1, 4]
- * # 'y' is [2, 5]
- * # 'z' is [3, 6]
- * pack([x, y, z]) => [[1, 4], [2, 5], [3, 6]]  # Pack along first dim.
- * pack([x, y, z], axis=1) => [[1, 2, 3], [4, 5, 6]]
- * }</pre>
- * This is the opposite of `unpack`.
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator
-public final class Stack<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.core.Stack}
-   */
-  public static class Options {
-    
-    /**
-     * @param axis Dimension along which to pack.  Negative values wrap around, so the
-     * valid range is `[-(R+1), R+1)`.
-     */
-    public Options axis(Long axis) {
-      this.axis = axis;
-      return this;
-    }
-    
-    private Long axis;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new Stack operation.
-   * 
-   * @param scope current scope
-   * @param values Must be of same shape and type.
-   * @param options carries optional attributes values
-   * @return a new instance of Stack
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> Stack<T> create(Scope scope, Iterable<Operand<T>> values, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Pack", scope.makeOpName("Stack"));
-    opBuilder.addInputList(Operands.asOutputs(scope, values));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.axis != null) {
-          opBuilder.setAttr("axis", opts.axis);
-        }
-      }
-    }
-    return new Stack<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param axis Dimension along which to pack.  Negative values wrap around, so the
-   * valid range is `[-(R+1), R+1)`.
-   */
-  public static Options axis(Long axis) {
-    return new Options().axis(axis);
-  }
-  
-  /**
-   * The packed tensor.
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Pack";
-  
-  private Output<T> output;
-  
-  private Stack(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Stage.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Stage.java
deleted file mode 100644
index ee809f3e79f..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Stage.java
+++ /dev/null
@@ -1,157 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.Operands;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-
-/**
- * Stage values similar to a lightweight Enqueue.
- * <p>
- * The basic functionality of this Op is similar to a queue with many
- * fewer capabilities and options.  This Op is optimized for performance.
- */
-@Operator
-public final class Stage extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.core.Stage}
-   */
-  public static class Options {
-    
-    /**
-     * @param capacity Maximum number of elements in the Staging Area. If > 0, inserts
-     * on the container will block when the capacity is reached.
-     */
-    public Options capacity(Long capacity) {
-      this.capacity = capacity;
-      return this;
-    }
-    
-    /**
-     * @param memoryLimit The maximum number of bytes allowed for Tensors in the Staging Area.
-     * If > 0, inserts will block until sufficient space is available.
-     */
-    public Options memoryLimit(Long memoryLimit) {
-      this.memoryLimit = memoryLimit;
-      return this;
-    }
-    
-    /**
-     * @param container If non-empty, this queue is placed in the given container. Otherwise,
-     * a default container is used.
-     */
-    public Options container(String container) {
-      this.container = container;
-      return this;
-    }
-    
-    /**
-     * @param sharedName It is necessary to match this name to the matching Unstage Op.
-     */
-    public Options sharedName(String sharedName) {
-      this.sharedName = sharedName;
-      return this;
-    }
-    
-    private Long capacity;
-    private Long memoryLimit;
-    private String container;
-    private String sharedName;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new Stage operation.
-   * 
-   * @param scope current scope
-   * @param values a list of tensors
-   * dtypes A list of data types that inserted values should adhere to.
-   * @param options carries optional attributes values
-   * @return a new instance of Stage
-   */
-  @Endpoint(describeByClass = true)
-  public static Stage create(Scope scope, Iterable<Operand<?>> values, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Stage", scope.makeOpName("Stage"));
-    opBuilder.addInputList(Operands.asOutputs(scope, values));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.capacity != null) {
-          opBuilder.setAttr("capacity", opts.capacity);
-        }
-        if (opts.memoryLimit != null) {
-          opBuilder.setAttr("memory_limit", opts.memoryLimit);
-        }
-        if (opts.container != null) {
-          opBuilder.setAttr("container", opts.container);
-        }
-        if (opts.sharedName != null) {
-          opBuilder.setAttr("shared_name", opts.sharedName);
-        }
-      }
-    }
-    return new Stage(opBuilder.build());
-  }
-  
-  /**
-   * @param capacity Maximum number of elements in the Staging Area. If > 0, inserts
-   * on the container will block when the capacity is reached.
-   */
-  public static Options capacity(Long capacity) {
-    return new Options().capacity(capacity);
-  }
-  
-  /**
-   * @param memoryLimit The maximum number of bytes allowed for Tensors in the Staging Area.
-   * If > 0, inserts will block until sufficient space is available.
-   */
-  public static Options memoryLimit(Long memoryLimit) {
-    return new Options().memoryLimit(memoryLimit);
-  }
-  
-  /**
-   * @param container If non-empty, this queue is placed in the given container. Otherwise,
-   * a default container is used.
-   */
-  public static Options container(String container) {
-    return new Options().container(container);
-  }
-  
-  /**
-   * @param sharedName It is necessary to match this name to the matching Unstage Op.
-   */
-  public static Options sharedName(String sharedName) {
-    return new Options().sharedName(sharedName);
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Stage";
-  
-  private Stage(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/StageClear.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/StageClear.java
deleted file mode 100644
index 11175fc9e4d..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/StageClear.java
+++ /dev/null
@@ -1,150 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import java.util.List;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-
-/**
- * Op removes all elements in the underlying container.
- */
-@Operator
-public final class StageClear extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.core.StageClear}
-   */
-  public static class Options {
-    
-    /**
-     * @param capacity 
-     */
-    public Options capacity(Long capacity) {
-      this.capacity = capacity;
-      return this;
-    }
-    
-    /**
-     * @param memoryLimit 
-     */
-    public Options memoryLimit(Long memoryLimit) {
-      this.memoryLimit = memoryLimit;
-      return this;
-    }
-    
-    /**
-     * @param container 
-     */
-    public Options container(String container) {
-      this.container = container;
-      return this;
-    }
-    
-    /**
-     * @param sharedName 
-     */
-    public Options sharedName(String sharedName) {
-      this.sharedName = sharedName;
-      return this;
-    }
-    
-    private Long capacity;
-    private Long memoryLimit;
-    private String container;
-    private String sharedName;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new StageClear operation.
-   * 
-   * @param scope current scope
-   * @param dtypes 
-   * @param options carries optional attributes values
-   * @return a new instance of StageClear
-   */
-  @Endpoint(describeByClass = true)
-  public static StageClear create(Scope scope, List<Class<?>> dtypes, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("StageClear", scope.makeOpName("StageClear"));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    Class[] dtypesArray = new Class[dtypes.size()];
-    for (int i = 0; i < dtypesArray.length; ++i) {
-      dtypesArray[i] = dtypes.get(i);
-    }
-    opBuilder.setAttr("dtypes", dtypesArray);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.capacity != null) {
-          opBuilder.setAttr("capacity", opts.capacity);
-        }
-        if (opts.memoryLimit != null) {
-          opBuilder.setAttr("memory_limit", opts.memoryLimit);
-        }
-        if (opts.container != null) {
-          opBuilder.setAttr("container", opts.container);
-        }
-        if (opts.sharedName != null) {
-          opBuilder.setAttr("shared_name", opts.sharedName);
-        }
-      }
-    }
-    return new StageClear(opBuilder.build());
-  }
-  
-  /**
-   * @param capacity 
-   */
-  public static Options capacity(Long capacity) {
-    return new Options().capacity(capacity);
-  }
-  
-  /**
-   * @param memoryLimit 
-   */
-  public static Options memoryLimit(Long memoryLimit) {
-    return new Options().memoryLimit(memoryLimit);
-  }
-  
-  /**
-   * @param container 
-   */
-  public static Options container(String container) {
-    return new Options().container(container);
-  }
-  
-  /**
-   * @param sharedName 
-   */
-  public static Options sharedName(String sharedName) {
-    return new Options().sharedName(sharedName);
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "StageClear";
-  
-  private StageClear(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/StagePeek.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/StagePeek.java
deleted file mode 100644
index 646d721a4e9..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/StagePeek.java
+++ /dev/null
@@ -1,180 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import java.util.Arrays;
-import java.util.Iterator;
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.family.TType;
-
-/**
- * Op peeks at the values at the specified index.  If the
- * <p>
- * underlying container does not contain sufficient elements
- * this op will block until it does.   This Op is optimized for
- * performance.
- */
-@Operator
-public final class StagePeek extends RawOp implements Iterable<Operand<TType>> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.core.StagePeek}
-   */
-  public static class Options {
-    
-    /**
-     * @param capacity 
-     */
-    public Options capacity(Long capacity) {
-      this.capacity = capacity;
-      return this;
-    }
-    
-    /**
-     * @param memoryLimit 
-     */
-    public Options memoryLimit(Long memoryLimit) {
-      this.memoryLimit = memoryLimit;
-      return this;
-    }
-    
-    /**
-     * @param container 
-     */
-    public Options container(String container) {
-      this.container = container;
-      return this;
-    }
-    
-    /**
-     * @param sharedName 
-     */
-    public Options sharedName(String sharedName) {
-      this.sharedName = sharedName;
-      return this;
-    }
-    
-    private Long capacity;
-    private Long memoryLimit;
-    private String container;
-    private String sharedName;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new StagePeek operation.
-   * 
-   * @param scope current scope
-   * @param index 
-   * @param dtypes 
-   * @param options carries optional attributes values
-   * @return a new instance of StagePeek
-   */
-  @Endpoint(describeByClass = true)
-  public static StagePeek create(Scope scope, Operand<TInt32> index, List<Class<?>> dtypes, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("StagePeek", scope.makeOpName("StagePeek"));
-    opBuilder.addInput(index.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    Class[] dtypesArray = new Class[dtypes.size()];
-    for (int i = 0; i < dtypesArray.length; ++i) {
-      dtypesArray[i] = dtypes.get(i);
-    }
-    opBuilder.setAttr("dtypes", dtypesArray);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.capacity != null) {
-          opBuilder.setAttr("capacity", opts.capacity);
-        }
-        if (opts.memoryLimit != null) {
-          opBuilder.setAttr("memory_limit", opts.memoryLimit);
-        }
-        if (opts.container != null) {
-          opBuilder.setAttr("container", opts.container);
-        }
-        if (opts.sharedName != null) {
-          opBuilder.setAttr("shared_name", opts.sharedName);
-        }
-      }
-    }
-    return new StagePeek(opBuilder.build());
-  }
-  
-  /**
-   * @param capacity 
-   */
-  public static Options capacity(Long capacity) {
-    return new Options().capacity(capacity);
-  }
-  
-  /**
-   * @param memoryLimit 
-   */
-  public static Options memoryLimit(Long memoryLimit) {
-    return new Options().memoryLimit(memoryLimit);
-  }
-  
-  /**
-   * @param container 
-   */
-  public static Options container(String container) {
-    return new Options().container(container);
-  }
-  
-  /**
-   * @param sharedName 
-   */
-  public static Options sharedName(String sharedName) {
-    return new Options().sharedName(sharedName);
-  }
-  
-  /**
-   */
-  public List<Output<?>> values() {
-    return values;
-  }
-  
-  @Override
-  @SuppressWarnings({"rawtypes", "unchecked"})
-  public Iterator<Operand<TType>> iterator() {
-    return (Iterator) values.iterator();
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "StagePeek";
-  
-  private List<Output<?>> values;
-  
-  private StagePeek(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    int valuesLength = operation.outputListLength("values");
-    values = Arrays.asList(operation.outputList(outputIdx, valuesLength));
-    outputIdx += valuesLength;
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/StageSize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/StageSize.java
deleted file mode 100644
index b4d06a08bb4..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/StageSize.java
+++ /dev/null
@@ -1,168 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
-
-/**
- * Op returns the number of elements in the underlying container.
- */
-@Operator
-public final class StageSize extends RawOp implements Operand<TInt32> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.core.StageSize}
-   */
-  public static class Options {
-    
-    /**
-     * @param capacity 
-     */
-    public Options capacity(Long capacity) {
-      this.capacity = capacity;
-      return this;
-    }
-    
-    /**
-     * @param memoryLimit 
-     */
-    public Options memoryLimit(Long memoryLimit) {
-      this.memoryLimit = memoryLimit;
-      return this;
-    }
-    
-    /**
-     * @param container 
-     */
-    public Options container(String container) {
-      this.container = container;
-      return this;
-    }
-    
-    /**
-     * @param sharedName 
-     */
-    public Options sharedName(String sharedName) {
-      this.sharedName = sharedName;
-      return this;
-    }
-    
-    private Long capacity;
-    private Long memoryLimit;
-    private String container;
-    private String sharedName;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new StageSize operation.
-   * 
-   * @param scope current scope
-   * @param dtypes 
-   * @param options carries optional attributes values
-   * @return a new instance of StageSize
-   */
-  @Endpoint(describeByClass = true)
-  public static StageSize create(Scope scope, List<Class<?>> dtypes, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("StageSize", scope.makeOpName("StageSize"));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    Class[] dtypesArray = new Class[dtypes.size()];
-    for (int i = 0; i < dtypesArray.length; ++i) {
-      dtypesArray[i] = dtypes.get(i);
-    }
-    opBuilder.setAttr("dtypes", dtypesArray);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.capacity != null) {
-          opBuilder.setAttr("capacity", opts.capacity);
-        }
-        if (opts.memoryLimit != null) {
-          opBuilder.setAttr("memory_limit", opts.memoryLimit);
-        }
-        if (opts.container != null) {
-          opBuilder.setAttr("container", opts.container);
-        }
-        if (opts.sharedName != null) {
-          opBuilder.setAttr("shared_name", opts.sharedName);
-        }
-      }
-    }
-    return new StageSize(opBuilder.build());
-  }
-  
-  /**
-   * @param capacity 
-   */
-  public static Options capacity(Long capacity) {
-    return new Options().capacity(capacity);
-  }
-  
-  /**
-   * @param memoryLimit 
-   */
-  public static Options memoryLimit(Long memoryLimit) {
-    return new Options().memoryLimit(memoryLimit);
-  }
-  
-  /**
-   * @param container 
-   */
-  public static Options container(String container) {
-    return new Options().container(container);
-  }
-  
-  /**
-   * @param sharedName 
-   */
-  public static Options sharedName(String sharedName) {
-    return new Options().sharedName(sharedName);
-  }
-  
-  /**
-   */
-  public Output<TInt32> size() {
-    return size;
-  }
-  
-  @Override
-  public Output<TInt32> asOutput(Scope scope) {
-    return size;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "StageSize";
-  
-  private Output<TInt32> size;
-  
-  private StageSize(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    size = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/StopGradient.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/StopGradient.java
deleted file mode 100644
index f57784f8f1f..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/StopGradient.java
+++ /dev/null
@@ -1,99 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Stops gradient computation.
- * <p>
- * When executed in a graph, this op outputs its input tensor as-is.
- * <p>
- * When building ops to compute gradients, this op prevents the contribution of
- * its inputs to be taken into account.  Normally, the gradient generator adds ops
- * to a graph to compute the derivatives of a specified 'loss' by recursively
- * finding out inputs that contributed to its computation.  If you insert this op
- * in the graph it inputs are masked from the gradient generator.  They are not
- * taken into account for computing gradients.
- * <p>
- * This is useful any time you want to compute a value with TensorFlow but need
- * to pretend that the value was a constant. Some examples include:
- * <ul>
- * <li>
- * The <i>EM</i> algorithm where the <i>M-step</i> should not involve backpropagation
- *    through the output of the <i>E-step</i>.
- * </li>
- * <li>
- * Contrastive divergence training of Boltzmann machines where, when
- *    differentiating the energy function, the training must not backpropagate
- *    through the graph that generated the samples from the model.
- * </li>
- * <li>
- * Adversarial training, where no backprop should happen through the adversarial
- *    example generation process.
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator
-public final class StopGradient<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new StopGradient operation.
-   * 
-   * @param scope current scope
-   * @param input 
-   * @return a new instance of StopGradient
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> StopGradient<T> create(Scope scope, Operand<T> input) {
-    OperationBuilder opBuilder = scope.env().opBuilder("StopGradient", scope.makeOpName("StopGradient"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new StopGradient<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "StopGradient";
-  
-  private Output<T> output;
-  
-  private StopGradient(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/StridedSlice.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/StridedSlice.java
deleted file mode 100644
index 5fb0ee67946..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/StridedSlice.java
+++ /dev/null
@@ -1,317 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * Return a strided slice from `input`.
- * <p>
- * Note, most python users will want to use the Python `Tensor.__getitem__`
- * or `Variable.__getitem__` rather than this op directly.
- * <p>
- * The goal of this op is to produce a new tensor with a subset of
- * the elements from the `n` dimensional `input` tensor. The subset is chosen using
- * a sequence of `m` sparse range specifications encoded into the arguments
- * of this function. Note, in some cases
- * `m` could be equal to `n`, but this need not be the case. Each
- * range specification entry can be one of the following:
- * <p>
- * - An ellipsis (...). Ellipses are used to imply zero or more
- *   dimensions of full-dimension selection and are produced using
- *   `ellipsis_mask`. For example, `foo[...]` is the identity slice.
- * <p>
- * - A new axis. This is used to insert a new shape=1 dimension and is
- *   produced using `new_axis_mask`. For example, `foo[:, ...]` where
- *   `foo` is shape `(3, 4)` produces a `(1, 3, 4)` tensor.
- * <p>
- * - A range `begin:end:stride`. This is used to specify how much to choose from
- *   a given dimension. `stride` can be any integer but 0.  `begin` is an integer
- *   which represents the index of the first value to select while `end` represents
- *   the index of the last value to select. The number of values selected in each
- *   dimension is `end - begin` if `stride > 0` and `begin - end` if `stride < 0`.
- *   `begin` and `end` can be negative where `-1` is the last element, `-2` is
- *   the second to last. `begin_mask` controls whether to replace the explicitly
- *   given `begin` with an implicit effective value of `0` if `stride > 0` and
- *   `-1` if `stride < 0`. `end_mask` is analogous but produces the number
- *   required to create the largest open interval. For example, given a shape
- *   `(3,)` tensor `foo[:]`, the effective `begin` and `end` are `0` and `3`. Do
- *   not assume this is equivalent to `foo[0:-1]` which has an effective `begin`
- *   and `end` of `0` and `2`. Another example is `foo[-2::-1]` which reverses the
- *   first dimension of a tensor while dropping the last two (in the original
- *   order elements). For example `foo = [1,2,3,4]; foo[-2::-1]` is `[4,3]`.
- * <p>
- * - A single index. This is used to keep only elements that have a given
- *   index. For example (`foo[2, :]` on a shape `(5,6)` tensor produces a
- *   shape `(6,)` tensor. This is encoded in `begin` and `end` and
- *   `shrink_axis_mask`.
- * <p>
- * Each conceptual range specification is encoded in the op's argument. This
- * encoding is best understand by considering a non-trivial example. In
- * particular,
- * `foo[1, 2:4, None, ..., :-3:-1, :]` will be encoded as
- * <pre>{@code
- * begin = [1, 2, x, x, 0, x] # x denotes don't care (usually 0)
- * end = [2, 4, x, x, -3, x]
- * strides = [1, 1, x, x, -1, 1]
- * begin_mask = 1<<4 | 1<<5 = 48
- * end_mask = 1<<5 = 32
- * ellipsis_mask = 1<<3 = 8
- * new_axis_mask = 1<<2 = 4
- * shrink_axis_mask = 1<<0 = 1
- * }</pre>
- * In this case if `foo.shape` is (5, 5, 5, 5, 5, 5) the final shape of
- * the slice becomes (2, 1, 5, 5, 2, 5).
- * Let us walk step by step through each argument specification.
- * <p>
- * 1.  The first argument in the example slice is turned into `begin = 1` and
- * `end = begin + 1 = 2`. To disambiguate from the original spec `2:4` we
- * also set the appropriate bit in `shrink_axis_mask`.
- * <p>
- * 2. `2:4` is contributes 2, 4, 1 to begin, end, and stride. All masks have
- * zero bits contributed.
- * <p>
- * 3. None is a synonym for `tf.newaxis`. This means insert a dimension of size 1
- * dimension in the final shape. Dummy values are contributed to begin,
- * end and stride, while the new_axis_mask bit is set.
- * <p>
- * 4. `...` grab the full ranges from as many dimensions as needed to
- * fully specify a slice for every dimension of the input shape.
- * <p>
- * 5. `:-3:-1` shows the use of negative indices. A negative index `i` associated
- * with a dimension that has shape `s` is converted to a positive index
- * `s + i`. So `-1` becomes `s-1` (i.e. the last element). This conversion
- * is done internally so begin, end and strides receive x, -3, and -1.
- * The appropriate begin_mask bit is set to indicate the start range is the
- * full range (ignoring the x).
- * <p>
- * 6. `:` indicates that the entire contents of the corresponding dimension
- * is selected. This is equivalent to `::` or `0::1`. begin, end, and strides
- * receive 0, 0, and 1, respectively. The appropriate bits in `begin_mask` and
- * `end_mask` are also set.
- * <p>
- * <i>Requirements</i>:
- *   `0 != strides[i] for i in [0, m)`
- *   `ellipsis_mask must be a power of two (only one ellipsis)`
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator
-public final class StridedSlice<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.core.StridedSlice}
-   */
-  public static class Options {
-    
-    /**
-     * @param beginMask a bitmask where a bit i being 1 means to ignore the begin
-     * value and instead use the largest interval possible. At runtime
-     * begin[i] will be replaced with `[0, n-1)` if `stride[i] > 0` or
-     * `[-1, n-1]` if `stride[i] < 0`
-     */
-    public Options beginMask(Long beginMask) {
-      this.beginMask = beginMask;
-      return this;
-    }
-    
-    /**
-     * @param endMask analogous to `begin_mask`
-     */
-    public Options endMask(Long endMask) {
-      this.endMask = endMask;
-      return this;
-    }
-    
-    /**
-     * @param ellipsisMask a bitmask where bit `i` being 1 means the `i`th
-     * position is actually an ellipsis. One bit at most can be 1.
-     * If `ellipsis_mask == 0`, then an implicit ellipsis mask of `1 << (m+1)`
-     * is provided. This means that `foo[3:5] == foo[3:5, ...]`. An ellipsis
-     * implicitly creates as many range specifications as necessary to fully
-     * specify the sliced range for every dimension. For example for a 4-dimensional
-     * tensor `foo` the slice `foo[2, ..., 5:8]` implies `foo[2, :, :, 5:8]`.
-     */
-    public Options ellipsisMask(Long ellipsisMask) {
-      this.ellipsisMask = ellipsisMask;
-      return this;
-    }
-    
-    /**
-     * @param newAxisMask a bitmask where bit `i` being 1 means the `i`th
-     * specification creates a new shape 1 dimension. For example
-     * `foo[:4, tf.newaxis, :2]` would produce a shape `(4, 1, 2)` tensor.
-     */
-    public Options newAxisMask(Long newAxisMask) {
-      this.newAxisMask = newAxisMask;
-      return this;
-    }
-    
-    /**
-     * @param shrinkAxisMask a bitmask where bit `i` implies that the `i`th
-     * specification should shrink the dimensionality. begin and end
-     * must imply a slice of size 1 in the dimension. For example in
-     * python one might do `foo[:, 3, :]` which would result in
-     * `shrink_axis_mask` being 2.
-     */
-    public Options shrinkAxisMask(Long shrinkAxisMask) {
-      this.shrinkAxisMask = shrinkAxisMask;
-      return this;
-    }
-    
-    private Long beginMask;
-    private Long endMask;
-    private Long ellipsisMask;
-    private Long newAxisMask;
-    private Long shrinkAxisMask;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new StridedSlice operation.
-   * 
-   * @param scope current scope
-   * @param input 
-   * @param begin `begin[k]` specifies the offset into the `k`th range specification.
-   * The exact dimension this corresponds to will be determined by context.
-   * Out-of-bounds values will be silently clamped. If the `k`th bit of
-   * `begin_mask` then `begin[k]` is ignored and the full range of the
-   * appropriate dimension is used instead. Negative values causes indexing
-   * to start from the highest element e.g. If `foo==[1,2,3]` then `foo[-1]==3`.
-   * @param end `end[i]` is like `begin` with the exception that `end_mask` is
-   * used to determine full ranges.
-   * @param strides `strides[i]` specifies the increment in the `i`th specification
-   * after extracting a given element. Negative indices will reverse
-   * the original order. Out or range values are
-   * clamped to `[0,dim[i]) if slice[i]>0` or `[-1,dim[i]-1] if slice[i] < 0`
-   * @param options carries optional attributes values
-   * @return a new instance of StridedSlice
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> StridedSlice<T> create(Scope scope, Operand<T> input, Operand<U> begin, Operand<U> end, Operand<U> strides, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("StridedSlice", scope.makeOpName("StridedSlice"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder.addInput(begin.asOutput(scope));
-    opBuilder.addInput(end.asOutput(scope));
-    opBuilder.addInput(strides.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.beginMask != null) {
-          opBuilder.setAttr("begin_mask", opts.beginMask);
-        }
-        if (opts.endMask != null) {
-          opBuilder.setAttr("end_mask", opts.endMask);
-        }
-        if (opts.ellipsisMask != null) {
-          opBuilder.setAttr("ellipsis_mask", opts.ellipsisMask);
-        }
-        if (opts.newAxisMask != null) {
-          opBuilder.setAttr("new_axis_mask", opts.newAxisMask);
-        }
-        if (opts.shrinkAxisMask != null) {
-          opBuilder.setAttr("shrink_axis_mask", opts.shrinkAxisMask);
-        }
-      }
-    }
-    return new StridedSlice<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param beginMask a bitmask where a bit i being 1 means to ignore the begin
-   * value and instead use the largest interval possible. At runtime
-   * begin[i] will be replaced with `[0, n-1)` if `stride[i] > 0` or
-   * `[-1, n-1]` if `stride[i] < 0`
-   */
-  public static Options beginMask(Long beginMask) {
-    return new Options().beginMask(beginMask);
-  }
-  
-  /**
-   * @param endMask analogous to `begin_mask`
-   */
-  public static Options endMask(Long endMask) {
-    return new Options().endMask(endMask);
-  }
-  
-  /**
-   * @param ellipsisMask a bitmask where bit `i` being 1 means the `i`th
-   * position is actually an ellipsis. One bit at most can be 1.
-   * If `ellipsis_mask == 0`, then an implicit ellipsis mask of `1 << (m+1)`
-   * is provided. This means that `foo[3:5] == foo[3:5, ...]`. An ellipsis
-   * implicitly creates as many range specifications as necessary to fully
-   * specify the sliced range for every dimension. For example for a 4-dimensional
-   * tensor `foo` the slice `foo[2, ..., 5:8]` implies `foo[2, :, :, 5:8]`.
-   */
-  public static Options ellipsisMask(Long ellipsisMask) {
-    return new Options().ellipsisMask(ellipsisMask);
-  }
-  
-  /**
-   * @param newAxisMask a bitmask where bit `i` being 1 means the `i`th
-   * specification creates a new shape 1 dimension. For example
-   * `foo[:4, tf.newaxis, :2]` would produce a shape `(4, 1, 2)` tensor.
-   */
-  public static Options newAxisMask(Long newAxisMask) {
-    return new Options().newAxisMask(newAxisMask);
-  }
-  
-  /**
-   * @param shrinkAxisMask a bitmask where bit `i` implies that the `i`th
-   * specification should shrink the dimensionality. begin and end
-   * must imply a slice of size 1 in the dimension. For example in
-   * python one might do `foo[:, 3, :]` which would result in
-   * `shrink_axis_mask` being 2.
-   */
-  public static Options shrinkAxisMask(Long shrinkAxisMask) {
-    return new Options().shrinkAxisMask(shrinkAxisMask);
-  }
-  
-  /**
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "StridedSlice";
-  
-  private Output<T> output;
-  
-  private StridedSlice(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/StridedSliceAssign.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/StridedSliceAssign.java
deleted file mode 100644
index 0cb552ea895..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/StridedSliceAssign.java
+++ /dev/null
@@ -1,200 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * Assign `value` to the sliced l-value reference of `ref`.
- * <p>
- * The values of `value` are assigned to the positions in the variable
- * `ref` that are selected by the slice parameters. The slice parameters
- * `begin`, `end`, `strides`, etc. work exactly as in `StridedSlice`.
- * <p>
- * NOTE this op currently does not support broadcasting and so `value`'s
- * shape must be exactly the shape produced by the slice of `ref`.
- * 
- * @param <T> data type for {@code outputRef()} output
- */
-@Operator
-public final class StridedSliceAssign<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.core.StridedSliceAssign}
-   */
-  public static class Options {
-    
-    /**
-     * @param beginMask 
-     */
-    public Options beginMask(Long beginMask) {
-      this.beginMask = beginMask;
-      return this;
-    }
-    
-    /**
-     * @param endMask 
-     */
-    public Options endMask(Long endMask) {
-      this.endMask = endMask;
-      return this;
-    }
-    
-    /**
-     * @param ellipsisMask 
-     */
-    public Options ellipsisMask(Long ellipsisMask) {
-      this.ellipsisMask = ellipsisMask;
-      return this;
-    }
-    
-    /**
-     * @param newAxisMask 
-     */
-    public Options newAxisMask(Long newAxisMask) {
-      this.newAxisMask = newAxisMask;
-      return this;
-    }
-    
-    /**
-     * @param shrinkAxisMask 
-     */
-    public Options shrinkAxisMask(Long shrinkAxisMask) {
-      this.shrinkAxisMask = shrinkAxisMask;
-      return this;
-    }
-    
-    private Long beginMask;
-    private Long endMask;
-    private Long ellipsisMask;
-    private Long newAxisMask;
-    private Long shrinkAxisMask;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new StridedSliceAssign operation.
-   * 
-   * @param scope current scope
-   * @param ref 
-   * @param begin 
-   * @param end 
-   * @param strides 
-   * @param value 
-   * @param options carries optional attributes values
-   * @return a new instance of StridedSliceAssign
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> StridedSliceAssign<T> create(Scope scope, Operand<T> ref, Operand<U> begin, Operand<U> end, Operand<U> strides, Operand<T> value, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("StridedSliceAssign", scope.makeOpName("StridedSliceAssign"));
-    opBuilder.addInput(ref.asOutput(scope));
-    opBuilder.addInput(begin.asOutput(scope));
-    opBuilder.addInput(end.asOutput(scope));
-    opBuilder.addInput(strides.asOutput(scope));
-    opBuilder.addInput(value.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.beginMask != null) {
-          opBuilder.setAttr("begin_mask", opts.beginMask);
-        }
-        if (opts.endMask != null) {
-          opBuilder.setAttr("end_mask", opts.endMask);
-        }
-        if (opts.ellipsisMask != null) {
-          opBuilder.setAttr("ellipsis_mask", opts.ellipsisMask);
-        }
-        if (opts.newAxisMask != null) {
-          opBuilder.setAttr("new_axis_mask", opts.newAxisMask);
-        }
-        if (opts.shrinkAxisMask != null) {
-          opBuilder.setAttr("shrink_axis_mask", opts.shrinkAxisMask);
-        }
-      }
-    }
-    return new StridedSliceAssign<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param beginMask 
-   */
-  public static Options beginMask(Long beginMask) {
-    return new Options().beginMask(beginMask);
-  }
-  
-  /**
-   * @param endMask 
-   */
-  public static Options endMask(Long endMask) {
-    return new Options().endMask(endMask);
-  }
-  
-  /**
-   * @param ellipsisMask 
-   */
-  public static Options ellipsisMask(Long ellipsisMask) {
-    return new Options().ellipsisMask(ellipsisMask);
-  }
-  
-  /**
-   * @param newAxisMask 
-   */
-  public static Options newAxisMask(Long newAxisMask) {
-    return new Options().newAxisMask(newAxisMask);
-  }
-  
-  /**
-   * @param shrinkAxisMask 
-   */
-  public static Options shrinkAxisMask(Long shrinkAxisMask) {
-    return new Options().shrinkAxisMask(shrinkAxisMask);
-  }
-  
-  /**
-   */
-  public Output<T> outputRef() {
-    return outputRef;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return outputRef;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "StridedSliceAssign";
-  
-  private Output<T> outputRef;
-  
-  private StridedSliceAssign(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    outputRef = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/StridedSliceGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/StridedSliceGrad.java
deleted file mode 100644
index bbf0e022ca6..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/StridedSliceGrad.java
+++ /dev/null
@@ -1,202 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * Returns the gradient of `StridedSlice`.
- * <p>
- * Since `StridedSlice` cuts out pieces of its `input` which is size
- * `shape`, its gradient will have the same shape (which is passed here
- * as `shape`). The gradient will be zero in any element that the slice
- * does not select.
- * <p>
- * Arguments are the same as StridedSliceGrad with the exception that
- * `dy` is the input gradient to be propagated and `shape` is the
- * shape of `StridedSlice`'s `input`.
- * 
- * @param <U> data type for {@code output()} output
- */
-@Operator
-public final class StridedSliceGrad<U extends TType> extends RawOp implements Operand<U> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.core.StridedSliceGrad}
-   */
-  public static class Options {
-    
-    /**
-     * @param beginMask 
-     */
-    public Options beginMask(Long beginMask) {
-      this.beginMask = beginMask;
-      return this;
-    }
-    
-    /**
-     * @param endMask 
-     */
-    public Options endMask(Long endMask) {
-      this.endMask = endMask;
-      return this;
-    }
-    
-    /**
-     * @param ellipsisMask 
-     */
-    public Options ellipsisMask(Long ellipsisMask) {
-      this.ellipsisMask = ellipsisMask;
-      return this;
-    }
-    
-    /**
-     * @param newAxisMask 
-     */
-    public Options newAxisMask(Long newAxisMask) {
-      this.newAxisMask = newAxisMask;
-      return this;
-    }
-    
-    /**
-     * @param shrinkAxisMask 
-     */
-    public Options shrinkAxisMask(Long shrinkAxisMask) {
-      this.shrinkAxisMask = shrinkAxisMask;
-      return this;
-    }
-    
-    private Long beginMask;
-    private Long endMask;
-    private Long ellipsisMask;
-    private Long newAxisMask;
-    private Long shrinkAxisMask;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new StridedSliceGrad operation.
-   * 
-   * @param scope current scope
-   * @param shape 
-   * @param begin 
-   * @param end 
-   * @param strides 
-   * @param dy 
-   * @param options carries optional attributes values
-   * @return a new instance of StridedSliceGrad
-   */
-  @Endpoint(describeByClass = true)
-  public static <U extends TType, T extends TNumber> StridedSliceGrad<U> create(Scope scope, Operand<T> shape, Operand<T> begin, Operand<T> end, Operand<T> strides, Operand<U> dy, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("StridedSliceGrad", scope.makeOpName("StridedSliceGrad"));
-    opBuilder.addInput(shape.asOutput(scope));
-    opBuilder.addInput(begin.asOutput(scope));
-    opBuilder.addInput(end.asOutput(scope));
-    opBuilder.addInput(strides.asOutput(scope));
-    opBuilder.addInput(dy.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.beginMask != null) {
-          opBuilder.setAttr("begin_mask", opts.beginMask);
-        }
-        if (opts.endMask != null) {
-          opBuilder.setAttr("end_mask", opts.endMask);
-        }
-        if (opts.ellipsisMask != null) {
-          opBuilder.setAttr("ellipsis_mask", opts.ellipsisMask);
-        }
-        if (opts.newAxisMask != null) {
-          opBuilder.setAttr("new_axis_mask", opts.newAxisMask);
-        }
-        if (opts.shrinkAxisMask != null) {
-          opBuilder.setAttr("shrink_axis_mask", opts.shrinkAxisMask);
-        }
-      }
-    }
-    return new StridedSliceGrad<U>(opBuilder.build());
-  }
-  
-  /**
-   * @param beginMask 
-   */
-  public static Options beginMask(Long beginMask) {
-    return new Options().beginMask(beginMask);
-  }
-  
-  /**
-   * @param endMask 
-   */
-  public static Options endMask(Long endMask) {
-    return new Options().endMask(endMask);
-  }
-  
-  /**
-   * @param ellipsisMask 
-   */
-  public static Options ellipsisMask(Long ellipsisMask) {
-    return new Options().ellipsisMask(ellipsisMask);
-  }
-  
-  /**
-   * @param newAxisMask 
-   */
-  public static Options newAxisMask(Long newAxisMask) {
-    return new Options().newAxisMask(newAxisMask);
-  }
-  
-  /**
-   * @param shrinkAxisMask 
-   */
-  public static Options shrinkAxisMask(Long shrinkAxisMask) {
-    return new Options().shrinkAxisMask(shrinkAxisMask);
-  }
-  
-  /**
-   */
-  public Output<U> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<U> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "StridedSliceGrad";
-  
-  private Output<U> output;
-  
-  private StridedSliceGrad(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Sum.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Sum.java
deleted file mode 100644
index ae5eca98bc7..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Sum.java
+++ /dev/null
@@ -1,118 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * Computes the sum of elements across dimensions of a tensor.
- * <p>
- * Reduces `input` along the dimensions given in `axis`. Unless
- * `keep_dims` is true, the rank of the tensor is reduced by 1 for each entry in
- * `axis`. If `keep_dims` is true, the reduced dimensions are
- * retained with length 1.
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator
-public final class Sum<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.core.Sum}
-   */
-  public static class Options {
-    
-    /**
-     * @param keepDims If true, retain reduced dimensions with length 1.
-     */
-    public Options keepDims(Boolean keepDims) {
-      this.keepDims = keepDims;
-      return this;
-    }
-    
-    private Boolean keepDims;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new Sum operation.
-   * 
-   * @param scope current scope
-   * @param input The tensor to reduce.
-   * @param axis The dimensions to reduce. Must be in the range
-   * `[-rank(input), rank(input))`.
-   * @param options carries optional attributes values
-   * @return a new instance of Sum
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> Sum<T> create(Scope scope, Operand<T> input, Operand<U> axis, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Sum", scope.makeOpName("Sum"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder.addInput(axis.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.keepDims != null) {
-          opBuilder.setAttr("keep_dims", opts.keepDims);
-        }
-      }
-    }
-    return new Sum<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param keepDims If true, retain reduced dimensions with length 1.
-   */
-  public static Options keepDims(Boolean keepDims) {
-    return new Options().keepDims(keepDims);
-  }
-  
-  /**
-   * The reduced tensor.
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Sum";
-  
-  private Output<T> output;
-  
-  private Sum(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/SwitchCond.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/SwitchCond.java
deleted file mode 100644
index 49bd0008e47..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/SwitchCond.java
+++ /dev/null
@@ -1,87 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TBool;
-import org.tensorflow.types.family.TType;
-
-/**
- * Forwards `data` to the output port determined by `pred`.
- * <p>
- * If `pred` is true, the `data` input is forwarded to `output_true`. Otherwise,
- * the data goes to `output_false`.
- * <p>
- * See also `RefSwitch` and `Merge`.
- * 
- * @param <T> data type for {@code outputFalse()} output
- */
-@Operator
-public final class SwitchCond<T extends TType> extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new SwitchCond operation.
-   * 
-   * @param scope current scope
-   * @param data The tensor to be forwarded to the appropriate output.
-   * @param pred A scalar that specifies which output port will receive data.
-   * @return a new instance of SwitchCond
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> SwitchCond<T> create(Scope scope, Operand<T> data, Operand<TBool> pred) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Switch", scope.makeOpName("SwitchCond"));
-    opBuilder.addInput(data.asOutput(scope));
-    opBuilder.addInput(pred.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new SwitchCond<T>(opBuilder.build());
-  }
-  
-  /**
-   * If `pred` is false, data will be forwarded to this output.
-   */
-  public Output<T> outputFalse() {
-    return outputFalse;
-  }
-  
-  /**
-   * If `pred` is true, data will be forwarded to this output.
-   */
-  public Output<T> outputTrue() {
-    return outputTrue;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Switch";
-  
-  private Output<T> outputFalse;
-  private Output<T> outputTrue;
-  
-  private SwitchCond(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    outputFalse = operation.output(outputIdx++);
-    outputTrue = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TemporaryVariable.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TemporaryVariable.java
deleted file mode 100644
index 97b76053a4a..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TemporaryVariable.java
+++ /dev/null
@@ -1,129 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Returns a tensor that may be mutated, but only persists within a single step.
- * <p>
- * This is an experimental op for internal use only and it is possible to use this
- * op in unsafe ways.  DO NOT USE unless you fully understand the risks.
- * <p>
- * It is the caller's responsibility to ensure that 'ref' is eventually passed to a
- * matching 'DestroyTemporaryVariable' op after all other uses have completed.
- * <p>
- * Outputs a ref to the tensor state so it may be read or modified.
- * <p>
- *   E.g.
- *       var = state_ops._temporary_variable([1, 2], types.float_)
- *       var_name = var.op.name
- *       var = state_ops.assign(var, [[4.0, 5.0]])
- *       var = state_ops.assign_add(var, [[6.0, 7.0]])
- *       final = state_ops._destroy_temporary_variable(var, var_name=var_name)
- * 
- * @param <T> data type for {@code ref()} output
- */
-@Operator
-public final class TemporaryVariable<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.core.TemporaryVariable}
-   */
-  public static class Options {
-    
-    /**
-     * @param varName Overrides the name used for the temporary variable resource. Default
-     * value is the name of the 'TemporaryVariable' op (which is guaranteed unique).
-     */
-    public Options varName(String varName) {
-      this.varName = varName;
-      return this;
-    }
-    
-    private String varName;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new TemporaryVariable operation.
-   * 
-   * @param scope current scope
-   * @param shape The shape of the variable tensor.
-   * @param dtype The type of elements in the variable tensor.
-   * @param options carries optional attributes values
-   * @return a new instance of TemporaryVariable
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> TemporaryVariable<T> create(Scope scope, Shape shape, Class<T> dtype, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("TemporaryVariable", scope.makeOpName("TemporaryVariable"));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("shape", shape);
-    opBuilder.setAttr("dtype", dtype);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.varName != null) {
-          opBuilder.setAttr("var_name", opts.varName);
-        }
-      }
-    }
-    return new TemporaryVariable<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param varName Overrides the name used for the temporary variable resource. Default
-   * value is the name of the 'TemporaryVariable' op (which is guaranteed unique).
-   */
-  public static Options varName(String varName) {
-    return new Options().varName(varName);
-  }
-  
-  /**
-   * A reference to the variable tensor.
-   */
-  public Output<T> ref() {
-    return ref;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return ref;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "TemporaryVariable";
-  
-  private Output<T> ref;
-  
-  private TemporaryVariable(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    ref = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArray.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArray.java
deleted file mode 100644
index 41cff2dce10..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArray.java
+++ /dev/null
@@ -1,218 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.family.TType;
-
-/**
- * An array of Tensors of given size.
- * <p>
- * Write data via Write and read via Read or Pack.
- */
-@Operator
-public final class TensorArray extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.core.TensorArray}
-   */
-  public static class Options {
-    
-    /**
-     * @param elementShape The expected shape of an element, if known. Used to
-     * validate the shapes of TensorArray elements. If this shape is not
-     * fully specified, gathering zero-size TensorArrays is an error.
-     */
-    public Options elementShape(Shape elementShape) {
-      this.elementShape = elementShape;
-      return this;
-    }
-    
-    /**
-     * @param dynamicSize A boolean that determines whether writes to the TensorArray
-     * are allowed to grow the size.  By default, this is not allowed.
-     */
-    public Options dynamicSize(Boolean dynamicSize) {
-      this.dynamicSize = dynamicSize;
-      return this;
-    }
-    
-    /**
-     * @param clearAfterRead If true (default), Tensors in the TensorArray are cleared
-     * after being read.  This disables multiple read semantics but allows early
-     * release of memory.
-     */
-    public Options clearAfterRead(Boolean clearAfterRead) {
-      this.clearAfterRead = clearAfterRead;
-      return this;
-    }
-    
-    /**
-     * @param identicalElementShapes If true (default is false), then all
-     * elements in the TensorArray will be expected to have have identical shapes.
-     * This allows certain behaviors, like dynamically checking for
-     * consistent shapes on write, and being able to fill in properly
-     * shaped zero tensors on stack -- even if the element_shape attribute
-     * is not fully defined.
-     */
-    public Options identicalElementShapes(Boolean identicalElementShapes) {
-      this.identicalElementShapes = identicalElementShapes;
-      return this;
-    }
-    
-    /**
-     * @param tensorArrayName Overrides the name used for the temporary tensor_array
-     * resource. Default value is the name of the 'TensorArray' op (which
-     * is guaranteed unique).
-     */
-    public Options tensorArrayName(String tensorArrayName) {
-      this.tensorArrayName = tensorArrayName;
-      return this;
-    }
-    
-    private Shape elementShape;
-    private Boolean dynamicSize;
-    private Boolean clearAfterRead;
-    private Boolean identicalElementShapes;
-    private String tensorArrayName;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new TensorArray operation.
-   * 
-   * @param scope current scope
-   * @param size The size of the array.
-   * @param dtype The type of the elements on the tensor_array.
-   * @param options carries optional attributes values
-   * @return a new instance of TensorArray
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> TensorArray create(Scope scope, Operand<TInt32> size, Class<T> dtype, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("TensorArrayV3", scope.makeOpName("TensorArray"));
-    opBuilder.addInput(size.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("dtype", dtype);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.elementShape != null) {
-          opBuilder.setAttr("element_shape", opts.elementShape);
-        }
-        if (opts.dynamicSize != null) {
-          opBuilder.setAttr("dynamic_size", opts.dynamicSize);
-        }
-        if (opts.clearAfterRead != null) {
-          opBuilder.setAttr("clear_after_read", opts.clearAfterRead);
-        }
-        if (opts.identicalElementShapes != null) {
-          opBuilder.setAttr("identical_element_shapes", opts.identicalElementShapes);
-        }
-        if (opts.tensorArrayName != null) {
-          opBuilder.setAttr("tensor_array_name", opts.tensorArrayName);
-        }
-      }
-    }
-    return new TensorArray(opBuilder.build());
-  }
-  
-  /**
-   * @param elementShape The expected shape of an element, if known. Used to
-   * validate the shapes of TensorArray elements. If this shape is not
-   * fully specified, gathering zero-size TensorArrays is an error.
-   */
-  public static Options elementShape(Shape elementShape) {
-    return new Options().elementShape(elementShape);
-  }
-  
-  /**
-   * @param dynamicSize A boolean that determines whether writes to the TensorArray
-   * are allowed to grow the size.  By default, this is not allowed.
-   */
-  public static Options dynamicSize(Boolean dynamicSize) {
-    return new Options().dynamicSize(dynamicSize);
-  }
-  
-  /**
-   * @param clearAfterRead If true (default), Tensors in the TensorArray are cleared
-   * after being read.  This disables multiple read semantics but allows early
-   * release of memory.
-   */
-  public static Options clearAfterRead(Boolean clearAfterRead) {
-    return new Options().clearAfterRead(clearAfterRead);
-  }
-  
-  /**
-   * @param identicalElementShapes If true (default is false), then all
-   * elements in the TensorArray will be expected to have have identical shapes.
-   * This allows certain behaviors, like dynamically checking for
-   * consistent shapes on write, and being able to fill in properly
-   * shaped zero tensors on stack -- even if the element_shape attribute
-   * is not fully defined.
-   */
-  public static Options identicalElementShapes(Boolean identicalElementShapes) {
-    return new Options().identicalElementShapes(identicalElementShapes);
-  }
-  
-  /**
-   * @param tensorArrayName Overrides the name used for the temporary tensor_array
-   * resource. Default value is the name of the 'TensorArray' op (which
-   * is guaranteed unique).
-   */
-  public static Options tensorArrayName(String tensorArrayName) {
-    return new Options().tensorArrayName(tensorArrayName);
-  }
-  
-  /**
-   * The handle to the TensorArray.
-   */
-  public Output<?> handle() {
-    return handle;
-  }
-  
-  /**
-   * A scalar used to control gradient flow.
-   */
-  public Output<TFloat32> flow() {
-    return flow;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "TensorArrayV3";
-  
-  private Output<?> handle;
-  private Output<TFloat32> flow;
-  
-  private TensorArray(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    handle = operation.output(outputIdx++);
-    flow = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayClose.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayClose.java
deleted file mode 100644
index 047dda8e506..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayClose.java
+++ /dev/null
@@ -1,58 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-
-/**
- * Delete the TensorArray from its resource container.
- * <p>
- * This enables the user to close and release the resource in the middle
- * of a step/run.
- */
-@Operator
-public final class TensorArrayClose extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new TensorArrayClose operation.
-   * 
-   * @param scope current scope
-   * @param handle The handle to a TensorArray (output of TensorArray or TensorArrayGrad).
-   * @return a new instance of TensorArrayClose
-   */
-  @Endpoint(describeByClass = true)
-  public static TensorArrayClose create(Scope scope, Operand<?> handle) {
-    OperationBuilder opBuilder = scope.env().opBuilder("TensorArrayCloseV3", scope.makeOpName("TensorArrayClose"));
-    opBuilder.addInput(handle.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new TensorArrayClose(opBuilder.build());
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "TensorArrayCloseV3";
-  
-  private TensorArrayClose(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayConcat.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayConcat.java
deleted file mode 100644
index 3151585cf31..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayConcat.java
+++ /dev/null
@@ -1,140 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
-
-/**
- * Concat the elements from the TensorArray into value `value`.
- * <p>
- * Takes `T` elements of shapes
- * <p>
- *   <pre>{@code
- *   (n0 x d0 x d1 x ...), (n1 x d0 x d1 x ...), ..., (n(T-1) x d0 x d1 x ...)
- *   }</pre>
- * and concatenates them into a Tensor of shape:
- * <p>
- *   <pre>{@code
- * (n0 + n1 + ... + n(T-1) x d0 x d1 x ...)}</pre>
- * All elements must have the same shape (excepting the first dimension).
- * 
- * @param <T> data type for {@code value()} output
- */
-@Operator
-public final class TensorArrayConcat<T extends TType> extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.core.TensorArrayConcat}
-   */
-  public static class Options {
-    
-    /**
-     * @param elementShapeExcept0 The expected shape of an element, if known,
-     * excluding the first dimension. Used to validate the shapes of
-     * TensorArray elements. If this shape is not fully specified, concatenating
-     * zero-size TensorArrays is an error.
-     */
-    public Options elementShapeExcept0(Shape elementShapeExcept0) {
-      this.elementShapeExcept0 = elementShapeExcept0;
-      return this;
-    }
-    
-    private Shape elementShapeExcept0;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new TensorArrayConcat operation.
-   * 
-   * @param scope current scope
-   * @param handle The handle to a TensorArray.
-   * @param flowIn A float scalar that enforces proper chaining of operations.
-   * @param dtype The type of the elem that is returned.
-   * @param options carries optional attributes values
-   * @return a new instance of TensorArrayConcat
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> TensorArrayConcat<T> create(Scope scope, Operand<?> handle, Operand<TFloat32> flowIn, Class<T> dtype, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("TensorArrayConcatV3", scope.makeOpName("TensorArrayConcat"));
-    opBuilder.addInput(handle.asOutput(scope));
-    opBuilder.addInput(flowIn.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("dtype", dtype);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.elementShapeExcept0 != null) {
-          opBuilder.setAttr("element_shape_except0", opts.elementShapeExcept0);
-        }
-      }
-    }
-    return new TensorArrayConcat<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param elementShapeExcept0 The expected shape of an element, if known,
-   * excluding the first dimension. Used to validate the shapes of
-   * TensorArray elements. If this shape is not fully specified, concatenating
-   * zero-size TensorArrays is an error.
-   */
-  public static Options elementShapeExcept0(Shape elementShapeExcept0) {
-    return new Options().elementShapeExcept0(elementShapeExcept0);
-  }
-  
-  /**
-   * All of the elements in the TensorArray, concatenated along the first
-   * axis.
-   */
-  public Output<T> value() {
-    return value;
-  }
-  
-  /**
-   * A vector of the row sizes of the original T elements in the
-   * value output.  In the example above, this would be the values:
-   * `(n1, n2, ..., n(T-1))`.
-   */
-  public Output<TInt64> lengths() {
-    return lengths;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "TensorArrayConcatV3";
-  
-  private Output<T> value;
-  private Output<TInt64> lengths;
-  
-  private TensorArrayConcat(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    value = operation.output(outputIdx++);
-    lengths = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayGather.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayGather.java
deleted file mode 100644
index 69fdd98af0d..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayGather.java
+++ /dev/null
@@ -1,125 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.family.TType;
-
-/**
- * Gather specific elements from the TensorArray into output `value`.
- * <p>
- * All elements selected by `indices` must have the same shape.
- * 
- * @param <T> data type for {@code value()} output
- */
-@Operator
-public final class TensorArrayGather<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.core.TensorArrayGather}
-   */
-  public static class Options {
-    
-    /**
-     * @param elementShape The expected shape of an element, if known. Used to
-     * validate the shapes of TensorArray elements. If this shape is not
-     * fully specified, gathering zero-size TensorArrays is an error.
-     */
-    public Options elementShape(Shape elementShape) {
-      this.elementShape = elementShape;
-      return this;
-    }
-    
-    private Shape elementShape;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new TensorArrayGather operation.
-   * 
-   * @param scope current scope
-   * @param handle The handle to a TensorArray.
-   * @param indices The locations in the TensorArray from which to read tensor elements.
-   * @param flowIn A float scalar that enforces proper chaining of operations.
-   * @param dtype The type of the elem that is returned.
-   * @param options carries optional attributes values
-   * @return a new instance of TensorArrayGather
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> TensorArrayGather<T> create(Scope scope, Operand<?> handle, Operand<TInt32> indices, Operand<TFloat32> flowIn, Class<T> dtype, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("TensorArrayGatherV3", scope.makeOpName("TensorArrayGather"));
-    opBuilder.addInput(handle.asOutput(scope));
-    opBuilder.addInput(indices.asOutput(scope));
-    opBuilder.addInput(flowIn.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("dtype", dtype);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.elementShape != null) {
-          opBuilder.setAttr("element_shape", opts.elementShape);
-        }
-      }
-    }
-    return new TensorArrayGather<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param elementShape The expected shape of an element, if known. Used to
-   * validate the shapes of TensorArray elements. If this shape is not
-   * fully specified, gathering zero-size TensorArrays is an error.
-   */
-  public static Options elementShape(Shape elementShape) {
-    return new Options().elementShape(elementShape);
-  }
-  
-  /**
-   * All of the elements in the TensorArray, concatenated along a new
-   * axis (the new dimension 0).
-   */
-  public Output<T> value() {
-    return value;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return value;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "TensorArrayGatherV3";
-  
-  private Output<T> value;
-  
-  private TensorArrayGather(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    value = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayGrad.java
deleted file mode 100644
index 961a09cb278..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayGrad.java
+++ /dev/null
@@ -1,117 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-
-/**
- * Creates a TensorArray for storing the gradients of values in the given handle.
- * <p>
- * If the given TensorArray gradient already exists, returns a reference to it.
- * <p>
- * Locks the size of the original TensorArray by disabling its dynamic size flag.
- * <p>
- * **A note about the input flow_in:**
- * <p>
- * The handle flow_in forces the execution of the gradient lookup to occur
- * only after certain other operations have occurred.  For example, when
- * the forward TensorArray is dynamically sized, writes to this TensorArray
- * may resize the object.  The gradient TensorArray is statically sized based
- * on the size of the forward TensorArray when this operation executes.
- * Furthermore, the size of the forward TensorArray is frozen by this call.
- * As a result, the flow is used to ensure that the call to generate the gradient
- * TensorArray only happens after all writes are executed.
- * <p>
- * In the case of dynamically sized TensorArrays, gradient computation should
- * only be performed on read operations that have themselves been chained via
- * flow to occur only after all writes have executed. That way the final size
- * of the forward TensorArray is known when this operation is called.
- * <p>
- * **A note about the source attribute:**
- * <p>
- * TensorArray gradient calls use an accumulator TensorArray object.  If
- * multiple gradients are calculated and run in the same session, the multiple
- * gradient nodes may accidentally flow through the same accumulator TensorArray.
- * This double counts and generally breaks the TensorArray gradient flow.
- * <p>
- * The solution is to identify which gradient call this particular
- * TensorArray gradient is being called in.  This is performed by identifying
- * a unique string (e.g. "gradients", "gradients_1", ...) from the input
- * gradient Tensor's name.  This string is used as a suffix when creating
- * the TensorArray gradient object here (the attribute `source`).
- * <p>
- * The attribute `source` is added as a suffix to the forward TensorArray's
- * name when performing the creation / lookup, so that each separate gradient
- * calculation gets its own TensorArray accumulator.
- */
-@Operator
-public final class TensorArrayGrad extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new TensorArrayGrad operation.
-   * 
-   * @param scope current scope
-   * @param handle The handle to the forward TensorArray.
-   * @param flowIn A float scalar that enforces proper chaining of operations.
-   * @param source The gradient source string, used to decide which gradient TensorArray
-   * to return.
-   * @return a new instance of TensorArrayGrad
-   */
-  @Endpoint(describeByClass = true)
-  public static TensorArrayGrad create(Scope scope, Operand<?> handle, Operand<TFloat32> flowIn, String source) {
-    OperationBuilder opBuilder = scope.env().opBuilder("TensorArrayGradV3", scope.makeOpName("TensorArrayGrad"));
-    opBuilder.addInput(handle.asOutput(scope));
-    opBuilder.addInput(flowIn.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("source", source);
-    return new TensorArrayGrad(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<?> gradHandle() {
-    return gradHandle;
-  }
-  
-  /**
-   */
-  public Output<TFloat32> flowOut() {
-    return flowOut;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "TensorArrayGradV3";
-  
-  private Output<?> gradHandle;
-  private Output<TFloat32> flowOut;
-  
-  private TensorArrayGrad(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    gradHandle = operation.output(outputIdx++);
-    flowOut = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayGradWithShape.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayGradWithShape.java
deleted file mode 100644
index 6420095d6c5..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayGradWithShape.java
+++ /dev/null
@@ -1,90 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.TInt32;
-
-/**
- * Creates a TensorArray for storing multiple gradients of values in the given handle.
- * <p>
- * Similar to TensorArrayGradV3. However it creates an accumulator with an
- * expanded shape compared to the input TensorArray whose gradient is being
- * computed. This enables multiple gradients for the same TensorArray to be
- * calculated using the same accumulator.
- */
-@Operator
-public final class TensorArrayGradWithShape extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new TensorArrayGradWithShape operation.
-   * 
-   * @param scope current scope
-   * @param handle The handle to the forward TensorArray.
-   * @param flowIn A float scalar that enforces proper chaining of operations.
-   * @param shapeToPrepend An int32 vector representing a shape. Elements in the gradient accumulator will
-   * have shape which is this shape_to_prepend value concatenated with shape of the
-   * elements in the TensorArray corresponding to the input handle.
-   * @param source The gradient source string, used to decide which gradient TensorArray
-   * to return.
-   * @return a new instance of TensorArrayGradWithShape
-   */
-  @Endpoint(describeByClass = true)
-  public static TensorArrayGradWithShape create(Scope scope, Operand<?> handle, Operand<TFloat32> flowIn, Operand<TInt32> shapeToPrepend, String source) {
-    OperationBuilder opBuilder = scope.env().opBuilder("TensorArrayGradWithShape", scope.makeOpName("TensorArrayGradWithShape"));
-    opBuilder.addInput(handle.asOutput(scope));
-    opBuilder.addInput(flowIn.asOutput(scope));
-    opBuilder.addInput(shapeToPrepend.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("source", source);
-    return new TensorArrayGradWithShape(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<?> gradHandle() {
-    return gradHandle;
-  }
-  
-  /**
-   */
-  public Output<TFloat32> flowOut() {
-    return flowOut;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "TensorArrayGradWithShape";
-  
-  private Output<?> gradHandle;
-  private Output<TFloat32> flowOut;
-  
-  private TensorArrayGradWithShape(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    gradHandle = operation.output(outputIdx++);
-    flowOut = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayPack.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayPack.java
deleted file mode 100644
index dbed7f00a84..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayPack.java
+++ /dev/null
@@ -1,113 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.TString;
-import org.tensorflow.types.family.TType;
-
-/**
- * @param <T> data type for {@code value()} output
- */
-@Operator
-public final class TensorArrayPack<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.core.TensorArrayPack}
-   */
-  public static class Options {
-    
-    /**
-     * @param elementShape 
-     */
-    public Options elementShape(Shape elementShape) {
-      this.elementShape = elementShape;
-      return this;
-    }
-    
-    private Shape elementShape;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new TensorArrayPack operation.
-   * 
-   * @param scope current scope
-   * @param handle 
-   * @param flowIn 
-   * @param dtype 
-   * @param options carries optional attributes values
-   * @return a new instance of TensorArrayPack
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> TensorArrayPack<T> create(Scope scope, Operand<TString> handle, Operand<TFloat32> flowIn, Class<T> dtype, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("TensorArrayPack", scope.makeOpName("TensorArrayPack"));
-    opBuilder.addInput(handle.asOutput(scope));
-    opBuilder.addInput(flowIn.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("dtype", dtype);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.elementShape != null) {
-          opBuilder.setAttr("element_shape", opts.elementShape);
-        }
-      }
-    }
-    return new TensorArrayPack<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param elementShape 
-   */
-  public static Options elementShape(Shape elementShape) {
-    return new Options().elementShape(elementShape);
-  }
-  
-  /**
-   */
-  public Output<T> value() {
-    return value;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return value;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "TensorArrayPack";
-  
-  private Output<T> value;
-  
-  private TensorArrayPack(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    value = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayRead.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayRead.java
deleted file mode 100644
index bbefe81a7bf..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayRead.java
+++ /dev/null
@@ -1,83 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.family.TType;
-
-/**
- * Read an element from the TensorArray into output `value`.
- * 
- * @param <T> data type for {@code value()} output
- */
-@Operator
-public final class TensorArrayRead<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new TensorArrayRead operation.
-   * 
-   * @param scope current scope
-   * @param handle The handle to a TensorArray.
-   * @param index 
-   * @param flowIn A float scalar that enforces proper chaining of operations.
-   * @param dtype The type of the elem that is returned.
-   * @return a new instance of TensorArrayRead
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> TensorArrayRead<T> create(Scope scope, Operand<?> handle, Operand<TInt32> index, Operand<TFloat32> flowIn, Class<T> dtype) {
-    OperationBuilder opBuilder = scope.env().opBuilder("TensorArrayReadV3", scope.makeOpName("TensorArrayRead"));
-    opBuilder.addInput(handle.asOutput(scope));
-    opBuilder.addInput(index.asOutput(scope));
-    opBuilder.addInput(flowIn.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("dtype", dtype);
-    return new TensorArrayRead<T>(opBuilder.build());
-  }
-  
-  /**
-   * The tensor that is read from the TensorArray.
-   */
-  public Output<T> value() {
-    return value;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return value;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "TensorArrayReadV3";
-  
-  private Output<T> value;
-  
-  private TensorArrayRead(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    value = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayScatter.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayScatter.java
deleted file mode 100644
index 6053ce43fbe..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayScatter.java
+++ /dev/null
@@ -1,83 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.family.TType;
-
-/**
- * Scatter the data from the input value into specific TensorArray elements.
- * <p>
- * `indices` must be a vector, its length must match the first dim of `value`.
- */
-@Operator
-public final class TensorArrayScatter extends RawOp implements Operand<TFloat32> {
-  
-  /**
-   * Factory method to create a class wrapping a new TensorArrayScatter operation.
-   * 
-   * @param scope current scope
-   * @param handle The handle to a TensorArray.
-   * @param indices The locations at which to write the tensor elements.
-   * @param value The concatenated tensor to write to the TensorArray.
-   * @param flowIn A float scalar that enforces proper chaining of operations.
-   * @return a new instance of TensorArrayScatter
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> TensorArrayScatter create(Scope scope, Operand<?> handle, Operand<TInt32> indices, Operand<T> value, Operand<TFloat32> flowIn) {
-    OperationBuilder opBuilder = scope.env().opBuilder("TensorArrayScatterV3", scope.makeOpName("TensorArrayScatter"));
-    opBuilder.addInput(handle.asOutput(scope));
-    opBuilder.addInput(indices.asOutput(scope));
-    opBuilder.addInput(value.asOutput(scope));
-    opBuilder.addInput(flowIn.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new TensorArrayScatter(opBuilder.build());
-  }
-  
-  /**
-   * A float scalar that enforces proper chaining of operations.
-   */
-  public Output<TFloat32> flowOut() {
-    return flowOut;
-  }
-  
-  @Override
-  public Output<TFloat32> asOutput(Scope scope) {
-    return flowOut;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "TensorArrayScatterV3";
-  
-  private Output<TFloat32> flowOut;
-  
-  private TensorArrayScatter(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    flowOut = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArraySize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArraySize.java
deleted file mode 100644
index 41883931ca6..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArraySize.java
+++ /dev/null
@@ -1,76 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.TInt32;
-
-/**
- * Get the current size of the TensorArray.
- */
-@Operator
-public final class TensorArraySize extends RawOp implements Operand<TInt32> {
-  
-  /**
-   * Factory method to create a class wrapping a new TensorArraySize operation.
-   * 
-   * @param scope current scope
-   * @param handle The handle to a TensorArray (output of TensorArray or TensorArrayGrad).
-   * @param flowIn A float scalar that enforces proper chaining of operations.
-   * @return a new instance of TensorArraySize
-   */
-  @Endpoint(describeByClass = true)
-  public static TensorArraySize create(Scope scope, Operand<?> handle, Operand<TFloat32> flowIn) {
-    OperationBuilder opBuilder = scope.env().opBuilder("TensorArraySizeV3", scope.makeOpName("TensorArraySize"));
-    opBuilder.addInput(handle.asOutput(scope));
-    opBuilder.addInput(flowIn.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new TensorArraySize(opBuilder.build());
-  }
-  
-  /**
-   * The current size of the TensorArray.
-   */
-  public Output<TInt32> size() {
-    return size;
-  }
-  
-  @Override
-  public Output<TInt32> asOutput(Scope scope) {
-    return size;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "TensorArraySizeV3";
-  
-  private Output<TInt32> size;
-  
-  private TensorArraySize(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    size = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArraySplit.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArraySplit.java
deleted file mode 100644
index 4b9bc963fac..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArraySplit.java
+++ /dev/null
@@ -1,104 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
-
-/**
- * Split the data from the input value into TensorArray elements.
- * <p>
- * Assuming that `lengths` takes on values
- * <p>
- *   <pre>{@code
- * (n0, n1, ..., n(T-1))}</pre>
- * and that `value` has shape
- * <p>
- *   <pre>{@code
- * (n0 + n1 + ... + n(T-1) x d0 x d1 x ...)}</pre>
- * ,
- * <p>
- * this splits values into a TensorArray with T tensors.
- * <p>
- * TensorArray index t will be the subtensor of values with starting position
- * <p>
- *   <pre>{@code
- * (n0 + n1 + ... + n(t-1), 0, 0, ...)}</pre>
- * and having size
- * <p>
- *   <pre>{@code
- * nt x d0 x d1 x ...}</pre>
- * 
- */
-@Operator
-public final class TensorArraySplit extends RawOp implements Operand<TFloat32> {
-  
-  /**
-   * Factory method to create a class wrapping a new TensorArraySplit operation.
-   * 
-   * @param scope current scope
-   * @param handle The handle to a TensorArray.
-   * @param value The concatenated tensor to write to the TensorArray.
-   * @param lengths The vector of lengths, how to split the rows of value into the
-   * TensorArray.
-   * @param flowIn A float scalar that enforces proper chaining of operations.
-   * @return a new instance of TensorArraySplit
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> TensorArraySplit create(Scope scope, Operand<?> handle, Operand<T> value, Operand<TInt64> lengths, Operand<TFloat32> flowIn) {
-    OperationBuilder opBuilder = scope.env().opBuilder("TensorArraySplitV3", scope.makeOpName("TensorArraySplit"));
-    opBuilder.addInput(handle.asOutput(scope));
-    opBuilder.addInput(value.asOutput(scope));
-    opBuilder.addInput(lengths.asOutput(scope));
-    opBuilder.addInput(flowIn.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new TensorArraySplit(opBuilder.build());
-  }
-  
-  /**
-   * A float scalar that enforces proper chaining of operations.
-   */
-  public Output<TFloat32> flowOut() {
-    return flowOut;
-  }
-  
-  @Override
-  public Output<TFloat32> asOutput(Scope scope) {
-    return flowOut;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "TensorArraySplitV3";
-  
-  private Output<TFloat32> flowOut;
-  
-  private TensorArraySplit(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    flowOut = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayUnpack.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayUnpack.java
deleted file mode 100644
index f179119693c..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayUnpack.java
+++ /dev/null
@@ -1,77 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.TString;
-import org.tensorflow.types.family.TType;
-
-/**
- */
-@Operator
-public final class TensorArrayUnpack extends RawOp implements Operand<TFloat32> {
-  
-  /**
-   * Factory method to create a class wrapping a new TensorArrayUnpack operation.
-   * 
-   * @param scope current scope
-   * @param handle 
-   * @param value 
-   * @param flowIn 
-   * @return a new instance of TensorArrayUnpack
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> TensorArrayUnpack create(Scope scope, Operand<TString> handle, Operand<T> value, Operand<TFloat32> flowIn) {
-    OperationBuilder opBuilder = scope.env().opBuilder("TensorArrayUnpack", scope.makeOpName("TensorArrayUnpack"));
-    opBuilder.addInput(handle.asOutput(scope));
-    opBuilder.addInput(value.asOutput(scope));
-    opBuilder.addInput(flowIn.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new TensorArrayUnpack(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<TFloat32> flowOut() {
-    return flowOut;
-  }
-  
-  @Override
-  public Output<TFloat32> asOutput(Scope scope) {
-    return flowOut;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "TensorArrayUnpack";
-  
-  private Output<TFloat32> flowOut;
-  
-  private TensorArrayUnpack(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    flowOut = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayWrite.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayWrite.java
deleted file mode 100644
index 1a37e6474a4..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayWrite.java
+++ /dev/null
@@ -1,81 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.family.TType;
-
-/**
- * Push an element onto the tensor_array.
- */
-@Operator
-public final class TensorArrayWrite extends RawOp implements Operand<TFloat32> {
-  
-  /**
-   * Factory method to create a class wrapping a new TensorArrayWrite operation.
-   * 
-   * @param scope current scope
-   * @param handle The handle to a TensorArray.
-   * @param index The position to write to inside the TensorArray.
-   * @param value The tensor to write to the TensorArray.
-   * @param flowIn A float scalar that enforces proper chaining of operations.
-   * @return a new instance of TensorArrayWrite
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> TensorArrayWrite create(Scope scope, Operand<?> handle, Operand<TInt32> index, Operand<T> value, Operand<TFloat32> flowIn) {
-    OperationBuilder opBuilder = scope.env().opBuilder("TensorArrayWriteV3", scope.makeOpName("TensorArrayWrite"));
-    opBuilder.addInput(handle.asOutput(scope));
-    opBuilder.addInput(index.asOutput(scope));
-    opBuilder.addInput(value.asOutput(scope));
-    opBuilder.addInput(flowIn.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new TensorArrayWrite(opBuilder.build());
-  }
-  
-  /**
-   * A float scalar that enforces proper chaining of operations.
-   */
-  public Output<TFloat32> flowOut() {
-    return flowOut;
-  }
-  
-  @Override
-  public Output<TFloat32> asOutput(Scope scope) {
-    return flowOut;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "TensorArrayWriteV3";
-  
-  private Output<TFloat32> flowOut;
-  
-  private TensorArrayWrite(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    flowOut = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorForestCreateTreeVariable.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorForestCreateTreeVariable.java
deleted file mode 100644
index 3daad106a9e..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorForestCreateTreeVariable.java
+++ /dev/null
@@ -1,57 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TString;
-
-/**
- * Creates a tree resource and returns a handle to it.
- */
-public final class TensorForestCreateTreeVariable extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new TensorForestCreateTreeVariable operation.
-   * 
-   * @param scope current scope
-   * @param treeHandle Handle to the tree resource to be created.
-   * @param treeConfig Serialized proto string of the boosted_trees.Tree.
-   * @return a new instance of TensorForestCreateTreeVariable
-   */
-  @Endpoint(describeByClass = true)
-  public static TensorForestCreateTreeVariable create(Scope scope, Operand<?> treeHandle, Operand<TString> treeConfig) {
-    OperationBuilder opBuilder = scope.env().opBuilder("TensorForestCreateTreeVariable", scope.makeOpName("TensorForestCreateTreeVariable"));
-    opBuilder.addInput(treeHandle.asOutput(scope));
-    opBuilder.addInput(treeConfig.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new TensorForestCreateTreeVariable(opBuilder.build());
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "TensorForestCreateTreeVariable";
-  
-  private TensorForestCreateTreeVariable(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorForestTreeDeserialize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorForestTreeDeserialize.java
deleted file mode 100644
index cdbbc60af61..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorForestTreeDeserialize.java
+++ /dev/null
@@ -1,57 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TString;
-
-/**
- * Deserializes a proto into the tree handle
- */
-public final class TensorForestTreeDeserialize extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new TensorForestTreeDeserialize operation.
-   * 
-   * @param scope current scope
-   * @param treeHandle Handle to the tree resource to be restored.
-   * @param treeConfig Serialied proto string of the boosted_trees.Tree proto.
-   * @return a new instance of TensorForestTreeDeserialize
-   */
-  @Endpoint(describeByClass = true)
-  public static TensorForestTreeDeserialize create(Scope scope, Operand<?> treeHandle, Operand<TString> treeConfig) {
-    OperationBuilder opBuilder = scope.env().opBuilder("TensorForestTreeDeserialize", scope.makeOpName("TensorForestTreeDeserialize"));
-    opBuilder.addInput(treeHandle.asOutput(scope));
-    opBuilder.addInput(treeConfig.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new TensorForestTreeDeserialize(opBuilder.build());
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "TensorForestTreeDeserialize";
-  
-  private TensorForestTreeDeserialize(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorForestTreeIsInitializedOp.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorForestTreeIsInitializedOp.java
deleted file mode 100644
index f15882776a6..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorForestTreeIsInitializedOp.java
+++ /dev/null
@@ -1,72 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TBool;
-
-/**
- * Checks whether a tree has been initialized.
- */
-public final class TensorForestTreeIsInitializedOp extends RawOp implements Operand<TBool> {
-  
-  /**
-   * Factory method to create a class wrapping a new TensorForestTreeIsInitializedOp operation.
-   * 
-   * @param scope current scope
-   * @param treeHandle Handle to the tree.
-   * @return a new instance of TensorForestTreeIsInitializedOp
-   */
-  @Endpoint(describeByClass = true)
-  public static TensorForestTreeIsInitializedOp create(Scope scope, Operand<?> treeHandle) {
-    OperationBuilder opBuilder = scope.env().opBuilder("TensorForestTreeIsInitializedOp", scope.makeOpName("TensorForestTreeIsInitializedOp"));
-    opBuilder.addInput(treeHandle.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new TensorForestTreeIsInitializedOp(opBuilder.build());
-  }
-  
-  /**
-   * Whether the tree is initialized.
-   */
-  public Output<TBool> isInitialized() {
-    return isInitialized;
-  }
-  
-  @Override
-  public Output<TBool> asOutput(Scope scope) {
-    return isInitialized;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "TensorForestTreeIsInitializedOp";
-  
-  private Output<TBool> isInitialized;
-  
-  private TensorForestTreeIsInitializedOp(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    isInitialized = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorForestTreePredict.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorForestTreePredict.java
deleted file mode 100644
index 4b8268d0258..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorForestTreePredict.java
+++ /dev/null
@@ -1,76 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-
-/**
- * Output the logits for the given input data
- */
-public final class TensorForestTreePredict extends RawOp implements Operand<TFloat32> {
-  
-  /**
-   * Factory method to create a class wrapping a new TensorForestTreePredict operation.
-   * 
-   * @param scope current scope
-   * @param treeHandle Handle to the tree resource.
-   * @param denseFeatures Rank 2 dense features tensor.
-   * @param logitsDimension Scalar, dimension of the logits.
-   * @return a new instance of TensorForestTreePredict
-   */
-  @Endpoint(describeByClass = true)
-  public static TensorForestTreePredict create(Scope scope, Operand<?> treeHandle, Operand<TFloat32> denseFeatures, Long logitsDimension) {
-    OperationBuilder opBuilder = scope.env().opBuilder("TensorForestTreePredict", scope.makeOpName("TensorForestTreePredict"));
-    opBuilder.addInput(treeHandle.asOutput(scope));
-    opBuilder.addInput(denseFeatures.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("logits_dimension", logitsDimension);
-    return new TensorForestTreePredict(opBuilder.build());
-  }
-  
-  /**
-   * The logits predictions from the tree for each instance in the batch.
-   */
-  public Output<TFloat32> logits() {
-    return logits;
-  }
-  
-  @Override
-  public Output<TFloat32> asOutput(Scope scope) {
-    return logits;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "TensorForestTreePredict";
-  
-  private Output<TFloat32> logits;
-  
-  private TensorForestTreePredict(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    logits = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorForestTreeResourceHandleOp.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorForestTreeResourceHandleOp.java
deleted file mode 100644
index d2a7541b05c..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorForestTreeResourceHandleOp.java
+++ /dev/null
@@ -1,123 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Creates a handle to a TensorForestTreeResource
- */
-public final class TensorForestTreeResourceHandleOp extends RawOp implements Operand<TType> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.core.TensorForestTreeResourceHandleOp}
-   */
-  public static class Options {
-    
-    /**
-     * @param container 
-     */
-    public Options container(String container) {
-      this.container = container;
-      return this;
-    }
-    
-    /**
-     * @param sharedName 
-     */
-    public Options sharedName(String sharedName) {
-      this.sharedName = sharedName;
-      return this;
-    }
-    
-    private String container;
-    private String sharedName;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new TensorForestTreeResourceHandleOp operation.
-   * 
-   * @param scope current scope
-   * @param options carries optional attributes values
-   * @return a new instance of TensorForestTreeResourceHandleOp
-   */
-  @Endpoint(describeByClass = true)
-  public static TensorForestTreeResourceHandleOp create(Scope scope, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("TensorForestTreeResourceHandleOp", scope.makeOpName("TensorForestTreeResourceHandleOp"));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.container != null) {
-          opBuilder.setAttr("container", opts.container);
-        }
-        if (opts.sharedName != null) {
-          opBuilder.setAttr("shared_name", opts.sharedName);
-        }
-      }
-    }
-    return new TensorForestTreeResourceHandleOp(opBuilder.build());
-  }
-  
-  /**
-   * @param container 
-   */
-  public static Options container(String container) {
-    return new Options().container(container);
-  }
-  
-  /**
-   * @param sharedName 
-   */
-  public static Options sharedName(String sharedName) {
-    return new Options().sharedName(sharedName);
-  }
-  
-  /**
-   */
-  public Output<?> resource() {
-    return resource;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) resource;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "TensorForestTreeResourceHandleOp";
-  
-  private Output<?> resource;
-  
-  private TensorForestTreeResourceHandleOp(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    resource = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorForestTreeSerialize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorForestTreeSerialize.java
deleted file mode 100644
index fbb182ff56d..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorForestTreeSerialize.java
+++ /dev/null
@@ -1,72 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TString;
-
-/**
- * Serializes the tree handle to a proto
- */
-public final class TensorForestTreeSerialize extends RawOp implements Operand<TString> {
-  
-  /**
-   * Factory method to create a class wrapping a new TensorForestTreeSerialize operation.
-   * 
-   * @param scope current scope
-   * @param treeHandle Handle to the tree resource to be serialized.
-   * @return a new instance of TensorForestTreeSerialize
-   */
-  @Endpoint(describeByClass = true)
-  public static TensorForestTreeSerialize create(Scope scope, Operand<?> treeHandle) {
-    OperationBuilder opBuilder = scope.env().opBuilder("TensorForestTreeSerialize", scope.makeOpName("TensorForestTreeSerialize"));
-    opBuilder.addInput(treeHandle.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new TensorForestTreeSerialize(opBuilder.build());
-  }
-  
-  /**
-   * Serialied proto string of the tree resource.
-   */
-  public Output<TString> treeConfig() {
-    return treeConfig;
-  }
-  
-  @Override
-  public Output<TString> asOutput(Scope scope) {
-    return treeConfig;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "TensorForestTreeSerialize";
-  
-  private Output<TString> treeConfig;
-  
-  private TensorForestTreeSerialize(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    treeConfig = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorForestTreeSize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorForestTreeSize.java
deleted file mode 100644
index ce4abb39160..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorForestTreeSize.java
+++ /dev/null
@@ -1,72 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
-
-/**
- * Get the number of nodes in a tree
- */
-public final class TensorForestTreeSize extends RawOp implements Operand<TInt32> {
-  
-  /**
-   * Factory method to create a class wrapping a new TensorForestTreeSize operation.
-   * 
-   * @param scope current scope
-   * @param treeHandle Handle to the tree resource.
-   * @return a new instance of TensorForestTreeSize
-   */
-  @Endpoint(describeByClass = true)
-  public static TensorForestTreeSize create(Scope scope, Operand<?> treeHandle) {
-    OperationBuilder opBuilder = scope.env().opBuilder("TensorForestTreeSize", scope.makeOpName("TensorForestTreeSize"));
-    opBuilder.addInput(treeHandle.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new TensorForestTreeSize(opBuilder.build());
-  }
-  
-  /**
-   * The size of the tree.
-   */
-  public Output<TInt32> treeSize() {
-    return treeSize;
-  }
-  
-  @Override
-  public Output<TInt32> asOutput(Scope scope) {
-    return treeSize;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "TensorForestTreeSize";
-  
-  private Output<TInt32> treeSize;
-  
-  private TensorForestTreeSize(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    treeSize = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListConcat.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListConcat.java
deleted file mode 100644
index 387cf6db10b..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListConcat.java
+++ /dev/null
@@ -1,98 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * Concats all tensors in the list along the 0th dimension.
- * <p>
- * Requires that all tensors have the same shape except the first dimension.
- * <p>
- * input_handle: The input list.
- * element_shape: The shape of the uninitialized elements in the list. If the first
- *   dimension is not -1, it is assumed that all list elements have the same
- *   leading dim.
- * leading_dims: The list of leading dims of uninitialized list elements. Used if
- *   the leading dim of input_handle.element_shape or the element_shape input arg
- *   is not already set.
- * tensor: The concated result.
- * lengths: Output tensor containing sizes of the 0th dimension of tensors in the list, used for computing the gradient.
- * 
- * 
- * @param <U> data type for {@code tensor()} output
- */
-@Operator
-public final class TensorListConcat<U extends TType> extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new TensorListConcat operation.
-   * 
-   * @param scope current scope
-   * @param inputHandle 
-   * @param elementShape 
-   * @param leadingDims 
-   * @param elementDtype 
-   * @return a new instance of TensorListConcat
-   */
-  @Endpoint(describeByClass = true)
-  public static <U extends TType, T extends TNumber> TensorListConcat<U> create(Scope scope, Operand<?> inputHandle, Operand<T> elementShape, Operand<TInt64> leadingDims, Class<U> elementDtype) {
-    OperationBuilder opBuilder = scope.env().opBuilder("TensorListConcatV2", scope.makeOpName("TensorListConcat"));
-    opBuilder.addInput(inputHandle.asOutput(scope));
-    opBuilder.addInput(elementShape.asOutput(scope));
-    opBuilder.addInput(leadingDims.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("element_dtype", elementDtype);
-    return new TensorListConcat<U>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<U> tensor() {
-    return tensor;
-  }
-  
-  /**
-   */
-  public Output<TInt64> lengths() {
-    return lengths;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "TensorListConcatV2";
-  
-  private Output<U> tensor;
-  private Output<TInt64> lengths;
-  
-  private TensorListConcat(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    tensor = operation.output(outputIdx++);
-    lengths = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListConcatLists.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListConcatLists.java
deleted file mode 100644
index 20a4ee32dff..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListConcatLists.java
+++ /dev/null
@@ -1,76 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- */
-@Operator
-public final class TensorListConcatLists extends RawOp implements Operand<TType> {
-  
-  /**
-   * Factory method to create a class wrapping a new TensorListConcatLists operation.
-   * 
-   * @param scope current scope
-   * @param inputA 
-   * @param inputB 
-   * @param elementDtype 
-   * @return a new instance of TensorListConcatLists
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> TensorListConcatLists create(Scope scope, Operand<?> inputA, Operand<?> inputB, Class<T> elementDtype) {
-    OperationBuilder opBuilder = scope.env().opBuilder("TensorListConcatLists", scope.makeOpName("TensorListConcatLists"));
-    opBuilder.addInput(inputA.asOutput(scope));
-    opBuilder.addInput(inputB.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("element_dtype", elementDtype);
-    return new TensorListConcatLists(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<?> output() {
-    return output;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "TensorListConcatLists";
-  
-  private Output<?> output;
-  
-  private TensorListConcatLists(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListElementShape.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListElementShape.java
deleted file mode 100644
index 0099f14585c..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListElementShape.java
+++ /dev/null
@@ -1,79 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * The shape of the elements of the given list, as a tensor.
- * <p>
- *   input_handle: the list
- *   element_shape: the shape of elements of the list
- * 
- * @param <T> data type for {@code elementShape()} output
- */
-@Operator
-public final class TensorListElementShape<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new TensorListElementShape operation.
-   * 
-   * @param scope current scope
-   * @param inputHandle 
-   * @param shapeType 
-   * @return a new instance of TensorListElementShape
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> TensorListElementShape<T> create(Scope scope, Operand<?> inputHandle, Class<T> shapeType) {
-    OperationBuilder opBuilder = scope.env().opBuilder("TensorListElementShape", scope.makeOpName("TensorListElementShape"));
-    opBuilder.addInput(inputHandle.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("shape_type", shapeType);
-    return new TensorListElementShape<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> elementShape() {
-    return elementShape;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return elementShape;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "TensorListElementShape";
-  
-  private Output<T> elementShape;
-  
-  private TensorListElementShape(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    elementShape = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListFromTensor.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListFromTensor.java
deleted file mode 100644
index d95c1718653..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListFromTensor.java
+++ /dev/null
@@ -1,81 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * Creates a TensorList which, when stacked, has the value of `tensor`.
- * <p>
- * Each tensor in the result list corresponds to one row of the input tensor.
- * <p>
- * tensor: The input tensor.
- * output_handle: The list.
- */
-@Operator
-public final class TensorListFromTensor extends RawOp implements Operand<TType> {
-  
-  /**
-   * Factory method to create a class wrapping a new TensorListFromTensor operation.
-   * 
-   * @param scope current scope
-   * @param tensor 
-   * @param elementShape 
-   * @return a new instance of TensorListFromTensor
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> TensorListFromTensor create(Scope scope, Operand<T> tensor, Operand<U> elementShape) {
-    OperationBuilder opBuilder = scope.env().opBuilder("TensorListFromTensor", scope.makeOpName("TensorListFromTensor"));
-    opBuilder.addInput(tensor.asOutput(scope));
-    opBuilder.addInput(elementShape.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new TensorListFromTensor(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<?> outputHandle() {
-    return outputHandle;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) outputHandle;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "TensorListFromTensor";
-  
-  private Output<?> outputHandle;
-  
-  private TensorListFromTensor(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    outputHandle = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListGather.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListGather.java
deleted file mode 100644
index 1062c76ad17..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListGather.java
+++ /dev/null
@@ -1,88 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.family.TType;
-
-/**
- * Creates a Tensor by indexing into the TensorList.
- * <p>
- * Each row in the produced Tensor corresponds to the element in the TensorList
- * specified by the given index (see `tf.gather`).
- * <p>
- * input_handle: The input tensor list.
- * indices: The indices used to index into the list.
- * values: The tensor.
- * 
- * @param <T> data type for {@code values()} output
- */
-@Operator
-public final class TensorListGather<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new TensorListGather operation.
-   * 
-   * @param scope current scope
-   * @param inputHandle 
-   * @param indices 
-   * @param elementShape 
-   * @param elementDtype 
-   * @return a new instance of TensorListGather
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> TensorListGather<T> create(Scope scope, Operand<?> inputHandle, Operand<TInt32> indices, Operand<TInt32> elementShape, Class<T> elementDtype) {
-    OperationBuilder opBuilder = scope.env().opBuilder("TensorListGather", scope.makeOpName("TensorListGather"));
-    opBuilder.addInput(inputHandle.asOutput(scope));
-    opBuilder.addInput(indices.asOutput(scope));
-    opBuilder.addInput(elementShape.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("element_dtype", elementDtype);
-    return new TensorListGather<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> values() {
-    return values;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return values;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "TensorListGather";
-  
-  private Output<T> values;
-  
-  private TensorListGather(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    values = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListGetItem.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListGetItem.java
deleted file mode 100644
index eb2ab5423eb..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListGetItem.java
+++ /dev/null
@@ -1,79 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.family.TType;
-
-/**
- * @param <T> data type for {@code item()} output
- */
-@Operator
-public final class TensorListGetItem<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new TensorListGetItem operation.
-   * 
-   * @param scope current scope
-   * @param inputHandle 
-   * @param index 
-   * @param elementShape 
-   * @param elementDtype 
-   * @return a new instance of TensorListGetItem
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> TensorListGetItem<T> create(Scope scope, Operand<?> inputHandle, Operand<TInt32> index, Operand<TInt32> elementShape, Class<T> elementDtype) {
-    OperationBuilder opBuilder = scope.env().opBuilder("TensorListGetItem", scope.makeOpName("TensorListGetItem"));
-    opBuilder.addInput(inputHandle.asOutput(scope));
-    opBuilder.addInput(index.asOutput(scope));
-    opBuilder.addInput(elementShape.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("element_dtype", elementDtype);
-    return new TensorListGetItem<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> item() {
-    return item;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return item;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "TensorListGetItem";
-  
-  private Output<T> item;
-  
-  private TensorListGetItem(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    item = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListLength.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListLength.java
deleted file mode 100644
index 2b4e4055630..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListLength.java
+++ /dev/null
@@ -1,75 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
-
-/**
- * Returns the number of tensors in the input tensor list.
- * <p>
- * input_handle: the input list
- * length: the number of tensors in the list
- */
-@Operator
-public final class TensorListLength extends RawOp implements Operand<TInt32> {
-  
-  /**
-   * Factory method to create a class wrapping a new TensorListLength operation.
-   * 
-   * @param scope current scope
-   * @param inputHandle 
-   * @return a new instance of TensorListLength
-   */
-  @Endpoint(describeByClass = true)
-  public static TensorListLength create(Scope scope, Operand<?> inputHandle) {
-    OperationBuilder opBuilder = scope.env().opBuilder("TensorListLength", scope.makeOpName("TensorListLength"));
-    opBuilder.addInput(inputHandle.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new TensorListLength(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<TInt32> length() {
-    return length;
-  }
-  
-  @Override
-  public Output<TInt32> asOutput(Scope scope) {
-    return length;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "TensorListLength";
-  
-  private Output<TInt32> length;
-  
-  private TensorListLength(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    length = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListPopBack.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListPopBack.java
deleted file mode 100644
index d2df132000c..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListPopBack.java
+++ /dev/null
@@ -1,89 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.family.TType;
-
-/**
- * Returns the last element of the input list as well as a list with all but that element.
- * <p>
- * Fails if the list is empty.
- * <p>
- * input_handle: the input list
- * tensor: the withdrawn last element of the list
- * element_dtype: the type of elements in the list
- * element_shape: the shape of the output tensor
- * 
- * @param <T> data type for {@code tensor()} output
- */
-@Operator
-public final class TensorListPopBack<T extends TType> extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new TensorListPopBack operation.
-   * 
-   * @param scope current scope
-   * @param inputHandle 
-   * @param elementShape 
-   * @param elementDtype 
-   * @return a new instance of TensorListPopBack
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> TensorListPopBack<T> create(Scope scope, Operand<?> inputHandle, Operand<TInt32> elementShape, Class<T> elementDtype) {
-    OperationBuilder opBuilder = scope.env().opBuilder("TensorListPopBack", scope.makeOpName("TensorListPopBack"));
-    opBuilder.addInput(inputHandle.asOutput(scope));
-    opBuilder.addInput(elementShape.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("element_dtype", elementDtype);
-    return new TensorListPopBack<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<?> outputHandle() {
-    return outputHandle;
-  }
-  
-  /**
-   */
-  public Output<T> tensor() {
-    return tensor;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "TensorListPopBack";
-  
-  private Output<?> outputHandle;
-  private Output<T> tensor;
-  
-  private TensorListPopBack(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    outputHandle = operation.output(outputIdx++);
-    tensor = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListPushBack.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListPushBack.java
deleted file mode 100644
index ad8bac3d00a..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListPushBack.java
+++ /dev/null
@@ -1,81 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Returns a list which has the passed-in `Tensor` as last element and the other elements of the given list in `input_handle`.
- * <p>
- * tensor: The tensor to put on the list.
- * input_handle: The old list.
- * output_handle: A list with the elements of the old list followed by tensor.
- * element_dtype: the type of elements in the list.
- * element_shape: a shape compatible with that of elements in the list.
- */
-@Operator
-public final class TensorListPushBack extends RawOp implements Operand<TType> {
-  
-  /**
-   * Factory method to create a class wrapping a new TensorListPushBack operation.
-   * 
-   * @param scope current scope
-   * @param inputHandle 
-   * @param tensor 
-   * @return a new instance of TensorListPushBack
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> TensorListPushBack create(Scope scope, Operand<?> inputHandle, Operand<T> tensor) {
-    OperationBuilder opBuilder = scope.env().opBuilder("TensorListPushBack", scope.makeOpName("TensorListPushBack"));
-    opBuilder.addInput(inputHandle.asOutput(scope));
-    opBuilder.addInput(tensor.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new TensorListPushBack(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<?> outputHandle() {
-    return outputHandle;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) outputHandle;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "TensorListPushBack";
-  
-  private Output<?> outputHandle;
-  
-  private TensorListPushBack(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    outputHandle = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListPushBackBatch.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListPushBackBatch.java
deleted file mode 100644
index 74f3ca489fd..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListPushBackBatch.java
+++ /dev/null
@@ -1,74 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- */
-@Operator
-public final class TensorListPushBackBatch extends RawOp implements Operand<TType> {
-  
-  /**
-   * Factory method to create a class wrapping a new TensorListPushBackBatch operation.
-   * 
-   * @param scope current scope
-   * @param inputHandles 
-   * @param tensor 
-   * @return a new instance of TensorListPushBackBatch
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> TensorListPushBackBatch create(Scope scope, Operand<?> inputHandles, Operand<T> tensor) {
-    OperationBuilder opBuilder = scope.env().opBuilder("TensorListPushBackBatch", scope.makeOpName("TensorListPushBackBatch"));
-    opBuilder.addInput(inputHandles.asOutput(scope));
-    opBuilder.addInput(tensor.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new TensorListPushBackBatch(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<?> outputHandles() {
-    return outputHandles;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) outputHandles;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "TensorListPushBackBatch";
-  
-  private Output<?> outputHandles;
-  
-  private TensorListPushBackBatch(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    outputHandles = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListReserve.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListReserve.java
deleted file mode 100644
index 13f68b7492f..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListReserve.java
+++ /dev/null
@@ -1,84 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * List of the given size with empty elements.
- * <p>
- * element_shape: the shape of the future elements of the list
- * num_elements: the number of elements to reserve
- * handle: the output list
- * element_dtype: the desired type of elements in the list.
- */
-@Operator
-public final class TensorListReserve extends RawOp implements Operand<TType> {
-  
-  /**
-   * Factory method to create a class wrapping a new TensorListReserve operation.
-   * 
-   * @param scope current scope
-   * @param elementShape 
-   * @param numElements 
-   * @param elementDtype 
-   * @return a new instance of TensorListReserve
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber, U extends TType> TensorListReserve create(Scope scope, Operand<T> elementShape, Operand<TInt32> numElements, Class<U> elementDtype) {
-    OperationBuilder opBuilder = scope.env().opBuilder("TensorListReserve", scope.makeOpName("TensorListReserve"));
-    opBuilder.addInput(elementShape.asOutput(scope));
-    opBuilder.addInput(numElements.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("element_dtype", elementDtype);
-    return new TensorListReserve(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<?> handle() {
-    return handle;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) handle;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "TensorListReserve";
-  
-  private Output<?> handle;
-  
-  private TensorListReserve(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    handle = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListResize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListResize.java
deleted file mode 100644
index 9bd592af106..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListResize.java
+++ /dev/null
@@ -1,81 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.family.TType;
-
-/**
- * Resizes the list.
- * <p>
- * 
- * input_handle: the input list
- * size: size of the output list
- * 
- */
-@Operator
-public final class TensorListResize extends RawOp implements Operand<TType> {
-  
-  /**
-   * Factory method to create a class wrapping a new TensorListResize operation.
-   * 
-   * @param scope current scope
-   * @param inputHandle 
-   * @param size 
-   * @return a new instance of TensorListResize
-   */
-  @Endpoint(describeByClass = true)
-  public static TensorListResize create(Scope scope, Operand<?> inputHandle, Operand<TInt32> size) {
-    OperationBuilder opBuilder = scope.env().opBuilder("TensorListResize", scope.makeOpName("TensorListResize"));
-    opBuilder.addInput(inputHandle.asOutput(scope));
-    opBuilder.addInput(size.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new TensorListResize(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<?> outputHandle() {
-    return outputHandle;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) outputHandle;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "TensorListResize";
-  
-  private Output<?> outputHandle;
-  
-  private TensorListResize(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    outputHandle = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListScatter.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListScatter.java
deleted file mode 100644
index 173ccd4c7f9..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListScatter.java
+++ /dev/null
@@ -1,93 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * Creates a TensorList by indexing into a Tensor.
- * <p>
- * Each member of the TensorList corresponds to one row of the input tensor,
- * specified by the given index (see `tf.gather`).
- * <p>
- * tensor: The input tensor.
- * indices: The indices used to index into the list.
- * element_shape: The shape of the elements in the list (can be less specified than
- *   the shape of the tensor).
- * num_elements: The size of the output list. Must be large enough to accommodate
- *   the largest index in indices. If -1, the list is just large enough to include
- *   the largest index in indices.
- * output_handle: The TensorList.
- */
-@Operator
-public final class TensorListScatter extends RawOp implements Operand<TType> {
-  
-  /**
-   * Factory method to create a class wrapping a new TensorListScatter operation.
-   * 
-   * @param scope current scope
-   * @param tensor 
-   * @param indices 
-   * @param elementShape 
-   * @param numElements 
-   * @return a new instance of TensorListScatter
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> TensorListScatter create(Scope scope, Operand<T> tensor, Operand<TInt32> indices, Operand<U> elementShape, Operand<TInt32> numElements) {
-    OperationBuilder opBuilder = scope.env().opBuilder("TensorListScatterV2", scope.makeOpName("TensorListScatter"));
-    opBuilder.addInput(tensor.asOutput(scope));
-    opBuilder.addInput(indices.asOutput(scope));
-    opBuilder.addInput(elementShape.asOutput(scope));
-    opBuilder.addInput(numElements.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new TensorListScatter(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<?> outputHandle() {
-    return outputHandle;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) outputHandle;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "TensorListScatterV2";
-  
-  private Output<?> outputHandle;
-  
-  private TensorListScatter(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    outputHandle = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListScatterIntoExistingList.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListScatterIntoExistingList.java
deleted file mode 100644
index 7350939b6d7..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListScatterIntoExistingList.java
+++ /dev/null
@@ -1,86 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.family.TType;
-
-/**
- * Scatters tensor at indices in an input list.
- * <p>
- * Each member of the TensorList corresponds to one row of the input tensor,
- * specified by the given index (see `tf.gather`).
- * <p>
- * input_handle: The list to scatter into.
- * tensor: The input tensor.
- * indices: The indices used to index into the list.
- * output_handle: The TensorList.
- */
-@Operator
-public final class TensorListScatterIntoExistingList extends RawOp implements Operand<TType> {
-  
-  /**
-   * Factory method to create a class wrapping a new TensorListScatterIntoExistingList operation.
-   * 
-   * @param scope current scope
-   * @param inputHandle 
-   * @param tensor 
-   * @param indices 
-   * @return a new instance of TensorListScatterIntoExistingList
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> TensorListScatterIntoExistingList create(Scope scope, Operand<?> inputHandle, Operand<T> tensor, Operand<TInt32> indices) {
-    OperationBuilder opBuilder = scope.env().opBuilder("TensorListScatterIntoExistingList", scope.makeOpName("TensorListScatterIntoExistingList"));
-    opBuilder.addInput(inputHandle.asOutput(scope));
-    opBuilder.addInput(tensor.asOutput(scope));
-    opBuilder.addInput(indices.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new TensorListScatterIntoExistingList(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<?> outputHandle() {
-    return outputHandle;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) outputHandle;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "TensorListScatterIntoExistingList";
-  
-  private Output<?> outputHandle;
-  
-  private TensorListScatterIntoExistingList(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    outputHandle = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListSetItem.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListSetItem.java
deleted file mode 100644
index 8a3ceb52d01..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListSetItem.java
+++ /dev/null
@@ -1,77 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.family.TType;
-
-/**
- */
-@Operator
-public final class TensorListSetItem extends RawOp implements Operand<TType> {
-  
-  /**
-   * Factory method to create a class wrapping a new TensorListSetItem operation.
-   * 
-   * @param scope current scope
-   * @param inputHandle 
-   * @param index 
-   * @param item 
-   * @return a new instance of TensorListSetItem
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> TensorListSetItem create(Scope scope, Operand<?> inputHandle, Operand<TInt32> index, Operand<T> item) {
-    OperationBuilder opBuilder = scope.env().opBuilder("TensorListSetItem", scope.makeOpName("TensorListSetItem"));
-    opBuilder.addInput(inputHandle.asOutput(scope));
-    opBuilder.addInput(index.asOutput(scope));
-    opBuilder.addInput(item.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new TensorListSetItem(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<?> outputHandle() {
-    return outputHandle;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) outputHandle;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "TensorListSetItem";
-  
-  private Output<?> outputHandle;
-  
-  private TensorListSetItem(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    outputHandle = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListSplit.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListSplit.java
deleted file mode 100644
index 209eb066259..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListSplit.java
+++ /dev/null
@@ -1,87 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * Splits a tensor into a list.
- * <p>
- * list[i] corresponds to lengths[i] tensors from the input tensor.
- * The tensor must have rank at least 1 and contain exactly sum(lengths) elements.
- * <p>
- * tensor: The input tensor.
- * element_shape: A shape compatible with that of elements in the tensor.
- * lengths: Vector of sizes of the 0th dimension of tensors in the list.
- * output_handle: The list.
- */
-@Operator
-public final class TensorListSplit extends RawOp implements Operand<TType> {
-  
-  /**
-   * Factory method to create a class wrapping a new TensorListSplit operation.
-   * 
-   * @param scope current scope
-   * @param tensor 
-   * @param elementShape 
-   * @param lengths 
-   * @return a new instance of TensorListSplit
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> TensorListSplit create(Scope scope, Operand<T> tensor, Operand<U> elementShape, Operand<TInt64> lengths) {
-    OperationBuilder opBuilder = scope.env().opBuilder("TensorListSplit", scope.makeOpName("TensorListSplit"));
-    opBuilder.addInput(tensor.asOutput(scope));
-    opBuilder.addInput(elementShape.asOutput(scope));
-    opBuilder.addInput(lengths.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new TensorListSplit(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<?> outputHandle() {
-    return outputHandle;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) outputHandle;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "TensorListSplit";
-  
-  private Output<?> outputHandle;
-  
-  private TensorListSplit(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    outputHandle = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListStack.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListStack.java
deleted file mode 100644
index 42dc682cda6..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListStack.java
+++ /dev/null
@@ -1,120 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.family.TType;
-
-/**
- * Stacks all tensors in the list.
- * <p>
- * Requires that all tensors have the same shape.
- * <p>
- * input_handle: the input list
- * tensor: the gathered result
- * num_elements: optional. If not -1, the number of elements in the list.
- * 
- * 
- * @param <T> data type for {@code tensor()} output
- */
-@Operator
-public final class TensorListStack<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.core.TensorListStack}
-   */
-  public static class Options {
-    
-    /**
-     * @param numElements 
-     */
-    public Options numElements(Long numElements) {
-      this.numElements = numElements;
-      return this;
-    }
-    
-    private Long numElements;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new TensorListStack operation.
-   * 
-   * @param scope current scope
-   * @param inputHandle 
-   * @param elementShape 
-   * @param elementDtype 
-   * @param options carries optional attributes values
-   * @return a new instance of TensorListStack
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> TensorListStack<T> create(Scope scope, Operand<?> inputHandle, Operand<TInt32> elementShape, Class<T> elementDtype, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("TensorListStack", scope.makeOpName("TensorListStack"));
-    opBuilder.addInput(inputHandle.asOutput(scope));
-    opBuilder.addInput(elementShape.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("element_dtype", elementDtype);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.numElements != null) {
-          opBuilder.setAttr("num_elements", opts.numElements);
-        }
-      }
-    }
-    return new TensorListStack<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param numElements 
-   */
-  public static Options numElements(Long numElements) {
-    return new Options().numElements(numElements);
-  }
-  
-  /**
-   */
-  public Output<T> tensor() {
-    return tensor;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return tensor;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "TensorListStack";
-  
-  private Output<T> tensor;
-  
-  private TensorListStack(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    tensor = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorScatterNdAdd.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorScatterNdAdd.java
deleted file mode 100644
index 85647215fa2..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorScatterNdAdd.java
+++ /dev/null
@@ -1,140 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * Adds sparse `updates` to an existing tensor according to `indices`.
- * <p>
- * This operation creates a new tensor by adding sparse `updates` to the passed
- * in `tensor`.
- * This operation is very similar to `tf.scatter_nd_add`, except that the updates
- * are added onto an existing tensor (as opposed to a variable). If the memory
- * for the existing tensor cannot be re-used, a copy is made and updated.
- * <p>
- * `indices` is an integer tensor containing indices into a new tensor of shape
- * `tensor.shape`.  The last dimension of `indices` can be at most the rank of
- * `tensor.shape`:
- * <p>
- *     indices.shape[-1] <= tensor.shape.rank
- * <p>
- * The last dimension of `indices` corresponds to indices into elements
- * (if `indices.shape[-1] = tensor.shape.rank`) or slices
- * (if `indices.shape[-1] < tensor.shape.rank`) along dimension
- * `indices.shape[-1]` of `tensor.shape`.  `updates` is a tensor with shape
- * <p>
- *     indices.shape[:-1] + tensor.shape[indices.shape[-1]:]
- * <p>
- * The simplest form of tensor_scatter_add is to add individual elements to a
- * tensor by index. For example, say we want to add 4 elements in a rank-1
- * tensor with 8 elements.
- * <p>
- * In Python, this scatter add operation would look like this:
- * <pre>{@code
- *     indices = tf.constant([[4], [3], [1], [7]])
- *     updates = tf.constant([9, 10, 11, 12])
- *     tensor = tf.ones([8], dtype=tf.int32)
- *     updated = tf.tensor_scatter_nd_add(tensor, indices, updates)
- *     print(updated)
- * }</pre>
- * The resulting tensor would look like this:
- * <p>
- *     [1, 12, 1, 11, 10, 1, 1, 13]
- * <p>
- * We can also, insert entire slices of a higher rank tensor all at once. For
- * example, if we wanted to insert two slices in the first dimension of a
- * rank-3 tensor with two matrices of new values.
- * <p>
- * In Python, this scatter add operation would look like this:
- * <pre>{@code
- *     indices = tf.constant([[0], [2]])
- *     updates = tf.constant([[[5, 5, 5, 5], [6, 6, 6, 6],
- *                             [7, 7, 7, 7], [8, 8, 8, 8]],
- *                            [[5, 5, 5, 5], [6, 6, 6, 6],
- *                             [7, 7, 7, 7], [8, 8, 8, 8]]])
- *     tensor = tf.ones([4, 4, 4],dtype=tf.int32)
- *     updated = tf.tensor_scatter_nd_add(tensor, indices, updates)
- *     print(updated)
- * }</pre>
- * The resulting tensor would look like this:
- * <p>
- *     [[[6, 6, 6, 6], [7, 7, 7, 7], [8, 8, 8, 8], [9, 9, 9, 9]],
- *      [[1, 1, 1, 1], [1, 1, 1, 1], [1, 1, 1, 1], [1, 1, 1, 1]],
- *      [[6, 6, 6, 6], [7, 7, 7, 7], [8, 8, 8, 8], [9, 9, 9, 9]],
- *      [[1, 1, 1, 1], [1, 1, 1, 1], [1, 1, 1, 1], [1, 1, 1, 1]]]
- * <p>
- * Note that on CPU, if an out of bound index is found, an error is returned.
- * On GPU, if an out of bound index is found, the index is ignored.
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator
-public final class TensorScatterNdAdd<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new TensorScatterNdAdd operation.
-   * 
-   * @param scope current scope
-   * @param tensor Tensor to copy/update.
-   * @param indices Index tensor.
-   * @param updates Updates to scatter into output.
-   * @return a new instance of TensorScatterNdAdd
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> TensorScatterNdAdd<T> create(Scope scope, Operand<T> tensor, Operand<U> indices, Operand<T> updates) {
-    OperationBuilder opBuilder = scope.env().opBuilder("TensorScatterAdd", scope.makeOpName("TensorScatterNdAdd"));
-    opBuilder.addInput(tensor.asOutput(scope));
-    opBuilder.addInput(indices.asOutput(scope));
-    opBuilder.addInput(updates.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new TensorScatterNdAdd<T>(opBuilder.build());
-  }
-  
-  /**
-   * A new tensor copied from tensor and updates added according to the indices.
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "TensorScatterAdd";
-  
-  private Output<T> output;
-  
-  private TensorScatterNdAdd(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorScatterNdMax.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorScatterNdMax.java
deleted file mode 100644
index 55d07a26a4e..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorScatterNdMax.java
+++ /dev/null
@@ -1,78 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * @param <T> data type for {@code output()} output
- */
-@Operator
-public final class TensorScatterNdMax<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new TensorScatterNdMax operation.
-   * 
-   * @param scope current scope
-   * @param tensor Tensor to update.
-   * @param indices Index tensor.
-   * @param updates Updates to scatter into output.
-   * @return a new instance of TensorScatterNdMax
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> TensorScatterNdMax<T> create(Scope scope, Operand<T> tensor, Operand<U> indices, Operand<T> updates) {
-    OperationBuilder opBuilder = scope.env().opBuilder("TensorScatterMax", scope.makeOpName("TensorScatterNdMax"));
-    opBuilder.addInput(tensor.asOutput(scope));
-    opBuilder.addInput(indices.asOutput(scope));
-    opBuilder.addInput(updates.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new TensorScatterNdMax<T>(opBuilder.build());
-  }
-  
-  /**
-   * A new tensor copied from tensor whose values are element-wise maximum between tensor and updates according to the indices.
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "TensorScatterMax";
-  
-  private Output<T> output;
-  
-  private TensorScatterNdMax(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorScatterNdMin.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorScatterNdMin.java
deleted file mode 100644
index a021b1cfc03..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorScatterNdMin.java
+++ /dev/null
@@ -1,78 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * @param <T> data type for {@code output()} output
- */
-@Operator
-public final class TensorScatterNdMin<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new TensorScatterNdMin operation.
-   * 
-   * @param scope current scope
-   * @param tensor Tensor to update.
-   * @param indices Index tensor.
-   * @param updates Updates to scatter into output.
-   * @return a new instance of TensorScatterNdMin
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> TensorScatterNdMin<T> create(Scope scope, Operand<T> tensor, Operand<U> indices, Operand<T> updates) {
-    OperationBuilder opBuilder = scope.env().opBuilder("TensorScatterMin", scope.makeOpName("TensorScatterNdMin"));
-    opBuilder.addInput(tensor.asOutput(scope));
-    opBuilder.addInput(indices.asOutput(scope));
-    opBuilder.addInput(updates.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new TensorScatterNdMin<T>(opBuilder.build());
-  }
-  
-  /**
-   * A new tensor copied from tensor whose values are element-wise minimum between tensor and updates according to the indices.
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "TensorScatterMin";
-  
-  private Output<T> output;
-  
-  private TensorScatterNdMin(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorScatterNdSub.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorScatterNdSub.java
deleted file mode 100644
index a97d99fc197..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorScatterNdSub.java
+++ /dev/null
@@ -1,139 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * Subtracts sparse `updates` from an existing tensor according to `indices`.
- * <p>
- * This operation creates a new tensor by subtracting sparse `updates` from the
- * passed in `tensor`.
- * This operation is very similar to `tf.scatter_nd_sub`, except that the updates
- * are subtracted from an existing tensor (as opposed to a variable). If the memory
- * for the existing tensor cannot be re-used, a copy is made and updated.
- * <p>
- * `indices` is an integer tensor containing indices into a new tensor of shape
- * `shape`.  The last dimension of `indices` can be at most the rank of `shape`:
- * <p>
- *     indices.shape[-1] <= shape.rank
- * <p>
- * The last dimension of `indices` corresponds to indices into elements
- * (if `indices.shape[-1] = shape.rank`) or slices
- * (if `indices.shape[-1] < shape.rank`) along dimension `indices.shape[-1]` of
- * `shape`.  `updates` is a tensor with shape
- * <p>
- *     indices.shape[:-1] + shape[indices.shape[-1]:]
- * <p>
- * The simplest form of tensor_scatter_sub is to subtract individual elements
- * from a tensor by index. For example, say we want to insert 4 scattered elements
- * in a rank-1 tensor with 8 elements.
- * <p>
- * In Python, this scatter subtract operation would look like this:
- * <pre>{@code
- *     indices = tf.constant([[4], [3], [1], [7]])
- *     updates = tf.constant([9, 10, 11, 12])
- *     tensor = tf.ones([8], dtype=tf.int32)
- *     updated = tf.tensor_scatter_nd_sub(tensor, indices, updates)
- *     print(updated)
- * }</pre>
- * The resulting tensor would look like this:
- * <p>
- *     [1, -10, 1, -9, -8, 1, 1, -11]
- * <p>
- * We can also, insert entire slices of a higher rank tensor all at once. For
- * example, if we wanted to insert two slices in the first dimension of a
- * rank-3 tensor with two matrices of new values.
- * <p>
- * In Python, this scatter add operation would look like this:
- * <pre>{@code
- *     indices = tf.constant([[0], [2]])
- *     updates = tf.constant([[[5, 5, 5, 5], [6, 6, 6, 6],
- *                             [7, 7, 7, 7], [8, 8, 8, 8]],
- *                            [[5, 5, 5, 5], [6, 6, 6, 6],
- *                             [7, 7, 7, 7], [8, 8, 8, 8]]])
- *     tensor = tf.ones([4, 4, 4],dtype=tf.int32)
- *     updated = tf.tensor_scatter_nd_sub(tensor, indices, updates)
- *     print(updated)
- * }</pre>
- * The resulting tensor would look like this:
- * <p>
- *     [[[-4, -4, -4, -4], [-5, -5, -5, -5], [-6, -6, -6, -6], [-7, -7, -7, -7]],
- *      [[1, 1, 1, 1], [1, 1, 1, 1], [1, 1, 1, 1], [1, 1, 1, 1]],
- *      [[-4, -4, -4, -4], [-5, -5, -5, -5], [-6, -6, -6, -6], [-7, -7, -7, -7]],
- *      [[1, 1, 1, 1], [1, 1, 1, 1], [1, 1, 1, 1], [1, 1, 1, 1]]]
- * <p>
- * Note that on CPU, if an out of bound index is found, an error is returned.
- * On GPU, if an out of bound index is found, the index is ignored.
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator
-public final class TensorScatterNdSub<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new TensorScatterNdSub operation.
-   * 
-   * @param scope current scope
-   * @param tensor Tensor to copy/update.
-   * @param indices Index tensor.
-   * @param updates Updates to scatter into output.
-   * @return a new instance of TensorScatterNdSub
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> TensorScatterNdSub<T> create(Scope scope, Operand<T> tensor, Operand<U> indices, Operand<T> updates) {
-    OperationBuilder opBuilder = scope.env().opBuilder("TensorScatterSub", scope.makeOpName("TensorScatterNdSub"));
-    opBuilder.addInput(tensor.asOutput(scope));
-    opBuilder.addInput(indices.asOutput(scope));
-    opBuilder.addInput(updates.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new TensorScatterNdSub<T>(opBuilder.build());
-  }
-  
-  /**
-   * A new tensor copied from tensor and updates subtracted according to the indices.
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "TensorScatterSub";
-  
-  private Output<T> output;
-  
-  private TensorScatterNdSub(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorScatterNdUpdate.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorScatterNdUpdate.java
deleted file mode 100644
index 0c4cf483e57..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorScatterNdUpdate.java
+++ /dev/null
@@ -1,155 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * Scatter `updates` into an existing tensor according to `indices`.
- * <p>
- * This operation creates a new tensor by applying sparse `updates` to the passed
- * in `tensor`.
- * This operation is very similar to `tf.scatter_nd`, except that the updates are
- * scattered onto an existing tensor (as opposed to a zero-tensor). If the memory
- * for the existing tensor cannot be re-used, a copy is made and updated.
- * <p>
- * If `indices` contains duplicates, then their updates are accumulated (summed).
- * <p>
- * <b>WARNING</b>: The order in which updates are applied is nondeterministic, so the
- * output will be nondeterministic if `indices` contains duplicates -- because
- * of some numerical approximation issues, numbers summed in different order
- * may yield different results.
- * <p>
- * `indices` is an integer tensor containing indices into a new tensor of shape
- * `shape`.  The last dimension of `indices` can be at most the rank of `shape`:
- * <p>
- *     indices.shape[-1] <= shape.rank
- * <p>
- * The last dimension of `indices` corresponds to indices into elements
- * (if `indices.shape[-1] = shape.rank`) or slices
- * (if `indices.shape[-1] < shape.rank`) along dimension `indices.shape[-1]` of
- * `shape`.  `updates` is a tensor with shape
- * <p>
- *     indices.shape[:-1] + shape[indices.shape[-1]:]
- * <p>
- * The simplest form of scatter is to insert individual elements in a tensor by
- * index. For example, say we want to insert 4 scattered elements in a rank-1
- * tensor with 8 elements.
- * <p>
- * <div style="width:70%; margin:auto; margin-bottom:10px; margin-top:20px;">
- * <img style="width:100%" src="https://www.tensorflow.org/images/ScatterNd1.png" alt>
- * </div>
- * <p>
- * In Python, this scatter operation would look like this:
- * <p>
- *     >>> indices = tf.constant([[4], [3], [1], [7]])
- *     >>> updates = tf.constant([9, 10, 11, 12])
- *     >>> tensor = tf.ones([8], dtype=tf.int32)
- *     >>> print(tf.tensor_scatter_nd_update(tensor, indices, updates))
- *     tf.Tensor([ 1 11  1 10  9  1  1 12], shape=(8,), dtype=int32)
- * <p>
- * We can also, insert entire slices of a higher rank tensor all at once. For
- * example, if we wanted to insert two slices in the first dimension of a
- * rank-3 tensor with two matrices of new values.
- * <p>
- * In Python, this scatter operation would look like this:
- * <p>
- *     >>> indices = tf.constant([[0], [2]])
- *     >>> updates = tf.constant([[[5, 5, 5, 5], [6, 6, 6, 6],
- *     ...                         [7, 7, 7, 7], [8, 8, 8, 8]],
- *     ...                        [[5, 5, 5, 5], [6, 6, 6, 6],
- *     ...                         [7, 7, 7, 7], [8, 8, 8, 8]]])
- *     >>> tensor = tf.ones([4, 4, 4], dtype=tf.int32)
- *     >>> print(tf.tensor_scatter_nd_update(tensor, indices, updates).numpy())
- *     [[[5 5 5 5]
- *       [6 6 6 6]
- *       [7 7 7 7]
- *       [8 8 8 8]]
- *      [[1 1 1 1]
- *       [1 1 1 1]
- *       [1 1 1 1]
- *       [1 1 1 1]]
- *      [[5 5 5 5]
- *       [6 6 6 6]
- *       [7 7 7 7]
- *       [8 8 8 8]]
- *      [[1 1 1 1]
- *       [1 1 1 1]
- *       [1 1 1 1]
- *       [1 1 1 1]]]
- * <p>
- * Note that on CPU, if an out of bound index is found, an error is returned.
- * On GPU, if an out of bound index is found, the index is ignored.
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator
-public final class TensorScatterNdUpdate<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new TensorScatterNdUpdate operation.
-   * 
-   * @param scope current scope
-   * @param tensor Tensor to copy/update.
-   * @param indices Index tensor.
-   * @param updates Updates to scatter into output.
-   * @return a new instance of TensorScatterNdUpdate
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> TensorScatterNdUpdate<T> create(Scope scope, Operand<T> tensor, Operand<U> indices, Operand<T> updates) {
-    OperationBuilder opBuilder = scope.env().opBuilder("TensorScatterUpdate", scope.makeOpName("TensorScatterNdUpdate"));
-    opBuilder.addInput(tensor.asOutput(scope));
-    opBuilder.addInput(indices.asOutput(scope));
-    opBuilder.addInput(updates.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new TensorScatterNdUpdate<T>(opBuilder.build());
-  }
-  
-  /**
-   * A new tensor with the given shape and updates applied according
-   * to the indices.
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "TensorScatterUpdate";
-  
-  private Output<T> output;
-  
-  private TensorScatterNdUpdate(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorStridedSliceUpdate.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorStridedSliceUpdate.java
deleted file mode 100644
index ff0777f607d..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorStridedSliceUpdate.java
+++ /dev/null
@@ -1,200 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * Assign `value` to the sliced l-value reference of `input`.
- * <p>
- * The values of `value` are assigned to the positions in the tensor `input` that
- * are selected by the slice parameters. The slice parameters `begin` `end`
- * `strides` etc. work exactly as in `StridedSlice`.
- * <p>
- * NOTE this op currently does not support broadcasting and so `value`'s shape
- * must be exactly the shape produced by the slice of `input`.
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator
-public final class TensorStridedSliceUpdate<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.core.TensorStridedSliceUpdate}
-   */
-  public static class Options {
-    
-    /**
-     * @param beginMask 
-     */
-    public Options beginMask(Long beginMask) {
-      this.beginMask = beginMask;
-      return this;
-    }
-    
-    /**
-     * @param endMask 
-     */
-    public Options endMask(Long endMask) {
-      this.endMask = endMask;
-      return this;
-    }
-    
-    /**
-     * @param ellipsisMask 
-     */
-    public Options ellipsisMask(Long ellipsisMask) {
-      this.ellipsisMask = ellipsisMask;
-      return this;
-    }
-    
-    /**
-     * @param newAxisMask 
-     */
-    public Options newAxisMask(Long newAxisMask) {
-      this.newAxisMask = newAxisMask;
-      return this;
-    }
-    
-    /**
-     * @param shrinkAxisMask 
-     */
-    public Options shrinkAxisMask(Long shrinkAxisMask) {
-      this.shrinkAxisMask = shrinkAxisMask;
-      return this;
-    }
-    
-    private Long beginMask;
-    private Long endMask;
-    private Long ellipsisMask;
-    private Long newAxisMask;
-    private Long shrinkAxisMask;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new TensorStridedSliceUpdate operation.
-   * 
-   * @param scope current scope
-   * @param input 
-   * @param begin 
-   * @param end 
-   * @param strides 
-   * @param value 
-   * @param options carries optional attributes values
-   * @return a new instance of TensorStridedSliceUpdate
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> TensorStridedSliceUpdate<T> create(Scope scope, Operand<T> input, Operand<U> begin, Operand<U> end, Operand<U> strides, Operand<T> value, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("TensorStridedSliceUpdate", scope.makeOpName("TensorStridedSliceUpdate"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder.addInput(begin.asOutput(scope));
-    opBuilder.addInput(end.asOutput(scope));
-    opBuilder.addInput(strides.asOutput(scope));
-    opBuilder.addInput(value.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.beginMask != null) {
-          opBuilder.setAttr("begin_mask", opts.beginMask);
-        }
-        if (opts.endMask != null) {
-          opBuilder.setAttr("end_mask", opts.endMask);
-        }
-        if (opts.ellipsisMask != null) {
-          opBuilder.setAttr("ellipsis_mask", opts.ellipsisMask);
-        }
-        if (opts.newAxisMask != null) {
-          opBuilder.setAttr("new_axis_mask", opts.newAxisMask);
-        }
-        if (opts.shrinkAxisMask != null) {
-          opBuilder.setAttr("shrink_axis_mask", opts.shrinkAxisMask);
-        }
-      }
-    }
-    return new TensorStridedSliceUpdate<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param beginMask 
-   */
-  public static Options beginMask(Long beginMask) {
-    return new Options().beginMask(beginMask);
-  }
-  
-  /**
-   * @param endMask 
-   */
-  public static Options endMask(Long endMask) {
-    return new Options().endMask(endMask);
-  }
-  
-  /**
-   * @param ellipsisMask 
-   */
-  public static Options ellipsisMask(Long ellipsisMask) {
-    return new Options().ellipsisMask(ellipsisMask);
-  }
-  
-  /**
-   * @param newAxisMask 
-   */
-  public static Options newAxisMask(Long newAxisMask) {
-    return new Options().newAxisMask(newAxisMask);
-  }
-  
-  /**
-   * @param shrinkAxisMask 
-   */
-  public static Options shrinkAxisMask(Long shrinkAxisMask) {
-    return new Options().shrinkAxisMask(shrinkAxisMask);
-  }
-  
-  /**
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "TensorStridedSliceUpdate";
-  
-  private Output<T> output;
-  
-  private TensorStridedSliceUpdate(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Tile.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Tile.java
deleted file mode 100644
index 82749c7897e..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Tile.java
+++ /dev/null
@@ -1,104 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * Constructs a tensor by tiling a given tensor.
- * <p>
- * This operation creates a new tensor by replicating `input` `multiples` times.
- * The output tensor's i'th dimension has `input.dims(i) * multiples[i]` elements,
- * and the values of `input` are replicated `multiples[i]` times along the 'i'th
- * dimension. For example, tiling `[a b c d]` by `[2]` produces
- * `[a b c d a b c d]`.
- * <p>
- * >>> a = tf.constant([[1,2,3],[4,5,6]], tf.int32)
- * >>> b = tf.constant([1,2], tf.int32)
- * >>> tf.tile(a, b)
- * <tf.Tensor: shape=(2, 6), dtype=int32, numpy=
- * array([[1, 2, 3, 1, 2, 3],
- *        [4, 5, 6, 4, 5, 6]], dtype=int32)>
- * >>> c = tf.constant([2,1], tf.int32)
- * >>> tf.tile(a, c)
- * <tf.Tensor: shape=(4, 3), dtype=int32, numpy=
- * array([[1, 2, 3],
- *        [4, 5, 6],
- *        [1, 2, 3],
- *        [4, 5, 6]], dtype=int32)>
- * >>> d = tf.constant([2,2], tf.int32)
- * >>> tf.tile(a, d)
- * <tf.Tensor: shape=(4, 6), dtype=int32, numpy=
- * array([[1, 2, 3, 1, 2, 3],
- *        [4, 5, 6, 4, 5, 6],
- *        [1, 2, 3, 1, 2, 3],
- *        [4, 5, 6, 4, 5, 6]], dtype=int32)>
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator
-public final class Tile<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new Tile operation.
-   * 
-   * @param scope current scope
-   * @param input 1-D or higher.
-   * @param multiples 1-D. Length must be the same as the number of dimensions in `input`
-   * @return a new instance of Tile
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> Tile<T> create(Scope scope, Operand<T> input, Operand<U> multiples) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Tile", scope.makeOpName("Tile"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder.addInput(multiples.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new Tile<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Tile";
-  
-  private Output<T> output;
-  
-  private Tile(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Timestamp.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Timestamp.java
deleted file mode 100644
index f0830c6ffe2..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Timestamp.java
+++ /dev/null
@@ -1,75 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat64;
-
-/**
- * Provides the time since epoch in seconds.
- * <p>
- * Returns the timestamp as a `float64` for seconds since the Unix epoch.
- * <p>
- * Note: the timestamp is computed when the op is executed, not when it is added
- * to the graph.
- */
-@Operator
-public final class Timestamp extends RawOp implements Operand<TFloat64> {
-  
-  /**
-   * Factory method to create a class wrapping a new Timestamp operation.
-   * 
-   * @param scope current scope
-   * @return a new instance of Timestamp
-   */
-  @Endpoint(describeByClass = true)
-  public static Timestamp create(Scope scope) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Timestamp", scope.makeOpName("Timestamp"));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new Timestamp(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<TFloat64> ts() {
-    return ts;
-  }
-  
-  @Override
-  public Output<TFloat64> asOutput(Scope scope) {
-    return ts;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Timestamp";
-  
-  private Output<TFloat64> ts;
-  
-  private Timestamp(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    ts = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TryRpc.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TryRpc.java
deleted file mode 100644
index 72e98564759..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TryRpc.java
+++ /dev/null
@@ -1,226 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.TString;
-
-/**
- * Perform batches of RPC requests.
- * <p>
- * This op asynchronously performs either a single RPC request, or a batch
- * of requests.  RPC requests are defined by three main parameters:
- * <p>
- *   - `address` (the host+port or BNS address of the request)
- *   - `method` (the method name for the request)
- *   - `request` (the serialized proto string, or vector of strings,
- *      of the RPC request argument).
- * <p>
- * For example, if you have an RPC service running on port localhost:2345,
- * and its interface is configured with the following proto declaration:
- * <pre>{@code
- * service MyService {
- *   rpc MyMethod(MyRequestProto) returns (MyResponseProto) {
- *   }
- * };
- * }</pre>
- * then call this op with arguments:
- * <pre>{@code
- * address = "localhost:2345"
- * method = "MyService/MyMethod"
- * }</pre>
- * The `request` tensor is a string tensor representing serialized `MyRequestProto`
- * strings; and the output string tensor `response` will have the same shape
- * and contain (upon successful completion) corresponding serialized
- * `MyResponseProto` strings.
- * <p>
- * For example, to send a single, empty, `MyRequestProto`, call
- * this op with `request = ""`.  To send 5 <b>parallel</b> empty requests,
- * call this op with `request = ["", "", "", "", ""]`.
- * <p>
- * More generally, one can create a batch of `MyRequestProto` serialized protos
- * from regular batched tensors using the `encode_proto` op, and convert
- * the response `MyResponseProto` serialized protos to batched tensors
- * using the `decode_proto` op.
- * <p>
- * <b>NOTE</b> Working with serialized proto strings is faster than instantiating
- * actual proto objects in memory, so no performance degradation is expected
- * compared to writing custom kernels for this workflow.
- * <p>
- * Unlike the standard `Rpc` op, if the connection fails or the remote worker
- * returns an error status, this op does <b>not</b> reraise the exception.
- * Instead, the `status_code` and `status_message` entry for the corresponding RPC
- * call is set with the error returned from the RPC call.  The `response` tensor
- * will contain valid response values for those minibatch entries whose RPCs did
- * not fail; the rest of the entries will have empty strings.
- */
-@Operator
-public final class TryRpc extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.core.TryRpc}
-   */
-  public static class Options {
-    
-    /**
-     * @param protocol RPC protocol to use.  Empty string means use the default protocol.
-     * Options include 'grpc'.
-     */
-    public Options protocol(String protocol) {
-      this.protocol = protocol;
-      return this;
-    }
-    
-    /**
-     * @param failFast `boolean`. If `true` (default), then failures to connect
-     * (i.e., the server does not immediately respond) cause an RPC failure.
-     */
-    public Options failFast(Boolean failFast) {
-      this.failFast = failFast;
-      return this;
-    }
-    
-    /**
-     * @param timeoutInMs `int`. If `0` (default), then the kernel will run the RPC
-     * request and only time out if the RPC deadline passes or the session times out.
-     * If this value is greater than `0`, then the op will raise an exception if
-     * the RPC takes longer than `timeout_in_ms`.
-     */
-    public Options timeoutInMs(Long timeoutInMs) {
-      this.timeoutInMs = timeoutInMs;
-      return this;
-    }
-    
-    private String protocol;
-    private Boolean failFast;
-    private Long timeoutInMs;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new TryRpc operation.
-   * 
-   * @param scope current scope
-   * @param address `0-D` or `1-D`.  The address (i.e. host_name:port) of the RPC server.
-   * If this tensor has more than 1 element, then multiple parallel rpc requests
-   * are sent.  This argument broadcasts with `method` and `request`.
-   * @param method `0-D` or `1-D`.  The method address on the RPC server.
-   * If this tensor has more than 1 element, then multiple parallel rpc requests
-   * are sent.  This argument broadcasts with `address` and `request`.
-   * @param request `0-D` or `1-D`.  Serialized proto strings: the rpc request argument.
-   * If this tensor has more than 1 element, then multiple parallel rpc requests
-   * are sent.  This argument broadcasts with `address` and `method`.
-   * @param options carries optional attributes values
-   * @return a new instance of TryRpc
-   */
-  @Endpoint(describeByClass = true)
-  public static TryRpc create(Scope scope, Operand<TString> address, Operand<TString> method, Operand<TString> request, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("TryRpc", scope.makeOpName("TryRpc"));
-    opBuilder.addInput(address.asOutput(scope));
-    opBuilder.addInput(method.asOutput(scope));
-    opBuilder.addInput(request.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.protocol != null) {
-          opBuilder.setAttr("protocol", opts.protocol);
-        }
-        if (opts.failFast != null) {
-          opBuilder.setAttr("fail_fast", opts.failFast);
-        }
-        if (opts.timeoutInMs != null) {
-          opBuilder.setAttr("timeout_in_ms", opts.timeoutInMs);
-        }
-      }
-    }
-    return new TryRpc(opBuilder.build());
-  }
-  
-  /**
-   * @param protocol RPC protocol to use.  Empty string means use the default protocol.
-   * Options include 'grpc'.
-   */
-  public static Options protocol(String protocol) {
-    return new Options().protocol(protocol);
-  }
-  
-  /**
-   * @param failFast `boolean`. If `true` (default), then failures to connect
-   * (i.e., the server does not immediately respond) cause an RPC failure.
-   */
-  public static Options failFast(Boolean failFast) {
-    return new Options().failFast(failFast);
-  }
-  
-  /**
-   * @param timeoutInMs `int`. If `0` (default), then the kernel will run the RPC
-   * request and only time out if the RPC deadline passes or the session times out.
-   * If this value is greater than `0`, then the op will raise an exception if
-   * the RPC takes longer than `timeout_in_ms`.
-   */
-  public static Options timeoutInMs(Long timeoutInMs) {
-    return new Options().timeoutInMs(timeoutInMs);
-  }
-  
-  /**
-   * Same shape as `request`. Serialized proto strings: the rpc responses.
-   */
-  public Output<TString> response() {
-    return response;
-  }
-  
-  /**
-   * Same shape as `request`.  Values correspond to tensorflow Status enum codes.
-   */
-  public Output<TInt32> statusCode() {
-    return statusCode;
-  }
-  
-  /**
-   * Same shape as `request`.  Values correspond to Status messages
-   * returned from the RPC calls.
-   */
-  public Output<TString> statusMessage() {
-    return statusMessage;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "TryRpc";
-  
-  private Output<TString> response;
-  private Output<TInt32> statusCode;
-  private Output<TString> statusMessage;
-  
-  private TryRpc(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    response = operation.output(outputIdx++);
-    statusCode = operation.output(outputIdx++);
-    statusMessage = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Unbatch.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Unbatch.java
deleted file mode 100644
index 266df8c4af5..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Unbatch.java
+++ /dev/null
@@ -1,153 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
-
-/**
- * Reverses the operation of Batch for a single output Tensor.
- * <p>
- * An instance of Unbatch either receives an empty batched_tensor, in which case it
- * asynchronously waits until the values become available from a concurrently
- * running instance of Unbatch with the same container and shared_name, or receives
- * a non-empty batched_tensor in which case it finalizes all other concurrently
- * running instances and outputs its own element from the batch.
- * <p>
- * batched_tensor: The possibly transformed output of Batch. The size of the first
- *  dimension should remain unchanged by the transformations for the operation to
- *  work.
- * batch_index: The matching batch_index obtained from Batch.
- * id: The id scalar emitted by Batch.
- * unbatched_tensor: The Tensor corresponding to this execution.
- * timeout_micros: Maximum amount of time (in microseconds) to wait to receive the
- *  batched input tensor associated with a given invocation of the op.
- * container: Container to control resource sharing.
- * shared_name: Instances of Unbatch with the same container and shared_name are
- *  assumed to possibly belong to the same batch. If left empty, the op name will
- *  be used as the shared name.
- * 
- * @param <T> data type for {@code unbatchedTensor()} output
- */
-@Operator
-public final class Unbatch<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.core.Unbatch}
-   */
-  public static class Options {
-    
-    /**
-     * @param container 
-     */
-    public Options container(String container) {
-      this.container = container;
-      return this;
-    }
-    
-    /**
-     * @param sharedName 
-     */
-    public Options sharedName(String sharedName) {
-      this.sharedName = sharedName;
-      return this;
-    }
-    
-    private String container;
-    private String sharedName;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new Unbatch operation.
-   * 
-   * @param scope current scope
-   * @param batchedTensor 
-   * @param batchIndex 
-   * @param id 
-   * @param timeoutMicros 
-   * @param options carries optional attributes values
-   * @return a new instance of Unbatch
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> Unbatch<T> create(Scope scope, Operand<T> batchedTensor, Operand<TInt64> batchIndex, Operand<TInt64> id, Long timeoutMicros, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Unbatch", scope.makeOpName("Unbatch"));
-    opBuilder.addInput(batchedTensor.asOutput(scope));
-    opBuilder.addInput(batchIndex.asOutput(scope));
-    opBuilder.addInput(id.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("timeout_micros", timeoutMicros);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.container != null) {
-          opBuilder.setAttr("container", opts.container);
-        }
-        if (opts.sharedName != null) {
-          opBuilder.setAttr("shared_name", opts.sharedName);
-        }
-      }
-    }
-    return new Unbatch<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param container 
-   */
-  public static Options container(String container) {
-    return new Options().container(container);
-  }
-  
-  /**
-   * @param sharedName 
-   */
-  public static Options sharedName(String sharedName) {
-    return new Options().sharedName(sharedName);
-  }
-  
-  /**
-   */
-  public Output<T> unbatchedTensor() {
-    return unbatchedTensor;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return unbatchedTensor;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Unbatch";
-  
-  private Output<T> unbatchedTensor;
-  
-  private Unbatch(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    unbatchedTensor = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/UnbatchGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/UnbatchGrad.java
deleted file mode 100644
index b821aaac605..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/UnbatchGrad.java
+++ /dev/null
@@ -1,149 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
-
-/**
- * Gradient of Unbatch.
- * <p>
- * Acts like Batch but using the given batch_index index of batching things as they
- * become available. This ensures that the gradients are propagated back in the
- * same session which did the forward pass.
- * <p>
- * original_input: The input to the Unbatch operation this is the gradient of.
- * batch_index: The batch_index given to the Unbatch operation this is the gradient
- * of.
- * grad: The downstream gradient.
- * id: The id scalar emitted by Batch.
- * batched_grad: The return value, either an empty tensor or the batched gradient.
- * container: Container to control resource sharing.
- * shared_name: Instances of UnbatchGrad with the same container and shared_name
- *  are assumed to possibly belong to the same batch. If left empty, the op name
- *  will be used as the shared name.
- * 
- * @param <T> data type for {@code batchedGrad()} output
- */
-@Operator
-public final class UnbatchGrad<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.core.UnbatchGrad}
-   */
-  public static class Options {
-    
-    /**
-     * @param container 
-     */
-    public Options container(String container) {
-      this.container = container;
-      return this;
-    }
-    
-    /**
-     * @param sharedName 
-     */
-    public Options sharedName(String sharedName) {
-      this.sharedName = sharedName;
-      return this;
-    }
-    
-    private String container;
-    private String sharedName;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new UnbatchGrad operation.
-   * 
-   * @param scope current scope
-   * @param originalInput 
-   * @param batchIndex 
-   * @param grad 
-   * @param id 
-   * @param options carries optional attributes values
-   * @return a new instance of UnbatchGrad
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> UnbatchGrad<T> create(Scope scope, Operand<T> originalInput, Operand<TInt64> batchIndex, Operand<T> grad, Operand<TInt64> id, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("UnbatchGrad", scope.makeOpName("UnbatchGrad"));
-    opBuilder.addInput(originalInput.asOutput(scope));
-    opBuilder.addInput(batchIndex.asOutput(scope));
-    opBuilder.addInput(grad.asOutput(scope));
-    opBuilder.addInput(id.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.container != null) {
-          opBuilder.setAttr("container", opts.container);
-        }
-        if (opts.sharedName != null) {
-          opBuilder.setAttr("shared_name", opts.sharedName);
-        }
-      }
-    }
-    return new UnbatchGrad<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param container 
-   */
-  public static Options container(String container) {
-    return new Options().container(container);
-  }
-  
-  /**
-   * @param sharedName 
-   */
-  public static Options sharedName(String sharedName) {
-    return new Options().sharedName(sharedName);
-  }
-  
-  /**
-   */
-  public Output<T> batchedGrad() {
-    return batchedGrad;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return batchedGrad;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "UnbatchGrad";
-  
-  private Output<T> batchedGrad;
-  
-  private UnbatchGrad(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    batchedGrad = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Unique.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Unique.java
deleted file mode 100644
index c4d56b4e8b6..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Unique.java
+++ /dev/null
@@ -1,142 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * Finds unique elements along an axis of a tensor.
- * <p>
- * This operation either returns a tensor `y` containing unique elements
- * along the `axis` of a tensor. The returned unique elements is sorted
- * in the same order as they occur along `axis` in `x`.
- * This operation also returns a tensor `idx` that is the same size as
- * the number of the elements in `x` along the `axis` dimension. It
- * contains the index in the unique output `y`.
- * In other words, for an `1-D` tensor `x` with `axis = None:
- * <p>
- * `y[idx[i]] = x[i] for i in [0, 1,...,rank(x) - 1]`
- * <p>
- * For example:
- * <pre>{@code
- * # tensor 'x' is [1, 1, 2, 4, 4, 4, 7, 8, 8]
- * y, idx = unique(x)
- * y ==> [1, 2, 4, 7, 8]
- * idx ==> [0, 0, 1, 2, 2, 2, 3, 4, 4]
- * }</pre>
- * For an `2-D` tensor `x` with `axis = 0`:
- * <pre>{@code
- * # tensor 'x' is [[1, 0, 0],
- * #                [1, 0, 0],
- * #                [2, 0, 0]]
- * y, idx = unique(x, axis=0)
- * y ==> [[1, 0, 0],
- *        [2, 0, 0]]
- * idx ==> [0, 0, 1]
- * }</pre>
- * For an `2-D` tensor `x` with `axis = 1`:
- * <pre>{@code
- * # tensor 'x' is [[1, 0, 0],
- * #                [1, 0, 0],
- * #                [2, 0, 0]]
- * y, idx = unique(x, axis=1)
- * y ==> [[1, 0],
- *        [1, 0],
- *        [2, 0]]
- * idx ==> [0, 1, 1]
- * }</pre>
- * 
- * 
- * @param <T> data type for {@code y()} output
- * @param <V> data type for {@code idx()} output
- */
-@Operator
-public final class Unique<T extends TType, V extends TNumber> extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new Unique operation.
-   * 
-   * @param scope current scope
-   * @param x A `Tensor`.
-   * @param axis A `Tensor` of type `int32` (default: None). The axis of the Tensor to
-   * find the unique elements.
-   * @param outIdx 
-   * @return a new instance of Unique
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType, V extends TNumber, U extends TNumber> Unique<T, V> create(Scope scope, Operand<T> x, Operand<U> axis, Class<V> outIdx) {
-    OperationBuilder opBuilder = scope.env().opBuilder("UniqueV2", scope.makeOpName("Unique"));
-    opBuilder.addInput(x.asOutput(scope));
-    opBuilder.addInput(axis.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("out_idx", outIdx);
-    return new Unique<T, V>(opBuilder.build());
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new Unique operation using default output types.
-   * 
-   * @param scope current scope
-   * @param x A `Tensor`.
-   * @param axis A `Tensor` of type `int32` (default: None). The axis of the Tensor to
-   * find the unique elements.
-   * @return a new instance of Unique
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> Unique<T, TInt32> create(Scope scope, Operand<T> x, Operand<U> axis) {
-    return create(scope, x, axis, TInt32.class);
-  }
-  
-  /**
-   * A `Tensor`. Unique elements along the `axis` of `Tensor` x.
-   */
-  public Output<T> y() {
-    return y;
-  }
-  
-  /**
-   * A 1-D Tensor. Has the same type as x that contains the index of each
-   * value of x in the output y.
-   */
-  public Output<V> idx() {
-    return idx;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "UniqueV2";
-  
-  private Output<T> y;
-  private Output<V> idx;
-  
-  private Unique(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    y = operation.output(outputIdx++);
-    idx = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/UniqueWithCounts.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/UniqueWithCounts.java
deleted file mode 100644
index 74046926294..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/UniqueWithCounts.java
+++ /dev/null
@@ -1,155 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * Finds unique elements along an axis of a tensor.
- * <p>
- * This operation either returns a tensor `y` containing unique elements
- * along the `axis` of a tensor. The returned unique elements is sorted
- * in the same order as they occur along `axis` in `x`.
- * This operation also returns a tensor `idx` and a tensor `count`
- * that are the same size as the number of the elements in `x` along the
- * `axis` dimension. The `idx` contains the index in the unique output `y`
- * and the `count` contains the count in the unique output `y`.
- * In other words, for an `1-D` tensor `x` with `axis = None:
- * <p>
- * `y[idx[i]] = x[i] for i in [0, 1,...,rank(x) - 1]`
- * <p>
- * For example:
- * <pre>{@code
- * # tensor 'x' is [1, 1, 2, 4, 4, 4, 7, 8, 8]
- * y, idx, count = unique_with_counts(x)
- * y ==> [1, 2, 4, 7, 8]
- * idx ==> [0, 0, 1, 2, 2, 2, 3, 4, 4]
- * count ==> [2, 1, 3, 1, 2]
- * }</pre>
- * For an `2-D` tensor `x` with `axis = 0`:
- * <pre>{@code
- * # tensor 'x' is [[1, 0, 0],
- * #                [1, 0, 0],
- * #                [2, 0, 0]]
- * y, idx, count = unique_with_counts(x, axis=0)
- * y ==> [[1, 0, 0],
- *        [2, 0, 0]]
- * idx ==> [0, 0, 1]
- * count ==> [2, 1]
- * }</pre>
- * For an `2-D` tensor `x` with `axis = 1`:
- * <pre>{@code
- * # tensor 'x' is [[1, 0, 0],
- * #                [1, 0, 0],
- * #                [2, 0, 0]]
- * y, idx, count = unique_with_counts(x, axis=1)
- * y ==> [[1, 0],
- *        [1, 0],
- *        [2, 0]]
- * idx ==> [0, 1, 1]
- * count ==> [1, 2]
- * }</pre>
- * 
- * 
- * @param <T> data type for {@code y()} output
- * @param <V> data type for {@code idx()} output
- */
-@Operator
-public final class UniqueWithCounts<T extends TType, V extends TNumber> extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new UniqueWithCounts operation.
-   * 
-   * @param scope current scope
-   * @param x A `Tensor`.
-   * @param axis A `Tensor` of type `int32` (default: None). The axis of the Tensor to
-   * find the unique elements.
-   * @param outIdx 
-   * @return a new instance of UniqueWithCounts
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType, V extends TNumber, U extends TNumber> UniqueWithCounts<T, V> create(Scope scope, Operand<T> x, Operand<U> axis, Class<V> outIdx) {
-    OperationBuilder opBuilder = scope.env().opBuilder("UniqueWithCountsV2", scope.makeOpName("UniqueWithCounts"));
-    opBuilder.addInput(x.asOutput(scope));
-    opBuilder.addInput(axis.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("out_idx", outIdx);
-    return new UniqueWithCounts<T, V>(opBuilder.build());
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new UniqueWithCounts operation using default output types.
-   * 
-   * @param scope current scope
-   * @param x A `Tensor`.
-   * @param axis A `Tensor` of type `int32` (default: None). The axis of the Tensor to
-   * find the unique elements.
-   * @return a new instance of UniqueWithCounts
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> UniqueWithCounts<T, TInt32> create(Scope scope, Operand<T> x, Operand<U> axis) {
-    return create(scope, x, axis, TInt32.class);
-  }
-  
-  /**
-   * A `Tensor`. Unique elements along the `axis` of `Tensor` x.
-   */
-  public Output<T> y() {
-    return y;
-  }
-  
-  /**
-   * A 1-D Tensor. Has the same type as x that contains the index of each
-   * value of x in the output y.
-   */
-  public Output<V> idx() {
-    return idx;
-  }
-  
-  /**
-   * A 1-D Tensor. The count of each value of x in the output y.
-   */
-  public Output<V> count() {
-    return count;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "UniqueWithCountsV2";
-  
-  private Output<T> y;
-  private Output<V> idx;
-  private Output<V> count;
-  
-  private UniqueWithCounts(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    y = operation.output(outputIdx++);
-    idx = operation.output(outputIdx++);
-    count = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/UnravelIndex.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/UnravelIndex.java
deleted file mode 100644
index 7f7c8b50bf8..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/UnravelIndex.java
+++ /dev/null
@@ -1,99 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Converts an array of flat indices into a tuple of coordinate arrays.
- * <p>
- * 
- * Example:
- * <pre>{@code
- * y = tf.unravel_index(indices=[2, 5, 7], dims=[3, 3])
- * # 'dims' represent a hypothetical (3, 3) tensor of indices:
- * # [[0, 1, *2*],
- * #  [3, 4, *5*],
- * #  [6, *7*, 8]]
- * # For each entry from 'indices', this operation returns
- * # its coordinates (marked with '*'), such as
- * # 2 ==> (0, 2)
- * # 5 ==> (1, 2)
- * # 7 ==> (2, 1)
- * y ==> [[0, 1, 2], [2, 2, 1]]
- * }</pre>
- * @compatibility(numpy)
- * Equivalent to np.unravel_index
- * @end_compatibility
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator
-public final class UnravelIndex<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new UnravelIndex operation.
-   * 
-   * @param scope current scope
-   * @param indices An 0-D or 1-D `int` Tensor whose elements are indices into the
-   * flattened version of an array of dimensions dims.
-   * @param dims An 1-D `int` Tensor. The shape of the array to use for unraveling
-   * indices.
-   * @return a new instance of UnravelIndex
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> UnravelIndex<T> create(Scope scope, Operand<T> indices, Operand<T> dims) {
-    OperationBuilder opBuilder = scope.env().opBuilder("UnravelIndex", scope.makeOpName("UnravelIndex"));
-    opBuilder.addInput(indices.asOutput(scope));
-    opBuilder.addInput(dims.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new UnravelIndex<T>(opBuilder.build());
-  }
-  
-  /**
-   * An 2-D (or 1-D if indices is 0-D) tensor where each row has the
-   * same shape as the indices array.
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "UnravelIndex";
-  
-  private Output<T> output;
-  
-  private UnravelIndex(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Unstack.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Unstack.java
deleted file mode 100644
index 4902acb7ae3..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Unstack.java
+++ /dev/null
@@ -1,133 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import java.util.Arrays;
-import java.util.Iterator;
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Unpacks a given dimension of a rank-`R` tensor into `num` rank-`(R-1)` tensors.
- * <p>
- * Unpacks `num` tensors from `value` by chipping it along the `axis` dimension.
- * For example, given a tensor of shape `(A, B, C, D)`;
- * <p>
- * If `axis == 0` then the i'th tensor in `output` is the slice `value[i, :, :, :]`
- *   and each tensor in `output` will have shape `(B, C, D)`. (Note that the
- *   dimension unpacked along is gone, unlike `split`).
- * <p>
- * If `axis == 1` then the i'th tensor in `output` is the slice `value[:, i, :, :]`
- *   and each tensor in `output` will have shape `(A, C, D)`.
- * Etc.
- * <p>
- * This is the opposite of `pack`.
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator
-public final class Unstack<T extends TType> extends RawOp implements Iterable<Operand<T>> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.core.Unstack}
-   */
-  public static class Options {
-    
-    /**
-     * @param axis Dimension along which to unpack.  Negative values wrap around, so the
-     * valid range is `[-R, R)`.
-     */
-    public Options axis(Long axis) {
-      this.axis = axis;
-      return this;
-    }
-    
-    private Long axis;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new Unstack operation.
-   * 
-   * @param scope current scope
-   * @param value 1-D or higher, with `axis` dimension size equal to `num`.
-   * @param num 
-   * @param options carries optional attributes values
-   * @return a new instance of Unstack
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> Unstack<T> create(Scope scope, Operand<T> value, Long num, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Unpack", scope.makeOpName("Unstack"));
-    opBuilder.addInput(value.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("num", num);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.axis != null) {
-          opBuilder.setAttr("axis", opts.axis);
-        }
-      }
-    }
-    return new Unstack<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param axis Dimension along which to unpack.  Negative values wrap around, so the
-   * valid range is `[-R, R)`.
-   */
-  public static Options axis(Long axis) {
-    return new Options().axis(axis);
-  }
-  
-  /**
-   * The list of tensors unpacked from `value`.
-   */
-  public List<Output<T>> output() {
-    return output;
-  }
-  
-  @Override
-  @SuppressWarnings({"rawtypes", "unchecked"})
-  public Iterator<Operand<T>> iterator() {
-    return (Iterator) output.iterator();
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Unpack";
-  
-  private List<Output<T>> output;
-  
-  @SuppressWarnings("unchecked")
-  private Unstack(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    int outputLength = operation.outputListLength("output");
-    output = Arrays.asList((Output<T>[])operation.outputList(outputIdx, outputLength));
-    outputIdx += outputLength;
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Unstage.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Unstage.java
deleted file mode 100644
index d92ec49bc2a..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Unstage.java
+++ /dev/null
@@ -1,176 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import java.util.Arrays;
-import java.util.Iterator;
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Op is similar to a lightweight Dequeue.
- * <p>
- * The basic functionality is similar to dequeue with many fewer
- * capabilities and options.  This Op is optimized for performance.
- */
-@Operator
-public final class Unstage extends RawOp implements Iterable<Operand<TType>> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.core.Unstage}
-   */
-  public static class Options {
-    
-    /**
-     * @param capacity 
-     */
-    public Options capacity(Long capacity) {
-      this.capacity = capacity;
-      return this;
-    }
-    
-    /**
-     * @param memoryLimit 
-     */
-    public Options memoryLimit(Long memoryLimit) {
-      this.memoryLimit = memoryLimit;
-      return this;
-    }
-    
-    /**
-     * @param container 
-     */
-    public Options container(String container) {
-      this.container = container;
-      return this;
-    }
-    
-    /**
-     * @param sharedName 
-     */
-    public Options sharedName(String sharedName) {
-      this.sharedName = sharedName;
-      return this;
-    }
-    
-    private Long capacity;
-    private Long memoryLimit;
-    private String container;
-    private String sharedName;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new Unstage operation.
-   * 
-   * @param scope current scope
-   * @param dtypes 
-   * @param options carries optional attributes values
-   * @return a new instance of Unstage
-   */
-  @Endpoint(describeByClass = true)
-  public static Unstage create(Scope scope, List<Class<?>> dtypes, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Unstage", scope.makeOpName("Unstage"));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    Class[] dtypesArray = new Class[dtypes.size()];
-    for (int i = 0; i < dtypesArray.length; ++i) {
-      dtypesArray[i] = dtypes.get(i);
-    }
-    opBuilder.setAttr("dtypes", dtypesArray);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.capacity != null) {
-          opBuilder.setAttr("capacity", opts.capacity);
-        }
-        if (opts.memoryLimit != null) {
-          opBuilder.setAttr("memory_limit", opts.memoryLimit);
-        }
-        if (opts.container != null) {
-          opBuilder.setAttr("container", opts.container);
-        }
-        if (opts.sharedName != null) {
-          opBuilder.setAttr("shared_name", opts.sharedName);
-        }
-      }
-    }
-    return new Unstage(opBuilder.build());
-  }
-  
-  /**
-   * @param capacity 
-   */
-  public static Options capacity(Long capacity) {
-    return new Options().capacity(capacity);
-  }
-  
-  /**
-   * @param memoryLimit 
-   */
-  public static Options memoryLimit(Long memoryLimit) {
-    return new Options().memoryLimit(memoryLimit);
-  }
-  
-  /**
-   * @param container 
-   */
-  public static Options container(String container) {
-    return new Options().container(container);
-  }
-  
-  /**
-   * @param sharedName 
-   */
-  public static Options sharedName(String sharedName) {
-    return new Options().sharedName(sharedName);
-  }
-  
-  /**
-   */
-  public List<Output<?>> values() {
-    return values;
-  }
-  
-  @Override
-  @SuppressWarnings({"rawtypes", "unchecked"})
-  public Iterator<Operand<TType>> iterator() {
-    return (Iterator) values.iterator();
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Unstage";
-  
-  private List<Output<?>> values;
-  
-  private Unstage(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    int valuesLength = operation.outputListLength("values");
-    values = Arrays.asList(operation.outputList(outputIdx, valuesLength));
-    outputIdx += valuesLength;
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/UpperBound.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/UpperBound.java
deleted file mode 100644
index 8e5b59b0fff..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/UpperBound.java
+++ /dev/null
@@ -1,115 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * Applies upper_bound(sorted_search_values, values) along each row.
- * <p>
- * Each set of rows with the same index in (sorted_inputs, values) is treated
- * independently.  The resulting row is the equivalent of calling
- * `np.searchsorted(sorted_inputs, values, side='right')`.
- * <p>
- * The result is not a global index to the entire
- * `Tensor`, but rather just the index in the last dimension.
- * <p>
- * A 2-D example:
- *   sorted_sequence = [[0, 3, 9, 9, 10],
- *                      [1, 2, 3, 4, 5]]
- *   values = [[2, 4, 9],
- *             [0, 2, 6]]
- * <p>
- *   result = UpperBound(sorted_sequence, values)
- * <p>
- *   result == [[1, 2, 4],
- *              [0, 2, 5]]
- * 
- * @param <U> data type for {@code output()} output
- */
-public final class UpperBound<U extends TNumber> extends RawOp implements Operand<U> {
-  
-  /**
-   * Factory method to create a class wrapping a new UpperBound operation.
-   * 
-   * @param scope current scope
-   * @param sortedInputs 2-D Tensor where each row is ordered.
-   * @param values 2-D Tensor with the same numbers of rows as `sorted_search_values`. Contains
-   * the values that will be searched for in `sorted_search_values`.
-   * @param outType 
-   * @return a new instance of UpperBound
-   */
-  @Endpoint(describeByClass = true)
-  public static <U extends TNumber, T extends TType> UpperBound<U> create(Scope scope, Operand<T> sortedInputs, Operand<T> values, Class<U> outType) {
-    OperationBuilder opBuilder = scope.env().opBuilder("UpperBound", scope.makeOpName("UpperBound"));
-    opBuilder.addInput(sortedInputs.asOutput(scope));
-    opBuilder.addInput(values.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("out_type", outType);
-    return new UpperBound<U>(opBuilder.build());
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new UpperBound operation using default output types.
-   * 
-   * @param scope current scope
-   * @param sortedInputs 2-D Tensor where each row is ordered.
-   * @param values 2-D Tensor with the same numbers of rows as `sorted_search_values`. Contains
-   * the values that will be searched for in `sorted_search_values`.
-   * @return a new instance of UpperBound
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> UpperBound<TInt32> create(Scope scope, Operand<T> sortedInputs, Operand<T> values) {
-    return create(scope, sortedInputs, values, TInt32.class);
-  }
-  
-  /**
-   * A `Tensor` with the same shape as `values`.  It contains the last scalar index
-   * into the last dimension where values can be inserted without changing the
-   * ordered property.
-   */
-  public Output<U> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<U> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "UpperBound";
-  
-  private Output<U> output;
-  
-  private UpperBound(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/VarHandleOp.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/VarHandleOp.java
deleted file mode 100644
index 65f8784c3ae..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/VarHandleOp.java
+++ /dev/null
@@ -1,156 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Creates a handle to a Variable resource.
- */
-@Operator
-public final class VarHandleOp extends RawOp implements Operand<TType> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.core.VarHandleOp}
-   */
-  public static class Options {
-    
-    /**
-     * @param container the container this variable is placed in.
-     */
-    public Options container(String container) {
-      this.container = container;
-      return this;
-    }
-    
-    /**
-     * @param sharedName the name by which this variable is referred to.
-     */
-    public Options sharedName(String sharedName) {
-      this.sharedName = sharedName;
-      return this;
-    }
-    
-    /**
-     * @param allowedDevices DEPRECATED. The allowed devices containing the resource variable. Set when the
-     * output ResourceHandle represents a per-replica/partitioned resource variable.
-     */
-    public Options allowedDevices(List<String> allowedDevices) {
-      this.allowedDevices = allowedDevices;
-      return this;
-    }
-    
-    private String container;
-    private String sharedName;
-    private List<String> allowedDevices;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new VarHandleOp operation.
-   * 
-   * @param scope current scope
-   * @param dtype the type of this variable. Must agree with the dtypes
-   * of all ops using this variable.
-   * @param shape The (possibly partially specified) shape of this variable.
-   * @param options carries optional attributes values
-   * @return a new instance of VarHandleOp
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> VarHandleOp create(Scope scope, Class<T> dtype, Shape shape, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("VarHandleOp", scope.makeOpName("VarHandleOp"));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("dtype", dtype);
-    opBuilder.setAttr("shape", shape);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.container != null) {
-          opBuilder.setAttr("container", opts.container);
-        }
-        if (opts.sharedName != null) {
-          opBuilder.setAttr("shared_name", opts.sharedName);
-        }
-        if (opts.allowedDevices != null) {
-          String[] allowedDevicesArray = new String[opts.allowedDevices.size()];
-          for (int i = 0; i < allowedDevicesArray.length; ++i) {
-            allowedDevicesArray[i] = opts.allowedDevices.get(i);
-          }
-          opBuilder.setAttr("allowed_devices", allowedDevicesArray);
-        }
-      }
-    }
-    return new VarHandleOp(opBuilder.build());
-  }
-  
-  /**
-   * @param container the container this variable is placed in.
-   */
-  public static Options container(String container) {
-    return new Options().container(container);
-  }
-  
-  /**
-   * @param sharedName the name by which this variable is referred to.
-   */
-  public static Options sharedName(String sharedName) {
-    return new Options().sharedName(sharedName);
-  }
-  
-  /**
-   * @param allowedDevices DEPRECATED. The allowed devices containing the resource variable. Set when the
-   * output ResourceHandle represents a per-replica/partitioned resource variable.
-   */
-  public static Options allowedDevices(List<String> allowedDevices) {
-    return new Options().allowedDevices(allowedDevices);
-  }
-  
-  /**
-   */
-  public Output<?> resource() {
-    return resource;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) resource;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "VarHandleOp";
-  
-  private Output<?> resource;
-  
-  private VarHandleOp(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    resource = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/VarIsInitializedOp.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/VarIsInitializedOp.java
deleted file mode 100644
index 6fa5ffc7162..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/VarIsInitializedOp.java
+++ /dev/null
@@ -1,74 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TBool;
-
-/**
- * Checks whether a resource handle-based variable has been initialized.
- */
-@Operator
-public final class VarIsInitializedOp extends RawOp implements Operand<TBool> {
-  
-  /**
-   * Factory method to create a class wrapping a new VarIsInitializedOp operation.
-   * 
-   * @param scope current scope
-   * @param resource the input resource handle.
-   * @return a new instance of VarIsInitializedOp
-   */
-  @Endpoint(describeByClass = true)
-  public static VarIsInitializedOp create(Scope scope, Operand<?> resource) {
-    OperationBuilder opBuilder = scope.env().opBuilder("VarIsInitializedOp", scope.makeOpName("VarIsInitializedOp"));
-    opBuilder.addInput(resource.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new VarIsInitializedOp(opBuilder.build());
-  }
-  
-  /**
-   * a scalar boolean which is true if the variable has been
-   * initialized.
-   */
-  public Output<TBool> isInitialized() {
-    return isInitialized;
-  }
-  
-  @Override
-  public Output<TBool> asOutput(Scope scope) {
-    return isInitialized;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "VarIsInitializedOp";
-  
-  private Output<TBool> isInitialized;
-  
-  private VarIsInitializedOp(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    isInitialized = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Variable.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Variable.java
deleted file mode 100644
index 7c89a2518e2..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Variable.java
+++ /dev/null
@@ -1,139 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Holds state in the form of a tensor that persists across steps.
- * <p>
- * Outputs a ref to the tensor state so it may be read or modified.
- * TODO(zhifengc/mrry): Adds a pointer to a more detail document
- * about sharing states in tensorflow.
- * 
- * @param <T> data type for {@code ref()} output
- */
-@Operator
-public final class Variable<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.core.Variable}
-   */
-  public static class Options {
-    
-    /**
-     * @param container If non-empty, this variable is placed in the given container.
-     * Otherwise, a default container is used.
-     */
-    public Options container(String container) {
-      this.container = container;
-      return this;
-    }
-    
-    /**
-     * @param sharedName If non-empty, this variable is named in the given bucket
-     * with this shared_name. Otherwise, the node name is used instead.
-     */
-    public Options sharedName(String sharedName) {
-      this.sharedName = sharedName;
-      return this;
-    }
-    
-    private String container;
-    private String sharedName;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new Variable operation.
-   * 
-   * @param scope current scope
-   * @param shape The shape of the variable tensor.
-   * @param dtype The type of elements in the variable tensor.
-   * @param options carries optional attributes values
-   * @return a new instance of Variable
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> Variable<T> create(Scope scope, Shape shape, Class<T> dtype, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("VariableV2", scope.makeOpName("Variable"));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("shape", shape);
-    opBuilder.setAttr("dtype", dtype);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.container != null) {
-          opBuilder.setAttr("container", opts.container);
-        }
-        if (opts.sharedName != null) {
-          opBuilder.setAttr("shared_name", opts.sharedName);
-        }
-      }
-    }
-    return new Variable<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param container If non-empty, this variable is placed in the given container.
-   * Otherwise, a default container is used.
-   */
-  public static Options container(String container) {
-    return new Options().container(container);
-  }
-  
-  /**
-   * @param sharedName If non-empty, this variable is named in the given bucket
-   * with this shared_name. Otherwise, the node name is used instead.
-   */
-  public static Options sharedName(String sharedName) {
-    return new Options().sharedName(sharedName);
-  }
-  
-  /**
-   * A reference to the variable tensor.
-   */
-  public Output<T> ref() {
-    return ref;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return ref;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "VariableV2";
-  
-  private Output<T> ref;
-  
-  private Variable(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    ref = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/VariableShape.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/VariableShape.java
deleted file mode 100644
index fa10b66ef00..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/VariableShape.java
+++ /dev/null
@@ -1,98 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Returns the shape of the variable pointed to by `resource`.
- * <p>
- * This operation returns a 1-D integer tensor representing the shape of `input`.
- * <p>
- * For example:
- * <pre>{@code
- * # 't' is [[[1, 1, 1], [2, 2, 2]], [[3, 3, 3], [4, 4, 4]]]
- * shape(t) ==> [2, 2, 3]
- * }</pre>
- * 
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator
-public final class VariableShape<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new VariableShape operation.
-   * 
-   * @param scope current scope
-   * @param input 
-   * @param outType 
-   * @return a new instance of VariableShape
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> VariableShape<T> create(Scope scope, Operand<?> input, Class<T> outType) {
-    OperationBuilder opBuilder = scope.env().opBuilder("VariableShape", scope.makeOpName("VariableShape"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("out_type", outType);
-    return new VariableShape<T>(opBuilder.build());
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new VariableShape operation using default output types.
-   * 
-   * @param scope current scope
-   * @param input 
-   * @return a new instance of VariableShape
-   */
-  @Endpoint(describeByClass = true)
-  public static VariableShape<TInt32> create(Scope scope, Operand<?> input) {
-    return create(scope, input, TInt32.class);
-  }
-  
-  /**
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "VariableShape";
-  
-  private Output<T> output;
-  
-  private VariableShape(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Where.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Where.java
deleted file mode 100644
index 791ba807489..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Where.java
+++ /dev/null
@@ -1,133 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
-
-/**
- * Returns locations of nonzero / true values in a tensor.
- * <p>
- * This operation returns the coordinates of true elements in `condition`. The
- * coordinates are returned in a 2-D tensor where the first dimension (rows)
- * represents the number of true elements, and the second dimension (columns)
- * represents the coordinates of the true elements. Keep in mind, the shape of
- * the output tensor can vary depending on how many true values there are in
- * `condition`. Indices are output in row-major order.
- * <p>
- * For example:
- * <pre>{@code
- * # 'input' tensor is [[True, False]
- * #                    [True, False]]
- * # 'input' has two true values, so output has two coordinates.
- * # 'input' has rank of 2, so coordinates have two indices.
- * where(input) ==> [[0, 0],
- *                   [1, 0]]
- * 
- * # `condition` tensor is [[[True, False]
- * #                     [True, False]]
- * #                    [[False, True]
- * #                     [False, True]]
- * #                    [[False, False]
- * #                     [False, True]]]
- * # 'input' has 5 true values, so output has 5 coordinates.
- * # 'input' has rank of 3, so coordinates have three indices.
- * where(input) ==> [[0, 0, 0],
- *                   [0, 1, 0],
- *                   [1, 0, 1],
- *                   [1, 1, 1],
- *                   [2, 1, 1]]
- * 
- * # `condition` tensor is [[[1.5,  0.0]
- * #                     [-0.5, 0.0]]
- * #                    [[0.0,  0.25]
- * #                     [0.0,  0.75]]
- * #                    [[0.0,  0.0]
- * #                     [0.0,  0.01]]]
- * # 'input' has 5 nonzero values, so output has 5 coordinates.
- * # 'input' has rank of 3, so coordinates have three indices.
- * where(input) ==> [[0, 0, 0],
- *                   [0, 1, 0],
- *                   [1, 0, 1],
- *                   [1, 1, 1],
- *                   [2, 1, 1]]
- * 
- * # `condition` tensor is [[[1.5 + 0.0j, 0.0  + 0.0j]
- * #                     [0.0 + 0.5j, 0.0  + 0.0j]]
- * #                    [[0.0 + 0.0j, 0.25 + 1.5j]
- * #                     [0.0 + 0.0j, 0.75 + 0.0j]]
- * #                    [[0.0 + 0.0j, 0.0  + 0.0j]
- * #                     [0.0 + 0.0j, 0.01 + 0.0j]]]
- * # 'input' has 5 nonzero magnitude values, so output has 5 coordinates.
- * # 'input' has rank of 3, so coordinates have three indices.
- * where(input) ==> [[0, 0, 0],
- *                   [0, 1, 0],
- *                   [1, 0, 1],
- *                   [1, 1, 1],
- *                   [2, 1, 1]]
- * }</pre>
- * 
- */
-@Operator
-public final class Where extends RawOp implements Operand<TInt64> {
-  
-  /**
-   * Factory method to create a class wrapping a new Where operation.
-   * 
-   * @param scope current scope
-   * @param condition 
-   * @return a new instance of Where
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> Where create(Scope scope, Operand<T> condition) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Where", scope.makeOpName("Where"));
-    opBuilder.addInput(condition.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new Where(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<TInt64> index() {
-    return index;
-  }
-  
-  @Override
-  public Output<TInt64> asOutput(Scope scope) {
-    return index;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Where";
-  
-  private Output<TInt64> index;
-  
-  private Where(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    index = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/XlaSpmdFullToShardShape.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/XlaSpmdFullToShardShape.java
deleted file mode 100644
index c28e44252b5..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/XlaSpmdFullToShardShape.java
+++ /dev/null
@@ -1,81 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * An op used by XLA SPMD partitioner to switch from automatic partitioning to
- * <p>
- * manual partitioning. It annotates the input (full-shape, to be automatically
- * partitioned) with the same sharding used by manual partitioning, and outputs a
- * shard-shaped tensor to be consumed by later manually-partitioned ops. If the
- * shape is not evenly partitionable, the padding region will be masked with 0s.
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator
-public final class XlaSpmdFullToShardShape<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new XlaSpmdFullToShardShape operation.
-   * 
-   * @param scope current scope
-   * @param input 
-   * @param manualSharding 
-   * @return a new instance of XlaSpmdFullToShardShape
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> XlaSpmdFullToShardShape<T> create(Scope scope, Operand<T> input, String manualSharding) {
-    OperationBuilder opBuilder = scope.env().opBuilder("XlaSpmdFullToShardShape", scope.makeOpName("XlaSpmdFullToShardShape"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("manual_sharding", manualSharding);
-    return new XlaSpmdFullToShardShape<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "XlaSpmdFullToShardShape";
-  
-  private Output<T> output;
-  
-  private XlaSpmdFullToShardShape(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/XlaSpmdShardToFullShape.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/XlaSpmdShardToFullShape.java
deleted file mode 100644
index ef9d7379d21..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/XlaSpmdShardToFullShape.java
+++ /dev/null
@@ -1,83 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * An op used by XLA SPMD partitioner to switch from manual partitioning to
- * <p>
- * automatic partitioning. It converts the shard-shaped, manually partitioned input
- * into full-shaped tensor to be partitioned automatically with the same sharding
- * used by manual partitioning.
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator
-public final class XlaSpmdShardToFullShape<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new XlaSpmdShardToFullShape operation.
-   * 
-   * @param scope current scope
-   * @param input 
-   * @param manualSharding 
-   * @param fullShape 
-   * @return a new instance of XlaSpmdShardToFullShape
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> XlaSpmdShardToFullShape<T> create(Scope scope, Operand<T> input, String manualSharding, Shape fullShape) {
-    OperationBuilder opBuilder = scope.env().opBuilder("XlaSpmdShardToFullShape", scope.makeOpName("XlaSpmdShardToFullShape"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("manual_sharding", manualSharding);
-    opBuilder.setAttr("full_shape", fullShape);
-    return new XlaSpmdShardToFullShape<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "XlaSpmdShardToFullShape";
-  
-  private Output<T> output;
-  
-  private XlaSpmdShardToFullShape(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ZerosLike.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ZerosLike.java
deleted file mode 100644
index 1d79777cc35..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ZerosLike.java
+++ /dev/null
@@ -1,75 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.core;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Returns a tensor of zeros with the same shape and type as x.
- * 
- * @param <T> data type for {@code y()} output
- */
-@Operator
-public final class ZerosLike<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new ZerosLike operation.
-   * 
-   * @param scope current scope
-   * @param x a tensor of type T.
-   * @return a new instance of ZerosLike
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> ZerosLike<T> create(Scope scope, Operand<T> x) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ZerosLike", scope.makeOpName("ZerosLike"));
-    opBuilder.addInput(x.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new ZerosLike<T>(opBuilder.build());
-  }
-  
-  /**
-   * a tensor of the same shape and type as x but filled with zeros.
-   */
-  public Output<T> y() {
-    return y;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return y;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ZerosLike";
-  
-  private Output<T> y;
-  
-  private ZerosLike(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    y = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/AnonymousIterator.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/AnonymousIterator.java
deleted file mode 100644
index ce6d0515fc2..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/AnonymousIterator.java
+++ /dev/null
@@ -1,90 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.data;
-
-import java.util.List;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-
-/**
- * A container for an iterator resource.
- */
-@Operator(group = "data")
-public final class AnonymousIterator extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new AnonymousIterator operation.
-   * 
-   * @param scope current scope
-   * @param outputTypes 
-   * @param outputShapes 
-   * @return a new instance of AnonymousIterator
-   */
-  @Endpoint(describeByClass = true)
-  public static AnonymousIterator create(Scope scope, List<Class<?>> outputTypes, List<Shape> outputShapes) {
-    OperationBuilder opBuilder = scope.env().opBuilder("AnonymousIteratorV2", scope.makeOpName("AnonymousIterator"));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    Class[] outputTypesArray = new Class[outputTypes.size()];
-    for (int i = 0; i < outputTypesArray.length; ++i) {
-      outputTypesArray[i] = outputTypes.get(i);
-    }
-    opBuilder.setAttr("output_types", outputTypesArray);
-    Shape[] outputShapesArray = new Shape[outputShapes.size()];
-    for (int i = 0; i < outputShapesArray.length; ++i) {
-      outputShapesArray[i] = outputShapes.get(i);
-    }
-    opBuilder.setAttr("output_shapes", outputShapesArray);
-    return new AnonymousIterator(opBuilder.build());
-  }
-  
-  /**
-   * A handle to the iterator that can be passed to a "MakeIterator" or
-   * "IteratorGetNext" op. In contrast to Iterator, AnonymousIterator prevents
-   * resource sharing by name, and does not keep a reference to the resource
-   * container.
-   */
-  public Output<?> handle() {
-    return handle;
-  }
-  
-  /**
-   * A variant deleter that should be passed into the op that deletes the iterator.
-   */
-  public Output<?> deleter() {
-    return deleter;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "AnonymousIteratorV2";
-  
-  private Output<?> handle;
-  private Output<?> deleter;
-  
-  private AnonymousIterator(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    handle = operation.output(outputIdx++);
-    deleter = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/AnonymousMemoryCache.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/AnonymousMemoryCache.java
deleted file mode 100644
index 4bc4523c1ea..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/AnonymousMemoryCache.java
+++ /dev/null
@@ -1,69 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.data;
-
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-
-/**
- */
-public final class AnonymousMemoryCache extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new AnonymousMemoryCache operation.
-   * 
-   * @param scope current scope
-   * @return a new instance of AnonymousMemoryCache
-   */
-  @Endpoint(describeByClass = true)
-  public static AnonymousMemoryCache create(Scope scope) {
-    OperationBuilder opBuilder = scope.env().opBuilder("AnonymousMemoryCache", scope.makeOpName("AnonymousMemoryCache"));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new AnonymousMemoryCache(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<?> handle() {
-    return handle;
-  }
-  
-  /**
-   */
-  public Output<?> deleter() {
-    return deleter;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "AnonymousMemoryCache";
-  
-  private Output<?> handle;
-  private Output<?> deleter;
-  
-  private AnonymousMemoryCache(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    handle = operation.output(outputIdx++);
-    deleter = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/AnonymousMultiDeviceIterator.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/AnonymousMultiDeviceIterator.java
deleted file mode 100644
index 8857254635f..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/AnonymousMultiDeviceIterator.java
+++ /dev/null
@@ -1,95 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.data;
-
-import java.util.List;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-
-/**
- * A container for a multi device iterator resource.
- */
-public final class AnonymousMultiDeviceIterator extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new AnonymousMultiDeviceIterator operation.
-   * 
-   * @param scope current scope
-   * @param devices 
-   * @param outputTypes 
-   * @param outputShapes 
-   * @return a new instance of AnonymousMultiDeviceIterator
-   */
-  @Endpoint(describeByClass = true)
-  public static AnonymousMultiDeviceIterator create(Scope scope, List<String> devices, List<Class<?>> outputTypes, List<Shape> outputShapes) {
-    OperationBuilder opBuilder = scope.env().opBuilder("AnonymousMultiDeviceIterator", scope.makeOpName("AnonymousMultiDeviceIterator"));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    String[] devicesArray = new String[devices.size()];
-    for (int i = 0; i < devicesArray.length; ++i) {
-      devicesArray[i] = devices.get(i);
-    }
-    opBuilder.setAttr("devices", devicesArray);
-    Class[] outputTypesArray = new Class[outputTypes.size()];
-    for (int i = 0; i < outputTypesArray.length; ++i) {
-      outputTypesArray[i] = outputTypes.get(i);
-    }
-    opBuilder.setAttr("output_types", outputTypesArray);
-    Shape[] outputShapesArray = new Shape[outputShapes.size()];
-    for (int i = 0; i < outputShapesArray.length; ++i) {
-      outputShapesArray[i] = outputShapes.get(i);
-    }
-    opBuilder.setAttr("output_shapes", outputShapesArray);
-    return new AnonymousMultiDeviceIterator(opBuilder.build());
-  }
-  
-  /**
-   * A handle to a multi device iterator that can be passed to a
-   * "MultiDeviceIteratorGetNextFromShard" op. In contrast to MultiDeviceIterator,
-   * AnonymousIterator prevents resource sharing by name, and does not keep a
-   * reference to the resource container.
-   */
-  public Output<?> handle() {
-    return handle;
-  }
-  
-  /**
-   * A variant deleter that should be passed into the op that deletes the iterator.
-   */
-  public Output<?> deleter() {
-    return deleter;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "AnonymousMultiDeviceIterator";
-  
-  private Output<?> handle;
-  private Output<?> deleter;
-  
-  private AnonymousMultiDeviceIterator(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    handle = operation.output(outputIdx++);
-    deleter = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/AssertNextDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/AssertNextDataset.java
deleted file mode 100644
index ba317ce9b11..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/AssertNextDataset.java
+++ /dev/null
@@ -1,100 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.data;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TString;
-import org.tensorflow.types.family.TType;
-
-/**
- * A transformation that asserts which transformations happen next.
- * <p>
- * This transformation checks whether the camel-case names (i.e. "FlatMap", not
- * "flat_map") of the transformations following this transformation match the list
- * of names in the `transformations` argument. If there is a mismatch, the
- * transformation raises an exception.
- * <p>
- * The check occurs when iterating over the contents of the dataset, which
- * means that the check happens <i>after</i> any static optimizations are applied
- * to the dataset graph.
- */
-public final class AssertNextDataset extends RawOp implements Operand<TType> {
-  
-  /**
-   * Factory method to create a class wrapping a new AssertNextDataset operation.
-   * 
-   * @param scope current scope
-   * @param inputDataset A variant tensor representing the input dataset.
-   * `data.AssertNextDataset` passes through the outputs of its input dataset.
-   * @param transformations A `tf.string` vector `tf.Tensor` identifying the transformations that are
-   * expected to happen next.
-   * @param outputTypes 
-   * @param outputShapes 
-   * @return a new instance of AssertNextDataset
-   */
-  @Endpoint(describeByClass = true)
-  public static AssertNextDataset create(Scope scope, Operand<?> inputDataset, Operand<TString> transformations, List<Class<?>> outputTypes, List<Shape> outputShapes) {
-    OperationBuilder opBuilder = scope.env().opBuilder("AssertNextDataset", scope.makeOpName("AssertNextDataset"));
-    opBuilder.addInput(inputDataset.asOutput(scope));
-    opBuilder.addInput(transformations.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    Class[] outputTypesArray = new Class[outputTypes.size()];
-    for (int i = 0; i < outputTypesArray.length; ++i) {
-      outputTypesArray[i] = outputTypes.get(i);
-    }
-    opBuilder.setAttr("output_types", outputTypesArray);
-    Shape[] outputShapesArray = new Shape[outputShapes.size()];
-    for (int i = 0; i < outputShapesArray.length; ++i) {
-      outputShapesArray[i] = outputShapes.get(i);
-    }
-    opBuilder.setAttr("output_shapes", outputShapesArray);
-    return new AssertNextDataset(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<?> handle() {
-    return handle;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) handle;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "AssertNextDataset";
-  
-  private Output<?> handle;
-  
-  private AssertNextDataset(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    handle = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/AutoShardDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/AutoShardDataset.java
deleted file mode 100644
index 818c94926cb..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/AutoShardDataset.java
+++ /dev/null
@@ -1,133 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.data;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
-
-/**
- * Creates a dataset that shards the input dataset.
- * <p>
- * Creates a dataset that shards the input dataset by num_workers, returning a
- * sharded dataset for the index-th worker. This attempts to automatically shard
- * a dataset by examining the Dataset graph and inserting a shard op before the
- * inputs to a reader Dataset (e.g. CSVDataset, TFRecordDataset).
- * <p>
- * This dataset will throw a NotFound error if we cannot shard the dataset
- * automatically.
- */
-public final class AutoShardDataset extends RawOp implements Operand<TType> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.data.AutoShardDataset}
-   */
-  public static class Options {
-    
-    /**
-     * @param autoShardPolicy 
-     */
-    public Options autoShardPolicy(Long autoShardPolicy) {
-      this.autoShardPolicy = autoShardPolicy;
-      return this;
-    }
-    
-    private Long autoShardPolicy;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new AutoShardDataset operation.
-   * 
-   * @param scope current scope
-   * @param inputDataset A variant tensor representing the input dataset.
-   * @param numWorkers A scalar representing the number of workers to distribute this dataset across.
-   * @param index A scalar representing the index of the current worker out of num_workers.
-   * @param outputTypes 
-   * @param outputShapes 
-   * @param options carries optional attributes values
-   * @return a new instance of AutoShardDataset
-   */
-  @Endpoint(describeByClass = true)
-  public static AutoShardDataset create(Scope scope, Operand<?> inputDataset, Operand<TInt64> numWorkers, Operand<TInt64> index, List<Class<?>> outputTypes, List<Shape> outputShapes, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("AutoShardDataset", scope.makeOpName("AutoShardDataset"));
-    opBuilder.addInput(inputDataset.asOutput(scope));
-    opBuilder.addInput(numWorkers.asOutput(scope));
-    opBuilder.addInput(index.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    Class[] outputTypesArray = new Class[outputTypes.size()];
-    for (int i = 0; i < outputTypesArray.length; ++i) {
-      outputTypesArray[i] = outputTypes.get(i);
-    }
-    opBuilder.setAttr("output_types", outputTypesArray);
-    Shape[] outputShapesArray = new Shape[outputShapes.size()];
-    for (int i = 0; i < outputShapesArray.length; ++i) {
-      outputShapesArray[i] = outputShapes.get(i);
-    }
-    opBuilder.setAttr("output_shapes", outputShapesArray);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.autoShardPolicy != null) {
-          opBuilder.setAttr("auto_shard_policy", opts.autoShardPolicy);
-        }
-      }
-    }
-    return new AutoShardDataset(opBuilder.build());
-  }
-  
-  /**
-   * @param autoShardPolicy 
-   */
-  public static Options autoShardPolicy(Long autoShardPolicy) {
-    return new Options().autoShardPolicy(autoShardPolicy);
-  }
-  
-  /**
-   */
-  public Output<?> handle() {
-    return handle;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) handle;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "AutoShardDataset";
-  
-  private Output<?> handle;
-  
-  private AutoShardDataset(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    handle = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/BatchDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/BatchDataset.java
deleted file mode 100644
index cb3a36abf99..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/BatchDataset.java
+++ /dev/null
@@ -1,128 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.data;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TBool;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
-
-/**
- * Creates a dataset that batches `batch_size` elements from `input_dataset`.
- */
-@Operator(group = "data")
-public final class BatchDataset extends RawOp implements Operand<TType> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.data.BatchDataset}
-   */
-  public static class Options {
-    
-    /**
-     * @param parallelCopy 
-     */
-    public Options parallelCopy(Boolean parallelCopy) {
-      this.parallelCopy = parallelCopy;
-      return this;
-    }
-    
-    private Boolean parallelCopy;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new BatchDataset operation.
-   * 
-   * @param scope current scope
-   * @param inputDataset 
-   * @param batchSize A scalar representing the number of elements to accumulate in a batch.
-   * @param dropRemainder A scalar representing whether the last batch should be dropped in case its size
-   * is smaller than desired.
-   * @param outputTypes 
-   * @param outputShapes 
-   * @param options carries optional attributes values
-   * @return a new instance of BatchDataset
-   */
-  @Endpoint(describeByClass = true)
-  public static BatchDataset create(Scope scope, Operand<?> inputDataset, Operand<TInt64> batchSize, Operand<TBool> dropRemainder, List<Class<?>> outputTypes, List<Shape> outputShapes, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("BatchDatasetV2", scope.makeOpName("BatchDataset"));
-    opBuilder.addInput(inputDataset.asOutput(scope));
-    opBuilder.addInput(batchSize.asOutput(scope));
-    opBuilder.addInput(dropRemainder.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    Class[] outputTypesArray = new Class[outputTypes.size()];
-    for (int i = 0; i < outputTypesArray.length; ++i) {
-      outputTypesArray[i] = outputTypes.get(i);
-    }
-    opBuilder.setAttr("output_types", outputTypesArray);
-    Shape[] outputShapesArray = new Shape[outputShapes.size()];
-    for (int i = 0; i < outputShapesArray.length; ++i) {
-      outputShapesArray[i] = outputShapes.get(i);
-    }
-    opBuilder.setAttr("output_shapes", outputShapesArray);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.parallelCopy != null) {
-          opBuilder.setAttr("parallel_copy", opts.parallelCopy);
-        }
-      }
-    }
-    return new BatchDataset(opBuilder.build());
-  }
-  
-  /**
-   * @param parallelCopy 
-   */
-  public static Options parallelCopy(Boolean parallelCopy) {
-    return new Options().parallelCopy(parallelCopy);
-  }
-  
-  /**
-   */
-  public Output<?> handle() {
-    return handle;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) handle;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "BatchDatasetV2";
-  
-  private Output<?> handle;
-  
-  private BatchDataset(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    handle = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/BytesProducedStatsDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/BytesProducedStatsDataset.java
deleted file mode 100644
index d6c33dbd767..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/BytesProducedStatsDataset.java
+++ /dev/null
@@ -1,89 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.data;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TString;
-import org.tensorflow.types.family.TType;
-
-/**
- * Records the bytes size of each element of `input_dataset` in a StatsAggregator.
- */
-public final class BytesProducedStatsDataset extends RawOp implements Operand<TType> {
-  
-  /**
-   * Factory method to create a class wrapping a new BytesProducedStatsDataset operation.
-   * 
-   * @param scope current scope
-   * @param inputDataset 
-   * @param tag 
-   * @param outputTypes 
-   * @param outputShapes 
-   * @return a new instance of BytesProducedStatsDataset
-   */
-  @Endpoint(describeByClass = true)
-  public static BytesProducedStatsDataset create(Scope scope, Operand<?> inputDataset, Operand<TString> tag, List<Class<?>> outputTypes, List<Shape> outputShapes) {
-    OperationBuilder opBuilder = scope.env().opBuilder("BytesProducedStatsDataset", scope.makeOpName("BytesProducedStatsDataset"));
-    opBuilder.addInput(inputDataset.asOutput(scope));
-    opBuilder.addInput(tag.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    Class[] outputTypesArray = new Class[outputTypes.size()];
-    for (int i = 0; i < outputTypesArray.length; ++i) {
-      outputTypesArray[i] = outputTypes.get(i);
-    }
-    opBuilder.setAttr("output_types", outputTypesArray);
-    Shape[] outputShapesArray = new Shape[outputShapes.size()];
-    for (int i = 0; i < outputShapesArray.length; ++i) {
-      outputShapesArray[i] = outputShapes.get(i);
-    }
-    opBuilder.setAttr("output_shapes", outputShapesArray);
-    return new BytesProducedStatsDataset(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<?> handle() {
-    return handle;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) handle;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "BytesProducedStatsDataset";
-  
-  private Output<?> handle;
-  
-  private BytesProducedStatsDataset(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    handle = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/CSVDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/CSVDataset.java
deleted file mode 100644
index 1466a507bfc..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/CSVDataset.java
+++ /dev/null
@@ -1,100 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.data;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.op.Operands;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TBool;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.TString;
-import org.tensorflow.types.family.TType;
-
-/**
- */
-@Operator(group = "data")
-public final class CSVDataset extends RawOp implements Operand<TType> {
-  
-  /**
-   * Factory method to create a class wrapping a new CSVDataset operation.
-   * 
-   * @param scope current scope
-   * @param filenames 
-   * @param compressionType 
-   * @param bufferSize 
-   * @param header 
-   * @param fieldDelim 
-   * @param useQuoteDelim 
-   * @param naValue 
-   * @param selectCols 
-   * @param recordDefaults 
-   * @param outputShapes 
-   * @return a new instance of CSVDataset
-   */
-  @Endpoint(describeByClass = true)
-  public static CSVDataset create(Scope scope, Operand<TString> filenames, Operand<TString> compressionType, Operand<TInt64> bufferSize, Operand<TBool> header, Operand<TString> fieldDelim, Operand<TBool> useQuoteDelim, Operand<TString> naValue, Operand<TInt64> selectCols, Iterable<Operand<?>> recordDefaults, List<Shape> outputShapes) {
-    OperationBuilder opBuilder = scope.env().opBuilder("CSVDataset", scope.makeOpName("CSVDataset"));
-    opBuilder.addInput(filenames.asOutput(scope));
-    opBuilder.addInput(compressionType.asOutput(scope));
-    opBuilder.addInput(bufferSize.asOutput(scope));
-    opBuilder.addInput(header.asOutput(scope));
-    opBuilder.addInput(fieldDelim.asOutput(scope));
-    opBuilder.addInput(useQuoteDelim.asOutput(scope));
-    opBuilder.addInput(naValue.asOutput(scope));
-    opBuilder.addInput(selectCols.asOutput(scope));
-    opBuilder.addInputList(Operands.asOutputs(scope, recordDefaults));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    Shape[] outputShapesArray = new Shape[outputShapes.size()];
-    for (int i = 0; i < outputShapesArray.length; ++i) {
-      outputShapesArray[i] = outputShapes.get(i);
-    }
-    opBuilder.setAttr("output_shapes", outputShapesArray);
-    return new CSVDataset(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<?> handle() {
-    return handle;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) handle;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "CSVDataset";
-  
-  private Output<?> handle;
-  
-  private CSVDataset(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    handle = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/CacheDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/CacheDataset.java
deleted file mode 100644
index e2c3d4444e6..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/CacheDataset.java
+++ /dev/null
@@ -1,95 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.data;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TString;
-import org.tensorflow.types.family.TType;
-
-/**
- * Creates a dataset that caches elements from `input_dataset`.
- * <p>
- * A CacheDataset will iterate over the input_dataset, and store tensors. If the
- * cache already exists, the cache will be used. If the cache is inappropriate
- * (e.g. cannot be opened, contains tensors of the wrong shape / size), an error
- * will the returned when used.
- */
-public final class CacheDataset extends RawOp implements Operand<TType> {
-  
-  /**
-   * Factory method to create a class wrapping a new CacheDataset operation.
-   * 
-   * @param scope current scope
-   * @param inputDataset 
-   * @param filename A path on the filesystem where we should cache the dataset. Note: this
-   * will be a directory.
-   * @param outputTypes 
-   * @param outputShapes 
-   * @return a new instance of CacheDataset
-   */
-  @Endpoint(describeByClass = true)
-  public static CacheDataset create(Scope scope, Operand<?> inputDataset, Operand<TString> filename, List<Class<?>> outputTypes, List<Shape> outputShapes) {
-    OperationBuilder opBuilder = scope.env().opBuilder("CacheDataset", scope.makeOpName("CacheDataset"));
-    opBuilder.addInput(inputDataset.asOutput(scope));
-    opBuilder.addInput(filename.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    Class[] outputTypesArray = new Class[outputTypes.size()];
-    for (int i = 0; i < outputTypesArray.length; ++i) {
-      outputTypesArray[i] = outputTypes.get(i);
-    }
-    opBuilder.setAttr("output_types", outputTypesArray);
-    Shape[] outputShapesArray = new Shape[outputShapes.size()];
-    for (int i = 0; i < outputShapesArray.length; ++i) {
-      outputShapesArray[i] = outputShapes.get(i);
-    }
-    opBuilder.setAttr("output_shapes", outputShapesArray);
-    return new CacheDataset(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<?> handle() {
-    return handle;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) handle;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "CacheDataset";
-  
-  private Output<?> handle;
-  
-  private CacheDataset(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    handle = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/CacheDatasetV2.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/CacheDatasetV2.java
deleted file mode 100644
index 7cd69f5fc66..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/CacheDatasetV2.java
+++ /dev/null
@@ -1,90 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.data;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TString;
-import org.tensorflow.types.family.TType;
-
-/**
- */
-public final class CacheDatasetV2 extends RawOp implements Operand<TType> {
-  
-  /**
-   * Factory method to create a class wrapping a new CacheDatasetV2 operation.
-   * 
-   * @param scope current scope
-   * @param inputDataset 
-   * @param filename 
-   * @param cache 
-   * @param outputTypes 
-   * @param outputShapes 
-   * @return a new instance of CacheDatasetV2
-   */
-  @Endpoint(describeByClass = true)
-  public static CacheDatasetV2 create(Scope scope, Operand<?> inputDataset, Operand<TString> filename, Operand<?> cache, List<Class<?>> outputTypes, List<Shape> outputShapes) {
-    OperationBuilder opBuilder = scope.env().opBuilder("CacheDatasetV2", scope.makeOpName("CacheDatasetV2"));
-    opBuilder.addInput(inputDataset.asOutput(scope));
-    opBuilder.addInput(filename.asOutput(scope));
-    opBuilder.addInput(cache.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    Class[] outputTypesArray = new Class[outputTypes.size()];
-    for (int i = 0; i < outputTypesArray.length; ++i) {
-      outputTypesArray[i] = outputTypes.get(i);
-    }
-    opBuilder.setAttr("output_types", outputTypesArray);
-    Shape[] outputShapesArray = new Shape[outputShapes.size()];
-    for (int i = 0; i < outputShapesArray.length; ++i) {
-      outputShapesArray[i] = outputShapes.get(i);
-    }
-    opBuilder.setAttr("output_shapes", outputShapesArray);
-    return new CacheDatasetV2(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<?> handle() {
-    return handle;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) handle;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "CacheDatasetV2";
-  
-  private Output<?> handle;
-  
-  private CacheDatasetV2(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    handle = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ChooseFastestDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ChooseFastestDataset.java
deleted file mode 100644
index d97ddf04970..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ChooseFastestDataset.java
+++ /dev/null
@@ -1,88 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.data;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.op.Operands;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- */
-public final class ChooseFastestDataset extends RawOp implements Operand<TType> {
-  
-  /**
-   * Factory method to create a class wrapping a new ChooseFastestDataset operation.
-   * 
-   * @param scope current scope
-   * @param inputDatasets 
-   * @param numExperiments 
-   * @param outputTypes 
-   * @param outputShapes 
-   * @return a new instance of ChooseFastestDataset
-   */
-  @Endpoint(describeByClass = true)
-  public static ChooseFastestDataset create(Scope scope, Iterable<Operand<?>> inputDatasets, Long numExperiments, List<Class<?>> outputTypes, List<Shape> outputShapes) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ChooseFastestDataset", scope.makeOpName("ChooseFastestDataset"));
-    opBuilder.addInputList(Operands.asOutputs(scope, inputDatasets));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("num_experiments", numExperiments);
-    Class[] outputTypesArray = new Class[outputTypes.size()];
-    for (int i = 0; i < outputTypesArray.length; ++i) {
-      outputTypesArray[i] = outputTypes.get(i);
-    }
-    opBuilder.setAttr("output_types", outputTypesArray);
-    Shape[] outputShapesArray = new Shape[outputShapes.size()];
-    for (int i = 0; i < outputShapesArray.length; ++i) {
-      outputShapesArray[i] = outputShapes.get(i);
-    }
-    opBuilder.setAttr("output_shapes", outputShapesArray);
-    return new ChooseFastestDataset(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<?> handle() {
-    return handle;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) handle;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ChooseFastestDataset";
-  
-  private Output<?> handle;
-  
-  private ChooseFastestDataset(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    handle = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ConcatenateDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ConcatenateDataset.java
deleted file mode 100644
index 5c566d50f0b..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ConcatenateDataset.java
+++ /dev/null
@@ -1,89 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.data;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Creates a dataset that concatenates `input_dataset` with `another_dataset`.
- */
-@Operator(group = "data")
-public final class ConcatenateDataset extends RawOp implements Operand<TType> {
-  
-  /**
-   * Factory method to create a class wrapping a new ConcatenateDataset operation.
-   * 
-   * @param scope current scope
-   * @param inputDataset 
-   * @param anotherDataset 
-   * @param outputTypes 
-   * @param outputShapes 
-   * @return a new instance of ConcatenateDataset
-   */
-  @Endpoint(describeByClass = true)
-  public static ConcatenateDataset create(Scope scope, Operand<?> inputDataset, Operand<?> anotherDataset, List<Class<?>> outputTypes, List<Shape> outputShapes) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ConcatenateDataset", scope.makeOpName("ConcatenateDataset"));
-    opBuilder.addInput(inputDataset.asOutput(scope));
-    opBuilder.addInput(anotherDataset.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    Class[] outputTypesArray = new Class[outputTypes.size()];
-    for (int i = 0; i < outputTypesArray.length; ++i) {
-      outputTypesArray[i] = outputTypes.get(i);
-    }
-    opBuilder.setAttr("output_types", outputTypesArray);
-    Shape[] outputShapesArray = new Shape[outputShapes.size()];
-    for (int i = 0; i < outputShapesArray.length; ++i) {
-      outputShapesArray[i] = outputShapes.get(i);
-    }
-    opBuilder.setAttr("output_shapes", outputShapesArray);
-    return new ConcatenateDataset(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<?> handle() {
-    return handle;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) handle;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ConcatenateDataset";
-  
-  private Output<?> handle;
-  
-  private ConcatenateDataset(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    handle = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DatasetCardinality.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DatasetCardinality.java
deleted file mode 100644
index 71b02bb2977..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DatasetCardinality.java
+++ /dev/null
@@ -1,75 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.data;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt64;
-
-/**
- * Returns the cardinality of `input_dataset`.
- * <p>
- * Returns the cardinality of `input_dataset`.
- */
-public final class DatasetCardinality extends RawOp implements Operand<TInt64> {
-  
-  /**
-   * Factory method to create a class wrapping a new DatasetCardinality operation.
-   * 
-   * @param scope current scope
-   * @param inputDataset A variant tensor representing the dataset to return cardinality for.
-   * @return a new instance of DatasetCardinality
-   */
-  @Endpoint(describeByClass = true)
-  public static DatasetCardinality create(Scope scope, Operand<?> inputDataset) {
-    OperationBuilder opBuilder = scope.env().opBuilder("DatasetCardinality", scope.makeOpName("DatasetCardinality"));
-    opBuilder.addInput(inputDataset.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new DatasetCardinality(opBuilder.build());
-  }
-  
-  /**
-   * The cardinality of `input_dataset`. Named constants are used to represent
-   * infinite and unknown cardinality.
-   */
-  public Output<TInt64> cardinality() {
-    return cardinality;
-  }
-  
-  @Override
-  public Output<TInt64> asOutput(Scope scope) {
-    return cardinality;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "DatasetCardinality";
-  
-  private Output<TInt64> cardinality;
-  
-  private DatasetCardinality(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    cardinality = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DatasetFromGraph.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DatasetFromGraph.java
deleted file mode 100644
index feaa06769fc..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DatasetFromGraph.java
+++ /dev/null
@@ -1,76 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.data;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TString;
-import org.tensorflow.types.family.TType;
-
-/**
- * Creates a dataset from the given `graph_def`.
- * <p>
- * Creates a dataset from the provided `graph_def`.
- */
-public final class DatasetFromGraph extends RawOp implements Operand<TType> {
-  
-  /**
-   * Factory method to create a class wrapping a new DatasetFromGraph operation.
-   * 
-   * @param scope current scope
-   * @param graphDef The graph representation of the dataset (as serialized GraphDef).
-   * @return a new instance of DatasetFromGraph
-   */
-  @Endpoint(describeByClass = true)
-  public static DatasetFromGraph create(Scope scope, Operand<TString> graphDef) {
-    OperationBuilder opBuilder = scope.env().opBuilder("DatasetFromGraph", scope.makeOpName("DatasetFromGraph"));
-    opBuilder.addInput(graphDef.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new DatasetFromGraph(opBuilder.build());
-  }
-  
-  /**
-   * A variant tensor representing the dataset.
-   */
-  public Output<?> handle() {
-    return handle;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) handle;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "DatasetFromGraph";
-  
-  private Output<?> handle;
-  
-  private DatasetFromGraph(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    handle = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DatasetToGraph.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DatasetToGraph.java
deleted file mode 100644
index 805cc39b424..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DatasetToGraph.java
+++ /dev/null
@@ -1,127 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.data;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TString;
-
-/**
- * Returns a serialized GraphDef representing `input_dataset`.
- * <p>
- * Returns a graph representation for `input_dataset`.
- */
-public final class DatasetToGraph extends RawOp implements Operand<TString> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.data.DatasetToGraph}
-   */
-  public static class Options {
-    
-    /**
-     * @param externalStatePolicy 
-     */
-    public Options externalStatePolicy(Long externalStatePolicy) {
-      this.externalStatePolicy = externalStatePolicy;
-      return this;
-    }
-    
-    /**
-     * @param stripDeviceAssignment 
-     */
-    public Options stripDeviceAssignment(Boolean stripDeviceAssignment) {
-      this.stripDeviceAssignment = stripDeviceAssignment;
-      return this;
-    }
-    
-    private Long externalStatePolicy;
-    private Boolean stripDeviceAssignment;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new DatasetToGraph operation.
-   * 
-   * @param scope current scope
-   * @param inputDataset A variant tensor representing the dataset to return the graph representation for.
-   * @param options carries optional attributes values
-   * @return a new instance of DatasetToGraph
-   */
-  @Endpoint(describeByClass = true)
-  public static DatasetToGraph create(Scope scope, Operand<?> inputDataset, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("DatasetToGraphV2", scope.makeOpName("DatasetToGraph"));
-    opBuilder.addInput(inputDataset.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.externalStatePolicy != null) {
-          opBuilder.setAttr("external_state_policy", opts.externalStatePolicy);
-        }
-        if (opts.stripDeviceAssignment != null) {
-          opBuilder.setAttr("strip_device_assignment", opts.stripDeviceAssignment);
-        }
-      }
-    }
-    return new DatasetToGraph(opBuilder.build());
-  }
-  
-  /**
-   * @param externalStatePolicy 
-   */
-  public static Options externalStatePolicy(Long externalStatePolicy) {
-    return new Options().externalStatePolicy(externalStatePolicy);
-  }
-  
-  /**
-   * @param stripDeviceAssignment 
-   */
-  public static Options stripDeviceAssignment(Boolean stripDeviceAssignment) {
-    return new Options().stripDeviceAssignment(stripDeviceAssignment);
-  }
-  
-  /**
-   * The graph representation of the dataset (as serialized GraphDef).
-   */
-  public Output<TString> graph() {
-    return graph;
-  }
-  
-  @Override
-  public Output<TString> asOutput(Scope scope) {
-    return graph;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "DatasetToGraphV2";
-  
-  private Output<TString> graph;
-  
-  private DatasetToGraph(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    graph = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DatasetToSingleElement.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DatasetToSingleElement.java
deleted file mode 100644
index ed49fe284d2..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DatasetToSingleElement.java
+++ /dev/null
@@ -1,91 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.data;
-
-import java.util.Arrays;
-import java.util.Iterator;
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Outputs the single element from the given dataset.
- */
-public final class DatasetToSingleElement extends RawOp implements Iterable<Operand<TType>> {
-  
-  /**
-   * Factory method to create a class wrapping a new DatasetToSingleElement operation.
-   * 
-   * @param scope current scope
-   * @param dataset A handle to a dataset that contains a single element.
-   * @param outputTypes 
-   * @param outputShapes 
-   * @return a new instance of DatasetToSingleElement
-   */
-  @Endpoint(describeByClass = true)
-  public static DatasetToSingleElement create(Scope scope, Operand<?> dataset, List<Class<?>> outputTypes, List<Shape> outputShapes) {
-    OperationBuilder opBuilder = scope.env().opBuilder("DatasetToSingleElement", scope.makeOpName("DatasetToSingleElement"));
-    opBuilder.addInput(dataset.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    Class[] outputTypesArray = new Class[outputTypes.size()];
-    for (int i = 0; i < outputTypesArray.length; ++i) {
-      outputTypesArray[i] = outputTypes.get(i);
-    }
-    opBuilder.setAttr("output_types", outputTypesArray);
-    Shape[] outputShapesArray = new Shape[outputShapes.size()];
-    for (int i = 0; i < outputShapesArray.length; ++i) {
-      outputShapesArray[i] = outputShapes.get(i);
-    }
-    opBuilder.setAttr("output_shapes", outputShapesArray);
-    return new DatasetToSingleElement(opBuilder.build());
-  }
-  
-  /**
-   * The components of the single element of `input`.
-   */
-  public List<Output<?>> components() {
-    return components;
-  }
-  
-  @Override
-  @SuppressWarnings({"rawtypes", "unchecked"})
-  public Iterator<Operand<TType>> iterator() {
-    return (Iterator) components.iterator();
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "DatasetToSingleElement";
-  
-  private List<Output<?>> components;
-  
-  private DatasetToSingleElement(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    int componentsLength = operation.outputListLength("components");
-    components = Arrays.asList(operation.outputList(outputIdx, componentsLength));
-    outputIdx += componentsLength;
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DatasetToTfRecord.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DatasetToTfRecord.java
deleted file mode 100644
index 016eec90e86..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DatasetToTfRecord.java
+++ /dev/null
@@ -1,60 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.data;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TString;
-
-/**
- * Writes the given dataset to the given file using the TFRecord format.
- */
-public final class DatasetToTfRecord extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new DatasetToTfRecord operation.
-   * 
-   * @param scope current scope
-   * @param inputDataset A variant tensor representing the dataset to write.
-   * @param filename A scalar string tensor representing the filename to use.
-   * @param compressionType A scalar string tensor containing either (i) the empty string (no
-   * compression), (ii) "ZLIB", or (iii) "GZIP".
-   * @return a new instance of DatasetToTfRecord
-   */
-  @Endpoint(describeByClass = true)
-  public static DatasetToTfRecord create(Scope scope, Operand<?> inputDataset, Operand<TString> filename, Operand<TString> compressionType) {
-    OperationBuilder opBuilder = scope.env().opBuilder("DatasetToTFRecord", scope.makeOpName("DatasetToTfRecord"));
-    opBuilder.addInput(inputDataset.asOutput(scope));
-    opBuilder.addInput(filename.asOutput(scope));
-    opBuilder.addInput(compressionType.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new DatasetToTfRecord(opBuilder.build());
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "DatasetToTFRecord";
-  
-  private DatasetToTfRecord(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DeleteIterator.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DeleteIterator.java
deleted file mode 100644
index 384f7f8369c..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DeleteIterator.java
+++ /dev/null
@@ -1,57 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.data;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-
-/**
- * A container for an iterator resource.
- */
-@Operator(group = "data")
-public final class DeleteIterator extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new DeleteIterator operation.
-   * 
-   * @param scope current scope
-   * @param handle A handle to the iterator to delete.
-   * @param deleter A variant deleter.
-   * @return a new instance of DeleteIterator
-   */
-  @Endpoint(describeByClass = true)
-  public static DeleteIterator create(Scope scope, Operand<?> handle, Operand<?> deleter) {
-    OperationBuilder opBuilder = scope.env().opBuilder("DeleteIterator", scope.makeOpName("DeleteIterator"));
-    opBuilder.addInput(handle.asOutput(scope));
-    opBuilder.addInput(deleter.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new DeleteIterator(opBuilder.build());
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "DeleteIterator";
-  
-  private DeleteIterator(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DeleteMemoryCache.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DeleteMemoryCache.java
deleted file mode 100644
index c83f6878dc2..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DeleteMemoryCache.java
+++ /dev/null
@@ -1,55 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.data;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-
-/**
- */
-public final class DeleteMemoryCache extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new DeleteMemoryCache operation.
-   * 
-   * @param scope current scope
-   * @param handle 
-   * @param deleter 
-   * @return a new instance of DeleteMemoryCache
-   */
-  @Endpoint(describeByClass = true)
-  public static DeleteMemoryCache create(Scope scope, Operand<?> handle, Operand<?> deleter) {
-    OperationBuilder opBuilder = scope.env().opBuilder("DeleteMemoryCache", scope.makeOpName("DeleteMemoryCache"));
-    opBuilder.addInput(handle.asOutput(scope));
-    opBuilder.addInput(deleter.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new DeleteMemoryCache(opBuilder.build());
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "DeleteMemoryCache";
-  
-  private DeleteMemoryCache(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DeleteMultiDeviceIterator.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DeleteMultiDeviceIterator.java
deleted file mode 100644
index de98ff6a78e..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DeleteMultiDeviceIterator.java
+++ /dev/null
@@ -1,59 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.data;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.Operands;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-
-/**
- * A container for an iterator resource.
- */
-public final class DeleteMultiDeviceIterator extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new DeleteMultiDeviceIterator operation.
-   * 
-   * @param scope current scope
-   * @param multiDeviceIterator A handle to the multi device iterator to delete.
-   * @param iterators A list of iterator handles (unused). This is added so that automatic control dependencies get added during function tracing that ensure this op runs after all the dependent iterators are deleted.
-   * @param deleter A variant deleter.
-   * @return a new instance of DeleteMultiDeviceIterator
-   */
-  @Endpoint(describeByClass = true)
-  public static DeleteMultiDeviceIterator create(Scope scope, Operand<?> multiDeviceIterator, Iterable<Operand<?>> iterators, Operand<?> deleter) {
-    OperationBuilder opBuilder = scope.env().opBuilder("DeleteMultiDeviceIterator", scope.makeOpName("DeleteMultiDeviceIterator"));
-    opBuilder.addInput(multiDeviceIterator.asOutput(scope));
-    opBuilder.addInputList(Operands.asOutputs(scope, iterators));
-    opBuilder.addInput(deleter.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new DeleteMultiDeviceIterator(opBuilder.build());
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "DeleteMultiDeviceIterator";
-  
-  private DeleteMultiDeviceIterator(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DenseToSparseBatchDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DenseToSparseBatchDataset.java
deleted file mode 100644
index a8566b736d5..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DenseToSparseBatchDataset.java
+++ /dev/null
@@ -1,94 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.data;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
-
-/**
- * Creates a dataset that batches input elements into a SparseTensor.
- */
-public final class DenseToSparseBatchDataset extends RawOp implements Operand<TType> {
-  
-  /**
-   * Factory method to create a class wrapping a new DenseToSparseBatchDataset operation.
-   * 
-   * @param scope current scope
-   * @param inputDataset A handle to an input dataset. Must have a single component.
-   * @param batchSize A scalar representing the number of elements to accumulate in a
-   * batch.
-   * @param rowShape A vector representing the dense shape of each row in the produced
-   * SparseTensor. The shape may be partially specified, using `-1` to indicate
-   * that a particular dimension should use the maximum size of all batch elements.
-   * @param outputTypes 
-   * @param outputShapes 
-   * @return a new instance of DenseToSparseBatchDataset
-   */
-  @Endpoint(describeByClass = true)
-  public static DenseToSparseBatchDataset create(Scope scope, Operand<?> inputDataset, Operand<TInt64> batchSize, Operand<TInt64> rowShape, List<Class<?>> outputTypes, List<Shape> outputShapes) {
-    OperationBuilder opBuilder = scope.env().opBuilder("DenseToSparseBatchDataset", scope.makeOpName("DenseToSparseBatchDataset"));
-    opBuilder.addInput(inputDataset.asOutput(scope));
-    opBuilder.addInput(batchSize.asOutput(scope));
-    opBuilder.addInput(rowShape.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    Class[] outputTypesArray = new Class[outputTypes.size()];
-    for (int i = 0; i < outputTypesArray.length; ++i) {
-      outputTypesArray[i] = outputTypes.get(i);
-    }
-    opBuilder.setAttr("output_types", outputTypesArray);
-    Shape[] outputShapesArray = new Shape[outputShapes.size()];
-    for (int i = 0; i < outputShapesArray.length; ++i) {
-      outputShapesArray[i] = outputShapes.get(i);
-    }
-    opBuilder.setAttr("output_shapes", outputShapesArray);
-    return new DenseToSparseBatchDataset(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<?> handle() {
-    return handle;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) handle;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "DenseToSparseBatchDataset";
-  
-  private Output<?> handle;
-  
-  private DenseToSparseBatchDataset(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    handle = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DeserializeIterator.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DeserializeIterator.java
deleted file mode 100644
index fb7873f6f42..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DeserializeIterator.java
+++ /dev/null
@@ -1,58 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.data;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-
-/**
- * Converts the given variant tensor to an iterator and stores it in the given resource.
- */
-@Operator(group = "data")
-public final class DeserializeIterator extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new DeserializeIterator operation.
-   * 
-   * @param scope current scope
-   * @param resourceHandle A handle to an iterator resource.
-   * @param serialized A variant tensor storing the state of the iterator contained in the
-   * resource.
-   * @return a new instance of DeserializeIterator
-   */
-  @Endpoint(describeByClass = true)
-  public static DeserializeIterator create(Scope scope, Operand<?> resourceHandle, Operand<?> serialized) {
-    OperationBuilder opBuilder = scope.env().opBuilder("DeserializeIterator", scope.makeOpName("DeserializeIterator"));
-    opBuilder.addInput(resourceHandle.asOutput(scope));
-    opBuilder.addInput(serialized.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new DeserializeIterator(opBuilder.build());
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "DeserializeIterator";
-  
-  private DeserializeIterator(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DirectedInterleaveDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DirectedInterleaveDataset.java
deleted file mode 100644
index eb196480822..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DirectedInterleaveDataset.java
+++ /dev/null
@@ -1,91 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.data;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.op.Operands;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * A substitute for `InterleaveDataset` on a fixed list of `N` datasets.
- */
-public final class DirectedInterleaveDataset extends RawOp implements Operand<TType> {
-  
-  /**
-   * Factory method to create a class wrapping a new DirectedInterleaveDataset operation.
-   * 
-   * @param scope current scope
-   * @param selectorInputDataset A dataset of scalar `DT_INT64` elements that determines which of the
-   * `N` data inputs should produce the next output element.
-   * @param dataInputDatasets `N` datasets with the same type that will be interleaved according to
-   * the values of `selector_input_dataset`.
-   * @param outputTypes 
-   * @param outputShapes 
-   * @return a new instance of DirectedInterleaveDataset
-   */
-  @Endpoint(describeByClass = true)
-  public static DirectedInterleaveDataset create(Scope scope, Operand<?> selectorInputDataset, Iterable<Operand<?>> dataInputDatasets, List<Class<?>> outputTypes, List<Shape> outputShapes) {
-    OperationBuilder opBuilder = scope.env().opBuilder("DirectedInterleaveDataset", scope.makeOpName("DirectedInterleaveDataset"));
-    opBuilder.addInput(selectorInputDataset.asOutput(scope));
-    opBuilder.addInputList(Operands.asOutputs(scope, dataInputDatasets));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    Class[] outputTypesArray = new Class[outputTypes.size()];
-    for (int i = 0; i < outputTypesArray.length; ++i) {
-      outputTypesArray[i] = outputTypes.get(i);
-    }
-    opBuilder.setAttr("output_types", outputTypesArray);
-    Shape[] outputShapesArray = new Shape[outputShapes.size()];
-    for (int i = 0; i < outputShapesArray.length; ++i) {
-      outputShapesArray[i] = outputShapes.get(i);
-    }
-    opBuilder.setAttr("output_shapes", outputShapesArray);
-    return new DirectedInterleaveDataset(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<?> handle() {
-    return handle;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) handle;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "DirectedInterleaveDataset";
-  
-  private Output<?> handle;
-  
-  private DirectedInterleaveDataset(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    handle = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/FilterByLastComponentDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/FilterByLastComponentDataset.java
deleted file mode 100644
index 618c6b07d2e..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/FilterByLastComponentDataset.java
+++ /dev/null
@@ -1,86 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.data;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Creates a dataset containing elements of first component of `input_dataset` having true in the last component.
- */
-public final class FilterByLastComponentDataset extends RawOp implements Operand<TType> {
-  
-  /**
-   * Factory method to create a class wrapping a new FilterByLastComponentDataset operation.
-   * 
-   * @param scope current scope
-   * @param inputDataset 
-   * @param outputTypes 
-   * @param outputShapes 
-   * @return a new instance of FilterByLastComponentDataset
-   */
-  @Endpoint(describeByClass = true)
-  public static FilterByLastComponentDataset create(Scope scope, Operand<?> inputDataset, List<Class<?>> outputTypes, List<Shape> outputShapes) {
-    OperationBuilder opBuilder = scope.env().opBuilder("FilterByLastComponentDataset", scope.makeOpName("FilterByLastComponentDataset"));
-    opBuilder.addInput(inputDataset.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    Class[] outputTypesArray = new Class[outputTypes.size()];
-    for (int i = 0; i < outputTypesArray.length; ++i) {
-      outputTypesArray[i] = outputTypes.get(i);
-    }
-    opBuilder.setAttr("output_types", outputTypesArray);
-    Shape[] outputShapesArray = new Shape[outputShapes.size()];
-    for (int i = 0; i < outputShapesArray.length; ++i) {
-      outputShapesArray[i] = outputShapes.get(i);
-    }
-    opBuilder.setAttr("output_shapes", outputShapesArray);
-    return new FilterByLastComponentDataset(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<?> output() {
-    return output;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "FilterByLastComponentDataset";
-  
-  private Output<?> output;
-  
-  private FilterByLastComponentDataset(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/FixedLengthRecordDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/FixedLengthRecordDataset.java
deleted file mode 100644
index 9dc927efb6f..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/FixedLengthRecordDataset.java
+++ /dev/null
@@ -1,83 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.data;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.TString;
-import org.tensorflow.types.family.TType;
-
-/**
- */
-public final class FixedLengthRecordDataset extends RawOp implements Operand<TType> {
-  
-  /**
-   * Factory method to create a class wrapping a new FixedLengthRecordDataset operation.
-   * 
-   * @param scope current scope
-   * @param filenames 
-   * @param headerBytes 
-   * @param recordBytes 
-   * @param footerBytes 
-   * @param bufferSize 
-   * @param compressionType 
-   * @return a new instance of FixedLengthRecordDataset
-   */
-  @Endpoint(describeByClass = true)
-  public static FixedLengthRecordDataset create(Scope scope, Operand<TString> filenames, Operand<TInt64> headerBytes, Operand<TInt64> recordBytes, Operand<TInt64> footerBytes, Operand<TInt64> bufferSize, Operand<TString> compressionType) {
-    OperationBuilder opBuilder = scope.env().opBuilder("FixedLengthRecordDatasetV2", scope.makeOpName("FixedLengthRecordDataset"));
-    opBuilder.addInput(filenames.asOutput(scope));
-    opBuilder.addInput(headerBytes.asOutput(scope));
-    opBuilder.addInput(recordBytes.asOutput(scope));
-    opBuilder.addInput(footerBytes.asOutput(scope));
-    opBuilder.addInput(bufferSize.asOutput(scope));
-    opBuilder.addInput(compressionType.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new FixedLengthRecordDataset(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<?> handle() {
-    return handle;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) handle;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "FixedLengthRecordDatasetV2";
-  
-  private Output<?> handle;
-  
-  private FixedLengthRecordDataset(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    handle = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/IgnoreErrorsDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/IgnoreErrorsDataset.java
deleted file mode 100644
index dbc01878164..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/IgnoreErrorsDataset.java
+++ /dev/null
@@ -1,86 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.data;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Creates a dataset that contains the elements of `input_dataset` ignoring errors.
- */
-public final class IgnoreErrorsDataset extends RawOp implements Operand<TType> {
-  
-  /**
-   * Factory method to create a class wrapping a new IgnoreErrorsDataset operation.
-   * 
-   * @param scope current scope
-   * @param inputDataset 
-   * @param outputTypes 
-   * @param outputShapes 
-   * @return a new instance of IgnoreErrorsDataset
-   */
-  @Endpoint(describeByClass = true)
-  public static IgnoreErrorsDataset create(Scope scope, Operand<?> inputDataset, List<Class<?>> outputTypes, List<Shape> outputShapes) {
-    OperationBuilder opBuilder = scope.env().opBuilder("IgnoreErrorsDataset", scope.makeOpName("IgnoreErrorsDataset"));
-    opBuilder.addInput(inputDataset.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    Class[] outputTypesArray = new Class[outputTypes.size()];
-    for (int i = 0; i < outputTypesArray.length; ++i) {
-      outputTypesArray[i] = outputTypes.get(i);
-    }
-    opBuilder.setAttr("output_types", outputTypesArray);
-    Shape[] outputShapesArray = new Shape[outputShapes.size()];
-    for (int i = 0; i < outputShapesArray.length; ++i) {
-      outputShapesArray[i] = outputShapes.get(i);
-    }
-    opBuilder.setAttr("output_shapes", outputShapesArray);
-    return new IgnoreErrorsDataset(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<?> handle() {
-    return handle;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) handle;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "IgnoreErrorsDataset";
-  
-  private Output<?> handle;
-  
-  private IgnoreErrorsDataset(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    handle = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/InitializeTableFromDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/InitializeTableFromDataset.java
deleted file mode 100644
index 5a1d007d572..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/InitializeTableFromDataset.java
+++ /dev/null
@@ -1,55 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.data;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-
-/**
- */
-public final class InitializeTableFromDataset extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new InitializeTableFromDataset operation.
-   * 
-   * @param scope current scope
-   * @param tableHandle 
-   * @param dataset 
-   * @return a new instance of InitializeTableFromDataset
-   */
-  @Endpoint(describeByClass = true)
-  public static InitializeTableFromDataset create(Scope scope, Operand<?> tableHandle, Operand<?> dataset) {
-    OperationBuilder opBuilder = scope.env().opBuilder("InitializeTableFromDataset", scope.makeOpName("InitializeTableFromDataset"));
-    opBuilder.addInput(tableHandle.asOutput(scope));
-    opBuilder.addInput(dataset.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new InitializeTableFromDataset(opBuilder.build());
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "InitializeTableFromDataset";
-  
-  private InitializeTableFromDataset(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/Iterator.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/Iterator.java
deleted file mode 100644
index 427120ded4d..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/Iterator.java
+++ /dev/null
@@ -1,88 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.data;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- */
-@Operator(group = "data")
-public final class Iterator extends RawOp implements Operand<TType> {
-  
-  /**
-   * Factory method to create a class wrapping a new Iterator operation.
-   * 
-   * @param scope current scope
-   * @param sharedName 
-   * @param container 
-   * @param outputTypes 
-   * @param outputShapes 
-   * @return a new instance of Iterator
-   */
-  @Endpoint(describeByClass = true)
-  public static Iterator create(Scope scope, String sharedName, String container, List<Class<?>> outputTypes, List<Shape> outputShapes) {
-    OperationBuilder opBuilder = scope.env().opBuilder("IteratorV2", scope.makeOpName("Iterator"));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("shared_name", sharedName);
-    opBuilder.setAttr("container", container);
-    Class[] outputTypesArray = new Class[outputTypes.size()];
-    for (int i = 0; i < outputTypesArray.length; ++i) {
-      outputTypesArray[i] = outputTypes.get(i);
-    }
-    opBuilder.setAttr("output_types", outputTypesArray);
-    Shape[] outputShapesArray = new Shape[outputShapes.size()];
-    for (int i = 0; i < outputShapesArray.length; ++i) {
-      outputShapesArray[i] = outputShapes.get(i);
-    }
-    opBuilder.setAttr("output_shapes", outputShapesArray);
-    return new Iterator(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<?> handle() {
-    return handle;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) handle;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "IteratorV2";
-  
-  private Output<?> handle;
-  
-  private Iterator(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    handle = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/IteratorFromStringHandle.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/IteratorFromStringHandle.java
deleted file mode 100644
index c36d2ea804d..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/IteratorFromStringHandle.java
+++ /dev/null
@@ -1,118 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.data;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TString;
-import org.tensorflow.types.family.TType;
-
-/**
- */
-public final class IteratorFromStringHandle extends RawOp implements Operand<TType> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.data.IteratorFromStringHandle}
-   */
-  public static class Options {
-    
-    /**
-     * @param outputShapes 
-     */
-    public Options outputShapes(List<Shape> outputShapes) {
-      this.outputShapes = outputShapes;
-      return this;
-    }
-    
-    private List<Shape> outputShapes;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new IteratorFromStringHandle operation.
-   * 
-   * @param scope current scope
-   * @param stringHandle 
-   * @param outputTypes 
-   * @param options carries optional attributes values
-   * @return a new instance of IteratorFromStringHandle
-   */
-  @Endpoint(describeByClass = true)
-  public static IteratorFromStringHandle create(Scope scope, Operand<TString> stringHandle, List<Class<?>> outputTypes, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("IteratorFromStringHandleV2", scope.makeOpName("IteratorFromStringHandle"));
-    opBuilder.addInput(stringHandle.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    Class[] outputTypesArray = new Class[outputTypes.size()];
-    for (int i = 0; i < outputTypesArray.length; ++i) {
-      outputTypesArray[i] = outputTypes.get(i);
-    }
-    opBuilder.setAttr("output_types", outputTypesArray);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.outputShapes != null) {
-          Shape[] outputShapesArray = new Shape[opts.outputShapes.size()];
-          for (int i = 0; i < outputShapesArray.length; ++i) {
-            outputShapesArray[i] = opts.outputShapes.get(i);
-          }
-          opBuilder.setAttr("output_shapes", outputShapesArray);
-        }
-      }
-    }
-    return new IteratorFromStringHandle(opBuilder.build());
-  }
-  
-  /**
-   * @param outputShapes 
-   */
-  public static Options outputShapes(List<Shape> outputShapes) {
-    return new Options().outputShapes(outputShapes);
-  }
-  
-  /**
-   */
-  public Output<?> resourceHandle() {
-    return resourceHandle;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) resourceHandle;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "IteratorFromStringHandleV2";
-  
-  private Output<?> resourceHandle;
-  
-  private IteratorFromStringHandle(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    resourceHandle = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/IteratorGetDevice.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/IteratorGetDevice.java
deleted file mode 100644
index d02fadac0f7..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/IteratorGetDevice.java
+++ /dev/null
@@ -1,71 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.data;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TString;
-
-/**
- * Returns the name of the device on which `resource` has been placed.
- */
-public final class IteratorGetDevice extends RawOp implements Operand<TString> {
-  
-  /**
-   * Factory method to create a class wrapping a new IteratorGetDevice operation.
-   * 
-   * @param scope current scope
-   * @param resource 
-   * @return a new instance of IteratorGetDevice
-   */
-  @Endpoint(describeByClass = true)
-  public static IteratorGetDevice create(Scope scope, Operand<?> resource) {
-    OperationBuilder opBuilder = scope.env().opBuilder("IteratorGetDevice", scope.makeOpName("IteratorGetDevice"));
-    opBuilder.addInput(resource.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new IteratorGetDevice(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<TString> device() {
-    return device;
-  }
-  
-  @Override
-  public Output<TString> asOutput(Scope scope) {
-    return device;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "IteratorGetDevice";
-  
-  private Output<TString> device;
-  
-  private IteratorGetDevice(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    device = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/IteratorGetNext.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/IteratorGetNext.java
deleted file mode 100644
index fb71fc18c97..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/IteratorGetNext.java
+++ /dev/null
@@ -1,91 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.data;
-
-import java.util.Arrays;
-import java.util.Iterator;
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Gets the next output from the given iterator .
- */
-@Operator(group = "data")
-public final class IteratorGetNext extends RawOp implements Iterable<Operand<TType>> {
-  
-  /**
-   * Factory method to create a class wrapping a new IteratorGetNext operation.
-   * 
-   * @param scope current scope
-   * @param iterator 
-   * @param outputTypes 
-   * @param outputShapes 
-   * @return a new instance of IteratorGetNext
-   */
-  @Endpoint(describeByClass = true)
-  public static IteratorGetNext create(Scope scope, Operand<?> iterator, List<Class<?>> outputTypes, List<Shape> outputShapes) {
-    OperationBuilder opBuilder = scope.env().opBuilder("IteratorGetNext", scope.makeOpName("IteratorGetNext"));
-    opBuilder.addInput(iterator.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    Class[] outputTypesArray = new Class[outputTypes.size()];
-    for (int i = 0; i < outputTypesArray.length; ++i) {
-      outputTypesArray[i] = outputTypes.get(i);
-    }
-    opBuilder.setAttr("output_types", outputTypesArray);
-    Shape[] outputShapesArray = new Shape[outputShapes.size()];
-    for (int i = 0; i < outputShapesArray.length; ++i) {
-      outputShapesArray[i] = outputShapes.get(i);
-    }
-    opBuilder.setAttr("output_shapes", outputShapesArray);
-    return new IteratorGetNext(opBuilder.build());
-  }
-  
-  /**
-   */
-  public List<Output<?>> components() {
-    return components;
-  }
-  
-  @Override
-  @SuppressWarnings({"rawtypes", "unchecked"})
-  public Iterator<Operand<TType>> iterator() {
-    return (Iterator) components.iterator();
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "IteratorGetNext";
-  
-  private List<Output<?>> components;
-  
-  private IteratorGetNext(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    int componentsLength = operation.outputListLength("components");
-    components = Arrays.asList(operation.outputList(outputIdx, componentsLength));
-    outputIdx += componentsLength;
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/IteratorGetNextAsOptional.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/IteratorGetNextAsOptional.java
deleted file mode 100644
index 70e0a91c8e2..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/IteratorGetNextAsOptional.java
+++ /dev/null
@@ -1,87 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.data;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Gets the next output from the given iterator as an Optional variant.
- */
-@Operator(group = "data")
-public final class IteratorGetNextAsOptional extends RawOp implements Operand<TType> {
-  
-  /**
-   * Factory method to create a class wrapping a new IteratorGetNextAsOptional operation.
-   * 
-   * @param scope current scope
-   * @param iterator 
-   * @param outputTypes 
-   * @param outputShapes 
-   * @return a new instance of IteratorGetNextAsOptional
-   */
-  @Endpoint(describeByClass = true)
-  public static IteratorGetNextAsOptional create(Scope scope, Operand<?> iterator, List<Class<?>> outputTypes, List<Shape> outputShapes) {
-    OperationBuilder opBuilder = scope.env().opBuilder("IteratorGetNextAsOptional", scope.makeOpName("IteratorGetNextAsOptional"));
-    opBuilder.addInput(iterator.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    Class[] outputTypesArray = new Class[outputTypes.size()];
-    for (int i = 0; i < outputTypesArray.length; ++i) {
-      outputTypesArray[i] = outputTypes.get(i);
-    }
-    opBuilder.setAttr("output_types", outputTypesArray);
-    Shape[] outputShapesArray = new Shape[outputShapes.size()];
-    for (int i = 0; i < outputShapesArray.length; ++i) {
-      outputShapesArray[i] = outputShapes.get(i);
-    }
-    opBuilder.setAttr("output_shapes", outputShapesArray);
-    return new IteratorGetNextAsOptional(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<?> optional() {
-    return optional;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) optional;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "IteratorGetNextAsOptional";
-  
-  private Output<?> optional;
-  
-  private IteratorGetNextAsOptional(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    optional = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/IteratorGetNextSync.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/IteratorGetNextSync.java
deleted file mode 100644
index b211a5b1d48..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/IteratorGetNextSync.java
+++ /dev/null
@@ -1,96 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.data;
-
-import java.util.Arrays;
-import java.util.Iterator;
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Gets the next output from the given iterator.
- * <p>
- * This operation is a synchronous version IteratorGetNext. It should only be used
- * in situations where the iterator does not block the calling thread, or where
- * the calling thread is not a member of the thread pool used to execute parallel
- * operations (e.g. in eager mode).
- */
-@Operator(group = "data")
-public final class IteratorGetNextSync extends RawOp implements Iterable<Operand<TType>> {
-  
-  /**
-   * Factory method to create a class wrapping a new IteratorGetNextSync operation.
-   * 
-   * @param scope current scope
-   * @param iterator 
-   * @param outputTypes 
-   * @param outputShapes 
-   * @return a new instance of IteratorGetNextSync
-   */
-  @Endpoint(describeByClass = true)
-  public static IteratorGetNextSync create(Scope scope, Operand<?> iterator, List<Class<?>> outputTypes, List<Shape> outputShapes) {
-    OperationBuilder opBuilder = scope.env().opBuilder("IteratorGetNextSync", scope.makeOpName("IteratorGetNextSync"));
-    opBuilder.addInput(iterator.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    Class[] outputTypesArray = new Class[outputTypes.size()];
-    for (int i = 0; i < outputTypesArray.length; ++i) {
-      outputTypesArray[i] = outputTypes.get(i);
-    }
-    opBuilder.setAttr("output_types", outputTypesArray);
-    Shape[] outputShapesArray = new Shape[outputShapes.size()];
-    for (int i = 0; i < outputShapesArray.length; ++i) {
-      outputShapesArray[i] = outputShapes.get(i);
-    }
-    opBuilder.setAttr("output_shapes", outputShapesArray);
-    return new IteratorGetNextSync(opBuilder.build());
-  }
-  
-  /**
-   */
-  public List<Output<?>> components() {
-    return components;
-  }
-  
-  @Override
-  @SuppressWarnings({"rawtypes", "unchecked"})
-  public Iterator<Operand<TType>> iterator() {
-    return (Iterator) components.iterator();
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "IteratorGetNextSync";
-  
-  private List<Output<?>> components;
-  
-  private IteratorGetNextSync(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    int componentsLength = operation.outputListLength("components");
-    components = Arrays.asList(operation.outputList(outputIdx, componentsLength));
-    outputIdx += componentsLength;
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/IteratorToStringHandle.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/IteratorToStringHandle.java
deleted file mode 100644
index b868405a984..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/IteratorToStringHandle.java
+++ /dev/null
@@ -1,73 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.data;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TString;
-
-/**
- * Converts the given `resource_handle` representing an iterator to a string.
- */
-@Operator(group = "data")
-public final class IteratorToStringHandle extends RawOp implements Operand<TString> {
-  
-  /**
-   * Factory method to create a class wrapping a new IteratorToStringHandle operation.
-   * 
-   * @param scope current scope
-   * @param resourceHandle A handle to an iterator resource.
-   * @return a new instance of IteratorToStringHandle
-   */
-  @Endpoint(describeByClass = true)
-  public static IteratorToStringHandle create(Scope scope, Operand<?> resourceHandle) {
-    OperationBuilder opBuilder = scope.env().opBuilder("IteratorToStringHandle", scope.makeOpName("IteratorToStringHandle"));
-    opBuilder.addInput(resourceHandle.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new IteratorToStringHandle(opBuilder.build());
-  }
-  
-  /**
-   * A string representation of the given handle.
-   */
-  public Output<TString> stringHandle() {
-    return stringHandle;
-  }
-  
-  @Override
-  public Output<TString> asOutput(Scope scope) {
-    return stringHandle;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "IteratorToStringHandle";
-  
-  private Output<TString> stringHandle;
-  
-  private IteratorToStringHandle(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    stringHandle = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/LMDBDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/LMDBDataset.java
deleted file mode 100644
index 51433bfe9cf..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/LMDBDataset.java
+++ /dev/null
@@ -1,99 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.data;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TString;
-import org.tensorflow.types.family.TType;
-
-/**
- * Creates a dataset that emits the key-value pairs in one or more LMDB files.
- * <p>
- * The Lightning Memory-Mapped Database Manager, or LMDB, is an embedded binary
- * key-value database. This dataset can read the contents of LMDB database files,
- * the names of which generally have the `.mdb` suffix.
- * <p>
- * Each output element consists of a key-value pair represented as a pair of
- * scalar string `Tensor`s, where the first `Tensor` contains the key and the
- * second `Tensor` contains the value.
- * <p>
- * LMDB uses different file formats on big- and little-endian machines.
- * `data.LMDBDataset` can only read files in the format of the host machine.
- */
-public final class LMDBDataset extends RawOp implements Operand<TType> {
-  
-  /**
-   * Factory method to create a class wrapping a new LMDBDataset operation.
-   * 
-   * @param scope current scope
-   * @param filenames A scalar or a vector containing the name(s) of the binary file(s) to be
-   * read.
-   * @param outputTypes 
-   * @param outputShapes 
-   * @return a new instance of LMDBDataset
-   */
-  @Endpoint(describeByClass = true)
-  public static LMDBDataset create(Scope scope, Operand<TString> filenames, List<Class<?>> outputTypes, List<Shape> outputShapes) {
-    OperationBuilder opBuilder = scope.env().opBuilder("LMDBDataset", scope.makeOpName("LMDBDataset"));
-    opBuilder.addInput(filenames.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    Class[] outputTypesArray = new Class[outputTypes.size()];
-    for (int i = 0; i < outputTypesArray.length; ++i) {
-      outputTypesArray[i] = outputTypes.get(i);
-    }
-    opBuilder.setAttr("output_types", outputTypesArray);
-    Shape[] outputShapesArray = new Shape[outputShapes.size()];
-    for (int i = 0; i < outputShapesArray.length; ++i) {
-      outputShapesArray[i] = outputShapes.get(i);
-    }
-    opBuilder.setAttr("output_shapes", outputShapesArray);
-    return new LMDBDataset(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<?> handle() {
-    return handle;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) handle;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "LMDBDataset";
-  
-  private Output<?> handle;
-  
-  private LMDBDataset(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    handle = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/LatencyStatsDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/LatencyStatsDataset.java
deleted file mode 100644
index 2f4ec7d69d7..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/LatencyStatsDataset.java
+++ /dev/null
@@ -1,89 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.data;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TString;
-import org.tensorflow.types.family.TType;
-
-/**
- * Records the latency of producing `input_dataset` elements in a StatsAggregator.
- */
-public final class LatencyStatsDataset extends RawOp implements Operand<TType> {
-  
-  /**
-   * Factory method to create a class wrapping a new LatencyStatsDataset operation.
-   * 
-   * @param scope current scope
-   * @param inputDataset 
-   * @param tag 
-   * @param outputTypes 
-   * @param outputShapes 
-   * @return a new instance of LatencyStatsDataset
-   */
-  @Endpoint(describeByClass = true)
-  public static LatencyStatsDataset create(Scope scope, Operand<?> inputDataset, Operand<TString> tag, List<Class<?>> outputTypes, List<Shape> outputShapes) {
-    OperationBuilder opBuilder = scope.env().opBuilder("LatencyStatsDataset", scope.makeOpName("LatencyStatsDataset"));
-    opBuilder.addInput(inputDataset.asOutput(scope));
-    opBuilder.addInput(tag.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    Class[] outputTypesArray = new Class[outputTypes.size()];
-    for (int i = 0; i < outputTypesArray.length; ++i) {
-      outputTypesArray[i] = outputTypes.get(i);
-    }
-    opBuilder.setAttr("output_types", outputTypesArray);
-    Shape[] outputShapesArray = new Shape[outputShapes.size()];
-    for (int i = 0; i < outputShapesArray.length; ++i) {
-      outputShapesArray[i] = outputShapes.get(i);
-    }
-    opBuilder.setAttr("output_shapes", outputShapesArray);
-    return new LatencyStatsDataset(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<?> handle() {
-    return handle;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) handle;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "LatencyStatsDataset";
-  
-  private Output<?> handle;
-  
-  private LatencyStatsDataset(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    handle = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/LeakyReluGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/LeakyReluGrad.java
deleted file mode 100644
index 46f9410dc26..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/LeakyReluGrad.java
+++ /dev/null
@@ -1,111 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.data;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Computes rectified linear gradients for a LeakyRelu operation.
- * 
- * @param <T> data type for {@code backprops()} output
- */
-public final class LeakyReluGrad<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.data.LeakyReluGrad}
-   */
-  public static class Options {
-    
-    /**
-     * @param alpha 
-     */
-    public Options alpha(Float alpha) {
-      this.alpha = alpha;
-      return this;
-    }
-    
-    private Float alpha;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new LeakyReluGrad operation.
-   * 
-   * @param scope current scope
-   * @param gradients The backpropagated gradients to the corresponding LeakyRelu operation.
-   * @param features The features passed as input to the corresponding LeakyRelu operation,
-   * OR the outputs of that operation (both work equivalently).
-   * @param options carries optional attributes values
-   * @return a new instance of LeakyReluGrad
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> LeakyReluGrad<T> create(Scope scope, Operand<T> gradients, Operand<T> features, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("LeakyReluGrad", scope.makeOpName("LeakyReluGrad"));
-    opBuilder.addInput(gradients.asOutput(scope));
-    opBuilder.addInput(features.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.alpha != null) {
-          opBuilder.setAttr("alpha", opts.alpha);
-        }
-      }
-    }
-    return new LeakyReluGrad<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param alpha 
-   */
-  public static Options alpha(Float alpha) {
-    return new Options().alpha(alpha);
-  }
-  
-  /**
-   * `gradients * (features > 0) + alpha * gradients * (features <= 0)`.
-   */
-  public Output<T> backprops() {
-    return backprops;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return backprops;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "LeakyReluGrad";
-  
-  private Output<T> backprops;
-  
-  private LeakyReluGrad(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    backprops = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MakeIterator.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MakeIterator.java
deleted file mode 100644
index 663d2f9e869..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MakeIterator.java
+++ /dev/null
@@ -1,60 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.data;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-
-/**
- * Makes a new iterator from the given `dataset` and stores it in `iterator`.
- * <p>
- * This operation may be executed multiple times. Each execution will reset the
- * iterator in `iterator` to the first element of `dataset`.
- */
-@Operator(group = "data")
-public final class MakeIterator extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new MakeIterator operation.
-   * 
-   * @param scope current scope
-   * @param dataset 
-   * @param iterator 
-   * @return a new instance of MakeIterator
-   */
-  @Endpoint(describeByClass = true)
-  public static MakeIterator create(Scope scope, Operand<?> dataset, Operand<?> iterator) {
-    OperationBuilder opBuilder = scope.env().opBuilder("MakeIterator", scope.makeOpName("MakeIterator"));
-    opBuilder.addInput(dataset.asOutput(scope));
-    opBuilder.addInput(iterator.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new MakeIterator(opBuilder.build());
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "MakeIterator";
-  
-  private MakeIterator(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MatchingFilesDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MatchingFilesDataset.java
deleted file mode 100644
index 884c14f3ef0..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MatchingFilesDataset.java
+++ /dev/null
@@ -1,72 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.data;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TString;
-import org.tensorflow.types.family.TType;
-
-/**
- */
-public final class MatchingFilesDataset extends RawOp implements Operand<TType> {
-  
-  /**
-   * Factory method to create a class wrapping a new MatchingFilesDataset operation.
-   * 
-   * @param scope current scope
-   * @param patterns 
-   * @return a new instance of MatchingFilesDataset
-   */
-  @Endpoint(describeByClass = true)
-  public static MatchingFilesDataset create(Scope scope, Operand<TString> patterns) {
-    OperationBuilder opBuilder = scope.env().opBuilder("MatchingFilesDataset", scope.makeOpName("MatchingFilesDataset"));
-    opBuilder.addInput(patterns.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new MatchingFilesDataset(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<?> handle() {
-    return handle;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) handle;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "MatchingFilesDataset";
-  
-  private Output<?> handle;
-  
-  private MatchingFilesDataset(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    handle = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MaxIntraOpParallelismDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MaxIntraOpParallelismDataset.java
deleted file mode 100644
index 71ae5c76dde..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MaxIntraOpParallelismDataset.java
+++ /dev/null
@@ -1,89 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.data;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
-
-/**
- * Creates a dataset that overrides the maximum intra-op parallelism.
- */
-public final class MaxIntraOpParallelismDataset extends RawOp implements Operand<TType> {
-  
-  /**
-   * Factory method to create a class wrapping a new MaxIntraOpParallelismDataset operation.
-   * 
-   * @param scope current scope
-   * @param inputDataset 
-   * @param maxIntraOpParallelism Identifies the maximum intra-op parallelism to use.
-   * @param outputTypes 
-   * @param outputShapes 
-   * @return a new instance of MaxIntraOpParallelismDataset
-   */
-  @Endpoint(describeByClass = true)
-  public static MaxIntraOpParallelismDataset create(Scope scope, Operand<?> inputDataset, Operand<TInt64> maxIntraOpParallelism, List<Class<?>> outputTypes, List<Shape> outputShapes) {
-    OperationBuilder opBuilder = scope.env().opBuilder("MaxIntraOpParallelismDataset", scope.makeOpName("MaxIntraOpParallelismDataset"));
-    opBuilder.addInput(inputDataset.asOutput(scope));
-    opBuilder.addInput(maxIntraOpParallelism.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    Class[] outputTypesArray = new Class[outputTypes.size()];
-    for (int i = 0; i < outputTypesArray.length; ++i) {
-      outputTypesArray[i] = outputTypes.get(i);
-    }
-    opBuilder.setAttr("output_types", outputTypesArray);
-    Shape[] outputShapesArray = new Shape[outputShapes.size()];
-    for (int i = 0; i < outputShapesArray.length; ++i) {
-      outputShapesArray[i] = outputShapes.get(i);
-    }
-    opBuilder.setAttr("output_shapes", outputShapesArray);
-    return new MaxIntraOpParallelismDataset(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<?> handle() {
-    return handle;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) handle;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "MaxIntraOpParallelismDataset";
-  
-  private Output<?> handle;
-  
-  private MaxIntraOpParallelismDataset(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    handle = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ModelDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ModelDataset.java
deleted file mode 100644
index 837a9ff4b46..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ModelDataset.java
+++ /dev/null
@@ -1,141 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.data;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Identity transformation that models performance.
- * <p>
- * Identity transformation that models performance.
- */
-public final class ModelDataset extends RawOp implements Operand<TType> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.data.ModelDataset}
-   */
-  public static class Options {
-    
-    /**
-     * @param algorithm 
-     */
-    public Options algorithm(Long algorithm) {
-      this.algorithm = algorithm;
-      return this;
-    }
-    
-    /**
-     * @param cpuBudget 
-     */
-    public Options cpuBudget(Long cpuBudget) {
-      this.cpuBudget = cpuBudget;
-      return this;
-    }
-    
-    private Long algorithm;
-    private Long cpuBudget;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new ModelDataset operation.
-   * 
-   * @param scope current scope
-   * @param inputDataset A variant tensor representing the input dataset.
-   * @param outputTypes 
-   * @param outputShapes 
-   * @param options carries optional attributes values
-   * @return a new instance of ModelDataset
-   */
-  @Endpoint(describeByClass = true)
-  public static ModelDataset create(Scope scope, Operand<?> inputDataset, List<Class<?>> outputTypes, List<Shape> outputShapes, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ModelDataset", scope.makeOpName("ModelDataset"));
-    opBuilder.addInput(inputDataset.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    Class[] outputTypesArray = new Class[outputTypes.size()];
-    for (int i = 0; i < outputTypesArray.length; ++i) {
-      outputTypesArray[i] = outputTypes.get(i);
-    }
-    opBuilder.setAttr("output_types", outputTypesArray);
-    Shape[] outputShapesArray = new Shape[outputShapes.size()];
-    for (int i = 0; i < outputShapesArray.length; ++i) {
-      outputShapesArray[i] = outputShapes.get(i);
-    }
-    opBuilder.setAttr("output_shapes", outputShapesArray);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.algorithm != null) {
-          opBuilder.setAttr("algorithm", opts.algorithm);
-        }
-        if (opts.cpuBudget != null) {
-          opBuilder.setAttr("cpu_budget", opts.cpuBudget);
-        }
-      }
-    }
-    return new ModelDataset(opBuilder.build());
-  }
-  
-  /**
-   * @param algorithm 
-   */
-  public static Options algorithm(Long algorithm) {
-    return new Options().algorithm(algorithm);
-  }
-  
-  /**
-   * @param cpuBudget 
-   */
-  public static Options cpuBudget(Long cpuBudget) {
-    return new Options().cpuBudget(cpuBudget);
-  }
-  
-  /**
-   */
-  public Output<?> handle() {
-    return handle;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) handle;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ModelDataset";
-  
-  private Output<?> handle;
-  
-  private ModelDataset(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    handle = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MultiDeviceIterator.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MultiDeviceIterator.java
deleted file mode 100644
index 5937c97a388..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MultiDeviceIterator.java
+++ /dev/null
@@ -1,97 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.data;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Creates a MultiDeviceIterator resource.
- */
-public final class MultiDeviceIterator extends RawOp implements Operand<TType> {
-  
-  /**
-   * Factory method to create a class wrapping a new MultiDeviceIterator operation.
-   * 
-   * @param scope current scope
-   * @param devices A list of devices the iterator works across.
-   * @param sharedName If non-empty, this resource will be shared under the given name
-   * across multiple sessions.
-   * @param container If non-empty, this resource is placed in the given container.
-   * Otherwise, a default container is used.
-   * @param outputTypes The type list for the return values.
-   * @param outputShapes The list of shapes being produced.
-   * @return a new instance of MultiDeviceIterator
-   */
-  @Endpoint(describeByClass = true)
-  public static MultiDeviceIterator create(Scope scope, List<String> devices, String sharedName, String container, List<Class<?>> outputTypes, List<Shape> outputShapes) {
-    OperationBuilder opBuilder = scope.env().opBuilder("MultiDeviceIterator", scope.makeOpName("MultiDeviceIterator"));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    String[] devicesArray = new String[devices.size()];
-    for (int i = 0; i < devicesArray.length; ++i) {
-      devicesArray[i] = devices.get(i);
-    }
-    opBuilder.setAttr("devices", devicesArray);
-    opBuilder.setAttr("shared_name", sharedName);
-    opBuilder.setAttr("container", container);
-    Class[] outputTypesArray = new Class[outputTypes.size()];
-    for (int i = 0; i < outputTypesArray.length; ++i) {
-      outputTypesArray[i] = outputTypes.get(i);
-    }
-    opBuilder.setAttr("output_types", outputTypesArray);
-    Shape[] outputShapesArray = new Shape[outputShapes.size()];
-    for (int i = 0; i < outputShapesArray.length; ++i) {
-      outputShapesArray[i] = outputShapes.get(i);
-    }
-    opBuilder.setAttr("output_shapes", outputShapesArray);
-    return new MultiDeviceIterator(opBuilder.build());
-  }
-  
-  /**
-   * Handle to the resource created.
-   */
-  public Output<?> handle() {
-    return handle;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) handle;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "MultiDeviceIterator";
-  
-  private Output<?> handle;
-  
-  private MultiDeviceIterator(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    handle = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MultiDeviceIteratorFromStringHandle.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MultiDeviceIteratorFromStringHandle.java
deleted file mode 100644
index 0c1ecab01d1..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MultiDeviceIteratorFromStringHandle.java
+++ /dev/null
@@ -1,120 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.data;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TString;
-import org.tensorflow.types.family.TType;
-
-/**
- * Generates a MultiDeviceIterator resource from its provided string handle.
- */
-public final class MultiDeviceIteratorFromStringHandle extends RawOp implements Operand<TType> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.data.MultiDeviceIteratorFromStringHandle}
-   */
-  public static class Options {
-    
-    /**
-     * @param outputShapes The list of shapes being produced.
-     */
-    public Options outputShapes(List<Shape> outputShapes) {
-      this.outputShapes = outputShapes;
-      return this;
-    }
-    
-    private List<Shape> outputShapes;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new MultiDeviceIteratorFromStringHandle operation.
-   * 
-   * @param scope current scope
-   * @param stringHandle String representing the resource.
-   * @param outputTypes The type list for the return values.
-   * @param options carries optional attributes values
-   * @return a new instance of MultiDeviceIteratorFromStringHandle
-   */
-  @Endpoint(describeByClass = true)
-  public static MultiDeviceIteratorFromStringHandle create(Scope scope, Operand<TString> stringHandle, List<Class<?>> outputTypes, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("MultiDeviceIteratorFromStringHandle", scope.makeOpName("MultiDeviceIteratorFromStringHandle"));
-    opBuilder.addInput(stringHandle.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    Class[] outputTypesArray = new Class[outputTypes.size()];
-    for (int i = 0; i < outputTypesArray.length; ++i) {
-      outputTypesArray[i] = outputTypes.get(i);
-    }
-    opBuilder.setAttr("output_types", outputTypesArray);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.outputShapes != null) {
-          Shape[] outputShapesArray = new Shape[opts.outputShapes.size()];
-          for (int i = 0; i < outputShapesArray.length; ++i) {
-            outputShapesArray[i] = opts.outputShapes.get(i);
-          }
-          opBuilder.setAttr("output_shapes", outputShapesArray);
-        }
-      }
-    }
-    return new MultiDeviceIteratorFromStringHandle(opBuilder.build());
-  }
-  
-  /**
-   * @param outputShapes The list of shapes being produced.
-   */
-  public static Options outputShapes(List<Shape> outputShapes) {
-    return new Options().outputShapes(outputShapes);
-  }
-  
-  /**
-   * A MultiDeviceIterator resource.
-   */
-  public Output<?> multiDeviceIterator() {
-    return multiDeviceIterator;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) multiDeviceIterator;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "MultiDeviceIteratorFromStringHandle";
-  
-  private Output<?> multiDeviceIterator;
-  
-  private MultiDeviceIteratorFromStringHandle(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    multiDeviceIterator = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MultiDeviceIteratorGetNextFromShard.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MultiDeviceIteratorGetNextFromShard.java
deleted file mode 100644
index 2156bab1b17..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MultiDeviceIteratorGetNextFromShard.java
+++ /dev/null
@@ -1,97 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.data;
-
-import java.util.Arrays;
-import java.util.Iterator;
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
-
-/**
- * Gets next element for the provided shard number.
- */
-public final class MultiDeviceIteratorGetNextFromShard extends RawOp implements Iterable<Operand<TType>> {
-  
-  /**
-   * Factory method to create a class wrapping a new MultiDeviceIteratorGetNextFromShard operation.
-   * 
-   * @param scope current scope
-   * @param multiDeviceIterator A MultiDeviceIterator resource.
-   * @param shardNum Integer representing which shard to fetch data for.
-   * @param incarnationId Which incarnation of the MultiDeviceIterator is running.
-   * @param outputTypes The type list for the return values.
-   * @param outputShapes The list of shapes being produced.
-   * @return a new instance of MultiDeviceIteratorGetNextFromShard
-   */
-  @Endpoint(describeByClass = true)
-  public static MultiDeviceIteratorGetNextFromShard create(Scope scope, Operand<?> multiDeviceIterator, Operand<TInt32> shardNum, Operand<TInt64> incarnationId, List<Class<?>> outputTypes, List<Shape> outputShapes) {
-    OperationBuilder opBuilder = scope.env().opBuilder("MultiDeviceIteratorGetNextFromShard", scope.makeOpName("MultiDeviceIteratorGetNextFromShard"));
-    opBuilder.addInput(multiDeviceIterator.asOutput(scope));
-    opBuilder.addInput(shardNum.asOutput(scope));
-    opBuilder.addInput(incarnationId.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    Class[] outputTypesArray = new Class[outputTypes.size()];
-    for (int i = 0; i < outputTypesArray.length; ++i) {
-      outputTypesArray[i] = outputTypes.get(i);
-    }
-    opBuilder.setAttr("output_types", outputTypesArray);
-    Shape[] outputShapesArray = new Shape[outputShapes.size()];
-    for (int i = 0; i < outputShapesArray.length; ++i) {
-      outputShapesArray[i] = outputShapes.get(i);
-    }
-    opBuilder.setAttr("output_shapes", outputShapesArray);
-    return new MultiDeviceIteratorGetNextFromShard(opBuilder.build());
-  }
-  
-  /**
-   * Result of the get_next on the dataset.
-   */
-  public List<Output<?>> components() {
-    return components;
-  }
-  
-  @Override
-  @SuppressWarnings({"rawtypes", "unchecked"})
-  public Iterator<Operand<TType>> iterator() {
-    return (Iterator) components.iterator();
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "MultiDeviceIteratorGetNextFromShard";
-  
-  private List<Output<?>> components;
-  
-  private MultiDeviceIteratorGetNextFromShard(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    int componentsLength = operation.outputListLength("components");
-    components = Arrays.asList(operation.outputList(outputIdx, componentsLength));
-    outputIdx += componentsLength;
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MultiDeviceIteratorInit.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MultiDeviceIteratorInit.java
deleted file mode 100644
index 0b1a70d6a8f..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MultiDeviceIteratorInit.java
+++ /dev/null
@@ -1,77 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.data;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt64;
-
-/**
- * Initializes the multi device iterator with the given dataset.
- */
-public final class MultiDeviceIteratorInit extends RawOp implements Operand<TInt64> {
-  
-  /**
-   * Factory method to create a class wrapping a new MultiDeviceIteratorInit operation.
-   * 
-   * @param scope current scope
-   * @param dataset Dataset to be iterated upon.
-   * @param multiDeviceIterator A MultiDeviceIteratorResource.
-   * @param maxBufferSize The maximum size of the host side per device buffer to keep.
-   * @return a new instance of MultiDeviceIteratorInit
-   */
-  @Endpoint(describeByClass = true)
-  public static MultiDeviceIteratorInit create(Scope scope, Operand<?> dataset, Operand<?> multiDeviceIterator, Operand<TInt64> maxBufferSize) {
-    OperationBuilder opBuilder = scope.env().opBuilder("MultiDeviceIteratorInit", scope.makeOpName("MultiDeviceIteratorInit"));
-    opBuilder.addInput(dataset.asOutput(scope));
-    opBuilder.addInput(multiDeviceIterator.asOutput(scope));
-    opBuilder.addInput(maxBufferSize.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new MultiDeviceIteratorInit(opBuilder.build());
-  }
-  
-  /**
-   * An int64 indicating which incarnation of the MultiDeviceIterator
-   * is running.
-   */
-  public Output<TInt64> incarnationId() {
-    return incarnationId;
-  }
-  
-  @Override
-  public Output<TInt64> asOutput(Scope scope) {
-    return incarnationId;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "MultiDeviceIteratorInit";
-  
-  private Output<TInt64> incarnationId;
-  
-  private MultiDeviceIteratorInit(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    incarnationId = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MultiDeviceIteratorToStringHandle.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MultiDeviceIteratorToStringHandle.java
deleted file mode 100644
index 3efeb3052e9..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MultiDeviceIteratorToStringHandle.java
+++ /dev/null
@@ -1,72 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.data;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TString;
-
-/**
- * Produces a string handle for the given MultiDeviceIterator.
- */
-public final class MultiDeviceIteratorToStringHandle extends RawOp implements Operand<TString> {
-  
-  /**
-   * Factory method to create a class wrapping a new MultiDeviceIteratorToStringHandle operation.
-   * 
-   * @param scope current scope
-   * @param multiDeviceIterator A MultiDeviceIterator resource.
-   * @return a new instance of MultiDeviceIteratorToStringHandle
-   */
-  @Endpoint(describeByClass = true)
-  public static MultiDeviceIteratorToStringHandle create(Scope scope, Operand<?> multiDeviceIterator) {
-    OperationBuilder opBuilder = scope.env().opBuilder("MultiDeviceIteratorToStringHandle", scope.makeOpName("MultiDeviceIteratorToStringHandle"));
-    opBuilder.addInput(multiDeviceIterator.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new MultiDeviceIteratorToStringHandle(opBuilder.build());
-  }
-  
-  /**
-   * A string representing the resource.
-   */
-  public Output<TString> stringHandle() {
-    return stringHandle;
-  }
-  
-  @Override
-  public Output<TString> asOutput(Scope scope) {
-    return stringHandle;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "MultiDeviceIteratorToStringHandle";
-  
-  private Output<TString> stringHandle;
-  
-  private MultiDeviceIteratorToStringHandle(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    stringHandle = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/NonSerializableDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/NonSerializableDataset.java
deleted file mode 100644
index f1be418af20..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/NonSerializableDataset.java
+++ /dev/null
@@ -1,85 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.data;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- */
-public final class NonSerializableDataset extends RawOp implements Operand<TType> {
-  
-  /**
-   * Factory method to create a class wrapping a new NonSerializableDataset operation.
-   * 
-   * @param scope current scope
-   * @param inputDataset 
-   * @param outputTypes 
-   * @param outputShapes 
-   * @return a new instance of NonSerializableDataset
-   */
-  @Endpoint(describeByClass = true)
-  public static NonSerializableDataset create(Scope scope, Operand<?> inputDataset, List<Class<?>> outputTypes, List<Shape> outputShapes) {
-    OperationBuilder opBuilder = scope.env().opBuilder("NonSerializableDataset", scope.makeOpName("NonSerializableDataset"));
-    opBuilder.addInput(inputDataset.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    Class[] outputTypesArray = new Class[outputTypes.size()];
-    for (int i = 0; i < outputTypesArray.length; ++i) {
-      outputTypesArray[i] = outputTypes.get(i);
-    }
-    opBuilder.setAttr("output_types", outputTypesArray);
-    Shape[] outputShapesArray = new Shape[outputShapes.size()];
-    for (int i = 0; i < outputShapesArray.length; ++i) {
-      outputShapesArray[i] = outputShapes.get(i);
-    }
-    opBuilder.setAttr("output_shapes", outputShapesArray);
-    return new NonSerializableDataset(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<?> handle() {
-    return handle;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) handle;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "NonSerializableDataset";
-  
-  private Output<?> handle;
-  
-  private NonSerializableDataset(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    handle = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/OptimizeDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/OptimizeDataset.java
deleted file mode 100644
index 41d6bf9d6c1..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/OptimizeDataset.java
+++ /dev/null
@@ -1,129 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.data;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TString;
-import org.tensorflow.types.family.TType;
-
-/**
- * Creates a dataset by applying optimizations to `input_dataset`.
- * <p>
- * Creates a dataset by applying optimizations to `input_dataset`.
- */
-public final class OptimizeDataset extends RawOp implements Operand<TType> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.data.OptimizeDataset}
-   */
-  public static class Options {
-    
-    /**
-     * @param optimizationConfigs 
-     */
-    public Options optimizationConfigs(List<String> optimizationConfigs) {
-      this.optimizationConfigs = optimizationConfigs;
-      return this;
-    }
-    
-    private List<String> optimizationConfigs;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new OptimizeDataset operation.
-   * 
-   * @param scope current scope
-   * @param inputDataset A variant tensor representing the input dataset.
-   * @param optimizations A `tf.string` vector `tf.Tensor` identifying optimizations to use.
-   * @param outputTypes 
-   * @param outputShapes 
-   * @param options carries optional attributes values
-   * @return a new instance of OptimizeDataset
-   */
-  @Endpoint(describeByClass = true)
-  public static OptimizeDataset create(Scope scope, Operand<?> inputDataset, Operand<TString> optimizations, List<Class<?>> outputTypes, List<Shape> outputShapes, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("OptimizeDataset", scope.makeOpName("OptimizeDataset"));
-    opBuilder.addInput(inputDataset.asOutput(scope));
-    opBuilder.addInput(optimizations.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    Class[] outputTypesArray = new Class[outputTypes.size()];
-    for (int i = 0; i < outputTypesArray.length; ++i) {
-      outputTypesArray[i] = outputTypes.get(i);
-    }
-    opBuilder.setAttr("output_types", outputTypesArray);
-    Shape[] outputShapesArray = new Shape[outputShapes.size()];
-    for (int i = 0; i < outputShapesArray.length; ++i) {
-      outputShapesArray[i] = outputShapes.get(i);
-    }
-    opBuilder.setAttr("output_shapes", outputShapesArray);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.optimizationConfigs != null) {
-          String[] optimizationConfigsArray = new String[opts.optimizationConfigs.size()];
-          for (int i = 0; i < optimizationConfigsArray.length; ++i) {
-            optimizationConfigsArray[i] = opts.optimizationConfigs.get(i);
-          }
-          opBuilder.setAttr("optimization_configs", optimizationConfigsArray);
-        }
-      }
-    }
-    return new OptimizeDataset(opBuilder.build());
-  }
-  
-  /**
-   * @param optimizationConfigs 
-   */
-  public static Options optimizationConfigs(List<String> optimizationConfigs) {
-    return new Options().optimizationConfigs(optimizationConfigs);
-  }
-  
-  /**
-   */
-  public Output<?> handle() {
-    return handle;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) handle;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "OptimizeDataset";
-  
-  private Output<?> handle;
-  
-  private OptimizeDataset(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    handle = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/OptionalFromValue.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/OptionalFromValue.java
deleted file mode 100644
index df33632dc0b..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/OptionalFromValue.java
+++ /dev/null
@@ -1,74 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.data;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.Operands;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Constructs an Optional variant from a tuple of tensors.
- */
-@Operator(group = "data")
-public final class OptionalFromValue extends RawOp implements Operand<TType> {
-  
-  /**
-   * Factory method to create a class wrapping a new OptionalFromValue operation.
-   * 
-   * @param scope current scope
-   * @param components 
-   * @return a new instance of OptionalFromValue
-   */
-  @Endpoint(describeByClass = true)
-  public static OptionalFromValue create(Scope scope, Iterable<Operand<?>> components) {
-    OperationBuilder opBuilder = scope.env().opBuilder("OptionalFromValue", scope.makeOpName("OptionalFromValue"));
-    opBuilder.addInputList(Operands.asOutputs(scope, components));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new OptionalFromValue(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<?> optional() {
-    return optional;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) optional;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "OptionalFromValue";
-  
-  private Output<?> optional;
-  
-  private OptionalFromValue(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    optional = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/OptionalGetValue.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/OptionalGetValue.java
deleted file mode 100644
index f3499e4a099..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/OptionalGetValue.java
+++ /dev/null
@@ -1,91 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.data;
-
-import java.util.Arrays;
-import java.util.Iterator;
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Returns the value stored in an Optional variant or raises an error if none exists.
- */
-@Operator(group = "data")
-public final class OptionalGetValue extends RawOp implements Iterable<Operand<TType>> {
-  
-  /**
-   * Factory method to create a class wrapping a new OptionalGetValue operation.
-   * 
-   * @param scope current scope
-   * @param optional 
-   * @param outputTypes 
-   * @param outputShapes 
-   * @return a new instance of OptionalGetValue
-   */
-  @Endpoint(describeByClass = true)
-  public static OptionalGetValue create(Scope scope, Operand<?> optional, List<Class<?>> outputTypes, List<Shape> outputShapes) {
-    OperationBuilder opBuilder = scope.env().opBuilder("OptionalGetValue", scope.makeOpName("OptionalGetValue"));
-    opBuilder.addInput(optional.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    Class[] outputTypesArray = new Class[outputTypes.size()];
-    for (int i = 0; i < outputTypesArray.length; ++i) {
-      outputTypesArray[i] = outputTypes.get(i);
-    }
-    opBuilder.setAttr("output_types", outputTypesArray);
-    Shape[] outputShapesArray = new Shape[outputShapes.size()];
-    for (int i = 0; i < outputShapesArray.length; ++i) {
-      outputShapesArray[i] = outputShapes.get(i);
-    }
-    opBuilder.setAttr("output_shapes", outputShapesArray);
-    return new OptionalGetValue(opBuilder.build());
-  }
-  
-  /**
-   */
-  public List<Output<?>> components() {
-    return components;
-  }
-  
-  @Override
-  @SuppressWarnings({"rawtypes", "unchecked"})
-  public Iterator<Operand<TType>> iterator() {
-    return (Iterator) components.iterator();
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "OptionalGetValue";
-  
-  private List<Output<?>> components;
-  
-  private OptionalGetValue(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    int componentsLength = operation.outputListLength("components");
-    components = Arrays.asList(operation.outputList(outputIdx, componentsLength));
-    outputIdx += componentsLength;
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/OptionalHasValue.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/OptionalHasValue.java
deleted file mode 100644
index d0fb8dabaef..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/OptionalHasValue.java
+++ /dev/null
@@ -1,72 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.data;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TBool;
-
-/**
- * Returns true if and only if the given Optional variant has a value.
- */
-@Operator(group = "data")
-public final class OptionalHasValue extends RawOp implements Operand<TBool> {
-  
-  /**
-   * Factory method to create a class wrapping a new OptionalHasValue operation.
-   * 
-   * @param scope current scope
-   * @param optional 
-   * @return a new instance of OptionalHasValue
-   */
-  @Endpoint(describeByClass = true)
-  public static OptionalHasValue create(Scope scope, Operand<?> optional) {
-    OperationBuilder opBuilder = scope.env().opBuilder("OptionalHasValue", scope.makeOpName("OptionalHasValue"));
-    opBuilder.addInput(optional.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new OptionalHasValue(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<TBool> hasValue() {
-    return hasValue;
-  }
-  
-  @Override
-  public Output<TBool> asOutput(Scope scope) {
-    return hasValue;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "OptionalHasValue";
-  
-  private Output<TBool> hasValue;
-  
-  private OptionalHasValue(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    hasValue = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/OptionalNone.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/OptionalNone.java
deleted file mode 100644
index be0d557c02e..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/OptionalNone.java
+++ /dev/null
@@ -1,71 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.data;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Creates an Optional variant with no value.
- */
-@Operator(group = "data")
-public final class OptionalNone extends RawOp implements Operand<TType> {
-  
-  /**
-   * Factory method to create a class wrapping a new OptionalNone operation.
-   * 
-   * @param scope current scope
-   * @return a new instance of OptionalNone
-   */
-  @Endpoint(describeByClass = true)
-  public static OptionalNone create(Scope scope) {
-    OperationBuilder opBuilder = scope.env().opBuilder("OptionalNone", scope.makeOpName("OptionalNone"));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new OptionalNone(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<?> optional() {
-    return optional;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) optional;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "OptionalNone";
-  
-  private Output<?> optional;
-  
-  private OptionalNone(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    optional = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/PaddedBatchDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/PaddedBatchDataset.java
deleted file mode 100644
index 97e3f9b5808..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/PaddedBatchDataset.java
+++ /dev/null
@@ -1,131 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.data;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.op.Operands;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TBool;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
-
-/**
- * Creates a dataset that batches and pads `batch_size` elements from the input.
- */
-public final class PaddedBatchDataset extends RawOp implements Operand<TType> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.data.PaddedBatchDataset}
-   */
-  public static class Options {
-    
-    /**
-     * @param parallelCopy 
-     */
-    public Options parallelCopy(Boolean parallelCopy) {
-      this.parallelCopy = parallelCopy;
-      return this;
-    }
-    
-    private Boolean parallelCopy;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new PaddedBatchDataset operation.
-   * 
-   * @param scope current scope
-   * @param inputDataset 
-   * @param batchSize A scalar representing the number of elements to accumulate in a
-   * batch.
-   * @param paddedShapes A list of int64 tensors representing the desired padded shapes
-   * of the corresponding output components. These shapes may be partially
-   * specified, using `-1` to indicate that a particular dimension should be
-   * padded to the maximum size of all batch elements.
-   * @param paddingValues A list of scalars containing the padding value to use for
-   * each of the outputs.
-   * @param dropRemainder A scalar representing whether the last batch should be dropped in case its size
-   * is smaller than desired.
-   * @param outputShapes 
-   * @param options carries optional attributes values
-   * @return a new instance of PaddedBatchDataset
-   */
-  @Endpoint(describeByClass = true)
-  public static PaddedBatchDataset create(Scope scope, Operand<?> inputDataset, Operand<TInt64> batchSize, Iterable<Operand<TInt64>> paddedShapes, Iterable<Operand<?>> paddingValues, Operand<TBool> dropRemainder, List<Shape> outputShapes, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("PaddedBatchDatasetV2", scope.makeOpName("PaddedBatchDataset"));
-    opBuilder.addInput(inputDataset.asOutput(scope));
-    opBuilder.addInput(batchSize.asOutput(scope));
-    opBuilder.addInputList(Operands.asOutputs(scope, paddedShapes));
-    opBuilder.addInputList(Operands.asOutputs(scope, paddingValues));
-    opBuilder.addInput(dropRemainder.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    Shape[] outputShapesArray = new Shape[outputShapes.size()];
-    for (int i = 0; i < outputShapesArray.length; ++i) {
-      outputShapesArray[i] = outputShapes.get(i);
-    }
-    opBuilder.setAttr("output_shapes", outputShapesArray);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.parallelCopy != null) {
-          opBuilder.setAttr("parallel_copy", opts.parallelCopy);
-        }
-      }
-    }
-    return new PaddedBatchDataset(opBuilder.build());
-  }
-  
-  /**
-   * @param parallelCopy 
-   */
-  public static Options parallelCopy(Boolean parallelCopy) {
-    return new Options().parallelCopy(parallelCopy);
-  }
-  
-  /**
-   */
-  public Output<?> handle() {
-    return handle;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) handle;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "PaddedBatchDatasetV2";
-  
-  private Output<?> handle;
-  
-  private PaddedBatchDataset(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    handle = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/PrefetchDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/PrefetchDataset.java
deleted file mode 100644
index c8d61ba16f3..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/PrefetchDataset.java
+++ /dev/null
@@ -1,143 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.data;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
-
-/**
- * Creates a dataset that asynchronously prefetches elements from `input_dataset`.
- */
-public final class PrefetchDataset extends RawOp implements Operand<TType> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.data.PrefetchDataset}
-   */
-  public static class Options {
-    
-    /**
-     * @param slackPeriod 
-     */
-    public Options slackPeriod(Long slackPeriod) {
-      this.slackPeriod = slackPeriod;
-      return this;
-    }
-    
-    /**
-     * @param legacyAutotune 
-     */
-    public Options legacyAutotune(Boolean legacyAutotune) {
-      this.legacyAutotune = legacyAutotune;
-      return this;
-    }
-    
-    private Long slackPeriod;
-    private Boolean legacyAutotune;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new PrefetchDataset operation.
-   * 
-   * @param scope current scope
-   * @param inputDataset 
-   * @param bufferSize The maximum number of elements to buffer in an iterator over
-   * this dataset.
-   * @param outputTypes 
-   * @param outputShapes 
-   * @param options carries optional attributes values
-   * @return a new instance of PrefetchDataset
-   */
-  @Endpoint(describeByClass = true)
-  public static PrefetchDataset create(Scope scope, Operand<?> inputDataset, Operand<TInt64> bufferSize, List<Class<?>> outputTypes, List<Shape> outputShapes, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("PrefetchDataset", scope.makeOpName("PrefetchDataset"));
-    opBuilder.addInput(inputDataset.asOutput(scope));
-    opBuilder.addInput(bufferSize.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    Class[] outputTypesArray = new Class[outputTypes.size()];
-    for (int i = 0; i < outputTypesArray.length; ++i) {
-      outputTypesArray[i] = outputTypes.get(i);
-    }
-    opBuilder.setAttr("output_types", outputTypesArray);
-    Shape[] outputShapesArray = new Shape[outputShapes.size()];
-    for (int i = 0; i < outputShapesArray.length; ++i) {
-      outputShapesArray[i] = outputShapes.get(i);
-    }
-    opBuilder.setAttr("output_shapes", outputShapesArray);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.slackPeriod != null) {
-          opBuilder.setAttr("slack_period", opts.slackPeriod);
-        }
-        if (opts.legacyAutotune != null) {
-          opBuilder.setAttr("legacy_autotune", opts.legacyAutotune);
-        }
-      }
-    }
-    return new PrefetchDataset(opBuilder.build());
-  }
-  
-  /**
-   * @param slackPeriod 
-   */
-  public static Options slackPeriod(Long slackPeriod) {
-    return new Options().slackPeriod(slackPeriod);
-  }
-  
-  /**
-   * @param legacyAutotune 
-   */
-  public static Options legacyAutotune(Boolean legacyAutotune) {
-    return new Options().legacyAutotune(legacyAutotune);
-  }
-  
-  /**
-   */
-  public Output<?> handle() {
-    return handle;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) handle;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "PrefetchDataset";
-  
-  private Output<?> handle;
-  
-  private PrefetchDataset(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    handle = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/PrivateThreadPoolDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/PrivateThreadPoolDataset.java
deleted file mode 100644
index 22d669b94b5..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/PrivateThreadPoolDataset.java
+++ /dev/null
@@ -1,89 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.data;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
-
-/**
- * Creates a dataset that uses a custom thread pool to compute `input_dataset`.
- */
-public final class PrivateThreadPoolDataset extends RawOp implements Operand<TType> {
-  
-  /**
-   * Factory method to create a class wrapping a new PrivateThreadPoolDataset operation.
-   * 
-   * @param scope current scope
-   * @param inputDataset 
-   * @param numThreads Identifies the number of threads to use for the private threadpool.
-   * @param outputTypes 
-   * @param outputShapes 
-   * @return a new instance of PrivateThreadPoolDataset
-   */
-  @Endpoint(describeByClass = true)
-  public static PrivateThreadPoolDataset create(Scope scope, Operand<?> inputDataset, Operand<TInt64> numThreads, List<Class<?>> outputTypes, List<Shape> outputShapes) {
-    OperationBuilder opBuilder = scope.env().opBuilder("PrivateThreadPoolDataset", scope.makeOpName("PrivateThreadPoolDataset"));
-    opBuilder.addInput(inputDataset.asOutput(scope));
-    opBuilder.addInput(numThreads.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    Class[] outputTypesArray = new Class[outputTypes.size()];
-    for (int i = 0; i < outputTypesArray.length; ++i) {
-      outputTypesArray[i] = outputTypes.get(i);
-    }
-    opBuilder.setAttr("output_types", outputTypesArray);
-    Shape[] outputShapesArray = new Shape[outputShapes.size()];
-    for (int i = 0; i < outputShapesArray.length; ++i) {
-      outputShapesArray[i] = outputShapes.get(i);
-    }
-    opBuilder.setAttr("output_shapes", outputShapesArray);
-    return new PrivateThreadPoolDataset(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<?> handle() {
-    return handle;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) handle;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "PrivateThreadPoolDataset";
-  
-  private Output<?> handle;
-  
-  private PrivateThreadPoolDataset(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    handle = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/RandomDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/RandomDataset.java
deleted file mode 100644
index 0de67cb18f0..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/RandomDataset.java
+++ /dev/null
@@ -1,102 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.data;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
-
-/**
- * Creates a Dataset that returns pseudorandom numbers.
- * <p>
- * Creates a Dataset that returns a stream of uniformly distributed
- * pseudorandom 64-bit signed integers.
- * <p>
- * In the TensorFlow Python API, you can instantiate this dataset via the
- * class `tf.data.experimental.RandomDataset`.
- * <p>
- * Instances of this dataset are also created as a result of the
- * `hoist_random_uniform` static optimization. Whether this optimization is
- * performed is determined by the `experimental_optimization.hoist_random_uniform`
- * option of `tf.data.Options`.
- */
-public final class RandomDataset extends RawOp implements Operand<TType> {
-  
-  /**
-   * Factory method to create a class wrapping a new RandomDataset operation.
-   * 
-   * @param scope current scope
-   * @param seed A scalar seed for the random number generator. If either seed or
-   * seed2 is set to be non-zero, the random number generator is seeded
-   * by the given seed.  Otherwise, a random seed is used.
-   * @param seed2 A second scalar seed to avoid seed collision.
-   * @param outputTypes 
-   * @param outputShapes 
-   * @return a new instance of RandomDataset
-   */
-  @Endpoint(describeByClass = true)
-  public static RandomDataset create(Scope scope, Operand<TInt64> seed, Operand<TInt64> seed2, List<Class<?>> outputTypes, List<Shape> outputShapes) {
-    OperationBuilder opBuilder = scope.env().opBuilder("RandomDataset", scope.makeOpName("RandomDataset"));
-    opBuilder.addInput(seed.asOutput(scope));
-    opBuilder.addInput(seed2.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    Class[] outputTypesArray = new Class[outputTypes.size()];
-    for (int i = 0; i < outputTypesArray.length; ++i) {
-      outputTypesArray[i] = outputTypes.get(i);
-    }
-    opBuilder.setAttr("output_types", outputTypesArray);
-    Shape[] outputShapesArray = new Shape[outputShapes.size()];
-    for (int i = 0; i < outputShapesArray.length; ++i) {
-      outputShapesArray[i] = outputShapes.get(i);
-    }
-    opBuilder.setAttr("output_shapes", outputShapesArray);
-    return new RandomDataset(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<?> handle() {
-    return handle;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) handle;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "RandomDataset";
-  
-  private Output<?> handle;
-  
-  private RandomDataset(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    handle = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/RangeDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/RangeDataset.java
deleted file mode 100644
index 6133f124725..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/RangeDataset.java
+++ /dev/null
@@ -1,92 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.data;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
-
-/**
- * Creates a dataset with a range of values. Corresponds to python's xrange.
- */
-@Operator(group = "data")
-public final class RangeDataset extends RawOp implements Operand<TType> {
-  
-  /**
-   * Factory method to create a class wrapping a new RangeDataset operation.
-   * 
-   * @param scope current scope
-   * @param start corresponds to start in python's xrange().
-   * @param stop corresponds to stop in python's xrange().
-   * @param step corresponds to step in python's xrange().
-   * @param outputTypes 
-   * @param outputShapes 
-   * @return a new instance of RangeDataset
-   */
-  @Endpoint(describeByClass = true)
-  public static RangeDataset create(Scope scope, Operand<TInt64> start, Operand<TInt64> stop, Operand<TInt64> step, List<Class<?>> outputTypes, List<Shape> outputShapes) {
-    OperationBuilder opBuilder = scope.env().opBuilder("RangeDataset", scope.makeOpName("RangeDataset"));
-    opBuilder.addInput(start.asOutput(scope));
-    opBuilder.addInput(stop.asOutput(scope));
-    opBuilder.addInput(step.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    Class[] outputTypesArray = new Class[outputTypes.size()];
-    for (int i = 0; i < outputTypesArray.length; ++i) {
-      outputTypesArray[i] = outputTypes.get(i);
-    }
-    opBuilder.setAttr("output_types", outputTypesArray);
-    Shape[] outputShapesArray = new Shape[outputShapes.size()];
-    for (int i = 0; i < outputShapesArray.length; ++i) {
-      outputShapesArray[i] = outputShapes.get(i);
-    }
-    opBuilder.setAttr("output_shapes", outputShapesArray);
-    return new RangeDataset(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<?> handle() {
-    return handle;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) handle;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "RangeDataset";
-  
-  private Output<?> handle;
-  
-  private RangeDataset(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    handle = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/RebatchDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/RebatchDataset.java
deleted file mode 100644
index 5340276fb24..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/RebatchDataset.java
+++ /dev/null
@@ -1,128 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.data;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
-
-/**
- * Creates a dataset that changes the batch size.
- * <p>
- * Creates a dataset that changes the batch size of the dataset to current batch
- * size // num_workers.
- */
-public final class RebatchDataset extends RawOp implements Operand<TType> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.data.RebatchDataset}
-   */
-  public static class Options {
-    
-    /**
-     * @param useFallback 
-     */
-    public Options useFallback(Boolean useFallback) {
-      this.useFallback = useFallback;
-      return this;
-    }
-    
-    private Boolean useFallback;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new RebatchDataset operation.
-   * 
-   * @param scope current scope
-   * @param inputDataset A variant tensor representing the input dataset.
-   * @param numReplicas A scalar representing the number of replicas to distribute this batch across. As
-   * a result of this transformation the current batch size would end up being
-   * divided  by this parameter.
-   * @param outputTypes 
-   * @param outputShapes 
-   * @param options carries optional attributes values
-   * @return a new instance of RebatchDataset
-   */
-  @Endpoint(describeByClass = true)
-  public static RebatchDataset create(Scope scope, Operand<?> inputDataset, Operand<TInt64> numReplicas, List<Class<?>> outputTypes, List<Shape> outputShapes, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("RebatchDataset", scope.makeOpName("RebatchDataset"));
-    opBuilder.addInput(inputDataset.asOutput(scope));
-    opBuilder.addInput(numReplicas.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    Class[] outputTypesArray = new Class[outputTypes.size()];
-    for (int i = 0; i < outputTypesArray.length; ++i) {
-      outputTypesArray[i] = outputTypes.get(i);
-    }
-    opBuilder.setAttr("output_types", outputTypesArray);
-    Shape[] outputShapesArray = new Shape[outputShapes.size()];
-    for (int i = 0; i < outputShapesArray.length; ++i) {
-      outputShapesArray[i] = outputShapes.get(i);
-    }
-    opBuilder.setAttr("output_shapes", outputShapesArray);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.useFallback != null) {
-          opBuilder.setAttr("use_fallback", opts.useFallback);
-        }
-      }
-    }
-    return new RebatchDataset(opBuilder.build());
-  }
-  
-  /**
-   * @param useFallback 
-   */
-  public static Options useFallback(Boolean useFallback) {
-    return new Options().useFallback(useFallback);
-  }
-  
-  /**
-   */
-  public Output<?> handle() {
-    return handle;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) handle;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "RebatchDataset";
-  
-  private Output<?> handle;
-  
-  private RebatchDataset(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    handle = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/RegisterDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/RegisterDataset.java
deleted file mode 100644
index fe07a1ef30c..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/RegisterDataset.java
+++ /dev/null
@@ -1,78 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.data;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.TString;
-
-/**
- * Registers a dataset with the tf.data service.
- */
-public final class RegisterDataset extends RawOp implements Operand<TInt64> {
-  
-  /**
-   * Factory method to create a class wrapping a new RegisterDataset operation.
-   * 
-   * @param scope current scope
-   * @param dataset 
-   * @param address 
-   * @param protocol 
-   * @param externalStatePolicy 
-   * @return a new instance of RegisterDataset
-   */
-  @Endpoint(describeByClass = true)
-  public static RegisterDataset create(Scope scope, Operand<?> dataset, Operand<TString> address, Operand<TString> protocol, Long externalStatePolicy) {
-    OperationBuilder opBuilder = scope.env().opBuilder("RegisterDataset", scope.makeOpName("RegisterDataset"));
-    opBuilder.addInput(dataset.asOutput(scope));
-    opBuilder.addInput(address.asOutput(scope));
-    opBuilder.addInput(protocol.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("external_state_policy", externalStatePolicy);
-    return new RegisterDataset(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<TInt64> datasetId() {
-    return datasetId;
-  }
-  
-  @Override
-  public Output<TInt64> asOutput(Scope scope) {
-    return datasetId;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "RegisterDataset";
-  
-  private Output<TInt64> datasetId;
-  
-  private RegisterDataset(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    datasetId = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/RepeatDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/RepeatDataset.java
deleted file mode 100644
index 4d5fffaf5fa..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/RepeatDataset.java
+++ /dev/null
@@ -1,91 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.data;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
-
-/**
- * Creates a dataset that emits the outputs of `input_dataset` `count` times.
- */
-@Operator(group = "data")
-public final class RepeatDataset extends RawOp implements Operand<TType> {
-  
-  /**
-   * Factory method to create a class wrapping a new RepeatDataset operation.
-   * 
-   * @param scope current scope
-   * @param inputDataset 
-   * @param count A scalar representing the number of times that `input_dataset` should
-   * be repeated. A value of `-1` indicates that it should be repeated infinitely.
-   * @param outputTypes 
-   * @param outputShapes 
-   * @return a new instance of RepeatDataset
-   */
-  @Endpoint(describeByClass = true)
-  public static RepeatDataset create(Scope scope, Operand<?> inputDataset, Operand<TInt64> count, List<Class<?>> outputTypes, List<Shape> outputShapes) {
-    OperationBuilder opBuilder = scope.env().opBuilder("RepeatDataset", scope.makeOpName("RepeatDataset"));
-    opBuilder.addInput(inputDataset.asOutput(scope));
-    opBuilder.addInput(count.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    Class[] outputTypesArray = new Class[outputTypes.size()];
-    for (int i = 0; i < outputTypesArray.length; ++i) {
-      outputTypesArray[i] = outputTypes.get(i);
-    }
-    opBuilder.setAttr("output_types", outputTypesArray);
-    Shape[] outputShapesArray = new Shape[outputShapes.size()];
-    for (int i = 0; i < outputShapesArray.length; ++i) {
-      outputShapesArray[i] = outputShapes.get(i);
-    }
-    opBuilder.setAttr("output_shapes", outputShapesArray);
-    return new RepeatDataset(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<?> handle() {
-    return handle;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) handle;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "RepeatDataset";
-  
-  private Output<?> handle;
-  
-  private RepeatDataset(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    handle = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SamplingDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SamplingDataset.java
deleted file mode 100644
index 649645e6dd3..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SamplingDataset.java
+++ /dev/null
@@ -1,101 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.data;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
-
-/**
- * Creates a dataset that takes a Bernoulli sample of the contents of another dataset.
- * <p>
- * There is no transformation in the `tf.data` Python API for creating this dataset.
- * Instead, it is created as a result of the `filter_with_random_uniform_fusion`
- * static optimization. Whether this optimization is performed is determined by the
- * `experimental_optimization.filter_with_random_uniform_fusion` option of
- * `tf.data.Options`.
- */
-public final class SamplingDataset extends RawOp implements Operand<TType> {
-  
-  /**
-   * Factory method to create a class wrapping a new SamplingDataset operation.
-   * 
-   * @param scope current scope
-   * @param inputDataset 
-   * @param rate A scalar representing the sample rate. Each element of `input_dataset` is
-   * retained with this probability, independent of all other elements.
-   * @param seed A scalar representing seed of random number generator.
-   * @param seed2 A scalar representing seed2 of random number generator.
-   * @param outputTypes 
-   * @param outputShapes 
-   * @return a new instance of SamplingDataset
-   */
-  @Endpoint(describeByClass = true)
-  public static SamplingDataset create(Scope scope, Operand<?> inputDataset, Operand<TFloat32> rate, Operand<TInt64> seed, Operand<TInt64> seed2, List<Class<?>> outputTypes, List<Shape> outputShapes) {
-    OperationBuilder opBuilder = scope.env().opBuilder("SamplingDataset", scope.makeOpName("SamplingDataset"));
-    opBuilder.addInput(inputDataset.asOutput(scope));
-    opBuilder.addInput(rate.asOutput(scope));
-    opBuilder.addInput(seed.asOutput(scope));
-    opBuilder.addInput(seed2.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    Class[] outputTypesArray = new Class[outputTypes.size()];
-    for (int i = 0; i < outputTypesArray.length; ++i) {
-      outputTypesArray[i] = outputTypes.get(i);
-    }
-    opBuilder.setAttr("output_types", outputTypesArray);
-    Shape[] outputShapesArray = new Shape[outputShapes.size()];
-    for (int i = 0; i < outputShapesArray.length; ++i) {
-      outputShapesArray[i] = outputShapes.get(i);
-    }
-    opBuilder.setAttr("output_shapes", outputShapesArray);
-    return new SamplingDataset(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<?> handle() {
-    return handle;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) handle;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "SamplingDataset";
-  
-  private Output<?> handle;
-  
-  private SamplingDataset(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    handle = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SerializeIterator.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SerializeIterator.java
deleted file mode 100644
index e4b8a08231e..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SerializeIterator.java
+++ /dev/null
@@ -1,109 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.data;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Converts the given `resource_handle` representing an iterator to a variant tensor.
- */
-@Operator(group = "data")
-public final class SerializeIterator extends RawOp implements Operand<TType> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.data.SerializeIterator}
-   */
-  public static class Options {
-    
-    /**
-     * @param externalStatePolicy 
-     */
-    public Options externalStatePolicy(Long externalStatePolicy) {
-      this.externalStatePolicy = externalStatePolicy;
-      return this;
-    }
-    
-    private Long externalStatePolicy;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new SerializeIterator operation.
-   * 
-   * @param scope current scope
-   * @param resourceHandle A handle to an iterator resource.
-   * @param options carries optional attributes values
-   * @return a new instance of SerializeIterator
-   */
-  @Endpoint(describeByClass = true)
-  public static SerializeIterator create(Scope scope, Operand<?> resourceHandle, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("SerializeIterator", scope.makeOpName("SerializeIterator"));
-    opBuilder.addInput(resourceHandle.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.externalStatePolicy != null) {
-          opBuilder.setAttr("external_state_policy", opts.externalStatePolicy);
-        }
-      }
-    }
-    return new SerializeIterator(opBuilder.build());
-  }
-  
-  /**
-   * @param externalStatePolicy 
-   */
-  public static Options externalStatePolicy(Long externalStatePolicy) {
-    return new Options().externalStatePolicy(externalStatePolicy);
-  }
-  
-  /**
-   * A variant tensor storing the state of the iterator contained in the
-   * resource.
-   */
-  public Output<?> serialized() {
-    return serialized;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) serialized;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "SerializeIterator";
-  
-  private Output<?> serialized;
-  
-  private SerializeIterator(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    serialized = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SetStatsAggregatorDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SetStatsAggregatorDataset.java
deleted file mode 100644
index 69fd92189eb..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SetStatsAggregatorDataset.java
+++ /dev/null
@@ -1,92 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.data;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TString;
-import org.tensorflow.types.family.TType;
-
-/**
- */
-public final class SetStatsAggregatorDataset extends RawOp implements Operand<TType> {
-  
-  /**
-   * Factory method to create a class wrapping a new SetStatsAggregatorDataset operation.
-   * 
-   * @param scope current scope
-   * @param inputDataset 
-   * @param statsAggregator 
-   * @param tag 
-   * @param counterPrefix 
-   * @param outputTypes 
-   * @param outputShapes 
-   * @return a new instance of SetStatsAggregatorDataset
-   */
-  @Endpoint(describeByClass = true)
-  public static SetStatsAggregatorDataset create(Scope scope, Operand<?> inputDataset, Operand<?> statsAggregator, Operand<TString> tag, Operand<TString> counterPrefix, List<Class<?>> outputTypes, List<Shape> outputShapes) {
-    OperationBuilder opBuilder = scope.env().opBuilder("SetStatsAggregatorDataset", scope.makeOpName("SetStatsAggregatorDataset"));
-    opBuilder.addInput(inputDataset.asOutput(scope));
-    opBuilder.addInput(statsAggregator.asOutput(scope));
-    opBuilder.addInput(tag.asOutput(scope));
-    opBuilder.addInput(counterPrefix.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    Class[] outputTypesArray = new Class[outputTypes.size()];
-    for (int i = 0; i < outputTypesArray.length; ++i) {
-      outputTypesArray[i] = outputTypes.get(i);
-    }
-    opBuilder.setAttr("output_types", outputTypesArray);
-    Shape[] outputShapesArray = new Shape[outputShapes.size()];
-    for (int i = 0; i < outputShapesArray.length; ++i) {
-      outputShapesArray[i] = outputShapes.get(i);
-    }
-    opBuilder.setAttr("output_shapes", outputShapesArray);
-    return new SetStatsAggregatorDataset(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<?> handle() {
-    return handle;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) handle;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "SetStatsAggregatorDataset";
-  
-  private Output<?> handle;
-  
-  private SetStatsAggregatorDataset(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    handle = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ShardDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ShardDataset.java
deleted file mode 100644
index 6ac6547cc41..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ShardDataset.java
+++ /dev/null
@@ -1,125 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.data;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
-
-/**
- * Creates a `Dataset` that includes only 1/`num_shards` of this dataset.
- */
-public final class ShardDataset extends RawOp implements Operand<TType> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.data.ShardDataset}
-   */
-  public static class Options {
-    
-    /**
-     * @param requireNonEmpty 
-     */
-    public Options requireNonEmpty(Boolean requireNonEmpty) {
-      this.requireNonEmpty = requireNonEmpty;
-      return this;
-    }
-    
-    private Boolean requireNonEmpty;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new ShardDataset operation.
-   * 
-   * @param scope current scope
-   * @param inputDataset 
-   * @param numShards An integer representing the number of shards operating in parallel.
-   * @param index An integer representing the current worker index.
-   * @param outputTypes 
-   * @param outputShapes 
-   * @param options carries optional attributes values
-   * @return a new instance of ShardDataset
-   */
-  @Endpoint(describeByClass = true)
-  public static ShardDataset create(Scope scope, Operand<?> inputDataset, Operand<TInt64> numShards, Operand<TInt64> index, List<Class<?>> outputTypes, List<Shape> outputShapes, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ShardDataset", scope.makeOpName("ShardDataset"));
-    opBuilder.addInput(inputDataset.asOutput(scope));
-    opBuilder.addInput(numShards.asOutput(scope));
-    opBuilder.addInput(index.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    Class[] outputTypesArray = new Class[outputTypes.size()];
-    for (int i = 0; i < outputTypesArray.length; ++i) {
-      outputTypesArray[i] = outputTypes.get(i);
-    }
-    opBuilder.setAttr("output_types", outputTypesArray);
-    Shape[] outputShapesArray = new Shape[outputShapes.size()];
-    for (int i = 0; i < outputShapesArray.length; ++i) {
-      outputShapesArray[i] = outputShapes.get(i);
-    }
-    opBuilder.setAttr("output_shapes", outputShapesArray);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.requireNonEmpty != null) {
-          opBuilder.setAttr("require_non_empty", opts.requireNonEmpty);
-        }
-      }
-    }
-    return new ShardDataset(opBuilder.build());
-  }
-  
-  /**
-   * @param requireNonEmpty 
-   */
-  public static Options requireNonEmpty(Boolean requireNonEmpty) {
-    return new Options().requireNonEmpty(requireNonEmpty);
-  }
-  
-  /**
-   */
-  public Output<?> handle() {
-    return handle;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) handle;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ShardDataset";
-  
-  private Output<?> handle;
-  
-  private ShardDataset(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    handle = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ShuffleAndRepeatDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ShuffleAndRepeatDataset.java
deleted file mode 100644
index 2c9ab0f4218..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ShuffleAndRepeatDataset.java
+++ /dev/null
@@ -1,130 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.data;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
-
-/**
- */
-public final class ShuffleAndRepeatDataset extends RawOp implements Operand<TType> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.data.ShuffleAndRepeatDataset}
-   */
-  public static class Options {
-    
-    /**
-     * @param reshuffleEachIteration 
-     */
-    public Options reshuffleEachIteration(Boolean reshuffleEachIteration) {
-      this.reshuffleEachIteration = reshuffleEachIteration;
-      return this;
-    }
-    
-    private Boolean reshuffleEachIteration;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new ShuffleAndRepeatDataset operation.
-   * 
-   * @param scope current scope
-   * @param inputDataset 
-   * @param bufferSize 
-   * @param seed 
-   * @param seed2 
-   * @param count 
-   * @param seedGenerator 
-   * @param outputTypes 
-   * @param outputShapes 
-   * @param options carries optional attributes values
-   * @return a new instance of ShuffleAndRepeatDataset
-   */
-  @Endpoint(describeByClass = true)
-  public static ShuffleAndRepeatDataset create(Scope scope, Operand<?> inputDataset, Operand<TInt64> bufferSize, Operand<TInt64> seed, Operand<TInt64> seed2, Operand<TInt64> count, Operand<?> seedGenerator, List<Class<?>> outputTypes, List<Shape> outputShapes, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ShuffleAndRepeatDatasetV2", scope.makeOpName("ShuffleAndRepeatDataset"));
-    opBuilder.addInput(inputDataset.asOutput(scope));
-    opBuilder.addInput(bufferSize.asOutput(scope));
-    opBuilder.addInput(seed.asOutput(scope));
-    opBuilder.addInput(seed2.asOutput(scope));
-    opBuilder.addInput(count.asOutput(scope));
-    opBuilder.addInput(seedGenerator.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    Class[] outputTypesArray = new Class[outputTypes.size()];
-    for (int i = 0; i < outputTypesArray.length; ++i) {
-      outputTypesArray[i] = outputTypes.get(i);
-    }
-    opBuilder.setAttr("output_types", outputTypesArray);
-    Shape[] outputShapesArray = new Shape[outputShapes.size()];
-    for (int i = 0; i < outputShapesArray.length; ++i) {
-      outputShapesArray[i] = outputShapes.get(i);
-    }
-    opBuilder.setAttr("output_shapes", outputShapesArray);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.reshuffleEachIteration != null) {
-          opBuilder.setAttr("reshuffle_each_iteration", opts.reshuffleEachIteration);
-        }
-      }
-    }
-    return new ShuffleAndRepeatDataset(opBuilder.build());
-  }
-  
-  /**
-   * @param reshuffleEachIteration 
-   */
-  public static Options reshuffleEachIteration(Boolean reshuffleEachIteration) {
-    return new Options().reshuffleEachIteration(reshuffleEachIteration);
-  }
-  
-  /**
-   */
-  public Output<?> handle() {
-    return handle;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) handle;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ShuffleAndRepeatDatasetV2";
-  
-  private Output<?> handle;
-  
-  private ShuffleAndRepeatDataset(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    handle = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ShuffleDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ShuffleDataset.java
deleted file mode 100644
index 018958cb2da..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ShuffleDataset.java
+++ /dev/null
@@ -1,128 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.data;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
-
-/**
- */
-public final class ShuffleDataset extends RawOp implements Operand<TType> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.data.ShuffleDataset}
-   */
-  public static class Options {
-    
-    /**
-     * @param reshuffleEachIteration 
-     */
-    public Options reshuffleEachIteration(Boolean reshuffleEachIteration) {
-      this.reshuffleEachIteration = reshuffleEachIteration;
-      return this;
-    }
-    
-    private Boolean reshuffleEachIteration;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new ShuffleDataset operation.
-   * 
-   * @param scope current scope
-   * @param inputDataset 
-   * @param bufferSize 
-   * @param seed 
-   * @param seed2 
-   * @param seedGenerator 
-   * @param outputTypes 
-   * @param outputShapes 
-   * @param options carries optional attributes values
-   * @return a new instance of ShuffleDataset
-   */
-  @Endpoint(describeByClass = true)
-  public static ShuffleDataset create(Scope scope, Operand<?> inputDataset, Operand<TInt64> bufferSize, Operand<TInt64> seed, Operand<TInt64> seed2, Operand<?> seedGenerator, List<Class<?>> outputTypes, List<Shape> outputShapes, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ShuffleDatasetV3", scope.makeOpName("ShuffleDataset"));
-    opBuilder.addInput(inputDataset.asOutput(scope));
-    opBuilder.addInput(bufferSize.asOutput(scope));
-    opBuilder.addInput(seed.asOutput(scope));
-    opBuilder.addInput(seed2.asOutput(scope));
-    opBuilder.addInput(seedGenerator.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    Class[] outputTypesArray = new Class[outputTypes.size()];
-    for (int i = 0; i < outputTypesArray.length; ++i) {
-      outputTypesArray[i] = outputTypes.get(i);
-    }
-    opBuilder.setAttr("output_types", outputTypesArray);
-    Shape[] outputShapesArray = new Shape[outputShapes.size()];
-    for (int i = 0; i < outputShapesArray.length; ++i) {
-      outputShapesArray[i] = outputShapes.get(i);
-    }
-    opBuilder.setAttr("output_shapes", outputShapesArray);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.reshuffleEachIteration != null) {
-          opBuilder.setAttr("reshuffle_each_iteration", opts.reshuffleEachIteration);
-        }
-      }
-    }
-    return new ShuffleDataset(opBuilder.build());
-  }
-  
-  /**
-   * @param reshuffleEachIteration 
-   */
-  public static Options reshuffleEachIteration(Boolean reshuffleEachIteration) {
-    return new Options().reshuffleEachIteration(reshuffleEachIteration);
-  }
-  
-  /**
-   */
-  public Output<?> handle() {
-    return handle;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) handle;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ShuffleDatasetV3";
-  
-  private Output<?> handle;
-  
-  private ShuffleDataset(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    handle = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SkipDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SkipDataset.java
deleted file mode 100644
index a106c523ec4..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SkipDataset.java
+++ /dev/null
@@ -1,91 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.data;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
-
-/**
- * Creates a dataset that skips `count` elements from the `input_dataset`.
- */
-@Operator(group = "data")
-public final class SkipDataset extends RawOp implements Operand<TType> {
-  
-  /**
-   * Factory method to create a class wrapping a new SkipDataset operation.
-   * 
-   * @param scope current scope
-   * @param inputDataset 
-   * @param count A scalar representing the number of elements from the `input_dataset`
-   * that should be skipped.  If count is -1, skips everything.
-   * @param outputTypes 
-   * @param outputShapes 
-   * @return a new instance of SkipDataset
-   */
-  @Endpoint(describeByClass = true)
-  public static SkipDataset create(Scope scope, Operand<?> inputDataset, Operand<TInt64> count, List<Class<?>> outputTypes, List<Shape> outputShapes) {
-    OperationBuilder opBuilder = scope.env().opBuilder("SkipDataset", scope.makeOpName("SkipDataset"));
-    opBuilder.addInput(inputDataset.asOutput(scope));
-    opBuilder.addInput(count.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    Class[] outputTypesArray = new Class[outputTypes.size()];
-    for (int i = 0; i < outputTypesArray.length; ++i) {
-      outputTypesArray[i] = outputTypes.get(i);
-    }
-    opBuilder.setAttr("output_types", outputTypesArray);
-    Shape[] outputShapesArray = new Shape[outputShapes.size()];
-    for (int i = 0; i < outputShapesArray.length; ++i) {
-      outputShapesArray[i] = outputShapes.get(i);
-    }
-    opBuilder.setAttr("output_shapes", outputShapesArray);
-    return new SkipDataset(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<?> handle() {
-    return handle;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) handle;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "SkipDataset";
-  
-  private Output<?> handle;
-  
-  private SkipDataset(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    handle = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SleepDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SleepDataset.java
deleted file mode 100644
index 3ced1cb7a24..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SleepDataset.java
+++ /dev/null
@@ -1,88 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.data;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
-
-/**
- */
-public final class SleepDataset extends RawOp implements Operand<TType> {
-  
-  /**
-   * Factory method to create a class wrapping a new SleepDataset operation.
-   * 
-   * @param scope current scope
-   * @param inputDataset 
-   * @param sleepMicroseconds 
-   * @param outputTypes 
-   * @param outputShapes 
-   * @return a new instance of SleepDataset
-   */
-  @Endpoint(describeByClass = true)
-  public static SleepDataset create(Scope scope, Operand<?> inputDataset, Operand<TInt64> sleepMicroseconds, List<Class<?>> outputTypes, List<Shape> outputShapes) {
-    OperationBuilder opBuilder = scope.env().opBuilder("SleepDataset", scope.makeOpName("SleepDataset"));
-    opBuilder.addInput(inputDataset.asOutput(scope));
-    opBuilder.addInput(sleepMicroseconds.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    Class[] outputTypesArray = new Class[outputTypes.size()];
-    for (int i = 0; i < outputTypesArray.length; ++i) {
-      outputTypesArray[i] = outputTypes.get(i);
-    }
-    opBuilder.setAttr("output_types", outputTypesArray);
-    Shape[] outputShapesArray = new Shape[outputShapes.size()];
-    for (int i = 0; i < outputShapesArray.length; ++i) {
-      outputShapesArray[i] = outputShapes.get(i);
-    }
-    opBuilder.setAttr("output_shapes", outputShapesArray);
-    return new SleepDataset(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<?> handle() {
-    return handle;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) handle;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "SleepDataset";
-  
-  private Output<?> handle;
-  
-  private SleepDataset(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    handle = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SlidingWindowDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SlidingWindowDataset.java
deleted file mode 100644
index fd539ce0c11..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SlidingWindowDataset.java
+++ /dev/null
@@ -1,96 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.data;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
-
-/**
- * Creates a dataset that passes a sliding window over `input_dataset`.
- */
-public final class SlidingWindowDataset extends RawOp implements Operand<TType> {
-  
-  /**
-   * Factory method to create a class wrapping a new SlidingWindowDataset operation.
-   * 
-   * @param scope current scope
-   * @param inputDataset 
-   * @param windowSize A scalar representing the number of elements in the
-   * sliding window.
-   * @param windowShift A scalar representing the steps moving the sliding window
-   * forward in one iteration. It must be positive.
-   * @param windowStride A scalar representing the stride of the input elements of the sliding window.
-   * It must be positive.
-   * @param outputTypes 
-   * @param outputShapes 
-   * @return a new instance of SlidingWindowDataset
-   */
-  @Endpoint(describeByClass = true)
-  public static SlidingWindowDataset create(Scope scope, Operand<?> inputDataset, Operand<TInt64> windowSize, Operand<TInt64> windowShift, Operand<TInt64> windowStride, List<Class<?>> outputTypes, List<Shape> outputShapes) {
-    OperationBuilder opBuilder = scope.env().opBuilder("SlidingWindowDataset", scope.makeOpName("SlidingWindowDataset"));
-    opBuilder.addInput(inputDataset.asOutput(scope));
-    opBuilder.addInput(windowSize.asOutput(scope));
-    opBuilder.addInput(windowShift.asOutput(scope));
-    opBuilder.addInput(windowStride.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    Class[] outputTypesArray = new Class[outputTypes.size()];
-    for (int i = 0; i < outputTypesArray.length; ++i) {
-      outputTypesArray[i] = outputTypes.get(i);
-    }
-    opBuilder.setAttr("output_types", outputTypesArray);
-    Shape[] outputShapesArray = new Shape[outputShapes.size()];
-    for (int i = 0; i < outputShapesArray.length; ++i) {
-      outputShapesArray[i] = outputShapes.get(i);
-    }
-    opBuilder.setAttr("output_shapes", outputShapesArray);
-    return new SlidingWindowDataset(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<?> handle() {
-    return handle;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) handle;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "SlidingWindowDataset";
-  
-  private Output<?> handle;
-  
-  private SlidingWindowDataset(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    handle = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SparseTensorSliceDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SparseTensorSliceDataset.java
deleted file mode 100644
index 23785d96b37..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SparseTensorSliceDataset.java
+++ /dev/null
@@ -1,77 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.data;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
-
-/**
- * Creates a dataset that splits a SparseTensor into elements row-wise.
- */
-public final class SparseTensorSliceDataset extends RawOp implements Operand<TType> {
-  
-  /**
-   * Factory method to create a class wrapping a new SparseTensorSliceDataset operation.
-   * 
-   * @param scope current scope
-   * @param indices 
-   * @param values 
-   * @param denseShape 
-   * @return a new instance of SparseTensorSliceDataset
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> SparseTensorSliceDataset create(Scope scope, Operand<TInt64> indices, Operand<T> values, Operand<TInt64> denseShape) {
-    OperationBuilder opBuilder = scope.env().opBuilder("SparseTensorSliceDataset", scope.makeOpName("SparseTensorSliceDataset"));
-    opBuilder.addInput(indices.asOutput(scope));
-    opBuilder.addInput(values.asOutput(scope));
-    opBuilder.addInput(denseShape.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new SparseTensorSliceDataset(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<?> handle() {
-    return handle;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) handle;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "SparseTensorSliceDataset";
-  
-  private Output<?> handle;
-  
-  private SparseTensorSliceDataset(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    handle = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SqlDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SqlDataset.java
deleted file mode 100644
index 3817e9926f7..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SqlDataset.java
+++ /dev/null
@@ -1,91 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.data;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TString;
-import org.tensorflow.types.family.TType;
-
-/**
- * Creates a dataset that executes a SQL query and emits rows of the result set.
- */
-public final class SqlDataset extends RawOp implements Operand<TType> {
-  
-  /**
-   * Factory method to create a class wrapping a new SqlDataset operation.
-   * 
-   * @param scope current scope
-   * @param driverName The database type. Currently, the only supported type is 'sqlite'.
-   * @param dataSourceName A connection string to connect to the database.
-   * @param query A SQL query to execute.
-   * @param outputTypes 
-   * @param outputShapes 
-   * @return a new instance of SqlDataset
-   */
-  @Endpoint(describeByClass = true)
-  public static SqlDataset create(Scope scope, Operand<TString> driverName, Operand<TString> dataSourceName, Operand<TString> query, List<Class<?>> outputTypes, List<Shape> outputShapes) {
-    OperationBuilder opBuilder = scope.env().opBuilder("SqlDataset", scope.makeOpName("SqlDataset"));
-    opBuilder.addInput(driverName.asOutput(scope));
-    opBuilder.addInput(dataSourceName.asOutput(scope));
-    opBuilder.addInput(query.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    Class[] outputTypesArray = new Class[outputTypes.size()];
-    for (int i = 0; i < outputTypesArray.length; ++i) {
-      outputTypesArray[i] = outputTypes.get(i);
-    }
-    opBuilder.setAttr("output_types", outputTypesArray);
-    Shape[] outputShapesArray = new Shape[outputShapes.size()];
-    for (int i = 0; i < outputShapesArray.length; ++i) {
-      outputShapesArray[i] = outputShapes.get(i);
-    }
-    opBuilder.setAttr("output_shapes", outputShapesArray);
-    return new SqlDataset(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<?> handle() {
-    return handle;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) handle;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "SqlDataset";
-  
-  private Output<?> handle;
-  
-  private SqlDataset(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    handle = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/StatsAggregatorHandle.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/StatsAggregatorHandle.java
deleted file mode 100644
index f96d0930a23..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/StatsAggregatorHandle.java
+++ /dev/null
@@ -1,123 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.data;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Creates a statistics manager resource.
- */
-public final class StatsAggregatorHandle extends RawOp implements Operand<TType> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.data.StatsAggregatorHandle}
-   */
-  public static class Options {
-    
-    /**
-     * @param container 
-     */
-    public Options container(String container) {
-      this.container = container;
-      return this;
-    }
-    
-    /**
-     * @param sharedName 
-     */
-    public Options sharedName(String sharedName) {
-      this.sharedName = sharedName;
-      return this;
-    }
-    
-    private String container;
-    private String sharedName;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new StatsAggregatorHandle operation.
-   * 
-   * @param scope current scope
-   * @param options carries optional attributes values
-   * @return a new instance of StatsAggregatorHandle
-   */
-  @Endpoint(describeByClass = true)
-  public static StatsAggregatorHandle create(Scope scope, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("StatsAggregatorHandle", scope.makeOpName("StatsAggregatorHandle"));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.container != null) {
-          opBuilder.setAttr("container", opts.container);
-        }
-        if (opts.sharedName != null) {
-          opBuilder.setAttr("shared_name", opts.sharedName);
-        }
-      }
-    }
-    return new StatsAggregatorHandle(opBuilder.build());
-  }
-  
-  /**
-   * @param container 
-   */
-  public static Options container(String container) {
-    return new Options().container(container);
-  }
-  
-  /**
-   * @param sharedName 
-   */
-  public static Options sharedName(String sharedName) {
-    return new Options().sharedName(sharedName);
-  }
-  
-  /**
-   */
-  public Output<?> handle() {
-    return handle;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) handle;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "StatsAggregatorHandle";
-  
-  private Output<?> handle;
-  
-  private StatsAggregatorHandle(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    handle = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/TakeDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/TakeDataset.java
deleted file mode 100644
index c0e8b37d4ce..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/TakeDataset.java
+++ /dev/null
@@ -1,92 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.data;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
-
-/**
- * Creates a dataset that contains `count` elements from the `input_dataset`.
- */
-@Operator(group = "data")
-public final class TakeDataset extends RawOp implements Operand<TType> {
-  
-  /**
-   * Factory method to create a class wrapping a new TakeDataset operation.
-   * 
-   * @param scope current scope
-   * @param inputDataset 
-   * @param count A scalar representing the number of elements from the `input_dataset`
-   * that should be taken. A value of `-1` indicates that all of `input_dataset`
-   * is taken.
-   * @param outputTypes 
-   * @param outputShapes 
-   * @return a new instance of TakeDataset
-   */
-  @Endpoint(describeByClass = true)
-  public static TakeDataset create(Scope scope, Operand<?> inputDataset, Operand<TInt64> count, List<Class<?>> outputTypes, List<Shape> outputShapes) {
-    OperationBuilder opBuilder = scope.env().opBuilder("TakeDataset", scope.makeOpName("TakeDataset"));
-    opBuilder.addInput(inputDataset.asOutput(scope));
-    opBuilder.addInput(count.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    Class[] outputTypesArray = new Class[outputTypes.size()];
-    for (int i = 0; i < outputTypesArray.length; ++i) {
-      outputTypesArray[i] = outputTypes.get(i);
-    }
-    opBuilder.setAttr("output_types", outputTypesArray);
-    Shape[] outputShapesArray = new Shape[outputShapes.size()];
-    for (int i = 0; i < outputShapesArray.length; ++i) {
-      outputShapesArray[i] = outputShapes.get(i);
-    }
-    opBuilder.setAttr("output_shapes", outputShapesArray);
-    return new TakeDataset(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<?> handle() {
-    return handle;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) handle;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "TakeDataset";
-  
-  private Output<?> handle;
-  
-  private TakeDataset(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    handle = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/TensorDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/TensorDataset.java
deleted file mode 100644
index 943ee03785e..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/TensorDataset.java
+++ /dev/null
@@ -1,81 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.data;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.op.Operands;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Creates a dataset that emits `components` as a tuple of tensors once.
- */
-public final class TensorDataset extends RawOp implements Operand<TType> {
-  
-  /**
-   * Factory method to create a class wrapping a new TensorDataset operation.
-   * 
-   * @param scope current scope
-   * @param components 
-   * @param outputShapes 
-   * @return a new instance of TensorDataset
-   */
-  @Endpoint(describeByClass = true)
-  public static TensorDataset create(Scope scope, Iterable<Operand<?>> components, List<Shape> outputShapes) {
-    OperationBuilder opBuilder = scope.env().opBuilder("TensorDataset", scope.makeOpName("TensorDataset"));
-    opBuilder.addInputList(Operands.asOutputs(scope, components));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    Shape[] outputShapesArray = new Shape[outputShapes.size()];
-    for (int i = 0; i < outputShapesArray.length; ++i) {
-      outputShapesArray[i] = outputShapes.get(i);
-    }
-    opBuilder.setAttr("output_shapes", outputShapesArray);
-    return new TensorDataset(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<?> handle() {
-    return handle;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) handle;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "TensorDataset";
-  
-  private Output<?> handle;
-  
-  private TensorDataset(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    handle = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/TensorSliceDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/TensorSliceDataset.java
deleted file mode 100644
index 038991f405b..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/TensorSliceDataset.java
+++ /dev/null
@@ -1,82 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.data;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.op.Operands;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Creates a dataset that emits each dim-0 slice of `components` once.
- */
-@Operator(group = "data")
-public final class TensorSliceDataset extends RawOp implements Operand<TType> {
-  
-  /**
-   * Factory method to create a class wrapping a new TensorSliceDataset operation.
-   * 
-   * @param scope current scope
-   * @param components 
-   * @param outputShapes 
-   * @return a new instance of TensorSliceDataset
-   */
-  @Endpoint(describeByClass = true)
-  public static TensorSliceDataset create(Scope scope, Iterable<Operand<?>> components, List<Shape> outputShapes) {
-    OperationBuilder opBuilder = scope.env().opBuilder("TensorSliceDataset", scope.makeOpName("TensorSliceDataset"));
-    opBuilder.addInputList(Operands.asOutputs(scope, components));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    Shape[] outputShapesArray = new Shape[outputShapes.size()];
-    for (int i = 0; i < outputShapesArray.length; ++i) {
-      outputShapesArray[i] = outputShapes.get(i);
-    }
-    opBuilder.setAttr("output_shapes", outputShapesArray);
-    return new TensorSliceDataset(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<?> handle() {
-    return handle;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) handle;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "TensorSliceDataset";
-  
-  private Output<?> handle;
-  
-  private TensorSliceDataset(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    handle = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/TextLineDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/TextLineDataset.java
deleted file mode 100644
index c5f0b5de678..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/TextLineDataset.java
+++ /dev/null
@@ -1,81 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.data;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.TString;
-import org.tensorflow.types.family.TType;
-
-/**
- * Creates a dataset that emits the lines of one or more text files.
- */
-@Operator(group = "data")
-public final class TextLineDataset extends RawOp implements Operand<TType> {
-  
-  /**
-   * Factory method to create a class wrapping a new TextLineDataset operation.
-   * 
-   * @param scope current scope
-   * @param filenames A scalar or a vector containing the name(s) of the file(s) to be
-   * read.
-   * @param compressionType A scalar containing either (i) the empty string (no
-   * compression), (ii) "ZLIB", or (iii) "GZIP".
-   * @param bufferSize A scalar containing the number of bytes to buffer.
-   * @return a new instance of TextLineDataset
-   */
-  @Endpoint(describeByClass = true)
-  public static TextLineDataset create(Scope scope, Operand<TString> filenames, Operand<TString> compressionType, Operand<TInt64> bufferSize) {
-    OperationBuilder opBuilder = scope.env().opBuilder("TextLineDataset", scope.makeOpName("TextLineDataset"));
-    opBuilder.addInput(filenames.asOutput(scope));
-    opBuilder.addInput(compressionType.asOutput(scope));
-    opBuilder.addInput(bufferSize.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new TextLineDataset(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<?> handle() {
-    return handle;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) handle;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "TextLineDataset";
-  
-  private Output<?> handle;
-  
-  private TextLineDataset(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    handle = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/TfRecordDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/TfRecordDataset.java
deleted file mode 100644
index d87da730c87..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/TfRecordDataset.java
+++ /dev/null
@@ -1,82 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.data;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.TString;
-import org.tensorflow.types.family.TType;
-
-/**
- * Creates a dataset that emits the records from one or more TFRecord files.
- */
-@Operator(group = "data")
-public final class TfRecordDataset extends RawOp implements Operand<TType> {
-  
-  /**
-   * Factory method to create a class wrapping a new TfRecordDataset operation.
-   * 
-   * @param scope current scope
-   * @param filenames A scalar or vector containing the name(s) of the file(s) to be
-   * read.
-   * @param compressionType A scalar containing either (i) the empty string (no
-   * compression), (ii) "ZLIB", or (iii) "GZIP".
-   * @param bufferSize A scalar representing the number of bytes to buffer. A value of
-   * 0 means no buffering will be performed.
-   * @return a new instance of TfRecordDataset
-   */
-  @Endpoint(describeByClass = true)
-  public static TfRecordDataset create(Scope scope, Operand<TString> filenames, Operand<TString> compressionType, Operand<TInt64> bufferSize) {
-    OperationBuilder opBuilder = scope.env().opBuilder("TFRecordDataset", scope.makeOpName("TfRecordDataset"));
-    opBuilder.addInput(filenames.asOutput(scope));
-    opBuilder.addInput(compressionType.asOutput(scope));
-    opBuilder.addInput(bufferSize.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new TfRecordDataset(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<?> handle() {
-    return handle;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) handle;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "TFRecordDataset";
-  
-  private Output<?> handle;
-  
-  private TfRecordDataset(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    handle = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ThreadPoolDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ThreadPoolDataset.java
deleted file mode 100644
index b42f03010a5..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ThreadPoolDataset.java
+++ /dev/null
@@ -1,88 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.data;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Creates a dataset that uses a custom thread pool to compute `input_dataset`.
- */
-public final class ThreadPoolDataset extends RawOp implements Operand<TType> {
-  
-  /**
-   * Factory method to create a class wrapping a new ThreadPoolDataset operation.
-   * 
-   * @param scope current scope
-   * @param inputDataset 
-   * @param threadPool A resource produced by the ThreadPoolHandle op.
-   * @param outputTypes 
-   * @param outputShapes 
-   * @return a new instance of ThreadPoolDataset
-   */
-  @Endpoint(describeByClass = true)
-  public static ThreadPoolDataset create(Scope scope, Operand<?> inputDataset, Operand<?> threadPool, List<Class<?>> outputTypes, List<Shape> outputShapes) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ThreadPoolDataset", scope.makeOpName("ThreadPoolDataset"));
-    opBuilder.addInput(inputDataset.asOutput(scope));
-    opBuilder.addInput(threadPool.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    Class[] outputTypesArray = new Class[outputTypes.size()];
-    for (int i = 0; i < outputTypesArray.length; ++i) {
-      outputTypesArray[i] = outputTypes.get(i);
-    }
-    opBuilder.setAttr("output_types", outputTypesArray);
-    Shape[] outputShapesArray = new Shape[outputShapes.size()];
-    for (int i = 0; i < outputShapesArray.length; ++i) {
-      outputShapesArray[i] = outputShapes.get(i);
-    }
-    opBuilder.setAttr("output_shapes", outputShapesArray);
-    return new ThreadPoolDataset(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<?> handle() {
-    return handle;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) handle;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ThreadPoolDataset";
-  
-  private Output<?> handle;
-  
-  private ThreadPoolDataset(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    handle = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ThreadPoolHandle.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ThreadPoolHandle.java
deleted file mode 100644
index ef7ccb677a9..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ThreadPoolHandle.java
+++ /dev/null
@@ -1,152 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.data;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Creates a dataset that uses a custom thread pool to compute `input_dataset`.
- */
-public final class ThreadPoolHandle extends RawOp implements Operand<TType> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.data.ThreadPoolHandle}
-   */
-  public static class Options {
-    
-    /**
-     * @param maxIntraOpParallelism The maximum degree of parallelism to use within operations that execute on this
-     * threadpool.
-     */
-    public Options maxIntraOpParallelism(Long maxIntraOpParallelism) {
-      this.maxIntraOpParallelism = maxIntraOpParallelism;
-      return this;
-    }
-    
-    /**
-     * @param container 
-     */
-    public Options container(String container) {
-      this.container = container;
-      return this;
-    }
-    
-    /**
-     * @param sharedName 
-     */
-    public Options sharedName(String sharedName) {
-      this.sharedName = sharedName;
-      return this;
-    }
-    
-    private Long maxIntraOpParallelism;
-    private String container;
-    private String sharedName;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new ThreadPoolHandle operation.
-   * 
-   * @param scope current scope
-   * @param numThreads The number of threads in the thread pool.
-   * @param displayName A human-readable name for the threads that may be visible in some
-   * visualizations.
-   * threadpool.
-   * @param options carries optional attributes values
-   * @return a new instance of ThreadPoolHandle
-   */
-  @Endpoint(describeByClass = true)
-  public static ThreadPoolHandle create(Scope scope, Long numThreads, String displayName, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ThreadPoolHandle", scope.makeOpName("ThreadPoolHandle"));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("num_threads", numThreads);
-    opBuilder.setAttr("display_name", displayName);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.maxIntraOpParallelism != null) {
-          opBuilder.setAttr("max_intra_op_parallelism", opts.maxIntraOpParallelism);
-        }
-        if (opts.container != null) {
-          opBuilder.setAttr("container", opts.container);
-        }
-        if (opts.sharedName != null) {
-          opBuilder.setAttr("shared_name", opts.sharedName);
-        }
-      }
-    }
-    return new ThreadPoolHandle(opBuilder.build());
-  }
-  
-  /**
-   * @param maxIntraOpParallelism The maximum degree of parallelism to use within operations that execute on this
-   * threadpool.
-   */
-  public static Options maxIntraOpParallelism(Long maxIntraOpParallelism) {
-    return new Options().maxIntraOpParallelism(maxIntraOpParallelism);
-  }
-  
-  /**
-   * @param container 
-   */
-  public static Options container(String container) {
-    return new Options().container(container);
-  }
-  
-  /**
-   * @param sharedName 
-   */
-  public static Options sharedName(String sharedName) {
-    return new Options().sharedName(sharedName);
-  }
-  
-  /**
-   * A resource that can be consumed by one or more ExperimentalThreadPoolDataset
-   * ops.
-   */
-  public Output<?> handle() {
-    return handle;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) handle;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ThreadPoolHandle";
-  
-  private Output<?> handle;
-  
-  private ThreadPoolHandle(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    handle = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/UnbatchDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/UnbatchDataset.java
deleted file mode 100644
index a808972d5f5..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/UnbatchDataset.java
+++ /dev/null
@@ -1,86 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.data;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * A dataset that splits the elements of its input into multiple elements.
- */
-public final class UnbatchDataset extends RawOp implements Operand<TType> {
-  
-  /**
-   * Factory method to create a class wrapping a new UnbatchDataset operation.
-   * 
-   * @param scope current scope
-   * @param inputDataset 
-   * @param outputTypes 
-   * @param outputShapes 
-   * @return a new instance of UnbatchDataset
-   */
-  @Endpoint(describeByClass = true)
-  public static UnbatchDataset create(Scope scope, Operand<?> inputDataset, List<Class<?>> outputTypes, List<Shape> outputShapes) {
-    OperationBuilder opBuilder = scope.env().opBuilder("UnbatchDataset", scope.makeOpName("UnbatchDataset"));
-    opBuilder.addInput(inputDataset.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    Class[] outputTypesArray = new Class[outputTypes.size()];
-    for (int i = 0; i < outputTypesArray.length; ++i) {
-      outputTypesArray[i] = outputTypes.get(i);
-    }
-    opBuilder.setAttr("output_types", outputTypesArray);
-    Shape[] outputShapesArray = new Shape[outputShapes.size()];
-    for (int i = 0; i < outputShapesArray.length; ++i) {
-      outputShapesArray[i] = outputShapes.get(i);
-    }
-    opBuilder.setAttr("output_shapes", outputShapesArray);
-    return new UnbatchDataset(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<?> handle() {
-    return handle;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) handle;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "UnbatchDataset";
-  
-  private Output<?> handle;
-  
-  private UnbatchDataset(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    handle = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/UniqueDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/UniqueDataset.java
deleted file mode 100644
index e00906012c8..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/UniqueDataset.java
+++ /dev/null
@@ -1,86 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.data;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Creates a dataset that contains the unique elements of `input_dataset`.
- */
-public final class UniqueDataset extends RawOp implements Operand<TType> {
-  
-  /**
-   * Factory method to create a class wrapping a new UniqueDataset operation.
-   * 
-   * @param scope current scope
-   * @param inputDataset 
-   * @param outputTypes 
-   * @param outputShapes 
-   * @return a new instance of UniqueDataset
-   */
-  @Endpoint(describeByClass = true)
-  public static UniqueDataset create(Scope scope, Operand<?> inputDataset, List<Class<?>> outputTypes, List<Shape> outputShapes) {
-    OperationBuilder opBuilder = scope.env().opBuilder("UniqueDataset", scope.makeOpName("UniqueDataset"));
-    opBuilder.addInput(inputDataset.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    Class[] outputTypesArray = new Class[outputTypes.size()];
-    for (int i = 0; i < outputTypesArray.length; ++i) {
-      outputTypesArray[i] = outputTypes.get(i);
-    }
-    opBuilder.setAttr("output_types", outputTypesArray);
-    Shape[] outputShapesArray = new Shape[outputShapes.size()];
-    for (int i = 0; i < outputShapesArray.length; ++i) {
-      outputShapesArray[i] = outputShapes.get(i);
-    }
-    opBuilder.setAttr("output_shapes", outputShapesArray);
-    return new UniqueDataset(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<?> handle() {
-    return handle;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) handle;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "UniqueDataset";
-  
-  private Output<?> handle;
-  
-  private UniqueDataset(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    handle = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/UnwrapDatasetVariant.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/UnwrapDatasetVariant.java
deleted file mode 100644
index 982d43e9168..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/UnwrapDatasetVariant.java
+++ /dev/null
@@ -1,71 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.data;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- */
-public final class UnwrapDatasetVariant extends RawOp implements Operand<TType> {
-  
-  /**
-   * Factory method to create a class wrapping a new UnwrapDatasetVariant operation.
-   * 
-   * @param scope current scope
-   * @param inputHandle 
-   * @return a new instance of UnwrapDatasetVariant
-   */
-  @Endpoint(describeByClass = true)
-  public static UnwrapDatasetVariant create(Scope scope, Operand<?> inputHandle) {
-    OperationBuilder opBuilder = scope.env().opBuilder("UnwrapDatasetVariant", scope.makeOpName("UnwrapDatasetVariant"));
-    opBuilder.addInput(inputHandle.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new UnwrapDatasetVariant(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<?> outputHandle() {
-    return outputHandle;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) outputHandle;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "UnwrapDatasetVariant";
-  
-  private Output<?> outputHandle;
-  
-  private UnwrapDatasetVariant(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    outputHandle = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/WindowDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/WindowDataset.java
deleted file mode 100644
index 33f679e2f31..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/WindowDataset.java
+++ /dev/null
@@ -1,143 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.data;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TBool;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
-
-/**
- *   Combines (nests of) input elements into a dataset of (nests of) windows.
- * <p>
- *   A "window" is a finite dataset of flat elements of size `size` (or possibly
- *   fewer if there are not enough input elements to fill the window and
- *   `drop_remainder` evaluates to false).
- * <p>
- *   The `shift` argument determines the number of input elements by which
- *   the window moves on each iteration.  The first element in the `k`th window
- *   will be element
- * <p>
- *   <pre>{@code
- *   1 + (k-1) * shift
- *   }</pre>
- * of the input dataset. In particular, the first element of the first window
- *   will always be the first element of the input dataset.  
- * <p>
- *   If the `stride` parameter is greater than 1, then each window will skip
- *   `(stride - 1)` input elements between each element that appears in the
- *   window. Output windows will still contain `size` elements regardless of
- *   the value of `stride`.
- * <p>
- *   The `stride` argument determines the stride of the input elements, and the
- *   `shift` argument determines the shift of the window.
- * <p>
- *   For example, letting `{...}` to represent a Dataset:
- * <p>
- *   - `tf.data.Dataset.range(7).window(2)` produces
- *     `{{0, 1}, {2, 3}, {4, 5}, {6}}`
- *   - `tf.data.Dataset.range(7).window(3, 2, 1, True)` produces
- *     `{{0, 1, 2}, {2, 3, 4}, {4, 5, 6}}`
- *   - `tf.data.Dataset.range(7).window(3, 1, 2, True)` produces
- *     `{{0, 2, 4}, {1, 3, 5}, {2, 4, 6}}`
- * <p>
- *   Note that when the `window` transformation is applied to a dataset of
- *   nested elements, it produces a dataset of nested windows.
- * <p>
- *   For example:
- * <p>
- *   - `tf.data.Dataset.from_tensor_slices((range(4), range(4))).window(2)`
- *     produces `{({0, 1}, {0, 1}), ({2, 3}, {2, 3})}`
- *   - `tf.data.Dataset.from_tensor_slices({"a": range(4)}).window(2)`
- *     produces `{{"a": {0, 1}}, {"a": {2, 3}}}`
- */
-public final class WindowDataset extends RawOp implements Operand<TType> {
-  
-  /**
-   * Factory method to create a class wrapping a new WindowDataset operation.
-   * 
-   * @param scope current scope
-   * @param inputDataset 
-   * @param size An integer scalar, representing the number of elements
-   * of the input dataset to combine into a window. Must be positive.
-   * @param shift An integer scalar, representing the number of input elements
-   * by which the window moves in each iteration.  Defaults to `size`.
-   * Must be positive.
-   * @param stride An integer scalar, representing the stride of the input elements
-   * in the sliding window. Must be positive. The default value of 1 means
-   * "retain every input element".
-   * @param dropRemainder A Boolean scalar, representing whether the last window should be
-   * dropped if its size is smaller than `window_size`.
-   * @param outputTypes 
-   * @param outputShapes 
-   * @return a new instance of WindowDataset
-   */
-  @Endpoint(describeByClass = true)
-  public static WindowDataset create(Scope scope, Operand<?> inputDataset, Operand<TInt64> size, Operand<TInt64> shift, Operand<TInt64> stride, Operand<TBool> dropRemainder, List<Class<?>> outputTypes, List<Shape> outputShapes) {
-    OperationBuilder opBuilder = scope.env().opBuilder("WindowDataset", scope.makeOpName("WindowDataset"));
-    opBuilder.addInput(inputDataset.asOutput(scope));
-    opBuilder.addInput(size.asOutput(scope));
-    opBuilder.addInput(shift.asOutput(scope));
-    opBuilder.addInput(stride.asOutput(scope));
-    opBuilder.addInput(dropRemainder.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    Class[] outputTypesArray = new Class[outputTypes.size()];
-    for (int i = 0; i < outputTypesArray.length; ++i) {
-      outputTypesArray[i] = outputTypes.get(i);
-    }
-    opBuilder.setAttr("output_types", outputTypesArray);
-    Shape[] outputShapesArray = new Shape[outputShapes.size()];
-    for (int i = 0; i < outputShapesArray.length; ++i) {
-      outputShapesArray[i] = outputShapes.get(i);
-    }
-    opBuilder.setAttr("output_shapes", outputShapesArray);
-    return new WindowDataset(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<?> handle() {
-    return handle;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) handle;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "WindowDataset";
-  
-  private Output<?> handle;
-  
-  private WindowDataset(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    handle = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/WrapDatasetVariant.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/WrapDatasetVariant.java
deleted file mode 100644
index 8a80750fcf5..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/WrapDatasetVariant.java
+++ /dev/null
@@ -1,71 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.data;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- */
-public final class WrapDatasetVariant extends RawOp implements Operand<TType> {
-  
-  /**
-   * Factory method to create a class wrapping a new WrapDatasetVariant operation.
-   * 
-   * @param scope current scope
-   * @param inputHandle 
-   * @return a new instance of WrapDatasetVariant
-   */
-  @Endpoint(describeByClass = true)
-  public static WrapDatasetVariant create(Scope scope, Operand<?> inputHandle) {
-    OperationBuilder opBuilder = scope.env().opBuilder("WrapDatasetVariant", scope.makeOpName("WrapDatasetVariant"));
-    opBuilder.addInput(inputHandle.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new WrapDatasetVariant(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<?> outputHandle() {
-    return outputHandle;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) outputHandle;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "WrapDatasetVariant";
-  
-  private Output<?> outputHandle;
-  
-  private WrapDatasetVariant(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    outputHandle = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ZipDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ZipDataset.java
deleted file mode 100644
index f4497d1c109..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ZipDataset.java
+++ /dev/null
@@ -1,94 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.data;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.op.Operands;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Creates a dataset that zips together `input_datasets`.
- * <p>
- * The elements of the resulting dataset are created by zipping corresponding
- * elements from each of the input datasets.
- * <p>
- * The size of the resulting dataset will match the size of the smallest input
- * dataset, and no error will be raised if input datasets have different sizes.
- */
-@Operator(group = "data")
-public final class ZipDataset extends RawOp implements Operand<TType> {
-  
-  /**
-   * Factory method to create a class wrapping a new ZipDataset operation.
-   * 
-   * @param scope current scope
-   * @param inputDatasets List of `N` variant Tensors representing datasets to be zipped together.
-   * @param outputTypes 
-   * @param outputShapes 
-   * @return a new instance of ZipDataset
-   */
-  @Endpoint(describeByClass = true)
-  public static ZipDataset create(Scope scope, Iterable<Operand<?>> inputDatasets, List<Class<?>> outputTypes, List<Shape> outputShapes) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ZipDataset", scope.makeOpName("ZipDataset"));
-    opBuilder.addInputList(Operands.asOutputs(scope, inputDatasets));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    Class[] outputTypesArray = new Class[outputTypes.size()];
-    for (int i = 0; i < outputTypesArray.length; ++i) {
-      outputTypesArray[i] = outputTypes.get(i);
-    }
-    opBuilder.setAttr("output_types", outputTypesArray);
-    Shape[] outputShapesArray = new Shape[outputShapes.size()];
-    for (int i = 0; i < outputShapesArray.length; ++i) {
-      outputShapesArray[i] = outputShapes.get(i);
-    }
-    opBuilder.setAttr("output_shapes", outputShapesArray);
-    return new ZipDataset(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<?> handle() {
-    return handle;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) handle;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ZipDataset";
-  
-  private Output<?> handle;
-  
-  private ZipDataset(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    handle = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/AssertCardinalityDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/AssertCardinalityDataset.java
deleted file mode 100644
index 69e7f0d769e..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/AssertCardinalityDataset.java
+++ /dev/null
@@ -1,88 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.data.experimental;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
-
-/**
- */
-public final class AssertCardinalityDataset extends RawOp implements Operand<TType> {
-  
-  /**
-   * Factory method to create a class wrapping a new AssertCardinalityDataset operation.
-   * 
-   * @param scope current scope
-   * @param inputDataset 
-   * @param cardinality 
-   * @param outputTypes 
-   * @param outputShapes 
-   * @return a new instance of AssertCardinalityDataset
-   */
-  @Endpoint(describeByClass = true)
-  public static AssertCardinalityDataset create(Scope scope, Operand<?> inputDataset, Operand<TInt64> cardinality, List<Class<?>> outputTypes, List<Shape> outputShapes) {
-    OperationBuilder opBuilder = scope.env().opBuilder("AssertCardinalityDataset", scope.makeOpName("AssertCardinalityDataset"));
-    opBuilder.addInput(inputDataset.asOutput(scope));
-    opBuilder.addInput(cardinality.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    Class[] outputTypesArray = new Class[outputTypes.size()];
-    for (int i = 0; i < outputTypesArray.length; ++i) {
-      outputTypesArray[i] = outputTypes.get(i);
-    }
-    opBuilder.setAttr("output_types", outputTypesArray);
-    Shape[] outputShapesArray = new Shape[outputShapes.size()];
-    for (int i = 0; i < outputShapesArray.length; ++i) {
-      outputShapesArray[i] = outputShapes.get(i);
-    }
-    opBuilder.setAttr("output_shapes", outputShapesArray);
-    return new AssertCardinalityDataset(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<?> handle() {
-    return handle;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) handle;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "AssertCardinalityDataset";
-  
-  private Output<?> handle;
-  
-  private AssertCardinalityDataset(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    handle = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/AssertNextDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/AssertNextDataset.java
deleted file mode 100644
index 213dd6428e2..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/AssertNextDataset.java
+++ /dev/null
@@ -1,88 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.data.experimental;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TString;
-import org.tensorflow.types.family.TType;
-
-/**
- */
-public final class AssertNextDataset extends RawOp implements Operand<TType> {
-  
-  /**
-   * Factory method to create a class wrapping a new AssertNextDataset operation.
-   * 
-   * @param scope current scope
-   * @param inputDataset 
-   * @param transformations 
-   * @param outputTypes 
-   * @param outputShapes 
-   * @return a new instance of AssertNextDataset
-   */
-  @Endpoint(describeByClass = true)
-  public static AssertNextDataset create(Scope scope, Operand<?> inputDataset, Operand<TString> transformations, List<Class<?>> outputTypes, List<Shape> outputShapes) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ExperimentalAssertNextDataset", scope.makeOpName("AssertNextDataset"));
-    opBuilder.addInput(inputDataset.asOutput(scope));
-    opBuilder.addInput(transformations.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    Class[] outputTypesArray = new Class[outputTypes.size()];
-    for (int i = 0; i < outputTypesArray.length; ++i) {
-      outputTypesArray[i] = outputTypes.get(i);
-    }
-    opBuilder.setAttr("output_types", outputTypesArray);
-    Shape[] outputShapesArray = new Shape[outputShapes.size()];
-    for (int i = 0; i < outputShapesArray.length; ++i) {
-      outputShapesArray[i] = outputShapes.get(i);
-    }
-    opBuilder.setAttr("output_shapes", outputShapesArray);
-    return new AssertNextDataset(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<?> handle() {
-    return handle;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) handle;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ExperimentalAssertNextDataset";
-  
-  private Output<?> handle;
-  
-  private AssertNextDataset(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    handle = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/AutoShardDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/AutoShardDataset.java
deleted file mode 100644
index 5cc6afb9abc..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/AutoShardDataset.java
+++ /dev/null
@@ -1,133 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.data.experimental;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
-
-/**
- * Creates a dataset that shards the input dataset.
- * <p>
- * Creates a dataset that shards the input dataset by num_workers, returning a
- * sharded dataset for the index-th worker. This attempts to automatically shard
- * a dataset by examining the Dataset graph and inserting a shard op before the
- * inputs to a reader Dataset (e.g. CSVDataset, TFRecordDataset).
- * <p>
- * This dataset will throw a NotFound error if we cannot shard the dataset
- * automatically.
- */
-public final class AutoShardDataset extends RawOp implements Operand<TType> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.data.experimental.AutoShardDataset}
-   */
-  public static class Options {
-    
-    /**
-     * @param autoShardPolicy 
-     */
-    public Options autoShardPolicy(Long autoShardPolicy) {
-      this.autoShardPolicy = autoShardPolicy;
-      return this;
-    }
-    
-    private Long autoShardPolicy;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new AutoShardDataset operation.
-   * 
-   * @param scope current scope
-   * @param inputDataset A variant tensor representing the input dataset.
-   * @param numWorkers A scalar representing the number of workers to distribute this dataset across.
-   * @param index A scalar representing the index of the current worker out of num_workers.
-   * @param outputTypes 
-   * @param outputShapes 
-   * @param options carries optional attributes values
-   * @return a new instance of AutoShardDataset
-   */
-  @Endpoint(describeByClass = true)
-  public static AutoShardDataset create(Scope scope, Operand<?> inputDataset, Operand<TInt64> numWorkers, Operand<TInt64> index, List<Class<?>> outputTypes, List<Shape> outputShapes, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ExperimentalAutoShardDataset", scope.makeOpName("AutoShardDataset"));
-    opBuilder.addInput(inputDataset.asOutput(scope));
-    opBuilder.addInput(numWorkers.asOutput(scope));
-    opBuilder.addInput(index.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    Class[] outputTypesArray = new Class[outputTypes.size()];
-    for (int i = 0; i < outputTypesArray.length; ++i) {
-      outputTypesArray[i] = outputTypes.get(i);
-    }
-    opBuilder.setAttr("output_types", outputTypesArray);
-    Shape[] outputShapesArray = new Shape[outputShapes.size()];
-    for (int i = 0; i < outputShapesArray.length; ++i) {
-      outputShapesArray[i] = outputShapes.get(i);
-    }
-    opBuilder.setAttr("output_shapes", outputShapesArray);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.autoShardPolicy != null) {
-          opBuilder.setAttr("auto_shard_policy", opts.autoShardPolicy);
-        }
-      }
-    }
-    return new AutoShardDataset(opBuilder.build());
-  }
-  
-  /**
-   * @param autoShardPolicy 
-   */
-  public static Options autoShardPolicy(Long autoShardPolicy) {
-    return new Options().autoShardPolicy(autoShardPolicy);
-  }
-  
-  /**
-   */
-  public Output<?> handle() {
-    return handle;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) handle;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ExperimentalAutoShardDataset";
-  
-  private Output<?> handle;
-  
-  private AutoShardDataset(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    handle = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/BytesProducedStatsDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/BytesProducedStatsDataset.java
deleted file mode 100644
index 6c102aef09d..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/BytesProducedStatsDataset.java
+++ /dev/null
@@ -1,89 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.data.experimental;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TString;
-import org.tensorflow.types.family.TType;
-
-/**
- * Records the bytes size of each element of `input_dataset` in a StatsAggregator.
- */
-public final class BytesProducedStatsDataset extends RawOp implements Operand<TType> {
-  
-  /**
-   * Factory method to create a class wrapping a new BytesProducedStatsDataset operation.
-   * 
-   * @param scope current scope
-   * @param inputDataset 
-   * @param tag 
-   * @param outputTypes 
-   * @param outputShapes 
-   * @return a new instance of BytesProducedStatsDataset
-   */
-  @Endpoint(describeByClass = true)
-  public static BytesProducedStatsDataset create(Scope scope, Operand<?> inputDataset, Operand<TString> tag, List<Class<?>> outputTypes, List<Shape> outputShapes) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ExperimentalBytesProducedStatsDataset", scope.makeOpName("BytesProducedStatsDataset"));
-    opBuilder.addInput(inputDataset.asOutput(scope));
-    opBuilder.addInput(tag.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    Class[] outputTypesArray = new Class[outputTypes.size()];
-    for (int i = 0; i < outputTypesArray.length; ++i) {
-      outputTypesArray[i] = outputTypes.get(i);
-    }
-    opBuilder.setAttr("output_types", outputTypesArray);
-    Shape[] outputShapesArray = new Shape[outputShapes.size()];
-    for (int i = 0; i < outputShapesArray.length; ++i) {
-      outputShapesArray[i] = outputShapes.get(i);
-    }
-    opBuilder.setAttr("output_shapes", outputShapesArray);
-    return new BytesProducedStatsDataset(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<?> handle() {
-    return handle;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) handle;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ExperimentalBytesProducedStatsDataset";
-  
-  private Output<?> handle;
-  
-  private BytesProducedStatsDataset(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    handle = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/CSVDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/CSVDataset.java
deleted file mode 100644
index 1fa07608599..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/CSVDataset.java
+++ /dev/null
@@ -1,99 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.data.experimental;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.op.Operands;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TBool;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.TString;
-import org.tensorflow.types.family.TType;
-
-/**
- */
-public final class CSVDataset extends RawOp implements Operand<TType> {
-  
-  /**
-   * Factory method to create a class wrapping a new CSVDataset operation.
-   * 
-   * @param scope current scope
-   * @param filenames 
-   * @param compressionType 
-   * @param bufferSize 
-   * @param header 
-   * @param fieldDelim 
-   * @param useQuoteDelim 
-   * @param naValue 
-   * @param selectCols 
-   * @param recordDefaults 
-   * @param outputShapes 
-   * @return a new instance of CSVDataset
-   */
-  @Endpoint(describeByClass = true)
-  public static CSVDataset create(Scope scope, Operand<TString> filenames, Operand<TString> compressionType, Operand<TInt64> bufferSize, Operand<TBool> header, Operand<TString> fieldDelim, Operand<TBool> useQuoteDelim, Operand<TString> naValue, Operand<TInt64> selectCols, Iterable<Operand<?>> recordDefaults, List<Shape> outputShapes) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ExperimentalCSVDataset", scope.makeOpName("CSVDataset"));
-    opBuilder.addInput(filenames.asOutput(scope));
-    opBuilder.addInput(compressionType.asOutput(scope));
-    opBuilder.addInput(bufferSize.asOutput(scope));
-    opBuilder.addInput(header.asOutput(scope));
-    opBuilder.addInput(fieldDelim.asOutput(scope));
-    opBuilder.addInput(useQuoteDelim.asOutput(scope));
-    opBuilder.addInput(naValue.asOutput(scope));
-    opBuilder.addInput(selectCols.asOutput(scope));
-    opBuilder.addInputList(Operands.asOutputs(scope, recordDefaults));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    Shape[] outputShapesArray = new Shape[outputShapes.size()];
-    for (int i = 0; i < outputShapesArray.length; ++i) {
-      outputShapesArray[i] = outputShapes.get(i);
-    }
-    opBuilder.setAttr("output_shapes", outputShapesArray);
-    return new CSVDataset(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<?> handle() {
-    return handle;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) handle;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ExperimentalCSVDataset";
-  
-  private Output<?> handle;
-  
-  private CSVDataset(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    handle = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/ChooseFastestDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/ChooseFastestDataset.java
deleted file mode 100644
index 25781e4afa1..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/ChooseFastestDataset.java
+++ /dev/null
@@ -1,88 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.data.experimental;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.op.Operands;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- */
-public final class ChooseFastestDataset extends RawOp implements Operand<TType> {
-  
-  /**
-   * Factory method to create a class wrapping a new ChooseFastestDataset operation.
-   * 
-   * @param scope current scope
-   * @param inputDatasets 
-   * @param numExperiments 
-   * @param outputTypes 
-   * @param outputShapes 
-   * @return a new instance of ChooseFastestDataset
-   */
-  @Endpoint(describeByClass = true)
-  public static ChooseFastestDataset create(Scope scope, Iterable<Operand<?>> inputDatasets, Long numExperiments, List<Class<?>> outputTypes, List<Shape> outputShapes) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ExperimentalChooseFastestDataset", scope.makeOpName("ChooseFastestDataset"));
-    opBuilder.addInputList(Operands.asOutputs(scope, inputDatasets));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("num_experiments", numExperiments);
-    Class[] outputTypesArray = new Class[outputTypes.size()];
-    for (int i = 0; i < outputTypesArray.length; ++i) {
-      outputTypesArray[i] = outputTypes.get(i);
-    }
-    opBuilder.setAttr("output_types", outputTypesArray);
-    Shape[] outputShapesArray = new Shape[outputShapes.size()];
-    for (int i = 0; i < outputShapesArray.length; ++i) {
-      outputShapesArray[i] = outputShapes.get(i);
-    }
-    opBuilder.setAttr("output_shapes", outputShapesArray);
-    return new ChooseFastestDataset(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<?> handle() {
-    return handle;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) handle;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ExperimentalChooseFastestDataset";
-  
-  private Output<?> handle;
-  
-  private ChooseFastestDataset(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    handle = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/CompressElement.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/CompressElement.java
deleted file mode 100644
index 0e853f754c3..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/CompressElement.java
+++ /dev/null
@@ -1,73 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.data.experimental;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.Operands;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Compresses a dataset element.
- */
-public final class CompressElement extends RawOp implements Operand<TType> {
-  
-  /**
-   * Factory method to create a class wrapping a new CompressElement operation.
-   * 
-   * @param scope current scope
-   * @param components 
-   * @return a new instance of CompressElement
-   */
-  @Endpoint(describeByClass = true)
-  public static CompressElement create(Scope scope, Iterable<Operand<?>> components) {
-    OperationBuilder opBuilder = scope.env().opBuilder("CompressElement", scope.makeOpName("CompressElement"));
-    opBuilder.addInputList(Operands.asOutputs(scope, components));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new CompressElement(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<?> compressed() {
-    return compressed;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) compressed;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "CompressElement";
-  
-  private Output<?> compressed;
-  
-  private CompressElement(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    compressed = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/DataServiceDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/DataServiceDataset.java
deleted file mode 100644
index da71533bd2b..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/DataServiceDataset.java
+++ /dev/null
@@ -1,134 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.data.experimental;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.TString;
-import org.tensorflow.types.family.TType;
-
-/**
- */
-@Operator(group = "data.experimental")
-public final class DataServiceDataset extends RawOp implements Operand<TType> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.data.experimental.DataServiceDataset}
-   */
-  public static class Options {
-    
-    /**
-     * @param taskRefreshIntervalHintMs 
-     */
-    public Options taskRefreshIntervalHintMs(Long taskRefreshIntervalHintMs) {
-      this.taskRefreshIntervalHintMs = taskRefreshIntervalHintMs;
-      return this;
-    }
-    
-    private Long taskRefreshIntervalHintMs;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new DataServiceDataset operation.
-   * 
-   * @param scope current scope
-   * @param datasetId 
-   * @param processingMode 
-   * @param address 
-   * @param protocol 
-   * @param jobName 
-   * @param maxOutstandingRequests 
-   * @param iterationCounter 
-   * @param outputTypes 
-   * @param outputShapes 
-   * @param options carries optional attributes values
-   * @return a new instance of DataServiceDataset
-   */
-  @Endpoint(describeByClass = true)
-  public static DataServiceDataset create(Scope scope, Operand<TInt64> datasetId, Operand<TString> processingMode, Operand<TString> address, Operand<TString> protocol, Operand<TString> jobName, Operand<TInt64> maxOutstandingRequests, Operand<?> iterationCounter, List<Class<?>> outputTypes, List<Shape> outputShapes, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("DataServiceDataset", scope.makeOpName("DataServiceDataset"));
-    opBuilder.addInput(datasetId.asOutput(scope));
-    opBuilder.addInput(processingMode.asOutput(scope));
-    opBuilder.addInput(address.asOutput(scope));
-    opBuilder.addInput(protocol.asOutput(scope));
-    opBuilder.addInput(jobName.asOutput(scope));
-    opBuilder.addInput(maxOutstandingRequests.asOutput(scope));
-    opBuilder.addInput(iterationCounter.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    Class[] outputTypesArray = new Class[outputTypes.size()];
-    for (int i = 0; i < outputTypesArray.length; ++i) {
-      outputTypesArray[i] = outputTypes.get(i);
-    }
-    opBuilder.setAttr("output_types", outputTypesArray);
-    Shape[] outputShapesArray = new Shape[outputShapes.size()];
-    for (int i = 0; i < outputShapesArray.length; ++i) {
-      outputShapesArray[i] = outputShapes.get(i);
-    }
-    opBuilder.setAttr("output_shapes", outputShapesArray);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.taskRefreshIntervalHintMs != null) {
-          opBuilder.setAttr("task_refresh_interval_hint_ms", opts.taskRefreshIntervalHintMs);
-        }
-      }
-    }
-    return new DataServiceDataset(opBuilder.build());
-  }
-  
-  /**
-   * @param taskRefreshIntervalHintMs 
-   */
-  public static Options taskRefreshIntervalHintMs(Long taskRefreshIntervalHintMs) {
-    return new Options().taskRefreshIntervalHintMs(taskRefreshIntervalHintMs);
-  }
-  
-  /**
-   */
-  public Output<?> handle() {
-    return handle;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) handle;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "DataServiceDataset";
-  
-  private Output<?> handle;
-  
-  private DataServiceDataset(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    handle = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/DatasetCardinality.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/DatasetCardinality.java
deleted file mode 100644
index 76aea531c3d..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/DatasetCardinality.java
+++ /dev/null
@@ -1,75 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.data.experimental;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt64;
-
-/**
- * Returns the cardinality of `input_dataset`.
- * <p>
- * Returns the cardinality of `input_dataset`.
- */
-public final class DatasetCardinality extends RawOp implements Operand<TInt64> {
-  
-  /**
-   * Factory method to create a class wrapping a new DatasetCardinality operation.
-   * 
-   * @param scope current scope
-   * @param inputDataset A variant tensor representing the dataset to return cardinality for.
-   * @return a new instance of DatasetCardinality
-   */
-  @Endpoint(describeByClass = true)
-  public static DatasetCardinality create(Scope scope, Operand<?> inputDataset) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ExperimentalDatasetCardinality", scope.makeOpName("DatasetCardinality"));
-    opBuilder.addInput(inputDataset.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new DatasetCardinality(opBuilder.build());
-  }
-  
-  /**
-   * The cardinality of `input_dataset`. Named constants are used to represent
-   * infinite and unknown cardinality.
-   */
-  public Output<TInt64> cardinality() {
-    return cardinality;
-  }
-  
-  @Override
-  public Output<TInt64> asOutput(Scope scope) {
-    return cardinality;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ExperimentalDatasetCardinality";
-  
-  private Output<TInt64> cardinality;
-  
-  private DatasetCardinality(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    cardinality = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/DatasetToTFRecord.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/DatasetToTFRecord.java
deleted file mode 100644
index ab2fda4e241..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/DatasetToTFRecord.java
+++ /dev/null
@@ -1,60 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.data.experimental;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TString;
-
-/**
- * Writes the given dataset to the given file using the TFRecord format.
- */
-public final class DatasetToTFRecord extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new DatasetToTFRecord operation.
-   * 
-   * @param scope current scope
-   * @param inputDataset A variant tensor representing the dataset to write.
-   * @param filename A scalar string tensor representing the filename to use.
-   * @param compressionType A scalar string tensor containing either (i) the empty string (no
-   * compression), (ii) "ZLIB", or (iii) "GZIP".
-   * @return a new instance of DatasetToTFRecord
-   */
-  @Endpoint(describeByClass = true)
-  public static DatasetToTFRecord create(Scope scope, Operand<?> inputDataset, Operand<TString> filename, Operand<TString> compressionType) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ExperimentalDatasetToTFRecord", scope.makeOpName("DatasetToTFRecord"));
-    opBuilder.addInput(inputDataset.asOutput(scope));
-    opBuilder.addInput(filename.asOutput(scope));
-    opBuilder.addInput(compressionType.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new DatasetToTFRecord(opBuilder.build());
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ExperimentalDatasetToTFRecord";
-  
-  private DatasetToTFRecord(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/DenseToSparseBatchDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/DenseToSparseBatchDataset.java
deleted file mode 100644
index a2b5f8a6db6..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/DenseToSparseBatchDataset.java
+++ /dev/null
@@ -1,94 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.data.experimental;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
-
-/**
- * Creates a dataset that batches input elements into a SparseTensor.
- */
-public final class DenseToSparseBatchDataset extends RawOp implements Operand<TType> {
-  
-  /**
-   * Factory method to create a class wrapping a new DenseToSparseBatchDataset operation.
-   * 
-   * @param scope current scope
-   * @param inputDataset A handle to an input dataset. Must have a single component.
-   * @param batchSize A scalar representing the number of elements to accumulate in a
-   * batch.
-   * @param rowShape A vector representing the dense shape of each row in the produced
-   * SparseTensor. The shape may be partially specified, using `-1` to indicate
-   * that a particular dimension should use the maximum size of all batch elements.
-   * @param outputTypes 
-   * @param outputShapes 
-   * @return a new instance of DenseToSparseBatchDataset
-   */
-  @Endpoint(describeByClass = true)
-  public static DenseToSparseBatchDataset create(Scope scope, Operand<?> inputDataset, Operand<TInt64> batchSize, Operand<TInt64> rowShape, List<Class<?>> outputTypes, List<Shape> outputShapes) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ExperimentalDenseToSparseBatchDataset", scope.makeOpName("DenseToSparseBatchDataset"));
-    opBuilder.addInput(inputDataset.asOutput(scope));
-    opBuilder.addInput(batchSize.asOutput(scope));
-    opBuilder.addInput(rowShape.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    Class[] outputTypesArray = new Class[outputTypes.size()];
-    for (int i = 0; i < outputTypesArray.length; ++i) {
-      outputTypesArray[i] = outputTypes.get(i);
-    }
-    opBuilder.setAttr("output_types", outputTypesArray);
-    Shape[] outputShapesArray = new Shape[outputShapes.size()];
-    for (int i = 0; i < outputShapesArray.length; ++i) {
-      outputShapesArray[i] = outputShapes.get(i);
-    }
-    opBuilder.setAttr("output_shapes", outputShapesArray);
-    return new DenseToSparseBatchDataset(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<?> handle() {
-    return handle;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) handle;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ExperimentalDenseToSparseBatchDataset";
-  
-  private Output<?> handle;
-  
-  private DenseToSparseBatchDataset(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    handle = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/DirectedInterleaveDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/DirectedInterleaveDataset.java
deleted file mode 100644
index b19d75ce552..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/DirectedInterleaveDataset.java
+++ /dev/null
@@ -1,91 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.data.experimental;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.op.Operands;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * A substitute for `InterleaveDataset` on a fixed list of `N` datasets.
- */
-public final class DirectedInterleaveDataset extends RawOp implements Operand<TType> {
-  
-  /**
-   * Factory method to create a class wrapping a new DirectedInterleaveDataset operation.
-   * 
-   * @param scope current scope
-   * @param selectorInputDataset A dataset of scalar `DT_INT64` elements that determines which of the
-   * `N` data inputs should produce the next output element.
-   * @param dataInputDatasets `N` datasets with the same type that will be interleaved according to
-   * the values of `selector_input_dataset`.
-   * @param outputTypes 
-   * @param outputShapes 
-   * @return a new instance of DirectedInterleaveDataset
-   */
-  @Endpoint(describeByClass = true)
-  public static DirectedInterleaveDataset create(Scope scope, Operand<?> selectorInputDataset, Iterable<Operand<?>> dataInputDatasets, List<Class<?>> outputTypes, List<Shape> outputShapes) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ExperimentalDirectedInterleaveDataset", scope.makeOpName("DirectedInterleaveDataset"));
-    opBuilder.addInput(selectorInputDataset.asOutput(scope));
-    opBuilder.addInputList(Operands.asOutputs(scope, dataInputDatasets));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    Class[] outputTypesArray = new Class[outputTypes.size()];
-    for (int i = 0; i < outputTypesArray.length; ++i) {
-      outputTypesArray[i] = outputTypes.get(i);
-    }
-    opBuilder.setAttr("output_types", outputTypesArray);
-    Shape[] outputShapesArray = new Shape[outputShapes.size()];
-    for (int i = 0; i < outputShapesArray.length; ++i) {
-      outputShapesArray[i] = outputShapes.get(i);
-    }
-    opBuilder.setAttr("output_shapes", outputShapesArray);
-    return new DirectedInterleaveDataset(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<?> handle() {
-    return handle;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) handle;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ExperimentalDirectedInterleaveDataset";
-  
-  private Output<?> handle;
-  
-  private DirectedInterleaveDataset(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    handle = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/DummyIterationCounter.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/DummyIterationCounter.java
deleted file mode 100644
index 9bc76fc0ba4..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/DummyIterationCounter.java
+++ /dev/null
@@ -1,69 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.data.experimental;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- */
-public final class DummyIterationCounter extends RawOp implements Operand<TType> {
-  
-  /**
-   * Factory method to create a class wrapping a new DummyIterationCounter operation.
-   * 
-   * @param scope current scope
-   * @return a new instance of DummyIterationCounter
-   */
-  @Endpoint(describeByClass = true)
-  public static DummyIterationCounter create(Scope scope) {
-    OperationBuilder opBuilder = scope.env().opBuilder("DummyIterationCounter", scope.makeOpName("DummyIterationCounter"));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new DummyIterationCounter(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<?> handle() {
-    return handle;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) handle;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "DummyIterationCounter";
-  
-  private Output<?> handle;
-  
-  private DummyIterationCounter(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    handle = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/IgnoreErrorsDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/IgnoreErrorsDataset.java
deleted file mode 100644
index 4ee758e24b8..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/IgnoreErrorsDataset.java
+++ /dev/null
@@ -1,86 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.data.experimental;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Creates a dataset that contains the elements of `input_dataset` ignoring errors.
- */
-public final class IgnoreErrorsDataset extends RawOp implements Operand<TType> {
-  
-  /**
-   * Factory method to create a class wrapping a new IgnoreErrorsDataset operation.
-   * 
-   * @param scope current scope
-   * @param inputDataset 
-   * @param outputTypes 
-   * @param outputShapes 
-   * @return a new instance of IgnoreErrorsDataset
-   */
-  @Endpoint(describeByClass = true)
-  public static IgnoreErrorsDataset create(Scope scope, Operand<?> inputDataset, List<Class<?>> outputTypes, List<Shape> outputShapes) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ExperimentalIgnoreErrorsDataset", scope.makeOpName("IgnoreErrorsDataset"));
-    opBuilder.addInput(inputDataset.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    Class[] outputTypesArray = new Class[outputTypes.size()];
-    for (int i = 0; i < outputTypesArray.length; ++i) {
-      outputTypesArray[i] = outputTypes.get(i);
-    }
-    opBuilder.setAttr("output_types", outputTypesArray);
-    Shape[] outputShapesArray = new Shape[outputShapes.size()];
-    for (int i = 0; i < outputShapesArray.length; ++i) {
-      outputShapesArray[i] = outputShapes.get(i);
-    }
-    opBuilder.setAttr("output_shapes", outputShapesArray);
-    return new IgnoreErrorsDataset(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<?> handle() {
-    return handle;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) handle;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ExperimentalIgnoreErrorsDataset";
-  
-  private Output<?> handle;
-  
-  private IgnoreErrorsDataset(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    handle = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/IteratorGetDevice.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/IteratorGetDevice.java
deleted file mode 100644
index f83241f5450..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/IteratorGetDevice.java
+++ /dev/null
@@ -1,71 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.data.experimental;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TString;
-
-/**
- * Returns the name of the device on which `resource` has been placed.
- */
-public final class IteratorGetDevice extends RawOp implements Operand<TString> {
-  
-  /**
-   * Factory method to create a class wrapping a new IteratorGetDevice operation.
-   * 
-   * @param scope current scope
-   * @param resource 
-   * @return a new instance of IteratorGetDevice
-   */
-  @Endpoint(describeByClass = true)
-  public static IteratorGetDevice create(Scope scope, Operand<?> resource) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ExperimentalIteratorGetDevice", scope.makeOpName("IteratorGetDevice"));
-    opBuilder.addInput(resource.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new IteratorGetDevice(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<TString> device() {
-    return device;
-  }
-  
-  @Override
-  public Output<TString> asOutput(Scope scope) {
-    return device;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ExperimentalIteratorGetDevice";
-  
-  private Output<TString> device;
-  
-  private IteratorGetDevice(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    device = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/LatencyStatsDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/LatencyStatsDataset.java
deleted file mode 100644
index 335d3a44a5e..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/LatencyStatsDataset.java
+++ /dev/null
@@ -1,89 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.data.experimental;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TString;
-import org.tensorflow.types.family.TType;
-
-/**
- * Records the latency of producing `input_dataset` elements in a StatsAggregator.
- */
-public final class LatencyStatsDataset extends RawOp implements Operand<TType> {
-  
-  /**
-   * Factory method to create a class wrapping a new LatencyStatsDataset operation.
-   * 
-   * @param scope current scope
-   * @param inputDataset 
-   * @param tag 
-   * @param outputTypes 
-   * @param outputShapes 
-   * @return a new instance of LatencyStatsDataset
-   */
-  @Endpoint(describeByClass = true)
-  public static LatencyStatsDataset create(Scope scope, Operand<?> inputDataset, Operand<TString> tag, List<Class<?>> outputTypes, List<Shape> outputShapes) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ExperimentalLatencyStatsDataset", scope.makeOpName("LatencyStatsDataset"));
-    opBuilder.addInput(inputDataset.asOutput(scope));
-    opBuilder.addInput(tag.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    Class[] outputTypesArray = new Class[outputTypes.size()];
-    for (int i = 0; i < outputTypesArray.length; ++i) {
-      outputTypesArray[i] = outputTypes.get(i);
-    }
-    opBuilder.setAttr("output_types", outputTypesArray);
-    Shape[] outputShapesArray = new Shape[outputShapes.size()];
-    for (int i = 0; i < outputShapesArray.length; ++i) {
-      outputShapesArray[i] = outputShapes.get(i);
-    }
-    opBuilder.setAttr("output_shapes", outputShapesArray);
-    return new LatencyStatsDataset(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<?> handle() {
-    return handle;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) handle;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ExperimentalLatencyStatsDataset";
-  
-  private Output<?> handle;
-  
-  private LatencyStatsDataset(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    handle = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/LmdbDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/LmdbDataset.java
deleted file mode 100644
index f0cba4ef849..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/LmdbDataset.java
+++ /dev/null
@@ -1,86 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.data.experimental;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TString;
-import org.tensorflow.types.family.TType;
-
-/**
- */
-public final class LmdbDataset extends RawOp implements Operand<TType> {
-  
-  /**
-   * Factory method to create a class wrapping a new LmdbDataset operation.
-   * 
-   * @param scope current scope
-   * @param filenames 
-   * @param outputTypes 
-   * @param outputShapes 
-   * @return a new instance of LmdbDataset
-   */
-  @Endpoint(describeByClass = true)
-  public static LmdbDataset create(Scope scope, Operand<TString> filenames, List<Class<?>> outputTypes, List<Shape> outputShapes) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ExperimentalLMDBDataset", scope.makeOpName("LmdbDataset"));
-    opBuilder.addInput(filenames.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    Class[] outputTypesArray = new Class[outputTypes.size()];
-    for (int i = 0; i < outputTypesArray.length; ++i) {
-      outputTypesArray[i] = outputTypes.get(i);
-    }
-    opBuilder.setAttr("output_types", outputTypesArray);
-    Shape[] outputShapesArray = new Shape[outputShapes.size()];
-    for (int i = 0; i < outputShapesArray.length; ++i) {
-      outputShapesArray[i] = outputShapes.get(i);
-    }
-    opBuilder.setAttr("output_shapes", outputShapesArray);
-    return new LmdbDataset(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<?> handle() {
-    return handle;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) handle;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ExperimentalLMDBDataset";
-  
-  private Output<?> handle;
-  
-  private LmdbDataset(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    handle = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/MatchingFilesDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/MatchingFilesDataset.java
deleted file mode 100644
index 9d4db8c1f39..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/MatchingFilesDataset.java
+++ /dev/null
@@ -1,72 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.data.experimental;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TString;
-import org.tensorflow.types.family.TType;
-
-/**
- */
-public final class MatchingFilesDataset extends RawOp implements Operand<TType> {
-  
-  /**
-   * Factory method to create a class wrapping a new MatchingFilesDataset operation.
-   * 
-   * @param scope current scope
-   * @param patterns 
-   * @return a new instance of MatchingFilesDataset
-   */
-  @Endpoint(describeByClass = true)
-  public static MatchingFilesDataset create(Scope scope, Operand<TString> patterns) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ExperimentalMatchingFilesDataset", scope.makeOpName("MatchingFilesDataset"));
-    opBuilder.addInput(patterns.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new MatchingFilesDataset(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<?> handle() {
-    return handle;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) handle;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ExperimentalMatchingFilesDataset";
-  
-  private Output<?> handle;
-  
-  private MatchingFilesDataset(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    handle = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/MaxIntraOpParallelismDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/MaxIntraOpParallelismDataset.java
deleted file mode 100644
index 62ae49c957d..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/MaxIntraOpParallelismDataset.java
+++ /dev/null
@@ -1,89 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.data.experimental;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
-
-/**
- * Creates a dataset that overrides the maximum intra-op parallelism.
- */
-public final class MaxIntraOpParallelismDataset extends RawOp implements Operand<TType> {
-  
-  /**
-   * Factory method to create a class wrapping a new MaxIntraOpParallelismDataset operation.
-   * 
-   * @param scope current scope
-   * @param inputDataset 
-   * @param maxIntraOpParallelism Identifies the maximum intra-op parallelism to use.
-   * @param outputTypes 
-   * @param outputShapes 
-   * @return a new instance of MaxIntraOpParallelismDataset
-   */
-  @Endpoint(describeByClass = true)
-  public static MaxIntraOpParallelismDataset create(Scope scope, Operand<?> inputDataset, Operand<TInt64> maxIntraOpParallelism, List<Class<?>> outputTypes, List<Shape> outputShapes) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ExperimentalMaxIntraOpParallelismDataset", scope.makeOpName("MaxIntraOpParallelismDataset"));
-    opBuilder.addInput(inputDataset.asOutput(scope));
-    opBuilder.addInput(maxIntraOpParallelism.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    Class[] outputTypesArray = new Class[outputTypes.size()];
-    for (int i = 0; i < outputTypesArray.length; ++i) {
-      outputTypesArray[i] = outputTypes.get(i);
-    }
-    opBuilder.setAttr("output_types", outputTypesArray);
-    Shape[] outputShapesArray = new Shape[outputShapes.size()];
-    for (int i = 0; i < outputShapesArray.length; ++i) {
-      outputShapesArray[i] = outputShapes.get(i);
-    }
-    opBuilder.setAttr("output_shapes", outputShapesArray);
-    return new MaxIntraOpParallelismDataset(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<?> handle() {
-    return handle;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) handle;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ExperimentalMaxIntraOpParallelismDataset";
-  
-  private Output<?> handle;
-  
-  private MaxIntraOpParallelismDataset(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    handle = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/NonSerializableDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/NonSerializableDataset.java
deleted file mode 100644
index e1a99b84aa3..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/NonSerializableDataset.java
+++ /dev/null
@@ -1,85 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.data.experimental;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- */
-public final class NonSerializableDataset extends RawOp implements Operand<TType> {
-  
-  /**
-   * Factory method to create a class wrapping a new NonSerializableDataset operation.
-   * 
-   * @param scope current scope
-   * @param inputDataset 
-   * @param outputTypes 
-   * @param outputShapes 
-   * @return a new instance of NonSerializableDataset
-   */
-  @Endpoint(describeByClass = true)
-  public static NonSerializableDataset create(Scope scope, Operand<?> inputDataset, List<Class<?>> outputTypes, List<Shape> outputShapes) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ExperimentalNonSerializableDataset", scope.makeOpName("NonSerializableDataset"));
-    opBuilder.addInput(inputDataset.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    Class[] outputTypesArray = new Class[outputTypes.size()];
-    for (int i = 0; i < outputTypesArray.length; ++i) {
-      outputTypesArray[i] = outputTypes.get(i);
-    }
-    opBuilder.setAttr("output_types", outputTypesArray);
-    Shape[] outputShapesArray = new Shape[outputShapes.size()];
-    for (int i = 0; i < outputShapesArray.length; ++i) {
-      outputShapesArray[i] = outputShapes.get(i);
-    }
-    opBuilder.setAttr("output_shapes", outputShapesArray);
-    return new NonSerializableDataset(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<?> handle() {
-    return handle;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) handle;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ExperimentalNonSerializableDataset";
-  
-  private Output<?> handle;
-  
-  private NonSerializableDataset(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    handle = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/ParseExampleDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/ParseExampleDataset.java
deleted file mode 100644
index f64b9741b13..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/ParseExampleDataset.java
+++ /dev/null
@@ -1,206 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.data.experimental;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.op.Operands;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
-
-/**
- * Transforms `input_dataset` containing `Example` protos as vectors of DT_STRING into a dataset of `Tensor` or `SparseTensor` objects representing the parsed features.
- */
-public final class ParseExampleDataset extends RawOp implements Operand<TType> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.data.experimental.ParseExampleDataset}
-   */
-  public static class Options {
-    
-    /**
-     * @param deterministic A string indicating the op-level determinism to use. Deterministic controls
-     * whether the dataset is allowed to return elements out of order if the next
-     * element to be returned isn't available, but a later element is. Options are
-     * "true", "false", and "default". "default" indicates that determinism should be
-     * decided by the `experimental_deterministic` parameter of `tf.data.Options`.
-     */
-    public Options deterministic(String deterministic) {
-      this.deterministic = deterministic;
-      return this;
-    }
-    
-    /**
-     * @param raggedKeys 
-     */
-    public Options raggedKeys(List<String> raggedKeys) {
-      this.raggedKeys = raggedKeys;
-      return this;
-    }
-    
-    private String deterministic;
-    private List<String> raggedKeys;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new ParseExampleDataset operation.
-   * 
-   * @param scope current scope
-   * @param inputDataset 
-   * @param numParallelCalls 
-   * @param denseDefaults A dict mapping string keys to `Tensor`s.
-   * The keys of the dict must match the dense_keys of the feature.
-   * @param sparseKeys A list of string keys in the examples features.
-   * The results for these keys will be returned as `SparseTensor` objects.
-   * @param denseKeys A list of Ndense string Tensors (scalars).
-   * The keys expected in the Examples features associated with dense values.
-   * @param sparseTypes A list of `DTypes` of the same length as `sparse_keys`.
-   * Only `tf.float32` (`FloatList`), `tf.int64` (`Int64List`),
-   * and `tf.string` (`BytesList`) are supported.
-   * @param denseShapes List of tuples with the same length as `dense_keys`.
-   * The shape of the data for each dense feature referenced by `dense_keys`.
-   * Required for any input tensors identified by `dense_keys`.  Must be
-   * either fully defined, or may contain an unknown first dimension.
-   * An unknown first dimension means the feature is treated as having
-   * a variable number of blocks, and the output shape along this dimension
-   * is considered unknown at graph build time.  Padding is applied for
-   * minibatch elements smaller than the maximum number of blocks for the
-   * given feature along this dimension.
-   * @param outputTypes The type list for the return values.
-   * @param outputShapes The list of shapes being produced.
-   * @param raggedValueTypes 
-   * @param raggedSplitTypes 
-   * @param options carries optional attributes values
-   * @return a new instance of ParseExampleDataset
-   */
-  @Endpoint(describeByClass = true)
-  public static ParseExampleDataset create(Scope scope, Operand<?> inputDataset, Operand<TInt64> numParallelCalls, Iterable<Operand<?>> denseDefaults, List<String> sparseKeys, List<String> denseKeys, List<Class<?>> sparseTypes, List<Shape> denseShapes, List<Class<?>> outputTypes, List<Shape> outputShapes, List<Class<?>> raggedValueTypes, List<Class<?>> raggedSplitTypes, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ParseExampleDatasetV2", scope.makeOpName("ParseExampleDataset"));
-    opBuilder.addInput(inputDataset.asOutput(scope));
-    opBuilder.addInput(numParallelCalls.asOutput(scope));
-    opBuilder.addInputList(Operands.asOutputs(scope, denseDefaults));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    String[] sparseKeysArray = new String[sparseKeys.size()];
-    for (int i = 0; i < sparseKeysArray.length; ++i) {
-      sparseKeysArray[i] = sparseKeys.get(i);
-    }
-    opBuilder.setAttr("sparse_keys", sparseKeysArray);
-    String[] denseKeysArray = new String[denseKeys.size()];
-    for (int i = 0; i < denseKeysArray.length; ++i) {
-      denseKeysArray[i] = denseKeys.get(i);
-    }
-    opBuilder.setAttr("dense_keys", denseKeysArray);
-    Class[] sparseTypesArray = new Class[sparseTypes.size()];
-    for (int i = 0; i < sparseTypesArray.length; ++i) {
-      sparseTypesArray[i] = sparseTypes.get(i);
-    }
-    opBuilder.setAttr("sparse_types", sparseTypesArray);
-    Shape[] denseShapesArray = new Shape[denseShapes.size()];
-    for (int i = 0; i < denseShapesArray.length; ++i) {
-      denseShapesArray[i] = denseShapes.get(i);
-    }
-    opBuilder.setAttr("dense_shapes", denseShapesArray);
-    Class[] outputTypesArray = new Class[outputTypes.size()];
-    for (int i = 0; i < outputTypesArray.length; ++i) {
-      outputTypesArray[i] = outputTypes.get(i);
-    }
-    opBuilder.setAttr("output_types", outputTypesArray);
-    Shape[] outputShapesArray = new Shape[outputShapes.size()];
-    for (int i = 0; i < outputShapesArray.length; ++i) {
-      outputShapesArray[i] = outputShapes.get(i);
-    }
-    opBuilder.setAttr("output_shapes", outputShapesArray);
-    Class[] raggedValueTypesArray = new Class[raggedValueTypes.size()];
-    for (int i = 0; i < raggedValueTypesArray.length; ++i) {
-      raggedValueTypesArray[i] = raggedValueTypes.get(i);
-    }
-    opBuilder.setAttr("ragged_value_types", raggedValueTypesArray);
-    Class[] raggedSplitTypesArray = new Class[raggedSplitTypes.size()];
-    for (int i = 0; i < raggedSplitTypesArray.length; ++i) {
-      raggedSplitTypesArray[i] = raggedSplitTypes.get(i);
-    }
-    opBuilder.setAttr("ragged_split_types", raggedSplitTypesArray);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.deterministic != null) {
-          opBuilder.setAttr("deterministic", opts.deterministic);
-        }
-        if (opts.raggedKeys != null) {
-          String[] raggedKeysArray = new String[opts.raggedKeys.size()];
-          for (int i = 0; i < raggedKeysArray.length; ++i) {
-            raggedKeysArray[i] = opts.raggedKeys.get(i);
-          }
-          opBuilder.setAttr("ragged_keys", raggedKeysArray);
-        }
-      }
-    }
-    return new ParseExampleDataset(opBuilder.build());
-  }
-  
-  /**
-   * @param deterministic A string indicating the op-level determinism to use. Deterministic controls
-   * whether the dataset is allowed to return elements out of order if the next
-   * element to be returned isn't available, but a later element is. Options are
-   * "true", "false", and "default". "default" indicates that determinism should be
-   * decided by the `experimental_deterministic` parameter of `tf.data.Options`.
-   */
-  public static Options deterministic(String deterministic) {
-    return new Options().deterministic(deterministic);
-  }
-  
-  /**
-   * @param raggedKeys 
-   */
-  public static Options raggedKeys(List<String> raggedKeys) {
-    return new Options().raggedKeys(raggedKeys);
-  }
-  
-  /**
-   */
-  public Output<?> handle() {
-    return handle;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) handle;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ParseExampleDatasetV2";
-  
-  private Output<?> handle;
-  
-  private ParseExampleDataset(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    handle = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/PrivateThreadPoolDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/PrivateThreadPoolDataset.java
deleted file mode 100644
index 0d972e9c9f4..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/PrivateThreadPoolDataset.java
+++ /dev/null
@@ -1,89 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.data.experimental;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
-
-/**
- * Creates a dataset that uses a custom thread pool to compute `input_dataset`.
- */
-public final class PrivateThreadPoolDataset extends RawOp implements Operand<TType> {
-  
-  /**
-   * Factory method to create a class wrapping a new PrivateThreadPoolDataset operation.
-   * 
-   * @param scope current scope
-   * @param inputDataset 
-   * @param numThreads Identifies the number of threads to use for the private threadpool.
-   * @param outputTypes 
-   * @param outputShapes 
-   * @return a new instance of PrivateThreadPoolDataset
-   */
-  @Endpoint(describeByClass = true)
-  public static PrivateThreadPoolDataset create(Scope scope, Operand<?> inputDataset, Operand<TInt64> numThreads, List<Class<?>> outputTypes, List<Shape> outputShapes) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ExperimentalPrivateThreadPoolDataset", scope.makeOpName("PrivateThreadPoolDataset"));
-    opBuilder.addInput(inputDataset.asOutput(scope));
-    opBuilder.addInput(numThreads.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    Class[] outputTypesArray = new Class[outputTypes.size()];
-    for (int i = 0; i < outputTypesArray.length; ++i) {
-      outputTypesArray[i] = outputTypes.get(i);
-    }
-    opBuilder.setAttr("output_types", outputTypesArray);
-    Shape[] outputShapesArray = new Shape[outputShapes.size()];
-    for (int i = 0; i < outputShapesArray.length; ++i) {
-      outputShapesArray[i] = outputShapes.get(i);
-    }
-    opBuilder.setAttr("output_shapes", outputShapesArray);
-    return new PrivateThreadPoolDataset(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<?> handle() {
-    return handle;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) handle;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ExperimentalPrivateThreadPoolDataset";
-  
-  private Output<?> handle;
-  
-  private PrivateThreadPoolDataset(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    handle = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/RandomDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/RandomDataset.java
deleted file mode 100644
index a00a6be2fe4..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/RandomDataset.java
+++ /dev/null
@@ -1,91 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.data.experimental;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
-
-/**
- * Creates a Dataset that returns pseudorandom numbers.
- */
-public final class RandomDataset extends RawOp implements Operand<TType> {
-  
-  /**
-   * Factory method to create a class wrapping a new RandomDataset operation.
-   * 
-   * @param scope current scope
-   * @param seed A scalar seed for the random number generator. If either seed or
-   * seed2 is set to be non-zero, the random number generator is seeded
-   * by the given seed.  Otherwise, a random seed is used.
-   * @param seed2 A second scalar seed to avoid seed collision.
-   * @param outputTypes 
-   * @param outputShapes 
-   * @return a new instance of RandomDataset
-   */
-  @Endpoint(describeByClass = true)
-  public static RandomDataset create(Scope scope, Operand<TInt64> seed, Operand<TInt64> seed2, List<Class<?>> outputTypes, List<Shape> outputShapes) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ExperimentalRandomDataset", scope.makeOpName("RandomDataset"));
-    opBuilder.addInput(seed.asOutput(scope));
-    opBuilder.addInput(seed2.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    Class[] outputTypesArray = new Class[outputTypes.size()];
-    for (int i = 0; i < outputTypesArray.length; ++i) {
-      outputTypesArray[i] = outputTypes.get(i);
-    }
-    opBuilder.setAttr("output_types", outputTypesArray);
-    Shape[] outputShapesArray = new Shape[outputShapes.size()];
-    for (int i = 0; i < outputShapesArray.length; ++i) {
-      outputShapesArray[i] = outputShapes.get(i);
-    }
-    opBuilder.setAttr("output_shapes", outputShapesArray);
-    return new RandomDataset(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<?> handle() {
-    return handle;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) handle;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ExperimentalRandomDataset";
-  
-  private Output<?> handle;
-  
-  private RandomDataset(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    handle = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/RebatchDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/RebatchDataset.java
deleted file mode 100644
index f59e00f2d48..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/RebatchDataset.java
+++ /dev/null
@@ -1,128 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.data.experimental;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
-
-/**
- * Creates a dataset that changes the batch size.
- * <p>
- * Creates a dataset that changes the batch size of the dataset to current batch
- * size // num_replicas.
- */
-public final class RebatchDataset extends RawOp implements Operand<TType> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.data.experimental.RebatchDataset}
-   */
-  public static class Options {
-    
-    /**
-     * @param useFallback 
-     */
-    public Options useFallback(Boolean useFallback) {
-      this.useFallback = useFallback;
-      return this;
-    }
-    
-    private Boolean useFallback;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new RebatchDataset operation.
-   * 
-   * @param scope current scope
-   * @param inputDataset A variant tensor representing the input dataset.
-   * @param numReplicas A scalar representing the number of replicas to distribute this batch across. As
-   * a result of this transformation the current batch size would end up being
-   * divided  by this parameter.
-   * @param outputTypes 
-   * @param outputShapes 
-   * @param options carries optional attributes values
-   * @return a new instance of RebatchDataset
-   */
-  @Endpoint(describeByClass = true)
-  public static RebatchDataset create(Scope scope, Operand<?> inputDataset, Operand<TInt64> numReplicas, List<Class<?>> outputTypes, List<Shape> outputShapes, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ExperimentalRebatchDataset", scope.makeOpName("RebatchDataset"));
-    opBuilder.addInput(inputDataset.asOutput(scope));
-    opBuilder.addInput(numReplicas.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    Class[] outputTypesArray = new Class[outputTypes.size()];
-    for (int i = 0; i < outputTypesArray.length; ++i) {
-      outputTypesArray[i] = outputTypes.get(i);
-    }
-    opBuilder.setAttr("output_types", outputTypesArray);
-    Shape[] outputShapesArray = new Shape[outputShapes.size()];
-    for (int i = 0; i < outputShapesArray.length; ++i) {
-      outputShapesArray[i] = outputShapes.get(i);
-    }
-    opBuilder.setAttr("output_shapes", outputShapesArray);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.useFallback != null) {
-          opBuilder.setAttr("use_fallback", opts.useFallback);
-        }
-      }
-    }
-    return new RebatchDataset(opBuilder.build());
-  }
-  
-  /**
-   * @param useFallback 
-   */
-  public static Options useFallback(Boolean useFallback) {
-    return new Options().useFallback(useFallback);
-  }
-  
-  /**
-   */
-  public Output<?> handle() {
-    return handle;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) handle;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ExperimentalRebatchDataset";
-  
-  private Output<?> handle;
-  
-  private RebatchDataset(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    handle = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/SetStatsAggregatorDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/SetStatsAggregatorDataset.java
deleted file mode 100644
index 1de0c59d921..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/SetStatsAggregatorDataset.java
+++ /dev/null
@@ -1,92 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.data.experimental;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TString;
-import org.tensorflow.types.family.TType;
-
-/**
- */
-public final class SetStatsAggregatorDataset extends RawOp implements Operand<TType> {
-  
-  /**
-   * Factory method to create a class wrapping a new SetStatsAggregatorDataset operation.
-   * 
-   * @param scope current scope
-   * @param inputDataset 
-   * @param statsAggregator 
-   * @param tag 
-   * @param counterPrefix 
-   * @param outputTypes 
-   * @param outputShapes 
-   * @return a new instance of SetStatsAggregatorDataset
-   */
-  @Endpoint(describeByClass = true)
-  public static SetStatsAggregatorDataset create(Scope scope, Operand<?> inputDataset, Operand<?> statsAggregator, Operand<TString> tag, Operand<TString> counterPrefix, List<Class<?>> outputTypes, List<Shape> outputShapes) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ExperimentalSetStatsAggregatorDataset", scope.makeOpName("SetStatsAggregatorDataset"));
-    opBuilder.addInput(inputDataset.asOutput(scope));
-    opBuilder.addInput(statsAggregator.asOutput(scope));
-    opBuilder.addInput(tag.asOutput(scope));
-    opBuilder.addInput(counterPrefix.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    Class[] outputTypesArray = new Class[outputTypes.size()];
-    for (int i = 0; i < outputTypesArray.length; ++i) {
-      outputTypesArray[i] = outputTypes.get(i);
-    }
-    opBuilder.setAttr("output_types", outputTypesArray);
-    Shape[] outputShapesArray = new Shape[outputShapes.size()];
-    for (int i = 0; i < outputShapesArray.length; ++i) {
-      outputShapesArray[i] = outputShapes.get(i);
-    }
-    opBuilder.setAttr("output_shapes", outputShapesArray);
-    return new SetStatsAggregatorDataset(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<?> handle() {
-    return handle;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) handle;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ExperimentalSetStatsAggregatorDataset";
-  
-  private Output<?> handle;
-  
-  private SetStatsAggregatorDataset(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    handle = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/SleepDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/SleepDataset.java
deleted file mode 100644
index e26f82bb21d..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/SleepDataset.java
+++ /dev/null
@@ -1,88 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.data.experimental;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
-
-/**
- */
-public final class SleepDataset extends RawOp implements Operand<TType> {
-  
-  /**
-   * Factory method to create a class wrapping a new SleepDataset operation.
-   * 
-   * @param scope current scope
-   * @param inputDataset 
-   * @param sleepMicroseconds 
-   * @param outputTypes 
-   * @param outputShapes 
-   * @return a new instance of SleepDataset
-   */
-  @Endpoint(describeByClass = true)
-  public static SleepDataset create(Scope scope, Operand<?> inputDataset, Operand<TInt64> sleepMicroseconds, List<Class<?>> outputTypes, List<Shape> outputShapes) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ExperimentalSleepDataset", scope.makeOpName("SleepDataset"));
-    opBuilder.addInput(inputDataset.asOutput(scope));
-    opBuilder.addInput(sleepMicroseconds.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    Class[] outputTypesArray = new Class[outputTypes.size()];
-    for (int i = 0; i < outputTypesArray.length; ++i) {
-      outputTypesArray[i] = outputTypes.get(i);
-    }
-    opBuilder.setAttr("output_types", outputTypesArray);
-    Shape[] outputShapesArray = new Shape[outputShapes.size()];
-    for (int i = 0; i < outputShapesArray.length; ++i) {
-      outputShapesArray[i] = outputShapes.get(i);
-    }
-    opBuilder.setAttr("output_shapes", outputShapesArray);
-    return new SleepDataset(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<?> handle() {
-    return handle;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) handle;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ExperimentalSleepDataset";
-  
-  private Output<?> handle;
-  
-  private SleepDataset(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    handle = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/SlidingWindowDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/SlidingWindowDataset.java
deleted file mode 100644
index 3ee0aaacd30..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/SlidingWindowDataset.java
+++ /dev/null
@@ -1,96 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.data.experimental;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
-
-/**
- * Creates a dataset that passes a sliding window over `input_dataset`.
- */
-public final class SlidingWindowDataset extends RawOp implements Operand<TType> {
-  
-  /**
-   * Factory method to create a class wrapping a new SlidingWindowDataset operation.
-   * 
-   * @param scope current scope
-   * @param inputDataset 
-   * @param windowSize A scalar representing the number of elements in the
-   * sliding window.
-   * @param windowShift A scalar representing the steps moving the sliding window
-   * forward in one iteration. It must be positive.
-   * @param windowStride A scalar representing the stride of the input elements of the sliding window.
-   * It must be positive.
-   * @param outputTypes 
-   * @param outputShapes 
-   * @return a new instance of SlidingWindowDataset
-   */
-  @Endpoint(describeByClass = true)
-  public static SlidingWindowDataset create(Scope scope, Operand<?> inputDataset, Operand<TInt64> windowSize, Operand<TInt64> windowShift, Operand<TInt64> windowStride, List<Class<?>> outputTypes, List<Shape> outputShapes) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ExperimentalSlidingWindowDataset", scope.makeOpName("SlidingWindowDataset"));
-    opBuilder.addInput(inputDataset.asOutput(scope));
-    opBuilder.addInput(windowSize.asOutput(scope));
-    opBuilder.addInput(windowShift.asOutput(scope));
-    opBuilder.addInput(windowStride.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    Class[] outputTypesArray = new Class[outputTypes.size()];
-    for (int i = 0; i < outputTypesArray.length; ++i) {
-      outputTypesArray[i] = outputTypes.get(i);
-    }
-    opBuilder.setAttr("output_types", outputTypesArray);
-    Shape[] outputShapesArray = new Shape[outputShapes.size()];
-    for (int i = 0; i < outputShapesArray.length; ++i) {
-      outputShapesArray[i] = outputShapes.get(i);
-    }
-    opBuilder.setAttr("output_shapes", outputShapesArray);
-    return new SlidingWindowDataset(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<?> handle() {
-    return handle;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) handle;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ExperimentalSlidingWindowDataset";
-  
-  private Output<?> handle;
-  
-  private SlidingWindowDataset(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    handle = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/SqlDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/SqlDataset.java
deleted file mode 100644
index 9e5f7d40144..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/SqlDataset.java
+++ /dev/null
@@ -1,91 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.data.experimental;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TString;
-import org.tensorflow.types.family.TType;
-
-/**
- * Creates a dataset that executes a SQL query and emits rows of the result set.
- */
-public final class SqlDataset extends RawOp implements Operand<TType> {
-  
-  /**
-   * Factory method to create a class wrapping a new SqlDataset operation.
-   * 
-   * @param scope current scope
-   * @param driverName The database type. Currently, the only supported type is 'sqlite'.
-   * @param dataSourceName A connection string to connect to the database.
-   * @param query A SQL query to execute.
-   * @param outputTypes 
-   * @param outputShapes 
-   * @return a new instance of SqlDataset
-   */
-  @Endpoint(describeByClass = true)
-  public static SqlDataset create(Scope scope, Operand<TString> driverName, Operand<TString> dataSourceName, Operand<TString> query, List<Class<?>> outputTypes, List<Shape> outputShapes) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ExperimentalSqlDataset", scope.makeOpName("SqlDataset"));
-    opBuilder.addInput(driverName.asOutput(scope));
-    opBuilder.addInput(dataSourceName.asOutput(scope));
-    opBuilder.addInput(query.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    Class[] outputTypesArray = new Class[outputTypes.size()];
-    for (int i = 0; i < outputTypesArray.length; ++i) {
-      outputTypesArray[i] = outputTypes.get(i);
-    }
-    opBuilder.setAttr("output_types", outputTypesArray);
-    Shape[] outputShapesArray = new Shape[outputShapes.size()];
-    for (int i = 0; i < outputShapesArray.length; ++i) {
-      outputShapesArray[i] = outputShapes.get(i);
-    }
-    opBuilder.setAttr("output_shapes", outputShapesArray);
-    return new SqlDataset(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<?> handle() {
-    return handle;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) handle;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ExperimentalSqlDataset";
-  
-  private Output<?> handle;
-  
-  private SqlDataset(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    handle = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/StatsAggregatorHandle.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/StatsAggregatorHandle.java
deleted file mode 100644
index c79ea93abb7..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/StatsAggregatorHandle.java
+++ /dev/null
@@ -1,122 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.data.experimental;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- */
-public final class StatsAggregatorHandle extends RawOp implements Operand<TType> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.data.experimental.StatsAggregatorHandle}
-   */
-  public static class Options {
-    
-    /**
-     * @param container 
-     */
-    public Options container(String container) {
-      this.container = container;
-      return this;
-    }
-    
-    /**
-     * @param sharedName 
-     */
-    public Options sharedName(String sharedName) {
-      this.sharedName = sharedName;
-      return this;
-    }
-    
-    private String container;
-    private String sharedName;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new StatsAggregatorHandle operation.
-   * 
-   * @param scope current scope
-   * @param options carries optional attributes values
-   * @return a new instance of StatsAggregatorHandle
-   */
-  @Endpoint(describeByClass = true)
-  public static StatsAggregatorHandle create(Scope scope, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("StatsAggregatorHandleV2", scope.makeOpName("StatsAggregatorHandle"));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.container != null) {
-          opBuilder.setAttr("container", opts.container);
-        }
-        if (opts.sharedName != null) {
-          opBuilder.setAttr("shared_name", opts.sharedName);
-        }
-      }
-    }
-    return new StatsAggregatorHandle(opBuilder.build());
-  }
-  
-  /**
-   * @param container 
-   */
-  public static Options container(String container) {
-    return new Options().container(container);
-  }
-  
-  /**
-   * @param sharedName 
-   */
-  public static Options sharedName(String sharedName) {
-    return new Options().sharedName(sharedName);
-  }
-  
-  /**
-   */
-  public Output<?> handle() {
-    return handle;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) handle;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "StatsAggregatorHandleV2";
-  
-  private Output<?> handle;
-  
-  private StatsAggregatorHandle(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    handle = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/StatsAggregatorSetSummaryWriter.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/StatsAggregatorSetSummaryWriter.java
deleted file mode 100644
index 52bf7e0be75..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/StatsAggregatorSetSummaryWriter.java
+++ /dev/null
@@ -1,56 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.data.experimental;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-
-/**
- * Set a summary_writer_interface to record statistics using given stats_aggregator.
- */
-public final class StatsAggregatorSetSummaryWriter extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new StatsAggregatorSetSummaryWriter operation.
-   * 
-   * @param scope current scope
-   * @param statsAggregator 
-   * @param summary 
-   * @return a new instance of StatsAggregatorSetSummaryWriter
-   */
-  @Endpoint(describeByClass = true)
-  public static StatsAggregatorSetSummaryWriter create(Scope scope, Operand<?> statsAggregator, Operand<?> summary) {
-    OperationBuilder opBuilder = scope.env().opBuilder("StatsAggregatorSetSummaryWriter", scope.makeOpName("StatsAggregatorSetSummaryWriter"));
-    opBuilder.addInput(statsAggregator.asOutput(scope));
-    opBuilder.addInput(summary.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new StatsAggregatorSetSummaryWriter(opBuilder.build());
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "StatsAggregatorSetSummaryWriter";
-  
-  private StatsAggregatorSetSummaryWriter(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/StatsAggregatorSummary.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/StatsAggregatorSummary.java
deleted file mode 100644
index ec12eaed6c7..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/StatsAggregatorSummary.java
+++ /dev/null
@@ -1,71 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.data.experimental;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TString;
-
-/**
- * Produces a summary of any statistics recorded by the given statistics manager.
- */
-public final class StatsAggregatorSummary extends RawOp implements Operand<TString> {
-  
-  /**
-   * Factory method to create a class wrapping a new StatsAggregatorSummary operation.
-   * 
-   * @param scope current scope
-   * @param iterator 
-   * @return a new instance of StatsAggregatorSummary
-   */
-  @Endpoint(describeByClass = true)
-  public static StatsAggregatorSummary create(Scope scope, Operand<?> iterator) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ExperimentalStatsAggregatorSummary", scope.makeOpName("StatsAggregatorSummary"));
-    opBuilder.addInput(iterator.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new StatsAggregatorSummary(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<TString> summary() {
-    return summary;
-  }
-  
-  @Override
-  public Output<TString> asOutput(Scope scope) {
-    return summary;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ExperimentalStatsAggregatorSummary";
-  
-  private Output<TString> summary;
-  
-  private StatsAggregatorSummary(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    summary = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/ThreadPoolDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/ThreadPoolDataset.java
deleted file mode 100644
index a6f66856cbc..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/ThreadPoolDataset.java
+++ /dev/null
@@ -1,88 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.data.experimental;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Creates a dataset that uses a custom thread pool to compute `input_dataset`.
- */
-public final class ThreadPoolDataset extends RawOp implements Operand<TType> {
-  
-  /**
-   * Factory method to create a class wrapping a new ThreadPoolDataset operation.
-   * 
-   * @param scope current scope
-   * @param inputDataset 
-   * @param threadPool A resource produced by the ThreadPoolHandle op.
-   * @param outputTypes 
-   * @param outputShapes 
-   * @return a new instance of ThreadPoolDataset
-   */
-  @Endpoint(describeByClass = true)
-  public static ThreadPoolDataset create(Scope scope, Operand<?> inputDataset, Operand<?> threadPool, List<Class<?>> outputTypes, List<Shape> outputShapes) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ExperimentalThreadPoolDataset", scope.makeOpName("ThreadPoolDataset"));
-    opBuilder.addInput(inputDataset.asOutput(scope));
-    opBuilder.addInput(threadPool.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    Class[] outputTypesArray = new Class[outputTypes.size()];
-    for (int i = 0; i < outputTypesArray.length; ++i) {
-      outputTypesArray[i] = outputTypes.get(i);
-    }
-    opBuilder.setAttr("output_types", outputTypesArray);
-    Shape[] outputShapesArray = new Shape[outputShapes.size()];
-    for (int i = 0; i < outputShapesArray.length; ++i) {
-      outputShapesArray[i] = outputShapes.get(i);
-    }
-    opBuilder.setAttr("output_shapes", outputShapesArray);
-    return new ThreadPoolDataset(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<?> handle() {
-    return handle;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) handle;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ExperimentalThreadPoolDataset";
-  
-  private Output<?> handle;
-  
-  private ThreadPoolDataset(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    handle = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/ThreadPoolHandle.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/ThreadPoolHandle.java
deleted file mode 100644
index 534c614c51b..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/ThreadPoolHandle.java
+++ /dev/null
@@ -1,152 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.data.experimental;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Creates a dataset that uses a custom thread pool to compute `input_dataset`.
- */
-public final class ThreadPoolHandle extends RawOp implements Operand<TType> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.data.experimental.ThreadPoolHandle}
-   */
-  public static class Options {
-    
-    /**
-     * @param maxIntraOpParallelism The maximum degree of parallelism to use within operations that execute on this
-     * threadpool.
-     */
-    public Options maxIntraOpParallelism(Long maxIntraOpParallelism) {
-      this.maxIntraOpParallelism = maxIntraOpParallelism;
-      return this;
-    }
-    
-    /**
-     * @param container 
-     */
-    public Options container(String container) {
-      this.container = container;
-      return this;
-    }
-    
-    /**
-     * @param sharedName 
-     */
-    public Options sharedName(String sharedName) {
-      this.sharedName = sharedName;
-      return this;
-    }
-    
-    private Long maxIntraOpParallelism;
-    private String container;
-    private String sharedName;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new ThreadPoolHandle operation.
-   * 
-   * @param scope current scope
-   * @param numThreads The number of threads in the thread pool.
-   * @param displayName A human-readable name for the threads that may be visible in some
-   * visualizations.
-   * threadpool.
-   * @param options carries optional attributes values
-   * @return a new instance of ThreadPoolHandle
-   */
-  @Endpoint(describeByClass = true)
-  public static ThreadPoolHandle create(Scope scope, Long numThreads, String displayName, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ExperimentalThreadPoolHandle", scope.makeOpName("ThreadPoolHandle"));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("num_threads", numThreads);
-    opBuilder.setAttr("display_name", displayName);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.maxIntraOpParallelism != null) {
-          opBuilder.setAttr("max_intra_op_parallelism", opts.maxIntraOpParallelism);
-        }
-        if (opts.container != null) {
-          opBuilder.setAttr("container", opts.container);
-        }
-        if (opts.sharedName != null) {
-          opBuilder.setAttr("shared_name", opts.sharedName);
-        }
-      }
-    }
-    return new ThreadPoolHandle(opBuilder.build());
-  }
-  
-  /**
-   * @param maxIntraOpParallelism The maximum degree of parallelism to use within operations that execute on this
-   * threadpool.
-   */
-  public static Options maxIntraOpParallelism(Long maxIntraOpParallelism) {
-    return new Options().maxIntraOpParallelism(maxIntraOpParallelism);
-  }
-  
-  /**
-   * @param container 
-   */
-  public static Options container(String container) {
-    return new Options().container(container);
-  }
-  
-  /**
-   * @param sharedName 
-   */
-  public static Options sharedName(String sharedName) {
-    return new Options().sharedName(sharedName);
-  }
-  
-  /**
-   * A resource that can be consumed by one or more ExperimentalThreadPoolDataset
-   * ops.
-   */
-  public Output<?> handle() {
-    return handle;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) handle;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ExperimentalThreadPoolHandle";
-  
-  private Output<?> handle;
-  
-  private ThreadPoolHandle(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    handle = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/UnbatchDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/UnbatchDataset.java
deleted file mode 100644
index b44a2ce1e25..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/UnbatchDataset.java
+++ /dev/null
@@ -1,86 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.data.experimental;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * A dataset that splits the elements of its input into multiple elements.
- */
-public final class UnbatchDataset extends RawOp implements Operand<TType> {
-  
-  /**
-   * Factory method to create a class wrapping a new UnbatchDataset operation.
-   * 
-   * @param scope current scope
-   * @param inputDataset 
-   * @param outputTypes 
-   * @param outputShapes 
-   * @return a new instance of UnbatchDataset
-   */
-  @Endpoint(describeByClass = true)
-  public static UnbatchDataset create(Scope scope, Operand<?> inputDataset, List<Class<?>> outputTypes, List<Shape> outputShapes) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ExperimentalUnbatchDataset", scope.makeOpName("UnbatchDataset"));
-    opBuilder.addInput(inputDataset.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    Class[] outputTypesArray = new Class[outputTypes.size()];
-    for (int i = 0; i < outputTypesArray.length; ++i) {
-      outputTypesArray[i] = outputTypes.get(i);
-    }
-    opBuilder.setAttr("output_types", outputTypesArray);
-    Shape[] outputShapesArray = new Shape[outputShapes.size()];
-    for (int i = 0; i < outputShapesArray.length; ++i) {
-      outputShapesArray[i] = outputShapes.get(i);
-    }
-    opBuilder.setAttr("output_shapes", outputShapesArray);
-    return new UnbatchDataset(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<?> handle() {
-    return handle;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) handle;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ExperimentalUnbatchDataset";
-  
-  private Output<?> handle;
-  
-  private UnbatchDataset(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    handle = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/UncompressElement.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/UncompressElement.java
deleted file mode 100644
index 384ba0d46d4..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/UncompressElement.java
+++ /dev/null
@@ -1,90 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.data.experimental;
-
-import java.util.Arrays;
-import java.util.Iterator;
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Uncompresses a compressed dataset element.
- */
-public final class UncompressElement extends RawOp implements Iterable<Operand<TType>> {
-  
-  /**
-   * Factory method to create a class wrapping a new UncompressElement operation.
-   * 
-   * @param scope current scope
-   * @param compressed 
-   * @param outputTypes 
-   * @param outputShapes 
-   * @return a new instance of UncompressElement
-   */
-  @Endpoint(describeByClass = true)
-  public static UncompressElement create(Scope scope, Operand<?> compressed, List<Class<?>> outputTypes, List<Shape> outputShapes) {
-    OperationBuilder opBuilder = scope.env().opBuilder("UncompressElement", scope.makeOpName("UncompressElement"));
-    opBuilder.addInput(compressed.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    Class[] outputTypesArray = new Class[outputTypes.size()];
-    for (int i = 0; i < outputTypesArray.length; ++i) {
-      outputTypesArray[i] = outputTypes.get(i);
-    }
-    opBuilder.setAttr("output_types", outputTypesArray);
-    Shape[] outputShapesArray = new Shape[outputShapes.size()];
-    for (int i = 0; i < outputShapesArray.length; ++i) {
-      outputShapesArray[i] = outputShapes.get(i);
-    }
-    opBuilder.setAttr("output_shapes", outputShapesArray);
-    return new UncompressElement(opBuilder.build());
-  }
-  
-  /**
-   */
-  public List<Output<?>> components() {
-    return components;
-  }
-  
-  @Override
-  @SuppressWarnings({"rawtypes", "unchecked"})
-  public Iterator<Operand<TType>> iterator() {
-    return (Iterator) components.iterator();
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "UncompressElement";
-  
-  private List<Output<?>> components;
-  
-  private UncompressElement(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    int componentsLength = operation.outputListLength("components");
-    components = Arrays.asList(operation.outputList(outputIdx, componentsLength));
-    outputIdx += componentsLength;
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/UniqueDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/UniqueDataset.java
deleted file mode 100644
index d59ddf296fd..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/UniqueDataset.java
+++ /dev/null
@@ -1,86 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.data.experimental;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Creates a dataset that contains the unique elements of `input_dataset`.
- */
-public final class UniqueDataset extends RawOp implements Operand<TType> {
-  
-  /**
-   * Factory method to create a class wrapping a new UniqueDataset operation.
-   * 
-   * @param scope current scope
-   * @param inputDataset 
-   * @param outputTypes 
-   * @param outputShapes 
-   * @return a new instance of UniqueDataset
-   */
-  @Endpoint(describeByClass = true)
-  public static UniqueDataset create(Scope scope, Operand<?> inputDataset, List<Class<?>> outputTypes, List<Shape> outputShapes) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ExperimentalUniqueDataset", scope.makeOpName("UniqueDataset"));
-    opBuilder.addInput(inputDataset.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    Class[] outputTypesArray = new Class[outputTypes.size()];
-    for (int i = 0; i < outputTypesArray.length; ++i) {
-      outputTypesArray[i] = outputTypes.get(i);
-    }
-    opBuilder.setAttr("output_types", outputTypesArray);
-    Shape[] outputShapesArray = new Shape[outputShapes.size()];
-    for (int i = 0; i < outputShapesArray.length; ++i) {
-      outputShapesArray[i] = outputShapes.get(i);
-    }
-    opBuilder.setAttr("output_shapes", outputShapesArray);
-    return new UniqueDataset(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<?> handle() {
-    return handle;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) handle;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ExperimentalUniqueDataset";
-  
-  private Output<?> handle;
-  
-  private UniqueDataset(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    handle = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/debugging/CheckNumerics.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/debugging/CheckNumerics.java
deleted file mode 100644
index d6285c89f54..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/debugging/CheckNumerics.java
+++ /dev/null
@@ -1,80 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.debugging;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Checks a tensor for NaN, -Inf and +Inf values.
- * <p>
- * When run, reports an `InvalidArgument` error if `tensor` has any values
- * that are not a number (NaN) or infinity (Inf). Otherwise, passes `tensor` as-is.
- * Unlike CheckNumerics (V1), CheckNumericsV2 distinguishes -Inf and +Inf in the
- * errors it throws.
- * 
- * @param <T> data type for {@code output()} output
- */
-public final class CheckNumerics<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new CheckNumerics operation.
-   * 
-   * @param scope current scope
-   * @param tensor 
-   * @param message Prefix of the error message.
-   * @return a new instance of CheckNumerics
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> CheckNumerics<T> create(Scope scope, Operand<T> tensor, String message) {
-    OperationBuilder opBuilder = scope.env().opBuilder("CheckNumericsV2", scope.makeOpName("CheckNumerics"));
-    opBuilder.addInput(tensor.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("message", message);
-    return new CheckNumerics<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "CheckNumericsV2";
-  
-  private Output<T> output;
-  
-  private CheckNumerics(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/debugging/DebugGradientIdentity.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/debugging/DebugGradientIdentity.java
deleted file mode 100644
index b851b34d426..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/debugging/DebugGradientIdentity.java
+++ /dev/null
@@ -1,77 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.debugging;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Identity op for gradient debugging.
- * <p>
- * This op is hidden from public in Python. It is used by TensorFlow Debugger to
- * register gradient tensors for gradient debugging.
- * This op operates on non-reference-type tensors.
- * 
- * @param <T> data type for {@code output()} output
- */
-public final class DebugGradientIdentity<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new DebugGradientIdentity operation.
-   * 
-   * @param scope current scope
-   * @param input 
-   * @return a new instance of DebugGradientIdentity
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> DebugGradientIdentity<T> create(Scope scope, Operand<T> input) {
-    OperationBuilder opBuilder = scope.env().opBuilder("DebugGradientIdentity", scope.makeOpName("DebugGradientIdentity"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new DebugGradientIdentity<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "DebugGradientIdentity";
-  
-  private Output<T> output;
-  
-  private DebugGradientIdentity(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/debugging/DebugGradientRefIdentity.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/debugging/DebugGradientRefIdentity.java
deleted file mode 100644
index 035ced72d8b..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/debugging/DebugGradientRefIdentity.java
+++ /dev/null
@@ -1,77 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.debugging;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Identity op for gradient debugging.
- * <p>
- * This op is hidden from public in Python. It is used by TensorFlow Debugger to
- * register gradient tensors for gradient debugging.
- * This op operates on reference-type tensors.
- * 
- * @param <T> data type for {@code output()} output
- */
-public final class DebugGradientRefIdentity<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new DebugGradientRefIdentity operation.
-   * 
-   * @param scope current scope
-   * @param input 
-   * @return a new instance of DebugGradientRefIdentity
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> DebugGradientRefIdentity<T> create(Scope scope, Operand<T> input) {
-    OperationBuilder opBuilder = scope.env().opBuilder("DebugGradientRefIdentity", scope.makeOpName("DebugGradientRefIdentity"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new DebugGradientRefIdentity<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "DebugGradientRefIdentity";
-  
-  private Output<T> output;
-  
-  private DebugGradientRefIdentity(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/debugging/DebugIdentity.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/debugging/DebugIdentity.java
deleted file mode 100644
index 4f075a98f6e..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/debugging/DebugIdentity.java
+++ /dev/null
@@ -1,242 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.debugging;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Debug Identity V2 Op.
- * <p>
- * Provides an identity mapping from input to output, while writing the content of
- * the input tensor by calling DebugEventsWriter.
- * <p>
- * The semantics of the input tensor depends on tensor_debug_mode. In typical
- * usage, the input tensor comes directly from the user computation only when
- * graph_debug_mode is FULL_TENSOR (see protobuf/debug_event.proto for a
- * list of all the possible values of graph_debug_mode). For the other debug modes,
- * the input tensor should be produced by an additional op or subgraph that
- * computes summary information about one or more tensors.
- * 
- * @param <T> data type for {@code output()} output
- */
-public final class DebugIdentity<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.debugging.DebugIdentity}
-   */
-  public static class Options {
-    
-    /**
-     * @param tfdbgContextId A tfdbg-generated ID for the context that the op belongs to,
-     *   e.g., a concrete compiled tf.function.
-     */
-    public Options tfdbgContextId(String tfdbgContextId) {
-      this.tfdbgContextId = tfdbgContextId;
-      return this;
-    }
-    
-    /**
-     * @param opName Optional. Name of the op that the debug op is concerned with.
-     *   Used only for single-tensor trace.
-     */
-    public Options opName(String opName) {
-      this.opName = opName;
-      return this;
-    }
-    
-    /**
-     * @param outputSlot Optional. Output slot index of the tensor that the debug op
-     *   is concerned with. Used only for single-tensor trace.
-     */
-    public Options outputSlot(Long outputSlot) {
-      this.outputSlot = outputSlot;
-      return this;
-    }
-    
-    /**
-     * @param tensorDebugMode TensorDebugMode enum value. See debug_event.proto for details.
-     */
-    public Options tensorDebugMode(Long tensorDebugMode) {
-      this.tensorDebugMode = tensorDebugMode;
-      return this;
-    }
-    
-    /**
-     * @param debugUrls List of URLs to debug targets, e.g., file:///foo/tfdbg_dump.
-     */
-    public Options debugUrls(List<String> debugUrls) {
-      this.debugUrls = debugUrls;
-      return this;
-    }
-    
-    /**
-     * @param circularBufferSize 
-     */
-    public Options circularBufferSize(Long circularBufferSize) {
-      this.circularBufferSize = circularBufferSize;
-      return this;
-    }
-    
-    /**
-     * @param tfdbgRunId 
-     */
-    public Options tfdbgRunId(String tfdbgRunId) {
-      this.tfdbgRunId = tfdbgRunId;
-      return this;
-    }
-    
-    private String tfdbgContextId;
-    private String opName;
-    private Long outputSlot;
-    private Long tensorDebugMode;
-    private List<String> debugUrls;
-    private Long circularBufferSize;
-    private String tfdbgRunId;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new DebugIdentity operation.
-   * 
-   * @param scope current scope
-   * @param input Input tensor, non-Reference type
-   * @param options carries optional attributes values
-   * @return a new instance of DebugIdentity
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> DebugIdentity<T> create(Scope scope, Operand<T> input, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("DebugIdentityV2", scope.makeOpName("DebugIdentity"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.tfdbgContextId != null) {
-          opBuilder.setAttr("tfdbg_context_id", opts.tfdbgContextId);
-        }
-        if (opts.opName != null) {
-          opBuilder.setAttr("op_name", opts.opName);
-        }
-        if (opts.outputSlot != null) {
-          opBuilder.setAttr("output_slot", opts.outputSlot);
-        }
-        if (opts.tensorDebugMode != null) {
-          opBuilder.setAttr("tensor_debug_mode", opts.tensorDebugMode);
-        }
-        if (opts.debugUrls != null) {
-          String[] debugUrlsArray = new String[opts.debugUrls.size()];
-          for (int i = 0; i < debugUrlsArray.length; ++i) {
-            debugUrlsArray[i] = opts.debugUrls.get(i);
-          }
-          opBuilder.setAttr("debug_urls", debugUrlsArray);
-        }
-        if (opts.circularBufferSize != null) {
-          opBuilder.setAttr("circular_buffer_size", opts.circularBufferSize);
-        }
-        if (opts.tfdbgRunId != null) {
-          opBuilder.setAttr("tfdbg_run_id", opts.tfdbgRunId);
-        }
-      }
-    }
-    return new DebugIdentity<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param tfdbgContextId A tfdbg-generated ID for the context that the op belongs to,
-   *   e.g., a concrete compiled tf.function.
-   */
-  public static Options tfdbgContextId(String tfdbgContextId) {
-    return new Options().tfdbgContextId(tfdbgContextId);
-  }
-  
-  /**
-   * @param opName Optional. Name of the op that the debug op is concerned with.
-   *   Used only for single-tensor trace.
-   */
-  public static Options opName(String opName) {
-    return new Options().opName(opName);
-  }
-  
-  /**
-   * @param outputSlot Optional. Output slot index of the tensor that the debug op
-   *   is concerned with. Used only for single-tensor trace.
-   */
-  public static Options outputSlot(Long outputSlot) {
-    return new Options().outputSlot(outputSlot);
-  }
-  
-  /**
-   * @param tensorDebugMode TensorDebugMode enum value. See debug_event.proto for details.
-   */
-  public static Options tensorDebugMode(Long tensorDebugMode) {
-    return new Options().tensorDebugMode(tensorDebugMode);
-  }
-  
-  /**
-   * @param debugUrls List of URLs to debug targets, e.g., file:///foo/tfdbg_dump.
-   */
-  public static Options debugUrls(List<String> debugUrls) {
-    return new Options().debugUrls(debugUrls);
-  }
-  
-  /**
-   * @param circularBufferSize 
-   */
-  public static Options circularBufferSize(Long circularBufferSize) {
-    return new Options().circularBufferSize(circularBufferSize);
-  }
-  
-  /**
-   * @param tfdbgRunId 
-   */
-  public static Options tfdbgRunId(String tfdbgRunId) {
-    return new Options().tfdbgRunId(tfdbgRunId);
-  }
-  
-  /**
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "DebugIdentityV2";
-  
-  private Output<T> output;
-  
-  private DebugIdentity(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/debugging/DebugNanCount.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/debugging/DebugNanCount.java
deleted file mode 100644
index 0c4666b1818..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/debugging/DebugNanCount.java
+++ /dev/null
@@ -1,182 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.debugging;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
-
-/**
- * Debug NaN Value Counter Op.
- * <p>
- * Counts number of NaNs in the input tensor, for debugging.
- */
-public final class DebugNanCount extends RawOp implements Operand<TInt64> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.debugging.DebugNanCount}
-   */
-  public static class Options {
-    
-    /**
-     * @param deviceName 
-     */
-    public Options deviceName(String deviceName) {
-      this.deviceName = deviceName;
-      return this;
-    }
-    
-    /**
-     * @param tensorName Name of the input tensor.
-     */
-    public Options tensorName(String tensorName) {
-      this.tensorName = tensorName;
-      return this;
-    }
-    
-    /**
-     * @param debugUrls List of URLs to debug targets, e.g.,
-     *   file:///foo/tfdbg_dump, grpc:://localhost:11011.
-     */
-    public Options debugUrls(List<String> debugUrls) {
-      this.debugUrls = debugUrls;
-      return this;
-    }
-    
-    /**
-     * @param gatedGrpc  Whether this op will be gated. If any of the debug_urls of this
-     *   debug node is of the grpc:// scheme, when the value of this attribute is set
-     *   to True, the data will not actually be sent via the grpc stream unless this
-     *   debug op has been enabled at the debug_url. If all of the debug_urls of this
-     *   debug node are of the grpc:// scheme and the debug op is enabled at none of
-     *   them, the output will be an empty Tensor.
-     */
-    public Options gatedGrpc(Boolean gatedGrpc) {
-      this.gatedGrpc = gatedGrpc;
-      return this;
-    }
-    
-    private String deviceName;
-    private String tensorName;
-    private List<String> debugUrls;
-    private Boolean gatedGrpc;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new DebugNanCount operation.
-   * 
-   * @param scope current scope
-   * @param input Input tensor, non-Reference type.
-   * @param options carries optional attributes values
-   * @return a new instance of DebugNanCount
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> DebugNanCount create(Scope scope, Operand<T> input, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("DebugNanCount", scope.makeOpName("DebugNanCount"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.deviceName != null) {
-          opBuilder.setAttr("device_name", opts.deviceName);
-        }
-        if (opts.tensorName != null) {
-          opBuilder.setAttr("tensor_name", opts.tensorName);
-        }
-        if (opts.debugUrls != null) {
-          String[] debugUrlsArray = new String[opts.debugUrls.size()];
-          for (int i = 0; i < debugUrlsArray.length; ++i) {
-            debugUrlsArray[i] = opts.debugUrls.get(i);
-          }
-          opBuilder.setAttr("debug_urls", debugUrlsArray);
-        }
-        if (opts.gatedGrpc != null) {
-          opBuilder.setAttr("gated_grpc", opts.gatedGrpc);
-        }
-      }
-    }
-    return new DebugNanCount(opBuilder.build());
-  }
-  
-  /**
-   * @param deviceName 
-   */
-  public static Options deviceName(String deviceName) {
-    return new Options().deviceName(deviceName);
-  }
-  
-  /**
-   * @param tensorName Name of the input tensor.
-   */
-  public static Options tensorName(String tensorName) {
-    return new Options().tensorName(tensorName);
-  }
-  
-  /**
-   * @param debugUrls List of URLs to debug targets, e.g.,
-   *   file:///foo/tfdbg_dump, grpc:://localhost:11011.
-   */
-  public static Options debugUrls(List<String> debugUrls) {
-    return new Options().debugUrls(debugUrls);
-  }
-  
-  /**
-   * @param gatedGrpc  Whether this op will be gated. If any of the debug_urls of this
-   *   debug node is of the grpc:// scheme, when the value of this attribute is set
-   *   to True, the data will not actually be sent via the grpc stream unless this
-   *   debug op has been enabled at the debug_url. If all of the debug_urls of this
-   *   debug node are of the grpc:// scheme and the debug op is enabled at none of
-   *   them, the output will be an empty Tensor.
-   */
-  public static Options gatedGrpc(Boolean gatedGrpc) {
-    return new Options().gatedGrpc(gatedGrpc);
-  }
-  
-  /**
-   */
-  public Output<TInt64> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<TInt64> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "DebugNanCount";
-  
-  private Output<TInt64> output;
-  
-  private DebugNanCount(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/debugging/DebugNumericsSummary.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/debugging/DebugNumericsSummary.java
deleted file mode 100644
index 6e5cd1fb011..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/debugging/DebugNumericsSummary.java
+++ /dev/null
@@ -1,253 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.debugging;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * Debug Numeric Summary V2 Op.
- * <p>
- * Computes a numeric summary of the input tensor. The shape of the output
- * depends on the tensor_debug_mode attribute.
- * This op is used internally by TensorFlow Debugger (tfdbg) v2.
- * 
- * @param <U> data type for {@code output()} output
- */
-public final class DebugNumericsSummary<U extends TNumber> extends RawOp implements Operand<U> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.debugging.DebugNumericsSummary}
-   */
-  public static class Options {
-    
-    /**
-     * @param tensorDebugMode Tensor debug mode: the mode in which the input tensor is summarized
-     *   by the op. See the TensorDebugMode enum in
-     *   tensorflow/core/protobuf/debug_event.proto for details.
-     * <p>
-     * Supported values:
-     *   2 (CURT_HEALTH): Output a float32/64 tensor of shape [2]. The 1st
-     *   element is the tensor_id, if provided, and -1 otherwise. The 2nd
-     *   element is a bit which is set to 1 if the input tensor has an
-     *   infinity or nan value, or zero otherwise.
-     * <p>
-     *   3 (CONCISE_HEALTH): Output a float32/64 tensor of shape [5]. The 1st
-     *   element is the tensor_id, if provided, and -1 otherwise. The
-     *   remaining four slots are the total number of elements, -infs,
-     *   +infs, and nans in the input tensor respectively.
-     * <p>
-     *   4 (FULL_HEALTH): Output a float32/64 tensor of shape [11]. The 1st
-     *   element is the tensor_id, if provided, and -1 otherwise. The 2nd
-     *   element is the device_id, if provided, and -1 otherwise. The 3rd
-     *   element holds the datatype value of the input tensor as according
-     *   to the enumerated type in tensorflow/core/framework/types.proto.
-     *   The remaining elements hold the total number of elements, -infs,
-     *   +infs, nans, negative finite numbers, zeros, and positive finite
-     *   numbers in the input tensor respectively.
-     * <p>
-     *   5 (SHAPE): Output a float32/64 tensor of shape [10]. The 1st
-     *   element is the tensor_id, if provided, and -1 otherwise. The 2nd
-     *   element holds the datatype value of the input tensor as according
-     *   to the enumerated type in tensorflow/core/framework/types.proto.
-     *   The 3rd element holds the rank of the tensor. The 4th element holds
-     *   the number of elements within the tensor. Finally the remaining 6
-     *   elements hold the shape of the tensor. If the rank of the tensor
-     *   is lower than 6, the shape is right padded with zeros. If the rank
-     *   is greater than 6, the head of the shape is truncated.
-     * <p>
-     *   6 (FULL_NUMERICS): Output a float32/64 tensor of shape [22]. The 1st
-     *   element is the tensor_id, if provided, and -1 otherwise. The 2nd
-     *   element is the device_id, if provided, and -1 otherwise. The 3rd
-     *   element holds the datatype value of the input tensor as according
-     *   to the enumerated type in tensorflow/core/framework/types.proto.
-     *   The 4th element holds the rank of the tensor. The 5th to 11th
-     *   elements hold the shape of the tensor. If the rank of the tensor
-     *   is lower than 6, the shape is right padded with zeros. If the rank
-     *   is greater than 6, the head of the shape is truncated. The 12th to
-     *   18th elements hold the number of elements, -infs, +infs, nans,
-     *   denormal floats, negative finite numbers, zeros, and positive
-     *   finite numbers in the input tensor respectively. The final four
-     *   elements hold the min value, max value, mean, and variance of the
-     *   input tensor.
-     * <p>
-     *   8 (REDUCE_INF_NAN_THREE_SLOTS): Output a float32/64 tensor of shape
-     *   [3]. The 1st element is -inf if any elements of the input tensor
-     *   is -inf, or zero otherwise. The 2nd element is +inf if any elements
-     *   of the input tensor is +inf, or zero otherwise.  The 3rd element is
-     *   nan if any element of the input tensor is nan, or zero otherwise.
-     */
-    public Options tensorDebugMode(Long tensorDebugMode) {
-      this.tensorDebugMode = tensorDebugMode;
-      return this;
-    }
-    
-    /**
-     * @param tensorId Optional. An integer identifier for the tensor being summarized by this op.
-     */
-    public Options tensorId(Long tensorId) {
-      this.tensorId = tensorId;
-      return this;
-    }
-    
-    private Long tensorDebugMode;
-    private Long tensorId;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new DebugNumericsSummary operation.
-   * 
-   * @param scope current scope
-   * @param input Input tensor, to be summarized by the op.
-   * @param outputDtype Optional. The type of the output. Can be float32 or float64 (default: float32).
-   * @param options carries optional attributes values
-   * @return a new instance of DebugNumericsSummary
-   */
-  @Endpoint(describeByClass = true)
-  public static <U extends TNumber, T extends TType> DebugNumericsSummary<U> create(Scope scope, Operand<T> input, Class<U> outputDtype, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("DebugNumericSummaryV2", scope.makeOpName("DebugNumericsSummary"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("output_dtype", outputDtype);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.tensorDebugMode != null) {
-          opBuilder.setAttr("tensor_debug_mode", opts.tensorDebugMode);
-        }
-        if (opts.tensorId != null) {
-          opBuilder.setAttr("tensor_id", opts.tensorId);
-        }
-      }
-    }
-    return new DebugNumericsSummary<U>(opBuilder.build());
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new DebugNumericsSummary operation using default output types.
-   * 
-   * @param scope current scope
-   * @param input Input tensor, to be summarized by the op.
-   * @param options carries optional attributes values
-   * @return a new instance of DebugNumericsSummary
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> DebugNumericsSummary<TFloat32> create(Scope scope, Operand<T> input, Options... options) {
-    return create(scope, input, TFloat32.class, options);
-  }
-  
-  /**
-   * @param tensorDebugMode Tensor debug mode: the mode in which the input tensor is summarized
-   *   by the op. See the TensorDebugMode enum in
-   *   tensorflow/core/protobuf/debug_event.proto for details.
-   * <p>
-   * Supported values:
-   *   2 (CURT_HEALTH): Output a float32/64 tensor of shape [2]. The 1st
-   *   element is the tensor_id, if provided, and -1 otherwise. The 2nd
-   *   element is a bit which is set to 1 if the input tensor has an
-   *   infinity or nan value, or zero otherwise.
-   * <p>
-   *   3 (CONCISE_HEALTH): Output a float32/64 tensor of shape [5]. The 1st
-   *   element is the tensor_id, if provided, and -1 otherwise. The
-   *   remaining four slots are the total number of elements, -infs,
-   *   +infs, and nans in the input tensor respectively.
-   * <p>
-   *   4 (FULL_HEALTH): Output a float32/64 tensor of shape [11]. The 1st
-   *   element is the tensor_id, if provided, and -1 otherwise. The 2nd
-   *   element is the device_id, if provided, and -1 otherwise. The 3rd
-   *   element holds the datatype value of the input tensor as according
-   *   to the enumerated type in tensorflow/core/framework/types.proto.
-   *   The remaining elements hold the total number of elements, -infs,
-   *   +infs, nans, negative finite numbers, zeros, and positive finite
-   *   numbers in the input tensor respectively.
-   * <p>
-   *   5 (SHAPE): Output a float32/64 tensor of shape [10]. The 1st
-   *   element is the tensor_id, if provided, and -1 otherwise. The 2nd
-   *   element holds the datatype value of the input tensor as according
-   *   to the enumerated type in tensorflow/core/framework/types.proto.
-   *   The 3rd element holds the rank of the tensor. The 4th element holds
-   *   the number of elements within the tensor. Finally the remaining 6
-   *   elements hold the shape of the tensor. If the rank of the tensor
-   *   is lower than 6, the shape is right padded with zeros. If the rank
-   *   is greater than 6, the head of the shape is truncated.
-   * <p>
-   *   6 (FULL_NUMERICS): Output a float32/64 tensor of shape [22]. The 1st
-   *   element is the tensor_id, if provided, and -1 otherwise. The 2nd
-   *   element is the device_id, if provided, and -1 otherwise. The 3rd
-   *   element holds the datatype value of the input tensor as according
-   *   to the enumerated type in tensorflow/core/framework/types.proto.
-   *   The 4th element holds the rank of the tensor. The 5th to 11th
-   *   elements hold the shape of the tensor. If the rank of the tensor
-   *   is lower than 6, the shape is right padded with zeros. If the rank
-   *   is greater than 6, the head of the shape is truncated. The 12th to
-   *   18th elements hold the number of elements, -infs, +infs, nans,
-   *   denormal floats, negative finite numbers, zeros, and positive
-   *   finite numbers in the input tensor respectively. The final four
-   *   elements hold the min value, max value, mean, and variance of the
-   *   input tensor.
-   * <p>
-   *   8 (REDUCE_INF_NAN_THREE_SLOTS): Output a float32/64 tensor of shape
-   *   [3]. The 1st element is -inf if any elements of the input tensor
-   *   is -inf, or zero otherwise. The 2nd element is +inf if any elements
-   *   of the input tensor is +inf, or zero otherwise.  The 3rd element is
-   *   nan if any element of the input tensor is nan, or zero otherwise.
-   */
-  public static Options tensorDebugMode(Long tensorDebugMode) {
-    return new Options().tensorDebugMode(tensorDebugMode);
-  }
-  
-  /**
-   * @param tensorId Optional. An integer identifier for the tensor being summarized by this op.
-   */
-  public static Options tensorId(Long tensorId) {
-    return new Options().tensorId(tensorId);
-  }
-  
-  /**
-   */
-  public Output<U> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<U> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "DebugNumericSummaryV2";
-  
-  private Output<U> output;
-  
-  private DebugNumericsSummary(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/dtypes/AsString.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/dtypes/AsString.java
deleted file mode 100644
index 17d4f15222b..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/dtypes/AsString.java
+++ /dev/null
@@ -1,206 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.dtypes;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TString;
-import org.tensorflow.types.family.TType;
-
-/**
- * Converts each entry in the given tensor to strings.
- * <p>
- * Supports many numeric types and boolean.
- * <p>
- * For Unicode, see the
- * [https://www.tensorflow.org/tutorials/representation/unicode](Working with Unicode text)
- * tutorial.
- * <p>
- * Examples:
- * <p>
- * >>> tf.strings.as_string([3, 2])
- * <tf.Tensor: shape=(2,), dtype=string, numpy=array([b'3', b'2'], dtype=object)>
- * >>> tf.strings.as_string([3.1415926, 2.71828], precision=2).numpy()
- * array([b'3.14', b'2.72'], dtype=object)
- */
-@Operator(group = "dtypes")
-public final class AsString extends RawOp implements Operand<TString> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.dtypes.AsString}
-   */
-  public static class Options {
-    
-    /**
-     * @param precision The post-decimal precision to use for floating point numbers.
-     * Only used if precision > -1.
-     */
-    public Options precision(Long precision) {
-      this.precision = precision;
-      return this;
-    }
-    
-    /**
-     * @param scientific Use scientific notation for floating point numbers.
-     */
-    public Options scientific(Boolean scientific) {
-      this.scientific = scientific;
-      return this;
-    }
-    
-    /**
-     * @param shortest Use shortest representation (either scientific or standard) for
-     * floating point numbers.
-     */
-    public Options shortest(Boolean shortest) {
-      this.shortest = shortest;
-      return this;
-    }
-    
-    /**
-     * @param width Pad pre-decimal numbers to this width.
-     * Applies to both floating point and integer numbers.
-     * Only used if width > -1.
-     */
-    public Options width(Long width) {
-      this.width = width;
-      return this;
-    }
-    
-    /**
-     * @param fill The value to pad if width > -1.  If empty, pads with spaces.
-     * Another typical value is '0'.  String cannot be longer than 1 character.
-     */
-    public Options fill(String fill) {
-      this.fill = fill;
-      return this;
-    }
-    
-    private Long precision;
-    private Boolean scientific;
-    private Boolean shortest;
-    private Long width;
-    private String fill;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new AsString operation.
-   * 
-   * @param scope current scope
-   * @param input 
-   * @param options carries optional attributes values
-   * @return a new instance of AsString
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> AsString create(Scope scope, Operand<T> input, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("AsString", scope.makeOpName("AsString"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.precision != null) {
-          opBuilder.setAttr("precision", opts.precision);
-        }
-        if (opts.scientific != null) {
-          opBuilder.setAttr("scientific", opts.scientific);
-        }
-        if (opts.shortest != null) {
-          opBuilder.setAttr("shortest", opts.shortest);
-        }
-        if (opts.width != null) {
-          opBuilder.setAttr("width", opts.width);
-        }
-        if (opts.fill != null) {
-          opBuilder.setAttr("fill", opts.fill);
-        }
-      }
-    }
-    return new AsString(opBuilder.build());
-  }
-  
-  /**
-   * @param precision The post-decimal precision to use for floating point numbers.
-   * Only used if precision > -1.
-   */
-  public static Options precision(Long precision) {
-    return new Options().precision(precision);
-  }
-  
-  /**
-   * @param scientific Use scientific notation for floating point numbers.
-   */
-  public static Options scientific(Boolean scientific) {
-    return new Options().scientific(scientific);
-  }
-  
-  /**
-   * @param shortest Use shortest representation (either scientific or standard) for
-   * floating point numbers.
-   */
-  public static Options shortest(Boolean shortest) {
-    return new Options().shortest(shortest);
-  }
-  
-  /**
-   * @param width Pad pre-decimal numbers to this width.
-   * Applies to both floating point and integer numbers.
-   * Only used if width > -1.
-   */
-  public static Options width(Long width) {
-    return new Options().width(width);
-  }
-  
-  /**
-   * @param fill The value to pad if width > -1.  If empty, pads with spaces.
-   * Another typical value is '0'.  String cannot be longer than 1 character.
-   */
-  public static Options fill(String fill) {
-    return new Options().fill(fill);
-  }
-  
-  /**
-   */
-  public Output<TString> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<TString> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "AsString";
-  
-  private Output<TString> output;
-  
-  private AsString(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/dtypes/Cast.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/dtypes/Cast.java
deleted file mode 100644
index 23fdd3d3a42..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/dtypes/Cast.java
+++ /dev/null
@@ -1,110 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.dtypes;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Cast x of type SrcT to y of DstT.
- * 
- * @param <U> data type for {@code y()} output
- */
-@Operator(group = "dtypes")
-public final class Cast<U extends TType> extends RawOp implements Operand<U> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.dtypes.Cast}
-   */
-  public static class Options {
-    
-    /**
-     * @param Truncate 
-     */
-    public Options Truncate(Boolean Truncate) {
-      this.Truncate = Truncate;
-      return this;
-    }
-    
-    private Boolean Truncate;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new Cast operation.
-   * 
-   * @param scope current scope
-   * @param x 
-   * @param DstT 
-   * @param options carries optional attributes values
-   * @return a new instance of Cast
-   */
-  @Endpoint(describeByClass = true)
-  public static <U extends TType, T extends TType> Cast<U> create(Scope scope, Operand<T> x, Class<U> DstT, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Cast", scope.makeOpName("Cast"));
-    opBuilder.addInput(x.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("DstT", DstT);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.Truncate != null) {
-          opBuilder.setAttr("Truncate", opts.Truncate);
-        }
-      }
-    }
-    return new Cast<U>(opBuilder.build());
-  }
-  
-  /**
-   * @param Truncate 
-   */
-  public static Options Truncate(Boolean Truncate) {
-    return new Options().Truncate(Truncate);
-  }
-  
-  /**
-   */
-  public Output<U> y() {
-    return y;
-  }
-  
-  @Override
-  public Output<U> asOutput(Scope scope) {
-    return y;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Cast";
-  
-  private Output<U> y;
-  
-  private Cast(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    y = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/dtypes/Complex.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/dtypes/Complex.java
deleted file mode 100644
index 17f10b67977..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/dtypes/Complex.java
+++ /dev/null
@@ -1,94 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.dtypes;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * Converts two real numbers to a complex number.
- * <p>
- * Given a tensor `real` representing the real part of a complex number, and a
- * tensor `imag` representing the imaginary part of a complex number, this
- * operation returns complex numbers elementwise of the form \\(a + bj\\), where
- * <i>a</i> represents the `real` part and <i>b</i> represents the `imag` part.
- * <p>
- * The input tensors `real` and `imag` must have the same shape.
- * <p>
- * For example:
- * <pre>{@code
- * # tensor 'real' is [2.25, 3.25]
- * # tensor `imag` is [4.75, 5.75]
- * tf.complex(real, imag) ==> [[2.25 + 4.75j], [3.25 + 5.75j]]
- * }</pre>
- * 
- * 
- * @param <U> data type for {@code out()} output
- */
-@Operator(group = "dtypes")
-public final class Complex<U extends TType> extends RawOp implements Operand<U> {
-  
-  /**
-   * Factory method to create a class wrapping a new Complex operation.
-   * 
-   * @param scope current scope
-   * @param real 
-   * @param imag 
-   * @param Tout 
-   * @return a new instance of Complex
-   */
-  @Endpoint(describeByClass = true)
-  public static <U extends TType, T extends TNumber> Complex<U> create(Scope scope, Operand<T> real, Operand<T> imag, Class<U> Tout) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Complex", scope.makeOpName("Complex"));
-    opBuilder.addInput(real.asOutput(scope));
-    opBuilder.addInput(imag.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("Tout", Tout);
-    return new Complex<U>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<U> out() {
-    return out;
-  }
-  
-  @Override
-  public Output<U> asOutput(Scope scope) {
-    return out;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Complex";
-  
-  private Output<U> out;
-  
-  private Complex(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    out = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/dtypes/ToBool.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/dtypes/ToBool.java
deleted file mode 100644
index 693100c20da..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/dtypes/ToBool.java
+++ /dev/null
@@ -1,85 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.dtypes;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TBool;
-import org.tensorflow.types.family.TType;
-
-/**
- * Converts a tensor to a scalar predicate.
- * <p>
- * Converts a tensor to a scalar predicate with the following rules:
- * <p>
- * - For 0D tensors, truthiness is determined by comparing against a "zero"
- *   value. For numerical types it is the obvious zero. For strings it is the
- *   empty string.
- * <p>
- * - For >0D tensors, truthiness is determined by looking at the number of
- *   elements. If has zero elements, then the result is false. Otherwise the
- *   result is true.
- * <p>
- * This matches the behavior of If and While for determining if a tensor counts
- * as true/false for a branch condition.
- */
-public final class ToBool extends RawOp implements Operand<TBool> {
-  
-  /**
-   * Factory method to create a class wrapping a new ToBool operation.
-   * 
-   * @param scope current scope
-   * @param input 
-   * @return a new instance of ToBool
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> ToBool create(Scope scope, Operand<T> input) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ToBool", scope.makeOpName("ToBool"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new ToBool(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<TBool> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<TBool> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ToBool";
-  
-  private Output<TBool> output;
-  
-  private ToBool(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesAggregateStats.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesAggregateStats.java
deleted file mode 100644
index 03ade5c79e1..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesAggregateStats.java
+++ /dev/null
@@ -1,86 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.estimator;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.TInt32;
-
-/**
- * Aggregates the summary of accumulated stats for the batch.
- * <p>
- * The summary stats contains gradients and hessians accumulated for each node, feature dimension id and bucket.
- */
-public final class BoostedTreesAggregateStats extends RawOp implements Operand<TFloat32> {
-  
-  /**
-   * Factory method to create a class wrapping a new BoostedTreesAggregateStats operation.
-   * 
-   * @param scope current scope
-   * @param nodeIds int32; Rank 1 Tensor containing node ids for each example, shape [batch_size].
-   * @param gradients float32; Rank 2 Tensor (shape=[batch_size, logits_dimension]) with gradients for each example.
-   * @param hessians float32; Rank 2 Tensor (shape=[batch_size, hessian_dimension]) with hessians for each example.
-   * @param feature int32; Rank 2 feature Tensors (shape=[batch_size, feature_dimension]).
-   * @param maxSplits int; the maximum number of splits possible in the whole tree.
-   * @param numBuckets int; equals to the maximum possible value of bucketized feature.
-   * @return a new instance of BoostedTreesAggregateStats
-   */
-  @Endpoint(describeByClass = true)
-  public static BoostedTreesAggregateStats create(Scope scope, Operand<TInt32> nodeIds, Operand<TFloat32> gradients, Operand<TFloat32> hessians, Operand<TInt32> feature, Long maxSplits, Long numBuckets) {
-    OperationBuilder opBuilder = scope.env().opBuilder("BoostedTreesAggregateStats", scope.makeOpName("BoostedTreesAggregateStats"));
-    opBuilder.addInput(nodeIds.asOutput(scope));
-    opBuilder.addInput(gradients.asOutput(scope));
-    opBuilder.addInput(hessians.asOutput(scope));
-    opBuilder.addInput(feature.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("max_splits", maxSplits);
-    opBuilder.setAttr("num_buckets", numBuckets);
-    return new BoostedTreesAggregateStats(opBuilder.build());
-  }
-  
-  /**
-   * output Rank 4 Tensor (shape=[splits, feature_dimension, buckets, logits_dimension + hessian_dimension])
-   * containing accumulated stats for each node, feature dimension and bucket.
-   */
-  public Output<TFloat32> statsSummary() {
-    return statsSummary;
-  }
-  
-  @Override
-  public Output<TFloat32> asOutput(Scope scope) {
-    return statsSummary;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "BoostedTreesAggregateStats";
-  
-  private Output<TFloat32> statsSummary;
-  
-  private BoostedTreesAggregateStats(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    statsSummary = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesBucketize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesBucketize.java
deleted file mode 100644
index f13498bd012..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesBucketize.java
+++ /dev/null
@@ -1,87 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.estimator;
-
-import java.util.Arrays;
-import java.util.Iterator;
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.Operands;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.TInt32;
-
-/**
- * Bucketize each feature based on bucket boundaries.
- * <p>
- * An op that returns a list of float tensors, where each tensor represents the
- * bucketized values for a single feature.
- */
-public final class BoostedTreesBucketize extends RawOp implements Iterable<Operand<TInt32>> {
-  
-  /**
-   * Factory method to create a class wrapping a new BoostedTreesBucketize operation.
-   * 
-   * @param scope current scope
-   * @param floatValues float; List of Rank 1 Tensor each containing float values for a single feature.
-   * @param bucketBoundaries float; List of Rank 1 Tensors each containing the bucket boundaries for a single
-   * feature.
-   * @return a new instance of BoostedTreesBucketize
-   */
-  @Endpoint(describeByClass = true)
-  public static BoostedTreesBucketize create(Scope scope, Iterable<Operand<TFloat32>> floatValues, Iterable<Operand<TFloat32>> bucketBoundaries) {
-    OperationBuilder opBuilder = scope.env().opBuilder("BoostedTreesBucketize", scope.makeOpName("BoostedTreesBucketize"));
-    opBuilder.addInputList(Operands.asOutputs(scope, floatValues));
-    opBuilder.addInputList(Operands.asOutputs(scope, bucketBoundaries));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new BoostedTreesBucketize(opBuilder.build());
-  }
-  
-  /**
-   * int; List of Rank 1 Tensors each containing the bucketized values for a single feature.
-   */
-  public List<Output<TInt32>> buckets() {
-    return buckets;
-  }
-  
-  @Override
-  @SuppressWarnings({"rawtypes", "unchecked"})
-  public Iterator<Operand<TInt32>> iterator() {
-    return (Iterator) buckets.iterator();
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "BoostedTreesBucketize";
-  
-  private List<Output<TInt32>> buckets;
-  
-  @SuppressWarnings("unchecked")
-  private BoostedTreesBucketize(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    int bucketsLength = operation.outputListLength("buckets");
-    buckets = Arrays.asList((Output<TInt32>[])operation.outputList(outputIdx, bucketsLength));
-    outputIdx += bucketsLength;
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesCalculateBestFeatureSplit.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesCalculateBestFeatureSplit.java
deleted file mode 100644
index 1e6fe616a4f..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesCalculateBestFeatureSplit.java
+++ /dev/null
@@ -1,179 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.estimator;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.TString;
-
-/**
- * Calculates gains for each feature and returns the best possible split information for the feature.
- * <p>
- * The split information is the best threshold (bucket id), gains and left/right node contributions per node for each feature.
- * <p>
- * It is possible that not all nodes can be split on each feature. Hence, the list of possible nodes can differ between the features. Therefore, we return `node_ids_list` for each feature, containing the list of nodes that this feature can be used to split.
- * <p>
- * In this manner, the output is the best split per features and per node, so that it needs to be combined later to produce the best split for each node (among all possible features).
- * <p>
- * The output shapes are compatible in a way that the first dimension of all tensors are the same and equal to the number of possible split nodes for each feature.
- */
-public final class BoostedTreesCalculateBestFeatureSplit extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.estimator.BoostedTreesCalculateBestFeatureSplit}
-   */
-  public static class Options {
-    
-    /**
-     * @param splitType A string indicating if this Op should perform inequality split or equality split.
-     */
-    public Options splitType(String splitType) {
-      this.splitType = splitType;
-      return this;
-    }
-    
-    private String splitType;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new BoostedTreesCalculateBestFeatureSplit operation.
-   * 
-   * @param scope current scope
-   * @param nodeIdRange A Rank 1 tensor (shape=[2]) to specify the range [first, last) of node ids to process within `stats_summary_list`. The nodes are iterated between the two nodes specified by the tensor, as like `for node_id in range(node_id_range[0], node_id_range[1])` (Note that the last index node_id_range[1] is exclusive).
-   * @param statsSummary A Rank 4 tensor (#shape=[max_splits, feature_dims, bucket, stats_dims]) for accumulated stats summary (gradient/hessian) per node, per dimension, per buckets for each feature.
-   * The first dimension of the tensor is the maximum number of splits, and thus not all elements of it will be used, but only the indexes specified by node_ids will be used.
-   * @param l1 l1 regularization factor on leaf weights, per instance based.
-   * @param l2 l2 regularization factor on leaf weights, per instance based.
-   * @param treeComplexity adjustment to the gain, per leaf based.
-   * @param minNodeWeight minimum avg of hessians in a node before required for the node to be considered for splitting.
-   * @param logitsDimension The dimension of logit, i.e., number of classes.
-   * @param options carries optional attributes values
-   * @return a new instance of BoostedTreesCalculateBestFeatureSplit
-   */
-  @Endpoint(describeByClass = true)
-  public static BoostedTreesCalculateBestFeatureSplit create(Scope scope, Operand<TInt32> nodeIdRange, Operand<TFloat32> statsSummary, Operand<TFloat32> l1, Operand<TFloat32> l2, Operand<TFloat32> treeComplexity, Operand<TFloat32> minNodeWeight, Long logitsDimension, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("BoostedTreesCalculateBestFeatureSplit", scope.makeOpName("BoostedTreesCalculateBestFeatureSplit"));
-    opBuilder.addInput(nodeIdRange.asOutput(scope));
-    opBuilder.addInput(statsSummary.asOutput(scope));
-    opBuilder.addInput(l1.asOutput(scope));
-    opBuilder.addInput(l2.asOutput(scope));
-    opBuilder.addInput(treeComplexity.asOutput(scope));
-    opBuilder.addInput(minNodeWeight.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("logits_dimension", logitsDimension);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.splitType != null) {
-          opBuilder.setAttr("split_type", opts.splitType);
-        }
-      }
-    }
-    return new BoostedTreesCalculateBestFeatureSplit(opBuilder.build());
-  }
-  
-  /**
-   * @param splitType A string indicating if this Op should perform inequality split or equality split.
-   */
-  public static Options splitType(String splitType) {
-    return new Options().splitType(splitType);
-  }
-  
-  /**
-   * A Rank 1 tensors indicating possible split node ids for each feature. The length of the list is num_features, but each tensor has different size as each feature provides different possible nodes. See above for details like shapes and sizes.
-   */
-  public Output<TInt32> nodeIds() {
-    return nodeIds;
-  }
-  
-  /**
-   * A Rank 1 tensors indicating the best gains for each feature to split for certain nodes. See above for details like shapes and sizes.
-   */
-  public Output<TFloat32> gains() {
-    return gains;
-  }
-  
-  /**
-   * A Rank 1 tensors indicating the best feature dimension for each feature to split for certain nodes if the feature is multi-dimension. See above for details like shapes and sizes.
-   */
-  public Output<TInt32> featureDimensions() {
-    return featureDimensions;
-  }
-  
-  /**
-   * A Rank 1 tensors indicating the bucket id to compare with (as a threshold) for split in each node. See above for details like shapes and sizes.
-   */
-  public Output<TInt32> thresholds() {
-    return thresholds;
-  }
-  
-  /**
-   * A Rank 2 tensors indicating the contribution of the left nodes when branching from parent nodes (given by the tensor element in the output node_ids_list) to the left direction by the given threshold for each feature. This value will be used to make the left node value by adding to the parent node value. Second dimension size is 1 for 1-dimensional logits, but would be larger for multi-class problems. See above for details like shapes and sizes.
-   */
-  public Output<TFloat32> leftNodeContribs() {
-    return leftNodeContribs;
-  }
-  
-  /**
-   * A Rank 2 tensors, with the same shape/conditions as left_node_contribs_list, but just that the value is for the right node.
-   */
-  public Output<TFloat32> rightNodeContribs() {
-    return rightNodeContribs;
-  }
-  
-  /**
-   * A Rank 1 tensors indicating the which direction to go if data is missing. See above for details like shapes and sizes.
-   * Inequality with default left returns 0, inequality with default right returns 1, equality with default right returns 2.
-   */
-  public Output<TString> splitWithDefaultDirections() {
-    return splitWithDefaultDirections;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "BoostedTreesCalculateBestFeatureSplit";
-  
-  private Output<TInt32> nodeIds;
-  private Output<TFloat32> gains;
-  private Output<TInt32> featureDimensions;
-  private Output<TInt32> thresholds;
-  private Output<TFloat32> leftNodeContribs;
-  private Output<TFloat32> rightNodeContribs;
-  private Output<TString> splitWithDefaultDirections;
-  
-  private BoostedTreesCalculateBestFeatureSplit(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    nodeIds = operation.output(outputIdx++);
-    gains = operation.output(outputIdx++);
-    featureDimensions = operation.output(outputIdx++);
-    thresholds = operation.output(outputIdx++);
-    leftNodeContribs = operation.output(outputIdx++);
-    rightNodeContribs = operation.output(outputIdx++);
-    splitWithDefaultDirections = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesCalculateBestFeatureSplitV2.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesCalculateBestFeatureSplitV2.java
deleted file mode 100644
index 85d6580d293..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesCalculateBestFeatureSplitV2.java
+++ /dev/null
@@ -1,159 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.estimator;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.Operands;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.TString;
-
-/**
- * Calculates gains for each feature and returns the best possible split information for each node. However, if no split is found, then no split information is returned for that node.
- * <p>
- * The split information is the best threshold (bucket id), gains and left/right node contributions per node for each feature.
- * <p>
- * It is possible that not all nodes can be split on each feature. Hence, the list of possible nodes can differ between the features. Therefore, we return `node_ids_list` for each feature, containing the list of nodes that this feature can be used to split.
- * <p>
- * In this manner, the output is the best split per features and per node, so that it needs to be combined later to produce the best split for each node (among all possible features).
- * <p>
- * The output shapes are compatible in a way that the first dimension of all tensors are the same and equal to the number of possible split nodes for each feature.
- */
-public final class BoostedTreesCalculateBestFeatureSplitV2 extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new BoostedTreesCalculateBestFeatureSplitV2 operation.
-   * 
-   * @param scope current scope
-   * @param nodeIdRange A Rank 1 tensor (shape=[2]) to specify the range [first, last) of node ids to process within `stats_summary_list`. The nodes are iterated between the two nodes specified by the tensor, as like `for node_id in range(node_id_range[0], node_id_range[1])` (Note that the last index node_id_range[1] is exclusive).
-   * @param statsSummariesList A list of Rank 4 tensor (#shape=[max_splits, feature_dims, bucket, stats_dims]) for accumulated stats summary (gradient/hessian) per node, per dimension, per buckets for each feature.
-   * The first dimension of the tensor is the maximum number of splits, and thus not all elements of it will be used, but only the indexes specified by node_ids will be used.
-   * @param splitTypes A Rank 1 tensor indicating if this Op should perform inequality split or equality split per feature.
-   * @param candidateFeatureIds Rank 1 tensor with ids for each feature. This is the real id of the feature.
-   * @param l1 l1 regularization factor on leaf weights, per instance based.
-   * @param l2 l2 regularization factor on leaf weights, per instance based.
-   * @param treeComplexity adjustment to the gain, per leaf based.
-   * @param minNodeWeight minimum avg of hessians in a node before required for the node to be considered for splitting.
-   * @param logitsDimension The dimension of logit, i.e., number of classes.
-   * @return a new instance of BoostedTreesCalculateBestFeatureSplitV2
-   */
-  @Endpoint(describeByClass = true)
-  public static BoostedTreesCalculateBestFeatureSplitV2 create(Scope scope, Operand<TInt32> nodeIdRange, Iterable<Operand<TFloat32>> statsSummariesList, Operand<TString> splitTypes, Operand<TInt32> candidateFeatureIds, Operand<TFloat32> l1, Operand<TFloat32> l2, Operand<TFloat32> treeComplexity, Operand<TFloat32> minNodeWeight, Long logitsDimension) {
-    OperationBuilder opBuilder = scope.env().opBuilder("BoostedTreesCalculateBestFeatureSplitV2", scope.makeOpName("BoostedTreesCalculateBestFeatureSplitV2"));
-    opBuilder.addInput(nodeIdRange.asOutput(scope));
-    opBuilder.addInputList(Operands.asOutputs(scope, statsSummariesList));
-    opBuilder.addInput(splitTypes.asOutput(scope));
-    opBuilder.addInput(candidateFeatureIds.asOutput(scope));
-    opBuilder.addInput(l1.asOutput(scope));
-    opBuilder.addInput(l2.asOutput(scope));
-    opBuilder.addInput(treeComplexity.asOutput(scope));
-    opBuilder.addInput(minNodeWeight.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("logits_dimension", logitsDimension);
-    return new BoostedTreesCalculateBestFeatureSplitV2(opBuilder.build());
-  }
-  
-  /**
-   * A Rank 1 tensors indicating possible split node ids for each feature. The length of the list is num_features, but each tensor has different size as each feature provides different possible nodes. See above for details like shapes and sizes.
-   */
-  public Output<TInt32> nodeIds() {
-    return nodeIds;
-  }
-  
-  /**
-   * A Rank 1 tensor indicating the best gains for each feature to split for certain nodes. See above for details like shapes and sizes.
-   */
-  public Output<TFloat32> gains() {
-    return gains;
-  }
-  
-  /**
-   * A Rank 1 tensors indicating the best feature id for each node. See above for details like shapes and sizes.
-   */
-  public Output<TInt32> featureIds() {
-    return featureIds;
-  }
-  
-  /**
-   * A Rank 1 tensors indicating the best feature dimension for each feature to split for certain nodes if the feature is multi-dimension. See above for details like shapes and sizes.
-   */
-  public Output<TInt32> featureDimensions() {
-    return featureDimensions;
-  }
-  
-  /**
-   * A Rank 1 tensors indicating the bucket id to compare with (as a threshold) for split in each node. See above for details like shapes and sizes.
-   */
-  public Output<TInt32> thresholds() {
-    return thresholds;
-  }
-  
-  /**
-   * A Rank 2 tensors indicating the contribution of the left nodes when branching from parent nodes (given by the tensor element in the output node_ids_list) to the left direction by the given threshold for each feature. This value will be used to make the left node value by adding to the parent node value. Second dimension size is 1 for 1-dimensional logits, but would be larger for multi-class problems. See above for details like shapes and sizes.
-   */
-  public Output<TFloat32> leftNodeContribs() {
-    return leftNodeContribs;
-  }
-  
-  /**
-   * A Rank 2 tensors, with the same shape/conditions as left_node_contribs_list, but just that the value is for the right node.
-   */
-  public Output<TFloat32> rightNodeContribs() {
-    return rightNodeContribs;
-  }
-  
-  /**
-   * A Rank 1 tensors indicating the which direction to go if data is missing. See above for details like shapes and sizes.
-   * Inequality with default left returns 0, inequality with default right returns 1, equality with default right returns 2.
-   */
-  public Output<TString> splitWithDefaultDirections() {
-    return splitWithDefaultDirections;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "BoostedTreesCalculateBestFeatureSplitV2";
-  
-  private Output<TInt32> nodeIds;
-  private Output<TFloat32> gains;
-  private Output<TInt32> featureIds;
-  private Output<TInt32> featureDimensions;
-  private Output<TInt32> thresholds;
-  private Output<TFloat32> leftNodeContribs;
-  private Output<TFloat32> rightNodeContribs;
-  private Output<TString> splitWithDefaultDirections;
-  
-  private BoostedTreesCalculateBestFeatureSplitV2(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    nodeIds = operation.output(outputIdx++);
-    gains = operation.output(outputIdx++);
-    featureIds = operation.output(outputIdx++);
-    featureDimensions = operation.output(outputIdx++);
-    thresholds = operation.output(outputIdx++);
-    leftNodeContribs = operation.output(outputIdx++);
-    rightNodeContribs = operation.output(outputIdx++);
-    splitWithDefaultDirections = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesCalculateBestGainsPerFeature.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesCalculateBestGainsPerFeature.java
deleted file mode 100644
index 5b4efb8d789..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesCalculateBestGainsPerFeature.java
+++ /dev/null
@@ -1,139 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.estimator;
-
-import java.util.Arrays;
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.Operands;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.TInt32;
-
-/**
- * Calculates gains for each feature and returns the best possible split information for the feature.
- * <p>
- * The split information is the best threshold (bucket id), gains and left/right node contributions per node for each feature.
- * <p>
- * It is possible that not all nodes can be split on each feature. Hence, the list of possible nodes can differ between the features. Therefore, we return `node_ids_list` for each feature, containing the list of nodes that this feature can be used to split.
- * <p>
- * In this manner, the output is the best split per features and per node, so that it needs to be combined later to produce the best split for each node (among all possible features).
- * <p>
- * The length of output lists are all of the same length, `num_features`.
- * The output shapes are compatible in a way that the first dimension of all tensors of all lists are the same and equal to the number of possible split nodes for each feature.
- */
-public final class BoostedTreesCalculateBestGainsPerFeature extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new BoostedTreesCalculateBestGainsPerFeature operation.
-   * 
-   * @param scope current scope
-   * @param nodeIdRange A Rank 1 tensor (shape=[2]) to specify the range [first, last) of node ids to process within `stats_summary_list`. The nodes are iterated between the two nodes specified by the tensor, as like `for node_id in range(node_id_range[0], node_id_range[1])` (Note that the last index node_id_range[1] is exclusive).
-   * @param statsSummaryList A list of Rank 3 tensor (#shape=[max_splits, bucket, 2]) for accumulated stats summary (gradient/hessian) per node per buckets for each feature. The first dimension of the tensor is the maximum number of splits, and thus not all elements of it will be used, but only the indexes specified by node_ids will be used.
-   * @param l1 l1 regularization factor on leaf weights, per instance based.
-   * @param l2 l2 regularization factor on leaf weights, per instance based.
-   * @param treeComplexity adjustment to the gain, per leaf based.
-   * @param minNodeWeight minimum avg of hessians in a node before required for the node to be considered for splitting.
-   * @param maxSplits the number of nodes that can be split in the whole tree. Used as a dimension of output tensors.
-   * @return a new instance of BoostedTreesCalculateBestGainsPerFeature
-   */
-  @Endpoint(describeByClass = true)
-  public static BoostedTreesCalculateBestGainsPerFeature create(Scope scope, Operand<TInt32> nodeIdRange, Iterable<Operand<TFloat32>> statsSummaryList, Operand<TFloat32> l1, Operand<TFloat32> l2, Operand<TFloat32> treeComplexity, Operand<TFloat32> minNodeWeight, Long maxSplits) {
-    OperationBuilder opBuilder = scope.env().opBuilder("BoostedTreesCalculateBestGainsPerFeature", scope.makeOpName("BoostedTreesCalculateBestGainsPerFeature"));
-    opBuilder.addInput(nodeIdRange.asOutput(scope));
-    opBuilder.addInputList(Operands.asOutputs(scope, statsSummaryList));
-    opBuilder.addInput(l1.asOutput(scope));
-    opBuilder.addInput(l2.asOutput(scope));
-    opBuilder.addInput(treeComplexity.asOutput(scope));
-    opBuilder.addInput(minNodeWeight.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("max_splits", maxSplits);
-    return new BoostedTreesCalculateBestGainsPerFeature(opBuilder.build());
-  }
-  
-  /**
-   * An output list of Rank 1 tensors indicating possible split node ids for each feature. The length of the list is num_features, but each tensor has different size as each feature provides different possible nodes. See above for details like shapes and sizes.
-   */
-  public List<Output<TInt32>> nodeIdsList() {
-    return nodeIdsList;
-  }
-  
-  /**
-   * An output list of Rank 1 tensors indicating the best gains for each feature to split for certain nodes. See above for details like shapes and sizes.
-   */
-  public List<Output<TFloat32>> gainsList() {
-    return gainsList;
-  }
-  
-  /**
-   * An output list of Rank 1 tensors indicating the bucket id to compare with (as a threshold) for split in each node. See above for details like shapes and sizes.
-   */
-  public List<Output<TInt32>> thresholdsList() {
-    return thresholdsList;
-  }
-  
-  /**
-   * A list of Rank 2 tensors indicating the contribution of the left nodes when branching from parent nodes (given by the tensor element in the output node_ids_list) to the left direction by the given threshold for each feature. This value will be used to make the left node value by adding to the parent node value. Second dimension size is 1 for 1-dimensional logits, but would be larger for multi-class problems. See above for details like shapes and sizes.
-   */
-  public List<Output<TFloat32>> leftNodeContribsList() {
-    return leftNodeContribsList;
-  }
-  
-  /**
-   * A list of Rank 2 tensors, with the same shape/conditions as left_node_contribs_list, but just that the value is for the right node.
-   */
-  public List<Output<TFloat32>> rightNodeContribsList() {
-    return rightNodeContribsList;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "BoostedTreesCalculateBestGainsPerFeature";
-  
-  private List<Output<TInt32>> nodeIdsList;
-  private List<Output<TFloat32>> gainsList;
-  private List<Output<TInt32>> thresholdsList;
-  private List<Output<TFloat32>> leftNodeContribsList;
-  private List<Output<TFloat32>> rightNodeContribsList;
-  
-  @SuppressWarnings("unchecked")
-  private BoostedTreesCalculateBestGainsPerFeature(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    int nodeIdsListLength = operation.outputListLength("node_ids_list");
-    nodeIdsList = Arrays.asList((Output<TInt32>[])operation.outputList(outputIdx, nodeIdsListLength));
-    outputIdx += nodeIdsListLength;
-    int gainsListLength = operation.outputListLength("gains_list");
-    gainsList = Arrays.asList((Output<TFloat32>[])operation.outputList(outputIdx, gainsListLength));
-    outputIdx += gainsListLength;
-    int thresholdsListLength = operation.outputListLength("thresholds_list");
-    thresholdsList = Arrays.asList((Output<TInt32>[])operation.outputList(outputIdx, thresholdsListLength));
-    outputIdx += thresholdsListLength;
-    int leftNodeContribsListLength = operation.outputListLength("left_node_contribs_list");
-    leftNodeContribsList = Arrays.asList((Output<TFloat32>[])operation.outputList(outputIdx, leftNodeContribsListLength));
-    outputIdx += leftNodeContribsListLength;
-    int rightNodeContribsListLength = operation.outputListLength("right_node_contribs_list");
-    rightNodeContribsList = Arrays.asList((Output<TFloat32>[])operation.outputList(outputIdx, rightNodeContribsListLength));
-    outputIdx += rightNodeContribsListLength;
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesCenterBias.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesCenterBias.java
deleted file mode 100644
index 9bea51c509c..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesCenterBias.java
+++ /dev/null
@@ -1,81 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.estimator;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TBool;
-import org.tensorflow.types.TFloat32;
-
-/**
- * Calculates the prior from the training data (the bias) and fills in the first node with the logits' prior. Returns a boolean indicating whether to continue centering.
- */
-public final class BoostedTreesCenterBias extends RawOp implements Operand<TBool> {
-  
-  /**
-   * Factory method to create a class wrapping a new BoostedTreesCenterBias operation.
-   * 
-   * @param scope current scope
-   * @param treeEnsembleHandle Handle to the tree ensemble.
-   * @param meanGradients A tensor with shape=[logits_dimension] with mean of gradients for a first node.
-   * @param meanHessians A tensor with shape=[logits_dimension] mean of hessians for a first node.
-   * @param l1 l1 regularization factor on leaf weights, per instance based.
-   * @param l2 l2 regularization factor on leaf weights, per instance based.
-   * @return a new instance of BoostedTreesCenterBias
-   */
-  @Endpoint(describeByClass = true)
-  public static BoostedTreesCenterBias create(Scope scope, Operand<?> treeEnsembleHandle, Operand<TFloat32> meanGradients, Operand<TFloat32> meanHessians, Operand<TFloat32> l1, Operand<TFloat32> l2) {
-    OperationBuilder opBuilder = scope.env().opBuilder("BoostedTreesCenterBias", scope.makeOpName("BoostedTreesCenterBias"));
-    opBuilder.addInput(treeEnsembleHandle.asOutput(scope));
-    opBuilder.addInput(meanGradients.asOutput(scope));
-    opBuilder.addInput(meanHessians.asOutput(scope));
-    opBuilder.addInput(l1.asOutput(scope));
-    opBuilder.addInput(l2.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new BoostedTreesCenterBias(opBuilder.build());
-  }
-  
-  /**
-   * Bool, whether to continue bias centering.
-   */
-  public Output<TBool> continueCentering() {
-    return continueCentering;
-  }
-  
-  @Override
-  public Output<TBool> asOutput(Scope scope) {
-    return continueCentering;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "BoostedTreesCenterBias";
-  
-  private Output<TBool> continueCentering;
-  
-  private BoostedTreesCenterBias(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    continueCentering = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesCreateEnsemble.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesCreateEnsemble.java
deleted file mode 100644
index b7d2ff14a67..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesCreateEnsemble.java
+++ /dev/null
@@ -1,60 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.estimator;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.TString;
-
-/**
- * Creates a tree ensemble model and returns a handle to it.
- */
-public final class BoostedTreesCreateEnsemble extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new BoostedTreesCreateEnsemble operation.
-   * 
-   * @param scope current scope
-   * @param treeEnsembleHandle Handle to the tree ensemble resource to be created.
-   * @param stampToken Token to use as the initial value of the resource stamp.
-   * @param treeEnsembleSerialized Serialized proto of the tree ensemble.
-   * @return a new instance of BoostedTreesCreateEnsemble
-   */
-  @Endpoint(describeByClass = true)
-  public static BoostedTreesCreateEnsemble create(Scope scope, Operand<?> treeEnsembleHandle, Operand<TInt64> stampToken, Operand<TString> treeEnsembleSerialized) {
-    OperationBuilder opBuilder = scope.env().opBuilder("BoostedTreesCreateEnsemble", scope.makeOpName("BoostedTreesCreateEnsemble"));
-    opBuilder.addInput(treeEnsembleHandle.asOutput(scope));
-    opBuilder.addInput(stampToken.asOutput(scope));
-    opBuilder.addInput(treeEnsembleSerialized.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new BoostedTreesCreateEnsemble(opBuilder.build());
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "BoostedTreesCreateEnsemble";
-  
-  private BoostedTreesCreateEnsemble(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesCreateQuantileStreamResource.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesCreateQuantileStreamResource.java
deleted file mode 100644
index 698f8356d53..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesCreateQuantileStreamResource.java
+++ /dev/null
@@ -1,94 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.estimator;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.TInt64;
-
-/**
- * Create the Resource for Quantile Streams.
- */
-public final class BoostedTreesCreateQuantileStreamResource extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.estimator.BoostedTreesCreateQuantileStreamResource}
-   */
-  public static class Options {
-    
-    /**
-     * @param maxElements int; The maximum number of data points that can be fed to the stream.
-     */
-    public Options maxElements(Long maxElements) {
-      this.maxElements = maxElements;
-      return this;
-    }
-    
-    private Long maxElements;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new BoostedTreesCreateQuantileStreamResource operation.
-   * 
-   * @param scope current scope
-   * @param quantileStreamResourceHandle resource; Handle to quantile stream resource.
-   * @param epsilon float; The required approximation error of the stream resource.
-   * @param numStreams int; The number of streams managed by the resource that shares the same epsilon.
-   * @param options carries optional attributes values
-   * @return a new instance of BoostedTreesCreateQuantileStreamResource
-   */
-  @Endpoint(describeByClass = true)
-  public static BoostedTreesCreateQuantileStreamResource create(Scope scope, Operand<?> quantileStreamResourceHandle, Operand<TFloat32> epsilon, Operand<TInt64> numStreams, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("BoostedTreesCreateQuantileStreamResource", scope.makeOpName("BoostedTreesCreateQuantileStreamResource"));
-    opBuilder.addInput(quantileStreamResourceHandle.asOutput(scope));
-    opBuilder.addInput(epsilon.asOutput(scope));
-    opBuilder.addInput(numStreams.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.maxElements != null) {
-          opBuilder.setAttr("max_elements", opts.maxElements);
-        }
-      }
-    }
-    return new BoostedTreesCreateQuantileStreamResource(opBuilder.build());
-  }
-  
-  /**
-   * @param maxElements int; The maximum number of data points that can be fed to the stream.
-   */
-  public static Options maxElements(Long maxElements) {
-    return new Options().maxElements(maxElements);
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "BoostedTreesCreateQuantileStreamResource";
-  
-  private BoostedTreesCreateQuantileStreamResource(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesDeserializeEnsemble.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesDeserializeEnsemble.java
deleted file mode 100644
index a9b02360384..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesDeserializeEnsemble.java
+++ /dev/null
@@ -1,62 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.estimator;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.TString;
-
-/**
- * Deserializes a serialized tree ensemble config and replaces current tree
- * <p>
- * ensemble.
- */
-public final class BoostedTreesDeserializeEnsemble extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new BoostedTreesDeserializeEnsemble operation.
-   * 
-   * @param scope current scope
-   * @param treeEnsembleHandle Handle to the tree ensemble.
-   * @param stampToken Token to use as the new value of the resource stamp.
-   * @param treeEnsembleSerialized Serialized proto of the ensemble.
-   * @return a new instance of BoostedTreesDeserializeEnsemble
-   */
-  @Endpoint(describeByClass = true)
-  public static BoostedTreesDeserializeEnsemble create(Scope scope, Operand<?> treeEnsembleHandle, Operand<TInt64> stampToken, Operand<TString> treeEnsembleSerialized) {
-    OperationBuilder opBuilder = scope.env().opBuilder("BoostedTreesDeserializeEnsemble", scope.makeOpName("BoostedTreesDeserializeEnsemble"));
-    opBuilder.addInput(treeEnsembleHandle.asOutput(scope));
-    opBuilder.addInput(stampToken.asOutput(scope));
-    opBuilder.addInput(treeEnsembleSerialized.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new BoostedTreesDeserializeEnsemble(opBuilder.build());
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "BoostedTreesDeserializeEnsemble";
-  
-  private BoostedTreesDeserializeEnsemble(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesEnsembleResourceHandleOp.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesEnsembleResourceHandleOp.java
deleted file mode 100644
index e2067f08a5b..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesEnsembleResourceHandleOp.java
+++ /dev/null
@@ -1,123 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.estimator;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Creates a handle to a BoostedTreesEnsembleResource
- */
-public final class BoostedTreesEnsembleResourceHandleOp extends RawOp implements Operand<TType> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.estimator.BoostedTreesEnsembleResourceHandleOp}
-   */
-  public static class Options {
-    
-    /**
-     * @param container 
-     */
-    public Options container(String container) {
-      this.container = container;
-      return this;
-    }
-    
-    /**
-     * @param sharedName 
-     */
-    public Options sharedName(String sharedName) {
-      this.sharedName = sharedName;
-      return this;
-    }
-    
-    private String container;
-    private String sharedName;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new BoostedTreesEnsembleResourceHandleOp operation.
-   * 
-   * @param scope current scope
-   * @param options carries optional attributes values
-   * @return a new instance of BoostedTreesEnsembleResourceHandleOp
-   */
-  @Endpoint(describeByClass = true)
-  public static BoostedTreesEnsembleResourceHandleOp create(Scope scope, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("BoostedTreesEnsembleResourceHandleOp", scope.makeOpName("BoostedTreesEnsembleResourceHandleOp"));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.container != null) {
-          opBuilder.setAttr("container", opts.container);
-        }
-        if (opts.sharedName != null) {
-          opBuilder.setAttr("shared_name", opts.sharedName);
-        }
-      }
-    }
-    return new BoostedTreesEnsembleResourceHandleOp(opBuilder.build());
-  }
-  
-  /**
-   * @param container 
-   */
-  public static Options container(String container) {
-    return new Options().container(container);
-  }
-  
-  /**
-   * @param sharedName 
-   */
-  public static Options sharedName(String sharedName) {
-    return new Options().sharedName(sharedName);
-  }
-  
-  /**
-   */
-  public Output<?> resource() {
-    return resource;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) resource;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "BoostedTreesEnsembleResourceHandleOp";
-  
-  private Output<?> resource;
-  
-  private BoostedTreesEnsembleResourceHandleOp(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    resource = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesExampleDebugOutputs.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesExampleDebugOutputs.java
deleted file mode 100644
index 058430e3315..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesExampleDebugOutputs.java
+++ /dev/null
@@ -1,84 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.estimator;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.Operands;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.TString;
-
-/**
- * Debugging/model interpretability outputs for each example.
- * <p>
- * It traverses all the trees and computes debug metrics for individual examples,
- * such as getting split feature ids and logits after each split along the decision
- * path used to compute directional feature contributions.
- */
-public final class BoostedTreesExampleDebugOutputs extends RawOp implements Operand<TString> {
-  
-  /**
-   * Factory method to create a class wrapping a new BoostedTreesExampleDebugOutputs operation.
-   * 
-   * @param scope current scope
-   * @param treeEnsembleHandle 
-   * @param bucketizedFeatures A list of rank 1 Tensors containing bucket id for each
-   * feature.
-   * @param logitsDimension scalar, dimension of the logits, to be used for constructing the protos in
-   * examples_debug_outputs_serialized.
-   * @return a new instance of BoostedTreesExampleDebugOutputs
-   */
-  @Endpoint(describeByClass = true)
-  public static BoostedTreesExampleDebugOutputs create(Scope scope, Operand<?> treeEnsembleHandle, Iterable<Operand<TInt32>> bucketizedFeatures, Long logitsDimension) {
-    OperationBuilder opBuilder = scope.env().opBuilder("BoostedTreesExampleDebugOutputs", scope.makeOpName("BoostedTreesExampleDebugOutputs"));
-    opBuilder.addInput(treeEnsembleHandle.asOutput(scope));
-    opBuilder.addInputList(Operands.asOutputs(scope, bucketizedFeatures));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("logits_dimension", logitsDimension);
-    return new BoostedTreesExampleDebugOutputs(opBuilder.build());
-  }
-  
-  /**
-   * Output rank 1 Tensor containing a proto serialized as a string for each example.
-   */
-  public Output<TString> examplesDebugOutputsSerialized() {
-    return examplesDebugOutputsSerialized;
-  }
-  
-  @Override
-  public Output<TString> asOutput(Scope scope) {
-    return examplesDebugOutputsSerialized;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "BoostedTreesExampleDebugOutputs";
-  
-  private Output<TString> examplesDebugOutputsSerialized;
-  
-  private BoostedTreesExampleDebugOutputs(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    examplesDebugOutputsSerialized = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesFlushQuantileSummaries.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesFlushQuantileSummaries.java
deleted file mode 100644
index f3e6381a923..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesFlushQuantileSummaries.java
+++ /dev/null
@@ -1,84 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.estimator;
-
-import java.util.Arrays;
-import java.util.Iterator;
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-
-/**
- * Flush the quantile summaries from each quantile stream resource.
- * <p>
- * An op that outputs a list of quantile summaries of a quantile stream resource.
- * Each summary Tensor is rank 2, containing summaries (value, weight, min_rank,
- * max_rank) for a single feature.
- */
-public final class BoostedTreesFlushQuantileSummaries extends RawOp implements Iterable<Operand<TFloat32>> {
-  
-  /**
-   * Factory method to create a class wrapping a new BoostedTreesFlushQuantileSummaries operation.
-   * 
-   * @param scope current scope
-   * @param quantileStreamResourceHandle resource handle referring to a QuantileStreamResource.
-   * @param numFeatures 
-   * @return a new instance of BoostedTreesFlushQuantileSummaries
-   */
-  @Endpoint(describeByClass = true)
-  public static BoostedTreesFlushQuantileSummaries create(Scope scope, Operand<?> quantileStreamResourceHandle, Long numFeatures) {
-    OperationBuilder opBuilder = scope.env().opBuilder("BoostedTreesFlushQuantileSummaries", scope.makeOpName("BoostedTreesFlushQuantileSummaries"));
-    opBuilder.addInput(quantileStreamResourceHandle.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("num_features", numFeatures);
-    return new BoostedTreesFlushQuantileSummaries(opBuilder.build());
-  }
-  
-  /**
-   */
-  public List<Output<TFloat32>> summaries() {
-    return summaries;
-  }
-  
-  @Override
-  @SuppressWarnings({"rawtypes", "unchecked"})
-  public Iterator<Operand<TFloat32>> iterator() {
-    return (Iterator) summaries.iterator();
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "BoostedTreesFlushQuantileSummaries";
-  
-  private List<Output<TFloat32>> summaries;
-  
-  @SuppressWarnings("unchecked")
-  private BoostedTreesFlushQuantileSummaries(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    int summariesLength = operation.outputListLength("summaries");
-    summaries = Arrays.asList((Output<TFloat32>[])operation.outputList(outputIdx, summariesLength));
-    outputIdx += summariesLength;
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesGetEnsembleStates.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesGetEnsembleStates.java
deleted file mode 100644
index deba0d48ee9..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesGetEnsembleStates.java
+++ /dev/null
@@ -1,105 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.estimator;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.TInt64;
-
-/**
- * Retrieves the tree ensemble resource stamp token, number of trees and growing statistics.
- */
-public final class BoostedTreesGetEnsembleStates extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new BoostedTreesGetEnsembleStates operation.
-   * 
-   * @param scope current scope
-   * @param treeEnsembleHandle Handle to the tree ensemble.
-   * @return a new instance of BoostedTreesGetEnsembleStates
-   */
-  @Endpoint(describeByClass = true)
-  public static BoostedTreesGetEnsembleStates create(Scope scope, Operand<?> treeEnsembleHandle) {
-    OperationBuilder opBuilder = scope.env().opBuilder("BoostedTreesGetEnsembleStates", scope.makeOpName("BoostedTreesGetEnsembleStates"));
-    opBuilder.addInput(treeEnsembleHandle.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new BoostedTreesGetEnsembleStates(opBuilder.build());
-  }
-  
-  /**
-   * Stamp token of the tree ensemble resource.
-   */
-  public Output<TInt64> stampToken() {
-    return stampToken;
-  }
-  
-  /**
-   * The number of trees in the tree ensemble resource.
-   */
-  public Output<TInt32> numTrees() {
-    return numTrees;
-  }
-  
-  /**
-   * The number of trees that were finished successfully.
-   */
-  public Output<TInt32> numFinalizedTrees() {
-    return numFinalizedTrees;
-  }
-  
-  /**
-   * The number of layers we attempted to build (but not necessarily succeeded).
-   */
-  public Output<TInt32> numAttemptedLayers() {
-    return numAttemptedLayers;
-  }
-  
-  /**
-   * Rank size 2 tensor that contains start and end ids of the nodes in the latest
-   * layer.
-   */
-  public Output<TInt32> lastLayerNodesRange() {
-    return lastLayerNodesRange;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "BoostedTreesGetEnsembleStates";
-  
-  private Output<TInt64> stampToken;
-  private Output<TInt32> numTrees;
-  private Output<TInt32> numFinalizedTrees;
-  private Output<TInt32> numAttemptedLayers;
-  private Output<TInt32> lastLayerNodesRange;
-  
-  private BoostedTreesGetEnsembleStates(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    stampToken = operation.output(outputIdx++);
-    numTrees = operation.output(outputIdx++);
-    numFinalizedTrees = operation.output(outputIdx++);
-    numAttemptedLayers = operation.output(outputIdx++);
-    lastLayerNodesRange = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesMakeQuantileSummaries.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesMakeQuantileSummaries.java
deleted file mode 100644
index cbf32069e02..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesMakeQuantileSummaries.java
+++ /dev/null
@@ -1,88 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.estimator;
-
-import java.util.Arrays;
-import java.util.Iterator;
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.Operands;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-
-/**
- * Makes the summary of quantiles for the batch.
- * <p>
- * An op that takes a list of tensors (one tensor per feature) and outputs the
- * quantile summaries for each tensor.
- */
-public final class BoostedTreesMakeQuantileSummaries extends RawOp implements Iterable<Operand<TFloat32>> {
-  
-  /**
-   * Factory method to create a class wrapping a new BoostedTreesMakeQuantileSummaries operation.
-   * 
-   * @param scope current scope
-   * @param floatValues float; List of Rank 1 Tensors each containing values for a single feature.
-   * @param exampleWeights float; Rank 1 Tensor with weights per instance.
-   * @param epsilon float; The required maximum approximation error.
-   * @return a new instance of BoostedTreesMakeQuantileSummaries
-   */
-  @Endpoint(describeByClass = true)
-  public static BoostedTreesMakeQuantileSummaries create(Scope scope, Iterable<Operand<TFloat32>> floatValues, Operand<TFloat32> exampleWeights, Operand<TFloat32> epsilon) {
-    OperationBuilder opBuilder = scope.env().opBuilder("BoostedTreesMakeQuantileSummaries", scope.makeOpName("BoostedTreesMakeQuantileSummaries"));
-    opBuilder.addInputList(Operands.asOutputs(scope, floatValues));
-    opBuilder.addInput(exampleWeights.asOutput(scope));
-    opBuilder.addInput(epsilon.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new BoostedTreesMakeQuantileSummaries(opBuilder.build());
-  }
-  
-  /**
-   * float; List of Rank 2 Tensors each containing the quantile summary
-   * (value, weight, min_rank, max_rank) of a single feature.
-   */
-  public List<Output<TFloat32>> summaries() {
-    return summaries;
-  }
-  
-  @Override
-  @SuppressWarnings({"rawtypes", "unchecked"})
-  public Iterator<Operand<TFloat32>> iterator() {
-    return (Iterator) summaries.iterator();
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "BoostedTreesMakeQuantileSummaries";
-  
-  private List<Output<TFloat32>> summaries;
-  
-  @SuppressWarnings("unchecked")
-  private BoostedTreesMakeQuantileSummaries(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    int summariesLength = operation.outputListLength("summaries");
-    summaries = Arrays.asList((Output<TFloat32>[])operation.outputList(outputIdx, summariesLength));
-    outputIdx += summariesLength;
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesMakeStatsSummary.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesMakeStatsSummary.java
deleted file mode 100644
index 6b73fd4903d..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesMakeStatsSummary.java
+++ /dev/null
@@ -1,86 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.estimator;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.Operands;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.TInt32;
-
-/**
- * Makes the summary of accumulated stats for the batch.
- * <p>
- * The summary stats contains gradients and hessians accumulated into the corresponding node and bucket for each example.
- */
-public final class BoostedTreesMakeStatsSummary extends RawOp implements Operand<TFloat32> {
-  
-  /**
-   * Factory method to create a class wrapping a new BoostedTreesMakeStatsSummary operation.
-   * 
-   * @param scope current scope
-   * @param nodeIds int32 Rank 1 Tensor containing node ids, which each example falls into for the requested layer.
-   * @param gradients float32; Rank 2 Tensor (shape=[#examples, 1]) for gradients.
-   * @param hessians float32; Rank 2 Tensor (shape=[#examples, 1]) for hessians.
-   * @param bucketizedFeaturesList int32 list of Rank 1 Tensors, each containing the bucketized feature (for each feature column).
-   * @param maxSplits int; the maximum number of splits possible in the whole tree.
-   * @param numBuckets int; equals to the maximum possible value of bucketized feature.
-   * @return a new instance of BoostedTreesMakeStatsSummary
-   */
-  @Endpoint(describeByClass = true)
-  public static BoostedTreesMakeStatsSummary create(Scope scope, Operand<TInt32> nodeIds, Operand<TFloat32> gradients, Operand<TFloat32> hessians, Iterable<Operand<TInt32>> bucketizedFeaturesList, Long maxSplits, Long numBuckets) {
-    OperationBuilder opBuilder = scope.env().opBuilder("BoostedTreesMakeStatsSummary", scope.makeOpName("BoostedTreesMakeStatsSummary"));
-    opBuilder.addInput(nodeIds.asOutput(scope));
-    opBuilder.addInput(gradients.asOutput(scope));
-    opBuilder.addInput(hessians.asOutput(scope));
-    opBuilder.addInputList(Operands.asOutputs(scope, bucketizedFeaturesList));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("max_splits", maxSplits);
-    opBuilder.setAttr("num_buckets", numBuckets);
-    return new BoostedTreesMakeStatsSummary(opBuilder.build());
-  }
-  
-  /**
-   * output Rank 4 Tensor (shape=[#features, #splits, #buckets, 2]) containing accumulated stats put into the corresponding node and bucket. The first index of 4th dimension refers to gradients, and the second to hessians.
-   */
-  public Output<TFloat32> statsSummary() {
-    return statsSummary;
-  }
-  
-  @Override
-  public Output<TFloat32> asOutput(Scope scope) {
-    return statsSummary;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "BoostedTreesMakeStatsSummary";
-  
-  private Output<TFloat32> statsSummary;
-  
-  private BoostedTreesMakeStatsSummary(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    statsSummary = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesPredict.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesPredict.java
deleted file mode 100644
index f5bdd283ae9..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesPredict.java
+++ /dev/null
@@ -1,83 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.estimator;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.Operands;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.TInt32;
-
-/**
- * Runs multiple additive regression ensemble predictors on input instances and
- * <p>
- * computes the logits. It is designed to be used during prediction.
- * It traverses all the trees and calculates the final score for each instance.
- */
-public final class BoostedTreesPredict extends RawOp implements Operand<TFloat32> {
-  
-  /**
-   * Factory method to create a class wrapping a new BoostedTreesPredict operation.
-   * 
-   * @param scope current scope
-   * @param treeEnsembleHandle 
-   * @param bucketizedFeatures A list of rank 1 Tensors containing bucket id for each
-   * feature.
-   * @param logitsDimension scalar, dimension of the logits, to be used for partial logits
-   * shape.
-   * @return a new instance of BoostedTreesPredict
-   */
-  @Endpoint(describeByClass = true)
-  public static BoostedTreesPredict create(Scope scope, Operand<?> treeEnsembleHandle, Iterable<Operand<TInt32>> bucketizedFeatures, Long logitsDimension) {
-    OperationBuilder opBuilder = scope.env().opBuilder("BoostedTreesPredict", scope.makeOpName("BoostedTreesPredict"));
-    opBuilder.addInput(treeEnsembleHandle.asOutput(scope));
-    opBuilder.addInputList(Operands.asOutputs(scope, bucketizedFeatures));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("logits_dimension", logitsDimension);
-    return new BoostedTreesPredict(opBuilder.build());
-  }
-  
-  /**
-   * Output rank 2 Tensor containing logits for each example.
-   */
-  public Output<TFloat32> logits() {
-    return logits;
-  }
-  
-  @Override
-  public Output<TFloat32> asOutput(Scope scope) {
-    return logits;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "BoostedTreesPredict";
-  
-  private Output<TFloat32> logits;
-  
-  private BoostedTreesPredict(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    logits = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesQuantileStreamResourceAddSummaries.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesQuantileStreamResourceAddSummaries.java
deleted file mode 100644
index 140bcc8f951..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesQuantileStreamResourceAddSummaries.java
+++ /dev/null
@@ -1,62 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.estimator;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.Operands;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-
-/**
- * Add the quantile summaries to each quantile stream resource.
- * <p>
- * An op that adds a list of quantile summaries to a quantile stream resource. Each
- * summary Tensor is rank 2, containing summaries (value, weight, min_rank, max_rank)
- * for a single feature.
- */
-public final class BoostedTreesQuantileStreamResourceAddSummaries extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new BoostedTreesQuantileStreamResourceAddSummaries operation.
-   * 
-   * @param scope current scope
-   * @param quantileStreamResourceHandle resource handle referring to a QuantileStreamResource.
-   * @param summaries string; List of Rank 2 Tensor each containing the summaries for a single feature.
-   * @return a new instance of BoostedTreesQuantileStreamResourceAddSummaries
-   */
-  @Endpoint(describeByClass = true)
-  public static BoostedTreesQuantileStreamResourceAddSummaries create(Scope scope, Operand<?> quantileStreamResourceHandle, Iterable<Operand<TFloat32>> summaries) {
-    OperationBuilder opBuilder = scope.env().opBuilder("BoostedTreesQuantileStreamResourceAddSummaries", scope.makeOpName("BoostedTreesQuantileStreamResourceAddSummaries"));
-    opBuilder.addInput(quantileStreamResourceHandle.asOutput(scope));
-    opBuilder.addInputList(Operands.asOutputs(scope, summaries));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new BoostedTreesQuantileStreamResourceAddSummaries(opBuilder.build());
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "BoostedTreesQuantileStreamResourceAddSummaries";
-  
-  private BoostedTreesQuantileStreamResourceAddSummaries(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesQuantileStreamResourceDeserialize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesQuantileStreamResourceDeserialize.java
deleted file mode 100644
index 11073dec531..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesQuantileStreamResourceDeserialize.java
+++ /dev/null
@@ -1,60 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.estimator;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.Operands;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-
-/**
- * Deserialize bucket boundaries and ready flag into current QuantileAccumulator.
- * <p>
- * An op that deserializes bucket boundaries and are boundaries ready flag into current QuantileAccumulator.
- */
-public final class BoostedTreesQuantileStreamResourceDeserialize extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new BoostedTreesQuantileStreamResourceDeserialize operation.
-   * 
-   * @param scope current scope
-   * @param quantileStreamResourceHandle resource handle referring to a QuantileStreamResource.
-   * @param bucketBoundaries float; List of Rank 1 Tensors each containing the bucket boundaries for a feature.
-   * @return a new instance of BoostedTreesQuantileStreamResourceDeserialize
-   */
-  @Endpoint(describeByClass = true)
-  public static BoostedTreesQuantileStreamResourceDeserialize create(Scope scope, Operand<?> quantileStreamResourceHandle, Iterable<Operand<TFloat32>> bucketBoundaries) {
-    OperationBuilder opBuilder = scope.env().opBuilder("BoostedTreesQuantileStreamResourceDeserialize", scope.makeOpName("BoostedTreesQuantileStreamResourceDeserialize"));
-    opBuilder.addInput(quantileStreamResourceHandle.asOutput(scope));
-    opBuilder.addInputList(Operands.asOutputs(scope, bucketBoundaries));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new BoostedTreesQuantileStreamResourceDeserialize(opBuilder.build());
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "BoostedTreesQuantileStreamResourceDeserialize";
-  
-  private BoostedTreesQuantileStreamResourceDeserialize(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesQuantileStreamResourceFlush.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesQuantileStreamResourceFlush.java
deleted file mode 100644
index 64163d59d9a..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesQuantileStreamResourceFlush.java
+++ /dev/null
@@ -1,103 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.estimator;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt64;
-
-/**
- * Flush the summaries for a quantile stream resource.
- * <p>
- * An op that flushes the summaries for a quantile stream resource.
- */
-public final class BoostedTreesQuantileStreamResourceFlush extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.estimator.BoostedTreesQuantileStreamResourceFlush}
-   */
-  public static class Options {
-    
-    /**
-     * @param generateQuantiles bool; If True, the output will be the num_quantiles for each stream where the ith
-     * entry is the ith quantile of the input with an approximation error of epsilon.
-     * Duplicate values may be present.
-     * If False, the output will be the points in the histogram that we got which roughly
-     * translates to 1/epsilon boundaries and without any duplicates.
-     * Default to False.
-     */
-    public Options generateQuantiles(Boolean generateQuantiles) {
-      this.generateQuantiles = generateQuantiles;
-      return this;
-    }
-    
-    private Boolean generateQuantiles;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new BoostedTreesQuantileStreamResourceFlush operation.
-   * 
-   * @param scope current scope
-   * @param quantileStreamResourceHandle resource handle referring to a QuantileStreamResource.
-   * @param numBuckets int; approximate number of buckets unless using generate_quantiles.
-   * @param options carries optional attributes values
-   * @return a new instance of BoostedTreesQuantileStreamResourceFlush
-   */
-  @Endpoint(describeByClass = true)
-  public static BoostedTreesQuantileStreamResourceFlush create(Scope scope, Operand<?> quantileStreamResourceHandle, Operand<TInt64> numBuckets, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("BoostedTreesQuantileStreamResourceFlush", scope.makeOpName("BoostedTreesQuantileStreamResourceFlush"));
-    opBuilder.addInput(quantileStreamResourceHandle.asOutput(scope));
-    opBuilder.addInput(numBuckets.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.generateQuantiles != null) {
-          opBuilder.setAttr("generate_quantiles", opts.generateQuantiles);
-        }
-      }
-    }
-    return new BoostedTreesQuantileStreamResourceFlush(opBuilder.build());
-  }
-  
-  /**
-   * @param generateQuantiles bool; If True, the output will be the num_quantiles for each stream where the ith
-   * entry is the ith quantile of the input with an approximation error of epsilon.
-   * Duplicate values may be present.
-   * If False, the output will be the points in the histogram that we got which roughly
-   * translates to 1/epsilon boundaries and without any duplicates.
-   * Default to False.
-   */
-  public static Options generateQuantiles(Boolean generateQuantiles) {
-    return new Options().generateQuantiles(generateQuantiles);
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "BoostedTreesQuantileStreamResourceFlush";
-  
-  private BoostedTreesQuantileStreamResourceFlush(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesQuantileStreamResourceGetBucketBoundaries.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesQuantileStreamResourceGetBucketBoundaries.java
deleted file mode 100644
index ac6be6f10c5..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesQuantileStreamResourceGetBucketBoundaries.java
+++ /dev/null
@@ -1,84 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.estimator;
-
-import java.util.Arrays;
-import java.util.Iterator;
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-
-/**
- * Generate the bucket boundaries for each feature based on accumulated summaries.
- * <p>
- * An op that returns a list of float tensors for a quantile stream resource. Each
- * tensor is Rank 1 containing bucket boundaries for a single feature.
- */
-public final class BoostedTreesQuantileStreamResourceGetBucketBoundaries extends RawOp implements Iterable<Operand<TFloat32>> {
-  
-  /**
-   * Factory method to create a class wrapping a new BoostedTreesQuantileStreamResourceGetBucketBoundaries operation.
-   * 
-   * @param scope current scope
-   * @param quantileStreamResourceHandle resource handle referring to a QuantileStreamResource.
-   * @param numFeatures inferred int; number of features to get bucket boundaries for.
-   * @return a new instance of BoostedTreesQuantileStreamResourceGetBucketBoundaries
-   */
-  @Endpoint(describeByClass = true)
-  public static BoostedTreesQuantileStreamResourceGetBucketBoundaries create(Scope scope, Operand<?> quantileStreamResourceHandle, Long numFeatures) {
-    OperationBuilder opBuilder = scope.env().opBuilder("BoostedTreesQuantileStreamResourceGetBucketBoundaries", scope.makeOpName("BoostedTreesQuantileStreamResourceGetBucketBoundaries"));
-    opBuilder.addInput(quantileStreamResourceHandle.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("num_features", numFeatures);
-    return new BoostedTreesQuantileStreamResourceGetBucketBoundaries(opBuilder.build());
-  }
-  
-  /**
-   * float; List of Rank 1 Tensors each containing the bucket boundaries for a feature.
-   */
-  public List<Output<TFloat32>> bucketBoundaries() {
-    return bucketBoundaries;
-  }
-  
-  @Override
-  @SuppressWarnings({"rawtypes", "unchecked"})
-  public Iterator<Operand<TFloat32>> iterator() {
-    return (Iterator) bucketBoundaries.iterator();
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "BoostedTreesQuantileStreamResourceGetBucketBoundaries";
-  
-  private List<Output<TFloat32>> bucketBoundaries;
-  
-  @SuppressWarnings("unchecked")
-  private BoostedTreesQuantileStreamResourceGetBucketBoundaries(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    int bucketBoundariesLength = operation.outputListLength("bucket_boundaries");
-    bucketBoundaries = Arrays.asList((Output<TFloat32>[])operation.outputList(outputIdx, bucketBoundariesLength));
-    outputIdx += bucketBoundariesLength;
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesQuantileStreamResourceHandleOp.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesQuantileStreamResourceHandleOp.java
deleted file mode 100644
index 20ee0ddcb71..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesQuantileStreamResourceHandleOp.java
+++ /dev/null
@@ -1,123 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.estimator;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Creates a handle to a BoostedTreesQuantileStreamResource.
- */
-public final class BoostedTreesQuantileStreamResourceHandleOp extends RawOp implements Operand<TType> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.estimator.BoostedTreesQuantileStreamResourceHandleOp}
-   */
-  public static class Options {
-    
-    /**
-     * @param container 
-     */
-    public Options container(String container) {
-      this.container = container;
-      return this;
-    }
-    
-    /**
-     * @param sharedName 
-     */
-    public Options sharedName(String sharedName) {
-      this.sharedName = sharedName;
-      return this;
-    }
-    
-    private String container;
-    private String sharedName;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new BoostedTreesQuantileStreamResourceHandleOp operation.
-   * 
-   * @param scope current scope
-   * @param options carries optional attributes values
-   * @return a new instance of BoostedTreesQuantileStreamResourceHandleOp
-   */
-  @Endpoint(describeByClass = true)
-  public static BoostedTreesQuantileStreamResourceHandleOp create(Scope scope, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("BoostedTreesQuantileStreamResourceHandleOp", scope.makeOpName("BoostedTreesQuantileStreamResourceHandleOp"));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.container != null) {
-          opBuilder.setAttr("container", opts.container);
-        }
-        if (opts.sharedName != null) {
-          opBuilder.setAttr("shared_name", opts.sharedName);
-        }
-      }
-    }
-    return new BoostedTreesQuantileStreamResourceHandleOp(opBuilder.build());
-  }
-  
-  /**
-   * @param container 
-   */
-  public static Options container(String container) {
-    return new Options().container(container);
-  }
-  
-  /**
-   * @param sharedName 
-   */
-  public static Options sharedName(String sharedName) {
-    return new Options().sharedName(sharedName);
-  }
-  
-  /**
-   */
-  public Output<?> resource() {
-    return resource;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) resource;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "BoostedTreesQuantileStreamResourceHandleOp";
-  
-  private Output<?> resource;
-  
-  private BoostedTreesQuantileStreamResourceHandleOp(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    resource = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesSerializeEnsemble.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesSerializeEnsemble.java
deleted file mode 100644
index 4755654dee8..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesSerializeEnsemble.java
+++ /dev/null
@@ -1,77 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.estimator;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.TString;
-
-/**
- * Serializes the tree ensemble to a proto.
- */
-public final class BoostedTreesSerializeEnsemble extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new BoostedTreesSerializeEnsemble operation.
-   * 
-   * @param scope current scope
-   * @param treeEnsembleHandle Handle to the tree ensemble.
-   * @return a new instance of BoostedTreesSerializeEnsemble
-   */
-  @Endpoint(describeByClass = true)
-  public static BoostedTreesSerializeEnsemble create(Scope scope, Operand<?> treeEnsembleHandle) {
-    OperationBuilder opBuilder = scope.env().opBuilder("BoostedTreesSerializeEnsemble", scope.makeOpName("BoostedTreesSerializeEnsemble"));
-    opBuilder.addInput(treeEnsembleHandle.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new BoostedTreesSerializeEnsemble(opBuilder.build());
-  }
-  
-  /**
-   * Stamp token of the tree ensemble resource.
-   */
-  public Output<TInt64> stampToken() {
-    return stampToken;
-  }
-  
-  /**
-   * Serialized proto of the ensemble.
-   */
-  public Output<TString> treeEnsembleSerialized() {
-    return treeEnsembleSerialized;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "BoostedTreesSerializeEnsemble";
-  
-  private Output<TInt64> stampToken;
-  private Output<TString> treeEnsembleSerialized;
-  
-  private BoostedTreesSerializeEnsemble(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    stampToken = operation.output(outputIdx++);
-    treeEnsembleSerialized = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesSparseAggregateStats.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesSparseAggregateStats.java
deleted file mode 100644
index b8e0ba4c030..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesSparseAggregateStats.java
+++ /dev/null
@@ -1,115 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.estimator;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.TInt32;
-
-/**
- * Aggregates the summary of accumulated stats for the batch.
- * <p>
- * The summary stats contains gradients and hessians accumulated for each node, bucket and dimension id.
- */
-public final class BoostedTreesSparseAggregateStats extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new BoostedTreesSparseAggregateStats operation.
-   * 
-   * @param scope current scope
-   * @param nodeIds int32; Rank 1 Tensor containing node ids for each example, shape [batch_size].
-   * @param gradients float32; Rank 2 Tensor (shape=[batch_size, logits_dimension]) with gradients for each example.
-   * @param hessians float32; Rank 2 Tensor (shape=[batch_size, hessian_dimension]) with hessians for each example.
-   * @param featureIndices int32; Rank 2 indices of feature sparse Tensors (shape=[number of sparse entries, 2]).
-   * Number of sparse entries across all instances from the batch. The first value is
-   * the index of the instance, the second is dimension of the feature. The second axis
-   * can only have 2 values, i.e., the input dense version of Tensor can only be matrix.
-   * @param featureValues int32; Rank 1 values of feature sparse Tensors (shape=[number of sparse entries]).
-   * Number of sparse entries across all instances from the batch. The first value is
-   * the index of the instance, the second is dimension of the feature.
-   * @param featureShape int32; Rank 1 dense shape of feature sparse Tensors (shape=[2]).
-   * The first axis can only have 2 values, [batch_size, feature_dimension].
-   * @param maxSplits int; the maximum number of splits possible in the whole tree.
-   * @param numBuckets int; equals to the maximum possible value of bucketized feature + 1.
-   * @return a new instance of BoostedTreesSparseAggregateStats
-   */
-  @Endpoint(describeByClass = true)
-  public static BoostedTreesSparseAggregateStats create(Scope scope, Operand<TInt32> nodeIds, Operand<TFloat32> gradients, Operand<TFloat32> hessians, Operand<TInt32> featureIndices, Operand<TInt32> featureValues, Operand<TInt32> featureShape, Long maxSplits, Long numBuckets) {
-    OperationBuilder opBuilder = scope.env().opBuilder("BoostedTreesSparseAggregateStats", scope.makeOpName("BoostedTreesSparseAggregateStats"));
-    opBuilder.addInput(nodeIds.asOutput(scope));
-    opBuilder.addInput(gradients.asOutput(scope));
-    opBuilder.addInput(hessians.asOutput(scope));
-    opBuilder.addInput(featureIndices.asOutput(scope));
-    opBuilder.addInput(featureValues.asOutput(scope));
-    opBuilder.addInput(featureShape.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("max_splits", maxSplits);
-    opBuilder.setAttr("num_buckets", numBuckets);
-    return new BoostedTreesSparseAggregateStats(opBuilder.build());
-  }
-  
-  /**
-   * int32; Rank 2 indices of summary sparse Tensors (shape=[number of non zero statistics, 4])
-   * The second axis can only be 4 including node id, feature dimension, bucket id, and statistics_dimension.
-   * statistics_dimension = logits_dimension + hessian_dimension.
-   */
-  public Output<TInt32> statsSummaryIndices() {
-    return statsSummaryIndices;
-  }
-  
-  /**
-   * output Rank 1 Tensor (shape=[number of non zero statistics])
-   */
-  public Output<TFloat32> statsSummaryValues() {
-    return statsSummaryValues;
-  }
-  
-  /**
-   * output Rank 1 Tensor (shape=[4])
-   * The tensor has following 4 values: [max_splits, feature_dimension, num_buckets, statistics_dimension],
-   * where statistics_dimension = gradient_dimension + hessian_dimension. gradient_dimension
-   * is the same as label_dimension, i.e., the output space. hessian_dimension can be the same
-   * as logits dimension when diagonal hessian is used, or label_dimension^2 when full
-   * hessian is used.
-   */
-  public Output<TInt32> statsSummaryShape() {
-    return statsSummaryShape;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "BoostedTreesSparseAggregateStats";
-  
-  private Output<TInt32> statsSummaryIndices;
-  private Output<TFloat32> statsSummaryValues;
-  private Output<TInt32> statsSummaryShape;
-  
-  private BoostedTreesSparseAggregateStats(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    statsSummaryIndices = operation.output(outputIdx++);
-    statsSummaryValues = operation.output(outputIdx++);
-    statsSummaryShape = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesSparseCalculateBestFeatureSplit.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesSparseCalculateBestFeatureSplit.java
deleted file mode 100644
index 3d20166d9cf..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesSparseCalculateBestFeatureSplit.java
+++ /dev/null
@@ -1,184 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.estimator;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.TString;
-
-/**
- * Calculates gains for each feature and returns the best possible split information for the feature.
- * <p>
- * The split information is the best threshold (bucket id), gains and left/right node contributions per node for each feature.
- * <p>
- * It is possible that not all nodes can be split on each feature. Hence, the list of possible nodes can differ between the features. Therefore, we return `node_ids_list` for each feature, containing the list of nodes that this feature can be used to split.
- * <p>
- * In this manner, the output is the best split per features and per node, so that it needs to be combined later to produce the best split for each node (among all possible features).
- * <p>
- * The output shapes are compatible in a way that the first dimension of all tensors are the same and equal to the number of possible split nodes for each feature.
- */
-public final class BoostedTreesSparseCalculateBestFeatureSplit extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.estimator.BoostedTreesSparseCalculateBestFeatureSplit}
-   */
-  public static class Options {
-    
-    /**
-     * @param splitType A string indicating if this Op should perform inequality split or equality split.
-     */
-    public Options splitType(String splitType) {
-      this.splitType = splitType;
-      return this;
-    }
-    
-    private String splitType;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new BoostedTreesSparseCalculateBestFeatureSplit operation.
-   * 
-   * @param scope current scope
-   * @param nodeIdRange A Rank 1 tensor (shape=[2]) to specify the range [first, last) of node ids to process within `stats_summary_list`. The nodes are iterated between the two nodes specified by the tensor, as like `for node_id in range(node_id_range[0], node_id_range[1])` (Note that the last index node_id_range[1] is exclusive).
-   * @param statsSummaryIndices A Rank 2 int64 tensor of dense shape [N, 4] (N specifies the number of non-zero values) for accumulated stats summary (gradient/hessian) per node per bucket for each feature. The second dimension contains node id, feature dimension, bucket id, and stats dim.
-   * stats dim is the sum of logits dimension and hessian dimension, hessian dimension can either be logits dimension if diagonal hessian is used, or logits dimension^2 if full hessian is used.
-   * @param statsSummaryValues A Rank 1 float tensor of dense shape [N] (N specifies the number of non-zero values), which supplies the values for each element in summary_indices.
-   * @param statsSummaryShape A Rank 1 float tensor of dense shape [4], which specifies the dense shape of the sparse tensor, which is [num tree nodes, feature dimensions, num buckets, stats dim].
-   * @param l1 l1 regularization factor on leaf weights, per instance based.
-   * @param l2 l2 regularization factor on leaf weights, per instance based.
-   * @param treeComplexity adjustment to the gain, per leaf based.
-   * @param minNodeWeight minimum avg of hessians in a node before required for the node to be considered for splitting.
-   * @param logitsDimension The dimension of logit, i.e., number of classes.
-   * @param options carries optional attributes values
-   * @return a new instance of BoostedTreesSparseCalculateBestFeatureSplit
-   */
-  @Endpoint(describeByClass = true)
-  public static BoostedTreesSparseCalculateBestFeatureSplit create(Scope scope, Operand<TInt32> nodeIdRange, Operand<TInt32> statsSummaryIndices, Operand<TFloat32> statsSummaryValues, Operand<TInt32> statsSummaryShape, Operand<TFloat32> l1, Operand<TFloat32> l2, Operand<TFloat32> treeComplexity, Operand<TFloat32> minNodeWeight, Long logitsDimension, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("BoostedTreesSparseCalculateBestFeatureSplit", scope.makeOpName("BoostedTreesSparseCalculateBestFeatureSplit"));
-    opBuilder.addInput(nodeIdRange.asOutput(scope));
-    opBuilder.addInput(statsSummaryIndices.asOutput(scope));
-    opBuilder.addInput(statsSummaryValues.asOutput(scope));
-    opBuilder.addInput(statsSummaryShape.asOutput(scope));
-    opBuilder.addInput(l1.asOutput(scope));
-    opBuilder.addInput(l2.asOutput(scope));
-    opBuilder.addInput(treeComplexity.asOutput(scope));
-    opBuilder.addInput(minNodeWeight.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("logits_dimension", logitsDimension);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.splitType != null) {
-          opBuilder.setAttr("split_type", opts.splitType);
-        }
-      }
-    }
-    return new BoostedTreesSparseCalculateBestFeatureSplit(opBuilder.build());
-  }
-  
-  /**
-   * @param splitType A string indicating if this Op should perform inequality split or equality split.
-   */
-  public static Options splitType(String splitType) {
-    return new Options().splitType(splitType);
-  }
-  
-  /**
-   * A Rank 1 tensor indicating possible node ids that can be split.
-   */
-  public Output<TInt32> nodeIds() {
-    return nodeIds;
-  }
-  
-  /**
-   * A Rank 1 tensor indicating the best gains to split each node.
-   */
-  public Output<TFloat32> gains() {
-    return gains;
-  }
-  
-  /**
-   * A Rank 1 tensor indicating the best feature dimension for each feature to split for each node.
-   */
-  public Output<TInt32> featureDimensions() {
-    return featureDimensions;
-  }
-  
-  /**
-   * A Rank 1 tensor indicating the bucket id to compare with (as a threshold) for split in each node.
-   */
-  public Output<TInt32> thresholds() {
-    return thresholds;
-  }
-  
-  /**
-   * A Rank 2 tensor indicating the contribution of the left nodes when branching from parent nodes to the left direction by the given threshold for each feature.
-   * This value will be used to make the left node value by adding to the parent node value. Second dimension size is logits dimension.
-   */
-  public Output<TFloat32> leftNodeContribs() {
-    return leftNodeContribs;
-  }
-  
-  /**
-   * A Rank 2 tensor, with the same shape/conditions as left_node_contribs_list, but just that the value is for the right node.
-   */
-  public Output<TFloat32> rightNodeContribs() {
-    return rightNodeContribs;
-  }
-  
-  /**
-   * A Rank 1 tensor indicating which direction to go if data is missing.
-   * Inequality with default left returns 0, inequality with default right returns 1, equality with default right returns 2.
-   */
-  public Output<TString> splitWithDefaultDirections() {
-    return splitWithDefaultDirections;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "BoostedTreesSparseCalculateBestFeatureSplit";
-  
-  private Output<TInt32> nodeIds;
-  private Output<TFloat32> gains;
-  private Output<TInt32> featureDimensions;
-  private Output<TInt32> thresholds;
-  private Output<TFloat32> leftNodeContribs;
-  private Output<TFloat32> rightNodeContribs;
-  private Output<TString> splitWithDefaultDirections;
-  
-  private BoostedTreesSparseCalculateBestFeatureSplit(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    nodeIds = operation.output(outputIdx++);
-    gains = operation.output(outputIdx++);
-    featureDimensions = operation.output(outputIdx++);
-    thresholds = operation.output(outputIdx++);
-    leftNodeContribs = operation.output(outputIdx++);
-    rightNodeContribs = operation.output(outputIdx++);
-    splitWithDefaultDirections = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesTrainingPredict.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesTrainingPredict.java
deleted file mode 100644
index 560e2b20948..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesTrainingPredict.java
+++ /dev/null
@@ -1,104 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.estimator;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.Operands;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.TInt32;
-
-/**
- * Runs multiple additive regression ensemble predictors on input instances and
- * <p>
- * computes the update to cached logits. It is designed to be used during training.
- * It traverses the trees starting from cached tree id and cached node id and
- * calculates the updates to be pushed to the cache.
- */
-public final class BoostedTreesTrainingPredict extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new BoostedTreesTrainingPredict operation.
-   * 
-   * @param scope current scope
-   * @param treeEnsembleHandle 
-   * @param cachedTreeIds Rank 1 Tensor containing cached tree ids which is the starting
-   * tree of prediction.
-   * @param cachedNodeIds Rank 1 Tensor containing cached node id which is the starting
-   * node of prediction.
-   * @param bucketizedFeatures A list of rank 1 Tensors containing bucket id for each
-   * feature.
-   * @param logitsDimension scalar, dimension of the logits, to be used for partial logits
-   * shape.
-   * @return a new instance of BoostedTreesTrainingPredict
-   */
-  @Endpoint(describeByClass = true)
-  public static BoostedTreesTrainingPredict create(Scope scope, Operand<?> treeEnsembleHandle, Operand<TInt32> cachedTreeIds, Operand<TInt32> cachedNodeIds, Iterable<Operand<TInt32>> bucketizedFeatures, Long logitsDimension) {
-    OperationBuilder opBuilder = scope.env().opBuilder("BoostedTreesTrainingPredict", scope.makeOpName("BoostedTreesTrainingPredict"));
-    opBuilder.addInput(treeEnsembleHandle.asOutput(scope));
-    opBuilder.addInput(cachedTreeIds.asOutput(scope));
-    opBuilder.addInput(cachedNodeIds.asOutput(scope));
-    opBuilder.addInputList(Operands.asOutputs(scope, bucketizedFeatures));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("logits_dimension", logitsDimension);
-    return new BoostedTreesTrainingPredict(opBuilder.build());
-  }
-  
-  /**
-   * Rank 2 Tensor containing logits update (with respect to cached
-   * values stored) for each example.
-   */
-  public Output<TFloat32> partialLogits() {
-    return partialLogits;
-  }
-  
-  /**
-   * Rank 1 Tensor containing new tree ids for each example.
-   */
-  public Output<TInt32> treeIds() {
-    return treeIds;
-  }
-  
-  /**
-   * Rank 1 Tensor containing new node ids in the new tree_ids.
-   */
-  public Output<TInt32> nodeIds() {
-    return nodeIds;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "BoostedTreesTrainingPredict";
-  
-  private Output<TFloat32> partialLogits;
-  private Output<TInt32> treeIds;
-  private Output<TInt32> nodeIds;
-  
-  private BoostedTreesTrainingPredict(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    partialLogits = operation.output(outputIdx++);
-    treeIds = operation.output(outputIdx++);
-    nodeIds = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesUpdateEnsemble.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesUpdateEnsemble.java
deleted file mode 100644
index 1e304c4319a..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesUpdateEnsemble.java
+++ /dev/null
@@ -1,85 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.estimator;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.Operands;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.TInt32;
-
-/**
- * Updates the tree ensemble by either adding a layer to the last tree being grown
- * <p>
- * or by starting a new tree.
- */
-public final class BoostedTreesUpdateEnsemble extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new BoostedTreesUpdateEnsemble operation.
-   * 
-   * @param scope current scope
-   * @param treeEnsembleHandle Handle to the ensemble variable.
-   * @param featureIds Rank 1 tensor with ids for each feature. This is the real id of
-   * the feature that will be used in the split.
-   * @param nodeIds List of rank 1 tensors representing the nodes for which this feature
-   * has a split.
-   * @param gains List of rank 1 tensors representing the gains for each of the feature's
-   * split.
-   * @param thresholds List of rank 1 tensors representing the thesholds for each of the
-   * feature's split.
-   * @param leftNodeContribs List of rank 2 tensors with left leaf contribs for each of
-   * the feature's splits. Will be added to the previous node values to constitute
-   * the values of the left nodes.
-   * @param rightNodeContribs List of rank 2 tensors with right leaf contribs for each
-   * of the feature's splits. Will be added to the previous node values to constitute
-   * the values of the right nodes.
-   * @param maxDepth Max depth of the tree to build.
-   * @param learningRate shrinkage const for each new tree.
-   * @param pruningMode 0-No pruning, 1-Pre-pruning, 2-Post-pruning.
-   * @return a new instance of BoostedTreesUpdateEnsemble
-   */
-  @Endpoint(describeByClass = true)
-  public static BoostedTreesUpdateEnsemble create(Scope scope, Operand<?> treeEnsembleHandle, Operand<TInt32> featureIds, Iterable<Operand<TInt32>> nodeIds, Iterable<Operand<TFloat32>> gains, Iterable<Operand<TInt32>> thresholds, Iterable<Operand<TFloat32>> leftNodeContribs, Iterable<Operand<TFloat32>> rightNodeContribs, Operand<TInt32> maxDepth, Operand<TFloat32> learningRate, Long pruningMode) {
-    OperationBuilder opBuilder = scope.env().opBuilder("BoostedTreesUpdateEnsemble", scope.makeOpName("BoostedTreesUpdateEnsemble"));
-    opBuilder.addInput(treeEnsembleHandle.asOutput(scope));
-    opBuilder.addInput(featureIds.asOutput(scope));
-    opBuilder.addInputList(Operands.asOutputs(scope, nodeIds));
-    opBuilder.addInputList(Operands.asOutputs(scope, gains));
-    opBuilder.addInputList(Operands.asOutputs(scope, thresholds));
-    opBuilder.addInputList(Operands.asOutputs(scope, leftNodeContribs));
-    opBuilder.addInputList(Operands.asOutputs(scope, rightNodeContribs));
-    opBuilder.addInput(maxDepth.asOutput(scope));
-    opBuilder.addInput(learningRate.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("pruning_mode", pruningMode);
-    return new BoostedTreesUpdateEnsemble(opBuilder.build());
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "BoostedTreesUpdateEnsemble";
-  
-  private BoostedTreesUpdateEnsemble(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesUpdateEnsembleV2.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesUpdateEnsembleV2.java
deleted file mode 100644
index 653cdb0a3ea..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesUpdateEnsembleV2.java
+++ /dev/null
@@ -1,124 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.estimator;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.Operands;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.TString;
-
-/**
- * Updates the tree ensemble by adding a layer to the last tree being grown
- * <p>
- * or by starting a new tree.
- */
-public final class BoostedTreesUpdateEnsembleV2 extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.estimator.BoostedTreesUpdateEnsembleV2}
-   */
-  public static class Options {
-    
-    /**
-     * @param logitsDimension scalar, dimension of the logits
-     */
-    public Options logitsDimension(Long logitsDimension) {
-      this.logitsDimension = logitsDimension;
-      return this;
-    }
-    
-    private Long logitsDimension;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new BoostedTreesUpdateEnsembleV2 operation.
-   * 
-   * @param scope current scope
-   * @param treeEnsembleHandle Handle to the ensemble variable.
-   * @param featureIds Rank 1 tensor with ids for each feature. This is the real id of
-   * the feature that will be used in the split.
-   * @param dimensionIds List of rank 1 tensors representing the dimension in each feature.
-   * @param nodeIds List of rank 1 tensors representing the nodes for which this feature
-   * has a split.
-   * @param gains List of rank 1 tensors representing the gains for each of the feature's
-   * split.
-   * @param thresholds List of rank 1 tensors representing the thesholds for each of the
-   * feature's split.
-   * @param leftNodeContribs List of rank 2 tensors with left leaf contribs for each of
-   * the feature's splits. Will be added to the previous node values to constitute
-   * the values of the left nodes.
-   * @param rightNodeContribs List of rank 2 tensors with right leaf contribs for each
-   * of the feature's splits. Will be added to the previous node values to constitute
-   * the values of the right nodes.
-   * @param splitTypes List of rank 1 tensors representing the split type for each feature.
-   * @param maxDepth Max depth of the tree to build.
-   * @param learningRate shrinkage const for each new tree.
-   * @param pruningMode 0-No pruning, 1-Pre-pruning, 2-Post-pruning.
-   * @param options carries optional attributes values
-   * @return a new instance of BoostedTreesUpdateEnsembleV2
-   */
-  @Endpoint(describeByClass = true)
-  public static BoostedTreesUpdateEnsembleV2 create(Scope scope, Operand<?> treeEnsembleHandle, Iterable<Operand<TInt32>> featureIds, Iterable<Operand<TInt32>> dimensionIds, Iterable<Operand<TInt32>> nodeIds, Iterable<Operand<TFloat32>> gains, Iterable<Operand<TInt32>> thresholds, Iterable<Operand<TFloat32>> leftNodeContribs, Iterable<Operand<TFloat32>> rightNodeContribs, Iterable<Operand<TString>> splitTypes, Operand<TInt32> maxDepth, Operand<TFloat32> learningRate, Operand<TInt32> pruningMode, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("BoostedTreesUpdateEnsembleV2", scope.makeOpName("BoostedTreesUpdateEnsembleV2"));
-    opBuilder.addInput(treeEnsembleHandle.asOutput(scope));
-    opBuilder.addInputList(Operands.asOutputs(scope, featureIds));
-    opBuilder.addInputList(Operands.asOutputs(scope, dimensionIds));
-    opBuilder.addInputList(Operands.asOutputs(scope, nodeIds));
-    opBuilder.addInputList(Operands.asOutputs(scope, gains));
-    opBuilder.addInputList(Operands.asOutputs(scope, thresholds));
-    opBuilder.addInputList(Operands.asOutputs(scope, leftNodeContribs));
-    opBuilder.addInputList(Operands.asOutputs(scope, rightNodeContribs));
-    opBuilder.addInputList(Operands.asOutputs(scope, splitTypes));
-    opBuilder.addInput(maxDepth.asOutput(scope));
-    opBuilder.addInput(learningRate.asOutput(scope));
-    opBuilder.addInput(pruningMode.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.logitsDimension != null) {
-          opBuilder.setAttr("logits_dimension", opts.logitsDimension);
-        }
-      }
-    }
-    return new BoostedTreesUpdateEnsembleV2(opBuilder.build());
-  }
-  
-  /**
-   * @param logitsDimension scalar, dimension of the logits
-   */
-  public static Options logitsDimension(Long logitsDimension) {
-    return new Options().logitsDimension(logitsDimension);
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "BoostedTreesUpdateEnsembleV2";
-  
-  private BoostedTreesUpdateEnsembleV2(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/IsBoostedTreesEnsembleInitialized.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/IsBoostedTreesEnsembleInitialized.java
deleted file mode 100644
index acccce061ff..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/IsBoostedTreesEnsembleInitialized.java
+++ /dev/null
@@ -1,72 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.estimator;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TBool;
-
-/**
- * Checks whether a tree ensemble has been initialized.
- */
-public final class IsBoostedTreesEnsembleInitialized extends RawOp implements Operand<TBool> {
-  
-  /**
-   * Factory method to create a class wrapping a new IsBoostedTreesEnsembleInitialized operation.
-   * 
-   * @param scope current scope
-   * @param treeEnsembleHandle Handle to the tree ensemble resource.
-   * @return a new instance of IsBoostedTreesEnsembleInitialized
-   */
-  @Endpoint(describeByClass = true)
-  public static IsBoostedTreesEnsembleInitialized create(Scope scope, Operand<?> treeEnsembleHandle) {
-    OperationBuilder opBuilder = scope.env().opBuilder("IsBoostedTreesEnsembleInitialized", scope.makeOpName("IsBoostedTreesEnsembleInitialized"));
-    opBuilder.addInput(treeEnsembleHandle.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new IsBoostedTreesEnsembleInitialized(opBuilder.build());
-  }
-  
-  /**
-   * output boolean on whether it is initialized or not.
-   */
-  public Output<TBool> isInitialized() {
-    return isInitialized;
-  }
-  
-  @Override
-  public Output<TBool> asOutput(Scope scope) {
-    return isInitialized;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "IsBoostedTreesEnsembleInitialized";
-  
-  private Output<TBool> isInitialized;
-  
-  private IsBoostedTreesEnsembleInitialized(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    isInitialized = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/IsBoostedTreesQuantileStreamResourceInitialized.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/IsBoostedTreesQuantileStreamResourceInitialized.java
deleted file mode 100644
index e16086691a8..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/IsBoostedTreesQuantileStreamResourceInitialized.java
+++ /dev/null
@@ -1,74 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.estimator;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TBool;
-
-/**
- * Checks whether a quantile stream has been initialized.
- * <p>
- * An Op that checks if quantile stream resource is initialized.
- */
-public final class IsBoostedTreesQuantileStreamResourceInitialized extends RawOp implements Operand<TBool> {
-  
-  /**
-   * Factory method to create a class wrapping a new IsBoostedTreesQuantileStreamResourceInitialized operation.
-   * 
-   * @param scope current scope
-   * @param quantileStreamResourceHandle resource; The reference to quantile stream resource handle.
-   * @return a new instance of IsBoostedTreesQuantileStreamResourceInitialized
-   */
-  @Endpoint(describeByClass = true)
-  public static IsBoostedTreesQuantileStreamResourceInitialized create(Scope scope, Operand<?> quantileStreamResourceHandle) {
-    OperationBuilder opBuilder = scope.env().opBuilder("IsBoostedTreesQuantileStreamResourceInitialized", scope.makeOpName("IsBoostedTreesQuantileStreamResourceInitialized"));
-    opBuilder.addInput(quantileStreamResourceHandle.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new IsBoostedTreesQuantileStreamResourceInitialized(opBuilder.build());
-  }
-  
-  /**
-   * bool; True if the resource is initialized, False otherwise.
-   */
-  public Output<TBool> isInitialized() {
-    return isInitialized;
-  }
-  
-  @Override
-  public Output<TBool> asOutput(Scope scope) {
-    return isInitialized;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "IsBoostedTreesQuantileStreamResourceInitialized";
-  
-  private Output<TBool> isInitialized;
-  
-  private IsBoostedTreesQuantileStreamResourceInitialized(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    isInitialized = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/AdjustContrast.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/AdjustContrast.java
deleted file mode 100644
index a5f824f5145..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/AdjustContrast.java
+++ /dev/null
@@ -1,88 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.image;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Adjust the contrast of one or more images.
- * <p>
- * `images` is a tensor of at least 3 dimensions.  The last 3 dimensions are
- * interpreted as `[height, width, channels]`.  The other dimensions only
- * represent a collection of images, such as `[batch, height, width, channels].`
- * <p>
- * Contrast is adjusted independently for each channel of each image.
- * <p>
- * For each channel, the Op first computes the mean of the image pixels in the
- * channel and then adjusts each component of each pixel to
- * `(x - mean) * contrast_factor + mean`.
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator(group = "image")
-public final class AdjustContrast<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new AdjustContrast operation.
-   * 
-   * @param scope current scope
-   * @param images Images to adjust.  At least 3-D.
-   * @param contrastFactor A float multiplier for adjusting contrast.
-   * @return a new instance of AdjustContrast
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> AdjustContrast<T> create(Scope scope, Operand<T> images, Operand<TFloat32> contrastFactor) {
-    OperationBuilder opBuilder = scope.env().opBuilder("AdjustContrastv2", scope.makeOpName("AdjustContrast"));
-    opBuilder.addInput(images.asOutput(scope));
-    opBuilder.addInput(contrastFactor.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new AdjustContrast<T>(opBuilder.build());
-  }
-  
-  /**
-   * The contrast-adjusted image or images.
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "AdjustContrastv2";
-  
-  private Output<T> output;
-  
-  private AdjustContrast(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/AdjustHue.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/AdjustHue.java
deleted file mode 100644
index 00d49018a7e..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/AdjustHue.java
+++ /dev/null
@@ -1,85 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.image;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Adjust the hue of one or more images.
- * <p>
- * `images` is a tensor of at least 3 dimensions.  The last dimension is
- * interpreted as channels, and must be three.
- * <p>
- * The input image is considered in the RGB colorspace. Conceptually, the RGB
- * colors are first mapped into HSV. A delta is then applied all the hue values,
- * and then remapped back to RGB colorspace.
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator(group = "image")
-public final class AdjustHue<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new AdjustHue operation.
-   * 
-   * @param scope current scope
-   * @param images Images to adjust.  At least 3-D.
-   * @param delta A float delta to add to the hue.
-   * @return a new instance of AdjustHue
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> AdjustHue<T> create(Scope scope, Operand<T> images, Operand<TFloat32> delta) {
-    OperationBuilder opBuilder = scope.env().opBuilder("AdjustHue", scope.makeOpName("AdjustHue"));
-    opBuilder.addInput(images.asOutput(scope));
-    opBuilder.addInput(delta.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new AdjustHue<T>(opBuilder.build());
-  }
-  
-  /**
-   * The hue-adjusted image or images.
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "AdjustHue";
-  
-  private Output<T> output;
-  
-  private AdjustHue(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/AdjustSaturation.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/AdjustSaturation.java
deleted file mode 100644
index d0100ce4f8a..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/AdjustSaturation.java
+++ /dev/null
@@ -1,85 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.image;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Adjust the saturation of one or more images.
- * <p>
- * `images` is a tensor of at least 3 dimensions.  The last dimension is
- * interpreted as channels, and must be three.
- * <p>
- * The input image is considered in the RGB colorspace. Conceptually, the RGB
- * colors are first mapped into HSV. A scale is then applied all the saturation
- * values, and then remapped back to RGB colorspace.
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator(group = "image")
-public final class AdjustSaturation<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new AdjustSaturation operation.
-   * 
-   * @param scope current scope
-   * @param images Images to adjust.  At least 3-D.
-   * @param scale A float scale to add to the saturation.
-   * @return a new instance of AdjustSaturation
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> AdjustSaturation<T> create(Scope scope, Operand<T> images, Operand<TFloat32> scale) {
-    OperationBuilder opBuilder = scope.env().opBuilder("AdjustSaturation", scope.makeOpName("AdjustSaturation"));
-    opBuilder.addInput(images.asOutput(scope));
-    opBuilder.addInput(scale.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new AdjustSaturation<T>(opBuilder.build());
-  }
-  
-  /**
-   * The hue-adjusted image or images.
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "AdjustSaturation";
-  
-  private Output<T> output;
-  
-  private AdjustSaturation(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/CombinedNonMaxSuppression.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/CombinedNonMaxSuppression.java
deleted file mode 100644
index 4868f373f58..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/CombinedNonMaxSuppression.java
+++ /dev/null
@@ -1,197 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.image;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.TInt32;
-
-/**
- * Greedily selects a subset of bounding boxes in descending order of score,
- * <p>
- * This operation performs non_max_suppression on the inputs per batch, across
- * all classes.
- * Prunes away boxes that have high intersection-over-union (IOU) overlap
- * with previously selected boxes.  Bounding boxes are supplied as
- * [y1, x1, y2, x2], where (y1, x1) and (y2, x2) are the coordinates of any
- * diagonal pair of box corners and the coordinates can be provided as normalized
- * (i.e., lying in the interval [0, 1]) or absolute.  Note that this algorithm
- * is agnostic to where the origin is in the coordinate system. Also note that
- * this algorithm is invariant to orthogonal transformations and translations
- * of the coordinate system; thus translating or reflections of the coordinate
- * system result in the same boxes being selected by the algorithm.
- * The output of this operation is the final boxes, scores and classes tensor
- * returned after performing non_max_suppression.
- */
-@Operator(group = "image")
-public final class CombinedNonMaxSuppression extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.image.CombinedNonMaxSuppression}
-   */
-  public static class Options {
-    
-    /**
-     * @param padPerClass If false, the output nmsed boxes, scores and classes
-     * are padded/clipped to `max_total_size`. If true, the
-     * output nmsed boxes, scores and classes are padded to be of length
-     * `max_size_per_class`*`num_classes`, unless it exceeds `max_total_size` in
-     * which case it is clipped to `max_total_size`. Defaults to false.
-     */
-    public Options padPerClass(Boolean padPerClass) {
-      this.padPerClass = padPerClass;
-      return this;
-    }
-    
-    /**
-     * @param clipBoxes If true, assume the box coordinates are between [0, 1] and clip the output boxes
-     * if they fall beyond [0, 1]. If false, do not do clipping and output the box
-     * coordinates as it is.
-     */
-    public Options clipBoxes(Boolean clipBoxes) {
-      this.clipBoxes = clipBoxes;
-      return this;
-    }
-    
-    private Boolean padPerClass;
-    private Boolean clipBoxes;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new CombinedNonMaxSuppression operation.
-   * 
-   * @param scope current scope
-   * @param boxes A 4-D float tensor of shape `[batch_size, num_boxes, q, 4]`. If `q` is 1 then
-   * same boxes are used for all classes otherwise, if `q` is equal to number of
-   * classes, class-specific boxes are used.
-   * @param scores A 3-D float tensor of shape `[batch_size, num_boxes, num_classes]`
-   * representing a single score corresponding to each box (each row of boxes).
-   * @param maxOutputSizePerClass A scalar integer tensor representing the maximum number of
-   * boxes to be selected by non max suppression per class
-   * @param maxTotalSize A scalar representing maximum number of boxes retained over all classes.
-   * @param iouThreshold A 0-D float tensor representing the threshold for deciding whether
-   * boxes overlap too much with respect to IOU.
-   * @param scoreThreshold A 0-D float tensor representing the threshold for deciding when to remove
-   * boxes based on score.
-   * @param options carries optional attributes values
-   * @return a new instance of CombinedNonMaxSuppression
-   */
-  @Endpoint(describeByClass = true)
-  public static CombinedNonMaxSuppression create(Scope scope, Operand<TFloat32> boxes, Operand<TFloat32> scores, Operand<TInt32> maxOutputSizePerClass, Operand<TInt32> maxTotalSize, Operand<TFloat32> iouThreshold, Operand<TFloat32> scoreThreshold, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("CombinedNonMaxSuppression", scope.makeOpName("CombinedNonMaxSuppression"));
-    opBuilder.addInput(boxes.asOutput(scope));
-    opBuilder.addInput(scores.asOutput(scope));
-    opBuilder.addInput(maxOutputSizePerClass.asOutput(scope));
-    opBuilder.addInput(maxTotalSize.asOutput(scope));
-    opBuilder.addInput(iouThreshold.asOutput(scope));
-    opBuilder.addInput(scoreThreshold.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.padPerClass != null) {
-          opBuilder.setAttr("pad_per_class", opts.padPerClass);
-        }
-        if (opts.clipBoxes != null) {
-          opBuilder.setAttr("clip_boxes", opts.clipBoxes);
-        }
-      }
-    }
-    return new CombinedNonMaxSuppression(opBuilder.build());
-  }
-  
-  /**
-   * @param padPerClass If false, the output nmsed boxes, scores and classes
-   * are padded/clipped to `max_total_size`. If true, the
-   * output nmsed boxes, scores and classes are padded to be of length
-   * `max_size_per_class`*`num_classes`, unless it exceeds `max_total_size` in
-   * which case it is clipped to `max_total_size`. Defaults to false.
-   */
-  public static Options padPerClass(Boolean padPerClass) {
-    return new Options().padPerClass(padPerClass);
-  }
-  
-  /**
-   * @param clipBoxes If true, assume the box coordinates are between [0, 1] and clip the output boxes
-   * if they fall beyond [0, 1]. If false, do not do clipping and output the box
-   * coordinates as it is.
-   */
-  public static Options clipBoxes(Boolean clipBoxes) {
-    return new Options().clipBoxes(clipBoxes);
-  }
-  
-  /**
-   * A [batch_size, max_detections, 4] float32 tensor
-   * containing the non-max suppressed boxes.
-   */
-  public Output<TFloat32> nmsedBoxes() {
-    return nmsedBoxes;
-  }
-  
-  /**
-   * A [batch_size, max_detections] float32 tensor
-   * containing the scores for the boxes.
-   */
-  public Output<TFloat32> nmsedScores() {
-    return nmsedScores;
-  }
-  
-  /**
-   * A [batch_size, max_detections] float32 tensor
-   * containing the classes for the boxes.
-   */
-  public Output<TFloat32> nmsedClasses() {
-    return nmsedClasses;
-  }
-  
-  /**
-   * A [batch_size] int32 tensor indicating the number of
-   * valid detections per batch item. Only the top num_detections[i] entries in
-   * nms_boxes[i], nms_scores[i] and nms_class[i] are valid. The rest of the
-   * entries are zero paddings.
-   */
-  public Output<TInt32> validDetections() {
-    return validDetections;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "CombinedNonMaxSuppression";
-  
-  private Output<TFloat32> nmsedBoxes;
-  private Output<TFloat32> nmsedScores;
-  private Output<TFloat32> nmsedClasses;
-  private Output<TInt32> validDetections;
-  
-  private CombinedNonMaxSuppression(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    nmsedBoxes = operation.output(outputIdx++);
-    nmsedScores = operation.output(outputIdx++);
-    nmsedClasses = operation.output(outputIdx++);
-    validDetections = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/CropAndResize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/CropAndResize.java
deleted file mode 100644
index bae22372323..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/CropAndResize.java
+++ /dev/null
@@ -1,168 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.image;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Extracts crops from the input image tensor and resizes them.
- * <p>
- * Extracts crops from the input image tensor and resizes them using bilinear
- * sampling or nearest neighbor sampling (possibly with aspect ratio change) to a
- * common output size specified by `crop_size`. This is more general than the
- * `crop_to_bounding_box` op which extracts a fixed size slice from the input image
- * and does not allow resizing or aspect ratio change.
- * <p>
- * Returns a tensor with `crops` from the input `image` at positions defined at the
- * bounding box locations in `boxes`. The cropped boxes are all resized (with
- * bilinear or nearest neighbor interpolation) to a fixed
- * `size = [crop_height, crop_width]`. The result is a 4-D tensor
- * `[num_boxes, crop_height, crop_width, depth]`. The resizing is corner aligned.
- * In particular, if `boxes = [[0, 0, 1, 1]]`, the method will give identical
- * results to using `tf.image.resize_bilinear()` or
- * `tf.image.resize_nearest_neighbor()`(depends on the `method` argument) with
- * `align_corners=True`.
- */
-@Operator(group = "image")
-public final class CropAndResize extends RawOp implements Operand<TFloat32> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.image.CropAndResize}
-   */
-  public static class Options {
-    
-    /**
-     * @param method A string specifying the sampling method for resizing. It can be either
-     * `"bilinear"` or `"nearest"` and default to `"bilinear"`. Currently two sampling
-     * methods are supported: Bilinear and Nearest Neighbor.
-     */
-    public Options method(String method) {
-      this.method = method;
-      return this;
-    }
-    
-    /**
-     * @param extrapolationValue Value used for extrapolation, when applicable.
-     */
-    public Options extrapolationValue(Float extrapolationValue) {
-      this.extrapolationValue = extrapolationValue;
-      return this;
-    }
-    
-    private String method;
-    private Float extrapolationValue;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new CropAndResize operation.
-   * 
-   * @param scope current scope
-   * @param image A 4-D tensor of shape `[batch, image_height, image_width, depth]`.
-   * Both `image_height` and `image_width` need to be positive.
-   * @param boxes A 2-D tensor of shape `[num_boxes, 4]`. The `i`-th row of the tensor
-   * specifies the coordinates of a box in the `box_ind[i]` image and is specified
-   * in normalized coordinates `[y1, x1, y2, x2]`. A normalized coordinate value of
-   * `y` is mapped to the image coordinate at `y * (image_height - 1)`, so as the
-   * `[0, 1]` interval of normalized image height is mapped to
-   * `[0, image_height - 1]` in image height coordinates. We do allow `y1` > `y2`, in
-   * which case the sampled crop is an up-down flipped version of the original
-   * image. The width dimension is treated similarly. Normalized coordinates
-   * outside the `[0, 1]` range are allowed, in which case we use
-   * `extrapolation_value` to extrapolate the input image values.
-   * @param boxInd A 1-D tensor of shape `[num_boxes]` with int32 values in `[0, batch)`.
-   * The value of `box_ind[i]` specifies the image that the `i`-th box refers to.
-   * @param cropSize A 1-D tensor of 2 elements, `size = [crop_height, crop_width]`. All
-   * cropped image patches are resized to this size. The aspect ratio of the image
-   * content is not preserved. Both `crop_height` and `crop_width` need to be
-   * positive.
-   * @param options carries optional attributes values
-   * @return a new instance of CropAndResize
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> CropAndResize create(Scope scope, Operand<T> image, Operand<TFloat32> boxes, Operand<TInt32> boxInd, Operand<TInt32> cropSize, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("CropAndResize", scope.makeOpName("CropAndResize"));
-    opBuilder.addInput(image.asOutput(scope));
-    opBuilder.addInput(boxes.asOutput(scope));
-    opBuilder.addInput(boxInd.asOutput(scope));
-    opBuilder.addInput(cropSize.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.method != null) {
-          opBuilder.setAttr("method", opts.method);
-        }
-        if (opts.extrapolationValue != null) {
-          opBuilder.setAttr("extrapolation_value", opts.extrapolationValue);
-        }
-      }
-    }
-    return new CropAndResize(opBuilder.build());
-  }
-  
-  /**
-   * @param method A string specifying the sampling method for resizing. It can be either
-   * `"bilinear"` or `"nearest"` and default to `"bilinear"`. Currently two sampling
-   * methods are supported: Bilinear and Nearest Neighbor.
-   */
-  public static Options method(String method) {
-    return new Options().method(method);
-  }
-  
-  /**
-   * @param extrapolationValue Value used for extrapolation, when applicable.
-   */
-  public static Options extrapolationValue(Float extrapolationValue) {
-    return new Options().extrapolationValue(extrapolationValue);
-  }
-  
-  /**
-   * A 4-D tensor of shape `[num_boxes, crop_height, crop_width, depth]`.
-   */
-  public Output<TFloat32> crops() {
-    return crops;
-  }
-  
-  @Override
-  public Output<TFloat32> asOutput(Scope scope) {
-    return crops;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "CropAndResize";
-  
-  private Output<TFloat32> crops;
-  
-  private CropAndResize(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    crops = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/CropAndResizeGradBoxes.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/CropAndResizeGradBoxes.java
deleted file mode 100644
index 3806d28fd2a..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/CropAndResizeGradBoxes.java
+++ /dev/null
@@ -1,128 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.image;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Computes the gradient of the crop_and_resize op wrt the input boxes tensor.
- */
-@Operator(group = "image")
-public final class CropAndResizeGradBoxes extends RawOp implements Operand<TFloat32> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.image.CropAndResizeGradBoxes}
-   */
-  public static class Options {
-    
-    /**
-     * @param method A string specifying the interpolation method. Only 'bilinear' is
-     * supported for now.
-     */
-    public Options method(String method) {
-      this.method = method;
-      return this;
-    }
-    
-    private String method;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new CropAndResizeGradBoxes operation.
-   * 
-   * @param scope current scope
-   * @param grads A 4-D tensor of shape `[num_boxes, crop_height, crop_width, depth]`.
-   * @param image A 4-D tensor of shape `[batch, image_height, image_width, depth]`.
-   * Both `image_height` and `image_width` need to be positive.
-   * @param boxes A 2-D tensor of shape `[num_boxes, 4]`. The `i`-th row of the tensor
-   * specifies the coordinates of a box in the `box_ind[i]` image and is specified
-   * in normalized coordinates `[y1, x1, y2, x2]`. A normalized coordinate value of
-   * `y` is mapped to the image coordinate at `y * (image_height - 1)`, so as the
-   * `[0, 1]` interval of normalized image height is mapped to
-   * `[0, image_height - 1] in image height coordinates. We do allow y1 > y2, in
-   * which case the sampled crop is an up-down flipped version of the original
-   * image. The width dimension is treated similarly. Normalized coordinates
-   * outside the `[0, 1]` range are allowed, in which case we use
-   * `extrapolation_value` to extrapolate the input image values.
-   * @param boxInd A 1-D tensor of shape `[num_boxes]` with int32 values in `[0, batch)`.
-   * The value of `box_ind[i]` specifies the image that the `i`-th box refers to.
-   * @param options carries optional attributes values
-   * @return a new instance of CropAndResizeGradBoxes
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> CropAndResizeGradBoxes create(Scope scope, Operand<TFloat32> grads, Operand<T> image, Operand<TFloat32> boxes, Operand<TInt32> boxInd, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("CropAndResizeGradBoxes", scope.makeOpName("CropAndResizeGradBoxes"));
-    opBuilder.addInput(grads.asOutput(scope));
-    opBuilder.addInput(image.asOutput(scope));
-    opBuilder.addInput(boxes.asOutput(scope));
-    opBuilder.addInput(boxInd.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.method != null) {
-          opBuilder.setAttr("method", opts.method);
-        }
-      }
-    }
-    return new CropAndResizeGradBoxes(opBuilder.build());
-  }
-  
-  /**
-   * @param method A string specifying the interpolation method. Only 'bilinear' is
-   * supported for now.
-   */
-  public static Options method(String method) {
-    return new Options().method(method);
-  }
-  
-  /**
-   * A 2-D tensor of shape `[num_boxes, 4]`.
-   */
-  public Output<TFloat32> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<TFloat32> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "CropAndResizeGradBoxes";
-  
-  private Output<TFloat32> output;
-  
-  private CropAndResizeGradBoxes(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/CropAndResizeGradImage.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/CropAndResizeGradImage.java
deleted file mode 100644
index b6a3830a3a6..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/CropAndResizeGradImage.java
+++ /dev/null
@@ -1,133 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.image;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Computes the gradient of the crop_and_resize op wrt the input image tensor.
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator(group = "image")
-public final class CropAndResizeGradImage<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.image.CropAndResizeGradImage}
-   */
-  public static class Options {
-    
-    /**
-     * @param method A string specifying the interpolation method. Only 'bilinear' is
-     * supported for now.
-     */
-    public Options method(String method) {
-      this.method = method;
-      return this;
-    }
-    
-    private String method;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new CropAndResizeGradImage operation.
-   * 
-   * @param scope current scope
-   * @param grads A 4-D tensor of shape `[num_boxes, crop_height, crop_width, depth]`.
-   * @param boxes A 2-D tensor of shape `[num_boxes, 4]`. The `i`-th row of the tensor
-   * specifies the coordinates of a box in the `box_ind[i]` image and is specified
-   * in normalized coordinates `[y1, x1, y2, x2]`. A normalized coordinate value of
-   * `y` is mapped to the image coordinate at `y * (image_height - 1)`, so as the
-   * `[0, 1]` interval of normalized image height is mapped to
-   * `[0, image_height - 1] in image height coordinates. We do allow y1 > y2, in
-   * which case the sampled crop is an up-down flipped version of the original
-   * image. The width dimension is treated similarly. Normalized coordinates
-   * outside the `[0, 1]` range are allowed, in which case we use
-   * `extrapolation_value` to extrapolate the input image values.
-   * @param boxInd A 1-D tensor of shape `[num_boxes]` with int32 values in `[0, batch)`.
-   * The value of `box_ind[i]` specifies the image that the `i`-th box refers to.
-   * @param imageSize A 1-D tensor with value `[batch, image_height, image_width, depth]`
-   * containing the original image size. Both `image_height` and `image_width` need
-   * to be positive.
-   * @param T 
-   * @param options carries optional attributes values
-   * @return a new instance of CropAndResizeGradImage
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> CropAndResizeGradImage<T> create(Scope scope, Operand<TFloat32> grads, Operand<TFloat32> boxes, Operand<TInt32> boxInd, Operand<TInt32> imageSize, Class<T> T, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("CropAndResizeGradImage", scope.makeOpName("CropAndResizeGradImage"));
-    opBuilder.addInput(grads.asOutput(scope));
-    opBuilder.addInput(boxes.asOutput(scope));
-    opBuilder.addInput(boxInd.asOutput(scope));
-    opBuilder.addInput(imageSize.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("T", T);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.method != null) {
-          opBuilder.setAttr("method", opts.method);
-        }
-      }
-    }
-    return new CropAndResizeGradImage<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param method A string specifying the interpolation method. Only 'bilinear' is
-   * supported for now.
-   */
-  public static Options method(String method) {
-    return new Options().method(method);
-  }
-  
-  /**
-   * A 4-D tensor of shape `[batch, image_height, image_width, depth]`.
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "CropAndResizeGradImage";
-  
-  private Output<T> output;
-  
-  private CropAndResizeGradImage(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/DecodeAndCropJpeg.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/DecodeAndCropJpeg.java
deleted file mode 100644
index dc8282ec77a..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/DecodeAndCropJpeg.java
+++ /dev/null
@@ -1,245 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.image;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.TString;
-import org.tensorflow.types.TUint8;
-
-/**
- * Decode and Crop a JPEG-encoded image to a uint8 tensor.
- * <p>
- * The attr `channels` indicates the desired number of color channels for the
- * decoded image.
- * <p>
- * Accepted values are:
- * <ul>
- * <li>
- * 0: Use the number of channels in the JPEG-encoded image.
- * </li>
- * <li>
- * 1: output a grayscale image.
- * </li>
- * <li>
- * 3: output an RGB image.
- * </li>
- * </ul>
- * If needed, the JPEG-encoded image is transformed to match the requested number
- * of color channels.
- * <p>
- * The attr `ratio` allows downscaling the image by an integer factor during
- * decoding.  Allowed values are: 1, 2, 4, and 8.  This is much faster than
- * downscaling the image later.
- * <p>
- * It is equivalent to a combination of decode and crop, but much faster by only
- * decoding partial jpeg image.
- */
-@Operator(group = "image")
-public final class DecodeAndCropJpeg extends RawOp implements Operand<TUint8> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.image.DecodeAndCropJpeg}
-   */
-  public static class Options {
-    
-    /**
-     * @param channels Number of color channels for the decoded image.
-     */
-    public Options channels(Long channels) {
-      this.channels = channels;
-      return this;
-    }
-    
-    /**
-     * @param ratio Downscaling ratio.
-     */
-    public Options ratio(Long ratio) {
-      this.ratio = ratio;
-      return this;
-    }
-    
-    /**
-     * @param fancyUpscaling If true use a slower but nicer upscaling of the
-     * chroma planes (yuv420/422 only).
-     */
-    public Options fancyUpscaling(Boolean fancyUpscaling) {
-      this.fancyUpscaling = fancyUpscaling;
-      return this;
-    }
-    
-    /**
-     * @param tryRecoverTruncated If true try to recover an image from truncated input.
-     */
-    public Options tryRecoverTruncated(Boolean tryRecoverTruncated) {
-      this.tryRecoverTruncated = tryRecoverTruncated;
-      return this;
-    }
-    
-    /**
-     * @param acceptableFraction The minimum required fraction of lines before a truncated
-     * input is accepted.
-     */
-    public Options acceptableFraction(Float acceptableFraction) {
-      this.acceptableFraction = acceptableFraction;
-      return this;
-    }
-    
-    /**
-     * @param dctMethod string specifying a hint about the algorithm used for
-     * decompression.  Defaults to "" which maps to a system-specific
-     * default.  Currently valid values are ["INTEGER_FAST",
-     * "INTEGER_ACCURATE"].  The hint may be ignored (e.g., the internal
-     * jpeg library changes to a version that does not have that specific
-     * option.)
-     */
-    public Options dctMethod(String dctMethod) {
-      this.dctMethod = dctMethod;
-      return this;
-    }
-    
-    private Long channels;
-    private Long ratio;
-    private Boolean fancyUpscaling;
-    private Boolean tryRecoverTruncated;
-    private Float acceptableFraction;
-    private String dctMethod;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new DecodeAndCropJpeg operation.
-   * 
-   * @param scope current scope
-   * @param contents 0-D.  The JPEG-encoded image.
-   * @param cropWindow 1-D.  The crop window: [crop_y, crop_x, crop_height, crop_width].
-   * @param options carries optional attributes values
-   * @return a new instance of DecodeAndCropJpeg
-   */
-  @Endpoint(describeByClass = true)
-  public static DecodeAndCropJpeg create(Scope scope, Operand<TString> contents, Operand<TInt32> cropWindow, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("DecodeAndCropJpeg", scope.makeOpName("DecodeAndCropJpeg"));
-    opBuilder.addInput(contents.asOutput(scope));
-    opBuilder.addInput(cropWindow.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.channels != null) {
-          opBuilder.setAttr("channels", opts.channels);
-        }
-        if (opts.ratio != null) {
-          opBuilder.setAttr("ratio", opts.ratio);
-        }
-        if (opts.fancyUpscaling != null) {
-          opBuilder.setAttr("fancy_upscaling", opts.fancyUpscaling);
-        }
-        if (opts.tryRecoverTruncated != null) {
-          opBuilder.setAttr("try_recover_truncated", opts.tryRecoverTruncated);
-        }
-        if (opts.acceptableFraction != null) {
-          opBuilder.setAttr("acceptable_fraction", opts.acceptableFraction);
-        }
-        if (opts.dctMethod != null) {
-          opBuilder.setAttr("dct_method", opts.dctMethod);
-        }
-      }
-    }
-    return new DecodeAndCropJpeg(opBuilder.build());
-  }
-  
-  /**
-   * @param channels Number of color channels for the decoded image.
-   */
-  public static Options channels(Long channels) {
-    return new Options().channels(channels);
-  }
-  
-  /**
-   * @param ratio Downscaling ratio.
-   */
-  public static Options ratio(Long ratio) {
-    return new Options().ratio(ratio);
-  }
-  
-  /**
-   * @param fancyUpscaling If true use a slower but nicer upscaling of the
-   * chroma planes (yuv420/422 only).
-   */
-  public static Options fancyUpscaling(Boolean fancyUpscaling) {
-    return new Options().fancyUpscaling(fancyUpscaling);
-  }
-  
-  /**
-   * @param tryRecoverTruncated If true try to recover an image from truncated input.
-   */
-  public static Options tryRecoverTruncated(Boolean tryRecoverTruncated) {
-    return new Options().tryRecoverTruncated(tryRecoverTruncated);
-  }
-  
-  /**
-   * @param acceptableFraction The minimum required fraction of lines before a truncated
-   * input is accepted.
-   */
-  public static Options acceptableFraction(Float acceptableFraction) {
-    return new Options().acceptableFraction(acceptableFraction);
-  }
-  
-  /**
-   * @param dctMethod string specifying a hint about the algorithm used for
-   * decompression.  Defaults to "" which maps to a system-specific
-   * default.  Currently valid values are ["INTEGER_FAST",
-   * "INTEGER_ACCURATE"].  The hint may be ignored (e.g., the internal
-   * jpeg library changes to a version that does not have that specific
-   * option.)
-   */
-  public static Options dctMethod(String dctMethod) {
-    return new Options().dctMethod(dctMethod);
-  }
-  
-  /**
-   * 3-D with shape `[height, width, channels]`..
-   */
-  public Output<TUint8> image() {
-    return image;
-  }
-  
-  @Override
-  public Output<TUint8> asOutput(Scope scope) {
-    return image;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "DecodeAndCropJpeg";
-  
-  private Output<TUint8> image;
-  
-  private DecodeAndCropJpeg(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    image = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/DecodeBmp.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/DecodeBmp.java
deleted file mode 100644
index 7ffa278060a..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/DecodeBmp.java
+++ /dev/null
@@ -1,122 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.image;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TString;
-import org.tensorflow.types.TUint8;
-
-/**
- * Decode the first frame of a BMP-encoded image to a uint8 tensor.
- * <p>
- * The attr `channels` indicates the desired number of color channels for the
- * decoded image.
- * <p>
- * Accepted values are:
- * <ul>
- * <li>
- * 0: Use the number of channels in the BMP-encoded image.
- * </li>
- * <li>
- * 3: output an RGB image.
- * </li>
- * <li>
- * 4: output an RGBA image.
- */
-@Operator(group = "image")
-public final class DecodeBmp extends RawOp implements Operand<TUint8> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.image.DecodeBmp}
-   */
-  public static class Options {
-    
-    /**
-     * @param channels 
-     */
-    public Options channels(Long channels) {
-      this.channels = channels;
-      return this;
-    }
-    
-    private Long channels;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new DecodeBmp operation.
-   * 
-   * @param scope current scope
-   * @param contents 0-D.  The BMP-encoded image.
-   * @param options carries optional attributes values
-   * @return a new instance of DecodeBmp
-   */
-  @Endpoint(describeByClass = true)
-  public static DecodeBmp create(Scope scope, Operand<TString> contents, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("DecodeBmp", scope.makeOpName("DecodeBmp"));
-    opBuilder.addInput(contents.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.channels != null) {
-          opBuilder.setAttr("channels", opts.channels);
-        }
-      }
-    }
-    return new DecodeBmp(opBuilder.build());
-  }
-  
-  /**
-   * @param channels 
-   */
-  public static Options channels(Long channels) {
-    return new Options().channels(channels);
-  }
-  
-  /**
-   * 3-D with shape `[height, width, channels]`. RGB order
-   */
-  public Output<TUint8> image() {
-    return image;
-  }
-  
-  @Override
-  public Output<TUint8> asOutput(Scope scope) {
-    return image;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "DecodeBmp";
-  
-  private Output<TUint8> image;
-  
-  private DecodeBmp(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    image = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/DecodeGif.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/DecodeGif.java
deleted file mode 100644
index 6ec523b5a34..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/DecodeGif.java
+++ /dev/null
@@ -1,83 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.image;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TString;
-import org.tensorflow.types.TUint8;
-
-/**
- * Decode the frame(s) of a GIF-encoded image to a uint8 tensor.
- * <p>
- * GIF images with frame or transparency compression are not supported.
- * On Linux and MacOS systems, convert animated GIFs from compressed to
- * uncompressed by running:
- * <p>
- *     convert $src.gif -coalesce $dst.gif
- * <p>
- * This op also supports decoding JPEGs and PNGs, though it is cleaner to use
- * `tf.io.decode_image`.
- */
-@Operator(group = "image")
-public final class DecodeGif extends RawOp implements Operand<TUint8> {
-  
-  /**
-   * Factory method to create a class wrapping a new DecodeGif operation.
-   * 
-   * @param scope current scope
-   * @param contents 0-D.  The GIF-encoded image.
-   * @return a new instance of DecodeGif
-   */
-  @Endpoint(describeByClass = true)
-  public static DecodeGif create(Scope scope, Operand<TString> contents) {
-    OperationBuilder opBuilder = scope.env().opBuilder("DecodeGif", scope.makeOpName("DecodeGif"));
-    opBuilder.addInput(contents.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new DecodeGif(opBuilder.build());
-  }
-  
-  /**
-   * 4-D with shape `[num_frames, height, width, 3]`. RGB channel order.
-   */
-  public Output<TUint8> image() {
-    return image;
-  }
-  
-  @Override
-  public Output<TUint8> asOutput(Scope scope) {
-    return image;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "DecodeGif";
-  
-  private Output<TUint8> image;
-  
-  private DecodeGif(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    image = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/DecodeJpeg.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/DecodeJpeg.java
deleted file mode 100644
index 854b391f83b..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/DecodeJpeg.java
+++ /dev/null
@@ -1,242 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.image;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TString;
-import org.tensorflow.types.TUint8;
-
-/**
- * Decode a JPEG-encoded image to a uint8 tensor.
- * <p>
- * The attr `channels` indicates the desired number of color channels for the
- * decoded image.
- * <p>
- * Accepted values are:
- * <ul>
- * <li>
- * 0: Use the number of channels in the JPEG-encoded image.
- * </li>
- * <li>
- * 1: output a grayscale image.
- * </li>
- * <li>
- * 3: output an RGB image.
- * </li>
- * </ul>
- * If needed, the JPEG-encoded image is transformed to match the requested number
- * of color channels.
- * <p>
- * The attr `ratio` allows downscaling the image by an integer factor during
- * decoding.  Allowed values are: 1, 2, 4, and 8.  This is much faster than
- * downscaling the image later.
- * <p>
- * This op also supports decoding PNGs and non-animated GIFs since the interface is
- * the same, though it is cleaner to use `tf.io.decode_image`.
- */
-@Operator(group = "image")
-public final class DecodeJpeg extends RawOp implements Operand<TUint8> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.image.DecodeJpeg}
-   */
-  public static class Options {
-    
-    /**
-     * @param channels Number of color channels for the decoded image.
-     */
-    public Options channels(Long channels) {
-      this.channels = channels;
-      return this;
-    }
-    
-    /**
-     * @param ratio Downscaling ratio.
-     */
-    public Options ratio(Long ratio) {
-      this.ratio = ratio;
-      return this;
-    }
-    
-    /**
-     * @param fancyUpscaling If true use a slower but nicer upscaling of the
-     * chroma planes (yuv420/422 only).
-     */
-    public Options fancyUpscaling(Boolean fancyUpscaling) {
-      this.fancyUpscaling = fancyUpscaling;
-      return this;
-    }
-    
-    /**
-     * @param tryRecoverTruncated If true try to recover an image from truncated input.
-     */
-    public Options tryRecoverTruncated(Boolean tryRecoverTruncated) {
-      this.tryRecoverTruncated = tryRecoverTruncated;
-      return this;
-    }
-    
-    /**
-     * @param acceptableFraction The minimum required fraction of lines before a truncated
-     * input is accepted.
-     */
-    public Options acceptableFraction(Float acceptableFraction) {
-      this.acceptableFraction = acceptableFraction;
-      return this;
-    }
-    
-    /**
-     * @param dctMethod string specifying a hint about the algorithm used for
-     * decompression.  Defaults to "" which maps to a system-specific
-     * default.  Currently valid values are ["INTEGER_FAST",
-     * "INTEGER_ACCURATE"].  The hint may be ignored (e.g., the internal
-     * jpeg library changes to a version that does not have that specific
-     * option.)
-     */
-    public Options dctMethod(String dctMethod) {
-      this.dctMethod = dctMethod;
-      return this;
-    }
-    
-    private Long channels;
-    private Long ratio;
-    private Boolean fancyUpscaling;
-    private Boolean tryRecoverTruncated;
-    private Float acceptableFraction;
-    private String dctMethod;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new DecodeJpeg operation.
-   * 
-   * @param scope current scope
-   * @param contents 0-D.  The JPEG-encoded image.
-   * @param options carries optional attributes values
-   * @return a new instance of DecodeJpeg
-   */
-  @Endpoint(describeByClass = true)
-  public static DecodeJpeg create(Scope scope, Operand<TString> contents, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("DecodeJpeg", scope.makeOpName("DecodeJpeg"));
-    opBuilder.addInput(contents.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.channels != null) {
-          opBuilder.setAttr("channels", opts.channels);
-        }
-        if (opts.ratio != null) {
-          opBuilder.setAttr("ratio", opts.ratio);
-        }
-        if (opts.fancyUpscaling != null) {
-          opBuilder.setAttr("fancy_upscaling", opts.fancyUpscaling);
-        }
-        if (opts.tryRecoverTruncated != null) {
-          opBuilder.setAttr("try_recover_truncated", opts.tryRecoverTruncated);
-        }
-        if (opts.acceptableFraction != null) {
-          opBuilder.setAttr("acceptable_fraction", opts.acceptableFraction);
-        }
-        if (opts.dctMethod != null) {
-          opBuilder.setAttr("dct_method", opts.dctMethod);
-        }
-      }
-    }
-    return new DecodeJpeg(opBuilder.build());
-  }
-  
-  /**
-   * @param channels Number of color channels for the decoded image.
-   */
-  public static Options channels(Long channels) {
-    return new Options().channels(channels);
-  }
-  
-  /**
-   * @param ratio Downscaling ratio.
-   */
-  public static Options ratio(Long ratio) {
-    return new Options().ratio(ratio);
-  }
-  
-  /**
-   * @param fancyUpscaling If true use a slower but nicer upscaling of the
-   * chroma planes (yuv420/422 only).
-   */
-  public static Options fancyUpscaling(Boolean fancyUpscaling) {
-    return new Options().fancyUpscaling(fancyUpscaling);
-  }
-  
-  /**
-   * @param tryRecoverTruncated If true try to recover an image from truncated input.
-   */
-  public static Options tryRecoverTruncated(Boolean tryRecoverTruncated) {
-    return new Options().tryRecoverTruncated(tryRecoverTruncated);
-  }
-  
-  /**
-   * @param acceptableFraction The minimum required fraction of lines before a truncated
-   * input is accepted.
-   */
-  public static Options acceptableFraction(Float acceptableFraction) {
-    return new Options().acceptableFraction(acceptableFraction);
-  }
-  
-  /**
-   * @param dctMethod string specifying a hint about the algorithm used for
-   * decompression.  Defaults to "" which maps to a system-specific
-   * default.  Currently valid values are ["INTEGER_FAST",
-   * "INTEGER_ACCURATE"].  The hint may be ignored (e.g., the internal
-   * jpeg library changes to a version that does not have that specific
-   * option.)
-   */
-  public static Options dctMethod(String dctMethod) {
-    return new Options().dctMethod(dctMethod);
-  }
-  
-  /**
-   * 3-D with shape `[height, width, channels]`..
-   */
-  public Output<TUint8> image() {
-    return image;
-  }
-  
-  @Override
-  public Output<TUint8> asOutput(Scope scope) {
-    return image;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "DecodeJpeg";
-  
-  private Output<TUint8> image;
-  
-  private DecodeJpeg(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    image = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/DecodePng.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/DecodePng.java
deleted file mode 100644
index 7cb68f21721..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/DecodePng.java
+++ /dev/null
@@ -1,150 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.image;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TString;
-import org.tensorflow.types.TUint8;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Decode a PNG-encoded image to a uint8 or uint16 tensor.
- * <p>
- * The attr `channels` indicates the desired number of color channels for the
- * decoded image.
- * <p>
- * Accepted values are:
- * <ul>
- * <li>
- * 0: Use the number of channels in the PNG-encoded image.
- * </li>
- * <li>
- * 1: output a grayscale image.
- * </li>
- * <li>
- * 3: output an RGB image.
- * </li>
- * <li>
- * 4: output an RGBA image.
- * </li>
- * </ul>
- * If needed, the PNG-encoded image is transformed to match the requested number
- * of color channels.
- * <p>
- * This op also supports decoding JPEGs and non-animated GIFs since the interface
- * is the same, though it is cleaner to use `tf.io.decode_image`.
- * 
- * @param <T> data type for {@code image()} output
- */
-@Operator(group = "image")
-public final class DecodePng<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.image.DecodePng}
-   */
-  public static class Options {
-    
-    /**
-     * @param channels Number of color channels for the decoded image.
-     */
-    public Options channels(Long channels) {
-      this.channels = channels;
-      return this;
-    }
-    
-    private Long channels;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new DecodePng operation.
-   * 
-   * @param scope current scope
-   * @param contents 0-D.  The PNG-encoded image.
-   * @param dtype 
-   * @param options carries optional attributes values
-   * @return a new instance of DecodePng
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> DecodePng<T> create(Scope scope, Operand<TString> contents, Class<T> dtype, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("DecodePng", scope.makeOpName("DecodePng"));
-    opBuilder.addInput(contents.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("dtype", dtype);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.channels != null) {
-          opBuilder.setAttr("channels", opts.channels);
-        }
-      }
-    }
-    return new DecodePng<T>(opBuilder.build());
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new DecodePng operation using default output types.
-   * 
-   * @param scope current scope
-   * @param contents 0-D.  The PNG-encoded image.
-   * @param options carries optional attributes values
-   * @return a new instance of DecodePng
-   */
-  @Endpoint(describeByClass = true)
-  public static DecodePng<TUint8> create(Scope scope, Operand<TString> contents, Options... options) {
-    return create(scope, contents, TUint8.class, options);
-  }
-  
-  /**
-   * @param channels Number of color channels for the decoded image.
-   */
-  public static Options channels(Long channels) {
-    return new Options().channels(channels);
-  }
-  
-  /**
-   * 3-D with shape `[height, width, channels]`.
-   */
-  public Output<T> image() {
-    return image;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return image;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "DecodePng";
-  
-  private Output<T> image;
-  
-  private DecodePng(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    image = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/DrawBoundingBoxes.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/DrawBoundingBoxes.java
deleted file mode 100644
index ad0e84f294d..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/DrawBoundingBoxes.java
+++ /dev/null
@@ -1,94 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.image;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Draw bounding boxes on a batch of images.
- * <p>
- * Outputs a copy of `images` but draws on top of the pixels zero or more bounding
- * boxes specified by the locations in `boxes`. The coordinates of the each
- * bounding box in `boxes` are encoded as `[y_min, x_min, y_max, x_max]`. The
- * bounding box coordinates are floats in `[0.0, 1.0]` relative to the width and
- * height of the underlying image.
- * <p>
- * For example, if an image is 100 x 200 pixels (height x width) and the bounding
- * box is `[0.1, 0.2, 0.5, 0.9]`, the upper-left and bottom-right coordinates of
- * the bounding box will be `(40, 10)` to `(100, 50)` (in (x,y) coordinates).
- * <p>
- * Parts of the bounding box may fall outside the image.
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator(group = "image")
-public final class DrawBoundingBoxes<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new DrawBoundingBoxes operation.
-   * 
-   * @param scope current scope
-   * @param images 4-D with shape `[batch, height, width, depth]`. A batch of images.
-   * @param boxes 3-D with shape `[batch, num_bounding_boxes, 4]` containing bounding
-   * boxes.
-   * @param colors 2-D. A list of RGBA colors to cycle through for the boxes.
-   * @return a new instance of DrawBoundingBoxes
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> DrawBoundingBoxes<T> create(Scope scope, Operand<T> images, Operand<TFloat32> boxes, Operand<TFloat32> colors) {
-    OperationBuilder opBuilder = scope.env().opBuilder("DrawBoundingBoxesV2", scope.makeOpName("DrawBoundingBoxes"));
-    opBuilder.addInput(images.asOutput(scope));
-    opBuilder.addInput(boxes.asOutput(scope));
-    opBuilder.addInput(colors.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new DrawBoundingBoxes<T>(opBuilder.build());
-  }
-  
-  /**
-   * 4-D with the same shape as `images`. The batch of input images with
-   * bounding boxes drawn on the images.
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "DrawBoundingBoxesV2";
-  
-  private Output<T> output;
-  
-  private DrawBoundingBoxes(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/EncodeJpeg.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/EncodeJpeg.java
deleted file mode 100644
index 6d504201441..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/EncodeJpeg.java
+++ /dev/null
@@ -1,288 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.image;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TString;
-import org.tensorflow.types.TUint8;
-
-/**
- * JPEG-encode an image.
- * <p>
- * `image` is a 3-D uint8 Tensor of shape `[height, width, channels]`.
- * <p>
- * The attr `format` can be used to override the color format of the encoded
- * output.  Values can be:
- * <ul>
- * <li>
- * `''`: Use a default format based on the number of channels in the image.
- * </li>
- * <li>
- * `grayscale`: Output a grayscale JPEG image.  The `channels` dimension
- *     of `image` must be 1.
- * </li>
- * <li>
- * `rgb`: Output an RGB JPEG image. The `channels` dimension
- *     of `image` must be 3.
- * </li>
- * </ul>
- * If `format` is not specified or is the empty string, a default format is picked
- * in function of the number of channels in `image`:
- * <ul>
- * <li>
- * 1: Output a grayscale image.
- * </li>
- * <li>
- * 3: Output an RGB image.
- */
-@Operator(group = "image")
-public final class EncodeJpeg extends RawOp implements Operand<TString> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.image.EncodeJpeg}
-   */
-  public static class Options {
-    
-    /**
-     * @param format Per pixel image format.
-     */
-    public Options format(String format) {
-      this.format = format;
-      return this;
-    }
-    
-    /**
-     * @param quality Quality of the compression from 0 to 100 (higher is better and slower).
-     */
-    public Options quality(Long quality) {
-      this.quality = quality;
-      return this;
-    }
-    
-    /**
-     * @param progressive If True, create a JPEG that loads progressively (coarse to fine).
-     */
-    public Options progressive(Boolean progressive) {
-      this.progressive = progressive;
-      return this;
-    }
-    
-    /**
-     * @param optimizeSize If True, spend CPU/RAM to reduce size with no quality change.
-     */
-    public Options optimizeSize(Boolean optimizeSize) {
-      this.optimizeSize = optimizeSize;
-      return this;
-    }
-    
-    /**
-     * @param chromaDownsampling See http://en.wikipedia.org/wiki/Chroma_subsampling.
-     */
-    public Options chromaDownsampling(Boolean chromaDownsampling) {
-      this.chromaDownsampling = chromaDownsampling;
-      return this;
-    }
-    
-    /**
-     * @param densityUnit Unit used to specify `x_density` and `y_density`:
-     * pixels per inch (`'in'`) or centimeter (`'cm'`).
-     */
-    public Options densityUnit(String densityUnit) {
-      this.densityUnit = densityUnit;
-      return this;
-    }
-    
-    /**
-     * @param xDensity Horizontal pixels per density unit.
-     */
-    public Options xDensity(Long xDensity) {
-      this.xDensity = xDensity;
-      return this;
-    }
-    
-    /**
-     * @param yDensity Vertical pixels per density unit.
-     */
-    public Options yDensity(Long yDensity) {
-      this.yDensity = yDensity;
-      return this;
-    }
-    
-    /**
-     * @param xmpMetadata If not empty, embed this XMP metadata in the image header.
-     */
-    public Options xmpMetadata(String xmpMetadata) {
-      this.xmpMetadata = xmpMetadata;
-      return this;
-    }
-    
-    private String format;
-    private Long quality;
-    private Boolean progressive;
-    private Boolean optimizeSize;
-    private Boolean chromaDownsampling;
-    private String densityUnit;
-    private Long xDensity;
-    private Long yDensity;
-    private String xmpMetadata;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new EncodeJpeg operation.
-   * 
-   * @param scope current scope
-   * @param image 3-D with shape `[height, width, channels]`.
-   * @param options carries optional attributes values
-   * @return a new instance of EncodeJpeg
-   */
-  @Endpoint(describeByClass = true)
-  public static EncodeJpeg create(Scope scope, Operand<TUint8> image, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("EncodeJpeg", scope.makeOpName("EncodeJpeg"));
-    opBuilder.addInput(image.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.format != null) {
-          opBuilder.setAttr("format", opts.format);
-        }
-        if (opts.quality != null) {
-          opBuilder.setAttr("quality", opts.quality);
-        }
-        if (opts.progressive != null) {
-          opBuilder.setAttr("progressive", opts.progressive);
-        }
-        if (opts.optimizeSize != null) {
-          opBuilder.setAttr("optimize_size", opts.optimizeSize);
-        }
-        if (opts.chromaDownsampling != null) {
-          opBuilder.setAttr("chroma_downsampling", opts.chromaDownsampling);
-        }
-        if (opts.densityUnit != null) {
-          opBuilder.setAttr("density_unit", opts.densityUnit);
-        }
-        if (opts.xDensity != null) {
-          opBuilder.setAttr("x_density", opts.xDensity);
-        }
-        if (opts.yDensity != null) {
-          opBuilder.setAttr("y_density", opts.yDensity);
-        }
-        if (opts.xmpMetadata != null) {
-          opBuilder.setAttr("xmp_metadata", opts.xmpMetadata);
-        }
-      }
-    }
-    return new EncodeJpeg(opBuilder.build());
-  }
-  
-  /**
-   * @param format Per pixel image format.
-   */
-  public static Options format(String format) {
-    return new Options().format(format);
-  }
-  
-  /**
-   * @param quality Quality of the compression from 0 to 100 (higher is better and slower).
-   */
-  public static Options quality(Long quality) {
-    return new Options().quality(quality);
-  }
-  
-  /**
-   * @param progressive If True, create a JPEG that loads progressively (coarse to fine).
-   */
-  public static Options progressive(Boolean progressive) {
-    return new Options().progressive(progressive);
-  }
-  
-  /**
-   * @param optimizeSize If True, spend CPU/RAM to reduce size with no quality change.
-   */
-  public static Options optimizeSize(Boolean optimizeSize) {
-    return new Options().optimizeSize(optimizeSize);
-  }
-  
-  /**
-   * @param chromaDownsampling See http://en.wikipedia.org/wiki/Chroma_subsampling.
-   */
-  public static Options chromaDownsampling(Boolean chromaDownsampling) {
-    return new Options().chromaDownsampling(chromaDownsampling);
-  }
-  
-  /**
-   * @param densityUnit Unit used to specify `x_density` and `y_density`:
-   * pixels per inch (`'in'`) or centimeter (`'cm'`).
-   */
-  public static Options densityUnit(String densityUnit) {
-    return new Options().densityUnit(densityUnit);
-  }
-  
-  /**
-   * @param xDensity Horizontal pixels per density unit.
-   */
-  public static Options xDensity(Long xDensity) {
-    return new Options().xDensity(xDensity);
-  }
-  
-  /**
-   * @param yDensity Vertical pixels per density unit.
-   */
-  public static Options yDensity(Long yDensity) {
-    return new Options().yDensity(yDensity);
-  }
-  
-  /**
-   * @param xmpMetadata If not empty, embed this XMP metadata in the image header.
-   */
-  public static Options xmpMetadata(String xmpMetadata) {
-    return new Options().xmpMetadata(xmpMetadata);
-  }
-  
-  /**
-   * 0-D. JPEG-encoded image.
-   */
-  public Output<TString> contents() {
-    return contents;
-  }
-  
-  @Override
-  public Output<TString> asOutput(Scope scope) {
-    return contents;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "EncodeJpeg";
-  
-  private Output<TString> contents;
-  
-  private EncodeJpeg(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    contents = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/EncodeJpegVariableQuality.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/EncodeJpegVariableQuality.java
deleted file mode 100644
index 84a39b998c1..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/EncodeJpegVariableQuality.java
+++ /dev/null
@@ -1,81 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.image;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.TString;
-import org.tensorflow.types.TUint8;
-
-/**
- * JPEG encode input image with provided compression quality.
- * <p>
- * `image` is a 3-D uint8 Tensor of shape `[height, width, channels]`.
- * `quality` is an int32 jpeg compression quality value between 0 and 100.
- * 
- */
-@Operator(group = "image")
-public final class EncodeJpegVariableQuality extends RawOp implements Operand<TString> {
-  
-  /**
-   * Factory method to create a class wrapping a new EncodeJpegVariableQuality operation.
-   * 
-   * @param scope current scope
-   * @param images Images to adjust.  At least 3-D.
-   * @param quality An int quality to encode to.
-   * @return a new instance of EncodeJpegVariableQuality
-   */
-  @Endpoint(describeByClass = true)
-  public static EncodeJpegVariableQuality create(Scope scope, Operand<TUint8> images, Operand<TInt32> quality) {
-    OperationBuilder opBuilder = scope.env().opBuilder("EncodeJpegVariableQuality", scope.makeOpName("EncodeJpegVariableQuality"));
-    opBuilder.addInput(images.asOutput(scope));
-    opBuilder.addInput(quality.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new EncodeJpegVariableQuality(opBuilder.build());
-  }
-  
-  /**
-   * 0-D. JPEG-encoded image.
-   */
-  public Output<TString> contents() {
-    return contents;
-  }
-  
-  @Override
-  public Output<TString> asOutput(Scope scope) {
-    return contents;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "EncodeJpegVariableQuality";
-  
-  private Output<TString> contents;
-  
-  private EncodeJpegVariableQuality(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    contents = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/EncodePng.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/EncodePng.java
deleted file mode 100644
index a43391c0264..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/EncodePng.java
+++ /dev/null
@@ -1,128 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.image;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TString;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * PNG-encode an image.
- * <p>
- * `image` is a 3-D uint8 or uint16 Tensor of shape `[height, width, channels]`
- * where `channels` is:
- * <ul>
- * <li>
- * 1: for grayscale.
- * </li>
- * <li>
- * 2: for grayscale + alpha.
- * </li>
- * <li>
- * 3: for RGB.
- * </li>
- * <li>
- * 4: for RGBA.
- * </li>
- * </ul>
- * The ZLIB compression level, `compression`, can be -1 for the PNG-encoder
- * default or a value from 0 to 9.  9 is the highest compression level, generating
- * the smallest output, but is slower.
- */
-@Operator(group = "image")
-public final class EncodePng extends RawOp implements Operand<TString> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.image.EncodePng}
-   */
-  public static class Options {
-    
-    /**
-     * @param compression Compression level.
-     */
-    public Options compression(Long compression) {
-      this.compression = compression;
-      return this;
-    }
-    
-    private Long compression;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new EncodePng operation.
-   * 
-   * @param scope current scope
-   * @param image 3-D with shape `[height, width, channels]`.
-   * @param options carries optional attributes values
-   * @return a new instance of EncodePng
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> EncodePng create(Scope scope, Operand<T> image, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("EncodePng", scope.makeOpName("EncodePng"));
-    opBuilder.addInput(image.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.compression != null) {
-          opBuilder.setAttr("compression", opts.compression);
-        }
-      }
-    }
-    return new EncodePng(opBuilder.build());
-  }
-  
-  /**
-   * @param compression Compression level.
-   */
-  public static Options compression(Long compression) {
-    return new Options().compression(compression);
-  }
-  
-  /**
-   * 0-D. PNG-encoded image.
-   */
-  public Output<TString> contents() {
-    return contents;
-  }
-  
-  @Override
-  public Output<TString> asOutput(Scope scope) {
-    return contents;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "EncodePng";
-  
-  private Output<TString> contents;
-  
-  private EncodePng(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    contents = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ExtractGlimpse.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ExtractGlimpse.java
deleted file mode 100644
index 59f46dd1947..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ExtractGlimpse.java
+++ /dev/null
@@ -1,211 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.image;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.TInt32;
-
-/**
- * Extracts a glimpse from the input tensor.
- * <p>
- * Returns a set of windows called glimpses extracted at location
- * `offsets` from the input tensor. If the windows only partially
- * overlaps the inputs, the non overlapping areas will be filled with
- * random noise.
- * <p>
- * The result is a 4-D tensor of shape `[batch_size, glimpse_height,
- * glimpse_width, channels]`. The channels and batch dimensions are the
- * same as that of the input tensor. The height and width of the output
- * windows are specified in the `size` parameter.
- * <p>
- * The argument `normalized` and `centered` controls how the windows are built:
- * <ul>
- * <li>
- * If the coordinates are normalized but not centered, 0.0 and 1.0
- *   correspond to the minimum and maximum of each height and width
- *   dimension.
- * </li>
- * <li>
- * If the coordinates are both normalized and centered, they range from
- *   -1.0 to 1.0. The coordinates (-1.0, -1.0) correspond to the upper
- *   left corner, the lower right corner is located at (1.0, 1.0) and the
- *   center is at (0, 0).
- * </li>
- * <li>
- * If the coordinates are not normalized they are interpreted as
- *   numbers of pixels.
- */
-public final class ExtractGlimpse extends RawOp implements Operand<TFloat32> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.image.ExtractGlimpse}
-   */
-  public static class Options {
-    
-    /**
-     * @param centered indicates if the offset coordinates are centered relative to
-     * the image, in which case the (0, 0) offset is relative to the center
-     * of the input images. If false, the (0,0) offset corresponds to the
-     * upper left corner of the input images.
-     */
-    public Options centered(Boolean centered) {
-      this.centered = centered;
-      return this;
-    }
-    
-    /**
-     * @param normalized indicates if the offset coordinates are normalized.
-     */
-    public Options normalized(Boolean normalized) {
-      this.normalized = normalized;
-      return this;
-    }
-    
-    /**
-     * @param uniformNoise indicates if the noise should be generated using a
-     * uniform distribution or a Gaussian distribution.
-     */
-    public Options uniformNoise(Boolean uniformNoise) {
-      this.uniformNoise = uniformNoise;
-      return this;
-    }
-    
-    /**
-     * @param noise indicates if the noise should `uniform`, `gaussian`, or
-     * `zero`. The default is `uniform` which means the the noise type
-     * will be decided by `uniform_noise`.
-     */
-    public Options noise(String noise) {
-      this.noise = noise;
-      return this;
-    }
-    
-    private Boolean centered;
-    private Boolean normalized;
-    private Boolean uniformNoise;
-    private String noise;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new ExtractGlimpse operation.
-   * 
-   * @param scope current scope
-   * @param input A 4-D float tensor of shape `[batch_size, height, width, channels]`.
-   * @param size A 1-D tensor of 2 elements containing the size of the glimpses
-   * to extract.  The glimpse height must be specified first, following
-   * by the glimpse width.
-   * @param offsets A 2-D integer tensor of shape `[batch_size, 2]` containing
-   * the y, x locations of the center of each window.
-   * @param options carries optional attributes values
-   * @return a new instance of ExtractGlimpse
-   */
-  @Endpoint(describeByClass = true)
-  public static ExtractGlimpse create(Scope scope, Operand<TFloat32> input, Operand<TInt32> size, Operand<TFloat32> offsets, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ExtractGlimpseV2", scope.makeOpName("ExtractGlimpse"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder.addInput(size.asOutput(scope));
-    opBuilder.addInput(offsets.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.centered != null) {
-          opBuilder.setAttr("centered", opts.centered);
-        }
-        if (opts.normalized != null) {
-          opBuilder.setAttr("normalized", opts.normalized);
-        }
-        if (opts.uniformNoise != null) {
-          opBuilder.setAttr("uniform_noise", opts.uniformNoise);
-        }
-        if (opts.noise != null) {
-          opBuilder.setAttr("noise", opts.noise);
-        }
-      }
-    }
-    return new ExtractGlimpse(opBuilder.build());
-  }
-  
-  /**
-   * @param centered indicates if the offset coordinates are centered relative to
-   * the image, in which case the (0, 0) offset is relative to the center
-   * of the input images. If false, the (0,0) offset corresponds to the
-   * upper left corner of the input images.
-   */
-  public static Options centered(Boolean centered) {
-    return new Options().centered(centered);
-  }
-  
-  /**
-   * @param normalized indicates if the offset coordinates are normalized.
-   */
-  public static Options normalized(Boolean normalized) {
-    return new Options().normalized(normalized);
-  }
-  
-  /**
-   * @param uniformNoise indicates if the noise should be generated using a
-   * uniform distribution or a Gaussian distribution.
-   */
-  public static Options uniformNoise(Boolean uniformNoise) {
-    return new Options().uniformNoise(uniformNoise);
-  }
-  
-  /**
-   * @param noise indicates if the noise should `uniform`, `gaussian`, or
-   * `zero`. The default is `uniform` which means the the noise type
-   * will be decided by `uniform_noise`.
-   */
-  public static Options noise(String noise) {
-    return new Options().noise(noise);
-  }
-  
-  /**
-   * A tensor representing the glimpses `[batch_size,
-   * glimpse_height, glimpse_width, channels]`.
-   */
-  public Output<TFloat32> glimpse() {
-    return glimpse;
-  }
-  
-  @Override
-  public Output<TFloat32> asOutput(Scope scope) {
-    return glimpse;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ExtractGlimpseV2";
-  
-  private Output<TFloat32> glimpse;
-  
-  private ExtractGlimpse(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    glimpse = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ExtractImagePatches.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ExtractImagePatches.java
deleted file mode 100644
index 7e9f055783f..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ExtractImagePatches.java
+++ /dev/null
@@ -1,105 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.image;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Extract `patches` from `images` and put them in the "depth" output dimension.
- * 
- * @param <T> data type for {@code patches()} output
- */
-@Operator(group = "image")
-public final class ExtractImagePatches<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new ExtractImagePatches operation.
-   * 
-   * @param scope current scope
-   * @param images 4-D Tensor with shape `[batch, in_rows, in_cols, depth]`.
-   * @param ksizes The size of the sliding window for each dimension of `images`.
-   * @param strides How far the centers of two consecutive patches are in
-   * the images. Must be: `[1, stride_rows, stride_cols, 1]`.
-   * @param rates Must be: `[1, rate_rows, rate_cols, 1]`. This is the
-   * input stride, specifying how far two consecutive patch samples are in the
-   * input. Equivalent to extracting patches with
-   * `patch_sizes_eff = patch_sizes + (patch_sizes - 1) * (rates - 1)`, followed by
-   * subsampling them spatially by a factor of `rates`. This is equivalent to
-   * `rate` in dilated (a.k.a. Atrous) convolutions.
-   * @param padding The type of padding algorithm to use.
-   * @return a new instance of ExtractImagePatches
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> ExtractImagePatches<T> create(Scope scope, Operand<T> images, List<Long> ksizes, List<Long> strides, List<Long> rates, String padding) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ExtractImagePatches", scope.makeOpName("ExtractImagePatches"));
-    opBuilder.addInput(images.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    long[] ksizesArray = new long[ksizes.size()];
-    for (int i = 0; i < ksizesArray.length; ++i) {
-      ksizesArray[i] = ksizes.get(i);
-    }
-    opBuilder.setAttr("ksizes", ksizesArray);
-    long[] stridesArray = new long[strides.size()];
-    for (int i = 0; i < stridesArray.length; ++i) {
-      stridesArray[i] = strides.get(i);
-    }
-    opBuilder.setAttr("strides", stridesArray);
-    long[] ratesArray = new long[rates.size()];
-    for (int i = 0; i < ratesArray.length; ++i) {
-      ratesArray[i] = rates.get(i);
-    }
-    opBuilder.setAttr("rates", ratesArray);
-    opBuilder.setAttr("padding", padding);
-    return new ExtractImagePatches<T>(opBuilder.build());
-  }
-  
-  /**
-   * 4-D Tensor with shape `[batch, out_rows, out_cols, ksize_rows *
-   * ksize_cols * depth]` containing image patches with size
-   * `ksize_rows x ksize_cols x depth` vectorized in the "depth" dimension. Note
-   * `out_rows` and `out_cols` are the dimensions of the output patches.
-   */
-  public Output<T> patches() {
-    return patches;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return patches;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ExtractImagePatches";
-  
-  private Output<T> patches;
-  
-  private ExtractImagePatches(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    patches = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ExtractJpegShape.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ExtractJpegShape.java
deleted file mode 100644
index 467fa8820ed..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ExtractJpegShape.java
+++ /dev/null
@@ -1,94 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.image;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.TString;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Extract the shape information of a JPEG-encoded image.
- * <p>
- * This op only parses the image header, so it is much faster than DecodeJpeg.
- * 
- * @param <T> data type for {@code imageShape()} output
- */
-@Operator(group = "image")
-public final class ExtractJpegShape<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new ExtractJpegShape operation.
-   * 
-   * @param scope current scope
-   * @param contents 0-D. The JPEG-encoded image.
-   * @param outputType (Optional) The output type of the operation (int32 or int64).
-   * Defaults to int32.
-   * @return a new instance of ExtractJpegShape
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> ExtractJpegShape<T> create(Scope scope, Operand<TString> contents, Class<T> outputType) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ExtractJpegShape", scope.makeOpName("ExtractJpegShape"));
-    opBuilder.addInput(contents.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("output_type", outputType);
-    return new ExtractJpegShape<T>(opBuilder.build());
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new ExtractJpegShape operation using default output types.
-   * 
-   * @param scope current scope
-   * @param contents 0-D. The JPEG-encoded image.
-   * @return a new instance of ExtractJpegShape
-   */
-  @Endpoint(describeByClass = true)
-  public static ExtractJpegShape<TInt32> create(Scope scope, Operand<TString> contents) {
-    return create(scope, contents, TInt32.class);
-  }
-  
-  /**
-   * 1-D. The image shape with format [height, width, channels].
-   */
-  public Output<T> imageShape() {
-    return imageShape;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return imageShape;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ExtractJpegShape";
-  
-  private Output<T> imageShape;
-  
-  private ExtractJpegShape(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    imageShape = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/GenerateBoundingBoxProposals.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/GenerateBoundingBoxProposals.java
deleted file mode 100644
index 4b83bea8c54..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/GenerateBoundingBoxProposals.java
+++ /dev/null
@@ -1,138 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.image;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.TInt32;
-
-/**
- * This op produces Region of Interests from given bounding boxes(bbox_deltas) encoded wrt anchors according to eq.2 in arXiv:1506.01497
- * <p>
- *       The op selects top `pre_nms_topn` scoring boxes, decodes them with respect to anchors,
- *       applies non-maximal suppression on overlapping boxes with higher than
- *       `nms_threshold` intersection-over-union (iou) value, discarding boxes where shorter
- *       side is less than `min_size`.
- *       Inputs:
- *       `scores`: A 4D tensor of shape [Batch, Height, Width, Num Anchors] containing the scores per anchor at given position
- *       `bbox_deltas`: is a tensor of shape [Batch, Height, Width, 4 x Num Anchors] boxes encoded to each anchor
- *       `anchors`: A 1D tensor of shape [4 x Num Anchors], representing the anchors.
- *       Outputs:
- *       `rois`: output RoIs, a 3D tensor of shape [Batch, post_nms_topn, 4], padded by 0 if less than post_nms_topn candidates found.
- *       `roi_probabilities`: probability scores of each roi in 'rois', a 2D tensor of shape [Batch,post_nms_topn], padded with 0 if needed, sorted by scores.
- */
-public final class GenerateBoundingBoxProposals extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.image.GenerateBoundingBoxProposals}
-   */
-  public static class Options {
-    
-    /**
-     * @param postNmsTopn An integer. Maximum number of rois in the output.
-     */
-    public Options postNmsTopn(Long postNmsTopn) {
-      this.postNmsTopn = postNmsTopn;
-      return this;
-    }
-    
-    private Long postNmsTopn;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new GenerateBoundingBoxProposals operation.
-   * 
-   * @param scope current scope
-   * @param scores A 4-D float tensor of shape `[num_images, height, width, num_achors]` containing scores of the boxes for given anchors, can be unsorted.
-   * @param bboxDeltas A 4-D float tensor of shape `[num_images, height, width, 4 x num_anchors]`. encoding boxes with respec to each anchor.
-   * Coordinates are given in the form [dy, dx, dh, dw].
-   * @param imageInfo A 2-D float tensor of shape `[num_images, 5]` containing image information Height, Width, Scale.
-   * @param anchors A 2-D float tensor of shape `[num_anchors, 4]` describing the anchor boxes. Boxes are formatted in the form [y1, x1, y2, x2].
-   * @param nmsThreshold A scalar float tensor for non-maximal-suppression threshold.
-   * @param preNmsTopn A scalar int tensor for the number of top scoring boxes to be used as input.
-   * @param minSize A scalar float tensor. Any box that has a smaller size than min_size will be discarded.
-   * @param options carries optional attributes values
-   * @return a new instance of GenerateBoundingBoxProposals
-   */
-  @Endpoint(describeByClass = true)
-  public static GenerateBoundingBoxProposals create(Scope scope, Operand<TFloat32> scores, Operand<TFloat32> bboxDeltas, Operand<TFloat32> imageInfo, Operand<TFloat32> anchors, Operand<TFloat32> nmsThreshold, Operand<TInt32> preNmsTopn, Operand<TFloat32> minSize, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("GenerateBoundingBoxProposals", scope.makeOpName("GenerateBoundingBoxProposals"));
-    opBuilder.addInput(scores.asOutput(scope));
-    opBuilder.addInput(bboxDeltas.asOutput(scope));
-    opBuilder.addInput(imageInfo.asOutput(scope));
-    opBuilder.addInput(anchors.asOutput(scope));
-    opBuilder.addInput(nmsThreshold.asOutput(scope));
-    opBuilder.addInput(preNmsTopn.asOutput(scope));
-    opBuilder.addInput(minSize.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.postNmsTopn != null) {
-          opBuilder.setAttr("post_nms_topn", opts.postNmsTopn);
-        }
-      }
-    }
-    return new GenerateBoundingBoxProposals(opBuilder.build());
-  }
-  
-  /**
-   * @param postNmsTopn An integer. Maximum number of rois in the output.
-   */
-  public static Options postNmsTopn(Long postNmsTopn) {
-    return new Options().postNmsTopn(postNmsTopn);
-  }
-  
-  /**
-   * A 3-D float tensor of shape `[num_images,post_nms_topn,4]` representing the selected
-   * region of interest boxes. Sorted in descending order in scores.
-   */
-  public Output<TFloat32> rois() {
-    return rois;
-  }
-  
-  /**
-   * A 2-D float tensor of shape `[num_images, post_nms_topn]` representing the score of the
-   * region of interest box in `rois` tensor at the same index.
-   */
-  public Output<TFloat32> roiProbabilities() {
-    return roiProbabilities;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "GenerateBoundingBoxProposals";
-  
-  private Output<TFloat32> rois;
-  private Output<TFloat32> roiProbabilities;
-  
-  private GenerateBoundingBoxProposals(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    rois = operation.output(outputIdx++);
-    roiProbabilities = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/HsvToRgb.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/HsvToRgb.java
deleted file mode 100644
index ac6876c570d..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/HsvToRgb.java
+++ /dev/null
@@ -1,81 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.image;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Convert one or more images from HSV to RGB.
- * <p>
- * Outputs a tensor of the same shape as the `images` tensor, containing the RGB
- * value of the pixels. The output is only well defined if the value in `images`
- * are in `[0,1]`.
- * <p>
- * See `rgb_to_hsv` for a description of the HSV encoding.
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator(group = "image")
-public final class HsvToRgb<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new HsvToRgb operation.
-   * 
-   * @param scope current scope
-   * @param images 1-D or higher rank. HSV data to convert. Last dimension must be size 3.
-   * @return a new instance of HsvToRgb
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> HsvToRgb<T> create(Scope scope, Operand<T> images) {
-    OperationBuilder opBuilder = scope.env().opBuilder("HSVToRGB", scope.makeOpName("HsvToRgb"));
-    opBuilder.addInput(images.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new HsvToRgb<T>(opBuilder.build());
-  }
-  
-  /**
-   * `images` converted to RGB.
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "HSVToRGB";
-  
-  private Output<T> output;
-  
-  private HsvToRgb(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ImageProjectiveTransformV2.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ImageProjectiveTransformV2.java
deleted file mode 100644
index 250e6659599..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ImageProjectiveTransformV2.java
+++ /dev/null
@@ -1,125 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.image;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Applies the given transform to each of the images.
- * <p>
- * If one row of `transforms` is `[a0, a1, a2, b0, b1, b2, c0, c1]`, then it maps
- * the <i>output</i> point `(x, y)` to a transformed <i>input</i> point
- * `(x', y') = ((a0 x + a1 y + a2) / k, (b0 x + b1 y + b2) / k)`, where
- * `k = c0 x + c1 y + 1`. If the transformed point lays outside of the input
- * image, the output pixel is set to 0.
- * 
- * @param <T> data type for {@code transformedImages()} output
- */
-public final class ImageProjectiveTransformV2<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.image.ImageProjectiveTransformV2}
-   */
-  public static class Options {
-    
-    /**
-     * @param fillMode Fill mode, "REFLECT", "WRAP", or "CONSTANT".
-     */
-    public Options fillMode(String fillMode) {
-      this.fillMode = fillMode;
-      return this;
-    }
-    
-    private String fillMode;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new ImageProjectiveTransformV2 operation.
-   * 
-   * @param scope current scope
-   * @param images 4-D with shape `[batch, height, width, channels]`.
-   * @param transforms 2-D Tensor, `[batch, 8]` or `[1, 8]` matrix, where each row corresponds to a 3 x 3
-   * projective transformation matrix, with the last entry assumed to be 1. If there
-   * is one row, the same transformation will be applied to all images.
-   * @param outputShape 1-D Tensor [new_height, new_width].
-   * @param interpolation Interpolation method, "NEAREST" or "BILINEAR".
-   * @param options carries optional attributes values
-   * @return a new instance of ImageProjectiveTransformV2
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> ImageProjectiveTransformV2<T> create(Scope scope, Operand<T> images, Operand<TFloat32> transforms, Operand<TInt32> outputShape, String interpolation, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ImageProjectiveTransformV2", scope.makeOpName("ImageProjectiveTransformV2"));
-    opBuilder.addInput(images.asOutput(scope));
-    opBuilder.addInput(transforms.asOutput(scope));
-    opBuilder.addInput(outputShape.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("interpolation", interpolation);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.fillMode != null) {
-          opBuilder.setAttr("fill_mode", opts.fillMode);
-        }
-      }
-    }
-    return new ImageProjectiveTransformV2<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param fillMode Fill mode, "REFLECT", "WRAP", or "CONSTANT".
-   */
-  public static Options fillMode(String fillMode) {
-    return new Options().fillMode(fillMode);
-  }
-  
-  /**
-   * 4-D with shape
-   * `[batch, new_height, new_width, channels]`.
-   */
-  public Output<T> transformedImages() {
-    return transformedImages;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return transformedImages;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ImageProjectiveTransformV2";
-  
-  private Output<T> transformedImages;
-  
-  private ImageProjectiveTransformV2(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    transformedImages = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/NearestNeighbors.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/NearestNeighbors.java
deleted file mode 100644
index 0dd8f6d45d7..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/NearestNeighbors.java
+++ /dev/null
@@ -1,88 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.image;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.TInt64;
-
-/**
- * Selects the k nearest centers for each point.
- * <p>
- * Rows of points are assumed to be input points. Rows of centers are assumed to be
- * the list of candidate centers. For each point, the k centers that have least L2
- * distance to it are computed.
- */
-public final class NearestNeighbors extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new NearestNeighbors operation.
-   * 
-   * @param scope current scope
-   * @param points Matrix of shape (n, d). Rows are assumed to be input points.
-   * @param centers Matrix of shape (m, d). Rows are assumed to be centers.
-   * @param k Number of nearest centers to return for each point. If k is larger than m, then
-   * only m centers are returned.
-   * @return a new instance of NearestNeighbors
-   */
-  @Endpoint(describeByClass = true)
-  public static NearestNeighbors create(Scope scope, Operand<TFloat32> points, Operand<TFloat32> centers, Operand<TInt64> k) {
-    OperationBuilder opBuilder = scope.env().opBuilder("NearestNeighbors", scope.makeOpName("NearestNeighbors"));
-    opBuilder.addInput(points.asOutput(scope));
-    opBuilder.addInput(centers.asOutput(scope));
-    opBuilder.addInput(k.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new NearestNeighbors(opBuilder.build());
-  }
-  
-  /**
-   * Matrix of shape (n, min(m, k)). Each row contains the indices of the centers
-   * closest to the corresponding point, ordered by increasing distance.
-   */
-  public Output<TInt64> nearestCenterIndices() {
-    return nearestCenterIndices;
-  }
-  
-  /**
-   * Matrix of shape (n, min(m, k)). Each row contains the squared L2 distance to the
-   * corresponding center in nearest_center_indices.
-   */
-  public Output<TFloat32> nearestCenterDistances() {
-    return nearestCenterDistances;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "NearestNeighbors";
-  
-  private Output<TInt64> nearestCenterIndices;
-  private Output<TFloat32> nearestCenterDistances;
-  
-  private NearestNeighbors(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    nearestCenterIndices = operation.output(outputIdx++);
-    nearestCenterDistances = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/NonMaxSuppression.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/NonMaxSuppression.java
deleted file mode 100644
index 12f9714a150..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/NonMaxSuppression.java
+++ /dev/null
@@ -1,169 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.image;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Greedily selects a subset of bounding boxes in descending order of score,
- * <p>
- * pruning away boxes that have high intersection-over-union (IOU) overlap
- * with previously selected boxes.  Bounding boxes with score less than
- * `score_threshold` are removed.  Bounding boxes are supplied as
- * [y1, x1, y2, x2], where (y1, x1) and (y2, x2) are the coordinates of any
- * diagonal pair of box corners and the coordinates can be provided as normalized
- * (i.e., lying in the interval [0, 1]) or absolute.  Note that this algorithm
- * is agnostic to where the origin is in the coordinate system and more
- * generally is invariant to orthogonal transformations and translations
- * of the coordinate system; thus translating or reflections of the coordinate
- * system result in the same boxes being selected by the algorithm.
- * The output of this operation is a set of integers indexing into the input
- * collection of bounding boxes representing the selected boxes.  The bounding
- * box coordinates corresponding to the selected indices can then be obtained
- * using the `tf.gather operation`.  For example:
- *   selected_indices = tf.image.non_max_suppression_v2(
- *       boxes, scores, max_output_size, iou_threshold, score_threshold)
- *   selected_boxes = tf.gather(boxes, selected_indices)
- * This op also supports a Soft-NMS (with Gaussian weighting) mode (c.f.
- * Bodla et al, https://arxiv.org/abs/1704.04503) where boxes reduce the score
- * of other overlapping boxes instead of directly causing them to be pruned.
- * To enable this Soft-NMS mode, set the `soft_nms_sigma` parameter to be
- * larger than 0.
- * 
- * @param <T> data type for {@code selectedScores()} output
- */
-@Operator(group = "image")
-public final class NonMaxSuppression<T extends TNumber> extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.image.NonMaxSuppression}
-   */
-  public static class Options {
-    
-    /**
-     * @param padToMaxOutputSize If true, the output `selected_indices` is padded to be of length
-     * `max_output_size`. Defaults to false.
-     */
-    public Options padToMaxOutputSize(Boolean padToMaxOutputSize) {
-      this.padToMaxOutputSize = padToMaxOutputSize;
-      return this;
-    }
-    
-    private Boolean padToMaxOutputSize;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new NonMaxSuppression operation.
-   * 
-   * @param scope current scope
-   * @param boxes A 2-D float tensor of shape `[num_boxes, 4]`.
-   * @param scores A 1-D float tensor of shape `[num_boxes]` representing a single
-   * score corresponding to each box (each row of boxes).
-   * @param maxOutputSize A scalar integer tensor representing the maximum number of
-   * boxes to be selected by non max suppression.
-   * @param iouThreshold A 0-D float tensor representing the threshold for deciding whether
-   * boxes overlap too much with respect to IOU.
-   * @param scoreThreshold A 0-D float tensor representing the threshold for deciding when to remove
-   * boxes based on score.
-   * @param softNmsSigma A 0-D float tensor representing the sigma parameter for Soft NMS; see Bodla et
-   * al (c.f. https://arxiv.org/abs/1704.04503).  When `soft_nms_sigma=0.0` (which
-   * is default), we fall back to standard (hard) NMS.
-   * @param options carries optional attributes values
-   * @return a new instance of NonMaxSuppression
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> NonMaxSuppression<T> create(Scope scope, Operand<T> boxes, Operand<T> scores, Operand<TInt32> maxOutputSize, Operand<T> iouThreshold, Operand<T> scoreThreshold, Operand<T> softNmsSigma, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("NonMaxSuppressionV5", scope.makeOpName("NonMaxSuppression"));
-    opBuilder.addInput(boxes.asOutput(scope));
-    opBuilder.addInput(scores.asOutput(scope));
-    opBuilder.addInput(maxOutputSize.asOutput(scope));
-    opBuilder.addInput(iouThreshold.asOutput(scope));
-    opBuilder.addInput(scoreThreshold.asOutput(scope));
-    opBuilder.addInput(softNmsSigma.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.padToMaxOutputSize != null) {
-          opBuilder.setAttr("pad_to_max_output_size", opts.padToMaxOutputSize);
-        }
-      }
-    }
-    return new NonMaxSuppression<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param padToMaxOutputSize If true, the output `selected_indices` is padded to be of length
-   * `max_output_size`. Defaults to false.
-   */
-  public static Options padToMaxOutputSize(Boolean padToMaxOutputSize) {
-    return new Options().padToMaxOutputSize(padToMaxOutputSize);
-  }
-  
-  /**
-   * A 1-D integer tensor of shape `[M]` representing the selected
-   * indices from the boxes tensor, where `M <= max_output_size`.
-   */
-  public Output<TInt32> selectedIndices() {
-    return selectedIndices;
-  }
-  
-  /**
-   * A 1-D float tensor of shape `[M]` representing the corresponding
-   * scores for each selected box, where `M <= max_output_size`.  Scores only differ
-   * from corresponding input scores when using Soft NMS (i.e. when
-   * `soft_nms_sigma>0`)
-   */
-  public Output<T> selectedScores() {
-    return selectedScores;
-  }
-  
-  /**
-   * A 0-D integer tensor representing the number of valid elements in
-   * `selected_indices`, with the valid elements appearing first.
-   */
-  public Output<TInt32> validOutputs() {
-    return validOutputs;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "NonMaxSuppressionV5";
-  
-  private Output<TInt32> selectedIndices;
-  private Output<T> selectedScores;
-  private Output<TInt32> validOutputs;
-  
-  private NonMaxSuppression(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    selectedIndices = operation.output(outputIdx++);
-    selectedScores = operation.output(outputIdx++);
-    validOutputs = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/NonMaxSuppressionWithOverlaps.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/NonMaxSuppressionWithOverlaps.java
deleted file mode 100644
index 952746a3108..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/NonMaxSuppressionWithOverlaps.java
+++ /dev/null
@@ -1,103 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.image;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.TInt32;
-
-/**
- * Greedily selects a subset of bounding boxes in descending order of score,
- * <p>
- * pruning away boxes that have high overlaps
- * with previously selected boxes.  Bounding boxes with score less than
- * `score_threshold` are removed. N-by-n overlap values are supplied as square matrix,
- * which allows for defining a custom overlap criterium (eg. intersection over union,
- * intersection over area, etc.).
- * <p>
- * The output of this operation is a set of integers indexing into the input
- * collection of bounding boxes representing the selected boxes.  The bounding
- * box coordinates corresponding to the selected indices can then be obtained
- * using the `tf.gather operation`.  For example:
- * <p>
- *   selected_indices = tf.image.non_max_suppression_with_overlaps(
- *       overlaps, scores, max_output_size, overlap_threshold, score_threshold)
- *   selected_boxes = tf.gather(boxes, selected_indices)
- */
-@Operator(group = "image")
-public final class NonMaxSuppressionWithOverlaps extends RawOp implements Operand<TInt32> {
-  
-  /**
-   * Factory method to create a class wrapping a new NonMaxSuppressionWithOverlaps operation.
-   * 
-   * @param scope current scope
-   * @param overlaps A 2-D float tensor of shape `[num_boxes, num_boxes]` representing
-   * the n-by-n box overlap values.
-   * @param scores A 1-D float tensor of shape `[num_boxes]` representing a single
-   * score corresponding to each box (each row of boxes).
-   * @param maxOutputSize A scalar integer tensor representing the maximum number of
-   * boxes to be selected by non max suppression.
-   * @param overlapThreshold A 0-D float tensor representing the threshold for deciding whether
-   * boxes overlap too.
-   * @param scoreThreshold A 0-D float tensor representing the threshold for deciding when to remove
-   * boxes based on score.
-   * @return a new instance of NonMaxSuppressionWithOverlaps
-   */
-  @Endpoint(describeByClass = true)
-  public static NonMaxSuppressionWithOverlaps create(Scope scope, Operand<TFloat32> overlaps, Operand<TFloat32> scores, Operand<TInt32> maxOutputSize, Operand<TFloat32> overlapThreshold, Operand<TFloat32> scoreThreshold) {
-    OperationBuilder opBuilder = scope.env().opBuilder("NonMaxSuppressionWithOverlaps", scope.makeOpName("NonMaxSuppressionWithOverlaps"));
-    opBuilder.addInput(overlaps.asOutput(scope));
-    opBuilder.addInput(scores.asOutput(scope));
-    opBuilder.addInput(maxOutputSize.asOutput(scope));
-    opBuilder.addInput(overlapThreshold.asOutput(scope));
-    opBuilder.addInput(scoreThreshold.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new NonMaxSuppressionWithOverlaps(opBuilder.build());
-  }
-  
-  /**
-   * A 1-D integer tensor of shape `[M]` representing the selected
-   * indices from the boxes tensor, where `M <= max_output_size`.
-   */
-  public Output<TInt32> selectedIndices() {
-    return selectedIndices;
-  }
-  
-  @Override
-  public Output<TInt32> asOutput(Scope scope) {
-    return selectedIndices;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "NonMaxSuppressionWithOverlaps";
-  
-  private Output<TInt32> selectedIndices;
-  
-  private NonMaxSuppressionWithOverlaps(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    selectedIndices = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/QuantizedResizeBilinear.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/QuantizedResizeBilinear.java
deleted file mode 100644
index 004caafdb5e..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/QuantizedResizeBilinear.java
+++ /dev/null
@@ -1,153 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.image;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.family.TType;
-
-/**
- * Resize quantized `images` to `size` using quantized bilinear interpolation.
- * <p>
- * Input images and output images must be quantized types.
- * 
- * @param <T> data type for {@code resizedImages()} output
- */
-@Operator(group = "image")
-public final class QuantizedResizeBilinear<T extends TType> extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.image.QuantizedResizeBilinear}
-   */
-  public static class Options {
-    
-    /**
-     * @param alignCorners If true, the centers of the 4 corner pixels of the input and output tensors are
-     * aligned, preserving the values at the corner pixels. Defaults to false.
-     */
-    public Options alignCorners(Boolean alignCorners) {
-      this.alignCorners = alignCorners;
-      return this;
-    }
-    
-    /**
-     * @param halfPixelCenters 
-     */
-    public Options halfPixelCenters(Boolean halfPixelCenters) {
-      this.halfPixelCenters = halfPixelCenters;
-      return this;
-    }
-    
-    private Boolean alignCorners;
-    private Boolean halfPixelCenters;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new QuantizedResizeBilinear operation.
-   * 
-   * @param scope current scope
-   * @param images 4-D with shape `[batch, height, width, channels]`.
-   * @param size = A 1-D int32 Tensor of 2 elements: `new_height, new_width`.  The
-   * new size for the images.
-   * @param min 
-   * @param max 
-   * @param options carries optional attributes values
-   * @return a new instance of QuantizedResizeBilinear
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> QuantizedResizeBilinear<T> create(Scope scope, Operand<T> images, Operand<TInt32> size, Operand<TFloat32> min, Operand<TFloat32> max, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("QuantizedResizeBilinear", scope.makeOpName("QuantizedResizeBilinear"));
-    opBuilder.addInput(images.asOutput(scope));
-    opBuilder.addInput(size.asOutput(scope));
-    opBuilder.addInput(min.asOutput(scope));
-    opBuilder.addInput(max.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.alignCorners != null) {
-          opBuilder.setAttr("align_corners", opts.alignCorners);
-        }
-        if (opts.halfPixelCenters != null) {
-          opBuilder.setAttr("half_pixel_centers", opts.halfPixelCenters);
-        }
-      }
-    }
-    return new QuantizedResizeBilinear<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param alignCorners If true, the centers of the 4 corner pixels of the input and output tensors are
-   * aligned, preserving the values at the corner pixels. Defaults to false.
-   */
-  public static Options alignCorners(Boolean alignCorners) {
-    return new Options().alignCorners(alignCorners);
-  }
-  
-  /**
-   * @param halfPixelCenters 
-   */
-  public static Options halfPixelCenters(Boolean halfPixelCenters) {
-    return new Options().halfPixelCenters(halfPixelCenters);
-  }
-  
-  /**
-   * 4-D with shape
-   * `[batch, new_height, new_width, channels]`.
-   */
-  public Output<T> resizedImages() {
-    return resizedImages;
-  }
-  
-  /**
-   */
-  public Output<TFloat32> outMin() {
-    return outMin;
-  }
-  
-  /**
-   */
-  public Output<TFloat32> outMax() {
-    return outMax;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "QuantizedResizeBilinear";
-  
-  private Output<T> resizedImages;
-  private Output<TFloat32> outMin;
-  private Output<TFloat32> outMax;
-  
-  private QuantizedResizeBilinear(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    resizedImages = operation.output(outputIdx++);
-    outMin = operation.output(outputIdx++);
-    outMax = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/RandomCrop.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/RandomCrop.java
deleted file mode 100644
index 1a5a2a4f47c..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/RandomCrop.java
+++ /dev/null
@@ -1,142 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.image;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Randomly crop `image`.
- * <p>
- * `size` is a 1-D int64 tensor with 2 elements representing the crop height and
- * width.  The values must be non negative.
- * <p>
- * This Op picks a random location in `image` and crops a `height` by `width`
- * rectangle from that location.  The random location is picked so the cropped
- * area will fit inside the original image.
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator(group = "image")
-public final class RandomCrop<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.image.RandomCrop}
-   */
-  public static class Options {
-    
-    /**
-     * @param seed If either seed or seed2 are set to be non-zero, the random number
-     * generator is seeded by the given seed.  Otherwise, it is seeded by a
-     * random seed.
-     */
-    public Options seed(Long seed) {
-      this.seed = seed;
-      return this;
-    }
-    
-    /**
-     * @param seed2 An second seed to avoid seed collision.
-     */
-    public Options seed2(Long seed2) {
-      this.seed2 = seed2;
-      return this;
-    }
-    
-    private Long seed;
-    private Long seed2;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new RandomCrop operation.
-   * 
-   * @param scope current scope
-   * @param image 3-D of shape `[height, width, channels]`.
-   * @param size 1-D of length 2 containing: `crop_height`, `crop_width`..
-   * @param options carries optional attributes values
-   * @return a new instance of RandomCrop
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> RandomCrop<T> create(Scope scope, Operand<T> image, Operand<TInt64> size, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("RandomCrop", scope.makeOpName("RandomCrop"));
-    opBuilder.addInput(image.asOutput(scope));
-    opBuilder.addInput(size.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.seed != null) {
-          opBuilder.setAttr("seed", opts.seed);
-        }
-        if (opts.seed2 != null) {
-          opBuilder.setAttr("seed2", opts.seed2);
-        }
-      }
-    }
-    return new RandomCrop<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param seed If either seed or seed2 are set to be non-zero, the random number
-   * generator is seeded by the given seed.  Otherwise, it is seeded by a
-   * random seed.
-   */
-  public static Options seed(Long seed) {
-    return new Options().seed(seed);
-  }
-  
-  /**
-   * @param seed2 An second seed to avoid seed collision.
-   */
-  public static Options seed2(Long seed2) {
-    return new Options().seed2(seed2);
-  }
-  
-  /**
-   * 3-D of shape `[crop_height, crop_width, channels].`
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "RandomCrop";
-  
-  private Output<T> output;
-  
-  private RandomCrop(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ResizeArea.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ResizeArea.java
deleted file mode 100644
index e4252030c3a..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ResizeArea.java
+++ /dev/null
@@ -1,127 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.image;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Resize `images` to `size` using area interpolation.
- * <p>
- * Input images can be of different types but output images are always float.
- * <p>
- * The range of pixel values for the output image might be slightly different
- * from the range for the input image because of limited numerical precision.
- * To guarantee an output range, for example `[0.0, 1.0]`, apply
- * `tf.clip_by_value` to the output.
- * <p>
- * Each output pixel is computed by first transforming the pixel's footprint into
- * the input tensor and then averaging the pixels that intersect the footprint. An
- * input pixel's contribution to the average is weighted by the fraction of its
- * area that intersects the footprint.  This is the same as OpenCV's INTER_AREA.
- */
-@Operator(group = "image")
-public final class ResizeArea extends RawOp implements Operand<TFloat32> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.image.ResizeArea}
-   */
-  public static class Options {
-    
-    /**
-     * @param alignCorners If true, the centers of the 4 corner pixels of the input and output tensors are
-     * aligned, preserving the values at the corner pixels. Defaults to false.
-     */
-    public Options alignCorners(Boolean alignCorners) {
-      this.alignCorners = alignCorners;
-      return this;
-    }
-    
-    private Boolean alignCorners;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new ResizeArea operation.
-   * 
-   * @param scope current scope
-   * @param images 4-D with shape `[batch, height, width, channels]`.
-   * @param size = A 1-D int32 Tensor of 2 elements: `new_height, new_width`.  The
-   * new size for the images.
-   * @param options carries optional attributes values
-   * @return a new instance of ResizeArea
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> ResizeArea create(Scope scope, Operand<T> images, Operand<TInt32> size, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ResizeArea", scope.makeOpName("ResizeArea"));
-    opBuilder.addInput(images.asOutput(scope));
-    opBuilder.addInput(size.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.alignCorners != null) {
-          opBuilder.setAttr("align_corners", opts.alignCorners);
-        }
-      }
-    }
-    return new ResizeArea(opBuilder.build());
-  }
-  
-  /**
-   * @param alignCorners If true, the centers of the 4 corner pixels of the input and output tensors are
-   * aligned, preserving the values at the corner pixels. Defaults to false.
-   */
-  public static Options alignCorners(Boolean alignCorners) {
-    return new Options().alignCorners(alignCorners);
-  }
-  
-  /**
-   * 4-D with shape
-   * `[batch, new_height, new_width, channels]`.
-   */
-  public Output<TFloat32> resizedImages() {
-    return resizedImages;
-  }
-  
-  @Override
-  public Output<TFloat32> asOutput(Scope scope) {
-    return resizedImages;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ResizeArea";
-  
-  private Output<TFloat32> resizedImages;
-  
-  private ResizeArea(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    resizedImages = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ResizeBicubic.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ResizeBicubic.java
deleted file mode 100644
index caf2a49dba5..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ResizeBicubic.java
+++ /dev/null
@@ -1,136 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.image;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Resize `images` to `size` using bicubic interpolation.
- * <p>
- * Input images can be of different types but output images are always float.
- */
-@Operator(group = "image")
-public final class ResizeBicubic extends RawOp implements Operand<TFloat32> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.image.ResizeBicubic}
-   */
-  public static class Options {
-    
-    /**
-     * @param alignCorners If true, the centers of the 4 corner pixels of the input and output tensors are
-     * aligned, preserving the values at the corner pixels. Defaults to false.
-     */
-    public Options alignCorners(Boolean alignCorners) {
-      this.alignCorners = alignCorners;
-      return this;
-    }
-    
-    /**
-     * @param halfPixelCenters 
-     */
-    public Options halfPixelCenters(Boolean halfPixelCenters) {
-      this.halfPixelCenters = halfPixelCenters;
-      return this;
-    }
-    
-    private Boolean alignCorners;
-    private Boolean halfPixelCenters;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new ResizeBicubic operation.
-   * 
-   * @param scope current scope
-   * @param images 4-D with shape `[batch, height, width, channels]`.
-   * @param size = A 1-D int32 Tensor of 2 elements: `new_height, new_width`.  The
-   * new size for the images.
-   * @param options carries optional attributes values
-   * @return a new instance of ResizeBicubic
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> ResizeBicubic create(Scope scope, Operand<T> images, Operand<TInt32> size, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ResizeBicubic", scope.makeOpName("ResizeBicubic"));
-    opBuilder.addInput(images.asOutput(scope));
-    opBuilder.addInput(size.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.alignCorners != null) {
-          opBuilder.setAttr("align_corners", opts.alignCorners);
-        }
-        if (opts.halfPixelCenters != null) {
-          opBuilder.setAttr("half_pixel_centers", opts.halfPixelCenters);
-        }
-      }
-    }
-    return new ResizeBicubic(opBuilder.build());
-  }
-  
-  /**
-   * @param alignCorners If true, the centers of the 4 corner pixels of the input and output tensors are
-   * aligned, preserving the values at the corner pixels. Defaults to false.
-   */
-  public static Options alignCorners(Boolean alignCorners) {
-    return new Options().alignCorners(alignCorners);
-  }
-  
-  /**
-   * @param halfPixelCenters 
-   */
-  public static Options halfPixelCenters(Boolean halfPixelCenters) {
-    return new Options().halfPixelCenters(halfPixelCenters);
-  }
-  
-  /**
-   * 4-D with shape
-   * `[batch, new_height, new_width, channels]`.
-   */
-  public Output<TFloat32> resizedImages() {
-    return resizedImages;
-  }
-  
-  @Override
-  public Output<TFloat32> asOutput(Scope scope) {
-    return resizedImages;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ResizeBicubic";
-  
-  private Output<TFloat32> resizedImages;
-  
-  private ResizeBicubic(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    resizedImages = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ResizeBicubicGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ResizeBicubicGrad.java
deleted file mode 100644
index 623941bbc98..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ResizeBicubicGrad.java
+++ /dev/null
@@ -1,135 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.image;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Computes the gradient of bicubic interpolation.
- * 
- * @param <T> data type for {@code output()} output
- */
-public final class ResizeBicubicGrad<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.image.ResizeBicubicGrad}
-   */
-  public static class Options {
-    
-    /**
-     * @param alignCorners If true, the centers of the 4 corner pixels of the input and grad tensors are
-     * aligned. Defaults to false.
-     */
-    public Options alignCorners(Boolean alignCorners) {
-      this.alignCorners = alignCorners;
-      return this;
-    }
-    
-    /**
-     * @param halfPixelCenters 
-     */
-    public Options halfPixelCenters(Boolean halfPixelCenters) {
-      this.halfPixelCenters = halfPixelCenters;
-      return this;
-    }
-    
-    private Boolean alignCorners;
-    private Boolean halfPixelCenters;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new ResizeBicubicGrad operation.
-   * 
-   * @param scope current scope
-   * @param grads 4-D with shape `[batch, height, width, channels]`.
-   * @param originalImage 4-D with shape `[batch, orig_height, orig_width, channels]`,
-   * The image tensor that was resized.
-   * @param options carries optional attributes values
-   * @return a new instance of ResizeBicubicGrad
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> ResizeBicubicGrad<T> create(Scope scope, Operand<TFloat32> grads, Operand<T> originalImage, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ResizeBicubicGrad", scope.makeOpName("ResizeBicubicGrad"));
-    opBuilder.addInput(grads.asOutput(scope));
-    opBuilder.addInput(originalImage.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.alignCorners != null) {
-          opBuilder.setAttr("align_corners", opts.alignCorners);
-        }
-        if (opts.halfPixelCenters != null) {
-          opBuilder.setAttr("half_pixel_centers", opts.halfPixelCenters);
-        }
-      }
-    }
-    return new ResizeBicubicGrad<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param alignCorners If true, the centers of the 4 corner pixels of the input and grad tensors are
-   * aligned. Defaults to false.
-   */
-  public static Options alignCorners(Boolean alignCorners) {
-    return new Options().alignCorners(alignCorners);
-  }
-  
-  /**
-   * @param halfPixelCenters 
-   */
-  public static Options halfPixelCenters(Boolean halfPixelCenters) {
-    return new Options().halfPixelCenters(halfPixelCenters);
-  }
-  
-  /**
-   * 4-D with shape `[batch, orig_height, orig_width, channels]`.
-   * Gradients with respect to the input image. Input image must have been
-   * float or double.
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ResizeBicubicGrad";
-  
-  private Output<T> output;
-  
-  private ResizeBicubicGrad(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ResizeBilinear.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ResizeBilinear.java
deleted file mode 100644
index c65c9c8d176..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ResizeBilinear.java
+++ /dev/null
@@ -1,136 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.image;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Resize `images` to `size` using bilinear interpolation.
- * <p>
- * Input images can be of different types but output images are always float.
- */
-@Operator(group = "image")
-public final class ResizeBilinear extends RawOp implements Operand<TFloat32> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.image.ResizeBilinear}
-   */
-  public static class Options {
-    
-    /**
-     * @param alignCorners If true, the centers of the 4 corner pixels of the input and output tensors are
-     * aligned, preserving the values at the corner pixels. Defaults to false.
-     */
-    public Options alignCorners(Boolean alignCorners) {
-      this.alignCorners = alignCorners;
-      return this;
-    }
-    
-    /**
-     * @param halfPixelCenters 
-     */
-    public Options halfPixelCenters(Boolean halfPixelCenters) {
-      this.halfPixelCenters = halfPixelCenters;
-      return this;
-    }
-    
-    private Boolean alignCorners;
-    private Boolean halfPixelCenters;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new ResizeBilinear operation.
-   * 
-   * @param scope current scope
-   * @param images 4-D with shape `[batch, height, width, channels]`.
-   * @param size = A 1-D int32 Tensor of 2 elements: `new_height, new_width`.  The
-   * new size for the images.
-   * @param options carries optional attributes values
-   * @return a new instance of ResizeBilinear
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> ResizeBilinear create(Scope scope, Operand<T> images, Operand<TInt32> size, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ResizeBilinear", scope.makeOpName("ResizeBilinear"));
-    opBuilder.addInput(images.asOutput(scope));
-    opBuilder.addInput(size.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.alignCorners != null) {
-          opBuilder.setAttr("align_corners", opts.alignCorners);
-        }
-        if (opts.halfPixelCenters != null) {
-          opBuilder.setAttr("half_pixel_centers", opts.halfPixelCenters);
-        }
-      }
-    }
-    return new ResizeBilinear(opBuilder.build());
-  }
-  
-  /**
-   * @param alignCorners If true, the centers of the 4 corner pixels of the input and output tensors are
-   * aligned, preserving the values at the corner pixels. Defaults to false.
-   */
-  public static Options alignCorners(Boolean alignCorners) {
-    return new Options().alignCorners(alignCorners);
-  }
-  
-  /**
-   * @param halfPixelCenters 
-   */
-  public static Options halfPixelCenters(Boolean halfPixelCenters) {
-    return new Options().halfPixelCenters(halfPixelCenters);
-  }
-  
-  /**
-   * 4-D with shape
-   * `[batch, new_height, new_width, channels]`.
-   */
-  public Output<TFloat32> resizedImages() {
-    return resizedImages;
-  }
-  
-  @Override
-  public Output<TFloat32> asOutput(Scope scope) {
-    return resizedImages;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ResizeBilinear";
-  
-  private Output<TFloat32> resizedImages;
-  
-  private ResizeBilinear(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    resizedImages = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ResizeBilinearGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ResizeBilinearGrad.java
deleted file mode 100644
index bacb51b9775..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ResizeBilinearGrad.java
+++ /dev/null
@@ -1,135 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.image;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Computes the gradient of bilinear interpolation.
- * 
- * @param <T> data type for {@code output()} output
- */
-public final class ResizeBilinearGrad<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.image.ResizeBilinearGrad}
-   */
-  public static class Options {
-    
-    /**
-     * @param alignCorners If true, the centers of the 4 corner pixels of the input and grad tensors are
-     * aligned. Defaults to false.
-     */
-    public Options alignCorners(Boolean alignCorners) {
-      this.alignCorners = alignCorners;
-      return this;
-    }
-    
-    /**
-     * @param halfPixelCenters 
-     */
-    public Options halfPixelCenters(Boolean halfPixelCenters) {
-      this.halfPixelCenters = halfPixelCenters;
-      return this;
-    }
-    
-    private Boolean alignCorners;
-    private Boolean halfPixelCenters;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new ResizeBilinearGrad operation.
-   * 
-   * @param scope current scope
-   * @param grads 4-D with shape `[batch, height, width, channels]`.
-   * @param originalImage 4-D with shape `[batch, orig_height, orig_width, channels]`,
-   * The image tensor that was resized.
-   * @param options carries optional attributes values
-   * @return a new instance of ResizeBilinearGrad
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> ResizeBilinearGrad<T> create(Scope scope, Operand<TFloat32> grads, Operand<T> originalImage, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ResizeBilinearGrad", scope.makeOpName("ResizeBilinearGrad"));
-    opBuilder.addInput(grads.asOutput(scope));
-    opBuilder.addInput(originalImage.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.alignCorners != null) {
-          opBuilder.setAttr("align_corners", opts.alignCorners);
-        }
-        if (opts.halfPixelCenters != null) {
-          opBuilder.setAttr("half_pixel_centers", opts.halfPixelCenters);
-        }
-      }
-    }
-    return new ResizeBilinearGrad<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param alignCorners If true, the centers of the 4 corner pixels of the input and grad tensors are
-   * aligned. Defaults to false.
-   */
-  public static Options alignCorners(Boolean alignCorners) {
-    return new Options().alignCorners(alignCorners);
-  }
-  
-  /**
-   * @param halfPixelCenters 
-   */
-  public static Options halfPixelCenters(Boolean halfPixelCenters) {
-    return new Options().halfPixelCenters(halfPixelCenters);
-  }
-  
-  /**
-   * 4-D with shape `[batch, orig_height, orig_width, channels]`.
-   * Gradients with respect to the input image. Input image must have been
-   * float or double.
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ResizeBilinearGrad";
-  
-  private Output<T> output;
-  
-  private ResizeBilinearGrad(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ResizeNearestNeighbor.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ResizeNearestNeighbor.java
deleted file mode 100644
index a7a8ceca11b..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ResizeNearestNeighbor.java
+++ /dev/null
@@ -1,135 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.image;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Resize `images` to `size` using nearest neighbor interpolation.
- * 
- * @param <T> data type for {@code resizedImages()} output
- */
-@Operator(group = "image")
-public final class ResizeNearestNeighbor<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.image.ResizeNearestNeighbor}
-   */
-  public static class Options {
-    
-    /**
-     * @param alignCorners If true, the centers of the 4 corner pixels of the input and output tensors are
-     * aligned, preserving the values at the corner pixels. Defaults to false.
-     */
-    public Options alignCorners(Boolean alignCorners) {
-      this.alignCorners = alignCorners;
-      return this;
-    }
-    
-    /**
-     * @param halfPixelCenters 
-     */
-    public Options halfPixelCenters(Boolean halfPixelCenters) {
-      this.halfPixelCenters = halfPixelCenters;
-      return this;
-    }
-    
-    private Boolean alignCorners;
-    private Boolean halfPixelCenters;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new ResizeNearestNeighbor operation.
-   * 
-   * @param scope current scope
-   * @param images 4-D with shape `[batch, height, width, channels]`.
-   * @param size = A 1-D int32 Tensor of 2 elements: `new_height, new_width`.  The
-   * new size for the images.
-   * @param options carries optional attributes values
-   * @return a new instance of ResizeNearestNeighbor
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> ResizeNearestNeighbor<T> create(Scope scope, Operand<T> images, Operand<TInt32> size, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ResizeNearestNeighbor", scope.makeOpName("ResizeNearestNeighbor"));
-    opBuilder.addInput(images.asOutput(scope));
-    opBuilder.addInput(size.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.alignCorners != null) {
-          opBuilder.setAttr("align_corners", opts.alignCorners);
-        }
-        if (opts.halfPixelCenters != null) {
-          opBuilder.setAttr("half_pixel_centers", opts.halfPixelCenters);
-        }
-      }
-    }
-    return new ResizeNearestNeighbor<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param alignCorners If true, the centers of the 4 corner pixels of the input and output tensors are
-   * aligned, preserving the values at the corner pixels. Defaults to false.
-   */
-  public static Options alignCorners(Boolean alignCorners) {
-    return new Options().alignCorners(alignCorners);
-  }
-  
-  /**
-   * @param halfPixelCenters 
-   */
-  public static Options halfPixelCenters(Boolean halfPixelCenters) {
-    return new Options().halfPixelCenters(halfPixelCenters);
-  }
-  
-  /**
-   * 4-D with shape
-   * `[batch, new_height, new_width, channels]`.
-   */
-  public Output<T> resizedImages() {
-    return resizedImages;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return resizedImages;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ResizeNearestNeighbor";
-  
-  private Output<T> resizedImages;
-  
-  private ResizeNearestNeighbor(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    resizedImages = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ResizeNearestNeighborGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ResizeNearestNeighborGrad.java
deleted file mode 100644
index fdee4f35ace..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ResizeNearestNeighborGrad.java
+++ /dev/null
@@ -1,134 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.image;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Computes the gradient of nearest neighbor interpolation.
- * 
- * @param <T> data type for {@code output()} output
- */
-public final class ResizeNearestNeighborGrad<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.image.ResizeNearestNeighborGrad}
-   */
-  public static class Options {
-    
-    /**
-     * @param alignCorners If true, the centers of the 4 corner pixels of the input and grad tensors are
-     * aligned. Defaults to false.
-     */
-    public Options alignCorners(Boolean alignCorners) {
-      this.alignCorners = alignCorners;
-      return this;
-    }
-    
-    /**
-     * @param halfPixelCenters 
-     */
-    public Options halfPixelCenters(Boolean halfPixelCenters) {
-      this.halfPixelCenters = halfPixelCenters;
-      return this;
-    }
-    
-    private Boolean alignCorners;
-    private Boolean halfPixelCenters;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new ResizeNearestNeighborGrad operation.
-   * 
-   * @param scope current scope
-   * @param grads 4-D with shape `[batch, height, width, channels]`.
-   * @param size = A 1-D int32 Tensor of 2 elements: `orig_height, orig_width`. The
-   * original input size.
-   * @param options carries optional attributes values
-   * @return a new instance of ResizeNearestNeighborGrad
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> ResizeNearestNeighborGrad<T> create(Scope scope, Operand<T> grads, Operand<TInt32> size, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ResizeNearestNeighborGrad", scope.makeOpName("ResizeNearestNeighborGrad"));
-    opBuilder.addInput(grads.asOutput(scope));
-    opBuilder.addInput(size.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.alignCorners != null) {
-          opBuilder.setAttr("align_corners", opts.alignCorners);
-        }
-        if (opts.halfPixelCenters != null) {
-          opBuilder.setAttr("half_pixel_centers", opts.halfPixelCenters);
-        }
-      }
-    }
-    return new ResizeNearestNeighborGrad<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param alignCorners If true, the centers of the 4 corner pixels of the input and grad tensors are
-   * aligned. Defaults to false.
-   */
-  public static Options alignCorners(Boolean alignCorners) {
-    return new Options().alignCorners(alignCorners);
-  }
-  
-  /**
-   * @param halfPixelCenters 
-   */
-  public static Options halfPixelCenters(Boolean halfPixelCenters) {
-    return new Options().halfPixelCenters(halfPixelCenters);
-  }
-  
-  /**
-   * 4-D with shape `[batch, orig_height, orig_width, channels]`. Gradients
-   * with respect to the input image.
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ResizeNearestNeighborGrad";
-  
-  private Output<T> output;
-  
-  private ResizeNearestNeighborGrad(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/RgbToHsv.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/RgbToHsv.java
deleted file mode 100644
index 9e86f495a7d..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/RgbToHsv.java
+++ /dev/null
@@ -1,95 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.image;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Converts one or more images from RGB to HSV.
- * <p>
- * Outputs a tensor of the same shape as the `images` tensor, containing the HSV
- * value of the pixels. The output is only well defined if the value in `images`
- * are in `[0,1]`.
- * <p>
- * `output[..., 0]` contains hue, `output[..., 1]` contains saturation, and
- * `output[..., 2]` contains value. All HSV values are in `[0,1]`. A hue of 0
- * corresponds to pure red, hue 1/3 is pure green, and 2/3 is pure blue.
- * <p>
- * Usage Example:
- * <p>
- * >>> blue_image = tf.stack([
- * ...    tf.zeros([5,5]),
- * ...    tf.zeros([5,5]),
- * ...    tf.ones([5,5])],
- * ...    axis=-1)
- * >>> blue_hsv_image = tf.image.rgb_to_hsv(blue_image)
- * >>> blue_hsv_image[0,0].numpy()
- * array([0.6666667, 1. , 1. ], dtype=float32)
- * 
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator(group = "image")
-public final class RgbToHsv<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new RgbToHsv operation.
-   * 
-   * @param scope current scope
-   * @param images 1-D or higher rank. RGB data to convert. Last dimension must be size 3.
-   * @return a new instance of RgbToHsv
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> RgbToHsv<T> create(Scope scope, Operand<T> images) {
-    OperationBuilder opBuilder = scope.env().opBuilder("RGBToHSV", scope.makeOpName("RgbToHsv"));
-    opBuilder.addInput(images.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new RgbToHsv<T>(opBuilder.build());
-  }
-  
-  /**
-   * `images` converted to HSV.
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "RGBToHSV";
-  
-  private Output<T> output;
-  
-  private RgbToHsv(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/SampleDistortedBoundingBox.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/SampleDistortedBoundingBox.java
deleted file mode 100644
index fe4bc34d9bd..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/SampleDistortedBoundingBox.java
+++ /dev/null
@@ -1,291 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.image;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Generate a single randomly distorted bounding box for an image.
- * <p>
- * Bounding box annotations are often supplied in addition to ground-truth labels
- * in image recognition or object localization tasks. A common technique for
- * training such a system is to randomly distort an image while preserving
- * its content, i.e. <i>data augmentation</i>. This Op outputs a randomly distorted
- * localization of an object, i.e. bounding box, given an `image_size`,
- * `bounding_boxes` and a series of constraints.
- * <p>
- * The output of this Op is a single bounding box that may be used to crop the
- * original image. The output is returned as 3 tensors: `begin`, `size` and
- * `bboxes`. The first 2 tensors can be fed directly into `tf.slice` to crop the
- * image. The latter may be supplied to `tf.image.draw_bounding_boxes` to visualize
- * what the bounding box looks like.
- * <p>
- * Bounding boxes are supplied and returned as `[y_min, x_min, y_max, x_max]`. The
- * bounding box coordinates are floats in `[0.0, 1.0]` relative to the width and
- * height of the underlying image.
- * <p>
- * For example,
- * <pre>{@code
- *     # Generate a single distorted bounding box.
- *     begin, size, bbox_for_draw = tf.image.sample_distorted_bounding_box(
- *         tf.shape(image),
- *         bounding_boxes=bounding_boxes)
- * 
- *     # Draw the bounding box in an image summary.
- *     image_with_box = tf.image.draw_bounding_boxes(tf.expand_dims(image, 0),
- *                                                   bbox_for_draw)
- *     tf.summary.image('images_with_box', image_with_box)
- * 
- *     # Employ the bounding box to distort the image.
- *     distorted_image = tf.slice(image, begin, size)
- * }</pre>
- * Note that if no bounding box information is available, setting
- * `use_image_if_no_bounding_boxes = true` will assume there is a single implicit
- * bounding box covering the whole image. If `use_image_if_no_bounding_boxes` is
- * false and no bounding boxes are supplied, an error is raised.
- * 
- * @param <T> data type for {@code begin()} output
- */
-@Operator(group = "image")
-public final class SampleDistortedBoundingBox<T extends TNumber> extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.image.SampleDistortedBoundingBox}
-   */
-  public static class Options {
-    
-    /**
-     * @param seed If either `seed` or `seed2` are set to non-zero, the random number
-     * generator is seeded by the given `seed`.  Otherwise, it is seeded by a random
-     * seed.
-     */
-    public Options seed(Long seed) {
-      this.seed = seed;
-      return this;
-    }
-    
-    /**
-     * @param seed2 A second seed to avoid seed collision.
-     */
-    public Options seed2(Long seed2) {
-      this.seed2 = seed2;
-      return this;
-    }
-    
-    /**
-     * @param aspectRatioRange The cropped area of the image must have an aspect ratio =
-     * width / height within this range.
-     */
-    public Options aspectRatioRange(List<Float> aspectRatioRange) {
-      this.aspectRatioRange = aspectRatioRange;
-      return this;
-    }
-    
-    /**
-     * @param areaRange The cropped area of the image must contain a fraction of the
-     * supplied image within this range.
-     */
-    public Options areaRange(List<Float> areaRange) {
-      this.areaRange = areaRange;
-      return this;
-    }
-    
-    /**
-     * @param maxAttempts Number of attempts at generating a cropped region of the image
-     * of the specified constraints. After `max_attempts` failures, return the entire
-     * image.
-     */
-    public Options maxAttempts(Long maxAttempts) {
-      this.maxAttempts = maxAttempts;
-      return this;
-    }
-    
-    /**
-     * @param useImageIfNoBoundingBoxes Controls behavior if no bounding boxes supplied.
-     * If true, assume an implicit bounding box covering the whole input. If false,
-     * raise an error.
-     */
-    public Options useImageIfNoBoundingBoxes(Boolean useImageIfNoBoundingBoxes) {
-      this.useImageIfNoBoundingBoxes = useImageIfNoBoundingBoxes;
-      return this;
-    }
-    
-    private Long seed;
-    private Long seed2;
-    private List<Float> aspectRatioRange;
-    private List<Float> areaRange;
-    private Long maxAttempts;
-    private Boolean useImageIfNoBoundingBoxes;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new SampleDistortedBoundingBox operation.
-   * 
-   * @param scope current scope
-   * @param imageSize 1-D, containing `[height, width, channels]`.
-   * @param boundingBoxes 3-D with shape `[batch, N, 4]` describing the N bounding boxes
-   * associated with the image.
-   * @param minObjectCovered The cropped area of the image must contain at least this
-   * fraction of any bounding box supplied. The value of this parameter should be
-   * non-negative. In the case of 0, the cropped area does not need to overlap
-   * any of the bounding boxes supplied.
-   * @param options carries optional attributes values
-   * @return a new instance of SampleDistortedBoundingBox
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> SampleDistortedBoundingBox<T> create(Scope scope, Operand<T> imageSize, Operand<TFloat32> boundingBoxes, Operand<TFloat32> minObjectCovered, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("SampleDistortedBoundingBoxV2", scope.makeOpName("SampleDistortedBoundingBox"));
-    opBuilder.addInput(imageSize.asOutput(scope));
-    opBuilder.addInput(boundingBoxes.asOutput(scope));
-    opBuilder.addInput(minObjectCovered.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.seed != null) {
-          opBuilder.setAttr("seed", opts.seed);
-        }
-        if (opts.seed2 != null) {
-          opBuilder.setAttr("seed2", opts.seed2);
-        }
-        if (opts.aspectRatioRange != null) {
-          float[] aspectRatioRangeArray = new float[opts.aspectRatioRange.size()];
-          for (int i = 0; i < aspectRatioRangeArray.length; ++i) {
-            aspectRatioRangeArray[i] = opts.aspectRatioRange.get(i);
-          }
-          opBuilder.setAttr("aspect_ratio_range", aspectRatioRangeArray);
-        }
-        if (opts.areaRange != null) {
-          float[] areaRangeArray = new float[opts.areaRange.size()];
-          for (int i = 0; i < areaRangeArray.length; ++i) {
-            areaRangeArray[i] = opts.areaRange.get(i);
-          }
-          opBuilder.setAttr("area_range", areaRangeArray);
-        }
-        if (opts.maxAttempts != null) {
-          opBuilder.setAttr("max_attempts", opts.maxAttempts);
-        }
-        if (opts.useImageIfNoBoundingBoxes != null) {
-          opBuilder.setAttr("use_image_if_no_bounding_boxes", opts.useImageIfNoBoundingBoxes);
-        }
-      }
-    }
-    return new SampleDistortedBoundingBox<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param seed If either `seed` or `seed2` are set to non-zero, the random number
-   * generator is seeded by the given `seed`.  Otherwise, it is seeded by a random
-   * seed.
-   */
-  public static Options seed(Long seed) {
-    return new Options().seed(seed);
-  }
-  
-  /**
-   * @param seed2 A second seed to avoid seed collision.
-   */
-  public static Options seed2(Long seed2) {
-    return new Options().seed2(seed2);
-  }
-  
-  /**
-   * @param aspectRatioRange The cropped area of the image must have an aspect ratio =
-   * width / height within this range.
-   */
-  public static Options aspectRatioRange(List<Float> aspectRatioRange) {
-    return new Options().aspectRatioRange(aspectRatioRange);
-  }
-  
-  /**
-   * @param areaRange The cropped area of the image must contain a fraction of the
-   * supplied image within this range.
-   */
-  public static Options areaRange(List<Float> areaRange) {
-    return new Options().areaRange(areaRange);
-  }
-  
-  /**
-   * @param maxAttempts Number of attempts at generating a cropped region of the image
-   * of the specified constraints. After `max_attempts` failures, return the entire
-   * image.
-   */
-  public static Options maxAttempts(Long maxAttempts) {
-    return new Options().maxAttempts(maxAttempts);
-  }
-  
-  /**
-   * @param useImageIfNoBoundingBoxes Controls behavior if no bounding boxes supplied.
-   * If true, assume an implicit bounding box covering the whole input. If false,
-   * raise an error.
-   */
-  public static Options useImageIfNoBoundingBoxes(Boolean useImageIfNoBoundingBoxes) {
-    return new Options().useImageIfNoBoundingBoxes(useImageIfNoBoundingBoxes);
-  }
-  
-  /**
-   * 1-D, containing `[offset_height, offset_width, 0]`. Provide as input to
-   * `tf.slice`.
-   */
-  public Output<T> begin() {
-    return begin;
-  }
-  
-  /**
-   * 1-D, containing `[target_height, target_width, -1]`. Provide as input to
-   * `tf.slice`.
-   */
-  public Output<T> size() {
-    return size;
-  }
-  
-  /**
-   * 3-D with shape `[1, 1, 4]` containing the distorted bounding box.
-   * Provide as input to `tf.image.draw_bounding_boxes`.
-   */
-  public Output<TFloat32> bboxes() {
-    return bboxes;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "SampleDistortedBoundingBoxV2";
-  
-  private Output<T> begin;
-  private Output<T> size;
-  private Output<TFloat32> bboxes;
-  
-  private SampleDistortedBoundingBox(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    begin = operation.output(outputIdx++);
-    size = operation.output(outputIdx++);
-    bboxes = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ScaleAndTranslate.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ScaleAndTranslate.java
deleted file mode 100644
index 12928534943..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ScaleAndTranslate.java
+++ /dev/null
@@ -1,132 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.image;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.family.TNumber;
-
-/**
- */
-@Operator(group = "image")
-public final class ScaleAndTranslate extends RawOp implements Operand<TFloat32> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.image.ScaleAndTranslate}
-   */
-  public static class Options {
-    
-    /**
-     * @param kernelType 
-     */
-    public Options kernelType(String kernelType) {
-      this.kernelType = kernelType;
-      return this;
-    }
-    
-    /**
-     * @param antialias 
-     */
-    public Options antialias(Boolean antialias) {
-      this.antialias = antialias;
-      return this;
-    }
-    
-    private String kernelType;
-    private Boolean antialias;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new ScaleAndTranslate operation.
-   * 
-   * @param scope current scope
-   * @param images 
-   * @param size 
-   * @param scale 
-   * @param translation 
-   * @param options carries optional attributes values
-   * @return a new instance of ScaleAndTranslate
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> ScaleAndTranslate create(Scope scope, Operand<T> images, Operand<TInt32> size, Operand<TFloat32> scale, Operand<TFloat32> translation, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ScaleAndTranslate", scope.makeOpName("ScaleAndTranslate"));
-    opBuilder.addInput(images.asOutput(scope));
-    opBuilder.addInput(size.asOutput(scope));
-    opBuilder.addInput(scale.asOutput(scope));
-    opBuilder.addInput(translation.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.kernelType != null) {
-          opBuilder.setAttr("kernel_type", opts.kernelType);
-        }
-        if (opts.antialias != null) {
-          opBuilder.setAttr("antialias", opts.antialias);
-        }
-      }
-    }
-    return new ScaleAndTranslate(opBuilder.build());
-  }
-  
-  /**
-   * @param kernelType 
-   */
-  public static Options kernelType(String kernelType) {
-    return new Options().kernelType(kernelType);
-  }
-  
-  /**
-   * @param antialias 
-   */
-  public static Options antialias(Boolean antialias) {
-    return new Options().antialias(antialias);
-  }
-  
-  /**
-   */
-  public Output<TFloat32> resizedImages() {
-    return resizedImages;
-  }
-  
-  @Override
-  public Output<TFloat32> asOutput(Scope scope) {
-    return resizedImages;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ScaleAndTranslate";
-  
-  private Output<TFloat32> resizedImages;
-  
-  private ScaleAndTranslate(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    resizedImages = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ScaleAndTranslateGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ScaleAndTranslateGrad.java
deleted file mode 100644
index 2c0c858b38c..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ScaleAndTranslateGrad.java
+++ /dev/null
@@ -1,131 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.image;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * @param <T> data type for {@code output()} output
- */
-public final class ScaleAndTranslateGrad<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.image.ScaleAndTranslateGrad}
-   */
-  public static class Options {
-    
-    /**
-     * @param kernelType 
-     */
-    public Options kernelType(String kernelType) {
-      this.kernelType = kernelType;
-      return this;
-    }
-    
-    /**
-     * @param antialias 
-     */
-    public Options antialias(Boolean antialias) {
-      this.antialias = antialias;
-      return this;
-    }
-    
-    private String kernelType;
-    private Boolean antialias;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new ScaleAndTranslateGrad operation.
-   * 
-   * @param scope current scope
-   * @param grads 
-   * @param originalImage 
-   * @param scale 
-   * @param translation 
-   * @param options carries optional attributes values
-   * @return a new instance of ScaleAndTranslateGrad
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> ScaleAndTranslateGrad<T> create(Scope scope, Operand<T> grads, Operand<T> originalImage, Operand<TFloat32> scale, Operand<TFloat32> translation, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ScaleAndTranslateGrad", scope.makeOpName("ScaleAndTranslateGrad"));
-    opBuilder.addInput(grads.asOutput(scope));
-    opBuilder.addInput(originalImage.asOutput(scope));
-    opBuilder.addInput(scale.asOutput(scope));
-    opBuilder.addInput(translation.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.kernelType != null) {
-          opBuilder.setAttr("kernel_type", opts.kernelType);
-        }
-        if (opts.antialias != null) {
-          opBuilder.setAttr("antialias", opts.antialias);
-        }
-      }
-    }
-    return new ScaleAndTranslateGrad<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param kernelType 
-   */
-  public static Options kernelType(String kernelType) {
-    return new Options().kernelType(kernelType);
-  }
-  
-  /**
-   * @param antialias 
-   */
-  public static Options antialias(Boolean antialias) {
-    return new Options().antialias(antialias);
-  }
-  
-  /**
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ScaleAndTranslateGrad";
-  
-  private Output<T> output;
-  
-  private ScaleAndTranslateGrad(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/DecodeBase64.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/DecodeBase64.java
deleted file mode 100644
index 4545ccd70c9..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/DecodeBase64.java
+++ /dev/null
@@ -1,76 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.io;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TString;
-
-/**
- * Decode web-safe base64-encoded strings.
- * <p>
- * Input may or may not have padding at the end. See EncodeBase64 for padding.
- * Web-safe means that input must use - and _ instead of + and /.
- */
-@Operator(group = "io")
-public final class DecodeBase64 extends RawOp implements Operand<TString> {
-  
-  /**
-   * Factory method to create a class wrapping a new DecodeBase64 operation.
-   * 
-   * @param scope current scope
-   * @param input Base64 strings to decode.
-   * @return a new instance of DecodeBase64
-   */
-  @Endpoint(describeByClass = true)
-  public static DecodeBase64 create(Scope scope, Operand<TString> input) {
-    OperationBuilder opBuilder = scope.env().opBuilder("DecodeBase64", scope.makeOpName("DecodeBase64"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new DecodeBase64(opBuilder.build());
-  }
-  
-  /**
-   * Decoded strings.
-   */
-  public Output<TString> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<TString> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "DecodeBase64";
-  
-  private Output<TString> output;
-  
-  private DecodeBase64(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/DecodeCompressed.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/DecodeCompressed.java
deleted file mode 100644
index b314e3b31b5..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/DecodeCompressed.java
+++ /dev/null
@@ -1,117 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.io;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TString;
-
-/**
- * Decompress strings.
- * <p>
- * This op decompresses each element of the `bytes` input `Tensor`, which
- * is assumed to be compressed using the given `compression_type`.
- * <p>
- * The `output` is a string `Tensor` of the same shape as `bytes`,
- * each element containing the decompressed data from the corresponding
- * element in `bytes`.
- */
-@Operator(group = "io")
-public final class DecodeCompressed extends RawOp implements Operand<TString> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.io.DecodeCompressed}
-   */
-  public static class Options {
-    
-    /**
-     * @param compressionType A scalar containing either (i) the empty string (no
-     * compression), (ii) "ZLIB", or (iii) "GZIP".
-     */
-    public Options compressionType(String compressionType) {
-      this.compressionType = compressionType;
-      return this;
-    }
-    
-    private String compressionType;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new DecodeCompressed operation.
-   * 
-   * @param scope current scope
-   * @param bytes A Tensor of string which is compressed.
-   * @param options carries optional attributes values
-   * @return a new instance of DecodeCompressed
-   */
-  @Endpoint(describeByClass = true)
-  public static DecodeCompressed create(Scope scope, Operand<TString> bytes, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("DecodeCompressed", scope.makeOpName("DecodeCompressed"));
-    opBuilder.addInput(bytes.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.compressionType != null) {
-          opBuilder.setAttr("compression_type", opts.compressionType);
-        }
-      }
-    }
-    return new DecodeCompressed(opBuilder.build());
-  }
-  
-  /**
-   * @param compressionType A scalar containing either (i) the empty string (no
-   * compression), (ii) "ZLIB", or (iii) "GZIP".
-   */
-  public static Options compressionType(String compressionType) {
-    return new Options().compressionType(compressionType);
-  }
-  
-  /**
-   * A Tensor with the same shape as input `bytes`, uncompressed
-   * from bytes.
-   */
-  public Output<TString> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<TString> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "DecodeCompressed";
-  
-  private Output<TString> output;
-  
-  private DecodeCompressed(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/DecodeCsv.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/DecodeCsv.java
deleted file mode 100644
index 84fe9d1d814..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/DecodeCsv.java
+++ /dev/null
@@ -1,189 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.io;
-
-import java.util.Arrays;
-import java.util.Iterator;
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.Operands;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TString;
-import org.tensorflow.types.family.TType;
-
-/**
- * Convert CSV records to tensors. Each column maps to one tensor.
- * <p>
- * RFC 4180 format is expected for the CSV records.
- * (https://tools.ietf.org/html/rfc4180)
- * Note that we allow leading and trailing spaces with int or float field.
- */
-@Operator(group = "io")
-public final class DecodeCsv extends RawOp implements Iterable<Operand<TType>> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.io.DecodeCsv}
-   */
-  public static class Options {
-    
-    /**
-     * @param fieldDelim char delimiter to separate fields in a record.
-     */
-    public Options fieldDelim(String fieldDelim) {
-      this.fieldDelim = fieldDelim;
-      return this;
-    }
-    
-    /**
-     * @param useQuoteDelim If false, treats double quotation marks as regular
-     * characters inside of the string fields (ignoring RFC 4180, Section 2,
-     * Bullet 5).
-     */
-    public Options useQuoteDelim(Boolean useQuoteDelim) {
-      this.useQuoteDelim = useQuoteDelim;
-      return this;
-    }
-    
-    /**
-     * @param naValue Additional string to recognize as NA/NaN.
-     */
-    public Options naValue(String naValue) {
-      this.naValue = naValue;
-      return this;
-    }
-    
-    /**
-     * @param selectCols 
-     */
-    public Options selectCols(List<Long> selectCols) {
-      this.selectCols = selectCols;
-      return this;
-    }
-    
-    private String fieldDelim;
-    private Boolean useQuoteDelim;
-    private String naValue;
-    private List<Long> selectCols;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new DecodeCsv operation.
-   * 
-   * @param scope current scope
-   * @param records Each string is a record/row in the csv and all records should have
-   * the same format.
-   * @param recordDefaults One tensor per column of the input record, with either a
-   * scalar default value for that column or an empty vector if the column is
-   * required.
-   * @param options carries optional attributes values
-   * @return a new instance of DecodeCsv
-   */
-  @Endpoint(describeByClass = true)
-  public static DecodeCsv create(Scope scope, Operand<TString> records, Iterable<Operand<?>> recordDefaults, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("DecodeCSV", scope.makeOpName("DecodeCsv"));
-    opBuilder.addInput(records.asOutput(scope));
-    opBuilder.addInputList(Operands.asOutputs(scope, recordDefaults));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.fieldDelim != null) {
-          opBuilder.setAttr("field_delim", opts.fieldDelim);
-        }
-        if (opts.useQuoteDelim != null) {
-          opBuilder.setAttr("use_quote_delim", opts.useQuoteDelim);
-        }
-        if (opts.naValue != null) {
-          opBuilder.setAttr("na_value", opts.naValue);
-        }
-        if (opts.selectCols != null) {
-          long[] selectColsArray = new long[opts.selectCols.size()];
-          for (int i = 0; i < selectColsArray.length; ++i) {
-            selectColsArray[i] = opts.selectCols.get(i);
-          }
-          opBuilder.setAttr("select_cols", selectColsArray);
-        }
-      }
-    }
-    return new DecodeCsv(opBuilder.build());
-  }
-  
-  /**
-   * @param fieldDelim char delimiter to separate fields in a record.
-   */
-  public static Options fieldDelim(String fieldDelim) {
-    return new Options().fieldDelim(fieldDelim);
-  }
-  
-  /**
-   * @param useQuoteDelim If false, treats double quotation marks as regular
-   * characters inside of the string fields (ignoring RFC 4180, Section 2,
-   * Bullet 5).
-   */
-  public static Options useQuoteDelim(Boolean useQuoteDelim) {
-    return new Options().useQuoteDelim(useQuoteDelim);
-  }
-  
-  /**
-   * @param naValue Additional string to recognize as NA/NaN.
-   */
-  public static Options naValue(String naValue) {
-    return new Options().naValue(naValue);
-  }
-  
-  /**
-   * @param selectCols 
-   */
-  public static Options selectCols(List<Long> selectCols) {
-    return new Options().selectCols(selectCols);
-  }
-  
-  /**
-   * Each tensor will have the same shape as records.
-   */
-  public List<Output<?>> output() {
-    return output;
-  }
-  
-  @Override
-  @SuppressWarnings({"rawtypes", "unchecked"})
-  public Iterator<Operand<TType>> iterator() {
-    return (Iterator) output.iterator();
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "DecodeCSV";
-  
-  private List<Output<?>> output;
-  
-  private DecodeCsv(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    int outputLength = operation.outputListLength("output");
-    output = Arrays.asList(operation.outputList(outputIdx, outputLength));
-    outputIdx += outputLength;
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/DecodeJsonExample.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/DecodeJsonExample.java
deleted file mode 100644
index c46049e8e9b..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/DecodeJsonExample.java
+++ /dev/null
@@ -1,82 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.io;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TString;
-
-/**
- * Convert JSON-encoded Example records to binary protocol buffer strings.
- * <p>
- * This op translates a tensor containing Example records, encoded using
- * the [standard JSON
- * mapping](https://developers.google.com/protocol-buffers/docs/proto3#json),
- * into a tensor containing the same records encoded as binary protocol
- * buffers. The resulting tensor can then be fed to any of the other
- * Example-parsing ops.
- */
-@Operator(group = "io")
-public final class DecodeJsonExample extends RawOp implements Operand<TString> {
-  
-  /**
-   * Factory method to create a class wrapping a new DecodeJsonExample operation.
-   * 
-   * @param scope current scope
-   * @param jsonExamples Each string is a JSON object serialized according to the JSON
-   * mapping of the Example proto.
-   * @return a new instance of DecodeJsonExample
-   */
-  @Endpoint(describeByClass = true)
-  public static DecodeJsonExample create(Scope scope, Operand<TString> jsonExamples) {
-    OperationBuilder opBuilder = scope.env().opBuilder("DecodeJSONExample", scope.makeOpName("DecodeJsonExample"));
-    opBuilder.addInput(jsonExamples.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new DecodeJsonExample(opBuilder.build());
-  }
-  
-  /**
-   * Each string is a binary Example protocol buffer corresponding
-   * to the respective element of `json_examples`.
-   */
-  public Output<TString> binaryExamples() {
-    return binaryExamples;
-  }
-  
-  @Override
-  public Output<TString> asOutput(Scope scope) {
-    return binaryExamples;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "DecodeJSONExample";
-  
-  private Output<TString> binaryExamples;
-  
-  private DecodeJsonExample(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    binaryExamples = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/DecodePaddedRaw.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/DecodePaddedRaw.java
deleted file mode 100644
index a7520073665..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/DecodePaddedRaw.java
+++ /dev/null
@@ -1,120 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.io;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.TString;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Reinterpret the bytes of a string as a vector of numbers.
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator(group = "io")
-public final class DecodePaddedRaw<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.io.DecodePaddedRaw}
-   */
-  public static class Options {
-    
-    /**
-     * @param littleEndian Whether the input `input_bytes` is in little-endian order. Ignored for
-     * `out_type` values that are stored in a single byte, like `uint8`
-     */
-    public Options littleEndian(Boolean littleEndian) {
-      this.littleEndian = littleEndian;
-      return this;
-    }
-    
-    private Boolean littleEndian;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new DecodePaddedRaw operation.
-   * 
-   * @param scope current scope
-   * @param inputBytes Tensor of string to be decoded.
-   * @param fixedLength Length in bytes for each element of the decoded output. Must be a multiple
-   * of the size of the output type.
-   * @param outType 
-   * @param options carries optional attributes values
-   * @return a new instance of DecodePaddedRaw
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> DecodePaddedRaw<T> create(Scope scope, Operand<TString> inputBytes, Operand<TInt32> fixedLength, Class<T> outType, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("DecodePaddedRaw", scope.makeOpName("DecodePaddedRaw"));
-    opBuilder.addInput(inputBytes.asOutput(scope));
-    opBuilder.addInput(fixedLength.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("out_type", outType);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.littleEndian != null) {
-          opBuilder.setAttr("little_endian", opts.littleEndian);
-        }
-      }
-    }
-    return new DecodePaddedRaw<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param littleEndian Whether the input `input_bytes` is in little-endian order. Ignored for
-   * `out_type` values that are stored in a single byte, like `uint8`
-   */
-  public static Options littleEndian(Boolean littleEndian) {
-    return new Options().littleEndian(littleEndian);
-  }
-  
-  /**
-   * A Tensor with one more dimension than the input `bytes`. The added dimension
-   * will have size equal to the length of the elements of `bytes` divided by the
-   * number of bytes to represent `out_type`.
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "DecodePaddedRaw";
-  
-  private Output<T> output;
-  
-  private DecodePaddedRaw(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/DecodeRaw.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/DecodeRaw.java
deleted file mode 100644
index 97f5543c8d4..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/DecodeRaw.java
+++ /dev/null
@@ -1,118 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.io;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TString;
-import org.tensorflow.types.family.TType;
-
-/**
- * Reinterpret the bytes of a string as a vector of numbers.
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator(group = "io")
-public final class DecodeRaw<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.io.DecodeRaw}
-   */
-  public static class Options {
-    
-    /**
-     * @param littleEndian Whether the input `bytes` are in little-endian order.
-     * Ignored for `out_type` values that are stored in a single byte like
-     * `uint8`.
-     */
-    public Options littleEndian(Boolean littleEndian) {
-      this.littleEndian = littleEndian;
-      return this;
-    }
-    
-    private Boolean littleEndian;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new DecodeRaw operation.
-   * 
-   * @param scope current scope
-   * @param bytes All the elements must have the same length.
-   * @param outType 
-   * @param options carries optional attributes values
-   * @return a new instance of DecodeRaw
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> DecodeRaw<T> create(Scope scope, Operand<TString> bytes, Class<T> outType, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("DecodeRaw", scope.makeOpName("DecodeRaw"));
-    opBuilder.addInput(bytes.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("out_type", outType);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.littleEndian != null) {
-          opBuilder.setAttr("little_endian", opts.littleEndian);
-        }
-      }
-    }
-    return new DecodeRaw<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param littleEndian Whether the input `bytes` are in little-endian order.
-   * Ignored for `out_type` values that are stored in a single byte like
-   * `uint8`.
-   */
-  public static Options littleEndian(Boolean littleEndian) {
-    return new Options().littleEndian(littleEndian);
-  }
-  
-  /**
-   * A Tensor with one more dimension than the input `bytes`.  The
-   * added dimension will have size equal to the length of the elements
-   * of `bytes` divided by the number of bytes to represent `out_type`.
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "DecodeRaw";
-  
-  private Output<T> output;
-  
-  private DecodeRaw(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/DeserializeManySparse.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/DeserializeManySparse.java
deleted file mode 100644
index 43d30a48a90..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/DeserializeManySparse.java
+++ /dev/null
@@ -1,132 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.io;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.TString;
-import org.tensorflow.types.family.TType;
-
-/**
- * Deserialize and concatenate `SparseTensors` from a serialized minibatch.
- * <p>
- * The input `serialized_sparse` must be a string matrix of shape `[N x 3]` where
- * `N` is the minibatch size and the rows correspond to packed outputs of
- * `SerializeSparse`.  The ranks of the original `SparseTensor` objects
- * must all match.  When the final `SparseTensor` is created, it has rank one
- * higher than the ranks of the incoming `SparseTensor` objects
- * (they have been concatenated along a new row dimension).
- * <p>
- * The output `SparseTensor` object's shape values for all dimensions but the
- * first are the max across the input `SparseTensor` objects' shape values
- * for the corresponding dimensions.  Its first shape value is `N`, the minibatch
- * size.
- * <p>
- * The input `SparseTensor` objects' indices are assumed ordered in
- * standard lexicographic order.  If this is not the case, after this
- * step run `SparseReorder` to restore index ordering.
- * <p>
- * For example, if the serialized input is a `[2 x 3]` matrix representing two
- * original `SparseTensor` objects:
- * <p>
- *     index = [ 0]
- *             [10]
- *             [20]
- *     values = [1, 2, 3]
- *     shape = [50]
- * <p>
- * and
- * <p>
- *     index = [ 2]
- *             [10]
- *     values = [4, 5]
- *     shape = [30]
- * <p>
- * then the final deserialized `SparseTensor` will be:
- * <p>
- *     index = [0  0]
- *             [0 10]
- *             [0 20]
- *             [1  2]
- *             [1 10]
- *     values = [1, 2, 3, 4, 5]
- *     shape = [2 50]
- * 
- * @param <T> data type for {@code sparseValues()} output
- */
-@Operator(group = "io")
-public final class DeserializeManySparse<T extends TType> extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new DeserializeManySparse operation.
-   * 
-   * @param scope current scope
-   * @param serializedSparse 2-D, The `N` serialized `SparseTensor` objects.
-   * Must have 3 columns.
-   * @param dtype The `dtype` of the serialized `SparseTensor` objects.
-   * @return a new instance of DeserializeManySparse
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> DeserializeManySparse<T> create(Scope scope, Operand<TString> serializedSparse, Class<T> dtype) {
-    OperationBuilder opBuilder = scope.env().opBuilder("DeserializeManySparse", scope.makeOpName("DeserializeManySparse"));
-    opBuilder.addInput(serializedSparse.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("dtype", dtype);
-    return new DeserializeManySparse<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<TInt64> sparseIndices() {
-    return sparseIndices;
-  }
-  
-  /**
-   */
-  public Output<T> sparseValues() {
-    return sparseValues;
-  }
-  
-  /**
-   */
-  public Output<TInt64> sparseShape() {
-    return sparseShape;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "DeserializeManySparse";
-  
-  private Output<TInt64> sparseIndices;
-  private Output<T> sparseValues;
-  private Output<TInt64> sparseShape;
-  
-  private DeserializeManySparse(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    sparseIndices = operation.output(outputIdx++);
-    sparseValues = operation.output(outputIdx++);
-    sparseShape = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/EncodeBase64.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/EncodeBase64.java
deleted file mode 100644
index 219ad0c422b..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/EncodeBase64.java
+++ /dev/null
@@ -1,114 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.io;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TString;
-
-/**
- * Encode strings into web-safe base64 format.
- * <p>
- * Refer to the following article for more information on base64 format:
- * en.wikipedia.org/wiki/Base64. Base64 strings may have padding with '=' at the
- * end so that the encoded has length multiple of 4. See Padding section of the
- * link above.
- * <p>
- * Web-safe means that the encoder uses - and _ instead of + and /.
- */
-@Operator(group = "io")
-public final class EncodeBase64 extends RawOp implements Operand<TString> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.io.EncodeBase64}
-   */
-  public static class Options {
-    
-    /**
-     * @param pad Bool whether padding is applied at the ends.
-     */
-    public Options pad(Boolean pad) {
-      this.pad = pad;
-      return this;
-    }
-    
-    private Boolean pad;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new EncodeBase64 operation.
-   * 
-   * @param scope current scope
-   * @param input Strings to be encoded.
-   * @param options carries optional attributes values
-   * @return a new instance of EncodeBase64
-   */
-  @Endpoint(describeByClass = true)
-  public static EncodeBase64 create(Scope scope, Operand<TString> input, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("EncodeBase64", scope.makeOpName("EncodeBase64"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.pad != null) {
-          opBuilder.setAttr("pad", opts.pad);
-        }
-      }
-    }
-    return new EncodeBase64(opBuilder.build());
-  }
-  
-  /**
-   * @param pad Bool whether padding is applied at the ends.
-   */
-  public static Options pad(Boolean pad) {
-    return new Options().pad(pad);
-  }
-  
-  /**
-   * Input strings encoded in base64.
-   */
-  public Output<TString> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<TString> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "EncodeBase64";
-  
-  private Output<TString> output;
-  
-  private EncodeBase64(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/FifoQueue.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/FifoQueue.java
deleted file mode 100644
index ca239776907..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/FifoQueue.java
+++ /dev/null
@@ -1,187 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.io;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * A queue that produces elements in first-in first-out order.
- */
-@Operator(group = "io")
-public final class FifoQueue extends RawOp implements Operand<TType> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.io.FifoQueue}
-   */
-  public static class Options {
-    
-    /**
-     * @param shapes The shape of each component in a value. The length of this attr must
-     * be either 0 or the same as the length of component_types. If the length of
-     * this attr is 0, the shapes of queue elements are not constrained, and
-     * only one element may be dequeued at a time.
-     */
-    public Options shapes(List<Shape> shapes) {
-      this.shapes = shapes;
-      return this;
-    }
-    
-    /**
-     * @param capacity The upper bound on the number of elements in this queue.
-     * Negative numbers mean no limit.
-     */
-    public Options capacity(Long capacity) {
-      this.capacity = capacity;
-      return this;
-    }
-    
-    /**
-     * @param container If non-empty, this queue is placed in the given container.
-     * Otherwise, a default container is used.
-     */
-    public Options container(String container) {
-      this.container = container;
-      return this;
-    }
-    
-    /**
-     * @param sharedName If non-empty, this queue will be shared under the given name
-     * across multiple sessions.
-     */
-    public Options sharedName(String sharedName) {
-      this.sharedName = sharedName;
-      return this;
-    }
-    
-    private List<Shape> shapes;
-    private Long capacity;
-    private String container;
-    private String sharedName;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new FifoQueue operation.
-   * 
-   * @param scope current scope
-   * @param componentTypes The type of each component in a value.
-   * @param options carries optional attributes values
-   * @return a new instance of FifoQueue
-   */
-  @Endpoint(describeByClass = true)
-  public static FifoQueue create(Scope scope, List<Class<?>> componentTypes, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("FIFOQueueV2", scope.makeOpName("FifoQueue"));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    Class[] componentTypesArray = new Class[componentTypes.size()];
-    for (int i = 0; i < componentTypesArray.length; ++i) {
-      componentTypesArray[i] = componentTypes.get(i);
-    }
-    opBuilder.setAttr("component_types", componentTypesArray);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.shapes != null) {
-          Shape[] shapesArray = new Shape[opts.shapes.size()];
-          for (int i = 0; i < shapesArray.length; ++i) {
-            shapesArray[i] = opts.shapes.get(i);
-          }
-          opBuilder.setAttr("shapes", shapesArray);
-        }
-        if (opts.capacity != null) {
-          opBuilder.setAttr("capacity", opts.capacity);
-        }
-        if (opts.container != null) {
-          opBuilder.setAttr("container", opts.container);
-        }
-        if (opts.sharedName != null) {
-          opBuilder.setAttr("shared_name", opts.sharedName);
-        }
-      }
-    }
-    return new FifoQueue(opBuilder.build());
-  }
-  
-  /**
-   * @param shapes The shape of each component in a value. The length of this attr must
-   * be either 0 or the same as the length of component_types. If the length of
-   * this attr is 0, the shapes of queue elements are not constrained, and
-   * only one element may be dequeued at a time.
-   */
-  public static Options shapes(List<Shape> shapes) {
-    return new Options().shapes(shapes);
-  }
-  
-  /**
-   * @param capacity The upper bound on the number of elements in this queue.
-   * Negative numbers mean no limit.
-   */
-  public static Options capacity(Long capacity) {
-    return new Options().capacity(capacity);
-  }
-  
-  /**
-   * @param container If non-empty, this queue is placed in the given container.
-   * Otherwise, a default container is used.
-   */
-  public static Options container(String container) {
-    return new Options().container(container);
-  }
-  
-  /**
-   * @param sharedName If non-empty, this queue will be shared under the given name
-   * across multiple sessions.
-   */
-  public static Options sharedName(String sharedName) {
-    return new Options().sharedName(sharedName);
-  }
-  
-  /**
-   * The handle to the queue.
-   */
-  public Output<?> handle() {
-    return handle;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) handle;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "FIFOQueueV2";
-  
-  private Output<?> handle;
-  
-  private FifoQueue(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    handle = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/FixedLengthRecordReader.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/FixedLengthRecordReader.java
deleted file mode 100644
index 2cc926ea765..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/FixedLengthRecordReader.java
+++ /dev/null
@@ -1,211 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.io;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * A Reader that outputs fixed-length records from a file.
- */
-@Operator(group = "io")
-public final class FixedLengthRecordReader extends RawOp implements Operand<TType> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.io.FixedLengthRecordReader}
-   */
-  public static class Options {
-    
-    /**
-     * @param headerBytes Number of bytes in the header, defaults to 0.
-     */
-    public Options headerBytes(Long headerBytes) {
-      this.headerBytes = headerBytes;
-      return this;
-    }
-    
-    /**
-     * @param footerBytes Number of bytes in the footer, defaults to 0.
-     */
-    public Options footerBytes(Long footerBytes) {
-      this.footerBytes = footerBytes;
-      return this;
-    }
-    
-    /**
-     * @param hopBytes Number of bytes to hop before each read. Default of 0 means using
-     * record_bytes.
-     */
-    public Options hopBytes(Long hopBytes) {
-      this.hopBytes = hopBytes;
-      return this;
-    }
-    
-    /**
-     * @param container If non-empty, this reader is placed in the given container.
-     * Otherwise, a default container is used.
-     */
-    public Options container(String container) {
-      this.container = container;
-      return this;
-    }
-    
-    /**
-     * @param sharedName If non-empty, this reader is named in the given bucket
-     * with this shared_name. Otherwise, the node name is used instead.
-     */
-    public Options sharedName(String sharedName) {
-      this.sharedName = sharedName;
-      return this;
-    }
-    
-    /**
-     * @param encoding The type of encoding for the file. Currently ZLIB and GZIP
-     * are supported. Defaults to none.
-     */
-    public Options encoding(String encoding) {
-      this.encoding = encoding;
-      return this;
-    }
-    
-    private Long headerBytes;
-    private Long footerBytes;
-    private Long hopBytes;
-    private String container;
-    private String sharedName;
-    private String encoding;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new FixedLengthRecordReader operation.
-   * 
-   * @param scope current scope
-   * @param recordBytes Number of bytes in the record.
-   * @param options carries optional attributes values
-   * @return a new instance of FixedLengthRecordReader
-   */
-  @Endpoint(describeByClass = true)
-  public static FixedLengthRecordReader create(Scope scope, Long recordBytes, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("FixedLengthRecordReaderV2", scope.makeOpName("FixedLengthRecordReader"));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("record_bytes", recordBytes);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.headerBytes != null) {
-          opBuilder.setAttr("header_bytes", opts.headerBytes);
-        }
-        if (opts.footerBytes != null) {
-          opBuilder.setAttr("footer_bytes", opts.footerBytes);
-        }
-        if (opts.hopBytes != null) {
-          opBuilder.setAttr("hop_bytes", opts.hopBytes);
-        }
-        if (opts.container != null) {
-          opBuilder.setAttr("container", opts.container);
-        }
-        if (opts.sharedName != null) {
-          opBuilder.setAttr("shared_name", opts.sharedName);
-        }
-        if (opts.encoding != null) {
-          opBuilder.setAttr("encoding", opts.encoding);
-        }
-      }
-    }
-    return new FixedLengthRecordReader(opBuilder.build());
-  }
-  
-  /**
-   * @param headerBytes Number of bytes in the header, defaults to 0.
-   */
-  public static Options headerBytes(Long headerBytes) {
-    return new Options().headerBytes(headerBytes);
-  }
-  
-  /**
-   * @param footerBytes Number of bytes in the footer, defaults to 0.
-   */
-  public static Options footerBytes(Long footerBytes) {
-    return new Options().footerBytes(footerBytes);
-  }
-  
-  /**
-   * @param hopBytes Number of bytes to hop before each read. Default of 0 means using
-   * record_bytes.
-   */
-  public static Options hopBytes(Long hopBytes) {
-    return new Options().hopBytes(hopBytes);
-  }
-  
-  /**
-   * @param container If non-empty, this reader is placed in the given container.
-   * Otherwise, a default container is used.
-   */
-  public static Options container(String container) {
-    return new Options().container(container);
-  }
-  
-  /**
-   * @param sharedName If non-empty, this reader is named in the given bucket
-   * with this shared_name. Otherwise, the node name is used instead.
-   */
-  public static Options sharedName(String sharedName) {
-    return new Options().sharedName(sharedName);
-  }
-  
-  /**
-   * @param encoding The type of encoding for the file. Currently ZLIB and GZIP
-   * are supported. Defaults to none.
-   */
-  public static Options encoding(String encoding) {
-    return new Options().encoding(encoding);
-  }
-  
-  /**
-   * The handle to reference the Reader.
-   */
-  public Output<?> readerHandle() {
-    return readerHandle;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) readerHandle;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "FixedLengthRecordReaderV2";
-  
-  private Output<?> readerHandle;
-  
-  private FixedLengthRecordReader(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    readerHandle = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/IdentityReader.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/IdentityReader.java
deleted file mode 100644
index 5649e2da49f..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/IdentityReader.java
+++ /dev/null
@@ -1,132 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.io;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * A Reader that outputs the queued work as both the key and value.
- * <p>
- * To use, enqueue strings in a Queue.  ReaderRead will take the front
- * work string and output (work, work).
- */
-@Operator(group = "io")
-public final class IdentityReader extends RawOp implements Operand<TType> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.io.IdentityReader}
-   */
-  public static class Options {
-    
-    /**
-     * @param container If non-empty, this reader is placed in the given container.
-     * Otherwise, a default container is used.
-     */
-    public Options container(String container) {
-      this.container = container;
-      return this;
-    }
-    
-    /**
-     * @param sharedName If non-empty, this reader is named in the given bucket
-     * with this shared_name. Otherwise, the node name is used instead.
-     */
-    public Options sharedName(String sharedName) {
-      this.sharedName = sharedName;
-      return this;
-    }
-    
-    private String container;
-    private String sharedName;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new IdentityReader operation.
-   * 
-   * @param scope current scope
-   * @param options carries optional attributes values
-   * @return a new instance of IdentityReader
-   */
-  @Endpoint(describeByClass = true)
-  public static IdentityReader create(Scope scope, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("IdentityReaderV2", scope.makeOpName("IdentityReader"));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.container != null) {
-          opBuilder.setAttr("container", opts.container);
-        }
-        if (opts.sharedName != null) {
-          opBuilder.setAttr("shared_name", opts.sharedName);
-        }
-      }
-    }
-    return new IdentityReader(opBuilder.build());
-  }
-  
-  /**
-   * @param container If non-empty, this reader is placed in the given container.
-   * Otherwise, a default container is used.
-   */
-  public static Options container(String container) {
-    return new Options().container(container);
-  }
-  
-  /**
-   * @param sharedName If non-empty, this reader is named in the given bucket
-   * with this shared_name. Otherwise, the node name is used instead.
-   */
-  public static Options sharedName(String sharedName) {
-    return new Options().sharedName(sharedName);
-  }
-  
-  /**
-   * The handle to reference the Reader.
-   */
-  public Output<?> readerHandle() {
-    return readerHandle;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) readerHandle;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "IdentityReaderV2";
-  
-  private Output<?> readerHandle;
-  
-  private IdentityReader(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    readerHandle = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/LmdbReader.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/LmdbReader.java
deleted file mode 100644
index affdd871f44..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/LmdbReader.java
+++ /dev/null
@@ -1,128 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.io;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TString;
-
-/**
- * A Reader that outputs the records from a LMDB file.
- */
-@Operator(group = "io")
-public final class LmdbReader extends RawOp implements Operand<TString> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.io.LmdbReader}
-   */
-  public static class Options {
-    
-    /**
-     * @param container If non-empty, this reader is placed in the given container.
-     * Otherwise, a default container is used.
-     */
-    public Options container(String container) {
-      this.container = container;
-      return this;
-    }
-    
-    /**
-     * @param sharedName If non-empty, this reader is named in the given bucket
-     * with this shared_name. Otherwise, the node name is used instead.
-     */
-    public Options sharedName(String sharedName) {
-      this.sharedName = sharedName;
-      return this;
-    }
-    
-    private String container;
-    private String sharedName;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new LmdbReader operation.
-   * 
-   * @param scope current scope
-   * @param options carries optional attributes values
-   * @return a new instance of LmdbReader
-   */
-  @Endpoint(describeByClass = true)
-  public static LmdbReader create(Scope scope, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("LMDBReader", scope.makeOpName("LmdbReader"));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.container != null) {
-          opBuilder.setAttr("container", opts.container);
-        }
-        if (opts.sharedName != null) {
-          opBuilder.setAttr("shared_name", opts.sharedName);
-        }
-      }
-    }
-    return new LmdbReader(opBuilder.build());
-  }
-  
-  /**
-   * @param container If non-empty, this reader is placed in the given container.
-   * Otherwise, a default container is used.
-   */
-  public static Options container(String container) {
-    return new Options().container(container);
-  }
-  
-  /**
-   * @param sharedName If non-empty, this reader is named in the given bucket
-   * with this shared_name. Otherwise, the node name is used instead.
-   */
-  public static Options sharedName(String sharedName) {
-    return new Options().sharedName(sharedName);
-  }
-  
-  /**
-   * The handle to reference the Reader.
-   */
-  public Output<TString> readerHandle() {
-    return readerHandle;
-  }
-  
-  @Override
-  public Output<TString> asOutput(Scope scope) {
-    return readerHandle;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "LMDBReader";
-  
-  private Output<TString> readerHandle;
-  
-  private LmdbReader(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    readerHandle = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/MatchingFiles.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/MatchingFiles.java
deleted file mode 100644
index 7441e1bd9eb..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/MatchingFiles.java
+++ /dev/null
@@ -1,77 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.io;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TString;
-
-/**
- * Returns the set of files matching one or more glob patterns.
- * <p>
- * Note that this routine only supports wildcard characters in the
- * basename portion of the pattern, not in the directory portion.
- * Note also that the order of filenames returned is deterministic.
- */
-@Operator(group = "io")
-public final class MatchingFiles extends RawOp implements Operand<TString> {
-  
-  /**
-   * Factory method to create a class wrapping a new MatchingFiles operation.
-   * 
-   * @param scope current scope
-   * @param pattern Shell wildcard pattern(s). Scalar or vector of type string.
-   * @return a new instance of MatchingFiles
-   */
-  @Endpoint(describeByClass = true)
-  public static MatchingFiles create(Scope scope, Operand<TString> pattern) {
-    OperationBuilder opBuilder = scope.env().opBuilder("MatchingFiles", scope.makeOpName("MatchingFiles"));
-    opBuilder.addInput(pattern.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new MatchingFiles(opBuilder.build());
-  }
-  
-  /**
-   * A vector of matching filenames.
-   */
-  public Output<TString> filenames() {
-    return filenames;
-  }
-  
-  @Override
-  public Output<TString> asOutput(Scope scope) {
-    return filenames;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "MatchingFiles";
-  
-  private Output<TString> filenames;
-  
-  private MatchingFiles(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    filenames = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/PaddingFifoQueue.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/PaddingFifoQueue.java
deleted file mode 100644
index ef993fa907d..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/PaddingFifoQueue.java
+++ /dev/null
@@ -1,199 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.io;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * A queue that produces elements in first-in first-out order.
- * <p>
- * Variable-size shapes are allowed by setting the corresponding shape dimensions
- * to 0 in the shape attr.  In this case DequeueMany will pad up to the maximum
- * size of any given element in the minibatch.  See below for details.
- */
-@Operator(group = "io")
-public final class PaddingFifoQueue extends RawOp implements Operand<TType> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.io.PaddingFifoQueue}
-   */
-  public static class Options {
-    
-    /**
-     * @param shapes The shape of each component in a value. The length of this attr must
-     * be either 0 or the same as the length of component_types.
-     * Shapes of fixed rank but variable size are allowed by setting
-     * any shape dimension to -1.  In this case, the inputs' shape may vary along
-     * the given dimension, and DequeueMany will pad the given dimension with
-     * zeros up to the maximum shape of all elements in the given batch.
-     * If the length of this attr is 0, different queue elements may have
-     * different ranks and shapes, but only one element may be dequeued at a time.
-     */
-    public Options shapes(List<Shape> shapes) {
-      this.shapes = shapes;
-      return this;
-    }
-    
-    /**
-     * @param capacity The upper bound on the number of elements in this queue.
-     * Negative numbers mean no limit.
-     */
-    public Options capacity(Long capacity) {
-      this.capacity = capacity;
-      return this;
-    }
-    
-    /**
-     * @param container If non-empty, this queue is placed in the given container.
-     * Otherwise, a default container is used.
-     */
-    public Options container(String container) {
-      this.container = container;
-      return this;
-    }
-    
-    /**
-     * @param sharedName If non-empty, this queue will be shared under the given name
-     * across multiple sessions.
-     */
-    public Options sharedName(String sharedName) {
-      this.sharedName = sharedName;
-      return this;
-    }
-    
-    private List<Shape> shapes;
-    private Long capacity;
-    private String container;
-    private String sharedName;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new PaddingFifoQueue operation.
-   * 
-   * @param scope current scope
-   * @param componentTypes The type of each component in a value.
-   * @param options carries optional attributes values
-   * @return a new instance of PaddingFifoQueue
-   */
-  @Endpoint(describeByClass = true)
-  public static PaddingFifoQueue create(Scope scope, List<Class<?>> componentTypes, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("PaddingFIFOQueueV2", scope.makeOpName("PaddingFifoQueue"));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    Class[] componentTypesArray = new Class[componentTypes.size()];
-    for (int i = 0; i < componentTypesArray.length; ++i) {
-      componentTypesArray[i] = componentTypes.get(i);
-    }
-    opBuilder.setAttr("component_types", componentTypesArray);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.shapes != null) {
-          Shape[] shapesArray = new Shape[opts.shapes.size()];
-          for (int i = 0; i < shapesArray.length; ++i) {
-            shapesArray[i] = opts.shapes.get(i);
-          }
-          opBuilder.setAttr("shapes", shapesArray);
-        }
-        if (opts.capacity != null) {
-          opBuilder.setAttr("capacity", opts.capacity);
-        }
-        if (opts.container != null) {
-          opBuilder.setAttr("container", opts.container);
-        }
-        if (opts.sharedName != null) {
-          opBuilder.setAttr("shared_name", opts.sharedName);
-        }
-      }
-    }
-    return new PaddingFifoQueue(opBuilder.build());
-  }
-  
-  /**
-   * @param shapes The shape of each component in a value. The length of this attr must
-   * be either 0 or the same as the length of component_types.
-   * Shapes of fixed rank but variable size are allowed by setting
-   * any shape dimension to -1.  In this case, the inputs' shape may vary along
-   * the given dimension, and DequeueMany will pad the given dimension with
-   * zeros up to the maximum shape of all elements in the given batch.
-   * If the length of this attr is 0, different queue elements may have
-   * different ranks and shapes, but only one element may be dequeued at a time.
-   */
-  public static Options shapes(List<Shape> shapes) {
-    return new Options().shapes(shapes);
-  }
-  
-  /**
-   * @param capacity The upper bound on the number of elements in this queue.
-   * Negative numbers mean no limit.
-   */
-  public static Options capacity(Long capacity) {
-    return new Options().capacity(capacity);
-  }
-  
-  /**
-   * @param container If non-empty, this queue is placed in the given container.
-   * Otherwise, a default container is used.
-   */
-  public static Options container(String container) {
-    return new Options().container(container);
-  }
-  
-  /**
-   * @param sharedName If non-empty, this queue will be shared under the given name
-   * across multiple sessions.
-   */
-  public static Options sharedName(String sharedName) {
-    return new Options().sharedName(sharedName);
-  }
-  
-  /**
-   * The handle to the queue.
-   */
-  public Output<?> handle() {
-    return handle;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) handle;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "PaddingFIFOQueueV2";
-  
-  private Output<?> handle;
-  
-  private PaddingFifoQueue(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    handle = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ParseExample.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ParseExample.java
deleted file mode 100644
index dddc8502a2b..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ParseExample.java
+++ /dev/null
@@ -1,200 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.io;
-
-import java.util.Arrays;
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.op.Operands;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.TString;
-
-/**
- * Transforms a vector of tf.Example protos (as strings) into typed tensors.
- */
-@Operator(group = "io")
-public final class ParseExample extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new ParseExample operation.
-   * 
-   * @param scope current scope
-   * @param serialized A scalar or vector containing binary serialized Example protos.
-   * @param names A tensor containing the names of the serialized protos.
-   * Corresponds 1:1 with the `serialized` tensor.
-   * May contain, for example, table key (descriptive) names for the
-   * corresponding serialized protos.  These are purely useful for debugging
-   * purposes, and the presence of values here has no effect on the output.
-   * May also be an empty vector if no names are available.
-   * If non-empty, this tensor must have the same shape as "serialized".
-   * @param sparseKeys Vector of strings.
-   * The keys expected in the Examples' features associated with sparse values.
-   * @param denseKeys Vector of strings.
-   * The keys expected in the Examples' features associated with dense values.
-   * @param raggedKeys Vector of strings.
-   * The keys expected in the Examples' features associated with ragged values.
-   * @param denseDefaults A list of Tensors (some may be empty).  Corresponds 1:1 with `dense_keys`.
-   * dense_defaults[j] provides default values
-   * when the example's feature_map lacks dense_key[j].  If an empty Tensor is
-   * provided for dense_defaults[j], then the Feature dense_keys[j] is required.
-   * The input type is inferred from dense_defaults[j], even when it's empty.
-   * If dense_defaults[j] is not empty, and dense_shapes[j] is fully defined,
-   * then the shape of dense_defaults[j] must match that of dense_shapes[j].
-   * If dense_shapes[j] has an undefined major dimension (variable strides dense
-   * feature), dense_defaults[j] must contain a single element:
-   * the padding element.
-   * @param numSparse The number of sparse keys.
-   * @param sparseTypes A list of `num_sparse` types; the data types of data in each Feature
-   * given in sparse_keys.
-   * Currently the ParseExample supports DT_FLOAT (FloatList),
-   * DT_INT64 (Int64List), and DT_STRING (BytesList).
-   * @param raggedValueTypes A list of `num_ragged` types; the data types of data in each Feature
-   * given in ragged_keys (where `num_ragged = sparse_keys.size()`).
-   * Currently the ParseExample supports DT_FLOAT (FloatList),
-   * DT_INT64 (Int64List), and DT_STRING (BytesList).
-   * @param raggedSplitTypes A list of `num_ragged` types; the data types of row_splits in each Feature
-   * given in ragged_keys (where `num_ragged = sparse_keys.size()`).
-   * May be DT_INT32 or DT_INT64.
-   * @param denseShapes A list of `num_dense` shapes; the shapes of data in each Feature
-   * given in dense_keys (where `num_dense = dense_keys.size()`).
-   * The number of elements in the Feature corresponding to dense_key[j]
-   * must always equal dense_shapes[j].NumEntries().
-   * If dense_shapes[j] == (D0, D1, ..., DN) then the shape of output
-   * Tensor dense_values[j] will be (|serialized|, D0, D1, ..., DN):
-   * The dense outputs are just the inputs row-stacked by batch.
-   * This works for dense_shapes[j] = (-1, D1, ..., DN).  In this case
-   * the shape of the output Tensor dense_values[j] will be
-   * (|serialized|, M, D1, .., DN), where M is the maximum number of blocks
-   * of elements of length D1 * .... * DN, across all minibatch entries
-   * in the input.  Any minibatch entry with less than M blocks of elements of
-   * length D1 * ... * DN will be padded with the corresponding default_value
-   * scalar element along the second dimension.
-   * @return a new instance of ParseExample
-   */
-  @Endpoint(describeByClass = true)
-  public static ParseExample create(Scope scope, Operand<TString> serialized, Operand<TString> names, Operand<TString> sparseKeys, Operand<TString> denseKeys, Operand<TString> raggedKeys, Iterable<Operand<?>> denseDefaults, Long numSparse, List<Class<?>> sparseTypes, List<Class<?>> raggedValueTypes, List<Class<?>> raggedSplitTypes, List<Shape> denseShapes) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ParseExampleV2", scope.makeOpName("ParseExample"));
-    opBuilder.addInput(serialized.asOutput(scope));
-    opBuilder.addInput(names.asOutput(scope));
-    opBuilder.addInput(sparseKeys.asOutput(scope));
-    opBuilder.addInput(denseKeys.asOutput(scope));
-    opBuilder.addInput(raggedKeys.asOutput(scope));
-    opBuilder.addInputList(Operands.asOutputs(scope, denseDefaults));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("num_sparse", numSparse);
-    Class[] sparseTypesArray = new Class[sparseTypes.size()];
-    for (int i = 0; i < sparseTypesArray.length; ++i) {
-      sparseTypesArray[i] = sparseTypes.get(i);
-    }
-    opBuilder.setAttr("sparse_types", sparseTypesArray);
-    Class[] raggedValueTypesArray = new Class[raggedValueTypes.size()];
-    for (int i = 0; i < raggedValueTypesArray.length; ++i) {
-      raggedValueTypesArray[i] = raggedValueTypes.get(i);
-    }
-    opBuilder.setAttr("ragged_value_types", raggedValueTypesArray);
-    Class[] raggedSplitTypesArray = new Class[raggedSplitTypes.size()];
-    for (int i = 0; i < raggedSplitTypesArray.length; ++i) {
-      raggedSplitTypesArray[i] = raggedSplitTypes.get(i);
-    }
-    opBuilder.setAttr("ragged_split_types", raggedSplitTypesArray);
-    Shape[] denseShapesArray = new Shape[denseShapes.size()];
-    for (int i = 0; i < denseShapesArray.length; ++i) {
-      denseShapesArray[i] = denseShapes.get(i);
-    }
-    opBuilder.setAttr("dense_shapes", denseShapesArray);
-    return new ParseExample(opBuilder.build());
-  }
-  
-  /**
-   */
-  public List<Output<TInt64>> sparseIndices() {
-    return sparseIndices;
-  }
-  
-  /**
-   */
-  public List<Output<?>> sparseValues() {
-    return sparseValues;
-  }
-  
-  /**
-   */
-  public List<Output<TInt64>> sparseShapes() {
-    return sparseShapes;
-  }
-  
-  /**
-   */
-  public List<Output<?>> denseValues() {
-    return denseValues;
-  }
-  
-  /**
-   */
-  public List<Output<?>> raggedValues() {
-    return raggedValues;
-  }
-  
-  /**
-   */
-  public List<Output<?>> raggedRowSplits() {
-    return raggedRowSplits;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ParseExampleV2";
-  
-  private List<Output<TInt64>> sparseIndices;
-  private List<Output<?>> sparseValues;
-  private List<Output<TInt64>> sparseShapes;
-  private List<Output<?>> denseValues;
-  private List<Output<?>> raggedValues;
-  private List<Output<?>> raggedRowSplits;
-  
-  @SuppressWarnings("unchecked")
-  private ParseExample(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    int sparseIndicesLength = operation.outputListLength("sparse_indices");
-    sparseIndices = Arrays.asList((Output<TInt64>[])operation.outputList(outputIdx, sparseIndicesLength));
-    outputIdx += sparseIndicesLength;
-    int sparseValuesLength = operation.outputListLength("sparse_values");
-    sparseValues = Arrays.asList(operation.outputList(outputIdx, sparseValuesLength));
-    outputIdx += sparseValuesLength;
-    int sparseShapesLength = operation.outputListLength("sparse_shapes");
-    sparseShapes = Arrays.asList((Output<TInt64>[])operation.outputList(outputIdx, sparseShapesLength));
-    outputIdx += sparseShapesLength;
-    int denseValuesLength = operation.outputListLength("dense_values");
-    denseValues = Arrays.asList(operation.outputList(outputIdx, denseValuesLength));
-    outputIdx += denseValuesLength;
-    int raggedValuesLength = operation.outputListLength("ragged_values");
-    raggedValues = Arrays.asList(operation.outputList(outputIdx, raggedValuesLength));
-    outputIdx += raggedValuesLength;
-    int raggedRowSplitsLength = operation.outputListLength("ragged_row_splits");
-    raggedRowSplits = Arrays.asList(operation.outputList(outputIdx, raggedRowSplitsLength));
-    outputIdx += raggedRowSplitsLength;
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ParseSequenceExample.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ParseSequenceExample.java
deleted file mode 100644
index fc3d57c9e49..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ParseSequenceExample.java
+++ /dev/null
@@ -1,424 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.io;
-
-import java.util.Arrays;
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.op.Operands;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TBool;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.TString;
-
-/**
- * Transforms a vector of tf.io.SequenceExample protos (as strings) into
- * typed tensors.
- */
-@Operator(group = "io")
-public final class ParseSequenceExample extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.io.ParseSequenceExample}
-   */
-  public static class Options {
-    
-    /**
-     * @param NcontextSparse 
-     */
-    public Options NcontextSparse(Long NcontextSparse) {
-      this.NcontextSparse = NcontextSparse;
-      return this;
-    }
-    
-    /**
-     * @param contextDenseShapes A list of Ncontext_dense shapes; the shapes of data in
-     * each context Feature given in context_dense_keys.
-     * The number of elements in the Feature corresponding to context_dense_key[j]
-     * must always equal context_dense_shapes[j].NumEntries().
-     * The shape of context_dense_values[j] will match context_dense_shapes[j].
-     */
-    public Options contextDenseShapes(List<Shape> contextDenseShapes) {
-      this.contextDenseShapes = contextDenseShapes;
-      return this;
-    }
-    
-    /**
-     * @param NfeatureListSparse 
-     */
-    public Options NfeatureListSparse(Long NfeatureListSparse) {
-      this.NfeatureListSparse = NfeatureListSparse;
-      return this;
-    }
-    
-    /**
-     * @param NfeatureListDense 
-     */
-    public Options NfeatureListDense(Long NfeatureListDense) {
-      this.NfeatureListDense = NfeatureListDense;
-      return this;
-    }
-    
-    /**
-     * @param featureListDenseShapes A list of Nfeature_list_dense shapes; the shapes of
-     * data in each FeatureList given in feature_list_dense_keys.
-     * The shape of each Feature in the FeatureList corresponding to
-     * feature_list_dense_key[j] must always equal
-     * feature_list_dense_shapes[j].NumEntries().
-     */
-    public Options featureListDenseShapes(List<Shape> featureListDenseShapes) {
-      this.featureListDenseShapes = featureListDenseShapes;
-      return this;
-    }
-    
-    private Long NcontextSparse;
-    private List<Shape> contextDenseShapes;
-    private Long NfeatureListSparse;
-    private Long NfeatureListDense;
-    private List<Shape> featureListDenseShapes;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new ParseSequenceExample operation.
-   * 
-   * @param scope current scope
-   * @param serialized A scalar or vector containing binary serialized SequenceExample protos.
-   * @param debugName A scalar or vector containing the names of the serialized protos.
-   * May contain, for example, table key (descriptive) name for the
-   * corresponding serialized proto.  This is purely useful for debugging
-   * purposes, and the presence of values here has no effect on the output.
-   * May also be an empty vector if no name is available.
-   * @param contextSparseKeys The keys expected in the Examples' features associated with context_sparse
-   * values.
-   * @param contextDenseKeys The keys expected in the SequenceExamples' context features associated with
-   * dense values.
-   * @param contextRaggedKeys The keys expected in the Examples' features associated with context_ragged
-   * values.
-   * @param featureListSparseKeys The keys expected in the FeatureLists associated with sparse values.
-   * @param featureListDenseKeys The keys expected in the SequenceExamples' feature_lists associated
-   * with lists of dense values.
-   * @param featureListRaggedKeys The keys expected in the FeatureLists associated with ragged values.
-   * @param featureListDenseMissingAssumedEmpty A vector corresponding 1:1 with feature_list_dense_keys, indicating which
-   * features may be missing from the SequenceExamples.  If the associated
-   * FeatureList is missing, it is treated as empty.
-   * @param contextDenseDefaults A list of Ncontext_dense Tensors (some may be empty).
-   * context_dense_defaults[j] provides default values
-   * when the SequenceExample's context map lacks context_dense_key[j].
-   * If an empty Tensor is provided for context_dense_defaults[j],
-   * then the Feature context_dense_keys[j] is required.
-   * The input type is inferred from context_dense_defaults[j], even when it's
-   * empty.  If context_dense_defaults[j] is not empty, its shape must match
-   * context_dense_shapes[j].
-   * @param contextSparseTypes A list of Ncontext_sparse types; the data types of data in
-   * each context Feature given in context_sparse_keys.
-   * Currently the ParseSingleSequenceExample supports DT_FLOAT (FloatList),
-   * DT_INT64 (Int64List), and DT_STRING (BytesList).
-   * @param contextRaggedValueTypes RaggedTensor.value dtypes for the ragged context features.
-   * @param contextRaggedSplitTypes RaggedTensor.row_split dtypes for the ragged context features.
-   * @param featureListDenseTypes 
-   * @param featureListSparseTypes A list of Nfeature_list_sparse types; the data types
-   * of data in each FeatureList given in feature_list_sparse_keys.
-   * Currently the ParseSingleSequenceExample supports DT_FLOAT (FloatList),
-   * DT_INT64 (Int64List), and DT_STRING (BytesList).
-   * @param featureListRaggedValueTypes RaggedTensor.value dtypes for the ragged FeatureList features.
-   * @param featureListRaggedSplitTypes RaggedTensor.row_split dtypes for the ragged FeatureList features.
-   * @param options carries optional attributes values
-   * @return a new instance of ParseSequenceExample
-   */
-  @Endpoint(describeByClass = true)
-  public static ParseSequenceExample create(Scope scope, Operand<TString> serialized, Operand<TString> debugName, Operand<TString> contextSparseKeys, Operand<TString> contextDenseKeys, Operand<TString> contextRaggedKeys, Operand<TString> featureListSparseKeys, Operand<TString> featureListDenseKeys, Operand<TString> featureListRaggedKeys, Operand<TBool> featureListDenseMissingAssumedEmpty, Iterable<Operand<?>> contextDenseDefaults, List<Class<?>> contextSparseTypes, List<Class<?>> contextRaggedValueTypes, List<Class<?>> contextRaggedSplitTypes, List<Class<?>> featureListDenseTypes, List<Class<?>> featureListSparseTypes, List<Class<?>> featureListRaggedValueTypes, List<Class<?>> featureListRaggedSplitTypes, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ParseSequenceExampleV2", scope.makeOpName("ParseSequenceExample"));
-    opBuilder.addInput(serialized.asOutput(scope));
-    opBuilder.addInput(debugName.asOutput(scope));
-    opBuilder.addInput(contextSparseKeys.asOutput(scope));
-    opBuilder.addInput(contextDenseKeys.asOutput(scope));
-    opBuilder.addInput(contextRaggedKeys.asOutput(scope));
-    opBuilder.addInput(featureListSparseKeys.asOutput(scope));
-    opBuilder.addInput(featureListDenseKeys.asOutput(scope));
-    opBuilder.addInput(featureListRaggedKeys.asOutput(scope));
-    opBuilder.addInput(featureListDenseMissingAssumedEmpty.asOutput(scope));
-    opBuilder.addInputList(Operands.asOutputs(scope, contextDenseDefaults));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    Class[] contextSparseTypesArray = new Class[contextSparseTypes.size()];
-    for (int i = 0; i < contextSparseTypesArray.length; ++i) {
-      contextSparseTypesArray[i] = contextSparseTypes.get(i);
-    }
-    opBuilder.setAttr("context_sparse_types", contextSparseTypesArray);
-    Class[] contextRaggedValueTypesArray = new Class[contextRaggedValueTypes.size()];
-    for (int i = 0; i < contextRaggedValueTypesArray.length; ++i) {
-      contextRaggedValueTypesArray[i] = contextRaggedValueTypes.get(i);
-    }
-    opBuilder.setAttr("context_ragged_value_types", contextRaggedValueTypesArray);
-    Class[] contextRaggedSplitTypesArray = new Class[contextRaggedSplitTypes.size()];
-    for (int i = 0; i < contextRaggedSplitTypesArray.length; ++i) {
-      contextRaggedSplitTypesArray[i] = contextRaggedSplitTypes.get(i);
-    }
-    opBuilder.setAttr("context_ragged_split_types", contextRaggedSplitTypesArray);
-    Class[] featureListDenseTypesArray = new Class[featureListDenseTypes.size()];
-    for (int i = 0; i < featureListDenseTypesArray.length; ++i) {
-      featureListDenseTypesArray[i] = featureListDenseTypes.get(i);
-    }
-    opBuilder.setAttr("feature_list_dense_types", featureListDenseTypesArray);
-    Class[] featureListSparseTypesArray = new Class[featureListSparseTypes.size()];
-    for (int i = 0; i < featureListSparseTypesArray.length; ++i) {
-      featureListSparseTypesArray[i] = featureListSparseTypes.get(i);
-    }
-    opBuilder.setAttr("feature_list_sparse_types", featureListSparseTypesArray);
-    Class[] featureListRaggedValueTypesArray = new Class[featureListRaggedValueTypes.size()];
-    for (int i = 0; i < featureListRaggedValueTypesArray.length; ++i) {
-      featureListRaggedValueTypesArray[i] = featureListRaggedValueTypes.get(i);
-    }
-    opBuilder.setAttr("feature_list_ragged_value_types", featureListRaggedValueTypesArray);
-    Class[] featureListRaggedSplitTypesArray = new Class[featureListRaggedSplitTypes.size()];
-    for (int i = 0; i < featureListRaggedSplitTypesArray.length; ++i) {
-      featureListRaggedSplitTypesArray[i] = featureListRaggedSplitTypes.get(i);
-    }
-    opBuilder.setAttr("feature_list_ragged_split_types", featureListRaggedSplitTypesArray);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.NcontextSparse != null) {
-          opBuilder.setAttr("Ncontext_sparse", opts.NcontextSparse);
-        }
-        if (opts.contextDenseShapes != null) {
-          Shape[] contextDenseShapesArray = new Shape[opts.contextDenseShapes.size()];
-          for (int i = 0; i < contextDenseShapesArray.length; ++i) {
-            contextDenseShapesArray[i] = opts.contextDenseShapes.get(i);
-          }
-          opBuilder.setAttr("context_dense_shapes", contextDenseShapesArray);
-        }
-        if (opts.NfeatureListSparse != null) {
-          opBuilder.setAttr("Nfeature_list_sparse", opts.NfeatureListSparse);
-        }
-        if (opts.NfeatureListDense != null) {
-          opBuilder.setAttr("Nfeature_list_dense", opts.NfeatureListDense);
-        }
-        if (opts.featureListDenseShapes != null) {
-          Shape[] featureListDenseShapesArray = new Shape[opts.featureListDenseShapes.size()];
-          for (int i = 0; i < featureListDenseShapesArray.length; ++i) {
-            featureListDenseShapesArray[i] = opts.featureListDenseShapes.get(i);
-          }
-          opBuilder.setAttr("feature_list_dense_shapes", featureListDenseShapesArray);
-        }
-      }
-    }
-    return new ParseSequenceExample(opBuilder.build());
-  }
-  
-  /**
-   * @param NcontextSparse 
-   */
-  public static Options NcontextSparse(Long NcontextSparse) {
-    return new Options().NcontextSparse(NcontextSparse);
-  }
-  
-  /**
-   * @param contextDenseShapes A list of Ncontext_dense shapes; the shapes of data in
-   * each context Feature given in context_dense_keys.
-   * The number of elements in the Feature corresponding to context_dense_key[j]
-   * must always equal context_dense_shapes[j].NumEntries().
-   * The shape of context_dense_values[j] will match context_dense_shapes[j].
-   */
-  public static Options contextDenseShapes(List<Shape> contextDenseShapes) {
-    return new Options().contextDenseShapes(contextDenseShapes);
-  }
-  
-  /**
-   * @param NfeatureListSparse 
-   */
-  public static Options NfeatureListSparse(Long NfeatureListSparse) {
-    return new Options().NfeatureListSparse(NfeatureListSparse);
-  }
-  
-  /**
-   * @param NfeatureListDense 
-   */
-  public static Options NfeatureListDense(Long NfeatureListDense) {
-    return new Options().NfeatureListDense(NfeatureListDense);
-  }
-  
-  /**
-   * @param featureListDenseShapes A list of Nfeature_list_dense shapes; the shapes of
-   * data in each FeatureList given in feature_list_dense_keys.
-   * The shape of each Feature in the FeatureList corresponding to
-   * feature_list_dense_key[j] must always equal
-   * feature_list_dense_shapes[j].NumEntries().
-   */
-  public static Options featureListDenseShapes(List<Shape> featureListDenseShapes) {
-    return new Options().featureListDenseShapes(featureListDenseShapes);
-  }
-  
-  /**
-   */
-  public List<Output<TInt64>> contextSparseIndices() {
-    return contextSparseIndices;
-  }
-  
-  /**
-   */
-  public List<Output<?>> contextSparseValues() {
-    return contextSparseValues;
-  }
-  
-  /**
-   */
-  public List<Output<TInt64>> contextSparseShapes() {
-    return contextSparseShapes;
-  }
-  
-  /**
-   */
-  public List<Output<?>> contextDenseValues() {
-    return contextDenseValues;
-  }
-  
-  /**
-   */
-  public List<Output<?>> contextRaggedValues() {
-    return contextRaggedValues;
-  }
-  
-  /**
-   */
-  public List<Output<?>> contextRaggedRowSplits() {
-    return contextRaggedRowSplits;
-  }
-  
-  /**
-   */
-  public List<Output<TInt64>> featureListSparseIndices() {
-    return featureListSparseIndices;
-  }
-  
-  /**
-   */
-  public List<Output<?>> featureListSparseValues() {
-    return featureListSparseValues;
-  }
-  
-  /**
-   */
-  public List<Output<TInt64>> featureListSparseShapes() {
-    return featureListSparseShapes;
-  }
-  
-  /**
-   */
-  public List<Output<?>> featureListDenseValues() {
-    return featureListDenseValues;
-  }
-  
-  /**
-   */
-  public List<Output<TInt64>> featureListDenseLengths() {
-    return featureListDenseLengths;
-  }
-  
-  /**
-   */
-  public List<Output<?>> featureListRaggedValues() {
-    return featureListRaggedValues;
-  }
-  
-  /**
-   */
-  public List<Output<?>> featureListRaggedOuterSplits() {
-    return featureListRaggedOuterSplits;
-  }
-  
-  /**
-   */
-  public List<Output<?>> featureListRaggedInnerSplits() {
-    return featureListRaggedInnerSplits;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ParseSequenceExampleV2";
-  
-  private List<Output<TInt64>> contextSparseIndices;
-  private List<Output<?>> contextSparseValues;
-  private List<Output<TInt64>> contextSparseShapes;
-  private List<Output<?>> contextDenseValues;
-  private List<Output<?>> contextRaggedValues;
-  private List<Output<?>> contextRaggedRowSplits;
-  private List<Output<TInt64>> featureListSparseIndices;
-  private List<Output<?>> featureListSparseValues;
-  private List<Output<TInt64>> featureListSparseShapes;
-  private List<Output<?>> featureListDenseValues;
-  private List<Output<TInt64>> featureListDenseLengths;
-  private List<Output<?>> featureListRaggedValues;
-  private List<Output<?>> featureListRaggedOuterSplits;
-  private List<Output<?>> featureListRaggedInnerSplits;
-  
-  @SuppressWarnings("unchecked")
-  private ParseSequenceExample(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    int contextSparseIndicesLength = operation.outputListLength("context_sparse_indices");
-    contextSparseIndices = Arrays.asList((Output<TInt64>[])operation.outputList(outputIdx, contextSparseIndicesLength));
-    outputIdx += contextSparseIndicesLength;
-    int contextSparseValuesLength = operation.outputListLength("context_sparse_values");
-    contextSparseValues = Arrays.asList(operation.outputList(outputIdx, contextSparseValuesLength));
-    outputIdx += contextSparseValuesLength;
-    int contextSparseShapesLength = operation.outputListLength("context_sparse_shapes");
-    contextSparseShapes = Arrays.asList((Output<TInt64>[])operation.outputList(outputIdx, contextSparseShapesLength));
-    outputIdx += contextSparseShapesLength;
-    int contextDenseValuesLength = operation.outputListLength("context_dense_values");
-    contextDenseValues = Arrays.asList(operation.outputList(outputIdx, contextDenseValuesLength));
-    outputIdx += contextDenseValuesLength;
-    int contextRaggedValuesLength = operation.outputListLength("context_ragged_values");
-    contextRaggedValues = Arrays.asList(operation.outputList(outputIdx, contextRaggedValuesLength));
-    outputIdx += contextRaggedValuesLength;
-    int contextRaggedRowSplitsLength = operation.outputListLength("context_ragged_row_splits");
-    contextRaggedRowSplits = Arrays.asList(operation.outputList(outputIdx, contextRaggedRowSplitsLength));
-    outputIdx += contextRaggedRowSplitsLength;
-    int featureListSparseIndicesLength = operation.outputListLength("feature_list_sparse_indices");
-    featureListSparseIndices = Arrays.asList((Output<TInt64>[])operation.outputList(outputIdx, featureListSparseIndicesLength));
-    outputIdx += featureListSparseIndicesLength;
-    int featureListSparseValuesLength = operation.outputListLength("feature_list_sparse_values");
-    featureListSparseValues = Arrays.asList(operation.outputList(outputIdx, featureListSparseValuesLength));
-    outputIdx += featureListSparseValuesLength;
-    int featureListSparseShapesLength = operation.outputListLength("feature_list_sparse_shapes");
-    featureListSparseShapes = Arrays.asList((Output<TInt64>[])operation.outputList(outputIdx, featureListSparseShapesLength));
-    outputIdx += featureListSparseShapesLength;
-    int featureListDenseValuesLength = operation.outputListLength("feature_list_dense_values");
-    featureListDenseValues = Arrays.asList(operation.outputList(outputIdx, featureListDenseValuesLength));
-    outputIdx += featureListDenseValuesLength;
-    int featureListDenseLengthsLength = operation.outputListLength("feature_list_dense_lengths");
-    featureListDenseLengths = Arrays.asList((Output<TInt64>[])operation.outputList(outputIdx, featureListDenseLengthsLength));
-    outputIdx += featureListDenseLengthsLength;
-    int featureListRaggedValuesLength = operation.outputListLength("feature_list_ragged_values");
-    featureListRaggedValues = Arrays.asList(operation.outputList(outputIdx, featureListRaggedValuesLength));
-    outputIdx += featureListRaggedValuesLength;
-    int featureListRaggedOuterSplitsLength = operation.outputListLength("feature_list_ragged_outer_splits");
-    featureListRaggedOuterSplits = Arrays.asList(operation.outputList(outputIdx, featureListRaggedOuterSplitsLength));
-    outputIdx += featureListRaggedOuterSplitsLength;
-    int featureListRaggedInnerSplitsLength = operation.outputListLength("feature_list_ragged_inner_splits");
-    featureListRaggedInnerSplits = Arrays.asList(operation.outputList(outputIdx, featureListRaggedInnerSplitsLength));
-    outputIdx += featureListRaggedInnerSplitsLength;
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ParseSingleExample.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ParseSingleExample.java
deleted file mode 100644
index a51abe0e89e..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ParseSingleExample.java
+++ /dev/null
@@ -1,156 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.io;
-
-import java.util.Arrays;
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.op.Operands;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.TString;
-
-/**
- * Transforms a tf.Example proto (as a string) into typed tensors.
- */
-@Operator(group = "io")
-public final class ParseSingleExample extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new ParseSingleExample operation.
-   * 
-   * @param scope current scope
-   * @param serialized A vector containing a batch of binary serialized Example protos.
-   * @param denseDefaults A list of Tensors (some may be empty), whose length matches
-   * the length of `dense_keys`. dense_defaults[j] provides default values
-   * when the example's feature_map lacks dense_key[j].  If an empty Tensor is
-   * provided for dense_defaults[j], then the Feature dense_keys[j] is required.
-   * The input type is inferred from dense_defaults[j], even when it's empty.
-   * If dense_defaults[j] is not empty, and dense_shapes[j] is fully defined,
-   * then the shape of dense_defaults[j] must match that of dense_shapes[j].
-   * If dense_shapes[j] has an undefined major dimension (variable strides dense
-   * feature), dense_defaults[j] must contain a single element:
-   * the padding element.
-   * @param numSparse The number of sparse features to be parsed from the example. This
-   * must match the lengths of `sparse_keys` and `sparse_types`.
-   * @param sparseKeys A list of `num_sparse` strings.
-   * The keys expected in the Examples' features associated with sparse values.
-   * @param denseKeys The keys expected in the Examples' features associated with dense
-   * values.
-   * @param sparseTypes A list of `num_sparse` types; the data types of data in each
-   * Feature given in sparse_keys.
-   * Currently the ParseSingleExample op supports DT_FLOAT (FloatList),
-   * DT_INT64 (Int64List), and DT_STRING (BytesList).
-   * @param denseShapes The shapes of data in each Feature given in dense_keys.
-   * The length of this list must match the length of `dense_keys`.  The
-   * number of elements in the Feature corresponding to dense_key[j] must
-   * always equal dense_shapes[j].NumEntries().  If dense_shapes[j] ==
-   * (D0, D1, ..., DN) then the shape of output Tensor dense_values[j]
-   * will be (D0, D1, ..., DN): In the case dense_shapes[j] = (-1, D1,
-   * ..., DN), the shape of the output Tensor dense_values[j] will be (M,
-   * D1, .., DN), where M is the number of blocks of elements of length
-   * D1 * .... * DN, in the input.
-   * @return a new instance of ParseSingleExample
-   */
-  @Endpoint(describeByClass = true)
-  public static ParseSingleExample create(Scope scope, Operand<TString> serialized, Iterable<Operand<?>> denseDefaults, Long numSparse, List<String> sparseKeys, List<String> denseKeys, List<Class<?>> sparseTypes, List<Shape> denseShapes) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ParseSingleExample", scope.makeOpName("ParseSingleExample"));
-    opBuilder.addInput(serialized.asOutput(scope));
-    opBuilder.addInputList(Operands.asOutputs(scope, denseDefaults));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("num_sparse", numSparse);
-    String[] sparseKeysArray = new String[sparseKeys.size()];
-    for (int i = 0; i < sparseKeysArray.length; ++i) {
-      sparseKeysArray[i] = sparseKeys.get(i);
-    }
-    opBuilder.setAttr("sparse_keys", sparseKeysArray);
-    String[] denseKeysArray = new String[denseKeys.size()];
-    for (int i = 0; i < denseKeysArray.length; ++i) {
-      denseKeysArray[i] = denseKeys.get(i);
-    }
-    opBuilder.setAttr("dense_keys", denseKeysArray);
-    Class[] sparseTypesArray = new Class[sparseTypes.size()];
-    for (int i = 0; i < sparseTypesArray.length; ++i) {
-      sparseTypesArray[i] = sparseTypes.get(i);
-    }
-    opBuilder.setAttr("sparse_types", sparseTypesArray);
-    Shape[] denseShapesArray = new Shape[denseShapes.size()];
-    for (int i = 0; i < denseShapesArray.length; ++i) {
-      denseShapesArray[i] = denseShapes.get(i);
-    }
-    opBuilder.setAttr("dense_shapes", denseShapesArray);
-    return new ParseSingleExample(opBuilder.build());
-  }
-  
-  /**
-   */
-  public List<Output<TInt64>> sparseIndices() {
-    return sparseIndices;
-  }
-  
-  /**
-   */
-  public List<Output<?>> sparseValues() {
-    return sparseValues;
-  }
-  
-  /**
-   */
-  public List<Output<TInt64>> sparseShapes() {
-    return sparseShapes;
-  }
-  
-  /**
-   */
-  public List<Output<?>> denseValues() {
-    return denseValues;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ParseSingleExample";
-  
-  private List<Output<TInt64>> sparseIndices;
-  private List<Output<?>> sparseValues;
-  private List<Output<TInt64>> sparseShapes;
-  private List<Output<?>> denseValues;
-  
-  @SuppressWarnings("unchecked")
-  private ParseSingleExample(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    int sparseIndicesLength = operation.outputListLength("sparse_indices");
-    sparseIndices = Arrays.asList((Output<TInt64>[])operation.outputList(outputIdx, sparseIndicesLength));
-    outputIdx += sparseIndicesLength;
-    int sparseValuesLength = operation.outputListLength("sparse_values");
-    sparseValues = Arrays.asList(operation.outputList(outputIdx, sparseValuesLength));
-    outputIdx += sparseValuesLength;
-    int sparseShapesLength = operation.outputListLength("sparse_shapes");
-    sparseShapes = Arrays.asList((Output<TInt64>[])operation.outputList(outputIdx, sparseShapesLength));
-    outputIdx += sparseShapesLength;
-    int denseValuesLength = operation.outputListLength("dense_values");
-    denseValues = Arrays.asList(operation.outputList(outputIdx, denseValuesLength));
-    outputIdx += denseValuesLength;
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ParseSingleSequenceExample.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ParseSingleSequenceExample.java
deleted file mode 100644
index 55fdbf10265..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ParseSingleSequenceExample.java
+++ /dev/null
@@ -1,282 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.io;
-
-import java.util.Arrays;
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.op.Operands;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.TString;
-
-/**
- * Transforms a scalar brain.SequenceExample proto (as strings) into typed tensors.
- */
-@Operator(group = "io")
-public final class ParseSingleSequenceExample extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.io.ParseSingleSequenceExample}
-   */
-  public static class Options {
-    
-    /**
-     * @param contextDenseShapes A list of Ncontext_dense shapes; the shapes of data in
-     * each context Feature given in context_dense_keys.
-     * The number of elements in the Feature corresponding to context_dense_key[j]
-     * must always equal context_dense_shapes[j].NumEntries().
-     * The shape of context_dense_values[j] will match context_dense_shapes[j].
-     */
-    public Options contextDenseShapes(List<Shape> contextDenseShapes) {
-      this.contextDenseShapes = contextDenseShapes;
-      return this;
-    }
-    
-    /**
-     * @param featureListDenseShapes A list of Nfeature_list_dense shapes; the shapes of
-     * data in each FeatureList given in feature_list_dense_keys.
-     * The shape of each Feature in the FeatureList corresponding to
-     * feature_list_dense_key[j] must always equal
-     * feature_list_dense_shapes[j].NumEntries().
-     */
-    public Options featureListDenseShapes(List<Shape> featureListDenseShapes) {
-      this.featureListDenseShapes = featureListDenseShapes;
-      return this;
-    }
-    
-    private List<Shape> contextDenseShapes;
-    private List<Shape> featureListDenseShapes;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new ParseSingleSequenceExample operation.
-   * 
-   * @param scope current scope
-   * @param serialized A scalar containing a binary serialized SequenceExample proto.
-   * @param featureListDenseMissingAssumedEmpty A vector listing the
-   * FeatureList keys which may be missing from the SequenceExample.  If the
-   * associated FeatureList is missing, it is treated as empty.  By default,
-   * any FeatureList not listed in this vector must exist in the SequenceExample.
-   * @param contextSparseKeys A list of Ncontext_sparse string Tensors (scalars).
-   * The keys expected in the Examples' features associated with context_sparse
-   * values.
-   * @param contextDenseKeys A list of Ncontext_dense string Tensors (scalars).
-   * The keys expected in the SequenceExamples' context features associated with
-   * dense values.
-   * @param featureListSparseKeys A list of Nfeature_list_sparse string Tensors
-   * (scalars).  The keys expected in the FeatureLists associated with sparse
-   * values.
-   * @param featureListDenseKeys A list of Nfeature_list_dense string Tensors (scalars).
-   * The keys expected in the SequenceExamples' feature_lists associated
-   * with lists of dense values.
-   * @param contextDenseDefaults A list of Ncontext_dense Tensors (some may be empty).
-   * context_dense_defaults[j] provides default values
-   * when the SequenceExample's context map lacks context_dense_key[j].
-   * If an empty Tensor is provided for context_dense_defaults[j],
-   * then the Feature context_dense_keys[j] is required.
-   * The input type is inferred from context_dense_defaults[j], even when it's
-   * empty.  If context_dense_defaults[j] is not empty, its shape must match
-   * context_dense_shapes[j].
-   * @param debugName A scalar containing the name of the serialized proto.
-   * May contain, for example, table key (descriptive) name for the
-   * corresponding serialized proto.  This is purely useful for debugging
-   * purposes, and the presence of values here has no effect on the output.
-   * May also be an empty scalar if no name is available.
-   * @param contextSparseTypes A list of Ncontext_sparse types; the data types of data in
-   * each context Feature given in context_sparse_keys.
-   * Currently the ParseSingleSequenceExample supports DT_FLOAT (FloatList),
-   * DT_INT64 (Int64List), and DT_STRING (BytesList).
-   * @param featureListDenseTypes 
-   * @param featureListSparseTypes A list of Nfeature_list_sparse types; the data types
-   * of data in each FeatureList given in feature_list_sparse_keys.
-   * Currently the ParseSingleSequenceExample supports DT_FLOAT (FloatList),
-   * DT_INT64 (Int64List), and DT_STRING (BytesList).
-   * @param options carries optional attributes values
-   * @return a new instance of ParseSingleSequenceExample
-   */
-  @Endpoint(describeByClass = true)
-  public static ParseSingleSequenceExample create(Scope scope, Operand<TString> serialized, Operand<TString> featureListDenseMissingAssumedEmpty, Iterable<Operand<TString>> contextSparseKeys, Iterable<Operand<TString>> contextDenseKeys, Iterable<Operand<TString>> featureListSparseKeys, Iterable<Operand<TString>> featureListDenseKeys, Iterable<Operand<?>> contextDenseDefaults, Operand<TString> debugName, List<Class<?>> contextSparseTypes, List<Class<?>> featureListDenseTypes, List<Class<?>> featureListSparseTypes, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ParseSingleSequenceExample", scope.makeOpName("ParseSingleSequenceExample"));
-    opBuilder.addInput(serialized.asOutput(scope));
-    opBuilder.addInput(featureListDenseMissingAssumedEmpty.asOutput(scope));
-    opBuilder.addInputList(Operands.asOutputs(scope, contextSparseKeys));
-    opBuilder.addInputList(Operands.asOutputs(scope, contextDenseKeys));
-    opBuilder.addInputList(Operands.asOutputs(scope, featureListSparseKeys));
-    opBuilder.addInputList(Operands.asOutputs(scope, featureListDenseKeys));
-    opBuilder.addInputList(Operands.asOutputs(scope, contextDenseDefaults));
-    opBuilder.addInput(debugName.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    Class[] contextSparseTypesArray = new Class[contextSparseTypes.size()];
-    for (int i = 0; i < contextSparseTypesArray.length; ++i) {
-      contextSparseTypesArray[i] = contextSparseTypes.get(i);
-    }
-    opBuilder.setAttr("context_sparse_types", contextSparseTypesArray);
-    Class[] featureListDenseTypesArray = new Class[featureListDenseTypes.size()];
-    for (int i = 0; i < featureListDenseTypesArray.length; ++i) {
-      featureListDenseTypesArray[i] = featureListDenseTypes.get(i);
-    }
-    opBuilder.setAttr("feature_list_dense_types", featureListDenseTypesArray);
-    Class[] featureListSparseTypesArray = new Class[featureListSparseTypes.size()];
-    for (int i = 0; i < featureListSparseTypesArray.length; ++i) {
-      featureListSparseTypesArray[i] = featureListSparseTypes.get(i);
-    }
-    opBuilder.setAttr("feature_list_sparse_types", featureListSparseTypesArray);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.contextDenseShapes != null) {
-          Shape[] contextDenseShapesArray = new Shape[opts.contextDenseShapes.size()];
-          for (int i = 0; i < contextDenseShapesArray.length; ++i) {
-            contextDenseShapesArray[i] = opts.contextDenseShapes.get(i);
-          }
-          opBuilder.setAttr("context_dense_shapes", contextDenseShapesArray);
-        }
-        if (opts.featureListDenseShapes != null) {
-          Shape[] featureListDenseShapesArray = new Shape[opts.featureListDenseShapes.size()];
-          for (int i = 0; i < featureListDenseShapesArray.length; ++i) {
-            featureListDenseShapesArray[i] = opts.featureListDenseShapes.get(i);
-          }
-          opBuilder.setAttr("feature_list_dense_shapes", featureListDenseShapesArray);
-        }
-      }
-    }
-    return new ParseSingleSequenceExample(opBuilder.build());
-  }
-  
-  /**
-   * @param contextDenseShapes A list of Ncontext_dense shapes; the shapes of data in
-   * each context Feature given in context_dense_keys.
-   * The number of elements in the Feature corresponding to context_dense_key[j]
-   * must always equal context_dense_shapes[j].NumEntries().
-   * The shape of context_dense_values[j] will match context_dense_shapes[j].
-   */
-  public static Options contextDenseShapes(List<Shape> contextDenseShapes) {
-    return new Options().contextDenseShapes(contextDenseShapes);
-  }
-  
-  /**
-   * @param featureListDenseShapes A list of Nfeature_list_dense shapes; the shapes of
-   * data in each FeatureList given in feature_list_dense_keys.
-   * The shape of each Feature in the FeatureList corresponding to
-   * feature_list_dense_key[j] must always equal
-   * feature_list_dense_shapes[j].NumEntries().
-   */
-  public static Options featureListDenseShapes(List<Shape> featureListDenseShapes) {
-    return new Options().featureListDenseShapes(featureListDenseShapes);
-  }
-  
-  /**
-   */
-  public List<Output<TInt64>> contextSparseIndices() {
-    return contextSparseIndices;
-  }
-  
-  /**
-   */
-  public List<Output<?>> contextSparseValues() {
-    return contextSparseValues;
-  }
-  
-  /**
-   */
-  public List<Output<TInt64>> contextSparseShapes() {
-    return contextSparseShapes;
-  }
-  
-  /**
-   */
-  public List<Output<?>> contextDenseValues() {
-    return contextDenseValues;
-  }
-  
-  /**
-   */
-  public List<Output<TInt64>> featureListSparseIndices() {
-    return featureListSparseIndices;
-  }
-  
-  /**
-   */
-  public List<Output<?>> featureListSparseValues() {
-    return featureListSparseValues;
-  }
-  
-  /**
-   */
-  public List<Output<TInt64>> featureListSparseShapes() {
-    return featureListSparseShapes;
-  }
-  
-  /**
-   */
-  public List<Output<?>> featureListDenseValues() {
-    return featureListDenseValues;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ParseSingleSequenceExample";
-  
-  private List<Output<TInt64>> contextSparseIndices;
-  private List<Output<?>> contextSparseValues;
-  private List<Output<TInt64>> contextSparseShapes;
-  private List<Output<?>> contextDenseValues;
-  private List<Output<TInt64>> featureListSparseIndices;
-  private List<Output<?>> featureListSparseValues;
-  private List<Output<TInt64>> featureListSparseShapes;
-  private List<Output<?>> featureListDenseValues;
-  
-  @SuppressWarnings("unchecked")
-  private ParseSingleSequenceExample(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    int contextSparseIndicesLength = operation.outputListLength("context_sparse_indices");
-    contextSparseIndices = Arrays.asList((Output<TInt64>[])operation.outputList(outputIdx, contextSparseIndicesLength));
-    outputIdx += contextSparseIndicesLength;
-    int contextSparseValuesLength = operation.outputListLength("context_sparse_values");
-    contextSparseValues = Arrays.asList(operation.outputList(outputIdx, contextSparseValuesLength));
-    outputIdx += contextSparseValuesLength;
-    int contextSparseShapesLength = operation.outputListLength("context_sparse_shapes");
-    contextSparseShapes = Arrays.asList((Output<TInt64>[])operation.outputList(outputIdx, contextSparseShapesLength));
-    outputIdx += contextSparseShapesLength;
-    int contextDenseValuesLength = operation.outputListLength("context_dense_values");
-    contextDenseValues = Arrays.asList(operation.outputList(outputIdx, contextDenseValuesLength));
-    outputIdx += contextDenseValuesLength;
-    int featureListSparseIndicesLength = operation.outputListLength("feature_list_sparse_indices");
-    featureListSparseIndices = Arrays.asList((Output<TInt64>[])operation.outputList(outputIdx, featureListSparseIndicesLength));
-    outputIdx += featureListSparseIndicesLength;
-    int featureListSparseValuesLength = operation.outputListLength("feature_list_sparse_values");
-    featureListSparseValues = Arrays.asList(operation.outputList(outputIdx, featureListSparseValuesLength));
-    outputIdx += featureListSparseValuesLength;
-    int featureListSparseShapesLength = operation.outputListLength("feature_list_sparse_shapes");
-    featureListSparseShapes = Arrays.asList((Output<TInt64>[])operation.outputList(outputIdx, featureListSparseShapesLength));
-    outputIdx += featureListSparseShapesLength;
-    int featureListDenseValuesLength = operation.outputListLength("feature_list_dense_values");
-    featureListDenseValues = Arrays.asList(operation.outputList(outputIdx, featureListDenseValuesLength));
-    outputIdx += featureListDenseValuesLength;
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ParseTensor.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ParseTensor.java
deleted file mode 100644
index 2a560908647..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ParseTensor.java
+++ /dev/null
@@ -1,79 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.io;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TString;
-import org.tensorflow.types.family.TType;
-
-/**
- * Transforms a serialized tensorflow.TensorProto proto into a Tensor.
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator(group = "io")
-public final class ParseTensor<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new ParseTensor operation.
-   * 
-   * @param scope current scope
-   * @param serialized A scalar string containing a serialized TensorProto proto.
-   * @param outType The type of the serialized tensor.  The provided type must match the
-   * type of the serialized tensor and no implicit conversion will take place.
-   * @return a new instance of ParseTensor
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> ParseTensor<T> create(Scope scope, Operand<TString> serialized, Class<T> outType) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ParseTensor", scope.makeOpName("ParseTensor"));
-    opBuilder.addInput(serialized.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("out_type", outType);
-    return new ParseTensor<T>(opBuilder.build());
-  }
-  
-  /**
-   * A Tensor of type `out_type`.
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ParseTensor";
-  
-  private Output<T> output;
-  
-  private ParseTensor(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/PriorityQueue.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/PriorityQueue.java
deleted file mode 100644
index 5e2bd63fec8..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/PriorityQueue.java
+++ /dev/null
@@ -1,173 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.io;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * A queue that produces elements sorted by the first component value.
- * <p>
- * Note that the PriorityQueue requires the first component of any element
- * to be a scalar int64, in addition to the other elements declared by
- * component_types.  Therefore calls to Enqueue and EnqueueMany (resp. Dequeue
- * and DequeueMany) on a PriorityQueue will all require (resp. output) one extra
- * entry in their input (resp. output) lists.
- */
-@Operator(group = "io")
-public final class PriorityQueue extends RawOp implements Operand<TType> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.io.PriorityQueue}
-   */
-  public static class Options {
-    
-    /**
-     * @param capacity The upper bound on the number of elements in this queue.
-     * Negative numbers mean no limit.
-     */
-    public Options capacity(Long capacity) {
-      this.capacity = capacity;
-      return this;
-    }
-    
-    /**
-     * @param container If non-empty, this queue is placed in the given container.
-     * Otherwise, a default container is used.
-     */
-    public Options container(String container) {
-      this.container = container;
-      return this;
-    }
-    
-    /**
-     * @param sharedName If non-empty, this queue will be shared under the given name
-     * across multiple sessions.
-     */
-    public Options sharedName(String sharedName) {
-      this.sharedName = sharedName;
-      return this;
-    }
-    
-    private Long capacity;
-    private String container;
-    private String sharedName;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new PriorityQueue operation.
-   * 
-   * @param scope current scope
-   * @param componentTypes The type of each component in a value.
-   * @param shapes The shape of each component in a value. The length of this attr must
-   * be either 0 or the same as the length of component_types. If the length of
-   * this attr is 0, the shapes of queue elements are not constrained, and
-   * only one element may be dequeued at a time.
-   * @param options carries optional attributes values
-   * @return a new instance of PriorityQueue
-   */
-  @Endpoint(describeByClass = true)
-  public static PriorityQueue create(Scope scope, List<Class<?>> componentTypes, List<Shape> shapes, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("PriorityQueueV2", scope.makeOpName("PriorityQueue"));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    Class[] componentTypesArray = new Class[componentTypes.size()];
-    for (int i = 0; i < componentTypesArray.length; ++i) {
-      componentTypesArray[i] = componentTypes.get(i);
-    }
-    opBuilder.setAttr("component_types", componentTypesArray);
-    Shape[] shapesArray = new Shape[shapes.size()];
-    for (int i = 0; i < shapesArray.length; ++i) {
-      shapesArray[i] = shapes.get(i);
-    }
-    opBuilder.setAttr("shapes", shapesArray);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.capacity != null) {
-          opBuilder.setAttr("capacity", opts.capacity);
-        }
-        if (opts.container != null) {
-          opBuilder.setAttr("container", opts.container);
-        }
-        if (opts.sharedName != null) {
-          opBuilder.setAttr("shared_name", opts.sharedName);
-        }
-      }
-    }
-    return new PriorityQueue(opBuilder.build());
-  }
-  
-  /**
-   * @param capacity The upper bound on the number of elements in this queue.
-   * Negative numbers mean no limit.
-   */
-  public static Options capacity(Long capacity) {
-    return new Options().capacity(capacity);
-  }
-  
-  /**
-   * @param container If non-empty, this queue is placed in the given container.
-   * Otherwise, a default container is used.
-   */
-  public static Options container(String container) {
-    return new Options().container(container);
-  }
-  
-  /**
-   * @param sharedName If non-empty, this queue will be shared under the given name
-   * across multiple sessions.
-   */
-  public static Options sharedName(String sharedName) {
-    return new Options().sharedName(sharedName);
-  }
-  
-  /**
-   * The handle to the queue.
-   */
-  public Output<?> handle() {
-    return handle;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) handle;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "PriorityQueueV2";
-  
-  private Output<?> handle;
-  
-  private PriorityQueue(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    handle = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/QueueClose.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/QueueClose.java
deleted file mode 100644
index dcbace2ecd3..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/QueueClose.java
+++ /dev/null
@@ -1,97 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.io;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-
-/**
- * Closes the given queue.
- * <p>
- * This operation signals that no more elements will be enqueued in the
- * given queue. Subsequent Enqueue(Many) operations will fail.
- * Subsequent Dequeue(Many) operations will continue to succeed if
- * sufficient elements remain in the queue. Subsequent Dequeue(Many)
- * operations that would block will fail immediately.
- */
-@Operator(group = "io")
-public final class QueueClose extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.io.QueueClose}
-   */
-  public static class Options {
-    
-    /**
-     * @param cancelPendingEnqueues If true, all pending enqueue requests that are
-     * blocked on the given queue will be canceled.
-     */
-    public Options cancelPendingEnqueues(Boolean cancelPendingEnqueues) {
-      this.cancelPendingEnqueues = cancelPendingEnqueues;
-      return this;
-    }
-    
-    private Boolean cancelPendingEnqueues;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new QueueClose operation.
-   * 
-   * @param scope current scope
-   * @param handle The handle to a queue.
-   * @param options carries optional attributes values
-   * @return a new instance of QueueClose
-   */
-  @Endpoint(describeByClass = true)
-  public static QueueClose create(Scope scope, Operand<?> handle, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("QueueCloseV2", scope.makeOpName("QueueClose"));
-    opBuilder.addInput(handle.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.cancelPendingEnqueues != null) {
-          opBuilder.setAttr("cancel_pending_enqueues", opts.cancelPendingEnqueues);
-        }
-      }
-    }
-    return new QueueClose(opBuilder.build());
-  }
-  
-  /**
-   * @param cancelPendingEnqueues If true, all pending enqueue requests that are
-   * blocked on the given queue will be canceled.
-   */
-  public static Options cancelPendingEnqueues(Boolean cancelPendingEnqueues) {
-    return new Options().cancelPendingEnqueues(cancelPendingEnqueues);
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "QueueCloseV2";
-  
-  private QueueClose(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/QueueDequeue.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/QueueDequeue.java
deleted file mode 100644
index 59ea5a8af61..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/QueueDequeue.java
+++ /dev/null
@@ -1,130 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.io;
-
-import java.util.Arrays;
-import java.util.Iterator;
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Dequeues a tuple of one or more tensors from the given queue.
- * <p>
- * This operation has k outputs, where k is the number of components
- * in the tuples stored in the given queue, and output i is the ith
- * component of the dequeued tuple.
- * <p>
- * N.B. If the queue is empty, this operation will block until an element
- * has been dequeued (or 'timeout_ms' elapses, if specified).
- */
-@Operator(group = "io")
-public final class QueueDequeue extends RawOp implements Iterable<Operand<TType>> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.io.QueueDequeue}
-   */
-  public static class Options {
-    
-    /**
-     * @param timeoutMs If the queue is empty, this operation will block for up to
-     * timeout_ms milliseconds.
-     * Note: This option is not supported yet.
-     */
-    public Options timeoutMs(Long timeoutMs) {
-      this.timeoutMs = timeoutMs;
-      return this;
-    }
-    
-    private Long timeoutMs;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new QueueDequeue operation.
-   * 
-   * @param scope current scope
-   * @param handle The handle to a queue.
-   * @param componentTypes The type of each component in a tuple.
-   * @param options carries optional attributes values
-   * @return a new instance of QueueDequeue
-   */
-  @Endpoint(describeByClass = true)
-  public static QueueDequeue create(Scope scope, Operand<?> handle, List<Class<?>> componentTypes, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("QueueDequeueV2", scope.makeOpName("QueueDequeue"));
-    opBuilder.addInput(handle.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    Class[] componentTypesArray = new Class[componentTypes.size()];
-    for (int i = 0; i < componentTypesArray.length; ++i) {
-      componentTypesArray[i] = componentTypes.get(i);
-    }
-    opBuilder.setAttr("component_types", componentTypesArray);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.timeoutMs != null) {
-          opBuilder.setAttr("timeout_ms", opts.timeoutMs);
-        }
-      }
-    }
-    return new QueueDequeue(opBuilder.build());
-  }
-  
-  /**
-   * @param timeoutMs If the queue is empty, this operation will block for up to
-   * timeout_ms milliseconds.
-   * Note: This option is not supported yet.
-   */
-  public static Options timeoutMs(Long timeoutMs) {
-    return new Options().timeoutMs(timeoutMs);
-  }
-  
-  /**
-   * One or more tensors that were dequeued as a tuple.
-   */
-  public List<Output<?>> components() {
-    return components;
-  }
-  
-  @Override
-  @SuppressWarnings({"rawtypes", "unchecked"})
-  public Iterator<Operand<TType>> iterator() {
-    return (Iterator) components.iterator();
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "QueueDequeueV2";
-  
-  private List<Output<?>> components;
-  
-  private QueueDequeue(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    int componentsLength = operation.outputListLength("components");
-    components = Arrays.asList(operation.outputList(outputIdx, componentsLength));
-    outputIdx += componentsLength;
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/QueueDequeueMany.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/QueueDequeueMany.java
deleted file mode 100644
index b40686b18e6..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/QueueDequeueMany.java
+++ /dev/null
@@ -1,140 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.io;
-
-import java.util.Arrays;
-import java.util.Iterator;
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.family.TType;
-
-/**
- * Dequeues `n` tuples of one or more tensors from the given queue.
- * <p>
- * If the queue is closed and there are fewer than `n` elements, then an
- * OutOfRange error is returned.
- * <p>
- * This operation concatenates queue-element component tensors along the
- * 0th dimension to make a single component tensor.  All of the components
- * in the dequeued tuple will have size `n` in the 0th dimension.
- * <p>
- * This operation has `k` outputs, where `k` is the number of components in
- * the tuples stored in the given queue, and output `i` is the ith
- * component of the dequeued tuple.
- * <p>
- * N.B. If the queue is empty, this operation will block until `n` elements
- * have been dequeued (or 'timeout_ms' elapses, if specified).
- */
-@Operator(group = "io")
-public final class QueueDequeueMany extends RawOp implements Iterable<Operand<TType>> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.io.QueueDequeueMany}
-   */
-  public static class Options {
-    
-    /**
-     * @param timeoutMs If the queue has fewer than n elements, this operation
-     * will block for up to timeout_ms milliseconds.
-     * Note: This option is not supported yet.
-     */
-    public Options timeoutMs(Long timeoutMs) {
-      this.timeoutMs = timeoutMs;
-      return this;
-    }
-    
-    private Long timeoutMs;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new QueueDequeueMany operation.
-   * 
-   * @param scope current scope
-   * @param handle The handle to a queue.
-   * @param n The number of tuples to dequeue.
-   * @param componentTypes The type of each component in a tuple.
-   * @param options carries optional attributes values
-   * @return a new instance of QueueDequeueMany
-   */
-  @Endpoint(describeByClass = true)
-  public static QueueDequeueMany create(Scope scope, Operand<?> handle, Operand<TInt32> n, List<Class<?>> componentTypes, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("QueueDequeueManyV2", scope.makeOpName("QueueDequeueMany"));
-    opBuilder.addInput(handle.asOutput(scope));
-    opBuilder.addInput(n.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    Class[] componentTypesArray = new Class[componentTypes.size()];
-    for (int i = 0; i < componentTypesArray.length; ++i) {
-      componentTypesArray[i] = componentTypes.get(i);
-    }
-    opBuilder.setAttr("component_types", componentTypesArray);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.timeoutMs != null) {
-          opBuilder.setAttr("timeout_ms", opts.timeoutMs);
-        }
-      }
-    }
-    return new QueueDequeueMany(opBuilder.build());
-  }
-  
-  /**
-   * @param timeoutMs If the queue has fewer than n elements, this operation
-   * will block for up to timeout_ms milliseconds.
-   * Note: This option is not supported yet.
-   */
-  public static Options timeoutMs(Long timeoutMs) {
-    return new Options().timeoutMs(timeoutMs);
-  }
-  
-  /**
-   * One or more tensors that were dequeued as a tuple.
-   */
-  public List<Output<?>> components() {
-    return components;
-  }
-  
-  @Override
-  @SuppressWarnings({"rawtypes", "unchecked"})
-  public Iterator<Operand<TType>> iterator() {
-    return (Iterator) components.iterator();
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "QueueDequeueManyV2";
-  
-  private List<Output<?>> components;
-  
-  private QueueDequeueMany(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    int componentsLength = operation.outputListLength("components");
-    components = Arrays.asList(operation.outputList(outputIdx, componentsLength));
-    outputIdx += componentsLength;
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/QueueDequeueUpTo.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/QueueDequeueUpTo.java
deleted file mode 100644
index 82109f028d9..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/QueueDequeueUpTo.java
+++ /dev/null
@@ -1,144 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.io;
-
-import java.util.Arrays;
-import java.util.Iterator;
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.family.TType;
-
-/**
- * Dequeues `n` tuples of one or more tensors from the given queue.
- * <p>
- * This operation is not supported by all queues.  If a queue does not support
- * DequeueUpTo, then an Unimplemented error is returned.
- * <p>
- * If the queue is closed and there are more than 0 but less than `n`
- * elements remaining, then instead of returning an OutOfRange error like
- * QueueDequeueMany, less than `n` elements are returned immediately.  If
- * the queue is closed and there are 0 elements left in the queue, then
- * an OutOfRange error is returned just like in QueueDequeueMany.
- * Otherwise the behavior is identical to QueueDequeueMany:
- * <p>
- * This operation concatenates queue-element component tensors along the
- * 0th dimension to make a single component tensor.  All of the components
- * in the dequeued tuple will have size n in the 0th dimension.
- * <p>
- * This operation has `k` outputs, where `k` is the number of components in
- * the tuples stored in the given queue, and output `i` is the ith
- * component of the dequeued tuple.
- */
-@Operator(group = "io")
-public final class QueueDequeueUpTo extends RawOp implements Iterable<Operand<TType>> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.io.QueueDequeueUpTo}
-   */
-  public static class Options {
-    
-    /**
-     * @param timeoutMs If the queue has fewer than n elements, this operation
-     * will block for up to timeout_ms milliseconds.
-     * Note: This option is not supported yet.
-     */
-    public Options timeoutMs(Long timeoutMs) {
-      this.timeoutMs = timeoutMs;
-      return this;
-    }
-    
-    private Long timeoutMs;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new QueueDequeueUpTo operation.
-   * 
-   * @param scope current scope
-   * @param handle The handle to a queue.
-   * @param n The number of tuples to dequeue.
-   * @param componentTypes The type of each component in a tuple.
-   * @param options carries optional attributes values
-   * @return a new instance of QueueDequeueUpTo
-   */
-  @Endpoint(describeByClass = true)
-  public static QueueDequeueUpTo create(Scope scope, Operand<?> handle, Operand<TInt32> n, List<Class<?>> componentTypes, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("QueueDequeueUpToV2", scope.makeOpName("QueueDequeueUpTo"));
-    opBuilder.addInput(handle.asOutput(scope));
-    opBuilder.addInput(n.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    Class[] componentTypesArray = new Class[componentTypes.size()];
-    for (int i = 0; i < componentTypesArray.length; ++i) {
-      componentTypesArray[i] = componentTypes.get(i);
-    }
-    opBuilder.setAttr("component_types", componentTypesArray);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.timeoutMs != null) {
-          opBuilder.setAttr("timeout_ms", opts.timeoutMs);
-        }
-      }
-    }
-    return new QueueDequeueUpTo(opBuilder.build());
-  }
-  
-  /**
-   * @param timeoutMs If the queue has fewer than n elements, this operation
-   * will block for up to timeout_ms milliseconds.
-   * Note: This option is not supported yet.
-   */
-  public static Options timeoutMs(Long timeoutMs) {
-    return new Options().timeoutMs(timeoutMs);
-  }
-  
-  /**
-   * One or more tensors that were dequeued as a tuple.
-   */
-  public List<Output<?>> components() {
-    return components;
-  }
-  
-  @Override
-  @SuppressWarnings({"rawtypes", "unchecked"})
-  public Iterator<Operand<TType>> iterator() {
-    return (Iterator) components.iterator();
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "QueueDequeueUpToV2";
-  
-  private List<Output<?>> components;
-  
-  private QueueDequeueUpTo(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    int componentsLength = operation.outputListLength("components");
-    components = Arrays.asList(operation.outputList(outputIdx, componentsLength));
-    outputIdx += componentsLength;
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/QueueEnqueue.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/QueueEnqueue.java
deleted file mode 100644
index b37f9d4f848..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/QueueEnqueue.java
+++ /dev/null
@@ -1,102 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.io;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.Operands;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-
-/**
- * Enqueues a tuple of one or more tensors in the given queue.
- * <p>
- * The components input has k elements, which correspond to the components of
- * tuples stored in the given queue.
- * <p>
- * N.B. If the queue is full, this operation will block until the given
- * element has been enqueued (or 'timeout_ms' elapses, if specified).
- */
-@Operator(group = "io")
-public final class QueueEnqueue extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.io.QueueEnqueue}
-   */
-  public static class Options {
-    
-    /**
-     * @param timeoutMs If the queue is full, this operation will block for up to
-     * timeout_ms milliseconds.
-     * Note: This option is not supported yet.
-     */
-    public Options timeoutMs(Long timeoutMs) {
-      this.timeoutMs = timeoutMs;
-      return this;
-    }
-    
-    private Long timeoutMs;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new QueueEnqueue operation.
-   * 
-   * @param scope current scope
-   * @param handle The handle to a queue.
-   * @param components One or more tensors from which the enqueued tensors should be taken.
-   * @param options carries optional attributes values
-   * @return a new instance of QueueEnqueue
-   */
-  @Endpoint(describeByClass = true)
-  public static QueueEnqueue create(Scope scope, Operand<?> handle, Iterable<Operand<?>> components, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("QueueEnqueueV2", scope.makeOpName("QueueEnqueue"));
-    opBuilder.addInput(handle.asOutput(scope));
-    opBuilder.addInputList(Operands.asOutputs(scope, components));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.timeoutMs != null) {
-          opBuilder.setAttr("timeout_ms", opts.timeoutMs);
-        }
-      }
-    }
-    return new QueueEnqueue(opBuilder.build());
-  }
-  
-  /**
-   * @param timeoutMs If the queue is full, this operation will block for up to
-   * timeout_ms milliseconds.
-   * Note: This option is not supported yet.
-   */
-  public static Options timeoutMs(Long timeoutMs) {
-    return new Options().timeoutMs(timeoutMs);
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "QueueEnqueueV2";
-  
-  private QueueEnqueue(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/QueueEnqueueMany.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/QueueEnqueueMany.java
deleted file mode 100644
index b308e7aa7d2..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/QueueEnqueueMany.java
+++ /dev/null
@@ -1,107 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.io;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.Operands;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-
-/**
- * Enqueues zero or more tuples of one or more tensors in the given queue.
- * <p>
- * This operation slices each component tensor along the 0th dimension to
- * make multiple queue elements. All of the tuple components must have the
- * same size in the 0th dimension.
- * <p>
- * The components input has k elements, which correspond to the components of
- * tuples stored in the given queue.
- * <p>
- * N.B. If the queue is full, this operation will block until the given
- * elements have been enqueued (or 'timeout_ms' elapses, if specified).
- */
-@Operator(group = "io")
-public final class QueueEnqueueMany extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.io.QueueEnqueueMany}
-   */
-  public static class Options {
-    
-    /**
-     * @param timeoutMs If the queue is too full, this operation will block for up
-     * to timeout_ms milliseconds.
-     * Note: This option is not supported yet.
-     */
-    public Options timeoutMs(Long timeoutMs) {
-      this.timeoutMs = timeoutMs;
-      return this;
-    }
-    
-    private Long timeoutMs;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new QueueEnqueueMany operation.
-   * 
-   * @param scope current scope
-   * @param handle The handle to a queue.
-   * @param components One or more tensors from which the enqueued tensors should
-   * be taken.
-   * @param options carries optional attributes values
-   * @return a new instance of QueueEnqueueMany
-   */
-  @Endpoint(describeByClass = true)
-  public static QueueEnqueueMany create(Scope scope, Operand<?> handle, Iterable<Operand<?>> components, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("QueueEnqueueManyV2", scope.makeOpName("QueueEnqueueMany"));
-    opBuilder.addInput(handle.asOutput(scope));
-    opBuilder.addInputList(Operands.asOutputs(scope, components));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.timeoutMs != null) {
-          opBuilder.setAttr("timeout_ms", opts.timeoutMs);
-        }
-      }
-    }
-    return new QueueEnqueueMany(opBuilder.build());
-  }
-  
-  /**
-   * @param timeoutMs If the queue is too full, this operation will block for up
-   * to timeout_ms milliseconds.
-   * Note: This option is not supported yet.
-   */
-  public static Options timeoutMs(Long timeoutMs) {
-    return new Options().timeoutMs(timeoutMs);
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "QueueEnqueueManyV2";
-  
-  private QueueEnqueueMany(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/QueueIsClosed.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/QueueIsClosed.java
deleted file mode 100644
index 0cb6ac9997b..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/QueueIsClosed.java
+++ /dev/null
@@ -1,75 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.io;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TBool;
-
-/**
- * Returns true if queue is closed.
- * <p>
- * This operation returns true if the queue is closed and false if the queue
- * is open.
- */
-@Operator(group = "io")
-public final class QueueIsClosed extends RawOp implements Operand<TBool> {
-  
-  /**
-   * Factory method to create a class wrapping a new QueueIsClosed operation.
-   * 
-   * @param scope current scope
-   * @param handle The handle to a queue.
-   * @return a new instance of QueueIsClosed
-   */
-  @Endpoint(describeByClass = true)
-  public static QueueIsClosed create(Scope scope, Operand<?> handle) {
-    OperationBuilder opBuilder = scope.env().opBuilder("QueueIsClosedV2", scope.makeOpName("QueueIsClosed"));
-    opBuilder.addInput(handle.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new QueueIsClosed(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<TBool> isClosed() {
-    return isClosed;
-  }
-  
-  @Override
-  public Output<TBool> asOutput(Scope scope) {
-    return isClosed;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "QueueIsClosedV2";
-  
-  private Output<TBool> isClosed;
-  
-  private QueueIsClosed(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    isClosed = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/QueueSize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/QueueSize.java
deleted file mode 100644
index cfb5c24dbaa..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/QueueSize.java
+++ /dev/null
@@ -1,73 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.io;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
-
-/**
- * Computes the number of elements in the given queue.
- */
-@Operator(group = "io")
-public final class QueueSize extends RawOp implements Operand<TInt32> {
-  
-  /**
-   * Factory method to create a class wrapping a new QueueSize operation.
-   * 
-   * @param scope current scope
-   * @param handle The handle to a queue.
-   * @return a new instance of QueueSize
-   */
-  @Endpoint(describeByClass = true)
-  public static QueueSize create(Scope scope, Operand<?> handle) {
-    OperationBuilder opBuilder = scope.env().opBuilder("QueueSizeV2", scope.makeOpName("QueueSize"));
-    opBuilder.addInput(handle.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new QueueSize(opBuilder.build());
-  }
-  
-  /**
-   * The number of elements in the given queue.
-   */
-  public Output<TInt32> size() {
-    return size;
-  }
-  
-  @Override
-  public Output<TInt32> asOutput(Scope scope) {
-    return size;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "QueueSizeV2";
-  
-  private Output<TInt32> size;
-  
-  private QueueSize(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    size = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/RandomShuffleQueue.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/RandomShuffleQueue.java
deleted file mode 100644
index 16dd63c9b92..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/RandomShuffleQueue.java
+++ /dev/null
@@ -1,250 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.io;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * A queue that randomizes the order of elements.
- */
-@Operator(group = "io")
-public final class RandomShuffleQueue extends RawOp implements Operand<TType> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.io.RandomShuffleQueue}
-   */
-  public static class Options {
-    
-    /**
-     * @param shapes The shape of each component in a value. The length of this attr must
-     * be either 0 or the same as the length of component_types. If the length of
-     * this attr is 0, the shapes of queue elements are not constrained, and
-     * only one element may be dequeued at a time.
-     */
-    public Options shapes(List<Shape> shapes) {
-      this.shapes = shapes;
-      return this;
-    }
-    
-    /**
-     * @param capacity The upper bound on the number of elements in this queue.
-     * Negative numbers mean no limit.
-     */
-    public Options capacity(Long capacity) {
-      this.capacity = capacity;
-      return this;
-    }
-    
-    /**
-     * @param minAfterDequeue Dequeue will block unless there would be this
-     * many elements after the dequeue or the queue is closed. This
-     * ensures a minimum level of mixing of elements.
-     */
-    public Options minAfterDequeue(Long minAfterDequeue) {
-      this.minAfterDequeue = minAfterDequeue;
-      return this;
-    }
-    
-    /**
-     * @param seed If either seed or seed2 is set to be non-zero, the random number
-     * generator is seeded by the given seed.  Otherwise, a random seed is used.
-     */
-    public Options seed(Long seed) {
-      this.seed = seed;
-      return this;
-    }
-    
-    /**
-     * @param seed2 A second seed to avoid seed collision.
-     */
-    public Options seed2(Long seed2) {
-      this.seed2 = seed2;
-      return this;
-    }
-    
-    /**
-     * @param container If non-empty, this queue is placed in the given container.
-     * Otherwise, a default container is used.
-     */
-    public Options container(String container) {
-      this.container = container;
-      return this;
-    }
-    
-    /**
-     * @param sharedName If non-empty, this queue will be shared under the given name
-     * across multiple sessions.
-     */
-    public Options sharedName(String sharedName) {
-      this.sharedName = sharedName;
-      return this;
-    }
-    
-    private List<Shape> shapes;
-    private Long capacity;
-    private Long minAfterDequeue;
-    private Long seed;
-    private Long seed2;
-    private String container;
-    private String sharedName;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new RandomShuffleQueue operation.
-   * 
-   * @param scope current scope
-   * @param componentTypes The type of each component in a value.
-   * @param options carries optional attributes values
-   * @return a new instance of RandomShuffleQueue
-   */
-  @Endpoint(describeByClass = true)
-  public static RandomShuffleQueue create(Scope scope, List<Class<?>> componentTypes, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("RandomShuffleQueueV2", scope.makeOpName("RandomShuffleQueue"));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    Class[] componentTypesArray = new Class[componentTypes.size()];
-    for (int i = 0; i < componentTypesArray.length; ++i) {
-      componentTypesArray[i] = componentTypes.get(i);
-    }
-    opBuilder.setAttr("component_types", componentTypesArray);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.shapes != null) {
-          Shape[] shapesArray = new Shape[opts.shapes.size()];
-          for (int i = 0; i < shapesArray.length; ++i) {
-            shapesArray[i] = opts.shapes.get(i);
-          }
-          opBuilder.setAttr("shapes", shapesArray);
-        }
-        if (opts.capacity != null) {
-          opBuilder.setAttr("capacity", opts.capacity);
-        }
-        if (opts.minAfterDequeue != null) {
-          opBuilder.setAttr("min_after_dequeue", opts.minAfterDequeue);
-        }
-        if (opts.seed != null) {
-          opBuilder.setAttr("seed", opts.seed);
-        }
-        if (opts.seed2 != null) {
-          opBuilder.setAttr("seed2", opts.seed2);
-        }
-        if (opts.container != null) {
-          opBuilder.setAttr("container", opts.container);
-        }
-        if (opts.sharedName != null) {
-          opBuilder.setAttr("shared_name", opts.sharedName);
-        }
-      }
-    }
-    return new RandomShuffleQueue(opBuilder.build());
-  }
-  
-  /**
-   * @param shapes The shape of each component in a value. The length of this attr must
-   * be either 0 or the same as the length of component_types. If the length of
-   * this attr is 0, the shapes of queue elements are not constrained, and
-   * only one element may be dequeued at a time.
-   */
-  public static Options shapes(List<Shape> shapes) {
-    return new Options().shapes(shapes);
-  }
-  
-  /**
-   * @param capacity The upper bound on the number of elements in this queue.
-   * Negative numbers mean no limit.
-   */
-  public static Options capacity(Long capacity) {
-    return new Options().capacity(capacity);
-  }
-  
-  /**
-   * @param minAfterDequeue Dequeue will block unless there would be this
-   * many elements after the dequeue or the queue is closed. This
-   * ensures a minimum level of mixing of elements.
-   */
-  public static Options minAfterDequeue(Long minAfterDequeue) {
-    return new Options().minAfterDequeue(minAfterDequeue);
-  }
-  
-  /**
-   * @param seed If either seed or seed2 is set to be non-zero, the random number
-   * generator is seeded by the given seed.  Otherwise, a random seed is used.
-   */
-  public static Options seed(Long seed) {
-    return new Options().seed(seed);
-  }
-  
-  /**
-   * @param seed2 A second seed to avoid seed collision.
-   */
-  public static Options seed2(Long seed2) {
-    return new Options().seed2(seed2);
-  }
-  
-  /**
-   * @param container If non-empty, this queue is placed in the given container.
-   * Otherwise, a default container is used.
-   */
-  public static Options container(String container) {
-    return new Options().container(container);
-  }
-  
-  /**
-   * @param sharedName If non-empty, this queue will be shared under the given name
-   * across multiple sessions.
-   */
-  public static Options sharedName(String sharedName) {
-    return new Options().sharedName(sharedName);
-  }
-  
-  /**
-   * The handle to the queue.
-   */
-  public Output<?> handle() {
-    return handle;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) handle;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "RandomShuffleQueueV2";
-  
-  private Output<?> handle;
-  
-  private RandomShuffleQueue(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    handle = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ReadFile.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ReadFile.java
deleted file mode 100644
index 0911125282c..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ReadFile.java
+++ /dev/null
@@ -1,72 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.io;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TString;
-
-/**
- * Reads and outputs the entire contents of the input filename.
- */
-@Operator(group = "io")
-public final class ReadFile extends RawOp implements Operand<TString> {
-  
-  /**
-   * Factory method to create a class wrapping a new ReadFile operation.
-   * 
-   * @param scope current scope
-   * @param filename 
-   * @return a new instance of ReadFile
-   */
-  @Endpoint(describeByClass = true)
-  public static ReadFile create(Scope scope, Operand<TString> filename) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ReadFile", scope.makeOpName("ReadFile"));
-    opBuilder.addInput(filename.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new ReadFile(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<TString> contents() {
-    return contents;
-  }
-  
-  @Override
-  public Output<TString> asOutput(Scope scope) {
-    return contents;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ReadFile";
-  
-  private Output<TString> contents;
-  
-  private ReadFile(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    contents = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ReaderNumRecordsProduced.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ReaderNumRecordsProduced.java
deleted file mode 100644
index 25c000c9a5e..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ReaderNumRecordsProduced.java
+++ /dev/null
@@ -1,75 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.io;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt64;
-
-/**
- * Returns the number of records this Reader has produced.
- * <p>
- * This is the same as the number of ReaderRead executions that have
- * succeeded.
- */
-@Operator(group = "io")
-public final class ReaderNumRecordsProduced extends RawOp implements Operand<TInt64> {
-  
-  /**
-   * Factory method to create a class wrapping a new ReaderNumRecordsProduced operation.
-   * 
-   * @param scope current scope
-   * @param readerHandle Handle to a Reader.
-   * @return a new instance of ReaderNumRecordsProduced
-   */
-  @Endpoint(describeByClass = true)
-  public static ReaderNumRecordsProduced create(Scope scope, Operand<?> readerHandle) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ReaderNumRecordsProducedV2", scope.makeOpName("ReaderNumRecordsProduced"));
-    opBuilder.addInput(readerHandle.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new ReaderNumRecordsProduced(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<TInt64> recordsProduced() {
-    return recordsProduced;
-  }
-  
-  @Override
-  public Output<TInt64> asOutput(Scope scope) {
-    return recordsProduced;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ReaderNumRecordsProducedV2";
-  
-  private Output<TInt64> recordsProduced;
-  
-  private ReaderNumRecordsProduced(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    recordsProduced = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ReaderNumWorkUnitsCompleted.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ReaderNumWorkUnitsCompleted.java
deleted file mode 100644
index 3f4b161768c..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ReaderNumWorkUnitsCompleted.java
+++ /dev/null
@@ -1,72 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.io;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt64;
-
-/**
- * Returns the number of work units this Reader has finished processing.
- */
-@Operator(group = "io")
-public final class ReaderNumWorkUnitsCompleted extends RawOp implements Operand<TInt64> {
-  
-  /**
-   * Factory method to create a class wrapping a new ReaderNumWorkUnitsCompleted operation.
-   * 
-   * @param scope current scope
-   * @param readerHandle Handle to a Reader.
-   * @return a new instance of ReaderNumWorkUnitsCompleted
-   */
-  @Endpoint(describeByClass = true)
-  public static ReaderNumWorkUnitsCompleted create(Scope scope, Operand<?> readerHandle) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ReaderNumWorkUnitsCompletedV2", scope.makeOpName("ReaderNumWorkUnitsCompleted"));
-    opBuilder.addInput(readerHandle.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new ReaderNumWorkUnitsCompleted(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<TInt64> unitsCompleted() {
-    return unitsCompleted;
-  }
-  
-  @Override
-  public Output<TInt64> asOutput(Scope scope) {
-    return unitsCompleted;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ReaderNumWorkUnitsCompletedV2";
-  
-  private Output<TInt64> unitsCompleted;
-  
-  private ReaderNumWorkUnitsCompleted(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    unitsCompleted = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ReaderRead.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ReaderRead.java
deleted file mode 100644
index 7c52b881b02..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ReaderRead.java
+++ /dev/null
@@ -1,83 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.io;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TString;
-
-/**
- * Returns the next record (key, value pair) produced by a Reader.
- * <p>
- * Will dequeue from the input queue if necessary (e.g. when the
- * Reader needs to start reading from a new file since it has finished
- * with the previous file).
- */
-@Operator(group = "io")
-public final class ReaderRead extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new ReaderRead operation.
-   * 
-   * @param scope current scope
-   * @param readerHandle Handle to a Reader.
-   * @param queueHandle Handle to a Queue, with string work items.
-   * @return a new instance of ReaderRead
-   */
-  @Endpoint(describeByClass = true)
-  public static ReaderRead create(Scope scope, Operand<?> readerHandle, Operand<?> queueHandle) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ReaderReadV2", scope.makeOpName("ReaderRead"));
-    opBuilder.addInput(readerHandle.asOutput(scope));
-    opBuilder.addInput(queueHandle.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new ReaderRead(opBuilder.build());
-  }
-  
-  /**
-   * A scalar.
-   */
-  public Output<TString> key() {
-    return key;
-  }
-  
-  /**
-   * A scalar.
-   */
-  public Output<TString> value() {
-    return value;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ReaderReadV2";
-  
-  private Output<TString> key;
-  private Output<TString> value;
-  
-  private ReaderRead(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    key = operation.output(outputIdx++);
-    value = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ReaderReadUpTo.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ReaderReadUpTo.java
deleted file mode 100644
index 168ebf8db6a..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ReaderReadUpTo.java
+++ /dev/null
@@ -1,87 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.io;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.TString;
-
-/**
- * Returns up to `num_records` (key, value) pairs produced by a Reader.
- * <p>
- * Will dequeue from the input queue if necessary (e.g. when the
- * Reader needs to start reading from a new file since it has finished
- * with the previous file).
- * It may return less than `num_records` even before the last batch.
- */
-@Operator(group = "io")
-public final class ReaderReadUpTo extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new ReaderReadUpTo operation.
-   * 
-   * @param scope current scope
-   * @param readerHandle Handle to a `Reader`.
-   * @param queueHandle Handle to a `Queue`, with string work items.
-   * @param numRecords number of records to read from `Reader`.
-   * @return a new instance of ReaderReadUpTo
-   */
-  @Endpoint(describeByClass = true)
-  public static ReaderReadUpTo create(Scope scope, Operand<?> readerHandle, Operand<?> queueHandle, Operand<TInt64> numRecords) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ReaderReadUpToV2", scope.makeOpName("ReaderReadUpTo"));
-    opBuilder.addInput(readerHandle.asOutput(scope));
-    opBuilder.addInput(queueHandle.asOutput(scope));
-    opBuilder.addInput(numRecords.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new ReaderReadUpTo(opBuilder.build());
-  }
-  
-  /**
-   * A 1-D tensor.
-   */
-  public Output<TString> keys() {
-    return keys;
-  }
-  
-  /**
-   * A 1-D tensor.
-   */
-  public Output<TString> values() {
-    return values;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ReaderReadUpToV2";
-  
-  private Output<TString> keys;
-  private Output<TString> values;
-  
-  private ReaderReadUpTo(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    keys = operation.output(outputIdx++);
-    values = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ReaderReset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ReaderReset.java
deleted file mode 100644
index ffb6898ac39..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ReaderReset.java
+++ /dev/null
@@ -1,55 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.io;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-
-/**
- * Restore a Reader to its initial clean state.
- */
-@Operator(group = "io")
-public final class ReaderReset extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new ReaderReset operation.
-   * 
-   * @param scope current scope
-   * @param readerHandle Handle to a Reader.
-   * @return a new instance of ReaderReset
-   */
-  @Endpoint(describeByClass = true)
-  public static ReaderReset create(Scope scope, Operand<?> readerHandle) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ReaderResetV2", scope.makeOpName("ReaderReset"));
-    opBuilder.addInput(readerHandle.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new ReaderReset(opBuilder.build());
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ReaderResetV2";
-  
-  private ReaderReset(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ReaderRestoreState.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ReaderRestoreState.java
deleted file mode 100644
index 8ecfd87479e..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ReaderRestoreState.java
+++ /dev/null
@@ -1,62 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.io;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TString;
-
-/**
- * Restore a reader to a previously saved state.
- * <p>
- * Not all Readers support being restored, so this can produce an
- * Unimplemented error.
- */
-@Operator(group = "io")
-public final class ReaderRestoreState extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new ReaderRestoreState operation.
-   * 
-   * @param scope current scope
-   * @param readerHandle Handle to a Reader.
-   * @param state Result of a ReaderSerializeState of a Reader with type
-   * matching reader_handle.
-   * @return a new instance of ReaderRestoreState
-   */
-  @Endpoint(describeByClass = true)
-  public static ReaderRestoreState create(Scope scope, Operand<?> readerHandle, Operand<TString> state) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ReaderRestoreStateV2", scope.makeOpName("ReaderRestoreState"));
-    opBuilder.addInput(readerHandle.asOutput(scope));
-    opBuilder.addInput(state.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new ReaderRestoreState(opBuilder.build());
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ReaderRestoreStateV2";
-  
-  private ReaderRestoreState(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ReaderSerializeState.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ReaderSerializeState.java
deleted file mode 100644
index affefa95016..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ReaderSerializeState.java
+++ /dev/null
@@ -1,75 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.io;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TString;
-
-/**
- * Produce a string tensor that encodes the state of a Reader.
- * <p>
- * Not all Readers support being serialized, so this can produce an
- * Unimplemented error.
- */
-@Operator(group = "io")
-public final class ReaderSerializeState extends RawOp implements Operand<TString> {
-  
-  /**
-   * Factory method to create a class wrapping a new ReaderSerializeState operation.
-   * 
-   * @param scope current scope
-   * @param readerHandle Handle to a Reader.
-   * @return a new instance of ReaderSerializeState
-   */
-  @Endpoint(describeByClass = true)
-  public static ReaderSerializeState create(Scope scope, Operand<?> readerHandle) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ReaderSerializeStateV2", scope.makeOpName("ReaderSerializeState"));
-    opBuilder.addInput(readerHandle.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new ReaderSerializeState(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<TString> state() {
-    return state;
-  }
-  
-  @Override
-  public Output<TString> asOutput(Scope scope) {
-    return state;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ReaderSerializeStateV2";
-  
-  private Output<TString> state;
-  
-  private ReaderSerializeState(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    state = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/SerializeManySparse.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/SerializeManySparse.java
deleted file mode 100644
index 2acc3cef171..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/SerializeManySparse.java
+++ /dev/null
@@ -1,105 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.io;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.TString;
-import org.tensorflow.types.family.TType;
-
-/**
- * Serialize an `N`-minibatch `SparseTensor` into an `[N, 3]` `Tensor` object.
- * <p>
- * The `SparseTensor` must have rank `R` greater than 1, and the first dimension
- * is treated as the minibatch dimension.  Elements of the `SparseTensor`
- * must be sorted in increasing order of this first dimension.  The serialized
- * `SparseTensor` objects going into each row of `serialized_sparse` will have
- * rank `R-1`.
- * <p>
- * The minibatch size `N` is extracted from `sparse_shape[0]`.
- * 
- * @param <U> data type for {@code serializedSparse()} output
- */
-@Operator(group = "io")
-public final class SerializeManySparse<U extends TType> extends RawOp implements Operand<U> {
-  
-  /**
-   * Factory method to create a class wrapping a new SerializeManySparse operation.
-   * 
-   * @param scope current scope
-   * @param sparseIndices 2-D.  The `indices` of the minibatch `SparseTensor`.
-   * @param sparseValues 1-D.  The `values` of the minibatch `SparseTensor`.
-   * @param sparseShape 1-D.  The `shape` of the minibatch `SparseTensor`.
-   * @param outType The `dtype` to use for serialization; the supported types are `string`
-   * (default) and `variant`.
-   * @return a new instance of SerializeManySparse
-   */
-  @Endpoint(describeByClass = true)
-  public static <U extends TType, T extends TType> SerializeManySparse<U> create(Scope scope, Operand<TInt64> sparseIndices, Operand<T> sparseValues, Operand<TInt64> sparseShape, Class<U> outType) {
-    OperationBuilder opBuilder = scope.env().opBuilder("SerializeManySparse", scope.makeOpName("SerializeManySparse"));
-    opBuilder.addInput(sparseIndices.asOutput(scope));
-    opBuilder.addInput(sparseValues.asOutput(scope));
-    opBuilder.addInput(sparseShape.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("out_type", outType);
-    return new SerializeManySparse<U>(opBuilder.build());
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new SerializeManySparse operation using default output types.
-   * 
-   * @param scope current scope
-   * @param sparseIndices 2-D.  The `indices` of the minibatch `SparseTensor`.
-   * @param sparseValues 1-D.  The `values` of the minibatch `SparseTensor`.
-   * @param sparseShape 1-D.  The `shape` of the minibatch `SparseTensor`.
-   * @return a new instance of SerializeManySparse
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> SerializeManySparse<TString> create(Scope scope, Operand<TInt64> sparseIndices, Operand<T> sparseValues, Operand<TInt64> sparseShape) {
-    return create(scope, sparseIndices, sparseValues, sparseShape, TString.class);
-  }
-  
-  /**
-   */
-  public Output<U> serializedSparse() {
-    return serializedSparse;
-  }
-  
-  @Override
-  public Output<U> asOutput(Scope scope) {
-    return serializedSparse;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "SerializeManySparse";
-  
-  private Output<U> serializedSparse;
-  
-  private SerializeManySparse(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    serializedSparse = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/SerializeSparse.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/SerializeSparse.java
deleted file mode 100644
index c6935770458..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/SerializeSparse.java
+++ /dev/null
@@ -1,97 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.io;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.TString;
-import org.tensorflow.types.family.TType;
-
-/**
- * Serialize a `SparseTensor` into a `[3]` `Tensor` object.
- * 
- * @param <U> data type for {@code serializedSparse()} output
- */
-@Operator(group = "io")
-public final class SerializeSparse<U extends TType> extends RawOp implements Operand<U> {
-  
-  /**
-   * Factory method to create a class wrapping a new SerializeSparse operation.
-   * 
-   * @param scope current scope
-   * @param sparseIndices 2-D.  The `indices` of the `SparseTensor`.
-   * @param sparseValues 1-D.  The `values` of the `SparseTensor`.
-   * @param sparseShape 1-D.  The `shape` of the `SparseTensor`.
-   * @param outType The `dtype` to use for serialization; the supported types are `string`
-   * (default) and `variant`.
-   * @return a new instance of SerializeSparse
-   */
-  @Endpoint(describeByClass = true)
-  public static <U extends TType, T extends TType> SerializeSparse<U> create(Scope scope, Operand<TInt64> sparseIndices, Operand<T> sparseValues, Operand<TInt64> sparseShape, Class<U> outType) {
-    OperationBuilder opBuilder = scope.env().opBuilder("SerializeSparse", scope.makeOpName("SerializeSparse"));
-    opBuilder.addInput(sparseIndices.asOutput(scope));
-    opBuilder.addInput(sparseValues.asOutput(scope));
-    opBuilder.addInput(sparseShape.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("out_type", outType);
-    return new SerializeSparse<U>(opBuilder.build());
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new SerializeSparse operation using default output types.
-   * 
-   * @param scope current scope
-   * @param sparseIndices 2-D.  The `indices` of the `SparseTensor`.
-   * @param sparseValues 1-D.  The `values` of the `SparseTensor`.
-   * @param sparseShape 1-D.  The `shape` of the `SparseTensor`.
-   * @return a new instance of SerializeSparse
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> SerializeSparse<TString> create(Scope scope, Operand<TInt64> sparseIndices, Operand<T> sparseValues, Operand<TInt64> sparseShape) {
-    return create(scope, sparseIndices, sparseValues, sparseShape, TString.class);
-  }
-  
-  /**
-   */
-  public Output<U> serializedSparse() {
-    return serializedSparse;
-  }
-  
-  @Override
-  public Output<U> asOutput(Scope scope) {
-    return serializedSparse;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "SerializeSparse";
-  
-  private Output<U> serializedSparse;
-  
-  private SerializeSparse(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    serializedSparse = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/SerializeTensor.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/SerializeTensor.java
deleted file mode 100644
index cc3b71be5d9..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/SerializeTensor.java
+++ /dev/null
@@ -1,74 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.io;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TString;
-import org.tensorflow.types.family.TType;
-
-/**
- * Transforms a Tensor into a serialized TensorProto proto.
- */
-@Operator(group = "io")
-public final class SerializeTensor extends RawOp implements Operand<TString> {
-  
-  /**
-   * Factory method to create a class wrapping a new SerializeTensor operation.
-   * 
-   * @param scope current scope
-   * @param tensor A Tensor of type `T`.
-   * @return a new instance of SerializeTensor
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> SerializeTensor create(Scope scope, Operand<T> tensor) {
-    OperationBuilder opBuilder = scope.env().opBuilder("SerializeTensor", scope.makeOpName("SerializeTensor"));
-    opBuilder.addInput(tensor.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new SerializeTensor(opBuilder.build());
-  }
-  
-  /**
-   * A serialized TensorProto proto of the input tensor.
-   */
-  public Output<TString> serialized() {
-    return serialized;
-  }
-  
-  @Override
-  public Output<TString> asOutput(Scope scope) {
-    return serialized;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "SerializeTensor";
-  
-  private Output<TString> serialized;
-  
-  private SerializeTensor(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    serialized = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ShardedFilename.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ShardedFilename.java
deleted file mode 100644
index 4ddc9959485..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ShardedFilename.java
+++ /dev/null
@@ -1,79 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.io;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.TString;
-
-/**
- * Generate a sharded filename. The filename is printf formatted as
- * <p>
- *    %s-%05d-of-%05d, basename, shard, num_shards.
- */
-@Operator(group = "io")
-public final class ShardedFilename extends RawOp implements Operand<TString> {
-  
-  /**
-   * Factory method to create a class wrapping a new ShardedFilename operation.
-   * 
-   * @param scope current scope
-   * @param basename 
-   * @param shard 
-   * @param numShards 
-   * @return a new instance of ShardedFilename
-   */
-  @Endpoint(describeByClass = true)
-  public static ShardedFilename create(Scope scope, Operand<TString> basename, Operand<TInt32> shard, Operand<TInt32> numShards) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ShardedFilename", scope.makeOpName("ShardedFilename"));
-    opBuilder.addInput(basename.asOutput(scope));
-    opBuilder.addInput(shard.asOutput(scope));
-    opBuilder.addInput(numShards.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new ShardedFilename(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<TString> filename() {
-    return filename;
-  }
-  
-  @Override
-  public Output<TString> asOutput(Scope scope) {
-    return filename;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ShardedFilename";
-  
-  private Output<TString> filename;
-  
-  private ShardedFilename(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    filename = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ShardedFilespec.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ShardedFilespec.java
deleted file mode 100644
index b48c4d37d88..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ShardedFilespec.java
+++ /dev/null
@@ -1,75 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.io;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.TString;
-
-/**
- * Generate a glob pattern matching all sharded file names.
- */
-@Operator(group = "io")
-public final class ShardedFilespec extends RawOp implements Operand<TString> {
-  
-  /**
-   * Factory method to create a class wrapping a new ShardedFilespec operation.
-   * 
-   * @param scope current scope
-   * @param basename 
-   * @param numShards 
-   * @return a new instance of ShardedFilespec
-   */
-  @Endpoint(describeByClass = true)
-  public static ShardedFilespec create(Scope scope, Operand<TString> basename, Operand<TInt32> numShards) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ShardedFilespec", scope.makeOpName("ShardedFilespec"));
-    opBuilder.addInput(basename.asOutput(scope));
-    opBuilder.addInput(numShards.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new ShardedFilespec(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<TString> filename() {
-    return filename;
-  }
-  
-  @Override
-  public Output<TString> asOutput(Scope scope) {
-    return filename;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ShardedFilespec";
-  
-  private Output<TString> filename;
-  
-  private ShardedFilespec(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    filename = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/TextLineReader.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/TextLineReader.java
deleted file mode 100644
index 5db34187e96..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/TextLineReader.java
+++ /dev/null
@@ -1,148 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.io;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * A Reader that outputs the lines of a file delimited by '\n'.
- */
-@Operator(group = "io")
-public final class TextLineReader extends RawOp implements Operand<TType> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.io.TextLineReader}
-   */
-  public static class Options {
-    
-    /**
-     * @param skipHeaderLines Number of lines to skip from the beginning of every file.
-     */
-    public Options skipHeaderLines(Long skipHeaderLines) {
-      this.skipHeaderLines = skipHeaderLines;
-      return this;
-    }
-    
-    /**
-     * @param container If non-empty, this reader is placed in the given container.
-     * Otherwise, a default container is used.
-     */
-    public Options container(String container) {
-      this.container = container;
-      return this;
-    }
-    
-    /**
-     * @param sharedName If non-empty, this reader is named in the given bucket
-     * with this shared_name. Otherwise, the node name is used instead.
-     */
-    public Options sharedName(String sharedName) {
-      this.sharedName = sharedName;
-      return this;
-    }
-    
-    private Long skipHeaderLines;
-    private String container;
-    private String sharedName;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new TextLineReader operation.
-   * 
-   * @param scope current scope
-   * @param options carries optional attributes values
-   * @return a new instance of TextLineReader
-   */
-  @Endpoint(describeByClass = true)
-  public static TextLineReader create(Scope scope, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("TextLineReaderV2", scope.makeOpName("TextLineReader"));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.skipHeaderLines != null) {
-          opBuilder.setAttr("skip_header_lines", opts.skipHeaderLines);
-        }
-        if (opts.container != null) {
-          opBuilder.setAttr("container", opts.container);
-        }
-        if (opts.sharedName != null) {
-          opBuilder.setAttr("shared_name", opts.sharedName);
-        }
-      }
-    }
-    return new TextLineReader(opBuilder.build());
-  }
-  
-  /**
-   * @param skipHeaderLines Number of lines to skip from the beginning of every file.
-   */
-  public static Options skipHeaderLines(Long skipHeaderLines) {
-    return new Options().skipHeaderLines(skipHeaderLines);
-  }
-  
-  /**
-   * @param container If non-empty, this reader is placed in the given container.
-   * Otherwise, a default container is used.
-   */
-  public static Options container(String container) {
-    return new Options().container(container);
-  }
-  
-  /**
-   * @param sharedName If non-empty, this reader is named in the given bucket
-   * with this shared_name. Otherwise, the node name is used instead.
-   */
-  public static Options sharedName(String sharedName) {
-    return new Options().sharedName(sharedName);
-  }
-  
-  /**
-   * The handle to reference the Reader.
-   */
-  public Output<?> readerHandle() {
-    return readerHandle;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) readerHandle;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "TextLineReaderV2";
-  
-  private Output<?> readerHandle;
-  
-  private TextLineReader(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    readerHandle = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/TfRecordReader.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/TfRecordReader.java
deleted file mode 100644
index 502ff7fe719..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/TfRecordReader.java
+++ /dev/null
@@ -1,148 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.io;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * A Reader that outputs the records from a TensorFlow Records file.
- */
-@Operator(group = "io")
-public final class TfRecordReader extends RawOp implements Operand<TType> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.io.TfRecordReader}
-   */
-  public static class Options {
-    
-    /**
-     * @param container If non-empty, this reader is placed in the given container.
-     * Otherwise, a default container is used.
-     */
-    public Options container(String container) {
-      this.container = container;
-      return this;
-    }
-    
-    /**
-     * @param sharedName If non-empty, this reader is named in the given bucket
-     * with this shared_name. Otherwise, the node name is used instead.
-     */
-    public Options sharedName(String sharedName) {
-      this.sharedName = sharedName;
-      return this;
-    }
-    
-    /**
-     * @param compressionType 
-     */
-    public Options compressionType(String compressionType) {
-      this.compressionType = compressionType;
-      return this;
-    }
-    
-    private String container;
-    private String sharedName;
-    private String compressionType;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new TfRecordReader operation.
-   * 
-   * @param scope current scope
-   * @param options carries optional attributes values
-   * @return a new instance of TfRecordReader
-   */
-  @Endpoint(describeByClass = true)
-  public static TfRecordReader create(Scope scope, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("TFRecordReaderV2", scope.makeOpName("TfRecordReader"));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.container != null) {
-          opBuilder.setAttr("container", opts.container);
-        }
-        if (opts.sharedName != null) {
-          opBuilder.setAttr("shared_name", opts.sharedName);
-        }
-        if (opts.compressionType != null) {
-          opBuilder.setAttr("compression_type", opts.compressionType);
-        }
-      }
-    }
-    return new TfRecordReader(opBuilder.build());
-  }
-  
-  /**
-   * @param container If non-empty, this reader is placed in the given container.
-   * Otherwise, a default container is used.
-   */
-  public static Options container(String container) {
-    return new Options().container(container);
-  }
-  
-  /**
-   * @param sharedName If non-empty, this reader is named in the given bucket
-   * with this shared_name. Otherwise, the node name is used instead.
-   */
-  public static Options sharedName(String sharedName) {
-    return new Options().sharedName(sharedName);
-  }
-  
-  /**
-   * @param compressionType 
-   */
-  public static Options compressionType(String compressionType) {
-    return new Options().compressionType(compressionType);
-  }
-  
-  /**
-   * The handle to reference the Reader.
-   */
-  public Output<?> readerHandle() {
-    return readerHandle;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) readerHandle;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "TFRecordReaderV2";
-  
-  private Output<?> readerHandle;
-  
-  private TfRecordReader(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    readerHandle = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/WholeFileReader.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/WholeFileReader.java
deleted file mode 100644
index 7be00710bdb..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/WholeFileReader.java
+++ /dev/null
@@ -1,132 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.io;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * A Reader that outputs the entire contents of a file as a value.
- * <p>
- * To use, enqueue filenames in a Queue.  The output of ReaderRead will
- * be a filename (key) and the contents of that file (value).
- */
-@Operator(group = "io")
-public final class WholeFileReader extends RawOp implements Operand<TType> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.io.WholeFileReader}
-   */
-  public static class Options {
-    
-    /**
-     * @param container If non-empty, this reader is placed in the given container.
-     * Otherwise, a default container is used.
-     */
-    public Options container(String container) {
-      this.container = container;
-      return this;
-    }
-    
-    /**
-     * @param sharedName If non-empty, this reader is named in the given bucket
-     * with this shared_name. Otherwise, the node name is used instead.
-     */
-    public Options sharedName(String sharedName) {
-      this.sharedName = sharedName;
-      return this;
-    }
-    
-    private String container;
-    private String sharedName;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new WholeFileReader operation.
-   * 
-   * @param scope current scope
-   * @param options carries optional attributes values
-   * @return a new instance of WholeFileReader
-   */
-  @Endpoint(describeByClass = true)
-  public static WholeFileReader create(Scope scope, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("WholeFileReaderV2", scope.makeOpName("WholeFileReader"));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.container != null) {
-          opBuilder.setAttr("container", opts.container);
-        }
-        if (opts.sharedName != null) {
-          opBuilder.setAttr("shared_name", opts.sharedName);
-        }
-      }
-    }
-    return new WholeFileReader(opBuilder.build());
-  }
-  
-  /**
-   * @param container If non-empty, this reader is placed in the given container.
-   * Otherwise, a default container is used.
-   */
-  public static Options container(String container) {
-    return new Options().container(container);
-  }
-  
-  /**
-   * @param sharedName If non-empty, this reader is named in the given bucket
-   * with this shared_name. Otherwise, the node name is used instead.
-   */
-  public static Options sharedName(String sharedName) {
-    return new Options().sharedName(sharedName);
-  }
-  
-  /**
-   * The handle to reference the Reader.
-   */
-  public Output<?> readerHandle() {
-    return readerHandle;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) readerHandle;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "WholeFileReaderV2";
-  
-  private Output<?> readerHandle;
-  
-  private WholeFileReader(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    readerHandle = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/WriteFile.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/WriteFile.java
deleted file mode 100644
index e43565c431a..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/WriteFile.java
+++ /dev/null
@@ -1,60 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.io;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TString;
-
-/**
- * Writes contents to the file at input filename. Creates file and recursively
- * <p>
- * creates directory if not existing.
- */
-@Operator(group = "io")
-public final class WriteFile extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new WriteFile operation.
-   * 
-   * @param scope current scope
-   * @param filename scalar. The name of the file to which we write the contents.
-   * @param contents scalar. The content to be written to the output file.
-   * @return a new instance of WriteFile
-   */
-  @Endpoint(describeByClass = true)
-  public static WriteFile create(Scope scope, Operand<TString> filename, Operand<TString> contents) {
-    OperationBuilder opBuilder = scope.env().opBuilder("WriteFile", scope.makeOpName("WriteFile"));
-    opBuilder.addInput(filename.asOutput(scope));
-    opBuilder.addInput(contents.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new WriteFile(opBuilder.build());
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "WriteFile";
-  
-  private WriteFile(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BandPart.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BandPart.java
deleted file mode 100644
index 9cebc6f40d1..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BandPart.java
+++ /dev/null
@@ -1,118 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.linalg;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * Copy a tensor setting everything outside a central band in each innermost matrix to zero.
- * <p>
- * The `band` part is computed as follows:
- * Assume `input` has `k` dimensions `[I, J, K, ..., M, N]`, then the output is a
- * tensor with the same shape where
- * <p>
- * `band[i, j, k, ..., m, n] = in_band(m, n) * input[i, j, k, ..., m, n]`.
- * <p>
- * The indicator function
- * <p>
- * `in_band(m, n) = (num_lower < 0 || (m-n) <= num_lower)) &&
- *                  (num_upper < 0 || (n-m) <= num_upper)`.
- * <p>
- * For example:
- * <pre>{@code
- * # if 'input' is [[ 0,  1,  2, 3]
- *                  [-1,  0,  1, 2]
- *                  [-2, -1,  0, 1]
- *                  [-3, -2, -1, 0]],
- * 
- * tf.matrix_band_part(input, 1, -1) ==> [[ 0,  1,  2, 3]
- *                                        [-1,  0,  1, 2]
- *                                        [ 0, -1,  0, 1]
- *                                        [ 0,  0, -1, 0]],
- * 
- * tf.matrix_band_part(input, 2, 1) ==> [[ 0,  1,  0, 0]
- *                                       [-1,  0,  1, 0]
- *                                       [-2, -1,  0, 1]
- *                                       [ 0, -2, -1, 0]]
- * }</pre>
- * Useful special cases:
- * <pre>{@code
- *  tf.matrix_band_part(input, 0, -1) ==> Upper triangular part.
- *  tf.matrix_band_part(input, -1, 0) ==> Lower triangular part.
- *  tf.matrix_band_part(input, 0, 0) ==> Diagonal.
- * }</pre>
- * 
- * 
- * @param <T> data type for {@code band()} output
- */
-@Operator(group = "linalg")
-public final class BandPart<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new BandPart operation.
-   * 
-   * @param scope current scope
-   * @param input Rank `k` tensor.
-   * @param numLower 0-D tensor. Number of subdiagonals to keep. If negative, keep entire
-   * lower triangle.
-   * @param numUpper 0-D tensor. Number of superdiagonals to keep. If negative, keep
-   * entire upper triangle.
-   * @return a new instance of BandPart
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> BandPart<T> create(Scope scope, Operand<T> input, Operand<U> numLower, Operand<U> numUpper) {
-    OperationBuilder opBuilder = scope.env().opBuilder("MatrixBandPart", scope.makeOpName("BandPart"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder.addInput(numLower.asOutput(scope));
-    opBuilder.addInput(numUpper.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new BandPart<T>(opBuilder.build());
-  }
-  
-  /**
-   * Rank `k` tensor of the same shape as input. The extracted banded tensor.
-   */
-  public Output<T> band() {
-    return band;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return band;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "MatrixBandPart";
-  
-  private Output<T> band;
-  
-  private BandPart(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    band = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BandedTriangularSolve.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BandedTriangularSolve.java
deleted file mode 100644
index cba49b57fb0..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BandedTriangularSolve.java
+++ /dev/null
@@ -1,126 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.linalg;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * @param <T> data type for {@code output()} output
- */
-public final class BandedTriangularSolve<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.linalg.BandedTriangularSolve}
-   */
-  public static class Options {
-    
-    /**
-     * @param lower 
-     */
-    public Options lower(Boolean lower) {
-      this.lower = lower;
-      return this;
-    }
-    
-    /**
-     * @param adjoint 
-     */
-    public Options adjoint(Boolean adjoint) {
-      this.adjoint = adjoint;
-      return this;
-    }
-    
-    private Boolean lower;
-    private Boolean adjoint;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new BandedTriangularSolve operation.
-   * 
-   * @param scope current scope
-   * @param matrix 
-   * @param rhs 
-   * @param options carries optional attributes values
-   * @return a new instance of BandedTriangularSolve
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> BandedTriangularSolve<T> create(Scope scope, Operand<T> matrix, Operand<T> rhs, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("BandedTriangularSolve", scope.makeOpName("BandedTriangularSolve"));
-    opBuilder.addInput(matrix.asOutput(scope));
-    opBuilder.addInput(rhs.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.lower != null) {
-          opBuilder.setAttr("lower", opts.lower);
-        }
-        if (opts.adjoint != null) {
-          opBuilder.setAttr("adjoint", opts.adjoint);
-        }
-      }
-    }
-    return new BandedTriangularSolve<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param lower 
-   */
-  public static Options lower(Boolean lower) {
-    return new Options().lower(lower);
-  }
-  
-  /**
-   * @param adjoint 
-   */
-  public static Options adjoint(Boolean adjoint) {
-    return new Options().adjoint(adjoint);
-  }
-  
-  /**
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "BandedTriangularSolve";
-  
-  private Output<T> output;
-  
-  private BandedTriangularSolve(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchCholesky.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchCholesky.java
deleted file mode 100644
index 088f7e118b0..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchCholesky.java
+++ /dev/null
@@ -1,72 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.linalg;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * @param <T> data type for {@code output()} output
- */
-@Operator(group = "linalg")
-public final class BatchCholesky<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new BatchCholesky operation.
-   * 
-   * @param scope current scope
-   * @param input 
-   * @return a new instance of BatchCholesky
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> BatchCholesky<T> create(Scope scope, Operand<T> input) {
-    OperationBuilder opBuilder = scope.env().opBuilder("BatchCholesky", scope.makeOpName("BatchCholesky"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new BatchCholesky<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "BatchCholesky";
-  
-  private Output<T> output;
-  
-  private BatchCholesky(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchCholeskyGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchCholeskyGrad.java
deleted file mode 100644
index 608b2e9d4a5..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchCholeskyGrad.java
+++ /dev/null
@@ -1,74 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.linalg;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * @param <T> data type for {@code output()} output
- */
-@Operator(group = "linalg")
-public final class BatchCholeskyGrad<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new BatchCholeskyGrad operation.
-   * 
-   * @param scope current scope
-   * @param l 
-   * @param grad 
-   * @return a new instance of BatchCholeskyGrad
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> BatchCholeskyGrad<T> create(Scope scope, Operand<T> l, Operand<T> grad) {
-    OperationBuilder opBuilder = scope.env().opBuilder("BatchCholeskyGrad", scope.makeOpName("BatchCholeskyGrad"));
-    opBuilder.addInput(l.asOutput(scope));
-    opBuilder.addInput(grad.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new BatchCholeskyGrad<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "BatchCholeskyGrad";
-  
-  private Output<T> output;
-  
-  private BatchCholeskyGrad(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixBandPart.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixBandPart.java
deleted file mode 100644
index 4ba6f583f74..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixBandPart.java
+++ /dev/null
@@ -1,77 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.linalg;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
-
-/**
- * @param <T> data type for {@code band()} output
- */
-@Operator(group = "linalg")
-public final class BatchMatrixBandPart<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new BatchMatrixBandPart operation.
-   * 
-   * @param scope current scope
-   * @param input 
-   * @param numLower 
-   * @param numUpper 
-   * @return a new instance of BatchMatrixBandPart
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> BatchMatrixBandPart<T> create(Scope scope, Operand<T> input, Operand<TInt64> numLower, Operand<TInt64> numUpper) {
-    OperationBuilder opBuilder = scope.env().opBuilder("BatchMatrixBandPart", scope.makeOpName("BatchMatrixBandPart"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder.addInput(numLower.asOutput(scope));
-    opBuilder.addInput(numUpper.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new BatchMatrixBandPart<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> band() {
-    return band;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return band;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "BatchMatrixBandPart";
-  
-  private Output<T> band;
-  
-  private BatchMatrixBandPart(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    band = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixDeterminant.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixDeterminant.java
deleted file mode 100644
index a29c53c36c4..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixDeterminant.java
+++ /dev/null
@@ -1,72 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.linalg;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * @param <T> data type for {@code output()} output
- */
-@Operator(group = "linalg")
-public final class BatchMatrixDeterminant<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new BatchMatrixDeterminant operation.
-   * 
-   * @param scope current scope
-   * @param input 
-   * @return a new instance of BatchMatrixDeterminant
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> BatchMatrixDeterminant<T> create(Scope scope, Operand<T> input) {
-    OperationBuilder opBuilder = scope.env().opBuilder("BatchMatrixDeterminant", scope.makeOpName("BatchMatrixDeterminant"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new BatchMatrixDeterminant<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "BatchMatrixDeterminant";
-  
-  private Output<T> output;
-  
-  private BatchMatrixDeterminant(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixDiag.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixDiag.java
deleted file mode 100644
index 78d4860ca43..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixDiag.java
+++ /dev/null
@@ -1,72 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.linalg;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * @param <T> data type for {@code output()} output
- */
-@Operator(group = "linalg")
-public final class BatchMatrixDiag<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new BatchMatrixDiag operation.
-   * 
-   * @param scope current scope
-   * @param diagonal 
-   * @return a new instance of BatchMatrixDiag
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> BatchMatrixDiag<T> create(Scope scope, Operand<T> diagonal) {
-    OperationBuilder opBuilder = scope.env().opBuilder("BatchMatrixDiag", scope.makeOpName("BatchMatrixDiag"));
-    opBuilder.addInput(diagonal.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new BatchMatrixDiag<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "BatchMatrixDiag";
-  
-  private Output<T> output;
-  
-  private BatchMatrixDiag(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixDiagPart.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixDiagPart.java
deleted file mode 100644
index 9ba0bacf370..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixDiagPart.java
+++ /dev/null
@@ -1,72 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.linalg;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * @param <T> data type for {@code diagonal()} output
- */
-@Operator(group = "linalg")
-public final class BatchMatrixDiagPart<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new BatchMatrixDiagPart operation.
-   * 
-   * @param scope current scope
-   * @param input 
-   * @return a new instance of BatchMatrixDiagPart
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> BatchMatrixDiagPart<T> create(Scope scope, Operand<T> input) {
-    OperationBuilder opBuilder = scope.env().opBuilder("BatchMatrixDiagPart", scope.makeOpName("BatchMatrixDiagPart"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new BatchMatrixDiagPart<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> diagonal() {
-    return diagonal;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return diagonal;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "BatchMatrixDiagPart";
-  
-  private Output<T> diagonal;
-  
-  private BatchMatrixDiagPart(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    diagonal = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixInverse.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixInverse.java
deleted file mode 100644
index 1590430cfe6..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixInverse.java
+++ /dev/null
@@ -1,106 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.linalg;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * @param <T> data type for {@code output()} output
- */
-@Operator(group = "linalg")
-public final class BatchMatrixInverse<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.linalg.BatchMatrixInverse}
-   */
-  public static class Options {
-    
-    /**
-     * @param adjoint 
-     */
-    public Options adjoint(Boolean adjoint) {
-      this.adjoint = adjoint;
-      return this;
-    }
-    
-    private Boolean adjoint;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new BatchMatrixInverse operation.
-   * 
-   * @param scope current scope
-   * @param input 
-   * @param options carries optional attributes values
-   * @return a new instance of BatchMatrixInverse
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> BatchMatrixInverse<T> create(Scope scope, Operand<T> input, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("BatchMatrixInverse", scope.makeOpName("BatchMatrixInverse"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.adjoint != null) {
-          opBuilder.setAttr("adjoint", opts.adjoint);
-        }
-      }
-    }
-    return new BatchMatrixInverse<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param adjoint 
-   */
-  public static Options adjoint(Boolean adjoint) {
-    return new Options().adjoint(adjoint);
-  }
-  
-  /**
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "BatchMatrixInverse";
-  
-  private Output<T> output;
-  
-  private BatchMatrixInverse(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixSetDiag.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixSetDiag.java
deleted file mode 100644
index b3a4ad61eb5..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixSetDiag.java
+++ /dev/null
@@ -1,74 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.linalg;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * @param <T> data type for {@code output()} output
- */
-@Operator(group = "linalg")
-public final class BatchMatrixSetDiag<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new BatchMatrixSetDiag operation.
-   * 
-   * @param scope current scope
-   * @param input 
-   * @param diagonal 
-   * @return a new instance of BatchMatrixSetDiag
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> BatchMatrixSetDiag<T> create(Scope scope, Operand<T> input, Operand<T> diagonal) {
-    OperationBuilder opBuilder = scope.env().opBuilder("BatchMatrixSetDiag", scope.makeOpName("BatchMatrixSetDiag"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder.addInput(diagonal.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new BatchMatrixSetDiag<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "BatchMatrixSetDiag";
-  
-  private Output<T> output;
-  
-  private BatchMatrixSetDiag(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixSolve.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixSolve.java
deleted file mode 100644
index 6ebfa446613..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixSolve.java
+++ /dev/null
@@ -1,108 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.linalg;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * @param <T> data type for {@code output()} output
- */
-@Operator(group = "linalg")
-public final class BatchMatrixSolve<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.linalg.BatchMatrixSolve}
-   */
-  public static class Options {
-    
-    /**
-     * @param adjoint 
-     */
-    public Options adjoint(Boolean adjoint) {
-      this.adjoint = adjoint;
-      return this;
-    }
-    
-    private Boolean adjoint;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new BatchMatrixSolve operation.
-   * 
-   * @param scope current scope
-   * @param matrix 
-   * @param rhs 
-   * @param options carries optional attributes values
-   * @return a new instance of BatchMatrixSolve
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> BatchMatrixSolve<T> create(Scope scope, Operand<T> matrix, Operand<T> rhs, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("BatchMatrixSolve", scope.makeOpName("BatchMatrixSolve"));
-    opBuilder.addInput(matrix.asOutput(scope));
-    opBuilder.addInput(rhs.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.adjoint != null) {
-          opBuilder.setAttr("adjoint", opts.adjoint);
-        }
-      }
-    }
-    return new BatchMatrixSolve<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param adjoint 
-   */
-  public static Options adjoint(Boolean adjoint) {
-    return new Options().adjoint(adjoint);
-  }
-  
-  /**
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "BatchMatrixSolve";
-  
-  private Output<T> output;
-  
-  private BatchMatrixSolve(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixSolveLs.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixSolveLs.java
deleted file mode 100644
index 722c3a5cf09..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixSolveLs.java
+++ /dev/null
@@ -1,111 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.linalg;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat64;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * @param <T> data type for {@code output()} output
- */
-@Operator(group = "linalg")
-public final class BatchMatrixSolveLs<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.linalg.BatchMatrixSolveLs}
-   */
-  public static class Options {
-    
-    /**
-     * @param fast 
-     */
-    public Options fast(Boolean fast) {
-      this.fast = fast;
-      return this;
-    }
-    
-    private Boolean fast;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new BatchMatrixSolveLs operation.
-   * 
-   * @param scope current scope
-   * @param matrix 
-   * @param rhs 
-   * @param l2Regularizer 
-   * @param options carries optional attributes values
-   * @return a new instance of BatchMatrixSolveLs
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> BatchMatrixSolveLs<T> create(Scope scope, Operand<T> matrix, Operand<T> rhs, Operand<TFloat64> l2Regularizer, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("BatchMatrixSolveLs", scope.makeOpName("BatchMatrixSolveLs"));
-    opBuilder.addInput(matrix.asOutput(scope));
-    opBuilder.addInput(rhs.asOutput(scope));
-    opBuilder.addInput(l2Regularizer.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.fast != null) {
-          opBuilder.setAttr("fast", opts.fast);
-        }
-      }
-    }
-    return new BatchMatrixSolveLs<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param fast 
-   */
-  public static Options fast(Boolean fast) {
-    return new Options().fast(fast);
-  }
-  
-  /**
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "BatchMatrixSolveLs";
-  
-  private Output<T> output;
-  
-  private BatchMatrixSolveLs(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixTriangularSolve.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixTriangularSolve.java
deleted file mode 100644
index 0062969a197..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixTriangularSolve.java
+++ /dev/null
@@ -1,127 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.linalg;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * @param <T> data type for {@code output()} output
- */
-@Operator(group = "linalg")
-public final class BatchMatrixTriangularSolve<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.linalg.BatchMatrixTriangularSolve}
-   */
-  public static class Options {
-    
-    /**
-     * @param lower 
-     */
-    public Options lower(Boolean lower) {
-      this.lower = lower;
-      return this;
-    }
-    
-    /**
-     * @param adjoint 
-     */
-    public Options adjoint(Boolean adjoint) {
-      this.adjoint = adjoint;
-      return this;
-    }
-    
-    private Boolean lower;
-    private Boolean adjoint;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new BatchMatrixTriangularSolve operation.
-   * 
-   * @param scope current scope
-   * @param matrix 
-   * @param rhs 
-   * @param options carries optional attributes values
-   * @return a new instance of BatchMatrixTriangularSolve
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> BatchMatrixTriangularSolve<T> create(Scope scope, Operand<T> matrix, Operand<T> rhs, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("BatchMatrixTriangularSolve", scope.makeOpName("BatchMatrixTriangularSolve"));
-    opBuilder.addInput(matrix.asOutput(scope));
-    opBuilder.addInput(rhs.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.lower != null) {
-          opBuilder.setAttr("lower", opts.lower);
-        }
-        if (opts.adjoint != null) {
-          opBuilder.setAttr("adjoint", opts.adjoint);
-        }
-      }
-    }
-    return new BatchMatrixTriangularSolve<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param lower 
-   */
-  public static Options lower(Boolean lower) {
-    return new Options().lower(lower);
-  }
-  
-  /**
-   * @param adjoint 
-   */
-  public static Options adjoint(Boolean adjoint) {
-    return new Options().adjoint(adjoint);
-  }
-  
-  /**
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "BatchMatrixTriangularSolve";
-  
-  private Output<T> output;
-  
-  private BatchMatrixTriangularSolve(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchSelfAdjointEig.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchSelfAdjointEig.java
deleted file mode 100644
index b9e5dd053c5..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchSelfAdjointEig.java
+++ /dev/null
@@ -1,109 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.linalg;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * @param <T> data type for {@code e()} output
- */
-@Operator(group = "linalg")
-public final class BatchSelfAdjointEig<T extends TNumber> extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.linalg.BatchSelfAdjointEig}
-   */
-  public static class Options {
-    
-    /**
-     * @param computeV 
-     */
-    public Options computeV(Boolean computeV) {
-      this.computeV = computeV;
-      return this;
-    }
-    
-    private Boolean computeV;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new BatchSelfAdjointEig operation.
-   * 
-   * @param scope current scope
-   * @param input 
-   * @param options carries optional attributes values
-   * @return a new instance of BatchSelfAdjointEig
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> BatchSelfAdjointEig<T> create(Scope scope, Operand<T> input, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("BatchSelfAdjointEigV2", scope.makeOpName("BatchSelfAdjointEig"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.computeV != null) {
-          opBuilder.setAttr("compute_v", opts.computeV);
-        }
-      }
-    }
-    return new BatchSelfAdjointEig<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param computeV 
-   */
-  public static Options computeV(Boolean computeV) {
-    return new Options().computeV(computeV);
-  }
-  
-  /**
-   */
-  public Output<T> e() {
-    return e;
-  }
-  
-  /**
-   */
-  public Output<T> v() {
-    return v;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "BatchSelfAdjointEigV2";
-  
-  private Output<T> e;
-  private Output<T> v;
-  
-  private BatchSelfAdjointEig(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    e = operation.output(outputIdx++);
-    v = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchSvd.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchSvd.java
deleted file mode 100644
index 76ed6341fce..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchSvd.java
+++ /dev/null
@@ -1,136 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.linalg;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * @param <T> data type for {@code s()} output
- */
-@Operator(group = "linalg")
-public final class BatchSvd<T extends TType> extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.linalg.BatchSvd}
-   */
-  public static class Options {
-    
-    /**
-     * @param computeUv 
-     */
-    public Options computeUv(Boolean computeUv) {
-      this.computeUv = computeUv;
-      return this;
-    }
-    
-    /**
-     * @param fullMatrices 
-     */
-    public Options fullMatrices(Boolean fullMatrices) {
-      this.fullMatrices = fullMatrices;
-      return this;
-    }
-    
-    private Boolean computeUv;
-    private Boolean fullMatrices;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new BatchSvd operation.
-   * 
-   * @param scope current scope
-   * @param input 
-   * @param options carries optional attributes values
-   * @return a new instance of BatchSvd
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> BatchSvd<T> create(Scope scope, Operand<T> input, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("BatchSvd", scope.makeOpName("BatchSvd"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.computeUv != null) {
-          opBuilder.setAttr("compute_uv", opts.computeUv);
-        }
-        if (opts.fullMatrices != null) {
-          opBuilder.setAttr("full_matrices", opts.fullMatrices);
-        }
-      }
-    }
-    return new BatchSvd<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param computeUv 
-   */
-  public static Options computeUv(Boolean computeUv) {
-    return new Options().computeUv(computeUv);
-  }
-  
-  /**
-   * @param fullMatrices 
-   */
-  public static Options fullMatrices(Boolean fullMatrices) {
-    return new Options().fullMatrices(fullMatrices);
-  }
-  
-  /**
-   */
-  public Output<T> s() {
-    return s;
-  }
-  
-  /**
-   */
-  public Output<T> u() {
-    return u;
-  }
-  
-  /**
-   */
-  public Output<T> v() {
-    return v;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "BatchSvd";
-  
-  private Output<T> s;
-  private Output<T> u;
-  private Output<T> v;
-  
-  private BatchSvd(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    s = operation.output(outputIdx++);
-    u = operation.output(outputIdx++);
-    v = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Cholesky.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Cholesky.java
deleted file mode 100644
index c38d8b21256..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Cholesky.java
+++ /dev/null
@@ -1,89 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.linalg;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Computes the Cholesky decomposition of one or more square matrices.
- * <p>
- * The input is a tensor of shape `[..., M, M]` whose inner-most 2 dimensions
- * form square matrices.
- * <p>
- * The input has to be symmetric and positive definite. Only the lower-triangular
- * part of the input will be used for this operation. The upper-triangular part
- * will not be read.
- * <p>
- * The output is a tensor of the same shape as the input
- * containing the Cholesky decompositions for all input submatrices `[..., :, :]`.
- * <p>
- * <b>Note</b>: The gradient computation on GPU is faster for large matrices but
- * not for large batch dimensions when the submatrices are small. In this
- * case it might be faster to use the CPU.
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator(group = "linalg")
-public final class Cholesky<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new Cholesky operation.
-   * 
-   * @param scope current scope
-   * @param input Shape is `[..., M, M]`.
-   * @return a new instance of Cholesky
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> Cholesky<T> create(Scope scope, Operand<T> input) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Cholesky", scope.makeOpName("Cholesky"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new Cholesky<T>(opBuilder.build());
-  }
-  
-  /**
-   * Shape is `[..., M, M]`.
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Cholesky";
-  
-  private Output<T> output;
-  
-  private Cholesky(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/CholeskyGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/CholeskyGrad.java
deleted file mode 100644
index e73fa93b7b2..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/CholeskyGrad.java
+++ /dev/null
@@ -1,84 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.linalg;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Computes the reverse mode backpropagated gradient of the Cholesky algorithm.
- * <p>
- * For an explanation see "Differentiation of the Cholesky algorithm" by
- * Iain Murray http://arxiv.org/abs/1602.07527.
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator(group = "linalg")
-public final class CholeskyGrad<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new CholeskyGrad operation.
-   * 
-   * @param scope current scope
-   * @param l Output of batch Cholesky algorithm l = cholesky(A). Shape is `[..., M, M]`.
-   * Algorithm depends only on lower triangular part of the innermost matrices of
-   * this tensor.
-   * @param grad df/dl where f is some scalar function. Shape is `[..., M, M]`.
-   * Algorithm depends only on lower triangular part of the innermost matrices of
-   * this tensor.
-   * @return a new instance of CholeskyGrad
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> CholeskyGrad<T> create(Scope scope, Operand<T> l, Operand<T> grad) {
-    OperationBuilder opBuilder = scope.env().opBuilder("CholeskyGrad", scope.makeOpName("CholeskyGrad"));
-    opBuilder.addInput(l.asOutput(scope));
-    opBuilder.addInput(grad.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new CholeskyGrad<T>(opBuilder.build());
-  }
-  
-  /**
-   * Symmetrized version of df/dA . Shape is `[..., M, M]`
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "CholeskyGrad";
-  
-  private Output<T> output;
-  
-  private CholeskyGrad(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/ConjugateTranspose.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/ConjugateTranspose.java
deleted file mode 100644
index b6219b24a6a..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/ConjugateTranspose.java
+++ /dev/null
@@ -1,81 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.linalg;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * Shuffle dimensions of x according to a permutation and conjugate the result.
- * <p>
- * The output `y` has the same rank as `x`. The shapes of `x` and `y` satisfy:
- *   `y.shape[i] == x.shape[perm[i]] for i in [0, 1, ..., rank(x) - 1]`
- *   `y[i,j,k,...,s,t,u] == conj(x[perm[i], perm[j], perm[k],...,perm[s], perm[t], perm[u]])`
- * 
- * @param <T> data type for {@code y()} output
- */
-@Operator(group = "linalg")
-public final class ConjugateTranspose<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new ConjugateTranspose operation.
-   * 
-   * @param scope current scope
-   * @param x 
-   * @param perm 
-   * @return a new instance of ConjugateTranspose
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> ConjugateTranspose<T> create(Scope scope, Operand<T> x, Operand<U> perm) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ConjugateTranspose", scope.makeOpName("ConjugateTranspose"));
-    opBuilder.addInput(x.asOutput(scope));
-    opBuilder.addInput(perm.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new ConjugateTranspose<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> y() {
-    return y;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return y;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ConjugateTranspose";
-  
-  private Output<T> y;
-  
-  private ConjugateTranspose(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    y = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Cross.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Cross.java
deleted file mode 100644
index 1ab0e9b3c0f..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Cross.java
+++ /dev/null
@@ -1,81 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.linalg;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Compute the pairwise cross product.
- * <p>
- * `a` and `b` must be the same shape; they can either be simple 3-element vectors,
- * or any shape where the innermost dimension is 3. In the latter case, each pair
- * of corresponding 3-element vectors is cross-multiplied independently.
- * 
- * @param <T> data type for {@code product()} output
- */
-@Operator(group = "linalg")
-public final class Cross<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new Cross operation.
-   * 
-   * @param scope current scope
-   * @param a A tensor containing 3-element vectors.
-   * @param b Another tensor, of same type and shape as `a`.
-   * @return a new instance of Cross
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> Cross<T> create(Scope scope, Operand<T> a, Operand<T> b) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Cross", scope.makeOpName("Cross"));
-    opBuilder.addInput(a.asOutput(scope));
-    opBuilder.addInput(b.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new Cross<T>(opBuilder.build());
-  }
-  
-  /**
-   * Pairwise cross product of the vectors in `a` and `b`.
-   */
-  public Output<T> product() {
-    return product;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return product;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Cross";
-  
-  private Output<T> product;
-  
-  private Cross(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    product = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Det.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Det.java
deleted file mode 100644
index d4f7dfe0eab..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Det.java
+++ /dev/null
@@ -1,79 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.linalg;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Computes the determinant of one or more square matrices.
- * <p>
- * The input is a tensor of shape `[..., M, M]` whose inner-most 2 dimensions
- * form square matrices. The output is a tensor containing the determinants
- * for all input submatrices `[..., :, :]`.
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator(group = "linalg")
-public final class Det<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new Det operation.
-   * 
-   * @param scope current scope
-   * @param input Shape is `[..., M, M]`.
-   * @return a new instance of Det
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> Det<T> create(Scope scope, Operand<T> input) {
-    OperationBuilder opBuilder = scope.env().opBuilder("MatrixDeterminant", scope.makeOpName("Det"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new Det<T>(opBuilder.build());
-  }
-  
-  /**
-   * Shape is `[...]`.
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "MatrixDeterminant";
-  
-  private Output<T> output;
-  
-  private Det(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Eig.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Eig.java
deleted file mode 100644
index 5e84611590b..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Eig.java
+++ /dev/null
@@ -1,129 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.linalg;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Computes the eigen decomposition of one or more square matrices.
- * <p>
- * Computes the eigenvalues and (optionally) right eigenvectors of each inner matrix in
- * `input` such that `input[..., :, :] = v[..., :, :] * diag(e[..., :])`. The eigenvalues
- * are sorted in non-decreasing order.
- * <pre>{@code
- * # a is a tensor.
- * # e is a tensor of eigenvalues.
- * # v is a tensor of eigenvectors.
- * e, v = eig(a)
- * e = eig(a, compute_v=False)
- * }</pre>
- * 
- * 
- * @param <U> data type for {@code e()} output
- */
-@Operator(group = "linalg")
-public final class Eig<U extends TType> extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.linalg.Eig}
-   */
-  public static class Options {
-    
-    /**
-     * @param computeV If `True` then eigenvectors will be computed and returned in `v`.
-     * Otherwise, only the eigenvalues will be computed.
-     */
-    public Options computeV(Boolean computeV) {
-      this.computeV = computeV;
-      return this;
-    }
-    
-    private Boolean computeV;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new Eig operation.
-   * 
-   * @param scope current scope
-   * @param input `Tensor` input of shape `[N, N]`.
-   * @param Tout 
-   * @param options carries optional attributes values
-   * @return a new instance of Eig
-   */
-  @Endpoint(describeByClass = true)
-  public static <U extends TType, T extends TType> Eig<U> create(Scope scope, Operand<T> input, Class<U> Tout, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Eig", scope.makeOpName("Eig"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("Tout", Tout);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.computeV != null) {
-          opBuilder.setAttr("compute_v", opts.computeV);
-        }
-      }
-    }
-    return new Eig<U>(opBuilder.build());
-  }
-  
-  /**
-   * @param computeV If `True` then eigenvectors will be computed and returned in `v`.
-   * Otherwise, only the eigenvalues will be computed.
-   */
-  public static Options computeV(Boolean computeV) {
-    return new Options().computeV(computeV);
-  }
-  
-  /**
-   * Eigenvalues. Shape is `[N]`.
-   */
-  public Output<U> e() {
-    return e;
-  }
-  
-  /**
-   * Eigenvectors. Shape is `[N, N]`.
-   */
-  public Output<U> v() {
-    return v;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Eig";
-  
-  private Output<U> e;
-  private Output<U> v;
-  
-  private Eig(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    e = operation.output(outputIdx++);
-    v = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Einsum.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Einsum.java
deleted file mode 100644
index 0e3091d05f2..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Einsum.java
+++ /dev/null
@@ -1,155 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.linalg;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.Operands;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Tensor contraction according to Einstein summation convention.
- * <p>
- * Implements generalized Tensor contraction and reduction. Each input Tensor must
- * have a corresponding input subscript appearing in the comma-separated left-hand
- * side of the equation. The right-hand side of the equation consists of the
- * output subscript. The input subscripts and the output subscript should consist
- * of zero or more named axis labels and at most one ellipsis (`...`).
- * <p>
- * The named axis labels may be any single character other than those having
- * special meaning, namely `,.->`. The behavior of this Op is undefined if it
- * receives an ill-formatted equation; since the validation is done at
- * graph-building time, we omit format validation checks at runtime.
- * <p>
- * Note: This Op is <i>not</i> intended to be called by the user; instead users should
- * call `tf.einsum` directly. It is a hidden Op used by `tf.einsum`.
- * <p>
- * Operations are applied to the input(s) according to the following rules:
- * <p>
- *  (a) Generalized Diagonals: For input dimensions corresponding to axis labels
- *      appearing more than once in the same input subscript, we take the
- *      generalized (`k`-dimensional) diagonal.
- *      For example, in the equation `iii->i` with input shape `[3, 3, 3]`, the
- *      generalized diagonal would consist of `3` elements at indices `(0, 0, 0)`,
- *      `(1, 1, 1)` and `(2, 2, 2)` to create a Tensor of shape `[3]`.
- * <p>
- *  (b) Reduction: Axes corresponding to labels appearing only in one input
- *      subscript but not in the output subscript are summed over prior to Tensor
- *      contraction.
- *      For example, in the equation `ab,bc->b`, the axis labels `a` and `c` are
- *      the reduction axis labels.
- * <p>
- *  (c) Batch Dimensions: Axes corresponding to labels appearing in each of the
- *      input subscripts and also in the output subscript make up the batch
- *      dimensions in Tensor contraction. Unnamed axis labels corresponding to
- *      ellipsis (`...`) also correspond to batch dimensions.
- *      For example, for the equation denoting batch matrix multiplication,
- *      `bij,bjk->bik`, the axis label `b` corresponds to a batch dimension.
- * <p>
- *  (d) Contraction: In case of binary einsum, axes corresponding to labels
- *      appearing in two different inputs (and not in the output) are contracted
- *      against each other.
- *      Considering the batch matrix multiplication equation again
- *      (`bij,bjk->bik`), the contracted axis label is `j`.
- * <p>
- *  (e) Expand Diagonal: If the output subscripts contain repeated (explicit) axis
- *      labels, the opposite operation of (a) is applied. For example, in the
- *      equation `i->iii`, and input shape `[3]`, the output of shape `[3, 3, 3]`
- *      are all zeros, except for the (generalized) diagonal which is populated
- *      with values from the input.
- *      Note: This operation is not supported by `np.einsum` or `tf.einsum`; it is
- *      provided to enable computing the symbolic gradient of `tf.einsum`.
- * <p>
- * The output subscripts must contain only labels appearing in at least one of the
- * input subscripts. Furthermore, all dimensions mapping to the same axis label
- * must be equal.
- * <p>
- * Any of the input and output subscripts may contain at most a single ellipsis
- * (`...`). These ellipsis are mapped against dimensions not corresponding to any
- * named axis label. If two inputs contain ellipsis, then they are broadcasted
- * according to standard NumPy broadcasting
- * [rules](http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html).
- * <p>
- * The broadcasted dimensions are placed in the corresponding location of the
- * ellipsis in the output subscript. If the broadcasted dimensions are non-empty
- * and the output subscripts do not contain ellipsis, then an InvalidArgument error
- * is raised.
- * <p>
- * @compatibility(numpy)
- * Similar to [`numpy.einsum`](https://docs.scipy.org/doc/numpy/reference/generated/numpy.einsum.html).
- * <p>
- * Comparison with `numpy.einsum`:
- * <p>
- *  * This Op only supports unary and binary forms of `numpy.einsum`.
- *  * This Op does not support implicit form. (i.e. equations without `->`).
- *  * This Op also supports repeated indices in the output subscript, which is not
- *    supported by `numpy.einsum`.
- * @end_compatibility
- * 
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator(group = "linalg")
-public final class Einsum<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new Einsum operation.
-   * 
-   * @param scope current scope
-   * @param inputs List of 1 or 2 Tensors.
-   * @param equation String describing the Einstein Summation operation; in the format of np.einsum.
-   * @return a new instance of Einsum
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> Einsum<T> create(Scope scope, Iterable<Operand<T>> inputs, String equation) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Einsum", scope.makeOpName("Einsum"));
-    opBuilder.addInputList(Operands.asOutputs(scope, inputs));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("equation", equation);
-    return new Einsum<T>(opBuilder.build());
-  }
-  
-  /**
-   * Output Tensor with shape depending upon `equation`.
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Einsum";
-  
-  private Output<T> output;
-  
-  private Einsum(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/EuclideanNorm.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/EuclideanNorm.java
deleted file mode 100644
index 62dae10d44b..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/EuclideanNorm.java
+++ /dev/null
@@ -1,118 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.linalg;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * Computes the euclidean norm of elements across dimensions of a tensor.
- * <p>
- * Reduces `input` along the dimensions given in `axis`. Unless
- * `keep_dims` is true, the rank of the tensor is reduced by 1 for each entry in
- * `axis`. If `keep_dims` is true, the reduced dimensions are
- * retained with length 1.
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator(group = "linalg")
-public final class EuclideanNorm<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.linalg.EuclideanNorm}
-   */
-  public static class Options {
-    
-    /**
-     * @param keepDims If true, retain reduced dimensions with length 1.
-     */
-    public Options keepDims(Boolean keepDims) {
-      this.keepDims = keepDims;
-      return this;
-    }
-    
-    private Boolean keepDims;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new EuclideanNorm operation.
-   * 
-   * @param scope current scope
-   * @param input The tensor to reduce.
-   * @param axis The dimensions to reduce. Must be in the range
-   * `[-rank(input), rank(input))`.
-   * @param options carries optional attributes values
-   * @return a new instance of EuclideanNorm
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> EuclideanNorm<T> create(Scope scope, Operand<T> input, Operand<U> axis, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("EuclideanNorm", scope.makeOpName("EuclideanNorm"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder.addInput(axis.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.keepDims != null) {
-          opBuilder.setAttr("keep_dims", opts.keepDims);
-        }
-      }
-    }
-    return new EuclideanNorm<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param keepDims If true, retain reduced dimensions with length 1.
-   */
-  public static Options keepDims(Boolean keepDims) {
-    return new Options().keepDims(keepDims);
-  }
-  
-  /**
-   * The reduced tensor.
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "EuclideanNorm";
-  
-  private Output<T> output;
-  
-  private EuclideanNorm(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Inv.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Inv.java
deleted file mode 100644
index dca78a8b23c..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Inv.java
+++ /dev/null
@@ -1,125 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.linalg;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Computes the inverse of one or more square invertible matrices or their
- * <p>
- * adjoints (conjugate transposes).
- * <p>
- * The input is a tensor of shape `[..., M, M]` whose inner-most 2 dimensions
- * form square matrices. The output is a tensor of the same shape as the input
- * containing the inverse for all input submatrices `[..., :, :]`.
- * <p>
- * The op uses LU decomposition with partial pivoting to compute the inverses.
- * <p>
- * If a matrix is not invertible there is no guarantee what the op does. It
- * may detect the condition and raise an exception or it may simply return a
- * garbage result.
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator(group = "linalg")
-public final class Inv<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.linalg.Inv}
-   */
-  public static class Options {
-    
-    /**
-     * @param adjoint 
-     */
-    public Options adjoint(Boolean adjoint) {
-      this.adjoint = adjoint;
-      return this;
-    }
-    
-    private Boolean adjoint;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new Inv operation.
-   * 
-   * @param scope current scope
-   * @param input Shape is `[..., M, M]`.
-   * @param options carries optional attributes values
-   * @return a new instance of Inv
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> Inv<T> create(Scope scope, Operand<T> input, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("MatrixInverse", scope.makeOpName("Inv"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.adjoint != null) {
-          opBuilder.setAttr("adjoint", opts.adjoint);
-        }
-      }
-    }
-    return new Inv<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param adjoint 
-   */
-  public static Options adjoint(Boolean adjoint) {
-    return new Options().adjoint(adjoint);
-  }
-  
-  /**
-   * Shape is `[..., M, M]`.
-   * <p>
-   * @compatibility(numpy)
-   * Equivalent to np.linalg.inv
-   * @end_compatibility
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "MatrixInverse";
-  
-  private Output<T> output;
-  
-  private Inv(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/LoadAndRemapMatrix.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/LoadAndRemapMatrix.java
deleted file mode 100644
index 31a4d445dc8..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/LoadAndRemapMatrix.java
+++ /dev/null
@@ -1,175 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.linalg;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.TString;
-
-/**
- * Loads a 2-D (matrix) `Tensor` with name `old_tensor_name` from the checkpoint
- * <p>
- * at `ckpt_path` and potentially reorders its rows and columns using the
- * specified remappings.
- * <p>
- * Most users should use one of the wrapper initializers (such as
- * `tf.contrib.framework.load_and_remap_matrix_initializer`) instead of this
- * function directly.
- * <p>
- * The remappings are 1-D tensors with the following properties:
- * <ul>
- * <li>
- * `row_remapping` must have exactly `num_rows` entries. Row `i` of the output
- *   matrix will be initialized from the row corresponding to index
- *   `row_remapping[i]` in the old `Tensor` from the checkpoint.
- * </li>
- * <li>
- * `col_remapping` must have either 0 entries (indicating that no column
- *   reordering is needed) or `num_cols` entries. If specified, column `j` of the
- *   output matrix will be initialized from the column corresponding to index
- *   `col_remapping[j]` in the old `Tensor` from the checkpoint.
- * </li>
- * <li>
- * A value of -1 in either of the remappings signifies a "missing" entry. In that
- *   case, values from the `initializing_values` tensor will be used to fill that
- *   missing row or column. If `row_remapping` has `r` missing entries and
- *   `col_remapping` has `c` missing entries, then the following condition must be
- *   true:
- * </li>
- * </ul>
- * `(r * num_cols) + (c * num_rows) - (r * c) == len(initializing_values)`
- * <p>
- * The remapping tensors can be generated using the GenerateVocabRemapping op.
- * <p>
- * As an example, with row_remapping = [1, 0, -1], col_remapping = [0, 2, -1],
- * initializing_values = [0.5, -0.5, 0.25, -0.25, 42], and w(i, j) representing
- * the value from row i, column j of the old tensor in the checkpoint, the output
- * matrix will look like the following:
- * <p>
- * [[w(1, 0),  w(1, 2),  0.5],
- *  [w(0, 0),  w(0, 2), -0.5],
- *  [0.25,    -0.25,      42]]
- */
-@Operator(group = "linalg")
-public final class LoadAndRemapMatrix extends RawOp implements Operand<TFloat32> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.linalg.LoadAndRemapMatrix}
-   */
-  public static class Options {
-    
-    /**
-     * @param maxRowsInMemory The maximum number of rows to load from the checkpoint at
-     * once. If less than or equal to 0, the entire matrix will be loaded into
-     * memory. Setting this arg trades increased disk reads for lower memory usage.
-     */
-    public Options maxRowsInMemory(Long maxRowsInMemory) {
-      this.maxRowsInMemory = maxRowsInMemory;
-      return this;
-    }
-    
-    private Long maxRowsInMemory;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new LoadAndRemapMatrix operation.
-   * 
-   * @param scope current scope
-   * @param ckptPath Path to the TensorFlow checkpoint (version 2, `TensorBundle`) from
-   * which the old matrix `Tensor` will be loaded.
-   * @param oldTensorName Name of the 2-D `Tensor` to load from checkpoint.
-   * @param rowRemapping An int `Tensor` of row remappings (generally created by
-   * `generate_vocab_remapping`).  Even if no row remapping is needed, this must
-   * still be an index-valued Tensor (e.g. [0, 1, 2, ...]), or a shifted
-   * index-valued `Tensor` (e.g. [8, 9, 10, ...], for partitioned `Variables`).
-   * @param colRemapping An int `Tensor` of column remappings (generally created by
-   * `generate_vocab_remapping`).  May be a size-0 `Tensor` if only row remapping
-   * is to be done (e.g. column ordering is the same).
-   * @param initializingValues A float `Tensor` containing  values to fill in for cells
-   * in the output matrix that are not loaded from the checkpoint. Length must be
-   * exactly the same as the number of missing / new cells.
-   * @param numRows Number of rows (length of the 1st dimension) in the output matrix.
-   * @param numCols Number of columns (length of the 2nd dimension) in the output matrix.
-   * @param options carries optional attributes values
-   * @return a new instance of LoadAndRemapMatrix
-   */
-  @Endpoint(describeByClass = true)
-  public static LoadAndRemapMatrix create(Scope scope, Operand<TString> ckptPath, Operand<TString> oldTensorName, Operand<TInt64> rowRemapping, Operand<TInt64> colRemapping, Operand<TFloat32> initializingValues, Long numRows, Long numCols, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("LoadAndRemapMatrix", scope.makeOpName("LoadAndRemapMatrix"));
-    opBuilder.addInput(ckptPath.asOutput(scope));
-    opBuilder.addInput(oldTensorName.asOutput(scope));
-    opBuilder.addInput(rowRemapping.asOutput(scope));
-    opBuilder.addInput(colRemapping.asOutput(scope));
-    opBuilder.addInput(initializingValues.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("num_rows", numRows);
-    opBuilder.setAttr("num_cols", numCols);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.maxRowsInMemory != null) {
-          opBuilder.setAttr("max_rows_in_memory", opts.maxRowsInMemory);
-        }
-      }
-    }
-    return new LoadAndRemapMatrix(opBuilder.build());
-  }
-  
-  /**
-   * @param maxRowsInMemory The maximum number of rows to load from the checkpoint at
-   * once. If less than or equal to 0, the entire matrix will be loaded into
-   * memory. Setting this arg trades increased disk reads for lower memory usage.
-   */
-  public static Options maxRowsInMemory(Long maxRowsInMemory) {
-    return new Options().maxRowsInMemory(maxRowsInMemory);
-  }
-  
-  /**
-   * Output matrix containing existing values loaded from the
-   * checkpoint, and with any missing values filled in from initializing_values.
-   */
-  public Output<TFloat32> outputMatrix() {
-    return outputMatrix;
-  }
-  
-  @Override
-  public Output<TFloat32> asOutput(Scope scope) {
-    return outputMatrix;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "LoadAndRemapMatrix";
-  
-  private Output<TFloat32> outputMatrix;
-  
-  private LoadAndRemapMatrix(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    outputMatrix = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/LogMatrixDeterminant.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/LogMatrixDeterminant.java
deleted file mode 100644
index de1e4a08bb1..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/LogMatrixDeterminant.java
+++ /dev/null
@@ -1,90 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.linalg;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Computes the sign and the log of the absolute value of the determinant of
- * <p>
- * one or more square matrices.
- * <p>
- * The input is a tensor of shape `[N, M, M]` whose inner-most 2 dimensions
- * form square matrices. The outputs are two tensors containing the signs and
- * absolute values of the log determinants for all N input submatrices
- * `[..., :, :]` such that the determinant = sign*exp(log_abs_determinant).
- * The log_abs_determinant is computed as det(P)*sum(log(diag(LU))) where LU
- * is the LU decomposition of the input and P is the corresponding
- * permutation matrix.
- * 
- * @param <T> data type for {@code sign()} output
- */
-@Operator(group = "linalg")
-public final class LogMatrixDeterminant<T extends TType> extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new LogMatrixDeterminant operation.
-   * 
-   * @param scope current scope
-   * @param input Shape is `[N, M, M]`.
-   * @return a new instance of LogMatrixDeterminant
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> LogMatrixDeterminant<T> create(Scope scope, Operand<T> input) {
-    OperationBuilder opBuilder = scope.env().opBuilder("LogMatrixDeterminant", scope.makeOpName("LogMatrixDeterminant"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new LogMatrixDeterminant<T>(opBuilder.build());
-  }
-  
-  /**
-   * The signs of the log determinants of the inputs. Shape is `[N]`.
-   */
-  public Output<T> sign() {
-    return sign;
-  }
-  
-  /**
-   * The logs of the absolute values of the determinants
-   * of the N input matrices.  Shape is `[N]`.
-   */
-  public Output<T> logAbsDeterminant() {
-    return logAbsDeterminant;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "LogMatrixDeterminant";
-  
-  private Output<T> sign;
-  private Output<T> logAbsDeterminant;
-  
-  private LogMatrixDeterminant(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    sign = operation.output(outputIdx++);
-    logAbsDeterminant = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Lu.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Lu.java
deleted file mode 100644
index abc37991725..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Lu.java
+++ /dev/null
@@ -1,125 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.linalg;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * Computes the LU decomposition of one or more square matrices.
- * <p>
- * The input is a tensor of shape `[..., M, M]` whose inner-most 2 dimensions
- * form square matrices.
- * <p>
- * The input has to be invertible.
- * <p>
- * The output consists of two tensors LU and P containing the LU decomposition
- * of all input submatrices `[..., :, :]`. LU encodes the lower triangular and
- * upper triangular factors.
- * <p>
- * For each input submatrix of shape `[M, M]`, L is a lower triangular matrix of
- * shape `[M, M]` with unit diagonal whose entries correspond to the strictly lower
- * triangular part of LU. U is a upper triangular matrix of shape `[M, M]` whose
- * entries correspond to the upper triangular part, including the diagonal, of LU.
- * <p>
- * P represents a permutation matrix encoded as a list of indices each between `0`
- * and `M-1`, inclusive. If P_mat denotes the permutation matrix corresponding to
- * P, then the L, U and P satisfies P_mat * input = L * U.
- * 
- * @param <T> data type for {@code lu()} output
- * @param <U> data type for {@code p()} output
- */
-@Operator(group = "linalg")
-public final class Lu<T extends TType, U extends TNumber> extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new Lu operation.
-   * 
-   * @param scope current scope
-   * @param input A tensor of shape `[..., M, M]` whose inner-most 2 dimensions form matrices of
-   * size `[M, M]`.
-   * @param outputIdxType 
-   * @return a new instance of Lu
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> Lu<T, U> create(Scope scope, Operand<T> input, Class<U> outputIdxType) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Lu", scope.makeOpName("Lu"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("output_idx_type", outputIdxType);
-    return new Lu<T, U>(opBuilder.build());
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new Lu operation using default output types.
-   * 
-   * @param scope current scope
-   * @param input A tensor of shape `[..., M, M]` whose inner-most 2 dimensions form matrices of
-   * size `[M, M]`.
-   * @return a new instance of Lu
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> Lu<T, TInt32> create(Scope scope, Operand<T> input) {
-    return create(scope, input, TInt32.class);
-  }
-  
-  /**
-   * A tensor of shape `[..., M, M]` whose strictly lower triangular part denotes the
-   * lower triangular factor `L` with unit diagonal, and whose upper triangular part
-   * denotes the upper triangular factor `U`.
-   */
-  public Output<T> lu() {
-    return lu;
-  }
-  
-  /**
-   * Permutation of the rows encoded as a list of indices in `0..M-1`. Shape is
-   * `[..., M]`.
-   * @compatibility(scipy)
-   * Similar to `scipy.linalg.lu`, except the triangular factors `L` and `U` are
-   * packed into a single tensor, the permutation is applied to `input` instead of
-   * the right hand side and the permutation `P` is returned as a list of indices
-   * instead of a permutation matrix.
-   * @end_compatibility
-   */
-  public Output<U> p() {
-    return p;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Lu";
-  
-  private Output<T> lu;
-  private Output<U> p;
-  
-  private Lu(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    lu = operation.output(outputIdx++);
-    p = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/MatMul.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/MatMul.java
deleted file mode 100644
index 700f03d4d39..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/MatMul.java
+++ /dev/null
@@ -1,137 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.linalg;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Multiply the matrix "a" by the matrix "b".
- * <p>
- * The inputs must be two-dimensional matrices and the inner dimension of
- * "a" (after being transposed if transpose_a is true) must match the
- * outer dimension of "b" (after being transposed if transposed_b is
- * true).
- * <p>
- * <i>Note</i>: The default kernel implementation for MatMul on GPUs uses
- * cublas.
- * 
- * @param <T> data type for {@code product()} output
- */
-@Operator(group = "linalg")
-public final class MatMul<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.linalg.MatMul}
-   */
-  public static class Options {
-    
-    /**
-     * @param transposeA If true, "a" is transposed before multiplication.
-     */
-    public Options transposeA(Boolean transposeA) {
-      this.transposeA = transposeA;
-      return this;
-    }
-    
-    /**
-     * @param transposeB If true, "b" is transposed before multiplication.
-     */
-    public Options transposeB(Boolean transposeB) {
-      this.transposeB = transposeB;
-      return this;
-    }
-    
-    private Boolean transposeA;
-    private Boolean transposeB;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new MatMul operation.
-   * 
-   * @param scope current scope
-   * @param a 
-   * @param b 
-   * @param options carries optional attributes values
-   * @return a new instance of MatMul
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> MatMul<T> create(Scope scope, Operand<T> a, Operand<T> b, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("MatMul", scope.makeOpName("MatMul"));
-    opBuilder.addInput(a.asOutput(scope));
-    opBuilder.addInput(b.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.transposeA != null) {
-          opBuilder.setAttr("transpose_a", opts.transposeA);
-        }
-        if (opts.transposeB != null) {
-          opBuilder.setAttr("transpose_b", opts.transposeB);
-        }
-      }
-    }
-    return new MatMul<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param transposeA If true, "a" is transposed before multiplication.
-   */
-  public static Options transposeA(Boolean transposeA) {
-    return new Options().transposeA(transposeA);
-  }
-  
-  /**
-   * @param transposeB If true, "b" is transposed before multiplication.
-   */
-  public static Options transposeB(Boolean transposeB) {
-    return new Options().transposeB(transposeB);
-  }
-  
-  /**
-   */
-  public Output<T> product() {
-    return product;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return product;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "MatMul";
-  
-  private Output<T> product;
-  
-  private MatMul(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    product = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/MatrixDiag.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/MatrixDiag.java
deleted file mode 100644
index 621564ab1be..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/MatrixDiag.java
+++ /dev/null
@@ -1,177 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.linalg;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.family.TType;
-
-/**
- * Returns a batched diagonal tensor with given batched diagonal values.
- * <p>
- * Returns a tensor with the contents in `diagonal` as `k[0]`-th to `k[1]`-th
- * diagonals of a matrix, with everything else padded with `padding`. `num_rows`
- * and `num_cols` specify the dimension of the innermost matrix of the output. If
- * both are not specified, the op assumes the innermost matrix is square and infers
- * its size from `k` and the innermost dimension of `diagonal`. If only one of them
- * is specified, the op assumes the unspecified value is the smallest possible
- * based on other criteria.
- * <p>
- * Let `diagonal` have `r` dimensions `[I, J, ..., L, M, N]`. The output tensor has
- * rank `r+1` with shape `[I, J, ..., L, M, num_rows, num_cols]` when only one
- * diagonal is given (`k` is an integer or `k[0] == k[1]`). Otherwise, it has rank
- * `r` with shape `[I, J, ..., L, num_rows, num_cols]`.
- * <p>
- * The second innermost dimension of `diagonal` has double meaning.
- * When `k` is scalar or `k[0] == k[1]`, `M` is part of the batch size
- * [I, J, ..., M], and the output tensor is:
- * <pre>{@code
- * output[i, j, ..., l, m, n]
- *   = diagonal[i, j, ..., l, n-max(d_upper, 0)] ; if n - m == d_upper
- *     padding_value                             ; otherwise
- * }</pre>
- * Otherwise, `M` is treated as the number of diagonals for the matrix in the
- * same batch (`M = k[1]-k[0]+1`), and the output tensor is:
- * <pre>{@code
- * output[i, j, ..., l, m, n]
- *   = diagonal[i, j, ..., l, diag_index, index_in_diag] ; if k[0] <= d <= k[1]
- *     padding_value                                     ; otherwise
- * }</pre>
- * where `d = n - m`, `diag_index = k[1] - d`, and `index_in_diag = n - max(d, 0)`.
- * <p>
- * For example:
- * <pre>{@code
- * # The main diagonal.
- * diagonal = np.array([[1, 2, 3, 4],            # Input shape: (2, 4)
- *                      [5, 6, 7, 8]])
- * tf.matrix_diag(diagonal) ==> [[[1, 0, 0, 0],  # Output shape: (2, 4, 4)
- *                                [0, 2, 0, 0],
- *                                [0, 0, 3, 0],
- *                                [0, 0, 0, 4]],
- *                               [[5, 0, 0, 0],
- *                                [0, 6, 0, 0],
- *                                [0, 0, 7, 0],
- *                                [0, 0, 0, 8]]]
- * 
- * # A superdiagonal (per batch).
- * diagonal = np.array([[1, 2, 3],  # Input shape: (2, 3)
- *                      [4, 5, 6]])
- * tf.matrix_diag(diagonal, k = 1)
- *   ==> [[[0, 1, 0, 0],  # Output shape: (2, 4, 4)
- *         [0, 0, 2, 0],
- *         [0, 0, 0, 3],
- *         [0, 0, 0, 0]],
- *        [[0, 4, 0, 0],
- *         [0, 0, 5, 0],
- *         [0, 0, 0, 6],
- *         [0, 0, 0, 0]]]
- * 
- * # A band of diagonals.
- * diagonals = np.array([[[1, 2, 3],  # Input shape: (2, 2, 3)
- *                        [4, 5, 0]],
- *                       [[6, 7, 9],
- *                        [9, 1, 0]]])
- * tf.matrix_diag(diagonals, k = (-1, 0))
- *   ==> [[[1, 0, 0],  # Output shape: (2, 3, 3)
- *         [4, 2, 0],
- *         [0, 5, 3]],
- *        [[6, 0, 0],
- *         [9, 7, 0],
- *         [0, 1, 9]]]
- * 
- * # Rectangular matrix.
- * diagonal = np.array([1, 2])  # Input shape: (2)
- * tf.matrix_diag(diagonal, k = -1, num_rows = 3, num_cols = 4)
- *   ==> [[0, 0, 0, 0],  # Output shape: (3, 4)
- *        [1, 0, 0, 0],
- *        [0, 2, 0, 0]]
- * 
- * # Rectangular matrix with inferred num_cols and padding_value = 9.
- * tf.matrix_diag(diagonal, k = -1, num_rows = 3, padding_value = 9)
- *   ==> [[9, 9],  # Output shape: (3, 2)
- *        [1, 9],
- *        [9, 2]]
- * }</pre>
- * 
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator(group = "linalg")
-public final class MatrixDiag<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new MatrixDiag operation.
-   * 
-   * @param scope current scope
-   * @param diagonal Rank `r`, where `r >= 1`
-   * @param k Diagonal offset(s). Positive value means superdiagonal, 0 refers to the main
-   * diagonal, and negative value means subdiagonals. `k` can be a single integer
-   * (for a single diagonal) or a pair of integers specifying the low and high ends
-   * of a matrix band. `k[0]` must not be larger than `k[1]`.
-   * @param numRows The number of rows of the output matrix. If it is not provided, the op assumes
-   * the output matrix is a square matrix and infers the matrix size from k and the
-   * innermost dimension of `diagonal`.
-   * @param numCols The number of columns of the output matrix. If it is not provided, the op
-   * assumes the output matrix is a square matrix and infers the matrix size from
-   * k and the innermost dimension of `diagonal`.
-   * @param paddingValue The number to fill the area outside the specified diagonal band with.
-   * Default is 0.
-   * @return a new instance of MatrixDiag
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> MatrixDiag<T> create(Scope scope, Operand<T> diagonal, Operand<TInt32> k, Operand<TInt32> numRows, Operand<TInt32> numCols, Operand<T> paddingValue) {
-    OperationBuilder opBuilder = scope.env().opBuilder("MatrixDiagV2", scope.makeOpName("MatrixDiag"));
-    opBuilder.addInput(diagonal.asOutput(scope));
-    opBuilder.addInput(k.asOutput(scope));
-    opBuilder.addInput(numRows.asOutput(scope));
-    opBuilder.addInput(numCols.asOutput(scope));
-    opBuilder.addInput(paddingValue.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new MatrixDiag<T>(opBuilder.build());
-  }
-  
-  /**
-   * Has rank `r+1` when `k` is an integer or `k[0] == k[1]`, rank `r` otherwise.
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "MatrixDiagV2";
-  
-  private Output<T> output;
-  
-  private MatrixDiag(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/MatrixDiagPart.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/MatrixDiagPart.java
deleted file mode 100644
index a9c11cf82dc..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/MatrixDiagPart.java
+++ /dev/null
@@ -1,151 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.linalg;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.family.TType;
-
-/**
- * Returns the batched diagonal part of a batched tensor.
- * <p>
- * Returns a tensor with the `k[0]`-th to `k[1]`-th diagonals of the batched
- * `input`.
- * <p>
- * Assume `input` has `r` dimensions `[I, J, ..., L, M, N]`.
- * Let `max_diag_len` be the maximum length among all diagonals to be extracted,
- * `max_diag_len = min(M + min(k[1], 0), N + min(-k[0], 0))`
- * Let `num_diags` be the number of diagonals to extract,
- * `num_diags = k[1] - k[0] + 1`.
- * <p>
- * If `num_diags == 1`, the output tensor is of rank `r - 1` with shape
- * `[I, J, ..., L, max_diag_len]` and values:
- * <pre>{@code
- * diagonal[i, j, ..., l, n]
- *   = input[i, j, ..., l, n+y, n+x] ; if 0 <= n+y < M and 0 <= n+x < N,
- *     padding_value                 ; otherwise.
- * }</pre>
- * where `y = max(-k[1], 0)`, `x = max(k[1], 0)`.
- * <p>
- * Otherwise, the output tensor has rank `r` with dimensions
- * `[I, J, ..., L, num_diags, max_diag_len]` with values:
- * <pre>{@code
- * diagonal[i, j, ..., l, m, n]
- *   = input[i, j, ..., l, n+y, n+x] ; if 0 <= n+y < M and 0 <= n+x < N,
- *     padding_value                 ; otherwise.
- * }</pre>
- * where `d = k[1] - m`, `y = max(-d, 0)`, and `x = max(d, 0)`.
- * <p>
- * The input must be at least a matrix.
- * <p>
- * For example:
- * <pre>{@code
- * input = np.array([[[1, 2, 3, 4],  # Input shape: (2, 3, 4)
- *                    [5, 6, 7, 8],
- *                    [9, 8, 7, 6]],
- *                   [[5, 4, 3, 2],
- *                    [1, 2, 3, 4],
- *                    [5, 6, 7, 8]]])
- * 
- * # A main diagonal from each batch.
- * tf.matrix_diag_part(input) ==> [[1, 6, 7],  # Output shape: (2, 3)
- *                                 [5, 2, 7]]
- * 
- * # A superdiagonal from each batch.
- * tf.matrix_diag_part(input, k = 1)
- *   ==> [[2, 7, 6],  # Output shape: (2, 3)
- *        [4, 3, 8]]
- * 
- * # A tridiagonal band from each batch.
- * tf.matrix_diag_part(input, k = (-1, 1))
- *   ==> [[[2, 7, 6],  # Output shape: (2, 3, 3)
- *         [1, 6, 7],
- *         [5, 8, 0]],
- *        [[4, 3, 8],
- *         [5, 2, 7],
- *         [1, 6, 0]]]
- * 
- * # Padding value = 9
- * tf.matrix_diag_part(input, k = (1, 3), padding_value = 9)
- *   ==> [[[4, 9, 9],  # Output shape: (2, 3, 3)
- *         [3, 8, 9],
- *         [2, 7, 6]],
- *        [[2, 9, 9],
- *         [3, 4, 9],
- *         [4, 3, 8]]]
- * }</pre>
- * 
- * 
- * @param <T> data type for {@code diagonal()} output
- */
-@Operator(group = "linalg")
-public final class MatrixDiagPart<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new MatrixDiagPart operation.
-   * 
-   * @param scope current scope
-   * @param input Rank `r` tensor where `r >= 2`.
-   * @param k Diagonal offset(s). Positive value means superdiagonal, 0 refers to the main
-   * diagonal, and negative value means subdiagonals. `k` can be a single integer
-   * (for a single diagonal) or a pair of integers specifying the low and high ends
-   * of a matrix band. `k[0]` must not be larger than `k[1]`.
-   * @param paddingValue The value to fill the area outside the specified diagonal band with.
-   * Default is 0.
-   * @return a new instance of MatrixDiagPart
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> MatrixDiagPart<T> create(Scope scope, Operand<T> input, Operand<TInt32> k, Operand<T> paddingValue) {
-    OperationBuilder opBuilder = scope.env().opBuilder("MatrixDiagPartV2", scope.makeOpName("MatrixDiagPart"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder.addInput(k.asOutput(scope));
-    opBuilder.addInput(paddingValue.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new MatrixDiagPart<T>(opBuilder.build());
-  }
-  
-  /**
-   * The extracted diagonal(s).
-   */
-  public Output<T> diagonal() {
-    return diagonal;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return diagonal;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "MatrixDiagPartV2";
-  
-  private Output<T> diagonal;
-  
-  private MatrixDiagPart(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    diagonal = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/MatrixDiagPartV3.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/MatrixDiagPartV3.java
deleted file mode 100644
index 1d671ac2c66..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/MatrixDiagPartV3.java
+++ /dev/null
@@ -1,228 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.linalg;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.family.TType;
-
-/**
- * Returns the batched diagonal part of a batched tensor.
- * <p>
- * Returns a tensor with the `k[0]`-th to `k[1]`-th diagonals of the batched
- * `input`.
- * <p>
- * Assume `input` has `r` dimensions `[I, J, ..., L, M, N]`.
- * Let `max_diag_len` be the maximum length among all diagonals to be extracted,
- * `max_diag_len = min(M + min(k[1], 0), N + min(-k[0], 0))`
- * Let `num_diags` be the number of diagonals to extract,
- * `num_diags = k[1] - k[0] + 1`.
- * <p>
- * If `num_diags == 1`, the output tensor is of rank `r - 1` with shape
- * `[I, J, ..., L, max_diag_len]` and values:
- * <pre>{@code
- * diagonal[i, j, ..., l, n]
- *   = input[i, j, ..., l, n+y, n+x] ; if 0 <= n+y < M and 0 <= n+x < N,
- *     padding_value                 ; otherwise.
- * }</pre>
- * where `y = max(-k[1], 0)`, `x = max(k[1], 0)`.
- * <p>
- * Otherwise, the output tensor has rank `r` with dimensions
- * `[I, J, ..., L, num_diags, max_diag_len]` with values:
- * <pre>{@code
- * diagonal[i, j, ..., l, m, n]
- *   = input[i, j, ..., l, n+y, n+x] ; if 0 <= n+y < M and 0 <= n+x < N,
- *     padding_value                 ; otherwise.
- * }</pre>
- * where `d = k[1] - m`, `y = max(-d, 0) - offset`, and `x = max(d, 0) - offset`.
- * <p>
- * `offset` is zero except when the alignment of the diagonal is to the right.
- * <pre>{@code
- * offset = max_diag_len - diag_len(d) ; if (`align` in {RIGHT_LEFT, RIGHT_RIGHT}
- *                                            and `d >= 0`) or
- *                                          (`align` in {LEFT_RIGHT, RIGHT_RIGHT}
- *                                            and `d <= 0`)
- *          0                          ; otherwise
- * }</pre>
- * where `diag_len(d) = min(cols - max(d, 0), rows + min(d, 0))`.
- * <p>
- * The input must be at least a matrix.
- * <p>
- * For example:
- * <pre>{@code
- * input = np.array([[[1, 2, 3, 4],  # Input shape: (2, 3, 4)
- *                    [5, 6, 7, 8],
- *                    [9, 8, 7, 6]],
- *                   [[5, 4, 3, 2],
- *                    [1, 2, 3, 4],
- *                    [5, 6, 7, 8]]])
- * 
- * # A main diagonal from each batch.
- * tf.matrix_diag_part(input) ==> [[1, 6, 7],  # Output shape: (2, 3)
- *                                 [5, 2, 7]]
- * 
- * # A superdiagonal from each batch.
- * tf.matrix_diag_part(input, k = 1)
- *   ==> [[2, 7, 6],  # Output shape: (2, 3)
- *        [4, 3, 8]]
- * 
- * # A band from each batch.
- * tf.matrix_diag_part(input, k = (-1, 2))
- *   ==> [[[0, 3, 8],  # Output shape: (2, 4, 3)
- *         [2, 7, 6],
- *         [1, 6, 7],
- *         [5, 8, 0]],
- *        [[0, 3, 4],
- *         [4, 3, 8],
- *         [5, 2, 7],
- *         [1, 6, 0]]]
- * 
- * # LEFT_RIGHT alignment.
- * tf.matrix_diag_part(input, k = (-1, 2), align="LEFT_RIGHT")
- *   ==> [[[3, 8, 0],  # Output shape: (2, 4, 3)
- *         [2, 7, 6],
- *         [1, 6, 7],
- *         [0, 5, 8]],
- *        [[3, 4, 0],
- *         [4, 3, 8],
- *         [5, 2, 7],
- *         [0, 1, 6]]]
- * 
- * # max_diag_len can be shorter than the main diagonal.
- * tf.matrix_diag_part(input, k = (-2, -1))
- *   ==> [[[5, 8],
- *         [9, 0]],
- *        [[1, 6],
- *         [5, 0]]]
- * 
- * # padding_value = 9
- * tf.matrix_diag_part(input, k = (1, 3), padding_value = 9)
- *   ==> [[[9, 9, 4],  # Output shape: (2, 3, 3)
- *         [9, 3, 8],
- *         [2, 7, 6]],
- *        [[9, 9, 2],
- *         [9, 3, 4],
- *         [4, 3, 8]]]
- * 
- * }</pre>
- * 
- * 
- * @param <T> data type for {@code diagonal()} output
- */
-@Operator(group = "linalg")
-public final class MatrixDiagPartV3<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.linalg.MatrixDiagPartV3}
-   */
-  public static class Options {
-    
-    /**
-     * @param align Some diagonals are shorter than `max_diag_len` and need to be padded. `align` is
-     * a string specifying how superdiagonals and subdiagonals should be aligned,
-     * respectively. There are four possible alignments: "RIGHT_LEFT" (default),
-     * "LEFT_RIGHT", "LEFT_LEFT", and "RIGHT_RIGHT". "RIGHT_LEFT" aligns superdiagonals
-     * to the right (left-pads the row) and subdiagonals to the left (right-pads the
-     * row). It is the packing format LAPACK uses. cuSPARSE uses "LEFT_RIGHT", which is
-     * the opposite alignment.
-     */
-    public Options align(String align) {
-      this.align = align;
-      return this;
-    }
-    
-    private String align;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new MatrixDiagPartV3 operation.
-   * 
-   * @param scope current scope
-   * @param input Rank `r` tensor where `r >= 2`.
-   * @param k Diagonal offset(s). Positive value means superdiagonal, 0 refers to the main
-   * diagonal, and negative value means subdiagonals. `k` can be a single integer
-   * (for a single diagonal) or a pair of integers specifying the low and high ends
-   * of a matrix band. `k[0]` must not be larger than `k[1]`.
-   * @param paddingValue The value to fill the area outside the specified diagonal band with.
-   * Default is 0.
-   * @param options carries optional attributes values
-   * @return a new instance of MatrixDiagPartV3
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> MatrixDiagPartV3<T> create(Scope scope, Operand<T> input, Operand<TInt32> k, Operand<T> paddingValue, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("MatrixDiagPartV3", scope.makeOpName("MatrixDiagPartV3"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder.addInput(k.asOutput(scope));
-    opBuilder.addInput(paddingValue.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.align != null) {
-          opBuilder.setAttr("align", opts.align);
-        }
-      }
-    }
-    return new MatrixDiagPartV3<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param align Some diagonals are shorter than `max_diag_len` and need to be padded. `align` is
-   * a string specifying how superdiagonals and subdiagonals should be aligned,
-   * respectively. There are four possible alignments: "RIGHT_LEFT" (default),
-   * "LEFT_RIGHT", "LEFT_LEFT", and "RIGHT_RIGHT". "RIGHT_LEFT" aligns superdiagonals
-   * to the right (left-pads the row) and subdiagonals to the left (right-pads the
-   * row). It is the packing format LAPACK uses. cuSPARSE uses "LEFT_RIGHT", which is
-   * the opposite alignment.
-   */
-  public static Options align(String align) {
-    return new Options().align(align);
-  }
-  
-  /**
-   * The extracted diagonal(s).
-   */
-  public Output<T> diagonal() {
-    return diagonal;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return diagonal;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "MatrixDiagPartV3";
-  
-  private Output<T> diagonal;
-  
-  private MatrixDiagPartV3(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    diagonal = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/MatrixDiagV3.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/MatrixDiagV3.java
deleted file mode 100644
index 204954aabbb..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/MatrixDiagV3.java
+++ /dev/null
@@ -1,252 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.linalg;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.family.TType;
-
-/**
- * Returns a batched diagonal tensor with given batched diagonal values.
- * <p>
- * Returns a tensor with the contents in `diagonal` as `k[0]`-th to `k[1]`-th
- * diagonals of a matrix, with everything else padded with `padding`. `num_rows`
- * and `num_cols` specify the dimension of the innermost matrix of the output. If
- * both are not specified, the op assumes the innermost matrix is square and infers
- * its size from `k` and the innermost dimension of `diagonal`. If only one of them
- * is specified, the op assumes the unspecified value is the smallest possible
- * based on other criteria.
- * <p>
- * Let `diagonal` have `r` dimensions `[I, J, ..., L, M, N]`. The output tensor has
- * rank `r+1` with shape `[I, J, ..., L, M, num_rows, num_cols]` when only one
- * diagonal is given (`k` is an integer or `k[0] == k[1]`). Otherwise, it has rank
- * `r` with shape `[I, J, ..., L, num_rows, num_cols]`.
- * <p>
- * The second innermost dimension of `diagonal` has double meaning.
- * When `k` is scalar or `k[0] == k[1]`, `M` is part of the batch size
- * [I, J, ..., M], and the output tensor is:
- * <pre>{@code
- * output[i, j, ..., l, m, n]
- *   = diagonal[i, j, ..., l, n-max(d_upper, 0)] ; if n - m == d_upper
- *     padding_value                             ; otherwise
- * }</pre>
- * Otherwise, `M` is treated as the number of diagonals for the matrix in the
- * same batch (`M = k[1]-k[0]+1`), and the output tensor is:
- * <pre>{@code
- * output[i, j, ..., l, m, n]
- *   = diagonal[i, j, ..., l, diag_index, index_in_diag] ; if k[0] <= d <= k[1]
- *     padding_value                                     ; otherwise
- * }</pre>
- * where `d = n - m`, `diag_index = [k] - d`, and
- * `index_in_diag = n - max(d, 0) + offset`.
- * <p>
- * `offset` is zero except when the alignment of the diagonal is to the right.
- * <pre>{@code
- * offset = max_diag_len - diag_len(d) ; if (`align` in {RIGHT_LEFT, RIGHT_RIGHT}
- *                                            and `d >= 0`) or
- *                                          (`align` in {LEFT_RIGHT, RIGHT_RIGHT}
- *                                            and `d <= 0`)
- *          0                          ; otherwise
- * }</pre>
- * where `diag_len(d) = min(cols - max(d, 0), rows + min(d, 0))`.
- * <p>
- * For example:
- * <pre>{@code
- * # The main diagonal.
- * diagonal = np.array([[1, 2, 3, 4],            # Input shape: (2, 4)
- *                      [5, 6, 7, 8]])
- * tf.matrix_diag(diagonal) ==> [[[1, 0, 0, 0],  # Output shape: (2, 4, 4)
- *                                [0, 2, 0, 0],
- *                                [0, 0, 3, 0],
- *                                [0, 0, 0, 4]],
- *                               [[5, 0, 0, 0],
- *                                [0, 6, 0, 0],
- *                                [0, 0, 7, 0],
- *                                [0, 0, 0, 8]]]
- * 
- * # A superdiagonal (per batch).
- * diagonal = np.array([[1, 2, 3],  # Input shape: (2, 3)
- *                      [4, 5, 6]])
- * tf.matrix_diag(diagonal, k = 1)
- *   ==> [[[0, 1, 0, 0],  # Output shape: (2, 4, 4)
- *         [0, 0, 2, 0],
- *         [0, 0, 0, 3],
- *         [0, 0, 0, 0]],
- *        [[0, 4, 0, 0],
- *         [0, 0, 5, 0],
- *         [0, 0, 0, 6],
- *         [0, 0, 0, 0]]]
- * 
- * # A tridiagonal band (per batch).
- * diagonals = np.array([[[0, 8, 9],  # Input shape: (2, 2, 3)
- *                        [1, 2, 3],
- *                        [4, 5, 0]],
- *                       [[0, 2, 3],
- *                        [6, 7, 9],
- *                        [9, 1, 0]]])
- * tf.matrix_diag(diagonals, k = (-1, 1))
- *   ==> [[[1, 8, 0],  # Output shape: (2, 3, 3)
- *         [4, 2, 9],
- *         [0, 5, 3]],
- *        [[6, 2, 0],
- *         [9, 7, 3],
- *         [0, 1, 9]]]
- * 
- * # LEFT_RIGHT alignment.
- * diagonals = np.array([[[8, 9, 0],  # Input shape: (2, 2, 3)
- *                        [1, 2, 3],
- *                        [0, 4, 5]],
- *                       [[2, 3, 0],
- *                        [6, 7, 9],
- *                        [0, 9, 1]]])
- * tf.matrix_diag(diagonals, k = (-1, 1), align="LEFT_RIGHT")
- *   ==> [[[1, 8, 0],  # Output shape: (2, 3, 3)
- *         [4, 2, 9],
- *         [0, 5, 3]],
- *        [[6, 2, 0],
- *         [9, 7, 3],
- *         [0, 1, 9]]]
- * 
- * # Rectangular matrix.
- * diagonal = np.array([1, 2])  # Input shape: (2)
- * tf.matrix_diag(diagonal, k = -1, num_rows = 3, num_cols = 4)
- *   ==> [[0, 0, 0, 0],  # Output shape: (3, 4)
- *        [1, 0, 0, 0],
- *        [0, 2, 0, 0]]
- * 
- * # Rectangular matrix with inferred num_cols and padding_value = 9.
- * tf.matrix_diag(diagonal, k = -1, num_rows = 3, padding_value = 9)
- *   ==> [[9, 9],  # Output shape: (3, 2)
- *        [1, 9],
- *        [9, 2]]
- * 
- * }</pre>
- * 
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator(group = "linalg")
-public final class MatrixDiagV3<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.linalg.MatrixDiagV3}
-   */
-  public static class Options {
-    
-    /**
-     * @param align Some diagonals are shorter than `max_diag_len` and need to be padded. `align` is
-     * a string specifying how superdiagonals and subdiagonals should be aligned,
-     * respectively. There are four possible alignments: "RIGHT_LEFT" (default),
-     * "LEFT_RIGHT", "LEFT_LEFT", and "RIGHT_RIGHT". "RIGHT_LEFT" aligns superdiagonals
-     * to the right (left-pads the row) and subdiagonals to the left (right-pads the
-     * row). It is the packing format LAPACK uses. cuSPARSE uses "LEFT_RIGHT", which is
-     * the opposite alignment.
-     */
-    public Options align(String align) {
-      this.align = align;
-      return this;
-    }
-    
-    private String align;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new MatrixDiagV3 operation.
-   * 
-   * @param scope current scope
-   * @param diagonal Rank `r`, where `r >= 1`
-   * @param k Diagonal offset(s). Positive value means superdiagonal, 0 refers to the main
-   * diagonal, and negative value means subdiagonals. `k` can be a single integer
-   * (for a single diagonal) or a pair of integers specifying the low and high ends
-   * of a matrix band. `k[0]` must not be larger than `k[1]`.
-   * @param numRows The number of rows of the output matrix. If it is not provided, the op assumes
-   * the output matrix is a square matrix and infers the matrix size from k and the
-   * innermost dimension of `diagonal`.
-   * @param numCols The number of columns of the output matrix. If it is not provided, the op
-   * assumes the output matrix is a square matrix and infers the matrix size from
-   * k and the innermost dimension of `diagonal`.
-   * @param paddingValue The number to fill the area outside the specified diagonal band with.
-   * Default is 0.
-   * @param options carries optional attributes values
-   * @return a new instance of MatrixDiagV3
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> MatrixDiagV3<T> create(Scope scope, Operand<T> diagonal, Operand<TInt32> k, Operand<TInt32> numRows, Operand<TInt32> numCols, Operand<T> paddingValue, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("MatrixDiagV3", scope.makeOpName("MatrixDiagV3"));
-    opBuilder.addInput(diagonal.asOutput(scope));
-    opBuilder.addInput(k.asOutput(scope));
-    opBuilder.addInput(numRows.asOutput(scope));
-    opBuilder.addInput(numCols.asOutput(scope));
-    opBuilder.addInput(paddingValue.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.align != null) {
-          opBuilder.setAttr("align", opts.align);
-        }
-      }
-    }
-    return new MatrixDiagV3<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param align Some diagonals are shorter than `max_diag_len` and need to be padded. `align` is
-   * a string specifying how superdiagonals and subdiagonals should be aligned,
-   * respectively. There are four possible alignments: "RIGHT_LEFT" (default),
-   * "LEFT_RIGHT", "LEFT_LEFT", and "RIGHT_RIGHT". "RIGHT_LEFT" aligns superdiagonals
-   * to the right (left-pads the row) and subdiagonals to the left (right-pads the
-   * row). It is the packing format LAPACK uses. cuSPARSE uses "LEFT_RIGHT", which is
-   * the opposite alignment.
-   */
-  public static Options align(String align) {
-    return new Options().align(align);
-  }
-  
-  /**
-   * Has rank `r+1` when `k` is an integer or `k[0] == k[1]`, rank `r` otherwise.
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "MatrixDiagV3";
-  
-  private Output<T> output;
-  
-  private MatrixDiagV3(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/MatrixLogarithm.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/MatrixLogarithm.java
deleted file mode 100644
index a43eabb415b..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/MatrixLogarithm.java
+++ /dev/null
@@ -1,94 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.linalg;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Computes the matrix logarithm of one or more square matrices:
- * <p>
- * 
- * \\(log(exp(A)) = A\\)
- * <p>
- * This op is only defined for complex matrices. If A is positive-definite and
- * real, then casting to a complex matrix, taking the logarithm and casting back
- * to a real matrix will give the correct result.
- * <p>
- * This function computes the matrix logarithm using the Schur-Parlett algorithm.
- * Details of the algorithm can be found in Section 11.6.2 of:
- * Nicholas J. Higham, Functions of Matrices: Theory and Computation, SIAM 2008.
- * ISBN 978-0-898716-46-7.
- * <p>
- * The input is a tensor of shape `[..., M, M]` whose inner-most 2 dimensions
- * form square matrices. The output is a tensor of the same shape as the input
- * containing the exponential for all input submatrices `[..., :, :]`.
- * 
- * @param <T> data type for {@code output()} output
- */
-public final class MatrixLogarithm<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new MatrixLogarithm operation.
-   * 
-   * @param scope current scope
-   * @param input Shape is `[..., M, M]`.
-   * @return a new instance of MatrixLogarithm
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> MatrixLogarithm<T> create(Scope scope, Operand<T> input) {
-    OperationBuilder opBuilder = scope.env().opBuilder("MatrixLogarithm", scope.makeOpName("MatrixLogarithm"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new MatrixLogarithm<T>(opBuilder.build());
-  }
-  
-  /**
-   * Shape is `[..., M, M]`.
-   * <p>
-   * @compatibility(scipy)
-   * Equivalent to scipy.linalg.logm
-   * @end_compatibility
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "MatrixLogarithm";
-  
-  private Output<T> output;
-  
-  private MatrixLogarithm(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/MatrixSetDiag.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/MatrixSetDiag.java
deleted file mode 100644
index d57dbb92fc8..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/MatrixSetDiag.java
+++ /dev/null
@@ -1,232 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.linalg;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.family.TType;
-
-/**
- * Returns a batched matrix tensor with new batched diagonal values.
- * <p>
- * Given `input` and `diagonal`, this operation returns a tensor with the
- * same shape and values as `input`, except for the specified diagonals of the
- * innermost matrices. These will be overwritten by the values in `diagonal`.
- * <p>
- * `input` has `r+1` dimensions `[I, J, ..., L, M, N]`. When `k` is scalar or
- * `k[0] == k[1]`, `diagonal` has `r` dimensions `[I, J, ..., L, max_diag_len]`.
- * Otherwise, it has `r+1` dimensions `[I, J, ..., L, num_diags, max_diag_len]`.
- * `num_diags` is the number of diagonals, `num_diags = k[1] - k[0] + 1`.
- * `max_diag_len` is the longest diagonal in the range `[k[0], k[1]]`,
- * `max_diag_len = min(M + min(k[1], 0), N + min(-k[0], 0))`
- * <p>
- * The output is a tensor of rank `k+1` with dimensions `[I, J, ..., L, M, N]`.
- * If `k` is scalar or `k[0] == k[1]`:
- * <pre>{@code
- * output[i, j, ..., l, m, n]
- *   = diagonal[i, j, ..., l, n-max(k[1], 0)] ; if n - m == k[1]
- *     input[i, j, ..., l, m, n]              ; otherwise
- * }</pre>
- * Otherwise,
- * <pre>{@code
- * output[i, j, ..., l, m, n]
- *   = diagonal[i, j, ..., l, diag_index, index_in_diag] ; if k[0] <= d <= k[1]
- *     input[i, j, ..., l, m, n]                         ; otherwise
- * }</pre>
- * where `d = n - m`, `diag_index = k[1] - d`, and
- * `index_in_diag = n - max(d, 0) + offset`.
- * <p>
- * `offset` is zero except when the alignment of the diagonal is to the right.
- * <pre>{@code
- * offset = max_diag_len - diag_len(d) ; if (`align` in {RIGHT_LEFT, RIGHT_RIGHT}
- *                                            and `d >= 0`) or
- *                                          (`align` in {LEFT_RIGHT, RIGHT_RIGHT}
- *                                            and `d <= 0`)
- *          0                          ; otherwise
- * }</pre>
- * where `diag_len(d) = min(cols - max(d, 0), rows + min(d, 0))`.
- * <p>
- * For example:
- * <pre>{@code
- * # The main diagonal.
- * input = np.array([[[7, 7, 7, 7],              # Input shape: (2, 3, 4)
- *                    [7, 7, 7, 7],
- *                    [7, 7, 7, 7]],
- *                   [[7, 7, 7, 7],
- *                    [7, 7, 7, 7],
- *                    [7, 7, 7, 7]]])
- * diagonal = np.array([[1, 2, 3],               # Diagonal shape: (2, 3)
- *                      [4, 5, 6]])
- * tf.matrix_set_diag(input, diagonal)
- *   ==> [[[1, 7, 7, 7],  # Output shape: (2, 3, 4)
- *         [7, 2, 7, 7],
- *         [7, 7, 3, 7]],
- *        [[4, 7, 7, 7],
- *         [7, 5, 7, 7],
- *         [7, 7, 6, 7]]]
- * 
- * # A superdiagonal (per batch).
- * tf.matrix_set_diag(input, diagonal, k = 1)
- *   ==> [[[7, 1, 7, 7],  # Output shape: (2, 3, 4)
- *         [7, 7, 2, 7],
- *         [7, 7, 7, 3]],
- *        [[7, 4, 7, 7],
- *         [7, 7, 5, 7],
- *         [7, 7, 7, 6]]]
- * 
- * # A band of diagonals.
- * diagonals = np.array([[[0, 9, 1],  # Diagonal shape: (2, 4, 3)
- *                        [6, 5, 8],
- *                        [1, 2, 3],
- *                        [4, 5, 0]],
- *                       [[0, 1, 2],
- *                        [5, 6, 4],
- *                        [6, 1, 2],
- *                        [3, 4, 0]]])
- * tf.matrix_set_diag(input, diagonals, k = (-1, 2))
- *   ==> [[[1, 6, 9, 7],  # Output shape: (2, 3, 4)
- *         [4, 2, 5, 1],
- *         [7, 5, 3, 8]],
- *        [[6, 5, 1, 7],
- *         [3, 1, 6, 2],
- *         [7, 4, 2, 4]]]
- * 
- * # LEFT_RIGHT alignment.
- * diagonals = np.array([[[9, 1, 0],  # Diagonal shape: (2, 4, 3)
- *                        [6, 5, 8],
- *                        [1, 2, 3],
- *                        [0, 4, 5]],
- *                       [[1, 2, 0],
- *                        [5, 6, 4],
- *                        [6, 1, 2],
- *                        [0, 3, 4]]])
- * tf.matrix_set_diag(input, diagonals, k = (-1, 2), align="LEFT_RIGHT")
- *   ==> [[[1, 6, 9, 7],  # Output shape: (2, 3, 4)
- *         [4, 2, 5, 1],
- *         [7, 5, 3, 8]],
- *        [[6, 5, 1, 7],
- *         [3, 1, 6, 2],
- *         [7, 4, 2, 4]]]
- * 
- * }</pre>
- * 
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator(group = "linalg")
-public final class MatrixSetDiag<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.linalg.MatrixSetDiag}
-   */
-  public static class Options {
-    
-    /**
-     * @param align Some diagonals are shorter than `max_diag_len` and need to be padded. `align` is
-     * a string specifying how superdiagonals and subdiagonals should be aligned,
-     * respectively. There are four possible alignments: "RIGHT_LEFT" (default),
-     * "LEFT_RIGHT", "LEFT_LEFT", and "RIGHT_RIGHT". "RIGHT_LEFT" aligns superdiagonals
-     * to the right (left-pads the row) and subdiagonals to the left (right-pads the
-     * row). It is the packing format LAPACK uses. cuSPARSE uses "LEFT_RIGHT", which is
-     * the opposite alignment.
-     */
-    public Options align(String align) {
-      this.align = align;
-      return this;
-    }
-    
-    private String align;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new MatrixSetDiag operation.
-   * 
-   * @param scope current scope
-   * @param input Rank `r+1`, where `r >= 1`.
-   * @param diagonal Rank `r` when `k` is an integer or `k[0] == k[1]`. Otherwise, it has rank `r+1`.
-   * `k >= 1`.
-   * @param k Diagonal offset(s). Positive value means superdiagonal, 0 refers to the main
-   * diagonal, and negative value means subdiagonals. `k` can be a single integer
-   * (for a single diagonal) or a pair of integers specifying the low and high ends
-   * of a matrix band. `k[0]` must not be larger than `k[1]`.
-   * @param options carries optional attributes values
-   * @return a new instance of MatrixSetDiag
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> MatrixSetDiag<T> create(Scope scope, Operand<T> input, Operand<T> diagonal, Operand<TInt32> k, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("MatrixSetDiagV3", scope.makeOpName("MatrixSetDiag"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder.addInput(diagonal.asOutput(scope));
-    opBuilder.addInput(k.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.align != null) {
-          opBuilder.setAttr("align", opts.align);
-        }
-      }
-    }
-    return new MatrixSetDiag<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param align Some diagonals are shorter than `max_diag_len` and need to be padded. `align` is
-   * a string specifying how superdiagonals and subdiagonals should be aligned,
-   * respectively. There are four possible alignments: "RIGHT_LEFT" (default),
-   * "LEFT_RIGHT", "LEFT_LEFT", and "RIGHT_RIGHT". "RIGHT_LEFT" aligns superdiagonals
-   * to the right (left-pads the row) and subdiagonals to the left (right-pads the
-   * row). It is the packing format LAPACK uses. cuSPARSE uses "LEFT_RIGHT", which is
-   * the opposite alignment.
-   */
-  public static Options align(String align) {
-    return new Options().align(align);
-  }
-  
-  /**
-   * Rank `r+1`, with `output.shape = input.shape`.
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "MatrixSetDiagV3";
-  
-  private Output<T> output;
-  
-  private MatrixSetDiag(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/MatrixSolveLs.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/MatrixSolveLs.java
deleted file mode 100644
index cea92eb0f45..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/MatrixSolveLs.java
+++ /dev/null
@@ -1,153 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.linalg;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat64;
-import org.tensorflow.types.family.TType;
-
-/**
- * Solves one or more linear least-squares problems.
- * <p>
- * `matrix` is a tensor of shape `[..., M, N]` whose inner-most 2 dimensions
- * form real or complex matrices of size `[M, N]`. `Rhs` is a tensor of the same
- * type as `matrix` and shape `[..., M, K]`.
- * The output is a tensor shape `[..., N, K]` where each output matrix solves
- * each of the equations
- * `matrix[..., :, :]` * `output[..., :, :]` = `rhs[..., :, :]`
- * in the least squares sense.
- * <p>
- * We use the following notation for (complex) matrix and right-hand sides
- * in the batch:
- * <p>
- * `matrix`=\\(A \in \mathbb{C}^{m \times n}\\),
- * `rhs`=\\(B  \in \mathbb{C}^{m \times k}\\),
- * `output`=\\(X  \in \mathbb{C}^{n \times k}\\),
- * `l2_regularizer`=\\(\lambda \in \mathbb{R}\\).
- * <p>
- * If `fast` is `True`, then the solution is computed by solving the normal
- * equations using Cholesky decomposition. Specifically, if \\(m \ge n\\) then
- * \\(X = (A^H A + \lambda I)^{-1} A^H B\\), which solves the least-squares
- * problem \\(X = \mathrm{argmin}_{Z \in \Re^{n \times k} } ||A Z - B||_F^2 + \lambda ||Z||_F^2\\).
- * If \\(m \lt n\\) then `output` is computed as
- * \\(X = A^H (A A^H + \lambda I)^{-1} B\\), which (for \\(\lambda = 0\\)) is the
- * minimum-norm solution to the under-determined linear system, i.e.
- * \\(X = \mathrm{argmin}_{Z \in \mathbb{C}^{n \times k} } ||Z||_F^2 \\),
- * subject to \\(A Z = B\\). Notice that the fast path is only numerically stable
- * when \\(A\\) is numerically full rank and has a condition number
- * \\(\mathrm{cond}(A) \lt \frac{1}{\sqrt{\epsilon_{mach} } }\\) or \\(\lambda\\) is
- * sufficiently large.
- * <p>
- * If `fast` is `False` an algorithm based on the numerically robust complete
- * orthogonal decomposition is used. This computes the minimum-norm
- * least-squares solution, even when \\(A\\) is rank deficient. This path is
- * typically 6-7 times slower than the fast path. If `fast` is `False` then
- * `l2_regularizer` is ignored.
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator(group = "linalg")
-public final class MatrixSolveLs<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.linalg.MatrixSolveLs}
-   */
-  public static class Options {
-    
-    /**
-     * @param fast 
-     */
-    public Options fast(Boolean fast) {
-      this.fast = fast;
-      return this;
-    }
-    
-    private Boolean fast;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new MatrixSolveLs operation.
-   * 
-   * @param scope current scope
-   * @param matrix Shape is `[..., M, N]`.
-   * @param rhs Shape is `[..., M, K]`.
-   * @param l2Regularizer Scalar tensor.
-   * <p>
-   * @compatibility(numpy)
-   * Equivalent to np.linalg.lstsq
-   * @end_compatibility
-   * @param options carries optional attributes values
-   * @return a new instance of MatrixSolveLs
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> MatrixSolveLs<T> create(Scope scope, Operand<T> matrix, Operand<T> rhs, Operand<TFloat64> l2Regularizer, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("MatrixSolveLs", scope.makeOpName("MatrixSolveLs"));
-    opBuilder.addInput(matrix.asOutput(scope));
-    opBuilder.addInput(rhs.asOutput(scope));
-    opBuilder.addInput(l2Regularizer.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.fast != null) {
-          opBuilder.setAttr("fast", opts.fast);
-        }
-      }
-    }
-    return new MatrixSolveLs<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param fast 
-   */
-  public static Options fast(Boolean fast) {
-    return new Options().fast(fast);
-  }
-  
-  /**
-   * Shape is `[..., N, K]`.
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "MatrixSolveLs";
-  
-  private Output<T> output;
-  
-  private MatrixSolveLs(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Qr.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Qr.java
deleted file mode 100644
index 483c709d75c..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Qr.java
+++ /dev/null
@@ -1,130 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.linalg;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Computes the QR decompositions of one or more matrices.
- * <p>
- * Computes the QR decomposition of each inner matrix in `tensor` such that
- * `tensor[..., :, :] = q[..., :, :] * r[..., :,:])`
- * <pre>{@code
- * # a is a tensor.
- * # q is a tensor of orthonormal matrices.
- * # r is a tensor of upper triangular matrices.
- * q, r = qr(a)
- * q_full, r_full = qr(a, full_matrices=True)
- * }</pre>
- * 
- * 
- * @param <T> data type for {@code q()} output
- */
-@Operator(group = "linalg")
-public final class Qr<T extends TType> extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.linalg.Qr}
-   */
-  public static class Options {
-    
-    /**
-     * @param fullMatrices If true, compute full-sized `q` and `r`. If false
-     * (the default), compute only the leading `P` columns of `q`.
-     */
-    public Options fullMatrices(Boolean fullMatrices) {
-      this.fullMatrices = fullMatrices;
-      return this;
-    }
-    
-    private Boolean fullMatrices;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new Qr operation.
-   * 
-   * @param scope current scope
-   * @param input A tensor of shape `[..., M, N]` whose inner-most 2 dimensions
-   * form matrices of size `[M, N]`. Let `P` be the minimum of `M` and `N`.
-   * @param options carries optional attributes values
-   * @return a new instance of Qr
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> Qr<T> create(Scope scope, Operand<T> input, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Qr", scope.makeOpName("Qr"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.fullMatrices != null) {
-          opBuilder.setAttr("full_matrices", opts.fullMatrices);
-        }
-      }
-    }
-    return new Qr<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param fullMatrices If true, compute full-sized `q` and `r`. If false
-   * (the default), compute only the leading `P` columns of `q`.
-   */
-  public static Options fullMatrices(Boolean fullMatrices) {
-    return new Options().fullMatrices(fullMatrices);
-  }
-  
-  /**
-   * Orthonormal basis for range of `a`. If `full_matrices` is `False` then
-   * shape is `[..., M, P]`; if `full_matrices` is `True` then shape is
-   * `[..., M, M]`.
-   */
-  public Output<T> q() {
-    return q;
-  }
-  
-  /**
-   * Triangular factor. If `full_matrices` is `False` then shape is
-   * `[..., P, N]`. If `full_matrices` is `True` then shape is `[..., M, N]`.
-   */
-  public Output<T> r() {
-    return r;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Qr";
-  
-  private Output<T> q;
-  private Output<T> r;
-  
-  private Qr(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    q = operation.output(outputIdx++);
-    r = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/QuantizedMatMul.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/QuantizedMatMul.java
deleted file mode 100644
index e8fd87c1fb3..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/QuantizedMatMul.java
+++ /dev/null
@@ -1,161 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.linalg;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.family.TType;
-
-/**
- * Perform a quantized matrix multiplication of  `a` by the matrix `b`.
- * <p>
- * The inputs must be two-dimensional matrices and the inner dimension of
- * `a` (after being transposed if `transpose_a` is non-zero) must match the
- * outer dimension of `b` (after being transposed if `transposed_b` is
- * non-zero).
- * 
- * @param <V> data type for {@code out()} output
- */
-@Operator(group = "linalg")
-public final class QuantizedMatMul<V extends TType> extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.linalg.QuantizedMatMul}
-   */
-  public static class Options {
-    
-    /**
-     * @param transposeA If true, `a` is transposed before multiplication.
-     */
-    public Options transposeA(Boolean transposeA) {
-      this.transposeA = transposeA;
-      return this;
-    }
-    
-    /**
-     * @param transposeB If true, `b` is transposed before multiplication.
-     */
-    public Options transposeB(Boolean transposeB) {
-      this.transposeB = transposeB;
-      return this;
-    }
-    
-    private Boolean transposeA;
-    private Boolean transposeB;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new QuantizedMatMul operation.
-   * 
-   * @param scope current scope
-   * @param a Must be a two-dimensional tensor.
-   * @param b Must be a two-dimensional tensor.
-   * @param minA The float value that the lowest quantized `a` value represents.
-   * @param maxA The float value that the highest quantized `a` value represents.
-   * @param minB The float value that the lowest quantized `b` value represents.
-   * @param maxB The float value that the highest quantized `b` value represents.
-   * @param Toutput 
-   * @param Tactivation The type of output produced by activation function
-   * following this operation.
-   * @param options carries optional attributes values
-   * @return a new instance of QuantizedMatMul
-   */
-  @Endpoint(describeByClass = true)
-  public static <V extends TType, T extends TType, U extends TType, W extends TType> QuantizedMatMul<V> create(Scope scope, Operand<T> a, Operand<U> b, Operand<TFloat32> minA, Operand<TFloat32> maxA, Operand<TFloat32> minB, Operand<TFloat32> maxB, Class<V> Toutput, Class<W> Tactivation, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("QuantizedMatMul", scope.makeOpName("QuantizedMatMul"));
-    opBuilder.addInput(a.asOutput(scope));
-    opBuilder.addInput(b.asOutput(scope));
-    opBuilder.addInput(minA.asOutput(scope));
-    opBuilder.addInput(maxA.asOutput(scope));
-    opBuilder.addInput(minB.asOutput(scope));
-    opBuilder.addInput(maxB.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("Toutput", Toutput);
-    opBuilder.setAttr("Tactivation", Tactivation);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.transposeA != null) {
-          opBuilder.setAttr("transpose_a", opts.transposeA);
-        }
-        if (opts.transposeB != null) {
-          opBuilder.setAttr("transpose_b", opts.transposeB);
-        }
-      }
-    }
-    return new QuantizedMatMul<V>(opBuilder.build());
-  }
-  
-  /**
-   * @param transposeA If true, `a` is transposed before multiplication.
-   */
-  public static Options transposeA(Boolean transposeA) {
-    return new Options().transposeA(transposeA);
-  }
-  
-  /**
-   * @param transposeB If true, `b` is transposed before multiplication.
-   */
-  public static Options transposeB(Boolean transposeB) {
-    return new Options().transposeB(transposeB);
-  }
-  
-  /**
-   */
-  public Output<V> out() {
-    return out;
-  }
-  
-  /**
-   * The float value that the lowest quantized output value represents.
-   */
-  public Output<TFloat32> minOut() {
-    return minOut;
-  }
-  
-  /**
-   * The float value that the highest quantized output value represents.
-   */
-  public Output<TFloat32> maxOut() {
-    return maxOut;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "QuantizedMatMul";
-  
-  private Output<V> out;
-  private Output<TFloat32> minOut;
-  private Output<TFloat32> maxOut;
-  
-  private QuantizedMatMul(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    out = operation.output(outputIdx++);
-    minOut = operation.output(outputIdx++);
-    maxOut = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/QuantizedMatMulWithBias.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/QuantizedMatMulWithBias.java
deleted file mode 100644
index faecc84f8cc..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/QuantizedMatMulWithBias.java
+++ /dev/null
@@ -1,181 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.linalg;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.family.TType;
-
-/**
- * Performs a quantized matrix multiplication of `a` by the matrix `b` with bias
- * add.
- * <p>
- * The inputs must be two-dimensional matrices and 1D bias vector. And the inner
- * dimension of `a` (after being transposed if `transpose_a` is non-zero) must
- * match the outer dimension of `b` (after being transposed if `transposed_b` is
- * non-zero). Then do broadcast add operation with bias values on the matrix
- * multiplication result. The bias size must match inner dimension of `b`.
- * 
- * @param <W> data type for {@code out()} output
- */
-public final class QuantizedMatMulWithBias<W extends TType> extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.linalg.QuantizedMatMulWithBias}
-   */
-  public static class Options {
-    
-    /**
-     * @param transposeA If true, `a` is transposed before multiplication.
-     */
-    public Options transposeA(Boolean transposeA) {
-      this.transposeA = transposeA;
-      return this;
-    }
-    
-    /**
-     * @param transposeB If true, `b` is transposed before multiplication.
-     */
-    public Options transposeB(Boolean transposeB) {
-      this.transposeB = transposeB;
-      return this;
-    }
-    
-    /**
-     * @param inputQuantMode Input data quantization mode. Either MIN_FIRST(default) or SCALED.
-     */
-    public Options inputQuantMode(String inputQuantMode) {
-      this.inputQuantMode = inputQuantMode;
-      return this;
-    }
-    
-    private Boolean transposeA;
-    private Boolean transposeB;
-    private String inputQuantMode;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new QuantizedMatMulWithBias operation.
-   * 
-   * @param scope current scope
-   * @param a A matrix to be multiplied. Must be a two-dimensional tensor of type `quint8`.
-   * @param b A matrix to be multiplied and must be a two-dimensional tensor of type `qint8`.
-   * @param bias A 1D bias tensor with size matching inner dimension of `b` (after being
-   * transposed if `transposed_b` is non-zero).
-   * @param minA The float value that the lowest quantized `a` value represents.
-   * @param maxA The float value that the highest quantized `a` value represents.
-   * @param minB The float value that the lowest quantized `b` value represents.
-   * @param maxB The float value that the highest quantized `b` value represents.
-   * @param Toutput 
-   * @param options carries optional attributes values
-   * @return a new instance of QuantizedMatMulWithBias
-   */
-  @Endpoint(describeByClass = true)
-  public static <W extends TType, T extends TType, U extends TType, V extends TType> QuantizedMatMulWithBias<W> create(Scope scope, Operand<T> a, Operand<U> b, Operand<V> bias, Operand<TFloat32> minA, Operand<TFloat32> maxA, Operand<TFloat32> minB, Operand<TFloat32> maxB, Class<W> Toutput, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("QuantizedMatMulWithBias", scope.makeOpName("QuantizedMatMulWithBias"));
-    opBuilder.addInput(a.asOutput(scope));
-    opBuilder.addInput(b.asOutput(scope));
-    opBuilder.addInput(bias.asOutput(scope));
-    opBuilder.addInput(minA.asOutput(scope));
-    opBuilder.addInput(maxA.asOutput(scope));
-    opBuilder.addInput(minB.asOutput(scope));
-    opBuilder.addInput(maxB.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("Toutput", Toutput);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.transposeA != null) {
-          opBuilder.setAttr("transpose_a", opts.transposeA);
-        }
-        if (opts.transposeB != null) {
-          opBuilder.setAttr("transpose_b", opts.transposeB);
-        }
-        if (opts.inputQuantMode != null) {
-          opBuilder.setAttr("input_quant_mode", opts.inputQuantMode);
-        }
-      }
-    }
-    return new QuantizedMatMulWithBias<W>(opBuilder.build());
-  }
-  
-  /**
-   * @param transposeA If true, `a` is transposed before multiplication.
-   */
-  public static Options transposeA(Boolean transposeA) {
-    return new Options().transposeA(transposeA);
-  }
-  
-  /**
-   * @param transposeB If true, `b` is transposed before multiplication.
-   */
-  public static Options transposeB(Boolean transposeB) {
-    return new Options().transposeB(transposeB);
-  }
-  
-  /**
-   * @param inputQuantMode Input data quantization mode. Either MIN_FIRST(default) or SCALED.
-   */
-  public static Options inputQuantMode(String inputQuantMode) {
-    return new Options().inputQuantMode(inputQuantMode);
-  }
-  
-  /**
-   */
-  public Output<W> out() {
-    return out;
-  }
-  
-  /**
-   * The float value that the lowest quantized output value represents.
-   */
-  public Output<TFloat32> minOut() {
-    return minOut;
-  }
-  
-  /**
-   * The float value that the highest quantized output value represents.
-   */
-  public Output<TFloat32> maxOut() {
-    return maxOut;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "QuantizedMatMulWithBias";
-  
-  private Output<W> out;
-  private Output<TFloat32> minOut;
-  private Output<TFloat32> maxOut;
-  
-  private QuantizedMatMulWithBias(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    out = operation.output(outputIdx++);
-    minOut = operation.output(outputIdx++);
-    maxOut = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/QuantizedMatMulWithBiasAndRelu.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/QuantizedMatMulWithBiasAndRelu.java
deleted file mode 100644
index 55dbb0ba73a..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/QuantizedMatMulWithBiasAndRelu.java
+++ /dev/null
@@ -1,182 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.linalg;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.family.TType;
-
-/**
- * Perform a quantized matrix multiplication of  `a` by the matrix `b` with bias
- * add and relu fusion.
- * <p>
- * The inputs must be two-dimensional matrices and 1D bias vector. And the inner
- * dimension of `a` (after being transposed if `transpose_a` is non-zero) must
- * match the outer dimension of `b` (after being transposed if `transposed_b` is
- * non-zero). Then do broadcast add operation with bias values on the matrix
- * multiplication result. The bias size must match inner dimension of `b`. Then do
- * relu activation to get non-negative result.
- * 
- * @param <V> data type for {@code out()} output
- */
-public final class QuantizedMatMulWithBiasAndRelu<V extends TType> extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.linalg.QuantizedMatMulWithBiasAndRelu}
-   */
-  public static class Options {
-    
-    /**
-     * @param transposeA If true, `a` is transposed before multiplication.
-     */
-    public Options transposeA(Boolean transposeA) {
-      this.transposeA = transposeA;
-      return this;
-    }
-    
-    /**
-     * @param transposeB If true, `b` is transposed before multiplication.
-     */
-    public Options transposeB(Boolean transposeB) {
-      this.transposeB = transposeB;
-      return this;
-    }
-    
-    /**
-     * @param inputQuantMode Input data quantization mode. Either MIN_FIRST(default) or SCALED.
-     */
-    public Options inputQuantMode(String inputQuantMode) {
-      this.inputQuantMode = inputQuantMode;
-      return this;
-    }
-    
-    private Boolean transposeA;
-    private Boolean transposeB;
-    private String inputQuantMode;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new QuantizedMatMulWithBiasAndRelu operation.
-   * 
-   * @param scope current scope
-   * @param a A matrix to be multiplied. Must be a two-dimensional tensor of type `quint8`.
-   * @param b A matrix to be multiplied and must be a two-dimensional tensor of type `qint8`.
-   * @param bias A 1D bias tensor with size matching with inner dimension of `b` (after being
-   * transposed if `transposed_b` is non-zero).
-   * @param minA The float value that the lowest quantized `a` value represents.
-   * @param maxA The float value that the highest quantized `a` value represents.
-   * @param minB The float value that the lowest quantized `b` value represents.
-   * @param maxB The float value that the highest quantized `b` value represents.
-   * @param Toutput 
-   * @param options carries optional attributes values
-   * @return a new instance of QuantizedMatMulWithBiasAndRelu
-   */
-  @Endpoint(describeByClass = true)
-  public static <V extends TType, T extends TType, U extends TType> QuantizedMatMulWithBiasAndRelu<V> create(Scope scope, Operand<T> a, Operand<U> b, Operand<TFloat32> bias, Operand<TFloat32> minA, Operand<TFloat32> maxA, Operand<TFloat32> minB, Operand<TFloat32> maxB, Class<V> Toutput, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("QuantizedMatMulWithBiasAndRelu", scope.makeOpName("QuantizedMatMulWithBiasAndRelu"));
-    opBuilder.addInput(a.asOutput(scope));
-    opBuilder.addInput(b.asOutput(scope));
-    opBuilder.addInput(bias.asOutput(scope));
-    opBuilder.addInput(minA.asOutput(scope));
-    opBuilder.addInput(maxA.asOutput(scope));
-    opBuilder.addInput(minB.asOutput(scope));
-    opBuilder.addInput(maxB.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("Toutput", Toutput);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.transposeA != null) {
-          opBuilder.setAttr("transpose_a", opts.transposeA);
-        }
-        if (opts.transposeB != null) {
-          opBuilder.setAttr("transpose_b", opts.transposeB);
-        }
-        if (opts.inputQuantMode != null) {
-          opBuilder.setAttr("input_quant_mode", opts.inputQuantMode);
-        }
-      }
-    }
-    return new QuantizedMatMulWithBiasAndRelu<V>(opBuilder.build());
-  }
-  
-  /**
-   * @param transposeA If true, `a` is transposed before multiplication.
-   */
-  public static Options transposeA(Boolean transposeA) {
-    return new Options().transposeA(transposeA);
-  }
-  
-  /**
-   * @param transposeB If true, `b` is transposed before multiplication.
-   */
-  public static Options transposeB(Boolean transposeB) {
-    return new Options().transposeB(transposeB);
-  }
-  
-  /**
-   * @param inputQuantMode Input data quantization mode. Either MIN_FIRST(default) or SCALED.
-   */
-  public static Options inputQuantMode(String inputQuantMode) {
-    return new Options().inputQuantMode(inputQuantMode);
-  }
-  
-  /**
-   */
-  public Output<V> out() {
-    return out;
-  }
-  
-  /**
-   * The float value that the lowest quantized output value represents.
-   */
-  public Output<TFloat32> minOut() {
-    return minOut;
-  }
-  
-  /**
-   * The float value that the highest quantized output value represents.
-   */
-  public Output<TFloat32> maxOut() {
-    return maxOut;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "QuantizedMatMulWithBiasAndRelu";
-  
-  private Output<V> out;
-  private Output<TFloat32> minOut;
-  private Output<TFloat32> maxOut;
-  
-  private QuantizedMatMulWithBiasAndRelu(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    out = operation.output(outputIdx++);
-    minOut = operation.output(outputIdx++);
-    maxOut = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/QuantizedMatMulWithBiasAndReluAndRequantize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/QuantizedMatMulWithBiasAndReluAndRequantize.java
deleted file mode 100644
index 8d0f81359f3..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/QuantizedMatMulWithBiasAndReluAndRequantize.java
+++ /dev/null
@@ -1,187 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.linalg;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.family.TType;
-
-/**
- * Perform a quantized matrix multiplication of  `a` by the matrix `b` with bias
- * add and relu and requantize fusion.
- * <p>
- * The inputs must be two-dimensional matrices and 1D bias vector. And the inner
- * dimension of `a` (after being transposed if `transpose_a` is non-zero) must
- * match the outer dimension of `b` (after being transposed if `transposed_b` is
- * non-zero). Then do broadcast add operation with bias values on the matrix
- * multiplication result. The bias size must match inner dimension of `b`.  Then do
- * relu activation to get non-negative result. Then do requantize operation to get
- * final uint8 result.
- * 
- * @param <W> data type for {@code out()} output
- */
-public final class QuantizedMatMulWithBiasAndReluAndRequantize<W extends TType> extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.linalg.QuantizedMatMulWithBiasAndReluAndRequantize}
-   */
-  public static class Options {
-    
-    /**
-     * @param transposeA If true, `a` is transposed before multiplication.
-     */
-    public Options transposeA(Boolean transposeA) {
-      this.transposeA = transposeA;
-      return this;
-    }
-    
-    /**
-     * @param transposeB If true, `b` is transposed before multiplication.
-     */
-    public Options transposeB(Boolean transposeB) {
-      this.transposeB = transposeB;
-      return this;
-    }
-    
-    /**
-     * @param inputQuantMode Input data quantization mode. Either MIN_FIRST(default) or SCALED.
-     */
-    public Options inputQuantMode(String inputQuantMode) {
-      this.inputQuantMode = inputQuantMode;
-      return this;
-    }
-    
-    private Boolean transposeA;
-    private Boolean transposeB;
-    private String inputQuantMode;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new QuantizedMatMulWithBiasAndReluAndRequantize operation.
-   * 
-   * @param scope current scope
-   * @param a A matrix to be multiplied. Must be a two-dimensional tensor of type `quint8`.
-   * @param b A matrix to be multiplied and must be a two-dimensional tensor of type `qint8`.
-   * @param bias A 1D bias tensor with size matching with inner dimension of `b` (after being
-   * transposed if `transposed_b` is non-zero).
-   * @param minA The float value that the lowest quantized `a` value represents.
-   * @param maxA The float value that the highest quantized `a` value represents.
-   * @param minB The float value that the lowest quantized `b` value represents.
-   * @param maxB The float value that the highest quantized `b` value represents.
-   * @param minFreezedOutput The float value that the highest quantized output value after requantize.
-   * @param maxFreezedOutput 
-   * @param Toutput 
-   * @param options carries optional attributes values
-   * @return a new instance of QuantizedMatMulWithBiasAndReluAndRequantize
-   */
-  @Endpoint(describeByClass = true)
-  public static <W extends TType, T extends TType, U extends TType, V extends TType> QuantizedMatMulWithBiasAndReluAndRequantize<W> create(Scope scope, Operand<T> a, Operand<U> b, Operand<V> bias, Operand<TFloat32> minA, Operand<TFloat32> maxA, Operand<TFloat32> minB, Operand<TFloat32> maxB, Operand<TFloat32> minFreezedOutput, Operand<TFloat32> maxFreezedOutput, Class<W> Toutput, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("QuantizedMatMulWithBiasAndReluAndRequantize", scope.makeOpName("QuantizedMatMulWithBiasAndReluAndRequantize"));
-    opBuilder.addInput(a.asOutput(scope));
-    opBuilder.addInput(b.asOutput(scope));
-    opBuilder.addInput(bias.asOutput(scope));
-    opBuilder.addInput(minA.asOutput(scope));
-    opBuilder.addInput(maxA.asOutput(scope));
-    opBuilder.addInput(minB.asOutput(scope));
-    opBuilder.addInput(maxB.asOutput(scope));
-    opBuilder.addInput(minFreezedOutput.asOutput(scope));
-    opBuilder.addInput(maxFreezedOutput.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("Toutput", Toutput);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.transposeA != null) {
-          opBuilder.setAttr("transpose_a", opts.transposeA);
-        }
-        if (opts.transposeB != null) {
-          opBuilder.setAttr("transpose_b", opts.transposeB);
-        }
-        if (opts.inputQuantMode != null) {
-          opBuilder.setAttr("input_quant_mode", opts.inputQuantMode);
-        }
-      }
-    }
-    return new QuantizedMatMulWithBiasAndReluAndRequantize<W>(opBuilder.build());
-  }
-  
-  /**
-   * @param transposeA If true, `a` is transposed before multiplication.
-   */
-  public static Options transposeA(Boolean transposeA) {
-    return new Options().transposeA(transposeA);
-  }
-  
-  /**
-   * @param transposeB If true, `b` is transposed before multiplication.
-   */
-  public static Options transposeB(Boolean transposeB) {
-    return new Options().transposeB(transposeB);
-  }
-  
-  /**
-   * @param inputQuantMode Input data quantization mode. Either MIN_FIRST(default) or SCALED.
-   */
-  public static Options inputQuantMode(String inputQuantMode) {
-    return new Options().inputQuantMode(inputQuantMode);
-  }
-  
-  /**
-   */
-  public Output<W> out() {
-    return out;
-  }
-  
-  /**
-   * The float value that the lowest quantized output value represents.
-   */
-  public Output<TFloat32> minOut() {
-    return minOut;
-  }
-  
-  /**
-   * The float value that the highest quantized output value represents.
-   */
-  public Output<TFloat32> maxOut() {
-    return maxOut;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "QuantizedMatMulWithBiasAndReluAndRequantize";
-  
-  private Output<W> out;
-  private Output<TFloat32> minOut;
-  private Output<TFloat32> maxOut;
-  
-  private QuantizedMatMulWithBiasAndReluAndRequantize(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    out = operation.output(outputIdx++);
-    minOut = operation.output(outputIdx++);
-    maxOut = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/SelfAdjointEig.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/SelfAdjointEig.java
deleted file mode 100644
index 954e94c10be..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/SelfAdjointEig.java
+++ /dev/null
@@ -1,127 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.linalg;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Computes the eigen decomposition of one or more square self-adjoint matrices.
- * <p>
- * Computes the eigenvalues and (optionally) eigenvectors of each inner matrix in
- * `input` such that `input[..., :, :] = v[..., :, :] * diag(e[..., :])`. The eigenvalues
- * are sorted in non-decreasing order.
- * <pre>{@code
- * # a is a tensor.
- * # e is a tensor of eigenvalues.
- * # v is a tensor of eigenvectors.
- * e, v = self_adjoint_eig(a)
- * e = self_adjoint_eig(a, compute_v=False)
- * }</pre>
- * 
- * 
- * @param <T> data type for {@code e()} output
- */
-@Operator(group = "linalg")
-public final class SelfAdjointEig<T extends TType> extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.linalg.SelfAdjointEig}
-   */
-  public static class Options {
-    
-    /**
-     * @param computeV If `True` then eigenvectors will be computed and returned in `v`.
-     * Otherwise, only the eigenvalues will be computed.
-     */
-    public Options computeV(Boolean computeV) {
-      this.computeV = computeV;
-      return this;
-    }
-    
-    private Boolean computeV;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new SelfAdjointEig operation.
-   * 
-   * @param scope current scope
-   * @param input `Tensor` input of shape `[N, N]`.
-   * @param options carries optional attributes values
-   * @return a new instance of SelfAdjointEig
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> SelfAdjointEig<T> create(Scope scope, Operand<T> input, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("SelfAdjointEigV2", scope.makeOpName("SelfAdjointEig"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.computeV != null) {
-          opBuilder.setAttr("compute_v", opts.computeV);
-        }
-      }
-    }
-    return new SelfAdjointEig<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param computeV If `True` then eigenvectors will be computed and returned in `v`.
-   * Otherwise, only the eigenvalues will be computed.
-   */
-  public static Options computeV(Boolean computeV) {
-    return new Options().computeV(computeV);
-  }
-  
-  /**
-   * Eigenvalues. Shape is `[N]`.
-   */
-  public Output<T> e() {
-    return e;
-  }
-  
-  /**
-   * Eigenvectors. Shape is `[N, N]`.
-   */
-  public Output<T> v() {
-    return v;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "SelfAdjointEigV2";
-  
-  private Output<T> e;
-  private Output<T> v;
-  
-  private SelfAdjointEig(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    e = operation.output(outputIdx++);
-    v = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Solve.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Solve.java
deleted file mode 100644
index ab85e800b71..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Solve.java
+++ /dev/null
@@ -1,120 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.linalg;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Solves systems of linear equations.
- * <p>
- * `Matrix` is a tensor of shape `[..., M, M]` whose inner-most 2 dimensions
- * form square matrices. `Rhs` is a tensor of shape `[..., M, K]`. The `output` is
- * a tensor shape `[..., M, K]`.  If `adjoint` is `False` then each output matrix
- * satisfies `matrix[..., :, :] * output[..., :, :] = rhs[..., :, :]`.
- * If `adjoint` is `True` then each output matrix satisfies
- * `adjoint(matrix[..., :, :]) * output[..., :, :] = rhs[..., :, :]`.
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator(group = "linalg")
-public final class Solve<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.linalg.Solve}
-   */
-  public static class Options {
-    
-    /**
-     * @param adjoint Boolean indicating whether to solve with `matrix` or its (block-wise)
-     * adjoint.
-     */
-    public Options adjoint(Boolean adjoint) {
-      this.adjoint = adjoint;
-      return this;
-    }
-    
-    private Boolean adjoint;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new Solve operation.
-   * 
-   * @param scope current scope
-   * @param matrix Shape is `[..., M, M]`.
-   * @param rhs Shape is `[..., M, K]`.
-   * @param options carries optional attributes values
-   * @return a new instance of Solve
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> Solve<T> create(Scope scope, Operand<T> matrix, Operand<T> rhs, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("MatrixSolve", scope.makeOpName("Solve"));
-    opBuilder.addInput(matrix.asOutput(scope));
-    opBuilder.addInput(rhs.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.adjoint != null) {
-          opBuilder.setAttr("adjoint", opts.adjoint);
-        }
-      }
-    }
-    return new Solve<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param adjoint Boolean indicating whether to solve with `matrix` or its (block-wise)
-   * adjoint.
-   */
-  public static Options adjoint(Boolean adjoint) {
-    return new Options().adjoint(adjoint);
-  }
-  
-  /**
-   * Shape is `[..., M, K]`.
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "MatrixSolve";
-  
-  private Output<T> output;
-  
-  private Solve(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Sqrtm.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Sqrtm.java
deleted file mode 100644
index 4334b4a1534..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Sqrtm.java
+++ /dev/null
@@ -1,95 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.linalg;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Computes the matrix square root of one or more square matrices:
- * <p>
- * matmul(sqrtm(A), sqrtm(A)) = A
- * <p>
- * The input matrix should be invertible. If the input matrix is real, it should
- * have no eigenvalues which are real and negative (pairs of complex conjugate
- * eigenvalues are allowed).
- * <p>
- * The matrix square root is computed by first reducing the matrix to
- * quasi-triangular form with the real Schur decomposition. The square root
- * of the quasi-triangular matrix is then computed directly. Details of
- * the algorithm can be found in: Nicholas J. Higham, "Computing real
- * square roots of a real matrix", Linear Algebra Appl., 1987.
- * <p>
- * The input is a tensor of shape `[..., M, M]` whose inner-most 2 dimensions
- * form square matrices. The output is a tensor of the same shape as the input
- * containing the matrix square root for all input submatrices `[..., :, :]`.
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator(group = "linalg")
-public final class Sqrtm<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new Sqrtm operation.
-   * 
-   * @param scope current scope
-   * @param input Shape is `[..., M, M]`.
-   * @return a new instance of Sqrtm
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> Sqrtm<T> create(Scope scope, Operand<T> input) {
-    OperationBuilder opBuilder = scope.env().opBuilder("MatrixSquareRoot", scope.makeOpName("Sqrtm"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new Sqrtm<T>(opBuilder.build());
-  }
-  
-  /**
-   * Shape is `[..., M, M]`.
-   * <p>
-   * @compatibility(scipy)
-   * Equivalent to scipy.linalg.sqrtm
-   * @end_compatibility
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "MatrixSquareRoot";
-  
-  private Output<T> output;
-  
-  private Sqrtm(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Svd.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Svd.java
deleted file mode 100644
index d020ab02358..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Svd.java
+++ /dev/null
@@ -1,166 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.linalg;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Computes the singular value decompositions of one or more matrices.
- * <p>
- * Computes the SVD of each inner matrix in `input` such that
- * `input[..., :, :] = u[..., :, :] * diag(s[..., :, :]) * transpose(v[..., :, :])`
- * <pre>{@code
- * # a is a tensor containing a batch of matrices.
- * # s is a tensor of singular values for each matrix.
- * # u is the tensor containing the left singular vectors for each matrix.
- * # v is the tensor containing the right singular vectors for each matrix.
- * s, u, v = svd(a)
- * s, _, _ = svd(a, compute_uv=False)
- * }</pre>
- * 
- * 
- * @param <T> data type for {@code s()} output
- */
-@Operator(group = "linalg")
-public final class Svd<T extends TType> extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.linalg.Svd}
-   */
-  public static class Options {
-    
-    /**
-     * @param computeUv If true, left and right singular vectors will be
-     * computed and returned in `u` and `v`, respectively.
-     * If false, `u` and `v` are not set and should never referenced.
-     */
-    public Options computeUv(Boolean computeUv) {
-      this.computeUv = computeUv;
-      return this;
-    }
-    
-    /**
-     * @param fullMatrices If true, compute full-sized `u` and `v`. If false
-     * (the default), compute only the leading `P` singular vectors.
-     * Ignored if `compute_uv` is `False`.
-     */
-    public Options fullMatrices(Boolean fullMatrices) {
-      this.fullMatrices = fullMatrices;
-      return this;
-    }
-    
-    private Boolean computeUv;
-    private Boolean fullMatrices;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new Svd operation.
-   * 
-   * @param scope current scope
-   * @param input A tensor of shape `[..., M, N]` whose inner-most 2 dimensions
-   * form matrices of size `[M, N]`. Let `P` be the minimum of `M` and `N`.
-   * @param options carries optional attributes values
-   * @return a new instance of Svd
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> Svd<T> create(Scope scope, Operand<T> input, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Svd", scope.makeOpName("Svd"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.computeUv != null) {
-          opBuilder.setAttr("compute_uv", opts.computeUv);
-        }
-        if (opts.fullMatrices != null) {
-          opBuilder.setAttr("full_matrices", opts.fullMatrices);
-        }
-      }
-    }
-    return new Svd<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param computeUv If true, left and right singular vectors will be
-   * computed and returned in `u` and `v`, respectively.
-   * If false, `u` and `v` are not set and should never referenced.
-   */
-  public static Options computeUv(Boolean computeUv) {
-    return new Options().computeUv(computeUv);
-  }
-  
-  /**
-   * @param fullMatrices If true, compute full-sized `u` and `v`. If false
-   * (the default), compute only the leading `P` singular vectors.
-   * Ignored if `compute_uv` is `False`.
-   */
-  public static Options fullMatrices(Boolean fullMatrices) {
-    return new Options().fullMatrices(fullMatrices);
-  }
-  
-  /**
-   * Singular values. Shape is `[..., P]`.
-   */
-  public Output<T> s() {
-    return s;
-  }
-  
-  /**
-   * Left singular vectors. If `full_matrices` is `False` then shape is
-   * `[..., M, P]`; if `full_matrices` is `True` then shape is
-   * `[..., M, M]`. Undefined if `compute_uv` is `False`.
-   */
-  public Output<T> u() {
-    return u;
-  }
-  
-  /**
-   * Left singular vectors. If `full_matrices` is `False` then shape is
-   * `[..., N, P]`. If `full_matrices` is `True` then shape is `[..., N, N]`.
-   * Undefined if `compute_uv` is false.
-   */
-  public Output<T> v() {
-    return v;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Svd";
-  
-  private Output<T> s;
-  private Output<T> u;
-  private Output<T> v;
-  
-  private Svd(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    s = operation.output(outputIdx++);
-    u = operation.output(outputIdx++);
-    v = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/TensorDiag.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/TensorDiag.java
deleted file mode 100644
index bcaac11e213..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/TensorDiag.java
+++ /dev/null
@@ -1,92 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.linalg;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Returns a diagonal tensor with a given diagonal values.
- * <p>
- * Given a `diagonal`, this operation returns a tensor with the `diagonal` and
- * everything else padded with zeros. The diagonal is computed as follows:
- * <p>
- * Assume `diagonal` has dimensions [D1,..., Dk], then the output is a tensor of
- * rank 2k with dimensions [D1,..., Dk, D1,..., Dk] where:
- * <p>
- * `output[i1,..., ik, i1,..., ik] = diagonal[i1, ..., ik]` and 0 everywhere else.
- * <p>
- * For example:
- * <pre>{@code
- * # 'diagonal' is [1, 2, 3, 4]
- * tf.diag(diagonal) ==> [[1, 0, 0, 0]
- *                        [0, 2, 0, 0]
- *                        [0, 0, 3, 0]
- *                        [0, 0, 0, 4]]
- * }</pre>
- * 
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator(group = "linalg")
-public final class TensorDiag<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new TensorDiag operation.
-   * 
-   * @param scope current scope
-   * @param diagonal Rank k tensor where k is at most 1.
-   * @return a new instance of TensorDiag
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> TensorDiag<T> create(Scope scope, Operand<T> diagonal) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Diag", scope.makeOpName("TensorDiag"));
-    opBuilder.addInput(diagonal.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new TensorDiag<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Diag";
-  
-  private Output<T> output;
-  
-  private TensorDiag(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/TensorDiagPart.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/TensorDiagPart.java
deleted file mode 100644
index ef2a3b258c5..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/TensorDiagPart.java
+++ /dev/null
@@ -1,94 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.linalg;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Returns the diagonal part of the tensor.
- * <p>
- * This operation returns a tensor with the `diagonal` part
- * of the `input`. The `diagonal` part is computed as follows:
- * <p>
- * Assume `input` has dimensions `[D1,..., Dk, D1,..., Dk]`, then the output is a
- * tensor of rank `k` with dimensions `[D1,..., Dk]` where:
- * <p>
- * `diagonal[i1,..., ik] = input[i1, ..., ik, i1,..., ik]`.
- * <p>
- * For example:
- * <pre>{@code
- * # 'input' is [[1, 0, 0, 0]
- *               [0, 2, 0, 0]
- *               [0, 0, 3, 0]
- *               [0, 0, 0, 4]]
- * 
- * tf.diag_part(input) ==> [1, 2, 3, 4]
- * }</pre>
- * 
- * 
- * @param <T> data type for {@code diagonal()} output
- */
-@Operator(group = "linalg")
-public final class TensorDiagPart<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new TensorDiagPart operation.
-   * 
-   * @param scope current scope
-   * @param input Rank k tensor where k is even and not zero.
-   * @return a new instance of TensorDiagPart
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> TensorDiagPart<T> create(Scope scope, Operand<T> input) {
-    OperationBuilder opBuilder = scope.env().opBuilder("DiagPart", scope.makeOpName("TensorDiagPart"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new TensorDiagPart<T>(opBuilder.build());
-  }
-  
-  /**
-   * The extracted diagonal.
-   */
-  public Output<T> diagonal() {
-    return diagonal;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return diagonal;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "DiagPart";
-  
-  private Output<T> diagonal;
-  
-  private TensorDiagPart(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    diagonal = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Transpose.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Transpose.java
deleted file mode 100644
index 7662ff4e3d6..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Transpose.java
+++ /dev/null
@@ -1,80 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.linalg;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * Shuffle dimensions of x according to a permutation.
- * <p>
- * The output `y` has the same rank as `x`. The shapes of `x` and `y` satisfy:
- *   `y.shape[i] == x.shape[perm[i]] for i in [0, 1, ..., rank(x) - 1]`
- * 
- * @param <T> data type for {@code y()} output
- */
-@Operator(group = "linalg")
-public final class Transpose<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new Transpose operation.
-   * 
-   * @param scope current scope
-   * @param x 
-   * @param perm 
-   * @return a new instance of Transpose
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> Transpose<T> create(Scope scope, Operand<T> x, Operand<U> perm) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Transpose", scope.makeOpName("Transpose"));
-    opBuilder.addInput(x.asOutput(scope));
-    opBuilder.addInput(perm.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new Transpose<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> y() {
-    return y;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return y;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Transpose";
-  
-  private Output<T> y;
-  
-  private Transpose(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    y = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/TriangularSolve.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/TriangularSolve.java
deleted file mode 100644
index 9e61c01a83a..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/TriangularSolve.java
+++ /dev/null
@@ -1,189 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.linalg;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Solves systems of linear equations with upper or lower triangular matrices by backsubstitution.
- * <p>
- * 
- * `matrix` is a tensor of shape `[..., M, M]` whose inner-most 2 dimensions form
- * square matrices. If `lower` is `True` then the strictly upper triangular part
- * of each inner-most matrix is assumed to be zero and not accessed.
- * If `lower` is False then the strictly lower triangular part of each inner-most
- * matrix is assumed to be zero and not accessed.
- * `rhs` is a tensor of shape `[..., M, N]`.
- * <p>
- * The output is a tensor of shape `[..., M, N]`. If `adjoint` is
- * `True` then the innermost matrices in `output` satisfy matrix equations
- * `matrix[..., :, :] * output[..., :, :] = rhs[..., :, :]`.
- * If `adjoint` is `False` then the strictly then the  innermost matrices in
- * `output` satisfy matrix equations
- * `adjoint(matrix[..., i, k]) * output[..., k, j] = rhs[..., i, j]`.
- * <p>
- * Note, the batch shapes for the inputs only need to broadcast.
- * <p>
- * Example:
- * <pre>{@code
- * a = tf.constant([[3,  0,  0,  0],
- *                  [2,  1,  0,  0],
- *                  [1,  0,  1,  0],
- *                  [1,  1,  1,  1]], dtype=tf.float32)
- * 
- * b = tf.constant([[4],
- *                  [2],
- *                  [4],
- *                  [2]], dtype=tf.float32)
- * 
- * x = tf.linalg.triangular_solve(a, b, lower=True)
- * x
- * # <tf.Tensor: shape=(4, 1), dtype=float32, numpy=
- * # array([[ 1.3333334 ],
- * #        [-0.66666675],
- * #        [ 2.6666665 ],
- * #        [-1.3333331 ]], dtype=float32)>
- * 
- * # in python3 one can use `a@x`
- * tf.matmul(a, x)
- * # <tf.Tensor: shape=(4, 1), dtype=float32, numpy=
- * # array([[4.       ],
- * #        [2.       ],
- * #        [4.       ],
- * #        [1.9999999]], dtype=float32)>
- * }</pre>
- * 
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator(group = "linalg")
-public final class TriangularSolve<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.linalg.TriangularSolve}
-   */
-  public static class Options {
-    
-    /**
-     * @param lower Boolean indicating whether the innermost matrices in `matrix` are
-     * lower or upper triangular.
-     */
-    public Options lower(Boolean lower) {
-      this.lower = lower;
-      return this;
-    }
-    
-    /**
-     * @param adjoint Boolean indicating whether to solve with `matrix` or its (block-wise)
-     *          adjoint.
-     * <p>
-     * @compatibility(numpy)
-     * Equivalent to scipy.linalg.solve_triangular
-     * @end_compatibility
-     */
-    public Options adjoint(Boolean adjoint) {
-      this.adjoint = adjoint;
-      return this;
-    }
-    
-    private Boolean lower;
-    private Boolean adjoint;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new TriangularSolve operation.
-   * 
-   * @param scope current scope
-   * @param matrix Shape is `[..., M, M]`.
-   * @param rhs Shape is `[..., M, K]`.
-   * @param options carries optional attributes values
-   * @return a new instance of TriangularSolve
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> TriangularSolve<T> create(Scope scope, Operand<T> matrix, Operand<T> rhs, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("MatrixTriangularSolve", scope.makeOpName("TriangularSolve"));
-    opBuilder.addInput(matrix.asOutput(scope));
-    opBuilder.addInput(rhs.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.lower != null) {
-          opBuilder.setAttr("lower", opts.lower);
-        }
-        if (opts.adjoint != null) {
-          opBuilder.setAttr("adjoint", opts.adjoint);
-        }
-      }
-    }
-    return new TriangularSolve<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param lower Boolean indicating whether the innermost matrices in `matrix` are
-   * lower or upper triangular.
-   */
-  public static Options lower(Boolean lower) {
-    return new Options().lower(lower);
-  }
-  
-  /**
-   * @param adjoint Boolean indicating whether to solve with `matrix` or its (block-wise)
-   *          adjoint.
-   * <p>
-   * @compatibility(numpy)
-   * Equivalent to scipy.linalg.solve_triangular
-   * @end_compatibility
-   */
-  public static Options adjoint(Boolean adjoint) {
-    return new Options().adjoint(adjoint);
-  }
-  
-  /**
-   * Shape is `[..., M, K]`.
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "MatrixTriangularSolve";
-  
-  private Output<T> output;
-  
-  private TriangularSolve(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/TridiagonalMatMul.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/TridiagonalMatMul.java
deleted file mode 100644
index c700a24f346..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/TridiagonalMatMul.java
+++ /dev/null
@@ -1,86 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.linalg;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Calculate product with tridiagonal matrix.
- * <p>
- * Calculates product of two matrices, where left matrix is a tridiagonal matrix.
- * 
- * @param <T> data type for {@code output()} output
- */
-public final class TridiagonalMatMul<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new TridiagonalMatMul operation.
-   * 
-   * @param scope current scope
-   * @param superdiag Tensor of shape `[..., 1, M]`, representing superdiagonals of
-   * tri-diagonal matrices to the left of multiplication. Last element is ignored.
-   * @param maindiag Tensor of shape `[..., 1, M]`, representing main diagonals of tri-diagonal
-   * matrices to the left of multiplication.
-   * @param subdiag Tensor of shape `[..., 1, M]`, representing subdiagonals of tri-diagonal
-   * matrices to the left of multiplication. First element is ignored.
-   * @param rhs Tensor of shape `[..., M, N]`, representing MxN matrices to the right of
-   * multiplication.
-   * @return a new instance of TridiagonalMatMul
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> TridiagonalMatMul<T> create(Scope scope, Operand<T> superdiag, Operand<T> maindiag, Operand<T> subdiag, Operand<T> rhs) {
-    OperationBuilder opBuilder = scope.env().opBuilder("TridiagonalMatMul", scope.makeOpName("TridiagonalMatMul"));
-    opBuilder.addInput(superdiag.asOutput(scope));
-    opBuilder.addInput(maindiag.asOutput(scope));
-    opBuilder.addInput(subdiag.asOutput(scope));
-    opBuilder.addInput(rhs.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new TridiagonalMatMul<T>(opBuilder.build());
-  }
-  
-  /**
-   * Tensor of shape `[..., M, N]` containing the product.
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "TridiagonalMatMul";
-  
-  private Output<T> output;
-  
-  private TridiagonalMatMul(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/TridiagonalSolve.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/TridiagonalSolve.java
deleted file mode 100644
index 6865f480ef2..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/TridiagonalSolve.java
+++ /dev/null
@@ -1,124 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.linalg;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Solves tridiagonal systems of equations.
- * <p>
- *   Solves tridiagonal systems of equations.
- *   Supports batch dimensions and multiple right-hand sides per each left-hand
- *   side.
- *   On CPU, solution is computed via Gaussian elimination with or without partial
- *   pivoting, depending on `partial_pivoting` attribute. On GPU, Nvidia's cuSPARSE
- *   library is used: https://docs.nvidia.com/cuda/cusparse/index.html#gtsv
- *   Partial pivoting is not yet supported by XLA backends.
- * 
- * @param <T> data type for {@code output()} output
- */
-public final class TridiagonalSolve<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.linalg.TridiagonalSolve}
-   */
-  public static class Options {
-    
-    /**
-     * @param partialPivoting Whether to apply partial pivoting. Partial pivoting makes the procedure more
-     * stable, but slower.
-     */
-    public Options partialPivoting(Boolean partialPivoting) {
-      this.partialPivoting = partialPivoting;
-      return this;
-    }
-    
-    private Boolean partialPivoting;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new TridiagonalSolve operation.
-   * 
-   * @param scope current scope
-   * @param diagonals Tensor of shape `[..., 3, M]` whose innermost 2 dimensions represent the
-   * tridiagonal matrices with three rows being the superdiagonal, diagonals, and
-   * subdiagonals, in order. The last element of the superdiagonal and the first
-   * element of the subdiagonal is ignored.
-   * @param rhs Tensor of shape `[..., M, K]`, representing K right-hand sides per each
-   * left-hand side.
-   * @param options carries optional attributes values
-   * @return a new instance of TridiagonalSolve
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> TridiagonalSolve<T> create(Scope scope, Operand<T> diagonals, Operand<T> rhs, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("TridiagonalSolve", scope.makeOpName("TridiagonalSolve"));
-    opBuilder.addInput(diagonals.asOutput(scope));
-    opBuilder.addInput(rhs.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.partialPivoting != null) {
-          opBuilder.setAttr("partial_pivoting", opts.partialPivoting);
-        }
-      }
-    }
-    return new TridiagonalSolve<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param partialPivoting Whether to apply partial pivoting. Partial pivoting makes the procedure more
-   * stable, but slower.
-   */
-  public static Options partialPivoting(Boolean partialPivoting) {
-    return new Options().partialPivoting(partialPivoting);
-  }
-  
-  /**
-   * Tensor of shape `[..., M, K]` containing the solutions
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "TridiagonalSolve";
-  
-  private Output<T> output;
-  
-  private TridiagonalSolve(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/CSRSparseMatrixComponents.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/CSRSparseMatrixComponents.java
deleted file mode 100644
index 732a5641ab0..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/CSRSparseMatrixComponents.java
+++ /dev/null
@@ -1,95 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.linalg.sparse;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.family.TType;
-
-/**
- * Reads out the CSR components at batch `index`.
- * <p>
- * This op is meant only for debugging / testing, and its interface is not expected
- * to be stable.
- * 
- * @param <T> data type for {@code values()} output
- */
-public final class CSRSparseMatrixComponents<T extends TType> extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new CSRSparseMatrixComponents operation.
-   * 
-   * @param scope current scope
-   * @param csrSparseMatrix A batched CSRSparseMatrix.
-   * @param index The index in `csr_sparse_matrix`'s batch.
-   * @param type 
-   * @return a new instance of CSRSparseMatrixComponents
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> CSRSparseMatrixComponents<T> create(Scope scope, Operand<?> csrSparseMatrix, Operand<TInt32> index, Class<T> type) {
-    OperationBuilder opBuilder = scope.env().opBuilder("CSRSparseMatrixComponents", scope.makeOpName("CSRSparseMatrixComponents"));
-    opBuilder.addInput(csrSparseMatrix.asOutput(scope));
-    opBuilder.addInput(index.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("type", type);
-    return new CSRSparseMatrixComponents<T>(opBuilder.build());
-  }
-  
-  /**
-   * An array containing CSR matrix row pointers.
-   */
-  public Output<TInt32> rowPtrs() {
-    return rowPtrs;
-  }
-  
-  /**
-   * An array containing CSR matrix column indices.
-   */
-  public Output<TInt32> colInds() {
-    return colInds;
-  }
-  
-  /**
-   * An array containing CSR matrix nonzero values.
-   */
-  public Output<T> values() {
-    return values;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "CSRSparseMatrixComponents";
-  
-  private Output<TInt32> rowPtrs;
-  private Output<TInt32> colInds;
-  private Output<T> values;
-  
-  private CSRSparseMatrixComponents(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    rowPtrs = operation.output(outputIdx++);
-    colInds = operation.output(outputIdx++);
-    values = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/CSRSparseMatrixToDense.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/CSRSparseMatrixToDense.java
deleted file mode 100644
index b0e0275d49c..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/CSRSparseMatrixToDense.java
+++ /dev/null
@@ -1,76 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.linalg.sparse;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Convert a (possibly batched) CSRSparseMatrix to dense.
- * 
- * @param <T> data type for {@code denseOutput()} output
- */
-public final class CSRSparseMatrixToDense<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new CSRSparseMatrixToDense operation.
-   * 
-   * @param scope current scope
-   * @param sparseInput A batched CSRSparseMatrix.
-   * @param type 
-   * @return a new instance of CSRSparseMatrixToDense
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> CSRSparseMatrixToDense<T> create(Scope scope, Operand<?> sparseInput, Class<T> type) {
-    OperationBuilder opBuilder = scope.env().opBuilder("CSRSparseMatrixToDense", scope.makeOpName("CSRSparseMatrixToDense"));
-    opBuilder.addInput(sparseInput.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("type", type);
-    return new CSRSparseMatrixToDense<T>(opBuilder.build());
-  }
-  
-  /**
-   * A dense tensor.
-   */
-  public Output<T> denseOutput() {
-    return denseOutput;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return denseOutput;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "CSRSparseMatrixToDense";
-  
-  private Output<T> denseOutput;
-  
-  private CSRSparseMatrixToDense(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    denseOutput = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/CSRSparseMatrixToSparseTensor.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/CSRSparseMatrixToSparseTensor.java
deleted file mode 100644
index 5b89bc86a70..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/CSRSparseMatrixToSparseTensor.java
+++ /dev/null
@@ -1,90 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.linalg.sparse;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
-
-/**
- * Converts a (possibly batched) CSRSparesMatrix to a SparseTensor.
- * 
- * @param <T> data type for {@code values()} output
- */
-public final class CSRSparseMatrixToSparseTensor<T extends TType> extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new CSRSparseMatrixToSparseTensor operation.
-   * 
-   * @param scope current scope
-   * @param sparseMatrix A (possibly batched) CSRSparseMatrix.
-   * @param type 
-   * @return a new instance of CSRSparseMatrixToSparseTensor
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> CSRSparseMatrixToSparseTensor<T> create(Scope scope, Operand<?> sparseMatrix, Class<T> type) {
-    OperationBuilder opBuilder = scope.env().opBuilder("CSRSparseMatrixToSparseTensor", scope.makeOpName("CSRSparseMatrixToSparseTensor"));
-    opBuilder.addInput(sparseMatrix.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("type", type);
-    return new CSRSparseMatrixToSparseTensor<T>(opBuilder.build());
-  }
-  
-  /**
-   * SparseTensor indices.
-   */
-  public Output<TInt64> indices() {
-    return indices;
-  }
-  
-  /**
-   * SparseTensor values.
-   */
-  public Output<T> values() {
-    return values;
-  }
-  
-  /**
-   * SparseTensor dense shape.
-   */
-  public Output<TInt64> denseShape() {
-    return denseShape;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "CSRSparseMatrixToSparseTensor";
-  
-  private Output<TInt64> indices;
-  private Output<T> values;
-  private Output<TInt64> denseShape;
-  
-  private CSRSparseMatrixToSparseTensor(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    indices = operation.output(outputIdx++);
-    values = operation.output(outputIdx++);
-    denseShape = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/DenseToCSRSparseMatrix.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/DenseToCSRSparseMatrix.java
deleted file mode 100644
index cdf128a932c..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/DenseToCSRSparseMatrix.java
+++ /dev/null
@@ -1,76 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.linalg.sparse;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
-
-/**
- * Converts a dense tensor to a (possibly batched) CSRSparseMatrix.
- */
-public final class DenseToCSRSparseMatrix extends RawOp implements Operand<TType> {
-  
-  /**
-   * Factory method to create a class wrapping a new DenseToCSRSparseMatrix operation.
-   * 
-   * @param scope current scope
-   * @param denseInput A Dense tensor.
-   * @param indices Indices of nonzero elements.
-   * @return a new instance of DenseToCSRSparseMatrix
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> DenseToCSRSparseMatrix create(Scope scope, Operand<T> denseInput, Operand<TInt64> indices) {
-    OperationBuilder opBuilder = scope.env().opBuilder("DenseToCSRSparseMatrix", scope.makeOpName("DenseToCSRSparseMatrix"));
-    opBuilder.addInput(denseInput.asOutput(scope));
-    opBuilder.addInput(indices.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new DenseToCSRSparseMatrix(opBuilder.build());
-  }
-  
-  /**
-   * A (possibly batched) CSRSparseMatrix.
-   */
-  public Output<?> sparseOutput() {
-    return sparseOutput;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) sparseOutput;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "DenseToCSRSparseMatrix";
-  
-  private Output<?> sparseOutput;
-  
-  private DenseToCSRSparseMatrix(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    sparseOutput = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixAdd.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixAdd.java
deleted file mode 100644
index 4b50062963d..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixAdd.java
+++ /dev/null
@@ -1,82 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.linalg.sparse;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Sparse addition of two CSR matrices, C = alpha * A + beta * B.
- * <p>
- * The gradients of SparseMatrixAdd outputs with respect to alpha and beta are not
- * currently defined (TensorFlow will return zeros for these entries).
- */
-public final class SparseMatrixAdd extends RawOp implements Operand<TType> {
-  
-  /**
-   * Factory method to create a class wrapping a new SparseMatrixAdd operation.
-   * 
-   * @param scope current scope
-   * @param a A CSRSparseMatrix.
-   * @param b A CSRSparseMatrix.
-   * @param alpha A constant scalar.
-   * @param beta A constant scalar.
-   * @return a new instance of SparseMatrixAdd
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> SparseMatrixAdd create(Scope scope, Operand<?> a, Operand<?> b, Operand<T> alpha, Operand<T> beta) {
-    OperationBuilder opBuilder = scope.env().opBuilder("SparseMatrixAdd", scope.makeOpName("SparseMatrixAdd"));
-    opBuilder.addInput(a.asOutput(scope));
-    opBuilder.addInput(b.asOutput(scope));
-    opBuilder.addInput(alpha.asOutput(scope));
-    opBuilder.addInput(beta.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new SparseMatrixAdd(opBuilder.build());
-  }
-  
-  /**
-   * A CSRSparseMatrix.
-   */
-  public Output<?> c() {
-    return c;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) c;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "SparseMatrixAdd";
-  
-  private Output<?> c;
-  
-  private SparseMatrixAdd(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    c = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixMatMul.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixMatMul.java
deleted file mode 100644
index 1668438105b..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixMatMul.java
+++ /dev/null
@@ -1,230 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.linalg.sparse;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Matrix-multiplies a sparse matrix with a dense matrix.
- * <p>
- * Returns a dense matrix.
- * For inputs A and B, where A is CSR and B is dense; this op returns a dense C;
- * <p>
- * If transpose_output is false, returns:
- * <pre>{@code
- *   C = A . B
- * }</pre>
- * If transpose_output is `true`, returns:
- * <pre>{@code
- *   C = transpose(A . B) = transpose(B) . transpose(A)
- * }</pre>
- * where the transposition is performed along the two innermost (matrix)
- * dimensions.
- * <p>
- * If conjugate_output is `true`, returns:
- * <pre>{@code
- *   C = conjugate(A . B) = conjugate(A) . conjugate(B)
- * }</pre>
- * If both conjugate_output and transpose_output are `true`, returns:
- * <pre>{@code
- *   C = conjugate(transpose(A . B)) = conjugate(transpose(B)) .
- *                                     conjugate(transpose(A))
- * }</pre>
- * 
- * 
- * @param <T> data type for {@code output()} output
- */
-public final class SparseMatrixMatMul<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.linalg.sparse.SparseMatrixMatMul}
-   */
-  public static class Options {
-    
-    /**
-     * @param transposeA Indicates whether `a` should be transposed.
-     */
-    public Options transposeA(Boolean transposeA) {
-      this.transposeA = transposeA;
-      return this;
-    }
-    
-    /**
-     * @param transposeB Indicates whether `b` should be transposed.
-     */
-    public Options transposeB(Boolean transposeB) {
-      this.transposeB = transposeB;
-      return this;
-    }
-    
-    /**
-     * @param adjointA Indicates whether `a` should be conjugate-transposed.
-     */
-    public Options adjointA(Boolean adjointA) {
-      this.adjointA = adjointA;
-      return this;
-    }
-    
-    /**
-     * @param adjointB Indicates whether `b` should be conjugate-transposed.
-     */
-    public Options adjointB(Boolean adjointB) {
-      this.adjointB = adjointB;
-      return this;
-    }
-    
-    /**
-     * @param transposeOutput Transposes the product of `a` and `b`.
-     */
-    public Options transposeOutput(Boolean transposeOutput) {
-      this.transposeOutput = transposeOutput;
-      return this;
-    }
-    
-    /**
-     * @param conjugateOutput Conjugates the product of `a` and `b`.
-     */
-    public Options conjugateOutput(Boolean conjugateOutput) {
-      this.conjugateOutput = conjugateOutput;
-      return this;
-    }
-    
-    private Boolean transposeA;
-    private Boolean transposeB;
-    private Boolean adjointA;
-    private Boolean adjointB;
-    private Boolean transposeOutput;
-    private Boolean conjugateOutput;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new SparseMatrixMatMul operation.
-   * 
-   * @param scope current scope
-   * @param a A CSRSparseMatrix.
-   * @param b A dense tensor.
-   * @param options carries optional attributes values
-   * @return a new instance of SparseMatrixMatMul
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> SparseMatrixMatMul<T> create(Scope scope, Operand<?> a, Operand<T> b, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("SparseMatrixMatMul", scope.makeOpName("SparseMatrixMatMul"));
-    opBuilder.addInput(a.asOutput(scope));
-    opBuilder.addInput(b.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.transposeA != null) {
-          opBuilder.setAttr("transpose_a", opts.transposeA);
-        }
-        if (opts.transposeB != null) {
-          opBuilder.setAttr("transpose_b", opts.transposeB);
-        }
-        if (opts.adjointA != null) {
-          opBuilder.setAttr("adjoint_a", opts.adjointA);
-        }
-        if (opts.adjointB != null) {
-          opBuilder.setAttr("adjoint_b", opts.adjointB);
-        }
-        if (opts.transposeOutput != null) {
-          opBuilder.setAttr("transpose_output", opts.transposeOutput);
-        }
-        if (opts.conjugateOutput != null) {
-          opBuilder.setAttr("conjugate_output", opts.conjugateOutput);
-        }
-      }
-    }
-    return new SparseMatrixMatMul<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param transposeA Indicates whether `a` should be transposed.
-   */
-  public static Options transposeA(Boolean transposeA) {
-    return new Options().transposeA(transposeA);
-  }
-  
-  /**
-   * @param transposeB Indicates whether `b` should be transposed.
-   */
-  public static Options transposeB(Boolean transposeB) {
-    return new Options().transposeB(transposeB);
-  }
-  
-  /**
-   * @param adjointA Indicates whether `a` should be conjugate-transposed.
-   */
-  public static Options adjointA(Boolean adjointA) {
-    return new Options().adjointA(adjointA);
-  }
-  
-  /**
-   * @param adjointB Indicates whether `b` should be conjugate-transposed.
-   */
-  public static Options adjointB(Boolean adjointB) {
-    return new Options().adjointB(adjointB);
-  }
-  
-  /**
-   * @param transposeOutput Transposes the product of `a` and `b`.
-   */
-  public static Options transposeOutput(Boolean transposeOutput) {
-    return new Options().transposeOutput(transposeOutput);
-  }
-  
-  /**
-   * @param conjugateOutput Conjugates the product of `a` and `b`.
-   */
-  public static Options conjugateOutput(Boolean conjugateOutput) {
-    return new Options().conjugateOutput(conjugateOutput);
-  }
-  
-  /**
-   * A dense output tensor.
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "SparseMatrixMatMul";
-  
-  private Output<T> output;
-  
-  private SparseMatrixMatMul(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixMul.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixMul.java
deleted file mode 100644
index d378dac7e35..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixMul.java
+++ /dev/null
@@ -1,84 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.linalg.sparse;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Element-wise multiplication of a sparse matrix with a dense tensor.
- * <p>
- * Returns a sparse matrix.
- * <p>
- * The dense tensor `b` may be either a scalar; otherwise `a` must be a rank-3
- * `SparseMatrix`; in this case `b` must be shaped `[batch_size, 1, 1]` and the
- * multiply operation broadcasts.
- * <p>
- * <b>NOTE</b> even if `b` is zero, the sparsity structure of the output does not
- * change.
- */
-public final class SparseMatrixMul extends RawOp implements Operand<TType> {
-  
-  /**
-   * Factory method to create a class wrapping a new SparseMatrixMul operation.
-   * 
-   * @param scope current scope
-   * @param a A CSRSparseMatrix.
-   * @param b A dense tensor.
-   * @return a new instance of SparseMatrixMul
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> SparseMatrixMul create(Scope scope, Operand<?> a, Operand<T> b) {
-    OperationBuilder opBuilder = scope.env().opBuilder("SparseMatrixMul", scope.makeOpName("SparseMatrixMul"));
-    opBuilder.addInput(a.asOutput(scope));
-    opBuilder.addInput(b.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new SparseMatrixMul(opBuilder.build());
-  }
-  
-  /**
-   * A dense output tensor.
-   */
-  public Output<?> output() {
-    return output;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "SparseMatrixMul";
-  
-  private Output<?> output;
-  
-  private SparseMatrixMul(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixNNZ.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixNNZ.java
deleted file mode 100644
index b17c3549720..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixNNZ.java
+++ /dev/null
@@ -1,72 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.linalg.sparse;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
-
-/**
- * Returns the number of nonzeroes of `sparse_matrix`.
- */
-public final class SparseMatrixNNZ extends RawOp implements Operand<TInt32> {
-  
-  /**
-   * Factory method to create a class wrapping a new SparseMatrixNNZ operation.
-   * 
-   * @param scope current scope
-   * @param sparseMatrix A CSRSparseMatrix.
-   * @return a new instance of SparseMatrixNNZ
-   */
-  @Endpoint(describeByClass = true)
-  public static SparseMatrixNNZ create(Scope scope, Operand<?> sparseMatrix) {
-    OperationBuilder opBuilder = scope.env().opBuilder("SparseMatrixNNZ", scope.makeOpName("SparseMatrixNNZ"));
-    opBuilder.addInput(sparseMatrix.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new SparseMatrixNNZ(opBuilder.build());
-  }
-  
-  /**
-   * The number of nonzeroes of `sparse_matrix`.
-   */
-  public Output<TInt32> nnz() {
-    return nnz;
-  }
-  
-  @Override
-  public Output<TInt32> asOutput(Scope scope) {
-    return nnz;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "SparseMatrixNNZ";
-  
-  private Output<TInt32> nnz;
-  
-  private SparseMatrixNNZ(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    nnz = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixOrderingAMD.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixOrderingAMD.java
deleted file mode 100644
index ad8264db7e3..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixOrderingAMD.java
+++ /dev/null
@@ -1,120 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.linalg.sparse;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
-
-/**
- * Computes the Approximate Minimum Degree (AMD) ordering of `input`.
- * <p>
- * Computes the Approximate Minimum Degree (AMD) ordering for a sparse matrix.
- * <p>
- * The returned permutation may be used to permute the rows and columns of the
- * given sparse matrix. This typically results in permuted sparse matrix's sparse
- * Cholesky (or other decompositions) in having fewer zero fill-in compared to
- * decomposition of the original matrix.
- * <p>
- * The input sparse matrix may have rank 2 or rank 3. The output Tensor,
- * representing would then have rank 1 or 2 respectively, with the same batch
- * shape as the input.
- * <p>
- * Each component of the input sparse matrix must represent a square symmetric
- * matrix; only the lower triangular part of the matrix is read. The values of the
- * sparse matrix does not affect the returned permutation, only the sparsity
- * pattern of the sparse matrix is used. Hence, a single AMD ordering may be
- * reused for the Cholesky decompositions of sparse matrices with the same sparsity
- * pattern but with possibly different values.
- * <p>
- * Each batch component of the output permutation represents a permutation of `N`
- * elements, where the input sparse matrix components each have `N` rows. That is,
- * the component contains each of the integers `{0, .. N-1}` exactly once. The
- * `i`th element represents the row index that the `i`th row maps to.
- * <p>
- * Usage example:
- * <pre>{@code
- *     from tensorflow.python.ops.linalg.sparse import sparse_csr_matrix_ops
- * 
- *     a_indices = np.array([[0, 0], [1, 1], [2, 1], [2, 2], [3, 3]])
- *     a_values = np.array([1.0, 2.0, 1.0, 3.0, 4.0], np.float32)
- *     a_dense_shape = [4, 4]
- * 
- *     with tf.Session() as sess:
- *       # Define (COO format) SparseTensor over Numpy array.
- *       a_st = tf.sparse.SparseTensor(a_indices, a_values, a_dense_shape)
- * 
- *       # Convert SparseTensors to CSR SparseMatrix.
- *       a_sm = sparse_csr_matrix_ops.sparse_tensor_to_csr_sparse_matrix(
- *           a_st.indices, a_st.values, a_st.dense_shape)
- * 
- *       # Obtain the AMD Ordering for the CSR SparseMatrix.
- *       ordering_amd = sparse_csr_matrix_ops.sparse_matrix_ordering_amd(sparse_matrix)
- * 
- *       ordering_amd_value = sess.run(ordering_amd)
- * }</pre>
- * `ordering_amd_value` stores the AMD ordering: `[1 2 3 0]`.
- * <p>
- * input: A `CSRSparseMatrix`.
- */
-public final class SparseMatrixOrderingAMD extends RawOp implements Operand<TInt32> {
-  
-  /**
-   * Factory method to create a class wrapping a new SparseMatrixOrderingAMD operation.
-   * 
-   * @param scope current scope
-   * @param input A `CSRSparseMatrix`.
-   * @return a new instance of SparseMatrixOrderingAMD
-   */
-  @Endpoint(describeByClass = true)
-  public static SparseMatrixOrderingAMD create(Scope scope, Operand<?> input) {
-    OperationBuilder opBuilder = scope.env().opBuilder("SparseMatrixOrderingAMD", scope.makeOpName("SparseMatrixOrderingAMD"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new SparseMatrixOrderingAMD(opBuilder.build());
-  }
-  
-  /**
-   * The Approximate Minimum Degree (AMD) ordering of `input`.
-   */
-  public Output<TInt32> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<TInt32> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "SparseMatrixOrderingAMD";
-  
-  private Output<TInt32> output;
-  
-  private SparseMatrixOrderingAMD(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixSoftmax.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixSoftmax.java
deleted file mode 100644
index 610706e5fa3..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixSoftmax.java
+++ /dev/null
@@ -1,82 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.linalg.sparse;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * Calculates the softmax of a CSRSparseMatrix.
- * <p>
- * Calculate the softmax of the innermost dimensions of a SparseMatrix.
- * <p>
- * Missing values are treated as `-inf` (i.e., logits of zero probability); and
- * the output has the same sparsity structure as the input (though missing values
- * in the output may now be treated as having probability zero).
- */
-public final class SparseMatrixSoftmax extends RawOp implements Operand<TType> {
-  
-  /**
-   * Factory method to create a class wrapping a new SparseMatrixSoftmax operation.
-   * 
-   * @param scope current scope
-   * @param logits A CSRSparseMatrix.
-   * @param type 
-   * @return a new instance of SparseMatrixSoftmax
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> SparseMatrixSoftmax create(Scope scope, Operand<?> logits, Class<T> type) {
-    OperationBuilder opBuilder = scope.env().opBuilder("SparseMatrixSoftmax", scope.makeOpName("SparseMatrixSoftmax"));
-    opBuilder.addInput(logits.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("type", type);
-    return new SparseMatrixSoftmax(opBuilder.build());
-  }
-  
-  /**
-   * A CSRSparseMatrix.
-   */
-  public Output<?> softmax() {
-    return softmax;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) softmax;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "SparseMatrixSoftmax";
-  
-  private Output<?> softmax;
-  
-  private SparseMatrixSoftmax(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    softmax = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixSoftmaxGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixSoftmaxGrad.java
deleted file mode 100644
index 4cfe45ff40d..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixSoftmaxGrad.java
+++ /dev/null
@@ -1,78 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.linalg.sparse;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * Calculates the gradient of the SparseMatrixSoftmax op.
- */
-public final class SparseMatrixSoftmaxGrad extends RawOp implements Operand<TType> {
-  
-  /**
-   * Factory method to create a class wrapping a new SparseMatrixSoftmaxGrad operation.
-   * 
-   * @param scope current scope
-   * @param softmax A CSRSparseMatrix.
-   * @param gradSoftmax The gradient of `softmax`.
-   * @param type 
-   * @return a new instance of SparseMatrixSoftmaxGrad
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> SparseMatrixSoftmaxGrad create(Scope scope, Operand<?> softmax, Operand<?> gradSoftmax, Class<T> type) {
-    OperationBuilder opBuilder = scope.env().opBuilder("SparseMatrixSoftmaxGrad", scope.makeOpName("SparseMatrixSoftmaxGrad"));
-    opBuilder.addInput(softmax.asOutput(scope));
-    opBuilder.addInput(gradSoftmax.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("type", type);
-    return new SparseMatrixSoftmaxGrad(opBuilder.build());
-  }
-  
-  /**
-   * The output gradient.
-   */
-  public Output<?> gradient() {
-    return gradient;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) gradient;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "SparseMatrixSoftmaxGrad";
-  
-  private Output<?> gradient;
-  
-  private SparseMatrixSoftmaxGrad(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    gradient = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixSparseCholesky.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixSparseCholesky.java
deleted file mode 100644
index 173275ccf57..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixSparseCholesky.java
+++ /dev/null
@@ -1,149 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.linalg.sparse;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.family.TType;
-
-/**
- * Computes the sparse Cholesky decomposition of `input`.
- * <p>
- * Computes the Sparse Cholesky decomposition of a sparse matrix, with the given
- * fill-in reducing permutation.
- * <p>
- * The input sparse matrix and the fill-in reducing permutation `permutation` must
- * have compatible shapes. If the sparse matrix has rank 3; with the batch
- * dimension `B`, then the `permutation` must be of rank 2; with the same batch
- * dimension `B`. There is no support for broadcasting.
- * <p>
- * Furthermore, each component vector of `permutation` must be of length `N`,
- * containing each of the integers {0, 1, ..., N - 1} exactly once, where `N` is
- * the number of rows of each component of the sparse matrix.
- * <p>
- * Each component of the input sparse matrix must represent a symmetric positive
- * definite (SPD) matrix; although only the lower triangular part of the matrix is
- * read. If any individual component is not SPD, then an InvalidArgument error is
- * thrown.
- * <p>
- * The returned sparse matrix has the same dense shape as the input sparse matrix.
- * For each component `A` of the input sparse matrix, the corresponding output
- * sparse matrix represents `L`, the lower triangular Cholesky factor satisfying
- * the following identity:
- * <pre>{@code
- *   A = L * Lt
- * }</pre>
- * where Lt denotes the transpose of L (or its conjugate transpose, if `type` is
- * `complex64` or `complex128`).
- * <p>
- * The `type` parameter denotes the type of the matrix elements. The supported
- * types are: `float32`, `float64`, `complex64` and `complex128`.
- * <p>
- * Usage example:
- * <pre>{@code
- *     from tensorflow.python.ops.linalg.sparse import sparse_csr_matrix_ops
- * 
- *     a_indices = np.array([[0, 0], [1, 1], [2, 1], [2, 2], [3, 3]])
- *     a_values = np.array([1.0, 2.0, 1.0, 3.0, 4.0], np.float32)
- *     a_dense_shape = [4, 4]
- * 
- *     with tf.Session() as sess:
- *       # Define (COO format) SparseTensor over Numpy array.
- *       a_st = tf.sparse.SparseTensor(a_indices, a_values, a_dense_shape)
- * 
- *       # Convert SparseTensors to CSR SparseMatrix.
- *       a_sm = sparse_csr_matrix_ops.sparse_tensor_to_csr_sparse_matrix(
- *           a_st.indices, a_st.values, a_st.dense_shape)
- * 
- *       # Obtain the Sparse Cholesky factor using AMD Ordering for reducing zero
- *       # fill-in (number of structural non-zeros in the sparse Cholesky factor).
- *       ordering_amd = sparse_csr_matrix_ops.sparse_matrix_ordering_amd(sparse_matrix)
- *       cholesky_sparse_matrices = (
- *           sparse_csr_matrix_ops.sparse_matrix_sparse_cholesky(
- *               sparse_matrix, ordering_amd, type=tf.float32))
- * 
- *       # Convert the CSRSparseMatrix Cholesky factor to a dense Tensor
- *       dense_cholesky = sparse_csr_matrix_ops.csr_sparse_matrix_to_dense(
- *           cholesky_sparse_matrices, tf.float32)
- * 
- *       # Evaluate the dense Tensor value.
- *       dense_cholesky_value = sess.run(dense_cholesky)
- * }</pre>
- * `dense_cholesky_value` stores the dense Cholesky factor:
- * <pre>{@code
- *     [[  1.  0.    0.    0.]
- *      [  0.  1.41  0.    0.]
- *      [  0.  0.70  1.58  0.]
- *      [  0.  0.    0.    2.]]
- * }</pre>
- * input: A `CSRSparseMatrix`.
- * permutation: A `Tensor`.
- * type: The type of `input`.
- */
-public final class SparseMatrixSparseCholesky extends RawOp implements Operand<TType> {
-  
-  /**
-   * Factory method to create a class wrapping a new SparseMatrixSparseCholesky operation.
-   * 
-   * @param scope current scope
-   * @param input A `CSRSparseMatrix`.
-   * @param permutation A fill-in reducing permutation matrix.
-   * @param type 
-   * @return a new instance of SparseMatrixSparseCholesky
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> SparseMatrixSparseCholesky create(Scope scope, Operand<?> input, Operand<TInt32> permutation, Class<T> type) {
-    OperationBuilder opBuilder = scope.env().opBuilder("SparseMatrixSparseCholesky", scope.makeOpName("SparseMatrixSparseCholesky"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder.addInput(permutation.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("type", type);
-    return new SparseMatrixSparseCholesky(opBuilder.build());
-  }
-  
-  /**
-   * The sparse Cholesky decompsition of `input`.
-   */
-  public Output<?> output() {
-    return output;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "SparseMatrixSparseCholesky";
-  
-  private Output<?> output;
-  
-  private SparseMatrixSparseCholesky(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixSparseMatMul.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixSparseMatMul.java
deleted file mode 100644
index ec46546119d..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixSparseMatMul.java
+++ /dev/null
@@ -1,241 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.linalg.sparse;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Sparse-matrix-multiplies two CSR matrices `a` and `b`.
- * <p>
- * Performs a matrix multiplication of a sparse matrix `a` with a sparse matrix
- * `b`; returns a sparse matrix `a * b`, unless either `a` or `b` is transposed or
- * adjointed.
- * <p>
- * Each matrix may be transposed or adjointed (conjugated and transposed)
- * according to the Boolean parameters `transpose_a`, `adjoint_a`, `transpose_b`
- * and `adjoint_b`. At most one of `transpose_a` or `adjoint_a` may be True.
- * Similarly, at most one of `transpose_b` or `adjoint_b` may be True.
- * <p>
- * The inputs must have compatible shapes. That is, the inner dimension of `a`
- * must be equal to the outer dimension of `b`. This requirement is adjusted
- * according to whether either `a` or `b` is transposed or adjointed.
- * <p>
- * The `type` parameter denotes the type of the matrix elements. Both `a` and `b`
- * must have the same type. The supported types are: `float32`, `float64`,
- * `complex64` and `complex128`.
- * <p>
- * Both `a` and `b` must have the same rank. Broadcasting is not supported. If they
- * have rank 3, each batch of 2D CSRSparseMatrices within `a` and `b` must have the
- * same dense shape.
- * <p>
- * The sparse matrix product may have numeric (non-structural) zeros.
- * TODO(anudhyan): Consider adding a boolean attribute to control whether to prune
- * zeros.
- * <p>
- * Usage example:
- * <pre>{@code
- *     from tensorflow.python.ops.linalg.sparse import sparse_csr_matrix_ops
- * 
- *     a_indices = np.array([[0, 0], [2, 3], [2, 4], [3, 0]])
- *     a_values = np.array([1.0, 5.0, -1.0, -2.0], np.float32)
- *     a_dense_shape = [4, 5]
- * 
- *     b_indices = np.array([[0, 0], [3, 0], [3, 1]])
- *     b_values = np.array([2.0, 7.0, 8.0], np.float32)
- *     b_dense_shape = [5, 3]
- * 
- *     with tf.Session() as sess:
- *       # Define (COO format) Sparse Tensors over Numpy arrays
- *       a_st = tf.sparse.SparseTensor(a_indices, a_values, a_dense_shape)
- *       b_st = tf.sparse.SparseTensor(b_indices, b_values, b_dense_shape)
- * 
- *       # Convert SparseTensors to CSR SparseMatrix
- *       a_sm = sparse_csr_matrix_ops.sparse_tensor_to_csr_sparse_matrix(
- *           a_st.indices, a_st.values, a_st.dense_shape)
- *       b_sm = sparse_csr_matrix_ops.sparse_tensor_to_csr_sparse_matrix(
- *           b_st.indices, b_st.values, b_st.dense_shape)
- * 
- *       # Compute the CSR SparseMatrix matrix multiplication
- *       c_sm = sparse_csr_matrix_ops.sparse_matrix_sparse_mat_mul(
- *           a=a_sm, b=b_sm, type=tf.float32)
- * 
- *       # Convert the CSR SparseMatrix product to a dense Tensor
- *       c_sm_dense = sparse_csr_matrix_ops.csr_sparse_matrix_to_dense(
- *           c_sm, tf.float32)
- *       # Evaluate the dense Tensor value
- *       c_sm_dense_value = sess.run(c_sm_dense)
- * }</pre>
- * `c_sm_dense_value` stores the dense matrix product:
- * <pre>{@code
- *     [[  2.   0.   0.]
- *      [  0.   0.   0.]
- *      [ 35.  40.   0.]
- *      [ -4.   0.   0.]]
- * }</pre>
- * a: A `CSRSparseMatrix`.
- * b: A `CSRSparseMatrix` with the same type and rank as `a`.
- * type: The type of both `a` and `b`.
- * transpose_a: If True, `a` transposed before multiplication.
- * transpose_b: If True, `b` transposed before multiplication.
- * adjoint_a: If True, `a` adjointed before multiplication.
- * adjoint_b: If True, `b` adjointed before multiplication.
- */
-public final class SparseMatrixSparseMatMul extends RawOp implements Operand<TType> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.linalg.sparse.SparseMatrixSparseMatMul}
-   */
-  public static class Options {
-    
-    /**
-     * @param transposeA Indicates whether `a` should be transposed.
-     */
-    public Options transposeA(Boolean transposeA) {
-      this.transposeA = transposeA;
-      return this;
-    }
-    
-    /**
-     * @param transposeB Indicates whether `b` should be transposed.
-     */
-    public Options transposeB(Boolean transposeB) {
-      this.transposeB = transposeB;
-      return this;
-    }
-    
-    /**
-     * @param adjointA Indicates whether `a` should be conjugate-transposed.
-     */
-    public Options adjointA(Boolean adjointA) {
-      this.adjointA = adjointA;
-      return this;
-    }
-    
-    /**
-     * @param adjointB Indicates whether `b` should be conjugate-transposed.
-     */
-    public Options adjointB(Boolean adjointB) {
-      this.adjointB = adjointB;
-      return this;
-    }
-    
-    private Boolean transposeA;
-    private Boolean transposeB;
-    private Boolean adjointA;
-    private Boolean adjointB;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new SparseMatrixSparseMatMul operation.
-   * 
-   * @param scope current scope
-   * @param a A CSRSparseMatrix.
-   * @param b A CSRSparseMatrix.
-   * @param type 
-   * @param options carries optional attributes values
-   * @return a new instance of SparseMatrixSparseMatMul
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> SparseMatrixSparseMatMul create(Scope scope, Operand<?> a, Operand<?> b, Class<T> type, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("SparseMatrixSparseMatMul", scope.makeOpName("SparseMatrixSparseMatMul"));
-    opBuilder.addInput(a.asOutput(scope));
-    opBuilder.addInput(b.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("type", type);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.transposeA != null) {
-          opBuilder.setAttr("transpose_a", opts.transposeA);
-        }
-        if (opts.transposeB != null) {
-          opBuilder.setAttr("transpose_b", opts.transposeB);
-        }
-        if (opts.adjointA != null) {
-          opBuilder.setAttr("adjoint_a", opts.adjointA);
-        }
-        if (opts.adjointB != null) {
-          opBuilder.setAttr("adjoint_b", opts.adjointB);
-        }
-      }
-    }
-    return new SparseMatrixSparseMatMul(opBuilder.build());
-  }
-  
-  /**
-   * @param transposeA Indicates whether `a` should be transposed.
-   */
-  public static Options transposeA(Boolean transposeA) {
-    return new Options().transposeA(transposeA);
-  }
-  
-  /**
-   * @param transposeB Indicates whether `b` should be transposed.
-   */
-  public static Options transposeB(Boolean transposeB) {
-    return new Options().transposeB(transposeB);
-  }
-  
-  /**
-   * @param adjointA Indicates whether `a` should be conjugate-transposed.
-   */
-  public static Options adjointA(Boolean adjointA) {
-    return new Options().adjointA(adjointA);
-  }
-  
-  /**
-   * @param adjointB Indicates whether `b` should be conjugate-transposed.
-   */
-  public static Options adjointB(Boolean adjointB) {
-    return new Options().adjointB(adjointB);
-  }
-  
-  /**
-   * A CSRSparseMatrix.
-   */
-  public Output<?> c() {
-    return c;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) c;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "SparseMatrixSparseMatMul";
-  
-  private Output<?> c;
-  
-  private SparseMatrixSparseMatMul(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    c = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixTranspose.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixTranspose.java
deleted file mode 100644
index 554c506be2a..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixTranspose.java
+++ /dev/null
@@ -1,112 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.linalg.sparse;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Transposes the inner (matrix) dimensions of a CSRSparseMatrix.
- * <p>
- * Transposes the inner (matrix) dimensions of a SparseMatrix and optionally
- * conjugates its values.
- */
-public final class SparseMatrixTranspose extends RawOp implements Operand<TType> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.linalg.sparse.SparseMatrixTranspose}
-   */
-  public static class Options {
-    
-    /**
-     * @param conjugate Indicates whether `input` should be conjugated.
-     */
-    public Options conjugate(Boolean conjugate) {
-      this.conjugate = conjugate;
-      return this;
-    }
-    
-    private Boolean conjugate;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new SparseMatrixTranspose operation.
-   * 
-   * @param scope current scope
-   * @param input A CSRSparseMatrix.
-   * @param type 
-   * @param options carries optional attributes values
-   * @return a new instance of SparseMatrixTranspose
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> SparseMatrixTranspose create(Scope scope, Operand<?> input, Class<T> type, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("SparseMatrixTranspose", scope.makeOpName("SparseMatrixTranspose"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("type", type);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.conjugate != null) {
-          opBuilder.setAttr("conjugate", opts.conjugate);
-        }
-      }
-    }
-    return new SparseMatrixTranspose(opBuilder.build());
-  }
-  
-  /**
-   * @param conjugate Indicates whether `input` should be conjugated.
-   */
-  public static Options conjugate(Boolean conjugate) {
-    return new Options().conjugate(conjugate);
-  }
-  
-  /**
-   * A CSRSparseMatrix.
-   */
-  public Output<?> output() {
-    return output;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "SparseMatrixTranspose";
-  
-  private Output<?> output;
-  
-  private SparseMatrixTranspose(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixZeros.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixZeros.java
deleted file mode 100644
index 33cac876a73..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixZeros.java
+++ /dev/null
@@ -1,76 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.linalg.sparse;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
-
-/**
- * Creates an all-zeros CSRSparseMatrix with shape `dense_shape`.
- */
-public final class SparseMatrixZeros extends RawOp implements Operand<TType> {
-  
-  /**
-   * Factory method to create a class wrapping a new SparseMatrixZeros operation.
-   * 
-   * @param scope current scope
-   * @param denseShape The desired matrix shape.
-   * @param type 
-   * @return a new instance of SparseMatrixZeros
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> SparseMatrixZeros create(Scope scope, Operand<TInt64> denseShape, Class<T> type) {
-    OperationBuilder opBuilder = scope.env().opBuilder("SparseMatrixZeros", scope.makeOpName("SparseMatrixZeros"));
-    opBuilder.addInput(denseShape.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("type", type);
-    return new SparseMatrixZeros(opBuilder.build());
-  }
-  
-  /**
-   * An empty CSR matrix with shape `dense_shape`.
-   */
-  public Output<?> sparseMatrix() {
-    return sparseMatrix;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) sparseMatrix;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "SparseMatrixZeros";
-  
-  private Output<?> sparseMatrix;
-  
-  private SparseMatrixZeros(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    sparseMatrix = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseTensorToCSRSparseMatrix.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseTensorToCSRSparseMatrix.java
deleted file mode 100644
index f3cf07dbb45..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseTensorToCSRSparseMatrix.java
+++ /dev/null
@@ -1,78 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.linalg.sparse;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
-
-/**
- * Converts a SparseTensor to a (possibly batched) CSRSparseMatrix.
- */
-public final class SparseTensorToCSRSparseMatrix extends RawOp implements Operand<TType> {
-  
-  /**
-   * Factory method to create a class wrapping a new SparseTensorToCSRSparseMatrix operation.
-   * 
-   * @param scope current scope
-   * @param indices SparseTensor indices.
-   * @param values SparseTensor values.
-   * @param denseShape SparseTensor dense shape.
-   * @return a new instance of SparseTensorToCSRSparseMatrix
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> SparseTensorToCSRSparseMatrix create(Scope scope, Operand<TInt64> indices, Operand<T> values, Operand<TInt64> denseShape) {
-    OperationBuilder opBuilder = scope.env().opBuilder("SparseTensorToCSRSparseMatrix", scope.makeOpName("SparseTensorToCSRSparseMatrix"));
-    opBuilder.addInput(indices.asOutput(scope));
-    opBuilder.addInput(values.asOutput(scope));
-    opBuilder.addInput(denseShape.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new SparseTensorToCSRSparseMatrix(opBuilder.build());
-  }
-  
-  /**
-   * A (possibly batched) CSRSparseMatrix.
-   */
-  public Output<?> sparseMatrix() {
-    return sparseMatrix;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) sparseMatrix;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "SparseTensorToCSRSparseMatrix";
-  
-  private Output<?> sparseMatrix;
-  
-  private SparseTensorToCSRSparseMatrix(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    sparseMatrix = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Abs.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Abs.java
deleted file mode 100644
index 3f52e09a1c5..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Abs.java
+++ /dev/null
@@ -1,78 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.math;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Computes the absolute value of a tensor.
- * <p>
- * Given a tensor `x`, this operation returns a tensor containing the absolute
- * value of each element in `x`. For example, if x is an input element and y is
- * an output element, this operation computes \\(y = |x|\\).
- * 
- * @param <T> data type for {@code y()} output
- */
-@Operator(group = "math")
-public final class Abs<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new Abs operation.
-   * 
-   * @param scope current scope
-   * @param x 
-   * @return a new instance of Abs
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> Abs<T> create(Scope scope, Operand<T> x) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Abs", scope.makeOpName("Abs"));
-    opBuilder.addInput(x.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new Abs<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> y() {
-    return y;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return y;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Abs";
-  
-  private Output<T> y;
-  
-  private Abs(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    y = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/AccumulateN.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/AccumulateN.java
deleted file mode 100644
index cdac6836f17..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/AccumulateN.java
+++ /dev/null
@@ -1,87 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.math;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.op.Operands;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Returns the element-wise sum of a list of tensors.
- * <p>
- * `tf.accumulate_n_v2` performs the same operation as `tf.add_n`, but does not
- * wait for all of its inputs to be ready before beginning to sum. This can
- * save memory if inputs are ready at different times, since minimum temporary
- * storage is proportional to the output size rather than the inputs size.
- * <p>
- * Unlike the original `accumulate_n`, `accumulate_n_v2` is differentiable.
- * <p>
- * Returns a `Tensor` of same shape and type as the elements of `inputs`.
- * 
- * @param <T> data type for {@code sum()} output
- */
-@Operator(group = "math")
-public final class AccumulateN<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new AccumulateN operation.
-   * 
-   * @param scope current scope
-   * @param inputs A list of `Tensor` objects, each with same shape and type.
-   * @param shape Shape of elements of `inputs`.
-   * @return a new instance of AccumulateN
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> AccumulateN<T> create(Scope scope, Iterable<Operand<T>> inputs, Shape shape) {
-    OperationBuilder opBuilder = scope.env().opBuilder("AccumulateNV2", scope.makeOpName("AccumulateN"));
-    opBuilder.addInputList(Operands.asOutputs(scope, inputs));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("shape", shape);
-    return new AccumulateN<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> sum() {
-    return sum;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return sum;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "AccumulateNV2";
-  
-  private Output<T> sum;
-  
-  private AccumulateN(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    sum = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Acos.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Acos.java
deleted file mode 100644
index e4a8200e5ea..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Acos.java
+++ /dev/null
@@ -1,74 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.math;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Computes acos of x element-wise.
- * 
- * @param <T> data type for {@code y()} output
- */
-@Operator(group = "math")
-public final class Acos<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new Acos operation.
-   * 
-   * @param scope current scope
-   * @param x 
-   * @return a new instance of Acos
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> Acos<T> create(Scope scope, Operand<T> x) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Acos", scope.makeOpName("Acos"));
-    opBuilder.addInput(x.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new Acos<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> y() {
-    return y;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return y;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Acos";
-  
-  private Output<T> y;
-  
-  private Acos(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    y = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Acosh.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Acosh.java
deleted file mode 100644
index 7f2bda57117..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Acosh.java
+++ /dev/null
@@ -1,82 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.math;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Computes inverse hyperbolic cosine of x element-wise.
- * <p>
- * Given an input tensor, the function computes inverse hyperbolic cosine of every element.
- * Input range is `[1, inf]`. It returns `nan` if the input lies outside the range.
- * <pre>{@code
- * x = tf.constant([-2, -0.5, 1, 1.2, 200, 10000, float("inf")])
- * tf.math.acosh(x) ==> [nan nan 0. 0.62236255 5.9914584 9.903487 inf]
- * }</pre>
- * 
- * 
- * @param <T> data type for {@code y()} output
- */
-@Operator(group = "math")
-public final class Acosh<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new Acosh operation.
-   * 
-   * @param scope current scope
-   * @param x 
-   * @return a new instance of Acosh
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> Acosh<T> create(Scope scope, Operand<T> x) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Acosh", scope.makeOpName("Acosh"));
-    opBuilder.addInput(x.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new Acosh<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> y() {
-    return y;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return y;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Acosh";
-  
-  private Output<T> y;
-  
-  private Acosh(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    y = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Add.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Add.java
deleted file mode 100644
index f2a374fb9e3..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Add.java
+++ /dev/null
@@ -1,79 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.math;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Returns x + y element-wise.
- * <p>
- * <i>NOTE</i>: `math.Add` supports broadcasting. `AddN` does not. More about broadcasting
- * [here](http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html)
- * 
- * @param <T> data type for {@code z()} output
- */
-@Operator(group = "math")
-public final class Add<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new Add operation.
-   * 
-   * @param scope current scope
-   * @param x 
-   * @param y 
-   * @return a new instance of Add
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> Add<T> create(Scope scope, Operand<T> x, Operand<T> y) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Add", scope.makeOpName("Add"));
-    opBuilder.addInput(x.asOutput(scope));
-    opBuilder.addInput(y.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new Add<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> z() {
-    return z;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return z;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Add";
-  
-  private Output<T> z;
-  
-  private Add(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    z = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/AddN.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/AddN.java
deleted file mode 100644
index 99815542073..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/AddN.java
+++ /dev/null
@@ -1,83 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.math;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.Operands;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Add all input tensors element wise.
- * <p>
- *   Inputs must be of same size and shape.
- * <p>
- *   <pre>{@code
- *   x = [9, 7, 10]
- *   tf.math.add_n(x) ==> 26
- *   }</pre>
- * 
- * 
- * @param <T> data type for {@code sum()} output
- */
-@Operator(group = "math")
-public final class AddN<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new AddN operation.
-   * 
-   * @param scope current scope
-   * @param inputs 
-   * @return a new instance of AddN
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> AddN<T> create(Scope scope, Iterable<Operand<T>> inputs) {
-    OperationBuilder opBuilder = scope.env().opBuilder("AddN", scope.makeOpName("AddN"));
-    opBuilder.addInputList(Operands.asOutputs(scope, inputs));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new AddN<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> sum() {
-    return sum;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return sum;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "AddN";
-  
-  private Output<T> sum;
-  
-  private AddN(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    sum = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Angle.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Angle.java
deleted file mode 100644
index 4d4d662f614..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Angle.java
+++ /dev/null
@@ -1,106 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.math;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * Returns the argument of a complex number.
- * <p>
- * Given a tensor `input` of complex numbers, this operation returns a tensor of
- * type `float` that is the argument of each element in `input`. All elements in
- * `input` must be complex numbers of the form \\(a + bj\\), where <i>a</i>
- * is the real part and <i>b</i> is the imaginary part.
- * <p>
- * The argument returned by this operation is of the form \\(atan2(b, a)\\).
- * <p>
- * For example:
- * <pre>{@code
- * # tensor 'input' is [-2.25 + 4.75j, 3.25 + 5.75j]
- * tf.angle(input) ==> [2.0132, 1.056]
- * }</pre>
- * @compatibility(numpy)
- * Equivalent to np.angle.
- * @end_compatibility
- * 
- * @param <U> data type for {@code output()} output
- */
-@Operator(group = "math")
-public final class Angle<U extends TNumber> extends RawOp implements Operand<U> {
-  
-  /**
-   * Factory method to create a class wrapping a new Angle operation.
-   * 
-   * @param scope current scope
-   * @param input 
-   * @param Tout 
-   * @return a new instance of Angle
-   */
-  @Endpoint(describeByClass = true)
-  public static <U extends TNumber, T extends TType> Angle<U> create(Scope scope, Operand<T> input, Class<U> Tout) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Angle", scope.makeOpName("Angle"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("Tout", Tout);
-    return new Angle<U>(opBuilder.build());
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new Angle operation using default output types.
-   * 
-   * @param scope current scope
-   * @param input 
-   * @return a new instance of Angle
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> Angle<TFloat32> create(Scope scope, Operand<T> input) {
-    return create(scope, input, TFloat32.class);
-  }
-  
-  /**
-   */
-  public Output<U> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<U> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Angle";
-  
-  private Output<U> output;
-  
-  private Angle(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/ApproximateEqual.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/ApproximateEqual.java
deleted file mode 100644
index 753f503f64b..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/ApproximateEqual.java
+++ /dev/null
@@ -1,109 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.math;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TBool;
-import org.tensorflow.types.family.TType;
-
-/**
- * Returns the truth value of abs(x-y) < tolerance element-wise.
- */
-@Operator(group = "math")
-public final class ApproximateEqual extends RawOp implements Operand<TBool> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.math.ApproximateEqual}
-   */
-  public static class Options {
-    
-    /**
-     * @param tolerance 
-     */
-    public Options tolerance(Float tolerance) {
-      this.tolerance = tolerance;
-      return this;
-    }
-    
-    private Float tolerance;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new ApproximateEqual operation.
-   * 
-   * @param scope current scope
-   * @param x 
-   * @param y 
-   * @param options carries optional attributes values
-   * @return a new instance of ApproximateEqual
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> ApproximateEqual create(Scope scope, Operand<T> x, Operand<T> y, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ApproximateEqual", scope.makeOpName("ApproximateEqual"));
-    opBuilder.addInput(x.asOutput(scope));
-    opBuilder.addInput(y.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.tolerance != null) {
-          opBuilder.setAttr("tolerance", opts.tolerance);
-        }
-      }
-    }
-    return new ApproximateEqual(opBuilder.build());
-  }
-  
-  /**
-   * @param tolerance 
-   */
-  public static Options tolerance(Float tolerance) {
-    return new Options().tolerance(tolerance);
-  }
-  
-  /**
-   */
-  public Output<TBool> z() {
-    return z;
-  }
-  
-  @Override
-  public Output<TBool> asOutput(Scope scope) {
-    return z;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ApproximateEqual";
-  
-  private Output<TBool> z;
-  
-  private ApproximateEqual(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    z = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/ArgMax.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/ArgMax.java
deleted file mode 100644
index f5348fd11d3..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/ArgMax.java
+++ /dev/null
@@ -1,110 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.math;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * Returns the index with the largest value across dimensions of a tensor.
- * <p>
- * Note that in case of ties the identity of the return value is not guaranteed.
- * <p>
- * Usage:
- *   <pre>{@code
- *   import tensorflow as tf
- *   a = [1, 10, 26.9, 2.8, 166.32, 62.3]
- *   b = tf.math.argmax(input = a)
- *   c = tf.keras.backend.eval(b)
- *   # c = 4
- *   # here a[4] = 166.32 which is the largest element of a across axis 0
- *   }</pre>
- * 
- * 
- * @param <V> data type for {@code output()} output
- */
-@Operator(group = "math")
-public final class ArgMax<V extends TNumber> extends RawOp implements Operand<V> {
-  
-  /**
-   * Factory method to create a class wrapping a new ArgMax operation.
-   * 
-   * @param scope current scope
-   * @param input 
-   * @param dimension int32 or int64, must be in the range `[-rank(input), rank(input))`.
-   * Describes which dimension of the input Tensor to reduce across. For vectors,
-   * use dimension = 0.
-   * @param outputType 
-   * @return a new instance of ArgMax
-   */
-  @Endpoint(describeByClass = true)
-  public static <V extends TNumber, T extends TType, U extends TNumber> ArgMax<V> create(Scope scope, Operand<T> input, Operand<U> dimension, Class<V> outputType) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ArgMax", scope.makeOpName("ArgMax"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder.addInput(dimension.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("output_type", outputType);
-    return new ArgMax<V>(opBuilder.build());
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new ArgMax operation using default output types.
-   * 
-   * @param scope current scope
-   * @param input 
-   * @param dimension int32 or int64, must be in the range `[-rank(input), rank(input))`.
-   * Describes which dimension of the input Tensor to reduce across. For vectors,
-   * use dimension = 0.
-   * @return a new instance of ArgMax
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> ArgMax<TInt64> create(Scope scope, Operand<T> input, Operand<U> dimension) {
-    return create(scope, input, dimension, TInt64.class);
-  }
-  
-  /**
-   */
-  public Output<V> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<V> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ArgMax";
-  
-  private Output<V> output;
-  
-  private ArgMax(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/ArgMin.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/ArgMin.java
deleted file mode 100644
index 5bb45ff03b5..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/ArgMin.java
+++ /dev/null
@@ -1,110 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.math;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * Returns the index with the smallest value across dimensions of a tensor.
- * <p>
- * Note that in case of ties the identity of the return value is not guaranteed.
- * <p>
- * Usage:
- *   <pre>{@code
- *   import tensorflow as tf
- *   a = [1, 10, 26.9, 2.8, 166.32, 62.3]
- *   b = tf.math.argmin(input = a)
- *   c = tf.keras.backend.eval(b)
- *   # c = 0
- *   # here a[0] = 1 which is the smallest element of a across axis 0
- *   }</pre>
- * 
- * 
- * @param <V> data type for {@code output()} output
- */
-@Operator(group = "math")
-public final class ArgMin<V extends TNumber> extends RawOp implements Operand<V> {
-  
-  /**
-   * Factory method to create a class wrapping a new ArgMin operation.
-   * 
-   * @param scope current scope
-   * @param input 
-   * @param dimension int32 or int64, must be in the range `[-rank(input), rank(input))`.
-   * Describes which dimension of the input Tensor to reduce across. For vectors,
-   * use dimension = 0.
-   * @param outputType 
-   * @return a new instance of ArgMin
-   */
-  @Endpoint(describeByClass = true)
-  public static <V extends TNumber, T extends TType, U extends TNumber> ArgMin<V> create(Scope scope, Operand<T> input, Operand<U> dimension, Class<V> outputType) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ArgMin", scope.makeOpName("ArgMin"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder.addInput(dimension.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("output_type", outputType);
-    return new ArgMin<V>(opBuilder.build());
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new ArgMin operation using default output types.
-   * 
-   * @param scope current scope
-   * @param input 
-   * @param dimension int32 or int64, must be in the range `[-rank(input), rank(input))`.
-   * Describes which dimension of the input Tensor to reduce across. For vectors,
-   * use dimension = 0.
-   * @return a new instance of ArgMin
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> ArgMin<TInt64> create(Scope scope, Operand<T> input, Operand<U> dimension) {
-    return create(scope, input, dimension, TInt64.class);
-  }
-  
-  /**
-   */
-  public Output<V> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<V> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ArgMin";
-  
-  private Output<V> output;
-  
-  private ArgMin(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Asin.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Asin.java
deleted file mode 100644
index 48750003eb1..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Asin.java
+++ /dev/null
@@ -1,90 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.math;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Computes the trignometric inverse sine of x element-wise.
- * <p>
- * The `tf.math.asin` operation returns the inverse of `tf.math.sin`, such that
- * if `y = tf.math.sin(x)` then, `x = tf.math.asin(y)`.
- * <p>
- * <b>Note</b>: The output of `tf.math.asin` will lie within the invertible range
- * of sine, i.e [-pi/2, pi/2].
- * <p>
- * For example:
- * <pre>{@code
- * # Note: [1.047, 0.785] ~= [(pi/3), (pi/4)]
- * x = tf.constant([1.047, 0.785])
- * y = tf.math.sin(x) # [0.8659266, 0.7068252]
- * 
- * tf.math.asin(y) # [1.047, 0.785] = x
- * }</pre>
- * 
- * 
- * @param <T> data type for {@code y()} output
- */
-@Operator(group = "math")
-public final class Asin<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new Asin operation.
-   * 
-   * @param scope current scope
-   * @param x 
-   * @return a new instance of Asin
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> Asin<T> create(Scope scope, Operand<T> x) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Asin", scope.makeOpName("Asin"));
-    opBuilder.addInput(x.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new Asin<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> y() {
-    return y;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return y;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Asin";
-  
-  private Output<T> y;
-  
-  private Asin(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    y = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Asinh.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Asinh.java
deleted file mode 100644
index ae067dffb7c..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Asinh.java
+++ /dev/null
@@ -1,84 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.math;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Computes inverse hyperbolic sine of x element-wise.
- * <p>
- *   Given an input tensor, this function computes inverse hyperbolic sine
- *   for every element in the tensor. Both input and output has a range of
- *   `[-inf, inf]`.
- * <p>
- *   <pre>{@code
- *   x = tf.constant([-float("inf"), -2, -0.5, 1, 1.2, 200, 10000, float("inf")])
- *   tf.math.asinh(x) ==> [-inf -1.4436355 -0.4812118 0.8813736 1.0159732 5.991471 9.903487 inf]
- *   }</pre>
- * 
- * 
- * @param <T> data type for {@code y()} output
- */
-@Operator(group = "math")
-public final class Asinh<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new Asinh operation.
-   * 
-   * @param scope current scope
-   * @param x 
-   * @return a new instance of Asinh
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> Asinh<T> create(Scope scope, Operand<T> x) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Asinh", scope.makeOpName("Asinh"));
-    opBuilder.addInput(x.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new Asinh<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> y() {
-    return y;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return y;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Asinh";
-  
-  private Output<T> y;
-  
-  private Asinh(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    y = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Atan.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Atan.java
deleted file mode 100644
index f78ec14e057..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Atan.java
+++ /dev/null
@@ -1,90 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.math;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Computes the trignometric inverse tangent of x element-wise.
- * <p>
- * The `tf.math.atan` operation returns the inverse of `tf.math.tan`, such that
- * if `y = tf.math.tan(x)` then, `x = tf.math.atan(y)`.
- * <p>
- * <b>Note</b>: The output of `tf.math.atan` will lie within the invertible range
- * of tan, i.e (-pi/2, pi/2).
- * <p>
- * For example:
- * <pre>{@code
- * # Note: [1.047, 0.785] ~= [(pi/3), (pi/4)]
- * x = tf.constant([1.047, 0.785])
- * y = tf.math.tan(x) # [1.731261, 0.99920404]
- * 
- * tf.math.atan(y) # [1.047, 0.785] = x
- * }</pre>
- * 
- * 
- * @param <T> data type for {@code y()} output
- */
-@Operator(group = "math")
-public final class Atan<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new Atan operation.
-   * 
-   * @param scope current scope
-   * @param x 
-   * @return a new instance of Atan
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> Atan<T> create(Scope scope, Operand<T> x) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Atan", scope.makeOpName("Atan"));
-    opBuilder.addInput(x.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new Atan<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> y() {
-    return y;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return y;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Atan";
-  
-  private Output<T> y;
-  
-  private Atan(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    y = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Atan2.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Atan2.java
deleted file mode 100644
index afb5011d4ce..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Atan2.java
+++ /dev/null
@@ -1,82 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.math;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Computes arctangent of `y/x` element-wise, respecting signs of the arguments.
- * <p>
- * This is the angle \( \theta \in [-\pi, \pi] \) such that
- * \[ x = r \cos(\theta) \]
- * and
- * \[ y = r \sin(\theta) \]
- * where \(r = \sqrt(x^2 + y^2) \).
- * 
- * @param <T> data type for {@code z()} output
- */
-@Operator(group = "math")
-public final class Atan2<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new Atan2 operation.
-   * 
-   * @param scope current scope
-   * @param y 
-   * @param x 
-   * @return a new instance of Atan2
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> Atan2<T> create(Scope scope, Operand<T> y, Operand<T> x) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Atan2", scope.makeOpName("Atan2"));
-    opBuilder.addInput(y.asOutput(scope));
-    opBuilder.addInput(x.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new Atan2<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> z() {
-    return z;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return z;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Atan2";
-  
-  private Output<T> z;
-  
-  private Atan2(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    z = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Atanh.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Atanh.java
deleted file mode 100644
index c8f3fbd8878..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Atanh.java
+++ /dev/null
@@ -1,86 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.math;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Computes inverse hyperbolic tangent of x element-wise.
- * <p>
- *   Given an input tensor, this function computes inverse hyperbolic tangent
- *   for every element in the tensor. Input range is `[-1,1]` and output range is
- *   `[-inf, inf]`. If input is `-1`, output will be `-inf` and if the
- *   input is `1`, output will be `inf`. Values outside the range will have
- *   `nan` as output.
- * <p>
- *   <pre>{@code
- *   x = tf.constant([-float("inf"), -1, -0.5, 1, 0, 0.5, 10, float("inf")])
- *   tf.math.atanh(x) ==> [nan -inf -0.54930615 inf  0. 0.54930615 nan nan]
- *   }</pre>
- * 
- * 
- * @param <T> data type for {@code y()} output
- */
-@Operator(group = "math")
-public final class Atanh<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new Atanh operation.
-   * 
-   * @param scope current scope
-   * @param x 
-   * @return a new instance of Atanh
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> Atanh<T> create(Scope scope, Operand<T> x) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Atanh", scope.makeOpName("Atanh"));
-    opBuilder.addInput(x.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new Atanh<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> y() {
-    return y;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return y;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Atanh";
-  
-  private Output<T> y;
-  
-  private Atanh(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    y = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/BesselI0.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/BesselI0.java
deleted file mode 100644
index ad238cc6aba..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/BesselI0.java
+++ /dev/null
@@ -1,71 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.math;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * @param <T> data type for {@code y()} output
- */
-public final class BesselI0<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new BesselI0 operation.
-   * 
-   * @param scope current scope
-   * @param x 
-   * @return a new instance of BesselI0
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> BesselI0<T> create(Scope scope, Operand<T> x) {
-    OperationBuilder opBuilder = scope.env().opBuilder("BesselI0", scope.makeOpName("BesselI0"));
-    opBuilder.addInput(x.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new BesselI0<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> y() {
-    return y;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return y;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "BesselI0";
-  
-  private Output<T> y;
-  
-  private BesselI0(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    y = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/BesselI0e.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/BesselI0e.java
deleted file mode 100644
index 7bb31bfb405..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/BesselI0e.java
+++ /dev/null
@@ -1,71 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.math;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * @param <T> data type for {@code y()} output
- */
-public final class BesselI0e<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new BesselI0e operation.
-   * 
-   * @param scope current scope
-   * @param x 
-   * @return a new instance of BesselI0e
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> BesselI0e<T> create(Scope scope, Operand<T> x) {
-    OperationBuilder opBuilder = scope.env().opBuilder("BesselI0e", scope.makeOpName("BesselI0e"));
-    opBuilder.addInput(x.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new BesselI0e<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> y() {
-    return y;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return y;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "BesselI0e";
-  
-  private Output<T> y;
-  
-  private BesselI0e(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    y = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/BesselI1.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/BesselI1.java
deleted file mode 100644
index 250d09d9d57..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/BesselI1.java
+++ /dev/null
@@ -1,71 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.math;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * @param <T> data type for {@code y()} output
- */
-public final class BesselI1<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new BesselI1 operation.
-   * 
-   * @param scope current scope
-   * @param x 
-   * @return a new instance of BesselI1
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> BesselI1<T> create(Scope scope, Operand<T> x) {
-    OperationBuilder opBuilder = scope.env().opBuilder("BesselI1", scope.makeOpName("BesselI1"));
-    opBuilder.addInput(x.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new BesselI1<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> y() {
-    return y;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return y;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "BesselI1";
-  
-  private Output<T> y;
-  
-  private BesselI1(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    y = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/BesselI1e.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/BesselI1e.java
deleted file mode 100644
index 7755807f940..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/BesselI1e.java
+++ /dev/null
@@ -1,71 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.math;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * @param <T> data type for {@code y()} output
- */
-public final class BesselI1e<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new BesselI1e operation.
-   * 
-   * @param scope current scope
-   * @param x 
-   * @return a new instance of BesselI1e
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> BesselI1e<T> create(Scope scope, Operand<T> x) {
-    OperationBuilder opBuilder = scope.env().opBuilder("BesselI1e", scope.makeOpName("BesselI1e"));
-    opBuilder.addInput(x.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new BesselI1e<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> y() {
-    return y;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return y;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "BesselI1e";
-  
-  private Output<T> y;
-  
-  private BesselI1e(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    y = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Betainc.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Betainc.java
deleted file mode 100644
index 066d1504311..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Betainc.java
+++ /dev/null
@@ -1,89 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.math;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Compute the regularized incomplete beta integral \\(I_x(a, b)\\).
- * <p>
- * The regularized incomplete beta integral is defined as:
- * <p>
- * \\(I_x(a, b) = \frac{B(x; a, b)}{B(a, b)}\\)
- * <p>
- * where
- * <p>
- * \\(B(x; a, b) = \int_0^x t^{a-1} (1 - t)^{b-1} dt\\)
- * <p>
- * is the incomplete beta function and \\(B(a, b)\\) is the <i>complete</i>
- * beta function.
- * 
- * @param <T> data type for {@code z()} output
- */
-@Operator(group = "math")
-public final class Betainc<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new Betainc operation.
-   * 
-   * @param scope current scope
-   * @param a 
-   * @param b 
-   * @param x 
-   * @return a new instance of Betainc
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> Betainc<T> create(Scope scope, Operand<T> a, Operand<T> b, Operand<T> x) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Betainc", scope.makeOpName("Betainc"));
-    opBuilder.addInput(a.asOutput(scope));
-    opBuilder.addInput(b.asOutput(scope));
-    opBuilder.addInput(x.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new Betainc<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> z() {
-    return z;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return z;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Betainc";
-  
-  private Output<T> z;
-  
-  private Betainc(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    z = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Bincount.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Bincount.java
deleted file mode 100644
index 0e15d315b76..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Bincount.java
+++ /dev/null
@@ -1,91 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.math;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Counts the number of occurrences of each value in an integer array.
- * <p>
- * Outputs a vector with length `size` and the same dtype as `weights`. If
- * `weights` are empty, then index `i` stores the number of times the value `i` is
- * counted in `arr`. If `weights` are non-empty, then index `i` stores the sum of
- * the value in `weights` at each index where the corresponding value in `arr` is
- * `i`.
- * <p>
- * Values in `arr` outside of the range [0, size) are ignored.
- * 
- * @param <T> data type for {@code bins()} output
- */
-@Operator(group = "math")
-public final class Bincount<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new Bincount operation.
-   * 
-   * @param scope current scope
-   * @param arr int32 `Tensor`.
-   * @param size non-negative int32 scalar `Tensor`.
-   * @param weights is an int32, int64, float32, or float64 `Tensor` with the same
-   * shape as `arr`, or a length-0 `Tensor`, in which case it acts as all weights
-   * equal to 1.
-   * @return a new instance of Bincount
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> Bincount<T> create(Scope scope, Operand<TInt32> arr, Operand<TInt32> size, Operand<T> weights) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Bincount", scope.makeOpName("Bincount"));
-    opBuilder.addInput(arr.asOutput(scope));
-    opBuilder.addInput(size.asOutput(scope));
-    opBuilder.addInput(weights.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new Bincount<T>(opBuilder.build());
-  }
-  
-  /**
-   * 1D `Tensor` with length equal to `size`. The counts or summed weights for
-   * each value in the range [0, size).
-   */
-  public Output<T> bins() {
-    return bins;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return bins;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Bincount";
-  
-  private Output<T> bins;
-  
-  private Bincount(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    bins = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Ceil.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Ceil.java
deleted file mode 100644
index 9f1081f9998..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Ceil.java
+++ /dev/null
@@ -1,74 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.math;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Returns element-wise smallest integer not less than x.
- * 
- * @param <T> data type for {@code y()} output
- */
-@Operator(group = "math")
-public final class Ceil<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new Ceil operation.
-   * 
-   * @param scope current scope
-   * @param x 
-   * @return a new instance of Ceil
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> Ceil<T> create(Scope scope, Operand<T> x) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Ceil", scope.makeOpName("Ceil"));
-    opBuilder.addInput(x.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new Ceil<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> y() {
-    return y;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return y;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Ceil";
-  
-  private Output<T> y;
-  
-  private Ceil(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    y = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/CompareAndBitpack.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/CompareAndBitpack.java
deleted file mode 100644
index 6ca6f9324aa..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/CompareAndBitpack.java
+++ /dev/null
@@ -1,98 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.math;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TUint8;
-import org.tensorflow.types.family.TType;
-
-/**
- * Compare values of `input` to `threshold` and pack resulting bits into a `uint8`.
- * <p>
- * Each comparison returns a boolean `true` (if `input_value > threshold`)
- * or and `false` otherwise.
- * <p>
- * This operation is useful for Locality-Sensitive-Hashing (LSH) and other
- * algorithms that use hashing approximations of cosine and `L2` distances;
- * codes can be generated from an input via:
- * <pre>{@code
- * codebook_size = 50
- * codebook_bits = codebook_size * 32
- * codebook = tf.get_variable('codebook', [x.shape[-1].value, codebook_bits],
- *                            dtype=x.dtype,
- *                            initializer=tf.orthogonal_initializer())
- * codes = compare_and_threshold(tf.matmul(x, codebook), threshold=0.)
- * codes = tf.bitcast(codes, tf.int32)  # go from uint8 to int32
- * # now codes has shape x.shape[:-1] + [codebook_size]
- * }</pre>
- * <b>NOTE</b>: Currently, the innermost dimension of the tensor must be divisible
- * by 8.
- * <p>
- * Given an `input` shaped `[s0, s1, ..., s_n]`, the output is
- * a `uint8` tensor shaped `[s0, s1, ..., s_n / 8]`.
- */
-@Operator(group = "math")
-public final class CompareAndBitpack extends RawOp implements Operand<TUint8> {
-  
-  /**
-   * Factory method to create a class wrapping a new CompareAndBitpack operation.
-   * 
-   * @param scope current scope
-   * @param input Values to compare against `threshold` and bitpack.
-   * @param threshold Threshold to compare against.
-   * @return a new instance of CompareAndBitpack
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> CompareAndBitpack create(Scope scope, Operand<T> input, Operand<T> threshold) {
-    OperationBuilder opBuilder = scope.env().opBuilder("CompareAndBitpack", scope.makeOpName("CompareAndBitpack"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder.addInput(threshold.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new CompareAndBitpack(opBuilder.build());
-  }
-  
-  /**
-   * The bitpacked comparisons.
-   */
-  public Output<TUint8> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<TUint8> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "CompareAndBitpack";
-  
-  private Output<TUint8> output;
-  
-  private CompareAndBitpack(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/ComplexAbs.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/ComplexAbs.java
deleted file mode 100644
index be130a37a7c..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/ComplexAbs.java
+++ /dev/null
@@ -1,95 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.math;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * Computes the complex absolute value of a tensor.
- * <p>
- * Given a tensor `x` of complex numbers, this operation returns a tensor of type
- * `float` or `double` that is the absolute value of each element in `x`. All
- * elements in `x` must be complex numbers of the form \\(a + bj\\). The absolute
- * value is computed as \\( \sqrt{a^2 + b^2}\\).
- * 
- * @param <U> data type for {@code y()} output
- */
-@Operator(group = "math")
-public final class ComplexAbs<U extends TNumber> extends RawOp implements Operand<U> {
-  
-  /**
-   * Factory method to create a class wrapping a new ComplexAbs operation.
-   * 
-   * @param scope current scope
-   * @param x 
-   * @param Tout 
-   * @return a new instance of ComplexAbs
-   */
-  @Endpoint(describeByClass = true)
-  public static <U extends TNumber, T extends TType> ComplexAbs<U> create(Scope scope, Operand<T> x, Class<U> Tout) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ComplexAbs", scope.makeOpName("ComplexAbs"));
-    opBuilder.addInput(x.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("Tout", Tout);
-    return new ComplexAbs<U>(opBuilder.build());
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new ComplexAbs operation using default output types.
-   * 
-   * @param scope current scope
-   * @param x 
-   * @return a new instance of ComplexAbs
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> ComplexAbs<TFloat32> create(Scope scope, Operand<T> x) {
-    return create(scope, x, TFloat32.class);
-  }
-  
-  /**
-   */
-  public Output<U> y() {
-    return y;
-  }
-  
-  @Override
-  public Output<U> asOutput(Scope scope) {
-    return y;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ComplexAbs";
-  
-  private Output<U> y;
-  
-  private ComplexAbs(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    y = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Conj.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Conj.java
deleted file mode 100644
index 27ca2ff1e88..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Conj.java
+++ /dev/null
@@ -1,88 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.math;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Returns the complex conjugate of a complex number.
- * <p>
- * Given a tensor `input` of complex numbers, this operation returns a tensor of
- * complex numbers that are the complex conjugate of each element in `input`. The
- * complex numbers in `input` must be of the form \\(a + bj\\), where <i>a</i> is the
- * real part and <i>b</i> is the imaginary part.
- * <p>
- * The complex conjugate returned by this operation is of the form \\(a - bj\\).
- * <p>
- * For example:
- * <pre>{@code
- * # tensor 'input' is [-2.25 + 4.75j, 3.25 + 5.75j]
- * tf.conj(input) ==> [-2.25 - 4.75j, 3.25 - 5.75j]
- * }</pre>
- * 
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator(group = "math")
-public final class Conj<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new Conj operation.
-   * 
-   * @param scope current scope
-   * @param input 
-   * @return a new instance of Conj
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> Conj<T> create(Scope scope, Operand<T> input) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Conj", scope.makeOpName("Conj"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new Conj<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Conj";
-  
-  private Output<T> output;
-  
-  private Conj(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Cos.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Cos.java
deleted file mode 100644
index f35c3eddb16..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Cos.java
+++ /dev/null
@@ -1,85 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.math;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Computes cos of x element-wise.
- * <p>
- *   Given an input tensor, this function computes cosine of every
- *   element in the tensor. Input range is `(-inf, inf)` and
- *   output range is `[-1,1]`. If input lies outside the boundary, `nan`
- *   is returned.
- * <p>
- *   <pre>{@code
- *   x = tf.constant([-float("inf"), -9, -0.5, 1, 1.2, 200, 10000, float("inf")])
- *   tf.math.cos(x) ==> [nan -0.91113025 0.87758255 0.5403023 0.36235774 0.48718765 -0.95215535 nan]
- *   }</pre>
- * 
- * 
- * @param <T> data type for {@code y()} output
- */
-@Operator(group = "math")
-public final class Cos<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new Cos operation.
-   * 
-   * @param scope current scope
-   * @param x 
-   * @return a new instance of Cos
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> Cos<T> create(Scope scope, Operand<T> x) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Cos", scope.makeOpName("Cos"));
-    opBuilder.addInput(x.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new Cos<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> y() {
-    return y;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return y;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Cos";
-  
-  private Output<T> y;
-  
-  private Cos(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    y = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Cosh.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Cosh.java
deleted file mode 100644
index 554123b820d..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Cosh.java
+++ /dev/null
@@ -1,84 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.math;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Computes hyperbolic cosine of x element-wise.
- * <p>
- *   Given an input tensor, this function computes hyperbolic cosine of every
- *   element in the tensor. Input range is `[-inf, inf]` and output range
- *   is `[1, inf]`.
- * <p>
- *   <pre>{@code
- *   x = tf.constant([-float("inf"), -9, -0.5, 1, 1.2, 2, 10, float("inf")])
- *   tf.math.cosh(x) ==> [inf 4.0515420e+03 1.1276259e+00 1.5430807e+00 1.8106556e+00 3.7621956e+00 1.1013233e+04 inf]
- *   }</pre>
- * 
- * 
- * @param <T> data type for {@code y()} output
- */
-@Operator(group = "math")
-public final class Cosh<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new Cosh operation.
-   * 
-   * @param scope current scope
-   * @param x 
-   * @return a new instance of Cosh
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> Cosh<T> create(Scope scope, Operand<T> x) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Cosh", scope.makeOpName("Cosh"));
-    opBuilder.addInput(x.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new Cosh<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> y() {
-    return y;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return y;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Cosh";
-  
-  private Output<T> y;
-  
-  private Cosh(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    y = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Cumprod.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Cumprod.java
deleted file mode 100644
index a2563630107..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Cumprod.java
+++ /dev/null
@@ -1,156 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.math;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * Compute the cumulative product of the tensor `x` along `axis`.
- * <p>
- * By default, this op performs an inclusive cumprod, which means that the first
- * element of the input is identical to the first element of the output:
- * <pre>{@code
- * tf.cumprod([a, b, c])  # => [a, a * b, a * b * c]
- * }</pre>
- * By setting the `exclusive` kwarg to `True`, an exclusive cumprod is
- * performed instead:
- * <pre>{@code
- * tf.cumprod([a, b, c], exclusive=True)  # => [1, a, a * b]
- * }</pre>
- * By setting the `reverse` kwarg to `True`, the cumprod is performed in the
- * opposite direction:
- * <pre>{@code
- * tf.cumprod([a, b, c], reverse=True)  # => [a * b * c, b * c, c]
- * }</pre>
- * This is more efficient than using separate `tf.reverse` ops.
- * <p>
- * The `reverse` and `exclusive` kwargs can also be combined:
- * <pre>{@code
- * tf.cumprod([a, b, c], exclusive=True, reverse=True)  # => [b * c, c, 1]
- * }</pre>
- * 
- * 
- * @param <T> data type for {@code out()} output
- */
-@Operator(group = "math")
-public final class Cumprod<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.math.Cumprod}
-   */
-  public static class Options {
-    
-    /**
-     * @param exclusive If `True`, perform exclusive cumprod.
-     */
-    public Options exclusive(Boolean exclusive) {
-      this.exclusive = exclusive;
-      return this;
-    }
-    
-    /**
-     * @param reverse A `bool` (default: False).
-     */
-    public Options reverse(Boolean reverse) {
-      this.reverse = reverse;
-      return this;
-    }
-    
-    private Boolean exclusive;
-    private Boolean reverse;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new Cumprod operation.
-   * 
-   * @param scope current scope
-   * @param x A `Tensor`. Must be one of the following types: `float32`, `float64`,
-   * `int64`, `int32`, `uint8`, `uint16`, `int16`, `int8`, `complex64`,
-   * `complex128`, `qint8`, `quint8`, `qint32`, `half`.
-   * @param axis A `Tensor` of type `int32` (default: 0). Must be in the range
-   * `[-rank(x), rank(x))`.
-   * @param options carries optional attributes values
-   * @return a new instance of Cumprod
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> Cumprod<T> create(Scope scope, Operand<T> x, Operand<U> axis, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Cumprod", scope.makeOpName("Cumprod"));
-    opBuilder.addInput(x.asOutput(scope));
-    opBuilder.addInput(axis.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.exclusive != null) {
-          opBuilder.setAttr("exclusive", opts.exclusive);
-        }
-        if (opts.reverse != null) {
-          opBuilder.setAttr("reverse", opts.reverse);
-        }
-      }
-    }
-    return new Cumprod<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param exclusive If `True`, perform exclusive cumprod.
-   */
-  public static Options exclusive(Boolean exclusive) {
-    return new Options().exclusive(exclusive);
-  }
-  
-  /**
-   * @param reverse A `bool` (default: False).
-   */
-  public static Options reverse(Boolean reverse) {
-    return new Options().reverse(reverse);
-  }
-  
-  /**
-   */
-  public Output<T> out() {
-    return out;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return out;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Cumprod";
-  
-  private Output<T> out;
-  
-  private Cumprod(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    out = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Cumsum.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Cumsum.java
deleted file mode 100644
index 47761661bbe..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Cumsum.java
+++ /dev/null
@@ -1,156 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.math;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * Compute the cumulative sum of the tensor `x` along `axis`.
- * <p>
- * By default, this op performs an inclusive cumsum, which means that the first
- * element of the input is identical to the first element of the output:
- * <pre>{@code
- * tf.cumsum([a, b, c])  # => [a, a + b, a + b + c]
- * }</pre>
- * By setting the `exclusive` kwarg to `True`, an exclusive cumsum is
- * performed instead:
- * <pre>{@code
- * tf.cumsum([a, b, c], exclusive=True)  # => [0, a, a + b]
- * }</pre>
- * By setting the `reverse` kwarg to `True`, the cumsum is performed in the
- * opposite direction:
- * <pre>{@code
- * tf.cumsum([a, b, c], reverse=True)  # => [a + b + c, b + c, c]
- * }</pre>
- * This is more efficient than using separate `tf.reverse` ops.
- * <p>
- * The `reverse` and `exclusive` kwargs can also be combined:
- * <pre>{@code
- * tf.cumsum([a, b, c], exclusive=True, reverse=True)  # => [b + c, c, 0]
- * }</pre>
- * 
- * 
- * @param <T> data type for {@code out()} output
- */
-@Operator(group = "math")
-public final class Cumsum<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.math.Cumsum}
-   */
-  public static class Options {
-    
-    /**
-     * @param exclusive If `True`, perform exclusive cumsum.
-     */
-    public Options exclusive(Boolean exclusive) {
-      this.exclusive = exclusive;
-      return this;
-    }
-    
-    /**
-     * @param reverse A `bool` (default: False).
-     */
-    public Options reverse(Boolean reverse) {
-      this.reverse = reverse;
-      return this;
-    }
-    
-    private Boolean exclusive;
-    private Boolean reverse;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new Cumsum operation.
-   * 
-   * @param scope current scope
-   * @param x A `Tensor`. Must be one of the following types: `float32`, `float64`,
-   * `int64`, `int32`, `uint8`, `uint16`, `int16`, `int8`, `complex64`,
-   * `complex128`, `qint8`, `quint8`, `qint32`, `half`.
-   * @param axis A `Tensor` of type `int32` (default: 0). Must be in the range
-   * `[-rank(x), rank(x))`.
-   * @param options carries optional attributes values
-   * @return a new instance of Cumsum
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> Cumsum<T> create(Scope scope, Operand<T> x, Operand<U> axis, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Cumsum", scope.makeOpName("Cumsum"));
-    opBuilder.addInput(x.asOutput(scope));
-    opBuilder.addInput(axis.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.exclusive != null) {
-          opBuilder.setAttr("exclusive", opts.exclusive);
-        }
-        if (opts.reverse != null) {
-          opBuilder.setAttr("reverse", opts.reverse);
-        }
-      }
-    }
-    return new Cumsum<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param exclusive If `True`, perform exclusive cumsum.
-   */
-  public static Options exclusive(Boolean exclusive) {
-    return new Options().exclusive(exclusive);
-  }
-  
-  /**
-   * @param reverse A `bool` (default: False).
-   */
-  public static Options reverse(Boolean reverse) {
-    return new Options().reverse(reverse);
-  }
-  
-  /**
-   */
-  public Output<T> out() {
-    return out;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return out;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Cumsum";
-  
-  private Output<T> out;
-  
-  private Cumsum(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    out = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/CumulativeLogsumexp.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/CumulativeLogsumexp.java
deleted file mode 100644
index 640d4cb05ca..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/CumulativeLogsumexp.java
+++ /dev/null
@@ -1,147 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.math;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Compute the cumulative product of the tensor `x` along `axis`.
- * <p>
- * By default, this op performs an inclusive cumulative log-sum-exp,
- * which means that the first
- * element of the input is identical to the first element of the output:
- * <pre>{@code
- * tf.math.cumulative_logsumexp([a, b, c])  # => [a, log(exp(a) + exp(b)), log(exp(a) + exp(b) + exp(c))]
- * }</pre>
- * By setting the `exclusive` kwarg to `True`, an exclusive cumulative log-sum-exp is
- * performed instead:
- * <pre>{@code
- * tf.cumulative_logsumexp([a, b, c], exclusive=True)  # => [-inf, a, log(exp(a) * exp(b))]
- * }</pre>
- * Note that the neutral element of the log-sum-exp operation is `-inf`,
- * however, for performance reasons, the minimal value representable by the
- * floating point type is used instead.
- * <p>
- * By setting the `reverse` kwarg to `True`, the cumulative log-sum-exp is performed in the
- * opposite direction.
- * 
- * @param <T> data type for {@code out()} output
- */
-public final class CumulativeLogsumexp<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.math.CumulativeLogsumexp}
-   */
-  public static class Options {
-    
-    /**
-     * @param exclusive If `True`, perform exclusive cumulative log-sum-exp.
-     */
-    public Options exclusive(Boolean exclusive) {
-      this.exclusive = exclusive;
-      return this;
-    }
-    
-    /**
-     * @param reverse A `bool` (default: False).
-     */
-    public Options reverse(Boolean reverse) {
-      this.reverse = reverse;
-      return this;
-    }
-    
-    private Boolean exclusive;
-    private Boolean reverse;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new CumulativeLogsumexp operation.
-   * 
-   * @param scope current scope
-   * @param x A `Tensor`. Must be one of the following types: `float16`, `float32`, `float64`.
-   * @param axis A `Tensor` of type `int32` (default: 0). Must be in the range
-   * `[-rank(x), rank(x))`.
-   * @param options carries optional attributes values
-   * @return a new instance of CumulativeLogsumexp
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber, U extends TNumber> CumulativeLogsumexp<T> create(Scope scope, Operand<T> x, Operand<U> axis, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("CumulativeLogsumexp", scope.makeOpName("CumulativeLogsumexp"));
-    opBuilder.addInput(x.asOutput(scope));
-    opBuilder.addInput(axis.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.exclusive != null) {
-          opBuilder.setAttr("exclusive", opts.exclusive);
-        }
-        if (opts.reverse != null) {
-          opBuilder.setAttr("reverse", opts.reverse);
-        }
-      }
-    }
-    return new CumulativeLogsumexp<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param exclusive If `True`, perform exclusive cumulative log-sum-exp.
-   */
-  public static Options exclusive(Boolean exclusive) {
-    return new Options().exclusive(exclusive);
-  }
-  
-  /**
-   * @param reverse A `bool` (default: False).
-   */
-  public static Options reverse(Boolean reverse) {
-    return new Options().reverse(reverse);
-  }
-  
-  /**
-   */
-  public Output<T> out() {
-    return out;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return out;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "CumulativeLogsumexp";
-  
-  private Output<T> out;
-  
-  private CumulativeLogsumexp(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    out = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/DenseBincount.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/DenseBincount.java
deleted file mode 100644
index 9cf288414a5..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/DenseBincount.java
+++ /dev/null
@@ -1,124 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.math;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Counts the number of occurrences of each value in an integer array.
- * <p>
- * Outputs a vector with length `size` and the same dtype as `weights`. If
- * `weights` are empty, then index `i` stores the number of times the value `i` is
- * counted in `arr`. If `weights` are non-empty, then index `i` stores the sum of
- * the value in `weights` at each index where the corresponding value in `arr` is
- * `i`.
- * <p>
- * Values in `arr` outside of the range [0, size) are ignored.
- * 
- * @param <U> data type for {@code output()} output
- */
-@Operator(group = "math")
-public final class DenseBincount<U extends TNumber> extends RawOp implements Operand<U> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.math.DenseBincount}
-   */
-  public static class Options {
-    
-    /**
-     * @param binaryOutput bool; Whether the kernel should count the appearance or number of occurrences.
-     */
-    public Options binaryOutput(Boolean binaryOutput) {
-      this.binaryOutput = binaryOutput;
-      return this;
-    }
-    
-    private Boolean binaryOutput;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new DenseBincount operation.
-   * 
-   * @param scope current scope
-   * @param input 1D or 2D int `Tensor`.
-   * @param size non-negative int scalar `Tensor`.
-   * @param weights is an int32, int64, float32, or float64 `Tensor` with the same
-   * shape as `arr`, or a length-0 `Tensor`, in which case it acts as all weights
-   * equal to 1.
-   * @param options carries optional attributes values
-   * @return a new instance of DenseBincount
-   */
-  @Endpoint(describeByClass = true)
-  public static <U extends TNumber, T extends TNumber> DenseBincount<U> create(Scope scope, Operand<T> input, Operand<T> size, Operand<U> weights, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("DenseBincount", scope.makeOpName("DenseBincount"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder.addInput(size.asOutput(scope));
-    opBuilder.addInput(weights.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.binaryOutput != null) {
-          opBuilder.setAttr("binary_output", opts.binaryOutput);
-        }
-      }
-    }
-    return new DenseBincount<U>(opBuilder.build());
-  }
-  
-  /**
-   * @param binaryOutput bool; Whether the kernel should count the appearance or number of occurrences.
-   */
-  public static Options binaryOutput(Boolean binaryOutput) {
-    return new Options().binaryOutput(binaryOutput);
-  }
-  
-  /**
-   * 1D `Tensor` with length equal to `size` or 2D `Tensor` with [batch_size, `size`].
-   * The counts or summed weights for each value in the range [0, size).
-   */
-  public Output<U> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<U> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "DenseBincount";
-  
-  private Output<U> output;
-  
-  private DenseBincount(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Digamma.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Digamma.java
deleted file mode 100644
index d1d87138af7..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Digamma.java
+++ /dev/null
@@ -1,76 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.math;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Computes Psi, the derivative of Lgamma (the log of the absolute value of
- * <p>
- * `Gamma(x)`), element-wise.
- * 
- * @param <T> data type for {@code y()} output
- */
-@Operator(group = "math")
-public final class Digamma<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new Digamma operation.
-   * 
-   * @param scope current scope
-   * @param x 
-   * @return a new instance of Digamma
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> Digamma<T> create(Scope scope, Operand<T> x) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Digamma", scope.makeOpName("Digamma"));
-    opBuilder.addInput(x.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new Digamma<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> y() {
-    return y;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return y;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Digamma";
-  
-  private Output<T> y;
-  
-  private Digamma(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    y = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Div.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Div.java
deleted file mode 100644
index f86f9909a49..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Div.java
+++ /dev/null
@@ -1,79 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.math;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Returns x / y element-wise.
- * <p>
- * <i>NOTE</i>: `math.Div` supports broadcasting. More about broadcasting
- * [here](http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html)
- * 
- * @param <T> data type for {@code z()} output
- */
-@Operator(group = "math")
-public final class Div<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new Div operation.
-   * 
-   * @param scope current scope
-   * @param x 
-   * @param y 
-   * @return a new instance of Div
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> Div<T> create(Scope scope, Operand<T> x, Operand<T> y) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Div", scope.makeOpName("Div"));
-    opBuilder.addInput(x.asOutput(scope));
-    opBuilder.addInput(y.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new Div<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> z() {
-    return z;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return z;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Div";
-  
-  private Output<T> z;
-  
-  private Div(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    z = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/DivNoNan.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/DivNoNan.java
deleted file mode 100644
index 3f4607025fd..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/DivNoNan.java
+++ /dev/null
@@ -1,80 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.math;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Returns 0 if the denominator is zero.
- * <p>
- * 
- * <i>NOTE</i>: `math.DivNoNan` supports broadcasting. More about broadcasting
- * [here](http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html)
- * 
- * @param <T> data type for {@code z()} output
- */
-@Operator(group = "math")
-public final class DivNoNan<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new DivNoNan operation.
-   * 
-   * @param scope current scope
-   * @param x 
-   * @param y 
-   * @return a new instance of DivNoNan
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> DivNoNan<T> create(Scope scope, Operand<T> x, Operand<T> y) {
-    OperationBuilder opBuilder = scope.env().opBuilder("DivNoNan", scope.makeOpName("DivNoNan"));
-    opBuilder.addInput(x.asOutput(scope));
-    opBuilder.addInput(y.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new DivNoNan<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> z() {
-    return z;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return z;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "DivNoNan";
-  
-  private Output<T> z;
-  
-  private DivNoNan(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    z = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Equal.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Equal.java
deleted file mode 100644
index 868bb3e0690..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Equal.java
+++ /dev/null
@@ -1,122 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.math;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TBool;
-import org.tensorflow.types.family.TType;
-
-/**
- * Returns the truth value of (x == y) element-wise.
- * <p>
- * <i>NOTE</i>: `math.Equal` supports broadcasting. More about broadcasting
- * [here](http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html)
- * <pre>{@code
- * x = tf.constant([2, 4])
- * y = tf.constant(2)
- * tf.math.equal(x, y) ==> array([True, False])
- * 
- * x = tf.constant([2, 4])
- * y = tf.constant([2, 4])
- * tf.math.equal(x, y) ==> array([True,  True])
- * }</pre>
- * 
- */
-@Operator(group = "math")
-public final class Equal extends RawOp implements Operand<TBool> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.math.Equal}
-   */
-  public static class Options {
-    
-    /**
-     * @param incompatibleShapeError 
-     */
-    public Options incompatibleShapeError(Boolean incompatibleShapeError) {
-      this.incompatibleShapeError = incompatibleShapeError;
-      return this;
-    }
-    
-    private Boolean incompatibleShapeError;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new Equal operation.
-   * 
-   * @param scope current scope
-   * @param x 
-   * @param y 
-   * @param options carries optional attributes values
-   * @return a new instance of Equal
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> Equal create(Scope scope, Operand<T> x, Operand<T> y, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Equal", scope.makeOpName("Equal"));
-    opBuilder.addInput(x.asOutput(scope));
-    opBuilder.addInput(y.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.incompatibleShapeError != null) {
-          opBuilder.setAttr("incompatible_shape_error", opts.incompatibleShapeError);
-        }
-      }
-    }
-    return new Equal(opBuilder.build());
-  }
-  
-  /**
-   * @param incompatibleShapeError 
-   */
-  public static Options incompatibleShapeError(Boolean incompatibleShapeError) {
-    return new Options().incompatibleShapeError(incompatibleShapeError);
-  }
-  
-  /**
-   */
-  public Output<TBool> z() {
-    return z;
-  }
-  
-  @Override
-  public Output<TBool> asOutput(Scope scope) {
-    return z;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Equal";
-  
-  private Output<TBool> z;
-  
-  private Equal(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    z = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Erf.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Erf.java
deleted file mode 100644
index f335ca9acf9..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Erf.java
+++ /dev/null
@@ -1,74 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.math;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Computes the Gauss error function of `x` element-wise.
- * 
- * @param <T> data type for {@code y()} output
- */
-@Operator(group = "math")
-public final class Erf<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new Erf operation.
-   * 
-   * @param scope current scope
-   * @param x 
-   * @return a new instance of Erf
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> Erf<T> create(Scope scope, Operand<T> x) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Erf", scope.makeOpName("Erf"));
-    opBuilder.addInput(x.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new Erf<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> y() {
-    return y;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return y;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Erf";
-  
-  private Output<T> y;
-  
-  private Erf(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    y = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Erfc.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Erfc.java
deleted file mode 100644
index 1eb9d1eb3d0..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Erfc.java
+++ /dev/null
@@ -1,74 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.math;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Computes the complementary error function of `x` element-wise.
- * 
- * @param <T> data type for {@code y()} output
- */
-@Operator(group = "math")
-public final class Erfc<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new Erfc operation.
-   * 
-   * @param scope current scope
-   * @param x 
-   * @return a new instance of Erfc
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> Erfc<T> create(Scope scope, Operand<T> x) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Erfc", scope.makeOpName("Erfc"));
-    opBuilder.addInput(x.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new Erfc<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> y() {
-    return y;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return y;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Erfc";
-  
-  private Output<T> y;
-  
-  private Erfc(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    y = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Exp.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Exp.java
deleted file mode 100644
index e37760b90f4..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Exp.java
+++ /dev/null
@@ -1,100 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.math;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Computes exponential of x element-wise.  \\(y = e^x\\).
- * <p>
- *   This function computes the exponential of every element in the input tensor.
- *   i.e. `exp(x)` or `e^(x)`, where `x` is the input tensor.
- *   `e` denotes Euler's number and is approximately equal to 2.718281.
- *   Output is positive for any real input.
- * <p>
- *   <pre>{@code
- *   x = tf.constant(2.0)
- *   tf.math.exp(x) ==> 7.389056
- * 
- *   x = tf.constant([2.0, 8.0])
- *   tf.math.exp(x) ==> array([7.389056, 2980.958], dtype=float32)
- *   }</pre>
- * For complex numbers, the exponential value is calculated as follows:
- * <p>
- *   <pre>{@code
- *   e^(x+iy) = e^x * e^iy = e^x * (cos y + i sin y)
- *   }</pre>
- * Let's consider complex number 1+1j as an example.
- *   e^1 * (cos 1 + i sin 1) = 2.7182818284590 * (0.54030230586+0.8414709848j)
- * <p>
- *   <pre>{@code
- *   x = tf.constant(1 + 1j)
- *   tf.math.exp(x) ==> 1.4686939399158851+2.2873552871788423j
- *   }</pre>
- * 
- * 
- * @param <T> data type for {@code y()} output
- */
-@Operator(group = "math")
-public final class Exp<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new Exp operation.
-   * 
-   * @param scope current scope
-   * @param x 
-   * @return a new instance of Exp
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> Exp<T> create(Scope scope, Operand<T> x) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Exp", scope.makeOpName("Exp"));
-    opBuilder.addInput(x.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new Exp<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> y() {
-    return y;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return y;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Exp";
-  
-  private Output<T> y;
-  
-  private Exp(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    y = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Expm1.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Expm1.java
deleted file mode 100644
index ad951dc71aa..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Expm1.java
+++ /dev/null
@@ -1,89 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.math;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Computes `exp(x) - 1` element-wise.
- * <p>
- *   i.e. `exp(x) - 1` or `e^(x) - 1`, where `x` is the input tensor.
- *   `e` denotes Euler's number and is approximately equal to 2.718281.
- * <p>
- *   <pre>{@code
- *   x = tf.constant(2.0)
- *   tf.math.expm1(x) ==> 6.389056
- * 
- *   x = tf.constant([2.0, 8.0])
- *   tf.math.expm1(x) ==> array([6.389056, 2979.958], dtype=float32)
- * 
- *   x = tf.constant(1 + 1j)
- *   tf.math.expm1(x) ==> (0.46869393991588515+2.2873552871788423j)
- *   }</pre>
- * 
- * 
- * @param <T> data type for {@code y()} output
- */
-@Operator(group = "math")
-public final class Expm1<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new Expm1 operation.
-   * 
-   * @param scope current scope
-   * @param x 
-   * @return a new instance of Expm1
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> Expm1<T> create(Scope scope, Operand<T> x) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Expm1", scope.makeOpName("Expm1"));
-    opBuilder.addInput(x.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new Expm1<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> y() {
-    return y;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return y;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Expm1";
-  
-  private Output<T> y;
-  
-  private Expm1(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    y = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Fact.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Fact.java
deleted file mode 100644
index dd199eb7759..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Fact.java
+++ /dev/null
@@ -1,70 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.math;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TString;
-
-/**
- * Output a fact about factorials.
- */
-@Operator(group = "math")
-public final class Fact extends RawOp implements Operand<TString> {
-  
-  /**
-   * Factory method to create a class wrapping a new Fact operation.
-   * 
-   * @param scope current scope
-   * @return a new instance of Fact
-   */
-  @Endpoint(describeByClass = true)
-  public static Fact create(Scope scope) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Fact", scope.makeOpName("Fact"));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new Fact(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<TString> fact() {
-    return fact;
-  }
-  
-  @Override
-  public Output<TString> asOutput(Scope scope) {
-    return fact;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Fact";
-  
-  private Output<TString> fact;
-  
-  private Fact(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    fact = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Floor.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Floor.java
deleted file mode 100644
index 190c1c3f7a1..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Floor.java
+++ /dev/null
@@ -1,74 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.math;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Returns element-wise largest integer not greater than x.
- * 
- * @param <T> data type for {@code y()} output
- */
-@Operator(group = "math")
-public final class Floor<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new Floor operation.
-   * 
-   * @param scope current scope
-   * @param x 
-   * @return a new instance of Floor
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> Floor<T> create(Scope scope, Operand<T> x) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Floor", scope.makeOpName("Floor"));
-    opBuilder.addInput(x.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new Floor<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> y() {
-    return y;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return y;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Floor";
-  
-  private Output<T> y;
-  
-  private Floor(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    y = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/FloorDiv.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/FloorDiv.java
deleted file mode 100644
index 2d92c796459..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/FloorDiv.java
+++ /dev/null
@@ -1,79 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.math;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Returns x // y element-wise.
- * <p>
- * <i>NOTE</i>: `math.FloorDiv` supports broadcasting. More about broadcasting
- * [here](http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html)
- * 
- * @param <T> data type for {@code z()} output
- */
-@Operator(group = "math")
-public final class FloorDiv<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new FloorDiv operation.
-   * 
-   * @param scope current scope
-   * @param x 
-   * @param y 
-   * @return a new instance of FloorDiv
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> FloorDiv<T> create(Scope scope, Operand<T> x, Operand<T> y) {
-    OperationBuilder opBuilder = scope.env().opBuilder("FloorDiv", scope.makeOpName("FloorDiv"));
-    opBuilder.addInput(x.asOutput(scope));
-    opBuilder.addInput(y.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new FloorDiv<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> z() {
-    return z;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return z;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "FloorDiv";
-  
-  private Output<T> z;
-  
-  private FloorDiv(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    z = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/FloorMod.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/FloorMod.java
deleted file mode 100644
index a8502331592..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/FloorMod.java
+++ /dev/null
@@ -1,82 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.math;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Returns element-wise remainder of division. When `x < 0` xor `y < 0` is
- * <p>
- * true, this follows Python semantics in that the result here is consistent
- * with a flooring divide. E.g. `floor(x / y) * y + mod(x, y) = x`.
- * <p>
- * <i>NOTE</i>: `math.FloorMod` supports broadcasting. More about broadcasting
- * [here](http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html)
- * 
- * @param <T> data type for {@code z()} output
- */
-@Operator(group = "math")
-public final class FloorMod<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new FloorMod operation.
-   * 
-   * @param scope current scope
-   * @param x 
-   * @param y 
-   * @return a new instance of FloorMod
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> FloorMod<T> create(Scope scope, Operand<T> x, Operand<T> y) {
-    OperationBuilder opBuilder = scope.env().opBuilder("FloorMod", scope.makeOpName("FloorMod"));
-    opBuilder.addInput(x.asOutput(scope));
-    opBuilder.addInput(y.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new FloorMod<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> z() {
-    return z;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return z;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "FloorMod";
-  
-  private Output<T> z;
-  
-  private FloorMod(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    z = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Greater.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Greater.java
deleted file mode 100644
index 5a5ab1ce6e8..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Greater.java
+++ /dev/null
@@ -1,90 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.math;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TBool;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Returns the truth value of (x > y) element-wise.
- * <p>
- * <i>NOTE</i>: `math.Greater` supports broadcasting. More about broadcasting
- * [here](http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html)
- * <p>
- * Example:
- * <pre>{@code
- * x = tf.constant([5, 4, 6])
- * y = tf.constant([5, 2, 5])
- * tf.math.greater(x, y) ==> [False, True, True]
- * 
- * x = tf.constant([5, 4, 6])
- * y = tf.constant([5])
- * tf.math.greater(x, y) ==> [False, False, True]
- * }</pre>
- * 
- */
-@Operator(group = "math")
-public final class Greater extends RawOp implements Operand<TBool> {
-  
-  /**
-   * Factory method to create a class wrapping a new Greater operation.
-   * 
-   * @param scope current scope
-   * @param x 
-   * @param y 
-   * @return a new instance of Greater
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> Greater create(Scope scope, Operand<T> x, Operand<T> y) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Greater", scope.makeOpName("Greater"));
-    opBuilder.addInput(x.asOutput(scope));
-    opBuilder.addInput(y.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new Greater(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<TBool> z() {
-    return z;
-  }
-  
-  @Override
-  public Output<TBool> asOutput(Scope scope) {
-    return z;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Greater";
-  
-  private Output<TBool> z;
-  
-  private Greater(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    z = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/GreaterEqual.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/GreaterEqual.java
deleted file mode 100644
index 88543131abb..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/GreaterEqual.java
+++ /dev/null
@@ -1,90 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.math;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TBool;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Returns the truth value of (x >= y) element-wise.
- * <p>
- * <i>NOTE</i>: `math.GreaterEqual` supports broadcasting. More about broadcasting
- * [here](http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html)
- * <p>
- * Example:
- * <pre>{@code
- * x = tf.constant([5, 4, 6, 7])
- * y = tf.constant([5, 2, 5, 10])
- * tf.math.greater_equal(x, y) ==> [True, True, True, False]
- * 
- * x = tf.constant([5, 4, 6, 7])
- * y = tf.constant([5])
- * tf.math.greater_equal(x, y) ==> [True, False, True, True]
- * }</pre>
- * 
- */
-@Operator(group = "math")
-public final class GreaterEqual extends RawOp implements Operand<TBool> {
-  
-  /**
-   * Factory method to create a class wrapping a new GreaterEqual operation.
-   * 
-   * @param scope current scope
-   * @param x 
-   * @param y 
-   * @return a new instance of GreaterEqual
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> GreaterEqual create(Scope scope, Operand<T> x, Operand<T> y) {
-    OperationBuilder opBuilder = scope.env().opBuilder("GreaterEqual", scope.makeOpName("GreaterEqual"));
-    opBuilder.addInput(x.asOutput(scope));
-    opBuilder.addInput(y.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new GreaterEqual(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<TBool> z() {
-    return z;
-  }
-  
-  @Override
-  public Output<TBool> asOutput(Scope scope) {
-    return z;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "GreaterEqual";
-  
-  private Output<TBool> z;
-  
-  private GreaterEqual(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    z = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Igamma.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Igamma.java
deleted file mode 100644
index 141cc603bae..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Igamma.java
+++ /dev/null
@@ -1,89 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.math;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Compute the lower regularized incomplete Gamma function `P(a, x)`.
- * <p>
- * The lower regularized incomplete Gamma function is defined as:
- * <p>
- * \\(P(a, x) = gamma(a, x) / Gamma(a) = 1 - Q(a, x)\\)
- * <p>
- * where
- * <p>
- * \\(gamma(a, x) = \\int_{0}^{x} t^{a-1} exp(-t) dt\\)
- * <p>
- * is the lower incomplete Gamma function.
- * <p>
- * Note, above `Q(a, x)` (`Igammac`) is the upper regularized complete
- * Gamma function.
- * 
- * @param <T> data type for {@code z()} output
- */
-@Operator(group = "math")
-public final class Igamma<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new Igamma operation.
-   * 
-   * @param scope current scope
-   * @param a 
-   * @param x 
-   * @return a new instance of Igamma
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> Igamma<T> create(Scope scope, Operand<T> a, Operand<T> x) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Igamma", scope.makeOpName("Igamma"));
-    opBuilder.addInput(a.asOutput(scope));
-    opBuilder.addInput(x.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new Igamma<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> z() {
-    return z;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return z;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Igamma";
-  
-  private Output<T> z;
-  
-  private Igamma(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    z = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/IgammaGradA.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/IgammaGradA.java
deleted file mode 100644
index 5320b437133..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/IgammaGradA.java
+++ /dev/null
@@ -1,75 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.math;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Computes the gradient of `igamma(a, x)` wrt `a`.
- * 
- * @param <T> data type for {@code z()} output
- */
-public final class IgammaGradA<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new IgammaGradA operation.
-   * 
-   * @param scope current scope
-   * @param a 
-   * @param x 
-   * @return a new instance of IgammaGradA
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> IgammaGradA<T> create(Scope scope, Operand<T> a, Operand<T> x) {
-    OperationBuilder opBuilder = scope.env().opBuilder("IgammaGradA", scope.makeOpName("IgammaGradA"));
-    opBuilder.addInput(a.asOutput(scope));
-    opBuilder.addInput(x.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new IgammaGradA<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> z() {
-    return z;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return z;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "IgammaGradA";
-  
-  private Output<T> z;
-  
-  private IgammaGradA(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    z = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Igammac.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Igammac.java
deleted file mode 100644
index 597e944d1fb..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Igammac.java
+++ /dev/null
@@ -1,89 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.math;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Compute the upper regularized incomplete Gamma function `Q(a, x)`.
- * <p>
- * The upper regularized incomplete Gamma function is defined as:
- * <p>
- * \\(Q(a, x) = Gamma(a, x) / Gamma(a) = 1 - P(a, x)\\)
- * <p>
- * where
- * <p>
- * \\(Gamma(a, x) = int_{x}^{\infty} t^{a-1} exp(-t) dt\\)
- * <p>
- * is the upper incomplete Gama function.
- * <p>
- * Note, above `P(a, x)` (`Igamma`) is the lower regularized complete
- * Gamma function.
- * 
- * @param <T> data type for {@code z()} output
- */
-@Operator(group = "math")
-public final class Igammac<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new Igammac operation.
-   * 
-   * @param scope current scope
-   * @param a 
-   * @param x 
-   * @return a new instance of Igammac
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> Igammac<T> create(Scope scope, Operand<T> a, Operand<T> x) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Igammac", scope.makeOpName("Igammac"));
-    opBuilder.addInput(a.asOutput(scope));
-    opBuilder.addInput(x.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new Igammac<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> z() {
-    return z;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return z;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Igammac";
-  
-  private Output<T> z;
-  
-  private Igammac(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    z = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Imag.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Imag.java
deleted file mode 100644
index ce18da2105d..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Imag.java
+++ /dev/null
@@ -1,102 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.math;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * Returns the imaginary part of a complex number.
- * <p>
- * Given a tensor `input` of complex numbers, this operation returns a tensor of
- * type `float` that is the imaginary part of each element in `input`. All
- * elements in `input` must be complex numbers of the form \\(a + bj\\), where <i>a</i>
- * is the real part and <i>b</i> is the imaginary part returned by this operation.
- * <p>
- * For example:
- * <pre>{@code
- * # tensor 'input' is [-2.25 + 4.75j, 3.25 + 5.75j]
- * tf.imag(input) ==> [4.75, 5.75]
- * }</pre>
- * 
- * 
- * @param <U> data type for {@code output()} output
- */
-@Operator(group = "math")
-public final class Imag<U extends TNumber> extends RawOp implements Operand<U> {
-  
-  /**
-   * Factory method to create a class wrapping a new Imag operation.
-   * 
-   * @param scope current scope
-   * @param input 
-   * @param Tout 
-   * @return a new instance of Imag
-   */
-  @Endpoint(describeByClass = true)
-  public static <U extends TNumber, T extends TType> Imag<U> create(Scope scope, Operand<T> input, Class<U> Tout) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Imag", scope.makeOpName("Imag"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("Tout", Tout);
-    return new Imag<U>(opBuilder.build());
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new Imag operation using default output types.
-   * 
-   * @param scope current scope
-   * @param input 
-   * @return a new instance of Imag
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> Imag<TFloat32> create(Scope scope, Operand<T> input) {
-    return create(scope, input, TFloat32.class);
-  }
-  
-  /**
-   */
-  public Output<U> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<U> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Imag";
-  
-  private Output<U> output;
-  
-  private Imag(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/InvertPermutation.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/InvertPermutation.java
deleted file mode 100644
index e6f38ded46b..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/InvertPermutation.java
+++ /dev/null
@@ -1,91 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.math;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Computes the inverse permutation of a tensor.
- * <p>
- * This operation computes the inverse of an index permutation. It takes a 1-D
- * integer tensor `x`, which represents the indices of a zero-based array, and
- * swaps each value with its index position. In other words, for an output tensor
- * `y` and an input tensor `x`, this operation computes the following:
- * <p>
- * `y[x[i]] = i for i in [0, 1, ..., len(x) - 1]`
- * <p>
- * The values must include 0. There can be no duplicate values or negative values.
- * <p>
- * For example:
- * <pre>{@code
- * # tensor `x` is [3, 4, 0, 2, 1]
- * invert_permutation(x) ==> [2, 4, 3, 0, 1]
- * }</pre>
- * 
- * 
- * @param <T> data type for {@code y()} output
- */
-@Operator(group = "math")
-public final class InvertPermutation<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new InvertPermutation operation.
-   * 
-   * @param scope current scope
-   * @param x 1-D.
-   * @return a new instance of InvertPermutation
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> InvertPermutation<T> create(Scope scope, Operand<T> x) {
-    OperationBuilder opBuilder = scope.env().opBuilder("InvertPermutation", scope.makeOpName("InvertPermutation"));
-    opBuilder.addInput(x.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new InvertPermutation<T>(opBuilder.build());
-  }
-  
-  /**
-   * 1-D.
-   */
-  public Output<T> y() {
-    return y;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return y;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "InvertPermutation";
-  
-  private Output<T> y;
-  
-  private InvertPermutation(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    y = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/IsFinite.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/IsFinite.java
deleted file mode 100644
index ec1566a22ae..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/IsFinite.java
+++ /dev/null
@@ -1,84 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.math;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TBool;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Returns which elements of x are finite.
- * <p>
- * @compatibility(numpy)
- * Equivalent to np.isfinite
- * @end_compatibility
- * <p>
- * Example:
- * <pre>{@code
- * x = tf.constant([5.0, 4.8, 6.8, np.inf, np.nan])
- * tf.math.is_finite(x) ==> [True, True, True, False, False]
- * }</pre>
- * 
- */
-@Operator(group = "math")
-public final class IsFinite extends RawOp implements Operand<TBool> {
-  
-  /**
-   * Factory method to create a class wrapping a new IsFinite operation.
-   * 
-   * @param scope current scope
-   * @param x 
-   * @return a new instance of IsFinite
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> IsFinite create(Scope scope, Operand<T> x) {
-    OperationBuilder opBuilder = scope.env().opBuilder("IsFinite", scope.makeOpName("IsFinite"));
-    opBuilder.addInput(x.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new IsFinite(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<TBool> y() {
-    return y;
-  }
-  
-  @Override
-  public Output<TBool> asOutput(Scope scope) {
-    return y;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "IsFinite";
-  
-  private Output<TBool> y;
-  
-  private IsFinite(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    y = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/IsInf.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/IsInf.java
deleted file mode 100644
index 7b91b7e4563..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/IsInf.java
+++ /dev/null
@@ -1,84 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.math;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TBool;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Returns which elements of x are Inf.
- * <p>
- * @compatibility(numpy)
- * Equivalent to np.isinf
- * @end_compatibility
- * <p>
- * Example:
- * <pre>{@code
- * x = tf.constant([5.0, np.inf, 6.8, np.inf])
- * tf.math.is_inf(x) ==> [False, True, False, True]
- * }</pre>
- * 
- */
-@Operator(group = "math")
-public final class IsInf extends RawOp implements Operand<TBool> {
-  
-  /**
-   * Factory method to create a class wrapping a new IsInf operation.
-   * 
-   * @param scope current scope
-   * @param x 
-   * @return a new instance of IsInf
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> IsInf create(Scope scope, Operand<T> x) {
-    OperationBuilder opBuilder = scope.env().opBuilder("IsInf", scope.makeOpName("IsInf"));
-    opBuilder.addInput(x.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new IsInf(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<TBool> y() {
-    return y;
-  }
-  
-  @Override
-  public Output<TBool> asOutput(Scope scope) {
-    return y;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "IsInf";
-  
-  private Output<TBool> y;
-  
-  private IsInf(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    y = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/IsNan.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/IsNan.java
deleted file mode 100644
index c02865ab759..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/IsNan.java
+++ /dev/null
@@ -1,84 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.math;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TBool;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Returns which elements of x are NaN.
- * <p>
- * @compatibility(numpy)
- * Equivalent to np.isnan
- * @end_compatibility
- * <p>
- * Example:
- * <pre>{@code
- * x = tf.constant([5.0, np.nan, 6.8, np.nan, np.inf])
- * tf.math.is_nan(x) ==> [False, True, False, True, False]
- * }</pre>
- * 
- */
-@Operator(group = "math")
-public final class IsNan extends RawOp implements Operand<TBool> {
-  
-  /**
-   * Factory method to create a class wrapping a new IsNan operation.
-   * 
-   * @param scope current scope
-   * @param x 
-   * @return a new instance of IsNan
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> IsNan create(Scope scope, Operand<T> x) {
-    OperationBuilder opBuilder = scope.env().opBuilder("IsNan", scope.makeOpName("IsNan"));
-    opBuilder.addInput(x.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new IsNan(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<TBool> y() {
-    return y;
-  }
-  
-  @Override
-  public Output<TBool> asOutput(Scope scope) {
-    return y;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "IsNan";
-  
-  private Output<TBool> y;
-  
-  private IsNan(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    y = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Less.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Less.java
deleted file mode 100644
index 7ce3046b5b7..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Less.java
+++ /dev/null
@@ -1,90 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.math;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TBool;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Returns the truth value of (x < y) element-wise.
- * <p>
- * <i>NOTE</i>: `math.Less` supports broadcasting. More about broadcasting
- * [here](http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html)
- * <p>
- * Example:
- * <pre>{@code
- * x = tf.constant([5, 4, 6])
- * y = tf.constant([5])
- * tf.math.less(x, y) ==> [False, True, False]
- * 
- * x = tf.constant([5, 4, 6])
- * y = tf.constant([5, 6, 7])
- * tf.math.less(x, y) ==> [False, True, True]
- * }</pre>
- * 
- */
-@Operator(group = "math")
-public final class Less extends RawOp implements Operand<TBool> {
-  
-  /**
-   * Factory method to create a class wrapping a new Less operation.
-   * 
-   * @param scope current scope
-   * @param x 
-   * @param y 
-   * @return a new instance of Less
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> Less create(Scope scope, Operand<T> x, Operand<T> y) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Less", scope.makeOpName("Less"));
-    opBuilder.addInput(x.asOutput(scope));
-    opBuilder.addInput(y.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new Less(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<TBool> z() {
-    return z;
-  }
-  
-  @Override
-  public Output<TBool> asOutput(Scope scope) {
-    return z;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Less";
-  
-  private Output<TBool> z;
-  
-  private Less(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    z = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/LessEqual.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/LessEqual.java
deleted file mode 100644
index 0e36cbc0223..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/LessEqual.java
+++ /dev/null
@@ -1,90 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.math;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TBool;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Returns the truth value of (x <= y) element-wise.
- * <p>
- * <i>NOTE</i>: `math.LessEqual` supports broadcasting. More about broadcasting
- * [here](http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html)
- * <p>
- * Example:
- * <pre>{@code
- * x = tf.constant([5, 4, 6])
- * y = tf.constant([5])
- * tf.math.less_equal(x, y) ==> [True, True, False]
- * 
- * x = tf.constant([5, 4, 6])
- * y = tf.constant([5, 6, 6])
- * tf.math.less_equal(x, y) ==> [True, True, True]
- * }</pre>
- * 
- */
-@Operator(group = "math")
-public final class LessEqual extends RawOp implements Operand<TBool> {
-  
-  /**
-   * Factory method to create a class wrapping a new LessEqual operation.
-   * 
-   * @param scope current scope
-   * @param x 
-   * @param y 
-   * @return a new instance of LessEqual
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> LessEqual create(Scope scope, Operand<T> x, Operand<T> y) {
-    OperationBuilder opBuilder = scope.env().opBuilder("LessEqual", scope.makeOpName("LessEqual"));
-    opBuilder.addInput(x.asOutput(scope));
-    opBuilder.addInput(y.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new LessEqual(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<TBool> z() {
-    return z;
-  }
-  
-  @Override
-  public Output<TBool> asOutput(Scope scope) {
-    return z;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "LessEqual";
-  
-  private Output<TBool> z;
-  
-  private LessEqual(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    z = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Lgamma.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Lgamma.java
deleted file mode 100644
index d37aced67e9..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Lgamma.java
+++ /dev/null
@@ -1,84 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.math;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Computes the log of the absolute value of `Gamma(x)` element-wise.
- * <p>
- *   For positive numbers, this function computes log((input - 1)!) for every element in the tensor.
- *   `lgamma(5) = log((5-1)!) = log(4!) = log(24) = 3.1780539`
- * <p>
- * Example:
- * <pre>{@code
- * x = tf.constant([0, 0.5, 1, 4.5, -4, -5.6])
- * tf.math.lgamma(x) ==> [inf, 0.5723649, 0., 2.4537368, inf, -4.6477685]
- * }</pre>
- * 
- * 
- * @param <T> data type for {@code y()} output
- */
-@Operator(group = "math")
-public final class Lgamma<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new Lgamma operation.
-   * 
-   * @param scope current scope
-   * @param x 
-   * @return a new instance of Lgamma
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> Lgamma<T> create(Scope scope, Operand<T> x) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Lgamma", scope.makeOpName("Lgamma"));
-    opBuilder.addInput(x.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new Lgamma<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> y() {
-    return y;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return y;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Lgamma";
-  
-  private Output<T> y;
-  
-  private Lgamma(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    y = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Log.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Log.java
deleted file mode 100644
index a4c7d04325d..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Log.java
+++ /dev/null
@@ -1,83 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.math;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Computes natural logarithm of x element-wise.
- * <p>
- * I.e., \\(y = \log_e x\\).
- * <p>
- * Example:
- * <pre>{@code
- * x = tf.constant([0, 0.5, 1, 5])
- * tf.math.log(x) ==> [-inf, -0.6931472,  0. ,  1.609438]
- * }</pre>
- * 
- * 
- * @param <T> data type for {@code y()} output
- */
-@Operator(group = "math")
-public final class Log<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new Log operation.
-   * 
-   * @param scope current scope
-   * @param x 
-   * @return a new instance of Log
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> Log<T> create(Scope scope, Operand<T> x) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Log", scope.makeOpName("Log"));
-    opBuilder.addInput(x.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new Log<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> y() {
-    return y;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return y;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Log";
-  
-  private Output<T> y;
-  
-  private Log(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    y = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Log1p.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Log1p.java
deleted file mode 100644
index 3f1c418d3ed..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Log1p.java
+++ /dev/null
@@ -1,83 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.math;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Computes natural logarithm of (1 + x) element-wise.
- * <p>
- * I.e., \\(y = \log_e (1 + x)\\).
- * <p>
- * Example:
- * <pre>{@code
- * x = tf.constant([0, 0.5, 1, 5])
- * tf.math.log1p(x) ==> [0., 0.4054651, 0.6931472, 1.7917595]
- * }</pre>
- * 
- * 
- * @param <T> data type for {@code y()} output
- */
-@Operator(group = "math")
-public final class Log1p<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new Log1p operation.
-   * 
-   * @param scope current scope
-   * @param x 
-   * @return a new instance of Log1p
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> Log1p<T> create(Scope scope, Operand<T> x) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Log1p", scope.makeOpName("Log1p"));
-    opBuilder.addInput(x.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new Log1p<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> y() {
-    return y;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return y;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Log1p";
-  
-  private Output<T> y;
-  
-  private Log1p(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    y = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/LogicalAnd.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/LogicalAnd.java
deleted file mode 100644
index a77dfbe0a73..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/LogicalAnd.java
+++ /dev/null
@@ -1,77 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.math;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TBool;
-
-/**
- * Returns the truth value of x AND y element-wise.
- * <p>
- * <i>NOTE</i>: `math.LogicalAnd` supports broadcasting. More about broadcasting
- * [here](http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html)
- */
-@Operator(group = "math")
-public final class LogicalAnd extends RawOp implements Operand<TBool> {
-  
-  /**
-   * Factory method to create a class wrapping a new LogicalAnd operation.
-   * 
-   * @param scope current scope
-   * @param x 
-   * @param y 
-   * @return a new instance of LogicalAnd
-   */
-  @Endpoint(describeByClass = true)
-  public static LogicalAnd create(Scope scope, Operand<TBool> x, Operand<TBool> y) {
-    OperationBuilder opBuilder = scope.env().opBuilder("LogicalAnd", scope.makeOpName("LogicalAnd"));
-    opBuilder.addInput(x.asOutput(scope));
-    opBuilder.addInput(y.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new LogicalAnd(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<TBool> z() {
-    return z;
-  }
-  
-  @Override
-  public Output<TBool> asOutput(Scope scope) {
-    return z;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "LogicalAnd";
-  
-  private Output<TBool> z;
-  
-  private LogicalAnd(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    z = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/LogicalNot.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/LogicalNot.java
deleted file mode 100644
index 58376781559..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/LogicalNot.java
+++ /dev/null
@@ -1,73 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.math;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TBool;
-
-/**
- * Returns the truth value of `NOT x` element-wise.
- */
-@Operator(group = "math")
-public final class LogicalNot extends RawOp implements Operand<TBool> {
-  
-  /**
-   * Factory method to create a class wrapping a new LogicalNot operation.
-   * 
-   * @param scope current scope
-   * @param x A `Tensor` of type `bool`.
-   * @return a new instance of LogicalNot
-   */
-  @Endpoint(describeByClass = true)
-  public static LogicalNot create(Scope scope, Operand<TBool> x) {
-    OperationBuilder opBuilder = scope.env().opBuilder("LogicalNot", scope.makeOpName("LogicalNot"));
-    opBuilder.addInput(x.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new LogicalNot(opBuilder.build());
-  }
-  
-  /**
-   * A `Tensor` of type `bool` with the same shape as `x`. The logical negation of `x`.
-   */
-  public Output<TBool> y() {
-    return y;
-  }
-  
-  @Override
-  public Output<TBool> asOutput(Scope scope) {
-    return y;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "LogicalNot";
-  
-  private Output<TBool> y;
-  
-  private LogicalNot(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    y = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/LogicalOr.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/LogicalOr.java
deleted file mode 100644
index 39ae7295a27..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/LogicalOr.java
+++ /dev/null
@@ -1,77 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.math;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TBool;
-
-/**
- * Returns the truth value of x OR y element-wise.
- * <p>
- * <i>NOTE</i>: `math.LogicalOr` supports broadcasting. More about broadcasting
- * [here](http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html)
- */
-@Operator(group = "math")
-public final class LogicalOr extends RawOp implements Operand<TBool> {
-  
-  /**
-   * Factory method to create a class wrapping a new LogicalOr operation.
-   * 
-   * @param scope current scope
-   * @param x 
-   * @param y 
-   * @return a new instance of LogicalOr
-   */
-  @Endpoint(describeByClass = true)
-  public static LogicalOr create(Scope scope, Operand<TBool> x, Operand<TBool> y) {
-    OperationBuilder opBuilder = scope.env().opBuilder("LogicalOr", scope.makeOpName("LogicalOr"));
-    opBuilder.addInput(x.asOutput(scope));
-    opBuilder.addInput(y.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new LogicalOr(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<TBool> z() {
-    return z;
-  }
-  
-  @Override
-  public Output<TBool> asOutput(Scope scope) {
-    return z;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "LogicalOr";
-  
-  private Output<TBool> z;
-  
-  private LogicalOr(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    z = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Maximum.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Maximum.java
deleted file mode 100644
index d9ea1d45870..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Maximum.java
+++ /dev/null
@@ -1,79 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.math;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Returns the max of x and y (i.e. x > y ? x : y) element-wise.
- * <p>
- * <i>NOTE</i>: `math.Maximum` supports broadcasting. More about broadcasting
- * [here](http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html)
- * 
- * @param <T> data type for {@code z()} output
- */
-@Operator(group = "math")
-public final class Maximum<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new Maximum operation.
-   * 
-   * @param scope current scope
-   * @param x 
-   * @param y 
-   * @return a new instance of Maximum
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> Maximum<T> create(Scope scope, Operand<T> x, Operand<T> y) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Maximum", scope.makeOpName("Maximum"));
-    opBuilder.addInput(x.asOutput(scope));
-    opBuilder.addInput(y.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new Maximum<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> z() {
-    return z;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return z;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Maximum";
-  
-  private Output<T> z;
-  
-  private Maximum(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    z = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Mean.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Mean.java
deleted file mode 100644
index 4923b136110..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Mean.java
+++ /dev/null
@@ -1,118 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.math;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * Computes the mean of elements across dimensions of a tensor.
- * <p>
- * Reduces `input` along the dimensions given in `axis`. Unless
- * `keep_dims` is true, the rank of the tensor is reduced by 1 for each entry in
- * `axis`. If `keep_dims` is true, the reduced dimensions are
- * retained with length 1.
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator(group = "math")
-public final class Mean<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.math.Mean}
-   */
-  public static class Options {
-    
-    /**
-     * @param keepDims If true, retain reduced dimensions with length 1.
-     */
-    public Options keepDims(Boolean keepDims) {
-      this.keepDims = keepDims;
-      return this;
-    }
-    
-    private Boolean keepDims;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new Mean operation.
-   * 
-   * @param scope current scope
-   * @param input The tensor to reduce.
-   * @param axis The dimensions to reduce. Must be in the range
-   * `[-rank(input), rank(input))`.
-   * @param options carries optional attributes values
-   * @return a new instance of Mean
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> Mean<T> create(Scope scope, Operand<T> input, Operand<U> axis, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Mean", scope.makeOpName("Mean"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder.addInput(axis.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.keepDims != null) {
-          opBuilder.setAttr("keep_dims", opts.keepDims);
-        }
-      }
-    }
-    return new Mean<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param keepDims If true, retain reduced dimensions with length 1.
-   */
-  public static Options keepDims(Boolean keepDims) {
-    return new Options().keepDims(keepDims);
-  }
-  
-  /**
-   * The reduced tensor.
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Mean";
-  
-  private Output<T> output;
-  
-  private Mean(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Minimum.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Minimum.java
deleted file mode 100644
index a9a7934a26a..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Minimum.java
+++ /dev/null
@@ -1,79 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.math;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Returns the min of x and y (i.e. x < y ? x : y) element-wise.
- * <p>
- * <i>NOTE</i>: `math.Minimum` supports broadcasting. More about broadcasting
- * [here](http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html)
- * 
- * @param <T> data type for {@code z()} output
- */
-@Operator(group = "math")
-public final class Minimum<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new Minimum operation.
-   * 
-   * @param scope current scope
-   * @param x 
-   * @param y 
-   * @return a new instance of Minimum
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> Minimum<T> create(Scope scope, Operand<T> x, Operand<T> y) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Minimum", scope.makeOpName("Minimum"));
-    opBuilder.addInput(x.asOutput(scope));
-    opBuilder.addInput(y.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new Minimum<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> z() {
-    return z;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return z;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Minimum";
-  
-  private Output<T> z;
-  
-  private Minimum(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    z = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Mod.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Mod.java
deleted file mode 100644
index db3c2dbd24e..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Mod.java
+++ /dev/null
@@ -1,82 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.math;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Returns element-wise remainder of division. This emulates C semantics in that
- * <p>
- * the result here is consistent with a truncating divide. E.g.
- * `tf.truncatediv(x, y) * y + truncate_mod(x, y) = x`.
- * <p>
- * <i>NOTE</i>: `math.Mod` supports broadcasting. More about broadcasting
- * [here](http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html)
- * 
- * @param <T> data type for {@code z()} output
- */
-@Operator(group = "math")
-public final class Mod<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new Mod operation.
-   * 
-   * @param scope current scope
-   * @param x 
-   * @param y 
-   * @return a new instance of Mod
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> Mod<T> create(Scope scope, Operand<T> x, Operand<T> y) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Mod", scope.makeOpName("Mod"));
-    opBuilder.addInput(x.asOutput(scope));
-    opBuilder.addInput(y.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new Mod<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> z() {
-    return z;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return z;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Mod";
-  
-  private Output<T> z;
-  
-  private Mod(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    z = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Mul.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Mul.java
deleted file mode 100644
index 778a19ceb4c..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Mul.java
+++ /dev/null
@@ -1,79 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.math;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Returns x * y element-wise.
- * <p>
- * <i>NOTE</i>: `math.Mul` supports broadcasting. More about broadcasting
- * [here](http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html)
- * 
- * @param <T> data type for {@code z()} output
- */
-@Operator(group = "math")
-public final class Mul<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new Mul operation.
-   * 
-   * @param scope current scope
-   * @param x 
-   * @param y 
-   * @return a new instance of Mul
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> Mul<T> create(Scope scope, Operand<T> x, Operand<T> y) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Mul", scope.makeOpName("Mul"));
-    opBuilder.addInput(x.asOutput(scope));
-    opBuilder.addInput(y.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new Mul<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> z() {
-    return z;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return z;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Mul";
-  
-  private Output<T> z;
-  
-  private Mul(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    z = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/MulNoNan.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/MulNoNan.java
deleted file mode 100644
index 4c73331b7ca..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/MulNoNan.java
+++ /dev/null
@@ -1,79 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.math;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Returns x * y element-wise. Returns zero if y is zero, even if x if infinite or NaN.
- * <p>
- * <i>NOTE</i>: `math.MulNoNan` supports broadcasting. More about broadcasting
- * [here](http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html)
- * 
- * @param <T> data type for {@code z()} output
- */
-@Operator(group = "math")
-public final class MulNoNan<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new MulNoNan operation.
-   * 
-   * @param scope current scope
-   * @param x 
-   * @param y 
-   * @return a new instance of MulNoNan
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> MulNoNan<T> create(Scope scope, Operand<T> x, Operand<T> y) {
-    OperationBuilder opBuilder = scope.env().opBuilder("MulNoNan", scope.makeOpName("MulNoNan"));
-    opBuilder.addInput(x.asOutput(scope));
-    opBuilder.addInput(y.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new MulNoNan<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> z() {
-    return z;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return z;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "MulNoNan";
-  
-  private Output<T> z;
-  
-  private MulNoNan(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    z = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Ndtri.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Ndtri.java
deleted file mode 100644
index 72108ba8d51..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Ndtri.java
+++ /dev/null
@@ -1,72 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.math;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * @param <T> data type for {@code y()} output
- */
-@Operator(group = "math")
-public final class Ndtri<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new Ndtri operation.
-   * 
-   * @param scope current scope
-   * @param x 
-   * @return a new instance of Ndtri
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> Ndtri<T> create(Scope scope, Operand<T> x) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Ndtri", scope.makeOpName("Ndtri"));
-    opBuilder.addInput(x.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new Ndtri<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> y() {
-    return y;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return y;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Ndtri";
-  
-  private Output<T> y;
-  
-  private Ndtri(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    y = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Neg.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Neg.java
deleted file mode 100644
index 90615ef3cbe..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Neg.java
+++ /dev/null
@@ -1,76 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.math;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Computes numerical negative value element-wise.
- * <p>
- * I.e., \\(y = -x\\).
- * 
- * @param <T> data type for {@code y()} output
- */
-@Operator(group = "math")
-public final class Neg<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new Neg operation.
-   * 
-   * @param scope current scope
-   * @param x 
-   * @return a new instance of Neg
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> Neg<T> create(Scope scope, Operand<T> x) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Neg", scope.makeOpName("Neg"));
-    opBuilder.addInput(x.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new Neg<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> y() {
-    return y;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return y;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Neg";
-  
-  private Output<T> y;
-  
-  private Neg(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    y = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/NextAfter.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/NextAfter.java
deleted file mode 100644
index 11bb2c1699a..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/NextAfter.java
+++ /dev/null
@@ -1,84 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.math;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Returns the next representable value of `x1` in the direction of `x2`, element-wise.
- * <p>
- * This operation returns the same result as the C++ std::nextafter function.
- * <p>
- * It can also return a subnormal number.
- * <p>
- * @compatibility(cpp)
- * Equivalent to C++ std::nextafter function.
- * @end_compatibility
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator(group = "math")
-public final class NextAfter<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new NextAfter operation.
-   * 
-   * @param scope current scope
-   * @param x1 
-   * @param x2 
-   * @return a new instance of NextAfter
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> NextAfter<T> create(Scope scope, Operand<T> x1, Operand<T> x2) {
-    OperationBuilder opBuilder = scope.env().opBuilder("NextAfter", scope.makeOpName("NextAfter"));
-    opBuilder.addInput(x1.asOutput(scope));
-    opBuilder.addInput(x2.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new NextAfter<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "NextAfter";
-  
-  private Output<T> output;
-  
-  private NextAfter(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/NotEqual.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/NotEqual.java
deleted file mode 100644
index 0ca58afe9b9..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/NotEqual.java
+++ /dev/null
@@ -1,112 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.math;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TBool;
-import org.tensorflow.types.family.TType;
-
-/**
- * Returns the truth value of (x != y) element-wise.
- * <p>
- * <i>NOTE</i>: `math.NotEqual` supports broadcasting. More about broadcasting
- * [here](http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html)
- */
-@Operator(group = "math")
-public final class NotEqual extends RawOp implements Operand<TBool> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.math.NotEqual}
-   */
-  public static class Options {
-    
-    /**
-     * @param incompatibleShapeError 
-     */
-    public Options incompatibleShapeError(Boolean incompatibleShapeError) {
-      this.incompatibleShapeError = incompatibleShapeError;
-      return this;
-    }
-    
-    private Boolean incompatibleShapeError;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new NotEqual operation.
-   * 
-   * @param scope current scope
-   * @param x 
-   * @param y 
-   * @param options carries optional attributes values
-   * @return a new instance of NotEqual
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> NotEqual create(Scope scope, Operand<T> x, Operand<T> y, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("NotEqual", scope.makeOpName("NotEqual"));
-    opBuilder.addInput(x.asOutput(scope));
-    opBuilder.addInput(y.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.incompatibleShapeError != null) {
-          opBuilder.setAttr("incompatible_shape_error", opts.incompatibleShapeError);
-        }
-      }
-    }
-    return new NotEqual(opBuilder.build());
-  }
-  
-  /**
-   * @param incompatibleShapeError 
-   */
-  public static Options incompatibleShapeError(Boolean incompatibleShapeError) {
-    return new Options().incompatibleShapeError(incompatibleShapeError);
-  }
-  
-  /**
-   */
-  public Output<TBool> z() {
-    return z;
-  }
-  
-  @Override
-  public Output<TBool> asOutput(Scope scope) {
-    return z;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "NotEqual";
-  
-  private Output<TBool> z;
-  
-  private NotEqual(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    z = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Polygamma.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Polygamma.java
deleted file mode 100644
index 792f45e251b..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Polygamma.java
+++ /dev/null
@@ -1,83 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.math;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Compute the polygamma function \\(\psi^{(n)}(x)\\).
- * <p>
- * The polygamma function is defined as:
- * <p>
- * \\(\psi^{(a)}(x) = \frac{d^a}{dx^a} \psi(x)\\)
- * <p>
- * where \\(\psi(x)\\) is the digamma function.
- * The polygamma function is defined only for non-negative integer orders \\a\\.
- * 
- * @param <T> data type for {@code z()} output
- */
-@Operator(group = "math")
-public final class Polygamma<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new Polygamma operation.
-   * 
-   * @param scope current scope
-   * @param a 
-   * @param x 
-   * @return a new instance of Polygamma
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> Polygamma<T> create(Scope scope, Operand<T> a, Operand<T> x) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Polygamma", scope.makeOpName("Polygamma"));
-    opBuilder.addInput(a.asOutput(scope));
-    opBuilder.addInput(x.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new Polygamma<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> z() {
-    return z;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return z;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Polygamma";
-  
-  private Output<T> z;
-  
-  private Polygamma(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    z = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/PopulationCount.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/PopulationCount.java
deleted file mode 100644
index aa56a025b24..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/PopulationCount.java
+++ /dev/null
@@ -1,80 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.math;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TUint8;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Computes element-wise population count (a.k.a. popcount, bitsum, bitcount).
- * <p>
- * For each entry in `x`, calculates the number of `1` (on) bits in the binary
- * representation of that entry.
- * <p>
- * <b>NOTE</b>: It is more efficient to first `tf.bitcast` your tensors into
- * `int32` or `int64` and perform the bitcount on the result, than to feed in
- * 8- or 16-bit inputs and then aggregate the resulting counts.
- */
-@Operator(group = "math")
-public final class PopulationCount extends RawOp implements Operand<TUint8> {
-  
-  /**
-   * Factory method to create a class wrapping a new PopulationCount operation.
-   * 
-   * @param scope current scope
-   * @param x 
-   * @return a new instance of PopulationCount
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> PopulationCount create(Scope scope, Operand<T> x) {
-    OperationBuilder opBuilder = scope.env().opBuilder("PopulationCount", scope.makeOpName("PopulationCount"));
-    opBuilder.addInput(x.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new PopulationCount(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<TUint8> y() {
-    return y;
-  }
-  
-  @Override
-  public Output<TUint8> asOutput(Scope scope) {
-    return y;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "PopulationCount";
-  
-  private Output<TUint8> y;
-  
-  private PopulationCount(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    y = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Pow.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Pow.java
deleted file mode 100644
index d632c6b6bb4..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Pow.java
+++ /dev/null
@@ -1,85 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.math;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Computes the power of one value to another.
- * <p>
- * Given a tensor `x` and a tensor `y`, this operation computes \\(x^y\\) for
- * corresponding elements in `x` and `y`. For example:
- * <pre>{@code
- * # tensor 'x' is [[2, 2]], [3, 3]]
- * # tensor 'y' is [[8, 16], [2, 3]]
- * tf.pow(x, y) ==> [[256, 65536], [9, 27]]
- * }</pre>
- * 
- * 
- * @param <T> data type for {@code z()} output
- */
-@Operator(group = "math")
-public final class Pow<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new Pow operation.
-   * 
-   * @param scope current scope
-   * @param x 
-   * @param y 
-   * @return a new instance of Pow
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> Pow<T> create(Scope scope, Operand<T> x, Operand<T> y) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Pow", scope.makeOpName("Pow"));
-    opBuilder.addInput(x.asOutput(scope));
-    opBuilder.addInput(y.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new Pow<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> z() {
-    return z;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return z;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Pow";
-  
-  private Output<T> z;
-  
-  private Pow(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    z = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/QuantizedAdd.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/QuantizedAdd.java
deleted file mode 100644
index cc36c2bfdd6..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/QuantizedAdd.java
+++ /dev/null
@@ -1,103 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.math;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.family.TType;
-
-/**
- * Returns x + y element-wise, working on quantized buffers.
- * 
- * @param <V> data type for {@code z()} output
- */
-@Operator(group = "math")
-public final class QuantizedAdd<V extends TType> extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new QuantizedAdd operation.
-   * 
-   * @param scope current scope
-   * @param x 
-   * @param y 
-   * @param minX The float value that the lowest quantized `x` value represents.
-   * @param maxX The float value that the highest quantized `x` value represents.
-   * @param minY The float value that the lowest quantized `y` value represents.
-   * @param maxY The float value that the highest quantized `y` value represents.
-   * @param Toutput 
-   * @return a new instance of QuantizedAdd
-   */
-  @Endpoint(describeByClass = true)
-  public static <V extends TType, T extends TType, U extends TType> QuantizedAdd<V> create(Scope scope, Operand<T> x, Operand<U> y, Operand<TFloat32> minX, Operand<TFloat32> maxX, Operand<TFloat32> minY, Operand<TFloat32> maxY, Class<V> Toutput) {
-    OperationBuilder opBuilder = scope.env().opBuilder("QuantizedAdd", scope.makeOpName("QuantizedAdd"));
-    opBuilder.addInput(x.asOutput(scope));
-    opBuilder.addInput(y.asOutput(scope));
-    opBuilder.addInput(minX.asOutput(scope));
-    opBuilder.addInput(maxX.asOutput(scope));
-    opBuilder.addInput(minY.asOutput(scope));
-    opBuilder.addInput(maxY.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("Toutput", Toutput);
-    return new QuantizedAdd<V>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<V> z() {
-    return z;
-  }
-  
-  /**
-   * The float value that the lowest quantized output value represents.
-   */
-  public Output<TFloat32> minZ() {
-    return minZ;
-  }
-  
-  /**
-   * The float value that the highest quantized output value represents.
-   * <p>
-   * <i>NOTE</i>: `math.QuantizedAdd` supports limited forms of broadcasting. More about
-   * broadcasting [here](http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html)
-   */
-  public Output<TFloat32> maxZ() {
-    return maxZ;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "QuantizedAdd";
-  
-  private Output<V> z;
-  private Output<TFloat32> minZ;
-  private Output<TFloat32> maxZ;
-  
-  private QuantizedAdd(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    z = operation.output(outputIdx++);
-    minZ = operation.output(outputIdx++);
-    maxZ = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/QuantizedMul.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/QuantizedMul.java
deleted file mode 100644
index 907e9721316..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/QuantizedMul.java
+++ /dev/null
@@ -1,103 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.math;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.family.TType;
-
-/**
- * Returns x * y element-wise, working on quantized buffers.
- * 
- * @param <V> data type for {@code z()} output
- */
-@Operator(group = "math")
-public final class QuantizedMul<V extends TType> extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new QuantizedMul operation.
-   * 
-   * @param scope current scope
-   * @param x 
-   * @param y 
-   * @param minX The float value that the lowest quantized `x` value represents.
-   * @param maxX The float value that the highest quantized `x` value represents.
-   * @param minY The float value that the lowest quantized `y` value represents.
-   * @param maxY The float value that the highest quantized `y` value represents.
-   * @param Toutput 
-   * @return a new instance of QuantizedMul
-   */
-  @Endpoint(describeByClass = true)
-  public static <V extends TType, T extends TType, U extends TType> QuantizedMul<V> create(Scope scope, Operand<T> x, Operand<U> y, Operand<TFloat32> minX, Operand<TFloat32> maxX, Operand<TFloat32> minY, Operand<TFloat32> maxY, Class<V> Toutput) {
-    OperationBuilder opBuilder = scope.env().opBuilder("QuantizedMul", scope.makeOpName("QuantizedMul"));
-    opBuilder.addInput(x.asOutput(scope));
-    opBuilder.addInput(y.asOutput(scope));
-    opBuilder.addInput(minX.asOutput(scope));
-    opBuilder.addInput(maxX.asOutput(scope));
-    opBuilder.addInput(minY.asOutput(scope));
-    opBuilder.addInput(maxY.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("Toutput", Toutput);
-    return new QuantizedMul<V>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<V> z() {
-    return z;
-  }
-  
-  /**
-   * The float value that the lowest quantized output value represents.
-   */
-  public Output<TFloat32> minZ() {
-    return minZ;
-  }
-  
-  /**
-   * The float value that the highest quantized output value represents.
-   * <p>
-   * <i>NOTE</i>: `math.QuantizedMul` supports limited forms of broadcasting. More about
-   * broadcasting [here](http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html)
-   */
-  public Output<TFloat32> maxZ() {
-    return maxZ;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "QuantizedMul";
-  
-  private Output<V> z;
-  private Output<TFloat32> minZ;
-  private Output<TFloat32> maxZ;
-  
-  private QuantizedMul(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    z = operation.output(outputIdx++);
-    minZ = operation.output(outputIdx++);
-    maxZ = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Real.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Real.java
deleted file mode 100644
index 72004717894..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Real.java
+++ /dev/null
@@ -1,102 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.math;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * Returns the real part of a complex number.
- * <p>
- * Given a tensor `input` of complex numbers, this operation returns a tensor of
- * type `float` that is the real part of each element in `input`. All elements in
- * `input` must be complex numbers of the form \\(a + bj\\), where <i>a</i> is the real
- *  part returned by this operation and <i>b</i> is the imaginary part.
- * <p>
- * For example:
- * <pre>{@code
- * # tensor 'input' is [-2.25 + 4.75j, 3.25 + 5.75j]
- * tf.real(input) ==> [-2.25, 3.25]
- * }</pre>
- * 
- * 
- * @param <U> data type for {@code output()} output
- */
-@Operator(group = "math")
-public final class Real<U extends TNumber> extends RawOp implements Operand<U> {
-  
-  /**
-   * Factory method to create a class wrapping a new Real operation.
-   * 
-   * @param scope current scope
-   * @param input 
-   * @param Tout 
-   * @return a new instance of Real
-   */
-  @Endpoint(describeByClass = true)
-  public static <U extends TNumber, T extends TType> Real<U> create(Scope scope, Operand<T> input, Class<U> Tout) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Real", scope.makeOpName("Real"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("Tout", Tout);
-    return new Real<U>(opBuilder.build());
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new Real operation using default output types.
-   * 
-   * @param scope current scope
-   * @param input 
-   * @return a new instance of Real
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> Real<TFloat32> create(Scope scope, Operand<T> input) {
-    return create(scope, input, TFloat32.class);
-  }
-  
-  /**
-   */
-  public Output<U> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<U> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Real";
-  
-  private Output<U> output;
-  
-  private Real(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/RealDiv.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/RealDiv.java
deleted file mode 100644
index 472cbb80e1f..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/RealDiv.java
+++ /dev/null
@@ -1,81 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.math;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Returns x / y element-wise for real types.
- * <p>
- * If `x` and `y` are reals, this will return the floating-point division.
- * <p>
- * <i>NOTE</i>: `Div` supports broadcasting. More about broadcasting
- * [here](http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html)
- * 
- * @param <T> data type for {@code z()} output
- */
-@Operator(group = "math")
-public final class RealDiv<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new RealDiv operation.
-   * 
-   * @param scope current scope
-   * @param x 
-   * @param y 
-   * @return a new instance of RealDiv
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> RealDiv<T> create(Scope scope, Operand<T> x, Operand<T> y) {
-    OperationBuilder opBuilder = scope.env().opBuilder("RealDiv", scope.makeOpName("RealDiv"));
-    opBuilder.addInput(x.asOutput(scope));
-    opBuilder.addInput(y.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new RealDiv<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> z() {
-    return z;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return z;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "RealDiv";
-  
-  private Output<T> z;
-  
-  private RealDiv(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    z = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Reciprocal.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Reciprocal.java
deleted file mode 100644
index 187fa436621..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Reciprocal.java
+++ /dev/null
@@ -1,76 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.math;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Computes the reciprocal of x element-wise.
- * <p>
- * I.e., \\(y = 1 / x\\).
- * 
- * @param <T> data type for {@code y()} output
- */
-@Operator(group = "math")
-public final class Reciprocal<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new Reciprocal operation.
-   * 
-   * @param scope current scope
-   * @param x 
-   * @return a new instance of Reciprocal
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> Reciprocal<T> create(Scope scope, Operand<T> x) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Reciprocal", scope.makeOpName("Reciprocal"));
-    opBuilder.addInput(x.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new Reciprocal<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> y() {
-    return y;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return y;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Reciprocal";
-  
-  private Output<T> y;
-  
-  private Reciprocal(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    y = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/ReciprocalGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/ReciprocalGrad.java
deleted file mode 100644
index 8ee6127d7cc..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/ReciprocalGrad.java
+++ /dev/null
@@ -1,78 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.math;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Computes the gradient for the inverse of `x` wrt its input.
- * <p>
- * Specifically, `grad = -dy <i> y</i>y`, where `y = 1/x`, and `dy`
- * is the corresponding input gradient.
- * 
- * @param <T> data type for {@code z()} output
- */
-public final class ReciprocalGrad<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new ReciprocalGrad operation.
-   * 
-   * @param scope current scope
-   * @param y 
-   * @param dy 
-   * @return a new instance of ReciprocalGrad
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> ReciprocalGrad<T> create(Scope scope, Operand<T> y, Operand<T> dy) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ReciprocalGrad", scope.makeOpName("ReciprocalGrad"));
-    opBuilder.addInput(y.asOutput(scope));
-    opBuilder.addInput(dy.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new ReciprocalGrad<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> z() {
-    return z;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return z;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ReciprocalGrad";
-  
-  private Output<T> z;
-  
-  private ReciprocalGrad(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    z = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/RequantizationRangePerChannel.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/RequantizationRangePerChannel.java
deleted file mode 100644
index 5dfaaa2ceab..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/RequantizationRangePerChannel.java
+++ /dev/null
@@ -1,84 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.math;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.family.TType;
-
-/**
- * Computes requantization range per channel.
- */
-public final class RequantizationRangePerChannel extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new RequantizationRangePerChannel operation.
-   * 
-   * @param scope current scope
-   * @param input The original input tensor.
-   * @param inputMin The minimum value of the input tensor
-   * @param inputMax The maximum value of the input tensor.
-   * @param clipValueMax The maximum value of the output that needs to be clipped.
-   * Example: set this to 6 for Relu6.
-   * @return a new instance of RequantizationRangePerChannel
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> RequantizationRangePerChannel create(Scope scope, Operand<T> input, Operand<TFloat32> inputMin, Operand<TFloat32> inputMax, Float clipValueMax) {
-    OperationBuilder opBuilder = scope.env().opBuilder("RequantizationRangePerChannel", scope.makeOpName("RequantizationRangePerChannel"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder.addInput(inputMin.asOutput(scope));
-    opBuilder.addInput(inputMax.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("clip_value_max", clipValueMax);
-    return new RequantizationRangePerChannel(opBuilder.build());
-  }
-  
-  /**
-   * The minimum value of the final output tensor
-   */
-  public Output<TFloat32> outputMin() {
-    return outputMin;
-  }
-  
-  /**
-   * The maximum value of the final output tensor.
-   */
-  public Output<TFloat32> outputMax() {
-    return outputMax;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "RequantizationRangePerChannel";
-  
-  private Output<TFloat32> outputMin;
-  private Output<TFloat32> outputMax;
-  
-  private RequantizationRangePerChannel(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    outputMin = operation.output(outputIdx++);
-    outputMax = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/RequantizePerChannel.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/RequantizePerChannel.java
deleted file mode 100644
index 81b5c105c5d..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/RequantizePerChannel.java
+++ /dev/null
@@ -1,98 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.math;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.family.TType;
-
-/**
- * Requantizes input with min and max values known per channel.
- * 
- * @param <U> data type for {@code output()} output
- */
-public final class RequantizePerChannel<U extends TType> extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new RequantizePerChannel operation.
-   * 
-   * @param scope current scope
-   * @param input The original input tensor.
-   * @param inputMin The minimum value of the input tensor
-   * @param inputMax The maximum value of the input tensor.
-   * @param requestedOutputMin The minimum value of the output tensor requested.
-   * @param requestedOutputMax The maximum value of the output tensor requested.
-   * @param outType The quantized type of output tensor that needs to be converted.
-   * @return a new instance of RequantizePerChannel
-   */
-  @Endpoint(describeByClass = true)
-  public static <U extends TType, T extends TType> RequantizePerChannel<U> create(Scope scope, Operand<T> input, Operand<TFloat32> inputMin, Operand<TFloat32> inputMax, Operand<TFloat32> requestedOutputMin, Operand<TFloat32> requestedOutputMax, Class<U> outType) {
-    OperationBuilder opBuilder = scope.env().opBuilder("RequantizePerChannel", scope.makeOpName("RequantizePerChannel"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder.addInput(inputMin.asOutput(scope));
-    opBuilder.addInput(inputMax.asOutput(scope));
-    opBuilder.addInput(requestedOutputMin.asOutput(scope));
-    opBuilder.addInput(requestedOutputMax.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("out_type", outType);
-    return new RequantizePerChannel<U>(opBuilder.build());
-  }
-  
-  /**
-   * Output tensor.
-   */
-  public Output<U> output() {
-    return output;
-  }
-  
-  /**
-   * The minimum value of the final output tensor
-   */
-  public Output<TFloat32> outputMin() {
-    return outputMin;
-  }
-  
-  /**
-   * The maximum value of the final output tensor.
-   */
-  public Output<TFloat32> outputMax() {
-    return outputMax;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "RequantizePerChannel";
-  
-  private Output<U> output;
-  private Output<TFloat32> outputMin;
-  private Output<TFloat32> outputMax;
-  
-  private RequantizePerChannel(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-    outputMin = operation.output(outputIdx++);
-    outputMax = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Rint.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Rint.java
deleted file mode 100644
index 5ebd67d43ae..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Rint.java
+++ /dev/null
@@ -1,84 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.math;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Returns element-wise integer closest to x.
- * <p>
- * If the result is midway between two representable values,
- * the even representable is chosen.
- * For example:
- * <pre>{@code
- * rint(-1.5) ==> -2.0
- * rint(0.5000001) ==> 1.0
- * rint([-1.7, -1.5, -0.2, 0.2, 1.5, 1.7, 2.0]) ==> [-2., -2., -0., 0., 2., 2., 2.]
- * }</pre>
- * 
- * 
- * @param <T> data type for {@code y()} output
- */
-@Operator(group = "math")
-public final class Rint<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new Rint operation.
-   * 
-   * @param scope current scope
-   * @param x 
-   * @return a new instance of Rint
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> Rint<T> create(Scope scope, Operand<T> x) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Rint", scope.makeOpName("Rint"));
-    opBuilder.addInput(x.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new Rint<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> y() {
-    return y;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return y;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Rint";
-  
-  private Output<T> y;
-  
-  private Rint(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    y = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Round.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Round.java
deleted file mode 100644
index ea8e3cfb878..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Round.java
+++ /dev/null
@@ -1,77 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.math;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Rounds the values of a tensor to the nearest integer, element-wise.
- * <p>
- * Rounds half to even.  Also known as bankers rounding. If you want to round
- * according to the current system rounding mode use std::cint.
- * 
- * @param <T> data type for {@code y()} output
- */
-@Operator(group = "math")
-public final class Round<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new Round operation.
-   * 
-   * @param scope current scope
-   * @param x 
-   * @return a new instance of Round
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> Round<T> create(Scope scope, Operand<T> x) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Round", scope.makeOpName("Round"));
-    opBuilder.addInput(x.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new Round<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> y() {
-    return y;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return y;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Round";
-  
-  private Output<T> y;
-  
-  private Round(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    y = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Rsqrt.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Rsqrt.java
deleted file mode 100644
index 198e2ac0f9e..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Rsqrt.java
+++ /dev/null
@@ -1,76 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.math;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Computes reciprocal of square root of x element-wise.
- * <p>
- * I.e., \\(y = 1 / \sqrt{x}\\).
- * 
- * @param <T> data type for {@code y()} output
- */
-@Operator(group = "math")
-public final class Rsqrt<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new Rsqrt operation.
-   * 
-   * @param scope current scope
-   * @param x 
-   * @return a new instance of Rsqrt
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> Rsqrt<T> create(Scope scope, Operand<T> x) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Rsqrt", scope.makeOpName("Rsqrt"));
-    opBuilder.addInput(x.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new Rsqrt<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> y() {
-    return y;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return y;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Rsqrt";
-  
-  private Output<T> y;
-  
-  private Rsqrt(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    y = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/RsqrtGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/RsqrtGrad.java
deleted file mode 100644
index c7fd2ca1502..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/RsqrtGrad.java
+++ /dev/null
@@ -1,78 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.math;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Computes the gradient for the rsqrt of `x` wrt its input.
- * <p>
- * Specifically, `grad = dy * -0.5 * y^3`, where `y = rsqrt(x)`, and `dy`
- * is the corresponding input gradient.
- * 
- * @param <T> data type for {@code z()} output
- */
-public final class RsqrtGrad<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new RsqrtGrad operation.
-   * 
-   * @param scope current scope
-   * @param y 
-   * @param dy 
-   * @return a new instance of RsqrtGrad
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> RsqrtGrad<T> create(Scope scope, Operand<T> y, Operand<T> dy) {
-    OperationBuilder opBuilder = scope.env().opBuilder("RsqrtGrad", scope.makeOpName("RsqrtGrad"));
-    opBuilder.addInput(y.asOutput(scope));
-    opBuilder.addInput(dy.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new RsqrtGrad<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> z() {
-    return z;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return z;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "RsqrtGrad";
-  
-  private Output<T> z;
-  
-  private RsqrtGrad(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    z = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SegmentMax.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SegmentMax.java
deleted file mode 100644
index ff02ffa3bbc..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SegmentMax.java
+++ /dev/null
@@ -1,102 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.math;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Computes the maximum along segments of a tensor.
- * <p>
- * Read
- * [the section on segmentation](https://tensorflow.org/api_docs/python/tf/math#Segmentation)
- * for an explanation of segments.
- * <p>
- * Computes a tensor such that
- * \\(output_i = \max_j(data_j)\\) where `max` is over `j` such
- * that `segment_ids[j] == i`.
- * <p>
- * If the max is empty for a given segment ID `i`, `output[i] = 0`.
- * <p>
- * <div style="width:70%; margin:auto; margin-bottom:10px; margin-top:20px;">
- * <img style="width:100%" src="https://www.tensorflow.org/images/SegmentMax.png" alt>
- * </div>
- * <p>
- * For example:
- * <pre>{@code
- * c = tf.constant([[1,2,3,4], [4, 3, 2, 1], [5,6,7,8]])
- * tf.segment_max(c, tf.constant([0, 0, 1]))
- * # ==> [[4, 3, 3, 4],
- * #      [5, 6, 7, 8]]
- * }</pre>
- * 
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator(group = "math")
-public final class SegmentMax<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new SegmentMax operation.
-   * 
-   * @param scope current scope
-   * @param data 
-   * @param segmentIds A 1-D tensor whose size is equal to the size of `data`'s
-   * first dimension.  Values should be sorted and can be repeated.
-   * @return a new instance of SegmentMax
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber, U extends TNumber> SegmentMax<T> create(Scope scope, Operand<T> data, Operand<U> segmentIds) {
-    OperationBuilder opBuilder = scope.env().opBuilder("SegmentMax", scope.makeOpName("SegmentMax"));
-    opBuilder.addInput(data.asOutput(scope));
-    opBuilder.addInput(segmentIds.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new SegmentMax<T>(opBuilder.build());
-  }
-  
-  /**
-   * Has same shape as data, except for dimension 0 which
-   * has size `k`, the number of segments.
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "SegmentMax";
-  
-  private Output<T> output;
-  
-  private SegmentMax(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SegmentMean.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SegmentMean.java
deleted file mode 100644
index cdd1acc7fa3..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SegmentMean.java
+++ /dev/null
@@ -1,104 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.math;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * Computes the mean along segments of a tensor.
- * <p>
- * Read
- * [the section on segmentation](https://tensorflow.org/api_docs/python/tf/math#Segmentation)
- * for an explanation of segments.
- * <p>
- * Computes a tensor such that
- * \\(output_i = \frac{\sum_j data_j}{N}\\) where `mean` is
- * over `j` such that `segment_ids[j] == i` and `N` is the total number of
- * values summed.
- * <p>
- * If the mean is empty for a given segment ID `i`, `output[i] = 0`.
- * <p>
- * <div style="width:70%; margin:auto; margin-bottom:10px; margin-top:20px;">
- * <img style="width:100%" src="https://www.tensorflow.org/images/SegmentMean.png" alt>
- * </div>
- * <p>
- * For example:
- * <pre>{@code
- * c = tf.constant([[1.0,2,3,4], [4, 3, 2, 1], [5,6,7,8]])
- * tf.segment_mean(c, tf.constant([0, 0, 1]))
- * # ==> [[2.5, 2.5, 2.5, 2.5],
- * #      [5, 6, 7, 8]]
- * }</pre>
- * 
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator(group = "math")
-public final class SegmentMean<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new SegmentMean operation.
-   * 
-   * @param scope current scope
-   * @param data 
-   * @param segmentIds A 1-D tensor whose size is equal to the size of `data`'s
-   * first dimension.  Values should be sorted and can be repeated.
-   * @return a new instance of SegmentMean
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> SegmentMean<T> create(Scope scope, Operand<T> data, Operand<U> segmentIds) {
-    OperationBuilder opBuilder = scope.env().opBuilder("SegmentMean", scope.makeOpName("SegmentMean"));
-    opBuilder.addInput(data.asOutput(scope));
-    opBuilder.addInput(segmentIds.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new SegmentMean<T>(opBuilder.build());
-  }
-  
-  /**
-   * Has same shape as data, except for dimension 0 which
-   * has size `k`, the number of segments.
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "SegmentMean";
-  
-  private Output<T> output;
-  
-  private SegmentMean(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SegmentMin.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SegmentMin.java
deleted file mode 100644
index e40683c8aa3..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SegmentMin.java
+++ /dev/null
@@ -1,102 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.math;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Computes the minimum along segments of a tensor.
- * <p>
- * Read
- * [the section on segmentation](https://tensorflow.org/api_docs/python/tf/math#Segmentation)
- * for an explanation of segments.
- * <p>
- * Computes a tensor such that
- * \\(output_i = \min_j(data_j)\\) where `min` is over `j` such
- * that `segment_ids[j] == i`.
- * <p>
- * If the min is empty for a given segment ID `i`, `output[i] = 0`.
- * <p>
- * <div style="width:70%; margin:auto; margin-bottom:10px; margin-top:20px;">
- * <img style="width:100%" src="https://www.tensorflow.org/images/SegmentMin.png" alt>
- * </div>
- * <p>
- * For example:
- * <pre>{@code
- * c = tf.constant([[1,2,3,4], [4, 3, 2, 1], [5,6,7,8]])
- * tf.segment_min(c, tf.constant([0, 0, 1]))
- * # ==> [[1, 2, 2, 1],
- * #      [5, 6, 7, 8]]
- * }</pre>
- * 
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator(group = "math")
-public final class SegmentMin<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new SegmentMin operation.
-   * 
-   * @param scope current scope
-   * @param data 
-   * @param segmentIds A 1-D tensor whose size is equal to the size of `data`'s
-   * first dimension.  Values should be sorted and can be repeated.
-   * @return a new instance of SegmentMin
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber, U extends TNumber> SegmentMin<T> create(Scope scope, Operand<T> data, Operand<U> segmentIds) {
-    OperationBuilder opBuilder = scope.env().opBuilder("SegmentMin", scope.makeOpName("SegmentMin"));
-    opBuilder.addInput(data.asOutput(scope));
-    opBuilder.addInput(segmentIds.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new SegmentMin<T>(opBuilder.build());
-  }
-  
-  /**
-   * Has same shape as data, except for dimension 0 which
-   * has size `k`, the number of segments.
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "SegmentMin";
-  
-  private Output<T> output;
-  
-  private SegmentMin(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SegmentProd.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SegmentProd.java
deleted file mode 100644
index b5757ecb249..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SegmentProd.java
+++ /dev/null
@@ -1,103 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.math;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * Computes the product along segments of a tensor.
- * <p>
- * Read
- * [the section on segmentation](https://tensorflow.org/api_docs/python/tf/math#Segmentation)
- * for an explanation of segments.
- * <p>
- * Computes a tensor such that
- * \\(output_i = \prod_j data_j\\) where the product is over `j` such
- * that `segment_ids[j] == i`.
- * <p>
- * If the product is empty for a given segment ID `i`, `output[i] = 1`.
- * <p>
- * <div style="width:70%; margin:auto; margin-bottom:10px; margin-top:20px;">
- * <img style="width:100%" src="https://www.tensorflow.org/images/SegmentProd.png" alt>
- * </div>
- * <p>
- * For example:
- * <pre>{@code
- * c = tf.constant([[1,2,3,4], [4, 3, 2, 1], [5,6,7,8]])
- * tf.segment_prod(c, tf.constant([0, 0, 1]))
- * # ==> [[4, 6, 6, 4],
- * #      [5, 6, 7, 8]]
- * }</pre>
- * 
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator(group = "math")
-public final class SegmentProd<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new SegmentProd operation.
-   * 
-   * @param scope current scope
-   * @param data 
-   * @param segmentIds A 1-D tensor whose size is equal to the size of `data`'s
-   * first dimension.  Values should be sorted and can be repeated.
-   * @return a new instance of SegmentProd
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> SegmentProd<T> create(Scope scope, Operand<T> data, Operand<U> segmentIds) {
-    OperationBuilder opBuilder = scope.env().opBuilder("SegmentProd", scope.makeOpName("SegmentProd"));
-    opBuilder.addInput(data.asOutput(scope));
-    opBuilder.addInput(segmentIds.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new SegmentProd<T>(opBuilder.build());
-  }
-  
-  /**
-   * Has same shape as data, except for dimension 0 which
-   * has size `k`, the number of segments.
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "SegmentProd";
-  
-  private Output<T> output;
-  
-  private SegmentProd(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SegmentSum.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SegmentSum.java
deleted file mode 100644
index 1bda10f8878..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SegmentSum.java
+++ /dev/null
@@ -1,103 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.math;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * Computes the sum along segments of a tensor.
- * <p>
- * Read
- * [the section on segmentation](https://tensorflow.org/api_docs/python/tf/math#Segmentation)
- * for an explanation of segments.
- * <p>
- * Computes a tensor such that
- * \\(output_i = \sum_j data_j\\) where sum is over `j` such
- * that `segment_ids[j] == i`.
- * <p>
- * If the sum is empty for a given segment ID `i`, `output[i] = 0`.
- * <p>
- * <div style="width:70%; margin:auto; margin-bottom:10px; margin-top:20px;">
- * <img style="width:100%" src="https://www.tensorflow.org/images/SegmentSum.png" alt>
- * </div>
- * <p>
- * For example:
- * <pre>{@code
- * c = tf.constant([[1,2,3,4], [4, 3, 2, 1], [5,6,7,8]])
- * tf.segment_sum(c, tf.constant([0, 0, 1]))
- * # ==> [[5, 5, 5, 5],
- * #      [5, 6, 7, 8]]
- * }</pre>
- * 
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator(group = "math")
-public final class SegmentSum<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new SegmentSum operation.
-   * 
-   * @param scope current scope
-   * @param data 
-   * @param segmentIds A 1-D tensor whose size is equal to the size of `data`'s
-   * first dimension.  Values should be sorted and can be repeated.
-   * @return a new instance of SegmentSum
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> SegmentSum<T> create(Scope scope, Operand<T> data, Operand<U> segmentIds) {
-    OperationBuilder opBuilder = scope.env().opBuilder("SegmentSum", scope.makeOpName("SegmentSum"));
-    opBuilder.addInput(data.asOutput(scope));
-    opBuilder.addInput(segmentIds.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new SegmentSum<T>(opBuilder.build());
-  }
-  
-  /**
-   * Has same shape as data, except for dimension 0 which
-   * has size `k`, the number of segments.
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "SegmentSum";
-  
-  private Output<T> output;
-  
-  private SegmentSum(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Sigmoid.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Sigmoid.java
deleted file mode 100644
index f8a06245fe5..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Sigmoid.java
+++ /dev/null
@@ -1,76 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.math;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Computes sigmoid of `x` element-wise.
- * <p>
- * Specifically, `y = 1 / (1 + exp(-x))`.
- * 
- * @param <T> data type for {@code y()} output
- */
-@Operator(group = "math")
-public final class Sigmoid<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new Sigmoid operation.
-   * 
-   * @param scope current scope
-   * @param x 
-   * @return a new instance of Sigmoid
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> Sigmoid<T> create(Scope scope, Operand<T> x) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Sigmoid", scope.makeOpName("Sigmoid"));
-    opBuilder.addInput(x.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new Sigmoid<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> y() {
-    return y;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return y;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Sigmoid";
-  
-  private Output<T> y;
-  
-  private Sigmoid(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    y = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SigmoidGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SigmoidGrad.java
deleted file mode 100644
index ea0613e0f1c..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SigmoidGrad.java
+++ /dev/null
@@ -1,78 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.math;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Computes the gradient of the sigmoid of `x` wrt its input.
- * <p>
- * Specifically, `grad = dy * y * (1 - y)`, where `y = sigmoid(x)`, and
- * `dy` is the corresponding input gradient.
- * 
- * @param <T> data type for {@code z()} output
- */
-public final class SigmoidGrad<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new SigmoidGrad operation.
-   * 
-   * @param scope current scope
-   * @param y 
-   * @param dy 
-   * @return a new instance of SigmoidGrad
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> SigmoidGrad<T> create(Scope scope, Operand<T> y, Operand<T> dy) {
-    OperationBuilder opBuilder = scope.env().opBuilder("SigmoidGrad", scope.makeOpName("SigmoidGrad"));
-    opBuilder.addInput(y.asOutput(scope));
-    opBuilder.addInput(dy.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new SigmoidGrad<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> z() {
-    return z;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return z;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "SigmoidGrad";
-  
-  private Output<T> z;
-  
-  private SigmoidGrad(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    z = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Sign.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Sign.java
deleted file mode 100644
index 7144fe432a9..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Sign.java
+++ /dev/null
@@ -1,82 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.math;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Returns an element-wise indication of the sign of a number.
- * <p>
- * `y = sign(x) = -1` if `x < 0`; 0 if `x == 0`; 1 if `x > 0`.
- * <p>
- * For complex numbers, `y = sign(x) = x / |x|` if `x != 0`, otherwise `y = 0`.
- * <p>
- * Example usage:
- * >>> tf.math.sign([0., 2., -3.])
- * <tf.Tensor: shape=(3,), dtype=float32, numpy=array([ 0.,  1., -1.], dtype=float32)>
- * 
- * @param <T> data type for {@code y()} output
- */
-@Operator(group = "math")
-public final class Sign<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new Sign operation.
-   * 
-   * @param scope current scope
-   * @param x 
-   * @return a new instance of Sign
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> Sign<T> create(Scope scope, Operand<T> x) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Sign", scope.makeOpName("Sign"));
-    opBuilder.addInput(x.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new Sign<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> y() {
-    return y;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return y;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Sign";
-  
-  private Output<T> y;
-  
-  private Sign(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    y = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Sin.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Sin.java
deleted file mode 100644
index 1a60f3d1df0..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Sin.java
+++ /dev/null
@@ -1,84 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.math;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Computes sine of x element-wise.
- * <p>
- *   Given an input tensor, this function computes sine of every
- *   element in the tensor. Input range is `(-inf, inf)` and
- *   output range is `[-1,1]`.
- * <p>
- *   <pre>{@code
- *   x = tf.constant([-float("inf"), -9, -0.5, 1, 1.2, 200, 10, float("inf")])
- *   tf.math.sin(x) ==> [nan -0.4121185 -0.47942555 0.84147096 0.9320391 -0.87329733 -0.54402107 nan]
- *   }</pre>
- * 
- * 
- * @param <T> data type for {@code y()} output
- */
-@Operator(group = "math")
-public final class Sin<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new Sin operation.
-   * 
-   * @param scope current scope
-   * @param x 
-   * @return a new instance of Sin
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> Sin<T> create(Scope scope, Operand<T> x) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Sin", scope.makeOpName("Sin"));
-    opBuilder.addInput(x.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new Sin<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> y() {
-    return y;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return y;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Sin";
-  
-  private Output<T> y;
-  
-  private Sin(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    y = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Sinh.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Sinh.java
deleted file mode 100644
index f82339505f3..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Sinh.java
+++ /dev/null
@@ -1,84 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.math;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Computes hyperbolic sine of x element-wise.
- * <p>
- *   Given an input tensor, this function computes hyperbolic sine of every
- *   element in the tensor. Input range is `[-inf,inf]` and output range
- *   is `[-inf,inf]`.
- * <p>
- *   <pre>{@code
- *   x = tf.constant([-float("inf"), -9, -0.5, 1, 1.2, 2, 10, float("inf")])
- *   tf.math.sinh(x) ==> [-inf -4.0515420e+03 -5.2109528e-01 1.1752012e+00 1.5094614e+00 3.6268604e+00 1.1013232e+04 inf]
- *   }</pre>
- * 
- * 
- * @param <T> data type for {@code y()} output
- */
-@Operator(group = "math")
-public final class Sinh<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new Sinh operation.
-   * 
-   * @param scope current scope
-   * @param x 
-   * @return a new instance of Sinh
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> Sinh<T> create(Scope scope, Operand<T> x) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Sinh", scope.makeOpName("Sinh"));
-    opBuilder.addInput(x.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new Sinh<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> y() {
-    return y;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return y;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Sinh";
-  
-  private Output<T> y;
-  
-  private Sinh(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    y = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SobolSample.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SobolSample.java
deleted file mode 100644
index f67da4dedf9..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SobolSample.java
+++ /dev/null
@@ -1,103 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.math;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Generates points from the Sobol sequence.
- * <p>
- * Creates a Sobol sequence with `num_results` samples. Each sample has dimension
- * `dim`. Skips the first `skip` samples.
- * 
- * @param <T> data type for {@code samples()} output
- */
-public final class SobolSample<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new SobolSample operation.
-   * 
-   * @param scope current scope
-   * @param dim Positive scalar `Tensor` representing each sample's dimension.
-   * @param numResults Positive scalar `Tensor` of dtype int32. The number of Sobol points to return
-   * in the output.
-   * @param skip Positive scalar `Tensor` of dtype int32. The number of initial points of the
-   * Sobol sequence to skip.
-   * @param dtype The type of the sample. One of: `float32` or `float64`.
-   * @return a new instance of SobolSample
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> SobolSample<T> create(Scope scope, Operand<TInt32> dim, Operand<TInt32> numResults, Operand<TInt32> skip, Class<T> dtype) {
-    OperationBuilder opBuilder = scope.env().opBuilder("SobolSample", scope.makeOpName("SobolSample"));
-    opBuilder.addInput(dim.asOutput(scope));
-    opBuilder.addInput(numResults.asOutput(scope));
-    opBuilder.addInput(skip.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("dtype", dtype);
-    return new SobolSample<T>(opBuilder.build());
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new SobolSample operation using default output types.
-   * 
-   * @param scope current scope
-   * @param dim Positive scalar `Tensor` representing each sample's dimension.
-   * @param numResults Positive scalar `Tensor` of dtype int32. The number of Sobol points to return
-   * in the output.
-   * @param skip Positive scalar `Tensor` of dtype int32. The number of initial points of the
-   * Sobol sequence to skip.
-   * @return a new instance of SobolSample
-   */
-  @Endpoint(describeByClass = true)
-  public static SobolSample<TFloat32> create(Scope scope, Operand<TInt32> dim, Operand<TInt32> numResults, Operand<TInt32> skip) {
-    return create(scope, dim, numResults, skip, TFloat32.class);
-  }
-  
-  /**
-   * `Tensor` of samples from Sobol sequence with `shape` [num_results, dim].
-   */
-  public Output<T> samples() {
-    return samples;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return samples;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "SobolSample";
-  
-  private Output<T> samples;
-  
-  private SobolSample(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    samples = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Softplus.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Softplus.java
deleted file mode 100644
index 1a07ecfb747..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Softplus.java
+++ /dev/null
@@ -1,74 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.math;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Computes softplus: `log(exp(features) + 1)`.
- * 
- * @param <T> data type for {@code activations()} output
- */
-@Operator(group = "math")
-public final class Softplus<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new Softplus operation.
-   * 
-   * @param scope current scope
-   * @param features 
-   * @return a new instance of Softplus
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> Softplus<T> create(Scope scope, Operand<T> features) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Softplus", scope.makeOpName("Softplus"));
-    opBuilder.addInput(features.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new Softplus<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> activations() {
-    return activations;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return activations;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Softplus";
-  
-  private Output<T> activations;
-  
-  private Softplus(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    activations = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SoftplusGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SoftplusGrad.java
deleted file mode 100644
index 9bc0aee229f..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SoftplusGrad.java
+++ /dev/null
@@ -1,76 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.math;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Computes softplus gradients for a softplus operation.
- * 
- * @param <T> data type for {@code backprops()} output
- */
-public final class SoftplusGrad<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new SoftplusGrad operation.
-   * 
-   * @param scope current scope
-   * @param gradients The backpropagated gradients to the corresponding softplus operation.
-   * @param features The features passed as input to the corresponding softplus operation.
-   * @return a new instance of SoftplusGrad
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> SoftplusGrad<T> create(Scope scope, Operand<T> gradients, Operand<T> features) {
-    OperationBuilder opBuilder = scope.env().opBuilder("SoftplusGrad", scope.makeOpName("SoftplusGrad"));
-    opBuilder.addInput(gradients.asOutput(scope));
-    opBuilder.addInput(features.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new SoftplusGrad<T>(opBuilder.build());
-  }
-  
-  /**
-   * The gradients: `gradients / (1 + exp(-features))`.
-   */
-  public Output<T> backprops() {
-    return backprops;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return backprops;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "SoftplusGrad";
-  
-  private Output<T> backprops;
-  
-  private SoftplusGrad(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    backprops = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Sqrt.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Sqrt.java
deleted file mode 100644
index dad9c1d6cec..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Sqrt.java
+++ /dev/null
@@ -1,76 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.math;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Computes square root of x element-wise.
- * <p>
- * I.e., \\(y = \sqrt{x} = x^{1/2}\\).
- * 
- * @param <T> data type for {@code y()} output
- */
-@Operator(group = "math")
-public final class Sqrt<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new Sqrt operation.
-   * 
-   * @param scope current scope
-   * @param x 
-   * @return a new instance of Sqrt
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> Sqrt<T> create(Scope scope, Operand<T> x) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Sqrt", scope.makeOpName("Sqrt"));
-    opBuilder.addInput(x.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new Sqrt<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> y() {
-    return y;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return y;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Sqrt";
-  
-  private Output<T> y;
-  
-  private Sqrt(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    y = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SqrtGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SqrtGrad.java
deleted file mode 100644
index 0206518124c..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SqrtGrad.java
+++ /dev/null
@@ -1,78 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.math;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Computes the gradient for the sqrt of `x` wrt its input.
- * <p>
- * Specifically, `grad = dy * 0.5 / y`, where `y = sqrt(x)`, and `dy`
- * is the corresponding input gradient.
- * 
- * @param <T> data type for {@code z()} output
- */
-public final class SqrtGrad<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new SqrtGrad operation.
-   * 
-   * @param scope current scope
-   * @param y 
-   * @param dy 
-   * @return a new instance of SqrtGrad
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> SqrtGrad<T> create(Scope scope, Operand<T> y, Operand<T> dy) {
-    OperationBuilder opBuilder = scope.env().opBuilder("SqrtGrad", scope.makeOpName("SqrtGrad"));
-    opBuilder.addInput(y.asOutput(scope));
-    opBuilder.addInput(dy.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new SqrtGrad<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> z() {
-    return z;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return z;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "SqrtGrad";
-  
-  private Output<T> z;
-  
-  private SqrtGrad(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    z = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Square.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Square.java
deleted file mode 100644
index 896cf528ca3..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Square.java
+++ /dev/null
@@ -1,76 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.math;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Computes square of x element-wise.
- * <p>
- * I.e., \\(y = x * x = x^2\\).
- * 
- * @param <T> data type for {@code y()} output
- */
-@Operator(group = "math")
-public final class Square<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new Square operation.
-   * 
-   * @param scope current scope
-   * @param x 
-   * @return a new instance of Square
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> Square<T> create(Scope scope, Operand<T> x) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Square", scope.makeOpName("Square"));
-    opBuilder.addInput(x.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new Square<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> y() {
-    return y;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return y;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Square";
-  
-  private Output<T> y;
-  
-  private Square(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    y = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SquaredDifference.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SquaredDifference.java
deleted file mode 100644
index b0cc11688d7..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SquaredDifference.java
+++ /dev/null
@@ -1,79 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.math;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Returns (x - y)(x - y) element-wise.
- * <p>
- * <i>NOTE</i>: `math.SquaredDifference` supports broadcasting. More about broadcasting
- * [here](http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html)
- * 
- * @param <T> data type for {@code z()} output
- */
-@Operator(group = "math")
-public final class SquaredDifference<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new SquaredDifference operation.
-   * 
-   * @param scope current scope
-   * @param x 
-   * @param y 
-   * @return a new instance of SquaredDifference
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> SquaredDifference<T> create(Scope scope, Operand<T> x, Operand<T> y) {
-    OperationBuilder opBuilder = scope.env().opBuilder("SquaredDifference", scope.makeOpName("SquaredDifference"));
-    opBuilder.addInput(x.asOutput(scope));
-    opBuilder.addInput(y.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new SquaredDifference<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> z() {
-    return z;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return z;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "SquaredDifference";
-  
-  private Output<T> z;
-  
-  private SquaredDifference(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    z = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Sub.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Sub.java
deleted file mode 100644
index cff65b6c38b..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Sub.java
+++ /dev/null
@@ -1,79 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.math;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Returns x - y element-wise.
- * <p>
- * <i>NOTE</i>: `math.Sub` supports broadcasting. More about broadcasting
- * [here](http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html)
- * 
- * @param <T> data type for {@code z()} output
- */
-@Operator(group = "math")
-public final class Sub<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new Sub operation.
-   * 
-   * @param scope current scope
-   * @param x 
-   * @param y 
-   * @return a new instance of Sub
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> Sub<T> create(Scope scope, Operand<T> x, Operand<T> y) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Sub", scope.makeOpName("Sub"));
-    opBuilder.addInput(x.asOutput(scope));
-    opBuilder.addInput(y.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new Sub<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> z() {
-    return z;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return z;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Sub";
-  
-  private Output<T> z;
-  
-  private Sub(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    z = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Tan.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Tan.java
deleted file mode 100644
index 07e2f9ddf06..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Tan.java
+++ /dev/null
@@ -1,85 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.math;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Computes tan of x element-wise.
- * <p>
- *   Given an input tensor, this function computes tangent of every
- *   element in the tensor. Input range is `(-inf, inf)` and
- *   output range is `(-inf, inf)`. If input lies outside the boundary, `nan`
- *   is returned.
- * <p>
- *   <pre>{@code
- *   x = tf.constant([-float("inf"), -9, -0.5, 1, 1.2, 200, 10000, float("inf")])
- *   tf.math.tan(x) ==> [nan 0.45231566 -0.5463025 1.5574077 2.572152 -1.7925274 0.32097113 nan]
- *   }</pre>
- * 
- * 
- * @param <T> data type for {@code y()} output
- */
-@Operator(group = "math")
-public final class Tan<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new Tan operation.
-   * 
-   * @param scope current scope
-   * @param x 
-   * @return a new instance of Tan
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> Tan<T> create(Scope scope, Operand<T> x) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Tan", scope.makeOpName("Tan"));
-    opBuilder.addInput(x.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new Tan<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> y() {
-    return y;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return y;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Tan";
-  
-  private Output<T> y;
-  
-  private Tan(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    y = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Tanh.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Tanh.java
deleted file mode 100644
index 66ae2614338..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Tanh.java
+++ /dev/null
@@ -1,84 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.math;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Computes hyperbolic tangent of `x` element-wise.
- * <p>
- *   Given an input tensor, this function computes hyperbolic tangent of every
- *   element in the tensor. Input range is `[-inf, inf]` and
- *   output range is `[-1,1]`.
- * <p>
- *   <pre>{@code
- *   x = tf.constant([-float("inf"), -5, -0.5, 1, 1.2, 2, 3, float("inf")])
- *   tf.math.tanh(x) ==> [-1. -0.99990916 -0.46211717 0.7615942 0.8336547 0.9640276 0.9950547 1.]
- *   }</pre>
- * 
- * 
- * @param <T> data type for {@code y()} output
- */
-@Operator(group = "math")
-public final class Tanh<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new Tanh operation.
-   * 
-   * @param scope current scope
-   * @param x 
-   * @return a new instance of Tanh
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> Tanh<T> create(Scope scope, Operand<T> x) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Tanh", scope.makeOpName("Tanh"));
-    opBuilder.addInput(x.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new Tanh<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> y() {
-    return y;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return y;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Tanh";
-  
-  private Output<T> y;
-  
-  private Tanh(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    y = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/TanhGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/TanhGrad.java
deleted file mode 100644
index 4e8b54ddba8..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/TanhGrad.java
+++ /dev/null
@@ -1,78 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.math;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Computes the gradient for the tanh of `x` wrt its input.
- * <p>
- * Specifically, `grad = dy <i> (1 - y</i>y)`, where `y = tanh(x)`, and `dy`
- * is the corresponding input gradient.
- * 
- * @param <T> data type for {@code z()} output
- */
-public final class TanhGrad<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new TanhGrad operation.
-   * 
-   * @param scope current scope
-   * @param y 
-   * @param dy 
-   * @return a new instance of TanhGrad
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> TanhGrad<T> create(Scope scope, Operand<T> y, Operand<T> dy) {
-    OperationBuilder opBuilder = scope.env().opBuilder("TanhGrad", scope.makeOpName("TanhGrad"));
-    opBuilder.addInput(y.asOutput(scope));
-    opBuilder.addInput(dy.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new TanhGrad<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> z() {
-    return z;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return z;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "TanhGrad";
-  
-  private Output<T> z;
-  
-  private TanhGrad(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    z = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/TruncateDiv.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/TruncateDiv.java
deleted file mode 100644
index 8ee52866c66..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/TruncateDiv.java
+++ /dev/null
@@ -1,84 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.math;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Returns x / y element-wise for integer types.
- * <p>
- * Truncation designates that negative numbers will round fractional quantities
- * toward zero. I.e. -7 / 5 = -1. This matches C semantics but it is different
- * than Python semantics. See `FloorDiv` for a division function that matches
- * Python Semantics.
- * <p>
- * <i>NOTE</i>: `math.TruncateDiv` supports broadcasting. More about broadcasting
- * [here](http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html)
- * 
- * @param <T> data type for {@code z()} output
- */
-@Operator(group = "math")
-public final class TruncateDiv<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new TruncateDiv operation.
-   * 
-   * @param scope current scope
-   * @param x 
-   * @param y 
-   * @return a new instance of TruncateDiv
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> TruncateDiv<T> create(Scope scope, Operand<T> x, Operand<T> y) {
-    OperationBuilder opBuilder = scope.env().opBuilder("TruncateDiv", scope.makeOpName("TruncateDiv"));
-    opBuilder.addInput(x.asOutput(scope));
-    opBuilder.addInput(y.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new TruncateDiv<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> z() {
-    return z;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return z;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "TruncateDiv";
-  
-  private Output<T> z;
-  
-  private TruncateDiv(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    z = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/TruncateMod.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/TruncateMod.java
deleted file mode 100644
index 8ff26efff5b..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/TruncateMod.java
+++ /dev/null
@@ -1,82 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.math;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Returns element-wise remainder of division. This emulates C semantics in that
- * <p>
- * the result here is consistent with a truncating divide. E.g. `truncate(x / y) *
- * y + truncate_mod(x, y) = x`.
- * <p>
- * <i>NOTE</i>: `math.TruncateMod` supports broadcasting. More about broadcasting
- * [here](http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html)
- * 
- * @param <T> data type for {@code z()} output
- */
-@Operator(group = "math")
-public final class TruncateMod<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new TruncateMod operation.
-   * 
-   * @param scope current scope
-   * @param x 
-   * @param y 
-   * @return a new instance of TruncateMod
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> TruncateMod<T> create(Scope scope, Operand<T> x, Operand<T> y) {
-    OperationBuilder opBuilder = scope.env().opBuilder("TruncateMod", scope.makeOpName("TruncateMod"));
-    opBuilder.addInput(x.asOutput(scope));
-    opBuilder.addInput(y.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new TruncateMod<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> z() {
-    return z;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return z;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "TruncateMod";
-  
-  private Output<T> z;
-  
-  private TruncateMod(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    z = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/UnsortedSegmentMax.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/UnsortedSegmentMax.java
deleted file mode 100644
index bdf89cd2ad8..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/UnsortedSegmentMax.java
+++ /dev/null
@@ -1,112 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.math;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Computes the maximum along segments of a tensor.
- * <p>
- * Read
- * [the section on segmentation](https://tensorflow.org/api_docs/python/tf/math#Segmentation)
- * for an explanation of segments.
- * <p>
- * This operator is similar to the unsorted segment sum operator found
- * [(here)](../../../api_docs/python/math_ops.md#UnsortedSegmentSum).
- * Instead of computing the sum over segments, it computes the maximum such that:
- * <p>
- * \\(output_i = \max_{j...} data[j...]\\) where max is over tuples `j...` such
- * that `segment_ids[j...] == i`.
- * <p>
- * If the maximum is empty for a given segment ID `i`, it outputs the smallest
- * possible value for the specific numeric type,
- * `output[i] = numeric_limits<T>::lowest()`.
- * <p>
- * If the given segment ID `i` is negative, then the corresponding value is
- * dropped, and will not be included in the result.
- * <p>
- * <div style="width:70%; margin:auto; margin-bottom:10px; margin-top:20px;">
- * <img style="width:100%" src="https://www.tensorflow.org/images/UnsortedSegmentMax.png" alt>
- * </div>
- * <p>
- * For example:
- * <pre>{@code
- * c = tf.constant([[1,2,3,4], [5,6,7,8], [4,3,2,1]])
- * tf.unsorted_segment_max(c, tf.constant([0, 1, 0]), num_segments=2)
- * # ==> [[ 4,  3, 3, 4],
- * #       [5,  6, 7, 8]]
- * }</pre>
- * 
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator(group = "math")
-public final class UnsortedSegmentMax<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new UnsortedSegmentMax operation.
-   * 
-   * @param scope current scope
-   * @param data 
-   * @param segmentIds A tensor whose shape is a prefix of `data.shape`.
-   * @param numSegments 
-   * @return a new instance of UnsortedSegmentMax
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber, U extends TNumber, V extends TNumber> UnsortedSegmentMax<T> create(Scope scope, Operand<T> data, Operand<U> segmentIds, Operand<V> numSegments) {
-    OperationBuilder opBuilder = scope.env().opBuilder("UnsortedSegmentMax", scope.makeOpName("UnsortedSegmentMax"));
-    opBuilder.addInput(data.asOutput(scope));
-    opBuilder.addInput(segmentIds.asOutput(scope));
-    opBuilder.addInput(numSegments.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new UnsortedSegmentMax<T>(opBuilder.build());
-  }
-  
-  /**
-   * Has same shape as data, except for the first `segment_ids.rank`
-   * dimensions, which are replaced with a single dimension which has size
-   * `num_segments`.
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "UnsortedSegmentMax";
-  
-  private Output<T> output;
-  
-  private UnsortedSegmentMax(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/UnsortedSegmentMin.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/UnsortedSegmentMin.java
deleted file mode 100644
index 5c237ccadac..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/UnsortedSegmentMin.java
+++ /dev/null
@@ -1,106 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.math;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Computes the minimum along segments of a tensor.
- * <p>
- * Read
- * [the section on segmentation](https://tensorflow.org/api_docs/python/tf/math#Segmentation)
- * for an explanation of segments.
- * <p>
- * This operator is similar to the unsorted segment sum operator found
- * [(here)](../../../api_docs/python/math_ops.md#UnsortedSegmentSum).
- * Instead of computing the sum over segments, it computes the minimum such that:
- * <p>
- * \\(output_i = \min_{j...} data_[j...]\\) where min is over tuples `j...` such
- * that `segment_ids[j...] == i`.
- * <p>
- * If the minimum is empty for a given segment ID `i`, it outputs the largest
- * possible value for the specific numeric type,
- * `output[i] = numeric_limits<T>::max()`.
- * <p>
- * For example:
- * <pre>{@code
- * c = tf.constant([[1,2,3,4], [5,6,7,8], [4,3,2,1]])
- * tf.unsorted_segment_min(c, tf.constant([0, 1, 0]), num_segments=2)
- * # ==> [[ 1,  2, 2, 1],
- * #       [5,  6, 7, 8]]
- * }</pre>
- * If the given segment ID `i` is negative, then the corresponding value is
- * dropped, and will not be included in the result.
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator(group = "math")
-public final class UnsortedSegmentMin<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new UnsortedSegmentMin operation.
-   * 
-   * @param scope current scope
-   * @param data 
-   * @param segmentIds A tensor whose shape is a prefix of `data.shape`.
-   * @param numSegments 
-   * @return a new instance of UnsortedSegmentMin
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber, U extends TNumber, V extends TNumber> UnsortedSegmentMin<T> create(Scope scope, Operand<T> data, Operand<U> segmentIds, Operand<V> numSegments) {
-    OperationBuilder opBuilder = scope.env().opBuilder("UnsortedSegmentMin", scope.makeOpName("UnsortedSegmentMin"));
-    opBuilder.addInput(data.asOutput(scope));
-    opBuilder.addInput(segmentIds.asOutput(scope));
-    opBuilder.addInput(numSegments.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new UnsortedSegmentMin<T>(opBuilder.build());
-  }
-  
-  /**
-   * Has same shape as data, except for the first `segment_ids.rank`
-   * dimensions, which are replaced with a single dimension which has size
-   * `num_segments`.
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "UnsortedSegmentMin";
-  
-  private Output<T> output;
-  
-  private UnsortedSegmentMin(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/UnsortedSegmentProd.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/UnsortedSegmentProd.java
deleted file mode 100644
index d2eefdc418e..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/UnsortedSegmentProd.java
+++ /dev/null
@@ -1,106 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.math;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * Computes the product along segments of a tensor.
- * <p>
- * Read
- * [the section on segmentation](https://tensorflow.org/api_docs/python/tf/math#Segmentation)
- * for an explanation of segments.
- * <p>
- * This operator is similar to the unsorted segment sum operator found
- * [(here)](../../../api_docs/python/math_ops.md#UnsortedSegmentSum).
- * Instead of computing the sum over segments, it computes the product of all
- * entries belonging to a segment such that:
- * <p>
- * \\(output_i = \prod_{j...} data[j...]\\) where the product is over tuples
- * `j...` such that `segment_ids[j...] == i`.
- * <p>
- * For example:
- * <pre>{@code
- * c = tf.constant([[1,2,3,4], [5,6,7,8], [4,3,2,1]])
- * tf.unsorted_segment_prod(c, tf.constant([0, 1, 0]), num_segments=2)
- * # ==> [[ 4,  6, 6, 4],
- * #       [5,  6, 7, 8]]
- * }</pre>
- * If there is no entry for a given segment ID `i`, it outputs 1.
- * <p>
- * If the given segment ID `i` is negative, then the corresponding value is
- * dropped, and will not be included in the result.
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator(group = "math")
-public final class UnsortedSegmentProd<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new UnsortedSegmentProd operation.
-   * 
-   * @param scope current scope
-   * @param data 
-   * @param segmentIds A tensor whose shape is a prefix of `data.shape`.
-   * @param numSegments 
-   * @return a new instance of UnsortedSegmentProd
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber, V extends TNumber> UnsortedSegmentProd<T> create(Scope scope, Operand<T> data, Operand<U> segmentIds, Operand<V> numSegments) {
-    OperationBuilder opBuilder = scope.env().opBuilder("UnsortedSegmentProd", scope.makeOpName("UnsortedSegmentProd"));
-    opBuilder.addInput(data.asOutput(scope));
-    opBuilder.addInput(segmentIds.asOutput(scope));
-    opBuilder.addInput(numSegments.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new UnsortedSegmentProd<T>(opBuilder.build());
-  }
-  
-  /**
-   * Has same shape as data, except for the first `segment_ids.rank`
-   * dimensions, which are replaced with a single dimension which has size
-   * `num_segments`.
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "UnsortedSegmentProd";
-  
-  private Output<T> output;
-  
-  private UnsortedSegmentProd(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/UnsortedSegmentSum.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/UnsortedSegmentSum.java
deleted file mode 100644
index 7108e10e216..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/UnsortedSegmentSum.java
+++ /dev/null
@@ -1,109 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.math;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * Computes the sum along segments of a tensor.
- * <p>
- * Read
- * [the section on segmentation](https://tensorflow.org/api_docs/python/tf/math#Segmentation)
- * for an explanation of segments.
- * <p>
- * Computes a tensor such that
- * \\(output[i] = \sum_{j...} data[j...]\\) where the sum is over tuples `j...` such
- * that `segment_ids[j...] == i`.  Unlike `SegmentSum`, `segment_ids`
- * need not be sorted and need not cover all values in the full
- * range of valid values.
- * <p>
- * If the sum is empty for a given segment ID `i`, `output[i] = 0`.
- * If the given segment ID `i` is negative, the value is dropped and will not be
- * added to the sum of the segment.
- * <p>
- * `num_segments` should equal the number of distinct segment IDs.
- * <p>
- * <div style="width:70%; margin:auto; margin-bottom:10px; margin-top:20px;">
- * <img style="width:100%" src="https://www.tensorflow.org/images/UnsortedSegmentSum.png" alt>
- * </div>
- * <pre>{@code
- * c = tf.constant([[1,2,3,4], [5,6,7,8], [4,3,2,1]])
- * tf.unsorted_segment_sum(c, tf.constant([0, 1, 0]), num_segments=2)
- * # ==> [[ 5,  5, 5, 5],
- * #       [5,  6, 7, 8]]
- * }</pre>
- * 
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator(group = "math")
-public final class UnsortedSegmentSum<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new UnsortedSegmentSum operation.
-   * 
-   * @param scope current scope
-   * @param data 
-   * @param segmentIds A tensor whose shape is a prefix of `data.shape`.
-   * @param numSegments 
-   * @return a new instance of UnsortedSegmentSum
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber, V extends TNumber> UnsortedSegmentSum<T> create(Scope scope, Operand<T> data, Operand<U> segmentIds, Operand<V> numSegments) {
-    OperationBuilder opBuilder = scope.env().opBuilder("UnsortedSegmentSum", scope.makeOpName("UnsortedSegmentSum"));
-    opBuilder.addInput(data.asOutput(scope));
-    opBuilder.addInput(segmentIds.asOutput(scope));
-    opBuilder.addInput(numSegments.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new UnsortedSegmentSum<T>(opBuilder.build());
-  }
-  
-  /**
-   * Has same shape as data, except for the first `segment_ids.rank`
-   * dimensions, which are replaced with a single dimension which has size
-   * `num_segments`.
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "UnsortedSegmentSum";
-  
-  private Output<T> output;
-  
-  private UnsortedSegmentSum(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Xdivy.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Xdivy.java
deleted file mode 100644
index 5c825f899d5..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Xdivy.java
+++ /dev/null
@@ -1,76 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.math;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Returns 0 if x == 0, and x / y otherwise, elementwise.
- * 
- * @param <T> data type for {@code z()} output
- */
-@Operator(group = "math")
-public final class Xdivy<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new Xdivy operation.
-   * 
-   * @param scope current scope
-   * @param x 
-   * @param y 
-   * @return a new instance of Xdivy
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> Xdivy<T> create(Scope scope, Operand<T> x, Operand<T> y) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Xdivy", scope.makeOpName("Xdivy"));
-    opBuilder.addInput(x.asOutput(scope));
-    opBuilder.addInput(y.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new Xdivy<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> z() {
-    return z;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return z;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Xdivy";
-  
-  private Output<T> z;
-  
-  private Xdivy(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    z = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Xlog1py.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Xlog1py.java
deleted file mode 100644
index c53e83bbc51..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Xlog1py.java
+++ /dev/null
@@ -1,76 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.math;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Returns 0 if x == 0, and x * log1p(y) otherwise, elementwise.
- * 
- * @param <T> data type for {@code z()} output
- */
-@Operator(group = "math")
-public final class Xlog1py<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new Xlog1py operation.
-   * 
-   * @param scope current scope
-   * @param x 
-   * @param y 
-   * @return a new instance of Xlog1py
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> Xlog1py<T> create(Scope scope, Operand<T> x, Operand<T> y) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Xlog1py", scope.makeOpName("Xlog1py"));
-    opBuilder.addInput(x.asOutput(scope));
-    opBuilder.addInput(y.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new Xlog1py<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> z() {
-    return z;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return z;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Xlog1py";
-  
-  private Output<T> z;
-  
-  private Xlog1py(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    z = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Xlogy.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Xlogy.java
deleted file mode 100644
index 8985e6cc7e7..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Xlogy.java
+++ /dev/null
@@ -1,76 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.math;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Returns 0 if x == 0, and x * log(y) otherwise, elementwise.
- * 
- * @param <T> data type for {@code z()} output
- */
-@Operator(group = "math")
-public final class Xlogy<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new Xlogy operation.
-   * 
-   * @param scope current scope
-   * @param x 
-   * @param y 
-   * @return a new instance of Xlogy
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> Xlogy<T> create(Scope scope, Operand<T> x, Operand<T> y) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Xlogy", scope.makeOpName("Xlogy"));
-    opBuilder.addInput(x.asOutput(scope));
-    opBuilder.addInput(y.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new Xlogy<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> z() {
-    return z;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return z;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Xlogy";
-  
-  private Output<T> z;
-  
-  private Xlogy(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    z = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Zeta.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Zeta.java
deleted file mode 100644
index 47cdb43465a..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Zeta.java
+++ /dev/null
@@ -1,80 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.math;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Compute the Hurwitz zeta function \\(\zeta(x, q)\\).
- * <p>
- * The Hurwitz zeta function is defined as:
- * <p>
- * \\(\zeta(x, q) = \sum_{n=0}^{\infty} (q + n)^{-x}\\)
- * 
- * @param <T> data type for {@code z()} output
- */
-@Operator(group = "math")
-public final class Zeta<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new Zeta operation.
-   * 
-   * @param scope current scope
-   * @param x 
-   * @param q 
-   * @return a new instance of Zeta
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> Zeta<T> create(Scope scope, Operand<T> x, Operand<T> q) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Zeta", scope.makeOpName("Zeta"));
-    opBuilder.addInput(x.asOutput(scope));
-    opBuilder.addInput(q.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new Zeta<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> z() {
-    return z;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return z;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Zeta";
-  
-  private Output<T> z;
-  
-  private Zeta(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    z = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/erfinv.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/erfinv.java
deleted file mode 100644
index 837d55d977b..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/erfinv.java
+++ /dev/null
@@ -1,72 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.math;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * @param <T> data type for {@code y()} output
- */
-@Operator(group = "math")
-public final class erfinv<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new erfinv operation.
-   * 
-   * @param scope current scope
-   * @param x 
-   * @return a new instance of erfinv
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> erfinv<T> create(Scope scope, Operand<T> x) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Erfinv", scope.makeOpName("erfinv"));
-    opBuilder.addInput(x.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new erfinv<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> y() {
-    return y;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return y;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Erfinv";
-  
-  private Output<T> y;
-  
-  private erfinv(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    y = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/BesselJ0.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/BesselJ0.java
deleted file mode 100644
index 7f532e14f32..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/BesselJ0.java
+++ /dev/null
@@ -1,71 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.math.special;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * @param <T> data type for {@code y()} output
- */
-public final class BesselJ0<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new BesselJ0 operation.
-   * 
-   * @param scope current scope
-   * @param x 
-   * @return a new instance of BesselJ0
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> BesselJ0<T> create(Scope scope, Operand<T> x) {
-    OperationBuilder opBuilder = scope.env().opBuilder("BesselJ0", scope.makeOpName("BesselJ0"));
-    opBuilder.addInput(x.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new BesselJ0<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> y() {
-    return y;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return y;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "BesselJ0";
-  
-  private Output<T> y;
-  
-  private BesselJ0(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    y = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/BesselJ1.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/BesselJ1.java
deleted file mode 100644
index d43d61e2a7b..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/BesselJ1.java
+++ /dev/null
@@ -1,71 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.math.special;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * @param <T> data type for {@code y()} output
- */
-public final class BesselJ1<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new BesselJ1 operation.
-   * 
-   * @param scope current scope
-   * @param x 
-   * @return a new instance of BesselJ1
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> BesselJ1<T> create(Scope scope, Operand<T> x) {
-    OperationBuilder opBuilder = scope.env().opBuilder("BesselJ1", scope.makeOpName("BesselJ1"));
-    opBuilder.addInput(x.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new BesselJ1<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> y() {
-    return y;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return y;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "BesselJ1";
-  
-  private Output<T> y;
-  
-  private BesselJ1(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    y = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/BesselK0.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/BesselK0.java
deleted file mode 100644
index 23c71bde19d..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/BesselK0.java
+++ /dev/null
@@ -1,71 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.math.special;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * @param <T> data type for {@code y()} output
- */
-public final class BesselK0<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new BesselK0 operation.
-   * 
-   * @param scope current scope
-   * @param x 
-   * @return a new instance of BesselK0
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> BesselK0<T> create(Scope scope, Operand<T> x) {
-    OperationBuilder opBuilder = scope.env().opBuilder("BesselK0", scope.makeOpName("BesselK0"));
-    opBuilder.addInput(x.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new BesselK0<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> y() {
-    return y;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return y;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "BesselK0";
-  
-  private Output<T> y;
-  
-  private BesselK0(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    y = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/BesselK0e.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/BesselK0e.java
deleted file mode 100644
index 854c4c11c7d..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/BesselK0e.java
+++ /dev/null
@@ -1,71 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.math.special;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * @param <T> data type for {@code y()} output
- */
-public final class BesselK0e<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new BesselK0e operation.
-   * 
-   * @param scope current scope
-   * @param x 
-   * @return a new instance of BesselK0e
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> BesselK0e<T> create(Scope scope, Operand<T> x) {
-    OperationBuilder opBuilder = scope.env().opBuilder("BesselK0e", scope.makeOpName("BesselK0e"));
-    opBuilder.addInput(x.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new BesselK0e<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> y() {
-    return y;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return y;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "BesselK0e";
-  
-  private Output<T> y;
-  
-  private BesselK0e(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    y = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/BesselK1.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/BesselK1.java
deleted file mode 100644
index 7e4a9b3339a..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/BesselK1.java
+++ /dev/null
@@ -1,71 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.math.special;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * @param <T> data type for {@code y()} output
- */
-public final class BesselK1<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new BesselK1 operation.
-   * 
-   * @param scope current scope
-   * @param x 
-   * @return a new instance of BesselK1
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> BesselK1<T> create(Scope scope, Operand<T> x) {
-    OperationBuilder opBuilder = scope.env().opBuilder("BesselK1", scope.makeOpName("BesselK1"));
-    opBuilder.addInput(x.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new BesselK1<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> y() {
-    return y;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return y;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "BesselK1";
-  
-  private Output<T> y;
-  
-  private BesselK1(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    y = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/BesselK1e.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/BesselK1e.java
deleted file mode 100644
index 0e012259687..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/BesselK1e.java
+++ /dev/null
@@ -1,71 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.math.special;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * @param <T> data type for {@code y()} output
- */
-public final class BesselK1e<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new BesselK1e operation.
-   * 
-   * @param scope current scope
-   * @param x 
-   * @return a new instance of BesselK1e
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> BesselK1e<T> create(Scope scope, Operand<T> x) {
-    OperationBuilder opBuilder = scope.env().opBuilder("BesselK1e", scope.makeOpName("BesselK1e"));
-    opBuilder.addInput(x.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new BesselK1e<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> y() {
-    return y;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return y;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "BesselK1e";
-  
-  private Output<T> y;
-  
-  private BesselK1e(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    y = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/BesselY0.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/BesselY0.java
deleted file mode 100644
index 318fd9d490a..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/BesselY0.java
+++ /dev/null
@@ -1,71 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.math.special;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * @param <T> data type for {@code y()} output
- */
-public final class BesselY0<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new BesselY0 operation.
-   * 
-   * @param scope current scope
-   * @param x 
-   * @return a new instance of BesselY0
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> BesselY0<T> create(Scope scope, Operand<T> x) {
-    OperationBuilder opBuilder = scope.env().opBuilder("BesselY0", scope.makeOpName("BesselY0"));
-    opBuilder.addInput(x.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new BesselY0<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> y() {
-    return y;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return y;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "BesselY0";
-  
-  private Output<T> y;
-  
-  private BesselY0(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    y = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/BesselY1.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/BesselY1.java
deleted file mode 100644
index 954bc8a7768..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/BesselY1.java
+++ /dev/null
@@ -1,71 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.math.special;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * @param <T> data type for {@code y()} output
- */
-public final class BesselY1<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new BesselY1 operation.
-   * 
-   * @param scope current scope
-   * @param x 
-   * @return a new instance of BesselY1
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> BesselY1<T> create(Scope scope, Operand<T> x) {
-    OperationBuilder opBuilder = scope.env().opBuilder("BesselY1", scope.makeOpName("BesselY1"));
-    opBuilder.addInput(x.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new BesselY1<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> y() {
-    return y;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return y;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "BesselY1";
-  
-  private Output<T> y;
-  
-  private BesselY1(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    y = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/Dawsn.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/Dawsn.java
deleted file mode 100644
index a57ede341f8..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/Dawsn.java
+++ /dev/null
@@ -1,71 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.math.special;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * @param <T> data type for {@code y()} output
- */
-public final class Dawsn<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new Dawsn operation.
-   * 
-   * @param scope current scope
-   * @param x 
-   * @return a new instance of Dawsn
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> Dawsn<T> create(Scope scope, Operand<T> x) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Dawsn", scope.makeOpName("Dawsn"));
-    opBuilder.addInput(x.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new Dawsn<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> y() {
-    return y;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return y;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Dawsn";
-  
-  private Output<T> y;
-  
-  private Dawsn(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    y = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/Expint.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/Expint.java
deleted file mode 100644
index c3ba2e7f3ed..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/Expint.java
+++ /dev/null
@@ -1,71 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.math.special;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * @param <T> data type for {@code y()} output
- */
-public final class Expint<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new Expint operation.
-   * 
-   * @param scope current scope
-   * @param x 
-   * @return a new instance of Expint
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> Expint<T> create(Scope scope, Operand<T> x) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Expint", scope.makeOpName("Expint"));
-    opBuilder.addInput(x.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new Expint<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> y() {
-    return y;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return y;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Expint";
-  
-  private Output<T> y;
-  
-  private Expint(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    y = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/FresnelCos.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/FresnelCos.java
deleted file mode 100644
index 43e19ecdf72..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/FresnelCos.java
+++ /dev/null
@@ -1,71 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.math.special;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * @param <T> data type for {@code y()} output
- */
-public final class FresnelCos<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new FresnelCos operation.
-   * 
-   * @param scope current scope
-   * @param x 
-   * @return a new instance of FresnelCos
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> FresnelCos<T> create(Scope scope, Operand<T> x) {
-    OperationBuilder opBuilder = scope.env().opBuilder("FresnelCos", scope.makeOpName("FresnelCos"));
-    opBuilder.addInput(x.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new FresnelCos<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> y() {
-    return y;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return y;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "FresnelCos";
-  
-  private Output<T> y;
-  
-  private FresnelCos(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    y = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/FresnelSin.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/FresnelSin.java
deleted file mode 100644
index dc35737b1b3..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/FresnelSin.java
+++ /dev/null
@@ -1,71 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.math.special;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * @param <T> data type for {@code y()} output
- */
-public final class FresnelSin<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new FresnelSin operation.
-   * 
-   * @param scope current scope
-   * @param x 
-   * @return a new instance of FresnelSin
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> FresnelSin<T> create(Scope scope, Operand<T> x) {
-    OperationBuilder opBuilder = scope.env().opBuilder("FresnelSin", scope.makeOpName("FresnelSin"));
-    opBuilder.addInput(x.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new FresnelSin<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> y() {
-    return y;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return y;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "FresnelSin";
-  
-  private Output<T> y;
-  
-  private FresnelSin(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    y = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/Spence.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/Spence.java
deleted file mode 100644
index 1e820823960..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/Spence.java
+++ /dev/null
@@ -1,71 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.math.special;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * @param <T> data type for {@code y()} output
- */
-public final class Spence<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new Spence operation.
-   * 
-   * @param scope current scope
-   * @param x 
-   * @return a new instance of Spence
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> Spence<T> create(Scope scope, Operand<T> x) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Spence", scope.makeOpName("Spence"));
-    opBuilder.addInput(x.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new Spence<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> y() {
-    return y;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return y;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Spence";
-  
-  private Output<T> y;
-  
-  private Spence(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    y = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/AvgPool.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/AvgPool.java
deleted file mode 100644
index d3417d490b0..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/AvgPool.java
+++ /dev/null
@@ -1,135 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.nn;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Performs average pooling on the input.
- * <p>
- * Each entry in `output` is the mean of the corresponding size `ksize`
- * window in `value`.
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator(group = "nn")
-public final class AvgPool<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.nn.AvgPool}
-   */
-  public static class Options {
-    
-    /**
-     * @param dataFormat Specify the data format of the input and output data. With the
-     * default format "NHWC", the data is stored in the order of:
-     *     [batch, in_height, in_width, in_channels].
-     * Alternatively, the format could be "NCHW", the data storage order of:
-     *     [batch, in_channels, in_height, in_width].
-     */
-    public Options dataFormat(String dataFormat) {
-      this.dataFormat = dataFormat;
-      return this;
-    }
-    
-    private String dataFormat;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new AvgPool operation.
-   * 
-   * @param scope current scope
-   * @param value 4-D with shape `[batch, height, width, channels]`.
-   * @param ksize The size of the sliding window for each dimension of `value`.
-   * @param strides The stride of the sliding window for each dimension of `value`.
-   * @param padding The type of padding algorithm to use.
-   * @param options carries optional attributes values
-   * @return a new instance of AvgPool
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> AvgPool<T> create(Scope scope, Operand<T> value, List<Long> ksize, List<Long> strides, String padding, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("AvgPool", scope.makeOpName("AvgPool"));
-    opBuilder.addInput(value.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    long[] ksizeArray = new long[ksize.size()];
-    for (int i = 0; i < ksizeArray.length; ++i) {
-      ksizeArray[i] = ksize.get(i);
-    }
-    opBuilder.setAttr("ksize", ksizeArray);
-    long[] stridesArray = new long[strides.size()];
-    for (int i = 0; i < stridesArray.length; ++i) {
-      stridesArray[i] = strides.get(i);
-    }
-    opBuilder.setAttr("strides", stridesArray);
-    opBuilder.setAttr("padding", padding);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.dataFormat != null) {
-          opBuilder.setAttr("data_format", opts.dataFormat);
-        }
-      }
-    }
-    return new AvgPool<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param dataFormat Specify the data format of the input and output data. With the
-   * default format "NHWC", the data is stored in the order of:
-   *     [batch, in_height, in_width, in_channels].
-   * Alternatively, the format could be "NCHW", the data storage order of:
-   *     [batch, in_channels, in_height, in_width].
-   */
-  public static Options dataFormat(String dataFormat) {
-    return new Options().dataFormat(dataFormat);
-  }
-  
-  /**
-   * The average pooled output tensor.
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "AvgPool";
-  
-  private Output<T> output;
-  
-  private AvgPool(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/AvgPool3d.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/AvgPool3d.java
deleted file mode 100644
index 0b9fac0931d..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/AvgPool3d.java
+++ /dev/null
@@ -1,134 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.nn;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Performs 3D average pooling on the input.
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator(group = "nn")
-public final class AvgPool3d<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.nn.AvgPool3d}
-   */
-  public static class Options {
-    
-    /**
-     * @param dataFormat The data format of the input and output data. With the
-     * default format "NDHWC", the data is stored in the order of:
-     *     [batch, in_depth, in_height, in_width, in_channels].
-     * Alternatively, the format could be "NCDHW", the data storage order is:
-     *     [batch, in_channels, in_depth, in_height, in_width].
-     */
-    public Options dataFormat(String dataFormat) {
-      this.dataFormat = dataFormat;
-      return this;
-    }
-    
-    private String dataFormat;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new AvgPool3d operation.
-   * 
-   * @param scope current scope
-   * @param input Shape `[batch, depth, rows, cols, channels]` tensor to pool over.
-   * @param ksize 1-D tensor of length 5. The size of the window for each dimension of
-   * the input tensor. Must have `ksize[0] = ksize[4] = 1`.
-   * @param strides 1-D tensor of length 5. The stride of the sliding window for each
-   * dimension of `input`. Must have `strides[0] = strides[4] = 1`.
-   * @param padding The type of padding algorithm to use.
-   * @param options carries optional attributes values
-   * @return a new instance of AvgPool3d
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> AvgPool3d<T> create(Scope scope, Operand<T> input, List<Long> ksize, List<Long> strides, String padding, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("AvgPool3D", scope.makeOpName("AvgPool3d"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    long[] ksizeArray = new long[ksize.size()];
-    for (int i = 0; i < ksizeArray.length; ++i) {
-      ksizeArray[i] = ksize.get(i);
-    }
-    opBuilder.setAttr("ksize", ksizeArray);
-    long[] stridesArray = new long[strides.size()];
-    for (int i = 0; i < stridesArray.length; ++i) {
-      stridesArray[i] = strides.get(i);
-    }
-    opBuilder.setAttr("strides", stridesArray);
-    opBuilder.setAttr("padding", padding);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.dataFormat != null) {
-          opBuilder.setAttr("data_format", opts.dataFormat);
-        }
-      }
-    }
-    return new AvgPool3d<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param dataFormat The data format of the input and output data. With the
-   * default format "NDHWC", the data is stored in the order of:
-   *     [batch, in_depth, in_height, in_width, in_channels].
-   * Alternatively, the format could be "NCDHW", the data storage order is:
-   *     [batch, in_channels, in_depth, in_height, in_width].
-   */
-  public static Options dataFormat(String dataFormat) {
-    return new Options().dataFormat(dataFormat);
-  }
-  
-  /**
-   * The average pooled output tensor.
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "AvgPool3D";
-  
-  private Output<T> output;
-  
-  private AvgPool3d(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/AvgPool3dGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/AvgPool3dGrad.java
deleted file mode 100644
index 4465248b03a..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/AvgPool3dGrad.java
+++ /dev/null
@@ -1,137 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.nn;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Computes gradients of average pooling function.
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator(group = "nn")
-public final class AvgPool3dGrad<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.nn.AvgPool3dGrad}
-   */
-  public static class Options {
-    
-    /**
-     * @param dataFormat The data format of the input and output data. With the
-     * default format "NDHWC", the data is stored in the order of:
-     *     [batch, in_depth, in_height, in_width, in_channels].
-     * Alternatively, the format could be "NCDHW", the data storage order is:
-     *     [batch, in_channels, in_depth, in_height, in_width].
-     */
-    public Options dataFormat(String dataFormat) {
-      this.dataFormat = dataFormat;
-      return this;
-    }
-    
-    private String dataFormat;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new AvgPool3dGrad operation.
-   * 
-   * @param scope current scope
-   * @param origInputShape The original input dimensions.
-   * @param grad Output backprop of shape `[batch, depth, rows, cols, channels]`.
-   * @param ksize 1-D tensor of length 5. The size of the window for each dimension of
-   * the input tensor. Must have `ksize[0] = ksize[4] = 1`.
-   * @param strides 1-D tensor of length 5. The stride of the sliding window for each
-   * dimension of `input`. Must have `strides[0] = strides[4] = 1`.
-   * @param padding The type of padding algorithm to use.
-   * @param options carries optional attributes values
-   * @return a new instance of AvgPool3dGrad
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> AvgPool3dGrad<T> create(Scope scope, Operand<TInt32> origInputShape, Operand<T> grad, List<Long> ksize, List<Long> strides, String padding, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("AvgPool3DGrad", scope.makeOpName("AvgPool3dGrad"));
-    opBuilder.addInput(origInputShape.asOutput(scope));
-    opBuilder.addInput(grad.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    long[] ksizeArray = new long[ksize.size()];
-    for (int i = 0; i < ksizeArray.length; ++i) {
-      ksizeArray[i] = ksize.get(i);
-    }
-    opBuilder.setAttr("ksize", ksizeArray);
-    long[] stridesArray = new long[strides.size()];
-    for (int i = 0; i < stridesArray.length; ++i) {
-      stridesArray[i] = strides.get(i);
-    }
-    opBuilder.setAttr("strides", stridesArray);
-    opBuilder.setAttr("padding", padding);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.dataFormat != null) {
-          opBuilder.setAttr("data_format", opts.dataFormat);
-        }
-      }
-    }
-    return new AvgPool3dGrad<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param dataFormat The data format of the input and output data. With the
-   * default format "NDHWC", the data is stored in the order of:
-   *     [batch, in_depth, in_height, in_width, in_channels].
-   * Alternatively, the format could be "NCDHW", the data storage order is:
-   *     [batch, in_channels, in_depth, in_height, in_width].
-   */
-  public static Options dataFormat(String dataFormat) {
-    return new Options().dataFormat(dataFormat);
-  }
-  
-  /**
-   * The backprop for input.
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "AvgPool3DGrad";
-  
-  private Output<T> output;
-  
-  private AvgPool3dGrad(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/AvgPoolGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/AvgPoolGrad.java
deleted file mode 100644
index 490bf1c6e89..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/AvgPoolGrad.java
+++ /dev/null
@@ -1,135 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.nn;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Computes gradients of the average pooling function.
- * 
- * @param <T> data type for {@code output()} output
- */
-public final class AvgPoolGrad<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.nn.AvgPoolGrad}
-   */
-  public static class Options {
-    
-    /**
-     * @param dataFormat Specify the data format of the input and output data. With the
-     * default format "NHWC", the data is stored in the order of:
-     *     [batch, in_height, in_width, in_channels].
-     * Alternatively, the format could be "NCHW", the data storage order of:
-     *     [batch, in_channels, in_height, in_width].
-     */
-    public Options dataFormat(String dataFormat) {
-      this.dataFormat = dataFormat;
-      return this;
-    }
-    
-    private String dataFormat;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new AvgPoolGrad operation.
-   * 
-   * @param scope current scope
-   * @param origInputShape 1-D.  Shape of the original input to `avg_pool`.
-   * @param grad 4-D with shape `[batch, height, width, channels]`.  Gradients w.r.t.
-   * the output of `avg_pool`.
-   * @param ksize The size of the sliding window for each dimension of the input.
-   * @param strides The stride of the sliding window for each dimension of the input.
-   * @param padding The type of padding algorithm to use.
-   * @param options carries optional attributes values
-   * @return a new instance of AvgPoolGrad
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> AvgPoolGrad<T> create(Scope scope, Operand<TInt32> origInputShape, Operand<T> grad, List<Long> ksize, List<Long> strides, String padding, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("AvgPoolGrad", scope.makeOpName("AvgPoolGrad"));
-    opBuilder.addInput(origInputShape.asOutput(scope));
-    opBuilder.addInput(grad.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    long[] ksizeArray = new long[ksize.size()];
-    for (int i = 0; i < ksizeArray.length; ++i) {
-      ksizeArray[i] = ksize.get(i);
-    }
-    opBuilder.setAttr("ksize", ksizeArray);
-    long[] stridesArray = new long[strides.size()];
-    for (int i = 0; i < stridesArray.length; ++i) {
-      stridesArray[i] = strides.get(i);
-    }
-    opBuilder.setAttr("strides", stridesArray);
-    opBuilder.setAttr("padding", padding);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.dataFormat != null) {
-          opBuilder.setAttr("data_format", opts.dataFormat);
-        }
-      }
-    }
-    return new AvgPoolGrad<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param dataFormat Specify the data format of the input and output data. With the
-   * default format "NHWC", the data is stored in the order of:
-   *     [batch, in_height, in_width, in_channels].
-   * Alternatively, the format could be "NCHW", the data storage order of:
-   *     [batch, in_channels, in_height, in_width].
-   */
-  public static Options dataFormat(String dataFormat) {
-    return new Options().dataFormat(dataFormat);
-  }
-  
-  /**
-   * 4-D.  Gradients w.r.t. the input of `avg_pool`.
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "AvgPoolGrad";
-  
-  private Output<T> output;
-  
-  private AvgPoolGrad(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/BatchNormWithGlobalNormalization.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/BatchNormWithGlobalNormalization.java
deleted file mode 100644
index 7b09a43f339..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/BatchNormWithGlobalNormalization.java
+++ /dev/null
@@ -1,96 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.nn;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Batch normalization.
- * <p>
- * This op is deprecated. Prefer `tf.nn.batch_normalization`.
- * 
- * @param <T> data type for {@code result()} output
- */
-@Operator(group = "nn")
-public final class BatchNormWithGlobalNormalization<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new BatchNormWithGlobalNormalization operation.
-   * 
-   * @param scope current scope
-   * @param t A 4D input Tensor.
-   * @param m A 1D mean Tensor with size matching the last dimension of t.
-   * This is the first output from tf.nn.moments,
-   * or a saved moving average thereof.
-   * @param v A 1D variance Tensor with size matching the last dimension of t.
-   * This is the second output from tf.nn.moments,
-   * or a saved moving average thereof.
-   * @param beta A 1D beta Tensor with size matching the last dimension of t.
-   * An offset to be added to the normalized tensor.
-   * @param gamma A 1D gamma Tensor with size matching the last dimension of t.
-   * If "scale_after_normalization" is true, this tensor will be multiplied
-   * with the normalized tensor.
-   * @param varianceEpsilon A small float number to avoid dividing by 0.
-   * @param scaleAfterNormalization A bool indicating whether the resulted tensor
-   * needs to be multiplied with gamma.
-   * @return a new instance of BatchNormWithGlobalNormalization
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> BatchNormWithGlobalNormalization<T> create(Scope scope, Operand<T> t, Operand<T> m, Operand<T> v, Operand<T> beta, Operand<T> gamma, Float varianceEpsilon, Boolean scaleAfterNormalization) {
-    OperationBuilder opBuilder = scope.env().opBuilder("BatchNormWithGlobalNormalization", scope.makeOpName("BatchNormWithGlobalNormalization"));
-    opBuilder.addInput(t.asOutput(scope));
-    opBuilder.addInput(m.asOutput(scope));
-    opBuilder.addInput(v.asOutput(scope));
-    opBuilder.addInput(beta.asOutput(scope));
-    opBuilder.addInput(gamma.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("variance_epsilon", varianceEpsilon);
-    opBuilder.setAttr("scale_after_normalization", scaleAfterNormalization);
-    return new BatchNormWithGlobalNormalization<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> result() {
-    return result;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return result;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "BatchNormWithGlobalNormalization";
-  
-  private Output<T> result;
-  
-  private BatchNormWithGlobalNormalization(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    result = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/BatchNormWithGlobalNormalizationGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/BatchNormWithGlobalNormalizationGrad.java
deleted file mode 100644
index 812058fff34..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/BatchNormWithGlobalNormalizationGrad.java
+++ /dev/null
@@ -1,127 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.nn;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Gradients for batch normalization.
- * <p>
- * This op is deprecated. See `tf.nn.batch_normalization`.
- * 
- * @param <T> data type for {@code dx()} output
- */
-@Operator(group = "nn")
-public final class BatchNormWithGlobalNormalizationGrad<T extends TType> extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new BatchNormWithGlobalNormalizationGrad operation.
-   * 
-   * @param scope current scope
-   * @param t A 4D input Tensor.
-   * @param m A 1D mean Tensor with size matching the last dimension of t.
-   * This is the first output from tf.nn.moments,
-   * or a saved moving average thereof.
-   * @param v A 1D variance Tensor with size matching the last dimension of t.
-   * This is the second output from tf.nn.moments,
-   * or a saved moving average thereof.
-   * @param gamma A 1D gamma Tensor with size matching the last dimension of t.
-   * If "scale_after_normalization" is true, this Tensor will be multiplied
-   * with the normalized Tensor.
-   * @param backprop 4D backprop Tensor.
-   * @param varianceEpsilon A small float number to avoid dividing by 0.
-   * @param scaleAfterNormalization A bool indicating whether the resulted tensor
-   * needs to be multiplied with gamma.
-   * @return a new instance of BatchNormWithGlobalNormalizationGrad
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> BatchNormWithGlobalNormalizationGrad<T> create(Scope scope, Operand<T> t, Operand<T> m, Operand<T> v, Operand<T> gamma, Operand<T> backprop, Float varianceEpsilon, Boolean scaleAfterNormalization) {
-    OperationBuilder opBuilder = scope.env().opBuilder("BatchNormWithGlobalNormalizationGrad", scope.makeOpName("BatchNormWithGlobalNormalizationGrad"));
-    opBuilder.addInput(t.asOutput(scope));
-    opBuilder.addInput(m.asOutput(scope));
-    opBuilder.addInput(v.asOutput(scope));
-    opBuilder.addInput(gamma.asOutput(scope));
-    opBuilder.addInput(backprop.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("variance_epsilon", varianceEpsilon);
-    opBuilder.setAttr("scale_after_normalization", scaleAfterNormalization);
-    return new BatchNormWithGlobalNormalizationGrad<T>(opBuilder.build());
-  }
-  
-  /**
-   * 4D backprop tensor for input.
-   */
-  public Output<T> dx() {
-    return dx;
-  }
-  
-  /**
-   * 1D backprop tensor for mean.
-   */
-  public Output<T> dm() {
-    return dm;
-  }
-  
-  /**
-   * 1D backprop tensor for variance.
-   */
-  public Output<T> dv() {
-    return dv;
-  }
-  
-  /**
-   * 1D backprop tensor for beta.
-   */
-  public Output<T> db() {
-    return db;
-  }
-  
-  /**
-   * 1D backprop tensor for gamma.
-   */
-  public Output<T> dg() {
-    return dg;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "BatchNormWithGlobalNormalizationGrad";
-  
-  private Output<T> dx;
-  private Output<T> dm;
-  private Output<T> dv;
-  private Output<T> db;
-  private Output<T> dg;
-  
-  private BatchNormWithGlobalNormalizationGrad(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    dx = operation.output(outputIdx++);
-    dm = operation.output(outputIdx++);
-    dv = operation.output(outputIdx++);
-    db = operation.output(outputIdx++);
-    dg = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/BiasAdd.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/BiasAdd.java
deleted file mode 100644
index d57cf9c67e1..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/BiasAdd.java
+++ /dev/null
@@ -1,126 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.nn;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Adds `bias` to `value`.
- * <p>
- * This is a special case of `tf.add` where `bias` is restricted to be 1-D.
- * Broadcasting is supported, so `value` may have any number of dimensions.
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator(group = "nn")
-public final class BiasAdd<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.nn.BiasAdd}
-   */
-  public static class Options {
-    
-    /**
-     * @param dataFormat Specify the data format of the input and output data. With the
-     * default format "NHWC", the bias tensor will be added to the last dimension
-     * of the value tensor.
-     * Alternatively, the format could be "NCHW", the data storage order of:
-     *     [batch, in_channels, in_height, in_width].
-     * The tensor will be added to "in_channels", the third-to-the-last
-     *     dimension.
-     */
-    public Options dataFormat(String dataFormat) {
-      this.dataFormat = dataFormat;
-      return this;
-    }
-    
-    private String dataFormat;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new BiasAdd operation.
-   * 
-   * @param scope current scope
-   * @param value Any number of dimensions.
-   * @param bias 1-D with size the last dimension of `value`.
-   * @param options carries optional attributes values
-   * @return a new instance of BiasAdd
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> BiasAdd<T> create(Scope scope, Operand<T> value, Operand<T> bias, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("BiasAdd", scope.makeOpName("BiasAdd"));
-    opBuilder.addInput(value.asOutput(scope));
-    opBuilder.addInput(bias.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.dataFormat != null) {
-          opBuilder.setAttr("data_format", opts.dataFormat);
-        }
-      }
-    }
-    return new BiasAdd<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param dataFormat Specify the data format of the input and output data. With the
-   * default format "NHWC", the bias tensor will be added to the last dimension
-   * of the value tensor.
-   * Alternatively, the format could be "NCHW", the data storage order of:
-   *     [batch, in_channels, in_height, in_width].
-   * The tensor will be added to "in_channels", the third-to-the-last
-   *     dimension.
-   */
-  public static Options dataFormat(String dataFormat) {
-    return new Options().dataFormat(dataFormat);
-  }
-  
-  /**
-   * Broadcasted sum of `value` and `bias`.
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "BiasAdd";
-  
-  private Output<T> output;
-  
-  private BiasAdd(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/BiasAddGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/BiasAddGrad.java
deleted file mode 100644
index ea4aadd373a..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/BiasAddGrad.java
+++ /dev/null
@@ -1,125 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.nn;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * The backward operation for "BiasAdd" on the "bias" tensor.
- * <p>
- * It accumulates all the values from out_backprop into the feature dimension.
- * For NHWC data format, the feature dimension is the last. For NCHW data format,
- * the feature dimension is the third-to-last.
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator(group = "nn")
-public final class BiasAddGrad<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.nn.BiasAddGrad}
-   */
-  public static class Options {
-    
-    /**
-     * @param dataFormat Specify the data format of the input and output data. With the
-     * default format "NHWC", the bias tensor will be added to the last dimension
-     * of the value tensor.
-     * Alternatively, the format could be "NCHW", the data storage order of:
-     *     [batch, in_channels, in_height, in_width].
-     * The tensor will be added to "in_channels", the third-to-the-last
-     *     dimension.
-     */
-    public Options dataFormat(String dataFormat) {
-      this.dataFormat = dataFormat;
-      return this;
-    }
-    
-    private String dataFormat;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new BiasAddGrad operation.
-   * 
-   * @param scope current scope
-   * @param outBackprop Any number of dimensions.
-   * @param options carries optional attributes values
-   * @return a new instance of BiasAddGrad
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> BiasAddGrad<T> create(Scope scope, Operand<T> outBackprop, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("BiasAddGrad", scope.makeOpName("BiasAddGrad"));
-    opBuilder.addInput(outBackprop.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.dataFormat != null) {
-          opBuilder.setAttr("data_format", opts.dataFormat);
-        }
-      }
-    }
-    return new BiasAddGrad<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param dataFormat Specify the data format of the input and output data. With the
-   * default format "NHWC", the bias tensor will be added to the last dimension
-   * of the value tensor.
-   * Alternatively, the format could be "NCHW", the data storage order of:
-   *     [batch, in_channels, in_height, in_width].
-   * The tensor will be added to "in_channels", the third-to-the-last
-   *     dimension.
-   */
-  public static Options dataFormat(String dataFormat) {
-    return new Options().dataFormat(dataFormat);
-  }
-  
-  /**
-   * 1-D with size the feature dimension of `out_backprop`.
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "BiasAddGrad";
-  
-  private Output<T> output;
-  
-  private BiasAddGrad(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/BlockLSTM.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/BlockLSTM.java
deleted file mode 100644
index 88ec1a64daf..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/BlockLSTM.java
+++ /dev/null
@@ -1,216 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.nn;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Computes the LSTM cell forward propagation for all the time steps.
- * <p>
- * This is equivalent to applying LSTMBlockCell in a loop, like so:
- * <pre>{@code
- * for x1 in unpack(x):
- *   i1, cs1, f1, o1, ci1, co1, h1 = LSTMBlock(
- *     x1, cs_prev, h_prev, w, wci, wcf, wco, b)
- *   cs_prev = cs1
- *   h_prev = h1
- *   i.append(i1)
- *   cs.append(cs1)
- *   f.append(f1)
- *   o.append(o1)
- *   ci.append(ci1)
- *   co.append(co1)
- *   h.append(h1)
- * return pack(i), pack(cs), pack(f), pack(o), pack(ci), pack(ch), pack(h)
- * 
- * Note that unlike LSTMBlockCell (and BlockLSTM) which uses ICFO gate layout,
- * this op uses IFCO. So in order for the following snippet to be equivalent
- * all gate-related outputs should be reordered.
- * }</pre>
- * 
- * 
- * @param <T> data type for {@code i()} output
- */
-public final class BlockLSTM<T extends TNumber> extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.nn.BlockLSTM}
-   */
-  public static class Options {
-    
-    /**
-     * @param cellClip Value to clip the 'cs' value to.
-     */
-    public Options cellClip(Float cellClip) {
-      this.cellClip = cellClip;
-      return this;
-    }
-    
-    /**
-     * @param usePeephole Whether to use peephole weights.
-     */
-    public Options usePeephole(Boolean usePeephole) {
-      this.usePeephole = usePeephole;
-      return this;
-    }
-    
-    private Float cellClip;
-    private Boolean usePeephole;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new BlockLSTM operation.
-   * 
-   * @param scope current scope
-   * @param seqLenMax Maximum time length actually used by this input. Outputs are padded
-   * with zeros beyond this length.
-   * @param x The sequence input to the LSTM, shape (timelen, batch_size, num_inputs).
-   * @param csPrev Value of the initial cell state.
-   * @param hPrev Initial output of cell (to be used for peephole).
-   * @param w The weight matrix.
-   * @param wci The weight matrix for input gate peephole connection.
-   * @param wcf The weight matrix for forget gate peephole connection.
-   * @param wco The weight matrix for output gate peephole connection.
-   * @param b The bias vector.
-   * @param options carries optional attributes values
-   * @return a new instance of BlockLSTM
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> BlockLSTM<T> create(Scope scope, Operand<TInt64> seqLenMax, Operand<T> x, Operand<T> csPrev, Operand<T> hPrev, Operand<T> w, Operand<T> wci, Operand<T> wcf, Operand<T> wco, Operand<T> b, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("BlockLSTMV2", scope.makeOpName("BlockLSTM"));
-    opBuilder.addInput(seqLenMax.asOutput(scope));
-    opBuilder.addInput(x.asOutput(scope));
-    opBuilder.addInput(csPrev.asOutput(scope));
-    opBuilder.addInput(hPrev.asOutput(scope));
-    opBuilder.addInput(w.asOutput(scope));
-    opBuilder.addInput(wci.asOutput(scope));
-    opBuilder.addInput(wcf.asOutput(scope));
-    opBuilder.addInput(wco.asOutput(scope));
-    opBuilder.addInput(b.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.cellClip != null) {
-          opBuilder.setAttr("cell_clip", opts.cellClip);
-        }
-        if (opts.usePeephole != null) {
-          opBuilder.setAttr("use_peephole", opts.usePeephole);
-        }
-      }
-    }
-    return new BlockLSTM<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param cellClip Value to clip the 'cs' value to.
-   */
-  public static Options cellClip(Float cellClip) {
-    return new Options().cellClip(cellClip);
-  }
-  
-  /**
-   * @param usePeephole Whether to use peephole weights.
-   */
-  public static Options usePeephole(Boolean usePeephole) {
-    return new Options().usePeephole(usePeephole);
-  }
-  
-  /**
-   * The input gate over the whole time sequence.
-   */
-  public Output<T> i() {
-    return i;
-  }
-  
-  /**
-   * The cell state before the tanh over the whole time sequence.
-   */
-  public Output<T> cs() {
-    return cs;
-  }
-  
-  /**
-   * The forget gate over the whole time sequence.
-   */
-  public Output<T> f() {
-    return f;
-  }
-  
-  /**
-   * The output gate over the whole time sequence.
-   */
-  public Output<T> o() {
-    return o;
-  }
-  
-  /**
-   * The cell input over the whole time sequence.
-   */
-  public Output<T> ci() {
-    return ci;
-  }
-  
-  /**
-   * The cell after the tanh over the whole time sequence.
-   */
-  public Output<T> co() {
-    return co;
-  }
-  
-  /**
-   * The output h vector over the whole time sequence.
-   */
-  public Output<T> h() {
-    return h;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "BlockLSTMV2";
-  
-  private Output<T> i;
-  private Output<T> cs;
-  private Output<T> f;
-  private Output<T> o;
-  private Output<T> ci;
-  private Output<T> co;
-  private Output<T> h;
-  
-  private BlockLSTM(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    i = operation.output(outputIdx++);
-    cs = operation.output(outputIdx++);
-    f = operation.output(outputIdx++);
-    o = operation.output(outputIdx++);
-    ci = operation.output(outputIdx++);
-    co = operation.output(outputIdx++);
-    h = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/BlockLSTMGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/BlockLSTMGrad.java
deleted file mode 100644
index 4f8907f3ee4..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/BlockLSTMGrad.java
+++ /dev/null
@@ -1,172 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.nn;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Computes the LSTM cell backward propagation for the entire time sequence.
- * <p>
- * This implementation is to be used in conjunction of BlockLSTMV2.
- * 
- * @param <T> data type for {@code xGrad()} output
- */
-public final class BlockLSTMGrad<T extends TNumber> extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new BlockLSTMGrad operation.
-   * 
-   * @param scope current scope
-   * @param seqLenMax Maximum time length actually used by this input. Outputs are padded
-   * with zeros beyond this length.
-   * @param x The sequence input to the LSTM, shape (timelen, batch_size, num_inputs).
-   * @param csPrev Value of the initial cell state.
-   * @param hPrev Initial output of cell (to be used for peephole).
-   * @param w The weight matrix.
-   * @param wci The weight matrix for input gate peephole connection.
-   * @param wcf The weight matrix for forget gate peephole connection.
-   * @param wco The weight matrix for output gate peephole connection.
-   * @param b The bias vector.
-   * @param i The input gate over the whole time sequence.
-   * @param cs The cell state before the tanh over the whole time sequence.
-   * @param f The forget gate over the whole time sequence.
-   * @param o The output gate over the whole time sequence.
-   * @param ci The cell input over the whole time sequence.
-   * @param co The cell after the tanh over the whole time sequence.
-   * @param h The output h vector over the whole time sequence.
-   * @param csGrad The current gradient of cs.
-   * @param hGrad The gradient of h vector.
-   * @param usePeephole Whether to use peephole weights.
-   * @return a new instance of BlockLSTMGrad
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> BlockLSTMGrad<T> create(Scope scope, Operand<TInt64> seqLenMax, Operand<T> x, Operand<T> csPrev, Operand<T> hPrev, Operand<T> w, Operand<T> wci, Operand<T> wcf, Operand<T> wco, Operand<T> b, Operand<T> i, Operand<T> cs, Operand<T> f, Operand<T> o, Operand<T> ci, Operand<T> co, Operand<T> h, Operand<T> csGrad, Operand<T> hGrad, Boolean usePeephole) {
-    OperationBuilder opBuilder = scope.env().opBuilder("BlockLSTMGradV2", scope.makeOpName("BlockLSTMGrad"));
-    opBuilder.addInput(seqLenMax.asOutput(scope));
-    opBuilder.addInput(x.asOutput(scope));
-    opBuilder.addInput(csPrev.asOutput(scope));
-    opBuilder.addInput(hPrev.asOutput(scope));
-    opBuilder.addInput(w.asOutput(scope));
-    opBuilder.addInput(wci.asOutput(scope));
-    opBuilder.addInput(wcf.asOutput(scope));
-    opBuilder.addInput(wco.asOutput(scope));
-    opBuilder.addInput(b.asOutput(scope));
-    opBuilder.addInput(i.asOutput(scope));
-    opBuilder.addInput(cs.asOutput(scope));
-    opBuilder.addInput(f.asOutput(scope));
-    opBuilder.addInput(o.asOutput(scope));
-    opBuilder.addInput(ci.asOutput(scope));
-    opBuilder.addInput(co.asOutput(scope));
-    opBuilder.addInput(h.asOutput(scope));
-    opBuilder.addInput(csGrad.asOutput(scope));
-    opBuilder.addInput(hGrad.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("use_peephole", usePeephole);
-    return new BlockLSTMGrad<T>(opBuilder.build());
-  }
-  
-  /**
-   * The gradient of x to be back-propped.
-   */
-  public Output<T> xGrad() {
-    return xGrad;
-  }
-  
-  /**
-   * The gradient of cs_prev to be back-propped.
-   */
-  public Output<T> csPrevGrad() {
-    return csPrevGrad;
-  }
-  
-  /**
-   * The gradient of h_prev to be back-propped.
-   */
-  public Output<T> hPrevGrad() {
-    return hPrevGrad;
-  }
-  
-  /**
-   * The gradient for w to be back-propped.
-   */
-  public Output<T> wGrad() {
-    return wGrad;
-  }
-  
-  /**
-   * The gradient for wci to be back-propped.
-   */
-  public Output<T> wciGrad() {
-    return wciGrad;
-  }
-  
-  /**
-   * The gradient for wcf to be back-propped.
-   */
-  public Output<T> wcfGrad() {
-    return wcfGrad;
-  }
-  
-  /**
-   * The gradient for wco to be back-propped.
-   */
-  public Output<T> wcoGrad() {
-    return wcoGrad;
-  }
-  
-  /**
-   * The gradient for w to be back-propped.
-   */
-  public Output<T> bGrad() {
-    return bGrad;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "BlockLSTMGradV2";
-  
-  private Output<T> xGrad;
-  private Output<T> csPrevGrad;
-  private Output<T> hPrevGrad;
-  private Output<T> wGrad;
-  private Output<T> wciGrad;
-  private Output<T> wcfGrad;
-  private Output<T> wcoGrad;
-  private Output<T> bGrad;
-  
-  private BlockLSTMGrad(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    xGrad = operation.output(outputIdx++);
-    csPrevGrad = operation.output(outputIdx++);
-    hPrevGrad = operation.output(outputIdx++);
-    wGrad = operation.output(outputIdx++);
-    wciGrad = operation.output(outputIdx++);
-    wcfGrad = operation.output(outputIdx++);
-    wcoGrad = operation.output(outputIdx++);
-    bGrad = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CTCLossV2.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CTCLossV2.java
deleted file mode 100644
index c79ab1414fd..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CTCLossV2.java
+++ /dev/null
@@ -1,173 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.nn;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.TInt64;
-
-/**
- * Calculates the CTC Loss (log probability) for each batch entry.  Also calculates
- * <p>
- * the gradient.  This class performs the softmax operation for you, so inputs
- * should be e.g. linear projections of outputs by an LSTM.
- */
-public final class CTCLossV2 extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.nn.CTCLossV2}
-   */
-  public static class Options {
-    
-    /**
-     * @param preprocessCollapseRepeated Scalar, if true then repeated labels are
-     * collapsed prior to the CTC calculation.
-     */
-    public Options preprocessCollapseRepeated(Boolean preprocessCollapseRepeated) {
-      this.preprocessCollapseRepeated = preprocessCollapseRepeated;
-      return this;
-    }
-    
-    /**
-     * @param ctcMergeRepeated Scalar.  If set to false, <i>during</i> CTC calculation
-     * repeated non-blank labels will not be merged and are interpreted as
-     * individual labels.  This is a simplified version of CTC.
-     */
-    public Options ctcMergeRepeated(Boolean ctcMergeRepeated) {
-      this.ctcMergeRepeated = ctcMergeRepeated;
-      return this;
-    }
-    
-    /**
-     * @param ignoreLongerOutputsThanInputs Scalar. If set to true, during CTC
-     * calculation, items that have longer output sequences than input sequences
-     * are skipped: they don't contribute to the loss term and have zero-gradient.
-     */
-    public Options ignoreLongerOutputsThanInputs(Boolean ignoreLongerOutputsThanInputs) {
-      this.ignoreLongerOutputsThanInputs = ignoreLongerOutputsThanInputs;
-      return this;
-    }
-    
-    private Boolean preprocessCollapseRepeated;
-    private Boolean ctcMergeRepeated;
-    private Boolean ignoreLongerOutputsThanInputs;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new CTCLossV2 operation.
-   * 
-   * @param scope current scope
-   * @param inputs 3-D, shape: `(max_time x batch_size x num_classes)`, the logits. Default blank
-   * label is 0 rather num_classes - 1.
-   * @param labelsIndices The indices of a `SparseTensor<int32, 2>`.
-   * `labels_indices(i, :) == [b, t]` means `labels_values(i)` stores the id for
-   * `(batch b, time t)`.
-   * @param labelsValues The values (labels) associated with the given batch and time.
-   * @param sequenceLength A vector containing sequence lengths (batch).
-   * @param options carries optional attributes values
-   * @return a new instance of CTCLossV2
-   */
-  @Endpoint(describeByClass = true)
-  public static CTCLossV2 create(Scope scope, Operand<TFloat32> inputs, Operand<TInt64> labelsIndices, Operand<TInt32> labelsValues, Operand<TInt32> sequenceLength, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("CTCLossV2", scope.makeOpName("CTCLossV2"));
-    opBuilder.addInput(inputs.asOutput(scope));
-    opBuilder.addInput(labelsIndices.asOutput(scope));
-    opBuilder.addInput(labelsValues.asOutput(scope));
-    opBuilder.addInput(sequenceLength.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.preprocessCollapseRepeated != null) {
-          opBuilder.setAttr("preprocess_collapse_repeated", opts.preprocessCollapseRepeated);
-        }
-        if (opts.ctcMergeRepeated != null) {
-          opBuilder.setAttr("ctc_merge_repeated", opts.ctcMergeRepeated);
-        }
-        if (opts.ignoreLongerOutputsThanInputs != null) {
-          opBuilder.setAttr("ignore_longer_outputs_than_inputs", opts.ignoreLongerOutputsThanInputs);
-        }
-      }
-    }
-    return new CTCLossV2(opBuilder.build());
-  }
-  
-  /**
-   * @param preprocessCollapseRepeated Scalar, if true then repeated labels are
-   * collapsed prior to the CTC calculation.
-   */
-  public static Options preprocessCollapseRepeated(Boolean preprocessCollapseRepeated) {
-    return new Options().preprocessCollapseRepeated(preprocessCollapseRepeated);
-  }
-  
-  /**
-   * @param ctcMergeRepeated Scalar.  If set to false, <i>during</i> CTC calculation
-   * repeated non-blank labels will not be merged and are interpreted as
-   * individual labels.  This is a simplified version of CTC.
-   */
-  public static Options ctcMergeRepeated(Boolean ctcMergeRepeated) {
-    return new Options().ctcMergeRepeated(ctcMergeRepeated);
-  }
-  
-  /**
-   * @param ignoreLongerOutputsThanInputs Scalar. If set to true, during CTC
-   * calculation, items that have longer output sequences than input sequences
-   * are skipped: they don't contribute to the loss term and have zero-gradient.
-   */
-  public static Options ignoreLongerOutputsThanInputs(Boolean ignoreLongerOutputsThanInputs) {
-    return new Options().ignoreLongerOutputsThanInputs(ignoreLongerOutputsThanInputs);
-  }
-  
-  /**
-   * A vector (batch) containing log-probabilities.
-   */
-  public Output<TFloat32> loss() {
-    return loss;
-  }
-  
-  /**
-   * The gradient of `loss`.  3-D, shape:
-   * `(max_time x batch_size x num_classes)`.
-   */
-  public Output<TFloat32> gradient() {
-    return gradient;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "CTCLossV2";
-  
-  private Output<TFloat32> loss;
-  private Output<TFloat32> gradient;
-  
-  private CTCLossV2(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    loss = operation.output(outputIdx++);
-    gradient = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/ComputeAccidentalHits.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/ComputeAccidentalHits.java
deleted file mode 100644
index 6329a45cb74..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/ComputeAccidentalHits.java
+++ /dev/null
@@ -1,156 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.nn;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.TInt64;
-
-/**
- * Computes the ids of the positions in sampled_candidates that match true_labels.
- * <p>
- * When doing log-odds NCE, the result of this op should be passed through a
- * SparseToDense op, then added to the logits of the sampled candidates. This has
- * the effect of 'removing' the sampled labels that match the true labels by
- * making the classifier sure that they are sampled labels.
- */
-@Operator(group = "nn")
-public final class ComputeAccidentalHits extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.nn.ComputeAccidentalHits}
-   */
-  public static class Options {
-    
-    /**
-     * @param seed If either seed or seed2 are set to be non-zero, the random number
-     * generator is seeded by the given seed.  Otherwise, it is seeded by a
-     * random seed.
-     */
-    public Options seed(Long seed) {
-      this.seed = seed;
-      return this;
-    }
-    
-    /**
-     * @param seed2 An second seed to avoid seed collision.
-     */
-    public Options seed2(Long seed2) {
-      this.seed2 = seed2;
-      return this;
-    }
-    
-    private Long seed;
-    private Long seed2;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new ComputeAccidentalHits operation.
-   * 
-   * @param scope current scope
-   * @param trueClasses The true_classes output of UnpackSparseLabels.
-   * @param sampledCandidates The sampled_candidates output of CandidateSampler.
-   * @param numTrue Number of true labels per context.
-   * @param options carries optional attributes values
-   * @return a new instance of ComputeAccidentalHits
-   */
-  @Endpoint(describeByClass = true)
-  public static ComputeAccidentalHits create(Scope scope, Operand<TInt64> trueClasses, Operand<TInt64> sampledCandidates, Long numTrue, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ComputeAccidentalHits", scope.makeOpName("ComputeAccidentalHits"));
-    opBuilder.addInput(trueClasses.asOutput(scope));
-    opBuilder.addInput(sampledCandidates.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("num_true", numTrue);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.seed != null) {
-          opBuilder.setAttr("seed", opts.seed);
-        }
-        if (opts.seed2 != null) {
-          opBuilder.setAttr("seed2", opts.seed2);
-        }
-      }
-    }
-    return new ComputeAccidentalHits(opBuilder.build());
-  }
-  
-  /**
-   * @param seed If either seed or seed2 are set to be non-zero, the random number
-   * generator is seeded by the given seed.  Otherwise, it is seeded by a
-   * random seed.
-   */
-  public static Options seed(Long seed) {
-    return new Options().seed(seed);
-  }
-  
-  /**
-   * @param seed2 An second seed to avoid seed collision.
-   */
-  public static Options seed2(Long seed2) {
-    return new Options().seed2(seed2);
-  }
-  
-  /**
-   * A vector of indices corresponding to rows of true_candidates.
-   */
-  public Output<TInt32> indices() {
-    return indices;
-  }
-  
-  /**
-   * A vector of IDs of positions in sampled_candidates that match a true_label
-   * for the row with the corresponding index in indices.
-   */
-  public Output<TInt64> ids() {
-    return ids;
-  }
-  
-  /**
-   * A vector of the same length as indices and ids, in which each element
-   * is -FLOAT_MAX.
-   */
-  public Output<TFloat32> weights() {
-    return weights;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ComputeAccidentalHits";
-  
-  private Output<TInt32> indices;
-  private Output<TInt64> ids;
-  private Output<TFloat32> weights;
-  
-  private ComputeAccidentalHits(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    indices = operation.output(outputIdx++);
-    ids = operation.output(outputIdx++);
-    weights = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Conv2d.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Conv2d.java
deleted file mode 100644
index af5f16acbb2..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Conv2d.java
+++ /dev/null
@@ -1,234 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.nn;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Computes a 2-D convolution given 4-D `input` and `filter` tensors.
- * <p>
- * Given an input tensor of shape `[batch, in_height, in_width, in_channels]`
- * and a filter / kernel tensor of shape
- * `[filter_height, filter_width, in_channels, out_channels]`, this op
- * performs the following:
- * <p>
- * 1. Flattens the filter to a 2-D matrix with shape
- *    `[filter_height * filter_width * in_channels, output_channels]`.
- * 2. Extracts image patches from the input tensor to form a <i>virtual</i>
- *    tensor of shape `[batch, out_height, out_width,
- *    filter_height * filter_width * in_channels]`.
- * 3. For each patch, right-multiplies the filter matrix and the image patch
- *    vector.
- * <p>
- * In detail, with the default NHWC format,
- * <p>
- *     output[b, i, j, k] =
- *         sum_{di, dj, q} input[b, strides[1] * i + di, strides[2] * j + dj, q] *
- *                         filter[di, dj, q, k]
- * <p>
- * Must have `strides[0] = strides[3] = 1`.  For the most common case of the same
- * horizontal and vertices strides, `strides = [1, stride, stride, 1]`.
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator(group = "nn")
-public final class Conv2d<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.nn.Conv2d}
-   */
-  public static class Options {
-    
-    /**
-     * @param useCudnnOnGpu 
-     */
-    public Options useCudnnOnGpu(Boolean useCudnnOnGpu) {
-      this.useCudnnOnGpu = useCudnnOnGpu;
-      return this;
-    }
-    
-    /**
-     * @param explicitPaddings If `padding` is `"EXPLICIT"`, the list of explicit padding amounts. For the ith
-     * dimension, the amount of padding inserted before and after the dimension is
-     * `explicit_paddings[2 * i]` and `explicit_paddings[2 * i + 1]`, respectively. If
-     * `padding` is not `"EXPLICIT"`, `explicit_paddings` must be empty.
-     */
-    public Options explicitPaddings(List<Long> explicitPaddings) {
-      this.explicitPaddings = explicitPaddings;
-      return this;
-    }
-    
-    /**
-     * @param dataFormat Specify the data format of the input and output data. With the
-     * default format "NHWC", the data is stored in the order of:
-     *     [batch, height, width, channels].
-     * Alternatively, the format could be "NCHW", the data storage order of:
-     *     [batch, channels, height, width].
-     */
-    public Options dataFormat(String dataFormat) {
-      this.dataFormat = dataFormat;
-      return this;
-    }
-    
-    /**
-     * @param dilations 1-D tensor of length 4.  The dilation factor for each dimension of
-     * `input`. If set to k > 1, there will be k-1 skipped cells between each
-     * filter element on that dimension. The dimension order is determined by the
-     * value of `data_format`, see above for details. Dilations in the batch and
-     * depth dimensions must be 1.
-     */
-    public Options dilations(List<Long> dilations) {
-      this.dilations = dilations;
-      return this;
-    }
-    
-    private Boolean useCudnnOnGpu;
-    private List<Long> explicitPaddings;
-    private String dataFormat;
-    private List<Long> dilations;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new Conv2d operation.
-   * 
-   * @param scope current scope
-   * @param input A 4-D tensor. The dimension order is interpreted according to the value
-   * of `data_format`, see below for details.
-   * @param filter A 4-D tensor of shape
-   * `[filter_height, filter_width, in_channels, out_channels]`
-   * @param strides 1-D tensor of length 4.  The stride of the sliding window for each
-   * dimension of `input`. The dimension order is determined by the value of
-   * `data_format`, see below for details.
-   * @param padding The type of padding algorithm to use.
-   * @param options carries optional attributes values
-   * @return a new instance of Conv2d
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> Conv2d<T> create(Scope scope, Operand<T> input, Operand<T> filter, List<Long> strides, String padding, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Conv2D", scope.makeOpName("Conv2d"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder.addInput(filter.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    long[] stridesArray = new long[strides.size()];
-    for (int i = 0; i < stridesArray.length; ++i) {
-      stridesArray[i] = strides.get(i);
-    }
-    opBuilder.setAttr("strides", stridesArray);
-    opBuilder.setAttr("padding", padding);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.useCudnnOnGpu != null) {
-          opBuilder.setAttr("use_cudnn_on_gpu", opts.useCudnnOnGpu);
-        }
-        if (opts.explicitPaddings != null) {
-          long[] explicitPaddingsArray = new long[opts.explicitPaddings.size()];
-          for (int i = 0; i < explicitPaddingsArray.length; ++i) {
-            explicitPaddingsArray[i] = opts.explicitPaddings.get(i);
-          }
-          opBuilder.setAttr("explicit_paddings", explicitPaddingsArray);
-        }
-        if (opts.dataFormat != null) {
-          opBuilder.setAttr("data_format", opts.dataFormat);
-        }
-        if (opts.dilations != null) {
-          long[] dilationsArray = new long[opts.dilations.size()];
-          for (int i = 0; i < dilationsArray.length; ++i) {
-            dilationsArray[i] = opts.dilations.get(i);
-          }
-          opBuilder.setAttr("dilations", dilationsArray);
-        }
-      }
-    }
-    return new Conv2d<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param useCudnnOnGpu 
-   */
-  public static Options useCudnnOnGpu(Boolean useCudnnOnGpu) {
-    return new Options().useCudnnOnGpu(useCudnnOnGpu);
-  }
-  
-  /**
-   * @param explicitPaddings If `padding` is `"EXPLICIT"`, the list of explicit padding amounts. For the ith
-   * dimension, the amount of padding inserted before and after the dimension is
-   * `explicit_paddings[2 * i]` and `explicit_paddings[2 * i + 1]`, respectively. If
-   * `padding` is not `"EXPLICIT"`, `explicit_paddings` must be empty.
-   */
-  public static Options explicitPaddings(List<Long> explicitPaddings) {
-    return new Options().explicitPaddings(explicitPaddings);
-  }
-  
-  /**
-   * @param dataFormat Specify the data format of the input and output data. With the
-   * default format "NHWC", the data is stored in the order of:
-   *     [batch, height, width, channels].
-   * Alternatively, the format could be "NCHW", the data storage order of:
-   *     [batch, channels, height, width].
-   */
-  public static Options dataFormat(String dataFormat) {
-    return new Options().dataFormat(dataFormat);
-  }
-  
-  /**
-   * @param dilations 1-D tensor of length 4.  The dilation factor for each dimension of
-   * `input`. If set to k > 1, there will be k-1 skipped cells between each
-   * filter element on that dimension. The dimension order is determined by the
-   * value of `data_format`, see above for details. Dilations in the batch and
-   * depth dimensions must be 1.
-   */
-  public static Options dilations(List<Long> dilations) {
-    return new Options().dilations(dilations);
-  }
-  
-  /**
-   * A 4-D tensor. The dimension order is determined by the value of
-   * `data_format`, see below for details.
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Conv2D";
-  
-  private Output<T> output;
-  
-  private Conv2d(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Conv2dBackpropFilter.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Conv2dBackpropFilter.java
deleted file mode 100644
index 17a7b97f78e..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Conv2dBackpropFilter.java
+++ /dev/null
@@ -1,217 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.nn;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Computes the gradients of convolution with respect to the filter.
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator(group = "nn")
-public final class Conv2dBackpropFilter<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.nn.Conv2dBackpropFilter}
-   */
-  public static class Options {
-    
-    /**
-     * @param useCudnnOnGpu 
-     */
-    public Options useCudnnOnGpu(Boolean useCudnnOnGpu) {
-      this.useCudnnOnGpu = useCudnnOnGpu;
-      return this;
-    }
-    
-    /**
-     * @param explicitPaddings If `padding` is `"EXPLICIT"`, the list of explicit padding amounts. For the ith
-     * dimension, the amount of padding inserted before and after the dimension is
-     * `explicit_paddings[2 * i]` and `explicit_paddings[2 * i + 1]`, respectively. If
-     * `padding` is not `"EXPLICIT"`, `explicit_paddings` must be empty.
-     */
-    public Options explicitPaddings(List<Long> explicitPaddings) {
-      this.explicitPaddings = explicitPaddings;
-      return this;
-    }
-    
-    /**
-     * @param dataFormat Specify the data format of the input and output data. With the
-     * default format "NHWC", the data is stored in the order of:
-     *     [batch, in_height, in_width, in_channels].
-     * Alternatively, the format could be "NCHW", the data storage order of:
-     *     [batch, in_channels, in_height, in_width].
-     */
-    public Options dataFormat(String dataFormat) {
-      this.dataFormat = dataFormat;
-      return this;
-    }
-    
-    /**
-     * @param dilations 1-D tensor of length 4.  The dilation factor for each dimension of
-     * `input`. If set to k > 1, there will be k-1 skipped cells between each filter
-     * element on that dimension. The dimension order is determined by the value of
-     * `data_format`, see above for details. Dilations in the batch and depth
-     * dimensions must be 1.
-     */
-    public Options dilations(List<Long> dilations) {
-      this.dilations = dilations;
-      return this;
-    }
-    
-    private Boolean useCudnnOnGpu;
-    private List<Long> explicitPaddings;
-    private String dataFormat;
-    private List<Long> dilations;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new Conv2dBackpropFilter operation.
-   * 
-   * @param scope current scope
-   * @param input 4-D with shape `[batch, in_height, in_width, in_channels]`.
-   * @param filterSizes An integer vector representing the tensor shape of `filter`,
-   * where `filter` is a 4-D
-   * `[filter_height, filter_width, in_channels, out_channels]` tensor.
-   * @param outBackprop 4-D with shape `[batch, out_height, out_width, out_channels]`.
-   * Gradients w.r.t. the output of the convolution.
-   * @param strides The stride of the sliding window for each dimension of the input
-   * of the convolution. Must be in the same order as the dimension specified with
-   * format.
-   * @param padding The type of padding algorithm to use.
-   * @param options carries optional attributes values
-   * @return a new instance of Conv2dBackpropFilter
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> Conv2dBackpropFilter<T> create(Scope scope, Operand<T> input, Operand<TInt32> filterSizes, Operand<T> outBackprop, List<Long> strides, String padding, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Conv2DBackpropFilter", scope.makeOpName("Conv2dBackpropFilter"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder.addInput(filterSizes.asOutput(scope));
-    opBuilder.addInput(outBackprop.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    long[] stridesArray = new long[strides.size()];
-    for (int i = 0; i < stridesArray.length; ++i) {
-      stridesArray[i] = strides.get(i);
-    }
-    opBuilder.setAttr("strides", stridesArray);
-    opBuilder.setAttr("padding", padding);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.useCudnnOnGpu != null) {
-          opBuilder.setAttr("use_cudnn_on_gpu", opts.useCudnnOnGpu);
-        }
-        if (opts.explicitPaddings != null) {
-          long[] explicitPaddingsArray = new long[opts.explicitPaddings.size()];
-          for (int i = 0; i < explicitPaddingsArray.length; ++i) {
-            explicitPaddingsArray[i] = opts.explicitPaddings.get(i);
-          }
-          opBuilder.setAttr("explicit_paddings", explicitPaddingsArray);
-        }
-        if (opts.dataFormat != null) {
-          opBuilder.setAttr("data_format", opts.dataFormat);
-        }
-        if (opts.dilations != null) {
-          long[] dilationsArray = new long[opts.dilations.size()];
-          for (int i = 0; i < dilationsArray.length; ++i) {
-            dilationsArray[i] = opts.dilations.get(i);
-          }
-          opBuilder.setAttr("dilations", dilationsArray);
-        }
-      }
-    }
-    return new Conv2dBackpropFilter<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param useCudnnOnGpu 
-   */
-  public static Options useCudnnOnGpu(Boolean useCudnnOnGpu) {
-    return new Options().useCudnnOnGpu(useCudnnOnGpu);
-  }
-  
-  /**
-   * @param explicitPaddings If `padding` is `"EXPLICIT"`, the list of explicit padding amounts. For the ith
-   * dimension, the amount of padding inserted before and after the dimension is
-   * `explicit_paddings[2 * i]` and `explicit_paddings[2 * i + 1]`, respectively. If
-   * `padding` is not `"EXPLICIT"`, `explicit_paddings` must be empty.
-   */
-  public static Options explicitPaddings(List<Long> explicitPaddings) {
-    return new Options().explicitPaddings(explicitPaddings);
-  }
-  
-  /**
-   * @param dataFormat Specify the data format of the input and output data. With the
-   * default format "NHWC", the data is stored in the order of:
-   *     [batch, in_height, in_width, in_channels].
-   * Alternatively, the format could be "NCHW", the data storage order of:
-   *     [batch, in_channels, in_height, in_width].
-   */
-  public static Options dataFormat(String dataFormat) {
-    return new Options().dataFormat(dataFormat);
-  }
-  
-  /**
-   * @param dilations 1-D tensor of length 4.  The dilation factor for each dimension of
-   * `input`. If set to k > 1, there will be k-1 skipped cells between each filter
-   * element on that dimension. The dimension order is determined by the value of
-   * `data_format`, see above for details. Dilations in the batch and depth
-   * dimensions must be 1.
-   */
-  public static Options dilations(List<Long> dilations) {
-    return new Options().dilations(dilations);
-  }
-  
-  /**
-   * 4-D with shape
-   * `[filter_height, filter_width, in_channels, out_channels]`.  Gradient w.r.t.
-   * the `filter` input of the convolution.
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Conv2DBackpropFilter";
-  
-  private Output<T> output;
-  
-  private Conv2dBackpropFilter(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Conv2dBackpropInput.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Conv2dBackpropInput.java
deleted file mode 100644
index 6e95b5906ff..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Conv2dBackpropInput.java
+++ /dev/null
@@ -1,216 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.nn;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Computes the gradients of convolution with respect to the input.
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator(group = "nn")
-public final class Conv2dBackpropInput<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.nn.Conv2dBackpropInput}
-   */
-  public static class Options {
-    
-    /**
-     * @param useCudnnOnGpu 
-     */
-    public Options useCudnnOnGpu(Boolean useCudnnOnGpu) {
-      this.useCudnnOnGpu = useCudnnOnGpu;
-      return this;
-    }
-    
-    /**
-     * @param explicitPaddings If `padding` is `"EXPLICIT"`, the list of explicit padding amounts. For the ith
-     * dimension, the amount of padding inserted before and after the dimension is
-     * `explicit_paddings[2 * i]` and `explicit_paddings[2 * i + 1]`, respectively. If
-     * `padding` is not `"EXPLICIT"`, `explicit_paddings` must be empty.
-     */
-    public Options explicitPaddings(List<Long> explicitPaddings) {
-      this.explicitPaddings = explicitPaddings;
-      return this;
-    }
-    
-    /**
-     * @param dataFormat Specify the data format of the input and output data. With the
-     * default format "NHWC", the data is stored in the order of:
-     *     [batch, in_height, in_width, in_channels].
-     * Alternatively, the format could be "NCHW", the data storage order of:
-     *     [batch, in_channels, in_height, in_width].
-     */
-    public Options dataFormat(String dataFormat) {
-      this.dataFormat = dataFormat;
-      return this;
-    }
-    
-    /**
-     * @param dilations 1-D tensor of length 4.  The dilation factor for each dimension of
-     * `input`. If set to k > 1, there will be k-1 skipped cells between each filter
-     * element on that dimension. The dimension order is determined by the value of
-     * `data_format`, see above for details. Dilations in the batch and depth
-     * dimensions must be 1.
-     */
-    public Options dilations(List<Long> dilations) {
-      this.dilations = dilations;
-      return this;
-    }
-    
-    private Boolean useCudnnOnGpu;
-    private List<Long> explicitPaddings;
-    private String dataFormat;
-    private List<Long> dilations;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new Conv2dBackpropInput operation.
-   * 
-   * @param scope current scope
-   * @param inputSizes An integer vector representing the shape of `input`,
-   * where `input` is a 4-D `[batch, height, width, channels]` tensor.
-   * @param filter 4-D with shape
-   * `[filter_height, filter_width, in_channels, out_channels]`.
-   * @param outBackprop 4-D with shape `[batch, out_height, out_width, out_channels]`.
-   * Gradients w.r.t. the output of the convolution.
-   * @param strides The stride of the sliding window for each dimension of the input
-   * of the convolution. Must be in the same order as the dimension specified with
-   * format.
-   * @param padding The type of padding algorithm to use.
-   * @param options carries optional attributes values
-   * @return a new instance of Conv2dBackpropInput
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> Conv2dBackpropInput<T> create(Scope scope, Operand<TInt32> inputSizes, Operand<T> filter, Operand<T> outBackprop, List<Long> strides, String padding, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Conv2DBackpropInput", scope.makeOpName("Conv2dBackpropInput"));
-    opBuilder.addInput(inputSizes.asOutput(scope));
-    opBuilder.addInput(filter.asOutput(scope));
-    opBuilder.addInput(outBackprop.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    long[] stridesArray = new long[strides.size()];
-    for (int i = 0; i < stridesArray.length; ++i) {
-      stridesArray[i] = strides.get(i);
-    }
-    opBuilder.setAttr("strides", stridesArray);
-    opBuilder.setAttr("padding", padding);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.useCudnnOnGpu != null) {
-          opBuilder.setAttr("use_cudnn_on_gpu", opts.useCudnnOnGpu);
-        }
-        if (opts.explicitPaddings != null) {
-          long[] explicitPaddingsArray = new long[opts.explicitPaddings.size()];
-          for (int i = 0; i < explicitPaddingsArray.length; ++i) {
-            explicitPaddingsArray[i] = opts.explicitPaddings.get(i);
-          }
-          opBuilder.setAttr("explicit_paddings", explicitPaddingsArray);
-        }
-        if (opts.dataFormat != null) {
-          opBuilder.setAttr("data_format", opts.dataFormat);
-        }
-        if (opts.dilations != null) {
-          long[] dilationsArray = new long[opts.dilations.size()];
-          for (int i = 0; i < dilationsArray.length; ++i) {
-            dilationsArray[i] = opts.dilations.get(i);
-          }
-          opBuilder.setAttr("dilations", dilationsArray);
-        }
-      }
-    }
-    return new Conv2dBackpropInput<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param useCudnnOnGpu 
-   */
-  public static Options useCudnnOnGpu(Boolean useCudnnOnGpu) {
-    return new Options().useCudnnOnGpu(useCudnnOnGpu);
-  }
-  
-  /**
-   * @param explicitPaddings If `padding` is `"EXPLICIT"`, the list of explicit padding amounts. For the ith
-   * dimension, the amount of padding inserted before and after the dimension is
-   * `explicit_paddings[2 * i]` and `explicit_paddings[2 * i + 1]`, respectively. If
-   * `padding` is not `"EXPLICIT"`, `explicit_paddings` must be empty.
-   */
-  public static Options explicitPaddings(List<Long> explicitPaddings) {
-    return new Options().explicitPaddings(explicitPaddings);
-  }
-  
-  /**
-   * @param dataFormat Specify the data format of the input and output data. With the
-   * default format "NHWC", the data is stored in the order of:
-   *     [batch, in_height, in_width, in_channels].
-   * Alternatively, the format could be "NCHW", the data storage order of:
-   *     [batch, in_channels, in_height, in_width].
-   */
-  public static Options dataFormat(String dataFormat) {
-    return new Options().dataFormat(dataFormat);
-  }
-  
-  /**
-   * @param dilations 1-D tensor of length 4.  The dilation factor for each dimension of
-   * `input`. If set to k > 1, there will be k-1 skipped cells between each filter
-   * element on that dimension. The dimension order is determined by the value of
-   * `data_format`, see above for details. Dilations in the batch and depth
-   * dimensions must be 1.
-   */
-  public static Options dilations(List<Long> dilations) {
-    return new Options().dilations(dilations);
-  }
-  
-  /**
-   * 4-D with shape `[batch, in_height, in_width, in_channels]`.  Gradient
-   * w.r.t. the input of the convolution.
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Conv2DBackpropInput";
-  
-  private Output<T> output;
-  
-  private Conv2dBackpropInput(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Conv3d.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Conv3d.java
deleted file mode 100644
index 20e44b309f6..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Conv3d.java
+++ /dev/null
@@ -1,166 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.nn;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Computes a 3-D convolution given 5-D `input` and `filter` tensors.
- * <p>
- * In signal processing, cross-correlation is a measure of similarity of
- * two waveforms as a function of a time-lag applied to one of them. This
- * is also known as a sliding dot product or sliding inner-product.
- * <p>
- * Our Conv3D implements a form of cross-correlation.
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator(group = "nn")
-public final class Conv3d<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.nn.Conv3d}
-   */
-  public static class Options {
-    
-    /**
-     * @param dataFormat The data format of the input and output data. With the
-     * default format "NDHWC", the data is stored in the order of:
-     *     [batch, in_depth, in_height, in_width, in_channels].
-     * Alternatively, the format could be "NCDHW", the data storage order is:
-     *     [batch, in_channels, in_depth, in_height, in_width].
-     */
-    public Options dataFormat(String dataFormat) {
-      this.dataFormat = dataFormat;
-      return this;
-    }
-    
-    /**
-     * @param dilations 1-D tensor of length 5.  The dilation factor for each dimension of
-     * `input`. If set to k > 1, there will be k-1 skipped cells between each
-     * filter element on that dimension. The dimension order is determined by the
-     * value of `data_format`, see above for details. Dilations in the batch and
-     * depth dimensions must be 1.
-     */
-    public Options dilations(List<Long> dilations) {
-      this.dilations = dilations;
-      return this;
-    }
-    
-    private String dataFormat;
-    private List<Long> dilations;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new Conv3d operation.
-   * 
-   * @param scope current scope
-   * @param input Shape `[batch, in_depth, in_height, in_width, in_channels]`.
-   * @param filter Shape `[filter_depth, filter_height, filter_width, in_channels,
-   * out_channels]`. `in_channels` must match between `input` and `filter`.
-   * @param strides 1-D tensor of length 5. The stride of the sliding window for each
-   * dimension of `input`. Must have `strides[0] = strides[4] = 1`.
-   * @param padding The type of padding algorithm to use.
-   * @param options carries optional attributes values
-   * @return a new instance of Conv3d
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> Conv3d<T> create(Scope scope, Operand<T> input, Operand<T> filter, List<Long> strides, String padding, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Conv3D", scope.makeOpName("Conv3d"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder.addInput(filter.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    long[] stridesArray = new long[strides.size()];
-    for (int i = 0; i < stridesArray.length; ++i) {
-      stridesArray[i] = strides.get(i);
-    }
-    opBuilder.setAttr("strides", stridesArray);
-    opBuilder.setAttr("padding", padding);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.dataFormat != null) {
-          opBuilder.setAttr("data_format", opts.dataFormat);
-        }
-        if (opts.dilations != null) {
-          long[] dilationsArray = new long[opts.dilations.size()];
-          for (int i = 0; i < dilationsArray.length; ++i) {
-            dilationsArray[i] = opts.dilations.get(i);
-          }
-          opBuilder.setAttr("dilations", dilationsArray);
-        }
-      }
-    }
-    return new Conv3d<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param dataFormat The data format of the input and output data. With the
-   * default format "NDHWC", the data is stored in the order of:
-   *     [batch, in_depth, in_height, in_width, in_channels].
-   * Alternatively, the format could be "NCDHW", the data storage order is:
-   *     [batch, in_channels, in_depth, in_height, in_width].
-   */
-  public static Options dataFormat(String dataFormat) {
-    return new Options().dataFormat(dataFormat);
-  }
-  
-  /**
-   * @param dilations 1-D tensor of length 5.  The dilation factor for each dimension of
-   * `input`. If set to k > 1, there will be k-1 skipped cells between each
-   * filter element on that dimension. The dimension order is determined by the
-   * value of `data_format`, see above for details. Dilations in the batch and
-   * depth dimensions must be 1.
-   */
-  public static Options dilations(List<Long> dilations) {
-    return new Options().dilations(dilations);
-  }
-  
-  /**
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Conv3D";
-  
-  private Output<T> output;
-  
-  private Conv3d(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Conv3dBackpropFilter.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Conv3dBackpropFilter.java
deleted file mode 100644
index 94c4dbd8a01..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Conv3dBackpropFilter.java
+++ /dev/null
@@ -1,166 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.nn;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Computes the gradients of 3-D convolution with respect to the filter.
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator(group = "nn")
-public final class Conv3dBackpropFilter<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.nn.Conv3dBackpropFilter}
-   */
-  public static class Options {
-    
-    /**
-     * @param dataFormat The data format of the input and output data. With the
-     * default format "NDHWC", the data is stored in the order of:
-     *     [batch, in_depth, in_height, in_width, in_channels].
-     * Alternatively, the format could be "NCDHW", the data storage order is:
-     *     [batch, in_channels, in_depth, in_height, in_width].
-     */
-    public Options dataFormat(String dataFormat) {
-      this.dataFormat = dataFormat;
-      return this;
-    }
-    
-    /**
-     * @param dilations 1-D tensor of length 5.  The dilation factor for each dimension of
-     * `input`. If set to k > 1, there will be k-1 skipped cells between each
-     * filter element on that dimension. The dimension order is determined by the
-     * value of `data_format`, see above for details. Dilations in the batch and
-     * depth dimensions must be 1.
-     */
-    public Options dilations(List<Long> dilations) {
-      this.dilations = dilations;
-      return this;
-    }
-    
-    private String dataFormat;
-    private List<Long> dilations;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new Conv3dBackpropFilter operation.
-   * 
-   * @param scope current scope
-   * @param input Shape `[batch, depth, rows, cols, in_channels]`.
-   * @param filterSizes An integer vector representing the tensor shape of `filter`,
-   * where `filter` is a 5-D
-   * `[filter_depth, filter_height, filter_width, in_channels, out_channels]`
-   * tensor.
-   * @param outBackprop Backprop signal of shape `[batch, out_depth, out_rows, out_cols,
-   * out_channels]`.
-   * @param strides 1-D tensor of length 5. The stride of the sliding window for each
-   * dimension of `input`. Must have `strides[0] = strides[4] = 1`.
-   * @param padding The type of padding algorithm to use.
-   * @param options carries optional attributes values
-   * @return a new instance of Conv3dBackpropFilter
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> Conv3dBackpropFilter<T> create(Scope scope, Operand<T> input, Operand<TInt32> filterSizes, Operand<T> outBackprop, List<Long> strides, String padding, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Conv3DBackpropFilterV2", scope.makeOpName("Conv3dBackpropFilter"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder.addInput(filterSizes.asOutput(scope));
-    opBuilder.addInput(outBackprop.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    long[] stridesArray = new long[strides.size()];
-    for (int i = 0; i < stridesArray.length; ++i) {
-      stridesArray[i] = strides.get(i);
-    }
-    opBuilder.setAttr("strides", stridesArray);
-    opBuilder.setAttr("padding", padding);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.dataFormat != null) {
-          opBuilder.setAttr("data_format", opts.dataFormat);
-        }
-        if (opts.dilations != null) {
-          long[] dilationsArray = new long[opts.dilations.size()];
-          for (int i = 0; i < dilationsArray.length; ++i) {
-            dilationsArray[i] = opts.dilations.get(i);
-          }
-          opBuilder.setAttr("dilations", dilationsArray);
-        }
-      }
-    }
-    return new Conv3dBackpropFilter<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param dataFormat The data format of the input and output data. With the
-   * default format "NDHWC", the data is stored in the order of:
-   *     [batch, in_depth, in_height, in_width, in_channels].
-   * Alternatively, the format could be "NCDHW", the data storage order is:
-   *     [batch, in_channels, in_depth, in_height, in_width].
-   */
-  public static Options dataFormat(String dataFormat) {
-    return new Options().dataFormat(dataFormat);
-  }
-  
-  /**
-   * @param dilations 1-D tensor of length 5.  The dilation factor for each dimension of
-   * `input`. If set to k > 1, there will be k-1 skipped cells between each
-   * filter element on that dimension. The dimension order is determined by the
-   * value of `data_format`, see above for details. Dilations in the batch and
-   * depth dimensions must be 1.
-   */
-  public static Options dilations(List<Long> dilations) {
-    return new Options().dilations(dilations);
-  }
-  
-  /**
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Conv3DBackpropFilterV2";
-  
-  private Output<T> output;
-  
-  private Conv3dBackpropFilter(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Conv3dBackpropInput.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Conv3dBackpropInput.java
deleted file mode 100644
index 102305387a7..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Conv3dBackpropInput.java
+++ /dev/null
@@ -1,165 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.nn;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Computes the gradients of 3-D convolution with respect to the input.
- * 
- * @param <U> data type for {@code output()} output
- */
-@Operator(group = "nn")
-public final class Conv3dBackpropInput<U extends TNumber> extends RawOp implements Operand<U> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.nn.Conv3dBackpropInput}
-   */
-  public static class Options {
-    
-    /**
-     * @param dataFormat The data format of the input and output data. With the
-     * default format "NDHWC", the data is stored in the order of:
-     *     [batch, in_depth, in_height, in_width, in_channels].
-     * Alternatively, the format could be "NCDHW", the data storage order is:
-     *     [batch, in_channels, in_depth, in_height, in_width].
-     */
-    public Options dataFormat(String dataFormat) {
-      this.dataFormat = dataFormat;
-      return this;
-    }
-    
-    /**
-     * @param dilations 1-D tensor of length 5.  The dilation factor for each dimension of
-     * `input`. If set to k > 1, there will be k-1 skipped cells between each
-     * filter element on that dimension. The dimension order is determined by the
-     * value of `data_format`, see above for details. Dilations in the batch and
-     * depth dimensions must be 1.
-     */
-    public Options dilations(List<Long> dilations) {
-      this.dilations = dilations;
-      return this;
-    }
-    
-    private String dataFormat;
-    private List<Long> dilations;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new Conv3dBackpropInput operation.
-   * 
-   * @param scope current scope
-   * @param inputSizes An integer vector representing the tensor shape of `input`,
-   * where `input` is a 5-D
-   * `[batch, depth, rows, cols, in_channels]` tensor.
-   * @param filter Shape `[depth, rows, cols, in_channels, out_channels]`.
-   * `in_channels` must match between `input` and `filter`.
-   * @param outBackprop Backprop signal of shape `[batch, out_depth, out_rows, out_cols,
-   * out_channels]`.
-   * @param strides 1-D tensor of length 5. The stride of the sliding window for each
-   * dimension of `input`. Must have `strides[0] = strides[4] = 1`.
-   * @param padding The type of padding algorithm to use.
-   * @param options carries optional attributes values
-   * @return a new instance of Conv3dBackpropInput
-   */
-  @Endpoint(describeByClass = true)
-  public static <U extends TNumber, T extends TNumber> Conv3dBackpropInput<U> create(Scope scope, Operand<T> inputSizes, Operand<U> filter, Operand<U> outBackprop, List<Long> strides, String padding, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Conv3DBackpropInputV2", scope.makeOpName("Conv3dBackpropInput"));
-    opBuilder.addInput(inputSizes.asOutput(scope));
-    opBuilder.addInput(filter.asOutput(scope));
-    opBuilder.addInput(outBackprop.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    long[] stridesArray = new long[strides.size()];
-    for (int i = 0; i < stridesArray.length; ++i) {
-      stridesArray[i] = strides.get(i);
-    }
-    opBuilder.setAttr("strides", stridesArray);
-    opBuilder.setAttr("padding", padding);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.dataFormat != null) {
-          opBuilder.setAttr("data_format", opts.dataFormat);
-        }
-        if (opts.dilations != null) {
-          long[] dilationsArray = new long[opts.dilations.size()];
-          for (int i = 0; i < dilationsArray.length; ++i) {
-            dilationsArray[i] = opts.dilations.get(i);
-          }
-          opBuilder.setAttr("dilations", dilationsArray);
-        }
-      }
-    }
-    return new Conv3dBackpropInput<U>(opBuilder.build());
-  }
-  
-  /**
-   * @param dataFormat The data format of the input and output data. With the
-   * default format "NDHWC", the data is stored in the order of:
-   *     [batch, in_depth, in_height, in_width, in_channels].
-   * Alternatively, the format could be "NCDHW", the data storage order is:
-   *     [batch, in_channels, in_depth, in_height, in_width].
-   */
-  public static Options dataFormat(String dataFormat) {
-    return new Options().dataFormat(dataFormat);
-  }
-  
-  /**
-   * @param dilations 1-D tensor of length 5.  The dilation factor for each dimension of
-   * `input`. If set to k > 1, there will be k-1 skipped cells between each
-   * filter element on that dimension. The dimension order is determined by the
-   * value of `data_format`, see above for details. Dilations in the batch and
-   * depth dimensions must be 1.
-   */
-  public static Options dilations(List<Long> dilations) {
-    return new Options().dilations(dilations);
-  }
-  
-  /**
-   */
-  public Output<U> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<U> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Conv3DBackpropInputV2";
-  
-  private Output<U> output;
-  
-  private Conv3dBackpropInput(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CtcBeamSearchDecoder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CtcBeamSearchDecoder.java
deleted file mode 100644
index b9952e52777..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CtcBeamSearchDecoder.java
+++ /dev/null
@@ -1,161 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.nn;
-
-import java.util.Arrays;
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Performs beam search decoding on the logits given in input.
- * <p>
- * A note about the attribute merge_repeated: For the beam search decoder,
- * this means that if consecutive entries in a beam are the same, only
- * the first of these is emitted.  That is, when the top path is "A B B B B",
- * "A B" is returned if merge_repeated = True but "A B B B B" is
- * returned if merge_repeated = False.
- * 
- * @param <T> data type for {@code logProbability()} output
- */
-@Operator(group = "nn")
-public final class CtcBeamSearchDecoder<T extends TNumber> extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.nn.CtcBeamSearchDecoder}
-   */
-  public static class Options {
-    
-    /**
-     * @param mergeRepeated If true, merge repeated classes in output.
-     */
-    public Options mergeRepeated(Boolean mergeRepeated) {
-      this.mergeRepeated = mergeRepeated;
-      return this;
-    }
-    
-    private Boolean mergeRepeated;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new CtcBeamSearchDecoder operation.
-   * 
-   * @param scope current scope
-   * @param inputs 3-D, shape: `(max_time x batch_size x num_classes)`, the logits.
-   * @param sequenceLength A vector containing sequence lengths, size `(batch)`.
-   * @param beamWidth A scalar >= 0 (beam search beam width).
-   * @param topPaths A scalar >= 0, <= beam_width (controls output size).
-   * @param options carries optional attributes values
-   * @return a new instance of CtcBeamSearchDecoder
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> CtcBeamSearchDecoder<T> create(Scope scope, Operand<T> inputs, Operand<TInt32> sequenceLength, Long beamWidth, Long topPaths, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("CTCBeamSearchDecoder", scope.makeOpName("CtcBeamSearchDecoder"));
-    opBuilder.addInput(inputs.asOutput(scope));
-    opBuilder.addInput(sequenceLength.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("beam_width", beamWidth);
-    opBuilder.setAttr("top_paths", topPaths);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.mergeRepeated != null) {
-          opBuilder.setAttr("merge_repeated", opts.mergeRepeated);
-        }
-      }
-    }
-    return new CtcBeamSearchDecoder<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param mergeRepeated If true, merge repeated classes in output.
-   */
-  public static Options mergeRepeated(Boolean mergeRepeated) {
-    return new Options().mergeRepeated(mergeRepeated);
-  }
-  
-  /**
-   * A list (length: top_paths) of indices matrices.  Matrix j,
-   * size `(total_decoded_outputs[j] x 2)`, has indices of a
-   * `SparseTensor<int64, 2>`.  The rows store: [batch, time].
-   */
-  public List<Output<TInt64>> decodedIndices() {
-    return decodedIndices;
-  }
-  
-  /**
-   * A list (length: top_paths) of values vectors.  Vector j,
-   * size `(length total_decoded_outputs[j])`, has the values of a
-   * `SparseTensor<int64, 2>`.  The vector stores the decoded classes for beam j.
-   */
-  public List<Output<TInt64>> decodedValues() {
-    return decodedValues;
-  }
-  
-  /**
-   * A list (length: top_paths) of shape vector.  Vector j,
-   * size `(2)`, stores the shape of the decoded `SparseTensor[j]`.
-   * Its values are: `[batch_size, max_decoded_length[j]]`.
-   */
-  public List<Output<TInt64>> decodedShape() {
-    return decodedShape;
-  }
-  
-  /**
-   * A matrix, shaped: `(batch_size x top_paths)`.  The
-   * sequence log-probabilities.
-   */
-  public Output<T> logProbability() {
-    return logProbability;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "CTCBeamSearchDecoder";
-  
-  private List<Output<TInt64>> decodedIndices;
-  private List<Output<TInt64>> decodedValues;
-  private List<Output<TInt64>> decodedShape;
-  private Output<T> logProbability;
-  
-  @SuppressWarnings("unchecked")
-  private CtcBeamSearchDecoder(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    int decodedIndicesLength = operation.outputListLength("decoded_indices");
-    decodedIndices = Arrays.asList((Output<TInt64>[])operation.outputList(outputIdx, decodedIndicesLength));
-    outputIdx += decodedIndicesLength;
-    int decodedValuesLength = operation.outputListLength("decoded_values");
-    decodedValues = Arrays.asList((Output<TInt64>[])operation.outputList(outputIdx, decodedValuesLength));
-    outputIdx += decodedValuesLength;
-    int decodedShapeLength = operation.outputListLength("decoded_shape");
-    decodedShape = Arrays.asList((Output<TInt64>[])operation.outputList(outputIdx, decodedShapeLength));
-    outputIdx += decodedShapeLength;
-    logProbability = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CtcGreedyDecoder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CtcGreedyDecoder.java
deleted file mode 100644
index 8ba71cd466e..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CtcGreedyDecoder.java
+++ /dev/null
@@ -1,149 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.nn;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Performs greedy decoding on the logits given in inputs.
- * <p>
- * A note about the attribute merge_repeated: if enabled, when
- * consecutive logits' maximum indices are the same, only the first of
- * these is emitted.  Labeling the blank '*', the sequence "A B B * B B"
- * becomes "A B B" if merge_repeated = True and "A B B B B" if
- * merge_repeated = False.
- * <p>
- * Regardless of the value of merge_repeated, if the maximum index of a given
- * time and batch corresponds to the blank, index `(num_classes - 1)`, no new
- * element is emitted.
- * 
- * @param <T> data type for {@code logProbability()} output
- */
-@Operator(group = "nn")
-public final class CtcGreedyDecoder<T extends TNumber> extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.nn.CtcGreedyDecoder}
-   */
-  public static class Options {
-    
-    /**
-     * @param mergeRepeated If True, merge repeated classes in output.
-     */
-    public Options mergeRepeated(Boolean mergeRepeated) {
-      this.mergeRepeated = mergeRepeated;
-      return this;
-    }
-    
-    private Boolean mergeRepeated;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new CtcGreedyDecoder operation.
-   * 
-   * @param scope current scope
-   * @param inputs 3-D, shape: `(max_time x batch_size x num_classes)`, the logits.
-   * @param sequenceLength A vector containing sequence lengths, size `(batch_size)`.
-   * @param options carries optional attributes values
-   * @return a new instance of CtcGreedyDecoder
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> CtcGreedyDecoder<T> create(Scope scope, Operand<T> inputs, Operand<TInt32> sequenceLength, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("CTCGreedyDecoder", scope.makeOpName("CtcGreedyDecoder"));
-    opBuilder.addInput(inputs.asOutput(scope));
-    opBuilder.addInput(sequenceLength.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.mergeRepeated != null) {
-          opBuilder.setAttr("merge_repeated", opts.mergeRepeated);
-        }
-      }
-    }
-    return new CtcGreedyDecoder<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param mergeRepeated If True, merge repeated classes in output.
-   */
-  public static Options mergeRepeated(Boolean mergeRepeated) {
-    return new Options().mergeRepeated(mergeRepeated);
-  }
-  
-  /**
-   * Indices matrix, size `(total_decoded_outputs x 2)`,
-   * of a `SparseTensor<int64, 2>`.  The rows store: [batch, time].
-   */
-  public Output<TInt64> decodedIndices() {
-    return decodedIndices;
-  }
-  
-  /**
-   * Values vector, size: `(total_decoded_outputs)`,
-   * of a `SparseTensor<int64, 2>`.  The vector stores the decoded classes.
-   */
-  public Output<TInt64> decodedValues() {
-    return decodedValues;
-  }
-  
-  /**
-   * Shape vector, size `(2)`, of the decoded SparseTensor.
-   * Values are: `[batch_size, max_decoded_length]`.
-   */
-  public Output<TInt64> decodedShape() {
-    return decodedShape;
-  }
-  
-  /**
-   * Matrix, size `(batch_size x 1)`, containing sequence
-   * log-probabilities.
-   */
-  public Output<T> logProbability() {
-    return logProbability;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "CTCGreedyDecoder";
-  
-  private Output<TInt64> decodedIndices;
-  private Output<TInt64> decodedValues;
-  private Output<TInt64> decodedShape;
-  private Output<T> logProbability;
-  
-  private CtcGreedyDecoder(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    decodedIndices = operation.output(outputIdx++);
-    decodedValues = operation.output(outputIdx++);
-    decodedShape = operation.output(outputIdx++);
-    logProbability = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CtcLoss.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CtcLoss.java
deleted file mode 100644
index db4df7fe909..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CtcLoss.java
+++ /dev/null
@@ -1,175 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.nn;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Calculates the CTC Loss (log probability) for each batch entry.  Also calculates
- * <p>
- * the gradient.  This class performs the softmax operation for you, so inputs
- * should be e.g. linear projections of outputs by an LSTM.
- * 
- * @param <T> data type for {@code loss()} output
- */
-@Operator(group = "nn")
-public final class CtcLoss<T extends TNumber> extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.nn.CtcLoss}
-   */
-  public static class Options {
-    
-    /**
-     * @param preprocessCollapseRepeated Scalar, if true then repeated labels are
-     * collapsed prior to the CTC calculation.
-     */
-    public Options preprocessCollapseRepeated(Boolean preprocessCollapseRepeated) {
-      this.preprocessCollapseRepeated = preprocessCollapseRepeated;
-      return this;
-    }
-    
-    /**
-     * @param ctcMergeRepeated Scalar.  If set to false, <i>during</i> CTC calculation
-     * repeated non-blank labels will not be merged and are interpreted as
-     * individual labels.  This is a simplified version of CTC.
-     */
-    public Options ctcMergeRepeated(Boolean ctcMergeRepeated) {
-      this.ctcMergeRepeated = ctcMergeRepeated;
-      return this;
-    }
-    
-    /**
-     * @param ignoreLongerOutputsThanInputs Scalar. If set to true, during CTC
-     * calculation, items that have longer output sequences than input sequences
-     * are skipped: they don't contribute to the loss term and have zero-gradient.
-     */
-    public Options ignoreLongerOutputsThanInputs(Boolean ignoreLongerOutputsThanInputs) {
-      this.ignoreLongerOutputsThanInputs = ignoreLongerOutputsThanInputs;
-      return this;
-    }
-    
-    private Boolean preprocessCollapseRepeated;
-    private Boolean ctcMergeRepeated;
-    private Boolean ignoreLongerOutputsThanInputs;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new CtcLoss operation.
-   * 
-   * @param scope current scope
-   * @param inputs 3-D, shape: `(max_time x batch_size x num_classes)`, the logits.
-   * @param labelsIndices The indices of a `SparseTensor<int32, 2>`.
-   * `labels_indices(i, :) == [b, t]` means `labels_values(i)` stores the id for
-   * `(batch b, time t)`.
-   * @param labelsValues The values (labels) associated with the given batch and time.
-   * @param sequenceLength A vector containing sequence lengths (batch).
-   * @param options carries optional attributes values
-   * @return a new instance of CtcLoss
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> CtcLoss<T> create(Scope scope, Operand<T> inputs, Operand<TInt64> labelsIndices, Operand<TInt32> labelsValues, Operand<TInt32> sequenceLength, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("CTCLoss", scope.makeOpName("CtcLoss"));
-    opBuilder.addInput(inputs.asOutput(scope));
-    opBuilder.addInput(labelsIndices.asOutput(scope));
-    opBuilder.addInput(labelsValues.asOutput(scope));
-    opBuilder.addInput(sequenceLength.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.preprocessCollapseRepeated != null) {
-          opBuilder.setAttr("preprocess_collapse_repeated", opts.preprocessCollapseRepeated);
-        }
-        if (opts.ctcMergeRepeated != null) {
-          opBuilder.setAttr("ctc_merge_repeated", opts.ctcMergeRepeated);
-        }
-        if (opts.ignoreLongerOutputsThanInputs != null) {
-          opBuilder.setAttr("ignore_longer_outputs_than_inputs", opts.ignoreLongerOutputsThanInputs);
-        }
-      }
-    }
-    return new CtcLoss<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param preprocessCollapseRepeated Scalar, if true then repeated labels are
-   * collapsed prior to the CTC calculation.
-   */
-  public static Options preprocessCollapseRepeated(Boolean preprocessCollapseRepeated) {
-    return new Options().preprocessCollapseRepeated(preprocessCollapseRepeated);
-  }
-  
-  /**
-   * @param ctcMergeRepeated Scalar.  If set to false, <i>during</i> CTC calculation
-   * repeated non-blank labels will not be merged and are interpreted as
-   * individual labels.  This is a simplified version of CTC.
-   */
-  public static Options ctcMergeRepeated(Boolean ctcMergeRepeated) {
-    return new Options().ctcMergeRepeated(ctcMergeRepeated);
-  }
-  
-  /**
-   * @param ignoreLongerOutputsThanInputs Scalar. If set to true, during CTC
-   * calculation, items that have longer output sequences than input sequences
-   * are skipped: they don't contribute to the loss term and have zero-gradient.
-   */
-  public static Options ignoreLongerOutputsThanInputs(Boolean ignoreLongerOutputsThanInputs) {
-    return new Options().ignoreLongerOutputsThanInputs(ignoreLongerOutputsThanInputs);
-  }
-  
-  /**
-   * A vector (batch) containing log-probabilities.
-   */
-  public Output<T> loss() {
-    return loss;
-  }
-  
-  /**
-   * The gradient of `loss`.  3-D, shape:
-   * `(max_time x batch_size x num_classes)`.
-   */
-  public Output<T> gradient() {
-    return gradient;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "CTCLoss";
-  
-  private Output<T> loss;
-  private Output<T> gradient;
-  
-  private CtcLoss(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    loss = operation.output(outputIdx++);
-    gradient = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CudnnRNN.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CudnnRNN.java
deleted file mode 100644
index a3fc18e8ec5..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CudnnRNN.java
+++ /dev/null
@@ -1,333 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.nn;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * A RNN backed by cuDNN.
- * <p>
- * Computes the RNN from the input and initial states, with respect to the params
- * buffer. Accepts one extra input "sequence_lengths" than CudnnRNN.
- * <p>
- * rnn_mode: Indicates the type of the RNN model.
- * input_mode: Indicates whether there is a linear projection between the input and
- *   the actual computation before the first layer. 'skip_input' is only allowed
- *   when input_size == num_units; 'auto_select' implies 'skip_input' when
- *   input_size == num_units; otherwise, it implies 'linear_input'.
- * direction: Indicates whether a bidirectional model will be used. Should be
- *   "unidirectional" or "bidirectional".
- * dropout: Dropout probability. When set to 0., dropout is disabled.
- * seed: The 1st part of a seed to initialize dropout.
- * seed2: The 2nd part of a seed to initialize dropout.
- * input: If time_major is true, this is a 3-D tensor with the shape of
- *     [seq_length, batch_size, input_size]. If time_major is false, the shape is
- *     [batch_size, seq_length, input_size].
- * input_h: If time_major is true, this is a 3-D tensor with the shape of
- *     [num_layer * dir, batch_size, num_units]. If time_major is false, the shape
- *     is [batch_size, num_layer * dir, num_units].
- * input_c: For LSTM, a 3-D tensor with the shape of
- *     [num_layer * dir, batch, num_units]. For other models, it is ignored.
- * params: A 1-D tensor that contains the weights and biases in an opaque layout.
- *     The size must be created through CudnnRNNParamsSize, and initialized
- *     separately. Note that they might not be compatible across different
- *     generations. So it is a good idea to save and restore
- * sequence_lengths: a vector of lengths of each input sequence.
- * output: If time_major is true, this is a 3-D tensor with the shape of
- *     [seq_length, batch_size, dir * num_units]. If time_major is false, the
- *     shape is [batch_size, seq_length, dir * num_units].
- * output_h: The same shape has input_h.
- * output_c: The same shape as input_c for LSTM. An empty tensor for other models.
- * is_training: Indicates whether this operation is used for inference or
- *   training.
- * time_major: Indicates whether the input/output format is time major or batch
- *     major.
- * reserve_space: An opaque tensor that can be used in backprop calculation. It
- *   is only produced if is_training is true.
- * 
- * @param <T> data type for {@code output()} output
- */
-public final class CudnnRNN<T extends TNumber> extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.nn.CudnnRNN}
-   */
-  public static class Options {
-    
-    /**
-     * @param rnnMode 
-     */
-    public Options rnnMode(String rnnMode) {
-      this.rnnMode = rnnMode;
-      return this;
-    }
-    
-    /**
-     * @param inputMode 
-     */
-    public Options inputMode(String inputMode) {
-      this.inputMode = inputMode;
-      return this;
-    }
-    
-    /**
-     * @param direction 
-     */
-    public Options direction(String direction) {
-      this.direction = direction;
-      return this;
-    }
-    
-    /**
-     * @param dropout 
-     */
-    public Options dropout(Float dropout) {
-      this.dropout = dropout;
-      return this;
-    }
-    
-    /**
-     * @param seed 
-     */
-    public Options seed(Long seed) {
-      this.seed = seed;
-      return this;
-    }
-    
-    /**
-     * @param seed2 
-     */
-    public Options seed2(Long seed2) {
-      this.seed2 = seed2;
-      return this;
-    }
-    
-    /**
-     * @param numProj 
-     */
-    public Options numProj(Long numProj) {
-      this.numProj = numProj;
-      return this;
-    }
-    
-    /**
-     * @param isTraining 
-     */
-    public Options isTraining(Boolean isTraining) {
-      this.isTraining = isTraining;
-      return this;
-    }
-    
-    /**
-     * @param timeMajor 
-     */
-    public Options timeMajor(Boolean timeMajor) {
-      this.timeMajor = timeMajor;
-      return this;
-    }
-    
-    private String rnnMode;
-    private String inputMode;
-    private String direction;
-    private Float dropout;
-    private Long seed;
-    private Long seed2;
-    private Long numProj;
-    private Boolean isTraining;
-    private Boolean timeMajor;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new CudnnRNN operation.
-   * 
-   * @param scope current scope
-   * @param input 
-   * @param inputH 
-   * @param inputC 
-   * @param params 
-   * @param sequenceLengths 
-   * @param options carries optional attributes values
-   * @return a new instance of CudnnRNN
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> CudnnRNN<T> create(Scope scope, Operand<T> input, Operand<T> inputH, Operand<T> inputC, Operand<T> params, Operand<TInt32> sequenceLengths, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("CudnnRNNV3", scope.makeOpName("CudnnRNN"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder.addInput(inputH.asOutput(scope));
-    opBuilder.addInput(inputC.asOutput(scope));
-    opBuilder.addInput(params.asOutput(scope));
-    opBuilder.addInput(sequenceLengths.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.rnnMode != null) {
-          opBuilder.setAttr("rnn_mode", opts.rnnMode);
-        }
-        if (opts.inputMode != null) {
-          opBuilder.setAttr("input_mode", opts.inputMode);
-        }
-        if (opts.direction != null) {
-          opBuilder.setAttr("direction", opts.direction);
-        }
-        if (opts.dropout != null) {
-          opBuilder.setAttr("dropout", opts.dropout);
-        }
-        if (opts.seed != null) {
-          opBuilder.setAttr("seed", opts.seed);
-        }
-        if (opts.seed2 != null) {
-          opBuilder.setAttr("seed2", opts.seed2);
-        }
-        if (opts.numProj != null) {
-          opBuilder.setAttr("num_proj", opts.numProj);
-        }
-        if (opts.isTraining != null) {
-          opBuilder.setAttr("is_training", opts.isTraining);
-        }
-        if (opts.timeMajor != null) {
-          opBuilder.setAttr("time_major", opts.timeMajor);
-        }
-      }
-    }
-    return new CudnnRNN<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param rnnMode 
-   */
-  public static Options rnnMode(String rnnMode) {
-    return new Options().rnnMode(rnnMode);
-  }
-  
-  /**
-   * @param inputMode 
-   */
-  public static Options inputMode(String inputMode) {
-    return new Options().inputMode(inputMode);
-  }
-  
-  /**
-   * @param direction 
-   */
-  public static Options direction(String direction) {
-    return new Options().direction(direction);
-  }
-  
-  /**
-   * @param dropout 
-   */
-  public static Options dropout(Float dropout) {
-    return new Options().dropout(dropout);
-  }
-  
-  /**
-   * @param seed 
-   */
-  public static Options seed(Long seed) {
-    return new Options().seed(seed);
-  }
-  
-  /**
-   * @param seed2 
-   */
-  public static Options seed2(Long seed2) {
-    return new Options().seed2(seed2);
-  }
-  
-  /**
-   * @param numProj 
-   */
-  public static Options numProj(Long numProj) {
-    return new Options().numProj(numProj);
-  }
-  
-  /**
-   * @param isTraining 
-   */
-  public static Options isTraining(Boolean isTraining) {
-    return new Options().isTraining(isTraining);
-  }
-  
-  /**
-   * @param timeMajor 
-   */
-  public static Options timeMajor(Boolean timeMajor) {
-    return new Options().timeMajor(timeMajor);
-  }
-  
-  /**
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  /**
-   */
-  public Output<T> outputH() {
-    return outputH;
-  }
-  
-  /**
-   */
-  public Output<T> outputC() {
-    return outputC;
-  }
-  
-  /**
-   */
-  public Output<T> reserveSpace() {
-    return reserveSpace;
-  }
-  
-  /**
-   */
-  public Output<?> hostReserved() {
-    return hostReserved;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "CudnnRNNV3";
-  
-  private Output<T> output;
-  private Output<T> outputH;
-  private Output<T> outputC;
-  private Output<T> reserveSpace;
-  private Output<?> hostReserved;
-  
-  private CudnnRNN(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-    outputH = operation.output(outputIdx++);
-    outputC = operation.output(outputIdx++);
-    reserveSpace = operation.output(outputIdx++);
-    hostReserved = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CudnnRNNBackprop.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CudnnRNNBackprop.java
deleted file mode 100644
index 53c32de11cc..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CudnnRNNBackprop.java
+++ /dev/null
@@ -1,332 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.nn;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Backprop step of CudnnRNNV3.
- * <p>
- * Compute the backprop of both data and weights in a RNN. Takes an extra
- *     "sequence_lengths" input than CudnnRNNBackprop.
- * <p>
- * rnn_mode: Indicates the type of the RNN model.
- * input_mode: Indicates whether there is a linear projection between the input and
- *     the actual computation before the first layer. 'skip_input' is only allowed
- *     when input_size == num_units; 'auto_select' implies 'skip_input' when
- *     input_size == num_units; otherwise, it implies 'linear_input'.
- * direction: Indicates whether a bidirectional model will be used. Should be
- *   "unidirectional" or "bidirectional".
- * dropout: Dropout probability. When set to 0., dropout is disabled.
- * seed: The 1st part of a seed to initialize dropout.
- * seed2: The 2nd part of a seed to initialize dropout.
- * input: If time_major is true, this is a 3-D tensor with the shape of
- *     [seq_length, batch_size, input_size]. If time_major is false, the shape is
- *     [batch_size, seq_length, input_size].
- * input_h: If time_major is true, this is a 3-D tensor with the shape of
- *     [num_layer * dir, batch_size, num_units]. If time_major is false, the shape
- *     is [batch_size, num_layer * dir, num_units].
- * input_c: For LSTM, a 3-D tensor with the shape of
- *     [num_layer * dir, batch, num_units]. For other models, it is ignored.
- * params: A 1-D tensor that contains the weights and biases in an opaque layout.
- *     The size must be created through CudnnRNNParamsSize, and initialized
- *     separately. Note that they might not be compatible across different
- *     generations. So it is a good idea to save and restore
- * sequence_lengths: a vector of lengths of each input sequence.
- * output: If time_major is true, this is a 3-D tensor with the shape of
- *     [seq_length, batch_size, dir * num_units]. If time_major is false, the
- *     shape is [batch_size, seq_length, dir * num_units].
- * output_h: The same shape has input_h.
- * output_c: The same shape as input_c for LSTM. An empty tensor for other models.
- * output_backprop: A 3-D tensor with the same shape as output in the forward pass.
- * output_h_backprop: A 3-D tensor with the same shape as output_h in the forward
- *     pass.
- * output_c_backprop: A 3-D tensor with the same shape as output_c in the forward
- *     pass.
- * time_major: Indicates whether the input/output format is time major or batch
- *     major.
- * reserve_space: The same reserve_space produced in the forward operation.
- * input_backprop: The backprop to input in the forward pass. Has the same shape
- *     as input.
- * input_h_backprop: The backprop to input_h in the forward pass. Has the same
- *     shape as input_h.
- * input_c_backprop: The backprop to input_c in the forward pass. Has the same
- *     shape as input_c.
- * params_backprop: The backprop to the params buffer in the forward pass. Has the
- *     same shape as params.
- * 
- * @param <T> data type for {@code inputBackprop()} output
- */
-public final class CudnnRNNBackprop<T extends TNumber> extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.nn.CudnnRNNBackprop}
-   */
-  public static class Options {
-    
-    /**
-     * @param rnnMode 
-     */
-    public Options rnnMode(String rnnMode) {
-      this.rnnMode = rnnMode;
-      return this;
-    }
-    
-    /**
-     * @param inputMode 
-     */
-    public Options inputMode(String inputMode) {
-      this.inputMode = inputMode;
-      return this;
-    }
-    
-    /**
-     * @param direction 
-     */
-    public Options direction(String direction) {
-      this.direction = direction;
-      return this;
-    }
-    
-    /**
-     * @param dropout 
-     */
-    public Options dropout(Float dropout) {
-      this.dropout = dropout;
-      return this;
-    }
-    
-    /**
-     * @param seed 
-     */
-    public Options seed(Long seed) {
-      this.seed = seed;
-      return this;
-    }
-    
-    /**
-     * @param seed2 
-     */
-    public Options seed2(Long seed2) {
-      this.seed2 = seed2;
-      return this;
-    }
-    
-    /**
-     * @param numProj 
-     */
-    public Options numProj(Long numProj) {
-      this.numProj = numProj;
-      return this;
-    }
-    
-    /**
-     * @param timeMajor 
-     */
-    public Options timeMajor(Boolean timeMajor) {
-      this.timeMajor = timeMajor;
-      return this;
-    }
-    
-    private String rnnMode;
-    private String inputMode;
-    private String direction;
-    private Float dropout;
-    private Long seed;
-    private Long seed2;
-    private Long numProj;
-    private Boolean timeMajor;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new CudnnRNNBackprop operation.
-   * 
-   * @param scope current scope
-   * @param input 
-   * @param inputH 
-   * @param inputC 
-   * @param params 
-   * @param sequenceLengths 
-   * @param output 
-   * @param outputH 
-   * @param outputC 
-   * @param outputBackprop 
-   * @param outputHBackprop 
-   * @param outputCBackprop 
-   * @param reserveSpace 
-   * @param hostReserved 
-   * @param options carries optional attributes values
-   * @return a new instance of CudnnRNNBackprop
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> CudnnRNNBackprop<T> create(Scope scope, Operand<T> input, Operand<T> inputH, Operand<T> inputC, Operand<T> params, Operand<TInt32> sequenceLengths, Operand<T> output, Operand<T> outputH, Operand<T> outputC, Operand<T> outputBackprop, Operand<T> outputHBackprop, Operand<T> outputCBackprop, Operand<T> reserveSpace, Operand<?> hostReserved, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("CudnnRNNBackpropV3", scope.makeOpName("CudnnRNNBackprop"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder.addInput(inputH.asOutput(scope));
-    opBuilder.addInput(inputC.asOutput(scope));
-    opBuilder.addInput(params.asOutput(scope));
-    opBuilder.addInput(sequenceLengths.asOutput(scope));
-    opBuilder.addInput(output.asOutput(scope));
-    opBuilder.addInput(outputH.asOutput(scope));
-    opBuilder.addInput(outputC.asOutput(scope));
-    opBuilder.addInput(outputBackprop.asOutput(scope));
-    opBuilder.addInput(outputHBackprop.asOutput(scope));
-    opBuilder.addInput(outputCBackprop.asOutput(scope));
-    opBuilder.addInput(reserveSpace.asOutput(scope));
-    opBuilder.addInput(hostReserved.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.rnnMode != null) {
-          opBuilder.setAttr("rnn_mode", opts.rnnMode);
-        }
-        if (opts.inputMode != null) {
-          opBuilder.setAttr("input_mode", opts.inputMode);
-        }
-        if (opts.direction != null) {
-          opBuilder.setAttr("direction", opts.direction);
-        }
-        if (opts.dropout != null) {
-          opBuilder.setAttr("dropout", opts.dropout);
-        }
-        if (opts.seed != null) {
-          opBuilder.setAttr("seed", opts.seed);
-        }
-        if (opts.seed2 != null) {
-          opBuilder.setAttr("seed2", opts.seed2);
-        }
-        if (opts.numProj != null) {
-          opBuilder.setAttr("num_proj", opts.numProj);
-        }
-        if (opts.timeMajor != null) {
-          opBuilder.setAttr("time_major", opts.timeMajor);
-        }
-      }
-    }
-    return new CudnnRNNBackprop<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param rnnMode 
-   */
-  public static Options rnnMode(String rnnMode) {
-    return new Options().rnnMode(rnnMode);
-  }
-  
-  /**
-   * @param inputMode 
-   */
-  public static Options inputMode(String inputMode) {
-    return new Options().inputMode(inputMode);
-  }
-  
-  /**
-   * @param direction 
-   */
-  public static Options direction(String direction) {
-    return new Options().direction(direction);
-  }
-  
-  /**
-   * @param dropout 
-   */
-  public static Options dropout(Float dropout) {
-    return new Options().dropout(dropout);
-  }
-  
-  /**
-   * @param seed 
-   */
-  public static Options seed(Long seed) {
-    return new Options().seed(seed);
-  }
-  
-  /**
-   * @param seed2 
-   */
-  public static Options seed2(Long seed2) {
-    return new Options().seed2(seed2);
-  }
-  
-  /**
-   * @param numProj 
-   */
-  public static Options numProj(Long numProj) {
-    return new Options().numProj(numProj);
-  }
-  
-  /**
-   * @param timeMajor 
-   */
-  public static Options timeMajor(Boolean timeMajor) {
-    return new Options().timeMajor(timeMajor);
-  }
-  
-  /**
-   */
-  public Output<T> inputBackprop() {
-    return inputBackprop;
-  }
-  
-  /**
-   */
-  public Output<T> inputHBackprop() {
-    return inputHBackprop;
-  }
-  
-  /**
-   */
-  public Output<T> inputCBackprop() {
-    return inputCBackprop;
-  }
-  
-  /**
-   */
-  public Output<T> paramsBackprop() {
-    return paramsBackprop;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "CudnnRNNBackpropV3";
-  
-  private Output<T> inputBackprop;
-  private Output<T> inputHBackprop;
-  private Output<T> inputCBackprop;
-  private Output<T> paramsBackprop;
-  
-  private CudnnRNNBackprop(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    inputBackprop = operation.output(outputIdx++);
-    inputHBackprop = operation.output(outputIdx++);
-    inputCBackprop = operation.output(outputIdx++);
-    paramsBackprop = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CudnnRNNCanonicalToParams.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CudnnRNNCanonicalToParams.java
deleted file mode 100644
index c92c6e8c8d0..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CudnnRNNCanonicalToParams.java
+++ /dev/null
@@ -1,263 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.nn;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.Operands;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Converts CudnnRNN params from canonical form to usable form. It supports the projection in LSTM.
- * <p>
- * Writes a set of weights into the opaque params buffer so they can be used in
- * upcoming training or inferences.
- * <p>
- * Note that the params buffer may not be compatible across different GPUs. So any
- * save and restoration should be converted to and from the canonical weights and
- * biases.
- * <p>
- * num_layers: Specifies the number of layers in the RNN model.
- * num_units: Specifies the size of the hidden state.
- * input_size: Specifies the size of the input state.
- * weights: the canonical form of weights that can be used for saving
- *     and restoration. They are more likely to be compatible across different
- *     generations.
- * biases: the canonical form of biases that can be used for saving
- *     and restoration. They are more likely to be compatible across different
- *     generations.
- * num_params_weights: number of weight parameter matrix for all layers.
- * num_params_biases: number of bias parameter vector for all layers.
- * rnn_mode: Indicates the type of the RNN model.
- * input_mode: Indicate whether there is a linear projection between the input and
- *     The actual computation before the first layer. 'skip_input' is only allowed
- *     when input_size == num_units; 'auto_select' implies 'skip_input' when
- *     input_size == num_units; otherwise, it implies 'linear_input'.
- * direction: Indicates whether a bidirectional model will be used.
- *     dir = (direction == bidirectional) ? 2 : 1
- * dropout: dropout probability. When set to 0., dropout is disabled.
- * seed: the 1st part of a seed to initialize dropout.
- * seed2: the 2nd part of a seed to initialize dropout.
- * num_proj: The output dimensionality for the projection matrices. If None or 0,
- *     no projection is performed.
- * 
- * @param <T> data type for {@code params()} output
- */
-@Operator(group = "nn")
-public final class CudnnRNNCanonicalToParams<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.nn.CudnnRNNCanonicalToParams}
-   */
-  public static class Options {
-    
-    /**
-     * @param rnnMode 
-     */
-    public Options rnnMode(String rnnMode) {
-      this.rnnMode = rnnMode;
-      return this;
-    }
-    
-    /**
-     * @param inputMode 
-     */
-    public Options inputMode(String inputMode) {
-      this.inputMode = inputMode;
-      return this;
-    }
-    
-    /**
-     * @param direction 
-     */
-    public Options direction(String direction) {
-      this.direction = direction;
-      return this;
-    }
-    
-    /**
-     * @param dropout 
-     */
-    public Options dropout(Float dropout) {
-      this.dropout = dropout;
-      return this;
-    }
-    
-    /**
-     * @param seed 
-     */
-    public Options seed(Long seed) {
-      this.seed = seed;
-      return this;
-    }
-    
-    /**
-     * @param seed2 
-     */
-    public Options seed2(Long seed2) {
-      this.seed2 = seed2;
-      return this;
-    }
-    
-    /**
-     * @param numProj 
-     */
-    public Options numProj(Long numProj) {
-      this.numProj = numProj;
-      return this;
-    }
-    
-    private String rnnMode;
-    private String inputMode;
-    private String direction;
-    private Float dropout;
-    private Long seed;
-    private Long seed2;
-    private Long numProj;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new CudnnRNNCanonicalToParams operation.
-   * 
-   * @param scope current scope
-   * @param numLayers 
-   * @param numUnits 
-   * @param inputSize 
-   * @param weights 
-   * @param biases 
-   * @param options carries optional attributes values
-   * @return a new instance of CudnnRNNCanonicalToParams
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> CudnnRNNCanonicalToParams<T> create(Scope scope, Operand<TInt32> numLayers, Operand<TInt32> numUnits, Operand<TInt32> inputSize, Iterable<Operand<T>> weights, Iterable<Operand<T>> biases, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("CudnnRNNCanonicalToParamsV2", scope.makeOpName("CudnnRNNCanonicalToParams"));
-    opBuilder.addInput(numLayers.asOutput(scope));
-    opBuilder.addInput(numUnits.asOutput(scope));
-    opBuilder.addInput(inputSize.asOutput(scope));
-    opBuilder.addInputList(Operands.asOutputs(scope, weights));
-    opBuilder.addInputList(Operands.asOutputs(scope, biases));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.rnnMode != null) {
-          opBuilder.setAttr("rnn_mode", opts.rnnMode);
-        }
-        if (opts.inputMode != null) {
-          opBuilder.setAttr("input_mode", opts.inputMode);
-        }
-        if (opts.direction != null) {
-          opBuilder.setAttr("direction", opts.direction);
-        }
-        if (opts.dropout != null) {
-          opBuilder.setAttr("dropout", opts.dropout);
-        }
-        if (opts.seed != null) {
-          opBuilder.setAttr("seed", opts.seed);
-        }
-        if (opts.seed2 != null) {
-          opBuilder.setAttr("seed2", opts.seed2);
-        }
-        if (opts.numProj != null) {
-          opBuilder.setAttr("num_proj", opts.numProj);
-        }
-      }
-    }
-    return new CudnnRNNCanonicalToParams<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param rnnMode 
-   */
-  public static Options rnnMode(String rnnMode) {
-    return new Options().rnnMode(rnnMode);
-  }
-  
-  /**
-   * @param inputMode 
-   */
-  public static Options inputMode(String inputMode) {
-    return new Options().inputMode(inputMode);
-  }
-  
-  /**
-   * @param direction 
-   */
-  public static Options direction(String direction) {
-    return new Options().direction(direction);
-  }
-  
-  /**
-   * @param dropout 
-   */
-  public static Options dropout(Float dropout) {
-    return new Options().dropout(dropout);
-  }
-  
-  /**
-   * @param seed 
-   */
-  public static Options seed(Long seed) {
-    return new Options().seed(seed);
-  }
-  
-  /**
-   * @param seed2 
-   */
-  public static Options seed2(Long seed2) {
-    return new Options().seed2(seed2);
-  }
-  
-  /**
-   * @param numProj 
-   */
-  public static Options numProj(Long numProj) {
-    return new Options().numProj(numProj);
-  }
-  
-  /**
-   */
-  public Output<T> params() {
-    return params;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return params;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "CudnnRNNCanonicalToParamsV2";
-  
-  private Output<T> params;
-  
-  private CudnnRNNCanonicalToParams(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    params = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CudnnRNNParamsToCanonical.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CudnnRNNParamsToCanonical.java
deleted file mode 100644
index c76cef9ced0..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CudnnRNNParamsToCanonical.java
+++ /dev/null
@@ -1,274 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.nn;
-
-import java.util.Arrays;
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Retrieves CudnnRNN params in canonical form. It supports the projection in LSTM.
- * <p>
- * Retrieves a set of weights from the opaque params buffer that can be saved and
- * restored in a way compatible with future runs.
- * <p>
- * Note that the params buffer may not be compatible across different GPUs. So any
- * save and restoration should be converted to and from the canonical weights and
- * biases.
- * <p>
- * num_layers: Specifies the number of layers in the RNN model.
- * num_units: Specifies the size of the hidden state.
- * input_size: Specifies the size of the input state.
- * num_params_weights: number of weight parameter matrix for all layers.
- * num_params_biases: number of bias parameter vector for all layers.
- * weights: the canonical form of weights that can be used for saving
- *     and restoration. They are more likely to be compatible across different
- *     generations.
- * biases: the canonical form of biases that can be used for saving
- *     and restoration. They are more likely to be compatible across different
- *     generations.
- * rnn_mode: Indicates the type of the RNN model.
- * input_mode: Indicate whether there is a linear projection between the input and
- *     The actual computation before the first layer. 'skip_input' is only allowed
- *     when input_size == num_units; 'auto_select' implies 'skip_input' when
- *     input_size == num_units; otherwise, it implies 'linear_input'.
- * direction: Indicates whether a bidirectional model will be used.
- *     dir = (direction == bidirectional) ? 2 : 1
- * dropout: dropout probability. When set to 0., dropout is disabled.
- * seed: the 1st part of a seed to initialize dropout.
- * seed2: the 2nd part of a seed to initialize dropout.
- * num_proj: The output dimensionality for the projection matrices. If None or 0,
- *     no projection is performed.
- * 
- * @param <T> data type for {@code weights()} output
- */
-@Operator(group = "nn")
-public final class CudnnRNNParamsToCanonical<T extends TNumber> extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.nn.CudnnRNNParamsToCanonical}
-   */
-  public static class Options {
-    
-    /**
-     * @param rnnMode 
-     */
-    public Options rnnMode(String rnnMode) {
-      this.rnnMode = rnnMode;
-      return this;
-    }
-    
-    /**
-     * @param inputMode 
-     */
-    public Options inputMode(String inputMode) {
-      this.inputMode = inputMode;
-      return this;
-    }
-    
-    /**
-     * @param direction 
-     */
-    public Options direction(String direction) {
-      this.direction = direction;
-      return this;
-    }
-    
-    /**
-     * @param dropout 
-     */
-    public Options dropout(Float dropout) {
-      this.dropout = dropout;
-      return this;
-    }
-    
-    /**
-     * @param seed 
-     */
-    public Options seed(Long seed) {
-      this.seed = seed;
-      return this;
-    }
-    
-    /**
-     * @param seed2 
-     */
-    public Options seed2(Long seed2) {
-      this.seed2 = seed2;
-      return this;
-    }
-    
-    /**
-     * @param numProj 
-     */
-    public Options numProj(Long numProj) {
-      this.numProj = numProj;
-      return this;
-    }
-    
-    private String rnnMode;
-    private String inputMode;
-    private String direction;
-    private Float dropout;
-    private Long seed;
-    private Long seed2;
-    private Long numProj;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new CudnnRNNParamsToCanonical operation.
-   * 
-   * @param scope current scope
-   * @param numLayers 
-   * @param numUnits 
-   * @param inputSize 
-   * @param params 
-   * @param numParamsWeights 
-   * @param numParamsBiases 
-   * @param options carries optional attributes values
-   * @return a new instance of CudnnRNNParamsToCanonical
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> CudnnRNNParamsToCanonical<T> create(Scope scope, Operand<TInt32> numLayers, Operand<TInt32> numUnits, Operand<TInt32> inputSize, Operand<T> params, Long numParamsWeights, Long numParamsBiases, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("CudnnRNNParamsToCanonicalV2", scope.makeOpName("CudnnRNNParamsToCanonical"));
-    opBuilder.addInput(numLayers.asOutput(scope));
-    opBuilder.addInput(numUnits.asOutput(scope));
-    opBuilder.addInput(inputSize.asOutput(scope));
-    opBuilder.addInput(params.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("num_params_weights", numParamsWeights);
-    opBuilder.setAttr("num_params_biases", numParamsBiases);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.rnnMode != null) {
-          opBuilder.setAttr("rnn_mode", opts.rnnMode);
-        }
-        if (opts.inputMode != null) {
-          opBuilder.setAttr("input_mode", opts.inputMode);
-        }
-        if (opts.direction != null) {
-          opBuilder.setAttr("direction", opts.direction);
-        }
-        if (opts.dropout != null) {
-          opBuilder.setAttr("dropout", opts.dropout);
-        }
-        if (opts.seed != null) {
-          opBuilder.setAttr("seed", opts.seed);
-        }
-        if (opts.seed2 != null) {
-          opBuilder.setAttr("seed2", opts.seed2);
-        }
-        if (opts.numProj != null) {
-          opBuilder.setAttr("num_proj", opts.numProj);
-        }
-      }
-    }
-    return new CudnnRNNParamsToCanonical<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param rnnMode 
-   */
-  public static Options rnnMode(String rnnMode) {
-    return new Options().rnnMode(rnnMode);
-  }
-  
-  /**
-   * @param inputMode 
-   */
-  public static Options inputMode(String inputMode) {
-    return new Options().inputMode(inputMode);
-  }
-  
-  /**
-   * @param direction 
-   */
-  public static Options direction(String direction) {
-    return new Options().direction(direction);
-  }
-  
-  /**
-   * @param dropout 
-   */
-  public static Options dropout(Float dropout) {
-    return new Options().dropout(dropout);
-  }
-  
-  /**
-   * @param seed 
-   */
-  public static Options seed(Long seed) {
-    return new Options().seed(seed);
-  }
-  
-  /**
-   * @param seed2 
-   */
-  public static Options seed2(Long seed2) {
-    return new Options().seed2(seed2);
-  }
-  
-  /**
-   * @param numProj 
-   */
-  public static Options numProj(Long numProj) {
-    return new Options().numProj(numProj);
-  }
-  
-  /**
-   */
-  public List<Output<T>> weights() {
-    return weights;
-  }
-  
-  /**
-   */
-  public List<Output<T>> biases() {
-    return biases;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "CudnnRNNParamsToCanonicalV2";
-  
-  private List<Output<T>> weights;
-  private List<Output<T>> biases;
-  
-  @SuppressWarnings("unchecked")
-  private CudnnRNNParamsToCanonical(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    int weightsLength = operation.outputListLength("weights");
-    weights = Arrays.asList((Output<T>[])operation.outputList(outputIdx, weightsLength));
-    outputIdx += weightsLength;
-    int biasesLength = operation.outputListLength("biases");
-    biases = Arrays.asList((Output<T>[])operation.outputList(outputIdx, biasesLength));
-    outputIdx += biasesLength;
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CudnnRnnParamsSize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CudnnRnnParamsSize.java
deleted file mode 100644
index 3372899ee2f..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CudnnRnnParamsSize.java
+++ /dev/null
@@ -1,253 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.nn;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Computes size of weights that can be used by a Cudnn RNN model.
- * <p>
- * Return the params size that can be used by the Cudnn RNN model. Subsequent
- * weight allocation and initialization should use this size.
- * <p>
- * num_layers: Specifies the number of layers in the RNN model.
- * num_units: Specifies the size of the hidden state.
- * input_size: Specifies the size of the input state.
- * rnn_mode: Indicates the type of the RNN model.
- * input_mode: Indicate whether there is a linear projection between the input and
- *   The actual computation before the first layer. 'skip_input' is only allowed
- *   when input_size == num_units; 'auto_select' implies 'skip_input' when
- *   input_size == num_units; otherwise, it implies 'linear_input'.
- * direction: Indicates whether a bidirectional model will be used.
- *   dir = (direction == bidirectional) ? 2 : 1
- * dropout: dropout probability. When set to 0., dropout is disabled.
- * seed: the 1st part of a seed to initialize dropout.
- * seed2: the 2nd part of a seed to initialize dropout.
- * params_size: The size of the params buffer that should be allocated and
- *   initialized for this RNN model. Note that this params buffer may not be
- *   compatible across GPUs. Please use CudnnRNNParamsWeights and
- *   CudnnRNNParamsBiases to save and restore them in a way that is compatible
- *   across different runs.
- * 
- * @param <U> data type for {@code paramsSize()} output
- */
-@Operator(group = "nn")
-public final class CudnnRnnParamsSize<U extends TNumber> extends RawOp implements Operand<U> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.nn.CudnnRnnParamsSize}
-   */
-  public static class Options {
-    
-    /**
-     * @param rnnMode 
-     */
-    public Options rnnMode(String rnnMode) {
-      this.rnnMode = rnnMode;
-      return this;
-    }
-    
-    /**
-     * @param inputMode 
-     */
-    public Options inputMode(String inputMode) {
-      this.inputMode = inputMode;
-      return this;
-    }
-    
-    /**
-     * @param direction 
-     */
-    public Options direction(String direction) {
-      this.direction = direction;
-      return this;
-    }
-    
-    /**
-     * @param dropout 
-     */
-    public Options dropout(Float dropout) {
-      this.dropout = dropout;
-      return this;
-    }
-    
-    /**
-     * @param seed 
-     */
-    public Options seed(Long seed) {
-      this.seed = seed;
-      return this;
-    }
-    
-    /**
-     * @param seed2 
-     */
-    public Options seed2(Long seed2) {
-      this.seed2 = seed2;
-      return this;
-    }
-    
-    /**
-     * @param numProj 
-     */
-    public Options numProj(Long numProj) {
-      this.numProj = numProj;
-      return this;
-    }
-    
-    private String rnnMode;
-    private String inputMode;
-    private String direction;
-    private Float dropout;
-    private Long seed;
-    private Long seed2;
-    private Long numProj;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new CudnnRnnParamsSize operation.
-   * 
-   * @param scope current scope
-   * @param numLayers 
-   * @param numUnits 
-   * @param inputSize 
-   * @param T 
-   * @param S 
-   * @param options carries optional attributes values
-   * @return a new instance of CudnnRnnParamsSize
-   */
-  @Endpoint(describeByClass = true)
-  public static <U extends TNumber, T extends TNumber> CudnnRnnParamsSize<U> create(Scope scope, Operand<TInt32> numLayers, Operand<TInt32> numUnits, Operand<TInt32> inputSize, Class<T> T, Class<U> S, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("CudnnRNNParamsSize", scope.makeOpName("CudnnRnnParamsSize"));
-    opBuilder.addInput(numLayers.asOutput(scope));
-    opBuilder.addInput(numUnits.asOutput(scope));
-    opBuilder.addInput(inputSize.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("T", T);
-    opBuilder.setAttr("S", S);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.rnnMode != null) {
-          opBuilder.setAttr("rnn_mode", opts.rnnMode);
-        }
-        if (opts.inputMode != null) {
-          opBuilder.setAttr("input_mode", opts.inputMode);
-        }
-        if (opts.direction != null) {
-          opBuilder.setAttr("direction", opts.direction);
-        }
-        if (opts.dropout != null) {
-          opBuilder.setAttr("dropout", opts.dropout);
-        }
-        if (opts.seed != null) {
-          opBuilder.setAttr("seed", opts.seed);
-        }
-        if (opts.seed2 != null) {
-          opBuilder.setAttr("seed2", opts.seed2);
-        }
-        if (opts.numProj != null) {
-          opBuilder.setAttr("num_proj", opts.numProj);
-        }
-      }
-    }
-    return new CudnnRnnParamsSize<U>(opBuilder.build());
-  }
-  
-  /**
-   * @param rnnMode 
-   */
-  public static Options rnnMode(String rnnMode) {
-    return new Options().rnnMode(rnnMode);
-  }
-  
-  /**
-   * @param inputMode 
-   */
-  public static Options inputMode(String inputMode) {
-    return new Options().inputMode(inputMode);
-  }
-  
-  /**
-   * @param direction 
-   */
-  public static Options direction(String direction) {
-    return new Options().direction(direction);
-  }
-  
-  /**
-   * @param dropout 
-   */
-  public static Options dropout(Float dropout) {
-    return new Options().dropout(dropout);
-  }
-  
-  /**
-   * @param seed 
-   */
-  public static Options seed(Long seed) {
-    return new Options().seed(seed);
-  }
-  
-  /**
-   * @param seed2 
-   */
-  public static Options seed2(Long seed2) {
-    return new Options().seed2(seed2);
-  }
-  
-  /**
-   * @param numProj 
-   */
-  public static Options numProj(Long numProj) {
-    return new Options().numProj(numProj);
-  }
-  
-  /**
-   */
-  public Output<U> paramsSize() {
-    return paramsSize;
-  }
-  
-  @Override
-  public Output<U> asOutput(Scope scope) {
-    return paramsSize;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "CudnnRNNParamsSize";
-  
-  private Output<U> paramsSize;
-  
-  private CudnnRnnParamsSize(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    paramsSize = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/DataFormatDimMap.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/DataFormatDimMap.java
deleted file mode 100644
index 8090d201190..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/DataFormatDimMap.java
+++ /dev/null
@@ -1,131 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.nn;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Returns the dimension index in the destination data format given the one in
- * <p>
- * the source data format.
- * 
- * @param <T> data type for {@code y()} output
- */
-@Operator(group = "nn")
-public final class DataFormatDimMap<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.nn.DataFormatDimMap}
-   */
-  public static class Options {
-    
-    /**
-     * @param srcFormat source data format.
-     */
-    public Options srcFormat(String srcFormat) {
-      this.srcFormat = srcFormat;
-      return this;
-    }
-    
-    /**
-     * @param dstFormat destination data format.
-     */
-    public Options dstFormat(String dstFormat) {
-      this.dstFormat = dstFormat;
-      return this;
-    }
-    
-    private String srcFormat;
-    private String dstFormat;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new DataFormatDimMap operation.
-   * 
-   * @param scope current scope
-   * @param x A Tensor with each element as a dimension index in source data format.
-   * Must be in the range [-4, 4).
-   * @param options carries optional attributes values
-   * @return a new instance of DataFormatDimMap
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> DataFormatDimMap<T> create(Scope scope, Operand<T> x, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("DataFormatDimMap", scope.makeOpName("DataFormatDimMap"));
-    opBuilder.addInput(x.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.srcFormat != null) {
-          opBuilder.setAttr("src_format", opts.srcFormat);
-        }
-        if (opts.dstFormat != null) {
-          opBuilder.setAttr("dst_format", opts.dstFormat);
-        }
-      }
-    }
-    return new DataFormatDimMap<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param srcFormat source data format.
-   */
-  public static Options srcFormat(String srcFormat) {
-    return new Options().srcFormat(srcFormat);
-  }
-  
-  /**
-   * @param dstFormat destination data format.
-   */
-  public static Options dstFormat(String dstFormat) {
-    return new Options().dstFormat(dstFormat);
-  }
-  
-  /**
-   * A Tensor with each element as a dimension index in destination data format.
-   */
-  public Output<T> y() {
-    return y;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return y;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "DataFormatDimMap";
-  
-  private Output<T> y;
-  
-  private DataFormatDimMap(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    y = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/DataFormatVecPermute.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/DataFormatVecPermute.java
deleted file mode 100644
index f9b7f096ed3..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/DataFormatVecPermute.java
+++ /dev/null
@@ -1,130 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.nn;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Returns the permuted vector/tensor in the destination data format given the
- * <p>
- * one in the source data format.
- * 
- * @param <T> data type for {@code y()} output
- */
-@Operator(group = "nn")
-public final class DataFormatVecPermute<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.nn.DataFormatVecPermute}
-   */
-  public static class Options {
-    
-    /**
-     * @param srcFormat source data format.
-     */
-    public Options srcFormat(String srcFormat) {
-      this.srcFormat = srcFormat;
-      return this;
-    }
-    
-    /**
-     * @param dstFormat destination data format.
-     */
-    public Options dstFormat(String dstFormat) {
-      this.dstFormat = dstFormat;
-      return this;
-    }
-    
-    private String srcFormat;
-    private String dstFormat;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new DataFormatVecPermute operation.
-   * 
-   * @param scope current scope
-   * @param x Vector of size 4 or Tensor of shape (4, 2) in source data format.
-   * @param options carries optional attributes values
-   * @return a new instance of DataFormatVecPermute
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> DataFormatVecPermute<T> create(Scope scope, Operand<T> x, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("DataFormatVecPermute", scope.makeOpName("DataFormatVecPermute"));
-    opBuilder.addInput(x.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.srcFormat != null) {
-          opBuilder.setAttr("src_format", opts.srcFormat);
-        }
-        if (opts.dstFormat != null) {
-          opBuilder.setAttr("dst_format", opts.dstFormat);
-        }
-      }
-    }
-    return new DataFormatVecPermute<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param srcFormat source data format.
-   */
-  public static Options srcFormat(String srcFormat) {
-    return new Options().srcFormat(srcFormat);
-  }
-  
-  /**
-   * @param dstFormat destination data format.
-   */
-  public static Options dstFormat(String dstFormat) {
-    return new Options().dstFormat(dstFormat);
-  }
-  
-  /**
-   * Vector of size 4 or Tensor of shape (4, 2) in destination data format.
-   */
-  public Output<T> y() {
-    return y;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return y;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "DataFormatVecPermute";
-  
-  private Output<T> y;
-  
-  private DataFormatVecPermute(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    y = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/DepthToSpace.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/DepthToSpace.java
deleted file mode 100644
index a35e6e9e474..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/DepthToSpace.java
+++ /dev/null
@@ -1,190 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.nn;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * DepthToSpace for tensors of type T.
- * <p>
- * Rearranges data from depth into blocks of spatial data.
- * This is the reverse transformation of SpaceToDepth. More specifically,
- * this op outputs a copy of the input tensor where values from the `depth`
- * dimension are moved in spatial blocks to the `height` and `width` dimensions.
- * The attr `block_size` indicates the input block size and how the data is moved.
- * <p>
- *   * Chunks of data of size `block_size * block_size` from depth are rearranged
- *     into non-overlapping blocks of size `block_size x block_size`
- *   * The width the output tensor is `input_depth * block_size`, whereas the
- *     height is `input_height * block_size`.
- *   * The Y, X coordinates within each block of the output image are determined
- *     by the high order component of the input channel index.
- *   * The depth of the input tensor must be divisible by
- *     `block_size * block_size`.
- * <p>
- * The `data_format` attr specifies the layout of the input and output tensors
- * with the following options:
- *   "NHWC": `[ batch, height, width, channels ]`
- *   "NCHW": `[ batch, channels, height, width ]`
- *   "NCHW_VECT_C":
- *       `qint8 [ batch, channels / 4, height, width, 4 ]`
- * <p>
- * It is useful to consider the operation as transforming a 6-D Tensor.
- * e.g. for data_format = NHWC,
- *      Each element in the input tensor can be specified via 6 coordinates,
- *      ordered by decreasing memory layout significance as:
- *      n,iY,iX,bY,bX,oC  (where n=batch index, iX, iY means X or Y coordinates
- *                         within the input image, bX, bY means coordinates
- *                         within the output block, oC means output channels).
- *      The output would be the input transposed to the following layout:
- *      n,iY,bY,iX,bX,oC
- * <p>
- * This operation is useful for resizing the activations between convolutions
- * (but keeping all data), e.g. instead of pooling. It is also useful for training
- * purely convolutional models.
- * <p>
- * For example, given an input of shape `[1, 1, 1, 4]`, data_format = "NHWC" and
- * block_size = 2:
- * <pre>{@code
- * x = [[[[1, 2, 3, 4]]]]
- * 
- * }</pre>
- * This operation will output a tensor of shape `[1, 2, 2, 1]`:
- * <pre>{@code
- *    [[[[1], [2]],
- *      [[3], [4]]]]
- * }</pre>
- * Here, the input has a batch of 1 and each batch element has shape `[1, 1, 4]`,
- * the corresponding output will have 2x2 elements and will have a depth of
- * 1 channel (1 = `4 / (block_size * block_size)`).
- * The output element shape is `[2, 2, 1]`.
- * <p>
- * For an input tensor with larger depth, here of shape `[1, 1, 1, 12]`, e.g.
- * <pre>{@code
- * x = [[[[1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12]]]]
- * }</pre>
- * This operation, for block size of 2, will return the following tensor of shape
- * `[1, 2, 2, 3]`
- * <pre>{@code
- *    [[[[1, 2, 3], [4, 5, 6]],
- *      [[7, 8, 9], [10, 11, 12]]]]
- * 
- * }</pre>
- * Similarly, for the following input of shape `[1 2 2 4]`, and a block size of 2:
- * <pre>{@code
- * x =  [[[[1, 2, 3, 4],
- *        [5, 6, 7, 8]],
- *       [[9, 10, 11, 12],
- *        [13, 14, 15, 16]]]]
- * }</pre>
- * the operator will return the following tensor of shape `[1 4 4 1]`:
- * <pre>{@code
- * x = [[[ [1],   [2],  [5],  [6]],
- *       [ [3],   [4],  [7],  [8]],
- *       [ [9],  [10], [13],  [14]],
- *       [ [11], [12], [15],  [16]]]]
- * 
- * }</pre>
- * 
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator(group = "nn")
-public final class DepthToSpace<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.nn.DepthToSpace}
-   */
-  public static class Options {
-    
-    /**
-     * @param dataFormat 
-     */
-    public Options dataFormat(String dataFormat) {
-      this.dataFormat = dataFormat;
-      return this;
-    }
-    
-    private String dataFormat;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new DepthToSpace operation.
-   * 
-   * @param scope current scope
-   * @param input 
-   * @param blockSize The size of the spatial block, same as in Space2Depth.
-   * @param options carries optional attributes values
-   * @return a new instance of DepthToSpace
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> DepthToSpace<T> create(Scope scope, Operand<T> input, Long blockSize, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("DepthToSpace", scope.makeOpName("DepthToSpace"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("block_size", blockSize);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.dataFormat != null) {
-          opBuilder.setAttr("data_format", opts.dataFormat);
-        }
-      }
-    }
-    return new DepthToSpace<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param dataFormat 
-   */
-  public static Options dataFormat(String dataFormat) {
-    return new Options().dataFormat(dataFormat);
-  }
-  
-  /**
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "DepthToSpace";
-  
-  private Output<T> output;
-  
-  private DepthToSpace(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/DepthwiseConv2dNative.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/DepthwiseConv2dNative.java
deleted file mode 100644
index 5e799acd0a5..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/DepthwiseConv2dNative.java
+++ /dev/null
@@ -1,199 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.nn;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Computes a 2-D depthwise convolution given 4-D `input` and `filter` tensors.
- * <p>
- * Given an input tensor of shape `[batch, in_height, in_width, in_channels]`
- * and a filter / kernel tensor of shape
- * `[filter_height, filter_width, in_channels, channel_multiplier]`, containing
- * `in_channels` convolutional filters of depth 1, `depthwise_conv2d` applies
- * a different filter to each input channel (expanding from 1 channel to
- * `channel_multiplier` channels for each), then concatenates the results
- * together. Thus, the output has `in_channels * channel_multiplier` channels.
- * <pre>{@code
- * for k in 0..in_channels-1
- *   for q in 0..channel_multiplier-1
- *     output[b, i, j, k * channel_multiplier + q] =
- *       sum_{di, dj} input[b, strides[1] * i + di, strides[2] * j + dj, k] *
- *                         filter[di, dj, k, q]
- * }</pre>
- * Must have `strides[0] = strides[3] = 1`.  For the most common case of the same
- * horizontal and vertices strides, `strides = [1, stride, stride, 1]`.
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator(group = "nn")
-public final class DepthwiseConv2dNative<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.nn.DepthwiseConv2dNative}
-   */
-  public static class Options {
-    
-    /**
-     * @param explicitPaddings 
-     */
-    public Options explicitPaddings(List<Long> explicitPaddings) {
-      this.explicitPaddings = explicitPaddings;
-      return this;
-    }
-    
-    /**
-     * @param dataFormat Specify the data format of the input and output data. With the
-     * default format "NHWC", the data is stored in the order of:
-     *     [batch, height, width, channels].
-     * Alternatively, the format could be "NCHW", the data storage order of:
-     *     [batch, channels, height, width].
-     */
-    public Options dataFormat(String dataFormat) {
-      this.dataFormat = dataFormat;
-      return this;
-    }
-    
-    /**
-     * @param dilations 1-D tensor of length 4.  The dilation factor for each dimension of
-     * `input`. If set to k > 1, there will be k-1 skipped cells between each filter
-     * element on that dimension. The dimension order is determined by the value of
-     * `data_format`, see above for details. Dilations in the batch and depth
-     * dimensions must be 1.
-     */
-    public Options dilations(List<Long> dilations) {
-      this.dilations = dilations;
-      return this;
-    }
-    
-    private List<Long> explicitPaddings;
-    private String dataFormat;
-    private List<Long> dilations;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new DepthwiseConv2dNative operation.
-   * 
-   * @param scope current scope
-   * @param input 
-   * @param filter 
-   * @param strides 1-D of length 4.  The stride of the sliding window for each dimension
-   * of `input`.
-   * @param padding The type of padding algorithm to use.
-   * @param options carries optional attributes values
-   * @return a new instance of DepthwiseConv2dNative
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> DepthwiseConv2dNative<T> create(Scope scope, Operand<T> input, Operand<T> filter, List<Long> strides, String padding, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("DepthwiseConv2dNative", scope.makeOpName("DepthwiseConv2dNative"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder.addInput(filter.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    long[] stridesArray = new long[strides.size()];
-    for (int i = 0; i < stridesArray.length; ++i) {
-      stridesArray[i] = strides.get(i);
-    }
-    opBuilder.setAttr("strides", stridesArray);
-    opBuilder.setAttr("padding", padding);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.explicitPaddings != null) {
-          long[] explicitPaddingsArray = new long[opts.explicitPaddings.size()];
-          for (int i = 0; i < explicitPaddingsArray.length; ++i) {
-            explicitPaddingsArray[i] = opts.explicitPaddings.get(i);
-          }
-          opBuilder.setAttr("explicit_paddings", explicitPaddingsArray);
-        }
-        if (opts.dataFormat != null) {
-          opBuilder.setAttr("data_format", opts.dataFormat);
-        }
-        if (opts.dilations != null) {
-          long[] dilationsArray = new long[opts.dilations.size()];
-          for (int i = 0; i < dilationsArray.length; ++i) {
-            dilationsArray[i] = opts.dilations.get(i);
-          }
-          opBuilder.setAttr("dilations", dilationsArray);
-        }
-      }
-    }
-    return new DepthwiseConv2dNative<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param explicitPaddings 
-   */
-  public static Options explicitPaddings(List<Long> explicitPaddings) {
-    return new Options().explicitPaddings(explicitPaddings);
-  }
-  
-  /**
-   * @param dataFormat Specify the data format of the input and output data. With the
-   * default format "NHWC", the data is stored in the order of:
-   *     [batch, height, width, channels].
-   * Alternatively, the format could be "NCHW", the data storage order of:
-   *     [batch, channels, height, width].
-   */
-  public static Options dataFormat(String dataFormat) {
-    return new Options().dataFormat(dataFormat);
-  }
-  
-  /**
-   * @param dilations 1-D tensor of length 4.  The dilation factor for each dimension of
-   * `input`. If set to k > 1, there will be k-1 skipped cells between each filter
-   * element on that dimension. The dimension order is determined by the value of
-   * `data_format`, see above for details. Dilations in the batch and depth
-   * dimensions must be 1.
-   */
-  public static Options dilations(List<Long> dilations) {
-    return new Options().dilations(dilations);
-  }
-  
-  /**
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "DepthwiseConv2dNative";
-  
-  private Output<T> output;
-  
-  private DepthwiseConv2dNative(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/DepthwiseConv2dNativeBackpropFilter.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/DepthwiseConv2dNativeBackpropFilter.java
deleted file mode 100644
index bee307ed16b..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/DepthwiseConv2dNativeBackpropFilter.java
+++ /dev/null
@@ -1,195 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.nn;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Computes the gradients of depthwise convolution with respect to the filter.
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator(group = "nn")
-public final class DepthwiseConv2dNativeBackpropFilter<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.nn.DepthwiseConv2dNativeBackpropFilter}
-   */
-  public static class Options {
-    
-    /**
-     * @param explicitPaddings 
-     */
-    public Options explicitPaddings(List<Long> explicitPaddings) {
-      this.explicitPaddings = explicitPaddings;
-      return this;
-    }
-    
-    /**
-     * @param dataFormat Specify the data format of the input and output data. With the
-     * default format "NHWC", the data is stored in the order of:
-     *     [batch, height, width, channels].
-     * Alternatively, the format could be "NCHW", the data storage order of:
-     *     [batch, channels, height, width].
-     */
-    public Options dataFormat(String dataFormat) {
-      this.dataFormat = dataFormat;
-      return this;
-    }
-    
-    /**
-     * @param dilations 1-D tensor of length 4.  The dilation factor for each dimension of
-     * `input`. If set to k > 1, there will be k-1 skipped cells between each filter
-     * element on that dimension. The dimension order is determined by the value of
-     * `data_format`, see above for details. Dilations in the batch and depth
-     * dimensions must be 1.
-     */
-    public Options dilations(List<Long> dilations) {
-      this.dilations = dilations;
-      return this;
-    }
-    
-    private List<Long> explicitPaddings;
-    private String dataFormat;
-    private List<Long> dilations;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new DepthwiseConv2dNativeBackpropFilter operation.
-   * 
-   * @param scope current scope
-   * @param input 4-D with shape based on `data_format`.  For example, if
-   * `data_format` is 'NHWC' then `input` is a 4-D `[batch, in_height,
-   * in_width, in_channels]` tensor.
-   * @param filterSizes An integer vector representing the tensor shape of `filter`,
-   * where `filter` is a 4-D
-   * `[filter_height, filter_width, in_channels, depthwise_multiplier]` tensor.
-   * @param outBackprop 4-D with shape  based on `data_format`.
-   * For example, if `data_format` is 'NHWC' then
-   * out_backprop shape is `[batch, out_height, out_width, out_channels]`.
-   * Gradients w.r.t. the output of the convolution.
-   * @param strides The stride of the sliding window for each dimension of the input
-   * of the convolution.
-   * @param padding The type of padding algorithm to use.
-   * @param options carries optional attributes values
-   * @return a new instance of DepthwiseConv2dNativeBackpropFilter
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> DepthwiseConv2dNativeBackpropFilter<T> create(Scope scope, Operand<T> input, Operand<TInt32> filterSizes, Operand<T> outBackprop, List<Long> strides, String padding, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("DepthwiseConv2dNativeBackpropFilter", scope.makeOpName("DepthwiseConv2dNativeBackpropFilter"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder.addInput(filterSizes.asOutput(scope));
-    opBuilder.addInput(outBackprop.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    long[] stridesArray = new long[strides.size()];
-    for (int i = 0; i < stridesArray.length; ++i) {
-      stridesArray[i] = strides.get(i);
-    }
-    opBuilder.setAttr("strides", stridesArray);
-    opBuilder.setAttr("padding", padding);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.explicitPaddings != null) {
-          long[] explicitPaddingsArray = new long[opts.explicitPaddings.size()];
-          for (int i = 0; i < explicitPaddingsArray.length; ++i) {
-            explicitPaddingsArray[i] = opts.explicitPaddings.get(i);
-          }
-          opBuilder.setAttr("explicit_paddings", explicitPaddingsArray);
-        }
-        if (opts.dataFormat != null) {
-          opBuilder.setAttr("data_format", opts.dataFormat);
-        }
-        if (opts.dilations != null) {
-          long[] dilationsArray = new long[opts.dilations.size()];
-          for (int i = 0; i < dilationsArray.length; ++i) {
-            dilationsArray[i] = opts.dilations.get(i);
-          }
-          opBuilder.setAttr("dilations", dilationsArray);
-        }
-      }
-    }
-    return new DepthwiseConv2dNativeBackpropFilter<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param explicitPaddings 
-   */
-  public static Options explicitPaddings(List<Long> explicitPaddings) {
-    return new Options().explicitPaddings(explicitPaddings);
-  }
-  
-  /**
-   * @param dataFormat Specify the data format of the input and output data. With the
-   * default format "NHWC", the data is stored in the order of:
-   *     [batch, height, width, channels].
-   * Alternatively, the format could be "NCHW", the data storage order of:
-   *     [batch, channels, height, width].
-   */
-  public static Options dataFormat(String dataFormat) {
-    return new Options().dataFormat(dataFormat);
-  }
-  
-  /**
-   * @param dilations 1-D tensor of length 4.  The dilation factor for each dimension of
-   * `input`. If set to k > 1, there will be k-1 skipped cells between each filter
-   * element on that dimension. The dimension order is determined by the value of
-   * `data_format`, see above for details. Dilations in the batch and depth
-   * dimensions must be 1.
-   */
-  public static Options dilations(List<Long> dilations) {
-    return new Options().dilations(dilations);
-  }
-  
-  /**
-   * 4-D with shape
-   * `[filter_height, filter_width, in_channels, out_channels]`.  Gradient w.r.t.
-   * the `filter` input of the convolution.
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "DepthwiseConv2dNativeBackpropFilter";
-  
-  private Output<T> output;
-  
-  private DepthwiseConv2dNativeBackpropFilter(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/DepthwiseConv2dNativeBackpropInput.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/DepthwiseConv2dNativeBackpropInput.java
deleted file mode 100644
index 99d8e2933f2..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/DepthwiseConv2dNativeBackpropInput.java
+++ /dev/null
@@ -1,195 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.nn;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Computes the gradients of depthwise convolution with respect to the input.
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator(group = "nn")
-public final class DepthwiseConv2dNativeBackpropInput<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.nn.DepthwiseConv2dNativeBackpropInput}
-   */
-  public static class Options {
-    
-    /**
-     * @param explicitPaddings 
-     */
-    public Options explicitPaddings(List<Long> explicitPaddings) {
-      this.explicitPaddings = explicitPaddings;
-      return this;
-    }
-    
-    /**
-     * @param dataFormat Specify the data format of the input and output data. With the
-     * default format "NHWC", the data is stored in the order of:
-     *     [batch, height, width, channels].
-     * Alternatively, the format could be "NCHW", the data storage order of:
-     *     [batch, channels, height, width].
-     */
-    public Options dataFormat(String dataFormat) {
-      this.dataFormat = dataFormat;
-      return this;
-    }
-    
-    /**
-     * @param dilations 1-D tensor of length 4.  The dilation factor for each dimension of
-     * `input`. If set to k > 1, there will be k-1 skipped cells between each filter
-     * element on that dimension. The dimension order is determined by the value of
-     * `data_format`, see above for details. Dilations in the batch and depth
-     * dimensions must be 1.
-     */
-    public Options dilations(List<Long> dilations) {
-      this.dilations = dilations;
-      return this;
-    }
-    
-    private List<Long> explicitPaddings;
-    private String dataFormat;
-    private List<Long> dilations;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new DepthwiseConv2dNativeBackpropInput operation.
-   * 
-   * @param scope current scope
-   * @param inputSizes An integer vector representing the shape of `input`, based
-   * on `data_format`.  For example, if `data_format` is 'NHWC' then
-   *  `input` is a 4-D `[batch, height, width, channels]` tensor.
-   * @param filter 4-D with shape
-   * `[filter_height, filter_width, in_channels, depthwise_multiplier]`.
-   * @param outBackprop 4-D with shape  based on `data_format`.
-   * For example, if `data_format` is 'NHWC' then
-   * out_backprop shape is `[batch, out_height, out_width, out_channels]`.
-   * Gradients w.r.t. the output of the convolution.
-   * @param strides The stride of the sliding window for each dimension of the input
-   * of the convolution.
-   * @param padding The type of padding algorithm to use.
-   * @param options carries optional attributes values
-   * @return a new instance of DepthwiseConv2dNativeBackpropInput
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> DepthwiseConv2dNativeBackpropInput<T> create(Scope scope, Operand<TInt32> inputSizes, Operand<T> filter, Operand<T> outBackprop, List<Long> strides, String padding, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("DepthwiseConv2dNativeBackpropInput", scope.makeOpName("DepthwiseConv2dNativeBackpropInput"));
-    opBuilder.addInput(inputSizes.asOutput(scope));
-    opBuilder.addInput(filter.asOutput(scope));
-    opBuilder.addInput(outBackprop.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    long[] stridesArray = new long[strides.size()];
-    for (int i = 0; i < stridesArray.length; ++i) {
-      stridesArray[i] = strides.get(i);
-    }
-    opBuilder.setAttr("strides", stridesArray);
-    opBuilder.setAttr("padding", padding);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.explicitPaddings != null) {
-          long[] explicitPaddingsArray = new long[opts.explicitPaddings.size()];
-          for (int i = 0; i < explicitPaddingsArray.length; ++i) {
-            explicitPaddingsArray[i] = opts.explicitPaddings.get(i);
-          }
-          opBuilder.setAttr("explicit_paddings", explicitPaddingsArray);
-        }
-        if (opts.dataFormat != null) {
-          opBuilder.setAttr("data_format", opts.dataFormat);
-        }
-        if (opts.dilations != null) {
-          long[] dilationsArray = new long[opts.dilations.size()];
-          for (int i = 0; i < dilationsArray.length; ++i) {
-            dilationsArray[i] = opts.dilations.get(i);
-          }
-          opBuilder.setAttr("dilations", dilationsArray);
-        }
-      }
-    }
-    return new DepthwiseConv2dNativeBackpropInput<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param explicitPaddings 
-   */
-  public static Options explicitPaddings(List<Long> explicitPaddings) {
-    return new Options().explicitPaddings(explicitPaddings);
-  }
-  
-  /**
-   * @param dataFormat Specify the data format of the input and output data. With the
-   * default format "NHWC", the data is stored in the order of:
-   *     [batch, height, width, channels].
-   * Alternatively, the format could be "NCHW", the data storage order of:
-   *     [batch, channels, height, width].
-   */
-  public static Options dataFormat(String dataFormat) {
-    return new Options().dataFormat(dataFormat);
-  }
-  
-  /**
-   * @param dilations 1-D tensor of length 4.  The dilation factor for each dimension of
-   * `input`. If set to k > 1, there will be k-1 skipped cells between each filter
-   * element on that dimension. The dimension order is determined by the value of
-   * `data_format`, see above for details. Dilations in the batch and depth
-   * dimensions must be 1.
-   */
-  public static Options dilations(List<Long> dilations) {
-    return new Options().dilations(dilations);
-  }
-  
-  /**
-   * 4-D with shape according to `data_format`.  For example, if
-   * `data_format` is 'NHWC', output shape is `[batch, in_height,
-   * in_width, in_channels]`.  Gradient w.r.t. the input of the
-   * convolution.
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "DepthwiseConv2dNativeBackpropInput";
-  
-  private Output<T> output;
-  
-  private DepthwiseConv2dNativeBackpropInput(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Dilation2d.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Dilation2d.java
deleted file mode 100644
index e07c5ca6047..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Dilation2d.java
+++ /dev/null
@@ -1,118 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.nn;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Computes the grayscale dilation of 4-D `input` and 3-D `filter` tensors.
- * <p>
- * The `input` tensor has shape `[batch, in_height, in_width, depth]` and the
- * `filter` tensor has shape `[filter_height, filter_width, depth]`, i.e., each
- * input channel is processed independently of the others with its own structuring
- * function. The `output` tensor has shape
- * `[batch, out_height, out_width, depth]`. The spatial dimensions of the output
- * tensor depend on the `padding` algorithm. We currently only support the default
- * "NHWC" `data_format`.
- * <p>
- * In detail, the grayscale morphological 2-D dilation is the max-sum correlation
- * (for consistency with `conv2d`, we use unmirrored filters):
- * <p>
- *     output[b, y, x, c] =
- *        max_{dy, dx} input[b,
- *                           strides[1] * y + rates[1] * dy,
- *                           strides[2] * x + rates[2] * dx,
- *                           c] +
- *                     filter[dy, dx, c]
- * <p>
- * Max-pooling is a special case when the filter has size equal to the pooling
- * kernel size and contains all zeros.
- * <p>
- * Note on duality: The dilation of `input` by the `filter` is equal to the
- * negation of the erosion of `-input` by the reflected `filter`.
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator(group = "nn")
-public final class Dilation2d<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new Dilation2d operation.
-   * 
-   * @param scope current scope
-   * @param input 4-D with shape `[batch, in_height, in_width, depth]`.
-   * @param filter 3-D with shape `[filter_height, filter_width, depth]`.
-   * @param strides The stride of the sliding window for each dimension of the input
-   * tensor. Must be: `[1, stride_height, stride_width, 1]`.
-   * @param rates The input stride for atrous morphological dilation. Must be:
-   * `[1, rate_height, rate_width, 1]`.
-   * @param padding The type of padding algorithm to use.
-   * @return a new instance of Dilation2d
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> Dilation2d<T> create(Scope scope, Operand<T> input, Operand<T> filter, List<Long> strides, List<Long> rates, String padding) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Dilation2D", scope.makeOpName("Dilation2d"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder.addInput(filter.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    long[] stridesArray = new long[strides.size()];
-    for (int i = 0; i < stridesArray.length; ++i) {
-      stridesArray[i] = strides.get(i);
-    }
-    opBuilder.setAttr("strides", stridesArray);
-    long[] ratesArray = new long[rates.size()];
-    for (int i = 0; i < ratesArray.length; ++i) {
-      ratesArray[i] = rates.get(i);
-    }
-    opBuilder.setAttr("rates", ratesArray);
-    opBuilder.setAttr("padding", padding);
-    return new Dilation2d<T>(opBuilder.build());
-  }
-  
-  /**
-   * 4-D with shape `[batch, out_height, out_width, depth]`.
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Dilation2D";
-  
-  private Output<T> output;
-  
-  private Dilation2d(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Dilation2dBackpropFilter.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Dilation2dBackpropFilter.java
deleted file mode 100644
index b9f59b07796..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Dilation2dBackpropFilter.java
+++ /dev/null
@@ -1,96 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.nn;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Computes the gradient of morphological 2-D dilation with respect to the filter.
- * 
- * @param <T> data type for {@code filterBackprop()} output
- */
-@Operator(group = "nn")
-public final class Dilation2dBackpropFilter<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new Dilation2dBackpropFilter operation.
-   * 
-   * @param scope current scope
-   * @param input 4-D with shape `[batch, in_height, in_width, depth]`.
-   * @param filter 3-D with shape `[filter_height, filter_width, depth]`.
-   * @param outBackprop 4-D with shape `[batch, out_height, out_width, depth]`.
-   * @param strides 1-D of length 4. The stride of the sliding window for each dimension of
-   * the input tensor. Must be: `[1, stride_height, stride_width, 1]`.
-   * @param rates 1-D of length 4. The input stride for atrous morphological dilation.
-   * Must be: `[1, rate_height, rate_width, 1]`.
-   * @param padding The type of padding algorithm to use.
-   * @return a new instance of Dilation2dBackpropFilter
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> Dilation2dBackpropFilter<T> create(Scope scope, Operand<T> input, Operand<T> filter, Operand<T> outBackprop, List<Long> strides, List<Long> rates, String padding) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Dilation2DBackpropFilter", scope.makeOpName("Dilation2dBackpropFilter"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder.addInput(filter.asOutput(scope));
-    opBuilder.addInput(outBackprop.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    long[] stridesArray = new long[strides.size()];
-    for (int i = 0; i < stridesArray.length; ++i) {
-      stridesArray[i] = strides.get(i);
-    }
-    opBuilder.setAttr("strides", stridesArray);
-    long[] ratesArray = new long[rates.size()];
-    for (int i = 0; i < ratesArray.length; ++i) {
-      ratesArray[i] = rates.get(i);
-    }
-    opBuilder.setAttr("rates", ratesArray);
-    opBuilder.setAttr("padding", padding);
-    return new Dilation2dBackpropFilter<T>(opBuilder.build());
-  }
-  
-  /**
-   * 3-D with shape `[filter_height, filter_width, depth]`.
-   */
-  public Output<T> filterBackprop() {
-    return filterBackprop;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return filterBackprop;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Dilation2DBackpropFilter";
-  
-  private Output<T> filterBackprop;
-  
-  private Dilation2dBackpropFilter(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    filterBackprop = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Dilation2dBackpropInput.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Dilation2dBackpropInput.java
deleted file mode 100644
index 7634c99cb6a..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Dilation2dBackpropInput.java
+++ /dev/null
@@ -1,96 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.nn;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Computes the gradient of morphological 2-D dilation with respect to the input.
- * 
- * @param <T> data type for {@code inBackprop()} output
- */
-@Operator(group = "nn")
-public final class Dilation2dBackpropInput<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new Dilation2dBackpropInput operation.
-   * 
-   * @param scope current scope
-   * @param input 4-D with shape `[batch, in_height, in_width, depth]`.
-   * @param filter 3-D with shape `[filter_height, filter_width, depth]`.
-   * @param outBackprop 4-D with shape `[batch, out_height, out_width, depth]`.
-   * @param strides 1-D of length 4. The stride of the sliding window for each dimension of
-   * the input tensor. Must be: `[1, stride_height, stride_width, 1]`.
-   * @param rates 1-D of length 4. The input stride for atrous morphological dilation.
-   * Must be: `[1, rate_height, rate_width, 1]`.
-   * @param padding The type of padding algorithm to use.
-   * @return a new instance of Dilation2dBackpropInput
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> Dilation2dBackpropInput<T> create(Scope scope, Operand<T> input, Operand<T> filter, Operand<T> outBackprop, List<Long> strides, List<Long> rates, String padding) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Dilation2DBackpropInput", scope.makeOpName("Dilation2dBackpropInput"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder.addInput(filter.asOutput(scope));
-    opBuilder.addInput(outBackprop.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    long[] stridesArray = new long[strides.size()];
-    for (int i = 0; i < stridesArray.length; ++i) {
-      stridesArray[i] = strides.get(i);
-    }
-    opBuilder.setAttr("strides", stridesArray);
-    long[] ratesArray = new long[rates.size()];
-    for (int i = 0; i < ratesArray.length; ++i) {
-      ratesArray[i] = rates.get(i);
-    }
-    opBuilder.setAttr("rates", ratesArray);
-    opBuilder.setAttr("padding", padding);
-    return new Dilation2dBackpropInput<T>(opBuilder.build());
-  }
-  
-  /**
-   * 4-D with shape `[batch, in_height, in_width, depth]`.
-   */
-  public Output<T> inBackprop() {
-    return inBackprop;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return inBackprop;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Dilation2DBackpropInput";
-  
-  private Output<T> inBackprop;
-  
-  private Dilation2dBackpropInput(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    inBackprop = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Elu.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Elu.java
deleted file mode 100644
index 6c855873ce4..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Elu.java
+++ /dev/null
@@ -1,77 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.nn;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Computes exponential linear: `exp(features) - 1` if < 0, `features` otherwise.
- * <p>
- * See [Fast and Accurate Deep Network Learning by Exponential Linear Units (ELUs)
- * ](http://arxiv.org/abs/1511.07289)
- * 
- * @param <T> data type for {@code activations()} output
- */
-@Operator(group = "nn")
-public final class Elu<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new Elu operation.
-   * 
-   * @param scope current scope
-   * @param features 
-   * @return a new instance of Elu
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> Elu<T> create(Scope scope, Operand<T> features) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Elu", scope.makeOpName("Elu"));
-    opBuilder.addInput(features.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new Elu<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> activations() {
-    return activations;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return activations;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Elu";
-  
-  private Output<T> activations;
-  
-  private Elu(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    activations = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/EluGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/EluGrad.java
deleted file mode 100644
index 9776b7c2f23..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/EluGrad.java
+++ /dev/null
@@ -1,77 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.nn;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Computes gradients for the exponential linear (Elu) operation.
- * 
- * @param <T> data type for {@code backprops()} output
- */
-public final class EluGrad<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new EluGrad operation.
-   * 
-   * @param scope current scope
-   * @param gradients The backpropagated gradients to the corresponding Elu operation.
-   * @param outputs The outputs of the corresponding Elu operation.
-   * @return a new instance of EluGrad
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> EluGrad<T> create(Scope scope, Operand<T> gradients, Operand<T> outputs) {
-    OperationBuilder opBuilder = scope.env().opBuilder("EluGrad", scope.makeOpName("EluGrad"));
-    opBuilder.addInput(gradients.asOutput(scope));
-    opBuilder.addInput(outputs.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new EluGrad<T>(opBuilder.build());
-  }
-  
-  /**
-   * The gradients: `gradients * (outputs + 1)` if outputs < 0,
-   * `gradients` otherwise.
-   */
-  public Output<T> backprops() {
-    return backprops;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return backprops;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "EluGrad";
-  
-  private Output<T> backprops;
-  
-  private EluGrad(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    backprops = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/FixedUnigramCandidateSampler.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/FixedUnigramCandidateSampler.java
deleted file mode 100644
index 50255afbf49..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/FixedUnigramCandidateSampler.java
+++ /dev/null
@@ -1,329 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.nn;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.TInt64;
-
-/**
- * Generates labels for candidate sampling with a learned unigram distribution.
- * <p>
- * A unigram sampler could use a fixed unigram distribution read from a
- * file or passed in as an in-memory array instead of building up the distribution
- * from data on the fly. There is also an option to skew the distribution by
- * applying a distortion power to the weights.
- * <p>
- * The vocabulary file should be in CSV-like format, with the last field
- * being the weight associated with the word.
- * <p>
- * For each batch, this op picks a single set of sampled candidate labels.
- * <p>
- * The advantages of sampling candidates per-batch are simplicity and the
- * possibility of efficient dense matrix multiplication. The disadvantage is that
- * the sampled candidates must be chosen independently of the context and of the
- * true labels.
- */
-@Operator(group = "nn")
-public final class FixedUnigramCandidateSampler extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.nn.FixedUnigramCandidateSampler}
-   */
-  public static class Options {
-    
-    /**
-     * @param vocabFile Each valid line in this file (which should have a CSV-like format)
-     * corresponds to a valid word ID. IDs are in sequential order, starting from
-     * num_reserved_ids. The last entry in each line is expected to be a value
-     * corresponding to the count or relative probability. Exactly one of vocab_file
-     * and unigrams needs to be passed to this op.
-     */
-    public Options vocabFile(String vocabFile) {
-      this.vocabFile = vocabFile;
-      return this;
-    }
-    
-    /**
-     * @param distortion The distortion is used to skew the unigram probability distribution.
-     * Each weight is first raised to the distortion's power before adding to the
-     * internal unigram distribution. As a result, distortion = 1.0 gives regular
-     * unigram sampling (as defined by the vocab file), and distortion = 0.0 gives
-     * a uniform distribution.
-     */
-    public Options distortion(Float distortion) {
-      this.distortion = distortion;
-      return this;
-    }
-    
-    /**
-     * @param numReservedIds Optionally some reserved IDs can be added in the range [0,
-     * ..., num_reserved_ids) by the users. One use case is that a special unknown
-     * word token is used as ID 0. These IDs will have a sampling probability of 0.
-     */
-    public Options numReservedIds(Long numReservedIds) {
-      this.numReservedIds = numReservedIds;
-      return this;
-    }
-    
-    /**
-     * @param numShards A sampler can be used to sample from a subset of the original range
-     * in order to speed up the whole computation through parallelism. This parameter
-     * (together with 'shard') indicates the number of partitions that are being
-     * used in the overall computation.
-     */
-    public Options numShards(Long numShards) {
-      this.numShards = numShards;
-      return this;
-    }
-    
-    /**
-     * @param shard A sampler can be used to sample from a subset of the original range
-     * in order to speed up the whole computation through parallelism. This parameter
-     * (together with 'num_shards') indicates the particular partition number of a
-     * sampler op, when partitioning is being used.
-     */
-    public Options shard(Long shard) {
-      this.shard = shard;
-      return this;
-    }
-    
-    /**
-     * @param unigrams A list of unigram counts or probabilities, one per ID in sequential
-     * order. Exactly one of vocab_file and unigrams should be passed to this op.
-     */
-    public Options unigrams(List<Float> unigrams) {
-      this.unigrams = unigrams;
-      return this;
-    }
-    
-    /**
-     * @param seed If either seed or seed2 are set to be non-zero, the random number
-     * generator is seeded by the given seed.  Otherwise, it is seeded by a
-     * random seed.
-     */
-    public Options seed(Long seed) {
-      this.seed = seed;
-      return this;
-    }
-    
-    /**
-     * @param seed2 An second seed to avoid seed collision.
-     */
-    public Options seed2(Long seed2) {
-      this.seed2 = seed2;
-      return this;
-    }
-    
-    private String vocabFile;
-    private Float distortion;
-    private Long numReservedIds;
-    private Long numShards;
-    private Long shard;
-    private List<Float> unigrams;
-    private Long seed;
-    private Long seed2;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new FixedUnigramCandidateSampler operation.
-   * 
-   * @param scope current scope
-   * @param trueClasses A batch_size * num_true matrix, in which each row contains the
-   * IDs of the num_true target_classes in the corresponding original label.
-   * @param numTrue Number of true labels per context.
-   * @param numSampled Number of candidates to randomly sample.
-   * @param unique If unique is true, we sample with rejection, so that all sampled
-   * candidates in a batch are unique. This requires some approximation to
-   * estimate the post-rejection sampling probabilities.
-   * @param rangeMax The sampler will sample integers from the interval [0, range_max).
-   * @param options carries optional attributes values
-   * @return a new instance of FixedUnigramCandidateSampler
-   */
-  @Endpoint(describeByClass = true)
-  public static FixedUnigramCandidateSampler create(Scope scope, Operand<TInt64> trueClasses, Long numTrue, Long numSampled, Boolean unique, Long rangeMax, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("FixedUnigramCandidateSampler", scope.makeOpName("FixedUnigramCandidateSampler"));
-    opBuilder.addInput(trueClasses.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("num_true", numTrue);
-    opBuilder.setAttr("num_sampled", numSampled);
-    opBuilder.setAttr("unique", unique);
-    opBuilder.setAttr("range_max", rangeMax);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.vocabFile != null) {
-          opBuilder.setAttr("vocab_file", opts.vocabFile);
-        }
-        if (opts.distortion != null) {
-          opBuilder.setAttr("distortion", opts.distortion);
-        }
-        if (opts.numReservedIds != null) {
-          opBuilder.setAttr("num_reserved_ids", opts.numReservedIds);
-        }
-        if (opts.numShards != null) {
-          opBuilder.setAttr("num_shards", opts.numShards);
-        }
-        if (opts.shard != null) {
-          opBuilder.setAttr("shard", opts.shard);
-        }
-        if (opts.unigrams != null) {
-          float[] unigramsArray = new float[opts.unigrams.size()];
-          for (int i = 0; i < unigramsArray.length; ++i) {
-            unigramsArray[i] = opts.unigrams.get(i);
-          }
-          opBuilder.setAttr("unigrams", unigramsArray);
-        }
-        if (opts.seed != null) {
-          opBuilder.setAttr("seed", opts.seed);
-        }
-        if (opts.seed2 != null) {
-          opBuilder.setAttr("seed2", opts.seed2);
-        }
-      }
-    }
-    return new FixedUnigramCandidateSampler(opBuilder.build());
-  }
-  
-  /**
-   * @param vocabFile Each valid line in this file (which should have a CSV-like format)
-   * corresponds to a valid word ID. IDs are in sequential order, starting from
-   * num_reserved_ids. The last entry in each line is expected to be a value
-   * corresponding to the count or relative probability. Exactly one of vocab_file
-   * and unigrams needs to be passed to this op.
-   */
-  public static Options vocabFile(String vocabFile) {
-    return new Options().vocabFile(vocabFile);
-  }
-  
-  /**
-   * @param distortion The distortion is used to skew the unigram probability distribution.
-   * Each weight is first raised to the distortion's power before adding to the
-   * internal unigram distribution. As a result, distortion = 1.0 gives regular
-   * unigram sampling (as defined by the vocab file), and distortion = 0.0 gives
-   * a uniform distribution.
-   */
-  public static Options distortion(Float distortion) {
-    return new Options().distortion(distortion);
-  }
-  
-  /**
-   * @param numReservedIds Optionally some reserved IDs can be added in the range [0,
-   * ..., num_reserved_ids) by the users. One use case is that a special unknown
-   * word token is used as ID 0. These IDs will have a sampling probability of 0.
-   */
-  public static Options numReservedIds(Long numReservedIds) {
-    return new Options().numReservedIds(numReservedIds);
-  }
-  
-  /**
-   * @param numShards A sampler can be used to sample from a subset of the original range
-   * in order to speed up the whole computation through parallelism. This parameter
-   * (together with 'shard') indicates the number of partitions that are being
-   * used in the overall computation.
-   */
-  public static Options numShards(Long numShards) {
-    return new Options().numShards(numShards);
-  }
-  
-  /**
-   * @param shard A sampler can be used to sample from a subset of the original range
-   * in order to speed up the whole computation through parallelism. This parameter
-   * (together with 'num_shards') indicates the particular partition number of a
-   * sampler op, when partitioning is being used.
-   */
-  public static Options shard(Long shard) {
-    return new Options().shard(shard);
-  }
-  
-  /**
-   * @param unigrams A list of unigram counts or probabilities, one per ID in sequential
-   * order. Exactly one of vocab_file and unigrams should be passed to this op.
-   */
-  public static Options unigrams(List<Float> unigrams) {
-    return new Options().unigrams(unigrams);
-  }
-  
-  /**
-   * @param seed If either seed or seed2 are set to be non-zero, the random number
-   * generator is seeded by the given seed.  Otherwise, it is seeded by a
-   * random seed.
-   */
-  public static Options seed(Long seed) {
-    return new Options().seed(seed);
-  }
-  
-  /**
-   * @param seed2 An second seed to avoid seed collision.
-   */
-  public static Options seed2(Long seed2) {
-    return new Options().seed2(seed2);
-  }
-  
-  /**
-   * A vector of length num_sampled, in which each element is
-   * the ID of a sampled candidate.
-   */
-  public Output<TInt64> sampledCandidates() {
-    return sampledCandidates;
-  }
-  
-  /**
-   * A batch_size * num_true matrix, representing
-   * the number of times each candidate is expected to occur in a batch
-   * of sampled candidates. If unique=true, then this is a probability.
-   */
-  public Output<TFloat32> trueExpectedCount() {
-    return trueExpectedCount;
-  }
-  
-  /**
-   * A vector of length num_sampled, for each sampled
-   * candidate representing the number of times the candidate is expected
-   * to occur in a batch of sampled candidates.  If unique=true, then this is a
-   * probability.
-   */
-  public Output<TFloat32> sampledExpectedCount() {
-    return sampledExpectedCount;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "FixedUnigramCandidateSampler";
-  
-  private Output<TInt64> sampledCandidates;
-  private Output<TFloat32> trueExpectedCount;
-  private Output<TFloat32> sampledExpectedCount;
-  
-  private FixedUnigramCandidateSampler(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    sampledCandidates = operation.output(outputIdx++);
-    trueExpectedCount = operation.output(outputIdx++);
-    sampledExpectedCount = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/FractionalAvgPool.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/FractionalAvgPool.java
deleted file mode 100644
index 2c3888ab6a5..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/FractionalAvgPool.java
+++ /dev/null
@@ -1,246 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.nn;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Performs fractional average pooling on the input.
- * <p>
- * Fractional average pooling is similar to Fractional max pooling in the pooling
- * region generation step. The only difference is that after pooling regions are
- * generated, a mean operation is performed instead of a max operation in each
- * pooling region.
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator(group = "nn")
-public final class FractionalAvgPool<T extends TNumber> extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.nn.FractionalAvgPool}
-   */
-  public static class Options {
-    
-    /**
-     * @param pseudoRandom When set to True, generates the pooling sequence in a
-     * pseudorandom fashion, otherwise, in a random fashion. Check paper [Benjamin
-     * Graham, Fractional Max-Pooling](http://arxiv.org/abs/1412.6071) for
-     * difference between pseudorandom and random.
-     */
-    public Options pseudoRandom(Boolean pseudoRandom) {
-      this.pseudoRandom = pseudoRandom;
-      return this;
-    }
-    
-    /**
-     * @param overlapping When set to True, it means when pooling, the values at the boundary
-     * of adjacent pooling cells are used by both cells. For example:
-     * <p>
-     * `index  0  1  2  3  4`
-     * <p>
-     * `value  20 5  16 3  7`
-     * <p>
-     * If the pooling sequence is [0, 2, 4], then 16, at index 2 will be used twice.
-     * The result would be [41/3, 26/3] for fractional avg pooling.
-     */
-    public Options overlapping(Boolean overlapping) {
-      this.overlapping = overlapping;
-      return this;
-    }
-    
-    /**
-     * @param deterministic When set to True, a fixed pooling region will be used when
-     * iterating over a FractionalAvgPool node in the computation graph. Mainly used
-     * in unit test to make FractionalAvgPool deterministic.
-     */
-    public Options deterministic(Boolean deterministic) {
-      this.deterministic = deterministic;
-      return this;
-    }
-    
-    /**
-     * @param seed If either seed or seed2 are set to be non-zero, the random number
-     * generator is seeded by the given seed.  Otherwise, it is seeded by a
-     * random seed.
-     */
-    public Options seed(Long seed) {
-      this.seed = seed;
-      return this;
-    }
-    
-    /**
-     * @param seed2 An second seed to avoid seed collision.
-     */
-    public Options seed2(Long seed2) {
-      this.seed2 = seed2;
-      return this;
-    }
-    
-    private Boolean pseudoRandom;
-    private Boolean overlapping;
-    private Boolean deterministic;
-    private Long seed;
-    private Long seed2;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new FractionalAvgPool operation.
-   * 
-   * @param scope current scope
-   * @param value 4-D with shape `[batch, height, width, channels]`.
-   * @param poolingRatio Pooling ratio for each dimension of `value`, currently only
-   * supports row and col dimension and should be >= 1.0. For example, a valid
-   * pooling ratio looks like [1.0, 1.44, 1.73, 1.0]. The first and last elements
-   * must be 1.0 because we don't allow pooling on batch and channels
-   * dimensions. 1.44 and 1.73 are pooling ratio on height and width dimensions
-   * respectively.
-   * @param options carries optional attributes values
-   * @return a new instance of FractionalAvgPool
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> FractionalAvgPool<T> create(Scope scope, Operand<T> value, List<Float> poolingRatio, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("FractionalAvgPool", scope.makeOpName("FractionalAvgPool"));
-    opBuilder.addInput(value.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    float[] poolingRatioArray = new float[poolingRatio.size()];
-    for (int i = 0; i < poolingRatioArray.length; ++i) {
-      poolingRatioArray[i] = poolingRatio.get(i);
-    }
-    opBuilder.setAttr("pooling_ratio", poolingRatioArray);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.pseudoRandom != null) {
-          opBuilder.setAttr("pseudo_random", opts.pseudoRandom);
-        }
-        if (opts.overlapping != null) {
-          opBuilder.setAttr("overlapping", opts.overlapping);
-        }
-        if (opts.deterministic != null) {
-          opBuilder.setAttr("deterministic", opts.deterministic);
-        }
-        if (opts.seed != null) {
-          opBuilder.setAttr("seed", opts.seed);
-        }
-        if (opts.seed2 != null) {
-          opBuilder.setAttr("seed2", opts.seed2);
-        }
-      }
-    }
-    return new FractionalAvgPool<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param pseudoRandom When set to True, generates the pooling sequence in a
-   * pseudorandom fashion, otherwise, in a random fashion. Check paper [Benjamin
-   * Graham, Fractional Max-Pooling](http://arxiv.org/abs/1412.6071) for
-   * difference between pseudorandom and random.
-   */
-  public static Options pseudoRandom(Boolean pseudoRandom) {
-    return new Options().pseudoRandom(pseudoRandom);
-  }
-  
-  /**
-   * @param overlapping When set to True, it means when pooling, the values at the boundary
-   * of adjacent pooling cells are used by both cells. For example:
-   * <p>
-   * `index  0  1  2  3  4`
-   * <p>
-   * `value  20 5  16 3  7`
-   * <p>
-   * If the pooling sequence is [0, 2, 4], then 16, at index 2 will be used twice.
-   * The result would be [41/3, 26/3] for fractional avg pooling.
-   */
-  public static Options overlapping(Boolean overlapping) {
-    return new Options().overlapping(overlapping);
-  }
-  
-  /**
-   * @param deterministic When set to True, a fixed pooling region will be used when
-   * iterating over a FractionalAvgPool node in the computation graph. Mainly used
-   * in unit test to make FractionalAvgPool deterministic.
-   */
-  public static Options deterministic(Boolean deterministic) {
-    return new Options().deterministic(deterministic);
-  }
-  
-  /**
-   * @param seed If either seed or seed2 are set to be non-zero, the random number
-   * generator is seeded by the given seed.  Otherwise, it is seeded by a
-   * random seed.
-   */
-  public static Options seed(Long seed) {
-    return new Options().seed(seed);
-  }
-  
-  /**
-   * @param seed2 An second seed to avoid seed collision.
-   */
-  public static Options seed2(Long seed2) {
-    return new Options().seed2(seed2);
-  }
-  
-  /**
-   * output tensor after fractional avg pooling.
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  /**
-   * row pooling sequence, needed to calculate gradient.
-   */
-  public Output<TInt64> rowPoolingSequence() {
-    return rowPoolingSequence;
-  }
-  
-  /**
-   * column pooling sequence, needed to calculate gradient.
-   */
-  public Output<TInt64> colPoolingSequence() {
-    return colPoolingSequence;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "FractionalAvgPool";
-  
-  private Output<T> output;
-  private Output<TInt64> rowPoolingSequence;
-  private Output<TInt64> colPoolingSequence;
-  
-  private FractionalAvgPool(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-    rowPoolingSequence = operation.output(outputIdx++);
-    colPoolingSequence = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/FractionalAvgPoolGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/FractionalAvgPoolGrad.java
deleted file mode 100644
index 0d052aee581..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/FractionalAvgPoolGrad.java
+++ /dev/null
@@ -1,140 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.nn;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Computes gradient of the FractionalAvgPool function.
- * <p>
- * Unlike FractionalMaxPoolGrad, we don't need to find arg_max for
- * FractionalAvgPoolGrad, we just need to evenly back-propagate each element of
- * out_backprop to those indices that form the same pooling cell. Therefore, we
- * just need to know the shape of original input tensor, instead of the whole
- * tensor.
- * 
- * @param <T> data type for {@code output()} output
- */
-public final class FractionalAvgPoolGrad<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.nn.FractionalAvgPoolGrad}
-   */
-  public static class Options {
-    
-    /**
-     * @param overlapping When set to True, it means when pooling, the values at the boundary
-     * of adjacent pooling cells are used by both cells. For example:
-     * <p>
-     * `index  0  1  2  3  4`
-     * <p>
-     * `value  20 5  16 3  7`
-     * <p>
-     * If the pooling sequence is [0, 2, 4], then 16, at index 2 will be used twice.
-     * The result would be [41/3, 26/3] for fractional avg pooling.
-     */
-    public Options overlapping(Boolean overlapping) {
-      this.overlapping = overlapping;
-      return this;
-    }
-    
-    private Boolean overlapping;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new FractionalAvgPoolGrad operation.
-   * 
-   * @param scope current scope
-   * @param origInputTensorShape Original input tensor shape for `fractional_avg_pool`
-   * @param outBackprop 4-D with shape `[batch, height, width, channels]`.  Gradients
-   * w.r.t. the output of `fractional_avg_pool`.
-   * @param rowPoolingSequence row pooling sequence, form pooling region with
-   * col_pooling_sequence.
-   * @param colPoolingSequence column pooling sequence, form pooling region with
-   * row_pooling sequence.
-   * @param options carries optional attributes values
-   * @return a new instance of FractionalAvgPoolGrad
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> FractionalAvgPoolGrad<T> create(Scope scope, Operand<TInt64> origInputTensorShape, Operand<T> outBackprop, Operand<TInt64> rowPoolingSequence, Operand<TInt64> colPoolingSequence, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("FractionalAvgPoolGrad", scope.makeOpName("FractionalAvgPoolGrad"));
-    opBuilder.addInput(origInputTensorShape.asOutput(scope));
-    opBuilder.addInput(outBackprop.asOutput(scope));
-    opBuilder.addInput(rowPoolingSequence.asOutput(scope));
-    opBuilder.addInput(colPoolingSequence.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.overlapping != null) {
-          opBuilder.setAttr("overlapping", opts.overlapping);
-        }
-      }
-    }
-    return new FractionalAvgPoolGrad<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param overlapping When set to True, it means when pooling, the values at the boundary
-   * of adjacent pooling cells are used by both cells. For example:
-   * <p>
-   * `index  0  1  2  3  4`
-   * <p>
-   * `value  20 5  16 3  7`
-   * <p>
-   * If the pooling sequence is [0, 2, 4], then 16, at index 2 will be used twice.
-   * The result would be [41/3, 26/3] for fractional avg pooling.
-   */
-  public static Options overlapping(Boolean overlapping) {
-    return new Options().overlapping(overlapping);
-  }
-  
-  /**
-   * 4-D.  Gradients w.r.t. the input of `fractional_avg_pool`.
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "FractionalAvgPoolGrad";
-  
-  private Output<T> output;
-  
-  private FractionalAvgPoolGrad(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/FractionalMaxPool.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/FractionalMaxPool.java
deleted file mode 100644
index 234ec684451..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/FractionalMaxPool.java
+++ /dev/null
@@ -1,270 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.nn;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Performs fractional max pooling on the input.
- * <p>
- * Fractional max pooling is slightly different than regular max pooling.  In
- * regular max pooling, you downsize an input set by taking the maximum value of
- * smaller N x N subsections of the set (often 2x2), and try to reduce the set by
- * a factor of N, where N is an integer.  Fractional max pooling, as you might
- * expect from the word "fractional", means that the overall reduction ratio N
- * does not have to be an integer.
- * <p>
- * The sizes of the pooling regions are generated randomly but are fairly uniform.
- * For example, let's look at the height dimension, and the constraints on the
- * list of rows that will be pool boundaries.
- * <p>
- * First we define the following:
- * <p>
- * 1.  input_row_length : the number of rows from the input set
- * 2.  output_row_length : which will be smaller than the input
- * 3.  alpha = input_row_length / output_row_length : our reduction ratio
- * 4.  K = floor(alpha)
- * 5.  row_pooling_sequence : this is the result list of pool boundary rows
- * <p>
- * Then, row_pooling_sequence should satisfy:
- * <p>
- * 1.  a[0] = 0 : the first value of the sequence is 0
- * 2.  a[end] = input_row_length : the last value of the sequence is the size
- * 3.  K <= (a[i+1] - a[i]) <= K+1 : all intervals are K or K+1 size
- * 4.  length(row_pooling_sequence) = output_row_length+1
- * <p>
- * For more details on fractional max pooling, see this paper:
- * [Benjamin Graham, Fractional Max-Pooling](http://arxiv.org/abs/1412.6071)
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator(group = "nn")
-public final class FractionalMaxPool<T extends TNumber> extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.nn.FractionalMaxPool}
-   */
-  public static class Options {
-    
-    /**
-     * @param pseudoRandom When set to True, generates the pooling sequence in a
-     * pseudorandom fashion, otherwise, in a random fashion. Check paper [Benjamin
-     * Graham, Fractional Max-Pooling](http://arxiv.org/abs/1412.6071) for
-     * difference between pseudorandom and random.
-     */
-    public Options pseudoRandom(Boolean pseudoRandom) {
-      this.pseudoRandom = pseudoRandom;
-      return this;
-    }
-    
-    /**
-     * @param overlapping When set to True, it means when pooling, the values at the boundary
-     * of adjacent pooling cells are used by both cells. For example:
-     * <p>
-     * `index  0  1  2  3  4`
-     * <p>
-     * `value  20 5  16 3  7`
-     * <p>
-     * If the pooling sequence is [0, 2, 4], then 16, at index 2 will be used twice.
-     * The result would be [20, 16] for fractional max pooling.
-     */
-    public Options overlapping(Boolean overlapping) {
-      this.overlapping = overlapping;
-      return this;
-    }
-    
-    /**
-     * @param deterministic When set to True, a fixed pooling region will be used when
-     * iterating over a FractionalMaxPool node in the computation graph. Mainly used
-     * in unit test to make FractionalMaxPool deterministic.
-     */
-    public Options deterministic(Boolean deterministic) {
-      this.deterministic = deterministic;
-      return this;
-    }
-    
-    /**
-     * @param seed If either seed or seed2 are set to be non-zero, the random number
-     * generator is seeded by the given seed.  Otherwise, it is seeded by a
-     * random seed.
-     */
-    public Options seed(Long seed) {
-      this.seed = seed;
-      return this;
-    }
-    
-    /**
-     * @param seed2 An second seed to avoid seed collision.
-     */
-    public Options seed2(Long seed2) {
-      this.seed2 = seed2;
-      return this;
-    }
-    
-    private Boolean pseudoRandom;
-    private Boolean overlapping;
-    private Boolean deterministic;
-    private Long seed;
-    private Long seed2;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new FractionalMaxPool operation.
-   * 
-   * @param scope current scope
-   * @param value 4-D with shape `[batch, height, width, channels]`.
-   * @param poolingRatio Pooling ratio for each dimension of `value`, currently only
-   * supports row and col dimension and should be >= 1.0. For example, a valid
-   * pooling ratio looks like [1.0, 1.44, 1.73, 1.0]. The first and last elements
-   * must be 1.0 because we don't allow pooling on batch and channels
-   * dimensions. 1.44 and 1.73 are pooling ratio on height and width dimensions
-   * respectively.
-   * @param options carries optional attributes values
-   * @return a new instance of FractionalMaxPool
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> FractionalMaxPool<T> create(Scope scope, Operand<T> value, List<Float> poolingRatio, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("FractionalMaxPool", scope.makeOpName("FractionalMaxPool"));
-    opBuilder.addInput(value.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    float[] poolingRatioArray = new float[poolingRatio.size()];
-    for (int i = 0; i < poolingRatioArray.length; ++i) {
-      poolingRatioArray[i] = poolingRatio.get(i);
-    }
-    opBuilder.setAttr("pooling_ratio", poolingRatioArray);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.pseudoRandom != null) {
-          opBuilder.setAttr("pseudo_random", opts.pseudoRandom);
-        }
-        if (opts.overlapping != null) {
-          opBuilder.setAttr("overlapping", opts.overlapping);
-        }
-        if (opts.deterministic != null) {
-          opBuilder.setAttr("deterministic", opts.deterministic);
-        }
-        if (opts.seed != null) {
-          opBuilder.setAttr("seed", opts.seed);
-        }
-        if (opts.seed2 != null) {
-          opBuilder.setAttr("seed2", opts.seed2);
-        }
-      }
-    }
-    return new FractionalMaxPool<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param pseudoRandom When set to True, generates the pooling sequence in a
-   * pseudorandom fashion, otherwise, in a random fashion. Check paper [Benjamin
-   * Graham, Fractional Max-Pooling](http://arxiv.org/abs/1412.6071) for
-   * difference between pseudorandom and random.
-   */
-  public static Options pseudoRandom(Boolean pseudoRandom) {
-    return new Options().pseudoRandom(pseudoRandom);
-  }
-  
-  /**
-   * @param overlapping When set to True, it means when pooling, the values at the boundary
-   * of adjacent pooling cells are used by both cells. For example:
-   * <p>
-   * `index  0  1  2  3  4`
-   * <p>
-   * `value  20 5  16 3  7`
-   * <p>
-   * If the pooling sequence is [0, 2, 4], then 16, at index 2 will be used twice.
-   * The result would be [20, 16] for fractional max pooling.
-   */
-  public static Options overlapping(Boolean overlapping) {
-    return new Options().overlapping(overlapping);
-  }
-  
-  /**
-   * @param deterministic When set to True, a fixed pooling region will be used when
-   * iterating over a FractionalMaxPool node in the computation graph. Mainly used
-   * in unit test to make FractionalMaxPool deterministic.
-   */
-  public static Options deterministic(Boolean deterministic) {
-    return new Options().deterministic(deterministic);
-  }
-  
-  /**
-   * @param seed If either seed or seed2 are set to be non-zero, the random number
-   * generator is seeded by the given seed.  Otherwise, it is seeded by a
-   * random seed.
-   */
-  public static Options seed(Long seed) {
-    return new Options().seed(seed);
-  }
-  
-  /**
-   * @param seed2 An second seed to avoid seed collision.
-   */
-  public static Options seed2(Long seed2) {
-    return new Options().seed2(seed2);
-  }
-  
-  /**
-   * output tensor after fractional max pooling.
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  /**
-   * row pooling sequence, needed to calculate gradient.
-   */
-  public Output<TInt64> rowPoolingSequence() {
-    return rowPoolingSequence;
-  }
-  
-  /**
-   * column pooling sequence, needed to calculate gradient.
-   */
-  public Output<TInt64> colPoolingSequence() {
-    return colPoolingSequence;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "FractionalMaxPool";
-  
-  private Output<T> output;
-  private Output<TInt64> rowPoolingSequence;
-  private Output<TInt64> colPoolingSequence;
-  
-  private FractionalMaxPool(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-    rowPoolingSequence = operation.output(outputIdx++);
-    colPoolingSequence = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/FractionalMaxPoolGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/FractionalMaxPoolGrad.java
deleted file mode 100644
index 3a9be5c7cd2..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/FractionalMaxPoolGrad.java
+++ /dev/null
@@ -1,136 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.nn;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Computes gradient of the FractionalMaxPool function.
- * 
- * @param <T> data type for {@code output()} output
- */
-public final class FractionalMaxPoolGrad<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.nn.FractionalMaxPoolGrad}
-   */
-  public static class Options {
-    
-    /**
-     * @param overlapping When set to True, it means when pooling, the values at the boundary
-     * of adjacent pooling cells are used by both cells. For example:
-     * <p>
-     * `index  0  1  2  3  4`
-     * <p>
-     * `value  20 5  16 3  7`
-     * <p>
-     * If the pooling sequence is [0, 2, 4], then 16, at index 2 will be used twice.
-     * The result would be [20, 16] for fractional max pooling.
-     */
-    public Options overlapping(Boolean overlapping) {
-      this.overlapping = overlapping;
-      return this;
-    }
-    
-    private Boolean overlapping;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new FractionalMaxPoolGrad operation.
-   * 
-   * @param scope current scope
-   * @param origInput Original input for `fractional_max_pool`
-   * @param origOutput Original output for `fractional_max_pool`
-   * @param outBackprop 4-D with shape `[batch, height, width, channels]`.  Gradients
-   * w.r.t. the output of `fractional_max_pool`.
-   * @param rowPoolingSequence row pooling sequence, form pooling region with
-   * col_pooling_sequence.
-   * @param colPoolingSequence column pooling sequence, form pooling region with
-   * row_pooling sequence.
-   * @param options carries optional attributes values
-   * @return a new instance of FractionalMaxPoolGrad
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> FractionalMaxPoolGrad<T> create(Scope scope, Operand<T> origInput, Operand<T> origOutput, Operand<T> outBackprop, Operand<TInt64> rowPoolingSequence, Operand<TInt64> colPoolingSequence, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("FractionalMaxPoolGrad", scope.makeOpName("FractionalMaxPoolGrad"));
-    opBuilder.addInput(origInput.asOutput(scope));
-    opBuilder.addInput(origOutput.asOutput(scope));
-    opBuilder.addInput(outBackprop.asOutput(scope));
-    opBuilder.addInput(rowPoolingSequence.asOutput(scope));
-    opBuilder.addInput(colPoolingSequence.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.overlapping != null) {
-          opBuilder.setAttr("overlapping", opts.overlapping);
-        }
-      }
-    }
-    return new FractionalMaxPoolGrad<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param overlapping When set to True, it means when pooling, the values at the boundary
-   * of adjacent pooling cells are used by both cells. For example:
-   * <p>
-   * `index  0  1  2  3  4`
-   * <p>
-   * `value  20 5  16 3  7`
-   * <p>
-   * If the pooling sequence is [0, 2, 4], then 16, at index 2 will be used twice.
-   * The result would be [20, 16] for fractional max pooling.
-   */
-  public static Options overlapping(Boolean overlapping) {
-    return new Options().overlapping(overlapping);
-  }
-  
-  /**
-   * 4-D.  Gradients w.r.t. the input of `fractional_max_pool`.
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "FractionalMaxPoolGrad";
-  
-  private Output<T> output;
-  
-  private FractionalMaxPoolGrad(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/FusedBatchNorm.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/FusedBatchNorm.java
deleted file mode 100644
index 1e61f4e0060..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/FusedBatchNorm.java
+++ /dev/null
@@ -1,227 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.nn;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Batch normalization.
- * <p>
- * Note that the size of 4D Tensors are defined by either "NHWC" or "NCHW".
- * The size of 1D Tensors matches the dimension C of the 4D Tensors.
- * 
- * @param <T> data type for {@code y()} output
- * @param <U> data type for {@code batchMean()} output
- */
-@Operator(group = "nn")
-public final class FusedBatchNorm<T extends TNumber, U extends TNumber> extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.nn.FusedBatchNorm}
-   */
-  public static class Options {
-    
-    /**
-     * @param epsilon A small float number added to the variance of x.
-     */
-    public Options epsilon(Float epsilon) {
-      this.epsilon = epsilon;
-      return this;
-    }
-    
-    /**
-     * @param exponentialAvgFactor 
-     */
-    public Options exponentialAvgFactor(Float exponentialAvgFactor) {
-      this.exponentialAvgFactor = exponentialAvgFactor;
-      return this;
-    }
-    
-    /**
-     * @param dataFormat The data format for x and y. Either "NHWC" (default) or "NCHW".
-     */
-    public Options dataFormat(String dataFormat) {
-      this.dataFormat = dataFormat;
-      return this;
-    }
-    
-    /**
-     * @param isTraining A bool value to indicate the operation is for training (default)
-     * or inference.
-     */
-    public Options isTraining(Boolean isTraining) {
-      this.isTraining = isTraining;
-      return this;
-    }
-    
-    private Float epsilon;
-    private Float exponentialAvgFactor;
-    private String dataFormat;
-    private Boolean isTraining;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new FusedBatchNorm operation.
-   * 
-   * @param scope current scope
-   * @param x A 4D Tensor for input data.
-   * @param scale A 1D Tensor for scaling factor, to scale the normalized x.
-   * @param offset A 1D Tensor for offset, to shift to the normalized x.
-   * @param mean A 1D Tensor for population mean. Used for inference only;
-   * must be empty for training.
-   * @param variance A 1D Tensor for population variance. Used for inference only;
-   * must be empty for training.
-   * @param options carries optional attributes values
-   * @return a new instance of FusedBatchNorm
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber, U extends TNumber> FusedBatchNorm<T, U> create(Scope scope, Operand<T> x, Operand<U> scale, Operand<U> offset, Operand<U> mean, Operand<U> variance, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("FusedBatchNormV3", scope.makeOpName("FusedBatchNorm"));
-    opBuilder.addInput(x.asOutput(scope));
-    opBuilder.addInput(scale.asOutput(scope));
-    opBuilder.addInput(offset.asOutput(scope));
-    opBuilder.addInput(mean.asOutput(scope));
-    opBuilder.addInput(variance.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.epsilon != null) {
-          opBuilder.setAttr("epsilon", opts.epsilon);
-        }
-        if (opts.exponentialAvgFactor != null) {
-          opBuilder.setAttr("exponential_avg_factor", opts.exponentialAvgFactor);
-        }
-        if (opts.dataFormat != null) {
-          opBuilder.setAttr("data_format", opts.dataFormat);
-        }
-        if (opts.isTraining != null) {
-          opBuilder.setAttr("is_training", opts.isTraining);
-        }
-      }
-    }
-    return new FusedBatchNorm<T, U>(opBuilder.build());
-  }
-  
-  /**
-   * @param epsilon A small float number added to the variance of x.
-   */
-  public static Options epsilon(Float epsilon) {
-    return new Options().epsilon(epsilon);
-  }
-  
-  /**
-   * @param exponentialAvgFactor 
-   */
-  public static Options exponentialAvgFactor(Float exponentialAvgFactor) {
-    return new Options().exponentialAvgFactor(exponentialAvgFactor);
-  }
-  
-  /**
-   * @param dataFormat The data format for x and y. Either "NHWC" (default) or "NCHW".
-   */
-  public static Options dataFormat(String dataFormat) {
-    return new Options().dataFormat(dataFormat);
-  }
-  
-  /**
-   * @param isTraining A bool value to indicate the operation is for training (default)
-   * or inference.
-   */
-  public static Options isTraining(Boolean isTraining) {
-    return new Options().isTraining(isTraining);
-  }
-  
-  /**
-   * A 4D Tensor for output data.
-   */
-  public Output<T> y() {
-    return y;
-  }
-  
-  /**
-   * A 1D Tensor for the computed batch mean, to be used by TensorFlow
-   * to compute the running mean.
-   */
-  public Output<U> batchMean() {
-    return batchMean;
-  }
-  
-  /**
-   * A 1D Tensor for the computed batch variance, to be used by
-   * TensorFlow to compute the running variance.
-   */
-  public Output<U> batchVariance() {
-    return batchVariance;
-  }
-  
-  /**
-   * A 1D Tensor for the computed batch mean, to be reused
-   * in the gradient computation.
-   */
-  public Output<U> reserveSpace1() {
-    return reserveSpace1;
-  }
-  
-  /**
-   * A 1D Tensor for the computed batch variance (inverted variance
-   * in the cuDNN case), to be reused in the gradient computation.
-   */
-  public Output<U> reserveSpace2() {
-    return reserveSpace2;
-  }
-  
-  /**
-   * A 1D Tensor for some intermediate results, to be reused in the gradient
-   * computation for better efficiency.
-   */
-  public Output<U> reserveSpace3() {
-    return reserveSpace3;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "FusedBatchNormV3";
-  
-  private Output<T> y;
-  private Output<U> batchMean;
-  private Output<U> batchVariance;
-  private Output<U> reserveSpace1;
-  private Output<U> reserveSpace2;
-  private Output<U> reserveSpace3;
-  
-  private FusedBatchNorm(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    y = operation.output(outputIdx++);
-    batchMean = operation.output(outputIdx++);
-    batchVariance = operation.output(outputIdx++);
-    reserveSpace1 = operation.output(outputIdx++);
-    reserveSpace2 = operation.output(outputIdx++);
-    reserveSpace3 = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/FusedBatchNormGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/FusedBatchNormGrad.java
deleted file mode 100644
index abeb9214bab..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/FusedBatchNormGrad.java
+++ /dev/null
@@ -1,207 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.nn;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Gradient for batch normalization.
- * <p>
- * Note that the size of 4D Tensors are defined by either "NHWC" or "NCHW".
- * The size of 1D Tensors matches the dimension C of the 4D Tensors.
- * 
- * @param <T> data type for {@code xBackprop()} output
- * @param <U> data type for {@code scaleBackprop()} output
- */
-@Operator(group = "nn")
-public final class FusedBatchNormGrad<T extends TNumber, U extends TNumber> extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.nn.FusedBatchNormGrad}
-   */
-  public static class Options {
-    
-    /**
-     * @param epsilon A small float number added to the variance of x.
-     */
-    public Options epsilon(Float epsilon) {
-      this.epsilon = epsilon;
-      return this;
-    }
-    
-    /**
-     * @param dataFormat The data format for y_backprop, x, x_backprop.
-     * Either "NHWC" (default) or "NCHW".
-     */
-    public Options dataFormat(String dataFormat) {
-      this.dataFormat = dataFormat;
-      return this;
-    }
-    
-    /**
-     * @param isTraining A bool value to indicate the operation is for training (default)
-     * or inference.
-     */
-    public Options isTraining(Boolean isTraining) {
-      this.isTraining = isTraining;
-      return this;
-    }
-    
-    private Float epsilon;
-    private String dataFormat;
-    private Boolean isTraining;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new FusedBatchNormGrad operation.
-   * 
-   * @param scope current scope
-   * @param yBackprop A 4D Tensor for the gradient with respect to y.
-   * @param x A 4D Tensor for input data.
-   * @param scale A 1D Tensor for scaling factor, to scale the normalized x.
-   * @param reserveSpace1 When is_training is True, a 1D Tensor for the computed batch
-   * mean to be reused in gradient computation. When is_training is
-   * False, a 1D Tensor for the population mean to be reused in both
-   * 1st and 2nd order gradient computation.
-   * @param reserveSpace2 When is_training is True, a 1D Tensor for the computed batch
-   * variance (inverted variance in the cuDNN case) to be reused in
-   * gradient computation. When is_training is False, a 1D Tensor
-   * for the population variance to be reused in both 1st and 2nd
-   * order gradient computation.
-   * @param reserveSpace3 When is_training is True, a 1D Tensor for some intermediate results to be reused
-   * in gradient computation. When is_training is False, a dummy empty Tensor will be
-   * created.
-   * @param options carries optional attributes values
-   * @return a new instance of FusedBatchNormGrad
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber, U extends TNumber> FusedBatchNormGrad<T, U> create(Scope scope, Operand<T> yBackprop, Operand<T> x, Operand<TFloat32> scale, Operand<U> reserveSpace1, Operand<U> reserveSpace2, Operand<U> reserveSpace3, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("FusedBatchNormGradV3", scope.makeOpName("FusedBatchNormGrad"));
-    opBuilder.addInput(yBackprop.asOutput(scope));
-    opBuilder.addInput(x.asOutput(scope));
-    opBuilder.addInput(scale.asOutput(scope));
-    opBuilder.addInput(reserveSpace1.asOutput(scope));
-    opBuilder.addInput(reserveSpace2.asOutput(scope));
-    opBuilder.addInput(reserveSpace3.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.epsilon != null) {
-          opBuilder.setAttr("epsilon", opts.epsilon);
-        }
-        if (opts.dataFormat != null) {
-          opBuilder.setAttr("data_format", opts.dataFormat);
-        }
-        if (opts.isTraining != null) {
-          opBuilder.setAttr("is_training", opts.isTraining);
-        }
-      }
-    }
-    return new FusedBatchNormGrad<T, U>(opBuilder.build());
-  }
-  
-  /**
-   * @param epsilon A small float number added to the variance of x.
-   */
-  public static Options epsilon(Float epsilon) {
-    return new Options().epsilon(epsilon);
-  }
-  
-  /**
-   * @param dataFormat The data format for y_backprop, x, x_backprop.
-   * Either "NHWC" (default) or "NCHW".
-   */
-  public static Options dataFormat(String dataFormat) {
-    return new Options().dataFormat(dataFormat);
-  }
-  
-  /**
-   * @param isTraining A bool value to indicate the operation is for training (default)
-   * or inference.
-   */
-  public static Options isTraining(Boolean isTraining) {
-    return new Options().isTraining(isTraining);
-  }
-  
-  /**
-   * A 4D Tensor for the gradient with respect to x.
-   */
-  public Output<T> xBackprop() {
-    return xBackprop;
-  }
-  
-  /**
-   * A 1D Tensor for the gradient with respect to scale.
-   */
-  public Output<U> scaleBackprop() {
-    return scaleBackprop;
-  }
-  
-  /**
-   * A 1D Tensor for the gradient with respect to offset.
-   */
-  public Output<U> offsetBackprop() {
-    return offsetBackprop;
-  }
-  
-  /**
-   * Unused placeholder to match the mean input in FusedBatchNorm.
-   */
-  public Output<U> reserveSpace4() {
-    return reserveSpace4;
-  }
-  
-  /**
-   * Unused placeholder to match the variance input
-   * in FusedBatchNorm.
-   */
-  public Output<U> reserveSpace5() {
-    return reserveSpace5;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "FusedBatchNormGradV3";
-  
-  private Output<T> xBackprop;
-  private Output<U> scaleBackprop;
-  private Output<U> offsetBackprop;
-  private Output<U> reserveSpace4;
-  private Output<U> reserveSpace5;
-  
-  private FusedBatchNormGrad(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    xBackprop = operation.output(outputIdx++);
-    scaleBackprop = operation.output(outputIdx++);
-    offsetBackprop = operation.output(outputIdx++);
-    reserveSpace4 = operation.output(outputIdx++);
-    reserveSpace5 = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/FusedPadConv2d.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/FusedPadConv2d.java
deleted file mode 100644
index 2bd86f38e37..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/FusedPadConv2d.java
+++ /dev/null
@@ -1,105 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.nn;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Performs a padding as a preprocess during a convolution.
- * <p>
- * Similar to FusedResizeAndPadConv2d, this op allows for an optimized
- * implementation where the spatial padding transformation stage is fused with the
- * im2col lookup, but in this case without the bilinear filtering required for
- * resizing. Fusing the padding prevents the need to write out the intermediate
- * results as whole tensors, reducing memory pressure, and we can get some latency
- * gains by merging the transformation calculations.
- * The data_format attribute for Conv2D isn't supported by this op, and 'NHWC'
- * order is used instead.
- * Internally this op uses a single per-graph scratch buffer, which means that it
- * will block if multiple versions are being run in parallel. This is because this
- * operator is primarily an optimization to minimize memory usage.
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator(group = "nn")
-public final class FusedPadConv2d<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new FusedPadConv2d operation.
-   * 
-   * @param scope current scope
-   * @param input 4-D with shape `[batch, in_height, in_width, in_channels]`.
-   * @param paddings A two-column matrix specifying the padding sizes. The number of
-   * rows must be the same as the rank of `input`.
-   * @param filter 4-D with shape
-   * `[filter_height, filter_width, in_channels, out_channels]`.
-   * @param mode 
-   * @param strides 1-D of length 4.  The stride of the sliding window for each dimension
-   * of `input`. Must be in the same order as the dimension specified with format.
-   * @param padding The type of padding algorithm to use.
-   * @return a new instance of FusedPadConv2d
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> FusedPadConv2d<T> create(Scope scope, Operand<T> input, Operand<TInt32> paddings, Operand<T> filter, String mode, List<Long> strides, String padding) {
-    OperationBuilder opBuilder = scope.env().opBuilder("FusedPadConv2D", scope.makeOpName("FusedPadConv2d"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder.addInput(paddings.asOutput(scope));
-    opBuilder.addInput(filter.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("mode", mode);
-    long[] stridesArray = new long[strides.size()];
-    for (int i = 0; i < stridesArray.length; ++i) {
-      stridesArray[i] = strides.get(i);
-    }
-    opBuilder.setAttr("strides", stridesArray);
-    opBuilder.setAttr("padding", padding);
-    return new FusedPadConv2d<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "FusedPadConv2D";
-  
-  private Output<T> output;
-  
-  private FusedPadConv2d(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/FusedResizeAndPadConv2d.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/FusedResizeAndPadConv2d.java
deleted file mode 100644
index e4f8fa54396..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/FusedResizeAndPadConv2d.java
+++ /dev/null
@@ -1,143 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.nn;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Performs a resize and padding as a preprocess during a convolution.
- * <p>
- * It's often possible to do spatial transformations more efficiently as part of
- * the packing stage of a convolution, so this op allows for an optimized
- * implementation where these stages are fused together. This prevents the need to
- * write out the intermediate results as whole tensors, reducing memory pressure,
- * and we can get some latency gains by merging the transformation calculations.
- * The data_format attribute for Conv2D isn't supported by this op, and defaults to
- * 'NHWC' order.
- * Internally this op uses a single per-graph scratch buffer, which means that it
- * will block if multiple versions are being run in parallel. This is because this
- * operator is primarily an optimization to minimize memory usage.
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator(group = "nn")
-public final class FusedResizeAndPadConv2d<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.nn.FusedResizeAndPadConv2d}
-   */
-  public static class Options {
-    
-    /**
-     * @param resizeAlignCorners If true, the centers of the 4 corner pixels of the input and output tensors are
-     * aligned, preserving the values at the corner pixels. Defaults to false.
-     */
-    public Options resizeAlignCorners(Boolean resizeAlignCorners) {
-      this.resizeAlignCorners = resizeAlignCorners;
-      return this;
-    }
-    
-    private Boolean resizeAlignCorners;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new FusedResizeAndPadConv2d operation.
-   * 
-   * @param scope current scope
-   * @param input 4-D with shape `[batch, in_height, in_width, in_channels]`.
-   * @param size A 1-D int32 Tensor of 2 elements: `new_height, new_width`.  The
-   * new size for the images.
-   * @param paddings A two-column matrix specifying the padding sizes. The number of
-   * rows must be the same as the rank of `input`.
-   * @param filter 4-D with shape
-   * `[filter_height, filter_width, in_channels, out_channels]`.
-   * @param mode 
-   * @param strides 1-D of length 4.  The stride of the sliding window for each dimension
-   * of `input`. Must be in the same order as the dimension specified with format.
-   * @param padding The type of padding algorithm to use.
-   * @param options carries optional attributes values
-   * @return a new instance of FusedResizeAndPadConv2d
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> FusedResizeAndPadConv2d<T> create(Scope scope, Operand<T> input, Operand<TInt32> size, Operand<TInt32> paddings, Operand<T> filter, String mode, List<Long> strides, String padding, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("FusedResizeAndPadConv2D", scope.makeOpName("FusedResizeAndPadConv2d"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder.addInput(size.asOutput(scope));
-    opBuilder.addInput(paddings.asOutput(scope));
-    opBuilder.addInput(filter.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("mode", mode);
-    long[] stridesArray = new long[strides.size()];
-    for (int i = 0; i < stridesArray.length; ++i) {
-      stridesArray[i] = strides.get(i);
-    }
-    opBuilder.setAttr("strides", stridesArray);
-    opBuilder.setAttr("padding", padding);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.resizeAlignCorners != null) {
-          opBuilder.setAttr("resize_align_corners", opts.resizeAlignCorners);
-        }
-      }
-    }
-    return new FusedResizeAndPadConv2d<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param resizeAlignCorners If true, the centers of the 4 corner pixels of the input and output tensors are
-   * aligned, preserving the values at the corner pixels. Defaults to false.
-   */
-  public static Options resizeAlignCorners(Boolean resizeAlignCorners) {
-    return new Options().resizeAlignCorners(resizeAlignCorners);
-  }
-  
-  /**
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "FusedResizeAndPadConv2D";
-  
-  private Output<T> output;
-  
-  private FusedResizeAndPadConv2d(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/GRUBlockCell.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/GRUBlockCell.java
deleted file mode 100644
index c51e1db1bdd..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/GRUBlockCell.java
+++ /dev/null
@@ -1,147 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.nn;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Computes the GRU cell forward propagation for 1 time step.
- * <p>
- * Args
- *     x: Input to the GRU cell.
- *     h_prev: State input from the previous GRU cell.
- *     w_ru: Weight matrix for the reset and update gate.
- *     w_c: Weight matrix for the cell connection gate.
- *     b_ru: Bias vector for the reset and update gate.
- *     b_c: Bias vector for the cell connection gate.
- * <p>
- * Returns
- *     r: Output of the reset gate.
- *     u: Output of the update gate.
- *     c: Output of the cell connection gate.
- *     h: Current state of the GRU cell.
- * <p>
- * Note on notation of the variables:
- * <p>
- * Concatenation of a and b is represented by a_b
- * Element-wise dot product of a and b is represented by ab
- * Element-wise dot product is represented by \circ
- * Matrix multiplication is represented by *
- * <p>
- * Biases are initialized with :
- * `b_ru` - constant_initializer(1.0)
- * `b_c` - constant_initializer(0.0)
- * <p>
- * This kernel op implements the following mathematical equations:
- * <pre>{@code
- * x_h_prev = [x, h_prev]
- * 
- * [r_bar u_bar] = x_h_prev * w_ru + b_ru
- * 
- * r = sigmoid(r_bar)
- * u = sigmoid(u_bar)
- * 
- * h_prevr = h_prev \circ r
- * 
- * x_h_prevr = [x h_prevr]
- * 
- * c_bar = x_h_prevr * w_c + b_c
- * c = tanh(c_bar)
- * 
- * h = (1-u) \circ c + u \circ h_prev
- * }</pre>
- * 
- * 
- * @param <T> data type for {@code r()} output
- */
-public final class GRUBlockCell<T extends TNumber> extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new GRUBlockCell operation.
-   * 
-   * @param scope current scope
-   * @param x 
-   * @param hPrev 
-   * @param wRu 
-   * @param wC 
-   * @param bRu 
-   * @param bC 
-   * @return a new instance of GRUBlockCell
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> GRUBlockCell<T> create(Scope scope, Operand<T> x, Operand<T> hPrev, Operand<T> wRu, Operand<T> wC, Operand<T> bRu, Operand<T> bC) {
-    OperationBuilder opBuilder = scope.env().opBuilder("GRUBlockCell", scope.makeOpName("GRUBlockCell"));
-    opBuilder.addInput(x.asOutput(scope));
-    opBuilder.addInput(hPrev.asOutput(scope));
-    opBuilder.addInput(wRu.asOutput(scope));
-    opBuilder.addInput(wC.asOutput(scope));
-    opBuilder.addInput(bRu.asOutput(scope));
-    opBuilder.addInput(bC.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new GRUBlockCell<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> r() {
-    return r;
-  }
-  
-  /**
-   */
-  public Output<T> u() {
-    return u;
-  }
-  
-  /**
-   */
-  public Output<T> c() {
-    return c;
-  }
-  
-  /**
-   */
-  public Output<T> h() {
-    return h;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "GRUBlockCell";
-  
-  private Output<T> r;
-  private Output<T> u;
-  private Output<T> c;
-  private Output<T> h;
-  
-  private GRUBlockCell(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    r = operation.output(outputIdx++);
-    u = operation.output(outputIdx++);
-    c = operation.output(outputIdx++);
-    h = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/GRUBlockCellGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/GRUBlockCellGrad.java
deleted file mode 100644
index 1e93416bcc6..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/GRUBlockCellGrad.java
+++ /dev/null
@@ -1,191 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.nn;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Computes the GRU cell back-propagation for 1 time step.
- * <p>
- * Args
- *     x: Input to the GRU cell.
- *     h_prev: State input from the previous GRU cell.
- *     w_ru: Weight matrix for the reset and update gate.
- *     w_c: Weight matrix for the cell connection gate.
- *     b_ru: Bias vector for the reset and update gate.
- *     b_c: Bias vector for the cell connection gate.
- *     r: Output of the reset gate.
- *     u: Output of the update gate.
- *     c: Output of the cell connection gate.
- *     d_h: Gradients of the h_new wrt to objective function.
- * <p>
- * Returns
- *     d_x: Gradients of the x wrt to objective function.
- *     d_h_prev: Gradients of the h wrt to objective function.
- *     d_c_bar Gradients of the c_bar wrt to objective function.
- *     d_r_bar_u_bar Gradients of the r_bar & u_bar wrt to objective function.
- * <p>
- * This kernel op implements the following mathematical equations:
- * <p>
- * Note on notation of the variables:
- * <p>
- * Concatenation of a and b is represented by a_b
- * Element-wise dot product of a and b is represented by ab
- * Element-wise dot product is represented by \circ
- * Matrix multiplication is represented by *
- * <p>
- * Additional notes for clarity:
- * <p>
- * `w_ru` can be segmented into 4 different matrices.
- * <pre>{@code
- * w_ru = [w_r_x w_u_x
- *         w_r_h_prev w_u_h_prev]
- * }</pre>
- * Similarly, `w_c` can be segmented into 2 different matrices.
- * <pre>{@code
- * w_c = [w_c_x w_c_h_prevr]
- * }</pre>
- * Same goes for biases.
- * <pre>{@code
- * b_ru = [b_ru_x b_ru_h]
- * b_c = [b_c_x b_c_h]
- * }</pre>
- * Another note on notation:
- * <pre>{@code
- * d_x = d_x_component_1 + d_x_component_2
- * 
- * where d_x_component_1 = d_r_bar * w_r_x^T + d_u_bar * w_r_x^T
- * and d_x_component_2 = d_c_bar * w_c_x^T
- * 
- * d_h_prev = d_h_prev_component_1 + d_h_prevr \circ r + d_h \circ u
- * where d_h_prev_componenet_1 = d_r_bar * w_r_h_prev^T + d_u_bar * w_r_h_prev^T
- * }</pre>
- * Mathematics behind the Gradients below:
- * <pre>{@code
- * d_c_bar = d_h \circ (1-u) \circ (1-c \circ c)
- * d_u_bar = d_h \circ (h-c) \circ u \circ (1-u)
- * 
- * d_r_bar_u_bar = [d_r_bar d_u_bar]
- * 
- * [d_x_component_1 d_h_prev_component_1] = d_r_bar_u_bar * w_ru^T
- * 
- * [d_x_component_2 d_h_prevr] = d_c_bar * w_c^T
- * 
- * d_x = d_x_component_1 + d_x_component_2
- * 
- * d_h_prev = d_h_prev_component_1 + d_h_prevr \circ r + u
- * }</pre>
- * Below calculation is performed in the python wrapper for the Gradients
- * (not in the gradient kernel.)
- * <pre>{@code
- * d_w_ru = x_h_prevr^T * d_c_bar
- * 
- * d_w_c = x_h_prev^T * d_r_bar_u_bar
- * 
- * d_b_ru = sum of d_r_bar_u_bar along axis = 0
- * 
- * d_b_c = sum of d_c_bar along axis = 0
- * }</pre>
- * 
- * 
- * @param <T> data type for {@code dX()} output
- */
-public final class GRUBlockCellGrad<T extends TNumber> extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new GRUBlockCellGrad operation.
-   * 
-   * @param scope current scope
-   * @param x 
-   * @param hPrev 
-   * @param wRu 
-   * @param wC 
-   * @param bRu 
-   * @param bC 
-   * @param r 
-   * @param u 
-   * @param c 
-   * @param dH 
-   * @return a new instance of GRUBlockCellGrad
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> GRUBlockCellGrad<T> create(Scope scope, Operand<T> x, Operand<T> hPrev, Operand<T> wRu, Operand<T> wC, Operand<T> bRu, Operand<T> bC, Operand<T> r, Operand<T> u, Operand<T> c, Operand<T> dH) {
-    OperationBuilder opBuilder = scope.env().opBuilder("GRUBlockCellGrad", scope.makeOpName("GRUBlockCellGrad"));
-    opBuilder.addInput(x.asOutput(scope));
-    opBuilder.addInput(hPrev.asOutput(scope));
-    opBuilder.addInput(wRu.asOutput(scope));
-    opBuilder.addInput(wC.asOutput(scope));
-    opBuilder.addInput(bRu.asOutput(scope));
-    opBuilder.addInput(bC.asOutput(scope));
-    opBuilder.addInput(r.asOutput(scope));
-    opBuilder.addInput(u.asOutput(scope));
-    opBuilder.addInput(c.asOutput(scope));
-    opBuilder.addInput(dH.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new GRUBlockCellGrad<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> dX() {
-    return dX;
-  }
-  
-  /**
-   */
-  public Output<T> dHPrev() {
-    return dHPrev;
-  }
-  
-  /**
-   */
-  public Output<T> dCBar() {
-    return dCBar;
-  }
-  
-  /**
-   */
-  public Output<T> dRBarUBar() {
-    return dRBarUBar;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "GRUBlockCellGrad";
-  
-  private Output<T> dX;
-  private Output<T> dHPrev;
-  private Output<T> dCBar;
-  private Output<T> dRBarUBar;
-  
-  private GRUBlockCellGrad(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    dX = operation.output(outputIdx++);
-    dHPrev = operation.output(outputIdx++);
-    dCBar = operation.output(outputIdx++);
-    dRBarUBar = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/InTopK.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/InTopK.java
deleted file mode 100644
index 28a53b5b13f..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/InTopK.java
+++ /dev/null
@@ -1,94 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.nn;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TBool;
-import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Says whether the targets are in the top `K` predictions.
- * <p>
- * This outputs a `batch_size` bool array, an entry `out[i]` is `true` if the
- * prediction for the target class is among the top `k` predictions among
- * all predictions for example `i`. Note that the behavior of `InTopK` differs
- * from the `TopK` op in its handling of ties; if multiple classes have the
- * same prediction value and straddle the top-`k` boundary, all of those
- * classes are considered to be in the top `k`.
- * <p>
- * More formally, let
- * <p>
- *   \\(predictions_i\\) be the predictions for all classes for example `i`,
- *   \\(targets_i\\) be the target class for example `i`,
- *   \\(out_i\\) be the output for example `i`,
- * <p>
- * $$out_i = predictions_{i, targets_i} \in TopKIncludingTies(predictions_i)$$
- */
-@Operator(group = "nn")
-public final class InTopK extends RawOp implements Operand<TBool> {
-  
-  /**
-   * Factory method to create a class wrapping a new InTopK operation.
-   * 
-   * @param scope current scope
-   * @param predictions A `batch_size` x `classes` tensor.
-   * @param targets A `batch_size` vector of class ids.
-   * @param k Number of top elements to look at for computing precision.
-   * @return a new instance of InTopK
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> InTopK create(Scope scope, Operand<TFloat32> predictions, Operand<T> targets, Operand<T> k) {
-    OperationBuilder opBuilder = scope.env().opBuilder("InTopKV2", scope.makeOpName("InTopK"));
-    opBuilder.addInput(predictions.asOutput(scope));
-    opBuilder.addInput(targets.asOutput(scope));
-    opBuilder.addInput(k.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new InTopK(opBuilder.build());
-  }
-  
-  /**
-   * Computed precision at `k` as a `bool Tensor`.
-   */
-  public Output<TBool> precision() {
-    return precision;
-  }
-  
-  @Override
-  public Output<TBool> asOutput(Scope scope) {
-    return precision;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "InTopKV2";
-  
-  private Output<TBool> precision;
-  
-  private InTopK(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    precision = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/InvGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/InvGrad.java
deleted file mode 100644
index c1143025c38..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/InvGrad.java
+++ /dev/null
@@ -1,78 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.nn;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Computes the gradient for the inverse of `x` wrt its input.
- * <p>
- * Specifically, `grad = -dy <i> y</i>y`, where `y = 1/x`, and `dy`
- * is the corresponding input gradient.
- * 
- * @param <T> data type for {@code z()} output
- */
-public final class InvGrad<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new InvGrad operation.
-   * 
-   * @param scope current scope
-   * @param y 
-   * @param dy 
-   * @return a new instance of InvGrad
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> InvGrad<T> create(Scope scope, Operand<T> y, Operand<T> dy) {
-    OperationBuilder opBuilder = scope.env().opBuilder("InvGrad", scope.makeOpName("InvGrad"));
-    opBuilder.addInput(y.asOutput(scope));
-    opBuilder.addInput(dy.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new InvGrad<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> z() {
-    return z;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return z;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "InvGrad";
-  
-  private Output<T> z;
-  
-  private InvGrad(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    z = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/L2Loss.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/L2Loss.java
deleted file mode 100644
index 2c72e84e897..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/L2Loss.java
+++ /dev/null
@@ -1,79 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.nn;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * L2 Loss.
- * <p>
- * Computes half the L2 norm of a tensor without the `sqrt`:
- * <p>
- *     output = sum(t ** 2) / 2
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator(group = "nn")
-public final class L2Loss<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new L2Loss operation.
-   * 
-   * @param scope current scope
-   * @param t Typically 2-D, but may have any dimensions.
-   * @return a new instance of L2Loss
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> L2Loss<T> create(Scope scope, Operand<T> t) {
-    OperationBuilder opBuilder = scope.env().opBuilder("L2Loss", scope.makeOpName("L2Loss"));
-    opBuilder.addInput(t.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new L2Loss<T>(opBuilder.build());
-  }
-  
-  /**
-   * 0-D.
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "L2Loss";
-  
-  private Output<T> output;
-  
-  private L2Loss(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/LSTMBlockCell.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/LSTMBlockCell.java
deleted file mode 100644
index d733379eac1..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/LSTMBlockCell.java
+++ /dev/null
@@ -1,234 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.nn;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Computes the LSTM cell forward propagation for 1 time step.
- * <p>
- * This implementation uses 1 weight matrix and 1 bias vector, and there's an
- * optional peephole connection.
- * <p>
- * This kernel op implements the following mathematical equations:
- * <pre>{@code
- * xh = [x, h_prev]
- * [i, f, ci, o] = xh * w + b
- * f = f + forget_bias
- * 
- * if not use_peephole:
- *   wci = wcf = wco = 0
- * 
- * i = sigmoid(cs_prev * wci + i)
- * f = sigmoid(cs_prev * wcf + f)
- * ci = tanh(ci)
- * 
- * cs = ci .* i + cs_prev .* f
- * cs = clip(cs, cell_clip)
- * 
- * o = sigmoid(cs * wco + o)
- * co = tanh(cs)
- * h = co .* o
- * }</pre>
- * 
- * 
- * @param <T> data type for {@code i()} output
- */
-public final class LSTMBlockCell<T extends TNumber> extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.nn.LSTMBlockCell}
-   */
-  public static class Options {
-    
-    /**
-     * @param forgetBias The forget gate bias.
-     */
-    public Options forgetBias(Float forgetBias) {
-      this.forgetBias = forgetBias;
-      return this;
-    }
-    
-    /**
-     * @param cellClip Value to clip the 'cs' value to.
-     */
-    public Options cellClip(Float cellClip) {
-      this.cellClip = cellClip;
-      return this;
-    }
-    
-    /**
-     * @param usePeephole Whether to use peephole weights.
-     */
-    public Options usePeephole(Boolean usePeephole) {
-      this.usePeephole = usePeephole;
-      return this;
-    }
-    
-    private Float forgetBias;
-    private Float cellClip;
-    private Boolean usePeephole;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new LSTMBlockCell operation.
-   * 
-   * @param scope current scope
-   * @param x The input to the LSTM cell, shape (batch_size, num_inputs).
-   * @param csPrev Value of the cell state at previous time step.
-   * @param hPrev Output of the previous cell at previous time step.
-   * @param w The weight matrix.
-   * @param wci The weight matrix for input gate peephole connection.
-   * @param wcf The weight matrix for forget gate peephole connection.
-   * @param wco The weight matrix for output gate peephole connection.
-   * @param b The bias vector.
-   * @param options carries optional attributes values
-   * @return a new instance of LSTMBlockCell
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> LSTMBlockCell<T> create(Scope scope, Operand<T> x, Operand<T> csPrev, Operand<T> hPrev, Operand<T> w, Operand<T> wci, Operand<T> wcf, Operand<T> wco, Operand<T> b, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("LSTMBlockCell", scope.makeOpName("LSTMBlockCell"));
-    opBuilder.addInput(x.asOutput(scope));
-    opBuilder.addInput(csPrev.asOutput(scope));
-    opBuilder.addInput(hPrev.asOutput(scope));
-    opBuilder.addInput(w.asOutput(scope));
-    opBuilder.addInput(wci.asOutput(scope));
-    opBuilder.addInput(wcf.asOutput(scope));
-    opBuilder.addInput(wco.asOutput(scope));
-    opBuilder.addInput(b.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.forgetBias != null) {
-          opBuilder.setAttr("forget_bias", opts.forgetBias);
-        }
-        if (opts.cellClip != null) {
-          opBuilder.setAttr("cell_clip", opts.cellClip);
-        }
-        if (opts.usePeephole != null) {
-          opBuilder.setAttr("use_peephole", opts.usePeephole);
-        }
-      }
-    }
-    return new LSTMBlockCell<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param forgetBias The forget gate bias.
-   */
-  public static Options forgetBias(Float forgetBias) {
-    return new Options().forgetBias(forgetBias);
-  }
-  
-  /**
-   * @param cellClip Value to clip the 'cs' value to.
-   */
-  public static Options cellClip(Float cellClip) {
-    return new Options().cellClip(cellClip);
-  }
-  
-  /**
-   * @param usePeephole Whether to use peephole weights.
-   */
-  public static Options usePeephole(Boolean usePeephole) {
-    return new Options().usePeephole(usePeephole);
-  }
-  
-  /**
-   * The input gate.
-   */
-  public Output<T> i() {
-    return i;
-  }
-  
-  /**
-   * The cell state before the tanh.
-   */
-  public Output<T> cs() {
-    return cs;
-  }
-  
-  /**
-   * The forget gate.
-   */
-  public Output<T> f() {
-    return f;
-  }
-  
-  /**
-   * The output gate.
-   */
-  public Output<T> o() {
-    return o;
-  }
-  
-  /**
-   * The cell input.
-   */
-  public Output<T> ci() {
-    return ci;
-  }
-  
-  /**
-   * The cell after the tanh.
-   */
-  public Output<T> co() {
-    return co;
-  }
-  
-  /**
-   * The output h vector.
-   */
-  public Output<T> h() {
-    return h;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "LSTMBlockCell";
-  
-  private Output<T> i;
-  private Output<T> cs;
-  private Output<T> f;
-  private Output<T> o;
-  private Output<T> ci;
-  private Output<T> co;
-  private Output<T> h;
-  
-  private LSTMBlockCell(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    i = operation.output(outputIdx++);
-    cs = operation.output(outputIdx++);
-    f = operation.output(outputIdx++);
-    o = operation.output(outputIdx++);
-    ci = operation.output(outputIdx++);
-    co = operation.output(outputIdx++);
-    h = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/LSTMBlockCellGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/LSTMBlockCellGrad.java
deleted file mode 100644
index e515f99e9e9..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/LSTMBlockCellGrad.java
+++ /dev/null
@@ -1,139 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.nn;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Computes the LSTM cell backward propagation for 1 timestep.
- * <p>
- * This implementation is to be used in conjunction of LSTMBlockCell.
- * 
- * @param <T> data type for {@code csPrevGrad()} output
- */
-public final class LSTMBlockCellGrad<T extends TNumber> extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new LSTMBlockCellGrad operation.
-   * 
-   * @param scope current scope
-   * @param x The input to the LSTM cell, shape (batch_size, num_inputs).
-   * @param csPrev The previous cell state.
-   * @param hPrev The previous h state.
-   * @param w The weight matrix.
-   * @param wci The weight matrix for input gate peephole connection.
-   * @param wcf The weight matrix for forget gate peephole connection.
-   * @param wco The weight matrix for output gate peephole connection.
-   * @param b The bias vector.
-   * @param i The input gate.
-   * @param cs The cell state before the tanh.
-   * @param f The forget gate.
-   * @param o The output gate.
-   * @param ci The cell input.
-   * @param co The cell after the tanh.
-   * @param csGrad The current gradient of cs.
-   * @param hGrad The gradient of h vector.
-   * @param usePeephole Whether the cell uses peephole connections.
-   * @return a new instance of LSTMBlockCellGrad
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> LSTMBlockCellGrad<T> create(Scope scope, Operand<T> x, Operand<T> csPrev, Operand<T> hPrev, Operand<T> w, Operand<T> wci, Operand<T> wcf, Operand<T> wco, Operand<T> b, Operand<T> i, Operand<T> cs, Operand<T> f, Operand<T> o, Operand<T> ci, Operand<T> co, Operand<T> csGrad, Operand<T> hGrad, Boolean usePeephole) {
-    OperationBuilder opBuilder = scope.env().opBuilder("LSTMBlockCellGrad", scope.makeOpName("LSTMBlockCellGrad"));
-    opBuilder.addInput(x.asOutput(scope));
-    opBuilder.addInput(csPrev.asOutput(scope));
-    opBuilder.addInput(hPrev.asOutput(scope));
-    opBuilder.addInput(w.asOutput(scope));
-    opBuilder.addInput(wci.asOutput(scope));
-    opBuilder.addInput(wcf.asOutput(scope));
-    opBuilder.addInput(wco.asOutput(scope));
-    opBuilder.addInput(b.asOutput(scope));
-    opBuilder.addInput(i.asOutput(scope));
-    opBuilder.addInput(cs.asOutput(scope));
-    opBuilder.addInput(f.asOutput(scope));
-    opBuilder.addInput(o.asOutput(scope));
-    opBuilder.addInput(ci.asOutput(scope));
-    opBuilder.addInput(co.asOutput(scope));
-    opBuilder.addInput(csGrad.asOutput(scope));
-    opBuilder.addInput(hGrad.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("use_peephole", usePeephole);
-    return new LSTMBlockCellGrad<T>(opBuilder.build());
-  }
-  
-  /**
-   * The gradient of cs to be back-propped.
-   */
-  public Output<T> csPrevGrad() {
-    return csPrevGrad;
-  }
-  
-  /**
-   * The derivative wrt to [i, cs, f, o].
-   */
-  public Output<T> dicfo() {
-    return dicfo;
-  }
-  
-  /**
-   * The gradient for wci to be back-propped.
-   */
-  public Output<T> wciGrad() {
-    return wciGrad;
-  }
-  
-  /**
-   * The gradient for wcf to be back-propped.
-   */
-  public Output<T> wcfGrad() {
-    return wcfGrad;
-  }
-  
-  /**
-   * The gradient for wco to be back-propped.
-   */
-  public Output<T> wcoGrad() {
-    return wcoGrad;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "LSTMBlockCellGrad";
-  
-  private Output<T> csPrevGrad;
-  private Output<T> dicfo;
-  private Output<T> wciGrad;
-  private Output<T> wcfGrad;
-  private Output<T> wcoGrad;
-  
-  private LSTMBlockCellGrad(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    csPrevGrad = operation.output(outputIdx++);
-    dicfo = operation.output(outputIdx++);
-    wciGrad = operation.output(outputIdx++);
-    wcfGrad = operation.output(outputIdx++);
-    wcoGrad = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/LearnedUnigramCandidateSampler.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/LearnedUnigramCandidateSampler.java
deleted file mode 100644
index 668c1cd3b1e..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/LearnedUnigramCandidateSampler.java
+++ /dev/null
@@ -1,171 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.nn;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.TInt64;
-
-/**
- * Generates labels for candidate sampling with a learned unigram distribution.
- * <p>
- * See explanations of candidate sampling and the data formats at
- * go/candidate-sampling.
- * <p>
- * For each batch, this op picks a single set of sampled candidate labels.
- * <p>
- * The advantages of sampling candidates per-batch are simplicity and the
- * possibility of efficient dense matrix multiplication. The disadvantage is that
- * the sampled candidates must be chosen independently of the context and of the
- * true labels.
- */
-@Operator(group = "nn")
-public final class LearnedUnigramCandidateSampler extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.nn.LearnedUnigramCandidateSampler}
-   */
-  public static class Options {
-    
-    /**
-     * @param seed If either seed or seed2 are set to be non-zero, the random number
-     * generator is seeded by the given seed.  Otherwise, it is seeded by a
-     * random seed.
-     */
-    public Options seed(Long seed) {
-      this.seed = seed;
-      return this;
-    }
-    
-    /**
-     * @param seed2 An second seed to avoid seed collision.
-     */
-    public Options seed2(Long seed2) {
-      this.seed2 = seed2;
-      return this;
-    }
-    
-    private Long seed;
-    private Long seed2;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new LearnedUnigramCandidateSampler operation.
-   * 
-   * @param scope current scope
-   * @param trueClasses A batch_size * num_true matrix, in which each row contains the
-   * IDs of the num_true target_classes in the corresponding original label.
-   * @param numTrue Number of true labels per context.
-   * @param numSampled Number of candidates to randomly sample.
-   * @param unique If unique is true, we sample with rejection, so that all sampled
-   * candidates in a batch are unique. This requires some approximation to
-   * estimate the post-rejection sampling probabilities.
-   * @param rangeMax The sampler will sample integers from the interval [0, range_max).
-   * @param options carries optional attributes values
-   * @return a new instance of LearnedUnigramCandidateSampler
-   */
-  @Endpoint(describeByClass = true)
-  public static LearnedUnigramCandidateSampler create(Scope scope, Operand<TInt64> trueClasses, Long numTrue, Long numSampled, Boolean unique, Long rangeMax, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("LearnedUnigramCandidateSampler", scope.makeOpName("LearnedUnigramCandidateSampler"));
-    opBuilder.addInput(trueClasses.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("num_true", numTrue);
-    opBuilder.setAttr("num_sampled", numSampled);
-    opBuilder.setAttr("unique", unique);
-    opBuilder.setAttr("range_max", rangeMax);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.seed != null) {
-          opBuilder.setAttr("seed", opts.seed);
-        }
-        if (opts.seed2 != null) {
-          opBuilder.setAttr("seed2", opts.seed2);
-        }
-      }
-    }
-    return new LearnedUnigramCandidateSampler(opBuilder.build());
-  }
-  
-  /**
-   * @param seed If either seed or seed2 are set to be non-zero, the random number
-   * generator is seeded by the given seed.  Otherwise, it is seeded by a
-   * random seed.
-   */
-  public static Options seed(Long seed) {
-    return new Options().seed(seed);
-  }
-  
-  /**
-   * @param seed2 An second seed to avoid seed collision.
-   */
-  public static Options seed2(Long seed2) {
-    return new Options().seed2(seed2);
-  }
-  
-  /**
-   * A vector of length num_sampled, in which each element is
-   * the ID of a sampled candidate.
-   */
-  public Output<TInt64> sampledCandidates() {
-    return sampledCandidates;
-  }
-  
-  /**
-   * A batch_size * num_true matrix, representing
-   * the number of times each candidate is expected to occur in a batch
-   * of sampled candidates. If unique=true, then this is a probability.
-   */
-  public Output<TFloat32> trueExpectedCount() {
-    return trueExpectedCount;
-  }
-  
-  /**
-   * A vector of length num_sampled, for each sampled
-   * candidate representing the number of times the candidate is expected
-   * to occur in a batch of sampled candidates.  If unique=true, then this is a
-   * probability.
-   */
-  public Output<TFloat32> sampledExpectedCount() {
-    return sampledExpectedCount;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "LearnedUnigramCandidateSampler";
-  
-  private Output<TInt64> sampledCandidates;
-  private Output<TFloat32> trueExpectedCount;
-  private Output<TFloat32> sampledExpectedCount;
-  
-  private LearnedUnigramCandidateSampler(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    sampledCandidates = operation.output(outputIdx++);
-    trueExpectedCount = operation.output(outputIdx++);
-    sampledExpectedCount = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/LocalResponseNormalization.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/LocalResponseNormalization.java
deleted file mode 100644
index 9347b69f656..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/LocalResponseNormalization.java
+++ /dev/null
@@ -1,177 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.nn;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Local Response Normalization.
- * <p>
- * The 4-D `input` tensor is treated as a 3-D array of 1-D vectors (along the last
- * dimension), and each vector is normalized independently.  Within a given vector,
- * each component is divided by the weighted, squared sum of inputs within
- * `depth_radius`.  In detail,
- * <p>
- *     sqr_sum[a, b, c, d] =
- *         sum(input[a, b, c, d - depth_radius : d + depth_radius + 1] ** 2)
- *     output = input / (bias + alpha * sqr_sum) ** beta
- * <p>
- * For details, see [Krizhevsky et al., ImageNet classification with deep
- * convolutional neural networks (NIPS 2012)](http://papers.nips.cc/paper/4824-imagenet-classification-with-deep-convolutional-neural-networks).
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator(group = "nn")
-public final class LocalResponseNormalization<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.nn.LocalResponseNormalization}
-   */
-  public static class Options {
-    
-    /**
-     * @param depthRadius 0-D.  Half-width of the 1-D normalization window.
-     */
-    public Options depthRadius(Long depthRadius) {
-      this.depthRadius = depthRadius;
-      return this;
-    }
-    
-    /**
-     * @param bias An offset (usually positive to avoid dividing by 0).
-     */
-    public Options bias(Float bias) {
-      this.bias = bias;
-      return this;
-    }
-    
-    /**
-     * @param alpha A scale factor, usually positive.
-     */
-    public Options alpha(Float alpha) {
-      this.alpha = alpha;
-      return this;
-    }
-    
-    /**
-     * @param beta An exponent.
-     */
-    public Options beta(Float beta) {
-      this.beta = beta;
-      return this;
-    }
-    
-    private Long depthRadius;
-    private Float bias;
-    private Float alpha;
-    private Float beta;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new LocalResponseNormalization operation.
-   * 
-   * @param scope current scope
-   * @param input 4-D.
-   * @param options carries optional attributes values
-   * @return a new instance of LocalResponseNormalization
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> LocalResponseNormalization<T> create(Scope scope, Operand<T> input, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("LRN", scope.makeOpName("LocalResponseNormalization"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.depthRadius != null) {
-          opBuilder.setAttr("depth_radius", opts.depthRadius);
-        }
-        if (opts.bias != null) {
-          opBuilder.setAttr("bias", opts.bias);
-        }
-        if (opts.alpha != null) {
-          opBuilder.setAttr("alpha", opts.alpha);
-        }
-        if (opts.beta != null) {
-          opBuilder.setAttr("beta", opts.beta);
-        }
-      }
-    }
-    return new LocalResponseNormalization<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param depthRadius 0-D.  Half-width of the 1-D normalization window.
-   */
-  public static Options depthRadius(Long depthRadius) {
-    return new Options().depthRadius(depthRadius);
-  }
-  
-  /**
-   * @param bias An offset (usually positive to avoid dividing by 0).
-   */
-  public static Options bias(Float bias) {
-    return new Options().bias(bias);
-  }
-  
-  /**
-   * @param alpha A scale factor, usually positive.
-   */
-  public static Options alpha(Float alpha) {
-    return new Options().alpha(alpha);
-  }
-  
-  /**
-   * @param beta An exponent.
-   */
-  public static Options beta(Float beta) {
-    return new Options().beta(beta);
-  }
-  
-  /**
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "LRN";
-  
-  private Output<T> output;
-  
-  private LocalResponseNormalization(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/LocalResponseNormalizationGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/LocalResponseNormalizationGrad.java
deleted file mode 100644
index 086ea52f53e..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/LocalResponseNormalizationGrad.java
+++ /dev/null
@@ -1,169 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.nn;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Gradients for Local Response Normalization.
- * 
- * @param <T> data type for {@code output()} output
- */
-public final class LocalResponseNormalizationGrad<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.nn.LocalResponseNormalizationGrad}
-   */
-  public static class Options {
-    
-    /**
-     * @param depthRadius A depth radius.
-     */
-    public Options depthRadius(Long depthRadius) {
-      this.depthRadius = depthRadius;
-      return this;
-    }
-    
-    /**
-     * @param bias An offset (usually > 0 to avoid dividing by 0).
-     */
-    public Options bias(Float bias) {
-      this.bias = bias;
-      return this;
-    }
-    
-    /**
-     * @param alpha A scale factor, usually positive.
-     */
-    public Options alpha(Float alpha) {
-      this.alpha = alpha;
-      return this;
-    }
-    
-    /**
-     * @param beta An exponent.
-     */
-    public Options beta(Float beta) {
-      this.beta = beta;
-      return this;
-    }
-    
-    private Long depthRadius;
-    private Float bias;
-    private Float alpha;
-    private Float beta;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new LocalResponseNormalizationGrad operation.
-   * 
-   * @param scope current scope
-   * @param inputGrads 4-D with shape `[batch, height, width, channels]`.
-   * @param inputImage 4-D with shape `[batch, height, width, channels]`.
-   * @param outputImage 4-D with shape `[batch, height, width, channels]`.
-   * @param options carries optional attributes values
-   * @return a new instance of LocalResponseNormalizationGrad
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> LocalResponseNormalizationGrad<T> create(Scope scope, Operand<T> inputGrads, Operand<T> inputImage, Operand<T> outputImage, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("LRNGrad", scope.makeOpName("LocalResponseNormalizationGrad"));
-    opBuilder.addInput(inputGrads.asOutput(scope));
-    opBuilder.addInput(inputImage.asOutput(scope));
-    opBuilder.addInput(outputImage.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.depthRadius != null) {
-          opBuilder.setAttr("depth_radius", opts.depthRadius);
-        }
-        if (opts.bias != null) {
-          opBuilder.setAttr("bias", opts.bias);
-        }
-        if (opts.alpha != null) {
-          opBuilder.setAttr("alpha", opts.alpha);
-        }
-        if (opts.beta != null) {
-          opBuilder.setAttr("beta", opts.beta);
-        }
-      }
-    }
-    return new LocalResponseNormalizationGrad<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param depthRadius A depth radius.
-   */
-  public static Options depthRadius(Long depthRadius) {
-    return new Options().depthRadius(depthRadius);
-  }
-  
-  /**
-   * @param bias An offset (usually > 0 to avoid dividing by 0).
-   */
-  public static Options bias(Float bias) {
-    return new Options().bias(bias);
-  }
-  
-  /**
-   * @param alpha A scale factor, usually positive.
-   */
-  public static Options alpha(Float alpha) {
-    return new Options().alpha(alpha);
-  }
-  
-  /**
-   * @param beta An exponent.
-   */
-  public static Options beta(Float beta) {
-    return new Options().beta(beta);
-  }
-  
-  /**
-   * The gradients for LRN.
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "LRNGrad";
-  
-  private Output<T> output;
-  
-  private LocalResponseNormalizationGrad(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/LogSoftmax.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/LogSoftmax.java
deleted file mode 100644
index dd035488742..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/LogSoftmax.java
+++ /dev/null
@@ -1,79 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.nn;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Computes log softmax activations.
- * <p>
- * For each batch `i` and class `j` we have
- * <p>
- *     logsoftmax[i, j] = logits[i, j] - log(sum(exp(logits[i])))
- * 
- * @param <T> data type for {@code logsoftmax()} output
- */
-@Operator(group = "nn")
-public final class LogSoftmax<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new LogSoftmax operation.
-   * 
-   * @param scope current scope
-   * @param logits 2-D with shape `[batch_size, num_classes]`.
-   * @return a new instance of LogSoftmax
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> LogSoftmax<T> create(Scope scope, Operand<T> logits) {
-    OperationBuilder opBuilder = scope.env().opBuilder("LogSoftmax", scope.makeOpName("LogSoftmax"));
-    opBuilder.addInput(logits.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new LogSoftmax<T>(opBuilder.build());
-  }
-  
-  /**
-   * Same shape as `logits`.
-   */
-  public Output<T> logsoftmax() {
-    return logsoftmax;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return logsoftmax;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "LogSoftmax";
-  
-  private Output<T> logsoftmax;
-  
-  private LogSoftmax(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    logsoftmax = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPool.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPool.java
deleted file mode 100644
index c2e5977e1c7..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPool.java
+++ /dev/null
@@ -1,125 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.nn;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.family.TType;
-
-/**
- * Performs max pooling on the input.
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator(group = "nn")
-public final class MaxPool<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.nn.MaxPool}
-   */
-  public static class Options {
-    
-    /**
-     * @param dataFormat Specify the data format of the input and output data. With the
-     * default format "NHWC", the data is stored in the order of:
-     *     [batch, in_height, in_width, in_channels].
-     * Alternatively, the format could be "NCHW", the data storage order of:
-     *     [batch, in_channels, in_height, in_width].
-     */
-    public Options dataFormat(String dataFormat) {
-      this.dataFormat = dataFormat;
-      return this;
-    }
-    
-    private String dataFormat;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new MaxPool operation.
-   * 
-   * @param scope current scope
-   * @param input 4-D input to pool over.
-   * @param ksize The size of the window for each dimension of the input tensor.
-   * @param strides The stride of the sliding window for each dimension of the
-   * input tensor.
-   * @param padding The type of padding algorithm to use.
-   * @param options carries optional attributes values
-   * @return a new instance of MaxPool
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> MaxPool<T> create(Scope scope, Operand<T> input, Operand<TInt32> ksize, Operand<TInt32> strides, String padding, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("MaxPoolV2", scope.makeOpName("MaxPool"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder.addInput(ksize.asOutput(scope));
-    opBuilder.addInput(strides.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("padding", padding);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.dataFormat != null) {
-          opBuilder.setAttr("data_format", opts.dataFormat);
-        }
-      }
-    }
-    return new MaxPool<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param dataFormat Specify the data format of the input and output data. With the
-   * default format "NHWC", the data is stored in the order of:
-   *     [batch, in_height, in_width, in_channels].
-   * Alternatively, the format could be "NCHW", the data storage order of:
-   *     [batch, in_channels, in_height, in_width].
-   */
-  public static Options dataFormat(String dataFormat) {
-    return new Options().dataFormat(dataFormat);
-  }
-  
-  /**
-   * The max pooled output tensor.
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "MaxPoolV2";
-  
-  private Output<T> output;
-  
-  private MaxPool(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPool3d.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPool3d.java
deleted file mode 100644
index 72a8241fd93..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPool3d.java
+++ /dev/null
@@ -1,134 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.nn;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Performs 3D max pooling on the input.
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator(group = "nn")
-public final class MaxPool3d<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.nn.MaxPool3d}
-   */
-  public static class Options {
-    
-    /**
-     * @param dataFormat The data format of the input and output data. With the
-     * default format "NDHWC", the data is stored in the order of:
-     *     [batch, in_depth, in_height, in_width, in_channels].
-     * Alternatively, the format could be "NCDHW", the data storage order is:
-     *     [batch, in_channels, in_depth, in_height, in_width].
-     */
-    public Options dataFormat(String dataFormat) {
-      this.dataFormat = dataFormat;
-      return this;
-    }
-    
-    private String dataFormat;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new MaxPool3d operation.
-   * 
-   * @param scope current scope
-   * @param input Shape `[batch, depth, rows, cols, channels]` tensor to pool over.
-   * @param ksize 1-D tensor of length 5. The size of the window for each dimension of
-   * the input tensor. Must have `ksize[0] = ksize[4] = 1`.
-   * @param strides 1-D tensor of length 5. The stride of the sliding window for each
-   * dimension of `input`. Must have `strides[0] = strides[4] = 1`.
-   * @param padding The type of padding algorithm to use.
-   * @param options carries optional attributes values
-   * @return a new instance of MaxPool3d
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> MaxPool3d<T> create(Scope scope, Operand<T> input, List<Long> ksize, List<Long> strides, String padding, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("MaxPool3D", scope.makeOpName("MaxPool3d"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    long[] ksizeArray = new long[ksize.size()];
-    for (int i = 0; i < ksizeArray.length; ++i) {
-      ksizeArray[i] = ksize.get(i);
-    }
-    opBuilder.setAttr("ksize", ksizeArray);
-    long[] stridesArray = new long[strides.size()];
-    for (int i = 0; i < stridesArray.length; ++i) {
-      stridesArray[i] = strides.get(i);
-    }
-    opBuilder.setAttr("strides", stridesArray);
-    opBuilder.setAttr("padding", padding);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.dataFormat != null) {
-          opBuilder.setAttr("data_format", opts.dataFormat);
-        }
-      }
-    }
-    return new MaxPool3d<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param dataFormat The data format of the input and output data. With the
-   * default format "NDHWC", the data is stored in the order of:
-   *     [batch, in_depth, in_height, in_width, in_channels].
-   * Alternatively, the format could be "NCDHW", the data storage order is:
-   *     [batch, in_channels, in_depth, in_height, in_width].
-   */
-  public static Options dataFormat(String dataFormat) {
-    return new Options().dataFormat(dataFormat);
-  }
-  
-  /**
-   * The max pooled output tensor.
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "MaxPool3D";
-  
-  private Output<T> output;
-  
-  private MaxPool3d(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPool3dGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPool3dGrad.java
deleted file mode 100644
index 3fd6f901792..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPool3dGrad.java
+++ /dev/null
@@ -1,137 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.nn;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Computes gradients of 3D max pooling function.
- * 
- * @param <U> data type for {@code output()} output
- */
-@Operator(group = "nn")
-public final class MaxPool3dGrad<U extends TNumber> extends RawOp implements Operand<U> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.nn.MaxPool3dGrad}
-   */
-  public static class Options {
-    
-    /**
-     * @param dataFormat The data format of the input and output data. With the
-     * default format "NDHWC", the data is stored in the order of:
-     *     [batch, in_depth, in_height, in_width, in_channels].
-     * Alternatively, the format could be "NCDHW", the data storage order is:
-     *     [batch, in_channels, in_depth, in_height, in_width].
-     */
-    public Options dataFormat(String dataFormat) {
-      this.dataFormat = dataFormat;
-      return this;
-    }
-    
-    private String dataFormat;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new MaxPool3dGrad operation.
-   * 
-   * @param scope current scope
-   * @param origInput The original input tensor.
-   * @param origOutput The original output tensor.
-   * @param grad Output backprop of shape `[batch, depth, rows, cols, channels]`.
-   * @param ksize 1-D tensor of length 5. The size of the window for each dimension of
-   * the input tensor. Must have `ksize[0] = ksize[4] = 1`.
-   * @param strides 1-D tensor of length 5. The stride of the sliding window for each
-   * dimension of `input`. Must have `strides[0] = strides[4] = 1`.
-   * @param padding The type of padding algorithm to use.
-   * @param options carries optional attributes values
-   * @return a new instance of MaxPool3dGrad
-   */
-  @Endpoint(describeByClass = true)
-  public static <U extends TNumber, T extends TNumber> MaxPool3dGrad<U> create(Scope scope, Operand<T> origInput, Operand<T> origOutput, Operand<U> grad, List<Long> ksize, List<Long> strides, String padding, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("MaxPool3DGrad", scope.makeOpName("MaxPool3dGrad"));
-    opBuilder.addInput(origInput.asOutput(scope));
-    opBuilder.addInput(origOutput.asOutput(scope));
-    opBuilder.addInput(grad.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    long[] ksizeArray = new long[ksize.size()];
-    for (int i = 0; i < ksizeArray.length; ++i) {
-      ksizeArray[i] = ksize.get(i);
-    }
-    opBuilder.setAttr("ksize", ksizeArray);
-    long[] stridesArray = new long[strides.size()];
-    for (int i = 0; i < stridesArray.length; ++i) {
-      stridesArray[i] = strides.get(i);
-    }
-    opBuilder.setAttr("strides", stridesArray);
-    opBuilder.setAttr("padding", padding);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.dataFormat != null) {
-          opBuilder.setAttr("data_format", opts.dataFormat);
-        }
-      }
-    }
-    return new MaxPool3dGrad<U>(opBuilder.build());
-  }
-  
-  /**
-   * @param dataFormat The data format of the input and output data. With the
-   * default format "NDHWC", the data is stored in the order of:
-   *     [batch, in_depth, in_height, in_width, in_channels].
-   * Alternatively, the format could be "NCDHW", the data storage order is:
-   *     [batch, in_channels, in_depth, in_height, in_width].
-   */
-  public static Options dataFormat(String dataFormat) {
-    return new Options().dataFormat(dataFormat);
-  }
-  
-  /**
-   */
-  public Output<U> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<U> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "MaxPool3DGrad";
-  
-  private Output<U> output;
-  
-  private MaxPool3dGrad(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPool3dGradGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPool3dGradGrad.java
deleted file mode 100644
index e9ccd76db0d..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPool3dGradGrad.java
+++ /dev/null
@@ -1,138 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.nn;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Computes second-order gradients of the maxpooling function.
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator(group = "nn")
-public final class MaxPool3dGradGrad<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.nn.MaxPool3dGradGrad}
-   */
-  public static class Options {
-    
-    /**
-     * @param dataFormat The data format of the input and output data. With the
-     * default format "NDHWC", the data is stored in the order of:
-     *     [batch, in_depth, in_height, in_width, in_channels].
-     * Alternatively, the format could be "NCDHW", the data storage order is:
-     *     [batch, in_channels, in_depth, in_height, in_width].
-     */
-    public Options dataFormat(String dataFormat) {
-      this.dataFormat = dataFormat;
-      return this;
-    }
-    
-    private String dataFormat;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new MaxPool3dGradGrad operation.
-   * 
-   * @param scope current scope
-   * @param origInput The original input tensor.
-   * @param origOutput The original output tensor.
-   * @param grad Output backprop of shape `[batch, depth, rows, cols, channels]`.
-   * @param ksize 1-D tensor of length 5. The size of the window for each dimension of
-   * the input tensor. Must have `ksize[0] = ksize[4] = 1`.
-   * @param strides 1-D tensor of length 5. The stride of the sliding window for each
-   * dimension of `input`. Must have `strides[0] = strides[4] = 1`.
-   * @param padding The type of padding algorithm to use.
-   * @param options carries optional attributes values
-   * @return a new instance of MaxPool3dGradGrad
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> MaxPool3dGradGrad<T> create(Scope scope, Operand<T> origInput, Operand<T> origOutput, Operand<T> grad, List<Long> ksize, List<Long> strides, String padding, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("MaxPool3DGradGrad", scope.makeOpName("MaxPool3dGradGrad"));
-    opBuilder.addInput(origInput.asOutput(scope));
-    opBuilder.addInput(origOutput.asOutput(scope));
-    opBuilder.addInput(grad.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    long[] ksizeArray = new long[ksize.size()];
-    for (int i = 0; i < ksizeArray.length; ++i) {
-      ksizeArray[i] = ksize.get(i);
-    }
-    opBuilder.setAttr("ksize", ksizeArray);
-    long[] stridesArray = new long[strides.size()];
-    for (int i = 0; i < stridesArray.length; ++i) {
-      stridesArray[i] = strides.get(i);
-    }
-    opBuilder.setAttr("strides", stridesArray);
-    opBuilder.setAttr("padding", padding);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.dataFormat != null) {
-          opBuilder.setAttr("data_format", opts.dataFormat);
-        }
-      }
-    }
-    return new MaxPool3dGradGrad<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param dataFormat The data format of the input and output data. With the
-   * default format "NDHWC", the data is stored in the order of:
-   *     [batch, in_depth, in_height, in_width, in_channels].
-   * Alternatively, the format could be "NCDHW", the data storage order is:
-   *     [batch, in_channels, in_depth, in_height, in_width].
-   */
-  public static Options dataFormat(String dataFormat) {
-    return new Options().dataFormat(dataFormat);
-  }
-  
-  /**
-   * Gradients of gradients w.r.t. the input to `max_pool`.
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "MaxPool3DGradGrad";
-  
-  private Output<T> output;
-  
-  private MaxPool3dGradGrad(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPoolGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPoolGrad.java
deleted file mode 100644
index c1951a034f9..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPoolGrad.java
+++ /dev/null
@@ -1,129 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.nn;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Computes gradients of the maxpooling function.
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator(group = "nn")
-public final class MaxPoolGrad<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.nn.MaxPoolGrad}
-   */
-  public static class Options {
-    
-    /**
-     * @param dataFormat Specify the data format of the input and output data. With the
-     * default format "NHWC", the data is stored in the order of:
-     *     [batch, in_height, in_width, in_channels].
-     * Alternatively, the format could be "NCHW", the data storage order of:
-     *     [batch, in_channels, in_height, in_width].
-     */
-    public Options dataFormat(String dataFormat) {
-      this.dataFormat = dataFormat;
-      return this;
-    }
-    
-    private String dataFormat;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new MaxPoolGrad operation.
-   * 
-   * @param scope current scope
-   * @param origInput The original input tensor.
-   * @param origOutput The original output tensor.
-   * @param grad 4-D.  Gradients w.r.t. the output of `max_pool`.
-   * @param ksize The size of the window for each dimension of the input tensor.
-   * @param strides The stride of the sliding window for each dimension of the
-   * input tensor.
-   * @param padding The type of padding algorithm to use.
-   * @param options carries optional attributes values
-   * @return a new instance of MaxPoolGrad
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> MaxPoolGrad<T> create(Scope scope, Operand<T> origInput, Operand<T> origOutput, Operand<T> grad, Operand<TInt32> ksize, Operand<TInt32> strides, String padding, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("MaxPoolGradV2", scope.makeOpName("MaxPoolGrad"));
-    opBuilder.addInput(origInput.asOutput(scope));
-    opBuilder.addInput(origOutput.asOutput(scope));
-    opBuilder.addInput(grad.asOutput(scope));
-    opBuilder.addInput(ksize.asOutput(scope));
-    opBuilder.addInput(strides.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("padding", padding);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.dataFormat != null) {
-          opBuilder.setAttr("data_format", opts.dataFormat);
-        }
-      }
-    }
-    return new MaxPoolGrad<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param dataFormat Specify the data format of the input and output data. With the
-   * default format "NHWC", the data is stored in the order of:
-   *     [batch, in_height, in_width, in_channels].
-   * Alternatively, the format could be "NCHW", the data storage order of:
-   *     [batch, in_channels, in_height, in_width].
-   */
-  public static Options dataFormat(String dataFormat) {
-    return new Options().dataFormat(dataFormat);
-  }
-  
-  /**
-   * Gradients w.r.t. the input to `max_pool`.
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "MaxPoolGradV2";
-  
-  private Output<T> output;
-  
-  private MaxPoolGrad(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPoolGradGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPoolGradGrad.java
deleted file mode 100644
index 3810b6845a5..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPoolGradGrad.java
+++ /dev/null
@@ -1,129 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.nn;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Computes second-order gradients of the maxpooling function.
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator(group = "nn")
-public final class MaxPoolGradGrad<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.nn.MaxPoolGradGrad}
-   */
-  public static class Options {
-    
-    /**
-     * @param dataFormat Specify the data format of the input and output data. With the
-     * default format "NHWC", the data is stored in the order of:
-     *     [batch, in_height, in_width, in_channels].
-     * Alternatively, the format could be "NCHW", the data storage order of:
-     *     [batch, in_channels, in_height, in_width].
-     */
-    public Options dataFormat(String dataFormat) {
-      this.dataFormat = dataFormat;
-      return this;
-    }
-    
-    private String dataFormat;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new MaxPoolGradGrad operation.
-   * 
-   * @param scope current scope
-   * @param origInput The original input tensor.
-   * @param origOutput The original output tensor.
-   * @param grad 4-D.  Gradients of gradients w.r.t. the input of `max_pool`.
-   * @param ksize The size of the window for each dimension of the input tensor.
-   * @param strides The stride of the sliding window for each dimension of the
-   * input tensor.
-   * @param padding The type of padding algorithm to use.
-   * @param options carries optional attributes values
-   * @return a new instance of MaxPoolGradGrad
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> MaxPoolGradGrad<T> create(Scope scope, Operand<T> origInput, Operand<T> origOutput, Operand<T> grad, Operand<TInt32> ksize, Operand<TInt32> strides, String padding, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("MaxPoolGradGradV2", scope.makeOpName("MaxPoolGradGrad"));
-    opBuilder.addInput(origInput.asOutput(scope));
-    opBuilder.addInput(origOutput.asOutput(scope));
-    opBuilder.addInput(grad.asOutput(scope));
-    opBuilder.addInput(ksize.asOutput(scope));
-    opBuilder.addInput(strides.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("padding", padding);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.dataFormat != null) {
-          opBuilder.setAttr("data_format", opts.dataFormat);
-        }
-      }
-    }
-    return new MaxPoolGradGrad<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param dataFormat Specify the data format of the input and output data. With the
-   * default format "NHWC", the data is stored in the order of:
-   *     [batch, in_height, in_width, in_channels].
-   * Alternatively, the format could be "NCHW", the data storage order of:
-   *     [batch, in_channels, in_height, in_width].
-   */
-  public static Options dataFormat(String dataFormat) {
-    return new Options().dataFormat(dataFormat);
-  }
-  
-  /**
-   * Gradients of gradients w.r.t. the input to `max_pool`.
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "MaxPoolGradGradV2";
-  
-  private Output<T> output;
-  
-  private MaxPoolGradGrad(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPoolGradGradWithArgmax.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPoolGradGradWithArgmax.java
deleted file mode 100644
index 8f22d65be65..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPoolGradGradWithArgmax.java
+++ /dev/null
@@ -1,130 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.nn;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Computes second-order gradients of the maxpooling function.
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator(group = "nn")
-public final class MaxPoolGradGradWithArgmax<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.nn.MaxPoolGradGradWithArgmax}
-   */
-  public static class Options {
-    
-    /**
-     * @param includeBatchInIndex Whether to include batch dimension in flattened index of `argmax`.
-     */
-    public Options includeBatchInIndex(Boolean includeBatchInIndex) {
-      this.includeBatchInIndex = includeBatchInIndex;
-      return this;
-    }
-    
-    private Boolean includeBatchInIndex;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new MaxPoolGradGradWithArgmax operation.
-   * 
-   * @param scope current scope
-   * @param input The original input.
-   * @param grad 4-D with shape `[batch, height, width, channels]`.  Gradients w.r.t. the
-   * input of `max_pool`.
-   * @param argmax The indices of the maximum values chosen for each output of `max_pool`.
-   * @param ksize The size of the window for each dimension of the input tensor.
-   * @param strides The stride of the sliding window for each dimension of the
-   * input tensor.
-   * @param padding The type of padding algorithm to use.
-   * @param options carries optional attributes values
-   * @return a new instance of MaxPoolGradGradWithArgmax
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber, U extends TNumber> MaxPoolGradGradWithArgmax<T> create(Scope scope, Operand<T> input, Operand<T> grad, Operand<U> argmax, List<Long> ksize, List<Long> strides, String padding, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("MaxPoolGradGradWithArgmax", scope.makeOpName("MaxPoolGradGradWithArgmax"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder.addInput(grad.asOutput(scope));
-    opBuilder.addInput(argmax.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    long[] ksizeArray = new long[ksize.size()];
-    for (int i = 0; i < ksizeArray.length; ++i) {
-      ksizeArray[i] = ksize.get(i);
-    }
-    opBuilder.setAttr("ksize", ksizeArray);
-    long[] stridesArray = new long[strides.size()];
-    for (int i = 0; i < stridesArray.length; ++i) {
-      stridesArray[i] = strides.get(i);
-    }
-    opBuilder.setAttr("strides", stridesArray);
-    opBuilder.setAttr("padding", padding);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.includeBatchInIndex != null) {
-          opBuilder.setAttr("include_batch_in_index", opts.includeBatchInIndex);
-        }
-      }
-    }
-    return new MaxPoolGradGradWithArgmax<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param includeBatchInIndex Whether to include batch dimension in flattened index of `argmax`.
-   */
-  public static Options includeBatchInIndex(Boolean includeBatchInIndex) {
-    return new Options().includeBatchInIndex(includeBatchInIndex);
-  }
-  
-  /**
-   * Gradients of gradients w.r.t. the input of `max_pool`.
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "MaxPoolGradGradWithArgmax";
-  
-  private Output<T> output;
-  
-  private MaxPoolGradGradWithArgmax(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPoolGradWithArgmax.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPoolGradWithArgmax.java
deleted file mode 100644
index 551b3ef0594..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPoolGradWithArgmax.java
+++ /dev/null
@@ -1,129 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.nn;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Computes gradients of the maxpooling function.
- * 
- * @param <T> data type for {@code output()} output
- */
-public final class MaxPoolGradWithArgmax<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.nn.MaxPoolGradWithArgmax}
-   */
-  public static class Options {
-    
-    /**
-     * @param includeBatchInIndex Whether to include batch dimension in flattened index of `argmax`.
-     */
-    public Options includeBatchInIndex(Boolean includeBatchInIndex) {
-      this.includeBatchInIndex = includeBatchInIndex;
-      return this;
-    }
-    
-    private Boolean includeBatchInIndex;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new MaxPoolGradWithArgmax operation.
-   * 
-   * @param scope current scope
-   * @param input The original input.
-   * @param grad 4-D with shape `[batch, height, width, channels]`.  Gradients w.r.t. the
-   * output of `max_pool`.
-   * @param argmax The indices of the maximum values chosen for each output of `max_pool`.
-   * @param ksize The size of the window for each dimension of the input tensor.
-   * @param strides The stride of the sliding window for each dimension of the
-   * input tensor.
-   * @param padding The type of padding algorithm to use.
-   * @param options carries optional attributes values
-   * @return a new instance of MaxPoolGradWithArgmax
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber, U extends TNumber> MaxPoolGradWithArgmax<T> create(Scope scope, Operand<T> input, Operand<T> grad, Operand<U> argmax, List<Long> ksize, List<Long> strides, String padding, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("MaxPoolGradWithArgmax", scope.makeOpName("MaxPoolGradWithArgmax"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder.addInput(grad.asOutput(scope));
-    opBuilder.addInput(argmax.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    long[] ksizeArray = new long[ksize.size()];
-    for (int i = 0; i < ksizeArray.length; ++i) {
-      ksizeArray[i] = ksize.get(i);
-    }
-    opBuilder.setAttr("ksize", ksizeArray);
-    long[] stridesArray = new long[strides.size()];
-    for (int i = 0; i < stridesArray.length; ++i) {
-      stridesArray[i] = strides.get(i);
-    }
-    opBuilder.setAttr("strides", stridesArray);
-    opBuilder.setAttr("padding", padding);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.includeBatchInIndex != null) {
-          opBuilder.setAttr("include_batch_in_index", opts.includeBatchInIndex);
-        }
-      }
-    }
-    return new MaxPoolGradWithArgmax<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param includeBatchInIndex Whether to include batch dimension in flattened index of `argmax`.
-   */
-  public static Options includeBatchInIndex(Boolean includeBatchInIndex) {
-    return new Options().includeBatchInIndex(includeBatchInIndex);
-  }
-  
-  /**
-   * Gradients w.r.t. the input of `max_pool`.
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "MaxPoolGradWithArgmax";
-  
-  private Output<T> output;
-  
-  private MaxPoolGradWithArgmax(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPoolWithArgmax.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPoolWithArgmax.java
deleted file mode 100644
index 870e56fd600..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPoolWithArgmax.java
+++ /dev/null
@@ -1,160 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.nn;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Performs max pooling on the input and outputs both max values and indices.
- * <p>
- * The indices in `argmax` are flattened, so that a maximum value at position
- * `[b, y, x, c]` becomes flattened index:
- * `(y * width + x) * channels + c` if `include_batch_in_index` is False;
- * `((b * height + y) * width + x) * channels + c` if `include_batch_in_index` is True.
- * <p>
- * The indices returned are always in `[0, height) x [0, width)` before flattening,
- * even if padding is involved and the mathematically correct answer is outside
- * (either negative or too large).  This is a bug, but fixing it is difficult to do
- * in a safe backwards compatible way, especially due to flattening.
- * 
- * @param <T> data type for {@code output()} output
- * @param <U> data type for {@code argmax()} output
- */
-@Operator(group = "nn")
-public final class MaxPoolWithArgmax<T extends TNumber, U extends TNumber> extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.nn.MaxPoolWithArgmax}
-   */
-  public static class Options {
-    
-    /**
-     * @param includeBatchInIndex Whether to include batch dimension in flattened index of `argmax`.
-     */
-    public Options includeBatchInIndex(Boolean includeBatchInIndex) {
-      this.includeBatchInIndex = includeBatchInIndex;
-      return this;
-    }
-    
-    private Boolean includeBatchInIndex;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new MaxPoolWithArgmax operation.
-   * 
-   * @param scope current scope
-   * @param input 4-D with shape `[batch, height, width, channels]`.  Input to pool over.
-   * @param ksize The size of the window for each dimension of the input tensor.
-   * @param strides The stride of the sliding window for each dimension of the
-   * input tensor.
-   * @param Targmax 
-   * @param padding The type of padding algorithm to use.
-   * @param options carries optional attributes values
-   * @return a new instance of MaxPoolWithArgmax
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber, U extends TNumber> MaxPoolWithArgmax<T, U> create(Scope scope, Operand<T> input, List<Long> ksize, List<Long> strides, Class<U> Targmax, String padding, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("MaxPoolWithArgmax", scope.makeOpName("MaxPoolWithArgmax"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    long[] ksizeArray = new long[ksize.size()];
-    for (int i = 0; i < ksizeArray.length; ++i) {
-      ksizeArray[i] = ksize.get(i);
-    }
-    opBuilder.setAttr("ksize", ksizeArray);
-    long[] stridesArray = new long[strides.size()];
-    for (int i = 0; i < stridesArray.length; ++i) {
-      stridesArray[i] = strides.get(i);
-    }
-    opBuilder.setAttr("strides", stridesArray);
-    opBuilder.setAttr("Targmax", Targmax);
-    opBuilder.setAttr("padding", padding);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.includeBatchInIndex != null) {
-          opBuilder.setAttr("include_batch_in_index", opts.includeBatchInIndex);
-        }
-      }
-    }
-    return new MaxPoolWithArgmax<T, U>(opBuilder.build());
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new MaxPoolWithArgmax operation using default output types.
-   * 
-   * @param scope current scope
-   * @param input 4-D with shape `[batch, height, width, channels]`.  Input to pool over.
-   * @param ksize The size of the window for each dimension of the input tensor.
-   * @param strides The stride of the sliding window for each dimension of the
-   * input tensor.
-   * @param padding The type of padding algorithm to use.
-   * @param options carries optional attributes values
-   * @return a new instance of MaxPoolWithArgmax
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> MaxPoolWithArgmax<T, TInt64> create(Scope scope, Operand<T> input, List<Long> ksize, List<Long> strides, String padding, Options... options) {
-    return create(scope, input, ksize, strides, TInt64.class, padding, options);
-  }
-  
-  /**
-   * @param includeBatchInIndex Whether to include batch dimension in flattened index of `argmax`.
-   */
-  public static Options includeBatchInIndex(Boolean includeBatchInIndex) {
-    return new Options().includeBatchInIndex(includeBatchInIndex);
-  }
-  
-  /**
-   * The max pooled output tensor.
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  /**
-   * 4-D.  The flattened indices of the max values chosen for each output.
-   */
-  public Output<U> argmax() {
-    return argmax;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "MaxPoolWithArgmax";
-  
-  private Output<T> output;
-  private Output<U> argmax;
-  
-  private MaxPoolWithArgmax(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-    argmax = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/NthElement.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/NthElement.java
deleted file mode 100644
index 9f55dce0be7..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/NthElement.java
+++ /dev/null
@@ -1,123 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.nn;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Finds values of the `n`-th order statistic for the last dimension.
- * <p>
- * If the input is a vector (rank-1), finds the entries which is the nth-smallest
- * value in the vector and outputs their values as scalar tensor.
- * <p>
- * For matrices (resp. higher rank input), computes the entries which is the
- * nth-smallest value in each row (resp. vector along the last dimension). Thus,
- * <p>
- *     values.shape = input.shape[:-1]
- * 
- * @param <T> data type for {@code values()} output
- */
-@Operator(group = "nn")
-public final class NthElement<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.nn.NthElement}
-   */
-  public static class Options {
-    
-    /**
-     * @param reverse When set to True, find the nth-largest value in the vector and vice
-     * versa.
-     */
-    public Options reverse(Boolean reverse) {
-      this.reverse = reverse;
-      return this;
-    }
-    
-    private Boolean reverse;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new NthElement operation.
-   * 
-   * @param scope current scope
-   * @param input 1-D or higher with last dimension at least `n+1`.
-   * @param n 0-D. Position of sorted vector to select along the last dimension (along
-   * each row for matrices). Valid range of n is `[0, input.shape[:-1])`
-   * @param options carries optional attributes values
-   * @return a new instance of NthElement
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> NthElement<T> create(Scope scope, Operand<T> input, Operand<TInt32> n, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("NthElement", scope.makeOpName("NthElement"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder.addInput(n.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.reverse != null) {
-          opBuilder.setAttr("reverse", opts.reverse);
-        }
-      }
-    }
-    return new NthElement<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param reverse When set to True, find the nth-largest value in the vector and vice
-   * versa.
-   */
-  public static Options reverse(Boolean reverse) {
-    return new Options().reverse(reverse);
-  }
-  
-  /**
-   * The `n`-th order statistic along each last dimensional slice.
-   */
-  public Output<T> values() {
-    return values;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return values;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "NthElement";
-  
-  private Output<T> values;
-  
-  private NthElement(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    values = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedAvgPool.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedAvgPool.java
deleted file mode 100644
index f573d560413..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedAvgPool.java
+++ /dev/null
@@ -1,109 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.nn;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.family.TType;
-
-/**
- * Produces the average pool of the input tensor for quantized types.
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator(group = "nn")
-public final class QuantizedAvgPool<T extends TType> extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new QuantizedAvgPool operation.
-   * 
-   * @param scope current scope
-   * @param input 4-D with shape `[batch, height, width, channels]`.
-   * @param minInput The float value that the lowest quantized input value represents.
-   * @param maxInput The float value that the highest quantized input value represents.
-   * @param ksize The size of the window for each dimension of the input tensor.
-   * The length must be 4 to match the number of dimensions of the input.
-   * @param strides The stride of the sliding window for each dimension of the input
-   * tensor.  The length must be 4 to match the number of dimensions of the input.
-   * @param padding The type of padding algorithm to use.
-   * @return a new instance of QuantizedAvgPool
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> QuantizedAvgPool<T> create(Scope scope, Operand<T> input, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, List<Long> ksize, List<Long> strides, String padding) {
-    OperationBuilder opBuilder = scope.env().opBuilder("QuantizedAvgPool", scope.makeOpName("QuantizedAvgPool"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder.addInput(minInput.asOutput(scope));
-    opBuilder.addInput(maxInput.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    long[] ksizeArray = new long[ksize.size()];
-    for (int i = 0; i < ksizeArray.length; ++i) {
-      ksizeArray[i] = ksize.get(i);
-    }
-    opBuilder.setAttr("ksize", ksizeArray);
-    long[] stridesArray = new long[strides.size()];
-    for (int i = 0; i < stridesArray.length; ++i) {
-      stridesArray[i] = strides.get(i);
-    }
-    opBuilder.setAttr("strides", stridesArray);
-    opBuilder.setAttr("padding", padding);
-    return new QuantizedAvgPool<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  /**
-   * The float value that the lowest quantized output value represents.
-   */
-  public Output<TFloat32> minOutput() {
-    return minOutput;
-  }
-  
-  /**
-   * The float value that the highest quantized output value represents.
-   */
-  public Output<TFloat32> maxOutput() {
-    return maxOutput;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "QuantizedAvgPool";
-  
-  private Output<T> output;
-  private Output<TFloat32> minOutput;
-  private Output<TFloat32> maxOutput;
-  
-  private QuantizedAvgPool(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-    minOutput = operation.output(outputIdx++);
-    maxOutput = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedBatchNormWithGlobalNormalization.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedBatchNormWithGlobalNormalization.java
deleted file mode 100644
index 07caa55bf86..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedBatchNormWithGlobalNormalization.java
+++ /dev/null
@@ -1,131 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.nn;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.family.TType;
-
-/**
- * Quantized Batch normalization.
- * <p>
- * This op is deprecated and will be removed in the future. Prefer
- * `tf.nn.batch_normalization`.
- * 
- * @param <U> data type for {@code result()} output
- */
-@Operator(group = "nn")
-public final class QuantizedBatchNormWithGlobalNormalization<U extends TType> extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new QuantizedBatchNormWithGlobalNormalization operation.
-   * 
-   * @param scope current scope
-   * @param t A 4D input Tensor.
-   * @param tMin The value represented by the lowest quantized input.
-   * @param tMax The value represented by the highest quantized input.
-   * @param m A 1D mean Tensor with size matching the last dimension of t.
-   * This is the first output from tf.nn.moments,
-   * or a saved moving average thereof.
-   * @param mMin The value represented by the lowest quantized mean.
-   * @param mMax The value represented by the highest quantized mean.
-   * @param v A 1D variance Tensor with size matching the last dimension of t.
-   * This is the second output from tf.nn.moments,
-   * or a saved moving average thereof.
-   * @param vMin The value represented by the lowest quantized variance.
-   * @param vMax The value represented by the highest quantized variance.
-   * @param beta A 1D beta Tensor with size matching the last dimension of t.
-   * An offset to be added to the normalized tensor.
-   * @param betaMin The value represented by the lowest quantized offset.
-   * @param betaMax The value represented by the highest quantized offset.
-   * @param gamma A 1D gamma Tensor with size matching the last dimension of t.
-   * If "scale_after_normalization" is true, this tensor will be multiplied
-   * with the normalized tensor.
-   * @param gammaMin The value represented by the lowest quantized gamma.
-   * @param gammaMax The value represented by the highest quantized gamma.
-   * @param outType 
-   * @param varianceEpsilon A small float number to avoid dividing by 0.
-   * @param scaleAfterNormalization A bool indicating whether the resulted tensor
-   * needs to be multiplied with gamma.
-   * @return a new instance of QuantizedBatchNormWithGlobalNormalization
-   */
-  @Endpoint(describeByClass = true)
-  public static <U extends TType, T extends TType> QuantizedBatchNormWithGlobalNormalization<U> create(Scope scope, Operand<T> t, Operand<TFloat32> tMin, Operand<TFloat32> tMax, Operand<T> m, Operand<TFloat32> mMin, Operand<TFloat32> mMax, Operand<T> v, Operand<TFloat32> vMin, Operand<TFloat32> vMax, Operand<T> beta, Operand<TFloat32> betaMin, Operand<TFloat32> betaMax, Operand<T> gamma, Operand<TFloat32> gammaMin, Operand<TFloat32> gammaMax, Class<U> outType, Float varianceEpsilon, Boolean scaleAfterNormalization) {
-    OperationBuilder opBuilder = scope.env().opBuilder("QuantizedBatchNormWithGlobalNormalization", scope.makeOpName("QuantizedBatchNormWithGlobalNormalization"));
-    opBuilder.addInput(t.asOutput(scope));
-    opBuilder.addInput(tMin.asOutput(scope));
-    opBuilder.addInput(tMax.asOutput(scope));
-    opBuilder.addInput(m.asOutput(scope));
-    opBuilder.addInput(mMin.asOutput(scope));
-    opBuilder.addInput(mMax.asOutput(scope));
-    opBuilder.addInput(v.asOutput(scope));
-    opBuilder.addInput(vMin.asOutput(scope));
-    opBuilder.addInput(vMax.asOutput(scope));
-    opBuilder.addInput(beta.asOutput(scope));
-    opBuilder.addInput(betaMin.asOutput(scope));
-    opBuilder.addInput(betaMax.asOutput(scope));
-    opBuilder.addInput(gamma.asOutput(scope));
-    opBuilder.addInput(gammaMin.asOutput(scope));
-    opBuilder.addInput(gammaMax.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("out_type", outType);
-    opBuilder.setAttr("variance_epsilon", varianceEpsilon);
-    opBuilder.setAttr("scale_after_normalization", scaleAfterNormalization);
-    return new QuantizedBatchNormWithGlobalNormalization<U>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<U> result() {
-    return result;
-  }
-  
-  /**
-   */
-  public Output<TFloat32> resultMin() {
-    return resultMin;
-  }
-  
-  /**
-   */
-  public Output<TFloat32> resultMax() {
-    return resultMax;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "QuantizedBatchNormWithGlobalNormalization";
-  
-  private Output<U> result;
-  private Output<TFloat32> resultMin;
-  private Output<TFloat32> resultMax;
-  
-  private QuantizedBatchNormWithGlobalNormalization(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    result = operation.output(outputIdx++);
-    resultMin = operation.output(outputIdx++);
-    resultMax = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedBiasAdd.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedBiasAdd.java
deleted file mode 100644
index fddae6e6272..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedBiasAdd.java
+++ /dev/null
@@ -1,102 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.nn;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.family.TType;
-
-/**
- * Adds Tensor 'bias' to Tensor 'input' for Quantized types.
- * <p>
- * Broadcasts the values of bias on dimensions 0..N-2 of 'input'.
- * 
- * @param <V> data type for {@code output()} output
- */
-@Operator(group = "nn")
-public final class QuantizedBiasAdd<V extends TType> extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new QuantizedBiasAdd operation.
-   * 
-   * @param scope current scope
-   * @param input 
-   * @param bias A 1D bias Tensor with size matching the last dimension of 'input'.
-   * @param minInput The float value that the lowest quantized input value represents.
-   * @param maxInput The float value that the highest quantized input value represents.
-   * @param minBias The float value that the lowest quantized bias value represents.
-   * @param maxBias The float value that the highest quantized bias value represents.
-   * @param outType 
-   * @return a new instance of QuantizedBiasAdd
-   */
-  @Endpoint(describeByClass = true)
-  public static <V extends TType, T extends TType, U extends TType> QuantizedBiasAdd<V> create(Scope scope, Operand<T> input, Operand<U> bias, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, Operand<TFloat32> minBias, Operand<TFloat32> maxBias, Class<V> outType) {
-    OperationBuilder opBuilder = scope.env().opBuilder("QuantizedBiasAdd", scope.makeOpName("QuantizedBiasAdd"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder.addInput(bias.asOutput(scope));
-    opBuilder.addInput(minInput.asOutput(scope));
-    opBuilder.addInput(maxInput.asOutput(scope));
-    opBuilder.addInput(minBias.asOutput(scope));
-    opBuilder.addInput(maxBias.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("out_type", outType);
-    return new QuantizedBiasAdd<V>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<V> output() {
-    return output;
-  }
-  
-  /**
-   * The float value that the lowest quantized output value represents.
-   */
-  public Output<TFloat32> minOut() {
-    return minOut;
-  }
-  
-  /**
-   * The float value that the highest quantized output value represents.
-   */
-  public Output<TFloat32> maxOut() {
-    return maxOut;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "QuantizedBiasAdd";
-  
-  private Output<V> output;
-  private Output<TFloat32> minOut;
-  private Output<TFloat32> maxOut;
-  
-  private QuantizedBiasAdd(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-    minOut = operation.output(outputIdx++);
-    maxOut = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DAndRelu.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DAndRelu.java
deleted file mode 100644
index 071f2b0269e..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DAndRelu.java
+++ /dev/null
@@ -1,165 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.nn;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.family.TType;
-
-/**
- * @param <V> data type for {@code output()} output
- */
-public final class QuantizedConv2DAndRelu<V extends TType> extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.nn.QuantizedConv2DAndRelu}
-   */
-  public static class Options {
-    
-    /**
-     * @param dilations 
-     */
-    public Options dilations(List<Long> dilations) {
-      this.dilations = dilations;
-      return this;
-    }
-    
-    /**
-     * @param paddingList 
-     */
-    public Options paddingList(List<Long> paddingList) {
-      this.paddingList = paddingList;
-      return this;
-    }
-    
-    private List<Long> dilations;
-    private List<Long> paddingList;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new QuantizedConv2DAndRelu operation.
-   * 
-   * @param scope current scope
-   * @param input 
-   * @param filter 
-   * @param minInput 
-   * @param maxInput 
-   * @param minFilter 
-   * @param maxFilter 
-   * @param outType 
-   * @param strides 
-   * @param padding 
-   * @param options carries optional attributes values
-   * @return a new instance of QuantizedConv2DAndRelu
-   */
-  @Endpoint(describeByClass = true)
-  public static <V extends TType, T extends TType, U extends TType> QuantizedConv2DAndRelu<V> create(Scope scope, Operand<T> input, Operand<U> filter, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, Operand<TFloat32> minFilter, Operand<TFloat32> maxFilter, Class<V> outType, List<Long> strides, String padding, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("QuantizedConv2DAndRelu", scope.makeOpName("QuantizedConv2DAndRelu"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder.addInput(filter.asOutput(scope));
-    opBuilder.addInput(minInput.asOutput(scope));
-    opBuilder.addInput(maxInput.asOutput(scope));
-    opBuilder.addInput(minFilter.asOutput(scope));
-    opBuilder.addInput(maxFilter.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("out_type", outType);
-    long[] stridesArray = new long[strides.size()];
-    for (int i = 0; i < stridesArray.length; ++i) {
-      stridesArray[i] = strides.get(i);
-    }
-    opBuilder.setAttr("strides", stridesArray);
-    opBuilder.setAttr("padding", padding);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.dilations != null) {
-          long[] dilationsArray = new long[opts.dilations.size()];
-          for (int i = 0; i < dilationsArray.length; ++i) {
-            dilationsArray[i] = opts.dilations.get(i);
-          }
-          opBuilder.setAttr("dilations", dilationsArray);
-        }
-        if (opts.paddingList != null) {
-          long[] paddingListArray = new long[opts.paddingList.size()];
-          for (int i = 0; i < paddingListArray.length; ++i) {
-            paddingListArray[i] = opts.paddingList.get(i);
-          }
-          opBuilder.setAttr("padding_list", paddingListArray);
-        }
-      }
-    }
-    return new QuantizedConv2DAndRelu<V>(opBuilder.build());
-  }
-  
-  /**
-   * @param dilations 
-   */
-  public static Options dilations(List<Long> dilations) {
-    return new Options().dilations(dilations);
-  }
-  
-  /**
-   * @param paddingList 
-   */
-  public static Options paddingList(List<Long> paddingList) {
-    return new Options().paddingList(paddingList);
-  }
-  
-  /**
-   */
-  public Output<V> output() {
-    return output;
-  }
-  
-  /**
-   */
-  public Output<TFloat32> minOutput() {
-    return minOutput;
-  }
-  
-  /**
-   */
-  public Output<TFloat32> maxOutput() {
-    return maxOutput;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "QuantizedConv2DAndRelu";
-  
-  private Output<V> output;
-  private Output<TFloat32> minOutput;
-  private Output<TFloat32> maxOutput;
-  
-  private QuantizedConv2DAndRelu(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-    minOutput = operation.output(outputIdx++);
-    maxOutput = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DAndReluAndRequantize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DAndReluAndRequantize.java
deleted file mode 100644
index e1909629558..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DAndReluAndRequantize.java
+++ /dev/null
@@ -1,169 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.nn;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.family.TType;
-
-/**
- * @param <V> data type for {@code output()} output
- */
-public final class QuantizedConv2DAndReluAndRequantize<V extends TType> extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.nn.QuantizedConv2DAndReluAndRequantize}
-   */
-  public static class Options {
-    
-    /**
-     * @param dilations 
-     */
-    public Options dilations(List<Long> dilations) {
-      this.dilations = dilations;
-      return this;
-    }
-    
-    /**
-     * @param paddingList 
-     */
-    public Options paddingList(List<Long> paddingList) {
-      this.paddingList = paddingList;
-      return this;
-    }
-    
-    private List<Long> dilations;
-    private List<Long> paddingList;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new QuantizedConv2DAndReluAndRequantize operation.
-   * 
-   * @param scope current scope
-   * @param input 
-   * @param filter 
-   * @param minInput 
-   * @param maxInput 
-   * @param minFilter 
-   * @param maxFilter 
-   * @param minFreezedOutput 
-   * @param maxFreezedOutput 
-   * @param outType 
-   * @param strides 
-   * @param padding 
-   * @param options carries optional attributes values
-   * @return a new instance of QuantizedConv2DAndReluAndRequantize
-   */
-  @Endpoint(describeByClass = true)
-  public static <V extends TType, T extends TType, U extends TType> QuantizedConv2DAndReluAndRequantize<V> create(Scope scope, Operand<T> input, Operand<U> filter, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, Operand<TFloat32> minFilter, Operand<TFloat32> maxFilter, Operand<TFloat32> minFreezedOutput, Operand<TFloat32> maxFreezedOutput, Class<V> outType, List<Long> strides, String padding, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("QuantizedConv2DAndReluAndRequantize", scope.makeOpName("QuantizedConv2DAndReluAndRequantize"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder.addInput(filter.asOutput(scope));
-    opBuilder.addInput(minInput.asOutput(scope));
-    opBuilder.addInput(maxInput.asOutput(scope));
-    opBuilder.addInput(minFilter.asOutput(scope));
-    opBuilder.addInput(maxFilter.asOutput(scope));
-    opBuilder.addInput(minFreezedOutput.asOutput(scope));
-    opBuilder.addInput(maxFreezedOutput.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("out_type", outType);
-    long[] stridesArray = new long[strides.size()];
-    for (int i = 0; i < stridesArray.length; ++i) {
-      stridesArray[i] = strides.get(i);
-    }
-    opBuilder.setAttr("strides", stridesArray);
-    opBuilder.setAttr("padding", padding);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.dilations != null) {
-          long[] dilationsArray = new long[opts.dilations.size()];
-          for (int i = 0; i < dilationsArray.length; ++i) {
-            dilationsArray[i] = opts.dilations.get(i);
-          }
-          opBuilder.setAttr("dilations", dilationsArray);
-        }
-        if (opts.paddingList != null) {
-          long[] paddingListArray = new long[opts.paddingList.size()];
-          for (int i = 0; i < paddingListArray.length; ++i) {
-            paddingListArray[i] = opts.paddingList.get(i);
-          }
-          opBuilder.setAttr("padding_list", paddingListArray);
-        }
-      }
-    }
-    return new QuantizedConv2DAndReluAndRequantize<V>(opBuilder.build());
-  }
-  
-  /**
-   * @param dilations 
-   */
-  public static Options dilations(List<Long> dilations) {
-    return new Options().dilations(dilations);
-  }
-  
-  /**
-   * @param paddingList 
-   */
-  public static Options paddingList(List<Long> paddingList) {
-    return new Options().paddingList(paddingList);
-  }
-  
-  /**
-   */
-  public Output<V> output() {
-    return output;
-  }
-  
-  /**
-   */
-  public Output<TFloat32> minOutput() {
-    return minOutput;
-  }
-  
-  /**
-   */
-  public Output<TFloat32> maxOutput() {
-    return maxOutput;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "QuantizedConv2DAndReluAndRequantize";
-  
-  private Output<V> output;
-  private Output<TFloat32> minOutput;
-  private Output<TFloat32> maxOutput;
-  
-  private QuantizedConv2DAndReluAndRequantize(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-    minOutput = operation.output(outputIdx++);
-    maxOutput = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DAndRequantize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DAndRequantize.java
deleted file mode 100644
index 0964c0e98fa..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DAndRequantize.java
+++ /dev/null
@@ -1,169 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.nn;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.family.TType;
-
-/**
- * @param <V> data type for {@code output()} output
- */
-public final class QuantizedConv2DAndRequantize<V extends TType> extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.nn.QuantizedConv2DAndRequantize}
-   */
-  public static class Options {
-    
-    /**
-     * @param dilations 
-     */
-    public Options dilations(List<Long> dilations) {
-      this.dilations = dilations;
-      return this;
-    }
-    
-    /**
-     * @param paddingList 
-     */
-    public Options paddingList(List<Long> paddingList) {
-      this.paddingList = paddingList;
-      return this;
-    }
-    
-    private List<Long> dilations;
-    private List<Long> paddingList;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new QuantizedConv2DAndRequantize operation.
-   * 
-   * @param scope current scope
-   * @param input 
-   * @param filter 
-   * @param minInput 
-   * @param maxInput 
-   * @param minFilter 
-   * @param maxFilter 
-   * @param minFreezedOutput 
-   * @param maxFreezedOutput 
-   * @param outType 
-   * @param strides 
-   * @param padding 
-   * @param options carries optional attributes values
-   * @return a new instance of QuantizedConv2DAndRequantize
-   */
-  @Endpoint(describeByClass = true)
-  public static <V extends TType, T extends TType, U extends TType> QuantizedConv2DAndRequantize<V> create(Scope scope, Operand<T> input, Operand<U> filter, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, Operand<TFloat32> minFilter, Operand<TFloat32> maxFilter, Operand<TFloat32> minFreezedOutput, Operand<TFloat32> maxFreezedOutput, Class<V> outType, List<Long> strides, String padding, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("QuantizedConv2DAndRequantize", scope.makeOpName("QuantizedConv2DAndRequantize"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder.addInput(filter.asOutput(scope));
-    opBuilder.addInput(minInput.asOutput(scope));
-    opBuilder.addInput(maxInput.asOutput(scope));
-    opBuilder.addInput(minFilter.asOutput(scope));
-    opBuilder.addInput(maxFilter.asOutput(scope));
-    opBuilder.addInput(minFreezedOutput.asOutput(scope));
-    opBuilder.addInput(maxFreezedOutput.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("out_type", outType);
-    long[] stridesArray = new long[strides.size()];
-    for (int i = 0; i < stridesArray.length; ++i) {
-      stridesArray[i] = strides.get(i);
-    }
-    opBuilder.setAttr("strides", stridesArray);
-    opBuilder.setAttr("padding", padding);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.dilations != null) {
-          long[] dilationsArray = new long[opts.dilations.size()];
-          for (int i = 0; i < dilationsArray.length; ++i) {
-            dilationsArray[i] = opts.dilations.get(i);
-          }
-          opBuilder.setAttr("dilations", dilationsArray);
-        }
-        if (opts.paddingList != null) {
-          long[] paddingListArray = new long[opts.paddingList.size()];
-          for (int i = 0; i < paddingListArray.length; ++i) {
-            paddingListArray[i] = opts.paddingList.get(i);
-          }
-          opBuilder.setAttr("padding_list", paddingListArray);
-        }
-      }
-    }
-    return new QuantizedConv2DAndRequantize<V>(opBuilder.build());
-  }
-  
-  /**
-   * @param dilations 
-   */
-  public static Options dilations(List<Long> dilations) {
-    return new Options().dilations(dilations);
-  }
-  
-  /**
-   * @param paddingList 
-   */
-  public static Options paddingList(List<Long> paddingList) {
-    return new Options().paddingList(paddingList);
-  }
-  
-  /**
-   */
-  public Output<V> output() {
-    return output;
-  }
-  
-  /**
-   */
-  public Output<TFloat32> minOutput() {
-    return minOutput;
-  }
-  
-  /**
-   */
-  public Output<TFloat32> maxOutput() {
-    return maxOutput;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "QuantizedConv2DAndRequantize";
-  
-  private Output<V> output;
-  private Output<TFloat32> minOutput;
-  private Output<TFloat32> maxOutput;
-  
-  private QuantizedConv2DAndRequantize(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-    minOutput = operation.output(outputIdx++);
-    maxOutput = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DPerChannel.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DPerChannel.java
deleted file mode 100644
index 51799cff652..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DPerChannel.java
+++ /dev/null
@@ -1,147 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.nn;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.family.TType;
-
-/**
- * Computes QuantizedConv2D per channel.
- * 
- * @param <V> data type for {@code output()} output
- */
-public final class QuantizedConv2DPerChannel<V extends TType> extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.nn.QuantizedConv2DPerChannel}
-   */
-  public static class Options {
-    
-    /**
-     * @param dilations list of dilation values.
-     */
-    public Options dilations(List<Long> dilations) {
-      this.dilations = dilations;
-      return this;
-    }
-    
-    private List<Long> dilations;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new QuantizedConv2DPerChannel operation.
-   * 
-   * @param scope current scope
-   * @param input The original input tensor.
-   * @param filter The original filter tensor.
-   * @param minInput The minimum value of the input tensor
-   * @param maxInput The maximum value of the input tensor.
-   * @param minFilter The minimum value of the filter tensor.
-   * @param maxFilter The maximum value of the filter tensor.
-   * @param outType The quantized type of output tensor that needs to be converted.
-   * @param strides list of stride values.
-   * @param padding 
-   * @param options carries optional attributes values
-   * @return a new instance of QuantizedConv2DPerChannel
-   */
-  @Endpoint(describeByClass = true)
-  public static <V extends TType, T extends TType, U extends TType> QuantizedConv2DPerChannel<V> create(Scope scope, Operand<T> input, Operand<U> filter, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, Operand<TFloat32> minFilter, Operand<TFloat32> maxFilter, Class<V> outType, List<Long> strides, String padding, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("QuantizedConv2DPerChannel", scope.makeOpName("QuantizedConv2DPerChannel"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder.addInput(filter.asOutput(scope));
-    opBuilder.addInput(minInput.asOutput(scope));
-    opBuilder.addInput(maxInput.asOutput(scope));
-    opBuilder.addInput(minFilter.asOutput(scope));
-    opBuilder.addInput(maxFilter.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("out_type", outType);
-    long[] stridesArray = new long[strides.size()];
-    for (int i = 0; i < stridesArray.length; ++i) {
-      stridesArray[i] = strides.get(i);
-    }
-    opBuilder.setAttr("strides", stridesArray);
-    opBuilder.setAttr("padding", padding);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.dilations != null) {
-          long[] dilationsArray = new long[opts.dilations.size()];
-          for (int i = 0; i < dilationsArray.length; ++i) {
-            dilationsArray[i] = opts.dilations.get(i);
-          }
-          opBuilder.setAttr("dilations", dilationsArray);
-        }
-      }
-    }
-    return new QuantizedConv2DPerChannel<V>(opBuilder.build());
-  }
-  
-  /**
-   * @param dilations list of dilation values.
-   */
-  public static Options dilations(List<Long> dilations) {
-    return new Options().dilations(dilations);
-  }
-  
-  /**
-   * The output tensor.
-   */
-  public Output<V> output() {
-    return output;
-  }
-  
-  /**
-   * The minimum value of the final output tensor.
-   */
-  public Output<TFloat32> minOutput() {
-    return minOutput;
-  }
-  
-  /**
-   * The maximum value of the final output tensor.
-   */
-  public Output<TFloat32> maxOutput() {
-    return maxOutput;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "QuantizedConv2DPerChannel";
-  
-  private Output<V> output;
-  private Output<TFloat32> minOutput;
-  private Output<TFloat32> maxOutput;
-  
-  private QuantizedConv2DPerChannel(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-    minOutput = operation.output(outputIdx++);
-    maxOutput = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DWithBias.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DWithBias.java
deleted file mode 100644
index 561c1854190..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DWithBias.java
+++ /dev/null
@@ -1,167 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.nn;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.family.TType;
-
-/**
- * @param <V> data type for {@code output()} output
- */
-public final class QuantizedConv2DWithBias<V extends TType> extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.nn.QuantizedConv2DWithBias}
-   */
-  public static class Options {
-    
-    /**
-     * @param dilations 
-     */
-    public Options dilations(List<Long> dilations) {
-      this.dilations = dilations;
-      return this;
-    }
-    
-    /**
-     * @param paddingList 
-     */
-    public Options paddingList(List<Long> paddingList) {
-      this.paddingList = paddingList;
-      return this;
-    }
-    
-    private List<Long> dilations;
-    private List<Long> paddingList;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new QuantizedConv2DWithBias operation.
-   * 
-   * @param scope current scope
-   * @param input 
-   * @param filter 
-   * @param bias 
-   * @param minInput 
-   * @param maxInput 
-   * @param minFilter 
-   * @param maxFilter 
-   * @param outType 
-   * @param strides 
-   * @param padding 
-   * @param options carries optional attributes values
-   * @return a new instance of QuantizedConv2DWithBias
-   */
-  @Endpoint(describeByClass = true)
-  public static <V extends TType, T extends TType, U extends TType> QuantizedConv2DWithBias<V> create(Scope scope, Operand<T> input, Operand<U> filter, Operand<TFloat32> bias, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, Operand<TFloat32> minFilter, Operand<TFloat32> maxFilter, Class<V> outType, List<Long> strides, String padding, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("QuantizedConv2DWithBias", scope.makeOpName("QuantizedConv2DWithBias"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder.addInput(filter.asOutput(scope));
-    opBuilder.addInput(bias.asOutput(scope));
-    opBuilder.addInput(minInput.asOutput(scope));
-    opBuilder.addInput(maxInput.asOutput(scope));
-    opBuilder.addInput(minFilter.asOutput(scope));
-    opBuilder.addInput(maxFilter.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("out_type", outType);
-    long[] stridesArray = new long[strides.size()];
-    for (int i = 0; i < stridesArray.length; ++i) {
-      stridesArray[i] = strides.get(i);
-    }
-    opBuilder.setAttr("strides", stridesArray);
-    opBuilder.setAttr("padding", padding);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.dilations != null) {
-          long[] dilationsArray = new long[opts.dilations.size()];
-          for (int i = 0; i < dilationsArray.length; ++i) {
-            dilationsArray[i] = opts.dilations.get(i);
-          }
-          opBuilder.setAttr("dilations", dilationsArray);
-        }
-        if (opts.paddingList != null) {
-          long[] paddingListArray = new long[opts.paddingList.size()];
-          for (int i = 0; i < paddingListArray.length; ++i) {
-            paddingListArray[i] = opts.paddingList.get(i);
-          }
-          opBuilder.setAttr("padding_list", paddingListArray);
-        }
-      }
-    }
-    return new QuantizedConv2DWithBias<V>(opBuilder.build());
-  }
-  
-  /**
-   * @param dilations 
-   */
-  public static Options dilations(List<Long> dilations) {
-    return new Options().dilations(dilations);
-  }
-  
-  /**
-   * @param paddingList 
-   */
-  public static Options paddingList(List<Long> paddingList) {
-    return new Options().paddingList(paddingList);
-  }
-  
-  /**
-   */
-  public Output<V> output() {
-    return output;
-  }
-  
-  /**
-   */
-  public Output<TFloat32> minOutput() {
-    return minOutput;
-  }
-  
-  /**
-   */
-  public Output<TFloat32> maxOutput() {
-    return maxOutput;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "QuantizedConv2DWithBias";
-  
-  private Output<V> output;
-  private Output<TFloat32> minOutput;
-  private Output<TFloat32> maxOutput;
-  
-  private QuantizedConv2DWithBias(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-    minOutput = operation.output(outputIdx++);
-    maxOutput = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DWithBiasAndRelu.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DWithBiasAndRelu.java
deleted file mode 100644
index 6f39f8d6291..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DWithBiasAndRelu.java
+++ /dev/null
@@ -1,167 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.nn;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.family.TType;
-
-/**
- * @param <V> data type for {@code output()} output
- */
-public final class QuantizedConv2DWithBiasAndRelu<V extends TType> extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.nn.QuantizedConv2DWithBiasAndRelu}
-   */
-  public static class Options {
-    
-    /**
-     * @param dilations 
-     */
-    public Options dilations(List<Long> dilations) {
-      this.dilations = dilations;
-      return this;
-    }
-    
-    /**
-     * @param paddingList 
-     */
-    public Options paddingList(List<Long> paddingList) {
-      this.paddingList = paddingList;
-      return this;
-    }
-    
-    private List<Long> dilations;
-    private List<Long> paddingList;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new QuantizedConv2DWithBiasAndRelu operation.
-   * 
-   * @param scope current scope
-   * @param input 
-   * @param filter 
-   * @param bias 
-   * @param minInput 
-   * @param maxInput 
-   * @param minFilter 
-   * @param maxFilter 
-   * @param outType 
-   * @param strides 
-   * @param padding 
-   * @param options carries optional attributes values
-   * @return a new instance of QuantizedConv2DWithBiasAndRelu
-   */
-  @Endpoint(describeByClass = true)
-  public static <V extends TType, T extends TType, U extends TType> QuantizedConv2DWithBiasAndRelu<V> create(Scope scope, Operand<T> input, Operand<U> filter, Operand<TFloat32> bias, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, Operand<TFloat32> minFilter, Operand<TFloat32> maxFilter, Class<V> outType, List<Long> strides, String padding, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("QuantizedConv2DWithBiasAndRelu", scope.makeOpName("QuantizedConv2DWithBiasAndRelu"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder.addInput(filter.asOutput(scope));
-    opBuilder.addInput(bias.asOutput(scope));
-    opBuilder.addInput(minInput.asOutput(scope));
-    opBuilder.addInput(maxInput.asOutput(scope));
-    opBuilder.addInput(minFilter.asOutput(scope));
-    opBuilder.addInput(maxFilter.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("out_type", outType);
-    long[] stridesArray = new long[strides.size()];
-    for (int i = 0; i < stridesArray.length; ++i) {
-      stridesArray[i] = strides.get(i);
-    }
-    opBuilder.setAttr("strides", stridesArray);
-    opBuilder.setAttr("padding", padding);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.dilations != null) {
-          long[] dilationsArray = new long[opts.dilations.size()];
-          for (int i = 0; i < dilationsArray.length; ++i) {
-            dilationsArray[i] = opts.dilations.get(i);
-          }
-          opBuilder.setAttr("dilations", dilationsArray);
-        }
-        if (opts.paddingList != null) {
-          long[] paddingListArray = new long[opts.paddingList.size()];
-          for (int i = 0; i < paddingListArray.length; ++i) {
-            paddingListArray[i] = opts.paddingList.get(i);
-          }
-          opBuilder.setAttr("padding_list", paddingListArray);
-        }
-      }
-    }
-    return new QuantizedConv2DWithBiasAndRelu<V>(opBuilder.build());
-  }
-  
-  /**
-   * @param dilations 
-   */
-  public static Options dilations(List<Long> dilations) {
-    return new Options().dilations(dilations);
-  }
-  
-  /**
-   * @param paddingList 
-   */
-  public static Options paddingList(List<Long> paddingList) {
-    return new Options().paddingList(paddingList);
-  }
-  
-  /**
-   */
-  public Output<V> output() {
-    return output;
-  }
-  
-  /**
-   */
-  public Output<TFloat32> minOutput() {
-    return minOutput;
-  }
-  
-  /**
-   */
-  public Output<TFloat32> maxOutput() {
-    return maxOutput;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "QuantizedConv2DWithBiasAndRelu";
-  
-  private Output<V> output;
-  private Output<TFloat32> minOutput;
-  private Output<TFloat32> maxOutput;
-  
-  private QuantizedConv2DWithBiasAndRelu(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-    minOutput = operation.output(outputIdx++);
-    maxOutput = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DWithBiasAndReluAndRequantize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DWithBiasAndReluAndRequantize.java
deleted file mode 100644
index 3b95469edb7..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DWithBiasAndReluAndRequantize.java
+++ /dev/null
@@ -1,171 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.nn;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.family.TType;
-
-/**
- * @param <W> data type for {@code output()} output
- */
-public final class QuantizedConv2DWithBiasAndReluAndRequantize<W extends TType> extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.nn.QuantizedConv2DWithBiasAndReluAndRequantize}
-   */
-  public static class Options {
-    
-    /**
-     * @param dilations 
-     */
-    public Options dilations(List<Long> dilations) {
-      this.dilations = dilations;
-      return this;
-    }
-    
-    /**
-     * @param paddingList 
-     */
-    public Options paddingList(List<Long> paddingList) {
-      this.paddingList = paddingList;
-      return this;
-    }
-    
-    private List<Long> dilations;
-    private List<Long> paddingList;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new QuantizedConv2DWithBiasAndReluAndRequantize operation.
-   * 
-   * @param scope current scope
-   * @param input 
-   * @param filter 
-   * @param bias 
-   * @param minInput 
-   * @param maxInput 
-   * @param minFilter 
-   * @param maxFilter 
-   * @param minFreezedOutput 
-   * @param maxFreezedOutput 
-   * @param outType 
-   * @param strides 
-   * @param padding 
-   * @param options carries optional attributes values
-   * @return a new instance of QuantizedConv2DWithBiasAndReluAndRequantize
-   */
-  @Endpoint(describeByClass = true)
-  public static <W extends TType, T extends TType, U extends TType, V extends TType> QuantizedConv2DWithBiasAndReluAndRequantize<W> create(Scope scope, Operand<T> input, Operand<U> filter, Operand<V> bias, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, Operand<TFloat32> minFilter, Operand<TFloat32> maxFilter, Operand<TFloat32> minFreezedOutput, Operand<TFloat32> maxFreezedOutput, Class<W> outType, List<Long> strides, String padding, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("QuantizedConv2DWithBiasAndReluAndRequantize", scope.makeOpName("QuantizedConv2DWithBiasAndReluAndRequantize"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder.addInput(filter.asOutput(scope));
-    opBuilder.addInput(bias.asOutput(scope));
-    opBuilder.addInput(minInput.asOutput(scope));
-    opBuilder.addInput(maxInput.asOutput(scope));
-    opBuilder.addInput(minFilter.asOutput(scope));
-    opBuilder.addInput(maxFilter.asOutput(scope));
-    opBuilder.addInput(minFreezedOutput.asOutput(scope));
-    opBuilder.addInput(maxFreezedOutput.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("out_type", outType);
-    long[] stridesArray = new long[strides.size()];
-    for (int i = 0; i < stridesArray.length; ++i) {
-      stridesArray[i] = strides.get(i);
-    }
-    opBuilder.setAttr("strides", stridesArray);
-    opBuilder.setAttr("padding", padding);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.dilations != null) {
-          long[] dilationsArray = new long[opts.dilations.size()];
-          for (int i = 0; i < dilationsArray.length; ++i) {
-            dilationsArray[i] = opts.dilations.get(i);
-          }
-          opBuilder.setAttr("dilations", dilationsArray);
-        }
-        if (opts.paddingList != null) {
-          long[] paddingListArray = new long[opts.paddingList.size()];
-          for (int i = 0; i < paddingListArray.length; ++i) {
-            paddingListArray[i] = opts.paddingList.get(i);
-          }
-          opBuilder.setAttr("padding_list", paddingListArray);
-        }
-      }
-    }
-    return new QuantizedConv2DWithBiasAndReluAndRequantize<W>(opBuilder.build());
-  }
-  
-  /**
-   * @param dilations 
-   */
-  public static Options dilations(List<Long> dilations) {
-    return new Options().dilations(dilations);
-  }
-  
-  /**
-   * @param paddingList 
-   */
-  public static Options paddingList(List<Long> paddingList) {
-    return new Options().paddingList(paddingList);
-  }
-  
-  /**
-   */
-  public Output<W> output() {
-    return output;
-  }
-  
-  /**
-   */
-  public Output<TFloat32> minOutput() {
-    return minOutput;
-  }
-  
-  /**
-   */
-  public Output<TFloat32> maxOutput() {
-    return maxOutput;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "QuantizedConv2DWithBiasAndReluAndRequantize";
-  
-  private Output<W> output;
-  private Output<TFloat32> minOutput;
-  private Output<TFloat32> maxOutput;
-  
-  private QuantizedConv2DWithBiasAndReluAndRequantize(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-    minOutput = operation.output(outputIdx++);
-    maxOutput = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DWithBiasAndRequantize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DWithBiasAndRequantize.java
deleted file mode 100644
index ac5d066e2d7..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DWithBiasAndRequantize.java
+++ /dev/null
@@ -1,171 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.nn;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.family.TType;
-
-/**
- * @param <W> data type for {@code output()} output
- */
-public final class QuantizedConv2DWithBiasAndRequantize<W extends TType> extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.nn.QuantizedConv2DWithBiasAndRequantize}
-   */
-  public static class Options {
-    
-    /**
-     * @param dilations 
-     */
-    public Options dilations(List<Long> dilations) {
-      this.dilations = dilations;
-      return this;
-    }
-    
-    /**
-     * @param paddingList 
-     */
-    public Options paddingList(List<Long> paddingList) {
-      this.paddingList = paddingList;
-      return this;
-    }
-    
-    private List<Long> dilations;
-    private List<Long> paddingList;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new QuantizedConv2DWithBiasAndRequantize operation.
-   * 
-   * @param scope current scope
-   * @param input 
-   * @param filter 
-   * @param bias 
-   * @param minInput 
-   * @param maxInput 
-   * @param minFilter 
-   * @param maxFilter 
-   * @param minFreezedOutput 
-   * @param maxFreezedOutput 
-   * @param outType 
-   * @param strides 
-   * @param padding 
-   * @param options carries optional attributes values
-   * @return a new instance of QuantizedConv2DWithBiasAndRequantize
-   */
-  @Endpoint(describeByClass = true)
-  public static <W extends TType, T extends TType, U extends TType, V extends TType> QuantizedConv2DWithBiasAndRequantize<W> create(Scope scope, Operand<T> input, Operand<U> filter, Operand<V> bias, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, Operand<TFloat32> minFilter, Operand<TFloat32> maxFilter, Operand<TFloat32> minFreezedOutput, Operand<TFloat32> maxFreezedOutput, Class<W> outType, List<Long> strides, String padding, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("QuantizedConv2DWithBiasAndRequantize", scope.makeOpName("QuantizedConv2DWithBiasAndRequantize"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder.addInput(filter.asOutput(scope));
-    opBuilder.addInput(bias.asOutput(scope));
-    opBuilder.addInput(minInput.asOutput(scope));
-    opBuilder.addInput(maxInput.asOutput(scope));
-    opBuilder.addInput(minFilter.asOutput(scope));
-    opBuilder.addInput(maxFilter.asOutput(scope));
-    opBuilder.addInput(minFreezedOutput.asOutput(scope));
-    opBuilder.addInput(maxFreezedOutput.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("out_type", outType);
-    long[] stridesArray = new long[strides.size()];
-    for (int i = 0; i < stridesArray.length; ++i) {
-      stridesArray[i] = strides.get(i);
-    }
-    opBuilder.setAttr("strides", stridesArray);
-    opBuilder.setAttr("padding", padding);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.dilations != null) {
-          long[] dilationsArray = new long[opts.dilations.size()];
-          for (int i = 0; i < dilationsArray.length; ++i) {
-            dilationsArray[i] = opts.dilations.get(i);
-          }
-          opBuilder.setAttr("dilations", dilationsArray);
-        }
-        if (opts.paddingList != null) {
-          long[] paddingListArray = new long[opts.paddingList.size()];
-          for (int i = 0; i < paddingListArray.length; ++i) {
-            paddingListArray[i] = opts.paddingList.get(i);
-          }
-          opBuilder.setAttr("padding_list", paddingListArray);
-        }
-      }
-    }
-    return new QuantizedConv2DWithBiasAndRequantize<W>(opBuilder.build());
-  }
-  
-  /**
-   * @param dilations 
-   */
-  public static Options dilations(List<Long> dilations) {
-    return new Options().dilations(dilations);
-  }
-  
-  /**
-   * @param paddingList 
-   */
-  public static Options paddingList(List<Long> paddingList) {
-    return new Options().paddingList(paddingList);
-  }
-  
-  /**
-   */
-  public Output<W> output() {
-    return output;
-  }
-  
-  /**
-   */
-  public Output<TFloat32> minOutput() {
-    return minOutput;
-  }
-  
-  /**
-   */
-  public Output<TFloat32> maxOutput() {
-    return maxOutput;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "QuantizedConv2DWithBiasAndRequantize";
-  
-  private Output<W> output;
-  private Output<TFloat32> minOutput;
-  private Output<TFloat32> maxOutput;
-  
-  private QuantizedConv2DWithBiasAndRequantize(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-    minOutput = operation.output(outputIdx++);
-    maxOutput = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DWithBiasSignedSumAndReluAndRequantize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DWithBiasSignedSumAndReluAndRequantize.java
deleted file mode 100644
index 42aef6a8258..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DWithBiasSignedSumAndReluAndRequantize.java
+++ /dev/null
@@ -1,177 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.nn;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.family.TType;
-
-/**
- * @param <X> data type for {@code output()} output
- */
-public final class QuantizedConv2DWithBiasSignedSumAndReluAndRequantize<X extends TType> extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.nn.QuantizedConv2DWithBiasSignedSumAndReluAndRequantize}
-   */
-  public static class Options {
-    
-    /**
-     * @param dilations 
-     */
-    public Options dilations(List<Long> dilations) {
-      this.dilations = dilations;
-      return this;
-    }
-    
-    /**
-     * @param paddingList 
-     */
-    public Options paddingList(List<Long> paddingList) {
-      this.paddingList = paddingList;
-      return this;
-    }
-    
-    private List<Long> dilations;
-    private List<Long> paddingList;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new QuantizedConv2DWithBiasSignedSumAndReluAndRequantize operation.
-   * 
-   * @param scope current scope
-   * @param input 
-   * @param filter 
-   * @param bias 
-   * @param minInput 
-   * @param maxInput 
-   * @param minFilter 
-   * @param maxFilter 
-   * @param minFreezedOutput 
-   * @param maxFreezedOutput 
-   * @param summand 
-   * @param minSummand 
-   * @param maxSummand 
-   * @param outType 
-   * @param strides 
-   * @param padding 
-   * @param options carries optional attributes values
-   * @return a new instance of QuantizedConv2DWithBiasSignedSumAndReluAndRequantize
-   */
-  @Endpoint(describeByClass = true)
-  public static <X extends TType, T extends TType, U extends TType, V extends TType, W extends TType> QuantizedConv2DWithBiasSignedSumAndReluAndRequantize<X> create(Scope scope, Operand<T> input, Operand<U> filter, Operand<V> bias, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, Operand<TFloat32> minFilter, Operand<TFloat32> maxFilter, Operand<TFloat32> minFreezedOutput, Operand<TFloat32> maxFreezedOutput, Operand<W> summand, Operand<TFloat32> minSummand, Operand<TFloat32> maxSummand, Class<X> outType, List<Long> strides, String padding, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("QuantizedConv2DWithBiasSignedSumAndReluAndRequantize", scope.makeOpName("QuantizedConv2DWithBiasSignedSumAndReluAndRequantize"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder.addInput(filter.asOutput(scope));
-    opBuilder.addInput(bias.asOutput(scope));
-    opBuilder.addInput(minInput.asOutput(scope));
-    opBuilder.addInput(maxInput.asOutput(scope));
-    opBuilder.addInput(minFilter.asOutput(scope));
-    opBuilder.addInput(maxFilter.asOutput(scope));
-    opBuilder.addInput(minFreezedOutput.asOutput(scope));
-    opBuilder.addInput(maxFreezedOutput.asOutput(scope));
-    opBuilder.addInput(summand.asOutput(scope));
-    opBuilder.addInput(minSummand.asOutput(scope));
-    opBuilder.addInput(maxSummand.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("out_type", outType);
-    long[] stridesArray = new long[strides.size()];
-    for (int i = 0; i < stridesArray.length; ++i) {
-      stridesArray[i] = strides.get(i);
-    }
-    opBuilder.setAttr("strides", stridesArray);
-    opBuilder.setAttr("padding", padding);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.dilations != null) {
-          long[] dilationsArray = new long[opts.dilations.size()];
-          for (int i = 0; i < dilationsArray.length; ++i) {
-            dilationsArray[i] = opts.dilations.get(i);
-          }
-          opBuilder.setAttr("dilations", dilationsArray);
-        }
-        if (opts.paddingList != null) {
-          long[] paddingListArray = new long[opts.paddingList.size()];
-          for (int i = 0; i < paddingListArray.length; ++i) {
-            paddingListArray[i] = opts.paddingList.get(i);
-          }
-          opBuilder.setAttr("padding_list", paddingListArray);
-        }
-      }
-    }
-    return new QuantizedConv2DWithBiasSignedSumAndReluAndRequantize<X>(opBuilder.build());
-  }
-  
-  /**
-   * @param dilations 
-   */
-  public static Options dilations(List<Long> dilations) {
-    return new Options().dilations(dilations);
-  }
-  
-  /**
-   * @param paddingList 
-   */
-  public static Options paddingList(List<Long> paddingList) {
-    return new Options().paddingList(paddingList);
-  }
-  
-  /**
-   */
-  public Output<X> output() {
-    return output;
-  }
-  
-  /**
-   */
-  public Output<TFloat32> minOutput() {
-    return minOutput;
-  }
-  
-  /**
-   */
-  public Output<TFloat32> maxOutput() {
-    return maxOutput;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "QuantizedConv2DWithBiasSignedSumAndReluAndRequantize";
-  
-  private Output<X> output;
-  private Output<TFloat32> minOutput;
-  private Output<TFloat32> maxOutput;
-  
-  private QuantizedConv2DWithBiasSignedSumAndReluAndRequantize(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-    minOutput = operation.output(outputIdx++);
-    maxOutput = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DWithBiasSumAndRelu.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DWithBiasSumAndRelu.java
deleted file mode 100644
index d2e9508074c..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DWithBiasSumAndRelu.java
+++ /dev/null
@@ -1,169 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.nn;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.family.TType;
-
-/**
- * @param <V> data type for {@code output()} output
- */
-public final class QuantizedConv2DWithBiasSumAndRelu<V extends TType> extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.nn.QuantizedConv2DWithBiasSumAndRelu}
-   */
-  public static class Options {
-    
-    /**
-     * @param dilations 
-     */
-    public Options dilations(List<Long> dilations) {
-      this.dilations = dilations;
-      return this;
-    }
-    
-    /**
-     * @param paddingList 
-     */
-    public Options paddingList(List<Long> paddingList) {
-      this.paddingList = paddingList;
-      return this;
-    }
-    
-    private List<Long> dilations;
-    private List<Long> paddingList;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new QuantizedConv2DWithBiasSumAndRelu operation.
-   * 
-   * @param scope current scope
-   * @param input 
-   * @param filter 
-   * @param bias 
-   * @param minInput 
-   * @param maxInput 
-   * @param minFilter 
-   * @param maxFilter 
-   * @param summand 
-   * @param outType 
-   * @param strides 
-   * @param padding 
-   * @param options carries optional attributes values
-   * @return a new instance of QuantizedConv2DWithBiasSumAndRelu
-   */
-  @Endpoint(describeByClass = true)
-  public static <V extends TType, T extends TType, U extends TType> QuantizedConv2DWithBiasSumAndRelu<V> create(Scope scope, Operand<T> input, Operand<U> filter, Operand<TFloat32> bias, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, Operand<TFloat32> minFilter, Operand<TFloat32> maxFilter, Operand<TFloat32> summand, Class<V> outType, List<Long> strides, String padding, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("QuantizedConv2DWithBiasSumAndRelu", scope.makeOpName("QuantizedConv2DWithBiasSumAndRelu"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder.addInput(filter.asOutput(scope));
-    opBuilder.addInput(bias.asOutput(scope));
-    opBuilder.addInput(minInput.asOutput(scope));
-    opBuilder.addInput(maxInput.asOutput(scope));
-    opBuilder.addInput(minFilter.asOutput(scope));
-    opBuilder.addInput(maxFilter.asOutput(scope));
-    opBuilder.addInput(summand.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("out_type", outType);
-    long[] stridesArray = new long[strides.size()];
-    for (int i = 0; i < stridesArray.length; ++i) {
-      stridesArray[i] = strides.get(i);
-    }
-    opBuilder.setAttr("strides", stridesArray);
-    opBuilder.setAttr("padding", padding);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.dilations != null) {
-          long[] dilationsArray = new long[opts.dilations.size()];
-          for (int i = 0; i < dilationsArray.length; ++i) {
-            dilationsArray[i] = opts.dilations.get(i);
-          }
-          opBuilder.setAttr("dilations", dilationsArray);
-        }
-        if (opts.paddingList != null) {
-          long[] paddingListArray = new long[opts.paddingList.size()];
-          for (int i = 0; i < paddingListArray.length; ++i) {
-            paddingListArray[i] = opts.paddingList.get(i);
-          }
-          opBuilder.setAttr("padding_list", paddingListArray);
-        }
-      }
-    }
-    return new QuantizedConv2DWithBiasSumAndRelu<V>(opBuilder.build());
-  }
-  
-  /**
-   * @param dilations 
-   */
-  public static Options dilations(List<Long> dilations) {
-    return new Options().dilations(dilations);
-  }
-  
-  /**
-   * @param paddingList 
-   */
-  public static Options paddingList(List<Long> paddingList) {
-    return new Options().paddingList(paddingList);
-  }
-  
-  /**
-   */
-  public Output<V> output() {
-    return output;
-  }
-  
-  /**
-   */
-  public Output<TFloat32> minOutput() {
-    return minOutput;
-  }
-  
-  /**
-   */
-  public Output<TFloat32> maxOutput() {
-    return maxOutput;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "QuantizedConv2DWithBiasSumAndRelu";
-  
-  private Output<V> output;
-  private Output<TFloat32> minOutput;
-  private Output<TFloat32> maxOutput;
-  
-  private QuantizedConv2DWithBiasSumAndRelu(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-    minOutput = operation.output(outputIdx++);
-    maxOutput = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DWithBiasSumAndReluAndRequantize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DWithBiasSumAndReluAndRequantize.java
deleted file mode 100644
index cfd4aff55cd..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DWithBiasSumAndReluAndRequantize.java
+++ /dev/null
@@ -1,177 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.nn;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.family.TType;
-
-/**
- * @param <X> data type for {@code output()} output
- */
-public final class QuantizedConv2DWithBiasSumAndReluAndRequantize<X extends TType> extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.nn.QuantizedConv2DWithBiasSumAndReluAndRequantize}
-   */
-  public static class Options {
-    
-    /**
-     * @param dilations 
-     */
-    public Options dilations(List<Long> dilations) {
-      this.dilations = dilations;
-      return this;
-    }
-    
-    /**
-     * @param paddingList 
-     */
-    public Options paddingList(List<Long> paddingList) {
-      this.paddingList = paddingList;
-      return this;
-    }
-    
-    private List<Long> dilations;
-    private List<Long> paddingList;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new QuantizedConv2DWithBiasSumAndReluAndRequantize operation.
-   * 
-   * @param scope current scope
-   * @param input 
-   * @param filter 
-   * @param bias 
-   * @param minInput 
-   * @param maxInput 
-   * @param minFilter 
-   * @param maxFilter 
-   * @param minFreezedOutput 
-   * @param maxFreezedOutput 
-   * @param summand 
-   * @param minSummand 
-   * @param maxSummand 
-   * @param outType 
-   * @param strides 
-   * @param padding 
-   * @param options carries optional attributes values
-   * @return a new instance of QuantizedConv2DWithBiasSumAndReluAndRequantize
-   */
-  @Endpoint(describeByClass = true)
-  public static <X extends TType, T extends TType, U extends TType, V extends TType, W extends TType> QuantizedConv2DWithBiasSumAndReluAndRequantize<X> create(Scope scope, Operand<T> input, Operand<U> filter, Operand<V> bias, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, Operand<TFloat32> minFilter, Operand<TFloat32> maxFilter, Operand<TFloat32> minFreezedOutput, Operand<TFloat32> maxFreezedOutput, Operand<W> summand, Operand<TFloat32> minSummand, Operand<TFloat32> maxSummand, Class<X> outType, List<Long> strides, String padding, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("QuantizedConv2DWithBiasSumAndReluAndRequantize", scope.makeOpName("QuantizedConv2DWithBiasSumAndReluAndRequantize"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder.addInput(filter.asOutput(scope));
-    opBuilder.addInput(bias.asOutput(scope));
-    opBuilder.addInput(minInput.asOutput(scope));
-    opBuilder.addInput(maxInput.asOutput(scope));
-    opBuilder.addInput(minFilter.asOutput(scope));
-    opBuilder.addInput(maxFilter.asOutput(scope));
-    opBuilder.addInput(minFreezedOutput.asOutput(scope));
-    opBuilder.addInput(maxFreezedOutput.asOutput(scope));
-    opBuilder.addInput(summand.asOutput(scope));
-    opBuilder.addInput(minSummand.asOutput(scope));
-    opBuilder.addInput(maxSummand.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("out_type", outType);
-    long[] stridesArray = new long[strides.size()];
-    for (int i = 0; i < stridesArray.length; ++i) {
-      stridesArray[i] = strides.get(i);
-    }
-    opBuilder.setAttr("strides", stridesArray);
-    opBuilder.setAttr("padding", padding);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.dilations != null) {
-          long[] dilationsArray = new long[opts.dilations.size()];
-          for (int i = 0; i < dilationsArray.length; ++i) {
-            dilationsArray[i] = opts.dilations.get(i);
-          }
-          opBuilder.setAttr("dilations", dilationsArray);
-        }
-        if (opts.paddingList != null) {
-          long[] paddingListArray = new long[opts.paddingList.size()];
-          for (int i = 0; i < paddingListArray.length; ++i) {
-            paddingListArray[i] = opts.paddingList.get(i);
-          }
-          opBuilder.setAttr("padding_list", paddingListArray);
-        }
-      }
-    }
-    return new QuantizedConv2DWithBiasSumAndReluAndRequantize<X>(opBuilder.build());
-  }
-  
-  /**
-   * @param dilations 
-   */
-  public static Options dilations(List<Long> dilations) {
-    return new Options().dilations(dilations);
-  }
-  
-  /**
-   * @param paddingList 
-   */
-  public static Options paddingList(List<Long> paddingList) {
-    return new Options().paddingList(paddingList);
-  }
-  
-  /**
-   */
-  public Output<X> output() {
-    return output;
-  }
-  
-  /**
-   */
-  public Output<TFloat32> minOutput() {
-    return minOutput;
-  }
-  
-  /**
-   */
-  public Output<TFloat32> maxOutput() {
-    return maxOutput;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "QuantizedConv2DWithBiasSumAndReluAndRequantize";
-  
-  private Output<X> output;
-  private Output<TFloat32> minOutput;
-  private Output<TFloat32> maxOutput;
-  
-  private QuantizedConv2DWithBiasSumAndReluAndRequantize(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-    minOutput = operation.output(outputIdx++);
-    maxOutput = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2d.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2d.java
deleted file mode 100644
index 7ed9b36502d..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2d.java
+++ /dev/null
@@ -1,161 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.nn;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.family.TType;
-
-/**
- * Computes a 2D convolution given quantized 4D input and filter tensors.
- * <p>
- * The inputs are quantized tensors where the lowest value represents the real
- * number of the associated minimum, and the highest represents the maximum.
- * This means that you can only interpret the quantized output in the same way, by
- * taking the returned minimum and maximum values into account.
- * 
- * @param <V> data type for {@code output()} output
- */
-@Operator(group = "nn")
-public final class QuantizedConv2d<V extends TType> extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.nn.QuantizedConv2d}
-   */
-  public static class Options {
-    
-    /**
-     * @param dilations 1-D tensor of length 4.  The dilation factor for each dimension of
-     * `input`. If set to k > 1, there will be k-1 skipped cells between each
-     * filter element on that dimension. The dimension order is determined by the
-     * value of `data_format`, see above for details. Dilations in the batch and
-     * depth dimensions must be 1.
-     */
-    public Options dilations(List<Long> dilations) {
-      this.dilations = dilations;
-      return this;
-    }
-    
-    private List<Long> dilations;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new QuantizedConv2d operation.
-   * 
-   * @param scope current scope
-   * @param input 
-   * @param filter filter's input_depth dimension must match input's depth dimensions.
-   * @param minInput The float value that the lowest quantized input value represents.
-   * @param maxInput The float value that the highest quantized input value represents.
-   * @param minFilter The float value that the lowest quantized filter value represents.
-   * @param maxFilter The float value that the highest quantized filter value represents.
-   * @param outType 
-   * @param strides The stride of the sliding window for each dimension of the input
-   * tensor.
-   * @param padding The type of padding algorithm to use.
-   * @param options carries optional attributes values
-   * @return a new instance of QuantizedConv2d
-   */
-  @Endpoint(describeByClass = true)
-  public static <V extends TType, T extends TType, U extends TType> QuantizedConv2d<V> create(Scope scope, Operand<T> input, Operand<U> filter, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, Operand<TFloat32> minFilter, Operand<TFloat32> maxFilter, Class<V> outType, List<Long> strides, String padding, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("QuantizedConv2D", scope.makeOpName("QuantizedConv2d"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder.addInput(filter.asOutput(scope));
-    opBuilder.addInput(minInput.asOutput(scope));
-    opBuilder.addInput(maxInput.asOutput(scope));
-    opBuilder.addInput(minFilter.asOutput(scope));
-    opBuilder.addInput(maxFilter.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("out_type", outType);
-    long[] stridesArray = new long[strides.size()];
-    for (int i = 0; i < stridesArray.length; ++i) {
-      stridesArray[i] = strides.get(i);
-    }
-    opBuilder.setAttr("strides", stridesArray);
-    opBuilder.setAttr("padding", padding);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.dilations != null) {
-          long[] dilationsArray = new long[opts.dilations.size()];
-          for (int i = 0; i < dilationsArray.length; ++i) {
-            dilationsArray[i] = opts.dilations.get(i);
-          }
-          opBuilder.setAttr("dilations", dilationsArray);
-        }
-      }
-    }
-    return new QuantizedConv2d<V>(opBuilder.build());
-  }
-  
-  /**
-   * @param dilations 1-D tensor of length 4.  The dilation factor for each dimension of
-   * `input`. If set to k > 1, there will be k-1 skipped cells between each
-   * filter element on that dimension. The dimension order is determined by the
-   * value of `data_format`, see above for details. Dilations in the batch and
-   * depth dimensions must be 1.
-   */
-  public static Options dilations(List<Long> dilations) {
-    return new Options().dilations(dilations);
-  }
-  
-  /**
-   */
-  public Output<V> output() {
-    return output;
-  }
-  
-  /**
-   * The float value that the lowest quantized output value represents.
-   */
-  public Output<TFloat32> minOutput() {
-    return minOutput;
-  }
-  
-  /**
-   * The float value that the highest quantized output value represents.
-   */
-  public Output<TFloat32> maxOutput() {
-    return maxOutput;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "QuantizedConv2D";
-  
-  private Output<V> output;
-  private Output<TFloat32> minOutput;
-  private Output<TFloat32> maxOutput;
-  
-  private QuantizedConv2d(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-    minOutput = operation.output(outputIdx++);
-    maxOutput = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedDepthwiseConv2D.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedDepthwiseConv2D.java
deleted file mode 100644
index a480adc28f5..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedDepthwiseConv2D.java
+++ /dev/null
@@ -1,147 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.nn;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.family.TType;
-
-/**
- * Computes quantized depthwise Conv2D.
- * 
- * @param <V> data type for {@code output()} output
- */
-public final class QuantizedDepthwiseConv2D<V extends TType> extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.nn.QuantizedDepthwiseConv2D}
-   */
-  public static class Options {
-    
-    /**
-     * @param dilations List of dilation values.
-     */
-    public Options dilations(List<Long> dilations) {
-      this.dilations = dilations;
-      return this;
-    }
-    
-    private List<Long> dilations;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new QuantizedDepthwiseConv2D operation.
-   * 
-   * @param scope current scope
-   * @param input The original input tensor.
-   * @param filter The original filter tensor.
-   * @param minInput The float value that the minimum quantized input value represents.
-   * @param maxInput The float value that the maximum quantized input value represents.
-   * @param minFilter The float value that the minimum quantized filter value represents.
-   * @param maxFilter The float value that the maximum quantized filter value represents.
-   * @param outType The type of the output.
-   * @param strides List of stride values.
-   * @param padding 
-   * @param options carries optional attributes values
-   * @return a new instance of QuantizedDepthwiseConv2D
-   */
-  @Endpoint(describeByClass = true)
-  public static <V extends TType, T extends TType, U extends TType> QuantizedDepthwiseConv2D<V> create(Scope scope, Operand<T> input, Operand<U> filter, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, Operand<TFloat32> minFilter, Operand<TFloat32> maxFilter, Class<V> outType, List<Long> strides, String padding, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("QuantizedDepthwiseConv2D", scope.makeOpName("QuantizedDepthwiseConv2D"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder.addInput(filter.asOutput(scope));
-    opBuilder.addInput(minInput.asOutput(scope));
-    opBuilder.addInput(maxInput.asOutput(scope));
-    opBuilder.addInput(minFilter.asOutput(scope));
-    opBuilder.addInput(maxFilter.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("out_type", outType);
-    long[] stridesArray = new long[strides.size()];
-    for (int i = 0; i < stridesArray.length; ++i) {
-      stridesArray[i] = strides.get(i);
-    }
-    opBuilder.setAttr("strides", stridesArray);
-    opBuilder.setAttr("padding", padding);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.dilations != null) {
-          long[] dilationsArray = new long[opts.dilations.size()];
-          for (int i = 0; i < dilationsArray.length; ++i) {
-            dilationsArray[i] = opts.dilations.get(i);
-          }
-          opBuilder.setAttr("dilations", dilationsArray);
-        }
-      }
-    }
-    return new QuantizedDepthwiseConv2D<V>(opBuilder.build());
-  }
-  
-  /**
-   * @param dilations List of dilation values.
-   */
-  public static Options dilations(List<Long> dilations) {
-    return new Options().dilations(dilations);
-  }
-  
-  /**
-   * The output tensor.
-   */
-  public Output<V> output() {
-    return output;
-  }
-  
-  /**
-   * The float value that the minimum quantized output value represents.
-   */
-  public Output<TFloat32> minOutput() {
-    return minOutput;
-  }
-  
-  /**
-   * The float value that the maximum quantized output value represents.
-   */
-  public Output<TFloat32> maxOutput() {
-    return maxOutput;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "QuantizedDepthwiseConv2D";
-  
-  private Output<V> output;
-  private Output<TFloat32> minOutput;
-  private Output<TFloat32> maxOutput;
-  
-  private QuantizedDepthwiseConv2D(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-    minOutput = operation.output(outputIdx++);
-    maxOutput = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedDepthwiseConv2DWithBias.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedDepthwiseConv2DWithBias.java
deleted file mode 100644
index 4b3326a0f18..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedDepthwiseConv2DWithBias.java
+++ /dev/null
@@ -1,149 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.nn;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.family.TType;
-
-/**
- * Computes quantized depthwise Conv2D with Bias.
- * 
- * @param <V> data type for {@code output()} output
- */
-public final class QuantizedDepthwiseConv2DWithBias<V extends TType> extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.nn.QuantizedDepthwiseConv2DWithBias}
-   */
-  public static class Options {
-    
-    /**
-     * @param dilations List of dilation values.
-     */
-    public Options dilations(List<Long> dilations) {
-      this.dilations = dilations;
-      return this;
-    }
-    
-    private List<Long> dilations;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new QuantizedDepthwiseConv2DWithBias operation.
-   * 
-   * @param scope current scope
-   * @param input The original input tensor.
-   * @param filter The original filter tensor.
-   * @param bias The original bias tensor.
-   * @param minInput The float value that the minimum quantized input value represents.
-   * @param maxInput The float value that the maximum quantized input value represents.
-   * @param minFilter The float value that the minimum quantized filter value represents.
-   * @param maxFilter The float value that the maximum quantized filter value represents.
-   * @param outType The type of the output.
-   * @param strides List of stride values.
-   * @param padding 
-   * @param options carries optional attributes values
-   * @return a new instance of QuantizedDepthwiseConv2DWithBias
-   */
-  @Endpoint(describeByClass = true)
-  public static <V extends TType, T extends TType, U extends TType> QuantizedDepthwiseConv2DWithBias<V> create(Scope scope, Operand<T> input, Operand<U> filter, Operand<TFloat32> bias, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, Operand<TFloat32> minFilter, Operand<TFloat32> maxFilter, Class<V> outType, List<Long> strides, String padding, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("QuantizedDepthwiseConv2DWithBias", scope.makeOpName("QuantizedDepthwiseConv2DWithBias"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder.addInput(filter.asOutput(scope));
-    opBuilder.addInput(bias.asOutput(scope));
-    opBuilder.addInput(minInput.asOutput(scope));
-    opBuilder.addInput(maxInput.asOutput(scope));
-    opBuilder.addInput(minFilter.asOutput(scope));
-    opBuilder.addInput(maxFilter.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("out_type", outType);
-    long[] stridesArray = new long[strides.size()];
-    for (int i = 0; i < stridesArray.length; ++i) {
-      stridesArray[i] = strides.get(i);
-    }
-    opBuilder.setAttr("strides", stridesArray);
-    opBuilder.setAttr("padding", padding);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.dilations != null) {
-          long[] dilationsArray = new long[opts.dilations.size()];
-          for (int i = 0; i < dilationsArray.length; ++i) {
-            dilationsArray[i] = opts.dilations.get(i);
-          }
-          opBuilder.setAttr("dilations", dilationsArray);
-        }
-      }
-    }
-    return new QuantizedDepthwiseConv2DWithBias<V>(opBuilder.build());
-  }
-  
-  /**
-   * @param dilations List of dilation values.
-   */
-  public static Options dilations(List<Long> dilations) {
-    return new Options().dilations(dilations);
-  }
-  
-  /**
-   * The output tensor.
-   */
-  public Output<V> output() {
-    return output;
-  }
-  
-  /**
-   * The float value that the minimum quantized output value represents.
-   */
-  public Output<TFloat32> minOutput() {
-    return minOutput;
-  }
-  
-  /**
-   * The float value that the maximum quantized output value represents.
-   */
-  public Output<TFloat32> maxOutput() {
-    return maxOutput;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "QuantizedDepthwiseConv2DWithBias";
-  
-  private Output<V> output;
-  private Output<TFloat32> minOutput;
-  private Output<TFloat32> maxOutput;
-  
-  private QuantizedDepthwiseConv2DWithBias(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-    minOutput = operation.output(outputIdx++);
-    maxOutput = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedDepthwiseConv2DWithBiasAndRelu.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedDepthwiseConv2DWithBiasAndRelu.java
deleted file mode 100644
index b7c82b9ca57..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedDepthwiseConv2DWithBiasAndRelu.java
+++ /dev/null
@@ -1,172 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.nn;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.family.TType;
-
-/**
- * Computes quantized depthwise Conv2D with Bias and Relu.
- * 
- * @param <V> data type for {@code output()} output
- */
-public final class QuantizedDepthwiseConv2DWithBiasAndRelu<V extends TType> extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.nn.QuantizedDepthwiseConv2DWithBiasAndRelu}
-   */
-  public static class Options {
-    
-    /**
-     * @param dilations List of dilation values.
-     */
-    public Options dilations(List<Long> dilations) {
-      this.dilations = dilations;
-      return this;
-    }
-    
-    /**
-     * @param paddingList 
-     */
-    public Options paddingList(List<Long> paddingList) {
-      this.paddingList = paddingList;
-      return this;
-    }
-    
-    private List<Long> dilations;
-    private List<Long> paddingList;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new QuantizedDepthwiseConv2DWithBiasAndRelu operation.
-   * 
-   * @param scope current scope
-   * @param input The original input tensor.
-   * @param filter The original filter tensor.
-   * @param bias The original bias tensor.
-   * @param minInput The float value that the minimum quantized input value represents.
-   * @param maxInput The float value that the maximum quantized input value represents.
-   * @param minFilter The float value that the minimum quantized filter value represents.
-   * @param maxFilter The float value that the maximum quantized filter value represents.
-   * @param outType The type of the output.
-   * @param strides List of stride values.
-   * @param padding 
-   * @param options carries optional attributes values
-   * @return a new instance of QuantizedDepthwiseConv2DWithBiasAndRelu
-   */
-  @Endpoint(describeByClass = true)
-  public static <V extends TType, T extends TType, U extends TType> QuantizedDepthwiseConv2DWithBiasAndRelu<V> create(Scope scope, Operand<T> input, Operand<U> filter, Operand<TFloat32> bias, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, Operand<TFloat32> minFilter, Operand<TFloat32> maxFilter, Class<V> outType, List<Long> strides, String padding, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("QuantizedDepthwiseConv2DWithBiasAndRelu", scope.makeOpName("QuantizedDepthwiseConv2DWithBiasAndRelu"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder.addInput(filter.asOutput(scope));
-    opBuilder.addInput(bias.asOutput(scope));
-    opBuilder.addInput(minInput.asOutput(scope));
-    opBuilder.addInput(maxInput.asOutput(scope));
-    opBuilder.addInput(minFilter.asOutput(scope));
-    opBuilder.addInput(maxFilter.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("out_type", outType);
-    long[] stridesArray = new long[strides.size()];
-    for (int i = 0; i < stridesArray.length; ++i) {
-      stridesArray[i] = strides.get(i);
-    }
-    opBuilder.setAttr("strides", stridesArray);
-    opBuilder.setAttr("padding", padding);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.dilations != null) {
-          long[] dilationsArray = new long[opts.dilations.size()];
-          for (int i = 0; i < dilationsArray.length; ++i) {
-            dilationsArray[i] = opts.dilations.get(i);
-          }
-          opBuilder.setAttr("dilations", dilationsArray);
-        }
-        if (opts.paddingList != null) {
-          long[] paddingListArray = new long[opts.paddingList.size()];
-          for (int i = 0; i < paddingListArray.length; ++i) {
-            paddingListArray[i] = opts.paddingList.get(i);
-          }
-          opBuilder.setAttr("padding_list", paddingListArray);
-        }
-      }
-    }
-    return new QuantizedDepthwiseConv2DWithBiasAndRelu<V>(opBuilder.build());
-  }
-  
-  /**
-   * @param dilations List of dilation values.
-   */
-  public static Options dilations(List<Long> dilations) {
-    return new Options().dilations(dilations);
-  }
-  
-  /**
-   * @param paddingList 
-   */
-  public static Options paddingList(List<Long> paddingList) {
-    return new Options().paddingList(paddingList);
-  }
-  
-  /**
-   * The output tensor.
-   */
-  public Output<V> output() {
-    return output;
-  }
-  
-  /**
-   * The float value that the minimum quantized output value represents.
-   */
-  public Output<TFloat32> minOutput() {
-    return minOutput;
-  }
-  
-  /**
-   * The float value that the maximum quantized output value represents.
-   */
-  public Output<TFloat32> maxOutput() {
-    return maxOutput;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "QuantizedDepthwiseConv2DWithBiasAndRelu";
-  
-  private Output<V> output;
-  private Output<TFloat32> minOutput;
-  private Output<TFloat32> maxOutput;
-  
-  private QuantizedDepthwiseConv2DWithBiasAndRelu(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-    minOutput = operation.output(outputIdx++);
-    maxOutput = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedDepthwiseConv2DWithBiasAndReluAndRequantize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedDepthwiseConv2DWithBiasAndReluAndRequantize.java
deleted file mode 100644
index 0046acce74b..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedDepthwiseConv2DWithBiasAndReluAndRequantize.java
+++ /dev/null
@@ -1,176 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.nn;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.family.TType;
-
-/**
- * Computes quantized depthwise Conv2D with Bias, Relu and Requantize.
- * 
- * @param <W> data type for {@code output()} output
- */
-public final class QuantizedDepthwiseConv2DWithBiasAndReluAndRequantize<W extends TType> extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.nn.QuantizedDepthwiseConv2DWithBiasAndReluAndRequantize}
-   */
-  public static class Options {
-    
-    /**
-     * @param dilations List of dilation values.
-     */
-    public Options dilations(List<Long> dilations) {
-      this.dilations = dilations;
-      return this;
-    }
-    
-    /**
-     * @param paddingList 
-     */
-    public Options paddingList(List<Long> paddingList) {
-      this.paddingList = paddingList;
-      return this;
-    }
-    
-    private List<Long> dilations;
-    private List<Long> paddingList;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new QuantizedDepthwiseConv2DWithBiasAndReluAndRequantize operation.
-   * 
-   * @param scope current scope
-   * @param input The original input tensor.
-   * @param filter The original filter tensor.
-   * @param bias The original bias tensor.
-   * @param minInput The float value that the minimum quantized input value represents.
-   * @param maxInput The float value that the maximum quantized input value represents.
-   * @param minFilter The float value that the minimum quantized filter value represents.
-   * @param maxFilter The float value that the maximum quantized filter value represents.
-   * @param minFreezedOutput The minimum float value of the output tensor.
-   * @param maxFreezedOutput The maximum float value of the output tensor.
-   * @param outType The type of the output.
-   * @param strides List of stride values.
-   * @param padding 
-   * @param options carries optional attributes values
-   * @return a new instance of QuantizedDepthwiseConv2DWithBiasAndReluAndRequantize
-   */
-  @Endpoint(describeByClass = true)
-  public static <W extends TType, T extends TType, U extends TType, V extends TType> QuantizedDepthwiseConv2DWithBiasAndReluAndRequantize<W> create(Scope scope, Operand<T> input, Operand<U> filter, Operand<V> bias, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, Operand<TFloat32> minFilter, Operand<TFloat32> maxFilter, Operand<TFloat32> minFreezedOutput, Operand<TFloat32> maxFreezedOutput, Class<W> outType, List<Long> strides, String padding, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("QuantizedDepthwiseConv2DWithBiasAndReluAndRequantize", scope.makeOpName("QuantizedDepthwiseConv2DWithBiasAndReluAndRequantize"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder.addInput(filter.asOutput(scope));
-    opBuilder.addInput(bias.asOutput(scope));
-    opBuilder.addInput(minInput.asOutput(scope));
-    opBuilder.addInput(maxInput.asOutput(scope));
-    opBuilder.addInput(minFilter.asOutput(scope));
-    opBuilder.addInput(maxFilter.asOutput(scope));
-    opBuilder.addInput(minFreezedOutput.asOutput(scope));
-    opBuilder.addInput(maxFreezedOutput.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("out_type", outType);
-    long[] stridesArray = new long[strides.size()];
-    for (int i = 0; i < stridesArray.length; ++i) {
-      stridesArray[i] = strides.get(i);
-    }
-    opBuilder.setAttr("strides", stridesArray);
-    opBuilder.setAttr("padding", padding);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.dilations != null) {
-          long[] dilationsArray = new long[opts.dilations.size()];
-          for (int i = 0; i < dilationsArray.length; ++i) {
-            dilationsArray[i] = opts.dilations.get(i);
-          }
-          opBuilder.setAttr("dilations", dilationsArray);
-        }
-        if (opts.paddingList != null) {
-          long[] paddingListArray = new long[opts.paddingList.size()];
-          for (int i = 0; i < paddingListArray.length; ++i) {
-            paddingListArray[i] = opts.paddingList.get(i);
-          }
-          opBuilder.setAttr("padding_list", paddingListArray);
-        }
-      }
-    }
-    return new QuantizedDepthwiseConv2DWithBiasAndReluAndRequantize<W>(opBuilder.build());
-  }
-  
-  /**
-   * @param dilations List of dilation values.
-   */
-  public static Options dilations(List<Long> dilations) {
-    return new Options().dilations(dilations);
-  }
-  
-  /**
-   * @param paddingList 
-   */
-  public static Options paddingList(List<Long> paddingList) {
-    return new Options().paddingList(paddingList);
-  }
-  
-  /**
-   * The output tensor.
-   */
-  public Output<W> output() {
-    return output;
-  }
-  
-  /**
-   * The float value that the minimum quantized output value represents.
-   */
-  public Output<TFloat32> minOutput() {
-    return minOutput;
-  }
-  
-  /**
-   * The float value that the maximum quantized output value represents.
-   */
-  public Output<TFloat32> maxOutput() {
-    return maxOutput;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "QuantizedDepthwiseConv2DWithBiasAndReluAndRequantize";
-  
-  private Output<W> output;
-  private Output<TFloat32> minOutput;
-  private Output<TFloat32> maxOutput;
-  
-  private QuantizedDepthwiseConv2DWithBiasAndReluAndRequantize(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-    minOutput = operation.output(outputIdx++);
-    maxOutput = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedInstanceNorm.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedInstanceNorm.java
deleted file mode 100644
index 5c12e5a6b71..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedInstanceNorm.java
+++ /dev/null
@@ -1,207 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.nn;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.family.TType;
-
-/**
- * Quantized Instance normalization.
- * 
- * @param <T> data type for {@code y()} output
- */
-@Operator(group = "nn")
-public final class QuantizedInstanceNorm<T extends TType> extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.nn.QuantizedInstanceNorm}
-   */
-  public static class Options {
-    
-    /**
-     * @param outputRangeGiven If True, `given_y_min` and `given_y_min`
-     * and `given_y_max` are used as the output range. Otherwise,
-     * the implementation computes the output range.
-     */
-    public Options outputRangeGiven(Boolean outputRangeGiven) {
-      this.outputRangeGiven = outputRangeGiven;
-      return this;
-    }
-    
-    /**
-     * @param givenYMin Output in `y_min` if `output_range_given` is True.
-     */
-    public Options givenYMin(Float givenYMin) {
-      this.givenYMin = givenYMin;
-      return this;
-    }
-    
-    /**
-     * @param givenYMax Output in `y_max` if `output_range_given` is True.
-     */
-    public Options givenYMax(Float givenYMax) {
-      this.givenYMax = givenYMax;
-      return this;
-    }
-    
-    /**
-     * @param varianceEpsilon A small float number to avoid dividing by 0.
-     */
-    public Options varianceEpsilon(Float varianceEpsilon) {
-      this.varianceEpsilon = varianceEpsilon;
-      return this;
-    }
-    
-    /**
-     * @param minSeparation Minimum value of `y_max - y_min`
-     */
-    public Options minSeparation(Float minSeparation) {
-      this.minSeparation = minSeparation;
-      return this;
-    }
-    
-    private Boolean outputRangeGiven;
-    private Float givenYMin;
-    private Float givenYMax;
-    private Float varianceEpsilon;
-    private Float minSeparation;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new QuantizedInstanceNorm operation.
-   * 
-   * @param scope current scope
-   * @param x A 4D input Tensor.
-   * @param xMin The value represented by the lowest quantized input.
-   * @param xMax The value represented by the highest quantized input.
-   * @param options carries optional attributes values
-   * @return a new instance of QuantizedInstanceNorm
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> QuantizedInstanceNorm<T> create(Scope scope, Operand<T> x, Operand<TFloat32> xMin, Operand<TFloat32> xMax, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("QuantizedInstanceNorm", scope.makeOpName("QuantizedInstanceNorm"));
-    opBuilder.addInput(x.asOutput(scope));
-    opBuilder.addInput(xMin.asOutput(scope));
-    opBuilder.addInput(xMax.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.outputRangeGiven != null) {
-          opBuilder.setAttr("output_range_given", opts.outputRangeGiven);
-        }
-        if (opts.givenYMin != null) {
-          opBuilder.setAttr("given_y_min", opts.givenYMin);
-        }
-        if (opts.givenYMax != null) {
-          opBuilder.setAttr("given_y_max", opts.givenYMax);
-        }
-        if (opts.varianceEpsilon != null) {
-          opBuilder.setAttr("variance_epsilon", opts.varianceEpsilon);
-        }
-        if (opts.minSeparation != null) {
-          opBuilder.setAttr("min_separation", opts.minSeparation);
-        }
-      }
-    }
-    return new QuantizedInstanceNorm<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param outputRangeGiven If True, `given_y_min` and `given_y_min`
-   * and `given_y_max` are used as the output range. Otherwise,
-   * the implementation computes the output range.
-   */
-  public static Options outputRangeGiven(Boolean outputRangeGiven) {
-    return new Options().outputRangeGiven(outputRangeGiven);
-  }
-  
-  /**
-   * @param givenYMin Output in `y_min` if `output_range_given` is True.
-   */
-  public static Options givenYMin(Float givenYMin) {
-    return new Options().givenYMin(givenYMin);
-  }
-  
-  /**
-   * @param givenYMax Output in `y_max` if `output_range_given` is True.
-   */
-  public static Options givenYMax(Float givenYMax) {
-    return new Options().givenYMax(givenYMax);
-  }
-  
-  /**
-   * @param varianceEpsilon A small float number to avoid dividing by 0.
-   */
-  public static Options varianceEpsilon(Float varianceEpsilon) {
-    return new Options().varianceEpsilon(varianceEpsilon);
-  }
-  
-  /**
-   * @param minSeparation Minimum value of `y_max - y_min`
-   */
-  public static Options minSeparation(Float minSeparation) {
-    return new Options().minSeparation(minSeparation);
-  }
-  
-  /**
-   * A 4D Tensor.
-   */
-  public Output<T> y() {
-    return y;
-  }
-  
-  /**
-   * The value represented by the lowest quantized output.
-   */
-  public Output<TFloat32> yMin() {
-    return yMin;
-  }
-  
-  /**
-   * The value represented by the highest quantized output.
-   */
-  public Output<TFloat32> yMax() {
-    return yMax;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "QuantizedInstanceNorm";
-  
-  private Output<T> y;
-  private Output<TFloat32> yMin;
-  private Output<TFloat32> yMax;
-  
-  private QuantizedInstanceNorm(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    y = operation.output(outputIdx++);
-    yMin = operation.output(outputIdx++);
-    yMax = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedMaxPool.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedMaxPool.java
deleted file mode 100644
index aaea42bd906..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedMaxPool.java
+++ /dev/null
@@ -1,109 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.nn;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.family.TType;
-
-/**
- * Produces the max pool of the input tensor for quantized types.
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator(group = "nn")
-public final class QuantizedMaxPool<T extends TType> extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new QuantizedMaxPool operation.
-   * 
-   * @param scope current scope
-   * @param input The 4D (batch x rows x cols x depth) Tensor to MaxReduce over.
-   * @param minInput The float value that the lowest quantized input value represents.
-   * @param maxInput The float value that the highest quantized input value represents.
-   * @param ksize The size of the window for each dimension of the input tensor.
-   * The length must be 4 to match the number of dimensions of the input.
-   * @param strides The stride of the sliding window for each dimension of the input
-   * tensor. The length must be 4 to match the number of dimensions of the input.
-   * @param padding The type of padding algorithm to use.
-   * @return a new instance of QuantizedMaxPool
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> QuantizedMaxPool<T> create(Scope scope, Operand<T> input, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, List<Long> ksize, List<Long> strides, String padding) {
-    OperationBuilder opBuilder = scope.env().opBuilder("QuantizedMaxPool", scope.makeOpName("QuantizedMaxPool"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder.addInput(minInput.asOutput(scope));
-    opBuilder.addInput(maxInput.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    long[] ksizeArray = new long[ksize.size()];
-    for (int i = 0; i < ksizeArray.length; ++i) {
-      ksizeArray[i] = ksize.get(i);
-    }
-    opBuilder.setAttr("ksize", ksizeArray);
-    long[] stridesArray = new long[strides.size()];
-    for (int i = 0; i < stridesArray.length; ++i) {
-      stridesArray[i] = strides.get(i);
-    }
-    opBuilder.setAttr("strides", stridesArray);
-    opBuilder.setAttr("padding", padding);
-    return new QuantizedMaxPool<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  /**
-   * The float value that the lowest quantized output value represents.
-   */
-  public Output<TFloat32> minOutput() {
-    return minOutput;
-  }
-  
-  /**
-   * The float value that the highest quantized output value represents.
-   */
-  public Output<TFloat32> maxOutput() {
-    return maxOutput;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "QuantizedMaxPool";
-  
-  private Output<T> output;
-  private Output<TFloat32> minOutput;
-  private Output<TFloat32> maxOutput;
-  
-  private QuantizedMaxPool(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-    minOutput = operation.output(outputIdx++);
-    maxOutput = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedRelu.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedRelu.java
deleted file mode 100644
index 60115bdf49e..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedRelu.java
+++ /dev/null
@@ -1,95 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.nn;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.family.TType;
-
-/**
- * Computes Quantized Rectified Linear: `max(features, 0)`
- * 
- * @param <U> data type for {@code activations()} output
- */
-@Operator(group = "nn")
-public final class QuantizedRelu<U extends TType> extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new QuantizedRelu operation.
-   * 
-   * @param scope current scope
-   * @param features 
-   * @param minFeatures The float value that the lowest quantized value represents.
-   * @param maxFeatures The float value that the highest quantized value represents.
-   * @param outType 
-   * @return a new instance of QuantizedRelu
-   */
-  @Endpoint(describeByClass = true)
-  public static <U extends TType, T extends TType> QuantizedRelu<U> create(Scope scope, Operand<T> features, Operand<TFloat32> minFeatures, Operand<TFloat32> maxFeatures, Class<U> outType) {
-    OperationBuilder opBuilder = scope.env().opBuilder("QuantizedRelu", scope.makeOpName("QuantizedRelu"));
-    opBuilder.addInput(features.asOutput(scope));
-    opBuilder.addInput(minFeatures.asOutput(scope));
-    opBuilder.addInput(maxFeatures.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("out_type", outType);
-    return new QuantizedRelu<U>(opBuilder.build());
-  }
-  
-  /**
-   * Has the same output shape as "features".
-   */
-  public Output<U> activations() {
-    return activations;
-  }
-  
-  /**
-   * The float value that the lowest quantized value represents.
-   */
-  public Output<TFloat32> minActivations() {
-    return minActivations;
-  }
-  
-  /**
-   * The float value that the highest quantized value represents.
-   */
-  public Output<TFloat32> maxActivations() {
-    return maxActivations;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "QuantizedRelu";
-  
-  private Output<U> activations;
-  private Output<TFloat32> minActivations;
-  private Output<TFloat32> maxActivations;
-  
-  private QuantizedRelu(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    activations = operation.output(outputIdx++);
-    minActivations = operation.output(outputIdx++);
-    maxActivations = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedRelu6.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedRelu6.java
deleted file mode 100644
index 3cf1ee8f7c5..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedRelu6.java
+++ /dev/null
@@ -1,95 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.nn;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.family.TType;
-
-/**
- * Computes Quantized Rectified Linear 6: `min(max(features, 0), 6)`
- * 
- * @param <U> data type for {@code activations()} output
- */
-@Operator(group = "nn")
-public final class QuantizedRelu6<U extends TType> extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new QuantizedRelu6 operation.
-   * 
-   * @param scope current scope
-   * @param features 
-   * @param minFeatures The float value that the lowest quantized value represents.
-   * @param maxFeatures The float value that the highest quantized value represents.
-   * @param outType 
-   * @return a new instance of QuantizedRelu6
-   */
-  @Endpoint(describeByClass = true)
-  public static <U extends TType, T extends TType> QuantizedRelu6<U> create(Scope scope, Operand<T> features, Operand<TFloat32> minFeatures, Operand<TFloat32> maxFeatures, Class<U> outType) {
-    OperationBuilder opBuilder = scope.env().opBuilder("QuantizedRelu6", scope.makeOpName("QuantizedRelu6"));
-    opBuilder.addInput(features.asOutput(scope));
-    opBuilder.addInput(minFeatures.asOutput(scope));
-    opBuilder.addInput(maxFeatures.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("out_type", outType);
-    return new QuantizedRelu6<U>(opBuilder.build());
-  }
-  
-  /**
-   * Has the same output shape as "features".
-   */
-  public Output<U> activations() {
-    return activations;
-  }
-  
-  /**
-   * The float value that the lowest quantized value represents.
-   */
-  public Output<TFloat32> minActivations() {
-    return minActivations;
-  }
-  
-  /**
-   * The float value that the highest quantized value represents.
-   */
-  public Output<TFloat32> maxActivations() {
-    return maxActivations;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "QuantizedRelu6";
-  
-  private Output<U> activations;
-  private Output<TFloat32> minActivations;
-  private Output<TFloat32> maxActivations;
-  
-  private QuantizedRelu6(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    activations = operation.output(outputIdx++);
-    minActivations = operation.output(outputIdx++);
-    maxActivations = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedReluX.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedReluX.java
deleted file mode 100644
index 23e1092e20d..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedReluX.java
+++ /dev/null
@@ -1,97 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.nn;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.family.TType;
-
-/**
- * Computes Quantized Rectified Linear X: `min(max(features, 0), max_value)`
- * 
- * @param <U> data type for {@code activations()} output
- */
-@Operator(group = "nn")
-public final class QuantizedReluX<U extends TType> extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new QuantizedReluX operation.
-   * 
-   * @param scope current scope
-   * @param features 
-   * @param maxValue 
-   * @param minFeatures The float value that the lowest quantized value represents.
-   * @param maxFeatures The float value that the highest quantized value represents.
-   * @param outType 
-   * @return a new instance of QuantizedReluX
-   */
-  @Endpoint(describeByClass = true)
-  public static <U extends TType, T extends TType> QuantizedReluX<U> create(Scope scope, Operand<T> features, Operand<TFloat32> maxValue, Operand<TFloat32> minFeatures, Operand<TFloat32> maxFeatures, Class<U> outType) {
-    OperationBuilder opBuilder = scope.env().opBuilder("QuantizedReluX", scope.makeOpName("QuantizedReluX"));
-    opBuilder.addInput(features.asOutput(scope));
-    opBuilder.addInput(maxValue.asOutput(scope));
-    opBuilder.addInput(minFeatures.asOutput(scope));
-    opBuilder.addInput(maxFeatures.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("out_type", outType);
-    return new QuantizedReluX<U>(opBuilder.build());
-  }
-  
-  /**
-   * Has the same output shape as "features".
-   */
-  public Output<U> activations() {
-    return activations;
-  }
-  
-  /**
-   * The float value that the lowest quantized value represents.
-   */
-  public Output<TFloat32> minActivations() {
-    return minActivations;
-  }
-  
-  /**
-   * The float value that the highest quantized value represents.
-   */
-  public Output<TFloat32> maxActivations() {
-    return maxActivations;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "QuantizedReluX";
-  
-  private Output<U> activations;
-  private Output<TFloat32> minActivations;
-  private Output<TFloat32> maxActivations;
-  
-  private QuantizedReluX(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    activations = operation.output(outputIdx++);
-    minActivations = operation.output(outputIdx++);
-    maxActivations = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Relu.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Relu.java
deleted file mode 100644
index 3482ad1f7e4..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Relu.java
+++ /dev/null
@@ -1,79 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.nn;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Computes rectified linear: `max(features, 0)`.
- * <p>
- * See: https://en.wikipedia.org/wiki/Rectifier_(neural_networks)
- * Example usage:
- * >>> tf.nn.relu([-2., 0., -0., 3.]).numpy()
- * array([ 0.,  0., -0.,  3.], dtype=float32)
- * 
- * @param <T> data type for {@code activations()} output
- */
-@Operator(group = "nn")
-public final class Relu<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new Relu operation.
-   * 
-   * @param scope current scope
-   * @param features 
-   * @return a new instance of Relu
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> Relu<T> create(Scope scope, Operand<T> features) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Relu", scope.makeOpName("Relu"));
-    opBuilder.addInput(features.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new Relu<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> activations() {
-    return activations;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return activations;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Relu";
-  
-  private Output<T> activations;
-  
-  private Relu(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    activations = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Relu6.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Relu6.java
deleted file mode 100644
index f497fec9130..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Relu6.java
+++ /dev/null
@@ -1,74 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.nn;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Computes rectified linear 6: `min(max(features, 0), 6)`.
- * 
- * @param <T> data type for {@code activations()} output
- */
-@Operator(group = "nn")
-public final class Relu6<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new Relu6 operation.
-   * 
-   * @param scope current scope
-   * @param features 
-   * @return a new instance of Relu6
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> Relu6<T> create(Scope scope, Operand<T> features) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Relu6", scope.makeOpName("Relu6"));
-    opBuilder.addInput(features.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new Relu6<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> activations() {
-    return activations;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return activations;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Relu6";
-  
-  private Output<T> activations;
-  
-  private Relu6(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    activations = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Relu6Grad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Relu6Grad.java
deleted file mode 100644
index 77fa55e828a..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Relu6Grad.java
+++ /dev/null
@@ -1,78 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.nn;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Computes rectified linear 6 gradients for a Relu6 operation.
- * 
- * @param <T> data type for {@code backprops()} output
- */
-public final class Relu6Grad<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new Relu6Grad operation.
-   * 
-   * @param scope current scope
-   * @param gradients The backpropagated gradients to the corresponding Relu6 operation.
-   * @param features The features passed as input to the corresponding Relu6 operation, or
-   * its output; using either one produces the same result.
-   * @return a new instance of Relu6Grad
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> Relu6Grad<T> create(Scope scope, Operand<T> gradients, Operand<T> features) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Relu6Grad", scope.makeOpName("Relu6Grad"));
-    opBuilder.addInput(gradients.asOutput(scope));
-    opBuilder.addInput(features.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new Relu6Grad<T>(opBuilder.build());
-  }
-  
-  /**
-   * The gradients:
-   * `gradients * (features > 0) * (features < 6)`.
-   */
-  public Output<T> backprops() {
-    return backprops;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return backprops;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Relu6Grad";
-  
-  private Output<T> backprops;
-  
-  private Relu6Grad(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    backprops = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/ReluGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/ReluGrad.java
deleted file mode 100644
index 936a7a8df6b..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/ReluGrad.java
+++ /dev/null
@@ -1,77 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.nn;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Computes rectified linear gradients for a Relu operation.
- * 
- * @param <T> data type for {@code backprops()} output
- */
-public final class ReluGrad<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new ReluGrad operation.
-   * 
-   * @param scope current scope
-   * @param gradients The backpropagated gradients to the corresponding Relu operation.
-   * @param features The features passed as input to the corresponding Relu operation, OR
-   * the outputs of that operation (both work equivalently).
-   * @return a new instance of ReluGrad
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> ReluGrad<T> create(Scope scope, Operand<T> gradients, Operand<T> features) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ReluGrad", scope.makeOpName("ReluGrad"));
-    opBuilder.addInput(gradients.asOutput(scope));
-    opBuilder.addInput(features.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new ReluGrad<T>(opBuilder.build());
-  }
-  
-  /**
-   * `gradients * (features > 0)`.
-   */
-  public Output<T> backprops() {
-    return backprops;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return backprops;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ReluGrad";
-  
-  private Output<T> backprops;
-  
-  private ReluGrad(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    backprops = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Selu.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Selu.java
deleted file mode 100644
index 9a1cc72c687..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Selu.java
+++ /dev/null
@@ -1,82 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.nn;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Computes scaled exponential linear: `scale * alpha * (exp(features) - 1)`
- * <p>
- * if < 0, `scale * features` otherwise.
- * <p>
- * To be used together with
- * `initializer = tf.variance_scaling_initializer(factor=1.0, mode='FAN_IN')`.
- * For correct dropout, use `tf.contrib.nn.alpha_dropout`.
- * <p>
- * See [Self-Normalizing Neural Networks](https://arxiv.org/abs/1706.02515)
- * 
- * @param <T> data type for {@code activations()} output
- */
-@Operator(group = "nn")
-public final class Selu<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new Selu operation.
-   * 
-   * @param scope current scope
-   * @param features 
-   * @return a new instance of Selu
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> Selu<T> create(Scope scope, Operand<T> features) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Selu", scope.makeOpName("Selu"));
-    opBuilder.addInput(features.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new Selu<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> activations() {
-    return activations;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return activations;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Selu";
-  
-  private Output<T> activations;
-  
-  private Selu(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    activations = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/SeluGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/SeluGrad.java
deleted file mode 100644
index 5acf2b14f93..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/SeluGrad.java
+++ /dev/null
@@ -1,77 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.nn;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Computes gradients for the scaled exponential linear (Selu) operation.
- * 
- * @param <T> data type for {@code backprops()} output
- */
-public final class SeluGrad<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new SeluGrad operation.
-   * 
-   * @param scope current scope
-   * @param gradients The backpropagated gradients to the corresponding Selu operation.
-   * @param outputs The outputs of the corresponding Selu operation.
-   * @return a new instance of SeluGrad
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> SeluGrad<T> create(Scope scope, Operand<T> gradients, Operand<T> outputs) {
-    OperationBuilder opBuilder = scope.env().opBuilder("SeluGrad", scope.makeOpName("SeluGrad"));
-    opBuilder.addInput(gradients.asOutput(scope));
-    opBuilder.addInput(outputs.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new SeluGrad<T>(opBuilder.build());
-  }
-  
-  /**
-   * The gradients: `gradients * (outputs + scale * alpha)`
-   * if outputs < 0, `scale * gradients` otherwise.
-   */
-  public Output<T> backprops() {
-    return backprops;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return backprops;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "SeluGrad";
-  
-  private Output<T> backprops;
-  
-  private SeluGrad(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    backprops = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Softmax.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Softmax.java
deleted file mode 100644
index 3cef45fd7a7..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Softmax.java
+++ /dev/null
@@ -1,79 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.nn;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Computes softmax activations.
- * <p>
- * For each batch `i` and class `j` we have
- * <p>
- *     $$softmax[i, j] = exp(logits[i, j]) / sum_j(exp(logits[i, j]))$$
- * 
- * @param <T> data type for {@code softmax()} output
- */
-@Operator(group = "nn")
-public final class Softmax<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new Softmax operation.
-   * 
-   * @param scope current scope
-   * @param logits 2-D with shape `[batch_size, num_classes]`.
-   * @return a new instance of Softmax
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> Softmax<T> create(Scope scope, Operand<T> logits) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Softmax", scope.makeOpName("Softmax"));
-    opBuilder.addInput(logits.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new Softmax<T>(opBuilder.build());
-  }
-  
-  /**
-   * Same shape as `logits`.
-   */
-  public Output<T> softmax() {
-    return softmax;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return softmax;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Softmax";
-  
-  private Output<T> softmax;
-  
-  private Softmax(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    softmax = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Softsign.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Softsign.java
deleted file mode 100644
index b3494d281a7..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Softsign.java
+++ /dev/null
@@ -1,74 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.nn;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Computes softsign: `features / (abs(features) + 1)`.
- * 
- * @param <T> data type for {@code activations()} output
- */
-@Operator(group = "nn")
-public final class Softsign<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new Softsign operation.
-   * 
-   * @param scope current scope
-   * @param features 
-   * @return a new instance of Softsign
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> Softsign<T> create(Scope scope, Operand<T> features) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Softsign", scope.makeOpName("Softsign"));
-    opBuilder.addInput(features.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new Softsign<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> activations() {
-    return activations;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return activations;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Softsign";
-  
-  private Output<T> activations;
-  
-  private Softsign(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    activations = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/SoftsignGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/SoftsignGrad.java
deleted file mode 100644
index 70f6a9077fb..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/SoftsignGrad.java
+++ /dev/null
@@ -1,76 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.nn;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Computes softsign gradients for a softsign operation.
- * 
- * @param <T> data type for {@code backprops()} output
- */
-public final class SoftsignGrad<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new SoftsignGrad operation.
-   * 
-   * @param scope current scope
-   * @param gradients The backpropagated gradients to the corresponding softsign operation.
-   * @param features The features passed as input to the corresponding softsign operation.
-   * @return a new instance of SoftsignGrad
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> SoftsignGrad<T> create(Scope scope, Operand<T> gradients, Operand<T> features) {
-    OperationBuilder opBuilder = scope.env().opBuilder("SoftsignGrad", scope.makeOpName("SoftsignGrad"));
-    opBuilder.addInput(gradients.asOutput(scope));
-    opBuilder.addInput(features.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new SoftsignGrad<T>(opBuilder.build());
-  }
-  
-  /**
-   * The gradients: `gradients / (1 + abs(features)) ** 2`.
-   */
-  public Output<T> backprops() {
-    return backprops;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return backprops;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "SoftsignGrad";
-  
-  private Output<T> backprops;
-  
-  private SoftsignGrad(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    backprops = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/SpaceToBatch.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/SpaceToBatch.java
deleted file mode 100644
index 1c51c37b53b..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/SpaceToBatch.java
+++ /dev/null
@@ -1,155 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.nn;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * SpaceToBatch for 4-D tensors of type T.
- * <p>
- * This is a legacy version of the more general SpaceToBatchND.
- * <p>
- * Zero-pads and then rearranges (permutes) blocks of spatial data into batch.
- * More specifically, this op outputs a copy of the input tensor where values from
- * the `height` and `width` dimensions are moved to the `batch` dimension. After
- * the zero-padding, both `height` and `width` of the input must be divisible by the
- * block size.
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator(group = "nn")
-public final class SpaceToBatch<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new SpaceToBatch operation.
-   * 
-   * @param scope current scope
-   * @param input 4-D with shape `[batch, height, width, depth]`.
-   * @param paddings 2-D tensor of non-negative integers with shape `[2, 2]`. It specifies
-   *   the padding of the input with zeros across the spatial dimensions as follows:
-   * <p>
-   *       paddings = [[pad_top, pad_bottom], [pad_left, pad_right]]
-   * <p>
-   *   The effective spatial dimensions of the zero-padded input tensor will be:
-   * <p>
-   *       height_pad = pad_top + height + pad_bottom
-   *       width_pad = pad_left + width + pad_right
-   * <p>
-   * The attr `block_size` must be greater than one. It indicates the block size.
-   * <p>
-   *   * Non-overlapping blocks of size `block_size x block size` in the height and
-   *     width dimensions are rearranged into the batch dimension at each location.
-   *   * The batch of the output tensor is `batch * block_size * block_size`.
-   *   * Both height_pad and width_pad must be divisible by block_size.
-   * <p>
-   * The shape of the output will be:
-   * <p>
-   *     [batch<i>block_size</i>block_size, height_pad/block_size, width_pad/block_size,
-   *      depth]
-   * <p>
-   * Some examples:
-   * <p>
-   * (1) For the following input of shape `[1, 2, 2, 1]` and block_size of 2:
-   * <pre>{@code
-   * x = [[[[1], [2]], [[3], [4]]]]
-   * }</pre>
-   * The output tensor has shape `[4, 1, 1, 1]` and value:
-   * <pre>{@code
-   * [[[[1]]], [[[2]]], [[[3]]], [[[4]]]]
-   * }</pre>
-   * (2) For the following input of shape `[1, 2, 2, 3]` and block_size of 2:
-   * <pre>{@code
-   * x = [[[[1, 2, 3], [4, 5, 6]],
-   *       [[7, 8, 9], [10, 11, 12]]]]
-   * }</pre>
-   * The output tensor has shape `[4, 1, 1, 3]` and value:
-   * <pre>{@code
-   * [[[[1, 2, 3]]], [[[4, 5, 6]]], [[[7, 8, 9]]], [[[10, 11, 12]]]]
-   * }</pre>
-   * (3) For the following input of shape `[1, 4, 4, 1]` and block_size of 2:
-   * <pre>{@code
-   * x = [[[[1],   [2],  [3],  [4]],
-   *       [[5],   [6],  [7],  [8]],
-   *       [[9],  [10], [11],  [12]],
-   *       [[13], [14], [15],  [16]]]]
-   * }</pre>
-   * The output tensor has shape `[4, 2, 2, 1]` and value:
-   * <pre>{@code
-   * x = [[[[1], [3]], [[9], [11]]],
-   *      [[[2], [4]], [[10], [12]]],
-   *      [[[5], [7]], [[13], [15]]],
-   *      [[[6], [8]], [[14], [16]]]]
-   * }</pre>
-   * (4) For the following input of shape `[2, 2, 4, 1]` and block_size of 2:
-   * <pre>{@code
-   * x = [[[[1],   [2],  [3],  [4]],
-   *       [[5],   [6],  [7],  [8]]],
-   *      [[[9],  [10], [11],  [12]],
-   *       [[13], [14], [15],  [16]]]]
-   * }</pre>
-   * The output tensor has shape `[8, 1, 2, 1]` and value:
-   * <pre>{@code
-   * x = [[[[1], [3]]], [[[9], [11]]], [[[2], [4]]], [[[10], [12]]],
-   *      [[[5], [7]]], [[[13], [15]]], [[[6], [8]]], [[[14], [16]]]]
-   * }</pre>
-   * Among others, this operation is useful for reducing atrous convolution into
-   * regular convolution.
-   * @param blockSize 
-   * @return a new instance of SpaceToBatch
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> SpaceToBatch<T> create(Scope scope, Operand<T> input, Operand<U> paddings, Long blockSize) {
-    OperationBuilder opBuilder = scope.env().opBuilder("SpaceToBatch", scope.makeOpName("SpaceToBatch"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder.addInput(paddings.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("block_size", blockSize);
-    return new SpaceToBatch<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "SpaceToBatch";
-  
-  private Output<T> output;
-  
-  private SpaceToBatch(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/SpaceToDepth.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/SpaceToDepth.java
deleted file mode 100644
index 404c97f889e..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/SpaceToDepth.java
+++ /dev/null
@@ -1,184 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.nn;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * SpaceToDepth for tensors of type T.
- * <p>
- * Rearranges blocks of spatial data, into depth. More specifically,
- * this op outputs a copy of the input tensor where values from the `height`
- * and `width` dimensions are moved to the `depth` dimension.
- * The attr `block_size` indicates the input block size.
- * <p>
- *   * Non-overlapping blocks of size `block_size x block size` are rearranged
- *     into depth at each location.
- *   * The depth of the output tensor is `block_size * block_size * input_depth`.
- *   * The Y, X coordinates within each block of the input become the high order
- *     component of the output channel index.
- *   * The input tensor's height and width must be divisible by block_size.
- * <p>
- * The `data_format` attr specifies the layout of the input and output tensors
- * with the following options:
- *   "NHWC": `[ batch, height, width, channels ]`
- *   "NCHW": `[ batch, channels, height, width ]`
- *   "NCHW_VECT_C":
- *       `qint8 [ batch, channels / 4, height, width, 4 ]`
- * <p>
- * It is useful to consider the operation as transforming a 6-D Tensor.
- * e.g. for data_format = NHWC,
- *      Each element in the input tensor can be specified via 6 coordinates,
- *      ordered by decreasing memory layout significance as:
- *      n,oY,bY,oX,bX,iC  (where n=batch index, oX, oY means X or Y coordinates
- *                         within the output image, bX, bY means coordinates
- *                         within the input block, iC means input channels).
- *      The output would be a transpose to the following layout:
- *      n,oY,oX,bY,bX,iC
- * <p>
- * This operation is useful for resizing the activations between convolutions
- * (but keeping all data), e.g. instead of pooling. It is also useful for training
- * purely convolutional models.
- * <p>
- * For example, given an input of shape `[1, 2, 2, 1]`, data_format = "NHWC" and
- * block_size = 2:
- * <pre>{@code
- * x = [[[[1], [2]],
- *       [[3], [4]]]]
- * }</pre>
- * This operation will output a tensor of shape `[1, 1, 1, 4]`:
- * <pre>{@code
- * [[[[1, 2, 3, 4]]]]
- * }</pre>
- * Here, the input has a batch of 1 and each batch element has shape `[2, 2, 1]`,
- * the corresponding output will have a single element (i.e. width and height are
- * both 1) and will have a depth of 4 channels (1 * block_size * block_size).
- * The output element shape is `[1, 1, 4]`.
- * <p>
- * For an input tensor with larger depth, here of shape `[1, 2, 2, 3]`, e.g.
- * <pre>{@code
- * x = [[[[1, 2, 3], [4, 5, 6]],
- *       [[7, 8, 9], [10, 11, 12]]]]
- * }</pre>
- * This operation, for block_size of 2, will return the following tensor of shape
- * `[1, 1, 1, 12]`
- * <pre>{@code
- * [[[[1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12]]]]
- * }</pre>
- * Similarly, for the following input of shape `[1 4 4 1]`, and a block size of 2:
- * <pre>{@code
- * x = [[[[1],   [2],  [5],  [6]],
- *       [[3],   [4],  [7],  [8]],
- *       [[9],  [10], [13],  [14]],
- *       [[11], [12], [15],  [16]]]]
- * }</pre>
- * the operator will return the following tensor of shape `[1 2 2 4]`:
- * <pre>{@code
- * x = [[[[1, 2, 3, 4],
- *        [5, 6, 7, 8]],
- *       [[9, 10, 11, 12],
- *        [13, 14, 15, 16]]]]
- * }</pre>
- * 
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator(group = "nn")
-public final class SpaceToDepth<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.nn.SpaceToDepth}
-   */
-  public static class Options {
-    
-    /**
-     * @param dataFormat 
-     */
-    public Options dataFormat(String dataFormat) {
-      this.dataFormat = dataFormat;
-      return this;
-    }
-    
-    private String dataFormat;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new SpaceToDepth operation.
-   * 
-   * @param scope current scope
-   * @param input 
-   * @param blockSize The size of the spatial block.
-   * @param options carries optional attributes values
-   * @return a new instance of SpaceToDepth
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> SpaceToDepth<T> create(Scope scope, Operand<T> input, Long blockSize, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("SpaceToDepth", scope.makeOpName("SpaceToDepth"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("block_size", blockSize);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.dataFormat != null) {
-          opBuilder.setAttr("data_format", opts.dataFormat);
-        }
-      }
-    }
-    return new SpaceToDepth<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param dataFormat 
-   */
-  public static Options dataFormat(String dataFormat) {
-    return new Options().dataFormat(dataFormat);
-  }
-  
-  /**
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "SpaceToDepth";
-  
-  private Output<T> output;
-  
-  private SpaceToDepth(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/TopK.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/TopK.java
deleted file mode 100644
index cd87a1374bb..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/TopK.java
+++ /dev/null
@@ -1,130 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.nn;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Finds values and indices of the `k` largest elements for the last dimension.
- * <p>
- * If the input is a vector (rank-1), finds the `k` largest entries in the vector
- * and outputs their values and indices as vectors.  Thus `values[j]` is the
- * `j`-th largest entry in `input`, and its index is `indices[j]`.
- * <p>
- * For matrices (resp. higher rank input), computes the top `k` entries in each
- * row (resp. vector along the last dimension).  Thus,
- * <p>
- *     values.shape = indices.shape = input.shape[:-1] + [k]
- * <p>
- * If two elements are equal, the lower-index element appears first.
- * 
- * @param <T> data type for {@code values()} output
- */
-@Operator(group = "nn")
-public final class TopK<T extends TNumber> extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.nn.TopK}
-   */
-  public static class Options {
-    
-    /**
-     * @param sorted If true the resulting `k` elements will be sorted by the values in
-     * descending order.
-     */
-    public Options sorted(Boolean sorted) {
-      this.sorted = sorted;
-      return this;
-    }
-    
-    private Boolean sorted;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new TopK operation.
-   * 
-   * @param scope current scope
-   * @param input 1-D or higher with last dimension at least `k`.
-   * @param k 0-D.  Number of top elements to look for along the last dimension (along each
-   * row for matrices).
-   * @param options carries optional attributes values
-   * @return a new instance of TopK
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> TopK<T> create(Scope scope, Operand<T> input, Operand<TInt32> k, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("TopKV2", scope.makeOpName("TopK"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder.addInput(k.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.sorted != null) {
-          opBuilder.setAttr("sorted", opts.sorted);
-        }
-      }
-    }
-    return new TopK<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param sorted If true the resulting `k` elements will be sorted by the values in
-   * descending order.
-   */
-  public static Options sorted(Boolean sorted) {
-    return new Options().sorted(sorted);
-  }
-  
-  /**
-   * The `k` largest elements along each last dimensional slice.
-   */
-  public Output<T> values() {
-    return values;
-  }
-  
-  /**
-   * The indices of `values` within the last dimension of `input`.
-   */
-  public Output<TInt32> indices() {
-    return indices;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "TopKV2";
-  
-  private Output<T> values;
-  private Output<TInt32> indices;
-  
-  private TopK(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    values = operation.output(outputIdx++);
-    indices = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/raw/SoftmaxCrossEntropyWithLogits.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/raw/SoftmaxCrossEntropyWithLogits.java
deleted file mode 100644
index 7f400f6650c..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/raw/SoftmaxCrossEntropyWithLogits.java
+++ /dev/null
@@ -1,85 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.nn.raw;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Computes softmax cross entropy cost and gradients to backpropagate.
- * <p>
- * Inputs are the logits, not probabilities.
- * 
- * @param <T> data type for {@code loss()} output
- */
-@Operator(group = "nn.raw")
-public final class SoftmaxCrossEntropyWithLogits<T extends TNumber> extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new SoftmaxCrossEntropyWithLogits operation.
-   * 
-   * @param scope current scope
-   * @param features batch_size x num_classes matrix
-   * @param labels batch_size x num_classes matrix
-   * The caller must ensure that each batch of labels represents a valid
-   * probability distribution.
-   * @return a new instance of SoftmaxCrossEntropyWithLogits
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> SoftmaxCrossEntropyWithLogits<T> create(Scope scope, Operand<T> features, Operand<T> labels) {
-    OperationBuilder opBuilder = scope.env().opBuilder("SoftmaxCrossEntropyWithLogits", scope.makeOpName("SoftmaxCrossEntropyWithLogits"));
-    opBuilder.addInput(features.asOutput(scope));
-    opBuilder.addInput(labels.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new SoftmaxCrossEntropyWithLogits<T>(opBuilder.build());
-  }
-  
-  /**
-   * Per example loss (batch_size vector).
-   */
-  public Output<T> loss() {
-    return loss;
-  }
-  
-  /**
-   * backpropagated gradients (batch_size x num_classes matrix).
-   */
-  public Output<T> backprop() {
-    return backprop;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "SoftmaxCrossEntropyWithLogits";
-  
-  private Output<T> loss;
-  private Output<T> backprop;
-  
-  private SoftmaxCrossEntropyWithLogits(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    loss = operation.output(outputIdx++);
-    backprop = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/raw/SparseSoftmaxCrossEntropyWithLogits.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/raw/SparseSoftmaxCrossEntropyWithLogits.java
deleted file mode 100644
index 2a2c4de517e..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/raw/SparseSoftmaxCrossEntropyWithLogits.java
+++ /dev/null
@@ -1,89 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.nn.raw;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Computes softmax cross entropy cost and gradients to backpropagate.
- * <p>
- * Unlike `SoftmaxCrossEntropyWithLogits`, this operation does not accept
- * a matrix of label probabilities, but rather a single label per row
- * of features.  This label is considered to have probability 1.0 for the
- * given row.
- * <p>
- * Inputs are the logits, not probabilities.
- * 
- * @param <T> data type for {@code loss()} output
- */
-@Operator(group = "nn.raw")
-public final class SparseSoftmaxCrossEntropyWithLogits<T extends TNumber> extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new SparseSoftmaxCrossEntropyWithLogits operation.
-   * 
-   * @param scope current scope
-   * @param features batch_size x num_classes matrix
-   * @param labels batch_size vector with values in [0, num_classes).
-   * This is the label for the given minibatch entry.
-   * @return a new instance of SparseSoftmaxCrossEntropyWithLogits
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber, U extends TNumber> SparseSoftmaxCrossEntropyWithLogits<T> create(Scope scope, Operand<T> features, Operand<U> labels) {
-    OperationBuilder opBuilder = scope.env().opBuilder("SparseSoftmaxCrossEntropyWithLogits", scope.makeOpName("SparseSoftmaxCrossEntropyWithLogits"));
-    opBuilder.addInput(features.asOutput(scope));
-    opBuilder.addInput(labels.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new SparseSoftmaxCrossEntropyWithLogits<T>(opBuilder.build());
-  }
-  
-  /**
-   * Per example loss (batch_size vector).
-   */
-  public Output<T> loss() {
-    return loss;
-  }
-  
-  /**
-   * backpropagated gradients (batch_size x num_classes matrix).
-   */
-  public Output<T> backprop() {
-    return backprop;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "SparseSoftmaxCrossEntropyWithLogits";
-  
-  private Output<T> loss;
-  private Output<T> backprop;
-  
-  private SparseSoftmaxCrossEntropyWithLogits(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    loss = operation.output(outputIdx++);
-    backprop = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/Dequantize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/Dequantize.java
deleted file mode 100644
index ce4071e349c..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/Dequantize.java
+++ /dev/null
@@ -1,219 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.quantization;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * Dequantize the 'input' tensor into a float or bfloat16 Tensor.
- * <p>
- * [min_range, max_range] are scalar floats that specify the range for
- * the output. The 'mode' attribute controls exactly which calculations are
- * used to convert the float values to their quantized equivalents.
- * <p>
- * In 'MIN_COMBINED' mode, each value of the tensor will undergo the following:
- * <pre>{@code
- * if T == qint8: in[i] += (range(T) + 1)/ 2.0
- * out[i] = min_range + (in[i]* (max_range - min_range) / range(T))
- * }</pre>
- * here `range(T) = numeric_limits<T>::max() - numeric_limits<T>::min()`
- * <p>
- * <i>MIN_COMBINED Mode Example</i>
- * <p>
- * If the input comes from a QuantizedRelu6, the output type is
- * quint8 (range of 0-255) but the possible range of QuantizedRelu6 is
- * 0-6.  The min_range and max_range values are therefore 0.0 and 6.0.
- * Dequantize on quint8 will take each value, cast to float, and multiply
- * by 6 / 255.
- * Note that if quantizedtype is qint8, the operation will additionally add
- * each value by 128 prior to casting.
- * <p>
- * If the mode is 'MIN_FIRST', then this approach is used:
- * <pre>{@code
- * num_discrete_values = 1 << (# of bits in T)
- * range_adjust = num_discrete_values / (num_discrete_values - 1)
- * range = (range_max - range_min) * range_adjust
- * range_scale = range / num_discrete_values
- * const double offset_input = static_cast<double>(input) - lowest_quantized;
- * result = range_min + ((input - numeric_limits<T>::min()) * range_scale)
- * }</pre>
- * If the mode is `SCALED`, dequantization is performed by multiplying each
- * input value by a scaling_factor. (Thus an input of 0 always maps to 0.0).
- * <p>
- * The scaling_factor is determined from `min_range`, `max_range`, and
- * `narrow_range` in a way that is compatible with `QuantizeAndDequantize{V2|V3}`
- * and `QuantizeV2`, using the following algorithm:
- * <pre>{@code
- *   const int min_expected_T = std::numeric_limits<T>::min() +
- *     (narrow_range ? 1 : 0);
- *   const int max_expected_T = std::numeric_limits<T>::max();
- *   const float max_expected_T = std::numeric_limits<float>::max();
- * 
- *   const float scale_factor =
- *     (std::numeric_limits<T>::min() == 0) ? (max_range / max_expected_T)
- *                                          : std::max(min_range / min_expected_T,
- *                                                     max_range / max_expected_T);
- * }</pre>
- * 
- * 
- * @param <U> data type for {@code output()} output
- */
-@Operator(group = "quantization")
-public final class Dequantize<U extends TNumber> extends RawOp implements Operand<U> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.quantization.Dequantize}
-   */
-  public static class Options {
-    
-    /**
-     * @param mode 
-     */
-    public Options mode(String mode) {
-      this.mode = mode;
-      return this;
-    }
-    
-    /**
-     * @param narrowRange 
-     */
-    public Options narrowRange(Boolean narrowRange) {
-      this.narrowRange = narrowRange;
-      return this;
-    }
-    
-    /**
-     * @param axis 
-     */
-    public Options axis(Long axis) {
-      this.axis = axis;
-      return this;
-    }
-    
-    private String mode;
-    private Boolean narrowRange;
-    private Long axis;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new Dequantize operation.
-   * 
-   * @param scope current scope
-   * @param input 
-   * @param minRange The minimum scalar value possibly produced for the input.
-   * @param maxRange The maximum scalar value possibly produced for the input.
-   * @param dtype Type of the output tensor. Currently Dequantize supports float and bfloat16.
-   * If 'dtype' is 'bfloat16', it only supports 'MIN_COMBINED' mode.
-   * @param options carries optional attributes values
-   * @return a new instance of Dequantize
-   */
-  @Endpoint(describeByClass = true)
-  public static <U extends TNumber, T extends TType> Dequantize<U> create(Scope scope, Operand<T> input, Operand<TFloat32> minRange, Operand<TFloat32> maxRange, Class<U> dtype, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Dequantize", scope.makeOpName("Dequantize"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder.addInput(minRange.asOutput(scope));
-    opBuilder.addInput(maxRange.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("dtype", dtype);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.mode != null) {
-          opBuilder.setAttr("mode", opts.mode);
-        }
-        if (opts.narrowRange != null) {
-          opBuilder.setAttr("narrow_range", opts.narrowRange);
-        }
-        if (opts.axis != null) {
-          opBuilder.setAttr("axis", opts.axis);
-        }
-      }
-    }
-    return new Dequantize<U>(opBuilder.build());
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new Dequantize operation using default output types.
-   * 
-   * @param scope current scope
-   * @param input 
-   * @param minRange The minimum scalar value possibly produced for the input.
-   * @param maxRange The maximum scalar value possibly produced for the input.
-   * @param options carries optional attributes values
-   * @return a new instance of Dequantize
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> Dequantize<TFloat32> create(Scope scope, Operand<T> input, Operand<TFloat32> minRange, Operand<TFloat32> maxRange, Options... options) {
-    return create(scope, input, minRange, maxRange, TFloat32.class, options);
-  }
-  
-  /**
-   * @param mode 
-   */
-  public static Options mode(String mode) {
-    return new Options().mode(mode);
-  }
-  
-  /**
-   * @param narrowRange 
-   */
-  public static Options narrowRange(Boolean narrowRange) {
-    return new Options().narrowRange(narrowRange);
-  }
-  
-  /**
-   * @param axis 
-   */
-  public static Options axis(Long axis) {
-    return new Options().axis(axis);
-  }
-  
-  /**
-   */
-  public Output<U> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<U> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Dequantize";
-  
-  private Output<U> output;
-  
-  private Dequantize(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/FakeQuantWithMinMaxArgs.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/FakeQuantWithMinMaxArgs.java
deleted file mode 100644
index 3cbac1bdf23..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/FakeQuantWithMinMaxArgs.java
+++ /dev/null
@@ -1,196 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.quantization;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-
-/**
- * Fake-quantize the 'inputs' tensor, type float to 'outputs' tensor of same type.
- * <p>
- * Attributes
- * <ul>
- * <li>
- * `[min; max]` define the clamping range for the `inputs` data.
- * </li>
- * <li>
- * `inputs` values are quantized into the quantization range (
- * `[0; 2^num_bits - 1]` when `narrow_range` is false and `[1; 2^num_bits - 1]`
- * when it is true) and then de-quantized and output as floats in `[min; max]`
- * interval.
- * </li>
- * <li>
- * `num_bits` is the bitwidth of the quantization; between 2 and 16, inclusive.
- * </li>
- * </ul>
- * Before quantization, `min` and `max` values are adjusted with the following
- * logic.
- * It is suggested to have `min <= 0 <= max`. If `0` is not in the range of values,
- * the behavior can be unexpected:
- * <ul>
- * <li>
- * If `0 < min < max`: `min_adj = 0` and `max_adj = max - min`.
- * </li>
- * <li>
- * If `min < max < 0`: `min_adj = min - max` and `max_adj = 0`.
- * </li>
- * <li>
- * If `min <= 0 <= max`: `scale = (max - min) / (2^num_bits - 1) `,
- * `min_adj = scale * round(min / scale)` and `max_adj = max + min_adj - min`.
- * </li>
- * </ul>
- * Quantization is called fake since the output is still in floating point.
- */
-@Operator(group = "quantization")
-public final class FakeQuantWithMinMaxArgs extends RawOp implements Operand<TFloat32> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.quantization.FakeQuantWithMinMaxArgs}
-   */
-  public static class Options {
-    
-    /**
-     * @param min 
-     */
-    public Options min(Float min) {
-      this.min = min;
-      return this;
-    }
-    
-    /**
-     * @param max 
-     */
-    public Options max(Float max) {
-      this.max = max;
-      return this;
-    }
-    
-    /**
-     * @param numBits 
-     */
-    public Options numBits(Long numBits) {
-      this.numBits = numBits;
-      return this;
-    }
-    
-    /**
-     * @param narrowRange 
-     */
-    public Options narrowRange(Boolean narrowRange) {
-      this.narrowRange = narrowRange;
-      return this;
-    }
-    
-    private Float min;
-    private Float max;
-    private Long numBits;
-    private Boolean narrowRange;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new FakeQuantWithMinMaxArgs operation.
-   * 
-   * @param scope current scope
-   * @param inputs 
-   * @param options carries optional attributes values
-   * @return a new instance of FakeQuantWithMinMaxArgs
-   */
-  @Endpoint(describeByClass = true)
-  public static FakeQuantWithMinMaxArgs create(Scope scope, Operand<TFloat32> inputs, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("FakeQuantWithMinMaxArgs", scope.makeOpName("FakeQuantWithMinMaxArgs"));
-    opBuilder.addInput(inputs.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.min != null) {
-          opBuilder.setAttr("min", opts.min);
-        }
-        if (opts.max != null) {
-          opBuilder.setAttr("max", opts.max);
-        }
-        if (opts.numBits != null) {
-          opBuilder.setAttr("num_bits", opts.numBits);
-        }
-        if (opts.narrowRange != null) {
-          opBuilder.setAttr("narrow_range", opts.narrowRange);
-        }
-      }
-    }
-    return new FakeQuantWithMinMaxArgs(opBuilder.build());
-  }
-  
-  /**
-   * @param min 
-   */
-  public static Options min(Float min) {
-    return new Options().min(min);
-  }
-  
-  /**
-   * @param max 
-   */
-  public static Options max(Float max) {
-    return new Options().max(max);
-  }
-  
-  /**
-   * @param numBits 
-   */
-  public static Options numBits(Long numBits) {
-    return new Options().numBits(numBits);
-  }
-  
-  /**
-   * @param narrowRange 
-   */
-  public static Options narrowRange(Boolean narrowRange) {
-    return new Options().narrowRange(narrowRange);
-  }
-  
-  /**
-   */
-  public Output<TFloat32> outputs() {
-    return outputs;
-  }
-  
-  @Override
-  public Output<TFloat32> asOutput(Scope scope) {
-    return outputs;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "FakeQuantWithMinMaxArgs";
-  
-  private Output<TFloat32> outputs;
-  
-  private FakeQuantWithMinMaxArgs(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    outputs = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/FakeQuantWithMinMaxArgsGradient.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/FakeQuantWithMinMaxArgsGradient.java
deleted file mode 100644
index 031d0d71a69..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/FakeQuantWithMinMaxArgsGradient.java
+++ /dev/null
@@ -1,167 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.quantization;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-
-/**
- * Compute gradients for a FakeQuantWithMinMaxArgs operation.
- */
-@Operator(group = "quantization")
-public final class FakeQuantWithMinMaxArgsGradient extends RawOp implements Operand<TFloat32> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.quantization.FakeQuantWithMinMaxArgsGradient}
-   */
-  public static class Options {
-    
-    /**
-     * @param min 
-     */
-    public Options min(Float min) {
-      this.min = min;
-      return this;
-    }
-    
-    /**
-     * @param max 
-     */
-    public Options max(Float max) {
-      this.max = max;
-      return this;
-    }
-    
-    /**
-     * @param numBits 
-     */
-    public Options numBits(Long numBits) {
-      this.numBits = numBits;
-      return this;
-    }
-    
-    /**
-     * @param narrowRange 
-     */
-    public Options narrowRange(Boolean narrowRange) {
-      this.narrowRange = narrowRange;
-      return this;
-    }
-    
-    private Float min;
-    private Float max;
-    private Long numBits;
-    private Boolean narrowRange;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new FakeQuantWithMinMaxArgsGradient operation.
-   * 
-   * @param scope current scope
-   * @param gradients Backpropagated gradients above the FakeQuantWithMinMaxArgs operation.
-   * @param inputs Values passed as inputs to the FakeQuantWithMinMaxArgs operation.
-   * @param options carries optional attributes values
-   * @return a new instance of FakeQuantWithMinMaxArgsGradient
-   */
-  @Endpoint(describeByClass = true)
-  public static FakeQuantWithMinMaxArgsGradient create(Scope scope, Operand<TFloat32> gradients, Operand<TFloat32> inputs, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("FakeQuantWithMinMaxArgsGradient", scope.makeOpName("FakeQuantWithMinMaxArgsGradient"));
-    opBuilder.addInput(gradients.asOutput(scope));
-    opBuilder.addInput(inputs.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.min != null) {
-          opBuilder.setAttr("min", opts.min);
-        }
-        if (opts.max != null) {
-          opBuilder.setAttr("max", opts.max);
-        }
-        if (opts.numBits != null) {
-          opBuilder.setAttr("num_bits", opts.numBits);
-        }
-        if (opts.narrowRange != null) {
-          opBuilder.setAttr("narrow_range", opts.narrowRange);
-        }
-      }
-    }
-    return new FakeQuantWithMinMaxArgsGradient(opBuilder.build());
-  }
-  
-  /**
-   * @param min 
-   */
-  public static Options min(Float min) {
-    return new Options().min(min);
-  }
-  
-  /**
-   * @param max 
-   */
-  public static Options max(Float max) {
-    return new Options().max(max);
-  }
-  
-  /**
-   * @param numBits 
-   */
-  public static Options numBits(Long numBits) {
-    return new Options().numBits(numBits);
-  }
-  
-  /**
-   * @param narrowRange 
-   */
-  public static Options narrowRange(Boolean narrowRange) {
-    return new Options().narrowRange(narrowRange);
-  }
-  
-  /**
-   * Backpropagated gradients below the FakeQuantWithMinMaxArgs operation:
-   * `gradients * (inputs >= min && inputs <= max)`.
-   */
-  public Output<TFloat32> backprops() {
-    return backprops;
-  }
-  
-  @Override
-  public Output<TFloat32> asOutput(Scope scope) {
-    return backprops;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "FakeQuantWithMinMaxArgsGradient";
-  
-  private Output<TFloat32> backprops;
-  
-  private FakeQuantWithMinMaxArgsGradient(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    backprops = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/FakeQuantWithMinMaxVars.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/FakeQuantWithMinMaxVars.java
deleted file mode 100644
index 53bdcc5dc0f..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/FakeQuantWithMinMaxVars.java
+++ /dev/null
@@ -1,166 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.quantization;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-
-/**
- * Fake-quantize the 'inputs' tensor of type float via global float scalars
- * <p>
- * Fake-quantize the `inputs` tensor of type float via global float scalars
- * `min` and `max` to `outputs` tensor of same shape as `inputs`.
- * <p>
- * Attributes
- * <ul>
- * <li>
- * `[min; max]` define the clamping range for the `inputs` data.
- * </li>
- * <li>
- * `inputs` values are quantized into the quantization range (
- * `[0; 2^num_bits - 1]` when `narrow_range` is false and `[1; 2^num_bits - 1]`
- * when it is true) and then de-quantized and output as floats in `[min; max]`
- * interval.
- * </li>
- * <li>
- * `num_bits` is the bitwidth of the quantization; between 2 and 16, inclusive.
- * </li>
- * </ul>
- * Before quantization, `min` and `max` values are adjusted with the following
- * logic.
- * It is suggested to have `min <= 0 <= max`. If `0` is not in the range of values,
- * the behavior can be unexpected:
- * <ul>
- * <li>
- * If `0 < min < max`: `min_adj = 0` and `max_adj = max - min`.
- * </li>
- * <li>
- * If `min < max < 0`: `min_adj = min - max` and `max_adj = 0`.
- * </li>
- * <li>
- * If `min <= 0 <= max`: `scale = (max - min) / (2^num_bits - 1) `,
- * `min_adj = scale * round(min / scale)` and `max_adj = max + min_adj - min`.
- * </li>
- * </ul>
- * This operation has a gradient and thus allows for training `min` and `max`
- * values.
- */
-@Operator(group = "quantization")
-public final class FakeQuantWithMinMaxVars extends RawOp implements Operand<TFloat32> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.quantization.FakeQuantWithMinMaxVars}
-   */
-  public static class Options {
-    
-    /**
-     * @param numBits 
-     */
-    public Options numBits(Long numBits) {
-      this.numBits = numBits;
-      return this;
-    }
-    
-    /**
-     * @param narrowRange 
-     */
-    public Options narrowRange(Boolean narrowRange) {
-      this.narrowRange = narrowRange;
-      return this;
-    }
-    
-    private Long numBits;
-    private Boolean narrowRange;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new FakeQuantWithMinMaxVars operation.
-   * 
-   * @param scope current scope
-   * @param inputs 
-   * @param min 
-   * @param max 
-   * @param options carries optional attributes values
-   * @return a new instance of FakeQuantWithMinMaxVars
-   */
-  @Endpoint(describeByClass = true)
-  public static FakeQuantWithMinMaxVars create(Scope scope, Operand<TFloat32> inputs, Operand<TFloat32> min, Operand<TFloat32> max, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("FakeQuantWithMinMaxVars", scope.makeOpName("FakeQuantWithMinMaxVars"));
-    opBuilder.addInput(inputs.asOutput(scope));
-    opBuilder.addInput(min.asOutput(scope));
-    opBuilder.addInput(max.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.numBits != null) {
-          opBuilder.setAttr("num_bits", opts.numBits);
-        }
-        if (opts.narrowRange != null) {
-          opBuilder.setAttr("narrow_range", opts.narrowRange);
-        }
-      }
-    }
-    return new FakeQuantWithMinMaxVars(opBuilder.build());
-  }
-  
-  /**
-   * @param numBits 
-   */
-  public static Options numBits(Long numBits) {
-    return new Options().numBits(numBits);
-  }
-  
-  /**
-   * @param narrowRange 
-   */
-  public static Options narrowRange(Boolean narrowRange) {
-    return new Options().narrowRange(narrowRange);
-  }
-  
-  /**
-   */
-  public Output<TFloat32> outputs() {
-    return outputs;
-  }
-  
-  @Override
-  public Output<TFloat32> asOutput(Scope scope) {
-    return outputs;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "FakeQuantWithMinMaxVars";
-  
-  private Output<TFloat32> outputs;
-  
-  private FakeQuantWithMinMaxVars(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    outputs = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/FakeQuantWithMinMaxVarsGradient.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/FakeQuantWithMinMaxVarsGradient.java
deleted file mode 100644
index f487a5bb806..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/FakeQuantWithMinMaxVarsGradient.java
+++ /dev/null
@@ -1,149 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.quantization;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-
-/**
- * Compute gradients for a FakeQuantWithMinMaxVars operation.
- */
-@Operator(group = "quantization")
-public final class FakeQuantWithMinMaxVarsGradient extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.quantization.FakeQuantWithMinMaxVarsGradient}
-   */
-  public static class Options {
-    
-    /**
-     * @param numBits The bitwidth of the quantization; between 2 and 8, inclusive.
-     */
-    public Options numBits(Long numBits) {
-      this.numBits = numBits;
-      return this;
-    }
-    
-    /**
-     * @param narrowRange Whether to quantize into 2^num_bits - 1 distinct values.
-     */
-    public Options narrowRange(Boolean narrowRange) {
-      this.narrowRange = narrowRange;
-      return this;
-    }
-    
-    private Long numBits;
-    private Boolean narrowRange;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new FakeQuantWithMinMaxVarsGradient operation.
-   * 
-   * @param scope current scope
-   * @param gradients Backpropagated gradients above the FakeQuantWithMinMaxVars operation.
-   * @param inputs Values passed as inputs to the FakeQuantWithMinMaxVars operation.
-   * min, max: Quantization interval, scalar floats.
-   * @param min 
-   * @param max 
-   * @param options carries optional attributes values
-   * @return a new instance of FakeQuantWithMinMaxVarsGradient
-   */
-  @Endpoint(describeByClass = true)
-  public static FakeQuantWithMinMaxVarsGradient create(Scope scope, Operand<TFloat32> gradients, Operand<TFloat32> inputs, Operand<TFloat32> min, Operand<TFloat32> max, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("FakeQuantWithMinMaxVarsGradient", scope.makeOpName("FakeQuantWithMinMaxVarsGradient"));
-    opBuilder.addInput(gradients.asOutput(scope));
-    opBuilder.addInput(inputs.asOutput(scope));
-    opBuilder.addInput(min.asOutput(scope));
-    opBuilder.addInput(max.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.numBits != null) {
-          opBuilder.setAttr("num_bits", opts.numBits);
-        }
-        if (opts.narrowRange != null) {
-          opBuilder.setAttr("narrow_range", opts.narrowRange);
-        }
-      }
-    }
-    return new FakeQuantWithMinMaxVarsGradient(opBuilder.build());
-  }
-  
-  /**
-   * @param numBits The bitwidth of the quantization; between 2 and 8, inclusive.
-   */
-  public static Options numBits(Long numBits) {
-    return new Options().numBits(numBits);
-  }
-  
-  /**
-   * @param narrowRange Whether to quantize into 2^num_bits - 1 distinct values.
-   */
-  public static Options narrowRange(Boolean narrowRange) {
-    return new Options().narrowRange(narrowRange);
-  }
-  
-  /**
-   * Backpropagated gradients w.r.t. inputs:
-   * `gradients * (inputs >= min && inputs <= max)`.
-   */
-  public Output<TFloat32> backpropsWrtInput() {
-    return backpropsWrtInput;
-  }
-  
-  /**
-   * Backpropagated gradients w.r.t. min parameter:
-   * `sum(gradients * (inputs < min))`.
-   */
-  public Output<TFloat32> backpropWrtMin() {
-    return backpropWrtMin;
-  }
-  
-  /**
-   * Backpropagated gradients w.r.t. max parameter:
-   * `sum(gradients * (inputs > max))`.
-   */
-  public Output<TFloat32> backpropWrtMax() {
-    return backpropWrtMax;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "FakeQuantWithMinMaxVarsGradient";
-  
-  private Output<TFloat32> backpropsWrtInput;
-  private Output<TFloat32> backpropWrtMin;
-  private Output<TFloat32> backpropWrtMax;
-  
-  private FakeQuantWithMinMaxVarsGradient(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    backpropsWrtInput = operation.output(outputIdx++);
-    backpropWrtMin = operation.output(outputIdx++);
-    backpropWrtMax = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/FakeQuantWithMinMaxVarsPerChannel.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/FakeQuantWithMinMaxVarsPerChannel.java
deleted file mode 100644
index 62a5fbe4554..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/FakeQuantWithMinMaxVarsPerChannel.java
+++ /dev/null
@@ -1,167 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.quantization;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-
-/**
- * Fake-quantize the 'inputs' tensor of type float via per-channel floats
- * <p>
- * Fake-quantize the `inputs` tensor of type float per-channel and one of the
- * shapes: `[d]`, `[b, d]` `[b, h, w, d]` via per-channel floats `min` and `max`
- * of shape `[d]` to `outputs` tensor of same shape as `inputs`.
- * <p>
- * Attributes
- * <ul>
- * <li>
- * `[min; max]` define the clamping range for the `inputs` data.
- * </li>
- * <li>
- * `inputs` values are quantized into the quantization range (
- * `[0; 2^num_bits - 1]` when `narrow_range` is false and `[1; 2^num_bits - 1]`
- * when it is true) and then de-quantized and output as floats in `[min; max]`
- * interval.
- * </li>
- * <li>
- * `num_bits` is the bitwidth of the quantization; between 2 and 16, inclusive.
- * </li>
- * </ul>
- * Before quantization, `min` and `max` values are adjusted with the following
- * logic.
- * It is suggested to have `min <= 0 <= max`. If `0` is not in the range of values,
- * the behavior can be unexpected:
- * <ul>
- * <li>
- * If `0 < min < max`: `min_adj = 0` and `max_adj = max - min`.
- * </li>
- * <li>
- * If `min < max < 0`: `min_adj = min - max` and `max_adj = 0`.
- * </li>
- * <li>
- * If `min <= 0 <= max`: `scale = (max - min) / (2^num_bits - 1) `,
- * `min_adj = scale * round(min / scale)` and `max_adj = max + min_adj - min`.
- * </li>
- * </ul>
- * This operation has a gradient and thus allows for training `min` and `max`
- * values.
- */
-@Operator(group = "quantization")
-public final class FakeQuantWithMinMaxVarsPerChannel extends RawOp implements Operand<TFloat32> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.quantization.FakeQuantWithMinMaxVarsPerChannel}
-   */
-  public static class Options {
-    
-    /**
-     * @param numBits 
-     */
-    public Options numBits(Long numBits) {
-      this.numBits = numBits;
-      return this;
-    }
-    
-    /**
-     * @param narrowRange 
-     */
-    public Options narrowRange(Boolean narrowRange) {
-      this.narrowRange = narrowRange;
-      return this;
-    }
-    
-    private Long numBits;
-    private Boolean narrowRange;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new FakeQuantWithMinMaxVarsPerChannel operation.
-   * 
-   * @param scope current scope
-   * @param inputs 
-   * @param min 
-   * @param max 
-   * @param options carries optional attributes values
-   * @return a new instance of FakeQuantWithMinMaxVarsPerChannel
-   */
-  @Endpoint(describeByClass = true)
-  public static FakeQuantWithMinMaxVarsPerChannel create(Scope scope, Operand<TFloat32> inputs, Operand<TFloat32> min, Operand<TFloat32> max, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("FakeQuantWithMinMaxVarsPerChannel", scope.makeOpName("FakeQuantWithMinMaxVarsPerChannel"));
-    opBuilder.addInput(inputs.asOutput(scope));
-    opBuilder.addInput(min.asOutput(scope));
-    opBuilder.addInput(max.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.numBits != null) {
-          opBuilder.setAttr("num_bits", opts.numBits);
-        }
-        if (opts.narrowRange != null) {
-          opBuilder.setAttr("narrow_range", opts.narrowRange);
-        }
-      }
-    }
-    return new FakeQuantWithMinMaxVarsPerChannel(opBuilder.build());
-  }
-  
-  /**
-   * @param numBits 
-   */
-  public static Options numBits(Long numBits) {
-    return new Options().numBits(numBits);
-  }
-  
-  /**
-   * @param narrowRange 
-   */
-  public static Options narrowRange(Boolean narrowRange) {
-    return new Options().narrowRange(narrowRange);
-  }
-  
-  /**
-   */
-  public Output<TFloat32> outputs() {
-    return outputs;
-  }
-  
-  @Override
-  public Output<TFloat32> asOutput(Scope scope) {
-    return outputs;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "FakeQuantWithMinMaxVarsPerChannel";
-  
-  private Output<TFloat32> outputs;
-  
-  private FakeQuantWithMinMaxVarsPerChannel(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    outputs = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/FakeQuantWithMinMaxVarsPerChannelGradient.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/FakeQuantWithMinMaxVarsPerChannelGradient.java
deleted file mode 100644
index 02fd3200c68..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/FakeQuantWithMinMaxVarsPerChannelGradient.java
+++ /dev/null
@@ -1,152 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.quantization;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-
-/**
- * Compute gradients for a FakeQuantWithMinMaxVarsPerChannel operation.
- */
-@Operator(group = "quantization")
-public final class FakeQuantWithMinMaxVarsPerChannelGradient extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.quantization.FakeQuantWithMinMaxVarsPerChannelGradient}
-   */
-  public static class Options {
-    
-    /**
-     * @param numBits The bitwidth of the quantization; between 2 and 16, inclusive.
-     */
-    public Options numBits(Long numBits) {
-      this.numBits = numBits;
-      return this;
-    }
-    
-    /**
-     * @param narrowRange Whether to quantize into 2^num_bits - 1 distinct values.
-     */
-    public Options narrowRange(Boolean narrowRange) {
-      this.narrowRange = narrowRange;
-      return this;
-    }
-    
-    private Long numBits;
-    private Boolean narrowRange;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new FakeQuantWithMinMaxVarsPerChannelGradient operation.
-   * 
-   * @param scope current scope
-   * @param gradients Backpropagated gradients above the FakeQuantWithMinMaxVars operation,
-   * shape one of: `[d]`, `[b, d]`,  `[b, h, w, d]`.
-   * @param inputs Values passed as inputs to the FakeQuantWithMinMaxVars operation, shape
-   *   same as `gradients`.
-   * min, max: Quantization interval, floats of shape `[d]`.
-   * @param min 
-   * @param max 
-   * @param options carries optional attributes values
-   * @return a new instance of FakeQuantWithMinMaxVarsPerChannelGradient
-   */
-  @Endpoint(describeByClass = true)
-  public static FakeQuantWithMinMaxVarsPerChannelGradient create(Scope scope, Operand<TFloat32> gradients, Operand<TFloat32> inputs, Operand<TFloat32> min, Operand<TFloat32> max, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("FakeQuantWithMinMaxVarsPerChannelGradient", scope.makeOpName("FakeQuantWithMinMaxVarsPerChannelGradient"));
-    opBuilder.addInput(gradients.asOutput(scope));
-    opBuilder.addInput(inputs.asOutput(scope));
-    opBuilder.addInput(min.asOutput(scope));
-    opBuilder.addInput(max.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.numBits != null) {
-          opBuilder.setAttr("num_bits", opts.numBits);
-        }
-        if (opts.narrowRange != null) {
-          opBuilder.setAttr("narrow_range", opts.narrowRange);
-        }
-      }
-    }
-    return new FakeQuantWithMinMaxVarsPerChannelGradient(opBuilder.build());
-  }
-  
-  /**
-   * @param numBits The bitwidth of the quantization; between 2 and 16, inclusive.
-   */
-  public static Options numBits(Long numBits) {
-    return new Options().numBits(numBits);
-  }
-  
-  /**
-   * @param narrowRange Whether to quantize into 2^num_bits - 1 distinct values.
-   */
-  public static Options narrowRange(Boolean narrowRange) {
-    return new Options().narrowRange(narrowRange);
-  }
-  
-  /**
-   * Backpropagated gradients w.r.t. inputs, shape same as
-   * `inputs`:
-   *   `gradients * (inputs >= min && inputs <= max)`.
-   */
-  public Output<TFloat32> backpropsWrtInput() {
-    return backpropsWrtInput;
-  }
-  
-  /**
-   * Backpropagated gradients w.r.t. min parameter, shape `[d]`:
-   * `sum_per_d(gradients * (inputs < min))`.
-   */
-  public Output<TFloat32> backpropWrtMin() {
-    return backpropWrtMin;
-  }
-  
-  /**
-   * Backpropagated gradients w.r.t. max parameter, shape `[d]`:
-   * `sum_per_d(gradients * (inputs > max))`.
-   */
-  public Output<TFloat32> backpropWrtMax() {
-    return backpropWrtMax;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "FakeQuantWithMinMaxVarsPerChannelGradient";
-  
-  private Output<TFloat32> backpropsWrtInput;
-  private Output<TFloat32> backpropWrtMin;
-  private Output<TFloat32> backpropWrtMax;
-  
-  private FakeQuantWithMinMaxVarsPerChannelGradient(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    backpropsWrtInput = operation.output(outputIdx++);
-    backpropWrtMin = operation.output(outputIdx++);
-    backpropWrtMax = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/Quantize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/Quantize.java
deleted file mode 100644
index 30c609e5b6d..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/Quantize.java
+++ /dev/null
@@ -1,327 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.quantization;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.family.TType;
-
-/**
- * Quantize the 'input' tensor of type float to 'output' tensor of type 'T'.
- * <p>
- * [min_range, max_range] are scalar floats that specify the range for
- * the 'input' data. The 'mode' attribute controls exactly which calculations are
- * used to convert the float values to their quantized equivalents.  The
- * 'round_mode' attribute controls which rounding tie-breaking algorithm is used
- * when rounding float values to their quantized equivalents.
- * <p>
- * In 'MIN_COMBINED' mode, each value of the tensor will undergo the following:
- * <pre>{@code
- * out[i] = (in[i] - min_range) * range(T) / (max_range - min_range)
- * if T == qint8: out[i] -= (range(T) + 1) / 2.0
- * }</pre>
- * here `range(T) = numeric_limits<T>::max() - numeric_limits<T>::min()`
- * <p>
- * <i>MIN_COMBINED Mode Example</i>
- * <p>
- * Assume the input is type float and has a possible range of [0.0, 6.0] and the
- * output type is quint8 ([0, 255]). The min_range and max_range values should be
- * specified as 0.0 and 6.0. Quantizing from float to quint8 will multiply each
- * value of the input by 255/6 and cast to quint8.
- * <p>
- * If the output type was qint8 ([-128, 127]), the operation will additionally
- * subtract each value by 128 prior to casting, so that the range of values aligns
- * with the range of qint8.
- * <p>
- * If the mode is 'MIN_FIRST', then this approach is used:
- * <pre>{@code
- * num_discrete_values = 1 << (# of bits in T)
- * range_adjust = num_discrete_values / (num_discrete_values - 1)
- * range = (range_max - range_min) * range_adjust
- * range_scale = num_discrete_values / range
- * quantized = round(input * range_scale) - round(range_min * range_scale) +
- *   numeric_limits<T>::min()
- * quantized = max(quantized, numeric_limits<T>::min())
- * quantized = min(quantized, numeric_limits<T>::max())
- * }</pre>
- * The biggest difference between this and MIN_COMBINED is that the minimum range
- * is rounded first, before it's subtracted from the rounded value. With
- * MIN_COMBINED, a small bias is introduced where repeated iterations of quantizing
- * and dequantizing will introduce a larger and larger error.
- * <p>
- * <i>SCALED mode Example</i>
- * <p>
- * `SCALED` mode matches the quantization approach used in
- * `QuantizeAndDequantize{V2|V3}`.
- * <p>
- * If the mode is `SCALED`, the quantization is performed by multiplying each
- * input value by a scaling_factor.
- * The scaling_factor is determined from `min_range` and `max_range` to be as large
- * as possible such that the range from `min_range` to `max_range` is representable
- * within values of type T.
- * <pre>{@code
- *   const int min_T = std::numeric_limits<T>::min();
- *   const int max_T = std::numeric_limits<T>::max();
- *   const float max_float = std::numeric_limits<float>::max();
- * 
- *   const float scale_factor_from_min_side =
- *       (min_T * min_range > 0) ? min_T / min_range : max_float;
- *   const float scale_factor_from_max_side =
- *       (max_T * max_range > 0) ? max_T / max_range : max_float;
- * 
- *   const float scale_factor = std::min(scale_factor_from_min_side,
- *                                       scale_factor_from_max_side);
- * }</pre>
- * We next use the scale_factor to adjust min_range and max_range as follows:
- * <pre>{@code
- *       min_range = min_T / scale_factor;
- *       max_range = max_T / scale_factor;
- * }</pre>
- * e.g. if T = qint8, and initially min_range = -10, and max_range = 9, we would
- * compare -128/-10.0 = 12.8 to 127/9.0 = 14.11, and set scaling_factor = 12.8
- * In this case, min_range would remain -10, but max_range would be adjusted to
- * 127 / 12.8 = 9.921875
- * <p>
- * So we will quantize input values in the range (-10, 9.921875) to (-128, 127).
- * <p>
- * The input tensor can now be quantized by clipping values to the range
- * `min_range` to `max_range`, then multiplying by scale_factor as follows:
- * <pre>{@code
- * result = round(min(max_range, max(min_range, input)) * scale_factor)
- * }</pre>
- * The adjusted `min_range` and `max_range` are returned as outputs 2 and 3 of
- * this operation. These outputs should be used as the range for any further
- * calculations.
- * <p>
- * <i>narrow_range (bool) attribute</i>
- * <p>
- * If true, we do not use the minimum quantized value.
- * i.e. for int8 the quantized output, it would be restricted to the range
- * -127..127 instead of the full -128..127 range.
- * This is provided for compatibility with certain inference backends.
- * (Only applies to SCALED mode)
- * <p>
- * <i>axis (int) attribute</i>
- * <p>
- * An optional `axis` attribute can specify a dimension index of the input tensor,
- * such that quantization ranges will be calculated and applied separately for each
- * slice of the tensor along that dimension. This is useful for per-channel
- * quantization.
- * <p>
- * If axis is specified, min_range and max_range
- * <p>
- * if `axis`=None, per-tensor quantization is performed as normal.
- * <p>
- * <i>ensure_minimum_range (float) attribute</i>
- * <p>
- * Ensures the minimum quantization range is at least this value.
- * The legacy default value for this is 0.01, but it is strongly suggested to
- * set it to 0 for new uses.
- * 
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator(group = "quantization")
-public final class Quantize<T extends TType> extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.quantization.Quantize}
-   */
-  public static class Options {
-    
-    /**
-     * @param mode 
-     */
-    public Options mode(String mode) {
-      this.mode = mode;
-      return this;
-    }
-    
-    /**
-     * @param roundMode 
-     */
-    public Options roundMode(String roundMode) {
-      this.roundMode = roundMode;
-      return this;
-    }
-    
-    /**
-     * @param narrowRange 
-     */
-    public Options narrowRange(Boolean narrowRange) {
-      this.narrowRange = narrowRange;
-      return this;
-    }
-    
-    /**
-     * @param axis 
-     */
-    public Options axis(Long axis) {
-      this.axis = axis;
-      return this;
-    }
-    
-    /**
-     * @param ensureMinimumRange 
-     */
-    public Options ensureMinimumRange(Float ensureMinimumRange) {
-      this.ensureMinimumRange = ensureMinimumRange;
-      return this;
-    }
-    
-    private String mode;
-    private String roundMode;
-    private Boolean narrowRange;
-    private Long axis;
-    private Float ensureMinimumRange;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new Quantize operation.
-   * 
-   * @param scope current scope
-   * @param input 
-   * @param minRange The minimum value of the quantization range. This value may be adjusted by the
-   * op depending on other parameters. The adjusted value is written to `output_min`.
-   * If the `axis` attribute is specified, this must be a 1-D tensor whose size
-   * matches the `axis` dimension of the input and output tensors.
-   * @param maxRange The maximum value of the quantization range. This value may be adjusted by the
-   * op depending on other parameters. The adjusted value is written to `output_max`.
-   * If the `axis` attribute is specified, this must be a 1-D tensor whose size
-   * matches the `axis` dimension of the input and output tensors.
-   * @param T 
-   * @param options carries optional attributes values
-   * @return a new instance of Quantize
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> Quantize<T> create(Scope scope, Operand<TFloat32> input, Operand<TFloat32> minRange, Operand<TFloat32> maxRange, Class<T> T, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("QuantizeV2", scope.makeOpName("Quantize"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder.addInput(minRange.asOutput(scope));
-    opBuilder.addInput(maxRange.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("T", T);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.mode != null) {
-          opBuilder.setAttr("mode", opts.mode);
-        }
-        if (opts.roundMode != null) {
-          opBuilder.setAttr("round_mode", opts.roundMode);
-        }
-        if (opts.narrowRange != null) {
-          opBuilder.setAttr("narrow_range", opts.narrowRange);
-        }
-        if (opts.axis != null) {
-          opBuilder.setAttr("axis", opts.axis);
-        }
-        if (opts.ensureMinimumRange != null) {
-          opBuilder.setAttr("ensure_minimum_range", opts.ensureMinimumRange);
-        }
-      }
-    }
-    return new Quantize<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param mode 
-   */
-  public static Options mode(String mode) {
-    return new Options().mode(mode);
-  }
-  
-  /**
-   * @param roundMode 
-   */
-  public static Options roundMode(String roundMode) {
-    return new Options().roundMode(roundMode);
-  }
-  
-  /**
-   * @param narrowRange 
-   */
-  public static Options narrowRange(Boolean narrowRange) {
-    return new Options().narrowRange(narrowRange);
-  }
-  
-  /**
-   * @param axis 
-   */
-  public static Options axis(Long axis) {
-    return new Options().axis(axis);
-  }
-  
-  /**
-   * @param ensureMinimumRange 
-   */
-  public static Options ensureMinimumRange(Float ensureMinimumRange) {
-    return new Options().ensureMinimumRange(ensureMinimumRange);
-  }
-  
-  /**
-   * The quantized data produced from the float input.
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  /**
-   * The final quantization range minimum, used to clip input values before scaling
-   * and rounding them to quantized values.
-   * If the `axis` attribute is specified, this will be a 1-D tensor whose size
-   * matches the `axis` dimension of the input and output tensors.
-   */
-  public Output<TFloat32> outputMin() {
-    return outputMin;
-  }
-  
-  /**
-   * The final quantization range maximum, used to clip input values before scaling
-   * and rounding them to quantized values.
-   * If the `axis` attribute is specified, this will be a 1-D tensor whose size
-   * matches the `axis` dimension of the input and output tensors.
-   */
-  public Output<TFloat32> outputMax() {
-    return outputMax;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "QuantizeV2";
-  
-  private Output<T> output;
-  private Output<TFloat32> outputMin;
-  private Output<TFloat32> outputMax;
-  
-  private Quantize(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-    outputMin = operation.output(outputIdx++);
-    outputMax = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/QuantizeAndDequantize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/QuantizeAndDequantize.java
deleted file mode 100644
index aefb818c77d..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/QuantizeAndDequantize.java
+++ /dev/null
@@ -1,175 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.quantization;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Quantizes then dequantizes a tensor.
- * <p>
- * This is almost identical to QuantizeAndDequantizeV2, except that num_bits is a
- * tensor, so its value can change during training.
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator(group = "quantization")
-public final class QuantizeAndDequantize<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.quantization.QuantizeAndDequantize}
-   */
-  public static class Options {
-    
-    /**
-     * @param signedInput 
-     */
-    public Options signedInput(Boolean signedInput) {
-      this.signedInput = signedInput;
-      return this;
-    }
-    
-    /**
-     * @param rangeGiven 
-     */
-    public Options rangeGiven(Boolean rangeGiven) {
-      this.rangeGiven = rangeGiven;
-      return this;
-    }
-    
-    /**
-     * @param narrowRange 
-     */
-    public Options narrowRange(Boolean narrowRange) {
-      this.narrowRange = narrowRange;
-      return this;
-    }
-    
-    /**
-     * @param axis 
-     */
-    public Options axis(Long axis) {
-      this.axis = axis;
-      return this;
-    }
-    
-    private Boolean signedInput;
-    private Boolean rangeGiven;
-    private Boolean narrowRange;
-    private Long axis;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new QuantizeAndDequantize operation.
-   * 
-   * @param scope current scope
-   * @param input 
-   * @param inputMin 
-   * @param inputMax 
-   * @param numBits 
-   * @param options carries optional attributes values
-   * @return a new instance of QuantizeAndDequantize
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> QuantizeAndDequantize<T> create(Scope scope, Operand<T> input, Operand<T> inputMin, Operand<T> inputMax, Operand<TInt32> numBits, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("QuantizeAndDequantizeV3", scope.makeOpName("QuantizeAndDequantize"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder.addInput(inputMin.asOutput(scope));
-    opBuilder.addInput(inputMax.asOutput(scope));
-    opBuilder.addInput(numBits.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.signedInput != null) {
-          opBuilder.setAttr("signed_input", opts.signedInput);
-        }
-        if (opts.rangeGiven != null) {
-          opBuilder.setAttr("range_given", opts.rangeGiven);
-        }
-        if (opts.narrowRange != null) {
-          opBuilder.setAttr("narrow_range", opts.narrowRange);
-        }
-        if (opts.axis != null) {
-          opBuilder.setAttr("axis", opts.axis);
-        }
-      }
-    }
-    return new QuantizeAndDequantize<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param signedInput 
-   */
-  public static Options signedInput(Boolean signedInput) {
-    return new Options().signedInput(signedInput);
-  }
-  
-  /**
-   * @param rangeGiven 
-   */
-  public static Options rangeGiven(Boolean rangeGiven) {
-    return new Options().rangeGiven(rangeGiven);
-  }
-  
-  /**
-   * @param narrowRange 
-   */
-  public static Options narrowRange(Boolean narrowRange) {
-    return new Options().narrowRange(narrowRange);
-  }
-  
-  /**
-   * @param axis 
-   */
-  public static Options axis(Long axis) {
-    return new Options().axis(axis);
-  }
-  
-  /**
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "QuantizeAndDequantizeV3";
-  
-  private Output<T> output;
-  
-  private QuantizeAndDequantize(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/QuantizeDownAndShrinkRange.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/QuantizeDownAndShrinkRange.java
deleted file mode 100644
index 15497432bed..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/QuantizeDownAndShrinkRange.java
+++ /dev/null
@@ -1,117 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.quantization;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.family.TType;
-
-/**
- * Convert the quantized 'input' tensor into a lower-precision 'output', using the
- * <p>
- * actual distribution of the values to maximize the usage of the lower bit depth
- * and adjusting the output min and max ranges accordingly.
- * <p>
- * [input_min, input_max] are scalar floats that specify the range for the float
- * interpretation of the 'input' data. For example, if input_min is -1.0f and
- * input_max is 1.0f, and we are dealing with quint16 quantized data, then a 0
- * value in the 16-bit data should be interpreted as -1.0f, and a 65535 means 1.0f.
- * <p>
- * This operator tries to squeeze as much precision as possible into an output with
- * a lower bit depth by calculating the actual min and max values found in the
- * data. For example, maybe that quint16 input has no values lower than 16,384 and
- * none higher than 49,152. That means only half the range is actually needed, all
- * the float interpretations are between -0.5f and 0.5f, so if we want to compress
- * the data into a quint8 output, we can use that range rather than the theoretical
- * -1.0f to 1.0f that is suggested by the input min and max.
- * <p>
- * In practice, this is most useful for taking output from operations like
- * QuantizedMatMul that can produce higher bit-depth outputs than their inputs and
- * may have large potential output ranges, but in practice have a distribution of
- * input values that only uses a small fraction of the possible range. By feeding
- * that output into this operator, we can reduce it from 32 bits down to 8 with
- * minimal loss of accuracy.
- * 
- * @param <U> data type for {@code output()} output
- */
-@Operator(group = "quantization")
-public final class QuantizeDownAndShrinkRange<U extends TType> extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new QuantizeDownAndShrinkRange operation.
-   * 
-   * @param scope current scope
-   * @param input 
-   * @param inputMin The float value that the minimum quantized input value represents.
-   * @param inputMax The float value that the maximum quantized input value represents.
-   * @param outType The type of the output. Should be a lower bit depth than Tinput.
-   * @return a new instance of QuantizeDownAndShrinkRange
-   */
-  @Endpoint(describeByClass = true)
-  public static <U extends TType, T extends TType> QuantizeDownAndShrinkRange<U> create(Scope scope, Operand<T> input, Operand<TFloat32> inputMin, Operand<TFloat32> inputMax, Class<U> outType) {
-    OperationBuilder opBuilder = scope.env().opBuilder("QuantizeDownAndShrinkRange", scope.makeOpName("QuantizeDownAndShrinkRange"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder.addInput(inputMin.asOutput(scope));
-    opBuilder.addInput(inputMax.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("out_type", outType);
-    return new QuantizeDownAndShrinkRange<U>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<U> output() {
-    return output;
-  }
-  
-  /**
-   * The float value that the minimum quantized output value represents.
-   */
-  public Output<TFloat32> outputMin() {
-    return outputMin;
-  }
-  
-  /**
-   * The float value that the maximum quantized output value represents.
-   */
-  public Output<TFloat32> outputMax() {
-    return outputMax;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "QuantizeDownAndShrinkRange";
-  
-  private Output<U> output;
-  private Output<TFloat32> outputMin;
-  private Output<TFloat32> outputMax;
-  
-  private QuantizeDownAndShrinkRange(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-    outputMin = operation.output(outputIdx++);
-    outputMax = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/QuantizedConcat.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/QuantizedConcat.java
deleted file mode 100644
index 714e49fa347..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/QuantizedConcat.java
+++ /dev/null
@@ -1,101 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.quantization;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.Operands;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.family.TType;
-
-/**
- * Concatenates quantized tensors along one dimension.
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator(group = "quantization")
-public final class QuantizedConcat<T extends TType> extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new QuantizedConcat operation.
-   * 
-   * @param scope current scope
-   * @param concatDim 0-D.  The dimension along which to concatenate.  Must be in the
-   * range [0, rank(values)).
-   * @param values The `N` Tensors to concatenate. Their ranks and types must match,
-   * and their sizes must match in all dimensions except `concat_dim`.
-   * @param inputMins The minimum scalar values for each of the input tensors.
-   * @param inputMaxes The maximum scalar values for each of the input tensors.
-   * @return a new instance of QuantizedConcat
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> QuantizedConcat<T> create(Scope scope, Operand<TInt32> concatDim, Iterable<Operand<T>> values, Iterable<Operand<TFloat32>> inputMins, Iterable<Operand<TFloat32>> inputMaxes) {
-    OperationBuilder opBuilder = scope.env().opBuilder("QuantizedConcat", scope.makeOpName("QuantizedConcat"));
-    opBuilder.addInput(concatDim.asOutput(scope));
-    opBuilder.addInputList(Operands.asOutputs(scope, values));
-    opBuilder.addInputList(Operands.asOutputs(scope, inputMins));
-    opBuilder.addInputList(Operands.asOutputs(scope, inputMaxes));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new QuantizedConcat<T>(opBuilder.build());
-  }
-  
-  /**
-   * A `Tensor` with the concatenation of values stacked along the
-   * `concat_dim` dimension.  This tensor's shape matches that of `values` except
-   * in `concat_dim` where it has the sum of the sizes.
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  /**
-   * The float value that the minimum quantized output value represents.
-   */
-  public Output<TFloat32> outputMin() {
-    return outputMin;
-  }
-  
-  /**
-   * The float value that the maximum quantized output value represents.
-   */
-  public Output<TFloat32> outputMax() {
-    return outputMax;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "QuantizedConcat";
-  
-  private Output<T> output;
-  private Output<TFloat32> outputMin;
-  private Output<TFloat32> outputMax;
-  
-  private QuantizedConcat(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-    outputMin = operation.output(outputIdx++);
-    outputMax = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/QuantizedMatMulWithBiasAndDequantize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/QuantizedMatMulWithBiasAndDequantize.java
deleted file mode 100644
index 0ab6bd11b79..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/QuantizedMatMulWithBiasAndDequantize.java
+++ /dev/null
@@ -1,163 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.quantization;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * @param <W> data type for {@code out()} output
- */
-public final class QuantizedMatMulWithBiasAndDequantize<W extends TNumber> extends RawOp implements Operand<W> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.quantization.QuantizedMatMulWithBiasAndDequantize}
-   */
-  public static class Options {
-    
-    /**
-     * @param transposeA 
-     */
-    public Options transposeA(Boolean transposeA) {
-      this.transposeA = transposeA;
-      return this;
-    }
-    
-    /**
-     * @param transposeB 
-     */
-    public Options transposeB(Boolean transposeB) {
-      this.transposeB = transposeB;
-      return this;
-    }
-    
-    /**
-     * @param inputQuantMode 
-     */
-    public Options inputQuantMode(String inputQuantMode) {
-      this.inputQuantMode = inputQuantMode;
-      return this;
-    }
-    
-    private Boolean transposeA;
-    private Boolean transposeB;
-    private String inputQuantMode;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new QuantizedMatMulWithBiasAndDequantize operation.
-   * 
-   * @param scope current scope
-   * @param a 
-   * @param b 
-   * @param bias 
-   * @param minA 
-   * @param maxA 
-   * @param minB 
-   * @param maxB 
-   * @param minFreezedOutput 
-   * @param maxFreezedOutput 
-   * @param Toutput 
-   * @param options carries optional attributes values
-   * @return a new instance of QuantizedMatMulWithBiasAndDequantize
-   */
-  @Endpoint(describeByClass = true)
-  public static <W extends TNumber, T extends TType, U extends TType, V extends TType> QuantizedMatMulWithBiasAndDequantize<W> create(Scope scope, Operand<T> a, Operand<U> b, Operand<V> bias, Operand<TFloat32> minA, Operand<TFloat32> maxA, Operand<TFloat32> minB, Operand<TFloat32> maxB, Operand<TFloat32> minFreezedOutput, Operand<TFloat32> maxFreezedOutput, Class<W> Toutput, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("QuantizedMatMulWithBiasAndDequantize", scope.makeOpName("QuantizedMatMulWithBiasAndDequantize"));
-    opBuilder.addInput(a.asOutput(scope));
-    opBuilder.addInput(b.asOutput(scope));
-    opBuilder.addInput(bias.asOutput(scope));
-    opBuilder.addInput(minA.asOutput(scope));
-    opBuilder.addInput(maxA.asOutput(scope));
-    opBuilder.addInput(minB.asOutput(scope));
-    opBuilder.addInput(maxB.asOutput(scope));
-    opBuilder.addInput(minFreezedOutput.asOutput(scope));
-    opBuilder.addInput(maxFreezedOutput.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("Toutput", Toutput);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.transposeA != null) {
-          opBuilder.setAttr("transpose_a", opts.transposeA);
-        }
-        if (opts.transposeB != null) {
-          opBuilder.setAttr("transpose_b", opts.transposeB);
-        }
-        if (opts.inputQuantMode != null) {
-          opBuilder.setAttr("input_quant_mode", opts.inputQuantMode);
-        }
-      }
-    }
-    return new QuantizedMatMulWithBiasAndDequantize<W>(opBuilder.build());
-  }
-  
-  /**
-   * @param transposeA 
-   */
-  public static Options transposeA(Boolean transposeA) {
-    return new Options().transposeA(transposeA);
-  }
-  
-  /**
-   * @param transposeB 
-   */
-  public static Options transposeB(Boolean transposeB) {
-    return new Options().transposeB(transposeB);
-  }
-  
-  /**
-   * @param inputQuantMode 
-   */
-  public static Options inputQuantMode(String inputQuantMode) {
-    return new Options().inputQuantMode(inputQuantMode);
-  }
-  
-  /**
-   */
-  public Output<W> out() {
-    return out;
-  }
-  
-  @Override
-  public Output<W> asOutput(Scope scope) {
-    return out;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "QuantizedMatMulWithBiasAndDequantize";
-  
-  private Output<W> out;
-  
-  private QuantizedMatMulWithBiasAndDequantize(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    out = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/QuantizedMatMulWithBiasAndRequantize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/QuantizedMatMulWithBiasAndRequantize.java
deleted file mode 100644
index 51341159d76..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/QuantizedMatMulWithBiasAndRequantize.java
+++ /dev/null
@@ -1,173 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.quantization;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.family.TType;
-
-/**
- * @param <W> data type for {@code out()} output
- */
-public final class QuantizedMatMulWithBiasAndRequantize<W extends TType> extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.quantization.QuantizedMatMulWithBiasAndRequantize}
-   */
-  public static class Options {
-    
-    /**
-     * @param transposeA 
-     */
-    public Options transposeA(Boolean transposeA) {
-      this.transposeA = transposeA;
-      return this;
-    }
-    
-    /**
-     * @param transposeB 
-     */
-    public Options transposeB(Boolean transposeB) {
-      this.transposeB = transposeB;
-      return this;
-    }
-    
-    /**
-     * @param inputQuantMode 
-     */
-    public Options inputQuantMode(String inputQuantMode) {
-      this.inputQuantMode = inputQuantMode;
-      return this;
-    }
-    
-    private Boolean transposeA;
-    private Boolean transposeB;
-    private String inputQuantMode;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new QuantizedMatMulWithBiasAndRequantize operation.
-   * 
-   * @param scope current scope
-   * @param a 
-   * @param b 
-   * @param bias 
-   * @param minA 
-   * @param maxA 
-   * @param minB 
-   * @param maxB 
-   * @param minFreezedOutput 
-   * @param maxFreezedOutput 
-   * @param Toutput 
-   * @param options carries optional attributes values
-   * @return a new instance of QuantizedMatMulWithBiasAndRequantize
-   */
-  @Endpoint(describeByClass = true)
-  public static <W extends TType, T extends TType, U extends TType, V extends TType> QuantizedMatMulWithBiasAndRequantize<W> create(Scope scope, Operand<T> a, Operand<U> b, Operand<V> bias, Operand<TFloat32> minA, Operand<TFloat32> maxA, Operand<TFloat32> minB, Operand<TFloat32> maxB, Operand<TFloat32> minFreezedOutput, Operand<TFloat32> maxFreezedOutput, Class<W> Toutput, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("QuantizedMatMulWithBiasAndRequantize", scope.makeOpName("QuantizedMatMulWithBiasAndRequantize"));
-    opBuilder.addInput(a.asOutput(scope));
-    opBuilder.addInput(b.asOutput(scope));
-    opBuilder.addInput(bias.asOutput(scope));
-    opBuilder.addInput(minA.asOutput(scope));
-    opBuilder.addInput(maxA.asOutput(scope));
-    opBuilder.addInput(minB.asOutput(scope));
-    opBuilder.addInput(maxB.asOutput(scope));
-    opBuilder.addInput(minFreezedOutput.asOutput(scope));
-    opBuilder.addInput(maxFreezedOutput.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("Toutput", Toutput);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.transposeA != null) {
-          opBuilder.setAttr("transpose_a", opts.transposeA);
-        }
-        if (opts.transposeB != null) {
-          opBuilder.setAttr("transpose_b", opts.transposeB);
-        }
-        if (opts.inputQuantMode != null) {
-          opBuilder.setAttr("input_quant_mode", opts.inputQuantMode);
-        }
-      }
-    }
-    return new QuantizedMatMulWithBiasAndRequantize<W>(opBuilder.build());
-  }
-  
-  /**
-   * @param transposeA 
-   */
-  public static Options transposeA(Boolean transposeA) {
-    return new Options().transposeA(transposeA);
-  }
-  
-  /**
-   * @param transposeB 
-   */
-  public static Options transposeB(Boolean transposeB) {
-    return new Options().transposeB(transposeB);
-  }
-  
-  /**
-   * @param inputQuantMode 
-   */
-  public static Options inputQuantMode(String inputQuantMode) {
-    return new Options().inputQuantMode(inputQuantMode);
-  }
-  
-  /**
-   */
-  public Output<W> out() {
-    return out;
-  }
-  
-  /**
-   */
-  public Output<TFloat32> minOut() {
-    return minOut;
-  }
-  
-  /**
-   */
-  public Output<TFloat32> maxOut() {
-    return maxOut;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "QuantizedMatMulWithBiasAndRequantize";
-  
-  private Output<W> out;
-  private Output<TFloat32> minOut;
-  private Output<TFloat32> maxOut;
-  
-  private QuantizedMatMulWithBiasAndRequantize(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    out = operation.output(outputIdx++);
-    minOut = operation.output(outputIdx++);
-    maxOut = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/RequantizationRange.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/RequantizationRange.java
deleted file mode 100644
index 2a57d9b435d..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/RequantizationRange.java
+++ /dev/null
@@ -1,87 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.quantization;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.family.TType;
-
-/**
- * Computes a range that covers the actual values present in a quantized tensor.
- * <p>
- * Given a quantized tensor described by `(input, input_min, input_max)`, outputs a
- * range that covers the actual values present in that tensor. This op is typically
- * used to produce the `requested_output_min` and `requested_output_max` for
- * `Requantize`.
- */
-@Operator(group = "quantization")
-public final class RequantizationRange extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new RequantizationRange operation.
-   * 
-   * @param scope current scope
-   * @param input 
-   * @param inputMin The float value that the minimum quantized input value represents.
-   * @param inputMax The float value that the maximum quantized input value represents.
-   * @return a new instance of RequantizationRange
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> RequantizationRange create(Scope scope, Operand<T> input, Operand<TFloat32> inputMin, Operand<TFloat32> inputMax) {
-    OperationBuilder opBuilder = scope.env().opBuilder("RequantizationRange", scope.makeOpName("RequantizationRange"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder.addInput(inputMin.asOutput(scope));
-    opBuilder.addInput(inputMax.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new RequantizationRange(opBuilder.build());
-  }
-  
-  /**
-   * The computed min output.
-   */
-  public Output<TFloat32> outputMin() {
-    return outputMin;
-  }
-  
-  /**
-   * the computed max output.
-   */
-  public Output<TFloat32> outputMax() {
-    return outputMax;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "RequantizationRange";
-  
-  private Output<TFloat32> outputMin;
-  private Output<TFloat32> outputMax;
-  
-  private RequantizationRange(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    outputMin = operation.output(outputIdx++);
-    outputMax = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/Requantize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/Requantize.java
deleted file mode 100644
index 1de8c064658..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/Requantize.java
+++ /dev/null
@@ -1,106 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.quantization;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.family.TType;
-
-/**
- * Converts the quantized `input` tensor into a lower-precision `output`.
- * <p>
- * Converts the quantized `input` tensor into a lower-precision `output`, using the
- * output range specified with `requested_output_min` and `requested_output_max`.
- * <p>
- * `[input_min, input_max]` are scalar floats that specify the range for the float
- * interpretation of the `input` data. For example, if `input_min` is -1.0f and
- * `input_max` is 1.0f, and we are dealing with `quint16` quantized data, then a 0
- * value in the 16-bit data should be interpreted as -1.0f, and a 65535 means 1.0f.
- * 
- * @param <U> data type for {@code output()} output
- */
-@Operator(group = "quantization")
-public final class Requantize<U extends TType> extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new Requantize operation.
-   * 
-   * @param scope current scope
-   * @param input 
-   * @param inputMin The float value that the minimum quantized input value represents.
-   * @param inputMax The float value that the maximum quantized input value represents.
-   * @param requestedOutputMin The float value that the minimum quantized output value represents.
-   * @param requestedOutputMax The float value that the maximum quantized output value represents.
-   * @param outType The type of the output. Should be a lower bit depth than Tinput.
-   * @return a new instance of Requantize
-   */
-  @Endpoint(describeByClass = true)
-  public static <U extends TType, T extends TType> Requantize<U> create(Scope scope, Operand<T> input, Operand<TFloat32> inputMin, Operand<TFloat32> inputMax, Operand<TFloat32> requestedOutputMin, Operand<TFloat32> requestedOutputMax, Class<U> outType) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Requantize", scope.makeOpName("Requantize"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder.addInput(inputMin.asOutput(scope));
-    opBuilder.addInput(inputMax.asOutput(scope));
-    opBuilder.addInput(requestedOutputMin.asOutput(scope));
-    opBuilder.addInput(requestedOutputMax.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("out_type", outType);
-    return new Requantize<U>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<U> output() {
-    return output;
-  }
-  
-  /**
-   * The requested_output_min value is copied into this output.
-   */
-  public Output<TFloat32> outputMin() {
-    return outputMin;
-  }
-  
-  /**
-   * The requested_output_max value is copied into this output.
-   */
-  public Output<TFloat32> outputMax() {
-    return outputMax;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Requantize";
-  
-  private Output<U> output;
-  private Output<TFloat32> outputMin;
-  private Output<TFloat32> outputMax;
-  
-  private Requantize(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-    outputMin = operation.output(outputIdx++);
-    outputMax = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/ragged/RaggedBincount.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/ragged/RaggedBincount.java
deleted file mode 100644
index 40aceaed5c8..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/ragged/RaggedBincount.java
+++ /dev/null
@@ -1,127 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.ragged;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Counts the number of occurrences of each value in an integer array.
- * <p>
- * Outputs a vector with length `size` and the same dtype as `weights`. If
- * `weights` are empty, then index `i` stores the number of times the value `i` is
- * counted in `arr`. If `weights` are non-empty, then index `i` stores the sum of
- * the value in `weights` at each index where the corresponding value in `arr` is
- * `i`.
- * <p>
- * Values in `arr` outside of the range [0, size) are ignored.
- * 
- * @param <U> data type for {@code output()} output
- */
-@Operator(group = "ragged")
-public final class RaggedBincount<U extends TNumber> extends RawOp implements Operand<U> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.ragged.RaggedBincount}
-   */
-  public static class Options {
-    
-    /**
-     * @param binaryOutput bool; Whether the kernel should count the appearance or number of occurrences.
-     */
-    public Options binaryOutput(Boolean binaryOutput) {
-      this.binaryOutput = binaryOutput;
-      return this;
-    }
-    
-    private Boolean binaryOutput;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new RaggedBincount operation.
-   * 
-   * @param scope current scope
-   * @param splits 1D int64 `Tensor`.
-   * @param values 2D int `Tensor`.
-   * @param size non-negative int scalar `Tensor`.
-   * @param weights is an int32, int64, float32, or float64 `Tensor` with the same
-   * shape as `input`, or a length-0 `Tensor`, in which case it acts as all weights
-   * equal to 1.
-   * @param options carries optional attributes values
-   * @return a new instance of RaggedBincount
-   */
-  @Endpoint(describeByClass = true)
-  public static <U extends TNumber, T extends TNumber> RaggedBincount<U> create(Scope scope, Operand<TInt64> splits, Operand<T> values, Operand<T> size, Operand<U> weights, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("RaggedBincount", scope.makeOpName("RaggedBincount"));
-    opBuilder.addInput(splits.asOutput(scope));
-    opBuilder.addInput(values.asOutput(scope));
-    opBuilder.addInput(size.asOutput(scope));
-    opBuilder.addInput(weights.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.binaryOutput != null) {
-          opBuilder.setAttr("binary_output", opts.binaryOutput);
-        }
-      }
-    }
-    return new RaggedBincount<U>(opBuilder.build());
-  }
-  
-  /**
-   * @param binaryOutput bool; Whether the kernel should count the appearance or number of occurrences.
-   */
-  public static Options binaryOutput(Boolean binaryOutput) {
-    return new Options().binaryOutput(binaryOutput);
-  }
-  
-  /**
-   * 1D `Tensor` with length equal to `size` or 2D `Tensor` with [batch_size, `size`].
-   * The counts or summed weights for each value in the range [0, size).
-   */
-  public Output<U> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<U> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "RaggedBincount";
-  
-  private Output<U> output;
-  
-  private RaggedBincount(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/ragged/RaggedCountSparseOutput.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/ragged/RaggedCountSparseOutput.java
deleted file mode 100644
index b747211fa5b..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/ragged/RaggedCountSparseOutput.java
+++ /dev/null
@@ -1,156 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.ragged;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Performs sparse-output bin counting for a ragged tensor input.
- * <p>
- *   Counts the number of times each value occurs in the input.
- * 
- * @param <U> data type for {@code outputValues()} output
- */
-public final class RaggedCountSparseOutput<U extends TNumber> extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.ragged.RaggedCountSparseOutput}
-   */
-  public static class Options {
-    
-    /**
-     * @param minlength Minimum value to count. Can be set to -1 for no minimum.
-     */
-    public Options minlength(Long minlength) {
-      this.minlength = minlength;
-      return this;
-    }
-    
-    /**
-     * @param maxlength Maximum value to count. Can be set to -1 for no maximum.
-     */
-    public Options maxlength(Long maxlength) {
-      this.maxlength = maxlength;
-      return this;
-    }
-    
-    private Long minlength;
-    private Long maxlength;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new RaggedCountSparseOutput operation.
-   * 
-   * @param scope current scope
-   * @param splits Tensor containing the row splits of the ragged tensor to count.
-   * @param values Tensor containing values of the sparse tensor to count.
-   * @param weights A Tensor of the same shape as indices containing per-index weight values.
-   * May also be the empty tensor if no weights are used.
-   * @param binaryOutput Whether to output the number of occurrences of each value or 1.
-   * @param options carries optional attributes values
-   * @return a new instance of RaggedCountSparseOutput
-   */
-  @Endpoint(describeByClass = true)
-  public static <U extends TNumber, T extends TNumber> RaggedCountSparseOutput<U> create(Scope scope, Operand<TInt64> splits, Operand<T> values, Operand<U> weights, Boolean binaryOutput, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("RaggedCountSparseOutput", scope.makeOpName("RaggedCountSparseOutput"));
-    opBuilder.addInput(splits.asOutput(scope));
-    opBuilder.addInput(values.asOutput(scope));
-    opBuilder.addInput(weights.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("binary_output", binaryOutput);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.minlength != null) {
-          opBuilder.setAttr("minlength", opts.minlength);
-        }
-        if (opts.maxlength != null) {
-          opBuilder.setAttr("maxlength", opts.maxlength);
-        }
-      }
-    }
-    return new RaggedCountSparseOutput<U>(opBuilder.build());
-  }
-  
-  /**
-   * @param minlength Minimum value to count. Can be set to -1 for no minimum.
-   */
-  public static Options minlength(Long minlength) {
-    return new Options().minlength(minlength);
-  }
-  
-  /**
-   * @param maxlength Maximum value to count. Can be set to -1 for no maximum.
-   */
-  public static Options maxlength(Long maxlength) {
-    return new Options().maxlength(maxlength);
-  }
-  
-  /**
-   * Indices tensor for the resulting sparse tensor object.
-   */
-  public Output<TInt64> outputIndices() {
-    return outputIndices;
-  }
-  
-  /**
-   * Values tensor for the resulting sparse tensor object.
-   */
-  public Output<U> outputValues() {
-    return outputValues;
-  }
-  
-  /**
-   * Shape tensor for the resulting sparse tensor object.
-   *   END
-   *   }
-   *   attr {
-   *     name: "T"
-   *     description: <<END
-   * Dtype of the input values tensor.
-   */
-  public Output<TInt64> outputDenseShape() {
-    return outputDenseShape;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "RaggedCountSparseOutput";
-  
-  private Output<TInt64> outputIndices;
-  private Output<U> outputValues;
-  private Output<TInt64> outputDenseShape;
-  
-  private RaggedCountSparseOutput(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    outputIndices = operation.output(outputIdx++);
-    outputValues = operation.output(outputIdx++);
-    outputDenseShape = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/ragged/RaggedCross.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/ragged/RaggedCross.java
deleted file mode 100644
index 2d13027a88a..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/ragged/RaggedCross.java
+++ /dev/null
@@ -1,108 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.ragged;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.Operands;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * Generates a feature cross from a list of tensors, and returns it as a
- * RaggedTensor.  See `tf.ragged.cross` for more details.
- * 
- * @param <T> data type for {@code outputValues()} output
- * @param <U> data type for {@code outputRowSplits()} output
- */
-public final class RaggedCross<T extends TType, U extends TNumber> extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new RaggedCross operation.
-   * 
-   * @param scope current scope
-   * @param raggedValues The values tensor for each RaggedTensor input.
-   * @param raggedRowSplits The row_splits tensor for each RaggedTensor input.
-   * @param sparseIndices The indices tensor for each SparseTensor input.
-   * @param sparseValues The values tensor for each SparseTensor input.
-   * @param sparseShape The dense_shape tensor for each SparseTensor input.
-   * @param denseInputs The tf.Tensor inputs.
-   * @param inputOrder String specifying the tensor type for each input.  The `i`th character in
-   * this string specifies the type of the `i`th input, and is one of: 'R' (ragged),
-   * 'D' (dense), or 'S' (sparse).  This attr is used to ensure that the crossed
-   * values are combined in the order of the inputs from the call to tf.ragged.cross.
-   * @param hashedOutput 
-   * @param numBuckets 
-   * @param hashKey 
-   * @param outValuesType 
-   * @param outRowSplitsType 
-   * @return a new instance of RaggedCross
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> RaggedCross<T, U> create(Scope scope, Iterable<Operand<?>> raggedValues, Iterable<Operand<?>> raggedRowSplits, Iterable<Operand<TInt64>> sparseIndices, Iterable<Operand<?>> sparseValues, Iterable<Operand<TInt64>> sparseShape, Iterable<Operand<?>> denseInputs, String inputOrder, Boolean hashedOutput, Long numBuckets, Long hashKey, Class<T> outValuesType, Class<U> outRowSplitsType) {
-    OperationBuilder opBuilder = scope.env().opBuilder("RaggedCross", scope.makeOpName("RaggedCross"));
-    opBuilder.addInputList(Operands.asOutputs(scope, raggedValues));
-    opBuilder.addInputList(Operands.asOutputs(scope, raggedRowSplits));
-    opBuilder.addInputList(Operands.asOutputs(scope, sparseIndices));
-    opBuilder.addInputList(Operands.asOutputs(scope, sparseValues));
-    opBuilder.addInputList(Operands.asOutputs(scope, sparseShape));
-    opBuilder.addInputList(Operands.asOutputs(scope, denseInputs));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("input_order", inputOrder);
-    opBuilder.setAttr("hashed_output", hashedOutput);
-    opBuilder.setAttr("num_buckets", numBuckets);
-    opBuilder.setAttr("hash_key", hashKey);
-    opBuilder.setAttr("out_values_type", outValuesType);
-    opBuilder.setAttr("out_row_splits_type", outRowSplitsType);
-    return new RaggedCross<T, U>(opBuilder.build());
-  }
-  
-  /**
-   * The `values` for the returned `RaggedTensor`.
-   */
-  public Output<T> outputValues() {
-    return outputValues;
-  }
-  
-  /**
-   * The `row_splits` for the returned `RaggedTensor`.
-   */
-  public Output<U> outputRowSplits() {
-    return outputRowSplits;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "RaggedCross";
-  
-  private Output<T> outputValues;
-  private Output<U> outputRowSplits;
-  
-  private RaggedCross(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    outputValues = operation.output(outputIdx++);
-    outputRowSplits = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/ragged/RaggedGather.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/ragged/RaggedGather.java
deleted file mode 100644
index fe59b2e4373..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/ragged/RaggedGather.java
+++ /dev/null
@@ -1,127 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.ragged;
-
-import java.util.Arrays;
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.Operands;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * Gather ragged slices from `params` axis `0` according to `indices`.
- * <p>
- * Outputs a `RaggedTensor` output composed from `output_dense_values` and
- * `output_nested_splits`, such that:
- * <pre>{@code
- * output.shape = indices.shape + params.shape[1:]
- * output.ragged_rank = indices.shape.ndims + params.ragged_rank
- * output[i...j, d0...dn] = params[indices[i...j], d0...dn]
- * }</pre>
- * where
- * <ul>
- * <li>
- * `params =
- *    ragged.from_nested_row_splits(params_dense_values, params_nested_splits)`
- *    provides the values that should be gathered.
- * </li>
- * <li>
- * `indices` ia a dense tensor with dtype `int32` or `int64`, indicating which
- *    values should be gathered.
- * </li>
- * <li>
- * `output =
- *    ragged.from_nested_row_splits(output_dense_values, output_nested_splits)`
- *    is the output tensor.
- * </li>
- * </ul>
- * (Note: This c++ op is used to implement the higher-level python
- * `tf.ragged.gather` op, which also supports ragged indices.)
- * 
- * 
- * @param <T> data type for {@code outputNestedSplits()} output
- * @param <U> data type for {@code outputDenseValues()} output
- */
-public final class RaggedGather<T extends TNumber, U extends TType> extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new RaggedGather operation.
-   * 
-   * @param scope current scope
-   * @param paramsNestedSplits The `nested_row_splits` tensors that define the row-partitioning for the
-   * `params` RaggedTensor input.
-   * @param paramsDenseValues The `flat_values` for the `params` RaggedTensor. There was a terminology change
-   * at the python level from dense_values to flat_values, so dense_values is the
-   * deprecated name.
-   * @param indices Indices in the outermost dimension of `params` of the values that should be
-   * gathered.
-   * @param OUTPUTRAGGEDRANK The ragged rank of the output RaggedTensor. `output_nested_splits` will contain
-   * this number of `row_splits` tensors. This value should equal
-   * `indices.shape.ndims + params.ragged_rank - 1`.
-   * @return a new instance of RaggedGather
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber, U extends TType, V extends TNumber> RaggedGather<T, U> create(Scope scope, Iterable<Operand<T>> paramsNestedSplits, Operand<U> paramsDenseValues, Operand<V> indices, Long OUTPUTRAGGEDRANK) {
-    OperationBuilder opBuilder = scope.env().opBuilder("RaggedGather", scope.makeOpName("RaggedGather"));
-    opBuilder.addInputList(Operands.asOutputs(scope, paramsNestedSplits));
-    opBuilder.addInput(paramsDenseValues.asOutput(scope));
-    opBuilder.addInput(indices.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("OUTPUT_RAGGED_RANK", OUTPUTRAGGEDRANK);
-    return new RaggedGather<T, U>(opBuilder.build());
-  }
-  
-  /**
-   * The `nested_row_splits` tensors that define the row-partitioning for the
-   * returned RaggedTensor.
-   */
-  public List<Output<T>> outputNestedSplits() {
-    return outputNestedSplits;
-  }
-  
-  /**
-   * The `flat_values` for the returned RaggedTensor.
-   */
-  public Output<U> outputDenseValues() {
-    return outputDenseValues;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "RaggedGather";
-  
-  private List<Output<T>> outputNestedSplits;
-  private Output<U> outputDenseValues;
-  
-  @SuppressWarnings("unchecked")
-  private RaggedGather(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    int outputNestedSplitsLength = operation.outputListLength("output_nested_splits");
-    outputNestedSplits = Arrays.asList((Output<T>[])operation.outputList(outputIdx, outputNestedSplitsLength));
-    outputIdx += outputNestedSplitsLength;
-    outputDenseValues = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/ragged/RaggedRange.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/ragged/RaggedRange.java
deleted file mode 100644
index 327b8a3c1d8..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/ragged/RaggedRange.java
+++ /dev/null
@@ -1,115 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.ragged;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Returns a `RaggedTensor` containing the specified sequences of numbers.
- * <p>
- * 
- * Returns a `RaggedTensor` `result` composed from `rt_dense_values` and
- * `rt_nested_splits`, such that
- * `result[i] = range(starts[i], limits[i], deltas[i])`.
- * <pre>{@code
- * (rt_nested_splits, rt_dense_values) = ragged_range(
- *       starts=[2, 5, 8], limits=[3, 5, 12], deltas=1)
- * result = tf.ragged.from_row_splits(rt_dense_values, rt_nested_splits)
- * print(result)
- * <tf.RaggedTensor [[2], [], [8, 9, 10, 11]] >
- * }</pre>
- * The input tensors `starts`, `limits`, and `deltas` may be scalars or vectors.
- * The vector inputs must all have the same size.  Scalar inputs are broadcast
- * to match the size of the vector inputs.
- * 
- * @param <U> data type for {@code rtNestedSplits()} output
- * @param <T> data type for {@code rtDenseValues()} output
- */
-public final class RaggedRange<U extends TNumber, T extends TNumber> extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new RaggedRange operation.
-   * 
-   * @param scope current scope
-   * @param starts The starts of each range.
-   * @param limits The limits of each range.
-   * @param deltas The deltas of each range.
-   * @param Tsplits 
-   * @return a new instance of RaggedRange
-   */
-  @Endpoint(describeByClass = true)
-  public static <U extends TNumber, T extends TNumber> RaggedRange<U, T> create(Scope scope, Operand<T> starts, Operand<T> limits, Operand<T> deltas, Class<U> Tsplits) {
-    OperationBuilder opBuilder = scope.env().opBuilder("RaggedRange", scope.makeOpName("RaggedRange"));
-    opBuilder.addInput(starts.asOutput(scope));
-    opBuilder.addInput(limits.asOutput(scope));
-    opBuilder.addInput(deltas.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("Tsplits", Tsplits);
-    return new RaggedRange<U, T>(opBuilder.build());
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new RaggedRange operation using default output types.
-   * 
-   * @param scope current scope
-   * @param starts The starts of each range.
-   * @param limits The limits of each range.
-   * @param deltas The deltas of each range.
-   * @return a new instance of RaggedRange
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> RaggedRange<TInt64, T> create(Scope scope, Operand<T> starts, Operand<T> limits, Operand<T> deltas) {
-    return create(scope, starts, limits, deltas, TInt64.class);
-  }
-  
-  /**
-   * The `row_splits` for the returned `RaggedTensor`.
-   */
-  public Output<U> rtNestedSplits() {
-    return rtNestedSplits;
-  }
-  
-  /**
-   * The `flat_values` for the returned `RaggedTensor`.
-   */
-  public Output<T> rtDenseValues() {
-    return rtDenseValues;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "RaggedRange";
-  
-  private Output<U> rtNestedSplits;
-  private Output<T> rtDenseValues;
-  
-  private RaggedRange(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    rtNestedSplits = operation.output(outputIdx++);
-    rtDenseValues = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/ragged/RaggedTensorFromVariant.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/ragged/RaggedTensorFromVariant.java
deleted file mode 100644
index 736b6c47b21..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/ragged/RaggedTensorFromVariant.java
+++ /dev/null
@@ -1,127 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.ragged;
-
-import java.util.Arrays;
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * Decodes a `variant` Tensor into a `RaggedTensor`.
- * <p>
- * Decodes the given `variant` Tensor and returns a `RaggedTensor`. The input
- * could be a scalar, meaning it encodes a single `RaggedTensor` with ragged_rank
- * `output_ragged_rank`. It could also have an arbitrary rank, in which case each
- * element is decoded into a `RaggedTensor` with ragged_rank `input_ragged_rank`
- * and these are then stacked according to the input shape to output a single
- * `RaggedTensor` with ragged_rank `output_ragged_rank`. Each `variant` element in
- * the input Tensor is decoded by retrieving from the element a 1-D `variant`
- * Tensor with `input_ragged_rank + 1` Tensors, corresponding to the splits and
- * values of the decoded `RaggedTensor`. If `input_ragged_rank` is -1, then it is
- * inferred as `output_ragged_rank` - `rank(encoded_ragged)`. See
- * `RaggedTensorToVariant` for the corresponding encoding logic.
- * 
- * 
- * @param <U> data type for {@code outputNestedSplits()} output
- * @param <T> data type for {@code outputDenseValues()} output
- */
-public final class RaggedTensorFromVariant<U extends TNumber, T extends TType> extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new RaggedTensorFromVariant operation.
-   * 
-   * @param scope current scope
-   * @param encodedRagged A `variant` Tensor containing encoded `RaggedTensor`s.
-   * @param inputRaggedRank The ragged rank of each encoded `RaggedTensor` component in the input. If set to
-   * -1, this is inferred as `output_ragged_rank` - `rank(encoded_ragged)`
-   * @param outputRaggedRank The expected ragged rank of the output `RaggedTensor`. The following must hold:
-   * `output_ragged_rank = rank(encoded_ragged) + input_ragged_rank`.
-   * @param Tvalues 
-   * @param Tsplits 
-   * @return a new instance of RaggedTensorFromVariant
-   */
-  @Endpoint(describeByClass = true)
-  public static <U extends TNumber, T extends TType> RaggedTensorFromVariant<U, T> create(Scope scope, Operand<?> encodedRagged, Long inputRaggedRank, Long outputRaggedRank, Class<T> Tvalues, Class<U> Tsplits) {
-    OperationBuilder opBuilder = scope.env().opBuilder("RaggedTensorFromVariant", scope.makeOpName("RaggedTensorFromVariant"));
-    opBuilder.addInput(encodedRagged.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("input_ragged_rank", inputRaggedRank);
-    opBuilder.setAttr("output_ragged_rank", outputRaggedRank);
-    opBuilder.setAttr("Tvalues", Tvalues);
-    opBuilder.setAttr("Tsplits", Tsplits);
-    return new RaggedTensorFromVariant<U, T>(opBuilder.build());
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new RaggedTensorFromVariant operation using default output types.
-   * 
-   * @param scope current scope
-   * @param encodedRagged A `variant` Tensor containing encoded `RaggedTensor`s.
-   * @param inputRaggedRank The ragged rank of each encoded `RaggedTensor` component in the input. If set to
-   * -1, this is inferred as `output_ragged_rank` - `rank(encoded_ragged)`
-   * @param outputRaggedRank The expected ragged rank of the output `RaggedTensor`. The following must hold:
-   * `output_ragged_rank = rank(encoded_ragged) + input_ragged_rank`.
-   * @param Tvalues 
-   * @return a new instance of RaggedTensorFromVariant
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> RaggedTensorFromVariant<TInt64, T> create(Scope scope, Operand<?> encodedRagged, Long inputRaggedRank, Long outputRaggedRank, Class<T> Tvalues) {
-    return create(scope, encodedRagged, inputRaggedRank, outputRaggedRank, Tvalues, TInt64.class);
-  }
-  
-  /**
-   * A list of one or more Tensors representing the splits of the output
-   * `RaggedTensor`.
-   */
-  public List<Output<U>> outputNestedSplits() {
-    return outputNestedSplits;
-  }
-  
-  /**
-   * A Tensor representing the values of the output `RaggedTensor`.
-   */
-  public Output<T> outputDenseValues() {
-    return outputDenseValues;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "RaggedTensorFromVariant";
-  
-  private List<Output<U>> outputNestedSplits;
-  private Output<T> outputDenseValues;
-  
-  @SuppressWarnings("unchecked")
-  private RaggedTensorFromVariant(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    int outputNestedSplitsLength = operation.outputListLength("output_nested_splits");
-    outputNestedSplits = Arrays.asList((Output<U>[])operation.outputList(outputIdx, outputNestedSplitsLength));
-    outputIdx += outputNestedSplitsLength;
-    outputDenseValues = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/ragged/RaggedTensorToSparse.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/ragged/RaggedTensorToSparse.java
deleted file mode 100644
index 5ca5d799c67..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/ragged/RaggedTensorToSparse.java
+++ /dev/null
@@ -1,96 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.ragged;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.Operands;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * Converts a `RaggedTensor` into a `SparseTensor` with the same values.
- * <p>
- * input=ragged.from_nested_row_splits(rt_dense_values, rt_nested_splits)
- * output=SparseTensor(indices=sparse_indices, values=sparse_values,
- *                     dense_shape=sparse_dense_shape)
- * 
- * @param <U> data type for {@code sparseValues()} output
- */
-public final class RaggedTensorToSparse<U extends TType> extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new RaggedTensorToSparse operation.
-   * 
-   * @param scope current scope
-   * @param rtNestedSplits The `row_splits` for the `RaggedTensor`.
-   * @param rtDenseValues The `flat_values` for the `RaggedTensor`.
-   * @return a new instance of RaggedTensorToSparse
-   */
-  @Endpoint(describeByClass = true)
-  public static <U extends TType, T extends TNumber> RaggedTensorToSparse<U> create(Scope scope, Iterable<Operand<T>> rtNestedSplits, Operand<U> rtDenseValues) {
-    OperationBuilder opBuilder = scope.env().opBuilder("RaggedTensorToSparse", scope.makeOpName("RaggedTensorToSparse"));
-    opBuilder.addInputList(Operands.asOutputs(scope, rtNestedSplits));
-    opBuilder.addInput(rtDenseValues.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new RaggedTensorToSparse<U>(opBuilder.build());
-  }
-  
-  /**
-   * The indices for the `SparseTensor`.
-   */
-  public Output<TInt64> sparseIndices() {
-    return sparseIndices;
-  }
-  
-  /**
-   * The values of the `SparseTensor`.
-   */
-  public Output<U> sparseValues() {
-    return sparseValues;
-  }
-  
-  /**
-   * `sparse_dense_shape` is a tight bounding box of the input `RaggedTensor`.
-   */
-  public Output<TInt64> sparseDenseShape() {
-    return sparseDenseShape;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "RaggedTensorToSparse";
-  
-  private Output<TInt64> sparseIndices;
-  private Output<U> sparseValues;
-  private Output<TInt64> sparseDenseShape;
-  
-  private RaggedTensorToSparse(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    sparseIndices = operation.output(outputIdx++);
-    sparseValues = operation.output(outputIdx++);
-    sparseDenseShape = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/ragged/RaggedTensorToTensor.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/ragged/RaggedTensorToTensor.java
deleted file mode 100644
index 589c0a5e100..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/ragged/RaggedTensorToTensor.java
+++ /dev/null
@@ -1,146 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.ragged;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.Operands;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * Create a dense tensor from a ragged tensor, possibly altering its shape.
- * <p>
- * The `ragged_to_dense` op creates a dense tensor from a list of row partition
- * tensors, a value vector, and default values. If the shape is unspecified, the
- * minimal shape required to contain all the elements in the ragged tensor (the
- * natural shape) will be used. If some dimensions are left unspecified, then the
- * size of the natural shape is used in that dimension.
- * <p>
- * The default_value will be broadcast to the output shape. After that, the values
- * from the ragged tensor overwrite the default values. Note that the default_value
- * must have less dimensions than the value.
- * <p>
- * The row partition tensors are in the order of the dimensions.
- * At present, the types can be:
- * <ul>
- * <li>
- * "ROW_SPLITS": the row_splits tensor from the ragged tensor.
- * </li>
- * <li>
- * "VALUE_ROWIDS": the value_rowids tensor from the ragged tensor.
- * </li>
- * <li>
- * "FIRST_DIM_SIZE": if value_rowids is used for the first dimension, then it
- *   is preceded by "FIRST_DIM_SIZE".
- * 
- * @param <U> data type for {@code result()} output
- */
-public final class RaggedTensorToTensor<U extends TType> extends RawOp implements Operand<U> {
-  
-  /**
-   * Factory method to create a class wrapping a new RaggedTensorToTensor operation.
-   * 
-   * @param scope current scope
-   * @param shape The desired shape of the the output tensor. If left unspecified (empty),
-   * the minimal shape required to contain all the elements in the ragged tensor
-   * (the natural shape) will be used. If some dimensions are left unspecified, then
-   * the size of the natural shape is used in that dimension.
-   * <p>
-   * Note that dense dimensions cannot be modified by the shape argument. Trying to
-   * change the size of a dense dimension will cause the op to fail.
-   * Examples:
-   * natural shape: [4, 5, 6]
-   * shape: -1
-   * output shape: [4, 5, 6]
-   * <p>
-   * natural shape: [4, 5, 6]
-   * shape: [3, -1, 2]
-   * output shape: [3, 5, 2]
-   * <p>
-   * natural shape: [4, 5, 6]
-   * shape: [3, 7, 2]
-   * output shape: [3, 7, 2]
-   * 
-   * @param values A 1D tensor representing the values of the ragged tensor.
-   * @param defaultValue The default_value when the shape is larger than the ragged tensor. The
-   * default_value is broadcast until it is the shape of the output tensor, and
-   * then overwritten by values in the ragged tensor. The default value must be
-   * compatible with this broadcast operation, and must have fewer dimensions than
-   * the value tensor.
-   * @param rowPartitionTensors 
-   * @param rowPartitionTypes The types of the row partition tensors. At present, these can be:
-   * <ul>
-   * <li>
-   * "ROW_SPLITS": the row_splits tensor from the ragged tensor.
-   * </li>
-   * <li>
-   * "VALUE_ROWIDS": the value_rowids tensor from the ragged tensor.
-   * </li>
-   * <li>
-   * "FIRST_DIM_SIZE": if value_rowids is used for the first dimension, then it
-   *   is preceeded by "FIRST_DIM_SIZE".
-   * The tensors are in the order of the dimensions.
-   * @return a new instance of RaggedTensorToTensor
-   */
-  @Endpoint(describeByClass = true)
-  public static <U extends TType, T extends TNumber, V extends TNumber> RaggedTensorToTensor<U> create(Scope scope, Operand<T> shape, Operand<U> values, Operand<U> defaultValue, Iterable<Operand<V>> rowPartitionTensors, List<String> rowPartitionTypes) {
-    OperationBuilder opBuilder = scope.env().opBuilder("RaggedTensorToTensor", scope.makeOpName("RaggedTensorToTensor"));
-    opBuilder.addInput(shape.asOutput(scope));
-    opBuilder.addInput(values.asOutput(scope));
-    opBuilder.addInput(defaultValue.asOutput(scope));
-    opBuilder.addInputList(Operands.asOutputs(scope, rowPartitionTensors));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    String[] rowPartitionTypesArray = new String[rowPartitionTypes.size()];
-    for (int i = 0; i < rowPartitionTypesArray.length; ++i) {
-      rowPartitionTypesArray[i] = rowPartitionTypes.get(i);
-    }
-    opBuilder.setAttr("row_partition_types", rowPartitionTypesArray);
-    return new RaggedTensorToTensor<U>(opBuilder.build());
-  }
-  
-  /**
-   * The resulting dense tensor.
-   */
-  public Output<U> result() {
-    return result;
-  }
-  
-  @Override
-  public Output<U> asOutput(Scope scope) {
-    return result;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "RaggedTensorToTensor";
-  
-  private Output<U> result;
-  
-  private RaggedTensorToTensor(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    result = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/ragged/RaggedTensorToVariant.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/ragged/RaggedTensorToVariant.java
deleted file mode 100644
index 1a48b91b9f5..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/ragged/RaggedTensorToVariant.java
+++ /dev/null
@@ -1,93 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.ragged;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.Operands;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * Encodes a `RaggedTensor` into a `variant` Tensor.
- * <p>
- * 
- * Encodes the given `RaggedTensor` and returns a `variant` Tensor. If
- * `batched_input` is True, then input `RaggedTensor` is unbatched along the
- * zero-th dimension, each component `RaggedTensor` is encoded into a scalar
- * `variant` Tensor, and these are stacked to return a 1-D `variant` Tensor.
- * If `batched_input` is False, then the input `RaggedTensor` is encoded as is and
- * a scalar `variant` Tensor is returned. A `RaggedTensor` is encoded by first
- * creating a 1-D `variant` Tensor with `ragged_rank + 1` elements, containing the
- * splits and values Tensors of the `RaggedTensor`. Then the 1-D `variant` Tensor
- * is wrapped in a scalar `variant` Tensor. See `RaggedTensorFromVariant` for the
- * corresponding decoding logic.
- * 
- */
-public final class RaggedTensorToVariant extends RawOp implements Operand<TType> {
-  
-  /**
-   * Factory method to create a class wrapping a new RaggedTensorToVariant operation.
-   * 
-   * @param scope current scope
-   * @param rtNestedSplits A list of one or more Tensors representing the splits of the input
-   * `RaggedTensor`.
-   * @param rtDenseValues A Tensor representing the values of the input `RaggedTensor`.
-   * @param batchedInput A `bool` denoting whether the input is a batched `RaggedTensor`.
-   * @return a new instance of RaggedTensorToVariant
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber, U extends TType> RaggedTensorToVariant create(Scope scope, Iterable<Operand<T>> rtNestedSplits, Operand<U> rtDenseValues, Boolean batchedInput) {
-    OperationBuilder opBuilder = scope.env().opBuilder("RaggedTensorToVariant", scope.makeOpName("RaggedTensorToVariant"));
-    opBuilder.addInputList(Operands.asOutputs(scope, rtNestedSplits));
-    opBuilder.addInput(rtDenseValues.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("batched_input", batchedInput);
-    return new RaggedTensorToVariant(opBuilder.build());
-  }
-  
-  /**
-   * A `variant` Tensor that containing encoded `RaggedTensor`.
-   */
-  public Output<?> encodedRagged() {
-    return encodedRagged;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) encodedRagged;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "RaggedTensorToVariant";
-  
-  private Output<?> encodedRagged;
-  
-  private RaggedTensorToVariant(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    encodedRagged = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/AllCandidateSampler.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/AllCandidateSampler.java
deleted file mode 100644
index 1f495de2599..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/AllCandidateSampler.java
+++ /dev/null
@@ -1,169 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.random;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.TInt64;
-
-/**
- * Generates labels for candidate sampling with a learned unigram distribution.
- * <p>
- * See explanations of candidate sampling and the data formats at
- * go/candidate-sampling.
- * <p>
- * For each batch, this op picks a single set of sampled candidate labels.
- * <p>
- * The advantages of sampling candidates per-batch are simplicity and the
- * possibility of efficient dense matrix multiplication. The disadvantage is that
- * the sampled candidates must be chosen independently of the context and of the
- * true labels.
- */
-@Operator(group = "random")
-public final class AllCandidateSampler extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.random.AllCandidateSampler}
-   */
-  public static class Options {
-    
-    /**
-     * @param seed If either seed or seed2 are set to be non-zero, the random number
-     * generator is seeded by the given seed.  Otherwise, it is seeded by a
-     * random seed.
-     */
-    public Options seed(Long seed) {
-      this.seed = seed;
-      return this;
-    }
-    
-    /**
-     * @param seed2 An second seed to avoid seed collision.
-     */
-    public Options seed2(Long seed2) {
-      this.seed2 = seed2;
-      return this;
-    }
-    
-    private Long seed;
-    private Long seed2;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new AllCandidateSampler operation.
-   * 
-   * @param scope current scope
-   * @param trueClasses A batch_size * num_true matrix, in which each row contains the
-   * IDs of the num_true target_classes in the corresponding original label.
-   * @param numTrue Number of true labels per context.
-   * @param numSampled Number of candidates to produce.
-   * @param unique If unique is true, we sample with rejection, so that all sampled
-   * candidates in a batch are unique. This requires some approximation to
-   * estimate the post-rejection sampling probabilities.
-   * @param options carries optional attributes values
-   * @return a new instance of AllCandidateSampler
-   */
-  @Endpoint(describeByClass = true)
-  public static AllCandidateSampler create(Scope scope, Operand<TInt64> trueClasses, Long numTrue, Long numSampled, Boolean unique, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("AllCandidateSampler", scope.makeOpName("AllCandidateSampler"));
-    opBuilder.addInput(trueClasses.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("num_true", numTrue);
-    opBuilder.setAttr("num_sampled", numSampled);
-    opBuilder.setAttr("unique", unique);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.seed != null) {
-          opBuilder.setAttr("seed", opts.seed);
-        }
-        if (opts.seed2 != null) {
-          opBuilder.setAttr("seed2", opts.seed2);
-        }
-      }
-    }
-    return new AllCandidateSampler(opBuilder.build());
-  }
-  
-  /**
-   * @param seed If either seed or seed2 are set to be non-zero, the random number
-   * generator is seeded by the given seed.  Otherwise, it is seeded by a
-   * random seed.
-   */
-  public static Options seed(Long seed) {
-    return new Options().seed(seed);
-  }
-  
-  /**
-   * @param seed2 An second seed to avoid seed collision.
-   */
-  public static Options seed2(Long seed2) {
-    return new Options().seed2(seed2);
-  }
-  
-  /**
-   * A vector of length num_sampled, in which each element is
-   * the ID of a sampled candidate.
-   */
-  public Output<TInt64> sampledCandidates() {
-    return sampledCandidates;
-  }
-  
-  /**
-   * A batch_size * num_true matrix, representing
-   * the number of times each candidate is expected to occur in a batch
-   * of sampled candidates. If unique=true, then this is a probability.
-   */
-  public Output<TFloat32> trueExpectedCount() {
-    return trueExpectedCount;
-  }
-  
-  /**
-   * A vector of length num_sampled, for each sampled
-   * candidate representing the number of times the candidate is expected
-   * to occur in a batch of sampled candidates.  If unique=true, then this is a
-   * probability.
-   */
-  public Output<TFloat32> sampledExpectedCount() {
-    return sampledExpectedCount;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "AllCandidateSampler";
-  
-  private Output<TInt64> sampledCandidates;
-  private Output<TFloat32> trueExpectedCount;
-  private Output<TFloat32> sampledExpectedCount;
-  
-  private AllCandidateSampler(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    sampledCandidates = operation.output(outputIdx++);
-    trueExpectedCount = operation.output(outputIdx++);
-    sampledExpectedCount = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/AnonymousRandomSeedGenerator.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/AnonymousRandomSeedGenerator.java
deleted file mode 100644
index 3e5f52a8c9b..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/AnonymousRandomSeedGenerator.java
+++ /dev/null
@@ -1,75 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.random;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt64;
-
-/**
- */
-public final class AnonymousRandomSeedGenerator extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new AnonymousRandomSeedGenerator operation.
-   * 
-   * @param scope current scope
-   * @param seed 
-   * @param seed2 
-   * @return a new instance of AnonymousRandomSeedGenerator
-   */
-  @Endpoint(describeByClass = true)
-  public static AnonymousRandomSeedGenerator create(Scope scope, Operand<TInt64> seed, Operand<TInt64> seed2) {
-    OperationBuilder opBuilder = scope.env().opBuilder("AnonymousRandomSeedGenerator", scope.makeOpName("AnonymousRandomSeedGenerator"));
-    opBuilder.addInput(seed.asOutput(scope));
-    opBuilder.addInput(seed2.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new AnonymousRandomSeedGenerator(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<?> handle() {
-    return handle;
-  }
-  
-  /**
-   */
-  public Output<?> deleter() {
-    return deleter;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "AnonymousRandomSeedGenerator";
-  
-  private Output<?> handle;
-  private Output<?> deleter;
-  
-  private AnonymousRandomSeedGenerator(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    handle = operation.output(outputIdx++);
-    deleter = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/AnonymousSeedGenerator.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/AnonymousSeedGenerator.java
deleted file mode 100644
index 2929a630525..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/AnonymousSeedGenerator.java
+++ /dev/null
@@ -1,78 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.random;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TBool;
-import org.tensorflow.types.TInt64;
-
-/**
- */
-public final class AnonymousSeedGenerator extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new AnonymousSeedGenerator operation.
-   * 
-   * @param scope current scope
-   * @param seed 
-   * @param seed2 
-   * @param reshuffle 
-   * @return a new instance of AnonymousSeedGenerator
-   */
-  @Endpoint(describeByClass = true)
-  public static AnonymousSeedGenerator create(Scope scope, Operand<TInt64> seed, Operand<TInt64> seed2, Operand<TBool> reshuffle) {
-    OperationBuilder opBuilder = scope.env().opBuilder("AnonymousSeedGenerator", scope.makeOpName("AnonymousSeedGenerator"));
-    opBuilder.addInput(seed.asOutput(scope));
-    opBuilder.addInput(seed2.asOutput(scope));
-    opBuilder.addInput(reshuffle.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new AnonymousSeedGenerator(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<?> handle() {
-    return handle;
-  }
-  
-  /**
-   */
-  public Output<?> deleter() {
-    return deleter;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "AnonymousSeedGenerator";
-  
-  private Output<?> handle;
-  private Output<?> deleter;
-  
-  private AnonymousSeedGenerator(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    handle = operation.output(outputIdx++);
-    deleter = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/DeleteRandomSeedGenerator.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/DeleteRandomSeedGenerator.java
deleted file mode 100644
index 9b86275f453..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/DeleteRandomSeedGenerator.java
+++ /dev/null
@@ -1,55 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.random;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-
-/**
- */
-public final class DeleteRandomSeedGenerator extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new DeleteRandomSeedGenerator operation.
-   * 
-   * @param scope current scope
-   * @param handle 
-   * @param deleter 
-   * @return a new instance of DeleteRandomSeedGenerator
-   */
-  @Endpoint(describeByClass = true)
-  public static DeleteRandomSeedGenerator create(Scope scope, Operand<?> handle, Operand<?> deleter) {
-    OperationBuilder opBuilder = scope.env().opBuilder("DeleteRandomSeedGenerator", scope.makeOpName("DeleteRandomSeedGenerator"));
-    opBuilder.addInput(handle.asOutput(scope));
-    opBuilder.addInput(deleter.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new DeleteRandomSeedGenerator(opBuilder.build());
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "DeleteRandomSeedGenerator";
-  
-  private DeleteRandomSeedGenerator(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/DeleteSeedGenerator.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/DeleteSeedGenerator.java
deleted file mode 100644
index 5bc2c0dd1b7..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/DeleteSeedGenerator.java
+++ /dev/null
@@ -1,55 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.random;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-
-/**
- */
-public final class DeleteSeedGenerator extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new DeleteSeedGenerator operation.
-   * 
-   * @param scope current scope
-   * @param handle 
-   * @param deleter 
-   * @return a new instance of DeleteSeedGenerator
-   */
-  @Endpoint(describeByClass = true)
-  public static DeleteSeedGenerator create(Scope scope, Operand<?> handle, Operand<?> deleter) {
-    OperationBuilder opBuilder = scope.env().opBuilder("DeleteSeedGenerator", scope.makeOpName("DeleteSeedGenerator"));
-    opBuilder.addInput(handle.asOutput(scope));
-    opBuilder.addInput(deleter.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new DeleteSeedGenerator(opBuilder.build());
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "DeleteSeedGenerator";
-  
-  private DeleteSeedGenerator(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/LogUniformCandidateSampler.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/LogUniformCandidateSampler.java
deleted file mode 100644
index 2e493705a72..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/LogUniformCandidateSampler.java
+++ /dev/null
@@ -1,171 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.random;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.TInt64;
-
-/**
- * Generates labels for candidate sampling with a log-uniform distribution.
- * <p>
- * See explanations of candidate sampling and the data formats at
- * go/candidate-sampling.
- * <p>
- * For each batch, this op picks a single set of sampled candidate labels.
- * <p>
- * The advantages of sampling candidates per-batch are simplicity and the
- * possibility of efficient dense matrix multiplication. The disadvantage is that
- * the sampled candidates must be chosen independently of the context and of the
- * true labels.
- */
-@Operator(group = "random")
-public final class LogUniformCandidateSampler extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.random.LogUniformCandidateSampler}
-   */
-  public static class Options {
-    
-    /**
-     * @param seed If either seed or seed2 are set to be non-zero, the random number
-     * generator is seeded by the given seed.  Otherwise, it is seeded by a
-     * random seed.
-     */
-    public Options seed(Long seed) {
-      this.seed = seed;
-      return this;
-    }
-    
-    /**
-     * @param seed2 An second seed to avoid seed collision.
-     */
-    public Options seed2(Long seed2) {
-      this.seed2 = seed2;
-      return this;
-    }
-    
-    private Long seed;
-    private Long seed2;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new LogUniformCandidateSampler operation.
-   * 
-   * @param scope current scope
-   * @param trueClasses A batch_size * num_true matrix, in which each row contains the
-   * IDs of the num_true target_classes in the corresponding original label.
-   * @param numTrue Number of true labels per context.
-   * @param numSampled Number of candidates to randomly sample.
-   * @param unique If unique is true, we sample with rejection, so that all sampled
-   * candidates in a batch are unique. This requires some approximation to
-   * estimate the post-rejection sampling probabilities.
-   * @param rangeMax The sampler will sample integers from the interval [0, range_max).
-   * @param options carries optional attributes values
-   * @return a new instance of LogUniformCandidateSampler
-   */
-  @Endpoint(describeByClass = true)
-  public static LogUniformCandidateSampler create(Scope scope, Operand<TInt64> trueClasses, Long numTrue, Long numSampled, Boolean unique, Long rangeMax, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("LogUniformCandidateSampler", scope.makeOpName("LogUniformCandidateSampler"));
-    opBuilder.addInput(trueClasses.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("num_true", numTrue);
-    opBuilder.setAttr("num_sampled", numSampled);
-    opBuilder.setAttr("unique", unique);
-    opBuilder.setAttr("range_max", rangeMax);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.seed != null) {
-          opBuilder.setAttr("seed", opts.seed);
-        }
-        if (opts.seed2 != null) {
-          opBuilder.setAttr("seed2", opts.seed2);
-        }
-      }
-    }
-    return new LogUniformCandidateSampler(opBuilder.build());
-  }
-  
-  /**
-   * @param seed If either seed or seed2 are set to be non-zero, the random number
-   * generator is seeded by the given seed.  Otherwise, it is seeded by a
-   * random seed.
-   */
-  public static Options seed(Long seed) {
-    return new Options().seed(seed);
-  }
-  
-  /**
-   * @param seed2 An second seed to avoid seed collision.
-   */
-  public static Options seed2(Long seed2) {
-    return new Options().seed2(seed2);
-  }
-  
-  /**
-   * A vector of length num_sampled, in which each element is
-   * the ID of a sampled candidate.
-   */
-  public Output<TInt64> sampledCandidates() {
-    return sampledCandidates;
-  }
-  
-  /**
-   * A batch_size * num_true matrix, representing
-   * the number of times each candidate is expected to occur in a batch
-   * of sampled candidates. If unique=true, then this is a probability.
-   */
-  public Output<TFloat32> trueExpectedCount() {
-    return trueExpectedCount;
-  }
-  
-  /**
-   * A vector of length num_sampled, for each sampled
-   * candidate representing the number of times the candidate is expected
-   * to occur in a batch of sampled candidates.  If unique=true, then this is a
-   * probability.
-   */
-  public Output<TFloat32> sampledExpectedCount() {
-    return sampledExpectedCount;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "LogUniformCandidateSampler";
-  
-  private Output<TInt64> sampledCandidates;
-  private Output<TFloat32> trueExpectedCount;
-  private Output<TFloat32> sampledExpectedCount;
-  
-  private LogUniformCandidateSampler(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    sampledCandidates = operation.output(outputIdx++);
-    trueExpectedCount = operation.output(outputIdx++);
-    sampledExpectedCount = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/Multinomial.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/Multinomial.java
deleted file mode 100644
index bdff3925130..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/Multinomial.java
+++ /dev/null
@@ -1,153 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.random;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Draws samples from a multinomial distribution.
- * 
- * @param <U> data type for {@code output()} output
- */
-@Operator(group = "random")
-public final class Multinomial<U extends TNumber> extends RawOp implements Operand<U> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.random.Multinomial}
-   */
-  public static class Options {
-    
-    /**
-     * @param seed If either seed or seed2 is set to be non-zero, the internal random number
-     * generator is seeded by the given seed.  Otherwise, a random seed is used.
-     */
-    public Options seed(Long seed) {
-      this.seed = seed;
-      return this;
-    }
-    
-    /**
-     * @param seed2 A second seed to avoid seed collision.
-     */
-    public Options seed2(Long seed2) {
-      this.seed2 = seed2;
-      return this;
-    }
-    
-    private Long seed;
-    private Long seed2;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new Multinomial operation.
-   * 
-   * @param scope current scope
-   * @param logits 2-D Tensor with shape `[batch_size, num_classes]`.  Each slice `[i, :]`
-   * represents the unnormalized log probabilities for all classes.
-   * @param numSamples 0-D.  Number of independent samples to draw for each row slice.
-   * @param outputDtype 
-   * @param options carries optional attributes values
-   * @return a new instance of Multinomial
-   */
-  @Endpoint(describeByClass = true)
-  public static <U extends TNumber, T extends TNumber> Multinomial<U> create(Scope scope, Operand<T> logits, Operand<TInt32> numSamples, Class<U> outputDtype, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Multinomial", scope.makeOpName("Multinomial"));
-    opBuilder.addInput(logits.asOutput(scope));
-    opBuilder.addInput(numSamples.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("output_dtype", outputDtype);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.seed != null) {
-          opBuilder.setAttr("seed", opts.seed);
-        }
-        if (opts.seed2 != null) {
-          opBuilder.setAttr("seed2", opts.seed2);
-        }
-      }
-    }
-    return new Multinomial<U>(opBuilder.build());
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new Multinomial operation using default output types.
-   * 
-   * @param scope current scope
-   * @param logits 2-D Tensor with shape `[batch_size, num_classes]`.  Each slice `[i, :]`
-   * represents the unnormalized log probabilities for all classes.
-   * @param numSamples 0-D.  Number of independent samples to draw for each row slice.
-   * @param options carries optional attributes values
-   * @return a new instance of Multinomial
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> Multinomial<TInt64> create(Scope scope, Operand<T> logits, Operand<TInt32> numSamples, Options... options) {
-    return create(scope, logits, numSamples, TInt64.class, options);
-  }
-  
-  /**
-   * @param seed If either seed or seed2 is set to be non-zero, the internal random number
-   * generator is seeded by the given seed.  Otherwise, a random seed is used.
-   */
-  public static Options seed(Long seed) {
-    return new Options().seed(seed);
-  }
-  
-  /**
-   * @param seed2 A second seed to avoid seed collision.
-   */
-  public static Options seed2(Long seed2) {
-    return new Options().seed2(seed2);
-  }
-  
-  /**
-   * 2-D Tensor with shape `[batch_size, num_samples]`.  Each slice `[i, :]`
-   * contains the drawn class labels with range `[0, num_classes)`.
-   */
-  public Output<U> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<U> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Multinomial";
-  
-  private Output<U> output;
-  
-  private Multinomial(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/NonDeterministicInts.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/NonDeterministicInts.java
deleted file mode 100644
index 21b3ab97a77..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/NonDeterministicInts.java
+++ /dev/null
@@ -1,91 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.random;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
-
-/**
- * Non-deterministically generates some integers.
- * <p>
- * This op may use some OS-provided source of non-determinism (e.g. an RNG), so each execution will give different results.
- * 
- * @param <U> data type for {@code output()} output
- */
-public final class NonDeterministicInts<U extends TType> extends RawOp implements Operand<U> {
-  
-  /**
-   * Factory method to create a class wrapping a new NonDeterministicInts operation.
-   * 
-   * @param scope current scope
-   * @param shape The shape of the output tensor.
-   * @param dtype The type of the output.
-   * @return a new instance of NonDeterministicInts
-   */
-  @Endpoint(describeByClass = true)
-  public static <U extends TType, T extends TType> NonDeterministicInts<U> create(Scope scope, Operand<T> shape, Class<U> dtype) {
-    OperationBuilder opBuilder = scope.env().opBuilder("NonDeterministicInts", scope.makeOpName("NonDeterministicInts"));
-    opBuilder.addInput(shape.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("dtype", dtype);
-    return new NonDeterministicInts<U>(opBuilder.build());
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new NonDeterministicInts operation using default output types.
-   * 
-   * @param scope current scope
-   * @param shape The shape of the output tensor.
-   * @return a new instance of NonDeterministicInts
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> NonDeterministicInts<TInt64> create(Scope scope, Operand<T> shape) {
-    return create(scope, shape, TInt64.class);
-  }
-  
-  /**
-   * Non-deterministic integer values with specified shape.
-   */
-  public Output<U> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<U> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "NonDeterministicInts";
-  
-  private Output<U> output;
-  
-  private NonDeterministicInts(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/ParameterizedTruncatedNormal.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/ParameterizedTruncatedNormal.java
deleted file mode 100644
index 76dc7d3417c..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/ParameterizedTruncatedNormal.java
+++ /dev/null
@@ -1,145 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.random;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Outputs random values from a normal distribution. The parameters may each be a
- * <p>
- * scalar which applies to the entire output, or a vector of length shape[0] which
- * stores the parameters for each batch.
- * 
- * @param <U> data type for {@code output()} output
- */
-@Operator(group = "random")
-public final class ParameterizedTruncatedNormal<U extends TNumber> extends RawOp implements Operand<U> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.random.ParameterizedTruncatedNormal}
-   */
-  public static class Options {
-    
-    /**
-     * @param seed If either `seed` or `seed2` are set to be non-zero, the random number
-     * generator is seeded by the given seed.  Otherwise, it is seeded by a
-     * random seed.
-     */
-    public Options seed(Long seed) {
-      this.seed = seed;
-      return this;
-    }
-    
-    /**
-     * @param seed2 A second seed to avoid seed collision.
-     */
-    public Options seed2(Long seed2) {
-      this.seed2 = seed2;
-      return this;
-    }
-    
-    private Long seed;
-    private Long seed2;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new ParameterizedTruncatedNormal operation.
-   * 
-   * @param scope current scope
-   * @param shape The shape of the output tensor. Batches are indexed by the 0th dimension.
-   * @param means The mean parameter of each batch.
-   * @param stdevs The standard deviation parameter of each batch. Must be greater than 0.
-   * @param minvals The minimum cutoff. May be -infinity.
-   * @param maxvals The maximum cutoff. May be +infinity, and must be more than the minval
-   * for each batch.
-   * @param options carries optional attributes values
-   * @return a new instance of ParameterizedTruncatedNormal
-   */
-  @Endpoint(describeByClass = true)
-  public static <U extends TNumber, T extends TNumber> ParameterizedTruncatedNormal<U> create(Scope scope, Operand<T> shape, Operand<U> means, Operand<U> stdevs, Operand<U> minvals, Operand<U> maxvals, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ParameterizedTruncatedNormal", scope.makeOpName("ParameterizedTruncatedNormal"));
-    opBuilder.addInput(shape.asOutput(scope));
-    opBuilder.addInput(means.asOutput(scope));
-    opBuilder.addInput(stdevs.asOutput(scope));
-    opBuilder.addInput(minvals.asOutput(scope));
-    opBuilder.addInput(maxvals.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.seed != null) {
-          opBuilder.setAttr("seed", opts.seed);
-        }
-        if (opts.seed2 != null) {
-          opBuilder.setAttr("seed2", opts.seed2);
-        }
-      }
-    }
-    return new ParameterizedTruncatedNormal<U>(opBuilder.build());
-  }
-  
-  /**
-   * @param seed If either `seed` or `seed2` are set to be non-zero, the random number
-   * generator is seeded by the given seed.  Otherwise, it is seeded by a
-   * random seed.
-   */
-  public static Options seed(Long seed) {
-    return new Options().seed(seed);
-  }
-  
-  /**
-   * @param seed2 A second seed to avoid seed collision.
-   */
-  public static Options seed2(Long seed2) {
-    return new Options().seed2(seed2);
-  }
-  
-  /**
-   * A matrix of shape num_batches x samples_per_batch, filled with random
-   * truncated normal values using the parameters for each row.
-   */
-  public Output<U> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<U> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ParameterizedTruncatedNormal";
-  
-  private Output<U> output;
-  
-  private ParameterizedTruncatedNormal(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RandomGamma.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RandomGamma.java
deleted file mode 100644
index 0f9fc373468..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RandomGamma.java
+++ /dev/null
@@ -1,142 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.random;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Outputs random values from the Gamma distribution(s) described by alpha.
- * <p>
- * This op uses the algorithm by Marsaglia et al. to acquire samples via
- * transformation-rejection from pairs of uniform and normal random variables.
- * See http://dl.acm.org/citation.cfm?id=358414
- * 
- * @param <U> data type for {@code output()} output
- */
-@Operator(group = "random")
-public final class RandomGamma<U extends TNumber> extends RawOp implements Operand<U> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.random.RandomGamma}
-   */
-  public static class Options {
-    
-    /**
-     * @param seed If either `seed` or `seed2` are set to be non-zero, the random number
-     * generator is seeded by the given seed.  Otherwise, it is seeded by a
-     * random seed.
-     */
-    public Options seed(Long seed) {
-      this.seed = seed;
-      return this;
-    }
-    
-    /**
-     * @param seed2 A second seed to avoid seed collision.
-     */
-    public Options seed2(Long seed2) {
-      this.seed2 = seed2;
-      return this;
-    }
-    
-    private Long seed;
-    private Long seed2;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new RandomGamma operation.
-   * 
-   * @param scope current scope
-   * @param shape 1-D integer tensor. Shape of independent samples to draw from each
-   * distribution described by the shape parameters given in alpha.
-   * @param alpha A tensor in which each scalar is a "shape" parameter describing the
-   * associated gamma distribution.
-   * @param options carries optional attributes values
-   * @return a new instance of RandomGamma
-   */
-  @Endpoint(describeByClass = true)
-  public static <U extends TNumber, T extends TNumber> RandomGamma<U> create(Scope scope, Operand<T> shape, Operand<U> alpha, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("RandomGamma", scope.makeOpName("RandomGamma"));
-    opBuilder.addInput(shape.asOutput(scope));
-    opBuilder.addInput(alpha.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.seed != null) {
-          opBuilder.setAttr("seed", opts.seed);
-        }
-        if (opts.seed2 != null) {
-          opBuilder.setAttr("seed2", opts.seed2);
-        }
-      }
-    }
-    return new RandomGamma<U>(opBuilder.build());
-  }
-  
-  /**
-   * @param seed If either `seed` or `seed2` are set to be non-zero, the random number
-   * generator is seeded by the given seed.  Otherwise, it is seeded by a
-   * random seed.
-   */
-  public static Options seed(Long seed) {
-    return new Options().seed(seed);
-  }
-  
-  /**
-   * @param seed2 A second seed to avoid seed collision.
-   */
-  public static Options seed2(Long seed2) {
-    return new Options().seed2(seed2);
-  }
-  
-  /**
-   * A tensor with shape `shape + shape(alpha)`. Each slice
-   * `[:, ..., :, i0, i1, ...iN]` contains the samples drawn for
-   * `alpha[i0, i1, ...iN]`. The dtype of the output matches the dtype of alpha.
-   */
-  public Output<U> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<U> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "RandomGamma";
-  
-  private Output<U> output;
-  
-  private RandomGamma(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RandomGammaGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RandomGammaGrad.java
deleted file mode 100644
index 822cb158ebd..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RandomGammaGrad.java
+++ /dev/null
@@ -1,75 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.random;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Computes the derivative of a Gamma random sample w.r.t. `alpha`.
- * 
- * @param <T> data type for {@code output()} output
- */
-public final class RandomGammaGrad<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new RandomGammaGrad operation.
-   * 
-   * @param scope current scope
-   * @param alpha 
-   * @param sample 
-   * @return a new instance of RandomGammaGrad
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> RandomGammaGrad<T> create(Scope scope, Operand<T> alpha, Operand<T> sample) {
-    OperationBuilder opBuilder = scope.env().opBuilder("RandomGammaGrad", scope.makeOpName("RandomGammaGrad"));
-    opBuilder.addInput(alpha.asOutput(scope));
-    opBuilder.addInput(sample.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new RandomGammaGrad<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "RandomGammaGrad";
-  
-  private Output<T> output;
-  
-  private RandomGammaGrad(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RandomPoisson.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RandomPoisson.java
deleted file mode 100644
index d0b62eae694..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RandomPoisson.java
+++ /dev/null
@@ -1,167 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.random;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Outputs random values from the Poisson distribution(s) described by rate.
- * <p>
- * This op uses two algorithms, depending on rate. If rate >= 10, then
- * the algorithm by Hormann is used to acquire samples via
- * transformation-rejection.
- * See http://www.sciencedirect.com/science/article/pii/0167668793909974.
- * <p>
- * Otherwise, Knuth's algorithm is used to acquire samples via multiplying uniform
- * random variables.
- * See Donald E. Knuth (1969). Seminumerical Algorithms. The Art of Computer
- * Programming, Volume 2. Addison Wesley
- * 
- * @param <V> data type for {@code output()} output
- */
-@Operator(group = "random")
-public final class RandomPoisson<V extends TNumber> extends RawOp implements Operand<V> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.random.RandomPoisson}
-   */
-  public static class Options {
-    
-    /**
-     * @param seed If either `seed` or `seed2` are set to be non-zero, the random number
-     * generator is seeded by the given seed.  Otherwise, it is seeded by a
-     * random seed.
-     */
-    public Options seed(Long seed) {
-      this.seed = seed;
-      return this;
-    }
-    
-    /**
-     * @param seed2 A second seed to avoid seed collision.
-     */
-    public Options seed2(Long seed2) {
-      this.seed2 = seed2;
-      return this;
-    }
-    
-    private Long seed;
-    private Long seed2;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new RandomPoisson operation.
-   * 
-   * @param scope current scope
-   * @param shape 1-D integer tensor. Shape of independent samples to draw from each
-   * distribution described by the shape parameters given in rate.
-   * @param rate A tensor in which each scalar is a "rate" parameter describing the
-   * associated poisson distribution.
-   * @param dtype 
-   * @param options carries optional attributes values
-   * @return a new instance of RandomPoisson
-   */
-  @Endpoint(describeByClass = true)
-  public static <V extends TNumber, T extends TNumber, U extends TNumber> RandomPoisson<V> create(Scope scope, Operand<T> shape, Operand<U> rate, Class<V> dtype, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("RandomPoissonV2", scope.makeOpName("RandomPoisson"));
-    opBuilder.addInput(shape.asOutput(scope));
-    opBuilder.addInput(rate.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("dtype", dtype);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.seed != null) {
-          opBuilder.setAttr("seed", opts.seed);
-        }
-        if (opts.seed2 != null) {
-          opBuilder.setAttr("seed2", opts.seed2);
-        }
-      }
-    }
-    return new RandomPoisson<V>(opBuilder.build());
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new RandomPoisson operation using default output types.
-   * 
-   * @param scope current scope
-   * @param shape 1-D integer tensor. Shape of independent samples to draw from each
-   * distribution described by the shape parameters given in rate.
-   * @param rate A tensor in which each scalar is a "rate" parameter describing the
-   * associated poisson distribution.
-   * @param options carries optional attributes values
-   * @return a new instance of RandomPoisson
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber, U extends TNumber> RandomPoisson<TInt64> create(Scope scope, Operand<T> shape, Operand<U> rate, Options... options) {
-    return create(scope, shape, rate, TInt64.class, options);
-  }
-  
-  /**
-   * @param seed If either `seed` or `seed2` are set to be non-zero, the random number
-   * generator is seeded by the given seed.  Otherwise, it is seeded by a
-   * random seed.
-   */
-  public static Options seed(Long seed) {
-    return new Options().seed(seed);
-  }
-  
-  /**
-   * @param seed2 A second seed to avoid seed collision.
-   */
-  public static Options seed2(Long seed2) {
-    return new Options().seed2(seed2);
-  }
-  
-  /**
-   * A tensor with shape `shape + shape(rate)`. Each slice
-   * `[:, ..., :, i0, i1, ...iN]` contains the samples drawn for
-   * `rate[i0, i1, ...iN]`.
-   */
-  public Output<V> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<V> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "RandomPoissonV2";
-  
-  private Output<V> output;
-  
-  private RandomPoisson(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RandomShuffle.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RandomShuffle.java
deleted file mode 100644
index 2d7adf20727..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RandomShuffle.java
+++ /dev/null
@@ -1,143 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.random;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Randomly shuffles a tensor along its first dimension.
- * <p>
- *   The tensor is shuffled along dimension 0, such that each `value[j]` is mapped
- *   to one and only one `output[i]`. For example, a mapping that might occur for a
- *   3x2 tensor is:
- * <pre>{@code
- * [[1, 2],       [[5, 6],
- *  [3, 4],  ==>   [1, 2],
- *  [5, 6]]        [3, 4]]
- * }</pre>
- * 
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator(group = "random")
-public final class RandomShuffle<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.random.RandomShuffle}
-   */
-  public static class Options {
-    
-    /**
-     * @param seed If either `seed` or `seed2` are set to be non-zero, the random number
-     * generator is seeded by the given seed.  Otherwise, it is seeded by a
-     * random seed.
-     */
-    public Options seed(Long seed) {
-      this.seed = seed;
-      return this;
-    }
-    
-    /**
-     * @param seed2 A second seed to avoid seed collision.
-     */
-    public Options seed2(Long seed2) {
-      this.seed2 = seed2;
-      return this;
-    }
-    
-    private Long seed;
-    private Long seed2;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new RandomShuffle operation.
-   * 
-   * @param scope current scope
-   * @param value The tensor to be shuffled.
-   * @param options carries optional attributes values
-   * @return a new instance of RandomShuffle
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> RandomShuffle<T> create(Scope scope, Operand<T> value, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("RandomShuffle", scope.makeOpName("RandomShuffle"));
-    opBuilder.addInput(value.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.seed != null) {
-          opBuilder.setAttr("seed", opts.seed);
-        }
-        if (opts.seed2 != null) {
-          opBuilder.setAttr("seed2", opts.seed2);
-        }
-      }
-    }
-    return new RandomShuffle<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param seed If either `seed` or `seed2` are set to be non-zero, the random number
-   * generator is seeded by the given seed.  Otherwise, it is seeded by a
-   * random seed.
-   */
-  public static Options seed(Long seed) {
-    return new Options().seed(seed);
-  }
-  
-  /**
-   * @param seed2 A second seed to avoid seed collision.
-   */
-  public static Options seed2(Long seed2) {
-    return new Options().seed2(seed2);
-  }
-  
-  /**
-   * A tensor of same shape and type as `value`, shuffled along its first
-   * dimension.
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "RandomShuffle";
-  
-  private Output<T> output;
-  
-  private RandomShuffle(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RandomStandardNormal.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RandomStandardNormal.java
deleted file mode 100644
index 0743f6860df..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RandomStandardNormal.java
+++ /dev/null
@@ -1,136 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.random;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Outputs random values from a normal distribution.
- * <p>
- * The generated values will have mean 0 and standard deviation 1.
- * 
- * @param <U> data type for {@code output()} output
- */
-@Operator(group = "random")
-public final class RandomStandardNormal<U extends TNumber> extends RawOp implements Operand<U> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.random.RandomStandardNormal}
-   */
-  public static class Options {
-    
-    /**
-     * @param seed If either `seed` or `seed2` are set to be non-zero, the random number
-     * generator is seeded by the given seed.  Otherwise, it is seeded by a
-     * random seed.
-     */
-    public Options seed(Long seed) {
-      this.seed = seed;
-      return this;
-    }
-    
-    /**
-     * @param seed2 A second seed to avoid seed collision.
-     */
-    public Options seed2(Long seed2) {
-      this.seed2 = seed2;
-      return this;
-    }
-    
-    private Long seed;
-    private Long seed2;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new RandomStandardNormal operation.
-   * 
-   * @param scope current scope
-   * @param shape The shape of the output tensor.
-   * @param dtype The type of the output.
-   * @param options carries optional attributes values
-   * @return a new instance of RandomStandardNormal
-   */
-  @Endpoint(describeByClass = true)
-  public static <U extends TNumber, T extends TNumber> RandomStandardNormal<U> create(Scope scope, Operand<T> shape, Class<U> dtype, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("RandomStandardNormal", scope.makeOpName("RandomStandardNormal"));
-    opBuilder.addInput(shape.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("dtype", dtype);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.seed != null) {
-          opBuilder.setAttr("seed", opts.seed);
-        }
-        if (opts.seed2 != null) {
-          opBuilder.setAttr("seed2", opts.seed2);
-        }
-      }
-    }
-    return new RandomStandardNormal<U>(opBuilder.build());
-  }
-  
-  /**
-   * @param seed If either `seed` or `seed2` are set to be non-zero, the random number
-   * generator is seeded by the given seed.  Otherwise, it is seeded by a
-   * random seed.
-   */
-  public static Options seed(Long seed) {
-    return new Options().seed(seed);
-  }
-  
-  /**
-   * @param seed2 A second seed to avoid seed collision.
-   */
-  public static Options seed2(Long seed2) {
-    return new Options().seed2(seed2);
-  }
-  
-  /**
-   * A tensor of the specified shape filled with random normal values.
-   */
-  public Output<U> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<U> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "RandomStandardNormal";
-  
-  private Output<U> output;
-  
-  private RandomStandardNormal(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RandomUniform.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RandomUniform.java
deleted file mode 100644
index fcd57fc4158..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RandomUniform.java
+++ /dev/null
@@ -1,137 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.random;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Outputs random values from a uniform distribution.
- * <p>
- * The generated values follow a uniform distribution in the range `[0, 1)`. The
- * lower bound 0 is included in the range, while the upper bound 1 is excluded.
- * 
- * @param <U> data type for {@code output()} output
- */
-@Operator(group = "random")
-public final class RandomUniform<U extends TNumber> extends RawOp implements Operand<U> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.random.RandomUniform}
-   */
-  public static class Options {
-    
-    /**
-     * @param seed If either `seed` or `seed2` are set to be non-zero, the random number
-     * generator is seeded by the given seed.  Otherwise, it is seeded by a
-     * random seed.
-     */
-    public Options seed(Long seed) {
-      this.seed = seed;
-      return this;
-    }
-    
-    /**
-     * @param seed2 A second seed to avoid seed collision.
-     */
-    public Options seed2(Long seed2) {
-      this.seed2 = seed2;
-      return this;
-    }
-    
-    private Long seed;
-    private Long seed2;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new RandomUniform operation.
-   * 
-   * @param scope current scope
-   * @param shape The shape of the output tensor.
-   * @param dtype The type of the output.
-   * @param options carries optional attributes values
-   * @return a new instance of RandomUniform
-   */
-  @Endpoint(describeByClass = true)
-  public static <U extends TNumber, T extends TNumber> RandomUniform<U> create(Scope scope, Operand<T> shape, Class<U> dtype, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("RandomUniform", scope.makeOpName("RandomUniform"));
-    opBuilder.addInput(shape.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("dtype", dtype);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.seed != null) {
-          opBuilder.setAttr("seed", opts.seed);
-        }
-        if (opts.seed2 != null) {
-          opBuilder.setAttr("seed2", opts.seed2);
-        }
-      }
-    }
-    return new RandomUniform<U>(opBuilder.build());
-  }
-  
-  /**
-   * @param seed If either `seed` or `seed2` are set to be non-zero, the random number
-   * generator is seeded by the given seed.  Otherwise, it is seeded by a
-   * random seed.
-   */
-  public static Options seed(Long seed) {
-    return new Options().seed(seed);
-  }
-  
-  /**
-   * @param seed2 A second seed to avoid seed collision.
-   */
-  public static Options seed2(Long seed2) {
-    return new Options().seed2(seed2);
-  }
-  
-  /**
-   * A tensor of the specified shape filled with uniform random values.
-   */
-  public Output<U> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<U> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "RandomUniform";
-  
-  private Output<U> output;
-  
-  private RandomUniform(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RandomUniformInt.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RandomUniformInt.java
deleted file mode 100644
index ecabd1fa6f8..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RandomUniformInt.java
+++ /dev/null
@@ -1,144 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.random;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Outputs random integers from a uniform distribution.
- * <p>
- * The generated values are uniform integers in the range `[minval, maxval)`.
- * The lower bound `minval` is included in the range, while the upper bound
- * `maxval` is excluded.
- * <p>
- * The random integers are slightly biased unless `maxval - minval` is an exact
- * power of two.  The bias is small for values of `maxval - minval` significantly
- * smaller than the range of the output (either `2^32` or `2^64`).
- * 
- * @param <U> data type for {@code output()} output
- */
-@Operator(group = "random")
-public final class RandomUniformInt<U extends TNumber> extends RawOp implements Operand<U> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.random.RandomUniformInt}
-   */
-  public static class Options {
-    
-    /**
-     * @param seed If either `seed` or `seed2` are set to be non-zero, the random number
-     * generator is seeded by the given seed.  Otherwise, it is seeded by a
-     * random seed.
-     */
-    public Options seed(Long seed) {
-      this.seed = seed;
-      return this;
-    }
-    
-    /**
-     * @param seed2 A second seed to avoid seed collision.
-     */
-    public Options seed2(Long seed2) {
-      this.seed2 = seed2;
-      return this;
-    }
-    
-    private Long seed;
-    private Long seed2;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new RandomUniformInt operation.
-   * 
-   * @param scope current scope
-   * @param shape The shape of the output tensor.
-   * @param minval 0-D.  Inclusive lower bound on the generated integers.
-   * @param maxval 0-D.  Exclusive upper bound on the generated integers.
-   * @param options carries optional attributes values
-   * @return a new instance of RandomUniformInt
-   */
-  @Endpoint(describeByClass = true)
-  public static <U extends TNumber, T extends TNumber> RandomUniformInt<U> create(Scope scope, Operand<T> shape, Operand<U> minval, Operand<U> maxval, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("RandomUniformInt", scope.makeOpName("RandomUniformInt"));
-    opBuilder.addInput(shape.asOutput(scope));
-    opBuilder.addInput(minval.asOutput(scope));
-    opBuilder.addInput(maxval.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.seed != null) {
-          opBuilder.setAttr("seed", opts.seed);
-        }
-        if (opts.seed2 != null) {
-          opBuilder.setAttr("seed2", opts.seed2);
-        }
-      }
-    }
-    return new RandomUniformInt<U>(opBuilder.build());
-  }
-  
-  /**
-   * @param seed If either `seed` or `seed2` are set to be non-zero, the random number
-   * generator is seeded by the given seed.  Otherwise, it is seeded by a
-   * random seed.
-   */
-  public static Options seed(Long seed) {
-    return new Options().seed(seed);
-  }
-  
-  /**
-   * @param seed2 A second seed to avoid seed collision.
-   */
-  public static Options seed2(Long seed2) {
-    return new Options().seed2(seed2);
-  }
-  
-  /**
-   * A tensor of the specified shape filled with uniform random integers.
-   */
-  public Output<U> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<U> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "RandomUniformInt";
-  
-  private Output<U> output;
-  
-  private RandomUniformInt(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RecordInput.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RecordInput.java
deleted file mode 100644
index b409b0f4674..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RecordInput.java
+++ /dev/null
@@ -1,206 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.random;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TString;
-
-/**
- * Emits randomized records.
- */
-@Operator(group = "random")
-public final class RecordInput extends RawOp implements Operand<TString> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.random.RecordInput}
-   */
-  public static class Options {
-    
-    /**
-     * @param fileRandomSeed Random seeds used to produce randomized records.
-     */
-    public Options fileRandomSeed(Long fileRandomSeed) {
-      this.fileRandomSeed = fileRandomSeed;
-      return this;
-    }
-    
-    /**
-     * @param fileShuffleShiftRatio Shifts the list of files after the list is randomly
-     * shuffled.
-     */
-    public Options fileShuffleShiftRatio(Float fileShuffleShiftRatio) {
-      this.fileShuffleShiftRatio = fileShuffleShiftRatio;
-      return this;
-    }
-    
-    /**
-     * @param fileBufferSize The randomization shuffling buffer.
-     */
-    public Options fileBufferSize(Long fileBufferSize) {
-      this.fileBufferSize = fileBufferSize;
-      return this;
-    }
-    
-    /**
-     * @param fileParallelism How many sstables are opened and concurrently iterated over.
-     */
-    public Options fileParallelism(Long fileParallelism) {
-      this.fileParallelism = fileParallelism;
-      return this;
-    }
-    
-    /**
-     * @param batchSize The batch size.
-     */
-    public Options batchSize(Long batchSize) {
-      this.batchSize = batchSize;
-      return this;
-    }
-    
-    /**
-     * @param compressionType The type of compression for the file. Currently ZLIB and
-     * GZIP are supported. Defaults to none.
-     */
-    public Options compressionType(String compressionType) {
-      this.compressionType = compressionType;
-      return this;
-    }
-    
-    private Long fileRandomSeed;
-    private Float fileShuffleShiftRatio;
-    private Long fileBufferSize;
-    private Long fileParallelism;
-    private Long batchSize;
-    private String compressionType;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new RecordInput operation.
-   * 
-   * @param scope current scope
-   * @param filePattern Glob pattern for the data files.
-   * @param options carries optional attributes values
-   * @return a new instance of RecordInput
-   */
-  @Endpoint(describeByClass = true)
-  public static RecordInput create(Scope scope, String filePattern, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("RecordInput", scope.makeOpName("RecordInput"));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("file_pattern", filePattern);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.fileRandomSeed != null) {
-          opBuilder.setAttr("file_random_seed", opts.fileRandomSeed);
-        }
-        if (opts.fileShuffleShiftRatio != null) {
-          opBuilder.setAttr("file_shuffle_shift_ratio", opts.fileShuffleShiftRatio);
-        }
-        if (opts.fileBufferSize != null) {
-          opBuilder.setAttr("file_buffer_size", opts.fileBufferSize);
-        }
-        if (opts.fileParallelism != null) {
-          opBuilder.setAttr("file_parallelism", opts.fileParallelism);
-        }
-        if (opts.batchSize != null) {
-          opBuilder.setAttr("batch_size", opts.batchSize);
-        }
-        if (opts.compressionType != null) {
-          opBuilder.setAttr("compression_type", opts.compressionType);
-        }
-      }
-    }
-    return new RecordInput(opBuilder.build());
-  }
-  
-  /**
-   * @param fileRandomSeed Random seeds used to produce randomized records.
-   */
-  public static Options fileRandomSeed(Long fileRandomSeed) {
-    return new Options().fileRandomSeed(fileRandomSeed);
-  }
-  
-  /**
-   * @param fileShuffleShiftRatio Shifts the list of files after the list is randomly
-   * shuffled.
-   */
-  public static Options fileShuffleShiftRatio(Float fileShuffleShiftRatio) {
-    return new Options().fileShuffleShiftRatio(fileShuffleShiftRatio);
-  }
-  
-  /**
-   * @param fileBufferSize The randomization shuffling buffer.
-   */
-  public static Options fileBufferSize(Long fileBufferSize) {
-    return new Options().fileBufferSize(fileBufferSize);
-  }
-  
-  /**
-   * @param fileParallelism How many sstables are opened and concurrently iterated over.
-   */
-  public static Options fileParallelism(Long fileParallelism) {
-    return new Options().fileParallelism(fileParallelism);
-  }
-  
-  /**
-   * @param batchSize The batch size.
-   */
-  public static Options batchSize(Long batchSize) {
-    return new Options().batchSize(batchSize);
-  }
-  
-  /**
-   * @param compressionType The type of compression for the file. Currently ZLIB and
-   * GZIP are supported. Defaults to none.
-   */
-  public static Options compressionType(String compressionType) {
-    return new Options().compressionType(compressionType);
-  }
-  
-  /**
-   * A tensor of shape [batch_size].
-   */
-  public Output<TString> records() {
-    return records;
-  }
-  
-  @Override
-  public Output<TString> asOutput(Scope scope) {
-    return records;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "RecordInput";
-  
-  private Output<TString> records;
-  
-  private RecordInput(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    records = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RngSkip.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RngSkip.java
deleted file mode 100644
index a353b0e861c..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RngSkip.java
+++ /dev/null
@@ -1,64 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.random;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt64;
-
-/**
- * Advance the counter of a counter-based RNG.
- * <p>
- * The state of the RNG after
- * `rng_skip(n)` will be the same as that after `stateful_uniform([n])`
- * (or any other distribution). The actual increment added to the
- * counter is an unspecified implementation detail.
- */
-public final class RngSkip extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new RngSkip operation.
-   * 
-   * @param scope current scope
-   * @param resource The handle of the resource variable that stores the state of the RNG.
-   * @param algorithm The RNG algorithm.
-   * @param delta The amount of advancement.
-   * @return a new instance of RngSkip
-   */
-  @Endpoint(describeByClass = true)
-  public static RngSkip create(Scope scope, Operand<?> resource, Operand<TInt64> algorithm, Operand<TInt64> delta) {
-    OperationBuilder opBuilder = scope.env().opBuilder("RngSkip", scope.makeOpName("RngSkip"));
-    opBuilder.addInput(resource.asOutput(scope));
-    opBuilder.addInput(algorithm.asOutput(scope));
-    opBuilder.addInput(delta.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new RngSkip(opBuilder.build());
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "RngSkip";
-  
-  private RngSkip(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatefulRandomBinomial.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatefulRandomBinomial.java
deleted file mode 100644
index cb2f012ca95..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatefulRandomBinomial.java
+++ /dev/null
@@ -1,99 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.random;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * @param <V> data type for {@code output()} output
- */
-@Operator(group = "random")
-public final class StatefulRandomBinomial<V extends TNumber> extends RawOp implements Operand<V> {
-  
-  /**
-   * Factory method to create a class wrapping a new StatefulRandomBinomial operation.
-   * 
-   * @param scope current scope
-   * @param resource 
-   * @param algorithm 
-   * @param shape 
-   * @param counts 
-   * @param probs 
-   * @param dtype 
-   * @return a new instance of StatefulRandomBinomial
-   */
-  @Endpoint(describeByClass = true)
-  public static <V extends TNumber, T extends TNumber, U extends TNumber> StatefulRandomBinomial<V> create(Scope scope, Operand<?> resource, Operand<TInt64> algorithm, Operand<T> shape, Operand<U> counts, Operand<U> probs, Class<V> dtype) {
-    OperationBuilder opBuilder = scope.env().opBuilder("StatefulRandomBinomial", scope.makeOpName("StatefulRandomBinomial"));
-    opBuilder.addInput(resource.asOutput(scope));
-    opBuilder.addInput(algorithm.asOutput(scope));
-    opBuilder.addInput(shape.asOutput(scope));
-    opBuilder.addInput(counts.asOutput(scope));
-    opBuilder.addInput(probs.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("dtype", dtype);
-    return new StatefulRandomBinomial<V>(opBuilder.build());
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new StatefulRandomBinomial operation using default output types.
-   * 
-   * @param scope current scope
-   * @param resource 
-   * @param algorithm 
-   * @param shape 
-   * @param counts 
-   * @param probs 
-   * @return a new instance of StatefulRandomBinomial
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber, U extends TNumber> StatefulRandomBinomial<TInt64> create(Scope scope, Operand<?> resource, Operand<TInt64> algorithm, Operand<T> shape, Operand<U> counts, Operand<U> probs) {
-    return create(scope, resource, algorithm, shape, counts, probs, TInt64.class);
-  }
-  
-  /**
-   */
-  public Output<V> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<V> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "StatefulRandomBinomial";
-  
-  private Output<V> output;
-  
-  private StatefulRandomBinomial(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatefulStandardNormal.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatefulStandardNormal.java
deleted file mode 100644
index 779bd1b55b8..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatefulStandardNormal.java
+++ /dev/null
@@ -1,99 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.random;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
-
-/**
- * Outputs random values from a normal distribution.
- * <p>
- * The generated values will have mean 0 and standard deviation 1.
- * 
- * @param <U> data type for {@code output()} output
- */
-@Operator(group = "random")
-public final class StatefulStandardNormal<U extends TType> extends RawOp implements Operand<U> {
-  
-  /**
-   * Factory method to create a class wrapping a new StatefulStandardNormal operation.
-   * 
-   * @param scope current scope
-   * @param resource The handle of the resource variable that stores the state of the RNG.
-   * @param algorithm The RNG algorithm.
-   * @param shape The shape of the output tensor.
-   * @param dtype The type of the output.
-   * @return a new instance of StatefulStandardNormal
-   */
-  @Endpoint(describeByClass = true)
-  public static <U extends TType, T extends TType> StatefulStandardNormal<U> create(Scope scope, Operand<?> resource, Operand<TInt64> algorithm, Operand<T> shape, Class<U> dtype) {
-    OperationBuilder opBuilder = scope.env().opBuilder("StatefulStandardNormalV2", scope.makeOpName("StatefulStandardNormal"));
-    opBuilder.addInput(resource.asOutput(scope));
-    opBuilder.addInput(algorithm.asOutput(scope));
-    opBuilder.addInput(shape.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("dtype", dtype);
-    return new StatefulStandardNormal<U>(opBuilder.build());
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new StatefulStandardNormal operation using default output types.
-   * 
-   * @param scope current scope
-   * @param resource The handle of the resource variable that stores the state of the RNG.
-   * @param algorithm The RNG algorithm.
-   * @param shape The shape of the output tensor.
-   * @return a new instance of StatefulStandardNormal
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> StatefulStandardNormal<TFloat32> create(Scope scope, Operand<?> resource, Operand<TInt64> algorithm, Operand<T> shape) {
-    return create(scope, resource, algorithm, shape, TFloat32.class);
-  }
-  
-  /**
-   * A tensor of the specified shape filled with random normal values.
-   */
-  public Output<U> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<U> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "StatefulStandardNormalV2";
-  
-  private Output<U> output;
-  
-  private StatefulStandardNormal(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatefulTruncatedNormal.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatefulTruncatedNormal.java
deleted file mode 100644
index 807c6ed335c..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatefulTruncatedNormal.java
+++ /dev/null
@@ -1,100 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.random;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
-
-/**
- * Outputs random values from a truncated normal distribution.
- * <p>
- * The generated values follow a normal distribution with mean 0 and standard
- * deviation 1, except that values whose magnitude is more than 2 standard
- * deviations from the mean are dropped and re-picked.
- * 
- * @param <U> data type for {@code output()} output
- */
-public final class StatefulTruncatedNormal<U extends TType> extends RawOp implements Operand<U> {
-  
-  /**
-   * Factory method to create a class wrapping a new StatefulTruncatedNormal operation.
-   * 
-   * @param scope current scope
-   * @param resource The handle of the resource variable that stores the state of the RNG.
-   * @param algorithm The RNG algorithm.
-   * @param shape The shape of the output tensor.
-   * @param dtype The type of the output.
-   * @return a new instance of StatefulTruncatedNormal
-   */
-  @Endpoint(describeByClass = true)
-  public static <U extends TType, T extends TType> StatefulTruncatedNormal<U> create(Scope scope, Operand<?> resource, Operand<TInt64> algorithm, Operand<T> shape, Class<U> dtype) {
-    OperationBuilder opBuilder = scope.env().opBuilder("StatefulTruncatedNormal", scope.makeOpName("StatefulTruncatedNormal"));
-    opBuilder.addInput(resource.asOutput(scope));
-    opBuilder.addInput(algorithm.asOutput(scope));
-    opBuilder.addInput(shape.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("dtype", dtype);
-    return new StatefulTruncatedNormal<U>(opBuilder.build());
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new StatefulTruncatedNormal operation using default output types.
-   * 
-   * @param scope current scope
-   * @param resource The handle of the resource variable that stores the state of the RNG.
-   * @param algorithm The RNG algorithm.
-   * @param shape The shape of the output tensor.
-   * @return a new instance of StatefulTruncatedNormal
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> StatefulTruncatedNormal<TFloat32> create(Scope scope, Operand<?> resource, Operand<TInt64> algorithm, Operand<T> shape) {
-    return create(scope, resource, algorithm, shape, TFloat32.class);
-  }
-  
-  /**
-   * Random values with specified shape.
-   */
-  public Output<U> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<U> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "StatefulTruncatedNormal";
-  
-  private Output<U> output;
-  
-  private StatefulTruncatedNormal(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatefulUniform.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatefulUniform.java
deleted file mode 100644
index bd8d035be1a..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatefulUniform.java
+++ /dev/null
@@ -1,99 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.random;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
-
-/**
- * Outputs random values from a uniform distribution.
- * <p>
- * The generated values follow a uniform distribution in the range `[0, 1)`. The
- * lower bound 0 is included in the range, while the upper bound 1 is excluded.
- * 
- * @param <U> data type for {@code output()} output
- */
-public final class StatefulUniform<U extends TType> extends RawOp implements Operand<U> {
-  
-  /**
-   * Factory method to create a class wrapping a new StatefulUniform operation.
-   * 
-   * @param scope current scope
-   * @param resource The handle of the resource variable that stores the state of the RNG.
-   * @param algorithm The RNG algorithm.
-   * @param shape The shape of the output tensor.
-   * @param dtype The type of the output.
-   * @return a new instance of StatefulUniform
-   */
-  @Endpoint(describeByClass = true)
-  public static <U extends TType, T extends TType> StatefulUniform<U> create(Scope scope, Operand<?> resource, Operand<TInt64> algorithm, Operand<T> shape, Class<U> dtype) {
-    OperationBuilder opBuilder = scope.env().opBuilder("StatefulUniform", scope.makeOpName("StatefulUniform"));
-    opBuilder.addInput(resource.asOutput(scope));
-    opBuilder.addInput(algorithm.asOutput(scope));
-    opBuilder.addInput(shape.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("dtype", dtype);
-    return new StatefulUniform<U>(opBuilder.build());
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new StatefulUniform operation using default output types.
-   * 
-   * @param scope current scope
-   * @param resource The handle of the resource variable that stores the state of the RNG.
-   * @param algorithm The RNG algorithm.
-   * @param shape The shape of the output tensor.
-   * @return a new instance of StatefulUniform
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> StatefulUniform<TFloat32> create(Scope scope, Operand<?> resource, Operand<TInt64> algorithm, Operand<T> shape) {
-    return create(scope, resource, algorithm, shape, TFloat32.class);
-  }
-  
-  /**
-   * Random values with specified shape.
-   */
-  public Output<U> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<U> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "StatefulUniform";
-  
-  private Output<U> output;
-  
-  private StatefulUniform(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatefulUniformFullInt.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatefulUniformFullInt.java
deleted file mode 100644
index 722764bfb7c..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatefulUniformFullInt.java
+++ /dev/null
@@ -1,83 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.random;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
-
-/**
- * Outputs random integers from a uniform distribution.
- * <p>
- * The generated values are uniform integers covering the whole range of `dtype`.
- * 
- * @param <U> data type for {@code output()} output
- */
-public final class StatefulUniformFullInt<U extends TType> extends RawOp implements Operand<U> {
-  
-  /**
-   * Factory method to create a class wrapping a new StatefulUniformFullInt operation.
-   * 
-   * @param scope current scope
-   * @param resource The handle of the resource variable that stores the state of the RNG.
-   * @param algorithm The RNG algorithm.
-   * @param shape The shape of the output tensor.
-   * @param dtype The type of the output.
-   * @return a new instance of StatefulUniformFullInt
-   */
-  @Endpoint(describeByClass = true)
-  public static <U extends TType, T extends TType> StatefulUniformFullInt<U> create(Scope scope, Operand<?> resource, Operand<TInt64> algorithm, Operand<T> shape, Class<U> dtype) {
-    OperationBuilder opBuilder = scope.env().opBuilder("StatefulUniformFullInt", scope.makeOpName("StatefulUniformFullInt"));
-    opBuilder.addInput(resource.asOutput(scope));
-    opBuilder.addInput(algorithm.asOutput(scope));
-    opBuilder.addInput(shape.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("dtype", dtype);
-    return new StatefulUniformFullInt<U>(opBuilder.build());
-  }
-  
-  /**
-   * Random values with specified shape.
-   */
-  public Output<U> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<U> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "StatefulUniformFullInt";
-  
-  private Output<U> output;
-  
-  private StatefulUniformFullInt(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatefulUniformInt.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatefulUniformInt.java
deleted file mode 100644
index 14edb24fd00..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatefulUniformInt.java
+++ /dev/null
@@ -1,91 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.random;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
-
-/**
- * Outputs random integers from a uniform distribution.
- * <p>
- * The generated values are uniform integers in the range `[minval, maxval)`.
- * The lower bound `minval` is included in the range, while the upper bound
- * `maxval` is excluded.
- * <p>
- * The random integers are slightly biased unless `maxval - minval` is an exact
- * power of two.  The bias is small for values of `maxval - minval` significantly
- * smaller than the range of the output (either `2^32` or `2^64`).
- * 
- * @param <U> data type for {@code output()} output
- */
-public final class StatefulUniformInt<U extends TType> extends RawOp implements Operand<U> {
-  
-  /**
-   * Factory method to create a class wrapping a new StatefulUniformInt operation.
-   * 
-   * @param scope current scope
-   * @param resource The handle of the resource variable that stores the state of the RNG.
-   * @param algorithm The RNG algorithm.
-   * @param shape The shape of the output tensor.
-   * @param minval Minimum value (inclusive, scalar).
-   * @param maxval Maximum value (exclusive, scalar).
-   * @return a new instance of StatefulUniformInt
-   */
-  @Endpoint(describeByClass = true)
-  public static <U extends TType, T extends TType> StatefulUniformInt<U> create(Scope scope, Operand<?> resource, Operand<TInt64> algorithm, Operand<T> shape, Operand<U> minval, Operand<U> maxval) {
-    OperationBuilder opBuilder = scope.env().opBuilder("StatefulUniformInt", scope.makeOpName("StatefulUniformInt"));
-    opBuilder.addInput(resource.asOutput(scope));
-    opBuilder.addInput(algorithm.asOutput(scope));
-    opBuilder.addInput(shape.asOutput(scope));
-    opBuilder.addInput(minval.asOutput(scope));
-    opBuilder.addInput(maxval.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new StatefulUniformInt<U>(opBuilder.build());
-  }
-  
-  /**
-   * Random values with specified shape.
-   */
-  public Output<U> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<U> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "StatefulUniformInt";
-  
-  private Output<U> output;
-  
-  private StatefulUniformInt(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessMultinomial.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessMultinomial.java
deleted file mode 100644
index 442d7b9db0e..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessMultinomial.java
+++ /dev/null
@@ -1,100 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.random;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Draws samples from a multinomial distribution.
- * 
- * @param <V> data type for {@code output()} output
- */
-@Operator(group = "random")
-public final class StatelessMultinomial<V extends TNumber> extends RawOp implements Operand<V> {
-  
-  /**
-   * Factory method to create a class wrapping a new StatelessMultinomial operation.
-   * 
-   * @param scope current scope
-   * @param logits 2-D Tensor with shape `[batch_size, num_classes]`.  Each slice `[i, :]`
-   * represents the unnormalized log probabilities for all classes.
-   * @param numSamples 0-D.  Number of independent samples to draw for each row slice.
-   * @param seed 2 seeds (shape [2]).
-   * @param outputDtype 
-   * @return a new instance of StatelessMultinomial
-   */
-  @Endpoint(describeByClass = true)
-  public static <V extends TNumber, T extends TNumber, U extends TNumber> StatelessMultinomial<V> create(Scope scope, Operand<T> logits, Operand<TInt32> numSamples, Operand<U> seed, Class<V> outputDtype) {
-    OperationBuilder opBuilder = scope.env().opBuilder("StatelessMultinomial", scope.makeOpName("StatelessMultinomial"));
-    opBuilder.addInput(logits.asOutput(scope));
-    opBuilder.addInput(numSamples.asOutput(scope));
-    opBuilder.addInput(seed.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("output_dtype", outputDtype);
-    return new StatelessMultinomial<V>(opBuilder.build());
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new StatelessMultinomial operation using default output types.
-   * 
-   * @param scope current scope
-   * @param logits 2-D Tensor with shape `[batch_size, num_classes]`.  Each slice `[i, :]`
-   * represents the unnormalized log probabilities for all classes.
-   * @param numSamples 0-D.  Number of independent samples to draw for each row slice.
-   * @param seed 2 seeds (shape [2]).
-   * @return a new instance of StatelessMultinomial
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber, U extends TNumber> StatelessMultinomial<TInt64> create(Scope scope, Operand<T> logits, Operand<TInt32> numSamples, Operand<U> seed) {
-    return create(scope, logits, numSamples, seed, TInt64.class);
-  }
-  
-  /**
-   * 2-D Tensor with shape `[batch_size, num_samples]`.  Each slice `[i, :]`
-   * contains the drawn class labels with range `[0, num_classes)`.
-   */
-  public Output<V> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<V> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "StatelessMultinomial";
-  
-  private Output<V> output;
-  
-  private StatelessMultinomial(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessParameterizedTruncatedNormal.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessParameterizedTruncatedNormal.java
deleted file mode 100644
index 6ad43c17fe2..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessParameterizedTruncatedNormal.java
+++ /dev/null
@@ -1,84 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.random;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * @param <V> data type for {@code output()} output
- */
-public final class StatelessParameterizedTruncatedNormal<V extends TNumber> extends RawOp implements Operand<V> {
-  
-  /**
-   * Factory method to create a class wrapping a new StatelessParameterizedTruncatedNormal operation.
-   * 
-   * @param scope current scope
-   * @param shape The shape of the output tensor.
-   * @param seed 2 seeds (shape [2]).
-   * @param means The mean parameter of each batch.
-   * @param stddevs The standard deviation parameter of each batch. Must be greater than 0.
-   * @param minvals The minimum cutoff. May be -infinity.
-   * @param maxvals The maximum cutoff. May be +infinity, and must be more than the minval
-   * for each batch.
-   * @return a new instance of StatelessParameterizedTruncatedNormal
-   */
-  @Endpoint(describeByClass = true)
-  public static <V extends TNumber, T extends TNumber, U extends TNumber> StatelessParameterizedTruncatedNormal<V> create(Scope scope, Operand<T> shape, Operand<U> seed, Operand<V> means, Operand<V> stddevs, Operand<V> minvals, Operand<V> maxvals) {
-    OperationBuilder opBuilder = scope.env().opBuilder("StatelessParameterizedTruncatedNormal", scope.makeOpName("StatelessParameterizedTruncatedNormal"));
-    opBuilder.addInput(shape.asOutput(scope));
-    opBuilder.addInput(seed.asOutput(scope));
-    opBuilder.addInput(means.asOutput(scope));
-    opBuilder.addInput(stddevs.asOutput(scope));
-    opBuilder.addInput(minvals.asOutput(scope));
-    opBuilder.addInput(maxvals.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new StatelessParameterizedTruncatedNormal<V>(opBuilder.build());
-  }
-  
-  /**
-   * The outputs are truncated normal samples and are a deterministic function of
-   * `shape`, `seed`, `minvals`, `maxvals`, `means` and `stddevs`.
-   */
-  public Output<V> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<V> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "StatelessParameterizedTruncatedNormal";
-  
-  private Output<V> output;
-  
-  private StatelessParameterizedTruncatedNormal(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessRandomBinomial.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessRandomBinomial.java
deleted file mode 100644
index 870f995c6f0..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessRandomBinomial.java
+++ /dev/null
@@ -1,106 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.random;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Outputs deterministic pseudorandom random numbers from a binomial distribution.
- * <p>
- * Outputs random values from a binomial distribution.
- * <p>
- * The outputs are a deterministic function of `shape`, `seed`, `counts`, and `probs`.
- * 
- * @param <W> data type for {@code output()} output
- */
-public final class StatelessRandomBinomial<W extends TNumber> extends RawOp implements Operand<W> {
-  
-  /**
-   * Factory method to create a class wrapping a new StatelessRandomBinomial operation.
-   * 
-   * @param scope current scope
-   * @param shape The shape of the output tensor.
-   * @param seed 2 seeds (shape [2]).
-   * @param counts The counts of the binomial distribution. Must be broadcastable with `probs`,
-   * and broadcastable with the rightmost dimensions of `shape`.
-   * @param probs The probability of success for the binomial distribution. Must be broadcastable
-   * with `counts` and broadcastable with the rightmost dimensions of `shape`.
-   * @param dtype The type of the output.
-   * @return a new instance of StatelessRandomBinomial
-   */
-  @Endpoint(describeByClass = true)
-  public static <W extends TNumber, T extends TNumber, U extends TNumber, V extends TNumber> StatelessRandomBinomial<W> create(Scope scope, Operand<T> shape, Operand<U> seed, Operand<V> counts, Operand<V> probs, Class<W> dtype) {
-    OperationBuilder opBuilder = scope.env().opBuilder("StatelessRandomBinomial", scope.makeOpName("StatelessRandomBinomial"));
-    opBuilder.addInput(shape.asOutput(scope));
-    opBuilder.addInput(seed.asOutput(scope));
-    opBuilder.addInput(counts.asOutput(scope));
-    opBuilder.addInput(probs.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("dtype", dtype);
-    return new StatelessRandomBinomial<W>(opBuilder.build());
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new StatelessRandomBinomial operation using default output types.
-   * 
-   * @param scope current scope
-   * @param shape The shape of the output tensor.
-   * @param seed 2 seeds (shape [2]).
-   * @param counts The counts of the binomial distribution. Must be broadcastable with `probs`,
-   * and broadcastable with the rightmost dimensions of `shape`.
-   * @param probs The probability of success for the binomial distribution. Must be broadcastable
-   * with `counts` and broadcastable with the rightmost dimensions of `shape`.
-   * @return a new instance of StatelessRandomBinomial
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber, U extends TNumber, V extends TNumber> StatelessRandomBinomial<TInt64> create(Scope scope, Operand<T> shape, Operand<U> seed, Operand<V> counts, Operand<V> probs) {
-    return create(scope, shape, seed, counts, probs, TInt64.class);
-  }
-  
-  /**
-   * Random values with specified shape.
-   */
-  public Output<W> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<W> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "StatelessRandomBinomial";
-  
-  private Output<W> output;
-  
-  private StatelessRandomBinomial(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessRandomGamma.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessRandomGamma.java
deleted file mode 100644
index 30800534761..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessRandomGamma.java
+++ /dev/null
@@ -1,83 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.random;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Outputs deterministic pseudorandom random numbers from a gamma distribution.
- * <p>
- * Outputs random values from a gamma distribution.
- * <p>
- * The outputs are a deterministic function of `shape`, `seed`, and `alpha`.
- * 
- * @param <V> data type for {@code output()} output
- */
-public final class StatelessRandomGamma<V extends TNumber> extends RawOp implements Operand<V> {
-  
-  /**
-   * Factory method to create a class wrapping a new StatelessRandomGamma operation.
-   * 
-   * @param scope current scope
-   * @param shape The shape of the output tensor.
-   * @param seed 2 seeds (shape [2]).
-   * @param alpha The concentration of the gamma distribution. Shape must match the rightmost
-   * dimensions of `shape`.
-   * @return a new instance of StatelessRandomGamma
-   */
-  @Endpoint(describeByClass = true)
-  public static <V extends TNumber, T extends TNumber, U extends TNumber> StatelessRandomGamma<V> create(Scope scope, Operand<T> shape, Operand<U> seed, Operand<V> alpha) {
-    OperationBuilder opBuilder = scope.env().opBuilder("StatelessRandomGammaV2", scope.makeOpName("StatelessRandomGamma"));
-    opBuilder.addInput(shape.asOutput(scope));
-    opBuilder.addInput(seed.asOutput(scope));
-    opBuilder.addInput(alpha.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new StatelessRandomGamma<V>(opBuilder.build());
-  }
-  
-  /**
-   * Random values with specified shape.
-   */
-  public Output<V> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<V> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "StatelessRandomGammaV2";
-  
-  private Output<V> output;
-  
-  private StatelessRandomGamma(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessRandomNormal.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessRandomNormal.java
deleted file mode 100644
index bd7eb1e5fe8..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessRandomNormal.java
+++ /dev/null
@@ -1,97 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.random;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Outputs deterministic pseudorandom values from a normal distribution.
- * <p>
- * The generated values will have mean 0 and standard deviation 1.
- * <p>
- * The outputs are a deterministic function of `shape` and `seed`.
- * 
- * @param <V> data type for {@code output()} output
- */
-@Operator(group = "random")
-public final class StatelessRandomNormal<V extends TNumber> extends RawOp implements Operand<V> {
-  
-  /**
-   * Factory method to create a class wrapping a new StatelessRandomNormal operation.
-   * 
-   * @param scope current scope
-   * @param shape The shape of the output tensor.
-   * @param seed 2 seeds (shape [2]).
-   * @param dtype The type of the output.
-   * @return a new instance of StatelessRandomNormal
-   */
-  @Endpoint(describeByClass = true)
-  public static <V extends TNumber, T extends TNumber, U extends TNumber> StatelessRandomNormal<V> create(Scope scope, Operand<T> shape, Operand<U> seed, Class<V> dtype) {
-    OperationBuilder opBuilder = scope.env().opBuilder("StatelessRandomNormal", scope.makeOpName("StatelessRandomNormal"));
-    opBuilder.addInput(shape.asOutput(scope));
-    opBuilder.addInput(seed.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("dtype", dtype);
-    return new StatelessRandomNormal<V>(opBuilder.build());
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new StatelessRandomNormal operation using default output types.
-   * 
-   * @param scope current scope
-   * @param shape The shape of the output tensor.
-   * @param seed 2 seeds (shape [2]).
-   * @return a new instance of StatelessRandomNormal
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber, U extends TNumber> StatelessRandomNormal<TFloat32> create(Scope scope, Operand<T> shape, Operand<U> seed) {
-    return create(scope, shape, seed, TFloat32.class);
-  }
-  
-  /**
-   * Random values with specified shape.
-   */
-  public Output<V> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<V> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "StatelessRandomNormal";
-  
-  private Output<V> output;
-  
-  private StatelessRandomNormal(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessRandomPoisson.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessRandomPoisson.java
deleted file mode 100644
index 13cb97ada3c..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessRandomPoisson.java
+++ /dev/null
@@ -1,85 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.random;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Outputs deterministic pseudorandom random numbers from a Poisson distribution.
- * <p>
- * Outputs random values from a Poisson distribution.
- * <p>
- * The outputs are a deterministic function of `shape`, `seed`, and `lam`.
- * 
- * @param <W> data type for {@code output()} output
- */
-public final class StatelessRandomPoisson<W extends TNumber> extends RawOp implements Operand<W> {
-  
-  /**
-   * Factory method to create a class wrapping a new StatelessRandomPoisson operation.
-   * 
-   * @param scope current scope
-   * @param shape The shape of the output tensor.
-   * @param seed 2 seeds (shape [2]).
-   * @param lam The rate of the Poisson distribution. Shape must match the rightmost dimensions
-   * of `shape`.
-   * @param dtype The type of the output.
-   * @return a new instance of StatelessRandomPoisson
-   */
-  @Endpoint(describeByClass = true)
-  public static <W extends TNumber, T extends TNumber, U extends TNumber, V extends TNumber> StatelessRandomPoisson<W> create(Scope scope, Operand<T> shape, Operand<U> seed, Operand<V> lam, Class<W> dtype) {
-    OperationBuilder opBuilder = scope.env().opBuilder("StatelessRandomPoisson", scope.makeOpName("StatelessRandomPoisson"));
-    opBuilder.addInput(shape.asOutput(scope));
-    opBuilder.addInput(seed.asOutput(scope));
-    opBuilder.addInput(lam.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("dtype", dtype);
-    return new StatelessRandomPoisson<W>(opBuilder.build());
-  }
-  
-  /**
-   * Random values with specified shape.
-   */
-  public Output<W> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<W> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "StatelessRandomPoisson";
-  
-  private Output<W> output;
-  
-  private StatelessRandomPoisson(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessRandomUniform.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessRandomUniform.java
deleted file mode 100644
index 797f660371f..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessRandomUniform.java
+++ /dev/null
@@ -1,98 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.random;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Outputs deterministic pseudorandom random values from a uniform distribution.
- * <p>
- * The generated values follow a uniform distribution in the range `[0, 1)`. The
- * lower bound 0 is included in the range, while the upper bound 1 is excluded.
- * <p>
- * The outputs are a deterministic function of `shape` and `seed`.
- * 
- * @param <V> data type for {@code output()} output
- */
-@Operator(group = "random")
-public final class StatelessRandomUniform<V extends TNumber> extends RawOp implements Operand<V> {
-  
-  /**
-   * Factory method to create a class wrapping a new StatelessRandomUniform operation.
-   * 
-   * @param scope current scope
-   * @param shape The shape of the output tensor.
-   * @param seed 2 seeds (shape [2]).
-   * @param dtype The type of the output.
-   * @return a new instance of StatelessRandomUniform
-   */
-  @Endpoint(describeByClass = true)
-  public static <V extends TNumber, T extends TNumber, U extends TNumber> StatelessRandomUniform<V> create(Scope scope, Operand<T> shape, Operand<U> seed, Class<V> dtype) {
-    OperationBuilder opBuilder = scope.env().opBuilder("StatelessRandomUniform", scope.makeOpName("StatelessRandomUniform"));
-    opBuilder.addInput(shape.asOutput(scope));
-    opBuilder.addInput(seed.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("dtype", dtype);
-    return new StatelessRandomUniform<V>(opBuilder.build());
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new StatelessRandomUniform operation using default output types.
-   * 
-   * @param scope current scope
-   * @param shape The shape of the output tensor.
-   * @param seed 2 seeds (shape [2]).
-   * @return a new instance of StatelessRandomUniform
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber, U extends TNumber> StatelessRandomUniform<TFloat32> create(Scope scope, Operand<T> shape, Operand<U> seed) {
-    return create(scope, shape, seed, TFloat32.class);
-  }
-  
-  /**
-   * Random values with specified shape.
-   */
-  public Output<V> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<V> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "StatelessRandomUniform";
-  
-  private Output<V> output;
-  
-  private StatelessRandomUniform(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessRandomUniformFullInt.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessRandomUniformFullInt.java
deleted file mode 100644
index 5132de53d4d..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessRandomUniformFullInt.java
+++ /dev/null
@@ -1,82 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.random;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Outputs deterministic pseudorandom random integers from a uniform distribution.
- * <p>
- * The generated values are uniform integers covering the whole range of `dtype`.
- * <p>
- * The outputs are a deterministic function of `shape` and `seed`.
- * 
- * @param <V> data type for {@code output()} output
- */
-public final class StatelessRandomUniformFullInt<V extends TNumber> extends RawOp implements Operand<V> {
-  
-  /**
-   * Factory method to create a class wrapping a new StatelessRandomUniformFullInt operation.
-   * 
-   * @param scope current scope
-   * @param shape The shape of the output tensor.
-   * @param seed 2 seeds (shape [2]).
-   * @param dtype The type of the output.
-   * @return a new instance of StatelessRandomUniformFullInt
-   */
-  @Endpoint(describeByClass = true)
-  public static <V extends TNumber, T extends TNumber, U extends TNumber> StatelessRandomUniformFullInt<V> create(Scope scope, Operand<T> shape, Operand<U> seed, Class<V> dtype) {
-    OperationBuilder opBuilder = scope.env().opBuilder("StatelessRandomUniformFullInt", scope.makeOpName("StatelessRandomUniformFullInt"));
-    opBuilder.addInput(shape.asOutput(scope));
-    opBuilder.addInput(seed.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("dtype", dtype);
-    return new StatelessRandomUniformFullInt<V>(opBuilder.build());
-  }
-  
-  /**
-   * Random values with specified shape.
-   */
-  public Output<V> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<V> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "StatelessRandomUniformFullInt";
-  
-  private Output<V> output;
-  
-  private StatelessRandomUniformFullInt(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessRandomUniformInt.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessRandomUniformInt.java
deleted file mode 100644
index 29b5d589a76..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessRandomUniformInt.java
+++ /dev/null
@@ -1,84 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.random;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Outputs deterministic pseudorandom random integers from a uniform distribution.
- * <p>
- * The generated values follow a uniform distribution in the range `[minval, maxval)`.
- * <p>
- * The outputs are a deterministic function of `shape`, `seed`, `minval`, and `maxval`.
- * 
- * @param <V> data type for {@code output()} output
- */
-public final class StatelessRandomUniformInt<V extends TNumber> extends RawOp implements Operand<V> {
-  
-  /**
-   * Factory method to create a class wrapping a new StatelessRandomUniformInt operation.
-   * 
-   * @param scope current scope
-   * @param shape The shape of the output tensor.
-   * @param seed 2 seeds (shape [2]).
-   * @param minval Minimum value (inclusive, scalar).
-   * @param maxval Maximum value (exclusive, scalar).
-   * @return a new instance of StatelessRandomUniformInt
-   */
-  @Endpoint(describeByClass = true)
-  public static <V extends TNumber, T extends TNumber, U extends TNumber> StatelessRandomUniformInt<V> create(Scope scope, Operand<T> shape, Operand<U> seed, Operand<V> minval, Operand<V> maxval) {
-    OperationBuilder opBuilder = scope.env().opBuilder("StatelessRandomUniformInt", scope.makeOpName("StatelessRandomUniformInt"));
-    opBuilder.addInput(shape.asOutput(scope));
-    opBuilder.addInput(seed.asOutput(scope));
-    opBuilder.addInput(minval.asOutput(scope));
-    opBuilder.addInput(maxval.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new StatelessRandomUniformInt<V>(opBuilder.build());
-  }
-  
-  /**
-   * Random values with specified shape.
-   */
-  public Output<V> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<V> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "StatelessRandomUniformInt";
-  
-  private Output<V> output;
-  
-  private StatelessRandomUniformInt(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessTruncatedNormal.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessTruncatedNormal.java
deleted file mode 100644
index 81dfc469250..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessTruncatedNormal.java
+++ /dev/null
@@ -1,99 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.random;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Outputs deterministic pseudorandom values from a truncated normal distribution.
- * <p>
- * The generated values follow a normal distribution with mean 0 and standard
- * deviation 1, except that values whose magnitude is more than 2 standard
- * deviations from the mean are dropped and re-picked.
- * <p>
- * The outputs are a deterministic function of `shape` and `seed`.
- * 
- * @param <V> data type for {@code output()} output
- */
-@Operator(group = "random")
-public final class StatelessTruncatedNormal<V extends TNumber> extends RawOp implements Operand<V> {
-  
-  /**
-   * Factory method to create a class wrapping a new StatelessTruncatedNormal operation.
-   * 
-   * @param scope current scope
-   * @param shape The shape of the output tensor.
-   * @param seed 2 seeds (shape [2]).
-   * @param dtype The type of the output.
-   * @return a new instance of StatelessTruncatedNormal
-   */
-  @Endpoint(describeByClass = true)
-  public static <V extends TNumber, T extends TNumber, U extends TNumber> StatelessTruncatedNormal<V> create(Scope scope, Operand<T> shape, Operand<U> seed, Class<V> dtype) {
-    OperationBuilder opBuilder = scope.env().opBuilder("StatelessTruncatedNormal", scope.makeOpName("StatelessTruncatedNormal"));
-    opBuilder.addInput(shape.asOutput(scope));
-    opBuilder.addInput(seed.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("dtype", dtype);
-    return new StatelessTruncatedNormal<V>(opBuilder.build());
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new StatelessTruncatedNormal operation using default output types.
-   * 
-   * @param scope current scope
-   * @param shape The shape of the output tensor.
-   * @param seed 2 seeds (shape [2]).
-   * @return a new instance of StatelessTruncatedNormal
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber, U extends TNumber> StatelessTruncatedNormal<TFloat32> create(Scope scope, Operand<T> shape, Operand<U> seed) {
-    return create(scope, shape, seed, TFloat32.class);
-  }
-  
-  /**
-   * Random values with specified shape.
-   */
-  public Output<V> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<V> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "StatelessTruncatedNormal";
-  
-  private Output<V> output;
-  
-  private StatelessTruncatedNormal(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/TruncatedNormal.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/TruncatedNormal.java
deleted file mode 100644
index 32be4a817a4..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/TruncatedNormal.java
+++ /dev/null
@@ -1,139 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.random;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Outputs random values from a truncated normal distribution.
- * <p>
- * The generated values follow a normal distribution with mean 0 and standard
- * deviation 1, except that values whose magnitude is more than 2 standard
- * deviations from the mean are dropped and re-picked.
- * 
- * @param <U> data type for {@code output()} output
- */
-@Operator(group = "random")
-public final class TruncatedNormal<U extends TNumber> extends RawOp implements Operand<U> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.random.TruncatedNormal}
-   */
-  public static class Options {
-    
-    /**
-     * @param seed If either `seed` or `seed2` are set to be non-zero, the random number
-     * generator is seeded by the given seed.  Otherwise, it is seeded by a
-     * random seed.
-     */
-    public Options seed(Long seed) {
-      this.seed = seed;
-      return this;
-    }
-    
-    /**
-     * @param seed2 A second seed to avoid seed collision.
-     */
-    public Options seed2(Long seed2) {
-      this.seed2 = seed2;
-      return this;
-    }
-    
-    private Long seed;
-    private Long seed2;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new TruncatedNormal operation.
-   * 
-   * @param scope current scope
-   * @param shape The shape of the output tensor.
-   * @param dtype The type of the output.
-   * @param options carries optional attributes values
-   * @return a new instance of TruncatedNormal
-   */
-  @Endpoint(describeByClass = true)
-  public static <U extends TNumber, T extends TNumber> TruncatedNormal<U> create(Scope scope, Operand<T> shape, Class<U> dtype, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("TruncatedNormal", scope.makeOpName("TruncatedNormal"));
-    opBuilder.addInput(shape.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("dtype", dtype);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.seed != null) {
-          opBuilder.setAttr("seed", opts.seed);
-        }
-        if (opts.seed2 != null) {
-          opBuilder.setAttr("seed2", opts.seed2);
-        }
-      }
-    }
-    return new TruncatedNormal<U>(opBuilder.build());
-  }
-  
-  /**
-   * @param seed If either `seed` or `seed2` are set to be non-zero, the random number
-   * generator is seeded by the given seed.  Otherwise, it is seeded by a
-   * random seed.
-   */
-  public static Options seed(Long seed) {
-    return new Options().seed(seed);
-  }
-  
-  /**
-   * @param seed2 A second seed to avoid seed collision.
-   */
-  public static Options seed2(Long seed2) {
-    return new Options().seed2(seed2);
-  }
-  
-  /**
-   * A tensor of the specified shape filled with random truncated normal
-   * values.
-   */
-  public Output<U> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<U> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "TruncatedNormal";
-  
-  private Output<U> output;
-  
-  private TruncatedNormal(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/UniformCandidateSampler.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/UniformCandidateSampler.java
deleted file mode 100644
index 9a0b724f8ff..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/UniformCandidateSampler.java
+++ /dev/null
@@ -1,171 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.random;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.TInt64;
-
-/**
- * Generates labels for candidate sampling with a uniform distribution.
- * <p>
- * See explanations of candidate sampling and the data formats at
- * go/candidate-sampling.
- * <p>
- * For each batch, this op picks a single set of sampled candidate labels.
- * <p>
- * The advantages of sampling candidates per-batch are simplicity and the
- * possibility of efficient dense matrix multiplication. The disadvantage is that
- * the sampled candidates must be chosen independently of the context and of the
- * true labels.
- */
-@Operator(group = "random")
-public final class UniformCandidateSampler extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.random.UniformCandidateSampler}
-   */
-  public static class Options {
-    
-    /**
-     * @param seed If either seed or seed2 are set to be non-zero, the random number
-     * generator is seeded by the given seed.  Otherwise, it is seeded by a
-     * random seed.
-     */
-    public Options seed(Long seed) {
-      this.seed = seed;
-      return this;
-    }
-    
-    /**
-     * @param seed2 An second seed to avoid seed collision.
-     */
-    public Options seed2(Long seed2) {
-      this.seed2 = seed2;
-      return this;
-    }
-    
-    private Long seed;
-    private Long seed2;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new UniformCandidateSampler operation.
-   * 
-   * @param scope current scope
-   * @param trueClasses A batch_size * num_true matrix, in which each row contains the
-   * IDs of the num_true target_classes in the corresponding original label.
-   * @param numTrue Number of true labels per context.
-   * @param numSampled Number of candidates to randomly sample.
-   * @param unique If unique is true, we sample with rejection, so that all sampled
-   * candidates in a batch are unique. This requires some approximation to
-   * estimate the post-rejection sampling probabilities.
-   * @param rangeMax The sampler will sample integers from the interval [0, range_max).
-   * @param options carries optional attributes values
-   * @return a new instance of UniformCandidateSampler
-   */
-  @Endpoint(describeByClass = true)
-  public static UniformCandidateSampler create(Scope scope, Operand<TInt64> trueClasses, Long numTrue, Long numSampled, Boolean unique, Long rangeMax, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("UniformCandidateSampler", scope.makeOpName("UniformCandidateSampler"));
-    opBuilder.addInput(trueClasses.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("num_true", numTrue);
-    opBuilder.setAttr("num_sampled", numSampled);
-    opBuilder.setAttr("unique", unique);
-    opBuilder.setAttr("range_max", rangeMax);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.seed != null) {
-          opBuilder.setAttr("seed", opts.seed);
-        }
-        if (opts.seed2 != null) {
-          opBuilder.setAttr("seed2", opts.seed2);
-        }
-      }
-    }
-    return new UniformCandidateSampler(opBuilder.build());
-  }
-  
-  /**
-   * @param seed If either seed or seed2 are set to be non-zero, the random number
-   * generator is seeded by the given seed.  Otherwise, it is seeded by a
-   * random seed.
-   */
-  public static Options seed(Long seed) {
-    return new Options().seed(seed);
-  }
-  
-  /**
-   * @param seed2 An second seed to avoid seed collision.
-   */
-  public static Options seed2(Long seed2) {
-    return new Options().seed2(seed2);
-  }
-  
-  /**
-   * A vector of length num_sampled, in which each element is
-   * the ID of a sampled candidate.
-   */
-  public Output<TInt64> sampledCandidates() {
-    return sampledCandidates;
-  }
-  
-  /**
-   * A batch_size * num_true matrix, representing
-   * the number of times each candidate is expected to occur in a batch
-   * of sampled candidates. If unique=true, then this is a probability.
-   */
-  public Output<TFloat32> trueExpectedCount() {
-    return trueExpectedCount;
-  }
-  
-  /**
-   * A vector of length num_sampled, for each sampled
-   * candidate representing the number of times the candidate is expected
-   * to occur in a batch of sampled candidates.  If unique=true, then this is a
-   * probability.
-   */
-  public Output<TFloat32> sampledExpectedCount() {
-    return sampledExpectedCount;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "UniformCandidateSampler";
-  
-  private Output<TInt64> sampledCandidates;
-  private Output<TFloat32> trueExpectedCount;
-  private Output<TFloat32> sampledExpectedCount;
-  
-  private UniformCandidateSampler(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    sampledCandidates = operation.output(outputIdx++);
-    trueExpectedCount = operation.output(outputIdx++);
-    sampledExpectedCount = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/experimental/DummySeedGenerator.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/experimental/DummySeedGenerator.java
deleted file mode 100644
index dff26846e0f..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/experimental/DummySeedGenerator.java
+++ /dev/null
@@ -1,69 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.random.experimental;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- */
-public final class DummySeedGenerator extends RawOp implements Operand<TType> {
-  
-  /**
-   * Factory method to create a class wrapping a new DummySeedGenerator operation.
-   * 
-   * @param scope current scope
-   * @return a new instance of DummySeedGenerator
-   */
-  @Endpoint(describeByClass = true)
-  public static DummySeedGenerator create(Scope scope) {
-    OperationBuilder opBuilder = scope.env().opBuilder("DummySeedGenerator", scope.makeOpName("DummySeedGenerator"));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new DummySeedGenerator(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<?> handle() {
-    return handle;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) handle;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "DummySeedGenerator";
-  
-  private Output<?> handle;
-  
-  private DummySeedGenerator(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    handle = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/BatchFft.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/BatchFft.java
deleted file mode 100644
index 2e9d7f0da88..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/BatchFft.java
+++ /dev/null
@@ -1,72 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.signal;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- */
-@Operator(group = "signal")
-public final class BatchFft extends RawOp implements Operand<TType> {
-  
-  /**
-   * Factory method to create a class wrapping a new BatchFft operation.
-   * 
-   * @param scope current scope
-   * @param input 
-   * @return a new instance of BatchFft
-   */
-  @Endpoint(describeByClass = true)
-  public static BatchFft create(Scope scope, Operand<?> input) {
-    OperationBuilder opBuilder = scope.env().opBuilder("BatchFFT", scope.makeOpName("BatchFft"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new BatchFft(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<?> output() {
-    return output;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "BatchFFT";
-  
-  private Output<?> output;
-  
-  private BatchFft(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/BatchFft2d.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/BatchFft2d.java
deleted file mode 100644
index aa8d61a34ef..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/BatchFft2d.java
+++ /dev/null
@@ -1,72 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.signal;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- */
-@Operator(group = "signal")
-public final class BatchFft2d extends RawOp implements Operand<TType> {
-  
-  /**
-   * Factory method to create a class wrapping a new BatchFft2d operation.
-   * 
-   * @param scope current scope
-   * @param input 
-   * @return a new instance of BatchFft2d
-   */
-  @Endpoint(describeByClass = true)
-  public static BatchFft2d create(Scope scope, Operand<?> input) {
-    OperationBuilder opBuilder = scope.env().opBuilder("BatchFFT2D", scope.makeOpName("BatchFft2d"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new BatchFft2d(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<?> output() {
-    return output;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "BatchFFT2D";
-  
-  private Output<?> output;
-  
-  private BatchFft2d(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/BatchFft3d.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/BatchFft3d.java
deleted file mode 100644
index 57d61e6a060..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/BatchFft3d.java
+++ /dev/null
@@ -1,72 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.signal;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- */
-@Operator(group = "signal")
-public final class BatchFft3d extends RawOp implements Operand<TType> {
-  
-  /**
-   * Factory method to create a class wrapping a new BatchFft3d operation.
-   * 
-   * @param scope current scope
-   * @param input 
-   * @return a new instance of BatchFft3d
-   */
-  @Endpoint(describeByClass = true)
-  public static BatchFft3d create(Scope scope, Operand<?> input) {
-    OperationBuilder opBuilder = scope.env().opBuilder("BatchFFT3D", scope.makeOpName("BatchFft3d"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new BatchFft3d(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<?> output() {
-    return output;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "BatchFFT3D";
-  
-  private Output<?> output;
-  
-  private BatchFft3d(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/BatchIfft.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/BatchIfft.java
deleted file mode 100644
index 2033092ed75..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/BatchIfft.java
+++ /dev/null
@@ -1,72 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.signal;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- */
-@Operator(group = "signal")
-public final class BatchIfft extends RawOp implements Operand<TType> {
-  
-  /**
-   * Factory method to create a class wrapping a new BatchIfft operation.
-   * 
-   * @param scope current scope
-   * @param input 
-   * @return a new instance of BatchIfft
-   */
-  @Endpoint(describeByClass = true)
-  public static BatchIfft create(Scope scope, Operand<?> input) {
-    OperationBuilder opBuilder = scope.env().opBuilder("BatchIFFT", scope.makeOpName("BatchIfft"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new BatchIfft(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<?> output() {
-    return output;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "BatchIFFT";
-  
-  private Output<?> output;
-  
-  private BatchIfft(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/BatchIfft2d.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/BatchIfft2d.java
deleted file mode 100644
index fa6dcf4ba52..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/BatchIfft2d.java
+++ /dev/null
@@ -1,72 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.signal;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- */
-@Operator(group = "signal")
-public final class BatchIfft2d extends RawOp implements Operand<TType> {
-  
-  /**
-   * Factory method to create a class wrapping a new BatchIfft2d operation.
-   * 
-   * @param scope current scope
-   * @param input 
-   * @return a new instance of BatchIfft2d
-   */
-  @Endpoint(describeByClass = true)
-  public static BatchIfft2d create(Scope scope, Operand<?> input) {
-    OperationBuilder opBuilder = scope.env().opBuilder("BatchIFFT2D", scope.makeOpName("BatchIfft2d"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new BatchIfft2d(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<?> output() {
-    return output;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "BatchIFFT2D";
-  
-  private Output<?> output;
-  
-  private BatchIfft2d(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/BatchIfft3d.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/BatchIfft3d.java
deleted file mode 100644
index 1ed5c57bb1b..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/BatchIfft3d.java
+++ /dev/null
@@ -1,72 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.signal;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- */
-@Operator(group = "signal")
-public final class BatchIfft3d extends RawOp implements Operand<TType> {
-  
-  /**
-   * Factory method to create a class wrapping a new BatchIfft3d operation.
-   * 
-   * @param scope current scope
-   * @param input 
-   * @return a new instance of BatchIfft3d
-   */
-  @Endpoint(describeByClass = true)
-  public static BatchIfft3d create(Scope scope, Operand<?> input) {
-    OperationBuilder opBuilder = scope.env().opBuilder("BatchIFFT3D", scope.makeOpName("BatchIfft3d"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new BatchIfft3d(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<?> output() {
-    return output;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "BatchIFFT3D";
-  
-  private Output<?> output;
-  
-  private BatchIfft3d(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Fft.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Fft.java
deleted file mode 100644
index 2c9fb8d16b9..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Fft.java
+++ /dev/null
@@ -1,83 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.signal;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Fast Fourier transform.
- * <p>
- * Computes the 1-dimensional discrete Fourier transform over the inner-most
- * dimension of `input`.
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator(group = "signal")
-public final class Fft<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new Fft operation.
-   * 
-   * @param scope current scope
-   * @param input A complex tensor.
-   * @return a new instance of Fft
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> Fft<T> create(Scope scope, Operand<T> input) {
-    OperationBuilder opBuilder = scope.env().opBuilder("FFT", scope.makeOpName("Fft"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new Fft<T>(opBuilder.build());
-  }
-  
-  /**
-   * A complex tensor of the same shape as `input`. The inner-most
-   *   dimension of `input` is replaced with its 1D Fourier transform.
-   * <p>
-   * @compatibility(numpy)
-   * Equivalent to np.fft.fft
-   * @end_compatibility
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "FFT";
-  
-  private Output<T> output;
-  
-  private Fft(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Fft2d.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Fft2d.java
deleted file mode 100644
index 226d5b313d6..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Fft2d.java
+++ /dev/null
@@ -1,83 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.signal;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * 2D fast Fourier transform.
- * <p>
- * Computes the 2-dimensional discrete Fourier transform over the inner-most
- * 2 dimensions of `input`.
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator(group = "signal")
-public final class Fft2d<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new Fft2d operation.
-   * 
-   * @param scope current scope
-   * @param input A complex tensor.
-   * @return a new instance of Fft2d
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> Fft2d<T> create(Scope scope, Operand<T> input) {
-    OperationBuilder opBuilder = scope.env().opBuilder("FFT2D", scope.makeOpName("Fft2d"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new Fft2d<T>(opBuilder.build());
-  }
-  
-  /**
-   * A complex tensor of the same shape as `input`. The inner-most 2
-   *   dimensions of `input` are replaced with their 2D Fourier transform.
-   * <p>
-   * @compatibility(numpy)
-   * Equivalent to np.fft.fft2
-   * @end_compatibility
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "FFT2D";
-  
-  private Output<T> output;
-  
-  private Fft2d(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Fft3d.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Fft3d.java
deleted file mode 100644
index 4dc5e420356..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Fft3d.java
+++ /dev/null
@@ -1,83 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.signal;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * 3D fast Fourier transform.
- * <p>
- * Computes the 3-dimensional discrete Fourier transform over the inner-most 3
- * dimensions of `input`.
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator(group = "signal")
-public final class Fft3d<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new Fft3d operation.
-   * 
-   * @param scope current scope
-   * @param input A complex tensor.
-   * @return a new instance of Fft3d
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> Fft3d<T> create(Scope scope, Operand<T> input) {
-    OperationBuilder opBuilder = scope.env().opBuilder("FFT3D", scope.makeOpName("Fft3d"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new Fft3d<T>(opBuilder.build());
-  }
-  
-  /**
-   * A complex tensor of the same shape as `input`. The inner-most 3
-   *   dimensions of `input` are replaced with their 3D Fourier transform.
-   * <p>
-   * @compatibility(numpy)
-   * Equivalent to np.fft.fftn with 3 dimensions.
-   * @end_compatibility
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "FFT3D";
-  
-  private Output<T> output;
-  
-  private Fft3d(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Ifft.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Ifft.java
deleted file mode 100644
index fb65e5a5745..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Ifft.java
+++ /dev/null
@@ -1,83 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.signal;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Inverse fast Fourier transform.
- * <p>
- * Computes the inverse 1-dimensional discrete Fourier transform over the
- * inner-most dimension of `input`.
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator(group = "signal")
-public final class Ifft<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new Ifft operation.
-   * 
-   * @param scope current scope
-   * @param input A complex tensor.
-   * @return a new instance of Ifft
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> Ifft<T> create(Scope scope, Operand<T> input) {
-    OperationBuilder opBuilder = scope.env().opBuilder("IFFT", scope.makeOpName("Ifft"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new Ifft<T>(opBuilder.build());
-  }
-  
-  /**
-   * A complex tensor of the same shape as `input`. The inner-most
-   *   dimension of `input` is replaced with its inverse 1D Fourier transform.
-   * <p>
-   * @compatibility(numpy)
-   * Equivalent to np.fft.ifft
-   * @end_compatibility
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "IFFT";
-  
-  private Output<T> output;
-  
-  private Ifft(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Ifft2d.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Ifft2d.java
deleted file mode 100644
index 6d4bf5c8504..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Ifft2d.java
+++ /dev/null
@@ -1,83 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.signal;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Inverse 2D fast Fourier transform.
- * <p>
- * Computes the inverse 2-dimensional discrete Fourier transform over the
- * inner-most 2 dimensions of `input`.
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator(group = "signal")
-public final class Ifft2d<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new Ifft2d operation.
-   * 
-   * @param scope current scope
-   * @param input A complex tensor.
-   * @return a new instance of Ifft2d
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> Ifft2d<T> create(Scope scope, Operand<T> input) {
-    OperationBuilder opBuilder = scope.env().opBuilder("IFFT2D", scope.makeOpName("Ifft2d"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new Ifft2d<T>(opBuilder.build());
-  }
-  
-  /**
-   * A complex tensor of the same shape as `input`. The inner-most 2
-   *   dimensions of `input` are replaced with their inverse 2D Fourier transform.
-   * <p>
-   * @compatibility(numpy)
-   * Equivalent to np.fft.ifft2
-   * @end_compatibility
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "IFFT2D";
-  
-  private Output<T> output;
-  
-  private Ifft2d(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Ifft3d.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Ifft3d.java
deleted file mode 100644
index 1ee196c2a73..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Ifft3d.java
+++ /dev/null
@@ -1,83 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.signal;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Inverse 3D fast Fourier transform.
- * <p>
- * Computes the inverse 3-dimensional discrete Fourier transform over the
- * inner-most 3 dimensions of `input`.
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator(group = "signal")
-public final class Ifft3d<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new Ifft3d operation.
-   * 
-   * @param scope current scope
-   * @param input A complex tensor.
-   * @return a new instance of Ifft3d
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> Ifft3d<T> create(Scope scope, Operand<T> input) {
-    OperationBuilder opBuilder = scope.env().opBuilder("IFFT3D", scope.makeOpName("Ifft3d"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new Ifft3d<T>(opBuilder.build());
-  }
-  
-  /**
-   * A complex tensor of the same shape as `input`. The inner-most 3
-   *   dimensions of `input` are replaced with their inverse 3D Fourier transform.
-   * <p>
-   * @compatibility(numpy)
-   * Equivalent to np.fft.ifftn with 3 dimensions.
-   * @end_compatibility
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "IFFT3D";
-  
-  private Output<T> output;
-  
-  private Ifft3d(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Irfft.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Irfft.java
deleted file mode 100644
index 712e3b714e2..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Irfft.java
+++ /dev/null
@@ -1,115 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.signal;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * Inverse real-valued fast Fourier transform.
- * <p>
- * Computes the inverse 1-dimensional discrete Fourier transform of a real-valued
- * signal over the inner-most dimension of `input`.
- * <p>
- * The inner-most dimension of `input` is assumed to be the result of `RFFT`: the
- * `fft_length / 2 + 1` unique components of the DFT of a real-valued signal. If
- * `fft_length` is not provided, it is computed from the size of the inner-most
- * dimension of `input` (`fft_length = 2 * (inner - 1)`). If the FFT length used to
- * compute `input` is odd, it should be provided since it cannot be inferred
- * properly.
- * <p>
- * Along the axis `signal.Irfft` is computed on, if `fft_length / 2 + 1` is smaller
- * than the corresponding dimension of `input`, the dimension is cropped. If it is
- * larger, the dimension is padded with zeros.
- * 
- * @param <U> data type for {@code output()} output
- */
-@Operator(group = "signal")
-public final class Irfft<U extends TNumber> extends RawOp implements Operand<U> {
-  
-  /**
-   * Factory method to create a class wrapping a new Irfft operation.
-   * 
-   * @param scope current scope
-   * @param input A complex tensor.
-   * @param fftLength An int32 tensor of shape [1]. The FFT length.
-   * @param Treal 
-   * @return a new instance of Irfft
-   */
-  @Endpoint(describeByClass = true)
-  public static <U extends TNumber, T extends TType> Irfft<U> create(Scope scope, Operand<T> input, Operand<TInt32> fftLength, Class<U> Treal) {
-    OperationBuilder opBuilder = scope.env().opBuilder("IRFFT", scope.makeOpName("Irfft"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder.addInput(fftLength.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("Treal", Treal);
-    return new Irfft<U>(opBuilder.build());
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new Irfft operation using default output types.
-   * 
-   * @param scope current scope
-   * @param input A complex tensor.
-   * @param fftLength An int32 tensor of shape [1]. The FFT length.
-   * @return a new instance of Irfft
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> Irfft<TFloat32> create(Scope scope, Operand<T> input, Operand<TInt32> fftLength) {
-    return create(scope, input, fftLength, TFloat32.class);
-  }
-  
-  /**
-   * A float32 tensor of the same rank as `input`. The inner-most
-   *   dimension of `input` is replaced with the `fft_length` samples of its inverse
-   *   1D Fourier transform.
-   * <p>
-   * @compatibility(numpy)
-   * Equivalent to np.fft.irfft
-   * @end_compatibility
-   */
-  public Output<U> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<U> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "IRFFT";
-  
-  private Output<U> output;
-  
-  private Irfft(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Irfft2d.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Irfft2d.java
deleted file mode 100644
index bc3a340d210..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Irfft2d.java
+++ /dev/null
@@ -1,116 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.signal;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * Inverse 2D real-valued fast Fourier transform.
- * <p>
- * Computes the inverse 2-dimensional discrete Fourier transform of a real-valued
- * signal over the inner-most 2 dimensions of `input`.
- * <p>
- * The inner-most 2 dimensions of `input` are assumed to be the result of `RFFT2D`:
- * The inner-most dimension contains the `fft_length / 2 + 1` unique components of
- * the DFT of a real-valued signal. If `fft_length` is not provided, it is computed
- * from the size of the inner-most 2 dimensions of `input`. If the FFT length used
- * to compute `input` is odd, it should be provided since it cannot be inferred
- * properly.
- * <p>
- * Along each axis `signal.Irfft2d` is computed on, if `fft_length` (or
- * `fft_length / 2 + 1` for the inner-most dimension) is smaller than the
- * corresponding dimension of `input`, the dimension is cropped. If it is larger,
- * the dimension is padded with zeros.
- * 
- * @param <U> data type for {@code output()} output
- */
-@Operator(group = "signal")
-public final class Irfft2d<U extends TNumber> extends RawOp implements Operand<U> {
-  
-  /**
-   * Factory method to create a class wrapping a new Irfft2d operation.
-   * 
-   * @param scope current scope
-   * @param input A complex tensor.
-   * @param fftLength An int32 tensor of shape [2]. The FFT length for each dimension.
-   * @param Treal 
-   * @return a new instance of Irfft2d
-   */
-  @Endpoint(describeByClass = true)
-  public static <U extends TNumber, T extends TType> Irfft2d<U> create(Scope scope, Operand<T> input, Operand<TInt32> fftLength, Class<U> Treal) {
-    OperationBuilder opBuilder = scope.env().opBuilder("IRFFT2D", scope.makeOpName("Irfft2d"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder.addInput(fftLength.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("Treal", Treal);
-    return new Irfft2d<U>(opBuilder.build());
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new Irfft2d operation using default output types.
-   * 
-   * @param scope current scope
-   * @param input A complex tensor.
-   * @param fftLength An int32 tensor of shape [2]. The FFT length for each dimension.
-   * @return a new instance of Irfft2d
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> Irfft2d<TFloat32> create(Scope scope, Operand<T> input, Operand<TInt32> fftLength) {
-    return create(scope, input, fftLength, TFloat32.class);
-  }
-  
-  /**
-   * A float32 tensor of the same rank as `input`. The inner-most 2
-   *   dimensions of `input` are replaced with the `fft_length` samples of their
-   *   inverse 2D Fourier transform.
-   * <p>
-   * @compatibility(numpy)
-   * Equivalent to np.fft.irfft2
-   * @end_compatibility
-   */
-  public Output<U> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<U> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "IRFFT2D";
-  
-  private Output<U> output;
-  
-  private Irfft2d(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Irfft3d.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Irfft3d.java
deleted file mode 100644
index d4306fe1c2a..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Irfft3d.java
+++ /dev/null
@@ -1,116 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.signal;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * Inverse 3D real-valued fast Fourier transform.
- * <p>
- * Computes the inverse 3-dimensional discrete Fourier transform of a real-valued
- * signal over the inner-most 3 dimensions of `input`.
- * <p>
- * The inner-most 3 dimensions of `input` are assumed to be the result of `RFFT3D`:
- * The inner-most dimension contains the `fft_length / 2 + 1` unique components of
- * the DFT of a real-valued signal. If `fft_length` is not provided, it is computed
- * from the size of the inner-most 3 dimensions of `input`. If the FFT length used
- * to compute `input` is odd, it should be provided since it cannot be inferred
- * properly.
- * <p>
- * Along each axis `signal.Irfft3d` is computed on, if `fft_length` (or
- * `fft_length / 2 + 1` for the inner-most dimension) is smaller than the
- * corresponding dimension of `input`, the dimension is cropped. If it is larger,
- * the dimension is padded with zeros.
- * 
- * @param <U> data type for {@code output()} output
- */
-@Operator(group = "signal")
-public final class Irfft3d<U extends TNumber> extends RawOp implements Operand<U> {
-  
-  /**
-   * Factory method to create a class wrapping a new Irfft3d operation.
-   * 
-   * @param scope current scope
-   * @param input A complex tensor.
-   * @param fftLength An int32 tensor of shape [3]. The FFT length for each dimension.
-   * @param Treal 
-   * @return a new instance of Irfft3d
-   */
-  @Endpoint(describeByClass = true)
-  public static <U extends TNumber, T extends TType> Irfft3d<U> create(Scope scope, Operand<T> input, Operand<TInt32> fftLength, Class<U> Treal) {
-    OperationBuilder opBuilder = scope.env().opBuilder("IRFFT3D", scope.makeOpName("Irfft3d"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder.addInput(fftLength.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("Treal", Treal);
-    return new Irfft3d<U>(opBuilder.build());
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new Irfft3d operation using default output types.
-   * 
-   * @param scope current scope
-   * @param input A complex tensor.
-   * @param fftLength An int32 tensor of shape [3]. The FFT length for each dimension.
-   * @return a new instance of Irfft3d
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> Irfft3d<TFloat32> create(Scope scope, Operand<T> input, Operand<TInt32> fftLength) {
-    return create(scope, input, fftLength, TFloat32.class);
-  }
-  
-  /**
-   * A float32 tensor of the same rank as `input`. The inner-most 3
-   *   dimensions of `input` are replaced with the `fft_length` samples of their
-   *   inverse 3D real Fourier transform.
-   * <p>
-   * @compatibility(numpy)
-   * Equivalent to np.irfftn with 3 dimensions.
-   * @end_compatibility
-   */
-  public Output<U> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<U> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "IRFFT3D";
-  
-  private Output<U> output;
-  
-  private Irfft3d(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Rfft.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Rfft.java
deleted file mode 100644
index c4a47a60b8d..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Rfft.java
+++ /dev/null
@@ -1,98 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.signal;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * Real-valued fast Fourier transform.
- * <p>
- * Computes the 1-dimensional discrete Fourier transform of a real-valued signal
- * over the inner-most dimension of `input`.
- * <p>
- * Since the DFT of a real signal is Hermitian-symmetric, `signal.Rfft` only returns the
- * `fft_length / 2 + 1` unique components of the FFT: the zero-frequency term,
- * followed by the `fft_length / 2` positive-frequency terms.
- * <p>
- * Along the axis `signal.Rfft` is computed on, if `fft_length` is smaller than the
- * corresponding dimension of `input`, the dimension is cropped. If it is larger,
- * the dimension is padded with zeros.
- * 
- * @param <U> data type for {@code output()} output
- */
-@Operator(group = "signal")
-public final class Rfft<U extends TType> extends RawOp implements Operand<U> {
-  
-  /**
-   * Factory method to create a class wrapping a new Rfft operation.
-   * 
-   * @param scope current scope
-   * @param input A float32 tensor.
-   * @param fftLength An int32 tensor of shape [1]. The FFT length.
-   * @param Tcomplex 
-   * @return a new instance of Rfft
-   */
-  @Endpoint(describeByClass = true)
-  public static <U extends TType, T extends TNumber> Rfft<U> create(Scope scope, Operand<T> input, Operand<TInt32> fftLength, Class<U> Tcomplex) {
-    OperationBuilder opBuilder = scope.env().opBuilder("RFFT", scope.makeOpName("Rfft"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder.addInput(fftLength.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("Tcomplex", Tcomplex);
-    return new Rfft<U>(opBuilder.build());
-  }
-  
-  /**
-   * A complex64 tensor of the same rank as `input`. The inner-most
-   *   dimension of `input` is replaced with the `fft_length / 2 + 1` unique
-   *   frequency components of its 1D Fourier transform.
-   * <p>
-   * @compatibility(numpy)
-   * Equivalent to np.fft.rfft
-   * @end_compatibility
-   */
-  public Output<U> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<U> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "RFFT";
-  
-  private Output<U> output;
-  
-  private Rfft(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Rfft2d.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Rfft2d.java
deleted file mode 100644
index 173b09c53fd..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Rfft2d.java
+++ /dev/null
@@ -1,100 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.signal;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * 2D real-valued fast Fourier transform.
- * <p>
- * Computes the 2-dimensional discrete Fourier transform of a real-valued signal
- * over the inner-most 2 dimensions of `input`.
- * <p>
- * Since the DFT of a real signal is Hermitian-symmetric, `signal.Rfft2d` only returns the
- * `fft_length / 2 + 1` unique components of the FFT for the inner-most dimension
- * of `output`: the zero-frequency term, followed by the `fft_length / 2`
- * positive-frequency terms.
- * <p>
- * Along each axis `signal.Rfft2d` is computed on, if `fft_length` is smaller than the
- * corresponding dimension of `input`, the dimension is cropped. If it is larger,
- * the dimension is padded with zeros.
- * 
- * @param <U> data type for {@code output()} output
- */
-@Operator(group = "signal")
-public final class Rfft2d<U extends TType> extends RawOp implements Operand<U> {
-  
-  /**
-   * Factory method to create a class wrapping a new Rfft2d operation.
-   * 
-   * @param scope current scope
-   * @param input A float32 tensor.
-   * @param fftLength An int32 tensor of shape [2]. The FFT length for each dimension.
-   * @param Tcomplex 
-   * @return a new instance of Rfft2d
-   */
-  @Endpoint(describeByClass = true)
-  public static <U extends TType, T extends TNumber> Rfft2d<U> create(Scope scope, Operand<T> input, Operand<TInt32> fftLength, Class<U> Tcomplex) {
-    OperationBuilder opBuilder = scope.env().opBuilder("RFFT2D", scope.makeOpName("Rfft2d"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder.addInput(fftLength.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("Tcomplex", Tcomplex);
-    return new Rfft2d<U>(opBuilder.build());
-  }
-  
-  /**
-   * A complex64 tensor of the same rank as `input`. The inner-most 2
-   *   dimensions of `input` are replaced with their 2D Fourier transform. The
-   *   inner-most dimension contains `fft_length / 2 + 1` unique frequency
-   *   components.
-   * <p>
-   * @compatibility(numpy)
-   * Equivalent to np.fft.rfft2
-   * @end_compatibility
-   */
-  public Output<U> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<U> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "RFFT2D";
-  
-  private Output<U> output;
-  
-  private Rfft2d(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Rfft3d.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Rfft3d.java
deleted file mode 100644
index 48620690a7e..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Rfft3d.java
+++ /dev/null
@@ -1,100 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.signal;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * 3D real-valued fast Fourier transform.
- * <p>
- * Computes the 3-dimensional discrete Fourier transform of a real-valued signal
- * over the inner-most 3 dimensions of `input`.
- * <p>
- * Since the DFT of a real signal is Hermitian-symmetric, `signal.Rfft3d` only returns the
- * `fft_length / 2 + 1` unique components of the FFT for the inner-most dimension
- * of `output`: the zero-frequency term, followed by the `fft_length / 2`
- * positive-frequency terms.
- * <p>
- * Along each axis `signal.Rfft3d` is computed on, if `fft_length` is smaller than the
- * corresponding dimension of `input`, the dimension is cropped. If it is larger,
- * the dimension is padded with zeros.
- * 
- * @param <U> data type for {@code output()} output
- */
-@Operator(group = "signal")
-public final class Rfft3d<U extends TType> extends RawOp implements Operand<U> {
-  
-  /**
-   * Factory method to create a class wrapping a new Rfft3d operation.
-   * 
-   * @param scope current scope
-   * @param input A float32 tensor.
-   * @param fftLength An int32 tensor of shape [3]. The FFT length for each dimension.
-   * @param Tcomplex 
-   * @return a new instance of Rfft3d
-   */
-  @Endpoint(describeByClass = true)
-  public static <U extends TType, T extends TNumber> Rfft3d<U> create(Scope scope, Operand<T> input, Operand<TInt32> fftLength, Class<U> Tcomplex) {
-    OperationBuilder opBuilder = scope.env().opBuilder("RFFT3D", scope.makeOpName("Rfft3d"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder.addInput(fftLength.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("Tcomplex", Tcomplex);
-    return new Rfft3d<U>(opBuilder.build());
-  }
-  
-  /**
-   * A complex64 tensor of the same rank as `input`. The inner-most 3
-   *   dimensions of `input` are replaced with the their 3D Fourier transform. The
-   *   inner-most dimension contains `fft_length / 2 + 1` unique frequency
-   *   components.
-   * <p>
-   * @compatibility(numpy)
-   * Equivalent to np.fft.rfftn with 3 dimensions.
-   * @end_compatibility
-   */
-  public Output<U> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<U> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "RFFT3D";
-  
-  private Output<U> output;
-  
-  private Rfft3d(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/AddManySparseToTensorsMap.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/AddManySparseToTensorsMap.java
deleted file mode 100644
index 1a5adb2e8c0..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/AddManySparseToTensorsMap.java
+++ /dev/null
@@ -1,158 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.sparse;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
-
-/**
- * Add an `N`-minibatch `SparseTensor` to a `SparseTensorsMap`, return `N` handles.
- * <p>
- * A `SparseTensor` of rank `R` is represented by three tensors: `sparse_indices`,
- * `sparse_values`, and `sparse_shape`, where
- * <pre>{@code
- * sparse_indices.shape[1] == sparse_shape.shape[0] == R}</pre>
- * An `N`-minibatch of `SparseTensor` objects is represented as a `SparseTensor`
- * having a first `sparse_indices` column taking values between `[0, N)`, where
- * the minibatch size `N == sparse_shape[0]`.
- * <p>
- * The input `SparseTensor` must have rank `R` greater than 1, and the first
- * dimension is treated as the minibatch dimension.  Elements of the `SparseTensor`
- * must be sorted in increasing order of this first dimension.  The stored
- * `SparseTensor` objects pointed to by each row of the output `sparse_handles`
- * will have rank `R-1`.
- * <p>
- * The `SparseTensor` values can then be read out as part of a minibatch by passing
- * the given keys as vector elements to `TakeManySparseFromTensorsMap`.  To ensure
- * the correct `SparseTensorsMap` is accessed, ensure that the same
- * `container` and `shared_name` are passed to that Op.  If no `shared_name`
- * is provided here, instead use the <i>name</i> of the Operation created by calling
- * `sparse.AddManySparseToTensorsMap` as the `shared_name` passed to
- * `TakeManySparseFromTensorsMap`.  Ensure the Operations are colocated.
- */
-@Operator(group = "sparse")
-public final class AddManySparseToTensorsMap extends RawOp implements Operand<TInt64> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.sparse.AddManySparseToTensorsMap}
-   */
-  public static class Options {
-    
-    /**
-     * @param container The container name for the `SparseTensorsMap` created by this op.
-     */
-    public Options container(String container) {
-      this.container = container;
-      return this;
-    }
-    
-    /**
-     * @param sharedName The shared name for the `SparseTensorsMap` created by this op.
-     * If blank, the new Operation's unique name is used.
-     */
-    public Options sharedName(String sharedName) {
-      this.sharedName = sharedName;
-      return this;
-    }
-    
-    private String container;
-    private String sharedName;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new AddManySparseToTensorsMap operation.
-   * 
-   * @param scope current scope
-   * @param sparseIndices 2-D.  The `indices` of the minibatch `SparseTensor`.
-   * `sparse_indices[:, 0]` must be ordered values in `[0, N)`.
-   * @param sparseValues 1-D.  The `values` of the minibatch `SparseTensor`.
-   * @param sparseShape 1-D.  The `shape` of the minibatch `SparseTensor`.
-   * The minibatch size `N == sparse_shape[0]`.
-   * @param options carries optional attributes values
-   * @return a new instance of AddManySparseToTensorsMap
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> AddManySparseToTensorsMap create(Scope scope, Operand<TInt64> sparseIndices, Operand<T> sparseValues, Operand<TInt64> sparseShape, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("AddManySparseToTensorsMap", scope.makeOpName("AddManySparseToTensorsMap"));
-    opBuilder.addInput(sparseIndices.asOutput(scope));
-    opBuilder.addInput(sparseValues.asOutput(scope));
-    opBuilder.addInput(sparseShape.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.container != null) {
-          opBuilder.setAttr("container", opts.container);
-        }
-        if (opts.sharedName != null) {
-          opBuilder.setAttr("shared_name", opts.sharedName);
-        }
-      }
-    }
-    return new AddManySparseToTensorsMap(opBuilder.build());
-  }
-  
-  /**
-   * @param container The container name for the `SparseTensorsMap` created by this op.
-   */
-  public static Options container(String container) {
-    return new Options().container(container);
-  }
-  
-  /**
-   * @param sharedName The shared name for the `SparseTensorsMap` created by this op.
-   * If blank, the new Operation's unique name is used.
-   */
-  public static Options sharedName(String sharedName) {
-    return new Options().sharedName(sharedName);
-  }
-  
-  /**
-   * 1-D.  The handles of the `SparseTensor` now stored in the
-   * `SparseTensorsMap`.  Shape: `[N]`.
-   */
-  public Output<TInt64> sparseHandles() {
-    return sparseHandles;
-  }
-  
-  @Override
-  public Output<TInt64> asOutput(Scope scope) {
-    return sparseHandles;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "AddManySparseToTensorsMap";
-  
-  private Output<TInt64> sparseHandles;
-  
-  private AddManySparseToTensorsMap(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    sparseHandles = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/AddSparseToTensorsMap.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/AddSparseToTensorsMap.java
deleted file mode 100644
index 936f21954db..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/AddSparseToTensorsMap.java
+++ /dev/null
@@ -1,149 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.sparse;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
-
-/**
- * Add a `SparseTensor` to a `SparseTensorsMap` return its handle.
- * <p>
- * A `SparseTensor` is represented by three tensors: `sparse_indices`,
- * `sparse_values`, and `sparse_shape`.
- * <p>
- * This operator takes the given `SparseTensor` and adds it to a container
- * object (a `SparseTensorsMap`).  A unique key within this container is generated
- * in the form of an `int64`, and this is the value that is returned.
- * <p>
- * The `SparseTensor` can then be read out as part of a minibatch by passing
- * the key as a vector element to `TakeManySparseFromTensorsMap`.  To ensure
- * the correct `SparseTensorsMap` is accessed, ensure that the same
- * `container` and `shared_name` are passed to that Op.  If no `shared_name`
- * is provided here, instead use the <i>name</i> of the Operation created by calling
- * `sparse.AddSparseToTensorsMap` as the `shared_name` passed to
- * `TakeManySparseFromTensorsMap`.  Ensure the Operations are colocated.
- */
-@Operator(group = "sparse")
-public final class AddSparseToTensorsMap extends RawOp implements Operand<TInt64> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.sparse.AddSparseToTensorsMap}
-   */
-  public static class Options {
-    
-    /**
-     * @param container The container name for the `SparseTensorsMap` created by this op.
-     */
-    public Options container(String container) {
-      this.container = container;
-      return this;
-    }
-    
-    /**
-     * @param sharedName The shared name for the `SparseTensorsMap` created by this op.
-     * If blank, the new Operation's unique name is used.
-     */
-    public Options sharedName(String sharedName) {
-      this.sharedName = sharedName;
-      return this;
-    }
-    
-    private String container;
-    private String sharedName;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new AddSparseToTensorsMap operation.
-   * 
-   * @param scope current scope
-   * @param sparseIndices 2-D.  The `indices` of the `SparseTensor`.
-   * @param sparseValues 1-D.  The `values` of the `SparseTensor`.
-   * @param sparseShape 1-D.  The `shape` of the `SparseTensor`.
-   * @param options carries optional attributes values
-   * @return a new instance of AddSparseToTensorsMap
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> AddSparseToTensorsMap create(Scope scope, Operand<TInt64> sparseIndices, Operand<T> sparseValues, Operand<TInt64> sparseShape, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("AddSparseToTensorsMap", scope.makeOpName("AddSparseToTensorsMap"));
-    opBuilder.addInput(sparseIndices.asOutput(scope));
-    opBuilder.addInput(sparseValues.asOutput(scope));
-    opBuilder.addInput(sparseShape.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.container != null) {
-          opBuilder.setAttr("container", opts.container);
-        }
-        if (opts.sharedName != null) {
-          opBuilder.setAttr("shared_name", opts.sharedName);
-        }
-      }
-    }
-    return new AddSparseToTensorsMap(opBuilder.build());
-  }
-  
-  /**
-   * @param container The container name for the `SparseTensorsMap` created by this op.
-   */
-  public static Options container(String container) {
-    return new Options().container(container);
-  }
-  
-  /**
-   * @param sharedName The shared name for the `SparseTensorsMap` created by this op.
-   * If blank, the new Operation's unique name is used.
-   */
-  public static Options sharedName(String sharedName) {
-    return new Options().sharedName(sharedName);
-  }
-  
-  /**
-   * 0-D.  The handle of the `SparseTensor` now stored in the
-   * `SparseTensorsMap`.
-   */
-  public Output<TInt64> sparseHandle() {
-    return sparseHandle;
-  }
-  
-  @Override
-  public Output<TInt64> asOutput(Scope scope) {
-    return sparseHandle;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "AddSparseToTensorsMap";
-  
-  private Output<TInt64> sparseHandle;
-  
-  private AddSparseToTensorsMap(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    sparseHandle = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/DenseCountSparseOutput.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/DenseCountSparseOutput.java
deleted file mode 100644
index e59f5dbe644..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/DenseCountSparseOutput.java
+++ /dev/null
@@ -1,148 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.sparse;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Performs sparse-output bin counting for a tf.tensor input.
- * <p>
- *   Counts the number of times each value occurs in the input.
- * 
- * @param <U> data type for {@code outputValues()} output
- */
-public final class DenseCountSparseOutput<U extends TNumber> extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.sparse.DenseCountSparseOutput}
-   */
-  public static class Options {
-    
-    /**
-     * @param minlength Minimum value to count. Can be set to -1 for no minimum.
-     */
-    public Options minlength(Long minlength) {
-      this.minlength = minlength;
-      return this;
-    }
-    
-    /**
-     * @param maxlength Maximum value to count. Can be set to -1 for no maximum.
-     */
-    public Options maxlength(Long maxlength) {
-      this.maxlength = maxlength;
-      return this;
-    }
-    
-    private Long minlength;
-    private Long maxlength;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new DenseCountSparseOutput operation.
-   * 
-   * @param scope current scope
-   * @param values Tensor containing data to count.
-   * @param weights A Tensor of the same shape as indices containing per-index weight values. May
-   * also be the empty tensor if no weights are used.
-   * @param binaryOutput Whether to output the number of occurrences of each value or 1.
-   * @param options carries optional attributes values
-   * @return a new instance of DenseCountSparseOutput
-   */
-  @Endpoint(describeByClass = true)
-  public static <U extends TNumber, T extends TNumber> DenseCountSparseOutput<U> create(Scope scope, Operand<T> values, Operand<U> weights, Boolean binaryOutput, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("DenseCountSparseOutput", scope.makeOpName("DenseCountSparseOutput"));
-    opBuilder.addInput(values.asOutput(scope));
-    opBuilder.addInput(weights.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("binary_output", binaryOutput);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.minlength != null) {
-          opBuilder.setAttr("minlength", opts.minlength);
-        }
-        if (opts.maxlength != null) {
-          opBuilder.setAttr("maxlength", opts.maxlength);
-        }
-      }
-    }
-    return new DenseCountSparseOutput<U>(opBuilder.build());
-  }
-  
-  /**
-   * @param minlength Minimum value to count. Can be set to -1 for no minimum.
-   */
-  public static Options minlength(Long minlength) {
-    return new Options().minlength(minlength);
-  }
-  
-  /**
-   * @param maxlength Maximum value to count. Can be set to -1 for no maximum.
-   */
-  public static Options maxlength(Long maxlength) {
-    return new Options().maxlength(maxlength);
-  }
-  
-  /**
-   * Indices tensor for the resulting sparse tensor object.
-   */
-  public Output<TInt64> outputIndices() {
-    return outputIndices;
-  }
-  
-  /**
-   * Values tensor for the resulting sparse tensor object.
-   */
-  public Output<U> outputValues() {
-    return outputValues;
-  }
-  
-  /**
-   * Shape tensor for the resulting sparse tensor object.
-   */
-  public Output<TInt64> outputDenseShape() {
-    return outputDenseShape;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "DenseCountSparseOutput";
-  
-  private Output<TInt64> outputIndices;
-  private Output<U> outputValues;
-  private Output<TInt64> outputDenseShape;
-  
-  private DenseCountSparseOutput(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    outputIndices = operation.output(outputIdx++);
-    outputValues = operation.output(outputIdx++);
-    outputDenseShape = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/DenseToDenseSetOperation.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/DenseToDenseSetOperation.java
deleted file mode 100644
index dacf7725a02..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/DenseToDenseSetOperation.java
+++ /dev/null
@@ -1,139 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.sparse;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
-
-/**
- * Applies set operation along last dimension of 2 `Tensor` inputs.
- * <p>
- * See SetOperationOp::SetOperationFromContext for values of `set_operation`.
- * <p>
- * Output `result` is a `SparseTensor` represented by `result_indices`,
- * `result_values`, and `result_shape`. For `set1` and `set2` ranked `n`, this
- * has rank `n` and the same 1st `n-1` dimensions as `set1` and `set2`. The `nth`
- * dimension contains the result of `set_operation` applied to the corresponding
- * `[0...n-1]` dimension of `set`.
- * 
- * @param <T> data type for {@code resultValues()} output
- */
-@Operator(group = "sparse")
-public final class DenseToDenseSetOperation<T extends TType> extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.sparse.DenseToDenseSetOperation}
-   */
-  public static class Options {
-    
-    /**
-     * @param validateIndices 
-     */
-    public Options validateIndices(Boolean validateIndices) {
-      this.validateIndices = validateIndices;
-      return this;
-    }
-    
-    private Boolean validateIndices;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new DenseToDenseSetOperation operation.
-   * 
-   * @param scope current scope
-   * @param set1 `Tensor` with rank `n`. 1st `n-1` dimensions must be the same as `set2`.
-   * Dimension `n` contains values in a set, duplicates are allowed but ignored.
-   * @param set2 `Tensor` with rank `n`. 1st `n-1` dimensions must be the same as `set1`.
-   * Dimension `n` contains values in a set, duplicates are allowed but ignored.
-   * @param setOperation 
-   * @param options carries optional attributes values
-   * @return a new instance of DenseToDenseSetOperation
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> DenseToDenseSetOperation<T> create(Scope scope, Operand<T> set1, Operand<T> set2, String setOperation, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("DenseToDenseSetOperation", scope.makeOpName("DenseToDenseSetOperation"));
-    opBuilder.addInput(set1.asOutput(scope));
-    opBuilder.addInput(set2.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("set_operation", setOperation);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.validateIndices != null) {
-          opBuilder.setAttr("validate_indices", opts.validateIndices);
-        }
-      }
-    }
-    return new DenseToDenseSetOperation<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param validateIndices 
-   */
-  public static Options validateIndices(Boolean validateIndices) {
-    return new Options().validateIndices(validateIndices);
-  }
-  
-  /**
-   * 2D indices of a `SparseTensor`.
-   */
-  public Output<TInt64> resultIndices() {
-    return resultIndices;
-  }
-  
-  /**
-   * 1D values of a `SparseTensor`.
-   */
-  public Output<T> resultValues() {
-    return resultValues;
-  }
-  
-  /**
-   * 1D `Tensor` shape of a `SparseTensor`. `result_shape[0...n-1]` is
-   * the same as the 1st `n-1` dimensions of `set1` and `set2`, `result_shape[n]`
-   * is the max result set size across all `0...n-1` dimensions.
-   */
-  public Output<TInt64> resultShape() {
-    return resultShape;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "DenseToDenseSetOperation";
-  
-  private Output<TInt64> resultIndices;
-  private Output<T> resultValues;
-  private Output<TInt64> resultShape;
-  
-  private DenseToDenseSetOperation(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    resultIndices = operation.output(outputIdx++);
-    resultValues = operation.output(outputIdx++);
-    resultShape = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/DenseToSparseSetOperation.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/DenseToSparseSetOperation.java
deleted file mode 100644
index f47e96674f8..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/DenseToSparseSetOperation.java
+++ /dev/null
@@ -1,154 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.sparse;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
-
-/**
- * Applies set operation along last dimension of `Tensor` and `SparseTensor`.
- * <p>
- * See SetOperationOp::SetOperationFromContext for values of `set_operation`.
- * <p>
- * Input `set2` is a `SparseTensor` represented by `set2_indices`, `set2_values`,
- * and `set2_shape`. For `set2` ranked `n`, 1st `n-1` dimensions must be the same
- * as `set1`. Dimension `n` contains values in a set, duplicates are allowed but
- * ignored.
- * <p>
- * If `validate_indices` is `True`, this op validates the order and range of `set2`
- * indices.
- * <p>
- * Output `result` is a `SparseTensor` represented by `result_indices`,
- * `result_values`, and `result_shape`. For `set1` and `set2` ranked `n`, this
- * has rank `n` and the same 1st `n-1` dimensions as `set1` and `set2`. The `nth`
- * dimension contains the result of `set_operation` applied to the corresponding
- * `[0...n-1]` dimension of `set`.
- * 
- * @param <T> data type for {@code resultValues()} output
- */
-@Operator(group = "sparse")
-public final class DenseToSparseSetOperation<T extends TType> extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.sparse.DenseToSparseSetOperation}
-   */
-  public static class Options {
-    
-    /**
-     * @param validateIndices 
-     */
-    public Options validateIndices(Boolean validateIndices) {
-      this.validateIndices = validateIndices;
-      return this;
-    }
-    
-    private Boolean validateIndices;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new DenseToSparseSetOperation operation.
-   * 
-   * @param scope current scope
-   * @param set1 `Tensor` with rank `n`. 1st `n-1` dimensions must be the same as `set2`.
-   * Dimension `n` contains values in a set, duplicates are allowed but ignored.
-   * @param set2Indices 2D `Tensor`, indices of a `SparseTensor`. Must be in row-major
-   * order.
-   * @param set2Values 1D `Tensor`, values of a `SparseTensor`. Must be in row-major
-   * order.
-   * @param set2Shape 1D `Tensor`, shape of a `SparseTensor`. `set2_shape[0...n-1]` must
-   * be the same as the 1st `n-1` dimensions of `set1`, `result_shape[n]` is the
-   * max set size across `n-1` dimensions.
-   * @param setOperation 
-   * @param options carries optional attributes values
-   * @return a new instance of DenseToSparseSetOperation
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> DenseToSparseSetOperation<T> create(Scope scope, Operand<T> set1, Operand<TInt64> set2Indices, Operand<T> set2Values, Operand<TInt64> set2Shape, String setOperation, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("DenseToSparseSetOperation", scope.makeOpName("DenseToSparseSetOperation"));
-    opBuilder.addInput(set1.asOutput(scope));
-    opBuilder.addInput(set2Indices.asOutput(scope));
-    opBuilder.addInput(set2Values.asOutput(scope));
-    opBuilder.addInput(set2Shape.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("set_operation", setOperation);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.validateIndices != null) {
-          opBuilder.setAttr("validate_indices", opts.validateIndices);
-        }
-      }
-    }
-    return new DenseToSparseSetOperation<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param validateIndices 
-   */
-  public static Options validateIndices(Boolean validateIndices) {
-    return new Options().validateIndices(validateIndices);
-  }
-  
-  /**
-   * 2D indices of a `SparseTensor`.
-   */
-  public Output<TInt64> resultIndices() {
-    return resultIndices;
-  }
-  
-  /**
-   * 1D values of a `SparseTensor`.
-   */
-  public Output<T> resultValues() {
-    return resultValues;
-  }
-  
-  /**
-   * 1D `Tensor` shape of a `SparseTensor`. `result_shape[0...n-1]` is
-   * the same as the 1st `n-1` dimensions of `set1` and `set2`, `result_shape[n]`
-   * is the max result set size across all `0...n-1` dimensions.
-   */
-  public Output<TInt64> resultShape() {
-    return resultShape;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "DenseToSparseSetOperation";
-  
-  private Output<TInt64> resultIndices;
-  private Output<T> resultValues;
-  private Output<TInt64> resultShape;
-  
-  private DenseToSparseSetOperation(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    resultIndices = operation.output(outputIdx++);
-    resultValues = operation.output(outputIdx++);
-    resultShape = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/DeserializeSparse.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/DeserializeSparse.java
deleted file mode 100644
index b079981f9b1..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/DeserializeSparse.java
+++ /dev/null
@@ -1,131 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.sparse;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
-
-/**
- * Deserialize `SparseTensor` objects.
- * <p>
- * The input `serialized_sparse` must have the shape `[?, ?, ..., ?, 3]` where
- * the last dimension stores serialized `SparseTensor` objects and the other N
- * dimensions (N >= 0) correspond to a batch. The ranks of the original
- * `SparseTensor` objects must all match. When the final `SparseTensor` is
- * created, its rank is the rank of the incoming `SparseTensor` objects plus N;
- * the sparse tensors have been concatenated along new dimensions, one for each
- * batch.
- * <p>
- * The output `SparseTensor` object's shape values for the original dimensions
- * are the max across the input `SparseTensor` objects' shape values for the
- * corresponding dimensions. The new dimensions match the size of the batch.
- * <p>
- * The input `SparseTensor` objects' indices are assumed ordered in
- * standard lexicographic order.  If this is not the case, after this
- * step run `SparseReorder` to restore index ordering.
- * <p>
- * For example, if the serialized input is a `[2 x 3]` matrix representing two
- * original `SparseTensor` objects:
- * <p>
- *     index = [ 0]
- *             [10]
- *             [20]
- *     values = [1, 2, 3]
- *     shape = [50]
- * <p>
- * and
- * <p>
- *     index = [ 2]
- *             [10]
- *     values = [4, 5]
- *     shape = [30]
- * <p>
- * then the final deserialized `SparseTensor` will be:
- * <p>
- *     index = [0  0]
- *             [0 10]
- *             [0 20]
- *             [1  2]
- *             [1 10]
- *     values = [1, 2, 3, 4, 5]
- *     shape = [2 50]
- * 
- * @param <U> data type for {@code sparseValues()} output
- */
-@Operator(group = "sparse")
-public final class DeserializeSparse<U extends TType> extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new DeserializeSparse operation.
-   * 
-   * @param scope current scope
-   * @param serializedSparse The serialized `SparseTensor` objects. The last dimension
-   * must have 3 columns.
-   * @param dtype The `dtype` of the serialized `SparseTensor` objects.
-   * @return a new instance of DeserializeSparse
-   */
-  @Endpoint(describeByClass = true)
-  public static <U extends TType, T extends TType> DeserializeSparse<U> create(Scope scope, Operand<T> serializedSparse, Class<U> dtype) {
-    OperationBuilder opBuilder = scope.env().opBuilder("DeserializeSparse", scope.makeOpName("DeserializeSparse"));
-    opBuilder.addInput(serializedSparse.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("dtype", dtype);
-    return new DeserializeSparse<U>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<TInt64> sparseIndices() {
-    return sparseIndices;
-  }
-  
-  /**
-   */
-  public Output<U> sparseValues() {
-    return sparseValues;
-  }
-  
-  /**
-   */
-  public Output<TInt64> sparseShape() {
-    return sparseShape;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "DeserializeSparse";
-  
-  private Output<TInt64> sparseIndices;
-  private Output<U> sparseValues;
-  private Output<TInt64> sparseShape;
-  
-  private DeserializeSparse(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    sparseIndices = operation.output(outputIdx++);
-    sparseValues = operation.output(outputIdx++);
-    sparseShape = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseAccumulatorApplyGradient.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseAccumulatorApplyGradient.java
deleted file mode 100644
index 99bf0938930..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseAccumulatorApplyGradient.java
+++ /dev/null
@@ -1,75 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.sparse;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.TString;
-import org.tensorflow.types.family.TType;
-
-/**
- * Applies a sparse gradient to a given accumulator.
- * <p>
- * Does not add if local_step is smaller than the accumulator's
- * global_step.
- */
-@Operator(group = "sparse")
-public final class SparseAccumulatorApplyGradient extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new SparseAccumulatorApplyGradient operation.
-   * 
-   * @param scope current scope
-   * @param handle The handle to a accumulator.
-   * @param localStep The local_step value at which the sparse gradient was computed.
-   * @param gradientIndices Indices of the sparse gradient to be accumulated. Must be a
-   * vector.
-   * @param gradientValues Values are the non-zero slices of the gradient, and must have
-   * the same first dimension as indices, i.e., the nnz represented by indices and
-   * values must be consistent.
-   * @param gradientShape Shape of the sparse gradient to be accumulated.
-   * @param hasKnownShape Boolean indicating whether gradient_shape is unknown, in which
-   * case the input is ignored during validation.
-   * @return a new instance of SparseAccumulatorApplyGradient
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> SparseAccumulatorApplyGradient create(Scope scope, Operand<TString> handle, Operand<TInt64> localStep, Operand<TInt64> gradientIndices, Operand<T> gradientValues, Operand<TInt64> gradientShape, Boolean hasKnownShape) {
-    OperationBuilder opBuilder = scope.env().opBuilder("SparseAccumulatorApplyGradient", scope.makeOpName("SparseAccumulatorApplyGradient"));
-    opBuilder.addInput(handle.asOutput(scope));
-    opBuilder.addInput(localStep.asOutput(scope));
-    opBuilder.addInput(gradientIndices.asOutput(scope));
-    opBuilder.addInput(gradientValues.asOutput(scope));
-    opBuilder.addInput(gradientShape.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("has_known_shape", hasKnownShape);
-    return new SparseAccumulatorApplyGradient(opBuilder.build());
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "SparseAccumulatorApplyGradient";
-  
-  private SparseAccumulatorApplyGradient(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseAccumulatorTakeGradient.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseAccumulatorTakeGradient.java
deleted file mode 100644
index 75f34a7155c..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseAccumulatorTakeGradient.java
+++ /dev/null
@@ -1,103 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.sparse;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.TString;
-import org.tensorflow.types.family.TType;
-
-/**
- * Extracts the average sparse gradient in a SparseConditionalAccumulator.
- * <p>
- * The op will blocks until sufficient (i.e., more than num_required)
- * gradients have been accumulated. If the accumulator has already
- * aggregated more than num_required gradients, it will return its
- * average of the accumulated gradients.  Also automatically increments
- * the recorded global_step in the accumulator by 1, and resets the
- * aggregate to 0.
- * 
- * @param <T> data type for {@code values()} output
- */
-@Operator(group = "sparse")
-public final class SparseAccumulatorTakeGradient<T extends TType> extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new SparseAccumulatorTakeGradient operation.
-   * 
-   * @param scope current scope
-   * @param handle The handle to a SparseConditionalAccumulator.
-   * @param numRequired Number of gradients required before we return an aggregate.
-   * @param dtype The data type of accumulated gradients. Needs to correspond to the type
-   * of the accumulator.
-   * @return a new instance of SparseAccumulatorTakeGradient
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> SparseAccumulatorTakeGradient<T> create(Scope scope, Operand<TString> handle, Operand<TInt32> numRequired, Class<T> dtype) {
-    OperationBuilder opBuilder = scope.env().opBuilder("SparseAccumulatorTakeGradient", scope.makeOpName("SparseAccumulatorTakeGradient"));
-    opBuilder.addInput(handle.asOutput(scope));
-    opBuilder.addInput(numRequired.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("dtype", dtype);
-    return new SparseAccumulatorTakeGradient<T>(opBuilder.build());
-  }
-  
-  /**
-   * Indices of the average of the accumulated sparse gradients.
-   */
-  public Output<TInt64> indices() {
-    return indices;
-  }
-  
-  /**
-   * Values of the average of the accumulated sparse gradients.
-   */
-  public Output<T> values() {
-    return values;
-  }
-  
-  /**
-   * Shape of the average of the accumulated sparse gradients.
-   */
-  public Output<TInt64> shape() {
-    return shape;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "SparseAccumulatorTakeGradient";
-  
-  private Output<TInt64> indices;
-  private Output<T> values;
-  private Output<TInt64> shape;
-  
-  private SparseAccumulatorTakeGradient(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    indices = operation.output(outputIdx++);
-    values = operation.output(outputIdx++);
-    shape = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseAdd.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseAdd.java
deleted file mode 100644
index 7377c27b799..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseAdd.java
+++ /dev/null
@@ -1,114 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.sparse;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * Adds two `SparseTensor` objects to produce another `SparseTensor`.
- * <p>
- * The input `SparseTensor` objects' indices are assumed ordered in standard
- * lexicographic order.  If this is not the case, before this step run
- * `SparseReorder` to restore index ordering.
- * <p>
- * By default, if two values sum to zero at some index, the output `SparseTensor`
- * would still include that particular location in its index, storing a zero in the
- * corresponding value slot.  To override this, callers can specify `thresh`,
- * indicating that if the sum has a magnitude strictly smaller than `thresh`, its
- * corresponding value and index would then not be included.  In particular,
- * `thresh == 0` (default) means everything is kept and actual thresholding happens
- * only for a positive value.
- * <p>
- * In the following shapes, `nnz` is the count after taking `thresh` into account.
- * 
- * @param <T> data type for {@code sumValues()} output
- */
-@Operator(group = "sparse")
-public final class SparseAdd<T extends TType> extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new SparseAdd operation.
-   * 
-   * @param scope current scope
-   * @param aIndices 2-D.  The `indices` of the first `SparseTensor`, size `[nnz, ndims]` Matrix.
-   * @param aValues 1-D.  The `values` of the first `SparseTensor`, size `[nnz]` Vector.
-   * @param aShape 1-D.  The `shape` of the first `SparseTensor`, size `[ndims]` Vector.
-   * @param bIndices 2-D.  The `indices` of the second `SparseTensor`, size `[nnz, ndims]` Matrix.
-   * @param bValues 1-D.  The `values` of the second `SparseTensor`, size `[nnz]` Vector.
-   * @param bShape 1-D.  The `shape` of the second `SparseTensor`, size `[ndims]` Vector.
-   * @param thresh 0-D.  The magnitude threshold that determines if an output value/index
-   * pair takes space.
-   * @return a new instance of SparseAdd
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> SparseAdd<T> create(Scope scope, Operand<TInt64> aIndices, Operand<T> aValues, Operand<TInt64> aShape, Operand<TInt64> bIndices, Operand<T> bValues, Operand<TInt64> bShape, Operand<U> thresh) {
-    OperationBuilder opBuilder = scope.env().opBuilder("SparseAdd", scope.makeOpName("SparseAdd"));
-    opBuilder.addInput(aIndices.asOutput(scope));
-    opBuilder.addInput(aValues.asOutput(scope));
-    opBuilder.addInput(aShape.asOutput(scope));
-    opBuilder.addInput(bIndices.asOutput(scope));
-    opBuilder.addInput(bValues.asOutput(scope));
-    opBuilder.addInput(bShape.asOutput(scope));
-    opBuilder.addInput(thresh.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new SparseAdd<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<TInt64> sumIndices() {
-    return sumIndices;
-  }
-  
-  /**
-   */
-  public Output<T> sumValues() {
-    return sumValues;
-  }
-  
-  /**
-   */
-  public Output<TInt64> sumShape() {
-    return sumShape;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "SparseAdd";
-  
-  private Output<TInt64> sumIndices;
-  private Output<T> sumValues;
-  private Output<TInt64> sumShape;
-  
-  private SparseAdd(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    sumIndices = operation.output(outputIdx++);
-    sumValues = operation.output(outputIdx++);
-    sumShape = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseAddGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseAddGrad.java
deleted file mode 100644
index f7f84d2be77..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseAddGrad.java
+++ /dev/null
@@ -1,95 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.sparse;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
-
-/**
- * The gradient operator for the SparseAdd op.
- * <p>
- * The SparseAdd op calculates A + B, where A, B, and the sum are all represented
- * as `SparseTensor` objects.  This op takes in the upstream gradient w.r.t.
- * non-empty values of the sum, and outputs the gradients w.r.t. the non-empty
- * values of A and B.
- * 
- * @param <T> data type for {@code aValGrad()} output
- */
-@Operator(group = "sparse")
-public final class SparseAddGrad<T extends TType> extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new SparseAddGrad operation.
-   * 
-   * @param scope current scope
-   * @param backpropValGrad 1-D with shape `[nnz(sum)]`.  The gradient with respect to
-   * the non-empty values of the sum.
-   * @param aIndices 2-D.  The `indices` of the `SparseTensor` A, size `[nnz(A), ndims]`.
-   * @param bIndices 2-D.  The `indices` of the `SparseTensor` B, size `[nnz(B), ndims]`.
-   * @param sumIndices 2-D.  The `indices` of the sum `SparseTensor`, size
-   * `[nnz(sum), ndims]`.
-   * @return a new instance of SparseAddGrad
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> SparseAddGrad<T> create(Scope scope, Operand<T> backpropValGrad, Operand<TInt64> aIndices, Operand<TInt64> bIndices, Operand<TInt64> sumIndices) {
-    OperationBuilder opBuilder = scope.env().opBuilder("SparseAddGrad", scope.makeOpName("SparseAddGrad"));
-    opBuilder.addInput(backpropValGrad.asOutput(scope));
-    opBuilder.addInput(aIndices.asOutput(scope));
-    opBuilder.addInput(bIndices.asOutput(scope));
-    opBuilder.addInput(sumIndices.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new SparseAddGrad<T>(opBuilder.build());
-  }
-  
-  /**
-   * 1-D with shape `[nnz(A)]`. The gradient with respect to the
-   * non-empty values of A.
-   */
-  public Output<T> aValGrad() {
-    return aValGrad;
-  }
-  
-  /**
-   * 1-D with shape `[nnz(B)]`. The gradient with respect to the
-   * non-empty values of B.
-   */
-  public Output<T> bValGrad() {
-    return bValGrad;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "SparseAddGrad";
-  
-  private Output<T> aValGrad;
-  private Output<T> bValGrad;
-  
-  private SparseAddGrad(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    aValGrad = operation.output(outputIdx++);
-    bValGrad = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseBincount.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseBincount.java
deleted file mode 100644
index 3f41ac7bb37..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseBincount.java
+++ /dev/null
@@ -1,129 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.sparse;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Counts the number of occurrences of each value in an integer array.
- * <p>
- * Outputs a vector with length `size` and the same dtype as `weights`. If
- * `weights` are empty, then index `i` stores the number of times the value `i` is
- * counted in `arr`. If `weights` are non-empty, then index `i` stores the sum of
- * the value in `weights` at each index where the corresponding value in `arr` is
- * `i`.
- * <p>
- * Values in `arr` outside of the range [0, size) are ignored.
- * 
- * @param <U> data type for {@code output()} output
- */
-@Operator(group = "sparse")
-public final class SparseBincount<U extends TNumber> extends RawOp implements Operand<U> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.sparse.SparseBincount}
-   */
-  public static class Options {
-    
-    /**
-     * @param binaryOutput bool; Whether the kernel should count the appearance or number of occurrences.
-     */
-    public Options binaryOutput(Boolean binaryOutput) {
-      this.binaryOutput = binaryOutput;
-      return this;
-    }
-    
-    private Boolean binaryOutput;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new SparseBincount operation.
-   * 
-   * @param scope current scope
-   * @param indices 2D int64 `Tensor`.
-   * @param values 1D int `Tensor`.
-   * @param denseShape 1D int64 `Tensor`.
-   * @param size non-negative int scalar `Tensor`.
-   * @param weights is an int32, int64, float32, or float64 `Tensor` with the same
-   * shape as `input`, or a length-0 `Tensor`, in which case it acts as all weights
-   * equal to 1.
-   * @param options carries optional attributes values
-   * @return a new instance of SparseBincount
-   */
-  @Endpoint(describeByClass = true)
-  public static <U extends TNumber, T extends TNumber> SparseBincount<U> create(Scope scope, Operand<TInt64> indices, Operand<T> values, Operand<TInt64> denseShape, Operand<T> size, Operand<U> weights, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("SparseBincount", scope.makeOpName("SparseBincount"));
-    opBuilder.addInput(indices.asOutput(scope));
-    opBuilder.addInput(values.asOutput(scope));
-    opBuilder.addInput(denseShape.asOutput(scope));
-    opBuilder.addInput(size.asOutput(scope));
-    opBuilder.addInput(weights.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.binaryOutput != null) {
-          opBuilder.setAttr("binary_output", opts.binaryOutput);
-        }
-      }
-    }
-    return new SparseBincount<U>(opBuilder.build());
-  }
-  
-  /**
-   * @param binaryOutput bool; Whether the kernel should count the appearance or number of occurrences.
-   */
-  public static Options binaryOutput(Boolean binaryOutput) {
-    return new Options().binaryOutput(binaryOutput);
-  }
-  
-  /**
-   * 1D `Tensor` with length equal to `size` or 2D `Tensor` with [batch_size, `size`].
-   * The counts or summed weights for each value in the range [0, size).
-   */
-  public Output<U> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<U> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "SparseBincount";
-  
-  private Output<U> output;
-  
-  private SparseBincount(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseConcat.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseConcat.java
deleted file mode 100644
index 53eecb4b967..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseConcat.java
+++ /dev/null
@@ -1,139 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.sparse;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.Operands;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
-
-/**
- * Concatenates a list of `SparseTensor` along the specified dimension.
- * <p>
- * Concatenation is with respect to the dense versions of these sparse tensors.
- * It is assumed that each input is a `SparseTensor` whose elements are ordered
- * along increasing dimension number.
- * <p>
- * All inputs' shapes must match, except for the concat dimension.  The
- * `indices`, `values`, and `shapes` lists must have the same length.
- * <p>
- * The output shape is identical to the inputs', except along the concat
- * dimension, where it is the sum of the inputs' sizes along that dimension.
- * <p>
- * The output elements will be resorted to preserve the sort order along
- * increasing dimension number.
- * <p>
- * This op runs in `O(M log M)` time, where `M` is the total number of non-empty
- * values across all inputs. This is due to the need for an internal sort in
- * order to concatenate efficiently across an arbitrary dimension.
- * <p>
- * For example, if `concat_dim = 1` and the inputs are
- * <p>
- *     sp_inputs[0]: shape = [2, 3]
- *     [0, 2]: "a"
- *     [1, 0]: "b"
- *     [1, 1]: "c"
- * <p>
- *     sp_inputs[1]: shape = [2, 4]
- *     [0, 1]: "d"
- *     [0, 2]: "e"
- * <p>
- * then the output will be
- * <p>
- *     shape = [2, 7]
- *     [0, 2]: "a"
- *     [0, 4]: "d"
- *     [0, 5]: "e"
- *     [1, 0]: "b"
- *     [1, 1]: "c"
- * <p>
- * Graphically this is equivalent to doing
- * <p>
- *     [    a] concat [  d e  ] = [    a   d e  ]
- *     [b c  ]        [       ]   [b c          ]
- * 
- * @param <T> data type for {@code outputValues()} output
- */
-@Operator(group = "sparse")
-public final class SparseConcat<T extends TType> extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new SparseConcat operation.
-   * 
-   * @param scope current scope
-   * @param indices 2-D.  Indices of each input `SparseTensor`.
-   * @param values 1-D.  Non-empty values of each `SparseTensor`.
-   * @param shapes 1-D.  Shapes of each `SparseTensor`.
-   * @param concatDim Dimension to concatenate along. Must be in range [-rank, rank),
-   * where rank is the number of dimensions in each input `SparseTensor`.
-   * @return a new instance of SparseConcat
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> SparseConcat<T> create(Scope scope, Iterable<Operand<TInt64>> indices, Iterable<Operand<T>> values, Iterable<Operand<TInt64>> shapes, Long concatDim) {
-    OperationBuilder opBuilder = scope.env().opBuilder("SparseConcat", scope.makeOpName("SparseConcat"));
-    opBuilder.addInputList(Operands.asOutputs(scope, indices));
-    opBuilder.addInputList(Operands.asOutputs(scope, values));
-    opBuilder.addInputList(Operands.asOutputs(scope, shapes));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("concat_dim", concatDim);
-    return new SparseConcat<T>(opBuilder.build());
-  }
-  
-  /**
-   * 2-D.  Indices of the concatenated `SparseTensor`.
-   */
-  public Output<TInt64> outputIndices() {
-    return outputIndices;
-  }
-  
-  /**
-   * 1-D.  Non-empty values of the concatenated `SparseTensor`.
-   */
-  public Output<T> outputValues() {
-    return outputValues;
-  }
-  
-  /**
-   * 1-D.  Shape of the concatenated `SparseTensor`.
-   */
-  public Output<TInt64> outputShape() {
-    return outputShape;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "SparseConcat";
-  
-  private Output<TInt64> outputIndices;
-  private Output<T> outputValues;
-  private Output<TInt64> outputShape;
-  
-  private SparseConcat(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    outputIndices = operation.output(outputIdx++);
-    outputValues = operation.output(outputIdx++);
-    outputShape = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseConditionalAccumulator.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseConditionalAccumulator.java
deleted file mode 100644
index 045169e2f9a..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseConditionalAccumulator.java
+++ /dev/null
@@ -1,160 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.sparse;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TString;
-import org.tensorflow.types.family.TType;
-
-/**
- * A conditional accumulator for aggregating sparse gradients.
- * <p>
- * The accumulator accepts gradients marked with local_step greater or
- * equal to the most recent global_step known to the accumulator. The
- * average can be extracted from the accumulator, provided sufficient
- * gradients have been accumulated. Extracting the average automatically
- * resets the aggregate to 0, and increments the global_step recorded by
- * the accumulator.
- */
-@Operator(group = "sparse")
-public final class SparseConditionalAccumulator extends RawOp implements Operand<TString> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.sparse.SparseConditionalAccumulator}
-   */
-  public static class Options {
-    
-    /**
-     * @param container If non-empty, this accumulator is placed in the given container.
-     * Otherwise, a default container is used.
-     */
-    public Options container(String container) {
-      this.container = container;
-      return this;
-    }
-    
-    /**
-     * @param sharedName If non-empty, this accumulator will be shared under the given name
-     * across multiple sessions.
-     */
-    public Options sharedName(String sharedName) {
-      this.sharedName = sharedName;
-      return this;
-    }
-    
-    /**
-     * @param reductionType 
-     */
-    public Options reductionType(String reductionType) {
-      this.reductionType = reductionType;
-      return this;
-    }
-    
-    private String container;
-    private String sharedName;
-    private String reductionType;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new SparseConditionalAccumulator operation.
-   * 
-   * @param scope current scope
-   * @param dtype The type of the value being accumulated.
-   * @param shape The shape of the values.
-   * @param options carries optional attributes values
-   * @return a new instance of SparseConditionalAccumulator
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> SparseConditionalAccumulator create(Scope scope, Class<T> dtype, Shape shape, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("SparseConditionalAccumulator", scope.makeOpName("SparseConditionalAccumulator"));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("dtype", dtype);
-    opBuilder.setAttr("shape", shape);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.container != null) {
-          opBuilder.setAttr("container", opts.container);
-        }
-        if (opts.sharedName != null) {
-          opBuilder.setAttr("shared_name", opts.sharedName);
-        }
-        if (opts.reductionType != null) {
-          opBuilder.setAttr("reduction_type", opts.reductionType);
-        }
-      }
-    }
-    return new SparseConditionalAccumulator(opBuilder.build());
-  }
-  
-  /**
-   * @param container If non-empty, this accumulator is placed in the given container.
-   * Otherwise, a default container is used.
-   */
-  public static Options container(String container) {
-    return new Options().container(container);
-  }
-  
-  /**
-   * @param sharedName If non-empty, this accumulator will be shared under the given name
-   * across multiple sessions.
-   */
-  public static Options sharedName(String sharedName) {
-    return new Options().sharedName(sharedName);
-  }
-  
-  /**
-   * @param reductionType 
-   */
-  public static Options reductionType(String reductionType) {
-    return new Options().reductionType(reductionType);
-  }
-  
-  /**
-   * The handle to the accumulator.
-   */
-  public Output<TString> handle() {
-    return handle;
-  }
-  
-  @Override
-  public Output<TString> asOutput(Scope scope) {
-    return handle;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "SparseConditionalAccumulator";
-  
-  private Output<TString> handle;
-  
-  private SparseConditionalAccumulator(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    handle = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseCountSparseOutput.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseCountSparseOutput.java
deleted file mode 100644
index 452a1830055..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseCountSparseOutput.java
+++ /dev/null
@@ -1,152 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.sparse;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Performs sparse-output bin counting for a sparse tensor input.
- * <p>
- *   Counts the number of times each value occurs in the input.
- * 
- * @param <U> data type for {@code outputValues()} output
- */
-public final class SparseCountSparseOutput<U extends TNumber> extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.sparse.SparseCountSparseOutput}
-   */
-  public static class Options {
-    
-    /**
-     * @param minlength Minimum value to count. Can be set to -1 for no minimum.
-     */
-    public Options minlength(Long minlength) {
-      this.minlength = minlength;
-      return this;
-    }
-    
-    /**
-     * @param maxlength Maximum value to count. Can be set to -1 for no maximum.
-     */
-    public Options maxlength(Long maxlength) {
-      this.maxlength = maxlength;
-      return this;
-    }
-    
-    private Long minlength;
-    private Long maxlength;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new SparseCountSparseOutput operation.
-   * 
-   * @param scope current scope
-   * @param indices Tensor containing the indices of the sparse tensor to count.
-   * @param values Tensor containing values of the sparse tensor to count.
-   * @param denseShape Tensor containing the dense shape of the sparse tensor to count.
-   * @param weights A Tensor of the same shape as indices containing per-index weight values.
-   * May also be the empty tensor if no weights are used.
-   * @param binaryOutput Whether to output the number of occurrences of each value or 1.
-   * @param options carries optional attributes values
-   * @return a new instance of SparseCountSparseOutput
-   */
-  @Endpoint(describeByClass = true)
-  public static <U extends TNumber, T extends TNumber> SparseCountSparseOutput<U> create(Scope scope, Operand<TInt64> indices, Operand<T> values, Operand<TInt64> denseShape, Operand<U> weights, Boolean binaryOutput, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("SparseCountSparseOutput", scope.makeOpName("SparseCountSparseOutput"));
-    opBuilder.addInput(indices.asOutput(scope));
-    opBuilder.addInput(values.asOutput(scope));
-    opBuilder.addInput(denseShape.asOutput(scope));
-    opBuilder.addInput(weights.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("binary_output", binaryOutput);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.minlength != null) {
-          opBuilder.setAttr("minlength", opts.minlength);
-        }
-        if (opts.maxlength != null) {
-          opBuilder.setAttr("maxlength", opts.maxlength);
-        }
-      }
-    }
-    return new SparseCountSparseOutput<U>(opBuilder.build());
-  }
-  
-  /**
-   * @param minlength Minimum value to count. Can be set to -1 for no minimum.
-   */
-  public static Options minlength(Long minlength) {
-    return new Options().minlength(minlength);
-  }
-  
-  /**
-   * @param maxlength Maximum value to count. Can be set to -1 for no maximum.
-   */
-  public static Options maxlength(Long maxlength) {
-    return new Options().maxlength(maxlength);
-  }
-  
-  /**
-   * Indices tensor for the resulting sparse tensor object.
-   */
-  public Output<TInt64> outputIndices() {
-    return outputIndices;
-  }
-  
-  /**
-   * Values tensor for the resulting sparse tensor object.
-   */
-  public Output<U> outputValues() {
-    return outputValues;
-  }
-  
-  /**
-   * Shape tensor for the resulting sparse tensor object.
-   */
-  public Output<TInt64> outputDenseShape() {
-    return outputDenseShape;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "SparseCountSparseOutput";
-  
-  private Output<TInt64> outputIndices;
-  private Output<U> outputValues;
-  private Output<TInt64> outputDenseShape;
-  
-  private SparseCountSparseOutput(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    outputIndices = operation.output(outputIdx++);
-    outputValues = operation.output(outputIdx++);
-    outputDenseShape = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseCross.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseCross.java
deleted file mode 100644
index 0b3a25c3eb7..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseCross.java
+++ /dev/null
@@ -1,134 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.sparse;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.Operands;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.TString;
-
-/**
- * Generates sparse cross from a list of sparse and dense tensors.
- * <p>
- * The op takes two lists, one of 2D `SparseTensor` and one of 2D `Tensor`, each
- * representing features of one feature column. It outputs a 2D `SparseTensor` with
- * the batchwise crosses of these features.
- * <p>
- * For example, if the inputs are
- * <p>
- *     inputs[0]: SparseTensor with shape = [2, 2]
- *     [0, 0]: "a"
- *     [1, 0]: "b"
- *     [1, 1]: "c"
- * <p>
- *     inputs[1]: SparseTensor with shape = [2, 1]
- *     [0, 0]: "d"
- *     [1, 0]: "e"
- * <p>
- *     inputs[2]: Tensor [["f"], ["g"]]
- * <p>
- * then the output will be
- * <p>
- *     shape = [2, 2]
- *     [0, 0]: "a_X_d_X_f"
- *     [1, 0]: "b_X_e_X_g"
- *     [1, 1]: "c_X_e_X_g"
- * <p>
- * if hashed_output=true then the output will be
- * <p>
- *     shape = [2, 2]
- *     [0, 0]: FingerprintCat64(
- *                 Fingerprint64("f"), FingerprintCat64(
- *                     Fingerprint64("d"), Fingerprint64("a")))
- *     [1, 0]: FingerprintCat64(
- *                 Fingerprint64("g"), FingerprintCat64(
- *                     Fingerprint64("e"), Fingerprint64("b")))
- *     [1, 1]: FingerprintCat64(
- *                 Fingerprint64("g"), FingerprintCat64(
- *                     Fingerprint64("e"), Fingerprint64("c")))
- */
-@Operator(group = "sparse")
-public final class SparseCross extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new SparseCross operation.
-   * 
-   * @param scope current scope
-   * @param indices 2-D.  Indices of each input `SparseTensor`.
-   * @param values 1-D.   values of each `SparseTensor`.
-   * @param shapes 1-D.   Shapes of each `SparseTensor`.
-   * @param denseInputs 2-D.    Columns represented by dense `Tensor`.
-   * @param sep string used when joining a list of string inputs, can be used as separator later.
-   * @return a new instance of SparseCross
-   */
-  @Endpoint(describeByClass = true)
-  public static SparseCross create(Scope scope, Iterable<Operand<TInt64>> indices, Iterable<Operand<?>> values, Iterable<Operand<TInt64>> shapes, Iterable<Operand<?>> denseInputs, Operand<TString> sep) {
-    OperationBuilder opBuilder = scope.env().opBuilder("SparseCrossV2", scope.makeOpName("SparseCross"));
-    opBuilder.addInputList(Operands.asOutputs(scope, indices));
-    opBuilder.addInputList(Operands.asOutputs(scope, values));
-    opBuilder.addInputList(Operands.asOutputs(scope, shapes));
-    opBuilder.addInputList(Operands.asOutputs(scope, denseInputs));
-    opBuilder.addInput(sep.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new SparseCross(opBuilder.build());
-  }
-  
-  /**
-   * 2-D.  Indices of the concatenated `SparseTensor`.
-   */
-  public Output<TInt64> outputIndices() {
-    return outputIndices;
-  }
-  
-  /**
-   * 1-D.  Non-empty values of the concatenated or hashed
-   * `SparseTensor`.
-   */
-  public Output<TString> outputValues() {
-    return outputValues;
-  }
-  
-  /**
-   * 1-D.  Shape of the concatenated `SparseTensor`.
-   */
-  public Output<TInt64> outputShape() {
-    return outputShape;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "SparseCrossV2";
-  
-  private Output<TInt64> outputIndices;
-  private Output<TString> outputValues;
-  private Output<TInt64> outputShape;
-  
-  private SparseCross(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    outputIndices = operation.output(outputIdx++);
-    outputValues = operation.output(outputIdx++);
-    outputShape = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseCrossHashed.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseCrossHashed.java
deleted file mode 100644
index 9b65e55d0b4..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseCrossHashed.java
+++ /dev/null
@@ -1,139 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.sparse;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.Operands;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TBool;
-import org.tensorflow.types.TInt64;
-
-/**
- * Generates sparse cross from a list of sparse and dense tensors.
- * <p>
- * The op takes two lists, one of 2D `SparseTensor` and one of 2D `Tensor`, each
- * representing features of one feature column. It outputs a 2D `SparseTensor` with
- * the batchwise crosses of these features.
- * <p>
- * For example, if the inputs are
- * <p>
- *     inputs[0]: SparseTensor with shape = [2, 2]
- *     [0, 0]: "a"
- *     [1, 0]: "b"
- *     [1, 1]: "c"
- * <p>
- *     inputs[1]: SparseTensor with shape = [2, 1]
- *     [0, 0]: "d"
- *     [1, 0]: "e"
- * <p>
- *     inputs[2]: Tensor [["f"], ["g"]]
- * <p>
- * then the output will be
- * <p>
- *     shape = [2, 2]
- *     [0, 0]: "a_X_d_X_f"
- *     [1, 0]: "b_X_e_X_g"
- *     [1, 1]: "c_X_e_X_g"
- * <p>
- * if hashed_output=true then the output will be
- * <p>
- *     shape = [2, 2]
- *     [0, 0]: FingerprintCat64(
- *                 Fingerprint64("f"), FingerprintCat64(
- *                     Fingerprint64("d"), Fingerprint64("a")))
- *     [1, 0]: FingerprintCat64(
- *                 Fingerprint64("g"), FingerprintCat64(
- *                     Fingerprint64("e"), Fingerprint64("b")))
- *     [1, 1]: FingerprintCat64(
- *                 Fingerprint64("g"), FingerprintCat64(
- *                     Fingerprint64("e"), Fingerprint64("c")))
- */
-@Operator(group = "sparse")
-public final class SparseCrossHashed extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new SparseCrossHashed operation.
-   * 
-   * @param scope current scope
-   * @param indices 2-D.  Indices of each input `SparseTensor`.
-   * @param values 1-D.   values of each `SparseTensor`.
-   * @param shapes 1-D.   Shapes of each `SparseTensor`.
-   * @param denseInputs 2-D.    Columns represented by dense `Tensor`.
-   * @param numBuckets It is used if hashed_output is true.
-   * output = hashed_value%num_buckets if num_buckets > 0 else hashed_value.
-   * @param strongHash boolean, if true, siphash with salt will be used instead of farmhash.
-   * @param salt Specify the salt that will be used by the siphash function.
-   * @return a new instance of SparseCrossHashed
-   */
-  @Endpoint(describeByClass = true)
-  public static SparseCrossHashed create(Scope scope, Iterable<Operand<TInt64>> indices, Iterable<Operand<?>> values, Iterable<Operand<TInt64>> shapes, Iterable<Operand<?>> denseInputs, Operand<TInt64> numBuckets, Operand<TBool> strongHash, Operand<TInt64> salt) {
-    OperationBuilder opBuilder = scope.env().opBuilder("SparseCrossHashed", scope.makeOpName("SparseCrossHashed"));
-    opBuilder.addInputList(Operands.asOutputs(scope, indices));
-    opBuilder.addInputList(Operands.asOutputs(scope, values));
-    opBuilder.addInputList(Operands.asOutputs(scope, shapes));
-    opBuilder.addInputList(Operands.asOutputs(scope, denseInputs));
-    opBuilder.addInput(numBuckets.asOutput(scope));
-    opBuilder.addInput(strongHash.asOutput(scope));
-    opBuilder.addInput(salt.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new SparseCrossHashed(opBuilder.build());
-  }
-  
-  /**
-   * 2-D.  Indices of the concatenated `SparseTensor`.
-   */
-  public Output<TInt64> outputIndices() {
-    return outputIndices;
-  }
-  
-  /**
-   * 1-D.  Non-empty values of the concatenated or hashed
-   * `SparseTensor`.
-   */
-  public Output<TInt64> outputValues() {
-    return outputValues;
-  }
-  
-  /**
-   * 1-D.  Shape of the concatenated `SparseTensor`.
-   */
-  public Output<TInt64> outputShape() {
-    return outputShape;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "SparseCrossHashed";
-  
-  private Output<TInt64> outputIndices;
-  private Output<TInt64> outputValues;
-  private Output<TInt64> outputShape;
-  
-  private SparseCrossHashed(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    outputIndices = operation.output(outputIdx++);
-    outputValues = operation.output(outputIdx++);
-    outputShape = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseDenseCwiseAdd.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseDenseCwiseAdd.java
deleted file mode 100644
index f3ceaa81bbc..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseDenseCwiseAdd.java
+++ /dev/null
@@ -1,92 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.sparse;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
-
-/**
- * Adds up a SparseTensor and a dense Tensor, using these special rules:
- * <p>
- * (1) Broadcasts the dense side to have the same shape as the sparse side, if
- *     eligible;
- * (2) Then, only the dense values pointed to by the indices of the SparseTensor
- *     participate in the cwise addition.
- * <p>
- * By these rules, the result is a logical SparseTensor with exactly the same
- * indices and shape, but possibly with different non-zero values.  The output of
- * this Op is the resultant non-zero values.
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator(group = "sparse")
-public final class SparseDenseCwiseAdd<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new SparseDenseCwiseAdd operation.
-   * 
-   * @param scope current scope
-   * @param spIndices 2-D.  `N x R` matrix with the indices of non-empty values in a
-   * SparseTensor, possibly not in canonical ordering.
-   * @param spValues 1-D.  `N` non-empty values corresponding to `sp_indices`.
-   * @param spShape 1-D.  Shape of the input SparseTensor.
-   * @param dense `R`-D.  The dense Tensor operand.
-   * @return a new instance of SparseDenseCwiseAdd
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> SparseDenseCwiseAdd<T> create(Scope scope, Operand<TInt64> spIndices, Operand<T> spValues, Operand<TInt64> spShape, Operand<T> dense) {
-    OperationBuilder opBuilder = scope.env().opBuilder("SparseDenseCwiseAdd", scope.makeOpName("SparseDenseCwiseAdd"));
-    opBuilder.addInput(spIndices.asOutput(scope));
-    opBuilder.addInput(spValues.asOutput(scope));
-    opBuilder.addInput(spShape.asOutput(scope));
-    opBuilder.addInput(dense.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new SparseDenseCwiseAdd<T>(opBuilder.build());
-  }
-  
-  /**
-   * 1-D.  The `N` values that are operated on.
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "SparseDenseCwiseAdd";
-  
-  private Output<T> output;
-  
-  private SparseDenseCwiseAdd(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseDenseCwiseDiv.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseDenseCwiseDiv.java
deleted file mode 100644
index 3d377ac1485..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseDenseCwiseDiv.java
+++ /dev/null
@@ -1,86 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.sparse;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
-
-/**
- * Component-wise divides a SparseTensor by a dense Tensor.
- * <p>
- * <i>Limitation</i>: this Op only broadcasts the dense side to the sparse side, but not
- * the other direction.
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator(group = "sparse")
-public final class SparseDenseCwiseDiv<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new SparseDenseCwiseDiv operation.
-   * 
-   * @param scope current scope
-   * @param spIndices 2-D.  `N x R` matrix with the indices of non-empty values in a
-   * SparseTensor, possibly not in canonical ordering.
-   * @param spValues 1-D.  `N` non-empty values corresponding to `sp_indices`.
-   * @param spShape 1-D.  Shape of the input SparseTensor.
-   * @param dense `R`-D.  The dense Tensor operand.
-   * @return a new instance of SparseDenseCwiseDiv
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> SparseDenseCwiseDiv<T> create(Scope scope, Operand<TInt64> spIndices, Operand<T> spValues, Operand<TInt64> spShape, Operand<T> dense) {
-    OperationBuilder opBuilder = scope.env().opBuilder("SparseDenseCwiseDiv", scope.makeOpName("SparseDenseCwiseDiv"));
-    opBuilder.addInput(spIndices.asOutput(scope));
-    opBuilder.addInput(spValues.asOutput(scope));
-    opBuilder.addInput(spShape.asOutput(scope));
-    opBuilder.addInput(dense.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new SparseDenseCwiseDiv<T>(opBuilder.build());
-  }
-  
-  /**
-   * 1-D.  The `N` values that are operated on.
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "SparseDenseCwiseDiv";
-  
-  private Output<T> output;
-  
-  private SparseDenseCwiseDiv(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseDenseCwiseMul.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseDenseCwiseMul.java
deleted file mode 100644
index f6d0b868c90..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseDenseCwiseMul.java
+++ /dev/null
@@ -1,90 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.sparse;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
-
-/**
- * Component-wise multiplies a SparseTensor by a dense Tensor.
- * <p>
- * The output locations corresponding to the implicitly zero elements in the sparse
- * tensor will be zero (i.e., will not take up storage space), regardless of the
- * contents of the dense tensor (even if it's +/-INF and that INF<i>0 == NaN).
- * <p>
- * </i>Limitation*: this Op only broadcasts the dense side to the sparse side, but not
- * the other direction.
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator(group = "sparse")
-public final class SparseDenseCwiseMul<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new SparseDenseCwiseMul operation.
-   * 
-   * @param scope current scope
-   * @param spIndices 2-D.  `N x R` matrix with the indices of non-empty values in a
-   * SparseTensor, possibly not in canonical ordering.
-   * @param spValues 1-D.  `N` non-empty values corresponding to `sp_indices`.
-   * @param spShape 1-D.  Shape of the input SparseTensor.
-   * @param dense `R`-D.  The dense Tensor operand.
-   * @return a new instance of SparseDenseCwiseMul
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> SparseDenseCwiseMul<T> create(Scope scope, Operand<TInt64> spIndices, Operand<T> spValues, Operand<TInt64> spShape, Operand<T> dense) {
-    OperationBuilder opBuilder = scope.env().opBuilder("SparseDenseCwiseMul", scope.makeOpName("SparseDenseCwiseMul"));
-    opBuilder.addInput(spIndices.asOutput(scope));
-    opBuilder.addInput(spValues.asOutput(scope));
-    opBuilder.addInput(spShape.asOutput(scope));
-    opBuilder.addInput(dense.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new SparseDenseCwiseMul<T>(opBuilder.build());
-  }
-  
-  /**
-   * 1-D.  The `N` values that are operated on.
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "SparseDenseCwiseMul";
-  
-  private Output<T> output;
-  
-  private SparseDenseCwiseMul(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseFillEmptyRows.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseFillEmptyRows.java
deleted file mode 100644
index c54860b62ca..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseFillEmptyRows.java
+++ /dev/null
@@ -1,144 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.sparse;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TBool;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
-
-/**
- * Fills empty rows in the input 2-D `SparseTensor` with a default value.
- * <p>
- * The input `SparseTensor` is represented via the tuple of inputs
- * (`indices`, `values`, `dense_shape`).  The output `SparseTensor` has the
- * same `dense_shape` but with indices `output_indices` and values
- * `output_values`.
- * <p>
- * This op inserts a single entry for every row that doesn't have any values.
- * The index is created as `[row, 0, ..., 0]` and the inserted value
- * is `default_value`.
- * <p>
- * For example, suppose `sp_input` has shape `[5, 6]` and non-empty values:
- * <p>
- *     [0, 1]: a
- *     [0, 3]: b
- *     [2, 0]: c
- *     [3, 1]: d
- * <p>
- * Rows 1 and 4 are empty, so the output will be of shape `[5, 6]` with values:
- * <p>
- *     [0, 1]: a
- *     [0, 3]: b
- *     [1, 0]: default_value
- *     [2, 0]: c
- *     [3, 1]: d
- *     [4, 0]: default_value
- * <p>
- * The output `SparseTensor` will be in row-major order and will have the
- * same shape as the input.
- * <p>
- * This op also returns an indicator vector shaped `[dense_shape[0]]` such that
- * <p>
- *     empty_row_indicator[i] = True iff row i was an empty row.
- * <p>
- * And a reverse index map vector shaped `[indices.shape[0]]` that is used during
- * backpropagation,
- * <p>
- *     reverse_index_map[j] = out_j s.t. indices[j, :] == output_indices[out_j, :]
- * 
- * @param <T> data type for {@code outputValues()} output
- */
-@Operator(group = "sparse")
-public final class SparseFillEmptyRows<T extends TType> extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new SparseFillEmptyRows operation.
-   * 
-   * @param scope current scope
-   * @param indices 2-D. the indices of the sparse tensor.
-   * @param values 1-D. the values of the sparse tensor.
-   * @param denseShape 1-D. the shape of the sparse tensor.
-   * @param defaultValue 0-D. default value to insert into location `[row, 0, ..., 0]`
-   *   for rows missing from the input sparse tensor.
-   * output indices: 2-D. the indices of the filled sparse tensor.
-   * @return a new instance of SparseFillEmptyRows
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> SparseFillEmptyRows<T> create(Scope scope, Operand<TInt64> indices, Operand<T> values, Operand<TInt64> denseShape, Operand<T> defaultValue) {
-    OperationBuilder opBuilder = scope.env().opBuilder("SparseFillEmptyRows", scope.makeOpName("SparseFillEmptyRows"));
-    opBuilder.addInput(indices.asOutput(scope));
-    opBuilder.addInput(values.asOutput(scope));
-    opBuilder.addInput(denseShape.asOutput(scope));
-    opBuilder.addInput(defaultValue.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new SparseFillEmptyRows<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<TInt64> outputIndices() {
-    return outputIndices;
-  }
-  
-  /**
-   * 1-D. the values of the filled sparse tensor.
-   */
-  public Output<T> outputValues() {
-    return outputValues;
-  }
-  
-  /**
-   * 1-D. whether the dense row was missing in the
-   * input sparse tensor.
-   */
-  public Output<TBool> emptyRowIndicator() {
-    return emptyRowIndicator;
-  }
-  
-  /**
-   * 1-D. a map from the input indices to the output indices.
-   */
-  public Output<TInt64> reverseIndexMap() {
-    return reverseIndexMap;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "SparseFillEmptyRows";
-  
-  private Output<TInt64> outputIndices;
-  private Output<T> outputValues;
-  private Output<TBool> emptyRowIndicator;
-  private Output<TInt64> reverseIndexMap;
-  
-  private SparseFillEmptyRows(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    outputIndices = operation.output(outputIdx++);
-    outputValues = operation.output(outputIdx++);
-    emptyRowIndicator = operation.output(outputIdx++);
-    reverseIndexMap = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseFillEmptyRowsGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseFillEmptyRowsGrad.java
deleted file mode 100644
index bf1123de6a2..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseFillEmptyRowsGrad.java
+++ /dev/null
@@ -1,91 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.sparse;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
-
-/**
- * The gradient of SparseFillEmptyRows.
- * <p>
- * Takes vectors reverse_index_map, shaped `[N]`, and grad_values,
- * shaped `[N_full]`, where `N_full >= N` and copies data into either
- * `d_values` or `d_default_value`.  Here `d_values` is shaped `[N]` and
- * `d_default_value` is a scalar.
- * <p>
- *   d_values[j] = grad_values[reverse_index_map[j]]
- *   d_default_value = sum_{k : 0 .. N_full - 1} (
- *      grad_values[k] * 1{k not in reverse_index_map})
- * 
- * @param <T> data type for {@code dValues()} output
- */
-@Operator(group = "sparse")
-public final class SparseFillEmptyRowsGrad<T extends TType> extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new SparseFillEmptyRowsGrad operation.
-   * 
-   * @param scope current scope
-   * @param reverseIndexMap 1-D.  The reverse index map from SparseFillEmptyRows.
-   * @param gradValues 1-D.  The gradients from backprop.
-   * @return a new instance of SparseFillEmptyRowsGrad
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> SparseFillEmptyRowsGrad<T> create(Scope scope, Operand<TInt64> reverseIndexMap, Operand<T> gradValues) {
-    OperationBuilder opBuilder = scope.env().opBuilder("SparseFillEmptyRowsGrad", scope.makeOpName("SparseFillEmptyRowsGrad"));
-    opBuilder.addInput(reverseIndexMap.asOutput(scope));
-    opBuilder.addInput(gradValues.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new SparseFillEmptyRowsGrad<T>(opBuilder.build());
-  }
-  
-  /**
-   * 1-D.  The backprop into values.
-   */
-  public Output<T> dValues() {
-    return dValues;
-  }
-  
-  /**
-   * 0-D.  The backprop into default_value.
-   */
-  public Output<T> dDefaultValue() {
-    return dDefaultValue;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "SparseFillEmptyRowsGrad";
-  
-  private Output<T> dValues;
-  private Output<T> dDefaultValue;
-  
-  private SparseFillEmptyRowsGrad(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    dValues = operation.output(outputIdx++);
-    dDefaultValue = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseMatMul.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseMatMul.java
deleted file mode 100644
index 45ba5453179..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseMatMul.java
+++ /dev/null
@@ -1,176 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.sparse;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Multiply matrix "a" by matrix "b".
- * <p>
- * The inputs must be two-dimensional matrices and the inner dimension of "a" must
- * match the outer dimension of "b". Both "a" and "b" must be `Tensor`s not
- * `SparseTensor`s.  This op is optimized for the case where at least one of "a" or
- * "b" is sparse, in the sense that they have a large proportion of zero values.
- * The breakeven for using this versus a dense matrix multiply on one platform was
- * 30% zero values in the sparse matrix.
- * <p>
- * The gradient computation of this operation will only take advantage of sparsity
- * in the input gradient when that gradient comes from a Relu.
- */
-@Operator(group = "sparse")
-public final class SparseMatMul extends RawOp implements Operand<TFloat32> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.sparse.SparseMatMul}
-   */
-  public static class Options {
-    
-    /**
-     * @param transposeA 
-     */
-    public Options transposeA(Boolean transposeA) {
-      this.transposeA = transposeA;
-      return this;
-    }
-    
-    /**
-     * @param transposeB 
-     */
-    public Options transposeB(Boolean transposeB) {
-      this.transposeB = transposeB;
-      return this;
-    }
-    
-    /**
-     * @param aIsSparse 
-     */
-    public Options aIsSparse(Boolean aIsSparse) {
-      this.aIsSparse = aIsSparse;
-      return this;
-    }
-    
-    /**
-     * @param bIsSparse 
-     */
-    public Options bIsSparse(Boolean bIsSparse) {
-      this.bIsSparse = bIsSparse;
-      return this;
-    }
-    
-    private Boolean transposeA;
-    private Boolean transposeB;
-    private Boolean aIsSparse;
-    private Boolean bIsSparse;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new SparseMatMul operation.
-   * 
-   * @param scope current scope
-   * @param a 
-   * @param b 
-   * @param options carries optional attributes values
-   * @return a new instance of SparseMatMul
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber, U extends TNumber> SparseMatMul create(Scope scope, Operand<T> a, Operand<U> b, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("SparseMatMul", scope.makeOpName("SparseMatMul"));
-    opBuilder.addInput(a.asOutput(scope));
-    opBuilder.addInput(b.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.transposeA != null) {
-          opBuilder.setAttr("transpose_a", opts.transposeA);
-        }
-        if (opts.transposeB != null) {
-          opBuilder.setAttr("transpose_b", opts.transposeB);
-        }
-        if (opts.aIsSparse != null) {
-          opBuilder.setAttr("a_is_sparse", opts.aIsSparse);
-        }
-        if (opts.bIsSparse != null) {
-          opBuilder.setAttr("b_is_sparse", opts.bIsSparse);
-        }
-      }
-    }
-    return new SparseMatMul(opBuilder.build());
-  }
-  
-  /**
-   * @param transposeA 
-   */
-  public static Options transposeA(Boolean transposeA) {
-    return new Options().transposeA(transposeA);
-  }
-  
-  /**
-   * @param transposeB 
-   */
-  public static Options transposeB(Boolean transposeB) {
-    return new Options().transposeB(transposeB);
-  }
-  
-  /**
-   * @param aIsSparse 
-   */
-  public static Options aIsSparse(Boolean aIsSparse) {
-    return new Options().aIsSparse(aIsSparse);
-  }
-  
-  /**
-   * @param bIsSparse 
-   */
-  public static Options bIsSparse(Boolean bIsSparse) {
-    return new Options().bIsSparse(bIsSparse);
-  }
-  
-  /**
-   */
-  public Output<TFloat32> product() {
-    return product;
-  }
-  
-  @Override
-  public Output<TFloat32> asOutput(Scope scope) {
-    return product;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "SparseMatMul";
-  
-  private Output<TFloat32> product;
-  
-  private SparseMatMul(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    product = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseReduceMax.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseReduceMax.java
deleted file mode 100644
index 1f4803c2d1a..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseReduceMax.java
+++ /dev/null
@@ -1,131 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.sparse;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Computes the max of elements across dimensions of a SparseTensor.
- * <p>
- * This Op takes a SparseTensor and is the sparse counterpart to
- * `tf.reduce_max()`.  In particular, this Op also returns a dense `Tensor`
- * instead of a sparse one.
- * <p>
- * Reduces `sp_input` along the dimensions given in `reduction_axes`.  Unless
- * `keep_dims` is true, the rank of the tensor is reduced by 1 for each entry in
- * `reduction_axes`. If `keep_dims` is true, the reduced dimensions are retained
- * with length 1.
- * <p>
- * If `reduction_axes` has no entries, all dimensions are reduced, and a tensor
- * with a single element is returned.  Additionally, the axes can be negative,
- * which are interpreted according to the indexing rules in Python.
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator(group = "sparse")
-public final class SparseReduceMax<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.sparse.SparseReduceMax}
-   */
-  public static class Options {
-    
-    /**
-     * @param keepDims If true, retain reduced dimensions with length 1.
-     */
-    public Options keepDims(Boolean keepDims) {
-      this.keepDims = keepDims;
-      return this;
-    }
-    
-    private Boolean keepDims;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new SparseReduceMax operation.
-   * 
-   * @param scope current scope
-   * @param inputIndices 2-D.  `N x R` matrix with the indices of non-empty values in a
-   * SparseTensor, possibly not in canonical ordering.
-   * @param inputValues 1-D.  `N` non-empty values corresponding to `input_indices`.
-   * @param inputShape 1-D.  Shape of the input SparseTensor.
-   * @param reductionAxes 1-D.  Length-`K` vector containing the reduction axes.
-   * @param options carries optional attributes values
-   * @return a new instance of SparseReduceMax
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> SparseReduceMax<T> create(Scope scope, Operand<TInt64> inputIndices, Operand<T> inputValues, Operand<TInt64> inputShape, Operand<TInt32> reductionAxes, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("SparseReduceMax", scope.makeOpName("SparseReduceMax"));
-    opBuilder.addInput(inputIndices.asOutput(scope));
-    opBuilder.addInput(inputValues.asOutput(scope));
-    opBuilder.addInput(inputShape.asOutput(scope));
-    opBuilder.addInput(reductionAxes.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.keepDims != null) {
-          opBuilder.setAttr("keep_dims", opts.keepDims);
-        }
-      }
-    }
-    return new SparseReduceMax<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param keepDims If true, retain reduced dimensions with length 1.
-   */
-  public static Options keepDims(Boolean keepDims) {
-    return new Options().keepDims(keepDims);
-  }
-  
-  /**
-   * `R-K`-D.  The reduced Tensor.
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "SparseReduceMax";
-  
-  private Output<T> output;
-  
-  private SparseReduceMax(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseReduceMaxSparse.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseReduceMaxSparse.java
deleted file mode 100644
index 448d70903b6..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseReduceMaxSparse.java
+++ /dev/null
@@ -1,141 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.sparse;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Computes the max of elements across dimensions of a SparseTensor.
- * <p>
- * This Op takes a SparseTensor and is the sparse counterpart to
- * `tf.reduce_max()`.  In contrast to SparseReduceMax, this Op returns a
- * SparseTensor.
- * <p>
- * Reduces `sp_input` along the dimensions given in `reduction_axes`.  Unless
- * `keep_dims` is true, the rank of the tensor is reduced by 1 for each entry in
- * `reduction_axes`. If `keep_dims` is true, the reduced dimensions are retained
- * with length 1.
- * <p>
- * If `reduction_axes` has no entries, all dimensions are reduced, and a tensor
- * with a single element is returned.  Additionally, the axes can be negative,
- * which are interpreted according to the indexing rules in Python.
- * 
- * @param <T> data type for {@code outputValues()} output
- */
-@Operator(group = "sparse")
-public final class SparseReduceMaxSparse<T extends TNumber> extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.sparse.SparseReduceMaxSparse}
-   */
-  public static class Options {
-    
-    /**
-     * @param keepDims If true, retain reduced dimensions with length 1.
-     */
-    public Options keepDims(Boolean keepDims) {
-      this.keepDims = keepDims;
-      return this;
-    }
-    
-    private Boolean keepDims;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new SparseReduceMaxSparse operation.
-   * 
-   * @param scope current scope
-   * @param inputIndices 2-D.  `N x R` matrix with the indices of non-empty values in a
-   * SparseTensor, possibly not in canonical ordering.
-   * @param inputValues 1-D.  `N` non-empty values corresponding to `input_indices`.
-   * @param inputShape 1-D.  Shape of the input SparseTensor.
-   * @param reductionAxes 1-D.  Length-`K` vector containing the reduction axes.
-   * @param options carries optional attributes values
-   * @return a new instance of SparseReduceMaxSparse
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> SparseReduceMaxSparse<T> create(Scope scope, Operand<TInt64> inputIndices, Operand<T> inputValues, Operand<TInt64> inputShape, Operand<TInt32> reductionAxes, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("SparseReduceMaxSparse", scope.makeOpName("SparseReduceMaxSparse"));
-    opBuilder.addInput(inputIndices.asOutput(scope));
-    opBuilder.addInput(inputValues.asOutput(scope));
-    opBuilder.addInput(inputShape.asOutput(scope));
-    opBuilder.addInput(reductionAxes.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.keepDims != null) {
-          opBuilder.setAttr("keep_dims", opts.keepDims);
-        }
-      }
-    }
-    return new SparseReduceMaxSparse<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param keepDims If true, retain reduced dimensions with length 1.
-   */
-  public static Options keepDims(Boolean keepDims) {
-    return new Options().keepDims(keepDims);
-  }
-  
-  /**
-   */
-  public Output<TInt64> outputIndices() {
-    return outputIndices;
-  }
-  
-  /**
-   */
-  public Output<T> outputValues() {
-    return outputValues;
-  }
-  
-  /**
-   */
-  public Output<TInt64> outputShape() {
-    return outputShape;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "SparseReduceMaxSparse";
-  
-  private Output<TInt64> outputIndices;
-  private Output<T> outputValues;
-  private Output<TInt64> outputShape;
-  
-  private SparseReduceMaxSparse(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    outputIndices = operation.output(outputIdx++);
-    outputValues = operation.output(outputIdx++);
-    outputShape = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseReduceSum.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseReduceSum.java
deleted file mode 100644
index 1415dbdd867..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseReduceSum.java
+++ /dev/null
@@ -1,131 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.sparse;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
-
-/**
- * Computes the sum of elements across dimensions of a SparseTensor.
- * <p>
- * This Op takes a SparseTensor and is the sparse counterpart to
- * `tf.reduce_sum()`.  In particular, this Op also returns a dense `Tensor`
- * instead of a sparse one.
- * <p>
- * Reduces `sp_input` along the dimensions given in `reduction_axes`.  Unless
- * `keep_dims` is true, the rank of the tensor is reduced by 1 for each entry in
- * `reduction_axes`. If `keep_dims` is true, the reduced dimensions are retained
- * with length 1.
- * <p>
- * If `reduction_axes` has no entries, all dimensions are reduced, and a tensor
- * with a single element is returned.  Additionally, the axes can be negative,
- * which are interpreted according to the indexing rules in Python.
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator(group = "sparse")
-public final class SparseReduceSum<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.sparse.SparseReduceSum}
-   */
-  public static class Options {
-    
-    /**
-     * @param keepDims If true, retain reduced dimensions with length 1.
-     */
-    public Options keepDims(Boolean keepDims) {
-      this.keepDims = keepDims;
-      return this;
-    }
-    
-    private Boolean keepDims;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new SparseReduceSum operation.
-   * 
-   * @param scope current scope
-   * @param inputIndices 2-D.  `N x R` matrix with the indices of non-empty values in a
-   * SparseTensor, possibly not in canonical ordering.
-   * @param inputValues 1-D.  `N` non-empty values corresponding to `input_indices`.
-   * @param inputShape 1-D.  Shape of the input SparseTensor.
-   * @param reductionAxes 1-D.  Length-`K` vector containing the reduction axes.
-   * @param options carries optional attributes values
-   * @return a new instance of SparseReduceSum
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> SparseReduceSum<T> create(Scope scope, Operand<TInt64> inputIndices, Operand<T> inputValues, Operand<TInt64> inputShape, Operand<TInt32> reductionAxes, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("SparseReduceSum", scope.makeOpName("SparseReduceSum"));
-    opBuilder.addInput(inputIndices.asOutput(scope));
-    opBuilder.addInput(inputValues.asOutput(scope));
-    opBuilder.addInput(inputShape.asOutput(scope));
-    opBuilder.addInput(reductionAxes.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.keepDims != null) {
-          opBuilder.setAttr("keep_dims", opts.keepDims);
-        }
-      }
-    }
-    return new SparseReduceSum<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param keepDims If true, retain reduced dimensions with length 1.
-   */
-  public static Options keepDims(Boolean keepDims) {
-    return new Options().keepDims(keepDims);
-  }
-  
-  /**
-   * `R-K`-D.  The reduced Tensor.
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "SparseReduceSum";
-  
-  private Output<T> output;
-  
-  private SparseReduceSum(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseReduceSumSparse.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseReduceSumSparse.java
deleted file mode 100644
index b5ae71e226f..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseReduceSumSparse.java
+++ /dev/null
@@ -1,141 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.sparse;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
-
-/**
- * Computes the sum of elements across dimensions of a SparseTensor.
- * <p>
- * This Op takes a SparseTensor and is the sparse counterpart to
- * `tf.reduce_sum()`.  In contrast to SparseReduceSum, this Op returns a
- * SparseTensor.
- * <p>
- * Reduces `sp_input` along the dimensions given in `reduction_axes`.  Unless
- * `keep_dims` is true, the rank of the tensor is reduced by 1 for each entry in
- * `reduction_axes`. If `keep_dims` is true, the reduced dimensions are retained
- * with length 1.
- * <p>
- * If `reduction_axes` has no entries, all dimensions are reduced, and a tensor
- * with a single element is returned.  Additionally, the axes can be negative,
- * which are interpreted according to the indexing rules in Python.
- * 
- * @param <T> data type for {@code outputValues()} output
- */
-@Operator(group = "sparse")
-public final class SparseReduceSumSparse<T extends TType> extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.sparse.SparseReduceSumSparse}
-   */
-  public static class Options {
-    
-    /**
-     * @param keepDims If true, retain reduced dimensions with length 1.
-     */
-    public Options keepDims(Boolean keepDims) {
-      this.keepDims = keepDims;
-      return this;
-    }
-    
-    private Boolean keepDims;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new SparseReduceSumSparse operation.
-   * 
-   * @param scope current scope
-   * @param inputIndices 2-D.  `N x R` matrix with the indices of non-empty values in a
-   * SparseTensor, possibly not in canonical ordering.
-   * @param inputValues 1-D.  `N` non-empty values corresponding to `input_indices`.
-   * @param inputShape 1-D.  Shape of the input SparseTensor.
-   * @param reductionAxes 1-D.  Length-`K` vector containing the reduction axes.
-   * @param options carries optional attributes values
-   * @return a new instance of SparseReduceSumSparse
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> SparseReduceSumSparse<T> create(Scope scope, Operand<TInt64> inputIndices, Operand<T> inputValues, Operand<TInt64> inputShape, Operand<TInt32> reductionAxes, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("SparseReduceSumSparse", scope.makeOpName("SparseReduceSumSparse"));
-    opBuilder.addInput(inputIndices.asOutput(scope));
-    opBuilder.addInput(inputValues.asOutput(scope));
-    opBuilder.addInput(inputShape.asOutput(scope));
-    opBuilder.addInput(reductionAxes.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.keepDims != null) {
-          opBuilder.setAttr("keep_dims", opts.keepDims);
-        }
-      }
-    }
-    return new SparseReduceSumSparse<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param keepDims If true, retain reduced dimensions with length 1.
-   */
-  public static Options keepDims(Boolean keepDims) {
-    return new Options().keepDims(keepDims);
-  }
-  
-  /**
-   */
-  public Output<TInt64> outputIndices() {
-    return outputIndices;
-  }
-  
-  /**
-   */
-  public Output<T> outputValues() {
-    return outputValues;
-  }
-  
-  /**
-   */
-  public Output<TInt64> outputShape() {
-    return outputShape;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "SparseReduceSumSparse";
-  
-  private Output<TInt64> outputIndices;
-  private Output<T> outputValues;
-  private Output<TInt64> outputShape;
-  
-  private SparseReduceSumSparse(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    outputIndices = operation.output(outputIdx++);
-    outputValues = operation.output(outputIdx++);
-    outputShape = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseReorder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseReorder.java
deleted file mode 100644
index 75ed7a3e04a..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseReorder.java
+++ /dev/null
@@ -1,95 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.sparse;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
-
-/**
- * Reorders a SparseTensor into the canonical, row-major ordering.
- * <p>
- * Note that by convention, all sparse ops preserve the canonical ordering along
- * increasing dimension number. The only time ordering can be violated is during
- * manual manipulation of the indices and values vectors to add entries.
- * <p>
- * Reordering does not affect the shape of the SparseTensor.
- * <p>
- * If the tensor has rank `R` and `N` non-empty values, `input_indices` has
- * shape `[N, R]`, input_values has length `N`, and input_shape has length `R`.
- * 
- * @param <T> data type for {@code outputValues()} output
- */
-@Operator(group = "sparse")
-public final class SparseReorder<T extends TType> extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new SparseReorder operation.
-   * 
-   * @param scope current scope
-   * @param inputIndices 2-D.  `N x R` matrix with the indices of non-empty values in a
-   * SparseTensor, possibly not in canonical ordering.
-   * @param inputValues 1-D.  `N` non-empty values corresponding to `input_indices`.
-   * @param inputShape 1-D.  Shape of the input SparseTensor.
-   * @return a new instance of SparseReorder
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> SparseReorder<T> create(Scope scope, Operand<TInt64> inputIndices, Operand<T> inputValues, Operand<TInt64> inputShape) {
-    OperationBuilder opBuilder = scope.env().opBuilder("SparseReorder", scope.makeOpName("SparseReorder"));
-    opBuilder.addInput(inputIndices.asOutput(scope));
-    opBuilder.addInput(inputValues.asOutput(scope));
-    opBuilder.addInput(inputShape.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new SparseReorder<T>(opBuilder.build());
-  }
-  
-  /**
-   * 2-D.  `N x R` matrix with the same indices as input_indices, but
-   * in canonical row-major ordering.
-   */
-  public Output<TInt64> outputIndices() {
-    return outputIndices;
-  }
-  
-  /**
-   * 1-D.  `N` non-empty values corresponding to `output_indices`.
-   */
-  public Output<T> outputValues() {
-    return outputValues;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "SparseReorder";
-  
-  private Output<TInt64> outputIndices;
-  private Output<T> outputValues;
-  
-  private SparseReorder(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    outputIndices = operation.output(outputIdx++);
-    outputValues = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseReshape.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseReshape.java
deleted file mode 100644
index e9a6673b759..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseReshape.java
+++ /dev/null
@@ -1,101 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.sparse;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt64;
-
-/**
- * Reshapes a SparseTensor to represent values in a new dense shape.
- * <p>
- * This operation has the same semantics as reshape on the represented dense
- * tensor.  The `input_indices` are recomputed based on the requested `new_shape`.
- * <p>
- * If one component of `new_shape` is the special value -1, the size of that
- * dimension is computed so that the total dense size remains constant.  At
- * most one component of `new_shape` can be -1.  The number of dense elements
- * implied by `new_shape` must be the same as the number of dense elements
- * originally implied by `input_shape`.
- * <p>
- * Reshaping does not affect the order of values in the SparseTensor.
- * <p>
- * If the input tensor has rank `R_in` and `N` non-empty values, and `new_shape`
- * has length `R_out`, then `input_indices` has shape `[N, R_in]`,
- * `input_shape` has length `R_in`, `output_indices` has shape `[N, R_out]`, and
- * `output_shape` has length `R_out`.
- */
-@Operator(group = "sparse")
-public final class SparseReshape extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new SparseReshape operation.
-   * 
-   * @param scope current scope
-   * @param inputIndices 2-D.  `N x R_in` matrix with the indices of non-empty values in a
-   * SparseTensor.
-   * @param inputShape 1-D.  `R_in` vector with the input SparseTensor's dense shape.
-   * @param newShape 1-D.  `R_out` vector with the requested new dense shape.
-   * @return a new instance of SparseReshape
-   */
-  @Endpoint(describeByClass = true)
-  public static SparseReshape create(Scope scope, Operand<TInt64> inputIndices, Operand<TInt64> inputShape, Operand<TInt64> newShape) {
-    OperationBuilder opBuilder = scope.env().opBuilder("SparseReshape", scope.makeOpName("SparseReshape"));
-    opBuilder.addInput(inputIndices.asOutput(scope));
-    opBuilder.addInput(inputShape.asOutput(scope));
-    opBuilder.addInput(newShape.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new SparseReshape(opBuilder.build());
-  }
-  
-  /**
-   * 2-D.  `N x R_out` matrix with the updated indices of non-empty
-   * values in the output SparseTensor.
-   */
-  public Output<TInt64> outputIndices() {
-    return outputIndices;
-  }
-  
-  /**
-   * 1-D.  `R_out` vector with the full dense shape of the output
-   * SparseTensor.  This is the same as `new_shape` but with any -1 dimensions
-   * filled in.
-   */
-  public Output<TInt64> outputShape() {
-    return outputShape;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "SparseReshape";
-  
-  private Output<TInt64> outputIndices;
-  private Output<TInt64> outputShape;
-  
-  private SparseReshape(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    outputIndices = operation.output(outputIdx++);
-    outputShape = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentMean.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentMean.java
deleted file mode 100644
index 9287fb2688a..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentMean.java
+++ /dev/null
@@ -1,85 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.sparse;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Computes the mean along sparse segments of a tensor.
- * <p>
- * See `tf.sparse.segment_sum` for usage examples.
- * <p>
- * Like `SegmentMean`, but `segment_ids` can have rank less than `data`'s first
- * dimension, selecting a subset of dimension 0, specified by `indices`.
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator(group = "sparse")
-public final class SparseSegmentMean<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new SparseSegmentMean operation.
-   * 
-   * @param scope current scope
-   * @param data 
-   * @param indices A 1-D tensor. Has same rank as `segment_ids`.
-   * @param segmentIds A 1-D tensor. Values should be sorted and can be repeated.
-   * @return a new instance of SparseSegmentMean
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber, U extends TNumber, V extends TNumber> SparseSegmentMean<T> create(Scope scope, Operand<T> data, Operand<U> indices, Operand<V> segmentIds) {
-    OperationBuilder opBuilder = scope.env().opBuilder("SparseSegmentMean", scope.makeOpName("SparseSegmentMean"));
-    opBuilder.addInput(data.asOutput(scope));
-    opBuilder.addInput(indices.asOutput(scope));
-    opBuilder.addInput(segmentIds.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new SparseSegmentMean<T>(opBuilder.build());
-  }
-  
-  /**
-   * Has same shape as data, except for dimension 0 which
-   * has size `k`, the number of segments.
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "SparseSegmentMean";
-  
-  private Output<T> output;
-  
-  private SparseSegmentMean(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentMeanGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentMeanGrad.java
deleted file mode 100644
index 4aa3800c0d4..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentMeanGrad.java
+++ /dev/null
@@ -1,84 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.sparse;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Computes gradients for SparseSegmentMean.
- * <p>
- * Returns tensor "output" with same shape as grad, except for dimension 0 whose
- * value is output_dim0.
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator(group = "sparse")
-public final class SparseSegmentMeanGrad<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new SparseSegmentMeanGrad operation.
-   * 
-   * @param scope current scope
-   * @param grad gradient propagated to the SparseSegmentMean op.
-   * @param indices indices passed to the corresponding SparseSegmentMean op.
-   * @param segmentIds segment_ids passed to the corresponding SparseSegmentMean op.
-   * @param outputDim0 dimension 0 of "data" passed to SparseSegmentMean op.
-   * @return a new instance of SparseSegmentMeanGrad
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber, U extends TNumber, V extends TNumber> SparseSegmentMeanGrad<T> create(Scope scope, Operand<T> grad, Operand<U> indices, Operand<V> segmentIds, Operand<TInt32> outputDim0) {
-    OperationBuilder opBuilder = scope.env().opBuilder("SparseSegmentMeanGrad", scope.makeOpName("SparseSegmentMeanGrad"));
-    opBuilder.addInput(grad.asOutput(scope));
-    opBuilder.addInput(indices.asOutput(scope));
-    opBuilder.addInput(segmentIds.asOutput(scope));
-    opBuilder.addInput(outputDim0.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new SparseSegmentMeanGrad<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "SparseSegmentMeanGrad";
-  
-  private Output<T> output;
-  
-  private SparseSegmentMeanGrad(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentMeanWithNumSegments.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentMeanWithNumSegments.java
deleted file mode 100644
index 14c069899dc..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentMeanWithNumSegments.java
+++ /dev/null
@@ -1,89 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.sparse;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Computes the mean along sparse segments of a tensor.
- * <p>
- * Like `SparseSegmentMean`, but allows missing ids in `segment_ids`. If an id is
- * missing, the `output` tensor at that position will be zeroed.
- * <p>
- * Read
- * [the section on segmentation](https://tensorflow.org/api_docs/python/tf/math#Segmentation)
- * for an explanation of segments.
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator(group = "sparse")
-public final class SparseSegmentMeanWithNumSegments<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new SparseSegmentMeanWithNumSegments operation.
-   * 
-   * @param scope current scope
-   * @param data 
-   * @param indices A 1-D tensor. Has same rank as `segment_ids`.
-   * @param segmentIds A 1-D tensor. Values should be sorted and can be repeated.
-   * @param numSegments Should equal the number of distinct segment IDs.
-   * @return a new instance of SparseSegmentMeanWithNumSegments
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber, U extends TNumber, V extends TNumber, W extends TNumber> SparseSegmentMeanWithNumSegments<T> create(Scope scope, Operand<T> data, Operand<U> indices, Operand<V> segmentIds, Operand<W> numSegments) {
-    OperationBuilder opBuilder = scope.env().opBuilder("SparseSegmentMeanWithNumSegments", scope.makeOpName("SparseSegmentMeanWithNumSegments"));
-    opBuilder.addInput(data.asOutput(scope));
-    opBuilder.addInput(indices.asOutput(scope));
-    opBuilder.addInput(segmentIds.asOutput(scope));
-    opBuilder.addInput(numSegments.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new SparseSegmentMeanWithNumSegments<T>(opBuilder.build());
-  }
-  
-  /**
-   * Has same shape as data, except for dimension 0 which has size
-   * `num_segments`.
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "SparseSegmentMeanWithNumSegments";
-  
-  private Output<T> output;
-  
-  private SparseSegmentMeanWithNumSegments(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentSqrtN.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentSqrtN.java
deleted file mode 100644
index 93851b0abf1..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentSqrtN.java
+++ /dev/null
@@ -1,85 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.sparse;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Computes the sum along sparse segments of a tensor divided by the sqrt of N.
- * <p>
- * N is the size of the segment being reduced.
- * <p>
- * See `tf.sparse.segment_sum` for usage examples.
- * 
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator(group = "sparse")
-public final class SparseSegmentSqrtN<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new SparseSegmentSqrtN operation.
-   * 
-   * @param scope current scope
-   * @param data 
-   * @param indices A 1-D tensor. Has same rank as `segment_ids`.
-   * @param segmentIds A 1-D tensor. Values should be sorted and can be repeated.
-   * @return a new instance of SparseSegmentSqrtN
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber, U extends TNumber, V extends TNumber> SparseSegmentSqrtN<T> create(Scope scope, Operand<T> data, Operand<U> indices, Operand<V> segmentIds) {
-    OperationBuilder opBuilder = scope.env().opBuilder("SparseSegmentSqrtN", scope.makeOpName("SparseSegmentSqrtN"));
-    opBuilder.addInput(data.asOutput(scope));
-    opBuilder.addInput(indices.asOutput(scope));
-    opBuilder.addInput(segmentIds.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new SparseSegmentSqrtN<T>(opBuilder.build());
-  }
-  
-  /**
-   * Has same shape as data, except for dimension 0 which
-   * has size `k`, the number of segments.
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "SparseSegmentSqrtN";
-  
-  private Output<T> output;
-  
-  private SparseSegmentSqrtN(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentSqrtNGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentSqrtNGrad.java
deleted file mode 100644
index 22a6610f717..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentSqrtNGrad.java
+++ /dev/null
@@ -1,84 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.sparse;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Computes gradients for SparseSegmentSqrtN.
- * <p>
- * Returns tensor "output" with same shape as grad, except for dimension 0 whose
- * value is output_dim0.
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator(group = "sparse")
-public final class SparseSegmentSqrtNGrad<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new SparseSegmentSqrtNGrad operation.
-   * 
-   * @param scope current scope
-   * @param grad gradient propagated to the SparseSegmentSqrtN op.
-   * @param indices indices passed to the corresponding SparseSegmentSqrtN op.
-   * @param segmentIds segment_ids passed to the corresponding SparseSegmentSqrtN op.
-   * @param outputDim0 dimension 0 of "data" passed to SparseSegmentSqrtN op.
-   * @return a new instance of SparseSegmentSqrtNGrad
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber, U extends TNumber, V extends TNumber> SparseSegmentSqrtNGrad<T> create(Scope scope, Operand<T> grad, Operand<U> indices, Operand<V> segmentIds, Operand<TInt32> outputDim0) {
-    OperationBuilder opBuilder = scope.env().opBuilder("SparseSegmentSqrtNGrad", scope.makeOpName("SparseSegmentSqrtNGrad"));
-    opBuilder.addInput(grad.asOutput(scope));
-    opBuilder.addInput(indices.asOutput(scope));
-    opBuilder.addInput(segmentIds.asOutput(scope));
-    opBuilder.addInput(outputDim0.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new SparseSegmentSqrtNGrad<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "SparseSegmentSqrtNGrad";
-  
-  private Output<T> output;
-  
-  private SparseSegmentSqrtNGrad(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentSqrtNWithNumSegments.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentSqrtNWithNumSegments.java
deleted file mode 100644
index 8b87a35353a..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentSqrtNWithNumSegments.java
+++ /dev/null
@@ -1,91 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.sparse;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Computes the sum along sparse segments of a tensor divided by the sqrt of N.
- * <p>
- * N is the size of the segment being reduced.
- * <p>
- * Like `SparseSegmentSqrtN`, but allows missing ids in `segment_ids`. If an id is
- * missing, the `output` tensor at that position will be zeroed.
- * <p>
- * Read
- * [the section on segmentation](https://tensorflow.org/api_docs/python/tf/math#Segmentation)
- * for an explanation of segments.
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator(group = "sparse")
-public final class SparseSegmentSqrtNWithNumSegments<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new SparseSegmentSqrtNWithNumSegments operation.
-   * 
-   * @param scope current scope
-   * @param data 
-   * @param indices A 1-D tensor. Has same rank as `segment_ids`.
-   * @param segmentIds A 1-D tensor. Values should be sorted and can be repeated.
-   * @param numSegments Should equal the number of distinct segment IDs.
-   * @return a new instance of SparseSegmentSqrtNWithNumSegments
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber, U extends TNumber, V extends TNumber, W extends TNumber> SparseSegmentSqrtNWithNumSegments<T> create(Scope scope, Operand<T> data, Operand<U> indices, Operand<V> segmentIds, Operand<W> numSegments) {
-    OperationBuilder opBuilder = scope.env().opBuilder("SparseSegmentSqrtNWithNumSegments", scope.makeOpName("SparseSegmentSqrtNWithNumSegments"));
-    opBuilder.addInput(data.asOutput(scope));
-    opBuilder.addInput(indices.asOutput(scope));
-    opBuilder.addInput(segmentIds.asOutput(scope));
-    opBuilder.addInput(numSegments.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new SparseSegmentSqrtNWithNumSegments<T>(opBuilder.build());
-  }
-  
-  /**
-   * Has same shape as data, except for dimension 0 which
-   * has size `k`, the number of segments.
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "SparseSegmentSqrtNWithNumSegments";
-  
-  private Output<T> output;
-  
-  private SparseSegmentSqrtNWithNumSegments(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentSum.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentSum.java
deleted file mode 100644
index b393d51e753..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentSum.java
+++ /dev/null
@@ -1,110 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.sparse;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Computes the sum along sparse segments of a tensor.
- * <p>
- * Read
- * [the section on segmentation](https://tensorflow.org/api_docs/python/tf/math#Segmentation)
- * for an explanation of segments.
- * <p>
- * Like `SegmentSum`, but `segment_ids` can have rank less than `data`'s first
- * dimension, selecting a subset of dimension 0, specified by `indices`.
- * <p>
- * For example:
- * <pre>{@code
- * c = tf.constant([[1,2,3,4], [-1,-2,-3,-4], [5,6,7,8]])
- * 
- * # Select two rows, one segment.
- * tf.sparse_segment_sum(c, tf.constant([0, 1]), tf.constant([0, 0]))
- * # => [[0 0 0 0]]
- * 
- * # Select two rows, two segment.
- * tf.sparse_segment_sum(c, tf.constant([0, 1]), tf.constant([0, 1]))
- * # => [[ 1  2  3  4]
- * #     [-1 -2 -3 -4]]
- * 
- * # Select all rows, two segments.
- * tf.sparse_segment_sum(c, tf.constant([0, 1, 2]), tf.constant([0, 0, 1]))
- * # => [[0 0 0 0]
- * #     [5 6 7 8]]
- * 
- * # Which is equivalent to:
- * tf.segment_sum(c, tf.constant([0, 0, 1]))
- * }</pre>
- * 
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator(group = "sparse")
-public final class SparseSegmentSum<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new SparseSegmentSum operation.
-   * 
-   * @param scope current scope
-   * @param data 
-   * @param indices A 1-D tensor. Has same rank as `segment_ids`.
-   * @param segmentIds A 1-D tensor. Values should be sorted and can be repeated.
-   * @return a new instance of SparseSegmentSum
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber, U extends TNumber, V extends TNumber> SparseSegmentSum<T> create(Scope scope, Operand<T> data, Operand<U> indices, Operand<V> segmentIds) {
-    OperationBuilder opBuilder = scope.env().opBuilder("SparseSegmentSum", scope.makeOpName("SparseSegmentSum"));
-    opBuilder.addInput(data.asOutput(scope));
-    opBuilder.addInput(indices.asOutput(scope));
-    opBuilder.addInput(segmentIds.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new SparseSegmentSum<T>(opBuilder.build());
-  }
-  
-  /**
-   * Has same shape as data, except for dimension 0 which
-   * has size `k`, the number of segments.
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "SparseSegmentSum";
-  
-  private Output<T> output;
-  
-  private SparseSegmentSum(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentSumWithNumSegments.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentSumWithNumSegments.java
deleted file mode 100644
index 1cf4df85132..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentSumWithNumSegments.java
+++ /dev/null
@@ -1,110 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.sparse;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Computes the sum along sparse segments of a tensor.
- * <p>
- * Like `SparseSegmentSum`, but allows missing ids in `segment_ids`. If an id is
- * missing, the `output` tensor at that position will be zeroed.
- * <p>
- * Read
- * [the section on segmentation](https://tensorflow.org/api_docs/python/tf/sparse#Segmentation)
- * for an explanation of segments.
- * <p>
- * For example:
- * <pre>{@code
- * c = tf.constant([[1,2,3,4], [-1,-2,-3,-4], [5,6,7,8]])
- * 
- * tf.sparse_segment_sum_with_num_segments(
- *     c, tf.constant([0, 1]), tf.constant([0, 0]), num_segments=3)
- * # => [[0 0 0 0]
- * #     [0 0 0 0]
- * #     [0 0 0 0]]
- * 
- * tf.sparse_segment_sum_with_num_segments(c,
- *                                         tf.constant([0, 1]),
- *                                         tf.constant([0, 2],
- *                                         num_segments=4))
- * # => [[ 1  2  3  4]
- * #     [ 0  0  0  0]
- * #     [-1 -2 -3 -4]
- * #     [ 0  0  0  0]]
- * }</pre>
- * 
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator(group = "sparse")
-public final class SparseSegmentSumWithNumSegments<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new SparseSegmentSumWithNumSegments operation.
-   * 
-   * @param scope current scope
-   * @param data 
-   * @param indices A 1-D tensor. Has same rank as `segment_ids`.
-   * @param segmentIds A 1-D tensor. Values should be sorted and can be repeated.
-   * @param numSegments Should equal the number of distinct segment IDs.
-   * @return a new instance of SparseSegmentSumWithNumSegments
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber, U extends TNumber, V extends TNumber, W extends TNumber> SparseSegmentSumWithNumSegments<T> create(Scope scope, Operand<T> data, Operand<U> indices, Operand<V> segmentIds, Operand<W> numSegments) {
-    OperationBuilder opBuilder = scope.env().opBuilder("SparseSegmentSumWithNumSegments", scope.makeOpName("SparseSegmentSumWithNumSegments"));
-    opBuilder.addInput(data.asOutput(scope));
-    opBuilder.addInput(indices.asOutput(scope));
-    opBuilder.addInput(segmentIds.asOutput(scope));
-    opBuilder.addInput(numSegments.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new SparseSegmentSumWithNumSegments<T>(opBuilder.build());
-  }
-  
-  /**
-   * Has same shape as data, except for dimension 0 which
-   * has size `num_segments`.
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "SparseSegmentSumWithNumSegments";
-  
-  private Output<T> output;
-  
-  private SparseSegmentSumWithNumSegments(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSlice.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSlice.java
deleted file mode 100644
index 3cebea9baa6..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSlice.java
+++ /dev/null
@@ -1,116 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.sparse;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
-
-/**
- * Slice a `SparseTensor` based on the `start` and `size`.
- * <p>
- * For example, if the input is
- * <p>
- *     input_tensor = shape = [2, 7]
- *     [    a   d e  ]
- *     [b c          ]
- * <p>
- * Graphically the output tensors are:
- * <p>
- *     sparse_slice([0, 0], [2, 4]) = shape = [2, 4]
- *     [    a  ]
- *     [b c    ]
- * <p>
- *     sparse_slice([0, 4], [2, 3]) = shape = [2, 3]
- *     [ d e  ]
- *     [      ]
- * 
- * @param <T> data type for {@code outputValues()} output
- */
-@Operator(group = "sparse")
-public final class SparseSlice<T extends TType> extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new SparseSlice operation.
-   * 
-   * @param scope current scope
-   * @param indices 2-D tensor represents the indices of the sparse tensor.
-   * @param values 1-D tensor represents the values of the sparse tensor.
-   * @param shape 1-D. tensor represents the shape of the sparse tensor.
-   * @param start 1-D. tensor represents the start of the slice.
-   * @param size 1-D. tensor represents the size of the slice.
-   * output indices: A list of 1-D tensors represents the indices of the output
-   * sparse tensors.
-   * @return a new instance of SparseSlice
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> SparseSlice<T> create(Scope scope, Operand<TInt64> indices, Operand<T> values, Operand<TInt64> shape, Operand<TInt64> start, Operand<TInt64> size) {
-    OperationBuilder opBuilder = scope.env().opBuilder("SparseSlice", scope.makeOpName("SparseSlice"));
-    opBuilder.addInput(indices.asOutput(scope));
-    opBuilder.addInput(values.asOutput(scope));
-    opBuilder.addInput(shape.asOutput(scope));
-    opBuilder.addInput(start.asOutput(scope));
-    opBuilder.addInput(size.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new SparseSlice<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<TInt64> outputIndices() {
-    return outputIndices;
-  }
-  
-  /**
-   * A list of 1-D tensors represents the values of the output sparse
-   * tensors.
-   */
-  public Output<T> outputValues() {
-    return outputValues;
-  }
-  
-  /**
-   * A list of 1-D tensors represents the shape of the output sparse
-   * tensors.
-   */
-  public Output<TInt64> outputShape() {
-    return outputShape;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "SparseSlice";
-  
-  private Output<TInt64> outputIndices;
-  private Output<T> outputValues;
-  private Output<TInt64> outputShape;
-  
-  private SparseSlice(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    outputIndices = operation.output(outputIdx++);
-    outputValues = operation.output(outputIdx++);
-    outputShape = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSliceGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSliceGrad.java
deleted file mode 100644
index 8b951621d2b..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSliceGrad.java
+++ /dev/null
@@ -1,87 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.sparse;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
-
-/**
- * The gradient operator for the SparseSlice op.
- * <p>
- * This op takes in the upstream gradient w.r.t. non-empty values of
- * the sliced `SparseTensor`, and outputs the gradients w.r.t.
- * the non-empty values of input `SparseTensor`.
- * 
- * @param <T> data type for {@code valGrad()} output
- */
-@Operator(group = "sparse")
-public final class SparseSliceGrad<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new SparseSliceGrad operation.
-   * 
-   * @param scope current scope
-   * @param backpropValGrad 1-D. The gradient with respect to
-   * the non-empty values of the sliced `SparseTensor`.
-   * @param inputIndices 2-D.  The `indices` of the input `SparseTensor`.
-   * @param inputStart 1-D. tensor represents the start of the slice.
-   * @param outputIndices 2-D.  The `indices` of the sliced `SparseTensor`.
-   * @return a new instance of SparseSliceGrad
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> SparseSliceGrad<T> create(Scope scope, Operand<T> backpropValGrad, Operand<TInt64> inputIndices, Operand<TInt64> inputStart, Operand<TInt64> outputIndices) {
-    OperationBuilder opBuilder = scope.env().opBuilder("SparseSliceGrad", scope.makeOpName("SparseSliceGrad"));
-    opBuilder.addInput(backpropValGrad.asOutput(scope));
-    opBuilder.addInput(inputIndices.asOutput(scope));
-    opBuilder.addInput(inputStart.asOutput(scope));
-    opBuilder.addInput(outputIndices.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new SparseSliceGrad<T>(opBuilder.build());
-  }
-  
-  /**
-   * 1-D. The gradient with respect to the non-empty values of input `SparseTensor`.
-   */
-  public Output<T> valGrad() {
-    return valGrad;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return valGrad;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "SparseSliceGrad";
-  
-  private Output<T> valGrad;
-  
-  private SparseSliceGrad(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    valGrad = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSoftmax.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSoftmax.java
deleted file mode 100644
index e9a60b85f3d..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSoftmax.java
+++ /dev/null
@@ -1,97 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.sparse;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Applies softmax to a batched N-D `SparseTensor`.
- * <p>
- * The inputs represent an N-D SparseTensor  with logical shape `[..., B, C]`
- * (where `N >= 2`), and with indices sorted in the canonical lexicographic order.
- * <p>
- * This op is equivalent to applying the normal `tf.nn.softmax()` to each innermost
- * logical submatrix with shape `[B, C]`, but with the catch that <i>the implicitly
- * zero elements do not participate</i>.  Specifically, the algorithm is equivalent
- * to the following:
- * <p>
- *   (1) Applies `tf.nn.softmax()` to a densified view of each innermost submatrix
- *       with shape `[B, C]`, along the size-C dimension;
- *   (2) Masks out the original implicitly-zero locations;
- *   (3) Renormalizes the remaining elements.
- * <p>
- * Hence, the `SparseTensor` result has exactly the same non-zero indices and
- * shape.
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator(group = "sparse")
-public final class SparseSoftmax<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new SparseSoftmax operation.
-   * 
-   * @param scope current scope
-   * @param spIndices 2-D.  `NNZ x R` matrix with the indices of non-empty values in a
-   * SparseTensor, in canonical ordering.
-   * @param spValues 1-D.  `NNZ` non-empty values corresponding to `sp_indices`.
-   * @param spShape 1-D.  Shape of the input SparseTensor.
-   * @return a new instance of SparseSoftmax
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> SparseSoftmax<T> create(Scope scope, Operand<TInt64> spIndices, Operand<T> spValues, Operand<TInt64> spShape) {
-    OperationBuilder opBuilder = scope.env().opBuilder("SparseSoftmax", scope.makeOpName("SparseSoftmax"));
-    opBuilder.addInput(spIndices.asOutput(scope));
-    opBuilder.addInput(spValues.asOutput(scope));
-    opBuilder.addInput(spShape.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new SparseSoftmax<T>(opBuilder.build());
-  }
-  
-  /**
-   * 1-D.  The `NNZ` values for the result `SparseTensor`.
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "SparseSoftmax";
-  
-  private Output<T> output;
-  
-  private SparseSoftmax(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSparseMaximum.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSparseMaximum.java
deleted file mode 100644
index 01cccab7b55..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSparseMaximum.java
+++ /dev/null
@@ -1,93 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.sparse;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Returns the element-wise max of two SparseTensors.
- * <p>
- * Assumes the two SparseTensors have the same shape, i.e., no broadcasting.
- * 
- * @param <T> data type for {@code outputValues()} output
- */
-@Operator(group = "sparse")
-public final class SparseSparseMaximum<T extends TNumber> extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new SparseSparseMaximum operation.
-   * 
-   * @param scope current scope
-   * @param aIndices 2-D.  `N x R` matrix with the indices of non-empty values in a
-   * SparseTensor, in the canonical lexicographic ordering.
-   * @param aValues 1-D.  `N` non-empty values corresponding to `a_indices`.
-   * @param aShape 1-D.  Shape of the input SparseTensor.
-   * @param bIndices counterpart to `a_indices` for the other operand.
-   * @param bValues counterpart to `a_values` for the other operand; must be of the same dtype.
-   * @param bShape counterpart to `a_shape` for the other operand; the two shapes must be equal.
-   * @return a new instance of SparseSparseMaximum
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> SparseSparseMaximum<T> create(Scope scope, Operand<TInt64> aIndices, Operand<T> aValues, Operand<TInt64> aShape, Operand<TInt64> bIndices, Operand<T> bValues, Operand<TInt64> bShape) {
-    OperationBuilder opBuilder = scope.env().opBuilder("SparseSparseMaximum", scope.makeOpName("SparseSparseMaximum"));
-    opBuilder.addInput(aIndices.asOutput(scope));
-    opBuilder.addInput(aValues.asOutput(scope));
-    opBuilder.addInput(aShape.asOutput(scope));
-    opBuilder.addInput(bIndices.asOutput(scope));
-    opBuilder.addInput(bValues.asOutput(scope));
-    opBuilder.addInput(bShape.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new SparseSparseMaximum<T>(opBuilder.build());
-  }
-  
-  /**
-   * 2-D.  The indices of the output SparseTensor.
-   */
-  public Output<TInt64> outputIndices() {
-    return outputIndices;
-  }
-  
-  /**
-   * 1-D.  The values of the output SparseTensor.
-   */
-  public Output<T> outputValues() {
-    return outputValues;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "SparseSparseMaximum";
-  
-  private Output<TInt64> outputIndices;
-  private Output<T> outputValues;
-  
-  private SparseSparseMaximum(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    outputIndices = operation.output(outputIdx++);
-    outputValues = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSparseMinimum.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSparseMinimum.java
deleted file mode 100644
index cdfdaf03255..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSparseMinimum.java
+++ /dev/null
@@ -1,93 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.sparse;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
-
-/**
- * Returns the element-wise min of two SparseTensors.
- * <p>
- * Assumes the two SparseTensors have the same shape, i.e., no broadcasting.
- * 
- * @param <T> data type for {@code outputValues()} output
- */
-@Operator(group = "sparse")
-public final class SparseSparseMinimum<T extends TType> extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new SparseSparseMinimum operation.
-   * 
-   * @param scope current scope
-   * @param aIndices 2-D.  `N x R` matrix with the indices of non-empty values in a
-   * SparseTensor, in the canonical lexicographic ordering.
-   * @param aValues 1-D.  `N` non-empty values corresponding to `a_indices`.
-   * @param aShape 1-D.  Shape of the input SparseTensor.
-   * @param bIndices counterpart to `a_indices` for the other operand.
-   * @param bValues counterpart to `a_values` for the other operand; must be of the same dtype.
-   * @param bShape counterpart to `a_shape` for the other operand; the two shapes must be equal.
-   * @return a new instance of SparseSparseMinimum
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> SparseSparseMinimum<T> create(Scope scope, Operand<TInt64> aIndices, Operand<T> aValues, Operand<TInt64> aShape, Operand<TInt64> bIndices, Operand<T> bValues, Operand<TInt64> bShape) {
-    OperationBuilder opBuilder = scope.env().opBuilder("SparseSparseMinimum", scope.makeOpName("SparseSparseMinimum"));
-    opBuilder.addInput(aIndices.asOutput(scope));
-    opBuilder.addInput(aValues.asOutput(scope));
-    opBuilder.addInput(aShape.asOutput(scope));
-    opBuilder.addInput(bIndices.asOutput(scope));
-    opBuilder.addInput(bValues.asOutput(scope));
-    opBuilder.addInput(bShape.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new SparseSparseMinimum<T>(opBuilder.build());
-  }
-  
-  /**
-   * 2-D.  The indices of the output SparseTensor.
-   */
-  public Output<TInt64> outputIndices() {
-    return outputIndices;
-  }
-  
-  /**
-   * 1-D.  The values of the output SparseTensor.
-   */
-  public Output<T> outputValues() {
-    return outputValues;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "SparseSparseMinimum";
-  
-  private Output<TInt64> outputIndices;
-  private Output<T> outputValues;
-  
-  private SparseSparseMinimum(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    outputIndices = operation.output(outputIdx++);
-    outputValues = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSplit.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSplit.java
deleted file mode 100644
index ca60f9e528a..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSplit.java
+++ /dev/null
@@ -1,128 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.sparse;
-
-import java.util.Arrays;
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
-
-/**
- * Split a `SparseTensor` into `num_split` tensors along one dimension.
- * <p>
- * If the `shape[split_dim]` is not an integer multiple of `num_split`. Slices
- * `[0 : shape[split_dim] % num_split]` gets one extra dimension.
- * For example, if `split_dim = 1` and `num_split = 2` and the input is
- * <p>
- *     input_tensor = shape = [2, 7]
- *     [    a   d e  ]
- *     [b c          ]
- * <p>
- * Graphically the output tensors are:
- * <p>
- *     output_tensor[0] = shape = [2, 4]
- *     [    a  ]
- *     [b c    ]
- * <p>
- *     output_tensor[1] = shape = [2, 3]
- *     [ d e  ]
- *     [      ]
- * 
- * @param <T> data type for {@code outputValues()} output
- */
-@Operator(group = "sparse")
-public final class SparseSplit<T extends TType> extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new SparseSplit operation.
-   * 
-   * @param scope current scope
-   * @param splitDim 0-D.  The dimension along which to split.  Must be in the range
-   * `[0, rank(shape))`.
-   * @param indices 2-D tensor represents the indices of the sparse tensor.
-   * @param values 1-D tensor represents the values of the sparse tensor.
-   * @param shape 1-D. tensor represents the shape of the sparse tensor.
-   * output indices: A list of 1-D tensors represents the indices of the output
-   * sparse tensors.
-   * @param numSplit The number of ways to split.
-   * @return a new instance of SparseSplit
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> SparseSplit<T> create(Scope scope, Operand<TInt64> splitDim, Operand<TInt64> indices, Operand<T> values, Operand<TInt64> shape, Long numSplit) {
-    OperationBuilder opBuilder = scope.env().opBuilder("SparseSplit", scope.makeOpName("SparseSplit"));
-    opBuilder.addInput(splitDim.asOutput(scope));
-    opBuilder.addInput(indices.asOutput(scope));
-    opBuilder.addInput(values.asOutput(scope));
-    opBuilder.addInput(shape.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("num_split", numSplit);
-    return new SparseSplit<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public List<Output<TInt64>> outputIndices() {
-    return outputIndices;
-  }
-  
-  /**
-   * A list of 1-D tensors represents the values of the output sparse
-   * tensors.
-   */
-  public List<Output<T>> outputValues() {
-    return outputValues;
-  }
-  
-  /**
-   * A list of 1-D tensors represents the shape of the output sparse
-   * tensors.
-   */
-  public List<Output<TInt64>> outputShape() {
-    return outputShape;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "SparseSplit";
-  
-  private List<Output<TInt64>> outputIndices;
-  private List<Output<T>> outputValues;
-  private List<Output<TInt64>> outputShape;
-  
-  @SuppressWarnings("unchecked")
-  private SparseSplit(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    int outputIndicesLength = operation.outputListLength("output_indices");
-    outputIndices = Arrays.asList((Output<TInt64>[])operation.outputList(outputIdx, outputIndicesLength));
-    outputIdx += outputIndicesLength;
-    int outputValuesLength = operation.outputListLength("output_values");
-    outputValues = Arrays.asList((Output<T>[])operation.outputList(outputIdx, outputValuesLength));
-    outputIdx += outputValuesLength;
-    int outputShapeLength = operation.outputListLength("output_shape");
-    outputShape = Arrays.asList((Output<TInt64>[])operation.outputList(outputIdx, outputShapeLength));
-    outputIdx += outputShapeLength;
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseTensorDenseAdd.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseTensorDenseAdd.java
deleted file mode 100644
index 527f2b3a759..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseTensorDenseAdd.java
+++ /dev/null
@@ -1,83 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.sparse;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * Adds up a `SparseTensor` and a dense `Tensor`, producing a dense `Tensor`.
- * <p>
- * This Op does not require `a_indices` be sorted in standard lexicographic order.
- * 
- * @param <U> data type for {@code output()} output
- */
-@Operator(group = "sparse")
-public final class SparseTensorDenseAdd<U extends TType> extends RawOp implements Operand<U> {
-  
-  /**
-   * Factory method to create a class wrapping a new SparseTensorDenseAdd operation.
-   * 
-   * @param scope current scope
-   * @param aIndices 2-D.  The `indices` of the `SparseTensor`, with shape `[nnz, ndims]`.
-   * @param aValues 1-D.  The `values` of the `SparseTensor`, with shape `[nnz]`.
-   * @param aShape 1-D.  The `shape` of the `SparseTensor`, with shape `[ndims]`.
-   * @param b `ndims`-D Tensor.  With shape `a_shape`.
-   * @return a new instance of SparseTensorDenseAdd
-   */
-  @Endpoint(describeByClass = true)
-  public static <U extends TType, T extends TNumber> SparseTensorDenseAdd<U> create(Scope scope, Operand<T> aIndices, Operand<U> aValues, Operand<T> aShape, Operand<U> b) {
-    OperationBuilder opBuilder = scope.env().opBuilder("SparseTensorDenseAdd", scope.makeOpName("SparseTensorDenseAdd"));
-    opBuilder.addInput(aIndices.asOutput(scope));
-    opBuilder.addInput(aValues.asOutput(scope));
-    opBuilder.addInput(aShape.asOutput(scope));
-    opBuilder.addInput(b.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new SparseTensorDenseAdd<U>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<U> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<U> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "SparseTensorDenseAdd";
-  
-  private Output<U> output;
-  
-  private SparseTensorDenseAdd(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseTensorDenseMatMul.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseTensorDenseMatMul.java
deleted file mode 100644
index c8b245d515d..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseTensorDenseMatMul.java
+++ /dev/null
@@ -1,149 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.sparse;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * Multiply SparseTensor (of rank 2) "A" by dense matrix "B".
- * <p>
- * No validity checking is performed on the indices of A.  However, the following
- * input format is recommended for optimal behavior:
- * <p>
- * if adjoint_a == false:
- *   A should be sorted in lexicographically increasing order.  Use SparseReorder
- *   if you're not sure.
- * if adjoint_a == true:
- *   A should be sorted in order of increasing dimension 1 (i.e., "column major"
- *   order instead of "row major" order).
- * 
- * @param <U> data type for {@code product()} output
- */
-@Operator(group = "sparse")
-public final class SparseTensorDenseMatMul<U extends TType> extends RawOp implements Operand<U> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.sparse.SparseTensorDenseMatMul}
-   */
-  public static class Options {
-    
-    /**
-     * @param adjointA Use the adjoint of A in the matrix multiply.  If A is complex, this
-     * is transpose(conj(A)).  Otherwise it's transpose(A).
-     */
-    public Options adjointA(Boolean adjointA) {
-      this.adjointA = adjointA;
-      return this;
-    }
-    
-    /**
-     * @param adjointB Use the adjoint of B in the matrix multiply.  If B is complex, this
-     * is transpose(conj(B)).  Otherwise it's transpose(B).
-     */
-    public Options adjointB(Boolean adjointB) {
-      this.adjointB = adjointB;
-      return this;
-    }
-    
-    private Boolean adjointA;
-    private Boolean adjointB;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new SparseTensorDenseMatMul operation.
-   * 
-   * @param scope current scope
-   * @param aIndices 2-D.  The `indices` of the `SparseTensor`, size `[nnz, 2]` Matrix.
-   * @param aValues 1-D.  The `values` of the `SparseTensor`, size `[nnz]` Vector.
-   * @param aShape 1-D.  The `shape` of the `SparseTensor`, size `[2]` Vector.
-   * @param b 2-D.  A dense Matrix.
-   * @param options carries optional attributes values
-   * @return a new instance of SparseTensorDenseMatMul
-   */
-  @Endpoint(describeByClass = true)
-  public static <U extends TType, T extends TNumber> SparseTensorDenseMatMul<U> create(Scope scope, Operand<T> aIndices, Operand<U> aValues, Operand<TInt64> aShape, Operand<U> b, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("SparseTensorDenseMatMul", scope.makeOpName("SparseTensorDenseMatMul"));
-    opBuilder.addInput(aIndices.asOutput(scope));
-    opBuilder.addInput(aValues.asOutput(scope));
-    opBuilder.addInput(aShape.asOutput(scope));
-    opBuilder.addInput(b.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.adjointA != null) {
-          opBuilder.setAttr("adjoint_a", opts.adjointA);
-        }
-        if (opts.adjointB != null) {
-          opBuilder.setAttr("adjoint_b", opts.adjointB);
-        }
-      }
-    }
-    return new SparseTensorDenseMatMul<U>(opBuilder.build());
-  }
-  
-  /**
-   * @param adjointA Use the adjoint of A in the matrix multiply.  If A is complex, this
-   * is transpose(conj(A)).  Otherwise it's transpose(A).
-   */
-  public static Options adjointA(Boolean adjointA) {
-    return new Options().adjointA(adjointA);
-  }
-  
-  /**
-   * @param adjointB Use the adjoint of B in the matrix multiply.  If B is complex, this
-   * is transpose(conj(B)).  Otherwise it's transpose(B).
-   */
-  public static Options adjointB(Boolean adjointB) {
-    return new Options().adjointB(adjointB);
-  }
-  
-  /**
-   */
-  public Output<U> product() {
-    return product;
-  }
-  
-  @Override
-  public Output<U> asOutput(Scope scope) {
-    return product;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "SparseTensorDenseMatMul";
-  
-  private Output<U> product;
-  
-  private SparseTensorDenseMatMul(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    product = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseToDense.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseToDense.java
deleted file mode 100644
index 27f5f67a4e9..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseToDense.java
+++ /dev/null
@@ -1,139 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.sparse;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * Converts a sparse representation into a dense tensor.
- * <p>
- * Builds an array `dense` with shape `output_shape` such that
- * <pre>{@code
- * # If sparse_indices is scalar
- * dense[i] = (i == sparse_indices ? sparse_values : default_value)
- * 
- * # If sparse_indices is a vector, then for each i
- * dense[sparse_indices[i]] = sparse_values[i]
- * 
- * # If sparse_indices is an n by d matrix, then for each i in [0, n)
- * dense[sparse_indices[i][0], ..., sparse_indices[i][d-1]] = sparse_values[i]
- * }</pre>
- * All other values in `dense` are set to `default_value`.  If `sparse_values` is a
- * scalar, all sparse indices are set to this single value.
- * <p>
- * Indices should be sorted in lexicographic order, and indices must not
- * contain any repeats. If `validate_indices` is true, these properties
- * are checked during execution.
- * 
- * @param <U> data type for {@code dense()} output
- */
-@Operator(group = "sparse")
-public final class SparseToDense<U extends TType> extends RawOp implements Operand<U> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.sparse.SparseToDense}
-   */
-  public static class Options {
-    
-    /**
-     * @param validateIndices If true, indices are checked to make sure they are sorted in
-     * lexicographic order and that there are no repeats.
-     */
-    public Options validateIndices(Boolean validateIndices) {
-      this.validateIndices = validateIndices;
-      return this;
-    }
-    
-    private Boolean validateIndices;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new SparseToDense operation.
-   * 
-   * @param scope current scope
-   * @param sparseIndices 0-D, 1-D, or 2-D.  `sparse_indices[i]` contains the complete
-   * index where `sparse_values[i]` will be placed.
-   * @param outputShape 1-D.  Shape of the dense output tensor.
-   * @param sparseValues 1-D.  Values corresponding to each row of `sparse_indices`,
-   * or a scalar value to be used for all sparse indices.
-   * @param defaultValue Scalar value to set for indices not specified in
-   * `sparse_indices`.
-   * @param options carries optional attributes values
-   * @return a new instance of SparseToDense
-   */
-  @Endpoint(describeByClass = true)
-  public static <U extends TType, T extends TNumber> SparseToDense<U> create(Scope scope, Operand<T> sparseIndices, Operand<T> outputShape, Operand<U> sparseValues, Operand<U> defaultValue, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("SparseToDense", scope.makeOpName("SparseToDense"));
-    opBuilder.addInput(sparseIndices.asOutput(scope));
-    opBuilder.addInput(outputShape.asOutput(scope));
-    opBuilder.addInput(sparseValues.asOutput(scope));
-    opBuilder.addInput(defaultValue.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.validateIndices != null) {
-          opBuilder.setAttr("validate_indices", opts.validateIndices);
-        }
-      }
-    }
-    return new SparseToDense<U>(opBuilder.build());
-  }
-  
-  /**
-   * @param validateIndices If true, indices are checked to make sure they are sorted in
-   * lexicographic order and that there are no repeats.
-   */
-  public static Options validateIndices(Boolean validateIndices) {
-    return new Options().validateIndices(validateIndices);
-  }
-  
-  /**
-   * Dense output tensor of shape `output_shape`.
-   */
-  public Output<U> dense() {
-    return dense;
-  }
-  
-  @Override
-  public Output<U> asOutput(Scope scope) {
-    return dense;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "SparseToDense";
-  
-  private Output<U> dense;
-  
-  private SparseToDense(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    dense = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseToSparseSetOperation.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseToSparseSetOperation.java
deleted file mode 100644
index 1d084cb04f7..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseToSparseSetOperation.java
+++ /dev/null
@@ -1,169 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.sparse;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
-
-/**
- * Applies set operation along last dimension of 2 `SparseTensor` inputs.
- * <p>
- * See SetOperationOp::SetOperationFromContext for values of `set_operation`.
- * <p>
- * If `validate_indices` is `True`, `sparse.SparseToSparseSetOperation` validates the
- * order and range of `set1` and `set2` indices.
- * <p>
- * Input `set1` is a `SparseTensor` represented by `set1_indices`, `set1_values`,
- * and `set1_shape`. For `set1` ranked `n`, 1st `n-1` dimensions must be the same
- * as `set2`. Dimension `n` contains values in a set, duplicates are allowed but
- * ignored.
- * <p>
- * Input `set2` is a `SparseTensor` represented by `set2_indices`, `set2_values`,
- * and `set2_shape`. For `set2` ranked `n`, 1st `n-1` dimensions must be the same
- * as `set1`. Dimension `n` contains values in a set, duplicates are allowed but
- * ignored.
- * <p>
- * If `validate_indices` is `True`, this op validates the order and range of `set1`
- * and `set2` indices.
- * <p>
- * Output `result` is a `SparseTensor` represented by `result_indices`,
- * `result_values`, and `result_shape`. For `set1` and `set2` ranked `n`, this
- * has rank `n` and the same 1st `n-1` dimensions as `set1` and `set2`. The `nth`
- * dimension contains the result of `set_operation` applied to the corresponding
- * `[0...n-1]` dimension of `set`.
- * 
- * @param <T> data type for {@code resultValues()} output
- */
-@Operator(group = "sparse")
-public final class SparseToSparseSetOperation<T extends TType> extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.sparse.SparseToSparseSetOperation}
-   */
-  public static class Options {
-    
-    /**
-     * @param validateIndices 
-     */
-    public Options validateIndices(Boolean validateIndices) {
-      this.validateIndices = validateIndices;
-      return this;
-    }
-    
-    private Boolean validateIndices;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new SparseToSparseSetOperation operation.
-   * 
-   * @param scope current scope
-   * @param set1Indices 2D `Tensor`, indices of a `SparseTensor`. Must be in row-major
-   * order.
-   * @param set1Values 1D `Tensor`, values of a `SparseTensor`. Must be in row-major
-   * order.
-   * @param set1Shape 1D `Tensor`, shape of a `SparseTensor`. `set1_shape[0...n-1]` must
-   * be the same as `set2_shape[0...n-1]`, `set1_shape[n]` is the
-   * max set size across `0...n-1` dimensions.
-   * @param set2Indices 2D `Tensor`, indices of a `SparseTensor`. Must be in row-major
-   * order.
-   * @param set2Values 1D `Tensor`, values of a `SparseTensor`. Must be in row-major
-   * order.
-   * @param set2Shape 1D `Tensor`, shape of a `SparseTensor`. `set2_shape[0...n-1]` must
-   * be the same as `set1_shape[0...n-1]`, `set2_shape[n]` is the
-   * max set size across `0...n-1` dimensions.
-   * @param setOperation 
-   * @param options carries optional attributes values
-   * @return a new instance of SparseToSparseSetOperation
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> SparseToSparseSetOperation<T> create(Scope scope, Operand<TInt64> set1Indices, Operand<T> set1Values, Operand<TInt64> set1Shape, Operand<TInt64> set2Indices, Operand<T> set2Values, Operand<TInt64> set2Shape, String setOperation, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("SparseToSparseSetOperation", scope.makeOpName("SparseToSparseSetOperation"));
-    opBuilder.addInput(set1Indices.asOutput(scope));
-    opBuilder.addInput(set1Values.asOutput(scope));
-    opBuilder.addInput(set1Shape.asOutput(scope));
-    opBuilder.addInput(set2Indices.asOutput(scope));
-    opBuilder.addInput(set2Values.asOutput(scope));
-    opBuilder.addInput(set2Shape.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("set_operation", setOperation);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.validateIndices != null) {
-          opBuilder.setAttr("validate_indices", opts.validateIndices);
-        }
-      }
-    }
-    return new SparseToSparseSetOperation<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param validateIndices 
-   */
-  public static Options validateIndices(Boolean validateIndices) {
-    return new Options().validateIndices(validateIndices);
-  }
-  
-  /**
-   * 2D indices of a `SparseTensor`.
-   */
-  public Output<TInt64> resultIndices() {
-    return resultIndices;
-  }
-  
-  /**
-   * 1D values of a `SparseTensor`.
-   */
-  public Output<T> resultValues() {
-    return resultValues;
-  }
-  
-  /**
-   * 1D `Tensor` shape of a `SparseTensor`. `result_shape[0...n-1]` is
-   * the same as the 1st `n-1` dimensions of `set1` and `set2`, `result_shape[n]`
-   * is the max result set size across all `0...n-1` dimensions.
-   */
-  public Output<TInt64> resultShape() {
-    return resultShape;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "SparseToSparseSetOperation";
-  
-  private Output<TInt64> resultIndices;
-  private Output<T> resultValues;
-  private Output<TInt64> resultShape;
-  
-  private SparseToSparseSetOperation(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    resultIndices = operation.output(outputIdx++);
-    resultValues = operation.output(outputIdx++);
-    resultShape = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/TakeManySparseFromTensorsMap.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/TakeManySparseFromTensorsMap.java
deleted file mode 100644
index 7cf6141873f..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/TakeManySparseFromTensorsMap.java
+++ /dev/null
@@ -1,195 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.sparse;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
-
-/**
- * Read `SparseTensors` from a `SparseTensorsMap` and concatenate them.
- * <p>
- * The input `sparse_handles` must be an `int64` matrix of shape `[N, 1]` where
- * `N` is the minibatch size and the rows correspond to the output handles of
- * `AddSparseToTensorsMap` or `AddManySparseToTensorsMap`.  The ranks of the
- * original `SparseTensor` objects that went into the given input ops must all
- * match.  When the final `SparseTensor` is created, it has rank one
- * higher than the ranks of the incoming `SparseTensor` objects
- * (they have been concatenated along a new row dimension on the left).
- * <p>
- * The output `SparseTensor` object's shape values for all dimensions but the
- * first are the max across the input `SparseTensor` objects' shape values
- * for the corresponding dimensions.  Its first shape value is `N`, the minibatch
- * size.
- * <p>
- * The input `SparseTensor` objects' indices are assumed ordered in
- * standard lexicographic order.  If this is not the case, after this
- * step run `SparseReorder` to restore index ordering.
- * <p>
- * For example, if the handles represent an input, which is a `[2, 3]` matrix
- * representing two original `SparseTensor` objects:
- * <pre>{@code
- *     index = [ 0]
- *             [10]
- *             [20]
- *     values = [1, 2, 3]
- *     shape = [50]
- * }</pre>
- * and
- * <pre>{@code
- *     index = [ 2]
- *             [10]
- *     values = [4, 5]
- *     shape = [30]
- * }</pre>
- * then the final `SparseTensor` will be:
- * <pre>{@code
- *     index = [0  0]
- *             [0 10]
- *             [0 20]
- *             [1  2]
- *             [1 10]
- *     values = [1, 2, 3, 4, 5]
- *     shape = [2 50]
- * }</pre>
- * 
- * 
- * @param <T> data type for {@code sparseValues()} output
- */
-@Operator(group = "sparse")
-public final class TakeManySparseFromTensorsMap<T extends TType> extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.sparse.TakeManySparseFromTensorsMap}
-   */
-  public static class Options {
-    
-    /**
-     * @param container The container name for the `SparseTensorsMap` read by this op.
-     */
-    public Options container(String container) {
-      this.container = container;
-      return this;
-    }
-    
-    /**
-     * @param sharedName The shared name for the `SparseTensorsMap` read by this op.
-     * It should not be blank; rather the `shared_name` or unique Operation name
-     * of the Op that created the original `SparseTensorsMap` should be used.
-     */
-    public Options sharedName(String sharedName) {
-      this.sharedName = sharedName;
-      return this;
-    }
-    
-    private String container;
-    private String sharedName;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new TakeManySparseFromTensorsMap operation.
-   * 
-   * @param scope current scope
-   * @param sparseHandles 1-D, The `N` serialized `SparseTensor` objects.
-   * Shape: `[N]`.
-   * @param dtype The `dtype` of the `SparseTensor` objects stored in the
-   * `SparseTensorsMap`.
-   * @param options carries optional attributes values
-   * @return a new instance of TakeManySparseFromTensorsMap
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> TakeManySparseFromTensorsMap<T> create(Scope scope, Operand<TInt64> sparseHandles, Class<T> dtype, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("TakeManySparseFromTensorsMap", scope.makeOpName("TakeManySparseFromTensorsMap"));
-    opBuilder.addInput(sparseHandles.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("dtype", dtype);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.container != null) {
-          opBuilder.setAttr("container", opts.container);
-        }
-        if (opts.sharedName != null) {
-          opBuilder.setAttr("shared_name", opts.sharedName);
-        }
-      }
-    }
-    return new TakeManySparseFromTensorsMap<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param container The container name for the `SparseTensorsMap` read by this op.
-   */
-  public static Options container(String container) {
-    return new Options().container(container);
-  }
-  
-  /**
-   * @param sharedName The shared name for the `SparseTensorsMap` read by this op.
-   * It should not be blank; rather the `shared_name` or unique Operation name
-   * of the Op that created the original `SparseTensorsMap` should be used.
-   */
-  public static Options sharedName(String sharedName) {
-    return new Options().sharedName(sharedName);
-  }
-  
-  /**
-   * 2-D.  The `indices` of the minibatch `SparseTensor`.
-   */
-  public Output<TInt64> sparseIndices() {
-    return sparseIndices;
-  }
-  
-  /**
-   * 1-D.  The `values` of the minibatch `SparseTensor`.
-   */
-  public Output<T> sparseValues() {
-    return sparseValues;
-  }
-  
-  /**
-   * 1-D.  The `shape` of the minibatch `SparseTensor`.
-   */
-  public Output<TInt64> sparseShape() {
-    return sparseShape;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "TakeManySparseFromTensorsMap";
-  
-  private Output<TInt64> sparseIndices;
-  private Output<T> sparseValues;
-  private Output<TInt64> sparseShape;
-  
-  private TakeManySparseFromTensorsMap(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    sparseIndices = operation.output(outputIdx++);
-    sparseValues = operation.output(outputIdx++);
-    sparseShape = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/Join.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/Join.java
deleted file mode 100644
index f52eff740ef..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/Join.java
+++ /dev/null
@@ -1,117 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.strings;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.Operands;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TString;
-
-/**
- * Joins the strings in the given list of string tensors into one tensor;
- * <p>
- * with the given separator (default is an empty separator).
- * <p>
- * Examples:
- * <p>
- * >>> s = ["hello", "world", "tensorflow"]
- * >>> tf.strings.join(s, " ")
- * <tf.Tensor: shape=(), dtype=string, numpy=b'hello world tensorflow'>
- */
-@Operator(group = "strings")
-public final class Join extends RawOp implements Operand<TString> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.strings.Join}
-   */
-  public static class Options {
-    
-    /**
-     * @param separator string, an optional join separator.
-     */
-    public Options separator(String separator) {
-      this.separator = separator;
-      return this;
-    }
-    
-    private String separator;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new Join operation.
-   * 
-   * @param scope current scope
-   * @param inputs A list of string tensors.  The tensors must all have the same shape,
-   * or be scalars.  Scalars may be mixed in; these will be broadcast to the shape
-   * of non-scalar inputs.
-   * @param options carries optional attributes values
-   * @return a new instance of Join
-   */
-  @Endpoint(describeByClass = true)
-  public static Join create(Scope scope, Iterable<Operand<TString>> inputs, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("StringJoin", scope.makeOpName("Join"));
-    opBuilder.addInputList(Operands.asOutputs(scope, inputs));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.separator != null) {
-          opBuilder.setAttr("separator", opts.separator);
-        }
-      }
-    }
-    return new Join(opBuilder.build());
-  }
-  
-  /**
-   * @param separator string, an optional join separator.
-   */
-  public static Options separator(String separator) {
-    return new Options().separator(separator);
-  }
-  
-  /**
-   */
-  public Output<TString> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<TString> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "StringJoin";
-  
-  private Output<TString> output;
-  
-  private Join(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/Lower.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/Lower.java
deleted file mode 100644
index 634fa32914e..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/Lower.java
+++ /dev/null
@@ -1,112 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.strings;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TString;
-
-/**
- * Converts all uppercase characters into their respective lowercase replacements.
- * <p>
- * Example:
- * <p>
- * >>> tf.strings.lower("CamelCase string and ALL CAPS")
- * <tf.Tensor: shape=(), dtype=string, numpy=b'camelcase string and all caps'>
- * 
- */
-@Operator(group = "strings")
-public final class Lower extends RawOp implements Operand<TString> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.strings.Lower}
-   */
-  public static class Options {
-    
-    /**
-     * @param encoding 
-     */
-    public Options encoding(String encoding) {
-      this.encoding = encoding;
-      return this;
-    }
-    
-    private String encoding;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new Lower operation.
-   * 
-   * @param scope current scope
-   * @param input 
-   * @param options carries optional attributes values
-   * @return a new instance of Lower
-   */
-  @Endpoint(describeByClass = true)
-  public static Lower create(Scope scope, Operand<TString> input, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("StringLower", scope.makeOpName("Lower"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.encoding != null) {
-          opBuilder.setAttr("encoding", opts.encoding);
-        }
-      }
-    }
-    return new Lower(opBuilder.build());
-  }
-  
-  /**
-   * @param encoding 
-   */
-  public static Options encoding(String encoding) {
-    return new Options().encoding(encoding);
-  }
-  
-  /**
-   */
-  public Output<TString> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<TString> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "StringLower";
-  
-  private Output<TString> output;
-  
-  private Lower(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/ReduceJoin.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/ReduceJoin.java
deleted file mode 100644
index b07f92d0ccc..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/ReduceJoin.java
+++ /dev/null
@@ -1,156 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.strings;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.TString;
-
-/**
- * Joins a string Tensor across the given dimensions.
- * <p>
- * Computes the string join across dimensions in the given string Tensor of shape
- * `[\\(d_0, d_1, ..., d_{n-1}\\)]`.  Returns a new Tensor created by joining the input
- * strings with the given separator (default: empty string).  Negative indices are
- * counted backwards from the end, with `-1` being equivalent to `n - 1`.  If
- * indices are not specified, joins across all dimensions beginning from `n - 1`
- * through `0`.
- * <p>
- * For example:
- * <pre>{@code
- * # tensor `a` is [["a", "b"], ["c", "d"]]
- * tf.reduce_join(a, 0) ==> ["ac", "bd"]
- * tf.reduce_join(a, 1) ==> ["ab", "cd"]
- * tf.reduce_join(a, -2) = tf.reduce_join(a, 0) ==> ["ac", "bd"]
- * tf.reduce_join(a, -1) = tf.reduce_join(a, 1) ==> ["ab", "cd"]
- * tf.reduce_join(a, 0, keep_dims=True) ==> [["ac", "bd"]]
- * tf.reduce_join(a, 1, keep_dims=True) ==> [["ab"], ["cd"]]
- * tf.reduce_join(a, 0, separator=".") ==> ["a.c", "b.d"]
- * tf.reduce_join(a, [0, 1]) ==> "acbd"
- * tf.reduce_join(a, [1, 0]) ==> "abcd"
- * tf.reduce_join(a, []) ==> [["a", "b"], ["c", "d"]]
- * tf.reduce_join(a) = tf.reduce_join(a, [1, 0]) ==> "abcd"
- * }</pre>
- * 
- */
-@Operator(group = "strings")
-public final class ReduceJoin extends RawOp implements Operand<TString> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.strings.ReduceJoin}
-   */
-  public static class Options {
-    
-    /**
-     * @param keepDims If `True`, retain reduced dimensions with length `1`.
-     */
-    public Options keepDims(Boolean keepDims) {
-      this.keepDims = keepDims;
-      return this;
-    }
-    
-    /**
-     * @param separator The separator to use when joining.
-     */
-    public Options separator(String separator) {
-      this.separator = separator;
-      return this;
-    }
-    
-    private Boolean keepDims;
-    private String separator;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new ReduceJoin operation.
-   * 
-   * @param scope current scope
-   * @param inputs The input to be joined.  All reduced indices must have non-zero size.
-   * @param reductionIndices The dimensions to reduce over.  Dimensions are reduced in the
-   * order specified.  Omitting `reduction_indices` is equivalent to passing
-   * `[n-1, n-2, ..., 0]`.  Negative indices from `-n` to `-1` are supported.
-   * @param options carries optional attributes values
-   * @return a new instance of ReduceJoin
-   */
-  @Endpoint(describeByClass = true)
-  public static ReduceJoin create(Scope scope, Operand<TString> inputs, Operand<TInt32> reductionIndices, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ReduceJoin", scope.makeOpName("ReduceJoin"));
-    opBuilder.addInput(inputs.asOutput(scope));
-    opBuilder.addInput(reductionIndices.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.keepDims != null) {
-          opBuilder.setAttr("keep_dims", opts.keepDims);
-        }
-        if (opts.separator != null) {
-          opBuilder.setAttr("separator", opts.separator);
-        }
-      }
-    }
-    return new ReduceJoin(opBuilder.build());
-  }
-  
-  /**
-   * @param keepDims If `True`, retain reduced dimensions with length `1`.
-   */
-  public static Options keepDims(Boolean keepDims) {
-    return new Options().keepDims(keepDims);
-  }
-  
-  /**
-   * @param separator The separator to use when joining.
-   */
-  public static Options separator(String separator) {
-    return new Options().separator(separator);
-  }
-  
-  /**
-   * Has shape equal to that of the input with reduced dimensions removed or
-   * set to `1` depending on `keep_dims`.
-   */
-  public Output<TString> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<TString> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ReduceJoin";
-  
-  private Output<TString> output;
-  
-  private ReduceJoin(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/RegexFullMatch.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/RegexFullMatch.java
deleted file mode 100644
index a5462464baa..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/RegexFullMatch.java
+++ /dev/null
@@ -1,90 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.strings;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TBool;
-import org.tensorflow.types.TString;
-
-/**
- * Check if the input matches the regex pattern.
- * <p>
- * The input is a string tensor of any shape. The pattern is a scalar
- * string tensor which is applied to every element of the input tensor.
- * The boolean values (True or False) of the output tensor indicate
- * if the input matches the regex pattern provided.
- * <p>
- * The pattern follows the re2 syntax (https://github.com/google/re2/wiki/Syntax)
- * <p>
- * Examples:
- * <p>
- * >>> tf.strings.regex_full_match(["TF lib", "lib TF"], ".*lib$")
- * <tf.Tensor: shape=(2,), dtype=bool, numpy=array([ True, False])>
- * >>> tf.strings.regex_full_match(["TF lib", "lib TF"], ".*TF$")
- * <tf.Tensor: shape=(2,), dtype=bool, numpy=array([False,  True])>
- */
-@Operator(group = "strings")
-public final class RegexFullMatch extends RawOp implements Operand<TBool> {
-  
-  /**
-   * Factory method to create a class wrapping a new RegexFullMatch operation.
-   * 
-   * @param scope current scope
-   * @param input A string tensor of the text to be processed.
-   * @param pattern A scalar string tensor containing the regular expression to match the input.
-   * @return a new instance of RegexFullMatch
-   */
-  @Endpoint(describeByClass = true)
-  public static RegexFullMatch create(Scope scope, Operand<TString> input, Operand<TString> pattern) {
-    OperationBuilder opBuilder = scope.env().opBuilder("RegexFullMatch", scope.makeOpName("RegexFullMatch"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder.addInput(pattern.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new RegexFullMatch(opBuilder.build());
-  }
-  
-  /**
-   * A bool tensor with the same shape as `input`.
-   */
-  public Output<TBool> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<TBool> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "RegexFullMatch";
-  
-  private Output<TBool> output;
-  
-  private RegexFullMatch(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/RegexReplace.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/RegexReplace.java
deleted file mode 100644
index 475eaf4035c..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/RegexReplace.java
+++ /dev/null
@@ -1,119 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.strings;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TString;
-
-/**
- * Replaces matches of the `pattern` regular expression in `input` with the
- * replacement string provided in `rewrite`.
- * <p>
- * It follows the re2 syntax (https://github.com/google/re2/wiki/Syntax)
- */
-@Operator(group = "strings")
-public final class RegexReplace extends RawOp implements Operand<TString> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.strings.RegexReplace}
-   */
-  public static class Options {
-    
-    /**
-     * @param replaceGlobal If True, the replacement is global (that is, all matches of the `pattern` regular
-     * expression in each input string are rewritten), otherwise the `rewrite`
-     * substitution is only made for the first `pattern` match.
-     */
-    public Options replaceGlobal(Boolean replaceGlobal) {
-      this.replaceGlobal = replaceGlobal;
-      return this;
-    }
-    
-    private Boolean replaceGlobal;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new RegexReplace operation.
-   * 
-   * @param scope current scope
-   * @param input The text to be processed.
-   * @param pattern The regular expression to be matched in the `input` strings.
-   * @param rewrite The rewrite string to be substituted for the `pattern` expression where it is
-   * matched in the `input` strings.
-   * @param options carries optional attributes values
-   * @return a new instance of RegexReplace
-   */
-  @Endpoint(describeByClass = true)
-  public static RegexReplace create(Scope scope, Operand<TString> input, Operand<TString> pattern, Operand<TString> rewrite, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("RegexReplace", scope.makeOpName("RegexReplace"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder.addInput(pattern.asOutput(scope));
-    opBuilder.addInput(rewrite.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.replaceGlobal != null) {
-          opBuilder.setAttr("replace_global", opts.replaceGlobal);
-        }
-      }
-    }
-    return new RegexReplace(opBuilder.build());
-  }
-  
-  /**
-   * @param replaceGlobal If True, the replacement is global (that is, all matches of the `pattern` regular
-   * expression in each input string are rewritten), otherwise the `rewrite`
-   * substitution is only made for the first `pattern` match.
-   */
-  public static Options replaceGlobal(Boolean replaceGlobal) {
-    return new Options().replaceGlobal(replaceGlobal);
-  }
-  
-  /**
-   * The text after applying pattern match and rewrite substitution.
-   */
-  public Output<TString> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<TString> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "RegexReplace";
-  
-  private Output<TString> output;
-  
-  private RegexReplace(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/StaticRegexFullMatch.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/StaticRegexFullMatch.java
deleted file mode 100644
index db75da8afc5..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/StaticRegexFullMatch.java
+++ /dev/null
@@ -1,82 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.strings;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TBool;
-import org.tensorflow.types.TString;
-
-/**
- * Check if the input matches the regex pattern.
- * <p>
- * The input is a string tensor of any shape. The pattern is the
- * regular expression to be matched with every element of the input tensor.
- * The boolean values (True or False) of the output tensor indicate
- * if the input matches the regex pattern provided.
- * <p>
- * The pattern follows the re2 syntax (https://github.com/google/re2/wiki/Syntax)
- */
-public final class StaticRegexFullMatch extends RawOp implements Operand<TBool> {
-  
-  /**
-   * Factory method to create a class wrapping a new StaticRegexFullMatch operation.
-   * 
-   * @param scope current scope
-   * @param input A string tensor of the text to be processed.
-   * @param pattern The regular expression to match the input.
-   * @return a new instance of StaticRegexFullMatch
-   */
-  @Endpoint(describeByClass = true)
-  public static StaticRegexFullMatch create(Scope scope, Operand<TString> input, String pattern) {
-    OperationBuilder opBuilder = scope.env().opBuilder("StaticRegexFullMatch", scope.makeOpName("StaticRegexFullMatch"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("pattern", pattern);
-    return new StaticRegexFullMatch(opBuilder.build());
-  }
-  
-  /**
-   * A bool tensor with the same shape as `input`.
-   */
-  public Output<TBool> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<TBool> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "StaticRegexFullMatch";
-  
-  private Output<TBool> output;
-  
-  private StaticRegexFullMatch(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/StaticRegexReplace.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/StaticRegexReplace.java
deleted file mode 100644
index ee7a1df0994..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/StaticRegexReplace.java
+++ /dev/null
@@ -1,114 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.strings;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TString;
-
-/**
- * Replaces the match of pattern in input with rewrite.
- * <p>
- * It follows the re2 syntax (https://github.com/google/re2/wiki/Syntax)
- */
-public final class StaticRegexReplace extends RawOp implements Operand<TString> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.strings.StaticRegexReplace}
-   */
-  public static class Options {
-    
-    /**
-     * @param replaceGlobal If True, the replacement is global, otherwise the replacement
-     * is done only on the first match.
-     */
-    public Options replaceGlobal(Boolean replaceGlobal) {
-      this.replaceGlobal = replaceGlobal;
-      return this;
-    }
-    
-    private Boolean replaceGlobal;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new StaticRegexReplace operation.
-   * 
-   * @param scope current scope
-   * @param input The text to be processed.
-   * @param pattern The regular expression to match the input.
-   * @param rewrite The rewrite to be applied to the matched expression.
-   * @param options carries optional attributes values
-   * @return a new instance of StaticRegexReplace
-   */
-  @Endpoint(describeByClass = true)
-  public static StaticRegexReplace create(Scope scope, Operand<TString> input, String pattern, String rewrite, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("StaticRegexReplace", scope.makeOpName("StaticRegexReplace"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("pattern", pattern);
-    opBuilder.setAttr("rewrite", rewrite);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.replaceGlobal != null) {
-          opBuilder.setAttr("replace_global", opts.replaceGlobal);
-        }
-      }
-    }
-    return new StaticRegexReplace(opBuilder.build());
-  }
-  
-  /**
-   * @param replaceGlobal If True, the replacement is global, otherwise the replacement
-   * is done only on the first match.
-   */
-  public static Options replaceGlobal(Boolean replaceGlobal) {
-    return new Options().replaceGlobal(replaceGlobal);
-  }
-  
-  /**
-   * The text after applying pattern and rewrite.
-   */
-  public Output<TString> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<TString> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "StaticRegexReplace";
-  
-  private Output<TString> output;
-  
-  private StaticRegexReplace(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/StringFormat.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/StringFormat.java
deleted file mode 100644
index 706f9d769ea..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/StringFormat.java
+++ /dev/null
@@ -1,148 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.strings;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.Operands;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TString;
-
-/**
- * Formats a string template using a list of tensors.
- * <p>
- * Formats a string template using a list of tensors, pretty-printing tensor summaries.
- */
-@Operator(group = "strings")
-public final class StringFormat extends RawOp implements Operand<TString> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.strings.StringFormat}
-   */
-  public static class Options {
-    
-    /**
-     * @param template A string, the template to format tensor summaries into.
-     */
-    public Options template(String template) {
-      this.template = template;
-      return this;
-    }
-    
-    /**
-     * @param placeholder A string, at each placeholder in the template a subsequent tensor summary will be inserted.
-     */
-    public Options placeholder(String placeholder) {
-      this.placeholder = placeholder;
-      return this;
-    }
-    
-    /**
-     * @param summarize When formatting the tensor summaries print the first and last summarize entries of each tensor dimension.
-     */
-    public Options summarize(Long summarize) {
-      this.summarize = summarize;
-      return this;
-    }
-    
-    private String template;
-    private String placeholder;
-    private Long summarize;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new StringFormat operation.
-   * 
-   * @param scope current scope
-   * @param inputs The list of tensors to format into the placeholder string.
-   * @param options carries optional attributes values
-   * @return a new instance of StringFormat
-   */
-  @Endpoint(describeByClass = true)
-  public static StringFormat create(Scope scope, Iterable<Operand<?>> inputs, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("StringFormat", scope.makeOpName("StringFormat"));
-    opBuilder.addInputList(Operands.asOutputs(scope, inputs));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.template != null) {
-          opBuilder.setAttr("template", opts.template);
-        }
-        if (opts.placeholder != null) {
-          opBuilder.setAttr("placeholder", opts.placeholder);
-        }
-        if (opts.summarize != null) {
-          opBuilder.setAttr("summarize", opts.summarize);
-        }
-      }
-    }
-    return new StringFormat(opBuilder.build());
-  }
-  
-  /**
-   * @param template A string, the template to format tensor summaries into.
-   */
-  public static Options template(String template) {
-    return new Options().template(template);
-  }
-  
-  /**
-   * @param placeholder A string, at each placeholder in the template a subsequent tensor summary will be inserted.
-   */
-  public static Options placeholder(String placeholder) {
-    return new Options().placeholder(placeholder);
-  }
-  
-  /**
-   * @param summarize When formatting the tensor summaries print the first and last summarize entries of each tensor dimension.
-   */
-  public static Options summarize(Long summarize) {
-    return new Options().summarize(summarize);
-  }
-  
-  /**
-   * = The resulting string scalar.
-   */
-  public Output<TString> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<TString> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "StringFormat";
-  
-  private Output<TString> output;
-  
-  private StringFormat(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/StringLength.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/StringLength.java
deleted file mode 100644
index c92ed01631b..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/StringLength.java
+++ /dev/null
@@ -1,126 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.strings;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.TString;
-
-/**
- * String lengths of `input`.
- * <p>
- * Computes the length of each string given in the input tensor.
- * <p>
- * >>> strings = tf.constant(['Hello','TensorFlow', '\U0001F642'])
- * >>> tf.strings.length(strings).numpy() # default counts bytes
- * array([ 5, 10, 4], dtype=int32)
- * >>> tf.strings.length(strings, unit="UTF8_CHAR").numpy()
- * array([ 5, 10, 1], dtype=int32)
- * 
- */
-@Operator(group = "strings")
-public final class StringLength extends RawOp implements Operand<TInt32> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.strings.StringLength}
-   */
-  public static class Options {
-    
-    /**
-     * @param unit The unit that is counted to compute string length.  One of: `"BYTE"` (for
-     * the number of bytes in each string) or `"UTF8_CHAR"` (for the number of UTF-8
-     * encoded Unicode code points in each string).  Results are undefined
-     * if `unit=UTF8_CHAR` and the `input` strings do not contain structurally
-     * valid UTF-8.
-     */
-    public Options unit(String unit) {
-      this.unit = unit;
-      return this;
-    }
-    
-    private String unit;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new StringLength operation.
-   * 
-   * @param scope current scope
-   * @param input The strings for which to compute the length for each element.
-   * @param options carries optional attributes values
-   * @return a new instance of StringLength
-   */
-  @Endpoint(describeByClass = true)
-  public static StringLength create(Scope scope, Operand<TString> input, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("StringLength", scope.makeOpName("StringLength"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.unit != null) {
-          opBuilder.setAttr("unit", opts.unit);
-        }
-      }
-    }
-    return new StringLength(opBuilder.build());
-  }
-  
-  /**
-   * @param unit The unit that is counted to compute string length.  One of: `"BYTE"` (for
-   * the number of bytes in each string) or `"UTF8_CHAR"` (for the number of UTF-8
-   * encoded Unicode code points in each string).  Results are undefined
-   * if `unit=UTF8_CHAR` and the `input` strings do not contain structurally
-   * valid UTF-8.
-   */
-  public static Options unit(String unit) {
-    return new Options().unit(unit);
-  }
-  
-  /**
-   * Integer tensor that has the same shape as `input`. The output contains the
-   * element-wise string lengths of `input`.
-   */
-  public Output<TInt32> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<TInt32> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "StringLength";
-  
-  private Output<TInt32> output;
-  
-  private StringLength(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/StringNGrams.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/StringNGrams.java
deleted file mode 100644
index fe7b1463c46..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/StringNGrams.java
+++ /dev/null
@@ -1,109 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.strings;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TString;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Creates ngrams from ragged string data.
- * <p>
- * This op accepts a ragged tensor with 1 ragged dimension containing only
- * strings and outputs a ragged tensor with 1 ragged dimension containing ngrams
- * of that string, joined along the innermost axis.
- * 
- * @param <T> data type for {@code ngramsSplits()} output
- */
-@Operator(group = "strings")
-public final class StringNGrams<T extends TNumber> extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new StringNGrams operation.
-   * 
-   * @param scope current scope
-   * @param data The values tensor of the ragged string tensor to make ngrams out of. Must be a
-   * 1D string tensor.
-   * @param dataSplits The splits tensor of the ragged string tensor to make ngrams out of.
-   * @param separator The string to append between elements of the token. Use "" for no separator.
-   * @param ngramWidths The sizes of the ngrams to create.
-   * @param leftPad The string to use to pad the left side of the ngram sequence. Only used if
-   * pad_width != 0.
-   * @param rightPad The string to use to pad the right side of the ngram sequence. Only used if
-   * pad_width != 0.
-   * @param padWidth The number of padding elements to add to each side of each
-   * sequence. Note that padding will never be greater than 'ngram_widths'-1
-   * regardless of this value. If `pad_width=-1`, then add `max(ngram_widths)-1`
-   * elements.
-   * @param preserveShortSequences 
-   * @return a new instance of StringNGrams
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> StringNGrams<T> create(Scope scope, Operand<TString> data, Operand<T> dataSplits, String separator, List<Long> ngramWidths, String leftPad, String rightPad, Long padWidth, Boolean preserveShortSequences) {
-    OperationBuilder opBuilder = scope.env().opBuilder("StringNGrams", scope.makeOpName("StringNGrams"));
-    opBuilder.addInput(data.asOutput(scope));
-    opBuilder.addInput(dataSplits.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("separator", separator);
-    long[] ngramWidthsArray = new long[ngramWidths.size()];
-    for (int i = 0; i < ngramWidthsArray.length; ++i) {
-      ngramWidthsArray[i] = ngramWidths.get(i);
-    }
-    opBuilder.setAttr("ngram_widths", ngramWidthsArray);
-    opBuilder.setAttr("left_pad", leftPad);
-    opBuilder.setAttr("right_pad", rightPad);
-    opBuilder.setAttr("pad_width", padWidth);
-    opBuilder.setAttr("preserve_short_sequences", preserveShortSequences);
-    return new StringNGrams<T>(opBuilder.build());
-  }
-  
-  /**
-   * The values tensor of the output ngrams ragged tensor.
-   */
-  public Output<TString> ngrams() {
-    return ngrams;
-  }
-  
-  /**
-   * The splits tensor of the output ngrams ragged tensor.
-   */
-  public Output<T> ngramsSplits() {
-    return ngramsSplits;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "StringNGrams";
-  
-  private Output<TString> ngrams;
-  private Output<T> ngramsSplits;
-  
-  private StringNGrams(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    ngrams = operation.output(outputIdx++);
-    ngramsSplits = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/StringSplit.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/StringSplit.java
deleted file mode 100644
index cc349f0aac3..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/StringSplit.java
+++ /dev/null
@@ -1,144 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.strings;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.TString;
-
-/**
- * Split elements of `source` based on `sep` into a `SparseTensor`.
- * <p>
- * Let N be the size of source (typically N will be the batch size). Split each
- * element of `source` based on `sep` and return a `SparseTensor`
- * containing the split tokens. Empty tokens are ignored.
- * <p>
- * For example, N = 2, source[0] is 'hello world' and source[1] is 'a b c',
- * then the output will be
- * <pre>{@code
- * st.indices = [0, 0;
- *               0, 1;
- *               1, 0;
- *               1, 1;
- *               1, 2]
- * st.shape = [2, 3]
- * st.values = ['hello', 'world', 'a', 'b', 'c']
- * }</pre>
- * If `sep` is given, consecutive delimiters are not grouped together and are
- * deemed to delimit empty strings. For example, source of `"1<>2<><>3"` and
- * sep of `"<>"` returns `["1", "2", "", "3"]`. If `sep` is None or an empty
- * string, consecutive whitespace are regarded as a single separator, and the
- * result will contain no empty strings at the startor end if the string has
- * leading or trailing whitespace.
- * <p>
- * Note that the above mentioned behavior matches python's str.split.
- */
-@Operator(group = "strings")
-public final class StringSplit extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.strings.StringSplit}
-   */
-  public static class Options {
-    
-    /**
-     * @param maxsplit An `int`. If `maxsplit > 0`, limit of the split of the result.
-     */
-    public Options maxsplit(Long maxsplit) {
-      this.maxsplit = maxsplit;
-      return this;
-    }
-    
-    private Long maxsplit;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new StringSplit operation.
-   * 
-   * @param scope current scope
-   * @param input `1-D` string `Tensor`, the strings to split.
-   * @param sep `0-D` string `Tensor`, the delimiter character.
-   * @param options carries optional attributes values
-   * @return a new instance of StringSplit
-   */
-  @Endpoint(describeByClass = true)
-  public static StringSplit create(Scope scope, Operand<TString> input, Operand<TString> sep, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("StringSplitV2", scope.makeOpName("StringSplit"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder.addInput(sep.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.maxsplit != null) {
-          opBuilder.setAttr("maxsplit", opts.maxsplit);
-        }
-      }
-    }
-    return new StringSplit(opBuilder.build());
-  }
-  
-  /**
-   * @param maxsplit An `int`. If `maxsplit > 0`, limit of the split of the result.
-   */
-  public static Options maxsplit(Long maxsplit) {
-    return new Options().maxsplit(maxsplit);
-  }
-  
-  /**
-   */
-  public Output<TInt64> indices() {
-    return indices;
-  }
-  
-  /**
-   */
-  public Output<TString> values() {
-    return values;
-  }
-  
-  /**
-   */
-  public Output<TInt64> shape() {
-    return shape;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "StringSplitV2";
-  
-  private Output<TInt64> indices;
-  private Output<TString> values;
-  private Output<TInt64> shape;
-  
-  private StringSplit(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    indices = operation.output(outputIdx++);
-    values = operation.output(outputIdx++);
-    shape = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/Strip.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/Strip.java
deleted file mode 100644
index 0a6abf749c0..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/Strip.java
+++ /dev/null
@@ -1,78 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.strings;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TString;
-
-/**
- * Strip leading and trailing whitespaces from the Tensor.
- */
-@Operator(group = "strings")
-public final class Strip extends RawOp implements Operand<TString> {
-  
-  /**
-   * Factory method to create a class wrapping a new Strip operation.
-   * 
-   * @param scope current scope
-   * @param input A string `Tensor` of any shape.
-   * @return a new instance of Strip
-   */
-  @Endpoint(describeByClass = true)
-  public static Strip create(Scope scope, Operand<TString> input) {
-    OperationBuilder opBuilder = scope.env().opBuilder("StringStrip", scope.makeOpName("Strip"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new Strip(opBuilder.build());
-  }
-  
-  /**
-   * A string `Tensor` of the same shape as the input.
-   * <p>
-   * Examples:
-   * <p>
-   * >>> tf.strings.strip(["\nTensorFlow", "     The python library    "]).numpy()
-   * array([b'TensorFlow', b'The python library'], dtype=object)
-   */
-  public Output<TString> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<TString> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "StringStrip";
-  
-  private Output<TString> output;
-  
-  private Strip(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/Substr.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/Substr.java
deleted file mode 100644
index ed30aef5aec..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/Substr.java
+++ /dev/null
@@ -1,196 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.strings;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TString;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Return substrings from `Tensor` of strings.
- * <p>
- * For each string in the input `Tensor`, creates a substring starting at index
- * `pos` with a total length of `len`.
- * <p>
- * If `len` defines a substring that would extend beyond the length of the input
- * string, or if `len` is negative, then as many characters as possible are used.
- * <p>
- * A negative `pos` indicates distance within the string backwards from the end.
- * <p>
- * If `pos` specifies an index which is out of range for any of the input strings,
- * then an `InvalidArgumentError` is thrown.
- * <p>
- * `pos` and `len` must have the same shape, otherwise a `ValueError` is thrown on
- * Op creation.
- * <p>
- * <i>NOTE</i>: `strings.Substr` supports broadcasting up to two dimensions. More about
- * broadcasting
- * [here](http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html)
- * <p>
- * ---
- * <p>
- * Examples
- * <p>
- * Using scalar `pos` and `len`:
- * <pre>{@code
- * input = [b'Hello', b'World']
- * position = 1
- * length = 3
- * 
- * output = [b'ell', b'orl']
- * }</pre>
- * Using `pos` and `len` with same shape as `input`:
- * <pre>{@code
- * input = [[b'ten', b'eleven', b'twelve'],
- *          [b'thirteen', b'fourteen', b'fifteen'],
- *          [b'sixteen', b'seventeen', b'eighteen']]
- * position = [[1, 2, 3],
- *             [1, 2, 3],
- *             [1, 2, 3]]
- * length =   [[2, 3, 4],
- *             [4, 3, 2],
- *             [5, 5, 5]]
- * 
- * output = [[b'en', b'eve', b'lve'],
- *           [b'hirt', b'urt', b'te'],
- *           [b'ixtee', b'vente', b'hteen']]
- * }</pre>
- * Broadcasting `pos` and `len` onto `input`:
- * <pre>{@code
- * input = [[b'ten', b'eleven', b'twelve'],
- *          [b'thirteen', b'fourteen', b'fifteen'],
- *          [b'sixteen', b'seventeen', b'eighteen'],
- *          [b'nineteen', b'twenty', b'twentyone']]
- * position = [1, 2, 3]
- * length =   [1, 2, 3]
- * 
- * output = [[b'e', b'ev', b'lve'],
- *           [b'h', b'ur', b'tee'],
- *           [b'i', b've', b'hte'],
- *           [b'i', b'en', b'nty']]
- * }</pre>
- * Broadcasting `input` onto `pos` and `len`:
- * <pre>{@code
- * input = b'thirteen'
- * position = [1, 5, 7]
- * length =   [3, 2, 1]
- * 
- * output = [b'hir', b'ee', b'n']
- * }</pre>
- * Raises:
- * <p>
- *   * `ValueError`: If the first argument cannot be converted to a
- *      Tensor of `dtype string`.
- *   * `InvalidArgumentError`: If indices are out of range.
- *   * `ValueError`: If `pos` and `len` are not the same shape.
- * 
- */
-@Operator(group = "strings")
-public final class Substr extends RawOp implements Operand<TString> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.strings.Substr}
-   */
-  public static class Options {
-    
-    /**
-     * @param unit The unit that is used to create the substring.  One of: `"BYTE"` (for
-     * defining position and length by bytes) or `"UTF8_CHAR"` (for the UTF-8
-     * encoded Unicode code points).  The default is `"BYTE"`. Results are undefined if
-     * `unit=UTF8_CHAR` and the `input` strings do not contain structurally valid
-     * UTF-8.
-     */
-    public Options unit(String unit) {
-      this.unit = unit;
-      return this;
-    }
-    
-    private String unit;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new Substr operation.
-   * 
-   * @param scope current scope
-   * @param input Tensor of strings
-   * @param pos Scalar defining the position of first character in each substring
-   * @param len Scalar defining the number of characters to include in each substring
-   * @param options carries optional attributes values
-   * @return a new instance of Substr
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> Substr create(Scope scope, Operand<TString> input, Operand<T> pos, Operand<T> len, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Substr", scope.makeOpName("Substr"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder.addInput(pos.asOutput(scope));
-    opBuilder.addInput(len.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.unit != null) {
-          opBuilder.setAttr("unit", opts.unit);
-        }
-      }
-    }
-    return new Substr(opBuilder.build());
-  }
-  
-  /**
-   * @param unit The unit that is used to create the substring.  One of: `"BYTE"` (for
-   * defining position and length by bytes) or `"UTF8_CHAR"` (for the UTF-8
-   * encoded Unicode code points).  The default is `"BYTE"`. Results are undefined if
-   * `unit=UTF8_CHAR` and the `input` strings do not contain structurally valid
-   * UTF-8.
-   */
-  public static Options unit(String unit) {
-    return new Options().unit(unit);
-  }
-  
-  /**
-   * Tensor of substrings
-   */
-  public Output<TString> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<TString> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Substr";
-  
-  private Output<TString> output;
-  
-  private Substr(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/ToHashBucket.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/ToHashBucket.java
deleted file mode 100644
index 5b0d752274f..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/ToHashBucket.java
+++ /dev/null
@@ -1,83 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.strings;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.TString;
-
-/**
- * Converts each string in the input Tensor to its hash mod by a number of buckets.
- * <p>
- * The hash function is deterministic on the content of the string within the
- * process.
- * <p>
- * Note that the hash function may change from time to time.
- * This functionality will be deprecated and it's recommended to use
- * `tf.string_to_hash_bucket_fast()` or `tf.string_to_hash_bucket_strong()`.
- */
-@Operator(group = "strings")
-public final class ToHashBucket extends RawOp implements Operand<TInt64> {
-  
-  /**
-   * Factory method to create a class wrapping a new ToHashBucket operation.
-   * 
-   * @param scope current scope
-   * @param stringTensor 
-   * @param numBuckets The number of buckets.
-   * @return a new instance of ToHashBucket
-   */
-  @Endpoint(describeByClass = true)
-  public static ToHashBucket create(Scope scope, Operand<TString> stringTensor, Long numBuckets) {
-    OperationBuilder opBuilder = scope.env().opBuilder("StringToHashBucket", scope.makeOpName("ToHashBucket"));
-    opBuilder.addInput(stringTensor.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("num_buckets", numBuckets);
-    return new ToHashBucket(opBuilder.build());
-  }
-  
-  /**
-   * A Tensor of the same shape as the input `string_tensor`.
-   */
-  public Output<TInt64> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<TInt64> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "StringToHashBucket";
-  
-  private Output<TInt64> output;
-  
-  private ToHashBucket(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/ToHashBucketFast.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/ToHashBucketFast.java
deleted file mode 100644
index 77085c6eca4..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/ToHashBucketFast.java
+++ /dev/null
@@ -1,88 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.strings;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.TString;
-
-/**
- * Converts each string in the input Tensor to its hash mod by a number of buckets.
- * <p>
- * The hash function is deterministic on the content of the string within the
- * process and will never change. However, it is not suitable for cryptography.
- * This function may be used when CPU time is scarce and inputs are trusted or
- * unimportant. There is a risk of adversaries constructing inputs that all hash
- * to the same bucket. To prevent this problem, use a strong hash function with
- * `tf.string_to_hash_bucket_strong`.
- * <p>
- * Examples:
- * <p>
- * >>> tf.strings.to_hash_bucket_fast(["Hello", "TensorFlow", "2.x"], 3).numpy()
- * array([0, 2, 2])
- */
-@Operator(group = "strings")
-public final class ToHashBucketFast extends RawOp implements Operand<TInt64> {
-  
-  /**
-   * Factory method to create a class wrapping a new ToHashBucketFast operation.
-   * 
-   * @param scope current scope
-   * @param input The strings to assign a hash bucket.
-   * @param numBuckets The number of buckets.
-   * @return a new instance of ToHashBucketFast
-   */
-  @Endpoint(describeByClass = true)
-  public static ToHashBucketFast create(Scope scope, Operand<TString> input, Long numBuckets) {
-    OperationBuilder opBuilder = scope.env().opBuilder("StringToHashBucketFast", scope.makeOpName("ToHashBucketFast"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("num_buckets", numBuckets);
-    return new ToHashBucketFast(opBuilder.build());
-  }
-  
-  /**
-   * A Tensor of the same shape as the input `string_tensor`.
-   */
-  public Output<TInt64> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<TInt64> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "StringToHashBucketFast";
-  
-  private Output<TInt64> output;
-  
-  private ToHashBucketFast(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/ToHashBucketStrong.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/ToHashBucketStrong.java
deleted file mode 100644
index ce408cde7e2..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/ToHashBucketStrong.java
+++ /dev/null
@@ -1,103 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.strings;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.TString;
-
-/**
- * Converts each string in the input Tensor to its hash mod by a number of buckets.
- * <p>
- * The hash function is deterministic on the content of the string within the
- * process. The hash function is a keyed hash function, where attribute `key`
- * defines the key of the hash function. `key` is an array of 2 elements.
- * <p>
- * A strong hash is important when inputs may be malicious, e.g. URLs with
- * additional components. Adversaries could try to make their inputs hash to the
- * same bucket for a denial-of-service attack or to skew the results. A strong
- * hash can be used to make it difficult to find inputs with a skewed hash value
- * distribution over buckets. This requires that the hash function is
- * seeded by a high-entropy (random) "key" unknown to the adversary.
- * <p>
- * The additional robustness comes at a cost of roughly 4x higher compute
- * time than `tf.string_to_hash_bucket_fast`.
- * <p>
- * Examples:
- * <p>
- * >>> tf.strings.to_hash_bucket_strong(["Hello", "TF"], 3, [1, 2]).numpy()
- * array([2, 0])
- */
-@Operator(group = "strings")
-public final class ToHashBucketStrong extends RawOp implements Operand<TInt64> {
-  
-  /**
-   * Factory method to create a class wrapping a new ToHashBucketStrong operation.
-   * 
-   * @param scope current scope
-   * @param input The strings to assign a hash bucket.
-   * @param numBuckets The number of buckets.
-   * @param key The key used to seed the hash function, passed as a list of two uint64
-   * elements.
-   * @return a new instance of ToHashBucketStrong
-   */
-  @Endpoint(describeByClass = true)
-  public static ToHashBucketStrong create(Scope scope, Operand<TString> input, Long numBuckets, List<Long> key) {
-    OperationBuilder opBuilder = scope.env().opBuilder("StringToHashBucketStrong", scope.makeOpName("ToHashBucketStrong"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("num_buckets", numBuckets);
-    long[] keyArray = new long[key.size()];
-    for (int i = 0; i < keyArray.length; ++i) {
-      keyArray[i] = key.get(i);
-    }
-    opBuilder.setAttr("key", keyArray);
-    return new ToHashBucketStrong(opBuilder.build());
-  }
-  
-  /**
-   * A Tensor of the same shape as the input `string_tensor`.
-   */
-  public Output<TInt64> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<TInt64> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "StringToHashBucketStrong";
-  
-  private Output<TInt64> output;
-  
-  private ToHashBucketStrong(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/ToNumber.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/ToNumber.java
deleted file mode 100644
index 9c70d36f10b..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/ToNumber.java
+++ /dev/null
@@ -1,101 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.strings;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.TString;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Converts each string in the input Tensor to the specified numeric type.
- * <p>
- * (Note that int32 overflow results in an error while float overflow
- * results in a rounded value.)
- * <p>
- * Example:
- * <p>
- * >>> strings = ["5.0", "3.0", "7.0"]
- * >>> tf.strings.to_number(strings)
- * <tf.Tensor: shape=(3,), dtype=float32, numpy=array([5., 3., 7.], dtype=float32)>
- * 
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator(group = "strings")
-public final class ToNumber<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new ToNumber operation.
-   * 
-   * @param scope current scope
-   * @param stringTensor 
-   * @param outType The numeric type to interpret each string in `string_tensor` as.
-   * @return a new instance of ToNumber
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> ToNumber<T> create(Scope scope, Operand<TString> stringTensor, Class<T> outType) {
-    OperationBuilder opBuilder = scope.env().opBuilder("StringToNumber", scope.makeOpName("ToNumber"));
-    opBuilder.addInput(stringTensor.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("out_type", outType);
-    return new ToNumber<T>(opBuilder.build());
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new ToNumber operation using default output types.
-   * 
-   * @param scope current scope
-   * @param stringTensor 
-   * @return a new instance of ToNumber
-   */
-  @Endpoint(describeByClass = true)
-  public static ToNumber<TFloat32> create(Scope scope, Operand<TString> stringTensor) {
-    return create(scope, stringTensor, TFloat32.class);
-  }
-  
-  /**
-   * A Tensor of the same shape as the input `string_tensor`.
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "StringToNumber";
-  
-  private Output<T> output;
-  
-  private ToNumber(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/UnicodeDecode.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/UnicodeDecode.java
deleted file mode 100644
index 2b573f03eba..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/UnicodeDecode.java
+++ /dev/null
@@ -1,212 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.strings;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.TString;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Decodes each string in `input` into a sequence of Unicode code points.
- * <p>
- * The character codepoints for all strings are returned using a single vector
- * `char_values`, with strings expanded to characters in row-major order.
- * <p>
- * The `row_splits` tensor indicates where the codepoints for
- * each input string begin and end within the `char_values` tensor.
- * In particular, the values for the `i`th
- * string (in row-major order) are stored in the slice
- * `[row_splits[i]:row_splits[i+1]]`. Thus:
- * <ul>
- * <li>
- * `char_values[row_splits[i]+j]` is the Unicode codepoint for the `j`th
- *   character in the `i`th string (in row-major order).
- * </li>
- * <li>
- * `row_splits[i+1] - row_splits[i]` is the number of characters in the `i`th
- *   string (in row-major order).
- * 
- * @param <T> data type for {@code rowSplits()} output
- */
-public final class UnicodeDecode<T extends TNumber> extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.strings.UnicodeDecode}
-   */
-  public static class Options {
-    
-    /**
-     * @param errors Error handling policy when there is invalid formatting found in the input.
-     * The value of 'strict' will cause the operation to produce a InvalidArgument
-     * error on any invalid input formatting. A value of 'replace' (the default) will
-     * cause the operation to replace any invalid formatting in the input with the
-     * `replacement_char` codepoint. A value of 'ignore' will cause the operation to
-     * skip any invalid formatting in the input and produce no corresponding output
-     * character.
-     */
-    public Options errors(String errors) {
-      this.errors = errors;
-      return this;
-    }
-    
-    /**
-     * @param replacementChar The replacement character codepoint to be used in place of any invalid
-     * formatting in the input when `errors='replace'`. Any valid unicode codepoint may
-     * be used. The default value is the default unicode replacement character is
-     * 0xFFFD or U+65533.)
-     */
-    public Options replacementChar(Long replacementChar) {
-      this.replacementChar = replacementChar;
-      return this;
-    }
-    
-    /**
-     * @param replaceControlCharacters Whether to replace the C0 control characters (00-1F) with the
-     * `replacement_char`. Default is false.
-     */
-    public Options replaceControlCharacters(Boolean replaceControlCharacters) {
-      this.replaceControlCharacters = replaceControlCharacters;
-      return this;
-    }
-    
-    private String errors;
-    private Long replacementChar;
-    private Boolean replaceControlCharacters;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new UnicodeDecode operation.
-   * 
-   * @param scope current scope
-   * @param input The text to be decoded. Can have any shape. Note that the output is flattened
-   * to a vector of char values.
-   * @param inputEncoding Text encoding of the input strings. This is any of the encodings supported
-   * by ICU ucnv algorithmic converters. Examples: `"UTF-16", "US ASCII", "UTF-8"`.
-   * @param Tsplits 
-   * @param options carries optional attributes values
-   * @return a new instance of UnicodeDecode
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> UnicodeDecode<T> create(Scope scope, Operand<TString> input, String inputEncoding, Class<T> Tsplits, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("UnicodeDecode", scope.makeOpName("UnicodeDecode"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("input_encoding", inputEncoding);
-    opBuilder.setAttr("Tsplits", Tsplits);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.errors != null) {
-          opBuilder.setAttr("errors", opts.errors);
-        }
-        if (opts.replacementChar != null) {
-          opBuilder.setAttr("replacement_char", opts.replacementChar);
-        }
-        if (opts.replaceControlCharacters != null) {
-          opBuilder.setAttr("replace_control_characters", opts.replaceControlCharacters);
-        }
-      }
-    }
-    return new UnicodeDecode<T>(opBuilder.build());
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new UnicodeDecode operation using default output types.
-   * 
-   * @param scope current scope
-   * @param input The text to be decoded. Can have any shape. Note that the output is flattened
-   * to a vector of char values.
-   * @param inputEncoding Text encoding of the input strings. This is any of the encodings supported
-   * by ICU ucnv algorithmic converters. Examples: `"UTF-16", "US ASCII", "UTF-8"`.
-   * @param options carries optional attributes values
-   * @return a new instance of UnicodeDecode
-   */
-  @Endpoint(describeByClass = true)
-  public static UnicodeDecode<TInt64> create(Scope scope, Operand<TString> input, String inputEncoding, Options... options) {
-    return create(scope, input, inputEncoding, TInt64.class, options);
-  }
-  
-  /**
-   * @param errors Error handling policy when there is invalid formatting found in the input.
-   * The value of 'strict' will cause the operation to produce a InvalidArgument
-   * error on any invalid input formatting. A value of 'replace' (the default) will
-   * cause the operation to replace any invalid formatting in the input with the
-   * `replacement_char` codepoint. A value of 'ignore' will cause the operation to
-   * skip any invalid formatting in the input and produce no corresponding output
-   * character.
-   */
-  public static Options errors(String errors) {
-    return new Options().errors(errors);
-  }
-  
-  /**
-   * @param replacementChar The replacement character codepoint to be used in place of any invalid
-   * formatting in the input when `errors='replace'`. Any valid unicode codepoint may
-   * be used. The default value is the default unicode replacement character is
-   * 0xFFFD or U+65533.)
-   */
-  public static Options replacementChar(Long replacementChar) {
-    return new Options().replacementChar(replacementChar);
-  }
-  
-  /**
-   * @param replaceControlCharacters Whether to replace the C0 control characters (00-1F) with the
-   * `replacement_char`. Default is false.
-   */
-  public static Options replaceControlCharacters(Boolean replaceControlCharacters) {
-    return new Options().replaceControlCharacters(replaceControlCharacters);
-  }
-  
-  /**
-   * A 1D int32 tensor containing the row splits.
-   */
-  public Output<T> rowSplits() {
-    return rowSplits;
-  }
-  
-  /**
-   * A 1D int32 Tensor containing the decoded codepoints.
-   */
-  public Output<TInt32> charValues() {
-    return charValues;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "UnicodeDecode";
-  
-  private Output<T> rowSplits;
-  private Output<TInt32> charValues;
-  
-  private UnicodeDecode(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    rowSplits = operation.output(outputIdx++);
-    charValues = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/UnicodeDecodeWithOffsets.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/UnicodeDecodeWithOffsets.java
deleted file mode 100644
index a13775dba39..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/UnicodeDecodeWithOffsets.java
+++ /dev/null
@@ -1,228 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.strings;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.TString;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Decodes each string in `input` into a sequence of Unicode code points.
- * <p>
- * The character codepoints for all strings are returned using a single vector
- * `char_values`, with strings expanded to characters in row-major order.
- * Similarly, the character start byte offsets are returned using a single vector
- * `char_to_byte_starts`, with strings expanded in row-major order.
- * <p>
- * The `row_splits` tensor indicates where the codepoints and start offsets for
- * each input string begin and end within the `char_values` and
- * `char_to_byte_starts` tensors.  In particular, the values for the `i`th
- * string (in row-major order) are stored in the slice
- * `[row_splits[i]:row_splits[i+1]]`. Thus:
- * <ul>
- * <li>
- * `char_values[row_splits[i]+j]` is the Unicode codepoint for the `j`th
- *   character in the `i`th string (in row-major order).
- * </li>
- * <li>
- * `char_to_bytes_starts[row_splits[i]+j]` is the start byte offset for the `j`th
- *   character in the `i`th string (in row-major order).
- * </li>
- * <li>
- * `row_splits[i+1] - row_splits[i]` is the number of characters in the `i`th
- *   string (in row-major order).
- * 
- * @param <T> data type for {@code rowSplits()} output
- */
-public final class UnicodeDecodeWithOffsets<T extends TNumber> extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.strings.UnicodeDecodeWithOffsets}
-   */
-  public static class Options {
-    
-    /**
-     * @param errors Error handling policy when there is invalid formatting found in the input.
-     * The value of 'strict' will cause the operation to produce a InvalidArgument
-     * error on any invalid input formatting. A value of 'replace' (the default) will
-     * cause the operation to replace any invalid formatting in the input with the
-     * `replacement_char` codepoint. A value of 'ignore' will cause the operation to
-     * skip any invalid formatting in the input and produce no corresponding output
-     * character.
-     */
-    public Options errors(String errors) {
-      this.errors = errors;
-      return this;
-    }
-    
-    /**
-     * @param replacementChar The replacement character codepoint to be used in place of any invalid
-     * formatting in the input when `errors='replace'`. Any valid unicode codepoint may
-     * be used. The default value is the default unicode replacement character is
-     * 0xFFFD or U+65533.)
-     */
-    public Options replacementChar(Long replacementChar) {
-      this.replacementChar = replacementChar;
-      return this;
-    }
-    
-    /**
-     * @param replaceControlCharacters Whether to replace the C0 control characters (00-1F) with the
-     * `replacement_char`. Default is false.
-     */
-    public Options replaceControlCharacters(Boolean replaceControlCharacters) {
-      this.replaceControlCharacters = replaceControlCharacters;
-      return this;
-    }
-    
-    private String errors;
-    private Long replacementChar;
-    private Boolean replaceControlCharacters;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new UnicodeDecodeWithOffsets operation.
-   * 
-   * @param scope current scope
-   * @param input The text to be decoded. Can have any shape. Note that the output is flattened
-   * to a vector of char values.
-   * @param inputEncoding Text encoding of the input strings. This is any of the encodings supported
-   * by ICU ucnv algorithmic converters. Examples: `"UTF-16", "US ASCII", "UTF-8"`.
-   * @param Tsplits 
-   * @param options carries optional attributes values
-   * @return a new instance of UnicodeDecodeWithOffsets
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> UnicodeDecodeWithOffsets<T> create(Scope scope, Operand<TString> input, String inputEncoding, Class<T> Tsplits, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("UnicodeDecodeWithOffsets", scope.makeOpName("UnicodeDecodeWithOffsets"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("input_encoding", inputEncoding);
-    opBuilder.setAttr("Tsplits", Tsplits);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.errors != null) {
-          opBuilder.setAttr("errors", opts.errors);
-        }
-        if (opts.replacementChar != null) {
-          opBuilder.setAttr("replacement_char", opts.replacementChar);
-        }
-        if (opts.replaceControlCharacters != null) {
-          opBuilder.setAttr("replace_control_characters", opts.replaceControlCharacters);
-        }
-      }
-    }
-    return new UnicodeDecodeWithOffsets<T>(opBuilder.build());
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new UnicodeDecodeWithOffsets operation using default output types.
-   * 
-   * @param scope current scope
-   * @param input The text to be decoded. Can have any shape. Note that the output is flattened
-   * to a vector of char values.
-   * @param inputEncoding Text encoding of the input strings. This is any of the encodings supported
-   * by ICU ucnv algorithmic converters. Examples: `"UTF-16", "US ASCII", "UTF-8"`.
-   * @param options carries optional attributes values
-   * @return a new instance of UnicodeDecodeWithOffsets
-   */
-  @Endpoint(describeByClass = true)
-  public static UnicodeDecodeWithOffsets<TInt64> create(Scope scope, Operand<TString> input, String inputEncoding, Options... options) {
-    return create(scope, input, inputEncoding, TInt64.class, options);
-  }
-  
-  /**
-   * @param errors Error handling policy when there is invalid formatting found in the input.
-   * The value of 'strict' will cause the operation to produce a InvalidArgument
-   * error on any invalid input formatting. A value of 'replace' (the default) will
-   * cause the operation to replace any invalid formatting in the input with the
-   * `replacement_char` codepoint. A value of 'ignore' will cause the operation to
-   * skip any invalid formatting in the input and produce no corresponding output
-   * character.
-   */
-  public static Options errors(String errors) {
-    return new Options().errors(errors);
-  }
-  
-  /**
-   * @param replacementChar The replacement character codepoint to be used in place of any invalid
-   * formatting in the input when `errors='replace'`. Any valid unicode codepoint may
-   * be used. The default value is the default unicode replacement character is
-   * 0xFFFD or U+65533.)
-   */
-  public static Options replacementChar(Long replacementChar) {
-    return new Options().replacementChar(replacementChar);
-  }
-  
-  /**
-   * @param replaceControlCharacters Whether to replace the C0 control characters (00-1F) with the
-   * `replacement_char`. Default is false.
-   */
-  public static Options replaceControlCharacters(Boolean replaceControlCharacters) {
-    return new Options().replaceControlCharacters(replaceControlCharacters);
-  }
-  
-  /**
-   * A 1D int32 tensor containing the row splits.
-   */
-  public Output<T> rowSplits() {
-    return rowSplits;
-  }
-  
-  /**
-   * A 1D int32 Tensor containing the decoded codepoints.
-   */
-  public Output<TInt32> charValues() {
-    return charValues;
-  }
-  
-  /**
-   * A 1D int32 Tensor containing the byte index in the input string where each
-   * character in `char_values` starts.
-   */
-  public Output<TInt64> charToByteStarts() {
-    return charToByteStarts;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "UnicodeDecodeWithOffsets";
-  
-  private Output<T> rowSplits;
-  private Output<TInt32> charValues;
-  private Output<TInt64> charToByteStarts;
-  
-  private UnicodeDecodeWithOffsets(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    rowSplits = operation.output(outputIdx++);
-    charValues = operation.output(outputIdx++);
-    charToByteStarts = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/UnicodeEncode.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/UnicodeEncode.java
deleted file mode 100644
index f0629e9e60a..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/UnicodeEncode.java
+++ /dev/null
@@ -1,168 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.strings;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.TString;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Encode a tensor of ints into unicode strings.
- * <p>
- * Returns a vector of strings, where `output[i]` is constructed by encoding the
- * Unicode codepoints in `input_values[input_splits[i]:input_splits[i+1]]`
- * using `output_encoding`.
- * <p>
- * ---
- * <p>
- * Example:
- * <pre>{@code
- * input_values = [72, 101, 108, 108, 111, 87, 111, 114, 108, 100]
- * input_splits = [0, 5, 10]
- * output_encoding = 'UTF-8'
- * 
- * output = ['Hello', 'World']
- * }</pre>
- * 
- */
-public final class UnicodeEncode extends RawOp implements Operand<TString> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.strings.UnicodeEncode}
-   */
-  public static class Options {
-    
-    /**
-     * @param errors Error handling policy when there is invalid formatting found in the input.
-     * The value of 'strict' will cause the operation to produce a InvalidArgument
-     * error on any invalid input formatting. A value of 'replace' (the default) will
-     * cause the operation to replace any invalid formatting in the input with the
-     * `replacement_char` codepoint. A value of 'ignore' will cause the operation to
-     * skip any invalid formatting in the input and produce no corresponding output
-     * character.
-     */
-    public Options errors(String errors) {
-      this.errors = errors;
-      return this;
-    }
-    
-    /**
-     * @param replacementChar The replacement character codepoint to be used in place of any invalid
-     * formatting in the input when `errors='replace'`. Any valid unicode codepoint may
-     * be used. The default value is the default unicode replacement character is
-     * 0xFFFD (U+65533).
-     */
-    public Options replacementChar(Long replacementChar) {
-      this.replacementChar = replacementChar;
-      return this;
-    }
-    
-    private String errors;
-    private Long replacementChar;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new UnicodeEncode operation.
-   * 
-   * @param scope current scope
-   * @param inputValues A 1D tensor containing the unicode codepoints that should be encoded.
-   * @param inputSplits A 1D tensor specifying how the unicode codepoints should be split into strings.
-   * In particular, `output[i]` is constructed by encoding the codepoints in the
-   * slice `input_values[input_splits[i]:input_splits[i+1]]`.
-   * @param outputEncoding Unicode encoding of the output strings. Valid encodings are: `"UTF-8",
-   * "UTF-16-BE", and "UTF-32-BE"`.
-   * @param options carries optional attributes values
-   * @return a new instance of UnicodeEncode
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> UnicodeEncode create(Scope scope, Operand<TInt32> inputValues, Operand<T> inputSplits, String outputEncoding, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("UnicodeEncode", scope.makeOpName("UnicodeEncode"));
-    opBuilder.addInput(inputValues.asOutput(scope));
-    opBuilder.addInput(inputSplits.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("output_encoding", outputEncoding);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.errors != null) {
-          opBuilder.setAttr("errors", opts.errors);
-        }
-        if (opts.replacementChar != null) {
-          opBuilder.setAttr("replacement_char", opts.replacementChar);
-        }
-      }
-    }
-    return new UnicodeEncode(opBuilder.build());
-  }
-  
-  /**
-   * @param errors Error handling policy when there is invalid formatting found in the input.
-   * The value of 'strict' will cause the operation to produce a InvalidArgument
-   * error on any invalid input formatting. A value of 'replace' (the default) will
-   * cause the operation to replace any invalid formatting in the input with the
-   * `replacement_char` codepoint. A value of 'ignore' will cause the operation to
-   * skip any invalid formatting in the input and produce no corresponding output
-   * character.
-   */
-  public static Options errors(String errors) {
-    return new Options().errors(errors);
-  }
-  
-  /**
-   * @param replacementChar The replacement character codepoint to be used in place of any invalid
-   * formatting in the input when `errors='replace'`. Any valid unicode codepoint may
-   * be used. The default value is the default unicode replacement character is
-   * 0xFFFD (U+65533).
-   */
-  public static Options replacementChar(Long replacementChar) {
-    return new Options().replacementChar(replacementChar);
-  }
-  
-  /**
-   * The 1-D Tensor of strings encoded from the provided unicode codepoints.
-   */
-  public Output<TString> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<TString> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "UnicodeEncode";
-  
-  private Output<TString> output;
-  
-  private UnicodeEncode(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/UnicodeScript.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/UnicodeScript.java
deleted file mode 100644
index 39b43b34d45..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/UnicodeScript.java
+++ /dev/null
@@ -1,84 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.strings;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
-
-/**
- * Determine the script codes of a given tensor of Unicode integer code points.
- * <p>
- * This operation converts Unicode code points to script codes corresponding to
- * each code point. Script codes correspond to International Components for
- * Unicode (ICU) UScriptCode values. See http://icu-project.org/apiref/icu4c/uscript_8h.html.
- * Returns -1 (USCRIPT_INVALID_CODE) for invalid codepoints. Output shape will
- * match input shape.
- * <p>
- * Examples:
- * <p>
- * >>> tf.strings.unicode_script([1, 31, 38])
- * <tf.Tensor: shape=(3,), dtype=int32, numpy=array([0, 0, 0], dtype=int32)>
- */
-@Operator(group = "strings")
-public final class UnicodeScript extends RawOp implements Operand<TInt32> {
-  
-  /**
-   * Factory method to create a class wrapping a new UnicodeScript operation.
-   * 
-   * @param scope current scope
-   * @param input A Tensor of int32 Unicode code points.
-   * @return a new instance of UnicodeScript
-   */
-  @Endpoint(describeByClass = true)
-  public static UnicodeScript create(Scope scope, Operand<TInt32> input) {
-    OperationBuilder opBuilder = scope.env().opBuilder("UnicodeScript", scope.makeOpName("UnicodeScript"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new UnicodeScript(opBuilder.build());
-  }
-  
-  /**
-   * A Tensor of int32 script codes corresponding to each input code point.
-   */
-  public Output<TInt32> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<TInt32> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "UnicodeScript";
-  
-  private Output<TInt32> output;
-  
-  private UnicodeScript(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/UnicodeTranscode.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/UnicodeTranscode.java
deleted file mode 100644
index 8d666a585f1..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/UnicodeTranscode.java
+++ /dev/null
@@ -1,216 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.strings;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TString;
-
-/**
- * Transcode the input text from a source encoding to a destination encoding.
- * <p>
- * The input is a string tensor of any shape. The output is a string tensor of
- * the same shape containing the transcoded strings. Output strings are always
- * valid unicode. If the input contains invalid encoding positions, the
- * `errors` attribute sets the policy for how to deal with them. If the default
- * error-handling policy is used, invalid formatting will be substituted in the
- * output by the `replacement_char`. If the errors policy is to `ignore`, any
- * invalid encoding positions in the input are skipped and not included in the
- * output. If it set to `strict` then any invalid formatting will result in an
- * InvalidArgument error.
- * <p>
- * This operation can be used with `output_encoding = input_encoding` to enforce
- * correct formatting for inputs even if they are already in the desired encoding.
- * <p>
- * If the input is prefixed by a Byte Order Mark needed to determine encoding
- * (e.g. if the encoding is UTF-16 and the BOM indicates big-endian), then that
- * BOM will be consumed and not emitted into the output. If the input encoding
- * is marked with an explicit endianness (e.g. UTF-16-BE), then the BOM is
- * interpreted as a non-breaking-space and is preserved in the output (including
- * always for UTF-8).
- * <p>
- * The end result is that if the input is marked as an explicit endianness the
- * transcoding is faithful to all codepoints in the source. If it is not marked
- * with an explicit endianness, the BOM is not considered part of the string itself
- * but as metadata, and so is not preserved in the output.
- * <p>
- * Examples:
- * <p>
- * >>> tf.strings.unicode_transcode(["Hello", "TensorFlow", "2.x"], "UTF-8", "UTF-16-BE")
- * <tf.Tensor: shape=(3,), dtype=string, numpy=
- * array([b'\x00H\x00e\x00l\x00l\x00o',
- *        b'\x00T\x00e\x00n\x00s\x00o\x00r\x00F\x00l\x00o\x00w',
- *        b'\x002\x00.\x00x'], dtype=object)>
- * >>> tf.strings.unicode_transcode(["A", "B", "C"], "US ASCII", "UTF-8").numpy()
- * array([b'A', b'B', b'C'], dtype=object)
- */
-@Operator(group = "strings")
-public final class UnicodeTranscode extends RawOp implements Operand<TString> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.strings.UnicodeTranscode}
-   */
-  public static class Options {
-    
-    /**
-     * @param errors Error handling policy when there is invalid formatting found in the input.
-     * The value of 'strict' will cause the operation to produce a InvalidArgument
-     * error on any invalid input formatting. A value of 'replace' (the default) will
-     * cause the operation to replace any invalid formatting in the input with the
-     * `replacement_char` codepoint. A value of 'ignore' will cause the operation to
-     * skip any invalid formatting in the input and produce no corresponding output
-     * character.
-     */
-    public Options errors(String errors) {
-      this.errors = errors;
-      return this;
-    }
-    
-    /**
-     * @param replacementChar The replacement character codepoint to be used in place of any invalid
-     * formatting in the input when `errors='replace'`. Any valid unicode codepoint may
-     * be used. The default value is the default unicode replacement character is
-     * 0xFFFD or U+65533.)
-     * <p>
-     * Note that for UTF-8, passing a replacement character expressible in 1 byte, such
-     * as ' ', will preserve string alignment to the source since invalid bytes will be
-     * replaced with a 1-byte replacement. For UTF-16-BE and UTF-16-LE, any 1 or 2 byte
-     * replacement character will preserve byte alignment to the source.
-     */
-    public Options replacementChar(Long replacementChar) {
-      this.replacementChar = replacementChar;
-      return this;
-    }
-    
-    /**
-     * @param replaceControlCharacters Whether to replace the C0 control characters (00-1F) with the
-     * `replacement_char`. Default is false.
-     */
-    public Options replaceControlCharacters(Boolean replaceControlCharacters) {
-      this.replaceControlCharacters = replaceControlCharacters;
-      return this;
-    }
-    
-    private String errors;
-    private Long replacementChar;
-    private Boolean replaceControlCharacters;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new UnicodeTranscode operation.
-   * 
-   * @param scope current scope
-   * @param input The text to be processed. Can have any shape.
-   * @param inputEncoding Text encoding of the input strings. This is any of the encodings supported
-   * by ICU ucnv algorithmic converters. Examples: `"UTF-16", "US ASCII", "UTF-8"`.
-   * @param outputEncoding The unicode encoding to use in the output. Must be one of
-   * `"UTF-8", "UTF-16-BE", "UTF-32-BE"`. Multi-byte encodings will be big-endian.
-   * @param options carries optional attributes values
-   * @return a new instance of UnicodeTranscode
-   */
-  @Endpoint(describeByClass = true)
-  public static UnicodeTranscode create(Scope scope, Operand<TString> input, String inputEncoding, String outputEncoding, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("UnicodeTranscode", scope.makeOpName("UnicodeTranscode"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("input_encoding", inputEncoding);
-    opBuilder.setAttr("output_encoding", outputEncoding);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.errors != null) {
-          opBuilder.setAttr("errors", opts.errors);
-        }
-        if (opts.replacementChar != null) {
-          opBuilder.setAttr("replacement_char", opts.replacementChar);
-        }
-        if (opts.replaceControlCharacters != null) {
-          opBuilder.setAttr("replace_control_characters", opts.replaceControlCharacters);
-        }
-      }
-    }
-    return new UnicodeTranscode(opBuilder.build());
-  }
-  
-  /**
-   * @param errors Error handling policy when there is invalid formatting found in the input.
-   * The value of 'strict' will cause the operation to produce a InvalidArgument
-   * error on any invalid input formatting. A value of 'replace' (the default) will
-   * cause the operation to replace any invalid formatting in the input with the
-   * `replacement_char` codepoint. A value of 'ignore' will cause the operation to
-   * skip any invalid formatting in the input and produce no corresponding output
-   * character.
-   */
-  public static Options errors(String errors) {
-    return new Options().errors(errors);
-  }
-  
-  /**
-   * @param replacementChar The replacement character codepoint to be used in place of any invalid
-   * formatting in the input when `errors='replace'`. Any valid unicode codepoint may
-   * be used. The default value is the default unicode replacement character is
-   * 0xFFFD or U+65533.)
-   * <p>
-   * Note that for UTF-8, passing a replacement character expressible in 1 byte, such
-   * as ' ', will preserve string alignment to the source since invalid bytes will be
-   * replaced with a 1-byte replacement. For UTF-16-BE and UTF-16-LE, any 1 or 2 byte
-   * replacement character will preserve byte alignment to the source.
-   */
-  public static Options replacementChar(Long replacementChar) {
-    return new Options().replacementChar(replacementChar);
-  }
-  
-  /**
-   * @param replaceControlCharacters Whether to replace the C0 control characters (00-1F) with the
-   * `replacement_char`. Default is false.
-   */
-  public static Options replaceControlCharacters(Boolean replaceControlCharacters) {
-    return new Options().replaceControlCharacters(replaceControlCharacters);
-  }
-  
-  /**
-   * A string tensor containing unicode text encoded using `output_encoding`.
-   */
-  public Output<TString> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<TString> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "UnicodeTranscode";
-  
-  private Output<TString> output;
-  
-  private UnicodeTranscode(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/UnsortedSegmentJoin.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/UnsortedSegmentJoin.java
deleted file mode 100644
index 150511e769d..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/UnsortedSegmentJoin.java
+++ /dev/null
@@ -1,139 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.strings;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TString;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Joins the elements of `inputs` based on `segment_ids`.
- * <p>
- * Computes the string join along segments of a tensor.
- * Given `segment_ids` with rank `N` and `data` with rank `N+M`:
- * <p>
- *     `output[i, k1...kM] = strings.join([data[j1...jN, k1...kM])`
- * <p>
- * where the join is over all [j1...jN] such that segment_ids[j1...jN] = i.
- * Strings are joined in row-major order.
- * <p>
- * For example:
- * <pre>{@code
- * inputs = [['Y', 'q', 'c'], ['Y', '6', '6'], ['p', 'G', 'a']]
- * output_array = string_ops.unsorted_segment_join(inputs=inputs,
- *                                                 segment_ids=[1, 0, 1],
- *                                                 num_segments=2,
- *                                                 separator=':'))
- * # output_array ==> [['Y', '6', '6'], ['Y:p', 'q:G', 'c:a']]
- * 
- * 
- * inputs = ['this', 'is', 'a', 'test']
- * output_array = string_ops.unsorted_segment_join(inputs=inputs,
- *                                                 segment_ids=[0, 0, 0, 0],
- *                                                 num_segments=1,
- *                                                 separator=':'))
- * # output_array ==> ['this:is:a:test']
- * }</pre>
- * 
- */
-@Operator(group = "strings")
-public final class UnsortedSegmentJoin extends RawOp implements Operand<TString> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.strings.UnsortedSegmentJoin}
-   */
-  public static class Options {
-    
-    /**
-     * @param separator The separator to use when joining.
-     */
-    public Options separator(String separator) {
-      this.separator = separator;
-      return this;
-    }
-    
-    private String separator;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new UnsortedSegmentJoin operation.
-   * 
-   * @param scope current scope
-   * @param inputs The input to be joined.
-   * @param segmentIds A tensor whose shape is a prefix of data.shape.  Negative segment ids are not
-   * supported.
-   * @param numSegments A scalar.
-   * @param options carries optional attributes values
-   * @return a new instance of UnsortedSegmentJoin
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber, U extends TNumber> UnsortedSegmentJoin create(Scope scope, Operand<TString> inputs, Operand<T> segmentIds, Operand<U> numSegments, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("UnsortedSegmentJoin", scope.makeOpName("UnsortedSegmentJoin"));
-    opBuilder.addInput(inputs.asOutput(scope));
-    opBuilder.addInput(segmentIds.asOutput(scope));
-    opBuilder.addInput(numSegments.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.separator != null) {
-          opBuilder.setAttr("separator", opts.separator);
-        }
-      }
-    }
-    return new UnsortedSegmentJoin(opBuilder.build());
-  }
-  
-  /**
-   * @param separator The separator to use when joining.
-   */
-  public static Options separator(String separator) {
-    return new Options().separator(separator);
-  }
-  
-  /**
-   */
-  public Output<TString> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<TString> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "UnsortedSegmentJoin";
-  
-  private Output<TString> output;
-  
-  private UnsortedSegmentJoin(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/Upper.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/Upper.java
deleted file mode 100644
index 44443dca328..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/Upper.java
+++ /dev/null
@@ -1,112 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.strings;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TString;
-
-/**
- * Converts all lowercase characters into their respective uppercase replacements.
- * <p>
- * Example:
- * <p>
- * >>> tf.strings.upper("CamelCase string and ALL CAPS")
- * <tf.Tensor: shape=(), dtype=string, numpy=b'CAMELCASE STRING AND ALL CAPS'>
- * 
- */
-@Operator(group = "strings")
-public final class Upper extends RawOp implements Operand<TString> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.strings.Upper}
-   */
-  public static class Options {
-    
-    /**
-     * @param encoding 
-     */
-    public Options encoding(String encoding) {
-      this.encoding = encoding;
-      return this;
-    }
-    
-    private String encoding;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new Upper operation.
-   * 
-   * @param scope current scope
-   * @param input 
-   * @param options carries optional attributes values
-   * @return a new instance of Upper
-   */
-  @Endpoint(describeByClass = true)
-  public static Upper create(Scope scope, Operand<TString> input, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("StringUpper", scope.makeOpName("Upper"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.encoding != null) {
-          opBuilder.setAttr("encoding", opts.encoding);
-        }
-      }
-    }
-    return new Upper(opBuilder.build());
-  }
-  
-  /**
-   * @param encoding 
-   */
-  public static Options encoding(String encoding) {
-    return new Options().encoding(encoding);
-  }
-  
-  /**
-   */
-  public Output<TString> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<TString> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "StringUpper";
-  
-  private Output<TString> output;
-  
-  private Upper(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/AudioSummary.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/AudioSummary.java
deleted file mode 100644
index 11d0a8cdf6a..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/AudioSummary.java
+++ /dev/null
@@ -1,127 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.summary;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.TString;
-
-/**
- * Outputs a `Summary` protocol buffer with audio.
- * <p>
- * The summary has up to `max_outputs` summary values containing audio. The
- * audio is built from `tensor` which must be 3-D with shape `[batch_size,
- * frames, channels]` or 2-D with shape `[batch_size, frames]`. The values are
- * assumed to be in the range of `[-1.0, 1.0]` with a sample rate of `sample_rate`.
- * <p>
- * The `tag` argument is a scalar `Tensor` of type `string`.  It is used to
- * build the `tag` of the summary values:
- * <ul>
- * <li>
- * If `max_outputs` is 1, the summary value tag is '<i>tag</i>/audio'.
- * </li>
- * <li>
- * If `max_outputs` is greater than 1, the summary value tags are
- *    generated sequentially as '<i>tag</i>/audio/0', '<i>tag</i>/audio/1', etc.
- */
-@Operator(group = "summary")
-public final class AudioSummary extends RawOp implements Operand<TString> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.summary.AudioSummary}
-   */
-  public static class Options {
-    
-    /**
-     * @param maxOutputs Max number of batch elements to generate audio for.
-     */
-    public Options maxOutputs(Long maxOutputs) {
-      this.maxOutputs = maxOutputs;
-      return this;
-    }
-    
-    private Long maxOutputs;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new AudioSummary operation.
-   * 
-   * @param scope current scope
-   * @param tag Scalar. Used to build the `tag` attribute of the summary values.
-   * @param tensor 2-D of shape `[batch_size, frames]`.
-   * @param sampleRate The sample rate of the signal in hertz.
-   * @param options carries optional attributes values
-   * @return a new instance of AudioSummary
-   */
-  @Endpoint(describeByClass = true)
-  public static AudioSummary create(Scope scope, Operand<TString> tag, Operand<TFloat32> tensor, Operand<TFloat32> sampleRate, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("AudioSummaryV2", scope.makeOpName("AudioSummary"));
-    opBuilder.addInput(tag.asOutput(scope));
-    opBuilder.addInput(tensor.asOutput(scope));
-    opBuilder.addInput(sampleRate.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.maxOutputs != null) {
-          opBuilder.setAttr("max_outputs", opts.maxOutputs);
-        }
-      }
-    }
-    return new AudioSummary(opBuilder.build());
-  }
-  
-  /**
-   * @param maxOutputs Max number of batch elements to generate audio for.
-   */
-  public static Options maxOutputs(Long maxOutputs) {
-    return new Options().maxOutputs(maxOutputs);
-  }
-  
-  /**
-   * Scalar. Serialized `Summary` protocol buffer.
-   */
-  public Output<TString> summary() {
-    return summary;
-  }
-  
-  @Override
-  public Output<TString> asOutput(Scope scope) {
-    return summary;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "AudioSummaryV2";
-  
-  private Output<TString> summary;
-  
-  private AudioSummary(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    summary = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/CloseSummaryWriter.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/CloseSummaryWriter.java
deleted file mode 100644
index defd036d9ea..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/CloseSummaryWriter.java
+++ /dev/null
@@ -1,53 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.summary;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-
-/**
- */
-public final class CloseSummaryWriter extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new CloseSummaryWriter operation.
-   * 
-   * @param scope current scope
-   * @param writer 
-   * @return a new instance of CloseSummaryWriter
-   */
-  @Endpoint(describeByClass = true)
-  public static CloseSummaryWriter create(Scope scope, Operand<?> writer) {
-    OperationBuilder opBuilder = scope.env().opBuilder("CloseSummaryWriter", scope.makeOpName("CloseSummaryWriter"));
-    opBuilder.addInput(writer.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new CloseSummaryWriter(opBuilder.build());
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "CloseSummaryWriter";
-  
-  private CloseSummaryWriter(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/CreateSummaryDbWriter.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/CreateSummaryDbWriter.java
deleted file mode 100644
index 16dd23622d5..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/CreateSummaryDbWriter.java
+++ /dev/null
@@ -1,62 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.summary;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TString;
-
-/**
- */
-public final class CreateSummaryDbWriter extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new CreateSummaryDbWriter operation.
-   * 
-   * @param scope current scope
-   * @param writer 
-   * @param dbUri 
-   * @param experimentName 
-   * @param runName 
-   * @param userName 
-   * @return a new instance of CreateSummaryDbWriter
-   */
-  @Endpoint(describeByClass = true)
-  public static CreateSummaryDbWriter create(Scope scope, Operand<?> writer, Operand<TString> dbUri, Operand<TString> experimentName, Operand<TString> runName, Operand<TString> userName) {
-    OperationBuilder opBuilder = scope.env().opBuilder("CreateSummaryDbWriter", scope.makeOpName("CreateSummaryDbWriter"));
-    opBuilder.addInput(writer.asOutput(scope));
-    opBuilder.addInput(dbUri.asOutput(scope));
-    opBuilder.addInput(experimentName.asOutput(scope));
-    opBuilder.addInput(runName.asOutput(scope));
-    opBuilder.addInput(userName.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new CreateSummaryDbWriter(opBuilder.build());
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "CreateSummaryDbWriter";
-  
-  private CreateSummaryDbWriter(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/CreateSummaryFileWriter.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/CreateSummaryFileWriter.java
deleted file mode 100644
index a9fd623e23d..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/CreateSummaryFileWriter.java
+++ /dev/null
@@ -1,63 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.summary;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.TString;
-
-/**
- */
-public final class CreateSummaryFileWriter extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new CreateSummaryFileWriter operation.
-   * 
-   * @param scope current scope
-   * @param writer 
-   * @param logdir 
-   * @param maxQueue 
-   * @param flushMillis 
-   * @param filenameSuffix 
-   * @return a new instance of CreateSummaryFileWriter
-   */
-  @Endpoint(describeByClass = true)
-  public static CreateSummaryFileWriter create(Scope scope, Operand<?> writer, Operand<TString> logdir, Operand<TInt32> maxQueue, Operand<TInt32> flushMillis, Operand<TString> filenameSuffix) {
-    OperationBuilder opBuilder = scope.env().opBuilder("CreateSummaryFileWriter", scope.makeOpName("CreateSummaryFileWriter"));
-    opBuilder.addInput(writer.asOutput(scope));
-    opBuilder.addInput(logdir.asOutput(scope));
-    opBuilder.addInput(maxQueue.asOutput(scope));
-    opBuilder.addInput(flushMillis.asOutput(scope));
-    opBuilder.addInput(filenameSuffix.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new CreateSummaryFileWriter(opBuilder.build());
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "CreateSummaryFileWriter";
-  
-  private CreateSummaryFileWriter(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/FlushSummaryWriter.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/FlushSummaryWriter.java
deleted file mode 100644
index 05040764d4e..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/FlushSummaryWriter.java
+++ /dev/null
@@ -1,53 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.summary;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-
-/**
- */
-public final class FlushSummaryWriter extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new FlushSummaryWriter operation.
-   * 
-   * @param scope current scope
-   * @param writer 
-   * @return a new instance of FlushSummaryWriter
-   */
-  @Endpoint(describeByClass = true)
-  public static FlushSummaryWriter create(Scope scope, Operand<?> writer) {
-    OperationBuilder opBuilder = scope.env().opBuilder("FlushSummaryWriter", scope.makeOpName("FlushSummaryWriter"));
-    opBuilder.addInput(writer.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new FlushSummaryWriter(opBuilder.build());
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "FlushSummaryWriter";
-  
-  private FlushSummaryWriter(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/HistogramSummary.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/HistogramSummary.java
deleted file mode 100644
index 271c2793c42..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/HistogramSummary.java
+++ /dev/null
@@ -1,82 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.summary;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TString;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Outputs a `Summary` protocol buffer with a histogram.
- * <p>
- * The generated
- * [`Summary`](https://www.tensorflow.org/code/tensorflow/core/framework/summary.proto)
- * has one summary value containing a histogram for `values`.
- * <p>
- * This op reports an `InvalidArgument` error if any value is not finite.
- */
-@Operator(group = "summary")
-public final class HistogramSummary extends RawOp implements Operand<TString> {
-  
-  /**
-   * Factory method to create a class wrapping a new HistogramSummary operation.
-   * 
-   * @param scope current scope
-   * @param tag Scalar.  Tag to use for the `Summary.Value`.
-   * @param values Any shape. Values to use to build the histogram.
-   * @return a new instance of HistogramSummary
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> HistogramSummary create(Scope scope, Operand<TString> tag, Operand<T> values) {
-    OperationBuilder opBuilder = scope.env().opBuilder("HistogramSummary", scope.makeOpName("HistogramSummary"));
-    opBuilder.addInput(tag.asOutput(scope));
-    opBuilder.addInput(values.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new HistogramSummary(opBuilder.build());
-  }
-  
-  /**
-   * Scalar. Serialized `Summary` protocol buffer.
-   */
-  public Output<TString> summary() {
-    return summary;
-  }
-  
-  @Override
-  public Output<TString> asOutput(Scope scope) {
-    return summary;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "HistogramSummary";
-  
-  private Output<TString> summary;
-  
-  private HistogramSummary(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    summary = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/ImageSummary.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/ImageSummary.java
deleted file mode 100644
index 188074c9899..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/ImageSummary.java
+++ /dev/null
@@ -1,178 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.summary;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.Tensor;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TString;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Outputs a `Summary` protocol buffer with images.
- * <p>
- * The summary has up to `max_images` summary values containing images. The
- * images are built from `tensor` which must be 4-D with shape `[batch_size,
- * height, width, channels]` and where `channels` can be:
- * <ul>
- * <li>
- * 1: `tensor` is interpreted as Grayscale.
- * </li>
- * <li>
- * 3: `tensor` is interpreted as RGB.
- * </li>
- * <li>
- * 4: `tensor` is interpreted as RGBA.
- * </li>
- * </ul>
- * The images have the same number of channels as the input tensor. For float
- * input, the values are normalized one image at a time to fit in the range
- * `[0, 255]`.  `uint8` values are unchanged.  The op uses two different
- * normalization algorithms:
- * <ul>
- * <li>
- * If the input values are all positive, they are rescaled so the largest one
- *    is 255.
- * </li>
- * <li>
- * If any input value is negative, the values are shifted so input value 0.0
- *    is at 127.  They are then rescaled so that either the smallest value is 0,
- *    or the largest one is 255.
- * </li>
- * </ul>
- * The `tag` argument is a scalar `Tensor` of type `string`.  It is used to
- * build the `tag` of the summary values:
- * <ul>
- * <li>
- * If `max_images` is 1, the summary value tag is '<i>tag</i>/image'.
- * </li>
- * <li>
- * If `max_images` is greater than 1, the summary value tags are
- *    generated sequentially as '<i>tag</i>/image/0', '<i>tag</i>/image/1', etc.
- * </li>
- * </ul>
- * The `bad_color` argument is the color to use in the generated images for
- * non-finite input values.  It is a `uint8` 1-D tensor of length `channels`.
- * Each element must be in the range `[0, 255]` (It represents the value of a
- * pixel in the output image).  Non-finite values in the input tensor are
- * replaced by this tensor in the output image.  The default value is the color
- * red.
- */
-@Operator(group = "summary")
-public final class ImageSummary extends RawOp implements Operand<TString> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.summary.ImageSummary}
-   */
-  public static class Options {
-    
-    /**
-     * @param maxImages Max number of batch elements to generate images for.
-     */
-    public Options maxImages(Long maxImages) {
-      this.maxImages = maxImages;
-      return this;
-    }
-    
-    /**
-     * @param badColor Color to use for pixels with non-finite values.
-     */
-    public Options badColor(Tensor<?> badColor) {
-      this.badColor = badColor;
-      return this;
-    }
-    
-    private Long maxImages;
-    private Tensor<?> badColor;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new ImageSummary operation.
-   * 
-   * @param scope current scope
-   * @param tag Scalar. Used to build the `tag` attribute of the summary values.
-   * @param tensor 4-D of shape `[batch_size, height, width, channels]` where
-   * `channels` is 1, 3, or 4.
-   * @param options carries optional attributes values
-   * @return a new instance of ImageSummary
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> ImageSummary create(Scope scope, Operand<TString> tag, Operand<T> tensor, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ImageSummary", scope.makeOpName("ImageSummary"));
-    opBuilder.addInput(tag.asOutput(scope));
-    opBuilder.addInput(tensor.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.maxImages != null) {
-          opBuilder.setAttr("max_images", opts.maxImages);
-        }
-        if (opts.badColor != null) {
-          opBuilder.setAttr("bad_color", opts.badColor);
-        }
-      }
-    }
-    return new ImageSummary(opBuilder.build());
-  }
-  
-  /**
-   * @param maxImages Max number of batch elements to generate images for.
-   */
-  public static Options maxImages(Long maxImages) {
-    return new Options().maxImages(maxImages);
-  }
-  
-  /**
-   * @param badColor Color to use for pixels with non-finite values.
-   */
-  public static Options badColor(Tensor<?> badColor) {
-    return new Options().badColor(badColor);
-  }
-  
-  /**
-   * Scalar. Serialized `Summary` protocol buffer.
-   */
-  public Output<TString> summary() {
-    return summary;
-  }
-  
-  @Override
-  public Output<TString> asOutput(Scope scope) {
-    return summary;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ImageSummary";
-  
-  private Output<TString> summary;
-  
-  private ImageSummary(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    summary = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/ImportEvent.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/ImportEvent.java
deleted file mode 100644
index 0a3276f246a..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/ImportEvent.java
+++ /dev/null
@@ -1,56 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.summary;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TString;
-
-/**
- */
-public final class ImportEvent extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new ImportEvent operation.
-   * 
-   * @param scope current scope
-   * @param writer 
-   * @param event 
-   * @return a new instance of ImportEvent
-   */
-  @Endpoint(describeByClass = true)
-  public static ImportEvent create(Scope scope, Operand<?> writer, Operand<TString> event) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ImportEvent", scope.makeOpName("ImportEvent"));
-    opBuilder.addInput(writer.asOutput(scope));
-    opBuilder.addInput(event.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new ImportEvent(opBuilder.build());
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ImportEvent";
-  
-  private ImportEvent(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/MergeSummary.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/MergeSummary.java
deleted file mode 100644
index d561f574d8f..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/MergeSummary.java
+++ /dev/null
@@ -1,83 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.summary;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.Operands;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TString;
-
-/**
- * Merges summaries.
- * <p>
- * This op creates a
- * [`Summary`](https://www.tensorflow.org/code/tensorflow/core/framework/summary.proto)
- * protocol buffer that contains the union of all the values in the input
- * summaries.
- * <p>
- * When the Op is run, it reports an `InvalidArgument` error if multiple values
- * in the summaries to merge use the same tag.
- */
-@Operator(group = "summary")
-public final class MergeSummary extends RawOp implements Operand<TString> {
-  
-  /**
-   * Factory method to create a class wrapping a new MergeSummary operation.
-   * 
-   * @param scope current scope
-   * @param inputs Can be of any shape.  Each must contain serialized `Summary` protocol
-   * buffers.
-   * @return a new instance of MergeSummary
-   */
-  @Endpoint(describeByClass = true)
-  public static MergeSummary create(Scope scope, Iterable<Operand<TString>> inputs) {
-    OperationBuilder opBuilder = scope.env().opBuilder("MergeSummary", scope.makeOpName("MergeSummary"));
-    opBuilder.addInputList(Operands.asOutputs(scope, inputs));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new MergeSummary(opBuilder.build());
-  }
-  
-  /**
-   * Scalar. Serialized `Summary` protocol buffer.
-   */
-  public Output<TString> summary() {
-    return summary;
-  }
-  
-  @Override
-  public Output<TString> asOutput(Scope scope) {
-    return summary;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "MergeSummary";
-  
-  private Output<TString> summary;
-  
-  private MergeSummary(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    summary = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/ScalarSummary.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/ScalarSummary.java
deleted file mode 100644
index 5e8f8ad82e2..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/ScalarSummary.java
+++ /dev/null
@@ -1,79 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.summary;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TString;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Outputs a `Summary` protocol buffer with scalar values.
- * <p>
- * The input `tags` and `values` must have the same shape.  The generated summary
- * has a summary value for each tag-value pair in `tags` and `values`.
- */
-@Operator(group = "summary")
-public final class ScalarSummary extends RawOp implements Operand<TString> {
-  
-  /**
-   * Factory method to create a class wrapping a new ScalarSummary operation.
-   * 
-   * @param scope current scope
-   * @param tags Tags for the summary.
-   * @param values Same shape as `tags.  Values for the summary.
-   * @return a new instance of ScalarSummary
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> ScalarSummary create(Scope scope, Operand<TString> tags, Operand<T> values) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ScalarSummary", scope.makeOpName("ScalarSummary"));
-    opBuilder.addInput(tags.asOutput(scope));
-    opBuilder.addInput(values.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new ScalarSummary(opBuilder.build());
-  }
-  
-  /**
-   * Scalar.  Serialized `Summary` protocol buffer.
-   */
-  public Output<TString> summary() {
-    return summary;
-  }
-  
-  @Override
-  public Output<TString> asOutput(Scope scope) {
-    return summary;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ScalarSummary";
-  
-  private Output<TString> summary;
-  
-  private ScalarSummary(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    summary = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/StatsAggregatorSummary.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/StatsAggregatorSummary.java
deleted file mode 100644
index 576487c0972..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/StatsAggregatorSummary.java
+++ /dev/null
@@ -1,71 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.summary;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TString;
-
-/**
- * Produces a summary of any statistics recorded by the given statistics manager.
- */
-public final class StatsAggregatorSummary extends RawOp implements Operand<TString> {
-  
-  /**
-   * Factory method to create a class wrapping a new StatsAggregatorSummary operation.
-   * 
-   * @param scope current scope
-   * @param iterator 
-   * @return a new instance of StatsAggregatorSummary
-   */
-  @Endpoint(describeByClass = true)
-  public static StatsAggregatorSummary create(Scope scope, Operand<?> iterator) {
-    OperationBuilder opBuilder = scope.env().opBuilder("StatsAggregatorSummary", scope.makeOpName("StatsAggregatorSummary"));
-    opBuilder.addInput(iterator.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new StatsAggregatorSummary(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<TString> summary() {
-    return summary;
-  }
-  
-  @Override
-  public Output<TString> asOutput(Scope scope) {
-    return summary;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "StatsAggregatorSummary";
-  
-  private Output<TString> summary;
-  
-  private StatsAggregatorSummary(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    summary = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/SummaryWriter.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/SummaryWriter.java
deleted file mode 100644
index 641683d7d38..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/SummaryWriter.java
+++ /dev/null
@@ -1,122 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.summary;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- */
-public final class SummaryWriter extends RawOp implements Operand<TType> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.summary.SummaryWriter}
-   */
-  public static class Options {
-    
-    /**
-     * @param sharedName 
-     */
-    public Options sharedName(String sharedName) {
-      this.sharedName = sharedName;
-      return this;
-    }
-    
-    /**
-     * @param container 
-     */
-    public Options container(String container) {
-      this.container = container;
-      return this;
-    }
-    
-    private String sharedName;
-    private String container;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new SummaryWriter operation.
-   * 
-   * @param scope current scope
-   * @param options carries optional attributes values
-   * @return a new instance of SummaryWriter
-   */
-  @Endpoint(describeByClass = true)
-  public static SummaryWriter create(Scope scope, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("SummaryWriter", scope.makeOpName("SummaryWriter"));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.sharedName != null) {
-          opBuilder.setAttr("shared_name", opts.sharedName);
-        }
-        if (opts.container != null) {
-          opBuilder.setAttr("container", opts.container);
-        }
-      }
-    }
-    return new SummaryWriter(opBuilder.build());
-  }
-  
-  /**
-   * @param sharedName 
-   */
-  public static Options sharedName(String sharedName) {
-    return new Options().sharedName(sharedName);
-  }
-  
-  /**
-   * @param container 
-   */
-  public static Options container(String container) {
-    return new Options().container(container);
-  }
-  
-  /**
-   */
-  public Output<?> writer() {
-    return writer;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) writer;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "SummaryWriter";
-  
-  private Output<?> writer;
-  
-  private SummaryWriter(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    writer = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/TensorSummary.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/TensorSummary.java
deleted file mode 100644
index 7bb21e42e5f..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/TensorSummary.java
+++ /dev/null
@@ -1,78 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.summary;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TString;
-import org.tensorflow.types.family.TType;
-
-/**
- * Outputs a `Summary` protocol buffer with a tensor and per-plugin data.
- */
-@Operator(group = "summary")
-public final class TensorSummary extends RawOp implements Operand<TString> {
-  
-  /**
-   * Factory method to create a class wrapping a new TensorSummary operation.
-   * 
-   * @param scope current scope
-   * @param tag A string attached to this summary. Used for organization in TensorBoard.
-   * @param tensor A tensor to serialize.
-   * @param serializedSummaryMetadata A serialized SummaryMetadata proto. Contains plugin
-   * data.
-   * @return a new instance of TensorSummary
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> TensorSummary create(Scope scope, Operand<TString> tag, Operand<T> tensor, Operand<TString> serializedSummaryMetadata) {
-    OperationBuilder opBuilder = scope.env().opBuilder("TensorSummaryV2", scope.makeOpName("TensorSummary"));
-    opBuilder.addInput(tag.asOutput(scope));
-    opBuilder.addInput(tensor.asOutput(scope));
-    opBuilder.addInput(serializedSummaryMetadata.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new TensorSummary(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<TString> summary() {
-    return summary;
-  }
-  
-  @Override
-  public Output<TString> asOutput(Scope scope) {
-    return summary;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "TensorSummaryV2";
-  
-  private Output<TString> summary;
-  
-  private TensorSummary(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    summary = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/WriteAudioSummary.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/WriteAudioSummary.java
deleted file mode 100644
index f2f444cf5c4..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/WriteAudioSummary.java
+++ /dev/null
@@ -1,98 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.summary;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.TString;
-
-/**
- */
-public final class WriteAudioSummary extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.summary.WriteAudioSummary}
-   */
-  public static class Options {
-    
-    /**
-     * @param maxOutputs 
-     */
-    public Options maxOutputs(Long maxOutputs) {
-      this.maxOutputs = maxOutputs;
-      return this;
-    }
-    
-    private Long maxOutputs;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new WriteAudioSummary operation.
-   * 
-   * @param scope current scope
-   * @param writer 
-   * @param step 
-   * @param tag 
-   * @param tensor 
-   * @param sampleRate 
-   * @param options carries optional attributes values
-   * @return a new instance of WriteAudioSummary
-   */
-  @Endpoint(describeByClass = true)
-  public static WriteAudioSummary create(Scope scope, Operand<?> writer, Operand<TInt64> step, Operand<TString> tag, Operand<TFloat32> tensor, Operand<TFloat32> sampleRate, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("WriteAudioSummary", scope.makeOpName("WriteAudioSummary"));
-    opBuilder.addInput(writer.asOutput(scope));
-    opBuilder.addInput(step.asOutput(scope));
-    opBuilder.addInput(tag.asOutput(scope));
-    opBuilder.addInput(tensor.asOutput(scope));
-    opBuilder.addInput(sampleRate.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.maxOutputs != null) {
-          opBuilder.setAttr("max_outputs", opts.maxOutputs);
-        }
-      }
-    }
-    return new WriteAudioSummary(opBuilder.build());
-  }
-  
-  /**
-   * @param maxOutputs 
-   */
-  public static Options maxOutputs(Long maxOutputs) {
-    return new Options().maxOutputs(maxOutputs);
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "WriteAudioSummary";
-  
-  private WriteAudioSummary(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/WriteGraphSummary.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/WriteGraphSummary.java
deleted file mode 100644
index 36ccabbe82b..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/WriteGraphSummary.java
+++ /dev/null
@@ -1,59 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.summary;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.TString;
-
-/**
- */
-public final class WriteGraphSummary extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new WriteGraphSummary operation.
-   * 
-   * @param scope current scope
-   * @param writer 
-   * @param step 
-   * @param tensor 
-   * @return a new instance of WriteGraphSummary
-   */
-  @Endpoint(describeByClass = true)
-  public static WriteGraphSummary create(Scope scope, Operand<?> writer, Operand<TInt64> step, Operand<TString> tensor) {
-    OperationBuilder opBuilder = scope.env().opBuilder("WriteGraphSummary", scope.makeOpName("WriteGraphSummary"));
-    opBuilder.addInput(writer.asOutput(scope));
-    opBuilder.addInput(step.asOutput(scope));
-    opBuilder.addInput(tensor.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new WriteGraphSummary(opBuilder.build());
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "WriteGraphSummary";
-  
-  private WriteGraphSummary(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/WriteHistogramSummary.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/WriteHistogramSummary.java
deleted file mode 100644
index 30bd22b8448..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/WriteHistogramSummary.java
+++ /dev/null
@@ -1,62 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.summary;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.TString;
-import org.tensorflow.types.family.TNumber;
-
-/**
- */
-public final class WriteHistogramSummary extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new WriteHistogramSummary operation.
-   * 
-   * @param scope current scope
-   * @param writer 
-   * @param step 
-   * @param tag 
-   * @param values 
-   * @return a new instance of WriteHistogramSummary
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> WriteHistogramSummary create(Scope scope, Operand<?> writer, Operand<TInt64> step, Operand<TString> tag, Operand<T> values) {
-    OperationBuilder opBuilder = scope.env().opBuilder("WriteHistogramSummary", scope.makeOpName("WriteHistogramSummary"));
-    opBuilder.addInput(writer.asOutput(scope));
-    opBuilder.addInput(step.asOutput(scope));
-    opBuilder.addInput(tag.asOutput(scope));
-    opBuilder.addInput(values.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new WriteHistogramSummary(opBuilder.build());
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "WriteHistogramSummary";
-  
-  private WriteHistogramSummary(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/WriteImageSummary.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/WriteImageSummary.java
deleted file mode 100644
index 80942e3b53f..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/WriteImageSummary.java
+++ /dev/null
@@ -1,99 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.summary;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.TString;
-import org.tensorflow.types.TUint8;
-import org.tensorflow.types.family.TNumber;
-
-/**
- */
-public final class WriteImageSummary extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.summary.WriteImageSummary}
-   */
-  public static class Options {
-    
-    /**
-     * @param maxImages 
-     */
-    public Options maxImages(Long maxImages) {
-      this.maxImages = maxImages;
-      return this;
-    }
-    
-    private Long maxImages;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new WriteImageSummary operation.
-   * 
-   * @param scope current scope
-   * @param writer 
-   * @param step 
-   * @param tag 
-   * @param tensor 
-   * @param badColor 
-   * @param options carries optional attributes values
-   * @return a new instance of WriteImageSummary
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> WriteImageSummary create(Scope scope, Operand<?> writer, Operand<TInt64> step, Operand<TString> tag, Operand<T> tensor, Operand<TUint8> badColor, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("WriteImageSummary", scope.makeOpName("WriteImageSummary"));
-    opBuilder.addInput(writer.asOutput(scope));
-    opBuilder.addInput(step.asOutput(scope));
-    opBuilder.addInput(tag.asOutput(scope));
-    opBuilder.addInput(tensor.asOutput(scope));
-    opBuilder.addInput(badColor.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.maxImages != null) {
-          opBuilder.setAttr("max_images", opts.maxImages);
-        }
-      }
-    }
-    return new WriteImageSummary(opBuilder.build());
-  }
-  
-  /**
-   * @param maxImages 
-   */
-  public static Options maxImages(Long maxImages) {
-    return new Options().maxImages(maxImages);
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "WriteImageSummary";
-  
-  private WriteImageSummary(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/WriteRawProtoSummary.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/WriteRawProtoSummary.java
deleted file mode 100644
index c1dd19a57f5..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/WriteRawProtoSummary.java
+++ /dev/null
@@ -1,59 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.summary;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.TString;
-
-/**
- */
-public final class WriteRawProtoSummary extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new WriteRawProtoSummary operation.
-   * 
-   * @param scope current scope
-   * @param writer 
-   * @param step 
-   * @param tensor 
-   * @return a new instance of WriteRawProtoSummary
-   */
-  @Endpoint(describeByClass = true)
-  public static WriteRawProtoSummary create(Scope scope, Operand<?> writer, Operand<TInt64> step, Operand<TString> tensor) {
-    OperationBuilder opBuilder = scope.env().opBuilder("WriteRawProtoSummary", scope.makeOpName("WriteRawProtoSummary"));
-    opBuilder.addInput(writer.asOutput(scope));
-    opBuilder.addInput(step.asOutput(scope));
-    opBuilder.addInput(tensor.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new WriteRawProtoSummary(opBuilder.build());
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "WriteRawProtoSummary";
-  
-  private WriteRawProtoSummary(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/WriteScalarSummary.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/WriteScalarSummary.java
deleted file mode 100644
index 4878a15d68c..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/WriteScalarSummary.java
+++ /dev/null
@@ -1,62 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.summary;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.TString;
-import org.tensorflow.types.family.TNumber;
-
-/**
- */
-public final class WriteScalarSummary extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new WriteScalarSummary operation.
-   * 
-   * @param scope current scope
-   * @param writer 
-   * @param step 
-   * @param tag 
-   * @param value 
-   * @return a new instance of WriteScalarSummary
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> WriteScalarSummary create(Scope scope, Operand<?> writer, Operand<TInt64> step, Operand<TString> tag, Operand<T> value) {
-    OperationBuilder opBuilder = scope.env().opBuilder("WriteScalarSummary", scope.makeOpName("WriteScalarSummary"));
-    opBuilder.addInput(writer.asOutput(scope));
-    opBuilder.addInput(step.asOutput(scope));
-    opBuilder.addInput(tag.asOutput(scope));
-    opBuilder.addInput(value.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new WriteScalarSummary(opBuilder.build());
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "WriteScalarSummary";
-  
-  private WriteScalarSummary(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/WriteSummary.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/WriteSummary.java
deleted file mode 100644
index a5629306255..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/WriteSummary.java
+++ /dev/null
@@ -1,64 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.summary;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.TString;
-import org.tensorflow.types.family.TType;
-
-/**
- */
-public final class WriteSummary extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new WriteSummary operation.
-   * 
-   * @param scope current scope
-   * @param writer 
-   * @param step 
-   * @param tensor 
-   * @param tag 
-   * @param summaryMetadata 
-   * @return a new instance of WriteSummary
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> WriteSummary create(Scope scope, Operand<?> writer, Operand<TInt64> step, Operand<T> tensor, Operand<TString> tag, Operand<TString> summaryMetadata) {
-    OperationBuilder opBuilder = scope.env().opBuilder("WriteSummary", scope.makeOpName("WriteSummary"));
-    opBuilder.addInput(writer.asOutput(scope));
-    opBuilder.addInput(step.asOutput(scope));
-    opBuilder.addInput(tensor.asOutput(scope));
-    opBuilder.addInput(tag.asOutput(scope));
-    opBuilder.addInput(summaryMetadata.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new WriteSummary(opBuilder.build());
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "WriteSummary";
-  
-  private WriteSummary(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/AllToAll.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/AllToAll.java
deleted file mode 100644
index 71f38709d77..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/AllToAll.java
+++ /dev/null
@@ -1,103 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.tpu;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.family.TType;
-
-/**
- * An Op to exchange data across TPU replicas.
- * <p>
- * On each replica, the input is split into `split_count` blocks along
- * `split_dimension` and send to the other replicas given group_assignment. After
- * receiving `split_count` - 1 blocks from other replicas, we concatenate the
- * blocks along `concat_dimension` as the output.
- * <p>
- * For example, suppose there are 2 TPU replicas:
- * replica 0 receives input: `[[A, B]]`
- * replica 1 receives input: `[[C, D]]`
- * <p>
- * group_assignment=`[[0, 1]]`
- * concat_dimension=0
- * split_dimension=1
- * split_count=2
- * <p>
- * replica 0's output: `[[A], [C]]`
- * replica 1's output: `[[B], [D]]`
- * 
- * @param <T> data type for {@code output()} output
- */
-public final class AllToAll<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new AllToAll operation.
-   * 
-   * @param scope current scope
-   * @param input The local input to the sum.
-   * @param groupAssignment An int32 tensor with shape
-   * [num_groups, num_replicas_per_group]. `group_assignment[i]` represents the
-   * replica ids in the ith subgroup.
-   * @param concatDimension The dimension number to concatenate.
-   * @param splitDimension The dimension number to split.
-   * @param splitCount The number of splits, this number must equal to the sub-group
-   * size(group_assignment.get_shape()[1])
-   * @return a new instance of AllToAll
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> AllToAll<T> create(Scope scope, Operand<T> input, Operand<TInt32> groupAssignment, Long concatDimension, Long splitDimension, Long splitCount) {
-    OperationBuilder opBuilder = scope.env().opBuilder("AllToAll", scope.makeOpName("AllToAll"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder.addInput(groupAssignment.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("concat_dimension", concatDimension);
-    opBuilder.setAttr("split_dimension", splitDimension);
-    opBuilder.setAttr("split_count", splitCount);
-    return new AllToAll<T>(opBuilder.build());
-  }
-  
-  /**
-   * The exchanged result.
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "AllToAll";
-  
-  private Output<T> output;
-  
-  private AllToAll(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/CollectivePermute.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/CollectivePermute.java
deleted file mode 100644
index 58a583fa5c8..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/CollectivePermute.java
+++ /dev/null
@@ -1,84 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.tpu;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.family.TType;
-
-/**
- * An Op to permute tensors across replicated TPU instances.
- * <p>
- * Each instance supplies its own input.
- * <p>
- * For example, suppose there are 4 TPU instances: `[A, B, C, D]`. Passing
- * source_target_pairs=`[[0,1],[1,2],[2,3],[3,0]]` gets the outputs:
- * `[D, A, B, C]`.
- * 
- * @param <T> data type for {@code output()} output
- */
-public final class CollectivePermute<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new CollectivePermute operation.
-   * 
-   * @param scope current scope
-   * @param input The local input to be permuted. Currently only supports float and
-   * bfloat16.
-   * @param sourceTargetPairs A tensor with shape [num_pairs, 2].
-   * @return a new instance of CollectivePermute
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> CollectivePermute<T> create(Scope scope, Operand<T> input, Operand<TInt32> sourceTargetPairs) {
-    OperationBuilder opBuilder = scope.env().opBuilder("CollectivePermute", scope.makeOpName("CollectivePermute"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder.addInput(sourceTargetPairs.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new CollectivePermute<T>(opBuilder.build());
-  }
-  
-  /**
-   * The permuted input.
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "CollectivePermute";
-  
-  private Output<T> output;
-  
-  private CollectivePermute(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/ConfigureDistributedTPU.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/ConfigureDistributedTPU.java
deleted file mode 100644
index dd01f7135a3..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/ConfigureDistributedTPU.java
+++ /dev/null
@@ -1,183 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.tpu;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TString;
-
-/**
- * Sets up the centralized structures for a distributed TPU system.
- */
-public final class ConfigureDistributedTPU extends RawOp implements Operand<TString> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.tpu.ConfigureDistributedTPU}
-   */
-  public static class Options {
-    
-    /**
-     * @param embeddingConfig Reserved. Do not use.
-     */
-    public Options embeddingConfig(String embeddingConfig) {
-      this.embeddingConfig = embeddingConfig;
-      return this;
-    }
-    
-    /**
-     * @param tpuEmbeddingConfig Serialized tensorflow.tpu.TPUEmbeddingConfiguration that
-     * describes the embedding lookups of the program.
-     */
-    public Options tpuEmbeddingConfig(String tpuEmbeddingConfig) {
-      this.tpuEmbeddingConfig = tpuEmbeddingConfig;
-      return this;
-    }
-    
-    /**
-     * @param isGlobalInit Reserved. Do not use.
-     */
-    public Options isGlobalInit(Boolean isGlobalInit) {
-      this.isGlobalInit = isGlobalInit;
-      return this;
-    }
-    
-    /**
-     * @param enableWholeMeshCompilations 
-     */
-    public Options enableWholeMeshCompilations(Boolean enableWholeMeshCompilations) {
-      this.enableWholeMeshCompilations = enableWholeMeshCompilations;
-      return this;
-    }
-    
-    /**
-     * @param compilationFailureClosesChips 
-     */
-    public Options compilationFailureClosesChips(Boolean compilationFailureClosesChips) {
-      this.compilationFailureClosesChips = compilationFailureClosesChips;
-      return this;
-    }
-    
-    private String embeddingConfig;
-    private String tpuEmbeddingConfig;
-    private Boolean isGlobalInit;
-    private Boolean enableWholeMeshCompilations;
-    private Boolean compilationFailureClosesChips;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new ConfigureDistributedTPU operation.
-   * 
-   * @param scope current scope
-   * @param options carries optional attributes values
-   * @return a new instance of ConfigureDistributedTPU
-   */
-  @Endpoint(describeByClass = true)
-  public static ConfigureDistributedTPU create(Scope scope, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ConfigureDistributedTPU", scope.makeOpName("ConfigureDistributedTPU"));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.embeddingConfig != null) {
-          opBuilder.setAttr("embedding_config", opts.embeddingConfig);
-        }
-        if (opts.tpuEmbeddingConfig != null) {
-          opBuilder.setAttr("tpu_embedding_config", opts.tpuEmbeddingConfig);
-        }
-        if (opts.isGlobalInit != null) {
-          opBuilder.setAttr("is_global_init", opts.isGlobalInit);
-        }
-        if (opts.enableWholeMeshCompilations != null) {
-          opBuilder.setAttr("enable_whole_mesh_compilations", opts.enableWholeMeshCompilations);
-        }
-        if (opts.compilationFailureClosesChips != null) {
-          opBuilder.setAttr("compilation_failure_closes_chips", opts.compilationFailureClosesChips);
-        }
-      }
-    }
-    return new ConfigureDistributedTPU(opBuilder.build());
-  }
-  
-  /**
-   * @param embeddingConfig Reserved. Do not use.
-   */
-  public static Options embeddingConfig(String embeddingConfig) {
-    return new Options().embeddingConfig(embeddingConfig);
-  }
-  
-  /**
-   * @param tpuEmbeddingConfig Serialized tensorflow.tpu.TPUEmbeddingConfiguration that
-   * describes the embedding lookups of the program.
-   */
-  public static Options tpuEmbeddingConfig(String tpuEmbeddingConfig) {
-    return new Options().tpuEmbeddingConfig(tpuEmbeddingConfig);
-  }
-  
-  /**
-   * @param isGlobalInit Reserved. Do not use.
-   */
-  public static Options isGlobalInit(Boolean isGlobalInit) {
-    return new Options().isGlobalInit(isGlobalInit);
-  }
-  
-  /**
-   * @param enableWholeMeshCompilations 
-   */
-  public static Options enableWholeMeshCompilations(Boolean enableWholeMeshCompilations) {
-    return new Options().enableWholeMeshCompilations(enableWholeMeshCompilations);
-  }
-  
-  /**
-   * @param compilationFailureClosesChips 
-   */
-  public static Options compilationFailureClosesChips(Boolean compilationFailureClosesChips) {
-    return new Options().compilationFailureClosesChips(compilationFailureClosesChips);
-  }
-  
-  /**
-   * A serialized tensorflow.tpu.TopologyProto that describes the TPU
-   * topology.
-   */
-  public Output<TString> topology() {
-    return topology;
-  }
-  
-  @Override
-  public Output<TString> asOutput(Scope scope) {
-    return topology;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ConfigureDistributedTPU";
-  
-  private Output<TString> topology;
-  
-  private ConfigureDistributedTPU(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    topology = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/ConfigureTPUEmbedding.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/ConfigureTPUEmbedding.java
deleted file mode 100644
index 76bccd51f83..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/ConfigureTPUEmbedding.java
+++ /dev/null
@@ -1,54 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.tpu;
-
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-
-/**
- * Sets up TPUEmbedding in a distributed TPU system.
- */
-public final class ConfigureTPUEmbedding extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new ConfigureTPUEmbedding operation.
-   * 
-   * @param scope current scope
-   * @param config Serialized tensorflow.tpu.TPUEmbeddingConfiguration that
-   * describes the embedding lookups of the program.
-   * @return a new instance of ConfigureTPUEmbedding
-   */
-  @Endpoint(describeByClass = true)
-  public static ConfigureTPUEmbedding create(Scope scope, String config) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ConfigureTPUEmbedding", scope.makeOpName("ConfigureTPUEmbedding"));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("config", config);
-    return new ConfigureTPUEmbedding(opBuilder.build());
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ConfigureTPUEmbedding";
-  
-  private ConfigureTPUEmbedding(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/CrossReplicaSum.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/CrossReplicaSum.java
deleted file mode 100644
index 1ddf061e4e9..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/CrossReplicaSum.java
+++ /dev/null
@@ -1,86 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.tpu;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * An Op to sum inputs across replicated TPU instances.
- * <p>
- * Each instance supplies its own input.
- * <p>
- * For example, suppose there are 8 TPU instances: `[A, B, C, D, E, F, G, H]`.
- * Passing group_assignment=`[[0,2,4,6],[1,3,5,7]]` sets `A, C, E, G` as group 0,
- * and `B, D, F, H` as group 1. Thus we get the outputs:
- * `[A+C+E+G, B+D+F+H, A+C+E+G, B+D+F+H, A+C+E+G, B+D+F+H, A+C+E+G, B+D+F+H]`.
- * 
- * @param <T> data type for {@code output()} output
- */
-public final class CrossReplicaSum<T extends TNumber> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new CrossReplicaSum operation.
-   * 
-   * @param scope current scope
-   * @param input The local input to the sum.
-   * @param groupAssignment An int32 tensor with shape
-   * [num_groups, num_replicas_per_group]. `group_assignment[i]` represents the
-   * replica ids in the ith subgroup.
-   * @return a new instance of CrossReplicaSum
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber> CrossReplicaSum<T> create(Scope scope, Operand<T> input, Operand<TInt32> groupAssignment) {
-    OperationBuilder opBuilder = scope.env().opBuilder("CrossReplicaSum", scope.makeOpName("CrossReplicaSum"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder.addInput(groupAssignment.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new CrossReplicaSum<T>(opBuilder.build());
-  }
-  
-  /**
-   * The sum of all the distributed inputs.
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "CrossReplicaSum";
-  
-  private Output<T> output;
-  
-  private CrossReplicaSum(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/EnqueueTPUEmbeddingIntegerBatch.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/EnqueueTPUEmbeddingIntegerBatch.java
deleted file mode 100644
index e39002954f3..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/EnqueueTPUEmbeddingIntegerBatch.java
+++ /dev/null
@@ -1,99 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.tpu;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.Operands;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.TString;
-
-/**
- * An op that enqueues a list of input batch tensors to TPUEmbedding.
- */
-public final class EnqueueTPUEmbeddingIntegerBatch extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.tpu.EnqueueTPUEmbeddingIntegerBatch}
-   */
-  public static class Options {
-    
-    /**
-     * @param deviceOrdinal The TPU device to use. Should be >= 0 and less than the number
-     * of TPU cores in the task on which the node is placed.
-     */
-    public Options deviceOrdinal(Long deviceOrdinal) {
-      this.deviceOrdinal = deviceOrdinal;
-      return this;
-    }
-    
-    private Long deviceOrdinal;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new EnqueueTPUEmbeddingIntegerBatch operation.
-   * 
-   * @param scope current scope
-   * @param batch A list of 1D tensors, one for each embedding table, containing the
-   * indices into the tables.
-   * @param modeOverride A string input that overrides the mode specified in the
-   * TPUEmbeddingConfiguration. Supported values are {'unspecified', 'inference',
-   * 'training', 'backward_pass_only'}. When set to 'unspecified', the mode set
-   * in TPUEmbeddingConfiguration is used, otherwise mode_override is used.
-   * @param options carries optional attributes values
-   * @return a new instance of EnqueueTPUEmbeddingIntegerBatch
-   */
-  @Endpoint(describeByClass = true)
-  public static EnqueueTPUEmbeddingIntegerBatch create(Scope scope, Iterable<Operand<TInt32>> batch, Operand<TString> modeOverride, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("EnqueueTPUEmbeddingIntegerBatch", scope.makeOpName("EnqueueTPUEmbeddingIntegerBatch"));
-    opBuilder.addInputList(Operands.asOutputs(scope, batch));
-    opBuilder.addInput(modeOverride.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.deviceOrdinal != null) {
-          opBuilder.setAttr("device_ordinal", opts.deviceOrdinal);
-        }
-      }
-    }
-    return new EnqueueTPUEmbeddingIntegerBatch(opBuilder.build());
-  }
-  
-  /**
-   * @param deviceOrdinal The TPU device to use. Should be >= 0 and less than the number
-   * of TPU cores in the task on which the node is placed.
-   */
-  public static Options deviceOrdinal(Long deviceOrdinal) {
-    return new Options().deviceOrdinal(deviceOrdinal);
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "EnqueueTPUEmbeddingIntegerBatch";
-  
-  private EnqueueTPUEmbeddingIntegerBatch(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/EnqueueTPUEmbeddingRaggedTensorBatch.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/EnqueueTPUEmbeddingRaggedTensorBatch.java
deleted file mode 100644
index c4d1286e441..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/EnqueueTPUEmbeddingRaggedTensorBatch.java
+++ /dev/null
@@ -1,185 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.tpu;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.Operands;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TString;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Eases the porting of code that uses tf.nn.embedding_lookup().
- * <p>
- * sample_splits[i], embedding_indices[i] and aggregation_weights[i] correspond
- * to the ith feature. table_ids[i] indicates which embedding table to look up ith
- * feature.
- * <p>
- * The tensors at corresponding positions in two of the input lists,
- * embedding_indices and aggregation_weights, must have the same shape, i.e. rank 1
- * with dim_size() equal to the total number of lookups into the table described by
- * the corresponding feature.
- */
-public final class EnqueueTPUEmbeddingRaggedTensorBatch extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.tpu.EnqueueTPUEmbeddingRaggedTensorBatch}
-   */
-  public static class Options {
-    
-    /**
-     * @param deviceOrdinal The TPU device to use. Should be >= 0 and less than the number
-     * of TPU cores in the task on which the node is placed.
-     */
-    public Options deviceOrdinal(Long deviceOrdinal) {
-      this.deviceOrdinal = deviceOrdinal;
-      return this;
-    }
-    
-    /**
-     * @param combiners A list of string scalars, one for each embedding table that specify
-     * how to normalize the embedding activations after weighted summation.
-     * Supported combiners are 'mean', 'sum', or 'sqrtn'. It is invalid to have
-     * the sum of the weights be 0 for 'mean' or the sum of the squared weights be
-     * 0 for 'sqrtn'. If combiners isn't passed, the default is to use 'sum' for
-     * all tables.
-     */
-    public Options combiners(List<String> combiners) {
-      this.combiners = combiners;
-      return this;
-    }
-    
-    /**
-     * @param maxSequenceLengths 
-     */
-    public Options maxSequenceLengths(List<Long> maxSequenceLengths) {
-      this.maxSequenceLengths = maxSequenceLengths;
-      return this;
-    }
-    
-    private Long deviceOrdinal;
-    private List<String> combiners;
-    private List<Long> maxSequenceLengths;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new EnqueueTPUEmbeddingRaggedTensorBatch operation.
-   * 
-   * @param scope current scope
-   * @param sampleSplits A list of rank 1 Tensors specifying the break points for splitting
-   * embedding_indices and aggregation_weights into rows.
-   * It corresponds to ids.row_splits in embedding_lookup(), when ids is a
-   * RaggedTensor.
-   * @param embeddingIndices A list of rank 1 Tensors, indices into the embedding tables.
-   * It corresponds to ids.values in embedding_lookup(), when ids is a RaggedTensor.
-   * @param aggregationWeights A list of rank 1 Tensors containing per training example
-   * aggregation weights. It corresponds to the values field of a RaggedTensor
-   * with the same row_splits as ids in embedding_lookup(), when ids is a
-   * RaggedTensor.
-   * @param modeOverride A string input that overrides the mode specified in the
-   * TPUEmbeddingConfiguration. Supported values are {'unspecified', 'inference',
-   * 'training', 'backward_pass_only'}. When set to 'unspecified', the mode set
-   * in TPUEmbeddingConfiguration is used, otherwise mode_override is used.
-   * @param tableIds A list of integers specifying the identifier of the embedding table
-   * (offset of TableDescriptor in the TPUEmbeddingConfiguration) to lookup the
-   * corresponding input. The ith input is looked up using table_ids[i]. The size
-   * of the table_ids list must be equal to that of sample_indices,
-   * embedding_indices and aggregation_weights.
-   * @param options carries optional attributes values
-   * @return a new instance of EnqueueTPUEmbeddingRaggedTensorBatch
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber, U extends TNumber, V extends TNumber> EnqueueTPUEmbeddingRaggedTensorBatch create(Scope scope, Iterable<Operand<T>> sampleSplits, Iterable<Operand<U>> embeddingIndices, Iterable<Operand<V>> aggregationWeights, Operand<TString> modeOverride, List<Long> tableIds, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("EnqueueTPUEmbeddingRaggedTensorBatch", scope.makeOpName("EnqueueTPUEmbeddingRaggedTensorBatch"));
-    opBuilder.addInputList(Operands.asOutputs(scope, sampleSplits));
-    opBuilder.addInputList(Operands.asOutputs(scope, embeddingIndices));
-    opBuilder.addInputList(Operands.asOutputs(scope, aggregationWeights));
-    opBuilder.addInput(modeOverride.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    long[] tableIdsArray = new long[tableIds.size()];
-    for (int i = 0; i < tableIdsArray.length; ++i) {
-      tableIdsArray[i] = tableIds.get(i);
-    }
-    opBuilder.setAttr("table_ids", tableIdsArray);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.deviceOrdinal != null) {
-          opBuilder.setAttr("device_ordinal", opts.deviceOrdinal);
-        }
-        if (opts.combiners != null) {
-          String[] combinersArray = new String[opts.combiners.size()];
-          for (int i = 0; i < combinersArray.length; ++i) {
-            combinersArray[i] = opts.combiners.get(i);
-          }
-          opBuilder.setAttr("combiners", combinersArray);
-        }
-        if (opts.maxSequenceLengths != null) {
-          long[] maxSequenceLengthsArray = new long[opts.maxSequenceLengths.size()];
-          for (int i = 0; i < maxSequenceLengthsArray.length; ++i) {
-            maxSequenceLengthsArray[i] = opts.maxSequenceLengths.get(i);
-          }
-          opBuilder.setAttr("max_sequence_lengths", maxSequenceLengthsArray);
-        }
-      }
-    }
-    return new EnqueueTPUEmbeddingRaggedTensorBatch(opBuilder.build());
-  }
-  
-  /**
-   * @param deviceOrdinal The TPU device to use. Should be >= 0 and less than the number
-   * of TPU cores in the task on which the node is placed.
-   */
-  public static Options deviceOrdinal(Long deviceOrdinal) {
-    return new Options().deviceOrdinal(deviceOrdinal);
-  }
-  
-  /**
-   * @param combiners A list of string scalars, one for each embedding table that specify
-   * how to normalize the embedding activations after weighted summation.
-   * Supported combiners are 'mean', 'sum', or 'sqrtn'. It is invalid to have
-   * the sum of the weights be 0 for 'mean' or the sum of the squared weights be
-   * 0 for 'sqrtn'. If combiners isn't passed, the default is to use 'sum' for
-   * all tables.
-   */
-  public static Options combiners(List<String> combiners) {
-    return new Options().combiners(combiners);
-  }
-  
-  /**
-   * @param maxSequenceLengths 
-   */
-  public static Options maxSequenceLengths(List<Long> maxSequenceLengths) {
-    return new Options().maxSequenceLengths(maxSequenceLengths);
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "EnqueueTPUEmbeddingRaggedTensorBatch";
-  
-  private EnqueueTPUEmbeddingRaggedTensorBatch(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/EnqueueTPUEmbeddingSparseBatch.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/EnqueueTPUEmbeddingSparseBatch.java
deleted file mode 100644
index a4607d1ae2a..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/EnqueueTPUEmbeddingSparseBatch.java
+++ /dev/null
@@ -1,151 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.tpu;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.Operands;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TString;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * An op that enqueues TPUEmbedding input indices from a SparseTensor.
- * <p>
- * This Op eases the porting of code that uses embedding_lookup_sparse(),
- * although some Python preprocessing of the SparseTensor arguments to
- * embedding_lookup_sparse() is required to produce the arguments to this Op,
- * since only a single EnqueueTPUEmbeddingSparseBatch Op is allowed per training
- * step.
- * <p>
- * The tensors at corresponding positions in the three input lists
- * must have the same shape, i.e. rank 1 with dim_size() equal to the total
- * number of lookups into the table described by the corresponding table_id.
- */
-public final class EnqueueTPUEmbeddingSparseBatch extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.tpu.EnqueueTPUEmbeddingSparseBatch}
-   */
-  public static class Options {
-    
-    /**
-     * @param deviceOrdinal The TPU device to use. Should be >= 0 and less than the number
-     * of TPU cores in the task on which the node is placed.
-     */
-    public Options deviceOrdinal(Long deviceOrdinal) {
-      this.deviceOrdinal = deviceOrdinal;
-      return this;
-    }
-    
-    /**
-     * @param combiners A list of string scalars, one for each embedding table that specify
-     * how to normalize the embedding activations after weighted summation.
-     * Supported combiners are 'mean', 'sum', or 'sqrtn'. It is invalid to have
-     * the sum of the weights be 0 for 'mean' or the sum of the squared weights be
-     * 0 for 'sqrtn'. If combiners isn't passed, the default is to use 'sum' for
-     * all tables.
-     */
-    public Options combiners(List<String> combiners) {
-      this.combiners = combiners;
-      return this;
-    }
-    
-    private Long deviceOrdinal;
-    private List<String> combiners;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new EnqueueTPUEmbeddingSparseBatch operation.
-   * 
-   * @param scope current scope
-   * @param sampleIndices A list of rank 1 Tensors specifying the training example and
-   * feature to which the corresponding embedding_indices and aggregation_weights
-   * values belong. sample_indices[i] must equal b * nf + f, where nf is the
-   * number of features from the corresponding table, f is in [0, nf), and
-   * b is in [0, batch size).
-   * @param embeddingIndices A list of rank 1 Tensors, indices into the embedding tables.
-   * @param aggregationWeights A list of rank 1 Tensors containing per sample -- i.e. per
-   * (training example, feature) -- aggregation weights.
-   * @param modeOverride A string input that overrides the mode specified in the
-   * TPUEmbeddingConfiguration. Supported values are {'unspecified', 'inference',
-   * 'training', 'backward_pass_only'}. When set to 'unspecified', the mode set
-   * in TPUEmbeddingConfiguration is used, otherwise mode_override is used.
-   * @param options carries optional attributes values
-   * @return a new instance of EnqueueTPUEmbeddingSparseBatch
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber, U extends TNumber, V extends TNumber> EnqueueTPUEmbeddingSparseBatch create(Scope scope, Iterable<Operand<T>> sampleIndices, Iterable<Operand<U>> embeddingIndices, Iterable<Operand<V>> aggregationWeights, Operand<TString> modeOverride, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("EnqueueTPUEmbeddingSparseBatch", scope.makeOpName("EnqueueTPUEmbeddingSparseBatch"));
-    opBuilder.addInputList(Operands.asOutputs(scope, sampleIndices));
-    opBuilder.addInputList(Operands.asOutputs(scope, embeddingIndices));
-    opBuilder.addInputList(Operands.asOutputs(scope, aggregationWeights));
-    opBuilder.addInput(modeOverride.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.deviceOrdinal != null) {
-          opBuilder.setAttr("device_ordinal", opts.deviceOrdinal);
-        }
-        if (opts.combiners != null) {
-          String[] combinersArray = new String[opts.combiners.size()];
-          for (int i = 0; i < combinersArray.length; ++i) {
-            combinersArray[i] = opts.combiners.get(i);
-          }
-          opBuilder.setAttr("combiners", combinersArray);
-        }
-      }
-    }
-    return new EnqueueTPUEmbeddingSparseBatch(opBuilder.build());
-  }
-  
-  /**
-   * @param deviceOrdinal The TPU device to use. Should be >= 0 and less than the number
-   * of TPU cores in the task on which the node is placed.
-   */
-  public static Options deviceOrdinal(Long deviceOrdinal) {
-    return new Options().deviceOrdinal(deviceOrdinal);
-  }
-  
-  /**
-   * @param combiners A list of string scalars, one for each embedding table that specify
-   * how to normalize the embedding activations after weighted summation.
-   * Supported combiners are 'mean', 'sum', or 'sqrtn'. It is invalid to have
-   * the sum of the weights be 0 for 'mean' or the sum of the squared weights be
-   * 0 for 'sqrtn'. If combiners isn't passed, the default is to use 'sum' for
-   * all tables.
-   */
-  public static Options combiners(List<String> combiners) {
-    return new Options().combiners(combiners);
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "EnqueueTPUEmbeddingSparseBatch";
-  
-  private EnqueueTPUEmbeddingSparseBatch(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/EnqueueTPUEmbeddingSparseTensorBatch.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/EnqueueTPUEmbeddingSparseTensorBatch.java
deleted file mode 100644
index cf0f2bf80f5..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/EnqueueTPUEmbeddingSparseTensorBatch.java
+++ /dev/null
@@ -1,183 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.tpu;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.Operands;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TString;
-import org.tensorflow.types.family.TNumber;
-
-/**
- * Eases the porting of code that uses tf.nn.embedding_lookup_sparse().
- * <p>
- * sample_indices[i], embedding_indices[i] and aggregation_weights[i] correspond
- * to the ith feature. table_ids[i] indicates which embedding table to look up ith
- * feature.
- * <p>
- * The tensors at corresponding positions in the three input lists (sample_indices,
- * embedding_indices and aggregation_weights) must have the same shape, i.e. rank 1
- * with dim_size() equal to the total number of lookups into the table described by
- * the corresponding feature.
- */
-public final class EnqueueTPUEmbeddingSparseTensorBatch extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.tpu.EnqueueTPUEmbeddingSparseTensorBatch}
-   */
-  public static class Options {
-    
-    /**
-     * @param deviceOrdinal The TPU device to use. Should be >= 0 and less than the number
-     * of TPU cores in the task on which the node is placed.
-     */
-    public Options deviceOrdinal(Long deviceOrdinal) {
-      this.deviceOrdinal = deviceOrdinal;
-      return this;
-    }
-    
-    /**
-     * @param combiners A list of string scalars, one for each embedding table that specify
-     * how to normalize the embedding activations after weighted summation.
-     * Supported combiners are 'mean', 'sum', or 'sqrtn'. It is invalid to have
-     * the sum of the weights be 0 for 'mean' or the sum of the squared weights be
-     * 0 for 'sqrtn'. If combiners isn't passed, the default is to use 'sum' for
-     * all tables.
-     */
-    public Options combiners(List<String> combiners) {
-      this.combiners = combiners;
-      return this;
-    }
-    
-    /**
-     * @param maxSequenceLengths 
-     */
-    public Options maxSequenceLengths(List<Long> maxSequenceLengths) {
-      this.maxSequenceLengths = maxSequenceLengths;
-      return this;
-    }
-    
-    private Long deviceOrdinal;
-    private List<String> combiners;
-    private List<Long> maxSequenceLengths;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new EnqueueTPUEmbeddingSparseTensorBatch operation.
-   * 
-   * @param scope current scope
-   * @param sampleIndices A list of rank 1 Tensors specifying the training example to
-   * which the corresponding embedding_indices and aggregation_weights values
-   * belong. It corresponds to sp_ids.indices[:,0] in  embedding_lookup_sparse().
-   * @param embeddingIndices A list of rank 1 Tensors, indices into the embedding tables.
-   * It corresponds to sp_ids.values in embedding_lookup_sparse().
-   * @param aggregationWeights A list of rank 1 Tensors containing per training example
-   * aggregation weights. It corresponds to sp_weights.values in
-   * embedding_lookup_sparse().
-   * @param modeOverride A string input that overrides the mode specified in the
-   * TPUEmbeddingConfiguration. Supported values are {'unspecified', 'inference',
-   * 'training', 'backward_pass_only'}. When set to 'unspecified', the mode set
-   * in TPUEmbeddingConfiguration is used, otherwise mode_override is used.
-   * @param tableIds A list of integers specifying the identifier of the embedding table
-   * (offset of TableDescriptor in the TPUEmbeddingConfiguration) to lookup the
-   * corresponding input. The ith input is looked up using table_ids[i]. The size
-   * of the table_ids list must be equal to that of sample_indices,
-   * embedding_indices and aggregation_weights.
-   * @param options carries optional attributes values
-   * @return a new instance of EnqueueTPUEmbeddingSparseTensorBatch
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber, U extends TNumber, V extends TNumber> EnqueueTPUEmbeddingSparseTensorBatch create(Scope scope, Iterable<Operand<T>> sampleIndices, Iterable<Operand<U>> embeddingIndices, Iterable<Operand<V>> aggregationWeights, Operand<TString> modeOverride, List<Long> tableIds, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("EnqueueTPUEmbeddingSparseTensorBatch", scope.makeOpName("EnqueueTPUEmbeddingSparseTensorBatch"));
-    opBuilder.addInputList(Operands.asOutputs(scope, sampleIndices));
-    opBuilder.addInputList(Operands.asOutputs(scope, embeddingIndices));
-    opBuilder.addInputList(Operands.asOutputs(scope, aggregationWeights));
-    opBuilder.addInput(modeOverride.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    long[] tableIdsArray = new long[tableIds.size()];
-    for (int i = 0; i < tableIdsArray.length; ++i) {
-      tableIdsArray[i] = tableIds.get(i);
-    }
-    opBuilder.setAttr("table_ids", tableIdsArray);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.deviceOrdinal != null) {
-          opBuilder.setAttr("device_ordinal", opts.deviceOrdinal);
-        }
-        if (opts.combiners != null) {
-          String[] combinersArray = new String[opts.combiners.size()];
-          for (int i = 0; i < combinersArray.length; ++i) {
-            combinersArray[i] = opts.combiners.get(i);
-          }
-          opBuilder.setAttr("combiners", combinersArray);
-        }
-        if (opts.maxSequenceLengths != null) {
-          long[] maxSequenceLengthsArray = new long[opts.maxSequenceLengths.size()];
-          for (int i = 0; i < maxSequenceLengthsArray.length; ++i) {
-            maxSequenceLengthsArray[i] = opts.maxSequenceLengths.get(i);
-          }
-          opBuilder.setAttr("max_sequence_lengths", maxSequenceLengthsArray);
-        }
-      }
-    }
-    return new EnqueueTPUEmbeddingSparseTensorBatch(opBuilder.build());
-  }
-  
-  /**
-   * @param deviceOrdinal The TPU device to use. Should be >= 0 and less than the number
-   * of TPU cores in the task on which the node is placed.
-   */
-  public static Options deviceOrdinal(Long deviceOrdinal) {
-    return new Options().deviceOrdinal(deviceOrdinal);
-  }
-  
-  /**
-   * @param combiners A list of string scalars, one for each embedding table that specify
-   * how to normalize the embedding activations after weighted summation.
-   * Supported combiners are 'mean', 'sum', or 'sqrtn'. It is invalid to have
-   * the sum of the weights be 0 for 'mean' or the sum of the squared weights be
-   * 0 for 'sqrtn'. If combiners isn't passed, the default is to use 'sum' for
-   * all tables.
-   */
-  public static Options combiners(List<String> combiners) {
-    return new Options().combiners(combiners);
-  }
-  
-  /**
-   * @param maxSequenceLengths 
-   */
-  public static Options maxSequenceLengths(List<Long> maxSequenceLengths) {
-    return new Options().maxSequenceLengths(maxSequenceLengths);
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "EnqueueTPUEmbeddingSparseTensorBatch";
-  
-  private EnqueueTPUEmbeddingSparseTensorBatch(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/InfeedDequeue.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/InfeedDequeue.java
deleted file mode 100644
index ae5e04f0276..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/InfeedDequeue.java
+++ /dev/null
@@ -1,77 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.tpu;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * A placeholder op for a value that will be fed into the computation.
- * 
- * @param <T> data type for {@code output()} output
- */
-public final class InfeedDequeue<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new InfeedDequeue operation.
-   * 
-   * @param scope current scope
-   * @param dtype The type of elements in the tensor.
-   * @param shape The shape of the tensor.
-   * @return a new instance of InfeedDequeue
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> InfeedDequeue<T> create(Scope scope, Class<T> dtype, Shape shape) {
-    OperationBuilder opBuilder = scope.env().opBuilder("InfeedDequeue", scope.makeOpName("InfeedDequeue"));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("dtype", dtype);
-    opBuilder.setAttr("shape", shape);
-    return new InfeedDequeue<T>(opBuilder.build());
-  }
-  
-  /**
-   * A tensor that will be provided using the infeed mechanism.
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "InfeedDequeue";
-  
-  private Output<T> output;
-  
-  private InfeedDequeue(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/InfeedDequeueTuple.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/InfeedDequeueTuple.java
deleted file mode 100644
index 95f6b023f59..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/InfeedDequeueTuple.java
+++ /dev/null
@@ -1,89 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.tpu;
-
-import java.util.Arrays;
-import java.util.Iterator;
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Fetches multiple values from infeed as an XLA tuple.
- */
-public final class InfeedDequeueTuple extends RawOp implements Iterable<Operand<TType>> {
-  
-  /**
-   * Factory method to create a class wrapping a new InfeedDequeueTuple operation.
-   * 
-   * @param scope current scope
-   * @param dtypes The element types of each element in `outputs`.
-   * @param shapes The shapes of each tensor in `outputs`.
-   * @return a new instance of InfeedDequeueTuple
-   */
-  @Endpoint(describeByClass = true)
-  public static InfeedDequeueTuple create(Scope scope, List<Class<?>> dtypes, List<Shape> shapes) {
-    OperationBuilder opBuilder = scope.env().opBuilder("InfeedDequeueTuple", scope.makeOpName("InfeedDequeueTuple"));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    Class[] dtypesArray = new Class[dtypes.size()];
-    for (int i = 0; i < dtypesArray.length; ++i) {
-      dtypesArray[i] = dtypes.get(i);
-    }
-    opBuilder.setAttr("dtypes", dtypesArray);
-    Shape[] shapesArray = new Shape[shapes.size()];
-    for (int i = 0; i < shapesArray.length; ++i) {
-      shapesArray[i] = shapes.get(i);
-    }
-    opBuilder.setAttr("shapes", shapesArray);
-    return new InfeedDequeueTuple(opBuilder.build());
-  }
-  
-  /**
-   * A list of tensors that will be provided using the infeed mechanism.
-   */
-  public List<Output<?>> outputs() {
-    return outputs;
-  }
-  
-  @Override
-  @SuppressWarnings({"rawtypes", "unchecked"})
-  public Iterator<Operand<TType>> iterator() {
-    return (Iterator) outputs.iterator();
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "InfeedDequeueTuple";
-  
-  private List<Output<?>> outputs;
-  
-  private InfeedDequeueTuple(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    int outputsLength = operation.outputListLength("outputs");
-    outputs = Arrays.asList(operation.outputList(outputIdx, outputsLength));
-    outputIdx += outputsLength;
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/InfeedEnqueue.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/InfeedEnqueue.java
deleted file mode 100644
index 07c2452e051..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/InfeedEnqueue.java
+++ /dev/null
@@ -1,141 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.tpu;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * An op which feeds a single Tensor value into the computation.
- */
-public final class InfeedEnqueue extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.tpu.InfeedEnqueue}
-   */
-  public static class Options {
-    
-    /**
-     * @param shape The shape of the tensor.
-     */
-    public Options shape(Shape shape) {
-      this.shape = shape;
-      return this;
-    }
-    
-    /**
-     * @param layout A vector holding the requested layout in minor-to-major sequence.
-     * If a layout attribute is passed, but its values are all -1, the layout will
-     * be computed by the infeed operation.
-     */
-    public Options layout(List<Long> layout) {
-      this.layout = layout;
-      return this;
-    }
-    
-    /**
-     * @param deviceOrdinal The TPU device to use. This should be -1 when the Op
-     * is running on a TPU device, and >= 0 when the Op is running on the CPU
-     * device.
-     */
-    public Options deviceOrdinal(Long deviceOrdinal) {
-      this.deviceOrdinal = deviceOrdinal;
-      return this;
-    }
-    
-    private Shape shape;
-    private List<Long> layout;
-    private Long deviceOrdinal;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new InfeedEnqueue operation.
-   * 
-   * @param scope current scope
-   * @param input A tensor that will be provided using the infeed mechanism.
-   * @param options carries optional attributes values
-   * @return a new instance of InfeedEnqueue
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> InfeedEnqueue create(Scope scope, Operand<T> input, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("InfeedEnqueue", scope.makeOpName("InfeedEnqueue"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.shape != null) {
-          opBuilder.setAttr("shape", opts.shape);
-        }
-        if (opts.layout != null) {
-          long[] layoutArray = new long[opts.layout.size()];
-          for (int i = 0; i < layoutArray.length; ++i) {
-            layoutArray[i] = opts.layout.get(i);
-          }
-          opBuilder.setAttr("layout", layoutArray);
-        }
-        if (opts.deviceOrdinal != null) {
-          opBuilder.setAttr("device_ordinal", opts.deviceOrdinal);
-        }
-      }
-    }
-    return new InfeedEnqueue(opBuilder.build());
-  }
-  
-  /**
-   * @param shape The shape of the tensor.
-   */
-  public static Options shape(Shape shape) {
-    return new Options().shape(shape);
-  }
-  
-  /**
-   * @param layout A vector holding the requested layout in minor-to-major sequence.
-   * If a layout attribute is passed, but its values are all -1, the layout will
-   * be computed by the infeed operation.
-   */
-  public static Options layout(List<Long> layout) {
-    return new Options().layout(layout);
-  }
-  
-  /**
-   * @param deviceOrdinal The TPU device to use. This should be -1 when the Op
-   * is running on a TPU device, and >= 0 when the Op is running on the CPU
-   * device.
-   */
-  public static Options deviceOrdinal(Long deviceOrdinal) {
-    return new Options().deviceOrdinal(deviceOrdinal);
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "InfeedEnqueue";
-  
-  private InfeedEnqueue(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/InfeedEnqueuePrelinearizedBuffer.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/InfeedEnqueuePrelinearizedBuffer.java
deleted file mode 100644
index 2d3b7277185..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/InfeedEnqueuePrelinearizedBuffer.java
+++ /dev/null
@@ -1,90 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.tpu;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-
-/**
- * An op which enqueues prelinearized buffer into TPU infeed.
- */
-public final class InfeedEnqueuePrelinearizedBuffer extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.tpu.InfeedEnqueuePrelinearizedBuffer}
-   */
-  public static class Options {
-    
-    /**
-     * @param deviceOrdinal The TPU device to use. This should be -1 when the Op is running on a TPU device
-     * and = 0 when the Op is running on the CPU device.
-     */
-    public Options deviceOrdinal(Long deviceOrdinal) {
-      this.deviceOrdinal = deviceOrdinal;
-      return this;
-    }
-    
-    private Long deviceOrdinal;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new InfeedEnqueuePrelinearizedBuffer operation.
-   * 
-   * @param scope current scope
-   * @param input A variant tensor representing linearized output.
-   * @param options carries optional attributes values
-   * @return a new instance of InfeedEnqueuePrelinearizedBuffer
-   */
-  @Endpoint(describeByClass = true)
-  public static InfeedEnqueuePrelinearizedBuffer create(Scope scope, Operand<?> input, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("InfeedEnqueuePrelinearizedBuffer", scope.makeOpName("InfeedEnqueuePrelinearizedBuffer"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.deviceOrdinal != null) {
-          opBuilder.setAttr("device_ordinal", opts.deviceOrdinal);
-        }
-      }
-    }
-    return new InfeedEnqueuePrelinearizedBuffer(opBuilder.build());
-  }
-  
-  /**
-   * @param deviceOrdinal The TPU device to use. This should be -1 when the Op is running on a TPU device
-   * and = 0 when the Op is running on the CPU device.
-   */
-  public static Options deviceOrdinal(Long deviceOrdinal) {
-    return new Options().deviceOrdinal(deviceOrdinal);
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "InfeedEnqueuePrelinearizedBuffer";
-  
-  private InfeedEnqueuePrelinearizedBuffer(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/InfeedEnqueueTuple.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/InfeedEnqueueTuple.java
deleted file mode 100644
index a7a8d3e3be5..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/InfeedEnqueueTuple.java
+++ /dev/null
@@ -1,130 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.tpu;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.op.Operands;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-
-/**
- * Feeds multiple Tensor values into the computation as an XLA tuple.
- */
-public final class InfeedEnqueueTuple extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.tpu.InfeedEnqueueTuple}
-   */
-  public static class Options {
-    
-    /**
-     * @param layouts A vector holding the requested layout in minor-to-major sequence for
-     * all the tuple shapes, in the order the shapes appear in the "shapes" input.
-     * The layout elements for a sub-shape can be set to -1, in which case the
-     * corresponding layout will be computed by the infeed operation.
-     */
-    public Options layouts(List<Long> layouts) {
-      this.layouts = layouts;
-      return this;
-    }
-    
-    /**
-     * @param deviceOrdinal The TPU device to use. This should be -1 when the Op
-     * is running on a TPU device, and >= 0 when the Op is running on the CPU
-     * device.
-     */
-    public Options deviceOrdinal(Long deviceOrdinal) {
-      this.deviceOrdinal = deviceOrdinal;
-      return this;
-    }
-    
-    private List<Long> layouts;
-    private Long deviceOrdinal;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new InfeedEnqueueTuple operation.
-   * 
-   * @param scope current scope
-   * @param inputs A list of tensors that will be provided using the infeed mechanism.
-   * @param shapes The shapes of each tensor in `inputs`.
-   * @param options carries optional attributes values
-   * @return a new instance of InfeedEnqueueTuple
-   */
-  @Endpoint(describeByClass = true)
-  public static InfeedEnqueueTuple create(Scope scope, Iterable<Operand<?>> inputs, List<Shape> shapes, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("InfeedEnqueueTuple", scope.makeOpName("InfeedEnqueueTuple"));
-    opBuilder.addInputList(Operands.asOutputs(scope, inputs));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    Shape[] shapesArray = new Shape[shapes.size()];
-    for (int i = 0; i < shapesArray.length; ++i) {
-      shapesArray[i] = shapes.get(i);
-    }
-    opBuilder.setAttr("shapes", shapesArray);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.layouts != null) {
-          long[] layoutsArray = new long[opts.layouts.size()];
-          for (int i = 0; i < layoutsArray.length; ++i) {
-            layoutsArray[i] = opts.layouts.get(i);
-          }
-          opBuilder.setAttr("layouts", layoutsArray);
-        }
-        if (opts.deviceOrdinal != null) {
-          opBuilder.setAttr("device_ordinal", opts.deviceOrdinal);
-        }
-      }
-    }
-    return new InfeedEnqueueTuple(opBuilder.build());
-  }
-  
-  /**
-   * @param layouts A vector holding the requested layout in minor-to-major sequence for
-   * all the tuple shapes, in the order the shapes appear in the "shapes" input.
-   * The layout elements for a sub-shape can be set to -1, in which case the
-   * corresponding layout will be computed by the infeed operation.
-   */
-  public static Options layouts(List<Long> layouts) {
-    return new Options().layouts(layouts);
-  }
-  
-  /**
-   * @param deviceOrdinal The TPU device to use. This should be -1 when the Op
-   * is running on a TPU device, and >= 0 when the Op is running on the CPU
-   * device.
-   */
-  public static Options deviceOrdinal(Long deviceOrdinal) {
-    return new Options().deviceOrdinal(deviceOrdinal);
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "InfeedEnqueueTuple";
-  
-  private InfeedEnqueueTuple(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingADAMParameters.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingADAMParameters.java
deleted file mode 100644
index 8b29e1022f1..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingADAMParameters.java
+++ /dev/null
@@ -1,141 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.tpu;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-
-/**
- * Load ADAM embedding parameters.
- * <p>
- * An op that loads optimization parameters into HBM for embedding. Must be
- * preceded by a ConfigureTPUEmbeddingHost op that sets up the correct
- * embedding table configuration. For example, this op is used to install
- * parameters that are loaded from a checkpoint before a training loop is
- * executed.
- */
-public final class LoadTPUEmbeddingADAMParameters extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.tpu.LoadTPUEmbeddingADAMParameters}
-   */
-  public static class Options {
-    
-    /**
-     * @param tableId 
-     */
-    public Options tableId(Long tableId) {
-      this.tableId = tableId;
-      return this;
-    }
-    
-    /**
-     * @param tableName 
-     */
-    public Options tableName(String tableName) {
-      this.tableName = tableName;
-      return this;
-    }
-    
-    /**
-     * @param config 
-     */
-    public Options config(String config) {
-      this.config = config;
-      return this;
-    }
-    
-    private Long tableId;
-    private String tableName;
-    private String config;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new LoadTPUEmbeddingADAMParameters operation.
-   * 
-   * @param scope current scope
-   * @param parameters Value of parameters used in the ADAM optimization algorithm.
-   * @param momenta Value of momenta used in the ADAM optimization algorithm.
-   * @param velocities Value of velocities used in the ADAM optimization algorithm.
-   * @param numShards 
-   * @param shardId 
-   * @param options carries optional attributes values
-   * @return a new instance of LoadTPUEmbeddingADAMParameters
-   */
-  @Endpoint(describeByClass = true)
-  public static LoadTPUEmbeddingADAMParameters create(Scope scope, Operand<TFloat32> parameters, Operand<TFloat32> momenta, Operand<TFloat32> velocities, Long numShards, Long shardId, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("LoadTPUEmbeddingADAMParameters", scope.makeOpName("LoadTPUEmbeddingADAMParameters"));
-    opBuilder.addInput(parameters.asOutput(scope));
-    opBuilder.addInput(momenta.asOutput(scope));
-    opBuilder.addInput(velocities.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("num_shards", numShards);
-    opBuilder.setAttr("shard_id", shardId);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.tableId != null) {
-          opBuilder.setAttr("table_id", opts.tableId);
-        }
-        if (opts.tableName != null) {
-          opBuilder.setAttr("table_name", opts.tableName);
-        }
-        if (opts.config != null) {
-          opBuilder.setAttr("config", opts.config);
-        }
-      }
-    }
-    return new LoadTPUEmbeddingADAMParameters(opBuilder.build());
-  }
-  
-  /**
-   * @param tableId 
-   */
-  public static Options tableId(Long tableId) {
-    return new Options().tableId(tableId);
-  }
-  
-  /**
-   * @param tableName 
-   */
-  public static Options tableName(String tableName) {
-    return new Options().tableName(tableName);
-  }
-  
-  /**
-   * @param config 
-   */
-  public static Options config(String config) {
-    return new Options().config(config);
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "LoadTPUEmbeddingADAMParameters";
-  
-  private LoadTPUEmbeddingADAMParameters(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingADAMParametersGradAccumDebug.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingADAMParametersGradAccumDebug.java
deleted file mode 100644
index 44117323874..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingADAMParametersGradAccumDebug.java
+++ /dev/null
@@ -1,143 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.tpu;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-
-/**
- * Load ADAM embedding parameters with debug support.
- * <p>
- * An op that loads optimization parameters into HBM for embedding. Must be
- * preceded by a ConfigureTPUEmbeddingHost op that sets up the correct
- * embedding table configuration. For example, this op is used to install
- * parameters that are loaded from a checkpoint before a training loop is
- * executed.
- */
-public final class LoadTPUEmbeddingADAMParametersGradAccumDebug extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.tpu.LoadTPUEmbeddingADAMParametersGradAccumDebug}
-   */
-  public static class Options {
-    
-    /**
-     * @param tableId 
-     */
-    public Options tableId(Long tableId) {
-      this.tableId = tableId;
-      return this;
-    }
-    
-    /**
-     * @param tableName 
-     */
-    public Options tableName(String tableName) {
-      this.tableName = tableName;
-      return this;
-    }
-    
-    /**
-     * @param config 
-     */
-    public Options config(String config) {
-      this.config = config;
-      return this;
-    }
-    
-    private Long tableId;
-    private String tableName;
-    private String config;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new LoadTPUEmbeddingADAMParametersGradAccumDebug operation.
-   * 
-   * @param scope current scope
-   * @param parameters Value of parameters used in the ADAM optimization algorithm.
-   * @param momenta Value of momenta used in the ADAM optimization algorithm.
-   * @param velocities Value of velocities used in the ADAM optimization algorithm.
-   * @param gradientAccumulators Value of gradient_accumulators used in the ADAM optimization algorithm.
-   * @param numShards 
-   * @param shardId 
-   * @param options carries optional attributes values
-   * @return a new instance of LoadTPUEmbeddingADAMParametersGradAccumDebug
-   */
-  @Endpoint(describeByClass = true)
-  public static LoadTPUEmbeddingADAMParametersGradAccumDebug create(Scope scope, Operand<TFloat32> parameters, Operand<TFloat32> momenta, Operand<TFloat32> velocities, Operand<TFloat32> gradientAccumulators, Long numShards, Long shardId, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("LoadTPUEmbeddingADAMParametersGradAccumDebug", scope.makeOpName("LoadTPUEmbeddingADAMParametersGradAccumDebug"));
-    opBuilder.addInput(parameters.asOutput(scope));
-    opBuilder.addInput(momenta.asOutput(scope));
-    opBuilder.addInput(velocities.asOutput(scope));
-    opBuilder.addInput(gradientAccumulators.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("num_shards", numShards);
-    opBuilder.setAttr("shard_id", shardId);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.tableId != null) {
-          opBuilder.setAttr("table_id", opts.tableId);
-        }
-        if (opts.tableName != null) {
-          opBuilder.setAttr("table_name", opts.tableName);
-        }
-        if (opts.config != null) {
-          opBuilder.setAttr("config", opts.config);
-        }
-      }
-    }
-    return new LoadTPUEmbeddingADAMParametersGradAccumDebug(opBuilder.build());
-  }
-  
-  /**
-   * @param tableId 
-   */
-  public static Options tableId(Long tableId) {
-    return new Options().tableId(tableId);
-  }
-  
-  /**
-   * @param tableName 
-   */
-  public static Options tableName(String tableName) {
-    return new Options().tableName(tableName);
-  }
-  
-  /**
-   * @param config 
-   */
-  public static Options config(String config) {
-    return new Options().config(config);
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "LoadTPUEmbeddingADAMParametersGradAccumDebug";
-  
-  private LoadTPUEmbeddingADAMParametersGradAccumDebug(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingAdadeltaParameters.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingAdadeltaParameters.java
deleted file mode 100644
index 43befd10f20..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingAdadeltaParameters.java
+++ /dev/null
@@ -1,141 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.tpu;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-
-/**
- * Load Adadelta embedding parameters.
- * <p>
- * An op that loads optimization parameters into HBM for embedding. Must be
- * preceded by a ConfigureTPUEmbeddingHost op that sets up the correct
- * embedding table configuration. For example, this op is used to install
- * parameters that are loaded from a checkpoint before a training loop is
- * executed.
- */
-public final class LoadTPUEmbeddingAdadeltaParameters extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.tpu.LoadTPUEmbeddingAdadeltaParameters}
-   */
-  public static class Options {
-    
-    /**
-     * @param tableId 
-     */
-    public Options tableId(Long tableId) {
-      this.tableId = tableId;
-      return this;
-    }
-    
-    /**
-     * @param tableName 
-     */
-    public Options tableName(String tableName) {
-      this.tableName = tableName;
-      return this;
-    }
-    
-    /**
-     * @param config 
-     */
-    public Options config(String config) {
-      this.config = config;
-      return this;
-    }
-    
-    private Long tableId;
-    private String tableName;
-    private String config;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new LoadTPUEmbeddingAdadeltaParameters operation.
-   * 
-   * @param scope current scope
-   * @param parameters Value of parameters used in the Adadelta optimization algorithm.
-   * @param accumulators Value of accumulators used in the Adadelta optimization algorithm.
-   * @param updates Value of updates used in the Adadelta optimization algorithm.
-   * @param numShards 
-   * @param shardId 
-   * @param options carries optional attributes values
-   * @return a new instance of LoadTPUEmbeddingAdadeltaParameters
-   */
-  @Endpoint(describeByClass = true)
-  public static LoadTPUEmbeddingAdadeltaParameters create(Scope scope, Operand<TFloat32> parameters, Operand<TFloat32> accumulators, Operand<TFloat32> updates, Long numShards, Long shardId, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("LoadTPUEmbeddingAdadeltaParameters", scope.makeOpName("LoadTPUEmbeddingAdadeltaParameters"));
-    opBuilder.addInput(parameters.asOutput(scope));
-    opBuilder.addInput(accumulators.asOutput(scope));
-    opBuilder.addInput(updates.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("num_shards", numShards);
-    opBuilder.setAttr("shard_id", shardId);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.tableId != null) {
-          opBuilder.setAttr("table_id", opts.tableId);
-        }
-        if (opts.tableName != null) {
-          opBuilder.setAttr("table_name", opts.tableName);
-        }
-        if (opts.config != null) {
-          opBuilder.setAttr("config", opts.config);
-        }
-      }
-    }
-    return new LoadTPUEmbeddingAdadeltaParameters(opBuilder.build());
-  }
-  
-  /**
-   * @param tableId 
-   */
-  public static Options tableId(Long tableId) {
-    return new Options().tableId(tableId);
-  }
-  
-  /**
-   * @param tableName 
-   */
-  public static Options tableName(String tableName) {
-    return new Options().tableName(tableName);
-  }
-  
-  /**
-   * @param config 
-   */
-  public static Options config(String config) {
-    return new Options().config(config);
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "LoadTPUEmbeddingAdadeltaParameters";
-  
-  private LoadTPUEmbeddingAdadeltaParameters(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingAdadeltaParametersGradAccumDebug.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingAdadeltaParametersGradAccumDebug.java
deleted file mode 100644
index 3c9b92513da..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingAdadeltaParametersGradAccumDebug.java
+++ /dev/null
@@ -1,143 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.tpu;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-
-/**
- * Load Adadelta parameters with debug support.
- * <p>
- * An op that loads optimization parameters into HBM for embedding. Must be
- * preceded by a ConfigureTPUEmbeddingHost op that sets up the correct
- * embedding table configuration. For example, this op is used to install
- * parameters that are loaded from a checkpoint before a training loop is
- * executed.
- */
-public final class LoadTPUEmbeddingAdadeltaParametersGradAccumDebug extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.tpu.LoadTPUEmbeddingAdadeltaParametersGradAccumDebug}
-   */
-  public static class Options {
-    
-    /**
-     * @param tableId 
-     */
-    public Options tableId(Long tableId) {
-      this.tableId = tableId;
-      return this;
-    }
-    
-    /**
-     * @param tableName 
-     */
-    public Options tableName(String tableName) {
-      this.tableName = tableName;
-      return this;
-    }
-    
-    /**
-     * @param config 
-     */
-    public Options config(String config) {
-      this.config = config;
-      return this;
-    }
-    
-    private Long tableId;
-    private String tableName;
-    private String config;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new LoadTPUEmbeddingAdadeltaParametersGradAccumDebug operation.
-   * 
-   * @param scope current scope
-   * @param parameters Value of parameters used in the Adadelta optimization algorithm.
-   * @param accumulators Value of accumulators used in the Adadelta optimization algorithm.
-   * @param updates Value of updates used in the Adadelta optimization algorithm.
-   * @param gradientAccumulators Value of gradient_accumulators used in the Adadelta optimization algorithm.
-   * @param numShards 
-   * @param shardId 
-   * @param options carries optional attributes values
-   * @return a new instance of LoadTPUEmbeddingAdadeltaParametersGradAccumDebug
-   */
-  @Endpoint(describeByClass = true)
-  public static LoadTPUEmbeddingAdadeltaParametersGradAccumDebug create(Scope scope, Operand<TFloat32> parameters, Operand<TFloat32> accumulators, Operand<TFloat32> updates, Operand<TFloat32> gradientAccumulators, Long numShards, Long shardId, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("LoadTPUEmbeddingAdadeltaParametersGradAccumDebug", scope.makeOpName("LoadTPUEmbeddingAdadeltaParametersGradAccumDebug"));
-    opBuilder.addInput(parameters.asOutput(scope));
-    opBuilder.addInput(accumulators.asOutput(scope));
-    opBuilder.addInput(updates.asOutput(scope));
-    opBuilder.addInput(gradientAccumulators.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("num_shards", numShards);
-    opBuilder.setAttr("shard_id", shardId);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.tableId != null) {
-          opBuilder.setAttr("table_id", opts.tableId);
-        }
-        if (opts.tableName != null) {
-          opBuilder.setAttr("table_name", opts.tableName);
-        }
-        if (opts.config != null) {
-          opBuilder.setAttr("config", opts.config);
-        }
-      }
-    }
-    return new LoadTPUEmbeddingAdadeltaParametersGradAccumDebug(opBuilder.build());
-  }
-  
-  /**
-   * @param tableId 
-   */
-  public static Options tableId(Long tableId) {
-    return new Options().tableId(tableId);
-  }
-  
-  /**
-   * @param tableName 
-   */
-  public static Options tableName(String tableName) {
-    return new Options().tableName(tableName);
-  }
-  
-  /**
-   * @param config 
-   */
-  public static Options config(String config) {
-    return new Options().config(config);
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "LoadTPUEmbeddingAdadeltaParametersGradAccumDebug";
-  
-  private LoadTPUEmbeddingAdadeltaParametersGradAccumDebug(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingAdagradParameters.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingAdagradParameters.java
deleted file mode 100644
index 9695b8815e5..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingAdagradParameters.java
+++ /dev/null
@@ -1,139 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.tpu;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-
-/**
- * Load Adagrad embedding parameters.
- * <p>
- * An op that loads optimization parameters into HBM for embedding. Must be
- * preceded by a ConfigureTPUEmbeddingHost op that sets up the correct
- * embedding table configuration. For example, this op is used to install
- * parameters that are loaded from a checkpoint before a training loop is
- * executed.
- */
-public final class LoadTPUEmbeddingAdagradParameters extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.tpu.LoadTPUEmbeddingAdagradParameters}
-   */
-  public static class Options {
-    
-    /**
-     * @param tableId 
-     */
-    public Options tableId(Long tableId) {
-      this.tableId = tableId;
-      return this;
-    }
-    
-    /**
-     * @param tableName 
-     */
-    public Options tableName(String tableName) {
-      this.tableName = tableName;
-      return this;
-    }
-    
-    /**
-     * @param config 
-     */
-    public Options config(String config) {
-      this.config = config;
-      return this;
-    }
-    
-    private Long tableId;
-    private String tableName;
-    private String config;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new LoadTPUEmbeddingAdagradParameters operation.
-   * 
-   * @param scope current scope
-   * @param parameters Value of parameters used in the Adagrad optimization algorithm.
-   * @param accumulators Value of accumulators used in the Adagrad optimization algorithm.
-   * @param numShards 
-   * @param shardId 
-   * @param options carries optional attributes values
-   * @return a new instance of LoadTPUEmbeddingAdagradParameters
-   */
-  @Endpoint(describeByClass = true)
-  public static LoadTPUEmbeddingAdagradParameters create(Scope scope, Operand<TFloat32> parameters, Operand<TFloat32> accumulators, Long numShards, Long shardId, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("LoadTPUEmbeddingAdagradParameters", scope.makeOpName("LoadTPUEmbeddingAdagradParameters"));
-    opBuilder.addInput(parameters.asOutput(scope));
-    opBuilder.addInput(accumulators.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("num_shards", numShards);
-    opBuilder.setAttr("shard_id", shardId);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.tableId != null) {
-          opBuilder.setAttr("table_id", opts.tableId);
-        }
-        if (opts.tableName != null) {
-          opBuilder.setAttr("table_name", opts.tableName);
-        }
-        if (opts.config != null) {
-          opBuilder.setAttr("config", opts.config);
-        }
-      }
-    }
-    return new LoadTPUEmbeddingAdagradParameters(opBuilder.build());
-  }
-  
-  /**
-   * @param tableId 
-   */
-  public static Options tableId(Long tableId) {
-    return new Options().tableId(tableId);
-  }
-  
-  /**
-   * @param tableName 
-   */
-  public static Options tableName(String tableName) {
-    return new Options().tableName(tableName);
-  }
-  
-  /**
-   * @param config 
-   */
-  public static Options config(String config) {
-    return new Options().config(config);
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "LoadTPUEmbeddingAdagradParameters";
-  
-  private LoadTPUEmbeddingAdagradParameters(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingAdagradParametersGradAccumDebug.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingAdagradParametersGradAccumDebug.java
deleted file mode 100644
index edc56619b76..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingAdagradParametersGradAccumDebug.java
+++ /dev/null
@@ -1,141 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.tpu;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-
-/**
- * Load Adagrad embedding parameters with debug support.
- * <p>
- * An op that loads optimization parameters into HBM for embedding. Must be
- * preceded by a ConfigureTPUEmbeddingHost op that sets up the correct
- * embedding table configuration. For example, this op is used to install
- * parameters that are loaded from a checkpoint before a training loop is
- * executed.
- */
-public final class LoadTPUEmbeddingAdagradParametersGradAccumDebug extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.tpu.LoadTPUEmbeddingAdagradParametersGradAccumDebug}
-   */
-  public static class Options {
-    
-    /**
-     * @param tableId 
-     */
-    public Options tableId(Long tableId) {
-      this.tableId = tableId;
-      return this;
-    }
-    
-    /**
-     * @param tableName 
-     */
-    public Options tableName(String tableName) {
-      this.tableName = tableName;
-      return this;
-    }
-    
-    /**
-     * @param config 
-     */
-    public Options config(String config) {
-      this.config = config;
-      return this;
-    }
-    
-    private Long tableId;
-    private String tableName;
-    private String config;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new LoadTPUEmbeddingAdagradParametersGradAccumDebug operation.
-   * 
-   * @param scope current scope
-   * @param parameters Value of parameters used in the Adagrad optimization algorithm.
-   * @param accumulators Value of accumulators used in the Adagrad optimization algorithm.
-   * @param gradientAccumulators Value of gradient_accumulators used in the Adagrad optimization algorithm.
-   * @param numShards 
-   * @param shardId 
-   * @param options carries optional attributes values
-   * @return a new instance of LoadTPUEmbeddingAdagradParametersGradAccumDebug
-   */
-  @Endpoint(describeByClass = true)
-  public static LoadTPUEmbeddingAdagradParametersGradAccumDebug create(Scope scope, Operand<TFloat32> parameters, Operand<TFloat32> accumulators, Operand<TFloat32> gradientAccumulators, Long numShards, Long shardId, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("LoadTPUEmbeddingAdagradParametersGradAccumDebug", scope.makeOpName("LoadTPUEmbeddingAdagradParametersGradAccumDebug"));
-    opBuilder.addInput(parameters.asOutput(scope));
-    opBuilder.addInput(accumulators.asOutput(scope));
-    opBuilder.addInput(gradientAccumulators.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("num_shards", numShards);
-    opBuilder.setAttr("shard_id", shardId);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.tableId != null) {
-          opBuilder.setAttr("table_id", opts.tableId);
-        }
-        if (opts.tableName != null) {
-          opBuilder.setAttr("table_name", opts.tableName);
-        }
-        if (opts.config != null) {
-          opBuilder.setAttr("config", opts.config);
-        }
-      }
-    }
-    return new LoadTPUEmbeddingAdagradParametersGradAccumDebug(opBuilder.build());
-  }
-  
-  /**
-   * @param tableId 
-   */
-  public static Options tableId(Long tableId) {
-    return new Options().tableId(tableId);
-  }
-  
-  /**
-   * @param tableName 
-   */
-  public static Options tableName(String tableName) {
-    return new Options().tableName(tableName);
-  }
-  
-  /**
-   * @param config 
-   */
-  public static Options config(String config) {
-    return new Options().config(config);
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "LoadTPUEmbeddingAdagradParametersGradAccumDebug";
-  
-  private LoadTPUEmbeddingAdagradParametersGradAccumDebug(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingCenteredRMSPropParameters.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingCenteredRMSPropParameters.java
deleted file mode 100644
index d727ad24e1f..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingCenteredRMSPropParameters.java
+++ /dev/null
@@ -1,143 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.tpu;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-
-/**
- * Load centered RMSProp embedding parameters.
- * <p>
- * An op that loads optimization parameters into HBM for embedding. Must be
- * preceded by a ConfigureTPUEmbeddingHost op that sets up the correct
- * embedding table configuration. For example, this op is used to install
- * parameters that are loaded from a checkpoint before a training loop is
- * executed.
- */
-public final class LoadTPUEmbeddingCenteredRMSPropParameters extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.tpu.LoadTPUEmbeddingCenteredRMSPropParameters}
-   */
-  public static class Options {
-    
-    /**
-     * @param tableId 
-     */
-    public Options tableId(Long tableId) {
-      this.tableId = tableId;
-      return this;
-    }
-    
-    /**
-     * @param tableName 
-     */
-    public Options tableName(String tableName) {
-      this.tableName = tableName;
-      return this;
-    }
-    
-    /**
-     * @param config 
-     */
-    public Options config(String config) {
-      this.config = config;
-      return this;
-    }
-    
-    private Long tableId;
-    private String tableName;
-    private String config;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new LoadTPUEmbeddingCenteredRMSPropParameters operation.
-   * 
-   * @param scope current scope
-   * @param parameters Value of parameters used in the centered RMSProp optimization algorithm.
-   * @param ms Value of ms used in the centered RMSProp optimization algorithm.
-   * @param mom Value of mom used in the centered RMSProp optimization algorithm.
-   * @param mg Value of mg used in the centered RMSProp optimization algorithm.
-   * @param numShards 
-   * @param shardId 
-   * @param options carries optional attributes values
-   * @return a new instance of LoadTPUEmbeddingCenteredRMSPropParameters
-   */
-  @Endpoint(describeByClass = true)
-  public static LoadTPUEmbeddingCenteredRMSPropParameters create(Scope scope, Operand<TFloat32> parameters, Operand<TFloat32> ms, Operand<TFloat32> mom, Operand<TFloat32> mg, Long numShards, Long shardId, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("LoadTPUEmbeddingCenteredRMSPropParameters", scope.makeOpName("LoadTPUEmbeddingCenteredRMSPropParameters"));
-    opBuilder.addInput(parameters.asOutput(scope));
-    opBuilder.addInput(ms.asOutput(scope));
-    opBuilder.addInput(mom.asOutput(scope));
-    opBuilder.addInput(mg.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("num_shards", numShards);
-    opBuilder.setAttr("shard_id", shardId);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.tableId != null) {
-          opBuilder.setAttr("table_id", opts.tableId);
-        }
-        if (opts.tableName != null) {
-          opBuilder.setAttr("table_name", opts.tableName);
-        }
-        if (opts.config != null) {
-          opBuilder.setAttr("config", opts.config);
-        }
-      }
-    }
-    return new LoadTPUEmbeddingCenteredRMSPropParameters(opBuilder.build());
-  }
-  
-  /**
-   * @param tableId 
-   */
-  public static Options tableId(Long tableId) {
-    return new Options().tableId(tableId);
-  }
-  
-  /**
-   * @param tableName 
-   */
-  public static Options tableName(String tableName) {
-    return new Options().tableName(tableName);
-  }
-  
-  /**
-   * @param config 
-   */
-  public static Options config(String config) {
-    return new Options().config(config);
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "LoadTPUEmbeddingCenteredRMSPropParameters";
-  
-  private LoadTPUEmbeddingCenteredRMSPropParameters(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingFTRLParameters.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingFTRLParameters.java
deleted file mode 100644
index 242282fdd7a..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingFTRLParameters.java
+++ /dev/null
@@ -1,141 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.tpu;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-
-/**
- * Load FTRL embedding parameters.
- * <p>
- * An op that loads optimization parameters into HBM for embedding. Must be
- * preceded by a ConfigureTPUEmbeddingHost op that sets up the correct
- * embedding table configuration. For example, this op is used to install
- * parameters that are loaded from a checkpoint before a training loop is
- * executed.
- */
-public final class LoadTPUEmbeddingFTRLParameters extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.tpu.LoadTPUEmbeddingFTRLParameters}
-   */
-  public static class Options {
-    
-    /**
-     * @param tableId 
-     */
-    public Options tableId(Long tableId) {
-      this.tableId = tableId;
-      return this;
-    }
-    
-    /**
-     * @param tableName 
-     */
-    public Options tableName(String tableName) {
-      this.tableName = tableName;
-      return this;
-    }
-    
-    /**
-     * @param config 
-     */
-    public Options config(String config) {
-      this.config = config;
-      return this;
-    }
-    
-    private Long tableId;
-    private String tableName;
-    private String config;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new LoadTPUEmbeddingFTRLParameters operation.
-   * 
-   * @param scope current scope
-   * @param parameters Value of parameters used in the FTRL optimization algorithm.
-   * @param accumulators Value of accumulators used in the FTRL optimization algorithm.
-   * @param linears Value of linears used in the FTRL optimization algorithm.
-   * @param numShards 
-   * @param shardId 
-   * @param options carries optional attributes values
-   * @return a new instance of LoadTPUEmbeddingFTRLParameters
-   */
-  @Endpoint(describeByClass = true)
-  public static LoadTPUEmbeddingFTRLParameters create(Scope scope, Operand<TFloat32> parameters, Operand<TFloat32> accumulators, Operand<TFloat32> linears, Long numShards, Long shardId, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("LoadTPUEmbeddingFTRLParameters", scope.makeOpName("LoadTPUEmbeddingFTRLParameters"));
-    opBuilder.addInput(parameters.asOutput(scope));
-    opBuilder.addInput(accumulators.asOutput(scope));
-    opBuilder.addInput(linears.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("num_shards", numShards);
-    opBuilder.setAttr("shard_id", shardId);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.tableId != null) {
-          opBuilder.setAttr("table_id", opts.tableId);
-        }
-        if (opts.tableName != null) {
-          opBuilder.setAttr("table_name", opts.tableName);
-        }
-        if (opts.config != null) {
-          opBuilder.setAttr("config", opts.config);
-        }
-      }
-    }
-    return new LoadTPUEmbeddingFTRLParameters(opBuilder.build());
-  }
-  
-  /**
-   * @param tableId 
-   */
-  public static Options tableId(Long tableId) {
-    return new Options().tableId(tableId);
-  }
-  
-  /**
-   * @param tableName 
-   */
-  public static Options tableName(String tableName) {
-    return new Options().tableName(tableName);
-  }
-  
-  /**
-   * @param config 
-   */
-  public static Options config(String config) {
-    return new Options().config(config);
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "LoadTPUEmbeddingFTRLParameters";
-  
-  private LoadTPUEmbeddingFTRLParameters(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingFTRLParametersGradAccumDebug.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingFTRLParametersGradAccumDebug.java
deleted file mode 100644
index f5fe30a6c27..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingFTRLParametersGradAccumDebug.java
+++ /dev/null
@@ -1,143 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.tpu;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-
-/**
- * Load FTRL embedding parameters with debug support.
- * <p>
- * An op that loads optimization parameters into HBM for embedding. Must be
- * preceded by a ConfigureTPUEmbeddingHost op that sets up the correct
- * embedding table configuration. For example, this op is used to install
- * parameters that are loaded from a checkpoint before a training loop is
- * executed.
- */
-public final class LoadTPUEmbeddingFTRLParametersGradAccumDebug extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.tpu.LoadTPUEmbeddingFTRLParametersGradAccumDebug}
-   */
-  public static class Options {
-    
-    /**
-     * @param tableId 
-     */
-    public Options tableId(Long tableId) {
-      this.tableId = tableId;
-      return this;
-    }
-    
-    /**
-     * @param tableName 
-     */
-    public Options tableName(String tableName) {
-      this.tableName = tableName;
-      return this;
-    }
-    
-    /**
-     * @param config 
-     */
-    public Options config(String config) {
-      this.config = config;
-      return this;
-    }
-    
-    private Long tableId;
-    private String tableName;
-    private String config;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new LoadTPUEmbeddingFTRLParametersGradAccumDebug operation.
-   * 
-   * @param scope current scope
-   * @param parameters Value of parameters used in the FTRL optimization algorithm.
-   * @param accumulators Value of accumulators used in the FTRL optimization algorithm.
-   * @param linears Value of linears used in the FTRL optimization algorithm.
-   * @param gradientAccumulators Value of gradient_accumulators used in the FTRL optimization algorithm.
-   * @param numShards 
-   * @param shardId 
-   * @param options carries optional attributes values
-   * @return a new instance of LoadTPUEmbeddingFTRLParametersGradAccumDebug
-   */
-  @Endpoint(describeByClass = true)
-  public static LoadTPUEmbeddingFTRLParametersGradAccumDebug create(Scope scope, Operand<TFloat32> parameters, Operand<TFloat32> accumulators, Operand<TFloat32> linears, Operand<TFloat32> gradientAccumulators, Long numShards, Long shardId, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("LoadTPUEmbeddingFTRLParametersGradAccumDebug", scope.makeOpName("LoadTPUEmbeddingFTRLParametersGradAccumDebug"));
-    opBuilder.addInput(parameters.asOutput(scope));
-    opBuilder.addInput(accumulators.asOutput(scope));
-    opBuilder.addInput(linears.asOutput(scope));
-    opBuilder.addInput(gradientAccumulators.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("num_shards", numShards);
-    opBuilder.setAttr("shard_id", shardId);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.tableId != null) {
-          opBuilder.setAttr("table_id", opts.tableId);
-        }
-        if (opts.tableName != null) {
-          opBuilder.setAttr("table_name", opts.tableName);
-        }
-        if (opts.config != null) {
-          opBuilder.setAttr("config", opts.config);
-        }
-      }
-    }
-    return new LoadTPUEmbeddingFTRLParametersGradAccumDebug(opBuilder.build());
-  }
-  
-  /**
-   * @param tableId 
-   */
-  public static Options tableId(Long tableId) {
-    return new Options().tableId(tableId);
-  }
-  
-  /**
-   * @param tableName 
-   */
-  public static Options tableName(String tableName) {
-    return new Options().tableName(tableName);
-  }
-  
-  /**
-   * @param config 
-   */
-  public static Options config(String config) {
-    return new Options().config(config);
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "LoadTPUEmbeddingFTRLParametersGradAccumDebug";
-  
-  private LoadTPUEmbeddingFTRLParametersGradAccumDebug(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingMDLAdagradLightParameters.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingMDLAdagradLightParameters.java
deleted file mode 100644
index 28983395b7c..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingMDLAdagradLightParameters.java
+++ /dev/null
@@ -1,143 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.tpu;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-
-/**
- * Load MDL Adagrad Light embedding parameters.
- * <p>
- * An op that loads optimization parameters into HBM for embedding. Must be
- * preceded by a ConfigureTPUEmbeddingHost op that sets up the correct
- * embedding table configuration. For example, this op is used to install
- * parameters that are loaded from a checkpoint before a training loop is
- * executed.
- */
-public final class LoadTPUEmbeddingMDLAdagradLightParameters extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.tpu.LoadTPUEmbeddingMDLAdagradLightParameters}
-   */
-  public static class Options {
-    
-    /**
-     * @param tableId 
-     */
-    public Options tableId(Long tableId) {
-      this.tableId = tableId;
-      return this;
-    }
-    
-    /**
-     * @param tableName 
-     */
-    public Options tableName(String tableName) {
-      this.tableName = tableName;
-      return this;
-    }
-    
-    /**
-     * @param config 
-     */
-    public Options config(String config) {
-      this.config = config;
-      return this;
-    }
-    
-    private Long tableId;
-    private String tableName;
-    private String config;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new LoadTPUEmbeddingMDLAdagradLightParameters operation.
-   * 
-   * @param scope current scope
-   * @param parameters Value of parameters used in the MDL Adagrad Light optimization algorithm.
-   * @param accumulators Value of accumulators used in the MDL Adagrad Light optimization algorithm.
-   * @param weights Value of weights used in the MDL Adagrad Light optimization algorithm.
-   * @param benefits Value of benefits used in the MDL Adagrad Light optimization algorithm.
-   * @param numShards 
-   * @param shardId 
-   * @param options carries optional attributes values
-   * @return a new instance of LoadTPUEmbeddingMDLAdagradLightParameters
-   */
-  @Endpoint(describeByClass = true)
-  public static LoadTPUEmbeddingMDLAdagradLightParameters create(Scope scope, Operand<TFloat32> parameters, Operand<TFloat32> accumulators, Operand<TFloat32> weights, Operand<TFloat32> benefits, Long numShards, Long shardId, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("LoadTPUEmbeddingMDLAdagradLightParameters", scope.makeOpName("LoadTPUEmbeddingMDLAdagradLightParameters"));
-    opBuilder.addInput(parameters.asOutput(scope));
-    opBuilder.addInput(accumulators.asOutput(scope));
-    opBuilder.addInput(weights.asOutput(scope));
-    opBuilder.addInput(benefits.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("num_shards", numShards);
-    opBuilder.setAttr("shard_id", shardId);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.tableId != null) {
-          opBuilder.setAttr("table_id", opts.tableId);
-        }
-        if (opts.tableName != null) {
-          opBuilder.setAttr("table_name", opts.tableName);
-        }
-        if (opts.config != null) {
-          opBuilder.setAttr("config", opts.config);
-        }
-      }
-    }
-    return new LoadTPUEmbeddingMDLAdagradLightParameters(opBuilder.build());
-  }
-  
-  /**
-   * @param tableId 
-   */
-  public static Options tableId(Long tableId) {
-    return new Options().tableId(tableId);
-  }
-  
-  /**
-   * @param tableName 
-   */
-  public static Options tableName(String tableName) {
-    return new Options().tableName(tableName);
-  }
-  
-  /**
-   * @param config 
-   */
-  public static Options config(String config) {
-    return new Options().config(config);
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "LoadTPUEmbeddingMDLAdagradLightParameters";
-  
-  private LoadTPUEmbeddingMDLAdagradLightParameters(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingMomentumParameters.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingMomentumParameters.java
deleted file mode 100644
index 27341769cf3..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingMomentumParameters.java
+++ /dev/null
@@ -1,139 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.tpu;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-
-/**
- * Load Momentum embedding parameters.
- * <p>
- * An op that loads optimization parameters into HBM for embedding. Must be
- * preceded by a ConfigureTPUEmbeddingHost op that sets up the correct
- * embedding table configuration. For example, this op is used to install
- * parameters that are loaded from a checkpoint before a training loop is
- * executed.
- */
-public final class LoadTPUEmbeddingMomentumParameters extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.tpu.LoadTPUEmbeddingMomentumParameters}
-   */
-  public static class Options {
-    
-    /**
-     * @param tableId 
-     */
-    public Options tableId(Long tableId) {
-      this.tableId = tableId;
-      return this;
-    }
-    
-    /**
-     * @param tableName 
-     */
-    public Options tableName(String tableName) {
-      this.tableName = tableName;
-      return this;
-    }
-    
-    /**
-     * @param config 
-     */
-    public Options config(String config) {
-      this.config = config;
-      return this;
-    }
-    
-    private Long tableId;
-    private String tableName;
-    private String config;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new LoadTPUEmbeddingMomentumParameters operation.
-   * 
-   * @param scope current scope
-   * @param parameters Value of parameters used in the Momentum optimization algorithm.
-   * @param momenta Value of momenta used in the Momentum optimization algorithm.
-   * @param numShards 
-   * @param shardId 
-   * @param options carries optional attributes values
-   * @return a new instance of LoadTPUEmbeddingMomentumParameters
-   */
-  @Endpoint(describeByClass = true)
-  public static LoadTPUEmbeddingMomentumParameters create(Scope scope, Operand<TFloat32> parameters, Operand<TFloat32> momenta, Long numShards, Long shardId, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("LoadTPUEmbeddingMomentumParameters", scope.makeOpName("LoadTPUEmbeddingMomentumParameters"));
-    opBuilder.addInput(parameters.asOutput(scope));
-    opBuilder.addInput(momenta.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("num_shards", numShards);
-    opBuilder.setAttr("shard_id", shardId);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.tableId != null) {
-          opBuilder.setAttr("table_id", opts.tableId);
-        }
-        if (opts.tableName != null) {
-          opBuilder.setAttr("table_name", opts.tableName);
-        }
-        if (opts.config != null) {
-          opBuilder.setAttr("config", opts.config);
-        }
-      }
-    }
-    return new LoadTPUEmbeddingMomentumParameters(opBuilder.build());
-  }
-  
-  /**
-   * @param tableId 
-   */
-  public static Options tableId(Long tableId) {
-    return new Options().tableId(tableId);
-  }
-  
-  /**
-   * @param tableName 
-   */
-  public static Options tableName(String tableName) {
-    return new Options().tableName(tableName);
-  }
-  
-  /**
-   * @param config 
-   */
-  public static Options config(String config) {
-    return new Options().config(config);
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "LoadTPUEmbeddingMomentumParameters";
-  
-  private LoadTPUEmbeddingMomentumParameters(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingMomentumParametersGradAccumDebug.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingMomentumParametersGradAccumDebug.java
deleted file mode 100644
index 6da5dc0fe72..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingMomentumParametersGradAccumDebug.java
+++ /dev/null
@@ -1,141 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.tpu;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-
-/**
- * Load Momentum embedding parameters with debug support.
- * <p>
- * An op that loads optimization parameters into HBM for embedding. Must be
- * preceded by a ConfigureTPUEmbeddingHost op that sets up the correct
- * embedding table configuration. For example, this op is used to install
- * parameters that are loaded from a checkpoint before a training loop is
- * executed.
- */
-public final class LoadTPUEmbeddingMomentumParametersGradAccumDebug extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.tpu.LoadTPUEmbeddingMomentumParametersGradAccumDebug}
-   */
-  public static class Options {
-    
-    /**
-     * @param tableId 
-     */
-    public Options tableId(Long tableId) {
-      this.tableId = tableId;
-      return this;
-    }
-    
-    /**
-     * @param tableName 
-     */
-    public Options tableName(String tableName) {
-      this.tableName = tableName;
-      return this;
-    }
-    
-    /**
-     * @param config 
-     */
-    public Options config(String config) {
-      this.config = config;
-      return this;
-    }
-    
-    private Long tableId;
-    private String tableName;
-    private String config;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new LoadTPUEmbeddingMomentumParametersGradAccumDebug operation.
-   * 
-   * @param scope current scope
-   * @param parameters Value of parameters used in the Momentum optimization algorithm.
-   * @param momenta Value of momenta used in the Momentum optimization algorithm.
-   * @param gradientAccumulators Value of gradient_accumulators used in the Momentum optimization algorithm.
-   * @param numShards 
-   * @param shardId 
-   * @param options carries optional attributes values
-   * @return a new instance of LoadTPUEmbeddingMomentumParametersGradAccumDebug
-   */
-  @Endpoint(describeByClass = true)
-  public static LoadTPUEmbeddingMomentumParametersGradAccumDebug create(Scope scope, Operand<TFloat32> parameters, Operand<TFloat32> momenta, Operand<TFloat32> gradientAccumulators, Long numShards, Long shardId, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("LoadTPUEmbeddingMomentumParametersGradAccumDebug", scope.makeOpName("LoadTPUEmbeddingMomentumParametersGradAccumDebug"));
-    opBuilder.addInput(parameters.asOutput(scope));
-    opBuilder.addInput(momenta.asOutput(scope));
-    opBuilder.addInput(gradientAccumulators.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("num_shards", numShards);
-    opBuilder.setAttr("shard_id", shardId);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.tableId != null) {
-          opBuilder.setAttr("table_id", opts.tableId);
-        }
-        if (opts.tableName != null) {
-          opBuilder.setAttr("table_name", opts.tableName);
-        }
-        if (opts.config != null) {
-          opBuilder.setAttr("config", opts.config);
-        }
-      }
-    }
-    return new LoadTPUEmbeddingMomentumParametersGradAccumDebug(opBuilder.build());
-  }
-  
-  /**
-   * @param tableId 
-   */
-  public static Options tableId(Long tableId) {
-    return new Options().tableId(tableId);
-  }
-  
-  /**
-   * @param tableName 
-   */
-  public static Options tableName(String tableName) {
-    return new Options().tableName(tableName);
-  }
-  
-  /**
-   * @param config 
-   */
-  public static Options config(String config) {
-    return new Options().config(config);
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "LoadTPUEmbeddingMomentumParametersGradAccumDebug";
-  
-  private LoadTPUEmbeddingMomentumParametersGradAccumDebug(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingProximalAdagradParameters.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingProximalAdagradParameters.java
deleted file mode 100644
index 9b332186332..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingProximalAdagradParameters.java
+++ /dev/null
@@ -1,139 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.tpu;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-
-/**
- * Load proximal Adagrad embedding parameters.
- * <p>
- * An op that loads optimization parameters into HBM for embedding. Must be
- * preceded by a ConfigureTPUEmbeddingHost op that sets up the correct
- * embedding table configuration. For example, this op is used to install
- * parameters that are loaded from a checkpoint before a training loop is
- * executed.
- */
-public final class LoadTPUEmbeddingProximalAdagradParameters extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.tpu.LoadTPUEmbeddingProximalAdagradParameters}
-   */
-  public static class Options {
-    
-    /**
-     * @param tableId 
-     */
-    public Options tableId(Long tableId) {
-      this.tableId = tableId;
-      return this;
-    }
-    
-    /**
-     * @param tableName 
-     */
-    public Options tableName(String tableName) {
-      this.tableName = tableName;
-      return this;
-    }
-    
-    /**
-     * @param config 
-     */
-    public Options config(String config) {
-      this.config = config;
-      return this;
-    }
-    
-    private Long tableId;
-    private String tableName;
-    private String config;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new LoadTPUEmbeddingProximalAdagradParameters operation.
-   * 
-   * @param scope current scope
-   * @param parameters Value of parameters used in the proximal Adagrad optimization algorithm.
-   * @param accumulators Value of accumulators used in the proximal Adagrad optimization algorithm.
-   * @param numShards 
-   * @param shardId 
-   * @param options carries optional attributes values
-   * @return a new instance of LoadTPUEmbeddingProximalAdagradParameters
-   */
-  @Endpoint(describeByClass = true)
-  public static LoadTPUEmbeddingProximalAdagradParameters create(Scope scope, Operand<TFloat32> parameters, Operand<TFloat32> accumulators, Long numShards, Long shardId, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("LoadTPUEmbeddingProximalAdagradParameters", scope.makeOpName("LoadTPUEmbeddingProximalAdagradParameters"));
-    opBuilder.addInput(parameters.asOutput(scope));
-    opBuilder.addInput(accumulators.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("num_shards", numShards);
-    opBuilder.setAttr("shard_id", shardId);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.tableId != null) {
-          opBuilder.setAttr("table_id", opts.tableId);
-        }
-        if (opts.tableName != null) {
-          opBuilder.setAttr("table_name", opts.tableName);
-        }
-        if (opts.config != null) {
-          opBuilder.setAttr("config", opts.config);
-        }
-      }
-    }
-    return new LoadTPUEmbeddingProximalAdagradParameters(opBuilder.build());
-  }
-  
-  /**
-   * @param tableId 
-   */
-  public static Options tableId(Long tableId) {
-    return new Options().tableId(tableId);
-  }
-  
-  /**
-   * @param tableName 
-   */
-  public static Options tableName(String tableName) {
-    return new Options().tableName(tableName);
-  }
-  
-  /**
-   * @param config 
-   */
-  public static Options config(String config) {
-    return new Options().config(config);
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "LoadTPUEmbeddingProximalAdagradParameters";
-  
-  private LoadTPUEmbeddingProximalAdagradParameters(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingProximalAdagradParametersGradAccumDebug.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingProximalAdagradParametersGradAccumDebug.java
deleted file mode 100644
index 76c8193545b..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingProximalAdagradParametersGradAccumDebug.java
+++ /dev/null
@@ -1,141 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.tpu;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-
-/**
- * Load proximal Adagrad embedding parameters with debug support.
- * <p>
- * An op that loads optimization parameters into HBM for embedding. Must be
- * preceded by a ConfigureTPUEmbeddingHost op that sets up the correct
- * embedding table configuration. For example, this op is used to install
- * parameters that are loaded from a checkpoint before a training loop is
- * executed.
- */
-public final class LoadTPUEmbeddingProximalAdagradParametersGradAccumDebug extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.tpu.LoadTPUEmbeddingProximalAdagradParametersGradAccumDebug}
-   */
-  public static class Options {
-    
-    /**
-     * @param tableId 
-     */
-    public Options tableId(Long tableId) {
-      this.tableId = tableId;
-      return this;
-    }
-    
-    /**
-     * @param tableName 
-     */
-    public Options tableName(String tableName) {
-      this.tableName = tableName;
-      return this;
-    }
-    
-    /**
-     * @param config 
-     */
-    public Options config(String config) {
-      this.config = config;
-      return this;
-    }
-    
-    private Long tableId;
-    private String tableName;
-    private String config;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new LoadTPUEmbeddingProximalAdagradParametersGradAccumDebug operation.
-   * 
-   * @param scope current scope
-   * @param parameters Value of parameters used in the proximal Adagrad optimization algorithm.
-   * @param accumulators Value of accumulators used in the proximal Adagrad optimization algorithm.
-   * @param gradientAccumulators Value of gradient_accumulators used in the proximal Adagrad optimization algorithm.
-   * @param numShards 
-   * @param shardId 
-   * @param options carries optional attributes values
-   * @return a new instance of LoadTPUEmbeddingProximalAdagradParametersGradAccumDebug
-   */
-  @Endpoint(describeByClass = true)
-  public static LoadTPUEmbeddingProximalAdagradParametersGradAccumDebug create(Scope scope, Operand<TFloat32> parameters, Operand<TFloat32> accumulators, Operand<TFloat32> gradientAccumulators, Long numShards, Long shardId, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("LoadTPUEmbeddingProximalAdagradParametersGradAccumDebug", scope.makeOpName("LoadTPUEmbeddingProximalAdagradParametersGradAccumDebug"));
-    opBuilder.addInput(parameters.asOutput(scope));
-    opBuilder.addInput(accumulators.asOutput(scope));
-    opBuilder.addInput(gradientAccumulators.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("num_shards", numShards);
-    opBuilder.setAttr("shard_id", shardId);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.tableId != null) {
-          opBuilder.setAttr("table_id", opts.tableId);
-        }
-        if (opts.tableName != null) {
-          opBuilder.setAttr("table_name", opts.tableName);
-        }
-        if (opts.config != null) {
-          opBuilder.setAttr("config", opts.config);
-        }
-      }
-    }
-    return new LoadTPUEmbeddingProximalAdagradParametersGradAccumDebug(opBuilder.build());
-  }
-  
-  /**
-   * @param tableId 
-   */
-  public static Options tableId(Long tableId) {
-    return new Options().tableId(tableId);
-  }
-  
-  /**
-   * @param tableName 
-   */
-  public static Options tableName(String tableName) {
-    return new Options().tableName(tableName);
-  }
-  
-  /**
-   * @param config 
-   */
-  public static Options config(String config) {
-    return new Options().config(config);
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "LoadTPUEmbeddingProximalAdagradParametersGradAccumDebug";
-  
-  private LoadTPUEmbeddingProximalAdagradParametersGradAccumDebug(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingProximalYogiParameters.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingProximalYogiParameters.java
deleted file mode 100644
index 233c64be04b..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingProximalYogiParameters.java
+++ /dev/null
@@ -1,134 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.tpu;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-
-/**
- */
-public final class LoadTPUEmbeddingProximalYogiParameters extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.tpu.LoadTPUEmbeddingProximalYogiParameters}
-   */
-  public static class Options {
-    
-    /**
-     * @param tableId 
-     */
-    public Options tableId(Long tableId) {
-      this.tableId = tableId;
-      return this;
-    }
-    
-    /**
-     * @param tableName 
-     */
-    public Options tableName(String tableName) {
-      this.tableName = tableName;
-      return this;
-    }
-    
-    /**
-     * @param config 
-     */
-    public Options config(String config) {
-      this.config = config;
-      return this;
-    }
-    
-    private Long tableId;
-    private String tableName;
-    private String config;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new LoadTPUEmbeddingProximalYogiParameters operation.
-   * 
-   * @param scope current scope
-   * @param parameters 
-   * @param v 
-   * @param m 
-   * @param numShards 
-   * @param shardId 
-   * @param options carries optional attributes values
-   * @return a new instance of LoadTPUEmbeddingProximalYogiParameters
-   */
-  @Endpoint(describeByClass = true)
-  public static LoadTPUEmbeddingProximalYogiParameters create(Scope scope, Operand<TFloat32> parameters, Operand<TFloat32> v, Operand<TFloat32> m, Long numShards, Long shardId, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("LoadTPUEmbeddingProximalYogiParameters", scope.makeOpName("LoadTPUEmbeddingProximalYogiParameters"));
-    opBuilder.addInput(parameters.asOutput(scope));
-    opBuilder.addInput(v.asOutput(scope));
-    opBuilder.addInput(m.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("num_shards", numShards);
-    opBuilder.setAttr("shard_id", shardId);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.tableId != null) {
-          opBuilder.setAttr("table_id", opts.tableId);
-        }
-        if (opts.tableName != null) {
-          opBuilder.setAttr("table_name", opts.tableName);
-        }
-        if (opts.config != null) {
-          opBuilder.setAttr("config", opts.config);
-        }
-      }
-    }
-    return new LoadTPUEmbeddingProximalYogiParameters(opBuilder.build());
-  }
-  
-  /**
-   * @param tableId 
-   */
-  public static Options tableId(Long tableId) {
-    return new Options().tableId(tableId);
-  }
-  
-  /**
-   * @param tableName 
-   */
-  public static Options tableName(String tableName) {
-    return new Options().tableName(tableName);
-  }
-  
-  /**
-   * @param config 
-   */
-  public static Options config(String config) {
-    return new Options().config(config);
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "LoadTPUEmbeddingProximalYogiParameters";
-  
-  private LoadTPUEmbeddingProximalYogiParameters(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingProximalYogiParametersGradAccumDebug.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingProximalYogiParametersGradAccumDebug.java
deleted file mode 100644
index 57160686c95..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingProximalYogiParametersGradAccumDebug.java
+++ /dev/null
@@ -1,136 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.tpu;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-
-/**
- */
-public final class LoadTPUEmbeddingProximalYogiParametersGradAccumDebug extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.tpu.LoadTPUEmbeddingProximalYogiParametersGradAccumDebug}
-   */
-  public static class Options {
-    
-    /**
-     * @param tableId 
-     */
-    public Options tableId(Long tableId) {
-      this.tableId = tableId;
-      return this;
-    }
-    
-    /**
-     * @param tableName 
-     */
-    public Options tableName(String tableName) {
-      this.tableName = tableName;
-      return this;
-    }
-    
-    /**
-     * @param config 
-     */
-    public Options config(String config) {
-      this.config = config;
-      return this;
-    }
-    
-    private Long tableId;
-    private String tableName;
-    private String config;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new LoadTPUEmbeddingProximalYogiParametersGradAccumDebug operation.
-   * 
-   * @param scope current scope
-   * @param parameters 
-   * @param v 
-   * @param m 
-   * @param gradientAccumulators 
-   * @param numShards 
-   * @param shardId 
-   * @param options carries optional attributes values
-   * @return a new instance of LoadTPUEmbeddingProximalYogiParametersGradAccumDebug
-   */
-  @Endpoint(describeByClass = true)
-  public static LoadTPUEmbeddingProximalYogiParametersGradAccumDebug create(Scope scope, Operand<TFloat32> parameters, Operand<TFloat32> v, Operand<TFloat32> m, Operand<TFloat32> gradientAccumulators, Long numShards, Long shardId, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("LoadTPUEmbeddingProximalYogiParametersGradAccumDebug", scope.makeOpName("LoadTPUEmbeddingProximalYogiParametersGradAccumDebug"));
-    opBuilder.addInput(parameters.asOutput(scope));
-    opBuilder.addInput(v.asOutput(scope));
-    opBuilder.addInput(m.asOutput(scope));
-    opBuilder.addInput(gradientAccumulators.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("num_shards", numShards);
-    opBuilder.setAttr("shard_id", shardId);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.tableId != null) {
-          opBuilder.setAttr("table_id", opts.tableId);
-        }
-        if (opts.tableName != null) {
-          opBuilder.setAttr("table_name", opts.tableName);
-        }
-        if (opts.config != null) {
-          opBuilder.setAttr("config", opts.config);
-        }
-      }
-    }
-    return new LoadTPUEmbeddingProximalYogiParametersGradAccumDebug(opBuilder.build());
-  }
-  
-  /**
-   * @param tableId 
-   */
-  public static Options tableId(Long tableId) {
-    return new Options().tableId(tableId);
-  }
-  
-  /**
-   * @param tableName 
-   */
-  public static Options tableName(String tableName) {
-    return new Options().tableName(tableName);
-  }
-  
-  /**
-   * @param config 
-   */
-  public static Options config(String config) {
-    return new Options().config(config);
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "LoadTPUEmbeddingProximalYogiParametersGradAccumDebug";
-  
-  private LoadTPUEmbeddingProximalYogiParametersGradAccumDebug(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingRMSPropParameters.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingRMSPropParameters.java
deleted file mode 100644
index 83753ee8b88..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingRMSPropParameters.java
+++ /dev/null
@@ -1,141 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.tpu;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-
-/**
- * Load RMSProp embedding parameters.
- * <p>
- * An op that loads optimization parameters into HBM for embedding. Must be
- * preceded by a ConfigureTPUEmbeddingHost op that sets up the correct
- * embedding table configuration. For example, this op is used to install
- * parameters that are loaded from a checkpoint before a training loop is
- * executed.
- */
-public final class LoadTPUEmbeddingRMSPropParameters extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.tpu.LoadTPUEmbeddingRMSPropParameters}
-   */
-  public static class Options {
-    
-    /**
-     * @param tableId 
-     */
-    public Options tableId(Long tableId) {
-      this.tableId = tableId;
-      return this;
-    }
-    
-    /**
-     * @param tableName 
-     */
-    public Options tableName(String tableName) {
-      this.tableName = tableName;
-      return this;
-    }
-    
-    /**
-     * @param config 
-     */
-    public Options config(String config) {
-      this.config = config;
-      return this;
-    }
-    
-    private Long tableId;
-    private String tableName;
-    private String config;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new LoadTPUEmbeddingRMSPropParameters operation.
-   * 
-   * @param scope current scope
-   * @param parameters Value of parameters used in the RMSProp optimization algorithm.
-   * @param ms Value of ms used in the RMSProp optimization algorithm.
-   * @param mom Value of mom used in the RMSProp optimization algorithm.
-   * @param numShards 
-   * @param shardId 
-   * @param options carries optional attributes values
-   * @return a new instance of LoadTPUEmbeddingRMSPropParameters
-   */
-  @Endpoint(describeByClass = true)
-  public static LoadTPUEmbeddingRMSPropParameters create(Scope scope, Operand<TFloat32> parameters, Operand<TFloat32> ms, Operand<TFloat32> mom, Long numShards, Long shardId, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("LoadTPUEmbeddingRMSPropParameters", scope.makeOpName("LoadTPUEmbeddingRMSPropParameters"));
-    opBuilder.addInput(parameters.asOutput(scope));
-    opBuilder.addInput(ms.asOutput(scope));
-    opBuilder.addInput(mom.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("num_shards", numShards);
-    opBuilder.setAttr("shard_id", shardId);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.tableId != null) {
-          opBuilder.setAttr("table_id", opts.tableId);
-        }
-        if (opts.tableName != null) {
-          opBuilder.setAttr("table_name", opts.tableName);
-        }
-        if (opts.config != null) {
-          opBuilder.setAttr("config", opts.config);
-        }
-      }
-    }
-    return new LoadTPUEmbeddingRMSPropParameters(opBuilder.build());
-  }
-  
-  /**
-   * @param tableId 
-   */
-  public static Options tableId(Long tableId) {
-    return new Options().tableId(tableId);
-  }
-  
-  /**
-   * @param tableName 
-   */
-  public static Options tableName(String tableName) {
-    return new Options().tableName(tableName);
-  }
-  
-  /**
-   * @param config 
-   */
-  public static Options config(String config) {
-    return new Options().config(config);
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "LoadTPUEmbeddingRMSPropParameters";
-  
-  private LoadTPUEmbeddingRMSPropParameters(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingRMSPropParametersGradAccumDebug.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingRMSPropParametersGradAccumDebug.java
deleted file mode 100644
index f562849e113..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingRMSPropParametersGradAccumDebug.java
+++ /dev/null
@@ -1,143 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.tpu;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-
-/**
- * Load RMSProp embedding parameters with debug support.
- * <p>
- * An op that loads optimization parameters into HBM for embedding. Must be
- * preceded by a ConfigureTPUEmbeddingHost op that sets up the correct
- * embedding table configuration. For example, this op is used to install
- * parameters that are loaded from a checkpoint before a training loop is
- * executed.
- */
-public final class LoadTPUEmbeddingRMSPropParametersGradAccumDebug extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.tpu.LoadTPUEmbeddingRMSPropParametersGradAccumDebug}
-   */
-  public static class Options {
-    
-    /**
-     * @param tableId 
-     */
-    public Options tableId(Long tableId) {
-      this.tableId = tableId;
-      return this;
-    }
-    
-    /**
-     * @param tableName 
-     */
-    public Options tableName(String tableName) {
-      this.tableName = tableName;
-      return this;
-    }
-    
-    /**
-     * @param config 
-     */
-    public Options config(String config) {
-      this.config = config;
-      return this;
-    }
-    
-    private Long tableId;
-    private String tableName;
-    private String config;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new LoadTPUEmbeddingRMSPropParametersGradAccumDebug operation.
-   * 
-   * @param scope current scope
-   * @param parameters Value of parameters used in the RMSProp optimization algorithm.
-   * @param ms Value of ms used in the RMSProp optimization algorithm.
-   * @param mom Value of mom used in the RMSProp optimization algorithm.
-   * @param gradientAccumulators Value of gradient_accumulators used in the RMSProp optimization algorithm.
-   * @param numShards 
-   * @param shardId 
-   * @param options carries optional attributes values
-   * @return a new instance of LoadTPUEmbeddingRMSPropParametersGradAccumDebug
-   */
-  @Endpoint(describeByClass = true)
-  public static LoadTPUEmbeddingRMSPropParametersGradAccumDebug create(Scope scope, Operand<TFloat32> parameters, Operand<TFloat32> ms, Operand<TFloat32> mom, Operand<TFloat32> gradientAccumulators, Long numShards, Long shardId, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("LoadTPUEmbeddingRMSPropParametersGradAccumDebug", scope.makeOpName("LoadTPUEmbeddingRMSPropParametersGradAccumDebug"));
-    opBuilder.addInput(parameters.asOutput(scope));
-    opBuilder.addInput(ms.asOutput(scope));
-    opBuilder.addInput(mom.asOutput(scope));
-    opBuilder.addInput(gradientAccumulators.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("num_shards", numShards);
-    opBuilder.setAttr("shard_id", shardId);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.tableId != null) {
-          opBuilder.setAttr("table_id", opts.tableId);
-        }
-        if (opts.tableName != null) {
-          opBuilder.setAttr("table_name", opts.tableName);
-        }
-        if (opts.config != null) {
-          opBuilder.setAttr("config", opts.config);
-        }
-      }
-    }
-    return new LoadTPUEmbeddingRMSPropParametersGradAccumDebug(opBuilder.build());
-  }
-  
-  /**
-   * @param tableId 
-   */
-  public static Options tableId(Long tableId) {
-    return new Options().tableId(tableId);
-  }
-  
-  /**
-   * @param tableName 
-   */
-  public static Options tableName(String tableName) {
-    return new Options().tableName(tableName);
-  }
-  
-  /**
-   * @param config 
-   */
-  public static Options config(String config) {
-    return new Options().config(config);
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "LoadTPUEmbeddingRMSPropParametersGradAccumDebug";
-  
-  private LoadTPUEmbeddingRMSPropParametersGradAccumDebug(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingStochasticGradientDescentParameters.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingStochasticGradientDescentParameters.java
deleted file mode 100644
index 5c3fb5d662d..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingStochasticGradientDescentParameters.java
+++ /dev/null
@@ -1,137 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.tpu;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-
-/**
- * Load SGD embedding parameters.
- * <p>
- * An op that loads optimization parameters into HBM for embedding. Must be
- * preceded by a ConfigureTPUEmbeddingHost op that sets up the correct
- * embedding table configuration. For example, this op is used to install
- * parameters that are loaded from a checkpoint before a training loop is
- * executed.
- */
-public final class LoadTPUEmbeddingStochasticGradientDescentParameters extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.tpu.LoadTPUEmbeddingStochasticGradientDescentParameters}
-   */
-  public static class Options {
-    
-    /**
-     * @param tableId 
-     */
-    public Options tableId(Long tableId) {
-      this.tableId = tableId;
-      return this;
-    }
-    
-    /**
-     * @param tableName 
-     */
-    public Options tableName(String tableName) {
-      this.tableName = tableName;
-      return this;
-    }
-    
-    /**
-     * @param config 
-     */
-    public Options config(String config) {
-      this.config = config;
-      return this;
-    }
-    
-    private Long tableId;
-    private String tableName;
-    private String config;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new LoadTPUEmbeddingStochasticGradientDescentParameters operation.
-   * 
-   * @param scope current scope
-   * @param parameters Value of parameters used in the stochastic gradient descent optimization algorithm.
-   * @param numShards 
-   * @param shardId 
-   * @param options carries optional attributes values
-   * @return a new instance of LoadTPUEmbeddingStochasticGradientDescentParameters
-   */
-  @Endpoint(describeByClass = true)
-  public static LoadTPUEmbeddingStochasticGradientDescentParameters create(Scope scope, Operand<TFloat32> parameters, Long numShards, Long shardId, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("LoadTPUEmbeddingStochasticGradientDescentParameters", scope.makeOpName("LoadTPUEmbeddingStochasticGradientDescentParameters"));
-    opBuilder.addInput(parameters.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("num_shards", numShards);
-    opBuilder.setAttr("shard_id", shardId);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.tableId != null) {
-          opBuilder.setAttr("table_id", opts.tableId);
-        }
-        if (opts.tableName != null) {
-          opBuilder.setAttr("table_name", opts.tableName);
-        }
-        if (opts.config != null) {
-          opBuilder.setAttr("config", opts.config);
-        }
-      }
-    }
-    return new LoadTPUEmbeddingStochasticGradientDescentParameters(opBuilder.build());
-  }
-  
-  /**
-   * @param tableId 
-   */
-  public static Options tableId(Long tableId) {
-    return new Options().tableId(tableId);
-  }
-  
-  /**
-   * @param tableName 
-   */
-  public static Options tableName(String tableName) {
-    return new Options().tableName(tableName);
-  }
-  
-  /**
-   * @param config 
-   */
-  public static Options config(String config) {
-    return new Options().config(config);
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "LoadTPUEmbeddingStochasticGradientDescentParameters";
-  
-  private LoadTPUEmbeddingStochasticGradientDescentParameters(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingStochasticGradientDescentParametersGradAccumDebug.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingStochasticGradientDescentParametersGradAccumDebug.java
deleted file mode 100644
index 4f22a6b9996..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingStochasticGradientDescentParametersGradAccumDebug.java
+++ /dev/null
@@ -1,139 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.tpu;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-
-/**
- * Load SGD embedding parameters.
- * <p>
- * An op that loads optimization parameters into HBM for embedding. Must be
- * preceded by a ConfigureTPUEmbeddingHost op that sets up the correct
- * embedding table configuration. For example, this op is used to install
- * parameters that are loaded from a checkpoint before a training loop is
- * executed.
- */
-public final class LoadTPUEmbeddingStochasticGradientDescentParametersGradAccumDebug extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.tpu.LoadTPUEmbeddingStochasticGradientDescentParametersGradAccumDebug}
-   */
-  public static class Options {
-    
-    /**
-     * @param tableId 
-     */
-    public Options tableId(Long tableId) {
-      this.tableId = tableId;
-      return this;
-    }
-    
-    /**
-     * @param tableName 
-     */
-    public Options tableName(String tableName) {
-      this.tableName = tableName;
-      return this;
-    }
-    
-    /**
-     * @param config 
-     */
-    public Options config(String config) {
-      this.config = config;
-      return this;
-    }
-    
-    private Long tableId;
-    private String tableName;
-    private String config;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new LoadTPUEmbeddingStochasticGradientDescentParametersGradAccumDebug operation.
-   * 
-   * @param scope current scope
-   * @param parameters Value of parameters used in the stochastic gradient descent optimization algorithm.
-   * @param gradientAccumulators Value of gradient_accumulators used in the Adadelta optimization algorithm.
-   * @param numShards 
-   * @param shardId 
-   * @param options carries optional attributes values
-   * @return a new instance of LoadTPUEmbeddingStochasticGradientDescentParametersGradAccumDebug
-   */
-  @Endpoint(describeByClass = true)
-  public static LoadTPUEmbeddingStochasticGradientDescentParametersGradAccumDebug create(Scope scope, Operand<TFloat32> parameters, Operand<TFloat32> gradientAccumulators, Long numShards, Long shardId, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("LoadTPUEmbeddingStochasticGradientDescentParametersGradAccumDebug", scope.makeOpName("LoadTPUEmbeddingStochasticGradientDescentParametersGradAccumDebug"));
-    opBuilder.addInput(parameters.asOutput(scope));
-    opBuilder.addInput(gradientAccumulators.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("num_shards", numShards);
-    opBuilder.setAttr("shard_id", shardId);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.tableId != null) {
-          opBuilder.setAttr("table_id", opts.tableId);
-        }
-        if (opts.tableName != null) {
-          opBuilder.setAttr("table_name", opts.tableName);
-        }
-        if (opts.config != null) {
-          opBuilder.setAttr("config", opts.config);
-        }
-      }
-    }
-    return new LoadTPUEmbeddingStochasticGradientDescentParametersGradAccumDebug(opBuilder.build());
-  }
-  
-  /**
-   * @param tableId 
-   */
-  public static Options tableId(Long tableId) {
-    return new Options().tableId(tableId);
-  }
-  
-  /**
-   * @param tableName 
-   */
-  public static Options tableName(String tableName) {
-    return new Options().tableName(tableName);
-  }
-  
-  /**
-   * @param config 
-   */
-  public static Options config(String config) {
-    return new Options().config(config);
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "LoadTPUEmbeddingStochasticGradientDescentParametersGradAccumDebug";
-  
-  private LoadTPUEmbeddingStochasticGradientDescentParametersGradAccumDebug(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/OutfeedDequeue.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/OutfeedDequeue.java
deleted file mode 100644
index c61ed2d5bad..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/OutfeedDequeue.java
+++ /dev/null
@@ -1,117 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.tpu;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Retrieves a single tensor from the computation outfeed.
- * <p>
- * This operation will block indefinitely until data is available.
- * 
- * @param <T> data type for {@code output()} output
- */
-public final class OutfeedDequeue<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.tpu.OutfeedDequeue}
-   */
-  public static class Options {
-    
-    /**
-     * @param deviceOrdinal The TPU device to use. This should be -1 when the Op
-     * is running on a TPU device, and >= 0 when the Op is running on the CPU
-     * device.
-     */
-    public Options deviceOrdinal(Long deviceOrdinal) {
-      this.deviceOrdinal = deviceOrdinal;
-      return this;
-    }
-    
-    private Long deviceOrdinal;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new OutfeedDequeue operation.
-   * 
-   * @param scope current scope
-   * @param dtype The type of elements in the tensor.
-   * @param shape The shape of the tensor.
-   * @param options carries optional attributes values
-   * @return a new instance of OutfeedDequeue
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> OutfeedDequeue<T> create(Scope scope, Class<T> dtype, Shape shape, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("OutfeedDequeue", scope.makeOpName("OutfeedDequeue"));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("dtype", dtype);
-    opBuilder.setAttr("shape", shape);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.deviceOrdinal != null) {
-          opBuilder.setAttr("device_ordinal", opts.deviceOrdinal);
-        }
-      }
-    }
-    return new OutfeedDequeue<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param deviceOrdinal The TPU device to use. This should be -1 when the Op
-   * is running on a TPU device, and >= 0 when the Op is running on the CPU
-   * device.
-   */
-  public static Options deviceOrdinal(Long deviceOrdinal) {
-    return new Options().deviceOrdinal(deviceOrdinal);
-  }
-  
-  /**
-   * A tensor that will be read from the device outfeed.
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "OutfeedDequeue";
-  
-  private Output<T> output;
-  
-  private OutfeedDequeue(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/OutfeedDequeueTuple.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/OutfeedDequeueTuple.java
deleted file mode 100644
index 068423a6f80..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/OutfeedDequeueTuple.java
+++ /dev/null
@@ -1,130 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.tpu;
-
-import java.util.Arrays;
-import java.util.Iterator;
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Retrieve multiple values from the computation outfeed.
- * <p>
- * This operation will block indefinitely until data is available. Output `i`
- * corresponds to XLA tuple element `i`.
- */
-public final class OutfeedDequeueTuple extends RawOp implements Iterable<Operand<TType>> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.tpu.OutfeedDequeueTuple}
-   */
-  public static class Options {
-    
-    /**
-     * @param deviceOrdinal The TPU device to use. This should be -1 when the Op
-     * is running on a TPU device, and >= 0 when the Op is running on the CPU
-     * device.
-     */
-    public Options deviceOrdinal(Long deviceOrdinal) {
-      this.deviceOrdinal = deviceOrdinal;
-      return this;
-    }
-    
-    private Long deviceOrdinal;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new OutfeedDequeueTuple operation.
-   * 
-   * @param scope current scope
-   * @param dtypes The element types of each element in `outputs`.
-   * @param shapes The shapes of each tensor in `outputs`.
-   * @param options carries optional attributes values
-   * @return a new instance of OutfeedDequeueTuple
-   */
-  @Endpoint(describeByClass = true)
-  public static OutfeedDequeueTuple create(Scope scope, List<Class<?>> dtypes, List<Shape> shapes, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("OutfeedDequeueTuple", scope.makeOpName("OutfeedDequeueTuple"));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    Class[] dtypesArray = new Class[dtypes.size()];
-    for (int i = 0; i < dtypesArray.length; ++i) {
-      dtypesArray[i] = dtypes.get(i);
-    }
-    opBuilder.setAttr("dtypes", dtypesArray);
-    Shape[] shapesArray = new Shape[shapes.size()];
-    for (int i = 0; i < shapesArray.length; ++i) {
-      shapesArray[i] = shapes.get(i);
-    }
-    opBuilder.setAttr("shapes", shapesArray);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.deviceOrdinal != null) {
-          opBuilder.setAttr("device_ordinal", opts.deviceOrdinal);
-        }
-      }
-    }
-    return new OutfeedDequeueTuple(opBuilder.build());
-  }
-  
-  /**
-   * @param deviceOrdinal The TPU device to use. This should be -1 when the Op
-   * is running on a TPU device, and >= 0 when the Op is running on the CPU
-   * device.
-   */
-  public static Options deviceOrdinal(Long deviceOrdinal) {
-    return new Options().deviceOrdinal(deviceOrdinal);
-  }
-  
-  /**
-   * A list of tensors that will be read from the outfeed.
-   */
-  public List<Output<?>> outputs() {
-    return outputs;
-  }
-  
-  @Override
-  @SuppressWarnings({"rawtypes", "unchecked"})
-  public Iterator<Operand<TType>> iterator() {
-    return (Iterator) outputs.iterator();
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "OutfeedDequeueTuple";
-  
-  private List<Output<?>> outputs;
-  
-  private OutfeedDequeueTuple(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    int outputsLength = operation.outputListLength("outputs");
-    outputs = Arrays.asList(operation.outputList(outputIdx, outputsLength));
-    outputIdx += outputsLength;
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/OutfeedEnqueue.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/OutfeedEnqueue.java
deleted file mode 100644
index 2cae7229fde..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/OutfeedEnqueue.java
+++ /dev/null
@@ -1,55 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.tpu;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Enqueue a Tensor on the computation outfeed.
- */
-public final class OutfeedEnqueue extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new OutfeedEnqueue operation.
-   * 
-   * @param scope current scope
-   * @param input A tensor that will be inserted into the outfeed queue.
-   * @return a new instance of OutfeedEnqueue
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> OutfeedEnqueue create(Scope scope, Operand<T> input) {
-    OperationBuilder opBuilder = scope.env().opBuilder("OutfeedEnqueue", scope.makeOpName("OutfeedEnqueue"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new OutfeedEnqueue(opBuilder.build());
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "OutfeedEnqueue";
-  
-  private OutfeedEnqueue(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/OutfeedEnqueueTuple.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/OutfeedEnqueueTuple.java
deleted file mode 100644
index 761b7f856ac..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/OutfeedEnqueueTuple.java
+++ /dev/null
@@ -1,56 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.tpu;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.Operands;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-
-/**
- * Enqueue multiple Tensor values on the computation outfeed.
- */
-public final class OutfeedEnqueueTuple extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new OutfeedEnqueueTuple operation.
-   * 
-   * @param scope current scope
-   * @param inputs A list of tensors that will be inserted into the outfeed queue as an
-   * XLA tuple.
-   * @return a new instance of OutfeedEnqueueTuple
-   */
-  @Endpoint(describeByClass = true)
-  public static OutfeedEnqueueTuple create(Scope scope, Iterable<Operand<?>> inputs) {
-    OperationBuilder opBuilder = scope.env().opBuilder("OutfeedEnqueueTuple", scope.makeOpName("OutfeedEnqueueTuple"));
-    opBuilder.addInputList(Operands.asOutputs(scope, inputs));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new OutfeedEnqueueTuple(opBuilder.build());
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "OutfeedEnqueueTuple";
-  
-  private OutfeedEnqueueTuple(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/Prelinearize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/Prelinearize.java
deleted file mode 100644
index 42eb810042a..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/Prelinearize.java
+++ /dev/null
@@ -1,135 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.tpu;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * An op which linearizes one Tensor value to an opaque variant tensor.
- */
-public final class Prelinearize extends RawOp implements Operand<TType> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.tpu.Prelinearize}
-   */
-  public static class Options {
-    
-    /**
-     * @param shape The shape of the tensor.
-     */
-    public Options shape(Shape shape) {
-      this.shape = shape;
-      return this;
-    }
-    
-    /**
-     * @param layout A vector holding the requested layout in minor-to-major sequence. If a layout
-     * attribute is passed but its values are all -1 the layout will be computed by
-     * the infeed operation.
-     */
-    public Options layout(List<Long> layout) {
-      this.layout = layout;
-      return this;
-    }
-    
-    private Shape shape;
-    private List<Long> layout;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new Prelinearize operation.
-   * 
-   * @param scope current scope
-   * @param input A tensor that will be linearized.
-   * @param options carries optional attributes values
-   * @return a new instance of Prelinearize
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> Prelinearize create(Scope scope, Operand<T> input, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("Prelinearize", scope.makeOpName("Prelinearize"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.shape != null) {
-          opBuilder.setAttr("shape", opts.shape);
-        }
-        if (opts.layout != null) {
-          long[] layoutArray = new long[opts.layout.size()];
-          for (int i = 0; i < layoutArray.length; ++i) {
-            layoutArray[i] = opts.layout.get(i);
-          }
-          opBuilder.setAttr("layout", layoutArray);
-        }
-      }
-    }
-    return new Prelinearize(opBuilder.build());
-  }
-  
-  /**
-   * @param shape The shape of the tensor.
-   */
-  public static Options shape(Shape shape) {
-    return new Options().shape(shape);
-  }
-  
-  /**
-   * @param layout A vector holding the requested layout in minor-to-major sequence. If a layout
-   * attribute is passed but its values are all -1 the layout will be computed by
-   * the infeed operation.
-   */
-  public static Options layout(List<Long> layout) {
-    return new Options().layout(layout);
-  }
-  
-  /**
-   */
-  public Output<?> output() {
-    return output;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "Prelinearize";
-  
-  private Output<?> output;
-  
-  private Prelinearize(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/PrelinearizeTuple.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/PrelinearizeTuple.java
deleted file mode 100644
index 427efc433bb..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/PrelinearizeTuple.java
+++ /dev/null
@@ -1,125 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.tpu;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.op.Operands;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * An op which linearizes multiple Tensor values to an opaque variant tensor.
- */
-public final class PrelinearizeTuple extends RawOp implements Operand<TType> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.tpu.PrelinearizeTuple}
-   */
-  public static class Options {
-    
-    /**
-     * @param layouts A vector holding the requested layout in minor-to-major sequence for all the
-     * tuple shapes in the order the shapes appear in the "shapes" input. The layout
-     * elements for a sub-shape can be set to -1 in which case the corresponding layout
-     * will be computed by the infeed operation.
-     */
-    public Options layouts(List<Long> layouts) {
-      this.layouts = layouts;
-      return this;
-    }
-    
-    private List<Long> layouts;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new PrelinearizeTuple operation.
-   * 
-   * @param scope current scope
-   * @param inputs A list of tensors that will be provided using the infeed mechanism.
-   * @param shapes The shapes of each tensor in `inputs`.
-   * @param options carries optional attributes values
-   * @return a new instance of PrelinearizeTuple
-   */
-  @Endpoint(describeByClass = true)
-  public static PrelinearizeTuple create(Scope scope, Iterable<Operand<?>> inputs, List<Shape> shapes, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("PrelinearizeTuple", scope.makeOpName("PrelinearizeTuple"));
-    opBuilder.addInputList(Operands.asOutputs(scope, inputs));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    Shape[] shapesArray = new Shape[shapes.size()];
-    for (int i = 0; i < shapesArray.length; ++i) {
-      shapesArray[i] = shapes.get(i);
-    }
-    opBuilder.setAttr("shapes", shapesArray);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.layouts != null) {
-          long[] layoutsArray = new long[opts.layouts.size()];
-          for (int i = 0; i < layoutsArray.length; ++i) {
-            layoutsArray[i] = opts.layouts.get(i);
-          }
-          opBuilder.setAttr("layouts", layoutsArray);
-        }
-      }
-    }
-    return new PrelinearizeTuple(opBuilder.build());
-  }
-  
-  /**
-   * @param layouts A vector holding the requested layout in minor-to-major sequence for all the
-   * tuple shapes in the order the shapes appear in the "shapes" input. The layout
-   * elements for a sub-shape can be set to -1 in which case the corresponding layout
-   * will be computed by the infeed operation.
-   */
-  public static Options layouts(List<Long> layouts) {
-    return new Options().layouts(layouts);
-  }
-  
-  /**
-   */
-  public Output<?> output() {
-    return output;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "PrelinearizeTuple";
-  
-  private Output<?> output;
-  
-  private PrelinearizeTuple(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RecvTPUEmbeddingActivations.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RecvTPUEmbeddingActivations.java
deleted file mode 100644
index e2bacade157..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RecvTPUEmbeddingActivations.java
+++ /dev/null
@@ -1,90 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.tpu;
-
-import java.util.Arrays;
-import java.util.Iterator;
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-
-/**
- * An op that receives embedding activations on the TPU.
- * <p>
- * The TPU system performs the embedding lookups and aggregations specified by
- * the arguments to TPUEmbeddingEnqueue(Integer/Sparse/SparseTensor)Batch. The
- * results of these aggregations are visible to the Tensorflow Graph as the
- * outputs of a RecvTPUEmbeddingActivations op. This op returns a list containing
- * one Tensor of activations per table specified in the model. There can be at
- * most one RecvTPUEmbeddingActivations op in the TPU graph.
- */
-public final class RecvTPUEmbeddingActivations extends RawOp implements Iterable<Operand<TFloat32>> {
-  
-  /**
-   * Factory method to create a class wrapping a new RecvTPUEmbeddingActivations operation.
-   * 
-   * @param scope current scope
-   * @param numOutputs The number of output activation tensors, equal to the number of
-   * embedding tables in the model.
-   * @param config Serialized TPUEmbeddingConfiguration proto.
-   * @return a new instance of RecvTPUEmbeddingActivations
-   */
-  @Endpoint(describeByClass = true)
-  public static RecvTPUEmbeddingActivations create(Scope scope, Long numOutputs, String config) {
-    OperationBuilder opBuilder = scope.env().opBuilder("RecvTPUEmbeddingActivations", scope.makeOpName("RecvTPUEmbeddingActivations"));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("num_outputs", numOutputs);
-    opBuilder.setAttr("config", config);
-    return new RecvTPUEmbeddingActivations(opBuilder.build());
-  }
-  
-  /**
-   * A TensorList of embedding activations containing one Tensor per
-   * embedding table in the model.
-   */
-  public List<Output<TFloat32>> outputs() {
-    return outputs;
-  }
-  
-  @Override
-  @SuppressWarnings({"rawtypes", "unchecked"})
-  public Iterator<Operand<TFloat32>> iterator() {
-    return (Iterator) outputs.iterator();
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "RecvTPUEmbeddingActivations";
-  
-  private List<Output<TFloat32>> outputs;
-  
-  @SuppressWarnings("unchecked")
-  private RecvTPUEmbeddingActivations(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    int outputsLength = operation.outputListLength("outputs");
-    outputs = Arrays.asList((Output<TFloat32>[])operation.outputList(outputIdx, outputsLength));
-    outputIdx += outputsLength;
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingADAMParameters.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingADAMParameters.java
deleted file mode 100644
index 6201ed4b9eb..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingADAMParameters.java
+++ /dev/null
@@ -1,163 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.tpu;
-
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-
-/**
- * Retrieve ADAM embedding parameters.
- * <p>
- * An op that retrieves optimization parameters from embedding to host
- * memory. Must be preceded by a ConfigureTPUEmbeddingHost op that sets up
- * the correct embedding table configuration. For example, this op is
- * used to retrieve updated parameters before saving a checkpoint.
- */
-public final class RetrieveTPUEmbeddingADAMParameters extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.tpu.RetrieveTPUEmbeddingADAMParameters}
-   */
-  public static class Options {
-    
-    /**
-     * @param tableId 
-     */
-    public Options tableId(Long tableId) {
-      this.tableId = tableId;
-      return this;
-    }
-    
-    /**
-     * @param tableName 
-     */
-    public Options tableName(String tableName) {
-      this.tableName = tableName;
-      return this;
-    }
-    
-    /**
-     * @param config 
-     */
-    public Options config(String config) {
-      this.config = config;
-      return this;
-    }
-    
-    private Long tableId;
-    private String tableName;
-    private String config;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new RetrieveTPUEmbeddingADAMParameters operation.
-   * 
-   * @param scope current scope
-   * @param numShards 
-   * @param shardId 
-   * @param options carries optional attributes values
-   * @return a new instance of RetrieveTPUEmbeddingADAMParameters
-   */
-  @Endpoint(describeByClass = true)
-  public static RetrieveTPUEmbeddingADAMParameters create(Scope scope, Long numShards, Long shardId, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("RetrieveTPUEmbeddingADAMParameters", scope.makeOpName("RetrieveTPUEmbeddingADAMParameters"));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("num_shards", numShards);
-    opBuilder.setAttr("shard_id", shardId);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.tableId != null) {
-          opBuilder.setAttr("table_id", opts.tableId);
-        }
-        if (opts.tableName != null) {
-          opBuilder.setAttr("table_name", opts.tableName);
-        }
-        if (opts.config != null) {
-          opBuilder.setAttr("config", opts.config);
-        }
-      }
-    }
-    return new RetrieveTPUEmbeddingADAMParameters(opBuilder.build());
-  }
-  
-  /**
-   * @param tableId 
-   */
-  public static Options tableId(Long tableId) {
-    return new Options().tableId(tableId);
-  }
-  
-  /**
-   * @param tableName 
-   */
-  public static Options tableName(String tableName) {
-    return new Options().tableName(tableName);
-  }
-  
-  /**
-   * @param config 
-   */
-  public static Options config(String config) {
-    return new Options().config(config);
-  }
-  
-  /**
-   * Parameter parameters updated by the ADAM optimization algorithm.
-   */
-  public Output<TFloat32> parameters() {
-    return parameters;
-  }
-  
-  /**
-   * Parameter momenta updated by the ADAM optimization algorithm.
-   */
-  public Output<TFloat32> momenta() {
-    return momenta;
-  }
-  
-  /**
-   * Parameter velocities updated by the ADAM optimization algorithm.
-   */
-  public Output<TFloat32> velocities() {
-    return velocities;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "RetrieveTPUEmbeddingADAMParameters";
-  
-  private Output<TFloat32> parameters;
-  private Output<TFloat32> momenta;
-  private Output<TFloat32> velocities;
-  
-  private RetrieveTPUEmbeddingADAMParameters(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    parameters = operation.output(outputIdx++);
-    momenta = operation.output(outputIdx++);
-    velocities = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingADAMParametersGradAccumDebug.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingADAMParametersGradAccumDebug.java
deleted file mode 100644
index 423fc97541e..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingADAMParametersGradAccumDebug.java
+++ /dev/null
@@ -1,172 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.tpu;
-
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-
-/**
- * Retrieve ADAM embedding parameters with debug support.
- * <p>
- * An op that retrieves optimization parameters from embedding to host
- * memory. Must be preceded by a ConfigureTPUEmbeddingHost op that sets up
- * the correct embedding table configuration. For example, this op is
- * used to retrieve updated parameters before saving a checkpoint.
- */
-public final class RetrieveTPUEmbeddingADAMParametersGradAccumDebug extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.tpu.RetrieveTPUEmbeddingADAMParametersGradAccumDebug}
-   */
-  public static class Options {
-    
-    /**
-     * @param tableId 
-     */
-    public Options tableId(Long tableId) {
-      this.tableId = tableId;
-      return this;
-    }
-    
-    /**
-     * @param tableName 
-     */
-    public Options tableName(String tableName) {
-      this.tableName = tableName;
-      return this;
-    }
-    
-    /**
-     * @param config 
-     */
-    public Options config(String config) {
-      this.config = config;
-      return this;
-    }
-    
-    private Long tableId;
-    private String tableName;
-    private String config;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new RetrieveTPUEmbeddingADAMParametersGradAccumDebug operation.
-   * 
-   * @param scope current scope
-   * @param numShards 
-   * @param shardId 
-   * @param options carries optional attributes values
-   * @return a new instance of RetrieveTPUEmbeddingADAMParametersGradAccumDebug
-   */
-  @Endpoint(describeByClass = true)
-  public static RetrieveTPUEmbeddingADAMParametersGradAccumDebug create(Scope scope, Long numShards, Long shardId, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("RetrieveTPUEmbeddingADAMParametersGradAccumDebug", scope.makeOpName("RetrieveTPUEmbeddingADAMParametersGradAccumDebug"));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("num_shards", numShards);
-    opBuilder.setAttr("shard_id", shardId);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.tableId != null) {
-          opBuilder.setAttr("table_id", opts.tableId);
-        }
-        if (opts.tableName != null) {
-          opBuilder.setAttr("table_name", opts.tableName);
-        }
-        if (opts.config != null) {
-          opBuilder.setAttr("config", opts.config);
-        }
-      }
-    }
-    return new RetrieveTPUEmbeddingADAMParametersGradAccumDebug(opBuilder.build());
-  }
-  
-  /**
-   * @param tableId 
-   */
-  public static Options tableId(Long tableId) {
-    return new Options().tableId(tableId);
-  }
-  
-  /**
-   * @param tableName 
-   */
-  public static Options tableName(String tableName) {
-    return new Options().tableName(tableName);
-  }
-  
-  /**
-   * @param config 
-   */
-  public static Options config(String config) {
-    return new Options().config(config);
-  }
-  
-  /**
-   * Parameter parameters updated by the ADAM optimization algorithm.
-   */
-  public Output<TFloat32> parameters() {
-    return parameters;
-  }
-  
-  /**
-   * Parameter momenta updated by the ADAM optimization algorithm.
-   */
-  public Output<TFloat32> momenta() {
-    return momenta;
-  }
-  
-  /**
-   * Parameter velocities updated by the ADAM optimization algorithm.
-   */
-  public Output<TFloat32> velocities() {
-    return velocities;
-  }
-  
-  /**
-   * Parameter gradient_accumulators updated by the ADAM optimization algorithm.
-   */
-  public Output<TFloat32> gradientAccumulators() {
-    return gradientAccumulators;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "RetrieveTPUEmbeddingADAMParametersGradAccumDebug";
-  
-  private Output<TFloat32> parameters;
-  private Output<TFloat32> momenta;
-  private Output<TFloat32> velocities;
-  private Output<TFloat32> gradientAccumulators;
-  
-  private RetrieveTPUEmbeddingADAMParametersGradAccumDebug(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    parameters = operation.output(outputIdx++);
-    momenta = operation.output(outputIdx++);
-    velocities = operation.output(outputIdx++);
-    gradientAccumulators = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingAdadeltaParameters.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingAdadeltaParameters.java
deleted file mode 100644
index 9a00080e017..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingAdadeltaParameters.java
+++ /dev/null
@@ -1,163 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.tpu;
-
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-
-/**
- * Retrieve Adadelta embedding parameters.
- * <p>
- * An op that retrieves optimization parameters from embedding to host
- * memory. Must be preceded by a ConfigureTPUEmbeddingHost op that sets up
- * the correct embedding table configuration. For example, this op is
- * used to retrieve updated parameters before saving a checkpoint.
- */
-public final class RetrieveTPUEmbeddingAdadeltaParameters extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.tpu.RetrieveTPUEmbeddingAdadeltaParameters}
-   */
-  public static class Options {
-    
-    /**
-     * @param tableId 
-     */
-    public Options tableId(Long tableId) {
-      this.tableId = tableId;
-      return this;
-    }
-    
-    /**
-     * @param tableName 
-     */
-    public Options tableName(String tableName) {
-      this.tableName = tableName;
-      return this;
-    }
-    
-    /**
-     * @param config 
-     */
-    public Options config(String config) {
-      this.config = config;
-      return this;
-    }
-    
-    private Long tableId;
-    private String tableName;
-    private String config;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new RetrieveTPUEmbeddingAdadeltaParameters operation.
-   * 
-   * @param scope current scope
-   * @param numShards 
-   * @param shardId 
-   * @param options carries optional attributes values
-   * @return a new instance of RetrieveTPUEmbeddingAdadeltaParameters
-   */
-  @Endpoint(describeByClass = true)
-  public static RetrieveTPUEmbeddingAdadeltaParameters create(Scope scope, Long numShards, Long shardId, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("RetrieveTPUEmbeddingAdadeltaParameters", scope.makeOpName("RetrieveTPUEmbeddingAdadeltaParameters"));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("num_shards", numShards);
-    opBuilder.setAttr("shard_id", shardId);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.tableId != null) {
-          opBuilder.setAttr("table_id", opts.tableId);
-        }
-        if (opts.tableName != null) {
-          opBuilder.setAttr("table_name", opts.tableName);
-        }
-        if (opts.config != null) {
-          opBuilder.setAttr("config", opts.config);
-        }
-      }
-    }
-    return new RetrieveTPUEmbeddingAdadeltaParameters(opBuilder.build());
-  }
-  
-  /**
-   * @param tableId 
-   */
-  public static Options tableId(Long tableId) {
-    return new Options().tableId(tableId);
-  }
-  
-  /**
-   * @param tableName 
-   */
-  public static Options tableName(String tableName) {
-    return new Options().tableName(tableName);
-  }
-  
-  /**
-   * @param config 
-   */
-  public static Options config(String config) {
-    return new Options().config(config);
-  }
-  
-  /**
-   * Parameter parameters updated by the Adadelta optimization algorithm.
-   */
-  public Output<TFloat32> parameters() {
-    return parameters;
-  }
-  
-  /**
-   * Parameter accumulators updated by the Adadelta optimization algorithm.
-   */
-  public Output<TFloat32> accumulators() {
-    return accumulators;
-  }
-  
-  /**
-   * Parameter updates updated by the Adadelta optimization algorithm.
-   */
-  public Output<TFloat32> updates() {
-    return updates;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "RetrieveTPUEmbeddingAdadeltaParameters";
-  
-  private Output<TFloat32> parameters;
-  private Output<TFloat32> accumulators;
-  private Output<TFloat32> updates;
-  
-  private RetrieveTPUEmbeddingAdadeltaParameters(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    parameters = operation.output(outputIdx++);
-    accumulators = operation.output(outputIdx++);
-    updates = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingAdadeltaParametersGradAccumDebug.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingAdadeltaParametersGradAccumDebug.java
deleted file mode 100644
index 3fec5d6a6f7..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingAdadeltaParametersGradAccumDebug.java
+++ /dev/null
@@ -1,172 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.tpu;
-
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-
-/**
- * Retrieve Adadelta embedding parameters with debug support.
- * <p>
- * An op that retrieves optimization parameters from embedding to host
- * memory. Must be preceded by a ConfigureTPUEmbeddingHost op that sets up
- * the correct embedding table configuration. For example, this op is
- * used to retrieve updated parameters before saving a checkpoint.
- */
-public final class RetrieveTPUEmbeddingAdadeltaParametersGradAccumDebug extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.tpu.RetrieveTPUEmbeddingAdadeltaParametersGradAccumDebug}
-   */
-  public static class Options {
-    
-    /**
-     * @param tableId 
-     */
-    public Options tableId(Long tableId) {
-      this.tableId = tableId;
-      return this;
-    }
-    
-    /**
-     * @param tableName 
-     */
-    public Options tableName(String tableName) {
-      this.tableName = tableName;
-      return this;
-    }
-    
-    /**
-     * @param config 
-     */
-    public Options config(String config) {
-      this.config = config;
-      return this;
-    }
-    
-    private Long tableId;
-    private String tableName;
-    private String config;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new RetrieveTPUEmbeddingAdadeltaParametersGradAccumDebug operation.
-   * 
-   * @param scope current scope
-   * @param numShards 
-   * @param shardId 
-   * @param options carries optional attributes values
-   * @return a new instance of RetrieveTPUEmbeddingAdadeltaParametersGradAccumDebug
-   */
-  @Endpoint(describeByClass = true)
-  public static RetrieveTPUEmbeddingAdadeltaParametersGradAccumDebug create(Scope scope, Long numShards, Long shardId, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("RetrieveTPUEmbeddingAdadeltaParametersGradAccumDebug", scope.makeOpName("RetrieveTPUEmbeddingAdadeltaParametersGradAccumDebug"));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("num_shards", numShards);
-    opBuilder.setAttr("shard_id", shardId);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.tableId != null) {
-          opBuilder.setAttr("table_id", opts.tableId);
-        }
-        if (opts.tableName != null) {
-          opBuilder.setAttr("table_name", opts.tableName);
-        }
-        if (opts.config != null) {
-          opBuilder.setAttr("config", opts.config);
-        }
-      }
-    }
-    return new RetrieveTPUEmbeddingAdadeltaParametersGradAccumDebug(opBuilder.build());
-  }
-  
-  /**
-   * @param tableId 
-   */
-  public static Options tableId(Long tableId) {
-    return new Options().tableId(tableId);
-  }
-  
-  /**
-   * @param tableName 
-   */
-  public static Options tableName(String tableName) {
-    return new Options().tableName(tableName);
-  }
-  
-  /**
-   * @param config 
-   */
-  public static Options config(String config) {
-    return new Options().config(config);
-  }
-  
-  /**
-   * Parameter parameters updated by the Adadelta optimization algorithm.
-   */
-  public Output<TFloat32> parameters() {
-    return parameters;
-  }
-  
-  /**
-   * Parameter accumulators updated by the Adadelta optimization algorithm.
-   */
-  public Output<TFloat32> accumulators() {
-    return accumulators;
-  }
-  
-  /**
-   * Parameter updates updated by the Adadelta optimization algorithm.
-   */
-  public Output<TFloat32> updates() {
-    return updates;
-  }
-  
-  /**
-   * Parameter gradient_accumulators updated by the Adadelta optimization algorithm.
-   */
-  public Output<TFloat32> gradientAccumulators() {
-    return gradientAccumulators;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "RetrieveTPUEmbeddingAdadeltaParametersGradAccumDebug";
-  
-  private Output<TFloat32> parameters;
-  private Output<TFloat32> accumulators;
-  private Output<TFloat32> updates;
-  private Output<TFloat32> gradientAccumulators;
-  
-  private RetrieveTPUEmbeddingAdadeltaParametersGradAccumDebug(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    parameters = operation.output(outputIdx++);
-    accumulators = operation.output(outputIdx++);
-    updates = operation.output(outputIdx++);
-    gradientAccumulators = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingAdagradParameters.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingAdagradParameters.java
deleted file mode 100644
index 07ed32ae1f9..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingAdagradParameters.java
+++ /dev/null
@@ -1,154 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.tpu;
-
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-
-/**
- * Retrieve Adagrad embedding parameters.
- * <p>
- * An op that retrieves optimization parameters from embedding to host
- * memory. Must be preceded by a ConfigureTPUEmbeddingHost op that sets up
- * the correct embedding table configuration. For example, this op is
- * used to retrieve updated parameters before saving a checkpoint.
- */
-public final class RetrieveTPUEmbeddingAdagradParameters extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.tpu.RetrieveTPUEmbeddingAdagradParameters}
-   */
-  public static class Options {
-    
-    /**
-     * @param tableId 
-     */
-    public Options tableId(Long tableId) {
-      this.tableId = tableId;
-      return this;
-    }
-    
-    /**
-     * @param tableName 
-     */
-    public Options tableName(String tableName) {
-      this.tableName = tableName;
-      return this;
-    }
-    
-    /**
-     * @param config 
-     */
-    public Options config(String config) {
-      this.config = config;
-      return this;
-    }
-    
-    private Long tableId;
-    private String tableName;
-    private String config;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new RetrieveTPUEmbeddingAdagradParameters operation.
-   * 
-   * @param scope current scope
-   * @param numShards 
-   * @param shardId 
-   * @param options carries optional attributes values
-   * @return a new instance of RetrieveTPUEmbeddingAdagradParameters
-   */
-  @Endpoint(describeByClass = true)
-  public static RetrieveTPUEmbeddingAdagradParameters create(Scope scope, Long numShards, Long shardId, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("RetrieveTPUEmbeddingAdagradParameters", scope.makeOpName("RetrieveTPUEmbeddingAdagradParameters"));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("num_shards", numShards);
-    opBuilder.setAttr("shard_id", shardId);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.tableId != null) {
-          opBuilder.setAttr("table_id", opts.tableId);
-        }
-        if (opts.tableName != null) {
-          opBuilder.setAttr("table_name", opts.tableName);
-        }
-        if (opts.config != null) {
-          opBuilder.setAttr("config", opts.config);
-        }
-      }
-    }
-    return new RetrieveTPUEmbeddingAdagradParameters(opBuilder.build());
-  }
-  
-  /**
-   * @param tableId 
-   */
-  public static Options tableId(Long tableId) {
-    return new Options().tableId(tableId);
-  }
-  
-  /**
-   * @param tableName 
-   */
-  public static Options tableName(String tableName) {
-    return new Options().tableName(tableName);
-  }
-  
-  /**
-   * @param config 
-   */
-  public static Options config(String config) {
-    return new Options().config(config);
-  }
-  
-  /**
-   * Parameter parameters updated by the Adagrad optimization algorithm.
-   */
-  public Output<TFloat32> parameters() {
-    return parameters;
-  }
-  
-  /**
-   * Parameter accumulators updated by the Adagrad optimization algorithm.
-   */
-  public Output<TFloat32> accumulators() {
-    return accumulators;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "RetrieveTPUEmbeddingAdagradParameters";
-  
-  private Output<TFloat32> parameters;
-  private Output<TFloat32> accumulators;
-  
-  private RetrieveTPUEmbeddingAdagradParameters(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    parameters = operation.output(outputIdx++);
-    accumulators = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingAdagradParametersGradAccumDebug.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingAdagradParametersGradAccumDebug.java
deleted file mode 100644
index 39a8a7ab791..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingAdagradParametersGradAccumDebug.java
+++ /dev/null
@@ -1,163 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.tpu;
-
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-
-/**
- * Retrieve Adagrad embedding parameters with debug support.
- * <p>
- * An op that retrieves optimization parameters from embedding to host
- * memory. Must be preceded by a ConfigureTPUEmbeddingHost op that sets up
- * the correct embedding table configuration. For example, this op is
- * used to retrieve updated parameters before saving a checkpoint.
- */
-public final class RetrieveTPUEmbeddingAdagradParametersGradAccumDebug extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.tpu.RetrieveTPUEmbeddingAdagradParametersGradAccumDebug}
-   */
-  public static class Options {
-    
-    /**
-     * @param tableId 
-     */
-    public Options tableId(Long tableId) {
-      this.tableId = tableId;
-      return this;
-    }
-    
-    /**
-     * @param tableName 
-     */
-    public Options tableName(String tableName) {
-      this.tableName = tableName;
-      return this;
-    }
-    
-    /**
-     * @param config 
-     */
-    public Options config(String config) {
-      this.config = config;
-      return this;
-    }
-    
-    private Long tableId;
-    private String tableName;
-    private String config;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new RetrieveTPUEmbeddingAdagradParametersGradAccumDebug operation.
-   * 
-   * @param scope current scope
-   * @param numShards 
-   * @param shardId 
-   * @param options carries optional attributes values
-   * @return a new instance of RetrieveTPUEmbeddingAdagradParametersGradAccumDebug
-   */
-  @Endpoint(describeByClass = true)
-  public static RetrieveTPUEmbeddingAdagradParametersGradAccumDebug create(Scope scope, Long numShards, Long shardId, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("RetrieveTPUEmbeddingAdagradParametersGradAccumDebug", scope.makeOpName("RetrieveTPUEmbeddingAdagradParametersGradAccumDebug"));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("num_shards", numShards);
-    opBuilder.setAttr("shard_id", shardId);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.tableId != null) {
-          opBuilder.setAttr("table_id", opts.tableId);
-        }
-        if (opts.tableName != null) {
-          opBuilder.setAttr("table_name", opts.tableName);
-        }
-        if (opts.config != null) {
-          opBuilder.setAttr("config", opts.config);
-        }
-      }
-    }
-    return new RetrieveTPUEmbeddingAdagradParametersGradAccumDebug(opBuilder.build());
-  }
-  
-  /**
-   * @param tableId 
-   */
-  public static Options tableId(Long tableId) {
-    return new Options().tableId(tableId);
-  }
-  
-  /**
-   * @param tableName 
-   */
-  public static Options tableName(String tableName) {
-    return new Options().tableName(tableName);
-  }
-  
-  /**
-   * @param config 
-   */
-  public static Options config(String config) {
-    return new Options().config(config);
-  }
-  
-  /**
-   * Parameter parameters updated by the Adagrad optimization algorithm.
-   */
-  public Output<TFloat32> parameters() {
-    return parameters;
-  }
-  
-  /**
-   * Parameter accumulators updated by the Adagrad optimization algorithm.
-   */
-  public Output<TFloat32> accumulators() {
-    return accumulators;
-  }
-  
-  /**
-   * Parameter gradient_accumulators updated by the Adagrad optimization algorithm.
-   */
-  public Output<TFloat32> gradientAccumulators() {
-    return gradientAccumulators;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "RetrieveTPUEmbeddingAdagradParametersGradAccumDebug";
-  
-  private Output<TFloat32> parameters;
-  private Output<TFloat32> accumulators;
-  private Output<TFloat32> gradientAccumulators;
-  
-  private RetrieveTPUEmbeddingAdagradParametersGradAccumDebug(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    parameters = operation.output(outputIdx++);
-    accumulators = operation.output(outputIdx++);
-    gradientAccumulators = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingCenteredRMSPropParameters.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingCenteredRMSPropParameters.java
deleted file mode 100644
index 800cbd10aff..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingCenteredRMSPropParameters.java
+++ /dev/null
@@ -1,172 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.tpu;
-
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-
-/**
- * Retrieve centered RMSProp embedding parameters.
- * <p>
- * An op that retrieves optimization parameters from embedding to host
- * memory. Must be preceded by a ConfigureTPUEmbeddingHost op that sets up
- * the correct embedding table configuration. For example, this op is
- * used to retrieve updated parameters before saving a checkpoint.
- */
-public final class RetrieveTPUEmbeddingCenteredRMSPropParameters extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.tpu.RetrieveTPUEmbeddingCenteredRMSPropParameters}
-   */
-  public static class Options {
-    
-    /**
-     * @param tableId 
-     */
-    public Options tableId(Long tableId) {
-      this.tableId = tableId;
-      return this;
-    }
-    
-    /**
-     * @param tableName 
-     */
-    public Options tableName(String tableName) {
-      this.tableName = tableName;
-      return this;
-    }
-    
-    /**
-     * @param config 
-     */
-    public Options config(String config) {
-      this.config = config;
-      return this;
-    }
-    
-    private Long tableId;
-    private String tableName;
-    private String config;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new RetrieveTPUEmbeddingCenteredRMSPropParameters operation.
-   * 
-   * @param scope current scope
-   * @param numShards 
-   * @param shardId 
-   * @param options carries optional attributes values
-   * @return a new instance of RetrieveTPUEmbeddingCenteredRMSPropParameters
-   */
-  @Endpoint(describeByClass = true)
-  public static RetrieveTPUEmbeddingCenteredRMSPropParameters create(Scope scope, Long numShards, Long shardId, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("RetrieveTPUEmbeddingCenteredRMSPropParameters", scope.makeOpName("RetrieveTPUEmbeddingCenteredRMSPropParameters"));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("num_shards", numShards);
-    opBuilder.setAttr("shard_id", shardId);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.tableId != null) {
-          opBuilder.setAttr("table_id", opts.tableId);
-        }
-        if (opts.tableName != null) {
-          opBuilder.setAttr("table_name", opts.tableName);
-        }
-        if (opts.config != null) {
-          opBuilder.setAttr("config", opts.config);
-        }
-      }
-    }
-    return new RetrieveTPUEmbeddingCenteredRMSPropParameters(opBuilder.build());
-  }
-  
-  /**
-   * @param tableId 
-   */
-  public static Options tableId(Long tableId) {
-    return new Options().tableId(tableId);
-  }
-  
-  /**
-   * @param tableName 
-   */
-  public static Options tableName(String tableName) {
-    return new Options().tableName(tableName);
-  }
-  
-  /**
-   * @param config 
-   */
-  public static Options config(String config) {
-    return new Options().config(config);
-  }
-  
-  /**
-   * Parameter parameters updated by the centered RMSProp optimization algorithm.
-   */
-  public Output<TFloat32> parameters() {
-    return parameters;
-  }
-  
-  /**
-   * Parameter ms updated by the centered RMSProp optimization algorithm.
-   */
-  public Output<TFloat32> ms() {
-    return ms;
-  }
-  
-  /**
-   * Parameter mom updated by the centered RMSProp optimization algorithm.
-   */
-  public Output<TFloat32> mom() {
-    return mom;
-  }
-  
-  /**
-   * Parameter mg updated by the centered RMSProp optimization algorithm.
-   */
-  public Output<TFloat32> mg() {
-    return mg;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "RetrieveTPUEmbeddingCenteredRMSPropParameters";
-  
-  private Output<TFloat32> parameters;
-  private Output<TFloat32> ms;
-  private Output<TFloat32> mom;
-  private Output<TFloat32> mg;
-  
-  private RetrieveTPUEmbeddingCenteredRMSPropParameters(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    parameters = operation.output(outputIdx++);
-    ms = operation.output(outputIdx++);
-    mom = operation.output(outputIdx++);
-    mg = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingFTRLParameters.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingFTRLParameters.java
deleted file mode 100644
index 0e7114d3744..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingFTRLParameters.java
+++ /dev/null
@@ -1,163 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.tpu;
-
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-
-/**
- * Retrieve FTRL embedding parameters.
- * <p>
- * An op that retrieves optimization parameters from embedding to host
- * memory. Must be preceded by a ConfigureTPUEmbeddingHost op that sets up
- * the correct embedding table configuration. For example, this op is
- * used to retrieve updated parameters before saving a checkpoint.
- */
-public final class RetrieveTPUEmbeddingFTRLParameters extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.tpu.RetrieveTPUEmbeddingFTRLParameters}
-   */
-  public static class Options {
-    
-    /**
-     * @param tableId 
-     */
-    public Options tableId(Long tableId) {
-      this.tableId = tableId;
-      return this;
-    }
-    
-    /**
-     * @param tableName 
-     */
-    public Options tableName(String tableName) {
-      this.tableName = tableName;
-      return this;
-    }
-    
-    /**
-     * @param config 
-     */
-    public Options config(String config) {
-      this.config = config;
-      return this;
-    }
-    
-    private Long tableId;
-    private String tableName;
-    private String config;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new RetrieveTPUEmbeddingFTRLParameters operation.
-   * 
-   * @param scope current scope
-   * @param numShards 
-   * @param shardId 
-   * @param options carries optional attributes values
-   * @return a new instance of RetrieveTPUEmbeddingFTRLParameters
-   */
-  @Endpoint(describeByClass = true)
-  public static RetrieveTPUEmbeddingFTRLParameters create(Scope scope, Long numShards, Long shardId, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("RetrieveTPUEmbeddingFTRLParameters", scope.makeOpName("RetrieveTPUEmbeddingFTRLParameters"));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("num_shards", numShards);
-    opBuilder.setAttr("shard_id", shardId);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.tableId != null) {
-          opBuilder.setAttr("table_id", opts.tableId);
-        }
-        if (opts.tableName != null) {
-          opBuilder.setAttr("table_name", opts.tableName);
-        }
-        if (opts.config != null) {
-          opBuilder.setAttr("config", opts.config);
-        }
-      }
-    }
-    return new RetrieveTPUEmbeddingFTRLParameters(opBuilder.build());
-  }
-  
-  /**
-   * @param tableId 
-   */
-  public static Options tableId(Long tableId) {
-    return new Options().tableId(tableId);
-  }
-  
-  /**
-   * @param tableName 
-   */
-  public static Options tableName(String tableName) {
-    return new Options().tableName(tableName);
-  }
-  
-  /**
-   * @param config 
-   */
-  public static Options config(String config) {
-    return new Options().config(config);
-  }
-  
-  /**
-   * Parameter parameters updated by the FTRL optimization algorithm.
-   */
-  public Output<TFloat32> parameters() {
-    return parameters;
-  }
-  
-  /**
-   * Parameter accumulators updated by the FTRL optimization algorithm.
-   */
-  public Output<TFloat32> accumulators() {
-    return accumulators;
-  }
-  
-  /**
-   * Parameter linears updated by the FTRL optimization algorithm.
-   */
-  public Output<TFloat32> linears() {
-    return linears;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "RetrieveTPUEmbeddingFTRLParameters";
-  
-  private Output<TFloat32> parameters;
-  private Output<TFloat32> accumulators;
-  private Output<TFloat32> linears;
-  
-  private RetrieveTPUEmbeddingFTRLParameters(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    parameters = operation.output(outputIdx++);
-    accumulators = operation.output(outputIdx++);
-    linears = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingFTRLParametersGradAccumDebug.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingFTRLParametersGradAccumDebug.java
deleted file mode 100644
index 2dbff3ea109..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingFTRLParametersGradAccumDebug.java
+++ /dev/null
@@ -1,172 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.tpu;
-
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-
-/**
- * Retrieve FTRL embedding parameters with debug support.
- * <p>
- * An op that retrieves optimization parameters from embedding to host
- * memory. Must be preceded by a ConfigureTPUEmbeddingHost op that sets up
- * the correct embedding table configuration. For example, this op is
- * used to retrieve updated parameters before saving a checkpoint.
- */
-public final class RetrieveTPUEmbeddingFTRLParametersGradAccumDebug extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.tpu.RetrieveTPUEmbeddingFTRLParametersGradAccumDebug}
-   */
-  public static class Options {
-    
-    /**
-     * @param tableId 
-     */
-    public Options tableId(Long tableId) {
-      this.tableId = tableId;
-      return this;
-    }
-    
-    /**
-     * @param tableName 
-     */
-    public Options tableName(String tableName) {
-      this.tableName = tableName;
-      return this;
-    }
-    
-    /**
-     * @param config 
-     */
-    public Options config(String config) {
-      this.config = config;
-      return this;
-    }
-    
-    private Long tableId;
-    private String tableName;
-    private String config;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new RetrieveTPUEmbeddingFTRLParametersGradAccumDebug operation.
-   * 
-   * @param scope current scope
-   * @param numShards 
-   * @param shardId 
-   * @param options carries optional attributes values
-   * @return a new instance of RetrieveTPUEmbeddingFTRLParametersGradAccumDebug
-   */
-  @Endpoint(describeByClass = true)
-  public static RetrieveTPUEmbeddingFTRLParametersGradAccumDebug create(Scope scope, Long numShards, Long shardId, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("RetrieveTPUEmbeddingFTRLParametersGradAccumDebug", scope.makeOpName("RetrieveTPUEmbeddingFTRLParametersGradAccumDebug"));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("num_shards", numShards);
-    opBuilder.setAttr("shard_id", shardId);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.tableId != null) {
-          opBuilder.setAttr("table_id", opts.tableId);
-        }
-        if (opts.tableName != null) {
-          opBuilder.setAttr("table_name", opts.tableName);
-        }
-        if (opts.config != null) {
-          opBuilder.setAttr("config", opts.config);
-        }
-      }
-    }
-    return new RetrieveTPUEmbeddingFTRLParametersGradAccumDebug(opBuilder.build());
-  }
-  
-  /**
-   * @param tableId 
-   */
-  public static Options tableId(Long tableId) {
-    return new Options().tableId(tableId);
-  }
-  
-  /**
-   * @param tableName 
-   */
-  public static Options tableName(String tableName) {
-    return new Options().tableName(tableName);
-  }
-  
-  /**
-   * @param config 
-   */
-  public static Options config(String config) {
-    return new Options().config(config);
-  }
-  
-  /**
-   * Parameter parameters updated by the FTRL optimization algorithm.
-   */
-  public Output<TFloat32> parameters() {
-    return parameters;
-  }
-  
-  /**
-   * Parameter accumulators updated by the FTRL optimization algorithm.
-   */
-  public Output<TFloat32> accumulators() {
-    return accumulators;
-  }
-  
-  /**
-   * Parameter linears updated by the FTRL optimization algorithm.
-   */
-  public Output<TFloat32> linears() {
-    return linears;
-  }
-  
-  /**
-   * Parameter gradient_accumulators updated by the FTRL optimization algorithm.
-   */
-  public Output<TFloat32> gradientAccumulators() {
-    return gradientAccumulators;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "RetrieveTPUEmbeddingFTRLParametersGradAccumDebug";
-  
-  private Output<TFloat32> parameters;
-  private Output<TFloat32> accumulators;
-  private Output<TFloat32> linears;
-  private Output<TFloat32> gradientAccumulators;
-  
-  private RetrieveTPUEmbeddingFTRLParametersGradAccumDebug(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    parameters = operation.output(outputIdx++);
-    accumulators = operation.output(outputIdx++);
-    linears = operation.output(outputIdx++);
-    gradientAccumulators = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingMDLAdagradLightParameters.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingMDLAdagradLightParameters.java
deleted file mode 100644
index aa4be2cb318..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingMDLAdagradLightParameters.java
+++ /dev/null
@@ -1,172 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.tpu;
-
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-
-/**
- * Retrieve MDL Adagrad Light embedding parameters.
- * <p>
- * An op that retrieves optimization parameters from embedding to host
- * memory. Must be preceded by a ConfigureTPUEmbeddingHost op that sets up
- * the correct embedding table configuration. For example, this op is
- * used to retrieve updated parameters before saving a checkpoint.
- */
-public final class RetrieveTPUEmbeddingMDLAdagradLightParameters extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.tpu.RetrieveTPUEmbeddingMDLAdagradLightParameters}
-   */
-  public static class Options {
-    
-    /**
-     * @param tableId 
-     */
-    public Options tableId(Long tableId) {
-      this.tableId = tableId;
-      return this;
-    }
-    
-    /**
-     * @param tableName 
-     */
-    public Options tableName(String tableName) {
-      this.tableName = tableName;
-      return this;
-    }
-    
-    /**
-     * @param config 
-     */
-    public Options config(String config) {
-      this.config = config;
-      return this;
-    }
-    
-    private Long tableId;
-    private String tableName;
-    private String config;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new RetrieveTPUEmbeddingMDLAdagradLightParameters operation.
-   * 
-   * @param scope current scope
-   * @param numShards 
-   * @param shardId 
-   * @param options carries optional attributes values
-   * @return a new instance of RetrieveTPUEmbeddingMDLAdagradLightParameters
-   */
-  @Endpoint(describeByClass = true)
-  public static RetrieveTPUEmbeddingMDLAdagradLightParameters create(Scope scope, Long numShards, Long shardId, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("RetrieveTPUEmbeddingMDLAdagradLightParameters", scope.makeOpName("RetrieveTPUEmbeddingMDLAdagradLightParameters"));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("num_shards", numShards);
-    opBuilder.setAttr("shard_id", shardId);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.tableId != null) {
-          opBuilder.setAttr("table_id", opts.tableId);
-        }
-        if (opts.tableName != null) {
-          opBuilder.setAttr("table_name", opts.tableName);
-        }
-        if (opts.config != null) {
-          opBuilder.setAttr("config", opts.config);
-        }
-      }
-    }
-    return new RetrieveTPUEmbeddingMDLAdagradLightParameters(opBuilder.build());
-  }
-  
-  /**
-   * @param tableId 
-   */
-  public static Options tableId(Long tableId) {
-    return new Options().tableId(tableId);
-  }
-  
-  /**
-   * @param tableName 
-   */
-  public static Options tableName(String tableName) {
-    return new Options().tableName(tableName);
-  }
-  
-  /**
-   * @param config 
-   */
-  public static Options config(String config) {
-    return new Options().config(config);
-  }
-  
-  /**
-   * Parameter parameters updated by the MDL Adagrad Light optimization algorithm.
-   */
-  public Output<TFloat32> parameters() {
-    return parameters;
-  }
-  
-  /**
-   * Parameter accumulators updated by the MDL Adagrad Light optimization algorithm.
-   */
-  public Output<TFloat32> accumulators() {
-    return accumulators;
-  }
-  
-  /**
-   * Parameter weights updated by the MDL Adagrad Light optimization algorithm.
-   */
-  public Output<TFloat32> weights() {
-    return weights;
-  }
-  
-  /**
-   * Parameter benefits updated by the MDL Adagrad Light optimization algorithm.
-   */
-  public Output<TFloat32> benefits() {
-    return benefits;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "RetrieveTPUEmbeddingMDLAdagradLightParameters";
-  
-  private Output<TFloat32> parameters;
-  private Output<TFloat32> accumulators;
-  private Output<TFloat32> weights;
-  private Output<TFloat32> benefits;
-  
-  private RetrieveTPUEmbeddingMDLAdagradLightParameters(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    parameters = operation.output(outputIdx++);
-    accumulators = operation.output(outputIdx++);
-    weights = operation.output(outputIdx++);
-    benefits = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingMomentumParameters.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingMomentumParameters.java
deleted file mode 100644
index 7f397adda1c..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingMomentumParameters.java
+++ /dev/null
@@ -1,154 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.tpu;
-
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-
-/**
- * Retrieve Momentum embedding parameters.
- * <p>
- * An op that retrieves optimization parameters from embedding to host
- * memory. Must be preceded by a ConfigureTPUEmbeddingHost op that sets up
- * the correct embedding table configuration. For example, this op is
- * used to retrieve updated parameters before saving a checkpoint.
- */
-public final class RetrieveTPUEmbeddingMomentumParameters extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.tpu.RetrieveTPUEmbeddingMomentumParameters}
-   */
-  public static class Options {
-    
-    /**
-     * @param tableId 
-     */
-    public Options tableId(Long tableId) {
-      this.tableId = tableId;
-      return this;
-    }
-    
-    /**
-     * @param tableName 
-     */
-    public Options tableName(String tableName) {
-      this.tableName = tableName;
-      return this;
-    }
-    
-    /**
-     * @param config 
-     */
-    public Options config(String config) {
-      this.config = config;
-      return this;
-    }
-    
-    private Long tableId;
-    private String tableName;
-    private String config;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new RetrieveTPUEmbeddingMomentumParameters operation.
-   * 
-   * @param scope current scope
-   * @param numShards 
-   * @param shardId 
-   * @param options carries optional attributes values
-   * @return a new instance of RetrieveTPUEmbeddingMomentumParameters
-   */
-  @Endpoint(describeByClass = true)
-  public static RetrieveTPUEmbeddingMomentumParameters create(Scope scope, Long numShards, Long shardId, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("RetrieveTPUEmbeddingMomentumParameters", scope.makeOpName("RetrieveTPUEmbeddingMomentumParameters"));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("num_shards", numShards);
-    opBuilder.setAttr("shard_id", shardId);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.tableId != null) {
-          opBuilder.setAttr("table_id", opts.tableId);
-        }
-        if (opts.tableName != null) {
-          opBuilder.setAttr("table_name", opts.tableName);
-        }
-        if (opts.config != null) {
-          opBuilder.setAttr("config", opts.config);
-        }
-      }
-    }
-    return new RetrieveTPUEmbeddingMomentumParameters(opBuilder.build());
-  }
-  
-  /**
-   * @param tableId 
-   */
-  public static Options tableId(Long tableId) {
-    return new Options().tableId(tableId);
-  }
-  
-  /**
-   * @param tableName 
-   */
-  public static Options tableName(String tableName) {
-    return new Options().tableName(tableName);
-  }
-  
-  /**
-   * @param config 
-   */
-  public static Options config(String config) {
-    return new Options().config(config);
-  }
-  
-  /**
-   * Parameter parameters updated by the Momentum optimization algorithm.
-   */
-  public Output<TFloat32> parameters() {
-    return parameters;
-  }
-  
-  /**
-   * Parameter momenta updated by the Momentum optimization algorithm.
-   */
-  public Output<TFloat32> momenta() {
-    return momenta;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "RetrieveTPUEmbeddingMomentumParameters";
-  
-  private Output<TFloat32> parameters;
-  private Output<TFloat32> momenta;
-  
-  private RetrieveTPUEmbeddingMomentumParameters(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    parameters = operation.output(outputIdx++);
-    momenta = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingMomentumParametersGradAccumDebug.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingMomentumParametersGradAccumDebug.java
deleted file mode 100644
index 15aceae7f2a..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingMomentumParametersGradAccumDebug.java
+++ /dev/null
@@ -1,163 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.tpu;
-
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-
-/**
- * Retrieve Momentum embedding parameters with debug support.
- * <p>
- * An op that retrieves optimization parameters from embedding to host
- * memory. Must be preceded by a ConfigureTPUEmbeddingHost op that sets up
- * the correct embedding table configuration. For example, this op is
- * used to retrieve updated parameters before saving a checkpoint.
- */
-public final class RetrieveTPUEmbeddingMomentumParametersGradAccumDebug extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.tpu.RetrieveTPUEmbeddingMomentumParametersGradAccumDebug}
-   */
-  public static class Options {
-    
-    /**
-     * @param tableId 
-     */
-    public Options tableId(Long tableId) {
-      this.tableId = tableId;
-      return this;
-    }
-    
-    /**
-     * @param tableName 
-     */
-    public Options tableName(String tableName) {
-      this.tableName = tableName;
-      return this;
-    }
-    
-    /**
-     * @param config 
-     */
-    public Options config(String config) {
-      this.config = config;
-      return this;
-    }
-    
-    private Long tableId;
-    private String tableName;
-    private String config;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new RetrieveTPUEmbeddingMomentumParametersGradAccumDebug operation.
-   * 
-   * @param scope current scope
-   * @param numShards 
-   * @param shardId 
-   * @param options carries optional attributes values
-   * @return a new instance of RetrieveTPUEmbeddingMomentumParametersGradAccumDebug
-   */
-  @Endpoint(describeByClass = true)
-  public static RetrieveTPUEmbeddingMomentumParametersGradAccumDebug create(Scope scope, Long numShards, Long shardId, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("RetrieveTPUEmbeddingMomentumParametersGradAccumDebug", scope.makeOpName("RetrieveTPUEmbeddingMomentumParametersGradAccumDebug"));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("num_shards", numShards);
-    opBuilder.setAttr("shard_id", shardId);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.tableId != null) {
-          opBuilder.setAttr("table_id", opts.tableId);
-        }
-        if (opts.tableName != null) {
-          opBuilder.setAttr("table_name", opts.tableName);
-        }
-        if (opts.config != null) {
-          opBuilder.setAttr("config", opts.config);
-        }
-      }
-    }
-    return new RetrieveTPUEmbeddingMomentumParametersGradAccumDebug(opBuilder.build());
-  }
-  
-  /**
-   * @param tableId 
-   */
-  public static Options tableId(Long tableId) {
-    return new Options().tableId(tableId);
-  }
-  
-  /**
-   * @param tableName 
-   */
-  public static Options tableName(String tableName) {
-    return new Options().tableName(tableName);
-  }
-  
-  /**
-   * @param config 
-   */
-  public static Options config(String config) {
-    return new Options().config(config);
-  }
-  
-  /**
-   * Parameter parameters updated by the Momentum optimization algorithm.
-   */
-  public Output<TFloat32> parameters() {
-    return parameters;
-  }
-  
-  /**
-   * Parameter momenta updated by the Momentum optimization algorithm.
-   */
-  public Output<TFloat32> momenta() {
-    return momenta;
-  }
-  
-  /**
-   * Parameter gradient_accumulators updated by the Momentum optimization algorithm.
-   */
-  public Output<TFloat32> gradientAccumulators() {
-    return gradientAccumulators;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "RetrieveTPUEmbeddingMomentumParametersGradAccumDebug";
-  
-  private Output<TFloat32> parameters;
-  private Output<TFloat32> momenta;
-  private Output<TFloat32> gradientAccumulators;
-  
-  private RetrieveTPUEmbeddingMomentumParametersGradAccumDebug(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    parameters = operation.output(outputIdx++);
-    momenta = operation.output(outputIdx++);
-    gradientAccumulators = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingProximalAdagradParameters.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingProximalAdagradParameters.java
deleted file mode 100644
index 4f20964f27f..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingProximalAdagradParameters.java
+++ /dev/null
@@ -1,154 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.tpu;
-
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-
-/**
- * Retrieve proximal Adagrad embedding parameters.
- * <p>
- * An op that retrieves optimization parameters from embedding to host
- * memory. Must be preceded by a ConfigureTPUEmbeddingHost op that sets up
- * the correct embedding table configuration. For example, this op is
- * used to retrieve updated parameters before saving a checkpoint.
- */
-public final class RetrieveTPUEmbeddingProximalAdagradParameters extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.tpu.RetrieveTPUEmbeddingProximalAdagradParameters}
-   */
-  public static class Options {
-    
-    /**
-     * @param tableId 
-     */
-    public Options tableId(Long tableId) {
-      this.tableId = tableId;
-      return this;
-    }
-    
-    /**
-     * @param tableName 
-     */
-    public Options tableName(String tableName) {
-      this.tableName = tableName;
-      return this;
-    }
-    
-    /**
-     * @param config 
-     */
-    public Options config(String config) {
-      this.config = config;
-      return this;
-    }
-    
-    private Long tableId;
-    private String tableName;
-    private String config;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new RetrieveTPUEmbeddingProximalAdagradParameters operation.
-   * 
-   * @param scope current scope
-   * @param numShards 
-   * @param shardId 
-   * @param options carries optional attributes values
-   * @return a new instance of RetrieveTPUEmbeddingProximalAdagradParameters
-   */
-  @Endpoint(describeByClass = true)
-  public static RetrieveTPUEmbeddingProximalAdagradParameters create(Scope scope, Long numShards, Long shardId, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("RetrieveTPUEmbeddingProximalAdagradParameters", scope.makeOpName("RetrieveTPUEmbeddingProximalAdagradParameters"));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("num_shards", numShards);
-    opBuilder.setAttr("shard_id", shardId);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.tableId != null) {
-          opBuilder.setAttr("table_id", opts.tableId);
-        }
-        if (opts.tableName != null) {
-          opBuilder.setAttr("table_name", opts.tableName);
-        }
-        if (opts.config != null) {
-          opBuilder.setAttr("config", opts.config);
-        }
-      }
-    }
-    return new RetrieveTPUEmbeddingProximalAdagradParameters(opBuilder.build());
-  }
-  
-  /**
-   * @param tableId 
-   */
-  public static Options tableId(Long tableId) {
-    return new Options().tableId(tableId);
-  }
-  
-  /**
-   * @param tableName 
-   */
-  public static Options tableName(String tableName) {
-    return new Options().tableName(tableName);
-  }
-  
-  /**
-   * @param config 
-   */
-  public static Options config(String config) {
-    return new Options().config(config);
-  }
-  
-  /**
-   * Parameter parameters updated by the proximal Adagrad optimization algorithm.
-   */
-  public Output<TFloat32> parameters() {
-    return parameters;
-  }
-  
-  /**
-   * Parameter accumulators updated by the proximal Adagrad optimization algorithm.
-   */
-  public Output<TFloat32> accumulators() {
-    return accumulators;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "RetrieveTPUEmbeddingProximalAdagradParameters";
-  
-  private Output<TFloat32> parameters;
-  private Output<TFloat32> accumulators;
-  
-  private RetrieveTPUEmbeddingProximalAdagradParameters(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    parameters = operation.output(outputIdx++);
-    accumulators = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingProximalAdagradParametersGradAccumDebug.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingProximalAdagradParametersGradAccumDebug.java
deleted file mode 100644
index abde61ec5a9..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingProximalAdagradParametersGradAccumDebug.java
+++ /dev/null
@@ -1,163 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.tpu;
-
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-
-/**
- * Retrieve proximal Adagrad embedding parameters with debug support.
- * <p>
- * An op that retrieves optimization parameters from embedding to host
- * memory. Must be preceded by a ConfigureTPUEmbeddingHost op that sets up
- * the correct embedding table configuration. For example, this op is
- * used to retrieve updated parameters before saving a checkpoint.
- */
-public final class RetrieveTPUEmbeddingProximalAdagradParametersGradAccumDebug extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.tpu.RetrieveTPUEmbeddingProximalAdagradParametersGradAccumDebug}
-   */
-  public static class Options {
-    
-    /**
-     * @param tableId 
-     */
-    public Options tableId(Long tableId) {
-      this.tableId = tableId;
-      return this;
-    }
-    
-    /**
-     * @param tableName 
-     */
-    public Options tableName(String tableName) {
-      this.tableName = tableName;
-      return this;
-    }
-    
-    /**
-     * @param config 
-     */
-    public Options config(String config) {
-      this.config = config;
-      return this;
-    }
-    
-    private Long tableId;
-    private String tableName;
-    private String config;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new RetrieveTPUEmbeddingProximalAdagradParametersGradAccumDebug operation.
-   * 
-   * @param scope current scope
-   * @param numShards 
-   * @param shardId 
-   * @param options carries optional attributes values
-   * @return a new instance of RetrieveTPUEmbeddingProximalAdagradParametersGradAccumDebug
-   */
-  @Endpoint(describeByClass = true)
-  public static RetrieveTPUEmbeddingProximalAdagradParametersGradAccumDebug create(Scope scope, Long numShards, Long shardId, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("RetrieveTPUEmbeddingProximalAdagradParametersGradAccumDebug", scope.makeOpName("RetrieveTPUEmbeddingProximalAdagradParametersGradAccumDebug"));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("num_shards", numShards);
-    opBuilder.setAttr("shard_id", shardId);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.tableId != null) {
-          opBuilder.setAttr("table_id", opts.tableId);
-        }
-        if (opts.tableName != null) {
-          opBuilder.setAttr("table_name", opts.tableName);
-        }
-        if (opts.config != null) {
-          opBuilder.setAttr("config", opts.config);
-        }
-      }
-    }
-    return new RetrieveTPUEmbeddingProximalAdagradParametersGradAccumDebug(opBuilder.build());
-  }
-  
-  /**
-   * @param tableId 
-   */
-  public static Options tableId(Long tableId) {
-    return new Options().tableId(tableId);
-  }
-  
-  /**
-   * @param tableName 
-   */
-  public static Options tableName(String tableName) {
-    return new Options().tableName(tableName);
-  }
-  
-  /**
-   * @param config 
-   */
-  public static Options config(String config) {
-    return new Options().config(config);
-  }
-  
-  /**
-   * Parameter parameters updated by the proximal Adagrad optimization algorithm.
-   */
-  public Output<TFloat32> parameters() {
-    return parameters;
-  }
-  
-  /**
-   * Parameter accumulators updated by the proximal Adagrad optimization algorithm.
-   */
-  public Output<TFloat32> accumulators() {
-    return accumulators;
-  }
-  
-  /**
-   * Parameter gradient_accumulators updated by the proximal Adagrad optimization algorithm.
-   */
-  public Output<TFloat32> gradientAccumulators() {
-    return gradientAccumulators;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "RetrieveTPUEmbeddingProximalAdagradParametersGradAccumDebug";
-  
-  private Output<TFloat32> parameters;
-  private Output<TFloat32> accumulators;
-  private Output<TFloat32> gradientAccumulators;
-  
-  private RetrieveTPUEmbeddingProximalAdagradParametersGradAccumDebug(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    parameters = operation.output(outputIdx++);
-    accumulators = operation.output(outputIdx++);
-    gradientAccumulators = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingProximalYogiParameters.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingProximalYogiParameters.java
deleted file mode 100644
index a46cae90359..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingProximalYogiParameters.java
+++ /dev/null
@@ -1,154 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.tpu;
-
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-
-/**
- */
-public final class RetrieveTPUEmbeddingProximalYogiParameters extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.tpu.RetrieveTPUEmbeddingProximalYogiParameters}
-   */
-  public static class Options {
-    
-    /**
-     * @param tableId 
-     */
-    public Options tableId(Long tableId) {
-      this.tableId = tableId;
-      return this;
-    }
-    
-    /**
-     * @param tableName 
-     */
-    public Options tableName(String tableName) {
-      this.tableName = tableName;
-      return this;
-    }
-    
-    /**
-     * @param config 
-     */
-    public Options config(String config) {
-      this.config = config;
-      return this;
-    }
-    
-    private Long tableId;
-    private String tableName;
-    private String config;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new RetrieveTPUEmbeddingProximalYogiParameters operation.
-   * 
-   * @param scope current scope
-   * @param numShards 
-   * @param shardId 
-   * @param options carries optional attributes values
-   * @return a new instance of RetrieveTPUEmbeddingProximalYogiParameters
-   */
-  @Endpoint(describeByClass = true)
-  public static RetrieveTPUEmbeddingProximalYogiParameters create(Scope scope, Long numShards, Long shardId, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("RetrieveTPUEmbeddingProximalYogiParameters", scope.makeOpName("RetrieveTPUEmbeddingProximalYogiParameters"));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("num_shards", numShards);
-    opBuilder.setAttr("shard_id", shardId);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.tableId != null) {
-          opBuilder.setAttr("table_id", opts.tableId);
-        }
-        if (opts.tableName != null) {
-          opBuilder.setAttr("table_name", opts.tableName);
-        }
-        if (opts.config != null) {
-          opBuilder.setAttr("config", opts.config);
-        }
-      }
-    }
-    return new RetrieveTPUEmbeddingProximalYogiParameters(opBuilder.build());
-  }
-  
-  /**
-   * @param tableId 
-   */
-  public static Options tableId(Long tableId) {
-    return new Options().tableId(tableId);
-  }
-  
-  /**
-   * @param tableName 
-   */
-  public static Options tableName(String tableName) {
-    return new Options().tableName(tableName);
-  }
-  
-  /**
-   * @param config 
-   */
-  public static Options config(String config) {
-    return new Options().config(config);
-  }
-  
-  /**
-   */
-  public Output<TFloat32> parameters() {
-    return parameters;
-  }
-  
-  /**
-   */
-  public Output<TFloat32> v() {
-    return v;
-  }
-  
-  /**
-   */
-  public Output<TFloat32> m() {
-    return m;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "RetrieveTPUEmbeddingProximalYogiParameters";
-  
-  private Output<TFloat32> parameters;
-  private Output<TFloat32> v;
-  private Output<TFloat32> m;
-  
-  private RetrieveTPUEmbeddingProximalYogiParameters(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    parameters = operation.output(outputIdx++);
-    v = operation.output(outputIdx++);
-    m = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingProximalYogiParametersGradAccumDebug.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingProximalYogiParametersGradAccumDebug.java
deleted file mode 100644
index 55535a573f6..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingProximalYogiParametersGradAccumDebug.java
+++ /dev/null
@@ -1,162 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.tpu;
-
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-
-/**
- */
-public final class RetrieveTPUEmbeddingProximalYogiParametersGradAccumDebug extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.tpu.RetrieveTPUEmbeddingProximalYogiParametersGradAccumDebug}
-   */
-  public static class Options {
-    
-    /**
-     * @param tableId 
-     */
-    public Options tableId(Long tableId) {
-      this.tableId = tableId;
-      return this;
-    }
-    
-    /**
-     * @param tableName 
-     */
-    public Options tableName(String tableName) {
-      this.tableName = tableName;
-      return this;
-    }
-    
-    /**
-     * @param config 
-     */
-    public Options config(String config) {
-      this.config = config;
-      return this;
-    }
-    
-    private Long tableId;
-    private String tableName;
-    private String config;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new RetrieveTPUEmbeddingProximalYogiParametersGradAccumDebug operation.
-   * 
-   * @param scope current scope
-   * @param numShards 
-   * @param shardId 
-   * @param options carries optional attributes values
-   * @return a new instance of RetrieveTPUEmbeddingProximalYogiParametersGradAccumDebug
-   */
-  @Endpoint(describeByClass = true)
-  public static RetrieveTPUEmbeddingProximalYogiParametersGradAccumDebug create(Scope scope, Long numShards, Long shardId, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("RetrieveTPUEmbeddingProximalYogiParametersGradAccumDebug", scope.makeOpName("RetrieveTPUEmbeddingProximalYogiParametersGradAccumDebug"));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("num_shards", numShards);
-    opBuilder.setAttr("shard_id", shardId);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.tableId != null) {
-          opBuilder.setAttr("table_id", opts.tableId);
-        }
-        if (opts.tableName != null) {
-          opBuilder.setAttr("table_name", opts.tableName);
-        }
-        if (opts.config != null) {
-          opBuilder.setAttr("config", opts.config);
-        }
-      }
-    }
-    return new RetrieveTPUEmbeddingProximalYogiParametersGradAccumDebug(opBuilder.build());
-  }
-  
-  /**
-   * @param tableId 
-   */
-  public static Options tableId(Long tableId) {
-    return new Options().tableId(tableId);
-  }
-  
-  /**
-   * @param tableName 
-   */
-  public static Options tableName(String tableName) {
-    return new Options().tableName(tableName);
-  }
-  
-  /**
-   * @param config 
-   */
-  public static Options config(String config) {
-    return new Options().config(config);
-  }
-  
-  /**
-   */
-  public Output<TFloat32> parameters() {
-    return parameters;
-  }
-  
-  /**
-   */
-  public Output<TFloat32> v() {
-    return v;
-  }
-  
-  /**
-   */
-  public Output<TFloat32> m() {
-    return m;
-  }
-  
-  /**
-   */
-  public Output<TFloat32> gradientAccumulators() {
-    return gradientAccumulators;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "RetrieveTPUEmbeddingProximalYogiParametersGradAccumDebug";
-  
-  private Output<TFloat32> parameters;
-  private Output<TFloat32> v;
-  private Output<TFloat32> m;
-  private Output<TFloat32> gradientAccumulators;
-  
-  private RetrieveTPUEmbeddingProximalYogiParametersGradAccumDebug(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    parameters = operation.output(outputIdx++);
-    v = operation.output(outputIdx++);
-    m = operation.output(outputIdx++);
-    gradientAccumulators = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingRMSPropParameters.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingRMSPropParameters.java
deleted file mode 100644
index fcc01589e38..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingRMSPropParameters.java
+++ /dev/null
@@ -1,163 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.tpu;
-
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-
-/**
- * Retrieve RMSProp embedding parameters.
- * <p>
- * An op that retrieves optimization parameters from embedding to host
- * memory. Must be preceded by a ConfigureTPUEmbeddingHost op that sets up
- * the correct embedding table configuration. For example, this op is
- * used to retrieve updated parameters before saving a checkpoint.
- */
-public final class RetrieveTPUEmbeddingRMSPropParameters extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.tpu.RetrieveTPUEmbeddingRMSPropParameters}
-   */
-  public static class Options {
-    
-    /**
-     * @param tableId 
-     */
-    public Options tableId(Long tableId) {
-      this.tableId = tableId;
-      return this;
-    }
-    
-    /**
-     * @param tableName 
-     */
-    public Options tableName(String tableName) {
-      this.tableName = tableName;
-      return this;
-    }
-    
-    /**
-     * @param config 
-     */
-    public Options config(String config) {
-      this.config = config;
-      return this;
-    }
-    
-    private Long tableId;
-    private String tableName;
-    private String config;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new RetrieveTPUEmbeddingRMSPropParameters operation.
-   * 
-   * @param scope current scope
-   * @param numShards 
-   * @param shardId 
-   * @param options carries optional attributes values
-   * @return a new instance of RetrieveTPUEmbeddingRMSPropParameters
-   */
-  @Endpoint(describeByClass = true)
-  public static RetrieveTPUEmbeddingRMSPropParameters create(Scope scope, Long numShards, Long shardId, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("RetrieveTPUEmbeddingRMSPropParameters", scope.makeOpName("RetrieveTPUEmbeddingRMSPropParameters"));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("num_shards", numShards);
-    opBuilder.setAttr("shard_id", shardId);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.tableId != null) {
-          opBuilder.setAttr("table_id", opts.tableId);
-        }
-        if (opts.tableName != null) {
-          opBuilder.setAttr("table_name", opts.tableName);
-        }
-        if (opts.config != null) {
-          opBuilder.setAttr("config", opts.config);
-        }
-      }
-    }
-    return new RetrieveTPUEmbeddingRMSPropParameters(opBuilder.build());
-  }
-  
-  /**
-   * @param tableId 
-   */
-  public static Options tableId(Long tableId) {
-    return new Options().tableId(tableId);
-  }
-  
-  /**
-   * @param tableName 
-   */
-  public static Options tableName(String tableName) {
-    return new Options().tableName(tableName);
-  }
-  
-  /**
-   * @param config 
-   */
-  public static Options config(String config) {
-    return new Options().config(config);
-  }
-  
-  /**
-   * Parameter parameters updated by the RMSProp optimization algorithm.
-   */
-  public Output<TFloat32> parameters() {
-    return parameters;
-  }
-  
-  /**
-   * Parameter ms updated by the RMSProp optimization algorithm.
-   */
-  public Output<TFloat32> ms() {
-    return ms;
-  }
-  
-  /**
-   * Parameter mom updated by the RMSProp optimization algorithm.
-   */
-  public Output<TFloat32> mom() {
-    return mom;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "RetrieveTPUEmbeddingRMSPropParameters";
-  
-  private Output<TFloat32> parameters;
-  private Output<TFloat32> ms;
-  private Output<TFloat32> mom;
-  
-  private RetrieveTPUEmbeddingRMSPropParameters(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    parameters = operation.output(outputIdx++);
-    ms = operation.output(outputIdx++);
-    mom = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingRMSPropParametersGradAccumDebug.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingRMSPropParametersGradAccumDebug.java
deleted file mode 100644
index 0c3814b23b4..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingRMSPropParametersGradAccumDebug.java
+++ /dev/null
@@ -1,172 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.tpu;
-
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-
-/**
- * Retrieve RMSProp embedding parameters with debug support.
- * <p>
- * An op that retrieves optimization parameters from embedding to host
- * memory. Must be preceded by a ConfigureTPUEmbeddingHost op that sets up
- * the correct embedding table configuration. For example, this op is
- * used to retrieve updated parameters before saving a checkpoint.
- */
-public final class RetrieveTPUEmbeddingRMSPropParametersGradAccumDebug extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.tpu.RetrieveTPUEmbeddingRMSPropParametersGradAccumDebug}
-   */
-  public static class Options {
-    
-    /**
-     * @param tableId 
-     */
-    public Options tableId(Long tableId) {
-      this.tableId = tableId;
-      return this;
-    }
-    
-    /**
-     * @param tableName 
-     */
-    public Options tableName(String tableName) {
-      this.tableName = tableName;
-      return this;
-    }
-    
-    /**
-     * @param config 
-     */
-    public Options config(String config) {
-      this.config = config;
-      return this;
-    }
-    
-    private Long tableId;
-    private String tableName;
-    private String config;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new RetrieveTPUEmbeddingRMSPropParametersGradAccumDebug operation.
-   * 
-   * @param scope current scope
-   * @param numShards 
-   * @param shardId 
-   * @param options carries optional attributes values
-   * @return a new instance of RetrieveTPUEmbeddingRMSPropParametersGradAccumDebug
-   */
-  @Endpoint(describeByClass = true)
-  public static RetrieveTPUEmbeddingRMSPropParametersGradAccumDebug create(Scope scope, Long numShards, Long shardId, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("RetrieveTPUEmbeddingRMSPropParametersGradAccumDebug", scope.makeOpName("RetrieveTPUEmbeddingRMSPropParametersGradAccumDebug"));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("num_shards", numShards);
-    opBuilder.setAttr("shard_id", shardId);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.tableId != null) {
-          opBuilder.setAttr("table_id", opts.tableId);
-        }
-        if (opts.tableName != null) {
-          opBuilder.setAttr("table_name", opts.tableName);
-        }
-        if (opts.config != null) {
-          opBuilder.setAttr("config", opts.config);
-        }
-      }
-    }
-    return new RetrieveTPUEmbeddingRMSPropParametersGradAccumDebug(opBuilder.build());
-  }
-  
-  /**
-   * @param tableId 
-   */
-  public static Options tableId(Long tableId) {
-    return new Options().tableId(tableId);
-  }
-  
-  /**
-   * @param tableName 
-   */
-  public static Options tableName(String tableName) {
-    return new Options().tableName(tableName);
-  }
-  
-  /**
-   * @param config 
-   */
-  public static Options config(String config) {
-    return new Options().config(config);
-  }
-  
-  /**
-   * Parameter parameters updated by the RMSProp optimization algorithm.
-   */
-  public Output<TFloat32> parameters() {
-    return parameters;
-  }
-  
-  /**
-   * Parameter ms updated by the RMSProp optimization algorithm.
-   */
-  public Output<TFloat32> ms() {
-    return ms;
-  }
-  
-  /**
-   * Parameter mom updated by the RMSProp optimization algorithm.
-   */
-  public Output<TFloat32> mom() {
-    return mom;
-  }
-  
-  /**
-   * Parameter gradient_accumulators updated by the RMSProp optimization algorithm.
-   */
-  public Output<TFloat32> gradientAccumulators() {
-    return gradientAccumulators;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "RetrieveTPUEmbeddingRMSPropParametersGradAccumDebug";
-  
-  private Output<TFloat32> parameters;
-  private Output<TFloat32> ms;
-  private Output<TFloat32> mom;
-  private Output<TFloat32> gradientAccumulators;
-  
-  private RetrieveTPUEmbeddingRMSPropParametersGradAccumDebug(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    parameters = operation.output(outputIdx++);
-    ms = operation.output(outputIdx++);
-    mom = operation.output(outputIdx++);
-    gradientAccumulators = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingStochasticGradientDescentParameters.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingStochasticGradientDescentParameters.java
deleted file mode 100644
index 4af3eff2fa1..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingStochasticGradientDescentParameters.java
+++ /dev/null
@@ -1,151 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.tpu;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-
-/**
- * Retrieve SGD embedding parameters.
- * <p>
- * An op that retrieves optimization parameters from embedding to host
- * memory. Must be preceded by a ConfigureTPUEmbeddingHost op that sets up
- * the correct embedding table configuration. For example, this op is
- * used to retrieve updated parameters before saving a checkpoint.
- */
-public final class RetrieveTPUEmbeddingStochasticGradientDescentParameters extends RawOp implements Operand<TFloat32> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.tpu.RetrieveTPUEmbeddingStochasticGradientDescentParameters}
-   */
-  public static class Options {
-    
-    /**
-     * @param tableId 
-     */
-    public Options tableId(Long tableId) {
-      this.tableId = tableId;
-      return this;
-    }
-    
-    /**
-     * @param tableName 
-     */
-    public Options tableName(String tableName) {
-      this.tableName = tableName;
-      return this;
-    }
-    
-    /**
-     * @param config 
-     */
-    public Options config(String config) {
-      this.config = config;
-      return this;
-    }
-    
-    private Long tableId;
-    private String tableName;
-    private String config;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new RetrieveTPUEmbeddingStochasticGradientDescentParameters operation.
-   * 
-   * @param scope current scope
-   * @param numShards 
-   * @param shardId 
-   * @param options carries optional attributes values
-   * @return a new instance of RetrieveTPUEmbeddingStochasticGradientDescentParameters
-   */
-  @Endpoint(describeByClass = true)
-  public static RetrieveTPUEmbeddingStochasticGradientDescentParameters create(Scope scope, Long numShards, Long shardId, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("RetrieveTPUEmbeddingStochasticGradientDescentParameters", scope.makeOpName("RetrieveTPUEmbeddingStochasticGradientDescentParameters"));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("num_shards", numShards);
-    opBuilder.setAttr("shard_id", shardId);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.tableId != null) {
-          opBuilder.setAttr("table_id", opts.tableId);
-        }
-        if (opts.tableName != null) {
-          opBuilder.setAttr("table_name", opts.tableName);
-        }
-        if (opts.config != null) {
-          opBuilder.setAttr("config", opts.config);
-        }
-      }
-    }
-    return new RetrieveTPUEmbeddingStochasticGradientDescentParameters(opBuilder.build());
-  }
-  
-  /**
-   * @param tableId 
-   */
-  public static Options tableId(Long tableId) {
-    return new Options().tableId(tableId);
-  }
-  
-  /**
-   * @param tableName 
-   */
-  public static Options tableName(String tableName) {
-    return new Options().tableName(tableName);
-  }
-  
-  /**
-   * @param config 
-   */
-  public static Options config(String config) {
-    return new Options().config(config);
-  }
-  
-  /**
-   * Parameter parameters updated by the stochastic gradient descent optimization algorithm.
-   */
-  public Output<TFloat32> parameters() {
-    return parameters;
-  }
-  
-  @Override
-  public Output<TFloat32> asOutput(Scope scope) {
-    return parameters;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "RetrieveTPUEmbeddingStochasticGradientDescentParameters";
-  
-  private Output<TFloat32> parameters;
-  
-  private RetrieveTPUEmbeddingStochasticGradientDescentParameters(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    parameters = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingStochasticGradientDescentParametersGradAccumDebug.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingStochasticGradientDescentParametersGradAccumDebug.java
deleted file mode 100644
index 9f35ffcd8e0..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingStochasticGradientDescentParametersGradAccumDebug.java
+++ /dev/null
@@ -1,154 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.tpu;
-
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-
-/**
- * Retrieve SGD embedding parameters with debug support.
- * <p>
- * An op that retrieves optimization parameters from embedding to host
- * memory. Must be preceded by a ConfigureTPUEmbeddingHost op that sets up
- * the correct embedding table configuration. For example, this op is
- * used to retrieve updated parameters before saving a checkpoint.
- */
-public final class RetrieveTPUEmbeddingStochasticGradientDescentParametersGradAccumDebug extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.tpu.RetrieveTPUEmbeddingStochasticGradientDescentParametersGradAccumDebug}
-   */
-  public static class Options {
-    
-    /**
-     * @param tableId 
-     */
-    public Options tableId(Long tableId) {
-      this.tableId = tableId;
-      return this;
-    }
-    
-    /**
-     * @param tableName 
-     */
-    public Options tableName(String tableName) {
-      this.tableName = tableName;
-      return this;
-    }
-    
-    /**
-     * @param config 
-     */
-    public Options config(String config) {
-      this.config = config;
-      return this;
-    }
-    
-    private Long tableId;
-    private String tableName;
-    private String config;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new RetrieveTPUEmbeddingStochasticGradientDescentParametersGradAccumDebug operation.
-   * 
-   * @param scope current scope
-   * @param numShards 
-   * @param shardId 
-   * @param options carries optional attributes values
-   * @return a new instance of RetrieveTPUEmbeddingStochasticGradientDescentParametersGradAccumDebug
-   */
-  @Endpoint(describeByClass = true)
-  public static RetrieveTPUEmbeddingStochasticGradientDescentParametersGradAccumDebug create(Scope scope, Long numShards, Long shardId, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("RetrieveTPUEmbeddingStochasticGradientDescentParametersGradAccumDebug", scope.makeOpName("RetrieveTPUEmbeddingStochasticGradientDescentParametersGradAccumDebug"));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("num_shards", numShards);
-    opBuilder.setAttr("shard_id", shardId);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.tableId != null) {
-          opBuilder.setAttr("table_id", opts.tableId);
-        }
-        if (opts.tableName != null) {
-          opBuilder.setAttr("table_name", opts.tableName);
-        }
-        if (opts.config != null) {
-          opBuilder.setAttr("config", opts.config);
-        }
-      }
-    }
-    return new RetrieveTPUEmbeddingStochasticGradientDescentParametersGradAccumDebug(opBuilder.build());
-  }
-  
-  /**
-   * @param tableId 
-   */
-  public static Options tableId(Long tableId) {
-    return new Options().tableId(tableId);
-  }
-  
-  /**
-   * @param tableName 
-   */
-  public static Options tableName(String tableName) {
-    return new Options().tableName(tableName);
-  }
-  
-  /**
-   * @param config 
-   */
-  public static Options config(String config) {
-    return new Options().config(config);
-  }
-  
-  /**
-   * Parameter parameters updated by the stochastic gradient descent optimization algorithm.
-   */
-  public Output<TFloat32> parameters() {
-    return parameters;
-  }
-  
-  /**
-   * Parameter gradient_accumulators updated by the Adadelta optimization algorithm.
-   */
-  public Output<TFloat32> gradientAccumulators() {
-    return gradientAccumulators;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "RetrieveTPUEmbeddingStochasticGradientDescentParametersGradAccumDebug";
-  
-  private Output<TFloat32> parameters;
-  private Output<TFloat32> gradientAccumulators;
-  
-  private RetrieveTPUEmbeddingStochasticGradientDescentParametersGradAccumDebug(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    parameters = operation.output(outputIdx++);
-    gradientAccumulators = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/SendTPUEmbeddingGradients.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/SendTPUEmbeddingGradients.java
deleted file mode 100644
index 2962d970bc3..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/SendTPUEmbeddingGradients.java
+++ /dev/null
@@ -1,70 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.tpu;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.Operands;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-
-/**
- * Performs gradient updates of embedding tables.
- */
-public final class SendTPUEmbeddingGradients extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new SendTPUEmbeddingGradients operation.
-   * 
-   * @param scope current scope
-   * @param inputs A TensorList of gradients with which to update embedding tables.
-   * This argument has the same length and shapes as the return value of
-   * RecvTPUEmbeddingActivations, but contains gradients of the model's loss
-   * with respect to the embedding activations. The embedding tables are updated
-   * from these gradients via the optimizer specified in the TPU embedding
-   * configuration given to tpu.initialize_system.
-   * @param learningRates A TensorList of float32 scalars, one for each dynamic learning
-   * rate tag: see the comments in
-   * //third_party/tensorflow/core/protobuf/tpu/optimization_parameters.proto.
-   * Multiple tables can share the same dynamic learning rate tag as specified
-   * in the configuration. If the learning rates for all tables are constant,
-   * this list should be empty.
-   * @param config Serialized TPUEmbeddingConfiguration proto.
-   * @return a new instance of SendTPUEmbeddingGradients
-   */
-  @Endpoint(describeByClass = true)
-  public static SendTPUEmbeddingGradients create(Scope scope, Iterable<Operand<TFloat32>> inputs, Iterable<Operand<TFloat32>> learningRates, String config) {
-    OperationBuilder opBuilder = scope.env().opBuilder("SendTPUEmbeddingGradients", scope.makeOpName("SendTPUEmbeddingGradients"));
-    opBuilder.addInputList(Operands.asOutputs(scope, inputs));
-    opBuilder.addInputList(Operands.asOutputs(scope, learningRates));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("config", config);
-    return new SendTPUEmbeddingGradients(opBuilder.build());
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "SendTPUEmbeddingGradients";
-  
-  private SendTPUEmbeddingGradients(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/ShutdownDistributedTPU.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/ShutdownDistributedTPU.java
deleted file mode 100644
index 9ce924d153a..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/ShutdownDistributedTPU.java
+++ /dev/null
@@ -1,53 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.tpu;
-
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-
-/**
- * Shuts down a running distributed TPU system.
- * <p>
- * The op returns an error if no system is running.
- */
-public final class ShutdownDistributedTPU extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new ShutdownDistributedTPU operation.
-   * 
-   * @param scope current scope
-   * @return a new instance of ShutdownDistributedTPU
-   */
-  @Endpoint(describeByClass = true)
-  public static ShutdownDistributedTPU create(Scope scope) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ShutdownDistributedTPU", scope.makeOpName("ShutdownDistributedTPU"));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new ShutdownDistributedTPU(opBuilder.build());
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ShutdownDistributedTPU";
-  
-  private ShutdownDistributedTPU(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/TPUCompilationResult.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/TPUCompilationResult.java
deleted file mode 100644
index 31423e6d6c7..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/TPUCompilationResult.java
+++ /dev/null
@@ -1,73 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.tpu;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TString;
-
-/**
- * Returns the result of a TPU compilation.
- * <p>
- * This operation returns the result of a TPU compilation as a serialized
- * CompilationResultProto, which holds a status and an error message if an error
- * occurred during compilation.
- */
-public final class TPUCompilationResult extends RawOp implements Operand<TString> {
-  
-  /**
-   * Factory method to create a class wrapping a new TPUCompilationResult operation.
-   * 
-   * @param scope current scope
-   * @return a new instance of TPUCompilationResult
-   */
-  @Endpoint(describeByClass = true)
-  public static TPUCompilationResult create(Scope scope) {
-    OperationBuilder opBuilder = scope.env().opBuilder("TPUCompilationResult", scope.makeOpName("TPUCompilationResult"));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new TPUCompilationResult(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<TString> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<TString> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "TPUCompilationResult";
-  
-  private Output<TString> output;
-  
-  private TPUCompilationResult(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/TPUEmbeddingActivations.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/TPUEmbeddingActivations.java
deleted file mode 100644
index f215a06df79..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/TPUEmbeddingActivations.java
+++ /dev/null
@@ -1,85 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.tpu;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-
-/**
- * An op enabling differentiation of TPU Embeddings.
- * <p>
- * This op simply returns its first input, which is assumed to have been sliced
- * from the Tensors returned by TPUEmbeddingDequeueActivations. The presence of
- * this op, and its first argument being a trainable Variable, enables automatic
- * differentiation of graphs containing embeddings via the TPU Embedding Python
- * libraries.
- */
-public final class TPUEmbeddingActivations extends RawOp implements Operand<TFloat32> {
-  
-  /**
-   * Factory method to create a class wrapping a new TPUEmbeddingActivations operation.
-   * 
-   * @param scope current scope
-   * @param embeddingVariable A trainable variable, enabling optimizers to find this op.
-   * @param slicedActivations The embedding activations Tensor to return.
-   * @param tableId The id of the table in the embedding layer configuration from which
-   * these activations were computed.
-   * @param lookupId Identifier of the set of embedding indices which produced these
-   * activations.
-   * @return a new instance of TPUEmbeddingActivations
-   */
-  @Endpoint(describeByClass = true)
-  public static TPUEmbeddingActivations create(Scope scope, Operand<TFloat32> embeddingVariable, Operand<TFloat32> slicedActivations, Long tableId, Long lookupId) {
-    OperationBuilder opBuilder = scope.env().opBuilder("TPUEmbeddingActivations", scope.makeOpName("TPUEmbeddingActivations"));
-    opBuilder.addInput(embeddingVariable.asOutput(scope));
-    opBuilder.addInput(slicedActivations.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("table_id", tableId);
-    opBuilder.setAttr("lookup_id", lookupId);
-    return new TPUEmbeddingActivations(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<TFloat32> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<TFloat32> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "TPUEmbeddingActivations";
-  
-  private Output<TFloat32> output;
-  
-  private TPUEmbeddingActivations(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/TPUOrdinalSelector.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/TPUOrdinalSelector.java
deleted file mode 100644
index 64ff5519c73..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/TPUOrdinalSelector.java
+++ /dev/null
@@ -1,74 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.tpu;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
-
-/**
- * A TPU core selector Op.
- * <p>
- * This Op produces a set of TPU cores (for warm-up) or a single TPU core
- * (for regular inference) to execute the TPU program on. The output is
- * consumed by TPUPartitionedCall.
- */
-public final class TPUOrdinalSelector extends RawOp implements Operand<TInt32> {
-  
-  /**
-   * Factory method to create a class wrapping a new TPUOrdinalSelector operation.
-   * 
-   * @param scope current scope
-   * @return a new instance of TPUOrdinalSelector
-   */
-  @Endpoint(describeByClass = true)
-  public static TPUOrdinalSelector create(Scope scope) {
-    OperationBuilder opBuilder = scope.env().opBuilder("TPUOrdinalSelector", scope.makeOpName("TPUOrdinalSelector"));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new TPUOrdinalSelector(opBuilder.build());
-  }
-  
-  /**
-   * A vector 1 or more TPU cores.
-   */
-  public Output<TInt32> deviceOrdinals() {
-    return deviceOrdinals;
-  }
-  
-  @Override
-  public Output<TInt32> asOutput(Scope scope) {
-    return deviceOrdinals;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "TPUOrdinalSelector";
-  
-  private Output<TInt32> deviceOrdinals;
-  
-  private TPUOrdinalSelector(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    deviceOrdinals = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/TPUReplicateMetadata.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/TPUReplicateMetadata.java
deleted file mode 100644
index 3fb9b782f9c..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/TPUReplicateMetadata.java
+++ /dev/null
@@ -1,258 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.tpu;
-
-import java.util.List;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-
-/**
- * Metadata indicating how the TPU computation should be replicated.
- * <p>
- * This operation holds the metadata common to operations of a `tpu.replicate()` computation subgraph.
- */
-public final class TPUReplicateMetadata extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.tpu.TPUReplicateMetadata}
-   */
-  public static class Options {
-    
-    /**
-     * @param numCoresPerReplica Number of cores per replica. Used for model parallelism.
-     */
-    public Options numCoresPerReplica(Long numCoresPerReplica) {
-      this.numCoresPerReplica = numCoresPerReplica;
-      return this;
-    }
-    
-    /**
-     * @param topology TopologyProto indicating the topology of the TPU pod slice.
-     */
-    public Options topology(String topology) {
-      this.topology = topology;
-      return this;
-    }
-    
-    /**
-     * @param useTpu Whether to place the computation on the TPU.
-     */
-    public Options useTpu(Boolean useTpu) {
-      this.useTpu = useTpu;
-      return this;
-    }
-    
-    /**
-     * @param deviceAssignment The assignment of devices for the computation.
-     */
-    public Options deviceAssignment(List<Long> deviceAssignment) {
-      this.deviceAssignment = deviceAssignment;
-      return this;
-    }
-    
-    /**
-     * @param computationShape DEPRECATED. Use num_cores_per_replica instead.
-     */
-    public Options computationShape(List<Long> computationShape) {
-      this.computationShape = computationShape;
-      return this;
-    }
-    
-    /**
-     * @param hostComputeCore 
-     */
-    public Options hostComputeCore(List<String> hostComputeCore) {
-      this.hostComputeCore = hostComputeCore;
-      return this;
-    }
-    
-    /**
-     * @param paddingMap 
-     */
-    public Options paddingMap(List<String> paddingMap) {
-      this.paddingMap = paddingMap;
-      return this;
-    }
-    
-    /**
-     * @param stepMarkerLocation 
-     */
-    public Options stepMarkerLocation(String stepMarkerLocation) {
-      this.stepMarkerLocation = stepMarkerLocation;
-      return this;
-    }
-    
-    /**
-     * @param allowSoftPlacement 
-     */
-    public Options allowSoftPlacement(Boolean allowSoftPlacement) {
-      this.allowSoftPlacement = allowSoftPlacement;
-      return this;
-    }
-    
-    private Long numCoresPerReplica;
-    private String topology;
-    private Boolean useTpu;
-    private List<Long> deviceAssignment;
-    private List<Long> computationShape;
-    private List<String> hostComputeCore;
-    private List<String> paddingMap;
-    private String stepMarkerLocation;
-    private Boolean allowSoftPlacement;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new TPUReplicateMetadata operation.
-   * 
-   * @param scope current scope
-   * @param numReplicas Number of replicas of the computation
-   * @param options carries optional attributes values
-   * @return a new instance of TPUReplicateMetadata
-   */
-  @Endpoint(describeByClass = true)
-  public static TPUReplicateMetadata create(Scope scope, Long numReplicas, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("TPUReplicateMetadata", scope.makeOpName("TPUReplicateMetadata"));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("num_replicas", numReplicas);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.numCoresPerReplica != null) {
-          opBuilder.setAttr("num_cores_per_replica", opts.numCoresPerReplica);
-        }
-        if (opts.topology != null) {
-          opBuilder.setAttr("topology", opts.topology);
-        }
-        if (opts.useTpu != null) {
-          opBuilder.setAttr("use_tpu", opts.useTpu);
-        }
-        if (opts.deviceAssignment != null) {
-          long[] deviceAssignmentArray = new long[opts.deviceAssignment.size()];
-          for (int i = 0; i < deviceAssignmentArray.length; ++i) {
-            deviceAssignmentArray[i] = opts.deviceAssignment.get(i);
-          }
-          opBuilder.setAttr("device_assignment", deviceAssignmentArray);
-        }
-        if (opts.computationShape != null) {
-          long[] computationShapeArray = new long[opts.computationShape.size()];
-          for (int i = 0; i < computationShapeArray.length; ++i) {
-            computationShapeArray[i] = opts.computationShape.get(i);
-          }
-          opBuilder.setAttr("computation_shape", computationShapeArray);
-        }
-        if (opts.hostComputeCore != null) {
-          String[] hostComputeCoreArray = new String[opts.hostComputeCore.size()];
-          for (int i = 0; i < hostComputeCoreArray.length; ++i) {
-            hostComputeCoreArray[i] = opts.hostComputeCore.get(i);
-          }
-          opBuilder.setAttr("host_compute_core", hostComputeCoreArray);
-        }
-        if (opts.paddingMap != null) {
-          String[] paddingMapArray = new String[opts.paddingMap.size()];
-          for (int i = 0; i < paddingMapArray.length; ++i) {
-            paddingMapArray[i] = opts.paddingMap.get(i);
-          }
-          opBuilder.setAttr("padding_map", paddingMapArray);
-        }
-        if (opts.stepMarkerLocation != null) {
-          opBuilder.setAttr("step_marker_location", opts.stepMarkerLocation);
-        }
-        if (opts.allowSoftPlacement != null) {
-          opBuilder.setAttr("allow_soft_placement", opts.allowSoftPlacement);
-        }
-      }
-    }
-    return new TPUReplicateMetadata(opBuilder.build());
-  }
-  
-  /**
-   * @param numCoresPerReplica Number of cores per replica. Used for model parallelism.
-   */
-  public static Options numCoresPerReplica(Long numCoresPerReplica) {
-    return new Options().numCoresPerReplica(numCoresPerReplica);
-  }
-  
-  /**
-   * @param topology TopologyProto indicating the topology of the TPU pod slice.
-   */
-  public static Options topology(String topology) {
-    return new Options().topology(topology);
-  }
-  
-  /**
-   * @param useTpu Whether to place the computation on the TPU.
-   */
-  public static Options useTpu(Boolean useTpu) {
-    return new Options().useTpu(useTpu);
-  }
-  
-  /**
-   * @param deviceAssignment The assignment of devices for the computation.
-   */
-  public static Options deviceAssignment(List<Long> deviceAssignment) {
-    return new Options().deviceAssignment(deviceAssignment);
-  }
-  
-  /**
-   * @param computationShape DEPRECATED. Use num_cores_per_replica instead.
-   */
-  public static Options computationShape(List<Long> computationShape) {
-    return new Options().computationShape(computationShape);
-  }
-  
-  /**
-   * @param hostComputeCore 
-   */
-  public static Options hostComputeCore(List<String> hostComputeCore) {
-    return new Options().hostComputeCore(hostComputeCore);
-  }
-  
-  /**
-   * @param paddingMap 
-   */
-  public static Options paddingMap(List<String> paddingMap) {
-    return new Options().paddingMap(paddingMap);
-  }
-  
-  /**
-   * @param stepMarkerLocation 
-   */
-  public static Options stepMarkerLocation(String stepMarkerLocation) {
-    return new Options().stepMarkerLocation(stepMarkerLocation);
-  }
-  
-  /**
-   * @param allowSoftPlacement 
-   */
-  public static Options allowSoftPlacement(Boolean allowSoftPlacement) {
-    return new Options().allowSoftPlacement(allowSoftPlacement);
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "TPUReplicateMetadata";
-  
-  private TPUReplicateMetadata(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/TPUReplicatedInput.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/TPUReplicatedInput.java
deleted file mode 100644
index 597865ffa6d..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/TPUReplicatedInput.java
+++ /dev/null
@@ -1,158 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.tpu;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.Operands;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Connects N inputs to an N-way replicated TPU computation.
- * <p>
- * This operation holds a replicated input to a `tpu.replicate()` computation subgraph.
- * Each replicated input has the same shape and type alongside the output.
- * <p>
- * For example:
- * <pre>{@code
- * %a = "tf.opA"()
- * %b = "tf.opB"()
- * %replicated_input = "tf.TPUReplicatedInput"(%a, %b)
- * %computation = "tf.Computation"(%replicated_input)
- * }</pre>
- * The above computation has a replicated input of two replicas.
- * 
- * @param <T> data type for {@code output()} output
- */
-public final class TPUReplicatedInput<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.tpu.TPUReplicatedInput}
-   */
-  public static class Options {
-    
-    /**
-     * @param isMirroredVariable 
-     */
-    public Options isMirroredVariable(Boolean isMirroredVariable) {
-      this.isMirroredVariable = isMirroredVariable;
-      return this;
-    }
-    
-    /**
-     * @param index 
-     */
-    public Options index(Long index) {
-      this.index = index;
-      return this;
-    }
-    
-    /**
-     * @param isPacked 
-     */
-    public Options isPacked(Boolean isPacked) {
-      this.isPacked = isPacked;
-      return this;
-    }
-    
-    private Boolean isMirroredVariable;
-    private Long index;
-    private Boolean isPacked;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new TPUReplicatedInput operation.
-   * 
-   * @param scope current scope
-   * @param inputs 
-   * @param options carries optional attributes values
-   * @return a new instance of TPUReplicatedInput
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> TPUReplicatedInput<T> create(Scope scope, Iterable<Operand<T>> inputs, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("TPUReplicatedInput", scope.makeOpName("TPUReplicatedInput"));
-    opBuilder.addInputList(Operands.asOutputs(scope, inputs));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.isMirroredVariable != null) {
-          opBuilder.setAttr("is_mirrored_variable", opts.isMirroredVariable);
-        }
-        if (opts.index != null) {
-          opBuilder.setAttr("index", opts.index);
-        }
-        if (opts.isPacked != null) {
-          opBuilder.setAttr("is_packed", opts.isPacked);
-        }
-      }
-    }
-    return new TPUReplicatedInput<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param isMirroredVariable 
-   */
-  public static Options isMirroredVariable(Boolean isMirroredVariable) {
-    return new Options().isMirroredVariable(isMirroredVariable);
-  }
-  
-  /**
-   * @param index 
-   */
-  public static Options index(Long index) {
-    return new Options().index(index);
-  }
-  
-  /**
-   * @param isPacked 
-   */
-  public static Options isPacked(Boolean isPacked) {
-    return new Options().isPacked(isPacked);
-  }
-  
-  /**
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "TPUReplicatedInput";
-  
-  private Output<T> output;
-  
-  private TPUReplicatedInput(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/TPUReplicatedOutput.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/TPUReplicatedOutput.java
deleted file mode 100644
index d7d2487b90c..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/TPUReplicatedOutput.java
+++ /dev/null
@@ -1,92 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.tpu;
-
-import java.util.Arrays;
-import java.util.Iterator;
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Connects N outputs from an N-way replicated TPU computation.
- * <p>
- * This operation holds a replicated output from a `tpu.replicate()` computation subgraph.
- * Each replicated output has the same shape and type alongside the input.
- * <p>
- * For example:
- * <pre>{@code
- * %computation = "tf.Computation"()
- * %replicated_output:2 = "tf.TPUReplicatedOutput"(%computation)
- * }</pre>
- * The above computation has a replicated output of two replicas.
- * 
- * @param <T> data type for {@code outputs()} output
- */
-public final class TPUReplicatedOutput<T extends TType> extends RawOp implements Iterable<Operand<T>> {
-  
-  /**
-   * Factory method to create a class wrapping a new TPUReplicatedOutput operation.
-   * 
-   * @param scope current scope
-   * @param input 
-   * @param numReplicas 
-   * @return a new instance of TPUReplicatedOutput
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> TPUReplicatedOutput<T> create(Scope scope, Operand<T> input, Long numReplicas) {
-    OperationBuilder opBuilder = scope.env().opBuilder("TPUReplicatedOutput", scope.makeOpName("TPUReplicatedOutput"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("num_replicas", numReplicas);
-    return new TPUReplicatedOutput<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public List<Output<T>> outputs() {
-    return outputs;
-  }
-  
-  @Override
-  @SuppressWarnings({"rawtypes", "unchecked"})
-  public Iterator<Operand<T>> iterator() {
-    return (Iterator) outputs.iterator();
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "TPUReplicatedOutput";
-  
-  private List<Output<T>> outputs;
-  
-  @SuppressWarnings("unchecked")
-  private TPUReplicatedOutput(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    int outputsLength = operation.outputListLength("outputs");
-    outputs = Arrays.asList((Output<T>[])operation.outputList(outputIdx, outputsLength));
-    outputIdx += outputsLength;
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/WorkerHeartbeat.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/WorkerHeartbeat.java
deleted file mode 100644
index 9ed34561956..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/WorkerHeartbeat.java
+++ /dev/null
@@ -1,75 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.tpu;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TString;
-
-/**
- * Worker heartbeat op.
- * <p>
- * Heartbeats may be sent periodically to indicate the coordinator is still active,
- * to retrieve the current worker status and to expedite shutdown when necessary.
- */
-public final class WorkerHeartbeat extends RawOp implements Operand<TString> {
-  
-  /**
-   * Factory method to create a class wrapping a new WorkerHeartbeat operation.
-   * 
-   * @param scope current scope
-   * @param request A string tensor containing a serialized WorkerHeartbeatRequest
-   * @return a new instance of WorkerHeartbeat
-   */
-  @Endpoint(describeByClass = true)
-  public static WorkerHeartbeat create(Scope scope, Operand<TString> request) {
-    OperationBuilder opBuilder = scope.env().opBuilder("WorkerHeartbeat", scope.makeOpName("WorkerHeartbeat"));
-    opBuilder.addInput(request.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new WorkerHeartbeat(opBuilder.build());
-  }
-  
-  /**
-   * A string tensor containing a serialized WorkerHeartbeatResponse
-   */
-  public Output<TString> response() {
-    return response;
-  }
-  
-  @Override
-  public Output<TString> asOutput(Scope scope) {
-    return response;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "WorkerHeartbeat";
-  
-  private Output<TString> response;
-  
-  private WorkerHeartbeat(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    response = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/AccumulatorApplyGradient.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/AccumulatorApplyGradient.java
deleted file mode 100644
index 685ebe0de74..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/AccumulatorApplyGradient.java
+++ /dev/null
@@ -1,64 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.train;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.TString;
-import org.tensorflow.types.family.TType;
-
-/**
- * Applies a gradient to a given accumulator.
- * <p>
- * Does not add if local_step is lesser than the accumulator's global_step.
- */
-@Operator(group = "train")
-public final class AccumulatorApplyGradient extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new AccumulatorApplyGradient operation.
-   * 
-   * @param scope current scope
-   * @param handle The handle to a accumulator.
-   * @param localStep The local_step value at which the gradient was computed.
-   * @param gradient A tensor of the gradient to be accumulated.
-   * @return a new instance of AccumulatorApplyGradient
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> AccumulatorApplyGradient create(Scope scope, Operand<TString> handle, Operand<TInt64> localStep, Operand<T> gradient) {
-    OperationBuilder opBuilder = scope.env().opBuilder("AccumulatorApplyGradient", scope.makeOpName("AccumulatorApplyGradient"));
-    opBuilder.addInput(handle.asOutput(scope));
-    opBuilder.addInput(localStep.asOutput(scope));
-    opBuilder.addInput(gradient.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new AccumulatorApplyGradient(opBuilder.build());
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "AccumulatorApplyGradient";
-  
-  private AccumulatorApplyGradient(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/AccumulatorNumAccumulated.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/AccumulatorNumAccumulated.java
deleted file mode 100644
index cb8fc93f7aa..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/AccumulatorNumAccumulated.java
+++ /dev/null
@@ -1,74 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.train;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.TString;
-
-/**
- * Returns the number of gradients aggregated in the given accumulators.
- */
-@Operator(group = "train")
-public final class AccumulatorNumAccumulated extends RawOp implements Operand<TInt32> {
-  
-  /**
-   * Factory method to create a class wrapping a new AccumulatorNumAccumulated operation.
-   * 
-   * @param scope current scope
-   * @param handle The handle to an accumulator.
-   * @return a new instance of AccumulatorNumAccumulated
-   */
-  @Endpoint(describeByClass = true)
-  public static AccumulatorNumAccumulated create(Scope scope, Operand<TString> handle) {
-    OperationBuilder opBuilder = scope.env().opBuilder("AccumulatorNumAccumulated", scope.makeOpName("AccumulatorNumAccumulated"));
-    opBuilder.addInput(handle.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new AccumulatorNumAccumulated(opBuilder.build());
-  }
-  
-  /**
-   * The number of gradients aggregated in the given accumulator.
-   */
-  public Output<TInt32> numAccumulated() {
-    return numAccumulated;
-  }
-  
-  @Override
-  public Output<TInt32> asOutput(Scope scope) {
-    return numAccumulated;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "AccumulatorNumAccumulated";
-  
-  private Output<TInt32> numAccumulated;
-  
-  private AccumulatorNumAccumulated(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    numAccumulated = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/AccumulatorSetGlobalStep.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/AccumulatorSetGlobalStep.java
deleted file mode 100644
index 8c354475de3..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/AccumulatorSetGlobalStep.java
+++ /dev/null
@@ -1,62 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.train;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.TString;
-
-/**
- * Updates the accumulator with a new value for global_step.
- * <p>
- * Logs warning if the accumulator's value is already higher than
- * new_global_step.
- */
-@Operator(group = "train")
-public final class AccumulatorSetGlobalStep extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new AccumulatorSetGlobalStep operation.
-   * 
-   * @param scope current scope
-   * @param handle The handle to an accumulator.
-   * @param newGlobalStep The new global_step value to set.
-   * @return a new instance of AccumulatorSetGlobalStep
-   */
-  @Endpoint(describeByClass = true)
-  public static AccumulatorSetGlobalStep create(Scope scope, Operand<TString> handle, Operand<TInt64> newGlobalStep) {
-    OperationBuilder opBuilder = scope.env().opBuilder("AccumulatorSetGlobalStep", scope.makeOpName("AccumulatorSetGlobalStep"));
-    opBuilder.addInput(handle.asOutput(scope));
-    opBuilder.addInput(newGlobalStep.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new AccumulatorSetGlobalStep(opBuilder.build());
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "AccumulatorSetGlobalStep";
-  
-  private AccumulatorSetGlobalStep(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/AccumulatorTakeGradient.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/AccumulatorTakeGradient.java
deleted file mode 100644
index 70c01e2ada6..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/AccumulatorTakeGradient.java
+++ /dev/null
@@ -1,88 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.train;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.TString;
-import org.tensorflow.types.family.TType;
-
-/**
- * Extracts the average gradient in the given ConditionalAccumulator.
- * <p>
- * The op blocks until sufficient (i.e., more than num_required)
- * gradients have been accumulated.  If the accumulator has already
- * aggregated more than num_required gradients, it returns the average of
- * the accumulated gradients.  Also automatically increments the recorded
- * global_step in the accumulator by 1, and resets the aggregate to 0.
- * 
- * @param <T> data type for {@code average()} output
- */
-@Operator(group = "train")
-public final class AccumulatorTakeGradient<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new AccumulatorTakeGradient operation.
-   * 
-   * @param scope current scope
-   * @param handle The handle to an accumulator.
-   * @param numRequired Number of gradients required before we return an aggregate.
-   * @param dtype The data type of accumulated gradients. Needs to correspond to the type
-   * of the accumulator.
-   * @return a new instance of AccumulatorTakeGradient
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> AccumulatorTakeGradient<T> create(Scope scope, Operand<TString> handle, Operand<TInt32> numRequired, Class<T> dtype) {
-    OperationBuilder opBuilder = scope.env().opBuilder("AccumulatorTakeGradient", scope.makeOpName("AccumulatorTakeGradient"));
-    opBuilder.addInput(handle.asOutput(scope));
-    opBuilder.addInput(numRequired.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("dtype", dtype);
-    return new AccumulatorTakeGradient<T>(opBuilder.build());
-  }
-  
-  /**
-   * The average of the accumulated gradients.
-   */
-  public Output<T> average() {
-    return average;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return average;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "AccumulatorTakeGradient";
-  
-  private Output<T> average;
-  
-  private AccumulatorTakeGradient(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    average = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyAdaMax.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyAdaMax.java
deleted file mode 100644
index 8529a8d695b..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyAdaMax.java
+++ /dev/null
@@ -1,132 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.train;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Update '*var' according to the AdaMax algorithm.
- * <p>
- * m_t <- beta1 * m_{t-1} + (1 - beta1) * g
- * v_t <- max(beta2 * v_{t-1}, abs(g))
- * variable <- variable - learning_rate / (1 - beta1^t) * m_t / (v_t + epsilon)
- * 
- * @param <T> data type for {@code out()} output
- */
-public final class ApplyAdaMax<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.train.ApplyAdaMax}
-   */
-  public static class Options {
-    
-    /**
-     * @param useLocking If `True`, updating of the var, m, and v tensors will be protected
-     * by a lock; otherwise the behavior is undefined, but may exhibit less
-     * contention.
-     */
-    public Options useLocking(Boolean useLocking) {
-      this.useLocking = useLocking;
-      return this;
-    }
-    
-    private Boolean useLocking;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new ApplyAdaMax operation.
-   * 
-   * @param scope current scope
-   * @param var Should be from a Variable().
-   * @param m Should be from a Variable().
-   * @param v Should be from a Variable().
-   * @param beta1Power Must be a scalar.
-   * @param lr Scaling factor. Must be a scalar.
-   * @param beta1 Momentum factor. Must be a scalar.
-   * @param beta2 Momentum factor. Must be a scalar.
-   * @param epsilon Ridge term. Must be a scalar.
-   * @param grad The gradient.
-   * @param options carries optional attributes values
-   * @return a new instance of ApplyAdaMax
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> ApplyAdaMax<T> create(Scope scope, Operand<T> var, Operand<T> m, Operand<T> v, Operand<T> beta1Power, Operand<T> lr, Operand<T> beta1, Operand<T> beta2, Operand<T> epsilon, Operand<T> grad, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ApplyAdaMax", scope.makeOpName("ApplyAdaMax"));
-    opBuilder.addInput(var.asOutput(scope));
-    opBuilder.addInput(m.asOutput(scope));
-    opBuilder.addInput(v.asOutput(scope));
-    opBuilder.addInput(beta1Power.asOutput(scope));
-    opBuilder.addInput(lr.asOutput(scope));
-    opBuilder.addInput(beta1.asOutput(scope));
-    opBuilder.addInput(beta2.asOutput(scope));
-    opBuilder.addInput(epsilon.asOutput(scope));
-    opBuilder.addInput(grad.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.useLocking != null) {
-          opBuilder.setAttr("use_locking", opts.useLocking);
-        }
-      }
-    }
-    return new ApplyAdaMax<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param useLocking If `True`, updating of the var, m, and v tensors will be protected
-   * by a lock; otherwise the behavior is undefined, but may exhibit less
-   * contention.
-   */
-  public static Options useLocking(Boolean useLocking) {
-    return new Options().useLocking(useLocking);
-  }
-  
-  /**
-   * Same as "var".
-   */
-  public Output<T> out() {
-    return out;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return out;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ApplyAdaMax";
-  
-  private Output<T> out;
-  
-  private ApplyAdaMax(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    out = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyAdadelta.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyAdadelta.java
deleted file mode 100644
index cef6fc27612..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyAdadelta.java
+++ /dev/null
@@ -1,128 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.train;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Update '*var' according to the adadelta scheme.
- * <p>
- * accum = rho() * accum + (1 - rho()) * grad.square();
- * update = (update_accum + epsilon).sqrt() * (accum + epsilon()).rsqrt() * grad;
- * update_accum = rho() * update_accum + (1 - rho()) * update.square();
- * var -= update;
- * 
- * @param <T> data type for {@code out()} output
- */
-@Operator(group = "train")
-public final class ApplyAdadelta<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.train.ApplyAdadelta}
-   */
-  public static class Options {
-    
-    /**
-     * @param useLocking If True, updating of the var, accum and update_accum tensors will be protected by
-     * a lock; otherwise the behavior is undefined, but may exhibit less contention.
-     */
-    public Options useLocking(Boolean useLocking) {
-      this.useLocking = useLocking;
-      return this;
-    }
-    
-    private Boolean useLocking;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new ApplyAdadelta operation.
-   * 
-   * @param scope current scope
-   * @param var Should be from a Variable().
-   * @param accum Should be from a Variable().
-   * @param accumUpdate Should be from a Variable().
-   * @param lr Scaling factor. Must be a scalar.
-   * @param rho Decay factor. Must be a scalar.
-   * @param epsilon Constant factor. Must be a scalar.
-   * @param grad The gradient.
-   * @param options carries optional attributes values
-   * @return a new instance of ApplyAdadelta
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> ApplyAdadelta<T> create(Scope scope, Operand<T> var, Operand<T> accum, Operand<T> accumUpdate, Operand<T> lr, Operand<T> rho, Operand<T> epsilon, Operand<T> grad, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ApplyAdadelta", scope.makeOpName("ApplyAdadelta"));
-    opBuilder.addInput(var.asOutput(scope));
-    opBuilder.addInput(accum.asOutput(scope));
-    opBuilder.addInput(accumUpdate.asOutput(scope));
-    opBuilder.addInput(lr.asOutput(scope));
-    opBuilder.addInput(rho.asOutput(scope));
-    opBuilder.addInput(epsilon.asOutput(scope));
-    opBuilder.addInput(grad.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.useLocking != null) {
-          opBuilder.setAttr("use_locking", opts.useLocking);
-        }
-      }
-    }
-    return new ApplyAdadelta<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param useLocking If True, updating of the var, accum and update_accum tensors will be protected by
-   * a lock; otherwise the behavior is undefined, but may exhibit less contention.
-   */
-  public static Options useLocking(Boolean useLocking) {
-    return new Options().useLocking(useLocking);
-  }
-  
-  /**
-   * Same as "var".
-   */
-  public Output<T> out() {
-    return out;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return out;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ApplyAdadelta";
-  
-  private Output<T> out;
-  
-  private ApplyAdadelta(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    out = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyAdagrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyAdagrad.java
deleted file mode 100644
index 725ebde8e91..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyAdagrad.java
+++ /dev/null
@@ -1,141 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.train;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Update '*var' according to the adagrad scheme.
- * <p>
- * accum += grad * grad
- * var -= lr * grad * (1 / sqrt(accum))
- * 
- * @param <T> data type for {@code out()} output
- */
-@Operator(group = "train")
-public final class ApplyAdagrad<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.train.ApplyAdagrad}
-   */
-  public static class Options {
-    
-    /**
-     * @param useLocking If `True`, updating of the var and accum tensors will be protected
-     * by a lock; otherwise the behavior is undefined, but may exhibit less
-     * contention.
-     */
-    public Options useLocking(Boolean useLocking) {
-      this.useLocking = useLocking;
-      return this;
-    }
-    
-    /**
-     * @param updateSlots 
-     */
-    public Options updateSlots(Boolean updateSlots) {
-      this.updateSlots = updateSlots;
-      return this;
-    }
-    
-    private Boolean useLocking;
-    private Boolean updateSlots;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new ApplyAdagrad operation.
-   * 
-   * @param scope current scope
-   * @param var Should be from a Variable().
-   * @param accum Should be from a Variable().
-   * @param lr Scaling factor. Must be a scalar.
-   * @param grad The gradient.
-   * @param options carries optional attributes values
-   * @return a new instance of ApplyAdagrad
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> ApplyAdagrad<T> create(Scope scope, Operand<T> var, Operand<T> accum, Operand<T> lr, Operand<T> grad, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ApplyAdagrad", scope.makeOpName("ApplyAdagrad"));
-    opBuilder.addInput(var.asOutput(scope));
-    opBuilder.addInput(accum.asOutput(scope));
-    opBuilder.addInput(lr.asOutput(scope));
-    opBuilder.addInput(grad.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.useLocking != null) {
-          opBuilder.setAttr("use_locking", opts.useLocking);
-        }
-        if (opts.updateSlots != null) {
-          opBuilder.setAttr("update_slots", opts.updateSlots);
-        }
-      }
-    }
-    return new ApplyAdagrad<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param useLocking If `True`, updating of the var and accum tensors will be protected
-   * by a lock; otherwise the behavior is undefined, but may exhibit less
-   * contention.
-   */
-  public static Options useLocking(Boolean useLocking) {
-    return new Options().useLocking(useLocking);
-  }
-  
-  /**
-   * @param updateSlots 
-   */
-  public static Options updateSlots(Boolean updateSlots) {
-    return new Options().updateSlots(updateSlots);
-  }
-  
-  /**
-   * Same as "var".
-   */
-  public Output<T> out() {
-    return out;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return out;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ApplyAdagrad";
-  
-  private Output<T> out;
-  
-  private ApplyAdagrad(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    out = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyAdagradDa.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyAdagradDa.java
deleted file mode 100644
index 4c53fb860fe..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyAdagradDa.java
+++ /dev/null
@@ -1,126 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.train;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
-
-/**
- * Update '*var' according to the proximal adagrad scheme.
- * 
- * @param <T> data type for {@code out()} output
- */
-@Operator(group = "train")
-public final class ApplyAdagradDa<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.train.ApplyAdagradDa}
-   */
-  public static class Options {
-    
-    /**
-     * @param useLocking If True, updating of the var and accum tensors will be protected by
-     * a lock; otherwise the behavior is undefined, but may exhibit less contention.
-     */
-    public Options useLocking(Boolean useLocking) {
-      this.useLocking = useLocking;
-      return this;
-    }
-    
-    private Boolean useLocking;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new ApplyAdagradDa operation.
-   * 
-   * @param scope current scope
-   * @param var Should be from a Variable().
-   * @param gradientAccumulator Should be from a Variable().
-   * @param gradientSquaredAccumulator Should be from a Variable().
-   * @param grad The gradient.
-   * @param lr Scaling factor. Must be a scalar.
-   * @param l1 L1 regularization. Must be a scalar.
-   * @param l2 L2 regularization. Must be a scalar.
-   * @param globalStep Training step number. Must be a scalar.
-   * @param options carries optional attributes values
-   * @return a new instance of ApplyAdagradDa
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> ApplyAdagradDa<T> create(Scope scope, Operand<T> var, Operand<T> gradientAccumulator, Operand<T> gradientSquaredAccumulator, Operand<T> grad, Operand<T> lr, Operand<T> l1, Operand<T> l2, Operand<TInt64> globalStep, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ApplyAdagradDA", scope.makeOpName("ApplyAdagradDa"));
-    opBuilder.addInput(var.asOutput(scope));
-    opBuilder.addInput(gradientAccumulator.asOutput(scope));
-    opBuilder.addInput(gradientSquaredAccumulator.asOutput(scope));
-    opBuilder.addInput(grad.asOutput(scope));
-    opBuilder.addInput(lr.asOutput(scope));
-    opBuilder.addInput(l1.asOutput(scope));
-    opBuilder.addInput(l2.asOutput(scope));
-    opBuilder.addInput(globalStep.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.useLocking != null) {
-          opBuilder.setAttr("use_locking", opts.useLocking);
-        }
-      }
-    }
-    return new ApplyAdagradDa<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param useLocking If True, updating of the var and accum tensors will be protected by
-   * a lock; otherwise the behavior is undefined, but may exhibit less contention.
-   */
-  public static Options useLocking(Boolean useLocking) {
-    return new Options().useLocking(useLocking);
-  }
-  
-  /**
-   * Same as "var".
-   */
-  public Output<T> out() {
-    return out;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return out;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ApplyAdagradDA";
-  
-  private Output<T> out;
-  
-  private ApplyAdagradDa(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    out = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyAdagradV2.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyAdagradV2.java
deleted file mode 100644
index 3d0e00bcfea..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyAdagradV2.java
+++ /dev/null
@@ -1,142 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.train;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Update '*var' according to the adagrad scheme.
- * <p>
- * accum += grad * grad
- * var -= lr * grad * (1 / sqrt(accum))
- * 
- * @param <T> data type for {@code out()} output
- */
-public final class ApplyAdagradV2<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.train.ApplyAdagradV2}
-   */
-  public static class Options {
-    
-    /**
-     * @param useLocking If `True`, updating of the var and accum tensors will be protected
-     * by a lock; otherwise the behavior is undefined, but may exhibit less
-     * contention.
-     */
-    public Options useLocking(Boolean useLocking) {
-      this.useLocking = useLocking;
-      return this;
-    }
-    
-    /**
-     * @param updateSlots 
-     */
-    public Options updateSlots(Boolean updateSlots) {
-      this.updateSlots = updateSlots;
-      return this;
-    }
-    
-    private Boolean useLocking;
-    private Boolean updateSlots;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new ApplyAdagradV2 operation.
-   * 
-   * @param scope current scope
-   * @param var Should be from a Variable().
-   * @param accum Should be from a Variable().
-   * @param lr Scaling factor. Must be a scalar.
-   * @param epsilon Constant factor. Must be a scalar.
-   * @param grad The gradient.
-   * @param options carries optional attributes values
-   * @return a new instance of ApplyAdagradV2
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> ApplyAdagradV2<T> create(Scope scope, Operand<T> var, Operand<T> accum, Operand<T> lr, Operand<T> epsilon, Operand<T> grad, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ApplyAdagradV2", scope.makeOpName("ApplyAdagradV2"));
-    opBuilder.addInput(var.asOutput(scope));
-    opBuilder.addInput(accum.asOutput(scope));
-    opBuilder.addInput(lr.asOutput(scope));
-    opBuilder.addInput(epsilon.asOutput(scope));
-    opBuilder.addInput(grad.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.useLocking != null) {
-          opBuilder.setAttr("use_locking", opts.useLocking);
-        }
-        if (opts.updateSlots != null) {
-          opBuilder.setAttr("update_slots", opts.updateSlots);
-        }
-      }
-    }
-    return new ApplyAdagradV2<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param useLocking If `True`, updating of the var and accum tensors will be protected
-   * by a lock; otherwise the behavior is undefined, but may exhibit less
-   * contention.
-   */
-  public static Options useLocking(Boolean useLocking) {
-    return new Options().useLocking(useLocking);
-  }
-  
-  /**
-   * @param updateSlots 
-   */
-  public static Options updateSlots(Boolean updateSlots) {
-    return new Options().updateSlots(updateSlots);
-  }
-  
-  /**
-   * Same as "var".
-   */
-  public Output<T> out() {
-    return out;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return out;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ApplyAdagradV2";
-  
-  private Output<T> out;
-  
-  private ApplyAdagradV2(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    out = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyAdam.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyAdam.java
deleted file mode 100644
index 1a212a4f9fd..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyAdam.java
+++ /dev/null
@@ -1,155 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.train;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Update '*var' according to the Adam algorithm.
- * <p>
- * $$lr_t := \text{learning\_rate} * \sqrt{1 - beta_2^t} / (1 - beta_1^t)$$
- * $$m_t := beta_1 * m_{t-1} + (1 - beta_1) * g$$
- * $$v_t := beta_2 * v_{t-1} + (1 - beta_2) * g * g$$
- * $$variable := variable - lr_t * m_t / (\sqrt{v_t} + \epsilon)$$
- * 
- * @param <T> data type for {@code out()} output
- */
-@Operator(group = "train")
-public final class ApplyAdam<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.train.ApplyAdam}
-   */
-  public static class Options {
-    
-    /**
-     * @param useLocking If `True`, updating of the var, m, and v tensors will be protected
-     * by a lock; otherwise the behavior is undefined, but may exhibit less
-     * contention.
-     */
-    public Options useLocking(Boolean useLocking) {
-      this.useLocking = useLocking;
-      return this;
-    }
-    
-    /**
-     * @param useNesterov If `True`, uses the nesterov update.
-     */
-    public Options useNesterov(Boolean useNesterov) {
-      this.useNesterov = useNesterov;
-      return this;
-    }
-    
-    private Boolean useLocking;
-    private Boolean useNesterov;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new ApplyAdam operation.
-   * 
-   * @param scope current scope
-   * @param var Should be from a Variable().
-   * @param m Should be from a Variable().
-   * @param v Should be from a Variable().
-   * @param beta1Power Must be a scalar.
-   * @param beta2Power Must be a scalar.
-   * @param lr Scaling factor. Must be a scalar.
-   * @param beta1 Momentum factor. Must be a scalar.
-   * @param beta2 Momentum factor. Must be a scalar.
-   * @param epsilon Ridge term. Must be a scalar.
-   * @param grad The gradient.
-   * @param options carries optional attributes values
-   * @return a new instance of ApplyAdam
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> ApplyAdam<T> create(Scope scope, Operand<T> var, Operand<T> m, Operand<T> v, Operand<T> beta1Power, Operand<T> beta2Power, Operand<T> lr, Operand<T> beta1, Operand<T> beta2, Operand<T> epsilon, Operand<T> grad, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ApplyAdam", scope.makeOpName("ApplyAdam"));
-    opBuilder.addInput(var.asOutput(scope));
-    opBuilder.addInput(m.asOutput(scope));
-    opBuilder.addInput(v.asOutput(scope));
-    opBuilder.addInput(beta1Power.asOutput(scope));
-    opBuilder.addInput(beta2Power.asOutput(scope));
-    opBuilder.addInput(lr.asOutput(scope));
-    opBuilder.addInput(beta1.asOutput(scope));
-    opBuilder.addInput(beta2.asOutput(scope));
-    opBuilder.addInput(epsilon.asOutput(scope));
-    opBuilder.addInput(grad.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.useLocking != null) {
-          opBuilder.setAttr("use_locking", opts.useLocking);
-        }
-        if (opts.useNesterov != null) {
-          opBuilder.setAttr("use_nesterov", opts.useNesterov);
-        }
-      }
-    }
-    return new ApplyAdam<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param useLocking If `True`, updating of the var, m, and v tensors will be protected
-   * by a lock; otherwise the behavior is undefined, but may exhibit less
-   * contention.
-   */
-  public static Options useLocking(Boolean useLocking) {
-    return new Options().useLocking(useLocking);
-  }
-  
-  /**
-   * @param useNesterov If `True`, uses the nesterov update.
-   */
-  public static Options useNesterov(Boolean useNesterov) {
-    return new Options().useNesterov(useNesterov);
-  }
-  
-  /**
-   * Same as "var".
-   */
-  public Output<T> out() {
-    return out;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return out;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ApplyAdam";
-  
-  private Output<T> out;
-  
-  private ApplyAdam(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    out = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyAddSign.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyAddSign.java
deleted file mode 100644
index ee3cb5e20a2..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyAddSign.java
+++ /dev/null
@@ -1,129 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.train;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Update '*var' according to the AddSign update.
- * <p>
- * m_t <- beta1 * m_{t-1} + (1 - beta1) * g
- * update <- (alpha + sign_decay * sign(g) *sign(m)) * g
- * variable <- variable - lr_t * update
- * 
- * @param <T> data type for {@code out()} output
- */
-@Operator(group = "train")
-public final class ApplyAddSign<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.train.ApplyAddSign}
-   */
-  public static class Options {
-    
-    /**
-     * @param useLocking If `True`, updating of the var and m tensors is
-     * protected by a lock; otherwise the behavior is undefined, but may exhibit less
-     * contention.
-     */
-    public Options useLocking(Boolean useLocking) {
-      this.useLocking = useLocking;
-      return this;
-    }
-    
-    private Boolean useLocking;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new ApplyAddSign operation.
-   * 
-   * @param scope current scope
-   * @param var Should be from a Variable().
-   * @param m Should be from a Variable().
-   * @param lr Scaling factor. Must be a scalar.
-   * @param alpha Must be a scalar.
-   * @param signDecay Must be a scalar.
-   * @param beta Must be a scalar.
-   * @param grad The gradient.
-   * @param options carries optional attributes values
-   * @return a new instance of ApplyAddSign
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> ApplyAddSign<T> create(Scope scope, Operand<T> var, Operand<T> m, Operand<T> lr, Operand<T> alpha, Operand<T> signDecay, Operand<T> beta, Operand<T> grad, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ApplyAddSign", scope.makeOpName("ApplyAddSign"));
-    opBuilder.addInput(var.asOutput(scope));
-    opBuilder.addInput(m.asOutput(scope));
-    opBuilder.addInput(lr.asOutput(scope));
-    opBuilder.addInput(alpha.asOutput(scope));
-    opBuilder.addInput(signDecay.asOutput(scope));
-    opBuilder.addInput(beta.asOutput(scope));
-    opBuilder.addInput(grad.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.useLocking != null) {
-          opBuilder.setAttr("use_locking", opts.useLocking);
-        }
-      }
-    }
-    return new ApplyAddSign<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param useLocking If `True`, updating of the var and m tensors is
-   * protected by a lock; otherwise the behavior is undefined, but may exhibit less
-   * contention.
-   */
-  public static Options useLocking(Boolean useLocking) {
-    return new Options().useLocking(useLocking);
-  }
-  
-  /**
-   * Same as "var".
-   */
-  public Output<T> out() {
-    return out;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return out;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ApplyAddSign";
-  
-  private Output<T> out;
-  
-  private ApplyAddSign(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    out = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyCenteredRmsProp.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyCenteredRmsProp.java
deleted file mode 100644
index 7150e30241b..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyCenteredRmsProp.java
+++ /dev/null
@@ -1,148 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.train;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Update '*var' according to the centered RMSProp algorithm.
- * <p>
- * The centered RMSProp algorithm uses an estimate of the centered second moment
- * (i.e., the variance) for normalization, as opposed to regular RMSProp, which
- * uses the (uncentered) second moment. This often helps with training, but is
- * slightly more expensive in terms of computation and memory.
- * <p>
- * Note that in dense implementation of this algorithm, mg, ms, and mom will
- * update even if the grad is zero, but in this sparse implementation, mg, ms,
- * and mom will not update in iterations during which the grad is zero.
- * <p>
- * mean_square = decay * mean_square + (1-decay) * gradient ** 2
- * mean_grad = decay * mean_grad + (1-decay) * gradient
- * <p>
- * Delta = learning_rate * gradient / sqrt(mean_square + epsilon - mean_grad ** 2)
- * <p>
- * mg <- rho * mg_{t-1} + (1-rho) * grad
- * ms <- rho * ms_{t-1} + (1-rho) * grad * grad
- * mom <- momentum * mom_{t-1} + lr * grad / sqrt(ms - mg * mg + epsilon)
- * var <- var - mom
- * 
- * @param <T> data type for {@code out()} output
- */
-@Operator(group = "train")
-public final class ApplyCenteredRmsProp<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.train.ApplyCenteredRmsProp}
-   */
-  public static class Options {
-    
-    /**
-     * @param useLocking If `True`, updating of the var, mg, ms, and mom tensors is
-     * protected by a lock; otherwise the behavior is undefined, but may exhibit less
-     * contention.
-     */
-    public Options useLocking(Boolean useLocking) {
-      this.useLocking = useLocking;
-      return this;
-    }
-    
-    private Boolean useLocking;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new ApplyCenteredRmsProp operation.
-   * 
-   * @param scope current scope
-   * @param var Should be from a Variable().
-   * @param mg Should be from a Variable().
-   * @param ms Should be from a Variable().
-   * @param mom Should be from a Variable().
-   * @param lr Scaling factor. Must be a scalar.
-   * @param rho Decay rate. Must be a scalar.
-   * @param momentum 
-   * @param epsilon Ridge term. Must be a scalar.
-   * @param grad The gradient.
-   * @param options carries optional attributes values
-   * @return a new instance of ApplyCenteredRmsProp
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> ApplyCenteredRmsProp<T> create(Scope scope, Operand<T> var, Operand<T> mg, Operand<T> ms, Operand<T> mom, Operand<T> lr, Operand<T> rho, Operand<T> momentum, Operand<T> epsilon, Operand<T> grad, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ApplyCenteredRMSProp", scope.makeOpName("ApplyCenteredRmsProp"));
-    opBuilder.addInput(var.asOutput(scope));
-    opBuilder.addInput(mg.asOutput(scope));
-    opBuilder.addInput(ms.asOutput(scope));
-    opBuilder.addInput(mom.asOutput(scope));
-    opBuilder.addInput(lr.asOutput(scope));
-    opBuilder.addInput(rho.asOutput(scope));
-    opBuilder.addInput(momentum.asOutput(scope));
-    opBuilder.addInput(epsilon.asOutput(scope));
-    opBuilder.addInput(grad.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.useLocking != null) {
-          opBuilder.setAttr("use_locking", opts.useLocking);
-        }
-      }
-    }
-    return new ApplyCenteredRmsProp<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param useLocking If `True`, updating of the var, mg, ms, and mom tensors is
-   * protected by a lock; otherwise the behavior is undefined, but may exhibit less
-   * contention.
-   */
-  public static Options useLocking(Boolean useLocking) {
-    return new Options().useLocking(useLocking);
-  }
-  
-  /**
-   * Same as "var".
-   */
-  public Output<T> out() {
-    return out;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return out;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ApplyCenteredRMSProp";
-  
-  private Output<T> out;
-  
-  private ApplyCenteredRmsProp(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    out = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyFtrl.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyFtrl.java
deleted file mode 100644
index b4e00bd8cc1..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyFtrl.java
+++ /dev/null
@@ -1,156 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.train;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Update '*var' according to the Ftrl-proximal scheme.
- * <p>
- * grad_with_shrinkage = grad + 2 * l2_shrinkage * var
- * accum_new = accum + grad * grad
- * linear += grad_with_shrinkage -
- *     (accum_new^(-lr_power) - accum^(-lr_power)) / lr * var
- * quadratic = 1.0 / (accum_new^(lr_power) * lr) + 2 * l2
- * var = (sign(linear) * l1 - linear) / quadratic if |linear| > l1 else 0.0
- * accum = accum_new
- * 
- * @param <T> data type for {@code out()} output
- */
-@Operator(group = "train")
-public final class ApplyFtrl<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.train.ApplyFtrl}
-   */
-  public static class Options {
-    
-    /**
-     * @param useLocking If `True`, updating of the var and accum tensors will be protected
-     * by a lock; otherwise the behavior is undefined, but may exhibit less
-     * contention.
-     */
-    public Options useLocking(Boolean useLocking) {
-      this.useLocking = useLocking;
-      return this;
-    }
-    
-    /**
-     * @param multiplyLinearByLr 
-     */
-    public Options multiplyLinearByLr(Boolean multiplyLinearByLr) {
-      this.multiplyLinearByLr = multiplyLinearByLr;
-      return this;
-    }
-    
-    private Boolean useLocking;
-    private Boolean multiplyLinearByLr;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new ApplyFtrl operation.
-   * 
-   * @param scope current scope
-   * @param var Should be from a Variable().
-   * @param accum Should be from a Variable().
-   * @param linear Should be from a Variable().
-   * @param grad The gradient.
-   * @param lr Scaling factor. Must be a scalar.
-   * @param l1 L1 regularization. Must be a scalar.
-   * @param l2 L2 shrinkage regularization. Must be a scalar.
-   * @param l2Shrinkage 
-   * @param lrPower Scaling factor. Must be a scalar.
-   * @param options carries optional attributes values
-   * @return a new instance of ApplyFtrl
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> ApplyFtrl<T> create(Scope scope, Operand<T> var, Operand<T> accum, Operand<T> linear, Operand<T> grad, Operand<T> lr, Operand<T> l1, Operand<T> l2, Operand<T> l2Shrinkage, Operand<T> lrPower, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ApplyFtrlV2", scope.makeOpName("ApplyFtrl"));
-    opBuilder.addInput(var.asOutput(scope));
-    opBuilder.addInput(accum.asOutput(scope));
-    opBuilder.addInput(linear.asOutput(scope));
-    opBuilder.addInput(grad.asOutput(scope));
-    opBuilder.addInput(lr.asOutput(scope));
-    opBuilder.addInput(l1.asOutput(scope));
-    opBuilder.addInput(l2.asOutput(scope));
-    opBuilder.addInput(l2Shrinkage.asOutput(scope));
-    opBuilder.addInput(lrPower.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.useLocking != null) {
-          opBuilder.setAttr("use_locking", opts.useLocking);
-        }
-        if (opts.multiplyLinearByLr != null) {
-          opBuilder.setAttr("multiply_linear_by_lr", opts.multiplyLinearByLr);
-        }
-      }
-    }
-    return new ApplyFtrl<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param useLocking If `True`, updating of the var and accum tensors will be protected
-   * by a lock; otherwise the behavior is undefined, but may exhibit less
-   * contention.
-   */
-  public static Options useLocking(Boolean useLocking) {
-    return new Options().useLocking(useLocking);
-  }
-  
-  /**
-   * @param multiplyLinearByLr 
-   */
-  public static Options multiplyLinearByLr(Boolean multiplyLinearByLr) {
-    return new Options().multiplyLinearByLr(multiplyLinearByLr);
-  }
-  
-  /**
-   * Same as "var".
-   */
-  public Output<T> out() {
-    return out;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return out;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ApplyFtrlV2";
-  
-  private Output<T> out;
-  
-  private ApplyFtrl(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    out = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyGradientDescent.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyGradientDescent.java
deleted file mode 100644
index 450adff5136..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyGradientDescent.java
+++ /dev/null
@@ -1,115 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.train;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Update '*var' by subtracting 'alpha' * 'delta' from it.
- * 
- * @param <T> data type for {@code out()} output
- */
-@Operator(group = "train")
-public final class ApplyGradientDescent<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.train.ApplyGradientDescent}
-   */
-  public static class Options {
-    
-    /**
-     * @param useLocking If `True`, the subtraction will be protected by a lock;
-     * otherwise the behavior is undefined, but may exhibit less contention.
-     */
-    public Options useLocking(Boolean useLocking) {
-      this.useLocking = useLocking;
-      return this;
-    }
-    
-    private Boolean useLocking;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new ApplyGradientDescent operation.
-   * 
-   * @param scope current scope
-   * @param var Should be from a Variable().
-   * @param alpha Scaling factor. Must be a scalar.
-   * @param delta The change.
-   * @param options carries optional attributes values
-   * @return a new instance of ApplyGradientDescent
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> ApplyGradientDescent<T> create(Scope scope, Operand<T> var, Operand<T> alpha, Operand<T> delta, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ApplyGradientDescent", scope.makeOpName("ApplyGradientDescent"));
-    opBuilder.addInput(var.asOutput(scope));
-    opBuilder.addInput(alpha.asOutput(scope));
-    opBuilder.addInput(delta.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.useLocking != null) {
-          opBuilder.setAttr("use_locking", opts.useLocking);
-        }
-      }
-    }
-    return new ApplyGradientDescent<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param useLocking If `True`, the subtraction will be protected by a lock;
-   * otherwise the behavior is undefined, but may exhibit less contention.
-   */
-  public static Options useLocking(Boolean useLocking) {
-    return new Options().useLocking(useLocking);
-  }
-  
-  /**
-   * Same as "var".
-   */
-  public Output<T> out() {
-    return out;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return out;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ApplyGradientDescent";
-  
-  private Output<T> out;
-  
-  private ApplyGradientDescent(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    out = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyMomentum.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyMomentum.java
deleted file mode 100644
index a3a4e1cde3c..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyMomentum.java
+++ /dev/null
@@ -1,149 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.train;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Update '*var' according to the momentum scheme.
- * <p>
- * Set use_nesterov = True if you want to use Nesterov momentum.
- * <p>
- * accum = accum * momentum + grad
- * var -= lr * accum
- * 
- * @param <T> data type for {@code out()} output
- */
-@Operator(group = "train")
-public final class ApplyMomentum<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.train.ApplyMomentum}
-   */
-  public static class Options {
-    
-    /**
-     * @param useLocking If `True`, updating of the var and accum tensors will be protected
-     * by a lock; otherwise the behavior is undefined, but may exhibit less
-     * contention.
-     */
-    public Options useLocking(Boolean useLocking) {
-      this.useLocking = useLocking;
-      return this;
-    }
-    
-    /**
-     * @param useNesterov If `True`, the tensor passed to compute grad will be
-     * var - lr * momentum * accum, so in the end, the var you get is actually
-     * var - lr * momentum * accum.
-     */
-    public Options useNesterov(Boolean useNesterov) {
-      this.useNesterov = useNesterov;
-      return this;
-    }
-    
-    private Boolean useLocking;
-    private Boolean useNesterov;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new ApplyMomentum operation.
-   * 
-   * @param scope current scope
-   * @param var Should be from a Variable().
-   * @param accum Should be from a Variable().
-   * @param lr Scaling factor. Must be a scalar.
-   * @param grad The gradient.
-   * @param momentum Momentum. Must be a scalar.
-   * @param options carries optional attributes values
-   * @return a new instance of ApplyMomentum
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> ApplyMomentum<T> create(Scope scope, Operand<T> var, Operand<T> accum, Operand<T> lr, Operand<T> grad, Operand<T> momentum, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ApplyMomentum", scope.makeOpName("ApplyMomentum"));
-    opBuilder.addInput(var.asOutput(scope));
-    opBuilder.addInput(accum.asOutput(scope));
-    opBuilder.addInput(lr.asOutput(scope));
-    opBuilder.addInput(grad.asOutput(scope));
-    opBuilder.addInput(momentum.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.useLocking != null) {
-          opBuilder.setAttr("use_locking", opts.useLocking);
-        }
-        if (opts.useNesterov != null) {
-          opBuilder.setAttr("use_nesterov", opts.useNesterov);
-        }
-      }
-    }
-    return new ApplyMomentum<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param useLocking If `True`, updating of the var and accum tensors will be protected
-   * by a lock; otherwise the behavior is undefined, but may exhibit less
-   * contention.
-   */
-  public static Options useLocking(Boolean useLocking) {
-    return new Options().useLocking(useLocking);
-  }
-  
-  /**
-   * @param useNesterov If `True`, the tensor passed to compute grad will be
-   * var - lr * momentum * accum, so in the end, the var you get is actually
-   * var - lr * momentum * accum.
-   */
-  public static Options useNesterov(Boolean useNesterov) {
-    return new Options().useNesterov(useNesterov);
-  }
-  
-  /**
-   * Same as "var".
-   */
-  public Output<T> out() {
-    return out;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return out;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ApplyMomentum";
-  
-  private Output<T> out;
-  
-  private ApplyMomentum(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    out = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyPowerSign.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyPowerSign.java
deleted file mode 100644
index e13fd58022d..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyPowerSign.java
+++ /dev/null
@@ -1,129 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.train;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Update '*var' according to the AddSign update.
- * <p>
- * m_t <- beta1 * m_{t-1} + (1 - beta1) * g
- * update <- exp(logbase * sign_decay * sign(g) * sign(m_t)) * g
- * variable <- variable - lr_t * update
- * 
- * @param <T> data type for {@code out()} output
- */
-@Operator(group = "train")
-public final class ApplyPowerSign<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.train.ApplyPowerSign}
-   */
-  public static class Options {
-    
-    /**
-     * @param useLocking If `True`, updating of the var and m tensors is
-     * protected by a lock; otherwise the behavior is undefined, but may exhibit less
-     * contention.
-     */
-    public Options useLocking(Boolean useLocking) {
-      this.useLocking = useLocking;
-      return this;
-    }
-    
-    private Boolean useLocking;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new ApplyPowerSign operation.
-   * 
-   * @param scope current scope
-   * @param var Should be from a Variable().
-   * @param m Should be from a Variable().
-   * @param lr Scaling factor. Must be a scalar.
-   * @param logbase Must be a scalar.
-   * @param signDecay Must be a scalar.
-   * @param beta Must be a scalar.
-   * @param grad The gradient.
-   * @param options carries optional attributes values
-   * @return a new instance of ApplyPowerSign
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> ApplyPowerSign<T> create(Scope scope, Operand<T> var, Operand<T> m, Operand<T> lr, Operand<T> logbase, Operand<T> signDecay, Operand<T> beta, Operand<T> grad, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ApplyPowerSign", scope.makeOpName("ApplyPowerSign"));
-    opBuilder.addInput(var.asOutput(scope));
-    opBuilder.addInput(m.asOutput(scope));
-    opBuilder.addInput(lr.asOutput(scope));
-    opBuilder.addInput(logbase.asOutput(scope));
-    opBuilder.addInput(signDecay.asOutput(scope));
-    opBuilder.addInput(beta.asOutput(scope));
-    opBuilder.addInput(grad.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.useLocking != null) {
-          opBuilder.setAttr("use_locking", opts.useLocking);
-        }
-      }
-    }
-    return new ApplyPowerSign<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param useLocking If `True`, updating of the var and m tensors is
-   * protected by a lock; otherwise the behavior is undefined, but may exhibit less
-   * contention.
-   */
-  public static Options useLocking(Boolean useLocking) {
-    return new Options().useLocking(useLocking);
-  }
-  
-  /**
-   * Same as "var".
-   */
-  public Output<T> out() {
-    return out;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return out;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ApplyPowerSign";
-  
-  private Output<T> out;
-  
-  private ApplyPowerSign(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    out = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyProximalAdagrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyProximalAdagrad.java
deleted file mode 100644
index ec8e57b7c5a..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyProximalAdagrad.java
+++ /dev/null
@@ -1,125 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.train;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Update '*var' and '*accum' according to FOBOS with Adagrad learning rate.
- * <p>
- * accum += grad <i> grad
- * prox_v = var - lr </i> grad <i> (1 / sqrt(accum))
- * var = sign(prox_v)/(1+lr</i>l2) <i> max{|prox_v|-lr</i>l1,0}
- * 
- * @param <T> data type for {@code out()} output
- */
-@Operator(group = "train")
-public final class ApplyProximalAdagrad<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.train.ApplyProximalAdagrad}
-   */
-  public static class Options {
-    
-    /**
-     * @param useLocking If True, updating of the var and accum tensors will be protected by
-     * a lock; otherwise the behavior is undefined, but may exhibit less contention.
-     */
-    public Options useLocking(Boolean useLocking) {
-      this.useLocking = useLocking;
-      return this;
-    }
-    
-    private Boolean useLocking;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new ApplyProximalAdagrad operation.
-   * 
-   * @param scope current scope
-   * @param var Should be from a Variable().
-   * @param accum Should be from a Variable().
-   * @param lr Scaling factor. Must be a scalar.
-   * @param l1 L1 regularization. Must be a scalar.
-   * @param l2 L2 regularization. Must be a scalar.
-   * @param grad The gradient.
-   * @param options carries optional attributes values
-   * @return a new instance of ApplyProximalAdagrad
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> ApplyProximalAdagrad<T> create(Scope scope, Operand<T> var, Operand<T> accum, Operand<T> lr, Operand<T> l1, Operand<T> l2, Operand<T> grad, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ApplyProximalAdagrad", scope.makeOpName("ApplyProximalAdagrad"));
-    opBuilder.addInput(var.asOutput(scope));
-    opBuilder.addInput(accum.asOutput(scope));
-    opBuilder.addInput(lr.asOutput(scope));
-    opBuilder.addInput(l1.asOutput(scope));
-    opBuilder.addInput(l2.asOutput(scope));
-    opBuilder.addInput(grad.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.useLocking != null) {
-          opBuilder.setAttr("use_locking", opts.useLocking);
-        }
-      }
-    }
-    return new ApplyProximalAdagrad<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param useLocking If True, updating of the var and accum tensors will be protected by
-   * a lock; otherwise the behavior is undefined, but may exhibit less contention.
-   */
-  public static Options useLocking(Boolean useLocking) {
-    return new Options().useLocking(useLocking);
-  }
-  
-  /**
-   * Same as "var".
-   */
-  public Output<T> out() {
-    return out;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return out;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ApplyProximalAdagrad";
-  
-  private Output<T> out;
-  
-  private ApplyProximalAdagrad(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    out = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyProximalGradientDescent.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyProximalGradientDescent.java
deleted file mode 100644
index 8ea81694471..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyProximalGradientDescent.java
+++ /dev/null
@@ -1,122 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.train;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Update '*var' as FOBOS algorithm with fixed learning rate.
- * <p>
- * prox_v = var - alpha <i> delta
- * var = sign(prox_v)/(1+alpha</i>l2) <i> max{|prox_v|-alpha</i>l1,0}
- * 
- * @param <T> data type for {@code out()} output
- */
-@Operator(group = "train")
-public final class ApplyProximalGradientDescent<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.train.ApplyProximalGradientDescent}
-   */
-  public static class Options {
-    
-    /**
-     * @param useLocking If True, the subtraction will be protected by a lock;
-     * otherwise the behavior is undefined, but may exhibit less contention.
-     */
-    public Options useLocking(Boolean useLocking) {
-      this.useLocking = useLocking;
-      return this;
-    }
-    
-    private Boolean useLocking;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new ApplyProximalGradientDescent operation.
-   * 
-   * @param scope current scope
-   * @param var Should be from a Variable().
-   * @param alpha Scaling factor. Must be a scalar.
-   * @param l1 L1 regularization. Must be a scalar.
-   * @param l2 L2 regularization. Must be a scalar.
-   * @param delta The change.
-   * @param options carries optional attributes values
-   * @return a new instance of ApplyProximalGradientDescent
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> ApplyProximalGradientDescent<T> create(Scope scope, Operand<T> var, Operand<T> alpha, Operand<T> l1, Operand<T> l2, Operand<T> delta, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ApplyProximalGradientDescent", scope.makeOpName("ApplyProximalGradientDescent"));
-    opBuilder.addInput(var.asOutput(scope));
-    opBuilder.addInput(alpha.asOutput(scope));
-    opBuilder.addInput(l1.asOutput(scope));
-    opBuilder.addInput(l2.asOutput(scope));
-    opBuilder.addInput(delta.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.useLocking != null) {
-          opBuilder.setAttr("use_locking", opts.useLocking);
-        }
-      }
-    }
-    return new ApplyProximalGradientDescent<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param useLocking If True, the subtraction will be protected by a lock;
-   * otherwise the behavior is undefined, but may exhibit less contention.
-   */
-  public static Options useLocking(Boolean useLocking) {
-    return new Options().useLocking(useLocking);
-  }
-  
-  /**
-   * Same as "var".
-   */
-  public Output<T> out() {
-    return out;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return out;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ApplyProximalGradientDescent";
-  
-  private Output<T> out;
-  
-  private ApplyProximalGradientDescent(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    out = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyRmsProp.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyRmsProp.java
deleted file mode 100644
index 9d5e1e4e449..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyRmsProp.java
+++ /dev/null
@@ -1,138 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.train;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Update '*var' according to the RMSProp algorithm.
- * <p>
- * Note that in dense implementation of this algorithm, ms and mom will
- * update even if the grad is zero, but in this sparse implementation, ms
- * and mom will not update in iterations during which the grad is zero.
- * <p>
- * mean_square = decay * mean_square + (1-decay) * gradient ** 2
- * Delta = learning_rate * gradient / sqrt(mean_square + epsilon)
- * <p>
- * ms <- rho * ms_{t-1} + (1-rho) * grad * grad
- * mom <- momentum * mom_{t-1} + lr * grad / sqrt(ms + epsilon)
- * var <- var - mom
- * 
- * @param <T> data type for {@code out()} output
- */
-@Operator(group = "train")
-public final class ApplyRmsProp<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.train.ApplyRmsProp}
-   */
-  public static class Options {
-    
-    /**
-     * @param useLocking If `True`, updating of the var, ms, and mom tensors is protected
-     * by a lock; otherwise the behavior is undefined, but may exhibit less
-     * contention.
-     */
-    public Options useLocking(Boolean useLocking) {
-      this.useLocking = useLocking;
-      return this;
-    }
-    
-    private Boolean useLocking;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new ApplyRmsProp operation.
-   * 
-   * @param scope current scope
-   * @param var Should be from a Variable().
-   * @param ms Should be from a Variable().
-   * @param mom Should be from a Variable().
-   * @param lr Scaling factor. Must be a scalar.
-   * @param rho Decay rate. Must be a scalar.
-   * @param momentum 
-   * @param epsilon Ridge term. Must be a scalar.
-   * @param grad The gradient.
-   * @param options carries optional attributes values
-   * @return a new instance of ApplyRmsProp
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> ApplyRmsProp<T> create(Scope scope, Operand<T> var, Operand<T> ms, Operand<T> mom, Operand<T> lr, Operand<T> rho, Operand<T> momentum, Operand<T> epsilon, Operand<T> grad, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ApplyRMSProp", scope.makeOpName("ApplyRmsProp"));
-    opBuilder.addInput(var.asOutput(scope));
-    opBuilder.addInput(ms.asOutput(scope));
-    opBuilder.addInput(mom.asOutput(scope));
-    opBuilder.addInput(lr.asOutput(scope));
-    opBuilder.addInput(rho.asOutput(scope));
-    opBuilder.addInput(momentum.asOutput(scope));
-    opBuilder.addInput(epsilon.asOutput(scope));
-    opBuilder.addInput(grad.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.useLocking != null) {
-          opBuilder.setAttr("use_locking", opts.useLocking);
-        }
-      }
-    }
-    return new ApplyRmsProp<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param useLocking If `True`, updating of the var, ms, and mom tensors is protected
-   * by a lock; otherwise the behavior is undefined, but may exhibit less
-   * contention.
-   */
-  public static Options useLocking(Boolean useLocking) {
-    return new Options().useLocking(useLocking);
-  }
-  
-  /**
-   * Same as "var".
-   */
-  public Output<T> out() {
-    return out;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return out;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ApplyRMSProp";
-  
-  private Output<T> out;
-  
-  private ApplyRmsProp(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    out = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/BatchMatMul.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/BatchMatMul.java
deleted file mode 100644
index b362fe5447a..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/BatchMatMul.java
+++ /dev/null
@@ -1,154 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.train;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Multiplies slices of two tensors in batches.
- * <p>
- * Multiplies all slices of `Tensor` `x` and `y` (each slice can be
- * viewed as an element of a batch), and arranges the individual results
- * in a single output tensor of the same batch size. Each of the
- * individual slices can optionally be adjointed (to adjoint a matrix
- * means to transpose and conjugate it) before multiplication by setting
- * the `adj_x` or `adj_y` flag to `True`, which are by default `False`.
- * <p>
- * The input tensors `x` and `y` are 2-D or higher with shape `[..., r_x, c_x]`
- * and `[..., r_y, c_y]`.
- * <p>
- * The output tensor is 2-D or higher with shape `[..., r_o, c_o]`, where:
- * <p>
- *     r_o = c_x if adj_x else r_x
- *     c_o = r_y if adj_y else c_y
- * <p>
- * It is computed as:
- * <p>
- *     output[..., :, :] = matrix(x[..., :, :]) * matrix(y[..., :, :])
- * <p>
- * <i>NOTE</i>: `train.BatchMatMul` supports broadcasting in the batch dimensions. More
- * about broadcasting
- * [here](http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html).
- * 
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator(group = "train")
-public final class BatchMatMul<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.train.BatchMatMul}
-   */
-  public static class Options {
-    
-    /**
-     * @param adjX If `True`, adjoint the slices of `x`. Defaults to `False`.
-     */
-    public Options adjX(Boolean adjX) {
-      this.adjX = adjX;
-      return this;
-    }
-    
-    /**
-     * @param adjY If `True`, adjoint the slices of `y`. Defaults to `False`.
-     */
-    public Options adjY(Boolean adjY) {
-      this.adjY = adjY;
-      return this;
-    }
-    
-    private Boolean adjX;
-    private Boolean adjY;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new BatchMatMul operation.
-   * 
-   * @param scope current scope
-   * @param x 2-D or higher with shape `[..., r_x, c_x]`.
-   * @param y 2-D or higher with shape `[..., r_y, c_y]`.
-   * @param options carries optional attributes values
-   * @return a new instance of BatchMatMul
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> BatchMatMul<T> create(Scope scope, Operand<T> x, Operand<T> y, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("BatchMatMulV2", scope.makeOpName("BatchMatMul"));
-    opBuilder.addInput(x.asOutput(scope));
-    opBuilder.addInput(y.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.adjX != null) {
-          opBuilder.setAttr("adj_x", opts.adjX);
-        }
-        if (opts.adjY != null) {
-          opBuilder.setAttr("adj_y", opts.adjY);
-        }
-      }
-    }
-    return new BatchMatMul<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param adjX If `True`, adjoint the slices of `x`. Defaults to `False`.
-   */
-  public static Options adjX(Boolean adjX) {
-    return new Options().adjX(adjX);
-  }
-  
-  /**
-   * @param adjY If `True`, adjoint the slices of `y`. Defaults to `False`.
-   */
-  public static Options adjY(Boolean adjY) {
-    return new Options().adjY(adjY);
-  }
-  
-  /**
-   * 3-D or higher with shape `[..., r_o, c_o]`
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "BatchMatMulV2";
-  
-  private Output<T> output;
-  
-  private BatchMatMul(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ConditionalAccumulator.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ConditionalAccumulator.java
deleted file mode 100644
index a3eeb3d0606..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ConditionalAccumulator.java
+++ /dev/null
@@ -1,160 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.train;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TString;
-import org.tensorflow.types.family.TType;
-
-/**
- * A conditional accumulator for aggregating gradients.
- * <p>
- * The accumulator accepts gradients marked with local_step greater or
- * equal to the most recent global_step known to the accumulator. The
- * average can be extracted from the accumulator, provided sufficient
- * gradients have been accumulated. Extracting the average automatically
- * resets the aggregate to 0, and increments the global_step recorded by
- * the accumulator.
- */
-@Operator(group = "train")
-public final class ConditionalAccumulator extends RawOp implements Operand<TString> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.train.ConditionalAccumulator}
-   */
-  public static class Options {
-    
-    /**
-     * @param container If non-empty, this accumulator is placed in the given container.
-     * Otherwise, a default container is used.
-     */
-    public Options container(String container) {
-      this.container = container;
-      return this;
-    }
-    
-    /**
-     * @param sharedName If non-empty, this accumulator will be shared under the
-     * given name across multiple sessions.
-     */
-    public Options sharedName(String sharedName) {
-      this.sharedName = sharedName;
-      return this;
-    }
-    
-    /**
-     * @param reductionType 
-     */
-    public Options reductionType(String reductionType) {
-      this.reductionType = reductionType;
-      return this;
-    }
-    
-    private String container;
-    private String sharedName;
-    private String reductionType;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new ConditionalAccumulator operation.
-   * 
-   * @param scope current scope
-   * @param dtype The type of the value being accumulated.
-   * @param shape The shape of the values, can be [], in which case shape is unknown.
-   * @param options carries optional attributes values
-   * @return a new instance of ConditionalAccumulator
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> ConditionalAccumulator create(Scope scope, Class<T> dtype, Shape shape, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ConditionalAccumulator", scope.makeOpName("ConditionalAccumulator"));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("dtype", dtype);
-    opBuilder.setAttr("shape", shape);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.container != null) {
-          opBuilder.setAttr("container", opts.container);
-        }
-        if (opts.sharedName != null) {
-          opBuilder.setAttr("shared_name", opts.sharedName);
-        }
-        if (opts.reductionType != null) {
-          opBuilder.setAttr("reduction_type", opts.reductionType);
-        }
-      }
-    }
-    return new ConditionalAccumulator(opBuilder.build());
-  }
-  
-  /**
-   * @param container If non-empty, this accumulator is placed in the given container.
-   * Otherwise, a default container is used.
-   */
-  public static Options container(String container) {
-    return new Options().container(container);
-  }
-  
-  /**
-   * @param sharedName If non-empty, this accumulator will be shared under the
-   * given name across multiple sessions.
-   */
-  public static Options sharedName(String sharedName) {
-    return new Options().sharedName(sharedName);
-  }
-  
-  /**
-   * @param reductionType 
-   */
-  public static Options reductionType(String reductionType) {
-    return new Options().reductionType(reductionType);
-  }
-  
-  /**
-   * The handle to the accumulator.
-   */
-  public Output<TString> handle() {
-    return handle;
-  }
-  
-  @Override
-  public Output<TString> asOutput(Scope scope) {
-    return handle;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ConditionalAccumulator";
-  
-  private Output<TString> handle;
-  
-  private ConditionalAccumulator(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    handle = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/GenerateVocabRemapping.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/GenerateVocabRemapping.java
deleted file mode 100644
index 7204dc32b0e..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/GenerateVocabRemapping.java
+++ /dev/null
@@ -1,152 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.train;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.TString;
-
-/**
- * Given a path to new and old vocabulary files, returns a remapping Tensor of
- * <p>
- * length `num_new_vocab`, where `remapping[i]` contains the row number in the old
- * vocabulary that corresponds to row `i` in the new vocabulary (starting at line
- * `new_vocab_offset` and up to `num_new_vocab` entities), or `-1` if entry `i`
- * in the new vocabulary is not in the old vocabulary.  The old vocabulary is
- * constrained to the first `old_vocab_size` entries if `old_vocab_size` is not the
- * default value of -1.
- * <p>
- * `num_vocab_offset` enables
- * use in the partitioned variable case, and should generally be set through
- * examining partitioning info.  The format of the files should be a text file,
- * with each line containing a single entity within the vocabulary.
- * <p>
- * For example, with `new_vocab_file` a text file containing each of the following
- * elements on a single line: `[f0, f1, f2, f3]`, old_vocab_file = [f1, f0, f3],
- * `num_new_vocab = 3, new_vocab_offset = 1`, the returned remapping would be
- * `[0, -1, 2]`.
- * <p>
- * The op also returns a count of how many entries in the new vocabulary
- * were present in the old vocabulary, which is used to calculate the number of
- * values to initialize in a weight matrix remapping
- * <p>
- * This functionality can be used to remap both row vocabularies (typically,
- * features) and column vocabularies (typically, classes) from TensorFlow
- * checkpoints.  Note that the partitioning logic relies on contiguous vocabularies
- * corresponding to div-partitioned variables.  Moreover, the underlying remapping
- * uses an IndexTable (as opposed to an inexact CuckooTable), so client code should
- * use the corresponding index_table_from_file() as the FeatureColumn framework
- * does (as opposed to tf.feature_to_id(), which uses a CuckooTable).
- */
-@Operator(group = "train")
-public final class GenerateVocabRemapping extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.train.GenerateVocabRemapping}
-   */
-  public static class Options {
-    
-    /**
-     * @param oldVocabSize Number of entries in the old vocab file to consider.  If -1,
-     * use the entire old vocabulary.
-     */
-    public Options oldVocabSize(Long oldVocabSize) {
-      this.oldVocabSize = oldVocabSize;
-      return this;
-    }
-    
-    private Long oldVocabSize;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new GenerateVocabRemapping operation.
-   * 
-   * @param scope current scope
-   * @param newVocabFile Path to the new vocab file.
-   * @param oldVocabFile Path to the old vocab file.
-   * @param newVocabOffset How many entries into the new vocab file to start reading.
-   * @param numNewVocab Number of entries in the new vocab file to remap.
-   * @param options carries optional attributes values
-   * @return a new instance of GenerateVocabRemapping
-   */
-  @Endpoint(describeByClass = true)
-  public static GenerateVocabRemapping create(Scope scope, Operand<TString> newVocabFile, Operand<TString> oldVocabFile, Long newVocabOffset, Long numNewVocab, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("GenerateVocabRemapping", scope.makeOpName("GenerateVocabRemapping"));
-    opBuilder.addInput(newVocabFile.asOutput(scope));
-    opBuilder.addInput(oldVocabFile.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("new_vocab_offset", newVocabOffset);
-    opBuilder.setAttr("num_new_vocab", numNewVocab);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.oldVocabSize != null) {
-          opBuilder.setAttr("old_vocab_size", opts.oldVocabSize);
-        }
-      }
-    }
-    return new GenerateVocabRemapping(opBuilder.build());
-  }
-  
-  /**
-   * @param oldVocabSize Number of entries in the old vocab file to consider.  If -1,
-   * use the entire old vocabulary.
-   */
-  public static Options oldVocabSize(Long oldVocabSize) {
-    return new Options().oldVocabSize(oldVocabSize);
-  }
-  
-  /**
-   * A Tensor of length num_new_vocab where the element at index i
-   * is equal to the old ID that maps to the new ID i.  This element is -1 for any
-   * new ID that is not found in the old vocabulary.
-   */
-  public Output<TInt64> remapping() {
-    return remapping;
-  }
-  
-  /**
-   * Number of new vocab entries found in old vocab.
-   */
-  public Output<TInt32> numPresent() {
-    return numPresent;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "GenerateVocabRemapping";
-  
-  private Output<TInt64> remapping;
-  private Output<TInt32> numPresent;
-  
-  private GenerateVocabRemapping(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    remapping = operation.output(outputIdx++);
-    numPresent = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/MergeV2Checkpoints.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/MergeV2Checkpoints.java
deleted file mode 100644
index dbd84af2f49..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/MergeV2Checkpoints.java
+++ /dev/null
@@ -1,102 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.train;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TString;
-
-/**
- * V2 format specific: merges the metadata files of sharded checkpoints.  The
- * <p>
- * result is one logical checkpoint, with one physical metadata file and renamed
- * data files.
- * <p>
- * Intended for "grouping" multiple checkpoints in a sharded checkpoint setup.
- * <p>
- * If delete_old_dirs is true, attempts to delete recursively the dirname of each
- * path in the input checkpoint_prefixes.  This is useful when those paths are non
- * user-facing temporary locations.
- */
-@Operator(group = "train")
-public final class MergeV2Checkpoints extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.train.MergeV2Checkpoints}
-   */
-  public static class Options {
-    
-    /**
-     * @param deleteOldDirs see above.
-     */
-    public Options deleteOldDirs(Boolean deleteOldDirs) {
-      this.deleteOldDirs = deleteOldDirs;
-      return this;
-    }
-    
-    private Boolean deleteOldDirs;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new MergeV2Checkpoints operation.
-   * 
-   * @param scope current scope
-   * @param checkpointPrefixes prefixes of V2 checkpoints to merge.
-   * @param destinationPrefix scalar.  The desired final prefix.  Allowed to be the same
-   * as one of the checkpoint_prefixes.
-   * @param options carries optional attributes values
-   * @return a new instance of MergeV2Checkpoints
-   */
-  @Endpoint(describeByClass = true)
-  public static MergeV2Checkpoints create(Scope scope, Operand<TString> checkpointPrefixes, Operand<TString> destinationPrefix, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("MergeV2Checkpoints", scope.makeOpName("MergeV2Checkpoints"));
-    opBuilder.addInput(checkpointPrefixes.asOutput(scope));
-    opBuilder.addInput(destinationPrefix.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.deleteOldDirs != null) {
-          opBuilder.setAttr("delete_old_dirs", opts.deleteOldDirs);
-        }
-      }
-    }
-    return new MergeV2Checkpoints(opBuilder.build());
-  }
-  
-  /**
-   * @param deleteOldDirs see above.
-   */
-  public static Options deleteOldDirs(Boolean deleteOldDirs) {
-    return new Options().deleteOldDirs(deleteOldDirs);
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "MergeV2Checkpoints";
-  
-  private MergeV2Checkpoints(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/NegTrain.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/NegTrain.java
deleted file mode 100644
index fc0cc638422..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/NegTrain.java
+++ /dev/null
@@ -1,74 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.train;
-
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.TInt32;
-
-/**
- * Training via negative sampling.
- */
-@Operator(group = "train")
-public final class NegTrain extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new NegTrain operation.
-   * 
-   * @param scope current scope
-   * @param wIn input word embedding.
-   * @param wOut output word embedding.
-   * @param examples A vector of word ids.
-   * @param labels A vector of word ids.
-   * @param lr 
-   * @param vocabCount Count of words in the vocabulary.
-   * @param numNegativeSamples Number of negative samples per example.
-   * @return a new instance of NegTrain
-   */
-  @Endpoint(describeByClass = true)
-  public static NegTrain create(Scope scope, Operand<TFloat32> wIn, Operand<TFloat32> wOut, Operand<TInt32> examples, Operand<TInt32> labels, Operand<TFloat32> lr, List<Long> vocabCount, Long numNegativeSamples) {
-    OperationBuilder opBuilder = scope.env().opBuilder("NegTrain", scope.makeOpName("NegTrain"));
-    opBuilder.addInput(wIn.asOutput(scope));
-    opBuilder.addInput(wOut.asOutput(scope));
-    opBuilder.addInput(examples.asOutput(scope));
-    opBuilder.addInput(labels.asOutput(scope));
-    opBuilder.addInput(lr.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    long[] vocabCountArray = new long[vocabCount.size()];
-    for (int i = 0; i < vocabCountArray.length; ++i) {
-      vocabCountArray[i] = vocabCount.get(i);
-    }
-    opBuilder.setAttr("vocab_count", vocabCountArray);
-    opBuilder.setAttr("num_negative_samples", numNegativeSamples);
-    return new NegTrain(opBuilder.build());
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "NegTrain";
-  
-  private NegTrain(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/PreventGradient.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/PreventGradient.java
deleted file mode 100644
index 46f03708cce..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/PreventGradient.java
+++ /dev/null
@@ -1,119 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.train;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * An identity op that triggers an error if a gradient is requested.
- * <p>
- * When executed in a graph, this op outputs its input tensor as-is.
- * <p>
- * When building ops to compute gradients, the TensorFlow gradient system
- * will return an error when trying to lookup the gradient of this op,
- * because no gradient must ever be registered for this function.  This
- * op exists to prevent subtle bugs from silently returning unimplemented
- * gradients in some corner cases.
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator(group = "train")
-public final class PreventGradient<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.train.PreventGradient}
-   */
-  public static class Options {
-    
-    /**
-     * @param message Will be printed in the error when anyone tries to differentiate
-     * this operation.
-     */
-    public Options message(String message) {
-      this.message = message;
-      return this;
-    }
-    
-    private String message;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new PreventGradient operation.
-   * 
-   * @param scope current scope
-   * @param input any tensor.
-   * @param options carries optional attributes values
-   * @return a new instance of PreventGradient
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> PreventGradient<T> create(Scope scope, Operand<T> input, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("PreventGradient", scope.makeOpName("PreventGradient"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.message != null) {
-          opBuilder.setAttr("message", opts.message);
-        }
-      }
-    }
-    return new PreventGradient<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param message Will be printed in the error when anyone tries to differentiate
-   * this operation.
-   */
-  public static Options message(String message) {
-    return new Options().message(message);
-  }
-  
-  /**
-   * the same input tensor.
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "PreventGradient";
-  
-  private Output<T> output;
-  
-  private PreventGradient(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceAccumulatorApplyGradient.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceAccumulatorApplyGradient.java
deleted file mode 100644
index 035969ce383..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceAccumulatorApplyGradient.java
+++ /dev/null
@@ -1,62 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.train;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
-
-/**
- * Applies a gradient to a given accumulator.
- * <p>
- * Does not add if local_step is lesser than the accumulator's global_step.
- */
-public final class ResourceAccumulatorApplyGradient extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new ResourceAccumulatorApplyGradient operation.
-   * 
-   * @param scope current scope
-   * @param handle The handle to a accumulator.
-   * @param localStep The local_step value at which the gradient was computed.
-   * @param gradient A tensor of the gradient to be accumulated.
-   * @return a new instance of ResourceAccumulatorApplyGradient
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> ResourceAccumulatorApplyGradient create(Scope scope, Operand<?> handle, Operand<TInt64> localStep, Operand<T> gradient) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ResourceAccumulatorApplyGradient", scope.makeOpName("ResourceAccumulatorApplyGradient"));
-    opBuilder.addInput(handle.asOutput(scope));
-    opBuilder.addInput(localStep.asOutput(scope));
-    opBuilder.addInput(gradient.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new ResourceAccumulatorApplyGradient(opBuilder.build());
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ResourceAccumulatorApplyGradient";
-  
-  private ResourceAccumulatorApplyGradient(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceAccumulatorNumAccumulated.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceAccumulatorNumAccumulated.java
deleted file mode 100644
index 41b4ddbd718..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceAccumulatorNumAccumulated.java
+++ /dev/null
@@ -1,72 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.train;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
-
-/**
- * Returns the number of gradients aggregated in the given accumulators.
- */
-public final class ResourceAccumulatorNumAccumulated extends RawOp implements Operand<TInt32> {
-  
-  /**
-   * Factory method to create a class wrapping a new ResourceAccumulatorNumAccumulated operation.
-   * 
-   * @param scope current scope
-   * @param handle The handle to an accumulator.
-   * @return a new instance of ResourceAccumulatorNumAccumulated
-   */
-  @Endpoint(describeByClass = true)
-  public static ResourceAccumulatorNumAccumulated create(Scope scope, Operand<?> handle) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ResourceAccumulatorNumAccumulated", scope.makeOpName("ResourceAccumulatorNumAccumulated"));
-    opBuilder.addInput(handle.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new ResourceAccumulatorNumAccumulated(opBuilder.build());
-  }
-  
-  /**
-   * The number of gradients aggregated in the given accumulator.
-   */
-  public Output<TInt32> numAccumulated() {
-    return numAccumulated;
-  }
-  
-  @Override
-  public Output<TInt32> asOutput(Scope scope) {
-    return numAccumulated;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ResourceAccumulatorNumAccumulated";
-  
-  private Output<TInt32> numAccumulated;
-  
-  private ResourceAccumulatorNumAccumulated(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    numAccumulated = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceAccumulatorSetGlobalStep.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceAccumulatorSetGlobalStep.java
deleted file mode 100644
index 46c8f2c9108..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceAccumulatorSetGlobalStep.java
+++ /dev/null
@@ -1,60 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.train;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt64;
-
-/**
- * Updates the accumulator with a new value for global_step.
- * <p>
- * Logs warning if the accumulator's value is already higher than
- * new_global_step.
- */
-public final class ResourceAccumulatorSetGlobalStep extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new ResourceAccumulatorSetGlobalStep operation.
-   * 
-   * @param scope current scope
-   * @param handle The handle to an accumulator.
-   * @param newGlobalStep The new global_step value to set.
-   * @return a new instance of ResourceAccumulatorSetGlobalStep
-   */
-  @Endpoint(describeByClass = true)
-  public static ResourceAccumulatorSetGlobalStep create(Scope scope, Operand<?> handle, Operand<TInt64> newGlobalStep) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ResourceAccumulatorSetGlobalStep", scope.makeOpName("ResourceAccumulatorSetGlobalStep"));
-    opBuilder.addInput(handle.asOutput(scope));
-    opBuilder.addInput(newGlobalStep.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new ResourceAccumulatorSetGlobalStep(opBuilder.build());
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ResourceAccumulatorSetGlobalStep";
-  
-  private ResourceAccumulatorSetGlobalStep(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceAccumulatorTakeGradient.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceAccumulatorTakeGradient.java
deleted file mode 100644
index 5a8f85ee16a..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceAccumulatorTakeGradient.java
+++ /dev/null
@@ -1,86 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.train;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.family.TType;
-
-/**
- * Extracts the average gradient in the given ConditionalAccumulator.
- * <p>
- * The op blocks until sufficient (i.e., more than num_required)
- * gradients have been accumulated.  If the accumulator has already
- * aggregated more than num_required gradients, it returns the average of
- * the accumulated gradients.  Also automatically increments the recorded
- * global_step in the accumulator by 1, and resets the aggregate to 0.
- * 
- * @param <T> data type for {@code average()} output
- */
-public final class ResourceAccumulatorTakeGradient<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new ResourceAccumulatorTakeGradient operation.
-   * 
-   * @param scope current scope
-   * @param handle The handle to an accumulator.
-   * @param numRequired Number of gradients required before we return an aggregate.
-   * @param dtype The data type of accumulated gradients. Needs to correspond to the type
-   * of the accumulator.
-   * @return a new instance of ResourceAccumulatorTakeGradient
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> ResourceAccumulatorTakeGradient<T> create(Scope scope, Operand<?> handle, Operand<TInt32> numRequired, Class<T> dtype) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ResourceAccumulatorTakeGradient", scope.makeOpName("ResourceAccumulatorTakeGradient"));
-    opBuilder.addInput(handle.asOutput(scope));
-    opBuilder.addInput(numRequired.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("dtype", dtype);
-    return new ResourceAccumulatorTakeGradient<T>(opBuilder.build());
-  }
-  
-  /**
-   * The average of the accumulated gradients.
-   */
-  public Output<T> average() {
-    return average;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return average;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ResourceAccumulatorTakeGradient";
-  
-  private Output<T> average;
-  
-  private ResourceAccumulatorTakeGradient(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    average = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyAdaMax.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyAdaMax.java
deleted file mode 100644
index 67d59ac422b..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyAdaMax.java
+++ /dev/null
@@ -1,113 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.train;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Update '*var' according to the AdaMax algorithm.
- * <p>
- * m_t <- beta1 * m_{t-1} + (1 - beta1) * g
- * v_t <- max(beta2 * v_{t-1}, abs(g))
- * variable <- variable - learning_rate / (1 - beta1^t) * m_t / (v_t + epsilon)
- */
-public final class ResourceApplyAdaMax extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.train.ResourceApplyAdaMax}
-   */
-  public static class Options {
-    
-    /**
-     * @param useLocking If `True`, updating of the var, m, and v tensors will be protected
-     * by a lock; otherwise the behavior is undefined, but may exhibit less
-     * contention.
-     */
-    public Options useLocking(Boolean useLocking) {
-      this.useLocking = useLocking;
-      return this;
-    }
-    
-    private Boolean useLocking;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new ResourceApplyAdaMax operation.
-   * 
-   * @param scope current scope
-   * @param var Should be from a Variable().
-   * @param m Should be from a Variable().
-   * @param v Should be from a Variable().
-   * @param beta1Power Must be a scalar.
-   * @param lr Scaling factor. Must be a scalar.
-   * @param beta1 Momentum factor. Must be a scalar.
-   * @param beta2 Momentum factor. Must be a scalar.
-   * @param epsilon Ridge term. Must be a scalar.
-   * @param grad The gradient.
-   * @param options carries optional attributes values
-   * @return a new instance of ResourceApplyAdaMax
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> ResourceApplyAdaMax create(Scope scope, Operand<?> var, Operand<?> m, Operand<?> v, Operand<T> beta1Power, Operand<T> lr, Operand<T> beta1, Operand<T> beta2, Operand<T> epsilon, Operand<T> grad, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ResourceApplyAdaMax", scope.makeOpName("ResourceApplyAdaMax"));
-    opBuilder.addInput(var.asOutput(scope));
-    opBuilder.addInput(m.asOutput(scope));
-    opBuilder.addInput(v.asOutput(scope));
-    opBuilder.addInput(beta1Power.asOutput(scope));
-    opBuilder.addInput(lr.asOutput(scope));
-    opBuilder.addInput(beta1.asOutput(scope));
-    opBuilder.addInput(beta2.asOutput(scope));
-    opBuilder.addInput(epsilon.asOutput(scope));
-    opBuilder.addInput(grad.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.useLocking != null) {
-          opBuilder.setAttr("use_locking", opts.useLocking);
-        }
-      }
-    }
-    return new ResourceApplyAdaMax(opBuilder.build());
-  }
-  
-  /**
-   * @param useLocking If `True`, updating of the var, m, and v tensors will be protected
-   * by a lock; otherwise the behavior is undefined, but may exhibit less
-   * contention.
-   */
-  public static Options useLocking(Boolean useLocking) {
-    return new Options().useLocking(useLocking);
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ResourceApplyAdaMax";
-  
-  private ResourceApplyAdaMax(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyAdadelta.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyAdadelta.java
deleted file mode 100644
index fdf35b45ce7..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyAdadelta.java
+++ /dev/null
@@ -1,109 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.train;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Update '*var' according to the adadelta scheme.
- * <p>
- * accum = rho() * accum + (1 - rho()) * grad.square();
- * update = (update_accum + epsilon).sqrt() * (accum + epsilon()).rsqrt() * grad;
- * update_accum = rho() * update_accum + (1 - rho()) * update.square();
- * var -= update;
- */
-@Operator(group = "train")
-public final class ResourceApplyAdadelta extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.train.ResourceApplyAdadelta}
-   */
-  public static class Options {
-    
-    /**
-     * @param useLocking If True, updating of the var, accum and update_accum tensors will be protected by
-     * a lock; otherwise the behavior is undefined, but may exhibit less contention.
-     */
-    public Options useLocking(Boolean useLocking) {
-      this.useLocking = useLocking;
-      return this;
-    }
-    
-    private Boolean useLocking;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new ResourceApplyAdadelta operation.
-   * 
-   * @param scope current scope
-   * @param var Should be from a Variable().
-   * @param accum Should be from a Variable().
-   * @param accumUpdate Should be from a Variable().
-   * @param lr Scaling factor. Must be a scalar.
-   * @param rho Decay factor. Must be a scalar.
-   * @param epsilon Constant factor. Must be a scalar.
-   * @param grad The gradient.
-   * @param options carries optional attributes values
-   * @return a new instance of ResourceApplyAdadelta
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> ResourceApplyAdadelta create(Scope scope, Operand<?> var, Operand<?> accum, Operand<?> accumUpdate, Operand<T> lr, Operand<T> rho, Operand<T> epsilon, Operand<T> grad, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ResourceApplyAdadelta", scope.makeOpName("ResourceApplyAdadelta"));
-    opBuilder.addInput(var.asOutput(scope));
-    opBuilder.addInput(accum.asOutput(scope));
-    opBuilder.addInput(accumUpdate.asOutput(scope));
-    opBuilder.addInput(lr.asOutput(scope));
-    opBuilder.addInput(rho.asOutput(scope));
-    opBuilder.addInput(epsilon.asOutput(scope));
-    opBuilder.addInput(grad.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.useLocking != null) {
-          opBuilder.setAttr("use_locking", opts.useLocking);
-        }
-      }
-    }
-    return new ResourceApplyAdadelta(opBuilder.build());
-  }
-  
-  /**
-   * @param useLocking If True, updating of the var, accum and update_accum tensors will be protected by
-   * a lock; otherwise the behavior is undefined, but may exhibit less contention.
-   */
-  public static Options useLocking(Boolean useLocking) {
-    return new Options().useLocking(useLocking);
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ResourceApplyAdadelta";
-  
-  private ResourceApplyAdadelta(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyAdagrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyAdagrad.java
deleted file mode 100644
index 8b6817db364..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyAdagrad.java
+++ /dev/null
@@ -1,123 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.train;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Update '*var' according to the adagrad scheme.
- * <p>
- * accum += grad * grad
- * var -= lr * grad * (1 / (sqrt(accum) + epsilon))
- */
-public final class ResourceApplyAdagrad extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.train.ResourceApplyAdagrad}
-   */
-  public static class Options {
-    
-    /**
-     * @param useLocking If `True`, updating of the var and accum tensors will be protected
-     * by a lock; otherwise the behavior is undefined, but may exhibit less
-     * contention.
-     */
-    public Options useLocking(Boolean useLocking) {
-      this.useLocking = useLocking;
-      return this;
-    }
-    
-    /**
-     * @param updateSlots 
-     */
-    public Options updateSlots(Boolean updateSlots) {
-      this.updateSlots = updateSlots;
-      return this;
-    }
-    
-    private Boolean useLocking;
-    private Boolean updateSlots;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new ResourceApplyAdagrad operation.
-   * 
-   * @param scope current scope
-   * @param var Should be from a Variable().
-   * @param accum Should be from a Variable().
-   * @param lr Scaling factor. Must be a scalar.
-   * @param epsilon Constant factor. Must be a scalar.
-   * @param grad The gradient.
-   * @param options carries optional attributes values
-   * @return a new instance of ResourceApplyAdagrad
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> ResourceApplyAdagrad create(Scope scope, Operand<?> var, Operand<?> accum, Operand<T> lr, Operand<T> epsilon, Operand<T> grad, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ResourceApplyAdagradV2", scope.makeOpName("ResourceApplyAdagrad"));
-    opBuilder.addInput(var.asOutput(scope));
-    opBuilder.addInput(accum.asOutput(scope));
-    opBuilder.addInput(lr.asOutput(scope));
-    opBuilder.addInput(epsilon.asOutput(scope));
-    opBuilder.addInput(grad.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.useLocking != null) {
-          opBuilder.setAttr("use_locking", opts.useLocking);
-        }
-        if (opts.updateSlots != null) {
-          opBuilder.setAttr("update_slots", opts.updateSlots);
-        }
-      }
-    }
-    return new ResourceApplyAdagrad(opBuilder.build());
-  }
-  
-  /**
-   * @param useLocking If `True`, updating of the var and accum tensors will be protected
-   * by a lock; otherwise the behavior is undefined, but may exhibit less
-   * contention.
-   */
-  public static Options useLocking(Boolean useLocking) {
-    return new Options().useLocking(useLocking);
-  }
-  
-  /**
-   * @param updateSlots 
-   */
-  public static Options updateSlots(Boolean updateSlots) {
-    return new Options().updateSlots(updateSlots);
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ResourceApplyAdagradV2";
-  
-  private ResourceApplyAdagrad(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyAdagradDa.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyAdagradDa.java
deleted file mode 100644
index a79c4e454b1..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyAdagradDa.java
+++ /dev/null
@@ -1,107 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.train;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TType;
-
-/**
- * Update '*var' according to the proximal adagrad scheme.
- */
-@Operator(group = "train")
-public final class ResourceApplyAdagradDa extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.train.ResourceApplyAdagradDa}
-   */
-  public static class Options {
-    
-    /**
-     * @param useLocking If True, updating of the var and accum tensors will be protected by
-     * a lock; otherwise the behavior is undefined, but may exhibit less contention.
-     */
-    public Options useLocking(Boolean useLocking) {
-      this.useLocking = useLocking;
-      return this;
-    }
-    
-    private Boolean useLocking;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new ResourceApplyAdagradDa operation.
-   * 
-   * @param scope current scope
-   * @param var Should be from a Variable().
-   * @param gradientAccumulator Should be from a Variable().
-   * @param gradientSquaredAccumulator Should be from a Variable().
-   * @param grad The gradient.
-   * @param lr Scaling factor. Must be a scalar.
-   * @param l1 L1 regularization. Must be a scalar.
-   * @param l2 L2 regularization. Must be a scalar.
-   * @param globalStep Training step number. Must be a scalar.
-   * @param options carries optional attributes values
-   * @return a new instance of ResourceApplyAdagradDa
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> ResourceApplyAdagradDa create(Scope scope, Operand<?> var, Operand<?> gradientAccumulator, Operand<?> gradientSquaredAccumulator, Operand<T> grad, Operand<T> lr, Operand<T> l1, Operand<T> l2, Operand<TInt64> globalStep, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ResourceApplyAdagradDA", scope.makeOpName("ResourceApplyAdagradDa"));
-    opBuilder.addInput(var.asOutput(scope));
-    opBuilder.addInput(gradientAccumulator.asOutput(scope));
-    opBuilder.addInput(gradientSquaredAccumulator.asOutput(scope));
-    opBuilder.addInput(grad.asOutput(scope));
-    opBuilder.addInput(lr.asOutput(scope));
-    opBuilder.addInput(l1.asOutput(scope));
-    opBuilder.addInput(l2.asOutput(scope));
-    opBuilder.addInput(globalStep.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.useLocking != null) {
-          opBuilder.setAttr("use_locking", opts.useLocking);
-        }
-      }
-    }
-    return new ResourceApplyAdagradDa(opBuilder.build());
-  }
-  
-  /**
-   * @param useLocking If True, updating of the var and accum tensors will be protected by
-   * a lock; otherwise the behavior is undefined, but may exhibit less contention.
-   */
-  public static Options useLocking(Boolean useLocking) {
-    return new Options().useLocking(useLocking);
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ResourceApplyAdagradDA";
-  
-  private ResourceApplyAdagradDa(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyAdam.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyAdam.java
deleted file mode 100644
index 8867adfa7f6..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyAdam.java
+++ /dev/null
@@ -1,136 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.train;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Update '*var' according to the Adam algorithm.
- * <p>
- * $$\text{lr}_t := \mathrm{learning_rate} * \sqrt{1 - \beta_2^t} / (1 - \beta_1^t)$$
- * $$m_t := \beta_1 * m_{t-1} + (1 - \beta_1) * g$$
- * $$v_t := \beta_2 * v_{t-1} + (1 - \beta_2) * g * g$$
- * $$\text{variable} := \text{variable} - \text{lr}_t * m_t / (\sqrt{v_t} + \epsilon)$$
- */
-@Operator(group = "train")
-public final class ResourceApplyAdam extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.train.ResourceApplyAdam}
-   */
-  public static class Options {
-    
-    /**
-     * @param useLocking If `True`, updating of the var, m, and v tensors will be protected
-     * by a lock; otherwise the behavior is undefined, but may exhibit less
-     * contention.
-     */
-    public Options useLocking(Boolean useLocking) {
-      this.useLocking = useLocking;
-      return this;
-    }
-    
-    /**
-     * @param useNesterov If `True`, uses the nesterov update.
-     */
-    public Options useNesterov(Boolean useNesterov) {
-      this.useNesterov = useNesterov;
-      return this;
-    }
-    
-    private Boolean useLocking;
-    private Boolean useNesterov;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new ResourceApplyAdam operation.
-   * 
-   * @param scope current scope
-   * @param var Should be from a Variable().
-   * @param m Should be from a Variable().
-   * @param v Should be from a Variable().
-   * @param beta1Power Must be a scalar.
-   * @param beta2Power Must be a scalar.
-   * @param lr Scaling factor. Must be a scalar.
-   * @param beta1 Momentum factor. Must be a scalar.
-   * @param beta2 Momentum factor. Must be a scalar.
-   * @param epsilon Ridge term. Must be a scalar.
-   * @param grad The gradient.
-   * @param options carries optional attributes values
-   * @return a new instance of ResourceApplyAdam
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> ResourceApplyAdam create(Scope scope, Operand<?> var, Operand<?> m, Operand<?> v, Operand<T> beta1Power, Operand<T> beta2Power, Operand<T> lr, Operand<T> beta1, Operand<T> beta2, Operand<T> epsilon, Operand<T> grad, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ResourceApplyAdam", scope.makeOpName("ResourceApplyAdam"));
-    opBuilder.addInput(var.asOutput(scope));
-    opBuilder.addInput(m.asOutput(scope));
-    opBuilder.addInput(v.asOutput(scope));
-    opBuilder.addInput(beta1Power.asOutput(scope));
-    opBuilder.addInput(beta2Power.asOutput(scope));
-    opBuilder.addInput(lr.asOutput(scope));
-    opBuilder.addInput(beta1.asOutput(scope));
-    opBuilder.addInput(beta2.asOutput(scope));
-    opBuilder.addInput(epsilon.asOutput(scope));
-    opBuilder.addInput(grad.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.useLocking != null) {
-          opBuilder.setAttr("use_locking", opts.useLocking);
-        }
-        if (opts.useNesterov != null) {
-          opBuilder.setAttr("use_nesterov", opts.useNesterov);
-        }
-      }
-    }
-    return new ResourceApplyAdam(opBuilder.build());
-  }
-  
-  /**
-   * @param useLocking If `True`, updating of the var, m, and v tensors will be protected
-   * by a lock; otherwise the behavior is undefined, but may exhibit less
-   * contention.
-   */
-  public static Options useLocking(Boolean useLocking) {
-    return new Options().useLocking(useLocking);
-  }
-  
-  /**
-   * @param useNesterov If `True`, uses the nesterov update.
-   */
-  public static Options useNesterov(Boolean useNesterov) {
-    return new Options().useNesterov(useNesterov);
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ResourceApplyAdam";
-  
-  private ResourceApplyAdam(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyAdamWithAmsgrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyAdamWithAmsgrad.java
deleted file mode 100644
index a9d5b7b72d2..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyAdamWithAmsgrad.java
+++ /dev/null
@@ -1,120 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.train;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Update '*var' according to the Adam algorithm.
- * <p>
- * $$\text{lr}_t := \mathrm{learning_rate} * \sqrt{1 - \beta_2^t} / (1 - \beta_1^t)$$
- * $$m_t := \beta_1 * m_{t-1} + (1 - \beta_1) * g$$
- * $$v_t := \beta_2 * v_{t-1} + (1 - \beta_2) * g * g$$
- * $$\hat{v}_t := max{\hat{v}_{t-1}, v_t}$$
- * $$\text{variable} := \text{variable} - \text{lr}_t * m_t / (\sqrt{\hat{v}_t} + \epsilon)$$
- */
-@Operator(group = "train")
-public final class ResourceApplyAdamWithAmsgrad extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.train.ResourceApplyAdamWithAmsgrad}
-   */
-  public static class Options {
-    
-    /**
-     * @param useLocking If `True`, updating of the var, m, and v tensors will be protected
-     * by a lock; otherwise the behavior is undefined, but may exhibit less
-     * contention.
-     */
-    public Options useLocking(Boolean useLocking) {
-      this.useLocking = useLocking;
-      return this;
-    }
-    
-    private Boolean useLocking;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new ResourceApplyAdamWithAmsgrad operation.
-   * 
-   * @param scope current scope
-   * @param var Should be from a Variable().
-   * @param m Should be from a Variable().
-   * @param v Should be from a Variable().
-   * @param vhat Should be from a Variable().
-   * @param beta1Power Must be a scalar.
-   * @param beta2Power Must be a scalar.
-   * @param lr Scaling factor. Must be a scalar.
-   * @param beta1 Momentum factor. Must be a scalar.
-   * @param beta2 Momentum factor. Must be a scalar.
-   * @param epsilon Ridge term. Must be a scalar.
-   * @param grad The gradient.
-   * @param options carries optional attributes values
-   * @return a new instance of ResourceApplyAdamWithAmsgrad
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> ResourceApplyAdamWithAmsgrad create(Scope scope, Operand<?> var, Operand<?> m, Operand<?> v, Operand<?> vhat, Operand<T> beta1Power, Operand<T> beta2Power, Operand<T> lr, Operand<T> beta1, Operand<T> beta2, Operand<T> epsilon, Operand<T> grad, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ResourceApplyAdamWithAmsgrad", scope.makeOpName("ResourceApplyAdamWithAmsgrad"));
-    opBuilder.addInput(var.asOutput(scope));
-    opBuilder.addInput(m.asOutput(scope));
-    opBuilder.addInput(v.asOutput(scope));
-    opBuilder.addInput(vhat.asOutput(scope));
-    opBuilder.addInput(beta1Power.asOutput(scope));
-    opBuilder.addInput(beta2Power.asOutput(scope));
-    opBuilder.addInput(lr.asOutput(scope));
-    opBuilder.addInput(beta1.asOutput(scope));
-    opBuilder.addInput(beta2.asOutput(scope));
-    opBuilder.addInput(epsilon.asOutput(scope));
-    opBuilder.addInput(grad.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.useLocking != null) {
-          opBuilder.setAttr("use_locking", opts.useLocking);
-        }
-      }
-    }
-    return new ResourceApplyAdamWithAmsgrad(opBuilder.build());
-  }
-  
-  /**
-   * @param useLocking If `True`, updating of the var, m, and v tensors will be protected
-   * by a lock; otherwise the behavior is undefined, but may exhibit less
-   * contention.
-   */
-  public static Options useLocking(Boolean useLocking) {
-    return new Options().useLocking(useLocking);
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ResourceApplyAdamWithAmsgrad";
-  
-  private ResourceApplyAdamWithAmsgrad(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyAddSign.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyAddSign.java
deleted file mode 100644
index e23409074f7..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyAddSign.java
+++ /dev/null
@@ -1,110 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.train;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Update '*var' according to the AddSign update.
- * <p>
- * m_t <- beta1 * m_{t-1} + (1 - beta1) * g
- * update <- (alpha + sign_decay * sign(g) *sign(m)) * g
- * variable <- variable - lr_t * update
- */
-@Operator(group = "train")
-public final class ResourceApplyAddSign extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.train.ResourceApplyAddSign}
-   */
-  public static class Options {
-    
-    /**
-     * @param useLocking If `True`, updating of the var and m tensors is
-     * protected by a lock; otherwise the behavior is undefined, but may exhibit less
-     * contention.
-     */
-    public Options useLocking(Boolean useLocking) {
-      this.useLocking = useLocking;
-      return this;
-    }
-    
-    private Boolean useLocking;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new ResourceApplyAddSign operation.
-   * 
-   * @param scope current scope
-   * @param var Should be from a Variable().
-   * @param m Should be from a Variable().
-   * @param lr Scaling factor. Must be a scalar.
-   * @param alpha Must be a scalar.
-   * @param signDecay Must be a scalar.
-   * @param beta Must be a scalar.
-   * @param grad The gradient.
-   * @param options carries optional attributes values
-   * @return a new instance of ResourceApplyAddSign
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> ResourceApplyAddSign create(Scope scope, Operand<?> var, Operand<?> m, Operand<T> lr, Operand<T> alpha, Operand<T> signDecay, Operand<T> beta, Operand<T> grad, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ResourceApplyAddSign", scope.makeOpName("ResourceApplyAddSign"));
-    opBuilder.addInput(var.asOutput(scope));
-    opBuilder.addInput(m.asOutput(scope));
-    opBuilder.addInput(lr.asOutput(scope));
-    opBuilder.addInput(alpha.asOutput(scope));
-    opBuilder.addInput(signDecay.asOutput(scope));
-    opBuilder.addInput(beta.asOutput(scope));
-    opBuilder.addInput(grad.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.useLocking != null) {
-          opBuilder.setAttr("use_locking", opts.useLocking);
-        }
-      }
-    }
-    return new ResourceApplyAddSign(opBuilder.build());
-  }
-  
-  /**
-   * @param useLocking If `True`, updating of the var and m tensors is
-   * protected by a lock; otherwise the behavior is undefined, but may exhibit less
-   * contention.
-   */
-  public static Options useLocking(Boolean useLocking) {
-    return new Options().useLocking(useLocking);
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ResourceApplyAddSign";
-  
-  private ResourceApplyAddSign(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyCenteredRmsProp.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyCenteredRmsProp.java
deleted file mode 100644
index 683dbef6a0f..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyCenteredRmsProp.java
+++ /dev/null
@@ -1,129 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.train;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Update '*var' according to the centered RMSProp algorithm.
- * <p>
- * The centered RMSProp algorithm uses an estimate of the centered second moment
- * (i.e., the variance) for normalization, as opposed to regular RMSProp, which
- * uses the (uncentered) second moment. This often helps with training, but is
- * slightly more expensive in terms of computation and memory.
- * <p>
- * Note that in dense implementation of this algorithm, mg, ms, and mom will
- * update even if the grad is zero, but in this sparse implementation, mg, ms,
- * and mom will not update in iterations during which the grad is zero.
- * <p>
- * mean_square = decay * mean_square + (1-decay) * gradient ** 2
- * mean_grad = decay * mean_grad + (1-decay) * gradient
- * <p>
- * Delta = learning_rate * gradient / sqrt(mean_square + epsilon - mean_grad ** 2)
- * <p>
- * mg <- rho * mg_{t-1} + (1-rho) * grad
- * ms <- rho * ms_{t-1} + (1-rho) * grad * grad
- * mom <- momentum * mom_{t-1} + lr * grad / sqrt(ms - mg * mg + epsilon)
- * var <- var - mom
- */
-@Operator(group = "train")
-public final class ResourceApplyCenteredRmsProp extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.train.ResourceApplyCenteredRmsProp}
-   */
-  public static class Options {
-    
-    /**
-     * @param useLocking If `True`, updating of the var, mg, ms, and mom tensors is
-     * protected by a lock; otherwise the behavior is undefined, but may exhibit less
-     * contention.
-     */
-    public Options useLocking(Boolean useLocking) {
-      this.useLocking = useLocking;
-      return this;
-    }
-    
-    private Boolean useLocking;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new ResourceApplyCenteredRmsProp operation.
-   * 
-   * @param scope current scope
-   * @param var Should be from a Variable().
-   * @param mg Should be from a Variable().
-   * @param ms Should be from a Variable().
-   * @param mom Should be from a Variable().
-   * @param lr Scaling factor. Must be a scalar.
-   * @param rho Decay rate. Must be a scalar.
-   * @param momentum 
-   * @param epsilon Ridge term. Must be a scalar.
-   * @param grad The gradient.
-   * @param options carries optional attributes values
-   * @return a new instance of ResourceApplyCenteredRmsProp
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> ResourceApplyCenteredRmsProp create(Scope scope, Operand<?> var, Operand<?> mg, Operand<?> ms, Operand<?> mom, Operand<T> lr, Operand<T> rho, Operand<T> momentum, Operand<T> epsilon, Operand<T> grad, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ResourceApplyCenteredRMSProp", scope.makeOpName("ResourceApplyCenteredRmsProp"));
-    opBuilder.addInput(var.asOutput(scope));
-    opBuilder.addInput(mg.asOutput(scope));
-    opBuilder.addInput(ms.asOutput(scope));
-    opBuilder.addInput(mom.asOutput(scope));
-    opBuilder.addInput(lr.asOutput(scope));
-    opBuilder.addInput(rho.asOutput(scope));
-    opBuilder.addInput(momentum.asOutput(scope));
-    opBuilder.addInput(epsilon.asOutput(scope));
-    opBuilder.addInput(grad.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.useLocking != null) {
-          opBuilder.setAttr("use_locking", opts.useLocking);
-        }
-      }
-    }
-    return new ResourceApplyCenteredRmsProp(opBuilder.build());
-  }
-  
-  /**
-   * @param useLocking If `True`, updating of the var, mg, ms, and mom tensors is
-   * protected by a lock; otherwise the behavior is undefined, but may exhibit less
-   * contention.
-   */
-  public static Options useLocking(Boolean useLocking) {
-    return new Options().useLocking(useLocking);
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ResourceApplyCenteredRMSProp";
-  
-  private ResourceApplyCenteredRmsProp(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyFtrl.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyFtrl.java
deleted file mode 100644
index 93d78c3a498..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyFtrl.java
+++ /dev/null
@@ -1,137 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.train;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Update '*var' according to the Ftrl-proximal scheme.
- * <p>
- * grad_with_shrinkage = grad + 2 * l2_shrinkage * var
- * accum_new = accum + grad_with_shrinkage * grad_with_shrinkage
- * linear += grad_with_shrinkage +
- *     (accum_new^(-lr_power) - accum^(-lr_power)) / lr * var
- * quadratic = 1.0 / (accum_new^(lr_power) * lr) + 2 * l2
- * var = (sign(linear) * l1 - linear) / quadratic if |linear| > l1 else 0.0
- * accum = accum_new
- */
-@Operator(group = "train")
-public final class ResourceApplyFtrl extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.train.ResourceApplyFtrl}
-   */
-  public static class Options {
-    
-    /**
-     * @param useLocking If `True`, updating of the var and accum tensors will be protected
-     * by a lock; otherwise the behavior is undefined, but may exhibit less
-     * contention.
-     */
-    public Options useLocking(Boolean useLocking) {
-      this.useLocking = useLocking;
-      return this;
-    }
-    
-    /**
-     * @param multiplyLinearByLr 
-     */
-    public Options multiplyLinearByLr(Boolean multiplyLinearByLr) {
-      this.multiplyLinearByLr = multiplyLinearByLr;
-      return this;
-    }
-    
-    private Boolean useLocking;
-    private Boolean multiplyLinearByLr;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new ResourceApplyFtrl operation.
-   * 
-   * @param scope current scope
-   * @param var Should be from a Variable().
-   * @param accum Should be from a Variable().
-   * @param linear Should be from a Variable().
-   * @param grad The gradient.
-   * @param lr Scaling factor. Must be a scalar.
-   * @param l1 L1 regularization. Must be a scalar.
-   * @param l2 L2 shrinkage regularization. Must be a scalar.
-   * @param l2Shrinkage 
-   * @param lrPower Scaling factor. Must be a scalar.
-   * @param options carries optional attributes values
-   * @return a new instance of ResourceApplyFtrl
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> ResourceApplyFtrl create(Scope scope, Operand<?> var, Operand<?> accum, Operand<?> linear, Operand<T> grad, Operand<T> lr, Operand<T> l1, Operand<T> l2, Operand<T> l2Shrinkage, Operand<T> lrPower, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ResourceApplyFtrlV2", scope.makeOpName("ResourceApplyFtrl"));
-    opBuilder.addInput(var.asOutput(scope));
-    opBuilder.addInput(accum.asOutput(scope));
-    opBuilder.addInput(linear.asOutput(scope));
-    opBuilder.addInput(grad.asOutput(scope));
-    opBuilder.addInput(lr.asOutput(scope));
-    opBuilder.addInput(l1.asOutput(scope));
-    opBuilder.addInput(l2.asOutput(scope));
-    opBuilder.addInput(l2Shrinkage.asOutput(scope));
-    opBuilder.addInput(lrPower.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.useLocking != null) {
-          opBuilder.setAttr("use_locking", opts.useLocking);
-        }
-        if (opts.multiplyLinearByLr != null) {
-          opBuilder.setAttr("multiply_linear_by_lr", opts.multiplyLinearByLr);
-        }
-      }
-    }
-    return new ResourceApplyFtrl(opBuilder.build());
-  }
-  
-  /**
-   * @param useLocking If `True`, updating of the var and accum tensors will be protected
-   * by a lock; otherwise the behavior is undefined, but may exhibit less
-   * contention.
-   */
-  public static Options useLocking(Boolean useLocking) {
-    return new Options().useLocking(useLocking);
-  }
-  
-  /**
-   * @param multiplyLinearByLr 
-   */
-  public static Options multiplyLinearByLr(Boolean multiplyLinearByLr) {
-    return new Options().multiplyLinearByLr(multiplyLinearByLr);
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ResourceApplyFtrlV2";
-  
-  private ResourceApplyFtrl(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyGradientDescent.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyGradientDescent.java
deleted file mode 100644
index 0098692cf89..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyGradientDescent.java
+++ /dev/null
@@ -1,96 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.train;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Update '*var' by subtracting 'alpha' * 'delta' from it.
- */
-@Operator(group = "train")
-public final class ResourceApplyGradientDescent extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.train.ResourceApplyGradientDescent}
-   */
-  public static class Options {
-    
-    /**
-     * @param useLocking If `True`, the subtraction will be protected by a lock;
-     * otherwise the behavior is undefined, but may exhibit less contention.
-     */
-    public Options useLocking(Boolean useLocking) {
-      this.useLocking = useLocking;
-      return this;
-    }
-    
-    private Boolean useLocking;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new ResourceApplyGradientDescent operation.
-   * 
-   * @param scope current scope
-   * @param var Should be from a Variable().
-   * @param alpha Scaling factor. Must be a scalar.
-   * @param delta The change.
-   * @param options carries optional attributes values
-   * @return a new instance of ResourceApplyGradientDescent
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> ResourceApplyGradientDescent create(Scope scope, Operand<?> var, Operand<T> alpha, Operand<T> delta, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ResourceApplyGradientDescent", scope.makeOpName("ResourceApplyGradientDescent"));
-    opBuilder.addInput(var.asOutput(scope));
-    opBuilder.addInput(alpha.asOutput(scope));
-    opBuilder.addInput(delta.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.useLocking != null) {
-          opBuilder.setAttr("use_locking", opts.useLocking);
-        }
-      }
-    }
-    return new ResourceApplyGradientDescent(opBuilder.build());
-  }
-  
-  /**
-   * @param useLocking If `True`, the subtraction will be protected by a lock;
-   * otherwise the behavior is undefined, but may exhibit less contention.
-   */
-  public static Options useLocking(Boolean useLocking) {
-    return new Options().useLocking(useLocking);
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ResourceApplyGradientDescent";
-  
-  private ResourceApplyGradientDescent(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyKerasMomentum.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyKerasMomentum.java
deleted file mode 100644
index 2f25475a048..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyKerasMomentum.java
+++ /dev/null
@@ -1,130 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.train;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Update '*var' according to the momentum scheme.
- * <p>
- * Set use_nesterov = True if you want to use Nesterov momentum.
- * <p>
- * accum = accum * momentum - lr * grad
- * var += accum
- */
-@Operator(group = "train")
-public final class ResourceApplyKerasMomentum extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.train.ResourceApplyKerasMomentum}
-   */
-  public static class Options {
-    
-    /**
-     * @param useLocking If `True`, updating of the var and accum tensors will be protected
-     * by a lock; otherwise the behavior is undefined, but may exhibit less
-     * contention.
-     */
-    public Options useLocking(Boolean useLocking) {
-      this.useLocking = useLocking;
-      return this;
-    }
-    
-    /**
-     * @param useNesterov If `True`, the tensor passed to compute grad will be
-     * var + momentum * accum, so in the end, the var you get is actually
-     * var + momentum * accum.
-     */
-    public Options useNesterov(Boolean useNesterov) {
-      this.useNesterov = useNesterov;
-      return this;
-    }
-    
-    private Boolean useLocking;
-    private Boolean useNesterov;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new ResourceApplyKerasMomentum operation.
-   * 
-   * @param scope current scope
-   * @param var Should be from a Variable().
-   * @param accum Should be from a Variable().
-   * @param lr Scaling factor. Must be a scalar.
-   * @param grad The gradient.
-   * @param momentum Momentum. Must be a scalar.
-   * @param options carries optional attributes values
-   * @return a new instance of ResourceApplyKerasMomentum
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> ResourceApplyKerasMomentum create(Scope scope, Operand<?> var, Operand<?> accum, Operand<T> lr, Operand<T> grad, Operand<T> momentum, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ResourceApplyKerasMomentum", scope.makeOpName("ResourceApplyKerasMomentum"));
-    opBuilder.addInput(var.asOutput(scope));
-    opBuilder.addInput(accum.asOutput(scope));
-    opBuilder.addInput(lr.asOutput(scope));
-    opBuilder.addInput(grad.asOutput(scope));
-    opBuilder.addInput(momentum.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.useLocking != null) {
-          opBuilder.setAttr("use_locking", opts.useLocking);
-        }
-        if (opts.useNesterov != null) {
-          opBuilder.setAttr("use_nesterov", opts.useNesterov);
-        }
-      }
-    }
-    return new ResourceApplyKerasMomentum(opBuilder.build());
-  }
-  
-  /**
-   * @param useLocking If `True`, updating of the var and accum tensors will be protected
-   * by a lock; otherwise the behavior is undefined, but may exhibit less
-   * contention.
-   */
-  public static Options useLocking(Boolean useLocking) {
-    return new Options().useLocking(useLocking);
-  }
-  
-  /**
-   * @param useNesterov If `True`, the tensor passed to compute grad will be
-   * var + momentum * accum, so in the end, the var you get is actually
-   * var + momentum * accum.
-   */
-  public static Options useNesterov(Boolean useNesterov) {
-    return new Options().useNesterov(useNesterov);
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ResourceApplyKerasMomentum";
-  
-  private ResourceApplyKerasMomentum(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyMomentum.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyMomentum.java
deleted file mode 100644
index b2e93b6dc17..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyMomentum.java
+++ /dev/null
@@ -1,130 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.train;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Update '*var' according to the momentum scheme.
- * <p>
- * Set use_nesterov = True if you want to use Nesterov momentum.
- * <p>
- * accum = accum * momentum + grad
- * var -= lr * accum
- */
-@Operator(group = "train")
-public final class ResourceApplyMomentum extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.train.ResourceApplyMomentum}
-   */
-  public static class Options {
-    
-    /**
-     * @param useLocking If `True`, updating of the var and accum tensors will be protected
-     * by a lock; otherwise the behavior is undefined, but may exhibit less
-     * contention.
-     */
-    public Options useLocking(Boolean useLocking) {
-      this.useLocking = useLocking;
-      return this;
-    }
-    
-    /**
-     * @param useNesterov If `True`, the tensor passed to compute grad will be
-     * var - lr * momentum * accum, so in the end, the var you get is actually
-     * var - lr * momentum * accum.
-     */
-    public Options useNesterov(Boolean useNesterov) {
-      this.useNesterov = useNesterov;
-      return this;
-    }
-    
-    private Boolean useLocking;
-    private Boolean useNesterov;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new ResourceApplyMomentum operation.
-   * 
-   * @param scope current scope
-   * @param var Should be from a Variable().
-   * @param accum Should be from a Variable().
-   * @param lr Scaling factor. Must be a scalar.
-   * @param grad The gradient.
-   * @param momentum Momentum. Must be a scalar.
-   * @param options carries optional attributes values
-   * @return a new instance of ResourceApplyMomentum
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> ResourceApplyMomentum create(Scope scope, Operand<?> var, Operand<?> accum, Operand<T> lr, Operand<T> grad, Operand<T> momentum, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ResourceApplyMomentum", scope.makeOpName("ResourceApplyMomentum"));
-    opBuilder.addInput(var.asOutput(scope));
-    opBuilder.addInput(accum.asOutput(scope));
-    opBuilder.addInput(lr.asOutput(scope));
-    opBuilder.addInput(grad.asOutput(scope));
-    opBuilder.addInput(momentum.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.useLocking != null) {
-          opBuilder.setAttr("use_locking", opts.useLocking);
-        }
-        if (opts.useNesterov != null) {
-          opBuilder.setAttr("use_nesterov", opts.useNesterov);
-        }
-      }
-    }
-    return new ResourceApplyMomentum(opBuilder.build());
-  }
-  
-  /**
-   * @param useLocking If `True`, updating of the var and accum tensors will be protected
-   * by a lock; otherwise the behavior is undefined, but may exhibit less
-   * contention.
-   */
-  public static Options useLocking(Boolean useLocking) {
-    return new Options().useLocking(useLocking);
-  }
-  
-  /**
-   * @param useNesterov If `True`, the tensor passed to compute grad will be
-   * var - lr * momentum * accum, so in the end, the var you get is actually
-   * var - lr * momentum * accum.
-   */
-  public static Options useNesterov(Boolean useNesterov) {
-    return new Options().useNesterov(useNesterov);
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ResourceApplyMomentum";
-  
-  private ResourceApplyMomentum(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyPowerSign.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyPowerSign.java
deleted file mode 100644
index b31285fb631..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyPowerSign.java
+++ /dev/null
@@ -1,110 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.train;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Update '*var' according to the AddSign update.
- * <p>
- * m_t <- beta1 * m_{t-1} + (1 - beta1) * g
- * update <- exp(logbase * sign_decay * sign(g) * sign(m_t)) * g
- * variable <- variable - lr_t * update
- */
-@Operator(group = "train")
-public final class ResourceApplyPowerSign extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.train.ResourceApplyPowerSign}
-   */
-  public static class Options {
-    
-    /**
-     * @param useLocking If `True`, updating of the var and m tensors is
-     * protected by a lock; otherwise the behavior is undefined, but may exhibit less
-     * contention.
-     */
-    public Options useLocking(Boolean useLocking) {
-      this.useLocking = useLocking;
-      return this;
-    }
-    
-    private Boolean useLocking;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new ResourceApplyPowerSign operation.
-   * 
-   * @param scope current scope
-   * @param var Should be from a Variable().
-   * @param m Should be from a Variable().
-   * @param lr Scaling factor. Must be a scalar.
-   * @param logbase Must be a scalar.
-   * @param signDecay Must be a scalar.
-   * @param beta Must be a scalar.
-   * @param grad The gradient.
-   * @param options carries optional attributes values
-   * @return a new instance of ResourceApplyPowerSign
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> ResourceApplyPowerSign create(Scope scope, Operand<?> var, Operand<?> m, Operand<T> lr, Operand<T> logbase, Operand<T> signDecay, Operand<T> beta, Operand<T> grad, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ResourceApplyPowerSign", scope.makeOpName("ResourceApplyPowerSign"));
-    opBuilder.addInput(var.asOutput(scope));
-    opBuilder.addInput(m.asOutput(scope));
-    opBuilder.addInput(lr.asOutput(scope));
-    opBuilder.addInput(logbase.asOutput(scope));
-    opBuilder.addInput(signDecay.asOutput(scope));
-    opBuilder.addInput(beta.asOutput(scope));
-    opBuilder.addInput(grad.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.useLocking != null) {
-          opBuilder.setAttr("use_locking", opts.useLocking);
-        }
-      }
-    }
-    return new ResourceApplyPowerSign(opBuilder.build());
-  }
-  
-  /**
-   * @param useLocking If `True`, updating of the var and m tensors is
-   * protected by a lock; otherwise the behavior is undefined, but may exhibit less
-   * contention.
-   */
-  public static Options useLocking(Boolean useLocking) {
-    return new Options().useLocking(useLocking);
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ResourceApplyPowerSign";
-  
-  private ResourceApplyPowerSign(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyProximalAdagrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyProximalAdagrad.java
deleted file mode 100644
index 38528f4054b..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyProximalAdagrad.java
+++ /dev/null
@@ -1,106 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.train;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Update '*var' and '*accum' according to FOBOS with Adagrad learning rate.
- * <p>
- * accum += grad <i> grad
- * prox_v = var - lr </i> grad <i> (1 / sqrt(accum))
- * var = sign(prox_v)/(1+lr</i>l2) <i> max{|prox_v|-lr</i>l1,0}
- */
-@Operator(group = "train")
-public final class ResourceApplyProximalAdagrad extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.train.ResourceApplyProximalAdagrad}
-   */
-  public static class Options {
-    
-    /**
-     * @param useLocking If True, updating of the var and accum tensors will be protected by
-     * a lock; otherwise the behavior is undefined, but may exhibit less contention.
-     */
-    public Options useLocking(Boolean useLocking) {
-      this.useLocking = useLocking;
-      return this;
-    }
-    
-    private Boolean useLocking;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new ResourceApplyProximalAdagrad operation.
-   * 
-   * @param scope current scope
-   * @param var Should be from a Variable().
-   * @param accum Should be from a Variable().
-   * @param lr Scaling factor. Must be a scalar.
-   * @param l1 L1 regularization. Must be a scalar.
-   * @param l2 L2 regularization. Must be a scalar.
-   * @param grad The gradient.
-   * @param options carries optional attributes values
-   * @return a new instance of ResourceApplyProximalAdagrad
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> ResourceApplyProximalAdagrad create(Scope scope, Operand<?> var, Operand<?> accum, Operand<T> lr, Operand<T> l1, Operand<T> l2, Operand<T> grad, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ResourceApplyProximalAdagrad", scope.makeOpName("ResourceApplyProximalAdagrad"));
-    opBuilder.addInput(var.asOutput(scope));
-    opBuilder.addInput(accum.asOutput(scope));
-    opBuilder.addInput(lr.asOutput(scope));
-    opBuilder.addInput(l1.asOutput(scope));
-    opBuilder.addInput(l2.asOutput(scope));
-    opBuilder.addInput(grad.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.useLocking != null) {
-          opBuilder.setAttr("use_locking", opts.useLocking);
-        }
-      }
-    }
-    return new ResourceApplyProximalAdagrad(opBuilder.build());
-  }
-  
-  /**
-   * @param useLocking If True, updating of the var and accum tensors will be protected by
-   * a lock; otherwise the behavior is undefined, but may exhibit less contention.
-   */
-  public static Options useLocking(Boolean useLocking) {
-    return new Options().useLocking(useLocking);
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ResourceApplyProximalAdagrad";
-  
-  private ResourceApplyProximalAdagrad(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyProximalGradientDescent.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyProximalGradientDescent.java
deleted file mode 100644
index bf781ae05f2..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyProximalGradientDescent.java
+++ /dev/null
@@ -1,103 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.train;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Update '*var' as FOBOS algorithm with fixed learning rate.
- * <p>
- * prox_v = var - alpha <i> delta
- * var = sign(prox_v)/(1+alpha</i>l2) <i> max{|prox_v|-alpha</i>l1,0}
- */
-@Operator(group = "train")
-public final class ResourceApplyProximalGradientDescent extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.train.ResourceApplyProximalGradientDescent}
-   */
-  public static class Options {
-    
-    /**
-     * @param useLocking If True, the subtraction will be protected by a lock;
-     * otherwise the behavior is undefined, but may exhibit less contention.
-     */
-    public Options useLocking(Boolean useLocking) {
-      this.useLocking = useLocking;
-      return this;
-    }
-    
-    private Boolean useLocking;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new ResourceApplyProximalGradientDescent operation.
-   * 
-   * @param scope current scope
-   * @param var Should be from a Variable().
-   * @param alpha Scaling factor. Must be a scalar.
-   * @param l1 L1 regularization. Must be a scalar.
-   * @param l2 L2 regularization. Must be a scalar.
-   * @param delta The change.
-   * @param options carries optional attributes values
-   * @return a new instance of ResourceApplyProximalGradientDescent
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> ResourceApplyProximalGradientDescent create(Scope scope, Operand<?> var, Operand<T> alpha, Operand<T> l1, Operand<T> l2, Operand<T> delta, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ResourceApplyProximalGradientDescent", scope.makeOpName("ResourceApplyProximalGradientDescent"));
-    opBuilder.addInput(var.asOutput(scope));
-    opBuilder.addInput(alpha.asOutput(scope));
-    opBuilder.addInput(l1.asOutput(scope));
-    opBuilder.addInput(l2.asOutput(scope));
-    opBuilder.addInput(delta.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.useLocking != null) {
-          opBuilder.setAttr("use_locking", opts.useLocking);
-        }
-      }
-    }
-    return new ResourceApplyProximalGradientDescent(opBuilder.build());
-  }
-  
-  /**
-   * @param useLocking If True, the subtraction will be protected by a lock;
-   * otherwise the behavior is undefined, but may exhibit less contention.
-   */
-  public static Options useLocking(Boolean useLocking) {
-    return new Options().useLocking(useLocking);
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ResourceApplyProximalGradientDescent";
-  
-  private ResourceApplyProximalGradientDescent(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyRmsProp.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyRmsProp.java
deleted file mode 100644
index 46ec2ea3ca3..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyRmsProp.java
+++ /dev/null
@@ -1,119 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.train;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Update '*var' according to the RMSProp algorithm.
- * <p>
- * Note that in dense implementation of this algorithm, ms and mom will
- * update even if the grad is zero, but in this sparse implementation, ms
- * and mom will not update in iterations during which the grad is zero.
- * <p>
- * mean_square = decay * mean_square + (1-decay) * gradient ** 2
- * Delta = learning_rate * gradient / sqrt(mean_square + epsilon)
- * <p>
- * ms <- rho * ms_{t-1} + (1-rho) * grad * grad
- * mom <- momentum * mom_{t-1} + lr * grad / sqrt(ms + epsilon)
- * var <- var - mom
- */
-@Operator(group = "train")
-public final class ResourceApplyRmsProp extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.train.ResourceApplyRmsProp}
-   */
-  public static class Options {
-    
-    /**
-     * @param useLocking If `True`, updating of the var, ms, and mom tensors is protected
-     * by a lock; otherwise the behavior is undefined, but may exhibit less
-     * contention.
-     */
-    public Options useLocking(Boolean useLocking) {
-      this.useLocking = useLocking;
-      return this;
-    }
-    
-    private Boolean useLocking;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new ResourceApplyRmsProp operation.
-   * 
-   * @param scope current scope
-   * @param var Should be from a Variable().
-   * @param ms Should be from a Variable().
-   * @param mom Should be from a Variable().
-   * @param lr Scaling factor. Must be a scalar.
-   * @param rho Decay rate. Must be a scalar.
-   * @param momentum 
-   * @param epsilon Ridge term. Must be a scalar.
-   * @param grad The gradient.
-   * @param options carries optional attributes values
-   * @return a new instance of ResourceApplyRmsProp
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> ResourceApplyRmsProp create(Scope scope, Operand<?> var, Operand<?> ms, Operand<?> mom, Operand<T> lr, Operand<T> rho, Operand<T> momentum, Operand<T> epsilon, Operand<T> grad, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ResourceApplyRMSProp", scope.makeOpName("ResourceApplyRmsProp"));
-    opBuilder.addInput(var.asOutput(scope));
-    opBuilder.addInput(ms.asOutput(scope));
-    opBuilder.addInput(mom.asOutput(scope));
-    opBuilder.addInput(lr.asOutput(scope));
-    opBuilder.addInput(rho.asOutput(scope));
-    opBuilder.addInput(momentum.asOutput(scope));
-    opBuilder.addInput(epsilon.asOutput(scope));
-    opBuilder.addInput(grad.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.useLocking != null) {
-          opBuilder.setAttr("use_locking", opts.useLocking);
-        }
-      }
-    }
-    return new ResourceApplyRmsProp(opBuilder.build());
-  }
-  
-  /**
-   * @param useLocking If `True`, updating of the var, ms, and mom tensors is protected
-   * by a lock; otherwise the behavior is undefined, but may exhibit less
-   * contention.
-   */
-  public static Options useLocking(Boolean useLocking) {
-    return new Options().useLocking(useLocking);
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ResourceApplyRMSProp";
-  
-  private ResourceApplyRmsProp(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceConditionalAccumulator.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceConditionalAccumulator.java
deleted file mode 100644
index a23d0c850e1..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceConditionalAccumulator.java
+++ /dev/null
@@ -1,161 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.train;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * A conditional accumulator for aggregating gradients.
- * <p>
- * The accumulator accepts gradients marked with local_step greater or
- * equal to the most recent global_step known to the accumulator. The
- * average can be extracted from the accumulator, provided sufficient
- * gradients have been accumulated. Extracting the average automatically
- * resets the aggregate to 0, and increments the global_step recorded by
- * the accumulator.
- * This is a resource version of ConditionalAccumulator that will work in TF2.0
- * with tf.cond version 2.
- */
-public final class ResourceConditionalAccumulator extends RawOp implements Operand<TType> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.train.ResourceConditionalAccumulator}
-   */
-  public static class Options {
-    
-    /**
-     * @param container If non-empty, this accumulator is placed in the given container.
-     * Otherwise, a default container is used.
-     */
-    public Options container(String container) {
-      this.container = container;
-      return this;
-    }
-    
-    /**
-     * @param sharedName If non-empty, this accumulator will be shared under the
-     * given name across multiple sessions.
-     */
-    public Options sharedName(String sharedName) {
-      this.sharedName = sharedName;
-      return this;
-    }
-    
-    /**
-     * @param reductionType 
-     */
-    public Options reductionType(String reductionType) {
-      this.reductionType = reductionType;
-      return this;
-    }
-    
-    private String container;
-    private String sharedName;
-    private String reductionType;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new ResourceConditionalAccumulator operation.
-   * 
-   * @param scope current scope
-   * @param dtype The type of the value being accumulated.
-   * @param shape The shape of the values, can be [], in which case shape is unknown.
-   * @param options carries optional attributes values
-   * @return a new instance of ResourceConditionalAccumulator
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> ResourceConditionalAccumulator create(Scope scope, Class<T> dtype, Shape shape, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ResourceConditionalAccumulator", scope.makeOpName("ResourceConditionalAccumulator"));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("dtype", dtype);
-    opBuilder.setAttr("shape", shape);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.container != null) {
-          opBuilder.setAttr("container", opts.container);
-        }
-        if (opts.sharedName != null) {
-          opBuilder.setAttr("shared_name", opts.sharedName);
-        }
-        if (opts.reductionType != null) {
-          opBuilder.setAttr("reduction_type", opts.reductionType);
-        }
-      }
-    }
-    return new ResourceConditionalAccumulator(opBuilder.build());
-  }
-  
-  /**
-   * @param container If non-empty, this accumulator is placed in the given container.
-   * Otherwise, a default container is used.
-   */
-  public static Options container(String container) {
-    return new Options().container(container);
-  }
-  
-  /**
-   * @param sharedName If non-empty, this accumulator will be shared under the
-   * given name across multiple sessions.
-   */
-  public static Options sharedName(String sharedName) {
-    return new Options().sharedName(sharedName);
-  }
-  
-  /**
-   * @param reductionType 
-   */
-  public static Options reductionType(String reductionType) {
-    return new Options().reductionType(reductionType);
-  }
-  
-  /**
-   * The handle to the accumulator.
-   */
-  public Output<?> handle() {
-    return handle;
-  }
-  
-  @Override
-  @SuppressWarnings("unchecked")
-  public Output<TType> asOutput(Scope scope) {
-    return (Output<TType>) handle;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ResourceConditionalAccumulator";
-  
-  private Output<?> handle;
-  
-  private ResourceConditionalAccumulator(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    handle = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyAdadelta.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyAdadelta.java
deleted file mode 100644
index 41813e0faac..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyAdadelta.java
+++ /dev/null
@@ -1,107 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.train;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * var: Should be from a Variable().
- */
-@Operator(group = "train")
-public final class ResourceSparseApplyAdadelta extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.train.ResourceSparseApplyAdadelta}
-   */
-  public static class Options {
-    
-    /**
-     * @param useLocking If True, updating of the var and accum tensors will be protected by
-     * a lock; otherwise the behavior is undefined, but may exhibit less contention.
-     */
-    public Options useLocking(Boolean useLocking) {
-      this.useLocking = useLocking;
-      return this;
-    }
-    
-    private Boolean useLocking;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new ResourceSparseApplyAdadelta operation.
-   * 
-   * @param scope current scope
-   * @param var 
-   * @param accum Should be from a Variable().
-   * @param accumUpdate : Should be from a Variable().
-   * @param lr Learning rate. Must be a scalar.
-   * @param rho Decay factor. Must be a scalar.
-   * @param epsilon Constant factor. Must be a scalar.
-   * @param grad The gradient.
-   * @param indices A vector of indices into the first dimension of var and accum.
-   * @param options carries optional attributes values
-   * @return a new instance of ResourceSparseApplyAdadelta
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> ResourceSparseApplyAdadelta create(Scope scope, Operand<?> var, Operand<?> accum, Operand<?> accumUpdate, Operand<T> lr, Operand<T> rho, Operand<T> epsilon, Operand<T> grad, Operand<U> indices, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ResourceSparseApplyAdadelta", scope.makeOpName("ResourceSparseApplyAdadelta"));
-    opBuilder.addInput(var.asOutput(scope));
-    opBuilder.addInput(accum.asOutput(scope));
-    opBuilder.addInput(accumUpdate.asOutput(scope));
-    opBuilder.addInput(lr.asOutput(scope));
-    opBuilder.addInput(rho.asOutput(scope));
-    opBuilder.addInput(epsilon.asOutput(scope));
-    opBuilder.addInput(grad.asOutput(scope));
-    opBuilder.addInput(indices.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.useLocking != null) {
-          opBuilder.setAttr("use_locking", opts.useLocking);
-        }
-      }
-    }
-    return new ResourceSparseApplyAdadelta(opBuilder.build());
-  }
-  
-  /**
-   * @param useLocking If True, updating of the var and accum tensors will be protected by
-   * a lock; otherwise the behavior is undefined, but may exhibit less contention.
-   */
-  public static Options useLocking(Boolean useLocking) {
-    return new Options().useLocking(useLocking);
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ResourceSparseApplyAdadelta";
-  
-  private ResourceSparseApplyAdadelta(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyAdagrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyAdagrad.java
deleted file mode 100644
index 68a3a21b4a5..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyAdagrad.java
+++ /dev/null
@@ -1,126 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.train;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * Update relevant entries in '*var' and '*accum' according to the adagrad scheme.
- * <p>
- * That is for rows we have grad for, we update var and accum as follows:
- * accum += grad * grad
- * var -= lr * grad * (1 / sqrt(accum))
- */
-@Operator(group = "train")
-public final class ResourceSparseApplyAdagrad extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.train.ResourceSparseApplyAdagrad}
-   */
-  public static class Options {
-    
-    /**
-     * @param useLocking If `True`, updating of the var and accum tensors will be protected
-     * by a lock; otherwise the behavior is undefined, but may exhibit less
-     * contention.
-     */
-    public Options useLocking(Boolean useLocking) {
-      this.useLocking = useLocking;
-      return this;
-    }
-    
-    /**
-     * @param updateSlots 
-     */
-    public Options updateSlots(Boolean updateSlots) {
-      this.updateSlots = updateSlots;
-      return this;
-    }
-    
-    private Boolean useLocking;
-    private Boolean updateSlots;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new ResourceSparseApplyAdagrad operation.
-   * 
-   * @param scope current scope
-   * @param var Should be from a Variable().
-   * @param accum Should be from a Variable().
-   * @param lr Learning rate. Must be a scalar.
-   * @param grad The gradient.
-   * @param indices A vector of indices into the first dimension of var and accum.
-   * @param options carries optional attributes values
-   * @return a new instance of ResourceSparseApplyAdagrad
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> ResourceSparseApplyAdagrad create(Scope scope, Operand<?> var, Operand<?> accum, Operand<T> lr, Operand<T> grad, Operand<U> indices, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ResourceSparseApplyAdagrad", scope.makeOpName("ResourceSparseApplyAdagrad"));
-    opBuilder.addInput(var.asOutput(scope));
-    opBuilder.addInput(accum.asOutput(scope));
-    opBuilder.addInput(lr.asOutput(scope));
-    opBuilder.addInput(grad.asOutput(scope));
-    opBuilder.addInput(indices.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.useLocking != null) {
-          opBuilder.setAttr("use_locking", opts.useLocking);
-        }
-        if (opts.updateSlots != null) {
-          opBuilder.setAttr("update_slots", opts.updateSlots);
-        }
-      }
-    }
-    return new ResourceSparseApplyAdagrad(opBuilder.build());
-  }
-  
-  /**
-   * @param useLocking If `True`, updating of the var and accum tensors will be protected
-   * by a lock; otherwise the behavior is undefined, but may exhibit less
-   * contention.
-   */
-  public static Options useLocking(Boolean useLocking) {
-    return new Options().useLocking(useLocking);
-  }
-  
-  /**
-   * @param updateSlots 
-   */
-  public static Options updateSlots(Boolean updateSlots) {
-    return new Options().updateSlots(updateSlots);
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ResourceSparseApplyAdagrad";
-  
-  private ResourceSparseApplyAdagrad(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyAdagradDa.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyAdagradDa.java
deleted file mode 100644
index 760e2cf1b47..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyAdagradDa.java
+++ /dev/null
@@ -1,110 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.train;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * Update entries in '*var' and '*accum' according to the proximal adagrad scheme.
- */
-@Operator(group = "train")
-public final class ResourceSparseApplyAdagradDa extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.train.ResourceSparseApplyAdagradDa}
-   */
-  public static class Options {
-    
-    /**
-     * @param useLocking If True, updating of the var and accum tensors will be protected by
-     * a lock; otherwise the behavior is undefined, but may exhibit less contention.
-     */
-    public Options useLocking(Boolean useLocking) {
-      this.useLocking = useLocking;
-      return this;
-    }
-    
-    private Boolean useLocking;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new ResourceSparseApplyAdagradDa operation.
-   * 
-   * @param scope current scope
-   * @param var Should be from a Variable().
-   * @param gradientAccumulator Should be from a Variable().
-   * @param gradientSquaredAccumulator Should be from a Variable().
-   * @param grad The gradient.
-   * @param indices A vector of indices into the first dimension of var and accum.
-   * @param lr Learning rate. Must be a scalar.
-   * @param l1 L1 regularization. Must be a scalar.
-   * @param l2 L2 regularization. Must be a scalar.
-   * @param globalStep Training step number. Must be a scalar.
-   * @param options carries optional attributes values
-   * @return a new instance of ResourceSparseApplyAdagradDa
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> ResourceSparseApplyAdagradDa create(Scope scope, Operand<?> var, Operand<?> gradientAccumulator, Operand<?> gradientSquaredAccumulator, Operand<T> grad, Operand<U> indices, Operand<T> lr, Operand<T> l1, Operand<T> l2, Operand<TInt64> globalStep, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ResourceSparseApplyAdagradDA", scope.makeOpName("ResourceSparseApplyAdagradDa"));
-    opBuilder.addInput(var.asOutput(scope));
-    opBuilder.addInput(gradientAccumulator.asOutput(scope));
-    opBuilder.addInput(gradientSquaredAccumulator.asOutput(scope));
-    opBuilder.addInput(grad.asOutput(scope));
-    opBuilder.addInput(indices.asOutput(scope));
-    opBuilder.addInput(lr.asOutput(scope));
-    opBuilder.addInput(l1.asOutput(scope));
-    opBuilder.addInput(l2.asOutput(scope));
-    opBuilder.addInput(globalStep.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.useLocking != null) {
-          opBuilder.setAttr("use_locking", opts.useLocking);
-        }
-      }
-    }
-    return new ResourceSparseApplyAdagradDa(opBuilder.build());
-  }
-  
-  /**
-   * @param useLocking If True, updating of the var and accum tensors will be protected by
-   * a lock; otherwise the behavior is undefined, but may exhibit less contention.
-   */
-  public static Options useLocking(Boolean useLocking) {
-    return new Options().useLocking(useLocking);
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ResourceSparseApplyAdagradDA";
-  
-  private ResourceSparseApplyAdagradDa(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyAdagradV2.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyAdagradV2.java
deleted file mode 100644
index 7572b71cd56..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyAdagradV2.java
+++ /dev/null
@@ -1,127 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.train;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * Update relevant entries in '*var' and '*accum' according to the adagrad scheme.
- * <p>
- * That is for rows we have grad for, we update var and accum as follows:
- * accum += grad * grad
- * var -= lr * grad * (1 / sqrt(accum))
- */
-public final class ResourceSparseApplyAdagradV2 extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.train.ResourceSparseApplyAdagradV2}
-   */
-  public static class Options {
-    
-    /**
-     * @param useLocking If `True`, updating of the var and accum tensors will be protected
-     * by a lock; otherwise the behavior is undefined, but may exhibit less
-     * contention.
-     */
-    public Options useLocking(Boolean useLocking) {
-      this.useLocking = useLocking;
-      return this;
-    }
-    
-    /**
-     * @param updateSlots 
-     */
-    public Options updateSlots(Boolean updateSlots) {
-      this.updateSlots = updateSlots;
-      return this;
-    }
-    
-    private Boolean useLocking;
-    private Boolean updateSlots;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new ResourceSparseApplyAdagradV2 operation.
-   * 
-   * @param scope current scope
-   * @param var Should be from a Variable().
-   * @param accum Should be from a Variable().
-   * @param lr Learning rate. Must be a scalar.
-   * @param epsilon Constant factor. Must be a scalar.
-   * @param grad The gradient.
-   * @param indices A vector of indices into the first dimension of var and accum.
-   * @param options carries optional attributes values
-   * @return a new instance of ResourceSparseApplyAdagradV2
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> ResourceSparseApplyAdagradV2 create(Scope scope, Operand<?> var, Operand<?> accum, Operand<T> lr, Operand<T> epsilon, Operand<T> grad, Operand<U> indices, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ResourceSparseApplyAdagradV2", scope.makeOpName("ResourceSparseApplyAdagradV2"));
-    opBuilder.addInput(var.asOutput(scope));
-    opBuilder.addInput(accum.asOutput(scope));
-    opBuilder.addInput(lr.asOutput(scope));
-    opBuilder.addInput(epsilon.asOutput(scope));
-    opBuilder.addInput(grad.asOutput(scope));
-    opBuilder.addInput(indices.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.useLocking != null) {
-          opBuilder.setAttr("use_locking", opts.useLocking);
-        }
-        if (opts.updateSlots != null) {
-          opBuilder.setAttr("update_slots", opts.updateSlots);
-        }
-      }
-    }
-    return new ResourceSparseApplyAdagradV2(opBuilder.build());
-  }
-  
-  /**
-   * @param useLocking If `True`, updating of the var and accum tensors will be protected
-   * by a lock; otherwise the behavior is undefined, but may exhibit less
-   * contention.
-   */
-  public static Options useLocking(Boolean useLocking) {
-    return new Options().useLocking(useLocking);
-  }
-  
-  /**
-   * @param updateSlots 
-   */
-  public static Options updateSlots(Boolean updateSlots) {
-    return new Options().updateSlots(updateSlots);
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ResourceSparseApplyAdagradV2";
-  
-  private ResourceSparseApplyAdagradV2(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyCenteredRmsProp.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyCenteredRmsProp.java
deleted file mode 100644
index a1015cadb11..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyCenteredRmsProp.java
+++ /dev/null
@@ -1,130 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.train;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * Update '*var' according to the centered RMSProp algorithm.
- * <p>
- * The centered RMSProp algorithm uses an estimate of the centered second moment
- * (i.e., the variance) for normalization, as opposed to regular RMSProp, which
- * uses the (uncentered) second moment. This often helps with training, but is
- * slightly more expensive in terms of computation and memory.
- * <p>
- * Note that in dense implementation of this algorithm, mg, ms, and mom will
- * update even if the grad is zero, but in this sparse implementation, mg, ms,
- * and mom will not update in iterations during which the grad is zero.
- * <p>
- * mean_square = decay * mean_square + (1-decay) * gradient ** 2
- * mean_grad = decay * mean_grad + (1-decay) * gradient
- * Delta = learning_rate * gradient / sqrt(mean_square + epsilon - mean_grad ** 2)
- * <p>
- * ms <- rho * ms_{t-1} + (1-rho) * grad * grad
- * mom <- momentum * mom_{t-1} + lr * grad / sqrt(ms + epsilon)
- * var <- var - mom
- */
-@Operator(group = "train")
-public final class ResourceSparseApplyCenteredRmsProp extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.train.ResourceSparseApplyCenteredRmsProp}
-   */
-  public static class Options {
-    
-    /**
-     * @param useLocking If `True`, updating of the var, mg, ms, and mom tensors is
-     * protected by a lock; otherwise the behavior is undefined, but may exhibit less
-     * contention.
-     */
-    public Options useLocking(Boolean useLocking) {
-      this.useLocking = useLocking;
-      return this;
-    }
-    
-    private Boolean useLocking;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new ResourceSparseApplyCenteredRmsProp operation.
-   * 
-   * @param scope current scope
-   * @param var Should be from a Variable().
-   * @param mg Should be from a Variable().
-   * @param ms Should be from a Variable().
-   * @param mom Should be from a Variable().
-   * @param lr Scaling factor. Must be a scalar.
-   * @param rho Decay rate. Must be a scalar.
-   * @param momentum 
-   * @param epsilon Ridge term. Must be a scalar.
-   * @param grad The gradient.
-   * @param indices A vector of indices into the first dimension of var, ms and mom.
-   * @param options carries optional attributes values
-   * @return a new instance of ResourceSparseApplyCenteredRmsProp
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> ResourceSparseApplyCenteredRmsProp create(Scope scope, Operand<?> var, Operand<?> mg, Operand<?> ms, Operand<?> mom, Operand<T> lr, Operand<T> rho, Operand<T> momentum, Operand<T> epsilon, Operand<T> grad, Operand<U> indices, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ResourceSparseApplyCenteredRMSProp", scope.makeOpName("ResourceSparseApplyCenteredRmsProp"));
-    opBuilder.addInput(var.asOutput(scope));
-    opBuilder.addInput(mg.asOutput(scope));
-    opBuilder.addInput(ms.asOutput(scope));
-    opBuilder.addInput(mom.asOutput(scope));
-    opBuilder.addInput(lr.asOutput(scope));
-    opBuilder.addInput(rho.asOutput(scope));
-    opBuilder.addInput(momentum.asOutput(scope));
-    opBuilder.addInput(epsilon.asOutput(scope));
-    opBuilder.addInput(grad.asOutput(scope));
-    opBuilder.addInput(indices.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.useLocking != null) {
-          opBuilder.setAttr("use_locking", opts.useLocking);
-        }
-      }
-    }
-    return new ResourceSparseApplyCenteredRmsProp(opBuilder.build());
-  }
-  
-  /**
-   * @param useLocking If `True`, updating of the var, mg, ms, and mom tensors is
-   * protected by a lock; otherwise the behavior is undefined, but may exhibit less
-   * contention.
-   */
-  public static Options useLocking(Boolean useLocking) {
-    return new Options().useLocking(useLocking);
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ResourceSparseApplyCenteredRMSProp";
-  
-  private ResourceSparseApplyCenteredRmsProp(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyFtrl.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyFtrl.java
deleted file mode 100644
index a04ab7aab5b..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyFtrl.java
+++ /dev/null
@@ -1,141 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.train;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * Update relevant entries in '*var' according to the Ftrl-proximal scheme.
- * <p>
- * That is for rows we have grad for, we update var, accum and linear as follows:
- * grad_with_shrinkage = grad + 2 * l2_shrinkage * var
- * accum_new = accum + grad_with_shrinkage * grad_with_shrinkage
- * linear += grad_with_shrinkage +
- *     (accum_new^(-lr_power) - accum^(-lr_power)) / lr * var
- * quadratic = 1.0 / (accum_new^(lr_power) * lr) + 2 * l2
- * var = (sign(linear) * l1 - linear) / quadratic if |linear| > l1 else 0.0
- * accum = accum_new
- */
-@Operator(group = "train")
-public final class ResourceSparseApplyFtrl extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.train.ResourceSparseApplyFtrl}
-   */
-  public static class Options {
-    
-    /**
-     * @param useLocking If `True`, updating of the var and accum tensors will be protected
-     * by a lock; otherwise the behavior is undefined, but may exhibit less
-     * contention.
-     */
-    public Options useLocking(Boolean useLocking) {
-      this.useLocking = useLocking;
-      return this;
-    }
-    
-    /**
-     * @param multiplyLinearByLr 
-     */
-    public Options multiplyLinearByLr(Boolean multiplyLinearByLr) {
-      this.multiplyLinearByLr = multiplyLinearByLr;
-      return this;
-    }
-    
-    private Boolean useLocking;
-    private Boolean multiplyLinearByLr;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new ResourceSparseApplyFtrl operation.
-   * 
-   * @param scope current scope
-   * @param var Should be from a Variable().
-   * @param accum Should be from a Variable().
-   * @param linear Should be from a Variable().
-   * @param grad The gradient.
-   * @param indices A vector of indices into the first dimension of var and accum.
-   * @param lr Scaling factor. Must be a scalar.
-   * @param l1 L1 regularization. Must be a scalar.
-   * @param l2 L2 shrinkage regularization. Must be a scalar.
-   * @param l2Shrinkage 
-   * @param lrPower Scaling factor. Must be a scalar.
-   * @param options carries optional attributes values
-   * @return a new instance of ResourceSparseApplyFtrl
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> ResourceSparseApplyFtrl create(Scope scope, Operand<?> var, Operand<?> accum, Operand<?> linear, Operand<T> grad, Operand<U> indices, Operand<T> lr, Operand<T> l1, Operand<T> l2, Operand<T> l2Shrinkage, Operand<T> lrPower, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ResourceSparseApplyFtrlV2", scope.makeOpName("ResourceSparseApplyFtrl"));
-    opBuilder.addInput(var.asOutput(scope));
-    opBuilder.addInput(accum.asOutput(scope));
-    opBuilder.addInput(linear.asOutput(scope));
-    opBuilder.addInput(grad.asOutput(scope));
-    opBuilder.addInput(indices.asOutput(scope));
-    opBuilder.addInput(lr.asOutput(scope));
-    opBuilder.addInput(l1.asOutput(scope));
-    opBuilder.addInput(l2.asOutput(scope));
-    opBuilder.addInput(l2Shrinkage.asOutput(scope));
-    opBuilder.addInput(lrPower.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.useLocking != null) {
-          opBuilder.setAttr("use_locking", opts.useLocking);
-        }
-        if (opts.multiplyLinearByLr != null) {
-          opBuilder.setAttr("multiply_linear_by_lr", opts.multiplyLinearByLr);
-        }
-      }
-    }
-    return new ResourceSparseApplyFtrl(opBuilder.build());
-  }
-  
-  /**
-   * @param useLocking If `True`, updating of the var and accum tensors will be protected
-   * by a lock; otherwise the behavior is undefined, but may exhibit less
-   * contention.
-   */
-  public static Options useLocking(Boolean useLocking) {
-    return new Options().useLocking(useLocking);
-  }
-  
-  /**
-   * @param multiplyLinearByLr 
-   */
-  public static Options multiplyLinearByLr(Boolean multiplyLinearByLr) {
-    return new Options().multiplyLinearByLr(multiplyLinearByLr);
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ResourceSparseApplyFtrlV2";
-  
-  private ResourceSparseApplyFtrl(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyKerasMomentum.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyKerasMomentum.java
deleted file mode 100644
index 05b6d72816c..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyKerasMomentum.java
+++ /dev/null
@@ -1,135 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.train;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * Update relevant entries in '*var' and '*accum' according to the momentum scheme.
- * <p>
- * Set use_nesterov = True if you want to use Nesterov momentum.
- * <p>
- * That is for rows we have grad for, we update var and accum as follows:
- * <p>
- * accum = accum * momentum - lr * grad
- * var += accum
- */
-@Operator(group = "train")
-public final class ResourceSparseApplyKerasMomentum extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.train.ResourceSparseApplyKerasMomentum}
-   */
-  public static class Options {
-    
-    /**
-     * @param useLocking If `True`, updating of the var and accum tensors will be protected
-     * by a lock; otherwise the behavior is undefined, but may exhibit less
-     * contention.
-     */
-    public Options useLocking(Boolean useLocking) {
-      this.useLocking = useLocking;
-      return this;
-    }
-    
-    /**
-     * @param useNesterov If `True`, the tensor passed to compute grad will be
-     * var + momentum * accum, so in the end, the var you get is actually
-     * var + momentum * accum.
-     */
-    public Options useNesterov(Boolean useNesterov) {
-      this.useNesterov = useNesterov;
-      return this;
-    }
-    
-    private Boolean useLocking;
-    private Boolean useNesterov;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new ResourceSparseApplyKerasMomentum operation.
-   * 
-   * @param scope current scope
-   * @param var Should be from a Variable().
-   * @param accum Should be from a Variable().
-   * @param lr Learning rate. Must be a scalar.
-   * @param grad The gradient.
-   * @param indices A vector of indices into the first dimension of var and accum.
-   * @param momentum Momentum. Must be a scalar.
-   * @param options carries optional attributes values
-   * @return a new instance of ResourceSparseApplyKerasMomentum
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> ResourceSparseApplyKerasMomentum create(Scope scope, Operand<?> var, Operand<?> accum, Operand<T> lr, Operand<T> grad, Operand<U> indices, Operand<T> momentum, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ResourceSparseApplyKerasMomentum", scope.makeOpName("ResourceSparseApplyKerasMomentum"));
-    opBuilder.addInput(var.asOutput(scope));
-    opBuilder.addInput(accum.asOutput(scope));
-    opBuilder.addInput(lr.asOutput(scope));
-    opBuilder.addInput(grad.asOutput(scope));
-    opBuilder.addInput(indices.asOutput(scope));
-    opBuilder.addInput(momentum.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.useLocking != null) {
-          opBuilder.setAttr("use_locking", opts.useLocking);
-        }
-        if (opts.useNesterov != null) {
-          opBuilder.setAttr("use_nesterov", opts.useNesterov);
-        }
-      }
-    }
-    return new ResourceSparseApplyKerasMomentum(opBuilder.build());
-  }
-  
-  /**
-   * @param useLocking If `True`, updating of the var and accum tensors will be protected
-   * by a lock; otherwise the behavior is undefined, but may exhibit less
-   * contention.
-   */
-  public static Options useLocking(Boolean useLocking) {
-    return new Options().useLocking(useLocking);
-  }
-  
-  /**
-   * @param useNesterov If `True`, the tensor passed to compute grad will be
-   * var + momentum * accum, so in the end, the var you get is actually
-   * var + momentum * accum.
-   */
-  public static Options useNesterov(Boolean useNesterov) {
-    return new Options().useNesterov(useNesterov);
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ResourceSparseApplyKerasMomentum";
-  
-  private ResourceSparseApplyKerasMomentum(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyMomentum.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyMomentum.java
deleted file mode 100644
index 160ddf01f60..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyMomentum.java
+++ /dev/null
@@ -1,135 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.train;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * Update relevant entries in '*var' and '*accum' according to the momentum scheme.
- * <p>
- * Set use_nesterov = True if you want to use Nesterov momentum.
- * <p>
- * That is for rows we have grad for, we update var and accum as follows:
- * <p>
- * accum = accum * momentum + grad
- * var -= lr * accum
- */
-@Operator(group = "train")
-public final class ResourceSparseApplyMomentum extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.train.ResourceSparseApplyMomentum}
-   */
-  public static class Options {
-    
-    /**
-     * @param useLocking If `True`, updating of the var and accum tensors will be protected
-     * by a lock; otherwise the behavior is undefined, but may exhibit less
-     * contention.
-     */
-    public Options useLocking(Boolean useLocking) {
-      this.useLocking = useLocking;
-      return this;
-    }
-    
-    /**
-     * @param useNesterov If `True`, the tensor passed to compute grad will be
-     * var - lr * momentum * accum, so in the end, the var you get is actually
-     * var - lr * momentum * accum.
-     */
-    public Options useNesterov(Boolean useNesterov) {
-      this.useNesterov = useNesterov;
-      return this;
-    }
-    
-    private Boolean useLocking;
-    private Boolean useNesterov;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new ResourceSparseApplyMomentum operation.
-   * 
-   * @param scope current scope
-   * @param var Should be from a Variable().
-   * @param accum Should be from a Variable().
-   * @param lr Learning rate. Must be a scalar.
-   * @param grad The gradient.
-   * @param indices A vector of indices into the first dimension of var and accum.
-   * @param momentum Momentum. Must be a scalar.
-   * @param options carries optional attributes values
-   * @return a new instance of ResourceSparseApplyMomentum
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> ResourceSparseApplyMomentum create(Scope scope, Operand<?> var, Operand<?> accum, Operand<T> lr, Operand<T> grad, Operand<U> indices, Operand<T> momentum, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ResourceSparseApplyMomentum", scope.makeOpName("ResourceSparseApplyMomentum"));
-    opBuilder.addInput(var.asOutput(scope));
-    opBuilder.addInput(accum.asOutput(scope));
-    opBuilder.addInput(lr.asOutput(scope));
-    opBuilder.addInput(grad.asOutput(scope));
-    opBuilder.addInput(indices.asOutput(scope));
-    opBuilder.addInput(momentum.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.useLocking != null) {
-          opBuilder.setAttr("use_locking", opts.useLocking);
-        }
-        if (opts.useNesterov != null) {
-          opBuilder.setAttr("use_nesterov", opts.useNesterov);
-        }
-      }
-    }
-    return new ResourceSparseApplyMomentum(opBuilder.build());
-  }
-  
-  /**
-   * @param useLocking If `True`, updating of the var and accum tensors will be protected
-   * by a lock; otherwise the behavior is undefined, but may exhibit less
-   * contention.
-   */
-  public static Options useLocking(Boolean useLocking) {
-    return new Options().useLocking(useLocking);
-  }
-  
-  /**
-   * @param useNesterov If `True`, the tensor passed to compute grad will be
-   * var - lr * momentum * accum, so in the end, the var you get is actually
-   * var - lr * momentum * accum.
-   */
-  public static Options useNesterov(Boolean useNesterov) {
-    return new Options().useNesterov(useNesterov);
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ResourceSparseApplyMomentum";
-  
-  private ResourceSparseApplyMomentum(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyProximalAdagrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyProximalAdagrad.java
deleted file mode 100644
index af51baac80e..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyProximalAdagrad.java
+++ /dev/null
@@ -1,111 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.train;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * Sparse update entries in '*var' and '*accum' according to FOBOS algorithm.
- * <p>
- * That is for rows we have grad for, we update var and accum as follows:
- * accum += grad <i> grad
- * prox_v = var
- * prox_v -= lr </i> grad <i> (1 / sqrt(accum))
- * var = sign(prox_v)/(1+lr</i>l2) <i> max{|prox_v|-lr</i>l1,0}
- */
-@Operator(group = "train")
-public final class ResourceSparseApplyProximalAdagrad extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.train.ResourceSparseApplyProximalAdagrad}
-   */
-  public static class Options {
-    
-    /**
-     * @param useLocking If True, updating of the var and accum tensors will be protected by
-     * a lock; otherwise the behavior is undefined, but may exhibit less contention.
-     */
-    public Options useLocking(Boolean useLocking) {
-      this.useLocking = useLocking;
-      return this;
-    }
-    
-    private Boolean useLocking;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new ResourceSparseApplyProximalAdagrad operation.
-   * 
-   * @param scope current scope
-   * @param var Should be from a Variable().
-   * @param accum Should be from a Variable().
-   * @param lr Learning rate. Must be a scalar.
-   * @param l1 L1 regularization. Must be a scalar.
-   * @param l2 L2 regularization. Must be a scalar.
-   * @param grad The gradient.
-   * @param indices A vector of indices into the first dimension of var and accum.
-   * @param options carries optional attributes values
-   * @return a new instance of ResourceSparseApplyProximalAdagrad
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> ResourceSparseApplyProximalAdagrad create(Scope scope, Operand<?> var, Operand<?> accum, Operand<T> lr, Operand<T> l1, Operand<T> l2, Operand<T> grad, Operand<U> indices, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ResourceSparseApplyProximalAdagrad", scope.makeOpName("ResourceSparseApplyProximalAdagrad"));
-    opBuilder.addInput(var.asOutput(scope));
-    opBuilder.addInput(accum.asOutput(scope));
-    opBuilder.addInput(lr.asOutput(scope));
-    opBuilder.addInput(l1.asOutput(scope));
-    opBuilder.addInput(l2.asOutput(scope));
-    opBuilder.addInput(grad.asOutput(scope));
-    opBuilder.addInput(indices.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.useLocking != null) {
-          opBuilder.setAttr("use_locking", opts.useLocking);
-        }
-      }
-    }
-    return new ResourceSparseApplyProximalAdagrad(opBuilder.build());
-  }
-  
-  /**
-   * @param useLocking If True, updating of the var and accum tensors will be protected by
-   * a lock; otherwise the behavior is undefined, but may exhibit less contention.
-   */
-  public static Options useLocking(Boolean useLocking) {
-    return new Options().useLocking(useLocking);
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ResourceSparseApplyProximalAdagrad";
-  
-  private ResourceSparseApplyProximalAdagrad(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyProximalGradientDescent.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyProximalGradientDescent.java
deleted file mode 100644
index eda3f87707b..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyProximalGradientDescent.java
+++ /dev/null
@@ -1,107 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.train;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * Sparse update '*var' as FOBOS algorithm with fixed learning rate.
- * <p>
- * That is for rows we have grad for, we update var as follows:
- * prox_v = var - alpha <i> grad
- * var = sign(prox_v)/(1+alpha</i>l2) <i> max{|prox_v|-alpha</i>l1,0}
- */
-@Operator(group = "train")
-public final class ResourceSparseApplyProximalGradientDescent extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.train.ResourceSparseApplyProximalGradientDescent}
-   */
-  public static class Options {
-    
-    /**
-     * @param useLocking If True, the subtraction will be protected by a lock;
-     * otherwise the behavior is undefined, but may exhibit less contention.
-     */
-    public Options useLocking(Boolean useLocking) {
-      this.useLocking = useLocking;
-      return this;
-    }
-    
-    private Boolean useLocking;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new ResourceSparseApplyProximalGradientDescent operation.
-   * 
-   * @param scope current scope
-   * @param var Should be from a Variable().
-   * @param alpha Scaling factor. Must be a scalar.
-   * @param l1 L1 regularization. Must be a scalar.
-   * @param l2 L2 regularization. Must be a scalar.
-   * @param grad The gradient.
-   * @param indices A vector of indices into the first dimension of var and accum.
-   * @param options carries optional attributes values
-   * @return a new instance of ResourceSparseApplyProximalGradientDescent
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> ResourceSparseApplyProximalGradientDescent create(Scope scope, Operand<?> var, Operand<T> alpha, Operand<T> l1, Operand<T> l2, Operand<T> grad, Operand<U> indices, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ResourceSparseApplyProximalGradientDescent", scope.makeOpName("ResourceSparseApplyProximalGradientDescent"));
-    opBuilder.addInput(var.asOutput(scope));
-    opBuilder.addInput(alpha.asOutput(scope));
-    opBuilder.addInput(l1.asOutput(scope));
-    opBuilder.addInput(l2.asOutput(scope));
-    opBuilder.addInput(grad.asOutput(scope));
-    opBuilder.addInput(indices.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.useLocking != null) {
-          opBuilder.setAttr("use_locking", opts.useLocking);
-        }
-      }
-    }
-    return new ResourceSparseApplyProximalGradientDescent(opBuilder.build());
-  }
-  
-  /**
-   * @param useLocking If True, the subtraction will be protected by a lock;
-   * otherwise the behavior is undefined, but may exhibit less contention.
-   */
-  public static Options useLocking(Boolean useLocking) {
-    return new Options().useLocking(useLocking);
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ResourceSparseApplyProximalGradientDescent";
-  
-  private ResourceSparseApplyProximalGradientDescent(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyRmsProp.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyRmsProp.java
deleted file mode 100644
index ae2678b3120..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyRmsProp.java
+++ /dev/null
@@ -1,122 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.train;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * Update '*var' according to the RMSProp algorithm.
- * <p>
- * Note that in dense implementation of this algorithm, ms and mom will
- * update even if the grad is zero, but in this sparse implementation, ms
- * and mom will not update in iterations during which the grad is zero.
- * <p>
- * mean_square = decay * mean_square + (1-decay) * gradient ** 2
- * Delta = learning_rate * gradient / sqrt(mean_square + epsilon)
- * <p>
- * ms <- rho * ms_{t-1} + (1-rho) * grad * grad
- * mom <- momentum * mom_{t-1} + lr * grad / sqrt(ms + epsilon)
- * var <- var - mom
- */
-@Operator(group = "train")
-public final class ResourceSparseApplyRmsProp extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.train.ResourceSparseApplyRmsProp}
-   */
-  public static class Options {
-    
-    /**
-     * @param useLocking If `True`, updating of the var, ms, and mom tensors is protected
-     * by a lock; otherwise the behavior is undefined, but may exhibit less
-     * contention.
-     */
-    public Options useLocking(Boolean useLocking) {
-      this.useLocking = useLocking;
-      return this;
-    }
-    
-    private Boolean useLocking;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new ResourceSparseApplyRmsProp operation.
-   * 
-   * @param scope current scope
-   * @param var Should be from a Variable().
-   * @param ms Should be from a Variable().
-   * @param mom Should be from a Variable().
-   * @param lr Scaling factor. Must be a scalar.
-   * @param rho Decay rate. Must be a scalar.
-   * @param momentum 
-   * @param epsilon Ridge term. Must be a scalar.
-   * @param grad The gradient.
-   * @param indices A vector of indices into the first dimension of var, ms and mom.
-   * @param options carries optional attributes values
-   * @return a new instance of ResourceSparseApplyRmsProp
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> ResourceSparseApplyRmsProp create(Scope scope, Operand<?> var, Operand<?> ms, Operand<?> mom, Operand<T> lr, Operand<T> rho, Operand<T> momentum, Operand<T> epsilon, Operand<T> grad, Operand<U> indices, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("ResourceSparseApplyRMSProp", scope.makeOpName("ResourceSparseApplyRmsProp"));
-    opBuilder.addInput(var.asOutput(scope));
-    opBuilder.addInput(ms.asOutput(scope));
-    opBuilder.addInput(mom.asOutput(scope));
-    opBuilder.addInput(lr.asOutput(scope));
-    opBuilder.addInput(rho.asOutput(scope));
-    opBuilder.addInput(momentum.asOutput(scope));
-    opBuilder.addInput(epsilon.asOutput(scope));
-    opBuilder.addInput(grad.asOutput(scope));
-    opBuilder.addInput(indices.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.useLocking != null) {
-          opBuilder.setAttr("use_locking", opts.useLocking);
-        }
-      }
-    }
-    return new ResourceSparseApplyRmsProp(opBuilder.build());
-  }
-  
-  /**
-   * @param useLocking If `True`, updating of the var, ms, and mom tensors is protected
-   * by a lock; otherwise the behavior is undefined, but may exhibit less
-   * contention.
-   */
-  public static Options useLocking(Boolean useLocking) {
-    return new Options().useLocking(useLocking);
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "ResourceSparseApplyRMSProp";
-  
-  private ResourceSparseApplyRmsProp(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/Restore.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/Restore.java
deleted file mode 100644
index b9d131d107b..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/Restore.java
+++ /dev/null
@@ -1,107 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.train;
-
-import java.util.Arrays;
-import java.util.Iterator;
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TString;
-import org.tensorflow.types.family.TType;
-
-/**
- * Restores tensors from a V2 checkpoint.
- * <p>
- * For backward compatibility with the V1 format, this Op currently allows
- * restoring from a V1 checkpoint as well:
- *   - This Op first attempts to find the V2 index file pointed to by "prefix", and
- *     if found proceed to read it as a V2 checkpoint;
- *   - Otherwise the V1 read path is invoked.
- * Relying on this behavior is not recommended, as the ability to fall back to read
- * V1 might be deprecated and eventually removed.
- * <p>
- * By default, restores the named tensors in full.  If the caller wishes to restore
- * specific slices of stored tensors, "shape_and_slices" should be non-empty
- * strings and correspondingly well-formed.
- * <p>
- * Callers must ensure all the named tensors are indeed stored in the checkpoint.
- */
-@Operator(group = "train")
-public final class Restore extends RawOp implements Iterable<Operand<TType>> {
-  
-  /**
-   * Factory method to create a class wrapping a new Restore operation.
-   * 
-   * @param scope current scope
-   * @param prefix Must have a single element.  The prefix of a V2 checkpoint.
-   * @param tensorNames shape {N}.  The names of the tensors to be restored.
-   * @param shapeAndSlices shape {N}.  The slice specs of the tensors to be restored.
-   * Empty strings indicate that they are non-partitioned tensors.
-   * @param dtypes shape {N}.  The list of expected dtype for the tensors.  Must match
-   * those stored in the checkpoint.
-   * @return a new instance of Restore
-   */
-  @Endpoint(describeByClass = true)
-  public static Restore create(Scope scope, Operand<TString> prefix, Operand<TString> tensorNames, Operand<TString> shapeAndSlices, List<Class<?>> dtypes) {
-    OperationBuilder opBuilder = scope.env().opBuilder("RestoreV2", scope.makeOpName("Restore"));
-    opBuilder.addInput(prefix.asOutput(scope));
-    opBuilder.addInput(tensorNames.asOutput(scope));
-    opBuilder.addInput(shapeAndSlices.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    Class[] dtypesArray = new Class[dtypes.size()];
-    for (int i = 0; i < dtypesArray.length; ++i) {
-      dtypesArray[i] = dtypes.get(i);
-    }
-    opBuilder.setAttr("dtypes", dtypesArray);
-    return new Restore(opBuilder.build());
-  }
-  
-  /**
-   * shape {N}.  The restored tensors, whose shapes are read from the
-   * checkpoint directly.
-   */
-  public List<Output<?>> tensors() {
-    return tensors;
-  }
-  
-  @Override
-  @SuppressWarnings({"rawtypes", "unchecked"})
-  public Iterator<Operand<TType>> iterator() {
-    return (Iterator) tensors.iterator();
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "RestoreV2";
-  
-  private List<Output<?>> tensors;
-  
-  private Restore(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    int tensorsLength = operation.outputListLength("tensors");
-    tensors = Arrays.asList(operation.outputList(outputIdx, tensorsLength));
-    outputIdx += tensorsLength;
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/RestoreSlice.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/RestoreSlice.java
deleted file mode 100644
index 11614978792..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/RestoreSlice.java
+++ /dev/null
@@ -1,128 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.train;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TString;
-import org.tensorflow.types.family.TType;
-
-/**
- * Restores a tensor from checkpoint files.
- * <p>
- * This is like `Restore` except that restored tensor can be listed as filling
- * only a slice of a larger tensor.  `shape_and_slice` specifies the shape of the
- * larger tensor and the slice that the restored tensor covers.
- * <p>
- * The `shape_and_slice` input has the same format as the
- * elements of the `shapes_and_slices` input of the `SaveSlices` op.
- * 
- * @param <T> data type for {@code tensor()} output
- */
-@Operator(group = "train")
-public final class RestoreSlice<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.train.RestoreSlice}
-   */
-  public static class Options {
-    
-    /**
-     * @param preferredShard Index of file to open first if multiple files match
-     * `file_pattern`. See the documentation for `Restore`.
-     */
-    public Options preferredShard(Long preferredShard) {
-      this.preferredShard = preferredShard;
-      return this;
-    }
-    
-    private Long preferredShard;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new RestoreSlice operation.
-   * 
-   * @param scope current scope
-   * @param filePattern Must have a single element. The pattern of the files from
-   * which we read the tensor.
-   * @param tensorName Must have a single element. The name of the tensor to be
-   * restored.
-   * @param shapeAndSlice Scalar. The shapes and slice specifications to use when
-   * restoring a tensors.
-   * @param dt The type of the tensor to be restored.
-   * @param options carries optional attributes values
-   * @return a new instance of RestoreSlice
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> RestoreSlice<T> create(Scope scope, Operand<TString> filePattern, Operand<TString> tensorName, Operand<TString> shapeAndSlice, Class<T> dt, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("RestoreSlice", scope.makeOpName("RestoreSlice"));
-    opBuilder.addInput(filePattern.asOutput(scope));
-    opBuilder.addInput(tensorName.asOutput(scope));
-    opBuilder.addInput(shapeAndSlice.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("dt", dt);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.preferredShard != null) {
-          opBuilder.setAttr("preferred_shard", opts.preferredShard);
-        }
-      }
-    }
-    return new RestoreSlice<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param preferredShard Index of file to open first if multiple files match
-   * `file_pattern`. See the documentation for `Restore`.
-   */
-  public static Options preferredShard(Long preferredShard) {
-    return new Options().preferredShard(preferredShard);
-  }
-  
-  /**
-   * The restored tensor.
-   */
-  public Output<T> tensor() {
-    return tensor;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return tensor;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "RestoreSlice";
-  
-  private Output<T> tensor;
-  
-  private RestoreSlice(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    tensor = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/Save.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/Save.java
deleted file mode 100644
index 21983a8fdc0..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/Save.java
+++ /dev/null
@@ -1,69 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.train;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.Operands;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TString;
-
-/**
- * Saves tensors in V2 checkpoint format.
- * <p>
- * By default, saves the named tensors in full.  If the caller wishes to save
- * specific slices of full tensors, "shape_and_slices" should be non-empty strings
- * and correspondingly well-formed.
- */
-@Operator(group = "train")
-public final class Save extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new Save operation.
-   * 
-   * @param scope current scope
-   * @param prefix Must have a single element. The prefix of the V2 checkpoint to which we
-   * write the tensors.
-   * @param tensorNames shape {N}. The names of the tensors to be saved.
-   * @param shapeAndSlices shape {N}.  The slice specs of the tensors to be saved.
-   * Empty strings indicate that they are non-partitioned tensors.
-   * @param tensors `N` tensors to save.
-   * @return a new instance of Save
-   */
-  @Endpoint(describeByClass = true)
-  public static Save create(Scope scope, Operand<TString> prefix, Operand<TString> tensorNames, Operand<TString> shapeAndSlices, Iterable<Operand<?>> tensors) {
-    OperationBuilder opBuilder = scope.env().opBuilder("SaveV2", scope.makeOpName("Save"));
-    opBuilder.addInput(prefix.asOutput(scope));
-    opBuilder.addInput(tensorNames.asOutput(scope));
-    opBuilder.addInput(shapeAndSlices.asOutput(scope));
-    opBuilder.addInputList(Operands.asOutputs(scope, tensors));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new Save(opBuilder.build());
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "SaveV2";
-  
-  private Save(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SaveSlices.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SaveSlices.java
deleted file mode 100644
index 348f8d983c2..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SaveSlices.java
+++ /dev/null
@@ -1,95 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.train;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.Operands;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TString;
-
-/**
- * Saves input tensors slices to disk.
- * <p>
- * This is like `Save` except that tensors can be listed in the saved file as being
- * a slice of a larger tensor.  `shapes_and_slices` specifies the shape of the
- * larger tensor and the slice that this tensor covers. `shapes_and_slices` must
- * have as many elements as `tensor_names`.
- * <p>
- * Elements of the `shapes_and_slices` input must either be:
- * <ul>
- * <li>
- * The empty string, in which case the corresponding tensor is
- *    saved normally.
- * </li>
- * <li>
- * A string of the form `dim0 dim1 ... dimN-1 slice-spec` where the
- *    `dimI` are the dimensions of the larger tensor and `slice-spec`
- *    specifies what part is covered by the tensor to save.
- * </li>
- * </ul>
- * `slice-spec` itself is a `:`-separated list: `slice0:slice1:...:sliceN-1`
- * where each `sliceI` is either:
- * <ul>
- * <li>
- * The string `-` meaning that the slice covers all indices of this dimension
- * </li>
- * <li>
- * `start,length` where `start` and `length` are integers.  In that
- *    case the slice covers `length` indices starting at `start`.
- * </li>
- * </ul>
- * See also `Save`.
- */
-@Operator(group = "train")
-public final class SaveSlices extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new SaveSlices operation.
-   * 
-   * @param scope current scope
-   * @param filename Must have a single element. The name of the file to which we write the
-   * tensor.
-   * @param tensorNames Shape `[N]`. The names of the tensors to be saved.
-   * @param shapesAndSlices Shape `[N]`.  The shapes and slice specifications to use when
-   * saving the tensors.
-   * @param data `N` tensors to save.
-   * @return a new instance of SaveSlices
-   */
-  @Endpoint(describeByClass = true)
-  public static SaveSlices create(Scope scope, Operand<TString> filename, Operand<TString> tensorNames, Operand<TString> shapesAndSlices, Iterable<Operand<?>> data) {
-    OperationBuilder opBuilder = scope.env().opBuilder("SaveSlices", scope.makeOpName("SaveSlices"));
-    opBuilder.addInput(filename.asOutput(scope));
-    opBuilder.addInput(tensorNames.asOutput(scope));
-    opBuilder.addInput(shapesAndSlices.asOutput(scope));
-    opBuilder.addInputList(Operands.asOutputs(scope, data));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new SaveSlices(opBuilder.build());
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "SaveSlices";
-  
-  private SaveSlices(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SdcaFprint.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SdcaFprint.java
deleted file mode 100644
index b162a071713..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SdcaFprint.java
+++ /dev/null
@@ -1,75 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.train;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.TString;
-
-/**
- * Computes fingerprints of the input strings.
- */
-@Operator(group = "train")
-public final class SdcaFprint extends RawOp implements Operand<TInt64> {
-  
-  /**
-   * Factory method to create a class wrapping a new SdcaFprint operation.
-   * 
-   * @param scope current scope
-   * @param input vector of strings to compute fingerprints on.
-   * @return a new instance of SdcaFprint
-   */
-  @Endpoint(describeByClass = true)
-  public static SdcaFprint create(Scope scope, Operand<TString> input) {
-    OperationBuilder opBuilder = scope.env().opBuilder("SdcaFprint", scope.makeOpName("SdcaFprint"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new SdcaFprint(opBuilder.build());
-  }
-  
-  /**
-   * a (N,2) shaped matrix where N is the number of elements in the input
-   * vector. Each row contains the low and high parts of the fingerprint.
-   */
-  public Output<TInt64> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<TInt64> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "SdcaFprint";
-  
-  private Output<TInt64> output;
-  
-  private SdcaFprint(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SdcaOptimizer.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SdcaOptimizer.java
deleted file mode 100644
index a97a4589134..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SdcaOptimizer.java
+++ /dev/null
@@ -1,185 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.train;
-
-import java.util.Arrays;
-import java.util.List;
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.Operands;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.TInt64;
-
-/**
- * Distributed version of Stochastic Dual Coordinate Ascent (SDCA) optimizer for
- * <p>
- * linear models with L1 + L2 regularization. As global optimization objective is
- * strongly-convex, the optimizer optimizes the dual objective at each step. The
- * optimizer applies each update one example at a time. Examples are sampled
- * uniformly, and the optimizer is learning rate free and enjoys linear convergence
- * rate.
- * <p>
- * [Proximal Stochastic Dual Coordinate Ascent](http://arxiv.org/pdf/1211.2717v1.pdf).<br>
- * Shai Shalev-Shwartz, Tong Zhang. 2012
- * <p>
- * $$Loss Objective = \sum f_{i} (wx_{i}) + (l2 / 2) * |w|^2 + l1 * |w|$$
- * <p>
- * [Adding vs. Averaging in Distributed Primal-Dual Optimization](http://arxiv.org/abs/1502.03508).<br>
- * Chenxin Ma, Virginia Smith, Martin Jaggi, Michael I. Jordan,
- * Peter Richtarik, Martin Takac. 2015
- * <p>
- * [Stochastic Dual Coordinate Ascent with Adaptive Probabilities](https://arxiv.org/abs/1502.08053).<br>
- * Dominik Csiba, Zheng Qu, Peter Richtarik. 2015
- */
-public final class SdcaOptimizer extends RawOp {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.train.SdcaOptimizer}
-   */
-  public static class Options {
-    
-    /**
-     * @param adaptive Whether to use Adaptive SDCA for the inner loop.
-     */
-    public Options adaptive(Boolean adaptive) {
-      this.adaptive = adaptive;
-      return this;
-    }
-    
-    private Boolean adaptive;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new SdcaOptimizer operation.
-   * 
-   * @param scope current scope
-   * @param sparseExampleIndices a list of vectors which contain example indices.
-   * @param sparseFeatureIndices a list of vectors which contain feature indices.
-   * @param sparseFeatureValues a list of vectors which contains feature value
-   * associated with each feature group.
-   * @param denseFeatures a list of matrices which contains the dense feature values.
-   * @param exampleWeights a vector which contains the weight associated with each
-   * example.
-   * @param exampleLabels a vector which contains the label/target associated with each
-   * example.
-   * @param sparseIndices a list of vectors where each value is the indices which has
-   * corresponding weights in sparse_weights. This field maybe omitted for the
-   * dense approach.
-   * @param sparseWeights a list of vectors where each value is the weight associated with
-   * a sparse feature group.
-   * @param denseWeights a list of vectors where the values are the weights associated
-   * with a dense feature group.
-   * @param exampleStateData a list of vectors containing the example state data.
-   * @param lossType Type of the primal loss. Currently SdcaSolver supports logistic,
-   * squared and hinge losses.
-   * @param l1 Symmetric l1 regularization strength.
-   * @param l2 Symmetric l2 regularization strength.
-   * @param numLossPartitions Number of partitions of the global loss function.
-   * @param numInnerIterations Number of iterations per mini-batch.
-   * @param options carries optional attributes values
-   * @return a new instance of SdcaOptimizer
-   */
-  @Endpoint(describeByClass = true)
-  public static SdcaOptimizer create(Scope scope, Iterable<Operand<TInt64>> sparseExampleIndices, Iterable<Operand<TInt64>> sparseFeatureIndices, Iterable<Operand<TFloat32>> sparseFeatureValues, Iterable<Operand<TFloat32>> denseFeatures, Operand<TFloat32> exampleWeights, Operand<TFloat32> exampleLabels, Iterable<Operand<TInt64>> sparseIndices, Iterable<Operand<TFloat32>> sparseWeights, Iterable<Operand<TFloat32>> denseWeights, Operand<TFloat32> exampleStateData, String lossType, Float l1, Float l2, Long numLossPartitions, Long numInnerIterations, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("SdcaOptimizerV2", scope.makeOpName("SdcaOptimizer"));
-    opBuilder.addInputList(Operands.asOutputs(scope, sparseExampleIndices));
-    opBuilder.addInputList(Operands.asOutputs(scope, sparseFeatureIndices));
-    opBuilder.addInputList(Operands.asOutputs(scope, sparseFeatureValues));
-    opBuilder.addInputList(Operands.asOutputs(scope, denseFeatures));
-    opBuilder.addInput(exampleWeights.asOutput(scope));
-    opBuilder.addInput(exampleLabels.asOutput(scope));
-    opBuilder.addInputList(Operands.asOutputs(scope, sparseIndices));
-    opBuilder.addInputList(Operands.asOutputs(scope, sparseWeights));
-    opBuilder.addInputList(Operands.asOutputs(scope, denseWeights));
-    opBuilder.addInput(exampleStateData.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("loss_type", lossType);
-    opBuilder.setAttr("l1", l1);
-    opBuilder.setAttr("l2", l2);
-    opBuilder.setAttr("num_loss_partitions", numLossPartitions);
-    opBuilder.setAttr("num_inner_iterations", numInnerIterations);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.adaptive != null) {
-          opBuilder.setAttr("adaptive", opts.adaptive);
-        }
-      }
-    }
-    return new SdcaOptimizer(opBuilder.build());
-  }
-  
-  /**
-   * @param adaptive Whether to use Adaptive SDCA for the inner loop.
-   */
-  public static Options adaptive(Boolean adaptive) {
-    return new Options().adaptive(adaptive);
-  }
-  
-  /**
-   * a list of vectors containing the updated example state
-   * data.
-   */
-  public Output<TFloat32> outExampleStateData() {
-    return outExampleStateData;
-  }
-  
-  /**
-   * a list of vectors where each value is the delta
-   * weights associated with a sparse feature group.
-   */
-  public List<Output<TFloat32>> outDeltaSparseWeights() {
-    return outDeltaSparseWeights;
-  }
-  
-  /**
-   * a list of vectors where the values are the delta
-   * weights associated with a dense feature group.
-   */
-  public List<Output<TFloat32>> outDeltaDenseWeights() {
-    return outDeltaDenseWeights;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "SdcaOptimizerV2";
-  
-  private Output<TFloat32> outExampleStateData;
-  private List<Output<TFloat32>> outDeltaSparseWeights;
-  private List<Output<TFloat32>> outDeltaDenseWeights;
-  
-  @SuppressWarnings("unchecked")
-  private SdcaOptimizer(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    outExampleStateData = operation.output(outputIdx++);
-    int outDeltaSparseWeightsLength = operation.outputListLength("out_delta_sparse_weights");
-    outDeltaSparseWeights = Arrays.asList((Output<TFloat32>[])operation.outputList(outputIdx, outDeltaSparseWeightsLength));
-    outputIdx += outDeltaSparseWeightsLength;
-    int outDeltaDenseWeightsLength = operation.outputListLength("out_delta_dense_weights");
-    outDeltaDenseWeights = Arrays.asList((Output<TFloat32>[])operation.outputList(outputIdx, outDeltaDenseWeightsLength));
-    outputIdx += outDeltaDenseWeightsLength;
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SdcaShrinkL1.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SdcaShrinkL1.java
deleted file mode 100644
index 54fad2cdcf6..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SdcaShrinkL1.java
+++ /dev/null
@@ -1,62 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.train;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.Operands;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TFloat32;
-
-/**
- * Applies L1 regularization shrink step on the parameters.
- */
-@Operator(group = "train")
-public final class SdcaShrinkL1 extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new SdcaShrinkL1 operation.
-   * 
-   * @param scope current scope
-   * @param weights a list of vectors where each value is the weight associated with a
-   * feature group.
-   * @param l1 Symmetric l1 regularization strength.
-   * @param l2 Symmetric l2 regularization strength. Should be a positive float.
-   * @return a new instance of SdcaShrinkL1
-   */
-  @Endpoint(describeByClass = true)
-  public static SdcaShrinkL1 create(Scope scope, Iterable<Operand<TFloat32>> weights, Float l1, Float l2) {
-    OperationBuilder opBuilder = scope.env().opBuilder("SdcaShrinkL1", scope.makeOpName("SdcaShrinkL1"));
-    opBuilder.addInputList(Operands.asOutputs(scope, weights));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("l1", l1);
-    opBuilder.setAttr("l2", l2);
-    return new SdcaShrinkL1(opBuilder.build());
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "SdcaShrinkL1";
-  
-  private SdcaShrinkL1(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyAdadelta.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyAdadelta.java
deleted file mode 100644
index 5f44aca4160..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyAdadelta.java
+++ /dev/null
@@ -1,126 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.train;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * var: Should be from a Variable().
- * 
- * @param <T> data type for {@code out()} output
- */
-@Operator(group = "train")
-public final class SparseApplyAdadelta<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.train.SparseApplyAdadelta}
-   */
-  public static class Options {
-    
-    /**
-     * @param useLocking If True, updating of the var and accum tensors will be protected by
-     * a lock; otherwise the behavior is undefined, but may exhibit less contention.
-     */
-    public Options useLocking(Boolean useLocking) {
-      this.useLocking = useLocking;
-      return this;
-    }
-    
-    private Boolean useLocking;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new SparseApplyAdadelta operation.
-   * 
-   * @param scope current scope
-   * @param var 
-   * @param accum Should be from a Variable().
-   * @param accumUpdate : Should be from a Variable().
-   * @param lr Learning rate. Must be a scalar.
-   * @param rho Decay factor. Must be a scalar.
-   * @param epsilon Constant factor. Must be a scalar.
-   * @param grad The gradient.
-   * @param indices A vector of indices into the first dimension of var and accum.
-   * @param options carries optional attributes values
-   * @return a new instance of SparseApplyAdadelta
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> SparseApplyAdadelta<T> create(Scope scope, Operand<T> var, Operand<T> accum, Operand<T> accumUpdate, Operand<T> lr, Operand<T> rho, Operand<T> epsilon, Operand<T> grad, Operand<U> indices, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("SparseApplyAdadelta", scope.makeOpName("SparseApplyAdadelta"));
-    opBuilder.addInput(var.asOutput(scope));
-    opBuilder.addInput(accum.asOutput(scope));
-    opBuilder.addInput(accumUpdate.asOutput(scope));
-    opBuilder.addInput(lr.asOutput(scope));
-    opBuilder.addInput(rho.asOutput(scope));
-    opBuilder.addInput(epsilon.asOutput(scope));
-    opBuilder.addInput(grad.asOutput(scope));
-    opBuilder.addInput(indices.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.useLocking != null) {
-          opBuilder.setAttr("use_locking", opts.useLocking);
-        }
-      }
-    }
-    return new SparseApplyAdadelta<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param useLocking If True, updating of the var and accum tensors will be protected by
-   * a lock; otherwise the behavior is undefined, but may exhibit less contention.
-   */
-  public static Options useLocking(Boolean useLocking) {
-    return new Options().useLocking(useLocking);
-  }
-  
-  /**
-   * Same as "var".
-   */
-  public Output<T> out() {
-    return out;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return out;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "SparseApplyAdadelta";
-  
-  private Output<T> out;
-  
-  private SparseApplyAdadelta(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    out = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyAdagrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyAdagrad.java
deleted file mode 100644
index a4c0eeb942a..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyAdagrad.java
+++ /dev/null
@@ -1,146 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.train;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * Update relevant entries in '*var' and '*accum' according to the adagrad scheme.
- * <p>
- * That is for rows we have grad for, we update var and accum as follows:
- * $$accum += grad * grad$$
- * $$var -= lr * grad * (1 / sqrt(accum))$$
- * 
- * @param <T> data type for {@code out()} output
- */
-public final class SparseApplyAdagrad<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.train.SparseApplyAdagrad}
-   */
-  public static class Options {
-    
-    /**
-     * @param useLocking If `True`, updating of the var and accum tensors will be protected
-     * by a lock; otherwise the behavior is undefined, but may exhibit less
-     * contention.
-     */
-    public Options useLocking(Boolean useLocking) {
-      this.useLocking = useLocking;
-      return this;
-    }
-    
-    /**
-     * @param updateSlots 
-     */
-    public Options updateSlots(Boolean updateSlots) {
-      this.updateSlots = updateSlots;
-      return this;
-    }
-    
-    private Boolean useLocking;
-    private Boolean updateSlots;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new SparseApplyAdagrad operation.
-   * 
-   * @param scope current scope
-   * @param var Should be from a Variable().
-   * @param accum Should be from a Variable().
-   * @param lr Learning rate. Must be a scalar.
-   * @param epsilon Constant factor. Must be a scalar.
-   * @param grad The gradient.
-   * @param indices A vector of indices into the first dimension of var and accum.
-   * @param options carries optional attributes values
-   * @return a new instance of SparseApplyAdagrad
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> SparseApplyAdagrad<T> create(Scope scope, Operand<T> var, Operand<T> accum, Operand<T> lr, Operand<T> epsilon, Operand<T> grad, Operand<U> indices, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("SparseApplyAdagradV2", scope.makeOpName("SparseApplyAdagrad"));
-    opBuilder.addInput(var.asOutput(scope));
-    opBuilder.addInput(accum.asOutput(scope));
-    opBuilder.addInput(lr.asOutput(scope));
-    opBuilder.addInput(epsilon.asOutput(scope));
-    opBuilder.addInput(grad.asOutput(scope));
-    opBuilder.addInput(indices.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.useLocking != null) {
-          opBuilder.setAttr("use_locking", opts.useLocking);
-        }
-        if (opts.updateSlots != null) {
-          opBuilder.setAttr("update_slots", opts.updateSlots);
-        }
-      }
-    }
-    return new SparseApplyAdagrad<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param useLocking If `True`, updating of the var and accum tensors will be protected
-   * by a lock; otherwise the behavior is undefined, but may exhibit less
-   * contention.
-   */
-  public static Options useLocking(Boolean useLocking) {
-    return new Options().useLocking(useLocking);
-  }
-  
-  /**
-   * @param updateSlots 
-   */
-  public static Options updateSlots(Boolean updateSlots) {
-    return new Options().updateSlots(updateSlots);
-  }
-  
-  /**
-   * Same as "var".
-   */
-  public Output<T> out() {
-    return out;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return out;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "SparseApplyAdagradV2";
-  
-  private Output<T> out;
-  
-  private SparseApplyAdagrad(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    out = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyAdagradDa.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyAdagradDa.java
deleted file mode 100644
index 6264bfaadd7..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyAdagradDa.java
+++ /dev/null
@@ -1,129 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.train;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * Update entries in '*var' and '*accum' according to the proximal adagrad scheme.
- * 
- * @param <T> data type for {@code out()} output
- */
-@Operator(group = "train")
-public final class SparseApplyAdagradDa<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.train.SparseApplyAdagradDa}
-   */
-  public static class Options {
-    
-    /**
-     * @param useLocking If True, updating of the var and accum tensors will be protected by
-     * a lock; otherwise the behavior is undefined, but may exhibit less contention.
-     */
-    public Options useLocking(Boolean useLocking) {
-      this.useLocking = useLocking;
-      return this;
-    }
-    
-    private Boolean useLocking;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new SparseApplyAdagradDa operation.
-   * 
-   * @param scope current scope
-   * @param var Should be from a Variable().
-   * @param gradientAccumulator Should be from a Variable().
-   * @param gradientSquaredAccumulator Should be from a Variable().
-   * @param grad The gradient.
-   * @param indices A vector of indices into the first dimension of var and accum.
-   * @param lr Learning rate. Must be a scalar.
-   * @param l1 L1 regularization. Must be a scalar.
-   * @param l2 L2 regularization. Must be a scalar.
-   * @param globalStep Training step number. Must be a scalar.
-   * @param options carries optional attributes values
-   * @return a new instance of SparseApplyAdagradDa
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> SparseApplyAdagradDa<T> create(Scope scope, Operand<T> var, Operand<T> gradientAccumulator, Operand<T> gradientSquaredAccumulator, Operand<T> grad, Operand<U> indices, Operand<T> lr, Operand<T> l1, Operand<T> l2, Operand<TInt64> globalStep, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("SparseApplyAdagradDA", scope.makeOpName("SparseApplyAdagradDa"));
-    opBuilder.addInput(var.asOutput(scope));
-    opBuilder.addInput(gradientAccumulator.asOutput(scope));
-    opBuilder.addInput(gradientSquaredAccumulator.asOutput(scope));
-    opBuilder.addInput(grad.asOutput(scope));
-    opBuilder.addInput(indices.asOutput(scope));
-    opBuilder.addInput(lr.asOutput(scope));
-    opBuilder.addInput(l1.asOutput(scope));
-    opBuilder.addInput(l2.asOutput(scope));
-    opBuilder.addInput(globalStep.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.useLocking != null) {
-          opBuilder.setAttr("use_locking", opts.useLocking);
-        }
-      }
-    }
-    return new SparseApplyAdagradDa<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param useLocking If True, updating of the var and accum tensors will be protected by
-   * a lock; otherwise the behavior is undefined, but may exhibit less contention.
-   */
-  public static Options useLocking(Boolean useLocking) {
-    return new Options().useLocking(useLocking);
-  }
-  
-  /**
-   * Same as "var".
-   */
-  public Output<T> out() {
-    return out;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return out;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "SparseApplyAdagradDA";
-  
-  private Output<T> out;
-  
-  private SparseApplyAdagradDa(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    out = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyCenteredRmsProp.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyCenteredRmsProp.java
deleted file mode 100644
index 8b4d0d9d78c..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyCenteredRmsProp.java
+++ /dev/null
@@ -1,149 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.train;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * Update '*var' according to the centered RMSProp algorithm.
- * <p>
- * The centered RMSProp algorithm uses an estimate of the centered second moment
- * (i.e., the variance) for normalization, as opposed to regular RMSProp, which
- * uses the (uncentered) second moment. This often helps with training, but is
- * slightly more expensive in terms of computation and memory.
- * <p>
- * Note that in dense implementation of this algorithm, mg, ms, and mom will
- * update even if the grad is zero, but in this sparse implementation, mg, ms,
- * and mom will not update in iterations during which the grad is zero.
- * <p>
- * mean_square = decay * mean_square + (1-decay) * gradient ** 2
- * mean_grad = decay * mean_grad + (1-decay) * gradient
- * Delta = learning_rate * gradient / sqrt(mean_square + epsilon - mean_grad ** 2)
- * <p>
- * $$ms <- rho * ms_{t-1} + (1-rho) * grad * grad$$
- * $$mom <- momentum * mom_{t-1} + lr * grad / sqrt(ms + epsilon)$$
- * $$var <- var - mom$$
- * 
- * @param <T> data type for {@code out()} output
- */
-@Operator(group = "train")
-public final class SparseApplyCenteredRmsProp<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.train.SparseApplyCenteredRmsProp}
-   */
-  public static class Options {
-    
-    /**
-     * @param useLocking If `True`, updating of the var, mg, ms, and mom tensors is
-     * protected by a lock; otherwise the behavior is undefined, but may exhibit less
-     * contention.
-     */
-    public Options useLocking(Boolean useLocking) {
-      this.useLocking = useLocking;
-      return this;
-    }
-    
-    private Boolean useLocking;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new SparseApplyCenteredRmsProp operation.
-   * 
-   * @param scope current scope
-   * @param var Should be from a Variable().
-   * @param mg Should be from a Variable().
-   * @param ms Should be from a Variable().
-   * @param mom Should be from a Variable().
-   * @param lr Scaling factor. Must be a scalar.
-   * @param rho Decay rate. Must be a scalar.
-   * @param momentum 
-   * @param epsilon Ridge term. Must be a scalar.
-   * @param grad The gradient.
-   * @param indices A vector of indices into the first dimension of var, ms and mom.
-   * @param options carries optional attributes values
-   * @return a new instance of SparseApplyCenteredRmsProp
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> SparseApplyCenteredRmsProp<T> create(Scope scope, Operand<T> var, Operand<T> mg, Operand<T> ms, Operand<T> mom, Operand<T> lr, Operand<T> rho, Operand<T> momentum, Operand<T> epsilon, Operand<T> grad, Operand<U> indices, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("SparseApplyCenteredRMSProp", scope.makeOpName("SparseApplyCenteredRmsProp"));
-    opBuilder.addInput(var.asOutput(scope));
-    opBuilder.addInput(mg.asOutput(scope));
-    opBuilder.addInput(ms.asOutput(scope));
-    opBuilder.addInput(mom.asOutput(scope));
-    opBuilder.addInput(lr.asOutput(scope));
-    opBuilder.addInput(rho.asOutput(scope));
-    opBuilder.addInput(momentum.asOutput(scope));
-    opBuilder.addInput(epsilon.asOutput(scope));
-    opBuilder.addInput(grad.asOutput(scope));
-    opBuilder.addInput(indices.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.useLocking != null) {
-          opBuilder.setAttr("use_locking", opts.useLocking);
-        }
-      }
-    }
-    return new SparseApplyCenteredRmsProp<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param useLocking If `True`, updating of the var, mg, ms, and mom tensors is
-   * protected by a lock; otherwise the behavior is undefined, but may exhibit less
-   * contention.
-   */
-  public static Options useLocking(Boolean useLocking) {
-    return new Options().useLocking(useLocking);
-  }
-  
-  /**
-   * Same as "var".
-   */
-  public Output<T> out() {
-    return out;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return out;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "SparseApplyCenteredRMSProp";
-  
-  private Output<T> out;
-  
-  private SparseApplyCenteredRmsProp(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    out = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyFtrl.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyFtrl.java
deleted file mode 100644
index 8102aea01ee..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyFtrl.java
+++ /dev/null
@@ -1,160 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.train;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * Update relevant entries in '*var' according to the Ftrl-proximal scheme.
- * <p>
- * That is for rows we have grad for, we update var, accum and linear as follows:
- * grad_with_shrinkage = grad + 2 * l2_shrinkage * var
- * accum_new = accum + grad * grad
- * linear += grad_with_shrinkage -
- *     (accum_new^(-lr_power) - accum^(-lr_power)) / lr * var
- * quadratic = 1.0 / (accum_new^(lr_power) * lr) + 2 * l2
- * var = (sign(linear) * l1 - linear) / quadratic if |linear| > l1 else 0.0
- * accum = accum_new
- * 
- * @param <T> data type for {@code out()} output
- */
-@Operator(group = "train")
-public final class SparseApplyFtrl<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.train.SparseApplyFtrl}
-   */
-  public static class Options {
-    
-    /**
-     * @param useLocking If `True`, updating of the var and accum tensors will be protected
-     * by a lock; otherwise the behavior is undefined, but may exhibit less
-     * contention.
-     */
-    public Options useLocking(Boolean useLocking) {
-      this.useLocking = useLocking;
-      return this;
-    }
-    
-    /**
-     * @param multiplyLinearByLr 
-     */
-    public Options multiplyLinearByLr(Boolean multiplyLinearByLr) {
-      this.multiplyLinearByLr = multiplyLinearByLr;
-      return this;
-    }
-    
-    private Boolean useLocking;
-    private Boolean multiplyLinearByLr;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new SparseApplyFtrl operation.
-   * 
-   * @param scope current scope
-   * @param var Should be from a Variable().
-   * @param accum Should be from a Variable().
-   * @param linear Should be from a Variable().
-   * @param grad The gradient.
-   * @param indices A vector of indices into the first dimension of var and accum.
-   * @param lr Scaling factor. Must be a scalar.
-   * @param l1 L1 regularization. Must be a scalar.
-   * @param l2 L2 shrinkage regularization. Must be a scalar.
-   * @param l2Shrinkage 
-   * @param lrPower Scaling factor. Must be a scalar.
-   * @param options carries optional attributes values
-   * @return a new instance of SparseApplyFtrl
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> SparseApplyFtrl<T> create(Scope scope, Operand<T> var, Operand<T> accum, Operand<T> linear, Operand<T> grad, Operand<U> indices, Operand<T> lr, Operand<T> l1, Operand<T> l2, Operand<T> l2Shrinkage, Operand<T> lrPower, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("SparseApplyFtrlV2", scope.makeOpName("SparseApplyFtrl"));
-    opBuilder.addInput(var.asOutput(scope));
-    opBuilder.addInput(accum.asOutput(scope));
-    opBuilder.addInput(linear.asOutput(scope));
-    opBuilder.addInput(grad.asOutput(scope));
-    opBuilder.addInput(indices.asOutput(scope));
-    opBuilder.addInput(lr.asOutput(scope));
-    opBuilder.addInput(l1.asOutput(scope));
-    opBuilder.addInput(l2.asOutput(scope));
-    opBuilder.addInput(l2Shrinkage.asOutput(scope));
-    opBuilder.addInput(lrPower.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.useLocking != null) {
-          opBuilder.setAttr("use_locking", opts.useLocking);
-        }
-        if (opts.multiplyLinearByLr != null) {
-          opBuilder.setAttr("multiply_linear_by_lr", opts.multiplyLinearByLr);
-        }
-      }
-    }
-    return new SparseApplyFtrl<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param useLocking If `True`, updating of the var and accum tensors will be protected
-   * by a lock; otherwise the behavior is undefined, but may exhibit less
-   * contention.
-   */
-  public static Options useLocking(Boolean useLocking) {
-    return new Options().useLocking(useLocking);
-  }
-  
-  /**
-   * @param multiplyLinearByLr 
-   */
-  public static Options multiplyLinearByLr(Boolean multiplyLinearByLr) {
-    return new Options().multiplyLinearByLr(multiplyLinearByLr);
-  }
-  
-  /**
-   * Same as "var".
-   */
-  public Output<T> out() {
-    return out;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return out;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "SparseApplyFtrlV2";
-  
-  private Output<T> out;
-  
-  private SparseApplyFtrl(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    out = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyMomentum.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyMomentum.java
deleted file mode 100644
index 84068e2d518..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyMomentum.java
+++ /dev/null
@@ -1,154 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.train;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * Update relevant entries in '*var' and '*accum' according to the momentum scheme.
- * <p>
- * Set use_nesterov = True if you want to use Nesterov momentum.
- * <p>
- * That is for rows we have grad for, we update var and accum as follows:
- * <p>
- * $$accum = accum * momentum + grad$$
- * $$var -= lr * accum$$
- * 
- * @param <T> data type for {@code out()} output
- */
-@Operator(group = "train")
-public final class SparseApplyMomentum<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.train.SparseApplyMomentum}
-   */
-  public static class Options {
-    
-    /**
-     * @param useLocking If `True`, updating of the var and accum tensors will be protected
-     * by a lock; otherwise the behavior is undefined, but may exhibit less
-     * contention.
-     */
-    public Options useLocking(Boolean useLocking) {
-      this.useLocking = useLocking;
-      return this;
-    }
-    
-    /**
-     * @param useNesterov If `True`, the tensor passed to compute grad will be
-     * var - lr * momentum * accum, so in the end, the var you get is actually
-     * var - lr * momentum * accum.
-     */
-    public Options useNesterov(Boolean useNesterov) {
-      this.useNesterov = useNesterov;
-      return this;
-    }
-    
-    private Boolean useLocking;
-    private Boolean useNesterov;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new SparseApplyMomentum operation.
-   * 
-   * @param scope current scope
-   * @param var Should be from a Variable().
-   * @param accum Should be from a Variable().
-   * @param lr Learning rate. Must be a scalar.
-   * @param grad The gradient.
-   * @param indices A vector of indices into the first dimension of var and accum.
-   * @param momentum Momentum. Must be a scalar.
-   * @param options carries optional attributes values
-   * @return a new instance of SparseApplyMomentum
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> SparseApplyMomentum<T> create(Scope scope, Operand<T> var, Operand<T> accum, Operand<T> lr, Operand<T> grad, Operand<U> indices, Operand<T> momentum, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("SparseApplyMomentum", scope.makeOpName("SparseApplyMomentum"));
-    opBuilder.addInput(var.asOutput(scope));
-    opBuilder.addInput(accum.asOutput(scope));
-    opBuilder.addInput(lr.asOutput(scope));
-    opBuilder.addInput(grad.asOutput(scope));
-    opBuilder.addInput(indices.asOutput(scope));
-    opBuilder.addInput(momentum.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.useLocking != null) {
-          opBuilder.setAttr("use_locking", opts.useLocking);
-        }
-        if (opts.useNesterov != null) {
-          opBuilder.setAttr("use_nesterov", opts.useNesterov);
-        }
-      }
-    }
-    return new SparseApplyMomentum<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param useLocking If `True`, updating of the var and accum tensors will be protected
-   * by a lock; otherwise the behavior is undefined, but may exhibit less
-   * contention.
-   */
-  public static Options useLocking(Boolean useLocking) {
-    return new Options().useLocking(useLocking);
-  }
-  
-  /**
-   * @param useNesterov If `True`, the tensor passed to compute grad will be
-   * var - lr * momentum * accum, so in the end, the var you get is actually
-   * var - lr * momentum * accum.
-   */
-  public static Options useNesterov(Boolean useNesterov) {
-    return new Options().useNesterov(useNesterov);
-  }
-  
-  /**
-   * Same as "var".
-   */
-  public Output<T> out() {
-    return out;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return out;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "SparseApplyMomentum";
-  
-  private Output<T> out;
-  
-  private SparseApplyMomentum(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    out = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyProximalAdagrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyProximalAdagrad.java
deleted file mode 100644
index 17933470a49..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyProximalAdagrad.java
+++ /dev/null
@@ -1,130 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.train;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * Sparse update entries in '*var' and '*accum' according to FOBOS algorithm.
- * <p>
- * That is for rows we have grad for, we update var and accum as follows:
- * $$accum += grad <i> grad$$
- * $$prox_v = var$$
- * $$prox_v -= lr </i> grad <i> (1 / sqrt(accum))$$
- * $$var = sign(prox_v)/(1+lr</i>l2) <i> max{|prox_v|-lr</i>l1,0}$$
- * 
- * @param <T> data type for {@code out()} output
- */
-@Operator(group = "train")
-public final class SparseApplyProximalAdagrad<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.train.SparseApplyProximalAdagrad}
-   */
-  public static class Options {
-    
-    /**
-     * @param useLocking If True, updating of the var and accum tensors will be protected by
-     * a lock; otherwise the behavior is undefined, but may exhibit less contention.
-     */
-    public Options useLocking(Boolean useLocking) {
-      this.useLocking = useLocking;
-      return this;
-    }
-    
-    private Boolean useLocking;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new SparseApplyProximalAdagrad operation.
-   * 
-   * @param scope current scope
-   * @param var Should be from a Variable().
-   * @param accum Should be from a Variable().
-   * @param lr Learning rate. Must be a scalar.
-   * @param l1 L1 regularization. Must be a scalar.
-   * @param l2 L2 regularization. Must be a scalar.
-   * @param grad The gradient.
-   * @param indices A vector of indices into the first dimension of var and accum.
-   * @param options carries optional attributes values
-   * @return a new instance of SparseApplyProximalAdagrad
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> SparseApplyProximalAdagrad<T> create(Scope scope, Operand<T> var, Operand<T> accum, Operand<T> lr, Operand<T> l1, Operand<T> l2, Operand<T> grad, Operand<U> indices, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("SparseApplyProximalAdagrad", scope.makeOpName("SparseApplyProximalAdagrad"));
-    opBuilder.addInput(var.asOutput(scope));
-    opBuilder.addInput(accum.asOutput(scope));
-    opBuilder.addInput(lr.asOutput(scope));
-    opBuilder.addInput(l1.asOutput(scope));
-    opBuilder.addInput(l2.asOutput(scope));
-    opBuilder.addInput(grad.asOutput(scope));
-    opBuilder.addInput(indices.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.useLocking != null) {
-          opBuilder.setAttr("use_locking", opts.useLocking);
-        }
-      }
-    }
-    return new SparseApplyProximalAdagrad<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param useLocking If True, updating of the var and accum tensors will be protected by
-   * a lock; otherwise the behavior is undefined, but may exhibit less contention.
-   */
-  public static Options useLocking(Boolean useLocking) {
-    return new Options().useLocking(useLocking);
-  }
-  
-  /**
-   * Same as "var".
-   */
-  public Output<T> out() {
-    return out;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return out;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "SparseApplyProximalAdagrad";
-  
-  private Output<T> out;
-  
-  private SparseApplyProximalAdagrad(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    out = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyProximalGradientDescent.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyProximalGradientDescent.java
deleted file mode 100644
index 5f3a6776c79..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyProximalGradientDescent.java
+++ /dev/null
@@ -1,126 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.train;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * Sparse update '*var' as FOBOS algorithm with fixed learning rate.
- * <p>
- * That is for rows we have grad for, we update var as follows:
- * $$prox_v = var - alpha <i> grad$$
- * $$var = sign(prox_v)/(1+alpha</i>l2) <i> max{|prox_v|-alpha</i>l1,0}$$
- * 
- * @param <T> data type for {@code out()} output
- */
-@Operator(group = "train")
-public final class SparseApplyProximalGradientDescent<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.train.SparseApplyProximalGradientDescent}
-   */
-  public static class Options {
-    
-    /**
-     * @param useLocking If True, the subtraction will be protected by a lock;
-     * otherwise the behavior is undefined, but may exhibit less contention.
-     */
-    public Options useLocking(Boolean useLocking) {
-      this.useLocking = useLocking;
-      return this;
-    }
-    
-    private Boolean useLocking;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new SparseApplyProximalGradientDescent operation.
-   * 
-   * @param scope current scope
-   * @param var Should be from a Variable().
-   * @param alpha Scaling factor. Must be a scalar.
-   * @param l1 L1 regularization. Must be a scalar.
-   * @param l2 L2 regularization. Must be a scalar.
-   * @param grad The gradient.
-   * @param indices A vector of indices into the first dimension of var and accum.
-   * @param options carries optional attributes values
-   * @return a new instance of SparseApplyProximalGradientDescent
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> SparseApplyProximalGradientDescent<T> create(Scope scope, Operand<T> var, Operand<T> alpha, Operand<T> l1, Operand<T> l2, Operand<T> grad, Operand<U> indices, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("SparseApplyProximalGradientDescent", scope.makeOpName("SparseApplyProximalGradientDescent"));
-    opBuilder.addInput(var.asOutput(scope));
-    opBuilder.addInput(alpha.asOutput(scope));
-    opBuilder.addInput(l1.asOutput(scope));
-    opBuilder.addInput(l2.asOutput(scope));
-    opBuilder.addInput(grad.asOutput(scope));
-    opBuilder.addInput(indices.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.useLocking != null) {
-          opBuilder.setAttr("use_locking", opts.useLocking);
-        }
-      }
-    }
-    return new SparseApplyProximalGradientDescent<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param useLocking If True, the subtraction will be protected by a lock;
-   * otherwise the behavior is undefined, but may exhibit less contention.
-   */
-  public static Options useLocking(Boolean useLocking) {
-    return new Options().useLocking(useLocking);
-  }
-  
-  /**
-   * Same as "var".
-   */
-  public Output<T> out() {
-    return out;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return out;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "SparseApplyProximalGradientDescent";
-  
-  private Output<T> out;
-  
-  private SparseApplyProximalGradientDescent(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    out = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyRmsProp.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyRmsProp.java
deleted file mode 100644
index a56246ca83f..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyRmsProp.java
+++ /dev/null
@@ -1,141 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.train;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * Update '*var' according to the RMSProp algorithm.
- * <p>
- * Note that in dense implementation of this algorithm, ms and mom will
- * update even if the grad is zero, but in this sparse implementation, ms
- * and mom will not update in iterations during which the grad is zero.
- * <p>
- * mean_square = decay * mean_square + (1-decay) * gradient ** 2
- * Delta = learning_rate * gradient / sqrt(mean_square + epsilon)
- * <p>
- * $$ms <- rho * ms_{t-1} + (1-rho) * grad * grad$$
- * $$mom <- momentum * mom_{t-1} + lr * grad / sqrt(ms + epsilon)$$
- * $$var <- var - mom$$
- * 
- * @param <T> data type for {@code out()} output
- */
-@Operator(group = "train")
-public final class SparseApplyRmsProp<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Optional attributes for {@link org.tensorflow.op.train.SparseApplyRmsProp}
-   */
-  public static class Options {
-    
-    /**
-     * @param useLocking If `True`, updating of the var, ms, and mom tensors is protected
-     * by a lock; otherwise the behavior is undefined, but may exhibit less
-     * contention.
-     */
-    public Options useLocking(Boolean useLocking) {
-      this.useLocking = useLocking;
-      return this;
-    }
-    
-    private Boolean useLocking;
-    
-    private Options() {
-    }
-  }
-  
-  /**
-   * Factory method to create a class wrapping a new SparseApplyRmsProp operation.
-   * 
-   * @param scope current scope
-   * @param var Should be from a Variable().
-   * @param ms Should be from a Variable().
-   * @param mom Should be from a Variable().
-   * @param lr Scaling factor. Must be a scalar.
-   * @param rho Decay rate. Must be a scalar.
-   * @param momentum 
-   * @param epsilon Ridge term. Must be a scalar.
-   * @param grad The gradient.
-   * @param indices A vector of indices into the first dimension of var, ms and mom.
-   * @param options carries optional attributes values
-   * @return a new instance of SparseApplyRmsProp
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> SparseApplyRmsProp<T> create(Scope scope, Operand<T> var, Operand<T> ms, Operand<T> mom, Operand<T> lr, Operand<T> rho, Operand<T> momentum, Operand<T> epsilon, Operand<T> grad, Operand<U> indices, Options... options) {
-    OperationBuilder opBuilder = scope.env().opBuilder("SparseApplyRMSProp", scope.makeOpName("SparseApplyRmsProp"));
-    opBuilder.addInput(var.asOutput(scope));
-    opBuilder.addInput(ms.asOutput(scope));
-    opBuilder.addInput(mom.asOutput(scope));
-    opBuilder.addInput(lr.asOutput(scope));
-    opBuilder.addInput(rho.asOutput(scope));
-    opBuilder.addInput(momentum.asOutput(scope));
-    opBuilder.addInput(epsilon.asOutput(scope));
-    opBuilder.addInput(grad.asOutput(scope));
-    opBuilder.addInput(indices.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    if (options != null) {
-      for (Options opts : options) {
-        if (opts.useLocking != null) {
-          opBuilder.setAttr("use_locking", opts.useLocking);
-        }
-      }
-    }
-    return new SparseApplyRmsProp<T>(opBuilder.build());
-  }
-  
-  /**
-   * @param useLocking If `True`, updating of the var, ms, and mom tensors is protected
-   * by a lock; otherwise the behavior is undefined, but may exhibit less
-   * contention.
-   */
-  public static Options useLocking(Boolean useLocking) {
-    return new Options().useLocking(useLocking);
-  }
-  
-  /**
-   * Same as "var".
-   */
-  public Output<T> out() {
-    return out;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return out;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "SparseApplyRMSProp";
-  
-  private Output<T> out;
-  
-  private SparseApplyRmsProp(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    out = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/TileGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/TileGrad.java
deleted file mode 100644
index 31b6a1dd486..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/TileGrad.java
+++ /dev/null
@@ -1,81 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.train;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.family.TType;
-
-/**
- * Returns the gradient of `Tile`.
- * <p>
- * Since `Tile` takes an input and repeats the input `multiples` times
- * along each dimension, `train.TileGrad` takes in `multiples` and aggregates
- * each repeated tile of `input` into `output`.
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator(group = "train")
-public final class TileGrad<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new TileGrad operation.
-   * 
-   * @param scope current scope
-   * @param input 
-   * @param multiples 
-   * @return a new instance of TileGrad
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> TileGrad<T> create(Scope scope, Operand<T> input, Operand<TInt32> multiples) {
-    OperationBuilder opBuilder = scope.env().opBuilder("TileGrad", scope.makeOpName("TileGrad"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder.addInput(multiples.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new TileGrad<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "TileGrad";
-  
-  private Output<T> output;
-  
-  private TileGrad(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/BroadcastHelper.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/BroadcastHelper.java
deleted file mode 100644
index 59c75e98fa2..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/BroadcastHelper.java
+++ /dev/null
@@ -1,88 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.xla;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * Helper operator for performing XLA-style broadcasts
- * <p>
- * Broadcasts `lhs` and `rhs` to the same rank, by adding size 1 dimensions to
- * whichever of `lhs` and `rhs` has the lower rank, using XLA's broadcasting rules
- * for binary operators.
- * 
- * @param <T> data type for {@code lhsOutput()} output
- */
-@Operator(group = "xla")
-public final class BroadcastHelper<T extends TType> extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new BroadcastHelper operation.
-   * 
-   * @param scope current scope
-   * @param lhs the LHS input tensor
-   * @param rhs the RHS input tensor
-   * @param broadcastDims an XLA-style broadcast dimension specification
-   * @return a new instance of BroadcastHelper
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> BroadcastHelper<T> create(Scope scope, Operand<T> lhs, Operand<T> rhs, Operand<U> broadcastDims) {
-    OperationBuilder opBuilder = scope.env().opBuilder("XlaBroadcastHelper", scope.makeOpName("BroadcastHelper"));
-    opBuilder.addInput(lhs.asOutput(scope));
-    opBuilder.addInput(rhs.asOutput(scope));
-    opBuilder.addInput(broadcastDims.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new BroadcastHelper<T>(opBuilder.build());
-  }
-  
-  /**
-   * the broadcasted LHS tensor
-   */
-  public Output<T> lhsOutput() {
-    return lhsOutput;
-  }
-  
-  /**
-   * the broadcasted RHS tensor
-   */
-  public Output<T> rhsOutput() {
-    return rhsOutput;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "XlaBroadcastHelper";
-  
-  private Output<T> lhsOutput;
-  private Output<T> rhsOutput;
-  
-  private BroadcastHelper(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    lhsOutput = operation.output(outputIdx++);
-    rhsOutput = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/ClusterOutput.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/ClusterOutput.java
deleted file mode 100644
index e9219f14b2a..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/ClusterOutput.java
+++ /dev/null
@@ -1,74 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.xla;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Operator that connects the output of an XLA computation to other consumer graph nodes.
- * 
- * @param <T> data type for {@code outputs()} output
- */
-@Operator(group = "xla")
-public final class ClusterOutput<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new ClusterOutput operation.
-   * 
-   * @param scope current scope
-   * @param input 
-   * @return a new instance of ClusterOutput
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> ClusterOutput<T> create(Scope scope, Operand<T> input) {
-    OperationBuilder opBuilder = scope.env().opBuilder("XlaClusterOutput", scope.makeOpName("ClusterOutput"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new ClusterOutput<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> outputs() {
-    return outputs;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return outputs;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "XlaClusterOutput";
-  
-  private Output<T> outputs;
-  
-  private ClusterOutput(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    outputs = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Conv.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Conv.java
deleted file mode 100644
index b6b94c44624..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Conv.java
+++ /dev/null
@@ -1,94 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.xla;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * Wraps the XLA ConvGeneralDilated operator, documented at
- * <p>
- *  https://www.tensorflow.org/performance/xla/operation_semantics#conv_convolution
- * .
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator(group = "xla")
-public final class Conv<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new Conv operation.
-   * 
-   * @param scope current scope
-   * @param lhs the input tensor
-   * @param rhs the kernel tensor
-   * @param windowStrides the inter-window strides
-   * @param padding the padding to apply at the start and end of each input dimensions
-   * @param lhsDilation dilation to apply between input elements
-   * @param rhsDilation dilation to apply between kernel elements
-   * @param featureGroupCount number of feature groups for grouped convolution.
-   * @param dimensionNumbers a serialized xla::ConvolutionDimensionNumbers proto.
-   * @param precisionConfig a serialized xla::PrecisionConfig proto.
-   * @return a new instance of Conv
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> Conv<T> create(Scope scope, Operand<T> lhs, Operand<T> rhs, Operand<U> windowStrides, Operand<U> padding, Operand<U> lhsDilation, Operand<U> rhsDilation, Operand<U> featureGroupCount, String dimensionNumbers, String precisionConfig) {
-    OperationBuilder opBuilder = scope.env().opBuilder("XlaConv", scope.makeOpName("Conv"));
-    opBuilder.addInput(lhs.asOutput(scope));
-    opBuilder.addInput(rhs.asOutput(scope));
-    opBuilder.addInput(windowStrides.asOutput(scope));
-    opBuilder.addInput(padding.asOutput(scope));
-    opBuilder.addInput(lhsDilation.asOutput(scope));
-    opBuilder.addInput(rhsDilation.asOutput(scope));
-    opBuilder.addInput(featureGroupCount.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("dimension_numbers", dimensionNumbers);
-    opBuilder.setAttr("precision_config", precisionConfig);
-    return new Conv<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "XlaConv";
-  
-  private Output<T> output;
-  
-  private Conv(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Dequantize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Dequantize.java
deleted file mode 100644
index 9cc4ca5a4f9..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Dequantize.java
+++ /dev/null
@@ -1,86 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.xla;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TBfloat16;
-
-/**
- * Takes the packed uint32 input and unpacks the input to uint8 to do
- * <p>
- * Dequantization on device.
- */
-@Operator(group = "xla")
-public final class Dequantize extends RawOp implements Operand<TBfloat16> {
-  
-  /**
-   * Factory method to create a class wrapping a new Dequantize operation.
-   * 
-   * @param scope current scope
-   * @param input Input tensors whose types is uint32, shape is [d0, ..., dn].
-   * @param minRange The minimum scalar value possibly produced for the input.
-   * @param maxRange The maximum scalar value possibly produced for the input.
-   * @param mode String to determine the dequantize mode in {"MIN_COMBINED", "MIN_FIRST", "SCALED"}.
-   * @param transposeOutput Boolean to determine if output is transposed. transpose_output
-   * is faster when input is large and rank of input is higher than 1.
-   * @return a new instance of Dequantize
-   */
-  @Endpoint(describeByClass = true)
-  public static Dequantize create(Scope scope, Operand<?> input, Float minRange, Float maxRange, String mode, Boolean transposeOutput) {
-    OperationBuilder opBuilder = scope.env().opBuilder("XlaDequantize", scope.makeOpName("Dequantize"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("min_range", minRange);
-    opBuilder.setAttr("max_range", maxRange);
-    opBuilder.setAttr("mode", mode);
-    opBuilder.setAttr("transpose_output", transposeOutput);
-    return new Dequantize(opBuilder.build());
-  }
-  
-  /**
-   * Output tensors whose types is bloat16. If transpose_output is true,
-   * output shape is [dn * 4, dn-1, ..., d1, d0]. If transpose_output
-   * is false, output shape is [d0,..., dn * 4].
-   */
-  public Output<TBfloat16> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<TBfloat16> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "XlaDequantize";
-  
-  private Output<TBfloat16> output;
-  
-  private Dequantize(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Dot.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Dot.java
deleted file mode 100644
index ee69f7bccac..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Dot.java
+++ /dev/null
@@ -1,83 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.xla;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Wraps the XLA DotGeneral operator, documented at
- * <p>
- *  https://www.tensorflow.org/performance/xla/operation_semantics#dotgeneral
- * .
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator(group = "xla")
-public final class Dot<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new Dot operation.
-   * 
-   * @param scope current scope
-   * @param lhs the LHS tensor
-   * @param rhs the RHS tensor
-   * @param dimensionNumbers a serialized xla::DotDimensionNumbers proto.
-   * @param precisionConfig a serialized xla::PrecisionConfig proto.
-   * @return a new instance of Dot
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> Dot<T> create(Scope scope, Operand<T> lhs, Operand<T> rhs, String dimensionNumbers, String precisionConfig) {
-    OperationBuilder opBuilder = scope.env().opBuilder("XlaDot", scope.makeOpName("Dot"));
-    opBuilder.addInput(lhs.asOutput(scope));
-    opBuilder.addInput(rhs.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("dimension_numbers", dimensionNumbers);
-    opBuilder.setAttr("precision_config", precisionConfig);
-    return new Dot<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "XlaDot";
-  
-  private Output<T> output;
-  
-  private Dot(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/DynamicSlice.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/DynamicSlice.java
deleted file mode 100644
index 249358166e1..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/DynamicSlice.java
+++ /dev/null
@@ -1,91 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.xla;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * Wraps the XLA DynamicSlice operator, documented at
- * <p>
- *  https://www.tensorflow.org/performance/xla/operation_semantics#dynamicslice
- * .
- * <p>
- * DynamicSlice extracts a sub-array from the input array at dynamic
- * start_indices. The size of the slice in each dimension is passed in
- * size_indices, which specify the end point of exclusive slice intervals in each
- * dimension -- [start, start + size). The shape of start_indices must have rank 1,
- * with dimension size equal to the rank of operand.
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator(group = "xla")
-public final class DynamicSlice<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new DynamicSlice operation.
-   * 
-   * @param scope current scope
-   * @param input A `Tensor` of type T.
-   * @param startIndices List of N integers containing the slice size for each
-   * dimension. Each value must be strictly greater than zero, and start + size
-   * must be less than or equal to the size of the dimension to avoid
-   * implementation defined behavior.
-   * @param sizeIndices 
-   * @return a new instance of DynamicSlice
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> DynamicSlice<T> create(Scope scope, Operand<T> input, Operand<U> startIndices, Operand<U> sizeIndices) {
-    OperationBuilder opBuilder = scope.env().opBuilder("XlaDynamicSlice", scope.makeOpName("DynamicSlice"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder.addInput(startIndices.asOutput(scope));
-    opBuilder.addInput(sizeIndices.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new DynamicSlice<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "XlaDynamicSlice";
-  
-  private Output<T> output;
-  
-  private DynamicSlice(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/DynamicUpdateSlice.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/DynamicUpdateSlice.java
deleted file mode 100644
index b94824d745e..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/DynamicUpdateSlice.java
+++ /dev/null
@@ -1,91 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.xla;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * Wraps the XLA DynamicUpdateSlice operator, documented at
- * <p>
- *  https://www.tensorflow.org/performance/xla/operation_semantics#dynamicupdateslice
- * .
- * <p>
- * XlaDynamicUpdateSlice generates a result which is the value of the `input`
- * operand, with a slice update overwritten at `indices`. The shape of `update`
- * determines the shape of the sub-array of the result which is updated. The shape
- * of indices must be rank == 1, with dimension size equal to the rank of `input`.
- * <p>
- * Handling of out-of-bounds slice indices is implementation-defined.
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator(group = "xla")
-public final class DynamicUpdateSlice<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new DynamicUpdateSlice operation.
-   * 
-   * @param scope current scope
-   * @param input A `Tensor` of type T.
-   * @param update A `Tensor` of type T. Same rank as `input`.
-   * @param indices A vector of indices into `input`. Must have length equal to the rank of
-   * `input`.
-   * @return a new instance of DynamicUpdateSlice
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> DynamicUpdateSlice<T> create(Scope scope, Operand<T> input, Operand<T> update, Operand<U> indices) {
-    OperationBuilder opBuilder = scope.env().opBuilder("XlaDynamicUpdateSlice", scope.makeOpName("DynamicUpdateSlice"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder.addInput(update.asOutput(scope));
-    opBuilder.addInput(indices.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new DynamicUpdateSlice<T>(opBuilder.build());
-  }
-  
-  /**
-   * A `Tensor` of type T.
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "XlaDynamicUpdateSlice";
-  
-  private Output<T> output;
-  
-  private DynamicUpdateSlice(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Einsum.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Einsum.java
deleted file mode 100644
index 28d2a86ed00..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Einsum.java
+++ /dev/null
@@ -1,81 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.xla;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * An op which supports basic einsum op with 2 inputs and 1 output.
- * <p>
- * This op has better TPU performance since it doesn't have explicitly reshape and
- * transpose operations as tf.einsum does.
- * 
- * @param <T> data type for {@code product()} output
- */
-@Operator(group = "xla")
-public final class Einsum<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new Einsum operation.
-   * 
-   * @param scope current scope
-   * @param a 
-   * @param b 
-   * @param equation 
-   * @return a new instance of Einsum
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> Einsum<T> create(Scope scope, Operand<T> a, Operand<T> b, String equation) {
-    OperationBuilder opBuilder = scope.env().opBuilder("XlaEinsum", scope.makeOpName("Einsum"));
-    opBuilder.addInput(a.asOutput(scope));
-    opBuilder.addInput(b.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("equation", equation);
-    return new Einsum<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> product() {
-    return product;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return product;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "XlaEinsum";
-  
-  private Output<T> product;
-  
-  private Einsum(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    product = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Gather.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Gather.java
deleted file mode 100644
index 1b7d3e4e627..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Gather.java
+++ /dev/null
@@ -1,85 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.xla;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * Wraps the XLA Gather operator documented at
- * <p>
- *   https://www.tensorflow.org/xla/operation_semantics#gather
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator(group = "xla")
-public final class Gather<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new Gather operation.
-   * 
-   * @param scope current scope
-   * @param operand The array we're gathering from.
-   * @param startIndices Array containing the starting indices of the slices we gather.
-   * @param sliceSizes slice_sizes[i] is the bounds for the slice on dimension i.
-   * @param dimensionNumbers A serialized xla::GatherDimensionNumbers proto.
-   * @param indicesAreSorted Boolean indicating if the indices are sorted.
-   * @return a new instance of Gather
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> Gather<T> create(Scope scope, Operand<T> operand, Operand<U> startIndices, Operand<U> sliceSizes, String dimensionNumbers, Boolean indicesAreSorted) {
-    OperationBuilder opBuilder = scope.env().opBuilder("XlaGather", scope.makeOpName("Gather"));
-    opBuilder.addInput(operand.asOutput(scope));
-    opBuilder.addInput(startIndices.asOutput(scope));
-    opBuilder.addInput(sliceSizes.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("dimension_numbers", dimensionNumbers);
-    opBuilder.setAttr("indices_are_sorted", indicesAreSorted);
-    return new Gather<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "XlaGather";
-  
-  private Output<T> output;
-  
-  private Gather(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/KeyValueSort.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/KeyValueSort.java
deleted file mode 100644
index 60b3de8d550..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/KeyValueSort.java
+++ /dev/null
@@ -1,88 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.xla;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * Wraps the XLA Sort operator, documented at
- * <p>
- *  https://www.tensorflow.org/performance/xla/operation_semantics#sort
- * .
- * <p>
- * Sorts a tensor. Currently only sorts in ascending order are supported.
- * 
- * @param <T> data type for {@code sortedKeys()} output
- * @param <U> data type for {@code sortedValues()} output
- */
-@Operator(group = "xla")
-public final class KeyValueSort<T extends TNumber, U extends TType> extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new KeyValueSort operation.
-   * 
-   * @param scope current scope
-   * @param keys A `Tensor` of type K.
-   * @param values A `Tensor` of type V.
-   * @return a new instance of KeyValueSort
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TNumber, U extends TType> KeyValueSort<T, U> create(Scope scope, Operand<T> keys, Operand<U> values) {
-    OperationBuilder opBuilder = scope.env().opBuilder("XlaKeyValueSort", scope.makeOpName("KeyValueSort"));
-    opBuilder.addInput(keys.asOutput(scope));
-    opBuilder.addInput(values.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new KeyValueSort<T, U>(opBuilder.build());
-  }
-  
-  /**
-   * A `Tensor` of type K.
-   */
-  public Output<T> sortedKeys() {
-    return sortedKeys;
-  }
-  
-  /**
-   * A `Tensor` of type V.
-   */
-  public Output<U> sortedValues() {
-    return sortedValues;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "XlaKeyValueSort";
-  
-  private Output<T> sortedKeys;
-  private Output<U> sortedValues;
-  
-  private KeyValueSort(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    sortedKeys = operation.output(outputIdx++);
-    sortedValues = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Pad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Pad.java
deleted file mode 100644
index 65c9d257ad5..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Pad.java
+++ /dev/null
@@ -1,87 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.xla;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-/**
- * Wraps the XLA Pad operator, documented at
- * <p>
- *  https://www.tensorflow.org/performance/xla/operation_semantics#pad
- * .
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator(group = "xla")
-public final class Pad<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new Pad operation.
-   * 
-   * @param scope current scope
-   * @param input A `Tensor` of type T.
-   * @param paddingValue A scalar `Tensor` of type T.
-   * @param paddingLow the padding to apply at the start of each input dimensions
-   * @param paddingHigh the padding to apply at the end of each input dimension.
-   * @param paddingInterior the padding to apply between each input element.
-   * @return a new instance of Pad
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType, U extends TNumber> Pad<T> create(Scope scope, Operand<T> input, Operand<T> paddingValue, Operand<U> paddingLow, Operand<U> paddingHigh, Operand<U> paddingInterior) {
-    OperationBuilder opBuilder = scope.env().opBuilder("XlaPad", scope.makeOpName("Pad"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder.addInput(paddingValue.asOutput(scope));
-    opBuilder.addInput(paddingLow.asOutput(scope));
-    opBuilder.addInput(paddingHigh.asOutput(scope));
-    opBuilder.addInput(paddingInterior.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new Pad<T>(opBuilder.build());
-  }
-  
-  /**
-   * A `Tensor` of type T.
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "XlaPad";
-  
-  private Output<T> output;
-  
-  private Pad(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Recv.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Recv.java
deleted file mode 100644
index eaedb5aaeef..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Recv.java
+++ /dev/null
@@ -1,83 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.xla;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Receives the named tensor from another XLA computation. Wraps the XLA Recv
- * <p>
- * operator documented at
- *  https://www.tensorflow.org/performance/xla/operation_semantics#recv .
- * 
- * @param <T> data type for {@code tensor()} output
- */
-@Operator(group = "xla")
-public final class Recv<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new Recv operation.
-   * 
-   * @param scope current scope
-   * @param dtype The type of the tensor.
-   * @param tensorName A string key that identifies the channel.
-   * @param shape The shape of the tensor.
-   * @return a new instance of Recv
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> Recv<T> create(Scope scope, Class<T> dtype, String tensorName, Shape shape) {
-    OperationBuilder opBuilder = scope.env().opBuilder("XlaRecv", scope.makeOpName("Recv"));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("dtype", dtype);
-    opBuilder.setAttr("tensor_name", tensorName);
-    opBuilder.setAttr("shape", shape);
-    return new Recv<T>(opBuilder.build());
-  }
-  
-  /**
-   * The tensor to receive.
-   */
-  public Output<T> tensor() {
-    return tensor;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return tensor;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "XlaRecv";
-  
-  private Output<T> tensor;
-  
-  private Recv(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    tensor = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/ReplicaId.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/ReplicaId.java
deleted file mode 100644
index 4d1b8e69a2b..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/ReplicaId.java
+++ /dev/null
@@ -1,70 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.xla;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.TInt32;
-
-/**
- * Replica ID.
- */
-@Operator(group = "xla")
-public final class ReplicaId extends RawOp implements Operand<TInt32> {
-  
-  /**
-   * Factory method to create a class wrapping a new ReplicaId operation.
-   * 
-   * @param scope current scope
-   * @return a new instance of ReplicaId
-   */
-  @Endpoint(describeByClass = true)
-  public static ReplicaId create(Scope scope) {
-    OperationBuilder opBuilder = scope.env().opBuilder("XlaReplicaId", scope.makeOpName("ReplicaId"));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new ReplicaId(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<TInt32> id() {
-    return id;
-  }
-  
-  @Override
-  public Output<TInt32> asOutput(Scope scope) {
-    return id;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "XlaReplicaId";
-  
-  private Output<TInt32> id;
-  
-  private ReplicaId(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    id = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/SelfAdjointEig.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/SelfAdjointEig.java
deleted file mode 100644
index 7acb1da628b..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/SelfAdjointEig.java
+++ /dev/null
@@ -1,97 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.xla;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Computes the eigen decomposition of a batch of self-adjoint matrices
- * <p>
- * (Note: Only real inputs are supported).
- * <p>
- * Computes the eigenvalues and eigenvectors of the innermost N-by-N matrices in
- * tensor such that tensor[...,:,:] * v[..., :,i] = e[..., i] * v[...,:,i], for
- * i=0...N-1.
- * 
- * @param <T> data type for {@code w()} output
- */
-@Operator(group = "xla")
-public final class SelfAdjointEig<T extends TType> extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new SelfAdjointEig operation.
-   * 
-   * @param scope current scope
-   * @param a the input tensor.
-   * @param lower a boolean specifies whether the calculation is done with the lower
-   * triangular part or the upper triangular part.
-   * @param maxIter maximum number of sweep update, i.e., the whole lower triangular
-   * part or upper triangular part based on parameter lower. Heuristically, it has
-   * been argued that approximately logN sweeps are needed in practice (Ref: Golub &
-   * van Loan "Matrix Computation").
-   * @param epsilon the tolerance ratio.
-   * @return a new instance of SelfAdjointEig
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> SelfAdjointEig<T> create(Scope scope, Operand<T> a, Boolean lower, Long maxIter, Float epsilon) {
-    OperationBuilder opBuilder = scope.env().opBuilder("XlaSelfAdjointEig", scope.makeOpName("SelfAdjointEig"));
-    opBuilder.addInput(a.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("lower", lower);
-    opBuilder.setAttr("max_iter", maxIter);
-    opBuilder.setAttr("epsilon", epsilon);
-    return new SelfAdjointEig<T>(opBuilder.build());
-  }
-  
-  /**
-   * The eigenvalues in ascending order, each repeated according to its
-   * multiplicity.
-   */
-  public Output<T> w() {
-    return w;
-  }
-  
-  /**
-   * The column v[..., :, i] is the normalized eigenvector corresponding to the
-   * eigenvalue w[..., i].
-   */
-  public Output<T> v() {
-    return v;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "XlaSelfAdjointEig";
-  
-  private Output<T> w;
-  private Output<T> v;
-  
-  private SelfAdjointEig(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    w = operation.output(outputIdx++);
-    v = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Send.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Send.java
deleted file mode 100644
index 73cc5acbd91..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Send.java
+++ /dev/null
@@ -1,61 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.xla;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Sends the named tensor to another XLA computation. Wraps the XLA Send operator
- * <p>
- * documented at
- *  https://www.tensorflow.org/performance/xla/operation_semantics#send .
- */
-@Operator(group = "xla")
-public final class Send extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new Send operation.
-   * 
-   * @param scope current scope
-   * @param tensor The tensor to send.
-   * @param tensorName A string key that identifies the channel.
-   * @return a new instance of Send
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> Send create(Scope scope, Operand<T> tensor, String tensorName) {
-    OperationBuilder opBuilder = scope.env().opBuilder("XlaSend", scope.makeOpName("Send"));
-    opBuilder.addInput(tensor.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("tensor_name", tensorName);
-    return new Send(opBuilder.build());
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "XlaSend";
-  
-  private Send(Operation operation) {
-    super(operation);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Sharding.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Sharding.java
deleted file mode 100644
index f3a982379fc..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Sharding.java
+++ /dev/null
@@ -1,74 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.xla;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * An op which shards the input based on the given sharding attribute.
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator(group = "xla")
-public final class Sharding<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new Sharding operation.
-   * 
-   * @param scope current scope
-   * @param input 
-   * @return a new instance of Sharding
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> Sharding<T> create(Scope scope, Operand<T> input) {
-    OperationBuilder opBuilder = scope.env().opBuilder("XlaSharding", scope.makeOpName("Sharding"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new Sharding<T>(opBuilder.build());
-  }
-  
-  /**
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "XlaSharding";
-  
-  private Output<T> output;
-  
-  private Sharding(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Sort.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Sort.java
deleted file mode 100644
index 3beaee0a3e9..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Sort.java
+++ /dev/null
@@ -1,80 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.xla;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Wraps the XLA Sort operator, documented at
- * <p>
- *  https://www.tensorflow.org/performance/xla/operation_semantics#sort
- * .
- * <p>
- * Sorts a tensor. Currently only sorts in ascending order are supported.
- * 
- * @param <T> data type for {@code output()} output
- */
-@Operator(group = "xla")
-public final class Sort<T extends TType> extends RawOp implements Operand<T> {
-  
-  /**
-   * Factory method to create a class wrapping a new Sort operation.
-   * 
-   * @param scope current scope
-   * @param input A `Tensor` of type T.
-   * @return a new instance of Sort
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> Sort<T> create(Scope scope, Operand<T> input) {
-    OperationBuilder opBuilder = scope.env().opBuilder("XlaSort", scope.makeOpName("Sort"));
-    opBuilder.addInput(input.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    return new Sort<T>(opBuilder.build());
-  }
-  
-  /**
-   * A `Tensor` of type T.
-   */
-  public Output<T> output() {
-    return output;
-  }
-  
-  @Override
-  public Output<T> asOutput(Scope scope) {
-    return output;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "XlaSort";
-  
-  private Output<T> output;
-  
-  private Sort(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    output = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Svd.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Svd.java
deleted file mode 100644
index 1e21a00b7f8..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Svd.java
+++ /dev/null
@@ -1,103 +0,0 @@
-/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-=======================================================================*/
-
-// This class has been generated, DO NOT EDIT!
-
-package org.tensorflow.op.xla;
-
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.OperationBuilder;
-import org.tensorflow.Output;
-import org.tensorflow.op.RawOp;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.annotation.Endpoint;
-import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.types.family.TType;
-
-/**
- * Computes the eigen decomposition of a batch of self-adjoint matrices
- * <p>
- * (Note: Only real inputs are supported).
- * <p>
- * Computes the eigenvalues and eigenvectors of the innermost M-by-N matrices in
- * tensor such that tensor[...,:,:] = u[..., :, :] * Diag(s[..., :]) * Transpose(v[...,:,:]).
- * 
- * @param <T> data type for {@code s()} output
- */
-@Operator(group = "xla")
-public final class Svd<T extends TType> extends RawOp {
-  
-  /**
-   * Factory method to create a class wrapping a new Svd operation.
-   * 
-   * @param scope current scope
-   * @param a the input tensor.
-   * @param maxIter maximum number of sweep update, i.e., the whole lower triangular
-   * part or upper triangular part based on parameter lower. Heuristically, it has
-   * been argued that approximately log(min (M, N)) sweeps are needed in practice
-   * (Ref: Golub & van Loan "Matrix Computation").
-   * @param epsilon the tolerance ratio.
-   * @param precisionConfig a serialized xla::PrecisionConfig proto.
-   * @return a new instance of Svd
-   */
-  @Endpoint(describeByClass = true)
-  public static <T extends TType> Svd<T> create(Scope scope, Operand<T> a, Long maxIter, Float epsilon, String precisionConfig) {
-    OperationBuilder opBuilder = scope.env().opBuilder("XlaSvd", scope.makeOpName("Svd"));
-    opBuilder.addInput(a.asOutput(scope));
-    opBuilder = scope.applyControlDependencies(opBuilder);
-    opBuilder.setAttr("max_iter", maxIter);
-    opBuilder.setAttr("epsilon", epsilon);
-    opBuilder.setAttr("precision_config", precisionConfig);
-    return new Svd<T>(opBuilder.build());
-  }
-  
-  /**
-   * Singular values. The values are sorted in reverse order of magnitude, so
-   * s[..., 0] is the largest value, s[..., 1] is the second largest, etc.
-   */
-  public Output<T> s() {
-    return s;
-  }
-  
-  /**
-   * Left singular vectors.
-   */
-  public Output<T> u() {
-    return u;
-  }
-  
-  /**
-   * Right singular vectors.
-   */
-  public Output<T> v() {
-    return v;
-  }
-  
-  /** The name of this op, as known by TensorFlow core engine */
-  public static final String OP_NAME = "XlaSvd";
-  
-  private Output<T> s;
-  private Output<T> u;
-  private Output<T> v;
-  
-  private Svd(Operation operation) {
-    super(operation);
-    int outputIdx = 0;
-    s = operation.output(outputIdx++);
-    u = operation.output(outputIdx++);
-    v = operation.output(outputIdx++);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/distruntime/ClusterDef.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/distruntime/ClusterDef.java
deleted file mode 100644
index f796e47d02f..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/distruntime/ClusterDef.java
+++ /dev/null
@@ -1,865 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/cluster.proto
-
-package org.tensorflow.proto.distruntime;
-
-/**
- * <pre>
- * Defines a TensorFlow cluster as a set of jobs.
- * </pre>
- *
- * Protobuf type {@code tensorflow.ClusterDef}
- */
-public  final class ClusterDef extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.ClusterDef)
-    ClusterDefOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use ClusterDef.newBuilder() to construct.
-  private ClusterDef(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private ClusterDef() {
-    job_ = java.util.Collections.emptyList();
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new ClusterDef();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private ClusterDef(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    int mutable_bitField0_ = 0;
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 10: {
-            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
-              job_ = new java.util.ArrayList<org.tensorflow.proto.distruntime.JobDef>();
-              mutable_bitField0_ |= 0x00000001;
-            }
-            job_.add(
-                input.readMessage(org.tensorflow.proto.distruntime.JobDef.parser(), extensionRegistry));
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      if (((mutable_bitField0_ & 0x00000001) != 0)) {
-        job_ = java.util.Collections.unmodifiableList(job_);
-      }
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.distruntime.ClusterProtos.internal_static_tensorflow_ClusterDef_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.distruntime.ClusterProtos.internal_static_tensorflow_ClusterDef_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.distruntime.ClusterDef.class, org.tensorflow.proto.distruntime.ClusterDef.Builder.class);
-  }
-
-  public static final int JOB_FIELD_NUMBER = 1;
-  private java.util.List<org.tensorflow.proto.distruntime.JobDef> job_;
-  /**
-   * <pre>
-   * The jobs that comprise the cluster.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.JobDef job = 1;</code>
-   */
-  public java.util.List<org.tensorflow.proto.distruntime.JobDef> getJobList() {
-    return job_;
-  }
-  /**
-   * <pre>
-   * The jobs that comprise the cluster.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.JobDef job = 1;</code>
-   */
-  public java.util.List<? extends org.tensorflow.proto.distruntime.JobDefOrBuilder> 
-      getJobOrBuilderList() {
-    return job_;
-  }
-  /**
-   * <pre>
-   * The jobs that comprise the cluster.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.JobDef job = 1;</code>
-   */
-  public int getJobCount() {
-    return job_.size();
-  }
-  /**
-   * <pre>
-   * The jobs that comprise the cluster.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.JobDef job = 1;</code>
-   */
-  public org.tensorflow.proto.distruntime.JobDef getJob(int index) {
-    return job_.get(index);
-  }
-  /**
-   * <pre>
-   * The jobs that comprise the cluster.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.JobDef job = 1;</code>
-   */
-  public org.tensorflow.proto.distruntime.JobDefOrBuilder getJobOrBuilder(
-      int index) {
-    return job_.get(index);
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    for (int i = 0; i < job_.size(); i++) {
-      output.writeMessage(1, job_.get(i));
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    for (int i = 0; i < job_.size(); i++) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(1, job_.get(i));
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.distruntime.ClusterDef)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.distruntime.ClusterDef other = (org.tensorflow.proto.distruntime.ClusterDef) obj;
-
-    if (!getJobList()
-        .equals(other.getJobList())) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    if (getJobCount() > 0) {
-      hash = (37 * hash) + JOB_FIELD_NUMBER;
-      hash = (53 * hash) + getJobList().hashCode();
-    }
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.distruntime.ClusterDef parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.distruntime.ClusterDef parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.distruntime.ClusterDef parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.distruntime.ClusterDef parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.distruntime.ClusterDef parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.distruntime.ClusterDef parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.distruntime.ClusterDef parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.distruntime.ClusterDef parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.distruntime.ClusterDef parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.distruntime.ClusterDef parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.distruntime.ClusterDef parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.distruntime.ClusterDef parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.distruntime.ClusterDef prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * <pre>
-   * Defines a TensorFlow cluster as a set of jobs.
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.ClusterDef}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.ClusterDef)
-      org.tensorflow.proto.distruntime.ClusterDefOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.distruntime.ClusterProtos.internal_static_tensorflow_ClusterDef_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.distruntime.ClusterProtos.internal_static_tensorflow_ClusterDef_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.distruntime.ClusterDef.class, org.tensorflow.proto.distruntime.ClusterDef.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.distruntime.ClusterDef.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-        getJobFieldBuilder();
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      if (jobBuilder_ == null) {
-        job_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000001);
-      } else {
-        jobBuilder_.clear();
-      }
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.distruntime.ClusterProtos.internal_static_tensorflow_ClusterDef_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.distruntime.ClusterDef getDefaultInstanceForType() {
-      return org.tensorflow.proto.distruntime.ClusterDef.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.distruntime.ClusterDef build() {
-      org.tensorflow.proto.distruntime.ClusterDef result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.distruntime.ClusterDef buildPartial() {
-      org.tensorflow.proto.distruntime.ClusterDef result = new org.tensorflow.proto.distruntime.ClusterDef(this);
-      int from_bitField0_ = bitField0_;
-      if (jobBuilder_ == null) {
-        if (((bitField0_ & 0x00000001) != 0)) {
-          job_ = java.util.Collections.unmodifiableList(job_);
-          bitField0_ = (bitField0_ & ~0x00000001);
-        }
-        result.job_ = job_;
-      } else {
-        result.job_ = jobBuilder_.build();
-      }
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.distruntime.ClusterDef) {
-        return mergeFrom((org.tensorflow.proto.distruntime.ClusterDef)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.distruntime.ClusterDef other) {
-      if (other == org.tensorflow.proto.distruntime.ClusterDef.getDefaultInstance()) return this;
-      if (jobBuilder_ == null) {
-        if (!other.job_.isEmpty()) {
-          if (job_.isEmpty()) {
-            job_ = other.job_;
-            bitField0_ = (bitField0_ & ~0x00000001);
-          } else {
-            ensureJobIsMutable();
-            job_.addAll(other.job_);
-          }
-          onChanged();
-        }
-      } else {
-        if (!other.job_.isEmpty()) {
-          if (jobBuilder_.isEmpty()) {
-            jobBuilder_.dispose();
-            jobBuilder_ = null;
-            job_ = other.job_;
-            bitField0_ = (bitField0_ & ~0x00000001);
-            jobBuilder_ = 
-              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
-                 getJobFieldBuilder() : null;
-          } else {
-            jobBuilder_.addAllMessages(other.job_);
-          }
-        }
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.distruntime.ClusterDef parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.distruntime.ClusterDef) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-    private int bitField0_;
-
-    private java.util.List<org.tensorflow.proto.distruntime.JobDef> job_ =
-      java.util.Collections.emptyList();
-    private void ensureJobIsMutable() {
-      if (!((bitField0_ & 0x00000001) != 0)) {
-        job_ = new java.util.ArrayList<org.tensorflow.proto.distruntime.JobDef>(job_);
-        bitField0_ |= 0x00000001;
-       }
-    }
-
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.distruntime.JobDef, org.tensorflow.proto.distruntime.JobDef.Builder, org.tensorflow.proto.distruntime.JobDefOrBuilder> jobBuilder_;
-
-    /**
-     * <pre>
-     * The jobs that comprise the cluster.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.JobDef job = 1;</code>
-     */
-    public java.util.List<org.tensorflow.proto.distruntime.JobDef> getJobList() {
-      if (jobBuilder_ == null) {
-        return java.util.Collections.unmodifiableList(job_);
-      } else {
-        return jobBuilder_.getMessageList();
-      }
-    }
-    /**
-     * <pre>
-     * The jobs that comprise the cluster.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.JobDef job = 1;</code>
-     */
-    public int getJobCount() {
-      if (jobBuilder_ == null) {
-        return job_.size();
-      } else {
-        return jobBuilder_.getCount();
-      }
-    }
-    /**
-     * <pre>
-     * The jobs that comprise the cluster.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.JobDef job = 1;</code>
-     */
-    public org.tensorflow.proto.distruntime.JobDef getJob(int index) {
-      if (jobBuilder_ == null) {
-        return job_.get(index);
-      } else {
-        return jobBuilder_.getMessage(index);
-      }
-    }
-    /**
-     * <pre>
-     * The jobs that comprise the cluster.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.JobDef job = 1;</code>
-     */
-    public Builder setJob(
-        int index, org.tensorflow.proto.distruntime.JobDef value) {
-      if (jobBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureJobIsMutable();
-        job_.set(index, value);
-        onChanged();
-      } else {
-        jobBuilder_.setMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * The jobs that comprise the cluster.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.JobDef job = 1;</code>
-     */
-    public Builder setJob(
-        int index, org.tensorflow.proto.distruntime.JobDef.Builder builderForValue) {
-      if (jobBuilder_ == null) {
-        ensureJobIsMutable();
-        job_.set(index, builderForValue.build());
-        onChanged();
-      } else {
-        jobBuilder_.setMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * The jobs that comprise the cluster.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.JobDef job = 1;</code>
-     */
-    public Builder addJob(org.tensorflow.proto.distruntime.JobDef value) {
-      if (jobBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureJobIsMutable();
-        job_.add(value);
-        onChanged();
-      } else {
-        jobBuilder_.addMessage(value);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * The jobs that comprise the cluster.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.JobDef job = 1;</code>
-     */
-    public Builder addJob(
-        int index, org.tensorflow.proto.distruntime.JobDef value) {
-      if (jobBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureJobIsMutable();
-        job_.add(index, value);
-        onChanged();
-      } else {
-        jobBuilder_.addMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * The jobs that comprise the cluster.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.JobDef job = 1;</code>
-     */
-    public Builder addJob(
-        org.tensorflow.proto.distruntime.JobDef.Builder builderForValue) {
-      if (jobBuilder_ == null) {
-        ensureJobIsMutable();
-        job_.add(builderForValue.build());
-        onChanged();
-      } else {
-        jobBuilder_.addMessage(builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * The jobs that comprise the cluster.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.JobDef job = 1;</code>
-     */
-    public Builder addJob(
-        int index, org.tensorflow.proto.distruntime.JobDef.Builder builderForValue) {
-      if (jobBuilder_ == null) {
-        ensureJobIsMutable();
-        job_.add(index, builderForValue.build());
-        onChanged();
-      } else {
-        jobBuilder_.addMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * The jobs that comprise the cluster.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.JobDef job = 1;</code>
-     */
-    public Builder addAllJob(
-        java.lang.Iterable<? extends org.tensorflow.proto.distruntime.JobDef> values) {
-      if (jobBuilder_ == null) {
-        ensureJobIsMutable();
-        com.google.protobuf.AbstractMessageLite.Builder.addAll(
-            values, job_);
-        onChanged();
-      } else {
-        jobBuilder_.addAllMessages(values);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * The jobs that comprise the cluster.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.JobDef job = 1;</code>
-     */
-    public Builder clearJob() {
-      if (jobBuilder_ == null) {
-        job_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000001);
-        onChanged();
-      } else {
-        jobBuilder_.clear();
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * The jobs that comprise the cluster.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.JobDef job = 1;</code>
-     */
-    public Builder removeJob(int index) {
-      if (jobBuilder_ == null) {
-        ensureJobIsMutable();
-        job_.remove(index);
-        onChanged();
-      } else {
-        jobBuilder_.remove(index);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * The jobs that comprise the cluster.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.JobDef job = 1;</code>
-     */
-    public org.tensorflow.proto.distruntime.JobDef.Builder getJobBuilder(
-        int index) {
-      return getJobFieldBuilder().getBuilder(index);
-    }
-    /**
-     * <pre>
-     * The jobs that comprise the cluster.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.JobDef job = 1;</code>
-     */
-    public org.tensorflow.proto.distruntime.JobDefOrBuilder getJobOrBuilder(
-        int index) {
-      if (jobBuilder_ == null) {
-        return job_.get(index);  } else {
-        return jobBuilder_.getMessageOrBuilder(index);
-      }
-    }
-    /**
-     * <pre>
-     * The jobs that comprise the cluster.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.JobDef job = 1;</code>
-     */
-    public java.util.List<? extends org.tensorflow.proto.distruntime.JobDefOrBuilder> 
-         getJobOrBuilderList() {
-      if (jobBuilder_ != null) {
-        return jobBuilder_.getMessageOrBuilderList();
-      } else {
-        return java.util.Collections.unmodifiableList(job_);
-      }
-    }
-    /**
-     * <pre>
-     * The jobs that comprise the cluster.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.JobDef job = 1;</code>
-     */
-    public org.tensorflow.proto.distruntime.JobDef.Builder addJobBuilder() {
-      return getJobFieldBuilder().addBuilder(
-          org.tensorflow.proto.distruntime.JobDef.getDefaultInstance());
-    }
-    /**
-     * <pre>
-     * The jobs that comprise the cluster.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.JobDef job = 1;</code>
-     */
-    public org.tensorflow.proto.distruntime.JobDef.Builder addJobBuilder(
-        int index) {
-      return getJobFieldBuilder().addBuilder(
-          index, org.tensorflow.proto.distruntime.JobDef.getDefaultInstance());
-    }
-    /**
-     * <pre>
-     * The jobs that comprise the cluster.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.JobDef job = 1;</code>
-     */
-    public java.util.List<org.tensorflow.proto.distruntime.JobDef.Builder> 
-         getJobBuilderList() {
-      return getJobFieldBuilder().getBuilderList();
-    }
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.distruntime.JobDef, org.tensorflow.proto.distruntime.JobDef.Builder, org.tensorflow.proto.distruntime.JobDefOrBuilder> 
-        getJobFieldBuilder() {
-      if (jobBuilder_ == null) {
-        jobBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
-            org.tensorflow.proto.distruntime.JobDef, org.tensorflow.proto.distruntime.JobDef.Builder, org.tensorflow.proto.distruntime.JobDefOrBuilder>(
-                job_,
-                ((bitField0_ & 0x00000001) != 0),
-                getParentForChildren(),
-                isClean());
-        job_ = null;
-      }
-      return jobBuilder_;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.ClusterDef)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.ClusterDef)
-  private static final org.tensorflow.proto.distruntime.ClusterDef DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.distruntime.ClusterDef();
-  }
-
-  public static org.tensorflow.proto.distruntime.ClusterDef getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<ClusterDef>
-      PARSER = new com.google.protobuf.AbstractParser<ClusterDef>() {
-    @java.lang.Override
-    public ClusterDef parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new ClusterDef(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<ClusterDef> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<ClusterDef> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.distruntime.ClusterDef getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/distruntime/ClusterDefOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/distruntime/ClusterDefOrBuilder.java
deleted file mode 100644
index 8cecd262cc2..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/distruntime/ClusterDefOrBuilder.java
+++ /dev/null
@@ -1,53 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/cluster.proto
-
-package org.tensorflow.proto.distruntime;
-
-public interface ClusterDefOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.ClusterDef)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <pre>
-   * The jobs that comprise the cluster.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.JobDef job = 1;</code>
-   */
-  java.util.List<org.tensorflow.proto.distruntime.JobDef> 
-      getJobList();
-  /**
-   * <pre>
-   * The jobs that comprise the cluster.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.JobDef job = 1;</code>
-   */
-  org.tensorflow.proto.distruntime.JobDef getJob(int index);
-  /**
-   * <pre>
-   * The jobs that comprise the cluster.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.JobDef job = 1;</code>
-   */
-  int getJobCount();
-  /**
-   * <pre>
-   * The jobs that comprise the cluster.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.JobDef job = 1;</code>
-   */
-  java.util.List<? extends org.tensorflow.proto.distruntime.JobDefOrBuilder> 
-      getJobOrBuilderList();
-  /**
-   * <pre>
-   * The jobs that comprise the cluster.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.JobDef job = 1;</code>
-   */
-  org.tensorflow.proto.distruntime.JobDefOrBuilder getJobOrBuilder(
-      int index);
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/distruntime/ClusterDeviceFilters.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/distruntime/ClusterDeviceFilters.java
deleted file mode 100644
index f9684c7e0e8..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/distruntime/ClusterDeviceFilters.java
+++ /dev/null
@@ -1,773 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/device_filters.proto
-
-package org.tensorflow.proto.distruntime;
-
-/**
- * <pre>
- * Defines the device filters for jobs in a cluster.
- * </pre>
- *
- * Protobuf type {@code tensorflow.ClusterDeviceFilters}
- */
-public  final class ClusterDeviceFilters extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.ClusterDeviceFilters)
-    ClusterDeviceFiltersOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use ClusterDeviceFilters.newBuilder() to construct.
-  private ClusterDeviceFilters(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private ClusterDeviceFilters() {
-    jobs_ = java.util.Collections.emptyList();
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new ClusterDeviceFilters();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private ClusterDeviceFilters(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    int mutable_bitField0_ = 0;
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 10: {
-            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
-              jobs_ = new java.util.ArrayList<org.tensorflow.proto.distruntime.JobDeviceFilters>();
-              mutable_bitField0_ |= 0x00000001;
-            }
-            jobs_.add(
-                input.readMessage(org.tensorflow.proto.distruntime.JobDeviceFilters.parser(), extensionRegistry));
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      if (((mutable_bitField0_ & 0x00000001) != 0)) {
-        jobs_ = java.util.Collections.unmodifiableList(jobs_);
-      }
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.distruntime.DeviceFiltersProtos.internal_static_tensorflow_ClusterDeviceFilters_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.distruntime.DeviceFiltersProtos.internal_static_tensorflow_ClusterDeviceFilters_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.distruntime.ClusterDeviceFilters.class, org.tensorflow.proto.distruntime.ClusterDeviceFilters.Builder.class);
-  }
-
-  public static final int JOBS_FIELD_NUMBER = 1;
-  private java.util.List<org.tensorflow.proto.distruntime.JobDeviceFilters> jobs_;
-  /**
-   * <code>repeated .tensorflow.JobDeviceFilters jobs = 1;</code>
-   */
-  public java.util.List<org.tensorflow.proto.distruntime.JobDeviceFilters> getJobsList() {
-    return jobs_;
-  }
-  /**
-   * <code>repeated .tensorflow.JobDeviceFilters jobs = 1;</code>
-   */
-  public java.util.List<? extends org.tensorflow.proto.distruntime.JobDeviceFiltersOrBuilder> 
-      getJobsOrBuilderList() {
-    return jobs_;
-  }
-  /**
-   * <code>repeated .tensorflow.JobDeviceFilters jobs = 1;</code>
-   */
-  public int getJobsCount() {
-    return jobs_.size();
-  }
-  /**
-   * <code>repeated .tensorflow.JobDeviceFilters jobs = 1;</code>
-   */
-  public org.tensorflow.proto.distruntime.JobDeviceFilters getJobs(int index) {
-    return jobs_.get(index);
-  }
-  /**
-   * <code>repeated .tensorflow.JobDeviceFilters jobs = 1;</code>
-   */
-  public org.tensorflow.proto.distruntime.JobDeviceFiltersOrBuilder getJobsOrBuilder(
-      int index) {
-    return jobs_.get(index);
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    for (int i = 0; i < jobs_.size(); i++) {
-      output.writeMessage(1, jobs_.get(i));
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    for (int i = 0; i < jobs_.size(); i++) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(1, jobs_.get(i));
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.distruntime.ClusterDeviceFilters)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.distruntime.ClusterDeviceFilters other = (org.tensorflow.proto.distruntime.ClusterDeviceFilters) obj;
-
-    if (!getJobsList()
-        .equals(other.getJobsList())) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    if (getJobsCount() > 0) {
-      hash = (37 * hash) + JOBS_FIELD_NUMBER;
-      hash = (53 * hash) + getJobsList().hashCode();
-    }
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.distruntime.ClusterDeviceFilters parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.distruntime.ClusterDeviceFilters parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.distruntime.ClusterDeviceFilters parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.distruntime.ClusterDeviceFilters parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.distruntime.ClusterDeviceFilters parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.distruntime.ClusterDeviceFilters parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.distruntime.ClusterDeviceFilters parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.distruntime.ClusterDeviceFilters parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.distruntime.ClusterDeviceFilters parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.distruntime.ClusterDeviceFilters parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.distruntime.ClusterDeviceFilters parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.distruntime.ClusterDeviceFilters parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.distruntime.ClusterDeviceFilters prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * <pre>
-   * Defines the device filters for jobs in a cluster.
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.ClusterDeviceFilters}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.ClusterDeviceFilters)
-      org.tensorflow.proto.distruntime.ClusterDeviceFiltersOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.distruntime.DeviceFiltersProtos.internal_static_tensorflow_ClusterDeviceFilters_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.distruntime.DeviceFiltersProtos.internal_static_tensorflow_ClusterDeviceFilters_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.distruntime.ClusterDeviceFilters.class, org.tensorflow.proto.distruntime.ClusterDeviceFilters.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.distruntime.ClusterDeviceFilters.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-        getJobsFieldBuilder();
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      if (jobsBuilder_ == null) {
-        jobs_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000001);
-      } else {
-        jobsBuilder_.clear();
-      }
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.distruntime.DeviceFiltersProtos.internal_static_tensorflow_ClusterDeviceFilters_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.distruntime.ClusterDeviceFilters getDefaultInstanceForType() {
-      return org.tensorflow.proto.distruntime.ClusterDeviceFilters.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.distruntime.ClusterDeviceFilters build() {
-      org.tensorflow.proto.distruntime.ClusterDeviceFilters result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.distruntime.ClusterDeviceFilters buildPartial() {
-      org.tensorflow.proto.distruntime.ClusterDeviceFilters result = new org.tensorflow.proto.distruntime.ClusterDeviceFilters(this);
-      int from_bitField0_ = bitField0_;
-      if (jobsBuilder_ == null) {
-        if (((bitField0_ & 0x00000001) != 0)) {
-          jobs_ = java.util.Collections.unmodifiableList(jobs_);
-          bitField0_ = (bitField0_ & ~0x00000001);
-        }
-        result.jobs_ = jobs_;
-      } else {
-        result.jobs_ = jobsBuilder_.build();
-      }
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.distruntime.ClusterDeviceFilters) {
-        return mergeFrom((org.tensorflow.proto.distruntime.ClusterDeviceFilters)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.distruntime.ClusterDeviceFilters other) {
-      if (other == org.tensorflow.proto.distruntime.ClusterDeviceFilters.getDefaultInstance()) return this;
-      if (jobsBuilder_ == null) {
-        if (!other.jobs_.isEmpty()) {
-          if (jobs_.isEmpty()) {
-            jobs_ = other.jobs_;
-            bitField0_ = (bitField0_ & ~0x00000001);
-          } else {
-            ensureJobsIsMutable();
-            jobs_.addAll(other.jobs_);
-          }
-          onChanged();
-        }
-      } else {
-        if (!other.jobs_.isEmpty()) {
-          if (jobsBuilder_.isEmpty()) {
-            jobsBuilder_.dispose();
-            jobsBuilder_ = null;
-            jobs_ = other.jobs_;
-            bitField0_ = (bitField0_ & ~0x00000001);
-            jobsBuilder_ = 
-              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
-                 getJobsFieldBuilder() : null;
-          } else {
-            jobsBuilder_.addAllMessages(other.jobs_);
-          }
-        }
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.distruntime.ClusterDeviceFilters parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.distruntime.ClusterDeviceFilters) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-    private int bitField0_;
-
-    private java.util.List<org.tensorflow.proto.distruntime.JobDeviceFilters> jobs_ =
-      java.util.Collections.emptyList();
-    private void ensureJobsIsMutable() {
-      if (!((bitField0_ & 0x00000001) != 0)) {
-        jobs_ = new java.util.ArrayList<org.tensorflow.proto.distruntime.JobDeviceFilters>(jobs_);
-        bitField0_ |= 0x00000001;
-       }
-    }
-
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.distruntime.JobDeviceFilters, org.tensorflow.proto.distruntime.JobDeviceFilters.Builder, org.tensorflow.proto.distruntime.JobDeviceFiltersOrBuilder> jobsBuilder_;
-
-    /**
-     * <code>repeated .tensorflow.JobDeviceFilters jobs = 1;</code>
-     */
-    public java.util.List<org.tensorflow.proto.distruntime.JobDeviceFilters> getJobsList() {
-      if (jobsBuilder_ == null) {
-        return java.util.Collections.unmodifiableList(jobs_);
-      } else {
-        return jobsBuilder_.getMessageList();
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.JobDeviceFilters jobs = 1;</code>
-     */
-    public int getJobsCount() {
-      if (jobsBuilder_ == null) {
-        return jobs_.size();
-      } else {
-        return jobsBuilder_.getCount();
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.JobDeviceFilters jobs = 1;</code>
-     */
-    public org.tensorflow.proto.distruntime.JobDeviceFilters getJobs(int index) {
-      if (jobsBuilder_ == null) {
-        return jobs_.get(index);
-      } else {
-        return jobsBuilder_.getMessage(index);
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.JobDeviceFilters jobs = 1;</code>
-     */
-    public Builder setJobs(
-        int index, org.tensorflow.proto.distruntime.JobDeviceFilters value) {
-      if (jobsBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureJobsIsMutable();
-        jobs_.set(index, value);
-        onChanged();
-      } else {
-        jobsBuilder_.setMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.JobDeviceFilters jobs = 1;</code>
-     */
-    public Builder setJobs(
-        int index, org.tensorflow.proto.distruntime.JobDeviceFilters.Builder builderForValue) {
-      if (jobsBuilder_ == null) {
-        ensureJobsIsMutable();
-        jobs_.set(index, builderForValue.build());
-        onChanged();
-      } else {
-        jobsBuilder_.setMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.JobDeviceFilters jobs = 1;</code>
-     */
-    public Builder addJobs(org.tensorflow.proto.distruntime.JobDeviceFilters value) {
-      if (jobsBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureJobsIsMutable();
-        jobs_.add(value);
-        onChanged();
-      } else {
-        jobsBuilder_.addMessage(value);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.JobDeviceFilters jobs = 1;</code>
-     */
-    public Builder addJobs(
-        int index, org.tensorflow.proto.distruntime.JobDeviceFilters value) {
-      if (jobsBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureJobsIsMutable();
-        jobs_.add(index, value);
-        onChanged();
-      } else {
-        jobsBuilder_.addMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.JobDeviceFilters jobs = 1;</code>
-     */
-    public Builder addJobs(
-        org.tensorflow.proto.distruntime.JobDeviceFilters.Builder builderForValue) {
-      if (jobsBuilder_ == null) {
-        ensureJobsIsMutable();
-        jobs_.add(builderForValue.build());
-        onChanged();
-      } else {
-        jobsBuilder_.addMessage(builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.JobDeviceFilters jobs = 1;</code>
-     */
-    public Builder addJobs(
-        int index, org.tensorflow.proto.distruntime.JobDeviceFilters.Builder builderForValue) {
-      if (jobsBuilder_ == null) {
-        ensureJobsIsMutable();
-        jobs_.add(index, builderForValue.build());
-        onChanged();
-      } else {
-        jobsBuilder_.addMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.JobDeviceFilters jobs = 1;</code>
-     */
-    public Builder addAllJobs(
-        java.lang.Iterable<? extends org.tensorflow.proto.distruntime.JobDeviceFilters> values) {
-      if (jobsBuilder_ == null) {
-        ensureJobsIsMutable();
-        com.google.protobuf.AbstractMessageLite.Builder.addAll(
-            values, jobs_);
-        onChanged();
-      } else {
-        jobsBuilder_.addAllMessages(values);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.JobDeviceFilters jobs = 1;</code>
-     */
-    public Builder clearJobs() {
-      if (jobsBuilder_ == null) {
-        jobs_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000001);
-        onChanged();
-      } else {
-        jobsBuilder_.clear();
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.JobDeviceFilters jobs = 1;</code>
-     */
-    public Builder removeJobs(int index) {
-      if (jobsBuilder_ == null) {
-        ensureJobsIsMutable();
-        jobs_.remove(index);
-        onChanged();
-      } else {
-        jobsBuilder_.remove(index);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.JobDeviceFilters jobs = 1;</code>
-     */
-    public org.tensorflow.proto.distruntime.JobDeviceFilters.Builder getJobsBuilder(
-        int index) {
-      return getJobsFieldBuilder().getBuilder(index);
-    }
-    /**
-     * <code>repeated .tensorflow.JobDeviceFilters jobs = 1;</code>
-     */
-    public org.tensorflow.proto.distruntime.JobDeviceFiltersOrBuilder getJobsOrBuilder(
-        int index) {
-      if (jobsBuilder_ == null) {
-        return jobs_.get(index);  } else {
-        return jobsBuilder_.getMessageOrBuilder(index);
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.JobDeviceFilters jobs = 1;</code>
-     */
-    public java.util.List<? extends org.tensorflow.proto.distruntime.JobDeviceFiltersOrBuilder> 
-         getJobsOrBuilderList() {
-      if (jobsBuilder_ != null) {
-        return jobsBuilder_.getMessageOrBuilderList();
-      } else {
-        return java.util.Collections.unmodifiableList(jobs_);
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.JobDeviceFilters jobs = 1;</code>
-     */
-    public org.tensorflow.proto.distruntime.JobDeviceFilters.Builder addJobsBuilder() {
-      return getJobsFieldBuilder().addBuilder(
-          org.tensorflow.proto.distruntime.JobDeviceFilters.getDefaultInstance());
-    }
-    /**
-     * <code>repeated .tensorflow.JobDeviceFilters jobs = 1;</code>
-     */
-    public org.tensorflow.proto.distruntime.JobDeviceFilters.Builder addJobsBuilder(
-        int index) {
-      return getJobsFieldBuilder().addBuilder(
-          index, org.tensorflow.proto.distruntime.JobDeviceFilters.getDefaultInstance());
-    }
-    /**
-     * <code>repeated .tensorflow.JobDeviceFilters jobs = 1;</code>
-     */
-    public java.util.List<org.tensorflow.proto.distruntime.JobDeviceFilters.Builder> 
-         getJobsBuilderList() {
-      return getJobsFieldBuilder().getBuilderList();
-    }
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.distruntime.JobDeviceFilters, org.tensorflow.proto.distruntime.JobDeviceFilters.Builder, org.tensorflow.proto.distruntime.JobDeviceFiltersOrBuilder> 
-        getJobsFieldBuilder() {
-      if (jobsBuilder_ == null) {
-        jobsBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
-            org.tensorflow.proto.distruntime.JobDeviceFilters, org.tensorflow.proto.distruntime.JobDeviceFilters.Builder, org.tensorflow.proto.distruntime.JobDeviceFiltersOrBuilder>(
-                jobs_,
-                ((bitField0_ & 0x00000001) != 0),
-                getParentForChildren(),
-                isClean());
-        jobs_ = null;
-      }
-      return jobsBuilder_;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.ClusterDeviceFilters)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.ClusterDeviceFilters)
-  private static final org.tensorflow.proto.distruntime.ClusterDeviceFilters DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.distruntime.ClusterDeviceFilters();
-  }
-
-  public static org.tensorflow.proto.distruntime.ClusterDeviceFilters getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<ClusterDeviceFilters>
-      PARSER = new com.google.protobuf.AbstractParser<ClusterDeviceFilters>() {
-    @java.lang.Override
-    public ClusterDeviceFilters parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new ClusterDeviceFilters(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<ClusterDeviceFilters> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<ClusterDeviceFilters> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.distruntime.ClusterDeviceFilters getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/distruntime/ClusterDeviceFiltersOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/distruntime/ClusterDeviceFiltersOrBuilder.java
deleted file mode 100644
index eaaf1635eec..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/distruntime/ClusterDeviceFiltersOrBuilder.java
+++ /dev/null
@@ -1,33 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/device_filters.proto
-
-package org.tensorflow.proto.distruntime;
-
-public interface ClusterDeviceFiltersOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.ClusterDeviceFilters)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <code>repeated .tensorflow.JobDeviceFilters jobs = 1;</code>
-   */
-  java.util.List<org.tensorflow.proto.distruntime.JobDeviceFilters> 
-      getJobsList();
-  /**
-   * <code>repeated .tensorflow.JobDeviceFilters jobs = 1;</code>
-   */
-  org.tensorflow.proto.distruntime.JobDeviceFilters getJobs(int index);
-  /**
-   * <code>repeated .tensorflow.JobDeviceFilters jobs = 1;</code>
-   */
-  int getJobsCount();
-  /**
-   * <code>repeated .tensorflow.JobDeviceFilters jobs = 1;</code>
-   */
-  java.util.List<? extends org.tensorflow.proto.distruntime.JobDeviceFiltersOrBuilder> 
-      getJobsOrBuilderList();
-  /**
-   * <code>repeated .tensorflow.JobDeviceFilters jobs = 1;</code>
-   */
-  org.tensorflow.proto.distruntime.JobDeviceFiltersOrBuilder getJobsOrBuilder(
-      int index);
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/distruntime/ClusterProtos.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/distruntime/ClusterProtos.java
deleted file mode 100644
index 06aa24f0049..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/distruntime/ClusterProtos.java
+++ /dev/null
@@ -1,76 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/cluster.proto
-
-package org.tensorflow.proto.distruntime;
-
-public final class ClusterProtos {
-  private ClusterProtos() {}
-  public static void registerAllExtensions(
-      com.google.protobuf.ExtensionRegistryLite registry) {
-  }
-
-  public static void registerAllExtensions(
-      com.google.protobuf.ExtensionRegistry registry) {
-    registerAllExtensions(
-        (com.google.protobuf.ExtensionRegistryLite) registry);
-  }
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_JobDef_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_JobDef_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_JobDef_TasksEntry_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_JobDef_TasksEntry_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_ClusterDef_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_ClusterDef_fieldAccessorTable;
-
-  public static com.google.protobuf.Descriptors.FileDescriptor
-      getDescriptor() {
-    return descriptor;
-  }
-  private static  com.google.protobuf.Descriptors.FileDescriptor
-      descriptor;
-  static {
-    java.lang.String[] descriptorData = {
-      "\n&tensorflow/core/protobuf/cluster.proto" +
-      "\022\ntensorflow\"r\n\006JobDef\022\014\n\004name\030\001 \001(\t\022,\n\005" +
-      "tasks\030\002 \003(\0132\035.tensorflow.JobDef.TasksEnt" +
-      "ry\032,\n\nTasksEntry\022\013\n\003key\030\001 \001(\005\022\r\n\005value\030\002" +
-      " \001(\t:\0028\001\"-\n\nClusterDef\022\037\n\003job\030\001 \003(\0132\022.te" +
-      "nsorflow.JobDefB\200\001\n org.tensorflow.proto" +
-      ".distruntimeB\rClusterProtosP\001ZHgithub.co" +
-      "m/tensorflow/tensorflow/tensorflow/go/co" +
-      "re/core_protos_go_proto\370\001\001b\006proto3"
-    };
-    descriptor = com.google.protobuf.Descriptors.FileDescriptor
-      .internalBuildGeneratedFileFrom(descriptorData,
-        new com.google.protobuf.Descriptors.FileDescriptor[] {
-        });
-    internal_static_tensorflow_JobDef_descriptor =
-      getDescriptor().getMessageTypes().get(0);
-    internal_static_tensorflow_JobDef_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_JobDef_descriptor,
-        new java.lang.String[] { "Name", "Tasks", });
-    internal_static_tensorflow_JobDef_TasksEntry_descriptor =
-      internal_static_tensorflow_JobDef_descriptor.getNestedTypes().get(0);
-    internal_static_tensorflow_JobDef_TasksEntry_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_JobDef_TasksEntry_descriptor,
-        new java.lang.String[] { "Key", "Value", });
-    internal_static_tensorflow_ClusterDef_descriptor =
-      getDescriptor().getMessageTypes().get(1);
-    internal_static_tensorflow_ClusterDef_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_ClusterDef_descriptor,
-        new java.lang.String[] { "Job", });
-  }
-
-  // @@protoc_insertion_point(outer_class_scope)
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/distruntime/DeviceFiltersProtos.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/distruntime/DeviceFiltersProtos.java
deleted file mode 100644
index 254d82a2a2a..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/distruntime/DeviceFiltersProtos.java
+++ /dev/null
@@ -1,91 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/device_filters.proto
-
-package org.tensorflow.proto.distruntime;
-
-public final class DeviceFiltersProtos {
-  private DeviceFiltersProtos() {}
-  public static void registerAllExtensions(
-      com.google.protobuf.ExtensionRegistryLite registry) {
-  }
-
-  public static void registerAllExtensions(
-      com.google.protobuf.ExtensionRegistry registry) {
-    registerAllExtensions(
-        (com.google.protobuf.ExtensionRegistryLite) registry);
-  }
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_TaskDeviceFilters_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_TaskDeviceFilters_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_JobDeviceFilters_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_JobDeviceFilters_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_JobDeviceFilters_TasksEntry_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_JobDeviceFilters_TasksEntry_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_ClusterDeviceFilters_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_ClusterDeviceFilters_fieldAccessorTable;
-
-  public static com.google.protobuf.Descriptors.FileDescriptor
-      getDescriptor() {
-    return descriptor;
-  }
-  private static  com.google.protobuf.Descriptors.FileDescriptor
-      descriptor;
-  static {
-    java.lang.String[] descriptorData = {
-      "\n-tensorflow/core/protobuf/device_filter" +
-      "s.proto\022\ntensorflow\"+\n\021TaskDeviceFilters" +
-      "\022\026\n\016device_filters\030\001 \003(\t\"\245\001\n\020JobDeviceFi" +
-      "lters\022\014\n\004name\030\001 \001(\t\0226\n\005tasks\030\002 \003(\0132\'.ten" +
-      "sorflow.JobDeviceFilters.TasksEntry\032K\n\nT" +
-      "asksEntry\022\013\n\003key\030\001 \001(\005\022,\n\005value\030\002 \001(\0132\035." +
-      "tensorflow.TaskDeviceFilters:\0028\001\"B\n\024Clus" +
-      "terDeviceFilters\022*\n\004jobs\030\001 \003(\0132\034.tensorf" +
-      "low.JobDeviceFiltersB\206\001\n org.tensorflow." +
-      "proto.distruntimeB\023DeviceFiltersProtosP\001" +
-      "ZHgithub.com/tensorflow/tensorflow/tenso" +
-      "rflow/go/core/core_protos_go_proto\370\001\001b\006p" +
-      "roto3"
-    };
-    descriptor = com.google.protobuf.Descriptors.FileDescriptor
-      .internalBuildGeneratedFileFrom(descriptorData,
-        new com.google.protobuf.Descriptors.FileDescriptor[] {
-        });
-    internal_static_tensorflow_TaskDeviceFilters_descriptor =
-      getDescriptor().getMessageTypes().get(0);
-    internal_static_tensorflow_TaskDeviceFilters_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_TaskDeviceFilters_descriptor,
-        new java.lang.String[] { "DeviceFilters", });
-    internal_static_tensorflow_JobDeviceFilters_descriptor =
-      getDescriptor().getMessageTypes().get(1);
-    internal_static_tensorflow_JobDeviceFilters_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_JobDeviceFilters_descriptor,
-        new java.lang.String[] { "Name", "Tasks", });
-    internal_static_tensorflow_JobDeviceFilters_TasksEntry_descriptor =
-      internal_static_tensorflow_JobDeviceFilters_descriptor.getNestedTypes().get(0);
-    internal_static_tensorflow_JobDeviceFilters_TasksEntry_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_JobDeviceFilters_TasksEntry_descriptor,
-        new java.lang.String[] { "Key", "Value", });
-    internal_static_tensorflow_ClusterDeviceFilters_descriptor =
-      getDescriptor().getMessageTypes().get(2);
-    internal_static_tensorflow_ClusterDeviceFilters_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_ClusterDeviceFilters_descriptor,
-        new java.lang.String[] { "Jobs", });
-  }
-
-  // @@protoc_insertion_point(outer_class_scope)
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/distruntime/JobDef.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/distruntime/JobDef.java
deleted file mode 100644
index c6766412507..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/distruntime/JobDef.java
+++ /dev/null
@@ -1,935 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/cluster.proto
-
-package org.tensorflow.proto.distruntime;
-
-/**
- * <pre>
- * Defines a single job in a TensorFlow cluster.
- * </pre>
- *
- * Protobuf type {@code tensorflow.JobDef}
- */
-public  final class JobDef extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.JobDef)
-    JobDefOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use JobDef.newBuilder() to construct.
-  private JobDef(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private JobDef() {
-    name_ = "";
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new JobDef();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private JobDef(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    int mutable_bitField0_ = 0;
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 10: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            name_ = s;
-            break;
-          }
-          case 18: {
-            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
-              tasks_ = com.google.protobuf.MapField.newMapField(
-                  TasksDefaultEntryHolder.defaultEntry);
-              mutable_bitField0_ |= 0x00000001;
-            }
-            com.google.protobuf.MapEntry<java.lang.Integer, java.lang.String>
-            tasks__ = input.readMessage(
-                TasksDefaultEntryHolder.defaultEntry.getParserForType(), extensionRegistry);
-            tasks_.getMutableMap().put(
-                tasks__.getKey(), tasks__.getValue());
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.distruntime.ClusterProtos.internal_static_tensorflow_JobDef_descriptor;
-  }
-
-  @SuppressWarnings({"rawtypes"})
-  @java.lang.Override
-  protected com.google.protobuf.MapField internalGetMapField(
-      int number) {
-    switch (number) {
-      case 2:
-        return internalGetTasks();
-      default:
-        throw new RuntimeException(
-            "Invalid map field number: " + number);
-    }
-  }
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.distruntime.ClusterProtos.internal_static_tensorflow_JobDef_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.distruntime.JobDef.class, org.tensorflow.proto.distruntime.JobDef.Builder.class);
-  }
-
-  public static final int NAME_FIELD_NUMBER = 1;
-  private volatile java.lang.Object name_;
-  /**
-   * <pre>
-   * The name of this job.
-   * </pre>
-   *
-   * <code>string name = 1;</code>
-   */
-  public java.lang.String getName() {
-    java.lang.Object ref = name_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      name_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * The name of this job.
-   * </pre>
-   *
-   * <code>string name = 1;</code>
-   */
-  public com.google.protobuf.ByteString
-      getNameBytes() {
-    java.lang.Object ref = name_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      name_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int TASKS_FIELD_NUMBER = 2;
-  private static final class TasksDefaultEntryHolder {
-    static final com.google.protobuf.MapEntry<
-        java.lang.Integer, java.lang.String> defaultEntry =
-            com.google.protobuf.MapEntry
-            .<java.lang.Integer, java.lang.String>newDefaultInstance(
-                org.tensorflow.proto.distruntime.ClusterProtos.internal_static_tensorflow_JobDef_TasksEntry_descriptor, 
-                com.google.protobuf.WireFormat.FieldType.INT32,
-                0,
-                com.google.protobuf.WireFormat.FieldType.STRING,
-                "");
-  }
-  private com.google.protobuf.MapField<
-      java.lang.Integer, java.lang.String> tasks_;
-  private com.google.protobuf.MapField<java.lang.Integer, java.lang.String>
-  internalGetTasks() {
-    if (tasks_ == null) {
-      return com.google.protobuf.MapField.emptyMapField(
-          TasksDefaultEntryHolder.defaultEntry);
-    }
-    return tasks_;
-  }
-
-  public int getTasksCount() {
-    return internalGetTasks().getMap().size();
-  }
-  /**
-   * <pre>
-   * Mapping from task ID to "hostname:port" string.
-   * If the `name` field contains "worker", and the `tasks` map contains a
-   * mapping from 7 to "example.org:2222", then the device prefix
-   * "/job:worker/task:7" will be assigned to "example.org:2222".
-   * </pre>
-   *
-   * <code>map&lt;int32, string&gt; tasks = 2;</code>
-   */
-
-  public boolean containsTasks(
-      int key) {
-    
-    return internalGetTasks().getMap().containsKey(key);
-  }
-  /**
-   * Use {@link #getTasksMap()} instead.
-   */
-  @java.lang.Deprecated
-  public java.util.Map<java.lang.Integer, java.lang.String> getTasks() {
-    return getTasksMap();
-  }
-  /**
-   * <pre>
-   * Mapping from task ID to "hostname:port" string.
-   * If the `name` field contains "worker", and the `tasks` map contains a
-   * mapping from 7 to "example.org:2222", then the device prefix
-   * "/job:worker/task:7" will be assigned to "example.org:2222".
-   * </pre>
-   *
-   * <code>map&lt;int32, string&gt; tasks = 2;</code>
-   */
-
-  public java.util.Map<java.lang.Integer, java.lang.String> getTasksMap() {
-    return internalGetTasks().getMap();
-  }
-  /**
-   * <pre>
-   * Mapping from task ID to "hostname:port" string.
-   * If the `name` field contains "worker", and the `tasks` map contains a
-   * mapping from 7 to "example.org:2222", then the device prefix
-   * "/job:worker/task:7" will be assigned to "example.org:2222".
-   * </pre>
-   *
-   * <code>map&lt;int32, string&gt; tasks = 2;</code>
-   */
-
-  public java.lang.String getTasksOrDefault(
-      int key,
-      java.lang.String defaultValue) {
-    
-    java.util.Map<java.lang.Integer, java.lang.String> map =
-        internalGetTasks().getMap();
-    return map.containsKey(key) ? map.get(key) : defaultValue;
-  }
-  /**
-   * <pre>
-   * Mapping from task ID to "hostname:port" string.
-   * If the `name` field contains "worker", and the `tasks` map contains a
-   * mapping from 7 to "example.org:2222", then the device prefix
-   * "/job:worker/task:7" will be assigned to "example.org:2222".
-   * </pre>
-   *
-   * <code>map&lt;int32, string&gt; tasks = 2;</code>
-   */
-
-  public java.lang.String getTasksOrThrow(
-      int key) {
-    
-    java.util.Map<java.lang.Integer, java.lang.String> map =
-        internalGetTasks().getMap();
-    if (!map.containsKey(key)) {
-      throw new java.lang.IllegalArgumentException();
-    }
-    return map.get(key);
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    if (!getNameBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 1, name_);
-    }
-    com.google.protobuf.GeneratedMessageV3
-      .serializeIntegerMapTo(
-        output,
-        internalGetTasks(),
-        TasksDefaultEntryHolder.defaultEntry,
-        2);
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (!getNameBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, name_);
-    }
-    for (java.util.Map.Entry<java.lang.Integer, java.lang.String> entry
-         : internalGetTasks().getMap().entrySet()) {
-      com.google.protobuf.MapEntry<java.lang.Integer, java.lang.String>
-      tasks__ = TasksDefaultEntryHolder.defaultEntry.newBuilderForType()
-          .setKey(entry.getKey())
-          .setValue(entry.getValue())
-          .build();
-      size += com.google.protobuf.CodedOutputStream
-          .computeMessageSize(2, tasks__);
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.distruntime.JobDef)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.distruntime.JobDef other = (org.tensorflow.proto.distruntime.JobDef) obj;
-
-    if (!getName()
-        .equals(other.getName())) return false;
-    if (!internalGetTasks().equals(
-        other.internalGetTasks())) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    hash = (37 * hash) + NAME_FIELD_NUMBER;
-    hash = (53 * hash) + getName().hashCode();
-    if (!internalGetTasks().getMap().isEmpty()) {
-      hash = (37 * hash) + TASKS_FIELD_NUMBER;
-      hash = (53 * hash) + internalGetTasks().hashCode();
-    }
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.distruntime.JobDef parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.distruntime.JobDef parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.distruntime.JobDef parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.distruntime.JobDef parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.distruntime.JobDef parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.distruntime.JobDef parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.distruntime.JobDef parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.distruntime.JobDef parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.distruntime.JobDef parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.distruntime.JobDef parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.distruntime.JobDef parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.distruntime.JobDef parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.distruntime.JobDef prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * <pre>
-   * Defines a single job in a TensorFlow cluster.
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.JobDef}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.JobDef)
-      org.tensorflow.proto.distruntime.JobDefOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.distruntime.ClusterProtos.internal_static_tensorflow_JobDef_descriptor;
-    }
-
-    @SuppressWarnings({"rawtypes"})
-    protected com.google.protobuf.MapField internalGetMapField(
-        int number) {
-      switch (number) {
-        case 2:
-          return internalGetTasks();
-        default:
-          throw new RuntimeException(
-              "Invalid map field number: " + number);
-      }
-    }
-    @SuppressWarnings({"rawtypes"})
-    protected com.google.protobuf.MapField internalGetMutableMapField(
-        int number) {
-      switch (number) {
-        case 2:
-          return internalGetMutableTasks();
-        default:
-          throw new RuntimeException(
-              "Invalid map field number: " + number);
-      }
-    }
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.distruntime.ClusterProtos.internal_static_tensorflow_JobDef_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.distruntime.JobDef.class, org.tensorflow.proto.distruntime.JobDef.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.distruntime.JobDef.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      name_ = "";
-
-      internalGetMutableTasks().clear();
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.distruntime.ClusterProtos.internal_static_tensorflow_JobDef_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.distruntime.JobDef getDefaultInstanceForType() {
-      return org.tensorflow.proto.distruntime.JobDef.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.distruntime.JobDef build() {
-      org.tensorflow.proto.distruntime.JobDef result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.distruntime.JobDef buildPartial() {
-      org.tensorflow.proto.distruntime.JobDef result = new org.tensorflow.proto.distruntime.JobDef(this);
-      int from_bitField0_ = bitField0_;
-      result.name_ = name_;
-      result.tasks_ = internalGetTasks();
-      result.tasks_.makeImmutable();
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.distruntime.JobDef) {
-        return mergeFrom((org.tensorflow.proto.distruntime.JobDef)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.distruntime.JobDef other) {
-      if (other == org.tensorflow.proto.distruntime.JobDef.getDefaultInstance()) return this;
-      if (!other.getName().isEmpty()) {
-        name_ = other.name_;
-        onChanged();
-      }
-      internalGetMutableTasks().mergeFrom(
-          other.internalGetTasks());
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.distruntime.JobDef parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.distruntime.JobDef) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-    private int bitField0_;
-
-    private java.lang.Object name_ = "";
-    /**
-     * <pre>
-     * The name of this job.
-     * </pre>
-     *
-     * <code>string name = 1;</code>
-     */
-    public java.lang.String getName() {
-      java.lang.Object ref = name_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        name_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * The name of this job.
-     * </pre>
-     *
-     * <code>string name = 1;</code>
-     */
-    public com.google.protobuf.ByteString
-        getNameBytes() {
-      java.lang.Object ref = name_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        name_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * The name of this job.
-     * </pre>
-     *
-     * <code>string name = 1;</code>
-     */
-    public Builder setName(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      name_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * The name of this job.
-     * </pre>
-     *
-     * <code>string name = 1;</code>
-     */
-    public Builder clearName() {
-      
-      name_ = getDefaultInstance().getName();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * The name of this job.
-     * </pre>
-     *
-     * <code>string name = 1;</code>
-     */
-    public Builder setNameBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      name_ = value;
-      onChanged();
-      return this;
-    }
-
-    private com.google.protobuf.MapField<
-        java.lang.Integer, java.lang.String> tasks_;
-    private com.google.protobuf.MapField<java.lang.Integer, java.lang.String>
-    internalGetTasks() {
-      if (tasks_ == null) {
-        return com.google.protobuf.MapField.emptyMapField(
-            TasksDefaultEntryHolder.defaultEntry);
-      }
-      return tasks_;
-    }
-    private com.google.protobuf.MapField<java.lang.Integer, java.lang.String>
-    internalGetMutableTasks() {
-      onChanged();;
-      if (tasks_ == null) {
-        tasks_ = com.google.protobuf.MapField.newMapField(
-            TasksDefaultEntryHolder.defaultEntry);
-      }
-      if (!tasks_.isMutable()) {
-        tasks_ = tasks_.copy();
-      }
-      return tasks_;
-    }
-
-    public int getTasksCount() {
-      return internalGetTasks().getMap().size();
-    }
-    /**
-     * <pre>
-     * Mapping from task ID to "hostname:port" string.
-     * If the `name` field contains "worker", and the `tasks` map contains a
-     * mapping from 7 to "example.org:2222", then the device prefix
-     * "/job:worker/task:7" will be assigned to "example.org:2222".
-     * </pre>
-     *
-     * <code>map&lt;int32, string&gt; tasks = 2;</code>
-     */
-
-    public boolean containsTasks(
-        int key) {
-      
-      return internalGetTasks().getMap().containsKey(key);
-    }
-    /**
-     * Use {@link #getTasksMap()} instead.
-     */
-    @java.lang.Deprecated
-    public java.util.Map<java.lang.Integer, java.lang.String> getTasks() {
-      return getTasksMap();
-    }
-    /**
-     * <pre>
-     * Mapping from task ID to "hostname:port" string.
-     * If the `name` field contains "worker", and the `tasks` map contains a
-     * mapping from 7 to "example.org:2222", then the device prefix
-     * "/job:worker/task:7" will be assigned to "example.org:2222".
-     * </pre>
-     *
-     * <code>map&lt;int32, string&gt; tasks = 2;</code>
-     */
-
-    public java.util.Map<java.lang.Integer, java.lang.String> getTasksMap() {
-      return internalGetTasks().getMap();
-    }
-    /**
-     * <pre>
-     * Mapping from task ID to "hostname:port" string.
-     * If the `name` field contains "worker", and the `tasks` map contains a
-     * mapping from 7 to "example.org:2222", then the device prefix
-     * "/job:worker/task:7" will be assigned to "example.org:2222".
-     * </pre>
-     *
-     * <code>map&lt;int32, string&gt; tasks = 2;</code>
-     */
-
-    public java.lang.String getTasksOrDefault(
-        int key,
-        java.lang.String defaultValue) {
-      
-      java.util.Map<java.lang.Integer, java.lang.String> map =
-          internalGetTasks().getMap();
-      return map.containsKey(key) ? map.get(key) : defaultValue;
-    }
-    /**
-     * <pre>
-     * Mapping from task ID to "hostname:port" string.
-     * If the `name` field contains "worker", and the `tasks` map contains a
-     * mapping from 7 to "example.org:2222", then the device prefix
-     * "/job:worker/task:7" will be assigned to "example.org:2222".
-     * </pre>
-     *
-     * <code>map&lt;int32, string&gt; tasks = 2;</code>
-     */
-
-    public java.lang.String getTasksOrThrow(
-        int key) {
-      
-      java.util.Map<java.lang.Integer, java.lang.String> map =
-          internalGetTasks().getMap();
-      if (!map.containsKey(key)) {
-        throw new java.lang.IllegalArgumentException();
-      }
-      return map.get(key);
-    }
-
-    public Builder clearTasks() {
-      internalGetMutableTasks().getMutableMap()
-          .clear();
-      return this;
-    }
-    /**
-     * <pre>
-     * Mapping from task ID to "hostname:port" string.
-     * If the `name` field contains "worker", and the `tasks` map contains a
-     * mapping from 7 to "example.org:2222", then the device prefix
-     * "/job:worker/task:7" will be assigned to "example.org:2222".
-     * </pre>
-     *
-     * <code>map&lt;int32, string&gt; tasks = 2;</code>
-     */
-
-    public Builder removeTasks(
-        int key) {
-      
-      internalGetMutableTasks().getMutableMap()
-          .remove(key);
-      return this;
-    }
-    /**
-     * Use alternate mutation accessors instead.
-     */
-    @java.lang.Deprecated
-    public java.util.Map<java.lang.Integer, java.lang.String>
-    getMutableTasks() {
-      return internalGetMutableTasks().getMutableMap();
-    }
-    /**
-     * <pre>
-     * Mapping from task ID to "hostname:port" string.
-     * If the `name` field contains "worker", and the `tasks` map contains a
-     * mapping from 7 to "example.org:2222", then the device prefix
-     * "/job:worker/task:7" will be assigned to "example.org:2222".
-     * </pre>
-     *
-     * <code>map&lt;int32, string&gt; tasks = 2;</code>
-     */
-    public Builder putTasks(
-        int key,
-        java.lang.String value) {
-      
-      if (value == null) { throw new java.lang.NullPointerException(); }
-      internalGetMutableTasks().getMutableMap()
-          .put(key, value);
-      return this;
-    }
-    /**
-     * <pre>
-     * Mapping from task ID to "hostname:port" string.
-     * If the `name` field contains "worker", and the `tasks` map contains a
-     * mapping from 7 to "example.org:2222", then the device prefix
-     * "/job:worker/task:7" will be assigned to "example.org:2222".
-     * </pre>
-     *
-     * <code>map&lt;int32, string&gt; tasks = 2;</code>
-     */
-
-    public Builder putAllTasks(
-        java.util.Map<java.lang.Integer, java.lang.String> values) {
-      internalGetMutableTasks().getMutableMap()
-          .putAll(values);
-      return this;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.JobDef)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.JobDef)
-  private static final org.tensorflow.proto.distruntime.JobDef DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.distruntime.JobDef();
-  }
-
-  public static org.tensorflow.proto.distruntime.JobDef getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<JobDef>
-      PARSER = new com.google.protobuf.AbstractParser<JobDef>() {
-    @java.lang.Override
-    public JobDef parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new JobDef(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<JobDef> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<JobDef> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.distruntime.JobDef getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/distruntime/JobDefOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/distruntime/JobDefOrBuilder.java
deleted file mode 100644
index a9c24b98a74..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/distruntime/JobDefOrBuilder.java
+++ /dev/null
@@ -1,96 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/cluster.proto
-
-package org.tensorflow.proto.distruntime;
-
-public interface JobDefOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.JobDef)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <pre>
-   * The name of this job.
-   * </pre>
-   *
-   * <code>string name = 1;</code>
-   */
-  java.lang.String getName();
-  /**
-   * <pre>
-   * The name of this job.
-   * </pre>
-   *
-   * <code>string name = 1;</code>
-   */
-  com.google.protobuf.ByteString
-      getNameBytes();
-
-  /**
-   * <pre>
-   * Mapping from task ID to "hostname:port" string.
-   * If the `name` field contains "worker", and the `tasks` map contains a
-   * mapping from 7 to "example.org:2222", then the device prefix
-   * "/job:worker/task:7" will be assigned to "example.org:2222".
-   * </pre>
-   *
-   * <code>map&lt;int32, string&gt; tasks = 2;</code>
-   */
-  int getTasksCount();
-  /**
-   * <pre>
-   * Mapping from task ID to "hostname:port" string.
-   * If the `name` field contains "worker", and the `tasks` map contains a
-   * mapping from 7 to "example.org:2222", then the device prefix
-   * "/job:worker/task:7" will be assigned to "example.org:2222".
-   * </pre>
-   *
-   * <code>map&lt;int32, string&gt; tasks = 2;</code>
-   */
-  boolean containsTasks(
-      int key);
-  /**
-   * Use {@link #getTasksMap()} instead.
-   */
-  @java.lang.Deprecated
-  java.util.Map<java.lang.Integer, java.lang.String>
-  getTasks();
-  /**
-   * <pre>
-   * Mapping from task ID to "hostname:port" string.
-   * If the `name` field contains "worker", and the `tasks` map contains a
-   * mapping from 7 to "example.org:2222", then the device prefix
-   * "/job:worker/task:7" will be assigned to "example.org:2222".
-   * </pre>
-   *
-   * <code>map&lt;int32, string&gt; tasks = 2;</code>
-   */
-  java.util.Map<java.lang.Integer, java.lang.String>
-  getTasksMap();
-  /**
-   * <pre>
-   * Mapping from task ID to "hostname:port" string.
-   * If the `name` field contains "worker", and the `tasks` map contains a
-   * mapping from 7 to "example.org:2222", then the device prefix
-   * "/job:worker/task:7" will be assigned to "example.org:2222".
-   * </pre>
-   *
-   * <code>map&lt;int32, string&gt; tasks = 2;</code>
-   */
-
-  java.lang.String getTasksOrDefault(
-      int key,
-      java.lang.String defaultValue);
-  /**
-   * <pre>
-   * Mapping from task ID to "hostname:port" string.
-   * If the `name` field contains "worker", and the `tasks` map contains a
-   * mapping from 7 to "example.org:2222", then the device prefix
-   * "/job:worker/task:7" will be assigned to "example.org:2222".
-   * </pre>
-   *
-   * <code>map&lt;int32, string&gt; tasks = 2;</code>
-   */
-
-  java.lang.String getTasksOrThrow(
-      int key);
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/distruntime/JobDeviceFilters.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/distruntime/JobDeviceFilters.java
deleted file mode 100644
index bbc7a6542be..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/distruntime/JobDeviceFilters.java
+++ /dev/null
@@ -1,902 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/device_filters.proto
-
-package org.tensorflow.proto.distruntime;
-
-/**
- * <pre>
- * Defines the device filters for tasks in a job.
- * </pre>
- *
- * Protobuf type {@code tensorflow.JobDeviceFilters}
- */
-public  final class JobDeviceFilters extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.JobDeviceFilters)
-    JobDeviceFiltersOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use JobDeviceFilters.newBuilder() to construct.
-  private JobDeviceFilters(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private JobDeviceFilters() {
-    name_ = "";
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new JobDeviceFilters();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private JobDeviceFilters(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    int mutable_bitField0_ = 0;
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 10: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            name_ = s;
-            break;
-          }
-          case 18: {
-            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
-              tasks_ = com.google.protobuf.MapField.newMapField(
-                  TasksDefaultEntryHolder.defaultEntry);
-              mutable_bitField0_ |= 0x00000001;
-            }
-            com.google.protobuf.MapEntry<java.lang.Integer, org.tensorflow.proto.distruntime.TaskDeviceFilters>
-            tasks__ = input.readMessage(
-                TasksDefaultEntryHolder.defaultEntry.getParserForType(), extensionRegistry);
-            tasks_.getMutableMap().put(
-                tasks__.getKey(), tasks__.getValue());
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.distruntime.DeviceFiltersProtos.internal_static_tensorflow_JobDeviceFilters_descriptor;
-  }
-
-  @SuppressWarnings({"rawtypes"})
-  @java.lang.Override
-  protected com.google.protobuf.MapField internalGetMapField(
-      int number) {
-    switch (number) {
-      case 2:
-        return internalGetTasks();
-      default:
-        throw new RuntimeException(
-            "Invalid map field number: " + number);
-    }
-  }
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.distruntime.DeviceFiltersProtos.internal_static_tensorflow_JobDeviceFilters_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.distruntime.JobDeviceFilters.class, org.tensorflow.proto.distruntime.JobDeviceFilters.Builder.class);
-  }
-
-  public static final int NAME_FIELD_NUMBER = 1;
-  private volatile java.lang.Object name_;
-  /**
-   * <pre>
-   * The name of this job.
-   * </pre>
-   *
-   * <code>string name = 1;</code>
-   */
-  public java.lang.String getName() {
-    java.lang.Object ref = name_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      name_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * The name of this job.
-   * </pre>
-   *
-   * <code>string name = 1;</code>
-   */
-  public com.google.protobuf.ByteString
-      getNameBytes() {
-    java.lang.Object ref = name_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      name_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int TASKS_FIELD_NUMBER = 2;
-  private static final class TasksDefaultEntryHolder {
-    static final com.google.protobuf.MapEntry<
-        java.lang.Integer, org.tensorflow.proto.distruntime.TaskDeviceFilters> defaultEntry =
-            com.google.protobuf.MapEntry
-            .<java.lang.Integer, org.tensorflow.proto.distruntime.TaskDeviceFilters>newDefaultInstance(
-                org.tensorflow.proto.distruntime.DeviceFiltersProtos.internal_static_tensorflow_JobDeviceFilters_TasksEntry_descriptor, 
-                com.google.protobuf.WireFormat.FieldType.INT32,
-                0,
-                com.google.protobuf.WireFormat.FieldType.MESSAGE,
-                org.tensorflow.proto.distruntime.TaskDeviceFilters.getDefaultInstance());
-  }
-  private com.google.protobuf.MapField<
-      java.lang.Integer, org.tensorflow.proto.distruntime.TaskDeviceFilters> tasks_;
-  private com.google.protobuf.MapField<java.lang.Integer, org.tensorflow.proto.distruntime.TaskDeviceFilters>
-  internalGetTasks() {
-    if (tasks_ == null) {
-      return com.google.protobuf.MapField.emptyMapField(
-          TasksDefaultEntryHolder.defaultEntry);
-    }
-    return tasks_;
-  }
-
-  public int getTasksCount() {
-    return internalGetTasks().getMap().size();
-  }
-  /**
-   * <pre>
-   * Mapping from task ID to task device filters.
-   * </pre>
-   *
-   * <code>map&lt;int32, .tensorflow.TaskDeviceFilters&gt; tasks = 2;</code>
-   */
-
-  public boolean containsTasks(
-      int key) {
-    
-    return internalGetTasks().getMap().containsKey(key);
-  }
-  /**
-   * Use {@link #getTasksMap()} instead.
-   */
-  @java.lang.Deprecated
-  public java.util.Map<java.lang.Integer, org.tensorflow.proto.distruntime.TaskDeviceFilters> getTasks() {
-    return getTasksMap();
-  }
-  /**
-   * <pre>
-   * Mapping from task ID to task device filters.
-   * </pre>
-   *
-   * <code>map&lt;int32, .tensorflow.TaskDeviceFilters&gt; tasks = 2;</code>
-   */
-
-  public java.util.Map<java.lang.Integer, org.tensorflow.proto.distruntime.TaskDeviceFilters> getTasksMap() {
-    return internalGetTasks().getMap();
-  }
-  /**
-   * <pre>
-   * Mapping from task ID to task device filters.
-   * </pre>
-   *
-   * <code>map&lt;int32, .tensorflow.TaskDeviceFilters&gt; tasks = 2;</code>
-   */
-
-  public org.tensorflow.proto.distruntime.TaskDeviceFilters getTasksOrDefault(
-      int key,
-      org.tensorflow.proto.distruntime.TaskDeviceFilters defaultValue) {
-    
-    java.util.Map<java.lang.Integer, org.tensorflow.proto.distruntime.TaskDeviceFilters> map =
-        internalGetTasks().getMap();
-    return map.containsKey(key) ? map.get(key) : defaultValue;
-  }
-  /**
-   * <pre>
-   * Mapping from task ID to task device filters.
-   * </pre>
-   *
-   * <code>map&lt;int32, .tensorflow.TaskDeviceFilters&gt; tasks = 2;</code>
-   */
-
-  public org.tensorflow.proto.distruntime.TaskDeviceFilters getTasksOrThrow(
-      int key) {
-    
-    java.util.Map<java.lang.Integer, org.tensorflow.proto.distruntime.TaskDeviceFilters> map =
-        internalGetTasks().getMap();
-    if (!map.containsKey(key)) {
-      throw new java.lang.IllegalArgumentException();
-    }
-    return map.get(key);
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    if (!getNameBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 1, name_);
-    }
-    com.google.protobuf.GeneratedMessageV3
-      .serializeIntegerMapTo(
-        output,
-        internalGetTasks(),
-        TasksDefaultEntryHolder.defaultEntry,
-        2);
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (!getNameBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, name_);
-    }
-    for (java.util.Map.Entry<java.lang.Integer, org.tensorflow.proto.distruntime.TaskDeviceFilters> entry
-         : internalGetTasks().getMap().entrySet()) {
-      com.google.protobuf.MapEntry<java.lang.Integer, org.tensorflow.proto.distruntime.TaskDeviceFilters>
-      tasks__ = TasksDefaultEntryHolder.defaultEntry.newBuilderForType()
-          .setKey(entry.getKey())
-          .setValue(entry.getValue())
-          .build();
-      size += com.google.protobuf.CodedOutputStream
-          .computeMessageSize(2, tasks__);
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.distruntime.JobDeviceFilters)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.distruntime.JobDeviceFilters other = (org.tensorflow.proto.distruntime.JobDeviceFilters) obj;
-
-    if (!getName()
-        .equals(other.getName())) return false;
-    if (!internalGetTasks().equals(
-        other.internalGetTasks())) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    hash = (37 * hash) + NAME_FIELD_NUMBER;
-    hash = (53 * hash) + getName().hashCode();
-    if (!internalGetTasks().getMap().isEmpty()) {
-      hash = (37 * hash) + TASKS_FIELD_NUMBER;
-      hash = (53 * hash) + internalGetTasks().hashCode();
-    }
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.distruntime.JobDeviceFilters parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.distruntime.JobDeviceFilters parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.distruntime.JobDeviceFilters parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.distruntime.JobDeviceFilters parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.distruntime.JobDeviceFilters parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.distruntime.JobDeviceFilters parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.distruntime.JobDeviceFilters parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.distruntime.JobDeviceFilters parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.distruntime.JobDeviceFilters parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.distruntime.JobDeviceFilters parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.distruntime.JobDeviceFilters parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.distruntime.JobDeviceFilters parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.distruntime.JobDeviceFilters prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * <pre>
-   * Defines the device filters for tasks in a job.
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.JobDeviceFilters}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.JobDeviceFilters)
-      org.tensorflow.proto.distruntime.JobDeviceFiltersOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.distruntime.DeviceFiltersProtos.internal_static_tensorflow_JobDeviceFilters_descriptor;
-    }
-
-    @SuppressWarnings({"rawtypes"})
-    protected com.google.protobuf.MapField internalGetMapField(
-        int number) {
-      switch (number) {
-        case 2:
-          return internalGetTasks();
-        default:
-          throw new RuntimeException(
-              "Invalid map field number: " + number);
-      }
-    }
-    @SuppressWarnings({"rawtypes"})
-    protected com.google.protobuf.MapField internalGetMutableMapField(
-        int number) {
-      switch (number) {
-        case 2:
-          return internalGetMutableTasks();
-        default:
-          throw new RuntimeException(
-              "Invalid map field number: " + number);
-      }
-    }
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.distruntime.DeviceFiltersProtos.internal_static_tensorflow_JobDeviceFilters_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.distruntime.JobDeviceFilters.class, org.tensorflow.proto.distruntime.JobDeviceFilters.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.distruntime.JobDeviceFilters.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      name_ = "";
-
-      internalGetMutableTasks().clear();
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.distruntime.DeviceFiltersProtos.internal_static_tensorflow_JobDeviceFilters_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.distruntime.JobDeviceFilters getDefaultInstanceForType() {
-      return org.tensorflow.proto.distruntime.JobDeviceFilters.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.distruntime.JobDeviceFilters build() {
-      org.tensorflow.proto.distruntime.JobDeviceFilters result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.distruntime.JobDeviceFilters buildPartial() {
-      org.tensorflow.proto.distruntime.JobDeviceFilters result = new org.tensorflow.proto.distruntime.JobDeviceFilters(this);
-      int from_bitField0_ = bitField0_;
-      result.name_ = name_;
-      result.tasks_ = internalGetTasks();
-      result.tasks_.makeImmutable();
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.distruntime.JobDeviceFilters) {
-        return mergeFrom((org.tensorflow.proto.distruntime.JobDeviceFilters)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.distruntime.JobDeviceFilters other) {
-      if (other == org.tensorflow.proto.distruntime.JobDeviceFilters.getDefaultInstance()) return this;
-      if (!other.getName().isEmpty()) {
-        name_ = other.name_;
-        onChanged();
-      }
-      internalGetMutableTasks().mergeFrom(
-          other.internalGetTasks());
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.distruntime.JobDeviceFilters parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.distruntime.JobDeviceFilters) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-    private int bitField0_;
-
-    private java.lang.Object name_ = "";
-    /**
-     * <pre>
-     * The name of this job.
-     * </pre>
-     *
-     * <code>string name = 1;</code>
-     */
-    public java.lang.String getName() {
-      java.lang.Object ref = name_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        name_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * The name of this job.
-     * </pre>
-     *
-     * <code>string name = 1;</code>
-     */
-    public com.google.protobuf.ByteString
-        getNameBytes() {
-      java.lang.Object ref = name_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        name_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * The name of this job.
-     * </pre>
-     *
-     * <code>string name = 1;</code>
-     */
-    public Builder setName(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      name_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * The name of this job.
-     * </pre>
-     *
-     * <code>string name = 1;</code>
-     */
-    public Builder clearName() {
-      
-      name_ = getDefaultInstance().getName();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * The name of this job.
-     * </pre>
-     *
-     * <code>string name = 1;</code>
-     */
-    public Builder setNameBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      name_ = value;
-      onChanged();
-      return this;
-    }
-
-    private com.google.protobuf.MapField<
-        java.lang.Integer, org.tensorflow.proto.distruntime.TaskDeviceFilters> tasks_;
-    private com.google.protobuf.MapField<java.lang.Integer, org.tensorflow.proto.distruntime.TaskDeviceFilters>
-    internalGetTasks() {
-      if (tasks_ == null) {
-        return com.google.protobuf.MapField.emptyMapField(
-            TasksDefaultEntryHolder.defaultEntry);
-      }
-      return tasks_;
-    }
-    private com.google.protobuf.MapField<java.lang.Integer, org.tensorflow.proto.distruntime.TaskDeviceFilters>
-    internalGetMutableTasks() {
-      onChanged();;
-      if (tasks_ == null) {
-        tasks_ = com.google.protobuf.MapField.newMapField(
-            TasksDefaultEntryHolder.defaultEntry);
-      }
-      if (!tasks_.isMutable()) {
-        tasks_ = tasks_.copy();
-      }
-      return tasks_;
-    }
-
-    public int getTasksCount() {
-      return internalGetTasks().getMap().size();
-    }
-    /**
-     * <pre>
-     * Mapping from task ID to task device filters.
-     * </pre>
-     *
-     * <code>map&lt;int32, .tensorflow.TaskDeviceFilters&gt; tasks = 2;</code>
-     */
-
-    public boolean containsTasks(
-        int key) {
-      
-      return internalGetTasks().getMap().containsKey(key);
-    }
-    /**
-     * Use {@link #getTasksMap()} instead.
-     */
-    @java.lang.Deprecated
-    public java.util.Map<java.lang.Integer, org.tensorflow.proto.distruntime.TaskDeviceFilters> getTasks() {
-      return getTasksMap();
-    }
-    /**
-     * <pre>
-     * Mapping from task ID to task device filters.
-     * </pre>
-     *
-     * <code>map&lt;int32, .tensorflow.TaskDeviceFilters&gt; tasks = 2;</code>
-     */
-
-    public java.util.Map<java.lang.Integer, org.tensorflow.proto.distruntime.TaskDeviceFilters> getTasksMap() {
-      return internalGetTasks().getMap();
-    }
-    /**
-     * <pre>
-     * Mapping from task ID to task device filters.
-     * </pre>
-     *
-     * <code>map&lt;int32, .tensorflow.TaskDeviceFilters&gt; tasks = 2;</code>
-     */
-
-    public org.tensorflow.proto.distruntime.TaskDeviceFilters getTasksOrDefault(
-        int key,
-        org.tensorflow.proto.distruntime.TaskDeviceFilters defaultValue) {
-      
-      java.util.Map<java.lang.Integer, org.tensorflow.proto.distruntime.TaskDeviceFilters> map =
-          internalGetTasks().getMap();
-      return map.containsKey(key) ? map.get(key) : defaultValue;
-    }
-    /**
-     * <pre>
-     * Mapping from task ID to task device filters.
-     * </pre>
-     *
-     * <code>map&lt;int32, .tensorflow.TaskDeviceFilters&gt; tasks = 2;</code>
-     */
-
-    public org.tensorflow.proto.distruntime.TaskDeviceFilters getTasksOrThrow(
-        int key) {
-      
-      java.util.Map<java.lang.Integer, org.tensorflow.proto.distruntime.TaskDeviceFilters> map =
-          internalGetTasks().getMap();
-      if (!map.containsKey(key)) {
-        throw new java.lang.IllegalArgumentException();
-      }
-      return map.get(key);
-    }
-
-    public Builder clearTasks() {
-      internalGetMutableTasks().getMutableMap()
-          .clear();
-      return this;
-    }
-    /**
-     * <pre>
-     * Mapping from task ID to task device filters.
-     * </pre>
-     *
-     * <code>map&lt;int32, .tensorflow.TaskDeviceFilters&gt; tasks = 2;</code>
-     */
-
-    public Builder removeTasks(
-        int key) {
-      
-      internalGetMutableTasks().getMutableMap()
-          .remove(key);
-      return this;
-    }
-    /**
-     * Use alternate mutation accessors instead.
-     */
-    @java.lang.Deprecated
-    public java.util.Map<java.lang.Integer, org.tensorflow.proto.distruntime.TaskDeviceFilters>
-    getMutableTasks() {
-      return internalGetMutableTasks().getMutableMap();
-    }
-    /**
-     * <pre>
-     * Mapping from task ID to task device filters.
-     * </pre>
-     *
-     * <code>map&lt;int32, .tensorflow.TaskDeviceFilters&gt; tasks = 2;</code>
-     */
-    public Builder putTasks(
-        int key,
-        org.tensorflow.proto.distruntime.TaskDeviceFilters value) {
-      
-      if (value == null) { throw new java.lang.NullPointerException(); }
-      internalGetMutableTasks().getMutableMap()
-          .put(key, value);
-      return this;
-    }
-    /**
-     * <pre>
-     * Mapping from task ID to task device filters.
-     * </pre>
-     *
-     * <code>map&lt;int32, .tensorflow.TaskDeviceFilters&gt; tasks = 2;</code>
-     */
-
-    public Builder putAllTasks(
-        java.util.Map<java.lang.Integer, org.tensorflow.proto.distruntime.TaskDeviceFilters> values) {
-      internalGetMutableTasks().getMutableMap()
-          .putAll(values);
-      return this;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.JobDeviceFilters)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.JobDeviceFilters)
-  private static final org.tensorflow.proto.distruntime.JobDeviceFilters DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.distruntime.JobDeviceFilters();
-  }
-
-  public static org.tensorflow.proto.distruntime.JobDeviceFilters getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<JobDeviceFilters>
-      PARSER = new com.google.protobuf.AbstractParser<JobDeviceFilters>() {
-    @java.lang.Override
-    public JobDeviceFilters parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new JobDeviceFilters(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<JobDeviceFilters> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<JobDeviceFilters> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.distruntime.JobDeviceFilters getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/distruntime/JobDeviceFiltersOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/distruntime/JobDeviceFiltersOrBuilder.java
deleted file mode 100644
index 92c3ba82935..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/distruntime/JobDeviceFiltersOrBuilder.java
+++ /dev/null
@@ -1,81 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/device_filters.proto
-
-package org.tensorflow.proto.distruntime;
-
-public interface JobDeviceFiltersOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.JobDeviceFilters)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <pre>
-   * The name of this job.
-   * </pre>
-   *
-   * <code>string name = 1;</code>
-   */
-  java.lang.String getName();
-  /**
-   * <pre>
-   * The name of this job.
-   * </pre>
-   *
-   * <code>string name = 1;</code>
-   */
-  com.google.protobuf.ByteString
-      getNameBytes();
-
-  /**
-   * <pre>
-   * Mapping from task ID to task device filters.
-   * </pre>
-   *
-   * <code>map&lt;int32, .tensorflow.TaskDeviceFilters&gt; tasks = 2;</code>
-   */
-  int getTasksCount();
-  /**
-   * <pre>
-   * Mapping from task ID to task device filters.
-   * </pre>
-   *
-   * <code>map&lt;int32, .tensorflow.TaskDeviceFilters&gt; tasks = 2;</code>
-   */
-  boolean containsTasks(
-      int key);
-  /**
-   * Use {@link #getTasksMap()} instead.
-   */
-  @java.lang.Deprecated
-  java.util.Map<java.lang.Integer, org.tensorflow.proto.distruntime.TaskDeviceFilters>
-  getTasks();
-  /**
-   * <pre>
-   * Mapping from task ID to task device filters.
-   * </pre>
-   *
-   * <code>map&lt;int32, .tensorflow.TaskDeviceFilters&gt; tasks = 2;</code>
-   */
-  java.util.Map<java.lang.Integer, org.tensorflow.proto.distruntime.TaskDeviceFilters>
-  getTasksMap();
-  /**
-   * <pre>
-   * Mapping from task ID to task device filters.
-   * </pre>
-   *
-   * <code>map&lt;int32, .tensorflow.TaskDeviceFilters&gt; tasks = 2;</code>
-   */
-
-  org.tensorflow.proto.distruntime.TaskDeviceFilters getTasksOrDefault(
-      int key,
-      org.tensorflow.proto.distruntime.TaskDeviceFilters defaultValue);
-  /**
-   * <pre>
-   * Mapping from task ID to task device filters.
-   * </pre>
-   *
-   * <code>map&lt;int32, .tensorflow.TaskDeviceFilters&gt; tasks = 2;</code>
-   */
-
-  org.tensorflow.proto.distruntime.TaskDeviceFilters getTasksOrThrow(
-      int key);
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/distruntime/ServerDef.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/distruntime/ServerDef.java
deleted file mode 100644
index 068264c171e..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/distruntime/ServerDef.java
+++ /dev/null
@@ -1,1611 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/tensorflow_server.proto
-
-package org.tensorflow.proto.distruntime;
-
-/**
- * <pre>
- * Defines the configuration of a single TensorFlow server.
- * </pre>
- *
- * Protobuf type {@code tensorflow.ServerDef}
- */
-public  final class ServerDef extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.ServerDef)
-    ServerDefOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use ServerDef.newBuilder() to construct.
-  private ServerDef(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private ServerDef() {
-    jobName_ = "";
-    protocol_ = "";
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new ServerDef();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private ServerDef(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 10: {
-            org.tensorflow.proto.distruntime.ClusterDef.Builder subBuilder = null;
-            if (cluster_ != null) {
-              subBuilder = cluster_.toBuilder();
-            }
-            cluster_ = input.readMessage(org.tensorflow.proto.distruntime.ClusterDef.parser(), extensionRegistry);
-            if (subBuilder != null) {
-              subBuilder.mergeFrom(cluster_);
-              cluster_ = subBuilder.buildPartial();
-            }
-
-            break;
-          }
-          case 18: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            jobName_ = s;
-            break;
-          }
-          case 24: {
-
-            taskIndex_ = input.readInt32();
-            break;
-          }
-          case 34: {
-            org.tensorflow.proto.framework.ConfigProto.Builder subBuilder = null;
-            if (defaultSessionConfig_ != null) {
-              subBuilder = defaultSessionConfig_.toBuilder();
-            }
-            defaultSessionConfig_ = input.readMessage(org.tensorflow.proto.framework.ConfigProto.parser(), extensionRegistry);
-            if (subBuilder != null) {
-              subBuilder.mergeFrom(defaultSessionConfig_);
-              defaultSessionConfig_ = subBuilder.buildPartial();
-            }
-
-            break;
-          }
-          case 42: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            protocol_ = s;
-            break;
-          }
-          case 48: {
-
-            port_ = input.readInt32();
-            break;
-          }
-          case 58: {
-            org.tensorflow.proto.distruntime.ClusterDeviceFilters.Builder subBuilder = null;
-            if (clusterDeviceFilters_ != null) {
-              subBuilder = clusterDeviceFilters_.toBuilder();
-            }
-            clusterDeviceFilters_ = input.readMessage(org.tensorflow.proto.distruntime.ClusterDeviceFilters.parser(), extensionRegistry);
-            if (subBuilder != null) {
-              subBuilder.mergeFrom(clusterDeviceFilters_);
-              clusterDeviceFilters_ = subBuilder.buildPartial();
-            }
-
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.distruntime.ServerProtos.internal_static_tensorflow_ServerDef_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.distruntime.ServerProtos.internal_static_tensorflow_ServerDef_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.distruntime.ServerDef.class, org.tensorflow.proto.distruntime.ServerDef.Builder.class);
-  }
-
-  public static final int CLUSTER_FIELD_NUMBER = 1;
-  private org.tensorflow.proto.distruntime.ClusterDef cluster_;
-  /**
-   * <pre>
-   * The cluster of which this server is a member.
-   * </pre>
-   *
-   * <code>.tensorflow.ClusterDef cluster = 1;</code>
-   */
-  public boolean hasCluster() {
-    return cluster_ != null;
-  }
-  /**
-   * <pre>
-   * The cluster of which this server is a member.
-   * </pre>
-   *
-   * <code>.tensorflow.ClusterDef cluster = 1;</code>
-   */
-  public org.tensorflow.proto.distruntime.ClusterDef getCluster() {
-    return cluster_ == null ? org.tensorflow.proto.distruntime.ClusterDef.getDefaultInstance() : cluster_;
-  }
-  /**
-   * <pre>
-   * The cluster of which this server is a member.
-   * </pre>
-   *
-   * <code>.tensorflow.ClusterDef cluster = 1;</code>
-   */
-  public org.tensorflow.proto.distruntime.ClusterDefOrBuilder getClusterOrBuilder() {
-    return getCluster();
-  }
-
-  public static final int JOB_NAME_FIELD_NUMBER = 2;
-  private volatile java.lang.Object jobName_;
-  /**
-   * <pre>
-   * The name of the job of which this server is a member.
-   * NOTE(mrry): The `cluster` field must contain a `JobDef` with a `name` field
-   * that matches this name.
-   * </pre>
-   *
-   * <code>string job_name = 2;</code>
-   */
-  public java.lang.String getJobName() {
-    java.lang.Object ref = jobName_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      jobName_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * The name of the job of which this server is a member.
-   * NOTE(mrry): The `cluster` field must contain a `JobDef` with a `name` field
-   * that matches this name.
-   * </pre>
-   *
-   * <code>string job_name = 2;</code>
-   */
-  public com.google.protobuf.ByteString
-      getJobNameBytes() {
-    java.lang.Object ref = jobName_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      jobName_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int TASK_INDEX_FIELD_NUMBER = 3;
-  private int taskIndex_;
-  /**
-   * <pre>
-   * The task index of this server in its job.
-   * NOTE: The `cluster` field must contain a `JobDef` with a matching `name`
-   * and a mapping in its `tasks` field for this index.
-   * </pre>
-   *
-   * <code>int32 task_index = 3;</code>
-   */
-  public int getTaskIndex() {
-    return taskIndex_;
-  }
-
-  public static final int DEFAULT_SESSION_CONFIG_FIELD_NUMBER = 4;
-  private org.tensorflow.proto.framework.ConfigProto defaultSessionConfig_;
-  /**
-   * <pre>
-   * The default configuration for sessions that run on this server.
-   * </pre>
-   *
-   * <code>.tensorflow.ConfigProto default_session_config = 4;</code>
-   */
-  public boolean hasDefaultSessionConfig() {
-    return defaultSessionConfig_ != null;
-  }
-  /**
-   * <pre>
-   * The default configuration for sessions that run on this server.
-   * </pre>
-   *
-   * <code>.tensorflow.ConfigProto default_session_config = 4;</code>
-   */
-  public org.tensorflow.proto.framework.ConfigProto getDefaultSessionConfig() {
-    return defaultSessionConfig_ == null ? org.tensorflow.proto.framework.ConfigProto.getDefaultInstance() : defaultSessionConfig_;
-  }
-  /**
-   * <pre>
-   * The default configuration for sessions that run on this server.
-   * </pre>
-   *
-   * <code>.tensorflow.ConfigProto default_session_config = 4;</code>
-   */
-  public org.tensorflow.proto.framework.ConfigProtoOrBuilder getDefaultSessionConfigOrBuilder() {
-    return getDefaultSessionConfig();
-  }
-
-  public static final int PROTOCOL_FIELD_NUMBER = 5;
-  private volatile java.lang.Object protocol_;
-  /**
-   * <pre>
-   * The protocol to be used by this server.
-   * Acceptable values include: "grpc", "grpc+verbs".
-   * </pre>
-   *
-   * <code>string protocol = 5;</code>
-   */
-  public java.lang.String getProtocol() {
-    java.lang.Object ref = protocol_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      protocol_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * The protocol to be used by this server.
-   * Acceptable values include: "grpc", "grpc+verbs".
-   * </pre>
-   *
-   * <code>string protocol = 5;</code>
-   */
-  public com.google.protobuf.ByteString
-      getProtocolBytes() {
-    java.lang.Object ref = protocol_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      protocol_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int PORT_FIELD_NUMBER = 6;
-  private int port_;
-  /**
-   * <pre>
-   * The server port. If not set, then we identify the port from the job_name.
-   * </pre>
-   *
-   * <code>int32 port = 6;</code>
-   */
-  public int getPort() {
-    return port_;
-  }
-
-  public static final int CLUSTER_DEVICE_FILTERS_FIELD_NUMBER = 7;
-  private org.tensorflow.proto.distruntime.ClusterDeviceFilters clusterDeviceFilters_;
-  /**
-   * <pre>
-   * Device filters for remote tasks in the cluster.
-   * NOTE: This is an experimental feature and only effective in TensorFlow 2.x.
-   * </pre>
-   *
-   * <code>.tensorflow.ClusterDeviceFilters cluster_device_filters = 7;</code>
-   */
-  public boolean hasClusterDeviceFilters() {
-    return clusterDeviceFilters_ != null;
-  }
-  /**
-   * <pre>
-   * Device filters for remote tasks in the cluster.
-   * NOTE: This is an experimental feature and only effective in TensorFlow 2.x.
-   * </pre>
-   *
-   * <code>.tensorflow.ClusterDeviceFilters cluster_device_filters = 7;</code>
-   */
-  public org.tensorflow.proto.distruntime.ClusterDeviceFilters getClusterDeviceFilters() {
-    return clusterDeviceFilters_ == null ? org.tensorflow.proto.distruntime.ClusterDeviceFilters.getDefaultInstance() : clusterDeviceFilters_;
-  }
-  /**
-   * <pre>
-   * Device filters for remote tasks in the cluster.
-   * NOTE: This is an experimental feature and only effective in TensorFlow 2.x.
-   * </pre>
-   *
-   * <code>.tensorflow.ClusterDeviceFilters cluster_device_filters = 7;</code>
-   */
-  public org.tensorflow.proto.distruntime.ClusterDeviceFiltersOrBuilder getClusterDeviceFiltersOrBuilder() {
-    return getClusterDeviceFilters();
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    if (cluster_ != null) {
-      output.writeMessage(1, getCluster());
-    }
-    if (!getJobNameBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 2, jobName_);
-    }
-    if (taskIndex_ != 0) {
-      output.writeInt32(3, taskIndex_);
-    }
-    if (defaultSessionConfig_ != null) {
-      output.writeMessage(4, getDefaultSessionConfig());
-    }
-    if (!getProtocolBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 5, protocol_);
-    }
-    if (port_ != 0) {
-      output.writeInt32(6, port_);
-    }
-    if (clusterDeviceFilters_ != null) {
-      output.writeMessage(7, getClusterDeviceFilters());
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (cluster_ != null) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(1, getCluster());
-    }
-    if (!getJobNameBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(2, jobName_);
-    }
-    if (taskIndex_ != 0) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt32Size(3, taskIndex_);
-    }
-    if (defaultSessionConfig_ != null) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(4, getDefaultSessionConfig());
-    }
-    if (!getProtocolBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(5, protocol_);
-    }
-    if (port_ != 0) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt32Size(6, port_);
-    }
-    if (clusterDeviceFilters_ != null) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(7, getClusterDeviceFilters());
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.distruntime.ServerDef)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.distruntime.ServerDef other = (org.tensorflow.proto.distruntime.ServerDef) obj;
-
-    if (hasCluster() != other.hasCluster()) return false;
-    if (hasCluster()) {
-      if (!getCluster()
-          .equals(other.getCluster())) return false;
-    }
-    if (!getJobName()
-        .equals(other.getJobName())) return false;
-    if (getTaskIndex()
-        != other.getTaskIndex()) return false;
-    if (hasDefaultSessionConfig() != other.hasDefaultSessionConfig()) return false;
-    if (hasDefaultSessionConfig()) {
-      if (!getDefaultSessionConfig()
-          .equals(other.getDefaultSessionConfig())) return false;
-    }
-    if (!getProtocol()
-        .equals(other.getProtocol())) return false;
-    if (getPort()
-        != other.getPort()) return false;
-    if (hasClusterDeviceFilters() != other.hasClusterDeviceFilters()) return false;
-    if (hasClusterDeviceFilters()) {
-      if (!getClusterDeviceFilters()
-          .equals(other.getClusterDeviceFilters())) return false;
-    }
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    if (hasCluster()) {
-      hash = (37 * hash) + CLUSTER_FIELD_NUMBER;
-      hash = (53 * hash) + getCluster().hashCode();
-    }
-    hash = (37 * hash) + JOB_NAME_FIELD_NUMBER;
-    hash = (53 * hash) + getJobName().hashCode();
-    hash = (37 * hash) + TASK_INDEX_FIELD_NUMBER;
-    hash = (53 * hash) + getTaskIndex();
-    if (hasDefaultSessionConfig()) {
-      hash = (37 * hash) + DEFAULT_SESSION_CONFIG_FIELD_NUMBER;
-      hash = (53 * hash) + getDefaultSessionConfig().hashCode();
-    }
-    hash = (37 * hash) + PROTOCOL_FIELD_NUMBER;
-    hash = (53 * hash) + getProtocol().hashCode();
-    hash = (37 * hash) + PORT_FIELD_NUMBER;
-    hash = (53 * hash) + getPort();
-    if (hasClusterDeviceFilters()) {
-      hash = (37 * hash) + CLUSTER_DEVICE_FILTERS_FIELD_NUMBER;
-      hash = (53 * hash) + getClusterDeviceFilters().hashCode();
-    }
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.distruntime.ServerDef parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.distruntime.ServerDef parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.distruntime.ServerDef parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.distruntime.ServerDef parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.distruntime.ServerDef parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.distruntime.ServerDef parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.distruntime.ServerDef parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.distruntime.ServerDef parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.distruntime.ServerDef parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.distruntime.ServerDef parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.distruntime.ServerDef parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.distruntime.ServerDef parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.distruntime.ServerDef prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * <pre>
-   * Defines the configuration of a single TensorFlow server.
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.ServerDef}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.ServerDef)
-      org.tensorflow.proto.distruntime.ServerDefOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.distruntime.ServerProtos.internal_static_tensorflow_ServerDef_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.distruntime.ServerProtos.internal_static_tensorflow_ServerDef_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.distruntime.ServerDef.class, org.tensorflow.proto.distruntime.ServerDef.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.distruntime.ServerDef.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      if (clusterBuilder_ == null) {
-        cluster_ = null;
-      } else {
-        cluster_ = null;
-        clusterBuilder_ = null;
-      }
-      jobName_ = "";
-
-      taskIndex_ = 0;
-
-      if (defaultSessionConfigBuilder_ == null) {
-        defaultSessionConfig_ = null;
-      } else {
-        defaultSessionConfig_ = null;
-        defaultSessionConfigBuilder_ = null;
-      }
-      protocol_ = "";
-
-      port_ = 0;
-
-      if (clusterDeviceFiltersBuilder_ == null) {
-        clusterDeviceFilters_ = null;
-      } else {
-        clusterDeviceFilters_ = null;
-        clusterDeviceFiltersBuilder_ = null;
-      }
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.distruntime.ServerProtos.internal_static_tensorflow_ServerDef_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.distruntime.ServerDef getDefaultInstanceForType() {
-      return org.tensorflow.proto.distruntime.ServerDef.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.distruntime.ServerDef build() {
-      org.tensorflow.proto.distruntime.ServerDef result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.distruntime.ServerDef buildPartial() {
-      org.tensorflow.proto.distruntime.ServerDef result = new org.tensorflow.proto.distruntime.ServerDef(this);
-      if (clusterBuilder_ == null) {
-        result.cluster_ = cluster_;
-      } else {
-        result.cluster_ = clusterBuilder_.build();
-      }
-      result.jobName_ = jobName_;
-      result.taskIndex_ = taskIndex_;
-      if (defaultSessionConfigBuilder_ == null) {
-        result.defaultSessionConfig_ = defaultSessionConfig_;
-      } else {
-        result.defaultSessionConfig_ = defaultSessionConfigBuilder_.build();
-      }
-      result.protocol_ = protocol_;
-      result.port_ = port_;
-      if (clusterDeviceFiltersBuilder_ == null) {
-        result.clusterDeviceFilters_ = clusterDeviceFilters_;
-      } else {
-        result.clusterDeviceFilters_ = clusterDeviceFiltersBuilder_.build();
-      }
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.distruntime.ServerDef) {
-        return mergeFrom((org.tensorflow.proto.distruntime.ServerDef)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.distruntime.ServerDef other) {
-      if (other == org.tensorflow.proto.distruntime.ServerDef.getDefaultInstance()) return this;
-      if (other.hasCluster()) {
-        mergeCluster(other.getCluster());
-      }
-      if (!other.getJobName().isEmpty()) {
-        jobName_ = other.jobName_;
-        onChanged();
-      }
-      if (other.getTaskIndex() != 0) {
-        setTaskIndex(other.getTaskIndex());
-      }
-      if (other.hasDefaultSessionConfig()) {
-        mergeDefaultSessionConfig(other.getDefaultSessionConfig());
-      }
-      if (!other.getProtocol().isEmpty()) {
-        protocol_ = other.protocol_;
-        onChanged();
-      }
-      if (other.getPort() != 0) {
-        setPort(other.getPort());
-      }
-      if (other.hasClusterDeviceFilters()) {
-        mergeClusterDeviceFilters(other.getClusterDeviceFilters());
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.distruntime.ServerDef parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.distruntime.ServerDef) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-
-    private org.tensorflow.proto.distruntime.ClusterDef cluster_;
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.distruntime.ClusterDef, org.tensorflow.proto.distruntime.ClusterDef.Builder, org.tensorflow.proto.distruntime.ClusterDefOrBuilder> clusterBuilder_;
-    /**
-     * <pre>
-     * The cluster of which this server is a member.
-     * </pre>
-     *
-     * <code>.tensorflow.ClusterDef cluster = 1;</code>
-     */
-    public boolean hasCluster() {
-      return clusterBuilder_ != null || cluster_ != null;
-    }
-    /**
-     * <pre>
-     * The cluster of which this server is a member.
-     * </pre>
-     *
-     * <code>.tensorflow.ClusterDef cluster = 1;</code>
-     */
-    public org.tensorflow.proto.distruntime.ClusterDef getCluster() {
-      if (clusterBuilder_ == null) {
-        return cluster_ == null ? org.tensorflow.proto.distruntime.ClusterDef.getDefaultInstance() : cluster_;
-      } else {
-        return clusterBuilder_.getMessage();
-      }
-    }
-    /**
-     * <pre>
-     * The cluster of which this server is a member.
-     * </pre>
-     *
-     * <code>.tensorflow.ClusterDef cluster = 1;</code>
-     */
-    public Builder setCluster(org.tensorflow.proto.distruntime.ClusterDef value) {
-      if (clusterBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        cluster_ = value;
-        onChanged();
-      } else {
-        clusterBuilder_.setMessage(value);
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * The cluster of which this server is a member.
-     * </pre>
-     *
-     * <code>.tensorflow.ClusterDef cluster = 1;</code>
-     */
-    public Builder setCluster(
-        org.tensorflow.proto.distruntime.ClusterDef.Builder builderForValue) {
-      if (clusterBuilder_ == null) {
-        cluster_ = builderForValue.build();
-        onChanged();
-      } else {
-        clusterBuilder_.setMessage(builderForValue.build());
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * The cluster of which this server is a member.
-     * </pre>
-     *
-     * <code>.tensorflow.ClusterDef cluster = 1;</code>
-     */
-    public Builder mergeCluster(org.tensorflow.proto.distruntime.ClusterDef value) {
-      if (clusterBuilder_ == null) {
-        if (cluster_ != null) {
-          cluster_ =
-            org.tensorflow.proto.distruntime.ClusterDef.newBuilder(cluster_).mergeFrom(value).buildPartial();
-        } else {
-          cluster_ = value;
-        }
-        onChanged();
-      } else {
-        clusterBuilder_.mergeFrom(value);
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * The cluster of which this server is a member.
-     * </pre>
-     *
-     * <code>.tensorflow.ClusterDef cluster = 1;</code>
-     */
-    public Builder clearCluster() {
-      if (clusterBuilder_ == null) {
-        cluster_ = null;
-        onChanged();
-      } else {
-        cluster_ = null;
-        clusterBuilder_ = null;
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * The cluster of which this server is a member.
-     * </pre>
-     *
-     * <code>.tensorflow.ClusterDef cluster = 1;</code>
-     */
-    public org.tensorflow.proto.distruntime.ClusterDef.Builder getClusterBuilder() {
-      
-      onChanged();
-      return getClusterFieldBuilder().getBuilder();
-    }
-    /**
-     * <pre>
-     * The cluster of which this server is a member.
-     * </pre>
-     *
-     * <code>.tensorflow.ClusterDef cluster = 1;</code>
-     */
-    public org.tensorflow.proto.distruntime.ClusterDefOrBuilder getClusterOrBuilder() {
-      if (clusterBuilder_ != null) {
-        return clusterBuilder_.getMessageOrBuilder();
-      } else {
-        return cluster_ == null ?
-            org.tensorflow.proto.distruntime.ClusterDef.getDefaultInstance() : cluster_;
-      }
-    }
-    /**
-     * <pre>
-     * The cluster of which this server is a member.
-     * </pre>
-     *
-     * <code>.tensorflow.ClusterDef cluster = 1;</code>
-     */
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.distruntime.ClusterDef, org.tensorflow.proto.distruntime.ClusterDef.Builder, org.tensorflow.proto.distruntime.ClusterDefOrBuilder> 
-        getClusterFieldBuilder() {
-      if (clusterBuilder_ == null) {
-        clusterBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-            org.tensorflow.proto.distruntime.ClusterDef, org.tensorflow.proto.distruntime.ClusterDef.Builder, org.tensorflow.proto.distruntime.ClusterDefOrBuilder>(
-                getCluster(),
-                getParentForChildren(),
-                isClean());
-        cluster_ = null;
-      }
-      return clusterBuilder_;
-    }
-
-    private java.lang.Object jobName_ = "";
-    /**
-     * <pre>
-     * The name of the job of which this server is a member.
-     * NOTE(mrry): The `cluster` field must contain a `JobDef` with a `name` field
-     * that matches this name.
-     * </pre>
-     *
-     * <code>string job_name = 2;</code>
-     */
-    public java.lang.String getJobName() {
-      java.lang.Object ref = jobName_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        jobName_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * The name of the job of which this server is a member.
-     * NOTE(mrry): The `cluster` field must contain a `JobDef` with a `name` field
-     * that matches this name.
-     * </pre>
-     *
-     * <code>string job_name = 2;</code>
-     */
-    public com.google.protobuf.ByteString
-        getJobNameBytes() {
-      java.lang.Object ref = jobName_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        jobName_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * The name of the job of which this server is a member.
-     * NOTE(mrry): The `cluster` field must contain a `JobDef` with a `name` field
-     * that matches this name.
-     * </pre>
-     *
-     * <code>string job_name = 2;</code>
-     */
-    public Builder setJobName(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      jobName_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * The name of the job of which this server is a member.
-     * NOTE(mrry): The `cluster` field must contain a `JobDef` with a `name` field
-     * that matches this name.
-     * </pre>
-     *
-     * <code>string job_name = 2;</code>
-     */
-    public Builder clearJobName() {
-      
-      jobName_ = getDefaultInstance().getJobName();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * The name of the job of which this server is a member.
-     * NOTE(mrry): The `cluster` field must contain a `JobDef` with a `name` field
-     * that matches this name.
-     * </pre>
-     *
-     * <code>string job_name = 2;</code>
-     */
-    public Builder setJobNameBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      jobName_ = value;
-      onChanged();
-      return this;
-    }
-
-    private int taskIndex_ ;
-    /**
-     * <pre>
-     * The task index of this server in its job.
-     * NOTE: The `cluster` field must contain a `JobDef` with a matching `name`
-     * and a mapping in its `tasks` field for this index.
-     * </pre>
-     *
-     * <code>int32 task_index = 3;</code>
-     */
-    public int getTaskIndex() {
-      return taskIndex_;
-    }
-    /**
-     * <pre>
-     * The task index of this server in its job.
-     * NOTE: The `cluster` field must contain a `JobDef` with a matching `name`
-     * and a mapping in its `tasks` field for this index.
-     * </pre>
-     *
-     * <code>int32 task_index = 3;</code>
-     */
-    public Builder setTaskIndex(int value) {
-      
-      taskIndex_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * The task index of this server in its job.
-     * NOTE: The `cluster` field must contain a `JobDef` with a matching `name`
-     * and a mapping in its `tasks` field for this index.
-     * </pre>
-     *
-     * <code>int32 task_index = 3;</code>
-     */
-    public Builder clearTaskIndex() {
-      
-      taskIndex_ = 0;
-      onChanged();
-      return this;
-    }
-
-    private org.tensorflow.proto.framework.ConfigProto defaultSessionConfig_;
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.ConfigProto, org.tensorflow.proto.framework.ConfigProto.Builder, org.tensorflow.proto.framework.ConfigProtoOrBuilder> defaultSessionConfigBuilder_;
-    /**
-     * <pre>
-     * The default configuration for sessions that run on this server.
-     * </pre>
-     *
-     * <code>.tensorflow.ConfigProto default_session_config = 4;</code>
-     */
-    public boolean hasDefaultSessionConfig() {
-      return defaultSessionConfigBuilder_ != null || defaultSessionConfig_ != null;
-    }
-    /**
-     * <pre>
-     * The default configuration for sessions that run on this server.
-     * </pre>
-     *
-     * <code>.tensorflow.ConfigProto default_session_config = 4;</code>
-     */
-    public org.tensorflow.proto.framework.ConfigProto getDefaultSessionConfig() {
-      if (defaultSessionConfigBuilder_ == null) {
-        return defaultSessionConfig_ == null ? org.tensorflow.proto.framework.ConfigProto.getDefaultInstance() : defaultSessionConfig_;
-      } else {
-        return defaultSessionConfigBuilder_.getMessage();
-      }
-    }
-    /**
-     * <pre>
-     * The default configuration for sessions that run on this server.
-     * </pre>
-     *
-     * <code>.tensorflow.ConfigProto default_session_config = 4;</code>
-     */
-    public Builder setDefaultSessionConfig(org.tensorflow.proto.framework.ConfigProto value) {
-      if (defaultSessionConfigBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        defaultSessionConfig_ = value;
-        onChanged();
-      } else {
-        defaultSessionConfigBuilder_.setMessage(value);
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * The default configuration for sessions that run on this server.
-     * </pre>
-     *
-     * <code>.tensorflow.ConfigProto default_session_config = 4;</code>
-     */
-    public Builder setDefaultSessionConfig(
-        org.tensorflow.proto.framework.ConfigProto.Builder builderForValue) {
-      if (defaultSessionConfigBuilder_ == null) {
-        defaultSessionConfig_ = builderForValue.build();
-        onChanged();
-      } else {
-        defaultSessionConfigBuilder_.setMessage(builderForValue.build());
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * The default configuration for sessions that run on this server.
-     * </pre>
-     *
-     * <code>.tensorflow.ConfigProto default_session_config = 4;</code>
-     */
-    public Builder mergeDefaultSessionConfig(org.tensorflow.proto.framework.ConfigProto value) {
-      if (defaultSessionConfigBuilder_ == null) {
-        if (defaultSessionConfig_ != null) {
-          defaultSessionConfig_ =
-            org.tensorflow.proto.framework.ConfigProto.newBuilder(defaultSessionConfig_).mergeFrom(value).buildPartial();
-        } else {
-          defaultSessionConfig_ = value;
-        }
-        onChanged();
-      } else {
-        defaultSessionConfigBuilder_.mergeFrom(value);
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * The default configuration for sessions that run on this server.
-     * </pre>
-     *
-     * <code>.tensorflow.ConfigProto default_session_config = 4;</code>
-     */
-    public Builder clearDefaultSessionConfig() {
-      if (defaultSessionConfigBuilder_ == null) {
-        defaultSessionConfig_ = null;
-        onChanged();
-      } else {
-        defaultSessionConfig_ = null;
-        defaultSessionConfigBuilder_ = null;
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * The default configuration for sessions that run on this server.
-     * </pre>
-     *
-     * <code>.tensorflow.ConfigProto default_session_config = 4;</code>
-     */
-    public org.tensorflow.proto.framework.ConfigProto.Builder getDefaultSessionConfigBuilder() {
-      
-      onChanged();
-      return getDefaultSessionConfigFieldBuilder().getBuilder();
-    }
-    /**
-     * <pre>
-     * The default configuration for sessions that run on this server.
-     * </pre>
-     *
-     * <code>.tensorflow.ConfigProto default_session_config = 4;</code>
-     */
-    public org.tensorflow.proto.framework.ConfigProtoOrBuilder getDefaultSessionConfigOrBuilder() {
-      if (defaultSessionConfigBuilder_ != null) {
-        return defaultSessionConfigBuilder_.getMessageOrBuilder();
-      } else {
-        return defaultSessionConfig_ == null ?
-            org.tensorflow.proto.framework.ConfigProto.getDefaultInstance() : defaultSessionConfig_;
-      }
-    }
-    /**
-     * <pre>
-     * The default configuration for sessions that run on this server.
-     * </pre>
-     *
-     * <code>.tensorflow.ConfigProto default_session_config = 4;</code>
-     */
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.ConfigProto, org.tensorflow.proto.framework.ConfigProto.Builder, org.tensorflow.proto.framework.ConfigProtoOrBuilder> 
-        getDefaultSessionConfigFieldBuilder() {
-      if (defaultSessionConfigBuilder_ == null) {
-        defaultSessionConfigBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-            org.tensorflow.proto.framework.ConfigProto, org.tensorflow.proto.framework.ConfigProto.Builder, org.tensorflow.proto.framework.ConfigProtoOrBuilder>(
-                getDefaultSessionConfig(),
-                getParentForChildren(),
-                isClean());
-        defaultSessionConfig_ = null;
-      }
-      return defaultSessionConfigBuilder_;
-    }
-
-    private java.lang.Object protocol_ = "";
-    /**
-     * <pre>
-     * The protocol to be used by this server.
-     * Acceptable values include: "grpc", "grpc+verbs".
-     * </pre>
-     *
-     * <code>string protocol = 5;</code>
-     */
-    public java.lang.String getProtocol() {
-      java.lang.Object ref = protocol_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        protocol_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * The protocol to be used by this server.
-     * Acceptable values include: "grpc", "grpc+verbs".
-     * </pre>
-     *
-     * <code>string protocol = 5;</code>
-     */
-    public com.google.protobuf.ByteString
-        getProtocolBytes() {
-      java.lang.Object ref = protocol_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        protocol_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * The protocol to be used by this server.
-     * Acceptable values include: "grpc", "grpc+verbs".
-     * </pre>
-     *
-     * <code>string protocol = 5;</code>
-     */
-    public Builder setProtocol(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      protocol_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * The protocol to be used by this server.
-     * Acceptable values include: "grpc", "grpc+verbs".
-     * </pre>
-     *
-     * <code>string protocol = 5;</code>
-     */
-    public Builder clearProtocol() {
-      
-      protocol_ = getDefaultInstance().getProtocol();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * The protocol to be used by this server.
-     * Acceptable values include: "grpc", "grpc+verbs".
-     * </pre>
-     *
-     * <code>string protocol = 5;</code>
-     */
-    public Builder setProtocolBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      protocol_ = value;
-      onChanged();
-      return this;
-    }
-
-    private int port_ ;
-    /**
-     * <pre>
-     * The server port. If not set, then we identify the port from the job_name.
-     * </pre>
-     *
-     * <code>int32 port = 6;</code>
-     */
-    public int getPort() {
-      return port_;
-    }
-    /**
-     * <pre>
-     * The server port. If not set, then we identify the port from the job_name.
-     * </pre>
-     *
-     * <code>int32 port = 6;</code>
-     */
-    public Builder setPort(int value) {
-      
-      port_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * The server port. If not set, then we identify the port from the job_name.
-     * </pre>
-     *
-     * <code>int32 port = 6;</code>
-     */
-    public Builder clearPort() {
-      
-      port_ = 0;
-      onChanged();
-      return this;
-    }
-
-    private org.tensorflow.proto.distruntime.ClusterDeviceFilters clusterDeviceFilters_;
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.distruntime.ClusterDeviceFilters, org.tensorflow.proto.distruntime.ClusterDeviceFilters.Builder, org.tensorflow.proto.distruntime.ClusterDeviceFiltersOrBuilder> clusterDeviceFiltersBuilder_;
-    /**
-     * <pre>
-     * Device filters for remote tasks in the cluster.
-     * NOTE: This is an experimental feature and only effective in TensorFlow 2.x.
-     * </pre>
-     *
-     * <code>.tensorflow.ClusterDeviceFilters cluster_device_filters = 7;</code>
-     */
-    public boolean hasClusterDeviceFilters() {
-      return clusterDeviceFiltersBuilder_ != null || clusterDeviceFilters_ != null;
-    }
-    /**
-     * <pre>
-     * Device filters for remote tasks in the cluster.
-     * NOTE: This is an experimental feature and only effective in TensorFlow 2.x.
-     * </pre>
-     *
-     * <code>.tensorflow.ClusterDeviceFilters cluster_device_filters = 7;</code>
-     */
-    public org.tensorflow.proto.distruntime.ClusterDeviceFilters getClusterDeviceFilters() {
-      if (clusterDeviceFiltersBuilder_ == null) {
-        return clusterDeviceFilters_ == null ? org.tensorflow.proto.distruntime.ClusterDeviceFilters.getDefaultInstance() : clusterDeviceFilters_;
-      } else {
-        return clusterDeviceFiltersBuilder_.getMessage();
-      }
-    }
-    /**
-     * <pre>
-     * Device filters for remote tasks in the cluster.
-     * NOTE: This is an experimental feature and only effective in TensorFlow 2.x.
-     * </pre>
-     *
-     * <code>.tensorflow.ClusterDeviceFilters cluster_device_filters = 7;</code>
-     */
-    public Builder setClusterDeviceFilters(org.tensorflow.proto.distruntime.ClusterDeviceFilters value) {
-      if (clusterDeviceFiltersBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        clusterDeviceFilters_ = value;
-        onChanged();
-      } else {
-        clusterDeviceFiltersBuilder_.setMessage(value);
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Device filters for remote tasks in the cluster.
-     * NOTE: This is an experimental feature and only effective in TensorFlow 2.x.
-     * </pre>
-     *
-     * <code>.tensorflow.ClusterDeviceFilters cluster_device_filters = 7;</code>
-     */
-    public Builder setClusterDeviceFilters(
-        org.tensorflow.proto.distruntime.ClusterDeviceFilters.Builder builderForValue) {
-      if (clusterDeviceFiltersBuilder_ == null) {
-        clusterDeviceFilters_ = builderForValue.build();
-        onChanged();
-      } else {
-        clusterDeviceFiltersBuilder_.setMessage(builderForValue.build());
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Device filters for remote tasks in the cluster.
-     * NOTE: This is an experimental feature and only effective in TensorFlow 2.x.
-     * </pre>
-     *
-     * <code>.tensorflow.ClusterDeviceFilters cluster_device_filters = 7;</code>
-     */
-    public Builder mergeClusterDeviceFilters(org.tensorflow.proto.distruntime.ClusterDeviceFilters value) {
-      if (clusterDeviceFiltersBuilder_ == null) {
-        if (clusterDeviceFilters_ != null) {
-          clusterDeviceFilters_ =
-            org.tensorflow.proto.distruntime.ClusterDeviceFilters.newBuilder(clusterDeviceFilters_).mergeFrom(value).buildPartial();
-        } else {
-          clusterDeviceFilters_ = value;
-        }
-        onChanged();
-      } else {
-        clusterDeviceFiltersBuilder_.mergeFrom(value);
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Device filters for remote tasks in the cluster.
-     * NOTE: This is an experimental feature and only effective in TensorFlow 2.x.
-     * </pre>
-     *
-     * <code>.tensorflow.ClusterDeviceFilters cluster_device_filters = 7;</code>
-     */
-    public Builder clearClusterDeviceFilters() {
-      if (clusterDeviceFiltersBuilder_ == null) {
-        clusterDeviceFilters_ = null;
-        onChanged();
-      } else {
-        clusterDeviceFilters_ = null;
-        clusterDeviceFiltersBuilder_ = null;
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Device filters for remote tasks in the cluster.
-     * NOTE: This is an experimental feature and only effective in TensorFlow 2.x.
-     * </pre>
-     *
-     * <code>.tensorflow.ClusterDeviceFilters cluster_device_filters = 7;</code>
-     */
-    public org.tensorflow.proto.distruntime.ClusterDeviceFilters.Builder getClusterDeviceFiltersBuilder() {
-      
-      onChanged();
-      return getClusterDeviceFiltersFieldBuilder().getBuilder();
-    }
-    /**
-     * <pre>
-     * Device filters for remote tasks in the cluster.
-     * NOTE: This is an experimental feature and only effective in TensorFlow 2.x.
-     * </pre>
-     *
-     * <code>.tensorflow.ClusterDeviceFilters cluster_device_filters = 7;</code>
-     */
-    public org.tensorflow.proto.distruntime.ClusterDeviceFiltersOrBuilder getClusterDeviceFiltersOrBuilder() {
-      if (clusterDeviceFiltersBuilder_ != null) {
-        return clusterDeviceFiltersBuilder_.getMessageOrBuilder();
-      } else {
-        return clusterDeviceFilters_ == null ?
-            org.tensorflow.proto.distruntime.ClusterDeviceFilters.getDefaultInstance() : clusterDeviceFilters_;
-      }
-    }
-    /**
-     * <pre>
-     * Device filters for remote tasks in the cluster.
-     * NOTE: This is an experimental feature and only effective in TensorFlow 2.x.
-     * </pre>
-     *
-     * <code>.tensorflow.ClusterDeviceFilters cluster_device_filters = 7;</code>
-     */
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.distruntime.ClusterDeviceFilters, org.tensorflow.proto.distruntime.ClusterDeviceFilters.Builder, org.tensorflow.proto.distruntime.ClusterDeviceFiltersOrBuilder> 
-        getClusterDeviceFiltersFieldBuilder() {
-      if (clusterDeviceFiltersBuilder_ == null) {
-        clusterDeviceFiltersBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-            org.tensorflow.proto.distruntime.ClusterDeviceFilters, org.tensorflow.proto.distruntime.ClusterDeviceFilters.Builder, org.tensorflow.proto.distruntime.ClusterDeviceFiltersOrBuilder>(
-                getClusterDeviceFilters(),
-                getParentForChildren(),
-                isClean());
-        clusterDeviceFilters_ = null;
-      }
-      return clusterDeviceFiltersBuilder_;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.ServerDef)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.ServerDef)
-  private static final org.tensorflow.proto.distruntime.ServerDef DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.distruntime.ServerDef();
-  }
-
-  public static org.tensorflow.proto.distruntime.ServerDef getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<ServerDef>
-      PARSER = new com.google.protobuf.AbstractParser<ServerDef>() {
-    @java.lang.Override
-    public ServerDef parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new ServerDef(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<ServerDef> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<ServerDef> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.distruntime.ServerDef getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/distruntime/ServerDefOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/distruntime/ServerDefOrBuilder.java
deleted file mode 100644
index 6c1a12b002d..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/distruntime/ServerDefOrBuilder.java
+++ /dev/null
@@ -1,149 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/tensorflow_server.proto
-
-package org.tensorflow.proto.distruntime;
-
-public interface ServerDefOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.ServerDef)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <pre>
-   * The cluster of which this server is a member.
-   * </pre>
-   *
-   * <code>.tensorflow.ClusterDef cluster = 1;</code>
-   */
-  boolean hasCluster();
-  /**
-   * <pre>
-   * The cluster of which this server is a member.
-   * </pre>
-   *
-   * <code>.tensorflow.ClusterDef cluster = 1;</code>
-   */
-  org.tensorflow.proto.distruntime.ClusterDef getCluster();
-  /**
-   * <pre>
-   * The cluster of which this server is a member.
-   * </pre>
-   *
-   * <code>.tensorflow.ClusterDef cluster = 1;</code>
-   */
-  org.tensorflow.proto.distruntime.ClusterDefOrBuilder getClusterOrBuilder();
-
-  /**
-   * <pre>
-   * The name of the job of which this server is a member.
-   * NOTE(mrry): The `cluster` field must contain a `JobDef` with a `name` field
-   * that matches this name.
-   * </pre>
-   *
-   * <code>string job_name = 2;</code>
-   */
-  java.lang.String getJobName();
-  /**
-   * <pre>
-   * The name of the job of which this server is a member.
-   * NOTE(mrry): The `cluster` field must contain a `JobDef` with a `name` field
-   * that matches this name.
-   * </pre>
-   *
-   * <code>string job_name = 2;</code>
-   */
-  com.google.protobuf.ByteString
-      getJobNameBytes();
-
-  /**
-   * <pre>
-   * The task index of this server in its job.
-   * NOTE: The `cluster` field must contain a `JobDef` with a matching `name`
-   * and a mapping in its `tasks` field for this index.
-   * </pre>
-   *
-   * <code>int32 task_index = 3;</code>
-   */
-  int getTaskIndex();
-
-  /**
-   * <pre>
-   * The default configuration for sessions that run on this server.
-   * </pre>
-   *
-   * <code>.tensorflow.ConfigProto default_session_config = 4;</code>
-   */
-  boolean hasDefaultSessionConfig();
-  /**
-   * <pre>
-   * The default configuration for sessions that run on this server.
-   * </pre>
-   *
-   * <code>.tensorflow.ConfigProto default_session_config = 4;</code>
-   */
-  org.tensorflow.proto.framework.ConfigProto getDefaultSessionConfig();
-  /**
-   * <pre>
-   * The default configuration for sessions that run on this server.
-   * </pre>
-   *
-   * <code>.tensorflow.ConfigProto default_session_config = 4;</code>
-   */
-  org.tensorflow.proto.framework.ConfigProtoOrBuilder getDefaultSessionConfigOrBuilder();
-
-  /**
-   * <pre>
-   * The protocol to be used by this server.
-   * Acceptable values include: "grpc", "grpc+verbs".
-   * </pre>
-   *
-   * <code>string protocol = 5;</code>
-   */
-  java.lang.String getProtocol();
-  /**
-   * <pre>
-   * The protocol to be used by this server.
-   * Acceptable values include: "grpc", "grpc+verbs".
-   * </pre>
-   *
-   * <code>string protocol = 5;</code>
-   */
-  com.google.protobuf.ByteString
-      getProtocolBytes();
-
-  /**
-   * <pre>
-   * The server port. If not set, then we identify the port from the job_name.
-   * </pre>
-   *
-   * <code>int32 port = 6;</code>
-   */
-  int getPort();
-
-  /**
-   * <pre>
-   * Device filters for remote tasks in the cluster.
-   * NOTE: This is an experimental feature and only effective in TensorFlow 2.x.
-   * </pre>
-   *
-   * <code>.tensorflow.ClusterDeviceFilters cluster_device_filters = 7;</code>
-   */
-  boolean hasClusterDeviceFilters();
-  /**
-   * <pre>
-   * Device filters for remote tasks in the cluster.
-   * NOTE: This is an experimental feature and only effective in TensorFlow 2.x.
-   * </pre>
-   *
-   * <code>.tensorflow.ClusterDeviceFilters cluster_device_filters = 7;</code>
-   */
-  org.tensorflow.proto.distruntime.ClusterDeviceFilters getClusterDeviceFilters();
-  /**
-   * <pre>
-   * Device filters for remote tasks in the cluster.
-   * NOTE: This is an experimental feature and only effective in TensorFlow 2.x.
-   * </pre>
-   *
-   * <code>.tensorflow.ClusterDeviceFilters cluster_device_filters = 7;</code>
-   */
-  org.tensorflow.proto.distruntime.ClusterDeviceFiltersOrBuilder getClusterDeviceFiltersOrBuilder();
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/distruntime/ServerProtos.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/distruntime/ServerProtos.java
deleted file mode 100644
index d5b21d1500d..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/distruntime/ServerProtos.java
+++ /dev/null
@@ -1,66 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/tensorflow_server.proto
-
-package org.tensorflow.proto.distruntime;
-
-public final class ServerProtos {
-  private ServerProtos() {}
-  public static void registerAllExtensions(
-      com.google.protobuf.ExtensionRegistryLite registry) {
-  }
-
-  public static void registerAllExtensions(
-      com.google.protobuf.ExtensionRegistry registry) {
-    registerAllExtensions(
-        (com.google.protobuf.ExtensionRegistryLite) registry);
-  }
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_ServerDef_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_ServerDef_fieldAccessorTable;
-
-  public static com.google.protobuf.Descriptors.FileDescriptor
-      getDescriptor() {
-    return descriptor;
-  }
-  private static  com.google.protobuf.Descriptors.FileDescriptor
-      descriptor;
-  static {
-    java.lang.String[] descriptorData = {
-      "\n0tensorflow/core/protobuf/tensorflow_se" +
-      "rver.proto\022\ntensorflow\032&tensorflow/core/" +
-      "protobuf/cluster.proto\032%tensorflow/core/" +
-      "protobuf/config.proto\032-tensorflow/core/p" +
-      "rotobuf/device_filters.proto\"\365\001\n\tServerD" +
-      "ef\022\'\n\007cluster\030\001 \001(\0132\026.tensorflow.Cluster" +
-      "Def\022\020\n\010job_name\030\002 \001(\t\022\022\n\ntask_index\030\003 \001(" +
-      "\005\0227\n\026default_session_config\030\004 \001(\0132\027.tens" +
-      "orflow.ConfigProto\022\020\n\010protocol\030\005 \001(\t\022\014\n\004" +
-      "port\030\006 \001(\005\022@\n\026cluster_device_filters\030\007 \001" +
-      "(\0132 .tensorflow.ClusterDeviceFiltersB\177\n " +
-      "org.tensorflow.proto.distruntimeB\014Server" +
-      "ProtosP\001ZHgithub.com/tensorflow/tensorfl" +
-      "ow/tensorflow/go/core/core_protos_go_pro" +
-      "to\370\001\001b\006proto3"
-    };
-    descriptor = com.google.protobuf.Descriptors.FileDescriptor
-      .internalBuildGeneratedFileFrom(descriptorData,
-        new com.google.protobuf.Descriptors.FileDescriptor[] {
-          org.tensorflow.proto.distruntime.ClusterProtos.getDescriptor(),
-          org.tensorflow.proto.framework.ConfigProtos.getDescriptor(),
-          org.tensorflow.proto.distruntime.DeviceFiltersProtos.getDescriptor(),
-        });
-    internal_static_tensorflow_ServerDef_descriptor =
-      getDescriptor().getMessageTypes().get(0);
-    internal_static_tensorflow_ServerDef_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_ServerDef_descriptor,
-        new java.lang.String[] { "Cluster", "JobName", "TaskIndex", "DefaultSessionConfig", "Protocol", "Port", "ClusterDeviceFilters", });
-    org.tensorflow.proto.distruntime.ClusterProtos.getDescriptor();
-    org.tensorflow.proto.framework.ConfigProtos.getDescriptor();
-    org.tensorflow.proto.distruntime.DeviceFiltersProtos.getDescriptor();
-  }
-
-  // @@protoc_insertion_point(outer_class_scope)
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/distruntime/TaskDeviceFilters.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/distruntime/TaskDeviceFilters.java
deleted file mode 100644
index 50bc759c880..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/distruntime/TaskDeviceFilters.java
+++ /dev/null
@@ -1,600 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/device_filters.proto
-
-package org.tensorflow.proto.distruntime;
-
-/**
- * <pre>
- * Defines the device filters for a remote task.
- * </pre>
- *
- * Protobuf type {@code tensorflow.TaskDeviceFilters}
- */
-public  final class TaskDeviceFilters extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.TaskDeviceFilters)
-    TaskDeviceFiltersOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use TaskDeviceFilters.newBuilder() to construct.
-  private TaskDeviceFilters(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private TaskDeviceFilters() {
-    deviceFilters_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new TaskDeviceFilters();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private TaskDeviceFilters(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    int mutable_bitField0_ = 0;
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 10: {
-            java.lang.String s = input.readStringRequireUtf8();
-            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
-              deviceFilters_ = new com.google.protobuf.LazyStringArrayList();
-              mutable_bitField0_ |= 0x00000001;
-            }
-            deviceFilters_.add(s);
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      if (((mutable_bitField0_ & 0x00000001) != 0)) {
-        deviceFilters_ = deviceFilters_.getUnmodifiableView();
-      }
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.distruntime.DeviceFiltersProtos.internal_static_tensorflow_TaskDeviceFilters_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.distruntime.DeviceFiltersProtos.internal_static_tensorflow_TaskDeviceFilters_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.distruntime.TaskDeviceFilters.class, org.tensorflow.proto.distruntime.TaskDeviceFilters.Builder.class);
-  }
-
-  public static final int DEVICE_FILTERS_FIELD_NUMBER = 1;
-  private com.google.protobuf.LazyStringList deviceFilters_;
-  /**
-   * <code>repeated string device_filters = 1;</code>
-   */
-  public com.google.protobuf.ProtocolStringList
-      getDeviceFiltersList() {
-    return deviceFilters_;
-  }
-  /**
-   * <code>repeated string device_filters = 1;</code>
-   */
-  public int getDeviceFiltersCount() {
-    return deviceFilters_.size();
-  }
-  /**
-   * <code>repeated string device_filters = 1;</code>
-   */
-  public java.lang.String getDeviceFilters(int index) {
-    return deviceFilters_.get(index);
-  }
-  /**
-   * <code>repeated string device_filters = 1;</code>
-   */
-  public com.google.protobuf.ByteString
-      getDeviceFiltersBytes(int index) {
-    return deviceFilters_.getByteString(index);
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    for (int i = 0; i < deviceFilters_.size(); i++) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 1, deviceFilters_.getRaw(i));
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    {
-      int dataSize = 0;
-      for (int i = 0; i < deviceFilters_.size(); i++) {
-        dataSize += computeStringSizeNoTag(deviceFilters_.getRaw(i));
-      }
-      size += dataSize;
-      size += 1 * getDeviceFiltersList().size();
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.distruntime.TaskDeviceFilters)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.distruntime.TaskDeviceFilters other = (org.tensorflow.proto.distruntime.TaskDeviceFilters) obj;
-
-    if (!getDeviceFiltersList()
-        .equals(other.getDeviceFiltersList())) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    if (getDeviceFiltersCount() > 0) {
-      hash = (37 * hash) + DEVICE_FILTERS_FIELD_NUMBER;
-      hash = (53 * hash) + getDeviceFiltersList().hashCode();
-    }
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.distruntime.TaskDeviceFilters parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.distruntime.TaskDeviceFilters parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.distruntime.TaskDeviceFilters parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.distruntime.TaskDeviceFilters parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.distruntime.TaskDeviceFilters parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.distruntime.TaskDeviceFilters parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.distruntime.TaskDeviceFilters parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.distruntime.TaskDeviceFilters parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.distruntime.TaskDeviceFilters parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.distruntime.TaskDeviceFilters parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.distruntime.TaskDeviceFilters parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.distruntime.TaskDeviceFilters parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.distruntime.TaskDeviceFilters prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * <pre>
-   * Defines the device filters for a remote task.
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.TaskDeviceFilters}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.TaskDeviceFilters)
-      org.tensorflow.proto.distruntime.TaskDeviceFiltersOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.distruntime.DeviceFiltersProtos.internal_static_tensorflow_TaskDeviceFilters_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.distruntime.DeviceFiltersProtos.internal_static_tensorflow_TaskDeviceFilters_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.distruntime.TaskDeviceFilters.class, org.tensorflow.proto.distruntime.TaskDeviceFilters.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.distruntime.TaskDeviceFilters.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      deviceFilters_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-      bitField0_ = (bitField0_ & ~0x00000001);
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.distruntime.DeviceFiltersProtos.internal_static_tensorflow_TaskDeviceFilters_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.distruntime.TaskDeviceFilters getDefaultInstanceForType() {
-      return org.tensorflow.proto.distruntime.TaskDeviceFilters.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.distruntime.TaskDeviceFilters build() {
-      org.tensorflow.proto.distruntime.TaskDeviceFilters result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.distruntime.TaskDeviceFilters buildPartial() {
-      org.tensorflow.proto.distruntime.TaskDeviceFilters result = new org.tensorflow.proto.distruntime.TaskDeviceFilters(this);
-      int from_bitField0_ = bitField0_;
-      if (((bitField0_ & 0x00000001) != 0)) {
-        deviceFilters_ = deviceFilters_.getUnmodifiableView();
-        bitField0_ = (bitField0_ & ~0x00000001);
-      }
-      result.deviceFilters_ = deviceFilters_;
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.distruntime.TaskDeviceFilters) {
-        return mergeFrom((org.tensorflow.proto.distruntime.TaskDeviceFilters)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.distruntime.TaskDeviceFilters other) {
-      if (other == org.tensorflow.proto.distruntime.TaskDeviceFilters.getDefaultInstance()) return this;
-      if (!other.deviceFilters_.isEmpty()) {
-        if (deviceFilters_.isEmpty()) {
-          deviceFilters_ = other.deviceFilters_;
-          bitField0_ = (bitField0_ & ~0x00000001);
-        } else {
-          ensureDeviceFiltersIsMutable();
-          deviceFilters_.addAll(other.deviceFilters_);
-        }
-        onChanged();
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.distruntime.TaskDeviceFilters parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.distruntime.TaskDeviceFilters) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-    private int bitField0_;
-
-    private com.google.protobuf.LazyStringList deviceFilters_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-    private void ensureDeviceFiltersIsMutable() {
-      if (!((bitField0_ & 0x00000001) != 0)) {
-        deviceFilters_ = new com.google.protobuf.LazyStringArrayList(deviceFilters_);
-        bitField0_ |= 0x00000001;
-       }
-    }
-    /**
-     * <code>repeated string device_filters = 1;</code>
-     */
-    public com.google.protobuf.ProtocolStringList
-        getDeviceFiltersList() {
-      return deviceFilters_.getUnmodifiableView();
-    }
-    /**
-     * <code>repeated string device_filters = 1;</code>
-     */
-    public int getDeviceFiltersCount() {
-      return deviceFilters_.size();
-    }
-    /**
-     * <code>repeated string device_filters = 1;</code>
-     */
-    public java.lang.String getDeviceFilters(int index) {
-      return deviceFilters_.get(index);
-    }
-    /**
-     * <code>repeated string device_filters = 1;</code>
-     */
-    public com.google.protobuf.ByteString
-        getDeviceFiltersBytes(int index) {
-      return deviceFilters_.getByteString(index);
-    }
-    /**
-     * <code>repeated string device_filters = 1;</code>
-     */
-    public Builder setDeviceFilters(
-        int index, java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  ensureDeviceFiltersIsMutable();
-      deviceFilters_.set(index, value);
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>repeated string device_filters = 1;</code>
-     */
-    public Builder addDeviceFilters(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  ensureDeviceFiltersIsMutable();
-      deviceFilters_.add(value);
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>repeated string device_filters = 1;</code>
-     */
-    public Builder addAllDeviceFilters(
-        java.lang.Iterable<java.lang.String> values) {
-      ensureDeviceFiltersIsMutable();
-      com.google.protobuf.AbstractMessageLite.Builder.addAll(
-          values, deviceFilters_);
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>repeated string device_filters = 1;</code>
-     */
-    public Builder clearDeviceFilters() {
-      deviceFilters_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-      bitField0_ = (bitField0_ & ~0x00000001);
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>repeated string device_filters = 1;</code>
-     */
-    public Builder addDeviceFiltersBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      ensureDeviceFiltersIsMutable();
-      deviceFilters_.add(value);
-      onChanged();
-      return this;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.TaskDeviceFilters)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.TaskDeviceFilters)
-  private static final org.tensorflow.proto.distruntime.TaskDeviceFilters DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.distruntime.TaskDeviceFilters();
-  }
-
-  public static org.tensorflow.proto.distruntime.TaskDeviceFilters getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<TaskDeviceFilters>
-      PARSER = new com.google.protobuf.AbstractParser<TaskDeviceFilters>() {
-    @java.lang.Override
-    public TaskDeviceFilters parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new TaskDeviceFilters(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<TaskDeviceFilters> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<TaskDeviceFilters> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.distruntime.TaskDeviceFilters getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/distruntime/TaskDeviceFiltersOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/distruntime/TaskDeviceFiltersOrBuilder.java
deleted file mode 100644
index 2c359e1e049..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/distruntime/TaskDeviceFiltersOrBuilder.java
+++ /dev/null
@@ -1,28 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/device_filters.proto
-
-package org.tensorflow.proto.distruntime;
-
-public interface TaskDeviceFiltersOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.TaskDeviceFilters)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <code>repeated string device_filters = 1;</code>
-   */
-  java.util.List<java.lang.String>
-      getDeviceFiltersList();
-  /**
-   * <code>repeated string device_filters = 1;</code>
-   */
-  int getDeviceFiltersCount();
-  /**
-   * <code>repeated string device_filters = 1;</code>
-   */
-  java.lang.String getDeviceFilters(int index);
-  /**
-   * <code>repeated string device_filters = 1;</code>
-   */
-  com.google.protobuf.ByteString
-      getDeviceFiltersBytes(int index);
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/distruntime/TransportOptions.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/distruntime/TransportOptions.java
deleted file mode 100644
index f7498e56de5..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/distruntime/TransportOptions.java
+++ /dev/null
@@ -1,635 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/transport_options.proto
-
-package org.tensorflow.proto.distruntime;
-
-public final class TransportOptions {
-  private TransportOptions() {}
-  public static void registerAllExtensions(
-      com.google.protobuf.ExtensionRegistryLite registry) {
-  }
-
-  public static void registerAllExtensions(
-      com.google.protobuf.ExtensionRegistry registry) {
-    registerAllExtensions(
-        (com.google.protobuf.ExtensionRegistryLite) registry);
-  }
-  public interface RecvBufRespExtraOrBuilder extends
-      // @@protoc_insertion_point(interface_extends:tensorflow.RecvBufRespExtra)
-      com.google.protobuf.MessageOrBuilder {
-
-    /**
-     * <code>repeated bytes tensor_content = 1;</code>
-     */
-    java.util.List<com.google.protobuf.ByteString> getTensorContentList();
-    /**
-     * <code>repeated bytes tensor_content = 1;</code>
-     */
-    int getTensorContentCount();
-    /**
-     * <code>repeated bytes tensor_content = 1;</code>
-     */
-    com.google.protobuf.ByteString getTensorContent(int index);
-  }
-  /**
-   * <pre>
-   * Extra data needed on a non-RDMA RecvBufResponse.
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.RecvBufRespExtra}
-   */
-  public  static final class RecvBufRespExtra extends
-      com.google.protobuf.GeneratedMessageV3 implements
-      // @@protoc_insertion_point(message_implements:tensorflow.RecvBufRespExtra)
-      RecvBufRespExtraOrBuilder {
-  private static final long serialVersionUID = 0L;
-    // Use RecvBufRespExtra.newBuilder() to construct.
-    private RecvBufRespExtra(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-      super(builder);
-    }
-    private RecvBufRespExtra() {
-      tensorContent_ = java.util.Collections.emptyList();
-    }
-
-    @java.lang.Override
-    @SuppressWarnings({"unused"})
-    protected java.lang.Object newInstance(
-        UnusedPrivateParameter unused) {
-      return new RecvBufRespExtra();
-    }
-
-    @java.lang.Override
-    public final com.google.protobuf.UnknownFieldSet
-    getUnknownFields() {
-      return this.unknownFields;
-    }
-    private RecvBufRespExtra(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      this();
-      if (extensionRegistry == null) {
-        throw new java.lang.NullPointerException();
-      }
-      int mutable_bitField0_ = 0;
-      com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-          com.google.protobuf.UnknownFieldSet.newBuilder();
-      try {
-        boolean done = false;
-        while (!done) {
-          int tag = input.readTag();
-          switch (tag) {
-            case 0:
-              done = true;
-              break;
-            case 10: {
-              if (!((mutable_bitField0_ & 0x00000001) != 0)) {
-                tensorContent_ = new java.util.ArrayList<com.google.protobuf.ByteString>();
-                mutable_bitField0_ |= 0x00000001;
-              }
-              tensorContent_.add(input.readBytes());
-              break;
-            }
-            default: {
-              if (!parseUnknownField(
-                  input, unknownFields, extensionRegistry, tag)) {
-                done = true;
-              }
-              break;
-            }
-          }
-        }
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        throw e.setUnfinishedMessage(this);
-      } catch (java.io.IOException e) {
-        throw new com.google.protobuf.InvalidProtocolBufferException(
-            e).setUnfinishedMessage(this);
-      } finally {
-        if (((mutable_bitField0_ & 0x00000001) != 0)) {
-          tensorContent_ = java.util.Collections.unmodifiableList(tensorContent_); // C
-        }
-        this.unknownFields = unknownFields.build();
-        makeExtensionsImmutable();
-      }
-    }
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.distruntime.TransportOptions.internal_static_tensorflow_RecvBufRespExtra_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.distruntime.TransportOptions.internal_static_tensorflow_RecvBufRespExtra_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.distruntime.TransportOptions.RecvBufRespExtra.class, org.tensorflow.proto.distruntime.TransportOptions.RecvBufRespExtra.Builder.class);
-    }
-
-    public static final int TENSOR_CONTENT_FIELD_NUMBER = 1;
-    private java.util.List<com.google.protobuf.ByteString> tensorContent_;
-    /**
-     * <code>repeated bytes tensor_content = 1;</code>
-     */
-    public java.util.List<com.google.protobuf.ByteString>
-        getTensorContentList() {
-      return tensorContent_;
-    }
-    /**
-     * <code>repeated bytes tensor_content = 1;</code>
-     */
-    public int getTensorContentCount() {
-      return tensorContent_.size();
-    }
-    /**
-     * <code>repeated bytes tensor_content = 1;</code>
-     */
-    public com.google.protobuf.ByteString getTensorContent(int index) {
-      return tensorContent_.get(index);
-    }
-
-    private byte memoizedIsInitialized = -1;
-    @java.lang.Override
-    public final boolean isInitialized() {
-      byte isInitialized = memoizedIsInitialized;
-      if (isInitialized == 1) return true;
-      if (isInitialized == 0) return false;
-
-      memoizedIsInitialized = 1;
-      return true;
-    }
-
-    @java.lang.Override
-    public void writeTo(com.google.protobuf.CodedOutputStream output)
-                        throws java.io.IOException {
-      for (int i = 0; i < tensorContent_.size(); i++) {
-        output.writeBytes(1, tensorContent_.get(i));
-      }
-      unknownFields.writeTo(output);
-    }
-
-    @java.lang.Override
-    public int getSerializedSize() {
-      int size = memoizedSize;
-      if (size != -1) return size;
-
-      size = 0;
-      {
-        int dataSize = 0;
-        for (int i = 0; i < tensorContent_.size(); i++) {
-          dataSize += com.google.protobuf.CodedOutputStream
-            .computeBytesSizeNoTag(tensorContent_.get(i));
-        }
-        size += dataSize;
-        size += 1 * getTensorContentList().size();
-      }
-      size += unknownFields.getSerializedSize();
-      memoizedSize = size;
-      return size;
-    }
-
-    @java.lang.Override
-    public boolean equals(final java.lang.Object obj) {
-      if (obj == this) {
-       return true;
-      }
-      if (!(obj instanceof org.tensorflow.proto.distruntime.TransportOptions.RecvBufRespExtra)) {
-        return super.equals(obj);
-      }
-      org.tensorflow.proto.distruntime.TransportOptions.RecvBufRespExtra other = (org.tensorflow.proto.distruntime.TransportOptions.RecvBufRespExtra) obj;
-
-      if (!getTensorContentList()
-          .equals(other.getTensorContentList())) return false;
-      if (!unknownFields.equals(other.unknownFields)) return false;
-      return true;
-    }
-
-    @java.lang.Override
-    public int hashCode() {
-      if (memoizedHashCode != 0) {
-        return memoizedHashCode;
-      }
-      int hash = 41;
-      hash = (19 * hash) + getDescriptor().hashCode();
-      if (getTensorContentCount() > 0) {
-        hash = (37 * hash) + TENSOR_CONTENT_FIELD_NUMBER;
-        hash = (53 * hash) + getTensorContentList().hashCode();
-      }
-      hash = (29 * hash) + unknownFields.hashCode();
-      memoizedHashCode = hash;
-      return hash;
-    }
-
-    public static org.tensorflow.proto.distruntime.TransportOptions.RecvBufRespExtra parseFrom(
-        java.nio.ByteBuffer data)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data);
-    }
-    public static org.tensorflow.proto.distruntime.TransportOptions.RecvBufRespExtra parseFrom(
-        java.nio.ByteBuffer data,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data, extensionRegistry);
-    }
-    public static org.tensorflow.proto.distruntime.TransportOptions.RecvBufRespExtra parseFrom(
-        com.google.protobuf.ByteString data)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data);
-    }
-    public static org.tensorflow.proto.distruntime.TransportOptions.RecvBufRespExtra parseFrom(
-        com.google.protobuf.ByteString data,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data, extensionRegistry);
-    }
-    public static org.tensorflow.proto.distruntime.TransportOptions.RecvBufRespExtra parseFrom(byte[] data)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data);
-    }
-    public static org.tensorflow.proto.distruntime.TransportOptions.RecvBufRespExtra parseFrom(
-        byte[] data,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data, extensionRegistry);
-    }
-    public static org.tensorflow.proto.distruntime.TransportOptions.RecvBufRespExtra parseFrom(java.io.InputStream input)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input);
-    }
-    public static org.tensorflow.proto.distruntime.TransportOptions.RecvBufRespExtra parseFrom(
-        java.io.InputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input, extensionRegistry);
-    }
-    public static org.tensorflow.proto.distruntime.TransportOptions.RecvBufRespExtra parseDelimitedFrom(java.io.InputStream input)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseDelimitedWithIOException(PARSER, input);
-    }
-    public static org.tensorflow.proto.distruntime.TransportOptions.RecvBufRespExtra parseDelimitedFrom(
-        java.io.InputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-    }
-    public static org.tensorflow.proto.distruntime.TransportOptions.RecvBufRespExtra parseFrom(
-        com.google.protobuf.CodedInputStream input)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input);
-    }
-    public static org.tensorflow.proto.distruntime.TransportOptions.RecvBufRespExtra parseFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input, extensionRegistry);
-    }
-
-    @java.lang.Override
-    public Builder newBuilderForType() { return newBuilder(); }
-    public static Builder newBuilder() {
-      return DEFAULT_INSTANCE.toBuilder();
-    }
-    public static Builder newBuilder(org.tensorflow.proto.distruntime.TransportOptions.RecvBufRespExtra prototype) {
-      return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-    }
-    @java.lang.Override
-    public Builder toBuilder() {
-      return this == DEFAULT_INSTANCE
-          ? new Builder() : new Builder().mergeFrom(this);
-    }
-
-    @java.lang.Override
-    protected Builder newBuilderForType(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      Builder builder = new Builder(parent);
-      return builder;
-    }
-    /**
-     * <pre>
-     * Extra data needed on a non-RDMA RecvBufResponse.
-     * </pre>
-     *
-     * Protobuf type {@code tensorflow.RecvBufRespExtra}
-     */
-    public static final class Builder extends
-        com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-        // @@protoc_insertion_point(builder_implements:tensorflow.RecvBufRespExtra)
-        org.tensorflow.proto.distruntime.TransportOptions.RecvBufRespExtraOrBuilder {
-      public static final com.google.protobuf.Descriptors.Descriptor
-          getDescriptor() {
-        return org.tensorflow.proto.distruntime.TransportOptions.internal_static_tensorflow_RecvBufRespExtra_descriptor;
-      }
-
-      @java.lang.Override
-      protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-          internalGetFieldAccessorTable() {
-        return org.tensorflow.proto.distruntime.TransportOptions.internal_static_tensorflow_RecvBufRespExtra_fieldAccessorTable
-            .ensureFieldAccessorsInitialized(
-                org.tensorflow.proto.distruntime.TransportOptions.RecvBufRespExtra.class, org.tensorflow.proto.distruntime.TransportOptions.RecvBufRespExtra.Builder.class);
-      }
-
-      // Construct using org.tensorflow.proto.distruntime.TransportOptions.RecvBufRespExtra.newBuilder()
-      private Builder() {
-        maybeForceBuilderInitialization();
-      }
-
-      private Builder(
-          com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-        super(parent);
-        maybeForceBuilderInitialization();
-      }
-      private void maybeForceBuilderInitialization() {
-        if (com.google.protobuf.GeneratedMessageV3
-                .alwaysUseFieldBuilders) {
-        }
-      }
-      @java.lang.Override
-      public Builder clear() {
-        super.clear();
-        tensorContent_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000001);
-        return this;
-      }
-
-      @java.lang.Override
-      public com.google.protobuf.Descriptors.Descriptor
-          getDescriptorForType() {
-        return org.tensorflow.proto.distruntime.TransportOptions.internal_static_tensorflow_RecvBufRespExtra_descriptor;
-      }
-
-      @java.lang.Override
-      public org.tensorflow.proto.distruntime.TransportOptions.RecvBufRespExtra getDefaultInstanceForType() {
-        return org.tensorflow.proto.distruntime.TransportOptions.RecvBufRespExtra.getDefaultInstance();
-      }
-
-      @java.lang.Override
-      public org.tensorflow.proto.distruntime.TransportOptions.RecvBufRespExtra build() {
-        org.tensorflow.proto.distruntime.TransportOptions.RecvBufRespExtra result = buildPartial();
-        if (!result.isInitialized()) {
-          throw newUninitializedMessageException(result);
-        }
-        return result;
-      }
-
-      @java.lang.Override
-      public org.tensorflow.proto.distruntime.TransportOptions.RecvBufRespExtra buildPartial() {
-        org.tensorflow.proto.distruntime.TransportOptions.RecvBufRespExtra result = new org.tensorflow.proto.distruntime.TransportOptions.RecvBufRespExtra(this);
-        int from_bitField0_ = bitField0_;
-        if (((bitField0_ & 0x00000001) != 0)) {
-          tensorContent_ = java.util.Collections.unmodifiableList(tensorContent_);
-          bitField0_ = (bitField0_ & ~0x00000001);
-        }
-        result.tensorContent_ = tensorContent_;
-        onBuilt();
-        return result;
-      }
-
-      @java.lang.Override
-      public Builder clone() {
-        return super.clone();
-      }
-      @java.lang.Override
-      public Builder setField(
-          com.google.protobuf.Descriptors.FieldDescriptor field,
-          java.lang.Object value) {
-        return super.setField(field, value);
-      }
-      @java.lang.Override
-      public Builder clearField(
-          com.google.protobuf.Descriptors.FieldDescriptor field) {
-        return super.clearField(field);
-      }
-      @java.lang.Override
-      public Builder clearOneof(
-          com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-        return super.clearOneof(oneof);
-      }
-      @java.lang.Override
-      public Builder setRepeatedField(
-          com.google.protobuf.Descriptors.FieldDescriptor field,
-          int index, java.lang.Object value) {
-        return super.setRepeatedField(field, index, value);
-      }
-      @java.lang.Override
-      public Builder addRepeatedField(
-          com.google.protobuf.Descriptors.FieldDescriptor field,
-          java.lang.Object value) {
-        return super.addRepeatedField(field, value);
-      }
-      @java.lang.Override
-      public Builder mergeFrom(com.google.protobuf.Message other) {
-        if (other instanceof org.tensorflow.proto.distruntime.TransportOptions.RecvBufRespExtra) {
-          return mergeFrom((org.tensorflow.proto.distruntime.TransportOptions.RecvBufRespExtra)other);
-        } else {
-          super.mergeFrom(other);
-          return this;
-        }
-      }
-
-      public Builder mergeFrom(org.tensorflow.proto.distruntime.TransportOptions.RecvBufRespExtra other) {
-        if (other == org.tensorflow.proto.distruntime.TransportOptions.RecvBufRespExtra.getDefaultInstance()) return this;
-        if (!other.tensorContent_.isEmpty()) {
-          if (tensorContent_.isEmpty()) {
-            tensorContent_ = other.tensorContent_;
-            bitField0_ = (bitField0_ & ~0x00000001);
-          } else {
-            ensureTensorContentIsMutable();
-            tensorContent_.addAll(other.tensorContent_);
-          }
-          onChanged();
-        }
-        this.mergeUnknownFields(other.unknownFields);
-        onChanged();
-        return this;
-      }
-
-      @java.lang.Override
-      public final boolean isInitialized() {
-        return true;
-      }
-
-      @java.lang.Override
-      public Builder mergeFrom(
-          com.google.protobuf.CodedInputStream input,
-          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-          throws java.io.IOException {
-        org.tensorflow.proto.distruntime.TransportOptions.RecvBufRespExtra parsedMessage = null;
-        try {
-          parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-        } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-          parsedMessage = (org.tensorflow.proto.distruntime.TransportOptions.RecvBufRespExtra) e.getUnfinishedMessage();
-          throw e.unwrapIOException();
-        } finally {
-          if (parsedMessage != null) {
-            mergeFrom(parsedMessage);
-          }
-        }
-        return this;
-      }
-      private int bitField0_;
-
-      private java.util.List<com.google.protobuf.ByteString> tensorContent_ = java.util.Collections.emptyList();
-      private void ensureTensorContentIsMutable() {
-        if (!((bitField0_ & 0x00000001) != 0)) {
-          tensorContent_ = new java.util.ArrayList<com.google.protobuf.ByteString>(tensorContent_);
-          bitField0_ |= 0x00000001;
-         }
-      }
-      /**
-       * <code>repeated bytes tensor_content = 1;</code>
-       */
-      public java.util.List<com.google.protobuf.ByteString>
-          getTensorContentList() {
-        return ((bitField0_ & 0x00000001) != 0) ?
-                 java.util.Collections.unmodifiableList(tensorContent_) : tensorContent_;
-      }
-      /**
-       * <code>repeated bytes tensor_content = 1;</code>
-       */
-      public int getTensorContentCount() {
-        return tensorContent_.size();
-      }
-      /**
-       * <code>repeated bytes tensor_content = 1;</code>
-       */
-      public com.google.protobuf.ByteString getTensorContent(int index) {
-        return tensorContent_.get(index);
-      }
-      /**
-       * <code>repeated bytes tensor_content = 1;</code>
-       */
-      public Builder setTensorContent(
-          int index, com.google.protobuf.ByteString value) {
-        if (value == null) {
-    throw new NullPointerException();
-  }
-  ensureTensorContentIsMutable();
-        tensorContent_.set(index, value);
-        onChanged();
-        return this;
-      }
-      /**
-       * <code>repeated bytes tensor_content = 1;</code>
-       */
-      public Builder addTensorContent(com.google.protobuf.ByteString value) {
-        if (value == null) {
-    throw new NullPointerException();
-  }
-  ensureTensorContentIsMutable();
-        tensorContent_.add(value);
-        onChanged();
-        return this;
-      }
-      /**
-       * <code>repeated bytes tensor_content = 1;</code>
-       */
-      public Builder addAllTensorContent(
-          java.lang.Iterable<? extends com.google.protobuf.ByteString> values) {
-        ensureTensorContentIsMutable();
-        com.google.protobuf.AbstractMessageLite.Builder.addAll(
-            values, tensorContent_);
-        onChanged();
-        return this;
-      }
-      /**
-       * <code>repeated bytes tensor_content = 1;</code>
-       */
-      public Builder clearTensorContent() {
-        tensorContent_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000001);
-        onChanged();
-        return this;
-      }
-      @java.lang.Override
-      public final Builder setUnknownFields(
-          final com.google.protobuf.UnknownFieldSet unknownFields) {
-        return super.setUnknownFields(unknownFields);
-      }
-
-      @java.lang.Override
-      public final Builder mergeUnknownFields(
-          final com.google.protobuf.UnknownFieldSet unknownFields) {
-        return super.mergeUnknownFields(unknownFields);
-      }
-
-
-      // @@protoc_insertion_point(builder_scope:tensorflow.RecvBufRespExtra)
-    }
-
-    // @@protoc_insertion_point(class_scope:tensorflow.RecvBufRespExtra)
-    private static final org.tensorflow.proto.distruntime.TransportOptions.RecvBufRespExtra DEFAULT_INSTANCE;
-    static {
-      DEFAULT_INSTANCE = new org.tensorflow.proto.distruntime.TransportOptions.RecvBufRespExtra();
-    }
-
-    public static org.tensorflow.proto.distruntime.TransportOptions.RecvBufRespExtra getDefaultInstance() {
-      return DEFAULT_INSTANCE;
-    }
-
-    private static final com.google.protobuf.Parser<RecvBufRespExtra>
-        PARSER = new com.google.protobuf.AbstractParser<RecvBufRespExtra>() {
-      @java.lang.Override
-      public RecvBufRespExtra parsePartialFrom(
-          com.google.protobuf.CodedInputStream input,
-          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-          throws com.google.protobuf.InvalidProtocolBufferException {
-        return new RecvBufRespExtra(input, extensionRegistry);
-      }
-    };
-
-    public static com.google.protobuf.Parser<RecvBufRespExtra> parser() {
-      return PARSER;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Parser<RecvBufRespExtra> getParserForType() {
-      return PARSER;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.distruntime.TransportOptions.RecvBufRespExtra getDefaultInstanceForType() {
-      return DEFAULT_INSTANCE;
-    }
-
-  }
-
-  private static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_RecvBufRespExtra_descriptor;
-  private static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_RecvBufRespExtra_fieldAccessorTable;
-
-  public static com.google.protobuf.Descriptors.FileDescriptor
-      getDescriptor() {
-    return descriptor;
-  }
-  private static  com.google.protobuf.Descriptors.FileDescriptor
-      descriptor;
-  static {
-    java.lang.String[] descriptorData = {
-      "\n0tensorflow/core/protobuf/transport_opt" +
-      "ions.proto\022\ntensorflow\"*\n\020RecvBufRespExt" +
-      "ra\022\026\n\016tensor_content\030\001 \003(\014Bl\n org.tensor" +
-      "flow.proto.distruntimeZHgithub.com/tenso" +
-      "rflow/tensorflow/tensorflow/go/core/core" +
-      "_protos_go_protob\006proto3"
-    };
-    descriptor = com.google.protobuf.Descriptors.FileDescriptor
-      .internalBuildGeneratedFileFrom(descriptorData,
-        new com.google.protobuf.Descriptors.FileDescriptor[] {
-        });
-    internal_static_tensorflow_RecvBufRespExtra_descriptor =
-      getDescriptor().getMessageTypes().get(0);
-    internal_static_tensorflow_RecvBufRespExtra_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_RecvBufRespExtra_descriptor,
-        new java.lang.String[] { "TensorContent", });
-  }
-
-  // @@protoc_insertion_point(outer_class_scope)
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/BytesList.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/BytesList.java
deleted file mode 100644
index d8158b11f38..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/BytesList.java
+++ /dev/null
@@ -1,572 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/example/feature.proto
-
-package org.tensorflow.proto.example;
-
-/**
- * <pre>
- * Containers to hold repeated fundamental values.
- * </pre>
- *
- * Protobuf type {@code tensorflow.BytesList}
- */
-public  final class BytesList extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.BytesList)
-    BytesListOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use BytesList.newBuilder() to construct.
-  private BytesList(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private BytesList() {
-    value_ = java.util.Collections.emptyList();
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new BytesList();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private BytesList(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    int mutable_bitField0_ = 0;
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 10: {
-            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
-              value_ = new java.util.ArrayList<com.google.protobuf.ByteString>();
-              mutable_bitField0_ |= 0x00000001;
-            }
-            value_.add(input.readBytes());
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      if (((mutable_bitField0_ & 0x00000001) != 0)) {
-        value_ = java.util.Collections.unmodifiableList(value_); // C
-      }
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.example.FeatureProtos.internal_static_tensorflow_BytesList_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.example.FeatureProtos.internal_static_tensorflow_BytesList_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.example.BytesList.class, org.tensorflow.proto.example.BytesList.Builder.class);
-  }
-
-  public static final int VALUE_FIELD_NUMBER = 1;
-  private java.util.List<com.google.protobuf.ByteString> value_;
-  /**
-   * <code>repeated bytes value = 1;</code>
-   */
-  public java.util.List<com.google.protobuf.ByteString>
-      getValueList() {
-    return value_;
-  }
-  /**
-   * <code>repeated bytes value = 1;</code>
-   */
-  public int getValueCount() {
-    return value_.size();
-  }
-  /**
-   * <code>repeated bytes value = 1;</code>
-   */
-  public com.google.protobuf.ByteString getValue(int index) {
-    return value_.get(index);
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    for (int i = 0; i < value_.size(); i++) {
-      output.writeBytes(1, value_.get(i));
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    {
-      int dataSize = 0;
-      for (int i = 0; i < value_.size(); i++) {
-        dataSize += com.google.protobuf.CodedOutputStream
-          .computeBytesSizeNoTag(value_.get(i));
-      }
-      size += dataSize;
-      size += 1 * getValueList().size();
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.example.BytesList)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.example.BytesList other = (org.tensorflow.proto.example.BytesList) obj;
-
-    if (!getValueList()
-        .equals(other.getValueList())) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    if (getValueCount() > 0) {
-      hash = (37 * hash) + VALUE_FIELD_NUMBER;
-      hash = (53 * hash) + getValueList().hashCode();
-    }
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.example.BytesList parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.example.BytesList parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.example.BytesList parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.example.BytesList parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.example.BytesList parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.example.BytesList parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.example.BytesList parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.example.BytesList parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.example.BytesList parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.example.BytesList parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.example.BytesList parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.example.BytesList parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.example.BytesList prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * <pre>
-   * Containers to hold repeated fundamental values.
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.BytesList}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.BytesList)
-      org.tensorflow.proto.example.BytesListOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.example.FeatureProtos.internal_static_tensorflow_BytesList_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.example.FeatureProtos.internal_static_tensorflow_BytesList_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.example.BytesList.class, org.tensorflow.proto.example.BytesList.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.example.BytesList.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      value_ = java.util.Collections.emptyList();
-      bitField0_ = (bitField0_ & ~0x00000001);
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.example.FeatureProtos.internal_static_tensorflow_BytesList_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.example.BytesList getDefaultInstanceForType() {
-      return org.tensorflow.proto.example.BytesList.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.example.BytesList build() {
-      org.tensorflow.proto.example.BytesList result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.example.BytesList buildPartial() {
-      org.tensorflow.proto.example.BytesList result = new org.tensorflow.proto.example.BytesList(this);
-      int from_bitField0_ = bitField0_;
-      if (((bitField0_ & 0x00000001) != 0)) {
-        value_ = java.util.Collections.unmodifiableList(value_);
-        bitField0_ = (bitField0_ & ~0x00000001);
-      }
-      result.value_ = value_;
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.example.BytesList) {
-        return mergeFrom((org.tensorflow.proto.example.BytesList)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.example.BytesList other) {
-      if (other == org.tensorflow.proto.example.BytesList.getDefaultInstance()) return this;
-      if (!other.value_.isEmpty()) {
-        if (value_.isEmpty()) {
-          value_ = other.value_;
-          bitField0_ = (bitField0_ & ~0x00000001);
-        } else {
-          ensureValueIsMutable();
-          value_.addAll(other.value_);
-        }
-        onChanged();
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.example.BytesList parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.example.BytesList) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-    private int bitField0_;
-
-    private java.util.List<com.google.protobuf.ByteString> value_ = java.util.Collections.emptyList();
-    private void ensureValueIsMutable() {
-      if (!((bitField0_ & 0x00000001) != 0)) {
-        value_ = new java.util.ArrayList<com.google.protobuf.ByteString>(value_);
-        bitField0_ |= 0x00000001;
-       }
-    }
-    /**
-     * <code>repeated bytes value = 1;</code>
-     */
-    public java.util.List<com.google.protobuf.ByteString>
-        getValueList() {
-      return ((bitField0_ & 0x00000001) != 0) ?
-               java.util.Collections.unmodifiableList(value_) : value_;
-    }
-    /**
-     * <code>repeated bytes value = 1;</code>
-     */
-    public int getValueCount() {
-      return value_.size();
-    }
-    /**
-     * <code>repeated bytes value = 1;</code>
-     */
-    public com.google.protobuf.ByteString getValue(int index) {
-      return value_.get(index);
-    }
-    /**
-     * <code>repeated bytes value = 1;</code>
-     */
-    public Builder setValue(
-        int index, com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  ensureValueIsMutable();
-      value_.set(index, value);
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>repeated bytes value = 1;</code>
-     */
-    public Builder addValue(com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  ensureValueIsMutable();
-      value_.add(value);
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>repeated bytes value = 1;</code>
-     */
-    public Builder addAllValue(
-        java.lang.Iterable<? extends com.google.protobuf.ByteString> values) {
-      ensureValueIsMutable();
-      com.google.protobuf.AbstractMessageLite.Builder.addAll(
-          values, value_);
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>repeated bytes value = 1;</code>
-     */
-    public Builder clearValue() {
-      value_ = java.util.Collections.emptyList();
-      bitField0_ = (bitField0_ & ~0x00000001);
-      onChanged();
-      return this;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.BytesList)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.BytesList)
-  private static final org.tensorflow.proto.example.BytesList DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.example.BytesList();
-  }
-
-  public static org.tensorflow.proto.example.BytesList getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<BytesList>
-      PARSER = new com.google.protobuf.AbstractParser<BytesList>() {
-    @java.lang.Override
-    public BytesList parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new BytesList(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<BytesList> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<BytesList> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.example.BytesList getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/BytesListOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/BytesListOrBuilder.java
deleted file mode 100644
index 11151e1a947..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/BytesListOrBuilder.java
+++ /dev/null
@@ -1,22 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/example/feature.proto
-
-package org.tensorflow.proto.example;
-
-public interface BytesListOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.BytesList)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <code>repeated bytes value = 1;</code>
-   */
-  java.util.List<com.google.protobuf.ByteString> getValueList();
-  /**
-   * <code>repeated bytes value = 1;</code>
-   */
-  int getValueCount();
-  /**
-   * <code>repeated bytes value = 1;</code>
-   */
-  com.google.protobuf.ByteString getValue(int index);
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/Example.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/Example.java
deleted file mode 100644
index ea9d058e425..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/Example.java
+++ /dev/null
@@ -1,600 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/example/example.proto
-
-package org.tensorflow.proto.example;
-
-/**
- * Protobuf type {@code tensorflow.Example}
- */
-public  final class Example extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.Example)
-    ExampleOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use Example.newBuilder() to construct.
-  private Example(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private Example() {
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new Example();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private Example(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 10: {
-            org.tensorflow.proto.example.Features.Builder subBuilder = null;
-            if (features_ != null) {
-              subBuilder = features_.toBuilder();
-            }
-            features_ = input.readMessage(org.tensorflow.proto.example.Features.parser(), extensionRegistry);
-            if (subBuilder != null) {
-              subBuilder.mergeFrom(features_);
-              features_ = subBuilder.buildPartial();
-            }
-
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.example.ExampleProtos.internal_static_tensorflow_Example_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.example.ExampleProtos.internal_static_tensorflow_Example_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.example.Example.class, org.tensorflow.proto.example.Example.Builder.class);
-  }
-
-  public static final int FEATURES_FIELD_NUMBER = 1;
-  private org.tensorflow.proto.example.Features features_;
-  /**
-   * <code>.tensorflow.Features features = 1;</code>
-   */
-  public boolean hasFeatures() {
-    return features_ != null;
-  }
-  /**
-   * <code>.tensorflow.Features features = 1;</code>
-   */
-  public org.tensorflow.proto.example.Features getFeatures() {
-    return features_ == null ? org.tensorflow.proto.example.Features.getDefaultInstance() : features_;
-  }
-  /**
-   * <code>.tensorflow.Features features = 1;</code>
-   */
-  public org.tensorflow.proto.example.FeaturesOrBuilder getFeaturesOrBuilder() {
-    return getFeatures();
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    if (features_ != null) {
-      output.writeMessage(1, getFeatures());
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (features_ != null) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(1, getFeatures());
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.example.Example)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.example.Example other = (org.tensorflow.proto.example.Example) obj;
-
-    if (hasFeatures() != other.hasFeatures()) return false;
-    if (hasFeatures()) {
-      if (!getFeatures()
-          .equals(other.getFeatures())) return false;
-    }
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    if (hasFeatures()) {
-      hash = (37 * hash) + FEATURES_FIELD_NUMBER;
-      hash = (53 * hash) + getFeatures().hashCode();
-    }
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.example.Example parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.example.Example parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.example.Example parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.example.Example parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.example.Example parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.example.Example parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.example.Example parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.example.Example parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.example.Example parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.example.Example parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.example.Example parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.example.Example parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.example.Example prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * Protobuf type {@code tensorflow.Example}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.Example)
-      org.tensorflow.proto.example.ExampleOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.example.ExampleProtos.internal_static_tensorflow_Example_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.example.ExampleProtos.internal_static_tensorflow_Example_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.example.Example.class, org.tensorflow.proto.example.Example.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.example.Example.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      if (featuresBuilder_ == null) {
-        features_ = null;
-      } else {
-        features_ = null;
-        featuresBuilder_ = null;
-      }
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.example.ExampleProtos.internal_static_tensorflow_Example_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.example.Example getDefaultInstanceForType() {
-      return org.tensorflow.proto.example.Example.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.example.Example build() {
-      org.tensorflow.proto.example.Example result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.example.Example buildPartial() {
-      org.tensorflow.proto.example.Example result = new org.tensorflow.proto.example.Example(this);
-      if (featuresBuilder_ == null) {
-        result.features_ = features_;
-      } else {
-        result.features_ = featuresBuilder_.build();
-      }
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.example.Example) {
-        return mergeFrom((org.tensorflow.proto.example.Example)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.example.Example other) {
-      if (other == org.tensorflow.proto.example.Example.getDefaultInstance()) return this;
-      if (other.hasFeatures()) {
-        mergeFeatures(other.getFeatures());
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.example.Example parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.example.Example) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-
-    private org.tensorflow.proto.example.Features features_;
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.example.Features, org.tensorflow.proto.example.Features.Builder, org.tensorflow.proto.example.FeaturesOrBuilder> featuresBuilder_;
-    /**
-     * <code>.tensorflow.Features features = 1;</code>
-     */
-    public boolean hasFeatures() {
-      return featuresBuilder_ != null || features_ != null;
-    }
-    /**
-     * <code>.tensorflow.Features features = 1;</code>
-     */
-    public org.tensorflow.proto.example.Features getFeatures() {
-      if (featuresBuilder_ == null) {
-        return features_ == null ? org.tensorflow.proto.example.Features.getDefaultInstance() : features_;
-      } else {
-        return featuresBuilder_.getMessage();
-      }
-    }
-    /**
-     * <code>.tensorflow.Features features = 1;</code>
-     */
-    public Builder setFeatures(org.tensorflow.proto.example.Features value) {
-      if (featuresBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        features_ = value;
-        onChanged();
-      } else {
-        featuresBuilder_.setMessage(value);
-      }
-
-      return this;
-    }
-    /**
-     * <code>.tensorflow.Features features = 1;</code>
-     */
-    public Builder setFeatures(
-        org.tensorflow.proto.example.Features.Builder builderForValue) {
-      if (featuresBuilder_ == null) {
-        features_ = builderForValue.build();
-        onChanged();
-      } else {
-        featuresBuilder_.setMessage(builderForValue.build());
-      }
-
-      return this;
-    }
-    /**
-     * <code>.tensorflow.Features features = 1;</code>
-     */
-    public Builder mergeFeatures(org.tensorflow.proto.example.Features value) {
-      if (featuresBuilder_ == null) {
-        if (features_ != null) {
-          features_ =
-            org.tensorflow.proto.example.Features.newBuilder(features_).mergeFrom(value).buildPartial();
-        } else {
-          features_ = value;
-        }
-        onChanged();
-      } else {
-        featuresBuilder_.mergeFrom(value);
-      }
-
-      return this;
-    }
-    /**
-     * <code>.tensorflow.Features features = 1;</code>
-     */
-    public Builder clearFeatures() {
-      if (featuresBuilder_ == null) {
-        features_ = null;
-        onChanged();
-      } else {
-        features_ = null;
-        featuresBuilder_ = null;
-      }
-
-      return this;
-    }
-    /**
-     * <code>.tensorflow.Features features = 1;</code>
-     */
-    public org.tensorflow.proto.example.Features.Builder getFeaturesBuilder() {
-      
-      onChanged();
-      return getFeaturesFieldBuilder().getBuilder();
-    }
-    /**
-     * <code>.tensorflow.Features features = 1;</code>
-     */
-    public org.tensorflow.proto.example.FeaturesOrBuilder getFeaturesOrBuilder() {
-      if (featuresBuilder_ != null) {
-        return featuresBuilder_.getMessageOrBuilder();
-      } else {
-        return features_ == null ?
-            org.tensorflow.proto.example.Features.getDefaultInstance() : features_;
-      }
-    }
-    /**
-     * <code>.tensorflow.Features features = 1;</code>
-     */
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.example.Features, org.tensorflow.proto.example.Features.Builder, org.tensorflow.proto.example.FeaturesOrBuilder> 
-        getFeaturesFieldBuilder() {
-      if (featuresBuilder_ == null) {
-        featuresBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-            org.tensorflow.proto.example.Features, org.tensorflow.proto.example.Features.Builder, org.tensorflow.proto.example.FeaturesOrBuilder>(
-                getFeatures(),
-                getParentForChildren(),
-                isClean());
-        features_ = null;
-      }
-      return featuresBuilder_;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.Example)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.Example)
-  private static final org.tensorflow.proto.example.Example DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.example.Example();
-  }
-
-  public static org.tensorflow.proto.example.Example getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<Example>
-      PARSER = new com.google.protobuf.AbstractParser<Example>() {
-    @java.lang.Override
-    public Example parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new Example(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<Example> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<Example> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.example.Example getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/ExampleOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/ExampleOrBuilder.java
deleted file mode 100644
index 2026bc24075..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/ExampleOrBuilder.java
+++ /dev/null
@@ -1,22 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/example/example.proto
-
-package org.tensorflow.proto.example;
-
-public interface ExampleOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.Example)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <code>.tensorflow.Features features = 1;</code>
-   */
-  boolean hasFeatures();
-  /**
-   * <code>.tensorflow.Features features = 1;</code>
-   */
-  org.tensorflow.proto.example.Features getFeatures();
-  /**
-   * <code>.tensorflow.Features features = 1;</code>
-   */
-  org.tensorflow.proto.example.FeaturesOrBuilder getFeaturesOrBuilder();
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/ExampleParserConfiguration.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/ExampleParserConfiguration.java
deleted file mode 100644
index 4829d13d8ff..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/ExampleParserConfiguration.java
+++ /dev/null
@@ -1,695 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/example/example_parser_configuration.proto
-
-package org.tensorflow.proto.example;
-
-/**
- * Protobuf type {@code tensorflow.ExampleParserConfiguration}
- */
-public  final class ExampleParserConfiguration extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.ExampleParserConfiguration)
-    ExampleParserConfigurationOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use ExampleParserConfiguration.newBuilder() to construct.
-  private ExampleParserConfiguration(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private ExampleParserConfiguration() {
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new ExampleParserConfiguration();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private ExampleParserConfiguration(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    int mutable_bitField0_ = 0;
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 10: {
-            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
-              featureMap_ = com.google.protobuf.MapField.newMapField(
-                  FeatureMapDefaultEntryHolder.defaultEntry);
-              mutable_bitField0_ |= 0x00000001;
-            }
-            com.google.protobuf.MapEntry<java.lang.String, org.tensorflow.proto.example.FeatureConfiguration>
-            featureMap__ = input.readMessage(
-                FeatureMapDefaultEntryHolder.defaultEntry.getParserForType(), extensionRegistry);
-            featureMap_.getMutableMap().put(
-                featureMap__.getKey(), featureMap__.getValue());
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.example.ExampleParserConfigurationProtos.internal_static_tensorflow_ExampleParserConfiguration_descriptor;
-  }
-
-  @SuppressWarnings({"rawtypes"})
-  @java.lang.Override
-  protected com.google.protobuf.MapField internalGetMapField(
-      int number) {
-    switch (number) {
-      case 1:
-        return internalGetFeatureMap();
-      default:
-        throw new RuntimeException(
-            "Invalid map field number: " + number);
-    }
-  }
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.example.ExampleParserConfigurationProtos.internal_static_tensorflow_ExampleParserConfiguration_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.example.ExampleParserConfiguration.class, org.tensorflow.proto.example.ExampleParserConfiguration.Builder.class);
-  }
-
-  public static final int FEATURE_MAP_FIELD_NUMBER = 1;
-  private static final class FeatureMapDefaultEntryHolder {
-    static final com.google.protobuf.MapEntry<
-        java.lang.String, org.tensorflow.proto.example.FeatureConfiguration> defaultEntry =
-            com.google.protobuf.MapEntry
-            .<java.lang.String, org.tensorflow.proto.example.FeatureConfiguration>newDefaultInstance(
-                org.tensorflow.proto.example.ExampleParserConfigurationProtos.internal_static_tensorflow_ExampleParserConfiguration_FeatureMapEntry_descriptor, 
-                com.google.protobuf.WireFormat.FieldType.STRING,
-                "",
-                com.google.protobuf.WireFormat.FieldType.MESSAGE,
-                org.tensorflow.proto.example.FeatureConfiguration.getDefaultInstance());
-  }
-  private com.google.protobuf.MapField<
-      java.lang.String, org.tensorflow.proto.example.FeatureConfiguration> featureMap_;
-  private com.google.protobuf.MapField<java.lang.String, org.tensorflow.proto.example.FeatureConfiguration>
-  internalGetFeatureMap() {
-    if (featureMap_ == null) {
-      return com.google.protobuf.MapField.emptyMapField(
-          FeatureMapDefaultEntryHolder.defaultEntry);
-    }
-    return featureMap_;
-  }
-
-  public int getFeatureMapCount() {
-    return internalGetFeatureMap().getMap().size();
-  }
-  /**
-   * <code>map&lt;string, .tensorflow.FeatureConfiguration&gt; feature_map = 1;</code>
-   */
-
-  public boolean containsFeatureMap(
-      java.lang.String key) {
-    if (key == null) { throw new java.lang.NullPointerException(); }
-    return internalGetFeatureMap().getMap().containsKey(key);
-  }
-  /**
-   * Use {@link #getFeatureMapMap()} instead.
-   */
-  @java.lang.Deprecated
-  public java.util.Map<java.lang.String, org.tensorflow.proto.example.FeatureConfiguration> getFeatureMap() {
-    return getFeatureMapMap();
-  }
-  /**
-   * <code>map&lt;string, .tensorflow.FeatureConfiguration&gt; feature_map = 1;</code>
-   */
-
-  public java.util.Map<java.lang.String, org.tensorflow.proto.example.FeatureConfiguration> getFeatureMapMap() {
-    return internalGetFeatureMap().getMap();
-  }
-  /**
-   * <code>map&lt;string, .tensorflow.FeatureConfiguration&gt; feature_map = 1;</code>
-   */
-
-  public org.tensorflow.proto.example.FeatureConfiguration getFeatureMapOrDefault(
-      java.lang.String key,
-      org.tensorflow.proto.example.FeatureConfiguration defaultValue) {
-    if (key == null) { throw new java.lang.NullPointerException(); }
-    java.util.Map<java.lang.String, org.tensorflow.proto.example.FeatureConfiguration> map =
-        internalGetFeatureMap().getMap();
-    return map.containsKey(key) ? map.get(key) : defaultValue;
-  }
-  /**
-   * <code>map&lt;string, .tensorflow.FeatureConfiguration&gt; feature_map = 1;</code>
-   */
-
-  public org.tensorflow.proto.example.FeatureConfiguration getFeatureMapOrThrow(
-      java.lang.String key) {
-    if (key == null) { throw new java.lang.NullPointerException(); }
-    java.util.Map<java.lang.String, org.tensorflow.proto.example.FeatureConfiguration> map =
-        internalGetFeatureMap().getMap();
-    if (!map.containsKey(key)) {
-      throw new java.lang.IllegalArgumentException();
-    }
-    return map.get(key);
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    com.google.protobuf.GeneratedMessageV3
-      .serializeStringMapTo(
-        output,
-        internalGetFeatureMap(),
-        FeatureMapDefaultEntryHolder.defaultEntry,
-        1);
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    for (java.util.Map.Entry<java.lang.String, org.tensorflow.proto.example.FeatureConfiguration> entry
-         : internalGetFeatureMap().getMap().entrySet()) {
-      com.google.protobuf.MapEntry<java.lang.String, org.tensorflow.proto.example.FeatureConfiguration>
-      featureMap__ = FeatureMapDefaultEntryHolder.defaultEntry.newBuilderForType()
-          .setKey(entry.getKey())
-          .setValue(entry.getValue())
-          .build();
-      size += com.google.protobuf.CodedOutputStream
-          .computeMessageSize(1, featureMap__);
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.example.ExampleParserConfiguration)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.example.ExampleParserConfiguration other = (org.tensorflow.proto.example.ExampleParserConfiguration) obj;
-
-    if (!internalGetFeatureMap().equals(
-        other.internalGetFeatureMap())) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    if (!internalGetFeatureMap().getMap().isEmpty()) {
-      hash = (37 * hash) + FEATURE_MAP_FIELD_NUMBER;
-      hash = (53 * hash) + internalGetFeatureMap().hashCode();
-    }
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.example.ExampleParserConfiguration parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.example.ExampleParserConfiguration parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.example.ExampleParserConfiguration parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.example.ExampleParserConfiguration parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.example.ExampleParserConfiguration parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.example.ExampleParserConfiguration parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.example.ExampleParserConfiguration parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.example.ExampleParserConfiguration parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.example.ExampleParserConfiguration parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.example.ExampleParserConfiguration parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.example.ExampleParserConfiguration parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.example.ExampleParserConfiguration parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.example.ExampleParserConfiguration prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * Protobuf type {@code tensorflow.ExampleParserConfiguration}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.ExampleParserConfiguration)
-      org.tensorflow.proto.example.ExampleParserConfigurationOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.example.ExampleParserConfigurationProtos.internal_static_tensorflow_ExampleParserConfiguration_descriptor;
-    }
-
-    @SuppressWarnings({"rawtypes"})
-    protected com.google.protobuf.MapField internalGetMapField(
-        int number) {
-      switch (number) {
-        case 1:
-          return internalGetFeatureMap();
-        default:
-          throw new RuntimeException(
-              "Invalid map field number: " + number);
-      }
-    }
-    @SuppressWarnings({"rawtypes"})
-    protected com.google.protobuf.MapField internalGetMutableMapField(
-        int number) {
-      switch (number) {
-        case 1:
-          return internalGetMutableFeatureMap();
-        default:
-          throw new RuntimeException(
-              "Invalid map field number: " + number);
-      }
-    }
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.example.ExampleParserConfigurationProtos.internal_static_tensorflow_ExampleParserConfiguration_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.example.ExampleParserConfiguration.class, org.tensorflow.proto.example.ExampleParserConfiguration.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.example.ExampleParserConfiguration.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      internalGetMutableFeatureMap().clear();
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.example.ExampleParserConfigurationProtos.internal_static_tensorflow_ExampleParserConfiguration_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.example.ExampleParserConfiguration getDefaultInstanceForType() {
-      return org.tensorflow.proto.example.ExampleParserConfiguration.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.example.ExampleParserConfiguration build() {
-      org.tensorflow.proto.example.ExampleParserConfiguration result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.example.ExampleParserConfiguration buildPartial() {
-      org.tensorflow.proto.example.ExampleParserConfiguration result = new org.tensorflow.proto.example.ExampleParserConfiguration(this);
-      int from_bitField0_ = bitField0_;
-      result.featureMap_ = internalGetFeatureMap();
-      result.featureMap_.makeImmutable();
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.example.ExampleParserConfiguration) {
-        return mergeFrom((org.tensorflow.proto.example.ExampleParserConfiguration)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.example.ExampleParserConfiguration other) {
-      if (other == org.tensorflow.proto.example.ExampleParserConfiguration.getDefaultInstance()) return this;
-      internalGetMutableFeatureMap().mergeFrom(
-          other.internalGetFeatureMap());
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.example.ExampleParserConfiguration parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.example.ExampleParserConfiguration) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-    private int bitField0_;
-
-    private com.google.protobuf.MapField<
-        java.lang.String, org.tensorflow.proto.example.FeatureConfiguration> featureMap_;
-    private com.google.protobuf.MapField<java.lang.String, org.tensorflow.proto.example.FeatureConfiguration>
-    internalGetFeatureMap() {
-      if (featureMap_ == null) {
-        return com.google.protobuf.MapField.emptyMapField(
-            FeatureMapDefaultEntryHolder.defaultEntry);
-      }
-      return featureMap_;
-    }
-    private com.google.protobuf.MapField<java.lang.String, org.tensorflow.proto.example.FeatureConfiguration>
-    internalGetMutableFeatureMap() {
-      onChanged();;
-      if (featureMap_ == null) {
-        featureMap_ = com.google.protobuf.MapField.newMapField(
-            FeatureMapDefaultEntryHolder.defaultEntry);
-      }
-      if (!featureMap_.isMutable()) {
-        featureMap_ = featureMap_.copy();
-      }
-      return featureMap_;
-    }
-
-    public int getFeatureMapCount() {
-      return internalGetFeatureMap().getMap().size();
-    }
-    /**
-     * <code>map&lt;string, .tensorflow.FeatureConfiguration&gt; feature_map = 1;</code>
-     */
-
-    public boolean containsFeatureMap(
-        java.lang.String key) {
-      if (key == null) { throw new java.lang.NullPointerException(); }
-      return internalGetFeatureMap().getMap().containsKey(key);
-    }
-    /**
-     * Use {@link #getFeatureMapMap()} instead.
-     */
-    @java.lang.Deprecated
-    public java.util.Map<java.lang.String, org.tensorflow.proto.example.FeatureConfiguration> getFeatureMap() {
-      return getFeatureMapMap();
-    }
-    /**
-     * <code>map&lt;string, .tensorflow.FeatureConfiguration&gt; feature_map = 1;</code>
-     */
-
-    public java.util.Map<java.lang.String, org.tensorflow.proto.example.FeatureConfiguration> getFeatureMapMap() {
-      return internalGetFeatureMap().getMap();
-    }
-    /**
-     * <code>map&lt;string, .tensorflow.FeatureConfiguration&gt; feature_map = 1;</code>
-     */
-
-    public org.tensorflow.proto.example.FeatureConfiguration getFeatureMapOrDefault(
-        java.lang.String key,
-        org.tensorflow.proto.example.FeatureConfiguration defaultValue) {
-      if (key == null) { throw new java.lang.NullPointerException(); }
-      java.util.Map<java.lang.String, org.tensorflow.proto.example.FeatureConfiguration> map =
-          internalGetFeatureMap().getMap();
-      return map.containsKey(key) ? map.get(key) : defaultValue;
-    }
-    /**
-     * <code>map&lt;string, .tensorflow.FeatureConfiguration&gt; feature_map = 1;</code>
-     */
-
-    public org.tensorflow.proto.example.FeatureConfiguration getFeatureMapOrThrow(
-        java.lang.String key) {
-      if (key == null) { throw new java.lang.NullPointerException(); }
-      java.util.Map<java.lang.String, org.tensorflow.proto.example.FeatureConfiguration> map =
-          internalGetFeatureMap().getMap();
-      if (!map.containsKey(key)) {
-        throw new java.lang.IllegalArgumentException();
-      }
-      return map.get(key);
-    }
-
-    public Builder clearFeatureMap() {
-      internalGetMutableFeatureMap().getMutableMap()
-          .clear();
-      return this;
-    }
-    /**
-     * <code>map&lt;string, .tensorflow.FeatureConfiguration&gt; feature_map = 1;</code>
-     */
-
-    public Builder removeFeatureMap(
-        java.lang.String key) {
-      if (key == null) { throw new java.lang.NullPointerException(); }
-      internalGetMutableFeatureMap().getMutableMap()
-          .remove(key);
-      return this;
-    }
-    /**
-     * Use alternate mutation accessors instead.
-     */
-    @java.lang.Deprecated
-    public java.util.Map<java.lang.String, org.tensorflow.proto.example.FeatureConfiguration>
-    getMutableFeatureMap() {
-      return internalGetMutableFeatureMap().getMutableMap();
-    }
-    /**
-     * <code>map&lt;string, .tensorflow.FeatureConfiguration&gt; feature_map = 1;</code>
-     */
-    public Builder putFeatureMap(
-        java.lang.String key,
-        org.tensorflow.proto.example.FeatureConfiguration value) {
-      if (key == null) { throw new java.lang.NullPointerException(); }
-      if (value == null) { throw new java.lang.NullPointerException(); }
-      internalGetMutableFeatureMap().getMutableMap()
-          .put(key, value);
-      return this;
-    }
-    /**
-     * <code>map&lt;string, .tensorflow.FeatureConfiguration&gt; feature_map = 1;</code>
-     */
-
-    public Builder putAllFeatureMap(
-        java.util.Map<java.lang.String, org.tensorflow.proto.example.FeatureConfiguration> values) {
-      internalGetMutableFeatureMap().getMutableMap()
-          .putAll(values);
-      return this;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.ExampleParserConfiguration)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.ExampleParserConfiguration)
-  private static final org.tensorflow.proto.example.ExampleParserConfiguration DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.example.ExampleParserConfiguration();
-  }
-
-  public static org.tensorflow.proto.example.ExampleParserConfiguration getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<ExampleParserConfiguration>
-      PARSER = new com.google.protobuf.AbstractParser<ExampleParserConfiguration>() {
-    @java.lang.Override
-    public ExampleParserConfiguration parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new ExampleParserConfiguration(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<ExampleParserConfiguration> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<ExampleParserConfiguration> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.example.ExampleParserConfiguration getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/ExampleParserConfigurationOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/ExampleParserConfigurationOrBuilder.java
deleted file mode 100644
index 33df2a31a5b..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/ExampleParserConfigurationOrBuilder.java
+++ /dev/null
@@ -1,43 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/example/example_parser_configuration.proto
-
-package org.tensorflow.proto.example;
-
-public interface ExampleParserConfigurationOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.ExampleParserConfiguration)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <code>map&lt;string, .tensorflow.FeatureConfiguration&gt; feature_map = 1;</code>
-   */
-  int getFeatureMapCount();
-  /**
-   * <code>map&lt;string, .tensorflow.FeatureConfiguration&gt; feature_map = 1;</code>
-   */
-  boolean containsFeatureMap(
-      java.lang.String key);
-  /**
-   * Use {@link #getFeatureMapMap()} instead.
-   */
-  @java.lang.Deprecated
-  java.util.Map<java.lang.String, org.tensorflow.proto.example.FeatureConfiguration>
-  getFeatureMap();
-  /**
-   * <code>map&lt;string, .tensorflow.FeatureConfiguration&gt; feature_map = 1;</code>
-   */
-  java.util.Map<java.lang.String, org.tensorflow.proto.example.FeatureConfiguration>
-  getFeatureMapMap();
-  /**
-   * <code>map&lt;string, .tensorflow.FeatureConfiguration&gt; feature_map = 1;</code>
-   */
-
-  org.tensorflow.proto.example.FeatureConfiguration getFeatureMapOrDefault(
-      java.lang.String key,
-      org.tensorflow.proto.example.FeatureConfiguration defaultValue);
-  /**
-   * <code>map&lt;string, .tensorflow.FeatureConfiguration&gt; feature_map = 1;</code>
-   */
-
-  org.tensorflow.proto.example.FeatureConfiguration getFeatureMapOrThrow(
-      java.lang.String key);
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/ExampleParserConfigurationProtos.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/ExampleParserConfigurationProtos.java
deleted file mode 100644
index 9bc45ea35c4..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/ExampleParserConfigurationProtos.java
+++ /dev/null
@@ -1,123 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/example/example_parser_configuration.proto
-
-package org.tensorflow.proto.example;
-
-public final class ExampleParserConfigurationProtos {
-  private ExampleParserConfigurationProtos() {}
-  public static void registerAllExtensions(
-      com.google.protobuf.ExtensionRegistryLite registry) {
-  }
-
-  public static void registerAllExtensions(
-      com.google.protobuf.ExtensionRegistry registry) {
-    registerAllExtensions(
-        (com.google.protobuf.ExtensionRegistryLite) registry);
-  }
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_VarLenFeatureProto_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_VarLenFeatureProto_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_FixedLenFeatureProto_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_FixedLenFeatureProto_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_FeatureConfiguration_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_FeatureConfiguration_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_ExampleParserConfiguration_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_ExampleParserConfiguration_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_ExampleParserConfiguration_FeatureMapEntry_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_ExampleParserConfiguration_FeatureMapEntry_fieldAccessorTable;
-
-  public static com.google.protobuf.Descriptors.FileDescriptor
-      getDescriptor() {
-    return descriptor;
-  }
-  private static  com.google.protobuf.Descriptors.FileDescriptor
-      descriptor;
-  static {
-    java.lang.String[] descriptorData = {
-      "\n:tensorflow/core/example/example_parser" +
-      "_configuration.proto\022\ntensorflow\032&tensor" +
-      "flow/core/framework/tensor.proto\032,tensor" +
-      "flow/core/framework/tensor_shape.proto\032%" +
-      "tensorflow/core/framework/types.proto\"\243\001" +
-      "\n\022VarLenFeatureProto\022#\n\005dtype\030\001 \001(\0162\024.te" +
-      "nsorflow.DataType\022!\n\031values_output_tenso" +
-      "r_name\030\002 \001(\t\022\"\n\032indices_output_tensor_na" +
-      "me\030\003 \001(\t\022!\n\031shapes_output_tensor_name\030\004 " +
-      "\001(\t\"\273\001\n\024FixedLenFeatureProto\022#\n\005dtype\030\001 " +
-      "\001(\0162\024.tensorflow.DataType\022+\n\005shape\030\002 \001(\013" +
-      "2\034.tensorflow.TensorShapeProto\022.\n\rdefaul" +
-      "t_value\030\003 \001(\0132\027.tensorflow.TensorProto\022!" +
-      "\n\031values_output_tensor_name\030\004 \001(\t\"\232\001\n\024Fe" +
-      "atureConfiguration\022=\n\021fixed_len_feature\030" +
-      "\001 \001(\0132 .tensorflow.FixedLenFeatureProtoH" +
-      "\000\0229\n\017var_len_feature\030\002 \001(\0132\036.tensorflow." +
-      "VarLenFeatureProtoH\000B\010\n\006config\"\276\001\n\032Examp" +
-      "leParserConfiguration\022K\n\013feature_map\030\001 \003" +
-      "(\01326.tensorflow.ExampleParserConfigurati" +
-      "on.FeatureMapEntry\032S\n\017FeatureMapEntry\022\013\n" +
-      "\003key\030\001 \001(\t\022/\n\005value\030\002 \001(\0132 .tensorflow.F" +
-      "eatureConfiguration:\0028\001B\250\001\n\034org.tensorfl" +
-      "ow.proto.exampleB ExampleParserConfigura" +
-      "tionProtosP\001Zagithub.com/tensorflow/tens" +
-      "orflow/tensorflow/go/core/example/exampl" +
-      "e_parser_configuration_go_proto\370\001\001b\006prot" +
-      "o3"
-    };
-    descriptor = com.google.protobuf.Descriptors.FileDescriptor
-      .internalBuildGeneratedFileFrom(descriptorData,
-        new com.google.protobuf.Descriptors.FileDescriptor[] {
-          org.tensorflow.proto.framework.TensorProtos.getDescriptor(),
-          org.tensorflow.proto.framework.TensorShapeProtos.getDescriptor(),
-          org.tensorflow.proto.framework.TypesProtos.getDescriptor(),
-        });
-    internal_static_tensorflow_VarLenFeatureProto_descriptor =
-      getDescriptor().getMessageTypes().get(0);
-    internal_static_tensorflow_VarLenFeatureProto_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_VarLenFeatureProto_descriptor,
-        new java.lang.String[] { "Dtype", "ValuesOutputTensorName", "IndicesOutputTensorName", "ShapesOutputTensorName", });
-    internal_static_tensorflow_FixedLenFeatureProto_descriptor =
-      getDescriptor().getMessageTypes().get(1);
-    internal_static_tensorflow_FixedLenFeatureProto_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_FixedLenFeatureProto_descriptor,
-        new java.lang.String[] { "Dtype", "Shape", "DefaultValue", "ValuesOutputTensorName", });
-    internal_static_tensorflow_FeatureConfiguration_descriptor =
-      getDescriptor().getMessageTypes().get(2);
-    internal_static_tensorflow_FeatureConfiguration_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_FeatureConfiguration_descriptor,
-        new java.lang.String[] { "FixedLenFeature", "VarLenFeature", "Config", });
-    internal_static_tensorflow_ExampleParserConfiguration_descriptor =
-      getDescriptor().getMessageTypes().get(3);
-    internal_static_tensorflow_ExampleParserConfiguration_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_ExampleParserConfiguration_descriptor,
-        new java.lang.String[] { "FeatureMap", });
-    internal_static_tensorflow_ExampleParserConfiguration_FeatureMapEntry_descriptor =
-      internal_static_tensorflow_ExampleParserConfiguration_descriptor.getNestedTypes().get(0);
-    internal_static_tensorflow_ExampleParserConfiguration_FeatureMapEntry_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_ExampleParserConfiguration_FeatureMapEntry_descriptor,
-        new java.lang.String[] { "Key", "Value", });
-    org.tensorflow.proto.framework.TensorProtos.getDescriptor();
-    org.tensorflow.proto.framework.TensorShapeProtos.getDescriptor();
-    org.tensorflow.proto.framework.TypesProtos.getDescriptor();
-  }
-
-  // @@protoc_insertion_point(outer_class_scope)
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/ExampleProtos.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/ExampleProtos.java
deleted file mode 100644
index d5494a720c5..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/ExampleProtos.java
+++ /dev/null
@@ -1,68 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/example/example.proto
-
-package org.tensorflow.proto.example;
-
-public final class ExampleProtos {
-  private ExampleProtos() {}
-  public static void registerAllExtensions(
-      com.google.protobuf.ExtensionRegistryLite registry) {
-  }
-
-  public static void registerAllExtensions(
-      com.google.protobuf.ExtensionRegistry registry) {
-    registerAllExtensions(
-        (com.google.protobuf.ExtensionRegistryLite) registry);
-  }
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_Example_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_Example_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_SequenceExample_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_SequenceExample_fieldAccessorTable;
-
-  public static com.google.protobuf.Descriptors.FileDescriptor
-      getDescriptor() {
-    return descriptor;
-  }
-  private static  com.google.protobuf.Descriptors.FileDescriptor
-      descriptor;
-  static {
-    java.lang.String[] descriptorData = {
-      "\n%tensorflow/core/example/example.proto\022" +
-      "\ntensorflow\032%tensorflow/core/example/fea" +
-      "ture.proto\"1\n\007Example\022&\n\010features\030\001 \001(\0132" +
-      "\024.tensorflow.Features\"i\n\017SequenceExample" +
-      "\022%\n\007context\030\001 \001(\0132\024.tensorflow.Features\022" +
-      "/\n\rfeature_lists\030\002 \001(\0132\030.tensorflow.Feat" +
-      "ureListsB\207\001\n\034org.tensorflow.proto.exampl" +
-      "eB\rExampleProtosP\001ZSgithub.com/tensorflo" +
-      "w/tensorflow/tensorflow/go/core/example/" +
-      "example_protos_go_proto\370\001\001b\006proto3"
-    };
-    descriptor = com.google.protobuf.Descriptors.FileDescriptor
-      .internalBuildGeneratedFileFrom(descriptorData,
-        new com.google.protobuf.Descriptors.FileDescriptor[] {
-          org.tensorflow.proto.example.FeatureProtos.getDescriptor(),
-        });
-    internal_static_tensorflow_Example_descriptor =
-      getDescriptor().getMessageTypes().get(0);
-    internal_static_tensorflow_Example_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_Example_descriptor,
-        new java.lang.String[] { "Features", });
-    internal_static_tensorflow_SequenceExample_descriptor =
-      getDescriptor().getMessageTypes().get(1);
-    internal_static_tensorflow_SequenceExample_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_SequenceExample_descriptor,
-        new java.lang.String[] { "Context", "FeatureLists", });
-    org.tensorflow.proto.example.FeatureProtos.getDescriptor();
-  }
-
-  // @@protoc_insertion_point(outer_class_scope)
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/Feature.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/Feature.java
deleted file mode 100644
index a4aa0733f54..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/Feature.java
+++ /dev/null
@@ -1,1105 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/example/feature.proto
-
-package org.tensorflow.proto.example;
-
-/**
- * <pre>
- * Containers for non-sequential data.
- * </pre>
- *
- * Protobuf type {@code tensorflow.Feature}
- */
-public  final class Feature extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.Feature)
-    FeatureOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use Feature.newBuilder() to construct.
-  private Feature(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private Feature() {
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new Feature();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private Feature(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 10: {
-            org.tensorflow.proto.example.BytesList.Builder subBuilder = null;
-            if (kindCase_ == 1) {
-              subBuilder = ((org.tensorflow.proto.example.BytesList) kind_).toBuilder();
-            }
-            kind_ =
-                input.readMessage(org.tensorflow.proto.example.BytesList.parser(), extensionRegistry);
-            if (subBuilder != null) {
-              subBuilder.mergeFrom((org.tensorflow.proto.example.BytesList) kind_);
-              kind_ = subBuilder.buildPartial();
-            }
-            kindCase_ = 1;
-            break;
-          }
-          case 18: {
-            org.tensorflow.proto.example.FloatList.Builder subBuilder = null;
-            if (kindCase_ == 2) {
-              subBuilder = ((org.tensorflow.proto.example.FloatList) kind_).toBuilder();
-            }
-            kind_ =
-                input.readMessage(org.tensorflow.proto.example.FloatList.parser(), extensionRegistry);
-            if (subBuilder != null) {
-              subBuilder.mergeFrom((org.tensorflow.proto.example.FloatList) kind_);
-              kind_ = subBuilder.buildPartial();
-            }
-            kindCase_ = 2;
-            break;
-          }
-          case 26: {
-            org.tensorflow.proto.example.Int64List.Builder subBuilder = null;
-            if (kindCase_ == 3) {
-              subBuilder = ((org.tensorflow.proto.example.Int64List) kind_).toBuilder();
-            }
-            kind_ =
-                input.readMessage(org.tensorflow.proto.example.Int64List.parser(), extensionRegistry);
-            if (subBuilder != null) {
-              subBuilder.mergeFrom((org.tensorflow.proto.example.Int64List) kind_);
-              kind_ = subBuilder.buildPartial();
-            }
-            kindCase_ = 3;
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.example.FeatureProtos.internal_static_tensorflow_Feature_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.example.FeatureProtos.internal_static_tensorflow_Feature_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.example.Feature.class, org.tensorflow.proto.example.Feature.Builder.class);
-  }
-
-  private int kindCase_ = 0;
-  private java.lang.Object kind_;
-  public enum KindCase
-      implements com.google.protobuf.Internal.EnumLite {
-    BYTES_LIST(1),
-    FLOAT_LIST(2),
-    INT64_LIST(3),
-    KIND_NOT_SET(0);
-    private final int value;
-    private KindCase(int value) {
-      this.value = value;
-    }
-    /**
-     * @deprecated Use {@link #forNumber(int)} instead.
-     */
-    @java.lang.Deprecated
-    public static KindCase valueOf(int value) {
-      return forNumber(value);
-    }
-
-    public static KindCase forNumber(int value) {
-      switch (value) {
-        case 1: return BYTES_LIST;
-        case 2: return FLOAT_LIST;
-        case 3: return INT64_LIST;
-        case 0: return KIND_NOT_SET;
-        default: return null;
-      }
-    }
-    public int getNumber() {
-      return this.value;
-    }
-  };
-
-  public KindCase
-  getKindCase() {
-    return KindCase.forNumber(
-        kindCase_);
-  }
-
-  public static final int BYTES_LIST_FIELD_NUMBER = 1;
-  /**
-   * <code>.tensorflow.BytesList bytes_list = 1;</code>
-   */
-  public boolean hasBytesList() {
-    return kindCase_ == 1;
-  }
-  /**
-   * <code>.tensorflow.BytesList bytes_list = 1;</code>
-   */
-  public org.tensorflow.proto.example.BytesList getBytesList() {
-    if (kindCase_ == 1) {
-       return (org.tensorflow.proto.example.BytesList) kind_;
-    }
-    return org.tensorflow.proto.example.BytesList.getDefaultInstance();
-  }
-  /**
-   * <code>.tensorflow.BytesList bytes_list = 1;</code>
-   */
-  public org.tensorflow.proto.example.BytesListOrBuilder getBytesListOrBuilder() {
-    if (kindCase_ == 1) {
-       return (org.tensorflow.proto.example.BytesList) kind_;
-    }
-    return org.tensorflow.proto.example.BytesList.getDefaultInstance();
-  }
-
-  public static final int FLOAT_LIST_FIELD_NUMBER = 2;
-  /**
-   * <code>.tensorflow.FloatList float_list = 2;</code>
-   */
-  public boolean hasFloatList() {
-    return kindCase_ == 2;
-  }
-  /**
-   * <code>.tensorflow.FloatList float_list = 2;</code>
-   */
-  public org.tensorflow.proto.example.FloatList getFloatList() {
-    if (kindCase_ == 2) {
-       return (org.tensorflow.proto.example.FloatList) kind_;
-    }
-    return org.tensorflow.proto.example.FloatList.getDefaultInstance();
-  }
-  /**
-   * <code>.tensorflow.FloatList float_list = 2;</code>
-   */
-  public org.tensorflow.proto.example.FloatListOrBuilder getFloatListOrBuilder() {
-    if (kindCase_ == 2) {
-       return (org.tensorflow.proto.example.FloatList) kind_;
-    }
-    return org.tensorflow.proto.example.FloatList.getDefaultInstance();
-  }
-
-  public static final int INT64_LIST_FIELD_NUMBER = 3;
-  /**
-   * <code>.tensorflow.Int64List int64_list = 3;</code>
-   */
-  public boolean hasInt64List() {
-    return kindCase_ == 3;
-  }
-  /**
-   * <code>.tensorflow.Int64List int64_list = 3;</code>
-   */
-  public org.tensorflow.proto.example.Int64List getInt64List() {
-    if (kindCase_ == 3) {
-       return (org.tensorflow.proto.example.Int64List) kind_;
-    }
-    return org.tensorflow.proto.example.Int64List.getDefaultInstance();
-  }
-  /**
-   * <code>.tensorflow.Int64List int64_list = 3;</code>
-   */
-  public org.tensorflow.proto.example.Int64ListOrBuilder getInt64ListOrBuilder() {
-    if (kindCase_ == 3) {
-       return (org.tensorflow.proto.example.Int64List) kind_;
-    }
-    return org.tensorflow.proto.example.Int64List.getDefaultInstance();
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    if (kindCase_ == 1) {
-      output.writeMessage(1, (org.tensorflow.proto.example.BytesList) kind_);
-    }
-    if (kindCase_ == 2) {
-      output.writeMessage(2, (org.tensorflow.proto.example.FloatList) kind_);
-    }
-    if (kindCase_ == 3) {
-      output.writeMessage(3, (org.tensorflow.proto.example.Int64List) kind_);
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (kindCase_ == 1) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(1, (org.tensorflow.proto.example.BytesList) kind_);
-    }
-    if (kindCase_ == 2) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(2, (org.tensorflow.proto.example.FloatList) kind_);
-    }
-    if (kindCase_ == 3) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(3, (org.tensorflow.proto.example.Int64List) kind_);
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.example.Feature)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.example.Feature other = (org.tensorflow.proto.example.Feature) obj;
-
-    if (!getKindCase().equals(other.getKindCase())) return false;
-    switch (kindCase_) {
-      case 1:
-        if (!getBytesList()
-            .equals(other.getBytesList())) return false;
-        break;
-      case 2:
-        if (!getFloatList()
-            .equals(other.getFloatList())) return false;
-        break;
-      case 3:
-        if (!getInt64List()
-            .equals(other.getInt64List())) return false;
-        break;
-      case 0:
-      default:
-    }
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    switch (kindCase_) {
-      case 1:
-        hash = (37 * hash) + BYTES_LIST_FIELD_NUMBER;
-        hash = (53 * hash) + getBytesList().hashCode();
-        break;
-      case 2:
-        hash = (37 * hash) + FLOAT_LIST_FIELD_NUMBER;
-        hash = (53 * hash) + getFloatList().hashCode();
-        break;
-      case 3:
-        hash = (37 * hash) + INT64_LIST_FIELD_NUMBER;
-        hash = (53 * hash) + getInt64List().hashCode();
-        break;
-      case 0:
-      default:
-    }
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.example.Feature parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.example.Feature parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.example.Feature parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.example.Feature parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.example.Feature parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.example.Feature parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.example.Feature parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.example.Feature parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.example.Feature parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.example.Feature parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.example.Feature parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.example.Feature parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.example.Feature prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * <pre>
-   * Containers for non-sequential data.
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.Feature}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.Feature)
-      org.tensorflow.proto.example.FeatureOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.example.FeatureProtos.internal_static_tensorflow_Feature_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.example.FeatureProtos.internal_static_tensorflow_Feature_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.example.Feature.class, org.tensorflow.proto.example.Feature.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.example.Feature.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      kindCase_ = 0;
-      kind_ = null;
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.example.FeatureProtos.internal_static_tensorflow_Feature_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.example.Feature getDefaultInstanceForType() {
-      return org.tensorflow.proto.example.Feature.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.example.Feature build() {
-      org.tensorflow.proto.example.Feature result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.example.Feature buildPartial() {
-      org.tensorflow.proto.example.Feature result = new org.tensorflow.proto.example.Feature(this);
-      if (kindCase_ == 1) {
-        if (bytesListBuilder_ == null) {
-          result.kind_ = kind_;
-        } else {
-          result.kind_ = bytesListBuilder_.build();
-        }
-      }
-      if (kindCase_ == 2) {
-        if (floatListBuilder_ == null) {
-          result.kind_ = kind_;
-        } else {
-          result.kind_ = floatListBuilder_.build();
-        }
-      }
-      if (kindCase_ == 3) {
-        if (int64ListBuilder_ == null) {
-          result.kind_ = kind_;
-        } else {
-          result.kind_ = int64ListBuilder_.build();
-        }
-      }
-      result.kindCase_ = kindCase_;
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.example.Feature) {
-        return mergeFrom((org.tensorflow.proto.example.Feature)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.example.Feature other) {
-      if (other == org.tensorflow.proto.example.Feature.getDefaultInstance()) return this;
-      switch (other.getKindCase()) {
-        case BYTES_LIST: {
-          mergeBytesList(other.getBytesList());
-          break;
-        }
-        case FLOAT_LIST: {
-          mergeFloatList(other.getFloatList());
-          break;
-        }
-        case INT64_LIST: {
-          mergeInt64List(other.getInt64List());
-          break;
-        }
-        case KIND_NOT_SET: {
-          break;
-        }
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.example.Feature parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.example.Feature) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-    private int kindCase_ = 0;
-    private java.lang.Object kind_;
-    public KindCase
-        getKindCase() {
-      return KindCase.forNumber(
-          kindCase_);
-    }
-
-    public Builder clearKind() {
-      kindCase_ = 0;
-      kind_ = null;
-      onChanged();
-      return this;
-    }
-
-
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.example.BytesList, org.tensorflow.proto.example.BytesList.Builder, org.tensorflow.proto.example.BytesListOrBuilder> bytesListBuilder_;
-    /**
-     * <code>.tensorflow.BytesList bytes_list = 1;</code>
-     */
-    public boolean hasBytesList() {
-      return kindCase_ == 1;
-    }
-    /**
-     * <code>.tensorflow.BytesList bytes_list = 1;</code>
-     */
-    public org.tensorflow.proto.example.BytesList getBytesList() {
-      if (bytesListBuilder_ == null) {
-        if (kindCase_ == 1) {
-          return (org.tensorflow.proto.example.BytesList) kind_;
-        }
-        return org.tensorflow.proto.example.BytesList.getDefaultInstance();
-      } else {
-        if (kindCase_ == 1) {
-          return bytesListBuilder_.getMessage();
-        }
-        return org.tensorflow.proto.example.BytesList.getDefaultInstance();
-      }
-    }
-    /**
-     * <code>.tensorflow.BytesList bytes_list = 1;</code>
-     */
-    public Builder setBytesList(org.tensorflow.proto.example.BytesList value) {
-      if (bytesListBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        kind_ = value;
-        onChanged();
-      } else {
-        bytesListBuilder_.setMessage(value);
-      }
-      kindCase_ = 1;
-      return this;
-    }
-    /**
-     * <code>.tensorflow.BytesList bytes_list = 1;</code>
-     */
-    public Builder setBytesList(
-        org.tensorflow.proto.example.BytesList.Builder builderForValue) {
-      if (bytesListBuilder_ == null) {
-        kind_ = builderForValue.build();
-        onChanged();
-      } else {
-        bytesListBuilder_.setMessage(builderForValue.build());
-      }
-      kindCase_ = 1;
-      return this;
-    }
-    /**
-     * <code>.tensorflow.BytesList bytes_list = 1;</code>
-     */
-    public Builder mergeBytesList(org.tensorflow.proto.example.BytesList value) {
-      if (bytesListBuilder_ == null) {
-        if (kindCase_ == 1 &&
-            kind_ != org.tensorflow.proto.example.BytesList.getDefaultInstance()) {
-          kind_ = org.tensorflow.proto.example.BytesList.newBuilder((org.tensorflow.proto.example.BytesList) kind_)
-              .mergeFrom(value).buildPartial();
-        } else {
-          kind_ = value;
-        }
-        onChanged();
-      } else {
-        if (kindCase_ == 1) {
-          bytesListBuilder_.mergeFrom(value);
-        }
-        bytesListBuilder_.setMessage(value);
-      }
-      kindCase_ = 1;
-      return this;
-    }
-    /**
-     * <code>.tensorflow.BytesList bytes_list = 1;</code>
-     */
-    public Builder clearBytesList() {
-      if (bytesListBuilder_ == null) {
-        if (kindCase_ == 1) {
-          kindCase_ = 0;
-          kind_ = null;
-          onChanged();
-        }
-      } else {
-        if (kindCase_ == 1) {
-          kindCase_ = 0;
-          kind_ = null;
-        }
-        bytesListBuilder_.clear();
-      }
-      return this;
-    }
-    /**
-     * <code>.tensorflow.BytesList bytes_list = 1;</code>
-     */
-    public org.tensorflow.proto.example.BytesList.Builder getBytesListBuilder() {
-      return getBytesListFieldBuilder().getBuilder();
-    }
-    /**
-     * <code>.tensorflow.BytesList bytes_list = 1;</code>
-     */
-    public org.tensorflow.proto.example.BytesListOrBuilder getBytesListOrBuilder() {
-      if ((kindCase_ == 1) && (bytesListBuilder_ != null)) {
-        return bytesListBuilder_.getMessageOrBuilder();
-      } else {
-        if (kindCase_ == 1) {
-          return (org.tensorflow.proto.example.BytesList) kind_;
-        }
-        return org.tensorflow.proto.example.BytesList.getDefaultInstance();
-      }
-    }
-    /**
-     * <code>.tensorflow.BytesList bytes_list = 1;</code>
-     */
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.example.BytesList, org.tensorflow.proto.example.BytesList.Builder, org.tensorflow.proto.example.BytesListOrBuilder> 
-        getBytesListFieldBuilder() {
-      if (bytesListBuilder_ == null) {
-        if (!(kindCase_ == 1)) {
-          kind_ = org.tensorflow.proto.example.BytesList.getDefaultInstance();
-        }
-        bytesListBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-            org.tensorflow.proto.example.BytesList, org.tensorflow.proto.example.BytesList.Builder, org.tensorflow.proto.example.BytesListOrBuilder>(
-                (org.tensorflow.proto.example.BytesList) kind_,
-                getParentForChildren(),
-                isClean());
-        kind_ = null;
-      }
-      kindCase_ = 1;
-      onChanged();;
-      return bytesListBuilder_;
-    }
-
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.example.FloatList, org.tensorflow.proto.example.FloatList.Builder, org.tensorflow.proto.example.FloatListOrBuilder> floatListBuilder_;
-    /**
-     * <code>.tensorflow.FloatList float_list = 2;</code>
-     */
-    public boolean hasFloatList() {
-      return kindCase_ == 2;
-    }
-    /**
-     * <code>.tensorflow.FloatList float_list = 2;</code>
-     */
-    public org.tensorflow.proto.example.FloatList getFloatList() {
-      if (floatListBuilder_ == null) {
-        if (kindCase_ == 2) {
-          return (org.tensorflow.proto.example.FloatList) kind_;
-        }
-        return org.tensorflow.proto.example.FloatList.getDefaultInstance();
-      } else {
-        if (kindCase_ == 2) {
-          return floatListBuilder_.getMessage();
-        }
-        return org.tensorflow.proto.example.FloatList.getDefaultInstance();
-      }
-    }
-    /**
-     * <code>.tensorflow.FloatList float_list = 2;</code>
-     */
-    public Builder setFloatList(org.tensorflow.proto.example.FloatList value) {
-      if (floatListBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        kind_ = value;
-        onChanged();
-      } else {
-        floatListBuilder_.setMessage(value);
-      }
-      kindCase_ = 2;
-      return this;
-    }
-    /**
-     * <code>.tensorflow.FloatList float_list = 2;</code>
-     */
-    public Builder setFloatList(
-        org.tensorflow.proto.example.FloatList.Builder builderForValue) {
-      if (floatListBuilder_ == null) {
-        kind_ = builderForValue.build();
-        onChanged();
-      } else {
-        floatListBuilder_.setMessage(builderForValue.build());
-      }
-      kindCase_ = 2;
-      return this;
-    }
-    /**
-     * <code>.tensorflow.FloatList float_list = 2;</code>
-     */
-    public Builder mergeFloatList(org.tensorflow.proto.example.FloatList value) {
-      if (floatListBuilder_ == null) {
-        if (kindCase_ == 2 &&
-            kind_ != org.tensorflow.proto.example.FloatList.getDefaultInstance()) {
-          kind_ = org.tensorflow.proto.example.FloatList.newBuilder((org.tensorflow.proto.example.FloatList) kind_)
-              .mergeFrom(value).buildPartial();
-        } else {
-          kind_ = value;
-        }
-        onChanged();
-      } else {
-        if (kindCase_ == 2) {
-          floatListBuilder_.mergeFrom(value);
-        }
-        floatListBuilder_.setMessage(value);
-      }
-      kindCase_ = 2;
-      return this;
-    }
-    /**
-     * <code>.tensorflow.FloatList float_list = 2;</code>
-     */
-    public Builder clearFloatList() {
-      if (floatListBuilder_ == null) {
-        if (kindCase_ == 2) {
-          kindCase_ = 0;
-          kind_ = null;
-          onChanged();
-        }
-      } else {
-        if (kindCase_ == 2) {
-          kindCase_ = 0;
-          kind_ = null;
-        }
-        floatListBuilder_.clear();
-      }
-      return this;
-    }
-    /**
-     * <code>.tensorflow.FloatList float_list = 2;</code>
-     */
-    public org.tensorflow.proto.example.FloatList.Builder getFloatListBuilder() {
-      return getFloatListFieldBuilder().getBuilder();
-    }
-    /**
-     * <code>.tensorflow.FloatList float_list = 2;</code>
-     */
-    public org.tensorflow.proto.example.FloatListOrBuilder getFloatListOrBuilder() {
-      if ((kindCase_ == 2) && (floatListBuilder_ != null)) {
-        return floatListBuilder_.getMessageOrBuilder();
-      } else {
-        if (kindCase_ == 2) {
-          return (org.tensorflow.proto.example.FloatList) kind_;
-        }
-        return org.tensorflow.proto.example.FloatList.getDefaultInstance();
-      }
-    }
-    /**
-     * <code>.tensorflow.FloatList float_list = 2;</code>
-     */
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.example.FloatList, org.tensorflow.proto.example.FloatList.Builder, org.tensorflow.proto.example.FloatListOrBuilder> 
-        getFloatListFieldBuilder() {
-      if (floatListBuilder_ == null) {
-        if (!(kindCase_ == 2)) {
-          kind_ = org.tensorflow.proto.example.FloatList.getDefaultInstance();
-        }
-        floatListBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-            org.tensorflow.proto.example.FloatList, org.tensorflow.proto.example.FloatList.Builder, org.tensorflow.proto.example.FloatListOrBuilder>(
-                (org.tensorflow.proto.example.FloatList) kind_,
-                getParentForChildren(),
-                isClean());
-        kind_ = null;
-      }
-      kindCase_ = 2;
-      onChanged();;
-      return floatListBuilder_;
-    }
-
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.example.Int64List, org.tensorflow.proto.example.Int64List.Builder, org.tensorflow.proto.example.Int64ListOrBuilder> int64ListBuilder_;
-    /**
-     * <code>.tensorflow.Int64List int64_list = 3;</code>
-     */
-    public boolean hasInt64List() {
-      return kindCase_ == 3;
-    }
-    /**
-     * <code>.tensorflow.Int64List int64_list = 3;</code>
-     */
-    public org.tensorflow.proto.example.Int64List getInt64List() {
-      if (int64ListBuilder_ == null) {
-        if (kindCase_ == 3) {
-          return (org.tensorflow.proto.example.Int64List) kind_;
-        }
-        return org.tensorflow.proto.example.Int64List.getDefaultInstance();
-      } else {
-        if (kindCase_ == 3) {
-          return int64ListBuilder_.getMessage();
-        }
-        return org.tensorflow.proto.example.Int64List.getDefaultInstance();
-      }
-    }
-    /**
-     * <code>.tensorflow.Int64List int64_list = 3;</code>
-     */
-    public Builder setInt64List(org.tensorflow.proto.example.Int64List value) {
-      if (int64ListBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        kind_ = value;
-        onChanged();
-      } else {
-        int64ListBuilder_.setMessage(value);
-      }
-      kindCase_ = 3;
-      return this;
-    }
-    /**
-     * <code>.tensorflow.Int64List int64_list = 3;</code>
-     */
-    public Builder setInt64List(
-        org.tensorflow.proto.example.Int64List.Builder builderForValue) {
-      if (int64ListBuilder_ == null) {
-        kind_ = builderForValue.build();
-        onChanged();
-      } else {
-        int64ListBuilder_.setMessage(builderForValue.build());
-      }
-      kindCase_ = 3;
-      return this;
-    }
-    /**
-     * <code>.tensorflow.Int64List int64_list = 3;</code>
-     */
-    public Builder mergeInt64List(org.tensorflow.proto.example.Int64List value) {
-      if (int64ListBuilder_ == null) {
-        if (kindCase_ == 3 &&
-            kind_ != org.tensorflow.proto.example.Int64List.getDefaultInstance()) {
-          kind_ = org.tensorflow.proto.example.Int64List.newBuilder((org.tensorflow.proto.example.Int64List) kind_)
-              .mergeFrom(value).buildPartial();
-        } else {
-          kind_ = value;
-        }
-        onChanged();
-      } else {
-        if (kindCase_ == 3) {
-          int64ListBuilder_.mergeFrom(value);
-        }
-        int64ListBuilder_.setMessage(value);
-      }
-      kindCase_ = 3;
-      return this;
-    }
-    /**
-     * <code>.tensorflow.Int64List int64_list = 3;</code>
-     */
-    public Builder clearInt64List() {
-      if (int64ListBuilder_ == null) {
-        if (kindCase_ == 3) {
-          kindCase_ = 0;
-          kind_ = null;
-          onChanged();
-        }
-      } else {
-        if (kindCase_ == 3) {
-          kindCase_ = 0;
-          kind_ = null;
-        }
-        int64ListBuilder_.clear();
-      }
-      return this;
-    }
-    /**
-     * <code>.tensorflow.Int64List int64_list = 3;</code>
-     */
-    public org.tensorflow.proto.example.Int64List.Builder getInt64ListBuilder() {
-      return getInt64ListFieldBuilder().getBuilder();
-    }
-    /**
-     * <code>.tensorflow.Int64List int64_list = 3;</code>
-     */
-    public org.tensorflow.proto.example.Int64ListOrBuilder getInt64ListOrBuilder() {
-      if ((kindCase_ == 3) && (int64ListBuilder_ != null)) {
-        return int64ListBuilder_.getMessageOrBuilder();
-      } else {
-        if (kindCase_ == 3) {
-          return (org.tensorflow.proto.example.Int64List) kind_;
-        }
-        return org.tensorflow.proto.example.Int64List.getDefaultInstance();
-      }
-    }
-    /**
-     * <code>.tensorflow.Int64List int64_list = 3;</code>
-     */
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.example.Int64List, org.tensorflow.proto.example.Int64List.Builder, org.tensorflow.proto.example.Int64ListOrBuilder> 
-        getInt64ListFieldBuilder() {
-      if (int64ListBuilder_ == null) {
-        if (!(kindCase_ == 3)) {
-          kind_ = org.tensorflow.proto.example.Int64List.getDefaultInstance();
-        }
-        int64ListBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-            org.tensorflow.proto.example.Int64List, org.tensorflow.proto.example.Int64List.Builder, org.tensorflow.proto.example.Int64ListOrBuilder>(
-                (org.tensorflow.proto.example.Int64List) kind_,
-                getParentForChildren(),
-                isClean());
-        kind_ = null;
-      }
-      kindCase_ = 3;
-      onChanged();;
-      return int64ListBuilder_;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.Feature)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.Feature)
-  private static final org.tensorflow.proto.example.Feature DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.example.Feature();
-  }
-
-  public static org.tensorflow.proto.example.Feature getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<Feature>
-      PARSER = new com.google.protobuf.AbstractParser<Feature>() {
-    @java.lang.Override
-    public Feature parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new Feature(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<Feature> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<Feature> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.example.Feature getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/FeatureConfiguration.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/FeatureConfiguration.java
deleted file mode 100644
index 6034a97c93e..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/FeatureConfiguration.java
+++ /dev/null
@@ -1,893 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/example/example_parser_configuration.proto
-
-package org.tensorflow.proto.example;
-
-/**
- * Protobuf type {@code tensorflow.FeatureConfiguration}
- */
-public  final class FeatureConfiguration extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.FeatureConfiguration)
-    FeatureConfigurationOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use FeatureConfiguration.newBuilder() to construct.
-  private FeatureConfiguration(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private FeatureConfiguration() {
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new FeatureConfiguration();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private FeatureConfiguration(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 10: {
-            org.tensorflow.proto.example.FixedLenFeatureProto.Builder subBuilder = null;
-            if (configCase_ == 1) {
-              subBuilder = ((org.tensorflow.proto.example.FixedLenFeatureProto) config_).toBuilder();
-            }
-            config_ =
-                input.readMessage(org.tensorflow.proto.example.FixedLenFeatureProto.parser(), extensionRegistry);
-            if (subBuilder != null) {
-              subBuilder.mergeFrom((org.tensorflow.proto.example.FixedLenFeatureProto) config_);
-              config_ = subBuilder.buildPartial();
-            }
-            configCase_ = 1;
-            break;
-          }
-          case 18: {
-            org.tensorflow.proto.example.VarLenFeatureProto.Builder subBuilder = null;
-            if (configCase_ == 2) {
-              subBuilder = ((org.tensorflow.proto.example.VarLenFeatureProto) config_).toBuilder();
-            }
-            config_ =
-                input.readMessage(org.tensorflow.proto.example.VarLenFeatureProto.parser(), extensionRegistry);
-            if (subBuilder != null) {
-              subBuilder.mergeFrom((org.tensorflow.proto.example.VarLenFeatureProto) config_);
-              config_ = subBuilder.buildPartial();
-            }
-            configCase_ = 2;
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.example.ExampleParserConfigurationProtos.internal_static_tensorflow_FeatureConfiguration_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.example.ExampleParserConfigurationProtos.internal_static_tensorflow_FeatureConfiguration_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.example.FeatureConfiguration.class, org.tensorflow.proto.example.FeatureConfiguration.Builder.class);
-  }
-
-  private int configCase_ = 0;
-  private java.lang.Object config_;
-  public enum ConfigCase
-      implements com.google.protobuf.Internal.EnumLite {
-    FIXED_LEN_FEATURE(1),
-    VAR_LEN_FEATURE(2),
-    CONFIG_NOT_SET(0);
-    private final int value;
-    private ConfigCase(int value) {
-      this.value = value;
-    }
-    /**
-     * @deprecated Use {@link #forNumber(int)} instead.
-     */
-    @java.lang.Deprecated
-    public static ConfigCase valueOf(int value) {
-      return forNumber(value);
-    }
-
-    public static ConfigCase forNumber(int value) {
-      switch (value) {
-        case 1: return FIXED_LEN_FEATURE;
-        case 2: return VAR_LEN_FEATURE;
-        case 0: return CONFIG_NOT_SET;
-        default: return null;
-      }
-    }
-    public int getNumber() {
-      return this.value;
-    }
-  };
-
-  public ConfigCase
-  getConfigCase() {
-    return ConfigCase.forNumber(
-        configCase_);
-  }
-
-  public static final int FIXED_LEN_FEATURE_FIELD_NUMBER = 1;
-  /**
-   * <code>.tensorflow.FixedLenFeatureProto fixed_len_feature = 1;</code>
-   */
-  public boolean hasFixedLenFeature() {
-    return configCase_ == 1;
-  }
-  /**
-   * <code>.tensorflow.FixedLenFeatureProto fixed_len_feature = 1;</code>
-   */
-  public org.tensorflow.proto.example.FixedLenFeatureProto getFixedLenFeature() {
-    if (configCase_ == 1) {
-       return (org.tensorflow.proto.example.FixedLenFeatureProto) config_;
-    }
-    return org.tensorflow.proto.example.FixedLenFeatureProto.getDefaultInstance();
-  }
-  /**
-   * <code>.tensorflow.FixedLenFeatureProto fixed_len_feature = 1;</code>
-   */
-  public org.tensorflow.proto.example.FixedLenFeatureProtoOrBuilder getFixedLenFeatureOrBuilder() {
-    if (configCase_ == 1) {
-       return (org.tensorflow.proto.example.FixedLenFeatureProto) config_;
-    }
-    return org.tensorflow.proto.example.FixedLenFeatureProto.getDefaultInstance();
-  }
-
-  public static final int VAR_LEN_FEATURE_FIELD_NUMBER = 2;
-  /**
-   * <code>.tensorflow.VarLenFeatureProto var_len_feature = 2;</code>
-   */
-  public boolean hasVarLenFeature() {
-    return configCase_ == 2;
-  }
-  /**
-   * <code>.tensorflow.VarLenFeatureProto var_len_feature = 2;</code>
-   */
-  public org.tensorflow.proto.example.VarLenFeatureProto getVarLenFeature() {
-    if (configCase_ == 2) {
-       return (org.tensorflow.proto.example.VarLenFeatureProto) config_;
-    }
-    return org.tensorflow.proto.example.VarLenFeatureProto.getDefaultInstance();
-  }
-  /**
-   * <code>.tensorflow.VarLenFeatureProto var_len_feature = 2;</code>
-   */
-  public org.tensorflow.proto.example.VarLenFeatureProtoOrBuilder getVarLenFeatureOrBuilder() {
-    if (configCase_ == 2) {
-       return (org.tensorflow.proto.example.VarLenFeatureProto) config_;
-    }
-    return org.tensorflow.proto.example.VarLenFeatureProto.getDefaultInstance();
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    if (configCase_ == 1) {
-      output.writeMessage(1, (org.tensorflow.proto.example.FixedLenFeatureProto) config_);
-    }
-    if (configCase_ == 2) {
-      output.writeMessage(2, (org.tensorflow.proto.example.VarLenFeatureProto) config_);
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (configCase_ == 1) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(1, (org.tensorflow.proto.example.FixedLenFeatureProto) config_);
-    }
-    if (configCase_ == 2) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(2, (org.tensorflow.proto.example.VarLenFeatureProto) config_);
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.example.FeatureConfiguration)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.example.FeatureConfiguration other = (org.tensorflow.proto.example.FeatureConfiguration) obj;
-
-    if (!getConfigCase().equals(other.getConfigCase())) return false;
-    switch (configCase_) {
-      case 1:
-        if (!getFixedLenFeature()
-            .equals(other.getFixedLenFeature())) return false;
-        break;
-      case 2:
-        if (!getVarLenFeature()
-            .equals(other.getVarLenFeature())) return false;
-        break;
-      case 0:
-      default:
-    }
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    switch (configCase_) {
-      case 1:
-        hash = (37 * hash) + FIXED_LEN_FEATURE_FIELD_NUMBER;
-        hash = (53 * hash) + getFixedLenFeature().hashCode();
-        break;
-      case 2:
-        hash = (37 * hash) + VAR_LEN_FEATURE_FIELD_NUMBER;
-        hash = (53 * hash) + getVarLenFeature().hashCode();
-        break;
-      case 0:
-      default:
-    }
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.example.FeatureConfiguration parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.example.FeatureConfiguration parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.example.FeatureConfiguration parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.example.FeatureConfiguration parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.example.FeatureConfiguration parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.example.FeatureConfiguration parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.example.FeatureConfiguration parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.example.FeatureConfiguration parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.example.FeatureConfiguration parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.example.FeatureConfiguration parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.example.FeatureConfiguration parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.example.FeatureConfiguration parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.example.FeatureConfiguration prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * Protobuf type {@code tensorflow.FeatureConfiguration}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.FeatureConfiguration)
-      org.tensorflow.proto.example.FeatureConfigurationOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.example.ExampleParserConfigurationProtos.internal_static_tensorflow_FeatureConfiguration_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.example.ExampleParserConfigurationProtos.internal_static_tensorflow_FeatureConfiguration_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.example.FeatureConfiguration.class, org.tensorflow.proto.example.FeatureConfiguration.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.example.FeatureConfiguration.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      configCase_ = 0;
-      config_ = null;
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.example.ExampleParserConfigurationProtos.internal_static_tensorflow_FeatureConfiguration_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.example.FeatureConfiguration getDefaultInstanceForType() {
-      return org.tensorflow.proto.example.FeatureConfiguration.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.example.FeatureConfiguration build() {
-      org.tensorflow.proto.example.FeatureConfiguration result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.example.FeatureConfiguration buildPartial() {
-      org.tensorflow.proto.example.FeatureConfiguration result = new org.tensorflow.proto.example.FeatureConfiguration(this);
-      if (configCase_ == 1) {
-        if (fixedLenFeatureBuilder_ == null) {
-          result.config_ = config_;
-        } else {
-          result.config_ = fixedLenFeatureBuilder_.build();
-        }
-      }
-      if (configCase_ == 2) {
-        if (varLenFeatureBuilder_ == null) {
-          result.config_ = config_;
-        } else {
-          result.config_ = varLenFeatureBuilder_.build();
-        }
-      }
-      result.configCase_ = configCase_;
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.example.FeatureConfiguration) {
-        return mergeFrom((org.tensorflow.proto.example.FeatureConfiguration)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.example.FeatureConfiguration other) {
-      if (other == org.tensorflow.proto.example.FeatureConfiguration.getDefaultInstance()) return this;
-      switch (other.getConfigCase()) {
-        case FIXED_LEN_FEATURE: {
-          mergeFixedLenFeature(other.getFixedLenFeature());
-          break;
-        }
-        case VAR_LEN_FEATURE: {
-          mergeVarLenFeature(other.getVarLenFeature());
-          break;
-        }
-        case CONFIG_NOT_SET: {
-          break;
-        }
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.example.FeatureConfiguration parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.example.FeatureConfiguration) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-    private int configCase_ = 0;
-    private java.lang.Object config_;
-    public ConfigCase
-        getConfigCase() {
-      return ConfigCase.forNumber(
-          configCase_);
-    }
-
-    public Builder clearConfig() {
-      configCase_ = 0;
-      config_ = null;
-      onChanged();
-      return this;
-    }
-
-
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.example.FixedLenFeatureProto, org.tensorflow.proto.example.FixedLenFeatureProto.Builder, org.tensorflow.proto.example.FixedLenFeatureProtoOrBuilder> fixedLenFeatureBuilder_;
-    /**
-     * <code>.tensorflow.FixedLenFeatureProto fixed_len_feature = 1;</code>
-     */
-    public boolean hasFixedLenFeature() {
-      return configCase_ == 1;
-    }
-    /**
-     * <code>.tensorflow.FixedLenFeatureProto fixed_len_feature = 1;</code>
-     */
-    public org.tensorflow.proto.example.FixedLenFeatureProto getFixedLenFeature() {
-      if (fixedLenFeatureBuilder_ == null) {
-        if (configCase_ == 1) {
-          return (org.tensorflow.proto.example.FixedLenFeatureProto) config_;
-        }
-        return org.tensorflow.proto.example.FixedLenFeatureProto.getDefaultInstance();
-      } else {
-        if (configCase_ == 1) {
-          return fixedLenFeatureBuilder_.getMessage();
-        }
-        return org.tensorflow.proto.example.FixedLenFeatureProto.getDefaultInstance();
-      }
-    }
-    /**
-     * <code>.tensorflow.FixedLenFeatureProto fixed_len_feature = 1;</code>
-     */
-    public Builder setFixedLenFeature(org.tensorflow.proto.example.FixedLenFeatureProto value) {
-      if (fixedLenFeatureBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        config_ = value;
-        onChanged();
-      } else {
-        fixedLenFeatureBuilder_.setMessage(value);
-      }
-      configCase_ = 1;
-      return this;
-    }
-    /**
-     * <code>.tensorflow.FixedLenFeatureProto fixed_len_feature = 1;</code>
-     */
-    public Builder setFixedLenFeature(
-        org.tensorflow.proto.example.FixedLenFeatureProto.Builder builderForValue) {
-      if (fixedLenFeatureBuilder_ == null) {
-        config_ = builderForValue.build();
-        onChanged();
-      } else {
-        fixedLenFeatureBuilder_.setMessage(builderForValue.build());
-      }
-      configCase_ = 1;
-      return this;
-    }
-    /**
-     * <code>.tensorflow.FixedLenFeatureProto fixed_len_feature = 1;</code>
-     */
-    public Builder mergeFixedLenFeature(org.tensorflow.proto.example.FixedLenFeatureProto value) {
-      if (fixedLenFeatureBuilder_ == null) {
-        if (configCase_ == 1 &&
-            config_ != org.tensorflow.proto.example.FixedLenFeatureProto.getDefaultInstance()) {
-          config_ = org.tensorflow.proto.example.FixedLenFeatureProto.newBuilder((org.tensorflow.proto.example.FixedLenFeatureProto) config_)
-              .mergeFrom(value).buildPartial();
-        } else {
-          config_ = value;
-        }
-        onChanged();
-      } else {
-        if (configCase_ == 1) {
-          fixedLenFeatureBuilder_.mergeFrom(value);
-        }
-        fixedLenFeatureBuilder_.setMessage(value);
-      }
-      configCase_ = 1;
-      return this;
-    }
-    /**
-     * <code>.tensorflow.FixedLenFeatureProto fixed_len_feature = 1;</code>
-     */
-    public Builder clearFixedLenFeature() {
-      if (fixedLenFeatureBuilder_ == null) {
-        if (configCase_ == 1) {
-          configCase_ = 0;
-          config_ = null;
-          onChanged();
-        }
-      } else {
-        if (configCase_ == 1) {
-          configCase_ = 0;
-          config_ = null;
-        }
-        fixedLenFeatureBuilder_.clear();
-      }
-      return this;
-    }
-    /**
-     * <code>.tensorflow.FixedLenFeatureProto fixed_len_feature = 1;</code>
-     */
-    public org.tensorflow.proto.example.FixedLenFeatureProto.Builder getFixedLenFeatureBuilder() {
-      return getFixedLenFeatureFieldBuilder().getBuilder();
-    }
-    /**
-     * <code>.tensorflow.FixedLenFeatureProto fixed_len_feature = 1;</code>
-     */
-    public org.tensorflow.proto.example.FixedLenFeatureProtoOrBuilder getFixedLenFeatureOrBuilder() {
-      if ((configCase_ == 1) && (fixedLenFeatureBuilder_ != null)) {
-        return fixedLenFeatureBuilder_.getMessageOrBuilder();
-      } else {
-        if (configCase_ == 1) {
-          return (org.tensorflow.proto.example.FixedLenFeatureProto) config_;
-        }
-        return org.tensorflow.proto.example.FixedLenFeatureProto.getDefaultInstance();
-      }
-    }
-    /**
-     * <code>.tensorflow.FixedLenFeatureProto fixed_len_feature = 1;</code>
-     */
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.example.FixedLenFeatureProto, org.tensorflow.proto.example.FixedLenFeatureProto.Builder, org.tensorflow.proto.example.FixedLenFeatureProtoOrBuilder> 
-        getFixedLenFeatureFieldBuilder() {
-      if (fixedLenFeatureBuilder_ == null) {
-        if (!(configCase_ == 1)) {
-          config_ = org.tensorflow.proto.example.FixedLenFeatureProto.getDefaultInstance();
-        }
-        fixedLenFeatureBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-            org.tensorflow.proto.example.FixedLenFeatureProto, org.tensorflow.proto.example.FixedLenFeatureProto.Builder, org.tensorflow.proto.example.FixedLenFeatureProtoOrBuilder>(
-                (org.tensorflow.proto.example.FixedLenFeatureProto) config_,
-                getParentForChildren(),
-                isClean());
-        config_ = null;
-      }
-      configCase_ = 1;
-      onChanged();;
-      return fixedLenFeatureBuilder_;
-    }
-
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.example.VarLenFeatureProto, org.tensorflow.proto.example.VarLenFeatureProto.Builder, org.tensorflow.proto.example.VarLenFeatureProtoOrBuilder> varLenFeatureBuilder_;
-    /**
-     * <code>.tensorflow.VarLenFeatureProto var_len_feature = 2;</code>
-     */
-    public boolean hasVarLenFeature() {
-      return configCase_ == 2;
-    }
-    /**
-     * <code>.tensorflow.VarLenFeatureProto var_len_feature = 2;</code>
-     */
-    public org.tensorflow.proto.example.VarLenFeatureProto getVarLenFeature() {
-      if (varLenFeatureBuilder_ == null) {
-        if (configCase_ == 2) {
-          return (org.tensorflow.proto.example.VarLenFeatureProto) config_;
-        }
-        return org.tensorflow.proto.example.VarLenFeatureProto.getDefaultInstance();
-      } else {
-        if (configCase_ == 2) {
-          return varLenFeatureBuilder_.getMessage();
-        }
-        return org.tensorflow.proto.example.VarLenFeatureProto.getDefaultInstance();
-      }
-    }
-    /**
-     * <code>.tensorflow.VarLenFeatureProto var_len_feature = 2;</code>
-     */
-    public Builder setVarLenFeature(org.tensorflow.proto.example.VarLenFeatureProto value) {
-      if (varLenFeatureBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        config_ = value;
-        onChanged();
-      } else {
-        varLenFeatureBuilder_.setMessage(value);
-      }
-      configCase_ = 2;
-      return this;
-    }
-    /**
-     * <code>.tensorflow.VarLenFeatureProto var_len_feature = 2;</code>
-     */
-    public Builder setVarLenFeature(
-        org.tensorflow.proto.example.VarLenFeatureProto.Builder builderForValue) {
-      if (varLenFeatureBuilder_ == null) {
-        config_ = builderForValue.build();
-        onChanged();
-      } else {
-        varLenFeatureBuilder_.setMessage(builderForValue.build());
-      }
-      configCase_ = 2;
-      return this;
-    }
-    /**
-     * <code>.tensorflow.VarLenFeatureProto var_len_feature = 2;</code>
-     */
-    public Builder mergeVarLenFeature(org.tensorflow.proto.example.VarLenFeatureProto value) {
-      if (varLenFeatureBuilder_ == null) {
-        if (configCase_ == 2 &&
-            config_ != org.tensorflow.proto.example.VarLenFeatureProto.getDefaultInstance()) {
-          config_ = org.tensorflow.proto.example.VarLenFeatureProto.newBuilder((org.tensorflow.proto.example.VarLenFeatureProto) config_)
-              .mergeFrom(value).buildPartial();
-        } else {
-          config_ = value;
-        }
-        onChanged();
-      } else {
-        if (configCase_ == 2) {
-          varLenFeatureBuilder_.mergeFrom(value);
-        }
-        varLenFeatureBuilder_.setMessage(value);
-      }
-      configCase_ = 2;
-      return this;
-    }
-    /**
-     * <code>.tensorflow.VarLenFeatureProto var_len_feature = 2;</code>
-     */
-    public Builder clearVarLenFeature() {
-      if (varLenFeatureBuilder_ == null) {
-        if (configCase_ == 2) {
-          configCase_ = 0;
-          config_ = null;
-          onChanged();
-        }
-      } else {
-        if (configCase_ == 2) {
-          configCase_ = 0;
-          config_ = null;
-        }
-        varLenFeatureBuilder_.clear();
-      }
-      return this;
-    }
-    /**
-     * <code>.tensorflow.VarLenFeatureProto var_len_feature = 2;</code>
-     */
-    public org.tensorflow.proto.example.VarLenFeatureProto.Builder getVarLenFeatureBuilder() {
-      return getVarLenFeatureFieldBuilder().getBuilder();
-    }
-    /**
-     * <code>.tensorflow.VarLenFeatureProto var_len_feature = 2;</code>
-     */
-    public org.tensorflow.proto.example.VarLenFeatureProtoOrBuilder getVarLenFeatureOrBuilder() {
-      if ((configCase_ == 2) && (varLenFeatureBuilder_ != null)) {
-        return varLenFeatureBuilder_.getMessageOrBuilder();
-      } else {
-        if (configCase_ == 2) {
-          return (org.tensorflow.proto.example.VarLenFeatureProto) config_;
-        }
-        return org.tensorflow.proto.example.VarLenFeatureProto.getDefaultInstance();
-      }
-    }
-    /**
-     * <code>.tensorflow.VarLenFeatureProto var_len_feature = 2;</code>
-     */
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.example.VarLenFeatureProto, org.tensorflow.proto.example.VarLenFeatureProto.Builder, org.tensorflow.proto.example.VarLenFeatureProtoOrBuilder> 
-        getVarLenFeatureFieldBuilder() {
-      if (varLenFeatureBuilder_ == null) {
-        if (!(configCase_ == 2)) {
-          config_ = org.tensorflow.proto.example.VarLenFeatureProto.getDefaultInstance();
-        }
-        varLenFeatureBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-            org.tensorflow.proto.example.VarLenFeatureProto, org.tensorflow.proto.example.VarLenFeatureProto.Builder, org.tensorflow.proto.example.VarLenFeatureProtoOrBuilder>(
-                (org.tensorflow.proto.example.VarLenFeatureProto) config_,
-                getParentForChildren(),
-                isClean());
-        config_ = null;
-      }
-      configCase_ = 2;
-      onChanged();;
-      return varLenFeatureBuilder_;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.FeatureConfiguration)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.FeatureConfiguration)
-  private static final org.tensorflow.proto.example.FeatureConfiguration DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.example.FeatureConfiguration();
-  }
-
-  public static org.tensorflow.proto.example.FeatureConfiguration getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<FeatureConfiguration>
-      PARSER = new com.google.protobuf.AbstractParser<FeatureConfiguration>() {
-    @java.lang.Override
-    public FeatureConfiguration parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new FeatureConfiguration(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<FeatureConfiguration> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<FeatureConfiguration> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.example.FeatureConfiguration getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/FeatureConfigurationOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/FeatureConfigurationOrBuilder.java
deleted file mode 100644
index 917025caa25..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/FeatureConfigurationOrBuilder.java
+++ /dev/null
@@ -1,37 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/example/example_parser_configuration.proto
-
-package org.tensorflow.proto.example;
-
-public interface FeatureConfigurationOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.FeatureConfiguration)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <code>.tensorflow.FixedLenFeatureProto fixed_len_feature = 1;</code>
-   */
-  boolean hasFixedLenFeature();
-  /**
-   * <code>.tensorflow.FixedLenFeatureProto fixed_len_feature = 1;</code>
-   */
-  org.tensorflow.proto.example.FixedLenFeatureProto getFixedLenFeature();
-  /**
-   * <code>.tensorflow.FixedLenFeatureProto fixed_len_feature = 1;</code>
-   */
-  org.tensorflow.proto.example.FixedLenFeatureProtoOrBuilder getFixedLenFeatureOrBuilder();
-
-  /**
-   * <code>.tensorflow.VarLenFeatureProto var_len_feature = 2;</code>
-   */
-  boolean hasVarLenFeature();
-  /**
-   * <code>.tensorflow.VarLenFeatureProto var_len_feature = 2;</code>
-   */
-  org.tensorflow.proto.example.VarLenFeatureProto getVarLenFeature();
-  /**
-   * <code>.tensorflow.VarLenFeatureProto var_len_feature = 2;</code>
-   */
-  org.tensorflow.proto.example.VarLenFeatureProtoOrBuilder getVarLenFeatureOrBuilder();
-
-  public org.tensorflow.proto.example.FeatureConfiguration.ConfigCase getConfigCase();
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/FeatureList.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/FeatureList.java
deleted file mode 100644
index 6aa9f29864a..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/FeatureList.java
+++ /dev/null
@@ -1,781 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/example/feature.proto
-
-package org.tensorflow.proto.example;
-
-/**
- * <pre>
- * Containers for sequential data.
- * A FeatureList contains lists of Features.  These may hold zero or more
- * Feature values.
- * FeatureLists are organized into categories by name.  The FeatureLists message
- * contains the mapping from name to FeatureList.
- * </pre>
- *
- * Protobuf type {@code tensorflow.FeatureList}
- */
-public  final class FeatureList extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.FeatureList)
-    FeatureListOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use FeatureList.newBuilder() to construct.
-  private FeatureList(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private FeatureList() {
-    feature_ = java.util.Collections.emptyList();
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new FeatureList();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private FeatureList(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    int mutable_bitField0_ = 0;
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 10: {
-            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
-              feature_ = new java.util.ArrayList<org.tensorflow.proto.example.Feature>();
-              mutable_bitField0_ |= 0x00000001;
-            }
-            feature_.add(
-                input.readMessage(org.tensorflow.proto.example.Feature.parser(), extensionRegistry));
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      if (((mutable_bitField0_ & 0x00000001) != 0)) {
-        feature_ = java.util.Collections.unmodifiableList(feature_);
-      }
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.example.FeatureProtos.internal_static_tensorflow_FeatureList_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.example.FeatureProtos.internal_static_tensorflow_FeatureList_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.example.FeatureList.class, org.tensorflow.proto.example.FeatureList.Builder.class);
-  }
-
-  public static final int FEATURE_FIELD_NUMBER = 1;
-  private java.util.List<org.tensorflow.proto.example.Feature> feature_;
-  /**
-   * <code>repeated .tensorflow.Feature feature = 1;</code>
-   */
-  public java.util.List<org.tensorflow.proto.example.Feature> getFeatureList() {
-    return feature_;
-  }
-  /**
-   * <code>repeated .tensorflow.Feature feature = 1;</code>
-   */
-  public java.util.List<? extends org.tensorflow.proto.example.FeatureOrBuilder> 
-      getFeatureOrBuilderList() {
-    return feature_;
-  }
-  /**
-   * <code>repeated .tensorflow.Feature feature = 1;</code>
-   */
-  public int getFeatureCount() {
-    return feature_.size();
-  }
-  /**
-   * <code>repeated .tensorflow.Feature feature = 1;</code>
-   */
-  public org.tensorflow.proto.example.Feature getFeature(int index) {
-    return feature_.get(index);
-  }
-  /**
-   * <code>repeated .tensorflow.Feature feature = 1;</code>
-   */
-  public org.tensorflow.proto.example.FeatureOrBuilder getFeatureOrBuilder(
-      int index) {
-    return feature_.get(index);
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    for (int i = 0; i < feature_.size(); i++) {
-      output.writeMessage(1, feature_.get(i));
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    for (int i = 0; i < feature_.size(); i++) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(1, feature_.get(i));
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.example.FeatureList)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.example.FeatureList other = (org.tensorflow.proto.example.FeatureList) obj;
-
-    if (!getFeatureList()
-        .equals(other.getFeatureList())) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    if (getFeatureCount() > 0) {
-      hash = (37 * hash) + FEATURE_FIELD_NUMBER;
-      hash = (53 * hash) + getFeatureList().hashCode();
-    }
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.example.FeatureList parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.example.FeatureList parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.example.FeatureList parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.example.FeatureList parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.example.FeatureList parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.example.FeatureList parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.example.FeatureList parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.example.FeatureList parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.example.FeatureList parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.example.FeatureList parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.example.FeatureList parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.example.FeatureList parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.example.FeatureList prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * <pre>
-   * Containers for sequential data.
-   * A FeatureList contains lists of Features.  These may hold zero or more
-   * Feature values.
-   * FeatureLists are organized into categories by name.  The FeatureLists message
-   * contains the mapping from name to FeatureList.
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.FeatureList}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.FeatureList)
-      org.tensorflow.proto.example.FeatureListOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.example.FeatureProtos.internal_static_tensorflow_FeatureList_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.example.FeatureProtos.internal_static_tensorflow_FeatureList_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.example.FeatureList.class, org.tensorflow.proto.example.FeatureList.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.example.FeatureList.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-        getFeatureFieldBuilder();
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      if (featureBuilder_ == null) {
-        feature_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000001);
-      } else {
-        featureBuilder_.clear();
-      }
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.example.FeatureProtos.internal_static_tensorflow_FeatureList_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.example.FeatureList getDefaultInstanceForType() {
-      return org.tensorflow.proto.example.FeatureList.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.example.FeatureList build() {
-      org.tensorflow.proto.example.FeatureList result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.example.FeatureList buildPartial() {
-      org.tensorflow.proto.example.FeatureList result = new org.tensorflow.proto.example.FeatureList(this);
-      int from_bitField0_ = bitField0_;
-      if (featureBuilder_ == null) {
-        if (((bitField0_ & 0x00000001) != 0)) {
-          feature_ = java.util.Collections.unmodifiableList(feature_);
-          bitField0_ = (bitField0_ & ~0x00000001);
-        }
-        result.feature_ = feature_;
-      } else {
-        result.feature_ = featureBuilder_.build();
-      }
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.example.FeatureList) {
-        return mergeFrom((org.tensorflow.proto.example.FeatureList)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.example.FeatureList other) {
-      if (other == org.tensorflow.proto.example.FeatureList.getDefaultInstance()) return this;
-      if (featureBuilder_ == null) {
-        if (!other.feature_.isEmpty()) {
-          if (feature_.isEmpty()) {
-            feature_ = other.feature_;
-            bitField0_ = (bitField0_ & ~0x00000001);
-          } else {
-            ensureFeatureIsMutable();
-            feature_.addAll(other.feature_);
-          }
-          onChanged();
-        }
-      } else {
-        if (!other.feature_.isEmpty()) {
-          if (featureBuilder_.isEmpty()) {
-            featureBuilder_.dispose();
-            featureBuilder_ = null;
-            feature_ = other.feature_;
-            bitField0_ = (bitField0_ & ~0x00000001);
-            featureBuilder_ = 
-              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
-                 getFeatureFieldBuilder() : null;
-          } else {
-            featureBuilder_.addAllMessages(other.feature_);
-          }
-        }
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.example.FeatureList parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.example.FeatureList) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-    private int bitField0_;
-
-    private java.util.List<org.tensorflow.proto.example.Feature> feature_ =
-      java.util.Collections.emptyList();
-    private void ensureFeatureIsMutable() {
-      if (!((bitField0_ & 0x00000001) != 0)) {
-        feature_ = new java.util.ArrayList<org.tensorflow.proto.example.Feature>(feature_);
-        bitField0_ |= 0x00000001;
-       }
-    }
-
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.example.Feature, org.tensorflow.proto.example.Feature.Builder, org.tensorflow.proto.example.FeatureOrBuilder> featureBuilder_;
-
-    /**
-     * <code>repeated .tensorflow.Feature feature = 1;</code>
-     */
-    public java.util.List<org.tensorflow.proto.example.Feature> getFeatureList() {
-      if (featureBuilder_ == null) {
-        return java.util.Collections.unmodifiableList(feature_);
-      } else {
-        return featureBuilder_.getMessageList();
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.Feature feature = 1;</code>
-     */
-    public int getFeatureCount() {
-      if (featureBuilder_ == null) {
-        return feature_.size();
-      } else {
-        return featureBuilder_.getCount();
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.Feature feature = 1;</code>
-     */
-    public org.tensorflow.proto.example.Feature getFeature(int index) {
-      if (featureBuilder_ == null) {
-        return feature_.get(index);
-      } else {
-        return featureBuilder_.getMessage(index);
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.Feature feature = 1;</code>
-     */
-    public Builder setFeature(
-        int index, org.tensorflow.proto.example.Feature value) {
-      if (featureBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureFeatureIsMutable();
-        feature_.set(index, value);
-        onChanged();
-      } else {
-        featureBuilder_.setMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.Feature feature = 1;</code>
-     */
-    public Builder setFeature(
-        int index, org.tensorflow.proto.example.Feature.Builder builderForValue) {
-      if (featureBuilder_ == null) {
-        ensureFeatureIsMutable();
-        feature_.set(index, builderForValue.build());
-        onChanged();
-      } else {
-        featureBuilder_.setMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.Feature feature = 1;</code>
-     */
-    public Builder addFeature(org.tensorflow.proto.example.Feature value) {
-      if (featureBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureFeatureIsMutable();
-        feature_.add(value);
-        onChanged();
-      } else {
-        featureBuilder_.addMessage(value);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.Feature feature = 1;</code>
-     */
-    public Builder addFeature(
-        int index, org.tensorflow.proto.example.Feature value) {
-      if (featureBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureFeatureIsMutable();
-        feature_.add(index, value);
-        onChanged();
-      } else {
-        featureBuilder_.addMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.Feature feature = 1;</code>
-     */
-    public Builder addFeature(
-        org.tensorflow.proto.example.Feature.Builder builderForValue) {
-      if (featureBuilder_ == null) {
-        ensureFeatureIsMutable();
-        feature_.add(builderForValue.build());
-        onChanged();
-      } else {
-        featureBuilder_.addMessage(builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.Feature feature = 1;</code>
-     */
-    public Builder addFeature(
-        int index, org.tensorflow.proto.example.Feature.Builder builderForValue) {
-      if (featureBuilder_ == null) {
-        ensureFeatureIsMutable();
-        feature_.add(index, builderForValue.build());
-        onChanged();
-      } else {
-        featureBuilder_.addMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.Feature feature = 1;</code>
-     */
-    public Builder addAllFeature(
-        java.lang.Iterable<? extends org.tensorflow.proto.example.Feature> values) {
-      if (featureBuilder_ == null) {
-        ensureFeatureIsMutable();
-        com.google.protobuf.AbstractMessageLite.Builder.addAll(
-            values, feature_);
-        onChanged();
-      } else {
-        featureBuilder_.addAllMessages(values);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.Feature feature = 1;</code>
-     */
-    public Builder clearFeature() {
-      if (featureBuilder_ == null) {
-        feature_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000001);
-        onChanged();
-      } else {
-        featureBuilder_.clear();
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.Feature feature = 1;</code>
-     */
-    public Builder removeFeature(int index) {
-      if (featureBuilder_ == null) {
-        ensureFeatureIsMutable();
-        feature_.remove(index);
-        onChanged();
-      } else {
-        featureBuilder_.remove(index);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.Feature feature = 1;</code>
-     */
-    public org.tensorflow.proto.example.Feature.Builder getFeatureBuilder(
-        int index) {
-      return getFeatureFieldBuilder().getBuilder(index);
-    }
-    /**
-     * <code>repeated .tensorflow.Feature feature = 1;</code>
-     */
-    public org.tensorflow.proto.example.FeatureOrBuilder getFeatureOrBuilder(
-        int index) {
-      if (featureBuilder_ == null) {
-        return feature_.get(index);  } else {
-        return featureBuilder_.getMessageOrBuilder(index);
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.Feature feature = 1;</code>
-     */
-    public java.util.List<? extends org.tensorflow.proto.example.FeatureOrBuilder> 
-         getFeatureOrBuilderList() {
-      if (featureBuilder_ != null) {
-        return featureBuilder_.getMessageOrBuilderList();
-      } else {
-        return java.util.Collections.unmodifiableList(feature_);
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.Feature feature = 1;</code>
-     */
-    public org.tensorflow.proto.example.Feature.Builder addFeatureBuilder() {
-      return getFeatureFieldBuilder().addBuilder(
-          org.tensorflow.proto.example.Feature.getDefaultInstance());
-    }
-    /**
-     * <code>repeated .tensorflow.Feature feature = 1;</code>
-     */
-    public org.tensorflow.proto.example.Feature.Builder addFeatureBuilder(
-        int index) {
-      return getFeatureFieldBuilder().addBuilder(
-          index, org.tensorflow.proto.example.Feature.getDefaultInstance());
-    }
-    /**
-     * <code>repeated .tensorflow.Feature feature = 1;</code>
-     */
-    public java.util.List<org.tensorflow.proto.example.Feature.Builder> 
-         getFeatureBuilderList() {
-      return getFeatureFieldBuilder().getBuilderList();
-    }
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.example.Feature, org.tensorflow.proto.example.Feature.Builder, org.tensorflow.proto.example.FeatureOrBuilder> 
-        getFeatureFieldBuilder() {
-      if (featureBuilder_ == null) {
-        featureBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
-            org.tensorflow.proto.example.Feature, org.tensorflow.proto.example.Feature.Builder, org.tensorflow.proto.example.FeatureOrBuilder>(
-                feature_,
-                ((bitField0_ & 0x00000001) != 0),
-                getParentForChildren(),
-                isClean());
-        feature_ = null;
-      }
-      return featureBuilder_;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.FeatureList)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.FeatureList)
-  private static final org.tensorflow.proto.example.FeatureList DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.example.FeatureList();
-  }
-
-  public static org.tensorflow.proto.example.FeatureList getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<FeatureList>
-      PARSER = new com.google.protobuf.AbstractParser<FeatureList>() {
-    @java.lang.Override
-    public FeatureList parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new FeatureList(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<FeatureList> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<FeatureList> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.example.FeatureList getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/FeatureListOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/FeatureListOrBuilder.java
deleted file mode 100644
index 9302c3512db..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/FeatureListOrBuilder.java
+++ /dev/null
@@ -1,33 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/example/feature.proto
-
-package org.tensorflow.proto.example;
-
-public interface FeatureListOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.FeatureList)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <code>repeated .tensorflow.Feature feature = 1;</code>
-   */
-  java.util.List<org.tensorflow.proto.example.Feature> 
-      getFeatureList();
-  /**
-   * <code>repeated .tensorflow.Feature feature = 1;</code>
-   */
-  org.tensorflow.proto.example.Feature getFeature(int index);
-  /**
-   * <code>repeated .tensorflow.Feature feature = 1;</code>
-   */
-  int getFeatureCount();
-  /**
-   * <code>repeated .tensorflow.Feature feature = 1;</code>
-   */
-  java.util.List<? extends org.tensorflow.proto.example.FeatureOrBuilder> 
-      getFeatureOrBuilderList();
-  /**
-   * <code>repeated .tensorflow.Feature feature = 1;</code>
-   */
-  org.tensorflow.proto.example.FeatureOrBuilder getFeatureOrBuilder(
-      int index);
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/FeatureLists.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/FeatureLists.java
deleted file mode 100644
index 4b10f2b7d90..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/FeatureLists.java
+++ /dev/null
@@ -1,739 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/example/feature.proto
-
-package org.tensorflow.proto.example;
-
-/**
- * Protobuf type {@code tensorflow.FeatureLists}
- */
-public  final class FeatureLists extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.FeatureLists)
-    FeatureListsOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use FeatureLists.newBuilder() to construct.
-  private FeatureLists(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private FeatureLists() {
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new FeatureLists();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private FeatureLists(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    int mutable_bitField0_ = 0;
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 10: {
-            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
-              featureList_ = com.google.protobuf.MapField.newMapField(
-                  FeatureListDefaultEntryHolder.defaultEntry);
-              mutable_bitField0_ |= 0x00000001;
-            }
-            com.google.protobuf.MapEntry<java.lang.String, org.tensorflow.proto.example.FeatureList>
-            featureList__ = input.readMessage(
-                FeatureListDefaultEntryHolder.defaultEntry.getParserForType(), extensionRegistry);
-            featureList_.getMutableMap().put(
-                featureList__.getKey(), featureList__.getValue());
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.example.FeatureProtos.internal_static_tensorflow_FeatureLists_descriptor;
-  }
-
-  @SuppressWarnings({"rawtypes"})
-  @java.lang.Override
-  protected com.google.protobuf.MapField internalGetMapField(
-      int number) {
-    switch (number) {
-      case 1:
-        return internalGetFeatureList();
-      default:
-        throw new RuntimeException(
-            "Invalid map field number: " + number);
-    }
-  }
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.example.FeatureProtos.internal_static_tensorflow_FeatureLists_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.example.FeatureLists.class, org.tensorflow.proto.example.FeatureLists.Builder.class);
-  }
-
-  public static final int FEATURE_LIST_FIELD_NUMBER = 1;
-  private static final class FeatureListDefaultEntryHolder {
-    static final com.google.protobuf.MapEntry<
-        java.lang.String, org.tensorflow.proto.example.FeatureList> defaultEntry =
-            com.google.protobuf.MapEntry
-            .<java.lang.String, org.tensorflow.proto.example.FeatureList>newDefaultInstance(
-                org.tensorflow.proto.example.FeatureProtos.internal_static_tensorflow_FeatureLists_FeatureListEntry_descriptor, 
-                com.google.protobuf.WireFormat.FieldType.STRING,
-                "",
-                com.google.protobuf.WireFormat.FieldType.MESSAGE,
-                org.tensorflow.proto.example.FeatureList.getDefaultInstance());
-  }
-  private com.google.protobuf.MapField<
-      java.lang.String, org.tensorflow.proto.example.FeatureList> featureList_;
-  private com.google.protobuf.MapField<java.lang.String, org.tensorflow.proto.example.FeatureList>
-  internalGetFeatureList() {
-    if (featureList_ == null) {
-      return com.google.protobuf.MapField.emptyMapField(
-          FeatureListDefaultEntryHolder.defaultEntry);
-    }
-    return featureList_;
-  }
-
-  public int getFeatureListCount() {
-    return internalGetFeatureList().getMap().size();
-  }
-  /**
-   * <pre>
-   * Map from feature name to feature list.
-   * </pre>
-   *
-   * <code>map&lt;string, .tensorflow.FeatureList&gt; feature_list = 1;</code>
-   */
-
-  public boolean containsFeatureList(
-      java.lang.String key) {
-    if (key == null) { throw new java.lang.NullPointerException(); }
-    return internalGetFeatureList().getMap().containsKey(key);
-  }
-  /**
-   * Use {@link #getFeatureListMap()} instead.
-   */
-  @java.lang.Deprecated
-  public java.util.Map<java.lang.String, org.tensorflow.proto.example.FeatureList> getFeatureList() {
-    return getFeatureListMap();
-  }
-  /**
-   * <pre>
-   * Map from feature name to feature list.
-   * </pre>
-   *
-   * <code>map&lt;string, .tensorflow.FeatureList&gt; feature_list = 1;</code>
-   */
-
-  public java.util.Map<java.lang.String, org.tensorflow.proto.example.FeatureList> getFeatureListMap() {
-    return internalGetFeatureList().getMap();
-  }
-  /**
-   * <pre>
-   * Map from feature name to feature list.
-   * </pre>
-   *
-   * <code>map&lt;string, .tensorflow.FeatureList&gt; feature_list = 1;</code>
-   */
-
-  public org.tensorflow.proto.example.FeatureList getFeatureListOrDefault(
-      java.lang.String key,
-      org.tensorflow.proto.example.FeatureList defaultValue) {
-    if (key == null) { throw new java.lang.NullPointerException(); }
-    java.util.Map<java.lang.String, org.tensorflow.proto.example.FeatureList> map =
-        internalGetFeatureList().getMap();
-    return map.containsKey(key) ? map.get(key) : defaultValue;
-  }
-  /**
-   * <pre>
-   * Map from feature name to feature list.
-   * </pre>
-   *
-   * <code>map&lt;string, .tensorflow.FeatureList&gt; feature_list = 1;</code>
-   */
-
-  public org.tensorflow.proto.example.FeatureList getFeatureListOrThrow(
-      java.lang.String key) {
-    if (key == null) { throw new java.lang.NullPointerException(); }
-    java.util.Map<java.lang.String, org.tensorflow.proto.example.FeatureList> map =
-        internalGetFeatureList().getMap();
-    if (!map.containsKey(key)) {
-      throw new java.lang.IllegalArgumentException();
-    }
-    return map.get(key);
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    com.google.protobuf.GeneratedMessageV3
-      .serializeStringMapTo(
-        output,
-        internalGetFeatureList(),
-        FeatureListDefaultEntryHolder.defaultEntry,
-        1);
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    for (java.util.Map.Entry<java.lang.String, org.tensorflow.proto.example.FeatureList> entry
-         : internalGetFeatureList().getMap().entrySet()) {
-      com.google.protobuf.MapEntry<java.lang.String, org.tensorflow.proto.example.FeatureList>
-      featureList__ = FeatureListDefaultEntryHolder.defaultEntry.newBuilderForType()
-          .setKey(entry.getKey())
-          .setValue(entry.getValue())
-          .build();
-      size += com.google.protobuf.CodedOutputStream
-          .computeMessageSize(1, featureList__);
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.example.FeatureLists)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.example.FeatureLists other = (org.tensorflow.proto.example.FeatureLists) obj;
-
-    if (!internalGetFeatureList().equals(
-        other.internalGetFeatureList())) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    if (!internalGetFeatureList().getMap().isEmpty()) {
-      hash = (37 * hash) + FEATURE_LIST_FIELD_NUMBER;
-      hash = (53 * hash) + internalGetFeatureList().hashCode();
-    }
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.example.FeatureLists parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.example.FeatureLists parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.example.FeatureLists parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.example.FeatureLists parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.example.FeatureLists parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.example.FeatureLists parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.example.FeatureLists parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.example.FeatureLists parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.example.FeatureLists parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.example.FeatureLists parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.example.FeatureLists parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.example.FeatureLists parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.example.FeatureLists prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * Protobuf type {@code tensorflow.FeatureLists}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.FeatureLists)
-      org.tensorflow.proto.example.FeatureListsOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.example.FeatureProtos.internal_static_tensorflow_FeatureLists_descriptor;
-    }
-
-    @SuppressWarnings({"rawtypes"})
-    protected com.google.protobuf.MapField internalGetMapField(
-        int number) {
-      switch (number) {
-        case 1:
-          return internalGetFeatureList();
-        default:
-          throw new RuntimeException(
-              "Invalid map field number: " + number);
-      }
-    }
-    @SuppressWarnings({"rawtypes"})
-    protected com.google.protobuf.MapField internalGetMutableMapField(
-        int number) {
-      switch (number) {
-        case 1:
-          return internalGetMutableFeatureList();
-        default:
-          throw new RuntimeException(
-              "Invalid map field number: " + number);
-      }
-    }
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.example.FeatureProtos.internal_static_tensorflow_FeatureLists_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.example.FeatureLists.class, org.tensorflow.proto.example.FeatureLists.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.example.FeatureLists.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      internalGetMutableFeatureList().clear();
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.example.FeatureProtos.internal_static_tensorflow_FeatureLists_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.example.FeatureLists getDefaultInstanceForType() {
-      return org.tensorflow.proto.example.FeatureLists.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.example.FeatureLists build() {
-      org.tensorflow.proto.example.FeatureLists result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.example.FeatureLists buildPartial() {
-      org.tensorflow.proto.example.FeatureLists result = new org.tensorflow.proto.example.FeatureLists(this);
-      int from_bitField0_ = bitField0_;
-      result.featureList_ = internalGetFeatureList();
-      result.featureList_.makeImmutable();
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.example.FeatureLists) {
-        return mergeFrom((org.tensorflow.proto.example.FeatureLists)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.example.FeatureLists other) {
-      if (other == org.tensorflow.proto.example.FeatureLists.getDefaultInstance()) return this;
-      internalGetMutableFeatureList().mergeFrom(
-          other.internalGetFeatureList());
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.example.FeatureLists parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.example.FeatureLists) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-    private int bitField0_;
-
-    private com.google.protobuf.MapField<
-        java.lang.String, org.tensorflow.proto.example.FeatureList> featureList_;
-    private com.google.protobuf.MapField<java.lang.String, org.tensorflow.proto.example.FeatureList>
-    internalGetFeatureList() {
-      if (featureList_ == null) {
-        return com.google.protobuf.MapField.emptyMapField(
-            FeatureListDefaultEntryHolder.defaultEntry);
-      }
-      return featureList_;
-    }
-    private com.google.protobuf.MapField<java.lang.String, org.tensorflow.proto.example.FeatureList>
-    internalGetMutableFeatureList() {
-      onChanged();;
-      if (featureList_ == null) {
-        featureList_ = com.google.protobuf.MapField.newMapField(
-            FeatureListDefaultEntryHolder.defaultEntry);
-      }
-      if (!featureList_.isMutable()) {
-        featureList_ = featureList_.copy();
-      }
-      return featureList_;
-    }
-
-    public int getFeatureListCount() {
-      return internalGetFeatureList().getMap().size();
-    }
-    /**
-     * <pre>
-     * Map from feature name to feature list.
-     * </pre>
-     *
-     * <code>map&lt;string, .tensorflow.FeatureList&gt; feature_list = 1;</code>
-     */
-
-    public boolean containsFeatureList(
-        java.lang.String key) {
-      if (key == null) { throw new java.lang.NullPointerException(); }
-      return internalGetFeatureList().getMap().containsKey(key);
-    }
-    /**
-     * Use {@link #getFeatureListMap()} instead.
-     */
-    @java.lang.Deprecated
-    public java.util.Map<java.lang.String, org.tensorflow.proto.example.FeatureList> getFeatureList() {
-      return getFeatureListMap();
-    }
-    /**
-     * <pre>
-     * Map from feature name to feature list.
-     * </pre>
-     *
-     * <code>map&lt;string, .tensorflow.FeatureList&gt; feature_list = 1;</code>
-     */
-
-    public java.util.Map<java.lang.String, org.tensorflow.proto.example.FeatureList> getFeatureListMap() {
-      return internalGetFeatureList().getMap();
-    }
-    /**
-     * <pre>
-     * Map from feature name to feature list.
-     * </pre>
-     *
-     * <code>map&lt;string, .tensorflow.FeatureList&gt; feature_list = 1;</code>
-     */
-
-    public org.tensorflow.proto.example.FeatureList getFeatureListOrDefault(
-        java.lang.String key,
-        org.tensorflow.proto.example.FeatureList defaultValue) {
-      if (key == null) { throw new java.lang.NullPointerException(); }
-      java.util.Map<java.lang.String, org.tensorflow.proto.example.FeatureList> map =
-          internalGetFeatureList().getMap();
-      return map.containsKey(key) ? map.get(key) : defaultValue;
-    }
-    /**
-     * <pre>
-     * Map from feature name to feature list.
-     * </pre>
-     *
-     * <code>map&lt;string, .tensorflow.FeatureList&gt; feature_list = 1;</code>
-     */
-
-    public org.tensorflow.proto.example.FeatureList getFeatureListOrThrow(
-        java.lang.String key) {
-      if (key == null) { throw new java.lang.NullPointerException(); }
-      java.util.Map<java.lang.String, org.tensorflow.proto.example.FeatureList> map =
-          internalGetFeatureList().getMap();
-      if (!map.containsKey(key)) {
-        throw new java.lang.IllegalArgumentException();
-      }
-      return map.get(key);
-    }
-
-    public Builder clearFeatureList() {
-      internalGetMutableFeatureList().getMutableMap()
-          .clear();
-      return this;
-    }
-    /**
-     * <pre>
-     * Map from feature name to feature list.
-     * </pre>
-     *
-     * <code>map&lt;string, .tensorflow.FeatureList&gt; feature_list = 1;</code>
-     */
-
-    public Builder removeFeatureList(
-        java.lang.String key) {
-      if (key == null) { throw new java.lang.NullPointerException(); }
-      internalGetMutableFeatureList().getMutableMap()
-          .remove(key);
-      return this;
-    }
-    /**
-     * Use alternate mutation accessors instead.
-     */
-    @java.lang.Deprecated
-    public java.util.Map<java.lang.String, org.tensorflow.proto.example.FeatureList>
-    getMutableFeatureList() {
-      return internalGetMutableFeatureList().getMutableMap();
-    }
-    /**
-     * <pre>
-     * Map from feature name to feature list.
-     * </pre>
-     *
-     * <code>map&lt;string, .tensorflow.FeatureList&gt; feature_list = 1;</code>
-     */
-    public Builder putFeatureList(
-        java.lang.String key,
-        org.tensorflow.proto.example.FeatureList value) {
-      if (key == null) { throw new java.lang.NullPointerException(); }
-      if (value == null) { throw new java.lang.NullPointerException(); }
-      internalGetMutableFeatureList().getMutableMap()
-          .put(key, value);
-      return this;
-    }
-    /**
-     * <pre>
-     * Map from feature name to feature list.
-     * </pre>
-     *
-     * <code>map&lt;string, .tensorflow.FeatureList&gt; feature_list = 1;</code>
-     */
-
-    public Builder putAllFeatureList(
-        java.util.Map<java.lang.String, org.tensorflow.proto.example.FeatureList> values) {
-      internalGetMutableFeatureList().getMutableMap()
-          .putAll(values);
-      return this;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.FeatureLists)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.FeatureLists)
-  private static final org.tensorflow.proto.example.FeatureLists DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.example.FeatureLists();
-  }
-
-  public static org.tensorflow.proto.example.FeatureLists getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<FeatureLists>
-      PARSER = new com.google.protobuf.AbstractParser<FeatureLists>() {
-    @java.lang.Override
-    public FeatureLists parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new FeatureLists(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<FeatureLists> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<FeatureLists> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.example.FeatureLists getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/FeatureListsOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/FeatureListsOrBuilder.java
deleted file mode 100644
index 9f582b998ce..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/FeatureListsOrBuilder.java
+++ /dev/null
@@ -1,63 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/example/feature.proto
-
-package org.tensorflow.proto.example;
-
-public interface FeatureListsOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.FeatureLists)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <pre>
-   * Map from feature name to feature list.
-   * </pre>
-   *
-   * <code>map&lt;string, .tensorflow.FeatureList&gt; feature_list = 1;</code>
-   */
-  int getFeatureListCount();
-  /**
-   * <pre>
-   * Map from feature name to feature list.
-   * </pre>
-   *
-   * <code>map&lt;string, .tensorflow.FeatureList&gt; feature_list = 1;</code>
-   */
-  boolean containsFeatureList(
-      java.lang.String key);
-  /**
-   * Use {@link #getFeatureListMap()} instead.
-   */
-  @java.lang.Deprecated
-  java.util.Map<java.lang.String, org.tensorflow.proto.example.FeatureList>
-  getFeatureList();
-  /**
-   * <pre>
-   * Map from feature name to feature list.
-   * </pre>
-   *
-   * <code>map&lt;string, .tensorflow.FeatureList&gt; feature_list = 1;</code>
-   */
-  java.util.Map<java.lang.String, org.tensorflow.proto.example.FeatureList>
-  getFeatureListMap();
-  /**
-   * <pre>
-   * Map from feature name to feature list.
-   * </pre>
-   *
-   * <code>map&lt;string, .tensorflow.FeatureList&gt; feature_list = 1;</code>
-   */
-
-  org.tensorflow.proto.example.FeatureList getFeatureListOrDefault(
-      java.lang.String key,
-      org.tensorflow.proto.example.FeatureList defaultValue);
-  /**
-   * <pre>
-   * Map from feature name to feature list.
-   * </pre>
-   *
-   * <code>map&lt;string, .tensorflow.FeatureList&gt; feature_list = 1;</code>
-   */
-
-  org.tensorflow.proto.example.FeatureList getFeatureListOrThrow(
-      java.lang.String key);
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/FeatureOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/FeatureOrBuilder.java
deleted file mode 100644
index 92242bf9615..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/FeatureOrBuilder.java
+++ /dev/null
@@ -1,50 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/example/feature.proto
-
-package org.tensorflow.proto.example;
-
-public interface FeatureOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.Feature)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <code>.tensorflow.BytesList bytes_list = 1;</code>
-   */
-  boolean hasBytesList();
-  /**
-   * <code>.tensorflow.BytesList bytes_list = 1;</code>
-   */
-  org.tensorflow.proto.example.BytesList getBytesList();
-  /**
-   * <code>.tensorflow.BytesList bytes_list = 1;</code>
-   */
-  org.tensorflow.proto.example.BytesListOrBuilder getBytesListOrBuilder();
-
-  /**
-   * <code>.tensorflow.FloatList float_list = 2;</code>
-   */
-  boolean hasFloatList();
-  /**
-   * <code>.tensorflow.FloatList float_list = 2;</code>
-   */
-  org.tensorflow.proto.example.FloatList getFloatList();
-  /**
-   * <code>.tensorflow.FloatList float_list = 2;</code>
-   */
-  org.tensorflow.proto.example.FloatListOrBuilder getFloatListOrBuilder();
-
-  /**
-   * <code>.tensorflow.Int64List int64_list = 3;</code>
-   */
-  boolean hasInt64List();
-  /**
-   * <code>.tensorflow.Int64List int64_list = 3;</code>
-   */
-  org.tensorflow.proto.example.Int64List getInt64List();
-  /**
-   * <code>.tensorflow.Int64List int64_list = 3;</code>
-   */
-  org.tensorflow.proto.example.Int64ListOrBuilder getInt64ListOrBuilder();
-
-  public org.tensorflow.proto.example.Feature.KindCase getKindCase();
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/FeatureProtos.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/FeatureProtos.java
deleted file mode 100644
index a03e3fecc50..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/FeatureProtos.java
+++ /dev/null
@@ -1,153 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/example/feature.proto
-
-package org.tensorflow.proto.example;
-
-public final class FeatureProtos {
-  private FeatureProtos() {}
-  public static void registerAllExtensions(
-      com.google.protobuf.ExtensionRegistryLite registry) {
-  }
-
-  public static void registerAllExtensions(
-      com.google.protobuf.ExtensionRegistry registry) {
-    registerAllExtensions(
-        (com.google.protobuf.ExtensionRegistryLite) registry);
-  }
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_BytesList_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_BytesList_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_FloatList_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_FloatList_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_Int64List_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_Int64List_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_Feature_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_Feature_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_Features_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_Features_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_Features_FeatureEntry_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_Features_FeatureEntry_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_FeatureList_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_FeatureList_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_FeatureLists_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_FeatureLists_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_FeatureLists_FeatureListEntry_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_FeatureLists_FeatureListEntry_fieldAccessorTable;
-
-  public static com.google.protobuf.Descriptors.FileDescriptor
-      getDescriptor() {
-    return descriptor;
-  }
-  private static  com.google.protobuf.Descriptors.FileDescriptor
-      descriptor;
-  static {
-    java.lang.String[] descriptorData = {
-      "\n%tensorflow/core/example/feature.proto\022" +
-      "\ntensorflow\"\032\n\tBytesList\022\r\n\005value\030\001 \003(\014\"" +
-      "\036\n\tFloatList\022\021\n\005value\030\001 \003(\002B\002\020\001\"\036\n\tInt64" +
-      "List\022\021\n\005value\030\001 \003(\003B\002\020\001\"\230\001\n\007Feature\022+\n\nb" +
-      "ytes_list\030\001 \001(\0132\025.tensorflow.BytesListH\000" +
-      "\022+\n\nfloat_list\030\002 \001(\0132\025.tensorflow.FloatL" +
-      "istH\000\022+\n\nint64_list\030\003 \001(\0132\025.tensorflow.I" +
-      "nt64ListH\000B\006\n\004kind\"\203\001\n\010Features\0222\n\007featu" +
-      "re\030\001 \003(\0132!.tensorflow.Features.FeatureEn" +
-      "try\032C\n\014FeatureEntry\022\013\n\003key\030\001 \001(\t\022\"\n\005valu" +
-      "e\030\002 \001(\0132\023.tensorflow.Feature:\0028\001\"3\n\013Feat" +
-      "ureList\022$\n\007feature\030\001 \003(\0132\023.tensorflow.Fe" +
-      "ature\"\234\001\n\014FeatureLists\022?\n\014feature_list\030\001" +
-      " \003(\0132).tensorflow.FeatureLists.FeatureLi" +
-      "stEntry\032K\n\020FeatureListEntry\022\013\n\003key\030\001 \001(\t" +
-      "\022&\n\005value\030\002 \001(\0132\027.tensorflow.FeatureList" +
-      ":\0028\001B\207\001\n\034org.tensorflow.proto.exampleB\rF" +
-      "eatureProtosP\001ZSgithub.com/tensorflow/te" +
-      "nsorflow/tensorflow/go/core/example/exam" +
-      "ple_protos_go_proto\370\001\001b\006proto3"
-    };
-    descriptor = com.google.protobuf.Descriptors.FileDescriptor
-      .internalBuildGeneratedFileFrom(descriptorData,
-        new com.google.protobuf.Descriptors.FileDescriptor[] {
-        });
-    internal_static_tensorflow_BytesList_descriptor =
-      getDescriptor().getMessageTypes().get(0);
-    internal_static_tensorflow_BytesList_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_BytesList_descriptor,
-        new java.lang.String[] { "Value", });
-    internal_static_tensorflow_FloatList_descriptor =
-      getDescriptor().getMessageTypes().get(1);
-    internal_static_tensorflow_FloatList_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_FloatList_descriptor,
-        new java.lang.String[] { "Value", });
-    internal_static_tensorflow_Int64List_descriptor =
-      getDescriptor().getMessageTypes().get(2);
-    internal_static_tensorflow_Int64List_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_Int64List_descriptor,
-        new java.lang.String[] { "Value", });
-    internal_static_tensorflow_Feature_descriptor =
-      getDescriptor().getMessageTypes().get(3);
-    internal_static_tensorflow_Feature_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_Feature_descriptor,
-        new java.lang.String[] { "BytesList", "FloatList", "Int64List", "Kind", });
-    internal_static_tensorflow_Features_descriptor =
-      getDescriptor().getMessageTypes().get(4);
-    internal_static_tensorflow_Features_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_Features_descriptor,
-        new java.lang.String[] { "Feature", });
-    internal_static_tensorflow_Features_FeatureEntry_descriptor =
-      internal_static_tensorflow_Features_descriptor.getNestedTypes().get(0);
-    internal_static_tensorflow_Features_FeatureEntry_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_Features_FeatureEntry_descriptor,
-        new java.lang.String[] { "Key", "Value", });
-    internal_static_tensorflow_FeatureList_descriptor =
-      getDescriptor().getMessageTypes().get(5);
-    internal_static_tensorflow_FeatureList_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_FeatureList_descriptor,
-        new java.lang.String[] { "Feature", });
-    internal_static_tensorflow_FeatureLists_descriptor =
-      getDescriptor().getMessageTypes().get(6);
-    internal_static_tensorflow_FeatureLists_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_FeatureLists_descriptor,
-        new java.lang.String[] { "FeatureList", });
-    internal_static_tensorflow_FeatureLists_FeatureListEntry_descriptor =
-      internal_static_tensorflow_FeatureLists_descriptor.getNestedTypes().get(0);
-    internal_static_tensorflow_FeatureLists_FeatureListEntry_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_FeatureLists_FeatureListEntry_descriptor,
-        new java.lang.String[] { "Key", "Value", });
-  }
-
-  // @@protoc_insertion_point(outer_class_scope)
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/Features.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/Features.java
deleted file mode 100644
index 0e47b04941f..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/Features.java
+++ /dev/null
@@ -1,739 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/example/feature.proto
-
-package org.tensorflow.proto.example;
-
-/**
- * Protobuf type {@code tensorflow.Features}
- */
-public  final class Features extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.Features)
-    FeaturesOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use Features.newBuilder() to construct.
-  private Features(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private Features() {
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new Features();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private Features(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    int mutable_bitField0_ = 0;
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 10: {
-            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
-              feature_ = com.google.protobuf.MapField.newMapField(
-                  FeatureDefaultEntryHolder.defaultEntry);
-              mutable_bitField0_ |= 0x00000001;
-            }
-            com.google.protobuf.MapEntry<java.lang.String, org.tensorflow.proto.example.Feature>
-            feature__ = input.readMessage(
-                FeatureDefaultEntryHolder.defaultEntry.getParserForType(), extensionRegistry);
-            feature_.getMutableMap().put(
-                feature__.getKey(), feature__.getValue());
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.example.FeatureProtos.internal_static_tensorflow_Features_descriptor;
-  }
-
-  @SuppressWarnings({"rawtypes"})
-  @java.lang.Override
-  protected com.google.protobuf.MapField internalGetMapField(
-      int number) {
-    switch (number) {
-      case 1:
-        return internalGetFeature();
-      default:
-        throw new RuntimeException(
-            "Invalid map field number: " + number);
-    }
-  }
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.example.FeatureProtos.internal_static_tensorflow_Features_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.example.Features.class, org.tensorflow.proto.example.Features.Builder.class);
-  }
-
-  public static final int FEATURE_FIELD_NUMBER = 1;
-  private static final class FeatureDefaultEntryHolder {
-    static final com.google.protobuf.MapEntry<
-        java.lang.String, org.tensorflow.proto.example.Feature> defaultEntry =
-            com.google.protobuf.MapEntry
-            .<java.lang.String, org.tensorflow.proto.example.Feature>newDefaultInstance(
-                org.tensorflow.proto.example.FeatureProtos.internal_static_tensorflow_Features_FeatureEntry_descriptor, 
-                com.google.protobuf.WireFormat.FieldType.STRING,
-                "",
-                com.google.protobuf.WireFormat.FieldType.MESSAGE,
-                org.tensorflow.proto.example.Feature.getDefaultInstance());
-  }
-  private com.google.protobuf.MapField<
-      java.lang.String, org.tensorflow.proto.example.Feature> feature_;
-  private com.google.protobuf.MapField<java.lang.String, org.tensorflow.proto.example.Feature>
-  internalGetFeature() {
-    if (feature_ == null) {
-      return com.google.protobuf.MapField.emptyMapField(
-          FeatureDefaultEntryHolder.defaultEntry);
-    }
-    return feature_;
-  }
-
-  public int getFeatureCount() {
-    return internalGetFeature().getMap().size();
-  }
-  /**
-   * <pre>
-   * Map from feature name to feature.
-   * </pre>
-   *
-   * <code>map&lt;string, .tensorflow.Feature&gt; feature = 1;</code>
-   */
-
-  public boolean containsFeature(
-      java.lang.String key) {
-    if (key == null) { throw new java.lang.NullPointerException(); }
-    return internalGetFeature().getMap().containsKey(key);
-  }
-  /**
-   * Use {@link #getFeatureMap()} instead.
-   */
-  @java.lang.Deprecated
-  public java.util.Map<java.lang.String, org.tensorflow.proto.example.Feature> getFeature() {
-    return getFeatureMap();
-  }
-  /**
-   * <pre>
-   * Map from feature name to feature.
-   * </pre>
-   *
-   * <code>map&lt;string, .tensorflow.Feature&gt; feature = 1;</code>
-   */
-
-  public java.util.Map<java.lang.String, org.tensorflow.proto.example.Feature> getFeatureMap() {
-    return internalGetFeature().getMap();
-  }
-  /**
-   * <pre>
-   * Map from feature name to feature.
-   * </pre>
-   *
-   * <code>map&lt;string, .tensorflow.Feature&gt; feature = 1;</code>
-   */
-
-  public org.tensorflow.proto.example.Feature getFeatureOrDefault(
-      java.lang.String key,
-      org.tensorflow.proto.example.Feature defaultValue) {
-    if (key == null) { throw new java.lang.NullPointerException(); }
-    java.util.Map<java.lang.String, org.tensorflow.proto.example.Feature> map =
-        internalGetFeature().getMap();
-    return map.containsKey(key) ? map.get(key) : defaultValue;
-  }
-  /**
-   * <pre>
-   * Map from feature name to feature.
-   * </pre>
-   *
-   * <code>map&lt;string, .tensorflow.Feature&gt; feature = 1;</code>
-   */
-
-  public org.tensorflow.proto.example.Feature getFeatureOrThrow(
-      java.lang.String key) {
-    if (key == null) { throw new java.lang.NullPointerException(); }
-    java.util.Map<java.lang.String, org.tensorflow.proto.example.Feature> map =
-        internalGetFeature().getMap();
-    if (!map.containsKey(key)) {
-      throw new java.lang.IllegalArgumentException();
-    }
-    return map.get(key);
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    com.google.protobuf.GeneratedMessageV3
-      .serializeStringMapTo(
-        output,
-        internalGetFeature(),
-        FeatureDefaultEntryHolder.defaultEntry,
-        1);
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    for (java.util.Map.Entry<java.lang.String, org.tensorflow.proto.example.Feature> entry
-         : internalGetFeature().getMap().entrySet()) {
-      com.google.protobuf.MapEntry<java.lang.String, org.tensorflow.proto.example.Feature>
-      feature__ = FeatureDefaultEntryHolder.defaultEntry.newBuilderForType()
-          .setKey(entry.getKey())
-          .setValue(entry.getValue())
-          .build();
-      size += com.google.protobuf.CodedOutputStream
-          .computeMessageSize(1, feature__);
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.example.Features)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.example.Features other = (org.tensorflow.proto.example.Features) obj;
-
-    if (!internalGetFeature().equals(
-        other.internalGetFeature())) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    if (!internalGetFeature().getMap().isEmpty()) {
-      hash = (37 * hash) + FEATURE_FIELD_NUMBER;
-      hash = (53 * hash) + internalGetFeature().hashCode();
-    }
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.example.Features parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.example.Features parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.example.Features parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.example.Features parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.example.Features parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.example.Features parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.example.Features parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.example.Features parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.example.Features parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.example.Features parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.example.Features parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.example.Features parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.example.Features prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * Protobuf type {@code tensorflow.Features}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.Features)
-      org.tensorflow.proto.example.FeaturesOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.example.FeatureProtos.internal_static_tensorflow_Features_descriptor;
-    }
-
-    @SuppressWarnings({"rawtypes"})
-    protected com.google.protobuf.MapField internalGetMapField(
-        int number) {
-      switch (number) {
-        case 1:
-          return internalGetFeature();
-        default:
-          throw new RuntimeException(
-              "Invalid map field number: " + number);
-      }
-    }
-    @SuppressWarnings({"rawtypes"})
-    protected com.google.protobuf.MapField internalGetMutableMapField(
-        int number) {
-      switch (number) {
-        case 1:
-          return internalGetMutableFeature();
-        default:
-          throw new RuntimeException(
-              "Invalid map field number: " + number);
-      }
-    }
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.example.FeatureProtos.internal_static_tensorflow_Features_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.example.Features.class, org.tensorflow.proto.example.Features.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.example.Features.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      internalGetMutableFeature().clear();
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.example.FeatureProtos.internal_static_tensorflow_Features_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.example.Features getDefaultInstanceForType() {
-      return org.tensorflow.proto.example.Features.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.example.Features build() {
-      org.tensorflow.proto.example.Features result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.example.Features buildPartial() {
-      org.tensorflow.proto.example.Features result = new org.tensorflow.proto.example.Features(this);
-      int from_bitField0_ = bitField0_;
-      result.feature_ = internalGetFeature();
-      result.feature_.makeImmutable();
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.example.Features) {
-        return mergeFrom((org.tensorflow.proto.example.Features)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.example.Features other) {
-      if (other == org.tensorflow.proto.example.Features.getDefaultInstance()) return this;
-      internalGetMutableFeature().mergeFrom(
-          other.internalGetFeature());
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.example.Features parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.example.Features) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-    private int bitField0_;
-
-    private com.google.protobuf.MapField<
-        java.lang.String, org.tensorflow.proto.example.Feature> feature_;
-    private com.google.protobuf.MapField<java.lang.String, org.tensorflow.proto.example.Feature>
-    internalGetFeature() {
-      if (feature_ == null) {
-        return com.google.protobuf.MapField.emptyMapField(
-            FeatureDefaultEntryHolder.defaultEntry);
-      }
-      return feature_;
-    }
-    private com.google.protobuf.MapField<java.lang.String, org.tensorflow.proto.example.Feature>
-    internalGetMutableFeature() {
-      onChanged();;
-      if (feature_ == null) {
-        feature_ = com.google.protobuf.MapField.newMapField(
-            FeatureDefaultEntryHolder.defaultEntry);
-      }
-      if (!feature_.isMutable()) {
-        feature_ = feature_.copy();
-      }
-      return feature_;
-    }
-
-    public int getFeatureCount() {
-      return internalGetFeature().getMap().size();
-    }
-    /**
-     * <pre>
-     * Map from feature name to feature.
-     * </pre>
-     *
-     * <code>map&lt;string, .tensorflow.Feature&gt; feature = 1;</code>
-     */
-
-    public boolean containsFeature(
-        java.lang.String key) {
-      if (key == null) { throw new java.lang.NullPointerException(); }
-      return internalGetFeature().getMap().containsKey(key);
-    }
-    /**
-     * Use {@link #getFeatureMap()} instead.
-     */
-    @java.lang.Deprecated
-    public java.util.Map<java.lang.String, org.tensorflow.proto.example.Feature> getFeature() {
-      return getFeatureMap();
-    }
-    /**
-     * <pre>
-     * Map from feature name to feature.
-     * </pre>
-     *
-     * <code>map&lt;string, .tensorflow.Feature&gt; feature = 1;</code>
-     */
-
-    public java.util.Map<java.lang.String, org.tensorflow.proto.example.Feature> getFeatureMap() {
-      return internalGetFeature().getMap();
-    }
-    /**
-     * <pre>
-     * Map from feature name to feature.
-     * </pre>
-     *
-     * <code>map&lt;string, .tensorflow.Feature&gt; feature = 1;</code>
-     */
-
-    public org.tensorflow.proto.example.Feature getFeatureOrDefault(
-        java.lang.String key,
-        org.tensorflow.proto.example.Feature defaultValue) {
-      if (key == null) { throw new java.lang.NullPointerException(); }
-      java.util.Map<java.lang.String, org.tensorflow.proto.example.Feature> map =
-          internalGetFeature().getMap();
-      return map.containsKey(key) ? map.get(key) : defaultValue;
-    }
-    /**
-     * <pre>
-     * Map from feature name to feature.
-     * </pre>
-     *
-     * <code>map&lt;string, .tensorflow.Feature&gt; feature = 1;</code>
-     */
-
-    public org.tensorflow.proto.example.Feature getFeatureOrThrow(
-        java.lang.String key) {
-      if (key == null) { throw new java.lang.NullPointerException(); }
-      java.util.Map<java.lang.String, org.tensorflow.proto.example.Feature> map =
-          internalGetFeature().getMap();
-      if (!map.containsKey(key)) {
-        throw new java.lang.IllegalArgumentException();
-      }
-      return map.get(key);
-    }
-
-    public Builder clearFeature() {
-      internalGetMutableFeature().getMutableMap()
-          .clear();
-      return this;
-    }
-    /**
-     * <pre>
-     * Map from feature name to feature.
-     * </pre>
-     *
-     * <code>map&lt;string, .tensorflow.Feature&gt; feature = 1;</code>
-     */
-
-    public Builder removeFeature(
-        java.lang.String key) {
-      if (key == null) { throw new java.lang.NullPointerException(); }
-      internalGetMutableFeature().getMutableMap()
-          .remove(key);
-      return this;
-    }
-    /**
-     * Use alternate mutation accessors instead.
-     */
-    @java.lang.Deprecated
-    public java.util.Map<java.lang.String, org.tensorflow.proto.example.Feature>
-    getMutableFeature() {
-      return internalGetMutableFeature().getMutableMap();
-    }
-    /**
-     * <pre>
-     * Map from feature name to feature.
-     * </pre>
-     *
-     * <code>map&lt;string, .tensorflow.Feature&gt; feature = 1;</code>
-     */
-    public Builder putFeature(
-        java.lang.String key,
-        org.tensorflow.proto.example.Feature value) {
-      if (key == null) { throw new java.lang.NullPointerException(); }
-      if (value == null) { throw new java.lang.NullPointerException(); }
-      internalGetMutableFeature().getMutableMap()
-          .put(key, value);
-      return this;
-    }
-    /**
-     * <pre>
-     * Map from feature name to feature.
-     * </pre>
-     *
-     * <code>map&lt;string, .tensorflow.Feature&gt; feature = 1;</code>
-     */
-
-    public Builder putAllFeature(
-        java.util.Map<java.lang.String, org.tensorflow.proto.example.Feature> values) {
-      internalGetMutableFeature().getMutableMap()
-          .putAll(values);
-      return this;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.Features)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.Features)
-  private static final org.tensorflow.proto.example.Features DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.example.Features();
-  }
-
-  public static org.tensorflow.proto.example.Features getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<Features>
-      PARSER = new com.google.protobuf.AbstractParser<Features>() {
-    @java.lang.Override
-    public Features parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new Features(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<Features> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<Features> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.example.Features getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/FeaturesOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/FeaturesOrBuilder.java
deleted file mode 100644
index 2ee80ee66a0..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/FeaturesOrBuilder.java
+++ /dev/null
@@ -1,63 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/example/feature.proto
-
-package org.tensorflow.proto.example;
-
-public interface FeaturesOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.Features)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <pre>
-   * Map from feature name to feature.
-   * </pre>
-   *
-   * <code>map&lt;string, .tensorflow.Feature&gt; feature = 1;</code>
-   */
-  int getFeatureCount();
-  /**
-   * <pre>
-   * Map from feature name to feature.
-   * </pre>
-   *
-   * <code>map&lt;string, .tensorflow.Feature&gt; feature = 1;</code>
-   */
-  boolean containsFeature(
-      java.lang.String key);
-  /**
-   * Use {@link #getFeatureMap()} instead.
-   */
-  @java.lang.Deprecated
-  java.util.Map<java.lang.String, org.tensorflow.proto.example.Feature>
-  getFeature();
-  /**
-   * <pre>
-   * Map from feature name to feature.
-   * </pre>
-   *
-   * <code>map&lt;string, .tensorflow.Feature&gt; feature = 1;</code>
-   */
-  java.util.Map<java.lang.String, org.tensorflow.proto.example.Feature>
-  getFeatureMap();
-  /**
-   * <pre>
-   * Map from feature name to feature.
-   * </pre>
-   *
-   * <code>map&lt;string, .tensorflow.Feature&gt; feature = 1;</code>
-   */
-
-  org.tensorflow.proto.example.Feature getFeatureOrDefault(
-      java.lang.String key,
-      org.tensorflow.proto.example.Feature defaultValue);
-  /**
-   * <pre>
-   * Map from feature name to feature.
-   * </pre>
-   *
-   * <code>map&lt;string, .tensorflow.Feature&gt; feature = 1;</code>
-   */
-
-  org.tensorflow.proto.example.Feature getFeatureOrThrow(
-      java.lang.String key);
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/FixedLenFeatureProto.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/FixedLenFeatureProto.java
deleted file mode 100644
index 1f378121573..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/FixedLenFeatureProto.java
+++ /dev/null
@@ -1,993 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/example/example_parser_configuration.proto
-
-package org.tensorflow.proto.example;
-
-/**
- * Protobuf type {@code tensorflow.FixedLenFeatureProto}
- */
-public  final class FixedLenFeatureProto extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.FixedLenFeatureProto)
-    FixedLenFeatureProtoOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use FixedLenFeatureProto.newBuilder() to construct.
-  private FixedLenFeatureProto(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private FixedLenFeatureProto() {
-    dtype_ = 0;
-    valuesOutputTensorName_ = "";
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new FixedLenFeatureProto();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private FixedLenFeatureProto(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 8: {
-            int rawValue = input.readEnum();
-
-            dtype_ = rawValue;
-            break;
-          }
-          case 18: {
-            org.tensorflow.proto.framework.TensorShapeProto.Builder subBuilder = null;
-            if (shape_ != null) {
-              subBuilder = shape_.toBuilder();
-            }
-            shape_ = input.readMessage(org.tensorflow.proto.framework.TensorShapeProto.parser(), extensionRegistry);
-            if (subBuilder != null) {
-              subBuilder.mergeFrom(shape_);
-              shape_ = subBuilder.buildPartial();
-            }
-
-            break;
-          }
-          case 26: {
-            org.tensorflow.proto.framework.TensorProto.Builder subBuilder = null;
-            if (defaultValue_ != null) {
-              subBuilder = defaultValue_.toBuilder();
-            }
-            defaultValue_ = input.readMessage(org.tensorflow.proto.framework.TensorProto.parser(), extensionRegistry);
-            if (subBuilder != null) {
-              subBuilder.mergeFrom(defaultValue_);
-              defaultValue_ = subBuilder.buildPartial();
-            }
-
-            break;
-          }
-          case 34: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            valuesOutputTensorName_ = s;
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.example.ExampleParserConfigurationProtos.internal_static_tensorflow_FixedLenFeatureProto_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.example.ExampleParserConfigurationProtos.internal_static_tensorflow_FixedLenFeatureProto_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.example.FixedLenFeatureProto.class, org.tensorflow.proto.example.FixedLenFeatureProto.Builder.class);
-  }
-
-  public static final int DTYPE_FIELD_NUMBER = 1;
-  private int dtype_;
-  /**
-   * <code>.tensorflow.DataType dtype = 1;</code>
-   */
-  public int getDtypeValue() {
-    return dtype_;
-  }
-  /**
-   * <code>.tensorflow.DataType dtype = 1;</code>
-   */
-  public org.tensorflow.proto.framework.DataType getDtype() {
-    @SuppressWarnings("deprecation")
-    org.tensorflow.proto.framework.DataType result = org.tensorflow.proto.framework.DataType.valueOf(dtype_);
-    return result == null ? org.tensorflow.proto.framework.DataType.UNRECOGNIZED : result;
-  }
-
-  public static final int SHAPE_FIELD_NUMBER = 2;
-  private org.tensorflow.proto.framework.TensorShapeProto shape_;
-  /**
-   * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-   */
-  public boolean hasShape() {
-    return shape_ != null;
-  }
-  /**
-   * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-   */
-  public org.tensorflow.proto.framework.TensorShapeProto getShape() {
-    return shape_ == null ? org.tensorflow.proto.framework.TensorShapeProto.getDefaultInstance() : shape_;
-  }
-  /**
-   * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-   */
-  public org.tensorflow.proto.framework.TensorShapeProtoOrBuilder getShapeOrBuilder() {
-    return getShape();
-  }
-
-  public static final int DEFAULT_VALUE_FIELD_NUMBER = 3;
-  private org.tensorflow.proto.framework.TensorProto defaultValue_;
-  /**
-   * <code>.tensorflow.TensorProto default_value = 3;</code>
-   */
-  public boolean hasDefaultValue() {
-    return defaultValue_ != null;
-  }
-  /**
-   * <code>.tensorflow.TensorProto default_value = 3;</code>
-   */
-  public org.tensorflow.proto.framework.TensorProto getDefaultValue() {
-    return defaultValue_ == null ? org.tensorflow.proto.framework.TensorProto.getDefaultInstance() : defaultValue_;
-  }
-  /**
-   * <code>.tensorflow.TensorProto default_value = 3;</code>
-   */
-  public org.tensorflow.proto.framework.TensorProtoOrBuilder getDefaultValueOrBuilder() {
-    return getDefaultValue();
-  }
-
-  public static final int VALUES_OUTPUT_TENSOR_NAME_FIELD_NUMBER = 4;
-  private volatile java.lang.Object valuesOutputTensorName_;
-  /**
-   * <code>string values_output_tensor_name = 4;</code>
-   */
-  public java.lang.String getValuesOutputTensorName() {
-    java.lang.Object ref = valuesOutputTensorName_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      valuesOutputTensorName_ = s;
-      return s;
-    }
-  }
-  /**
-   * <code>string values_output_tensor_name = 4;</code>
-   */
-  public com.google.protobuf.ByteString
-      getValuesOutputTensorNameBytes() {
-    java.lang.Object ref = valuesOutputTensorName_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      valuesOutputTensorName_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    if (dtype_ != org.tensorflow.proto.framework.DataType.DT_INVALID.getNumber()) {
-      output.writeEnum(1, dtype_);
-    }
-    if (shape_ != null) {
-      output.writeMessage(2, getShape());
-    }
-    if (defaultValue_ != null) {
-      output.writeMessage(3, getDefaultValue());
-    }
-    if (!getValuesOutputTensorNameBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 4, valuesOutputTensorName_);
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (dtype_ != org.tensorflow.proto.framework.DataType.DT_INVALID.getNumber()) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeEnumSize(1, dtype_);
-    }
-    if (shape_ != null) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(2, getShape());
-    }
-    if (defaultValue_ != null) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(3, getDefaultValue());
-    }
-    if (!getValuesOutputTensorNameBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(4, valuesOutputTensorName_);
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.example.FixedLenFeatureProto)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.example.FixedLenFeatureProto other = (org.tensorflow.proto.example.FixedLenFeatureProto) obj;
-
-    if (dtype_ != other.dtype_) return false;
-    if (hasShape() != other.hasShape()) return false;
-    if (hasShape()) {
-      if (!getShape()
-          .equals(other.getShape())) return false;
-    }
-    if (hasDefaultValue() != other.hasDefaultValue()) return false;
-    if (hasDefaultValue()) {
-      if (!getDefaultValue()
-          .equals(other.getDefaultValue())) return false;
-    }
-    if (!getValuesOutputTensorName()
-        .equals(other.getValuesOutputTensorName())) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    hash = (37 * hash) + DTYPE_FIELD_NUMBER;
-    hash = (53 * hash) + dtype_;
-    if (hasShape()) {
-      hash = (37 * hash) + SHAPE_FIELD_NUMBER;
-      hash = (53 * hash) + getShape().hashCode();
-    }
-    if (hasDefaultValue()) {
-      hash = (37 * hash) + DEFAULT_VALUE_FIELD_NUMBER;
-      hash = (53 * hash) + getDefaultValue().hashCode();
-    }
-    hash = (37 * hash) + VALUES_OUTPUT_TENSOR_NAME_FIELD_NUMBER;
-    hash = (53 * hash) + getValuesOutputTensorName().hashCode();
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.example.FixedLenFeatureProto parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.example.FixedLenFeatureProto parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.example.FixedLenFeatureProto parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.example.FixedLenFeatureProto parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.example.FixedLenFeatureProto parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.example.FixedLenFeatureProto parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.example.FixedLenFeatureProto parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.example.FixedLenFeatureProto parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.example.FixedLenFeatureProto parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.example.FixedLenFeatureProto parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.example.FixedLenFeatureProto parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.example.FixedLenFeatureProto parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.example.FixedLenFeatureProto prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * Protobuf type {@code tensorflow.FixedLenFeatureProto}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.FixedLenFeatureProto)
-      org.tensorflow.proto.example.FixedLenFeatureProtoOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.example.ExampleParserConfigurationProtos.internal_static_tensorflow_FixedLenFeatureProto_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.example.ExampleParserConfigurationProtos.internal_static_tensorflow_FixedLenFeatureProto_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.example.FixedLenFeatureProto.class, org.tensorflow.proto.example.FixedLenFeatureProto.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.example.FixedLenFeatureProto.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      dtype_ = 0;
-
-      if (shapeBuilder_ == null) {
-        shape_ = null;
-      } else {
-        shape_ = null;
-        shapeBuilder_ = null;
-      }
-      if (defaultValueBuilder_ == null) {
-        defaultValue_ = null;
-      } else {
-        defaultValue_ = null;
-        defaultValueBuilder_ = null;
-      }
-      valuesOutputTensorName_ = "";
-
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.example.ExampleParserConfigurationProtos.internal_static_tensorflow_FixedLenFeatureProto_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.example.FixedLenFeatureProto getDefaultInstanceForType() {
-      return org.tensorflow.proto.example.FixedLenFeatureProto.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.example.FixedLenFeatureProto build() {
-      org.tensorflow.proto.example.FixedLenFeatureProto result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.example.FixedLenFeatureProto buildPartial() {
-      org.tensorflow.proto.example.FixedLenFeatureProto result = new org.tensorflow.proto.example.FixedLenFeatureProto(this);
-      result.dtype_ = dtype_;
-      if (shapeBuilder_ == null) {
-        result.shape_ = shape_;
-      } else {
-        result.shape_ = shapeBuilder_.build();
-      }
-      if (defaultValueBuilder_ == null) {
-        result.defaultValue_ = defaultValue_;
-      } else {
-        result.defaultValue_ = defaultValueBuilder_.build();
-      }
-      result.valuesOutputTensorName_ = valuesOutputTensorName_;
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.example.FixedLenFeatureProto) {
-        return mergeFrom((org.tensorflow.proto.example.FixedLenFeatureProto)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.example.FixedLenFeatureProto other) {
-      if (other == org.tensorflow.proto.example.FixedLenFeatureProto.getDefaultInstance()) return this;
-      if (other.dtype_ != 0) {
-        setDtypeValue(other.getDtypeValue());
-      }
-      if (other.hasShape()) {
-        mergeShape(other.getShape());
-      }
-      if (other.hasDefaultValue()) {
-        mergeDefaultValue(other.getDefaultValue());
-      }
-      if (!other.getValuesOutputTensorName().isEmpty()) {
-        valuesOutputTensorName_ = other.valuesOutputTensorName_;
-        onChanged();
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.example.FixedLenFeatureProto parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.example.FixedLenFeatureProto) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-
-    private int dtype_ = 0;
-    /**
-     * <code>.tensorflow.DataType dtype = 1;</code>
-     */
-    public int getDtypeValue() {
-      return dtype_;
-    }
-    /**
-     * <code>.tensorflow.DataType dtype = 1;</code>
-     */
-    public Builder setDtypeValue(int value) {
-      dtype_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>.tensorflow.DataType dtype = 1;</code>
-     */
-    public org.tensorflow.proto.framework.DataType getDtype() {
-      @SuppressWarnings("deprecation")
-      org.tensorflow.proto.framework.DataType result = org.tensorflow.proto.framework.DataType.valueOf(dtype_);
-      return result == null ? org.tensorflow.proto.framework.DataType.UNRECOGNIZED : result;
-    }
-    /**
-     * <code>.tensorflow.DataType dtype = 1;</code>
-     */
-    public Builder setDtype(org.tensorflow.proto.framework.DataType value) {
-      if (value == null) {
-        throw new NullPointerException();
-      }
-      
-      dtype_ = value.getNumber();
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>.tensorflow.DataType dtype = 1;</code>
-     */
-    public Builder clearDtype() {
-      
-      dtype_ = 0;
-      onChanged();
-      return this;
-    }
-
-    private org.tensorflow.proto.framework.TensorShapeProto shape_;
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.TensorShapeProto, org.tensorflow.proto.framework.TensorShapeProto.Builder, org.tensorflow.proto.framework.TensorShapeProtoOrBuilder> shapeBuilder_;
-    /**
-     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-     */
-    public boolean hasShape() {
-      return shapeBuilder_ != null || shape_ != null;
-    }
-    /**
-     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-     */
-    public org.tensorflow.proto.framework.TensorShapeProto getShape() {
-      if (shapeBuilder_ == null) {
-        return shape_ == null ? org.tensorflow.proto.framework.TensorShapeProto.getDefaultInstance() : shape_;
-      } else {
-        return shapeBuilder_.getMessage();
-      }
-    }
-    /**
-     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-     */
-    public Builder setShape(org.tensorflow.proto.framework.TensorShapeProto value) {
-      if (shapeBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        shape_ = value;
-        onChanged();
-      } else {
-        shapeBuilder_.setMessage(value);
-      }
-
-      return this;
-    }
-    /**
-     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-     */
-    public Builder setShape(
-        org.tensorflow.proto.framework.TensorShapeProto.Builder builderForValue) {
-      if (shapeBuilder_ == null) {
-        shape_ = builderForValue.build();
-        onChanged();
-      } else {
-        shapeBuilder_.setMessage(builderForValue.build());
-      }
-
-      return this;
-    }
-    /**
-     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-     */
-    public Builder mergeShape(org.tensorflow.proto.framework.TensorShapeProto value) {
-      if (shapeBuilder_ == null) {
-        if (shape_ != null) {
-          shape_ =
-            org.tensorflow.proto.framework.TensorShapeProto.newBuilder(shape_).mergeFrom(value).buildPartial();
-        } else {
-          shape_ = value;
-        }
-        onChanged();
-      } else {
-        shapeBuilder_.mergeFrom(value);
-      }
-
-      return this;
-    }
-    /**
-     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-     */
-    public Builder clearShape() {
-      if (shapeBuilder_ == null) {
-        shape_ = null;
-        onChanged();
-      } else {
-        shape_ = null;
-        shapeBuilder_ = null;
-      }
-
-      return this;
-    }
-    /**
-     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-     */
-    public org.tensorflow.proto.framework.TensorShapeProto.Builder getShapeBuilder() {
-      
-      onChanged();
-      return getShapeFieldBuilder().getBuilder();
-    }
-    /**
-     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-     */
-    public org.tensorflow.proto.framework.TensorShapeProtoOrBuilder getShapeOrBuilder() {
-      if (shapeBuilder_ != null) {
-        return shapeBuilder_.getMessageOrBuilder();
-      } else {
-        return shape_ == null ?
-            org.tensorflow.proto.framework.TensorShapeProto.getDefaultInstance() : shape_;
-      }
-    }
-    /**
-     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-     */
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.TensorShapeProto, org.tensorflow.proto.framework.TensorShapeProto.Builder, org.tensorflow.proto.framework.TensorShapeProtoOrBuilder> 
-        getShapeFieldBuilder() {
-      if (shapeBuilder_ == null) {
-        shapeBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-            org.tensorflow.proto.framework.TensorShapeProto, org.tensorflow.proto.framework.TensorShapeProto.Builder, org.tensorflow.proto.framework.TensorShapeProtoOrBuilder>(
-                getShape(),
-                getParentForChildren(),
-                isClean());
-        shape_ = null;
-      }
-      return shapeBuilder_;
-    }
-
-    private org.tensorflow.proto.framework.TensorProto defaultValue_;
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.TensorProto, org.tensorflow.proto.framework.TensorProto.Builder, org.tensorflow.proto.framework.TensorProtoOrBuilder> defaultValueBuilder_;
-    /**
-     * <code>.tensorflow.TensorProto default_value = 3;</code>
-     */
-    public boolean hasDefaultValue() {
-      return defaultValueBuilder_ != null || defaultValue_ != null;
-    }
-    /**
-     * <code>.tensorflow.TensorProto default_value = 3;</code>
-     */
-    public org.tensorflow.proto.framework.TensorProto getDefaultValue() {
-      if (defaultValueBuilder_ == null) {
-        return defaultValue_ == null ? org.tensorflow.proto.framework.TensorProto.getDefaultInstance() : defaultValue_;
-      } else {
-        return defaultValueBuilder_.getMessage();
-      }
-    }
-    /**
-     * <code>.tensorflow.TensorProto default_value = 3;</code>
-     */
-    public Builder setDefaultValue(org.tensorflow.proto.framework.TensorProto value) {
-      if (defaultValueBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        defaultValue_ = value;
-        onChanged();
-      } else {
-        defaultValueBuilder_.setMessage(value);
-      }
-
-      return this;
-    }
-    /**
-     * <code>.tensorflow.TensorProto default_value = 3;</code>
-     */
-    public Builder setDefaultValue(
-        org.tensorflow.proto.framework.TensorProto.Builder builderForValue) {
-      if (defaultValueBuilder_ == null) {
-        defaultValue_ = builderForValue.build();
-        onChanged();
-      } else {
-        defaultValueBuilder_.setMessage(builderForValue.build());
-      }
-
-      return this;
-    }
-    /**
-     * <code>.tensorflow.TensorProto default_value = 3;</code>
-     */
-    public Builder mergeDefaultValue(org.tensorflow.proto.framework.TensorProto value) {
-      if (defaultValueBuilder_ == null) {
-        if (defaultValue_ != null) {
-          defaultValue_ =
-            org.tensorflow.proto.framework.TensorProto.newBuilder(defaultValue_).mergeFrom(value).buildPartial();
-        } else {
-          defaultValue_ = value;
-        }
-        onChanged();
-      } else {
-        defaultValueBuilder_.mergeFrom(value);
-      }
-
-      return this;
-    }
-    /**
-     * <code>.tensorflow.TensorProto default_value = 3;</code>
-     */
-    public Builder clearDefaultValue() {
-      if (defaultValueBuilder_ == null) {
-        defaultValue_ = null;
-        onChanged();
-      } else {
-        defaultValue_ = null;
-        defaultValueBuilder_ = null;
-      }
-
-      return this;
-    }
-    /**
-     * <code>.tensorflow.TensorProto default_value = 3;</code>
-     */
-    public org.tensorflow.proto.framework.TensorProto.Builder getDefaultValueBuilder() {
-      
-      onChanged();
-      return getDefaultValueFieldBuilder().getBuilder();
-    }
-    /**
-     * <code>.tensorflow.TensorProto default_value = 3;</code>
-     */
-    public org.tensorflow.proto.framework.TensorProtoOrBuilder getDefaultValueOrBuilder() {
-      if (defaultValueBuilder_ != null) {
-        return defaultValueBuilder_.getMessageOrBuilder();
-      } else {
-        return defaultValue_ == null ?
-            org.tensorflow.proto.framework.TensorProto.getDefaultInstance() : defaultValue_;
-      }
-    }
-    /**
-     * <code>.tensorflow.TensorProto default_value = 3;</code>
-     */
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.TensorProto, org.tensorflow.proto.framework.TensorProto.Builder, org.tensorflow.proto.framework.TensorProtoOrBuilder> 
-        getDefaultValueFieldBuilder() {
-      if (defaultValueBuilder_ == null) {
-        defaultValueBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-            org.tensorflow.proto.framework.TensorProto, org.tensorflow.proto.framework.TensorProto.Builder, org.tensorflow.proto.framework.TensorProtoOrBuilder>(
-                getDefaultValue(),
-                getParentForChildren(),
-                isClean());
-        defaultValue_ = null;
-      }
-      return defaultValueBuilder_;
-    }
-
-    private java.lang.Object valuesOutputTensorName_ = "";
-    /**
-     * <code>string values_output_tensor_name = 4;</code>
-     */
-    public java.lang.String getValuesOutputTensorName() {
-      java.lang.Object ref = valuesOutputTensorName_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        valuesOutputTensorName_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <code>string values_output_tensor_name = 4;</code>
-     */
-    public com.google.protobuf.ByteString
-        getValuesOutputTensorNameBytes() {
-      java.lang.Object ref = valuesOutputTensorName_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        valuesOutputTensorName_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <code>string values_output_tensor_name = 4;</code>
-     */
-    public Builder setValuesOutputTensorName(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      valuesOutputTensorName_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>string values_output_tensor_name = 4;</code>
-     */
-    public Builder clearValuesOutputTensorName() {
-      
-      valuesOutputTensorName_ = getDefaultInstance().getValuesOutputTensorName();
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>string values_output_tensor_name = 4;</code>
-     */
-    public Builder setValuesOutputTensorNameBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      valuesOutputTensorName_ = value;
-      onChanged();
-      return this;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.FixedLenFeatureProto)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.FixedLenFeatureProto)
-  private static final org.tensorflow.proto.example.FixedLenFeatureProto DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.example.FixedLenFeatureProto();
-  }
-
-  public static org.tensorflow.proto.example.FixedLenFeatureProto getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<FixedLenFeatureProto>
-      PARSER = new com.google.protobuf.AbstractParser<FixedLenFeatureProto>() {
-    @java.lang.Override
-    public FixedLenFeatureProto parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new FixedLenFeatureProto(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<FixedLenFeatureProto> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<FixedLenFeatureProto> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.example.FixedLenFeatureProto getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/FixedLenFeatureProtoOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/FixedLenFeatureProtoOrBuilder.java
deleted file mode 100644
index 0008b29583f..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/FixedLenFeatureProtoOrBuilder.java
+++ /dev/null
@@ -1,54 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/example/example_parser_configuration.proto
-
-package org.tensorflow.proto.example;
-
-public interface FixedLenFeatureProtoOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.FixedLenFeatureProto)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <code>.tensorflow.DataType dtype = 1;</code>
-   */
-  int getDtypeValue();
-  /**
-   * <code>.tensorflow.DataType dtype = 1;</code>
-   */
-  org.tensorflow.proto.framework.DataType getDtype();
-
-  /**
-   * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-   */
-  boolean hasShape();
-  /**
-   * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-   */
-  org.tensorflow.proto.framework.TensorShapeProto getShape();
-  /**
-   * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-   */
-  org.tensorflow.proto.framework.TensorShapeProtoOrBuilder getShapeOrBuilder();
-
-  /**
-   * <code>.tensorflow.TensorProto default_value = 3;</code>
-   */
-  boolean hasDefaultValue();
-  /**
-   * <code>.tensorflow.TensorProto default_value = 3;</code>
-   */
-  org.tensorflow.proto.framework.TensorProto getDefaultValue();
-  /**
-   * <code>.tensorflow.TensorProto default_value = 3;</code>
-   */
-  org.tensorflow.proto.framework.TensorProtoOrBuilder getDefaultValueOrBuilder();
-
-  /**
-   * <code>string values_output_tensor_name = 4;</code>
-   */
-  java.lang.String getValuesOutputTensorName();
-  /**
-   * <code>string values_output_tensor_name = 4;</code>
-   */
-  com.google.protobuf.ByteString
-      getValuesOutputTensorNameBytes();
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/FloatList.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/FloatList.java
deleted file mode 100644
index efad9ee2fb6..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/FloatList.java
+++ /dev/null
@@ -1,579 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/example/feature.proto
-
-package org.tensorflow.proto.example;
-
-/**
- * Protobuf type {@code tensorflow.FloatList}
- */
-public  final class FloatList extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.FloatList)
-    FloatListOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use FloatList.newBuilder() to construct.
-  private FloatList(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private FloatList() {
-    value_ = emptyFloatList();
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new FloatList();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private FloatList(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    int mutable_bitField0_ = 0;
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 13: {
-            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
-              value_ = newFloatList();
-              mutable_bitField0_ |= 0x00000001;
-            }
-            value_.addFloat(input.readFloat());
-            break;
-          }
-          case 10: {
-            int length = input.readRawVarint32();
-            int limit = input.pushLimit(length);
-            if (!((mutable_bitField0_ & 0x00000001) != 0) && input.getBytesUntilLimit() > 0) {
-              value_ = newFloatList();
-              mutable_bitField0_ |= 0x00000001;
-            }
-            while (input.getBytesUntilLimit() > 0) {
-              value_.addFloat(input.readFloat());
-            }
-            input.popLimit(limit);
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      if (((mutable_bitField0_ & 0x00000001) != 0)) {
-        value_.makeImmutable(); // C
-      }
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.example.FeatureProtos.internal_static_tensorflow_FloatList_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.example.FeatureProtos.internal_static_tensorflow_FloatList_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.example.FloatList.class, org.tensorflow.proto.example.FloatList.Builder.class);
-  }
-
-  public static final int VALUE_FIELD_NUMBER = 1;
-  private com.google.protobuf.Internal.FloatList value_;
-  /**
-   * <code>repeated float value = 1 [packed = true];</code>
-   */
-  public java.util.List<java.lang.Float>
-      getValueList() {
-    return value_;
-  }
-  /**
-   * <code>repeated float value = 1 [packed = true];</code>
-   */
-  public int getValueCount() {
-    return value_.size();
-  }
-  /**
-   * <code>repeated float value = 1 [packed = true];</code>
-   */
-  public float getValue(int index) {
-    return value_.getFloat(index);
-  }
-  private int valueMemoizedSerializedSize = -1;
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    getSerializedSize();
-    if (getValueList().size() > 0) {
-      output.writeUInt32NoTag(10);
-      output.writeUInt32NoTag(valueMemoizedSerializedSize);
-    }
-    for (int i = 0; i < value_.size(); i++) {
-      output.writeFloatNoTag(value_.getFloat(i));
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    {
-      int dataSize = 0;
-      dataSize = 4 * getValueList().size();
-      size += dataSize;
-      if (!getValueList().isEmpty()) {
-        size += 1;
-        size += com.google.protobuf.CodedOutputStream
-            .computeInt32SizeNoTag(dataSize);
-      }
-      valueMemoizedSerializedSize = dataSize;
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.example.FloatList)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.example.FloatList other = (org.tensorflow.proto.example.FloatList) obj;
-
-    if (!getValueList()
-        .equals(other.getValueList())) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    if (getValueCount() > 0) {
-      hash = (37 * hash) + VALUE_FIELD_NUMBER;
-      hash = (53 * hash) + getValueList().hashCode();
-    }
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.example.FloatList parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.example.FloatList parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.example.FloatList parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.example.FloatList parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.example.FloatList parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.example.FloatList parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.example.FloatList parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.example.FloatList parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.example.FloatList parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.example.FloatList parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.example.FloatList parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.example.FloatList parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.example.FloatList prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * Protobuf type {@code tensorflow.FloatList}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.FloatList)
-      org.tensorflow.proto.example.FloatListOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.example.FeatureProtos.internal_static_tensorflow_FloatList_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.example.FeatureProtos.internal_static_tensorflow_FloatList_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.example.FloatList.class, org.tensorflow.proto.example.FloatList.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.example.FloatList.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      value_ = emptyFloatList();
-      bitField0_ = (bitField0_ & ~0x00000001);
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.example.FeatureProtos.internal_static_tensorflow_FloatList_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.example.FloatList getDefaultInstanceForType() {
-      return org.tensorflow.proto.example.FloatList.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.example.FloatList build() {
-      org.tensorflow.proto.example.FloatList result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.example.FloatList buildPartial() {
-      org.tensorflow.proto.example.FloatList result = new org.tensorflow.proto.example.FloatList(this);
-      int from_bitField0_ = bitField0_;
-      if (((bitField0_ & 0x00000001) != 0)) {
-        value_.makeImmutable();
-        bitField0_ = (bitField0_ & ~0x00000001);
-      }
-      result.value_ = value_;
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.example.FloatList) {
-        return mergeFrom((org.tensorflow.proto.example.FloatList)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.example.FloatList other) {
-      if (other == org.tensorflow.proto.example.FloatList.getDefaultInstance()) return this;
-      if (!other.value_.isEmpty()) {
-        if (value_.isEmpty()) {
-          value_ = other.value_;
-          bitField0_ = (bitField0_ & ~0x00000001);
-        } else {
-          ensureValueIsMutable();
-          value_.addAll(other.value_);
-        }
-        onChanged();
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.example.FloatList parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.example.FloatList) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-    private int bitField0_;
-
-    private com.google.protobuf.Internal.FloatList value_ = emptyFloatList();
-    private void ensureValueIsMutable() {
-      if (!((bitField0_ & 0x00000001) != 0)) {
-        value_ = mutableCopy(value_);
-        bitField0_ |= 0x00000001;
-       }
-    }
-    /**
-     * <code>repeated float value = 1 [packed = true];</code>
-     */
-    public java.util.List<java.lang.Float>
-        getValueList() {
-      return ((bitField0_ & 0x00000001) != 0) ?
-               java.util.Collections.unmodifiableList(value_) : value_;
-    }
-    /**
-     * <code>repeated float value = 1 [packed = true];</code>
-     */
-    public int getValueCount() {
-      return value_.size();
-    }
-    /**
-     * <code>repeated float value = 1 [packed = true];</code>
-     */
-    public float getValue(int index) {
-      return value_.getFloat(index);
-    }
-    /**
-     * <code>repeated float value = 1 [packed = true];</code>
-     */
-    public Builder setValue(
-        int index, float value) {
-      ensureValueIsMutable();
-      value_.setFloat(index, value);
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>repeated float value = 1 [packed = true];</code>
-     */
-    public Builder addValue(float value) {
-      ensureValueIsMutable();
-      value_.addFloat(value);
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>repeated float value = 1 [packed = true];</code>
-     */
-    public Builder addAllValue(
-        java.lang.Iterable<? extends java.lang.Float> values) {
-      ensureValueIsMutable();
-      com.google.protobuf.AbstractMessageLite.Builder.addAll(
-          values, value_);
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>repeated float value = 1 [packed = true];</code>
-     */
-    public Builder clearValue() {
-      value_ = emptyFloatList();
-      bitField0_ = (bitField0_ & ~0x00000001);
-      onChanged();
-      return this;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.FloatList)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.FloatList)
-  private static final org.tensorflow.proto.example.FloatList DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.example.FloatList();
-  }
-
-  public static org.tensorflow.proto.example.FloatList getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<FloatList>
-      PARSER = new com.google.protobuf.AbstractParser<FloatList>() {
-    @java.lang.Override
-    public FloatList parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new FloatList(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<FloatList> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<FloatList> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.example.FloatList getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/FloatListOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/FloatListOrBuilder.java
deleted file mode 100644
index 65856cd5f4e..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/FloatListOrBuilder.java
+++ /dev/null
@@ -1,22 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/example/feature.proto
-
-package org.tensorflow.proto.example;
-
-public interface FloatListOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.FloatList)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <code>repeated float value = 1 [packed = true];</code>
-   */
-  java.util.List<java.lang.Float> getValueList();
-  /**
-   * <code>repeated float value = 1 [packed = true];</code>
-   */
-  int getValueCount();
-  /**
-   * <code>repeated float value = 1 [packed = true];</code>
-   */
-  float getValue(int index);
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/Int64List.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/Int64List.java
deleted file mode 100644
index 1cf4f104182..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/Int64List.java
+++ /dev/null
@@ -1,582 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/example/feature.proto
-
-package org.tensorflow.proto.example;
-
-/**
- * Protobuf type {@code tensorflow.Int64List}
- */
-public  final class Int64List extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.Int64List)
-    Int64ListOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use Int64List.newBuilder() to construct.
-  private Int64List(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private Int64List() {
-    value_ = emptyLongList();
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new Int64List();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private Int64List(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    int mutable_bitField0_ = 0;
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 8: {
-            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
-              value_ = newLongList();
-              mutable_bitField0_ |= 0x00000001;
-            }
-            value_.addLong(input.readInt64());
-            break;
-          }
-          case 10: {
-            int length = input.readRawVarint32();
-            int limit = input.pushLimit(length);
-            if (!((mutable_bitField0_ & 0x00000001) != 0) && input.getBytesUntilLimit() > 0) {
-              value_ = newLongList();
-              mutable_bitField0_ |= 0x00000001;
-            }
-            while (input.getBytesUntilLimit() > 0) {
-              value_.addLong(input.readInt64());
-            }
-            input.popLimit(limit);
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      if (((mutable_bitField0_ & 0x00000001) != 0)) {
-        value_.makeImmutable(); // C
-      }
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.example.FeatureProtos.internal_static_tensorflow_Int64List_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.example.FeatureProtos.internal_static_tensorflow_Int64List_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.example.Int64List.class, org.tensorflow.proto.example.Int64List.Builder.class);
-  }
-
-  public static final int VALUE_FIELD_NUMBER = 1;
-  private com.google.protobuf.Internal.LongList value_;
-  /**
-   * <code>repeated int64 value = 1 [packed = true];</code>
-   */
-  public java.util.List<java.lang.Long>
-      getValueList() {
-    return value_;
-  }
-  /**
-   * <code>repeated int64 value = 1 [packed = true];</code>
-   */
-  public int getValueCount() {
-    return value_.size();
-  }
-  /**
-   * <code>repeated int64 value = 1 [packed = true];</code>
-   */
-  public long getValue(int index) {
-    return value_.getLong(index);
-  }
-  private int valueMemoizedSerializedSize = -1;
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    getSerializedSize();
-    if (getValueList().size() > 0) {
-      output.writeUInt32NoTag(10);
-      output.writeUInt32NoTag(valueMemoizedSerializedSize);
-    }
-    for (int i = 0; i < value_.size(); i++) {
-      output.writeInt64NoTag(value_.getLong(i));
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    {
-      int dataSize = 0;
-      for (int i = 0; i < value_.size(); i++) {
-        dataSize += com.google.protobuf.CodedOutputStream
-          .computeInt64SizeNoTag(value_.getLong(i));
-      }
-      size += dataSize;
-      if (!getValueList().isEmpty()) {
-        size += 1;
-        size += com.google.protobuf.CodedOutputStream
-            .computeInt32SizeNoTag(dataSize);
-      }
-      valueMemoizedSerializedSize = dataSize;
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.example.Int64List)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.example.Int64List other = (org.tensorflow.proto.example.Int64List) obj;
-
-    if (!getValueList()
-        .equals(other.getValueList())) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    if (getValueCount() > 0) {
-      hash = (37 * hash) + VALUE_FIELD_NUMBER;
-      hash = (53 * hash) + getValueList().hashCode();
-    }
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.example.Int64List parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.example.Int64List parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.example.Int64List parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.example.Int64List parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.example.Int64List parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.example.Int64List parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.example.Int64List parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.example.Int64List parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.example.Int64List parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.example.Int64List parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.example.Int64List parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.example.Int64List parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.example.Int64List prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * Protobuf type {@code tensorflow.Int64List}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.Int64List)
-      org.tensorflow.proto.example.Int64ListOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.example.FeatureProtos.internal_static_tensorflow_Int64List_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.example.FeatureProtos.internal_static_tensorflow_Int64List_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.example.Int64List.class, org.tensorflow.proto.example.Int64List.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.example.Int64List.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      value_ = emptyLongList();
-      bitField0_ = (bitField0_ & ~0x00000001);
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.example.FeatureProtos.internal_static_tensorflow_Int64List_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.example.Int64List getDefaultInstanceForType() {
-      return org.tensorflow.proto.example.Int64List.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.example.Int64List build() {
-      org.tensorflow.proto.example.Int64List result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.example.Int64List buildPartial() {
-      org.tensorflow.proto.example.Int64List result = new org.tensorflow.proto.example.Int64List(this);
-      int from_bitField0_ = bitField0_;
-      if (((bitField0_ & 0x00000001) != 0)) {
-        value_.makeImmutable();
-        bitField0_ = (bitField0_ & ~0x00000001);
-      }
-      result.value_ = value_;
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.example.Int64List) {
-        return mergeFrom((org.tensorflow.proto.example.Int64List)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.example.Int64List other) {
-      if (other == org.tensorflow.proto.example.Int64List.getDefaultInstance()) return this;
-      if (!other.value_.isEmpty()) {
-        if (value_.isEmpty()) {
-          value_ = other.value_;
-          bitField0_ = (bitField0_ & ~0x00000001);
-        } else {
-          ensureValueIsMutable();
-          value_.addAll(other.value_);
-        }
-        onChanged();
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.example.Int64List parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.example.Int64List) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-    private int bitField0_;
-
-    private com.google.protobuf.Internal.LongList value_ = emptyLongList();
-    private void ensureValueIsMutable() {
-      if (!((bitField0_ & 0x00000001) != 0)) {
-        value_ = mutableCopy(value_);
-        bitField0_ |= 0x00000001;
-       }
-    }
-    /**
-     * <code>repeated int64 value = 1 [packed = true];</code>
-     */
-    public java.util.List<java.lang.Long>
-        getValueList() {
-      return ((bitField0_ & 0x00000001) != 0) ?
-               java.util.Collections.unmodifiableList(value_) : value_;
-    }
-    /**
-     * <code>repeated int64 value = 1 [packed = true];</code>
-     */
-    public int getValueCount() {
-      return value_.size();
-    }
-    /**
-     * <code>repeated int64 value = 1 [packed = true];</code>
-     */
-    public long getValue(int index) {
-      return value_.getLong(index);
-    }
-    /**
-     * <code>repeated int64 value = 1 [packed = true];</code>
-     */
-    public Builder setValue(
-        int index, long value) {
-      ensureValueIsMutable();
-      value_.setLong(index, value);
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>repeated int64 value = 1 [packed = true];</code>
-     */
-    public Builder addValue(long value) {
-      ensureValueIsMutable();
-      value_.addLong(value);
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>repeated int64 value = 1 [packed = true];</code>
-     */
-    public Builder addAllValue(
-        java.lang.Iterable<? extends java.lang.Long> values) {
-      ensureValueIsMutable();
-      com.google.protobuf.AbstractMessageLite.Builder.addAll(
-          values, value_);
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>repeated int64 value = 1 [packed = true];</code>
-     */
-    public Builder clearValue() {
-      value_ = emptyLongList();
-      bitField0_ = (bitField0_ & ~0x00000001);
-      onChanged();
-      return this;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.Int64List)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.Int64List)
-  private static final org.tensorflow.proto.example.Int64List DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.example.Int64List();
-  }
-
-  public static org.tensorflow.proto.example.Int64List getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<Int64List>
-      PARSER = new com.google.protobuf.AbstractParser<Int64List>() {
-    @java.lang.Override
-    public Int64List parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new Int64List(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<Int64List> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<Int64List> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.example.Int64List getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/Int64ListOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/Int64ListOrBuilder.java
deleted file mode 100644
index 6ca9f168a7d..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/Int64ListOrBuilder.java
+++ /dev/null
@@ -1,22 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/example/feature.proto
-
-package org.tensorflow.proto.example;
-
-public interface Int64ListOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.Int64List)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <code>repeated int64 value = 1 [packed = true];</code>
-   */
-  java.util.List<java.lang.Long> getValueList();
-  /**
-   * <code>repeated int64 value = 1 [packed = true];</code>
-   */
-  int getValueCount();
-  /**
-   * <code>repeated int64 value = 1 [packed = true];</code>
-   */
-  long getValue(int index);
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/SequenceExample.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/SequenceExample.java
deleted file mode 100644
index b88e6f12e34..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/SequenceExample.java
+++ /dev/null
@@ -1,781 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/example/example.proto
-
-package org.tensorflow.proto.example;
-
-/**
- * Protobuf type {@code tensorflow.SequenceExample}
- */
-public  final class SequenceExample extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.SequenceExample)
-    SequenceExampleOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use SequenceExample.newBuilder() to construct.
-  private SequenceExample(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private SequenceExample() {
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new SequenceExample();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private SequenceExample(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 10: {
-            org.tensorflow.proto.example.Features.Builder subBuilder = null;
-            if (context_ != null) {
-              subBuilder = context_.toBuilder();
-            }
-            context_ = input.readMessage(org.tensorflow.proto.example.Features.parser(), extensionRegistry);
-            if (subBuilder != null) {
-              subBuilder.mergeFrom(context_);
-              context_ = subBuilder.buildPartial();
-            }
-
-            break;
-          }
-          case 18: {
-            org.tensorflow.proto.example.FeatureLists.Builder subBuilder = null;
-            if (featureLists_ != null) {
-              subBuilder = featureLists_.toBuilder();
-            }
-            featureLists_ = input.readMessage(org.tensorflow.proto.example.FeatureLists.parser(), extensionRegistry);
-            if (subBuilder != null) {
-              subBuilder.mergeFrom(featureLists_);
-              featureLists_ = subBuilder.buildPartial();
-            }
-
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.example.ExampleProtos.internal_static_tensorflow_SequenceExample_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.example.ExampleProtos.internal_static_tensorflow_SequenceExample_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.example.SequenceExample.class, org.tensorflow.proto.example.SequenceExample.Builder.class);
-  }
-
-  public static final int CONTEXT_FIELD_NUMBER = 1;
-  private org.tensorflow.proto.example.Features context_;
-  /**
-   * <code>.tensorflow.Features context = 1;</code>
-   */
-  public boolean hasContext() {
-    return context_ != null;
-  }
-  /**
-   * <code>.tensorflow.Features context = 1;</code>
-   */
-  public org.tensorflow.proto.example.Features getContext() {
-    return context_ == null ? org.tensorflow.proto.example.Features.getDefaultInstance() : context_;
-  }
-  /**
-   * <code>.tensorflow.Features context = 1;</code>
-   */
-  public org.tensorflow.proto.example.FeaturesOrBuilder getContextOrBuilder() {
-    return getContext();
-  }
-
-  public static final int FEATURE_LISTS_FIELD_NUMBER = 2;
-  private org.tensorflow.proto.example.FeatureLists featureLists_;
-  /**
-   * <code>.tensorflow.FeatureLists feature_lists = 2;</code>
-   */
-  public boolean hasFeatureLists() {
-    return featureLists_ != null;
-  }
-  /**
-   * <code>.tensorflow.FeatureLists feature_lists = 2;</code>
-   */
-  public org.tensorflow.proto.example.FeatureLists getFeatureLists() {
-    return featureLists_ == null ? org.tensorflow.proto.example.FeatureLists.getDefaultInstance() : featureLists_;
-  }
-  /**
-   * <code>.tensorflow.FeatureLists feature_lists = 2;</code>
-   */
-  public org.tensorflow.proto.example.FeatureListsOrBuilder getFeatureListsOrBuilder() {
-    return getFeatureLists();
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    if (context_ != null) {
-      output.writeMessage(1, getContext());
-    }
-    if (featureLists_ != null) {
-      output.writeMessage(2, getFeatureLists());
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (context_ != null) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(1, getContext());
-    }
-    if (featureLists_ != null) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(2, getFeatureLists());
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.example.SequenceExample)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.example.SequenceExample other = (org.tensorflow.proto.example.SequenceExample) obj;
-
-    if (hasContext() != other.hasContext()) return false;
-    if (hasContext()) {
-      if (!getContext()
-          .equals(other.getContext())) return false;
-    }
-    if (hasFeatureLists() != other.hasFeatureLists()) return false;
-    if (hasFeatureLists()) {
-      if (!getFeatureLists()
-          .equals(other.getFeatureLists())) return false;
-    }
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    if (hasContext()) {
-      hash = (37 * hash) + CONTEXT_FIELD_NUMBER;
-      hash = (53 * hash) + getContext().hashCode();
-    }
-    if (hasFeatureLists()) {
-      hash = (37 * hash) + FEATURE_LISTS_FIELD_NUMBER;
-      hash = (53 * hash) + getFeatureLists().hashCode();
-    }
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.example.SequenceExample parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.example.SequenceExample parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.example.SequenceExample parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.example.SequenceExample parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.example.SequenceExample parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.example.SequenceExample parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.example.SequenceExample parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.example.SequenceExample parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.example.SequenceExample parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.example.SequenceExample parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.example.SequenceExample parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.example.SequenceExample parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.example.SequenceExample prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * Protobuf type {@code tensorflow.SequenceExample}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.SequenceExample)
-      org.tensorflow.proto.example.SequenceExampleOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.example.ExampleProtos.internal_static_tensorflow_SequenceExample_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.example.ExampleProtos.internal_static_tensorflow_SequenceExample_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.example.SequenceExample.class, org.tensorflow.proto.example.SequenceExample.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.example.SequenceExample.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      if (contextBuilder_ == null) {
-        context_ = null;
-      } else {
-        context_ = null;
-        contextBuilder_ = null;
-      }
-      if (featureListsBuilder_ == null) {
-        featureLists_ = null;
-      } else {
-        featureLists_ = null;
-        featureListsBuilder_ = null;
-      }
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.example.ExampleProtos.internal_static_tensorflow_SequenceExample_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.example.SequenceExample getDefaultInstanceForType() {
-      return org.tensorflow.proto.example.SequenceExample.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.example.SequenceExample build() {
-      org.tensorflow.proto.example.SequenceExample result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.example.SequenceExample buildPartial() {
-      org.tensorflow.proto.example.SequenceExample result = new org.tensorflow.proto.example.SequenceExample(this);
-      if (contextBuilder_ == null) {
-        result.context_ = context_;
-      } else {
-        result.context_ = contextBuilder_.build();
-      }
-      if (featureListsBuilder_ == null) {
-        result.featureLists_ = featureLists_;
-      } else {
-        result.featureLists_ = featureListsBuilder_.build();
-      }
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.example.SequenceExample) {
-        return mergeFrom((org.tensorflow.proto.example.SequenceExample)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.example.SequenceExample other) {
-      if (other == org.tensorflow.proto.example.SequenceExample.getDefaultInstance()) return this;
-      if (other.hasContext()) {
-        mergeContext(other.getContext());
-      }
-      if (other.hasFeatureLists()) {
-        mergeFeatureLists(other.getFeatureLists());
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.example.SequenceExample parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.example.SequenceExample) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-
-    private org.tensorflow.proto.example.Features context_;
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.example.Features, org.tensorflow.proto.example.Features.Builder, org.tensorflow.proto.example.FeaturesOrBuilder> contextBuilder_;
-    /**
-     * <code>.tensorflow.Features context = 1;</code>
-     */
-    public boolean hasContext() {
-      return contextBuilder_ != null || context_ != null;
-    }
-    /**
-     * <code>.tensorflow.Features context = 1;</code>
-     */
-    public org.tensorflow.proto.example.Features getContext() {
-      if (contextBuilder_ == null) {
-        return context_ == null ? org.tensorflow.proto.example.Features.getDefaultInstance() : context_;
-      } else {
-        return contextBuilder_.getMessage();
-      }
-    }
-    /**
-     * <code>.tensorflow.Features context = 1;</code>
-     */
-    public Builder setContext(org.tensorflow.proto.example.Features value) {
-      if (contextBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        context_ = value;
-        onChanged();
-      } else {
-        contextBuilder_.setMessage(value);
-      }
-
-      return this;
-    }
-    /**
-     * <code>.tensorflow.Features context = 1;</code>
-     */
-    public Builder setContext(
-        org.tensorflow.proto.example.Features.Builder builderForValue) {
-      if (contextBuilder_ == null) {
-        context_ = builderForValue.build();
-        onChanged();
-      } else {
-        contextBuilder_.setMessage(builderForValue.build());
-      }
-
-      return this;
-    }
-    /**
-     * <code>.tensorflow.Features context = 1;</code>
-     */
-    public Builder mergeContext(org.tensorflow.proto.example.Features value) {
-      if (contextBuilder_ == null) {
-        if (context_ != null) {
-          context_ =
-            org.tensorflow.proto.example.Features.newBuilder(context_).mergeFrom(value).buildPartial();
-        } else {
-          context_ = value;
-        }
-        onChanged();
-      } else {
-        contextBuilder_.mergeFrom(value);
-      }
-
-      return this;
-    }
-    /**
-     * <code>.tensorflow.Features context = 1;</code>
-     */
-    public Builder clearContext() {
-      if (contextBuilder_ == null) {
-        context_ = null;
-        onChanged();
-      } else {
-        context_ = null;
-        contextBuilder_ = null;
-      }
-
-      return this;
-    }
-    /**
-     * <code>.tensorflow.Features context = 1;</code>
-     */
-    public org.tensorflow.proto.example.Features.Builder getContextBuilder() {
-      
-      onChanged();
-      return getContextFieldBuilder().getBuilder();
-    }
-    /**
-     * <code>.tensorflow.Features context = 1;</code>
-     */
-    public org.tensorflow.proto.example.FeaturesOrBuilder getContextOrBuilder() {
-      if (contextBuilder_ != null) {
-        return contextBuilder_.getMessageOrBuilder();
-      } else {
-        return context_ == null ?
-            org.tensorflow.proto.example.Features.getDefaultInstance() : context_;
-      }
-    }
-    /**
-     * <code>.tensorflow.Features context = 1;</code>
-     */
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.example.Features, org.tensorflow.proto.example.Features.Builder, org.tensorflow.proto.example.FeaturesOrBuilder> 
-        getContextFieldBuilder() {
-      if (contextBuilder_ == null) {
-        contextBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-            org.tensorflow.proto.example.Features, org.tensorflow.proto.example.Features.Builder, org.tensorflow.proto.example.FeaturesOrBuilder>(
-                getContext(),
-                getParentForChildren(),
-                isClean());
-        context_ = null;
-      }
-      return contextBuilder_;
-    }
-
-    private org.tensorflow.proto.example.FeatureLists featureLists_;
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.example.FeatureLists, org.tensorflow.proto.example.FeatureLists.Builder, org.tensorflow.proto.example.FeatureListsOrBuilder> featureListsBuilder_;
-    /**
-     * <code>.tensorflow.FeatureLists feature_lists = 2;</code>
-     */
-    public boolean hasFeatureLists() {
-      return featureListsBuilder_ != null || featureLists_ != null;
-    }
-    /**
-     * <code>.tensorflow.FeatureLists feature_lists = 2;</code>
-     */
-    public org.tensorflow.proto.example.FeatureLists getFeatureLists() {
-      if (featureListsBuilder_ == null) {
-        return featureLists_ == null ? org.tensorflow.proto.example.FeatureLists.getDefaultInstance() : featureLists_;
-      } else {
-        return featureListsBuilder_.getMessage();
-      }
-    }
-    /**
-     * <code>.tensorflow.FeatureLists feature_lists = 2;</code>
-     */
-    public Builder setFeatureLists(org.tensorflow.proto.example.FeatureLists value) {
-      if (featureListsBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        featureLists_ = value;
-        onChanged();
-      } else {
-        featureListsBuilder_.setMessage(value);
-      }
-
-      return this;
-    }
-    /**
-     * <code>.tensorflow.FeatureLists feature_lists = 2;</code>
-     */
-    public Builder setFeatureLists(
-        org.tensorflow.proto.example.FeatureLists.Builder builderForValue) {
-      if (featureListsBuilder_ == null) {
-        featureLists_ = builderForValue.build();
-        onChanged();
-      } else {
-        featureListsBuilder_.setMessage(builderForValue.build());
-      }
-
-      return this;
-    }
-    /**
-     * <code>.tensorflow.FeatureLists feature_lists = 2;</code>
-     */
-    public Builder mergeFeatureLists(org.tensorflow.proto.example.FeatureLists value) {
-      if (featureListsBuilder_ == null) {
-        if (featureLists_ != null) {
-          featureLists_ =
-            org.tensorflow.proto.example.FeatureLists.newBuilder(featureLists_).mergeFrom(value).buildPartial();
-        } else {
-          featureLists_ = value;
-        }
-        onChanged();
-      } else {
-        featureListsBuilder_.mergeFrom(value);
-      }
-
-      return this;
-    }
-    /**
-     * <code>.tensorflow.FeatureLists feature_lists = 2;</code>
-     */
-    public Builder clearFeatureLists() {
-      if (featureListsBuilder_ == null) {
-        featureLists_ = null;
-        onChanged();
-      } else {
-        featureLists_ = null;
-        featureListsBuilder_ = null;
-      }
-
-      return this;
-    }
-    /**
-     * <code>.tensorflow.FeatureLists feature_lists = 2;</code>
-     */
-    public org.tensorflow.proto.example.FeatureLists.Builder getFeatureListsBuilder() {
-      
-      onChanged();
-      return getFeatureListsFieldBuilder().getBuilder();
-    }
-    /**
-     * <code>.tensorflow.FeatureLists feature_lists = 2;</code>
-     */
-    public org.tensorflow.proto.example.FeatureListsOrBuilder getFeatureListsOrBuilder() {
-      if (featureListsBuilder_ != null) {
-        return featureListsBuilder_.getMessageOrBuilder();
-      } else {
-        return featureLists_ == null ?
-            org.tensorflow.proto.example.FeatureLists.getDefaultInstance() : featureLists_;
-      }
-    }
-    /**
-     * <code>.tensorflow.FeatureLists feature_lists = 2;</code>
-     */
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.example.FeatureLists, org.tensorflow.proto.example.FeatureLists.Builder, org.tensorflow.proto.example.FeatureListsOrBuilder> 
-        getFeatureListsFieldBuilder() {
-      if (featureListsBuilder_ == null) {
-        featureListsBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-            org.tensorflow.proto.example.FeatureLists, org.tensorflow.proto.example.FeatureLists.Builder, org.tensorflow.proto.example.FeatureListsOrBuilder>(
-                getFeatureLists(),
-                getParentForChildren(),
-                isClean());
-        featureLists_ = null;
-      }
-      return featureListsBuilder_;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.SequenceExample)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.SequenceExample)
-  private static final org.tensorflow.proto.example.SequenceExample DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.example.SequenceExample();
-  }
-
-  public static org.tensorflow.proto.example.SequenceExample getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<SequenceExample>
-      PARSER = new com.google.protobuf.AbstractParser<SequenceExample>() {
-    @java.lang.Override
-    public SequenceExample parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new SequenceExample(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<SequenceExample> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<SequenceExample> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.example.SequenceExample getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/SequenceExampleOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/SequenceExampleOrBuilder.java
deleted file mode 100644
index 9b849c8c3c2..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/SequenceExampleOrBuilder.java
+++ /dev/null
@@ -1,35 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/example/example.proto
-
-package org.tensorflow.proto.example;
-
-public interface SequenceExampleOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.SequenceExample)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <code>.tensorflow.Features context = 1;</code>
-   */
-  boolean hasContext();
-  /**
-   * <code>.tensorflow.Features context = 1;</code>
-   */
-  org.tensorflow.proto.example.Features getContext();
-  /**
-   * <code>.tensorflow.Features context = 1;</code>
-   */
-  org.tensorflow.proto.example.FeaturesOrBuilder getContextOrBuilder();
-
-  /**
-   * <code>.tensorflow.FeatureLists feature_lists = 2;</code>
-   */
-  boolean hasFeatureLists();
-  /**
-   * <code>.tensorflow.FeatureLists feature_lists = 2;</code>
-   */
-  org.tensorflow.proto.example.FeatureLists getFeatureLists();
-  /**
-   * <code>.tensorflow.FeatureLists feature_lists = 2;</code>
-   */
-  org.tensorflow.proto.example.FeatureListsOrBuilder getFeatureListsOrBuilder();
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/VarLenFeatureProto.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/VarLenFeatureProto.java
deleted file mode 100644
index 8edd05b1f24..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/VarLenFeatureProto.java
+++ /dev/null
@@ -1,885 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/example/example_parser_configuration.proto
-
-package org.tensorflow.proto.example;
-
-/**
- * Protobuf type {@code tensorflow.VarLenFeatureProto}
- */
-public  final class VarLenFeatureProto extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.VarLenFeatureProto)
-    VarLenFeatureProtoOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use VarLenFeatureProto.newBuilder() to construct.
-  private VarLenFeatureProto(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private VarLenFeatureProto() {
-    dtype_ = 0;
-    valuesOutputTensorName_ = "";
-    indicesOutputTensorName_ = "";
-    shapesOutputTensorName_ = "";
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new VarLenFeatureProto();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private VarLenFeatureProto(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 8: {
-            int rawValue = input.readEnum();
-
-            dtype_ = rawValue;
-            break;
-          }
-          case 18: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            valuesOutputTensorName_ = s;
-            break;
-          }
-          case 26: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            indicesOutputTensorName_ = s;
-            break;
-          }
-          case 34: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            shapesOutputTensorName_ = s;
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.example.ExampleParserConfigurationProtos.internal_static_tensorflow_VarLenFeatureProto_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.example.ExampleParserConfigurationProtos.internal_static_tensorflow_VarLenFeatureProto_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.example.VarLenFeatureProto.class, org.tensorflow.proto.example.VarLenFeatureProto.Builder.class);
-  }
-
-  public static final int DTYPE_FIELD_NUMBER = 1;
-  private int dtype_;
-  /**
-   * <code>.tensorflow.DataType dtype = 1;</code>
-   */
-  public int getDtypeValue() {
-    return dtype_;
-  }
-  /**
-   * <code>.tensorflow.DataType dtype = 1;</code>
-   */
-  public org.tensorflow.proto.framework.DataType getDtype() {
-    @SuppressWarnings("deprecation")
-    org.tensorflow.proto.framework.DataType result = org.tensorflow.proto.framework.DataType.valueOf(dtype_);
-    return result == null ? org.tensorflow.proto.framework.DataType.UNRECOGNIZED : result;
-  }
-
-  public static final int VALUES_OUTPUT_TENSOR_NAME_FIELD_NUMBER = 2;
-  private volatile java.lang.Object valuesOutputTensorName_;
-  /**
-   * <code>string values_output_tensor_name = 2;</code>
-   */
-  public java.lang.String getValuesOutputTensorName() {
-    java.lang.Object ref = valuesOutputTensorName_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      valuesOutputTensorName_ = s;
-      return s;
-    }
-  }
-  /**
-   * <code>string values_output_tensor_name = 2;</code>
-   */
-  public com.google.protobuf.ByteString
-      getValuesOutputTensorNameBytes() {
-    java.lang.Object ref = valuesOutputTensorName_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      valuesOutputTensorName_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int INDICES_OUTPUT_TENSOR_NAME_FIELD_NUMBER = 3;
-  private volatile java.lang.Object indicesOutputTensorName_;
-  /**
-   * <code>string indices_output_tensor_name = 3;</code>
-   */
-  public java.lang.String getIndicesOutputTensorName() {
-    java.lang.Object ref = indicesOutputTensorName_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      indicesOutputTensorName_ = s;
-      return s;
-    }
-  }
-  /**
-   * <code>string indices_output_tensor_name = 3;</code>
-   */
-  public com.google.protobuf.ByteString
-      getIndicesOutputTensorNameBytes() {
-    java.lang.Object ref = indicesOutputTensorName_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      indicesOutputTensorName_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int SHAPES_OUTPUT_TENSOR_NAME_FIELD_NUMBER = 4;
-  private volatile java.lang.Object shapesOutputTensorName_;
-  /**
-   * <code>string shapes_output_tensor_name = 4;</code>
-   */
-  public java.lang.String getShapesOutputTensorName() {
-    java.lang.Object ref = shapesOutputTensorName_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      shapesOutputTensorName_ = s;
-      return s;
-    }
-  }
-  /**
-   * <code>string shapes_output_tensor_name = 4;</code>
-   */
-  public com.google.protobuf.ByteString
-      getShapesOutputTensorNameBytes() {
-    java.lang.Object ref = shapesOutputTensorName_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      shapesOutputTensorName_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    if (dtype_ != org.tensorflow.proto.framework.DataType.DT_INVALID.getNumber()) {
-      output.writeEnum(1, dtype_);
-    }
-    if (!getValuesOutputTensorNameBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 2, valuesOutputTensorName_);
-    }
-    if (!getIndicesOutputTensorNameBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 3, indicesOutputTensorName_);
-    }
-    if (!getShapesOutputTensorNameBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 4, shapesOutputTensorName_);
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (dtype_ != org.tensorflow.proto.framework.DataType.DT_INVALID.getNumber()) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeEnumSize(1, dtype_);
-    }
-    if (!getValuesOutputTensorNameBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(2, valuesOutputTensorName_);
-    }
-    if (!getIndicesOutputTensorNameBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(3, indicesOutputTensorName_);
-    }
-    if (!getShapesOutputTensorNameBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(4, shapesOutputTensorName_);
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.example.VarLenFeatureProto)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.example.VarLenFeatureProto other = (org.tensorflow.proto.example.VarLenFeatureProto) obj;
-
-    if (dtype_ != other.dtype_) return false;
-    if (!getValuesOutputTensorName()
-        .equals(other.getValuesOutputTensorName())) return false;
-    if (!getIndicesOutputTensorName()
-        .equals(other.getIndicesOutputTensorName())) return false;
-    if (!getShapesOutputTensorName()
-        .equals(other.getShapesOutputTensorName())) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    hash = (37 * hash) + DTYPE_FIELD_NUMBER;
-    hash = (53 * hash) + dtype_;
-    hash = (37 * hash) + VALUES_OUTPUT_TENSOR_NAME_FIELD_NUMBER;
-    hash = (53 * hash) + getValuesOutputTensorName().hashCode();
-    hash = (37 * hash) + INDICES_OUTPUT_TENSOR_NAME_FIELD_NUMBER;
-    hash = (53 * hash) + getIndicesOutputTensorName().hashCode();
-    hash = (37 * hash) + SHAPES_OUTPUT_TENSOR_NAME_FIELD_NUMBER;
-    hash = (53 * hash) + getShapesOutputTensorName().hashCode();
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.example.VarLenFeatureProto parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.example.VarLenFeatureProto parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.example.VarLenFeatureProto parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.example.VarLenFeatureProto parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.example.VarLenFeatureProto parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.example.VarLenFeatureProto parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.example.VarLenFeatureProto parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.example.VarLenFeatureProto parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.example.VarLenFeatureProto parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.example.VarLenFeatureProto parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.example.VarLenFeatureProto parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.example.VarLenFeatureProto parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.example.VarLenFeatureProto prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * Protobuf type {@code tensorflow.VarLenFeatureProto}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.VarLenFeatureProto)
-      org.tensorflow.proto.example.VarLenFeatureProtoOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.example.ExampleParserConfigurationProtos.internal_static_tensorflow_VarLenFeatureProto_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.example.ExampleParserConfigurationProtos.internal_static_tensorflow_VarLenFeatureProto_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.example.VarLenFeatureProto.class, org.tensorflow.proto.example.VarLenFeatureProto.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.example.VarLenFeatureProto.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      dtype_ = 0;
-
-      valuesOutputTensorName_ = "";
-
-      indicesOutputTensorName_ = "";
-
-      shapesOutputTensorName_ = "";
-
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.example.ExampleParserConfigurationProtos.internal_static_tensorflow_VarLenFeatureProto_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.example.VarLenFeatureProto getDefaultInstanceForType() {
-      return org.tensorflow.proto.example.VarLenFeatureProto.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.example.VarLenFeatureProto build() {
-      org.tensorflow.proto.example.VarLenFeatureProto result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.example.VarLenFeatureProto buildPartial() {
-      org.tensorflow.proto.example.VarLenFeatureProto result = new org.tensorflow.proto.example.VarLenFeatureProto(this);
-      result.dtype_ = dtype_;
-      result.valuesOutputTensorName_ = valuesOutputTensorName_;
-      result.indicesOutputTensorName_ = indicesOutputTensorName_;
-      result.shapesOutputTensorName_ = shapesOutputTensorName_;
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.example.VarLenFeatureProto) {
-        return mergeFrom((org.tensorflow.proto.example.VarLenFeatureProto)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.example.VarLenFeatureProto other) {
-      if (other == org.tensorflow.proto.example.VarLenFeatureProto.getDefaultInstance()) return this;
-      if (other.dtype_ != 0) {
-        setDtypeValue(other.getDtypeValue());
-      }
-      if (!other.getValuesOutputTensorName().isEmpty()) {
-        valuesOutputTensorName_ = other.valuesOutputTensorName_;
-        onChanged();
-      }
-      if (!other.getIndicesOutputTensorName().isEmpty()) {
-        indicesOutputTensorName_ = other.indicesOutputTensorName_;
-        onChanged();
-      }
-      if (!other.getShapesOutputTensorName().isEmpty()) {
-        shapesOutputTensorName_ = other.shapesOutputTensorName_;
-        onChanged();
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.example.VarLenFeatureProto parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.example.VarLenFeatureProto) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-
-    private int dtype_ = 0;
-    /**
-     * <code>.tensorflow.DataType dtype = 1;</code>
-     */
-    public int getDtypeValue() {
-      return dtype_;
-    }
-    /**
-     * <code>.tensorflow.DataType dtype = 1;</code>
-     */
-    public Builder setDtypeValue(int value) {
-      dtype_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>.tensorflow.DataType dtype = 1;</code>
-     */
-    public org.tensorflow.proto.framework.DataType getDtype() {
-      @SuppressWarnings("deprecation")
-      org.tensorflow.proto.framework.DataType result = org.tensorflow.proto.framework.DataType.valueOf(dtype_);
-      return result == null ? org.tensorflow.proto.framework.DataType.UNRECOGNIZED : result;
-    }
-    /**
-     * <code>.tensorflow.DataType dtype = 1;</code>
-     */
-    public Builder setDtype(org.tensorflow.proto.framework.DataType value) {
-      if (value == null) {
-        throw new NullPointerException();
-      }
-      
-      dtype_ = value.getNumber();
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>.tensorflow.DataType dtype = 1;</code>
-     */
-    public Builder clearDtype() {
-      
-      dtype_ = 0;
-      onChanged();
-      return this;
-    }
-
-    private java.lang.Object valuesOutputTensorName_ = "";
-    /**
-     * <code>string values_output_tensor_name = 2;</code>
-     */
-    public java.lang.String getValuesOutputTensorName() {
-      java.lang.Object ref = valuesOutputTensorName_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        valuesOutputTensorName_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <code>string values_output_tensor_name = 2;</code>
-     */
-    public com.google.protobuf.ByteString
-        getValuesOutputTensorNameBytes() {
-      java.lang.Object ref = valuesOutputTensorName_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        valuesOutputTensorName_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <code>string values_output_tensor_name = 2;</code>
-     */
-    public Builder setValuesOutputTensorName(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      valuesOutputTensorName_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>string values_output_tensor_name = 2;</code>
-     */
-    public Builder clearValuesOutputTensorName() {
-      
-      valuesOutputTensorName_ = getDefaultInstance().getValuesOutputTensorName();
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>string values_output_tensor_name = 2;</code>
-     */
-    public Builder setValuesOutputTensorNameBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      valuesOutputTensorName_ = value;
-      onChanged();
-      return this;
-    }
-
-    private java.lang.Object indicesOutputTensorName_ = "";
-    /**
-     * <code>string indices_output_tensor_name = 3;</code>
-     */
-    public java.lang.String getIndicesOutputTensorName() {
-      java.lang.Object ref = indicesOutputTensorName_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        indicesOutputTensorName_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <code>string indices_output_tensor_name = 3;</code>
-     */
-    public com.google.protobuf.ByteString
-        getIndicesOutputTensorNameBytes() {
-      java.lang.Object ref = indicesOutputTensorName_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        indicesOutputTensorName_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <code>string indices_output_tensor_name = 3;</code>
-     */
-    public Builder setIndicesOutputTensorName(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      indicesOutputTensorName_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>string indices_output_tensor_name = 3;</code>
-     */
-    public Builder clearIndicesOutputTensorName() {
-      
-      indicesOutputTensorName_ = getDefaultInstance().getIndicesOutputTensorName();
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>string indices_output_tensor_name = 3;</code>
-     */
-    public Builder setIndicesOutputTensorNameBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      indicesOutputTensorName_ = value;
-      onChanged();
-      return this;
-    }
-
-    private java.lang.Object shapesOutputTensorName_ = "";
-    /**
-     * <code>string shapes_output_tensor_name = 4;</code>
-     */
-    public java.lang.String getShapesOutputTensorName() {
-      java.lang.Object ref = shapesOutputTensorName_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        shapesOutputTensorName_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <code>string shapes_output_tensor_name = 4;</code>
-     */
-    public com.google.protobuf.ByteString
-        getShapesOutputTensorNameBytes() {
-      java.lang.Object ref = shapesOutputTensorName_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        shapesOutputTensorName_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <code>string shapes_output_tensor_name = 4;</code>
-     */
-    public Builder setShapesOutputTensorName(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      shapesOutputTensorName_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>string shapes_output_tensor_name = 4;</code>
-     */
-    public Builder clearShapesOutputTensorName() {
-      
-      shapesOutputTensorName_ = getDefaultInstance().getShapesOutputTensorName();
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>string shapes_output_tensor_name = 4;</code>
-     */
-    public Builder setShapesOutputTensorNameBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      shapesOutputTensorName_ = value;
-      onChanged();
-      return this;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.VarLenFeatureProto)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.VarLenFeatureProto)
-  private static final org.tensorflow.proto.example.VarLenFeatureProto DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.example.VarLenFeatureProto();
-  }
-
-  public static org.tensorflow.proto.example.VarLenFeatureProto getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<VarLenFeatureProto>
-      PARSER = new com.google.protobuf.AbstractParser<VarLenFeatureProto>() {
-    @java.lang.Override
-    public VarLenFeatureProto parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new VarLenFeatureProto(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<VarLenFeatureProto> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<VarLenFeatureProto> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.example.VarLenFeatureProto getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/VarLenFeatureProtoOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/VarLenFeatureProtoOrBuilder.java
deleted file mode 100644
index 820c2faa8f8..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/VarLenFeatureProtoOrBuilder.java
+++ /dev/null
@@ -1,48 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/example/example_parser_configuration.proto
-
-package org.tensorflow.proto.example;
-
-public interface VarLenFeatureProtoOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.VarLenFeatureProto)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <code>.tensorflow.DataType dtype = 1;</code>
-   */
-  int getDtypeValue();
-  /**
-   * <code>.tensorflow.DataType dtype = 1;</code>
-   */
-  org.tensorflow.proto.framework.DataType getDtype();
-
-  /**
-   * <code>string values_output_tensor_name = 2;</code>
-   */
-  java.lang.String getValuesOutputTensorName();
-  /**
-   * <code>string values_output_tensor_name = 2;</code>
-   */
-  com.google.protobuf.ByteString
-      getValuesOutputTensorNameBytes();
-
-  /**
-   * <code>string indices_output_tensor_name = 3;</code>
-   */
-  java.lang.String getIndicesOutputTensorName();
-  /**
-   * <code>string indices_output_tensor_name = 3;</code>
-   */
-  com.google.protobuf.ByteString
-      getIndicesOutputTensorNameBytes();
-
-  /**
-   * <code>string shapes_output_tensor_name = 4;</code>
-   */
-  java.lang.String getShapesOutputTensorName();
-  /**
-   * <code>string shapes_output_tensor_name = 4;</code>
-   */
-  com.google.protobuf.ByteString
-      getShapesOutputTensorNameBytes();
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/AllocationDescription.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/AllocationDescription.java
deleted file mode 100644
index 2abffe7549a..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/AllocationDescription.java
+++ /dev/null
@@ -1,944 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/framework/allocation_description.proto
-
-package org.tensorflow.proto.framework;
-
-/**
- * Protobuf type {@code tensorflow.AllocationDescription}
- */
-public  final class AllocationDescription extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.AllocationDescription)
-    AllocationDescriptionOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use AllocationDescription.newBuilder() to construct.
-  private AllocationDescription(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private AllocationDescription() {
-    allocatorName_ = "";
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new AllocationDescription();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private AllocationDescription(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 8: {
-
-            requestedBytes_ = input.readInt64();
-            break;
-          }
-          case 16: {
-
-            allocatedBytes_ = input.readInt64();
-            break;
-          }
-          case 26: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            allocatorName_ = s;
-            break;
-          }
-          case 32: {
-
-            allocationId_ = input.readInt64();
-            break;
-          }
-          case 40: {
-
-            hasSingleReference_ = input.readBool();
-            break;
-          }
-          case 48: {
-
-            ptr_ = input.readUInt64();
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.framework.AllocationDescriptionProtos.internal_static_tensorflow_AllocationDescription_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.framework.AllocationDescriptionProtos.internal_static_tensorflow_AllocationDescription_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.framework.AllocationDescription.class, org.tensorflow.proto.framework.AllocationDescription.Builder.class);
-  }
-
-  public static final int REQUESTED_BYTES_FIELD_NUMBER = 1;
-  private long requestedBytes_;
-  /**
-   * <pre>
-   * Total number of bytes requested
-   * </pre>
-   *
-   * <code>int64 requested_bytes = 1;</code>
-   */
-  public long getRequestedBytes() {
-    return requestedBytes_;
-  }
-
-  public static final int ALLOCATED_BYTES_FIELD_NUMBER = 2;
-  private long allocatedBytes_;
-  /**
-   * <pre>
-   * Total number of bytes allocated if known
-   * </pre>
-   *
-   * <code>int64 allocated_bytes = 2;</code>
-   */
-  public long getAllocatedBytes() {
-    return allocatedBytes_;
-  }
-
-  public static final int ALLOCATOR_NAME_FIELD_NUMBER = 3;
-  private volatile java.lang.Object allocatorName_;
-  /**
-   * <pre>
-   * Name of the allocator used
-   * </pre>
-   *
-   * <code>string allocator_name = 3;</code>
-   */
-  public java.lang.String getAllocatorName() {
-    java.lang.Object ref = allocatorName_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      allocatorName_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * Name of the allocator used
-   * </pre>
-   *
-   * <code>string allocator_name = 3;</code>
-   */
-  public com.google.protobuf.ByteString
-      getAllocatorNameBytes() {
-    java.lang.Object ref = allocatorName_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      allocatorName_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int ALLOCATION_ID_FIELD_NUMBER = 4;
-  private long allocationId_;
-  /**
-   * <pre>
-   * Identifier of the allocated buffer if known
-   * </pre>
-   *
-   * <code>int64 allocation_id = 4;</code>
-   */
-  public long getAllocationId() {
-    return allocationId_;
-  }
-
-  public static final int HAS_SINGLE_REFERENCE_FIELD_NUMBER = 5;
-  private boolean hasSingleReference_;
-  /**
-   * <pre>
-   * Set if this tensor only has one remaining reference
-   * </pre>
-   *
-   * <code>bool has_single_reference = 5;</code>
-   */
-  public boolean getHasSingleReference() {
-    return hasSingleReference_;
-  }
-
-  public static final int PTR_FIELD_NUMBER = 6;
-  private long ptr_;
-  /**
-   * <pre>
-   * Address of the allocation.
-   * </pre>
-   *
-   * <code>uint64 ptr = 6;</code>
-   */
-  public long getPtr() {
-    return ptr_;
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    if (requestedBytes_ != 0L) {
-      output.writeInt64(1, requestedBytes_);
-    }
-    if (allocatedBytes_ != 0L) {
-      output.writeInt64(2, allocatedBytes_);
-    }
-    if (!getAllocatorNameBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 3, allocatorName_);
-    }
-    if (allocationId_ != 0L) {
-      output.writeInt64(4, allocationId_);
-    }
-    if (hasSingleReference_ != false) {
-      output.writeBool(5, hasSingleReference_);
-    }
-    if (ptr_ != 0L) {
-      output.writeUInt64(6, ptr_);
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (requestedBytes_ != 0L) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt64Size(1, requestedBytes_);
-    }
-    if (allocatedBytes_ != 0L) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt64Size(2, allocatedBytes_);
-    }
-    if (!getAllocatorNameBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(3, allocatorName_);
-    }
-    if (allocationId_ != 0L) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt64Size(4, allocationId_);
-    }
-    if (hasSingleReference_ != false) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeBoolSize(5, hasSingleReference_);
-    }
-    if (ptr_ != 0L) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeUInt64Size(6, ptr_);
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.framework.AllocationDescription)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.framework.AllocationDescription other = (org.tensorflow.proto.framework.AllocationDescription) obj;
-
-    if (getRequestedBytes()
-        != other.getRequestedBytes()) return false;
-    if (getAllocatedBytes()
-        != other.getAllocatedBytes()) return false;
-    if (!getAllocatorName()
-        .equals(other.getAllocatorName())) return false;
-    if (getAllocationId()
-        != other.getAllocationId()) return false;
-    if (getHasSingleReference()
-        != other.getHasSingleReference()) return false;
-    if (getPtr()
-        != other.getPtr()) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    hash = (37 * hash) + REQUESTED_BYTES_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-        getRequestedBytes());
-    hash = (37 * hash) + ALLOCATED_BYTES_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-        getAllocatedBytes());
-    hash = (37 * hash) + ALLOCATOR_NAME_FIELD_NUMBER;
-    hash = (53 * hash) + getAllocatorName().hashCode();
-    hash = (37 * hash) + ALLOCATION_ID_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-        getAllocationId());
-    hash = (37 * hash) + HAS_SINGLE_REFERENCE_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashBoolean(
-        getHasSingleReference());
-    hash = (37 * hash) + PTR_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-        getPtr());
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.framework.AllocationDescription parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.AllocationDescription parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.AllocationDescription parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.AllocationDescription parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.AllocationDescription parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.AllocationDescription parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.AllocationDescription parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.AllocationDescription parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.AllocationDescription parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.AllocationDescription parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.AllocationDescription parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.AllocationDescription parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.framework.AllocationDescription prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * Protobuf type {@code tensorflow.AllocationDescription}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.AllocationDescription)
-      org.tensorflow.proto.framework.AllocationDescriptionOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.AllocationDescriptionProtos.internal_static_tensorflow_AllocationDescription_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.AllocationDescriptionProtos.internal_static_tensorflow_AllocationDescription_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.AllocationDescription.class, org.tensorflow.proto.framework.AllocationDescription.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.framework.AllocationDescription.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      requestedBytes_ = 0L;
-
-      allocatedBytes_ = 0L;
-
-      allocatorName_ = "";
-
-      allocationId_ = 0L;
-
-      hasSingleReference_ = false;
-
-      ptr_ = 0L;
-
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.framework.AllocationDescriptionProtos.internal_static_tensorflow_AllocationDescription_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.AllocationDescription getDefaultInstanceForType() {
-      return org.tensorflow.proto.framework.AllocationDescription.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.AllocationDescription build() {
-      org.tensorflow.proto.framework.AllocationDescription result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.AllocationDescription buildPartial() {
-      org.tensorflow.proto.framework.AllocationDescription result = new org.tensorflow.proto.framework.AllocationDescription(this);
-      result.requestedBytes_ = requestedBytes_;
-      result.allocatedBytes_ = allocatedBytes_;
-      result.allocatorName_ = allocatorName_;
-      result.allocationId_ = allocationId_;
-      result.hasSingleReference_ = hasSingleReference_;
-      result.ptr_ = ptr_;
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.framework.AllocationDescription) {
-        return mergeFrom((org.tensorflow.proto.framework.AllocationDescription)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.framework.AllocationDescription other) {
-      if (other == org.tensorflow.proto.framework.AllocationDescription.getDefaultInstance()) return this;
-      if (other.getRequestedBytes() != 0L) {
-        setRequestedBytes(other.getRequestedBytes());
-      }
-      if (other.getAllocatedBytes() != 0L) {
-        setAllocatedBytes(other.getAllocatedBytes());
-      }
-      if (!other.getAllocatorName().isEmpty()) {
-        allocatorName_ = other.allocatorName_;
-        onChanged();
-      }
-      if (other.getAllocationId() != 0L) {
-        setAllocationId(other.getAllocationId());
-      }
-      if (other.getHasSingleReference() != false) {
-        setHasSingleReference(other.getHasSingleReference());
-      }
-      if (other.getPtr() != 0L) {
-        setPtr(other.getPtr());
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.framework.AllocationDescription parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.framework.AllocationDescription) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-
-    private long requestedBytes_ ;
-    /**
-     * <pre>
-     * Total number of bytes requested
-     * </pre>
-     *
-     * <code>int64 requested_bytes = 1;</code>
-     */
-    public long getRequestedBytes() {
-      return requestedBytes_;
-    }
-    /**
-     * <pre>
-     * Total number of bytes requested
-     * </pre>
-     *
-     * <code>int64 requested_bytes = 1;</code>
-     */
-    public Builder setRequestedBytes(long value) {
-      
-      requestedBytes_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Total number of bytes requested
-     * </pre>
-     *
-     * <code>int64 requested_bytes = 1;</code>
-     */
-    public Builder clearRequestedBytes() {
-      
-      requestedBytes_ = 0L;
-      onChanged();
-      return this;
-    }
-
-    private long allocatedBytes_ ;
-    /**
-     * <pre>
-     * Total number of bytes allocated if known
-     * </pre>
-     *
-     * <code>int64 allocated_bytes = 2;</code>
-     */
-    public long getAllocatedBytes() {
-      return allocatedBytes_;
-    }
-    /**
-     * <pre>
-     * Total number of bytes allocated if known
-     * </pre>
-     *
-     * <code>int64 allocated_bytes = 2;</code>
-     */
-    public Builder setAllocatedBytes(long value) {
-      
-      allocatedBytes_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Total number of bytes allocated if known
-     * </pre>
-     *
-     * <code>int64 allocated_bytes = 2;</code>
-     */
-    public Builder clearAllocatedBytes() {
-      
-      allocatedBytes_ = 0L;
-      onChanged();
-      return this;
-    }
-
-    private java.lang.Object allocatorName_ = "";
-    /**
-     * <pre>
-     * Name of the allocator used
-     * </pre>
-     *
-     * <code>string allocator_name = 3;</code>
-     */
-    public java.lang.String getAllocatorName() {
-      java.lang.Object ref = allocatorName_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        allocatorName_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Name of the allocator used
-     * </pre>
-     *
-     * <code>string allocator_name = 3;</code>
-     */
-    public com.google.protobuf.ByteString
-        getAllocatorNameBytes() {
-      java.lang.Object ref = allocatorName_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        allocatorName_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Name of the allocator used
-     * </pre>
-     *
-     * <code>string allocator_name = 3;</code>
-     */
-    public Builder setAllocatorName(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      allocatorName_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Name of the allocator used
-     * </pre>
-     *
-     * <code>string allocator_name = 3;</code>
-     */
-    public Builder clearAllocatorName() {
-      
-      allocatorName_ = getDefaultInstance().getAllocatorName();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Name of the allocator used
-     * </pre>
-     *
-     * <code>string allocator_name = 3;</code>
-     */
-    public Builder setAllocatorNameBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      allocatorName_ = value;
-      onChanged();
-      return this;
-    }
-
-    private long allocationId_ ;
-    /**
-     * <pre>
-     * Identifier of the allocated buffer if known
-     * </pre>
-     *
-     * <code>int64 allocation_id = 4;</code>
-     */
-    public long getAllocationId() {
-      return allocationId_;
-    }
-    /**
-     * <pre>
-     * Identifier of the allocated buffer if known
-     * </pre>
-     *
-     * <code>int64 allocation_id = 4;</code>
-     */
-    public Builder setAllocationId(long value) {
-      
-      allocationId_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Identifier of the allocated buffer if known
-     * </pre>
-     *
-     * <code>int64 allocation_id = 4;</code>
-     */
-    public Builder clearAllocationId() {
-      
-      allocationId_ = 0L;
-      onChanged();
-      return this;
-    }
-
-    private boolean hasSingleReference_ ;
-    /**
-     * <pre>
-     * Set if this tensor only has one remaining reference
-     * </pre>
-     *
-     * <code>bool has_single_reference = 5;</code>
-     */
-    public boolean getHasSingleReference() {
-      return hasSingleReference_;
-    }
-    /**
-     * <pre>
-     * Set if this tensor only has one remaining reference
-     * </pre>
-     *
-     * <code>bool has_single_reference = 5;</code>
-     */
-    public Builder setHasSingleReference(boolean value) {
-      
-      hasSingleReference_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Set if this tensor only has one remaining reference
-     * </pre>
-     *
-     * <code>bool has_single_reference = 5;</code>
-     */
-    public Builder clearHasSingleReference() {
-      
-      hasSingleReference_ = false;
-      onChanged();
-      return this;
-    }
-
-    private long ptr_ ;
-    /**
-     * <pre>
-     * Address of the allocation.
-     * </pre>
-     *
-     * <code>uint64 ptr = 6;</code>
-     */
-    public long getPtr() {
-      return ptr_;
-    }
-    /**
-     * <pre>
-     * Address of the allocation.
-     * </pre>
-     *
-     * <code>uint64 ptr = 6;</code>
-     */
-    public Builder setPtr(long value) {
-      
-      ptr_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Address of the allocation.
-     * </pre>
-     *
-     * <code>uint64 ptr = 6;</code>
-     */
-    public Builder clearPtr() {
-      
-      ptr_ = 0L;
-      onChanged();
-      return this;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.AllocationDescription)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.AllocationDescription)
-  private static final org.tensorflow.proto.framework.AllocationDescription DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.AllocationDescription();
-  }
-
-  public static org.tensorflow.proto.framework.AllocationDescription getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<AllocationDescription>
-      PARSER = new com.google.protobuf.AbstractParser<AllocationDescription>() {
-    @java.lang.Override
-    public AllocationDescription parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new AllocationDescription(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<AllocationDescription> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<AllocationDescription> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.framework.AllocationDescription getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/AllocationDescriptionOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/AllocationDescriptionOrBuilder.java
deleted file mode 100644
index 9fbd5df0695..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/AllocationDescriptionOrBuilder.java
+++ /dev/null
@@ -1,72 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/framework/allocation_description.proto
-
-package org.tensorflow.proto.framework;
-
-public interface AllocationDescriptionOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.AllocationDescription)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <pre>
-   * Total number of bytes requested
-   * </pre>
-   *
-   * <code>int64 requested_bytes = 1;</code>
-   */
-  long getRequestedBytes();
-
-  /**
-   * <pre>
-   * Total number of bytes allocated if known
-   * </pre>
-   *
-   * <code>int64 allocated_bytes = 2;</code>
-   */
-  long getAllocatedBytes();
-
-  /**
-   * <pre>
-   * Name of the allocator used
-   * </pre>
-   *
-   * <code>string allocator_name = 3;</code>
-   */
-  java.lang.String getAllocatorName();
-  /**
-   * <pre>
-   * Name of the allocator used
-   * </pre>
-   *
-   * <code>string allocator_name = 3;</code>
-   */
-  com.google.protobuf.ByteString
-      getAllocatorNameBytes();
-
-  /**
-   * <pre>
-   * Identifier of the allocated buffer if known
-   * </pre>
-   *
-   * <code>int64 allocation_id = 4;</code>
-   */
-  long getAllocationId();
-
-  /**
-   * <pre>
-   * Set if this tensor only has one remaining reference
-   * </pre>
-   *
-   * <code>bool has_single_reference = 5;</code>
-   */
-  boolean getHasSingleReference();
-
-  /**
-   * <pre>
-   * Address of the allocation.
-   * </pre>
-   *
-   * <code>uint64 ptr = 6;</code>
-   */
-  long getPtr();
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/AllocationDescriptionProtos.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/AllocationDescriptionProtos.java
deleted file mode 100644
index f3af7b380fc..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/AllocationDescriptionProtos.java
+++ /dev/null
@@ -1,56 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/framework/allocation_description.proto
-
-package org.tensorflow.proto.framework;
-
-public final class AllocationDescriptionProtos {
-  private AllocationDescriptionProtos() {}
-  public static void registerAllExtensions(
-      com.google.protobuf.ExtensionRegistryLite registry) {
-  }
-
-  public static void registerAllExtensions(
-      com.google.protobuf.ExtensionRegistry registry) {
-    registerAllExtensions(
-        (com.google.protobuf.ExtensionRegistryLite) registry);
-  }
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_AllocationDescription_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_AllocationDescription_fieldAccessorTable;
-
-  public static com.google.protobuf.Descriptors.FileDescriptor
-      getDescriptor() {
-    return descriptor;
-  }
-  private static  com.google.protobuf.Descriptors.FileDescriptor
-      descriptor;
-  static {
-    java.lang.String[] descriptorData = {
-      "\n6tensorflow/core/framework/allocation_d" +
-      "escription.proto\022\ntensorflow\"\243\001\n\025Allocat" +
-      "ionDescription\022\027\n\017requested_bytes\030\001 \001(\003\022" +
-      "\027\n\017allocated_bytes\030\002 \001(\003\022\026\n\016allocator_na" +
-      "me\030\003 \001(\t\022\025\n\rallocation_id\030\004 \001(\003\022\034\n\024has_s" +
-      "ingle_reference\030\005 \001(\010\022\013\n\003ptr\030\006 \001(\004B\241\001\n\036o" +
-      "rg.tensorflow.proto.frameworkB\033Allocatio" +
-      "nDescriptionProtosP\001Z]github.com/tensorf" +
-      "low/tensorflow/tensorflow/go/core/framew" +
-      "ork/allocation_description_go_proto\370\001\001b\006" +
-      "proto3"
-    };
-    descriptor = com.google.protobuf.Descriptors.FileDescriptor
-      .internalBuildGeneratedFileFrom(descriptorData,
-        new com.google.protobuf.Descriptors.FileDescriptor[] {
-        });
-    internal_static_tensorflow_AllocationDescription_descriptor =
-      getDescriptor().getMessageTypes().get(0);
-    internal_static_tensorflow_AllocationDescription_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_AllocationDescription_descriptor,
-        new java.lang.String[] { "RequestedBytes", "AllocatedBytes", "AllocatorName", "AllocationId", "HasSingleReference", "Ptr", });
-  }
-
-  // @@protoc_insertion_point(outer_class_scope)
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/AllocationRecord.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/AllocationRecord.java
deleted file mode 100644
index 738f1117fbe..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/AllocationRecord.java
+++ /dev/null
@@ -1,575 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/framework/step_stats.proto
-
-package org.tensorflow.proto.framework;
-
-/**
- * <pre>
- * An allocation/de-allocation operation performed by the allocator.
- * </pre>
- *
- * Protobuf type {@code tensorflow.AllocationRecord}
- */
-public  final class AllocationRecord extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.AllocationRecord)
-    AllocationRecordOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use AllocationRecord.newBuilder() to construct.
-  private AllocationRecord(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private AllocationRecord() {
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new AllocationRecord();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private AllocationRecord(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 8: {
-
-            allocMicros_ = input.readInt64();
-            break;
-          }
-          case 16: {
-
-            allocBytes_ = input.readInt64();
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.framework.StepStatsProtos.internal_static_tensorflow_AllocationRecord_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.framework.StepStatsProtos.internal_static_tensorflow_AllocationRecord_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.framework.AllocationRecord.class, org.tensorflow.proto.framework.AllocationRecord.Builder.class);
-  }
-
-  public static final int ALLOC_MICROS_FIELD_NUMBER = 1;
-  private long allocMicros_;
-  /**
-   * <pre>
-   * The timestamp of the operation.
-   * </pre>
-   *
-   * <code>int64 alloc_micros = 1;</code>
-   */
-  public long getAllocMicros() {
-    return allocMicros_;
-  }
-
-  public static final int ALLOC_BYTES_FIELD_NUMBER = 2;
-  private long allocBytes_;
-  /**
-   * <pre>
-   * Number of bytes allocated, or de-allocated if negative.
-   * </pre>
-   *
-   * <code>int64 alloc_bytes = 2;</code>
-   */
-  public long getAllocBytes() {
-    return allocBytes_;
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    if (allocMicros_ != 0L) {
-      output.writeInt64(1, allocMicros_);
-    }
-    if (allocBytes_ != 0L) {
-      output.writeInt64(2, allocBytes_);
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (allocMicros_ != 0L) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt64Size(1, allocMicros_);
-    }
-    if (allocBytes_ != 0L) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt64Size(2, allocBytes_);
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.framework.AllocationRecord)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.framework.AllocationRecord other = (org.tensorflow.proto.framework.AllocationRecord) obj;
-
-    if (getAllocMicros()
-        != other.getAllocMicros()) return false;
-    if (getAllocBytes()
-        != other.getAllocBytes()) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    hash = (37 * hash) + ALLOC_MICROS_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-        getAllocMicros());
-    hash = (37 * hash) + ALLOC_BYTES_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-        getAllocBytes());
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.framework.AllocationRecord parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.AllocationRecord parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.AllocationRecord parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.AllocationRecord parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.AllocationRecord parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.AllocationRecord parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.AllocationRecord parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.AllocationRecord parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.AllocationRecord parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.AllocationRecord parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.AllocationRecord parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.AllocationRecord parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.framework.AllocationRecord prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * <pre>
-   * An allocation/de-allocation operation performed by the allocator.
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.AllocationRecord}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.AllocationRecord)
-      org.tensorflow.proto.framework.AllocationRecordOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.StepStatsProtos.internal_static_tensorflow_AllocationRecord_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.StepStatsProtos.internal_static_tensorflow_AllocationRecord_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.AllocationRecord.class, org.tensorflow.proto.framework.AllocationRecord.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.framework.AllocationRecord.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      allocMicros_ = 0L;
-
-      allocBytes_ = 0L;
-
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.framework.StepStatsProtos.internal_static_tensorflow_AllocationRecord_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.AllocationRecord getDefaultInstanceForType() {
-      return org.tensorflow.proto.framework.AllocationRecord.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.AllocationRecord build() {
-      org.tensorflow.proto.framework.AllocationRecord result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.AllocationRecord buildPartial() {
-      org.tensorflow.proto.framework.AllocationRecord result = new org.tensorflow.proto.framework.AllocationRecord(this);
-      result.allocMicros_ = allocMicros_;
-      result.allocBytes_ = allocBytes_;
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.framework.AllocationRecord) {
-        return mergeFrom((org.tensorflow.proto.framework.AllocationRecord)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.framework.AllocationRecord other) {
-      if (other == org.tensorflow.proto.framework.AllocationRecord.getDefaultInstance()) return this;
-      if (other.getAllocMicros() != 0L) {
-        setAllocMicros(other.getAllocMicros());
-      }
-      if (other.getAllocBytes() != 0L) {
-        setAllocBytes(other.getAllocBytes());
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.framework.AllocationRecord parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.framework.AllocationRecord) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-
-    private long allocMicros_ ;
-    /**
-     * <pre>
-     * The timestamp of the operation.
-     * </pre>
-     *
-     * <code>int64 alloc_micros = 1;</code>
-     */
-    public long getAllocMicros() {
-      return allocMicros_;
-    }
-    /**
-     * <pre>
-     * The timestamp of the operation.
-     * </pre>
-     *
-     * <code>int64 alloc_micros = 1;</code>
-     */
-    public Builder setAllocMicros(long value) {
-      
-      allocMicros_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * The timestamp of the operation.
-     * </pre>
-     *
-     * <code>int64 alloc_micros = 1;</code>
-     */
-    public Builder clearAllocMicros() {
-      
-      allocMicros_ = 0L;
-      onChanged();
-      return this;
-    }
-
-    private long allocBytes_ ;
-    /**
-     * <pre>
-     * Number of bytes allocated, or de-allocated if negative.
-     * </pre>
-     *
-     * <code>int64 alloc_bytes = 2;</code>
-     */
-    public long getAllocBytes() {
-      return allocBytes_;
-    }
-    /**
-     * <pre>
-     * Number of bytes allocated, or de-allocated if negative.
-     * </pre>
-     *
-     * <code>int64 alloc_bytes = 2;</code>
-     */
-    public Builder setAllocBytes(long value) {
-      
-      allocBytes_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Number of bytes allocated, or de-allocated if negative.
-     * </pre>
-     *
-     * <code>int64 alloc_bytes = 2;</code>
-     */
-    public Builder clearAllocBytes() {
-      
-      allocBytes_ = 0L;
-      onChanged();
-      return this;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.AllocationRecord)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.AllocationRecord)
-  private static final org.tensorflow.proto.framework.AllocationRecord DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.AllocationRecord();
-  }
-
-  public static org.tensorflow.proto.framework.AllocationRecord getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<AllocationRecord>
-      PARSER = new com.google.protobuf.AbstractParser<AllocationRecord>() {
-    @java.lang.Override
-    public AllocationRecord parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new AllocationRecord(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<AllocationRecord> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<AllocationRecord> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.framework.AllocationRecord getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/AllocationRecordOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/AllocationRecordOrBuilder.java
deleted file mode 100644
index c6c5dd502da..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/AllocationRecordOrBuilder.java
+++ /dev/null
@@ -1,27 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/framework/step_stats.proto
-
-package org.tensorflow.proto.framework;
-
-public interface AllocationRecordOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.AllocationRecord)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <pre>
-   * The timestamp of the operation.
-   * </pre>
-   *
-   * <code>int64 alloc_micros = 1;</code>
-   */
-  long getAllocMicros();
-
-  /**
-   * <pre>
-   * Number of bytes allocated, or de-allocated if negative.
-   * </pre>
-   *
-   * <code>int64 alloc_bytes = 2;</code>
-   */
-  long getAllocBytes();
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/AllocatorMemoryUsed.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/AllocatorMemoryUsed.java
deleted file mode 100644
index 3cb33266367..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/AllocatorMemoryUsed.java
+++ /dev/null
@@ -1,1268 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/framework/step_stats.proto
-
-package org.tensorflow.proto.framework;
-
-/**
- * Protobuf type {@code tensorflow.AllocatorMemoryUsed}
- */
-public  final class AllocatorMemoryUsed extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.AllocatorMemoryUsed)
-    AllocatorMemoryUsedOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use AllocatorMemoryUsed.newBuilder() to construct.
-  private AllocatorMemoryUsed(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private AllocatorMemoryUsed() {
-    allocatorName_ = "";
-    allocationRecords_ = java.util.Collections.emptyList();
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new AllocatorMemoryUsed();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private AllocatorMemoryUsed(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    int mutable_bitField0_ = 0;
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 10: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            allocatorName_ = s;
-            break;
-          }
-          case 16: {
-
-            totalBytes_ = input.readInt64();
-            break;
-          }
-          case 24: {
-
-            peakBytes_ = input.readInt64();
-            break;
-          }
-          case 32: {
-
-            liveBytes_ = input.readInt64();
-            break;
-          }
-          case 40: {
-
-            allocatorBytesInUse_ = input.readInt64();
-            break;
-          }
-          case 50: {
-            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
-              allocationRecords_ = new java.util.ArrayList<org.tensorflow.proto.framework.AllocationRecord>();
-              mutable_bitField0_ |= 0x00000001;
-            }
-            allocationRecords_.add(
-                input.readMessage(org.tensorflow.proto.framework.AllocationRecord.parser(), extensionRegistry));
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      if (((mutable_bitField0_ & 0x00000001) != 0)) {
-        allocationRecords_ = java.util.Collections.unmodifiableList(allocationRecords_);
-      }
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.framework.StepStatsProtos.internal_static_tensorflow_AllocatorMemoryUsed_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.framework.StepStatsProtos.internal_static_tensorflow_AllocatorMemoryUsed_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.framework.AllocatorMemoryUsed.class, org.tensorflow.proto.framework.AllocatorMemoryUsed.Builder.class);
-  }
-
-  public static final int ALLOCATOR_NAME_FIELD_NUMBER = 1;
-  private volatile java.lang.Object allocatorName_;
-  /**
-   * <code>string allocator_name = 1;</code>
-   */
-  public java.lang.String getAllocatorName() {
-    java.lang.Object ref = allocatorName_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      allocatorName_ = s;
-      return s;
-    }
-  }
-  /**
-   * <code>string allocator_name = 1;</code>
-   */
-  public com.google.protobuf.ByteString
-      getAllocatorNameBytes() {
-    java.lang.Object ref = allocatorName_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      allocatorName_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int TOTAL_BYTES_FIELD_NUMBER = 2;
-  private long totalBytes_;
-  /**
-   * <pre>
-   * These are per-node allocator memory stats.
-   * </pre>
-   *
-   * <code>int64 total_bytes = 2;</code>
-   */
-  public long getTotalBytes() {
-    return totalBytes_;
-  }
-
-  public static final int PEAK_BYTES_FIELD_NUMBER = 3;
-  private long peakBytes_;
-  /**
-   * <code>int64 peak_bytes = 3;</code>
-   */
-  public long getPeakBytes() {
-    return peakBytes_;
-  }
-
-  public static final int LIVE_BYTES_FIELD_NUMBER = 4;
-  private long liveBytes_;
-  /**
-   * <pre>
-   * The bytes that are not deallocated.
-   * </pre>
-   *
-   * <code>int64 live_bytes = 4;</code>
-   */
-  public long getLiveBytes() {
-    return liveBytes_;
-  }
-
-  public static final int ALLOCATION_RECORDS_FIELD_NUMBER = 6;
-  private java.util.List<org.tensorflow.proto.framework.AllocationRecord> allocationRecords_;
-  /**
-   * <pre>
-   * The allocation and deallocation timeline.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.AllocationRecord allocation_records = 6;</code>
-   */
-  public java.util.List<org.tensorflow.proto.framework.AllocationRecord> getAllocationRecordsList() {
-    return allocationRecords_;
-  }
-  /**
-   * <pre>
-   * The allocation and deallocation timeline.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.AllocationRecord allocation_records = 6;</code>
-   */
-  public java.util.List<? extends org.tensorflow.proto.framework.AllocationRecordOrBuilder> 
-      getAllocationRecordsOrBuilderList() {
-    return allocationRecords_;
-  }
-  /**
-   * <pre>
-   * The allocation and deallocation timeline.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.AllocationRecord allocation_records = 6;</code>
-   */
-  public int getAllocationRecordsCount() {
-    return allocationRecords_.size();
-  }
-  /**
-   * <pre>
-   * The allocation and deallocation timeline.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.AllocationRecord allocation_records = 6;</code>
-   */
-  public org.tensorflow.proto.framework.AllocationRecord getAllocationRecords(int index) {
-    return allocationRecords_.get(index);
-  }
-  /**
-   * <pre>
-   * The allocation and deallocation timeline.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.AllocationRecord allocation_records = 6;</code>
-   */
-  public org.tensorflow.proto.framework.AllocationRecordOrBuilder getAllocationRecordsOrBuilder(
-      int index) {
-    return allocationRecords_.get(index);
-  }
-
-  public static final int ALLOCATOR_BYTES_IN_USE_FIELD_NUMBER = 5;
-  private long allocatorBytesInUse_;
-  /**
-   * <pre>
-   * These are snapshots of the overall allocator memory stats.
-   * The number of live bytes currently allocated by the allocator.
-   * </pre>
-   *
-   * <code>int64 allocator_bytes_in_use = 5;</code>
-   */
-  public long getAllocatorBytesInUse() {
-    return allocatorBytesInUse_;
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    if (!getAllocatorNameBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 1, allocatorName_);
-    }
-    if (totalBytes_ != 0L) {
-      output.writeInt64(2, totalBytes_);
-    }
-    if (peakBytes_ != 0L) {
-      output.writeInt64(3, peakBytes_);
-    }
-    if (liveBytes_ != 0L) {
-      output.writeInt64(4, liveBytes_);
-    }
-    if (allocatorBytesInUse_ != 0L) {
-      output.writeInt64(5, allocatorBytesInUse_);
-    }
-    for (int i = 0; i < allocationRecords_.size(); i++) {
-      output.writeMessage(6, allocationRecords_.get(i));
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (!getAllocatorNameBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, allocatorName_);
-    }
-    if (totalBytes_ != 0L) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt64Size(2, totalBytes_);
-    }
-    if (peakBytes_ != 0L) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt64Size(3, peakBytes_);
-    }
-    if (liveBytes_ != 0L) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt64Size(4, liveBytes_);
-    }
-    if (allocatorBytesInUse_ != 0L) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt64Size(5, allocatorBytesInUse_);
-    }
-    for (int i = 0; i < allocationRecords_.size(); i++) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(6, allocationRecords_.get(i));
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.framework.AllocatorMemoryUsed)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.framework.AllocatorMemoryUsed other = (org.tensorflow.proto.framework.AllocatorMemoryUsed) obj;
-
-    if (!getAllocatorName()
-        .equals(other.getAllocatorName())) return false;
-    if (getTotalBytes()
-        != other.getTotalBytes()) return false;
-    if (getPeakBytes()
-        != other.getPeakBytes()) return false;
-    if (getLiveBytes()
-        != other.getLiveBytes()) return false;
-    if (!getAllocationRecordsList()
-        .equals(other.getAllocationRecordsList())) return false;
-    if (getAllocatorBytesInUse()
-        != other.getAllocatorBytesInUse()) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    hash = (37 * hash) + ALLOCATOR_NAME_FIELD_NUMBER;
-    hash = (53 * hash) + getAllocatorName().hashCode();
-    hash = (37 * hash) + TOTAL_BYTES_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-        getTotalBytes());
-    hash = (37 * hash) + PEAK_BYTES_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-        getPeakBytes());
-    hash = (37 * hash) + LIVE_BYTES_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-        getLiveBytes());
-    if (getAllocationRecordsCount() > 0) {
-      hash = (37 * hash) + ALLOCATION_RECORDS_FIELD_NUMBER;
-      hash = (53 * hash) + getAllocationRecordsList().hashCode();
-    }
-    hash = (37 * hash) + ALLOCATOR_BYTES_IN_USE_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-        getAllocatorBytesInUse());
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.framework.AllocatorMemoryUsed parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.AllocatorMemoryUsed parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.AllocatorMemoryUsed parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.AllocatorMemoryUsed parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.AllocatorMemoryUsed parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.AllocatorMemoryUsed parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.AllocatorMemoryUsed parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.AllocatorMemoryUsed parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.AllocatorMemoryUsed parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.AllocatorMemoryUsed parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.AllocatorMemoryUsed parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.AllocatorMemoryUsed parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.framework.AllocatorMemoryUsed prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * Protobuf type {@code tensorflow.AllocatorMemoryUsed}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.AllocatorMemoryUsed)
-      org.tensorflow.proto.framework.AllocatorMemoryUsedOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.StepStatsProtos.internal_static_tensorflow_AllocatorMemoryUsed_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.StepStatsProtos.internal_static_tensorflow_AllocatorMemoryUsed_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.AllocatorMemoryUsed.class, org.tensorflow.proto.framework.AllocatorMemoryUsed.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.framework.AllocatorMemoryUsed.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-        getAllocationRecordsFieldBuilder();
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      allocatorName_ = "";
-
-      totalBytes_ = 0L;
-
-      peakBytes_ = 0L;
-
-      liveBytes_ = 0L;
-
-      if (allocationRecordsBuilder_ == null) {
-        allocationRecords_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000001);
-      } else {
-        allocationRecordsBuilder_.clear();
-      }
-      allocatorBytesInUse_ = 0L;
-
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.framework.StepStatsProtos.internal_static_tensorflow_AllocatorMemoryUsed_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.AllocatorMemoryUsed getDefaultInstanceForType() {
-      return org.tensorflow.proto.framework.AllocatorMemoryUsed.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.AllocatorMemoryUsed build() {
-      org.tensorflow.proto.framework.AllocatorMemoryUsed result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.AllocatorMemoryUsed buildPartial() {
-      org.tensorflow.proto.framework.AllocatorMemoryUsed result = new org.tensorflow.proto.framework.AllocatorMemoryUsed(this);
-      int from_bitField0_ = bitField0_;
-      result.allocatorName_ = allocatorName_;
-      result.totalBytes_ = totalBytes_;
-      result.peakBytes_ = peakBytes_;
-      result.liveBytes_ = liveBytes_;
-      if (allocationRecordsBuilder_ == null) {
-        if (((bitField0_ & 0x00000001) != 0)) {
-          allocationRecords_ = java.util.Collections.unmodifiableList(allocationRecords_);
-          bitField0_ = (bitField0_ & ~0x00000001);
-        }
-        result.allocationRecords_ = allocationRecords_;
-      } else {
-        result.allocationRecords_ = allocationRecordsBuilder_.build();
-      }
-      result.allocatorBytesInUse_ = allocatorBytesInUse_;
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.framework.AllocatorMemoryUsed) {
-        return mergeFrom((org.tensorflow.proto.framework.AllocatorMemoryUsed)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.framework.AllocatorMemoryUsed other) {
-      if (other == org.tensorflow.proto.framework.AllocatorMemoryUsed.getDefaultInstance()) return this;
-      if (!other.getAllocatorName().isEmpty()) {
-        allocatorName_ = other.allocatorName_;
-        onChanged();
-      }
-      if (other.getTotalBytes() != 0L) {
-        setTotalBytes(other.getTotalBytes());
-      }
-      if (other.getPeakBytes() != 0L) {
-        setPeakBytes(other.getPeakBytes());
-      }
-      if (other.getLiveBytes() != 0L) {
-        setLiveBytes(other.getLiveBytes());
-      }
-      if (allocationRecordsBuilder_ == null) {
-        if (!other.allocationRecords_.isEmpty()) {
-          if (allocationRecords_.isEmpty()) {
-            allocationRecords_ = other.allocationRecords_;
-            bitField0_ = (bitField0_ & ~0x00000001);
-          } else {
-            ensureAllocationRecordsIsMutable();
-            allocationRecords_.addAll(other.allocationRecords_);
-          }
-          onChanged();
-        }
-      } else {
-        if (!other.allocationRecords_.isEmpty()) {
-          if (allocationRecordsBuilder_.isEmpty()) {
-            allocationRecordsBuilder_.dispose();
-            allocationRecordsBuilder_ = null;
-            allocationRecords_ = other.allocationRecords_;
-            bitField0_ = (bitField0_ & ~0x00000001);
-            allocationRecordsBuilder_ = 
-              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
-                 getAllocationRecordsFieldBuilder() : null;
-          } else {
-            allocationRecordsBuilder_.addAllMessages(other.allocationRecords_);
-          }
-        }
-      }
-      if (other.getAllocatorBytesInUse() != 0L) {
-        setAllocatorBytesInUse(other.getAllocatorBytesInUse());
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.framework.AllocatorMemoryUsed parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.framework.AllocatorMemoryUsed) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-    private int bitField0_;
-
-    private java.lang.Object allocatorName_ = "";
-    /**
-     * <code>string allocator_name = 1;</code>
-     */
-    public java.lang.String getAllocatorName() {
-      java.lang.Object ref = allocatorName_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        allocatorName_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <code>string allocator_name = 1;</code>
-     */
-    public com.google.protobuf.ByteString
-        getAllocatorNameBytes() {
-      java.lang.Object ref = allocatorName_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        allocatorName_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <code>string allocator_name = 1;</code>
-     */
-    public Builder setAllocatorName(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      allocatorName_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>string allocator_name = 1;</code>
-     */
-    public Builder clearAllocatorName() {
-      
-      allocatorName_ = getDefaultInstance().getAllocatorName();
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>string allocator_name = 1;</code>
-     */
-    public Builder setAllocatorNameBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      allocatorName_ = value;
-      onChanged();
-      return this;
-    }
-
-    private long totalBytes_ ;
-    /**
-     * <pre>
-     * These are per-node allocator memory stats.
-     * </pre>
-     *
-     * <code>int64 total_bytes = 2;</code>
-     */
-    public long getTotalBytes() {
-      return totalBytes_;
-    }
-    /**
-     * <pre>
-     * These are per-node allocator memory stats.
-     * </pre>
-     *
-     * <code>int64 total_bytes = 2;</code>
-     */
-    public Builder setTotalBytes(long value) {
-      
-      totalBytes_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * These are per-node allocator memory stats.
-     * </pre>
-     *
-     * <code>int64 total_bytes = 2;</code>
-     */
-    public Builder clearTotalBytes() {
-      
-      totalBytes_ = 0L;
-      onChanged();
-      return this;
-    }
-
-    private long peakBytes_ ;
-    /**
-     * <code>int64 peak_bytes = 3;</code>
-     */
-    public long getPeakBytes() {
-      return peakBytes_;
-    }
-    /**
-     * <code>int64 peak_bytes = 3;</code>
-     */
-    public Builder setPeakBytes(long value) {
-      
-      peakBytes_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>int64 peak_bytes = 3;</code>
-     */
-    public Builder clearPeakBytes() {
-      
-      peakBytes_ = 0L;
-      onChanged();
-      return this;
-    }
-
-    private long liveBytes_ ;
-    /**
-     * <pre>
-     * The bytes that are not deallocated.
-     * </pre>
-     *
-     * <code>int64 live_bytes = 4;</code>
-     */
-    public long getLiveBytes() {
-      return liveBytes_;
-    }
-    /**
-     * <pre>
-     * The bytes that are not deallocated.
-     * </pre>
-     *
-     * <code>int64 live_bytes = 4;</code>
-     */
-    public Builder setLiveBytes(long value) {
-      
-      liveBytes_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * The bytes that are not deallocated.
-     * </pre>
-     *
-     * <code>int64 live_bytes = 4;</code>
-     */
-    public Builder clearLiveBytes() {
-      
-      liveBytes_ = 0L;
-      onChanged();
-      return this;
-    }
-
-    private java.util.List<org.tensorflow.proto.framework.AllocationRecord> allocationRecords_ =
-      java.util.Collections.emptyList();
-    private void ensureAllocationRecordsIsMutable() {
-      if (!((bitField0_ & 0x00000001) != 0)) {
-        allocationRecords_ = new java.util.ArrayList<org.tensorflow.proto.framework.AllocationRecord>(allocationRecords_);
-        bitField0_ |= 0x00000001;
-       }
-    }
-
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.framework.AllocationRecord, org.tensorflow.proto.framework.AllocationRecord.Builder, org.tensorflow.proto.framework.AllocationRecordOrBuilder> allocationRecordsBuilder_;
-
-    /**
-     * <pre>
-     * The allocation and deallocation timeline.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.AllocationRecord allocation_records = 6;</code>
-     */
-    public java.util.List<org.tensorflow.proto.framework.AllocationRecord> getAllocationRecordsList() {
-      if (allocationRecordsBuilder_ == null) {
-        return java.util.Collections.unmodifiableList(allocationRecords_);
-      } else {
-        return allocationRecordsBuilder_.getMessageList();
-      }
-    }
-    /**
-     * <pre>
-     * The allocation and deallocation timeline.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.AllocationRecord allocation_records = 6;</code>
-     */
-    public int getAllocationRecordsCount() {
-      if (allocationRecordsBuilder_ == null) {
-        return allocationRecords_.size();
-      } else {
-        return allocationRecordsBuilder_.getCount();
-      }
-    }
-    /**
-     * <pre>
-     * The allocation and deallocation timeline.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.AllocationRecord allocation_records = 6;</code>
-     */
-    public org.tensorflow.proto.framework.AllocationRecord getAllocationRecords(int index) {
-      if (allocationRecordsBuilder_ == null) {
-        return allocationRecords_.get(index);
-      } else {
-        return allocationRecordsBuilder_.getMessage(index);
-      }
-    }
-    /**
-     * <pre>
-     * The allocation and deallocation timeline.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.AllocationRecord allocation_records = 6;</code>
-     */
-    public Builder setAllocationRecords(
-        int index, org.tensorflow.proto.framework.AllocationRecord value) {
-      if (allocationRecordsBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureAllocationRecordsIsMutable();
-        allocationRecords_.set(index, value);
-        onChanged();
-      } else {
-        allocationRecordsBuilder_.setMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * The allocation and deallocation timeline.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.AllocationRecord allocation_records = 6;</code>
-     */
-    public Builder setAllocationRecords(
-        int index, org.tensorflow.proto.framework.AllocationRecord.Builder builderForValue) {
-      if (allocationRecordsBuilder_ == null) {
-        ensureAllocationRecordsIsMutable();
-        allocationRecords_.set(index, builderForValue.build());
-        onChanged();
-      } else {
-        allocationRecordsBuilder_.setMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * The allocation and deallocation timeline.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.AllocationRecord allocation_records = 6;</code>
-     */
-    public Builder addAllocationRecords(org.tensorflow.proto.framework.AllocationRecord value) {
-      if (allocationRecordsBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureAllocationRecordsIsMutable();
-        allocationRecords_.add(value);
-        onChanged();
-      } else {
-        allocationRecordsBuilder_.addMessage(value);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * The allocation and deallocation timeline.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.AllocationRecord allocation_records = 6;</code>
-     */
-    public Builder addAllocationRecords(
-        int index, org.tensorflow.proto.framework.AllocationRecord value) {
-      if (allocationRecordsBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureAllocationRecordsIsMutable();
-        allocationRecords_.add(index, value);
-        onChanged();
-      } else {
-        allocationRecordsBuilder_.addMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * The allocation and deallocation timeline.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.AllocationRecord allocation_records = 6;</code>
-     */
-    public Builder addAllocationRecords(
-        org.tensorflow.proto.framework.AllocationRecord.Builder builderForValue) {
-      if (allocationRecordsBuilder_ == null) {
-        ensureAllocationRecordsIsMutable();
-        allocationRecords_.add(builderForValue.build());
-        onChanged();
-      } else {
-        allocationRecordsBuilder_.addMessage(builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * The allocation and deallocation timeline.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.AllocationRecord allocation_records = 6;</code>
-     */
-    public Builder addAllocationRecords(
-        int index, org.tensorflow.proto.framework.AllocationRecord.Builder builderForValue) {
-      if (allocationRecordsBuilder_ == null) {
-        ensureAllocationRecordsIsMutable();
-        allocationRecords_.add(index, builderForValue.build());
-        onChanged();
-      } else {
-        allocationRecordsBuilder_.addMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * The allocation and deallocation timeline.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.AllocationRecord allocation_records = 6;</code>
-     */
-    public Builder addAllAllocationRecords(
-        java.lang.Iterable<? extends org.tensorflow.proto.framework.AllocationRecord> values) {
-      if (allocationRecordsBuilder_ == null) {
-        ensureAllocationRecordsIsMutable();
-        com.google.protobuf.AbstractMessageLite.Builder.addAll(
-            values, allocationRecords_);
-        onChanged();
-      } else {
-        allocationRecordsBuilder_.addAllMessages(values);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * The allocation and deallocation timeline.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.AllocationRecord allocation_records = 6;</code>
-     */
-    public Builder clearAllocationRecords() {
-      if (allocationRecordsBuilder_ == null) {
-        allocationRecords_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000001);
-        onChanged();
-      } else {
-        allocationRecordsBuilder_.clear();
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * The allocation and deallocation timeline.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.AllocationRecord allocation_records = 6;</code>
-     */
-    public Builder removeAllocationRecords(int index) {
-      if (allocationRecordsBuilder_ == null) {
-        ensureAllocationRecordsIsMutable();
-        allocationRecords_.remove(index);
-        onChanged();
-      } else {
-        allocationRecordsBuilder_.remove(index);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * The allocation and deallocation timeline.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.AllocationRecord allocation_records = 6;</code>
-     */
-    public org.tensorflow.proto.framework.AllocationRecord.Builder getAllocationRecordsBuilder(
-        int index) {
-      return getAllocationRecordsFieldBuilder().getBuilder(index);
-    }
-    /**
-     * <pre>
-     * The allocation and deallocation timeline.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.AllocationRecord allocation_records = 6;</code>
-     */
-    public org.tensorflow.proto.framework.AllocationRecordOrBuilder getAllocationRecordsOrBuilder(
-        int index) {
-      if (allocationRecordsBuilder_ == null) {
-        return allocationRecords_.get(index);  } else {
-        return allocationRecordsBuilder_.getMessageOrBuilder(index);
-      }
-    }
-    /**
-     * <pre>
-     * The allocation and deallocation timeline.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.AllocationRecord allocation_records = 6;</code>
-     */
-    public java.util.List<? extends org.tensorflow.proto.framework.AllocationRecordOrBuilder> 
-         getAllocationRecordsOrBuilderList() {
-      if (allocationRecordsBuilder_ != null) {
-        return allocationRecordsBuilder_.getMessageOrBuilderList();
-      } else {
-        return java.util.Collections.unmodifiableList(allocationRecords_);
-      }
-    }
-    /**
-     * <pre>
-     * The allocation and deallocation timeline.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.AllocationRecord allocation_records = 6;</code>
-     */
-    public org.tensorflow.proto.framework.AllocationRecord.Builder addAllocationRecordsBuilder() {
-      return getAllocationRecordsFieldBuilder().addBuilder(
-          org.tensorflow.proto.framework.AllocationRecord.getDefaultInstance());
-    }
-    /**
-     * <pre>
-     * The allocation and deallocation timeline.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.AllocationRecord allocation_records = 6;</code>
-     */
-    public org.tensorflow.proto.framework.AllocationRecord.Builder addAllocationRecordsBuilder(
-        int index) {
-      return getAllocationRecordsFieldBuilder().addBuilder(
-          index, org.tensorflow.proto.framework.AllocationRecord.getDefaultInstance());
-    }
-    /**
-     * <pre>
-     * The allocation and deallocation timeline.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.AllocationRecord allocation_records = 6;</code>
-     */
-    public java.util.List<org.tensorflow.proto.framework.AllocationRecord.Builder> 
-         getAllocationRecordsBuilderList() {
-      return getAllocationRecordsFieldBuilder().getBuilderList();
-    }
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.framework.AllocationRecord, org.tensorflow.proto.framework.AllocationRecord.Builder, org.tensorflow.proto.framework.AllocationRecordOrBuilder> 
-        getAllocationRecordsFieldBuilder() {
-      if (allocationRecordsBuilder_ == null) {
-        allocationRecordsBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
-            org.tensorflow.proto.framework.AllocationRecord, org.tensorflow.proto.framework.AllocationRecord.Builder, org.tensorflow.proto.framework.AllocationRecordOrBuilder>(
-                allocationRecords_,
-                ((bitField0_ & 0x00000001) != 0),
-                getParentForChildren(),
-                isClean());
-        allocationRecords_ = null;
-      }
-      return allocationRecordsBuilder_;
-    }
-
-    private long allocatorBytesInUse_ ;
-    /**
-     * <pre>
-     * These are snapshots of the overall allocator memory stats.
-     * The number of live bytes currently allocated by the allocator.
-     * </pre>
-     *
-     * <code>int64 allocator_bytes_in_use = 5;</code>
-     */
-    public long getAllocatorBytesInUse() {
-      return allocatorBytesInUse_;
-    }
-    /**
-     * <pre>
-     * These are snapshots of the overall allocator memory stats.
-     * The number of live bytes currently allocated by the allocator.
-     * </pre>
-     *
-     * <code>int64 allocator_bytes_in_use = 5;</code>
-     */
-    public Builder setAllocatorBytesInUse(long value) {
-      
-      allocatorBytesInUse_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * These are snapshots of the overall allocator memory stats.
-     * The number of live bytes currently allocated by the allocator.
-     * </pre>
-     *
-     * <code>int64 allocator_bytes_in_use = 5;</code>
-     */
-    public Builder clearAllocatorBytesInUse() {
-      
-      allocatorBytesInUse_ = 0L;
-      onChanged();
-      return this;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.AllocatorMemoryUsed)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.AllocatorMemoryUsed)
-  private static final org.tensorflow.proto.framework.AllocatorMemoryUsed DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.AllocatorMemoryUsed();
-  }
-
-  public static org.tensorflow.proto.framework.AllocatorMemoryUsed getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<AllocatorMemoryUsed>
-      PARSER = new com.google.protobuf.AbstractParser<AllocatorMemoryUsed>() {
-    @java.lang.Override
-    public AllocatorMemoryUsed parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new AllocatorMemoryUsed(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<AllocatorMemoryUsed> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<AllocatorMemoryUsed> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.framework.AllocatorMemoryUsed getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/AllocatorMemoryUsedOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/AllocatorMemoryUsedOrBuilder.java
deleted file mode 100644
index 44dc47fa31d..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/AllocatorMemoryUsedOrBuilder.java
+++ /dev/null
@@ -1,96 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/framework/step_stats.proto
-
-package org.tensorflow.proto.framework;
-
-public interface AllocatorMemoryUsedOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.AllocatorMemoryUsed)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <code>string allocator_name = 1;</code>
-   */
-  java.lang.String getAllocatorName();
-  /**
-   * <code>string allocator_name = 1;</code>
-   */
-  com.google.protobuf.ByteString
-      getAllocatorNameBytes();
-
-  /**
-   * <pre>
-   * These are per-node allocator memory stats.
-   * </pre>
-   *
-   * <code>int64 total_bytes = 2;</code>
-   */
-  long getTotalBytes();
-
-  /**
-   * <code>int64 peak_bytes = 3;</code>
-   */
-  long getPeakBytes();
-
-  /**
-   * <pre>
-   * The bytes that are not deallocated.
-   * </pre>
-   *
-   * <code>int64 live_bytes = 4;</code>
-   */
-  long getLiveBytes();
-
-  /**
-   * <pre>
-   * The allocation and deallocation timeline.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.AllocationRecord allocation_records = 6;</code>
-   */
-  java.util.List<org.tensorflow.proto.framework.AllocationRecord> 
-      getAllocationRecordsList();
-  /**
-   * <pre>
-   * The allocation and deallocation timeline.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.AllocationRecord allocation_records = 6;</code>
-   */
-  org.tensorflow.proto.framework.AllocationRecord getAllocationRecords(int index);
-  /**
-   * <pre>
-   * The allocation and deallocation timeline.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.AllocationRecord allocation_records = 6;</code>
-   */
-  int getAllocationRecordsCount();
-  /**
-   * <pre>
-   * The allocation and deallocation timeline.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.AllocationRecord allocation_records = 6;</code>
-   */
-  java.util.List<? extends org.tensorflow.proto.framework.AllocationRecordOrBuilder> 
-      getAllocationRecordsOrBuilderList();
-  /**
-   * <pre>
-   * The allocation and deallocation timeline.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.AllocationRecord allocation_records = 6;</code>
-   */
-  org.tensorflow.proto.framework.AllocationRecordOrBuilder getAllocationRecordsOrBuilder(
-      int index);
-
-  /**
-   * <pre>
-   * These are snapshots of the overall allocator memory stats.
-   * The number of live bytes currently allocated by the allocator.
-   * </pre>
-   *
-   * <code>int64 allocator_bytes_in_use = 5;</code>
-   */
-  long getAllocatorBytesInUse();
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ApiDef.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ApiDef.java
deleted file mode 100644
index 99916fcf1f2..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ApiDef.java
+++ /dev/null
@@ -1,6303 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/framework/api_def.proto
-
-package org.tensorflow.proto.framework;
-
-/**
- * <pre>
- * Used to specify and override the default API &amp; behavior in the
- * generated code for client languages, from what you would get from
- * the OpDef alone. There will be a set of ApiDefs that are common
- * to all client languages, and another set per client language.
- * The per-client-language ApiDefs will inherit values from the
- * common ApiDefs which it can either replace or modify.
- * We separate the API definition from the OpDef so we can evolve the
- * API while remaining backwards compatible when interpretting old
- * graphs.  Overrides go in an "api_def.pbtxt" file with a text-format
- * ApiDefs message.
- * WARNING: Be *very* careful changing the API for any existing op --
- * you can change the semantics of existing code.  These changes may
- * need to wait until a major release of TensorFlow to avoid breaking
- * our compatibility promises.
- * </pre>
- *
- * Protobuf type {@code tensorflow.ApiDef}
- */
-public  final class ApiDef extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.ApiDef)
-    ApiDefOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use ApiDef.newBuilder() to construct.
-  private ApiDef(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private ApiDef() {
-    graphOpName_ = "";
-    deprecationMessage_ = "";
-    visibility_ = 0;
-    endpoint_ = java.util.Collections.emptyList();
-    inArg_ = java.util.Collections.emptyList();
-    outArg_ = java.util.Collections.emptyList();
-    argOrder_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-    attr_ = java.util.Collections.emptyList();
-    summary_ = "";
-    description_ = "";
-    descriptionPrefix_ = "";
-    descriptionSuffix_ = "";
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new ApiDef();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private ApiDef(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    int mutable_bitField0_ = 0;
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 10: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            graphOpName_ = s;
-            break;
-          }
-          case 16: {
-            int rawValue = input.readEnum();
-
-            visibility_ = rawValue;
-            break;
-          }
-          case 26: {
-            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
-              endpoint_ = new java.util.ArrayList<org.tensorflow.proto.framework.ApiDef.Endpoint>();
-              mutable_bitField0_ |= 0x00000001;
-            }
-            endpoint_.add(
-                input.readMessage(org.tensorflow.proto.framework.ApiDef.Endpoint.parser(), extensionRegistry));
-            break;
-          }
-          case 34: {
-            if (!((mutable_bitField0_ & 0x00000002) != 0)) {
-              inArg_ = new java.util.ArrayList<org.tensorflow.proto.framework.ApiDef.Arg>();
-              mutable_bitField0_ |= 0x00000002;
-            }
-            inArg_.add(
-                input.readMessage(org.tensorflow.proto.framework.ApiDef.Arg.parser(), extensionRegistry));
-            break;
-          }
-          case 42: {
-            if (!((mutable_bitField0_ & 0x00000004) != 0)) {
-              outArg_ = new java.util.ArrayList<org.tensorflow.proto.framework.ApiDef.Arg>();
-              mutable_bitField0_ |= 0x00000004;
-            }
-            outArg_.add(
-                input.readMessage(org.tensorflow.proto.framework.ApiDef.Arg.parser(), extensionRegistry));
-            break;
-          }
-          case 50: {
-            if (!((mutable_bitField0_ & 0x00000010) != 0)) {
-              attr_ = new java.util.ArrayList<org.tensorflow.proto.framework.ApiDef.Attr>();
-              mutable_bitField0_ |= 0x00000010;
-            }
-            attr_.add(
-                input.readMessage(org.tensorflow.proto.framework.ApiDef.Attr.parser(), extensionRegistry));
-            break;
-          }
-          case 58: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            summary_ = s;
-            break;
-          }
-          case 66: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            description_ = s;
-            break;
-          }
-          case 74: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            descriptionPrefix_ = s;
-            break;
-          }
-          case 82: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            descriptionSuffix_ = s;
-            break;
-          }
-          case 90: {
-            java.lang.String s = input.readStringRequireUtf8();
-            if (!((mutable_bitField0_ & 0x00000008) != 0)) {
-              argOrder_ = new com.google.protobuf.LazyStringArrayList();
-              mutable_bitField0_ |= 0x00000008;
-            }
-            argOrder_.add(s);
-            break;
-          }
-          case 98: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            deprecationMessage_ = s;
-            break;
-          }
-          case 104: {
-
-            deprecationVersion_ = input.readInt32();
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      if (((mutable_bitField0_ & 0x00000001) != 0)) {
-        endpoint_ = java.util.Collections.unmodifiableList(endpoint_);
-      }
-      if (((mutable_bitField0_ & 0x00000002) != 0)) {
-        inArg_ = java.util.Collections.unmodifiableList(inArg_);
-      }
-      if (((mutable_bitField0_ & 0x00000004) != 0)) {
-        outArg_ = java.util.Collections.unmodifiableList(outArg_);
-      }
-      if (((mutable_bitField0_ & 0x00000010) != 0)) {
-        attr_ = java.util.Collections.unmodifiableList(attr_);
-      }
-      if (((mutable_bitField0_ & 0x00000008) != 0)) {
-        argOrder_ = argOrder_.getUnmodifiableView();
-      }
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.framework.ApiDefProtos.internal_static_tensorflow_ApiDef_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.framework.ApiDefProtos.internal_static_tensorflow_ApiDef_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.framework.ApiDef.class, org.tensorflow.proto.framework.ApiDef.Builder.class);
-  }
-
-  /**
-   * Protobuf enum {@code tensorflow.ApiDef.Visibility}
-   */
-  public enum Visibility
-      implements com.google.protobuf.ProtocolMessageEnum {
-    /**
-     * <pre>
-     * Normally this is "VISIBLE" unless you are inheriting a
-     * different value from another ApiDef.
-     * </pre>
-     *
-     * <code>DEFAULT_VISIBILITY = 0;</code>
-     */
-    DEFAULT_VISIBILITY(0),
-    /**
-     * <pre>
-     * Publicly visible in the API.
-     * </pre>
-     *
-     * <code>VISIBLE = 1;</code>
-     */
-    VISIBLE(1),
-    /**
-     * <pre>
-     * Do not include this op in the generated API. If visibility is
-     * set to 'SKIP', other fields are ignored for this op.
-     * </pre>
-     *
-     * <code>SKIP = 2;</code>
-     */
-    SKIP(2),
-    /**
-     * <pre>
-     * Hide this op by putting it into an internal namespace (or whatever
-     * is appropriate in the target language).
-     * </pre>
-     *
-     * <code>HIDDEN = 3;</code>
-     */
-    HIDDEN(3),
-    UNRECOGNIZED(-1),
-    ;
-
-    /**
-     * <pre>
-     * Normally this is "VISIBLE" unless you are inheriting a
-     * different value from another ApiDef.
-     * </pre>
-     *
-     * <code>DEFAULT_VISIBILITY = 0;</code>
-     */
-    public static final int DEFAULT_VISIBILITY_VALUE = 0;
-    /**
-     * <pre>
-     * Publicly visible in the API.
-     * </pre>
-     *
-     * <code>VISIBLE = 1;</code>
-     */
-    public static final int VISIBLE_VALUE = 1;
-    /**
-     * <pre>
-     * Do not include this op in the generated API. If visibility is
-     * set to 'SKIP', other fields are ignored for this op.
-     * </pre>
-     *
-     * <code>SKIP = 2;</code>
-     */
-    public static final int SKIP_VALUE = 2;
-    /**
-     * <pre>
-     * Hide this op by putting it into an internal namespace (or whatever
-     * is appropriate in the target language).
-     * </pre>
-     *
-     * <code>HIDDEN = 3;</code>
-     */
-    public static final int HIDDEN_VALUE = 3;
-
-
-    public final int getNumber() {
-      if (this == UNRECOGNIZED) {
-        throw new java.lang.IllegalArgumentException(
-            "Can't get the number of an unknown enum value.");
-      }
-      return value;
-    }
-
-    /**
-     * @deprecated Use {@link #forNumber(int)} instead.
-     */
-    @java.lang.Deprecated
-    public static Visibility valueOf(int value) {
-      return forNumber(value);
-    }
-
-    public static Visibility forNumber(int value) {
-      switch (value) {
-        case 0: return DEFAULT_VISIBILITY;
-        case 1: return VISIBLE;
-        case 2: return SKIP;
-        case 3: return HIDDEN;
-        default: return null;
-      }
-    }
-
-    public static com.google.protobuf.Internal.EnumLiteMap<Visibility>
-        internalGetValueMap() {
-      return internalValueMap;
-    }
-    private static final com.google.protobuf.Internal.EnumLiteMap<
-        Visibility> internalValueMap =
-          new com.google.protobuf.Internal.EnumLiteMap<Visibility>() {
-            public Visibility findValueByNumber(int number) {
-              return Visibility.forNumber(number);
-            }
-          };
-
-    public final com.google.protobuf.Descriptors.EnumValueDescriptor
-        getValueDescriptor() {
-      return getDescriptor().getValues().get(ordinal());
-    }
-    public final com.google.protobuf.Descriptors.EnumDescriptor
-        getDescriptorForType() {
-      return getDescriptor();
-    }
-    public static final com.google.protobuf.Descriptors.EnumDescriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.ApiDef.getDescriptor().getEnumTypes().get(0);
-    }
-
-    private static final Visibility[] VALUES = values();
-
-    public static Visibility valueOf(
-        com.google.protobuf.Descriptors.EnumValueDescriptor desc) {
-      if (desc.getType() != getDescriptor()) {
-        throw new java.lang.IllegalArgumentException(
-          "EnumValueDescriptor is not for this type.");
-      }
-      if (desc.getIndex() == -1) {
-        return UNRECOGNIZED;
-      }
-      return VALUES[desc.getIndex()];
-    }
-
-    private final int value;
-
-    private Visibility(int value) {
-      this.value = value;
-    }
-
-    // @@protoc_insertion_point(enum_scope:tensorflow.ApiDef.Visibility)
-  }
-
-  public interface EndpointOrBuilder extends
-      // @@protoc_insertion_point(interface_extends:tensorflow.ApiDef.Endpoint)
-      com.google.protobuf.MessageOrBuilder {
-
-    /**
-     * <pre>
-     * Name should be either like "CamelCaseName" or
-     * "Package.CamelCaseName". Client-language-specific ApiDefs may
-     * use a snake_case convention instead of CamelCase.
-     * </pre>
-     *
-     * <code>string name = 1;</code>
-     */
-    java.lang.String getName();
-    /**
-     * <pre>
-     * Name should be either like "CamelCaseName" or
-     * "Package.CamelCaseName". Client-language-specific ApiDefs may
-     * use a snake_case convention instead of CamelCase.
-     * </pre>
-     *
-     * <code>string name = 1;</code>
-     */
-    com.google.protobuf.ByteString
-        getNameBytes();
-
-    /**
-     * <pre>
-     * Set if this endpoint is deprecated. If set to true, a message suggesting
-     * to use a non-deprecated endpoint instead will be printed. If all
-     * endpoints are deprecated, set deprecation_message in ApiDef instead.
-     * </pre>
-     *
-     * <code>bool deprecated = 3;</code>
-     */
-    boolean getDeprecated();
-
-    /**
-     * <pre>
-     * Major version when an endpoint will be deleted. For e.g. set this
-     * value to 2 if endpoint should be removed in TensorFlow 2.0 and
-     * deprecated in versions before that.
-     * </pre>
-     *
-     * <code>int32 deprecation_version = 4;</code>
-     */
-    int getDeprecationVersion();
-  }
-  /**
-   * <pre>
-   * If you specify any endpoint, this will replace all of the
-   * inherited endpoints.  The first endpoint should be the
-   * "canonical" endpoint, and should not be deprecated (unless all
-   * endpoints are deprecated).
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.ApiDef.Endpoint}
-   */
-  public  static final class Endpoint extends
-      com.google.protobuf.GeneratedMessageV3 implements
-      // @@protoc_insertion_point(message_implements:tensorflow.ApiDef.Endpoint)
-      EndpointOrBuilder {
-  private static final long serialVersionUID = 0L;
-    // Use Endpoint.newBuilder() to construct.
-    private Endpoint(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-      super(builder);
-    }
-    private Endpoint() {
-      name_ = "";
-    }
-
-    @java.lang.Override
-    @SuppressWarnings({"unused"})
-    protected java.lang.Object newInstance(
-        UnusedPrivateParameter unused) {
-      return new Endpoint();
-    }
-
-    @java.lang.Override
-    public final com.google.protobuf.UnknownFieldSet
-    getUnknownFields() {
-      return this.unknownFields;
-    }
-    private Endpoint(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      this();
-      if (extensionRegistry == null) {
-        throw new java.lang.NullPointerException();
-      }
-      com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-          com.google.protobuf.UnknownFieldSet.newBuilder();
-      try {
-        boolean done = false;
-        while (!done) {
-          int tag = input.readTag();
-          switch (tag) {
-            case 0:
-              done = true;
-              break;
-            case 10: {
-              java.lang.String s = input.readStringRequireUtf8();
-
-              name_ = s;
-              break;
-            }
-            case 24: {
-
-              deprecated_ = input.readBool();
-              break;
-            }
-            case 32: {
-
-              deprecationVersion_ = input.readInt32();
-              break;
-            }
-            default: {
-              if (!parseUnknownField(
-                  input, unknownFields, extensionRegistry, tag)) {
-                done = true;
-              }
-              break;
-            }
-          }
-        }
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        throw e.setUnfinishedMessage(this);
-      } catch (java.io.IOException e) {
-        throw new com.google.protobuf.InvalidProtocolBufferException(
-            e).setUnfinishedMessage(this);
-      } finally {
-        this.unknownFields = unknownFields.build();
-        makeExtensionsImmutable();
-      }
-    }
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.ApiDefProtos.internal_static_tensorflow_ApiDef_Endpoint_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.ApiDefProtos.internal_static_tensorflow_ApiDef_Endpoint_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.ApiDef.Endpoint.class, org.tensorflow.proto.framework.ApiDef.Endpoint.Builder.class);
-    }
-
-    public static final int NAME_FIELD_NUMBER = 1;
-    private volatile java.lang.Object name_;
-    /**
-     * <pre>
-     * Name should be either like "CamelCaseName" or
-     * "Package.CamelCaseName". Client-language-specific ApiDefs may
-     * use a snake_case convention instead of CamelCase.
-     * </pre>
-     *
-     * <code>string name = 1;</code>
-     */
-    public java.lang.String getName() {
-      java.lang.Object ref = name_;
-      if (ref instanceof java.lang.String) {
-        return (java.lang.String) ref;
-      } else {
-        com.google.protobuf.ByteString bs = 
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        name_ = s;
-        return s;
-      }
-    }
-    /**
-     * <pre>
-     * Name should be either like "CamelCaseName" or
-     * "Package.CamelCaseName". Client-language-specific ApiDefs may
-     * use a snake_case convention instead of CamelCase.
-     * </pre>
-     *
-     * <code>string name = 1;</code>
-     */
-    public com.google.protobuf.ByteString
-        getNameBytes() {
-      java.lang.Object ref = name_;
-      if (ref instanceof java.lang.String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        name_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-
-    public static final int DEPRECATED_FIELD_NUMBER = 3;
-    private boolean deprecated_;
-    /**
-     * <pre>
-     * Set if this endpoint is deprecated. If set to true, a message suggesting
-     * to use a non-deprecated endpoint instead will be printed. If all
-     * endpoints are deprecated, set deprecation_message in ApiDef instead.
-     * </pre>
-     *
-     * <code>bool deprecated = 3;</code>
-     */
-    public boolean getDeprecated() {
-      return deprecated_;
-    }
-
-    public static final int DEPRECATION_VERSION_FIELD_NUMBER = 4;
-    private int deprecationVersion_;
-    /**
-     * <pre>
-     * Major version when an endpoint will be deleted. For e.g. set this
-     * value to 2 if endpoint should be removed in TensorFlow 2.0 and
-     * deprecated in versions before that.
-     * </pre>
-     *
-     * <code>int32 deprecation_version = 4;</code>
-     */
-    public int getDeprecationVersion() {
-      return deprecationVersion_;
-    }
-
-    private byte memoizedIsInitialized = -1;
-    @java.lang.Override
-    public final boolean isInitialized() {
-      byte isInitialized = memoizedIsInitialized;
-      if (isInitialized == 1) return true;
-      if (isInitialized == 0) return false;
-
-      memoizedIsInitialized = 1;
-      return true;
-    }
-
-    @java.lang.Override
-    public void writeTo(com.google.protobuf.CodedOutputStream output)
-                        throws java.io.IOException {
-      if (!getNameBytes().isEmpty()) {
-        com.google.protobuf.GeneratedMessageV3.writeString(output, 1, name_);
-      }
-      if (deprecated_ != false) {
-        output.writeBool(3, deprecated_);
-      }
-      if (deprecationVersion_ != 0) {
-        output.writeInt32(4, deprecationVersion_);
-      }
-      unknownFields.writeTo(output);
-    }
-
-    @java.lang.Override
-    public int getSerializedSize() {
-      int size = memoizedSize;
-      if (size != -1) return size;
-
-      size = 0;
-      if (!getNameBytes().isEmpty()) {
-        size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, name_);
-      }
-      if (deprecated_ != false) {
-        size += com.google.protobuf.CodedOutputStream
-          .computeBoolSize(3, deprecated_);
-      }
-      if (deprecationVersion_ != 0) {
-        size += com.google.protobuf.CodedOutputStream
-          .computeInt32Size(4, deprecationVersion_);
-      }
-      size += unknownFields.getSerializedSize();
-      memoizedSize = size;
-      return size;
-    }
-
-    @java.lang.Override
-    public boolean equals(final java.lang.Object obj) {
-      if (obj == this) {
-       return true;
-      }
-      if (!(obj instanceof org.tensorflow.proto.framework.ApiDef.Endpoint)) {
-        return super.equals(obj);
-      }
-      org.tensorflow.proto.framework.ApiDef.Endpoint other = (org.tensorflow.proto.framework.ApiDef.Endpoint) obj;
-
-      if (!getName()
-          .equals(other.getName())) return false;
-      if (getDeprecated()
-          != other.getDeprecated()) return false;
-      if (getDeprecationVersion()
-          != other.getDeprecationVersion()) return false;
-      if (!unknownFields.equals(other.unknownFields)) return false;
-      return true;
-    }
-
-    @java.lang.Override
-    public int hashCode() {
-      if (memoizedHashCode != 0) {
-        return memoizedHashCode;
-      }
-      int hash = 41;
-      hash = (19 * hash) + getDescriptor().hashCode();
-      hash = (37 * hash) + NAME_FIELD_NUMBER;
-      hash = (53 * hash) + getName().hashCode();
-      hash = (37 * hash) + DEPRECATED_FIELD_NUMBER;
-      hash = (53 * hash) + com.google.protobuf.Internal.hashBoolean(
-          getDeprecated());
-      hash = (37 * hash) + DEPRECATION_VERSION_FIELD_NUMBER;
-      hash = (53 * hash) + getDeprecationVersion();
-      hash = (29 * hash) + unknownFields.hashCode();
-      memoizedHashCode = hash;
-      return hash;
-    }
-
-    public static org.tensorflow.proto.framework.ApiDef.Endpoint parseFrom(
-        java.nio.ByteBuffer data)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data);
-    }
-    public static org.tensorflow.proto.framework.ApiDef.Endpoint parseFrom(
-        java.nio.ByteBuffer data,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.ApiDef.Endpoint parseFrom(
-        com.google.protobuf.ByteString data)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data);
-    }
-    public static org.tensorflow.proto.framework.ApiDef.Endpoint parseFrom(
-        com.google.protobuf.ByteString data,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.ApiDef.Endpoint parseFrom(byte[] data)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data);
-    }
-    public static org.tensorflow.proto.framework.ApiDef.Endpoint parseFrom(
-        byte[] data,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.ApiDef.Endpoint parseFrom(java.io.InputStream input)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input);
-    }
-    public static org.tensorflow.proto.framework.ApiDef.Endpoint parseFrom(
-        java.io.InputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.ApiDef.Endpoint parseDelimitedFrom(java.io.InputStream input)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseDelimitedWithIOException(PARSER, input);
-    }
-    public static org.tensorflow.proto.framework.ApiDef.Endpoint parseDelimitedFrom(
-        java.io.InputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.ApiDef.Endpoint parseFrom(
-        com.google.protobuf.CodedInputStream input)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input);
-    }
-    public static org.tensorflow.proto.framework.ApiDef.Endpoint parseFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input, extensionRegistry);
-    }
-
-    @java.lang.Override
-    public Builder newBuilderForType() { return newBuilder(); }
-    public static Builder newBuilder() {
-      return DEFAULT_INSTANCE.toBuilder();
-    }
-    public static Builder newBuilder(org.tensorflow.proto.framework.ApiDef.Endpoint prototype) {
-      return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-    }
-    @java.lang.Override
-    public Builder toBuilder() {
-      return this == DEFAULT_INSTANCE
-          ? new Builder() : new Builder().mergeFrom(this);
-    }
-
-    @java.lang.Override
-    protected Builder newBuilderForType(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      Builder builder = new Builder(parent);
-      return builder;
-    }
-    /**
-     * <pre>
-     * If you specify any endpoint, this will replace all of the
-     * inherited endpoints.  The first endpoint should be the
-     * "canonical" endpoint, and should not be deprecated (unless all
-     * endpoints are deprecated).
-     * </pre>
-     *
-     * Protobuf type {@code tensorflow.ApiDef.Endpoint}
-     */
-    public static final class Builder extends
-        com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-        // @@protoc_insertion_point(builder_implements:tensorflow.ApiDef.Endpoint)
-        org.tensorflow.proto.framework.ApiDef.EndpointOrBuilder {
-      public static final com.google.protobuf.Descriptors.Descriptor
-          getDescriptor() {
-        return org.tensorflow.proto.framework.ApiDefProtos.internal_static_tensorflow_ApiDef_Endpoint_descriptor;
-      }
-
-      @java.lang.Override
-      protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-          internalGetFieldAccessorTable() {
-        return org.tensorflow.proto.framework.ApiDefProtos.internal_static_tensorflow_ApiDef_Endpoint_fieldAccessorTable
-            .ensureFieldAccessorsInitialized(
-                org.tensorflow.proto.framework.ApiDef.Endpoint.class, org.tensorflow.proto.framework.ApiDef.Endpoint.Builder.class);
-      }
-
-      // Construct using org.tensorflow.proto.framework.ApiDef.Endpoint.newBuilder()
-      private Builder() {
-        maybeForceBuilderInitialization();
-      }
-
-      private Builder(
-          com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-        super(parent);
-        maybeForceBuilderInitialization();
-      }
-      private void maybeForceBuilderInitialization() {
-        if (com.google.protobuf.GeneratedMessageV3
-                .alwaysUseFieldBuilders) {
-        }
-      }
-      @java.lang.Override
-      public Builder clear() {
-        super.clear();
-        name_ = "";
-
-        deprecated_ = false;
-
-        deprecationVersion_ = 0;
-
-        return this;
-      }
-
-      @java.lang.Override
-      public com.google.protobuf.Descriptors.Descriptor
-          getDescriptorForType() {
-        return org.tensorflow.proto.framework.ApiDefProtos.internal_static_tensorflow_ApiDef_Endpoint_descriptor;
-      }
-
-      @java.lang.Override
-      public org.tensorflow.proto.framework.ApiDef.Endpoint getDefaultInstanceForType() {
-        return org.tensorflow.proto.framework.ApiDef.Endpoint.getDefaultInstance();
-      }
-
-      @java.lang.Override
-      public org.tensorflow.proto.framework.ApiDef.Endpoint build() {
-        org.tensorflow.proto.framework.ApiDef.Endpoint result = buildPartial();
-        if (!result.isInitialized()) {
-          throw newUninitializedMessageException(result);
-        }
-        return result;
-      }
-
-      @java.lang.Override
-      public org.tensorflow.proto.framework.ApiDef.Endpoint buildPartial() {
-        org.tensorflow.proto.framework.ApiDef.Endpoint result = new org.tensorflow.proto.framework.ApiDef.Endpoint(this);
-        result.name_ = name_;
-        result.deprecated_ = deprecated_;
-        result.deprecationVersion_ = deprecationVersion_;
-        onBuilt();
-        return result;
-      }
-
-      @java.lang.Override
-      public Builder clone() {
-        return super.clone();
-      }
-      @java.lang.Override
-      public Builder setField(
-          com.google.protobuf.Descriptors.FieldDescriptor field,
-          java.lang.Object value) {
-        return super.setField(field, value);
-      }
-      @java.lang.Override
-      public Builder clearField(
-          com.google.protobuf.Descriptors.FieldDescriptor field) {
-        return super.clearField(field);
-      }
-      @java.lang.Override
-      public Builder clearOneof(
-          com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-        return super.clearOneof(oneof);
-      }
-      @java.lang.Override
-      public Builder setRepeatedField(
-          com.google.protobuf.Descriptors.FieldDescriptor field,
-          int index, java.lang.Object value) {
-        return super.setRepeatedField(field, index, value);
-      }
-      @java.lang.Override
-      public Builder addRepeatedField(
-          com.google.protobuf.Descriptors.FieldDescriptor field,
-          java.lang.Object value) {
-        return super.addRepeatedField(field, value);
-      }
-      @java.lang.Override
-      public Builder mergeFrom(com.google.protobuf.Message other) {
-        if (other instanceof org.tensorflow.proto.framework.ApiDef.Endpoint) {
-          return mergeFrom((org.tensorflow.proto.framework.ApiDef.Endpoint)other);
-        } else {
-          super.mergeFrom(other);
-          return this;
-        }
-      }
-
-      public Builder mergeFrom(org.tensorflow.proto.framework.ApiDef.Endpoint other) {
-        if (other == org.tensorflow.proto.framework.ApiDef.Endpoint.getDefaultInstance()) return this;
-        if (!other.getName().isEmpty()) {
-          name_ = other.name_;
-          onChanged();
-        }
-        if (other.getDeprecated() != false) {
-          setDeprecated(other.getDeprecated());
-        }
-        if (other.getDeprecationVersion() != 0) {
-          setDeprecationVersion(other.getDeprecationVersion());
-        }
-        this.mergeUnknownFields(other.unknownFields);
-        onChanged();
-        return this;
-      }
-
-      @java.lang.Override
-      public final boolean isInitialized() {
-        return true;
-      }
-
-      @java.lang.Override
-      public Builder mergeFrom(
-          com.google.protobuf.CodedInputStream input,
-          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-          throws java.io.IOException {
-        org.tensorflow.proto.framework.ApiDef.Endpoint parsedMessage = null;
-        try {
-          parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-        } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-          parsedMessage = (org.tensorflow.proto.framework.ApiDef.Endpoint) e.getUnfinishedMessage();
-          throw e.unwrapIOException();
-        } finally {
-          if (parsedMessage != null) {
-            mergeFrom(parsedMessage);
-          }
-        }
-        return this;
-      }
-
-      private java.lang.Object name_ = "";
-      /**
-       * <pre>
-       * Name should be either like "CamelCaseName" or
-       * "Package.CamelCaseName". Client-language-specific ApiDefs may
-       * use a snake_case convention instead of CamelCase.
-       * </pre>
-       *
-       * <code>string name = 1;</code>
-       */
-      public java.lang.String getName() {
-        java.lang.Object ref = name_;
-        if (!(ref instanceof java.lang.String)) {
-          com.google.protobuf.ByteString bs =
-              (com.google.protobuf.ByteString) ref;
-          java.lang.String s = bs.toStringUtf8();
-          name_ = s;
-          return s;
-        } else {
-          return (java.lang.String) ref;
-        }
-      }
-      /**
-       * <pre>
-       * Name should be either like "CamelCaseName" or
-       * "Package.CamelCaseName". Client-language-specific ApiDefs may
-       * use a snake_case convention instead of CamelCase.
-       * </pre>
-       *
-       * <code>string name = 1;</code>
-       */
-      public com.google.protobuf.ByteString
-          getNameBytes() {
-        java.lang.Object ref = name_;
-        if (ref instanceof String) {
-          com.google.protobuf.ByteString b = 
-              com.google.protobuf.ByteString.copyFromUtf8(
-                  (java.lang.String) ref);
-          name_ = b;
-          return b;
-        } else {
-          return (com.google.protobuf.ByteString) ref;
-        }
-      }
-      /**
-       * <pre>
-       * Name should be either like "CamelCaseName" or
-       * "Package.CamelCaseName". Client-language-specific ApiDefs may
-       * use a snake_case convention instead of CamelCase.
-       * </pre>
-       *
-       * <code>string name = 1;</code>
-       */
-      public Builder setName(
-          java.lang.String value) {
-        if (value == null) {
-    throw new NullPointerException();
-  }
-  
-        name_ = value;
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * Name should be either like "CamelCaseName" or
-       * "Package.CamelCaseName". Client-language-specific ApiDefs may
-       * use a snake_case convention instead of CamelCase.
-       * </pre>
-       *
-       * <code>string name = 1;</code>
-       */
-      public Builder clearName() {
-        
-        name_ = getDefaultInstance().getName();
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * Name should be either like "CamelCaseName" or
-       * "Package.CamelCaseName". Client-language-specific ApiDefs may
-       * use a snake_case convention instead of CamelCase.
-       * </pre>
-       *
-       * <code>string name = 1;</code>
-       */
-      public Builder setNameBytes(
-          com.google.protobuf.ByteString value) {
-        if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-        
-        name_ = value;
-        onChanged();
-        return this;
-      }
-
-      private boolean deprecated_ ;
-      /**
-       * <pre>
-       * Set if this endpoint is deprecated. If set to true, a message suggesting
-       * to use a non-deprecated endpoint instead will be printed. If all
-       * endpoints are deprecated, set deprecation_message in ApiDef instead.
-       * </pre>
-       *
-       * <code>bool deprecated = 3;</code>
-       */
-      public boolean getDeprecated() {
-        return deprecated_;
-      }
-      /**
-       * <pre>
-       * Set if this endpoint is deprecated. If set to true, a message suggesting
-       * to use a non-deprecated endpoint instead will be printed. If all
-       * endpoints are deprecated, set deprecation_message in ApiDef instead.
-       * </pre>
-       *
-       * <code>bool deprecated = 3;</code>
-       */
-      public Builder setDeprecated(boolean value) {
-        
-        deprecated_ = value;
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * Set if this endpoint is deprecated. If set to true, a message suggesting
-       * to use a non-deprecated endpoint instead will be printed. If all
-       * endpoints are deprecated, set deprecation_message in ApiDef instead.
-       * </pre>
-       *
-       * <code>bool deprecated = 3;</code>
-       */
-      public Builder clearDeprecated() {
-        
-        deprecated_ = false;
-        onChanged();
-        return this;
-      }
-
-      private int deprecationVersion_ ;
-      /**
-       * <pre>
-       * Major version when an endpoint will be deleted. For e.g. set this
-       * value to 2 if endpoint should be removed in TensorFlow 2.0 and
-       * deprecated in versions before that.
-       * </pre>
-       *
-       * <code>int32 deprecation_version = 4;</code>
-       */
-      public int getDeprecationVersion() {
-        return deprecationVersion_;
-      }
-      /**
-       * <pre>
-       * Major version when an endpoint will be deleted. For e.g. set this
-       * value to 2 if endpoint should be removed in TensorFlow 2.0 and
-       * deprecated in versions before that.
-       * </pre>
-       *
-       * <code>int32 deprecation_version = 4;</code>
-       */
-      public Builder setDeprecationVersion(int value) {
-        
-        deprecationVersion_ = value;
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * Major version when an endpoint will be deleted. For e.g. set this
-       * value to 2 if endpoint should be removed in TensorFlow 2.0 and
-       * deprecated in versions before that.
-       * </pre>
-       *
-       * <code>int32 deprecation_version = 4;</code>
-       */
-      public Builder clearDeprecationVersion() {
-        
-        deprecationVersion_ = 0;
-        onChanged();
-        return this;
-      }
-      @java.lang.Override
-      public final Builder setUnknownFields(
-          final com.google.protobuf.UnknownFieldSet unknownFields) {
-        return super.setUnknownFields(unknownFields);
-      }
-
-      @java.lang.Override
-      public final Builder mergeUnknownFields(
-          final com.google.protobuf.UnknownFieldSet unknownFields) {
-        return super.mergeUnknownFields(unknownFields);
-      }
-
-
-      // @@protoc_insertion_point(builder_scope:tensorflow.ApiDef.Endpoint)
-    }
-
-    // @@protoc_insertion_point(class_scope:tensorflow.ApiDef.Endpoint)
-    private static final org.tensorflow.proto.framework.ApiDef.Endpoint DEFAULT_INSTANCE;
-    static {
-      DEFAULT_INSTANCE = new org.tensorflow.proto.framework.ApiDef.Endpoint();
-    }
-
-    public static org.tensorflow.proto.framework.ApiDef.Endpoint getDefaultInstance() {
-      return DEFAULT_INSTANCE;
-    }
-
-    private static final com.google.protobuf.Parser<Endpoint>
-        PARSER = new com.google.protobuf.AbstractParser<Endpoint>() {
-      @java.lang.Override
-      public Endpoint parsePartialFrom(
-          com.google.protobuf.CodedInputStream input,
-          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-          throws com.google.protobuf.InvalidProtocolBufferException {
-        return new Endpoint(input, extensionRegistry);
-      }
-    };
-
-    public static com.google.protobuf.Parser<Endpoint> parser() {
-      return PARSER;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Parser<Endpoint> getParserForType() {
-      return PARSER;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.ApiDef.Endpoint getDefaultInstanceForType() {
-      return DEFAULT_INSTANCE;
-    }
-
-  }
-
-  public interface ArgOrBuilder extends
-      // @@protoc_insertion_point(interface_extends:tensorflow.ApiDef.Arg)
-      com.google.protobuf.MessageOrBuilder {
-
-    /**
-     * <code>string name = 1;</code>
-     */
-    java.lang.String getName();
-    /**
-     * <code>string name = 1;</code>
-     */
-    com.google.protobuf.ByteString
-        getNameBytes();
-
-    /**
-     * <pre>
-     * Change the name used to access this arg in the API from what
-     * is used in the GraphDef.  Note that these names in `backticks`
-     * will also be replaced in the summary &amp; description fields.
-     * </pre>
-     *
-     * <code>string rename_to = 2;</code>
-     */
-    java.lang.String getRenameTo();
-    /**
-     * <pre>
-     * Change the name used to access this arg in the API from what
-     * is used in the GraphDef.  Note that these names in `backticks`
-     * will also be replaced in the summary &amp; description fields.
-     * </pre>
-     *
-     * <code>string rename_to = 2;</code>
-     */
-    com.google.protobuf.ByteString
-        getRenameToBytes();
-
-    /**
-     * <pre>
-     * Note: this will replace any inherited arg doc. There is no
-     * current way of modifying arg descriptions (other than replacing
-     * them entirely) as can be done with op descriptions.
-     * </pre>
-     *
-     * <code>string description = 3;</code>
-     */
-    java.lang.String getDescription();
-    /**
-     * <pre>
-     * Note: this will replace any inherited arg doc. There is no
-     * current way of modifying arg descriptions (other than replacing
-     * them entirely) as can be done with op descriptions.
-     * </pre>
-     *
-     * <code>string description = 3;</code>
-     */
-    com.google.protobuf.ByteString
-        getDescriptionBytes();
-  }
-  /**
-   * Protobuf type {@code tensorflow.ApiDef.Arg}
-   */
-  public  static final class Arg extends
-      com.google.protobuf.GeneratedMessageV3 implements
-      // @@protoc_insertion_point(message_implements:tensorflow.ApiDef.Arg)
-      ArgOrBuilder {
-  private static final long serialVersionUID = 0L;
-    // Use Arg.newBuilder() to construct.
-    private Arg(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-      super(builder);
-    }
-    private Arg() {
-      name_ = "";
-      renameTo_ = "";
-      description_ = "";
-    }
-
-    @java.lang.Override
-    @SuppressWarnings({"unused"})
-    protected java.lang.Object newInstance(
-        UnusedPrivateParameter unused) {
-      return new Arg();
-    }
-
-    @java.lang.Override
-    public final com.google.protobuf.UnknownFieldSet
-    getUnknownFields() {
-      return this.unknownFields;
-    }
-    private Arg(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      this();
-      if (extensionRegistry == null) {
-        throw new java.lang.NullPointerException();
-      }
-      com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-          com.google.protobuf.UnknownFieldSet.newBuilder();
-      try {
-        boolean done = false;
-        while (!done) {
-          int tag = input.readTag();
-          switch (tag) {
-            case 0:
-              done = true;
-              break;
-            case 10: {
-              java.lang.String s = input.readStringRequireUtf8();
-
-              name_ = s;
-              break;
-            }
-            case 18: {
-              java.lang.String s = input.readStringRequireUtf8();
-
-              renameTo_ = s;
-              break;
-            }
-            case 26: {
-              java.lang.String s = input.readStringRequireUtf8();
-
-              description_ = s;
-              break;
-            }
-            default: {
-              if (!parseUnknownField(
-                  input, unknownFields, extensionRegistry, tag)) {
-                done = true;
-              }
-              break;
-            }
-          }
-        }
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        throw e.setUnfinishedMessage(this);
-      } catch (java.io.IOException e) {
-        throw new com.google.protobuf.InvalidProtocolBufferException(
-            e).setUnfinishedMessage(this);
-      } finally {
-        this.unknownFields = unknownFields.build();
-        makeExtensionsImmutable();
-      }
-    }
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.ApiDefProtos.internal_static_tensorflow_ApiDef_Arg_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.ApiDefProtos.internal_static_tensorflow_ApiDef_Arg_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.ApiDef.Arg.class, org.tensorflow.proto.framework.ApiDef.Arg.Builder.class);
-    }
-
-    public static final int NAME_FIELD_NUMBER = 1;
-    private volatile java.lang.Object name_;
-    /**
-     * <code>string name = 1;</code>
-     */
-    public java.lang.String getName() {
-      java.lang.Object ref = name_;
-      if (ref instanceof java.lang.String) {
-        return (java.lang.String) ref;
-      } else {
-        com.google.protobuf.ByteString bs = 
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        name_ = s;
-        return s;
-      }
-    }
-    /**
-     * <code>string name = 1;</code>
-     */
-    public com.google.protobuf.ByteString
-        getNameBytes() {
-      java.lang.Object ref = name_;
-      if (ref instanceof java.lang.String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        name_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-
-    public static final int RENAME_TO_FIELD_NUMBER = 2;
-    private volatile java.lang.Object renameTo_;
-    /**
-     * <pre>
-     * Change the name used to access this arg in the API from what
-     * is used in the GraphDef.  Note that these names in `backticks`
-     * will also be replaced in the summary &amp; description fields.
-     * </pre>
-     *
-     * <code>string rename_to = 2;</code>
-     */
-    public java.lang.String getRenameTo() {
-      java.lang.Object ref = renameTo_;
-      if (ref instanceof java.lang.String) {
-        return (java.lang.String) ref;
-      } else {
-        com.google.protobuf.ByteString bs = 
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        renameTo_ = s;
-        return s;
-      }
-    }
-    /**
-     * <pre>
-     * Change the name used to access this arg in the API from what
-     * is used in the GraphDef.  Note that these names in `backticks`
-     * will also be replaced in the summary &amp; description fields.
-     * </pre>
-     *
-     * <code>string rename_to = 2;</code>
-     */
-    public com.google.protobuf.ByteString
-        getRenameToBytes() {
-      java.lang.Object ref = renameTo_;
-      if (ref instanceof java.lang.String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        renameTo_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-
-    public static final int DESCRIPTION_FIELD_NUMBER = 3;
-    private volatile java.lang.Object description_;
-    /**
-     * <pre>
-     * Note: this will replace any inherited arg doc. There is no
-     * current way of modifying arg descriptions (other than replacing
-     * them entirely) as can be done with op descriptions.
-     * </pre>
-     *
-     * <code>string description = 3;</code>
-     */
-    public java.lang.String getDescription() {
-      java.lang.Object ref = description_;
-      if (ref instanceof java.lang.String) {
-        return (java.lang.String) ref;
-      } else {
-        com.google.protobuf.ByteString bs = 
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        description_ = s;
-        return s;
-      }
-    }
-    /**
-     * <pre>
-     * Note: this will replace any inherited arg doc. There is no
-     * current way of modifying arg descriptions (other than replacing
-     * them entirely) as can be done with op descriptions.
-     * </pre>
-     *
-     * <code>string description = 3;</code>
-     */
-    public com.google.protobuf.ByteString
-        getDescriptionBytes() {
-      java.lang.Object ref = description_;
-      if (ref instanceof java.lang.String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        description_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-
-    private byte memoizedIsInitialized = -1;
-    @java.lang.Override
-    public final boolean isInitialized() {
-      byte isInitialized = memoizedIsInitialized;
-      if (isInitialized == 1) return true;
-      if (isInitialized == 0) return false;
-
-      memoizedIsInitialized = 1;
-      return true;
-    }
-
-    @java.lang.Override
-    public void writeTo(com.google.protobuf.CodedOutputStream output)
-                        throws java.io.IOException {
-      if (!getNameBytes().isEmpty()) {
-        com.google.protobuf.GeneratedMessageV3.writeString(output, 1, name_);
-      }
-      if (!getRenameToBytes().isEmpty()) {
-        com.google.protobuf.GeneratedMessageV3.writeString(output, 2, renameTo_);
-      }
-      if (!getDescriptionBytes().isEmpty()) {
-        com.google.protobuf.GeneratedMessageV3.writeString(output, 3, description_);
-      }
-      unknownFields.writeTo(output);
-    }
-
-    @java.lang.Override
-    public int getSerializedSize() {
-      int size = memoizedSize;
-      if (size != -1) return size;
-
-      size = 0;
-      if (!getNameBytes().isEmpty()) {
-        size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, name_);
-      }
-      if (!getRenameToBytes().isEmpty()) {
-        size += com.google.protobuf.GeneratedMessageV3.computeStringSize(2, renameTo_);
-      }
-      if (!getDescriptionBytes().isEmpty()) {
-        size += com.google.protobuf.GeneratedMessageV3.computeStringSize(3, description_);
-      }
-      size += unknownFields.getSerializedSize();
-      memoizedSize = size;
-      return size;
-    }
-
-    @java.lang.Override
-    public boolean equals(final java.lang.Object obj) {
-      if (obj == this) {
-       return true;
-      }
-      if (!(obj instanceof org.tensorflow.proto.framework.ApiDef.Arg)) {
-        return super.equals(obj);
-      }
-      org.tensorflow.proto.framework.ApiDef.Arg other = (org.tensorflow.proto.framework.ApiDef.Arg) obj;
-
-      if (!getName()
-          .equals(other.getName())) return false;
-      if (!getRenameTo()
-          .equals(other.getRenameTo())) return false;
-      if (!getDescription()
-          .equals(other.getDescription())) return false;
-      if (!unknownFields.equals(other.unknownFields)) return false;
-      return true;
-    }
-
-    @java.lang.Override
-    public int hashCode() {
-      if (memoizedHashCode != 0) {
-        return memoizedHashCode;
-      }
-      int hash = 41;
-      hash = (19 * hash) + getDescriptor().hashCode();
-      hash = (37 * hash) + NAME_FIELD_NUMBER;
-      hash = (53 * hash) + getName().hashCode();
-      hash = (37 * hash) + RENAME_TO_FIELD_NUMBER;
-      hash = (53 * hash) + getRenameTo().hashCode();
-      hash = (37 * hash) + DESCRIPTION_FIELD_NUMBER;
-      hash = (53 * hash) + getDescription().hashCode();
-      hash = (29 * hash) + unknownFields.hashCode();
-      memoizedHashCode = hash;
-      return hash;
-    }
-
-    public static org.tensorflow.proto.framework.ApiDef.Arg parseFrom(
-        java.nio.ByteBuffer data)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data);
-    }
-    public static org.tensorflow.proto.framework.ApiDef.Arg parseFrom(
-        java.nio.ByteBuffer data,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.ApiDef.Arg parseFrom(
-        com.google.protobuf.ByteString data)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data);
-    }
-    public static org.tensorflow.proto.framework.ApiDef.Arg parseFrom(
-        com.google.protobuf.ByteString data,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.ApiDef.Arg parseFrom(byte[] data)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data);
-    }
-    public static org.tensorflow.proto.framework.ApiDef.Arg parseFrom(
-        byte[] data,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.ApiDef.Arg parseFrom(java.io.InputStream input)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input);
-    }
-    public static org.tensorflow.proto.framework.ApiDef.Arg parseFrom(
-        java.io.InputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.ApiDef.Arg parseDelimitedFrom(java.io.InputStream input)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseDelimitedWithIOException(PARSER, input);
-    }
-    public static org.tensorflow.proto.framework.ApiDef.Arg parseDelimitedFrom(
-        java.io.InputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.ApiDef.Arg parseFrom(
-        com.google.protobuf.CodedInputStream input)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input);
-    }
-    public static org.tensorflow.proto.framework.ApiDef.Arg parseFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input, extensionRegistry);
-    }
-
-    @java.lang.Override
-    public Builder newBuilderForType() { return newBuilder(); }
-    public static Builder newBuilder() {
-      return DEFAULT_INSTANCE.toBuilder();
-    }
-    public static Builder newBuilder(org.tensorflow.proto.framework.ApiDef.Arg prototype) {
-      return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-    }
-    @java.lang.Override
-    public Builder toBuilder() {
-      return this == DEFAULT_INSTANCE
-          ? new Builder() : new Builder().mergeFrom(this);
-    }
-
-    @java.lang.Override
-    protected Builder newBuilderForType(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      Builder builder = new Builder(parent);
-      return builder;
-    }
-    /**
-     * Protobuf type {@code tensorflow.ApiDef.Arg}
-     */
-    public static final class Builder extends
-        com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-        // @@protoc_insertion_point(builder_implements:tensorflow.ApiDef.Arg)
-        org.tensorflow.proto.framework.ApiDef.ArgOrBuilder {
-      public static final com.google.protobuf.Descriptors.Descriptor
-          getDescriptor() {
-        return org.tensorflow.proto.framework.ApiDefProtos.internal_static_tensorflow_ApiDef_Arg_descriptor;
-      }
-
-      @java.lang.Override
-      protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-          internalGetFieldAccessorTable() {
-        return org.tensorflow.proto.framework.ApiDefProtos.internal_static_tensorflow_ApiDef_Arg_fieldAccessorTable
-            .ensureFieldAccessorsInitialized(
-                org.tensorflow.proto.framework.ApiDef.Arg.class, org.tensorflow.proto.framework.ApiDef.Arg.Builder.class);
-      }
-
-      // Construct using org.tensorflow.proto.framework.ApiDef.Arg.newBuilder()
-      private Builder() {
-        maybeForceBuilderInitialization();
-      }
-
-      private Builder(
-          com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-        super(parent);
-        maybeForceBuilderInitialization();
-      }
-      private void maybeForceBuilderInitialization() {
-        if (com.google.protobuf.GeneratedMessageV3
-                .alwaysUseFieldBuilders) {
-        }
-      }
-      @java.lang.Override
-      public Builder clear() {
-        super.clear();
-        name_ = "";
-
-        renameTo_ = "";
-
-        description_ = "";
-
-        return this;
-      }
-
-      @java.lang.Override
-      public com.google.protobuf.Descriptors.Descriptor
-          getDescriptorForType() {
-        return org.tensorflow.proto.framework.ApiDefProtos.internal_static_tensorflow_ApiDef_Arg_descriptor;
-      }
-
-      @java.lang.Override
-      public org.tensorflow.proto.framework.ApiDef.Arg getDefaultInstanceForType() {
-        return org.tensorflow.proto.framework.ApiDef.Arg.getDefaultInstance();
-      }
-
-      @java.lang.Override
-      public org.tensorflow.proto.framework.ApiDef.Arg build() {
-        org.tensorflow.proto.framework.ApiDef.Arg result = buildPartial();
-        if (!result.isInitialized()) {
-          throw newUninitializedMessageException(result);
-        }
-        return result;
-      }
-
-      @java.lang.Override
-      public org.tensorflow.proto.framework.ApiDef.Arg buildPartial() {
-        org.tensorflow.proto.framework.ApiDef.Arg result = new org.tensorflow.proto.framework.ApiDef.Arg(this);
-        result.name_ = name_;
-        result.renameTo_ = renameTo_;
-        result.description_ = description_;
-        onBuilt();
-        return result;
-      }
-
-      @java.lang.Override
-      public Builder clone() {
-        return super.clone();
-      }
-      @java.lang.Override
-      public Builder setField(
-          com.google.protobuf.Descriptors.FieldDescriptor field,
-          java.lang.Object value) {
-        return super.setField(field, value);
-      }
-      @java.lang.Override
-      public Builder clearField(
-          com.google.protobuf.Descriptors.FieldDescriptor field) {
-        return super.clearField(field);
-      }
-      @java.lang.Override
-      public Builder clearOneof(
-          com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-        return super.clearOneof(oneof);
-      }
-      @java.lang.Override
-      public Builder setRepeatedField(
-          com.google.protobuf.Descriptors.FieldDescriptor field,
-          int index, java.lang.Object value) {
-        return super.setRepeatedField(field, index, value);
-      }
-      @java.lang.Override
-      public Builder addRepeatedField(
-          com.google.protobuf.Descriptors.FieldDescriptor field,
-          java.lang.Object value) {
-        return super.addRepeatedField(field, value);
-      }
-      @java.lang.Override
-      public Builder mergeFrom(com.google.protobuf.Message other) {
-        if (other instanceof org.tensorflow.proto.framework.ApiDef.Arg) {
-          return mergeFrom((org.tensorflow.proto.framework.ApiDef.Arg)other);
-        } else {
-          super.mergeFrom(other);
-          return this;
-        }
-      }
-
-      public Builder mergeFrom(org.tensorflow.proto.framework.ApiDef.Arg other) {
-        if (other == org.tensorflow.proto.framework.ApiDef.Arg.getDefaultInstance()) return this;
-        if (!other.getName().isEmpty()) {
-          name_ = other.name_;
-          onChanged();
-        }
-        if (!other.getRenameTo().isEmpty()) {
-          renameTo_ = other.renameTo_;
-          onChanged();
-        }
-        if (!other.getDescription().isEmpty()) {
-          description_ = other.description_;
-          onChanged();
-        }
-        this.mergeUnknownFields(other.unknownFields);
-        onChanged();
-        return this;
-      }
-
-      @java.lang.Override
-      public final boolean isInitialized() {
-        return true;
-      }
-
-      @java.lang.Override
-      public Builder mergeFrom(
-          com.google.protobuf.CodedInputStream input,
-          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-          throws java.io.IOException {
-        org.tensorflow.proto.framework.ApiDef.Arg parsedMessage = null;
-        try {
-          parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-        } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-          parsedMessage = (org.tensorflow.proto.framework.ApiDef.Arg) e.getUnfinishedMessage();
-          throw e.unwrapIOException();
-        } finally {
-          if (parsedMessage != null) {
-            mergeFrom(parsedMessage);
-          }
-        }
-        return this;
-      }
-
-      private java.lang.Object name_ = "";
-      /**
-       * <code>string name = 1;</code>
-       */
-      public java.lang.String getName() {
-        java.lang.Object ref = name_;
-        if (!(ref instanceof java.lang.String)) {
-          com.google.protobuf.ByteString bs =
-              (com.google.protobuf.ByteString) ref;
-          java.lang.String s = bs.toStringUtf8();
-          name_ = s;
-          return s;
-        } else {
-          return (java.lang.String) ref;
-        }
-      }
-      /**
-       * <code>string name = 1;</code>
-       */
-      public com.google.protobuf.ByteString
-          getNameBytes() {
-        java.lang.Object ref = name_;
-        if (ref instanceof String) {
-          com.google.protobuf.ByteString b = 
-              com.google.protobuf.ByteString.copyFromUtf8(
-                  (java.lang.String) ref);
-          name_ = b;
-          return b;
-        } else {
-          return (com.google.protobuf.ByteString) ref;
-        }
-      }
-      /**
-       * <code>string name = 1;</code>
-       */
-      public Builder setName(
-          java.lang.String value) {
-        if (value == null) {
-    throw new NullPointerException();
-  }
-  
-        name_ = value;
-        onChanged();
-        return this;
-      }
-      /**
-       * <code>string name = 1;</code>
-       */
-      public Builder clearName() {
-        
-        name_ = getDefaultInstance().getName();
-        onChanged();
-        return this;
-      }
-      /**
-       * <code>string name = 1;</code>
-       */
-      public Builder setNameBytes(
-          com.google.protobuf.ByteString value) {
-        if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-        
-        name_ = value;
-        onChanged();
-        return this;
-      }
-
-      private java.lang.Object renameTo_ = "";
-      /**
-       * <pre>
-       * Change the name used to access this arg in the API from what
-       * is used in the GraphDef.  Note that these names in `backticks`
-       * will also be replaced in the summary &amp; description fields.
-       * </pre>
-       *
-       * <code>string rename_to = 2;</code>
-       */
-      public java.lang.String getRenameTo() {
-        java.lang.Object ref = renameTo_;
-        if (!(ref instanceof java.lang.String)) {
-          com.google.protobuf.ByteString bs =
-              (com.google.protobuf.ByteString) ref;
-          java.lang.String s = bs.toStringUtf8();
-          renameTo_ = s;
-          return s;
-        } else {
-          return (java.lang.String) ref;
-        }
-      }
-      /**
-       * <pre>
-       * Change the name used to access this arg in the API from what
-       * is used in the GraphDef.  Note that these names in `backticks`
-       * will also be replaced in the summary &amp; description fields.
-       * </pre>
-       *
-       * <code>string rename_to = 2;</code>
-       */
-      public com.google.protobuf.ByteString
-          getRenameToBytes() {
-        java.lang.Object ref = renameTo_;
-        if (ref instanceof String) {
-          com.google.protobuf.ByteString b = 
-              com.google.protobuf.ByteString.copyFromUtf8(
-                  (java.lang.String) ref);
-          renameTo_ = b;
-          return b;
-        } else {
-          return (com.google.protobuf.ByteString) ref;
-        }
-      }
-      /**
-       * <pre>
-       * Change the name used to access this arg in the API from what
-       * is used in the GraphDef.  Note that these names in `backticks`
-       * will also be replaced in the summary &amp; description fields.
-       * </pre>
-       *
-       * <code>string rename_to = 2;</code>
-       */
-      public Builder setRenameTo(
-          java.lang.String value) {
-        if (value == null) {
-    throw new NullPointerException();
-  }
-  
-        renameTo_ = value;
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * Change the name used to access this arg in the API from what
-       * is used in the GraphDef.  Note that these names in `backticks`
-       * will also be replaced in the summary &amp; description fields.
-       * </pre>
-       *
-       * <code>string rename_to = 2;</code>
-       */
-      public Builder clearRenameTo() {
-        
-        renameTo_ = getDefaultInstance().getRenameTo();
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * Change the name used to access this arg in the API from what
-       * is used in the GraphDef.  Note that these names in `backticks`
-       * will also be replaced in the summary &amp; description fields.
-       * </pre>
-       *
-       * <code>string rename_to = 2;</code>
-       */
-      public Builder setRenameToBytes(
-          com.google.protobuf.ByteString value) {
-        if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-        
-        renameTo_ = value;
-        onChanged();
-        return this;
-      }
-
-      private java.lang.Object description_ = "";
-      /**
-       * <pre>
-       * Note: this will replace any inherited arg doc. There is no
-       * current way of modifying arg descriptions (other than replacing
-       * them entirely) as can be done with op descriptions.
-       * </pre>
-       *
-       * <code>string description = 3;</code>
-       */
-      public java.lang.String getDescription() {
-        java.lang.Object ref = description_;
-        if (!(ref instanceof java.lang.String)) {
-          com.google.protobuf.ByteString bs =
-              (com.google.protobuf.ByteString) ref;
-          java.lang.String s = bs.toStringUtf8();
-          description_ = s;
-          return s;
-        } else {
-          return (java.lang.String) ref;
-        }
-      }
-      /**
-       * <pre>
-       * Note: this will replace any inherited arg doc. There is no
-       * current way of modifying arg descriptions (other than replacing
-       * them entirely) as can be done with op descriptions.
-       * </pre>
-       *
-       * <code>string description = 3;</code>
-       */
-      public com.google.protobuf.ByteString
-          getDescriptionBytes() {
-        java.lang.Object ref = description_;
-        if (ref instanceof String) {
-          com.google.protobuf.ByteString b = 
-              com.google.protobuf.ByteString.copyFromUtf8(
-                  (java.lang.String) ref);
-          description_ = b;
-          return b;
-        } else {
-          return (com.google.protobuf.ByteString) ref;
-        }
-      }
-      /**
-       * <pre>
-       * Note: this will replace any inherited arg doc. There is no
-       * current way of modifying arg descriptions (other than replacing
-       * them entirely) as can be done with op descriptions.
-       * </pre>
-       *
-       * <code>string description = 3;</code>
-       */
-      public Builder setDescription(
-          java.lang.String value) {
-        if (value == null) {
-    throw new NullPointerException();
-  }
-  
-        description_ = value;
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * Note: this will replace any inherited arg doc. There is no
-       * current way of modifying arg descriptions (other than replacing
-       * them entirely) as can be done with op descriptions.
-       * </pre>
-       *
-       * <code>string description = 3;</code>
-       */
-      public Builder clearDescription() {
-        
-        description_ = getDefaultInstance().getDescription();
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * Note: this will replace any inherited arg doc. There is no
-       * current way of modifying arg descriptions (other than replacing
-       * them entirely) as can be done with op descriptions.
-       * </pre>
-       *
-       * <code>string description = 3;</code>
-       */
-      public Builder setDescriptionBytes(
-          com.google.protobuf.ByteString value) {
-        if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-        
-        description_ = value;
-        onChanged();
-        return this;
-      }
-      @java.lang.Override
-      public final Builder setUnknownFields(
-          final com.google.protobuf.UnknownFieldSet unknownFields) {
-        return super.setUnknownFields(unknownFields);
-      }
-
-      @java.lang.Override
-      public final Builder mergeUnknownFields(
-          final com.google.protobuf.UnknownFieldSet unknownFields) {
-        return super.mergeUnknownFields(unknownFields);
-      }
-
-
-      // @@protoc_insertion_point(builder_scope:tensorflow.ApiDef.Arg)
-    }
-
-    // @@protoc_insertion_point(class_scope:tensorflow.ApiDef.Arg)
-    private static final org.tensorflow.proto.framework.ApiDef.Arg DEFAULT_INSTANCE;
-    static {
-      DEFAULT_INSTANCE = new org.tensorflow.proto.framework.ApiDef.Arg();
-    }
-
-    public static org.tensorflow.proto.framework.ApiDef.Arg getDefaultInstance() {
-      return DEFAULT_INSTANCE;
-    }
-
-    private static final com.google.protobuf.Parser<Arg>
-        PARSER = new com.google.protobuf.AbstractParser<Arg>() {
-      @java.lang.Override
-      public Arg parsePartialFrom(
-          com.google.protobuf.CodedInputStream input,
-          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-          throws com.google.protobuf.InvalidProtocolBufferException {
-        return new Arg(input, extensionRegistry);
-      }
-    };
-
-    public static com.google.protobuf.Parser<Arg> parser() {
-      return PARSER;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Parser<Arg> getParserForType() {
-      return PARSER;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.ApiDef.Arg getDefaultInstanceForType() {
-      return DEFAULT_INSTANCE;
-    }
-
-  }
-
-  public interface AttrOrBuilder extends
-      // @@protoc_insertion_point(interface_extends:tensorflow.ApiDef.Attr)
-      com.google.protobuf.MessageOrBuilder {
-
-    /**
-     * <code>string name = 1;</code>
-     */
-    java.lang.String getName();
-    /**
-     * <code>string name = 1;</code>
-     */
-    com.google.protobuf.ByteString
-        getNameBytes();
-
-    /**
-     * <pre>
-     * Change the name used to access this attr in the API from what
-     * is used in the GraphDef.  Note that these names in `backticks`
-     * will also be replaced in the summary &amp; description fields.
-     * </pre>
-     *
-     * <code>string rename_to = 2;</code>
-     */
-    java.lang.String getRenameTo();
-    /**
-     * <pre>
-     * Change the name used to access this attr in the API from what
-     * is used in the GraphDef.  Note that these names in `backticks`
-     * will also be replaced in the summary &amp; description fields.
-     * </pre>
-     *
-     * <code>string rename_to = 2;</code>
-     */
-    com.google.protobuf.ByteString
-        getRenameToBytes();
-
-    /**
-     * <pre>
-     * Specify a new default value to use for this attr.  This default
-     * will be used when creating new graphs, as opposed to the
-     * default in the OpDef, which will be used when interpreting old
-     * GraphDefs.
-     * </pre>
-     *
-     * <code>.tensorflow.AttrValue default_value = 3;</code>
-     */
-    boolean hasDefaultValue();
-    /**
-     * <pre>
-     * Specify a new default value to use for this attr.  This default
-     * will be used when creating new graphs, as opposed to the
-     * default in the OpDef, which will be used when interpreting old
-     * GraphDefs.
-     * </pre>
-     *
-     * <code>.tensorflow.AttrValue default_value = 3;</code>
-     */
-    org.tensorflow.proto.framework.AttrValue getDefaultValue();
-    /**
-     * <pre>
-     * Specify a new default value to use for this attr.  This default
-     * will be used when creating new graphs, as opposed to the
-     * default in the OpDef, which will be used when interpreting old
-     * GraphDefs.
-     * </pre>
-     *
-     * <code>.tensorflow.AttrValue default_value = 3;</code>
-     */
-    org.tensorflow.proto.framework.AttrValueOrBuilder getDefaultValueOrBuilder();
-
-    /**
-     * <pre>
-     * Note: this will replace any inherited attr doc, there is no current
-     * way of modifying attr descriptions as can be done with op descriptions.
-     * </pre>
-     *
-     * <code>string description = 4;</code>
-     */
-    java.lang.String getDescription();
-    /**
-     * <pre>
-     * Note: this will replace any inherited attr doc, there is no current
-     * way of modifying attr descriptions as can be done with op descriptions.
-     * </pre>
-     *
-     * <code>string description = 4;</code>
-     */
-    com.google.protobuf.ByteString
-        getDescriptionBytes();
-  }
-  /**
-   * <pre>
-   * Description of the graph-construction-time configuration of this
-   * Op.  That is to say, this describes the attr fields that will
-   * be specified in the NodeDef.
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.ApiDef.Attr}
-   */
-  public  static final class Attr extends
-      com.google.protobuf.GeneratedMessageV3 implements
-      // @@protoc_insertion_point(message_implements:tensorflow.ApiDef.Attr)
-      AttrOrBuilder {
-  private static final long serialVersionUID = 0L;
-    // Use Attr.newBuilder() to construct.
-    private Attr(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-      super(builder);
-    }
-    private Attr() {
-      name_ = "";
-      renameTo_ = "";
-      description_ = "";
-    }
-
-    @java.lang.Override
-    @SuppressWarnings({"unused"})
-    protected java.lang.Object newInstance(
-        UnusedPrivateParameter unused) {
-      return new Attr();
-    }
-
-    @java.lang.Override
-    public final com.google.protobuf.UnknownFieldSet
-    getUnknownFields() {
-      return this.unknownFields;
-    }
-    private Attr(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      this();
-      if (extensionRegistry == null) {
-        throw new java.lang.NullPointerException();
-      }
-      com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-          com.google.protobuf.UnknownFieldSet.newBuilder();
-      try {
-        boolean done = false;
-        while (!done) {
-          int tag = input.readTag();
-          switch (tag) {
-            case 0:
-              done = true;
-              break;
-            case 10: {
-              java.lang.String s = input.readStringRequireUtf8();
-
-              name_ = s;
-              break;
-            }
-            case 18: {
-              java.lang.String s = input.readStringRequireUtf8();
-
-              renameTo_ = s;
-              break;
-            }
-            case 26: {
-              org.tensorflow.proto.framework.AttrValue.Builder subBuilder = null;
-              if (defaultValue_ != null) {
-                subBuilder = defaultValue_.toBuilder();
-              }
-              defaultValue_ = input.readMessage(org.tensorflow.proto.framework.AttrValue.parser(), extensionRegistry);
-              if (subBuilder != null) {
-                subBuilder.mergeFrom(defaultValue_);
-                defaultValue_ = subBuilder.buildPartial();
-              }
-
-              break;
-            }
-            case 34: {
-              java.lang.String s = input.readStringRequireUtf8();
-
-              description_ = s;
-              break;
-            }
-            default: {
-              if (!parseUnknownField(
-                  input, unknownFields, extensionRegistry, tag)) {
-                done = true;
-              }
-              break;
-            }
-          }
-        }
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        throw e.setUnfinishedMessage(this);
-      } catch (java.io.IOException e) {
-        throw new com.google.protobuf.InvalidProtocolBufferException(
-            e).setUnfinishedMessage(this);
-      } finally {
-        this.unknownFields = unknownFields.build();
-        makeExtensionsImmutable();
-      }
-    }
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.ApiDefProtos.internal_static_tensorflow_ApiDef_Attr_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.ApiDefProtos.internal_static_tensorflow_ApiDef_Attr_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.ApiDef.Attr.class, org.tensorflow.proto.framework.ApiDef.Attr.Builder.class);
-    }
-
-    public static final int NAME_FIELD_NUMBER = 1;
-    private volatile java.lang.Object name_;
-    /**
-     * <code>string name = 1;</code>
-     */
-    public java.lang.String getName() {
-      java.lang.Object ref = name_;
-      if (ref instanceof java.lang.String) {
-        return (java.lang.String) ref;
-      } else {
-        com.google.protobuf.ByteString bs = 
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        name_ = s;
-        return s;
-      }
-    }
-    /**
-     * <code>string name = 1;</code>
-     */
-    public com.google.protobuf.ByteString
-        getNameBytes() {
-      java.lang.Object ref = name_;
-      if (ref instanceof java.lang.String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        name_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-
-    public static final int RENAME_TO_FIELD_NUMBER = 2;
-    private volatile java.lang.Object renameTo_;
-    /**
-     * <pre>
-     * Change the name used to access this attr in the API from what
-     * is used in the GraphDef.  Note that these names in `backticks`
-     * will also be replaced in the summary &amp; description fields.
-     * </pre>
-     *
-     * <code>string rename_to = 2;</code>
-     */
-    public java.lang.String getRenameTo() {
-      java.lang.Object ref = renameTo_;
-      if (ref instanceof java.lang.String) {
-        return (java.lang.String) ref;
-      } else {
-        com.google.protobuf.ByteString bs = 
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        renameTo_ = s;
-        return s;
-      }
-    }
-    /**
-     * <pre>
-     * Change the name used to access this attr in the API from what
-     * is used in the GraphDef.  Note that these names in `backticks`
-     * will also be replaced in the summary &amp; description fields.
-     * </pre>
-     *
-     * <code>string rename_to = 2;</code>
-     */
-    public com.google.protobuf.ByteString
-        getRenameToBytes() {
-      java.lang.Object ref = renameTo_;
-      if (ref instanceof java.lang.String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        renameTo_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-
-    public static final int DEFAULT_VALUE_FIELD_NUMBER = 3;
-    private org.tensorflow.proto.framework.AttrValue defaultValue_;
-    /**
-     * <pre>
-     * Specify a new default value to use for this attr.  This default
-     * will be used when creating new graphs, as opposed to the
-     * default in the OpDef, which will be used when interpreting old
-     * GraphDefs.
-     * </pre>
-     *
-     * <code>.tensorflow.AttrValue default_value = 3;</code>
-     */
-    public boolean hasDefaultValue() {
-      return defaultValue_ != null;
-    }
-    /**
-     * <pre>
-     * Specify a new default value to use for this attr.  This default
-     * will be used when creating new graphs, as opposed to the
-     * default in the OpDef, which will be used when interpreting old
-     * GraphDefs.
-     * </pre>
-     *
-     * <code>.tensorflow.AttrValue default_value = 3;</code>
-     */
-    public org.tensorflow.proto.framework.AttrValue getDefaultValue() {
-      return defaultValue_ == null ? org.tensorflow.proto.framework.AttrValue.getDefaultInstance() : defaultValue_;
-    }
-    /**
-     * <pre>
-     * Specify a new default value to use for this attr.  This default
-     * will be used when creating new graphs, as opposed to the
-     * default in the OpDef, which will be used when interpreting old
-     * GraphDefs.
-     * </pre>
-     *
-     * <code>.tensorflow.AttrValue default_value = 3;</code>
-     */
-    public org.tensorflow.proto.framework.AttrValueOrBuilder getDefaultValueOrBuilder() {
-      return getDefaultValue();
-    }
-
-    public static final int DESCRIPTION_FIELD_NUMBER = 4;
-    private volatile java.lang.Object description_;
-    /**
-     * <pre>
-     * Note: this will replace any inherited attr doc, there is no current
-     * way of modifying attr descriptions as can be done with op descriptions.
-     * </pre>
-     *
-     * <code>string description = 4;</code>
-     */
-    public java.lang.String getDescription() {
-      java.lang.Object ref = description_;
-      if (ref instanceof java.lang.String) {
-        return (java.lang.String) ref;
-      } else {
-        com.google.protobuf.ByteString bs = 
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        description_ = s;
-        return s;
-      }
-    }
-    /**
-     * <pre>
-     * Note: this will replace any inherited attr doc, there is no current
-     * way of modifying attr descriptions as can be done with op descriptions.
-     * </pre>
-     *
-     * <code>string description = 4;</code>
-     */
-    public com.google.protobuf.ByteString
-        getDescriptionBytes() {
-      java.lang.Object ref = description_;
-      if (ref instanceof java.lang.String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        description_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-
-    private byte memoizedIsInitialized = -1;
-    @java.lang.Override
-    public final boolean isInitialized() {
-      byte isInitialized = memoizedIsInitialized;
-      if (isInitialized == 1) return true;
-      if (isInitialized == 0) return false;
-
-      memoizedIsInitialized = 1;
-      return true;
-    }
-
-    @java.lang.Override
-    public void writeTo(com.google.protobuf.CodedOutputStream output)
-                        throws java.io.IOException {
-      if (!getNameBytes().isEmpty()) {
-        com.google.protobuf.GeneratedMessageV3.writeString(output, 1, name_);
-      }
-      if (!getRenameToBytes().isEmpty()) {
-        com.google.protobuf.GeneratedMessageV3.writeString(output, 2, renameTo_);
-      }
-      if (defaultValue_ != null) {
-        output.writeMessage(3, getDefaultValue());
-      }
-      if (!getDescriptionBytes().isEmpty()) {
-        com.google.protobuf.GeneratedMessageV3.writeString(output, 4, description_);
-      }
-      unknownFields.writeTo(output);
-    }
-
-    @java.lang.Override
-    public int getSerializedSize() {
-      int size = memoizedSize;
-      if (size != -1) return size;
-
-      size = 0;
-      if (!getNameBytes().isEmpty()) {
-        size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, name_);
-      }
-      if (!getRenameToBytes().isEmpty()) {
-        size += com.google.protobuf.GeneratedMessageV3.computeStringSize(2, renameTo_);
-      }
-      if (defaultValue_ != null) {
-        size += com.google.protobuf.CodedOutputStream
-          .computeMessageSize(3, getDefaultValue());
-      }
-      if (!getDescriptionBytes().isEmpty()) {
-        size += com.google.protobuf.GeneratedMessageV3.computeStringSize(4, description_);
-      }
-      size += unknownFields.getSerializedSize();
-      memoizedSize = size;
-      return size;
-    }
-
-    @java.lang.Override
-    public boolean equals(final java.lang.Object obj) {
-      if (obj == this) {
-       return true;
-      }
-      if (!(obj instanceof org.tensorflow.proto.framework.ApiDef.Attr)) {
-        return super.equals(obj);
-      }
-      org.tensorflow.proto.framework.ApiDef.Attr other = (org.tensorflow.proto.framework.ApiDef.Attr) obj;
-
-      if (!getName()
-          .equals(other.getName())) return false;
-      if (!getRenameTo()
-          .equals(other.getRenameTo())) return false;
-      if (hasDefaultValue() != other.hasDefaultValue()) return false;
-      if (hasDefaultValue()) {
-        if (!getDefaultValue()
-            .equals(other.getDefaultValue())) return false;
-      }
-      if (!getDescription()
-          .equals(other.getDescription())) return false;
-      if (!unknownFields.equals(other.unknownFields)) return false;
-      return true;
-    }
-
-    @java.lang.Override
-    public int hashCode() {
-      if (memoizedHashCode != 0) {
-        return memoizedHashCode;
-      }
-      int hash = 41;
-      hash = (19 * hash) + getDescriptor().hashCode();
-      hash = (37 * hash) + NAME_FIELD_NUMBER;
-      hash = (53 * hash) + getName().hashCode();
-      hash = (37 * hash) + RENAME_TO_FIELD_NUMBER;
-      hash = (53 * hash) + getRenameTo().hashCode();
-      if (hasDefaultValue()) {
-        hash = (37 * hash) + DEFAULT_VALUE_FIELD_NUMBER;
-        hash = (53 * hash) + getDefaultValue().hashCode();
-      }
-      hash = (37 * hash) + DESCRIPTION_FIELD_NUMBER;
-      hash = (53 * hash) + getDescription().hashCode();
-      hash = (29 * hash) + unknownFields.hashCode();
-      memoizedHashCode = hash;
-      return hash;
-    }
-
-    public static org.tensorflow.proto.framework.ApiDef.Attr parseFrom(
-        java.nio.ByteBuffer data)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data);
-    }
-    public static org.tensorflow.proto.framework.ApiDef.Attr parseFrom(
-        java.nio.ByteBuffer data,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.ApiDef.Attr parseFrom(
-        com.google.protobuf.ByteString data)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data);
-    }
-    public static org.tensorflow.proto.framework.ApiDef.Attr parseFrom(
-        com.google.protobuf.ByteString data,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.ApiDef.Attr parseFrom(byte[] data)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data);
-    }
-    public static org.tensorflow.proto.framework.ApiDef.Attr parseFrom(
-        byte[] data,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.ApiDef.Attr parseFrom(java.io.InputStream input)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input);
-    }
-    public static org.tensorflow.proto.framework.ApiDef.Attr parseFrom(
-        java.io.InputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.ApiDef.Attr parseDelimitedFrom(java.io.InputStream input)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseDelimitedWithIOException(PARSER, input);
-    }
-    public static org.tensorflow.proto.framework.ApiDef.Attr parseDelimitedFrom(
-        java.io.InputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.ApiDef.Attr parseFrom(
-        com.google.protobuf.CodedInputStream input)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input);
-    }
-    public static org.tensorflow.proto.framework.ApiDef.Attr parseFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input, extensionRegistry);
-    }
-
-    @java.lang.Override
-    public Builder newBuilderForType() { return newBuilder(); }
-    public static Builder newBuilder() {
-      return DEFAULT_INSTANCE.toBuilder();
-    }
-    public static Builder newBuilder(org.tensorflow.proto.framework.ApiDef.Attr prototype) {
-      return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-    }
-    @java.lang.Override
-    public Builder toBuilder() {
-      return this == DEFAULT_INSTANCE
-          ? new Builder() : new Builder().mergeFrom(this);
-    }
-
-    @java.lang.Override
-    protected Builder newBuilderForType(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      Builder builder = new Builder(parent);
-      return builder;
-    }
-    /**
-     * <pre>
-     * Description of the graph-construction-time configuration of this
-     * Op.  That is to say, this describes the attr fields that will
-     * be specified in the NodeDef.
-     * </pre>
-     *
-     * Protobuf type {@code tensorflow.ApiDef.Attr}
-     */
-    public static final class Builder extends
-        com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-        // @@protoc_insertion_point(builder_implements:tensorflow.ApiDef.Attr)
-        org.tensorflow.proto.framework.ApiDef.AttrOrBuilder {
-      public static final com.google.protobuf.Descriptors.Descriptor
-          getDescriptor() {
-        return org.tensorflow.proto.framework.ApiDefProtos.internal_static_tensorflow_ApiDef_Attr_descriptor;
-      }
-
-      @java.lang.Override
-      protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-          internalGetFieldAccessorTable() {
-        return org.tensorflow.proto.framework.ApiDefProtos.internal_static_tensorflow_ApiDef_Attr_fieldAccessorTable
-            .ensureFieldAccessorsInitialized(
-                org.tensorflow.proto.framework.ApiDef.Attr.class, org.tensorflow.proto.framework.ApiDef.Attr.Builder.class);
-      }
-
-      // Construct using org.tensorflow.proto.framework.ApiDef.Attr.newBuilder()
-      private Builder() {
-        maybeForceBuilderInitialization();
-      }
-
-      private Builder(
-          com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-        super(parent);
-        maybeForceBuilderInitialization();
-      }
-      private void maybeForceBuilderInitialization() {
-        if (com.google.protobuf.GeneratedMessageV3
-                .alwaysUseFieldBuilders) {
-        }
-      }
-      @java.lang.Override
-      public Builder clear() {
-        super.clear();
-        name_ = "";
-
-        renameTo_ = "";
-
-        if (defaultValueBuilder_ == null) {
-          defaultValue_ = null;
-        } else {
-          defaultValue_ = null;
-          defaultValueBuilder_ = null;
-        }
-        description_ = "";
-
-        return this;
-      }
-
-      @java.lang.Override
-      public com.google.protobuf.Descriptors.Descriptor
-          getDescriptorForType() {
-        return org.tensorflow.proto.framework.ApiDefProtos.internal_static_tensorflow_ApiDef_Attr_descriptor;
-      }
-
-      @java.lang.Override
-      public org.tensorflow.proto.framework.ApiDef.Attr getDefaultInstanceForType() {
-        return org.tensorflow.proto.framework.ApiDef.Attr.getDefaultInstance();
-      }
-
-      @java.lang.Override
-      public org.tensorflow.proto.framework.ApiDef.Attr build() {
-        org.tensorflow.proto.framework.ApiDef.Attr result = buildPartial();
-        if (!result.isInitialized()) {
-          throw newUninitializedMessageException(result);
-        }
-        return result;
-      }
-
-      @java.lang.Override
-      public org.tensorflow.proto.framework.ApiDef.Attr buildPartial() {
-        org.tensorflow.proto.framework.ApiDef.Attr result = new org.tensorflow.proto.framework.ApiDef.Attr(this);
-        result.name_ = name_;
-        result.renameTo_ = renameTo_;
-        if (defaultValueBuilder_ == null) {
-          result.defaultValue_ = defaultValue_;
-        } else {
-          result.defaultValue_ = defaultValueBuilder_.build();
-        }
-        result.description_ = description_;
-        onBuilt();
-        return result;
-      }
-
-      @java.lang.Override
-      public Builder clone() {
-        return super.clone();
-      }
-      @java.lang.Override
-      public Builder setField(
-          com.google.protobuf.Descriptors.FieldDescriptor field,
-          java.lang.Object value) {
-        return super.setField(field, value);
-      }
-      @java.lang.Override
-      public Builder clearField(
-          com.google.protobuf.Descriptors.FieldDescriptor field) {
-        return super.clearField(field);
-      }
-      @java.lang.Override
-      public Builder clearOneof(
-          com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-        return super.clearOneof(oneof);
-      }
-      @java.lang.Override
-      public Builder setRepeatedField(
-          com.google.protobuf.Descriptors.FieldDescriptor field,
-          int index, java.lang.Object value) {
-        return super.setRepeatedField(field, index, value);
-      }
-      @java.lang.Override
-      public Builder addRepeatedField(
-          com.google.protobuf.Descriptors.FieldDescriptor field,
-          java.lang.Object value) {
-        return super.addRepeatedField(field, value);
-      }
-      @java.lang.Override
-      public Builder mergeFrom(com.google.protobuf.Message other) {
-        if (other instanceof org.tensorflow.proto.framework.ApiDef.Attr) {
-          return mergeFrom((org.tensorflow.proto.framework.ApiDef.Attr)other);
-        } else {
-          super.mergeFrom(other);
-          return this;
-        }
-      }
-
-      public Builder mergeFrom(org.tensorflow.proto.framework.ApiDef.Attr other) {
-        if (other == org.tensorflow.proto.framework.ApiDef.Attr.getDefaultInstance()) return this;
-        if (!other.getName().isEmpty()) {
-          name_ = other.name_;
-          onChanged();
-        }
-        if (!other.getRenameTo().isEmpty()) {
-          renameTo_ = other.renameTo_;
-          onChanged();
-        }
-        if (other.hasDefaultValue()) {
-          mergeDefaultValue(other.getDefaultValue());
-        }
-        if (!other.getDescription().isEmpty()) {
-          description_ = other.description_;
-          onChanged();
-        }
-        this.mergeUnknownFields(other.unknownFields);
-        onChanged();
-        return this;
-      }
-
-      @java.lang.Override
-      public final boolean isInitialized() {
-        return true;
-      }
-
-      @java.lang.Override
-      public Builder mergeFrom(
-          com.google.protobuf.CodedInputStream input,
-          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-          throws java.io.IOException {
-        org.tensorflow.proto.framework.ApiDef.Attr parsedMessage = null;
-        try {
-          parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-        } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-          parsedMessage = (org.tensorflow.proto.framework.ApiDef.Attr) e.getUnfinishedMessage();
-          throw e.unwrapIOException();
-        } finally {
-          if (parsedMessage != null) {
-            mergeFrom(parsedMessage);
-          }
-        }
-        return this;
-      }
-
-      private java.lang.Object name_ = "";
-      /**
-       * <code>string name = 1;</code>
-       */
-      public java.lang.String getName() {
-        java.lang.Object ref = name_;
-        if (!(ref instanceof java.lang.String)) {
-          com.google.protobuf.ByteString bs =
-              (com.google.protobuf.ByteString) ref;
-          java.lang.String s = bs.toStringUtf8();
-          name_ = s;
-          return s;
-        } else {
-          return (java.lang.String) ref;
-        }
-      }
-      /**
-       * <code>string name = 1;</code>
-       */
-      public com.google.protobuf.ByteString
-          getNameBytes() {
-        java.lang.Object ref = name_;
-        if (ref instanceof String) {
-          com.google.protobuf.ByteString b = 
-              com.google.protobuf.ByteString.copyFromUtf8(
-                  (java.lang.String) ref);
-          name_ = b;
-          return b;
-        } else {
-          return (com.google.protobuf.ByteString) ref;
-        }
-      }
-      /**
-       * <code>string name = 1;</code>
-       */
-      public Builder setName(
-          java.lang.String value) {
-        if (value == null) {
-    throw new NullPointerException();
-  }
-  
-        name_ = value;
-        onChanged();
-        return this;
-      }
-      /**
-       * <code>string name = 1;</code>
-       */
-      public Builder clearName() {
-        
-        name_ = getDefaultInstance().getName();
-        onChanged();
-        return this;
-      }
-      /**
-       * <code>string name = 1;</code>
-       */
-      public Builder setNameBytes(
-          com.google.protobuf.ByteString value) {
-        if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-        
-        name_ = value;
-        onChanged();
-        return this;
-      }
-
-      private java.lang.Object renameTo_ = "";
-      /**
-       * <pre>
-       * Change the name used to access this attr in the API from what
-       * is used in the GraphDef.  Note that these names in `backticks`
-       * will also be replaced in the summary &amp; description fields.
-       * </pre>
-       *
-       * <code>string rename_to = 2;</code>
-       */
-      public java.lang.String getRenameTo() {
-        java.lang.Object ref = renameTo_;
-        if (!(ref instanceof java.lang.String)) {
-          com.google.protobuf.ByteString bs =
-              (com.google.protobuf.ByteString) ref;
-          java.lang.String s = bs.toStringUtf8();
-          renameTo_ = s;
-          return s;
-        } else {
-          return (java.lang.String) ref;
-        }
-      }
-      /**
-       * <pre>
-       * Change the name used to access this attr in the API from what
-       * is used in the GraphDef.  Note that these names in `backticks`
-       * will also be replaced in the summary &amp; description fields.
-       * </pre>
-       *
-       * <code>string rename_to = 2;</code>
-       */
-      public com.google.protobuf.ByteString
-          getRenameToBytes() {
-        java.lang.Object ref = renameTo_;
-        if (ref instanceof String) {
-          com.google.protobuf.ByteString b = 
-              com.google.protobuf.ByteString.copyFromUtf8(
-                  (java.lang.String) ref);
-          renameTo_ = b;
-          return b;
-        } else {
-          return (com.google.protobuf.ByteString) ref;
-        }
-      }
-      /**
-       * <pre>
-       * Change the name used to access this attr in the API from what
-       * is used in the GraphDef.  Note that these names in `backticks`
-       * will also be replaced in the summary &amp; description fields.
-       * </pre>
-       *
-       * <code>string rename_to = 2;</code>
-       */
-      public Builder setRenameTo(
-          java.lang.String value) {
-        if (value == null) {
-    throw new NullPointerException();
-  }
-  
-        renameTo_ = value;
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * Change the name used to access this attr in the API from what
-       * is used in the GraphDef.  Note that these names in `backticks`
-       * will also be replaced in the summary &amp; description fields.
-       * </pre>
-       *
-       * <code>string rename_to = 2;</code>
-       */
-      public Builder clearRenameTo() {
-        
-        renameTo_ = getDefaultInstance().getRenameTo();
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * Change the name used to access this attr in the API from what
-       * is used in the GraphDef.  Note that these names in `backticks`
-       * will also be replaced in the summary &amp; description fields.
-       * </pre>
-       *
-       * <code>string rename_to = 2;</code>
-       */
-      public Builder setRenameToBytes(
-          com.google.protobuf.ByteString value) {
-        if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-        
-        renameTo_ = value;
-        onChanged();
-        return this;
-      }
-
-      private org.tensorflow.proto.framework.AttrValue defaultValue_;
-      private com.google.protobuf.SingleFieldBuilderV3<
-          org.tensorflow.proto.framework.AttrValue, org.tensorflow.proto.framework.AttrValue.Builder, org.tensorflow.proto.framework.AttrValueOrBuilder> defaultValueBuilder_;
-      /**
-       * <pre>
-       * Specify a new default value to use for this attr.  This default
-       * will be used when creating new graphs, as opposed to the
-       * default in the OpDef, which will be used when interpreting old
-       * GraphDefs.
-       * </pre>
-       *
-       * <code>.tensorflow.AttrValue default_value = 3;</code>
-       */
-      public boolean hasDefaultValue() {
-        return defaultValueBuilder_ != null || defaultValue_ != null;
-      }
-      /**
-       * <pre>
-       * Specify a new default value to use for this attr.  This default
-       * will be used when creating new graphs, as opposed to the
-       * default in the OpDef, which will be used when interpreting old
-       * GraphDefs.
-       * </pre>
-       *
-       * <code>.tensorflow.AttrValue default_value = 3;</code>
-       */
-      public org.tensorflow.proto.framework.AttrValue getDefaultValue() {
-        if (defaultValueBuilder_ == null) {
-          return defaultValue_ == null ? org.tensorflow.proto.framework.AttrValue.getDefaultInstance() : defaultValue_;
-        } else {
-          return defaultValueBuilder_.getMessage();
-        }
-      }
-      /**
-       * <pre>
-       * Specify a new default value to use for this attr.  This default
-       * will be used when creating new graphs, as opposed to the
-       * default in the OpDef, which will be used when interpreting old
-       * GraphDefs.
-       * </pre>
-       *
-       * <code>.tensorflow.AttrValue default_value = 3;</code>
-       */
-      public Builder setDefaultValue(org.tensorflow.proto.framework.AttrValue value) {
-        if (defaultValueBuilder_ == null) {
-          if (value == null) {
-            throw new NullPointerException();
-          }
-          defaultValue_ = value;
-          onChanged();
-        } else {
-          defaultValueBuilder_.setMessage(value);
-        }
-
-        return this;
-      }
-      /**
-       * <pre>
-       * Specify a new default value to use for this attr.  This default
-       * will be used when creating new graphs, as opposed to the
-       * default in the OpDef, which will be used when interpreting old
-       * GraphDefs.
-       * </pre>
-       *
-       * <code>.tensorflow.AttrValue default_value = 3;</code>
-       */
-      public Builder setDefaultValue(
-          org.tensorflow.proto.framework.AttrValue.Builder builderForValue) {
-        if (defaultValueBuilder_ == null) {
-          defaultValue_ = builderForValue.build();
-          onChanged();
-        } else {
-          defaultValueBuilder_.setMessage(builderForValue.build());
-        }
-
-        return this;
-      }
-      /**
-       * <pre>
-       * Specify a new default value to use for this attr.  This default
-       * will be used when creating new graphs, as opposed to the
-       * default in the OpDef, which will be used when interpreting old
-       * GraphDefs.
-       * </pre>
-       *
-       * <code>.tensorflow.AttrValue default_value = 3;</code>
-       */
-      public Builder mergeDefaultValue(org.tensorflow.proto.framework.AttrValue value) {
-        if (defaultValueBuilder_ == null) {
-          if (defaultValue_ != null) {
-            defaultValue_ =
-              org.tensorflow.proto.framework.AttrValue.newBuilder(defaultValue_).mergeFrom(value).buildPartial();
-          } else {
-            defaultValue_ = value;
-          }
-          onChanged();
-        } else {
-          defaultValueBuilder_.mergeFrom(value);
-        }
-
-        return this;
-      }
-      /**
-       * <pre>
-       * Specify a new default value to use for this attr.  This default
-       * will be used when creating new graphs, as opposed to the
-       * default in the OpDef, which will be used when interpreting old
-       * GraphDefs.
-       * </pre>
-       *
-       * <code>.tensorflow.AttrValue default_value = 3;</code>
-       */
-      public Builder clearDefaultValue() {
-        if (defaultValueBuilder_ == null) {
-          defaultValue_ = null;
-          onChanged();
-        } else {
-          defaultValue_ = null;
-          defaultValueBuilder_ = null;
-        }
-
-        return this;
-      }
-      /**
-       * <pre>
-       * Specify a new default value to use for this attr.  This default
-       * will be used when creating new graphs, as opposed to the
-       * default in the OpDef, which will be used when interpreting old
-       * GraphDefs.
-       * </pre>
-       *
-       * <code>.tensorflow.AttrValue default_value = 3;</code>
-       */
-      public org.tensorflow.proto.framework.AttrValue.Builder getDefaultValueBuilder() {
-        
-        onChanged();
-        return getDefaultValueFieldBuilder().getBuilder();
-      }
-      /**
-       * <pre>
-       * Specify a new default value to use for this attr.  This default
-       * will be used when creating new graphs, as opposed to the
-       * default in the OpDef, which will be used when interpreting old
-       * GraphDefs.
-       * </pre>
-       *
-       * <code>.tensorflow.AttrValue default_value = 3;</code>
-       */
-      public org.tensorflow.proto.framework.AttrValueOrBuilder getDefaultValueOrBuilder() {
-        if (defaultValueBuilder_ != null) {
-          return defaultValueBuilder_.getMessageOrBuilder();
-        } else {
-          return defaultValue_ == null ?
-              org.tensorflow.proto.framework.AttrValue.getDefaultInstance() : defaultValue_;
-        }
-      }
-      /**
-       * <pre>
-       * Specify a new default value to use for this attr.  This default
-       * will be used when creating new graphs, as opposed to the
-       * default in the OpDef, which will be used when interpreting old
-       * GraphDefs.
-       * </pre>
-       *
-       * <code>.tensorflow.AttrValue default_value = 3;</code>
-       */
-      private com.google.protobuf.SingleFieldBuilderV3<
-          org.tensorflow.proto.framework.AttrValue, org.tensorflow.proto.framework.AttrValue.Builder, org.tensorflow.proto.framework.AttrValueOrBuilder> 
-          getDefaultValueFieldBuilder() {
-        if (defaultValueBuilder_ == null) {
-          defaultValueBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-              org.tensorflow.proto.framework.AttrValue, org.tensorflow.proto.framework.AttrValue.Builder, org.tensorflow.proto.framework.AttrValueOrBuilder>(
-                  getDefaultValue(),
-                  getParentForChildren(),
-                  isClean());
-          defaultValue_ = null;
-        }
-        return defaultValueBuilder_;
-      }
-
-      private java.lang.Object description_ = "";
-      /**
-       * <pre>
-       * Note: this will replace any inherited attr doc, there is no current
-       * way of modifying attr descriptions as can be done with op descriptions.
-       * </pre>
-       *
-       * <code>string description = 4;</code>
-       */
-      public java.lang.String getDescription() {
-        java.lang.Object ref = description_;
-        if (!(ref instanceof java.lang.String)) {
-          com.google.protobuf.ByteString bs =
-              (com.google.protobuf.ByteString) ref;
-          java.lang.String s = bs.toStringUtf8();
-          description_ = s;
-          return s;
-        } else {
-          return (java.lang.String) ref;
-        }
-      }
-      /**
-       * <pre>
-       * Note: this will replace any inherited attr doc, there is no current
-       * way of modifying attr descriptions as can be done with op descriptions.
-       * </pre>
-       *
-       * <code>string description = 4;</code>
-       */
-      public com.google.protobuf.ByteString
-          getDescriptionBytes() {
-        java.lang.Object ref = description_;
-        if (ref instanceof String) {
-          com.google.protobuf.ByteString b = 
-              com.google.protobuf.ByteString.copyFromUtf8(
-                  (java.lang.String) ref);
-          description_ = b;
-          return b;
-        } else {
-          return (com.google.protobuf.ByteString) ref;
-        }
-      }
-      /**
-       * <pre>
-       * Note: this will replace any inherited attr doc, there is no current
-       * way of modifying attr descriptions as can be done with op descriptions.
-       * </pre>
-       *
-       * <code>string description = 4;</code>
-       */
-      public Builder setDescription(
-          java.lang.String value) {
-        if (value == null) {
-    throw new NullPointerException();
-  }
-  
-        description_ = value;
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * Note: this will replace any inherited attr doc, there is no current
-       * way of modifying attr descriptions as can be done with op descriptions.
-       * </pre>
-       *
-       * <code>string description = 4;</code>
-       */
-      public Builder clearDescription() {
-        
-        description_ = getDefaultInstance().getDescription();
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * Note: this will replace any inherited attr doc, there is no current
-       * way of modifying attr descriptions as can be done with op descriptions.
-       * </pre>
-       *
-       * <code>string description = 4;</code>
-       */
-      public Builder setDescriptionBytes(
-          com.google.protobuf.ByteString value) {
-        if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-        
-        description_ = value;
-        onChanged();
-        return this;
-      }
-      @java.lang.Override
-      public final Builder setUnknownFields(
-          final com.google.protobuf.UnknownFieldSet unknownFields) {
-        return super.setUnknownFields(unknownFields);
-      }
-
-      @java.lang.Override
-      public final Builder mergeUnknownFields(
-          final com.google.protobuf.UnknownFieldSet unknownFields) {
-        return super.mergeUnknownFields(unknownFields);
-      }
-
-
-      // @@protoc_insertion_point(builder_scope:tensorflow.ApiDef.Attr)
-    }
-
-    // @@protoc_insertion_point(class_scope:tensorflow.ApiDef.Attr)
-    private static final org.tensorflow.proto.framework.ApiDef.Attr DEFAULT_INSTANCE;
-    static {
-      DEFAULT_INSTANCE = new org.tensorflow.proto.framework.ApiDef.Attr();
-    }
-
-    public static org.tensorflow.proto.framework.ApiDef.Attr getDefaultInstance() {
-      return DEFAULT_INSTANCE;
-    }
-
-    private static final com.google.protobuf.Parser<Attr>
-        PARSER = new com.google.protobuf.AbstractParser<Attr>() {
-      @java.lang.Override
-      public Attr parsePartialFrom(
-          com.google.protobuf.CodedInputStream input,
-          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-          throws com.google.protobuf.InvalidProtocolBufferException {
-        return new Attr(input, extensionRegistry);
-      }
-    };
-
-    public static com.google.protobuf.Parser<Attr> parser() {
-      return PARSER;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Parser<Attr> getParserForType() {
-      return PARSER;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.ApiDef.Attr getDefaultInstanceForType() {
-      return DEFAULT_INSTANCE;
-    }
-
-  }
-
-  public static final int GRAPH_OP_NAME_FIELD_NUMBER = 1;
-  private volatile java.lang.Object graphOpName_;
-  /**
-   * <pre>
-   * Name of the op (in the OpDef) to specify the API for.
-   * </pre>
-   *
-   * <code>string graph_op_name = 1;</code>
-   */
-  public java.lang.String getGraphOpName() {
-    java.lang.Object ref = graphOpName_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      graphOpName_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * Name of the op (in the OpDef) to specify the API for.
-   * </pre>
-   *
-   * <code>string graph_op_name = 1;</code>
-   */
-  public com.google.protobuf.ByteString
-      getGraphOpNameBytes() {
-    java.lang.Object ref = graphOpName_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      graphOpName_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int DEPRECATION_MESSAGE_FIELD_NUMBER = 12;
-  private volatile java.lang.Object deprecationMessage_;
-  /**
-   * <pre>
-   * If this op is deprecated, set deprecation message to the message
-   * that should be logged when this op is used.
-   * The message should indicate alternative op to use, if any.
-   * </pre>
-   *
-   * <code>string deprecation_message = 12;</code>
-   */
-  public java.lang.String getDeprecationMessage() {
-    java.lang.Object ref = deprecationMessage_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      deprecationMessage_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * If this op is deprecated, set deprecation message to the message
-   * that should be logged when this op is used.
-   * The message should indicate alternative op to use, if any.
-   * </pre>
-   *
-   * <code>string deprecation_message = 12;</code>
-   */
-  public com.google.protobuf.ByteString
-      getDeprecationMessageBytes() {
-    java.lang.Object ref = deprecationMessage_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      deprecationMessage_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int DEPRECATION_VERSION_FIELD_NUMBER = 13;
-  private int deprecationVersion_;
-  /**
-   * <pre>
-   * Major version when the op will be deleted. For e.g. set this
-   * value to 2 if op API should be removed in TensorFlow 2.0 and
-   * deprecated in versions before that.
-   * </pre>
-   *
-   * <code>int32 deprecation_version = 13;</code>
-   */
-  public int getDeprecationVersion() {
-    return deprecationVersion_;
-  }
-
-  public static final int VISIBILITY_FIELD_NUMBER = 2;
-  private int visibility_;
-  /**
-   * <code>.tensorflow.ApiDef.Visibility visibility = 2;</code>
-   */
-  public int getVisibilityValue() {
-    return visibility_;
-  }
-  /**
-   * <code>.tensorflow.ApiDef.Visibility visibility = 2;</code>
-   */
-  public org.tensorflow.proto.framework.ApiDef.Visibility getVisibility() {
-    @SuppressWarnings("deprecation")
-    org.tensorflow.proto.framework.ApiDef.Visibility result = org.tensorflow.proto.framework.ApiDef.Visibility.valueOf(visibility_);
-    return result == null ? org.tensorflow.proto.framework.ApiDef.Visibility.UNRECOGNIZED : result;
-  }
-
-  public static final int ENDPOINT_FIELD_NUMBER = 3;
-  private java.util.List<org.tensorflow.proto.framework.ApiDef.Endpoint> endpoint_;
-  /**
-   * <code>repeated .tensorflow.ApiDef.Endpoint endpoint = 3;</code>
-   */
-  public java.util.List<org.tensorflow.proto.framework.ApiDef.Endpoint> getEndpointList() {
-    return endpoint_;
-  }
-  /**
-   * <code>repeated .tensorflow.ApiDef.Endpoint endpoint = 3;</code>
-   */
-  public java.util.List<? extends org.tensorflow.proto.framework.ApiDef.EndpointOrBuilder> 
-      getEndpointOrBuilderList() {
-    return endpoint_;
-  }
-  /**
-   * <code>repeated .tensorflow.ApiDef.Endpoint endpoint = 3;</code>
-   */
-  public int getEndpointCount() {
-    return endpoint_.size();
-  }
-  /**
-   * <code>repeated .tensorflow.ApiDef.Endpoint endpoint = 3;</code>
-   */
-  public org.tensorflow.proto.framework.ApiDef.Endpoint getEndpoint(int index) {
-    return endpoint_.get(index);
-  }
-  /**
-   * <code>repeated .tensorflow.ApiDef.Endpoint endpoint = 3;</code>
-   */
-  public org.tensorflow.proto.framework.ApiDef.EndpointOrBuilder getEndpointOrBuilder(
-      int index) {
-    return endpoint_.get(index);
-  }
-
-  public static final int IN_ARG_FIELD_NUMBER = 4;
-  private java.util.List<org.tensorflow.proto.framework.ApiDef.Arg> inArg_;
-  /**
-   * <code>repeated .tensorflow.ApiDef.Arg in_arg = 4;</code>
-   */
-  public java.util.List<org.tensorflow.proto.framework.ApiDef.Arg> getInArgList() {
-    return inArg_;
-  }
-  /**
-   * <code>repeated .tensorflow.ApiDef.Arg in_arg = 4;</code>
-   */
-  public java.util.List<? extends org.tensorflow.proto.framework.ApiDef.ArgOrBuilder> 
-      getInArgOrBuilderList() {
-    return inArg_;
-  }
-  /**
-   * <code>repeated .tensorflow.ApiDef.Arg in_arg = 4;</code>
-   */
-  public int getInArgCount() {
-    return inArg_.size();
-  }
-  /**
-   * <code>repeated .tensorflow.ApiDef.Arg in_arg = 4;</code>
-   */
-  public org.tensorflow.proto.framework.ApiDef.Arg getInArg(int index) {
-    return inArg_.get(index);
-  }
-  /**
-   * <code>repeated .tensorflow.ApiDef.Arg in_arg = 4;</code>
-   */
-  public org.tensorflow.proto.framework.ApiDef.ArgOrBuilder getInArgOrBuilder(
-      int index) {
-    return inArg_.get(index);
-  }
-
-  public static final int OUT_ARG_FIELD_NUMBER = 5;
-  private java.util.List<org.tensorflow.proto.framework.ApiDef.Arg> outArg_;
-  /**
-   * <code>repeated .tensorflow.ApiDef.Arg out_arg = 5;</code>
-   */
-  public java.util.List<org.tensorflow.proto.framework.ApiDef.Arg> getOutArgList() {
-    return outArg_;
-  }
-  /**
-   * <code>repeated .tensorflow.ApiDef.Arg out_arg = 5;</code>
-   */
-  public java.util.List<? extends org.tensorflow.proto.framework.ApiDef.ArgOrBuilder> 
-      getOutArgOrBuilderList() {
-    return outArg_;
-  }
-  /**
-   * <code>repeated .tensorflow.ApiDef.Arg out_arg = 5;</code>
-   */
-  public int getOutArgCount() {
-    return outArg_.size();
-  }
-  /**
-   * <code>repeated .tensorflow.ApiDef.Arg out_arg = 5;</code>
-   */
-  public org.tensorflow.proto.framework.ApiDef.Arg getOutArg(int index) {
-    return outArg_.get(index);
-  }
-  /**
-   * <code>repeated .tensorflow.ApiDef.Arg out_arg = 5;</code>
-   */
-  public org.tensorflow.proto.framework.ApiDef.ArgOrBuilder getOutArgOrBuilder(
-      int index) {
-    return outArg_.get(index);
-  }
-
-  public static final int ARG_ORDER_FIELD_NUMBER = 11;
-  private com.google.protobuf.LazyStringList argOrder_;
-  /**
-   * <pre>
-   * List of original in_arg names to specify new argument order.
-   * Length of arg_order should be either empty to keep current order
-   * or match size of in_arg.
-   * </pre>
-   *
-   * <code>repeated string arg_order = 11;</code>
-   */
-  public com.google.protobuf.ProtocolStringList
-      getArgOrderList() {
-    return argOrder_;
-  }
-  /**
-   * <pre>
-   * List of original in_arg names to specify new argument order.
-   * Length of arg_order should be either empty to keep current order
-   * or match size of in_arg.
-   * </pre>
-   *
-   * <code>repeated string arg_order = 11;</code>
-   */
-  public int getArgOrderCount() {
-    return argOrder_.size();
-  }
-  /**
-   * <pre>
-   * List of original in_arg names to specify new argument order.
-   * Length of arg_order should be either empty to keep current order
-   * or match size of in_arg.
-   * </pre>
-   *
-   * <code>repeated string arg_order = 11;</code>
-   */
-  public java.lang.String getArgOrder(int index) {
-    return argOrder_.get(index);
-  }
-  /**
-   * <pre>
-   * List of original in_arg names to specify new argument order.
-   * Length of arg_order should be either empty to keep current order
-   * or match size of in_arg.
-   * </pre>
-   *
-   * <code>repeated string arg_order = 11;</code>
-   */
-  public com.google.protobuf.ByteString
-      getArgOrderBytes(int index) {
-    return argOrder_.getByteString(index);
-  }
-
-  public static final int ATTR_FIELD_NUMBER = 6;
-  private java.util.List<org.tensorflow.proto.framework.ApiDef.Attr> attr_;
-  /**
-   * <code>repeated .tensorflow.ApiDef.Attr attr = 6;</code>
-   */
-  public java.util.List<org.tensorflow.proto.framework.ApiDef.Attr> getAttrList() {
-    return attr_;
-  }
-  /**
-   * <code>repeated .tensorflow.ApiDef.Attr attr = 6;</code>
-   */
-  public java.util.List<? extends org.tensorflow.proto.framework.ApiDef.AttrOrBuilder> 
-      getAttrOrBuilderList() {
-    return attr_;
-  }
-  /**
-   * <code>repeated .tensorflow.ApiDef.Attr attr = 6;</code>
-   */
-  public int getAttrCount() {
-    return attr_.size();
-  }
-  /**
-   * <code>repeated .tensorflow.ApiDef.Attr attr = 6;</code>
-   */
-  public org.tensorflow.proto.framework.ApiDef.Attr getAttr(int index) {
-    return attr_.get(index);
-  }
-  /**
-   * <code>repeated .tensorflow.ApiDef.Attr attr = 6;</code>
-   */
-  public org.tensorflow.proto.framework.ApiDef.AttrOrBuilder getAttrOrBuilder(
-      int index) {
-    return attr_.get(index);
-  }
-
-  public static final int SUMMARY_FIELD_NUMBER = 7;
-  private volatile java.lang.Object summary_;
-  /**
-   * <pre>
-   * One-line human-readable description of what the Op does.
-   * </pre>
-   *
-   * <code>string summary = 7;</code>
-   */
-  public java.lang.String getSummary() {
-    java.lang.Object ref = summary_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      summary_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * One-line human-readable description of what the Op does.
-   * </pre>
-   *
-   * <code>string summary = 7;</code>
-   */
-  public com.google.protobuf.ByteString
-      getSummaryBytes() {
-    java.lang.Object ref = summary_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      summary_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int DESCRIPTION_FIELD_NUMBER = 8;
-  private volatile java.lang.Object description_;
-  /**
-   * <pre>
-   * Additional, longer human-readable description of what the Op does.
-   * </pre>
-   *
-   * <code>string description = 8;</code>
-   */
-  public java.lang.String getDescription() {
-    java.lang.Object ref = description_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      description_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * Additional, longer human-readable description of what the Op does.
-   * </pre>
-   *
-   * <code>string description = 8;</code>
-   */
-  public com.google.protobuf.ByteString
-      getDescriptionBytes() {
-    java.lang.Object ref = description_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      description_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int DESCRIPTION_PREFIX_FIELD_NUMBER = 9;
-  private volatile java.lang.Object descriptionPrefix_;
-  /**
-   * <pre>
-   * Modify an existing/inherited description by adding text to the beginning
-   * or end.
-   * </pre>
-   *
-   * <code>string description_prefix = 9;</code>
-   */
-  public java.lang.String getDescriptionPrefix() {
-    java.lang.Object ref = descriptionPrefix_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      descriptionPrefix_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * Modify an existing/inherited description by adding text to the beginning
-   * or end.
-   * </pre>
-   *
-   * <code>string description_prefix = 9;</code>
-   */
-  public com.google.protobuf.ByteString
-      getDescriptionPrefixBytes() {
-    java.lang.Object ref = descriptionPrefix_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      descriptionPrefix_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int DESCRIPTION_SUFFIX_FIELD_NUMBER = 10;
-  private volatile java.lang.Object descriptionSuffix_;
-  /**
-   * <code>string description_suffix = 10;</code>
-   */
-  public java.lang.String getDescriptionSuffix() {
-    java.lang.Object ref = descriptionSuffix_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      descriptionSuffix_ = s;
-      return s;
-    }
-  }
-  /**
-   * <code>string description_suffix = 10;</code>
-   */
-  public com.google.protobuf.ByteString
-      getDescriptionSuffixBytes() {
-    java.lang.Object ref = descriptionSuffix_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      descriptionSuffix_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    if (!getGraphOpNameBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 1, graphOpName_);
-    }
-    if (visibility_ != org.tensorflow.proto.framework.ApiDef.Visibility.DEFAULT_VISIBILITY.getNumber()) {
-      output.writeEnum(2, visibility_);
-    }
-    for (int i = 0; i < endpoint_.size(); i++) {
-      output.writeMessage(3, endpoint_.get(i));
-    }
-    for (int i = 0; i < inArg_.size(); i++) {
-      output.writeMessage(4, inArg_.get(i));
-    }
-    for (int i = 0; i < outArg_.size(); i++) {
-      output.writeMessage(5, outArg_.get(i));
-    }
-    for (int i = 0; i < attr_.size(); i++) {
-      output.writeMessage(6, attr_.get(i));
-    }
-    if (!getSummaryBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 7, summary_);
-    }
-    if (!getDescriptionBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 8, description_);
-    }
-    if (!getDescriptionPrefixBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 9, descriptionPrefix_);
-    }
-    if (!getDescriptionSuffixBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 10, descriptionSuffix_);
-    }
-    for (int i = 0; i < argOrder_.size(); i++) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 11, argOrder_.getRaw(i));
-    }
-    if (!getDeprecationMessageBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 12, deprecationMessage_);
-    }
-    if (deprecationVersion_ != 0) {
-      output.writeInt32(13, deprecationVersion_);
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (!getGraphOpNameBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, graphOpName_);
-    }
-    if (visibility_ != org.tensorflow.proto.framework.ApiDef.Visibility.DEFAULT_VISIBILITY.getNumber()) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeEnumSize(2, visibility_);
-    }
-    for (int i = 0; i < endpoint_.size(); i++) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(3, endpoint_.get(i));
-    }
-    for (int i = 0; i < inArg_.size(); i++) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(4, inArg_.get(i));
-    }
-    for (int i = 0; i < outArg_.size(); i++) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(5, outArg_.get(i));
-    }
-    for (int i = 0; i < attr_.size(); i++) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(6, attr_.get(i));
-    }
-    if (!getSummaryBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(7, summary_);
-    }
-    if (!getDescriptionBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(8, description_);
-    }
-    if (!getDescriptionPrefixBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(9, descriptionPrefix_);
-    }
-    if (!getDescriptionSuffixBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(10, descriptionSuffix_);
-    }
-    {
-      int dataSize = 0;
-      for (int i = 0; i < argOrder_.size(); i++) {
-        dataSize += computeStringSizeNoTag(argOrder_.getRaw(i));
-      }
-      size += dataSize;
-      size += 1 * getArgOrderList().size();
-    }
-    if (!getDeprecationMessageBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(12, deprecationMessage_);
-    }
-    if (deprecationVersion_ != 0) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt32Size(13, deprecationVersion_);
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.framework.ApiDef)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.framework.ApiDef other = (org.tensorflow.proto.framework.ApiDef) obj;
-
-    if (!getGraphOpName()
-        .equals(other.getGraphOpName())) return false;
-    if (!getDeprecationMessage()
-        .equals(other.getDeprecationMessage())) return false;
-    if (getDeprecationVersion()
-        != other.getDeprecationVersion()) return false;
-    if (visibility_ != other.visibility_) return false;
-    if (!getEndpointList()
-        .equals(other.getEndpointList())) return false;
-    if (!getInArgList()
-        .equals(other.getInArgList())) return false;
-    if (!getOutArgList()
-        .equals(other.getOutArgList())) return false;
-    if (!getArgOrderList()
-        .equals(other.getArgOrderList())) return false;
-    if (!getAttrList()
-        .equals(other.getAttrList())) return false;
-    if (!getSummary()
-        .equals(other.getSummary())) return false;
-    if (!getDescription()
-        .equals(other.getDescription())) return false;
-    if (!getDescriptionPrefix()
-        .equals(other.getDescriptionPrefix())) return false;
-    if (!getDescriptionSuffix()
-        .equals(other.getDescriptionSuffix())) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    hash = (37 * hash) + GRAPH_OP_NAME_FIELD_NUMBER;
-    hash = (53 * hash) + getGraphOpName().hashCode();
-    hash = (37 * hash) + DEPRECATION_MESSAGE_FIELD_NUMBER;
-    hash = (53 * hash) + getDeprecationMessage().hashCode();
-    hash = (37 * hash) + DEPRECATION_VERSION_FIELD_NUMBER;
-    hash = (53 * hash) + getDeprecationVersion();
-    hash = (37 * hash) + VISIBILITY_FIELD_NUMBER;
-    hash = (53 * hash) + visibility_;
-    if (getEndpointCount() > 0) {
-      hash = (37 * hash) + ENDPOINT_FIELD_NUMBER;
-      hash = (53 * hash) + getEndpointList().hashCode();
-    }
-    if (getInArgCount() > 0) {
-      hash = (37 * hash) + IN_ARG_FIELD_NUMBER;
-      hash = (53 * hash) + getInArgList().hashCode();
-    }
-    if (getOutArgCount() > 0) {
-      hash = (37 * hash) + OUT_ARG_FIELD_NUMBER;
-      hash = (53 * hash) + getOutArgList().hashCode();
-    }
-    if (getArgOrderCount() > 0) {
-      hash = (37 * hash) + ARG_ORDER_FIELD_NUMBER;
-      hash = (53 * hash) + getArgOrderList().hashCode();
-    }
-    if (getAttrCount() > 0) {
-      hash = (37 * hash) + ATTR_FIELD_NUMBER;
-      hash = (53 * hash) + getAttrList().hashCode();
-    }
-    hash = (37 * hash) + SUMMARY_FIELD_NUMBER;
-    hash = (53 * hash) + getSummary().hashCode();
-    hash = (37 * hash) + DESCRIPTION_FIELD_NUMBER;
-    hash = (53 * hash) + getDescription().hashCode();
-    hash = (37 * hash) + DESCRIPTION_PREFIX_FIELD_NUMBER;
-    hash = (53 * hash) + getDescriptionPrefix().hashCode();
-    hash = (37 * hash) + DESCRIPTION_SUFFIX_FIELD_NUMBER;
-    hash = (53 * hash) + getDescriptionSuffix().hashCode();
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.framework.ApiDef parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.ApiDef parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.ApiDef parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.ApiDef parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.ApiDef parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.ApiDef parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.ApiDef parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.ApiDef parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.ApiDef parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.ApiDef parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.ApiDef parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.ApiDef parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.framework.ApiDef prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * <pre>
-   * Used to specify and override the default API &amp; behavior in the
-   * generated code for client languages, from what you would get from
-   * the OpDef alone. There will be a set of ApiDefs that are common
-   * to all client languages, and another set per client language.
-   * The per-client-language ApiDefs will inherit values from the
-   * common ApiDefs which it can either replace or modify.
-   * We separate the API definition from the OpDef so we can evolve the
-   * API while remaining backwards compatible when interpretting old
-   * graphs.  Overrides go in an "api_def.pbtxt" file with a text-format
-   * ApiDefs message.
-   * WARNING: Be *very* careful changing the API for any existing op --
-   * you can change the semantics of existing code.  These changes may
-   * need to wait until a major release of TensorFlow to avoid breaking
-   * our compatibility promises.
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.ApiDef}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.ApiDef)
-      org.tensorflow.proto.framework.ApiDefOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.ApiDefProtos.internal_static_tensorflow_ApiDef_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.ApiDefProtos.internal_static_tensorflow_ApiDef_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.ApiDef.class, org.tensorflow.proto.framework.ApiDef.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.framework.ApiDef.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-        getEndpointFieldBuilder();
-        getInArgFieldBuilder();
-        getOutArgFieldBuilder();
-        getAttrFieldBuilder();
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      graphOpName_ = "";
-
-      deprecationMessage_ = "";
-
-      deprecationVersion_ = 0;
-
-      visibility_ = 0;
-
-      if (endpointBuilder_ == null) {
-        endpoint_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000001);
-      } else {
-        endpointBuilder_.clear();
-      }
-      if (inArgBuilder_ == null) {
-        inArg_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000002);
-      } else {
-        inArgBuilder_.clear();
-      }
-      if (outArgBuilder_ == null) {
-        outArg_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000004);
-      } else {
-        outArgBuilder_.clear();
-      }
-      argOrder_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-      bitField0_ = (bitField0_ & ~0x00000008);
-      if (attrBuilder_ == null) {
-        attr_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000010);
-      } else {
-        attrBuilder_.clear();
-      }
-      summary_ = "";
-
-      description_ = "";
-
-      descriptionPrefix_ = "";
-
-      descriptionSuffix_ = "";
-
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.framework.ApiDefProtos.internal_static_tensorflow_ApiDef_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.ApiDef getDefaultInstanceForType() {
-      return org.tensorflow.proto.framework.ApiDef.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.ApiDef build() {
-      org.tensorflow.proto.framework.ApiDef result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.ApiDef buildPartial() {
-      org.tensorflow.proto.framework.ApiDef result = new org.tensorflow.proto.framework.ApiDef(this);
-      int from_bitField0_ = bitField0_;
-      result.graphOpName_ = graphOpName_;
-      result.deprecationMessage_ = deprecationMessage_;
-      result.deprecationVersion_ = deprecationVersion_;
-      result.visibility_ = visibility_;
-      if (endpointBuilder_ == null) {
-        if (((bitField0_ & 0x00000001) != 0)) {
-          endpoint_ = java.util.Collections.unmodifiableList(endpoint_);
-          bitField0_ = (bitField0_ & ~0x00000001);
-        }
-        result.endpoint_ = endpoint_;
-      } else {
-        result.endpoint_ = endpointBuilder_.build();
-      }
-      if (inArgBuilder_ == null) {
-        if (((bitField0_ & 0x00000002) != 0)) {
-          inArg_ = java.util.Collections.unmodifiableList(inArg_);
-          bitField0_ = (bitField0_ & ~0x00000002);
-        }
-        result.inArg_ = inArg_;
-      } else {
-        result.inArg_ = inArgBuilder_.build();
-      }
-      if (outArgBuilder_ == null) {
-        if (((bitField0_ & 0x00000004) != 0)) {
-          outArg_ = java.util.Collections.unmodifiableList(outArg_);
-          bitField0_ = (bitField0_ & ~0x00000004);
-        }
-        result.outArg_ = outArg_;
-      } else {
-        result.outArg_ = outArgBuilder_.build();
-      }
-      if (((bitField0_ & 0x00000008) != 0)) {
-        argOrder_ = argOrder_.getUnmodifiableView();
-        bitField0_ = (bitField0_ & ~0x00000008);
-      }
-      result.argOrder_ = argOrder_;
-      if (attrBuilder_ == null) {
-        if (((bitField0_ & 0x00000010) != 0)) {
-          attr_ = java.util.Collections.unmodifiableList(attr_);
-          bitField0_ = (bitField0_ & ~0x00000010);
-        }
-        result.attr_ = attr_;
-      } else {
-        result.attr_ = attrBuilder_.build();
-      }
-      result.summary_ = summary_;
-      result.description_ = description_;
-      result.descriptionPrefix_ = descriptionPrefix_;
-      result.descriptionSuffix_ = descriptionSuffix_;
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.framework.ApiDef) {
-        return mergeFrom((org.tensorflow.proto.framework.ApiDef)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.framework.ApiDef other) {
-      if (other == org.tensorflow.proto.framework.ApiDef.getDefaultInstance()) return this;
-      if (!other.getGraphOpName().isEmpty()) {
-        graphOpName_ = other.graphOpName_;
-        onChanged();
-      }
-      if (!other.getDeprecationMessage().isEmpty()) {
-        deprecationMessage_ = other.deprecationMessage_;
-        onChanged();
-      }
-      if (other.getDeprecationVersion() != 0) {
-        setDeprecationVersion(other.getDeprecationVersion());
-      }
-      if (other.visibility_ != 0) {
-        setVisibilityValue(other.getVisibilityValue());
-      }
-      if (endpointBuilder_ == null) {
-        if (!other.endpoint_.isEmpty()) {
-          if (endpoint_.isEmpty()) {
-            endpoint_ = other.endpoint_;
-            bitField0_ = (bitField0_ & ~0x00000001);
-          } else {
-            ensureEndpointIsMutable();
-            endpoint_.addAll(other.endpoint_);
-          }
-          onChanged();
-        }
-      } else {
-        if (!other.endpoint_.isEmpty()) {
-          if (endpointBuilder_.isEmpty()) {
-            endpointBuilder_.dispose();
-            endpointBuilder_ = null;
-            endpoint_ = other.endpoint_;
-            bitField0_ = (bitField0_ & ~0x00000001);
-            endpointBuilder_ = 
-              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
-                 getEndpointFieldBuilder() : null;
-          } else {
-            endpointBuilder_.addAllMessages(other.endpoint_);
-          }
-        }
-      }
-      if (inArgBuilder_ == null) {
-        if (!other.inArg_.isEmpty()) {
-          if (inArg_.isEmpty()) {
-            inArg_ = other.inArg_;
-            bitField0_ = (bitField0_ & ~0x00000002);
-          } else {
-            ensureInArgIsMutable();
-            inArg_.addAll(other.inArg_);
-          }
-          onChanged();
-        }
-      } else {
-        if (!other.inArg_.isEmpty()) {
-          if (inArgBuilder_.isEmpty()) {
-            inArgBuilder_.dispose();
-            inArgBuilder_ = null;
-            inArg_ = other.inArg_;
-            bitField0_ = (bitField0_ & ~0x00000002);
-            inArgBuilder_ = 
-              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
-                 getInArgFieldBuilder() : null;
-          } else {
-            inArgBuilder_.addAllMessages(other.inArg_);
-          }
-        }
-      }
-      if (outArgBuilder_ == null) {
-        if (!other.outArg_.isEmpty()) {
-          if (outArg_.isEmpty()) {
-            outArg_ = other.outArg_;
-            bitField0_ = (bitField0_ & ~0x00000004);
-          } else {
-            ensureOutArgIsMutable();
-            outArg_.addAll(other.outArg_);
-          }
-          onChanged();
-        }
-      } else {
-        if (!other.outArg_.isEmpty()) {
-          if (outArgBuilder_.isEmpty()) {
-            outArgBuilder_.dispose();
-            outArgBuilder_ = null;
-            outArg_ = other.outArg_;
-            bitField0_ = (bitField0_ & ~0x00000004);
-            outArgBuilder_ = 
-              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
-                 getOutArgFieldBuilder() : null;
-          } else {
-            outArgBuilder_.addAllMessages(other.outArg_);
-          }
-        }
-      }
-      if (!other.argOrder_.isEmpty()) {
-        if (argOrder_.isEmpty()) {
-          argOrder_ = other.argOrder_;
-          bitField0_ = (bitField0_ & ~0x00000008);
-        } else {
-          ensureArgOrderIsMutable();
-          argOrder_.addAll(other.argOrder_);
-        }
-        onChanged();
-      }
-      if (attrBuilder_ == null) {
-        if (!other.attr_.isEmpty()) {
-          if (attr_.isEmpty()) {
-            attr_ = other.attr_;
-            bitField0_ = (bitField0_ & ~0x00000010);
-          } else {
-            ensureAttrIsMutable();
-            attr_.addAll(other.attr_);
-          }
-          onChanged();
-        }
-      } else {
-        if (!other.attr_.isEmpty()) {
-          if (attrBuilder_.isEmpty()) {
-            attrBuilder_.dispose();
-            attrBuilder_ = null;
-            attr_ = other.attr_;
-            bitField0_ = (bitField0_ & ~0x00000010);
-            attrBuilder_ = 
-              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
-                 getAttrFieldBuilder() : null;
-          } else {
-            attrBuilder_.addAllMessages(other.attr_);
-          }
-        }
-      }
-      if (!other.getSummary().isEmpty()) {
-        summary_ = other.summary_;
-        onChanged();
-      }
-      if (!other.getDescription().isEmpty()) {
-        description_ = other.description_;
-        onChanged();
-      }
-      if (!other.getDescriptionPrefix().isEmpty()) {
-        descriptionPrefix_ = other.descriptionPrefix_;
-        onChanged();
-      }
-      if (!other.getDescriptionSuffix().isEmpty()) {
-        descriptionSuffix_ = other.descriptionSuffix_;
-        onChanged();
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.framework.ApiDef parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.framework.ApiDef) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-    private int bitField0_;
-
-    private java.lang.Object graphOpName_ = "";
-    /**
-     * <pre>
-     * Name of the op (in the OpDef) to specify the API for.
-     * </pre>
-     *
-     * <code>string graph_op_name = 1;</code>
-     */
-    public java.lang.String getGraphOpName() {
-      java.lang.Object ref = graphOpName_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        graphOpName_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Name of the op (in the OpDef) to specify the API for.
-     * </pre>
-     *
-     * <code>string graph_op_name = 1;</code>
-     */
-    public com.google.protobuf.ByteString
-        getGraphOpNameBytes() {
-      java.lang.Object ref = graphOpName_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        graphOpName_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Name of the op (in the OpDef) to specify the API for.
-     * </pre>
-     *
-     * <code>string graph_op_name = 1;</code>
-     */
-    public Builder setGraphOpName(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      graphOpName_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Name of the op (in the OpDef) to specify the API for.
-     * </pre>
-     *
-     * <code>string graph_op_name = 1;</code>
-     */
-    public Builder clearGraphOpName() {
-      
-      graphOpName_ = getDefaultInstance().getGraphOpName();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Name of the op (in the OpDef) to specify the API for.
-     * </pre>
-     *
-     * <code>string graph_op_name = 1;</code>
-     */
-    public Builder setGraphOpNameBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      graphOpName_ = value;
-      onChanged();
-      return this;
-    }
-
-    private java.lang.Object deprecationMessage_ = "";
-    /**
-     * <pre>
-     * If this op is deprecated, set deprecation message to the message
-     * that should be logged when this op is used.
-     * The message should indicate alternative op to use, if any.
-     * </pre>
-     *
-     * <code>string deprecation_message = 12;</code>
-     */
-    public java.lang.String getDeprecationMessage() {
-      java.lang.Object ref = deprecationMessage_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        deprecationMessage_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * If this op is deprecated, set deprecation message to the message
-     * that should be logged when this op is used.
-     * The message should indicate alternative op to use, if any.
-     * </pre>
-     *
-     * <code>string deprecation_message = 12;</code>
-     */
-    public com.google.protobuf.ByteString
-        getDeprecationMessageBytes() {
-      java.lang.Object ref = deprecationMessage_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        deprecationMessage_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * If this op is deprecated, set deprecation message to the message
-     * that should be logged when this op is used.
-     * The message should indicate alternative op to use, if any.
-     * </pre>
-     *
-     * <code>string deprecation_message = 12;</code>
-     */
-    public Builder setDeprecationMessage(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      deprecationMessage_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * If this op is deprecated, set deprecation message to the message
-     * that should be logged when this op is used.
-     * The message should indicate alternative op to use, if any.
-     * </pre>
-     *
-     * <code>string deprecation_message = 12;</code>
-     */
-    public Builder clearDeprecationMessage() {
-      
-      deprecationMessage_ = getDefaultInstance().getDeprecationMessage();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * If this op is deprecated, set deprecation message to the message
-     * that should be logged when this op is used.
-     * The message should indicate alternative op to use, if any.
-     * </pre>
-     *
-     * <code>string deprecation_message = 12;</code>
-     */
-    public Builder setDeprecationMessageBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      deprecationMessage_ = value;
-      onChanged();
-      return this;
-    }
-
-    private int deprecationVersion_ ;
-    /**
-     * <pre>
-     * Major version when the op will be deleted. For e.g. set this
-     * value to 2 if op API should be removed in TensorFlow 2.0 and
-     * deprecated in versions before that.
-     * </pre>
-     *
-     * <code>int32 deprecation_version = 13;</code>
-     */
-    public int getDeprecationVersion() {
-      return deprecationVersion_;
-    }
-    /**
-     * <pre>
-     * Major version when the op will be deleted. For e.g. set this
-     * value to 2 if op API should be removed in TensorFlow 2.0 and
-     * deprecated in versions before that.
-     * </pre>
-     *
-     * <code>int32 deprecation_version = 13;</code>
-     */
-    public Builder setDeprecationVersion(int value) {
-      
-      deprecationVersion_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Major version when the op will be deleted. For e.g. set this
-     * value to 2 if op API should be removed in TensorFlow 2.0 and
-     * deprecated in versions before that.
-     * </pre>
-     *
-     * <code>int32 deprecation_version = 13;</code>
-     */
-    public Builder clearDeprecationVersion() {
-      
-      deprecationVersion_ = 0;
-      onChanged();
-      return this;
-    }
-
-    private int visibility_ = 0;
-    /**
-     * <code>.tensorflow.ApiDef.Visibility visibility = 2;</code>
-     */
-    public int getVisibilityValue() {
-      return visibility_;
-    }
-    /**
-     * <code>.tensorflow.ApiDef.Visibility visibility = 2;</code>
-     */
-    public Builder setVisibilityValue(int value) {
-      visibility_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>.tensorflow.ApiDef.Visibility visibility = 2;</code>
-     */
-    public org.tensorflow.proto.framework.ApiDef.Visibility getVisibility() {
-      @SuppressWarnings("deprecation")
-      org.tensorflow.proto.framework.ApiDef.Visibility result = org.tensorflow.proto.framework.ApiDef.Visibility.valueOf(visibility_);
-      return result == null ? org.tensorflow.proto.framework.ApiDef.Visibility.UNRECOGNIZED : result;
-    }
-    /**
-     * <code>.tensorflow.ApiDef.Visibility visibility = 2;</code>
-     */
-    public Builder setVisibility(org.tensorflow.proto.framework.ApiDef.Visibility value) {
-      if (value == null) {
-        throw new NullPointerException();
-      }
-      
-      visibility_ = value.getNumber();
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>.tensorflow.ApiDef.Visibility visibility = 2;</code>
-     */
-    public Builder clearVisibility() {
-      
-      visibility_ = 0;
-      onChanged();
-      return this;
-    }
-
-    private java.util.List<org.tensorflow.proto.framework.ApiDef.Endpoint> endpoint_ =
-      java.util.Collections.emptyList();
-    private void ensureEndpointIsMutable() {
-      if (!((bitField0_ & 0x00000001) != 0)) {
-        endpoint_ = new java.util.ArrayList<org.tensorflow.proto.framework.ApiDef.Endpoint>(endpoint_);
-        bitField0_ |= 0x00000001;
-       }
-    }
-
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.framework.ApiDef.Endpoint, org.tensorflow.proto.framework.ApiDef.Endpoint.Builder, org.tensorflow.proto.framework.ApiDef.EndpointOrBuilder> endpointBuilder_;
-
-    /**
-     * <code>repeated .tensorflow.ApiDef.Endpoint endpoint = 3;</code>
-     */
-    public java.util.List<org.tensorflow.proto.framework.ApiDef.Endpoint> getEndpointList() {
-      if (endpointBuilder_ == null) {
-        return java.util.Collections.unmodifiableList(endpoint_);
-      } else {
-        return endpointBuilder_.getMessageList();
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.ApiDef.Endpoint endpoint = 3;</code>
-     */
-    public int getEndpointCount() {
-      if (endpointBuilder_ == null) {
-        return endpoint_.size();
-      } else {
-        return endpointBuilder_.getCount();
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.ApiDef.Endpoint endpoint = 3;</code>
-     */
-    public org.tensorflow.proto.framework.ApiDef.Endpoint getEndpoint(int index) {
-      if (endpointBuilder_ == null) {
-        return endpoint_.get(index);
-      } else {
-        return endpointBuilder_.getMessage(index);
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.ApiDef.Endpoint endpoint = 3;</code>
-     */
-    public Builder setEndpoint(
-        int index, org.tensorflow.proto.framework.ApiDef.Endpoint value) {
-      if (endpointBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureEndpointIsMutable();
-        endpoint_.set(index, value);
-        onChanged();
-      } else {
-        endpointBuilder_.setMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.ApiDef.Endpoint endpoint = 3;</code>
-     */
-    public Builder setEndpoint(
-        int index, org.tensorflow.proto.framework.ApiDef.Endpoint.Builder builderForValue) {
-      if (endpointBuilder_ == null) {
-        ensureEndpointIsMutable();
-        endpoint_.set(index, builderForValue.build());
-        onChanged();
-      } else {
-        endpointBuilder_.setMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.ApiDef.Endpoint endpoint = 3;</code>
-     */
-    public Builder addEndpoint(org.tensorflow.proto.framework.ApiDef.Endpoint value) {
-      if (endpointBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureEndpointIsMutable();
-        endpoint_.add(value);
-        onChanged();
-      } else {
-        endpointBuilder_.addMessage(value);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.ApiDef.Endpoint endpoint = 3;</code>
-     */
-    public Builder addEndpoint(
-        int index, org.tensorflow.proto.framework.ApiDef.Endpoint value) {
-      if (endpointBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureEndpointIsMutable();
-        endpoint_.add(index, value);
-        onChanged();
-      } else {
-        endpointBuilder_.addMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.ApiDef.Endpoint endpoint = 3;</code>
-     */
-    public Builder addEndpoint(
-        org.tensorflow.proto.framework.ApiDef.Endpoint.Builder builderForValue) {
-      if (endpointBuilder_ == null) {
-        ensureEndpointIsMutable();
-        endpoint_.add(builderForValue.build());
-        onChanged();
-      } else {
-        endpointBuilder_.addMessage(builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.ApiDef.Endpoint endpoint = 3;</code>
-     */
-    public Builder addEndpoint(
-        int index, org.tensorflow.proto.framework.ApiDef.Endpoint.Builder builderForValue) {
-      if (endpointBuilder_ == null) {
-        ensureEndpointIsMutable();
-        endpoint_.add(index, builderForValue.build());
-        onChanged();
-      } else {
-        endpointBuilder_.addMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.ApiDef.Endpoint endpoint = 3;</code>
-     */
-    public Builder addAllEndpoint(
-        java.lang.Iterable<? extends org.tensorflow.proto.framework.ApiDef.Endpoint> values) {
-      if (endpointBuilder_ == null) {
-        ensureEndpointIsMutable();
-        com.google.protobuf.AbstractMessageLite.Builder.addAll(
-            values, endpoint_);
-        onChanged();
-      } else {
-        endpointBuilder_.addAllMessages(values);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.ApiDef.Endpoint endpoint = 3;</code>
-     */
-    public Builder clearEndpoint() {
-      if (endpointBuilder_ == null) {
-        endpoint_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000001);
-        onChanged();
-      } else {
-        endpointBuilder_.clear();
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.ApiDef.Endpoint endpoint = 3;</code>
-     */
-    public Builder removeEndpoint(int index) {
-      if (endpointBuilder_ == null) {
-        ensureEndpointIsMutable();
-        endpoint_.remove(index);
-        onChanged();
-      } else {
-        endpointBuilder_.remove(index);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.ApiDef.Endpoint endpoint = 3;</code>
-     */
-    public org.tensorflow.proto.framework.ApiDef.Endpoint.Builder getEndpointBuilder(
-        int index) {
-      return getEndpointFieldBuilder().getBuilder(index);
-    }
-    /**
-     * <code>repeated .tensorflow.ApiDef.Endpoint endpoint = 3;</code>
-     */
-    public org.tensorflow.proto.framework.ApiDef.EndpointOrBuilder getEndpointOrBuilder(
-        int index) {
-      if (endpointBuilder_ == null) {
-        return endpoint_.get(index);  } else {
-        return endpointBuilder_.getMessageOrBuilder(index);
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.ApiDef.Endpoint endpoint = 3;</code>
-     */
-    public java.util.List<? extends org.tensorflow.proto.framework.ApiDef.EndpointOrBuilder> 
-         getEndpointOrBuilderList() {
-      if (endpointBuilder_ != null) {
-        return endpointBuilder_.getMessageOrBuilderList();
-      } else {
-        return java.util.Collections.unmodifiableList(endpoint_);
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.ApiDef.Endpoint endpoint = 3;</code>
-     */
-    public org.tensorflow.proto.framework.ApiDef.Endpoint.Builder addEndpointBuilder() {
-      return getEndpointFieldBuilder().addBuilder(
-          org.tensorflow.proto.framework.ApiDef.Endpoint.getDefaultInstance());
-    }
-    /**
-     * <code>repeated .tensorflow.ApiDef.Endpoint endpoint = 3;</code>
-     */
-    public org.tensorflow.proto.framework.ApiDef.Endpoint.Builder addEndpointBuilder(
-        int index) {
-      return getEndpointFieldBuilder().addBuilder(
-          index, org.tensorflow.proto.framework.ApiDef.Endpoint.getDefaultInstance());
-    }
-    /**
-     * <code>repeated .tensorflow.ApiDef.Endpoint endpoint = 3;</code>
-     */
-    public java.util.List<org.tensorflow.proto.framework.ApiDef.Endpoint.Builder> 
-         getEndpointBuilderList() {
-      return getEndpointFieldBuilder().getBuilderList();
-    }
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.framework.ApiDef.Endpoint, org.tensorflow.proto.framework.ApiDef.Endpoint.Builder, org.tensorflow.proto.framework.ApiDef.EndpointOrBuilder> 
-        getEndpointFieldBuilder() {
-      if (endpointBuilder_ == null) {
-        endpointBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
-            org.tensorflow.proto.framework.ApiDef.Endpoint, org.tensorflow.proto.framework.ApiDef.Endpoint.Builder, org.tensorflow.proto.framework.ApiDef.EndpointOrBuilder>(
-                endpoint_,
-                ((bitField0_ & 0x00000001) != 0),
-                getParentForChildren(),
-                isClean());
-        endpoint_ = null;
-      }
-      return endpointBuilder_;
-    }
-
-    private java.util.List<org.tensorflow.proto.framework.ApiDef.Arg> inArg_ =
-      java.util.Collections.emptyList();
-    private void ensureInArgIsMutable() {
-      if (!((bitField0_ & 0x00000002) != 0)) {
-        inArg_ = new java.util.ArrayList<org.tensorflow.proto.framework.ApiDef.Arg>(inArg_);
-        bitField0_ |= 0x00000002;
-       }
-    }
-
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.framework.ApiDef.Arg, org.tensorflow.proto.framework.ApiDef.Arg.Builder, org.tensorflow.proto.framework.ApiDef.ArgOrBuilder> inArgBuilder_;
-
-    /**
-     * <code>repeated .tensorflow.ApiDef.Arg in_arg = 4;</code>
-     */
-    public java.util.List<org.tensorflow.proto.framework.ApiDef.Arg> getInArgList() {
-      if (inArgBuilder_ == null) {
-        return java.util.Collections.unmodifiableList(inArg_);
-      } else {
-        return inArgBuilder_.getMessageList();
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.ApiDef.Arg in_arg = 4;</code>
-     */
-    public int getInArgCount() {
-      if (inArgBuilder_ == null) {
-        return inArg_.size();
-      } else {
-        return inArgBuilder_.getCount();
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.ApiDef.Arg in_arg = 4;</code>
-     */
-    public org.tensorflow.proto.framework.ApiDef.Arg getInArg(int index) {
-      if (inArgBuilder_ == null) {
-        return inArg_.get(index);
-      } else {
-        return inArgBuilder_.getMessage(index);
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.ApiDef.Arg in_arg = 4;</code>
-     */
-    public Builder setInArg(
-        int index, org.tensorflow.proto.framework.ApiDef.Arg value) {
-      if (inArgBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureInArgIsMutable();
-        inArg_.set(index, value);
-        onChanged();
-      } else {
-        inArgBuilder_.setMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.ApiDef.Arg in_arg = 4;</code>
-     */
-    public Builder setInArg(
-        int index, org.tensorflow.proto.framework.ApiDef.Arg.Builder builderForValue) {
-      if (inArgBuilder_ == null) {
-        ensureInArgIsMutable();
-        inArg_.set(index, builderForValue.build());
-        onChanged();
-      } else {
-        inArgBuilder_.setMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.ApiDef.Arg in_arg = 4;</code>
-     */
-    public Builder addInArg(org.tensorflow.proto.framework.ApiDef.Arg value) {
-      if (inArgBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureInArgIsMutable();
-        inArg_.add(value);
-        onChanged();
-      } else {
-        inArgBuilder_.addMessage(value);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.ApiDef.Arg in_arg = 4;</code>
-     */
-    public Builder addInArg(
-        int index, org.tensorflow.proto.framework.ApiDef.Arg value) {
-      if (inArgBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureInArgIsMutable();
-        inArg_.add(index, value);
-        onChanged();
-      } else {
-        inArgBuilder_.addMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.ApiDef.Arg in_arg = 4;</code>
-     */
-    public Builder addInArg(
-        org.tensorflow.proto.framework.ApiDef.Arg.Builder builderForValue) {
-      if (inArgBuilder_ == null) {
-        ensureInArgIsMutable();
-        inArg_.add(builderForValue.build());
-        onChanged();
-      } else {
-        inArgBuilder_.addMessage(builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.ApiDef.Arg in_arg = 4;</code>
-     */
-    public Builder addInArg(
-        int index, org.tensorflow.proto.framework.ApiDef.Arg.Builder builderForValue) {
-      if (inArgBuilder_ == null) {
-        ensureInArgIsMutable();
-        inArg_.add(index, builderForValue.build());
-        onChanged();
-      } else {
-        inArgBuilder_.addMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.ApiDef.Arg in_arg = 4;</code>
-     */
-    public Builder addAllInArg(
-        java.lang.Iterable<? extends org.tensorflow.proto.framework.ApiDef.Arg> values) {
-      if (inArgBuilder_ == null) {
-        ensureInArgIsMutable();
-        com.google.protobuf.AbstractMessageLite.Builder.addAll(
-            values, inArg_);
-        onChanged();
-      } else {
-        inArgBuilder_.addAllMessages(values);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.ApiDef.Arg in_arg = 4;</code>
-     */
-    public Builder clearInArg() {
-      if (inArgBuilder_ == null) {
-        inArg_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000002);
-        onChanged();
-      } else {
-        inArgBuilder_.clear();
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.ApiDef.Arg in_arg = 4;</code>
-     */
-    public Builder removeInArg(int index) {
-      if (inArgBuilder_ == null) {
-        ensureInArgIsMutable();
-        inArg_.remove(index);
-        onChanged();
-      } else {
-        inArgBuilder_.remove(index);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.ApiDef.Arg in_arg = 4;</code>
-     */
-    public org.tensorflow.proto.framework.ApiDef.Arg.Builder getInArgBuilder(
-        int index) {
-      return getInArgFieldBuilder().getBuilder(index);
-    }
-    /**
-     * <code>repeated .tensorflow.ApiDef.Arg in_arg = 4;</code>
-     */
-    public org.tensorflow.proto.framework.ApiDef.ArgOrBuilder getInArgOrBuilder(
-        int index) {
-      if (inArgBuilder_ == null) {
-        return inArg_.get(index);  } else {
-        return inArgBuilder_.getMessageOrBuilder(index);
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.ApiDef.Arg in_arg = 4;</code>
-     */
-    public java.util.List<? extends org.tensorflow.proto.framework.ApiDef.ArgOrBuilder> 
-         getInArgOrBuilderList() {
-      if (inArgBuilder_ != null) {
-        return inArgBuilder_.getMessageOrBuilderList();
-      } else {
-        return java.util.Collections.unmodifiableList(inArg_);
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.ApiDef.Arg in_arg = 4;</code>
-     */
-    public org.tensorflow.proto.framework.ApiDef.Arg.Builder addInArgBuilder() {
-      return getInArgFieldBuilder().addBuilder(
-          org.tensorflow.proto.framework.ApiDef.Arg.getDefaultInstance());
-    }
-    /**
-     * <code>repeated .tensorflow.ApiDef.Arg in_arg = 4;</code>
-     */
-    public org.tensorflow.proto.framework.ApiDef.Arg.Builder addInArgBuilder(
-        int index) {
-      return getInArgFieldBuilder().addBuilder(
-          index, org.tensorflow.proto.framework.ApiDef.Arg.getDefaultInstance());
-    }
-    /**
-     * <code>repeated .tensorflow.ApiDef.Arg in_arg = 4;</code>
-     */
-    public java.util.List<org.tensorflow.proto.framework.ApiDef.Arg.Builder> 
-         getInArgBuilderList() {
-      return getInArgFieldBuilder().getBuilderList();
-    }
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.framework.ApiDef.Arg, org.tensorflow.proto.framework.ApiDef.Arg.Builder, org.tensorflow.proto.framework.ApiDef.ArgOrBuilder> 
-        getInArgFieldBuilder() {
-      if (inArgBuilder_ == null) {
-        inArgBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
-            org.tensorflow.proto.framework.ApiDef.Arg, org.tensorflow.proto.framework.ApiDef.Arg.Builder, org.tensorflow.proto.framework.ApiDef.ArgOrBuilder>(
-                inArg_,
-                ((bitField0_ & 0x00000002) != 0),
-                getParentForChildren(),
-                isClean());
-        inArg_ = null;
-      }
-      return inArgBuilder_;
-    }
-
-    private java.util.List<org.tensorflow.proto.framework.ApiDef.Arg> outArg_ =
-      java.util.Collections.emptyList();
-    private void ensureOutArgIsMutable() {
-      if (!((bitField0_ & 0x00000004) != 0)) {
-        outArg_ = new java.util.ArrayList<org.tensorflow.proto.framework.ApiDef.Arg>(outArg_);
-        bitField0_ |= 0x00000004;
-       }
-    }
-
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.framework.ApiDef.Arg, org.tensorflow.proto.framework.ApiDef.Arg.Builder, org.tensorflow.proto.framework.ApiDef.ArgOrBuilder> outArgBuilder_;
-
-    /**
-     * <code>repeated .tensorflow.ApiDef.Arg out_arg = 5;</code>
-     */
-    public java.util.List<org.tensorflow.proto.framework.ApiDef.Arg> getOutArgList() {
-      if (outArgBuilder_ == null) {
-        return java.util.Collections.unmodifiableList(outArg_);
-      } else {
-        return outArgBuilder_.getMessageList();
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.ApiDef.Arg out_arg = 5;</code>
-     */
-    public int getOutArgCount() {
-      if (outArgBuilder_ == null) {
-        return outArg_.size();
-      } else {
-        return outArgBuilder_.getCount();
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.ApiDef.Arg out_arg = 5;</code>
-     */
-    public org.tensorflow.proto.framework.ApiDef.Arg getOutArg(int index) {
-      if (outArgBuilder_ == null) {
-        return outArg_.get(index);
-      } else {
-        return outArgBuilder_.getMessage(index);
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.ApiDef.Arg out_arg = 5;</code>
-     */
-    public Builder setOutArg(
-        int index, org.tensorflow.proto.framework.ApiDef.Arg value) {
-      if (outArgBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureOutArgIsMutable();
-        outArg_.set(index, value);
-        onChanged();
-      } else {
-        outArgBuilder_.setMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.ApiDef.Arg out_arg = 5;</code>
-     */
-    public Builder setOutArg(
-        int index, org.tensorflow.proto.framework.ApiDef.Arg.Builder builderForValue) {
-      if (outArgBuilder_ == null) {
-        ensureOutArgIsMutable();
-        outArg_.set(index, builderForValue.build());
-        onChanged();
-      } else {
-        outArgBuilder_.setMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.ApiDef.Arg out_arg = 5;</code>
-     */
-    public Builder addOutArg(org.tensorflow.proto.framework.ApiDef.Arg value) {
-      if (outArgBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureOutArgIsMutable();
-        outArg_.add(value);
-        onChanged();
-      } else {
-        outArgBuilder_.addMessage(value);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.ApiDef.Arg out_arg = 5;</code>
-     */
-    public Builder addOutArg(
-        int index, org.tensorflow.proto.framework.ApiDef.Arg value) {
-      if (outArgBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureOutArgIsMutable();
-        outArg_.add(index, value);
-        onChanged();
-      } else {
-        outArgBuilder_.addMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.ApiDef.Arg out_arg = 5;</code>
-     */
-    public Builder addOutArg(
-        org.tensorflow.proto.framework.ApiDef.Arg.Builder builderForValue) {
-      if (outArgBuilder_ == null) {
-        ensureOutArgIsMutable();
-        outArg_.add(builderForValue.build());
-        onChanged();
-      } else {
-        outArgBuilder_.addMessage(builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.ApiDef.Arg out_arg = 5;</code>
-     */
-    public Builder addOutArg(
-        int index, org.tensorflow.proto.framework.ApiDef.Arg.Builder builderForValue) {
-      if (outArgBuilder_ == null) {
-        ensureOutArgIsMutable();
-        outArg_.add(index, builderForValue.build());
-        onChanged();
-      } else {
-        outArgBuilder_.addMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.ApiDef.Arg out_arg = 5;</code>
-     */
-    public Builder addAllOutArg(
-        java.lang.Iterable<? extends org.tensorflow.proto.framework.ApiDef.Arg> values) {
-      if (outArgBuilder_ == null) {
-        ensureOutArgIsMutable();
-        com.google.protobuf.AbstractMessageLite.Builder.addAll(
-            values, outArg_);
-        onChanged();
-      } else {
-        outArgBuilder_.addAllMessages(values);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.ApiDef.Arg out_arg = 5;</code>
-     */
-    public Builder clearOutArg() {
-      if (outArgBuilder_ == null) {
-        outArg_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000004);
-        onChanged();
-      } else {
-        outArgBuilder_.clear();
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.ApiDef.Arg out_arg = 5;</code>
-     */
-    public Builder removeOutArg(int index) {
-      if (outArgBuilder_ == null) {
-        ensureOutArgIsMutable();
-        outArg_.remove(index);
-        onChanged();
-      } else {
-        outArgBuilder_.remove(index);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.ApiDef.Arg out_arg = 5;</code>
-     */
-    public org.tensorflow.proto.framework.ApiDef.Arg.Builder getOutArgBuilder(
-        int index) {
-      return getOutArgFieldBuilder().getBuilder(index);
-    }
-    /**
-     * <code>repeated .tensorflow.ApiDef.Arg out_arg = 5;</code>
-     */
-    public org.tensorflow.proto.framework.ApiDef.ArgOrBuilder getOutArgOrBuilder(
-        int index) {
-      if (outArgBuilder_ == null) {
-        return outArg_.get(index);  } else {
-        return outArgBuilder_.getMessageOrBuilder(index);
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.ApiDef.Arg out_arg = 5;</code>
-     */
-    public java.util.List<? extends org.tensorflow.proto.framework.ApiDef.ArgOrBuilder> 
-         getOutArgOrBuilderList() {
-      if (outArgBuilder_ != null) {
-        return outArgBuilder_.getMessageOrBuilderList();
-      } else {
-        return java.util.Collections.unmodifiableList(outArg_);
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.ApiDef.Arg out_arg = 5;</code>
-     */
-    public org.tensorflow.proto.framework.ApiDef.Arg.Builder addOutArgBuilder() {
-      return getOutArgFieldBuilder().addBuilder(
-          org.tensorflow.proto.framework.ApiDef.Arg.getDefaultInstance());
-    }
-    /**
-     * <code>repeated .tensorflow.ApiDef.Arg out_arg = 5;</code>
-     */
-    public org.tensorflow.proto.framework.ApiDef.Arg.Builder addOutArgBuilder(
-        int index) {
-      return getOutArgFieldBuilder().addBuilder(
-          index, org.tensorflow.proto.framework.ApiDef.Arg.getDefaultInstance());
-    }
-    /**
-     * <code>repeated .tensorflow.ApiDef.Arg out_arg = 5;</code>
-     */
-    public java.util.List<org.tensorflow.proto.framework.ApiDef.Arg.Builder> 
-         getOutArgBuilderList() {
-      return getOutArgFieldBuilder().getBuilderList();
-    }
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.framework.ApiDef.Arg, org.tensorflow.proto.framework.ApiDef.Arg.Builder, org.tensorflow.proto.framework.ApiDef.ArgOrBuilder> 
-        getOutArgFieldBuilder() {
-      if (outArgBuilder_ == null) {
-        outArgBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
-            org.tensorflow.proto.framework.ApiDef.Arg, org.tensorflow.proto.framework.ApiDef.Arg.Builder, org.tensorflow.proto.framework.ApiDef.ArgOrBuilder>(
-                outArg_,
-                ((bitField0_ & 0x00000004) != 0),
-                getParentForChildren(),
-                isClean());
-        outArg_ = null;
-      }
-      return outArgBuilder_;
-    }
-
-    private com.google.protobuf.LazyStringList argOrder_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-    private void ensureArgOrderIsMutable() {
-      if (!((bitField0_ & 0x00000008) != 0)) {
-        argOrder_ = new com.google.protobuf.LazyStringArrayList(argOrder_);
-        bitField0_ |= 0x00000008;
-       }
-    }
-    /**
-     * <pre>
-     * List of original in_arg names to specify new argument order.
-     * Length of arg_order should be either empty to keep current order
-     * or match size of in_arg.
-     * </pre>
-     *
-     * <code>repeated string arg_order = 11;</code>
-     */
-    public com.google.protobuf.ProtocolStringList
-        getArgOrderList() {
-      return argOrder_.getUnmodifiableView();
-    }
-    /**
-     * <pre>
-     * List of original in_arg names to specify new argument order.
-     * Length of arg_order should be either empty to keep current order
-     * or match size of in_arg.
-     * </pre>
-     *
-     * <code>repeated string arg_order = 11;</code>
-     */
-    public int getArgOrderCount() {
-      return argOrder_.size();
-    }
-    /**
-     * <pre>
-     * List of original in_arg names to specify new argument order.
-     * Length of arg_order should be either empty to keep current order
-     * or match size of in_arg.
-     * </pre>
-     *
-     * <code>repeated string arg_order = 11;</code>
-     */
-    public java.lang.String getArgOrder(int index) {
-      return argOrder_.get(index);
-    }
-    /**
-     * <pre>
-     * List of original in_arg names to specify new argument order.
-     * Length of arg_order should be either empty to keep current order
-     * or match size of in_arg.
-     * </pre>
-     *
-     * <code>repeated string arg_order = 11;</code>
-     */
-    public com.google.protobuf.ByteString
-        getArgOrderBytes(int index) {
-      return argOrder_.getByteString(index);
-    }
-    /**
-     * <pre>
-     * List of original in_arg names to specify new argument order.
-     * Length of arg_order should be either empty to keep current order
-     * or match size of in_arg.
-     * </pre>
-     *
-     * <code>repeated string arg_order = 11;</code>
-     */
-    public Builder setArgOrder(
-        int index, java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  ensureArgOrderIsMutable();
-      argOrder_.set(index, value);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * List of original in_arg names to specify new argument order.
-     * Length of arg_order should be either empty to keep current order
-     * or match size of in_arg.
-     * </pre>
-     *
-     * <code>repeated string arg_order = 11;</code>
-     */
-    public Builder addArgOrder(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  ensureArgOrderIsMutable();
-      argOrder_.add(value);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * List of original in_arg names to specify new argument order.
-     * Length of arg_order should be either empty to keep current order
-     * or match size of in_arg.
-     * </pre>
-     *
-     * <code>repeated string arg_order = 11;</code>
-     */
-    public Builder addAllArgOrder(
-        java.lang.Iterable<java.lang.String> values) {
-      ensureArgOrderIsMutable();
-      com.google.protobuf.AbstractMessageLite.Builder.addAll(
-          values, argOrder_);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * List of original in_arg names to specify new argument order.
-     * Length of arg_order should be either empty to keep current order
-     * or match size of in_arg.
-     * </pre>
-     *
-     * <code>repeated string arg_order = 11;</code>
-     */
-    public Builder clearArgOrder() {
-      argOrder_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-      bitField0_ = (bitField0_ & ~0x00000008);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * List of original in_arg names to specify new argument order.
-     * Length of arg_order should be either empty to keep current order
-     * or match size of in_arg.
-     * </pre>
-     *
-     * <code>repeated string arg_order = 11;</code>
-     */
-    public Builder addArgOrderBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      ensureArgOrderIsMutable();
-      argOrder_.add(value);
-      onChanged();
-      return this;
-    }
-
-    private java.util.List<org.tensorflow.proto.framework.ApiDef.Attr> attr_ =
-      java.util.Collections.emptyList();
-    private void ensureAttrIsMutable() {
-      if (!((bitField0_ & 0x00000010) != 0)) {
-        attr_ = new java.util.ArrayList<org.tensorflow.proto.framework.ApiDef.Attr>(attr_);
-        bitField0_ |= 0x00000010;
-       }
-    }
-
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.framework.ApiDef.Attr, org.tensorflow.proto.framework.ApiDef.Attr.Builder, org.tensorflow.proto.framework.ApiDef.AttrOrBuilder> attrBuilder_;
-
-    /**
-     * <code>repeated .tensorflow.ApiDef.Attr attr = 6;</code>
-     */
-    public java.util.List<org.tensorflow.proto.framework.ApiDef.Attr> getAttrList() {
-      if (attrBuilder_ == null) {
-        return java.util.Collections.unmodifiableList(attr_);
-      } else {
-        return attrBuilder_.getMessageList();
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.ApiDef.Attr attr = 6;</code>
-     */
-    public int getAttrCount() {
-      if (attrBuilder_ == null) {
-        return attr_.size();
-      } else {
-        return attrBuilder_.getCount();
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.ApiDef.Attr attr = 6;</code>
-     */
-    public org.tensorflow.proto.framework.ApiDef.Attr getAttr(int index) {
-      if (attrBuilder_ == null) {
-        return attr_.get(index);
-      } else {
-        return attrBuilder_.getMessage(index);
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.ApiDef.Attr attr = 6;</code>
-     */
-    public Builder setAttr(
-        int index, org.tensorflow.proto.framework.ApiDef.Attr value) {
-      if (attrBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureAttrIsMutable();
-        attr_.set(index, value);
-        onChanged();
-      } else {
-        attrBuilder_.setMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.ApiDef.Attr attr = 6;</code>
-     */
-    public Builder setAttr(
-        int index, org.tensorflow.proto.framework.ApiDef.Attr.Builder builderForValue) {
-      if (attrBuilder_ == null) {
-        ensureAttrIsMutable();
-        attr_.set(index, builderForValue.build());
-        onChanged();
-      } else {
-        attrBuilder_.setMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.ApiDef.Attr attr = 6;</code>
-     */
-    public Builder addAttr(org.tensorflow.proto.framework.ApiDef.Attr value) {
-      if (attrBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureAttrIsMutable();
-        attr_.add(value);
-        onChanged();
-      } else {
-        attrBuilder_.addMessage(value);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.ApiDef.Attr attr = 6;</code>
-     */
-    public Builder addAttr(
-        int index, org.tensorflow.proto.framework.ApiDef.Attr value) {
-      if (attrBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureAttrIsMutable();
-        attr_.add(index, value);
-        onChanged();
-      } else {
-        attrBuilder_.addMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.ApiDef.Attr attr = 6;</code>
-     */
-    public Builder addAttr(
-        org.tensorflow.proto.framework.ApiDef.Attr.Builder builderForValue) {
-      if (attrBuilder_ == null) {
-        ensureAttrIsMutable();
-        attr_.add(builderForValue.build());
-        onChanged();
-      } else {
-        attrBuilder_.addMessage(builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.ApiDef.Attr attr = 6;</code>
-     */
-    public Builder addAttr(
-        int index, org.tensorflow.proto.framework.ApiDef.Attr.Builder builderForValue) {
-      if (attrBuilder_ == null) {
-        ensureAttrIsMutable();
-        attr_.add(index, builderForValue.build());
-        onChanged();
-      } else {
-        attrBuilder_.addMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.ApiDef.Attr attr = 6;</code>
-     */
-    public Builder addAllAttr(
-        java.lang.Iterable<? extends org.tensorflow.proto.framework.ApiDef.Attr> values) {
-      if (attrBuilder_ == null) {
-        ensureAttrIsMutable();
-        com.google.protobuf.AbstractMessageLite.Builder.addAll(
-            values, attr_);
-        onChanged();
-      } else {
-        attrBuilder_.addAllMessages(values);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.ApiDef.Attr attr = 6;</code>
-     */
-    public Builder clearAttr() {
-      if (attrBuilder_ == null) {
-        attr_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000010);
-        onChanged();
-      } else {
-        attrBuilder_.clear();
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.ApiDef.Attr attr = 6;</code>
-     */
-    public Builder removeAttr(int index) {
-      if (attrBuilder_ == null) {
-        ensureAttrIsMutable();
-        attr_.remove(index);
-        onChanged();
-      } else {
-        attrBuilder_.remove(index);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.ApiDef.Attr attr = 6;</code>
-     */
-    public org.tensorflow.proto.framework.ApiDef.Attr.Builder getAttrBuilder(
-        int index) {
-      return getAttrFieldBuilder().getBuilder(index);
-    }
-    /**
-     * <code>repeated .tensorflow.ApiDef.Attr attr = 6;</code>
-     */
-    public org.tensorflow.proto.framework.ApiDef.AttrOrBuilder getAttrOrBuilder(
-        int index) {
-      if (attrBuilder_ == null) {
-        return attr_.get(index);  } else {
-        return attrBuilder_.getMessageOrBuilder(index);
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.ApiDef.Attr attr = 6;</code>
-     */
-    public java.util.List<? extends org.tensorflow.proto.framework.ApiDef.AttrOrBuilder> 
-         getAttrOrBuilderList() {
-      if (attrBuilder_ != null) {
-        return attrBuilder_.getMessageOrBuilderList();
-      } else {
-        return java.util.Collections.unmodifiableList(attr_);
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.ApiDef.Attr attr = 6;</code>
-     */
-    public org.tensorflow.proto.framework.ApiDef.Attr.Builder addAttrBuilder() {
-      return getAttrFieldBuilder().addBuilder(
-          org.tensorflow.proto.framework.ApiDef.Attr.getDefaultInstance());
-    }
-    /**
-     * <code>repeated .tensorflow.ApiDef.Attr attr = 6;</code>
-     */
-    public org.tensorflow.proto.framework.ApiDef.Attr.Builder addAttrBuilder(
-        int index) {
-      return getAttrFieldBuilder().addBuilder(
-          index, org.tensorflow.proto.framework.ApiDef.Attr.getDefaultInstance());
-    }
-    /**
-     * <code>repeated .tensorflow.ApiDef.Attr attr = 6;</code>
-     */
-    public java.util.List<org.tensorflow.proto.framework.ApiDef.Attr.Builder> 
-         getAttrBuilderList() {
-      return getAttrFieldBuilder().getBuilderList();
-    }
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.framework.ApiDef.Attr, org.tensorflow.proto.framework.ApiDef.Attr.Builder, org.tensorflow.proto.framework.ApiDef.AttrOrBuilder> 
-        getAttrFieldBuilder() {
-      if (attrBuilder_ == null) {
-        attrBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
-            org.tensorflow.proto.framework.ApiDef.Attr, org.tensorflow.proto.framework.ApiDef.Attr.Builder, org.tensorflow.proto.framework.ApiDef.AttrOrBuilder>(
-                attr_,
-                ((bitField0_ & 0x00000010) != 0),
-                getParentForChildren(),
-                isClean());
-        attr_ = null;
-      }
-      return attrBuilder_;
-    }
-
-    private java.lang.Object summary_ = "";
-    /**
-     * <pre>
-     * One-line human-readable description of what the Op does.
-     * </pre>
-     *
-     * <code>string summary = 7;</code>
-     */
-    public java.lang.String getSummary() {
-      java.lang.Object ref = summary_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        summary_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * One-line human-readable description of what the Op does.
-     * </pre>
-     *
-     * <code>string summary = 7;</code>
-     */
-    public com.google.protobuf.ByteString
-        getSummaryBytes() {
-      java.lang.Object ref = summary_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        summary_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * One-line human-readable description of what the Op does.
-     * </pre>
-     *
-     * <code>string summary = 7;</code>
-     */
-    public Builder setSummary(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      summary_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * One-line human-readable description of what the Op does.
-     * </pre>
-     *
-     * <code>string summary = 7;</code>
-     */
-    public Builder clearSummary() {
-      
-      summary_ = getDefaultInstance().getSummary();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * One-line human-readable description of what the Op does.
-     * </pre>
-     *
-     * <code>string summary = 7;</code>
-     */
-    public Builder setSummaryBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      summary_ = value;
-      onChanged();
-      return this;
-    }
-
-    private java.lang.Object description_ = "";
-    /**
-     * <pre>
-     * Additional, longer human-readable description of what the Op does.
-     * </pre>
-     *
-     * <code>string description = 8;</code>
-     */
-    public java.lang.String getDescription() {
-      java.lang.Object ref = description_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        description_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Additional, longer human-readable description of what the Op does.
-     * </pre>
-     *
-     * <code>string description = 8;</code>
-     */
-    public com.google.protobuf.ByteString
-        getDescriptionBytes() {
-      java.lang.Object ref = description_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        description_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Additional, longer human-readable description of what the Op does.
-     * </pre>
-     *
-     * <code>string description = 8;</code>
-     */
-    public Builder setDescription(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      description_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Additional, longer human-readable description of what the Op does.
-     * </pre>
-     *
-     * <code>string description = 8;</code>
-     */
-    public Builder clearDescription() {
-      
-      description_ = getDefaultInstance().getDescription();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Additional, longer human-readable description of what the Op does.
-     * </pre>
-     *
-     * <code>string description = 8;</code>
-     */
-    public Builder setDescriptionBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      description_ = value;
-      onChanged();
-      return this;
-    }
-
-    private java.lang.Object descriptionPrefix_ = "";
-    /**
-     * <pre>
-     * Modify an existing/inherited description by adding text to the beginning
-     * or end.
-     * </pre>
-     *
-     * <code>string description_prefix = 9;</code>
-     */
-    public java.lang.String getDescriptionPrefix() {
-      java.lang.Object ref = descriptionPrefix_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        descriptionPrefix_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Modify an existing/inherited description by adding text to the beginning
-     * or end.
-     * </pre>
-     *
-     * <code>string description_prefix = 9;</code>
-     */
-    public com.google.protobuf.ByteString
-        getDescriptionPrefixBytes() {
-      java.lang.Object ref = descriptionPrefix_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        descriptionPrefix_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Modify an existing/inherited description by adding text to the beginning
-     * or end.
-     * </pre>
-     *
-     * <code>string description_prefix = 9;</code>
-     */
-    public Builder setDescriptionPrefix(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      descriptionPrefix_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Modify an existing/inherited description by adding text to the beginning
-     * or end.
-     * </pre>
-     *
-     * <code>string description_prefix = 9;</code>
-     */
-    public Builder clearDescriptionPrefix() {
-      
-      descriptionPrefix_ = getDefaultInstance().getDescriptionPrefix();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Modify an existing/inherited description by adding text to the beginning
-     * or end.
-     * </pre>
-     *
-     * <code>string description_prefix = 9;</code>
-     */
-    public Builder setDescriptionPrefixBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      descriptionPrefix_ = value;
-      onChanged();
-      return this;
-    }
-
-    private java.lang.Object descriptionSuffix_ = "";
-    /**
-     * <code>string description_suffix = 10;</code>
-     */
-    public java.lang.String getDescriptionSuffix() {
-      java.lang.Object ref = descriptionSuffix_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        descriptionSuffix_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <code>string description_suffix = 10;</code>
-     */
-    public com.google.protobuf.ByteString
-        getDescriptionSuffixBytes() {
-      java.lang.Object ref = descriptionSuffix_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        descriptionSuffix_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <code>string description_suffix = 10;</code>
-     */
-    public Builder setDescriptionSuffix(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      descriptionSuffix_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>string description_suffix = 10;</code>
-     */
-    public Builder clearDescriptionSuffix() {
-      
-      descriptionSuffix_ = getDefaultInstance().getDescriptionSuffix();
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>string description_suffix = 10;</code>
-     */
-    public Builder setDescriptionSuffixBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      descriptionSuffix_ = value;
-      onChanged();
-      return this;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.ApiDef)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.ApiDef)
-  private static final org.tensorflow.proto.framework.ApiDef DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.ApiDef();
-  }
-
-  public static org.tensorflow.proto.framework.ApiDef getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<ApiDef>
-      PARSER = new com.google.protobuf.AbstractParser<ApiDef>() {
-    @java.lang.Override
-    public ApiDef parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new ApiDef(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<ApiDef> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<ApiDef> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.framework.ApiDef getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ApiDefOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ApiDefOrBuilder.java
deleted file mode 100644
index 1a435d93fdb..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ApiDefOrBuilder.java
+++ /dev/null
@@ -1,274 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/framework/api_def.proto
-
-package org.tensorflow.proto.framework;
-
-public interface ApiDefOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.ApiDef)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <pre>
-   * Name of the op (in the OpDef) to specify the API for.
-   * </pre>
-   *
-   * <code>string graph_op_name = 1;</code>
-   */
-  java.lang.String getGraphOpName();
-  /**
-   * <pre>
-   * Name of the op (in the OpDef) to specify the API for.
-   * </pre>
-   *
-   * <code>string graph_op_name = 1;</code>
-   */
-  com.google.protobuf.ByteString
-      getGraphOpNameBytes();
-
-  /**
-   * <pre>
-   * If this op is deprecated, set deprecation message to the message
-   * that should be logged when this op is used.
-   * The message should indicate alternative op to use, if any.
-   * </pre>
-   *
-   * <code>string deprecation_message = 12;</code>
-   */
-  java.lang.String getDeprecationMessage();
-  /**
-   * <pre>
-   * If this op is deprecated, set deprecation message to the message
-   * that should be logged when this op is used.
-   * The message should indicate alternative op to use, if any.
-   * </pre>
-   *
-   * <code>string deprecation_message = 12;</code>
-   */
-  com.google.protobuf.ByteString
-      getDeprecationMessageBytes();
-
-  /**
-   * <pre>
-   * Major version when the op will be deleted. For e.g. set this
-   * value to 2 if op API should be removed in TensorFlow 2.0 and
-   * deprecated in versions before that.
-   * </pre>
-   *
-   * <code>int32 deprecation_version = 13;</code>
-   */
-  int getDeprecationVersion();
-
-  /**
-   * <code>.tensorflow.ApiDef.Visibility visibility = 2;</code>
-   */
-  int getVisibilityValue();
-  /**
-   * <code>.tensorflow.ApiDef.Visibility visibility = 2;</code>
-   */
-  org.tensorflow.proto.framework.ApiDef.Visibility getVisibility();
-
-  /**
-   * <code>repeated .tensorflow.ApiDef.Endpoint endpoint = 3;</code>
-   */
-  java.util.List<org.tensorflow.proto.framework.ApiDef.Endpoint> 
-      getEndpointList();
-  /**
-   * <code>repeated .tensorflow.ApiDef.Endpoint endpoint = 3;</code>
-   */
-  org.tensorflow.proto.framework.ApiDef.Endpoint getEndpoint(int index);
-  /**
-   * <code>repeated .tensorflow.ApiDef.Endpoint endpoint = 3;</code>
-   */
-  int getEndpointCount();
-  /**
-   * <code>repeated .tensorflow.ApiDef.Endpoint endpoint = 3;</code>
-   */
-  java.util.List<? extends org.tensorflow.proto.framework.ApiDef.EndpointOrBuilder> 
-      getEndpointOrBuilderList();
-  /**
-   * <code>repeated .tensorflow.ApiDef.Endpoint endpoint = 3;</code>
-   */
-  org.tensorflow.proto.framework.ApiDef.EndpointOrBuilder getEndpointOrBuilder(
-      int index);
-
-  /**
-   * <code>repeated .tensorflow.ApiDef.Arg in_arg = 4;</code>
-   */
-  java.util.List<org.tensorflow.proto.framework.ApiDef.Arg> 
-      getInArgList();
-  /**
-   * <code>repeated .tensorflow.ApiDef.Arg in_arg = 4;</code>
-   */
-  org.tensorflow.proto.framework.ApiDef.Arg getInArg(int index);
-  /**
-   * <code>repeated .tensorflow.ApiDef.Arg in_arg = 4;</code>
-   */
-  int getInArgCount();
-  /**
-   * <code>repeated .tensorflow.ApiDef.Arg in_arg = 4;</code>
-   */
-  java.util.List<? extends org.tensorflow.proto.framework.ApiDef.ArgOrBuilder> 
-      getInArgOrBuilderList();
-  /**
-   * <code>repeated .tensorflow.ApiDef.Arg in_arg = 4;</code>
-   */
-  org.tensorflow.proto.framework.ApiDef.ArgOrBuilder getInArgOrBuilder(
-      int index);
-
-  /**
-   * <code>repeated .tensorflow.ApiDef.Arg out_arg = 5;</code>
-   */
-  java.util.List<org.tensorflow.proto.framework.ApiDef.Arg> 
-      getOutArgList();
-  /**
-   * <code>repeated .tensorflow.ApiDef.Arg out_arg = 5;</code>
-   */
-  org.tensorflow.proto.framework.ApiDef.Arg getOutArg(int index);
-  /**
-   * <code>repeated .tensorflow.ApiDef.Arg out_arg = 5;</code>
-   */
-  int getOutArgCount();
-  /**
-   * <code>repeated .tensorflow.ApiDef.Arg out_arg = 5;</code>
-   */
-  java.util.List<? extends org.tensorflow.proto.framework.ApiDef.ArgOrBuilder> 
-      getOutArgOrBuilderList();
-  /**
-   * <code>repeated .tensorflow.ApiDef.Arg out_arg = 5;</code>
-   */
-  org.tensorflow.proto.framework.ApiDef.ArgOrBuilder getOutArgOrBuilder(
-      int index);
-
-  /**
-   * <pre>
-   * List of original in_arg names to specify new argument order.
-   * Length of arg_order should be either empty to keep current order
-   * or match size of in_arg.
-   * </pre>
-   *
-   * <code>repeated string arg_order = 11;</code>
-   */
-  java.util.List<java.lang.String>
-      getArgOrderList();
-  /**
-   * <pre>
-   * List of original in_arg names to specify new argument order.
-   * Length of arg_order should be either empty to keep current order
-   * or match size of in_arg.
-   * </pre>
-   *
-   * <code>repeated string arg_order = 11;</code>
-   */
-  int getArgOrderCount();
-  /**
-   * <pre>
-   * List of original in_arg names to specify new argument order.
-   * Length of arg_order should be either empty to keep current order
-   * or match size of in_arg.
-   * </pre>
-   *
-   * <code>repeated string arg_order = 11;</code>
-   */
-  java.lang.String getArgOrder(int index);
-  /**
-   * <pre>
-   * List of original in_arg names to specify new argument order.
-   * Length of arg_order should be either empty to keep current order
-   * or match size of in_arg.
-   * </pre>
-   *
-   * <code>repeated string arg_order = 11;</code>
-   */
-  com.google.protobuf.ByteString
-      getArgOrderBytes(int index);
-
-  /**
-   * <code>repeated .tensorflow.ApiDef.Attr attr = 6;</code>
-   */
-  java.util.List<org.tensorflow.proto.framework.ApiDef.Attr> 
-      getAttrList();
-  /**
-   * <code>repeated .tensorflow.ApiDef.Attr attr = 6;</code>
-   */
-  org.tensorflow.proto.framework.ApiDef.Attr getAttr(int index);
-  /**
-   * <code>repeated .tensorflow.ApiDef.Attr attr = 6;</code>
-   */
-  int getAttrCount();
-  /**
-   * <code>repeated .tensorflow.ApiDef.Attr attr = 6;</code>
-   */
-  java.util.List<? extends org.tensorflow.proto.framework.ApiDef.AttrOrBuilder> 
-      getAttrOrBuilderList();
-  /**
-   * <code>repeated .tensorflow.ApiDef.Attr attr = 6;</code>
-   */
-  org.tensorflow.proto.framework.ApiDef.AttrOrBuilder getAttrOrBuilder(
-      int index);
-
-  /**
-   * <pre>
-   * One-line human-readable description of what the Op does.
-   * </pre>
-   *
-   * <code>string summary = 7;</code>
-   */
-  java.lang.String getSummary();
-  /**
-   * <pre>
-   * One-line human-readable description of what the Op does.
-   * </pre>
-   *
-   * <code>string summary = 7;</code>
-   */
-  com.google.protobuf.ByteString
-      getSummaryBytes();
-
-  /**
-   * <pre>
-   * Additional, longer human-readable description of what the Op does.
-   * </pre>
-   *
-   * <code>string description = 8;</code>
-   */
-  java.lang.String getDescription();
-  /**
-   * <pre>
-   * Additional, longer human-readable description of what the Op does.
-   * </pre>
-   *
-   * <code>string description = 8;</code>
-   */
-  com.google.protobuf.ByteString
-      getDescriptionBytes();
-
-  /**
-   * <pre>
-   * Modify an existing/inherited description by adding text to the beginning
-   * or end.
-   * </pre>
-   *
-   * <code>string description_prefix = 9;</code>
-   */
-  java.lang.String getDescriptionPrefix();
-  /**
-   * <pre>
-   * Modify an existing/inherited description by adding text to the beginning
-   * or end.
-   * </pre>
-   *
-   * <code>string description_prefix = 9;</code>
-   */
-  com.google.protobuf.ByteString
-      getDescriptionPrefixBytes();
-
-  /**
-   * <code>string description_suffix = 10;</code>
-   */
-  java.lang.String getDescriptionSuffix();
-  /**
-   * <code>string description_suffix = 10;</code>
-   */
-  com.google.protobuf.ByteString
-      getDescriptionSuffixBytes();
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ApiDefProtos.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ApiDefProtos.java
deleted file mode 100644
index 6df5379d411..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ApiDefProtos.java
+++ /dev/null
@@ -1,117 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/framework/api_def.proto
-
-package org.tensorflow.proto.framework;
-
-public final class ApiDefProtos {
-  private ApiDefProtos() {}
-  public static void registerAllExtensions(
-      com.google.protobuf.ExtensionRegistryLite registry) {
-  }
-
-  public static void registerAllExtensions(
-      com.google.protobuf.ExtensionRegistry registry) {
-    registerAllExtensions(
-        (com.google.protobuf.ExtensionRegistryLite) registry);
-  }
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_ApiDef_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_ApiDef_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_ApiDef_Endpoint_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_ApiDef_Endpoint_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_ApiDef_Arg_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_ApiDef_Arg_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_ApiDef_Attr_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_ApiDef_Attr_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_ApiDefs_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_ApiDefs_fieldAccessorTable;
-
-  public static com.google.protobuf.Descriptors.FileDescriptor
-      getDescriptor() {
-    return descriptor;
-  }
-  private static  com.google.protobuf.Descriptors.FileDescriptor
-      descriptor;
-  static {
-    java.lang.String[] descriptorData = {
-      "\n\'tensorflow/core/framework/api_def.prot" +
-      "o\022\ntensorflow\032*tensorflow/core/framework" +
-      "/attr_value.proto\"\341\005\n\006ApiDef\022\025\n\rgraph_op" +
-      "_name\030\001 \001(\t\022\033\n\023deprecation_message\030\014 \001(\t" +
-      "\022\033\n\023deprecation_version\030\r \001(\005\0221\n\nvisibil" +
-      "ity\030\002 \001(\0162\035.tensorflow.ApiDef.Visibility" +
-      "\022-\n\010endpoint\030\003 \003(\0132\033.tensorflow.ApiDef.E" +
-      "ndpoint\022&\n\006in_arg\030\004 \003(\0132\026.tensorflow.Api" +
-      "Def.Arg\022\'\n\007out_arg\030\005 \003(\0132\026.tensorflow.Ap" +
-      "iDef.Arg\022\021\n\targ_order\030\013 \003(\t\022%\n\004attr\030\006 \003(" +
-      "\0132\027.tensorflow.ApiDef.Attr\022\017\n\007summary\030\007 " +
-      "\001(\t\022\023\n\013description\030\010 \001(\t\022\032\n\022description_" +
-      "prefix\030\t \001(\t\022\032\n\022description_suffix\030\n \001(\t" +
-      "\032I\n\010Endpoint\022\014\n\004name\030\001 \001(\t\022\022\n\ndeprecated" +
-      "\030\003 \001(\010\022\033\n\023deprecation_version\030\004 \001(\005\032;\n\003A" +
-      "rg\022\014\n\004name\030\001 \001(\t\022\021\n\trename_to\030\002 \001(\t\022\023\n\013d" +
-      "escription\030\003 \001(\t\032j\n\004Attr\022\014\n\004name\030\001 \001(\t\022\021" +
-      "\n\trename_to\030\002 \001(\t\022,\n\rdefault_value\030\003 \001(\013" +
-      "2\025.tensorflow.AttrValue\022\023\n\013description\030\004" +
-      " \001(\t\"G\n\nVisibility\022\026\n\022DEFAULT_VISIBILITY" +
-      "\020\000\022\013\n\007VISIBLE\020\001\022\010\n\004SKIP\020\002\022\n\n\006HIDDEN\020\003\")\n" +
-      "\007ApiDefs\022\036\n\002op\030\001 \003(\0132\022.tensorflow.ApiDef" +
-      "B\203\001\n\036org.tensorflow.proto.frameworkB\014Api" +
-      "DefProtosP\001ZNgithub.com/tensorflow/tenso" +
-      "rflow/tensorflow/go/core/framework/api_d" +
-      "ef_go_proto\370\001\001b\006proto3"
-    };
-    descriptor = com.google.protobuf.Descriptors.FileDescriptor
-      .internalBuildGeneratedFileFrom(descriptorData,
-        new com.google.protobuf.Descriptors.FileDescriptor[] {
-          org.tensorflow.proto.framework.AttrValueProtos.getDescriptor(),
-        });
-    internal_static_tensorflow_ApiDef_descriptor =
-      getDescriptor().getMessageTypes().get(0);
-    internal_static_tensorflow_ApiDef_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_ApiDef_descriptor,
-        new java.lang.String[] { "GraphOpName", "DeprecationMessage", "DeprecationVersion", "Visibility", "Endpoint", "InArg", "OutArg", "ArgOrder", "Attr", "Summary", "Description", "DescriptionPrefix", "DescriptionSuffix", });
-    internal_static_tensorflow_ApiDef_Endpoint_descriptor =
-      internal_static_tensorflow_ApiDef_descriptor.getNestedTypes().get(0);
-    internal_static_tensorflow_ApiDef_Endpoint_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_ApiDef_Endpoint_descriptor,
-        new java.lang.String[] { "Name", "Deprecated", "DeprecationVersion", });
-    internal_static_tensorflow_ApiDef_Arg_descriptor =
-      internal_static_tensorflow_ApiDef_descriptor.getNestedTypes().get(1);
-    internal_static_tensorflow_ApiDef_Arg_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_ApiDef_Arg_descriptor,
-        new java.lang.String[] { "Name", "RenameTo", "Description", });
-    internal_static_tensorflow_ApiDef_Attr_descriptor =
-      internal_static_tensorflow_ApiDef_descriptor.getNestedTypes().get(2);
-    internal_static_tensorflow_ApiDef_Attr_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_ApiDef_Attr_descriptor,
-        new java.lang.String[] { "Name", "RenameTo", "DefaultValue", "Description", });
-    internal_static_tensorflow_ApiDefs_descriptor =
-      getDescriptor().getMessageTypes().get(1);
-    internal_static_tensorflow_ApiDefs_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_ApiDefs_descriptor,
-        new java.lang.String[] { "Op", });
-    org.tensorflow.proto.framework.AttrValueProtos.getDescriptor();
-  }
-
-  // @@protoc_insertion_point(outer_class_scope)
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ApiDefs.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ApiDefs.java
deleted file mode 100644
index a4076265967..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ApiDefs.java
+++ /dev/null
@@ -1,765 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/framework/api_def.proto
-
-package org.tensorflow.proto.framework;
-
-/**
- * Protobuf type {@code tensorflow.ApiDefs}
- */
-public  final class ApiDefs extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.ApiDefs)
-    ApiDefsOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use ApiDefs.newBuilder() to construct.
-  private ApiDefs(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private ApiDefs() {
-    op_ = java.util.Collections.emptyList();
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new ApiDefs();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private ApiDefs(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    int mutable_bitField0_ = 0;
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 10: {
-            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
-              op_ = new java.util.ArrayList<org.tensorflow.proto.framework.ApiDef>();
-              mutable_bitField0_ |= 0x00000001;
-            }
-            op_.add(
-                input.readMessage(org.tensorflow.proto.framework.ApiDef.parser(), extensionRegistry));
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      if (((mutable_bitField0_ & 0x00000001) != 0)) {
-        op_ = java.util.Collections.unmodifiableList(op_);
-      }
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.framework.ApiDefProtos.internal_static_tensorflow_ApiDefs_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.framework.ApiDefProtos.internal_static_tensorflow_ApiDefs_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.framework.ApiDefs.class, org.tensorflow.proto.framework.ApiDefs.Builder.class);
-  }
-
-  public static final int OP_FIELD_NUMBER = 1;
-  private java.util.List<org.tensorflow.proto.framework.ApiDef> op_;
-  /**
-   * <code>repeated .tensorflow.ApiDef op = 1;</code>
-   */
-  public java.util.List<org.tensorflow.proto.framework.ApiDef> getOpList() {
-    return op_;
-  }
-  /**
-   * <code>repeated .tensorflow.ApiDef op = 1;</code>
-   */
-  public java.util.List<? extends org.tensorflow.proto.framework.ApiDefOrBuilder> 
-      getOpOrBuilderList() {
-    return op_;
-  }
-  /**
-   * <code>repeated .tensorflow.ApiDef op = 1;</code>
-   */
-  public int getOpCount() {
-    return op_.size();
-  }
-  /**
-   * <code>repeated .tensorflow.ApiDef op = 1;</code>
-   */
-  public org.tensorflow.proto.framework.ApiDef getOp(int index) {
-    return op_.get(index);
-  }
-  /**
-   * <code>repeated .tensorflow.ApiDef op = 1;</code>
-   */
-  public org.tensorflow.proto.framework.ApiDefOrBuilder getOpOrBuilder(
-      int index) {
-    return op_.get(index);
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    for (int i = 0; i < op_.size(); i++) {
-      output.writeMessage(1, op_.get(i));
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    for (int i = 0; i < op_.size(); i++) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(1, op_.get(i));
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.framework.ApiDefs)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.framework.ApiDefs other = (org.tensorflow.proto.framework.ApiDefs) obj;
-
-    if (!getOpList()
-        .equals(other.getOpList())) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    if (getOpCount() > 0) {
-      hash = (37 * hash) + OP_FIELD_NUMBER;
-      hash = (53 * hash) + getOpList().hashCode();
-    }
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.framework.ApiDefs parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.ApiDefs parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.ApiDefs parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.ApiDefs parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.ApiDefs parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.ApiDefs parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.ApiDefs parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.ApiDefs parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.ApiDefs parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.ApiDefs parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.ApiDefs parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.ApiDefs parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.framework.ApiDefs prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * Protobuf type {@code tensorflow.ApiDefs}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.ApiDefs)
-      org.tensorflow.proto.framework.ApiDefsOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.ApiDefProtos.internal_static_tensorflow_ApiDefs_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.ApiDefProtos.internal_static_tensorflow_ApiDefs_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.ApiDefs.class, org.tensorflow.proto.framework.ApiDefs.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.framework.ApiDefs.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-        getOpFieldBuilder();
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      if (opBuilder_ == null) {
-        op_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000001);
-      } else {
-        opBuilder_.clear();
-      }
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.framework.ApiDefProtos.internal_static_tensorflow_ApiDefs_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.ApiDefs getDefaultInstanceForType() {
-      return org.tensorflow.proto.framework.ApiDefs.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.ApiDefs build() {
-      org.tensorflow.proto.framework.ApiDefs result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.ApiDefs buildPartial() {
-      org.tensorflow.proto.framework.ApiDefs result = new org.tensorflow.proto.framework.ApiDefs(this);
-      int from_bitField0_ = bitField0_;
-      if (opBuilder_ == null) {
-        if (((bitField0_ & 0x00000001) != 0)) {
-          op_ = java.util.Collections.unmodifiableList(op_);
-          bitField0_ = (bitField0_ & ~0x00000001);
-        }
-        result.op_ = op_;
-      } else {
-        result.op_ = opBuilder_.build();
-      }
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.framework.ApiDefs) {
-        return mergeFrom((org.tensorflow.proto.framework.ApiDefs)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.framework.ApiDefs other) {
-      if (other == org.tensorflow.proto.framework.ApiDefs.getDefaultInstance()) return this;
-      if (opBuilder_ == null) {
-        if (!other.op_.isEmpty()) {
-          if (op_.isEmpty()) {
-            op_ = other.op_;
-            bitField0_ = (bitField0_ & ~0x00000001);
-          } else {
-            ensureOpIsMutable();
-            op_.addAll(other.op_);
-          }
-          onChanged();
-        }
-      } else {
-        if (!other.op_.isEmpty()) {
-          if (opBuilder_.isEmpty()) {
-            opBuilder_.dispose();
-            opBuilder_ = null;
-            op_ = other.op_;
-            bitField0_ = (bitField0_ & ~0x00000001);
-            opBuilder_ = 
-              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
-                 getOpFieldBuilder() : null;
-          } else {
-            opBuilder_.addAllMessages(other.op_);
-          }
-        }
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.framework.ApiDefs parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.framework.ApiDefs) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-    private int bitField0_;
-
-    private java.util.List<org.tensorflow.proto.framework.ApiDef> op_ =
-      java.util.Collections.emptyList();
-    private void ensureOpIsMutable() {
-      if (!((bitField0_ & 0x00000001) != 0)) {
-        op_ = new java.util.ArrayList<org.tensorflow.proto.framework.ApiDef>(op_);
-        bitField0_ |= 0x00000001;
-       }
-    }
-
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.framework.ApiDef, org.tensorflow.proto.framework.ApiDef.Builder, org.tensorflow.proto.framework.ApiDefOrBuilder> opBuilder_;
-
-    /**
-     * <code>repeated .tensorflow.ApiDef op = 1;</code>
-     */
-    public java.util.List<org.tensorflow.proto.framework.ApiDef> getOpList() {
-      if (opBuilder_ == null) {
-        return java.util.Collections.unmodifiableList(op_);
-      } else {
-        return opBuilder_.getMessageList();
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.ApiDef op = 1;</code>
-     */
-    public int getOpCount() {
-      if (opBuilder_ == null) {
-        return op_.size();
-      } else {
-        return opBuilder_.getCount();
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.ApiDef op = 1;</code>
-     */
-    public org.tensorflow.proto.framework.ApiDef getOp(int index) {
-      if (opBuilder_ == null) {
-        return op_.get(index);
-      } else {
-        return opBuilder_.getMessage(index);
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.ApiDef op = 1;</code>
-     */
-    public Builder setOp(
-        int index, org.tensorflow.proto.framework.ApiDef value) {
-      if (opBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureOpIsMutable();
-        op_.set(index, value);
-        onChanged();
-      } else {
-        opBuilder_.setMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.ApiDef op = 1;</code>
-     */
-    public Builder setOp(
-        int index, org.tensorflow.proto.framework.ApiDef.Builder builderForValue) {
-      if (opBuilder_ == null) {
-        ensureOpIsMutable();
-        op_.set(index, builderForValue.build());
-        onChanged();
-      } else {
-        opBuilder_.setMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.ApiDef op = 1;</code>
-     */
-    public Builder addOp(org.tensorflow.proto.framework.ApiDef value) {
-      if (opBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureOpIsMutable();
-        op_.add(value);
-        onChanged();
-      } else {
-        opBuilder_.addMessage(value);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.ApiDef op = 1;</code>
-     */
-    public Builder addOp(
-        int index, org.tensorflow.proto.framework.ApiDef value) {
-      if (opBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureOpIsMutable();
-        op_.add(index, value);
-        onChanged();
-      } else {
-        opBuilder_.addMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.ApiDef op = 1;</code>
-     */
-    public Builder addOp(
-        org.tensorflow.proto.framework.ApiDef.Builder builderForValue) {
-      if (opBuilder_ == null) {
-        ensureOpIsMutable();
-        op_.add(builderForValue.build());
-        onChanged();
-      } else {
-        opBuilder_.addMessage(builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.ApiDef op = 1;</code>
-     */
-    public Builder addOp(
-        int index, org.tensorflow.proto.framework.ApiDef.Builder builderForValue) {
-      if (opBuilder_ == null) {
-        ensureOpIsMutable();
-        op_.add(index, builderForValue.build());
-        onChanged();
-      } else {
-        opBuilder_.addMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.ApiDef op = 1;</code>
-     */
-    public Builder addAllOp(
-        java.lang.Iterable<? extends org.tensorflow.proto.framework.ApiDef> values) {
-      if (opBuilder_ == null) {
-        ensureOpIsMutable();
-        com.google.protobuf.AbstractMessageLite.Builder.addAll(
-            values, op_);
-        onChanged();
-      } else {
-        opBuilder_.addAllMessages(values);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.ApiDef op = 1;</code>
-     */
-    public Builder clearOp() {
-      if (opBuilder_ == null) {
-        op_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000001);
-        onChanged();
-      } else {
-        opBuilder_.clear();
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.ApiDef op = 1;</code>
-     */
-    public Builder removeOp(int index) {
-      if (opBuilder_ == null) {
-        ensureOpIsMutable();
-        op_.remove(index);
-        onChanged();
-      } else {
-        opBuilder_.remove(index);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.ApiDef op = 1;</code>
-     */
-    public org.tensorflow.proto.framework.ApiDef.Builder getOpBuilder(
-        int index) {
-      return getOpFieldBuilder().getBuilder(index);
-    }
-    /**
-     * <code>repeated .tensorflow.ApiDef op = 1;</code>
-     */
-    public org.tensorflow.proto.framework.ApiDefOrBuilder getOpOrBuilder(
-        int index) {
-      if (opBuilder_ == null) {
-        return op_.get(index);  } else {
-        return opBuilder_.getMessageOrBuilder(index);
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.ApiDef op = 1;</code>
-     */
-    public java.util.List<? extends org.tensorflow.proto.framework.ApiDefOrBuilder> 
-         getOpOrBuilderList() {
-      if (opBuilder_ != null) {
-        return opBuilder_.getMessageOrBuilderList();
-      } else {
-        return java.util.Collections.unmodifiableList(op_);
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.ApiDef op = 1;</code>
-     */
-    public org.tensorflow.proto.framework.ApiDef.Builder addOpBuilder() {
-      return getOpFieldBuilder().addBuilder(
-          org.tensorflow.proto.framework.ApiDef.getDefaultInstance());
-    }
-    /**
-     * <code>repeated .tensorflow.ApiDef op = 1;</code>
-     */
-    public org.tensorflow.proto.framework.ApiDef.Builder addOpBuilder(
-        int index) {
-      return getOpFieldBuilder().addBuilder(
-          index, org.tensorflow.proto.framework.ApiDef.getDefaultInstance());
-    }
-    /**
-     * <code>repeated .tensorflow.ApiDef op = 1;</code>
-     */
-    public java.util.List<org.tensorflow.proto.framework.ApiDef.Builder> 
-         getOpBuilderList() {
-      return getOpFieldBuilder().getBuilderList();
-    }
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.framework.ApiDef, org.tensorflow.proto.framework.ApiDef.Builder, org.tensorflow.proto.framework.ApiDefOrBuilder> 
-        getOpFieldBuilder() {
-      if (opBuilder_ == null) {
-        opBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
-            org.tensorflow.proto.framework.ApiDef, org.tensorflow.proto.framework.ApiDef.Builder, org.tensorflow.proto.framework.ApiDefOrBuilder>(
-                op_,
-                ((bitField0_ & 0x00000001) != 0),
-                getParentForChildren(),
-                isClean());
-        op_ = null;
-      }
-      return opBuilder_;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.ApiDefs)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.ApiDefs)
-  private static final org.tensorflow.proto.framework.ApiDefs DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.ApiDefs();
-  }
-
-  public static org.tensorflow.proto.framework.ApiDefs getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<ApiDefs>
-      PARSER = new com.google.protobuf.AbstractParser<ApiDefs>() {
-    @java.lang.Override
-    public ApiDefs parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new ApiDefs(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<ApiDefs> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<ApiDefs> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.framework.ApiDefs getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ApiDefsOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ApiDefsOrBuilder.java
deleted file mode 100644
index e3ba64964e7..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ApiDefsOrBuilder.java
+++ /dev/null
@@ -1,33 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/framework/api_def.proto
-
-package org.tensorflow.proto.framework;
-
-public interface ApiDefsOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.ApiDefs)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <code>repeated .tensorflow.ApiDef op = 1;</code>
-   */
-  java.util.List<org.tensorflow.proto.framework.ApiDef> 
-      getOpList();
-  /**
-   * <code>repeated .tensorflow.ApiDef op = 1;</code>
-   */
-  org.tensorflow.proto.framework.ApiDef getOp(int index);
-  /**
-   * <code>repeated .tensorflow.ApiDef op = 1;</code>
-   */
-  int getOpCount();
-  /**
-   * <code>repeated .tensorflow.ApiDef op = 1;</code>
-   */
-  java.util.List<? extends org.tensorflow.proto.framework.ApiDefOrBuilder> 
-      getOpOrBuilderList();
-  /**
-   * <code>repeated .tensorflow.ApiDef op = 1;</code>
-   */
-  org.tensorflow.proto.framework.ApiDefOrBuilder getOpOrBuilder(
-      int index);
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/AssetFileDef.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/AssetFileDef.java
deleted file mode 100644
index 2d4a4e4b828..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/AssetFileDef.java
+++ /dev/null
@@ -1,827 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/meta_graph.proto
-
-package org.tensorflow.proto.framework;
-
-/**
- * <pre>
- * An asset file def for a single file or a set of sharded files with the same
- * name.
- * </pre>
- *
- * Protobuf type {@code tensorflow.AssetFileDef}
- */
-public  final class AssetFileDef extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.AssetFileDef)
-    AssetFileDefOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use AssetFileDef.newBuilder() to construct.
-  private AssetFileDef(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private AssetFileDef() {
-    filename_ = "";
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new AssetFileDef();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private AssetFileDef(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 10: {
-            org.tensorflow.proto.framework.TensorInfo.Builder subBuilder = null;
-            if (tensorInfo_ != null) {
-              subBuilder = tensorInfo_.toBuilder();
-            }
-            tensorInfo_ = input.readMessage(org.tensorflow.proto.framework.TensorInfo.parser(), extensionRegistry);
-            if (subBuilder != null) {
-              subBuilder.mergeFrom(tensorInfo_);
-              tensorInfo_ = subBuilder.buildPartial();
-            }
-
-            break;
-          }
-          case 18: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            filename_ = s;
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.framework.MetaGraphProtos.internal_static_tensorflow_AssetFileDef_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.framework.MetaGraphProtos.internal_static_tensorflow_AssetFileDef_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.framework.AssetFileDef.class, org.tensorflow.proto.framework.AssetFileDef.Builder.class);
-  }
-
-  public static final int TENSOR_INFO_FIELD_NUMBER = 1;
-  private org.tensorflow.proto.framework.TensorInfo tensorInfo_;
-  /**
-   * <pre>
-   * The tensor to bind the asset filename to.
-   * </pre>
-   *
-   * <code>.tensorflow.TensorInfo tensor_info = 1;</code>
-   */
-  public boolean hasTensorInfo() {
-    return tensorInfo_ != null;
-  }
-  /**
-   * <pre>
-   * The tensor to bind the asset filename to.
-   * </pre>
-   *
-   * <code>.tensorflow.TensorInfo tensor_info = 1;</code>
-   */
-  public org.tensorflow.proto.framework.TensorInfo getTensorInfo() {
-    return tensorInfo_ == null ? org.tensorflow.proto.framework.TensorInfo.getDefaultInstance() : tensorInfo_;
-  }
-  /**
-   * <pre>
-   * The tensor to bind the asset filename to.
-   * </pre>
-   *
-   * <code>.tensorflow.TensorInfo tensor_info = 1;</code>
-   */
-  public org.tensorflow.proto.framework.TensorInfoOrBuilder getTensorInfoOrBuilder() {
-    return getTensorInfo();
-  }
-
-  public static final int FILENAME_FIELD_NUMBER = 2;
-  private volatile java.lang.Object filename_;
-  /**
-   * <pre>
-   * The filename within an assets directory. Note: does not include the path
-   * prefix, i.e. directories. For an asset at /tmp/path/vocab.txt, the filename
-   * would be "vocab.txt".
-   * </pre>
-   *
-   * <code>string filename = 2;</code>
-   */
-  public java.lang.String getFilename() {
-    java.lang.Object ref = filename_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      filename_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * The filename within an assets directory. Note: does not include the path
-   * prefix, i.e. directories. For an asset at /tmp/path/vocab.txt, the filename
-   * would be "vocab.txt".
-   * </pre>
-   *
-   * <code>string filename = 2;</code>
-   */
-  public com.google.protobuf.ByteString
-      getFilenameBytes() {
-    java.lang.Object ref = filename_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      filename_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    if (tensorInfo_ != null) {
-      output.writeMessage(1, getTensorInfo());
-    }
-    if (!getFilenameBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 2, filename_);
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (tensorInfo_ != null) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(1, getTensorInfo());
-    }
-    if (!getFilenameBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(2, filename_);
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.framework.AssetFileDef)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.framework.AssetFileDef other = (org.tensorflow.proto.framework.AssetFileDef) obj;
-
-    if (hasTensorInfo() != other.hasTensorInfo()) return false;
-    if (hasTensorInfo()) {
-      if (!getTensorInfo()
-          .equals(other.getTensorInfo())) return false;
-    }
-    if (!getFilename()
-        .equals(other.getFilename())) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    if (hasTensorInfo()) {
-      hash = (37 * hash) + TENSOR_INFO_FIELD_NUMBER;
-      hash = (53 * hash) + getTensorInfo().hashCode();
-    }
-    hash = (37 * hash) + FILENAME_FIELD_NUMBER;
-    hash = (53 * hash) + getFilename().hashCode();
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.framework.AssetFileDef parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.AssetFileDef parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.AssetFileDef parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.AssetFileDef parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.AssetFileDef parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.AssetFileDef parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.AssetFileDef parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.AssetFileDef parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.AssetFileDef parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.AssetFileDef parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.AssetFileDef parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.AssetFileDef parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.framework.AssetFileDef prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * <pre>
-   * An asset file def for a single file or a set of sharded files with the same
-   * name.
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.AssetFileDef}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.AssetFileDef)
-      org.tensorflow.proto.framework.AssetFileDefOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.MetaGraphProtos.internal_static_tensorflow_AssetFileDef_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.MetaGraphProtos.internal_static_tensorflow_AssetFileDef_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.AssetFileDef.class, org.tensorflow.proto.framework.AssetFileDef.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.framework.AssetFileDef.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      if (tensorInfoBuilder_ == null) {
-        tensorInfo_ = null;
-      } else {
-        tensorInfo_ = null;
-        tensorInfoBuilder_ = null;
-      }
-      filename_ = "";
-
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.framework.MetaGraphProtos.internal_static_tensorflow_AssetFileDef_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.AssetFileDef getDefaultInstanceForType() {
-      return org.tensorflow.proto.framework.AssetFileDef.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.AssetFileDef build() {
-      org.tensorflow.proto.framework.AssetFileDef result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.AssetFileDef buildPartial() {
-      org.tensorflow.proto.framework.AssetFileDef result = new org.tensorflow.proto.framework.AssetFileDef(this);
-      if (tensorInfoBuilder_ == null) {
-        result.tensorInfo_ = tensorInfo_;
-      } else {
-        result.tensorInfo_ = tensorInfoBuilder_.build();
-      }
-      result.filename_ = filename_;
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.framework.AssetFileDef) {
-        return mergeFrom((org.tensorflow.proto.framework.AssetFileDef)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.framework.AssetFileDef other) {
-      if (other == org.tensorflow.proto.framework.AssetFileDef.getDefaultInstance()) return this;
-      if (other.hasTensorInfo()) {
-        mergeTensorInfo(other.getTensorInfo());
-      }
-      if (!other.getFilename().isEmpty()) {
-        filename_ = other.filename_;
-        onChanged();
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.framework.AssetFileDef parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.framework.AssetFileDef) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-
-    private org.tensorflow.proto.framework.TensorInfo tensorInfo_;
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.TensorInfo, org.tensorflow.proto.framework.TensorInfo.Builder, org.tensorflow.proto.framework.TensorInfoOrBuilder> tensorInfoBuilder_;
-    /**
-     * <pre>
-     * The tensor to bind the asset filename to.
-     * </pre>
-     *
-     * <code>.tensorflow.TensorInfo tensor_info = 1;</code>
-     */
-    public boolean hasTensorInfo() {
-      return tensorInfoBuilder_ != null || tensorInfo_ != null;
-    }
-    /**
-     * <pre>
-     * The tensor to bind the asset filename to.
-     * </pre>
-     *
-     * <code>.tensorflow.TensorInfo tensor_info = 1;</code>
-     */
-    public org.tensorflow.proto.framework.TensorInfo getTensorInfo() {
-      if (tensorInfoBuilder_ == null) {
-        return tensorInfo_ == null ? org.tensorflow.proto.framework.TensorInfo.getDefaultInstance() : tensorInfo_;
-      } else {
-        return tensorInfoBuilder_.getMessage();
-      }
-    }
-    /**
-     * <pre>
-     * The tensor to bind the asset filename to.
-     * </pre>
-     *
-     * <code>.tensorflow.TensorInfo tensor_info = 1;</code>
-     */
-    public Builder setTensorInfo(org.tensorflow.proto.framework.TensorInfo value) {
-      if (tensorInfoBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        tensorInfo_ = value;
-        onChanged();
-      } else {
-        tensorInfoBuilder_.setMessage(value);
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * The tensor to bind the asset filename to.
-     * </pre>
-     *
-     * <code>.tensorflow.TensorInfo tensor_info = 1;</code>
-     */
-    public Builder setTensorInfo(
-        org.tensorflow.proto.framework.TensorInfo.Builder builderForValue) {
-      if (tensorInfoBuilder_ == null) {
-        tensorInfo_ = builderForValue.build();
-        onChanged();
-      } else {
-        tensorInfoBuilder_.setMessage(builderForValue.build());
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * The tensor to bind the asset filename to.
-     * </pre>
-     *
-     * <code>.tensorflow.TensorInfo tensor_info = 1;</code>
-     */
-    public Builder mergeTensorInfo(org.tensorflow.proto.framework.TensorInfo value) {
-      if (tensorInfoBuilder_ == null) {
-        if (tensorInfo_ != null) {
-          tensorInfo_ =
-            org.tensorflow.proto.framework.TensorInfo.newBuilder(tensorInfo_).mergeFrom(value).buildPartial();
-        } else {
-          tensorInfo_ = value;
-        }
-        onChanged();
-      } else {
-        tensorInfoBuilder_.mergeFrom(value);
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * The tensor to bind the asset filename to.
-     * </pre>
-     *
-     * <code>.tensorflow.TensorInfo tensor_info = 1;</code>
-     */
-    public Builder clearTensorInfo() {
-      if (tensorInfoBuilder_ == null) {
-        tensorInfo_ = null;
-        onChanged();
-      } else {
-        tensorInfo_ = null;
-        tensorInfoBuilder_ = null;
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * The tensor to bind the asset filename to.
-     * </pre>
-     *
-     * <code>.tensorflow.TensorInfo tensor_info = 1;</code>
-     */
-    public org.tensorflow.proto.framework.TensorInfo.Builder getTensorInfoBuilder() {
-      
-      onChanged();
-      return getTensorInfoFieldBuilder().getBuilder();
-    }
-    /**
-     * <pre>
-     * The tensor to bind the asset filename to.
-     * </pre>
-     *
-     * <code>.tensorflow.TensorInfo tensor_info = 1;</code>
-     */
-    public org.tensorflow.proto.framework.TensorInfoOrBuilder getTensorInfoOrBuilder() {
-      if (tensorInfoBuilder_ != null) {
-        return tensorInfoBuilder_.getMessageOrBuilder();
-      } else {
-        return tensorInfo_ == null ?
-            org.tensorflow.proto.framework.TensorInfo.getDefaultInstance() : tensorInfo_;
-      }
-    }
-    /**
-     * <pre>
-     * The tensor to bind the asset filename to.
-     * </pre>
-     *
-     * <code>.tensorflow.TensorInfo tensor_info = 1;</code>
-     */
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.TensorInfo, org.tensorflow.proto.framework.TensorInfo.Builder, org.tensorflow.proto.framework.TensorInfoOrBuilder> 
-        getTensorInfoFieldBuilder() {
-      if (tensorInfoBuilder_ == null) {
-        tensorInfoBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-            org.tensorflow.proto.framework.TensorInfo, org.tensorflow.proto.framework.TensorInfo.Builder, org.tensorflow.proto.framework.TensorInfoOrBuilder>(
-                getTensorInfo(),
-                getParentForChildren(),
-                isClean());
-        tensorInfo_ = null;
-      }
-      return tensorInfoBuilder_;
-    }
-
-    private java.lang.Object filename_ = "";
-    /**
-     * <pre>
-     * The filename within an assets directory. Note: does not include the path
-     * prefix, i.e. directories. For an asset at /tmp/path/vocab.txt, the filename
-     * would be "vocab.txt".
-     * </pre>
-     *
-     * <code>string filename = 2;</code>
-     */
-    public java.lang.String getFilename() {
-      java.lang.Object ref = filename_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        filename_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * The filename within an assets directory. Note: does not include the path
-     * prefix, i.e. directories. For an asset at /tmp/path/vocab.txt, the filename
-     * would be "vocab.txt".
-     * </pre>
-     *
-     * <code>string filename = 2;</code>
-     */
-    public com.google.protobuf.ByteString
-        getFilenameBytes() {
-      java.lang.Object ref = filename_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        filename_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * The filename within an assets directory. Note: does not include the path
-     * prefix, i.e. directories. For an asset at /tmp/path/vocab.txt, the filename
-     * would be "vocab.txt".
-     * </pre>
-     *
-     * <code>string filename = 2;</code>
-     */
-    public Builder setFilename(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      filename_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * The filename within an assets directory. Note: does not include the path
-     * prefix, i.e. directories. For an asset at /tmp/path/vocab.txt, the filename
-     * would be "vocab.txt".
-     * </pre>
-     *
-     * <code>string filename = 2;</code>
-     */
-    public Builder clearFilename() {
-      
-      filename_ = getDefaultInstance().getFilename();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * The filename within an assets directory. Note: does not include the path
-     * prefix, i.e. directories. For an asset at /tmp/path/vocab.txt, the filename
-     * would be "vocab.txt".
-     * </pre>
-     *
-     * <code>string filename = 2;</code>
-     */
-    public Builder setFilenameBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      filename_ = value;
-      onChanged();
-      return this;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.AssetFileDef)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.AssetFileDef)
-  private static final org.tensorflow.proto.framework.AssetFileDef DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.AssetFileDef();
-  }
-
-  public static org.tensorflow.proto.framework.AssetFileDef getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<AssetFileDef>
-      PARSER = new com.google.protobuf.AbstractParser<AssetFileDef>() {
-    @java.lang.Override
-    public AssetFileDef parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new AssetFileDef(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<AssetFileDef> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<AssetFileDef> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.framework.AssetFileDef getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/AssetFileDefOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/AssetFileDefOrBuilder.java
deleted file mode 100644
index dd8ec09799f..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/AssetFileDefOrBuilder.java
+++ /dev/null
@@ -1,56 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/meta_graph.proto
-
-package org.tensorflow.proto.framework;
-
-public interface AssetFileDefOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.AssetFileDef)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <pre>
-   * The tensor to bind the asset filename to.
-   * </pre>
-   *
-   * <code>.tensorflow.TensorInfo tensor_info = 1;</code>
-   */
-  boolean hasTensorInfo();
-  /**
-   * <pre>
-   * The tensor to bind the asset filename to.
-   * </pre>
-   *
-   * <code>.tensorflow.TensorInfo tensor_info = 1;</code>
-   */
-  org.tensorflow.proto.framework.TensorInfo getTensorInfo();
-  /**
-   * <pre>
-   * The tensor to bind the asset filename to.
-   * </pre>
-   *
-   * <code>.tensorflow.TensorInfo tensor_info = 1;</code>
-   */
-  org.tensorflow.proto.framework.TensorInfoOrBuilder getTensorInfoOrBuilder();
-
-  /**
-   * <pre>
-   * The filename within an assets directory. Note: does not include the path
-   * prefix, i.e. directories. For an asset at /tmp/path/vocab.txt, the filename
-   * would be "vocab.txt".
-   * </pre>
-   *
-   * <code>string filename = 2;</code>
-   */
-  java.lang.String getFilename();
-  /**
-   * <pre>
-   * The filename within an assets directory. Note: does not include the path
-   * prefix, i.e. directories. For an asset at /tmp/path/vocab.txt, the filename
-   * would be "vocab.txt".
-   * </pre>
-   *
-   * <code>string filename = 2;</code>
-   */
-  com.google.protobuf.ByteString
-      getFilenameBytes();
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/AttrValue.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/AttrValue.java
deleted file mode 100644
index f64f1a8b3ab..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/AttrValue.java
+++ /dev/null
@@ -1,5341 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/framework/attr_value.proto
-
-package org.tensorflow.proto.framework;
-
-/**
- * <pre>
- * Protocol buffer representing the value for an attr used to configure an Op.
- * Comment indicates the corresponding attr type.  Only the field matching the
- * attr type may be filled.
- * </pre>
- *
- * Protobuf type {@code tensorflow.AttrValue}
- */
-public  final class AttrValue extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.AttrValue)
-    AttrValueOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use AttrValue.newBuilder() to construct.
-  private AttrValue(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private AttrValue() {
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new AttrValue();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private AttrValue(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 10: {
-            org.tensorflow.proto.framework.AttrValue.ListValue.Builder subBuilder = null;
-            if (valueCase_ == 1) {
-              subBuilder = ((org.tensorflow.proto.framework.AttrValue.ListValue) value_).toBuilder();
-            }
-            value_ =
-                input.readMessage(org.tensorflow.proto.framework.AttrValue.ListValue.parser(), extensionRegistry);
-            if (subBuilder != null) {
-              subBuilder.mergeFrom((org.tensorflow.proto.framework.AttrValue.ListValue) value_);
-              value_ = subBuilder.buildPartial();
-            }
-            valueCase_ = 1;
-            break;
-          }
-          case 18: {
-            valueCase_ = 2;
-            value_ = input.readBytes();
-            break;
-          }
-          case 24: {
-            valueCase_ = 3;
-            value_ = input.readInt64();
-            break;
-          }
-          case 37: {
-            valueCase_ = 4;
-            value_ = input.readFloat();
-            break;
-          }
-          case 40: {
-            valueCase_ = 5;
-            value_ = input.readBool();
-            break;
-          }
-          case 48: {
-            int rawValue = input.readEnum();
-            valueCase_ = 6;
-            value_ = rawValue;
-            break;
-          }
-          case 58: {
-            org.tensorflow.proto.framework.TensorShapeProto.Builder subBuilder = null;
-            if (valueCase_ == 7) {
-              subBuilder = ((org.tensorflow.proto.framework.TensorShapeProto) value_).toBuilder();
-            }
-            value_ =
-                input.readMessage(org.tensorflow.proto.framework.TensorShapeProto.parser(), extensionRegistry);
-            if (subBuilder != null) {
-              subBuilder.mergeFrom((org.tensorflow.proto.framework.TensorShapeProto) value_);
-              value_ = subBuilder.buildPartial();
-            }
-            valueCase_ = 7;
-            break;
-          }
-          case 66: {
-            org.tensorflow.proto.framework.TensorProto.Builder subBuilder = null;
-            if (valueCase_ == 8) {
-              subBuilder = ((org.tensorflow.proto.framework.TensorProto) value_).toBuilder();
-            }
-            value_ =
-                input.readMessage(org.tensorflow.proto.framework.TensorProto.parser(), extensionRegistry);
-            if (subBuilder != null) {
-              subBuilder.mergeFrom((org.tensorflow.proto.framework.TensorProto) value_);
-              value_ = subBuilder.buildPartial();
-            }
-            valueCase_ = 8;
-            break;
-          }
-          case 74: {
-            java.lang.String s = input.readStringRequireUtf8();
-            valueCase_ = 9;
-            value_ = s;
-            break;
-          }
-          case 82: {
-            org.tensorflow.proto.framework.NameAttrList.Builder subBuilder = null;
-            if (valueCase_ == 10) {
-              subBuilder = ((org.tensorflow.proto.framework.NameAttrList) value_).toBuilder();
-            }
-            value_ =
-                input.readMessage(org.tensorflow.proto.framework.NameAttrList.parser(), extensionRegistry);
-            if (subBuilder != null) {
-              subBuilder.mergeFrom((org.tensorflow.proto.framework.NameAttrList) value_);
-              value_ = subBuilder.buildPartial();
-            }
-            valueCase_ = 10;
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.framework.AttrValueProtos.internal_static_tensorflow_AttrValue_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.framework.AttrValueProtos.internal_static_tensorflow_AttrValue_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.framework.AttrValue.class, org.tensorflow.proto.framework.AttrValue.Builder.class);
-  }
-
-  public interface ListValueOrBuilder extends
-      // @@protoc_insertion_point(interface_extends:tensorflow.AttrValue.ListValue)
-      com.google.protobuf.MessageOrBuilder {
-
-    /**
-     * <pre>
-     * "list(string)"
-     * </pre>
-     *
-     * <code>repeated bytes s = 2;</code>
-     */
-    java.util.List<com.google.protobuf.ByteString> getSList();
-    /**
-     * <pre>
-     * "list(string)"
-     * </pre>
-     *
-     * <code>repeated bytes s = 2;</code>
-     */
-    int getSCount();
-    /**
-     * <pre>
-     * "list(string)"
-     * </pre>
-     *
-     * <code>repeated bytes s = 2;</code>
-     */
-    com.google.protobuf.ByteString getS(int index);
-
-    /**
-     * <pre>
-     * "list(int)"
-     * </pre>
-     *
-     * <code>repeated int64 i = 3 [packed = true];</code>
-     */
-    java.util.List<java.lang.Long> getIList();
-    /**
-     * <pre>
-     * "list(int)"
-     * </pre>
-     *
-     * <code>repeated int64 i = 3 [packed = true];</code>
-     */
-    int getICount();
-    /**
-     * <pre>
-     * "list(int)"
-     * </pre>
-     *
-     * <code>repeated int64 i = 3 [packed = true];</code>
-     */
-    long getI(int index);
-
-    /**
-     * <pre>
-     * "list(float)"
-     * </pre>
-     *
-     * <code>repeated float f = 4 [packed = true];</code>
-     */
-    java.util.List<java.lang.Float> getFList();
-    /**
-     * <pre>
-     * "list(float)"
-     * </pre>
-     *
-     * <code>repeated float f = 4 [packed = true];</code>
-     */
-    int getFCount();
-    /**
-     * <pre>
-     * "list(float)"
-     * </pre>
-     *
-     * <code>repeated float f = 4 [packed = true];</code>
-     */
-    float getF(int index);
-
-    /**
-     * <pre>
-     * "list(bool)"
-     * </pre>
-     *
-     * <code>repeated bool b = 5 [packed = true];</code>
-     */
-    java.util.List<java.lang.Boolean> getBList();
-    /**
-     * <pre>
-     * "list(bool)"
-     * </pre>
-     *
-     * <code>repeated bool b = 5 [packed = true];</code>
-     */
-    int getBCount();
-    /**
-     * <pre>
-     * "list(bool)"
-     * </pre>
-     *
-     * <code>repeated bool b = 5 [packed = true];</code>
-     */
-    boolean getB(int index);
-
-    /**
-     * <pre>
-     * "list(type)"
-     * </pre>
-     *
-     * <code>repeated .tensorflow.DataType type = 6 [packed = true];</code>
-     */
-    java.util.List<org.tensorflow.proto.framework.DataType> getTypeList();
-    /**
-     * <pre>
-     * "list(type)"
-     * </pre>
-     *
-     * <code>repeated .tensorflow.DataType type = 6 [packed = true];</code>
-     */
-    int getTypeCount();
-    /**
-     * <pre>
-     * "list(type)"
-     * </pre>
-     *
-     * <code>repeated .tensorflow.DataType type = 6 [packed = true];</code>
-     */
-    org.tensorflow.proto.framework.DataType getType(int index);
-    /**
-     * <pre>
-     * "list(type)"
-     * </pre>
-     *
-     * <code>repeated .tensorflow.DataType type = 6 [packed = true];</code>
-     */
-    java.util.List<java.lang.Integer>
-    getTypeValueList();
-    /**
-     * <pre>
-     * "list(type)"
-     * </pre>
-     *
-     * <code>repeated .tensorflow.DataType type = 6 [packed = true];</code>
-     */
-    int getTypeValue(int index);
-
-    /**
-     * <pre>
-     * "list(shape)"
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorShapeProto shape = 7;</code>
-     */
-    java.util.List<org.tensorflow.proto.framework.TensorShapeProto> 
-        getShapeList();
-    /**
-     * <pre>
-     * "list(shape)"
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorShapeProto shape = 7;</code>
-     */
-    org.tensorflow.proto.framework.TensorShapeProto getShape(int index);
-    /**
-     * <pre>
-     * "list(shape)"
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorShapeProto shape = 7;</code>
-     */
-    int getShapeCount();
-    /**
-     * <pre>
-     * "list(shape)"
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorShapeProto shape = 7;</code>
-     */
-    java.util.List<? extends org.tensorflow.proto.framework.TensorShapeProtoOrBuilder> 
-        getShapeOrBuilderList();
-    /**
-     * <pre>
-     * "list(shape)"
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorShapeProto shape = 7;</code>
-     */
-    org.tensorflow.proto.framework.TensorShapeProtoOrBuilder getShapeOrBuilder(
-        int index);
-
-    /**
-     * <pre>
-     * "list(tensor)"
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorProto tensor = 8;</code>
-     */
-    java.util.List<org.tensorflow.proto.framework.TensorProto> 
-        getTensorList();
-    /**
-     * <pre>
-     * "list(tensor)"
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorProto tensor = 8;</code>
-     */
-    org.tensorflow.proto.framework.TensorProto getTensor(int index);
-    /**
-     * <pre>
-     * "list(tensor)"
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorProto tensor = 8;</code>
-     */
-    int getTensorCount();
-    /**
-     * <pre>
-     * "list(tensor)"
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorProto tensor = 8;</code>
-     */
-    java.util.List<? extends org.tensorflow.proto.framework.TensorProtoOrBuilder> 
-        getTensorOrBuilderList();
-    /**
-     * <pre>
-     * "list(tensor)"
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorProto tensor = 8;</code>
-     */
-    org.tensorflow.proto.framework.TensorProtoOrBuilder getTensorOrBuilder(
-        int index);
-
-    /**
-     * <pre>
-     * "list(attr)"
-     * </pre>
-     *
-     * <code>repeated .tensorflow.NameAttrList func = 9;</code>
-     */
-    java.util.List<org.tensorflow.proto.framework.NameAttrList> 
-        getFuncList();
-    /**
-     * <pre>
-     * "list(attr)"
-     * </pre>
-     *
-     * <code>repeated .tensorflow.NameAttrList func = 9;</code>
-     */
-    org.tensorflow.proto.framework.NameAttrList getFunc(int index);
-    /**
-     * <pre>
-     * "list(attr)"
-     * </pre>
-     *
-     * <code>repeated .tensorflow.NameAttrList func = 9;</code>
-     */
-    int getFuncCount();
-    /**
-     * <pre>
-     * "list(attr)"
-     * </pre>
-     *
-     * <code>repeated .tensorflow.NameAttrList func = 9;</code>
-     */
-    java.util.List<? extends org.tensorflow.proto.framework.NameAttrListOrBuilder> 
-        getFuncOrBuilderList();
-    /**
-     * <pre>
-     * "list(attr)"
-     * </pre>
-     *
-     * <code>repeated .tensorflow.NameAttrList func = 9;</code>
-     */
-    org.tensorflow.proto.framework.NameAttrListOrBuilder getFuncOrBuilder(
-        int index);
-  }
-  /**
-   * <pre>
-   * LINT.IfChange
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.AttrValue.ListValue}
-   */
-  public  static final class ListValue extends
-      com.google.protobuf.GeneratedMessageV3 implements
-      // @@protoc_insertion_point(message_implements:tensorflow.AttrValue.ListValue)
-      ListValueOrBuilder {
-  private static final long serialVersionUID = 0L;
-    // Use ListValue.newBuilder() to construct.
-    private ListValue(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-      super(builder);
-    }
-    private ListValue() {
-      s_ = java.util.Collections.emptyList();
-      i_ = emptyLongList();
-      f_ = emptyFloatList();
-      b_ = emptyBooleanList();
-      type_ = java.util.Collections.emptyList();
-      shape_ = java.util.Collections.emptyList();
-      tensor_ = java.util.Collections.emptyList();
-      func_ = java.util.Collections.emptyList();
-    }
-
-    @java.lang.Override
-    @SuppressWarnings({"unused"})
-    protected java.lang.Object newInstance(
-        UnusedPrivateParameter unused) {
-      return new ListValue();
-    }
-
-    @java.lang.Override
-    public final com.google.protobuf.UnknownFieldSet
-    getUnknownFields() {
-      return this.unknownFields;
-    }
-    private ListValue(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      this();
-      if (extensionRegistry == null) {
-        throw new java.lang.NullPointerException();
-      }
-      int mutable_bitField0_ = 0;
-      com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-          com.google.protobuf.UnknownFieldSet.newBuilder();
-      try {
-        boolean done = false;
-        while (!done) {
-          int tag = input.readTag();
-          switch (tag) {
-            case 0:
-              done = true;
-              break;
-            case 18: {
-              if (!((mutable_bitField0_ & 0x00000001) != 0)) {
-                s_ = new java.util.ArrayList<com.google.protobuf.ByteString>();
-                mutable_bitField0_ |= 0x00000001;
-              }
-              s_.add(input.readBytes());
-              break;
-            }
-            case 24: {
-              if (!((mutable_bitField0_ & 0x00000002) != 0)) {
-                i_ = newLongList();
-                mutable_bitField0_ |= 0x00000002;
-              }
-              i_.addLong(input.readInt64());
-              break;
-            }
-            case 26: {
-              int length = input.readRawVarint32();
-              int limit = input.pushLimit(length);
-              if (!((mutable_bitField0_ & 0x00000002) != 0) && input.getBytesUntilLimit() > 0) {
-                i_ = newLongList();
-                mutable_bitField0_ |= 0x00000002;
-              }
-              while (input.getBytesUntilLimit() > 0) {
-                i_.addLong(input.readInt64());
-              }
-              input.popLimit(limit);
-              break;
-            }
-            case 37: {
-              if (!((mutable_bitField0_ & 0x00000004) != 0)) {
-                f_ = newFloatList();
-                mutable_bitField0_ |= 0x00000004;
-              }
-              f_.addFloat(input.readFloat());
-              break;
-            }
-            case 34: {
-              int length = input.readRawVarint32();
-              int limit = input.pushLimit(length);
-              if (!((mutable_bitField0_ & 0x00000004) != 0) && input.getBytesUntilLimit() > 0) {
-                f_ = newFloatList();
-                mutable_bitField0_ |= 0x00000004;
-              }
-              while (input.getBytesUntilLimit() > 0) {
-                f_.addFloat(input.readFloat());
-              }
-              input.popLimit(limit);
-              break;
-            }
-            case 40: {
-              if (!((mutable_bitField0_ & 0x00000008) != 0)) {
-                b_ = newBooleanList();
-                mutable_bitField0_ |= 0x00000008;
-              }
-              b_.addBoolean(input.readBool());
-              break;
-            }
-            case 42: {
-              int length = input.readRawVarint32();
-              int limit = input.pushLimit(length);
-              if (!((mutable_bitField0_ & 0x00000008) != 0) && input.getBytesUntilLimit() > 0) {
-                b_ = newBooleanList();
-                mutable_bitField0_ |= 0x00000008;
-              }
-              while (input.getBytesUntilLimit() > 0) {
-                b_.addBoolean(input.readBool());
-              }
-              input.popLimit(limit);
-              break;
-            }
-            case 48: {
-              int rawValue = input.readEnum();
-              if (!((mutable_bitField0_ & 0x00000010) != 0)) {
-                type_ = new java.util.ArrayList<java.lang.Integer>();
-                mutable_bitField0_ |= 0x00000010;
-              }
-              type_.add(rawValue);
-              break;
-            }
-            case 50: {
-              int length = input.readRawVarint32();
-              int oldLimit = input.pushLimit(length);
-              while(input.getBytesUntilLimit() > 0) {
-                int rawValue = input.readEnum();
-                if (!((mutable_bitField0_ & 0x00000010) != 0)) {
-                  type_ = new java.util.ArrayList<java.lang.Integer>();
-                  mutable_bitField0_ |= 0x00000010;
-                }
-                type_.add(rawValue);
-              }
-              input.popLimit(oldLimit);
-              break;
-            }
-            case 58: {
-              if (!((mutable_bitField0_ & 0x00000020) != 0)) {
-                shape_ = new java.util.ArrayList<org.tensorflow.proto.framework.TensorShapeProto>();
-                mutable_bitField0_ |= 0x00000020;
-              }
-              shape_.add(
-                  input.readMessage(org.tensorflow.proto.framework.TensorShapeProto.parser(), extensionRegistry));
-              break;
-            }
-            case 66: {
-              if (!((mutable_bitField0_ & 0x00000040) != 0)) {
-                tensor_ = new java.util.ArrayList<org.tensorflow.proto.framework.TensorProto>();
-                mutable_bitField0_ |= 0x00000040;
-              }
-              tensor_.add(
-                  input.readMessage(org.tensorflow.proto.framework.TensorProto.parser(), extensionRegistry));
-              break;
-            }
-            case 74: {
-              if (!((mutable_bitField0_ & 0x00000080) != 0)) {
-                func_ = new java.util.ArrayList<org.tensorflow.proto.framework.NameAttrList>();
-                mutable_bitField0_ |= 0x00000080;
-              }
-              func_.add(
-                  input.readMessage(org.tensorflow.proto.framework.NameAttrList.parser(), extensionRegistry));
-              break;
-            }
-            default: {
-              if (!parseUnknownField(
-                  input, unknownFields, extensionRegistry, tag)) {
-                done = true;
-              }
-              break;
-            }
-          }
-        }
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        throw e.setUnfinishedMessage(this);
-      } catch (java.io.IOException e) {
-        throw new com.google.protobuf.InvalidProtocolBufferException(
-            e).setUnfinishedMessage(this);
-      } finally {
-        if (((mutable_bitField0_ & 0x00000001) != 0)) {
-          s_ = java.util.Collections.unmodifiableList(s_); // C
-        }
-        if (((mutable_bitField0_ & 0x00000002) != 0)) {
-          i_.makeImmutable(); // C
-        }
-        if (((mutable_bitField0_ & 0x00000004) != 0)) {
-          f_.makeImmutable(); // C
-        }
-        if (((mutable_bitField0_ & 0x00000008) != 0)) {
-          b_.makeImmutable(); // C
-        }
-        if (((mutable_bitField0_ & 0x00000010) != 0)) {
-          type_ = java.util.Collections.unmodifiableList(type_);
-        }
-        if (((mutable_bitField0_ & 0x00000020) != 0)) {
-          shape_ = java.util.Collections.unmodifiableList(shape_);
-        }
-        if (((mutable_bitField0_ & 0x00000040) != 0)) {
-          tensor_ = java.util.Collections.unmodifiableList(tensor_);
-        }
-        if (((mutable_bitField0_ & 0x00000080) != 0)) {
-          func_ = java.util.Collections.unmodifiableList(func_);
-        }
-        this.unknownFields = unknownFields.build();
-        makeExtensionsImmutable();
-      }
-    }
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.AttrValueProtos.internal_static_tensorflow_AttrValue_ListValue_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.AttrValueProtos.internal_static_tensorflow_AttrValue_ListValue_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.AttrValue.ListValue.class, org.tensorflow.proto.framework.AttrValue.ListValue.Builder.class);
-    }
-
-    public static final int S_FIELD_NUMBER = 2;
-    private java.util.List<com.google.protobuf.ByteString> s_;
-    /**
-     * <pre>
-     * "list(string)"
-     * </pre>
-     *
-     * <code>repeated bytes s = 2;</code>
-     */
-    public java.util.List<com.google.protobuf.ByteString>
-        getSList() {
-      return s_;
-    }
-    /**
-     * <pre>
-     * "list(string)"
-     * </pre>
-     *
-     * <code>repeated bytes s = 2;</code>
-     */
-    public int getSCount() {
-      return s_.size();
-    }
-    /**
-     * <pre>
-     * "list(string)"
-     * </pre>
-     *
-     * <code>repeated bytes s = 2;</code>
-     */
-    public com.google.protobuf.ByteString getS(int index) {
-      return s_.get(index);
-    }
-
-    public static final int I_FIELD_NUMBER = 3;
-    private com.google.protobuf.Internal.LongList i_;
-    /**
-     * <pre>
-     * "list(int)"
-     * </pre>
-     *
-     * <code>repeated int64 i = 3 [packed = true];</code>
-     */
-    public java.util.List<java.lang.Long>
-        getIList() {
-      return i_;
-    }
-    /**
-     * <pre>
-     * "list(int)"
-     * </pre>
-     *
-     * <code>repeated int64 i = 3 [packed = true];</code>
-     */
-    public int getICount() {
-      return i_.size();
-    }
-    /**
-     * <pre>
-     * "list(int)"
-     * </pre>
-     *
-     * <code>repeated int64 i = 3 [packed = true];</code>
-     */
-    public long getI(int index) {
-      return i_.getLong(index);
-    }
-    private int iMemoizedSerializedSize = -1;
-
-    public static final int F_FIELD_NUMBER = 4;
-    private com.google.protobuf.Internal.FloatList f_;
-    /**
-     * <pre>
-     * "list(float)"
-     * </pre>
-     *
-     * <code>repeated float f = 4 [packed = true];</code>
-     */
-    public java.util.List<java.lang.Float>
-        getFList() {
-      return f_;
-    }
-    /**
-     * <pre>
-     * "list(float)"
-     * </pre>
-     *
-     * <code>repeated float f = 4 [packed = true];</code>
-     */
-    public int getFCount() {
-      return f_.size();
-    }
-    /**
-     * <pre>
-     * "list(float)"
-     * </pre>
-     *
-     * <code>repeated float f = 4 [packed = true];</code>
-     */
-    public float getF(int index) {
-      return f_.getFloat(index);
-    }
-    private int fMemoizedSerializedSize = -1;
-
-    public static final int B_FIELD_NUMBER = 5;
-    private com.google.protobuf.Internal.BooleanList b_;
-    /**
-     * <pre>
-     * "list(bool)"
-     * </pre>
-     *
-     * <code>repeated bool b = 5 [packed = true];</code>
-     */
-    public java.util.List<java.lang.Boolean>
-        getBList() {
-      return b_;
-    }
-    /**
-     * <pre>
-     * "list(bool)"
-     * </pre>
-     *
-     * <code>repeated bool b = 5 [packed = true];</code>
-     */
-    public int getBCount() {
-      return b_.size();
-    }
-    /**
-     * <pre>
-     * "list(bool)"
-     * </pre>
-     *
-     * <code>repeated bool b = 5 [packed = true];</code>
-     */
-    public boolean getB(int index) {
-      return b_.getBoolean(index);
-    }
-    private int bMemoizedSerializedSize = -1;
-
-    public static final int TYPE_FIELD_NUMBER = 6;
-    private java.util.List<java.lang.Integer> type_;
-    private static final com.google.protobuf.Internal.ListAdapter.Converter<
-        java.lang.Integer, org.tensorflow.proto.framework.DataType> type_converter_ =
-            new com.google.protobuf.Internal.ListAdapter.Converter<
-                java.lang.Integer, org.tensorflow.proto.framework.DataType>() {
-              public org.tensorflow.proto.framework.DataType convert(java.lang.Integer from) {
-                @SuppressWarnings("deprecation")
-                org.tensorflow.proto.framework.DataType result = org.tensorflow.proto.framework.DataType.valueOf(from);
-                return result == null ? org.tensorflow.proto.framework.DataType.UNRECOGNIZED : result;
-              }
-            };
-    /**
-     * <pre>
-     * "list(type)"
-     * </pre>
-     *
-     * <code>repeated .tensorflow.DataType type = 6 [packed = true];</code>
-     */
-    public java.util.List<org.tensorflow.proto.framework.DataType> getTypeList() {
-      return new com.google.protobuf.Internal.ListAdapter<
-          java.lang.Integer, org.tensorflow.proto.framework.DataType>(type_, type_converter_);
-    }
-    /**
-     * <pre>
-     * "list(type)"
-     * </pre>
-     *
-     * <code>repeated .tensorflow.DataType type = 6 [packed = true];</code>
-     */
-    public int getTypeCount() {
-      return type_.size();
-    }
-    /**
-     * <pre>
-     * "list(type)"
-     * </pre>
-     *
-     * <code>repeated .tensorflow.DataType type = 6 [packed = true];</code>
-     */
-    public org.tensorflow.proto.framework.DataType getType(int index) {
-      return type_converter_.convert(type_.get(index));
-    }
-    /**
-     * <pre>
-     * "list(type)"
-     * </pre>
-     *
-     * <code>repeated .tensorflow.DataType type = 6 [packed = true];</code>
-     */
-    public java.util.List<java.lang.Integer>
-    getTypeValueList() {
-      return type_;
-    }
-    /**
-     * <pre>
-     * "list(type)"
-     * </pre>
-     *
-     * <code>repeated .tensorflow.DataType type = 6 [packed = true];</code>
-     */
-    public int getTypeValue(int index) {
-      return type_.get(index);
-    }
-    private int typeMemoizedSerializedSize;
-
-    public static final int SHAPE_FIELD_NUMBER = 7;
-    private java.util.List<org.tensorflow.proto.framework.TensorShapeProto> shape_;
-    /**
-     * <pre>
-     * "list(shape)"
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorShapeProto shape = 7;</code>
-     */
-    public java.util.List<org.tensorflow.proto.framework.TensorShapeProto> getShapeList() {
-      return shape_;
-    }
-    /**
-     * <pre>
-     * "list(shape)"
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorShapeProto shape = 7;</code>
-     */
-    public java.util.List<? extends org.tensorflow.proto.framework.TensorShapeProtoOrBuilder> 
-        getShapeOrBuilderList() {
-      return shape_;
-    }
-    /**
-     * <pre>
-     * "list(shape)"
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorShapeProto shape = 7;</code>
-     */
-    public int getShapeCount() {
-      return shape_.size();
-    }
-    /**
-     * <pre>
-     * "list(shape)"
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorShapeProto shape = 7;</code>
-     */
-    public org.tensorflow.proto.framework.TensorShapeProto getShape(int index) {
-      return shape_.get(index);
-    }
-    /**
-     * <pre>
-     * "list(shape)"
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorShapeProto shape = 7;</code>
-     */
-    public org.tensorflow.proto.framework.TensorShapeProtoOrBuilder getShapeOrBuilder(
-        int index) {
-      return shape_.get(index);
-    }
-
-    public static final int TENSOR_FIELD_NUMBER = 8;
-    private java.util.List<org.tensorflow.proto.framework.TensorProto> tensor_;
-    /**
-     * <pre>
-     * "list(tensor)"
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorProto tensor = 8;</code>
-     */
-    public java.util.List<org.tensorflow.proto.framework.TensorProto> getTensorList() {
-      return tensor_;
-    }
-    /**
-     * <pre>
-     * "list(tensor)"
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorProto tensor = 8;</code>
-     */
-    public java.util.List<? extends org.tensorflow.proto.framework.TensorProtoOrBuilder> 
-        getTensorOrBuilderList() {
-      return tensor_;
-    }
-    /**
-     * <pre>
-     * "list(tensor)"
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorProto tensor = 8;</code>
-     */
-    public int getTensorCount() {
-      return tensor_.size();
-    }
-    /**
-     * <pre>
-     * "list(tensor)"
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorProto tensor = 8;</code>
-     */
-    public org.tensorflow.proto.framework.TensorProto getTensor(int index) {
-      return tensor_.get(index);
-    }
-    /**
-     * <pre>
-     * "list(tensor)"
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorProto tensor = 8;</code>
-     */
-    public org.tensorflow.proto.framework.TensorProtoOrBuilder getTensorOrBuilder(
-        int index) {
-      return tensor_.get(index);
-    }
-
-    public static final int FUNC_FIELD_NUMBER = 9;
-    private java.util.List<org.tensorflow.proto.framework.NameAttrList> func_;
-    /**
-     * <pre>
-     * "list(attr)"
-     * </pre>
-     *
-     * <code>repeated .tensorflow.NameAttrList func = 9;</code>
-     */
-    public java.util.List<org.tensorflow.proto.framework.NameAttrList> getFuncList() {
-      return func_;
-    }
-    /**
-     * <pre>
-     * "list(attr)"
-     * </pre>
-     *
-     * <code>repeated .tensorflow.NameAttrList func = 9;</code>
-     */
-    public java.util.List<? extends org.tensorflow.proto.framework.NameAttrListOrBuilder> 
-        getFuncOrBuilderList() {
-      return func_;
-    }
-    /**
-     * <pre>
-     * "list(attr)"
-     * </pre>
-     *
-     * <code>repeated .tensorflow.NameAttrList func = 9;</code>
-     */
-    public int getFuncCount() {
-      return func_.size();
-    }
-    /**
-     * <pre>
-     * "list(attr)"
-     * </pre>
-     *
-     * <code>repeated .tensorflow.NameAttrList func = 9;</code>
-     */
-    public org.tensorflow.proto.framework.NameAttrList getFunc(int index) {
-      return func_.get(index);
-    }
-    /**
-     * <pre>
-     * "list(attr)"
-     * </pre>
-     *
-     * <code>repeated .tensorflow.NameAttrList func = 9;</code>
-     */
-    public org.tensorflow.proto.framework.NameAttrListOrBuilder getFuncOrBuilder(
-        int index) {
-      return func_.get(index);
-    }
-
-    private byte memoizedIsInitialized = -1;
-    @java.lang.Override
-    public final boolean isInitialized() {
-      byte isInitialized = memoizedIsInitialized;
-      if (isInitialized == 1) return true;
-      if (isInitialized == 0) return false;
-
-      memoizedIsInitialized = 1;
-      return true;
-    }
-
-    @java.lang.Override
-    public void writeTo(com.google.protobuf.CodedOutputStream output)
-                        throws java.io.IOException {
-      getSerializedSize();
-      for (int i = 0; i < s_.size(); i++) {
-        output.writeBytes(2, s_.get(i));
-      }
-      if (getIList().size() > 0) {
-        output.writeUInt32NoTag(26);
-        output.writeUInt32NoTag(iMemoizedSerializedSize);
-      }
-      for (int i = 0; i < i_.size(); i++) {
-        output.writeInt64NoTag(i_.getLong(i));
-      }
-      if (getFList().size() > 0) {
-        output.writeUInt32NoTag(34);
-        output.writeUInt32NoTag(fMemoizedSerializedSize);
-      }
-      for (int i = 0; i < f_.size(); i++) {
-        output.writeFloatNoTag(f_.getFloat(i));
-      }
-      if (getBList().size() > 0) {
-        output.writeUInt32NoTag(42);
-        output.writeUInt32NoTag(bMemoizedSerializedSize);
-      }
-      for (int i = 0; i < b_.size(); i++) {
-        output.writeBoolNoTag(b_.getBoolean(i));
-      }
-      if (getTypeList().size() > 0) {
-        output.writeUInt32NoTag(50);
-        output.writeUInt32NoTag(typeMemoizedSerializedSize);
-      }
-      for (int i = 0; i < type_.size(); i++) {
-        output.writeEnumNoTag(type_.get(i));
-      }
-      for (int i = 0; i < shape_.size(); i++) {
-        output.writeMessage(7, shape_.get(i));
-      }
-      for (int i = 0; i < tensor_.size(); i++) {
-        output.writeMessage(8, tensor_.get(i));
-      }
-      for (int i = 0; i < func_.size(); i++) {
-        output.writeMessage(9, func_.get(i));
-      }
-      unknownFields.writeTo(output);
-    }
-
-    @java.lang.Override
-    public int getSerializedSize() {
-      int size = memoizedSize;
-      if (size != -1) return size;
-
-      size = 0;
-      {
-        int dataSize = 0;
-        for (int i = 0; i < s_.size(); i++) {
-          dataSize += com.google.protobuf.CodedOutputStream
-            .computeBytesSizeNoTag(s_.get(i));
-        }
-        size += dataSize;
-        size += 1 * getSList().size();
-      }
-      {
-        int dataSize = 0;
-        for (int i = 0; i < i_.size(); i++) {
-          dataSize += com.google.protobuf.CodedOutputStream
-            .computeInt64SizeNoTag(i_.getLong(i));
-        }
-        size += dataSize;
-        if (!getIList().isEmpty()) {
-          size += 1;
-          size += com.google.protobuf.CodedOutputStream
-              .computeInt32SizeNoTag(dataSize);
-        }
-        iMemoizedSerializedSize = dataSize;
-      }
-      {
-        int dataSize = 0;
-        dataSize = 4 * getFList().size();
-        size += dataSize;
-        if (!getFList().isEmpty()) {
-          size += 1;
-          size += com.google.protobuf.CodedOutputStream
-              .computeInt32SizeNoTag(dataSize);
-        }
-        fMemoizedSerializedSize = dataSize;
-      }
-      {
-        int dataSize = 0;
-        dataSize = 1 * getBList().size();
-        size += dataSize;
-        if (!getBList().isEmpty()) {
-          size += 1;
-          size += com.google.protobuf.CodedOutputStream
-              .computeInt32SizeNoTag(dataSize);
-        }
-        bMemoizedSerializedSize = dataSize;
-      }
-      {
-        int dataSize = 0;
-        for (int i = 0; i < type_.size(); i++) {
-          dataSize += com.google.protobuf.CodedOutputStream
-            .computeEnumSizeNoTag(type_.get(i));
-        }
-        size += dataSize;
-        if (!getTypeList().isEmpty()) {  size += 1;
-          size += com.google.protobuf.CodedOutputStream
-            .computeUInt32SizeNoTag(dataSize);
-        }typeMemoizedSerializedSize = dataSize;
-      }
-      for (int i = 0; i < shape_.size(); i++) {
-        size += com.google.protobuf.CodedOutputStream
-          .computeMessageSize(7, shape_.get(i));
-      }
-      for (int i = 0; i < tensor_.size(); i++) {
-        size += com.google.protobuf.CodedOutputStream
-          .computeMessageSize(8, tensor_.get(i));
-      }
-      for (int i = 0; i < func_.size(); i++) {
-        size += com.google.protobuf.CodedOutputStream
-          .computeMessageSize(9, func_.get(i));
-      }
-      size += unknownFields.getSerializedSize();
-      memoizedSize = size;
-      return size;
-    }
-
-    @java.lang.Override
-    public boolean equals(final java.lang.Object obj) {
-      if (obj == this) {
-       return true;
-      }
-      if (!(obj instanceof org.tensorflow.proto.framework.AttrValue.ListValue)) {
-        return super.equals(obj);
-      }
-      org.tensorflow.proto.framework.AttrValue.ListValue other = (org.tensorflow.proto.framework.AttrValue.ListValue) obj;
-
-      if (!getSList()
-          .equals(other.getSList())) return false;
-      if (!getIList()
-          .equals(other.getIList())) return false;
-      if (!getFList()
-          .equals(other.getFList())) return false;
-      if (!getBList()
-          .equals(other.getBList())) return false;
-      if (!type_.equals(other.type_)) return false;
-      if (!getShapeList()
-          .equals(other.getShapeList())) return false;
-      if (!getTensorList()
-          .equals(other.getTensorList())) return false;
-      if (!getFuncList()
-          .equals(other.getFuncList())) return false;
-      if (!unknownFields.equals(other.unknownFields)) return false;
-      return true;
-    }
-
-    @java.lang.Override
-    public int hashCode() {
-      if (memoizedHashCode != 0) {
-        return memoizedHashCode;
-      }
-      int hash = 41;
-      hash = (19 * hash) + getDescriptor().hashCode();
-      if (getSCount() > 0) {
-        hash = (37 * hash) + S_FIELD_NUMBER;
-        hash = (53 * hash) + getSList().hashCode();
-      }
-      if (getICount() > 0) {
-        hash = (37 * hash) + I_FIELD_NUMBER;
-        hash = (53 * hash) + getIList().hashCode();
-      }
-      if (getFCount() > 0) {
-        hash = (37 * hash) + F_FIELD_NUMBER;
-        hash = (53 * hash) + getFList().hashCode();
-      }
-      if (getBCount() > 0) {
-        hash = (37 * hash) + B_FIELD_NUMBER;
-        hash = (53 * hash) + getBList().hashCode();
-      }
-      if (getTypeCount() > 0) {
-        hash = (37 * hash) + TYPE_FIELD_NUMBER;
-        hash = (53 * hash) + type_.hashCode();
-      }
-      if (getShapeCount() > 0) {
-        hash = (37 * hash) + SHAPE_FIELD_NUMBER;
-        hash = (53 * hash) + getShapeList().hashCode();
-      }
-      if (getTensorCount() > 0) {
-        hash = (37 * hash) + TENSOR_FIELD_NUMBER;
-        hash = (53 * hash) + getTensorList().hashCode();
-      }
-      if (getFuncCount() > 0) {
-        hash = (37 * hash) + FUNC_FIELD_NUMBER;
-        hash = (53 * hash) + getFuncList().hashCode();
-      }
-      hash = (29 * hash) + unknownFields.hashCode();
-      memoizedHashCode = hash;
-      return hash;
-    }
-
-    public static org.tensorflow.proto.framework.AttrValue.ListValue parseFrom(
-        java.nio.ByteBuffer data)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data);
-    }
-    public static org.tensorflow.proto.framework.AttrValue.ListValue parseFrom(
-        java.nio.ByteBuffer data,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.AttrValue.ListValue parseFrom(
-        com.google.protobuf.ByteString data)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data);
-    }
-    public static org.tensorflow.proto.framework.AttrValue.ListValue parseFrom(
-        com.google.protobuf.ByteString data,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.AttrValue.ListValue parseFrom(byte[] data)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data);
-    }
-    public static org.tensorflow.proto.framework.AttrValue.ListValue parseFrom(
-        byte[] data,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.AttrValue.ListValue parseFrom(java.io.InputStream input)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input);
-    }
-    public static org.tensorflow.proto.framework.AttrValue.ListValue parseFrom(
-        java.io.InputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.AttrValue.ListValue parseDelimitedFrom(java.io.InputStream input)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseDelimitedWithIOException(PARSER, input);
-    }
-    public static org.tensorflow.proto.framework.AttrValue.ListValue parseDelimitedFrom(
-        java.io.InputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.AttrValue.ListValue parseFrom(
-        com.google.protobuf.CodedInputStream input)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input);
-    }
-    public static org.tensorflow.proto.framework.AttrValue.ListValue parseFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input, extensionRegistry);
-    }
-
-    @java.lang.Override
-    public Builder newBuilderForType() { return newBuilder(); }
-    public static Builder newBuilder() {
-      return DEFAULT_INSTANCE.toBuilder();
-    }
-    public static Builder newBuilder(org.tensorflow.proto.framework.AttrValue.ListValue prototype) {
-      return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-    }
-    @java.lang.Override
-    public Builder toBuilder() {
-      return this == DEFAULT_INSTANCE
-          ? new Builder() : new Builder().mergeFrom(this);
-    }
-
-    @java.lang.Override
-    protected Builder newBuilderForType(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      Builder builder = new Builder(parent);
-      return builder;
-    }
-    /**
-     * <pre>
-     * LINT.IfChange
-     * </pre>
-     *
-     * Protobuf type {@code tensorflow.AttrValue.ListValue}
-     */
-    public static final class Builder extends
-        com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-        // @@protoc_insertion_point(builder_implements:tensorflow.AttrValue.ListValue)
-        org.tensorflow.proto.framework.AttrValue.ListValueOrBuilder {
-      public static final com.google.protobuf.Descriptors.Descriptor
-          getDescriptor() {
-        return org.tensorflow.proto.framework.AttrValueProtos.internal_static_tensorflow_AttrValue_ListValue_descriptor;
-      }
-
-      @java.lang.Override
-      protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-          internalGetFieldAccessorTable() {
-        return org.tensorflow.proto.framework.AttrValueProtos.internal_static_tensorflow_AttrValue_ListValue_fieldAccessorTable
-            .ensureFieldAccessorsInitialized(
-                org.tensorflow.proto.framework.AttrValue.ListValue.class, org.tensorflow.proto.framework.AttrValue.ListValue.Builder.class);
-      }
-
-      // Construct using org.tensorflow.proto.framework.AttrValue.ListValue.newBuilder()
-      private Builder() {
-        maybeForceBuilderInitialization();
-      }
-
-      private Builder(
-          com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-        super(parent);
-        maybeForceBuilderInitialization();
-      }
-      private void maybeForceBuilderInitialization() {
-        if (com.google.protobuf.GeneratedMessageV3
-                .alwaysUseFieldBuilders) {
-          getShapeFieldBuilder();
-          getTensorFieldBuilder();
-          getFuncFieldBuilder();
-        }
-      }
-      @java.lang.Override
-      public Builder clear() {
-        super.clear();
-        s_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000001);
-        i_ = emptyLongList();
-        bitField0_ = (bitField0_ & ~0x00000002);
-        f_ = emptyFloatList();
-        bitField0_ = (bitField0_ & ~0x00000004);
-        b_ = emptyBooleanList();
-        bitField0_ = (bitField0_ & ~0x00000008);
-        type_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000010);
-        if (shapeBuilder_ == null) {
-          shape_ = java.util.Collections.emptyList();
-          bitField0_ = (bitField0_ & ~0x00000020);
-        } else {
-          shapeBuilder_.clear();
-        }
-        if (tensorBuilder_ == null) {
-          tensor_ = java.util.Collections.emptyList();
-          bitField0_ = (bitField0_ & ~0x00000040);
-        } else {
-          tensorBuilder_.clear();
-        }
-        if (funcBuilder_ == null) {
-          func_ = java.util.Collections.emptyList();
-          bitField0_ = (bitField0_ & ~0x00000080);
-        } else {
-          funcBuilder_.clear();
-        }
-        return this;
-      }
-
-      @java.lang.Override
-      public com.google.protobuf.Descriptors.Descriptor
-          getDescriptorForType() {
-        return org.tensorflow.proto.framework.AttrValueProtos.internal_static_tensorflow_AttrValue_ListValue_descriptor;
-      }
-
-      @java.lang.Override
-      public org.tensorflow.proto.framework.AttrValue.ListValue getDefaultInstanceForType() {
-        return org.tensorflow.proto.framework.AttrValue.ListValue.getDefaultInstance();
-      }
-
-      @java.lang.Override
-      public org.tensorflow.proto.framework.AttrValue.ListValue build() {
-        org.tensorflow.proto.framework.AttrValue.ListValue result = buildPartial();
-        if (!result.isInitialized()) {
-          throw newUninitializedMessageException(result);
-        }
-        return result;
-      }
-
-      @java.lang.Override
-      public org.tensorflow.proto.framework.AttrValue.ListValue buildPartial() {
-        org.tensorflow.proto.framework.AttrValue.ListValue result = new org.tensorflow.proto.framework.AttrValue.ListValue(this);
-        int from_bitField0_ = bitField0_;
-        if (((bitField0_ & 0x00000001) != 0)) {
-          s_ = java.util.Collections.unmodifiableList(s_);
-          bitField0_ = (bitField0_ & ~0x00000001);
-        }
-        result.s_ = s_;
-        if (((bitField0_ & 0x00000002) != 0)) {
-          i_.makeImmutable();
-          bitField0_ = (bitField0_ & ~0x00000002);
-        }
-        result.i_ = i_;
-        if (((bitField0_ & 0x00000004) != 0)) {
-          f_.makeImmutable();
-          bitField0_ = (bitField0_ & ~0x00000004);
-        }
-        result.f_ = f_;
-        if (((bitField0_ & 0x00000008) != 0)) {
-          b_.makeImmutable();
-          bitField0_ = (bitField0_ & ~0x00000008);
-        }
-        result.b_ = b_;
-        if (((bitField0_ & 0x00000010) != 0)) {
-          type_ = java.util.Collections.unmodifiableList(type_);
-          bitField0_ = (bitField0_ & ~0x00000010);
-        }
-        result.type_ = type_;
-        if (shapeBuilder_ == null) {
-          if (((bitField0_ & 0x00000020) != 0)) {
-            shape_ = java.util.Collections.unmodifiableList(shape_);
-            bitField0_ = (bitField0_ & ~0x00000020);
-          }
-          result.shape_ = shape_;
-        } else {
-          result.shape_ = shapeBuilder_.build();
-        }
-        if (tensorBuilder_ == null) {
-          if (((bitField0_ & 0x00000040) != 0)) {
-            tensor_ = java.util.Collections.unmodifiableList(tensor_);
-            bitField0_ = (bitField0_ & ~0x00000040);
-          }
-          result.tensor_ = tensor_;
-        } else {
-          result.tensor_ = tensorBuilder_.build();
-        }
-        if (funcBuilder_ == null) {
-          if (((bitField0_ & 0x00000080) != 0)) {
-            func_ = java.util.Collections.unmodifiableList(func_);
-            bitField0_ = (bitField0_ & ~0x00000080);
-          }
-          result.func_ = func_;
-        } else {
-          result.func_ = funcBuilder_.build();
-        }
-        onBuilt();
-        return result;
-      }
-
-      @java.lang.Override
-      public Builder clone() {
-        return super.clone();
-      }
-      @java.lang.Override
-      public Builder setField(
-          com.google.protobuf.Descriptors.FieldDescriptor field,
-          java.lang.Object value) {
-        return super.setField(field, value);
-      }
-      @java.lang.Override
-      public Builder clearField(
-          com.google.protobuf.Descriptors.FieldDescriptor field) {
-        return super.clearField(field);
-      }
-      @java.lang.Override
-      public Builder clearOneof(
-          com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-        return super.clearOneof(oneof);
-      }
-      @java.lang.Override
-      public Builder setRepeatedField(
-          com.google.protobuf.Descriptors.FieldDescriptor field,
-          int index, java.lang.Object value) {
-        return super.setRepeatedField(field, index, value);
-      }
-      @java.lang.Override
-      public Builder addRepeatedField(
-          com.google.protobuf.Descriptors.FieldDescriptor field,
-          java.lang.Object value) {
-        return super.addRepeatedField(field, value);
-      }
-      @java.lang.Override
-      public Builder mergeFrom(com.google.protobuf.Message other) {
-        if (other instanceof org.tensorflow.proto.framework.AttrValue.ListValue) {
-          return mergeFrom((org.tensorflow.proto.framework.AttrValue.ListValue)other);
-        } else {
-          super.mergeFrom(other);
-          return this;
-        }
-      }
-
-      public Builder mergeFrom(org.tensorflow.proto.framework.AttrValue.ListValue other) {
-        if (other == org.tensorflow.proto.framework.AttrValue.ListValue.getDefaultInstance()) return this;
-        if (!other.s_.isEmpty()) {
-          if (s_.isEmpty()) {
-            s_ = other.s_;
-            bitField0_ = (bitField0_ & ~0x00000001);
-          } else {
-            ensureSIsMutable();
-            s_.addAll(other.s_);
-          }
-          onChanged();
-        }
-        if (!other.i_.isEmpty()) {
-          if (i_.isEmpty()) {
-            i_ = other.i_;
-            bitField0_ = (bitField0_ & ~0x00000002);
-          } else {
-            ensureIIsMutable();
-            i_.addAll(other.i_);
-          }
-          onChanged();
-        }
-        if (!other.f_.isEmpty()) {
-          if (f_.isEmpty()) {
-            f_ = other.f_;
-            bitField0_ = (bitField0_ & ~0x00000004);
-          } else {
-            ensureFIsMutable();
-            f_.addAll(other.f_);
-          }
-          onChanged();
-        }
-        if (!other.b_.isEmpty()) {
-          if (b_.isEmpty()) {
-            b_ = other.b_;
-            bitField0_ = (bitField0_ & ~0x00000008);
-          } else {
-            ensureBIsMutable();
-            b_.addAll(other.b_);
-          }
-          onChanged();
-        }
-        if (!other.type_.isEmpty()) {
-          if (type_.isEmpty()) {
-            type_ = other.type_;
-            bitField0_ = (bitField0_ & ~0x00000010);
-          } else {
-            ensureTypeIsMutable();
-            type_.addAll(other.type_);
-          }
-          onChanged();
-        }
-        if (shapeBuilder_ == null) {
-          if (!other.shape_.isEmpty()) {
-            if (shape_.isEmpty()) {
-              shape_ = other.shape_;
-              bitField0_ = (bitField0_ & ~0x00000020);
-            } else {
-              ensureShapeIsMutable();
-              shape_.addAll(other.shape_);
-            }
-            onChanged();
-          }
-        } else {
-          if (!other.shape_.isEmpty()) {
-            if (shapeBuilder_.isEmpty()) {
-              shapeBuilder_.dispose();
-              shapeBuilder_ = null;
-              shape_ = other.shape_;
-              bitField0_ = (bitField0_ & ~0x00000020);
-              shapeBuilder_ = 
-                com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
-                   getShapeFieldBuilder() : null;
-            } else {
-              shapeBuilder_.addAllMessages(other.shape_);
-            }
-          }
-        }
-        if (tensorBuilder_ == null) {
-          if (!other.tensor_.isEmpty()) {
-            if (tensor_.isEmpty()) {
-              tensor_ = other.tensor_;
-              bitField0_ = (bitField0_ & ~0x00000040);
-            } else {
-              ensureTensorIsMutable();
-              tensor_.addAll(other.tensor_);
-            }
-            onChanged();
-          }
-        } else {
-          if (!other.tensor_.isEmpty()) {
-            if (tensorBuilder_.isEmpty()) {
-              tensorBuilder_.dispose();
-              tensorBuilder_ = null;
-              tensor_ = other.tensor_;
-              bitField0_ = (bitField0_ & ~0x00000040);
-              tensorBuilder_ = 
-                com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
-                   getTensorFieldBuilder() : null;
-            } else {
-              tensorBuilder_.addAllMessages(other.tensor_);
-            }
-          }
-        }
-        if (funcBuilder_ == null) {
-          if (!other.func_.isEmpty()) {
-            if (func_.isEmpty()) {
-              func_ = other.func_;
-              bitField0_ = (bitField0_ & ~0x00000080);
-            } else {
-              ensureFuncIsMutable();
-              func_.addAll(other.func_);
-            }
-            onChanged();
-          }
-        } else {
-          if (!other.func_.isEmpty()) {
-            if (funcBuilder_.isEmpty()) {
-              funcBuilder_.dispose();
-              funcBuilder_ = null;
-              func_ = other.func_;
-              bitField0_ = (bitField0_ & ~0x00000080);
-              funcBuilder_ = 
-                com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
-                   getFuncFieldBuilder() : null;
-            } else {
-              funcBuilder_.addAllMessages(other.func_);
-            }
-          }
-        }
-        this.mergeUnknownFields(other.unknownFields);
-        onChanged();
-        return this;
-      }
-
-      @java.lang.Override
-      public final boolean isInitialized() {
-        return true;
-      }
-
-      @java.lang.Override
-      public Builder mergeFrom(
-          com.google.protobuf.CodedInputStream input,
-          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-          throws java.io.IOException {
-        org.tensorflow.proto.framework.AttrValue.ListValue parsedMessage = null;
-        try {
-          parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-        } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-          parsedMessage = (org.tensorflow.proto.framework.AttrValue.ListValue) e.getUnfinishedMessage();
-          throw e.unwrapIOException();
-        } finally {
-          if (parsedMessage != null) {
-            mergeFrom(parsedMessage);
-          }
-        }
-        return this;
-      }
-      private int bitField0_;
-
-      private java.util.List<com.google.protobuf.ByteString> s_ = java.util.Collections.emptyList();
-      private void ensureSIsMutable() {
-        if (!((bitField0_ & 0x00000001) != 0)) {
-          s_ = new java.util.ArrayList<com.google.protobuf.ByteString>(s_);
-          bitField0_ |= 0x00000001;
-         }
-      }
-      /**
-       * <pre>
-       * "list(string)"
-       * </pre>
-       *
-       * <code>repeated bytes s = 2;</code>
-       */
-      public java.util.List<com.google.protobuf.ByteString>
-          getSList() {
-        return ((bitField0_ & 0x00000001) != 0) ?
-                 java.util.Collections.unmodifiableList(s_) : s_;
-      }
-      /**
-       * <pre>
-       * "list(string)"
-       * </pre>
-       *
-       * <code>repeated bytes s = 2;</code>
-       */
-      public int getSCount() {
-        return s_.size();
-      }
-      /**
-       * <pre>
-       * "list(string)"
-       * </pre>
-       *
-       * <code>repeated bytes s = 2;</code>
-       */
-      public com.google.protobuf.ByteString getS(int index) {
-        return s_.get(index);
-      }
-      /**
-       * <pre>
-       * "list(string)"
-       * </pre>
-       *
-       * <code>repeated bytes s = 2;</code>
-       */
-      public Builder setS(
-          int index, com.google.protobuf.ByteString value) {
-        if (value == null) {
-    throw new NullPointerException();
-  }
-  ensureSIsMutable();
-        s_.set(index, value);
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * "list(string)"
-       * </pre>
-       *
-       * <code>repeated bytes s = 2;</code>
-       */
-      public Builder addS(com.google.protobuf.ByteString value) {
-        if (value == null) {
-    throw new NullPointerException();
-  }
-  ensureSIsMutable();
-        s_.add(value);
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * "list(string)"
-       * </pre>
-       *
-       * <code>repeated bytes s = 2;</code>
-       */
-      public Builder addAllS(
-          java.lang.Iterable<? extends com.google.protobuf.ByteString> values) {
-        ensureSIsMutable();
-        com.google.protobuf.AbstractMessageLite.Builder.addAll(
-            values, s_);
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * "list(string)"
-       * </pre>
-       *
-       * <code>repeated bytes s = 2;</code>
-       */
-      public Builder clearS() {
-        s_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000001);
-        onChanged();
-        return this;
-      }
-
-      private com.google.protobuf.Internal.LongList i_ = emptyLongList();
-      private void ensureIIsMutable() {
-        if (!((bitField0_ & 0x00000002) != 0)) {
-          i_ = mutableCopy(i_);
-          bitField0_ |= 0x00000002;
-         }
-      }
-      /**
-       * <pre>
-       * "list(int)"
-       * </pre>
-       *
-       * <code>repeated int64 i = 3 [packed = true];</code>
-       */
-      public java.util.List<java.lang.Long>
-          getIList() {
-        return ((bitField0_ & 0x00000002) != 0) ?
-                 java.util.Collections.unmodifiableList(i_) : i_;
-      }
-      /**
-       * <pre>
-       * "list(int)"
-       * </pre>
-       *
-       * <code>repeated int64 i = 3 [packed = true];</code>
-       */
-      public int getICount() {
-        return i_.size();
-      }
-      /**
-       * <pre>
-       * "list(int)"
-       * </pre>
-       *
-       * <code>repeated int64 i = 3 [packed = true];</code>
-       */
-      public long getI(int index) {
-        return i_.getLong(index);
-      }
-      /**
-       * <pre>
-       * "list(int)"
-       * </pre>
-       *
-       * <code>repeated int64 i = 3 [packed = true];</code>
-       */
-      public Builder setI(
-          int index, long value) {
-        ensureIIsMutable();
-        i_.setLong(index, value);
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * "list(int)"
-       * </pre>
-       *
-       * <code>repeated int64 i = 3 [packed = true];</code>
-       */
-      public Builder addI(long value) {
-        ensureIIsMutable();
-        i_.addLong(value);
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * "list(int)"
-       * </pre>
-       *
-       * <code>repeated int64 i = 3 [packed = true];</code>
-       */
-      public Builder addAllI(
-          java.lang.Iterable<? extends java.lang.Long> values) {
-        ensureIIsMutable();
-        com.google.protobuf.AbstractMessageLite.Builder.addAll(
-            values, i_);
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * "list(int)"
-       * </pre>
-       *
-       * <code>repeated int64 i = 3 [packed = true];</code>
-       */
-      public Builder clearI() {
-        i_ = emptyLongList();
-        bitField0_ = (bitField0_ & ~0x00000002);
-        onChanged();
-        return this;
-      }
-
-      private com.google.protobuf.Internal.FloatList f_ = emptyFloatList();
-      private void ensureFIsMutable() {
-        if (!((bitField0_ & 0x00000004) != 0)) {
-          f_ = mutableCopy(f_);
-          bitField0_ |= 0x00000004;
-         }
-      }
-      /**
-       * <pre>
-       * "list(float)"
-       * </pre>
-       *
-       * <code>repeated float f = 4 [packed = true];</code>
-       */
-      public java.util.List<java.lang.Float>
-          getFList() {
-        return ((bitField0_ & 0x00000004) != 0) ?
-                 java.util.Collections.unmodifiableList(f_) : f_;
-      }
-      /**
-       * <pre>
-       * "list(float)"
-       * </pre>
-       *
-       * <code>repeated float f = 4 [packed = true];</code>
-       */
-      public int getFCount() {
-        return f_.size();
-      }
-      /**
-       * <pre>
-       * "list(float)"
-       * </pre>
-       *
-       * <code>repeated float f = 4 [packed = true];</code>
-       */
-      public float getF(int index) {
-        return f_.getFloat(index);
-      }
-      /**
-       * <pre>
-       * "list(float)"
-       * </pre>
-       *
-       * <code>repeated float f = 4 [packed = true];</code>
-       */
-      public Builder setF(
-          int index, float value) {
-        ensureFIsMutable();
-        f_.setFloat(index, value);
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * "list(float)"
-       * </pre>
-       *
-       * <code>repeated float f = 4 [packed = true];</code>
-       */
-      public Builder addF(float value) {
-        ensureFIsMutable();
-        f_.addFloat(value);
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * "list(float)"
-       * </pre>
-       *
-       * <code>repeated float f = 4 [packed = true];</code>
-       */
-      public Builder addAllF(
-          java.lang.Iterable<? extends java.lang.Float> values) {
-        ensureFIsMutable();
-        com.google.protobuf.AbstractMessageLite.Builder.addAll(
-            values, f_);
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * "list(float)"
-       * </pre>
-       *
-       * <code>repeated float f = 4 [packed = true];</code>
-       */
-      public Builder clearF() {
-        f_ = emptyFloatList();
-        bitField0_ = (bitField0_ & ~0x00000004);
-        onChanged();
-        return this;
-      }
-
-      private com.google.protobuf.Internal.BooleanList b_ = emptyBooleanList();
-      private void ensureBIsMutable() {
-        if (!((bitField0_ & 0x00000008) != 0)) {
-          b_ = mutableCopy(b_);
-          bitField0_ |= 0x00000008;
-         }
-      }
-      /**
-       * <pre>
-       * "list(bool)"
-       * </pre>
-       *
-       * <code>repeated bool b = 5 [packed = true];</code>
-       */
-      public java.util.List<java.lang.Boolean>
-          getBList() {
-        return ((bitField0_ & 0x00000008) != 0) ?
-                 java.util.Collections.unmodifiableList(b_) : b_;
-      }
-      /**
-       * <pre>
-       * "list(bool)"
-       * </pre>
-       *
-       * <code>repeated bool b = 5 [packed = true];</code>
-       */
-      public int getBCount() {
-        return b_.size();
-      }
-      /**
-       * <pre>
-       * "list(bool)"
-       * </pre>
-       *
-       * <code>repeated bool b = 5 [packed = true];</code>
-       */
-      public boolean getB(int index) {
-        return b_.getBoolean(index);
-      }
-      /**
-       * <pre>
-       * "list(bool)"
-       * </pre>
-       *
-       * <code>repeated bool b = 5 [packed = true];</code>
-       */
-      public Builder setB(
-          int index, boolean value) {
-        ensureBIsMutable();
-        b_.setBoolean(index, value);
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * "list(bool)"
-       * </pre>
-       *
-       * <code>repeated bool b = 5 [packed = true];</code>
-       */
-      public Builder addB(boolean value) {
-        ensureBIsMutable();
-        b_.addBoolean(value);
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * "list(bool)"
-       * </pre>
-       *
-       * <code>repeated bool b = 5 [packed = true];</code>
-       */
-      public Builder addAllB(
-          java.lang.Iterable<? extends java.lang.Boolean> values) {
-        ensureBIsMutable();
-        com.google.protobuf.AbstractMessageLite.Builder.addAll(
-            values, b_);
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * "list(bool)"
-       * </pre>
-       *
-       * <code>repeated bool b = 5 [packed = true];</code>
-       */
-      public Builder clearB() {
-        b_ = emptyBooleanList();
-        bitField0_ = (bitField0_ & ~0x00000008);
-        onChanged();
-        return this;
-      }
-
-      private java.util.List<java.lang.Integer> type_ =
-        java.util.Collections.emptyList();
-      private void ensureTypeIsMutable() {
-        if (!((bitField0_ & 0x00000010) != 0)) {
-          type_ = new java.util.ArrayList<java.lang.Integer>(type_);
-          bitField0_ |= 0x00000010;
-        }
-      }
-      /**
-       * <pre>
-       * "list(type)"
-       * </pre>
-       *
-       * <code>repeated .tensorflow.DataType type = 6 [packed = true];</code>
-       */
-      public java.util.List<org.tensorflow.proto.framework.DataType> getTypeList() {
-        return new com.google.protobuf.Internal.ListAdapter<
-            java.lang.Integer, org.tensorflow.proto.framework.DataType>(type_, type_converter_);
-      }
-      /**
-       * <pre>
-       * "list(type)"
-       * </pre>
-       *
-       * <code>repeated .tensorflow.DataType type = 6 [packed = true];</code>
-       */
-      public int getTypeCount() {
-        return type_.size();
-      }
-      /**
-       * <pre>
-       * "list(type)"
-       * </pre>
-       *
-       * <code>repeated .tensorflow.DataType type = 6 [packed = true];</code>
-       */
-      public org.tensorflow.proto.framework.DataType getType(int index) {
-        return type_converter_.convert(type_.get(index));
-      }
-      /**
-       * <pre>
-       * "list(type)"
-       * </pre>
-       *
-       * <code>repeated .tensorflow.DataType type = 6 [packed = true];</code>
-       */
-      public Builder setType(
-          int index, org.tensorflow.proto.framework.DataType value) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureTypeIsMutable();
-        type_.set(index, value.getNumber());
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * "list(type)"
-       * </pre>
-       *
-       * <code>repeated .tensorflow.DataType type = 6 [packed = true];</code>
-       */
-      public Builder addType(org.tensorflow.proto.framework.DataType value) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureTypeIsMutable();
-        type_.add(value.getNumber());
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * "list(type)"
-       * </pre>
-       *
-       * <code>repeated .tensorflow.DataType type = 6 [packed = true];</code>
-       */
-      public Builder addAllType(
-          java.lang.Iterable<? extends org.tensorflow.proto.framework.DataType> values) {
-        ensureTypeIsMutable();
-        for (org.tensorflow.proto.framework.DataType value : values) {
-          type_.add(value.getNumber());
-        }
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * "list(type)"
-       * </pre>
-       *
-       * <code>repeated .tensorflow.DataType type = 6 [packed = true];</code>
-       */
-      public Builder clearType() {
-        type_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000010);
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * "list(type)"
-       * </pre>
-       *
-       * <code>repeated .tensorflow.DataType type = 6 [packed = true];</code>
-       */
-      public java.util.List<java.lang.Integer>
-      getTypeValueList() {
-        return java.util.Collections.unmodifiableList(type_);
-      }
-      /**
-       * <pre>
-       * "list(type)"
-       * </pre>
-       *
-       * <code>repeated .tensorflow.DataType type = 6 [packed = true];</code>
-       */
-      public int getTypeValue(int index) {
-        return type_.get(index);
-      }
-      /**
-       * <pre>
-       * "list(type)"
-       * </pre>
-       *
-       * <code>repeated .tensorflow.DataType type = 6 [packed = true];</code>
-       */
-      public Builder setTypeValue(
-          int index, int value) {
-        ensureTypeIsMutable();
-        type_.set(index, value);
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * "list(type)"
-       * </pre>
-       *
-       * <code>repeated .tensorflow.DataType type = 6 [packed = true];</code>
-       */
-      public Builder addTypeValue(int value) {
-        ensureTypeIsMutable();
-        type_.add(value);
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * "list(type)"
-       * </pre>
-       *
-       * <code>repeated .tensorflow.DataType type = 6 [packed = true];</code>
-       */
-      public Builder addAllTypeValue(
-          java.lang.Iterable<java.lang.Integer> values) {
-        ensureTypeIsMutable();
-        for (int value : values) {
-          type_.add(value);
-        }
-        onChanged();
-        return this;
-      }
-
-      private java.util.List<org.tensorflow.proto.framework.TensorShapeProto> shape_ =
-        java.util.Collections.emptyList();
-      private void ensureShapeIsMutable() {
-        if (!((bitField0_ & 0x00000020) != 0)) {
-          shape_ = new java.util.ArrayList<org.tensorflow.proto.framework.TensorShapeProto>(shape_);
-          bitField0_ |= 0x00000020;
-         }
-      }
-
-      private com.google.protobuf.RepeatedFieldBuilderV3<
-          org.tensorflow.proto.framework.TensorShapeProto, org.tensorflow.proto.framework.TensorShapeProto.Builder, org.tensorflow.proto.framework.TensorShapeProtoOrBuilder> shapeBuilder_;
-
-      /**
-       * <pre>
-       * "list(shape)"
-       * </pre>
-       *
-       * <code>repeated .tensorflow.TensorShapeProto shape = 7;</code>
-       */
-      public java.util.List<org.tensorflow.proto.framework.TensorShapeProto> getShapeList() {
-        if (shapeBuilder_ == null) {
-          return java.util.Collections.unmodifiableList(shape_);
-        } else {
-          return shapeBuilder_.getMessageList();
-        }
-      }
-      /**
-       * <pre>
-       * "list(shape)"
-       * </pre>
-       *
-       * <code>repeated .tensorflow.TensorShapeProto shape = 7;</code>
-       */
-      public int getShapeCount() {
-        if (shapeBuilder_ == null) {
-          return shape_.size();
-        } else {
-          return shapeBuilder_.getCount();
-        }
-      }
-      /**
-       * <pre>
-       * "list(shape)"
-       * </pre>
-       *
-       * <code>repeated .tensorflow.TensorShapeProto shape = 7;</code>
-       */
-      public org.tensorflow.proto.framework.TensorShapeProto getShape(int index) {
-        if (shapeBuilder_ == null) {
-          return shape_.get(index);
-        } else {
-          return shapeBuilder_.getMessage(index);
-        }
-      }
-      /**
-       * <pre>
-       * "list(shape)"
-       * </pre>
-       *
-       * <code>repeated .tensorflow.TensorShapeProto shape = 7;</code>
-       */
-      public Builder setShape(
-          int index, org.tensorflow.proto.framework.TensorShapeProto value) {
-        if (shapeBuilder_ == null) {
-          if (value == null) {
-            throw new NullPointerException();
-          }
-          ensureShapeIsMutable();
-          shape_.set(index, value);
-          onChanged();
-        } else {
-          shapeBuilder_.setMessage(index, value);
-        }
-        return this;
-      }
-      /**
-       * <pre>
-       * "list(shape)"
-       * </pre>
-       *
-       * <code>repeated .tensorflow.TensorShapeProto shape = 7;</code>
-       */
-      public Builder setShape(
-          int index, org.tensorflow.proto.framework.TensorShapeProto.Builder builderForValue) {
-        if (shapeBuilder_ == null) {
-          ensureShapeIsMutable();
-          shape_.set(index, builderForValue.build());
-          onChanged();
-        } else {
-          shapeBuilder_.setMessage(index, builderForValue.build());
-        }
-        return this;
-      }
-      /**
-       * <pre>
-       * "list(shape)"
-       * </pre>
-       *
-       * <code>repeated .tensorflow.TensorShapeProto shape = 7;</code>
-       */
-      public Builder addShape(org.tensorflow.proto.framework.TensorShapeProto value) {
-        if (shapeBuilder_ == null) {
-          if (value == null) {
-            throw new NullPointerException();
-          }
-          ensureShapeIsMutable();
-          shape_.add(value);
-          onChanged();
-        } else {
-          shapeBuilder_.addMessage(value);
-        }
-        return this;
-      }
-      /**
-       * <pre>
-       * "list(shape)"
-       * </pre>
-       *
-       * <code>repeated .tensorflow.TensorShapeProto shape = 7;</code>
-       */
-      public Builder addShape(
-          int index, org.tensorflow.proto.framework.TensorShapeProto value) {
-        if (shapeBuilder_ == null) {
-          if (value == null) {
-            throw new NullPointerException();
-          }
-          ensureShapeIsMutable();
-          shape_.add(index, value);
-          onChanged();
-        } else {
-          shapeBuilder_.addMessage(index, value);
-        }
-        return this;
-      }
-      /**
-       * <pre>
-       * "list(shape)"
-       * </pre>
-       *
-       * <code>repeated .tensorflow.TensorShapeProto shape = 7;</code>
-       */
-      public Builder addShape(
-          org.tensorflow.proto.framework.TensorShapeProto.Builder builderForValue) {
-        if (shapeBuilder_ == null) {
-          ensureShapeIsMutable();
-          shape_.add(builderForValue.build());
-          onChanged();
-        } else {
-          shapeBuilder_.addMessage(builderForValue.build());
-        }
-        return this;
-      }
-      /**
-       * <pre>
-       * "list(shape)"
-       * </pre>
-       *
-       * <code>repeated .tensorflow.TensorShapeProto shape = 7;</code>
-       */
-      public Builder addShape(
-          int index, org.tensorflow.proto.framework.TensorShapeProto.Builder builderForValue) {
-        if (shapeBuilder_ == null) {
-          ensureShapeIsMutable();
-          shape_.add(index, builderForValue.build());
-          onChanged();
-        } else {
-          shapeBuilder_.addMessage(index, builderForValue.build());
-        }
-        return this;
-      }
-      /**
-       * <pre>
-       * "list(shape)"
-       * </pre>
-       *
-       * <code>repeated .tensorflow.TensorShapeProto shape = 7;</code>
-       */
-      public Builder addAllShape(
-          java.lang.Iterable<? extends org.tensorflow.proto.framework.TensorShapeProto> values) {
-        if (shapeBuilder_ == null) {
-          ensureShapeIsMutable();
-          com.google.protobuf.AbstractMessageLite.Builder.addAll(
-              values, shape_);
-          onChanged();
-        } else {
-          shapeBuilder_.addAllMessages(values);
-        }
-        return this;
-      }
-      /**
-       * <pre>
-       * "list(shape)"
-       * </pre>
-       *
-       * <code>repeated .tensorflow.TensorShapeProto shape = 7;</code>
-       */
-      public Builder clearShape() {
-        if (shapeBuilder_ == null) {
-          shape_ = java.util.Collections.emptyList();
-          bitField0_ = (bitField0_ & ~0x00000020);
-          onChanged();
-        } else {
-          shapeBuilder_.clear();
-        }
-        return this;
-      }
-      /**
-       * <pre>
-       * "list(shape)"
-       * </pre>
-       *
-       * <code>repeated .tensorflow.TensorShapeProto shape = 7;</code>
-       */
-      public Builder removeShape(int index) {
-        if (shapeBuilder_ == null) {
-          ensureShapeIsMutable();
-          shape_.remove(index);
-          onChanged();
-        } else {
-          shapeBuilder_.remove(index);
-        }
-        return this;
-      }
-      /**
-       * <pre>
-       * "list(shape)"
-       * </pre>
-       *
-       * <code>repeated .tensorflow.TensorShapeProto shape = 7;</code>
-       */
-      public org.tensorflow.proto.framework.TensorShapeProto.Builder getShapeBuilder(
-          int index) {
-        return getShapeFieldBuilder().getBuilder(index);
-      }
-      /**
-       * <pre>
-       * "list(shape)"
-       * </pre>
-       *
-       * <code>repeated .tensorflow.TensorShapeProto shape = 7;</code>
-       */
-      public org.tensorflow.proto.framework.TensorShapeProtoOrBuilder getShapeOrBuilder(
-          int index) {
-        if (shapeBuilder_ == null) {
-          return shape_.get(index);  } else {
-          return shapeBuilder_.getMessageOrBuilder(index);
-        }
-      }
-      /**
-       * <pre>
-       * "list(shape)"
-       * </pre>
-       *
-       * <code>repeated .tensorflow.TensorShapeProto shape = 7;</code>
-       */
-      public java.util.List<? extends org.tensorflow.proto.framework.TensorShapeProtoOrBuilder> 
-           getShapeOrBuilderList() {
-        if (shapeBuilder_ != null) {
-          return shapeBuilder_.getMessageOrBuilderList();
-        } else {
-          return java.util.Collections.unmodifiableList(shape_);
-        }
-      }
-      /**
-       * <pre>
-       * "list(shape)"
-       * </pre>
-       *
-       * <code>repeated .tensorflow.TensorShapeProto shape = 7;</code>
-       */
-      public org.tensorflow.proto.framework.TensorShapeProto.Builder addShapeBuilder() {
-        return getShapeFieldBuilder().addBuilder(
-            org.tensorflow.proto.framework.TensorShapeProto.getDefaultInstance());
-      }
-      /**
-       * <pre>
-       * "list(shape)"
-       * </pre>
-       *
-       * <code>repeated .tensorflow.TensorShapeProto shape = 7;</code>
-       */
-      public org.tensorflow.proto.framework.TensorShapeProto.Builder addShapeBuilder(
-          int index) {
-        return getShapeFieldBuilder().addBuilder(
-            index, org.tensorflow.proto.framework.TensorShapeProto.getDefaultInstance());
-      }
-      /**
-       * <pre>
-       * "list(shape)"
-       * </pre>
-       *
-       * <code>repeated .tensorflow.TensorShapeProto shape = 7;</code>
-       */
-      public java.util.List<org.tensorflow.proto.framework.TensorShapeProto.Builder> 
-           getShapeBuilderList() {
-        return getShapeFieldBuilder().getBuilderList();
-      }
-      private com.google.protobuf.RepeatedFieldBuilderV3<
-          org.tensorflow.proto.framework.TensorShapeProto, org.tensorflow.proto.framework.TensorShapeProto.Builder, org.tensorflow.proto.framework.TensorShapeProtoOrBuilder> 
-          getShapeFieldBuilder() {
-        if (shapeBuilder_ == null) {
-          shapeBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
-              org.tensorflow.proto.framework.TensorShapeProto, org.tensorflow.proto.framework.TensorShapeProto.Builder, org.tensorflow.proto.framework.TensorShapeProtoOrBuilder>(
-                  shape_,
-                  ((bitField0_ & 0x00000020) != 0),
-                  getParentForChildren(),
-                  isClean());
-          shape_ = null;
-        }
-        return shapeBuilder_;
-      }
-
-      private java.util.List<org.tensorflow.proto.framework.TensorProto> tensor_ =
-        java.util.Collections.emptyList();
-      private void ensureTensorIsMutable() {
-        if (!((bitField0_ & 0x00000040) != 0)) {
-          tensor_ = new java.util.ArrayList<org.tensorflow.proto.framework.TensorProto>(tensor_);
-          bitField0_ |= 0x00000040;
-         }
-      }
-
-      private com.google.protobuf.RepeatedFieldBuilderV3<
-          org.tensorflow.proto.framework.TensorProto, org.tensorflow.proto.framework.TensorProto.Builder, org.tensorflow.proto.framework.TensorProtoOrBuilder> tensorBuilder_;
-
-      /**
-       * <pre>
-       * "list(tensor)"
-       * </pre>
-       *
-       * <code>repeated .tensorflow.TensorProto tensor = 8;</code>
-       */
-      public java.util.List<org.tensorflow.proto.framework.TensorProto> getTensorList() {
-        if (tensorBuilder_ == null) {
-          return java.util.Collections.unmodifiableList(tensor_);
-        } else {
-          return tensorBuilder_.getMessageList();
-        }
-      }
-      /**
-       * <pre>
-       * "list(tensor)"
-       * </pre>
-       *
-       * <code>repeated .tensorflow.TensorProto tensor = 8;</code>
-       */
-      public int getTensorCount() {
-        if (tensorBuilder_ == null) {
-          return tensor_.size();
-        } else {
-          return tensorBuilder_.getCount();
-        }
-      }
-      /**
-       * <pre>
-       * "list(tensor)"
-       * </pre>
-       *
-       * <code>repeated .tensorflow.TensorProto tensor = 8;</code>
-       */
-      public org.tensorflow.proto.framework.TensorProto getTensor(int index) {
-        if (tensorBuilder_ == null) {
-          return tensor_.get(index);
-        } else {
-          return tensorBuilder_.getMessage(index);
-        }
-      }
-      /**
-       * <pre>
-       * "list(tensor)"
-       * </pre>
-       *
-       * <code>repeated .tensorflow.TensorProto tensor = 8;</code>
-       */
-      public Builder setTensor(
-          int index, org.tensorflow.proto.framework.TensorProto value) {
-        if (tensorBuilder_ == null) {
-          if (value == null) {
-            throw new NullPointerException();
-          }
-          ensureTensorIsMutable();
-          tensor_.set(index, value);
-          onChanged();
-        } else {
-          tensorBuilder_.setMessage(index, value);
-        }
-        return this;
-      }
-      /**
-       * <pre>
-       * "list(tensor)"
-       * </pre>
-       *
-       * <code>repeated .tensorflow.TensorProto tensor = 8;</code>
-       */
-      public Builder setTensor(
-          int index, org.tensorflow.proto.framework.TensorProto.Builder builderForValue) {
-        if (tensorBuilder_ == null) {
-          ensureTensorIsMutable();
-          tensor_.set(index, builderForValue.build());
-          onChanged();
-        } else {
-          tensorBuilder_.setMessage(index, builderForValue.build());
-        }
-        return this;
-      }
-      /**
-       * <pre>
-       * "list(tensor)"
-       * </pre>
-       *
-       * <code>repeated .tensorflow.TensorProto tensor = 8;</code>
-       */
-      public Builder addTensor(org.tensorflow.proto.framework.TensorProto value) {
-        if (tensorBuilder_ == null) {
-          if (value == null) {
-            throw new NullPointerException();
-          }
-          ensureTensorIsMutable();
-          tensor_.add(value);
-          onChanged();
-        } else {
-          tensorBuilder_.addMessage(value);
-        }
-        return this;
-      }
-      /**
-       * <pre>
-       * "list(tensor)"
-       * </pre>
-       *
-       * <code>repeated .tensorflow.TensorProto tensor = 8;</code>
-       */
-      public Builder addTensor(
-          int index, org.tensorflow.proto.framework.TensorProto value) {
-        if (tensorBuilder_ == null) {
-          if (value == null) {
-            throw new NullPointerException();
-          }
-          ensureTensorIsMutable();
-          tensor_.add(index, value);
-          onChanged();
-        } else {
-          tensorBuilder_.addMessage(index, value);
-        }
-        return this;
-      }
-      /**
-       * <pre>
-       * "list(tensor)"
-       * </pre>
-       *
-       * <code>repeated .tensorflow.TensorProto tensor = 8;</code>
-       */
-      public Builder addTensor(
-          org.tensorflow.proto.framework.TensorProto.Builder builderForValue) {
-        if (tensorBuilder_ == null) {
-          ensureTensorIsMutable();
-          tensor_.add(builderForValue.build());
-          onChanged();
-        } else {
-          tensorBuilder_.addMessage(builderForValue.build());
-        }
-        return this;
-      }
-      /**
-       * <pre>
-       * "list(tensor)"
-       * </pre>
-       *
-       * <code>repeated .tensorflow.TensorProto tensor = 8;</code>
-       */
-      public Builder addTensor(
-          int index, org.tensorflow.proto.framework.TensorProto.Builder builderForValue) {
-        if (tensorBuilder_ == null) {
-          ensureTensorIsMutable();
-          tensor_.add(index, builderForValue.build());
-          onChanged();
-        } else {
-          tensorBuilder_.addMessage(index, builderForValue.build());
-        }
-        return this;
-      }
-      /**
-       * <pre>
-       * "list(tensor)"
-       * </pre>
-       *
-       * <code>repeated .tensorflow.TensorProto tensor = 8;</code>
-       */
-      public Builder addAllTensor(
-          java.lang.Iterable<? extends org.tensorflow.proto.framework.TensorProto> values) {
-        if (tensorBuilder_ == null) {
-          ensureTensorIsMutable();
-          com.google.protobuf.AbstractMessageLite.Builder.addAll(
-              values, tensor_);
-          onChanged();
-        } else {
-          tensorBuilder_.addAllMessages(values);
-        }
-        return this;
-      }
-      /**
-       * <pre>
-       * "list(tensor)"
-       * </pre>
-       *
-       * <code>repeated .tensorflow.TensorProto tensor = 8;</code>
-       */
-      public Builder clearTensor() {
-        if (tensorBuilder_ == null) {
-          tensor_ = java.util.Collections.emptyList();
-          bitField0_ = (bitField0_ & ~0x00000040);
-          onChanged();
-        } else {
-          tensorBuilder_.clear();
-        }
-        return this;
-      }
-      /**
-       * <pre>
-       * "list(tensor)"
-       * </pre>
-       *
-       * <code>repeated .tensorflow.TensorProto tensor = 8;</code>
-       */
-      public Builder removeTensor(int index) {
-        if (tensorBuilder_ == null) {
-          ensureTensorIsMutable();
-          tensor_.remove(index);
-          onChanged();
-        } else {
-          tensorBuilder_.remove(index);
-        }
-        return this;
-      }
-      /**
-       * <pre>
-       * "list(tensor)"
-       * </pre>
-       *
-       * <code>repeated .tensorflow.TensorProto tensor = 8;</code>
-       */
-      public org.tensorflow.proto.framework.TensorProto.Builder getTensorBuilder(
-          int index) {
-        return getTensorFieldBuilder().getBuilder(index);
-      }
-      /**
-       * <pre>
-       * "list(tensor)"
-       * </pre>
-       *
-       * <code>repeated .tensorflow.TensorProto tensor = 8;</code>
-       */
-      public org.tensorflow.proto.framework.TensorProtoOrBuilder getTensorOrBuilder(
-          int index) {
-        if (tensorBuilder_ == null) {
-          return tensor_.get(index);  } else {
-          return tensorBuilder_.getMessageOrBuilder(index);
-        }
-      }
-      /**
-       * <pre>
-       * "list(tensor)"
-       * </pre>
-       *
-       * <code>repeated .tensorflow.TensorProto tensor = 8;</code>
-       */
-      public java.util.List<? extends org.tensorflow.proto.framework.TensorProtoOrBuilder> 
-           getTensorOrBuilderList() {
-        if (tensorBuilder_ != null) {
-          return tensorBuilder_.getMessageOrBuilderList();
-        } else {
-          return java.util.Collections.unmodifiableList(tensor_);
-        }
-      }
-      /**
-       * <pre>
-       * "list(tensor)"
-       * </pre>
-       *
-       * <code>repeated .tensorflow.TensorProto tensor = 8;</code>
-       */
-      public org.tensorflow.proto.framework.TensorProto.Builder addTensorBuilder() {
-        return getTensorFieldBuilder().addBuilder(
-            org.tensorflow.proto.framework.TensorProto.getDefaultInstance());
-      }
-      /**
-       * <pre>
-       * "list(tensor)"
-       * </pre>
-       *
-       * <code>repeated .tensorflow.TensorProto tensor = 8;</code>
-       */
-      public org.tensorflow.proto.framework.TensorProto.Builder addTensorBuilder(
-          int index) {
-        return getTensorFieldBuilder().addBuilder(
-            index, org.tensorflow.proto.framework.TensorProto.getDefaultInstance());
-      }
-      /**
-       * <pre>
-       * "list(tensor)"
-       * </pre>
-       *
-       * <code>repeated .tensorflow.TensorProto tensor = 8;</code>
-       */
-      public java.util.List<org.tensorflow.proto.framework.TensorProto.Builder> 
-           getTensorBuilderList() {
-        return getTensorFieldBuilder().getBuilderList();
-      }
-      private com.google.protobuf.RepeatedFieldBuilderV3<
-          org.tensorflow.proto.framework.TensorProto, org.tensorflow.proto.framework.TensorProto.Builder, org.tensorflow.proto.framework.TensorProtoOrBuilder> 
-          getTensorFieldBuilder() {
-        if (tensorBuilder_ == null) {
-          tensorBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
-              org.tensorflow.proto.framework.TensorProto, org.tensorflow.proto.framework.TensorProto.Builder, org.tensorflow.proto.framework.TensorProtoOrBuilder>(
-                  tensor_,
-                  ((bitField0_ & 0x00000040) != 0),
-                  getParentForChildren(),
-                  isClean());
-          tensor_ = null;
-        }
-        return tensorBuilder_;
-      }
-
-      private java.util.List<org.tensorflow.proto.framework.NameAttrList> func_ =
-        java.util.Collections.emptyList();
-      private void ensureFuncIsMutable() {
-        if (!((bitField0_ & 0x00000080) != 0)) {
-          func_ = new java.util.ArrayList<org.tensorflow.proto.framework.NameAttrList>(func_);
-          bitField0_ |= 0x00000080;
-         }
-      }
-
-      private com.google.protobuf.RepeatedFieldBuilderV3<
-          org.tensorflow.proto.framework.NameAttrList, org.tensorflow.proto.framework.NameAttrList.Builder, org.tensorflow.proto.framework.NameAttrListOrBuilder> funcBuilder_;
-
-      /**
-       * <pre>
-       * "list(attr)"
-       * </pre>
-       *
-       * <code>repeated .tensorflow.NameAttrList func = 9;</code>
-       */
-      public java.util.List<org.tensorflow.proto.framework.NameAttrList> getFuncList() {
-        if (funcBuilder_ == null) {
-          return java.util.Collections.unmodifiableList(func_);
-        } else {
-          return funcBuilder_.getMessageList();
-        }
-      }
-      /**
-       * <pre>
-       * "list(attr)"
-       * </pre>
-       *
-       * <code>repeated .tensorflow.NameAttrList func = 9;</code>
-       */
-      public int getFuncCount() {
-        if (funcBuilder_ == null) {
-          return func_.size();
-        } else {
-          return funcBuilder_.getCount();
-        }
-      }
-      /**
-       * <pre>
-       * "list(attr)"
-       * </pre>
-       *
-       * <code>repeated .tensorflow.NameAttrList func = 9;</code>
-       */
-      public org.tensorflow.proto.framework.NameAttrList getFunc(int index) {
-        if (funcBuilder_ == null) {
-          return func_.get(index);
-        } else {
-          return funcBuilder_.getMessage(index);
-        }
-      }
-      /**
-       * <pre>
-       * "list(attr)"
-       * </pre>
-       *
-       * <code>repeated .tensorflow.NameAttrList func = 9;</code>
-       */
-      public Builder setFunc(
-          int index, org.tensorflow.proto.framework.NameAttrList value) {
-        if (funcBuilder_ == null) {
-          if (value == null) {
-            throw new NullPointerException();
-          }
-          ensureFuncIsMutable();
-          func_.set(index, value);
-          onChanged();
-        } else {
-          funcBuilder_.setMessage(index, value);
-        }
-        return this;
-      }
-      /**
-       * <pre>
-       * "list(attr)"
-       * </pre>
-       *
-       * <code>repeated .tensorflow.NameAttrList func = 9;</code>
-       */
-      public Builder setFunc(
-          int index, org.tensorflow.proto.framework.NameAttrList.Builder builderForValue) {
-        if (funcBuilder_ == null) {
-          ensureFuncIsMutable();
-          func_.set(index, builderForValue.build());
-          onChanged();
-        } else {
-          funcBuilder_.setMessage(index, builderForValue.build());
-        }
-        return this;
-      }
-      /**
-       * <pre>
-       * "list(attr)"
-       * </pre>
-       *
-       * <code>repeated .tensorflow.NameAttrList func = 9;</code>
-       */
-      public Builder addFunc(org.tensorflow.proto.framework.NameAttrList value) {
-        if (funcBuilder_ == null) {
-          if (value == null) {
-            throw new NullPointerException();
-          }
-          ensureFuncIsMutable();
-          func_.add(value);
-          onChanged();
-        } else {
-          funcBuilder_.addMessage(value);
-        }
-        return this;
-      }
-      /**
-       * <pre>
-       * "list(attr)"
-       * </pre>
-       *
-       * <code>repeated .tensorflow.NameAttrList func = 9;</code>
-       */
-      public Builder addFunc(
-          int index, org.tensorflow.proto.framework.NameAttrList value) {
-        if (funcBuilder_ == null) {
-          if (value == null) {
-            throw new NullPointerException();
-          }
-          ensureFuncIsMutable();
-          func_.add(index, value);
-          onChanged();
-        } else {
-          funcBuilder_.addMessage(index, value);
-        }
-        return this;
-      }
-      /**
-       * <pre>
-       * "list(attr)"
-       * </pre>
-       *
-       * <code>repeated .tensorflow.NameAttrList func = 9;</code>
-       */
-      public Builder addFunc(
-          org.tensorflow.proto.framework.NameAttrList.Builder builderForValue) {
-        if (funcBuilder_ == null) {
-          ensureFuncIsMutable();
-          func_.add(builderForValue.build());
-          onChanged();
-        } else {
-          funcBuilder_.addMessage(builderForValue.build());
-        }
-        return this;
-      }
-      /**
-       * <pre>
-       * "list(attr)"
-       * </pre>
-       *
-       * <code>repeated .tensorflow.NameAttrList func = 9;</code>
-       */
-      public Builder addFunc(
-          int index, org.tensorflow.proto.framework.NameAttrList.Builder builderForValue) {
-        if (funcBuilder_ == null) {
-          ensureFuncIsMutable();
-          func_.add(index, builderForValue.build());
-          onChanged();
-        } else {
-          funcBuilder_.addMessage(index, builderForValue.build());
-        }
-        return this;
-      }
-      /**
-       * <pre>
-       * "list(attr)"
-       * </pre>
-       *
-       * <code>repeated .tensorflow.NameAttrList func = 9;</code>
-       */
-      public Builder addAllFunc(
-          java.lang.Iterable<? extends org.tensorflow.proto.framework.NameAttrList> values) {
-        if (funcBuilder_ == null) {
-          ensureFuncIsMutable();
-          com.google.protobuf.AbstractMessageLite.Builder.addAll(
-              values, func_);
-          onChanged();
-        } else {
-          funcBuilder_.addAllMessages(values);
-        }
-        return this;
-      }
-      /**
-       * <pre>
-       * "list(attr)"
-       * </pre>
-       *
-       * <code>repeated .tensorflow.NameAttrList func = 9;</code>
-       */
-      public Builder clearFunc() {
-        if (funcBuilder_ == null) {
-          func_ = java.util.Collections.emptyList();
-          bitField0_ = (bitField0_ & ~0x00000080);
-          onChanged();
-        } else {
-          funcBuilder_.clear();
-        }
-        return this;
-      }
-      /**
-       * <pre>
-       * "list(attr)"
-       * </pre>
-       *
-       * <code>repeated .tensorflow.NameAttrList func = 9;</code>
-       */
-      public Builder removeFunc(int index) {
-        if (funcBuilder_ == null) {
-          ensureFuncIsMutable();
-          func_.remove(index);
-          onChanged();
-        } else {
-          funcBuilder_.remove(index);
-        }
-        return this;
-      }
-      /**
-       * <pre>
-       * "list(attr)"
-       * </pre>
-       *
-       * <code>repeated .tensorflow.NameAttrList func = 9;</code>
-       */
-      public org.tensorflow.proto.framework.NameAttrList.Builder getFuncBuilder(
-          int index) {
-        return getFuncFieldBuilder().getBuilder(index);
-      }
-      /**
-       * <pre>
-       * "list(attr)"
-       * </pre>
-       *
-       * <code>repeated .tensorflow.NameAttrList func = 9;</code>
-       */
-      public org.tensorflow.proto.framework.NameAttrListOrBuilder getFuncOrBuilder(
-          int index) {
-        if (funcBuilder_ == null) {
-          return func_.get(index);  } else {
-          return funcBuilder_.getMessageOrBuilder(index);
-        }
-      }
-      /**
-       * <pre>
-       * "list(attr)"
-       * </pre>
-       *
-       * <code>repeated .tensorflow.NameAttrList func = 9;</code>
-       */
-      public java.util.List<? extends org.tensorflow.proto.framework.NameAttrListOrBuilder> 
-           getFuncOrBuilderList() {
-        if (funcBuilder_ != null) {
-          return funcBuilder_.getMessageOrBuilderList();
-        } else {
-          return java.util.Collections.unmodifiableList(func_);
-        }
-      }
-      /**
-       * <pre>
-       * "list(attr)"
-       * </pre>
-       *
-       * <code>repeated .tensorflow.NameAttrList func = 9;</code>
-       */
-      public org.tensorflow.proto.framework.NameAttrList.Builder addFuncBuilder() {
-        return getFuncFieldBuilder().addBuilder(
-            org.tensorflow.proto.framework.NameAttrList.getDefaultInstance());
-      }
-      /**
-       * <pre>
-       * "list(attr)"
-       * </pre>
-       *
-       * <code>repeated .tensorflow.NameAttrList func = 9;</code>
-       */
-      public org.tensorflow.proto.framework.NameAttrList.Builder addFuncBuilder(
-          int index) {
-        return getFuncFieldBuilder().addBuilder(
-            index, org.tensorflow.proto.framework.NameAttrList.getDefaultInstance());
-      }
-      /**
-       * <pre>
-       * "list(attr)"
-       * </pre>
-       *
-       * <code>repeated .tensorflow.NameAttrList func = 9;</code>
-       */
-      public java.util.List<org.tensorflow.proto.framework.NameAttrList.Builder> 
-           getFuncBuilderList() {
-        return getFuncFieldBuilder().getBuilderList();
-      }
-      private com.google.protobuf.RepeatedFieldBuilderV3<
-          org.tensorflow.proto.framework.NameAttrList, org.tensorflow.proto.framework.NameAttrList.Builder, org.tensorflow.proto.framework.NameAttrListOrBuilder> 
-          getFuncFieldBuilder() {
-        if (funcBuilder_ == null) {
-          funcBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
-              org.tensorflow.proto.framework.NameAttrList, org.tensorflow.proto.framework.NameAttrList.Builder, org.tensorflow.proto.framework.NameAttrListOrBuilder>(
-                  func_,
-                  ((bitField0_ & 0x00000080) != 0),
-                  getParentForChildren(),
-                  isClean());
-          func_ = null;
-        }
-        return funcBuilder_;
-      }
-      @java.lang.Override
-      public final Builder setUnknownFields(
-          final com.google.protobuf.UnknownFieldSet unknownFields) {
-        return super.setUnknownFields(unknownFields);
-      }
-
-      @java.lang.Override
-      public final Builder mergeUnknownFields(
-          final com.google.protobuf.UnknownFieldSet unknownFields) {
-        return super.mergeUnknownFields(unknownFields);
-      }
-
-
-      // @@protoc_insertion_point(builder_scope:tensorflow.AttrValue.ListValue)
-    }
-
-    // @@protoc_insertion_point(class_scope:tensorflow.AttrValue.ListValue)
-    private static final org.tensorflow.proto.framework.AttrValue.ListValue DEFAULT_INSTANCE;
-    static {
-      DEFAULT_INSTANCE = new org.tensorflow.proto.framework.AttrValue.ListValue();
-    }
-
-    public static org.tensorflow.proto.framework.AttrValue.ListValue getDefaultInstance() {
-      return DEFAULT_INSTANCE;
-    }
-
-    private static final com.google.protobuf.Parser<ListValue>
-        PARSER = new com.google.protobuf.AbstractParser<ListValue>() {
-      @java.lang.Override
-      public ListValue parsePartialFrom(
-          com.google.protobuf.CodedInputStream input,
-          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-          throws com.google.protobuf.InvalidProtocolBufferException {
-        return new ListValue(input, extensionRegistry);
-      }
-    };
-
-    public static com.google.protobuf.Parser<ListValue> parser() {
-      return PARSER;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Parser<ListValue> getParserForType() {
-      return PARSER;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.AttrValue.ListValue getDefaultInstanceForType() {
-      return DEFAULT_INSTANCE;
-    }
-
-  }
-
-  private int valueCase_ = 0;
-  private java.lang.Object value_;
-  public enum ValueCase
-      implements com.google.protobuf.Internal.EnumLite {
-    S(2),
-    I(3),
-    F(4),
-    B(5),
-    TYPE(6),
-    SHAPE(7),
-    TENSOR(8),
-    LIST(1),
-    FUNC(10),
-    PLACEHOLDER(9),
-    VALUE_NOT_SET(0);
-    private final int value;
-    private ValueCase(int value) {
-      this.value = value;
-    }
-    /**
-     * @deprecated Use {@link #forNumber(int)} instead.
-     */
-    @java.lang.Deprecated
-    public static ValueCase valueOf(int value) {
-      return forNumber(value);
-    }
-
-    public static ValueCase forNumber(int value) {
-      switch (value) {
-        case 2: return S;
-        case 3: return I;
-        case 4: return F;
-        case 5: return B;
-        case 6: return TYPE;
-        case 7: return SHAPE;
-        case 8: return TENSOR;
-        case 1: return LIST;
-        case 10: return FUNC;
-        case 9: return PLACEHOLDER;
-        case 0: return VALUE_NOT_SET;
-        default: return null;
-      }
-    }
-    public int getNumber() {
-      return this.value;
-    }
-  };
-
-  public ValueCase
-  getValueCase() {
-    return ValueCase.forNumber(
-        valueCase_);
-  }
-
-  public static final int S_FIELD_NUMBER = 2;
-  /**
-   * <pre>
-   * "string"
-   * </pre>
-   *
-   * <code>bytes s = 2;</code>
-   */
-  public com.google.protobuf.ByteString getS() {
-    if (valueCase_ == 2) {
-      return (com.google.protobuf.ByteString) value_;
-    }
-    return com.google.protobuf.ByteString.EMPTY;
-  }
-
-  public static final int I_FIELD_NUMBER = 3;
-  /**
-   * <pre>
-   * "int"
-   * </pre>
-   *
-   * <code>int64 i = 3;</code>
-   */
-  public long getI() {
-    if (valueCase_ == 3) {
-      return (java.lang.Long) value_;
-    }
-    return 0L;
-  }
-
-  public static final int F_FIELD_NUMBER = 4;
-  /**
-   * <pre>
-   * "float"
-   * </pre>
-   *
-   * <code>float f = 4;</code>
-   */
-  public float getF() {
-    if (valueCase_ == 4) {
-      return (java.lang.Float) value_;
-    }
-    return 0F;
-  }
-
-  public static final int B_FIELD_NUMBER = 5;
-  /**
-   * <pre>
-   * "bool"
-   * </pre>
-   *
-   * <code>bool b = 5;</code>
-   */
-  public boolean getB() {
-    if (valueCase_ == 5) {
-      return (java.lang.Boolean) value_;
-    }
-    return false;
-  }
-
-  public static final int TYPE_FIELD_NUMBER = 6;
-  /**
-   * <pre>
-   * "type"
-   * </pre>
-   *
-   * <code>.tensorflow.DataType type = 6;</code>
-   */
-  public int getTypeValue() {
-    if (valueCase_ == 6) {
-      return (java.lang.Integer) value_;
-    }
-    return 0;
-  }
-  /**
-   * <pre>
-   * "type"
-   * </pre>
-   *
-   * <code>.tensorflow.DataType type = 6;</code>
-   */
-  public org.tensorflow.proto.framework.DataType getType() {
-    if (valueCase_ == 6) {
-      @SuppressWarnings("deprecation")
-      org.tensorflow.proto.framework.DataType result = org.tensorflow.proto.framework.DataType.valueOf(
-          (java.lang.Integer) value_);
-      return result == null ? org.tensorflow.proto.framework.DataType.UNRECOGNIZED : result;
-    }
-    return org.tensorflow.proto.framework.DataType.DT_INVALID;
-  }
-
-  public static final int SHAPE_FIELD_NUMBER = 7;
-  /**
-   * <pre>
-   * "shape"
-   * </pre>
-   *
-   * <code>.tensorflow.TensorShapeProto shape = 7;</code>
-   */
-  public boolean hasShape() {
-    return valueCase_ == 7;
-  }
-  /**
-   * <pre>
-   * "shape"
-   * </pre>
-   *
-   * <code>.tensorflow.TensorShapeProto shape = 7;</code>
-   */
-  public org.tensorflow.proto.framework.TensorShapeProto getShape() {
-    if (valueCase_ == 7) {
-       return (org.tensorflow.proto.framework.TensorShapeProto) value_;
-    }
-    return org.tensorflow.proto.framework.TensorShapeProto.getDefaultInstance();
-  }
-  /**
-   * <pre>
-   * "shape"
-   * </pre>
-   *
-   * <code>.tensorflow.TensorShapeProto shape = 7;</code>
-   */
-  public org.tensorflow.proto.framework.TensorShapeProtoOrBuilder getShapeOrBuilder() {
-    if (valueCase_ == 7) {
-       return (org.tensorflow.proto.framework.TensorShapeProto) value_;
-    }
-    return org.tensorflow.proto.framework.TensorShapeProto.getDefaultInstance();
-  }
-
-  public static final int TENSOR_FIELD_NUMBER = 8;
-  /**
-   * <pre>
-   * "tensor"
-   * </pre>
-   *
-   * <code>.tensorflow.TensorProto tensor = 8;</code>
-   */
-  public boolean hasTensor() {
-    return valueCase_ == 8;
-  }
-  /**
-   * <pre>
-   * "tensor"
-   * </pre>
-   *
-   * <code>.tensorflow.TensorProto tensor = 8;</code>
-   */
-  public org.tensorflow.proto.framework.TensorProto getTensor() {
-    if (valueCase_ == 8) {
-       return (org.tensorflow.proto.framework.TensorProto) value_;
-    }
-    return org.tensorflow.proto.framework.TensorProto.getDefaultInstance();
-  }
-  /**
-   * <pre>
-   * "tensor"
-   * </pre>
-   *
-   * <code>.tensorflow.TensorProto tensor = 8;</code>
-   */
-  public org.tensorflow.proto.framework.TensorProtoOrBuilder getTensorOrBuilder() {
-    if (valueCase_ == 8) {
-       return (org.tensorflow.proto.framework.TensorProto) value_;
-    }
-    return org.tensorflow.proto.framework.TensorProto.getDefaultInstance();
-  }
-
-  public static final int LIST_FIELD_NUMBER = 1;
-  /**
-   * <pre>
-   * any "list(...)"
-   * </pre>
-   *
-   * <code>.tensorflow.AttrValue.ListValue list = 1;</code>
-   */
-  public boolean hasList() {
-    return valueCase_ == 1;
-  }
-  /**
-   * <pre>
-   * any "list(...)"
-   * </pre>
-   *
-   * <code>.tensorflow.AttrValue.ListValue list = 1;</code>
-   */
-  public org.tensorflow.proto.framework.AttrValue.ListValue getList() {
-    if (valueCase_ == 1) {
-       return (org.tensorflow.proto.framework.AttrValue.ListValue) value_;
-    }
-    return org.tensorflow.proto.framework.AttrValue.ListValue.getDefaultInstance();
-  }
-  /**
-   * <pre>
-   * any "list(...)"
-   * </pre>
-   *
-   * <code>.tensorflow.AttrValue.ListValue list = 1;</code>
-   */
-  public org.tensorflow.proto.framework.AttrValue.ListValueOrBuilder getListOrBuilder() {
-    if (valueCase_ == 1) {
-       return (org.tensorflow.proto.framework.AttrValue.ListValue) value_;
-    }
-    return org.tensorflow.proto.framework.AttrValue.ListValue.getDefaultInstance();
-  }
-
-  public static final int FUNC_FIELD_NUMBER = 10;
-  /**
-   * <pre>
-   * "func" represents a function. func.name is a function's name or
-   * a primitive op's name. func.attr.first is the name of an attr
-   * defined for that function. func.attr.second is the value for
-   * that attr in the instantiation.
-   * </pre>
-   *
-   * <code>.tensorflow.NameAttrList func = 10;</code>
-   */
-  public boolean hasFunc() {
-    return valueCase_ == 10;
-  }
-  /**
-   * <pre>
-   * "func" represents a function. func.name is a function's name or
-   * a primitive op's name. func.attr.first is the name of an attr
-   * defined for that function. func.attr.second is the value for
-   * that attr in the instantiation.
-   * </pre>
-   *
-   * <code>.tensorflow.NameAttrList func = 10;</code>
-   */
-  public org.tensorflow.proto.framework.NameAttrList getFunc() {
-    if (valueCase_ == 10) {
-       return (org.tensorflow.proto.framework.NameAttrList) value_;
-    }
-    return org.tensorflow.proto.framework.NameAttrList.getDefaultInstance();
-  }
-  /**
-   * <pre>
-   * "func" represents a function. func.name is a function's name or
-   * a primitive op's name. func.attr.first is the name of an attr
-   * defined for that function. func.attr.second is the value for
-   * that attr in the instantiation.
-   * </pre>
-   *
-   * <code>.tensorflow.NameAttrList func = 10;</code>
-   */
-  public org.tensorflow.proto.framework.NameAttrListOrBuilder getFuncOrBuilder() {
-    if (valueCase_ == 10) {
-       return (org.tensorflow.proto.framework.NameAttrList) value_;
-    }
-    return org.tensorflow.proto.framework.NameAttrList.getDefaultInstance();
-  }
-
-  public static final int PLACEHOLDER_FIELD_NUMBER = 9;
-  /**
-   * <pre>
-   * This is a placeholder only used in nodes defined inside a
-   * function.  It indicates the attr value will be supplied when
-   * the function is instantiated.  For example, let us suppose a
-   * node "N" in function "FN". "N" has an attr "A" with value
-   * placeholder = "foo". When FN is instantiated with attr "foo"
-   * set to "bar", the instantiated node N's attr A will have been
-   * given the value "bar".
-   * </pre>
-   *
-   * <code>string placeholder = 9;</code>
-   */
-  public java.lang.String getPlaceholder() {
-    java.lang.Object ref = "";
-    if (valueCase_ == 9) {
-      ref = value_;
-    }
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      if (valueCase_ == 9) {
-        value_ = s;
-      }
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * This is a placeholder only used in nodes defined inside a
-   * function.  It indicates the attr value will be supplied when
-   * the function is instantiated.  For example, let us suppose a
-   * node "N" in function "FN". "N" has an attr "A" with value
-   * placeholder = "foo". When FN is instantiated with attr "foo"
-   * set to "bar", the instantiated node N's attr A will have been
-   * given the value "bar".
-   * </pre>
-   *
-   * <code>string placeholder = 9;</code>
-   */
-  public com.google.protobuf.ByteString
-      getPlaceholderBytes() {
-    java.lang.Object ref = "";
-    if (valueCase_ == 9) {
-      ref = value_;
-    }
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      if (valueCase_ == 9) {
-        value_ = b;
-      }
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    if (valueCase_ == 1) {
-      output.writeMessage(1, (org.tensorflow.proto.framework.AttrValue.ListValue) value_);
-    }
-    if (valueCase_ == 2) {
-      output.writeBytes(
-          2, (com.google.protobuf.ByteString) value_);
-    }
-    if (valueCase_ == 3) {
-      output.writeInt64(
-          3, (long)((java.lang.Long) value_));
-    }
-    if (valueCase_ == 4) {
-      output.writeFloat(
-          4, (float)((java.lang.Float) value_));
-    }
-    if (valueCase_ == 5) {
-      output.writeBool(
-          5, (boolean)((java.lang.Boolean) value_));
-    }
-    if (valueCase_ == 6) {
-      output.writeEnum(6, ((java.lang.Integer) value_));
-    }
-    if (valueCase_ == 7) {
-      output.writeMessage(7, (org.tensorflow.proto.framework.TensorShapeProto) value_);
-    }
-    if (valueCase_ == 8) {
-      output.writeMessage(8, (org.tensorflow.proto.framework.TensorProto) value_);
-    }
-    if (valueCase_ == 9) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 9, value_);
-    }
-    if (valueCase_ == 10) {
-      output.writeMessage(10, (org.tensorflow.proto.framework.NameAttrList) value_);
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (valueCase_ == 1) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(1, (org.tensorflow.proto.framework.AttrValue.ListValue) value_);
-    }
-    if (valueCase_ == 2) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeBytesSize(
-            2, (com.google.protobuf.ByteString) value_);
-    }
-    if (valueCase_ == 3) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt64Size(
-            3, (long)((java.lang.Long) value_));
-    }
-    if (valueCase_ == 4) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeFloatSize(
-            4, (float)((java.lang.Float) value_));
-    }
-    if (valueCase_ == 5) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeBoolSize(
-            5, (boolean)((java.lang.Boolean) value_));
-    }
-    if (valueCase_ == 6) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeEnumSize(6, ((java.lang.Integer) value_));
-    }
-    if (valueCase_ == 7) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(7, (org.tensorflow.proto.framework.TensorShapeProto) value_);
-    }
-    if (valueCase_ == 8) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(8, (org.tensorflow.proto.framework.TensorProto) value_);
-    }
-    if (valueCase_ == 9) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(9, value_);
-    }
-    if (valueCase_ == 10) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(10, (org.tensorflow.proto.framework.NameAttrList) value_);
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.framework.AttrValue)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.framework.AttrValue other = (org.tensorflow.proto.framework.AttrValue) obj;
-
-    if (!getValueCase().equals(other.getValueCase())) return false;
-    switch (valueCase_) {
-      case 2:
-        if (!getS()
-            .equals(other.getS())) return false;
-        break;
-      case 3:
-        if (getI()
-            != other.getI()) return false;
-        break;
-      case 4:
-        if (java.lang.Float.floatToIntBits(getF())
-            != java.lang.Float.floatToIntBits(
-                other.getF())) return false;
-        break;
-      case 5:
-        if (getB()
-            != other.getB()) return false;
-        break;
-      case 6:
-        if (getTypeValue()
-            != other.getTypeValue()) return false;
-        break;
-      case 7:
-        if (!getShape()
-            .equals(other.getShape())) return false;
-        break;
-      case 8:
-        if (!getTensor()
-            .equals(other.getTensor())) return false;
-        break;
-      case 1:
-        if (!getList()
-            .equals(other.getList())) return false;
-        break;
-      case 10:
-        if (!getFunc()
-            .equals(other.getFunc())) return false;
-        break;
-      case 9:
-        if (!getPlaceholder()
-            .equals(other.getPlaceholder())) return false;
-        break;
-      case 0:
-      default:
-    }
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    switch (valueCase_) {
-      case 2:
-        hash = (37 * hash) + S_FIELD_NUMBER;
-        hash = (53 * hash) + getS().hashCode();
-        break;
-      case 3:
-        hash = (37 * hash) + I_FIELD_NUMBER;
-        hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-            getI());
-        break;
-      case 4:
-        hash = (37 * hash) + F_FIELD_NUMBER;
-        hash = (53 * hash) + java.lang.Float.floatToIntBits(
-            getF());
-        break;
-      case 5:
-        hash = (37 * hash) + B_FIELD_NUMBER;
-        hash = (53 * hash) + com.google.protobuf.Internal.hashBoolean(
-            getB());
-        break;
-      case 6:
-        hash = (37 * hash) + TYPE_FIELD_NUMBER;
-        hash = (53 * hash) + getTypeValue();
-        break;
-      case 7:
-        hash = (37 * hash) + SHAPE_FIELD_NUMBER;
-        hash = (53 * hash) + getShape().hashCode();
-        break;
-      case 8:
-        hash = (37 * hash) + TENSOR_FIELD_NUMBER;
-        hash = (53 * hash) + getTensor().hashCode();
-        break;
-      case 1:
-        hash = (37 * hash) + LIST_FIELD_NUMBER;
-        hash = (53 * hash) + getList().hashCode();
-        break;
-      case 10:
-        hash = (37 * hash) + FUNC_FIELD_NUMBER;
-        hash = (53 * hash) + getFunc().hashCode();
-        break;
-      case 9:
-        hash = (37 * hash) + PLACEHOLDER_FIELD_NUMBER;
-        hash = (53 * hash) + getPlaceholder().hashCode();
-        break;
-      case 0:
-      default:
-    }
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.framework.AttrValue parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.AttrValue parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.AttrValue parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.AttrValue parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.AttrValue parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.AttrValue parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.AttrValue parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.AttrValue parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.AttrValue parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.AttrValue parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.AttrValue parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.AttrValue parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.framework.AttrValue prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * <pre>
-   * Protocol buffer representing the value for an attr used to configure an Op.
-   * Comment indicates the corresponding attr type.  Only the field matching the
-   * attr type may be filled.
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.AttrValue}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.AttrValue)
-      org.tensorflow.proto.framework.AttrValueOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.AttrValueProtos.internal_static_tensorflow_AttrValue_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.AttrValueProtos.internal_static_tensorflow_AttrValue_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.AttrValue.class, org.tensorflow.proto.framework.AttrValue.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.framework.AttrValue.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      valueCase_ = 0;
-      value_ = null;
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.framework.AttrValueProtos.internal_static_tensorflow_AttrValue_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.AttrValue getDefaultInstanceForType() {
-      return org.tensorflow.proto.framework.AttrValue.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.AttrValue build() {
-      org.tensorflow.proto.framework.AttrValue result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.AttrValue buildPartial() {
-      org.tensorflow.proto.framework.AttrValue result = new org.tensorflow.proto.framework.AttrValue(this);
-      if (valueCase_ == 2) {
-        result.value_ = value_;
-      }
-      if (valueCase_ == 3) {
-        result.value_ = value_;
-      }
-      if (valueCase_ == 4) {
-        result.value_ = value_;
-      }
-      if (valueCase_ == 5) {
-        result.value_ = value_;
-      }
-      if (valueCase_ == 6) {
-        result.value_ = value_;
-      }
-      if (valueCase_ == 7) {
-        if (shapeBuilder_ == null) {
-          result.value_ = value_;
-        } else {
-          result.value_ = shapeBuilder_.build();
-        }
-      }
-      if (valueCase_ == 8) {
-        if (tensorBuilder_ == null) {
-          result.value_ = value_;
-        } else {
-          result.value_ = tensorBuilder_.build();
-        }
-      }
-      if (valueCase_ == 1) {
-        if (listBuilder_ == null) {
-          result.value_ = value_;
-        } else {
-          result.value_ = listBuilder_.build();
-        }
-      }
-      if (valueCase_ == 10) {
-        if (funcBuilder_ == null) {
-          result.value_ = value_;
-        } else {
-          result.value_ = funcBuilder_.build();
-        }
-      }
-      if (valueCase_ == 9) {
-        result.value_ = value_;
-      }
-      result.valueCase_ = valueCase_;
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.framework.AttrValue) {
-        return mergeFrom((org.tensorflow.proto.framework.AttrValue)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.framework.AttrValue other) {
-      if (other == org.tensorflow.proto.framework.AttrValue.getDefaultInstance()) return this;
-      switch (other.getValueCase()) {
-        case S: {
-          setS(other.getS());
-          break;
-        }
-        case I: {
-          setI(other.getI());
-          break;
-        }
-        case F: {
-          setF(other.getF());
-          break;
-        }
-        case B: {
-          setB(other.getB());
-          break;
-        }
-        case TYPE: {
-          setTypeValue(other.getTypeValue());
-          break;
-        }
-        case SHAPE: {
-          mergeShape(other.getShape());
-          break;
-        }
-        case TENSOR: {
-          mergeTensor(other.getTensor());
-          break;
-        }
-        case LIST: {
-          mergeList(other.getList());
-          break;
-        }
-        case FUNC: {
-          mergeFunc(other.getFunc());
-          break;
-        }
-        case PLACEHOLDER: {
-          valueCase_ = 9;
-          value_ = other.value_;
-          onChanged();
-          break;
-        }
-        case VALUE_NOT_SET: {
-          break;
-        }
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.framework.AttrValue parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.framework.AttrValue) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-    private int valueCase_ = 0;
-    private java.lang.Object value_;
-    public ValueCase
-        getValueCase() {
-      return ValueCase.forNumber(
-          valueCase_);
-    }
-
-    public Builder clearValue() {
-      valueCase_ = 0;
-      value_ = null;
-      onChanged();
-      return this;
-    }
-
-
-    /**
-     * <pre>
-     * "string"
-     * </pre>
-     *
-     * <code>bytes s = 2;</code>
-     */
-    public com.google.protobuf.ByteString getS() {
-      if (valueCase_ == 2) {
-        return (com.google.protobuf.ByteString) value_;
-      }
-      return com.google.protobuf.ByteString.EMPTY;
-    }
-    /**
-     * <pre>
-     * "string"
-     * </pre>
-     *
-     * <code>bytes s = 2;</code>
-     */
-    public Builder setS(com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  valueCase_ = 2;
-      value_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * "string"
-     * </pre>
-     *
-     * <code>bytes s = 2;</code>
-     */
-    public Builder clearS() {
-      if (valueCase_ == 2) {
-        valueCase_ = 0;
-        value_ = null;
-        onChanged();
-      }
-      return this;
-    }
-
-    /**
-     * <pre>
-     * "int"
-     * </pre>
-     *
-     * <code>int64 i = 3;</code>
-     */
-    public long getI() {
-      if (valueCase_ == 3) {
-        return (java.lang.Long) value_;
-      }
-      return 0L;
-    }
-    /**
-     * <pre>
-     * "int"
-     * </pre>
-     *
-     * <code>int64 i = 3;</code>
-     */
-    public Builder setI(long value) {
-      valueCase_ = 3;
-      value_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * "int"
-     * </pre>
-     *
-     * <code>int64 i = 3;</code>
-     */
-    public Builder clearI() {
-      if (valueCase_ == 3) {
-        valueCase_ = 0;
-        value_ = null;
-        onChanged();
-      }
-      return this;
-    }
-
-    /**
-     * <pre>
-     * "float"
-     * </pre>
-     *
-     * <code>float f = 4;</code>
-     */
-    public float getF() {
-      if (valueCase_ == 4) {
-        return (java.lang.Float) value_;
-      }
-      return 0F;
-    }
-    /**
-     * <pre>
-     * "float"
-     * </pre>
-     *
-     * <code>float f = 4;</code>
-     */
-    public Builder setF(float value) {
-      valueCase_ = 4;
-      value_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * "float"
-     * </pre>
-     *
-     * <code>float f = 4;</code>
-     */
-    public Builder clearF() {
-      if (valueCase_ == 4) {
-        valueCase_ = 0;
-        value_ = null;
-        onChanged();
-      }
-      return this;
-    }
-
-    /**
-     * <pre>
-     * "bool"
-     * </pre>
-     *
-     * <code>bool b = 5;</code>
-     */
-    public boolean getB() {
-      if (valueCase_ == 5) {
-        return (java.lang.Boolean) value_;
-      }
-      return false;
-    }
-    /**
-     * <pre>
-     * "bool"
-     * </pre>
-     *
-     * <code>bool b = 5;</code>
-     */
-    public Builder setB(boolean value) {
-      valueCase_ = 5;
-      value_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * "bool"
-     * </pre>
-     *
-     * <code>bool b = 5;</code>
-     */
-    public Builder clearB() {
-      if (valueCase_ == 5) {
-        valueCase_ = 0;
-        value_ = null;
-        onChanged();
-      }
-      return this;
-    }
-
-    /**
-     * <pre>
-     * "type"
-     * </pre>
-     *
-     * <code>.tensorflow.DataType type = 6;</code>
-     */
-    public int getTypeValue() {
-      if (valueCase_ == 6) {
-        return ((java.lang.Integer) value_).intValue();
-      }
-      return 0;
-    }
-    /**
-     * <pre>
-     * "type"
-     * </pre>
-     *
-     * <code>.tensorflow.DataType type = 6;</code>
-     */
-    public Builder setTypeValue(int value) {
-      valueCase_ = 6;
-      value_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * "type"
-     * </pre>
-     *
-     * <code>.tensorflow.DataType type = 6;</code>
-     */
-    public org.tensorflow.proto.framework.DataType getType() {
-      if (valueCase_ == 6) {
-        @SuppressWarnings("deprecation")
-        org.tensorflow.proto.framework.DataType result = org.tensorflow.proto.framework.DataType.valueOf(
-            (java.lang.Integer) value_);
-        return result == null ? org.tensorflow.proto.framework.DataType.UNRECOGNIZED : result;
-      }
-      return org.tensorflow.proto.framework.DataType.DT_INVALID;
-    }
-    /**
-     * <pre>
-     * "type"
-     * </pre>
-     *
-     * <code>.tensorflow.DataType type = 6;</code>
-     */
-    public Builder setType(org.tensorflow.proto.framework.DataType value) {
-      if (value == null) {
-        throw new NullPointerException();
-      }
-      valueCase_ = 6;
-      value_ = value.getNumber();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * "type"
-     * </pre>
-     *
-     * <code>.tensorflow.DataType type = 6;</code>
-     */
-    public Builder clearType() {
-      if (valueCase_ == 6) {
-        valueCase_ = 0;
-        value_ = null;
-        onChanged();
-      }
-      return this;
-    }
-
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.TensorShapeProto, org.tensorflow.proto.framework.TensorShapeProto.Builder, org.tensorflow.proto.framework.TensorShapeProtoOrBuilder> shapeBuilder_;
-    /**
-     * <pre>
-     * "shape"
-     * </pre>
-     *
-     * <code>.tensorflow.TensorShapeProto shape = 7;</code>
-     */
-    public boolean hasShape() {
-      return valueCase_ == 7;
-    }
-    /**
-     * <pre>
-     * "shape"
-     * </pre>
-     *
-     * <code>.tensorflow.TensorShapeProto shape = 7;</code>
-     */
-    public org.tensorflow.proto.framework.TensorShapeProto getShape() {
-      if (shapeBuilder_ == null) {
-        if (valueCase_ == 7) {
-          return (org.tensorflow.proto.framework.TensorShapeProto) value_;
-        }
-        return org.tensorflow.proto.framework.TensorShapeProto.getDefaultInstance();
-      } else {
-        if (valueCase_ == 7) {
-          return shapeBuilder_.getMessage();
-        }
-        return org.tensorflow.proto.framework.TensorShapeProto.getDefaultInstance();
-      }
-    }
-    /**
-     * <pre>
-     * "shape"
-     * </pre>
-     *
-     * <code>.tensorflow.TensorShapeProto shape = 7;</code>
-     */
-    public Builder setShape(org.tensorflow.proto.framework.TensorShapeProto value) {
-      if (shapeBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        value_ = value;
-        onChanged();
-      } else {
-        shapeBuilder_.setMessage(value);
-      }
-      valueCase_ = 7;
-      return this;
-    }
-    /**
-     * <pre>
-     * "shape"
-     * </pre>
-     *
-     * <code>.tensorflow.TensorShapeProto shape = 7;</code>
-     */
-    public Builder setShape(
-        org.tensorflow.proto.framework.TensorShapeProto.Builder builderForValue) {
-      if (shapeBuilder_ == null) {
-        value_ = builderForValue.build();
-        onChanged();
-      } else {
-        shapeBuilder_.setMessage(builderForValue.build());
-      }
-      valueCase_ = 7;
-      return this;
-    }
-    /**
-     * <pre>
-     * "shape"
-     * </pre>
-     *
-     * <code>.tensorflow.TensorShapeProto shape = 7;</code>
-     */
-    public Builder mergeShape(org.tensorflow.proto.framework.TensorShapeProto value) {
-      if (shapeBuilder_ == null) {
-        if (valueCase_ == 7 &&
-            value_ != org.tensorflow.proto.framework.TensorShapeProto.getDefaultInstance()) {
-          value_ = org.tensorflow.proto.framework.TensorShapeProto.newBuilder((org.tensorflow.proto.framework.TensorShapeProto) value_)
-              .mergeFrom(value).buildPartial();
-        } else {
-          value_ = value;
-        }
-        onChanged();
-      } else {
-        if (valueCase_ == 7) {
-          shapeBuilder_.mergeFrom(value);
-        }
-        shapeBuilder_.setMessage(value);
-      }
-      valueCase_ = 7;
-      return this;
-    }
-    /**
-     * <pre>
-     * "shape"
-     * </pre>
-     *
-     * <code>.tensorflow.TensorShapeProto shape = 7;</code>
-     */
-    public Builder clearShape() {
-      if (shapeBuilder_ == null) {
-        if (valueCase_ == 7) {
-          valueCase_ = 0;
-          value_ = null;
-          onChanged();
-        }
-      } else {
-        if (valueCase_ == 7) {
-          valueCase_ = 0;
-          value_ = null;
-        }
-        shapeBuilder_.clear();
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * "shape"
-     * </pre>
-     *
-     * <code>.tensorflow.TensorShapeProto shape = 7;</code>
-     */
-    public org.tensorflow.proto.framework.TensorShapeProto.Builder getShapeBuilder() {
-      return getShapeFieldBuilder().getBuilder();
-    }
-    /**
-     * <pre>
-     * "shape"
-     * </pre>
-     *
-     * <code>.tensorflow.TensorShapeProto shape = 7;</code>
-     */
-    public org.tensorflow.proto.framework.TensorShapeProtoOrBuilder getShapeOrBuilder() {
-      if ((valueCase_ == 7) && (shapeBuilder_ != null)) {
-        return shapeBuilder_.getMessageOrBuilder();
-      } else {
-        if (valueCase_ == 7) {
-          return (org.tensorflow.proto.framework.TensorShapeProto) value_;
-        }
-        return org.tensorflow.proto.framework.TensorShapeProto.getDefaultInstance();
-      }
-    }
-    /**
-     * <pre>
-     * "shape"
-     * </pre>
-     *
-     * <code>.tensorflow.TensorShapeProto shape = 7;</code>
-     */
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.TensorShapeProto, org.tensorflow.proto.framework.TensorShapeProto.Builder, org.tensorflow.proto.framework.TensorShapeProtoOrBuilder> 
-        getShapeFieldBuilder() {
-      if (shapeBuilder_ == null) {
-        if (!(valueCase_ == 7)) {
-          value_ = org.tensorflow.proto.framework.TensorShapeProto.getDefaultInstance();
-        }
-        shapeBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-            org.tensorflow.proto.framework.TensorShapeProto, org.tensorflow.proto.framework.TensorShapeProto.Builder, org.tensorflow.proto.framework.TensorShapeProtoOrBuilder>(
-                (org.tensorflow.proto.framework.TensorShapeProto) value_,
-                getParentForChildren(),
-                isClean());
-        value_ = null;
-      }
-      valueCase_ = 7;
-      onChanged();;
-      return shapeBuilder_;
-    }
-
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.TensorProto, org.tensorflow.proto.framework.TensorProto.Builder, org.tensorflow.proto.framework.TensorProtoOrBuilder> tensorBuilder_;
-    /**
-     * <pre>
-     * "tensor"
-     * </pre>
-     *
-     * <code>.tensorflow.TensorProto tensor = 8;</code>
-     */
-    public boolean hasTensor() {
-      return valueCase_ == 8;
-    }
-    /**
-     * <pre>
-     * "tensor"
-     * </pre>
-     *
-     * <code>.tensorflow.TensorProto tensor = 8;</code>
-     */
-    public org.tensorflow.proto.framework.TensorProto getTensor() {
-      if (tensorBuilder_ == null) {
-        if (valueCase_ == 8) {
-          return (org.tensorflow.proto.framework.TensorProto) value_;
-        }
-        return org.tensorflow.proto.framework.TensorProto.getDefaultInstance();
-      } else {
-        if (valueCase_ == 8) {
-          return tensorBuilder_.getMessage();
-        }
-        return org.tensorflow.proto.framework.TensorProto.getDefaultInstance();
-      }
-    }
-    /**
-     * <pre>
-     * "tensor"
-     * </pre>
-     *
-     * <code>.tensorflow.TensorProto tensor = 8;</code>
-     */
-    public Builder setTensor(org.tensorflow.proto.framework.TensorProto value) {
-      if (tensorBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        value_ = value;
-        onChanged();
-      } else {
-        tensorBuilder_.setMessage(value);
-      }
-      valueCase_ = 8;
-      return this;
-    }
-    /**
-     * <pre>
-     * "tensor"
-     * </pre>
-     *
-     * <code>.tensorflow.TensorProto tensor = 8;</code>
-     */
-    public Builder setTensor(
-        org.tensorflow.proto.framework.TensorProto.Builder builderForValue) {
-      if (tensorBuilder_ == null) {
-        value_ = builderForValue.build();
-        onChanged();
-      } else {
-        tensorBuilder_.setMessage(builderForValue.build());
-      }
-      valueCase_ = 8;
-      return this;
-    }
-    /**
-     * <pre>
-     * "tensor"
-     * </pre>
-     *
-     * <code>.tensorflow.TensorProto tensor = 8;</code>
-     */
-    public Builder mergeTensor(org.tensorflow.proto.framework.TensorProto value) {
-      if (tensorBuilder_ == null) {
-        if (valueCase_ == 8 &&
-            value_ != org.tensorflow.proto.framework.TensorProto.getDefaultInstance()) {
-          value_ = org.tensorflow.proto.framework.TensorProto.newBuilder((org.tensorflow.proto.framework.TensorProto) value_)
-              .mergeFrom(value).buildPartial();
-        } else {
-          value_ = value;
-        }
-        onChanged();
-      } else {
-        if (valueCase_ == 8) {
-          tensorBuilder_.mergeFrom(value);
-        }
-        tensorBuilder_.setMessage(value);
-      }
-      valueCase_ = 8;
-      return this;
-    }
-    /**
-     * <pre>
-     * "tensor"
-     * </pre>
-     *
-     * <code>.tensorflow.TensorProto tensor = 8;</code>
-     */
-    public Builder clearTensor() {
-      if (tensorBuilder_ == null) {
-        if (valueCase_ == 8) {
-          valueCase_ = 0;
-          value_ = null;
-          onChanged();
-        }
-      } else {
-        if (valueCase_ == 8) {
-          valueCase_ = 0;
-          value_ = null;
-        }
-        tensorBuilder_.clear();
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * "tensor"
-     * </pre>
-     *
-     * <code>.tensorflow.TensorProto tensor = 8;</code>
-     */
-    public org.tensorflow.proto.framework.TensorProto.Builder getTensorBuilder() {
-      return getTensorFieldBuilder().getBuilder();
-    }
-    /**
-     * <pre>
-     * "tensor"
-     * </pre>
-     *
-     * <code>.tensorflow.TensorProto tensor = 8;</code>
-     */
-    public org.tensorflow.proto.framework.TensorProtoOrBuilder getTensorOrBuilder() {
-      if ((valueCase_ == 8) && (tensorBuilder_ != null)) {
-        return tensorBuilder_.getMessageOrBuilder();
-      } else {
-        if (valueCase_ == 8) {
-          return (org.tensorflow.proto.framework.TensorProto) value_;
-        }
-        return org.tensorflow.proto.framework.TensorProto.getDefaultInstance();
-      }
-    }
-    /**
-     * <pre>
-     * "tensor"
-     * </pre>
-     *
-     * <code>.tensorflow.TensorProto tensor = 8;</code>
-     */
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.TensorProto, org.tensorflow.proto.framework.TensorProto.Builder, org.tensorflow.proto.framework.TensorProtoOrBuilder> 
-        getTensorFieldBuilder() {
-      if (tensorBuilder_ == null) {
-        if (!(valueCase_ == 8)) {
-          value_ = org.tensorflow.proto.framework.TensorProto.getDefaultInstance();
-        }
-        tensorBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-            org.tensorflow.proto.framework.TensorProto, org.tensorflow.proto.framework.TensorProto.Builder, org.tensorflow.proto.framework.TensorProtoOrBuilder>(
-                (org.tensorflow.proto.framework.TensorProto) value_,
-                getParentForChildren(),
-                isClean());
-        value_ = null;
-      }
-      valueCase_ = 8;
-      onChanged();;
-      return tensorBuilder_;
-    }
-
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.AttrValue.ListValue, org.tensorflow.proto.framework.AttrValue.ListValue.Builder, org.tensorflow.proto.framework.AttrValue.ListValueOrBuilder> listBuilder_;
-    /**
-     * <pre>
-     * any "list(...)"
-     * </pre>
-     *
-     * <code>.tensorflow.AttrValue.ListValue list = 1;</code>
-     */
-    public boolean hasList() {
-      return valueCase_ == 1;
-    }
-    /**
-     * <pre>
-     * any "list(...)"
-     * </pre>
-     *
-     * <code>.tensorflow.AttrValue.ListValue list = 1;</code>
-     */
-    public org.tensorflow.proto.framework.AttrValue.ListValue getList() {
-      if (listBuilder_ == null) {
-        if (valueCase_ == 1) {
-          return (org.tensorflow.proto.framework.AttrValue.ListValue) value_;
-        }
-        return org.tensorflow.proto.framework.AttrValue.ListValue.getDefaultInstance();
-      } else {
-        if (valueCase_ == 1) {
-          return listBuilder_.getMessage();
-        }
-        return org.tensorflow.proto.framework.AttrValue.ListValue.getDefaultInstance();
-      }
-    }
-    /**
-     * <pre>
-     * any "list(...)"
-     * </pre>
-     *
-     * <code>.tensorflow.AttrValue.ListValue list = 1;</code>
-     */
-    public Builder setList(org.tensorflow.proto.framework.AttrValue.ListValue value) {
-      if (listBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        value_ = value;
-        onChanged();
-      } else {
-        listBuilder_.setMessage(value);
-      }
-      valueCase_ = 1;
-      return this;
-    }
-    /**
-     * <pre>
-     * any "list(...)"
-     * </pre>
-     *
-     * <code>.tensorflow.AttrValue.ListValue list = 1;</code>
-     */
-    public Builder setList(
-        org.tensorflow.proto.framework.AttrValue.ListValue.Builder builderForValue) {
-      if (listBuilder_ == null) {
-        value_ = builderForValue.build();
-        onChanged();
-      } else {
-        listBuilder_.setMessage(builderForValue.build());
-      }
-      valueCase_ = 1;
-      return this;
-    }
-    /**
-     * <pre>
-     * any "list(...)"
-     * </pre>
-     *
-     * <code>.tensorflow.AttrValue.ListValue list = 1;</code>
-     */
-    public Builder mergeList(org.tensorflow.proto.framework.AttrValue.ListValue value) {
-      if (listBuilder_ == null) {
-        if (valueCase_ == 1 &&
-            value_ != org.tensorflow.proto.framework.AttrValue.ListValue.getDefaultInstance()) {
-          value_ = org.tensorflow.proto.framework.AttrValue.ListValue.newBuilder((org.tensorflow.proto.framework.AttrValue.ListValue) value_)
-              .mergeFrom(value).buildPartial();
-        } else {
-          value_ = value;
-        }
-        onChanged();
-      } else {
-        if (valueCase_ == 1) {
-          listBuilder_.mergeFrom(value);
-        }
-        listBuilder_.setMessage(value);
-      }
-      valueCase_ = 1;
-      return this;
-    }
-    /**
-     * <pre>
-     * any "list(...)"
-     * </pre>
-     *
-     * <code>.tensorflow.AttrValue.ListValue list = 1;</code>
-     */
-    public Builder clearList() {
-      if (listBuilder_ == null) {
-        if (valueCase_ == 1) {
-          valueCase_ = 0;
-          value_ = null;
-          onChanged();
-        }
-      } else {
-        if (valueCase_ == 1) {
-          valueCase_ = 0;
-          value_ = null;
-        }
-        listBuilder_.clear();
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * any "list(...)"
-     * </pre>
-     *
-     * <code>.tensorflow.AttrValue.ListValue list = 1;</code>
-     */
-    public org.tensorflow.proto.framework.AttrValue.ListValue.Builder getListBuilder() {
-      return getListFieldBuilder().getBuilder();
-    }
-    /**
-     * <pre>
-     * any "list(...)"
-     * </pre>
-     *
-     * <code>.tensorflow.AttrValue.ListValue list = 1;</code>
-     */
-    public org.tensorflow.proto.framework.AttrValue.ListValueOrBuilder getListOrBuilder() {
-      if ((valueCase_ == 1) && (listBuilder_ != null)) {
-        return listBuilder_.getMessageOrBuilder();
-      } else {
-        if (valueCase_ == 1) {
-          return (org.tensorflow.proto.framework.AttrValue.ListValue) value_;
-        }
-        return org.tensorflow.proto.framework.AttrValue.ListValue.getDefaultInstance();
-      }
-    }
-    /**
-     * <pre>
-     * any "list(...)"
-     * </pre>
-     *
-     * <code>.tensorflow.AttrValue.ListValue list = 1;</code>
-     */
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.AttrValue.ListValue, org.tensorflow.proto.framework.AttrValue.ListValue.Builder, org.tensorflow.proto.framework.AttrValue.ListValueOrBuilder> 
-        getListFieldBuilder() {
-      if (listBuilder_ == null) {
-        if (!(valueCase_ == 1)) {
-          value_ = org.tensorflow.proto.framework.AttrValue.ListValue.getDefaultInstance();
-        }
-        listBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-            org.tensorflow.proto.framework.AttrValue.ListValue, org.tensorflow.proto.framework.AttrValue.ListValue.Builder, org.tensorflow.proto.framework.AttrValue.ListValueOrBuilder>(
-                (org.tensorflow.proto.framework.AttrValue.ListValue) value_,
-                getParentForChildren(),
-                isClean());
-        value_ = null;
-      }
-      valueCase_ = 1;
-      onChanged();;
-      return listBuilder_;
-    }
-
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.NameAttrList, org.tensorflow.proto.framework.NameAttrList.Builder, org.tensorflow.proto.framework.NameAttrListOrBuilder> funcBuilder_;
-    /**
-     * <pre>
-     * "func" represents a function. func.name is a function's name or
-     * a primitive op's name. func.attr.first is the name of an attr
-     * defined for that function. func.attr.second is the value for
-     * that attr in the instantiation.
-     * </pre>
-     *
-     * <code>.tensorflow.NameAttrList func = 10;</code>
-     */
-    public boolean hasFunc() {
-      return valueCase_ == 10;
-    }
-    /**
-     * <pre>
-     * "func" represents a function. func.name is a function's name or
-     * a primitive op's name. func.attr.first is the name of an attr
-     * defined for that function. func.attr.second is the value for
-     * that attr in the instantiation.
-     * </pre>
-     *
-     * <code>.tensorflow.NameAttrList func = 10;</code>
-     */
-    public org.tensorflow.proto.framework.NameAttrList getFunc() {
-      if (funcBuilder_ == null) {
-        if (valueCase_ == 10) {
-          return (org.tensorflow.proto.framework.NameAttrList) value_;
-        }
-        return org.tensorflow.proto.framework.NameAttrList.getDefaultInstance();
-      } else {
-        if (valueCase_ == 10) {
-          return funcBuilder_.getMessage();
-        }
-        return org.tensorflow.proto.framework.NameAttrList.getDefaultInstance();
-      }
-    }
-    /**
-     * <pre>
-     * "func" represents a function. func.name is a function's name or
-     * a primitive op's name. func.attr.first is the name of an attr
-     * defined for that function. func.attr.second is the value for
-     * that attr in the instantiation.
-     * </pre>
-     *
-     * <code>.tensorflow.NameAttrList func = 10;</code>
-     */
-    public Builder setFunc(org.tensorflow.proto.framework.NameAttrList value) {
-      if (funcBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        value_ = value;
-        onChanged();
-      } else {
-        funcBuilder_.setMessage(value);
-      }
-      valueCase_ = 10;
-      return this;
-    }
-    /**
-     * <pre>
-     * "func" represents a function. func.name is a function's name or
-     * a primitive op's name. func.attr.first is the name of an attr
-     * defined for that function. func.attr.second is the value for
-     * that attr in the instantiation.
-     * </pre>
-     *
-     * <code>.tensorflow.NameAttrList func = 10;</code>
-     */
-    public Builder setFunc(
-        org.tensorflow.proto.framework.NameAttrList.Builder builderForValue) {
-      if (funcBuilder_ == null) {
-        value_ = builderForValue.build();
-        onChanged();
-      } else {
-        funcBuilder_.setMessage(builderForValue.build());
-      }
-      valueCase_ = 10;
-      return this;
-    }
-    /**
-     * <pre>
-     * "func" represents a function. func.name is a function's name or
-     * a primitive op's name. func.attr.first is the name of an attr
-     * defined for that function. func.attr.second is the value for
-     * that attr in the instantiation.
-     * </pre>
-     *
-     * <code>.tensorflow.NameAttrList func = 10;</code>
-     */
-    public Builder mergeFunc(org.tensorflow.proto.framework.NameAttrList value) {
-      if (funcBuilder_ == null) {
-        if (valueCase_ == 10 &&
-            value_ != org.tensorflow.proto.framework.NameAttrList.getDefaultInstance()) {
-          value_ = org.tensorflow.proto.framework.NameAttrList.newBuilder((org.tensorflow.proto.framework.NameAttrList) value_)
-              .mergeFrom(value).buildPartial();
-        } else {
-          value_ = value;
-        }
-        onChanged();
-      } else {
-        if (valueCase_ == 10) {
-          funcBuilder_.mergeFrom(value);
-        }
-        funcBuilder_.setMessage(value);
-      }
-      valueCase_ = 10;
-      return this;
-    }
-    /**
-     * <pre>
-     * "func" represents a function. func.name is a function's name or
-     * a primitive op's name. func.attr.first is the name of an attr
-     * defined for that function. func.attr.second is the value for
-     * that attr in the instantiation.
-     * </pre>
-     *
-     * <code>.tensorflow.NameAttrList func = 10;</code>
-     */
-    public Builder clearFunc() {
-      if (funcBuilder_ == null) {
-        if (valueCase_ == 10) {
-          valueCase_ = 0;
-          value_ = null;
-          onChanged();
-        }
-      } else {
-        if (valueCase_ == 10) {
-          valueCase_ = 0;
-          value_ = null;
-        }
-        funcBuilder_.clear();
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * "func" represents a function. func.name is a function's name or
-     * a primitive op's name. func.attr.first is the name of an attr
-     * defined for that function. func.attr.second is the value for
-     * that attr in the instantiation.
-     * </pre>
-     *
-     * <code>.tensorflow.NameAttrList func = 10;</code>
-     */
-    public org.tensorflow.proto.framework.NameAttrList.Builder getFuncBuilder() {
-      return getFuncFieldBuilder().getBuilder();
-    }
-    /**
-     * <pre>
-     * "func" represents a function. func.name is a function's name or
-     * a primitive op's name. func.attr.first is the name of an attr
-     * defined for that function. func.attr.second is the value for
-     * that attr in the instantiation.
-     * </pre>
-     *
-     * <code>.tensorflow.NameAttrList func = 10;</code>
-     */
-    public org.tensorflow.proto.framework.NameAttrListOrBuilder getFuncOrBuilder() {
-      if ((valueCase_ == 10) && (funcBuilder_ != null)) {
-        return funcBuilder_.getMessageOrBuilder();
-      } else {
-        if (valueCase_ == 10) {
-          return (org.tensorflow.proto.framework.NameAttrList) value_;
-        }
-        return org.tensorflow.proto.framework.NameAttrList.getDefaultInstance();
-      }
-    }
-    /**
-     * <pre>
-     * "func" represents a function. func.name is a function's name or
-     * a primitive op's name. func.attr.first is the name of an attr
-     * defined for that function. func.attr.second is the value for
-     * that attr in the instantiation.
-     * </pre>
-     *
-     * <code>.tensorflow.NameAttrList func = 10;</code>
-     */
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.NameAttrList, org.tensorflow.proto.framework.NameAttrList.Builder, org.tensorflow.proto.framework.NameAttrListOrBuilder> 
-        getFuncFieldBuilder() {
-      if (funcBuilder_ == null) {
-        if (!(valueCase_ == 10)) {
-          value_ = org.tensorflow.proto.framework.NameAttrList.getDefaultInstance();
-        }
-        funcBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-            org.tensorflow.proto.framework.NameAttrList, org.tensorflow.proto.framework.NameAttrList.Builder, org.tensorflow.proto.framework.NameAttrListOrBuilder>(
-                (org.tensorflow.proto.framework.NameAttrList) value_,
-                getParentForChildren(),
-                isClean());
-        value_ = null;
-      }
-      valueCase_ = 10;
-      onChanged();;
-      return funcBuilder_;
-    }
-
-    /**
-     * <pre>
-     * This is a placeholder only used in nodes defined inside a
-     * function.  It indicates the attr value will be supplied when
-     * the function is instantiated.  For example, let us suppose a
-     * node "N" in function "FN". "N" has an attr "A" with value
-     * placeholder = "foo". When FN is instantiated with attr "foo"
-     * set to "bar", the instantiated node N's attr A will have been
-     * given the value "bar".
-     * </pre>
-     *
-     * <code>string placeholder = 9;</code>
-     */
-    public java.lang.String getPlaceholder() {
-      java.lang.Object ref = "";
-      if (valueCase_ == 9) {
-        ref = value_;
-      }
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        if (valueCase_ == 9) {
-          value_ = s;
-        }
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * This is a placeholder only used in nodes defined inside a
-     * function.  It indicates the attr value will be supplied when
-     * the function is instantiated.  For example, let us suppose a
-     * node "N" in function "FN". "N" has an attr "A" with value
-     * placeholder = "foo". When FN is instantiated with attr "foo"
-     * set to "bar", the instantiated node N's attr A will have been
-     * given the value "bar".
-     * </pre>
-     *
-     * <code>string placeholder = 9;</code>
-     */
-    public com.google.protobuf.ByteString
-        getPlaceholderBytes() {
-      java.lang.Object ref = "";
-      if (valueCase_ == 9) {
-        ref = value_;
-      }
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        if (valueCase_ == 9) {
-          value_ = b;
-        }
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * This is a placeholder only used in nodes defined inside a
-     * function.  It indicates the attr value will be supplied when
-     * the function is instantiated.  For example, let us suppose a
-     * node "N" in function "FN". "N" has an attr "A" with value
-     * placeholder = "foo". When FN is instantiated with attr "foo"
-     * set to "bar", the instantiated node N's attr A will have been
-     * given the value "bar".
-     * </pre>
-     *
-     * <code>string placeholder = 9;</code>
-     */
-    public Builder setPlaceholder(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  valueCase_ = 9;
-      value_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * This is a placeholder only used in nodes defined inside a
-     * function.  It indicates the attr value will be supplied when
-     * the function is instantiated.  For example, let us suppose a
-     * node "N" in function "FN". "N" has an attr "A" with value
-     * placeholder = "foo". When FN is instantiated with attr "foo"
-     * set to "bar", the instantiated node N's attr A will have been
-     * given the value "bar".
-     * </pre>
-     *
-     * <code>string placeholder = 9;</code>
-     */
-    public Builder clearPlaceholder() {
-      if (valueCase_ == 9) {
-        valueCase_ = 0;
-        value_ = null;
-        onChanged();
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * This is a placeholder only used in nodes defined inside a
-     * function.  It indicates the attr value will be supplied when
-     * the function is instantiated.  For example, let us suppose a
-     * node "N" in function "FN". "N" has an attr "A" with value
-     * placeholder = "foo". When FN is instantiated with attr "foo"
-     * set to "bar", the instantiated node N's attr A will have been
-     * given the value "bar".
-     * </pre>
-     *
-     * <code>string placeholder = 9;</code>
-     */
-    public Builder setPlaceholderBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      valueCase_ = 9;
-      value_ = value;
-      onChanged();
-      return this;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.AttrValue)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.AttrValue)
-  private static final org.tensorflow.proto.framework.AttrValue DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.AttrValue();
-  }
-
-  public static org.tensorflow.proto.framework.AttrValue getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<AttrValue>
-      PARSER = new com.google.protobuf.AbstractParser<AttrValue>() {
-    @java.lang.Override
-    public AttrValue parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new AttrValue(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<AttrValue> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<AttrValue> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.framework.AttrValue getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/AttrValueOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/AttrValueOrBuilder.java
deleted file mode 100644
index f2624691edf..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/AttrValueOrBuilder.java
+++ /dev/null
@@ -1,203 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/framework/attr_value.proto
-
-package org.tensorflow.proto.framework;
-
-public interface AttrValueOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.AttrValue)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <pre>
-   * "string"
-   * </pre>
-   *
-   * <code>bytes s = 2;</code>
-   */
-  com.google.protobuf.ByteString getS();
-
-  /**
-   * <pre>
-   * "int"
-   * </pre>
-   *
-   * <code>int64 i = 3;</code>
-   */
-  long getI();
-
-  /**
-   * <pre>
-   * "float"
-   * </pre>
-   *
-   * <code>float f = 4;</code>
-   */
-  float getF();
-
-  /**
-   * <pre>
-   * "bool"
-   * </pre>
-   *
-   * <code>bool b = 5;</code>
-   */
-  boolean getB();
-
-  /**
-   * <pre>
-   * "type"
-   * </pre>
-   *
-   * <code>.tensorflow.DataType type = 6;</code>
-   */
-  int getTypeValue();
-  /**
-   * <pre>
-   * "type"
-   * </pre>
-   *
-   * <code>.tensorflow.DataType type = 6;</code>
-   */
-  org.tensorflow.proto.framework.DataType getType();
-
-  /**
-   * <pre>
-   * "shape"
-   * </pre>
-   *
-   * <code>.tensorflow.TensorShapeProto shape = 7;</code>
-   */
-  boolean hasShape();
-  /**
-   * <pre>
-   * "shape"
-   * </pre>
-   *
-   * <code>.tensorflow.TensorShapeProto shape = 7;</code>
-   */
-  org.tensorflow.proto.framework.TensorShapeProto getShape();
-  /**
-   * <pre>
-   * "shape"
-   * </pre>
-   *
-   * <code>.tensorflow.TensorShapeProto shape = 7;</code>
-   */
-  org.tensorflow.proto.framework.TensorShapeProtoOrBuilder getShapeOrBuilder();
-
-  /**
-   * <pre>
-   * "tensor"
-   * </pre>
-   *
-   * <code>.tensorflow.TensorProto tensor = 8;</code>
-   */
-  boolean hasTensor();
-  /**
-   * <pre>
-   * "tensor"
-   * </pre>
-   *
-   * <code>.tensorflow.TensorProto tensor = 8;</code>
-   */
-  org.tensorflow.proto.framework.TensorProto getTensor();
-  /**
-   * <pre>
-   * "tensor"
-   * </pre>
-   *
-   * <code>.tensorflow.TensorProto tensor = 8;</code>
-   */
-  org.tensorflow.proto.framework.TensorProtoOrBuilder getTensorOrBuilder();
-
-  /**
-   * <pre>
-   * any "list(...)"
-   * </pre>
-   *
-   * <code>.tensorflow.AttrValue.ListValue list = 1;</code>
-   */
-  boolean hasList();
-  /**
-   * <pre>
-   * any "list(...)"
-   * </pre>
-   *
-   * <code>.tensorflow.AttrValue.ListValue list = 1;</code>
-   */
-  org.tensorflow.proto.framework.AttrValue.ListValue getList();
-  /**
-   * <pre>
-   * any "list(...)"
-   * </pre>
-   *
-   * <code>.tensorflow.AttrValue.ListValue list = 1;</code>
-   */
-  org.tensorflow.proto.framework.AttrValue.ListValueOrBuilder getListOrBuilder();
-
-  /**
-   * <pre>
-   * "func" represents a function. func.name is a function's name or
-   * a primitive op's name. func.attr.first is the name of an attr
-   * defined for that function. func.attr.second is the value for
-   * that attr in the instantiation.
-   * </pre>
-   *
-   * <code>.tensorflow.NameAttrList func = 10;</code>
-   */
-  boolean hasFunc();
-  /**
-   * <pre>
-   * "func" represents a function. func.name is a function's name or
-   * a primitive op's name. func.attr.first is the name of an attr
-   * defined for that function. func.attr.second is the value for
-   * that attr in the instantiation.
-   * </pre>
-   *
-   * <code>.tensorflow.NameAttrList func = 10;</code>
-   */
-  org.tensorflow.proto.framework.NameAttrList getFunc();
-  /**
-   * <pre>
-   * "func" represents a function. func.name is a function's name or
-   * a primitive op's name. func.attr.first is the name of an attr
-   * defined for that function. func.attr.second is the value for
-   * that attr in the instantiation.
-   * </pre>
-   *
-   * <code>.tensorflow.NameAttrList func = 10;</code>
-   */
-  org.tensorflow.proto.framework.NameAttrListOrBuilder getFuncOrBuilder();
-
-  /**
-   * <pre>
-   * This is a placeholder only used in nodes defined inside a
-   * function.  It indicates the attr value will be supplied when
-   * the function is instantiated.  For example, let us suppose a
-   * node "N" in function "FN". "N" has an attr "A" with value
-   * placeholder = "foo". When FN is instantiated with attr "foo"
-   * set to "bar", the instantiated node N's attr A will have been
-   * given the value "bar".
-   * </pre>
-   *
-   * <code>string placeholder = 9;</code>
-   */
-  java.lang.String getPlaceholder();
-  /**
-   * <pre>
-   * This is a placeholder only used in nodes defined inside a
-   * function.  It indicates the attr value will be supplied when
-   * the function is instantiated.  For example, let us suppose a
-   * node "N" in function "FN". "N" has an attr "A" with value
-   * placeholder = "foo". When FN is instantiated with attr "foo"
-   * set to "bar", the instantiated node N's attr A will have been
-   * given the value "bar".
-   * </pre>
-   *
-   * <code>string placeholder = 9;</code>
-   */
-  com.google.protobuf.ByteString
-      getPlaceholderBytes();
-
-  public org.tensorflow.proto.framework.AttrValue.ValueCase getValueCase();
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/AttrValueProtos.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/AttrValueProtos.java
deleted file mode 100644
index 2df18dbc3d0..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/AttrValueProtos.java
+++ /dev/null
@@ -1,110 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/framework/attr_value.proto
-
-package org.tensorflow.proto.framework;
-
-public final class AttrValueProtos {
-  private AttrValueProtos() {}
-  public static void registerAllExtensions(
-      com.google.protobuf.ExtensionRegistryLite registry) {
-  }
-
-  public static void registerAllExtensions(
-      com.google.protobuf.ExtensionRegistry registry) {
-    registerAllExtensions(
-        (com.google.protobuf.ExtensionRegistryLite) registry);
-  }
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_AttrValue_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_AttrValue_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_AttrValue_ListValue_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_AttrValue_ListValue_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_NameAttrList_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_NameAttrList_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_NameAttrList_AttrEntry_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_NameAttrList_AttrEntry_fieldAccessorTable;
-
-  public static com.google.protobuf.Descriptors.FileDescriptor
-      getDescriptor() {
-    return descriptor;
-  }
-  private static  com.google.protobuf.Descriptors.FileDescriptor
-      descriptor;
-  static {
-    java.lang.String[] descriptorData = {
-      "\n*tensorflow/core/framework/attr_value.p" +
-      "roto\022\ntensorflow\032&tensorflow/core/framew" +
-      "ork/tensor.proto\032,tensorflow/core/framew" +
-      "ork/tensor_shape.proto\032%tensorflow/core/" +
-      "framework/types.proto\"\246\004\n\tAttrValue\022\013\n\001s" +
-      "\030\002 \001(\014H\000\022\013\n\001i\030\003 \001(\003H\000\022\013\n\001f\030\004 \001(\002H\000\022\013\n\001b\030" +
-      "\005 \001(\010H\000\022$\n\004type\030\006 \001(\0162\024.tensorflow.DataT" +
-      "ypeH\000\022-\n\005shape\030\007 \001(\0132\034.tensorflow.Tensor" +
-      "ShapeProtoH\000\022)\n\006tensor\030\010 \001(\0132\027.tensorflo" +
-      "w.TensorProtoH\000\022/\n\004list\030\001 \001(\0132\037.tensorfl" +
-      "ow.AttrValue.ListValueH\000\022(\n\004func\030\n \001(\0132\030" +
-      ".tensorflow.NameAttrListH\000\022\025\n\013placeholde" +
-      "r\030\t \001(\tH\000\032\351\001\n\tListValue\022\t\n\001s\030\002 \003(\014\022\r\n\001i\030" +
-      "\003 \003(\003B\002\020\001\022\r\n\001f\030\004 \003(\002B\002\020\001\022\r\n\001b\030\005 \003(\010B\002\020\001\022" +
-      "&\n\004type\030\006 \003(\0162\024.tensorflow.DataTypeB\002\020\001\022" +
-      "+\n\005shape\030\007 \003(\0132\034.tensorflow.TensorShapeP" +
-      "roto\022\'\n\006tensor\030\010 \003(\0132\027.tensorflow.Tensor" +
-      "Proto\022&\n\004func\030\t \003(\0132\030.tensorflow.NameAtt" +
-      "rListB\007\n\005value\"\222\001\n\014NameAttrList\022\014\n\004name\030" +
-      "\001 \001(\t\0220\n\004attr\030\002 \003(\0132\".tensorflow.NameAtt" +
-      "rList.AttrEntry\032B\n\tAttrEntry\022\013\n\003key\030\001 \001(" +
-      "\t\022$\n\005value\030\002 \001(\0132\025.tensorflow.AttrValue:" +
-      "\0028\001B\211\001\n\036org.tensorflow.proto.frameworkB\017" +
-      "AttrValueProtosP\001ZQgithub.com/tensorflow" +
-      "/tensorflow/tensorflow/go/core/framework" +
-      "/attr_value_go_proto\370\001\001b\006proto3"
-    };
-    descriptor = com.google.protobuf.Descriptors.FileDescriptor
-      .internalBuildGeneratedFileFrom(descriptorData,
-        new com.google.protobuf.Descriptors.FileDescriptor[] {
-          org.tensorflow.proto.framework.TensorProtos.getDescriptor(),
-          org.tensorflow.proto.framework.TensorShapeProtos.getDescriptor(),
-          org.tensorflow.proto.framework.TypesProtos.getDescriptor(),
-        });
-    internal_static_tensorflow_AttrValue_descriptor =
-      getDescriptor().getMessageTypes().get(0);
-    internal_static_tensorflow_AttrValue_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_AttrValue_descriptor,
-        new java.lang.String[] { "S", "I", "F", "B", "Type", "Shape", "Tensor", "List", "Func", "Placeholder", "Value", });
-    internal_static_tensorflow_AttrValue_ListValue_descriptor =
-      internal_static_tensorflow_AttrValue_descriptor.getNestedTypes().get(0);
-    internal_static_tensorflow_AttrValue_ListValue_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_AttrValue_ListValue_descriptor,
-        new java.lang.String[] { "S", "I", "F", "B", "Type", "Shape", "Tensor", "Func", });
-    internal_static_tensorflow_NameAttrList_descriptor =
-      getDescriptor().getMessageTypes().get(1);
-    internal_static_tensorflow_NameAttrList_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_NameAttrList_descriptor,
-        new java.lang.String[] { "Name", "Attr", });
-    internal_static_tensorflow_NameAttrList_AttrEntry_descriptor =
-      internal_static_tensorflow_NameAttrList_descriptor.getNestedTypes().get(0);
-    internal_static_tensorflow_NameAttrList_AttrEntry_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_NameAttrList_AttrEntry_descriptor,
-        new java.lang.String[] { "Key", "Value", });
-    org.tensorflow.proto.framework.TensorProtos.getDescriptor();
-    org.tensorflow.proto.framework.TensorShapeProtos.getDescriptor();
-    org.tensorflow.proto.framework.TypesProtos.getDescriptor();
-  }
-
-  // @@protoc_insertion_point(outer_class_scope)
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/AutoParallelOptions.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/AutoParallelOptions.java
deleted file mode 100644
index cbb6bde48fe..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/AutoParallelOptions.java
+++ /dev/null
@@ -1,534 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/rewriter_config.proto
-
-package org.tensorflow.proto.framework;
-
-/**
- * Protobuf type {@code tensorflow.AutoParallelOptions}
- */
-public  final class AutoParallelOptions extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.AutoParallelOptions)
-    AutoParallelOptionsOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use AutoParallelOptions.newBuilder() to construct.
-  private AutoParallelOptions(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private AutoParallelOptions() {
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new AutoParallelOptions();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private AutoParallelOptions(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 8: {
-
-            enable_ = input.readBool();
-            break;
-          }
-          case 16: {
-
-            numReplicas_ = input.readInt32();
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.framework.RewriterConfigProtos.internal_static_tensorflow_AutoParallelOptions_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.framework.RewriterConfigProtos.internal_static_tensorflow_AutoParallelOptions_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.framework.AutoParallelOptions.class, org.tensorflow.proto.framework.AutoParallelOptions.Builder.class);
-  }
-
-  public static final int ENABLE_FIELD_NUMBER = 1;
-  private boolean enable_;
-  /**
-   * <code>bool enable = 1;</code>
-   */
-  public boolean getEnable() {
-    return enable_;
-  }
-
-  public static final int NUM_REPLICAS_FIELD_NUMBER = 2;
-  private int numReplicas_;
-  /**
-   * <code>int32 num_replicas = 2;</code>
-   */
-  public int getNumReplicas() {
-    return numReplicas_;
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    if (enable_ != false) {
-      output.writeBool(1, enable_);
-    }
-    if (numReplicas_ != 0) {
-      output.writeInt32(2, numReplicas_);
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (enable_ != false) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeBoolSize(1, enable_);
-    }
-    if (numReplicas_ != 0) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt32Size(2, numReplicas_);
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.framework.AutoParallelOptions)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.framework.AutoParallelOptions other = (org.tensorflow.proto.framework.AutoParallelOptions) obj;
-
-    if (getEnable()
-        != other.getEnable()) return false;
-    if (getNumReplicas()
-        != other.getNumReplicas()) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    hash = (37 * hash) + ENABLE_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashBoolean(
-        getEnable());
-    hash = (37 * hash) + NUM_REPLICAS_FIELD_NUMBER;
-    hash = (53 * hash) + getNumReplicas();
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.framework.AutoParallelOptions parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.AutoParallelOptions parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.AutoParallelOptions parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.AutoParallelOptions parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.AutoParallelOptions parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.AutoParallelOptions parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.AutoParallelOptions parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.AutoParallelOptions parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.AutoParallelOptions parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.AutoParallelOptions parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.AutoParallelOptions parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.AutoParallelOptions parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.framework.AutoParallelOptions prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * Protobuf type {@code tensorflow.AutoParallelOptions}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.AutoParallelOptions)
-      org.tensorflow.proto.framework.AutoParallelOptionsOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.RewriterConfigProtos.internal_static_tensorflow_AutoParallelOptions_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.RewriterConfigProtos.internal_static_tensorflow_AutoParallelOptions_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.AutoParallelOptions.class, org.tensorflow.proto.framework.AutoParallelOptions.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.framework.AutoParallelOptions.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      enable_ = false;
-
-      numReplicas_ = 0;
-
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.framework.RewriterConfigProtos.internal_static_tensorflow_AutoParallelOptions_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.AutoParallelOptions getDefaultInstanceForType() {
-      return org.tensorflow.proto.framework.AutoParallelOptions.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.AutoParallelOptions build() {
-      org.tensorflow.proto.framework.AutoParallelOptions result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.AutoParallelOptions buildPartial() {
-      org.tensorflow.proto.framework.AutoParallelOptions result = new org.tensorflow.proto.framework.AutoParallelOptions(this);
-      result.enable_ = enable_;
-      result.numReplicas_ = numReplicas_;
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.framework.AutoParallelOptions) {
-        return mergeFrom((org.tensorflow.proto.framework.AutoParallelOptions)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.framework.AutoParallelOptions other) {
-      if (other == org.tensorflow.proto.framework.AutoParallelOptions.getDefaultInstance()) return this;
-      if (other.getEnable() != false) {
-        setEnable(other.getEnable());
-      }
-      if (other.getNumReplicas() != 0) {
-        setNumReplicas(other.getNumReplicas());
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.framework.AutoParallelOptions parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.framework.AutoParallelOptions) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-
-    private boolean enable_ ;
-    /**
-     * <code>bool enable = 1;</code>
-     */
-    public boolean getEnable() {
-      return enable_;
-    }
-    /**
-     * <code>bool enable = 1;</code>
-     */
-    public Builder setEnable(boolean value) {
-      
-      enable_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>bool enable = 1;</code>
-     */
-    public Builder clearEnable() {
-      
-      enable_ = false;
-      onChanged();
-      return this;
-    }
-
-    private int numReplicas_ ;
-    /**
-     * <code>int32 num_replicas = 2;</code>
-     */
-    public int getNumReplicas() {
-      return numReplicas_;
-    }
-    /**
-     * <code>int32 num_replicas = 2;</code>
-     */
-    public Builder setNumReplicas(int value) {
-      
-      numReplicas_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>int32 num_replicas = 2;</code>
-     */
-    public Builder clearNumReplicas() {
-      
-      numReplicas_ = 0;
-      onChanged();
-      return this;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.AutoParallelOptions)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.AutoParallelOptions)
-  private static final org.tensorflow.proto.framework.AutoParallelOptions DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.AutoParallelOptions();
-  }
-
-  public static org.tensorflow.proto.framework.AutoParallelOptions getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<AutoParallelOptions>
-      PARSER = new com.google.protobuf.AbstractParser<AutoParallelOptions>() {
-    @java.lang.Override
-    public AutoParallelOptions parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new AutoParallelOptions(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<AutoParallelOptions> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<AutoParallelOptions> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.framework.AutoParallelOptions getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/AutoParallelOptionsOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/AutoParallelOptionsOrBuilder.java
deleted file mode 100644
index 0b954c6477c..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/AutoParallelOptionsOrBuilder.java
+++ /dev/null
@@ -1,19 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/rewriter_config.proto
-
-package org.tensorflow.proto.framework;
-
-public interface AutoParallelOptionsOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.AutoParallelOptions)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <code>bool enable = 1;</code>
-   */
-  boolean getEnable();
-
-  /**
-   * <code>int32 num_replicas = 2;</code>
-   */
-  int getNumReplicas();
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/BoundedTensorSpecProto.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/BoundedTensorSpecProto.java
deleted file mode 100644
index 2c21852a18c..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/BoundedTensorSpecProto.java
+++ /dev/null
@@ -1,1182 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/struct.proto
-
-package org.tensorflow.proto.framework;
-
-/**
- * <pre>
- * A protobuf to represent tf.BoundedTensorSpec.
- * </pre>
- *
- * Protobuf type {@code tensorflow.BoundedTensorSpecProto}
- */
-public  final class BoundedTensorSpecProto extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.BoundedTensorSpecProto)
-    BoundedTensorSpecProtoOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use BoundedTensorSpecProto.newBuilder() to construct.
-  private BoundedTensorSpecProto(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private BoundedTensorSpecProto() {
-    name_ = "";
-    dtype_ = 0;
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new BoundedTensorSpecProto();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private BoundedTensorSpecProto(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 10: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            name_ = s;
-            break;
-          }
-          case 18: {
-            org.tensorflow.proto.framework.TensorShapeProto.Builder subBuilder = null;
-            if (shape_ != null) {
-              subBuilder = shape_.toBuilder();
-            }
-            shape_ = input.readMessage(org.tensorflow.proto.framework.TensorShapeProto.parser(), extensionRegistry);
-            if (subBuilder != null) {
-              subBuilder.mergeFrom(shape_);
-              shape_ = subBuilder.buildPartial();
-            }
-
-            break;
-          }
-          case 24: {
-            int rawValue = input.readEnum();
-
-            dtype_ = rawValue;
-            break;
-          }
-          case 34: {
-            org.tensorflow.proto.framework.TensorProto.Builder subBuilder = null;
-            if (minimum_ != null) {
-              subBuilder = minimum_.toBuilder();
-            }
-            minimum_ = input.readMessage(org.tensorflow.proto.framework.TensorProto.parser(), extensionRegistry);
-            if (subBuilder != null) {
-              subBuilder.mergeFrom(minimum_);
-              minimum_ = subBuilder.buildPartial();
-            }
-
-            break;
-          }
-          case 42: {
-            org.tensorflow.proto.framework.TensorProto.Builder subBuilder = null;
-            if (maximum_ != null) {
-              subBuilder = maximum_.toBuilder();
-            }
-            maximum_ = input.readMessage(org.tensorflow.proto.framework.TensorProto.parser(), extensionRegistry);
-            if (subBuilder != null) {
-              subBuilder.mergeFrom(maximum_);
-              maximum_ = subBuilder.buildPartial();
-            }
-
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.framework.StructProtos.internal_static_tensorflow_BoundedTensorSpecProto_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.framework.StructProtos.internal_static_tensorflow_BoundedTensorSpecProto_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.framework.BoundedTensorSpecProto.class, org.tensorflow.proto.framework.BoundedTensorSpecProto.Builder.class);
-  }
-
-  public static final int NAME_FIELD_NUMBER = 1;
-  private volatile java.lang.Object name_;
-  /**
-   * <code>string name = 1;</code>
-   */
-  public java.lang.String getName() {
-    java.lang.Object ref = name_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      name_ = s;
-      return s;
-    }
-  }
-  /**
-   * <code>string name = 1;</code>
-   */
-  public com.google.protobuf.ByteString
-      getNameBytes() {
-    java.lang.Object ref = name_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      name_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int SHAPE_FIELD_NUMBER = 2;
-  private org.tensorflow.proto.framework.TensorShapeProto shape_;
-  /**
-   * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-   */
-  public boolean hasShape() {
-    return shape_ != null;
-  }
-  /**
-   * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-   */
-  public org.tensorflow.proto.framework.TensorShapeProto getShape() {
-    return shape_ == null ? org.tensorflow.proto.framework.TensorShapeProto.getDefaultInstance() : shape_;
-  }
-  /**
-   * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-   */
-  public org.tensorflow.proto.framework.TensorShapeProtoOrBuilder getShapeOrBuilder() {
-    return getShape();
-  }
-
-  public static final int DTYPE_FIELD_NUMBER = 3;
-  private int dtype_;
-  /**
-   * <code>.tensorflow.DataType dtype = 3;</code>
-   */
-  public int getDtypeValue() {
-    return dtype_;
-  }
-  /**
-   * <code>.tensorflow.DataType dtype = 3;</code>
-   */
-  public org.tensorflow.proto.framework.DataType getDtype() {
-    @SuppressWarnings("deprecation")
-    org.tensorflow.proto.framework.DataType result = org.tensorflow.proto.framework.DataType.valueOf(dtype_);
-    return result == null ? org.tensorflow.proto.framework.DataType.UNRECOGNIZED : result;
-  }
-
-  public static final int MINIMUM_FIELD_NUMBER = 4;
-  private org.tensorflow.proto.framework.TensorProto minimum_;
-  /**
-   * <code>.tensorflow.TensorProto minimum = 4;</code>
-   */
-  public boolean hasMinimum() {
-    return minimum_ != null;
-  }
-  /**
-   * <code>.tensorflow.TensorProto minimum = 4;</code>
-   */
-  public org.tensorflow.proto.framework.TensorProto getMinimum() {
-    return minimum_ == null ? org.tensorflow.proto.framework.TensorProto.getDefaultInstance() : minimum_;
-  }
-  /**
-   * <code>.tensorflow.TensorProto minimum = 4;</code>
-   */
-  public org.tensorflow.proto.framework.TensorProtoOrBuilder getMinimumOrBuilder() {
-    return getMinimum();
-  }
-
-  public static final int MAXIMUM_FIELD_NUMBER = 5;
-  private org.tensorflow.proto.framework.TensorProto maximum_;
-  /**
-   * <code>.tensorflow.TensorProto maximum = 5;</code>
-   */
-  public boolean hasMaximum() {
-    return maximum_ != null;
-  }
-  /**
-   * <code>.tensorflow.TensorProto maximum = 5;</code>
-   */
-  public org.tensorflow.proto.framework.TensorProto getMaximum() {
-    return maximum_ == null ? org.tensorflow.proto.framework.TensorProto.getDefaultInstance() : maximum_;
-  }
-  /**
-   * <code>.tensorflow.TensorProto maximum = 5;</code>
-   */
-  public org.tensorflow.proto.framework.TensorProtoOrBuilder getMaximumOrBuilder() {
-    return getMaximum();
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    if (!getNameBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 1, name_);
-    }
-    if (shape_ != null) {
-      output.writeMessage(2, getShape());
-    }
-    if (dtype_ != org.tensorflow.proto.framework.DataType.DT_INVALID.getNumber()) {
-      output.writeEnum(3, dtype_);
-    }
-    if (minimum_ != null) {
-      output.writeMessage(4, getMinimum());
-    }
-    if (maximum_ != null) {
-      output.writeMessage(5, getMaximum());
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (!getNameBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, name_);
-    }
-    if (shape_ != null) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(2, getShape());
-    }
-    if (dtype_ != org.tensorflow.proto.framework.DataType.DT_INVALID.getNumber()) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeEnumSize(3, dtype_);
-    }
-    if (minimum_ != null) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(4, getMinimum());
-    }
-    if (maximum_ != null) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(5, getMaximum());
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.framework.BoundedTensorSpecProto)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.framework.BoundedTensorSpecProto other = (org.tensorflow.proto.framework.BoundedTensorSpecProto) obj;
-
-    if (!getName()
-        .equals(other.getName())) return false;
-    if (hasShape() != other.hasShape()) return false;
-    if (hasShape()) {
-      if (!getShape()
-          .equals(other.getShape())) return false;
-    }
-    if (dtype_ != other.dtype_) return false;
-    if (hasMinimum() != other.hasMinimum()) return false;
-    if (hasMinimum()) {
-      if (!getMinimum()
-          .equals(other.getMinimum())) return false;
-    }
-    if (hasMaximum() != other.hasMaximum()) return false;
-    if (hasMaximum()) {
-      if (!getMaximum()
-          .equals(other.getMaximum())) return false;
-    }
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    hash = (37 * hash) + NAME_FIELD_NUMBER;
-    hash = (53 * hash) + getName().hashCode();
-    if (hasShape()) {
-      hash = (37 * hash) + SHAPE_FIELD_NUMBER;
-      hash = (53 * hash) + getShape().hashCode();
-    }
-    hash = (37 * hash) + DTYPE_FIELD_NUMBER;
-    hash = (53 * hash) + dtype_;
-    if (hasMinimum()) {
-      hash = (37 * hash) + MINIMUM_FIELD_NUMBER;
-      hash = (53 * hash) + getMinimum().hashCode();
-    }
-    if (hasMaximum()) {
-      hash = (37 * hash) + MAXIMUM_FIELD_NUMBER;
-      hash = (53 * hash) + getMaximum().hashCode();
-    }
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.framework.BoundedTensorSpecProto parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.BoundedTensorSpecProto parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.BoundedTensorSpecProto parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.BoundedTensorSpecProto parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.BoundedTensorSpecProto parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.BoundedTensorSpecProto parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.BoundedTensorSpecProto parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.BoundedTensorSpecProto parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.BoundedTensorSpecProto parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.BoundedTensorSpecProto parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.BoundedTensorSpecProto parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.BoundedTensorSpecProto parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.framework.BoundedTensorSpecProto prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * <pre>
-   * A protobuf to represent tf.BoundedTensorSpec.
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.BoundedTensorSpecProto}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.BoundedTensorSpecProto)
-      org.tensorflow.proto.framework.BoundedTensorSpecProtoOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.StructProtos.internal_static_tensorflow_BoundedTensorSpecProto_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.StructProtos.internal_static_tensorflow_BoundedTensorSpecProto_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.BoundedTensorSpecProto.class, org.tensorflow.proto.framework.BoundedTensorSpecProto.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.framework.BoundedTensorSpecProto.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      name_ = "";
-
-      if (shapeBuilder_ == null) {
-        shape_ = null;
-      } else {
-        shape_ = null;
-        shapeBuilder_ = null;
-      }
-      dtype_ = 0;
-
-      if (minimumBuilder_ == null) {
-        minimum_ = null;
-      } else {
-        minimum_ = null;
-        minimumBuilder_ = null;
-      }
-      if (maximumBuilder_ == null) {
-        maximum_ = null;
-      } else {
-        maximum_ = null;
-        maximumBuilder_ = null;
-      }
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.framework.StructProtos.internal_static_tensorflow_BoundedTensorSpecProto_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.BoundedTensorSpecProto getDefaultInstanceForType() {
-      return org.tensorflow.proto.framework.BoundedTensorSpecProto.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.BoundedTensorSpecProto build() {
-      org.tensorflow.proto.framework.BoundedTensorSpecProto result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.BoundedTensorSpecProto buildPartial() {
-      org.tensorflow.proto.framework.BoundedTensorSpecProto result = new org.tensorflow.proto.framework.BoundedTensorSpecProto(this);
-      result.name_ = name_;
-      if (shapeBuilder_ == null) {
-        result.shape_ = shape_;
-      } else {
-        result.shape_ = shapeBuilder_.build();
-      }
-      result.dtype_ = dtype_;
-      if (minimumBuilder_ == null) {
-        result.minimum_ = minimum_;
-      } else {
-        result.minimum_ = minimumBuilder_.build();
-      }
-      if (maximumBuilder_ == null) {
-        result.maximum_ = maximum_;
-      } else {
-        result.maximum_ = maximumBuilder_.build();
-      }
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.framework.BoundedTensorSpecProto) {
-        return mergeFrom((org.tensorflow.proto.framework.BoundedTensorSpecProto)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.framework.BoundedTensorSpecProto other) {
-      if (other == org.tensorflow.proto.framework.BoundedTensorSpecProto.getDefaultInstance()) return this;
-      if (!other.getName().isEmpty()) {
-        name_ = other.name_;
-        onChanged();
-      }
-      if (other.hasShape()) {
-        mergeShape(other.getShape());
-      }
-      if (other.dtype_ != 0) {
-        setDtypeValue(other.getDtypeValue());
-      }
-      if (other.hasMinimum()) {
-        mergeMinimum(other.getMinimum());
-      }
-      if (other.hasMaximum()) {
-        mergeMaximum(other.getMaximum());
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.framework.BoundedTensorSpecProto parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.framework.BoundedTensorSpecProto) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-
-    private java.lang.Object name_ = "";
-    /**
-     * <code>string name = 1;</code>
-     */
-    public java.lang.String getName() {
-      java.lang.Object ref = name_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        name_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <code>string name = 1;</code>
-     */
-    public com.google.protobuf.ByteString
-        getNameBytes() {
-      java.lang.Object ref = name_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        name_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <code>string name = 1;</code>
-     */
-    public Builder setName(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      name_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>string name = 1;</code>
-     */
-    public Builder clearName() {
-      
-      name_ = getDefaultInstance().getName();
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>string name = 1;</code>
-     */
-    public Builder setNameBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      name_ = value;
-      onChanged();
-      return this;
-    }
-
-    private org.tensorflow.proto.framework.TensorShapeProto shape_;
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.TensorShapeProto, org.tensorflow.proto.framework.TensorShapeProto.Builder, org.tensorflow.proto.framework.TensorShapeProtoOrBuilder> shapeBuilder_;
-    /**
-     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-     */
-    public boolean hasShape() {
-      return shapeBuilder_ != null || shape_ != null;
-    }
-    /**
-     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-     */
-    public org.tensorflow.proto.framework.TensorShapeProto getShape() {
-      if (shapeBuilder_ == null) {
-        return shape_ == null ? org.tensorflow.proto.framework.TensorShapeProto.getDefaultInstance() : shape_;
-      } else {
-        return shapeBuilder_.getMessage();
-      }
-    }
-    /**
-     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-     */
-    public Builder setShape(org.tensorflow.proto.framework.TensorShapeProto value) {
-      if (shapeBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        shape_ = value;
-        onChanged();
-      } else {
-        shapeBuilder_.setMessage(value);
-      }
-
-      return this;
-    }
-    /**
-     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-     */
-    public Builder setShape(
-        org.tensorflow.proto.framework.TensorShapeProto.Builder builderForValue) {
-      if (shapeBuilder_ == null) {
-        shape_ = builderForValue.build();
-        onChanged();
-      } else {
-        shapeBuilder_.setMessage(builderForValue.build());
-      }
-
-      return this;
-    }
-    /**
-     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-     */
-    public Builder mergeShape(org.tensorflow.proto.framework.TensorShapeProto value) {
-      if (shapeBuilder_ == null) {
-        if (shape_ != null) {
-          shape_ =
-            org.tensorflow.proto.framework.TensorShapeProto.newBuilder(shape_).mergeFrom(value).buildPartial();
-        } else {
-          shape_ = value;
-        }
-        onChanged();
-      } else {
-        shapeBuilder_.mergeFrom(value);
-      }
-
-      return this;
-    }
-    /**
-     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-     */
-    public Builder clearShape() {
-      if (shapeBuilder_ == null) {
-        shape_ = null;
-        onChanged();
-      } else {
-        shape_ = null;
-        shapeBuilder_ = null;
-      }
-
-      return this;
-    }
-    /**
-     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-     */
-    public org.tensorflow.proto.framework.TensorShapeProto.Builder getShapeBuilder() {
-      
-      onChanged();
-      return getShapeFieldBuilder().getBuilder();
-    }
-    /**
-     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-     */
-    public org.tensorflow.proto.framework.TensorShapeProtoOrBuilder getShapeOrBuilder() {
-      if (shapeBuilder_ != null) {
-        return shapeBuilder_.getMessageOrBuilder();
-      } else {
-        return shape_ == null ?
-            org.tensorflow.proto.framework.TensorShapeProto.getDefaultInstance() : shape_;
-      }
-    }
-    /**
-     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-     */
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.TensorShapeProto, org.tensorflow.proto.framework.TensorShapeProto.Builder, org.tensorflow.proto.framework.TensorShapeProtoOrBuilder> 
-        getShapeFieldBuilder() {
-      if (shapeBuilder_ == null) {
-        shapeBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-            org.tensorflow.proto.framework.TensorShapeProto, org.tensorflow.proto.framework.TensorShapeProto.Builder, org.tensorflow.proto.framework.TensorShapeProtoOrBuilder>(
-                getShape(),
-                getParentForChildren(),
-                isClean());
-        shape_ = null;
-      }
-      return shapeBuilder_;
-    }
-
-    private int dtype_ = 0;
-    /**
-     * <code>.tensorflow.DataType dtype = 3;</code>
-     */
-    public int getDtypeValue() {
-      return dtype_;
-    }
-    /**
-     * <code>.tensorflow.DataType dtype = 3;</code>
-     */
-    public Builder setDtypeValue(int value) {
-      dtype_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>.tensorflow.DataType dtype = 3;</code>
-     */
-    public org.tensorflow.proto.framework.DataType getDtype() {
-      @SuppressWarnings("deprecation")
-      org.tensorflow.proto.framework.DataType result = org.tensorflow.proto.framework.DataType.valueOf(dtype_);
-      return result == null ? org.tensorflow.proto.framework.DataType.UNRECOGNIZED : result;
-    }
-    /**
-     * <code>.tensorflow.DataType dtype = 3;</code>
-     */
-    public Builder setDtype(org.tensorflow.proto.framework.DataType value) {
-      if (value == null) {
-        throw new NullPointerException();
-      }
-      
-      dtype_ = value.getNumber();
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>.tensorflow.DataType dtype = 3;</code>
-     */
-    public Builder clearDtype() {
-      
-      dtype_ = 0;
-      onChanged();
-      return this;
-    }
-
-    private org.tensorflow.proto.framework.TensorProto minimum_;
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.TensorProto, org.tensorflow.proto.framework.TensorProto.Builder, org.tensorflow.proto.framework.TensorProtoOrBuilder> minimumBuilder_;
-    /**
-     * <code>.tensorflow.TensorProto minimum = 4;</code>
-     */
-    public boolean hasMinimum() {
-      return minimumBuilder_ != null || minimum_ != null;
-    }
-    /**
-     * <code>.tensorflow.TensorProto minimum = 4;</code>
-     */
-    public org.tensorflow.proto.framework.TensorProto getMinimum() {
-      if (minimumBuilder_ == null) {
-        return minimum_ == null ? org.tensorflow.proto.framework.TensorProto.getDefaultInstance() : minimum_;
-      } else {
-        return minimumBuilder_.getMessage();
-      }
-    }
-    /**
-     * <code>.tensorflow.TensorProto minimum = 4;</code>
-     */
-    public Builder setMinimum(org.tensorflow.proto.framework.TensorProto value) {
-      if (minimumBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        minimum_ = value;
-        onChanged();
-      } else {
-        minimumBuilder_.setMessage(value);
-      }
-
-      return this;
-    }
-    /**
-     * <code>.tensorflow.TensorProto minimum = 4;</code>
-     */
-    public Builder setMinimum(
-        org.tensorflow.proto.framework.TensorProto.Builder builderForValue) {
-      if (minimumBuilder_ == null) {
-        minimum_ = builderForValue.build();
-        onChanged();
-      } else {
-        minimumBuilder_.setMessage(builderForValue.build());
-      }
-
-      return this;
-    }
-    /**
-     * <code>.tensorflow.TensorProto minimum = 4;</code>
-     */
-    public Builder mergeMinimum(org.tensorflow.proto.framework.TensorProto value) {
-      if (minimumBuilder_ == null) {
-        if (minimum_ != null) {
-          minimum_ =
-            org.tensorflow.proto.framework.TensorProto.newBuilder(minimum_).mergeFrom(value).buildPartial();
-        } else {
-          minimum_ = value;
-        }
-        onChanged();
-      } else {
-        minimumBuilder_.mergeFrom(value);
-      }
-
-      return this;
-    }
-    /**
-     * <code>.tensorflow.TensorProto minimum = 4;</code>
-     */
-    public Builder clearMinimum() {
-      if (minimumBuilder_ == null) {
-        minimum_ = null;
-        onChanged();
-      } else {
-        minimum_ = null;
-        minimumBuilder_ = null;
-      }
-
-      return this;
-    }
-    /**
-     * <code>.tensorflow.TensorProto minimum = 4;</code>
-     */
-    public org.tensorflow.proto.framework.TensorProto.Builder getMinimumBuilder() {
-      
-      onChanged();
-      return getMinimumFieldBuilder().getBuilder();
-    }
-    /**
-     * <code>.tensorflow.TensorProto minimum = 4;</code>
-     */
-    public org.tensorflow.proto.framework.TensorProtoOrBuilder getMinimumOrBuilder() {
-      if (minimumBuilder_ != null) {
-        return minimumBuilder_.getMessageOrBuilder();
-      } else {
-        return minimum_ == null ?
-            org.tensorflow.proto.framework.TensorProto.getDefaultInstance() : minimum_;
-      }
-    }
-    /**
-     * <code>.tensorflow.TensorProto minimum = 4;</code>
-     */
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.TensorProto, org.tensorflow.proto.framework.TensorProto.Builder, org.tensorflow.proto.framework.TensorProtoOrBuilder> 
-        getMinimumFieldBuilder() {
-      if (minimumBuilder_ == null) {
-        minimumBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-            org.tensorflow.proto.framework.TensorProto, org.tensorflow.proto.framework.TensorProto.Builder, org.tensorflow.proto.framework.TensorProtoOrBuilder>(
-                getMinimum(),
-                getParentForChildren(),
-                isClean());
-        minimum_ = null;
-      }
-      return minimumBuilder_;
-    }
-
-    private org.tensorflow.proto.framework.TensorProto maximum_;
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.TensorProto, org.tensorflow.proto.framework.TensorProto.Builder, org.tensorflow.proto.framework.TensorProtoOrBuilder> maximumBuilder_;
-    /**
-     * <code>.tensorflow.TensorProto maximum = 5;</code>
-     */
-    public boolean hasMaximum() {
-      return maximumBuilder_ != null || maximum_ != null;
-    }
-    /**
-     * <code>.tensorflow.TensorProto maximum = 5;</code>
-     */
-    public org.tensorflow.proto.framework.TensorProto getMaximum() {
-      if (maximumBuilder_ == null) {
-        return maximum_ == null ? org.tensorflow.proto.framework.TensorProto.getDefaultInstance() : maximum_;
-      } else {
-        return maximumBuilder_.getMessage();
-      }
-    }
-    /**
-     * <code>.tensorflow.TensorProto maximum = 5;</code>
-     */
-    public Builder setMaximum(org.tensorflow.proto.framework.TensorProto value) {
-      if (maximumBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        maximum_ = value;
-        onChanged();
-      } else {
-        maximumBuilder_.setMessage(value);
-      }
-
-      return this;
-    }
-    /**
-     * <code>.tensorflow.TensorProto maximum = 5;</code>
-     */
-    public Builder setMaximum(
-        org.tensorflow.proto.framework.TensorProto.Builder builderForValue) {
-      if (maximumBuilder_ == null) {
-        maximum_ = builderForValue.build();
-        onChanged();
-      } else {
-        maximumBuilder_.setMessage(builderForValue.build());
-      }
-
-      return this;
-    }
-    /**
-     * <code>.tensorflow.TensorProto maximum = 5;</code>
-     */
-    public Builder mergeMaximum(org.tensorflow.proto.framework.TensorProto value) {
-      if (maximumBuilder_ == null) {
-        if (maximum_ != null) {
-          maximum_ =
-            org.tensorflow.proto.framework.TensorProto.newBuilder(maximum_).mergeFrom(value).buildPartial();
-        } else {
-          maximum_ = value;
-        }
-        onChanged();
-      } else {
-        maximumBuilder_.mergeFrom(value);
-      }
-
-      return this;
-    }
-    /**
-     * <code>.tensorflow.TensorProto maximum = 5;</code>
-     */
-    public Builder clearMaximum() {
-      if (maximumBuilder_ == null) {
-        maximum_ = null;
-        onChanged();
-      } else {
-        maximum_ = null;
-        maximumBuilder_ = null;
-      }
-
-      return this;
-    }
-    /**
-     * <code>.tensorflow.TensorProto maximum = 5;</code>
-     */
-    public org.tensorflow.proto.framework.TensorProto.Builder getMaximumBuilder() {
-      
-      onChanged();
-      return getMaximumFieldBuilder().getBuilder();
-    }
-    /**
-     * <code>.tensorflow.TensorProto maximum = 5;</code>
-     */
-    public org.tensorflow.proto.framework.TensorProtoOrBuilder getMaximumOrBuilder() {
-      if (maximumBuilder_ != null) {
-        return maximumBuilder_.getMessageOrBuilder();
-      } else {
-        return maximum_ == null ?
-            org.tensorflow.proto.framework.TensorProto.getDefaultInstance() : maximum_;
-      }
-    }
-    /**
-     * <code>.tensorflow.TensorProto maximum = 5;</code>
-     */
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.TensorProto, org.tensorflow.proto.framework.TensorProto.Builder, org.tensorflow.proto.framework.TensorProtoOrBuilder> 
-        getMaximumFieldBuilder() {
-      if (maximumBuilder_ == null) {
-        maximumBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-            org.tensorflow.proto.framework.TensorProto, org.tensorflow.proto.framework.TensorProto.Builder, org.tensorflow.proto.framework.TensorProtoOrBuilder>(
-                getMaximum(),
-                getParentForChildren(),
-                isClean());
-        maximum_ = null;
-      }
-      return maximumBuilder_;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.BoundedTensorSpecProto)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.BoundedTensorSpecProto)
-  private static final org.tensorflow.proto.framework.BoundedTensorSpecProto DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.BoundedTensorSpecProto();
-  }
-
-  public static org.tensorflow.proto.framework.BoundedTensorSpecProto getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<BoundedTensorSpecProto>
-      PARSER = new com.google.protobuf.AbstractParser<BoundedTensorSpecProto>() {
-    @java.lang.Override
-    public BoundedTensorSpecProto parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new BoundedTensorSpecProto(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<BoundedTensorSpecProto> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<BoundedTensorSpecProto> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.framework.BoundedTensorSpecProto getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/BoundedTensorSpecProtoOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/BoundedTensorSpecProtoOrBuilder.java
deleted file mode 100644
index d4b34e04e02..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/BoundedTensorSpecProtoOrBuilder.java
+++ /dev/null
@@ -1,67 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/struct.proto
-
-package org.tensorflow.proto.framework;
-
-public interface BoundedTensorSpecProtoOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.BoundedTensorSpecProto)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <code>string name = 1;</code>
-   */
-  java.lang.String getName();
-  /**
-   * <code>string name = 1;</code>
-   */
-  com.google.protobuf.ByteString
-      getNameBytes();
-
-  /**
-   * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-   */
-  boolean hasShape();
-  /**
-   * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-   */
-  org.tensorflow.proto.framework.TensorShapeProto getShape();
-  /**
-   * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-   */
-  org.tensorflow.proto.framework.TensorShapeProtoOrBuilder getShapeOrBuilder();
-
-  /**
-   * <code>.tensorflow.DataType dtype = 3;</code>
-   */
-  int getDtypeValue();
-  /**
-   * <code>.tensorflow.DataType dtype = 3;</code>
-   */
-  org.tensorflow.proto.framework.DataType getDtype();
-
-  /**
-   * <code>.tensorflow.TensorProto minimum = 4;</code>
-   */
-  boolean hasMinimum();
-  /**
-   * <code>.tensorflow.TensorProto minimum = 4;</code>
-   */
-  org.tensorflow.proto.framework.TensorProto getMinimum();
-  /**
-   * <code>.tensorflow.TensorProto minimum = 4;</code>
-   */
-  org.tensorflow.proto.framework.TensorProtoOrBuilder getMinimumOrBuilder();
-
-  /**
-   * <code>.tensorflow.TensorProto maximum = 5;</code>
-   */
-  boolean hasMaximum();
-  /**
-   * <code>.tensorflow.TensorProto maximum = 5;</code>
-   */
-  org.tensorflow.proto.framework.TensorProto getMaximum();
-  /**
-   * <code>.tensorflow.TensorProto maximum = 5;</code>
-   */
-  org.tensorflow.proto.framework.TensorProtoOrBuilder getMaximumOrBuilder();
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/CallableOptions.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/CallableOptions.java
deleted file mode 100644
index 1128fdc60f8..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/CallableOptions.java
+++ /dev/null
@@ -1,2902 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/config.proto
-
-package org.tensorflow.proto.framework;
-
-/**
- * <pre>
- * Defines a subgraph in another `GraphDef` as a set of feed points and nodes
- * to be fetched or executed.
- * Compare with the arguments to `Session::Run()`.
- * </pre>
- *
- * Protobuf type {@code tensorflow.CallableOptions}
- */
-public  final class CallableOptions extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.CallableOptions)
-    CallableOptionsOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use CallableOptions.newBuilder() to construct.
-  private CallableOptions(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private CallableOptions() {
-    feed_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-    fetch_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-    target_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-    tensorConnection_ = java.util.Collections.emptyList();
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new CallableOptions();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private CallableOptions(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    int mutable_bitField0_ = 0;
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 10: {
-            java.lang.String s = input.readStringRequireUtf8();
-            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
-              feed_ = new com.google.protobuf.LazyStringArrayList();
-              mutable_bitField0_ |= 0x00000001;
-            }
-            feed_.add(s);
-            break;
-          }
-          case 18: {
-            java.lang.String s = input.readStringRequireUtf8();
-            if (!((mutable_bitField0_ & 0x00000002) != 0)) {
-              fetch_ = new com.google.protobuf.LazyStringArrayList();
-              mutable_bitField0_ |= 0x00000002;
-            }
-            fetch_.add(s);
-            break;
-          }
-          case 26: {
-            java.lang.String s = input.readStringRequireUtf8();
-            if (!((mutable_bitField0_ & 0x00000004) != 0)) {
-              target_ = new com.google.protobuf.LazyStringArrayList();
-              mutable_bitField0_ |= 0x00000004;
-            }
-            target_.add(s);
-            break;
-          }
-          case 34: {
-            org.tensorflow.proto.framework.RunOptions.Builder subBuilder = null;
-            if (runOptions_ != null) {
-              subBuilder = runOptions_.toBuilder();
-            }
-            runOptions_ = input.readMessage(org.tensorflow.proto.framework.RunOptions.parser(), extensionRegistry);
-            if (subBuilder != null) {
-              subBuilder.mergeFrom(runOptions_);
-              runOptions_ = subBuilder.buildPartial();
-            }
-
-            break;
-          }
-          case 42: {
-            if (!((mutable_bitField0_ & 0x00000008) != 0)) {
-              tensorConnection_ = new java.util.ArrayList<org.tensorflow.proto.framework.TensorConnection>();
-              mutable_bitField0_ |= 0x00000008;
-            }
-            tensorConnection_.add(
-                input.readMessage(org.tensorflow.proto.framework.TensorConnection.parser(), extensionRegistry));
-            break;
-          }
-          case 50: {
-            if (!((mutable_bitField0_ & 0x00000010) != 0)) {
-              feedDevices_ = com.google.protobuf.MapField.newMapField(
-                  FeedDevicesDefaultEntryHolder.defaultEntry);
-              mutable_bitField0_ |= 0x00000010;
-            }
-            com.google.protobuf.MapEntry<java.lang.String, java.lang.String>
-            feedDevices__ = input.readMessage(
-                FeedDevicesDefaultEntryHolder.defaultEntry.getParserForType(), extensionRegistry);
-            feedDevices_.getMutableMap().put(
-                feedDevices__.getKey(), feedDevices__.getValue());
-            break;
-          }
-          case 58: {
-            if (!((mutable_bitField0_ & 0x00000020) != 0)) {
-              fetchDevices_ = com.google.protobuf.MapField.newMapField(
-                  FetchDevicesDefaultEntryHolder.defaultEntry);
-              mutable_bitField0_ |= 0x00000020;
-            }
-            com.google.protobuf.MapEntry<java.lang.String, java.lang.String>
-            fetchDevices__ = input.readMessage(
-                FetchDevicesDefaultEntryHolder.defaultEntry.getParserForType(), extensionRegistry);
-            fetchDevices_.getMutableMap().put(
-                fetchDevices__.getKey(), fetchDevices__.getValue());
-            break;
-          }
-          case 64: {
-
-            fetchSkipSync_ = input.readBool();
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      if (((mutable_bitField0_ & 0x00000001) != 0)) {
-        feed_ = feed_.getUnmodifiableView();
-      }
-      if (((mutable_bitField0_ & 0x00000002) != 0)) {
-        fetch_ = fetch_.getUnmodifiableView();
-      }
-      if (((mutable_bitField0_ & 0x00000004) != 0)) {
-        target_ = target_.getUnmodifiableView();
-      }
-      if (((mutable_bitField0_ & 0x00000008) != 0)) {
-        tensorConnection_ = java.util.Collections.unmodifiableList(tensorConnection_);
-      }
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_CallableOptions_descriptor;
-  }
-
-  @SuppressWarnings({"rawtypes"})
-  @java.lang.Override
-  protected com.google.protobuf.MapField internalGetMapField(
-      int number) {
-    switch (number) {
-      case 6:
-        return internalGetFeedDevices();
-      case 7:
-        return internalGetFetchDevices();
-      default:
-        throw new RuntimeException(
-            "Invalid map field number: " + number);
-    }
-  }
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_CallableOptions_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.framework.CallableOptions.class, org.tensorflow.proto.framework.CallableOptions.Builder.class);
-  }
-
-  public static final int FEED_FIELD_NUMBER = 1;
-  private com.google.protobuf.LazyStringList feed_;
-  /**
-   * <pre>
-   * Tensors to be fed in the callable. Each feed is the name of a tensor.
-   * </pre>
-   *
-   * <code>repeated string feed = 1;</code>
-   */
-  public com.google.protobuf.ProtocolStringList
-      getFeedList() {
-    return feed_;
-  }
-  /**
-   * <pre>
-   * Tensors to be fed in the callable. Each feed is the name of a tensor.
-   * </pre>
-   *
-   * <code>repeated string feed = 1;</code>
-   */
-  public int getFeedCount() {
-    return feed_.size();
-  }
-  /**
-   * <pre>
-   * Tensors to be fed in the callable. Each feed is the name of a tensor.
-   * </pre>
-   *
-   * <code>repeated string feed = 1;</code>
-   */
-  public java.lang.String getFeed(int index) {
-    return feed_.get(index);
-  }
-  /**
-   * <pre>
-   * Tensors to be fed in the callable. Each feed is the name of a tensor.
-   * </pre>
-   *
-   * <code>repeated string feed = 1;</code>
-   */
-  public com.google.protobuf.ByteString
-      getFeedBytes(int index) {
-    return feed_.getByteString(index);
-  }
-
-  public static final int FETCH_FIELD_NUMBER = 2;
-  private com.google.protobuf.LazyStringList fetch_;
-  /**
-   * <pre>
-   * Fetches. A list of tensor names. The caller of the callable expects a
-   * tensor to be returned for each fetch[i] (see RunStepResponse.tensor). The
-   * order of specified fetches does not change the execution order.
-   * </pre>
-   *
-   * <code>repeated string fetch = 2;</code>
-   */
-  public com.google.protobuf.ProtocolStringList
-      getFetchList() {
-    return fetch_;
-  }
-  /**
-   * <pre>
-   * Fetches. A list of tensor names. The caller of the callable expects a
-   * tensor to be returned for each fetch[i] (see RunStepResponse.tensor). The
-   * order of specified fetches does not change the execution order.
-   * </pre>
-   *
-   * <code>repeated string fetch = 2;</code>
-   */
-  public int getFetchCount() {
-    return fetch_.size();
-  }
-  /**
-   * <pre>
-   * Fetches. A list of tensor names. The caller of the callable expects a
-   * tensor to be returned for each fetch[i] (see RunStepResponse.tensor). The
-   * order of specified fetches does not change the execution order.
-   * </pre>
-   *
-   * <code>repeated string fetch = 2;</code>
-   */
-  public java.lang.String getFetch(int index) {
-    return fetch_.get(index);
-  }
-  /**
-   * <pre>
-   * Fetches. A list of tensor names. The caller of the callable expects a
-   * tensor to be returned for each fetch[i] (see RunStepResponse.tensor). The
-   * order of specified fetches does not change the execution order.
-   * </pre>
-   *
-   * <code>repeated string fetch = 2;</code>
-   */
-  public com.google.protobuf.ByteString
-      getFetchBytes(int index) {
-    return fetch_.getByteString(index);
-  }
-
-  public static final int TARGET_FIELD_NUMBER = 3;
-  private com.google.protobuf.LazyStringList target_;
-  /**
-   * <pre>
-   * Target Nodes. A list of node names. The named nodes will be run by the
-   * callable but their outputs will not be returned.
-   * </pre>
-   *
-   * <code>repeated string target = 3;</code>
-   */
-  public com.google.protobuf.ProtocolStringList
-      getTargetList() {
-    return target_;
-  }
-  /**
-   * <pre>
-   * Target Nodes. A list of node names. The named nodes will be run by the
-   * callable but their outputs will not be returned.
-   * </pre>
-   *
-   * <code>repeated string target = 3;</code>
-   */
-  public int getTargetCount() {
-    return target_.size();
-  }
-  /**
-   * <pre>
-   * Target Nodes. A list of node names. The named nodes will be run by the
-   * callable but their outputs will not be returned.
-   * </pre>
-   *
-   * <code>repeated string target = 3;</code>
-   */
-  public java.lang.String getTarget(int index) {
-    return target_.get(index);
-  }
-  /**
-   * <pre>
-   * Target Nodes. A list of node names. The named nodes will be run by the
-   * callable but their outputs will not be returned.
-   * </pre>
-   *
-   * <code>repeated string target = 3;</code>
-   */
-  public com.google.protobuf.ByteString
-      getTargetBytes(int index) {
-    return target_.getByteString(index);
-  }
-
-  public static final int RUN_OPTIONS_FIELD_NUMBER = 4;
-  private org.tensorflow.proto.framework.RunOptions runOptions_;
-  /**
-   * <pre>
-   * Options that will be applied to each run.
-   * </pre>
-   *
-   * <code>.tensorflow.RunOptions run_options = 4;</code>
-   */
-  public boolean hasRunOptions() {
-    return runOptions_ != null;
-  }
-  /**
-   * <pre>
-   * Options that will be applied to each run.
-   * </pre>
-   *
-   * <code>.tensorflow.RunOptions run_options = 4;</code>
-   */
-  public org.tensorflow.proto.framework.RunOptions getRunOptions() {
-    return runOptions_ == null ? org.tensorflow.proto.framework.RunOptions.getDefaultInstance() : runOptions_;
-  }
-  /**
-   * <pre>
-   * Options that will be applied to each run.
-   * </pre>
-   *
-   * <code>.tensorflow.RunOptions run_options = 4;</code>
-   */
-  public org.tensorflow.proto.framework.RunOptionsOrBuilder getRunOptionsOrBuilder() {
-    return getRunOptions();
-  }
-
-  public static final int TENSOR_CONNECTION_FIELD_NUMBER = 5;
-  private java.util.List<org.tensorflow.proto.framework.TensorConnection> tensorConnection_;
-  /**
-   * <pre>
-   * Tensors to be connected in the callable. Each TensorConnection denotes
-   * a pair of tensors in the graph, between which an edge will be created
-   * in the callable.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.TensorConnection tensor_connection = 5;</code>
-   */
-  public java.util.List<org.tensorflow.proto.framework.TensorConnection> getTensorConnectionList() {
-    return tensorConnection_;
-  }
-  /**
-   * <pre>
-   * Tensors to be connected in the callable. Each TensorConnection denotes
-   * a pair of tensors in the graph, between which an edge will be created
-   * in the callable.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.TensorConnection tensor_connection = 5;</code>
-   */
-  public java.util.List<? extends org.tensorflow.proto.framework.TensorConnectionOrBuilder> 
-      getTensorConnectionOrBuilderList() {
-    return tensorConnection_;
-  }
-  /**
-   * <pre>
-   * Tensors to be connected in the callable. Each TensorConnection denotes
-   * a pair of tensors in the graph, between which an edge will be created
-   * in the callable.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.TensorConnection tensor_connection = 5;</code>
-   */
-  public int getTensorConnectionCount() {
-    return tensorConnection_.size();
-  }
-  /**
-   * <pre>
-   * Tensors to be connected in the callable. Each TensorConnection denotes
-   * a pair of tensors in the graph, between which an edge will be created
-   * in the callable.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.TensorConnection tensor_connection = 5;</code>
-   */
-  public org.tensorflow.proto.framework.TensorConnection getTensorConnection(int index) {
-    return tensorConnection_.get(index);
-  }
-  /**
-   * <pre>
-   * Tensors to be connected in the callable. Each TensorConnection denotes
-   * a pair of tensors in the graph, between which an edge will be created
-   * in the callable.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.TensorConnection tensor_connection = 5;</code>
-   */
-  public org.tensorflow.proto.framework.TensorConnectionOrBuilder getTensorConnectionOrBuilder(
-      int index) {
-    return tensorConnection_.get(index);
-  }
-
-  public static final int FEED_DEVICES_FIELD_NUMBER = 6;
-  private static final class FeedDevicesDefaultEntryHolder {
-    static final com.google.protobuf.MapEntry<
-        java.lang.String, java.lang.String> defaultEntry =
-            com.google.protobuf.MapEntry
-            .<java.lang.String, java.lang.String>newDefaultInstance(
-                org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_CallableOptions_FeedDevicesEntry_descriptor, 
-                com.google.protobuf.WireFormat.FieldType.STRING,
-                "",
-                com.google.protobuf.WireFormat.FieldType.STRING,
-                "");
-  }
-  private com.google.protobuf.MapField<
-      java.lang.String, java.lang.String> feedDevices_;
-  private com.google.protobuf.MapField<java.lang.String, java.lang.String>
-  internalGetFeedDevices() {
-    if (feedDevices_ == null) {
-      return com.google.protobuf.MapField.emptyMapField(
-          FeedDevicesDefaultEntryHolder.defaultEntry);
-    }
-    return feedDevices_;
-  }
-
-  public int getFeedDevicesCount() {
-    return internalGetFeedDevices().getMap().size();
-  }
-  /**
-   * <pre>
-   * The Tensor objects fed in the callable and fetched from the callable
-   * are expected to be backed by host (CPU) memory by default.
-   * The options below allow changing that - feeding tensors backed by
-   * device memory, or returning tensors that are backed by device memory.
-   * The maps below map the name of a feed/fetch tensor (which appears in
-   * 'feed' or 'fetch' fields above), to the fully qualified name of the device
-   * owning the memory backing the contents of the tensor.
-   * For example, creating a callable with the following options:
-   * CallableOptions {
-   *   feed: "a:0"
-   *   feed: "b:0"
-   *   fetch: "x:0"
-   *   fetch: "y:0"
-   *   feed_devices: {
-   *     "a:0": "/job:localhost/replica:0/task:0/device:GPU:0"
-   *   }
-   *   fetch_devices: {
-   *     "y:0": "/job:localhost/replica:0/task:0/device:GPU:0"
-   *  }
-   * }
-   * means that the Callable expects:
-   * - The first argument ("a:0") is a Tensor backed by GPU memory.
-   * - The second argument ("b:0") is a Tensor backed by host memory.
-   * and of its return values:
-   * - The first output ("x:0") will be backed by host memory.
-   * - The second output ("y:0") will be backed by GPU memory.
-   * FEEDS:
-   * It is the responsibility of the caller to ensure that the memory of the fed
-   * tensors will be correctly initialized and synchronized before it is
-   * accessed by operations executed during the call to Session::RunCallable().
-   * This is typically ensured by using the TensorFlow memory allocators
-   * (Device::GetAllocator()) to create the Tensor to be fed.
-   * Alternatively, for CUDA-enabled GPU devices, this typically means that the
-   * operation that produced the contents of the tensor has completed, i.e., the
-   * CUDA stream has been synchronized (e.g., via cuCtxSynchronize() or
-   * cuStreamSynchronize()).
-   * </pre>
-   *
-   * <code>map&lt;string, string&gt; feed_devices = 6;</code>
-   */
-
-  public boolean containsFeedDevices(
-      java.lang.String key) {
-    if (key == null) { throw new java.lang.NullPointerException(); }
-    return internalGetFeedDevices().getMap().containsKey(key);
-  }
-  /**
-   * Use {@link #getFeedDevicesMap()} instead.
-   */
-  @java.lang.Deprecated
-  public java.util.Map<java.lang.String, java.lang.String> getFeedDevices() {
-    return getFeedDevicesMap();
-  }
-  /**
-   * <pre>
-   * The Tensor objects fed in the callable and fetched from the callable
-   * are expected to be backed by host (CPU) memory by default.
-   * The options below allow changing that - feeding tensors backed by
-   * device memory, or returning tensors that are backed by device memory.
-   * The maps below map the name of a feed/fetch tensor (which appears in
-   * 'feed' or 'fetch' fields above), to the fully qualified name of the device
-   * owning the memory backing the contents of the tensor.
-   * For example, creating a callable with the following options:
-   * CallableOptions {
-   *   feed: "a:0"
-   *   feed: "b:0"
-   *   fetch: "x:0"
-   *   fetch: "y:0"
-   *   feed_devices: {
-   *     "a:0": "/job:localhost/replica:0/task:0/device:GPU:0"
-   *   }
-   *   fetch_devices: {
-   *     "y:0": "/job:localhost/replica:0/task:0/device:GPU:0"
-   *  }
-   * }
-   * means that the Callable expects:
-   * - The first argument ("a:0") is a Tensor backed by GPU memory.
-   * - The second argument ("b:0") is a Tensor backed by host memory.
-   * and of its return values:
-   * - The first output ("x:0") will be backed by host memory.
-   * - The second output ("y:0") will be backed by GPU memory.
-   * FEEDS:
-   * It is the responsibility of the caller to ensure that the memory of the fed
-   * tensors will be correctly initialized and synchronized before it is
-   * accessed by operations executed during the call to Session::RunCallable().
-   * This is typically ensured by using the TensorFlow memory allocators
-   * (Device::GetAllocator()) to create the Tensor to be fed.
-   * Alternatively, for CUDA-enabled GPU devices, this typically means that the
-   * operation that produced the contents of the tensor has completed, i.e., the
-   * CUDA stream has been synchronized (e.g., via cuCtxSynchronize() or
-   * cuStreamSynchronize()).
-   * </pre>
-   *
-   * <code>map&lt;string, string&gt; feed_devices = 6;</code>
-   */
-
-  public java.util.Map<java.lang.String, java.lang.String> getFeedDevicesMap() {
-    return internalGetFeedDevices().getMap();
-  }
-  /**
-   * <pre>
-   * The Tensor objects fed in the callable and fetched from the callable
-   * are expected to be backed by host (CPU) memory by default.
-   * The options below allow changing that - feeding tensors backed by
-   * device memory, or returning tensors that are backed by device memory.
-   * The maps below map the name of a feed/fetch tensor (which appears in
-   * 'feed' or 'fetch' fields above), to the fully qualified name of the device
-   * owning the memory backing the contents of the tensor.
-   * For example, creating a callable with the following options:
-   * CallableOptions {
-   *   feed: "a:0"
-   *   feed: "b:0"
-   *   fetch: "x:0"
-   *   fetch: "y:0"
-   *   feed_devices: {
-   *     "a:0": "/job:localhost/replica:0/task:0/device:GPU:0"
-   *   }
-   *   fetch_devices: {
-   *     "y:0": "/job:localhost/replica:0/task:0/device:GPU:0"
-   *  }
-   * }
-   * means that the Callable expects:
-   * - The first argument ("a:0") is a Tensor backed by GPU memory.
-   * - The second argument ("b:0") is a Tensor backed by host memory.
-   * and of its return values:
-   * - The first output ("x:0") will be backed by host memory.
-   * - The second output ("y:0") will be backed by GPU memory.
-   * FEEDS:
-   * It is the responsibility of the caller to ensure that the memory of the fed
-   * tensors will be correctly initialized and synchronized before it is
-   * accessed by operations executed during the call to Session::RunCallable().
-   * This is typically ensured by using the TensorFlow memory allocators
-   * (Device::GetAllocator()) to create the Tensor to be fed.
-   * Alternatively, for CUDA-enabled GPU devices, this typically means that the
-   * operation that produced the contents of the tensor has completed, i.e., the
-   * CUDA stream has been synchronized (e.g., via cuCtxSynchronize() or
-   * cuStreamSynchronize()).
-   * </pre>
-   *
-   * <code>map&lt;string, string&gt; feed_devices = 6;</code>
-   */
-
-  public java.lang.String getFeedDevicesOrDefault(
-      java.lang.String key,
-      java.lang.String defaultValue) {
-    if (key == null) { throw new java.lang.NullPointerException(); }
-    java.util.Map<java.lang.String, java.lang.String> map =
-        internalGetFeedDevices().getMap();
-    return map.containsKey(key) ? map.get(key) : defaultValue;
-  }
-  /**
-   * <pre>
-   * The Tensor objects fed in the callable and fetched from the callable
-   * are expected to be backed by host (CPU) memory by default.
-   * The options below allow changing that - feeding tensors backed by
-   * device memory, or returning tensors that are backed by device memory.
-   * The maps below map the name of a feed/fetch tensor (which appears in
-   * 'feed' or 'fetch' fields above), to the fully qualified name of the device
-   * owning the memory backing the contents of the tensor.
-   * For example, creating a callable with the following options:
-   * CallableOptions {
-   *   feed: "a:0"
-   *   feed: "b:0"
-   *   fetch: "x:0"
-   *   fetch: "y:0"
-   *   feed_devices: {
-   *     "a:0": "/job:localhost/replica:0/task:0/device:GPU:0"
-   *   }
-   *   fetch_devices: {
-   *     "y:0": "/job:localhost/replica:0/task:0/device:GPU:0"
-   *  }
-   * }
-   * means that the Callable expects:
-   * - The first argument ("a:0") is a Tensor backed by GPU memory.
-   * - The second argument ("b:0") is a Tensor backed by host memory.
-   * and of its return values:
-   * - The first output ("x:0") will be backed by host memory.
-   * - The second output ("y:0") will be backed by GPU memory.
-   * FEEDS:
-   * It is the responsibility of the caller to ensure that the memory of the fed
-   * tensors will be correctly initialized and synchronized before it is
-   * accessed by operations executed during the call to Session::RunCallable().
-   * This is typically ensured by using the TensorFlow memory allocators
-   * (Device::GetAllocator()) to create the Tensor to be fed.
-   * Alternatively, for CUDA-enabled GPU devices, this typically means that the
-   * operation that produced the contents of the tensor has completed, i.e., the
-   * CUDA stream has been synchronized (e.g., via cuCtxSynchronize() or
-   * cuStreamSynchronize()).
-   * </pre>
-   *
-   * <code>map&lt;string, string&gt; feed_devices = 6;</code>
-   */
-
-  public java.lang.String getFeedDevicesOrThrow(
-      java.lang.String key) {
-    if (key == null) { throw new java.lang.NullPointerException(); }
-    java.util.Map<java.lang.String, java.lang.String> map =
-        internalGetFeedDevices().getMap();
-    if (!map.containsKey(key)) {
-      throw new java.lang.IllegalArgumentException();
-    }
-    return map.get(key);
-  }
-
-  public static final int FETCH_DEVICES_FIELD_NUMBER = 7;
-  private static final class FetchDevicesDefaultEntryHolder {
-    static final com.google.protobuf.MapEntry<
-        java.lang.String, java.lang.String> defaultEntry =
-            com.google.protobuf.MapEntry
-            .<java.lang.String, java.lang.String>newDefaultInstance(
-                org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_CallableOptions_FetchDevicesEntry_descriptor, 
-                com.google.protobuf.WireFormat.FieldType.STRING,
-                "",
-                com.google.protobuf.WireFormat.FieldType.STRING,
-                "");
-  }
-  private com.google.protobuf.MapField<
-      java.lang.String, java.lang.String> fetchDevices_;
-  private com.google.protobuf.MapField<java.lang.String, java.lang.String>
-  internalGetFetchDevices() {
-    if (fetchDevices_ == null) {
-      return com.google.protobuf.MapField.emptyMapField(
-          FetchDevicesDefaultEntryHolder.defaultEntry);
-    }
-    return fetchDevices_;
-  }
-
-  public int getFetchDevicesCount() {
-    return internalGetFetchDevices().getMap().size();
-  }
-  /**
-   * <code>map&lt;string, string&gt; fetch_devices = 7;</code>
-   */
-
-  public boolean containsFetchDevices(
-      java.lang.String key) {
-    if (key == null) { throw new java.lang.NullPointerException(); }
-    return internalGetFetchDevices().getMap().containsKey(key);
-  }
-  /**
-   * Use {@link #getFetchDevicesMap()} instead.
-   */
-  @java.lang.Deprecated
-  public java.util.Map<java.lang.String, java.lang.String> getFetchDevices() {
-    return getFetchDevicesMap();
-  }
-  /**
-   * <code>map&lt;string, string&gt; fetch_devices = 7;</code>
-   */
-
-  public java.util.Map<java.lang.String, java.lang.String> getFetchDevicesMap() {
-    return internalGetFetchDevices().getMap();
-  }
-  /**
-   * <code>map&lt;string, string&gt; fetch_devices = 7;</code>
-   */
-
-  public java.lang.String getFetchDevicesOrDefault(
-      java.lang.String key,
-      java.lang.String defaultValue) {
-    if (key == null) { throw new java.lang.NullPointerException(); }
-    java.util.Map<java.lang.String, java.lang.String> map =
-        internalGetFetchDevices().getMap();
-    return map.containsKey(key) ? map.get(key) : defaultValue;
-  }
-  /**
-   * <code>map&lt;string, string&gt; fetch_devices = 7;</code>
-   */
-
-  public java.lang.String getFetchDevicesOrThrow(
-      java.lang.String key) {
-    if (key == null) { throw new java.lang.NullPointerException(); }
-    java.util.Map<java.lang.String, java.lang.String> map =
-        internalGetFetchDevices().getMap();
-    if (!map.containsKey(key)) {
-      throw new java.lang.IllegalArgumentException();
-    }
-    return map.get(key);
-  }
-
-  public static final int FETCH_SKIP_SYNC_FIELD_NUMBER = 8;
-  private boolean fetchSkipSync_;
-  /**
-   * <pre>
-   * By default, RunCallable() will synchronize the GPU stream before returning
-   * fetched tensors on a GPU device, to ensure that the values in those tensors
-   * have been produced. This simplifies interacting with the tensors, but
-   * potentially incurs a performance hit.
-   * If this options is set to true, the caller is responsible for ensuring
-   * that the values in the fetched tensors have been produced before they are
-   * used. The caller can do this by invoking `Device::Sync()` on the underlying
-   * device(s), or by feeding the tensors back to the same Session using
-   * `feed_devices` with the same corresponding device name.
-   * </pre>
-   *
-   * <code>bool fetch_skip_sync = 8;</code>
-   */
-  public boolean getFetchSkipSync() {
-    return fetchSkipSync_;
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    for (int i = 0; i < feed_.size(); i++) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 1, feed_.getRaw(i));
-    }
-    for (int i = 0; i < fetch_.size(); i++) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 2, fetch_.getRaw(i));
-    }
-    for (int i = 0; i < target_.size(); i++) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 3, target_.getRaw(i));
-    }
-    if (runOptions_ != null) {
-      output.writeMessage(4, getRunOptions());
-    }
-    for (int i = 0; i < tensorConnection_.size(); i++) {
-      output.writeMessage(5, tensorConnection_.get(i));
-    }
-    com.google.protobuf.GeneratedMessageV3
-      .serializeStringMapTo(
-        output,
-        internalGetFeedDevices(),
-        FeedDevicesDefaultEntryHolder.defaultEntry,
-        6);
-    com.google.protobuf.GeneratedMessageV3
-      .serializeStringMapTo(
-        output,
-        internalGetFetchDevices(),
-        FetchDevicesDefaultEntryHolder.defaultEntry,
-        7);
-    if (fetchSkipSync_ != false) {
-      output.writeBool(8, fetchSkipSync_);
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    {
-      int dataSize = 0;
-      for (int i = 0; i < feed_.size(); i++) {
-        dataSize += computeStringSizeNoTag(feed_.getRaw(i));
-      }
-      size += dataSize;
-      size += 1 * getFeedList().size();
-    }
-    {
-      int dataSize = 0;
-      for (int i = 0; i < fetch_.size(); i++) {
-        dataSize += computeStringSizeNoTag(fetch_.getRaw(i));
-      }
-      size += dataSize;
-      size += 1 * getFetchList().size();
-    }
-    {
-      int dataSize = 0;
-      for (int i = 0; i < target_.size(); i++) {
-        dataSize += computeStringSizeNoTag(target_.getRaw(i));
-      }
-      size += dataSize;
-      size += 1 * getTargetList().size();
-    }
-    if (runOptions_ != null) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(4, getRunOptions());
-    }
-    for (int i = 0; i < tensorConnection_.size(); i++) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(5, tensorConnection_.get(i));
-    }
-    for (java.util.Map.Entry<java.lang.String, java.lang.String> entry
-         : internalGetFeedDevices().getMap().entrySet()) {
-      com.google.protobuf.MapEntry<java.lang.String, java.lang.String>
-      feedDevices__ = FeedDevicesDefaultEntryHolder.defaultEntry.newBuilderForType()
-          .setKey(entry.getKey())
-          .setValue(entry.getValue())
-          .build();
-      size += com.google.protobuf.CodedOutputStream
-          .computeMessageSize(6, feedDevices__);
-    }
-    for (java.util.Map.Entry<java.lang.String, java.lang.String> entry
-         : internalGetFetchDevices().getMap().entrySet()) {
-      com.google.protobuf.MapEntry<java.lang.String, java.lang.String>
-      fetchDevices__ = FetchDevicesDefaultEntryHolder.defaultEntry.newBuilderForType()
-          .setKey(entry.getKey())
-          .setValue(entry.getValue())
-          .build();
-      size += com.google.protobuf.CodedOutputStream
-          .computeMessageSize(7, fetchDevices__);
-    }
-    if (fetchSkipSync_ != false) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeBoolSize(8, fetchSkipSync_);
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.framework.CallableOptions)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.framework.CallableOptions other = (org.tensorflow.proto.framework.CallableOptions) obj;
-
-    if (!getFeedList()
-        .equals(other.getFeedList())) return false;
-    if (!getFetchList()
-        .equals(other.getFetchList())) return false;
-    if (!getTargetList()
-        .equals(other.getTargetList())) return false;
-    if (hasRunOptions() != other.hasRunOptions()) return false;
-    if (hasRunOptions()) {
-      if (!getRunOptions()
-          .equals(other.getRunOptions())) return false;
-    }
-    if (!getTensorConnectionList()
-        .equals(other.getTensorConnectionList())) return false;
-    if (!internalGetFeedDevices().equals(
-        other.internalGetFeedDevices())) return false;
-    if (!internalGetFetchDevices().equals(
-        other.internalGetFetchDevices())) return false;
-    if (getFetchSkipSync()
-        != other.getFetchSkipSync()) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    if (getFeedCount() > 0) {
-      hash = (37 * hash) + FEED_FIELD_NUMBER;
-      hash = (53 * hash) + getFeedList().hashCode();
-    }
-    if (getFetchCount() > 0) {
-      hash = (37 * hash) + FETCH_FIELD_NUMBER;
-      hash = (53 * hash) + getFetchList().hashCode();
-    }
-    if (getTargetCount() > 0) {
-      hash = (37 * hash) + TARGET_FIELD_NUMBER;
-      hash = (53 * hash) + getTargetList().hashCode();
-    }
-    if (hasRunOptions()) {
-      hash = (37 * hash) + RUN_OPTIONS_FIELD_NUMBER;
-      hash = (53 * hash) + getRunOptions().hashCode();
-    }
-    if (getTensorConnectionCount() > 0) {
-      hash = (37 * hash) + TENSOR_CONNECTION_FIELD_NUMBER;
-      hash = (53 * hash) + getTensorConnectionList().hashCode();
-    }
-    if (!internalGetFeedDevices().getMap().isEmpty()) {
-      hash = (37 * hash) + FEED_DEVICES_FIELD_NUMBER;
-      hash = (53 * hash) + internalGetFeedDevices().hashCode();
-    }
-    if (!internalGetFetchDevices().getMap().isEmpty()) {
-      hash = (37 * hash) + FETCH_DEVICES_FIELD_NUMBER;
-      hash = (53 * hash) + internalGetFetchDevices().hashCode();
-    }
-    hash = (37 * hash) + FETCH_SKIP_SYNC_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashBoolean(
-        getFetchSkipSync());
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.framework.CallableOptions parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.CallableOptions parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.CallableOptions parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.CallableOptions parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.CallableOptions parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.CallableOptions parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.CallableOptions parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.CallableOptions parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.CallableOptions parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.CallableOptions parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.CallableOptions parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.CallableOptions parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.framework.CallableOptions prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * <pre>
-   * Defines a subgraph in another `GraphDef` as a set of feed points and nodes
-   * to be fetched or executed.
-   * Compare with the arguments to `Session::Run()`.
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.CallableOptions}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.CallableOptions)
-      org.tensorflow.proto.framework.CallableOptionsOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_CallableOptions_descriptor;
-    }
-
-    @SuppressWarnings({"rawtypes"})
-    protected com.google.protobuf.MapField internalGetMapField(
-        int number) {
-      switch (number) {
-        case 6:
-          return internalGetFeedDevices();
-        case 7:
-          return internalGetFetchDevices();
-        default:
-          throw new RuntimeException(
-              "Invalid map field number: " + number);
-      }
-    }
-    @SuppressWarnings({"rawtypes"})
-    protected com.google.protobuf.MapField internalGetMutableMapField(
-        int number) {
-      switch (number) {
-        case 6:
-          return internalGetMutableFeedDevices();
-        case 7:
-          return internalGetMutableFetchDevices();
-        default:
-          throw new RuntimeException(
-              "Invalid map field number: " + number);
-      }
-    }
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_CallableOptions_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.CallableOptions.class, org.tensorflow.proto.framework.CallableOptions.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.framework.CallableOptions.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-        getTensorConnectionFieldBuilder();
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      feed_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-      bitField0_ = (bitField0_ & ~0x00000001);
-      fetch_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-      bitField0_ = (bitField0_ & ~0x00000002);
-      target_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-      bitField0_ = (bitField0_ & ~0x00000004);
-      if (runOptionsBuilder_ == null) {
-        runOptions_ = null;
-      } else {
-        runOptions_ = null;
-        runOptionsBuilder_ = null;
-      }
-      if (tensorConnectionBuilder_ == null) {
-        tensorConnection_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000008);
-      } else {
-        tensorConnectionBuilder_.clear();
-      }
-      internalGetMutableFeedDevices().clear();
-      internalGetMutableFetchDevices().clear();
-      fetchSkipSync_ = false;
-
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_CallableOptions_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.CallableOptions getDefaultInstanceForType() {
-      return org.tensorflow.proto.framework.CallableOptions.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.CallableOptions build() {
-      org.tensorflow.proto.framework.CallableOptions result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.CallableOptions buildPartial() {
-      org.tensorflow.proto.framework.CallableOptions result = new org.tensorflow.proto.framework.CallableOptions(this);
-      int from_bitField0_ = bitField0_;
-      if (((bitField0_ & 0x00000001) != 0)) {
-        feed_ = feed_.getUnmodifiableView();
-        bitField0_ = (bitField0_ & ~0x00000001);
-      }
-      result.feed_ = feed_;
-      if (((bitField0_ & 0x00000002) != 0)) {
-        fetch_ = fetch_.getUnmodifiableView();
-        bitField0_ = (bitField0_ & ~0x00000002);
-      }
-      result.fetch_ = fetch_;
-      if (((bitField0_ & 0x00000004) != 0)) {
-        target_ = target_.getUnmodifiableView();
-        bitField0_ = (bitField0_ & ~0x00000004);
-      }
-      result.target_ = target_;
-      if (runOptionsBuilder_ == null) {
-        result.runOptions_ = runOptions_;
-      } else {
-        result.runOptions_ = runOptionsBuilder_.build();
-      }
-      if (tensorConnectionBuilder_ == null) {
-        if (((bitField0_ & 0x00000008) != 0)) {
-          tensorConnection_ = java.util.Collections.unmodifiableList(tensorConnection_);
-          bitField0_ = (bitField0_ & ~0x00000008);
-        }
-        result.tensorConnection_ = tensorConnection_;
-      } else {
-        result.tensorConnection_ = tensorConnectionBuilder_.build();
-      }
-      result.feedDevices_ = internalGetFeedDevices();
-      result.feedDevices_.makeImmutable();
-      result.fetchDevices_ = internalGetFetchDevices();
-      result.fetchDevices_.makeImmutable();
-      result.fetchSkipSync_ = fetchSkipSync_;
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.framework.CallableOptions) {
-        return mergeFrom((org.tensorflow.proto.framework.CallableOptions)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.framework.CallableOptions other) {
-      if (other == org.tensorflow.proto.framework.CallableOptions.getDefaultInstance()) return this;
-      if (!other.feed_.isEmpty()) {
-        if (feed_.isEmpty()) {
-          feed_ = other.feed_;
-          bitField0_ = (bitField0_ & ~0x00000001);
-        } else {
-          ensureFeedIsMutable();
-          feed_.addAll(other.feed_);
-        }
-        onChanged();
-      }
-      if (!other.fetch_.isEmpty()) {
-        if (fetch_.isEmpty()) {
-          fetch_ = other.fetch_;
-          bitField0_ = (bitField0_ & ~0x00000002);
-        } else {
-          ensureFetchIsMutable();
-          fetch_.addAll(other.fetch_);
-        }
-        onChanged();
-      }
-      if (!other.target_.isEmpty()) {
-        if (target_.isEmpty()) {
-          target_ = other.target_;
-          bitField0_ = (bitField0_ & ~0x00000004);
-        } else {
-          ensureTargetIsMutable();
-          target_.addAll(other.target_);
-        }
-        onChanged();
-      }
-      if (other.hasRunOptions()) {
-        mergeRunOptions(other.getRunOptions());
-      }
-      if (tensorConnectionBuilder_ == null) {
-        if (!other.tensorConnection_.isEmpty()) {
-          if (tensorConnection_.isEmpty()) {
-            tensorConnection_ = other.tensorConnection_;
-            bitField0_ = (bitField0_ & ~0x00000008);
-          } else {
-            ensureTensorConnectionIsMutable();
-            tensorConnection_.addAll(other.tensorConnection_);
-          }
-          onChanged();
-        }
-      } else {
-        if (!other.tensorConnection_.isEmpty()) {
-          if (tensorConnectionBuilder_.isEmpty()) {
-            tensorConnectionBuilder_.dispose();
-            tensorConnectionBuilder_ = null;
-            tensorConnection_ = other.tensorConnection_;
-            bitField0_ = (bitField0_ & ~0x00000008);
-            tensorConnectionBuilder_ = 
-              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
-                 getTensorConnectionFieldBuilder() : null;
-          } else {
-            tensorConnectionBuilder_.addAllMessages(other.tensorConnection_);
-          }
-        }
-      }
-      internalGetMutableFeedDevices().mergeFrom(
-          other.internalGetFeedDevices());
-      internalGetMutableFetchDevices().mergeFrom(
-          other.internalGetFetchDevices());
-      if (other.getFetchSkipSync() != false) {
-        setFetchSkipSync(other.getFetchSkipSync());
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.framework.CallableOptions parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.framework.CallableOptions) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-    private int bitField0_;
-
-    private com.google.protobuf.LazyStringList feed_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-    private void ensureFeedIsMutable() {
-      if (!((bitField0_ & 0x00000001) != 0)) {
-        feed_ = new com.google.protobuf.LazyStringArrayList(feed_);
-        bitField0_ |= 0x00000001;
-       }
-    }
-    /**
-     * <pre>
-     * Tensors to be fed in the callable. Each feed is the name of a tensor.
-     * </pre>
-     *
-     * <code>repeated string feed = 1;</code>
-     */
-    public com.google.protobuf.ProtocolStringList
-        getFeedList() {
-      return feed_.getUnmodifiableView();
-    }
-    /**
-     * <pre>
-     * Tensors to be fed in the callable. Each feed is the name of a tensor.
-     * </pre>
-     *
-     * <code>repeated string feed = 1;</code>
-     */
-    public int getFeedCount() {
-      return feed_.size();
-    }
-    /**
-     * <pre>
-     * Tensors to be fed in the callable. Each feed is the name of a tensor.
-     * </pre>
-     *
-     * <code>repeated string feed = 1;</code>
-     */
-    public java.lang.String getFeed(int index) {
-      return feed_.get(index);
-    }
-    /**
-     * <pre>
-     * Tensors to be fed in the callable. Each feed is the name of a tensor.
-     * </pre>
-     *
-     * <code>repeated string feed = 1;</code>
-     */
-    public com.google.protobuf.ByteString
-        getFeedBytes(int index) {
-      return feed_.getByteString(index);
-    }
-    /**
-     * <pre>
-     * Tensors to be fed in the callable. Each feed is the name of a tensor.
-     * </pre>
-     *
-     * <code>repeated string feed = 1;</code>
-     */
-    public Builder setFeed(
-        int index, java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  ensureFeedIsMutable();
-      feed_.set(index, value);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Tensors to be fed in the callable. Each feed is the name of a tensor.
-     * </pre>
-     *
-     * <code>repeated string feed = 1;</code>
-     */
-    public Builder addFeed(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  ensureFeedIsMutable();
-      feed_.add(value);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Tensors to be fed in the callable. Each feed is the name of a tensor.
-     * </pre>
-     *
-     * <code>repeated string feed = 1;</code>
-     */
-    public Builder addAllFeed(
-        java.lang.Iterable<java.lang.String> values) {
-      ensureFeedIsMutable();
-      com.google.protobuf.AbstractMessageLite.Builder.addAll(
-          values, feed_);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Tensors to be fed in the callable. Each feed is the name of a tensor.
-     * </pre>
-     *
-     * <code>repeated string feed = 1;</code>
-     */
-    public Builder clearFeed() {
-      feed_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-      bitField0_ = (bitField0_ & ~0x00000001);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Tensors to be fed in the callable. Each feed is the name of a tensor.
-     * </pre>
-     *
-     * <code>repeated string feed = 1;</code>
-     */
-    public Builder addFeedBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      ensureFeedIsMutable();
-      feed_.add(value);
-      onChanged();
-      return this;
-    }
-
-    private com.google.protobuf.LazyStringList fetch_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-    private void ensureFetchIsMutable() {
-      if (!((bitField0_ & 0x00000002) != 0)) {
-        fetch_ = new com.google.protobuf.LazyStringArrayList(fetch_);
-        bitField0_ |= 0x00000002;
-       }
-    }
-    /**
-     * <pre>
-     * Fetches. A list of tensor names. The caller of the callable expects a
-     * tensor to be returned for each fetch[i] (see RunStepResponse.tensor). The
-     * order of specified fetches does not change the execution order.
-     * </pre>
-     *
-     * <code>repeated string fetch = 2;</code>
-     */
-    public com.google.protobuf.ProtocolStringList
-        getFetchList() {
-      return fetch_.getUnmodifiableView();
-    }
-    /**
-     * <pre>
-     * Fetches. A list of tensor names. The caller of the callable expects a
-     * tensor to be returned for each fetch[i] (see RunStepResponse.tensor). The
-     * order of specified fetches does not change the execution order.
-     * </pre>
-     *
-     * <code>repeated string fetch = 2;</code>
-     */
-    public int getFetchCount() {
-      return fetch_.size();
-    }
-    /**
-     * <pre>
-     * Fetches. A list of tensor names. The caller of the callable expects a
-     * tensor to be returned for each fetch[i] (see RunStepResponse.tensor). The
-     * order of specified fetches does not change the execution order.
-     * </pre>
-     *
-     * <code>repeated string fetch = 2;</code>
-     */
-    public java.lang.String getFetch(int index) {
-      return fetch_.get(index);
-    }
-    /**
-     * <pre>
-     * Fetches. A list of tensor names. The caller of the callable expects a
-     * tensor to be returned for each fetch[i] (see RunStepResponse.tensor). The
-     * order of specified fetches does not change the execution order.
-     * </pre>
-     *
-     * <code>repeated string fetch = 2;</code>
-     */
-    public com.google.protobuf.ByteString
-        getFetchBytes(int index) {
-      return fetch_.getByteString(index);
-    }
-    /**
-     * <pre>
-     * Fetches. A list of tensor names. The caller of the callable expects a
-     * tensor to be returned for each fetch[i] (see RunStepResponse.tensor). The
-     * order of specified fetches does not change the execution order.
-     * </pre>
-     *
-     * <code>repeated string fetch = 2;</code>
-     */
-    public Builder setFetch(
-        int index, java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  ensureFetchIsMutable();
-      fetch_.set(index, value);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Fetches. A list of tensor names. The caller of the callable expects a
-     * tensor to be returned for each fetch[i] (see RunStepResponse.tensor). The
-     * order of specified fetches does not change the execution order.
-     * </pre>
-     *
-     * <code>repeated string fetch = 2;</code>
-     */
-    public Builder addFetch(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  ensureFetchIsMutable();
-      fetch_.add(value);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Fetches. A list of tensor names. The caller of the callable expects a
-     * tensor to be returned for each fetch[i] (see RunStepResponse.tensor). The
-     * order of specified fetches does not change the execution order.
-     * </pre>
-     *
-     * <code>repeated string fetch = 2;</code>
-     */
-    public Builder addAllFetch(
-        java.lang.Iterable<java.lang.String> values) {
-      ensureFetchIsMutable();
-      com.google.protobuf.AbstractMessageLite.Builder.addAll(
-          values, fetch_);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Fetches. A list of tensor names. The caller of the callable expects a
-     * tensor to be returned for each fetch[i] (see RunStepResponse.tensor). The
-     * order of specified fetches does not change the execution order.
-     * </pre>
-     *
-     * <code>repeated string fetch = 2;</code>
-     */
-    public Builder clearFetch() {
-      fetch_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-      bitField0_ = (bitField0_ & ~0x00000002);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Fetches. A list of tensor names. The caller of the callable expects a
-     * tensor to be returned for each fetch[i] (see RunStepResponse.tensor). The
-     * order of specified fetches does not change the execution order.
-     * </pre>
-     *
-     * <code>repeated string fetch = 2;</code>
-     */
-    public Builder addFetchBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      ensureFetchIsMutable();
-      fetch_.add(value);
-      onChanged();
-      return this;
-    }
-
-    private com.google.protobuf.LazyStringList target_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-    private void ensureTargetIsMutable() {
-      if (!((bitField0_ & 0x00000004) != 0)) {
-        target_ = new com.google.protobuf.LazyStringArrayList(target_);
-        bitField0_ |= 0x00000004;
-       }
-    }
-    /**
-     * <pre>
-     * Target Nodes. A list of node names. The named nodes will be run by the
-     * callable but their outputs will not be returned.
-     * </pre>
-     *
-     * <code>repeated string target = 3;</code>
-     */
-    public com.google.protobuf.ProtocolStringList
-        getTargetList() {
-      return target_.getUnmodifiableView();
-    }
-    /**
-     * <pre>
-     * Target Nodes. A list of node names. The named nodes will be run by the
-     * callable but their outputs will not be returned.
-     * </pre>
-     *
-     * <code>repeated string target = 3;</code>
-     */
-    public int getTargetCount() {
-      return target_.size();
-    }
-    /**
-     * <pre>
-     * Target Nodes. A list of node names. The named nodes will be run by the
-     * callable but their outputs will not be returned.
-     * </pre>
-     *
-     * <code>repeated string target = 3;</code>
-     */
-    public java.lang.String getTarget(int index) {
-      return target_.get(index);
-    }
-    /**
-     * <pre>
-     * Target Nodes. A list of node names. The named nodes will be run by the
-     * callable but their outputs will not be returned.
-     * </pre>
-     *
-     * <code>repeated string target = 3;</code>
-     */
-    public com.google.protobuf.ByteString
-        getTargetBytes(int index) {
-      return target_.getByteString(index);
-    }
-    /**
-     * <pre>
-     * Target Nodes. A list of node names. The named nodes will be run by the
-     * callable but their outputs will not be returned.
-     * </pre>
-     *
-     * <code>repeated string target = 3;</code>
-     */
-    public Builder setTarget(
-        int index, java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  ensureTargetIsMutable();
-      target_.set(index, value);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Target Nodes. A list of node names. The named nodes will be run by the
-     * callable but their outputs will not be returned.
-     * </pre>
-     *
-     * <code>repeated string target = 3;</code>
-     */
-    public Builder addTarget(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  ensureTargetIsMutable();
-      target_.add(value);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Target Nodes. A list of node names. The named nodes will be run by the
-     * callable but their outputs will not be returned.
-     * </pre>
-     *
-     * <code>repeated string target = 3;</code>
-     */
-    public Builder addAllTarget(
-        java.lang.Iterable<java.lang.String> values) {
-      ensureTargetIsMutable();
-      com.google.protobuf.AbstractMessageLite.Builder.addAll(
-          values, target_);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Target Nodes. A list of node names. The named nodes will be run by the
-     * callable but their outputs will not be returned.
-     * </pre>
-     *
-     * <code>repeated string target = 3;</code>
-     */
-    public Builder clearTarget() {
-      target_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-      bitField0_ = (bitField0_ & ~0x00000004);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Target Nodes. A list of node names. The named nodes will be run by the
-     * callable but their outputs will not be returned.
-     * </pre>
-     *
-     * <code>repeated string target = 3;</code>
-     */
-    public Builder addTargetBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      ensureTargetIsMutable();
-      target_.add(value);
-      onChanged();
-      return this;
-    }
-
-    private org.tensorflow.proto.framework.RunOptions runOptions_;
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.RunOptions, org.tensorflow.proto.framework.RunOptions.Builder, org.tensorflow.proto.framework.RunOptionsOrBuilder> runOptionsBuilder_;
-    /**
-     * <pre>
-     * Options that will be applied to each run.
-     * </pre>
-     *
-     * <code>.tensorflow.RunOptions run_options = 4;</code>
-     */
-    public boolean hasRunOptions() {
-      return runOptionsBuilder_ != null || runOptions_ != null;
-    }
-    /**
-     * <pre>
-     * Options that will be applied to each run.
-     * </pre>
-     *
-     * <code>.tensorflow.RunOptions run_options = 4;</code>
-     */
-    public org.tensorflow.proto.framework.RunOptions getRunOptions() {
-      if (runOptionsBuilder_ == null) {
-        return runOptions_ == null ? org.tensorflow.proto.framework.RunOptions.getDefaultInstance() : runOptions_;
-      } else {
-        return runOptionsBuilder_.getMessage();
-      }
-    }
-    /**
-     * <pre>
-     * Options that will be applied to each run.
-     * </pre>
-     *
-     * <code>.tensorflow.RunOptions run_options = 4;</code>
-     */
-    public Builder setRunOptions(org.tensorflow.proto.framework.RunOptions value) {
-      if (runOptionsBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        runOptions_ = value;
-        onChanged();
-      } else {
-        runOptionsBuilder_.setMessage(value);
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Options that will be applied to each run.
-     * </pre>
-     *
-     * <code>.tensorflow.RunOptions run_options = 4;</code>
-     */
-    public Builder setRunOptions(
-        org.tensorflow.proto.framework.RunOptions.Builder builderForValue) {
-      if (runOptionsBuilder_ == null) {
-        runOptions_ = builderForValue.build();
-        onChanged();
-      } else {
-        runOptionsBuilder_.setMessage(builderForValue.build());
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Options that will be applied to each run.
-     * </pre>
-     *
-     * <code>.tensorflow.RunOptions run_options = 4;</code>
-     */
-    public Builder mergeRunOptions(org.tensorflow.proto.framework.RunOptions value) {
-      if (runOptionsBuilder_ == null) {
-        if (runOptions_ != null) {
-          runOptions_ =
-            org.tensorflow.proto.framework.RunOptions.newBuilder(runOptions_).mergeFrom(value).buildPartial();
-        } else {
-          runOptions_ = value;
-        }
-        onChanged();
-      } else {
-        runOptionsBuilder_.mergeFrom(value);
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Options that will be applied to each run.
-     * </pre>
-     *
-     * <code>.tensorflow.RunOptions run_options = 4;</code>
-     */
-    public Builder clearRunOptions() {
-      if (runOptionsBuilder_ == null) {
-        runOptions_ = null;
-        onChanged();
-      } else {
-        runOptions_ = null;
-        runOptionsBuilder_ = null;
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Options that will be applied to each run.
-     * </pre>
-     *
-     * <code>.tensorflow.RunOptions run_options = 4;</code>
-     */
-    public org.tensorflow.proto.framework.RunOptions.Builder getRunOptionsBuilder() {
-      
-      onChanged();
-      return getRunOptionsFieldBuilder().getBuilder();
-    }
-    /**
-     * <pre>
-     * Options that will be applied to each run.
-     * </pre>
-     *
-     * <code>.tensorflow.RunOptions run_options = 4;</code>
-     */
-    public org.tensorflow.proto.framework.RunOptionsOrBuilder getRunOptionsOrBuilder() {
-      if (runOptionsBuilder_ != null) {
-        return runOptionsBuilder_.getMessageOrBuilder();
-      } else {
-        return runOptions_ == null ?
-            org.tensorflow.proto.framework.RunOptions.getDefaultInstance() : runOptions_;
-      }
-    }
-    /**
-     * <pre>
-     * Options that will be applied to each run.
-     * </pre>
-     *
-     * <code>.tensorflow.RunOptions run_options = 4;</code>
-     */
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.RunOptions, org.tensorflow.proto.framework.RunOptions.Builder, org.tensorflow.proto.framework.RunOptionsOrBuilder> 
-        getRunOptionsFieldBuilder() {
-      if (runOptionsBuilder_ == null) {
-        runOptionsBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-            org.tensorflow.proto.framework.RunOptions, org.tensorflow.proto.framework.RunOptions.Builder, org.tensorflow.proto.framework.RunOptionsOrBuilder>(
-                getRunOptions(),
-                getParentForChildren(),
-                isClean());
-        runOptions_ = null;
-      }
-      return runOptionsBuilder_;
-    }
-
-    private java.util.List<org.tensorflow.proto.framework.TensorConnection> tensorConnection_ =
-      java.util.Collections.emptyList();
-    private void ensureTensorConnectionIsMutable() {
-      if (!((bitField0_ & 0x00000008) != 0)) {
-        tensorConnection_ = new java.util.ArrayList<org.tensorflow.proto.framework.TensorConnection>(tensorConnection_);
-        bitField0_ |= 0x00000008;
-       }
-    }
-
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.framework.TensorConnection, org.tensorflow.proto.framework.TensorConnection.Builder, org.tensorflow.proto.framework.TensorConnectionOrBuilder> tensorConnectionBuilder_;
-
-    /**
-     * <pre>
-     * Tensors to be connected in the callable. Each TensorConnection denotes
-     * a pair of tensors in the graph, between which an edge will be created
-     * in the callable.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorConnection tensor_connection = 5;</code>
-     */
-    public java.util.List<org.tensorflow.proto.framework.TensorConnection> getTensorConnectionList() {
-      if (tensorConnectionBuilder_ == null) {
-        return java.util.Collections.unmodifiableList(tensorConnection_);
-      } else {
-        return tensorConnectionBuilder_.getMessageList();
-      }
-    }
-    /**
-     * <pre>
-     * Tensors to be connected in the callable. Each TensorConnection denotes
-     * a pair of tensors in the graph, between which an edge will be created
-     * in the callable.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorConnection tensor_connection = 5;</code>
-     */
-    public int getTensorConnectionCount() {
-      if (tensorConnectionBuilder_ == null) {
-        return tensorConnection_.size();
-      } else {
-        return tensorConnectionBuilder_.getCount();
-      }
-    }
-    /**
-     * <pre>
-     * Tensors to be connected in the callable. Each TensorConnection denotes
-     * a pair of tensors in the graph, between which an edge will be created
-     * in the callable.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorConnection tensor_connection = 5;</code>
-     */
-    public org.tensorflow.proto.framework.TensorConnection getTensorConnection(int index) {
-      if (tensorConnectionBuilder_ == null) {
-        return tensorConnection_.get(index);
-      } else {
-        return tensorConnectionBuilder_.getMessage(index);
-      }
-    }
-    /**
-     * <pre>
-     * Tensors to be connected in the callable. Each TensorConnection denotes
-     * a pair of tensors in the graph, between which an edge will be created
-     * in the callable.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorConnection tensor_connection = 5;</code>
-     */
-    public Builder setTensorConnection(
-        int index, org.tensorflow.proto.framework.TensorConnection value) {
-      if (tensorConnectionBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureTensorConnectionIsMutable();
-        tensorConnection_.set(index, value);
-        onChanged();
-      } else {
-        tensorConnectionBuilder_.setMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Tensors to be connected in the callable. Each TensorConnection denotes
-     * a pair of tensors in the graph, between which an edge will be created
-     * in the callable.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorConnection tensor_connection = 5;</code>
-     */
-    public Builder setTensorConnection(
-        int index, org.tensorflow.proto.framework.TensorConnection.Builder builderForValue) {
-      if (tensorConnectionBuilder_ == null) {
-        ensureTensorConnectionIsMutable();
-        tensorConnection_.set(index, builderForValue.build());
-        onChanged();
-      } else {
-        tensorConnectionBuilder_.setMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Tensors to be connected in the callable. Each TensorConnection denotes
-     * a pair of tensors in the graph, between which an edge will be created
-     * in the callable.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorConnection tensor_connection = 5;</code>
-     */
-    public Builder addTensorConnection(org.tensorflow.proto.framework.TensorConnection value) {
-      if (tensorConnectionBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureTensorConnectionIsMutable();
-        tensorConnection_.add(value);
-        onChanged();
-      } else {
-        tensorConnectionBuilder_.addMessage(value);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Tensors to be connected in the callable. Each TensorConnection denotes
-     * a pair of tensors in the graph, between which an edge will be created
-     * in the callable.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorConnection tensor_connection = 5;</code>
-     */
-    public Builder addTensorConnection(
-        int index, org.tensorflow.proto.framework.TensorConnection value) {
-      if (tensorConnectionBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureTensorConnectionIsMutable();
-        tensorConnection_.add(index, value);
-        onChanged();
-      } else {
-        tensorConnectionBuilder_.addMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Tensors to be connected in the callable. Each TensorConnection denotes
-     * a pair of tensors in the graph, between which an edge will be created
-     * in the callable.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorConnection tensor_connection = 5;</code>
-     */
-    public Builder addTensorConnection(
-        org.tensorflow.proto.framework.TensorConnection.Builder builderForValue) {
-      if (tensorConnectionBuilder_ == null) {
-        ensureTensorConnectionIsMutable();
-        tensorConnection_.add(builderForValue.build());
-        onChanged();
-      } else {
-        tensorConnectionBuilder_.addMessage(builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Tensors to be connected in the callable. Each TensorConnection denotes
-     * a pair of tensors in the graph, between which an edge will be created
-     * in the callable.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorConnection tensor_connection = 5;</code>
-     */
-    public Builder addTensorConnection(
-        int index, org.tensorflow.proto.framework.TensorConnection.Builder builderForValue) {
-      if (tensorConnectionBuilder_ == null) {
-        ensureTensorConnectionIsMutable();
-        tensorConnection_.add(index, builderForValue.build());
-        onChanged();
-      } else {
-        tensorConnectionBuilder_.addMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Tensors to be connected in the callable. Each TensorConnection denotes
-     * a pair of tensors in the graph, between which an edge will be created
-     * in the callable.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorConnection tensor_connection = 5;</code>
-     */
-    public Builder addAllTensorConnection(
-        java.lang.Iterable<? extends org.tensorflow.proto.framework.TensorConnection> values) {
-      if (tensorConnectionBuilder_ == null) {
-        ensureTensorConnectionIsMutable();
-        com.google.protobuf.AbstractMessageLite.Builder.addAll(
-            values, tensorConnection_);
-        onChanged();
-      } else {
-        tensorConnectionBuilder_.addAllMessages(values);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Tensors to be connected in the callable. Each TensorConnection denotes
-     * a pair of tensors in the graph, between which an edge will be created
-     * in the callable.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorConnection tensor_connection = 5;</code>
-     */
-    public Builder clearTensorConnection() {
-      if (tensorConnectionBuilder_ == null) {
-        tensorConnection_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000008);
-        onChanged();
-      } else {
-        tensorConnectionBuilder_.clear();
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Tensors to be connected in the callable. Each TensorConnection denotes
-     * a pair of tensors in the graph, between which an edge will be created
-     * in the callable.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorConnection tensor_connection = 5;</code>
-     */
-    public Builder removeTensorConnection(int index) {
-      if (tensorConnectionBuilder_ == null) {
-        ensureTensorConnectionIsMutable();
-        tensorConnection_.remove(index);
-        onChanged();
-      } else {
-        tensorConnectionBuilder_.remove(index);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Tensors to be connected in the callable. Each TensorConnection denotes
-     * a pair of tensors in the graph, between which an edge will be created
-     * in the callable.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorConnection tensor_connection = 5;</code>
-     */
-    public org.tensorflow.proto.framework.TensorConnection.Builder getTensorConnectionBuilder(
-        int index) {
-      return getTensorConnectionFieldBuilder().getBuilder(index);
-    }
-    /**
-     * <pre>
-     * Tensors to be connected in the callable. Each TensorConnection denotes
-     * a pair of tensors in the graph, between which an edge will be created
-     * in the callable.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorConnection tensor_connection = 5;</code>
-     */
-    public org.tensorflow.proto.framework.TensorConnectionOrBuilder getTensorConnectionOrBuilder(
-        int index) {
-      if (tensorConnectionBuilder_ == null) {
-        return tensorConnection_.get(index);  } else {
-        return tensorConnectionBuilder_.getMessageOrBuilder(index);
-      }
-    }
-    /**
-     * <pre>
-     * Tensors to be connected in the callable. Each TensorConnection denotes
-     * a pair of tensors in the graph, between which an edge will be created
-     * in the callable.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorConnection tensor_connection = 5;</code>
-     */
-    public java.util.List<? extends org.tensorflow.proto.framework.TensorConnectionOrBuilder> 
-         getTensorConnectionOrBuilderList() {
-      if (tensorConnectionBuilder_ != null) {
-        return tensorConnectionBuilder_.getMessageOrBuilderList();
-      } else {
-        return java.util.Collections.unmodifiableList(tensorConnection_);
-      }
-    }
-    /**
-     * <pre>
-     * Tensors to be connected in the callable. Each TensorConnection denotes
-     * a pair of tensors in the graph, between which an edge will be created
-     * in the callable.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorConnection tensor_connection = 5;</code>
-     */
-    public org.tensorflow.proto.framework.TensorConnection.Builder addTensorConnectionBuilder() {
-      return getTensorConnectionFieldBuilder().addBuilder(
-          org.tensorflow.proto.framework.TensorConnection.getDefaultInstance());
-    }
-    /**
-     * <pre>
-     * Tensors to be connected in the callable. Each TensorConnection denotes
-     * a pair of tensors in the graph, between which an edge will be created
-     * in the callable.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorConnection tensor_connection = 5;</code>
-     */
-    public org.tensorflow.proto.framework.TensorConnection.Builder addTensorConnectionBuilder(
-        int index) {
-      return getTensorConnectionFieldBuilder().addBuilder(
-          index, org.tensorflow.proto.framework.TensorConnection.getDefaultInstance());
-    }
-    /**
-     * <pre>
-     * Tensors to be connected in the callable. Each TensorConnection denotes
-     * a pair of tensors in the graph, between which an edge will be created
-     * in the callable.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorConnection tensor_connection = 5;</code>
-     */
-    public java.util.List<org.tensorflow.proto.framework.TensorConnection.Builder> 
-         getTensorConnectionBuilderList() {
-      return getTensorConnectionFieldBuilder().getBuilderList();
-    }
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.framework.TensorConnection, org.tensorflow.proto.framework.TensorConnection.Builder, org.tensorflow.proto.framework.TensorConnectionOrBuilder> 
-        getTensorConnectionFieldBuilder() {
-      if (tensorConnectionBuilder_ == null) {
-        tensorConnectionBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
-            org.tensorflow.proto.framework.TensorConnection, org.tensorflow.proto.framework.TensorConnection.Builder, org.tensorflow.proto.framework.TensorConnectionOrBuilder>(
-                tensorConnection_,
-                ((bitField0_ & 0x00000008) != 0),
-                getParentForChildren(),
-                isClean());
-        tensorConnection_ = null;
-      }
-      return tensorConnectionBuilder_;
-    }
-
-    private com.google.protobuf.MapField<
-        java.lang.String, java.lang.String> feedDevices_;
-    private com.google.protobuf.MapField<java.lang.String, java.lang.String>
-    internalGetFeedDevices() {
-      if (feedDevices_ == null) {
-        return com.google.protobuf.MapField.emptyMapField(
-            FeedDevicesDefaultEntryHolder.defaultEntry);
-      }
-      return feedDevices_;
-    }
-    private com.google.protobuf.MapField<java.lang.String, java.lang.String>
-    internalGetMutableFeedDevices() {
-      onChanged();;
-      if (feedDevices_ == null) {
-        feedDevices_ = com.google.protobuf.MapField.newMapField(
-            FeedDevicesDefaultEntryHolder.defaultEntry);
-      }
-      if (!feedDevices_.isMutable()) {
-        feedDevices_ = feedDevices_.copy();
-      }
-      return feedDevices_;
-    }
-
-    public int getFeedDevicesCount() {
-      return internalGetFeedDevices().getMap().size();
-    }
-    /**
-     * <pre>
-     * The Tensor objects fed in the callable and fetched from the callable
-     * are expected to be backed by host (CPU) memory by default.
-     * The options below allow changing that - feeding tensors backed by
-     * device memory, or returning tensors that are backed by device memory.
-     * The maps below map the name of a feed/fetch tensor (which appears in
-     * 'feed' or 'fetch' fields above), to the fully qualified name of the device
-     * owning the memory backing the contents of the tensor.
-     * For example, creating a callable with the following options:
-     * CallableOptions {
-     *   feed: "a:0"
-     *   feed: "b:0"
-     *   fetch: "x:0"
-     *   fetch: "y:0"
-     *   feed_devices: {
-     *     "a:0": "/job:localhost/replica:0/task:0/device:GPU:0"
-     *   }
-     *   fetch_devices: {
-     *     "y:0": "/job:localhost/replica:0/task:0/device:GPU:0"
-     *  }
-     * }
-     * means that the Callable expects:
-     * - The first argument ("a:0") is a Tensor backed by GPU memory.
-     * - The second argument ("b:0") is a Tensor backed by host memory.
-     * and of its return values:
-     * - The first output ("x:0") will be backed by host memory.
-     * - The second output ("y:0") will be backed by GPU memory.
-     * FEEDS:
-     * It is the responsibility of the caller to ensure that the memory of the fed
-     * tensors will be correctly initialized and synchronized before it is
-     * accessed by operations executed during the call to Session::RunCallable().
-     * This is typically ensured by using the TensorFlow memory allocators
-     * (Device::GetAllocator()) to create the Tensor to be fed.
-     * Alternatively, for CUDA-enabled GPU devices, this typically means that the
-     * operation that produced the contents of the tensor has completed, i.e., the
-     * CUDA stream has been synchronized (e.g., via cuCtxSynchronize() or
-     * cuStreamSynchronize()).
-     * </pre>
-     *
-     * <code>map&lt;string, string&gt; feed_devices = 6;</code>
-     */
-
-    public boolean containsFeedDevices(
-        java.lang.String key) {
-      if (key == null) { throw new java.lang.NullPointerException(); }
-      return internalGetFeedDevices().getMap().containsKey(key);
-    }
-    /**
-     * Use {@link #getFeedDevicesMap()} instead.
-     */
-    @java.lang.Deprecated
-    public java.util.Map<java.lang.String, java.lang.String> getFeedDevices() {
-      return getFeedDevicesMap();
-    }
-    /**
-     * <pre>
-     * The Tensor objects fed in the callable and fetched from the callable
-     * are expected to be backed by host (CPU) memory by default.
-     * The options below allow changing that - feeding tensors backed by
-     * device memory, or returning tensors that are backed by device memory.
-     * The maps below map the name of a feed/fetch tensor (which appears in
-     * 'feed' or 'fetch' fields above), to the fully qualified name of the device
-     * owning the memory backing the contents of the tensor.
-     * For example, creating a callable with the following options:
-     * CallableOptions {
-     *   feed: "a:0"
-     *   feed: "b:0"
-     *   fetch: "x:0"
-     *   fetch: "y:0"
-     *   feed_devices: {
-     *     "a:0": "/job:localhost/replica:0/task:0/device:GPU:0"
-     *   }
-     *   fetch_devices: {
-     *     "y:0": "/job:localhost/replica:0/task:0/device:GPU:0"
-     *  }
-     * }
-     * means that the Callable expects:
-     * - The first argument ("a:0") is a Tensor backed by GPU memory.
-     * - The second argument ("b:0") is a Tensor backed by host memory.
-     * and of its return values:
-     * - The first output ("x:0") will be backed by host memory.
-     * - The second output ("y:0") will be backed by GPU memory.
-     * FEEDS:
-     * It is the responsibility of the caller to ensure that the memory of the fed
-     * tensors will be correctly initialized and synchronized before it is
-     * accessed by operations executed during the call to Session::RunCallable().
-     * This is typically ensured by using the TensorFlow memory allocators
-     * (Device::GetAllocator()) to create the Tensor to be fed.
-     * Alternatively, for CUDA-enabled GPU devices, this typically means that the
-     * operation that produced the contents of the tensor has completed, i.e., the
-     * CUDA stream has been synchronized (e.g., via cuCtxSynchronize() or
-     * cuStreamSynchronize()).
-     * </pre>
-     *
-     * <code>map&lt;string, string&gt; feed_devices = 6;</code>
-     */
-
-    public java.util.Map<java.lang.String, java.lang.String> getFeedDevicesMap() {
-      return internalGetFeedDevices().getMap();
-    }
-    /**
-     * <pre>
-     * The Tensor objects fed in the callable and fetched from the callable
-     * are expected to be backed by host (CPU) memory by default.
-     * The options below allow changing that - feeding tensors backed by
-     * device memory, or returning tensors that are backed by device memory.
-     * The maps below map the name of a feed/fetch tensor (which appears in
-     * 'feed' or 'fetch' fields above), to the fully qualified name of the device
-     * owning the memory backing the contents of the tensor.
-     * For example, creating a callable with the following options:
-     * CallableOptions {
-     *   feed: "a:0"
-     *   feed: "b:0"
-     *   fetch: "x:0"
-     *   fetch: "y:0"
-     *   feed_devices: {
-     *     "a:0": "/job:localhost/replica:0/task:0/device:GPU:0"
-     *   }
-     *   fetch_devices: {
-     *     "y:0": "/job:localhost/replica:0/task:0/device:GPU:0"
-     *  }
-     * }
-     * means that the Callable expects:
-     * - The first argument ("a:0") is a Tensor backed by GPU memory.
-     * - The second argument ("b:0") is a Tensor backed by host memory.
-     * and of its return values:
-     * - The first output ("x:0") will be backed by host memory.
-     * - The second output ("y:0") will be backed by GPU memory.
-     * FEEDS:
-     * It is the responsibility of the caller to ensure that the memory of the fed
-     * tensors will be correctly initialized and synchronized before it is
-     * accessed by operations executed during the call to Session::RunCallable().
-     * This is typically ensured by using the TensorFlow memory allocators
-     * (Device::GetAllocator()) to create the Tensor to be fed.
-     * Alternatively, for CUDA-enabled GPU devices, this typically means that the
-     * operation that produced the contents of the tensor has completed, i.e., the
-     * CUDA stream has been synchronized (e.g., via cuCtxSynchronize() or
-     * cuStreamSynchronize()).
-     * </pre>
-     *
-     * <code>map&lt;string, string&gt; feed_devices = 6;</code>
-     */
-
-    public java.lang.String getFeedDevicesOrDefault(
-        java.lang.String key,
-        java.lang.String defaultValue) {
-      if (key == null) { throw new java.lang.NullPointerException(); }
-      java.util.Map<java.lang.String, java.lang.String> map =
-          internalGetFeedDevices().getMap();
-      return map.containsKey(key) ? map.get(key) : defaultValue;
-    }
-    /**
-     * <pre>
-     * The Tensor objects fed in the callable and fetched from the callable
-     * are expected to be backed by host (CPU) memory by default.
-     * The options below allow changing that - feeding tensors backed by
-     * device memory, or returning tensors that are backed by device memory.
-     * The maps below map the name of a feed/fetch tensor (which appears in
-     * 'feed' or 'fetch' fields above), to the fully qualified name of the device
-     * owning the memory backing the contents of the tensor.
-     * For example, creating a callable with the following options:
-     * CallableOptions {
-     *   feed: "a:0"
-     *   feed: "b:0"
-     *   fetch: "x:0"
-     *   fetch: "y:0"
-     *   feed_devices: {
-     *     "a:0": "/job:localhost/replica:0/task:0/device:GPU:0"
-     *   }
-     *   fetch_devices: {
-     *     "y:0": "/job:localhost/replica:0/task:0/device:GPU:0"
-     *  }
-     * }
-     * means that the Callable expects:
-     * - The first argument ("a:0") is a Tensor backed by GPU memory.
-     * - The second argument ("b:0") is a Tensor backed by host memory.
-     * and of its return values:
-     * - The first output ("x:0") will be backed by host memory.
-     * - The second output ("y:0") will be backed by GPU memory.
-     * FEEDS:
-     * It is the responsibility of the caller to ensure that the memory of the fed
-     * tensors will be correctly initialized and synchronized before it is
-     * accessed by operations executed during the call to Session::RunCallable().
-     * This is typically ensured by using the TensorFlow memory allocators
-     * (Device::GetAllocator()) to create the Tensor to be fed.
-     * Alternatively, for CUDA-enabled GPU devices, this typically means that the
-     * operation that produced the contents of the tensor has completed, i.e., the
-     * CUDA stream has been synchronized (e.g., via cuCtxSynchronize() or
-     * cuStreamSynchronize()).
-     * </pre>
-     *
-     * <code>map&lt;string, string&gt; feed_devices = 6;</code>
-     */
-
-    public java.lang.String getFeedDevicesOrThrow(
-        java.lang.String key) {
-      if (key == null) { throw new java.lang.NullPointerException(); }
-      java.util.Map<java.lang.String, java.lang.String> map =
-          internalGetFeedDevices().getMap();
-      if (!map.containsKey(key)) {
-        throw new java.lang.IllegalArgumentException();
-      }
-      return map.get(key);
-    }
-
-    public Builder clearFeedDevices() {
-      internalGetMutableFeedDevices().getMutableMap()
-          .clear();
-      return this;
-    }
-    /**
-     * <pre>
-     * The Tensor objects fed in the callable and fetched from the callable
-     * are expected to be backed by host (CPU) memory by default.
-     * The options below allow changing that - feeding tensors backed by
-     * device memory, or returning tensors that are backed by device memory.
-     * The maps below map the name of a feed/fetch tensor (which appears in
-     * 'feed' or 'fetch' fields above), to the fully qualified name of the device
-     * owning the memory backing the contents of the tensor.
-     * For example, creating a callable with the following options:
-     * CallableOptions {
-     *   feed: "a:0"
-     *   feed: "b:0"
-     *   fetch: "x:0"
-     *   fetch: "y:0"
-     *   feed_devices: {
-     *     "a:0": "/job:localhost/replica:0/task:0/device:GPU:0"
-     *   }
-     *   fetch_devices: {
-     *     "y:0": "/job:localhost/replica:0/task:0/device:GPU:0"
-     *  }
-     * }
-     * means that the Callable expects:
-     * - The first argument ("a:0") is a Tensor backed by GPU memory.
-     * - The second argument ("b:0") is a Tensor backed by host memory.
-     * and of its return values:
-     * - The first output ("x:0") will be backed by host memory.
-     * - The second output ("y:0") will be backed by GPU memory.
-     * FEEDS:
-     * It is the responsibility of the caller to ensure that the memory of the fed
-     * tensors will be correctly initialized and synchronized before it is
-     * accessed by operations executed during the call to Session::RunCallable().
-     * This is typically ensured by using the TensorFlow memory allocators
-     * (Device::GetAllocator()) to create the Tensor to be fed.
-     * Alternatively, for CUDA-enabled GPU devices, this typically means that the
-     * operation that produced the contents of the tensor has completed, i.e., the
-     * CUDA stream has been synchronized (e.g., via cuCtxSynchronize() or
-     * cuStreamSynchronize()).
-     * </pre>
-     *
-     * <code>map&lt;string, string&gt; feed_devices = 6;</code>
-     */
-
-    public Builder removeFeedDevices(
-        java.lang.String key) {
-      if (key == null) { throw new java.lang.NullPointerException(); }
-      internalGetMutableFeedDevices().getMutableMap()
-          .remove(key);
-      return this;
-    }
-    /**
-     * Use alternate mutation accessors instead.
-     */
-    @java.lang.Deprecated
-    public java.util.Map<java.lang.String, java.lang.String>
-    getMutableFeedDevices() {
-      return internalGetMutableFeedDevices().getMutableMap();
-    }
-    /**
-     * <pre>
-     * The Tensor objects fed in the callable and fetched from the callable
-     * are expected to be backed by host (CPU) memory by default.
-     * The options below allow changing that - feeding tensors backed by
-     * device memory, or returning tensors that are backed by device memory.
-     * The maps below map the name of a feed/fetch tensor (which appears in
-     * 'feed' or 'fetch' fields above), to the fully qualified name of the device
-     * owning the memory backing the contents of the tensor.
-     * For example, creating a callable with the following options:
-     * CallableOptions {
-     *   feed: "a:0"
-     *   feed: "b:0"
-     *   fetch: "x:0"
-     *   fetch: "y:0"
-     *   feed_devices: {
-     *     "a:0": "/job:localhost/replica:0/task:0/device:GPU:0"
-     *   }
-     *   fetch_devices: {
-     *     "y:0": "/job:localhost/replica:0/task:0/device:GPU:0"
-     *  }
-     * }
-     * means that the Callable expects:
-     * - The first argument ("a:0") is a Tensor backed by GPU memory.
-     * - The second argument ("b:0") is a Tensor backed by host memory.
-     * and of its return values:
-     * - The first output ("x:0") will be backed by host memory.
-     * - The second output ("y:0") will be backed by GPU memory.
-     * FEEDS:
-     * It is the responsibility of the caller to ensure that the memory of the fed
-     * tensors will be correctly initialized and synchronized before it is
-     * accessed by operations executed during the call to Session::RunCallable().
-     * This is typically ensured by using the TensorFlow memory allocators
-     * (Device::GetAllocator()) to create the Tensor to be fed.
-     * Alternatively, for CUDA-enabled GPU devices, this typically means that the
-     * operation that produced the contents of the tensor has completed, i.e., the
-     * CUDA stream has been synchronized (e.g., via cuCtxSynchronize() or
-     * cuStreamSynchronize()).
-     * </pre>
-     *
-     * <code>map&lt;string, string&gt; feed_devices = 6;</code>
-     */
-    public Builder putFeedDevices(
-        java.lang.String key,
-        java.lang.String value) {
-      if (key == null) { throw new java.lang.NullPointerException(); }
-      if (value == null) { throw new java.lang.NullPointerException(); }
-      internalGetMutableFeedDevices().getMutableMap()
-          .put(key, value);
-      return this;
-    }
-    /**
-     * <pre>
-     * The Tensor objects fed in the callable and fetched from the callable
-     * are expected to be backed by host (CPU) memory by default.
-     * The options below allow changing that - feeding tensors backed by
-     * device memory, or returning tensors that are backed by device memory.
-     * The maps below map the name of a feed/fetch tensor (which appears in
-     * 'feed' or 'fetch' fields above), to the fully qualified name of the device
-     * owning the memory backing the contents of the tensor.
-     * For example, creating a callable with the following options:
-     * CallableOptions {
-     *   feed: "a:0"
-     *   feed: "b:0"
-     *   fetch: "x:0"
-     *   fetch: "y:0"
-     *   feed_devices: {
-     *     "a:0": "/job:localhost/replica:0/task:0/device:GPU:0"
-     *   }
-     *   fetch_devices: {
-     *     "y:0": "/job:localhost/replica:0/task:0/device:GPU:0"
-     *  }
-     * }
-     * means that the Callable expects:
-     * - The first argument ("a:0") is a Tensor backed by GPU memory.
-     * - The second argument ("b:0") is a Tensor backed by host memory.
-     * and of its return values:
-     * - The first output ("x:0") will be backed by host memory.
-     * - The second output ("y:0") will be backed by GPU memory.
-     * FEEDS:
-     * It is the responsibility of the caller to ensure that the memory of the fed
-     * tensors will be correctly initialized and synchronized before it is
-     * accessed by operations executed during the call to Session::RunCallable().
-     * This is typically ensured by using the TensorFlow memory allocators
-     * (Device::GetAllocator()) to create the Tensor to be fed.
-     * Alternatively, for CUDA-enabled GPU devices, this typically means that the
-     * operation that produced the contents of the tensor has completed, i.e., the
-     * CUDA stream has been synchronized (e.g., via cuCtxSynchronize() or
-     * cuStreamSynchronize()).
-     * </pre>
-     *
-     * <code>map&lt;string, string&gt; feed_devices = 6;</code>
-     */
-
-    public Builder putAllFeedDevices(
-        java.util.Map<java.lang.String, java.lang.String> values) {
-      internalGetMutableFeedDevices().getMutableMap()
-          .putAll(values);
-      return this;
-    }
-
-    private com.google.protobuf.MapField<
-        java.lang.String, java.lang.String> fetchDevices_;
-    private com.google.protobuf.MapField<java.lang.String, java.lang.String>
-    internalGetFetchDevices() {
-      if (fetchDevices_ == null) {
-        return com.google.protobuf.MapField.emptyMapField(
-            FetchDevicesDefaultEntryHolder.defaultEntry);
-      }
-      return fetchDevices_;
-    }
-    private com.google.protobuf.MapField<java.lang.String, java.lang.String>
-    internalGetMutableFetchDevices() {
-      onChanged();;
-      if (fetchDevices_ == null) {
-        fetchDevices_ = com.google.protobuf.MapField.newMapField(
-            FetchDevicesDefaultEntryHolder.defaultEntry);
-      }
-      if (!fetchDevices_.isMutable()) {
-        fetchDevices_ = fetchDevices_.copy();
-      }
-      return fetchDevices_;
-    }
-
-    public int getFetchDevicesCount() {
-      return internalGetFetchDevices().getMap().size();
-    }
-    /**
-     * <code>map&lt;string, string&gt; fetch_devices = 7;</code>
-     */
-
-    public boolean containsFetchDevices(
-        java.lang.String key) {
-      if (key == null) { throw new java.lang.NullPointerException(); }
-      return internalGetFetchDevices().getMap().containsKey(key);
-    }
-    /**
-     * Use {@link #getFetchDevicesMap()} instead.
-     */
-    @java.lang.Deprecated
-    public java.util.Map<java.lang.String, java.lang.String> getFetchDevices() {
-      return getFetchDevicesMap();
-    }
-    /**
-     * <code>map&lt;string, string&gt; fetch_devices = 7;</code>
-     */
-
-    public java.util.Map<java.lang.String, java.lang.String> getFetchDevicesMap() {
-      return internalGetFetchDevices().getMap();
-    }
-    /**
-     * <code>map&lt;string, string&gt; fetch_devices = 7;</code>
-     */
-
-    public java.lang.String getFetchDevicesOrDefault(
-        java.lang.String key,
-        java.lang.String defaultValue) {
-      if (key == null) { throw new java.lang.NullPointerException(); }
-      java.util.Map<java.lang.String, java.lang.String> map =
-          internalGetFetchDevices().getMap();
-      return map.containsKey(key) ? map.get(key) : defaultValue;
-    }
-    /**
-     * <code>map&lt;string, string&gt; fetch_devices = 7;</code>
-     */
-
-    public java.lang.String getFetchDevicesOrThrow(
-        java.lang.String key) {
-      if (key == null) { throw new java.lang.NullPointerException(); }
-      java.util.Map<java.lang.String, java.lang.String> map =
-          internalGetFetchDevices().getMap();
-      if (!map.containsKey(key)) {
-        throw new java.lang.IllegalArgumentException();
-      }
-      return map.get(key);
-    }
-
-    public Builder clearFetchDevices() {
-      internalGetMutableFetchDevices().getMutableMap()
-          .clear();
-      return this;
-    }
-    /**
-     * <code>map&lt;string, string&gt; fetch_devices = 7;</code>
-     */
-
-    public Builder removeFetchDevices(
-        java.lang.String key) {
-      if (key == null) { throw new java.lang.NullPointerException(); }
-      internalGetMutableFetchDevices().getMutableMap()
-          .remove(key);
-      return this;
-    }
-    /**
-     * Use alternate mutation accessors instead.
-     */
-    @java.lang.Deprecated
-    public java.util.Map<java.lang.String, java.lang.String>
-    getMutableFetchDevices() {
-      return internalGetMutableFetchDevices().getMutableMap();
-    }
-    /**
-     * <code>map&lt;string, string&gt; fetch_devices = 7;</code>
-     */
-    public Builder putFetchDevices(
-        java.lang.String key,
-        java.lang.String value) {
-      if (key == null) { throw new java.lang.NullPointerException(); }
-      if (value == null) { throw new java.lang.NullPointerException(); }
-      internalGetMutableFetchDevices().getMutableMap()
-          .put(key, value);
-      return this;
-    }
-    /**
-     * <code>map&lt;string, string&gt; fetch_devices = 7;</code>
-     */
-
-    public Builder putAllFetchDevices(
-        java.util.Map<java.lang.String, java.lang.String> values) {
-      internalGetMutableFetchDevices().getMutableMap()
-          .putAll(values);
-      return this;
-    }
-
-    private boolean fetchSkipSync_ ;
-    /**
-     * <pre>
-     * By default, RunCallable() will synchronize the GPU stream before returning
-     * fetched tensors on a GPU device, to ensure that the values in those tensors
-     * have been produced. This simplifies interacting with the tensors, but
-     * potentially incurs a performance hit.
-     * If this options is set to true, the caller is responsible for ensuring
-     * that the values in the fetched tensors have been produced before they are
-     * used. The caller can do this by invoking `Device::Sync()` on the underlying
-     * device(s), or by feeding the tensors back to the same Session using
-     * `feed_devices` with the same corresponding device name.
-     * </pre>
-     *
-     * <code>bool fetch_skip_sync = 8;</code>
-     */
-    public boolean getFetchSkipSync() {
-      return fetchSkipSync_;
-    }
-    /**
-     * <pre>
-     * By default, RunCallable() will synchronize the GPU stream before returning
-     * fetched tensors on a GPU device, to ensure that the values in those tensors
-     * have been produced. This simplifies interacting with the tensors, but
-     * potentially incurs a performance hit.
-     * If this options is set to true, the caller is responsible for ensuring
-     * that the values in the fetched tensors have been produced before they are
-     * used. The caller can do this by invoking `Device::Sync()` on the underlying
-     * device(s), or by feeding the tensors back to the same Session using
-     * `feed_devices` with the same corresponding device name.
-     * </pre>
-     *
-     * <code>bool fetch_skip_sync = 8;</code>
-     */
-    public Builder setFetchSkipSync(boolean value) {
-      
-      fetchSkipSync_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * By default, RunCallable() will synchronize the GPU stream before returning
-     * fetched tensors on a GPU device, to ensure that the values in those tensors
-     * have been produced. This simplifies interacting with the tensors, but
-     * potentially incurs a performance hit.
-     * If this options is set to true, the caller is responsible for ensuring
-     * that the values in the fetched tensors have been produced before they are
-     * used. The caller can do this by invoking `Device::Sync()` on the underlying
-     * device(s), or by feeding the tensors back to the same Session using
-     * `feed_devices` with the same corresponding device name.
-     * </pre>
-     *
-     * <code>bool fetch_skip_sync = 8;</code>
-     */
-    public Builder clearFetchSkipSync() {
-      
-      fetchSkipSync_ = false;
-      onChanged();
-      return this;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.CallableOptions)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.CallableOptions)
-  private static final org.tensorflow.proto.framework.CallableOptions DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.CallableOptions();
-  }
-
-  public static org.tensorflow.proto.framework.CallableOptions getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<CallableOptions>
-      PARSER = new com.google.protobuf.AbstractParser<CallableOptions>() {
-    @java.lang.Override
-    public CallableOptions parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new CallableOptions(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<CallableOptions> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<CallableOptions> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.framework.CallableOptions getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/CallableOptionsOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/CallableOptionsOrBuilder.java
deleted file mode 100644
index d65852d4843..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/CallableOptionsOrBuilder.java
+++ /dev/null
@@ -1,485 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/config.proto
-
-package org.tensorflow.proto.framework;
-
-public interface CallableOptionsOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.CallableOptions)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <pre>
-   * Tensors to be fed in the callable. Each feed is the name of a tensor.
-   * </pre>
-   *
-   * <code>repeated string feed = 1;</code>
-   */
-  java.util.List<java.lang.String>
-      getFeedList();
-  /**
-   * <pre>
-   * Tensors to be fed in the callable. Each feed is the name of a tensor.
-   * </pre>
-   *
-   * <code>repeated string feed = 1;</code>
-   */
-  int getFeedCount();
-  /**
-   * <pre>
-   * Tensors to be fed in the callable. Each feed is the name of a tensor.
-   * </pre>
-   *
-   * <code>repeated string feed = 1;</code>
-   */
-  java.lang.String getFeed(int index);
-  /**
-   * <pre>
-   * Tensors to be fed in the callable. Each feed is the name of a tensor.
-   * </pre>
-   *
-   * <code>repeated string feed = 1;</code>
-   */
-  com.google.protobuf.ByteString
-      getFeedBytes(int index);
-
-  /**
-   * <pre>
-   * Fetches. A list of tensor names. The caller of the callable expects a
-   * tensor to be returned for each fetch[i] (see RunStepResponse.tensor). The
-   * order of specified fetches does not change the execution order.
-   * </pre>
-   *
-   * <code>repeated string fetch = 2;</code>
-   */
-  java.util.List<java.lang.String>
-      getFetchList();
-  /**
-   * <pre>
-   * Fetches. A list of tensor names. The caller of the callable expects a
-   * tensor to be returned for each fetch[i] (see RunStepResponse.tensor). The
-   * order of specified fetches does not change the execution order.
-   * </pre>
-   *
-   * <code>repeated string fetch = 2;</code>
-   */
-  int getFetchCount();
-  /**
-   * <pre>
-   * Fetches. A list of tensor names. The caller of the callable expects a
-   * tensor to be returned for each fetch[i] (see RunStepResponse.tensor). The
-   * order of specified fetches does not change the execution order.
-   * </pre>
-   *
-   * <code>repeated string fetch = 2;</code>
-   */
-  java.lang.String getFetch(int index);
-  /**
-   * <pre>
-   * Fetches. A list of tensor names. The caller of the callable expects a
-   * tensor to be returned for each fetch[i] (see RunStepResponse.tensor). The
-   * order of specified fetches does not change the execution order.
-   * </pre>
-   *
-   * <code>repeated string fetch = 2;</code>
-   */
-  com.google.protobuf.ByteString
-      getFetchBytes(int index);
-
-  /**
-   * <pre>
-   * Target Nodes. A list of node names. The named nodes will be run by the
-   * callable but their outputs will not be returned.
-   * </pre>
-   *
-   * <code>repeated string target = 3;</code>
-   */
-  java.util.List<java.lang.String>
-      getTargetList();
-  /**
-   * <pre>
-   * Target Nodes. A list of node names. The named nodes will be run by the
-   * callable but their outputs will not be returned.
-   * </pre>
-   *
-   * <code>repeated string target = 3;</code>
-   */
-  int getTargetCount();
-  /**
-   * <pre>
-   * Target Nodes. A list of node names. The named nodes will be run by the
-   * callable but their outputs will not be returned.
-   * </pre>
-   *
-   * <code>repeated string target = 3;</code>
-   */
-  java.lang.String getTarget(int index);
-  /**
-   * <pre>
-   * Target Nodes. A list of node names. The named nodes will be run by the
-   * callable but their outputs will not be returned.
-   * </pre>
-   *
-   * <code>repeated string target = 3;</code>
-   */
-  com.google.protobuf.ByteString
-      getTargetBytes(int index);
-
-  /**
-   * <pre>
-   * Options that will be applied to each run.
-   * </pre>
-   *
-   * <code>.tensorflow.RunOptions run_options = 4;</code>
-   */
-  boolean hasRunOptions();
-  /**
-   * <pre>
-   * Options that will be applied to each run.
-   * </pre>
-   *
-   * <code>.tensorflow.RunOptions run_options = 4;</code>
-   */
-  org.tensorflow.proto.framework.RunOptions getRunOptions();
-  /**
-   * <pre>
-   * Options that will be applied to each run.
-   * </pre>
-   *
-   * <code>.tensorflow.RunOptions run_options = 4;</code>
-   */
-  org.tensorflow.proto.framework.RunOptionsOrBuilder getRunOptionsOrBuilder();
-
-  /**
-   * <pre>
-   * Tensors to be connected in the callable. Each TensorConnection denotes
-   * a pair of tensors in the graph, between which an edge will be created
-   * in the callable.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.TensorConnection tensor_connection = 5;</code>
-   */
-  java.util.List<org.tensorflow.proto.framework.TensorConnection> 
-      getTensorConnectionList();
-  /**
-   * <pre>
-   * Tensors to be connected in the callable. Each TensorConnection denotes
-   * a pair of tensors in the graph, between which an edge will be created
-   * in the callable.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.TensorConnection tensor_connection = 5;</code>
-   */
-  org.tensorflow.proto.framework.TensorConnection getTensorConnection(int index);
-  /**
-   * <pre>
-   * Tensors to be connected in the callable. Each TensorConnection denotes
-   * a pair of tensors in the graph, between which an edge will be created
-   * in the callable.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.TensorConnection tensor_connection = 5;</code>
-   */
-  int getTensorConnectionCount();
-  /**
-   * <pre>
-   * Tensors to be connected in the callable. Each TensorConnection denotes
-   * a pair of tensors in the graph, between which an edge will be created
-   * in the callable.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.TensorConnection tensor_connection = 5;</code>
-   */
-  java.util.List<? extends org.tensorflow.proto.framework.TensorConnectionOrBuilder> 
-      getTensorConnectionOrBuilderList();
-  /**
-   * <pre>
-   * Tensors to be connected in the callable. Each TensorConnection denotes
-   * a pair of tensors in the graph, between which an edge will be created
-   * in the callable.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.TensorConnection tensor_connection = 5;</code>
-   */
-  org.tensorflow.proto.framework.TensorConnectionOrBuilder getTensorConnectionOrBuilder(
-      int index);
-
-  /**
-   * <pre>
-   * The Tensor objects fed in the callable and fetched from the callable
-   * are expected to be backed by host (CPU) memory by default.
-   * The options below allow changing that - feeding tensors backed by
-   * device memory, or returning tensors that are backed by device memory.
-   * The maps below map the name of a feed/fetch tensor (which appears in
-   * 'feed' or 'fetch' fields above), to the fully qualified name of the device
-   * owning the memory backing the contents of the tensor.
-   * For example, creating a callable with the following options:
-   * CallableOptions {
-   *   feed: "a:0"
-   *   feed: "b:0"
-   *   fetch: "x:0"
-   *   fetch: "y:0"
-   *   feed_devices: {
-   *     "a:0": "/job:localhost/replica:0/task:0/device:GPU:0"
-   *   }
-   *   fetch_devices: {
-   *     "y:0": "/job:localhost/replica:0/task:0/device:GPU:0"
-   *  }
-   * }
-   * means that the Callable expects:
-   * - The first argument ("a:0") is a Tensor backed by GPU memory.
-   * - The second argument ("b:0") is a Tensor backed by host memory.
-   * and of its return values:
-   * - The first output ("x:0") will be backed by host memory.
-   * - The second output ("y:0") will be backed by GPU memory.
-   * FEEDS:
-   * It is the responsibility of the caller to ensure that the memory of the fed
-   * tensors will be correctly initialized and synchronized before it is
-   * accessed by operations executed during the call to Session::RunCallable().
-   * This is typically ensured by using the TensorFlow memory allocators
-   * (Device::GetAllocator()) to create the Tensor to be fed.
-   * Alternatively, for CUDA-enabled GPU devices, this typically means that the
-   * operation that produced the contents of the tensor has completed, i.e., the
-   * CUDA stream has been synchronized (e.g., via cuCtxSynchronize() or
-   * cuStreamSynchronize()).
-   * </pre>
-   *
-   * <code>map&lt;string, string&gt; feed_devices = 6;</code>
-   */
-  int getFeedDevicesCount();
-  /**
-   * <pre>
-   * The Tensor objects fed in the callable and fetched from the callable
-   * are expected to be backed by host (CPU) memory by default.
-   * The options below allow changing that - feeding tensors backed by
-   * device memory, or returning tensors that are backed by device memory.
-   * The maps below map the name of a feed/fetch tensor (which appears in
-   * 'feed' or 'fetch' fields above), to the fully qualified name of the device
-   * owning the memory backing the contents of the tensor.
-   * For example, creating a callable with the following options:
-   * CallableOptions {
-   *   feed: "a:0"
-   *   feed: "b:0"
-   *   fetch: "x:0"
-   *   fetch: "y:0"
-   *   feed_devices: {
-   *     "a:0": "/job:localhost/replica:0/task:0/device:GPU:0"
-   *   }
-   *   fetch_devices: {
-   *     "y:0": "/job:localhost/replica:0/task:0/device:GPU:0"
-   *  }
-   * }
-   * means that the Callable expects:
-   * - The first argument ("a:0") is a Tensor backed by GPU memory.
-   * - The second argument ("b:0") is a Tensor backed by host memory.
-   * and of its return values:
-   * - The first output ("x:0") will be backed by host memory.
-   * - The second output ("y:0") will be backed by GPU memory.
-   * FEEDS:
-   * It is the responsibility of the caller to ensure that the memory of the fed
-   * tensors will be correctly initialized and synchronized before it is
-   * accessed by operations executed during the call to Session::RunCallable().
-   * This is typically ensured by using the TensorFlow memory allocators
-   * (Device::GetAllocator()) to create the Tensor to be fed.
-   * Alternatively, for CUDA-enabled GPU devices, this typically means that the
-   * operation that produced the contents of the tensor has completed, i.e., the
-   * CUDA stream has been synchronized (e.g., via cuCtxSynchronize() or
-   * cuStreamSynchronize()).
-   * </pre>
-   *
-   * <code>map&lt;string, string&gt; feed_devices = 6;</code>
-   */
-  boolean containsFeedDevices(
-      java.lang.String key);
-  /**
-   * Use {@link #getFeedDevicesMap()} instead.
-   */
-  @java.lang.Deprecated
-  java.util.Map<java.lang.String, java.lang.String>
-  getFeedDevices();
-  /**
-   * <pre>
-   * The Tensor objects fed in the callable and fetched from the callable
-   * are expected to be backed by host (CPU) memory by default.
-   * The options below allow changing that - feeding tensors backed by
-   * device memory, or returning tensors that are backed by device memory.
-   * The maps below map the name of a feed/fetch tensor (which appears in
-   * 'feed' or 'fetch' fields above), to the fully qualified name of the device
-   * owning the memory backing the contents of the tensor.
-   * For example, creating a callable with the following options:
-   * CallableOptions {
-   *   feed: "a:0"
-   *   feed: "b:0"
-   *   fetch: "x:0"
-   *   fetch: "y:0"
-   *   feed_devices: {
-   *     "a:0": "/job:localhost/replica:0/task:0/device:GPU:0"
-   *   }
-   *   fetch_devices: {
-   *     "y:0": "/job:localhost/replica:0/task:0/device:GPU:0"
-   *  }
-   * }
-   * means that the Callable expects:
-   * - The first argument ("a:0") is a Tensor backed by GPU memory.
-   * - The second argument ("b:0") is a Tensor backed by host memory.
-   * and of its return values:
-   * - The first output ("x:0") will be backed by host memory.
-   * - The second output ("y:0") will be backed by GPU memory.
-   * FEEDS:
-   * It is the responsibility of the caller to ensure that the memory of the fed
-   * tensors will be correctly initialized and synchronized before it is
-   * accessed by operations executed during the call to Session::RunCallable().
-   * This is typically ensured by using the TensorFlow memory allocators
-   * (Device::GetAllocator()) to create the Tensor to be fed.
-   * Alternatively, for CUDA-enabled GPU devices, this typically means that the
-   * operation that produced the contents of the tensor has completed, i.e., the
-   * CUDA stream has been synchronized (e.g., via cuCtxSynchronize() or
-   * cuStreamSynchronize()).
-   * </pre>
-   *
-   * <code>map&lt;string, string&gt; feed_devices = 6;</code>
-   */
-  java.util.Map<java.lang.String, java.lang.String>
-  getFeedDevicesMap();
-  /**
-   * <pre>
-   * The Tensor objects fed in the callable and fetched from the callable
-   * are expected to be backed by host (CPU) memory by default.
-   * The options below allow changing that - feeding tensors backed by
-   * device memory, or returning tensors that are backed by device memory.
-   * The maps below map the name of a feed/fetch tensor (which appears in
-   * 'feed' or 'fetch' fields above), to the fully qualified name of the device
-   * owning the memory backing the contents of the tensor.
-   * For example, creating a callable with the following options:
-   * CallableOptions {
-   *   feed: "a:0"
-   *   feed: "b:0"
-   *   fetch: "x:0"
-   *   fetch: "y:0"
-   *   feed_devices: {
-   *     "a:0": "/job:localhost/replica:0/task:0/device:GPU:0"
-   *   }
-   *   fetch_devices: {
-   *     "y:0": "/job:localhost/replica:0/task:0/device:GPU:0"
-   *  }
-   * }
-   * means that the Callable expects:
-   * - The first argument ("a:0") is a Tensor backed by GPU memory.
-   * - The second argument ("b:0") is a Tensor backed by host memory.
-   * and of its return values:
-   * - The first output ("x:0") will be backed by host memory.
-   * - The second output ("y:0") will be backed by GPU memory.
-   * FEEDS:
-   * It is the responsibility of the caller to ensure that the memory of the fed
-   * tensors will be correctly initialized and synchronized before it is
-   * accessed by operations executed during the call to Session::RunCallable().
-   * This is typically ensured by using the TensorFlow memory allocators
-   * (Device::GetAllocator()) to create the Tensor to be fed.
-   * Alternatively, for CUDA-enabled GPU devices, this typically means that the
-   * operation that produced the contents of the tensor has completed, i.e., the
-   * CUDA stream has been synchronized (e.g., via cuCtxSynchronize() or
-   * cuStreamSynchronize()).
-   * </pre>
-   *
-   * <code>map&lt;string, string&gt; feed_devices = 6;</code>
-   */
-
-  java.lang.String getFeedDevicesOrDefault(
-      java.lang.String key,
-      java.lang.String defaultValue);
-  /**
-   * <pre>
-   * The Tensor objects fed in the callable and fetched from the callable
-   * are expected to be backed by host (CPU) memory by default.
-   * The options below allow changing that - feeding tensors backed by
-   * device memory, or returning tensors that are backed by device memory.
-   * The maps below map the name of a feed/fetch tensor (which appears in
-   * 'feed' or 'fetch' fields above), to the fully qualified name of the device
-   * owning the memory backing the contents of the tensor.
-   * For example, creating a callable with the following options:
-   * CallableOptions {
-   *   feed: "a:0"
-   *   feed: "b:0"
-   *   fetch: "x:0"
-   *   fetch: "y:0"
-   *   feed_devices: {
-   *     "a:0": "/job:localhost/replica:0/task:0/device:GPU:0"
-   *   }
-   *   fetch_devices: {
-   *     "y:0": "/job:localhost/replica:0/task:0/device:GPU:0"
-   *  }
-   * }
-   * means that the Callable expects:
-   * - The first argument ("a:0") is a Tensor backed by GPU memory.
-   * - The second argument ("b:0") is a Tensor backed by host memory.
-   * and of its return values:
-   * - The first output ("x:0") will be backed by host memory.
-   * - The second output ("y:0") will be backed by GPU memory.
-   * FEEDS:
-   * It is the responsibility of the caller to ensure that the memory of the fed
-   * tensors will be correctly initialized and synchronized before it is
-   * accessed by operations executed during the call to Session::RunCallable().
-   * This is typically ensured by using the TensorFlow memory allocators
-   * (Device::GetAllocator()) to create the Tensor to be fed.
-   * Alternatively, for CUDA-enabled GPU devices, this typically means that the
-   * operation that produced the contents of the tensor has completed, i.e., the
-   * CUDA stream has been synchronized (e.g., via cuCtxSynchronize() or
-   * cuStreamSynchronize()).
-   * </pre>
-   *
-   * <code>map&lt;string, string&gt; feed_devices = 6;</code>
-   */
-
-  java.lang.String getFeedDevicesOrThrow(
-      java.lang.String key);
-
-  /**
-   * <code>map&lt;string, string&gt; fetch_devices = 7;</code>
-   */
-  int getFetchDevicesCount();
-  /**
-   * <code>map&lt;string, string&gt; fetch_devices = 7;</code>
-   */
-  boolean containsFetchDevices(
-      java.lang.String key);
-  /**
-   * Use {@link #getFetchDevicesMap()} instead.
-   */
-  @java.lang.Deprecated
-  java.util.Map<java.lang.String, java.lang.String>
-  getFetchDevices();
-  /**
-   * <code>map&lt;string, string&gt; fetch_devices = 7;</code>
-   */
-  java.util.Map<java.lang.String, java.lang.String>
-  getFetchDevicesMap();
-  /**
-   * <code>map&lt;string, string&gt; fetch_devices = 7;</code>
-   */
-
-  java.lang.String getFetchDevicesOrDefault(
-      java.lang.String key,
-      java.lang.String defaultValue);
-  /**
-   * <code>map&lt;string, string&gt; fetch_devices = 7;</code>
-   */
-
-  java.lang.String getFetchDevicesOrThrow(
-      java.lang.String key);
-
-  /**
-   * <pre>
-   * By default, RunCallable() will synchronize the GPU stream before returning
-   * fetched tensors on a GPU device, to ensure that the values in those tensors
-   * have been produced. This simplifies interacting with the tensors, but
-   * potentially incurs a performance hit.
-   * If this options is set to true, the caller is responsible for ensuring
-   * that the values in the fetched tensors have been produced before they are
-   * used. The caller can do this by invoking `Device::Sync()` on the underlying
-   * device(s), or by feeding the tensors back to the same Session using
-   * `feed_devices` with the same corresponding device name.
-   * </pre>
-   *
-   * <code>bool fetch_skip_sync = 8;</code>
-   */
-  boolean getFetchSkipSync();
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/Code.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/Code.java
deleted file mode 100644
index c7eda03f92a..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/Code.java
+++ /dev/null
@@ -1,536 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/error_codes.proto
-
-package org.tensorflow.proto.framework;
-
-/**
- * <pre>
- * The canonical error codes for TensorFlow APIs.
- * Warnings:
- * -   Do not change any numeric assignments.
- * -   Changes to this list should only be made if there is a compelling
- *     need that can't be satisfied in another way.  Such changes
- *     must be approved by at least two OWNERS.
- * -   These error codes must match gRPC and protobuf error codes (except for
- *     DO_NOT_USE_RESERVED_FOR_FUTURE_EXPANSION_USE_DEFAULT_IN_SWITCH_INSTEAD_).
- * Sometimes multiple error codes may apply.  Services should return
- * the most specific error code that applies.  For example, prefer
- * OUT_OF_RANGE over FAILED_PRECONDITION if both codes apply.
- * Similarly prefer NOT_FOUND or ALREADY_EXISTS over FAILED_PRECONDITION.
- * </pre>
- *
- * Protobuf enum {@code tensorflow.error.Code}
- */
-public enum Code
-    implements com.google.protobuf.ProtocolMessageEnum {
-  /**
-   * <pre>
-   * Not an error; returned on success
-   * </pre>
-   *
-   * <code>OK = 0;</code>
-   */
-  OK(0),
-  /**
-   * <pre>
-   * The operation was cancelled (typically by the caller).
-   * </pre>
-   *
-   * <code>CANCELLED = 1;</code>
-   */
-  CANCELLED(1),
-  /**
-   * <pre>
-   * Unknown error.  An example of where this error may be returned is
-   * if a Status value received from another address space belongs to
-   * an error-space that is not known in this address space.  Also
-   * errors raised by APIs that do not return enough error information
-   * may be converted to this error.
-   * </pre>
-   *
-   * <code>UNKNOWN = 2;</code>
-   */
-  UNKNOWN(2),
-  /**
-   * <pre>
-   * Client specified an invalid argument.  Note that this differs
-   * from FAILED_PRECONDITION.  INVALID_ARGUMENT indicates arguments
-   * that are problematic regardless of the state of the system
-   * (e.g., a malformed file name).
-   * </pre>
-   *
-   * <code>INVALID_ARGUMENT = 3;</code>
-   */
-  INVALID_ARGUMENT(3),
-  /**
-   * <pre>
-   * Deadline expired before operation could complete.  For operations
-   * that change the state of the system, this error may be returned
-   * even if the operation has completed successfully.  For example, a
-   * successful response from a server could have been delayed long
-   * enough for the deadline to expire.
-   * </pre>
-   *
-   * <code>DEADLINE_EXCEEDED = 4;</code>
-   */
-  DEADLINE_EXCEEDED(4),
-  /**
-   * <pre>
-   * Some requested entity (e.g., file or directory) was not found.
-   * For privacy reasons, this code *may* be returned when the client
-   * does not have the access right to the entity.
-   * </pre>
-   *
-   * <code>NOT_FOUND = 5;</code>
-   */
-  NOT_FOUND(5),
-  /**
-   * <pre>
-   * Some entity that we attempted to create (e.g., file or directory)
-   * already exists.
-   * </pre>
-   *
-   * <code>ALREADY_EXISTS = 6;</code>
-   */
-  ALREADY_EXISTS(6),
-  /**
-   * <pre>
-   * The caller does not have permission to execute the specified
-   * operation.  PERMISSION_DENIED must not be used for rejections
-   * caused by exhausting some resource (use RESOURCE_EXHAUSTED
-   * instead for those errors).  PERMISSION_DENIED must not be
-   * used if the caller can not be identified (use UNAUTHENTICATED
-   * instead for those errors).
-   * </pre>
-   *
-   * <code>PERMISSION_DENIED = 7;</code>
-   */
-  PERMISSION_DENIED(7),
-  /**
-   * <pre>
-   * The request does not have valid authentication credentials for the
-   * operation.
-   * </pre>
-   *
-   * <code>UNAUTHENTICATED = 16;</code>
-   */
-  UNAUTHENTICATED(16),
-  /**
-   * <pre>
-   * Some resource has been exhausted, perhaps a per-user quota, or
-   * perhaps the entire file system is out of space.
-   * </pre>
-   *
-   * <code>RESOURCE_EXHAUSTED = 8;</code>
-   */
-  RESOURCE_EXHAUSTED(8),
-  /**
-   * <pre>
-   * Operation was rejected because the system is not in a state
-   * required for the operation's execution.  For example, directory
-   * to be deleted may be non-empty, an rmdir operation is applied to
-   * a non-directory, etc.
-   * A litmus test that may help a service implementor in deciding
-   * between FAILED_PRECONDITION, ABORTED, and UNAVAILABLE:
-   *  (a) Use UNAVAILABLE if the client can retry just the failing call.
-   *  (b) Use ABORTED if the client should retry at a higher-level
-   *      (e.g., restarting a read-modify-write sequence).
-   *  (c) Use FAILED_PRECONDITION if the client should not retry until
-   *      the system state has been explicitly fixed.  E.g., if an "rmdir"
-   *      fails because the directory is non-empty, FAILED_PRECONDITION
-   *      should be returned since the client should not retry unless
-   *      they have first fixed up the directory by deleting files from it.
-   *  (d) Use FAILED_PRECONDITION if the client performs conditional
-   *      REST Get/Update/Delete on a resource and the resource on the
-   *      server does not match the condition. E.g., conflicting
-   *      read-modify-write on the same resource.
-   * </pre>
-   *
-   * <code>FAILED_PRECONDITION = 9;</code>
-   */
-  FAILED_PRECONDITION(9),
-  /**
-   * <pre>
-   * The operation was aborted, typically due to a concurrency issue
-   * like sequencer check failures, transaction aborts, etc.
-   * See litmus test above for deciding between FAILED_PRECONDITION,
-   * ABORTED, and UNAVAILABLE.
-   * </pre>
-   *
-   * <code>ABORTED = 10;</code>
-   */
-  ABORTED(10),
-  /**
-   * <pre>
-   * Operation tried to iterate past the valid input range.  E.g., seeking or
-   * reading past end of file.
-   * Unlike INVALID_ARGUMENT, this error indicates a problem that may
-   * be fixed if the system state changes. For example, a 32-bit file
-   * system will generate INVALID_ARGUMENT if asked to read at an
-   * offset that is not in the range [0,2^32-1], but it will generate
-   * OUT_OF_RANGE if asked to read from an offset past the current
-   * file size.
-   * There is a fair bit of overlap between FAILED_PRECONDITION and
-   * OUT_OF_RANGE.  We recommend using OUT_OF_RANGE (the more specific
-   * error) when it applies so that callers who are iterating through
-   * a space can easily look for an OUT_OF_RANGE error to detect when
-   * they are done.
-   * </pre>
-   *
-   * <code>OUT_OF_RANGE = 11;</code>
-   */
-  OUT_OF_RANGE(11),
-  /**
-   * <pre>
-   * Operation is not implemented or not supported/enabled in this service.
-   * </pre>
-   *
-   * <code>UNIMPLEMENTED = 12;</code>
-   */
-  UNIMPLEMENTED(12),
-  /**
-   * <pre>
-   * Internal errors.  Means some invariant expected by the underlying
-   * system has been broken.  If you see one of these errors,
-   * something is very broken.
-   * </pre>
-   *
-   * <code>INTERNAL = 13;</code>
-   */
-  INTERNAL(13),
-  /**
-   * <pre>
-   * The service is currently unavailable.  This is a most likely a
-   * transient condition and may be corrected by retrying with
-   * a backoff.
-   * See litmus test above for deciding between FAILED_PRECONDITION,
-   * ABORTED, and UNAVAILABLE.
-   * </pre>
-   *
-   * <code>UNAVAILABLE = 14;</code>
-   */
-  UNAVAILABLE(14),
-  /**
-   * <pre>
-   * Unrecoverable data loss or corruption.
-   * </pre>
-   *
-   * <code>DATA_LOSS = 15;</code>
-   */
-  DATA_LOSS(15),
-  /**
-   * <pre>
-   * An extra enum entry to prevent people from writing code that
-   * fails to compile when a new code is added.
-   * Nobody should ever reference this enumeration entry. In particular,
-   * if you write C++ code that switches on this enumeration, add a default:
-   * case instead of a case that mentions this enumeration entry.
-   * Nobody should rely on the value (currently 20) listed here.  It
-   * may change in the future.
-   * </pre>
-   *
-   * <code>DO_NOT_USE_RESERVED_FOR_FUTURE_EXPANSION_USE_DEFAULT_IN_SWITCH_INSTEAD_ = 20;</code>
-   */
-  DO_NOT_USE_RESERVED_FOR_FUTURE_EXPANSION_USE_DEFAULT_IN_SWITCH_INSTEAD_(20),
-  UNRECOGNIZED(-1),
-  ;
-
-  /**
-   * <pre>
-   * Not an error; returned on success
-   * </pre>
-   *
-   * <code>OK = 0;</code>
-   */
-  public static final int OK_VALUE = 0;
-  /**
-   * <pre>
-   * The operation was cancelled (typically by the caller).
-   * </pre>
-   *
-   * <code>CANCELLED = 1;</code>
-   */
-  public static final int CANCELLED_VALUE = 1;
-  /**
-   * <pre>
-   * Unknown error.  An example of where this error may be returned is
-   * if a Status value received from another address space belongs to
-   * an error-space that is not known in this address space.  Also
-   * errors raised by APIs that do not return enough error information
-   * may be converted to this error.
-   * </pre>
-   *
-   * <code>UNKNOWN = 2;</code>
-   */
-  public static final int UNKNOWN_VALUE = 2;
-  /**
-   * <pre>
-   * Client specified an invalid argument.  Note that this differs
-   * from FAILED_PRECONDITION.  INVALID_ARGUMENT indicates arguments
-   * that are problematic regardless of the state of the system
-   * (e.g., a malformed file name).
-   * </pre>
-   *
-   * <code>INVALID_ARGUMENT = 3;</code>
-   */
-  public static final int INVALID_ARGUMENT_VALUE = 3;
-  /**
-   * <pre>
-   * Deadline expired before operation could complete.  For operations
-   * that change the state of the system, this error may be returned
-   * even if the operation has completed successfully.  For example, a
-   * successful response from a server could have been delayed long
-   * enough for the deadline to expire.
-   * </pre>
-   *
-   * <code>DEADLINE_EXCEEDED = 4;</code>
-   */
-  public static final int DEADLINE_EXCEEDED_VALUE = 4;
-  /**
-   * <pre>
-   * Some requested entity (e.g., file or directory) was not found.
-   * For privacy reasons, this code *may* be returned when the client
-   * does not have the access right to the entity.
-   * </pre>
-   *
-   * <code>NOT_FOUND = 5;</code>
-   */
-  public static final int NOT_FOUND_VALUE = 5;
-  /**
-   * <pre>
-   * Some entity that we attempted to create (e.g., file or directory)
-   * already exists.
-   * </pre>
-   *
-   * <code>ALREADY_EXISTS = 6;</code>
-   */
-  public static final int ALREADY_EXISTS_VALUE = 6;
-  /**
-   * <pre>
-   * The caller does not have permission to execute the specified
-   * operation.  PERMISSION_DENIED must not be used for rejections
-   * caused by exhausting some resource (use RESOURCE_EXHAUSTED
-   * instead for those errors).  PERMISSION_DENIED must not be
-   * used if the caller can not be identified (use UNAUTHENTICATED
-   * instead for those errors).
-   * </pre>
-   *
-   * <code>PERMISSION_DENIED = 7;</code>
-   */
-  public static final int PERMISSION_DENIED_VALUE = 7;
-  /**
-   * <pre>
-   * The request does not have valid authentication credentials for the
-   * operation.
-   * </pre>
-   *
-   * <code>UNAUTHENTICATED = 16;</code>
-   */
-  public static final int UNAUTHENTICATED_VALUE = 16;
-  /**
-   * <pre>
-   * Some resource has been exhausted, perhaps a per-user quota, or
-   * perhaps the entire file system is out of space.
-   * </pre>
-   *
-   * <code>RESOURCE_EXHAUSTED = 8;</code>
-   */
-  public static final int RESOURCE_EXHAUSTED_VALUE = 8;
-  /**
-   * <pre>
-   * Operation was rejected because the system is not in a state
-   * required for the operation's execution.  For example, directory
-   * to be deleted may be non-empty, an rmdir operation is applied to
-   * a non-directory, etc.
-   * A litmus test that may help a service implementor in deciding
-   * between FAILED_PRECONDITION, ABORTED, and UNAVAILABLE:
-   *  (a) Use UNAVAILABLE if the client can retry just the failing call.
-   *  (b) Use ABORTED if the client should retry at a higher-level
-   *      (e.g., restarting a read-modify-write sequence).
-   *  (c) Use FAILED_PRECONDITION if the client should not retry until
-   *      the system state has been explicitly fixed.  E.g., if an "rmdir"
-   *      fails because the directory is non-empty, FAILED_PRECONDITION
-   *      should be returned since the client should not retry unless
-   *      they have first fixed up the directory by deleting files from it.
-   *  (d) Use FAILED_PRECONDITION if the client performs conditional
-   *      REST Get/Update/Delete on a resource and the resource on the
-   *      server does not match the condition. E.g., conflicting
-   *      read-modify-write on the same resource.
-   * </pre>
-   *
-   * <code>FAILED_PRECONDITION = 9;</code>
-   */
-  public static final int FAILED_PRECONDITION_VALUE = 9;
-  /**
-   * <pre>
-   * The operation was aborted, typically due to a concurrency issue
-   * like sequencer check failures, transaction aborts, etc.
-   * See litmus test above for deciding between FAILED_PRECONDITION,
-   * ABORTED, and UNAVAILABLE.
-   * </pre>
-   *
-   * <code>ABORTED = 10;</code>
-   */
-  public static final int ABORTED_VALUE = 10;
-  /**
-   * <pre>
-   * Operation tried to iterate past the valid input range.  E.g., seeking or
-   * reading past end of file.
-   * Unlike INVALID_ARGUMENT, this error indicates a problem that may
-   * be fixed if the system state changes. For example, a 32-bit file
-   * system will generate INVALID_ARGUMENT if asked to read at an
-   * offset that is not in the range [0,2^32-1], but it will generate
-   * OUT_OF_RANGE if asked to read from an offset past the current
-   * file size.
-   * There is a fair bit of overlap between FAILED_PRECONDITION and
-   * OUT_OF_RANGE.  We recommend using OUT_OF_RANGE (the more specific
-   * error) when it applies so that callers who are iterating through
-   * a space can easily look for an OUT_OF_RANGE error to detect when
-   * they are done.
-   * </pre>
-   *
-   * <code>OUT_OF_RANGE = 11;</code>
-   */
-  public static final int OUT_OF_RANGE_VALUE = 11;
-  /**
-   * <pre>
-   * Operation is not implemented or not supported/enabled in this service.
-   * </pre>
-   *
-   * <code>UNIMPLEMENTED = 12;</code>
-   */
-  public static final int UNIMPLEMENTED_VALUE = 12;
-  /**
-   * <pre>
-   * Internal errors.  Means some invariant expected by the underlying
-   * system has been broken.  If you see one of these errors,
-   * something is very broken.
-   * </pre>
-   *
-   * <code>INTERNAL = 13;</code>
-   */
-  public static final int INTERNAL_VALUE = 13;
-  /**
-   * <pre>
-   * The service is currently unavailable.  This is a most likely a
-   * transient condition and may be corrected by retrying with
-   * a backoff.
-   * See litmus test above for deciding between FAILED_PRECONDITION,
-   * ABORTED, and UNAVAILABLE.
-   * </pre>
-   *
-   * <code>UNAVAILABLE = 14;</code>
-   */
-  public static final int UNAVAILABLE_VALUE = 14;
-  /**
-   * <pre>
-   * Unrecoverable data loss or corruption.
-   * </pre>
-   *
-   * <code>DATA_LOSS = 15;</code>
-   */
-  public static final int DATA_LOSS_VALUE = 15;
-  /**
-   * <pre>
-   * An extra enum entry to prevent people from writing code that
-   * fails to compile when a new code is added.
-   * Nobody should ever reference this enumeration entry. In particular,
-   * if you write C++ code that switches on this enumeration, add a default:
-   * case instead of a case that mentions this enumeration entry.
-   * Nobody should rely on the value (currently 20) listed here.  It
-   * may change in the future.
-   * </pre>
-   *
-   * <code>DO_NOT_USE_RESERVED_FOR_FUTURE_EXPANSION_USE_DEFAULT_IN_SWITCH_INSTEAD_ = 20;</code>
-   */
-  public static final int DO_NOT_USE_RESERVED_FOR_FUTURE_EXPANSION_USE_DEFAULT_IN_SWITCH_INSTEAD__VALUE = 20;
-
-
-  public final int getNumber() {
-    if (this == UNRECOGNIZED) {
-      throw new java.lang.IllegalArgumentException(
-          "Can't get the number of an unknown enum value.");
-    }
-    return value;
-  }
-
-  /**
-   * @deprecated Use {@link #forNumber(int)} instead.
-   */
-  @java.lang.Deprecated
-  public static Code valueOf(int value) {
-    return forNumber(value);
-  }
-
-  public static Code forNumber(int value) {
-    switch (value) {
-      case 0: return OK;
-      case 1: return CANCELLED;
-      case 2: return UNKNOWN;
-      case 3: return INVALID_ARGUMENT;
-      case 4: return DEADLINE_EXCEEDED;
-      case 5: return NOT_FOUND;
-      case 6: return ALREADY_EXISTS;
-      case 7: return PERMISSION_DENIED;
-      case 16: return UNAUTHENTICATED;
-      case 8: return RESOURCE_EXHAUSTED;
-      case 9: return FAILED_PRECONDITION;
-      case 10: return ABORTED;
-      case 11: return OUT_OF_RANGE;
-      case 12: return UNIMPLEMENTED;
-      case 13: return INTERNAL;
-      case 14: return UNAVAILABLE;
-      case 15: return DATA_LOSS;
-      case 20: return DO_NOT_USE_RESERVED_FOR_FUTURE_EXPANSION_USE_DEFAULT_IN_SWITCH_INSTEAD_;
-      default: return null;
-    }
-  }
-
-  public static com.google.protobuf.Internal.EnumLiteMap<Code>
-      internalGetValueMap() {
-    return internalValueMap;
-  }
-  private static final com.google.protobuf.Internal.EnumLiteMap<
-      Code> internalValueMap =
-        new com.google.protobuf.Internal.EnumLiteMap<Code>() {
-          public Code findValueByNumber(int number) {
-            return Code.forNumber(number);
-          }
-        };
-
-  public final com.google.protobuf.Descriptors.EnumValueDescriptor
-      getValueDescriptor() {
-    return getDescriptor().getValues().get(ordinal());
-  }
-  public final com.google.protobuf.Descriptors.EnumDescriptor
-      getDescriptorForType() {
-    return getDescriptor();
-  }
-  public static final com.google.protobuf.Descriptors.EnumDescriptor
-      getDescriptor() {
-    return org.tensorflow.proto.framework.ErrorCodesProtos.getDescriptor().getEnumTypes().get(0);
-  }
-
-  private static final Code[] VALUES = values();
-
-  public static Code valueOf(
-      com.google.protobuf.Descriptors.EnumValueDescriptor desc) {
-    if (desc.getType() != getDescriptor()) {
-      throw new java.lang.IllegalArgumentException(
-        "EnumValueDescriptor is not for this type.");
-    }
-    if (desc.getIndex() == -1) {
-      return UNRECOGNIZED;
-    }
-    return VALUES[desc.getIndex()];
-  }
-
-  private final int value;
-
-  private Code(int value) {
-    this.value = value;
-  }
-
-  // @@protoc_insertion_point(enum_scope:tensorflow.error.Code)
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/CollectionDef.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/CollectionDef.java
deleted file mode 100644
index f3094be10f6..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/CollectionDef.java
+++ /dev/null
@@ -1,4858 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/meta_graph.proto
-
-package org.tensorflow.proto.framework;
-
-/**
- * <pre>
- * CollectionDef should cover most collections.
- * To add a user-defined collection, do one of the following:
- * 1. For simple data types, such as string, int, float:
- *      tf.add_to_collection("your_collection_name", your_simple_value)
- *    strings will be stored as bytes_list.
- * 2. For Protobuf types, there are three ways to add them:
- *    1) tf.add_to_collection("your_collection_name",
- *         your_proto.SerializeToString())
- *       collection_def {
- *         key: "user_defined_bytes_collection"
- *         value {
- *           bytes_list {
- *             value: "queue_name: &#92;"test_queue&#92;"&#92;n"
- *           }
- *         }
- *       }
- *  or
- *    2) tf.add_to_collection("your_collection_name", str(your_proto))
- *       collection_def {
- *         key: "user_defined_string_collection"
- *         value {
- *          bytes_list {
- *             value: "&#92;n&#92;ntest_queue"
- *           }
- *         }
- *       }
- *  or
- *    3) any_buf = any_pb2.Any()
- *       tf.add_to_collection("your_collection_name",
- *         any_buf.Pack(your_proto))
- *       collection_def {
- *         key: "user_defined_any_collection"
- *         value {
- *           any_list {
- *             value {
- *               type_url: "type.googleapis.com/tensorflow.QueueRunnerDef"
- *               value: "&#92;n&#92;ntest_queue"
- *             }
- *           }
- *         }
- *       }
- * 3. For Python objects, implement to_proto() and from_proto(), and register
- *    them in the following manner:
- *    ops.register_proto_function("your_collection_name",
- *                                proto_type,
- *                                to_proto=YourPythonObject.to_proto,
- *                                from_proto=YourPythonObject.from_proto)
- *    These functions will be invoked to serialize and de-serialize the
- *    collection. For example,
- *    ops.register_proto_function(ops.GraphKeys.GLOBAL_VARIABLES,
- *                                proto_type=variable_pb2.VariableDef,
- *                                to_proto=Variable.to_proto,
- *                                from_proto=Variable.from_proto)
- * </pre>
- *
- * Protobuf type {@code tensorflow.CollectionDef}
- */
-public  final class CollectionDef extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.CollectionDef)
-    CollectionDefOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use CollectionDef.newBuilder() to construct.
-  private CollectionDef(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private CollectionDef() {
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new CollectionDef();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private CollectionDef(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 10: {
-            org.tensorflow.proto.framework.CollectionDef.NodeList.Builder subBuilder = null;
-            if (kindCase_ == 1) {
-              subBuilder = ((org.tensorflow.proto.framework.CollectionDef.NodeList) kind_).toBuilder();
-            }
-            kind_ =
-                input.readMessage(org.tensorflow.proto.framework.CollectionDef.NodeList.parser(), extensionRegistry);
-            if (subBuilder != null) {
-              subBuilder.mergeFrom((org.tensorflow.proto.framework.CollectionDef.NodeList) kind_);
-              kind_ = subBuilder.buildPartial();
-            }
-            kindCase_ = 1;
-            break;
-          }
-          case 18: {
-            org.tensorflow.proto.framework.CollectionDef.BytesList.Builder subBuilder = null;
-            if (kindCase_ == 2) {
-              subBuilder = ((org.tensorflow.proto.framework.CollectionDef.BytesList) kind_).toBuilder();
-            }
-            kind_ =
-                input.readMessage(org.tensorflow.proto.framework.CollectionDef.BytesList.parser(), extensionRegistry);
-            if (subBuilder != null) {
-              subBuilder.mergeFrom((org.tensorflow.proto.framework.CollectionDef.BytesList) kind_);
-              kind_ = subBuilder.buildPartial();
-            }
-            kindCase_ = 2;
-            break;
-          }
-          case 26: {
-            org.tensorflow.proto.framework.CollectionDef.Int64List.Builder subBuilder = null;
-            if (kindCase_ == 3) {
-              subBuilder = ((org.tensorflow.proto.framework.CollectionDef.Int64List) kind_).toBuilder();
-            }
-            kind_ =
-                input.readMessage(org.tensorflow.proto.framework.CollectionDef.Int64List.parser(), extensionRegistry);
-            if (subBuilder != null) {
-              subBuilder.mergeFrom((org.tensorflow.proto.framework.CollectionDef.Int64List) kind_);
-              kind_ = subBuilder.buildPartial();
-            }
-            kindCase_ = 3;
-            break;
-          }
-          case 34: {
-            org.tensorflow.proto.framework.CollectionDef.FloatList.Builder subBuilder = null;
-            if (kindCase_ == 4) {
-              subBuilder = ((org.tensorflow.proto.framework.CollectionDef.FloatList) kind_).toBuilder();
-            }
-            kind_ =
-                input.readMessage(org.tensorflow.proto.framework.CollectionDef.FloatList.parser(), extensionRegistry);
-            if (subBuilder != null) {
-              subBuilder.mergeFrom((org.tensorflow.proto.framework.CollectionDef.FloatList) kind_);
-              kind_ = subBuilder.buildPartial();
-            }
-            kindCase_ = 4;
-            break;
-          }
-          case 42: {
-            org.tensorflow.proto.framework.CollectionDef.AnyList.Builder subBuilder = null;
-            if (kindCase_ == 5) {
-              subBuilder = ((org.tensorflow.proto.framework.CollectionDef.AnyList) kind_).toBuilder();
-            }
-            kind_ =
-                input.readMessage(org.tensorflow.proto.framework.CollectionDef.AnyList.parser(), extensionRegistry);
-            if (subBuilder != null) {
-              subBuilder.mergeFrom((org.tensorflow.proto.framework.CollectionDef.AnyList) kind_);
-              kind_ = subBuilder.buildPartial();
-            }
-            kindCase_ = 5;
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.framework.MetaGraphProtos.internal_static_tensorflow_CollectionDef_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.framework.MetaGraphProtos.internal_static_tensorflow_CollectionDef_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.framework.CollectionDef.class, org.tensorflow.proto.framework.CollectionDef.Builder.class);
-  }
-
-  public interface NodeListOrBuilder extends
-      // @@protoc_insertion_point(interface_extends:tensorflow.CollectionDef.NodeList)
-      com.google.protobuf.MessageOrBuilder {
-
-    /**
-     * <code>repeated string value = 1;</code>
-     */
-    java.util.List<java.lang.String>
-        getValueList();
-    /**
-     * <code>repeated string value = 1;</code>
-     */
-    int getValueCount();
-    /**
-     * <code>repeated string value = 1;</code>
-     */
-    java.lang.String getValue(int index);
-    /**
-     * <code>repeated string value = 1;</code>
-     */
-    com.google.protobuf.ByteString
-        getValueBytes(int index);
-  }
-  /**
-   * <pre>
-   * NodeList is used for collecting nodes in graph. For example
-   * collection_def {
-   *   key: "summaries"
-   *   value {
-   *     node_list {
-   *       value: "input_producer/ScalarSummary:0"
-   *       value: "shuffle_batch/ScalarSummary:0"
-   *       value: "ImageSummary:0"
-   *     }
-   *   }
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.CollectionDef.NodeList}
-   */
-  public  static final class NodeList extends
-      com.google.protobuf.GeneratedMessageV3 implements
-      // @@protoc_insertion_point(message_implements:tensorflow.CollectionDef.NodeList)
-      NodeListOrBuilder {
-  private static final long serialVersionUID = 0L;
-    // Use NodeList.newBuilder() to construct.
-    private NodeList(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-      super(builder);
-    }
-    private NodeList() {
-      value_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-    }
-
-    @java.lang.Override
-    @SuppressWarnings({"unused"})
-    protected java.lang.Object newInstance(
-        UnusedPrivateParameter unused) {
-      return new NodeList();
-    }
-
-    @java.lang.Override
-    public final com.google.protobuf.UnknownFieldSet
-    getUnknownFields() {
-      return this.unknownFields;
-    }
-    private NodeList(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      this();
-      if (extensionRegistry == null) {
-        throw new java.lang.NullPointerException();
-      }
-      int mutable_bitField0_ = 0;
-      com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-          com.google.protobuf.UnknownFieldSet.newBuilder();
-      try {
-        boolean done = false;
-        while (!done) {
-          int tag = input.readTag();
-          switch (tag) {
-            case 0:
-              done = true;
-              break;
-            case 10: {
-              java.lang.String s = input.readStringRequireUtf8();
-              if (!((mutable_bitField0_ & 0x00000001) != 0)) {
-                value_ = new com.google.protobuf.LazyStringArrayList();
-                mutable_bitField0_ |= 0x00000001;
-              }
-              value_.add(s);
-              break;
-            }
-            default: {
-              if (!parseUnknownField(
-                  input, unknownFields, extensionRegistry, tag)) {
-                done = true;
-              }
-              break;
-            }
-          }
-        }
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        throw e.setUnfinishedMessage(this);
-      } catch (java.io.IOException e) {
-        throw new com.google.protobuf.InvalidProtocolBufferException(
-            e).setUnfinishedMessage(this);
-      } finally {
-        if (((mutable_bitField0_ & 0x00000001) != 0)) {
-          value_ = value_.getUnmodifiableView();
-        }
-        this.unknownFields = unknownFields.build();
-        makeExtensionsImmutable();
-      }
-    }
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.MetaGraphProtos.internal_static_tensorflow_CollectionDef_NodeList_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.MetaGraphProtos.internal_static_tensorflow_CollectionDef_NodeList_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.CollectionDef.NodeList.class, org.tensorflow.proto.framework.CollectionDef.NodeList.Builder.class);
-    }
-
-    public static final int VALUE_FIELD_NUMBER = 1;
-    private com.google.protobuf.LazyStringList value_;
-    /**
-     * <code>repeated string value = 1;</code>
-     */
-    public com.google.protobuf.ProtocolStringList
-        getValueList() {
-      return value_;
-    }
-    /**
-     * <code>repeated string value = 1;</code>
-     */
-    public int getValueCount() {
-      return value_.size();
-    }
-    /**
-     * <code>repeated string value = 1;</code>
-     */
-    public java.lang.String getValue(int index) {
-      return value_.get(index);
-    }
-    /**
-     * <code>repeated string value = 1;</code>
-     */
-    public com.google.protobuf.ByteString
-        getValueBytes(int index) {
-      return value_.getByteString(index);
-    }
-
-    private byte memoizedIsInitialized = -1;
-    @java.lang.Override
-    public final boolean isInitialized() {
-      byte isInitialized = memoizedIsInitialized;
-      if (isInitialized == 1) return true;
-      if (isInitialized == 0) return false;
-
-      memoizedIsInitialized = 1;
-      return true;
-    }
-
-    @java.lang.Override
-    public void writeTo(com.google.protobuf.CodedOutputStream output)
-                        throws java.io.IOException {
-      for (int i = 0; i < value_.size(); i++) {
-        com.google.protobuf.GeneratedMessageV3.writeString(output, 1, value_.getRaw(i));
-      }
-      unknownFields.writeTo(output);
-    }
-
-    @java.lang.Override
-    public int getSerializedSize() {
-      int size = memoizedSize;
-      if (size != -1) return size;
-
-      size = 0;
-      {
-        int dataSize = 0;
-        for (int i = 0; i < value_.size(); i++) {
-          dataSize += computeStringSizeNoTag(value_.getRaw(i));
-        }
-        size += dataSize;
-        size += 1 * getValueList().size();
-      }
-      size += unknownFields.getSerializedSize();
-      memoizedSize = size;
-      return size;
-    }
-
-    @java.lang.Override
-    public boolean equals(final java.lang.Object obj) {
-      if (obj == this) {
-       return true;
-      }
-      if (!(obj instanceof org.tensorflow.proto.framework.CollectionDef.NodeList)) {
-        return super.equals(obj);
-      }
-      org.tensorflow.proto.framework.CollectionDef.NodeList other = (org.tensorflow.proto.framework.CollectionDef.NodeList) obj;
-
-      if (!getValueList()
-          .equals(other.getValueList())) return false;
-      if (!unknownFields.equals(other.unknownFields)) return false;
-      return true;
-    }
-
-    @java.lang.Override
-    public int hashCode() {
-      if (memoizedHashCode != 0) {
-        return memoizedHashCode;
-      }
-      int hash = 41;
-      hash = (19 * hash) + getDescriptor().hashCode();
-      if (getValueCount() > 0) {
-        hash = (37 * hash) + VALUE_FIELD_NUMBER;
-        hash = (53 * hash) + getValueList().hashCode();
-      }
-      hash = (29 * hash) + unknownFields.hashCode();
-      memoizedHashCode = hash;
-      return hash;
-    }
-
-    public static org.tensorflow.proto.framework.CollectionDef.NodeList parseFrom(
-        java.nio.ByteBuffer data)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data);
-    }
-    public static org.tensorflow.proto.framework.CollectionDef.NodeList parseFrom(
-        java.nio.ByteBuffer data,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.CollectionDef.NodeList parseFrom(
-        com.google.protobuf.ByteString data)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data);
-    }
-    public static org.tensorflow.proto.framework.CollectionDef.NodeList parseFrom(
-        com.google.protobuf.ByteString data,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.CollectionDef.NodeList parseFrom(byte[] data)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data);
-    }
-    public static org.tensorflow.proto.framework.CollectionDef.NodeList parseFrom(
-        byte[] data,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.CollectionDef.NodeList parseFrom(java.io.InputStream input)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input);
-    }
-    public static org.tensorflow.proto.framework.CollectionDef.NodeList parseFrom(
-        java.io.InputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.CollectionDef.NodeList parseDelimitedFrom(java.io.InputStream input)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseDelimitedWithIOException(PARSER, input);
-    }
-    public static org.tensorflow.proto.framework.CollectionDef.NodeList parseDelimitedFrom(
-        java.io.InputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.CollectionDef.NodeList parseFrom(
-        com.google.protobuf.CodedInputStream input)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input);
-    }
-    public static org.tensorflow.proto.framework.CollectionDef.NodeList parseFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input, extensionRegistry);
-    }
-
-    @java.lang.Override
-    public Builder newBuilderForType() { return newBuilder(); }
-    public static Builder newBuilder() {
-      return DEFAULT_INSTANCE.toBuilder();
-    }
-    public static Builder newBuilder(org.tensorflow.proto.framework.CollectionDef.NodeList prototype) {
-      return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-    }
-    @java.lang.Override
-    public Builder toBuilder() {
-      return this == DEFAULT_INSTANCE
-          ? new Builder() : new Builder().mergeFrom(this);
-    }
-
-    @java.lang.Override
-    protected Builder newBuilderForType(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      Builder builder = new Builder(parent);
-      return builder;
-    }
-    /**
-     * <pre>
-     * NodeList is used for collecting nodes in graph. For example
-     * collection_def {
-     *   key: "summaries"
-     *   value {
-     *     node_list {
-     *       value: "input_producer/ScalarSummary:0"
-     *       value: "shuffle_batch/ScalarSummary:0"
-     *       value: "ImageSummary:0"
-     *     }
-     *   }
-     * </pre>
-     *
-     * Protobuf type {@code tensorflow.CollectionDef.NodeList}
-     */
-    public static final class Builder extends
-        com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-        // @@protoc_insertion_point(builder_implements:tensorflow.CollectionDef.NodeList)
-        org.tensorflow.proto.framework.CollectionDef.NodeListOrBuilder {
-      public static final com.google.protobuf.Descriptors.Descriptor
-          getDescriptor() {
-        return org.tensorflow.proto.framework.MetaGraphProtos.internal_static_tensorflow_CollectionDef_NodeList_descriptor;
-      }
-
-      @java.lang.Override
-      protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-          internalGetFieldAccessorTable() {
-        return org.tensorflow.proto.framework.MetaGraphProtos.internal_static_tensorflow_CollectionDef_NodeList_fieldAccessorTable
-            .ensureFieldAccessorsInitialized(
-                org.tensorflow.proto.framework.CollectionDef.NodeList.class, org.tensorflow.proto.framework.CollectionDef.NodeList.Builder.class);
-      }
-
-      // Construct using org.tensorflow.proto.framework.CollectionDef.NodeList.newBuilder()
-      private Builder() {
-        maybeForceBuilderInitialization();
-      }
-
-      private Builder(
-          com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-        super(parent);
-        maybeForceBuilderInitialization();
-      }
-      private void maybeForceBuilderInitialization() {
-        if (com.google.protobuf.GeneratedMessageV3
-                .alwaysUseFieldBuilders) {
-        }
-      }
-      @java.lang.Override
-      public Builder clear() {
-        super.clear();
-        value_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-        bitField0_ = (bitField0_ & ~0x00000001);
-        return this;
-      }
-
-      @java.lang.Override
-      public com.google.protobuf.Descriptors.Descriptor
-          getDescriptorForType() {
-        return org.tensorflow.proto.framework.MetaGraphProtos.internal_static_tensorflow_CollectionDef_NodeList_descriptor;
-      }
-
-      @java.lang.Override
-      public org.tensorflow.proto.framework.CollectionDef.NodeList getDefaultInstanceForType() {
-        return org.tensorflow.proto.framework.CollectionDef.NodeList.getDefaultInstance();
-      }
-
-      @java.lang.Override
-      public org.tensorflow.proto.framework.CollectionDef.NodeList build() {
-        org.tensorflow.proto.framework.CollectionDef.NodeList result = buildPartial();
-        if (!result.isInitialized()) {
-          throw newUninitializedMessageException(result);
-        }
-        return result;
-      }
-
-      @java.lang.Override
-      public org.tensorflow.proto.framework.CollectionDef.NodeList buildPartial() {
-        org.tensorflow.proto.framework.CollectionDef.NodeList result = new org.tensorflow.proto.framework.CollectionDef.NodeList(this);
-        int from_bitField0_ = bitField0_;
-        if (((bitField0_ & 0x00000001) != 0)) {
-          value_ = value_.getUnmodifiableView();
-          bitField0_ = (bitField0_ & ~0x00000001);
-        }
-        result.value_ = value_;
-        onBuilt();
-        return result;
-      }
-
-      @java.lang.Override
-      public Builder clone() {
-        return super.clone();
-      }
-      @java.lang.Override
-      public Builder setField(
-          com.google.protobuf.Descriptors.FieldDescriptor field,
-          java.lang.Object value) {
-        return super.setField(field, value);
-      }
-      @java.lang.Override
-      public Builder clearField(
-          com.google.protobuf.Descriptors.FieldDescriptor field) {
-        return super.clearField(field);
-      }
-      @java.lang.Override
-      public Builder clearOneof(
-          com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-        return super.clearOneof(oneof);
-      }
-      @java.lang.Override
-      public Builder setRepeatedField(
-          com.google.protobuf.Descriptors.FieldDescriptor field,
-          int index, java.lang.Object value) {
-        return super.setRepeatedField(field, index, value);
-      }
-      @java.lang.Override
-      public Builder addRepeatedField(
-          com.google.protobuf.Descriptors.FieldDescriptor field,
-          java.lang.Object value) {
-        return super.addRepeatedField(field, value);
-      }
-      @java.lang.Override
-      public Builder mergeFrom(com.google.protobuf.Message other) {
-        if (other instanceof org.tensorflow.proto.framework.CollectionDef.NodeList) {
-          return mergeFrom((org.tensorflow.proto.framework.CollectionDef.NodeList)other);
-        } else {
-          super.mergeFrom(other);
-          return this;
-        }
-      }
-
-      public Builder mergeFrom(org.tensorflow.proto.framework.CollectionDef.NodeList other) {
-        if (other == org.tensorflow.proto.framework.CollectionDef.NodeList.getDefaultInstance()) return this;
-        if (!other.value_.isEmpty()) {
-          if (value_.isEmpty()) {
-            value_ = other.value_;
-            bitField0_ = (bitField0_ & ~0x00000001);
-          } else {
-            ensureValueIsMutable();
-            value_.addAll(other.value_);
-          }
-          onChanged();
-        }
-        this.mergeUnknownFields(other.unknownFields);
-        onChanged();
-        return this;
-      }
-
-      @java.lang.Override
-      public final boolean isInitialized() {
-        return true;
-      }
-
-      @java.lang.Override
-      public Builder mergeFrom(
-          com.google.protobuf.CodedInputStream input,
-          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-          throws java.io.IOException {
-        org.tensorflow.proto.framework.CollectionDef.NodeList parsedMessage = null;
-        try {
-          parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-        } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-          parsedMessage = (org.tensorflow.proto.framework.CollectionDef.NodeList) e.getUnfinishedMessage();
-          throw e.unwrapIOException();
-        } finally {
-          if (parsedMessage != null) {
-            mergeFrom(parsedMessage);
-          }
-        }
-        return this;
-      }
-      private int bitField0_;
-
-      private com.google.protobuf.LazyStringList value_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-      private void ensureValueIsMutable() {
-        if (!((bitField0_ & 0x00000001) != 0)) {
-          value_ = new com.google.protobuf.LazyStringArrayList(value_);
-          bitField0_ |= 0x00000001;
-         }
-      }
-      /**
-       * <code>repeated string value = 1;</code>
-       */
-      public com.google.protobuf.ProtocolStringList
-          getValueList() {
-        return value_.getUnmodifiableView();
-      }
-      /**
-       * <code>repeated string value = 1;</code>
-       */
-      public int getValueCount() {
-        return value_.size();
-      }
-      /**
-       * <code>repeated string value = 1;</code>
-       */
-      public java.lang.String getValue(int index) {
-        return value_.get(index);
-      }
-      /**
-       * <code>repeated string value = 1;</code>
-       */
-      public com.google.protobuf.ByteString
-          getValueBytes(int index) {
-        return value_.getByteString(index);
-      }
-      /**
-       * <code>repeated string value = 1;</code>
-       */
-      public Builder setValue(
-          int index, java.lang.String value) {
-        if (value == null) {
-    throw new NullPointerException();
-  }
-  ensureValueIsMutable();
-        value_.set(index, value);
-        onChanged();
-        return this;
-      }
-      /**
-       * <code>repeated string value = 1;</code>
-       */
-      public Builder addValue(
-          java.lang.String value) {
-        if (value == null) {
-    throw new NullPointerException();
-  }
-  ensureValueIsMutable();
-        value_.add(value);
-        onChanged();
-        return this;
-      }
-      /**
-       * <code>repeated string value = 1;</code>
-       */
-      public Builder addAllValue(
-          java.lang.Iterable<java.lang.String> values) {
-        ensureValueIsMutable();
-        com.google.protobuf.AbstractMessageLite.Builder.addAll(
-            values, value_);
-        onChanged();
-        return this;
-      }
-      /**
-       * <code>repeated string value = 1;</code>
-       */
-      public Builder clearValue() {
-        value_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-        bitField0_ = (bitField0_ & ~0x00000001);
-        onChanged();
-        return this;
-      }
-      /**
-       * <code>repeated string value = 1;</code>
-       */
-      public Builder addValueBytes(
-          com.google.protobuf.ByteString value) {
-        if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-        ensureValueIsMutable();
-        value_.add(value);
-        onChanged();
-        return this;
-      }
-      @java.lang.Override
-      public final Builder setUnknownFields(
-          final com.google.protobuf.UnknownFieldSet unknownFields) {
-        return super.setUnknownFields(unknownFields);
-      }
-
-      @java.lang.Override
-      public final Builder mergeUnknownFields(
-          final com.google.protobuf.UnknownFieldSet unknownFields) {
-        return super.mergeUnknownFields(unknownFields);
-      }
-
-
-      // @@protoc_insertion_point(builder_scope:tensorflow.CollectionDef.NodeList)
-    }
-
-    // @@protoc_insertion_point(class_scope:tensorflow.CollectionDef.NodeList)
-    private static final org.tensorflow.proto.framework.CollectionDef.NodeList DEFAULT_INSTANCE;
-    static {
-      DEFAULT_INSTANCE = new org.tensorflow.proto.framework.CollectionDef.NodeList();
-    }
-
-    public static org.tensorflow.proto.framework.CollectionDef.NodeList getDefaultInstance() {
-      return DEFAULT_INSTANCE;
-    }
-
-    private static final com.google.protobuf.Parser<NodeList>
-        PARSER = new com.google.protobuf.AbstractParser<NodeList>() {
-      @java.lang.Override
-      public NodeList parsePartialFrom(
-          com.google.protobuf.CodedInputStream input,
-          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-          throws com.google.protobuf.InvalidProtocolBufferException {
-        return new NodeList(input, extensionRegistry);
-      }
-    };
-
-    public static com.google.protobuf.Parser<NodeList> parser() {
-      return PARSER;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Parser<NodeList> getParserForType() {
-      return PARSER;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.CollectionDef.NodeList getDefaultInstanceForType() {
-      return DEFAULT_INSTANCE;
-    }
-
-  }
-
-  public interface BytesListOrBuilder extends
-      // @@protoc_insertion_point(interface_extends:tensorflow.CollectionDef.BytesList)
-      com.google.protobuf.MessageOrBuilder {
-
-    /**
-     * <code>repeated bytes value = 1;</code>
-     */
-    java.util.List<com.google.protobuf.ByteString> getValueList();
-    /**
-     * <code>repeated bytes value = 1;</code>
-     */
-    int getValueCount();
-    /**
-     * <code>repeated bytes value = 1;</code>
-     */
-    com.google.protobuf.ByteString getValue(int index);
-  }
-  /**
-   * <pre>
-   * BytesList is used for collecting strings and serialized protobufs. For
-   * example:
-   * collection_def {
-   *   key: "trainable_variables"
-   *   value {
-   *     bytes_list {
-   *       value: "&#92;n&#92;017conv1/weights:0&#92;022&#92;024conv1/weights/Assign
-   *              &#92;032&#92;024conv1/weights/read:0"
-   *       value: "&#92;n&#92;016conv1/biases:0&#92;022&#92;023conv1/biases/Assign&#92;032
-   *              &#92;023conv1/biases/read:0"
-   *     }
-   *   }
-   * }
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.CollectionDef.BytesList}
-   */
-  public  static final class BytesList extends
-      com.google.protobuf.GeneratedMessageV3 implements
-      // @@protoc_insertion_point(message_implements:tensorflow.CollectionDef.BytesList)
-      BytesListOrBuilder {
-  private static final long serialVersionUID = 0L;
-    // Use BytesList.newBuilder() to construct.
-    private BytesList(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-      super(builder);
-    }
-    private BytesList() {
-      value_ = java.util.Collections.emptyList();
-    }
-
-    @java.lang.Override
-    @SuppressWarnings({"unused"})
-    protected java.lang.Object newInstance(
-        UnusedPrivateParameter unused) {
-      return new BytesList();
-    }
-
-    @java.lang.Override
-    public final com.google.protobuf.UnknownFieldSet
-    getUnknownFields() {
-      return this.unknownFields;
-    }
-    private BytesList(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      this();
-      if (extensionRegistry == null) {
-        throw new java.lang.NullPointerException();
-      }
-      int mutable_bitField0_ = 0;
-      com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-          com.google.protobuf.UnknownFieldSet.newBuilder();
-      try {
-        boolean done = false;
-        while (!done) {
-          int tag = input.readTag();
-          switch (tag) {
-            case 0:
-              done = true;
-              break;
-            case 10: {
-              if (!((mutable_bitField0_ & 0x00000001) != 0)) {
-                value_ = new java.util.ArrayList<com.google.protobuf.ByteString>();
-                mutable_bitField0_ |= 0x00000001;
-              }
-              value_.add(input.readBytes());
-              break;
-            }
-            default: {
-              if (!parseUnknownField(
-                  input, unknownFields, extensionRegistry, tag)) {
-                done = true;
-              }
-              break;
-            }
-          }
-        }
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        throw e.setUnfinishedMessage(this);
-      } catch (java.io.IOException e) {
-        throw new com.google.protobuf.InvalidProtocolBufferException(
-            e).setUnfinishedMessage(this);
-      } finally {
-        if (((mutable_bitField0_ & 0x00000001) != 0)) {
-          value_ = java.util.Collections.unmodifiableList(value_); // C
-        }
-        this.unknownFields = unknownFields.build();
-        makeExtensionsImmutable();
-      }
-    }
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.MetaGraphProtos.internal_static_tensorflow_CollectionDef_BytesList_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.MetaGraphProtos.internal_static_tensorflow_CollectionDef_BytesList_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.CollectionDef.BytesList.class, org.tensorflow.proto.framework.CollectionDef.BytesList.Builder.class);
-    }
-
-    public static final int VALUE_FIELD_NUMBER = 1;
-    private java.util.List<com.google.protobuf.ByteString> value_;
-    /**
-     * <code>repeated bytes value = 1;</code>
-     */
-    public java.util.List<com.google.protobuf.ByteString>
-        getValueList() {
-      return value_;
-    }
-    /**
-     * <code>repeated bytes value = 1;</code>
-     */
-    public int getValueCount() {
-      return value_.size();
-    }
-    /**
-     * <code>repeated bytes value = 1;</code>
-     */
-    public com.google.protobuf.ByteString getValue(int index) {
-      return value_.get(index);
-    }
-
-    private byte memoizedIsInitialized = -1;
-    @java.lang.Override
-    public final boolean isInitialized() {
-      byte isInitialized = memoizedIsInitialized;
-      if (isInitialized == 1) return true;
-      if (isInitialized == 0) return false;
-
-      memoizedIsInitialized = 1;
-      return true;
-    }
-
-    @java.lang.Override
-    public void writeTo(com.google.protobuf.CodedOutputStream output)
-                        throws java.io.IOException {
-      for (int i = 0; i < value_.size(); i++) {
-        output.writeBytes(1, value_.get(i));
-      }
-      unknownFields.writeTo(output);
-    }
-
-    @java.lang.Override
-    public int getSerializedSize() {
-      int size = memoizedSize;
-      if (size != -1) return size;
-
-      size = 0;
-      {
-        int dataSize = 0;
-        for (int i = 0; i < value_.size(); i++) {
-          dataSize += com.google.protobuf.CodedOutputStream
-            .computeBytesSizeNoTag(value_.get(i));
-        }
-        size += dataSize;
-        size += 1 * getValueList().size();
-      }
-      size += unknownFields.getSerializedSize();
-      memoizedSize = size;
-      return size;
-    }
-
-    @java.lang.Override
-    public boolean equals(final java.lang.Object obj) {
-      if (obj == this) {
-       return true;
-      }
-      if (!(obj instanceof org.tensorflow.proto.framework.CollectionDef.BytesList)) {
-        return super.equals(obj);
-      }
-      org.tensorflow.proto.framework.CollectionDef.BytesList other = (org.tensorflow.proto.framework.CollectionDef.BytesList) obj;
-
-      if (!getValueList()
-          .equals(other.getValueList())) return false;
-      if (!unknownFields.equals(other.unknownFields)) return false;
-      return true;
-    }
-
-    @java.lang.Override
-    public int hashCode() {
-      if (memoizedHashCode != 0) {
-        return memoizedHashCode;
-      }
-      int hash = 41;
-      hash = (19 * hash) + getDescriptor().hashCode();
-      if (getValueCount() > 0) {
-        hash = (37 * hash) + VALUE_FIELD_NUMBER;
-        hash = (53 * hash) + getValueList().hashCode();
-      }
-      hash = (29 * hash) + unknownFields.hashCode();
-      memoizedHashCode = hash;
-      return hash;
-    }
-
-    public static org.tensorflow.proto.framework.CollectionDef.BytesList parseFrom(
-        java.nio.ByteBuffer data)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data);
-    }
-    public static org.tensorflow.proto.framework.CollectionDef.BytesList parseFrom(
-        java.nio.ByteBuffer data,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.CollectionDef.BytesList parseFrom(
-        com.google.protobuf.ByteString data)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data);
-    }
-    public static org.tensorflow.proto.framework.CollectionDef.BytesList parseFrom(
-        com.google.protobuf.ByteString data,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.CollectionDef.BytesList parseFrom(byte[] data)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data);
-    }
-    public static org.tensorflow.proto.framework.CollectionDef.BytesList parseFrom(
-        byte[] data,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.CollectionDef.BytesList parseFrom(java.io.InputStream input)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input);
-    }
-    public static org.tensorflow.proto.framework.CollectionDef.BytesList parseFrom(
-        java.io.InputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.CollectionDef.BytesList parseDelimitedFrom(java.io.InputStream input)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseDelimitedWithIOException(PARSER, input);
-    }
-    public static org.tensorflow.proto.framework.CollectionDef.BytesList parseDelimitedFrom(
-        java.io.InputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.CollectionDef.BytesList parseFrom(
-        com.google.protobuf.CodedInputStream input)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input);
-    }
-    public static org.tensorflow.proto.framework.CollectionDef.BytesList parseFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input, extensionRegistry);
-    }
-
-    @java.lang.Override
-    public Builder newBuilderForType() { return newBuilder(); }
-    public static Builder newBuilder() {
-      return DEFAULT_INSTANCE.toBuilder();
-    }
-    public static Builder newBuilder(org.tensorflow.proto.framework.CollectionDef.BytesList prototype) {
-      return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-    }
-    @java.lang.Override
-    public Builder toBuilder() {
-      return this == DEFAULT_INSTANCE
-          ? new Builder() : new Builder().mergeFrom(this);
-    }
-
-    @java.lang.Override
-    protected Builder newBuilderForType(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      Builder builder = new Builder(parent);
-      return builder;
-    }
-    /**
-     * <pre>
-     * BytesList is used for collecting strings and serialized protobufs. For
-     * example:
-     * collection_def {
-     *   key: "trainable_variables"
-     *   value {
-     *     bytes_list {
-     *       value: "&#92;n&#92;017conv1/weights:0&#92;022&#92;024conv1/weights/Assign
-     *              &#92;032&#92;024conv1/weights/read:0"
-     *       value: "&#92;n&#92;016conv1/biases:0&#92;022&#92;023conv1/biases/Assign&#92;032
-     *              &#92;023conv1/biases/read:0"
-     *     }
-     *   }
-     * }
-     * </pre>
-     *
-     * Protobuf type {@code tensorflow.CollectionDef.BytesList}
-     */
-    public static final class Builder extends
-        com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-        // @@protoc_insertion_point(builder_implements:tensorflow.CollectionDef.BytesList)
-        org.tensorflow.proto.framework.CollectionDef.BytesListOrBuilder {
-      public static final com.google.protobuf.Descriptors.Descriptor
-          getDescriptor() {
-        return org.tensorflow.proto.framework.MetaGraphProtos.internal_static_tensorflow_CollectionDef_BytesList_descriptor;
-      }
-
-      @java.lang.Override
-      protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-          internalGetFieldAccessorTable() {
-        return org.tensorflow.proto.framework.MetaGraphProtos.internal_static_tensorflow_CollectionDef_BytesList_fieldAccessorTable
-            .ensureFieldAccessorsInitialized(
-                org.tensorflow.proto.framework.CollectionDef.BytesList.class, org.tensorflow.proto.framework.CollectionDef.BytesList.Builder.class);
-      }
-
-      // Construct using org.tensorflow.proto.framework.CollectionDef.BytesList.newBuilder()
-      private Builder() {
-        maybeForceBuilderInitialization();
-      }
-
-      private Builder(
-          com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-        super(parent);
-        maybeForceBuilderInitialization();
-      }
-      private void maybeForceBuilderInitialization() {
-        if (com.google.protobuf.GeneratedMessageV3
-                .alwaysUseFieldBuilders) {
-        }
-      }
-      @java.lang.Override
-      public Builder clear() {
-        super.clear();
-        value_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000001);
-        return this;
-      }
-
-      @java.lang.Override
-      public com.google.protobuf.Descriptors.Descriptor
-          getDescriptorForType() {
-        return org.tensorflow.proto.framework.MetaGraphProtos.internal_static_tensorflow_CollectionDef_BytesList_descriptor;
-      }
-
-      @java.lang.Override
-      public org.tensorflow.proto.framework.CollectionDef.BytesList getDefaultInstanceForType() {
-        return org.tensorflow.proto.framework.CollectionDef.BytesList.getDefaultInstance();
-      }
-
-      @java.lang.Override
-      public org.tensorflow.proto.framework.CollectionDef.BytesList build() {
-        org.tensorflow.proto.framework.CollectionDef.BytesList result = buildPartial();
-        if (!result.isInitialized()) {
-          throw newUninitializedMessageException(result);
-        }
-        return result;
-      }
-
-      @java.lang.Override
-      public org.tensorflow.proto.framework.CollectionDef.BytesList buildPartial() {
-        org.tensorflow.proto.framework.CollectionDef.BytesList result = new org.tensorflow.proto.framework.CollectionDef.BytesList(this);
-        int from_bitField0_ = bitField0_;
-        if (((bitField0_ & 0x00000001) != 0)) {
-          value_ = java.util.Collections.unmodifiableList(value_);
-          bitField0_ = (bitField0_ & ~0x00000001);
-        }
-        result.value_ = value_;
-        onBuilt();
-        return result;
-      }
-
-      @java.lang.Override
-      public Builder clone() {
-        return super.clone();
-      }
-      @java.lang.Override
-      public Builder setField(
-          com.google.protobuf.Descriptors.FieldDescriptor field,
-          java.lang.Object value) {
-        return super.setField(field, value);
-      }
-      @java.lang.Override
-      public Builder clearField(
-          com.google.protobuf.Descriptors.FieldDescriptor field) {
-        return super.clearField(field);
-      }
-      @java.lang.Override
-      public Builder clearOneof(
-          com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-        return super.clearOneof(oneof);
-      }
-      @java.lang.Override
-      public Builder setRepeatedField(
-          com.google.protobuf.Descriptors.FieldDescriptor field,
-          int index, java.lang.Object value) {
-        return super.setRepeatedField(field, index, value);
-      }
-      @java.lang.Override
-      public Builder addRepeatedField(
-          com.google.protobuf.Descriptors.FieldDescriptor field,
-          java.lang.Object value) {
-        return super.addRepeatedField(field, value);
-      }
-      @java.lang.Override
-      public Builder mergeFrom(com.google.protobuf.Message other) {
-        if (other instanceof org.tensorflow.proto.framework.CollectionDef.BytesList) {
-          return mergeFrom((org.tensorflow.proto.framework.CollectionDef.BytesList)other);
-        } else {
-          super.mergeFrom(other);
-          return this;
-        }
-      }
-
-      public Builder mergeFrom(org.tensorflow.proto.framework.CollectionDef.BytesList other) {
-        if (other == org.tensorflow.proto.framework.CollectionDef.BytesList.getDefaultInstance()) return this;
-        if (!other.value_.isEmpty()) {
-          if (value_.isEmpty()) {
-            value_ = other.value_;
-            bitField0_ = (bitField0_ & ~0x00000001);
-          } else {
-            ensureValueIsMutable();
-            value_.addAll(other.value_);
-          }
-          onChanged();
-        }
-        this.mergeUnknownFields(other.unknownFields);
-        onChanged();
-        return this;
-      }
-
-      @java.lang.Override
-      public final boolean isInitialized() {
-        return true;
-      }
-
-      @java.lang.Override
-      public Builder mergeFrom(
-          com.google.protobuf.CodedInputStream input,
-          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-          throws java.io.IOException {
-        org.tensorflow.proto.framework.CollectionDef.BytesList parsedMessage = null;
-        try {
-          parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-        } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-          parsedMessage = (org.tensorflow.proto.framework.CollectionDef.BytesList) e.getUnfinishedMessage();
-          throw e.unwrapIOException();
-        } finally {
-          if (parsedMessage != null) {
-            mergeFrom(parsedMessage);
-          }
-        }
-        return this;
-      }
-      private int bitField0_;
-
-      private java.util.List<com.google.protobuf.ByteString> value_ = java.util.Collections.emptyList();
-      private void ensureValueIsMutable() {
-        if (!((bitField0_ & 0x00000001) != 0)) {
-          value_ = new java.util.ArrayList<com.google.protobuf.ByteString>(value_);
-          bitField0_ |= 0x00000001;
-         }
-      }
-      /**
-       * <code>repeated bytes value = 1;</code>
-       */
-      public java.util.List<com.google.protobuf.ByteString>
-          getValueList() {
-        return ((bitField0_ & 0x00000001) != 0) ?
-                 java.util.Collections.unmodifiableList(value_) : value_;
-      }
-      /**
-       * <code>repeated bytes value = 1;</code>
-       */
-      public int getValueCount() {
-        return value_.size();
-      }
-      /**
-       * <code>repeated bytes value = 1;</code>
-       */
-      public com.google.protobuf.ByteString getValue(int index) {
-        return value_.get(index);
-      }
-      /**
-       * <code>repeated bytes value = 1;</code>
-       */
-      public Builder setValue(
-          int index, com.google.protobuf.ByteString value) {
-        if (value == null) {
-    throw new NullPointerException();
-  }
-  ensureValueIsMutable();
-        value_.set(index, value);
-        onChanged();
-        return this;
-      }
-      /**
-       * <code>repeated bytes value = 1;</code>
-       */
-      public Builder addValue(com.google.protobuf.ByteString value) {
-        if (value == null) {
-    throw new NullPointerException();
-  }
-  ensureValueIsMutable();
-        value_.add(value);
-        onChanged();
-        return this;
-      }
-      /**
-       * <code>repeated bytes value = 1;</code>
-       */
-      public Builder addAllValue(
-          java.lang.Iterable<? extends com.google.protobuf.ByteString> values) {
-        ensureValueIsMutable();
-        com.google.protobuf.AbstractMessageLite.Builder.addAll(
-            values, value_);
-        onChanged();
-        return this;
-      }
-      /**
-       * <code>repeated bytes value = 1;</code>
-       */
-      public Builder clearValue() {
-        value_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000001);
-        onChanged();
-        return this;
-      }
-      @java.lang.Override
-      public final Builder setUnknownFields(
-          final com.google.protobuf.UnknownFieldSet unknownFields) {
-        return super.setUnknownFields(unknownFields);
-      }
-
-      @java.lang.Override
-      public final Builder mergeUnknownFields(
-          final com.google.protobuf.UnknownFieldSet unknownFields) {
-        return super.mergeUnknownFields(unknownFields);
-      }
-
-
-      // @@protoc_insertion_point(builder_scope:tensorflow.CollectionDef.BytesList)
-    }
-
-    // @@protoc_insertion_point(class_scope:tensorflow.CollectionDef.BytesList)
-    private static final org.tensorflow.proto.framework.CollectionDef.BytesList DEFAULT_INSTANCE;
-    static {
-      DEFAULT_INSTANCE = new org.tensorflow.proto.framework.CollectionDef.BytesList();
-    }
-
-    public static org.tensorflow.proto.framework.CollectionDef.BytesList getDefaultInstance() {
-      return DEFAULT_INSTANCE;
-    }
-
-    private static final com.google.protobuf.Parser<BytesList>
-        PARSER = new com.google.protobuf.AbstractParser<BytesList>() {
-      @java.lang.Override
-      public BytesList parsePartialFrom(
-          com.google.protobuf.CodedInputStream input,
-          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-          throws com.google.protobuf.InvalidProtocolBufferException {
-        return new BytesList(input, extensionRegistry);
-      }
-    };
-
-    public static com.google.protobuf.Parser<BytesList> parser() {
-      return PARSER;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Parser<BytesList> getParserForType() {
-      return PARSER;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.CollectionDef.BytesList getDefaultInstanceForType() {
-      return DEFAULT_INSTANCE;
-    }
-
-  }
-
-  public interface Int64ListOrBuilder extends
-      // @@protoc_insertion_point(interface_extends:tensorflow.CollectionDef.Int64List)
-      com.google.protobuf.MessageOrBuilder {
-
-    /**
-     * <code>repeated int64 value = 1 [packed = true];</code>
-     */
-    java.util.List<java.lang.Long> getValueList();
-    /**
-     * <code>repeated int64 value = 1 [packed = true];</code>
-     */
-    int getValueCount();
-    /**
-     * <code>repeated int64 value = 1 [packed = true];</code>
-     */
-    long getValue(int index);
-  }
-  /**
-   * <pre>
-   * Int64List is used for collecting int, int64 and long values.
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.CollectionDef.Int64List}
-   */
-  public  static final class Int64List extends
-      com.google.protobuf.GeneratedMessageV3 implements
-      // @@protoc_insertion_point(message_implements:tensorflow.CollectionDef.Int64List)
-      Int64ListOrBuilder {
-  private static final long serialVersionUID = 0L;
-    // Use Int64List.newBuilder() to construct.
-    private Int64List(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-      super(builder);
-    }
-    private Int64List() {
-      value_ = emptyLongList();
-    }
-
-    @java.lang.Override
-    @SuppressWarnings({"unused"})
-    protected java.lang.Object newInstance(
-        UnusedPrivateParameter unused) {
-      return new Int64List();
-    }
-
-    @java.lang.Override
-    public final com.google.protobuf.UnknownFieldSet
-    getUnknownFields() {
-      return this.unknownFields;
-    }
-    private Int64List(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      this();
-      if (extensionRegistry == null) {
-        throw new java.lang.NullPointerException();
-      }
-      int mutable_bitField0_ = 0;
-      com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-          com.google.protobuf.UnknownFieldSet.newBuilder();
-      try {
-        boolean done = false;
-        while (!done) {
-          int tag = input.readTag();
-          switch (tag) {
-            case 0:
-              done = true;
-              break;
-            case 8: {
-              if (!((mutable_bitField0_ & 0x00000001) != 0)) {
-                value_ = newLongList();
-                mutable_bitField0_ |= 0x00000001;
-              }
-              value_.addLong(input.readInt64());
-              break;
-            }
-            case 10: {
-              int length = input.readRawVarint32();
-              int limit = input.pushLimit(length);
-              if (!((mutable_bitField0_ & 0x00000001) != 0) && input.getBytesUntilLimit() > 0) {
-                value_ = newLongList();
-                mutable_bitField0_ |= 0x00000001;
-              }
-              while (input.getBytesUntilLimit() > 0) {
-                value_.addLong(input.readInt64());
-              }
-              input.popLimit(limit);
-              break;
-            }
-            default: {
-              if (!parseUnknownField(
-                  input, unknownFields, extensionRegistry, tag)) {
-                done = true;
-              }
-              break;
-            }
-          }
-        }
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        throw e.setUnfinishedMessage(this);
-      } catch (java.io.IOException e) {
-        throw new com.google.protobuf.InvalidProtocolBufferException(
-            e).setUnfinishedMessage(this);
-      } finally {
-        if (((mutable_bitField0_ & 0x00000001) != 0)) {
-          value_.makeImmutable(); // C
-        }
-        this.unknownFields = unknownFields.build();
-        makeExtensionsImmutable();
-      }
-    }
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.MetaGraphProtos.internal_static_tensorflow_CollectionDef_Int64List_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.MetaGraphProtos.internal_static_tensorflow_CollectionDef_Int64List_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.CollectionDef.Int64List.class, org.tensorflow.proto.framework.CollectionDef.Int64List.Builder.class);
-    }
-
-    public static final int VALUE_FIELD_NUMBER = 1;
-    private com.google.protobuf.Internal.LongList value_;
-    /**
-     * <code>repeated int64 value = 1 [packed = true];</code>
-     */
-    public java.util.List<java.lang.Long>
-        getValueList() {
-      return value_;
-    }
-    /**
-     * <code>repeated int64 value = 1 [packed = true];</code>
-     */
-    public int getValueCount() {
-      return value_.size();
-    }
-    /**
-     * <code>repeated int64 value = 1 [packed = true];</code>
-     */
-    public long getValue(int index) {
-      return value_.getLong(index);
-    }
-    private int valueMemoizedSerializedSize = -1;
-
-    private byte memoizedIsInitialized = -1;
-    @java.lang.Override
-    public final boolean isInitialized() {
-      byte isInitialized = memoizedIsInitialized;
-      if (isInitialized == 1) return true;
-      if (isInitialized == 0) return false;
-
-      memoizedIsInitialized = 1;
-      return true;
-    }
-
-    @java.lang.Override
-    public void writeTo(com.google.protobuf.CodedOutputStream output)
-                        throws java.io.IOException {
-      getSerializedSize();
-      if (getValueList().size() > 0) {
-        output.writeUInt32NoTag(10);
-        output.writeUInt32NoTag(valueMemoizedSerializedSize);
-      }
-      for (int i = 0; i < value_.size(); i++) {
-        output.writeInt64NoTag(value_.getLong(i));
-      }
-      unknownFields.writeTo(output);
-    }
-
-    @java.lang.Override
-    public int getSerializedSize() {
-      int size = memoizedSize;
-      if (size != -1) return size;
-
-      size = 0;
-      {
-        int dataSize = 0;
-        for (int i = 0; i < value_.size(); i++) {
-          dataSize += com.google.protobuf.CodedOutputStream
-            .computeInt64SizeNoTag(value_.getLong(i));
-        }
-        size += dataSize;
-        if (!getValueList().isEmpty()) {
-          size += 1;
-          size += com.google.protobuf.CodedOutputStream
-              .computeInt32SizeNoTag(dataSize);
-        }
-        valueMemoizedSerializedSize = dataSize;
-      }
-      size += unknownFields.getSerializedSize();
-      memoizedSize = size;
-      return size;
-    }
-
-    @java.lang.Override
-    public boolean equals(final java.lang.Object obj) {
-      if (obj == this) {
-       return true;
-      }
-      if (!(obj instanceof org.tensorflow.proto.framework.CollectionDef.Int64List)) {
-        return super.equals(obj);
-      }
-      org.tensorflow.proto.framework.CollectionDef.Int64List other = (org.tensorflow.proto.framework.CollectionDef.Int64List) obj;
-
-      if (!getValueList()
-          .equals(other.getValueList())) return false;
-      if (!unknownFields.equals(other.unknownFields)) return false;
-      return true;
-    }
-
-    @java.lang.Override
-    public int hashCode() {
-      if (memoizedHashCode != 0) {
-        return memoizedHashCode;
-      }
-      int hash = 41;
-      hash = (19 * hash) + getDescriptor().hashCode();
-      if (getValueCount() > 0) {
-        hash = (37 * hash) + VALUE_FIELD_NUMBER;
-        hash = (53 * hash) + getValueList().hashCode();
-      }
-      hash = (29 * hash) + unknownFields.hashCode();
-      memoizedHashCode = hash;
-      return hash;
-    }
-
-    public static org.tensorflow.proto.framework.CollectionDef.Int64List parseFrom(
-        java.nio.ByteBuffer data)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data);
-    }
-    public static org.tensorflow.proto.framework.CollectionDef.Int64List parseFrom(
-        java.nio.ByteBuffer data,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.CollectionDef.Int64List parseFrom(
-        com.google.protobuf.ByteString data)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data);
-    }
-    public static org.tensorflow.proto.framework.CollectionDef.Int64List parseFrom(
-        com.google.protobuf.ByteString data,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.CollectionDef.Int64List parseFrom(byte[] data)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data);
-    }
-    public static org.tensorflow.proto.framework.CollectionDef.Int64List parseFrom(
-        byte[] data,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.CollectionDef.Int64List parseFrom(java.io.InputStream input)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input);
-    }
-    public static org.tensorflow.proto.framework.CollectionDef.Int64List parseFrom(
-        java.io.InputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.CollectionDef.Int64List parseDelimitedFrom(java.io.InputStream input)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseDelimitedWithIOException(PARSER, input);
-    }
-    public static org.tensorflow.proto.framework.CollectionDef.Int64List parseDelimitedFrom(
-        java.io.InputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.CollectionDef.Int64List parseFrom(
-        com.google.protobuf.CodedInputStream input)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input);
-    }
-    public static org.tensorflow.proto.framework.CollectionDef.Int64List parseFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input, extensionRegistry);
-    }
-
-    @java.lang.Override
-    public Builder newBuilderForType() { return newBuilder(); }
-    public static Builder newBuilder() {
-      return DEFAULT_INSTANCE.toBuilder();
-    }
-    public static Builder newBuilder(org.tensorflow.proto.framework.CollectionDef.Int64List prototype) {
-      return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-    }
-    @java.lang.Override
-    public Builder toBuilder() {
-      return this == DEFAULT_INSTANCE
-          ? new Builder() : new Builder().mergeFrom(this);
-    }
-
-    @java.lang.Override
-    protected Builder newBuilderForType(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      Builder builder = new Builder(parent);
-      return builder;
-    }
-    /**
-     * <pre>
-     * Int64List is used for collecting int, int64 and long values.
-     * </pre>
-     *
-     * Protobuf type {@code tensorflow.CollectionDef.Int64List}
-     */
-    public static final class Builder extends
-        com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-        // @@protoc_insertion_point(builder_implements:tensorflow.CollectionDef.Int64List)
-        org.tensorflow.proto.framework.CollectionDef.Int64ListOrBuilder {
-      public static final com.google.protobuf.Descriptors.Descriptor
-          getDescriptor() {
-        return org.tensorflow.proto.framework.MetaGraphProtos.internal_static_tensorflow_CollectionDef_Int64List_descriptor;
-      }
-
-      @java.lang.Override
-      protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-          internalGetFieldAccessorTable() {
-        return org.tensorflow.proto.framework.MetaGraphProtos.internal_static_tensorflow_CollectionDef_Int64List_fieldAccessorTable
-            .ensureFieldAccessorsInitialized(
-                org.tensorflow.proto.framework.CollectionDef.Int64List.class, org.tensorflow.proto.framework.CollectionDef.Int64List.Builder.class);
-      }
-
-      // Construct using org.tensorflow.proto.framework.CollectionDef.Int64List.newBuilder()
-      private Builder() {
-        maybeForceBuilderInitialization();
-      }
-
-      private Builder(
-          com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-        super(parent);
-        maybeForceBuilderInitialization();
-      }
-      private void maybeForceBuilderInitialization() {
-        if (com.google.protobuf.GeneratedMessageV3
-                .alwaysUseFieldBuilders) {
-        }
-      }
-      @java.lang.Override
-      public Builder clear() {
-        super.clear();
-        value_ = emptyLongList();
-        bitField0_ = (bitField0_ & ~0x00000001);
-        return this;
-      }
-
-      @java.lang.Override
-      public com.google.protobuf.Descriptors.Descriptor
-          getDescriptorForType() {
-        return org.tensorflow.proto.framework.MetaGraphProtos.internal_static_tensorflow_CollectionDef_Int64List_descriptor;
-      }
-
-      @java.lang.Override
-      public org.tensorflow.proto.framework.CollectionDef.Int64List getDefaultInstanceForType() {
-        return org.tensorflow.proto.framework.CollectionDef.Int64List.getDefaultInstance();
-      }
-
-      @java.lang.Override
-      public org.tensorflow.proto.framework.CollectionDef.Int64List build() {
-        org.tensorflow.proto.framework.CollectionDef.Int64List result = buildPartial();
-        if (!result.isInitialized()) {
-          throw newUninitializedMessageException(result);
-        }
-        return result;
-      }
-
-      @java.lang.Override
-      public org.tensorflow.proto.framework.CollectionDef.Int64List buildPartial() {
-        org.tensorflow.proto.framework.CollectionDef.Int64List result = new org.tensorflow.proto.framework.CollectionDef.Int64List(this);
-        int from_bitField0_ = bitField0_;
-        if (((bitField0_ & 0x00000001) != 0)) {
-          value_.makeImmutable();
-          bitField0_ = (bitField0_ & ~0x00000001);
-        }
-        result.value_ = value_;
-        onBuilt();
-        return result;
-      }
-
-      @java.lang.Override
-      public Builder clone() {
-        return super.clone();
-      }
-      @java.lang.Override
-      public Builder setField(
-          com.google.protobuf.Descriptors.FieldDescriptor field,
-          java.lang.Object value) {
-        return super.setField(field, value);
-      }
-      @java.lang.Override
-      public Builder clearField(
-          com.google.protobuf.Descriptors.FieldDescriptor field) {
-        return super.clearField(field);
-      }
-      @java.lang.Override
-      public Builder clearOneof(
-          com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-        return super.clearOneof(oneof);
-      }
-      @java.lang.Override
-      public Builder setRepeatedField(
-          com.google.protobuf.Descriptors.FieldDescriptor field,
-          int index, java.lang.Object value) {
-        return super.setRepeatedField(field, index, value);
-      }
-      @java.lang.Override
-      public Builder addRepeatedField(
-          com.google.protobuf.Descriptors.FieldDescriptor field,
-          java.lang.Object value) {
-        return super.addRepeatedField(field, value);
-      }
-      @java.lang.Override
-      public Builder mergeFrom(com.google.protobuf.Message other) {
-        if (other instanceof org.tensorflow.proto.framework.CollectionDef.Int64List) {
-          return mergeFrom((org.tensorflow.proto.framework.CollectionDef.Int64List)other);
-        } else {
-          super.mergeFrom(other);
-          return this;
-        }
-      }
-
-      public Builder mergeFrom(org.tensorflow.proto.framework.CollectionDef.Int64List other) {
-        if (other == org.tensorflow.proto.framework.CollectionDef.Int64List.getDefaultInstance()) return this;
-        if (!other.value_.isEmpty()) {
-          if (value_.isEmpty()) {
-            value_ = other.value_;
-            bitField0_ = (bitField0_ & ~0x00000001);
-          } else {
-            ensureValueIsMutable();
-            value_.addAll(other.value_);
-          }
-          onChanged();
-        }
-        this.mergeUnknownFields(other.unknownFields);
-        onChanged();
-        return this;
-      }
-
-      @java.lang.Override
-      public final boolean isInitialized() {
-        return true;
-      }
-
-      @java.lang.Override
-      public Builder mergeFrom(
-          com.google.protobuf.CodedInputStream input,
-          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-          throws java.io.IOException {
-        org.tensorflow.proto.framework.CollectionDef.Int64List parsedMessage = null;
-        try {
-          parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-        } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-          parsedMessage = (org.tensorflow.proto.framework.CollectionDef.Int64List) e.getUnfinishedMessage();
-          throw e.unwrapIOException();
-        } finally {
-          if (parsedMessage != null) {
-            mergeFrom(parsedMessage);
-          }
-        }
-        return this;
-      }
-      private int bitField0_;
-
-      private com.google.protobuf.Internal.LongList value_ = emptyLongList();
-      private void ensureValueIsMutable() {
-        if (!((bitField0_ & 0x00000001) != 0)) {
-          value_ = mutableCopy(value_);
-          bitField0_ |= 0x00000001;
-         }
-      }
-      /**
-       * <code>repeated int64 value = 1 [packed = true];</code>
-       */
-      public java.util.List<java.lang.Long>
-          getValueList() {
-        return ((bitField0_ & 0x00000001) != 0) ?
-                 java.util.Collections.unmodifiableList(value_) : value_;
-      }
-      /**
-       * <code>repeated int64 value = 1 [packed = true];</code>
-       */
-      public int getValueCount() {
-        return value_.size();
-      }
-      /**
-       * <code>repeated int64 value = 1 [packed = true];</code>
-       */
-      public long getValue(int index) {
-        return value_.getLong(index);
-      }
-      /**
-       * <code>repeated int64 value = 1 [packed = true];</code>
-       */
-      public Builder setValue(
-          int index, long value) {
-        ensureValueIsMutable();
-        value_.setLong(index, value);
-        onChanged();
-        return this;
-      }
-      /**
-       * <code>repeated int64 value = 1 [packed = true];</code>
-       */
-      public Builder addValue(long value) {
-        ensureValueIsMutable();
-        value_.addLong(value);
-        onChanged();
-        return this;
-      }
-      /**
-       * <code>repeated int64 value = 1 [packed = true];</code>
-       */
-      public Builder addAllValue(
-          java.lang.Iterable<? extends java.lang.Long> values) {
-        ensureValueIsMutable();
-        com.google.protobuf.AbstractMessageLite.Builder.addAll(
-            values, value_);
-        onChanged();
-        return this;
-      }
-      /**
-       * <code>repeated int64 value = 1 [packed = true];</code>
-       */
-      public Builder clearValue() {
-        value_ = emptyLongList();
-        bitField0_ = (bitField0_ & ~0x00000001);
-        onChanged();
-        return this;
-      }
-      @java.lang.Override
-      public final Builder setUnknownFields(
-          final com.google.protobuf.UnknownFieldSet unknownFields) {
-        return super.setUnknownFields(unknownFields);
-      }
-
-      @java.lang.Override
-      public final Builder mergeUnknownFields(
-          final com.google.protobuf.UnknownFieldSet unknownFields) {
-        return super.mergeUnknownFields(unknownFields);
-      }
-
-
-      // @@protoc_insertion_point(builder_scope:tensorflow.CollectionDef.Int64List)
-    }
-
-    // @@protoc_insertion_point(class_scope:tensorflow.CollectionDef.Int64List)
-    private static final org.tensorflow.proto.framework.CollectionDef.Int64List DEFAULT_INSTANCE;
-    static {
-      DEFAULT_INSTANCE = new org.tensorflow.proto.framework.CollectionDef.Int64List();
-    }
-
-    public static org.tensorflow.proto.framework.CollectionDef.Int64List getDefaultInstance() {
-      return DEFAULT_INSTANCE;
-    }
-
-    private static final com.google.protobuf.Parser<Int64List>
-        PARSER = new com.google.protobuf.AbstractParser<Int64List>() {
-      @java.lang.Override
-      public Int64List parsePartialFrom(
-          com.google.protobuf.CodedInputStream input,
-          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-          throws com.google.protobuf.InvalidProtocolBufferException {
-        return new Int64List(input, extensionRegistry);
-      }
-    };
-
-    public static com.google.protobuf.Parser<Int64List> parser() {
-      return PARSER;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Parser<Int64List> getParserForType() {
-      return PARSER;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.CollectionDef.Int64List getDefaultInstanceForType() {
-      return DEFAULT_INSTANCE;
-    }
-
-  }
-
-  public interface FloatListOrBuilder extends
-      // @@protoc_insertion_point(interface_extends:tensorflow.CollectionDef.FloatList)
-      com.google.protobuf.MessageOrBuilder {
-
-    /**
-     * <code>repeated float value = 1 [packed = true];</code>
-     */
-    java.util.List<java.lang.Float> getValueList();
-    /**
-     * <code>repeated float value = 1 [packed = true];</code>
-     */
-    int getValueCount();
-    /**
-     * <code>repeated float value = 1 [packed = true];</code>
-     */
-    float getValue(int index);
-  }
-  /**
-   * <pre>
-   * FloatList is used for collecting float values.
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.CollectionDef.FloatList}
-   */
-  public  static final class FloatList extends
-      com.google.protobuf.GeneratedMessageV3 implements
-      // @@protoc_insertion_point(message_implements:tensorflow.CollectionDef.FloatList)
-      FloatListOrBuilder {
-  private static final long serialVersionUID = 0L;
-    // Use FloatList.newBuilder() to construct.
-    private FloatList(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-      super(builder);
-    }
-    private FloatList() {
-      value_ = emptyFloatList();
-    }
-
-    @java.lang.Override
-    @SuppressWarnings({"unused"})
-    protected java.lang.Object newInstance(
-        UnusedPrivateParameter unused) {
-      return new FloatList();
-    }
-
-    @java.lang.Override
-    public final com.google.protobuf.UnknownFieldSet
-    getUnknownFields() {
-      return this.unknownFields;
-    }
-    private FloatList(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      this();
-      if (extensionRegistry == null) {
-        throw new java.lang.NullPointerException();
-      }
-      int mutable_bitField0_ = 0;
-      com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-          com.google.protobuf.UnknownFieldSet.newBuilder();
-      try {
-        boolean done = false;
-        while (!done) {
-          int tag = input.readTag();
-          switch (tag) {
-            case 0:
-              done = true;
-              break;
-            case 13: {
-              if (!((mutable_bitField0_ & 0x00000001) != 0)) {
-                value_ = newFloatList();
-                mutable_bitField0_ |= 0x00000001;
-              }
-              value_.addFloat(input.readFloat());
-              break;
-            }
-            case 10: {
-              int length = input.readRawVarint32();
-              int limit = input.pushLimit(length);
-              if (!((mutable_bitField0_ & 0x00000001) != 0) && input.getBytesUntilLimit() > 0) {
-                value_ = newFloatList();
-                mutable_bitField0_ |= 0x00000001;
-              }
-              while (input.getBytesUntilLimit() > 0) {
-                value_.addFloat(input.readFloat());
-              }
-              input.popLimit(limit);
-              break;
-            }
-            default: {
-              if (!parseUnknownField(
-                  input, unknownFields, extensionRegistry, tag)) {
-                done = true;
-              }
-              break;
-            }
-          }
-        }
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        throw e.setUnfinishedMessage(this);
-      } catch (java.io.IOException e) {
-        throw new com.google.protobuf.InvalidProtocolBufferException(
-            e).setUnfinishedMessage(this);
-      } finally {
-        if (((mutable_bitField0_ & 0x00000001) != 0)) {
-          value_.makeImmutable(); // C
-        }
-        this.unknownFields = unknownFields.build();
-        makeExtensionsImmutable();
-      }
-    }
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.MetaGraphProtos.internal_static_tensorflow_CollectionDef_FloatList_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.MetaGraphProtos.internal_static_tensorflow_CollectionDef_FloatList_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.CollectionDef.FloatList.class, org.tensorflow.proto.framework.CollectionDef.FloatList.Builder.class);
-    }
-
-    public static final int VALUE_FIELD_NUMBER = 1;
-    private com.google.protobuf.Internal.FloatList value_;
-    /**
-     * <code>repeated float value = 1 [packed = true];</code>
-     */
-    public java.util.List<java.lang.Float>
-        getValueList() {
-      return value_;
-    }
-    /**
-     * <code>repeated float value = 1 [packed = true];</code>
-     */
-    public int getValueCount() {
-      return value_.size();
-    }
-    /**
-     * <code>repeated float value = 1 [packed = true];</code>
-     */
-    public float getValue(int index) {
-      return value_.getFloat(index);
-    }
-    private int valueMemoizedSerializedSize = -1;
-
-    private byte memoizedIsInitialized = -1;
-    @java.lang.Override
-    public final boolean isInitialized() {
-      byte isInitialized = memoizedIsInitialized;
-      if (isInitialized == 1) return true;
-      if (isInitialized == 0) return false;
-
-      memoizedIsInitialized = 1;
-      return true;
-    }
-
-    @java.lang.Override
-    public void writeTo(com.google.protobuf.CodedOutputStream output)
-                        throws java.io.IOException {
-      getSerializedSize();
-      if (getValueList().size() > 0) {
-        output.writeUInt32NoTag(10);
-        output.writeUInt32NoTag(valueMemoizedSerializedSize);
-      }
-      for (int i = 0; i < value_.size(); i++) {
-        output.writeFloatNoTag(value_.getFloat(i));
-      }
-      unknownFields.writeTo(output);
-    }
-
-    @java.lang.Override
-    public int getSerializedSize() {
-      int size = memoizedSize;
-      if (size != -1) return size;
-
-      size = 0;
-      {
-        int dataSize = 0;
-        dataSize = 4 * getValueList().size();
-        size += dataSize;
-        if (!getValueList().isEmpty()) {
-          size += 1;
-          size += com.google.protobuf.CodedOutputStream
-              .computeInt32SizeNoTag(dataSize);
-        }
-        valueMemoizedSerializedSize = dataSize;
-      }
-      size += unknownFields.getSerializedSize();
-      memoizedSize = size;
-      return size;
-    }
-
-    @java.lang.Override
-    public boolean equals(final java.lang.Object obj) {
-      if (obj == this) {
-       return true;
-      }
-      if (!(obj instanceof org.tensorflow.proto.framework.CollectionDef.FloatList)) {
-        return super.equals(obj);
-      }
-      org.tensorflow.proto.framework.CollectionDef.FloatList other = (org.tensorflow.proto.framework.CollectionDef.FloatList) obj;
-
-      if (!getValueList()
-          .equals(other.getValueList())) return false;
-      if (!unknownFields.equals(other.unknownFields)) return false;
-      return true;
-    }
-
-    @java.lang.Override
-    public int hashCode() {
-      if (memoizedHashCode != 0) {
-        return memoizedHashCode;
-      }
-      int hash = 41;
-      hash = (19 * hash) + getDescriptor().hashCode();
-      if (getValueCount() > 0) {
-        hash = (37 * hash) + VALUE_FIELD_NUMBER;
-        hash = (53 * hash) + getValueList().hashCode();
-      }
-      hash = (29 * hash) + unknownFields.hashCode();
-      memoizedHashCode = hash;
-      return hash;
-    }
-
-    public static org.tensorflow.proto.framework.CollectionDef.FloatList parseFrom(
-        java.nio.ByteBuffer data)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data);
-    }
-    public static org.tensorflow.proto.framework.CollectionDef.FloatList parseFrom(
-        java.nio.ByteBuffer data,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.CollectionDef.FloatList parseFrom(
-        com.google.protobuf.ByteString data)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data);
-    }
-    public static org.tensorflow.proto.framework.CollectionDef.FloatList parseFrom(
-        com.google.protobuf.ByteString data,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.CollectionDef.FloatList parseFrom(byte[] data)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data);
-    }
-    public static org.tensorflow.proto.framework.CollectionDef.FloatList parseFrom(
-        byte[] data,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.CollectionDef.FloatList parseFrom(java.io.InputStream input)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input);
-    }
-    public static org.tensorflow.proto.framework.CollectionDef.FloatList parseFrom(
-        java.io.InputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.CollectionDef.FloatList parseDelimitedFrom(java.io.InputStream input)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseDelimitedWithIOException(PARSER, input);
-    }
-    public static org.tensorflow.proto.framework.CollectionDef.FloatList parseDelimitedFrom(
-        java.io.InputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.CollectionDef.FloatList parseFrom(
-        com.google.protobuf.CodedInputStream input)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input);
-    }
-    public static org.tensorflow.proto.framework.CollectionDef.FloatList parseFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input, extensionRegistry);
-    }
-
-    @java.lang.Override
-    public Builder newBuilderForType() { return newBuilder(); }
-    public static Builder newBuilder() {
-      return DEFAULT_INSTANCE.toBuilder();
-    }
-    public static Builder newBuilder(org.tensorflow.proto.framework.CollectionDef.FloatList prototype) {
-      return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-    }
-    @java.lang.Override
-    public Builder toBuilder() {
-      return this == DEFAULT_INSTANCE
-          ? new Builder() : new Builder().mergeFrom(this);
-    }
-
-    @java.lang.Override
-    protected Builder newBuilderForType(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      Builder builder = new Builder(parent);
-      return builder;
-    }
-    /**
-     * <pre>
-     * FloatList is used for collecting float values.
-     * </pre>
-     *
-     * Protobuf type {@code tensorflow.CollectionDef.FloatList}
-     */
-    public static final class Builder extends
-        com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-        // @@protoc_insertion_point(builder_implements:tensorflow.CollectionDef.FloatList)
-        org.tensorflow.proto.framework.CollectionDef.FloatListOrBuilder {
-      public static final com.google.protobuf.Descriptors.Descriptor
-          getDescriptor() {
-        return org.tensorflow.proto.framework.MetaGraphProtos.internal_static_tensorflow_CollectionDef_FloatList_descriptor;
-      }
-
-      @java.lang.Override
-      protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-          internalGetFieldAccessorTable() {
-        return org.tensorflow.proto.framework.MetaGraphProtos.internal_static_tensorflow_CollectionDef_FloatList_fieldAccessorTable
-            .ensureFieldAccessorsInitialized(
-                org.tensorflow.proto.framework.CollectionDef.FloatList.class, org.tensorflow.proto.framework.CollectionDef.FloatList.Builder.class);
-      }
-
-      // Construct using org.tensorflow.proto.framework.CollectionDef.FloatList.newBuilder()
-      private Builder() {
-        maybeForceBuilderInitialization();
-      }
-
-      private Builder(
-          com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-        super(parent);
-        maybeForceBuilderInitialization();
-      }
-      private void maybeForceBuilderInitialization() {
-        if (com.google.protobuf.GeneratedMessageV3
-                .alwaysUseFieldBuilders) {
-        }
-      }
-      @java.lang.Override
-      public Builder clear() {
-        super.clear();
-        value_ = emptyFloatList();
-        bitField0_ = (bitField0_ & ~0x00000001);
-        return this;
-      }
-
-      @java.lang.Override
-      public com.google.protobuf.Descriptors.Descriptor
-          getDescriptorForType() {
-        return org.tensorflow.proto.framework.MetaGraphProtos.internal_static_tensorflow_CollectionDef_FloatList_descriptor;
-      }
-
-      @java.lang.Override
-      public org.tensorflow.proto.framework.CollectionDef.FloatList getDefaultInstanceForType() {
-        return org.tensorflow.proto.framework.CollectionDef.FloatList.getDefaultInstance();
-      }
-
-      @java.lang.Override
-      public org.tensorflow.proto.framework.CollectionDef.FloatList build() {
-        org.tensorflow.proto.framework.CollectionDef.FloatList result = buildPartial();
-        if (!result.isInitialized()) {
-          throw newUninitializedMessageException(result);
-        }
-        return result;
-      }
-
-      @java.lang.Override
-      public org.tensorflow.proto.framework.CollectionDef.FloatList buildPartial() {
-        org.tensorflow.proto.framework.CollectionDef.FloatList result = new org.tensorflow.proto.framework.CollectionDef.FloatList(this);
-        int from_bitField0_ = bitField0_;
-        if (((bitField0_ & 0x00000001) != 0)) {
-          value_.makeImmutable();
-          bitField0_ = (bitField0_ & ~0x00000001);
-        }
-        result.value_ = value_;
-        onBuilt();
-        return result;
-      }
-
-      @java.lang.Override
-      public Builder clone() {
-        return super.clone();
-      }
-      @java.lang.Override
-      public Builder setField(
-          com.google.protobuf.Descriptors.FieldDescriptor field,
-          java.lang.Object value) {
-        return super.setField(field, value);
-      }
-      @java.lang.Override
-      public Builder clearField(
-          com.google.protobuf.Descriptors.FieldDescriptor field) {
-        return super.clearField(field);
-      }
-      @java.lang.Override
-      public Builder clearOneof(
-          com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-        return super.clearOneof(oneof);
-      }
-      @java.lang.Override
-      public Builder setRepeatedField(
-          com.google.protobuf.Descriptors.FieldDescriptor field,
-          int index, java.lang.Object value) {
-        return super.setRepeatedField(field, index, value);
-      }
-      @java.lang.Override
-      public Builder addRepeatedField(
-          com.google.protobuf.Descriptors.FieldDescriptor field,
-          java.lang.Object value) {
-        return super.addRepeatedField(field, value);
-      }
-      @java.lang.Override
-      public Builder mergeFrom(com.google.protobuf.Message other) {
-        if (other instanceof org.tensorflow.proto.framework.CollectionDef.FloatList) {
-          return mergeFrom((org.tensorflow.proto.framework.CollectionDef.FloatList)other);
-        } else {
-          super.mergeFrom(other);
-          return this;
-        }
-      }
-
-      public Builder mergeFrom(org.tensorflow.proto.framework.CollectionDef.FloatList other) {
-        if (other == org.tensorflow.proto.framework.CollectionDef.FloatList.getDefaultInstance()) return this;
-        if (!other.value_.isEmpty()) {
-          if (value_.isEmpty()) {
-            value_ = other.value_;
-            bitField0_ = (bitField0_ & ~0x00000001);
-          } else {
-            ensureValueIsMutable();
-            value_.addAll(other.value_);
-          }
-          onChanged();
-        }
-        this.mergeUnknownFields(other.unknownFields);
-        onChanged();
-        return this;
-      }
-
-      @java.lang.Override
-      public final boolean isInitialized() {
-        return true;
-      }
-
-      @java.lang.Override
-      public Builder mergeFrom(
-          com.google.protobuf.CodedInputStream input,
-          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-          throws java.io.IOException {
-        org.tensorflow.proto.framework.CollectionDef.FloatList parsedMessage = null;
-        try {
-          parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-        } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-          parsedMessage = (org.tensorflow.proto.framework.CollectionDef.FloatList) e.getUnfinishedMessage();
-          throw e.unwrapIOException();
-        } finally {
-          if (parsedMessage != null) {
-            mergeFrom(parsedMessage);
-          }
-        }
-        return this;
-      }
-      private int bitField0_;
-
-      private com.google.protobuf.Internal.FloatList value_ = emptyFloatList();
-      private void ensureValueIsMutable() {
-        if (!((bitField0_ & 0x00000001) != 0)) {
-          value_ = mutableCopy(value_);
-          bitField0_ |= 0x00000001;
-         }
-      }
-      /**
-       * <code>repeated float value = 1 [packed = true];</code>
-       */
-      public java.util.List<java.lang.Float>
-          getValueList() {
-        return ((bitField0_ & 0x00000001) != 0) ?
-                 java.util.Collections.unmodifiableList(value_) : value_;
-      }
-      /**
-       * <code>repeated float value = 1 [packed = true];</code>
-       */
-      public int getValueCount() {
-        return value_.size();
-      }
-      /**
-       * <code>repeated float value = 1 [packed = true];</code>
-       */
-      public float getValue(int index) {
-        return value_.getFloat(index);
-      }
-      /**
-       * <code>repeated float value = 1 [packed = true];</code>
-       */
-      public Builder setValue(
-          int index, float value) {
-        ensureValueIsMutable();
-        value_.setFloat(index, value);
-        onChanged();
-        return this;
-      }
-      /**
-       * <code>repeated float value = 1 [packed = true];</code>
-       */
-      public Builder addValue(float value) {
-        ensureValueIsMutable();
-        value_.addFloat(value);
-        onChanged();
-        return this;
-      }
-      /**
-       * <code>repeated float value = 1 [packed = true];</code>
-       */
-      public Builder addAllValue(
-          java.lang.Iterable<? extends java.lang.Float> values) {
-        ensureValueIsMutable();
-        com.google.protobuf.AbstractMessageLite.Builder.addAll(
-            values, value_);
-        onChanged();
-        return this;
-      }
-      /**
-       * <code>repeated float value = 1 [packed = true];</code>
-       */
-      public Builder clearValue() {
-        value_ = emptyFloatList();
-        bitField0_ = (bitField0_ & ~0x00000001);
-        onChanged();
-        return this;
-      }
-      @java.lang.Override
-      public final Builder setUnknownFields(
-          final com.google.protobuf.UnknownFieldSet unknownFields) {
-        return super.setUnknownFields(unknownFields);
-      }
-
-      @java.lang.Override
-      public final Builder mergeUnknownFields(
-          final com.google.protobuf.UnknownFieldSet unknownFields) {
-        return super.mergeUnknownFields(unknownFields);
-      }
-
-
-      // @@protoc_insertion_point(builder_scope:tensorflow.CollectionDef.FloatList)
-    }
-
-    // @@protoc_insertion_point(class_scope:tensorflow.CollectionDef.FloatList)
-    private static final org.tensorflow.proto.framework.CollectionDef.FloatList DEFAULT_INSTANCE;
-    static {
-      DEFAULT_INSTANCE = new org.tensorflow.proto.framework.CollectionDef.FloatList();
-    }
-
-    public static org.tensorflow.proto.framework.CollectionDef.FloatList getDefaultInstance() {
-      return DEFAULT_INSTANCE;
-    }
-
-    private static final com.google.protobuf.Parser<FloatList>
-        PARSER = new com.google.protobuf.AbstractParser<FloatList>() {
-      @java.lang.Override
-      public FloatList parsePartialFrom(
-          com.google.protobuf.CodedInputStream input,
-          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-          throws com.google.protobuf.InvalidProtocolBufferException {
-        return new FloatList(input, extensionRegistry);
-      }
-    };
-
-    public static com.google.protobuf.Parser<FloatList> parser() {
-      return PARSER;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Parser<FloatList> getParserForType() {
-      return PARSER;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.CollectionDef.FloatList getDefaultInstanceForType() {
-      return DEFAULT_INSTANCE;
-    }
-
-  }
-
-  public interface AnyListOrBuilder extends
-      // @@protoc_insertion_point(interface_extends:tensorflow.CollectionDef.AnyList)
-      com.google.protobuf.MessageOrBuilder {
-
-    /**
-     * <code>repeated .google.protobuf.Any value = 1;</code>
-     */
-    java.util.List<com.google.protobuf.Any> 
-        getValueList();
-    /**
-     * <code>repeated .google.protobuf.Any value = 1;</code>
-     */
-    com.google.protobuf.Any getValue(int index);
-    /**
-     * <code>repeated .google.protobuf.Any value = 1;</code>
-     */
-    int getValueCount();
-    /**
-     * <code>repeated .google.protobuf.Any value = 1;</code>
-     */
-    java.util.List<? extends com.google.protobuf.AnyOrBuilder> 
-        getValueOrBuilderList();
-    /**
-     * <code>repeated .google.protobuf.Any value = 1;</code>
-     */
-    com.google.protobuf.AnyOrBuilder getValueOrBuilder(
-        int index);
-  }
-  /**
-   * <pre>
-   * AnyList is used for collecting Any protos.
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.CollectionDef.AnyList}
-   */
-  public  static final class AnyList extends
-      com.google.protobuf.GeneratedMessageV3 implements
-      // @@protoc_insertion_point(message_implements:tensorflow.CollectionDef.AnyList)
-      AnyListOrBuilder {
-  private static final long serialVersionUID = 0L;
-    // Use AnyList.newBuilder() to construct.
-    private AnyList(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-      super(builder);
-    }
-    private AnyList() {
-      value_ = java.util.Collections.emptyList();
-    }
-
-    @java.lang.Override
-    @SuppressWarnings({"unused"})
-    protected java.lang.Object newInstance(
-        UnusedPrivateParameter unused) {
-      return new AnyList();
-    }
-
-    @java.lang.Override
-    public final com.google.protobuf.UnknownFieldSet
-    getUnknownFields() {
-      return this.unknownFields;
-    }
-    private AnyList(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      this();
-      if (extensionRegistry == null) {
-        throw new java.lang.NullPointerException();
-      }
-      int mutable_bitField0_ = 0;
-      com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-          com.google.protobuf.UnknownFieldSet.newBuilder();
-      try {
-        boolean done = false;
-        while (!done) {
-          int tag = input.readTag();
-          switch (tag) {
-            case 0:
-              done = true;
-              break;
-            case 10: {
-              if (!((mutable_bitField0_ & 0x00000001) != 0)) {
-                value_ = new java.util.ArrayList<com.google.protobuf.Any>();
-                mutable_bitField0_ |= 0x00000001;
-              }
-              value_.add(
-                  input.readMessage(com.google.protobuf.Any.parser(), extensionRegistry));
-              break;
-            }
-            default: {
-              if (!parseUnknownField(
-                  input, unknownFields, extensionRegistry, tag)) {
-                done = true;
-              }
-              break;
-            }
-          }
-        }
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        throw e.setUnfinishedMessage(this);
-      } catch (java.io.IOException e) {
-        throw new com.google.protobuf.InvalidProtocolBufferException(
-            e).setUnfinishedMessage(this);
-      } finally {
-        if (((mutable_bitField0_ & 0x00000001) != 0)) {
-          value_ = java.util.Collections.unmodifiableList(value_);
-        }
-        this.unknownFields = unknownFields.build();
-        makeExtensionsImmutable();
-      }
-    }
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.MetaGraphProtos.internal_static_tensorflow_CollectionDef_AnyList_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.MetaGraphProtos.internal_static_tensorflow_CollectionDef_AnyList_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.CollectionDef.AnyList.class, org.tensorflow.proto.framework.CollectionDef.AnyList.Builder.class);
-    }
-
-    public static final int VALUE_FIELD_NUMBER = 1;
-    private java.util.List<com.google.protobuf.Any> value_;
-    /**
-     * <code>repeated .google.protobuf.Any value = 1;</code>
-     */
-    public java.util.List<com.google.protobuf.Any> getValueList() {
-      return value_;
-    }
-    /**
-     * <code>repeated .google.protobuf.Any value = 1;</code>
-     */
-    public java.util.List<? extends com.google.protobuf.AnyOrBuilder> 
-        getValueOrBuilderList() {
-      return value_;
-    }
-    /**
-     * <code>repeated .google.protobuf.Any value = 1;</code>
-     */
-    public int getValueCount() {
-      return value_.size();
-    }
-    /**
-     * <code>repeated .google.protobuf.Any value = 1;</code>
-     */
-    public com.google.protobuf.Any getValue(int index) {
-      return value_.get(index);
-    }
-    /**
-     * <code>repeated .google.protobuf.Any value = 1;</code>
-     */
-    public com.google.protobuf.AnyOrBuilder getValueOrBuilder(
-        int index) {
-      return value_.get(index);
-    }
-
-    private byte memoizedIsInitialized = -1;
-    @java.lang.Override
-    public final boolean isInitialized() {
-      byte isInitialized = memoizedIsInitialized;
-      if (isInitialized == 1) return true;
-      if (isInitialized == 0) return false;
-
-      memoizedIsInitialized = 1;
-      return true;
-    }
-
-    @java.lang.Override
-    public void writeTo(com.google.protobuf.CodedOutputStream output)
-                        throws java.io.IOException {
-      for (int i = 0; i < value_.size(); i++) {
-        output.writeMessage(1, value_.get(i));
-      }
-      unknownFields.writeTo(output);
-    }
-
-    @java.lang.Override
-    public int getSerializedSize() {
-      int size = memoizedSize;
-      if (size != -1) return size;
-
-      size = 0;
-      for (int i = 0; i < value_.size(); i++) {
-        size += com.google.protobuf.CodedOutputStream
-          .computeMessageSize(1, value_.get(i));
-      }
-      size += unknownFields.getSerializedSize();
-      memoizedSize = size;
-      return size;
-    }
-
-    @java.lang.Override
-    public boolean equals(final java.lang.Object obj) {
-      if (obj == this) {
-       return true;
-      }
-      if (!(obj instanceof org.tensorflow.proto.framework.CollectionDef.AnyList)) {
-        return super.equals(obj);
-      }
-      org.tensorflow.proto.framework.CollectionDef.AnyList other = (org.tensorflow.proto.framework.CollectionDef.AnyList) obj;
-
-      if (!getValueList()
-          .equals(other.getValueList())) return false;
-      if (!unknownFields.equals(other.unknownFields)) return false;
-      return true;
-    }
-
-    @java.lang.Override
-    public int hashCode() {
-      if (memoizedHashCode != 0) {
-        return memoizedHashCode;
-      }
-      int hash = 41;
-      hash = (19 * hash) + getDescriptor().hashCode();
-      if (getValueCount() > 0) {
-        hash = (37 * hash) + VALUE_FIELD_NUMBER;
-        hash = (53 * hash) + getValueList().hashCode();
-      }
-      hash = (29 * hash) + unknownFields.hashCode();
-      memoizedHashCode = hash;
-      return hash;
-    }
-
-    public static org.tensorflow.proto.framework.CollectionDef.AnyList parseFrom(
-        java.nio.ByteBuffer data)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data);
-    }
-    public static org.tensorflow.proto.framework.CollectionDef.AnyList parseFrom(
-        java.nio.ByteBuffer data,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.CollectionDef.AnyList parseFrom(
-        com.google.protobuf.ByteString data)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data);
-    }
-    public static org.tensorflow.proto.framework.CollectionDef.AnyList parseFrom(
-        com.google.protobuf.ByteString data,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.CollectionDef.AnyList parseFrom(byte[] data)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data);
-    }
-    public static org.tensorflow.proto.framework.CollectionDef.AnyList parseFrom(
-        byte[] data,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.CollectionDef.AnyList parseFrom(java.io.InputStream input)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input);
-    }
-    public static org.tensorflow.proto.framework.CollectionDef.AnyList parseFrom(
-        java.io.InputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.CollectionDef.AnyList parseDelimitedFrom(java.io.InputStream input)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseDelimitedWithIOException(PARSER, input);
-    }
-    public static org.tensorflow.proto.framework.CollectionDef.AnyList parseDelimitedFrom(
-        java.io.InputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.CollectionDef.AnyList parseFrom(
-        com.google.protobuf.CodedInputStream input)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input);
-    }
-    public static org.tensorflow.proto.framework.CollectionDef.AnyList parseFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input, extensionRegistry);
-    }
-
-    @java.lang.Override
-    public Builder newBuilderForType() { return newBuilder(); }
-    public static Builder newBuilder() {
-      return DEFAULT_INSTANCE.toBuilder();
-    }
-    public static Builder newBuilder(org.tensorflow.proto.framework.CollectionDef.AnyList prototype) {
-      return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-    }
-    @java.lang.Override
-    public Builder toBuilder() {
-      return this == DEFAULT_INSTANCE
-          ? new Builder() : new Builder().mergeFrom(this);
-    }
-
-    @java.lang.Override
-    protected Builder newBuilderForType(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      Builder builder = new Builder(parent);
-      return builder;
-    }
-    /**
-     * <pre>
-     * AnyList is used for collecting Any protos.
-     * </pre>
-     *
-     * Protobuf type {@code tensorflow.CollectionDef.AnyList}
-     */
-    public static final class Builder extends
-        com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-        // @@protoc_insertion_point(builder_implements:tensorflow.CollectionDef.AnyList)
-        org.tensorflow.proto.framework.CollectionDef.AnyListOrBuilder {
-      public static final com.google.protobuf.Descriptors.Descriptor
-          getDescriptor() {
-        return org.tensorflow.proto.framework.MetaGraphProtos.internal_static_tensorflow_CollectionDef_AnyList_descriptor;
-      }
-
-      @java.lang.Override
-      protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-          internalGetFieldAccessorTable() {
-        return org.tensorflow.proto.framework.MetaGraphProtos.internal_static_tensorflow_CollectionDef_AnyList_fieldAccessorTable
-            .ensureFieldAccessorsInitialized(
-                org.tensorflow.proto.framework.CollectionDef.AnyList.class, org.tensorflow.proto.framework.CollectionDef.AnyList.Builder.class);
-      }
-
-      // Construct using org.tensorflow.proto.framework.CollectionDef.AnyList.newBuilder()
-      private Builder() {
-        maybeForceBuilderInitialization();
-      }
-
-      private Builder(
-          com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-        super(parent);
-        maybeForceBuilderInitialization();
-      }
-      private void maybeForceBuilderInitialization() {
-        if (com.google.protobuf.GeneratedMessageV3
-                .alwaysUseFieldBuilders) {
-          getValueFieldBuilder();
-        }
-      }
-      @java.lang.Override
-      public Builder clear() {
-        super.clear();
-        if (valueBuilder_ == null) {
-          value_ = java.util.Collections.emptyList();
-          bitField0_ = (bitField0_ & ~0x00000001);
-        } else {
-          valueBuilder_.clear();
-        }
-        return this;
-      }
-
-      @java.lang.Override
-      public com.google.protobuf.Descriptors.Descriptor
-          getDescriptorForType() {
-        return org.tensorflow.proto.framework.MetaGraphProtos.internal_static_tensorflow_CollectionDef_AnyList_descriptor;
-      }
-
-      @java.lang.Override
-      public org.tensorflow.proto.framework.CollectionDef.AnyList getDefaultInstanceForType() {
-        return org.tensorflow.proto.framework.CollectionDef.AnyList.getDefaultInstance();
-      }
-
-      @java.lang.Override
-      public org.tensorflow.proto.framework.CollectionDef.AnyList build() {
-        org.tensorflow.proto.framework.CollectionDef.AnyList result = buildPartial();
-        if (!result.isInitialized()) {
-          throw newUninitializedMessageException(result);
-        }
-        return result;
-      }
-
-      @java.lang.Override
-      public org.tensorflow.proto.framework.CollectionDef.AnyList buildPartial() {
-        org.tensorflow.proto.framework.CollectionDef.AnyList result = new org.tensorflow.proto.framework.CollectionDef.AnyList(this);
-        int from_bitField0_ = bitField0_;
-        if (valueBuilder_ == null) {
-          if (((bitField0_ & 0x00000001) != 0)) {
-            value_ = java.util.Collections.unmodifiableList(value_);
-            bitField0_ = (bitField0_ & ~0x00000001);
-          }
-          result.value_ = value_;
-        } else {
-          result.value_ = valueBuilder_.build();
-        }
-        onBuilt();
-        return result;
-      }
-
-      @java.lang.Override
-      public Builder clone() {
-        return super.clone();
-      }
-      @java.lang.Override
-      public Builder setField(
-          com.google.protobuf.Descriptors.FieldDescriptor field,
-          java.lang.Object value) {
-        return super.setField(field, value);
-      }
-      @java.lang.Override
-      public Builder clearField(
-          com.google.protobuf.Descriptors.FieldDescriptor field) {
-        return super.clearField(field);
-      }
-      @java.lang.Override
-      public Builder clearOneof(
-          com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-        return super.clearOneof(oneof);
-      }
-      @java.lang.Override
-      public Builder setRepeatedField(
-          com.google.protobuf.Descriptors.FieldDescriptor field,
-          int index, java.lang.Object value) {
-        return super.setRepeatedField(field, index, value);
-      }
-      @java.lang.Override
-      public Builder addRepeatedField(
-          com.google.protobuf.Descriptors.FieldDescriptor field,
-          java.lang.Object value) {
-        return super.addRepeatedField(field, value);
-      }
-      @java.lang.Override
-      public Builder mergeFrom(com.google.protobuf.Message other) {
-        if (other instanceof org.tensorflow.proto.framework.CollectionDef.AnyList) {
-          return mergeFrom((org.tensorflow.proto.framework.CollectionDef.AnyList)other);
-        } else {
-          super.mergeFrom(other);
-          return this;
-        }
-      }
-
-      public Builder mergeFrom(org.tensorflow.proto.framework.CollectionDef.AnyList other) {
-        if (other == org.tensorflow.proto.framework.CollectionDef.AnyList.getDefaultInstance()) return this;
-        if (valueBuilder_ == null) {
-          if (!other.value_.isEmpty()) {
-            if (value_.isEmpty()) {
-              value_ = other.value_;
-              bitField0_ = (bitField0_ & ~0x00000001);
-            } else {
-              ensureValueIsMutable();
-              value_.addAll(other.value_);
-            }
-            onChanged();
-          }
-        } else {
-          if (!other.value_.isEmpty()) {
-            if (valueBuilder_.isEmpty()) {
-              valueBuilder_.dispose();
-              valueBuilder_ = null;
-              value_ = other.value_;
-              bitField0_ = (bitField0_ & ~0x00000001);
-              valueBuilder_ = 
-                com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
-                   getValueFieldBuilder() : null;
-            } else {
-              valueBuilder_.addAllMessages(other.value_);
-            }
-          }
-        }
-        this.mergeUnknownFields(other.unknownFields);
-        onChanged();
-        return this;
-      }
-
-      @java.lang.Override
-      public final boolean isInitialized() {
-        return true;
-      }
-
-      @java.lang.Override
-      public Builder mergeFrom(
-          com.google.protobuf.CodedInputStream input,
-          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-          throws java.io.IOException {
-        org.tensorflow.proto.framework.CollectionDef.AnyList parsedMessage = null;
-        try {
-          parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-        } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-          parsedMessage = (org.tensorflow.proto.framework.CollectionDef.AnyList) e.getUnfinishedMessage();
-          throw e.unwrapIOException();
-        } finally {
-          if (parsedMessage != null) {
-            mergeFrom(parsedMessage);
-          }
-        }
-        return this;
-      }
-      private int bitField0_;
-
-      private java.util.List<com.google.protobuf.Any> value_ =
-        java.util.Collections.emptyList();
-      private void ensureValueIsMutable() {
-        if (!((bitField0_ & 0x00000001) != 0)) {
-          value_ = new java.util.ArrayList<com.google.protobuf.Any>(value_);
-          bitField0_ |= 0x00000001;
-         }
-      }
-
-      private com.google.protobuf.RepeatedFieldBuilderV3<
-          com.google.protobuf.Any, com.google.protobuf.Any.Builder, com.google.protobuf.AnyOrBuilder> valueBuilder_;
-
-      /**
-       * <code>repeated .google.protobuf.Any value = 1;</code>
-       */
-      public java.util.List<com.google.protobuf.Any> getValueList() {
-        if (valueBuilder_ == null) {
-          return java.util.Collections.unmodifiableList(value_);
-        } else {
-          return valueBuilder_.getMessageList();
-        }
-      }
-      /**
-       * <code>repeated .google.protobuf.Any value = 1;</code>
-       */
-      public int getValueCount() {
-        if (valueBuilder_ == null) {
-          return value_.size();
-        } else {
-          return valueBuilder_.getCount();
-        }
-      }
-      /**
-       * <code>repeated .google.protobuf.Any value = 1;</code>
-       */
-      public com.google.protobuf.Any getValue(int index) {
-        if (valueBuilder_ == null) {
-          return value_.get(index);
-        } else {
-          return valueBuilder_.getMessage(index);
-        }
-      }
-      /**
-       * <code>repeated .google.protobuf.Any value = 1;</code>
-       */
-      public Builder setValue(
-          int index, com.google.protobuf.Any value) {
-        if (valueBuilder_ == null) {
-          if (value == null) {
-            throw new NullPointerException();
-          }
-          ensureValueIsMutable();
-          value_.set(index, value);
-          onChanged();
-        } else {
-          valueBuilder_.setMessage(index, value);
-        }
-        return this;
-      }
-      /**
-       * <code>repeated .google.protobuf.Any value = 1;</code>
-       */
-      public Builder setValue(
-          int index, com.google.protobuf.Any.Builder builderForValue) {
-        if (valueBuilder_ == null) {
-          ensureValueIsMutable();
-          value_.set(index, builderForValue.build());
-          onChanged();
-        } else {
-          valueBuilder_.setMessage(index, builderForValue.build());
-        }
-        return this;
-      }
-      /**
-       * <code>repeated .google.protobuf.Any value = 1;</code>
-       */
-      public Builder addValue(com.google.protobuf.Any value) {
-        if (valueBuilder_ == null) {
-          if (value == null) {
-            throw new NullPointerException();
-          }
-          ensureValueIsMutable();
-          value_.add(value);
-          onChanged();
-        } else {
-          valueBuilder_.addMessage(value);
-        }
-        return this;
-      }
-      /**
-       * <code>repeated .google.protobuf.Any value = 1;</code>
-       */
-      public Builder addValue(
-          int index, com.google.protobuf.Any value) {
-        if (valueBuilder_ == null) {
-          if (value == null) {
-            throw new NullPointerException();
-          }
-          ensureValueIsMutable();
-          value_.add(index, value);
-          onChanged();
-        } else {
-          valueBuilder_.addMessage(index, value);
-        }
-        return this;
-      }
-      /**
-       * <code>repeated .google.protobuf.Any value = 1;</code>
-       */
-      public Builder addValue(
-          com.google.protobuf.Any.Builder builderForValue) {
-        if (valueBuilder_ == null) {
-          ensureValueIsMutable();
-          value_.add(builderForValue.build());
-          onChanged();
-        } else {
-          valueBuilder_.addMessage(builderForValue.build());
-        }
-        return this;
-      }
-      /**
-       * <code>repeated .google.protobuf.Any value = 1;</code>
-       */
-      public Builder addValue(
-          int index, com.google.protobuf.Any.Builder builderForValue) {
-        if (valueBuilder_ == null) {
-          ensureValueIsMutable();
-          value_.add(index, builderForValue.build());
-          onChanged();
-        } else {
-          valueBuilder_.addMessage(index, builderForValue.build());
-        }
-        return this;
-      }
-      /**
-       * <code>repeated .google.protobuf.Any value = 1;</code>
-       */
-      public Builder addAllValue(
-          java.lang.Iterable<? extends com.google.protobuf.Any> values) {
-        if (valueBuilder_ == null) {
-          ensureValueIsMutable();
-          com.google.protobuf.AbstractMessageLite.Builder.addAll(
-              values, value_);
-          onChanged();
-        } else {
-          valueBuilder_.addAllMessages(values);
-        }
-        return this;
-      }
-      /**
-       * <code>repeated .google.protobuf.Any value = 1;</code>
-       */
-      public Builder clearValue() {
-        if (valueBuilder_ == null) {
-          value_ = java.util.Collections.emptyList();
-          bitField0_ = (bitField0_ & ~0x00000001);
-          onChanged();
-        } else {
-          valueBuilder_.clear();
-        }
-        return this;
-      }
-      /**
-       * <code>repeated .google.protobuf.Any value = 1;</code>
-       */
-      public Builder removeValue(int index) {
-        if (valueBuilder_ == null) {
-          ensureValueIsMutable();
-          value_.remove(index);
-          onChanged();
-        } else {
-          valueBuilder_.remove(index);
-        }
-        return this;
-      }
-      /**
-       * <code>repeated .google.protobuf.Any value = 1;</code>
-       */
-      public com.google.protobuf.Any.Builder getValueBuilder(
-          int index) {
-        return getValueFieldBuilder().getBuilder(index);
-      }
-      /**
-       * <code>repeated .google.protobuf.Any value = 1;</code>
-       */
-      public com.google.protobuf.AnyOrBuilder getValueOrBuilder(
-          int index) {
-        if (valueBuilder_ == null) {
-          return value_.get(index);  } else {
-          return valueBuilder_.getMessageOrBuilder(index);
-        }
-      }
-      /**
-       * <code>repeated .google.protobuf.Any value = 1;</code>
-       */
-      public java.util.List<? extends com.google.protobuf.AnyOrBuilder> 
-           getValueOrBuilderList() {
-        if (valueBuilder_ != null) {
-          return valueBuilder_.getMessageOrBuilderList();
-        } else {
-          return java.util.Collections.unmodifiableList(value_);
-        }
-      }
-      /**
-       * <code>repeated .google.protobuf.Any value = 1;</code>
-       */
-      public com.google.protobuf.Any.Builder addValueBuilder() {
-        return getValueFieldBuilder().addBuilder(
-            com.google.protobuf.Any.getDefaultInstance());
-      }
-      /**
-       * <code>repeated .google.protobuf.Any value = 1;</code>
-       */
-      public com.google.protobuf.Any.Builder addValueBuilder(
-          int index) {
-        return getValueFieldBuilder().addBuilder(
-            index, com.google.protobuf.Any.getDefaultInstance());
-      }
-      /**
-       * <code>repeated .google.protobuf.Any value = 1;</code>
-       */
-      public java.util.List<com.google.protobuf.Any.Builder> 
-           getValueBuilderList() {
-        return getValueFieldBuilder().getBuilderList();
-      }
-      private com.google.protobuf.RepeatedFieldBuilderV3<
-          com.google.protobuf.Any, com.google.protobuf.Any.Builder, com.google.protobuf.AnyOrBuilder> 
-          getValueFieldBuilder() {
-        if (valueBuilder_ == null) {
-          valueBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
-              com.google.protobuf.Any, com.google.protobuf.Any.Builder, com.google.protobuf.AnyOrBuilder>(
-                  value_,
-                  ((bitField0_ & 0x00000001) != 0),
-                  getParentForChildren(),
-                  isClean());
-          value_ = null;
-        }
-        return valueBuilder_;
-      }
-      @java.lang.Override
-      public final Builder setUnknownFields(
-          final com.google.protobuf.UnknownFieldSet unknownFields) {
-        return super.setUnknownFields(unknownFields);
-      }
-
-      @java.lang.Override
-      public final Builder mergeUnknownFields(
-          final com.google.protobuf.UnknownFieldSet unknownFields) {
-        return super.mergeUnknownFields(unknownFields);
-      }
-
-
-      // @@protoc_insertion_point(builder_scope:tensorflow.CollectionDef.AnyList)
-    }
-
-    // @@protoc_insertion_point(class_scope:tensorflow.CollectionDef.AnyList)
-    private static final org.tensorflow.proto.framework.CollectionDef.AnyList DEFAULT_INSTANCE;
-    static {
-      DEFAULT_INSTANCE = new org.tensorflow.proto.framework.CollectionDef.AnyList();
-    }
-
-    public static org.tensorflow.proto.framework.CollectionDef.AnyList getDefaultInstance() {
-      return DEFAULT_INSTANCE;
-    }
-
-    private static final com.google.protobuf.Parser<AnyList>
-        PARSER = new com.google.protobuf.AbstractParser<AnyList>() {
-      @java.lang.Override
-      public AnyList parsePartialFrom(
-          com.google.protobuf.CodedInputStream input,
-          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-          throws com.google.protobuf.InvalidProtocolBufferException {
-        return new AnyList(input, extensionRegistry);
-      }
-    };
-
-    public static com.google.protobuf.Parser<AnyList> parser() {
-      return PARSER;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Parser<AnyList> getParserForType() {
-      return PARSER;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.CollectionDef.AnyList getDefaultInstanceForType() {
-      return DEFAULT_INSTANCE;
-    }
-
-  }
-
-  private int kindCase_ = 0;
-  private java.lang.Object kind_;
-  public enum KindCase
-      implements com.google.protobuf.Internal.EnumLite {
-    NODE_LIST(1),
-    BYTES_LIST(2),
-    INT64_LIST(3),
-    FLOAT_LIST(4),
-    ANY_LIST(5),
-    KIND_NOT_SET(0);
-    private final int value;
-    private KindCase(int value) {
-      this.value = value;
-    }
-    /**
-     * @deprecated Use {@link #forNumber(int)} instead.
-     */
-    @java.lang.Deprecated
-    public static KindCase valueOf(int value) {
-      return forNumber(value);
-    }
-
-    public static KindCase forNumber(int value) {
-      switch (value) {
-        case 1: return NODE_LIST;
-        case 2: return BYTES_LIST;
-        case 3: return INT64_LIST;
-        case 4: return FLOAT_LIST;
-        case 5: return ANY_LIST;
-        case 0: return KIND_NOT_SET;
-        default: return null;
-      }
-    }
-    public int getNumber() {
-      return this.value;
-    }
-  };
-
-  public KindCase
-  getKindCase() {
-    return KindCase.forNumber(
-        kindCase_);
-  }
-
-  public static final int NODE_LIST_FIELD_NUMBER = 1;
-  /**
-   * <code>.tensorflow.CollectionDef.NodeList node_list = 1;</code>
-   */
-  public boolean hasNodeList() {
-    return kindCase_ == 1;
-  }
-  /**
-   * <code>.tensorflow.CollectionDef.NodeList node_list = 1;</code>
-   */
-  public org.tensorflow.proto.framework.CollectionDef.NodeList getNodeList() {
-    if (kindCase_ == 1) {
-       return (org.tensorflow.proto.framework.CollectionDef.NodeList) kind_;
-    }
-    return org.tensorflow.proto.framework.CollectionDef.NodeList.getDefaultInstance();
-  }
-  /**
-   * <code>.tensorflow.CollectionDef.NodeList node_list = 1;</code>
-   */
-  public org.tensorflow.proto.framework.CollectionDef.NodeListOrBuilder getNodeListOrBuilder() {
-    if (kindCase_ == 1) {
-       return (org.tensorflow.proto.framework.CollectionDef.NodeList) kind_;
-    }
-    return org.tensorflow.proto.framework.CollectionDef.NodeList.getDefaultInstance();
-  }
-
-  public static final int BYTES_LIST_FIELD_NUMBER = 2;
-  /**
-   * <code>.tensorflow.CollectionDef.BytesList bytes_list = 2;</code>
-   */
-  public boolean hasBytesList() {
-    return kindCase_ == 2;
-  }
-  /**
-   * <code>.tensorflow.CollectionDef.BytesList bytes_list = 2;</code>
-   */
-  public org.tensorflow.proto.framework.CollectionDef.BytesList getBytesList() {
-    if (kindCase_ == 2) {
-       return (org.tensorflow.proto.framework.CollectionDef.BytesList) kind_;
-    }
-    return org.tensorflow.proto.framework.CollectionDef.BytesList.getDefaultInstance();
-  }
-  /**
-   * <code>.tensorflow.CollectionDef.BytesList bytes_list = 2;</code>
-   */
-  public org.tensorflow.proto.framework.CollectionDef.BytesListOrBuilder getBytesListOrBuilder() {
-    if (kindCase_ == 2) {
-       return (org.tensorflow.proto.framework.CollectionDef.BytesList) kind_;
-    }
-    return org.tensorflow.proto.framework.CollectionDef.BytesList.getDefaultInstance();
-  }
-
-  public static final int INT64_LIST_FIELD_NUMBER = 3;
-  /**
-   * <code>.tensorflow.CollectionDef.Int64List int64_list = 3;</code>
-   */
-  public boolean hasInt64List() {
-    return kindCase_ == 3;
-  }
-  /**
-   * <code>.tensorflow.CollectionDef.Int64List int64_list = 3;</code>
-   */
-  public org.tensorflow.proto.framework.CollectionDef.Int64List getInt64List() {
-    if (kindCase_ == 3) {
-       return (org.tensorflow.proto.framework.CollectionDef.Int64List) kind_;
-    }
-    return org.tensorflow.proto.framework.CollectionDef.Int64List.getDefaultInstance();
-  }
-  /**
-   * <code>.tensorflow.CollectionDef.Int64List int64_list = 3;</code>
-   */
-  public org.tensorflow.proto.framework.CollectionDef.Int64ListOrBuilder getInt64ListOrBuilder() {
-    if (kindCase_ == 3) {
-       return (org.tensorflow.proto.framework.CollectionDef.Int64List) kind_;
-    }
-    return org.tensorflow.proto.framework.CollectionDef.Int64List.getDefaultInstance();
-  }
-
-  public static final int FLOAT_LIST_FIELD_NUMBER = 4;
-  /**
-   * <code>.tensorflow.CollectionDef.FloatList float_list = 4;</code>
-   */
-  public boolean hasFloatList() {
-    return kindCase_ == 4;
-  }
-  /**
-   * <code>.tensorflow.CollectionDef.FloatList float_list = 4;</code>
-   */
-  public org.tensorflow.proto.framework.CollectionDef.FloatList getFloatList() {
-    if (kindCase_ == 4) {
-       return (org.tensorflow.proto.framework.CollectionDef.FloatList) kind_;
-    }
-    return org.tensorflow.proto.framework.CollectionDef.FloatList.getDefaultInstance();
-  }
-  /**
-   * <code>.tensorflow.CollectionDef.FloatList float_list = 4;</code>
-   */
-  public org.tensorflow.proto.framework.CollectionDef.FloatListOrBuilder getFloatListOrBuilder() {
-    if (kindCase_ == 4) {
-       return (org.tensorflow.proto.framework.CollectionDef.FloatList) kind_;
-    }
-    return org.tensorflow.proto.framework.CollectionDef.FloatList.getDefaultInstance();
-  }
-
-  public static final int ANY_LIST_FIELD_NUMBER = 5;
-  /**
-   * <code>.tensorflow.CollectionDef.AnyList any_list = 5;</code>
-   */
-  public boolean hasAnyList() {
-    return kindCase_ == 5;
-  }
-  /**
-   * <code>.tensorflow.CollectionDef.AnyList any_list = 5;</code>
-   */
-  public org.tensorflow.proto.framework.CollectionDef.AnyList getAnyList() {
-    if (kindCase_ == 5) {
-       return (org.tensorflow.proto.framework.CollectionDef.AnyList) kind_;
-    }
-    return org.tensorflow.proto.framework.CollectionDef.AnyList.getDefaultInstance();
-  }
-  /**
-   * <code>.tensorflow.CollectionDef.AnyList any_list = 5;</code>
-   */
-  public org.tensorflow.proto.framework.CollectionDef.AnyListOrBuilder getAnyListOrBuilder() {
-    if (kindCase_ == 5) {
-       return (org.tensorflow.proto.framework.CollectionDef.AnyList) kind_;
-    }
-    return org.tensorflow.proto.framework.CollectionDef.AnyList.getDefaultInstance();
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    if (kindCase_ == 1) {
-      output.writeMessage(1, (org.tensorflow.proto.framework.CollectionDef.NodeList) kind_);
-    }
-    if (kindCase_ == 2) {
-      output.writeMessage(2, (org.tensorflow.proto.framework.CollectionDef.BytesList) kind_);
-    }
-    if (kindCase_ == 3) {
-      output.writeMessage(3, (org.tensorflow.proto.framework.CollectionDef.Int64List) kind_);
-    }
-    if (kindCase_ == 4) {
-      output.writeMessage(4, (org.tensorflow.proto.framework.CollectionDef.FloatList) kind_);
-    }
-    if (kindCase_ == 5) {
-      output.writeMessage(5, (org.tensorflow.proto.framework.CollectionDef.AnyList) kind_);
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (kindCase_ == 1) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(1, (org.tensorflow.proto.framework.CollectionDef.NodeList) kind_);
-    }
-    if (kindCase_ == 2) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(2, (org.tensorflow.proto.framework.CollectionDef.BytesList) kind_);
-    }
-    if (kindCase_ == 3) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(3, (org.tensorflow.proto.framework.CollectionDef.Int64List) kind_);
-    }
-    if (kindCase_ == 4) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(4, (org.tensorflow.proto.framework.CollectionDef.FloatList) kind_);
-    }
-    if (kindCase_ == 5) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(5, (org.tensorflow.proto.framework.CollectionDef.AnyList) kind_);
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.framework.CollectionDef)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.framework.CollectionDef other = (org.tensorflow.proto.framework.CollectionDef) obj;
-
-    if (!getKindCase().equals(other.getKindCase())) return false;
-    switch (kindCase_) {
-      case 1:
-        if (!getNodeList()
-            .equals(other.getNodeList())) return false;
-        break;
-      case 2:
-        if (!getBytesList()
-            .equals(other.getBytesList())) return false;
-        break;
-      case 3:
-        if (!getInt64List()
-            .equals(other.getInt64List())) return false;
-        break;
-      case 4:
-        if (!getFloatList()
-            .equals(other.getFloatList())) return false;
-        break;
-      case 5:
-        if (!getAnyList()
-            .equals(other.getAnyList())) return false;
-        break;
-      case 0:
-      default:
-    }
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    switch (kindCase_) {
-      case 1:
-        hash = (37 * hash) + NODE_LIST_FIELD_NUMBER;
-        hash = (53 * hash) + getNodeList().hashCode();
-        break;
-      case 2:
-        hash = (37 * hash) + BYTES_LIST_FIELD_NUMBER;
-        hash = (53 * hash) + getBytesList().hashCode();
-        break;
-      case 3:
-        hash = (37 * hash) + INT64_LIST_FIELD_NUMBER;
-        hash = (53 * hash) + getInt64List().hashCode();
-        break;
-      case 4:
-        hash = (37 * hash) + FLOAT_LIST_FIELD_NUMBER;
-        hash = (53 * hash) + getFloatList().hashCode();
-        break;
-      case 5:
-        hash = (37 * hash) + ANY_LIST_FIELD_NUMBER;
-        hash = (53 * hash) + getAnyList().hashCode();
-        break;
-      case 0:
-      default:
-    }
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.framework.CollectionDef parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.CollectionDef parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.CollectionDef parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.CollectionDef parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.CollectionDef parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.CollectionDef parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.CollectionDef parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.CollectionDef parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.CollectionDef parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.CollectionDef parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.CollectionDef parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.CollectionDef parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.framework.CollectionDef prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * <pre>
-   * CollectionDef should cover most collections.
-   * To add a user-defined collection, do one of the following:
-   * 1. For simple data types, such as string, int, float:
-   *      tf.add_to_collection("your_collection_name", your_simple_value)
-   *    strings will be stored as bytes_list.
-   * 2. For Protobuf types, there are three ways to add them:
-   *    1) tf.add_to_collection("your_collection_name",
-   *         your_proto.SerializeToString())
-   *       collection_def {
-   *         key: "user_defined_bytes_collection"
-   *         value {
-   *           bytes_list {
-   *             value: "queue_name: &#92;"test_queue&#92;"&#92;n"
-   *           }
-   *         }
-   *       }
-   *  or
-   *    2) tf.add_to_collection("your_collection_name", str(your_proto))
-   *       collection_def {
-   *         key: "user_defined_string_collection"
-   *         value {
-   *          bytes_list {
-   *             value: "&#92;n&#92;ntest_queue"
-   *           }
-   *         }
-   *       }
-   *  or
-   *    3) any_buf = any_pb2.Any()
-   *       tf.add_to_collection("your_collection_name",
-   *         any_buf.Pack(your_proto))
-   *       collection_def {
-   *         key: "user_defined_any_collection"
-   *         value {
-   *           any_list {
-   *             value {
-   *               type_url: "type.googleapis.com/tensorflow.QueueRunnerDef"
-   *               value: "&#92;n&#92;ntest_queue"
-   *             }
-   *           }
-   *         }
-   *       }
-   * 3. For Python objects, implement to_proto() and from_proto(), and register
-   *    them in the following manner:
-   *    ops.register_proto_function("your_collection_name",
-   *                                proto_type,
-   *                                to_proto=YourPythonObject.to_proto,
-   *                                from_proto=YourPythonObject.from_proto)
-   *    These functions will be invoked to serialize and de-serialize the
-   *    collection. For example,
-   *    ops.register_proto_function(ops.GraphKeys.GLOBAL_VARIABLES,
-   *                                proto_type=variable_pb2.VariableDef,
-   *                                to_proto=Variable.to_proto,
-   *                                from_proto=Variable.from_proto)
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.CollectionDef}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.CollectionDef)
-      org.tensorflow.proto.framework.CollectionDefOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.MetaGraphProtos.internal_static_tensorflow_CollectionDef_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.MetaGraphProtos.internal_static_tensorflow_CollectionDef_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.CollectionDef.class, org.tensorflow.proto.framework.CollectionDef.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.framework.CollectionDef.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      kindCase_ = 0;
-      kind_ = null;
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.framework.MetaGraphProtos.internal_static_tensorflow_CollectionDef_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.CollectionDef getDefaultInstanceForType() {
-      return org.tensorflow.proto.framework.CollectionDef.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.CollectionDef build() {
-      org.tensorflow.proto.framework.CollectionDef result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.CollectionDef buildPartial() {
-      org.tensorflow.proto.framework.CollectionDef result = new org.tensorflow.proto.framework.CollectionDef(this);
-      if (kindCase_ == 1) {
-        if (nodeListBuilder_ == null) {
-          result.kind_ = kind_;
-        } else {
-          result.kind_ = nodeListBuilder_.build();
-        }
-      }
-      if (kindCase_ == 2) {
-        if (bytesListBuilder_ == null) {
-          result.kind_ = kind_;
-        } else {
-          result.kind_ = bytesListBuilder_.build();
-        }
-      }
-      if (kindCase_ == 3) {
-        if (int64ListBuilder_ == null) {
-          result.kind_ = kind_;
-        } else {
-          result.kind_ = int64ListBuilder_.build();
-        }
-      }
-      if (kindCase_ == 4) {
-        if (floatListBuilder_ == null) {
-          result.kind_ = kind_;
-        } else {
-          result.kind_ = floatListBuilder_.build();
-        }
-      }
-      if (kindCase_ == 5) {
-        if (anyListBuilder_ == null) {
-          result.kind_ = kind_;
-        } else {
-          result.kind_ = anyListBuilder_.build();
-        }
-      }
-      result.kindCase_ = kindCase_;
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.framework.CollectionDef) {
-        return mergeFrom((org.tensorflow.proto.framework.CollectionDef)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.framework.CollectionDef other) {
-      if (other == org.tensorflow.proto.framework.CollectionDef.getDefaultInstance()) return this;
-      switch (other.getKindCase()) {
-        case NODE_LIST: {
-          mergeNodeList(other.getNodeList());
-          break;
-        }
-        case BYTES_LIST: {
-          mergeBytesList(other.getBytesList());
-          break;
-        }
-        case INT64_LIST: {
-          mergeInt64List(other.getInt64List());
-          break;
-        }
-        case FLOAT_LIST: {
-          mergeFloatList(other.getFloatList());
-          break;
-        }
-        case ANY_LIST: {
-          mergeAnyList(other.getAnyList());
-          break;
-        }
-        case KIND_NOT_SET: {
-          break;
-        }
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.framework.CollectionDef parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.framework.CollectionDef) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-    private int kindCase_ = 0;
-    private java.lang.Object kind_;
-    public KindCase
-        getKindCase() {
-      return KindCase.forNumber(
-          kindCase_);
-    }
-
-    public Builder clearKind() {
-      kindCase_ = 0;
-      kind_ = null;
-      onChanged();
-      return this;
-    }
-
-
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.CollectionDef.NodeList, org.tensorflow.proto.framework.CollectionDef.NodeList.Builder, org.tensorflow.proto.framework.CollectionDef.NodeListOrBuilder> nodeListBuilder_;
-    /**
-     * <code>.tensorflow.CollectionDef.NodeList node_list = 1;</code>
-     */
-    public boolean hasNodeList() {
-      return kindCase_ == 1;
-    }
-    /**
-     * <code>.tensorflow.CollectionDef.NodeList node_list = 1;</code>
-     */
-    public org.tensorflow.proto.framework.CollectionDef.NodeList getNodeList() {
-      if (nodeListBuilder_ == null) {
-        if (kindCase_ == 1) {
-          return (org.tensorflow.proto.framework.CollectionDef.NodeList) kind_;
-        }
-        return org.tensorflow.proto.framework.CollectionDef.NodeList.getDefaultInstance();
-      } else {
-        if (kindCase_ == 1) {
-          return nodeListBuilder_.getMessage();
-        }
-        return org.tensorflow.proto.framework.CollectionDef.NodeList.getDefaultInstance();
-      }
-    }
-    /**
-     * <code>.tensorflow.CollectionDef.NodeList node_list = 1;</code>
-     */
-    public Builder setNodeList(org.tensorflow.proto.framework.CollectionDef.NodeList value) {
-      if (nodeListBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        kind_ = value;
-        onChanged();
-      } else {
-        nodeListBuilder_.setMessage(value);
-      }
-      kindCase_ = 1;
-      return this;
-    }
-    /**
-     * <code>.tensorflow.CollectionDef.NodeList node_list = 1;</code>
-     */
-    public Builder setNodeList(
-        org.tensorflow.proto.framework.CollectionDef.NodeList.Builder builderForValue) {
-      if (nodeListBuilder_ == null) {
-        kind_ = builderForValue.build();
-        onChanged();
-      } else {
-        nodeListBuilder_.setMessage(builderForValue.build());
-      }
-      kindCase_ = 1;
-      return this;
-    }
-    /**
-     * <code>.tensorflow.CollectionDef.NodeList node_list = 1;</code>
-     */
-    public Builder mergeNodeList(org.tensorflow.proto.framework.CollectionDef.NodeList value) {
-      if (nodeListBuilder_ == null) {
-        if (kindCase_ == 1 &&
-            kind_ != org.tensorflow.proto.framework.CollectionDef.NodeList.getDefaultInstance()) {
-          kind_ = org.tensorflow.proto.framework.CollectionDef.NodeList.newBuilder((org.tensorflow.proto.framework.CollectionDef.NodeList) kind_)
-              .mergeFrom(value).buildPartial();
-        } else {
-          kind_ = value;
-        }
-        onChanged();
-      } else {
-        if (kindCase_ == 1) {
-          nodeListBuilder_.mergeFrom(value);
-        }
-        nodeListBuilder_.setMessage(value);
-      }
-      kindCase_ = 1;
-      return this;
-    }
-    /**
-     * <code>.tensorflow.CollectionDef.NodeList node_list = 1;</code>
-     */
-    public Builder clearNodeList() {
-      if (nodeListBuilder_ == null) {
-        if (kindCase_ == 1) {
-          kindCase_ = 0;
-          kind_ = null;
-          onChanged();
-        }
-      } else {
-        if (kindCase_ == 1) {
-          kindCase_ = 0;
-          kind_ = null;
-        }
-        nodeListBuilder_.clear();
-      }
-      return this;
-    }
-    /**
-     * <code>.tensorflow.CollectionDef.NodeList node_list = 1;</code>
-     */
-    public org.tensorflow.proto.framework.CollectionDef.NodeList.Builder getNodeListBuilder() {
-      return getNodeListFieldBuilder().getBuilder();
-    }
-    /**
-     * <code>.tensorflow.CollectionDef.NodeList node_list = 1;</code>
-     */
-    public org.tensorflow.proto.framework.CollectionDef.NodeListOrBuilder getNodeListOrBuilder() {
-      if ((kindCase_ == 1) && (nodeListBuilder_ != null)) {
-        return nodeListBuilder_.getMessageOrBuilder();
-      } else {
-        if (kindCase_ == 1) {
-          return (org.tensorflow.proto.framework.CollectionDef.NodeList) kind_;
-        }
-        return org.tensorflow.proto.framework.CollectionDef.NodeList.getDefaultInstance();
-      }
-    }
-    /**
-     * <code>.tensorflow.CollectionDef.NodeList node_list = 1;</code>
-     */
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.CollectionDef.NodeList, org.tensorflow.proto.framework.CollectionDef.NodeList.Builder, org.tensorflow.proto.framework.CollectionDef.NodeListOrBuilder> 
-        getNodeListFieldBuilder() {
-      if (nodeListBuilder_ == null) {
-        if (!(kindCase_ == 1)) {
-          kind_ = org.tensorflow.proto.framework.CollectionDef.NodeList.getDefaultInstance();
-        }
-        nodeListBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-            org.tensorflow.proto.framework.CollectionDef.NodeList, org.tensorflow.proto.framework.CollectionDef.NodeList.Builder, org.tensorflow.proto.framework.CollectionDef.NodeListOrBuilder>(
-                (org.tensorflow.proto.framework.CollectionDef.NodeList) kind_,
-                getParentForChildren(),
-                isClean());
-        kind_ = null;
-      }
-      kindCase_ = 1;
-      onChanged();;
-      return nodeListBuilder_;
-    }
-
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.CollectionDef.BytesList, org.tensorflow.proto.framework.CollectionDef.BytesList.Builder, org.tensorflow.proto.framework.CollectionDef.BytesListOrBuilder> bytesListBuilder_;
-    /**
-     * <code>.tensorflow.CollectionDef.BytesList bytes_list = 2;</code>
-     */
-    public boolean hasBytesList() {
-      return kindCase_ == 2;
-    }
-    /**
-     * <code>.tensorflow.CollectionDef.BytesList bytes_list = 2;</code>
-     */
-    public org.tensorflow.proto.framework.CollectionDef.BytesList getBytesList() {
-      if (bytesListBuilder_ == null) {
-        if (kindCase_ == 2) {
-          return (org.tensorflow.proto.framework.CollectionDef.BytesList) kind_;
-        }
-        return org.tensorflow.proto.framework.CollectionDef.BytesList.getDefaultInstance();
-      } else {
-        if (kindCase_ == 2) {
-          return bytesListBuilder_.getMessage();
-        }
-        return org.tensorflow.proto.framework.CollectionDef.BytesList.getDefaultInstance();
-      }
-    }
-    /**
-     * <code>.tensorflow.CollectionDef.BytesList bytes_list = 2;</code>
-     */
-    public Builder setBytesList(org.tensorflow.proto.framework.CollectionDef.BytesList value) {
-      if (bytesListBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        kind_ = value;
-        onChanged();
-      } else {
-        bytesListBuilder_.setMessage(value);
-      }
-      kindCase_ = 2;
-      return this;
-    }
-    /**
-     * <code>.tensorflow.CollectionDef.BytesList bytes_list = 2;</code>
-     */
-    public Builder setBytesList(
-        org.tensorflow.proto.framework.CollectionDef.BytesList.Builder builderForValue) {
-      if (bytesListBuilder_ == null) {
-        kind_ = builderForValue.build();
-        onChanged();
-      } else {
-        bytesListBuilder_.setMessage(builderForValue.build());
-      }
-      kindCase_ = 2;
-      return this;
-    }
-    /**
-     * <code>.tensorflow.CollectionDef.BytesList bytes_list = 2;</code>
-     */
-    public Builder mergeBytesList(org.tensorflow.proto.framework.CollectionDef.BytesList value) {
-      if (bytesListBuilder_ == null) {
-        if (kindCase_ == 2 &&
-            kind_ != org.tensorflow.proto.framework.CollectionDef.BytesList.getDefaultInstance()) {
-          kind_ = org.tensorflow.proto.framework.CollectionDef.BytesList.newBuilder((org.tensorflow.proto.framework.CollectionDef.BytesList) kind_)
-              .mergeFrom(value).buildPartial();
-        } else {
-          kind_ = value;
-        }
-        onChanged();
-      } else {
-        if (kindCase_ == 2) {
-          bytesListBuilder_.mergeFrom(value);
-        }
-        bytesListBuilder_.setMessage(value);
-      }
-      kindCase_ = 2;
-      return this;
-    }
-    /**
-     * <code>.tensorflow.CollectionDef.BytesList bytes_list = 2;</code>
-     */
-    public Builder clearBytesList() {
-      if (bytesListBuilder_ == null) {
-        if (kindCase_ == 2) {
-          kindCase_ = 0;
-          kind_ = null;
-          onChanged();
-        }
-      } else {
-        if (kindCase_ == 2) {
-          kindCase_ = 0;
-          kind_ = null;
-        }
-        bytesListBuilder_.clear();
-      }
-      return this;
-    }
-    /**
-     * <code>.tensorflow.CollectionDef.BytesList bytes_list = 2;</code>
-     */
-    public org.tensorflow.proto.framework.CollectionDef.BytesList.Builder getBytesListBuilder() {
-      return getBytesListFieldBuilder().getBuilder();
-    }
-    /**
-     * <code>.tensorflow.CollectionDef.BytesList bytes_list = 2;</code>
-     */
-    public org.tensorflow.proto.framework.CollectionDef.BytesListOrBuilder getBytesListOrBuilder() {
-      if ((kindCase_ == 2) && (bytesListBuilder_ != null)) {
-        return bytesListBuilder_.getMessageOrBuilder();
-      } else {
-        if (kindCase_ == 2) {
-          return (org.tensorflow.proto.framework.CollectionDef.BytesList) kind_;
-        }
-        return org.tensorflow.proto.framework.CollectionDef.BytesList.getDefaultInstance();
-      }
-    }
-    /**
-     * <code>.tensorflow.CollectionDef.BytesList bytes_list = 2;</code>
-     */
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.CollectionDef.BytesList, org.tensorflow.proto.framework.CollectionDef.BytesList.Builder, org.tensorflow.proto.framework.CollectionDef.BytesListOrBuilder> 
-        getBytesListFieldBuilder() {
-      if (bytesListBuilder_ == null) {
-        if (!(kindCase_ == 2)) {
-          kind_ = org.tensorflow.proto.framework.CollectionDef.BytesList.getDefaultInstance();
-        }
-        bytesListBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-            org.tensorflow.proto.framework.CollectionDef.BytesList, org.tensorflow.proto.framework.CollectionDef.BytesList.Builder, org.tensorflow.proto.framework.CollectionDef.BytesListOrBuilder>(
-                (org.tensorflow.proto.framework.CollectionDef.BytesList) kind_,
-                getParentForChildren(),
-                isClean());
-        kind_ = null;
-      }
-      kindCase_ = 2;
-      onChanged();;
-      return bytesListBuilder_;
-    }
-
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.CollectionDef.Int64List, org.tensorflow.proto.framework.CollectionDef.Int64List.Builder, org.tensorflow.proto.framework.CollectionDef.Int64ListOrBuilder> int64ListBuilder_;
-    /**
-     * <code>.tensorflow.CollectionDef.Int64List int64_list = 3;</code>
-     */
-    public boolean hasInt64List() {
-      return kindCase_ == 3;
-    }
-    /**
-     * <code>.tensorflow.CollectionDef.Int64List int64_list = 3;</code>
-     */
-    public org.tensorflow.proto.framework.CollectionDef.Int64List getInt64List() {
-      if (int64ListBuilder_ == null) {
-        if (kindCase_ == 3) {
-          return (org.tensorflow.proto.framework.CollectionDef.Int64List) kind_;
-        }
-        return org.tensorflow.proto.framework.CollectionDef.Int64List.getDefaultInstance();
-      } else {
-        if (kindCase_ == 3) {
-          return int64ListBuilder_.getMessage();
-        }
-        return org.tensorflow.proto.framework.CollectionDef.Int64List.getDefaultInstance();
-      }
-    }
-    /**
-     * <code>.tensorflow.CollectionDef.Int64List int64_list = 3;</code>
-     */
-    public Builder setInt64List(org.tensorflow.proto.framework.CollectionDef.Int64List value) {
-      if (int64ListBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        kind_ = value;
-        onChanged();
-      } else {
-        int64ListBuilder_.setMessage(value);
-      }
-      kindCase_ = 3;
-      return this;
-    }
-    /**
-     * <code>.tensorflow.CollectionDef.Int64List int64_list = 3;</code>
-     */
-    public Builder setInt64List(
-        org.tensorflow.proto.framework.CollectionDef.Int64List.Builder builderForValue) {
-      if (int64ListBuilder_ == null) {
-        kind_ = builderForValue.build();
-        onChanged();
-      } else {
-        int64ListBuilder_.setMessage(builderForValue.build());
-      }
-      kindCase_ = 3;
-      return this;
-    }
-    /**
-     * <code>.tensorflow.CollectionDef.Int64List int64_list = 3;</code>
-     */
-    public Builder mergeInt64List(org.tensorflow.proto.framework.CollectionDef.Int64List value) {
-      if (int64ListBuilder_ == null) {
-        if (kindCase_ == 3 &&
-            kind_ != org.tensorflow.proto.framework.CollectionDef.Int64List.getDefaultInstance()) {
-          kind_ = org.tensorflow.proto.framework.CollectionDef.Int64List.newBuilder((org.tensorflow.proto.framework.CollectionDef.Int64List) kind_)
-              .mergeFrom(value).buildPartial();
-        } else {
-          kind_ = value;
-        }
-        onChanged();
-      } else {
-        if (kindCase_ == 3) {
-          int64ListBuilder_.mergeFrom(value);
-        }
-        int64ListBuilder_.setMessage(value);
-      }
-      kindCase_ = 3;
-      return this;
-    }
-    /**
-     * <code>.tensorflow.CollectionDef.Int64List int64_list = 3;</code>
-     */
-    public Builder clearInt64List() {
-      if (int64ListBuilder_ == null) {
-        if (kindCase_ == 3) {
-          kindCase_ = 0;
-          kind_ = null;
-          onChanged();
-        }
-      } else {
-        if (kindCase_ == 3) {
-          kindCase_ = 0;
-          kind_ = null;
-        }
-        int64ListBuilder_.clear();
-      }
-      return this;
-    }
-    /**
-     * <code>.tensorflow.CollectionDef.Int64List int64_list = 3;</code>
-     */
-    public org.tensorflow.proto.framework.CollectionDef.Int64List.Builder getInt64ListBuilder() {
-      return getInt64ListFieldBuilder().getBuilder();
-    }
-    /**
-     * <code>.tensorflow.CollectionDef.Int64List int64_list = 3;</code>
-     */
-    public org.tensorflow.proto.framework.CollectionDef.Int64ListOrBuilder getInt64ListOrBuilder() {
-      if ((kindCase_ == 3) && (int64ListBuilder_ != null)) {
-        return int64ListBuilder_.getMessageOrBuilder();
-      } else {
-        if (kindCase_ == 3) {
-          return (org.tensorflow.proto.framework.CollectionDef.Int64List) kind_;
-        }
-        return org.tensorflow.proto.framework.CollectionDef.Int64List.getDefaultInstance();
-      }
-    }
-    /**
-     * <code>.tensorflow.CollectionDef.Int64List int64_list = 3;</code>
-     */
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.CollectionDef.Int64List, org.tensorflow.proto.framework.CollectionDef.Int64List.Builder, org.tensorflow.proto.framework.CollectionDef.Int64ListOrBuilder> 
-        getInt64ListFieldBuilder() {
-      if (int64ListBuilder_ == null) {
-        if (!(kindCase_ == 3)) {
-          kind_ = org.tensorflow.proto.framework.CollectionDef.Int64List.getDefaultInstance();
-        }
-        int64ListBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-            org.tensorflow.proto.framework.CollectionDef.Int64List, org.tensorflow.proto.framework.CollectionDef.Int64List.Builder, org.tensorflow.proto.framework.CollectionDef.Int64ListOrBuilder>(
-                (org.tensorflow.proto.framework.CollectionDef.Int64List) kind_,
-                getParentForChildren(),
-                isClean());
-        kind_ = null;
-      }
-      kindCase_ = 3;
-      onChanged();;
-      return int64ListBuilder_;
-    }
-
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.CollectionDef.FloatList, org.tensorflow.proto.framework.CollectionDef.FloatList.Builder, org.tensorflow.proto.framework.CollectionDef.FloatListOrBuilder> floatListBuilder_;
-    /**
-     * <code>.tensorflow.CollectionDef.FloatList float_list = 4;</code>
-     */
-    public boolean hasFloatList() {
-      return kindCase_ == 4;
-    }
-    /**
-     * <code>.tensorflow.CollectionDef.FloatList float_list = 4;</code>
-     */
-    public org.tensorflow.proto.framework.CollectionDef.FloatList getFloatList() {
-      if (floatListBuilder_ == null) {
-        if (kindCase_ == 4) {
-          return (org.tensorflow.proto.framework.CollectionDef.FloatList) kind_;
-        }
-        return org.tensorflow.proto.framework.CollectionDef.FloatList.getDefaultInstance();
-      } else {
-        if (kindCase_ == 4) {
-          return floatListBuilder_.getMessage();
-        }
-        return org.tensorflow.proto.framework.CollectionDef.FloatList.getDefaultInstance();
-      }
-    }
-    /**
-     * <code>.tensorflow.CollectionDef.FloatList float_list = 4;</code>
-     */
-    public Builder setFloatList(org.tensorflow.proto.framework.CollectionDef.FloatList value) {
-      if (floatListBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        kind_ = value;
-        onChanged();
-      } else {
-        floatListBuilder_.setMessage(value);
-      }
-      kindCase_ = 4;
-      return this;
-    }
-    /**
-     * <code>.tensorflow.CollectionDef.FloatList float_list = 4;</code>
-     */
-    public Builder setFloatList(
-        org.tensorflow.proto.framework.CollectionDef.FloatList.Builder builderForValue) {
-      if (floatListBuilder_ == null) {
-        kind_ = builderForValue.build();
-        onChanged();
-      } else {
-        floatListBuilder_.setMessage(builderForValue.build());
-      }
-      kindCase_ = 4;
-      return this;
-    }
-    /**
-     * <code>.tensorflow.CollectionDef.FloatList float_list = 4;</code>
-     */
-    public Builder mergeFloatList(org.tensorflow.proto.framework.CollectionDef.FloatList value) {
-      if (floatListBuilder_ == null) {
-        if (kindCase_ == 4 &&
-            kind_ != org.tensorflow.proto.framework.CollectionDef.FloatList.getDefaultInstance()) {
-          kind_ = org.tensorflow.proto.framework.CollectionDef.FloatList.newBuilder((org.tensorflow.proto.framework.CollectionDef.FloatList) kind_)
-              .mergeFrom(value).buildPartial();
-        } else {
-          kind_ = value;
-        }
-        onChanged();
-      } else {
-        if (kindCase_ == 4) {
-          floatListBuilder_.mergeFrom(value);
-        }
-        floatListBuilder_.setMessage(value);
-      }
-      kindCase_ = 4;
-      return this;
-    }
-    /**
-     * <code>.tensorflow.CollectionDef.FloatList float_list = 4;</code>
-     */
-    public Builder clearFloatList() {
-      if (floatListBuilder_ == null) {
-        if (kindCase_ == 4) {
-          kindCase_ = 0;
-          kind_ = null;
-          onChanged();
-        }
-      } else {
-        if (kindCase_ == 4) {
-          kindCase_ = 0;
-          kind_ = null;
-        }
-        floatListBuilder_.clear();
-      }
-      return this;
-    }
-    /**
-     * <code>.tensorflow.CollectionDef.FloatList float_list = 4;</code>
-     */
-    public org.tensorflow.proto.framework.CollectionDef.FloatList.Builder getFloatListBuilder() {
-      return getFloatListFieldBuilder().getBuilder();
-    }
-    /**
-     * <code>.tensorflow.CollectionDef.FloatList float_list = 4;</code>
-     */
-    public org.tensorflow.proto.framework.CollectionDef.FloatListOrBuilder getFloatListOrBuilder() {
-      if ((kindCase_ == 4) && (floatListBuilder_ != null)) {
-        return floatListBuilder_.getMessageOrBuilder();
-      } else {
-        if (kindCase_ == 4) {
-          return (org.tensorflow.proto.framework.CollectionDef.FloatList) kind_;
-        }
-        return org.tensorflow.proto.framework.CollectionDef.FloatList.getDefaultInstance();
-      }
-    }
-    /**
-     * <code>.tensorflow.CollectionDef.FloatList float_list = 4;</code>
-     */
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.CollectionDef.FloatList, org.tensorflow.proto.framework.CollectionDef.FloatList.Builder, org.tensorflow.proto.framework.CollectionDef.FloatListOrBuilder> 
-        getFloatListFieldBuilder() {
-      if (floatListBuilder_ == null) {
-        if (!(kindCase_ == 4)) {
-          kind_ = org.tensorflow.proto.framework.CollectionDef.FloatList.getDefaultInstance();
-        }
-        floatListBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-            org.tensorflow.proto.framework.CollectionDef.FloatList, org.tensorflow.proto.framework.CollectionDef.FloatList.Builder, org.tensorflow.proto.framework.CollectionDef.FloatListOrBuilder>(
-                (org.tensorflow.proto.framework.CollectionDef.FloatList) kind_,
-                getParentForChildren(),
-                isClean());
-        kind_ = null;
-      }
-      kindCase_ = 4;
-      onChanged();;
-      return floatListBuilder_;
-    }
-
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.CollectionDef.AnyList, org.tensorflow.proto.framework.CollectionDef.AnyList.Builder, org.tensorflow.proto.framework.CollectionDef.AnyListOrBuilder> anyListBuilder_;
-    /**
-     * <code>.tensorflow.CollectionDef.AnyList any_list = 5;</code>
-     */
-    public boolean hasAnyList() {
-      return kindCase_ == 5;
-    }
-    /**
-     * <code>.tensorflow.CollectionDef.AnyList any_list = 5;</code>
-     */
-    public org.tensorflow.proto.framework.CollectionDef.AnyList getAnyList() {
-      if (anyListBuilder_ == null) {
-        if (kindCase_ == 5) {
-          return (org.tensorflow.proto.framework.CollectionDef.AnyList) kind_;
-        }
-        return org.tensorflow.proto.framework.CollectionDef.AnyList.getDefaultInstance();
-      } else {
-        if (kindCase_ == 5) {
-          return anyListBuilder_.getMessage();
-        }
-        return org.tensorflow.proto.framework.CollectionDef.AnyList.getDefaultInstance();
-      }
-    }
-    /**
-     * <code>.tensorflow.CollectionDef.AnyList any_list = 5;</code>
-     */
-    public Builder setAnyList(org.tensorflow.proto.framework.CollectionDef.AnyList value) {
-      if (anyListBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        kind_ = value;
-        onChanged();
-      } else {
-        anyListBuilder_.setMessage(value);
-      }
-      kindCase_ = 5;
-      return this;
-    }
-    /**
-     * <code>.tensorflow.CollectionDef.AnyList any_list = 5;</code>
-     */
-    public Builder setAnyList(
-        org.tensorflow.proto.framework.CollectionDef.AnyList.Builder builderForValue) {
-      if (anyListBuilder_ == null) {
-        kind_ = builderForValue.build();
-        onChanged();
-      } else {
-        anyListBuilder_.setMessage(builderForValue.build());
-      }
-      kindCase_ = 5;
-      return this;
-    }
-    /**
-     * <code>.tensorflow.CollectionDef.AnyList any_list = 5;</code>
-     */
-    public Builder mergeAnyList(org.tensorflow.proto.framework.CollectionDef.AnyList value) {
-      if (anyListBuilder_ == null) {
-        if (kindCase_ == 5 &&
-            kind_ != org.tensorflow.proto.framework.CollectionDef.AnyList.getDefaultInstance()) {
-          kind_ = org.tensorflow.proto.framework.CollectionDef.AnyList.newBuilder((org.tensorflow.proto.framework.CollectionDef.AnyList) kind_)
-              .mergeFrom(value).buildPartial();
-        } else {
-          kind_ = value;
-        }
-        onChanged();
-      } else {
-        if (kindCase_ == 5) {
-          anyListBuilder_.mergeFrom(value);
-        }
-        anyListBuilder_.setMessage(value);
-      }
-      kindCase_ = 5;
-      return this;
-    }
-    /**
-     * <code>.tensorflow.CollectionDef.AnyList any_list = 5;</code>
-     */
-    public Builder clearAnyList() {
-      if (anyListBuilder_ == null) {
-        if (kindCase_ == 5) {
-          kindCase_ = 0;
-          kind_ = null;
-          onChanged();
-        }
-      } else {
-        if (kindCase_ == 5) {
-          kindCase_ = 0;
-          kind_ = null;
-        }
-        anyListBuilder_.clear();
-      }
-      return this;
-    }
-    /**
-     * <code>.tensorflow.CollectionDef.AnyList any_list = 5;</code>
-     */
-    public org.tensorflow.proto.framework.CollectionDef.AnyList.Builder getAnyListBuilder() {
-      return getAnyListFieldBuilder().getBuilder();
-    }
-    /**
-     * <code>.tensorflow.CollectionDef.AnyList any_list = 5;</code>
-     */
-    public org.tensorflow.proto.framework.CollectionDef.AnyListOrBuilder getAnyListOrBuilder() {
-      if ((kindCase_ == 5) && (anyListBuilder_ != null)) {
-        return anyListBuilder_.getMessageOrBuilder();
-      } else {
-        if (kindCase_ == 5) {
-          return (org.tensorflow.proto.framework.CollectionDef.AnyList) kind_;
-        }
-        return org.tensorflow.proto.framework.CollectionDef.AnyList.getDefaultInstance();
-      }
-    }
-    /**
-     * <code>.tensorflow.CollectionDef.AnyList any_list = 5;</code>
-     */
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.CollectionDef.AnyList, org.tensorflow.proto.framework.CollectionDef.AnyList.Builder, org.tensorflow.proto.framework.CollectionDef.AnyListOrBuilder> 
-        getAnyListFieldBuilder() {
-      if (anyListBuilder_ == null) {
-        if (!(kindCase_ == 5)) {
-          kind_ = org.tensorflow.proto.framework.CollectionDef.AnyList.getDefaultInstance();
-        }
-        anyListBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-            org.tensorflow.proto.framework.CollectionDef.AnyList, org.tensorflow.proto.framework.CollectionDef.AnyList.Builder, org.tensorflow.proto.framework.CollectionDef.AnyListOrBuilder>(
-                (org.tensorflow.proto.framework.CollectionDef.AnyList) kind_,
-                getParentForChildren(),
-                isClean());
-        kind_ = null;
-      }
-      kindCase_ = 5;
-      onChanged();;
-      return anyListBuilder_;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.CollectionDef)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.CollectionDef)
-  private static final org.tensorflow.proto.framework.CollectionDef DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.CollectionDef();
-  }
-
-  public static org.tensorflow.proto.framework.CollectionDef getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<CollectionDef>
-      PARSER = new com.google.protobuf.AbstractParser<CollectionDef>() {
-    @java.lang.Override
-    public CollectionDef parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new CollectionDef(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<CollectionDef> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<CollectionDef> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.framework.CollectionDef getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/CollectionDefOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/CollectionDefOrBuilder.java
deleted file mode 100644
index f3d5fc7ce61..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/CollectionDefOrBuilder.java
+++ /dev/null
@@ -1,76 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/meta_graph.proto
-
-package org.tensorflow.proto.framework;
-
-public interface CollectionDefOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.CollectionDef)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <code>.tensorflow.CollectionDef.NodeList node_list = 1;</code>
-   */
-  boolean hasNodeList();
-  /**
-   * <code>.tensorflow.CollectionDef.NodeList node_list = 1;</code>
-   */
-  org.tensorflow.proto.framework.CollectionDef.NodeList getNodeList();
-  /**
-   * <code>.tensorflow.CollectionDef.NodeList node_list = 1;</code>
-   */
-  org.tensorflow.proto.framework.CollectionDef.NodeListOrBuilder getNodeListOrBuilder();
-
-  /**
-   * <code>.tensorflow.CollectionDef.BytesList bytes_list = 2;</code>
-   */
-  boolean hasBytesList();
-  /**
-   * <code>.tensorflow.CollectionDef.BytesList bytes_list = 2;</code>
-   */
-  org.tensorflow.proto.framework.CollectionDef.BytesList getBytesList();
-  /**
-   * <code>.tensorflow.CollectionDef.BytesList bytes_list = 2;</code>
-   */
-  org.tensorflow.proto.framework.CollectionDef.BytesListOrBuilder getBytesListOrBuilder();
-
-  /**
-   * <code>.tensorflow.CollectionDef.Int64List int64_list = 3;</code>
-   */
-  boolean hasInt64List();
-  /**
-   * <code>.tensorflow.CollectionDef.Int64List int64_list = 3;</code>
-   */
-  org.tensorflow.proto.framework.CollectionDef.Int64List getInt64List();
-  /**
-   * <code>.tensorflow.CollectionDef.Int64List int64_list = 3;</code>
-   */
-  org.tensorflow.proto.framework.CollectionDef.Int64ListOrBuilder getInt64ListOrBuilder();
-
-  /**
-   * <code>.tensorflow.CollectionDef.FloatList float_list = 4;</code>
-   */
-  boolean hasFloatList();
-  /**
-   * <code>.tensorflow.CollectionDef.FloatList float_list = 4;</code>
-   */
-  org.tensorflow.proto.framework.CollectionDef.FloatList getFloatList();
-  /**
-   * <code>.tensorflow.CollectionDef.FloatList float_list = 4;</code>
-   */
-  org.tensorflow.proto.framework.CollectionDef.FloatListOrBuilder getFloatListOrBuilder();
-
-  /**
-   * <code>.tensorflow.CollectionDef.AnyList any_list = 5;</code>
-   */
-  boolean hasAnyList();
-  /**
-   * <code>.tensorflow.CollectionDef.AnyList any_list = 5;</code>
-   */
-  org.tensorflow.proto.framework.CollectionDef.AnyList getAnyList();
-  /**
-   * <code>.tensorflow.CollectionDef.AnyList any_list = 5;</code>
-   */
-  org.tensorflow.proto.framework.CollectionDef.AnyListOrBuilder getAnyListOrBuilder();
-
-  public org.tensorflow.proto.framework.CollectionDef.KindCase getKindCase();
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/CondContextDef.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/CondContextDef.java
deleted file mode 100644
index 3ab6d340e10..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/CondContextDef.java
+++ /dev/null
@@ -1,1632 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/control_flow.proto
-
-package org.tensorflow.proto.framework;
-
-/**
- * <pre>
- * Protocol buffer representing a CondContext object.
- * </pre>
- *
- * Protobuf type {@code tensorflow.CondContextDef}
- */
-public  final class CondContextDef extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.CondContextDef)
-    CondContextDefOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use CondContextDef.newBuilder() to construct.
-  private CondContextDef(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private CondContextDef() {
-    contextName_ = "";
-    predName_ = "";
-    pivotName_ = "";
-    nestedContexts_ = java.util.Collections.emptyList();
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new CondContextDef();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private CondContextDef(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    int mutable_bitField0_ = 0;
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 10: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            contextName_ = s;
-            break;
-          }
-          case 18: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            predName_ = s;
-            break;
-          }
-          case 26: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            pivotName_ = s;
-            break;
-          }
-          case 32: {
-
-            branch_ = input.readInt32();
-            break;
-          }
-          case 42: {
-            org.tensorflow.proto.framework.ValuesDef.Builder subBuilder = null;
-            if (valuesDef_ != null) {
-              subBuilder = valuesDef_.toBuilder();
-            }
-            valuesDef_ = input.readMessage(org.tensorflow.proto.framework.ValuesDef.parser(), extensionRegistry);
-            if (subBuilder != null) {
-              subBuilder.mergeFrom(valuesDef_);
-              valuesDef_ = subBuilder.buildPartial();
-            }
-
-            break;
-          }
-          case 50: {
-            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
-              nestedContexts_ = new java.util.ArrayList<org.tensorflow.proto.framework.ControlFlowContextDef>();
-              mutable_bitField0_ |= 0x00000001;
-            }
-            nestedContexts_.add(
-                input.readMessage(org.tensorflow.proto.framework.ControlFlowContextDef.parser(), extensionRegistry));
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      if (((mutable_bitField0_ & 0x00000001) != 0)) {
-        nestedContexts_ = java.util.Collections.unmodifiableList(nestedContexts_);
-      }
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.framework.ControlFlowProtos.internal_static_tensorflow_CondContextDef_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.framework.ControlFlowProtos.internal_static_tensorflow_CondContextDef_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.framework.CondContextDef.class, org.tensorflow.proto.framework.CondContextDef.Builder.class);
-  }
-
-  public static final int CONTEXT_NAME_FIELD_NUMBER = 1;
-  private volatile java.lang.Object contextName_;
-  /**
-   * <pre>
-   * Name of the context.
-   * </pre>
-   *
-   * <code>string context_name = 1;</code>
-   */
-  public java.lang.String getContextName() {
-    java.lang.Object ref = contextName_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      contextName_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * Name of the context.
-   * </pre>
-   *
-   * <code>string context_name = 1;</code>
-   */
-  public com.google.protobuf.ByteString
-      getContextNameBytes() {
-    java.lang.Object ref = contextName_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      contextName_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int PRED_NAME_FIELD_NUMBER = 2;
-  private volatile java.lang.Object predName_;
-  /**
-   * <pre>
-   * Name of the pred tensor.
-   * </pre>
-   *
-   * <code>string pred_name = 2;</code>
-   */
-  public java.lang.String getPredName() {
-    java.lang.Object ref = predName_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      predName_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * Name of the pred tensor.
-   * </pre>
-   *
-   * <code>string pred_name = 2;</code>
-   */
-  public com.google.protobuf.ByteString
-      getPredNameBytes() {
-    java.lang.Object ref = predName_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      predName_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int PIVOT_NAME_FIELD_NUMBER = 3;
-  private volatile java.lang.Object pivotName_;
-  /**
-   * <pre>
-   * Name of the pivot tensor.
-   * </pre>
-   *
-   * <code>string pivot_name = 3;</code>
-   */
-  public java.lang.String getPivotName() {
-    java.lang.Object ref = pivotName_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      pivotName_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * Name of the pivot tensor.
-   * </pre>
-   *
-   * <code>string pivot_name = 3;</code>
-   */
-  public com.google.protobuf.ByteString
-      getPivotNameBytes() {
-    java.lang.Object ref = pivotName_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      pivotName_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int BRANCH_FIELD_NUMBER = 4;
-  private int branch_;
-  /**
-   * <pre>
-   * Branch prediction. 0 or 1.
-   * </pre>
-   *
-   * <code>int32 branch = 4;</code>
-   */
-  public int getBranch() {
-    return branch_;
-  }
-
-  public static final int VALUES_DEF_FIELD_NUMBER = 5;
-  private org.tensorflow.proto.framework.ValuesDef valuesDef_;
-  /**
-   * <pre>
-   * Values and external values in control flow context.
-   * </pre>
-   *
-   * <code>.tensorflow.ValuesDef values_def = 5;</code>
-   */
-  public boolean hasValuesDef() {
-    return valuesDef_ != null;
-  }
-  /**
-   * <pre>
-   * Values and external values in control flow context.
-   * </pre>
-   *
-   * <code>.tensorflow.ValuesDef values_def = 5;</code>
-   */
-  public org.tensorflow.proto.framework.ValuesDef getValuesDef() {
-    return valuesDef_ == null ? org.tensorflow.proto.framework.ValuesDef.getDefaultInstance() : valuesDef_;
-  }
-  /**
-   * <pre>
-   * Values and external values in control flow context.
-   * </pre>
-   *
-   * <code>.tensorflow.ValuesDef values_def = 5;</code>
-   */
-  public org.tensorflow.proto.framework.ValuesDefOrBuilder getValuesDefOrBuilder() {
-    return getValuesDef();
-  }
-
-  public static final int NESTED_CONTEXTS_FIELD_NUMBER = 6;
-  private java.util.List<org.tensorflow.proto.framework.ControlFlowContextDef> nestedContexts_;
-  /**
-   * <pre>
-   * Contexts contained inside this context (e.g. nested conds).
-   * </pre>
-   *
-   * <code>repeated .tensorflow.ControlFlowContextDef nested_contexts = 6;</code>
-   */
-  public java.util.List<org.tensorflow.proto.framework.ControlFlowContextDef> getNestedContextsList() {
-    return nestedContexts_;
-  }
-  /**
-   * <pre>
-   * Contexts contained inside this context (e.g. nested conds).
-   * </pre>
-   *
-   * <code>repeated .tensorflow.ControlFlowContextDef nested_contexts = 6;</code>
-   */
-  public java.util.List<? extends org.tensorflow.proto.framework.ControlFlowContextDefOrBuilder> 
-      getNestedContextsOrBuilderList() {
-    return nestedContexts_;
-  }
-  /**
-   * <pre>
-   * Contexts contained inside this context (e.g. nested conds).
-   * </pre>
-   *
-   * <code>repeated .tensorflow.ControlFlowContextDef nested_contexts = 6;</code>
-   */
-  public int getNestedContextsCount() {
-    return nestedContexts_.size();
-  }
-  /**
-   * <pre>
-   * Contexts contained inside this context (e.g. nested conds).
-   * </pre>
-   *
-   * <code>repeated .tensorflow.ControlFlowContextDef nested_contexts = 6;</code>
-   */
-  public org.tensorflow.proto.framework.ControlFlowContextDef getNestedContexts(int index) {
-    return nestedContexts_.get(index);
-  }
-  /**
-   * <pre>
-   * Contexts contained inside this context (e.g. nested conds).
-   * </pre>
-   *
-   * <code>repeated .tensorflow.ControlFlowContextDef nested_contexts = 6;</code>
-   */
-  public org.tensorflow.proto.framework.ControlFlowContextDefOrBuilder getNestedContextsOrBuilder(
-      int index) {
-    return nestedContexts_.get(index);
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    if (!getContextNameBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 1, contextName_);
-    }
-    if (!getPredNameBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 2, predName_);
-    }
-    if (!getPivotNameBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 3, pivotName_);
-    }
-    if (branch_ != 0) {
-      output.writeInt32(4, branch_);
-    }
-    if (valuesDef_ != null) {
-      output.writeMessage(5, getValuesDef());
-    }
-    for (int i = 0; i < nestedContexts_.size(); i++) {
-      output.writeMessage(6, nestedContexts_.get(i));
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (!getContextNameBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, contextName_);
-    }
-    if (!getPredNameBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(2, predName_);
-    }
-    if (!getPivotNameBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(3, pivotName_);
-    }
-    if (branch_ != 0) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt32Size(4, branch_);
-    }
-    if (valuesDef_ != null) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(5, getValuesDef());
-    }
-    for (int i = 0; i < nestedContexts_.size(); i++) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(6, nestedContexts_.get(i));
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.framework.CondContextDef)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.framework.CondContextDef other = (org.tensorflow.proto.framework.CondContextDef) obj;
-
-    if (!getContextName()
-        .equals(other.getContextName())) return false;
-    if (!getPredName()
-        .equals(other.getPredName())) return false;
-    if (!getPivotName()
-        .equals(other.getPivotName())) return false;
-    if (getBranch()
-        != other.getBranch()) return false;
-    if (hasValuesDef() != other.hasValuesDef()) return false;
-    if (hasValuesDef()) {
-      if (!getValuesDef()
-          .equals(other.getValuesDef())) return false;
-    }
-    if (!getNestedContextsList()
-        .equals(other.getNestedContextsList())) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    hash = (37 * hash) + CONTEXT_NAME_FIELD_NUMBER;
-    hash = (53 * hash) + getContextName().hashCode();
-    hash = (37 * hash) + PRED_NAME_FIELD_NUMBER;
-    hash = (53 * hash) + getPredName().hashCode();
-    hash = (37 * hash) + PIVOT_NAME_FIELD_NUMBER;
-    hash = (53 * hash) + getPivotName().hashCode();
-    hash = (37 * hash) + BRANCH_FIELD_NUMBER;
-    hash = (53 * hash) + getBranch();
-    if (hasValuesDef()) {
-      hash = (37 * hash) + VALUES_DEF_FIELD_NUMBER;
-      hash = (53 * hash) + getValuesDef().hashCode();
-    }
-    if (getNestedContextsCount() > 0) {
-      hash = (37 * hash) + NESTED_CONTEXTS_FIELD_NUMBER;
-      hash = (53 * hash) + getNestedContextsList().hashCode();
-    }
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.framework.CondContextDef parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.CondContextDef parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.CondContextDef parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.CondContextDef parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.CondContextDef parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.CondContextDef parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.CondContextDef parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.CondContextDef parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.CondContextDef parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.CondContextDef parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.CondContextDef parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.CondContextDef parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.framework.CondContextDef prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * <pre>
-   * Protocol buffer representing a CondContext object.
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.CondContextDef}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.CondContextDef)
-      org.tensorflow.proto.framework.CondContextDefOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.ControlFlowProtos.internal_static_tensorflow_CondContextDef_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.ControlFlowProtos.internal_static_tensorflow_CondContextDef_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.CondContextDef.class, org.tensorflow.proto.framework.CondContextDef.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.framework.CondContextDef.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-        getNestedContextsFieldBuilder();
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      contextName_ = "";
-
-      predName_ = "";
-
-      pivotName_ = "";
-
-      branch_ = 0;
-
-      if (valuesDefBuilder_ == null) {
-        valuesDef_ = null;
-      } else {
-        valuesDef_ = null;
-        valuesDefBuilder_ = null;
-      }
-      if (nestedContextsBuilder_ == null) {
-        nestedContexts_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000001);
-      } else {
-        nestedContextsBuilder_.clear();
-      }
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.framework.ControlFlowProtos.internal_static_tensorflow_CondContextDef_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.CondContextDef getDefaultInstanceForType() {
-      return org.tensorflow.proto.framework.CondContextDef.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.CondContextDef build() {
-      org.tensorflow.proto.framework.CondContextDef result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.CondContextDef buildPartial() {
-      org.tensorflow.proto.framework.CondContextDef result = new org.tensorflow.proto.framework.CondContextDef(this);
-      int from_bitField0_ = bitField0_;
-      result.contextName_ = contextName_;
-      result.predName_ = predName_;
-      result.pivotName_ = pivotName_;
-      result.branch_ = branch_;
-      if (valuesDefBuilder_ == null) {
-        result.valuesDef_ = valuesDef_;
-      } else {
-        result.valuesDef_ = valuesDefBuilder_.build();
-      }
-      if (nestedContextsBuilder_ == null) {
-        if (((bitField0_ & 0x00000001) != 0)) {
-          nestedContexts_ = java.util.Collections.unmodifiableList(nestedContexts_);
-          bitField0_ = (bitField0_ & ~0x00000001);
-        }
-        result.nestedContexts_ = nestedContexts_;
-      } else {
-        result.nestedContexts_ = nestedContextsBuilder_.build();
-      }
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.framework.CondContextDef) {
-        return mergeFrom((org.tensorflow.proto.framework.CondContextDef)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.framework.CondContextDef other) {
-      if (other == org.tensorflow.proto.framework.CondContextDef.getDefaultInstance()) return this;
-      if (!other.getContextName().isEmpty()) {
-        contextName_ = other.contextName_;
-        onChanged();
-      }
-      if (!other.getPredName().isEmpty()) {
-        predName_ = other.predName_;
-        onChanged();
-      }
-      if (!other.getPivotName().isEmpty()) {
-        pivotName_ = other.pivotName_;
-        onChanged();
-      }
-      if (other.getBranch() != 0) {
-        setBranch(other.getBranch());
-      }
-      if (other.hasValuesDef()) {
-        mergeValuesDef(other.getValuesDef());
-      }
-      if (nestedContextsBuilder_ == null) {
-        if (!other.nestedContexts_.isEmpty()) {
-          if (nestedContexts_.isEmpty()) {
-            nestedContexts_ = other.nestedContexts_;
-            bitField0_ = (bitField0_ & ~0x00000001);
-          } else {
-            ensureNestedContextsIsMutable();
-            nestedContexts_.addAll(other.nestedContexts_);
-          }
-          onChanged();
-        }
-      } else {
-        if (!other.nestedContexts_.isEmpty()) {
-          if (nestedContextsBuilder_.isEmpty()) {
-            nestedContextsBuilder_.dispose();
-            nestedContextsBuilder_ = null;
-            nestedContexts_ = other.nestedContexts_;
-            bitField0_ = (bitField0_ & ~0x00000001);
-            nestedContextsBuilder_ = 
-              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
-                 getNestedContextsFieldBuilder() : null;
-          } else {
-            nestedContextsBuilder_.addAllMessages(other.nestedContexts_);
-          }
-        }
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.framework.CondContextDef parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.framework.CondContextDef) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-    private int bitField0_;
-
-    private java.lang.Object contextName_ = "";
-    /**
-     * <pre>
-     * Name of the context.
-     * </pre>
-     *
-     * <code>string context_name = 1;</code>
-     */
-    public java.lang.String getContextName() {
-      java.lang.Object ref = contextName_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        contextName_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Name of the context.
-     * </pre>
-     *
-     * <code>string context_name = 1;</code>
-     */
-    public com.google.protobuf.ByteString
-        getContextNameBytes() {
-      java.lang.Object ref = contextName_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        contextName_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Name of the context.
-     * </pre>
-     *
-     * <code>string context_name = 1;</code>
-     */
-    public Builder setContextName(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      contextName_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Name of the context.
-     * </pre>
-     *
-     * <code>string context_name = 1;</code>
-     */
-    public Builder clearContextName() {
-      
-      contextName_ = getDefaultInstance().getContextName();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Name of the context.
-     * </pre>
-     *
-     * <code>string context_name = 1;</code>
-     */
-    public Builder setContextNameBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      contextName_ = value;
-      onChanged();
-      return this;
-    }
-
-    private java.lang.Object predName_ = "";
-    /**
-     * <pre>
-     * Name of the pred tensor.
-     * </pre>
-     *
-     * <code>string pred_name = 2;</code>
-     */
-    public java.lang.String getPredName() {
-      java.lang.Object ref = predName_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        predName_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Name of the pred tensor.
-     * </pre>
-     *
-     * <code>string pred_name = 2;</code>
-     */
-    public com.google.protobuf.ByteString
-        getPredNameBytes() {
-      java.lang.Object ref = predName_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        predName_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Name of the pred tensor.
-     * </pre>
-     *
-     * <code>string pred_name = 2;</code>
-     */
-    public Builder setPredName(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      predName_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Name of the pred tensor.
-     * </pre>
-     *
-     * <code>string pred_name = 2;</code>
-     */
-    public Builder clearPredName() {
-      
-      predName_ = getDefaultInstance().getPredName();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Name of the pred tensor.
-     * </pre>
-     *
-     * <code>string pred_name = 2;</code>
-     */
-    public Builder setPredNameBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      predName_ = value;
-      onChanged();
-      return this;
-    }
-
-    private java.lang.Object pivotName_ = "";
-    /**
-     * <pre>
-     * Name of the pivot tensor.
-     * </pre>
-     *
-     * <code>string pivot_name = 3;</code>
-     */
-    public java.lang.String getPivotName() {
-      java.lang.Object ref = pivotName_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        pivotName_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Name of the pivot tensor.
-     * </pre>
-     *
-     * <code>string pivot_name = 3;</code>
-     */
-    public com.google.protobuf.ByteString
-        getPivotNameBytes() {
-      java.lang.Object ref = pivotName_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        pivotName_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Name of the pivot tensor.
-     * </pre>
-     *
-     * <code>string pivot_name = 3;</code>
-     */
-    public Builder setPivotName(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      pivotName_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Name of the pivot tensor.
-     * </pre>
-     *
-     * <code>string pivot_name = 3;</code>
-     */
-    public Builder clearPivotName() {
-      
-      pivotName_ = getDefaultInstance().getPivotName();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Name of the pivot tensor.
-     * </pre>
-     *
-     * <code>string pivot_name = 3;</code>
-     */
-    public Builder setPivotNameBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      pivotName_ = value;
-      onChanged();
-      return this;
-    }
-
-    private int branch_ ;
-    /**
-     * <pre>
-     * Branch prediction. 0 or 1.
-     * </pre>
-     *
-     * <code>int32 branch = 4;</code>
-     */
-    public int getBranch() {
-      return branch_;
-    }
-    /**
-     * <pre>
-     * Branch prediction. 0 or 1.
-     * </pre>
-     *
-     * <code>int32 branch = 4;</code>
-     */
-    public Builder setBranch(int value) {
-      
-      branch_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Branch prediction. 0 or 1.
-     * </pre>
-     *
-     * <code>int32 branch = 4;</code>
-     */
-    public Builder clearBranch() {
-      
-      branch_ = 0;
-      onChanged();
-      return this;
-    }
-
-    private org.tensorflow.proto.framework.ValuesDef valuesDef_;
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.ValuesDef, org.tensorflow.proto.framework.ValuesDef.Builder, org.tensorflow.proto.framework.ValuesDefOrBuilder> valuesDefBuilder_;
-    /**
-     * <pre>
-     * Values and external values in control flow context.
-     * </pre>
-     *
-     * <code>.tensorflow.ValuesDef values_def = 5;</code>
-     */
-    public boolean hasValuesDef() {
-      return valuesDefBuilder_ != null || valuesDef_ != null;
-    }
-    /**
-     * <pre>
-     * Values and external values in control flow context.
-     * </pre>
-     *
-     * <code>.tensorflow.ValuesDef values_def = 5;</code>
-     */
-    public org.tensorflow.proto.framework.ValuesDef getValuesDef() {
-      if (valuesDefBuilder_ == null) {
-        return valuesDef_ == null ? org.tensorflow.proto.framework.ValuesDef.getDefaultInstance() : valuesDef_;
-      } else {
-        return valuesDefBuilder_.getMessage();
-      }
-    }
-    /**
-     * <pre>
-     * Values and external values in control flow context.
-     * </pre>
-     *
-     * <code>.tensorflow.ValuesDef values_def = 5;</code>
-     */
-    public Builder setValuesDef(org.tensorflow.proto.framework.ValuesDef value) {
-      if (valuesDefBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        valuesDef_ = value;
-        onChanged();
-      } else {
-        valuesDefBuilder_.setMessage(value);
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Values and external values in control flow context.
-     * </pre>
-     *
-     * <code>.tensorflow.ValuesDef values_def = 5;</code>
-     */
-    public Builder setValuesDef(
-        org.tensorflow.proto.framework.ValuesDef.Builder builderForValue) {
-      if (valuesDefBuilder_ == null) {
-        valuesDef_ = builderForValue.build();
-        onChanged();
-      } else {
-        valuesDefBuilder_.setMessage(builderForValue.build());
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Values and external values in control flow context.
-     * </pre>
-     *
-     * <code>.tensorflow.ValuesDef values_def = 5;</code>
-     */
-    public Builder mergeValuesDef(org.tensorflow.proto.framework.ValuesDef value) {
-      if (valuesDefBuilder_ == null) {
-        if (valuesDef_ != null) {
-          valuesDef_ =
-            org.tensorflow.proto.framework.ValuesDef.newBuilder(valuesDef_).mergeFrom(value).buildPartial();
-        } else {
-          valuesDef_ = value;
-        }
-        onChanged();
-      } else {
-        valuesDefBuilder_.mergeFrom(value);
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Values and external values in control flow context.
-     * </pre>
-     *
-     * <code>.tensorflow.ValuesDef values_def = 5;</code>
-     */
-    public Builder clearValuesDef() {
-      if (valuesDefBuilder_ == null) {
-        valuesDef_ = null;
-        onChanged();
-      } else {
-        valuesDef_ = null;
-        valuesDefBuilder_ = null;
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Values and external values in control flow context.
-     * </pre>
-     *
-     * <code>.tensorflow.ValuesDef values_def = 5;</code>
-     */
-    public org.tensorflow.proto.framework.ValuesDef.Builder getValuesDefBuilder() {
-      
-      onChanged();
-      return getValuesDefFieldBuilder().getBuilder();
-    }
-    /**
-     * <pre>
-     * Values and external values in control flow context.
-     * </pre>
-     *
-     * <code>.tensorflow.ValuesDef values_def = 5;</code>
-     */
-    public org.tensorflow.proto.framework.ValuesDefOrBuilder getValuesDefOrBuilder() {
-      if (valuesDefBuilder_ != null) {
-        return valuesDefBuilder_.getMessageOrBuilder();
-      } else {
-        return valuesDef_ == null ?
-            org.tensorflow.proto.framework.ValuesDef.getDefaultInstance() : valuesDef_;
-      }
-    }
-    /**
-     * <pre>
-     * Values and external values in control flow context.
-     * </pre>
-     *
-     * <code>.tensorflow.ValuesDef values_def = 5;</code>
-     */
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.ValuesDef, org.tensorflow.proto.framework.ValuesDef.Builder, org.tensorflow.proto.framework.ValuesDefOrBuilder> 
-        getValuesDefFieldBuilder() {
-      if (valuesDefBuilder_ == null) {
-        valuesDefBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-            org.tensorflow.proto.framework.ValuesDef, org.tensorflow.proto.framework.ValuesDef.Builder, org.tensorflow.proto.framework.ValuesDefOrBuilder>(
-                getValuesDef(),
-                getParentForChildren(),
-                isClean());
-        valuesDef_ = null;
-      }
-      return valuesDefBuilder_;
-    }
-
-    private java.util.List<org.tensorflow.proto.framework.ControlFlowContextDef> nestedContexts_ =
-      java.util.Collections.emptyList();
-    private void ensureNestedContextsIsMutable() {
-      if (!((bitField0_ & 0x00000001) != 0)) {
-        nestedContexts_ = new java.util.ArrayList<org.tensorflow.proto.framework.ControlFlowContextDef>(nestedContexts_);
-        bitField0_ |= 0x00000001;
-       }
-    }
-
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.framework.ControlFlowContextDef, org.tensorflow.proto.framework.ControlFlowContextDef.Builder, org.tensorflow.proto.framework.ControlFlowContextDefOrBuilder> nestedContextsBuilder_;
-
-    /**
-     * <pre>
-     * Contexts contained inside this context (e.g. nested conds).
-     * </pre>
-     *
-     * <code>repeated .tensorflow.ControlFlowContextDef nested_contexts = 6;</code>
-     */
-    public java.util.List<org.tensorflow.proto.framework.ControlFlowContextDef> getNestedContextsList() {
-      if (nestedContextsBuilder_ == null) {
-        return java.util.Collections.unmodifiableList(nestedContexts_);
-      } else {
-        return nestedContextsBuilder_.getMessageList();
-      }
-    }
-    /**
-     * <pre>
-     * Contexts contained inside this context (e.g. nested conds).
-     * </pre>
-     *
-     * <code>repeated .tensorflow.ControlFlowContextDef nested_contexts = 6;</code>
-     */
-    public int getNestedContextsCount() {
-      if (nestedContextsBuilder_ == null) {
-        return nestedContexts_.size();
-      } else {
-        return nestedContextsBuilder_.getCount();
-      }
-    }
-    /**
-     * <pre>
-     * Contexts contained inside this context (e.g. nested conds).
-     * </pre>
-     *
-     * <code>repeated .tensorflow.ControlFlowContextDef nested_contexts = 6;</code>
-     */
-    public org.tensorflow.proto.framework.ControlFlowContextDef getNestedContexts(int index) {
-      if (nestedContextsBuilder_ == null) {
-        return nestedContexts_.get(index);
-      } else {
-        return nestedContextsBuilder_.getMessage(index);
-      }
-    }
-    /**
-     * <pre>
-     * Contexts contained inside this context (e.g. nested conds).
-     * </pre>
-     *
-     * <code>repeated .tensorflow.ControlFlowContextDef nested_contexts = 6;</code>
-     */
-    public Builder setNestedContexts(
-        int index, org.tensorflow.proto.framework.ControlFlowContextDef value) {
-      if (nestedContextsBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureNestedContextsIsMutable();
-        nestedContexts_.set(index, value);
-        onChanged();
-      } else {
-        nestedContextsBuilder_.setMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Contexts contained inside this context (e.g. nested conds).
-     * </pre>
-     *
-     * <code>repeated .tensorflow.ControlFlowContextDef nested_contexts = 6;</code>
-     */
-    public Builder setNestedContexts(
-        int index, org.tensorflow.proto.framework.ControlFlowContextDef.Builder builderForValue) {
-      if (nestedContextsBuilder_ == null) {
-        ensureNestedContextsIsMutable();
-        nestedContexts_.set(index, builderForValue.build());
-        onChanged();
-      } else {
-        nestedContextsBuilder_.setMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Contexts contained inside this context (e.g. nested conds).
-     * </pre>
-     *
-     * <code>repeated .tensorflow.ControlFlowContextDef nested_contexts = 6;</code>
-     */
-    public Builder addNestedContexts(org.tensorflow.proto.framework.ControlFlowContextDef value) {
-      if (nestedContextsBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureNestedContextsIsMutable();
-        nestedContexts_.add(value);
-        onChanged();
-      } else {
-        nestedContextsBuilder_.addMessage(value);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Contexts contained inside this context (e.g. nested conds).
-     * </pre>
-     *
-     * <code>repeated .tensorflow.ControlFlowContextDef nested_contexts = 6;</code>
-     */
-    public Builder addNestedContexts(
-        int index, org.tensorflow.proto.framework.ControlFlowContextDef value) {
-      if (nestedContextsBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureNestedContextsIsMutable();
-        nestedContexts_.add(index, value);
-        onChanged();
-      } else {
-        nestedContextsBuilder_.addMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Contexts contained inside this context (e.g. nested conds).
-     * </pre>
-     *
-     * <code>repeated .tensorflow.ControlFlowContextDef nested_contexts = 6;</code>
-     */
-    public Builder addNestedContexts(
-        org.tensorflow.proto.framework.ControlFlowContextDef.Builder builderForValue) {
-      if (nestedContextsBuilder_ == null) {
-        ensureNestedContextsIsMutable();
-        nestedContexts_.add(builderForValue.build());
-        onChanged();
-      } else {
-        nestedContextsBuilder_.addMessage(builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Contexts contained inside this context (e.g. nested conds).
-     * </pre>
-     *
-     * <code>repeated .tensorflow.ControlFlowContextDef nested_contexts = 6;</code>
-     */
-    public Builder addNestedContexts(
-        int index, org.tensorflow.proto.framework.ControlFlowContextDef.Builder builderForValue) {
-      if (nestedContextsBuilder_ == null) {
-        ensureNestedContextsIsMutable();
-        nestedContexts_.add(index, builderForValue.build());
-        onChanged();
-      } else {
-        nestedContextsBuilder_.addMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Contexts contained inside this context (e.g. nested conds).
-     * </pre>
-     *
-     * <code>repeated .tensorflow.ControlFlowContextDef nested_contexts = 6;</code>
-     */
-    public Builder addAllNestedContexts(
-        java.lang.Iterable<? extends org.tensorflow.proto.framework.ControlFlowContextDef> values) {
-      if (nestedContextsBuilder_ == null) {
-        ensureNestedContextsIsMutable();
-        com.google.protobuf.AbstractMessageLite.Builder.addAll(
-            values, nestedContexts_);
-        onChanged();
-      } else {
-        nestedContextsBuilder_.addAllMessages(values);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Contexts contained inside this context (e.g. nested conds).
-     * </pre>
-     *
-     * <code>repeated .tensorflow.ControlFlowContextDef nested_contexts = 6;</code>
-     */
-    public Builder clearNestedContexts() {
-      if (nestedContextsBuilder_ == null) {
-        nestedContexts_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000001);
-        onChanged();
-      } else {
-        nestedContextsBuilder_.clear();
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Contexts contained inside this context (e.g. nested conds).
-     * </pre>
-     *
-     * <code>repeated .tensorflow.ControlFlowContextDef nested_contexts = 6;</code>
-     */
-    public Builder removeNestedContexts(int index) {
-      if (nestedContextsBuilder_ == null) {
-        ensureNestedContextsIsMutable();
-        nestedContexts_.remove(index);
-        onChanged();
-      } else {
-        nestedContextsBuilder_.remove(index);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Contexts contained inside this context (e.g. nested conds).
-     * </pre>
-     *
-     * <code>repeated .tensorflow.ControlFlowContextDef nested_contexts = 6;</code>
-     */
-    public org.tensorflow.proto.framework.ControlFlowContextDef.Builder getNestedContextsBuilder(
-        int index) {
-      return getNestedContextsFieldBuilder().getBuilder(index);
-    }
-    /**
-     * <pre>
-     * Contexts contained inside this context (e.g. nested conds).
-     * </pre>
-     *
-     * <code>repeated .tensorflow.ControlFlowContextDef nested_contexts = 6;</code>
-     */
-    public org.tensorflow.proto.framework.ControlFlowContextDefOrBuilder getNestedContextsOrBuilder(
-        int index) {
-      if (nestedContextsBuilder_ == null) {
-        return nestedContexts_.get(index);  } else {
-        return nestedContextsBuilder_.getMessageOrBuilder(index);
-      }
-    }
-    /**
-     * <pre>
-     * Contexts contained inside this context (e.g. nested conds).
-     * </pre>
-     *
-     * <code>repeated .tensorflow.ControlFlowContextDef nested_contexts = 6;</code>
-     */
-    public java.util.List<? extends org.tensorflow.proto.framework.ControlFlowContextDefOrBuilder> 
-         getNestedContextsOrBuilderList() {
-      if (nestedContextsBuilder_ != null) {
-        return nestedContextsBuilder_.getMessageOrBuilderList();
-      } else {
-        return java.util.Collections.unmodifiableList(nestedContexts_);
-      }
-    }
-    /**
-     * <pre>
-     * Contexts contained inside this context (e.g. nested conds).
-     * </pre>
-     *
-     * <code>repeated .tensorflow.ControlFlowContextDef nested_contexts = 6;</code>
-     */
-    public org.tensorflow.proto.framework.ControlFlowContextDef.Builder addNestedContextsBuilder() {
-      return getNestedContextsFieldBuilder().addBuilder(
-          org.tensorflow.proto.framework.ControlFlowContextDef.getDefaultInstance());
-    }
-    /**
-     * <pre>
-     * Contexts contained inside this context (e.g. nested conds).
-     * </pre>
-     *
-     * <code>repeated .tensorflow.ControlFlowContextDef nested_contexts = 6;</code>
-     */
-    public org.tensorflow.proto.framework.ControlFlowContextDef.Builder addNestedContextsBuilder(
-        int index) {
-      return getNestedContextsFieldBuilder().addBuilder(
-          index, org.tensorflow.proto.framework.ControlFlowContextDef.getDefaultInstance());
-    }
-    /**
-     * <pre>
-     * Contexts contained inside this context (e.g. nested conds).
-     * </pre>
-     *
-     * <code>repeated .tensorflow.ControlFlowContextDef nested_contexts = 6;</code>
-     */
-    public java.util.List<org.tensorflow.proto.framework.ControlFlowContextDef.Builder> 
-         getNestedContextsBuilderList() {
-      return getNestedContextsFieldBuilder().getBuilderList();
-    }
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.framework.ControlFlowContextDef, org.tensorflow.proto.framework.ControlFlowContextDef.Builder, org.tensorflow.proto.framework.ControlFlowContextDefOrBuilder> 
-        getNestedContextsFieldBuilder() {
-      if (nestedContextsBuilder_ == null) {
-        nestedContextsBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
-            org.tensorflow.proto.framework.ControlFlowContextDef, org.tensorflow.proto.framework.ControlFlowContextDef.Builder, org.tensorflow.proto.framework.ControlFlowContextDefOrBuilder>(
-                nestedContexts_,
-                ((bitField0_ & 0x00000001) != 0),
-                getParentForChildren(),
-                isClean());
-        nestedContexts_ = null;
-      }
-      return nestedContextsBuilder_;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.CondContextDef)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.CondContextDef)
-  private static final org.tensorflow.proto.framework.CondContextDef DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.CondContextDef();
-  }
-
-  public static org.tensorflow.proto.framework.CondContextDef getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<CondContextDef>
-      PARSER = new com.google.protobuf.AbstractParser<CondContextDef>() {
-    @java.lang.Override
-    public CondContextDef parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new CondContextDef(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<CondContextDef> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<CondContextDef> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.framework.CondContextDef getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/CondContextDefOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/CondContextDefOrBuilder.java
deleted file mode 100644
index 1dcdef5fd71..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/CondContextDefOrBuilder.java
+++ /dev/null
@@ -1,141 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/control_flow.proto
-
-package org.tensorflow.proto.framework;
-
-public interface CondContextDefOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.CondContextDef)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <pre>
-   * Name of the context.
-   * </pre>
-   *
-   * <code>string context_name = 1;</code>
-   */
-  java.lang.String getContextName();
-  /**
-   * <pre>
-   * Name of the context.
-   * </pre>
-   *
-   * <code>string context_name = 1;</code>
-   */
-  com.google.protobuf.ByteString
-      getContextNameBytes();
-
-  /**
-   * <pre>
-   * Name of the pred tensor.
-   * </pre>
-   *
-   * <code>string pred_name = 2;</code>
-   */
-  java.lang.String getPredName();
-  /**
-   * <pre>
-   * Name of the pred tensor.
-   * </pre>
-   *
-   * <code>string pred_name = 2;</code>
-   */
-  com.google.protobuf.ByteString
-      getPredNameBytes();
-
-  /**
-   * <pre>
-   * Name of the pivot tensor.
-   * </pre>
-   *
-   * <code>string pivot_name = 3;</code>
-   */
-  java.lang.String getPivotName();
-  /**
-   * <pre>
-   * Name of the pivot tensor.
-   * </pre>
-   *
-   * <code>string pivot_name = 3;</code>
-   */
-  com.google.protobuf.ByteString
-      getPivotNameBytes();
-
-  /**
-   * <pre>
-   * Branch prediction. 0 or 1.
-   * </pre>
-   *
-   * <code>int32 branch = 4;</code>
-   */
-  int getBranch();
-
-  /**
-   * <pre>
-   * Values and external values in control flow context.
-   * </pre>
-   *
-   * <code>.tensorflow.ValuesDef values_def = 5;</code>
-   */
-  boolean hasValuesDef();
-  /**
-   * <pre>
-   * Values and external values in control flow context.
-   * </pre>
-   *
-   * <code>.tensorflow.ValuesDef values_def = 5;</code>
-   */
-  org.tensorflow.proto.framework.ValuesDef getValuesDef();
-  /**
-   * <pre>
-   * Values and external values in control flow context.
-   * </pre>
-   *
-   * <code>.tensorflow.ValuesDef values_def = 5;</code>
-   */
-  org.tensorflow.proto.framework.ValuesDefOrBuilder getValuesDefOrBuilder();
-
-  /**
-   * <pre>
-   * Contexts contained inside this context (e.g. nested conds).
-   * </pre>
-   *
-   * <code>repeated .tensorflow.ControlFlowContextDef nested_contexts = 6;</code>
-   */
-  java.util.List<org.tensorflow.proto.framework.ControlFlowContextDef> 
-      getNestedContextsList();
-  /**
-   * <pre>
-   * Contexts contained inside this context (e.g. nested conds).
-   * </pre>
-   *
-   * <code>repeated .tensorflow.ControlFlowContextDef nested_contexts = 6;</code>
-   */
-  org.tensorflow.proto.framework.ControlFlowContextDef getNestedContexts(int index);
-  /**
-   * <pre>
-   * Contexts contained inside this context (e.g. nested conds).
-   * </pre>
-   *
-   * <code>repeated .tensorflow.ControlFlowContextDef nested_contexts = 6;</code>
-   */
-  int getNestedContextsCount();
-  /**
-   * <pre>
-   * Contexts contained inside this context (e.g. nested conds).
-   * </pre>
-   *
-   * <code>repeated .tensorflow.ControlFlowContextDef nested_contexts = 6;</code>
-   */
-  java.util.List<? extends org.tensorflow.proto.framework.ControlFlowContextDefOrBuilder> 
-      getNestedContextsOrBuilderList();
-  /**
-   * <pre>
-   * Contexts contained inside this context (e.g. nested conds).
-   * </pre>
-   *
-   * <code>repeated .tensorflow.ControlFlowContextDef nested_contexts = 6;</code>
-   */
-  org.tensorflow.proto.framework.ControlFlowContextDefOrBuilder getNestedContextsOrBuilder(
-      int index);
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ConfigProto.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ConfigProto.java
deleted file mode 100644
index a1c1e579953..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ConfigProto.java
+++ /dev/null
@@ -1,6064 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/config.proto
-
-package org.tensorflow.proto.framework;
-
-/**
- * <pre>
- * Session configuration parameters.
- * The system picks appropriate values for fields that are not set.
- * </pre>
- *
- * Protobuf type {@code tensorflow.ConfigProto}
- */
-public  final class ConfigProto extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.ConfigProto)
-    ConfigProtoOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use ConfigProto.newBuilder() to construct.
-  private ConfigProto(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private ConfigProto() {
-    sessionInterOpThreadPool_ = java.util.Collections.emptyList();
-    deviceFilters_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new ConfigProto();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private ConfigProto(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    int mutable_bitField0_ = 0;
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 10: {
-            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
-              deviceCount_ = com.google.protobuf.MapField.newMapField(
-                  DeviceCountDefaultEntryHolder.defaultEntry);
-              mutable_bitField0_ |= 0x00000001;
-            }
-            com.google.protobuf.MapEntry<java.lang.String, java.lang.Integer>
-            deviceCount__ = input.readMessage(
-                DeviceCountDefaultEntryHolder.defaultEntry.getParserForType(), extensionRegistry);
-            deviceCount_.getMutableMap().put(
-                deviceCount__.getKey(), deviceCount__.getValue());
-            break;
-          }
-          case 16: {
-
-            intraOpParallelismThreads_ = input.readInt32();
-            break;
-          }
-          case 24: {
-
-            placementPeriod_ = input.readInt32();
-            break;
-          }
-          case 34: {
-            java.lang.String s = input.readStringRequireUtf8();
-            if (!((mutable_bitField0_ & 0x00000004) != 0)) {
-              deviceFilters_ = new com.google.protobuf.LazyStringArrayList();
-              mutable_bitField0_ |= 0x00000004;
-            }
-            deviceFilters_.add(s);
-            break;
-          }
-          case 40: {
-
-            interOpParallelismThreads_ = input.readInt32();
-            break;
-          }
-          case 50: {
-            org.tensorflow.proto.framework.GPUOptions.Builder subBuilder = null;
-            if (gpuOptions_ != null) {
-              subBuilder = gpuOptions_.toBuilder();
-            }
-            gpuOptions_ = input.readMessage(org.tensorflow.proto.framework.GPUOptions.parser(), extensionRegistry);
-            if (subBuilder != null) {
-              subBuilder.mergeFrom(gpuOptions_);
-              gpuOptions_ = subBuilder.buildPartial();
-            }
-
-            break;
-          }
-          case 56: {
-
-            allowSoftPlacement_ = input.readBool();
-            break;
-          }
-          case 64: {
-
-            logDevicePlacement_ = input.readBool();
-            break;
-          }
-          case 72: {
-
-            usePerSessionThreads_ = input.readBool();
-            break;
-          }
-          case 82: {
-            org.tensorflow.proto.framework.GraphOptions.Builder subBuilder = null;
-            if (graphOptions_ != null) {
-              subBuilder = graphOptions_.toBuilder();
-            }
-            graphOptions_ = input.readMessage(org.tensorflow.proto.framework.GraphOptions.parser(), extensionRegistry);
-            if (subBuilder != null) {
-              subBuilder.mergeFrom(graphOptions_);
-              graphOptions_ = subBuilder.buildPartial();
-            }
-
-            break;
-          }
-          case 88: {
-
-            operationTimeoutInMs_ = input.readInt64();
-            break;
-          }
-          case 98: {
-            if (!((mutable_bitField0_ & 0x00000002) != 0)) {
-              sessionInterOpThreadPool_ = new java.util.ArrayList<org.tensorflow.proto.framework.ThreadPoolOptionProto>();
-              mutable_bitField0_ |= 0x00000002;
-            }
-            sessionInterOpThreadPool_.add(
-                input.readMessage(org.tensorflow.proto.framework.ThreadPoolOptionProto.parser(), extensionRegistry));
-            break;
-          }
-          case 106: {
-            org.tensorflow.proto.framework.RPCOptions.Builder subBuilder = null;
-            if (rpcOptions_ != null) {
-              subBuilder = rpcOptions_.toBuilder();
-            }
-            rpcOptions_ = input.readMessage(org.tensorflow.proto.framework.RPCOptions.parser(), extensionRegistry);
-            if (subBuilder != null) {
-              subBuilder.mergeFrom(rpcOptions_);
-              rpcOptions_ = subBuilder.buildPartial();
-            }
-
-            break;
-          }
-          case 114: {
-            org.tensorflow.proto.distruntime.ClusterDef.Builder subBuilder = null;
-            if (clusterDef_ != null) {
-              subBuilder = clusterDef_.toBuilder();
-            }
-            clusterDef_ = input.readMessage(org.tensorflow.proto.distruntime.ClusterDef.parser(), extensionRegistry);
-            if (subBuilder != null) {
-              subBuilder.mergeFrom(clusterDef_);
-              clusterDef_ = subBuilder.buildPartial();
-            }
-
-            break;
-          }
-          case 120: {
-
-            isolateSessionState_ = input.readBool();
-            break;
-          }
-          case 130: {
-            org.tensorflow.proto.framework.ConfigProto.Experimental.Builder subBuilder = null;
-            if (experimental_ != null) {
-              subBuilder = experimental_.toBuilder();
-            }
-            experimental_ = input.readMessage(org.tensorflow.proto.framework.ConfigProto.Experimental.parser(), extensionRegistry);
-            if (subBuilder != null) {
-              subBuilder.mergeFrom(experimental_);
-              experimental_ = subBuilder.buildPartial();
-            }
-
-            break;
-          }
-          case 136: {
-
-            shareClusterDevicesInSession_ = input.readBool();
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      if (((mutable_bitField0_ & 0x00000004) != 0)) {
-        deviceFilters_ = deviceFilters_.getUnmodifiableView();
-      }
-      if (((mutable_bitField0_ & 0x00000002) != 0)) {
-        sessionInterOpThreadPool_ = java.util.Collections.unmodifiableList(sessionInterOpThreadPool_);
-      }
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_ConfigProto_descriptor;
-  }
-
-  @SuppressWarnings({"rawtypes"})
-  @java.lang.Override
-  protected com.google.protobuf.MapField internalGetMapField(
-      int number) {
-    switch (number) {
-      case 1:
-        return internalGetDeviceCount();
-      default:
-        throw new RuntimeException(
-            "Invalid map field number: " + number);
-    }
-  }
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_ConfigProto_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.framework.ConfigProto.class, org.tensorflow.proto.framework.ConfigProto.Builder.class);
-  }
-
-  public interface ExperimentalOrBuilder extends
-      // @@protoc_insertion_point(interface_extends:tensorflow.ConfigProto.Experimental)
-      com.google.protobuf.MessageOrBuilder {
-
-    /**
-     * <pre>
-     * Task name for group resolution.
-     * </pre>
-     *
-     * <code>string collective_group_leader = 1;</code>
-     */
-    java.lang.String getCollectiveGroupLeader();
-    /**
-     * <pre>
-     * Task name for group resolution.
-     * </pre>
-     *
-     * <code>string collective_group_leader = 1;</code>
-     */
-    com.google.protobuf.ByteString
-        getCollectiveGroupLeaderBytes();
-
-    /**
-     * <pre>
-     * Which executor to use, the default executor will be used
-     * if it is an empty string or "DEFAULT"
-     * </pre>
-     *
-     * <code>string executor_type = 3;</code>
-     */
-    java.lang.String getExecutorType();
-    /**
-     * <pre>
-     * Which executor to use, the default executor will be used
-     * if it is an empty string or "DEFAULT"
-     * </pre>
-     *
-     * <code>string executor_type = 3;</code>
-     */
-    com.google.protobuf.ByteString
-        getExecutorTypeBytes();
-
-    /**
-     * <pre>
-     * Guidance to formatting of large RecvBuf fields for transfer.
-     * Any positive value sets the max chunk size.  0 defaults to 4096.
-     * Any negative value indicates no max, i.e. one chunk only.
-     * </pre>
-     *
-     * <code>int32 recv_buf_max_chunk = 4;</code>
-     */
-    int getRecvBufMaxChunk();
-
-    /**
-     * <pre>
-     * If true, and supported by the platform, the runtime will attempt to
-     * use NUMA affinity where applicable.  One consequence will be the
-     * existence of as many CPU devices as there are available NUMA nodes.
-     * </pre>
-     *
-     * <code>bool use_numa_affinity = 5;</code>
-     */
-    boolean getUseNumaAffinity();
-
-    /**
-     * <pre>
-     * If true, make collective op execution order sequential and deterministic
-     * for potentially concurrent collective instances.
-     * </pre>
-     *
-     * <code>bool collective_deterministic_sequential_execution = 6;</code>
-     */
-    boolean getCollectiveDeterministicSequentialExecution();
-
-    /**
-     * <pre>
-     * If true, use NCCL for CollectiveOps.  This feature is highly
-     * experimental.
-     * </pre>
-     *
-     * <code>bool collective_nccl = 7;</code>
-     */
-    boolean getCollectiveNccl();
-
-    /**
-     * <pre>
-     * In the following, session state means the value of a variable, elements
-     * in a hash table, or any other resource, accessible by worker sessions
-     * held by a TF server.
-     * When ClusterSpec propagation is enabled, the value of
-     * isolate_session_state is ignored when deciding whether to share session
-     * states in a TF server (for backwards compatibility reasons).
-     * - If share_session_state_in_clusterspec_propagation is true, the session
-     * states are shared.
-     * - If share_session_state_in_clusterspec_propagation is false, session
-     * states are isolated.
-     * When clusterspec propagation is not used, the value of
-     * share_session_state_in_clusterspec_propagation is ignored when deciding
-     * whether to share session states in a TF server.
-     * - If isolate_session_state is true, session states are isolated.
-     * - If isolate_session_state is false, session states are shared.
-     * TODO(b/129330037): Add a single API that consistently treats
-     * isolate_session_state and ClusterSpec propagation.
-     * </pre>
-     *
-     * <code>bool share_session_state_in_clusterspec_propagation = 8;</code>
-     */
-    boolean getShareSessionStateInClusterspecPropagation();
-
-    /**
-     * <pre>
-     * If using a direct session, disable spinning while waiting for work in
-     * the thread pool. This may result in higher latency for completing ops,
-     * but in the case where there is a lot of spinning may result in lower
-     * CPU usage.
-     * </pre>
-     *
-     * <code>bool disable_thread_spinning = 9;</code>
-     */
-    boolean getDisableThreadSpinning();
-
-    /**
-     * <pre>
-     * This was promoted to a non-experimental API. Please use
-     * ConfigProto.share_cluster_devices_in_session instead.
-     * </pre>
-     *
-     * <code>bool share_cluster_devices_in_session = 10;</code>
-     */
-    boolean getShareClusterDevicesInSession();
-
-    /**
-     * <pre>
-     * Metadata about the session.
-     * If set, this can be used by the runtime and the Ops for debugging,
-     * monitoring, etc.
-     * NOTE: This is currently used and propagated only by the direct session.
-     * </pre>
-     *
-     * <code>.tensorflow.SessionMetadata session_metadata = 11;</code>
-     */
-    boolean hasSessionMetadata();
-    /**
-     * <pre>
-     * Metadata about the session.
-     * If set, this can be used by the runtime and the Ops for debugging,
-     * monitoring, etc.
-     * NOTE: This is currently used and propagated only by the direct session.
-     * </pre>
-     *
-     * <code>.tensorflow.SessionMetadata session_metadata = 11;</code>
-     */
-    org.tensorflow.proto.framework.SessionMetadata getSessionMetadata();
-    /**
-     * <pre>
-     * Metadata about the session.
-     * If set, this can be used by the runtime and the Ops for debugging,
-     * monitoring, etc.
-     * NOTE: This is currently used and propagated only by the direct session.
-     * </pre>
-     *
-     * <code>.tensorflow.SessionMetadata session_metadata = 11;</code>
-     */
-    org.tensorflow.proto.framework.SessionMetadataOrBuilder getSessionMetadataOrBuilder();
-
-    /**
-     * <pre>
-     * If true, the session may treat the graph as being static for optimization
-     * purposes.
-     * If this option is set to true when a session is created, the full
-     * GraphDef must be passed in a single call to Session::Create(), and
-     * Session::Extend() may not be supported.
-     * </pre>
-     *
-     * <code>bool optimize_for_static_graph = 12;</code>
-     */
-    boolean getOptimizeForStaticGraph();
-
-    /**
-     * <pre>
-     * Whether to enable the MLIR-based TF-&gt;XLA bridge.
-     * This is a replacement to the existing bridge, and not ready for
-     * production usage yet.
-     * If this option is set to true when a session is created, MLIR is used to
-     * perform the set of graph transformations to put the graph in a form that
-     * can be executed with delegation of some computations to an accelerator.
-     * This builds on the model of XLA where a subset of the graph is
-     * encapsulated and attached to a "compile" operation, whose result is fed
-     * to an "execute" operation. The kernel for these operations is responsible
-     * to lower the encapsulated graph to a particular device.
-     * </pre>
-     *
-     * <code>bool enable_mlir_bridge = 13;</code>
-     */
-    boolean getEnableMlirBridge();
-
-    /**
-     * <pre>
-     * Whether to enable the MLIR-based Graph optimizations.
-     * This will become a part of standard Tensorflow graph optimization
-     * pipeline, currently this is only used for gradual migration and testing
-     * new passes that are replacing existing optimizations in Grappler.
-     * </pre>
-     *
-     * <code>bool enable_mlir_graph_optimization = 16;</code>
-     */
-    boolean getEnableMlirGraphOptimization();
-
-    /**
-     * <pre>
-     * If true, the session will not store an additional copy of the graph for
-     * each subgraph.
-     * If this option is set to true when a session is created, the
-     * `RunOptions.output_partition_graphs` options must not be set.
-     * </pre>
-     *
-     * <code>bool disable_output_partition_graphs = 14;</code>
-     */
-    boolean getDisableOutputPartitionGraphs();
-
-    /**
-     * <pre>
-     * Minimum number of batches run through the XLA graph before XLA fusion
-     * autotuner is enabled. Default value of zero disables the autotuner.
-     * The XLA fusion autotuner can improve performance by executing a heuristic
-     * search on the compiler parameters.
-     * </pre>
-     *
-     * <code>int64 xla_fusion_autotuner_thresh = 15;</code>
-     */
-    long getXlaFusionAutotunerThresh();
-  }
-  /**
-   * <pre>
-   * Everything inside Experimental is subject to change and is not subject
-   * to API stability guarantees in
-   * https://www.tensorflow.org/guide/version_compat.
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.ConfigProto.Experimental}
-   */
-  public  static final class Experimental extends
-      com.google.protobuf.GeneratedMessageV3 implements
-      // @@protoc_insertion_point(message_implements:tensorflow.ConfigProto.Experimental)
-      ExperimentalOrBuilder {
-  private static final long serialVersionUID = 0L;
-    // Use Experimental.newBuilder() to construct.
-    private Experimental(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-      super(builder);
-    }
-    private Experimental() {
-      collectiveGroupLeader_ = "";
-      executorType_ = "";
-    }
-
-    @java.lang.Override
-    @SuppressWarnings({"unused"})
-    protected java.lang.Object newInstance(
-        UnusedPrivateParameter unused) {
-      return new Experimental();
-    }
-
-    @java.lang.Override
-    public final com.google.protobuf.UnknownFieldSet
-    getUnknownFields() {
-      return this.unknownFields;
-    }
-    private Experimental(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      this();
-      if (extensionRegistry == null) {
-        throw new java.lang.NullPointerException();
-      }
-      com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-          com.google.protobuf.UnknownFieldSet.newBuilder();
-      try {
-        boolean done = false;
-        while (!done) {
-          int tag = input.readTag();
-          switch (tag) {
-            case 0:
-              done = true;
-              break;
-            case 10: {
-              java.lang.String s = input.readStringRequireUtf8();
-
-              collectiveGroupLeader_ = s;
-              break;
-            }
-            case 26: {
-              java.lang.String s = input.readStringRequireUtf8();
-
-              executorType_ = s;
-              break;
-            }
-            case 32: {
-
-              recvBufMaxChunk_ = input.readInt32();
-              break;
-            }
-            case 40: {
-
-              useNumaAffinity_ = input.readBool();
-              break;
-            }
-            case 48: {
-
-              collectiveDeterministicSequentialExecution_ = input.readBool();
-              break;
-            }
-            case 56: {
-
-              collectiveNccl_ = input.readBool();
-              break;
-            }
-            case 64: {
-
-              shareSessionStateInClusterspecPropagation_ = input.readBool();
-              break;
-            }
-            case 72: {
-
-              disableThreadSpinning_ = input.readBool();
-              break;
-            }
-            case 80: {
-
-              shareClusterDevicesInSession_ = input.readBool();
-              break;
-            }
-            case 90: {
-              org.tensorflow.proto.framework.SessionMetadata.Builder subBuilder = null;
-              if (sessionMetadata_ != null) {
-                subBuilder = sessionMetadata_.toBuilder();
-              }
-              sessionMetadata_ = input.readMessage(org.tensorflow.proto.framework.SessionMetadata.parser(), extensionRegistry);
-              if (subBuilder != null) {
-                subBuilder.mergeFrom(sessionMetadata_);
-                sessionMetadata_ = subBuilder.buildPartial();
-              }
-
-              break;
-            }
-            case 96: {
-
-              optimizeForStaticGraph_ = input.readBool();
-              break;
-            }
-            case 104: {
-
-              enableMlirBridge_ = input.readBool();
-              break;
-            }
-            case 112: {
-
-              disableOutputPartitionGraphs_ = input.readBool();
-              break;
-            }
-            case 120: {
-
-              xlaFusionAutotunerThresh_ = input.readInt64();
-              break;
-            }
-            case 128: {
-
-              enableMlirGraphOptimization_ = input.readBool();
-              break;
-            }
-            default: {
-              if (!parseUnknownField(
-                  input, unknownFields, extensionRegistry, tag)) {
-                done = true;
-              }
-              break;
-            }
-          }
-        }
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        throw e.setUnfinishedMessage(this);
-      } catch (java.io.IOException e) {
-        throw new com.google.protobuf.InvalidProtocolBufferException(
-            e).setUnfinishedMessage(this);
-      } finally {
-        this.unknownFields = unknownFields.build();
-        makeExtensionsImmutable();
-      }
-    }
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_ConfigProto_Experimental_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_ConfigProto_Experimental_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.ConfigProto.Experimental.class, org.tensorflow.proto.framework.ConfigProto.Experimental.Builder.class);
-    }
-
-    public static final int COLLECTIVE_GROUP_LEADER_FIELD_NUMBER = 1;
-    private volatile java.lang.Object collectiveGroupLeader_;
-    /**
-     * <pre>
-     * Task name for group resolution.
-     * </pre>
-     *
-     * <code>string collective_group_leader = 1;</code>
-     */
-    public java.lang.String getCollectiveGroupLeader() {
-      java.lang.Object ref = collectiveGroupLeader_;
-      if (ref instanceof java.lang.String) {
-        return (java.lang.String) ref;
-      } else {
-        com.google.protobuf.ByteString bs = 
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        collectiveGroupLeader_ = s;
-        return s;
-      }
-    }
-    /**
-     * <pre>
-     * Task name for group resolution.
-     * </pre>
-     *
-     * <code>string collective_group_leader = 1;</code>
-     */
-    public com.google.protobuf.ByteString
-        getCollectiveGroupLeaderBytes() {
-      java.lang.Object ref = collectiveGroupLeader_;
-      if (ref instanceof java.lang.String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        collectiveGroupLeader_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-
-    public static final int EXECUTOR_TYPE_FIELD_NUMBER = 3;
-    private volatile java.lang.Object executorType_;
-    /**
-     * <pre>
-     * Which executor to use, the default executor will be used
-     * if it is an empty string or "DEFAULT"
-     * </pre>
-     *
-     * <code>string executor_type = 3;</code>
-     */
-    public java.lang.String getExecutorType() {
-      java.lang.Object ref = executorType_;
-      if (ref instanceof java.lang.String) {
-        return (java.lang.String) ref;
-      } else {
-        com.google.protobuf.ByteString bs = 
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        executorType_ = s;
-        return s;
-      }
-    }
-    /**
-     * <pre>
-     * Which executor to use, the default executor will be used
-     * if it is an empty string or "DEFAULT"
-     * </pre>
-     *
-     * <code>string executor_type = 3;</code>
-     */
-    public com.google.protobuf.ByteString
-        getExecutorTypeBytes() {
-      java.lang.Object ref = executorType_;
-      if (ref instanceof java.lang.String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        executorType_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-
-    public static final int RECV_BUF_MAX_CHUNK_FIELD_NUMBER = 4;
-    private int recvBufMaxChunk_;
-    /**
-     * <pre>
-     * Guidance to formatting of large RecvBuf fields for transfer.
-     * Any positive value sets the max chunk size.  0 defaults to 4096.
-     * Any negative value indicates no max, i.e. one chunk only.
-     * </pre>
-     *
-     * <code>int32 recv_buf_max_chunk = 4;</code>
-     */
-    public int getRecvBufMaxChunk() {
-      return recvBufMaxChunk_;
-    }
-
-    public static final int USE_NUMA_AFFINITY_FIELD_NUMBER = 5;
-    private boolean useNumaAffinity_;
-    /**
-     * <pre>
-     * If true, and supported by the platform, the runtime will attempt to
-     * use NUMA affinity where applicable.  One consequence will be the
-     * existence of as many CPU devices as there are available NUMA nodes.
-     * </pre>
-     *
-     * <code>bool use_numa_affinity = 5;</code>
-     */
-    public boolean getUseNumaAffinity() {
-      return useNumaAffinity_;
-    }
-
-    public static final int COLLECTIVE_DETERMINISTIC_SEQUENTIAL_EXECUTION_FIELD_NUMBER = 6;
-    private boolean collectiveDeterministicSequentialExecution_;
-    /**
-     * <pre>
-     * If true, make collective op execution order sequential and deterministic
-     * for potentially concurrent collective instances.
-     * </pre>
-     *
-     * <code>bool collective_deterministic_sequential_execution = 6;</code>
-     */
-    public boolean getCollectiveDeterministicSequentialExecution() {
-      return collectiveDeterministicSequentialExecution_;
-    }
-
-    public static final int COLLECTIVE_NCCL_FIELD_NUMBER = 7;
-    private boolean collectiveNccl_;
-    /**
-     * <pre>
-     * If true, use NCCL for CollectiveOps.  This feature is highly
-     * experimental.
-     * </pre>
-     *
-     * <code>bool collective_nccl = 7;</code>
-     */
-    public boolean getCollectiveNccl() {
-      return collectiveNccl_;
-    }
-
-    public static final int SHARE_SESSION_STATE_IN_CLUSTERSPEC_PROPAGATION_FIELD_NUMBER = 8;
-    private boolean shareSessionStateInClusterspecPropagation_;
-    /**
-     * <pre>
-     * In the following, session state means the value of a variable, elements
-     * in a hash table, or any other resource, accessible by worker sessions
-     * held by a TF server.
-     * When ClusterSpec propagation is enabled, the value of
-     * isolate_session_state is ignored when deciding whether to share session
-     * states in a TF server (for backwards compatibility reasons).
-     * - If share_session_state_in_clusterspec_propagation is true, the session
-     * states are shared.
-     * - If share_session_state_in_clusterspec_propagation is false, session
-     * states are isolated.
-     * When clusterspec propagation is not used, the value of
-     * share_session_state_in_clusterspec_propagation is ignored when deciding
-     * whether to share session states in a TF server.
-     * - If isolate_session_state is true, session states are isolated.
-     * - If isolate_session_state is false, session states are shared.
-     * TODO(b/129330037): Add a single API that consistently treats
-     * isolate_session_state and ClusterSpec propagation.
-     * </pre>
-     *
-     * <code>bool share_session_state_in_clusterspec_propagation = 8;</code>
-     */
-    public boolean getShareSessionStateInClusterspecPropagation() {
-      return shareSessionStateInClusterspecPropagation_;
-    }
-
-    public static final int DISABLE_THREAD_SPINNING_FIELD_NUMBER = 9;
-    private boolean disableThreadSpinning_;
-    /**
-     * <pre>
-     * If using a direct session, disable spinning while waiting for work in
-     * the thread pool. This may result in higher latency for completing ops,
-     * but in the case where there is a lot of spinning may result in lower
-     * CPU usage.
-     * </pre>
-     *
-     * <code>bool disable_thread_spinning = 9;</code>
-     */
-    public boolean getDisableThreadSpinning() {
-      return disableThreadSpinning_;
-    }
-
-    public static final int SHARE_CLUSTER_DEVICES_IN_SESSION_FIELD_NUMBER = 10;
-    private boolean shareClusterDevicesInSession_;
-    /**
-     * <pre>
-     * This was promoted to a non-experimental API. Please use
-     * ConfigProto.share_cluster_devices_in_session instead.
-     * </pre>
-     *
-     * <code>bool share_cluster_devices_in_session = 10;</code>
-     */
-    public boolean getShareClusterDevicesInSession() {
-      return shareClusterDevicesInSession_;
-    }
-
-    public static final int SESSION_METADATA_FIELD_NUMBER = 11;
-    private org.tensorflow.proto.framework.SessionMetadata sessionMetadata_;
-    /**
-     * <pre>
-     * Metadata about the session.
-     * If set, this can be used by the runtime and the Ops for debugging,
-     * monitoring, etc.
-     * NOTE: This is currently used and propagated only by the direct session.
-     * </pre>
-     *
-     * <code>.tensorflow.SessionMetadata session_metadata = 11;</code>
-     */
-    public boolean hasSessionMetadata() {
-      return sessionMetadata_ != null;
-    }
-    /**
-     * <pre>
-     * Metadata about the session.
-     * If set, this can be used by the runtime and the Ops for debugging,
-     * monitoring, etc.
-     * NOTE: This is currently used and propagated only by the direct session.
-     * </pre>
-     *
-     * <code>.tensorflow.SessionMetadata session_metadata = 11;</code>
-     */
-    public org.tensorflow.proto.framework.SessionMetadata getSessionMetadata() {
-      return sessionMetadata_ == null ? org.tensorflow.proto.framework.SessionMetadata.getDefaultInstance() : sessionMetadata_;
-    }
-    /**
-     * <pre>
-     * Metadata about the session.
-     * If set, this can be used by the runtime and the Ops for debugging,
-     * monitoring, etc.
-     * NOTE: This is currently used and propagated only by the direct session.
-     * </pre>
-     *
-     * <code>.tensorflow.SessionMetadata session_metadata = 11;</code>
-     */
-    public org.tensorflow.proto.framework.SessionMetadataOrBuilder getSessionMetadataOrBuilder() {
-      return getSessionMetadata();
-    }
-
-    public static final int OPTIMIZE_FOR_STATIC_GRAPH_FIELD_NUMBER = 12;
-    private boolean optimizeForStaticGraph_;
-    /**
-     * <pre>
-     * If true, the session may treat the graph as being static for optimization
-     * purposes.
-     * If this option is set to true when a session is created, the full
-     * GraphDef must be passed in a single call to Session::Create(), and
-     * Session::Extend() may not be supported.
-     * </pre>
-     *
-     * <code>bool optimize_for_static_graph = 12;</code>
-     */
-    public boolean getOptimizeForStaticGraph() {
-      return optimizeForStaticGraph_;
-    }
-
-    public static final int ENABLE_MLIR_BRIDGE_FIELD_NUMBER = 13;
-    private boolean enableMlirBridge_;
-    /**
-     * <pre>
-     * Whether to enable the MLIR-based TF-&gt;XLA bridge.
-     * This is a replacement to the existing bridge, and not ready for
-     * production usage yet.
-     * If this option is set to true when a session is created, MLIR is used to
-     * perform the set of graph transformations to put the graph in a form that
-     * can be executed with delegation of some computations to an accelerator.
-     * This builds on the model of XLA where a subset of the graph is
-     * encapsulated and attached to a "compile" operation, whose result is fed
-     * to an "execute" operation. The kernel for these operations is responsible
-     * to lower the encapsulated graph to a particular device.
-     * </pre>
-     *
-     * <code>bool enable_mlir_bridge = 13;</code>
-     */
-    public boolean getEnableMlirBridge() {
-      return enableMlirBridge_;
-    }
-
-    public static final int ENABLE_MLIR_GRAPH_OPTIMIZATION_FIELD_NUMBER = 16;
-    private boolean enableMlirGraphOptimization_;
-    /**
-     * <pre>
-     * Whether to enable the MLIR-based Graph optimizations.
-     * This will become a part of standard Tensorflow graph optimization
-     * pipeline, currently this is only used for gradual migration and testing
-     * new passes that are replacing existing optimizations in Grappler.
-     * </pre>
-     *
-     * <code>bool enable_mlir_graph_optimization = 16;</code>
-     */
-    public boolean getEnableMlirGraphOptimization() {
-      return enableMlirGraphOptimization_;
-    }
-
-    public static final int DISABLE_OUTPUT_PARTITION_GRAPHS_FIELD_NUMBER = 14;
-    private boolean disableOutputPartitionGraphs_;
-    /**
-     * <pre>
-     * If true, the session will not store an additional copy of the graph for
-     * each subgraph.
-     * If this option is set to true when a session is created, the
-     * `RunOptions.output_partition_graphs` options must not be set.
-     * </pre>
-     *
-     * <code>bool disable_output_partition_graphs = 14;</code>
-     */
-    public boolean getDisableOutputPartitionGraphs() {
-      return disableOutputPartitionGraphs_;
-    }
-
-    public static final int XLA_FUSION_AUTOTUNER_THRESH_FIELD_NUMBER = 15;
-    private long xlaFusionAutotunerThresh_;
-    /**
-     * <pre>
-     * Minimum number of batches run through the XLA graph before XLA fusion
-     * autotuner is enabled. Default value of zero disables the autotuner.
-     * The XLA fusion autotuner can improve performance by executing a heuristic
-     * search on the compiler parameters.
-     * </pre>
-     *
-     * <code>int64 xla_fusion_autotuner_thresh = 15;</code>
-     */
-    public long getXlaFusionAutotunerThresh() {
-      return xlaFusionAutotunerThresh_;
-    }
-
-    private byte memoizedIsInitialized = -1;
-    @java.lang.Override
-    public final boolean isInitialized() {
-      byte isInitialized = memoizedIsInitialized;
-      if (isInitialized == 1) return true;
-      if (isInitialized == 0) return false;
-
-      memoizedIsInitialized = 1;
-      return true;
-    }
-
-    @java.lang.Override
-    public void writeTo(com.google.protobuf.CodedOutputStream output)
-                        throws java.io.IOException {
-      if (!getCollectiveGroupLeaderBytes().isEmpty()) {
-        com.google.protobuf.GeneratedMessageV3.writeString(output, 1, collectiveGroupLeader_);
-      }
-      if (!getExecutorTypeBytes().isEmpty()) {
-        com.google.protobuf.GeneratedMessageV3.writeString(output, 3, executorType_);
-      }
-      if (recvBufMaxChunk_ != 0) {
-        output.writeInt32(4, recvBufMaxChunk_);
-      }
-      if (useNumaAffinity_ != false) {
-        output.writeBool(5, useNumaAffinity_);
-      }
-      if (collectiveDeterministicSequentialExecution_ != false) {
-        output.writeBool(6, collectiveDeterministicSequentialExecution_);
-      }
-      if (collectiveNccl_ != false) {
-        output.writeBool(7, collectiveNccl_);
-      }
-      if (shareSessionStateInClusterspecPropagation_ != false) {
-        output.writeBool(8, shareSessionStateInClusterspecPropagation_);
-      }
-      if (disableThreadSpinning_ != false) {
-        output.writeBool(9, disableThreadSpinning_);
-      }
-      if (shareClusterDevicesInSession_ != false) {
-        output.writeBool(10, shareClusterDevicesInSession_);
-      }
-      if (sessionMetadata_ != null) {
-        output.writeMessage(11, getSessionMetadata());
-      }
-      if (optimizeForStaticGraph_ != false) {
-        output.writeBool(12, optimizeForStaticGraph_);
-      }
-      if (enableMlirBridge_ != false) {
-        output.writeBool(13, enableMlirBridge_);
-      }
-      if (disableOutputPartitionGraphs_ != false) {
-        output.writeBool(14, disableOutputPartitionGraphs_);
-      }
-      if (xlaFusionAutotunerThresh_ != 0L) {
-        output.writeInt64(15, xlaFusionAutotunerThresh_);
-      }
-      if (enableMlirGraphOptimization_ != false) {
-        output.writeBool(16, enableMlirGraphOptimization_);
-      }
-      unknownFields.writeTo(output);
-    }
-
-    @java.lang.Override
-    public int getSerializedSize() {
-      int size = memoizedSize;
-      if (size != -1) return size;
-
-      size = 0;
-      if (!getCollectiveGroupLeaderBytes().isEmpty()) {
-        size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, collectiveGroupLeader_);
-      }
-      if (!getExecutorTypeBytes().isEmpty()) {
-        size += com.google.protobuf.GeneratedMessageV3.computeStringSize(3, executorType_);
-      }
-      if (recvBufMaxChunk_ != 0) {
-        size += com.google.protobuf.CodedOutputStream
-          .computeInt32Size(4, recvBufMaxChunk_);
-      }
-      if (useNumaAffinity_ != false) {
-        size += com.google.protobuf.CodedOutputStream
-          .computeBoolSize(5, useNumaAffinity_);
-      }
-      if (collectiveDeterministicSequentialExecution_ != false) {
-        size += com.google.protobuf.CodedOutputStream
-          .computeBoolSize(6, collectiveDeterministicSequentialExecution_);
-      }
-      if (collectiveNccl_ != false) {
-        size += com.google.protobuf.CodedOutputStream
-          .computeBoolSize(7, collectiveNccl_);
-      }
-      if (shareSessionStateInClusterspecPropagation_ != false) {
-        size += com.google.protobuf.CodedOutputStream
-          .computeBoolSize(8, shareSessionStateInClusterspecPropagation_);
-      }
-      if (disableThreadSpinning_ != false) {
-        size += com.google.protobuf.CodedOutputStream
-          .computeBoolSize(9, disableThreadSpinning_);
-      }
-      if (shareClusterDevicesInSession_ != false) {
-        size += com.google.protobuf.CodedOutputStream
-          .computeBoolSize(10, shareClusterDevicesInSession_);
-      }
-      if (sessionMetadata_ != null) {
-        size += com.google.protobuf.CodedOutputStream
-          .computeMessageSize(11, getSessionMetadata());
-      }
-      if (optimizeForStaticGraph_ != false) {
-        size += com.google.protobuf.CodedOutputStream
-          .computeBoolSize(12, optimizeForStaticGraph_);
-      }
-      if (enableMlirBridge_ != false) {
-        size += com.google.protobuf.CodedOutputStream
-          .computeBoolSize(13, enableMlirBridge_);
-      }
-      if (disableOutputPartitionGraphs_ != false) {
-        size += com.google.protobuf.CodedOutputStream
-          .computeBoolSize(14, disableOutputPartitionGraphs_);
-      }
-      if (xlaFusionAutotunerThresh_ != 0L) {
-        size += com.google.protobuf.CodedOutputStream
-          .computeInt64Size(15, xlaFusionAutotunerThresh_);
-      }
-      if (enableMlirGraphOptimization_ != false) {
-        size += com.google.protobuf.CodedOutputStream
-          .computeBoolSize(16, enableMlirGraphOptimization_);
-      }
-      size += unknownFields.getSerializedSize();
-      memoizedSize = size;
-      return size;
-    }
-
-    @java.lang.Override
-    public boolean equals(final java.lang.Object obj) {
-      if (obj == this) {
-       return true;
-      }
-      if (!(obj instanceof org.tensorflow.proto.framework.ConfigProto.Experimental)) {
-        return super.equals(obj);
-      }
-      org.tensorflow.proto.framework.ConfigProto.Experimental other = (org.tensorflow.proto.framework.ConfigProto.Experimental) obj;
-
-      if (!getCollectiveGroupLeader()
-          .equals(other.getCollectiveGroupLeader())) return false;
-      if (!getExecutorType()
-          .equals(other.getExecutorType())) return false;
-      if (getRecvBufMaxChunk()
-          != other.getRecvBufMaxChunk()) return false;
-      if (getUseNumaAffinity()
-          != other.getUseNumaAffinity()) return false;
-      if (getCollectiveDeterministicSequentialExecution()
-          != other.getCollectiveDeterministicSequentialExecution()) return false;
-      if (getCollectiveNccl()
-          != other.getCollectiveNccl()) return false;
-      if (getShareSessionStateInClusterspecPropagation()
-          != other.getShareSessionStateInClusterspecPropagation()) return false;
-      if (getDisableThreadSpinning()
-          != other.getDisableThreadSpinning()) return false;
-      if (getShareClusterDevicesInSession()
-          != other.getShareClusterDevicesInSession()) return false;
-      if (hasSessionMetadata() != other.hasSessionMetadata()) return false;
-      if (hasSessionMetadata()) {
-        if (!getSessionMetadata()
-            .equals(other.getSessionMetadata())) return false;
-      }
-      if (getOptimizeForStaticGraph()
-          != other.getOptimizeForStaticGraph()) return false;
-      if (getEnableMlirBridge()
-          != other.getEnableMlirBridge()) return false;
-      if (getEnableMlirGraphOptimization()
-          != other.getEnableMlirGraphOptimization()) return false;
-      if (getDisableOutputPartitionGraphs()
-          != other.getDisableOutputPartitionGraphs()) return false;
-      if (getXlaFusionAutotunerThresh()
-          != other.getXlaFusionAutotunerThresh()) return false;
-      if (!unknownFields.equals(other.unknownFields)) return false;
-      return true;
-    }
-
-    @java.lang.Override
-    public int hashCode() {
-      if (memoizedHashCode != 0) {
-        return memoizedHashCode;
-      }
-      int hash = 41;
-      hash = (19 * hash) + getDescriptor().hashCode();
-      hash = (37 * hash) + COLLECTIVE_GROUP_LEADER_FIELD_NUMBER;
-      hash = (53 * hash) + getCollectiveGroupLeader().hashCode();
-      hash = (37 * hash) + EXECUTOR_TYPE_FIELD_NUMBER;
-      hash = (53 * hash) + getExecutorType().hashCode();
-      hash = (37 * hash) + RECV_BUF_MAX_CHUNK_FIELD_NUMBER;
-      hash = (53 * hash) + getRecvBufMaxChunk();
-      hash = (37 * hash) + USE_NUMA_AFFINITY_FIELD_NUMBER;
-      hash = (53 * hash) + com.google.protobuf.Internal.hashBoolean(
-          getUseNumaAffinity());
-      hash = (37 * hash) + COLLECTIVE_DETERMINISTIC_SEQUENTIAL_EXECUTION_FIELD_NUMBER;
-      hash = (53 * hash) + com.google.protobuf.Internal.hashBoolean(
-          getCollectiveDeterministicSequentialExecution());
-      hash = (37 * hash) + COLLECTIVE_NCCL_FIELD_NUMBER;
-      hash = (53 * hash) + com.google.protobuf.Internal.hashBoolean(
-          getCollectiveNccl());
-      hash = (37 * hash) + SHARE_SESSION_STATE_IN_CLUSTERSPEC_PROPAGATION_FIELD_NUMBER;
-      hash = (53 * hash) + com.google.protobuf.Internal.hashBoolean(
-          getShareSessionStateInClusterspecPropagation());
-      hash = (37 * hash) + DISABLE_THREAD_SPINNING_FIELD_NUMBER;
-      hash = (53 * hash) + com.google.protobuf.Internal.hashBoolean(
-          getDisableThreadSpinning());
-      hash = (37 * hash) + SHARE_CLUSTER_DEVICES_IN_SESSION_FIELD_NUMBER;
-      hash = (53 * hash) + com.google.protobuf.Internal.hashBoolean(
-          getShareClusterDevicesInSession());
-      if (hasSessionMetadata()) {
-        hash = (37 * hash) + SESSION_METADATA_FIELD_NUMBER;
-        hash = (53 * hash) + getSessionMetadata().hashCode();
-      }
-      hash = (37 * hash) + OPTIMIZE_FOR_STATIC_GRAPH_FIELD_NUMBER;
-      hash = (53 * hash) + com.google.protobuf.Internal.hashBoolean(
-          getOptimizeForStaticGraph());
-      hash = (37 * hash) + ENABLE_MLIR_BRIDGE_FIELD_NUMBER;
-      hash = (53 * hash) + com.google.protobuf.Internal.hashBoolean(
-          getEnableMlirBridge());
-      hash = (37 * hash) + ENABLE_MLIR_GRAPH_OPTIMIZATION_FIELD_NUMBER;
-      hash = (53 * hash) + com.google.protobuf.Internal.hashBoolean(
-          getEnableMlirGraphOptimization());
-      hash = (37 * hash) + DISABLE_OUTPUT_PARTITION_GRAPHS_FIELD_NUMBER;
-      hash = (53 * hash) + com.google.protobuf.Internal.hashBoolean(
-          getDisableOutputPartitionGraphs());
-      hash = (37 * hash) + XLA_FUSION_AUTOTUNER_THRESH_FIELD_NUMBER;
-      hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-          getXlaFusionAutotunerThresh());
-      hash = (29 * hash) + unknownFields.hashCode();
-      memoizedHashCode = hash;
-      return hash;
-    }
-
-    public static org.tensorflow.proto.framework.ConfigProto.Experimental parseFrom(
-        java.nio.ByteBuffer data)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data);
-    }
-    public static org.tensorflow.proto.framework.ConfigProto.Experimental parseFrom(
-        java.nio.ByteBuffer data,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.ConfigProto.Experimental parseFrom(
-        com.google.protobuf.ByteString data)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data);
-    }
-    public static org.tensorflow.proto.framework.ConfigProto.Experimental parseFrom(
-        com.google.protobuf.ByteString data,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.ConfigProto.Experimental parseFrom(byte[] data)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data);
-    }
-    public static org.tensorflow.proto.framework.ConfigProto.Experimental parseFrom(
-        byte[] data,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.ConfigProto.Experimental parseFrom(java.io.InputStream input)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input);
-    }
-    public static org.tensorflow.proto.framework.ConfigProto.Experimental parseFrom(
-        java.io.InputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.ConfigProto.Experimental parseDelimitedFrom(java.io.InputStream input)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseDelimitedWithIOException(PARSER, input);
-    }
-    public static org.tensorflow.proto.framework.ConfigProto.Experimental parseDelimitedFrom(
-        java.io.InputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.ConfigProto.Experimental parseFrom(
-        com.google.protobuf.CodedInputStream input)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input);
-    }
-    public static org.tensorflow.proto.framework.ConfigProto.Experimental parseFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input, extensionRegistry);
-    }
-
-    @java.lang.Override
-    public Builder newBuilderForType() { return newBuilder(); }
-    public static Builder newBuilder() {
-      return DEFAULT_INSTANCE.toBuilder();
-    }
-    public static Builder newBuilder(org.tensorflow.proto.framework.ConfigProto.Experimental prototype) {
-      return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-    }
-    @java.lang.Override
-    public Builder toBuilder() {
-      return this == DEFAULT_INSTANCE
-          ? new Builder() : new Builder().mergeFrom(this);
-    }
-
-    @java.lang.Override
-    protected Builder newBuilderForType(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      Builder builder = new Builder(parent);
-      return builder;
-    }
-    /**
-     * <pre>
-     * Everything inside Experimental is subject to change and is not subject
-     * to API stability guarantees in
-     * https://www.tensorflow.org/guide/version_compat.
-     * </pre>
-     *
-     * Protobuf type {@code tensorflow.ConfigProto.Experimental}
-     */
-    public static final class Builder extends
-        com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-        // @@protoc_insertion_point(builder_implements:tensorflow.ConfigProto.Experimental)
-        org.tensorflow.proto.framework.ConfigProto.ExperimentalOrBuilder {
-      public static final com.google.protobuf.Descriptors.Descriptor
-          getDescriptor() {
-        return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_ConfigProto_Experimental_descriptor;
-      }
-
-      @java.lang.Override
-      protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-          internalGetFieldAccessorTable() {
-        return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_ConfigProto_Experimental_fieldAccessorTable
-            .ensureFieldAccessorsInitialized(
-                org.tensorflow.proto.framework.ConfigProto.Experimental.class, org.tensorflow.proto.framework.ConfigProto.Experimental.Builder.class);
-      }
-
-      // Construct using org.tensorflow.proto.framework.ConfigProto.Experimental.newBuilder()
-      private Builder() {
-        maybeForceBuilderInitialization();
-      }
-
-      private Builder(
-          com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-        super(parent);
-        maybeForceBuilderInitialization();
-      }
-      private void maybeForceBuilderInitialization() {
-        if (com.google.protobuf.GeneratedMessageV3
-                .alwaysUseFieldBuilders) {
-        }
-      }
-      @java.lang.Override
-      public Builder clear() {
-        super.clear();
-        collectiveGroupLeader_ = "";
-
-        executorType_ = "";
-
-        recvBufMaxChunk_ = 0;
-
-        useNumaAffinity_ = false;
-
-        collectiveDeterministicSequentialExecution_ = false;
-
-        collectiveNccl_ = false;
-
-        shareSessionStateInClusterspecPropagation_ = false;
-
-        disableThreadSpinning_ = false;
-
-        shareClusterDevicesInSession_ = false;
-
-        if (sessionMetadataBuilder_ == null) {
-          sessionMetadata_ = null;
-        } else {
-          sessionMetadata_ = null;
-          sessionMetadataBuilder_ = null;
-        }
-        optimizeForStaticGraph_ = false;
-
-        enableMlirBridge_ = false;
-
-        enableMlirGraphOptimization_ = false;
-
-        disableOutputPartitionGraphs_ = false;
-
-        xlaFusionAutotunerThresh_ = 0L;
-
-        return this;
-      }
-
-      @java.lang.Override
-      public com.google.protobuf.Descriptors.Descriptor
-          getDescriptorForType() {
-        return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_ConfigProto_Experimental_descriptor;
-      }
-
-      @java.lang.Override
-      public org.tensorflow.proto.framework.ConfigProto.Experimental getDefaultInstanceForType() {
-        return org.tensorflow.proto.framework.ConfigProto.Experimental.getDefaultInstance();
-      }
-
-      @java.lang.Override
-      public org.tensorflow.proto.framework.ConfigProto.Experimental build() {
-        org.tensorflow.proto.framework.ConfigProto.Experimental result = buildPartial();
-        if (!result.isInitialized()) {
-          throw newUninitializedMessageException(result);
-        }
-        return result;
-      }
-
-      @java.lang.Override
-      public org.tensorflow.proto.framework.ConfigProto.Experimental buildPartial() {
-        org.tensorflow.proto.framework.ConfigProto.Experimental result = new org.tensorflow.proto.framework.ConfigProto.Experimental(this);
-        result.collectiveGroupLeader_ = collectiveGroupLeader_;
-        result.executorType_ = executorType_;
-        result.recvBufMaxChunk_ = recvBufMaxChunk_;
-        result.useNumaAffinity_ = useNumaAffinity_;
-        result.collectiveDeterministicSequentialExecution_ = collectiveDeterministicSequentialExecution_;
-        result.collectiveNccl_ = collectiveNccl_;
-        result.shareSessionStateInClusterspecPropagation_ = shareSessionStateInClusterspecPropagation_;
-        result.disableThreadSpinning_ = disableThreadSpinning_;
-        result.shareClusterDevicesInSession_ = shareClusterDevicesInSession_;
-        if (sessionMetadataBuilder_ == null) {
-          result.sessionMetadata_ = sessionMetadata_;
-        } else {
-          result.sessionMetadata_ = sessionMetadataBuilder_.build();
-        }
-        result.optimizeForStaticGraph_ = optimizeForStaticGraph_;
-        result.enableMlirBridge_ = enableMlirBridge_;
-        result.enableMlirGraphOptimization_ = enableMlirGraphOptimization_;
-        result.disableOutputPartitionGraphs_ = disableOutputPartitionGraphs_;
-        result.xlaFusionAutotunerThresh_ = xlaFusionAutotunerThresh_;
-        onBuilt();
-        return result;
-      }
-
-      @java.lang.Override
-      public Builder clone() {
-        return super.clone();
-      }
-      @java.lang.Override
-      public Builder setField(
-          com.google.protobuf.Descriptors.FieldDescriptor field,
-          java.lang.Object value) {
-        return super.setField(field, value);
-      }
-      @java.lang.Override
-      public Builder clearField(
-          com.google.protobuf.Descriptors.FieldDescriptor field) {
-        return super.clearField(field);
-      }
-      @java.lang.Override
-      public Builder clearOneof(
-          com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-        return super.clearOneof(oneof);
-      }
-      @java.lang.Override
-      public Builder setRepeatedField(
-          com.google.protobuf.Descriptors.FieldDescriptor field,
-          int index, java.lang.Object value) {
-        return super.setRepeatedField(field, index, value);
-      }
-      @java.lang.Override
-      public Builder addRepeatedField(
-          com.google.protobuf.Descriptors.FieldDescriptor field,
-          java.lang.Object value) {
-        return super.addRepeatedField(field, value);
-      }
-      @java.lang.Override
-      public Builder mergeFrom(com.google.protobuf.Message other) {
-        if (other instanceof org.tensorflow.proto.framework.ConfigProto.Experimental) {
-          return mergeFrom((org.tensorflow.proto.framework.ConfigProto.Experimental)other);
-        } else {
-          super.mergeFrom(other);
-          return this;
-        }
-      }
-
-      public Builder mergeFrom(org.tensorflow.proto.framework.ConfigProto.Experimental other) {
-        if (other == org.tensorflow.proto.framework.ConfigProto.Experimental.getDefaultInstance()) return this;
-        if (!other.getCollectiveGroupLeader().isEmpty()) {
-          collectiveGroupLeader_ = other.collectiveGroupLeader_;
-          onChanged();
-        }
-        if (!other.getExecutorType().isEmpty()) {
-          executorType_ = other.executorType_;
-          onChanged();
-        }
-        if (other.getRecvBufMaxChunk() != 0) {
-          setRecvBufMaxChunk(other.getRecvBufMaxChunk());
-        }
-        if (other.getUseNumaAffinity() != false) {
-          setUseNumaAffinity(other.getUseNumaAffinity());
-        }
-        if (other.getCollectiveDeterministicSequentialExecution() != false) {
-          setCollectiveDeterministicSequentialExecution(other.getCollectiveDeterministicSequentialExecution());
-        }
-        if (other.getCollectiveNccl() != false) {
-          setCollectiveNccl(other.getCollectiveNccl());
-        }
-        if (other.getShareSessionStateInClusterspecPropagation() != false) {
-          setShareSessionStateInClusterspecPropagation(other.getShareSessionStateInClusterspecPropagation());
-        }
-        if (other.getDisableThreadSpinning() != false) {
-          setDisableThreadSpinning(other.getDisableThreadSpinning());
-        }
-        if (other.getShareClusterDevicesInSession() != false) {
-          setShareClusterDevicesInSession(other.getShareClusterDevicesInSession());
-        }
-        if (other.hasSessionMetadata()) {
-          mergeSessionMetadata(other.getSessionMetadata());
-        }
-        if (other.getOptimizeForStaticGraph() != false) {
-          setOptimizeForStaticGraph(other.getOptimizeForStaticGraph());
-        }
-        if (other.getEnableMlirBridge() != false) {
-          setEnableMlirBridge(other.getEnableMlirBridge());
-        }
-        if (other.getEnableMlirGraphOptimization() != false) {
-          setEnableMlirGraphOptimization(other.getEnableMlirGraphOptimization());
-        }
-        if (other.getDisableOutputPartitionGraphs() != false) {
-          setDisableOutputPartitionGraphs(other.getDisableOutputPartitionGraphs());
-        }
-        if (other.getXlaFusionAutotunerThresh() != 0L) {
-          setXlaFusionAutotunerThresh(other.getXlaFusionAutotunerThresh());
-        }
-        this.mergeUnknownFields(other.unknownFields);
-        onChanged();
-        return this;
-      }
-
-      @java.lang.Override
-      public final boolean isInitialized() {
-        return true;
-      }
-
-      @java.lang.Override
-      public Builder mergeFrom(
-          com.google.protobuf.CodedInputStream input,
-          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-          throws java.io.IOException {
-        org.tensorflow.proto.framework.ConfigProto.Experimental parsedMessage = null;
-        try {
-          parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-        } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-          parsedMessage = (org.tensorflow.proto.framework.ConfigProto.Experimental) e.getUnfinishedMessage();
-          throw e.unwrapIOException();
-        } finally {
-          if (parsedMessage != null) {
-            mergeFrom(parsedMessage);
-          }
-        }
-        return this;
-      }
-
-      private java.lang.Object collectiveGroupLeader_ = "";
-      /**
-       * <pre>
-       * Task name for group resolution.
-       * </pre>
-       *
-       * <code>string collective_group_leader = 1;</code>
-       */
-      public java.lang.String getCollectiveGroupLeader() {
-        java.lang.Object ref = collectiveGroupLeader_;
-        if (!(ref instanceof java.lang.String)) {
-          com.google.protobuf.ByteString bs =
-              (com.google.protobuf.ByteString) ref;
-          java.lang.String s = bs.toStringUtf8();
-          collectiveGroupLeader_ = s;
-          return s;
-        } else {
-          return (java.lang.String) ref;
-        }
-      }
-      /**
-       * <pre>
-       * Task name for group resolution.
-       * </pre>
-       *
-       * <code>string collective_group_leader = 1;</code>
-       */
-      public com.google.protobuf.ByteString
-          getCollectiveGroupLeaderBytes() {
-        java.lang.Object ref = collectiveGroupLeader_;
-        if (ref instanceof String) {
-          com.google.protobuf.ByteString b = 
-              com.google.protobuf.ByteString.copyFromUtf8(
-                  (java.lang.String) ref);
-          collectiveGroupLeader_ = b;
-          return b;
-        } else {
-          return (com.google.protobuf.ByteString) ref;
-        }
-      }
-      /**
-       * <pre>
-       * Task name for group resolution.
-       * </pre>
-       *
-       * <code>string collective_group_leader = 1;</code>
-       */
-      public Builder setCollectiveGroupLeader(
-          java.lang.String value) {
-        if (value == null) {
-    throw new NullPointerException();
-  }
-  
-        collectiveGroupLeader_ = value;
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * Task name for group resolution.
-       * </pre>
-       *
-       * <code>string collective_group_leader = 1;</code>
-       */
-      public Builder clearCollectiveGroupLeader() {
-        
-        collectiveGroupLeader_ = getDefaultInstance().getCollectiveGroupLeader();
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * Task name for group resolution.
-       * </pre>
-       *
-       * <code>string collective_group_leader = 1;</code>
-       */
-      public Builder setCollectiveGroupLeaderBytes(
-          com.google.protobuf.ByteString value) {
-        if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-        
-        collectiveGroupLeader_ = value;
-        onChanged();
-        return this;
-      }
-
-      private java.lang.Object executorType_ = "";
-      /**
-       * <pre>
-       * Which executor to use, the default executor will be used
-       * if it is an empty string or "DEFAULT"
-       * </pre>
-       *
-       * <code>string executor_type = 3;</code>
-       */
-      public java.lang.String getExecutorType() {
-        java.lang.Object ref = executorType_;
-        if (!(ref instanceof java.lang.String)) {
-          com.google.protobuf.ByteString bs =
-              (com.google.protobuf.ByteString) ref;
-          java.lang.String s = bs.toStringUtf8();
-          executorType_ = s;
-          return s;
-        } else {
-          return (java.lang.String) ref;
-        }
-      }
-      /**
-       * <pre>
-       * Which executor to use, the default executor will be used
-       * if it is an empty string or "DEFAULT"
-       * </pre>
-       *
-       * <code>string executor_type = 3;</code>
-       */
-      public com.google.protobuf.ByteString
-          getExecutorTypeBytes() {
-        java.lang.Object ref = executorType_;
-        if (ref instanceof String) {
-          com.google.protobuf.ByteString b = 
-              com.google.protobuf.ByteString.copyFromUtf8(
-                  (java.lang.String) ref);
-          executorType_ = b;
-          return b;
-        } else {
-          return (com.google.protobuf.ByteString) ref;
-        }
-      }
-      /**
-       * <pre>
-       * Which executor to use, the default executor will be used
-       * if it is an empty string or "DEFAULT"
-       * </pre>
-       *
-       * <code>string executor_type = 3;</code>
-       */
-      public Builder setExecutorType(
-          java.lang.String value) {
-        if (value == null) {
-    throw new NullPointerException();
-  }
-  
-        executorType_ = value;
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * Which executor to use, the default executor will be used
-       * if it is an empty string or "DEFAULT"
-       * </pre>
-       *
-       * <code>string executor_type = 3;</code>
-       */
-      public Builder clearExecutorType() {
-        
-        executorType_ = getDefaultInstance().getExecutorType();
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * Which executor to use, the default executor will be used
-       * if it is an empty string or "DEFAULT"
-       * </pre>
-       *
-       * <code>string executor_type = 3;</code>
-       */
-      public Builder setExecutorTypeBytes(
-          com.google.protobuf.ByteString value) {
-        if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-        
-        executorType_ = value;
-        onChanged();
-        return this;
-      }
-
-      private int recvBufMaxChunk_ ;
-      /**
-       * <pre>
-       * Guidance to formatting of large RecvBuf fields for transfer.
-       * Any positive value sets the max chunk size.  0 defaults to 4096.
-       * Any negative value indicates no max, i.e. one chunk only.
-       * </pre>
-       *
-       * <code>int32 recv_buf_max_chunk = 4;</code>
-       */
-      public int getRecvBufMaxChunk() {
-        return recvBufMaxChunk_;
-      }
-      /**
-       * <pre>
-       * Guidance to formatting of large RecvBuf fields for transfer.
-       * Any positive value sets the max chunk size.  0 defaults to 4096.
-       * Any negative value indicates no max, i.e. one chunk only.
-       * </pre>
-       *
-       * <code>int32 recv_buf_max_chunk = 4;</code>
-       */
-      public Builder setRecvBufMaxChunk(int value) {
-        
-        recvBufMaxChunk_ = value;
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * Guidance to formatting of large RecvBuf fields for transfer.
-       * Any positive value sets the max chunk size.  0 defaults to 4096.
-       * Any negative value indicates no max, i.e. one chunk only.
-       * </pre>
-       *
-       * <code>int32 recv_buf_max_chunk = 4;</code>
-       */
-      public Builder clearRecvBufMaxChunk() {
-        
-        recvBufMaxChunk_ = 0;
-        onChanged();
-        return this;
-      }
-
-      private boolean useNumaAffinity_ ;
-      /**
-       * <pre>
-       * If true, and supported by the platform, the runtime will attempt to
-       * use NUMA affinity where applicable.  One consequence will be the
-       * existence of as many CPU devices as there are available NUMA nodes.
-       * </pre>
-       *
-       * <code>bool use_numa_affinity = 5;</code>
-       */
-      public boolean getUseNumaAffinity() {
-        return useNumaAffinity_;
-      }
-      /**
-       * <pre>
-       * If true, and supported by the platform, the runtime will attempt to
-       * use NUMA affinity where applicable.  One consequence will be the
-       * existence of as many CPU devices as there are available NUMA nodes.
-       * </pre>
-       *
-       * <code>bool use_numa_affinity = 5;</code>
-       */
-      public Builder setUseNumaAffinity(boolean value) {
-        
-        useNumaAffinity_ = value;
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * If true, and supported by the platform, the runtime will attempt to
-       * use NUMA affinity where applicable.  One consequence will be the
-       * existence of as many CPU devices as there are available NUMA nodes.
-       * </pre>
-       *
-       * <code>bool use_numa_affinity = 5;</code>
-       */
-      public Builder clearUseNumaAffinity() {
-        
-        useNumaAffinity_ = false;
-        onChanged();
-        return this;
-      }
-
-      private boolean collectiveDeterministicSequentialExecution_ ;
-      /**
-       * <pre>
-       * If true, make collective op execution order sequential and deterministic
-       * for potentially concurrent collective instances.
-       * </pre>
-       *
-       * <code>bool collective_deterministic_sequential_execution = 6;</code>
-       */
-      public boolean getCollectiveDeterministicSequentialExecution() {
-        return collectiveDeterministicSequentialExecution_;
-      }
-      /**
-       * <pre>
-       * If true, make collective op execution order sequential and deterministic
-       * for potentially concurrent collective instances.
-       * </pre>
-       *
-       * <code>bool collective_deterministic_sequential_execution = 6;</code>
-       */
-      public Builder setCollectiveDeterministicSequentialExecution(boolean value) {
-        
-        collectiveDeterministicSequentialExecution_ = value;
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * If true, make collective op execution order sequential and deterministic
-       * for potentially concurrent collective instances.
-       * </pre>
-       *
-       * <code>bool collective_deterministic_sequential_execution = 6;</code>
-       */
-      public Builder clearCollectiveDeterministicSequentialExecution() {
-        
-        collectiveDeterministicSequentialExecution_ = false;
-        onChanged();
-        return this;
-      }
-
-      private boolean collectiveNccl_ ;
-      /**
-       * <pre>
-       * If true, use NCCL for CollectiveOps.  This feature is highly
-       * experimental.
-       * </pre>
-       *
-       * <code>bool collective_nccl = 7;</code>
-       */
-      public boolean getCollectiveNccl() {
-        return collectiveNccl_;
-      }
-      /**
-       * <pre>
-       * If true, use NCCL for CollectiveOps.  This feature is highly
-       * experimental.
-       * </pre>
-       *
-       * <code>bool collective_nccl = 7;</code>
-       */
-      public Builder setCollectiveNccl(boolean value) {
-        
-        collectiveNccl_ = value;
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * If true, use NCCL for CollectiveOps.  This feature is highly
-       * experimental.
-       * </pre>
-       *
-       * <code>bool collective_nccl = 7;</code>
-       */
-      public Builder clearCollectiveNccl() {
-        
-        collectiveNccl_ = false;
-        onChanged();
-        return this;
-      }
-
-      private boolean shareSessionStateInClusterspecPropagation_ ;
-      /**
-       * <pre>
-       * In the following, session state means the value of a variable, elements
-       * in a hash table, or any other resource, accessible by worker sessions
-       * held by a TF server.
-       * When ClusterSpec propagation is enabled, the value of
-       * isolate_session_state is ignored when deciding whether to share session
-       * states in a TF server (for backwards compatibility reasons).
-       * - If share_session_state_in_clusterspec_propagation is true, the session
-       * states are shared.
-       * - If share_session_state_in_clusterspec_propagation is false, session
-       * states are isolated.
-       * When clusterspec propagation is not used, the value of
-       * share_session_state_in_clusterspec_propagation is ignored when deciding
-       * whether to share session states in a TF server.
-       * - If isolate_session_state is true, session states are isolated.
-       * - If isolate_session_state is false, session states are shared.
-       * TODO(b/129330037): Add a single API that consistently treats
-       * isolate_session_state and ClusterSpec propagation.
-       * </pre>
-       *
-       * <code>bool share_session_state_in_clusterspec_propagation = 8;</code>
-       */
-      public boolean getShareSessionStateInClusterspecPropagation() {
-        return shareSessionStateInClusterspecPropagation_;
-      }
-      /**
-       * <pre>
-       * In the following, session state means the value of a variable, elements
-       * in a hash table, or any other resource, accessible by worker sessions
-       * held by a TF server.
-       * When ClusterSpec propagation is enabled, the value of
-       * isolate_session_state is ignored when deciding whether to share session
-       * states in a TF server (for backwards compatibility reasons).
-       * - If share_session_state_in_clusterspec_propagation is true, the session
-       * states are shared.
-       * - If share_session_state_in_clusterspec_propagation is false, session
-       * states are isolated.
-       * When clusterspec propagation is not used, the value of
-       * share_session_state_in_clusterspec_propagation is ignored when deciding
-       * whether to share session states in a TF server.
-       * - If isolate_session_state is true, session states are isolated.
-       * - If isolate_session_state is false, session states are shared.
-       * TODO(b/129330037): Add a single API that consistently treats
-       * isolate_session_state and ClusterSpec propagation.
-       * </pre>
-       *
-       * <code>bool share_session_state_in_clusterspec_propagation = 8;</code>
-       */
-      public Builder setShareSessionStateInClusterspecPropagation(boolean value) {
-        
-        shareSessionStateInClusterspecPropagation_ = value;
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * In the following, session state means the value of a variable, elements
-       * in a hash table, or any other resource, accessible by worker sessions
-       * held by a TF server.
-       * When ClusterSpec propagation is enabled, the value of
-       * isolate_session_state is ignored when deciding whether to share session
-       * states in a TF server (for backwards compatibility reasons).
-       * - If share_session_state_in_clusterspec_propagation is true, the session
-       * states are shared.
-       * - If share_session_state_in_clusterspec_propagation is false, session
-       * states are isolated.
-       * When clusterspec propagation is not used, the value of
-       * share_session_state_in_clusterspec_propagation is ignored when deciding
-       * whether to share session states in a TF server.
-       * - If isolate_session_state is true, session states are isolated.
-       * - If isolate_session_state is false, session states are shared.
-       * TODO(b/129330037): Add a single API that consistently treats
-       * isolate_session_state and ClusterSpec propagation.
-       * </pre>
-       *
-       * <code>bool share_session_state_in_clusterspec_propagation = 8;</code>
-       */
-      public Builder clearShareSessionStateInClusterspecPropagation() {
-        
-        shareSessionStateInClusterspecPropagation_ = false;
-        onChanged();
-        return this;
-      }
-
-      private boolean disableThreadSpinning_ ;
-      /**
-       * <pre>
-       * If using a direct session, disable spinning while waiting for work in
-       * the thread pool. This may result in higher latency for completing ops,
-       * but in the case where there is a lot of spinning may result in lower
-       * CPU usage.
-       * </pre>
-       *
-       * <code>bool disable_thread_spinning = 9;</code>
-       */
-      public boolean getDisableThreadSpinning() {
-        return disableThreadSpinning_;
-      }
-      /**
-       * <pre>
-       * If using a direct session, disable spinning while waiting for work in
-       * the thread pool. This may result in higher latency for completing ops,
-       * but in the case where there is a lot of spinning may result in lower
-       * CPU usage.
-       * </pre>
-       *
-       * <code>bool disable_thread_spinning = 9;</code>
-       */
-      public Builder setDisableThreadSpinning(boolean value) {
-        
-        disableThreadSpinning_ = value;
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * If using a direct session, disable spinning while waiting for work in
-       * the thread pool. This may result in higher latency for completing ops,
-       * but in the case where there is a lot of spinning may result in lower
-       * CPU usage.
-       * </pre>
-       *
-       * <code>bool disable_thread_spinning = 9;</code>
-       */
-      public Builder clearDisableThreadSpinning() {
-        
-        disableThreadSpinning_ = false;
-        onChanged();
-        return this;
-      }
-
-      private boolean shareClusterDevicesInSession_ ;
-      /**
-       * <pre>
-       * This was promoted to a non-experimental API. Please use
-       * ConfigProto.share_cluster_devices_in_session instead.
-       * </pre>
-       *
-       * <code>bool share_cluster_devices_in_session = 10;</code>
-       */
-      public boolean getShareClusterDevicesInSession() {
-        return shareClusterDevicesInSession_;
-      }
-      /**
-       * <pre>
-       * This was promoted to a non-experimental API. Please use
-       * ConfigProto.share_cluster_devices_in_session instead.
-       * </pre>
-       *
-       * <code>bool share_cluster_devices_in_session = 10;</code>
-       */
-      public Builder setShareClusterDevicesInSession(boolean value) {
-        
-        shareClusterDevicesInSession_ = value;
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * This was promoted to a non-experimental API. Please use
-       * ConfigProto.share_cluster_devices_in_session instead.
-       * </pre>
-       *
-       * <code>bool share_cluster_devices_in_session = 10;</code>
-       */
-      public Builder clearShareClusterDevicesInSession() {
-        
-        shareClusterDevicesInSession_ = false;
-        onChanged();
-        return this;
-      }
-
-      private org.tensorflow.proto.framework.SessionMetadata sessionMetadata_;
-      private com.google.protobuf.SingleFieldBuilderV3<
-          org.tensorflow.proto.framework.SessionMetadata, org.tensorflow.proto.framework.SessionMetadata.Builder, org.tensorflow.proto.framework.SessionMetadataOrBuilder> sessionMetadataBuilder_;
-      /**
-       * <pre>
-       * Metadata about the session.
-       * If set, this can be used by the runtime and the Ops for debugging,
-       * monitoring, etc.
-       * NOTE: This is currently used and propagated only by the direct session.
-       * </pre>
-       *
-       * <code>.tensorflow.SessionMetadata session_metadata = 11;</code>
-       */
-      public boolean hasSessionMetadata() {
-        return sessionMetadataBuilder_ != null || sessionMetadata_ != null;
-      }
-      /**
-       * <pre>
-       * Metadata about the session.
-       * If set, this can be used by the runtime and the Ops for debugging,
-       * monitoring, etc.
-       * NOTE: This is currently used and propagated only by the direct session.
-       * </pre>
-       *
-       * <code>.tensorflow.SessionMetadata session_metadata = 11;</code>
-       */
-      public org.tensorflow.proto.framework.SessionMetadata getSessionMetadata() {
-        if (sessionMetadataBuilder_ == null) {
-          return sessionMetadata_ == null ? org.tensorflow.proto.framework.SessionMetadata.getDefaultInstance() : sessionMetadata_;
-        } else {
-          return sessionMetadataBuilder_.getMessage();
-        }
-      }
-      /**
-       * <pre>
-       * Metadata about the session.
-       * If set, this can be used by the runtime and the Ops for debugging,
-       * monitoring, etc.
-       * NOTE: This is currently used and propagated only by the direct session.
-       * </pre>
-       *
-       * <code>.tensorflow.SessionMetadata session_metadata = 11;</code>
-       */
-      public Builder setSessionMetadata(org.tensorflow.proto.framework.SessionMetadata value) {
-        if (sessionMetadataBuilder_ == null) {
-          if (value == null) {
-            throw new NullPointerException();
-          }
-          sessionMetadata_ = value;
-          onChanged();
-        } else {
-          sessionMetadataBuilder_.setMessage(value);
-        }
-
-        return this;
-      }
-      /**
-       * <pre>
-       * Metadata about the session.
-       * If set, this can be used by the runtime and the Ops for debugging,
-       * monitoring, etc.
-       * NOTE: This is currently used and propagated only by the direct session.
-       * </pre>
-       *
-       * <code>.tensorflow.SessionMetadata session_metadata = 11;</code>
-       */
-      public Builder setSessionMetadata(
-          org.tensorflow.proto.framework.SessionMetadata.Builder builderForValue) {
-        if (sessionMetadataBuilder_ == null) {
-          sessionMetadata_ = builderForValue.build();
-          onChanged();
-        } else {
-          sessionMetadataBuilder_.setMessage(builderForValue.build());
-        }
-
-        return this;
-      }
-      /**
-       * <pre>
-       * Metadata about the session.
-       * If set, this can be used by the runtime and the Ops for debugging,
-       * monitoring, etc.
-       * NOTE: This is currently used and propagated only by the direct session.
-       * </pre>
-       *
-       * <code>.tensorflow.SessionMetadata session_metadata = 11;</code>
-       */
-      public Builder mergeSessionMetadata(org.tensorflow.proto.framework.SessionMetadata value) {
-        if (sessionMetadataBuilder_ == null) {
-          if (sessionMetadata_ != null) {
-            sessionMetadata_ =
-              org.tensorflow.proto.framework.SessionMetadata.newBuilder(sessionMetadata_).mergeFrom(value).buildPartial();
-          } else {
-            sessionMetadata_ = value;
-          }
-          onChanged();
-        } else {
-          sessionMetadataBuilder_.mergeFrom(value);
-        }
-
-        return this;
-      }
-      /**
-       * <pre>
-       * Metadata about the session.
-       * If set, this can be used by the runtime and the Ops for debugging,
-       * monitoring, etc.
-       * NOTE: This is currently used and propagated only by the direct session.
-       * </pre>
-       *
-       * <code>.tensorflow.SessionMetadata session_metadata = 11;</code>
-       */
-      public Builder clearSessionMetadata() {
-        if (sessionMetadataBuilder_ == null) {
-          sessionMetadata_ = null;
-          onChanged();
-        } else {
-          sessionMetadata_ = null;
-          sessionMetadataBuilder_ = null;
-        }
-
-        return this;
-      }
-      /**
-       * <pre>
-       * Metadata about the session.
-       * If set, this can be used by the runtime and the Ops for debugging,
-       * monitoring, etc.
-       * NOTE: This is currently used and propagated only by the direct session.
-       * </pre>
-       *
-       * <code>.tensorflow.SessionMetadata session_metadata = 11;</code>
-       */
-      public org.tensorflow.proto.framework.SessionMetadata.Builder getSessionMetadataBuilder() {
-        
-        onChanged();
-        return getSessionMetadataFieldBuilder().getBuilder();
-      }
-      /**
-       * <pre>
-       * Metadata about the session.
-       * If set, this can be used by the runtime and the Ops for debugging,
-       * monitoring, etc.
-       * NOTE: This is currently used and propagated only by the direct session.
-       * </pre>
-       *
-       * <code>.tensorflow.SessionMetadata session_metadata = 11;</code>
-       */
-      public org.tensorflow.proto.framework.SessionMetadataOrBuilder getSessionMetadataOrBuilder() {
-        if (sessionMetadataBuilder_ != null) {
-          return sessionMetadataBuilder_.getMessageOrBuilder();
-        } else {
-          return sessionMetadata_ == null ?
-              org.tensorflow.proto.framework.SessionMetadata.getDefaultInstance() : sessionMetadata_;
-        }
-      }
-      /**
-       * <pre>
-       * Metadata about the session.
-       * If set, this can be used by the runtime and the Ops for debugging,
-       * monitoring, etc.
-       * NOTE: This is currently used and propagated only by the direct session.
-       * </pre>
-       *
-       * <code>.tensorflow.SessionMetadata session_metadata = 11;</code>
-       */
-      private com.google.protobuf.SingleFieldBuilderV3<
-          org.tensorflow.proto.framework.SessionMetadata, org.tensorflow.proto.framework.SessionMetadata.Builder, org.tensorflow.proto.framework.SessionMetadataOrBuilder> 
-          getSessionMetadataFieldBuilder() {
-        if (sessionMetadataBuilder_ == null) {
-          sessionMetadataBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-              org.tensorflow.proto.framework.SessionMetadata, org.tensorflow.proto.framework.SessionMetadata.Builder, org.tensorflow.proto.framework.SessionMetadataOrBuilder>(
-                  getSessionMetadata(),
-                  getParentForChildren(),
-                  isClean());
-          sessionMetadata_ = null;
-        }
-        return sessionMetadataBuilder_;
-      }
-
-      private boolean optimizeForStaticGraph_ ;
-      /**
-       * <pre>
-       * If true, the session may treat the graph as being static for optimization
-       * purposes.
-       * If this option is set to true when a session is created, the full
-       * GraphDef must be passed in a single call to Session::Create(), and
-       * Session::Extend() may not be supported.
-       * </pre>
-       *
-       * <code>bool optimize_for_static_graph = 12;</code>
-       */
-      public boolean getOptimizeForStaticGraph() {
-        return optimizeForStaticGraph_;
-      }
-      /**
-       * <pre>
-       * If true, the session may treat the graph as being static for optimization
-       * purposes.
-       * If this option is set to true when a session is created, the full
-       * GraphDef must be passed in a single call to Session::Create(), and
-       * Session::Extend() may not be supported.
-       * </pre>
-       *
-       * <code>bool optimize_for_static_graph = 12;</code>
-       */
-      public Builder setOptimizeForStaticGraph(boolean value) {
-        
-        optimizeForStaticGraph_ = value;
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * If true, the session may treat the graph as being static for optimization
-       * purposes.
-       * If this option is set to true when a session is created, the full
-       * GraphDef must be passed in a single call to Session::Create(), and
-       * Session::Extend() may not be supported.
-       * </pre>
-       *
-       * <code>bool optimize_for_static_graph = 12;</code>
-       */
-      public Builder clearOptimizeForStaticGraph() {
-        
-        optimizeForStaticGraph_ = false;
-        onChanged();
-        return this;
-      }
-
-      private boolean enableMlirBridge_ ;
-      /**
-       * <pre>
-       * Whether to enable the MLIR-based TF-&gt;XLA bridge.
-       * This is a replacement to the existing bridge, and not ready for
-       * production usage yet.
-       * If this option is set to true when a session is created, MLIR is used to
-       * perform the set of graph transformations to put the graph in a form that
-       * can be executed with delegation of some computations to an accelerator.
-       * This builds on the model of XLA where a subset of the graph is
-       * encapsulated and attached to a "compile" operation, whose result is fed
-       * to an "execute" operation. The kernel for these operations is responsible
-       * to lower the encapsulated graph to a particular device.
-       * </pre>
-       *
-       * <code>bool enable_mlir_bridge = 13;</code>
-       */
-      public boolean getEnableMlirBridge() {
-        return enableMlirBridge_;
-      }
-      /**
-       * <pre>
-       * Whether to enable the MLIR-based TF-&gt;XLA bridge.
-       * This is a replacement to the existing bridge, and not ready for
-       * production usage yet.
-       * If this option is set to true when a session is created, MLIR is used to
-       * perform the set of graph transformations to put the graph in a form that
-       * can be executed with delegation of some computations to an accelerator.
-       * This builds on the model of XLA where a subset of the graph is
-       * encapsulated and attached to a "compile" operation, whose result is fed
-       * to an "execute" operation. The kernel for these operations is responsible
-       * to lower the encapsulated graph to a particular device.
-       * </pre>
-       *
-       * <code>bool enable_mlir_bridge = 13;</code>
-       */
-      public Builder setEnableMlirBridge(boolean value) {
-        
-        enableMlirBridge_ = value;
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * Whether to enable the MLIR-based TF-&gt;XLA bridge.
-       * This is a replacement to the existing bridge, and not ready for
-       * production usage yet.
-       * If this option is set to true when a session is created, MLIR is used to
-       * perform the set of graph transformations to put the graph in a form that
-       * can be executed with delegation of some computations to an accelerator.
-       * This builds on the model of XLA where a subset of the graph is
-       * encapsulated and attached to a "compile" operation, whose result is fed
-       * to an "execute" operation. The kernel for these operations is responsible
-       * to lower the encapsulated graph to a particular device.
-       * </pre>
-       *
-       * <code>bool enable_mlir_bridge = 13;</code>
-       */
-      public Builder clearEnableMlirBridge() {
-        
-        enableMlirBridge_ = false;
-        onChanged();
-        return this;
-      }
-
-      private boolean enableMlirGraphOptimization_ ;
-      /**
-       * <pre>
-       * Whether to enable the MLIR-based Graph optimizations.
-       * This will become a part of standard Tensorflow graph optimization
-       * pipeline, currently this is only used for gradual migration and testing
-       * new passes that are replacing existing optimizations in Grappler.
-       * </pre>
-       *
-       * <code>bool enable_mlir_graph_optimization = 16;</code>
-       */
-      public boolean getEnableMlirGraphOptimization() {
-        return enableMlirGraphOptimization_;
-      }
-      /**
-       * <pre>
-       * Whether to enable the MLIR-based Graph optimizations.
-       * This will become a part of standard Tensorflow graph optimization
-       * pipeline, currently this is only used for gradual migration and testing
-       * new passes that are replacing existing optimizations in Grappler.
-       * </pre>
-       *
-       * <code>bool enable_mlir_graph_optimization = 16;</code>
-       */
-      public Builder setEnableMlirGraphOptimization(boolean value) {
-        
-        enableMlirGraphOptimization_ = value;
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * Whether to enable the MLIR-based Graph optimizations.
-       * This will become a part of standard Tensorflow graph optimization
-       * pipeline, currently this is only used for gradual migration and testing
-       * new passes that are replacing existing optimizations in Grappler.
-       * </pre>
-       *
-       * <code>bool enable_mlir_graph_optimization = 16;</code>
-       */
-      public Builder clearEnableMlirGraphOptimization() {
-        
-        enableMlirGraphOptimization_ = false;
-        onChanged();
-        return this;
-      }
-
-      private boolean disableOutputPartitionGraphs_ ;
-      /**
-       * <pre>
-       * If true, the session will not store an additional copy of the graph for
-       * each subgraph.
-       * If this option is set to true when a session is created, the
-       * `RunOptions.output_partition_graphs` options must not be set.
-       * </pre>
-       *
-       * <code>bool disable_output_partition_graphs = 14;</code>
-       */
-      public boolean getDisableOutputPartitionGraphs() {
-        return disableOutputPartitionGraphs_;
-      }
-      /**
-       * <pre>
-       * If true, the session will not store an additional copy of the graph for
-       * each subgraph.
-       * If this option is set to true when a session is created, the
-       * `RunOptions.output_partition_graphs` options must not be set.
-       * </pre>
-       *
-       * <code>bool disable_output_partition_graphs = 14;</code>
-       */
-      public Builder setDisableOutputPartitionGraphs(boolean value) {
-        
-        disableOutputPartitionGraphs_ = value;
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * If true, the session will not store an additional copy of the graph for
-       * each subgraph.
-       * If this option is set to true when a session is created, the
-       * `RunOptions.output_partition_graphs` options must not be set.
-       * </pre>
-       *
-       * <code>bool disable_output_partition_graphs = 14;</code>
-       */
-      public Builder clearDisableOutputPartitionGraphs() {
-        
-        disableOutputPartitionGraphs_ = false;
-        onChanged();
-        return this;
-      }
-
-      private long xlaFusionAutotunerThresh_ ;
-      /**
-       * <pre>
-       * Minimum number of batches run through the XLA graph before XLA fusion
-       * autotuner is enabled. Default value of zero disables the autotuner.
-       * The XLA fusion autotuner can improve performance by executing a heuristic
-       * search on the compiler parameters.
-       * </pre>
-       *
-       * <code>int64 xla_fusion_autotuner_thresh = 15;</code>
-       */
-      public long getXlaFusionAutotunerThresh() {
-        return xlaFusionAutotunerThresh_;
-      }
-      /**
-       * <pre>
-       * Minimum number of batches run through the XLA graph before XLA fusion
-       * autotuner is enabled. Default value of zero disables the autotuner.
-       * The XLA fusion autotuner can improve performance by executing a heuristic
-       * search on the compiler parameters.
-       * </pre>
-       *
-       * <code>int64 xla_fusion_autotuner_thresh = 15;</code>
-       */
-      public Builder setXlaFusionAutotunerThresh(long value) {
-        
-        xlaFusionAutotunerThresh_ = value;
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * Minimum number of batches run through the XLA graph before XLA fusion
-       * autotuner is enabled. Default value of zero disables the autotuner.
-       * The XLA fusion autotuner can improve performance by executing a heuristic
-       * search on the compiler parameters.
-       * </pre>
-       *
-       * <code>int64 xla_fusion_autotuner_thresh = 15;</code>
-       */
-      public Builder clearXlaFusionAutotunerThresh() {
-        
-        xlaFusionAutotunerThresh_ = 0L;
-        onChanged();
-        return this;
-      }
-      @java.lang.Override
-      public final Builder setUnknownFields(
-          final com.google.protobuf.UnknownFieldSet unknownFields) {
-        return super.setUnknownFields(unknownFields);
-      }
-
-      @java.lang.Override
-      public final Builder mergeUnknownFields(
-          final com.google.protobuf.UnknownFieldSet unknownFields) {
-        return super.mergeUnknownFields(unknownFields);
-      }
-
-
-      // @@protoc_insertion_point(builder_scope:tensorflow.ConfigProto.Experimental)
-    }
-
-    // @@protoc_insertion_point(class_scope:tensorflow.ConfigProto.Experimental)
-    private static final org.tensorflow.proto.framework.ConfigProto.Experimental DEFAULT_INSTANCE;
-    static {
-      DEFAULT_INSTANCE = new org.tensorflow.proto.framework.ConfigProto.Experimental();
-    }
-
-    public static org.tensorflow.proto.framework.ConfigProto.Experimental getDefaultInstance() {
-      return DEFAULT_INSTANCE;
-    }
-
-    private static final com.google.protobuf.Parser<Experimental>
-        PARSER = new com.google.protobuf.AbstractParser<Experimental>() {
-      @java.lang.Override
-      public Experimental parsePartialFrom(
-          com.google.protobuf.CodedInputStream input,
-          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-          throws com.google.protobuf.InvalidProtocolBufferException {
-        return new Experimental(input, extensionRegistry);
-      }
-    };
-
-    public static com.google.protobuf.Parser<Experimental> parser() {
-      return PARSER;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Parser<Experimental> getParserForType() {
-      return PARSER;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.ConfigProto.Experimental getDefaultInstanceForType() {
-      return DEFAULT_INSTANCE;
-    }
-
-  }
-
-  public static final int DEVICE_COUNT_FIELD_NUMBER = 1;
-  private static final class DeviceCountDefaultEntryHolder {
-    static final com.google.protobuf.MapEntry<
-        java.lang.String, java.lang.Integer> defaultEntry =
-            com.google.protobuf.MapEntry
-            .<java.lang.String, java.lang.Integer>newDefaultInstance(
-                org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_ConfigProto_DeviceCountEntry_descriptor, 
-                com.google.protobuf.WireFormat.FieldType.STRING,
-                "",
-                com.google.protobuf.WireFormat.FieldType.INT32,
-                0);
-  }
-  private com.google.protobuf.MapField<
-      java.lang.String, java.lang.Integer> deviceCount_;
-  private com.google.protobuf.MapField<java.lang.String, java.lang.Integer>
-  internalGetDeviceCount() {
-    if (deviceCount_ == null) {
-      return com.google.protobuf.MapField.emptyMapField(
-          DeviceCountDefaultEntryHolder.defaultEntry);
-    }
-    return deviceCount_;
-  }
-
-  public int getDeviceCountCount() {
-    return internalGetDeviceCount().getMap().size();
-  }
-  /**
-   * <pre>
-   * Map from device type name (e.g., "CPU" or "GPU" ) to maximum
-   * number of devices of that type to use.  If a particular device
-   * type is not found in the map, the system picks an appropriate
-   * number.
-   * </pre>
-   *
-   * <code>map&lt;string, int32&gt; device_count = 1;</code>
-   */
-
-  public boolean containsDeviceCount(
-      java.lang.String key) {
-    if (key == null) { throw new java.lang.NullPointerException(); }
-    return internalGetDeviceCount().getMap().containsKey(key);
-  }
-  /**
-   * Use {@link #getDeviceCountMap()} instead.
-   */
-  @java.lang.Deprecated
-  public java.util.Map<java.lang.String, java.lang.Integer> getDeviceCount() {
-    return getDeviceCountMap();
-  }
-  /**
-   * <pre>
-   * Map from device type name (e.g., "CPU" or "GPU" ) to maximum
-   * number of devices of that type to use.  If a particular device
-   * type is not found in the map, the system picks an appropriate
-   * number.
-   * </pre>
-   *
-   * <code>map&lt;string, int32&gt; device_count = 1;</code>
-   */
-
-  public java.util.Map<java.lang.String, java.lang.Integer> getDeviceCountMap() {
-    return internalGetDeviceCount().getMap();
-  }
-  /**
-   * <pre>
-   * Map from device type name (e.g., "CPU" or "GPU" ) to maximum
-   * number of devices of that type to use.  If a particular device
-   * type is not found in the map, the system picks an appropriate
-   * number.
-   * </pre>
-   *
-   * <code>map&lt;string, int32&gt; device_count = 1;</code>
-   */
-
-  public int getDeviceCountOrDefault(
-      java.lang.String key,
-      int defaultValue) {
-    if (key == null) { throw new java.lang.NullPointerException(); }
-    java.util.Map<java.lang.String, java.lang.Integer> map =
-        internalGetDeviceCount().getMap();
-    return map.containsKey(key) ? map.get(key) : defaultValue;
-  }
-  /**
-   * <pre>
-   * Map from device type name (e.g., "CPU" or "GPU" ) to maximum
-   * number of devices of that type to use.  If a particular device
-   * type is not found in the map, the system picks an appropriate
-   * number.
-   * </pre>
-   *
-   * <code>map&lt;string, int32&gt; device_count = 1;</code>
-   */
-
-  public int getDeviceCountOrThrow(
-      java.lang.String key) {
-    if (key == null) { throw new java.lang.NullPointerException(); }
-    java.util.Map<java.lang.String, java.lang.Integer> map =
-        internalGetDeviceCount().getMap();
-    if (!map.containsKey(key)) {
-      throw new java.lang.IllegalArgumentException();
-    }
-    return map.get(key);
-  }
-
-  public static final int INTRA_OP_PARALLELISM_THREADS_FIELD_NUMBER = 2;
-  private int intraOpParallelismThreads_;
-  /**
-   * <pre>
-   * The execution of an individual op (for some op types) can be
-   * parallelized on a pool of intra_op_parallelism_threads.
-   * 0 means the system picks an appropriate number.
-   * If you create an ordinary session, e.g., from Python or C++,
-   * then there is exactly one intra op thread pool per process.
-   * The first session created determines the number of threads in this pool.
-   * All subsequent sessions reuse/share this one global pool.
-   * There are notable exceptions to the default behavior describe above:
-   * 1. There is an environment variable  for overriding this thread pool,
-   *    named TF_OVERRIDE_GLOBAL_THREADPOOL.
-   * 2. When connecting to a server, such as a remote `tf.train.Server`
-   *    instance, then this option will be ignored altogether.
-   * </pre>
-   *
-   * <code>int32 intra_op_parallelism_threads = 2;</code>
-   */
-  public int getIntraOpParallelismThreads() {
-    return intraOpParallelismThreads_;
-  }
-
-  public static final int INTER_OP_PARALLELISM_THREADS_FIELD_NUMBER = 5;
-  private int interOpParallelismThreads_;
-  /**
-   * <pre>
-   * Nodes that perform blocking operations are enqueued on a pool of
-   * inter_op_parallelism_threads available in each process.
-   * 0 means the system picks an appropriate number.
-   * Negative means all operations are performed in caller's thread.
-   * Note that the first Session created in the process sets the
-   * number of threads for all future sessions unless use_per_session_threads is
-   * true or session_inter_op_thread_pool is configured.
-   * </pre>
-   *
-   * <code>int32 inter_op_parallelism_threads = 5;</code>
-   */
-  public int getInterOpParallelismThreads() {
-    return interOpParallelismThreads_;
-  }
-
-  public static final int USE_PER_SESSION_THREADS_FIELD_NUMBER = 9;
-  private boolean usePerSessionThreads_;
-  /**
-   * <pre>
-   * If true, use a new set of threads for this session rather than the global
-   * pool of threads. Only supported by direct sessions.
-   * If false, use the global threads created by the first session, or the
-   * per-session thread pools configured by session_inter_op_thread_pool.
-   * This option is deprecated. The same effect can be achieved by setting
-   * session_inter_op_thread_pool to have one element, whose num_threads equals
-   * inter_op_parallelism_threads.
-   * </pre>
-   *
-   * <code>bool use_per_session_threads = 9;</code>
-   */
-  public boolean getUsePerSessionThreads() {
-    return usePerSessionThreads_;
-  }
-
-  public static final int SESSION_INTER_OP_THREAD_POOL_FIELD_NUMBER = 12;
-  private java.util.List<org.tensorflow.proto.framework.ThreadPoolOptionProto> sessionInterOpThreadPool_;
-  /**
-   * <pre>
-   * This option is experimental - it may be replaced with a different mechanism
-   * in the future.
-   * Configures session thread pools. If this is configured, then RunOptions for
-   * a Run call can select the thread pool to use.
-   * The intended use is for when some session invocations need to run in a
-   * background pool limited to a small number of threads:
-   * - For example, a session may be configured to have one large pool (for
-   * regular compute) and one small pool (for periodic, low priority work);
-   * using the small pool is currently the mechanism for limiting the inter-op
-   * parallelism of the low priority work.  Note that it does not limit the
-   * parallelism of work spawned by a single op kernel implementation.
-   * - Using this setting is normally not needed in training, but may help some
-   * serving use cases.
-   * - It is also generally recommended to set the global_name field of this
-   * proto, to avoid creating multiple large pools. It is typically better to
-   * run the non-low-priority work, even across sessions, in a single large
-   * pool.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.ThreadPoolOptionProto session_inter_op_thread_pool = 12;</code>
-   */
-  public java.util.List<org.tensorflow.proto.framework.ThreadPoolOptionProto> getSessionInterOpThreadPoolList() {
-    return sessionInterOpThreadPool_;
-  }
-  /**
-   * <pre>
-   * This option is experimental - it may be replaced with a different mechanism
-   * in the future.
-   * Configures session thread pools. If this is configured, then RunOptions for
-   * a Run call can select the thread pool to use.
-   * The intended use is for when some session invocations need to run in a
-   * background pool limited to a small number of threads:
-   * - For example, a session may be configured to have one large pool (for
-   * regular compute) and one small pool (for periodic, low priority work);
-   * using the small pool is currently the mechanism for limiting the inter-op
-   * parallelism of the low priority work.  Note that it does not limit the
-   * parallelism of work spawned by a single op kernel implementation.
-   * - Using this setting is normally not needed in training, but may help some
-   * serving use cases.
-   * - It is also generally recommended to set the global_name field of this
-   * proto, to avoid creating multiple large pools. It is typically better to
-   * run the non-low-priority work, even across sessions, in a single large
-   * pool.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.ThreadPoolOptionProto session_inter_op_thread_pool = 12;</code>
-   */
-  public java.util.List<? extends org.tensorflow.proto.framework.ThreadPoolOptionProtoOrBuilder> 
-      getSessionInterOpThreadPoolOrBuilderList() {
-    return sessionInterOpThreadPool_;
-  }
-  /**
-   * <pre>
-   * This option is experimental - it may be replaced with a different mechanism
-   * in the future.
-   * Configures session thread pools. If this is configured, then RunOptions for
-   * a Run call can select the thread pool to use.
-   * The intended use is for when some session invocations need to run in a
-   * background pool limited to a small number of threads:
-   * - For example, a session may be configured to have one large pool (for
-   * regular compute) and one small pool (for periodic, low priority work);
-   * using the small pool is currently the mechanism for limiting the inter-op
-   * parallelism of the low priority work.  Note that it does not limit the
-   * parallelism of work spawned by a single op kernel implementation.
-   * - Using this setting is normally not needed in training, but may help some
-   * serving use cases.
-   * - It is also generally recommended to set the global_name field of this
-   * proto, to avoid creating multiple large pools. It is typically better to
-   * run the non-low-priority work, even across sessions, in a single large
-   * pool.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.ThreadPoolOptionProto session_inter_op_thread_pool = 12;</code>
-   */
-  public int getSessionInterOpThreadPoolCount() {
-    return sessionInterOpThreadPool_.size();
-  }
-  /**
-   * <pre>
-   * This option is experimental - it may be replaced with a different mechanism
-   * in the future.
-   * Configures session thread pools. If this is configured, then RunOptions for
-   * a Run call can select the thread pool to use.
-   * The intended use is for when some session invocations need to run in a
-   * background pool limited to a small number of threads:
-   * - For example, a session may be configured to have one large pool (for
-   * regular compute) and one small pool (for periodic, low priority work);
-   * using the small pool is currently the mechanism for limiting the inter-op
-   * parallelism of the low priority work.  Note that it does not limit the
-   * parallelism of work spawned by a single op kernel implementation.
-   * - Using this setting is normally not needed in training, but may help some
-   * serving use cases.
-   * - It is also generally recommended to set the global_name field of this
-   * proto, to avoid creating multiple large pools. It is typically better to
-   * run the non-low-priority work, even across sessions, in a single large
-   * pool.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.ThreadPoolOptionProto session_inter_op_thread_pool = 12;</code>
-   */
-  public org.tensorflow.proto.framework.ThreadPoolOptionProto getSessionInterOpThreadPool(int index) {
-    return sessionInterOpThreadPool_.get(index);
-  }
-  /**
-   * <pre>
-   * This option is experimental - it may be replaced with a different mechanism
-   * in the future.
-   * Configures session thread pools. If this is configured, then RunOptions for
-   * a Run call can select the thread pool to use.
-   * The intended use is for when some session invocations need to run in a
-   * background pool limited to a small number of threads:
-   * - For example, a session may be configured to have one large pool (for
-   * regular compute) and one small pool (for periodic, low priority work);
-   * using the small pool is currently the mechanism for limiting the inter-op
-   * parallelism of the low priority work.  Note that it does not limit the
-   * parallelism of work spawned by a single op kernel implementation.
-   * - Using this setting is normally not needed in training, but may help some
-   * serving use cases.
-   * - It is also generally recommended to set the global_name field of this
-   * proto, to avoid creating multiple large pools. It is typically better to
-   * run the non-low-priority work, even across sessions, in a single large
-   * pool.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.ThreadPoolOptionProto session_inter_op_thread_pool = 12;</code>
-   */
-  public org.tensorflow.proto.framework.ThreadPoolOptionProtoOrBuilder getSessionInterOpThreadPoolOrBuilder(
-      int index) {
-    return sessionInterOpThreadPool_.get(index);
-  }
-
-  public static final int PLACEMENT_PERIOD_FIELD_NUMBER = 3;
-  private int placementPeriod_;
-  /**
-   * <pre>
-   * Assignment of Nodes to Devices is recomputed every placement_period
-   * steps until the system warms up (at which point the recomputation
-   * typically slows down automatically).
-   * </pre>
-   *
-   * <code>int32 placement_period = 3;</code>
-   */
-  public int getPlacementPeriod() {
-    return placementPeriod_;
-  }
-
-  public static final int DEVICE_FILTERS_FIELD_NUMBER = 4;
-  private com.google.protobuf.LazyStringList deviceFilters_;
-  /**
-   * <pre>
-   * When any filters are present sessions will ignore all devices which do not
-   * match the filters. Each filter can be partially specified, e.g. "/job:ps"
-   * "/job:worker/replica:3", etc.
-   * </pre>
-   *
-   * <code>repeated string device_filters = 4;</code>
-   */
-  public com.google.protobuf.ProtocolStringList
-      getDeviceFiltersList() {
-    return deviceFilters_;
-  }
-  /**
-   * <pre>
-   * When any filters are present sessions will ignore all devices which do not
-   * match the filters. Each filter can be partially specified, e.g. "/job:ps"
-   * "/job:worker/replica:3", etc.
-   * </pre>
-   *
-   * <code>repeated string device_filters = 4;</code>
-   */
-  public int getDeviceFiltersCount() {
-    return deviceFilters_.size();
-  }
-  /**
-   * <pre>
-   * When any filters are present sessions will ignore all devices which do not
-   * match the filters. Each filter can be partially specified, e.g. "/job:ps"
-   * "/job:worker/replica:3", etc.
-   * </pre>
-   *
-   * <code>repeated string device_filters = 4;</code>
-   */
-  public java.lang.String getDeviceFilters(int index) {
-    return deviceFilters_.get(index);
-  }
-  /**
-   * <pre>
-   * When any filters are present sessions will ignore all devices which do not
-   * match the filters. Each filter can be partially specified, e.g. "/job:ps"
-   * "/job:worker/replica:3", etc.
-   * </pre>
-   *
-   * <code>repeated string device_filters = 4;</code>
-   */
-  public com.google.protobuf.ByteString
-      getDeviceFiltersBytes(int index) {
-    return deviceFilters_.getByteString(index);
-  }
-
-  public static final int GPU_OPTIONS_FIELD_NUMBER = 6;
-  private org.tensorflow.proto.framework.GPUOptions gpuOptions_;
-  /**
-   * <pre>
-   * Options that apply to all GPUs.
-   * </pre>
-   *
-   * <code>.tensorflow.GPUOptions gpu_options = 6;</code>
-   */
-  public boolean hasGpuOptions() {
-    return gpuOptions_ != null;
-  }
-  /**
-   * <pre>
-   * Options that apply to all GPUs.
-   * </pre>
-   *
-   * <code>.tensorflow.GPUOptions gpu_options = 6;</code>
-   */
-  public org.tensorflow.proto.framework.GPUOptions getGpuOptions() {
-    return gpuOptions_ == null ? org.tensorflow.proto.framework.GPUOptions.getDefaultInstance() : gpuOptions_;
-  }
-  /**
-   * <pre>
-   * Options that apply to all GPUs.
-   * </pre>
-   *
-   * <code>.tensorflow.GPUOptions gpu_options = 6;</code>
-   */
-  public org.tensorflow.proto.framework.GPUOptionsOrBuilder getGpuOptionsOrBuilder() {
-    return getGpuOptions();
-  }
-
-  public static final int ALLOW_SOFT_PLACEMENT_FIELD_NUMBER = 7;
-  private boolean allowSoftPlacement_;
-  /**
-   * <pre>
-   * Whether soft placement is allowed. If allow_soft_placement is true,
-   * an op will be placed on CPU if
-   *   1. there's no GPU implementation for the OP
-   * or
-   *   2. no GPU devices are known or registered
-   * or
-   *   3. need to co-locate with reftype input(s) which are from CPU.
-   * </pre>
-   *
-   * <code>bool allow_soft_placement = 7;</code>
-   */
-  public boolean getAllowSoftPlacement() {
-    return allowSoftPlacement_;
-  }
-
-  public static final int LOG_DEVICE_PLACEMENT_FIELD_NUMBER = 8;
-  private boolean logDevicePlacement_;
-  /**
-   * <pre>
-   * Whether device placements should be logged.
-   * </pre>
-   *
-   * <code>bool log_device_placement = 8;</code>
-   */
-  public boolean getLogDevicePlacement() {
-    return logDevicePlacement_;
-  }
-
-  public static final int GRAPH_OPTIONS_FIELD_NUMBER = 10;
-  private org.tensorflow.proto.framework.GraphOptions graphOptions_;
-  /**
-   * <pre>
-   * Options that apply to all graphs.
-   * </pre>
-   *
-   * <code>.tensorflow.GraphOptions graph_options = 10;</code>
-   */
-  public boolean hasGraphOptions() {
-    return graphOptions_ != null;
-  }
-  /**
-   * <pre>
-   * Options that apply to all graphs.
-   * </pre>
-   *
-   * <code>.tensorflow.GraphOptions graph_options = 10;</code>
-   */
-  public org.tensorflow.proto.framework.GraphOptions getGraphOptions() {
-    return graphOptions_ == null ? org.tensorflow.proto.framework.GraphOptions.getDefaultInstance() : graphOptions_;
-  }
-  /**
-   * <pre>
-   * Options that apply to all graphs.
-   * </pre>
-   *
-   * <code>.tensorflow.GraphOptions graph_options = 10;</code>
-   */
-  public org.tensorflow.proto.framework.GraphOptionsOrBuilder getGraphOptionsOrBuilder() {
-    return getGraphOptions();
-  }
-
-  public static final int OPERATION_TIMEOUT_IN_MS_FIELD_NUMBER = 11;
-  private long operationTimeoutInMs_;
-  /**
-   * <pre>
-   * Global timeout for all blocking operations in this session.  If non-zero,
-   * and not overridden on a per-operation basis, this value will be used as the
-   * deadline for all blocking operations.
-   * </pre>
-   *
-   * <code>int64 operation_timeout_in_ms = 11;</code>
-   */
-  public long getOperationTimeoutInMs() {
-    return operationTimeoutInMs_;
-  }
-
-  public static final int RPC_OPTIONS_FIELD_NUMBER = 13;
-  private org.tensorflow.proto.framework.RPCOptions rpcOptions_;
-  /**
-   * <pre>
-   * Options that apply when this session uses the distributed runtime.
-   * </pre>
-   *
-   * <code>.tensorflow.RPCOptions rpc_options = 13;</code>
-   */
-  public boolean hasRpcOptions() {
-    return rpcOptions_ != null;
-  }
-  /**
-   * <pre>
-   * Options that apply when this session uses the distributed runtime.
-   * </pre>
-   *
-   * <code>.tensorflow.RPCOptions rpc_options = 13;</code>
-   */
-  public org.tensorflow.proto.framework.RPCOptions getRpcOptions() {
-    return rpcOptions_ == null ? org.tensorflow.proto.framework.RPCOptions.getDefaultInstance() : rpcOptions_;
-  }
-  /**
-   * <pre>
-   * Options that apply when this session uses the distributed runtime.
-   * </pre>
-   *
-   * <code>.tensorflow.RPCOptions rpc_options = 13;</code>
-   */
-  public org.tensorflow.proto.framework.RPCOptionsOrBuilder getRpcOptionsOrBuilder() {
-    return getRpcOptions();
-  }
-
-  public static final int CLUSTER_DEF_FIELD_NUMBER = 14;
-  private org.tensorflow.proto.distruntime.ClusterDef clusterDef_;
-  /**
-   * <pre>
-   * Optional list of all workers to use in this session.
-   * </pre>
-   *
-   * <code>.tensorflow.ClusterDef cluster_def = 14;</code>
-   */
-  public boolean hasClusterDef() {
-    return clusterDef_ != null;
-  }
-  /**
-   * <pre>
-   * Optional list of all workers to use in this session.
-   * </pre>
-   *
-   * <code>.tensorflow.ClusterDef cluster_def = 14;</code>
-   */
-  public org.tensorflow.proto.distruntime.ClusterDef getClusterDef() {
-    return clusterDef_ == null ? org.tensorflow.proto.distruntime.ClusterDef.getDefaultInstance() : clusterDef_;
-  }
-  /**
-   * <pre>
-   * Optional list of all workers to use in this session.
-   * </pre>
-   *
-   * <code>.tensorflow.ClusterDef cluster_def = 14;</code>
-   */
-  public org.tensorflow.proto.distruntime.ClusterDefOrBuilder getClusterDefOrBuilder() {
-    return getClusterDef();
-  }
-
-  public static final int ISOLATE_SESSION_STATE_FIELD_NUMBER = 15;
-  private boolean isolateSessionState_;
-  /**
-   * <pre>
-   * If true, any resources such as Variables used in the session will not be
-   * shared with other sessions. However, when clusterspec propagation is
-   * enabled, this field is ignored and sessions are always isolated.
-   * </pre>
-   *
-   * <code>bool isolate_session_state = 15;</code>
-   */
-  public boolean getIsolateSessionState() {
-    return isolateSessionState_;
-  }
-
-  public static final int SHARE_CLUSTER_DEVICES_IN_SESSION_FIELD_NUMBER = 17;
-  private boolean shareClusterDevicesInSession_;
-  /**
-   * <pre>
-   * When true, WorkerSessions are created with device attributes from the
-   * full cluster.
-   * This is helpful when a worker wants to partition a graph
-   * (for example during a PartitionedCallOp).
-   * </pre>
-   *
-   * <code>bool share_cluster_devices_in_session = 17;</code>
-   */
-  public boolean getShareClusterDevicesInSession() {
-    return shareClusterDevicesInSession_;
-  }
-
-  public static final int EXPERIMENTAL_FIELD_NUMBER = 16;
-  private org.tensorflow.proto.framework.ConfigProto.Experimental experimental_;
-  /**
-   * <code>.tensorflow.ConfigProto.Experimental experimental = 16;</code>
-   */
-  public boolean hasExperimental() {
-    return experimental_ != null;
-  }
-  /**
-   * <code>.tensorflow.ConfigProto.Experimental experimental = 16;</code>
-   */
-  public org.tensorflow.proto.framework.ConfigProto.Experimental getExperimental() {
-    return experimental_ == null ? org.tensorflow.proto.framework.ConfigProto.Experimental.getDefaultInstance() : experimental_;
-  }
-  /**
-   * <code>.tensorflow.ConfigProto.Experimental experimental = 16;</code>
-   */
-  public org.tensorflow.proto.framework.ConfigProto.ExperimentalOrBuilder getExperimentalOrBuilder() {
-    return getExperimental();
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    com.google.protobuf.GeneratedMessageV3
-      .serializeStringMapTo(
-        output,
-        internalGetDeviceCount(),
-        DeviceCountDefaultEntryHolder.defaultEntry,
-        1);
-    if (intraOpParallelismThreads_ != 0) {
-      output.writeInt32(2, intraOpParallelismThreads_);
-    }
-    if (placementPeriod_ != 0) {
-      output.writeInt32(3, placementPeriod_);
-    }
-    for (int i = 0; i < deviceFilters_.size(); i++) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 4, deviceFilters_.getRaw(i));
-    }
-    if (interOpParallelismThreads_ != 0) {
-      output.writeInt32(5, interOpParallelismThreads_);
-    }
-    if (gpuOptions_ != null) {
-      output.writeMessage(6, getGpuOptions());
-    }
-    if (allowSoftPlacement_ != false) {
-      output.writeBool(7, allowSoftPlacement_);
-    }
-    if (logDevicePlacement_ != false) {
-      output.writeBool(8, logDevicePlacement_);
-    }
-    if (usePerSessionThreads_ != false) {
-      output.writeBool(9, usePerSessionThreads_);
-    }
-    if (graphOptions_ != null) {
-      output.writeMessage(10, getGraphOptions());
-    }
-    if (operationTimeoutInMs_ != 0L) {
-      output.writeInt64(11, operationTimeoutInMs_);
-    }
-    for (int i = 0; i < sessionInterOpThreadPool_.size(); i++) {
-      output.writeMessage(12, sessionInterOpThreadPool_.get(i));
-    }
-    if (rpcOptions_ != null) {
-      output.writeMessage(13, getRpcOptions());
-    }
-    if (clusterDef_ != null) {
-      output.writeMessage(14, getClusterDef());
-    }
-    if (isolateSessionState_ != false) {
-      output.writeBool(15, isolateSessionState_);
-    }
-    if (experimental_ != null) {
-      output.writeMessage(16, getExperimental());
-    }
-    if (shareClusterDevicesInSession_ != false) {
-      output.writeBool(17, shareClusterDevicesInSession_);
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    for (java.util.Map.Entry<java.lang.String, java.lang.Integer> entry
-         : internalGetDeviceCount().getMap().entrySet()) {
-      com.google.protobuf.MapEntry<java.lang.String, java.lang.Integer>
-      deviceCount__ = DeviceCountDefaultEntryHolder.defaultEntry.newBuilderForType()
-          .setKey(entry.getKey())
-          .setValue(entry.getValue())
-          .build();
-      size += com.google.protobuf.CodedOutputStream
-          .computeMessageSize(1, deviceCount__);
-    }
-    if (intraOpParallelismThreads_ != 0) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt32Size(2, intraOpParallelismThreads_);
-    }
-    if (placementPeriod_ != 0) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt32Size(3, placementPeriod_);
-    }
-    {
-      int dataSize = 0;
-      for (int i = 0; i < deviceFilters_.size(); i++) {
-        dataSize += computeStringSizeNoTag(deviceFilters_.getRaw(i));
-      }
-      size += dataSize;
-      size += 1 * getDeviceFiltersList().size();
-    }
-    if (interOpParallelismThreads_ != 0) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt32Size(5, interOpParallelismThreads_);
-    }
-    if (gpuOptions_ != null) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(6, getGpuOptions());
-    }
-    if (allowSoftPlacement_ != false) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeBoolSize(7, allowSoftPlacement_);
-    }
-    if (logDevicePlacement_ != false) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeBoolSize(8, logDevicePlacement_);
-    }
-    if (usePerSessionThreads_ != false) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeBoolSize(9, usePerSessionThreads_);
-    }
-    if (graphOptions_ != null) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(10, getGraphOptions());
-    }
-    if (operationTimeoutInMs_ != 0L) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt64Size(11, operationTimeoutInMs_);
-    }
-    for (int i = 0; i < sessionInterOpThreadPool_.size(); i++) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(12, sessionInterOpThreadPool_.get(i));
-    }
-    if (rpcOptions_ != null) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(13, getRpcOptions());
-    }
-    if (clusterDef_ != null) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(14, getClusterDef());
-    }
-    if (isolateSessionState_ != false) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeBoolSize(15, isolateSessionState_);
-    }
-    if (experimental_ != null) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(16, getExperimental());
-    }
-    if (shareClusterDevicesInSession_ != false) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeBoolSize(17, shareClusterDevicesInSession_);
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.framework.ConfigProto)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.framework.ConfigProto other = (org.tensorflow.proto.framework.ConfigProto) obj;
-
-    if (!internalGetDeviceCount().equals(
-        other.internalGetDeviceCount())) return false;
-    if (getIntraOpParallelismThreads()
-        != other.getIntraOpParallelismThreads()) return false;
-    if (getInterOpParallelismThreads()
-        != other.getInterOpParallelismThreads()) return false;
-    if (getUsePerSessionThreads()
-        != other.getUsePerSessionThreads()) return false;
-    if (!getSessionInterOpThreadPoolList()
-        .equals(other.getSessionInterOpThreadPoolList())) return false;
-    if (getPlacementPeriod()
-        != other.getPlacementPeriod()) return false;
-    if (!getDeviceFiltersList()
-        .equals(other.getDeviceFiltersList())) return false;
-    if (hasGpuOptions() != other.hasGpuOptions()) return false;
-    if (hasGpuOptions()) {
-      if (!getGpuOptions()
-          .equals(other.getGpuOptions())) return false;
-    }
-    if (getAllowSoftPlacement()
-        != other.getAllowSoftPlacement()) return false;
-    if (getLogDevicePlacement()
-        != other.getLogDevicePlacement()) return false;
-    if (hasGraphOptions() != other.hasGraphOptions()) return false;
-    if (hasGraphOptions()) {
-      if (!getGraphOptions()
-          .equals(other.getGraphOptions())) return false;
-    }
-    if (getOperationTimeoutInMs()
-        != other.getOperationTimeoutInMs()) return false;
-    if (hasRpcOptions() != other.hasRpcOptions()) return false;
-    if (hasRpcOptions()) {
-      if (!getRpcOptions()
-          .equals(other.getRpcOptions())) return false;
-    }
-    if (hasClusterDef() != other.hasClusterDef()) return false;
-    if (hasClusterDef()) {
-      if (!getClusterDef()
-          .equals(other.getClusterDef())) return false;
-    }
-    if (getIsolateSessionState()
-        != other.getIsolateSessionState()) return false;
-    if (getShareClusterDevicesInSession()
-        != other.getShareClusterDevicesInSession()) return false;
-    if (hasExperimental() != other.hasExperimental()) return false;
-    if (hasExperimental()) {
-      if (!getExperimental()
-          .equals(other.getExperimental())) return false;
-    }
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    if (!internalGetDeviceCount().getMap().isEmpty()) {
-      hash = (37 * hash) + DEVICE_COUNT_FIELD_NUMBER;
-      hash = (53 * hash) + internalGetDeviceCount().hashCode();
-    }
-    hash = (37 * hash) + INTRA_OP_PARALLELISM_THREADS_FIELD_NUMBER;
-    hash = (53 * hash) + getIntraOpParallelismThreads();
-    hash = (37 * hash) + INTER_OP_PARALLELISM_THREADS_FIELD_NUMBER;
-    hash = (53 * hash) + getInterOpParallelismThreads();
-    hash = (37 * hash) + USE_PER_SESSION_THREADS_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashBoolean(
-        getUsePerSessionThreads());
-    if (getSessionInterOpThreadPoolCount() > 0) {
-      hash = (37 * hash) + SESSION_INTER_OP_THREAD_POOL_FIELD_NUMBER;
-      hash = (53 * hash) + getSessionInterOpThreadPoolList().hashCode();
-    }
-    hash = (37 * hash) + PLACEMENT_PERIOD_FIELD_NUMBER;
-    hash = (53 * hash) + getPlacementPeriod();
-    if (getDeviceFiltersCount() > 0) {
-      hash = (37 * hash) + DEVICE_FILTERS_FIELD_NUMBER;
-      hash = (53 * hash) + getDeviceFiltersList().hashCode();
-    }
-    if (hasGpuOptions()) {
-      hash = (37 * hash) + GPU_OPTIONS_FIELD_NUMBER;
-      hash = (53 * hash) + getGpuOptions().hashCode();
-    }
-    hash = (37 * hash) + ALLOW_SOFT_PLACEMENT_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashBoolean(
-        getAllowSoftPlacement());
-    hash = (37 * hash) + LOG_DEVICE_PLACEMENT_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashBoolean(
-        getLogDevicePlacement());
-    if (hasGraphOptions()) {
-      hash = (37 * hash) + GRAPH_OPTIONS_FIELD_NUMBER;
-      hash = (53 * hash) + getGraphOptions().hashCode();
-    }
-    hash = (37 * hash) + OPERATION_TIMEOUT_IN_MS_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-        getOperationTimeoutInMs());
-    if (hasRpcOptions()) {
-      hash = (37 * hash) + RPC_OPTIONS_FIELD_NUMBER;
-      hash = (53 * hash) + getRpcOptions().hashCode();
-    }
-    if (hasClusterDef()) {
-      hash = (37 * hash) + CLUSTER_DEF_FIELD_NUMBER;
-      hash = (53 * hash) + getClusterDef().hashCode();
-    }
-    hash = (37 * hash) + ISOLATE_SESSION_STATE_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashBoolean(
-        getIsolateSessionState());
-    hash = (37 * hash) + SHARE_CLUSTER_DEVICES_IN_SESSION_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashBoolean(
-        getShareClusterDevicesInSession());
-    if (hasExperimental()) {
-      hash = (37 * hash) + EXPERIMENTAL_FIELD_NUMBER;
-      hash = (53 * hash) + getExperimental().hashCode();
-    }
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.framework.ConfigProto parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.ConfigProto parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.ConfigProto parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.ConfigProto parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.ConfigProto parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.ConfigProto parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.ConfigProto parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.ConfigProto parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.ConfigProto parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.ConfigProto parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.ConfigProto parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.ConfigProto parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.framework.ConfigProto prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * <pre>
-   * Session configuration parameters.
-   * The system picks appropriate values for fields that are not set.
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.ConfigProto}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.ConfigProto)
-      org.tensorflow.proto.framework.ConfigProtoOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_ConfigProto_descriptor;
-    }
-
-    @SuppressWarnings({"rawtypes"})
-    protected com.google.protobuf.MapField internalGetMapField(
-        int number) {
-      switch (number) {
-        case 1:
-          return internalGetDeviceCount();
-        default:
-          throw new RuntimeException(
-              "Invalid map field number: " + number);
-      }
-    }
-    @SuppressWarnings({"rawtypes"})
-    protected com.google.protobuf.MapField internalGetMutableMapField(
-        int number) {
-      switch (number) {
-        case 1:
-          return internalGetMutableDeviceCount();
-        default:
-          throw new RuntimeException(
-              "Invalid map field number: " + number);
-      }
-    }
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_ConfigProto_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.ConfigProto.class, org.tensorflow.proto.framework.ConfigProto.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.framework.ConfigProto.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-        getSessionInterOpThreadPoolFieldBuilder();
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      internalGetMutableDeviceCount().clear();
-      intraOpParallelismThreads_ = 0;
-
-      interOpParallelismThreads_ = 0;
-
-      usePerSessionThreads_ = false;
-
-      if (sessionInterOpThreadPoolBuilder_ == null) {
-        sessionInterOpThreadPool_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000002);
-      } else {
-        sessionInterOpThreadPoolBuilder_.clear();
-      }
-      placementPeriod_ = 0;
-
-      deviceFilters_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-      bitField0_ = (bitField0_ & ~0x00000004);
-      if (gpuOptionsBuilder_ == null) {
-        gpuOptions_ = null;
-      } else {
-        gpuOptions_ = null;
-        gpuOptionsBuilder_ = null;
-      }
-      allowSoftPlacement_ = false;
-
-      logDevicePlacement_ = false;
-
-      if (graphOptionsBuilder_ == null) {
-        graphOptions_ = null;
-      } else {
-        graphOptions_ = null;
-        graphOptionsBuilder_ = null;
-      }
-      operationTimeoutInMs_ = 0L;
-
-      if (rpcOptionsBuilder_ == null) {
-        rpcOptions_ = null;
-      } else {
-        rpcOptions_ = null;
-        rpcOptionsBuilder_ = null;
-      }
-      if (clusterDefBuilder_ == null) {
-        clusterDef_ = null;
-      } else {
-        clusterDef_ = null;
-        clusterDefBuilder_ = null;
-      }
-      isolateSessionState_ = false;
-
-      shareClusterDevicesInSession_ = false;
-
-      if (experimentalBuilder_ == null) {
-        experimental_ = null;
-      } else {
-        experimental_ = null;
-        experimentalBuilder_ = null;
-      }
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_ConfigProto_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.ConfigProto getDefaultInstanceForType() {
-      return org.tensorflow.proto.framework.ConfigProto.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.ConfigProto build() {
-      org.tensorflow.proto.framework.ConfigProto result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.ConfigProto buildPartial() {
-      org.tensorflow.proto.framework.ConfigProto result = new org.tensorflow.proto.framework.ConfigProto(this);
-      int from_bitField0_ = bitField0_;
-      result.deviceCount_ = internalGetDeviceCount();
-      result.deviceCount_.makeImmutable();
-      result.intraOpParallelismThreads_ = intraOpParallelismThreads_;
-      result.interOpParallelismThreads_ = interOpParallelismThreads_;
-      result.usePerSessionThreads_ = usePerSessionThreads_;
-      if (sessionInterOpThreadPoolBuilder_ == null) {
-        if (((bitField0_ & 0x00000002) != 0)) {
-          sessionInterOpThreadPool_ = java.util.Collections.unmodifiableList(sessionInterOpThreadPool_);
-          bitField0_ = (bitField0_ & ~0x00000002);
-        }
-        result.sessionInterOpThreadPool_ = sessionInterOpThreadPool_;
-      } else {
-        result.sessionInterOpThreadPool_ = sessionInterOpThreadPoolBuilder_.build();
-      }
-      result.placementPeriod_ = placementPeriod_;
-      if (((bitField0_ & 0x00000004) != 0)) {
-        deviceFilters_ = deviceFilters_.getUnmodifiableView();
-        bitField0_ = (bitField0_ & ~0x00000004);
-      }
-      result.deviceFilters_ = deviceFilters_;
-      if (gpuOptionsBuilder_ == null) {
-        result.gpuOptions_ = gpuOptions_;
-      } else {
-        result.gpuOptions_ = gpuOptionsBuilder_.build();
-      }
-      result.allowSoftPlacement_ = allowSoftPlacement_;
-      result.logDevicePlacement_ = logDevicePlacement_;
-      if (graphOptionsBuilder_ == null) {
-        result.graphOptions_ = graphOptions_;
-      } else {
-        result.graphOptions_ = graphOptionsBuilder_.build();
-      }
-      result.operationTimeoutInMs_ = operationTimeoutInMs_;
-      if (rpcOptionsBuilder_ == null) {
-        result.rpcOptions_ = rpcOptions_;
-      } else {
-        result.rpcOptions_ = rpcOptionsBuilder_.build();
-      }
-      if (clusterDefBuilder_ == null) {
-        result.clusterDef_ = clusterDef_;
-      } else {
-        result.clusterDef_ = clusterDefBuilder_.build();
-      }
-      result.isolateSessionState_ = isolateSessionState_;
-      result.shareClusterDevicesInSession_ = shareClusterDevicesInSession_;
-      if (experimentalBuilder_ == null) {
-        result.experimental_ = experimental_;
-      } else {
-        result.experimental_ = experimentalBuilder_.build();
-      }
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.framework.ConfigProto) {
-        return mergeFrom((org.tensorflow.proto.framework.ConfigProto)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.framework.ConfigProto other) {
-      if (other == org.tensorflow.proto.framework.ConfigProto.getDefaultInstance()) return this;
-      internalGetMutableDeviceCount().mergeFrom(
-          other.internalGetDeviceCount());
-      if (other.getIntraOpParallelismThreads() != 0) {
-        setIntraOpParallelismThreads(other.getIntraOpParallelismThreads());
-      }
-      if (other.getInterOpParallelismThreads() != 0) {
-        setInterOpParallelismThreads(other.getInterOpParallelismThreads());
-      }
-      if (other.getUsePerSessionThreads() != false) {
-        setUsePerSessionThreads(other.getUsePerSessionThreads());
-      }
-      if (sessionInterOpThreadPoolBuilder_ == null) {
-        if (!other.sessionInterOpThreadPool_.isEmpty()) {
-          if (sessionInterOpThreadPool_.isEmpty()) {
-            sessionInterOpThreadPool_ = other.sessionInterOpThreadPool_;
-            bitField0_ = (bitField0_ & ~0x00000002);
-          } else {
-            ensureSessionInterOpThreadPoolIsMutable();
-            sessionInterOpThreadPool_.addAll(other.sessionInterOpThreadPool_);
-          }
-          onChanged();
-        }
-      } else {
-        if (!other.sessionInterOpThreadPool_.isEmpty()) {
-          if (sessionInterOpThreadPoolBuilder_.isEmpty()) {
-            sessionInterOpThreadPoolBuilder_.dispose();
-            sessionInterOpThreadPoolBuilder_ = null;
-            sessionInterOpThreadPool_ = other.sessionInterOpThreadPool_;
-            bitField0_ = (bitField0_ & ~0x00000002);
-            sessionInterOpThreadPoolBuilder_ = 
-              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
-                 getSessionInterOpThreadPoolFieldBuilder() : null;
-          } else {
-            sessionInterOpThreadPoolBuilder_.addAllMessages(other.sessionInterOpThreadPool_);
-          }
-        }
-      }
-      if (other.getPlacementPeriod() != 0) {
-        setPlacementPeriod(other.getPlacementPeriod());
-      }
-      if (!other.deviceFilters_.isEmpty()) {
-        if (deviceFilters_.isEmpty()) {
-          deviceFilters_ = other.deviceFilters_;
-          bitField0_ = (bitField0_ & ~0x00000004);
-        } else {
-          ensureDeviceFiltersIsMutable();
-          deviceFilters_.addAll(other.deviceFilters_);
-        }
-        onChanged();
-      }
-      if (other.hasGpuOptions()) {
-        mergeGpuOptions(other.getGpuOptions());
-      }
-      if (other.getAllowSoftPlacement() != false) {
-        setAllowSoftPlacement(other.getAllowSoftPlacement());
-      }
-      if (other.getLogDevicePlacement() != false) {
-        setLogDevicePlacement(other.getLogDevicePlacement());
-      }
-      if (other.hasGraphOptions()) {
-        mergeGraphOptions(other.getGraphOptions());
-      }
-      if (other.getOperationTimeoutInMs() != 0L) {
-        setOperationTimeoutInMs(other.getOperationTimeoutInMs());
-      }
-      if (other.hasRpcOptions()) {
-        mergeRpcOptions(other.getRpcOptions());
-      }
-      if (other.hasClusterDef()) {
-        mergeClusterDef(other.getClusterDef());
-      }
-      if (other.getIsolateSessionState() != false) {
-        setIsolateSessionState(other.getIsolateSessionState());
-      }
-      if (other.getShareClusterDevicesInSession() != false) {
-        setShareClusterDevicesInSession(other.getShareClusterDevicesInSession());
-      }
-      if (other.hasExperimental()) {
-        mergeExperimental(other.getExperimental());
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.framework.ConfigProto parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.framework.ConfigProto) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-    private int bitField0_;
-
-    private com.google.protobuf.MapField<
-        java.lang.String, java.lang.Integer> deviceCount_;
-    private com.google.protobuf.MapField<java.lang.String, java.lang.Integer>
-    internalGetDeviceCount() {
-      if (deviceCount_ == null) {
-        return com.google.protobuf.MapField.emptyMapField(
-            DeviceCountDefaultEntryHolder.defaultEntry);
-      }
-      return deviceCount_;
-    }
-    private com.google.protobuf.MapField<java.lang.String, java.lang.Integer>
-    internalGetMutableDeviceCount() {
-      onChanged();;
-      if (deviceCount_ == null) {
-        deviceCount_ = com.google.protobuf.MapField.newMapField(
-            DeviceCountDefaultEntryHolder.defaultEntry);
-      }
-      if (!deviceCount_.isMutable()) {
-        deviceCount_ = deviceCount_.copy();
-      }
-      return deviceCount_;
-    }
-
-    public int getDeviceCountCount() {
-      return internalGetDeviceCount().getMap().size();
-    }
-    /**
-     * <pre>
-     * Map from device type name (e.g., "CPU" or "GPU" ) to maximum
-     * number of devices of that type to use.  If a particular device
-     * type is not found in the map, the system picks an appropriate
-     * number.
-     * </pre>
-     *
-     * <code>map&lt;string, int32&gt; device_count = 1;</code>
-     */
-
-    public boolean containsDeviceCount(
-        java.lang.String key) {
-      if (key == null) { throw new java.lang.NullPointerException(); }
-      return internalGetDeviceCount().getMap().containsKey(key);
-    }
-    /**
-     * Use {@link #getDeviceCountMap()} instead.
-     */
-    @java.lang.Deprecated
-    public java.util.Map<java.lang.String, java.lang.Integer> getDeviceCount() {
-      return getDeviceCountMap();
-    }
-    /**
-     * <pre>
-     * Map from device type name (e.g., "CPU" or "GPU" ) to maximum
-     * number of devices of that type to use.  If a particular device
-     * type is not found in the map, the system picks an appropriate
-     * number.
-     * </pre>
-     *
-     * <code>map&lt;string, int32&gt; device_count = 1;</code>
-     */
-
-    public java.util.Map<java.lang.String, java.lang.Integer> getDeviceCountMap() {
-      return internalGetDeviceCount().getMap();
-    }
-    /**
-     * <pre>
-     * Map from device type name (e.g., "CPU" or "GPU" ) to maximum
-     * number of devices of that type to use.  If a particular device
-     * type is not found in the map, the system picks an appropriate
-     * number.
-     * </pre>
-     *
-     * <code>map&lt;string, int32&gt; device_count = 1;</code>
-     */
-
-    public int getDeviceCountOrDefault(
-        java.lang.String key,
-        int defaultValue) {
-      if (key == null) { throw new java.lang.NullPointerException(); }
-      java.util.Map<java.lang.String, java.lang.Integer> map =
-          internalGetDeviceCount().getMap();
-      return map.containsKey(key) ? map.get(key) : defaultValue;
-    }
-    /**
-     * <pre>
-     * Map from device type name (e.g., "CPU" or "GPU" ) to maximum
-     * number of devices of that type to use.  If a particular device
-     * type is not found in the map, the system picks an appropriate
-     * number.
-     * </pre>
-     *
-     * <code>map&lt;string, int32&gt; device_count = 1;</code>
-     */
-
-    public int getDeviceCountOrThrow(
-        java.lang.String key) {
-      if (key == null) { throw new java.lang.NullPointerException(); }
-      java.util.Map<java.lang.String, java.lang.Integer> map =
-          internalGetDeviceCount().getMap();
-      if (!map.containsKey(key)) {
-        throw new java.lang.IllegalArgumentException();
-      }
-      return map.get(key);
-    }
-
-    public Builder clearDeviceCount() {
-      internalGetMutableDeviceCount().getMutableMap()
-          .clear();
-      return this;
-    }
-    /**
-     * <pre>
-     * Map from device type name (e.g., "CPU" or "GPU" ) to maximum
-     * number of devices of that type to use.  If a particular device
-     * type is not found in the map, the system picks an appropriate
-     * number.
-     * </pre>
-     *
-     * <code>map&lt;string, int32&gt; device_count = 1;</code>
-     */
-
-    public Builder removeDeviceCount(
-        java.lang.String key) {
-      if (key == null) { throw new java.lang.NullPointerException(); }
-      internalGetMutableDeviceCount().getMutableMap()
-          .remove(key);
-      return this;
-    }
-    /**
-     * Use alternate mutation accessors instead.
-     */
-    @java.lang.Deprecated
-    public java.util.Map<java.lang.String, java.lang.Integer>
-    getMutableDeviceCount() {
-      return internalGetMutableDeviceCount().getMutableMap();
-    }
-    /**
-     * <pre>
-     * Map from device type name (e.g., "CPU" or "GPU" ) to maximum
-     * number of devices of that type to use.  If a particular device
-     * type is not found in the map, the system picks an appropriate
-     * number.
-     * </pre>
-     *
-     * <code>map&lt;string, int32&gt; device_count = 1;</code>
-     */
-    public Builder putDeviceCount(
-        java.lang.String key,
-        int value) {
-      if (key == null) { throw new java.lang.NullPointerException(); }
-      
-      internalGetMutableDeviceCount().getMutableMap()
-          .put(key, value);
-      return this;
-    }
-    /**
-     * <pre>
-     * Map from device type name (e.g., "CPU" or "GPU" ) to maximum
-     * number of devices of that type to use.  If a particular device
-     * type is not found in the map, the system picks an appropriate
-     * number.
-     * </pre>
-     *
-     * <code>map&lt;string, int32&gt; device_count = 1;</code>
-     */
-
-    public Builder putAllDeviceCount(
-        java.util.Map<java.lang.String, java.lang.Integer> values) {
-      internalGetMutableDeviceCount().getMutableMap()
-          .putAll(values);
-      return this;
-    }
-
-    private int intraOpParallelismThreads_ ;
-    /**
-     * <pre>
-     * The execution of an individual op (for some op types) can be
-     * parallelized on a pool of intra_op_parallelism_threads.
-     * 0 means the system picks an appropriate number.
-     * If you create an ordinary session, e.g., from Python or C++,
-     * then there is exactly one intra op thread pool per process.
-     * The first session created determines the number of threads in this pool.
-     * All subsequent sessions reuse/share this one global pool.
-     * There are notable exceptions to the default behavior describe above:
-     * 1. There is an environment variable  for overriding this thread pool,
-     *    named TF_OVERRIDE_GLOBAL_THREADPOOL.
-     * 2. When connecting to a server, such as a remote `tf.train.Server`
-     *    instance, then this option will be ignored altogether.
-     * </pre>
-     *
-     * <code>int32 intra_op_parallelism_threads = 2;</code>
-     */
-    public int getIntraOpParallelismThreads() {
-      return intraOpParallelismThreads_;
-    }
-    /**
-     * <pre>
-     * The execution of an individual op (for some op types) can be
-     * parallelized on a pool of intra_op_parallelism_threads.
-     * 0 means the system picks an appropriate number.
-     * If you create an ordinary session, e.g., from Python or C++,
-     * then there is exactly one intra op thread pool per process.
-     * The first session created determines the number of threads in this pool.
-     * All subsequent sessions reuse/share this one global pool.
-     * There are notable exceptions to the default behavior describe above:
-     * 1. There is an environment variable  for overriding this thread pool,
-     *    named TF_OVERRIDE_GLOBAL_THREADPOOL.
-     * 2. When connecting to a server, such as a remote `tf.train.Server`
-     *    instance, then this option will be ignored altogether.
-     * </pre>
-     *
-     * <code>int32 intra_op_parallelism_threads = 2;</code>
-     */
-    public Builder setIntraOpParallelismThreads(int value) {
-      
-      intraOpParallelismThreads_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * The execution of an individual op (for some op types) can be
-     * parallelized on a pool of intra_op_parallelism_threads.
-     * 0 means the system picks an appropriate number.
-     * If you create an ordinary session, e.g., from Python or C++,
-     * then there is exactly one intra op thread pool per process.
-     * The first session created determines the number of threads in this pool.
-     * All subsequent sessions reuse/share this one global pool.
-     * There are notable exceptions to the default behavior describe above:
-     * 1. There is an environment variable  for overriding this thread pool,
-     *    named TF_OVERRIDE_GLOBAL_THREADPOOL.
-     * 2. When connecting to a server, such as a remote `tf.train.Server`
-     *    instance, then this option will be ignored altogether.
-     * </pre>
-     *
-     * <code>int32 intra_op_parallelism_threads = 2;</code>
-     */
-    public Builder clearIntraOpParallelismThreads() {
-      
-      intraOpParallelismThreads_ = 0;
-      onChanged();
-      return this;
-    }
-
-    private int interOpParallelismThreads_ ;
-    /**
-     * <pre>
-     * Nodes that perform blocking operations are enqueued on a pool of
-     * inter_op_parallelism_threads available in each process.
-     * 0 means the system picks an appropriate number.
-     * Negative means all operations are performed in caller's thread.
-     * Note that the first Session created in the process sets the
-     * number of threads for all future sessions unless use_per_session_threads is
-     * true or session_inter_op_thread_pool is configured.
-     * </pre>
-     *
-     * <code>int32 inter_op_parallelism_threads = 5;</code>
-     */
-    public int getInterOpParallelismThreads() {
-      return interOpParallelismThreads_;
-    }
-    /**
-     * <pre>
-     * Nodes that perform blocking operations are enqueued on a pool of
-     * inter_op_parallelism_threads available in each process.
-     * 0 means the system picks an appropriate number.
-     * Negative means all operations are performed in caller's thread.
-     * Note that the first Session created in the process sets the
-     * number of threads for all future sessions unless use_per_session_threads is
-     * true or session_inter_op_thread_pool is configured.
-     * </pre>
-     *
-     * <code>int32 inter_op_parallelism_threads = 5;</code>
-     */
-    public Builder setInterOpParallelismThreads(int value) {
-      
-      interOpParallelismThreads_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Nodes that perform blocking operations are enqueued on a pool of
-     * inter_op_parallelism_threads available in each process.
-     * 0 means the system picks an appropriate number.
-     * Negative means all operations are performed in caller's thread.
-     * Note that the first Session created in the process sets the
-     * number of threads for all future sessions unless use_per_session_threads is
-     * true or session_inter_op_thread_pool is configured.
-     * </pre>
-     *
-     * <code>int32 inter_op_parallelism_threads = 5;</code>
-     */
-    public Builder clearInterOpParallelismThreads() {
-      
-      interOpParallelismThreads_ = 0;
-      onChanged();
-      return this;
-    }
-
-    private boolean usePerSessionThreads_ ;
-    /**
-     * <pre>
-     * If true, use a new set of threads for this session rather than the global
-     * pool of threads. Only supported by direct sessions.
-     * If false, use the global threads created by the first session, or the
-     * per-session thread pools configured by session_inter_op_thread_pool.
-     * This option is deprecated. The same effect can be achieved by setting
-     * session_inter_op_thread_pool to have one element, whose num_threads equals
-     * inter_op_parallelism_threads.
-     * </pre>
-     *
-     * <code>bool use_per_session_threads = 9;</code>
-     */
-    public boolean getUsePerSessionThreads() {
-      return usePerSessionThreads_;
-    }
-    /**
-     * <pre>
-     * If true, use a new set of threads for this session rather than the global
-     * pool of threads. Only supported by direct sessions.
-     * If false, use the global threads created by the first session, or the
-     * per-session thread pools configured by session_inter_op_thread_pool.
-     * This option is deprecated. The same effect can be achieved by setting
-     * session_inter_op_thread_pool to have one element, whose num_threads equals
-     * inter_op_parallelism_threads.
-     * </pre>
-     *
-     * <code>bool use_per_session_threads = 9;</code>
-     */
-    public Builder setUsePerSessionThreads(boolean value) {
-      
-      usePerSessionThreads_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * If true, use a new set of threads for this session rather than the global
-     * pool of threads. Only supported by direct sessions.
-     * If false, use the global threads created by the first session, or the
-     * per-session thread pools configured by session_inter_op_thread_pool.
-     * This option is deprecated. The same effect can be achieved by setting
-     * session_inter_op_thread_pool to have one element, whose num_threads equals
-     * inter_op_parallelism_threads.
-     * </pre>
-     *
-     * <code>bool use_per_session_threads = 9;</code>
-     */
-    public Builder clearUsePerSessionThreads() {
-      
-      usePerSessionThreads_ = false;
-      onChanged();
-      return this;
-    }
-
-    private java.util.List<org.tensorflow.proto.framework.ThreadPoolOptionProto> sessionInterOpThreadPool_ =
-      java.util.Collections.emptyList();
-    private void ensureSessionInterOpThreadPoolIsMutable() {
-      if (!((bitField0_ & 0x00000002) != 0)) {
-        sessionInterOpThreadPool_ = new java.util.ArrayList<org.tensorflow.proto.framework.ThreadPoolOptionProto>(sessionInterOpThreadPool_);
-        bitField0_ |= 0x00000002;
-       }
-    }
-
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.framework.ThreadPoolOptionProto, org.tensorflow.proto.framework.ThreadPoolOptionProto.Builder, org.tensorflow.proto.framework.ThreadPoolOptionProtoOrBuilder> sessionInterOpThreadPoolBuilder_;
-
-    /**
-     * <pre>
-     * This option is experimental - it may be replaced with a different mechanism
-     * in the future.
-     * Configures session thread pools. If this is configured, then RunOptions for
-     * a Run call can select the thread pool to use.
-     * The intended use is for when some session invocations need to run in a
-     * background pool limited to a small number of threads:
-     * - For example, a session may be configured to have one large pool (for
-     * regular compute) and one small pool (for periodic, low priority work);
-     * using the small pool is currently the mechanism for limiting the inter-op
-     * parallelism of the low priority work.  Note that it does not limit the
-     * parallelism of work spawned by a single op kernel implementation.
-     * - Using this setting is normally not needed in training, but may help some
-     * serving use cases.
-     * - It is also generally recommended to set the global_name field of this
-     * proto, to avoid creating multiple large pools. It is typically better to
-     * run the non-low-priority work, even across sessions, in a single large
-     * pool.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.ThreadPoolOptionProto session_inter_op_thread_pool = 12;</code>
-     */
-    public java.util.List<org.tensorflow.proto.framework.ThreadPoolOptionProto> getSessionInterOpThreadPoolList() {
-      if (sessionInterOpThreadPoolBuilder_ == null) {
-        return java.util.Collections.unmodifiableList(sessionInterOpThreadPool_);
-      } else {
-        return sessionInterOpThreadPoolBuilder_.getMessageList();
-      }
-    }
-    /**
-     * <pre>
-     * This option is experimental - it may be replaced with a different mechanism
-     * in the future.
-     * Configures session thread pools. If this is configured, then RunOptions for
-     * a Run call can select the thread pool to use.
-     * The intended use is for when some session invocations need to run in a
-     * background pool limited to a small number of threads:
-     * - For example, a session may be configured to have one large pool (for
-     * regular compute) and one small pool (for periodic, low priority work);
-     * using the small pool is currently the mechanism for limiting the inter-op
-     * parallelism of the low priority work.  Note that it does not limit the
-     * parallelism of work spawned by a single op kernel implementation.
-     * - Using this setting is normally not needed in training, but may help some
-     * serving use cases.
-     * - It is also generally recommended to set the global_name field of this
-     * proto, to avoid creating multiple large pools. It is typically better to
-     * run the non-low-priority work, even across sessions, in a single large
-     * pool.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.ThreadPoolOptionProto session_inter_op_thread_pool = 12;</code>
-     */
-    public int getSessionInterOpThreadPoolCount() {
-      if (sessionInterOpThreadPoolBuilder_ == null) {
-        return sessionInterOpThreadPool_.size();
-      } else {
-        return sessionInterOpThreadPoolBuilder_.getCount();
-      }
-    }
-    /**
-     * <pre>
-     * This option is experimental - it may be replaced with a different mechanism
-     * in the future.
-     * Configures session thread pools. If this is configured, then RunOptions for
-     * a Run call can select the thread pool to use.
-     * The intended use is for when some session invocations need to run in a
-     * background pool limited to a small number of threads:
-     * - For example, a session may be configured to have one large pool (for
-     * regular compute) and one small pool (for periodic, low priority work);
-     * using the small pool is currently the mechanism for limiting the inter-op
-     * parallelism of the low priority work.  Note that it does not limit the
-     * parallelism of work spawned by a single op kernel implementation.
-     * - Using this setting is normally not needed in training, but may help some
-     * serving use cases.
-     * - It is also generally recommended to set the global_name field of this
-     * proto, to avoid creating multiple large pools. It is typically better to
-     * run the non-low-priority work, even across sessions, in a single large
-     * pool.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.ThreadPoolOptionProto session_inter_op_thread_pool = 12;</code>
-     */
-    public org.tensorflow.proto.framework.ThreadPoolOptionProto getSessionInterOpThreadPool(int index) {
-      if (sessionInterOpThreadPoolBuilder_ == null) {
-        return sessionInterOpThreadPool_.get(index);
-      } else {
-        return sessionInterOpThreadPoolBuilder_.getMessage(index);
-      }
-    }
-    /**
-     * <pre>
-     * This option is experimental - it may be replaced with a different mechanism
-     * in the future.
-     * Configures session thread pools. If this is configured, then RunOptions for
-     * a Run call can select the thread pool to use.
-     * The intended use is for when some session invocations need to run in a
-     * background pool limited to a small number of threads:
-     * - For example, a session may be configured to have one large pool (for
-     * regular compute) and one small pool (for periodic, low priority work);
-     * using the small pool is currently the mechanism for limiting the inter-op
-     * parallelism of the low priority work.  Note that it does not limit the
-     * parallelism of work spawned by a single op kernel implementation.
-     * - Using this setting is normally not needed in training, but may help some
-     * serving use cases.
-     * - It is also generally recommended to set the global_name field of this
-     * proto, to avoid creating multiple large pools. It is typically better to
-     * run the non-low-priority work, even across sessions, in a single large
-     * pool.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.ThreadPoolOptionProto session_inter_op_thread_pool = 12;</code>
-     */
-    public Builder setSessionInterOpThreadPool(
-        int index, org.tensorflow.proto.framework.ThreadPoolOptionProto value) {
-      if (sessionInterOpThreadPoolBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureSessionInterOpThreadPoolIsMutable();
-        sessionInterOpThreadPool_.set(index, value);
-        onChanged();
-      } else {
-        sessionInterOpThreadPoolBuilder_.setMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * This option is experimental - it may be replaced with a different mechanism
-     * in the future.
-     * Configures session thread pools. If this is configured, then RunOptions for
-     * a Run call can select the thread pool to use.
-     * The intended use is for when some session invocations need to run in a
-     * background pool limited to a small number of threads:
-     * - For example, a session may be configured to have one large pool (for
-     * regular compute) and one small pool (for periodic, low priority work);
-     * using the small pool is currently the mechanism for limiting the inter-op
-     * parallelism of the low priority work.  Note that it does not limit the
-     * parallelism of work spawned by a single op kernel implementation.
-     * - Using this setting is normally not needed in training, but may help some
-     * serving use cases.
-     * - It is also generally recommended to set the global_name field of this
-     * proto, to avoid creating multiple large pools. It is typically better to
-     * run the non-low-priority work, even across sessions, in a single large
-     * pool.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.ThreadPoolOptionProto session_inter_op_thread_pool = 12;</code>
-     */
-    public Builder setSessionInterOpThreadPool(
-        int index, org.tensorflow.proto.framework.ThreadPoolOptionProto.Builder builderForValue) {
-      if (sessionInterOpThreadPoolBuilder_ == null) {
-        ensureSessionInterOpThreadPoolIsMutable();
-        sessionInterOpThreadPool_.set(index, builderForValue.build());
-        onChanged();
-      } else {
-        sessionInterOpThreadPoolBuilder_.setMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * This option is experimental - it may be replaced with a different mechanism
-     * in the future.
-     * Configures session thread pools. If this is configured, then RunOptions for
-     * a Run call can select the thread pool to use.
-     * The intended use is for when some session invocations need to run in a
-     * background pool limited to a small number of threads:
-     * - For example, a session may be configured to have one large pool (for
-     * regular compute) and one small pool (for periodic, low priority work);
-     * using the small pool is currently the mechanism for limiting the inter-op
-     * parallelism of the low priority work.  Note that it does not limit the
-     * parallelism of work spawned by a single op kernel implementation.
-     * - Using this setting is normally not needed in training, but may help some
-     * serving use cases.
-     * - It is also generally recommended to set the global_name field of this
-     * proto, to avoid creating multiple large pools. It is typically better to
-     * run the non-low-priority work, even across sessions, in a single large
-     * pool.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.ThreadPoolOptionProto session_inter_op_thread_pool = 12;</code>
-     */
-    public Builder addSessionInterOpThreadPool(org.tensorflow.proto.framework.ThreadPoolOptionProto value) {
-      if (sessionInterOpThreadPoolBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureSessionInterOpThreadPoolIsMutable();
-        sessionInterOpThreadPool_.add(value);
-        onChanged();
-      } else {
-        sessionInterOpThreadPoolBuilder_.addMessage(value);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * This option is experimental - it may be replaced with a different mechanism
-     * in the future.
-     * Configures session thread pools. If this is configured, then RunOptions for
-     * a Run call can select the thread pool to use.
-     * The intended use is for when some session invocations need to run in a
-     * background pool limited to a small number of threads:
-     * - For example, a session may be configured to have one large pool (for
-     * regular compute) and one small pool (for periodic, low priority work);
-     * using the small pool is currently the mechanism for limiting the inter-op
-     * parallelism of the low priority work.  Note that it does not limit the
-     * parallelism of work spawned by a single op kernel implementation.
-     * - Using this setting is normally not needed in training, but may help some
-     * serving use cases.
-     * - It is also generally recommended to set the global_name field of this
-     * proto, to avoid creating multiple large pools. It is typically better to
-     * run the non-low-priority work, even across sessions, in a single large
-     * pool.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.ThreadPoolOptionProto session_inter_op_thread_pool = 12;</code>
-     */
-    public Builder addSessionInterOpThreadPool(
-        int index, org.tensorflow.proto.framework.ThreadPoolOptionProto value) {
-      if (sessionInterOpThreadPoolBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureSessionInterOpThreadPoolIsMutable();
-        sessionInterOpThreadPool_.add(index, value);
-        onChanged();
-      } else {
-        sessionInterOpThreadPoolBuilder_.addMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * This option is experimental - it may be replaced with a different mechanism
-     * in the future.
-     * Configures session thread pools. If this is configured, then RunOptions for
-     * a Run call can select the thread pool to use.
-     * The intended use is for when some session invocations need to run in a
-     * background pool limited to a small number of threads:
-     * - For example, a session may be configured to have one large pool (for
-     * regular compute) and one small pool (for periodic, low priority work);
-     * using the small pool is currently the mechanism for limiting the inter-op
-     * parallelism of the low priority work.  Note that it does not limit the
-     * parallelism of work spawned by a single op kernel implementation.
-     * - Using this setting is normally not needed in training, but may help some
-     * serving use cases.
-     * - It is also generally recommended to set the global_name field of this
-     * proto, to avoid creating multiple large pools. It is typically better to
-     * run the non-low-priority work, even across sessions, in a single large
-     * pool.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.ThreadPoolOptionProto session_inter_op_thread_pool = 12;</code>
-     */
-    public Builder addSessionInterOpThreadPool(
-        org.tensorflow.proto.framework.ThreadPoolOptionProto.Builder builderForValue) {
-      if (sessionInterOpThreadPoolBuilder_ == null) {
-        ensureSessionInterOpThreadPoolIsMutable();
-        sessionInterOpThreadPool_.add(builderForValue.build());
-        onChanged();
-      } else {
-        sessionInterOpThreadPoolBuilder_.addMessage(builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * This option is experimental - it may be replaced with a different mechanism
-     * in the future.
-     * Configures session thread pools. If this is configured, then RunOptions for
-     * a Run call can select the thread pool to use.
-     * The intended use is for when some session invocations need to run in a
-     * background pool limited to a small number of threads:
-     * - For example, a session may be configured to have one large pool (for
-     * regular compute) and one small pool (for periodic, low priority work);
-     * using the small pool is currently the mechanism for limiting the inter-op
-     * parallelism of the low priority work.  Note that it does not limit the
-     * parallelism of work spawned by a single op kernel implementation.
-     * - Using this setting is normally not needed in training, but may help some
-     * serving use cases.
-     * - It is also generally recommended to set the global_name field of this
-     * proto, to avoid creating multiple large pools. It is typically better to
-     * run the non-low-priority work, even across sessions, in a single large
-     * pool.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.ThreadPoolOptionProto session_inter_op_thread_pool = 12;</code>
-     */
-    public Builder addSessionInterOpThreadPool(
-        int index, org.tensorflow.proto.framework.ThreadPoolOptionProto.Builder builderForValue) {
-      if (sessionInterOpThreadPoolBuilder_ == null) {
-        ensureSessionInterOpThreadPoolIsMutable();
-        sessionInterOpThreadPool_.add(index, builderForValue.build());
-        onChanged();
-      } else {
-        sessionInterOpThreadPoolBuilder_.addMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * This option is experimental - it may be replaced with a different mechanism
-     * in the future.
-     * Configures session thread pools. If this is configured, then RunOptions for
-     * a Run call can select the thread pool to use.
-     * The intended use is for when some session invocations need to run in a
-     * background pool limited to a small number of threads:
-     * - For example, a session may be configured to have one large pool (for
-     * regular compute) and one small pool (for periodic, low priority work);
-     * using the small pool is currently the mechanism for limiting the inter-op
-     * parallelism of the low priority work.  Note that it does not limit the
-     * parallelism of work spawned by a single op kernel implementation.
-     * - Using this setting is normally not needed in training, but may help some
-     * serving use cases.
-     * - It is also generally recommended to set the global_name field of this
-     * proto, to avoid creating multiple large pools. It is typically better to
-     * run the non-low-priority work, even across sessions, in a single large
-     * pool.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.ThreadPoolOptionProto session_inter_op_thread_pool = 12;</code>
-     */
-    public Builder addAllSessionInterOpThreadPool(
-        java.lang.Iterable<? extends org.tensorflow.proto.framework.ThreadPoolOptionProto> values) {
-      if (sessionInterOpThreadPoolBuilder_ == null) {
-        ensureSessionInterOpThreadPoolIsMutable();
-        com.google.protobuf.AbstractMessageLite.Builder.addAll(
-            values, sessionInterOpThreadPool_);
-        onChanged();
-      } else {
-        sessionInterOpThreadPoolBuilder_.addAllMessages(values);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * This option is experimental - it may be replaced with a different mechanism
-     * in the future.
-     * Configures session thread pools. If this is configured, then RunOptions for
-     * a Run call can select the thread pool to use.
-     * The intended use is for when some session invocations need to run in a
-     * background pool limited to a small number of threads:
-     * - For example, a session may be configured to have one large pool (for
-     * regular compute) and one small pool (for periodic, low priority work);
-     * using the small pool is currently the mechanism for limiting the inter-op
-     * parallelism of the low priority work.  Note that it does not limit the
-     * parallelism of work spawned by a single op kernel implementation.
-     * - Using this setting is normally not needed in training, but may help some
-     * serving use cases.
-     * - It is also generally recommended to set the global_name field of this
-     * proto, to avoid creating multiple large pools. It is typically better to
-     * run the non-low-priority work, even across sessions, in a single large
-     * pool.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.ThreadPoolOptionProto session_inter_op_thread_pool = 12;</code>
-     */
-    public Builder clearSessionInterOpThreadPool() {
-      if (sessionInterOpThreadPoolBuilder_ == null) {
-        sessionInterOpThreadPool_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000002);
-        onChanged();
-      } else {
-        sessionInterOpThreadPoolBuilder_.clear();
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * This option is experimental - it may be replaced with a different mechanism
-     * in the future.
-     * Configures session thread pools. If this is configured, then RunOptions for
-     * a Run call can select the thread pool to use.
-     * The intended use is for when some session invocations need to run in a
-     * background pool limited to a small number of threads:
-     * - For example, a session may be configured to have one large pool (for
-     * regular compute) and one small pool (for periodic, low priority work);
-     * using the small pool is currently the mechanism for limiting the inter-op
-     * parallelism of the low priority work.  Note that it does not limit the
-     * parallelism of work spawned by a single op kernel implementation.
-     * - Using this setting is normally not needed in training, but may help some
-     * serving use cases.
-     * - It is also generally recommended to set the global_name field of this
-     * proto, to avoid creating multiple large pools. It is typically better to
-     * run the non-low-priority work, even across sessions, in a single large
-     * pool.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.ThreadPoolOptionProto session_inter_op_thread_pool = 12;</code>
-     */
-    public Builder removeSessionInterOpThreadPool(int index) {
-      if (sessionInterOpThreadPoolBuilder_ == null) {
-        ensureSessionInterOpThreadPoolIsMutable();
-        sessionInterOpThreadPool_.remove(index);
-        onChanged();
-      } else {
-        sessionInterOpThreadPoolBuilder_.remove(index);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * This option is experimental - it may be replaced with a different mechanism
-     * in the future.
-     * Configures session thread pools. If this is configured, then RunOptions for
-     * a Run call can select the thread pool to use.
-     * The intended use is for when some session invocations need to run in a
-     * background pool limited to a small number of threads:
-     * - For example, a session may be configured to have one large pool (for
-     * regular compute) and one small pool (for periodic, low priority work);
-     * using the small pool is currently the mechanism for limiting the inter-op
-     * parallelism of the low priority work.  Note that it does not limit the
-     * parallelism of work spawned by a single op kernel implementation.
-     * - Using this setting is normally not needed in training, but may help some
-     * serving use cases.
-     * - It is also generally recommended to set the global_name field of this
-     * proto, to avoid creating multiple large pools. It is typically better to
-     * run the non-low-priority work, even across sessions, in a single large
-     * pool.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.ThreadPoolOptionProto session_inter_op_thread_pool = 12;</code>
-     */
-    public org.tensorflow.proto.framework.ThreadPoolOptionProto.Builder getSessionInterOpThreadPoolBuilder(
-        int index) {
-      return getSessionInterOpThreadPoolFieldBuilder().getBuilder(index);
-    }
-    /**
-     * <pre>
-     * This option is experimental - it may be replaced with a different mechanism
-     * in the future.
-     * Configures session thread pools. If this is configured, then RunOptions for
-     * a Run call can select the thread pool to use.
-     * The intended use is for when some session invocations need to run in a
-     * background pool limited to a small number of threads:
-     * - For example, a session may be configured to have one large pool (for
-     * regular compute) and one small pool (for periodic, low priority work);
-     * using the small pool is currently the mechanism for limiting the inter-op
-     * parallelism of the low priority work.  Note that it does not limit the
-     * parallelism of work spawned by a single op kernel implementation.
-     * - Using this setting is normally not needed in training, but may help some
-     * serving use cases.
-     * - It is also generally recommended to set the global_name field of this
-     * proto, to avoid creating multiple large pools. It is typically better to
-     * run the non-low-priority work, even across sessions, in a single large
-     * pool.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.ThreadPoolOptionProto session_inter_op_thread_pool = 12;</code>
-     */
-    public org.tensorflow.proto.framework.ThreadPoolOptionProtoOrBuilder getSessionInterOpThreadPoolOrBuilder(
-        int index) {
-      if (sessionInterOpThreadPoolBuilder_ == null) {
-        return sessionInterOpThreadPool_.get(index);  } else {
-        return sessionInterOpThreadPoolBuilder_.getMessageOrBuilder(index);
-      }
-    }
-    /**
-     * <pre>
-     * This option is experimental - it may be replaced with a different mechanism
-     * in the future.
-     * Configures session thread pools. If this is configured, then RunOptions for
-     * a Run call can select the thread pool to use.
-     * The intended use is for when some session invocations need to run in a
-     * background pool limited to a small number of threads:
-     * - For example, a session may be configured to have one large pool (for
-     * regular compute) and one small pool (for periodic, low priority work);
-     * using the small pool is currently the mechanism for limiting the inter-op
-     * parallelism of the low priority work.  Note that it does not limit the
-     * parallelism of work spawned by a single op kernel implementation.
-     * - Using this setting is normally not needed in training, but may help some
-     * serving use cases.
-     * - It is also generally recommended to set the global_name field of this
-     * proto, to avoid creating multiple large pools. It is typically better to
-     * run the non-low-priority work, even across sessions, in a single large
-     * pool.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.ThreadPoolOptionProto session_inter_op_thread_pool = 12;</code>
-     */
-    public java.util.List<? extends org.tensorflow.proto.framework.ThreadPoolOptionProtoOrBuilder> 
-         getSessionInterOpThreadPoolOrBuilderList() {
-      if (sessionInterOpThreadPoolBuilder_ != null) {
-        return sessionInterOpThreadPoolBuilder_.getMessageOrBuilderList();
-      } else {
-        return java.util.Collections.unmodifiableList(sessionInterOpThreadPool_);
-      }
-    }
-    /**
-     * <pre>
-     * This option is experimental - it may be replaced with a different mechanism
-     * in the future.
-     * Configures session thread pools. If this is configured, then RunOptions for
-     * a Run call can select the thread pool to use.
-     * The intended use is for when some session invocations need to run in a
-     * background pool limited to a small number of threads:
-     * - For example, a session may be configured to have one large pool (for
-     * regular compute) and one small pool (for periodic, low priority work);
-     * using the small pool is currently the mechanism for limiting the inter-op
-     * parallelism of the low priority work.  Note that it does not limit the
-     * parallelism of work spawned by a single op kernel implementation.
-     * - Using this setting is normally not needed in training, but may help some
-     * serving use cases.
-     * - It is also generally recommended to set the global_name field of this
-     * proto, to avoid creating multiple large pools. It is typically better to
-     * run the non-low-priority work, even across sessions, in a single large
-     * pool.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.ThreadPoolOptionProto session_inter_op_thread_pool = 12;</code>
-     */
-    public org.tensorflow.proto.framework.ThreadPoolOptionProto.Builder addSessionInterOpThreadPoolBuilder() {
-      return getSessionInterOpThreadPoolFieldBuilder().addBuilder(
-          org.tensorflow.proto.framework.ThreadPoolOptionProto.getDefaultInstance());
-    }
-    /**
-     * <pre>
-     * This option is experimental - it may be replaced with a different mechanism
-     * in the future.
-     * Configures session thread pools. If this is configured, then RunOptions for
-     * a Run call can select the thread pool to use.
-     * The intended use is for when some session invocations need to run in a
-     * background pool limited to a small number of threads:
-     * - For example, a session may be configured to have one large pool (for
-     * regular compute) and one small pool (for periodic, low priority work);
-     * using the small pool is currently the mechanism for limiting the inter-op
-     * parallelism of the low priority work.  Note that it does not limit the
-     * parallelism of work spawned by a single op kernel implementation.
-     * - Using this setting is normally not needed in training, but may help some
-     * serving use cases.
-     * - It is also generally recommended to set the global_name field of this
-     * proto, to avoid creating multiple large pools. It is typically better to
-     * run the non-low-priority work, even across sessions, in a single large
-     * pool.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.ThreadPoolOptionProto session_inter_op_thread_pool = 12;</code>
-     */
-    public org.tensorflow.proto.framework.ThreadPoolOptionProto.Builder addSessionInterOpThreadPoolBuilder(
-        int index) {
-      return getSessionInterOpThreadPoolFieldBuilder().addBuilder(
-          index, org.tensorflow.proto.framework.ThreadPoolOptionProto.getDefaultInstance());
-    }
-    /**
-     * <pre>
-     * This option is experimental - it may be replaced with a different mechanism
-     * in the future.
-     * Configures session thread pools. If this is configured, then RunOptions for
-     * a Run call can select the thread pool to use.
-     * The intended use is for when some session invocations need to run in a
-     * background pool limited to a small number of threads:
-     * - For example, a session may be configured to have one large pool (for
-     * regular compute) and one small pool (for periodic, low priority work);
-     * using the small pool is currently the mechanism for limiting the inter-op
-     * parallelism of the low priority work.  Note that it does not limit the
-     * parallelism of work spawned by a single op kernel implementation.
-     * - Using this setting is normally not needed in training, but may help some
-     * serving use cases.
-     * - It is also generally recommended to set the global_name field of this
-     * proto, to avoid creating multiple large pools. It is typically better to
-     * run the non-low-priority work, even across sessions, in a single large
-     * pool.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.ThreadPoolOptionProto session_inter_op_thread_pool = 12;</code>
-     */
-    public java.util.List<org.tensorflow.proto.framework.ThreadPoolOptionProto.Builder> 
-         getSessionInterOpThreadPoolBuilderList() {
-      return getSessionInterOpThreadPoolFieldBuilder().getBuilderList();
-    }
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.framework.ThreadPoolOptionProto, org.tensorflow.proto.framework.ThreadPoolOptionProto.Builder, org.tensorflow.proto.framework.ThreadPoolOptionProtoOrBuilder> 
-        getSessionInterOpThreadPoolFieldBuilder() {
-      if (sessionInterOpThreadPoolBuilder_ == null) {
-        sessionInterOpThreadPoolBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
-            org.tensorflow.proto.framework.ThreadPoolOptionProto, org.tensorflow.proto.framework.ThreadPoolOptionProto.Builder, org.tensorflow.proto.framework.ThreadPoolOptionProtoOrBuilder>(
-                sessionInterOpThreadPool_,
-                ((bitField0_ & 0x00000002) != 0),
-                getParentForChildren(),
-                isClean());
-        sessionInterOpThreadPool_ = null;
-      }
-      return sessionInterOpThreadPoolBuilder_;
-    }
-
-    private int placementPeriod_ ;
-    /**
-     * <pre>
-     * Assignment of Nodes to Devices is recomputed every placement_period
-     * steps until the system warms up (at which point the recomputation
-     * typically slows down automatically).
-     * </pre>
-     *
-     * <code>int32 placement_period = 3;</code>
-     */
-    public int getPlacementPeriod() {
-      return placementPeriod_;
-    }
-    /**
-     * <pre>
-     * Assignment of Nodes to Devices is recomputed every placement_period
-     * steps until the system warms up (at which point the recomputation
-     * typically slows down automatically).
-     * </pre>
-     *
-     * <code>int32 placement_period = 3;</code>
-     */
-    public Builder setPlacementPeriod(int value) {
-      
-      placementPeriod_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Assignment of Nodes to Devices is recomputed every placement_period
-     * steps until the system warms up (at which point the recomputation
-     * typically slows down automatically).
-     * </pre>
-     *
-     * <code>int32 placement_period = 3;</code>
-     */
-    public Builder clearPlacementPeriod() {
-      
-      placementPeriod_ = 0;
-      onChanged();
-      return this;
-    }
-
-    private com.google.protobuf.LazyStringList deviceFilters_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-    private void ensureDeviceFiltersIsMutable() {
-      if (!((bitField0_ & 0x00000004) != 0)) {
-        deviceFilters_ = new com.google.protobuf.LazyStringArrayList(deviceFilters_);
-        bitField0_ |= 0x00000004;
-       }
-    }
-    /**
-     * <pre>
-     * When any filters are present sessions will ignore all devices which do not
-     * match the filters. Each filter can be partially specified, e.g. "/job:ps"
-     * "/job:worker/replica:3", etc.
-     * </pre>
-     *
-     * <code>repeated string device_filters = 4;</code>
-     */
-    public com.google.protobuf.ProtocolStringList
-        getDeviceFiltersList() {
-      return deviceFilters_.getUnmodifiableView();
-    }
-    /**
-     * <pre>
-     * When any filters are present sessions will ignore all devices which do not
-     * match the filters. Each filter can be partially specified, e.g. "/job:ps"
-     * "/job:worker/replica:3", etc.
-     * </pre>
-     *
-     * <code>repeated string device_filters = 4;</code>
-     */
-    public int getDeviceFiltersCount() {
-      return deviceFilters_.size();
-    }
-    /**
-     * <pre>
-     * When any filters are present sessions will ignore all devices which do not
-     * match the filters. Each filter can be partially specified, e.g. "/job:ps"
-     * "/job:worker/replica:3", etc.
-     * </pre>
-     *
-     * <code>repeated string device_filters = 4;</code>
-     */
-    public java.lang.String getDeviceFilters(int index) {
-      return deviceFilters_.get(index);
-    }
-    /**
-     * <pre>
-     * When any filters are present sessions will ignore all devices which do not
-     * match the filters. Each filter can be partially specified, e.g. "/job:ps"
-     * "/job:worker/replica:3", etc.
-     * </pre>
-     *
-     * <code>repeated string device_filters = 4;</code>
-     */
-    public com.google.protobuf.ByteString
-        getDeviceFiltersBytes(int index) {
-      return deviceFilters_.getByteString(index);
-    }
-    /**
-     * <pre>
-     * When any filters are present sessions will ignore all devices which do not
-     * match the filters. Each filter can be partially specified, e.g. "/job:ps"
-     * "/job:worker/replica:3", etc.
-     * </pre>
-     *
-     * <code>repeated string device_filters = 4;</code>
-     */
-    public Builder setDeviceFilters(
-        int index, java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  ensureDeviceFiltersIsMutable();
-      deviceFilters_.set(index, value);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * When any filters are present sessions will ignore all devices which do not
-     * match the filters. Each filter can be partially specified, e.g. "/job:ps"
-     * "/job:worker/replica:3", etc.
-     * </pre>
-     *
-     * <code>repeated string device_filters = 4;</code>
-     */
-    public Builder addDeviceFilters(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  ensureDeviceFiltersIsMutable();
-      deviceFilters_.add(value);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * When any filters are present sessions will ignore all devices which do not
-     * match the filters. Each filter can be partially specified, e.g. "/job:ps"
-     * "/job:worker/replica:3", etc.
-     * </pre>
-     *
-     * <code>repeated string device_filters = 4;</code>
-     */
-    public Builder addAllDeviceFilters(
-        java.lang.Iterable<java.lang.String> values) {
-      ensureDeviceFiltersIsMutable();
-      com.google.protobuf.AbstractMessageLite.Builder.addAll(
-          values, deviceFilters_);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * When any filters are present sessions will ignore all devices which do not
-     * match the filters. Each filter can be partially specified, e.g. "/job:ps"
-     * "/job:worker/replica:3", etc.
-     * </pre>
-     *
-     * <code>repeated string device_filters = 4;</code>
-     */
-    public Builder clearDeviceFilters() {
-      deviceFilters_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-      bitField0_ = (bitField0_ & ~0x00000004);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * When any filters are present sessions will ignore all devices which do not
-     * match the filters. Each filter can be partially specified, e.g. "/job:ps"
-     * "/job:worker/replica:3", etc.
-     * </pre>
-     *
-     * <code>repeated string device_filters = 4;</code>
-     */
-    public Builder addDeviceFiltersBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      ensureDeviceFiltersIsMutable();
-      deviceFilters_.add(value);
-      onChanged();
-      return this;
-    }
-
-    private org.tensorflow.proto.framework.GPUOptions gpuOptions_;
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.GPUOptions, org.tensorflow.proto.framework.GPUOptions.Builder, org.tensorflow.proto.framework.GPUOptionsOrBuilder> gpuOptionsBuilder_;
-    /**
-     * <pre>
-     * Options that apply to all GPUs.
-     * </pre>
-     *
-     * <code>.tensorflow.GPUOptions gpu_options = 6;</code>
-     */
-    public boolean hasGpuOptions() {
-      return gpuOptionsBuilder_ != null || gpuOptions_ != null;
-    }
-    /**
-     * <pre>
-     * Options that apply to all GPUs.
-     * </pre>
-     *
-     * <code>.tensorflow.GPUOptions gpu_options = 6;</code>
-     */
-    public org.tensorflow.proto.framework.GPUOptions getGpuOptions() {
-      if (gpuOptionsBuilder_ == null) {
-        return gpuOptions_ == null ? org.tensorflow.proto.framework.GPUOptions.getDefaultInstance() : gpuOptions_;
-      } else {
-        return gpuOptionsBuilder_.getMessage();
-      }
-    }
-    /**
-     * <pre>
-     * Options that apply to all GPUs.
-     * </pre>
-     *
-     * <code>.tensorflow.GPUOptions gpu_options = 6;</code>
-     */
-    public Builder setGpuOptions(org.tensorflow.proto.framework.GPUOptions value) {
-      if (gpuOptionsBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        gpuOptions_ = value;
-        onChanged();
-      } else {
-        gpuOptionsBuilder_.setMessage(value);
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Options that apply to all GPUs.
-     * </pre>
-     *
-     * <code>.tensorflow.GPUOptions gpu_options = 6;</code>
-     */
-    public Builder setGpuOptions(
-        org.tensorflow.proto.framework.GPUOptions.Builder builderForValue) {
-      if (gpuOptionsBuilder_ == null) {
-        gpuOptions_ = builderForValue.build();
-        onChanged();
-      } else {
-        gpuOptionsBuilder_.setMessage(builderForValue.build());
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Options that apply to all GPUs.
-     * </pre>
-     *
-     * <code>.tensorflow.GPUOptions gpu_options = 6;</code>
-     */
-    public Builder mergeGpuOptions(org.tensorflow.proto.framework.GPUOptions value) {
-      if (gpuOptionsBuilder_ == null) {
-        if (gpuOptions_ != null) {
-          gpuOptions_ =
-            org.tensorflow.proto.framework.GPUOptions.newBuilder(gpuOptions_).mergeFrom(value).buildPartial();
-        } else {
-          gpuOptions_ = value;
-        }
-        onChanged();
-      } else {
-        gpuOptionsBuilder_.mergeFrom(value);
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Options that apply to all GPUs.
-     * </pre>
-     *
-     * <code>.tensorflow.GPUOptions gpu_options = 6;</code>
-     */
-    public Builder clearGpuOptions() {
-      if (gpuOptionsBuilder_ == null) {
-        gpuOptions_ = null;
-        onChanged();
-      } else {
-        gpuOptions_ = null;
-        gpuOptionsBuilder_ = null;
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Options that apply to all GPUs.
-     * </pre>
-     *
-     * <code>.tensorflow.GPUOptions gpu_options = 6;</code>
-     */
-    public org.tensorflow.proto.framework.GPUOptions.Builder getGpuOptionsBuilder() {
-      
-      onChanged();
-      return getGpuOptionsFieldBuilder().getBuilder();
-    }
-    /**
-     * <pre>
-     * Options that apply to all GPUs.
-     * </pre>
-     *
-     * <code>.tensorflow.GPUOptions gpu_options = 6;</code>
-     */
-    public org.tensorflow.proto.framework.GPUOptionsOrBuilder getGpuOptionsOrBuilder() {
-      if (gpuOptionsBuilder_ != null) {
-        return gpuOptionsBuilder_.getMessageOrBuilder();
-      } else {
-        return gpuOptions_ == null ?
-            org.tensorflow.proto.framework.GPUOptions.getDefaultInstance() : gpuOptions_;
-      }
-    }
-    /**
-     * <pre>
-     * Options that apply to all GPUs.
-     * </pre>
-     *
-     * <code>.tensorflow.GPUOptions gpu_options = 6;</code>
-     */
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.GPUOptions, org.tensorflow.proto.framework.GPUOptions.Builder, org.tensorflow.proto.framework.GPUOptionsOrBuilder> 
-        getGpuOptionsFieldBuilder() {
-      if (gpuOptionsBuilder_ == null) {
-        gpuOptionsBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-            org.tensorflow.proto.framework.GPUOptions, org.tensorflow.proto.framework.GPUOptions.Builder, org.tensorflow.proto.framework.GPUOptionsOrBuilder>(
-                getGpuOptions(),
-                getParentForChildren(),
-                isClean());
-        gpuOptions_ = null;
-      }
-      return gpuOptionsBuilder_;
-    }
-
-    private boolean allowSoftPlacement_ ;
-    /**
-     * <pre>
-     * Whether soft placement is allowed. If allow_soft_placement is true,
-     * an op will be placed on CPU if
-     *   1. there's no GPU implementation for the OP
-     * or
-     *   2. no GPU devices are known or registered
-     * or
-     *   3. need to co-locate with reftype input(s) which are from CPU.
-     * </pre>
-     *
-     * <code>bool allow_soft_placement = 7;</code>
-     */
-    public boolean getAllowSoftPlacement() {
-      return allowSoftPlacement_;
-    }
-    /**
-     * <pre>
-     * Whether soft placement is allowed. If allow_soft_placement is true,
-     * an op will be placed on CPU if
-     *   1. there's no GPU implementation for the OP
-     * or
-     *   2. no GPU devices are known or registered
-     * or
-     *   3. need to co-locate with reftype input(s) which are from CPU.
-     * </pre>
-     *
-     * <code>bool allow_soft_placement = 7;</code>
-     */
-    public Builder setAllowSoftPlacement(boolean value) {
-      
-      allowSoftPlacement_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Whether soft placement is allowed. If allow_soft_placement is true,
-     * an op will be placed on CPU if
-     *   1. there's no GPU implementation for the OP
-     * or
-     *   2. no GPU devices are known or registered
-     * or
-     *   3. need to co-locate with reftype input(s) which are from CPU.
-     * </pre>
-     *
-     * <code>bool allow_soft_placement = 7;</code>
-     */
-    public Builder clearAllowSoftPlacement() {
-      
-      allowSoftPlacement_ = false;
-      onChanged();
-      return this;
-    }
-
-    private boolean logDevicePlacement_ ;
-    /**
-     * <pre>
-     * Whether device placements should be logged.
-     * </pre>
-     *
-     * <code>bool log_device_placement = 8;</code>
-     */
-    public boolean getLogDevicePlacement() {
-      return logDevicePlacement_;
-    }
-    /**
-     * <pre>
-     * Whether device placements should be logged.
-     * </pre>
-     *
-     * <code>bool log_device_placement = 8;</code>
-     */
-    public Builder setLogDevicePlacement(boolean value) {
-      
-      logDevicePlacement_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Whether device placements should be logged.
-     * </pre>
-     *
-     * <code>bool log_device_placement = 8;</code>
-     */
-    public Builder clearLogDevicePlacement() {
-      
-      logDevicePlacement_ = false;
-      onChanged();
-      return this;
-    }
-
-    private org.tensorflow.proto.framework.GraphOptions graphOptions_;
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.GraphOptions, org.tensorflow.proto.framework.GraphOptions.Builder, org.tensorflow.proto.framework.GraphOptionsOrBuilder> graphOptionsBuilder_;
-    /**
-     * <pre>
-     * Options that apply to all graphs.
-     * </pre>
-     *
-     * <code>.tensorflow.GraphOptions graph_options = 10;</code>
-     */
-    public boolean hasGraphOptions() {
-      return graphOptionsBuilder_ != null || graphOptions_ != null;
-    }
-    /**
-     * <pre>
-     * Options that apply to all graphs.
-     * </pre>
-     *
-     * <code>.tensorflow.GraphOptions graph_options = 10;</code>
-     */
-    public org.tensorflow.proto.framework.GraphOptions getGraphOptions() {
-      if (graphOptionsBuilder_ == null) {
-        return graphOptions_ == null ? org.tensorflow.proto.framework.GraphOptions.getDefaultInstance() : graphOptions_;
-      } else {
-        return graphOptionsBuilder_.getMessage();
-      }
-    }
-    /**
-     * <pre>
-     * Options that apply to all graphs.
-     * </pre>
-     *
-     * <code>.tensorflow.GraphOptions graph_options = 10;</code>
-     */
-    public Builder setGraphOptions(org.tensorflow.proto.framework.GraphOptions value) {
-      if (graphOptionsBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        graphOptions_ = value;
-        onChanged();
-      } else {
-        graphOptionsBuilder_.setMessage(value);
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Options that apply to all graphs.
-     * </pre>
-     *
-     * <code>.tensorflow.GraphOptions graph_options = 10;</code>
-     */
-    public Builder setGraphOptions(
-        org.tensorflow.proto.framework.GraphOptions.Builder builderForValue) {
-      if (graphOptionsBuilder_ == null) {
-        graphOptions_ = builderForValue.build();
-        onChanged();
-      } else {
-        graphOptionsBuilder_.setMessage(builderForValue.build());
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Options that apply to all graphs.
-     * </pre>
-     *
-     * <code>.tensorflow.GraphOptions graph_options = 10;</code>
-     */
-    public Builder mergeGraphOptions(org.tensorflow.proto.framework.GraphOptions value) {
-      if (graphOptionsBuilder_ == null) {
-        if (graphOptions_ != null) {
-          graphOptions_ =
-            org.tensorflow.proto.framework.GraphOptions.newBuilder(graphOptions_).mergeFrom(value).buildPartial();
-        } else {
-          graphOptions_ = value;
-        }
-        onChanged();
-      } else {
-        graphOptionsBuilder_.mergeFrom(value);
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Options that apply to all graphs.
-     * </pre>
-     *
-     * <code>.tensorflow.GraphOptions graph_options = 10;</code>
-     */
-    public Builder clearGraphOptions() {
-      if (graphOptionsBuilder_ == null) {
-        graphOptions_ = null;
-        onChanged();
-      } else {
-        graphOptions_ = null;
-        graphOptionsBuilder_ = null;
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Options that apply to all graphs.
-     * </pre>
-     *
-     * <code>.tensorflow.GraphOptions graph_options = 10;</code>
-     */
-    public org.tensorflow.proto.framework.GraphOptions.Builder getGraphOptionsBuilder() {
-      
-      onChanged();
-      return getGraphOptionsFieldBuilder().getBuilder();
-    }
-    /**
-     * <pre>
-     * Options that apply to all graphs.
-     * </pre>
-     *
-     * <code>.tensorflow.GraphOptions graph_options = 10;</code>
-     */
-    public org.tensorflow.proto.framework.GraphOptionsOrBuilder getGraphOptionsOrBuilder() {
-      if (graphOptionsBuilder_ != null) {
-        return graphOptionsBuilder_.getMessageOrBuilder();
-      } else {
-        return graphOptions_ == null ?
-            org.tensorflow.proto.framework.GraphOptions.getDefaultInstance() : graphOptions_;
-      }
-    }
-    /**
-     * <pre>
-     * Options that apply to all graphs.
-     * </pre>
-     *
-     * <code>.tensorflow.GraphOptions graph_options = 10;</code>
-     */
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.GraphOptions, org.tensorflow.proto.framework.GraphOptions.Builder, org.tensorflow.proto.framework.GraphOptionsOrBuilder> 
-        getGraphOptionsFieldBuilder() {
-      if (graphOptionsBuilder_ == null) {
-        graphOptionsBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-            org.tensorflow.proto.framework.GraphOptions, org.tensorflow.proto.framework.GraphOptions.Builder, org.tensorflow.proto.framework.GraphOptionsOrBuilder>(
-                getGraphOptions(),
-                getParentForChildren(),
-                isClean());
-        graphOptions_ = null;
-      }
-      return graphOptionsBuilder_;
-    }
-
-    private long operationTimeoutInMs_ ;
-    /**
-     * <pre>
-     * Global timeout for all blocking operations in this session.  If non-zero,
-     * and not overridden on a per-operation basis, this value will be used as the
-     * deadline for all blocking operations.
-     * </pre>
-     *
-     * <code>int64 operation_timeout_in_ms = 11;</code>
-     */
-    public long getOperationTimeoutInMs() {
-      return operationTimeoutInMs_;
-    }
-    /**
-     * <pre>
-     * Global timeout for all blocking operations in this session.  If non-zero,
-     * and not overridden on a per-operation basis, this value will be used as the
-     * deadline for all blocking operations.
-     * </pre>
-     *
-     * <code>int64 operation_timeout_in_ms = 11;</code>
-     */
-    public Builder setOperationTimeoutInMs(long value) {
-      
-      operationTimeoutInMs_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Global timeout for all blocking operations in this session.  If non-zero,
-     * and not overridden on a per-operation basis, this value will be used as the
-     * deadline for all blocking operations.
-     * </pre>
-     *
-     * <code>int64 operation_timeout_in_ms = 11;</code>
-     */
-    public Builder clearOperationTimeoutInMs() {
-      
-      operationTimeoutInMs_ = 0L;
-      onChanged();
-      return this;
-    }
-
-    private org.tensorflow.proto.framework.RPCOptions rpcOptions_;
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.RPCOptions, org.tensorflow.proto.framework.RPCOptions.Builder, org.tensorflow.proto.framework.RPCOptionsOrBuilder> rpcOptionsBuilder_;
-    /**
-     * <pre>
-     * Options that apply when this session uses the distributed runtime.
-     * </pre>
-     *
-     * <code>.tensorflow.RPCOptions rpc_options = 13;</code>
-     */
-    public boolean hasRpcOptions() {
-      return rpcOptionsBuilder_ != null || rpcOptions_ != null;
-    }
-    /**
-     * <pre>
-     * Options that apply when this session uses the distributed runtime.
-     * </pre>
-     *
-     * <code>.tensorflow.RPCOptions rpc_options = 13;</code>
-     */
-    public org.tensorflow.proto.framework.RPCOptions getRpcOptions() {
-      if (rpcOptionsBuilder_ == null) {
-        return rpcOptions_ == null ? org.tensorflow.proto.framework.RPCOptions.getDefaultInstance() : rpcOptions_;
-      } else {
-        return rpcOptionsBuilder_.getMessage();
-      }
-    }
-    /**
-     * <pre>
-     * Options that apply when this session uses the distributed runtime.
-     * </pre>
-     *
-     * <code>.tensorflow.RPCOptions rpc_options = 13;</code>
-     */
-    public Builder setRpcOptions(org.tensorflow.proto.framework.RPCOptions value) {
-      if (rpcOptionsBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        rpcOptions_ = value;
-        onChanged();
-      } else {
-        rpcOptionsBuilder_.setMessage(value);
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Options that apply when this session uses the distributed runtime.
-     * </pre>
-     *
-     * <code>.tensorflow.RPCOptions rpc_options = 13;</code>
-     */
-    public Builder setRpcOptions(
-        org.tensorflow.proto.framework.RPCOptions.Builder builderForValue) {
-      if (rpcOptionsBuilder_ == null) {
-        rpcOptions_ = builderForValue.build();
-        onChanged();
-      } else {
-        rpcOptionsBuilder_.setMessage(builderForValue.build());
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Options that apply when this session uses the distributed runtime.
-     * </pre>
-     *
-     * <code>.tensorflow.RPCOptions rpc_options = 13;</code>
-     */
-    public Builder mergeRpcOptions(org.tensorflow.proto.framework.RPCOptions value) {
-      if (rpcOptionsBuilder_ == null) {
-        if (rpcOptions_ != null) {
-          rpcOptions_ =
-            org.tensorflow.proto.framework.RPCOptions.newBuilder(rpcOptions_).mergeFrom(value).buildPartial();
-        } else {
-          rpcOptions_ = value;
-        }
-        onChanged();
-      } else {
-        rpcOptionsBuilder_.mergeFrom(value);
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Options that apply when this session uses the distributed runtime.
-     * </pre>
-     *
-     * <code>.tensorflow.RPCOptions rpc_options = 13;</code>
-     */
-    public Builder clearRpcOptions() {
-      if (rpcOptionsBuilder_ == null) {
-        rpcOptions_ = null;
-        onChanged();
-      } else {
-        rpcOptions_ = null;
-        rpcOptionsBuilder_ = null;
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Options that apply when this session uses the distributed runtime.
-     * </pre>
-     *
-     * <code>.tensorflow.RPCOptions rpc_options = 13;</code>
-     */
-    public org.tensorflow.proto.framework.RPCOptions.Builder getRpcOptionsBuilder() {
-      
-      onChanged();
-      return getRpcOptionsFieldBuilder().getBuilder();
-    }
-    /**
-     * <pre>
-     * Options that apply when this session uses the distributed runtime.
-     * </pre>
-     *
-     * <code>.tensorflow.RPCOptions rpc_options = 13;</code>
-     */
-    public org.tensorflow.proto.framework.RPCOptionsOrBuilder getRpcOptionsOrBuilder() {
-      if (rpcOptionsBuilder_ != null) {
-        return rpcOptionsBuilder_.getMessageOrBuilder();
-      } else {
-        return rpcOptions_ == null ?
-            org.tensorflow.proto.framework.RPCOptions.getDefaultInstance() : rpcOptions_;
-      }
-    }
-    /**
-     * <pre>
-     * Options that apply when this session uses the distributed runtime.
-     * </pre>
-     *
-     * <code>.tensorflow.RPCOptions rpc_options = 13;</code>
-     */
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.RPCOptions, org.tensorflow.proto.framework.RPCOptions.Builder, org.tensorflow.proto.framework.RPCOptionsOrBuilder> 
-        getRpcOptionsFieldBuilder() {
-      if (rpcOptionsBuilder_ == null) {
-        rpcOptionsBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-            org.tensorflow.proto.framework.RPCOptions, org.tensorflow.proto.framework.RPCOptions.Builder, org.tensorflow.proto.framework.RPCOptionsOrBuilder>(
-                getRpcOptions(),
-                getParentForChildren(),
-                isClean());
-        rpcOptions_ = null;
-      }
-      return rpcOptionsBuilder_;
-    }
-
-    private org.tensorflow.proto.distruntime.ClusterDef clusterDef_;
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.distruntime.ClusterDef, org.tensorflow.proto.distruntime.ClusterDef.Builder, org.tensorflow.proto.distruntime.ClusterDefOrBuilder> clusterDefBuilder_;
-    /**
-     * <pre>
-     * Optional list of all workers to use in this session.
-     * </pre>
-     *
-     * <code>.tensorflow.ClusterDef cluster_def = 14;</code>
-     */
-    public boolean hasClusterDef() {
-      return clusterDefBuilder_ != null || clusterDef_ != null;
-    }
-    /**
-     * <pre>
-     * Optional list of all workers to use in this session.
-     * </pre>
-     *
-     * <code>.tensorflow.ClusterDef cluster_def = 14;</code>
-     */
-    public org.tensorflow.proto.distruntime.ClusterDef getClusterDef() {
-      if (clusterDefBuilder_ == null) {
-        return clusterDef_ == null ? org.tensorflow.proto.distruntime.ClusterDef.getDefaultInstance() : clusterDef_;
-      } else {
-        return clusterDefBuilder_.getMessage();
-      }
-    }
-    /**
-     * <pre>
-     * Optional list of all workers to use in this session.
-     * </pre>
-     *
-     * <code>.tensorflow.ClusterDef cluster_def = 14;</code>
-     */
-    public Builder setClusterDef(org.tensorflow.proto.distruntime.ClusterDef value) {
-      if (clusterDefBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        clusterDef_ = value;
-        onChanged();
-      } else {
-        clusterDefBuilder_.setMessage(value);
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Optional list of all workers to use in this session.
-     * </pre>
-     *
-     * <code>.tensorflow.ClusterDef cluster_def = 14;</code>
-     */
-    public Builder setClusterDef(
-        org.tensorflow.proto.distruntime.ClusterDef.Builder builderForValue) {
-      if (clusterDefBuilder_ == null) {
-        clusterDef_ = builderForValue.build();
-        onChanged();
-      } else {
-        clusterDefBuilder_.setMessage(builderForValue.build());
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Optional list of all workers to use in this session.
-     * </pre>
-     *
-     * <code>.tensorflow.ClusterDef cluster_def = 14;</code>
-     */
-    public Builder mergeClusterDef(org.tensorflow.proto.distruntime.ClusterDef value) {
-      if (clusterDefBuilder_ == null) {
-        if (clusterDef_ != null) {
-          clusterDef_ =
-            org.tensorflow.proto.distruntime.ClusterDef.newBuilder(clusterDef_).mergeFrom(value).buildPartial();
-        } else {
-          clusterDef_ = value;
-        }
-        onChanged();
-      } else {
-        clusterDefBuilder_.mergeFrom(value);
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Optional list of all workers to use in this session.
-     * </pre>
-     *
-     * <code>.tensorflow.ClusterDef cluster_def = 14;</code>
-     */
-    public Builder clearClusterDef() {
-      if (clusterDefBuilder_ == null) {
-        clusterDef_ = null;
-        onChanged();
-      } else {
-        clusterDef_ = null;
-        clusterDefBuilder_ = null;
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Optional list of all workers to use in this session.
-     * </pre>
-     *
-     * <code>.tensorflow.ClusterDef cluster_def = 14;</code>
-     */
-    public org.tensorflow.proto.distruntime.ClusterDef.Builder getClusterDefBuilder() {
-      
-      onChanged();
-      return getClusterDefFieldBuilder().getBuilder();
-    }
-    /**
-     * <pre>
-     * Optional list of all workers to use in this session.
-     * </pre>
-     *
-     * <code>.tensorflow.ClusterDef cluster_def = 14;</code>
-     */
-    public org.tensorflow.proto.distruntime.ClusterDefOrBuilder getClusterDefOrBuilder() {
-      if (clusterDefBuilder_ != null) {
-        return clusterDefBuilder_.getMessageOrBuilder();
-      } else {
-        return clusterDef_ == null ?
-            org.tensorflow.proto.distruntime.ClusterDef.getDefaultInstance() : clusterDef_;
-      }
-    }
-    /**
-     * <pre>
-     * Optional list of all workers to use in this session.
-     * </pre>
-     *
-     * <code>.tensorflow.ClusterDef cluster_def = 14;</code>
-     */
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.distruntime.ClusterDef, org.tensorflow.proto.distruntime.ClusterDef.Builder, org.tensorflow.proto.distruntime.ClusterDefOrBuilder> 
-        getClusterDefFieldBuilder() {
-      if (clusterDefBuilder_ == null) {
-        clusterDefBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-            org.tensorflow.proto.distruntime.ClusterDef, org.tensorflow.proto.distruntime.ClusterDef.Builder, org.tensorflow.proto.distruntime.ClusterDefOrBuilder>(
-                getClusterDef(),
-                getParentForChildren(),
-                isClean());
-        clusterDef_ = null;
-      }
-      return clusterDefBuilder_;
-    }
-
-    private boolean isolateSessionState_ ;
-    /**
-     * <pre>
-     * If true, any resources such as Variables used in the session will not be
-     * shared with other sessions. However, when clusterspec propagation is
-     * enabled, this field is ignored and sessions are always isolated.
-     * </pre>
-     *
-     * <code>bool isolate_session_state = 15;</code>
-     */
-    public boolean getIsolateSessionState() {
-      return isolateSessionState_;
-    }
-    /**
-     * <pre>
-     * If true, any resources such as Variables used in the session will not be
-     * shared with other sessions. However, when clusterspec propagation is
-     * enabled, this field is ignored and sessions are always isolated.
-     * </pre>
-     *
-     * <code>bool isolate_session_state = 15;</code>
-     */
-    public Builder setIsolateSessionState(boolean value) {
-      
-      isolateSessionState_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * If true, any resources such as Variables used in the session will not be
-     * shared with other sessions. However, when clusterspec propagation is
-     * enabled, this field is ignored and sessions are always isolated.
-     * </pre>
-     *
-     * <code>bool isolate_session_state = 15;</code>
-     */
-    public Builder clearIsolateSessionState() {
-      
-      isolateSessionState_ = false;
-      onChanged();
-      return this;
-    }
-
-    private boolean shareClusterDevicesInSession_ ;
-    /**
-     * <pre>
-     * When true, WorkerSessions are created with device attributes from the
-     * full cluster.
-     * This is helpful when a worker wants to partition a graph
-     * (for example during a PartitionedCallOp).
-     * </pre>
-     *
-     * <code>bool share_cluster_devices_in_session = 17;</code>
-     */
-    public boolean getShareClusterDevicesInSession() {
-      return shareClusterDevicesInSession_;
-    }
-    /**
-     * <pre>
-     * When true, WorkerSessions are created with device attributes from the
-     * full cluster.
-     * This is helpful when a worker wants to partition a graph
-     * (for example during a PartitionedCallOp).
-     * </pre>
-     *
-     * <code>bool share_cluster_devices_in_session = 17;</code>
-     */
-    public Builder setShareClusterDevicesInSession(boolean value) {
-      
-      shareClusterDevicesInSession_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * When true, WorkerSessions are created with device attributes from the
-     * full cluster.
-     * This is helpful when a worker wants to partition a graph
-     * (for example during a PartitionedCallOp).
-     * </pre>
-     *
-     * <code>bool share_cluster_devices_in_session = 17;</code>
-     */
-    public Builder clearShareClusterDevicesInSession() {
-      
-      shareClusterDevicesInSession_ = false;
-      onChanged();
-      return this;
-    }
-
-    private org.tensorflow.proto.framework.ConfigProto.Experimental experimental_;
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.ConfigProto.Experimental, org.tensorflow.proto.framework.ConfigProto.Experimental.Builder, org.tensorflow.proto.framework.ConfigProto.ExperimentalOrBuilder> experimentalBuilder_;
-    /**
-     * <code>.tensorflow.ConfigProto.Experimental experimental = 16;</code>
-     */
-    public boolean hasExperimental() {
-      return experimentalBuilder_ != null || experimental_ != null;
-    }
-    /**
-     * <code>.tensorflow.ConfigProto.Experimental experimental = 16;</code>
-     */
-    public org.tensorflow.proto.framework.ConfigProto.Experimental getExperimental() {
-      if (experimentalBuilder_ == null) {
-        return experimental_ == null ? org.tensorflow.proto.framework.ConfigProto.Experimental.getDefaultInstance() : experimental_;
-      } else {
-        return experimentalBuilder_.getMessage();
-      }
-    }
-    /**
-     * <code>.tensorflow.ConfigProto.Experimental experimental = 16;</code>
-     */
-    public Builder setExperimental(org.tensorflow.proto.framework.ConfigProto.Experimental value) {
-      if (experimentalBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        experimental_ = value;
-        onChanged();
-      } else {
-        experimentalBuilder_.setMessage(value);
-      }
-
-      return this;
-    }
-    /**
-     * <code>.tensorflow.ConfigProto.Experimental experimental = 16;</code>
-     */
-    public Builder setExperimental(
-        org.tensorflow.proto.framework.ConfigProto.Experimental.Builder builderForValue) {
-      if (experimentalBuilder_ == null) {
-        experimental_ = builderForValue.build();
-        onChanged();
-      } else {
-        experimentalBuilder_.setMessage(builderForValue.build());
-      }
-
-      return this;
-    }
-    /**
-     * <code>.tensorflow.ConfigProto.Experimental experimental = 16;</code>
-     */
-    public Builder mergeExperimental(org.tensorflow.proto.framework.ConfigProto.Experimental value) {
-      if (experimentalBuilder_ == null) {
-        if (experimental_ != null) {
-          experimental_ =
-            org.tensorflow.proto.framework.ConfigProto.Experimental.newBuilder(experimental_).mergeFrom(value).buildPartial();
-        } else {
-          experimental_ = value;
-        }
-        onChanged();
-      } else {
-        experimentalBuilder_.mergeFrom(value);
-      }
-
-      return this;
-    }
-    /**
-     * <code>.tensorflow.ConfigProto.Experimental experimental = 16;</code>
-     */
-    public Builder clearExperimental() {
-      if (experimentalBuilder_ == null) {
-        experimental_ = null;
-        onChanged();
-      } else {
-        experimental_ = null;
-        experimentalBuilder_ = null;
-      }
-
-      return this;
-    }
-    /**
-     * <code>.tensorflow.ConfigProto.Experimental experimental = 16;</code>
-     */
-    public org.tensorflow.proto.framework.ConfigProto.Experimental.Builder getExperimentalBuilder() {
-      
-      onChanged();
-      return getExperimentalFieldBuilder().getBuilder();
-    }
-    /**
-     * <code>.tensorflow.ConfigProto.Experimental experimental = 16;</code>
-     */
-    public org.tensorflow.proto.framework.ConfigProto.ExperimentalOrBuilder getExperimentalOrBuilder() {
-      if (experimentalBuilder_ != null) {
-        return experimentalBuilder_.getMessageOrBuilder();
-      } else {
-        return experimental_ == null ?
-            org.tensorflow.proto.framework.ConfigProto.Experimental.getDefaultInstance() : experimental_;
-      }
-    }
-    /**
-     * <code>.tensorflow.ConfigProto.Experimental experimental = 16;</code>
-     */
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.ConfigProto.Experimental, org.tensorflow.proto.framework.ConfigProto.Experimental.Builder, org.tensorflow.proto.framework.ConfigProto.ExperimentalOrBuilder> 
-        getExperimentalFieldBuilder() {
-      if (experimentalBuilder_ == null) {
-        experimentalBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-            org.tensorflow.proto.framework.ConfigProto.Experimental, org.tensorflow.proto.framework.ConfigProto.Experimental.Builder, org.tensorflow.proto.framework.ConfigProto.ExperimentalOrBuilder>(
-                getExperimental(),
-                getParentForChildren(),
-                isClean());
-        experimental_ = null;
-      }
-      return experimentalBuilder_;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.ConfigProto)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.ConfigProto)
-  private static final org.tensorflow.proto.framework.ConfigProto DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.ConfigProto();
-  }
-
-  public static org.tensorflow.proto.framework.ConfigProto getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<ConfigProto>
-      PARSER = new com.google.protobuf.AbstractParser<ConfigProto>() {
-    @java.lang.Override
-    public ConfigProto parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new ConfigProto(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<ConfigProto> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<ConfigProto> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.framework.ConfigProto getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ConfigProtoOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ConfigProtoOrBuilder.java
deleted file mode 100644
index a13298b2bb5..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ConfigProtoOrBuilder.java
+++ /dev/null
@@ -1,477 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/config.proto
-
-package org.tensorflow.proto.framework;
-
-public interface ConfigProtoOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.ConfigProto)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <pre>
-   * Map from device type name (e.g., "CPU" or "GPU" ) to maximum
-   * number of devices of that type to use.  If a particular device
-   * type is not found in the map, the system picks an appropriate
-   * number.
-   * </pre>
-   *
-   * <code>map&lt;string, int32&gt; device_count = 1;</code>
-   */
-  int getDeviceCountCount();
-  /**
-   * <pre>
-   * Map from device type name (e.g., "CPU" or "GPU" ) to maximum
-   * number of devices of that type to use.  If a particular device
-   * type is not found in the map, the system picks an appropriate
-   * number.
-   * </pre>
-   *
-   * <code>map&lt;string, int32&gt; device_count = 1;</code>
-   */
-  boolean containsDeviceCount(
-      java.lang.String key);
-  /**
-   * Use {@link #getDeviceCountMap()} instead.
-   */
-  @java.lang.Deprecated
-  java.util.Map<java.lang.String, java.lang.Integer>
-  getDeviceCount();
-  /**
-   * <pre>
-   * Map from device type name (e.g., "CPU" or "GPU" ) to maximum
-   * number of devices of that type to use.  If a particular device
-   * type is not found in the map, the system picks an appropriate
-   * number.
-   * </pre>
-   *
-   * <code>map&lt;string, int32&gt; device_count = 1;</code>
-   */
-  java.util.Map<java.lang.String, java.lang.Integer>
-  getDeviceCountMap();
-  /**
-   * <pre>
-   * Map from device type name (e.g., "CPU" or "GPU" ) to maximum
-   * number of devices of that type to use.  If a particular device
-   * type is not found in the map, the system picks an appropriate
-   * number.
-   * </pre>
-   *
-   * <code>map&lt;string, int32&gt; device_count = 1;</code>
-   */
-
-  int getDeviceCountOrDefault(
-      java.lang.String key,
-      int defaultValue);
-  /**
-   * <pre>
-   * Map from device type name (e.g., "CPU" or "GPU" ) to maximum
-   * number of devices of that type to use.  If a particular device
-   * type is not found in the map, the system picks an appropriate
-   * number.
-   * </pre>
-   *
-   * <code>map&lt;string, int32&gt; device_count = 1;</code>
-   */
-
-  int getDeviceCountOrThrow(
-      java.lang.String key);
-
-  /**
-   * <pre>
-   * The execution of an individual op (for some op types) can be
-   * parallelized on a pool of intra_op_parallelism_threads.
-   * 0 means the system picks an appropriate number.
-   * If you create an ordinary session, e.g., from Python or C++,
-   * then there is exactly one intra op thread pool per process.
-   * The first session created determines the number of threads in this pool.
-   * All subsequent sessions reuse/share this one global pool.
-   * There are notable exceptions to the default behavior describe above:
-   * 1. There is an environment variable  for overriding this thread pool,
-   *    named TF_OVERRIDE_GLOBAL_THREADPOOL.
-   * 2. When connecting to a server, such as a remote `tf.train.Server`
-   *    instance, then this option will be ignored altogether.
-   * </pre>
-   *
-   * <code>int32 intra_op_parallelism_threads = 2;</code>
-   */
-  int getIntraOpParallelismThreads();
-
-  /**
-   * <pre>
-   * Nodes that perform blocking operations are enqueued on a pool of
-   * inter_op_parallelism_threads available in each process.
-   * 0 means the system picks an appropriate number.
-   * Negative means all operations are performed in caller's thread.
-   * Note that the first Session created in the process sets the
-   * number of threads for all future sessions unless use_per_session_threads is
-   * true or session_inter_op_thread_pool is configured.
-   * </pre>
-   *
-   * <code>int32 inter_op_parallelism_threads = 5;</code>
-   */
-  int getInterOpParallelismThreads();
-
-  /**
-   * <pre>
-   * If true, use a new set of threads for this session rather than the global
-   * pool of threads. Only supported by direct sessions.
-   * If false, use the global threads created by the first session, or the
-   * per-session thread pools configured by session_inter_op_thread_pool.
-   * This option is deprecated. The same effect can be achieved by setting
-   * session_inter_op_thread_pool to have one element, whose num_threads equals
-   * inter_op_parallelism_threads.
-   * </pre>
-   *
-   * <code>bool use_per_session_threads = 9;</code>
-   */
-  boolean getUsePerSessionThreads();
-
-  /**
-   * <pre>
-   * This option is experimental - it may be replaced with a different mechanism
-   * in the future.
-   * Configures session thread pools. If this is configured, then RunOptions for
-   * a Run call can select the thread pool to use.
-   * The intended use is for when some session invocations need to run in a
-   * background pool limited to a small number of threads:
-   * - For example, a session may be configured to have one large pool (for
-   * regular compute) and one small pool (for periodic, low priority work);
-   * using the small pool is currently the mechanism for limiting the inter-op
-   * parallelism of the low priority work.  Note that it does not limit the
-   * parallelism of work spawned by a single op kernel implementation.
-   * - Using this setting is normally not needed in training, but may help some
-   * serving use cases.
-   * - It is also generally recommended to set the global_name field of this
-   * proto, to avoid creating multiple large pools. It is typically better to
-   * run the non-low-priority work, even across sessions, in a single large
-   * pool.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.ThreadPoolOptionProto session_inter_op_thread_pool = 12;</code>
-   */
-  java.util.List<org.tensorflow.proto.framework.ThreadPoolOptionProto> 
-      getSessionInterOpThreadPoolList();
-  /**
-   * <pre>
-   * This option is experimental - it may be replaced with a different mechanism
-   * in the future.
-   * Configures session thread pools. If this is configured, then RunOptions for
-   * a Run call can select the thread pool to use.
-   * The intended use is for when some session invocations need to run in a
-   * background pool limited to a small number of threads:
-   * - For example, a session may be configured to have one large pool (for
-   * regular compute) and one small pool (for periodic, low priority work);
-   * using the small pool is currently the mechanism for limiting the inter-op
-   * parallelism of the low priority work.  Note that it does not limit the
-   * parallelism of work spawned by a single op kernel implementation.
-   * - Using this setting is normally not needed in training, but may help some
-   * serving use cases.
-   * - It is also generally recommended to set the global_name field of this
-   * proto, to avoid creating multiple large pools. It is typically better to
-   * run the non-low-priority work, even across sessions, in a single large
-   * pool.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.ThreadPoolOptionProto session_inter_op_thread_pool = 12;</code>
-   */
-  org.tensorflow.proto.framework.ThreadPoolOptionProto getSessionInterOpThreadPool(int index);
-  /**
-   * <pre>
-   * This option is experimental - it may be replaced with a different mechanism
-   * in the future.
-   * Configures session thread pools. If this is configured, then RunOptions for
-   * a Run call can select the thread pool to use.
-   * The intended use is for when some session invocations need to run in a
-   * background pool limited to a small number of threads:
-   * - For example, a session may be configured to have one large pool (for
-   * regular compute) and one small pool (for periodic, low priority work);
-   * using the small pool is currently the mechanism for limiting the inter-op
-   * parallelism of the low priority work.  Note that it does not limit the
-   * parallelism of work spawned by a single op kernel implementation.
-   * - Using this setting is normally not needed in training, but may help some
-   * serving use cases.
-   * - It is also generally recommended to set the global_name field of this
-   * proto, to avoid creating multiple large pools. It is typically better to
-   * run the non-low-priority work, even across sessions, in a single large
-   * pool.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.ThreadPoolOptionProto session_inter_op_thread_pool = 12;</code>
-   */
-  int getSessionInterOpThreadPoolCount();
-  /**
-   * <pre>
-   * This option is experimental - it may be replaced with a different mechanism
-   * in the future.
-   * Configures session thread pools. If this is configured, then RunOptions for
-   * a Run call can select the thread pool to use.
-   * The intended use is for when some session invocations need to run in a
-   * background pool limited to a small number of threads:
-   * - For example, a session may be configured to have one large pool (for
-   * regular compute) and one small pool (for periodic, low priority work);
-   * using the small pool is currently the mechanism for limiting the inter-op
-   * parallelism of the low priority work.  Note that it does not limit the
-   * parallelism of work spawned by a single op kernel implementation.
-   * - Using this setting is normally not needed in training, but may help some
-   * serving use cases.
-   * - It is also generally recommended to set the global_name field of this
-   * proto, to avoid creating multiple large pools. It is typically better to
-   * run the non-low-priority work, even across sessions, in a single large
-   * pool.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.ThreadPoolOptionProto session_inter_op_thread_pool = 12;</code>
-   */
-  java.util.List<? extends org.tensorflow.proto.framework.ThreadPoolOptionProtoOrBuilder> 
-      getSessionInterOpThreadPoolOrBuilderList();
-  /**
-   * <pre>
-   * This option is experimental - it may be replaced with a different mechanism
-   * in the future.
-   * Configures session thread pools. If this is configured, then RunOptions for
-   * a Run call can select the thread pool to use.
-   * The intended use is for when some session invocations need to run in a
-   * background pool limited to a small number of threads:
-   * - For example, a session may be configured to have one large pool (for
-   * regular compute) and one small pool (for periodic, low priority work);
-   * using the small pool is currently the mechanism for limiting the inter-op
-   * parallelism of the low priority work.  Note that it does not limit the
-   * parallelism of work spawned by a single op kernel implementation.
-   * - Using this setting is normally not needed in training, but may help some
-   * serving use cases.
-   * - It is also generally recommended to set the global_name field of this
-   * proto, to avoid creating multiple large pools. It is typically better to
-   * run the non-low-priority work, even across sessions, in a single large
-   * pool.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.ThreadPoolOptionProto session_inter_op_thread_pool = 12;</code>
-   */
-  org.tensorflow.proto.framework.ThreadPoolOptionProtoOrBuilder getSessionInterOpThreadPoolOrBuilder(
-      int index);
-
-  /**
-   * <pre>
-   * Assignment of Nodes to Devices is recomputed every placement_period
-   * steps until the system warms up (at which point the recomputation
-   * typically slows down automatically).
-   * </pre>
-   *
-   * <code>int32 placement_period = 3;</code>
-   */
-  int getPlacementPeriod();
-
-  /**
-   * <pre>
-   * When any filters are present sessions will ignore all devices which do not
-   * match the filters. Each filter can be partially specified, e.g. "/job:ps"
-   * "/job:worker/replica:3", etc.
-   * </pre>
-   *
-   * <code>repeated string device_filters = 4;</code>
-   */
-  java.util.List<java.lang.String>
-      getDeviceFiltersList();
-  /**
-   * <pre>
-   * When any filters are present sessions will ignore all devices which do not
-   * match the filters. Each filter can be partially specified, e.g. "/job:ps"
-   * "/job:worker/replica:3", etc.
-   * </pre>
-   *
-   * <code>repeated string device_filters = 4;</code>
-   */
-  int getDeviceFiltersCount();
-  /**
-   * <pre>
-   * When any filters are present sessions will ignore all devices which do not
-   * match the filters. Each filter can be partially specified, e.g. "/job:ps"
-   * "/job:worker/replica:3", etc.
-   * </pre>
-   *
-   * <code>repeated string device_filters = 4;</code>
-   */
-  java.lang.String getDeviceFilters(int index);
-  /**
-   * <pre>
-   * When any filters are present sessions will ignore all devices which do not
-   * match the filters. Each filter can be partially specified, e.g. "/job:ps"
-   * "/job:worker/replica:3", etc.
-   * </pre>
-   *
-   * <code>repeated string device_filters = 4;</code>
-   */
-  com.google.protobuf.ByteString
-      getDeviceFiltersBytes(int index);
-
-  /**
-   * <pre>
-   * Options that apply to all GPUs.
-   * </pre>
-   *
-   * <code>.tensorflow.GPUOptions gpu_options = 6;</code>
-   */
-  boolean hasGpuOptions();
-  /**
-   * <pre>
-   * Options that apply to all GPUs.
-   * </pre>
-   *
-   * <code>.tensorflow.GPUOptions gpu_options = 6;</code>
-   */
-  org.tensorflow.proto.framework.GPUOptions getGpuOptions();
-  /**
-   * <pre>
-   * Options that apply to all GPUs.
-   * </pre>
-   *
-   * <code>.tensorflow.GPUOptions gpu_options = 6;</code>
-   */
-  org.tensorflow.proto.framework.GPUOptionsOrBuilder getGpuOptionsOrBuilder();
-
-  /**
-   * <pre>
-   * Whether soft placement is allowed. If allow_soft_placement is true,
-   * an op will be placed on CPU if
-   *   1. there's no GPU implementation for the OP
-   * or
-   *   2. no GPU devices are known or registered
-   * or
-   *   3. need to co-locate with reftype input(s) which are from CPU.
-   * </pre>
-   *
-   * <code>bool allow_soft_placement = 7;</code>
-   */
-  boolean getAllowSoftPlacement();
-
-  /**
-   * <pre>
-   * Whether device placements should be logged.
-   * </pre>
-   *
-   * <code>bool log_device_placement = 8;</code>
-   */
-  boolean getLogDevicePlacement();
-
-  /**
-   * <pre>
-   * Options that apply to all graphs.
-   * </pre>
-   *
-   * <code>.tensorflow.GraphOptions graph_options = 10;</code>
-   */
-  boolean hasGraphOptions();
-  /**
-   * <pre>
-   * Options that apply to all graphs.
-   * </pre>
-   *
-   * <code>.tensorflow.GraphOptions graph_options = 10;</code>
-   */
-  org.tensorflow.proto.framework.GraphOptions getGraphOptions();
-  /**
-   * <pre>
-   * Options that apply to all graphs.
-   * </pre>
-   *
-   * <code>.tensorflow.GraphOptions graph_options = 10;</code>
-   */
-  org.tensorflow.proto.framework.GraphOptionsOrBuilder getGraphOptionsOrBuilder();
-
-  /**
-   * <pre>
-   * Global timeout for all blocking operations in this session.  If non-zero,
-   * and not overridden on a per-operation basis, this value will be used as the
-   * deadline for all blocking operations.
-   * </pre>
-   *
-   * <code>int64 operation_timeout_in_ms = 11;</code>
-   */
-  long getOperationTimeoutInMs();
-
-  /**
-   * <pre>
-   * Options that apply when this session uses the distributed runtime.
-   * </pre>
-   *
-   * <code>.tensorflow.RPCOptions rpc_options = 13;</code>
-   */
-  boolean hasRpcOptions();
-  /**
-   * <pre>
-   * Options that apply when this session uses the distributed runtime.
-   * </pre>
-   *
-   * <code>.tensorflow.RPCOptions rpc_options = 13;</code>
-   */
-  org.tensorflow.proto.framework.RPCOptions getRpcOptions();
-  /**
-   * <pre>
-   * Options that apply when this session uses the distributed runtime.
-   * </pre>
-   *
-   * <code>.tensorflow.RPCOptions rpc_options = 13;</code>
-   */
-  org.tensorflow.proto.framework.RPCOptionsOrBuilder getRpcOptionsOrBuilder();
-
-  /**
-   * <pre>
-   * Optional list of all workers to use in this session.
-   * </pre>
-   *
-   * <code>.tensorflow.ClusterDef cluster_def = 14;</code>
-   */
-  boolean hasClusterDef();
-  /**
-   * <pre>
-   * Optional list of all workers to use in this session.
-   * </pre>
-   *
-   * <code>.tensorflow.ClusterDef cluster_def = 14;</code>
-   */
-  org.tensorflow.proto.distruntime.ClusterDef getClusterDef();
-  /**
-   * <pre>
-   * Optional list of all workers to use in this session.
-   * </pre>
-   *
-   * <code>.tensorflow.ClusterDef cluster_def = 14;</code>
-   */
-  org.tensorflow.proto.distruntime.ClusterDefOrBuilder getClusterDefOrBuilder();
-
-  /**
-   * <pre>
-   * If true, any resources such as Variables used in the session will not be
-   * shared with other sessions. However, when clusterspec propagation is
-   * enabled, this field is ignored and sessions are always isolated.
-   * </pre>
-   *
-   * <code>bool isolate_session_state = 15;</code>
-   */
-  boolean getIsolateSessionState();
-
-  /**
-   * <pre>
-   * When true, WorkerSessions are created with device attributes from the
-   * full cluster.
-   * This is helpful when a worker wants to partition a graph
-   * (for example during a PartitionedCallOp).
-   * </pre>
-   *
-   * <code>bool share_cluster_devices_in_session = 17;</code>
-   */
-  boolean getShareClusterDevicesInSession();
-
-  /**
-   * <code>.tensorflow.ConfigProto.Experimental experimental = 16;</code>
-   */
-  boolean hasExperimental();
-  /**
-   * <code>.tensorflow.ConfigProto.Experimental experimental = 16;</code>
-   */
-  org.tensorflow.proto.framework.ConfigProto.Experimental getExperimental();
-  /**
-   * <code>.tensorflow.ConfigProto.Experimental experimental = 16;</code>
-   */
-  org.tensorflow.proto.framework.ConfigProto.ExperimentalOrBuilder getExperimentalOrBuilder();
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ConfigProtos.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ConfigProtos.java
deleted file mode 100644
index 9a2c79552d6..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ConfigProtos.java
+++ /dev/null
@@ -1,393 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/config.proto
-
-package org.tensorflow.proto.framework;
-
-public final class ConfigProtos {
-  private ConfigProtos() {}
-  public static void registerAllExtensions(
-      com.google.protobuf.ExtensionRegistryLite registry) {
-  }
-
-  public static void registerAllExtensions(
-      com.google.protobuf.ExtensionRegistry registry) {
-    registerAllExtensions(
-        (com.google.protobuf.ExtensionRegistryLite) registry);
-  }
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_GPUOptions_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_GPUOptions_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_GPUOptions_Experimental_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_GPUOptions_Experimental_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_GPUOptions_Experimental_VirtualDevices_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_GPUOptions_Experimental_VirtualDevices_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_OptimizerOptions_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_OptimizerOptions_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_GraphOptions_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_GraphOptions_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_ThreadPoolOptionProto_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_ThreadPoolOptionProto_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_RPCOptions_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_RPCOptions_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_SessionMetadata_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_SessionMetadata_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_ConfigProto_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_ConfigProto_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_ConfigProto_DeviceCountEntry_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_ConfigProto_DeviceCountEntry_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_ConfigProto_Experimental_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_ConfigProto_Experimental_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_RunOptions_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_RunOptions_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_RunOptions_Experimental_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_RunOptions_Experimental_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_RunOptions_Experimental_RunHandlerPoolOptions_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_RunOptions_Experimental_RunHandlerPoolOptions_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_RunMetadata_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_RunMetadata_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_RunMetadata_FunctionGraphs_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_RunMetadata_FunctionGraphs_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_TensorConnection_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_TensorConnection_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_CallableOptions_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_CallableOptions_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_CallableOptions_FeedDevicesEntry_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_CallableOptions_FeedDevicesEntry_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_CallableOptions_FetchDevicesEntry_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_CallableOptions_FetchDevicesEntry_fieldAccessorTable;
-
-  public static com.google.protobuf.Descriptors.FileDescriptor
-      getDescriptor() {
-    return descriptor;
-  }
-  private static  com.google.protobuf.Descriptors.FileDescriptor
-      descriptor;
-  static {
-    java.lang.String[] descriptorData = {
-      "\n%tensorflow/core/protobuf/config.proto\022" +
-      "\ntensorflow\032*tensorflow/core/framework/c" +
-      "ost_graph.proto\032%tensorflow/core/framewo" +
-      "rk/graph.proto\032*tensorflow/core/framewor" +
-      "k/step_stats.proto\032&tensorflow/core/prot" +
-      "obuf/cluster.proto\032$tensorflow/core/prot" +
-      "obuf/debug.proto\032.tensorflow/core/protob" +
-      "uf/rewriter_config.proto\"\311\005\n\nGPUOptions\022" +
-      "\'\n\037per_process_gpu_memory_fraction\030\001 \001(\001" +
-      "\022\024\n\014allow_growth\030\004 \001(\010\022\026\n\016allocator_type" +
-      "\030\002 \001(\t\022\037\n\027deferred_deletion_bytes\030\003 \001(\003\022" +
-      "\033\n\023visible_device_list\030\005 \001(\t\022\"\n\032polling_" +
-      "active_delay_usecs\030\006 \001(\005\022$\n\034polling_inac" +
-      "tive_delay_msecs\030\007 \001(\005\022\034\n\024force_gpu_comp" +
-      "atible\030\010 \001(\010\0229\n\014experimental\030\t \001(\0132#.ten" +
-      "sorflow.GPUOptions.Experimental\032\202\003\n\014Expe" +
-      "rimental\022K\n\017virtual_devices\030\001 \003(\01322.tens" +
-      "orflow.GPUOptions.Experimental.VirtualDe" +
-      "vices\022\032\n\022use_unified_memory\030\002 \001(\010\022#\n\033num" +
-      "_dev_to_dev_copy_streams\030\003 \001(\005\022\035\n\025collec" +
-      "tive_ring_order\030\004 \001(\t\022\035\n\025timestamped_all" +
-      "ocator\030\005 \001(\010\022#\n\033kernel_tracker_max_inter" +
-      "val\030\007 \001(\005\022 \n\030kernel_tracker_max_bytes\030\010 " +
-      "\001(\005\022\"\n\032kernel_tracker_max_pending\030\t \001(\005\032" +
-      ";\n\016VirtualDevices\022\027\n\017memory_limit_mb\030\001 \003" +
-      "(\002\022\020\n\010priority\030\002 \003(\005\"\205\003\n\020OptimizerOption" +
-      "s\022+\n#do_common_subexpression_elimination" +
-      "\030\001 \001(\010\022\033\n\023do_constant_folding\030\002 \001(\010\022$\n\034m" +
-      "ax_folded_constant_in_bytes\030\006 \001(\003\022\034\n\024do_" +
-      "function_inlining\030\004 \001(\010\0225\n\topt_level\030\003 \001" +
-      "(\0162\".tensorflow.OptimizerOptions.Level\022E" +
-      "\n\020global_jit_level\030\005 \001(\0162+.tensorflow.Op" +
-      "timizerOptions.GlobalJitLevel\" \n\005Level\022\006" +
-      "\n\002L1\020\000\022\017\n\002L0\020\377\377\377\377\377\377\377\377\377\001\"C\n\016GlobalJitLeve" +
-      "l\022\013\n\007DEFAULT\020\000\022\020\n\003OFF\020\377\377\377\377\377\377\377\377\377\001\022\010\n\004ON_1" +
-      "\020\001\022\010\n\004ON_2\020\002\"\356\002\n\014GraphOptions\022\036\n\026enable_" +
-      "recv_scheduling\030\002 \001(\010\0227\n\021optimizer_optio" +
-      "ns\030\003 \001(\0132\034.tensorflow.OptimizerOptions\022\030" +
-      "\n\020build_cost_model\030\004 \001(\003\022\036\n\026build_cost_m" +
-      "odel_after\030\t \001(\003\022\024\n\014infer_shapes\030\005 \001(\010\022\032" +
-      "\n\022place_pruned_graph\030\006 \001(\010\022 \n\030enable_bfl" +
-      "oat16_sendrecv\030\007 \001(\010\022\025\n\rtimeline_step\030\010 " +
-      "\001(\005\0223\n\017rewrite_options\030\n \001(\0132\032.tensorflo" +
-      "w.RewriterConfigJ\004\010\001\020\002R%skip_common_sube" +
-      "xpression_elimination\"A\n\025ThreadPoolOptio" +
-      "nProto\022\023\n\013num_threads\030\001 \001(\005\022\023\n\013global_na" +
-      "me\030\002 \001(\t\"\264\001\n\nRPCOptions\022$\n\034use_rpc_for_i" +
-      "nprocess_master\030\001 \001(\010\022\035\n\025compression_alg" +
-      "orithm\030\002 \001(\t\022\031\n\021compression_level\030\003 \001(\005\022" +
-      "\032\n\022cache_rpc_response\030\004 \001(\010\022*\n\"disable_s" +
-      "ession_connection_sharing\030\005 \001(\010\"0\n\017Sessi" +
-      "onMetadata\022\014\n\004name\030\001 \001(\t\022\017\n\007version\030\002 \001(" +
-      "\003\"\310\n\n\013ConfigProto\022>\n\014device_count\030\001 \003(\0132" +
-      "(.tensorflow.ConfigProto.DeviceCountEntr" +
-      "y\022$\n\034intra_op_parallelism_threads\030\002 \001(\005\022" +
-      "$\n\034inter_op_parallelism_threads\030\005 \001(\005\022\037\n" +
-      "\027use_per_session_threads\030\t \001(\010\022G\n\034sessio" +
-      "n_inter_op_thread_pool\030\014 \003(\0132!.tensorflo" +
-      "w.ThreadPoolOptionProto\022\030\n\020placement_per" +
-      "iod\030\003 \001(\005\022\026\n\016device_filters\030\004 \003(\t\022+\n\013gpu" +
-      "_options\030\006 \001(\0132\026.tensorflow.GPUOptions\022\034" +
-      "\n\024allow_soft_placement\030\007 \001(\010\022\034\n\024log_devi" +
-      "ce_placement\030\010 \001(\010\022/\n\rgraph_options\030\n \001(" +
-      "\0132\030.tensorflow.GraphOptions\022\037\n\027operation" +
-      "_timeout_in_ms\030\013 \001(\003\022+\n\013rpc_options\030\r \001(" +
-      "\0132\026.tensorflow.RPCOptions\022+\n\013cluster_def" +
-      "\030\016 \001(\0132\026.tensorflow.ClusterDef\022\035\n\025isolat" +
-      "e_session_state\030\017 \001(\010\022(\n share_cluster_d" +
-      "evices_in_session\030\021 \001(\010\022:\n\014experimental\030" +
-      "\020 \001(\0132$.tensorflow.ConfigProto.Experimen" +
-      "tal\0322\n\020DeviceCountEntry\022\013\n\003key\030\001 \001(\t\022\r\n\005" +
-      "value\030\002 \001(\005:\0028\001\032\302\004\n\014Experimental\022\037\n\027coll" +
-      "ective_group_leader\030\001 \001(\t\022\025\n\rexecutor_ty" +
-      "pe\030\003 \001(\t\022\032\n\022recv_buf_max_chunk\030\004 \001(\005\022\031\n\021" +
-      "use_numa_affinity\030\005 \001(\010\0225\n-collective_de" +
-      "terministic_sequential_execution\030\006 \001(\010\022\027" +
-      "\n\017collective_nccl\030\007 \001(\010\0226\n.share_session" +
-      "_state_in_clusterspec_propagation\030\010 \001(\010\022" +
-      "\037\n\027disable_thread_spinning\030\t \001(\010\022(\n shar" +
-      "e_cluster_devices_in_session\030\n \001(\010\0225\n\020se" +
-      "ssion_metadata\030\013 \001(\0132\033.tensorflow.Sessio" +
-      "nMetadata\022!\n\031optimize_for_static_graph\030\014" +
-      " \001(\010\022\032\n\022enable_mlir_bridge\030\r \001(\010\022&\n\036enab" +
-      "le_mlir_graph_optimization\030\020 \001(\010\022\'\n\037disa" +
-      "ble_output_partition_graphs\030\016 \001(\010\022#\n\033xla" +
-      "_fusion_autotuner_thresh\030\017 \001(\003J\004\010\002\020\003\"\341\004\n" +
-      "\nRunOptions\0226\n\013trace_level\030\001 \001(\0162!.tenso" +
-      "rflow.RunOptions.TraceLevel\022\025\n\rtimeout_i" +
-      "n_ms\030\002 \001(\003\022\034\n\024inter_op_thread_pool\030\003 \001(\005" +
-      "\022\037\n\027output_partition_graphs\030\005 \001(\010\022/\n\rdeb" +
-      "ug_options\030\006 \001(\0132\030.tensorflow.DebugOptio" +
-      "ns\022*\n\"report_tensor_allocations_upon_oom" +
-      "\030\007 \001(\010\0229\n\014experimental\030\010 \001(\0132#.tensorflo" +
-      "w.RunOptions.Experimental\032\322\001\n\014Experiment" +
-      "al\022\034\n\024collective_graph_key\030\001 \001(\003\022\034\n\024use_" +
-      "run_handler_pool\030\002 \001(\010\022[\n\030run_handler_po" +
-      "ol_options\030\003 \001(\01329.tensorflow.RunOptions" +
-      ".Experimental.RunHandlerPoolOptions\032)\n\025R" +
-      "unHandlerPoolOptions\022\020\n\010priority\030\001 \001(\003\"R" +
-      "\n\nTraceLevel\022\014\n\010NO_TRACE\020\000\022\022\n\016SOFTWARE_T" +
-      "RACE\020\001\022\022\n\016HARDWARE_TRACE\020\002\022\016\n\nFULL_TRACE" +
-      "\020\003J\004\010\004\020\005\"\207\003\n\013RunMetadata\022)\n\nstep_stats\030\001" +
-      " \001(\0132\025.tensorflow.StepStats\022,\n\ncost_grap" +
-      "h\030\002 \001(\0132\030.tensorflow.CostGraphDef\022.\n\020par" +
-      "tition_graphs\030\003 \003(\0132\024.tensorflow.GraphDe" +
-      "f\022?\n\017function_graphs\030\004 \003(\0132&.tensorflow." +
-      "RunMetadata.FunctionGraphs\032\255\001\n\016FunctionG" +
-      "raphs\022.\n\020partition_graphs\030\001 \003(\0132\024.tensor" +
-      "flow.GraphDef\0224\n\026pre_optimization_graph\030" +
-      "\002 \001(\0132\024.tensorflow.GraphDef\0225\n\027post_opti" +
-      "mization_graph\030\003 \001(\0132\024.tensorflow.GraphD" +
-      "ef\":\n\020TensorConnection\022\023\n\013from_tensor\030\001 " +
-      "\001(\t\022\021\n\tto_tensor\030\002 \001(\t\"\260\003\n\017CallableOptio" +
-      "ns\022\014\n\004feed\030\001 \003(\t\022\r\n\005fetch\030\002 \003(\t\022\016\n\006targe" +
-      "t\030\003 \003(\t\022+\n\013run_options\030\004 \001(\0132\026.tensorflo" +
-      "w.RunOptions\0227\n\021tensor_connection\030\005 \003(\0132" +
-      "\034.tensorflow.TensorConnection\022B\n\014feed_de" +
-      "vices\030\006 \003(\0132,.tensorflow.CallableOptions" +
-      ".FeedDevicesEntry\022D\n\rfetch_devices\030\007 \003(\013" +
-      "2-.tensorflow.CallableOptions.FetchDevic" +
-      "esEntry\022\027\n\017fetch_skip_sync\030\010 \001(\010\0322\n\020Feed" +
-      "DevicesEntry\022\013\n\003key\030\001 \001(\t\022\r\n\005value\030\002 \001(\t" +
-      ":\0028\001\0323\n\021FetchDevicesEntry\022\013\n\003key\030\001 \001(\t\022\r" +
-      "\n\005value\030\002 \001(\t:\0028\001B}\n\036org.tensorflow.prot" +
-      "o.frameworkB\014ConfigProtosP\001ZHgithub.com/" +
-      "tensorflow/tensorflow/tensorflow/go/core" +
-      "/core_protos_go_proto\370\001\001b\006proto3"
-    };
-    descriptor = com.google.protobuf.Descriptors.FileDescriptor
-      .internalBuildGeneratedFileFrom(descriptorData,
-        new com.google.protobuf.Descriptors.FileDescriptor[] {
-          org.tensorflow.proto.framework.CostGraphProtos.getDescriptor(),
-          org.tensorflow.proto.framework.GraphProtos.getDescriptor(),
-          org.tensorflow.proto.framework.StepStatsProtos.getDescriptor(),
-          org.tensorflow.proto.distruntime.ClusterProtos.getDescriptor(),
-          org.tensorflow.proto.framework.DebugProtos.getDescriptor(),
-          org.tensorflow.proto.framework.RewriterConfigProtos.getDescriptor(),
-        });
-    internal_static_tensorflow_GPUOptions_descriptor =
-      getDescriptor().getMessageTypes().get(0);
-    internal_static_tensorflow_GPUOptions_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_GPUOptions_descriptor,
-        new java.lang.String[] { "PerProcessGpuMemoryFraction", "AllowGrowth", "AllocatorType", "DeferredDeletionBytes", "VisibleDeviceList", "PollingActiveDelayUsecs", "PollingInactiveDelayMsecs", "ForceGpuCompatible", "Experimental", });
-    internal_static_tensorflow_GPUOptions_Experimental_descriptor =
-      internal_static_tensorflow_GPUOptions_descriptor.getNestedTypes().get(0);
-    internal_static_tensorflow_GPUOptions_Experimental_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_GPUOptions_Experimental_descriptor,
-        new java.lang.String[] { "VirtualDevices", "UseUnifiedMemory", "NumDevToDevCopyStreams", "CollectiveRingOrder", "TimestampedAllocator", "KernelTrackerMaxInterval", "KernelTrackerMaxBytes", "KernelTrackerMaxPending", });
-    internal_static_tensorflow_GPUOptions_Experimental_VirtualDevices_descriptor =
-      internal_static_tensorflow_GPUOptions_Experimental_descriptor.getNestedTypes().get(0);
-    internal_static_tensorflow_GPUOptions_Experimental_VirtualDevices_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_GPUOptions_Experimental_VirtualDevices_descriptor,
-        new java.lang.String[] { "MemoryLimitMb", "Priority", });
-    internal_static_tensorflow_OptimizerOptions_descriptor =
-      getDescriptor().getMessageTypes().get(1);
-    internal_static_tensorflow_OptimizerOptions_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_OptimizerOptions_descriptor,
-        new java.lang.String[] { "DoCommonSubexpressionElimination", "DoConstantFolding", "MaxFoldedConstantInBytes", "DoFunctionInlining", "OptLevel", "GlobalJitLevel", });
-    internal_static_tensorflow_GraphOptions_descriptor =
-      getDescriptor().getMessageTypes().get(2);
-    internal_static_tensorflow_GraphOptions_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_GraphOptions_descriptor,
-        new java.lang.String[] { "EnableRecvScheduling", "OptimizerOptions", "BuildCostModel", "BuildCostModelAfter", "InferShapes", "PlacePrunedGraph", "EnableBfloat16Sendrecv", "TimelineStep", "RewriteOptions", });
-    internal_static_tensorflow_ThreadPoolOptionProto_descriptor =
-      getDescriptor().getMessageTypes().get(3);
-    internal_static_tensorflow_ThreadPoolOptionProto_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_ThreadPoolOptionProto_descriptor,
-        new java.lang.String[] { "NumThreads", "GlobalName", });
-    internal_static_tensorflow_RPCOptions_descriptor =
-      getDescriptor().getMessageTypes().get(4);
-    internal_static_tensorflow_RPCOptions_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_RPCOptions_descriptor,
-        new java.lang.String[] { "UseRpcForInprocessMaster", "CompressionAlgorithm", "CompressionLevel", "CacheRpcResponse", "DisableSessionConnectionSharing", });
-    internal_static_tensorflow_SessionMetadata_descriptor =
-      getDescriptor().getMessageTypes().get(5);
-    internal_static_tensorflow_SessionMetadata_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_SessionMetadata_descriptor,
-        new java.lang.String[] { "Name", "Version", });
-    internal_static_tensorflow_ConfigProto_descriptor =
-      getDescriptor().getMessageTypes().get(6);
-    internal_static_tensorflow_ConfigProto_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_ConfigProto_descriptor,
-        new java.lang.String[] { "DeviceCount", "IntraOpParallelismThreads", "InterOpParallelismThreads", "UsePerSessionThreads", "SessionInterOpThreadPool", "PlacementPeriod", "DeviceFilters", "GpuOptions", "AllowSoftPlacement", "LogDevicePlacement", "GraphOptions", "OperationTimeoutInMs", "RpcOptions", "ClusterDef", "IsolateSessionState", "ShareClusterDevicesInSession", "Experimental", });
-    internal_static_tensorflow_ConfigProto_DeviceCountEntry_descriptor =
-      internal_static_tensorflow_ConfigProto_descriptor.getNestedTypes().get(0);
-    internal_static_tensorflow_ConfigProto_DeviceCountEntry_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_ConfigProto_DeviceCountEntry_descriptor,
-        new java.lang.String[] { "Key", "Value", });
-    internal_static_tensorflow_ConfigProto_Experimental_descriptor =
-      internal_static_tensorflow_ConfigProto_descriptor.getNestedTypes().get(1);
-    internal_static_tensorflow_ConfigProto_Experimental_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_ConfigProto_Experimental_descriptor,
-        new java.lang.String[] { "CollectiveGroupLeader", "ExecutorType", "RecvBufMaxChunk", "UseNumaAffinity", "CollectiveDeterministicSequentialExecution", "CollectiveNccl", "ShareSessionStateInClusterspecPropagation", "DisableThreadSpinning", "ShareClusterDevicesInSession", "SessionMetadata", "OptimizeForStaticGraph", "EnableMlirBridge", "EnableMlirGraphOptimization", "DisableOutputPartitionGraphs", "XlaFusionAutotunerThresh", });
-    internal_static_tensorflow_RunOptions_descriptor =
-      getDescriptor().getMessageTypes().get(7);
-    internal_static_tensorflow_RunOptions_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_RunOptions_descriptor,
-        new java.lang.String[] { "TraceLevel", "TimeoutInMs", "InterOpThreadPool", "OutputPartitionGraphs", "DebugOptions", "ReportTensorAllocationsUponOom", "Experimental", });
-    internal_static_tensorflow_RunOptions_Experimental_descriptor =
-      internal_static_tensorflow_RunOptions_descriptor.getNestedTypes().get(0);
-    internal_static_tensorflow_RunOptions_Experimental_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_RunOptions_Experimental_descriptor,
-        new java.lang.String[] { "CollectiveGraphKey", "UseRunHandlerPool", "RunHandlerPoolOptions", });
-    internal_static_tensorflow_RunOptions_Experimental_RunHandlerPoolOptions_descriptor =
-      internal_static_tensorflow_RunOptions_Experimental_descriptor.getNestedTypes().get(0);
-    internal_static_tensorflow_RunOptions_Experimental_RunHandlerPoolOptions_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_RunOptions_Experimental_RunHandlerPoolOptions_descriptor,
-        new java.lang.String[] { "Priority", });
-    internal_static_tensorflow_RunMetadata_descriptor =
-      getDescriptor().getMessageTypes().get(8);
-    internal_static_tensorflow_RunMetadata_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_RunMetadata_descriptor,
-        new java.lang.String[] { "StepStats", "CostGraph", "PartitionGraphs", "FunctionGraphs", });
-    internal_static_tensorflow_RunMetadata_FunctionGraphs_descriptor =
-      internal_static_tensorflow_RunMetadata_descriptor.getNestedTypes().get(0);
-    internal_static_tensorflow_RunMetadata_FunctionGraphs_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_RunMetadata_FunctionGraphs_descriptor,
-        new java.lang.String[] { "PartitionGraphs", "PreOptimizationGraph", "PostOptimizationGraph", });
-    internal_static_tensorflow_TensorConnection_descriptor =
-      getDescriptor().getMessageTypes().get(9);
-    internal_static_tensorflow_TensorConnection_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_TensorConnection_descriptor,
-        new java.lang.String[] { "FromTensor", "ToTensor", });
-    internal_static_tensorflow_CallableOptions_descriptor =
-      getDescriptor().getMessageTypes().get(10);
-    internal_static_tensorflow_CallableOptions_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_CallableOptions_descriptor,
-        new java.lang.String[] { "Feed", "Fetch", "Target", "RunOptions", "TensorConnection", "FeedDevices", "FetchDevices", "FetchSkipSync", });
-    internal_static_tensorflow_CallableOptions_FeedDevicesEntry_descriptor =
-      internal_static_tensorflow_CallableOptions_descriptor.getNestedTypes().get(0);
-    internal_static_tensorflow_CallableOptions_FeedDevicesEntry_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_CallableOptions_FeedDevicesEntry_descriptor,
-        new java.lang.String[] { "Key", "Value", });
-    internal_static_tensorflow_CallableOptions_FetchDevicesEntry_descriptor =
-      internal_static_tensorflow_CallableOptions_descriptor.getNestedTypes().get(1);
-    internal_static_tensorflow_CallableOptions_FetchDevicesEntry_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_CallableOptions_FetchDevicesEntry_descriptor,
-        new java.lang.String[] { "Key", "Value", });
-    org.tensorflow.proto.framework.CostGraphProtos.getDescriptor();
-    org.tensorflow.proto.framework.GraphProtos.getDescriptor();
-    org.tensorflow.proto.framework.StepStatsProtos.getDescriptor();
-    org.tensorflow.proto.distruntime.ClusterProtos.getDescriptor();
-    org.tensorflow.proto.framework.DebugProtos.getDescriptor();
-    org.tensorflow.proto.framework.RewriterConfigProtos.getDescriptor();
-  }
-
-  // @@protoc_insertion_point(outer_class_scope)
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ControlFlowContextDef.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ControlFlowContextDef.java
deleted file mode 100644
index 48b3b2cd638..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ControlFlowContextDef.java
+++ /dev/null
@@ -1,903 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/control_flow.proto
-
-package org.tensorflow.proto.framework;
-
-/**
- * <pre>
- * Container for any kind of control flow context. Any other control flow
- * contexts that are added below should also be added here.
- * </pre>
- *
- * Protobuf type {@code tensorflow.ControlFlowContextDef}
- */
-public  final class ControlFlowContextDef extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.ControlFlowContextDef)
-    ControlFlowContextDefOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use ControlFlowContextDef.newBuilder() to construct.
-  private ControlFlowContextDef(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private ControlFlowContextDef() {
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new ControlFlowContextDef();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private ControlFlowContextDef(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 10: {
-            org.tensorflow.proto.framework.CondContextDef.Builder subBuilder = null;
-            if (ctxtCase_ == 1) {
-              subBuilder = ((org.tensorflow.proto.framework.CondContextDef) ctxt_).toBuilder();
-            }
-            ctxt_ =
-                input.readMessage(org.tensorflow.proto.framework.CondContextDef.parser(), extensionRegistry);
-            if (subBuilder != null) {
-              subBuilder.mergeFrom((org.tensorflow.proto.framework.CondContextDef) ctxt_);
-              ctxt_ = subBuilder.buildPartial();
-            }
-            ctxtCase_ = 1;
-            break;
-          }
-          case 18: {
-            org.tensorflow.proto.framework.WhileContextDef.Builder subBuilder = null;
-            if (ctxtCase_ == 2) {
-              subBuilder = ((org.tensorflow.proto.framework.WhileContextDef) ctxt_).toBuilder();
-            }
-            ctxt_ =
-                input.readMessage(org.tensorflow.proto.framework.WhileContextDef.parser(), extensionRegistry);
-            if (subBuilder != null) {
-              subBuilder.mergeFrom((org.tensorflow.proto.framework.WhileContextDef) ctxt_);
-              ctxt_ = subBuilder.buildPartial();
-            }
-            ctxtCase_ = 2;
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.framework.ControlFlowProtos.internal_static_tensorflow_ControlFlowContextDef_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.framework.ControlFlowProtos.internal_static_tensorflow_ControlFlowContextDef_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.framework.ControlFlowContextDef.class, org.tensorflow.proto.framework.ControlFlowContextDef.Builder.class);
-  }
-
-  private int ctxtCase_ = 0;
-  private java.lang.Object ctxt_;
-  public enum CtxtCase
-      implements com.google.protobuf.Internal.EnumLite {
-    COND_CTXT(1),
-    WHILE_CTXT(2),
-    CTXT_NOT_SET(0);
-    private final int value;
-    private CtxtCase(int value) {
-      this.value = value;
-    }
-    /**
-     * @deprecated Use {@link #forNumber(int)} instead.
-     */
-    @java.lang.Deprecated
-    public static CtxtCase valueOf(int value) {
-      return forNumber(value);
-    }
-
-    public static CtxtCase forNumber(int value) {
-      switch (value) {
-        case 1: return COND_CTXT;
-        case 2: return WHILE_CTXT;
-        case 0: return CTXT_NOT_SET;
-        default: return null;
-      }
-    }
-    public int getNumber() {
-      return this.value;
-    }
-  };
-
-  public CtxtCase
-  getCtxtCase() {
-    return CtxtCase.forNumber(
-        ctxtCase_);
-  }
-
-  public static final int COND_CTXT_FIELD_NUMBER = 1;
-  /**
-   * <code>.tensorflow.CondContextDef cond_ctxt = 1;</code>
-   */
-  public boolean hasCondCtxt() {
-    return ctxtCase_ == 1;
-  }
-  /**
-   * <code>.tensorflow.CondContextDef cond_ctxt = 1;</code>
-   */
-  public org.tensorflow.proto.framework.CondContextDef getCondCtxt() {
-    if (ctxtCase_ == 1) {
-       return (org.tensorflow.proto.framework.CondContextDef) ctxt_;
-    }
-    return org.tensorflow.proto.framework.CondContextDef.getDefaultInstance();
-  }
-  /**
-   * <code>.tensorflow.CondContextDef cond_ctxt = 1;</code>
-   */
-  public org.tensorflow.proto.framework.CondContextDefOrBuilder getCondCtxtOrBuilder() {
-    if (ctxtCase_ == 1) {
-       return (org.tensorflow.proto.framework.CondContextDef) ctxt_;
-    }
-    return org.tensorflow.proto.framework.CondContextDef.getDefaultInstance();
-  }
-
-  public static final int WHILE_CTXT_FIELD_NUMBER = 2;
-  /**
-   * <code>.tensorflow.WhileContextDef while_ctxt = 2;</code>
-   */
-  public boolean hasWhileCtxt() {
-    return ctxtCase_ == 2;
-  }
-  /**
-   * <code>.tensorflow.WhileContextDef while_ctxt = 2;</code>
-   */
-  public org.tensorflow.proto.framework.WhileContextDef getWhileCtxt() {
-    if (ctxtCase_ == 2) {
-       return (org.tensorflow.proto.framework.WhileContextDef) ctxt_;
-    }
-    return org.tensorflow.proto.framework.WhileContextDef.getDefaultInstance();
-  }
-  /**
-   * <code>.tensorflow.WhileContextDef while_ctxt = 2;</code>
-   */
-  public org.tensorflow.proto.framework.WhileContextDefOrBuilder getWhileCtxtOrBuilder() {
-    if (ctxtCase_ == 2) {
-       return (org.tensorflow.proto.framework.WhileContextDef) ctxt_;
-    }
-    return org.tensorflow.proto.framework.WhileContextDef.getDefaultInstance();
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    if (ctxtCase_ == 1) {
-      output.writeMessage(1, (org.tensorflow.proto.framework.CondContextDef) ctxt_);
-    }
-    if (ctxtCase_ == 2) {
-      output.writeMessage(2, (org.tensorflow.proto.framework.WhileContextDef) ctxt_);
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (ctxtCase_ == 1) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(1, (org.tensorflow.proto.framework.CondContextDef) ctxt_);
-    }
-    if (ctxtCase_ == 2) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(2, (org.tensorflow.proto.framework.WhileContextDef) ctxt_);
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.framework.ControlFlowContextDef)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.framework.ControlFlowContextDef other = (org.tensorflow.proto.framework.ControlFlowContextDef) obj;
-
-    if (!getCtxtCase().equals(other.getCtxtCase())) return false;
-    switch (ctxtCase_) {
-      case 1:
-        if (!getCondCtxt()
-            .equals(other.getCondCtxt())) return false;
-        break;
-      case 2:
-        if (!getWhileCtxt()
-            .equals(other.getWhileCtxt())) return false;
-        break;
-      case 0:
-      default:
-    }
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    switch (ctxtCase_) {
-      case 1:
-        hash = (37 * hash) + COND_CTXT_FIELD_NUMBER;
-        hash = (53 * hash) + getCondCtxt().hashCode();
-        break;
-      case 2:
-        hash = (37 * hash) + WHILE_CTXT_FIELD_NUMBER;
-        hash = (53 * hash) + getWhileCtxt().hashCode();
-        break;
-      case 0:
-      default:
-    }
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.framework.ControlFlowContextDef parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.ControlFlowContextDef parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.ControlFlowContextDef parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.ControlFlowContextDef parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.ControlFlowContextDef parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.ControlFlowContextDef parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.ControlFlowContextDef parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.ControlFlowContextDef parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.ControlFlowContextDef parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.ControlFlowContextDef parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.ControlFlowContextDef parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.ControlFlowContextDef parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.framework.ControlFlowContextDef prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * <pre>
-   * Container for any kind of control flow context. Any other control flow
-   * contexts that are added below should also be added here.
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.ControlFlowContextDef}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.ControlFlowContextDef)
-      org.tensorflow.proto.framework.ControlFlowContextDefOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.ControlFlowProtos.internal_static_tensorflow_ControlFlowContextDef_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.ControlFlowProtos.internal_static_tensorflow_ControlFlowContextDef_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.ControlFlowContextDef.class, org.tensorflow.proto.framework.ControlFlowContextDef.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.framework.ControlFlowContextDef.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      ctxtCase_ = 0;
-      ctxt_ = null;
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.framework.ControlFlowProtos.internal_static_tensorflow_ControlFlowContextDef_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.ControlFlowContextDef getDefaultInstanceForType() {
-      return org.tensorflow.proto.framework.ControlFlowContextDef.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.ControlFlowContextDef build() {
-      org.tensorflow.proto.framework.ControlFlowContextDef result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.ControlFlowContextDef buildPartial() {
-      org.tensorflow.proto.framework.ControlFlowContextDef result = new org.tensorflow.proto.framework.ControlFlowContextDef(this);
-      if (ctxtCase_ == 1) {
-        if (condCtxtBuilder_ == null) {
-          result.ctxt_ = ctxt_;
-        } else {
-          result.ctxt_ = condCtxtBuilder_.build();
-        }
-      }
-      if (ctxtCase_ == 2) {
-        if (whileCtxtBuilder_ == null) {
-          result.ctxt_ = ctxt_;
-        } else {
-          result.ctxt_ = whileCtxtBuilder_.build();
-        }
-      }
-      result.ctxtCase_ = ctxtCase_;
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.framework.ControlFlowContextDef) {
-        return mergeFrom((org.tensorflow.proto.framework.ControlFlowContextDef)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.framework.ControlFlowContextDef other) {
-      if (other == org.tensorflow.proto.framework.ControlFlowContextDef.getDefaultInstance()) return this;
-      switch (other.getCtxtCase()) {
-        case COND_CTXT: {
-          mergeCondCtxt(other.getCondCtxt());
-          break;
-        }
-        case WHILE_CTXT: {
-          mergeWhileCtxt(other.getWhileCtxt());
-          break;
-        }
-        case CTXT_NOT_SET: {
-          break;
-        }
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.framework.ControlFlowContextDef parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.framework.ControlFlowContextDef) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-    private int ctxtCase_ = 0;
-    private java.lang.Object ctxt_;
-    public CtxtCase
-        getCtxtCase() {
-      return CtxtCase.forNumber(
-          ctxtCase_);
-    }
-
-    public Builder clearCtxt() {
-      ctxtCase_ = 0;
-      ctxt_ = null;
-      onChanged();
-      return this;
-    }
-
-
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.CondContextDef, org.tensorflow.proto.framework.CondContextDef.Builder, org.tensorflow.proto.framework.CondContextDefOrBuilder> condCtxtBuilder_;
-    /**
-     * <code>.tensorflow.CondContextDef cond_ctxt = 1;</code>
-     */
-    public boolean hasCondCtxt() {
-      return ctxtCase_ == 1;
-    }
-    /**
-     * <code>.tensorflow.CondContextDef cond_ctxt = 1;</code>
-     */
-    public org.tensorflow.proto.framework.CondContextDef getCondCtxt() {
-      if (condCtxtBuilder_ == null) {
-        if (ctxtCase_ == 1) {
-          return (org.tensorflow.proto.framework.CondContextDef) ctxt_;
-        }
-        return org.tensorflow.proto.framework.CondContextDef.getDefaultInstance();
-      } else {
-        if (ctxtCase_ == 1) {
-          return condCtxtBuilder_.getMessage();
-        }
-        return org.tensorflow.proto.framework.CondContextDef.getDefaultInstance();
-      }
-    }
-    /**
-     * <code>.tensorflow.CondContextDef cond_ctxt = 1;</code>
-     */
-    public Builder setCondCtxt(org.tensorflow.proto.framework.CondContextDef value) {
-      if (condCtxtBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ctxt_ = value;
-        onChanged();
-      } else {
-        condCtxtBuilder_.setMessage(value);
-      }
-      ctxtCase_ = 1;
-      return this;
-    }
-    /**
-     * <code>.tensorflow.CondContextDef cond_ctxt = 1;</code>
-     */
-    public Builder setCondCtxt(
-        org.tensorflow.proto.framework.CondContextDef.Builder builderForValue) {
-      if (condCtxtBuilder_ == null) {
-        ctxt_ = builderForValue.build();
-        onChanged();
-      } else {
-        condCtxtBuilder_.setMessage(builderForValue.build());
-      }
-      ctxtCase_ = 1;
-      return this;
-    }
-    /**
-     * <code>.tensorflow.CondContextDef cond_ctxt = 1;</code>
-     */
-    public Builder mergeCondCtxt(org.tensorflow.proto.framework.CondContextDef value) {
-      if (condCtxtBuilder_ == null) {
-        if (ctxtCase_ == 1 &&
-            ctxt_ != org.tensorflow.proto.framework.CondContextDef.getDefaultInstance()) {
-          ctxt_ = org.tensorflow.proto.framework.CondContextDef.newBuilder((org.tensorflow.proto.framework.CondContextDef) ctxt_)
-              .mergeFrom(value).buildPartial();
-        } else {
-          ctxt_ = value;
-        }
-        onChanged();
-      } else {
-        if (ctxtCase_ == 1) {
-          condCtxtBuilder_.mergeFrom(value);
-        }
-        condCtxtBuilder_.setMessage(value);
-      }
-      ctxtCase_ = 1;
-      return this;
-    }
-    /**
-     * <code>.tensorflow.CondContextDef cond_ctxt = 1;</code>
-     */
-    public Builder clearCondCtxt() {
-      if (condCtxtBuilder_ == null) {
-        if (ctxtCase_ == 1) {
-          ctxtCase_ = 0;
-          ctxt_ = null;
-          onChanged();
-        }
-      } else {
-        if (ctxtCase_ == 1) {
-          ctxtCase_ = 0;
-          ctxt_ = null;
-        }
-        condCtxtBuilder_.clear();
-      }
-      return this;
-    }
-    /**
-     * <code>.tensorflow.CondContextDef cond_ctxt = 1;</code>
-     */
-    public org.tensorflow.proto.framework.CondContextDef.Builder getCondCtxtBuilder() {
-      return getCondCtxtFieldBuilder().getBuilder();
-    }
-    /**
-     * <code>.tensorflow.CondContextDef cond_ctxt = 1;</code>
-     */
-    public org.tensorflow.proto.framework.CondContextDefOrBuilder getCondCtxtOrBuilder() {
-      if ((ctxtCase_ == 1) && (condCtxtBuilder_ != null)) {
-        return condCtxtBuilder_.getMessageOrBuilder();
-      } else {
-        if (ctxtCase_ == 1) {
-          return (org.tensorflow.proto.framework.CondContextDef) ctxt_;
-        }
-        return org.tensorflow.proto.framework.CondContextDef.getDefaultInstance();
-      }
-    }
-    /**
-     * <code>.tensorflow.CondContextDef cond_ctxt = 1;</code>
-     */
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.CondContextDef, org.tensorflow.proto.framework.CondContextDef.Builder, org.tensorflow.proto.framework.CondContextDefOrBuilder> 
-        getCondCtxtFieldBuilder() {
-      if (condCtxtBuilder_ == null) {
-        if (!(ctxtCase_ == 1)) {
-          ctxt_ = org.tensorflow.proto.framework.CondContextDef.getDefaultInstance();
-        }
-        condCtxtBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-            org.tensorflow.proto.framework.CondContextDef, org.tensorflow.proto.framework.CondContextDef.Builder, org.tensorflow.proto.framework.CondContextDefOrBuilder>(
-                (org.tensorflow.proto.framework.CondContextDef) ctxt_,
-                getParentForChildren(),
-                isClean());
-        ctxt_ = null;
-      }
-      ctxtCase_ = 1;
-      onChanged();;
-      return condCtxtBuilder_;
-    }
-
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.WhileContextDef, org.tensorflow.proto.framework.WhileContextDef.Builder, org.tensorflow.proto.framework.WhileContextDefOrBuilder> whileCtxtBuilder_;
-    /**
-     * <code>.tensorflow.WhileContextDef while_ctxt = 2;</code>
-     */
-    public boolean hasWhileCtxt() {
-      return ctxtCase_ == 2;
-    }
-    /**
-     * <code>.tensorflow.WhileContextDef while_ctxt = 2;</code>
-     */
-    public org.tensorflow.proto.framework.WhileContextDef getWhileCtxt() {
-      if (whileCtxtBuilder_ == null) {
-        if (ctxtCase_ == 2) {
-          return (org.tensorflow.proto.framework.WhileContextDef) ctxt_;
-        }
-        return org.tensorflow.proto.framework.WhileContextDef.getDefaultInstance();
-      } else {
-        if (ctxtCase_ == 2) {
-          return whileCtxtBuilder_.getMessage();
-        }
-        return org.tensorflow.proto.framework.WhileContextDef.getDefaultInstance();
-      }
-    }
-    /**
-     * <code>.tensorflow.WhileContextDef while_ctxt = 2;</code>
-     */
-    public Builder setWhileCtxt(org.tensorflow.proto.framework.WhileContextDef value) {
-      if (whileCtxtBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ctxt_ = value;
-        onChanged();
-      } else {
-        whileCtxtBuilder_.setMessage(value);
-      }
-      ctxtCase_ = 2;
-      return this;
-    }
-    /**
-     * <code>.tensorflow.WhileContextDef while_ctxt = 2;</code>
-     */
-    public Builder setWhileCtxt(
-        org.tensorflow.proto.framework.WhileContextDef.Builder builderForValue) {
-      if (whileCtxtBuilder_ == null) {
-        ctxt_ = builderForValue.build();
-        onChanged();
-      } else {
-        whileCtxtBuilder_.setMessage(builderForValue.build());
-      }
-      ctxtCase_ = 2;
-      return this;
-    }
-    /**
-     * <code>.tensorflow.WhileContextDef while_ctxt = 2;</code>
-     */
-    public Builder mergeWhileCtxt(org.tensorflow.proto.framework.WhileContextDef value) {
-      if (whileCtxtBuilder_ == null) {
-        if (ctxtCase_ == 2 &&
-            ctxt_ != org.tensorflow.proto.framework.WhileContextDef.getDefaultInstance()) {
-          ctxt_ = org.tensorflow.proto.framework.WhileContextDef.newBuilder((org.tensorflow.proto.framework.WhileContextDef) ctxt_)
-              .mergeFrom(value).buildPartial();
-        } else {
-          ctxt_ = value;
-        }
-        onChanged();
-      } else {
-        if (ctxtCase_ == 2) {
-          whileCtxtBuilder_.mergeFrom(value);
-        }
-        whileCtxtBuilder_.setMessage(value);
-      }
-      ctxtCase_ = 2;
-      return this;
-    }
-    /**
-     * <code>.tensorflow.WhileContextDef while_ctxt = 2;</code>
-     */
-    public Builder clearWhileCtxt() {
-      if (whileCtxtBuilder_ == null) {
-        if (ctxtCase_ == 2) {
-          ctxtCase_ = 0;
-          ctxt_ = null;
-          onChanged();
-        }
-      } else {
-        if (ctxtCase_ == 2) {
-          ctxtCase_ = 0;
-          ctxt_ = null;
-        }
-        whileCtxtBuilder_.clear();
-      }
-      return this;
-    }
-    /**
-     * <code>.tensorflow.WhileContextDef while_ctxt = 2;</code>
-     */
-    public org.tensorflow.proto.framework.WhileContextDef.Builder getWhileCtxtBuilder() {
-      return getWhileCtxtFieldBuilder().getBuilder();
-    }
-    /**
-     * <code>.tensorflow.WhileContextDef while_ctxt = 2;</code>
-     */
-    public org.tensorflow.proto.framework.WhileContextDefOrBuilder getWhileCtxtOrBuilder() {
-      if ((ctxtCase_ == 2) && (whileCtxtBuilder_ != null)) {
-        return whileCtxtBuilder_.getMessageOrBuilder();
-      } else {
-        if (ctxtCase_ == 2) {
-          return (org.tensorflow.proto.framework.WhileContextDef) ctxt_;
-        }
-        return org.tensorflow.proto.framework.WhileContextDef.getDefaultInstance();
-      }
-    }
-    /**
-     * <code>.tensorflow.WhileContextDef while_ctxt = 2;</code>
-     */
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.WhileContextDef, org.tensorflow.proto.framework.WhileContextDef.Builder, org.tensorflow.proto.framework.WhileContextDefOrBuilder> 
-        getWhileCtxtFieldBuilder() {
-      if (whileCtxtBuilder_ == null) {
-        if (!(ctxtCase_ == 2)) {
-          ctxt_ = org.tensorflow.proto.framework.WhileContextDef.getDefaultInstance();
-        }
-        whileCtxtBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-            org.tensorflow.proto.framework.WhileContextDef, org.tensorflow.proto.framework.WhileContextDef.Builder, org.tensorflow.proto.framework.WhileContextDefOrBuilder>(
-                (org.tensorflow.proto.framework.WhileContextDef) ctxt_,
-                getParentForChildren(),
-                isClean());
-        ctxt_ = null;
-      }
-      ctxtCase_ = 2;
-      onChanged();;
-      return whileCtxtBuilder_;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.ControlFlowContextDef)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.ControlFlowContextDef)
-  private static final org.tensorflow.proto.framework.ControlFlowContextDef DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.ControlFlowContextDef();
-  }
-
-  public static org.tensorflow.proto.framework.ControlFlowContextDef getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<ControlFlowContextDef>
-      PARSER = new com.google.protobuf.AbstractParser<ControlFlowContextDef>() {
-    @java.lang.Override
-    public ControlFlowContextDef parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new ControlFlowContextDef(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<ControlFlowContextDef> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<ControlFlowContextDef> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.framework.ControlFlowContextDef getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ControlFlowContextDefOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ControlFlowContextDefOrBuilder.java
deleted file mode 100644
index b4482047fd9..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ControlFlowContextDefOrBuilder.java
+++ /dev/null
@@ -1,37 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/control_flow.proto
-
-package org.tensorflow.proto.framework;
-
-public interface ControlFlowContextDefOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.ControlFlowContextDef)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <code>.tensorflow.CondContextDef cond_ctxt = 1;</code>
-   */
-  boolean hasCondCtxt();
-  /**
-   * <code>.tensorflow.CondContextDef cond_ctxt = 1;</code>
-   */
-  org.tensorflow.proto.framework.CondContextDef getCondCtxt();
-  /**
-   * <code>.tensorflow.CondContextDef cond_ctxt = 1;</code>
-   */
-  org.tensorflow.proto.framework.CondContextDefOrBuilder getCondCtxtOrBuilder();
-
-  /**
-   * <code>.tensorflow.WhileContextDef while_ctxt = 2;</code>
-   */
-  boolean hasWhileCtxt();
-  /**
-   * <code>.tensorflow.WhileContextDef while_ctxt = 2;</code>
-   */
-  org.tensorflow.proto.framework.WhileContextDef getWhileCtxt();
-  /**
-   * <code>.tensorflow.WhileContextDef while_ctxt = 2;</code>
-   */
-  org.tensorflow.proto.framework.WhileContextDefOrBuilder getWhileCtxtOrBuilder();
-
-  public org.tensorflow.proto.framework.ControlFlowContextDef.CtxtCase getCtxtCase();
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ControlFlowProtos.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ControlFlowProtos.java
deleted file mode 100644
index 97a565aa465..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ControlFlowProtos.java
+++ /dev/null
@@ -1,116 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/control_flow.proto
-
-package org.tensorflow.proto.framework;
-
-public final class ControlFlowProtos {
-  private ControlFlowProtos() {}
-  public static void registerAllExtensions(
-      com.google.protobuf.ExtensionRegistryLite registry) {
-  }
-
-  public static void registerAllExtensions(
-      com.google.protobuf.ExtensionRegistry registry) {
-    registerAllExtensions(
-        (com.google.protobuf.ExtensionRegistryLite) registry);
-  }
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_ValuesDef_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_ValuesDef_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_ValuesDef_ExternalValuesEntry_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_ValuesDef_ExternalValuesEntry_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_ControlFlowContextDef_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_ControlFlowContextDef_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_CondContextDef_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_CondContextDef_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_WhileContextDef_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_WhileContextDef_fieldAccessorTable;
-
-  public static com.google.protobuf.Descriptors.FileDescriptor
-      getDescriptor() {
-    return descriptor;
-  }
-  private static  com.google.protobuf.Descriptors.FileDescriptor
-      descriptor;
-  static {
-    java.lang.String[] descriptorData = {
-      "\n+tensorflow/core/protobuf/control_flow." +
-      "proto\022\ntensorflow\"\226\001\n\tValuesDef\022\016\n\006value" +
-      "s\030\001 \003(\t\022B\n\017external_values\030\002 \003(\0132).tenso" +
-      "rflow.ValuesDef.ExternalValuesEntry\0325\n\023E" +
-      "xternalValuesEntry\022\013\n\003key\030\001 \001(\t\022\r\n\005value" +
-      "\030\002 \001(\t:\0028\001\"\203\001\n\025ControlFlowContextDef\022/\n\t" +
-      "cond_ctxt\030\001 \001(\0132\032.tensorflow.CondContext" +
-      "DefH\000\0221\n\nwhile_ctxt\030\002 \001(\0132\033.tensorflow.W" +
-      "hileContextDefH\000B\006\n\004ctxt\"\304\001\n\016CondContext" +
-      "Def\022\024\n\014context_name\030\001 \001(\t\022\021\n\tpred_name\030\002" +
-      " \001(\t\022\022\n\npivot_name\030\003 \001(\t\022\016\n\006branch\030\004 \001(\005" +
-      "\022)\n\nvalues_def\030\005 \001(\0132\025.tensorflow.Values" +
-      "Def\022:\n\017nested_contexts\030\006 \003(\0132!.tensorflo" +
-      "w.ControlFlowContextDef\"\365\002\n\017WhileContext" +
-      "Def\022\024\n\014context_name\030\001 \001(\t\022\033\n\023parallel_it" +
-      "erations\030\002 \001(\005\022\021\n\tback_prop\030\003 \001(\010\022\023\n\013swa" +
-      "p_memory\030\004 \001(\010\022\022\n\npivot_name\030\005 \001(\t\022\033\n\023pi" +
-      "vot_for_pred_name\030\006 \001(\t\022\033\n\023pivot_for_bod" +
-      "y_name\030\007 \001(\t\022\027\n\017loop_exit_names\030\010 \003(\t\022\030\n" +
-      "\020loop_enter_names\030\n \003(\t\022)\n\nvalues_def\030\t " +
-      "\001(\0132\025.tensorflow.ValuesDef\022\037\n\027maximum_it" +
-      "erations_name\030\013 \001(\t\022:\n\017nested_contexts\030\014" +
-      " \003(\0132!.tensorflow.ControlFlowContextDefB" +
-      "\202\001\n\036org.tensorflow.proto.frameworkB\021Cont" +
-      "rolFlowProtosP\001ZHgithub.com/tensorflow/t" +
-      "ensorflow/tensorflow/go/core/core_protos" +
-      "_go_proto\370\001\001b\006proto3"
-    };
-    descriptor = com.google.protobuf.Descriptors.FileDescriptor
-      .internalBuildGeneratedFileFrom(descriptorData,
-        new com.google.protobuf.Descriptors.FileDescriptor[] {
-        });
-    internal_static_tensorflow_ValuesDef_descriptor =
-      getDescriptor().getMessageTypes().get(0);
-    internal_static_tensorflow_ValuesDef_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_ValuesDef_descriptor,
-        new java.lang.String[] { "Values", "ExternalValues", });
-    internal_static_tensorflow_ValuesDef_ExternalValuesEntry_descriptor =
-      internal_static_tensorflow_ValuesDef_descriptor.getNestedTypes().get(0);
-    internal_static_tensorflow_ValuesDef_ExternalValuesEntry_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_ValuesDef_ExternalValuesEntry_descriptor,
-        new java.lang.String[] { "Key", "Value", });
-    internal_static_tensorflow_ControlFlowContextDef_descriptor =
-      getDescriptor().getMessageTypes().get(1);
-    internal_static_tensorflow_ControlFlowContextDef_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_ControlFlowContextDef_descriptor,
-        new java.lang.String[] { "CondCtxt", "WhileCtxt", "Ctxt", });
-    internal_static_tensorflow_CondContextDef_descriptor =
-      getDescriptor().getMessageTypes().get(2);
-    internal_static_tensorflow_CondContextDef_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_CondContextDef_descriptor,
-        new java.lang.String[] { "ContextName", "PredName", "PivotName", "Branch", "ValuesDef", "NestedContexts", });
-    internal_static_tensorflow_WhileContextDef_descriptor =
-      getDescriptor().getMessageTypes().get(3);
-    internal_static_tensorflow_WhileContextDef_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_WhileContextDef_descriptor,
-        new java.lang.String[] { "ContextName", "ParallelIterations", "BackProp", "SwapMemory", "PivotName", "PivotForPredName", "PivotForBodyName", "LoopExitNames", "LoopEnterNames", "ValuesDef", "MaximumIterationsName", "NestedContexts", });
-  }
-
-  // @@protoc_insertion_point(outer_class_scope)
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/CostGraphDef.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/CostGraphDef.java
deleted file mode 100644
index c06cd1761fa..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/CostGraphDef.java
+++ /dev/null
@@ -1,5817 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/framework/cost_graph.proto
-
-package org.tensorflow.proto.framework;
-
-/**
- * Protobuf type {@code tensorflow.CostGraphDef}
- */
-public  final class CostGraphDef extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.CostGraphDef)
-    CostGraphDefOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use CostGraphDef.newBuilder() to construct.
-  private CostGraphDef(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private CostGraphDef() {
-    node_ = java.util.Collections.emptyList();
-    cost_ = java.util.Collections.emptyList();
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new CostGraphDef();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private CostGraphDef(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    int mutable_bitField0_ = 0;
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 10: {
-            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
-              node_ = new java.util.ArrayList<org.tensorflow.proto.framework.CostGraphDef.Node>();
-              mutable_bitField0_ |= 0x00000001;
-            }
-            node_.add(
-                input.readMessage(org.tensorflow.proto.framework.CostGraphDef.Node.parser(), extensionRegistry));
-            break;
-          }
-          case 18: {
-            if (!((mutable_bitField0_ & 0x00000002) != 0)) {
-              cost_ = new java.util.ArrayList<org.tensorflow.proto.framework.CostGraphDef.AggregatedCost>();
-              mutable_bitField0_ |= 0x00000002;
-            }
-            cost_.add(
-                input.readMessage(org.tensorflow.proto.framework.CostGraphDef.AggregatedCost.parser(), extensionRegistry));
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      if (((mutable_bitField0_ & 0x00000001) != 0)) {
-        node_ = java.util.Collections.unmodifiableList(node_);
-      }
-      if (((mutable_bitField0_ & 0x00000002) != 0)) {
-        cost_ = java.util.Collections.unmodifiableList(cost_);
-      }
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.framework.CostGraphProtos.internal_static_tensorflow_CostGraphDef_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.framework.CostGraphProtos.internal_static_tensorflow_CostGraphDef_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.framework.CostGraphDef.class, org.tensorflow.proto.framework.CostGraphDef.Builder.class);
-  }
-
-  public interface NodeOrBuilder extends
-      // @@protoc_insertion_point(interface_extends:tensorflow.CostGraphDef.Node)
-      com.google.protobuf.MessageOrBuilder {
-
-    /**
-     * <pre>
-     * The name of the node. Names are globally unique.
-     * </pre>
-     *
-     * <code>string name = 1;</code>
-     */
-    java.lang.String getName();
-    /**
-     * <pre>
-     * The name of the node. Names are globally unique.
-     * </pre>
-     *
-     * <code>string name = 1;</code>
-     */
-    com.google.protobuf.ByteString
-        getNameBytes();
-
-    /**
-     * <pre>
-     * The device of the node. Can be empty if the node is mapped to the
-     * default partition or partitioning hasn't been run yet.
-     * </pre>
-     *
-     * <code>string device = 2;</code>
-     */
-    java.lang.String getDevice();
-    /**
-     * <pre>
-     * The device of the node. Can be empty if the node is mapped to the
-     * default partition or partitioning hasn't been run yet.
-     * </pre>
-     *
-     * <code>string device = 2;</code>
-     */
-    com.google.protobuf.ByteString
-        getDeviceBytes();
-
-    /**
-     * <pre>
-     * The id of the node. Node ids are only unique inside a partition.
-     * </pre>
-     *
-     * <code>int32 id = 3;</code>
-     */
-    int getId();
-
-    /**
-     * <code>repeated .tensorflow.CostGraphDef.Node.InputInfo input_info = 4;</code>
-     */
-    java.util.List<org.tensorflow.proto.framework.CostGraphDef.Node.InputInfo> 
-        getInputInfoList();
-    /**
-     * <code>repeated .tensorflow.CostGraphDef.Node.InputInfo input_info = 4;</code>
-     */
-    org.tensorflow.proto.framework.CostGraphDef.Node.InputInfo getInputInfo(int index);
-    /**
-     * <code>repeated .tensorflow.CostGraphDef.Node.InputInfo input_info = 4;</code>
-     */
-    int getInputInfoCount();
-    /**
-     * <code>repeated .tensorflow.CostGraphDef.Node.InputInfo input_info = 4;</code>
-     */
-    java.util.List<? extends org.tensorflow.proto.framework.CostGraphDef.Node.InputInfoOrBuilder> 
-        getInputInfoOrBuilderList();
-    /**
-     * <code>repeated .tensorflow.CostGraphDef.Node.InputInfo input_info = 4;</code>
-     */
-    org.tensorflow.proto.framework.CostGraphDef.Node.InputInfoOrBuilder getInputInfoOrBuilder(
-        int index);
-
-    /**
-     * <code>repeated .tensorflow.CostGraphDef.Node.OutputInfo output_info = 5;</code>
-     */
-    java.util.List<org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfo> 
-        getOutputInfoList();
-    /**
-     * <code>repeated .tensorflow.CostGraphDef.Node.OutputInfo output_info = 5;</code>
-     */
-    org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfo getOutputInfo(int index);
-    /**
-     * <code>repeated .tensorflow.CostGraphDef.Node.OutputInfo output_info = 5;</code>
-     */
-    int getOutputInfoCount();
-    /**
-     * <code>repeated .tensorflow.CostGraphDef.Node.OutputInfo output_info = 5;</code>
-     */
-    java.util.List<? extends org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfoOrBuilder> 
-        getOutputInfoOrBuilderList();
-    /**
-     * <code>repeated .tensorflow.CostGraphDef.Node.OutputInfo output_info = 5;</code>
-     */
-    org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfoOrBuilder getOutputInfoOrBuilder(
-        int index);
-
-    /**
-     * <pre>
-     * Temporary memory used by this node.
-     * </pre>
-     *
-     * <code>int64 temporary_memory_size = 6;</code>
-     */
-    long getTemporaryMemorySize();
-
-    /**
-     * <pre>
-     * Persistent memory used by this node.
-     * </pre>
-     *
-     * <code>int64 persistent_memory_size = 12;</code>
-     */
-    long getPersistentMemorySize();
-
-    /**
-     * <code>int64 host_temp_memory_size = 10 [deprecated = true];</code>
-     */
-    @java.lang.Deprecated long getHostTempMemorySize();
-
-    /**
-     * <code>int64 device_temp_memory_size = 11 [deprecated = true];</code>
-     */
-    @java.lang.Deprecated long getDeviceTempMemorySize();
-
-    /**
-     * <code>int64 device_persistent_memory_size = 16 [deprecated = true];</code>
-     */
-    @java.lang.Deprecated long getDevicePersistentMemorySize();
-
-    /**
-     * <pre>
-     * Estimate of the computational cost of this node, in microseconds.
-     * </pre>
-     *
-     * <code>int64 compute_cost = 9;</code>
-     */
-    long getComputeCost();
-
-    /**
-     * <pre>
-     * Analytical estimate of the computational cost of this node, in
-     * microseconds.
-     * </pre>
-     *
-     * <code>int64 compute_time = 14;</code>
-     */
-    long getComputeTime();
-
-    /**
-     * <pre>
-     * Analytical estimate of the memory access cost of this node, in
-     * microseconds.
-     * </pre>
-     *
-     * <code>int64 memory_time = 15;</code>
-     */
-    long getMemoryTime();
-
-    /**
-     * <pre>
-     * If true, the output is permanent: it can't be discarded, because this
-     * node is part of the "final output". Nodes may depend on final nodes.
-     * </pre>
-     *
-     * <code>bool is_final = 7;</code>
-     */
-    boolean getIsFinal();
-
-    /**
-     * <pre>
-     * Ids of the control inputs for this node.
-     * </pre>
-     *
-     * <code>repeated int32 control_input = 8;</code>
-     */
-    java.util.List<java.lang.Integer> getControlInputList();
-    /**
-     * <pre>
-     * Ids of the control inputs for this node.
-     * </pre>
-     *
-     * <code>repeated int32 control_input = 8;</code>
-     */
-    int getControlInputCount();
-    /**
-     * <pre>
-     * Ids of the control inputs for this node.
-     * </pre>
-     *
-     * <code>repeated int32 control_input = 8;</code>
-     */
-    int getControlInput(int index);
-
-    /**
-     * <pre>
-     * Are the costs inaccurate?
-     * </pre>
-     *
-     * <code>bool inaccurate = 17;</code>
-     */
-    boolean getInaccurate();
-  }
-  /**
-   * Protobuf type {@code tensorflow.CostGraphDef.Node}
-   */
-  public  static final class Node extends
-      com.google.protobuf.GeneratedMessageV3 implements
-      // @@protoc_insertion_point(message_implements:tensorflow.CostGraphDef.Node)
-      NodeOrBuilder {
-  private static final long serialVersionUID = 0L;
-    // Use Node.newBuilder() to construct.
-    private Node(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-      super(builder);
-    }
-    private Node() {
-      name_ = "";
-      device_ = "";
-      inputInfo_ = java.util.Collections.emptyList();
-      outputInfo_ = java.util.Collections.emptyList();
-      controlInput_ = emptyIntList();
-    }
-
-    @java.lang.Override
-    @SuppressWarnings({"unused"})
-    protected java.lang.Object newInstance(
-        UnusedPrivateParameter unused) {
-      return new Node();
-    }
-
-    @java.lang.Override
-    public final com.google.protobuf.UnknownFieldSet
-    getUnknownFields() {
-      return this.unknownFields;
-    }
-    private Node(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      this();
-      if (extensionRegistry == null) {
-        throw new java.lang.NullPointerException();
-      }
-      int mutable_bitField0_ = 0;
-      com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-          com.google.protobuf.UnknownFieldSet.newBuilder();
-      try {
-        boolean done = false;
-        while (!done) {
-          int tag = input.readTag();
-          switch (tag) {
-            case 0:
-              done = true;
-              break;
-            case 10: {
-              java.lang.String s = input.readStringRequireUtf8();
-
-              name_ = s;
-              break;
-            }
-            case 18: {
-              java.lang.String s = input.readStringRequireUtf8();
-
-              device_ = s;
-              break;
-            }
-            case 24: {
-
-              id_ = input.readInt32();
-              break;
-            }
-            case 34: {
-              if (!((mutable_bitField0_ & 0x00000001) != 0)) {
-                inputInfo_ = new java.util.ArrayList<org.tensorflow.proto.framework.CostGraphDef.Node.InputInfo>();
-                mutable_bitField0_ |= 0x00000001;
-              }
-              inputInfo_.add(
-                  input.readMessage(org.tensorflow.proto.framework.CostGraphDef.Node.InputInfo.parser(), extensionRegistry));
-              break;
-            }
-            case 42: {
-              if (!((mutable_bitField0_ & 0x00000002) != 0)) {
-                outputInfo_ = new java.util.ArrayList<org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfo>();
-                mutable_bitField0_ |= 0x00000002;
-              }
-              outputInfo_.add(
-                  input.readMessage(org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfo.parser(), extensionRegistry));
-              break;
-            }
-            case 48: {
-
-              temporaryMemorySize_ = input.readInt64();
-              break;
-            }
-            case 56: {
-
-              isFinal_ = input.readBool();
-              break;
-            }
-            case 64: {
-              if (!((mutable_bitField0_ & 0x00000004) != 0)) {
-                controlInput_ = newIntList();
-                mutable_bitField0_ |= 0x00000004;
-              }
-              controlInput_.addInt(input.readInt32());
-              break;
-            }
-            case 66: {
-              int length = input.readRawVarint32();
-              int limit = input.pushLimit(length);
-              if (!((mutable_bitField0_ & 0x00000004) != 0) && input.getBytesUntilLimit() > 0) {
-                controlInput_ = newIntList();
-                mutable_bitField0_ |= 0x00000004;
-              }
-              while (input.getBytesUntilLimit() > 0) {
-                controlInput_.addInt(input.readInt32());
-              }
-              input.popLimit(limit);
-              break;
-            }
-            case 72: {
-
-              computeCost_ = input.readInt64();
-              break;
-            }
-            case 80: {
-
-              hostTempMemorySize_ = input.readInt64();
-              break;
-            }
-            case 88: {
-
-              deviceTempMemorySize_ = input.readInt64();
-              break;
-            }
-            case 96: {
-
-              persistentMemorySize_ = input.readInt64();
-              break;
-            }
-            case 112: {
-
-              computeTime_ = input.readInt64();
-              break;
-            }
-            case 120: {
-
-              memoryTime_ = input.readInt64();
-              break;
-            }
-            case 128: {
-
-              devicePersistentMemorySize_ = input.readInt64();
-              break;
-            }
-            case 136: {
-
-              inaccurate_ = input.readBool();
-              break;
-            }
-            default: {
-              if (!parseUnknownField(
-                  input, unknownFields, extensionRegistry, tag)) {
-                done = true;
-              }
-              break;
-            }
-          }
-        }
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        throw e.setUnfinishedMessage(this);
-      } catch (java.io.IOException e) {
-        throw new com.google.protobuf.InvalidProtocolBufferException(
-            e).setUnfinishedMessage(this);
-      } finally {
-        if (((mutable_bitField0_ & 0x00000001) != 0)) {
-          inputInfo_ = java.util.Collections.unmodifiableList(inputInfo_);
-        }
-        if (((mutable_bitField0_ & 0x00000002) != 0)) {
-          outputInfo_ = java.util.Collections.unmodifiableList(outputInfo_);
-        }
-        if (((mutable_bitField0_ & 0x00000004) != 0)) {
-          controlInput_.makeImmutable(); // C
-        }
-        this.unknownFields = unknownFields.build();
-        makeExtensionsImmutable();
-      }
-    }
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.CostGraphProtos.internal_static_tensorflow_CostGraphDef_Node_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.CostGraphProtos.internal_static_tensorflow_CostGraphDef_Node_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.CostGraphDef.Node.class, org.tensorflow.proto.framework.CostGraphDef.Node.Builder.class);
-    }
-
-    public interface InputInfoOrBuilder extends
-        // @@protoc_insertion_point(interface_extends:tensorflow.CostGraphDef.Node.InputInfo)
-        com.google.protobuf.MessageOrBuilder {
-
-      /**
-       * <code>int32 preceding_node = 1;</code>
-       */
-      int getPrecedingNode();
-
-      /**
-       * <code>int32 preceding_port = 2;</code>
-       */
-      int getPrecedingPort();
-    }
-    /**
-     * <pre>
-     * Inputs of this node. They must be executed before this node can be
-     * executed. An input is a particular output of another node, specified
-     * by the node id and the output index.
-     * </pre>
-     *
-     * Protobuf type {@code tensorflow.CostGraphDef.Node.InputInfo}
-     */
-    public  static final class InputInfo extends
-        com.google.protobuf.GeneratedMessageV3 implements
-        // @@protoc_insertion_point(message_implements:tensorflow.CostGraphDef.Node.InputInfo)
-        InputInfoOrBuilder {
-    private static final long serialVersionUID = 0L;
-      // Use InputInfo.newBuilder() to construct.
-      private InputInfo(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-        super(builder);
-      }
-      private InputInfo() {
-      }
-
-      @java.lang.Override
-      @SuppressWarnings({"unused"})
-      protected java.lang.Object newInstance(
-          UnusedPrivateParameter unused) {
-        return new InputInfo();
-      }
-
-      @java.lang.Override
-      public final com.google.protobuf.UnknownFieldSet
-      getUnknownFields() {
-        return this.unknownFields;
-      }
-      private InputInfo(
-          com.google.protobuf.CodedInputStream input,
-          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-          throws com.google.protobuf.InvalidProtocolBufferException {
-        this();
-        if (extensionRegistry == null) {
-          throw new java.lang.NullPointerException();
-        }
-        com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-            com.google.protobuf.UnknownFieldSet.newBuilder();
-        try {
-          boolean done = false;
-          while (!done) {
-            int tag = input.readTag();
-            switch (tag) {
-              case 0:
-                done = true;
-                break;
-              case 8: {
-
-                precedingNode_ = input.readInt32();
-                break;
-              }
-              case 16: {
-
-                precedingPort_ = input.readInt32();
-                break;
-              }
-              default: {
-                if (!parseUnknownField(
-                    input, unknownFields, extensionRegistry, tag)) {
-                  done = true;
-                }
-                break;
-              }
-            }
-          }
-        } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-          throw e.setUnfinishedMessage(this);
-        } catch (java.io.IOException e) {
-          throw new com.google.protobuf.InvalidProtocolBufferException(
-              e).setUnfinishedMessage(this);
-        } finally {
-          this.unknownFields = unknownFields.build();
-          makeExtensionsImmutable();
-        }
-      }
-      public static final com.google.protobuf.Descriptors.Descriptor
-          getDescriptor() {
-        return org.tensorflow.proto.framework.CostGraphProtos.internal_static_tensorflow_CostGraphDef_Node_InputInfo_descriptor;
-      }
-
-      @java.lang.Override
-      protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-          internalGetFieldAccessorTable() {
-        return org.tensorflow.proto.framework.CostGraphProtos.internal_static_tensorflow_CostGraphDef_Node_InputInfo_fieldAccessorTable
-            .ensureFieldAccessorsInitialized(
-                org.tensorflow.proto.framework.CostGraphDef.Node.InputInfo.class, org.tensorflow.proto.framework.CostGraphDef.Node.InputInfo.Builder.class);
-      }
-
-      public static final int PRECEDING_NODE_FIELD_NUMBER = 1;
-      private int precedingNode_;
-      /**
-       * <code>int32 preceding_node = 1;</code>
-       */
-      public int getPrecedingNode() {
-        return precedingNode_;
-      }
-
-      public static final int PRECEDING_PORT_FIELD_NUMBER = 2;
-      private int precedingPort_;
-      /**
-       * <code>int32 preceding_port = 2;</code>
-       */
-      public int getPrecedingPort() {
-        return precedingPort_;
-      }
-
-      private byte memoizedIsInitialized = -1;
-      @java.lang.Override
-      public final boolean isInitialized() {
-        byte isInitialized = memoizedIsInitialized;
-        if (isInitialized == 1) return true;
-        if (isInitialized == 0) return false;
-
-        memoizedIsInitialized = 1;
-        return true;
-      }
-
-      @java.lang.Override
-      public void writeTo(com.google.protobuf.CodedOutputStream output)
-                          throws java.io.IOException {
-        if (precedingNode_ != 0) {
-          output.writeInt32(1, precedingNode_);
-        }
-        if (precedingPort_ != 0) {
-          output.writeInt32(2, precedingPort_);
-        }
-        unknownFields.writeTo(output);
-      }
-
-      @java.lang.Override
-      public int getSerializedSize() {
-        int size = memoizedSize;
-        if (size != -1) return size;
-
-        size = 0;
-        if (precedingNode_ != 0) {
-          size += com.google.protobuf.CodedOutputStream
-            .computeInt32Size(1, precedingNode_);
-        }
-        if (precedingPort_ != 0) {
-          size += com.google.protobuf.CodedOutputStream
-            .computeInt32Size(2, precedingPort_);
-        }
-        size += unknownFields.getSerializedSize();
-        memoizedSize = size;
-        return size;
-      }
-
-      @java.lang.Override
-      public boolean equals(final java.lang.Object obj) {
-        if (obj == this) {
-         return true;
-        }
-        if (!(obj instanceof org.tensorflow.proto.framework.CostGraphDef.Node.InputInfo)) {
-          return super.equals(obj);
-        }
-        org.tensorflow.proto.framework.CostGraphDef.Node.InputInfo other = (org.tensorflow.proto.framework.CostGraphDef.Node.InputInfo) obj;
-
-        if (getPrecedingNode()
-            != other.getPrecedingNode()) return false;
-        if (getPrecedingPort()
-            != other.getPrecedingPort()) return false;
-        if (!unknownFields.equals(other.unknownFields)) return false;
-        return true;
-      }
-
-      @java.lang.Override
-      public int hashCode() {
-        if (memoizedHashCode != 0) {
-          return memoizedHashCode;
-        }
-        int hash = 41;
-        hash = (19 * hash) + getDescriptor().hashCode();
-        hash = (37 * hash) + PRECEDING_NODE_FIELD_NUMBER;
-        hash = (53 * hash) + getPrecedingNode();
-        hash = (37 * hash) + PRECEDING_PORT_FIELD_NUMBER;
-        hash = (53 * hash) + getPrecedingPort();
-        hash = (29 * hash) + unknownFields.hashCode();
-        memoizedHashCode = hash;
-        return hash;
-      }
-
-      public static org.tensorflow.proto.framework.CostGraphDef.Node.InputInfo parseFrom(
-          java.nio.ByteBuffer data)
-          throws com.google.protobuf.InvalidProtocolBufferException {
-        return PARSER.parseFrom(data);
-      }
-      public static org.tensorflow.proto.framework.CostGraphDef.Node.InputInfo parseFrom(
-          java.nio.ByteBuffer data,
-          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-          throws com.google.protobuf.InvalidProtocolBufferException {
-        return PARSER.parseFrom(data, extensionRegistry);
-      }
-      public static org.tensorflow.proto.framework.CostGraphDef.Node.InputInfo parseFrom(
-          com.google.protobuf.ByteString data)
-          throws com.google.protobuf.InvalidProtocolBufferException {
-        return PARSER.parseFrom(data);
-      }
-      public static org.tensorflow.proto.framework.CostGraphDef.Node.InputInfo parseFrom(
-          com.google.protobuf.ByteString data,
-          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-          throws com.google.protobuf.InvalidProtocolBufferException {
-        return PARSER.parseFrom(data, extensionRegistry);
-      }
-      public static org.tensorflow.proto.framework.CostGraphDef.Node.InputInfo parseFrom(byte[] data)
-          throws com.google.protobuf.InvalidProtocolBufferException {
-        return PARSER.parseFrom(data);
-      }
-      public static org.tensorflow.proto.framework.CostGraphDef.Node.InputInfo parseFrom(
-          byte[] data,
-          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-          throws com.google.protobuf.InvalidProtocolBufferException {
-        return PARSER.parseFrom(data, extensionRegistry);
-      }
-      public static org.tensorflow.proto.framework.CostGraphDef.Node.InputInfo parseFrom(java.io.InputStream input)
-          throws java.io.IOException {
-        return com.google.protobuf.GeneratedMessageV3
-            .parseWithIOException(PARSER, input);
-      }
-      public static org.tensorflow.proto.framework.CostGraphDef.Node.InputInfo parseFrom(
-          java.io.InputStream input,
-          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-          throws java.io.IOException {
-        return com.google.protobuf.GeneratedMessageV3
-            .parseWithIOException(PARSER, input, extensionRegistry);
-      }
-      public static org.tensorflow.proto.framework.CostGraphDef.Node.InputInfo parseDelimitedFrom(java.io.InputStream input)
-          throws java.io.IOException {
-        return com.google.protobuf.GeneratedMessageV3
-            .parseDelimitedWithIOException(PARSER, input);
-      }
-      public static org.tensorflow.proto.framework.CostGraphDef.Node.InputInfo parseDelimitedFrom(
-          java.io.InputStream input,
-          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-          throws java.io.IOException {
-        return com.google.protobuf.GeneratedMessageV3
-            .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-      }
-      public static org.tensorflow.proto.framework.CostGraphDef.Node.InputInfo parseFrom(
-          com.google.protobuf.CodedInputStream input)
-          throws java.io.IOException {
-        return com.google.protobuf.GeneratedMessageV3
-            .parseWithIOException(PARSER, input);
-      }
-      public static org.tensorflow.proto.framework.CostGraphDef.Node.InputInfo parseFrom(
-          com.google.protobuf.CodedInputStream input,
-          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-          throws java.io.IOException {
-        return com.google.protobuf.GeneratedMessageV3
-            .parseWithIOException(PARSER, input, extensionRegistry);
-      }
-
-      @java.lang.Override
-      public Builder newBuilderForType() { return newBuilder(); }
-      public static Builder newBuilder() {
-        return DEFAULT_INSTANCE.toBuilder();
-      }
-      public static Builder newBuilder(org.tensorflow.proto.framework.CostGraphDef.Node.InputInfo prototype) {
-        return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-      }
-      @java.lang.Override
-      public Builder toBuilder() {
-        return this == DEFAULT_INSTANCE
-            ? new Builder() : new Builder().mergeFrom(this);
-      }
-
-      @java.lang.Override
-      protected Builder newBuilderForType(
-          com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-        Builder builder = new Builder(parent);
-        return builder;
-      }
-      /**
-       * <pre>
-       * Inputs of this node. They must be executed before this node can be
-       * executed. An input is a particular output of another node, specified
-       * by the node id and the output index.
-       * </pre>
-       *
-       * Protobuf type {@code tensorflow.CostGraphDef.Node.InputInfo}
-       */
-      public static final class Builder extends
-          com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-          // @@protoc_insertion_point(builder_implements:tensorflow.CostGraphDef.Node.InputInfo)
-          org.tensorflow.proto.framework.CostGraphDef.Node.InputInfoOrBuilder {
-        public static final com.google.protobuf.Descriptors.Descriptor
-            getDescriptor() {
-          return org.tensorflow.proto.framework.CostGraphProtos.internal_static_tensorflow_CostGraphDef_Node_InputInfo_descriptor;
-        }
-
-        @java.lang.Override
-        protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-            internalGetFieldAccessorTable() {
-          return org.tensorflow.proto.framework.CostGraphProtos.internal_static_tensorflow_CostGraphDef_Node_InputInfo_fieldAccessorTable
-              .ensureFieldAccessorsInitialized(
-                  org.tensorflow.proto.framework.CostGraphDef.Node.InputInfo.class, org.tensorflow.proto.framework.CostGraphDef.Node.InputInfo.Builder.class);
-        }
-
-        // Construct using org.tensorflow.proto.framework.CostGraphDef.Node.InputInfo.newBuilder()
-        private Builder() {
-          maybeForceBuilderInitialization();
-        }
-
-        private Builder(
-            com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-          super(parent);
-          maybeForceBuilderInitialization();
-        }
-        private void maybeForceBuilderInitialization() {
-          if (com.google.protobuf.GeneratedMessageV3
-                  .alwaysUseFieldBuilders) {
-          }
-        }
-        @java.lang.Override
-        public Builder clear() {
-          super.clear();
-          precedingNode_ = 0;
-
-          precedingPort_ = 0;
-
-          return this;
-        }
-
-        @java.lang.Override
-        public com.google.protobuf.Descriptors.Descriptor
-            getDescriptorForType() {
-          return org.tensorflow.proto.framework.CostGraphProtos.internal_static_tensorflow_CostGraphDef_Node_InputInfo_descriptor;
-        }
-
-        @java.lang.Override
-        public org.tensorflow.proto.framework.CostGraphDef.Node.InputInfo getDefaultInstanceForType() {
-          return org.tensorflow.proto.framework.CostGraphDef.Node.InputInfo.getDefaultInstance();
-        }
-
-        @java.lang.Override
-        public org.tensorflow.proto.framework.CostGraphDef.Node.InputInfo build() {
-          org.tensorflow.proto.framework.CostGraphDef.Node.InputInfo result = buildPartial();
-          if (!result.isInitialized()) {
-            throw newUninitializedMessageException(result);
-          }
-          return result;
-        }
-
-        @java.lang.Override
-        public org.tensorflow.proto.framework.CostGraphDef.Node.InputInfo buildPartial() {
-          org.tensorflow.proto.framework.CostGraphDef.Node.InputInfo result = new org.tensorflow.proto.framework.CostGraphDef.Node.InputInfo(this);
-          result.precedingNode_ = precedingNode_;
-          result.precedingPort_ = precedingPort_;
-          onBuilt();
-          return result;
-        }
-
-        @java.lang.Override
-        public Builder clone() {
-          return super.clone();
-        }
-        @java.lang.Override
-        public Builder setField(
-            com.google.protobuf.Descriptors.FieldDescriptor field,
-            java.lang.Object value) {
-          return super.setField(field, value);
-        }
-        @java.lang.Override
-        public Builder clearField(
-            com.google.protobuf.Descriptors.FieldDescriptor field) {
-          return super.clearField(field);
-        }
-        @java.lang.Override
-        public Builder clearOneof(
-            com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-          return super.clearOneof(oneof);
-        }
-        @java.lang.Override
-        public Builder setRepeatedField(
-            com.google.protobuf.Descriptors.FieldDescriptor field,
-            int index, java.lang.Object value) {
-          return super.setRepeatedField(field, index, value);
-        }
-        @java.lang.Override
-        public Builder addRepeatedField(
-            com.google.protobuf.Descriptors.FieldDescriptor field,
-            java.lang.Object value) {
-          return super.addRepeatedField(field, value);
-        }
-        @java.lang.Override
-        public Builder mergeFrom(com.google.protobuf.Message other) {
-          if (other instanceof org.tensorflow.proto.framework.CostGraphDef.Node.InputInfo) {
-            return mergeFrom((org.tensorflow.proto.framework.CostGraphDef.Node.InputInfo)other);
-          } else {
-            super.mergeFrom(other);
-            return this;
-          }
-        }
-
-        public Builder mergeFrom(org.tensorflow.proto.framework.CostGraphDef.Node.InputInfo other) {
-          if (other == org.tensorflow.proto.framework.CostGraphDef.Node.InputInfo.getDefaultInstance()) return this;
-          if (other.getPrecedingNode() != 0) {
-            setPrecedingNode(other.getPrecedingNode());
-          }
-          if (other.getPrecedingPort() != 0) {
-            setPrecedingPort(other.getPrecedingPort());
-          }
-          this.mergeUnknownFields(other.unknownFields);
-          onChanged();
-          return this;
-        }
-
-        @java.lang.Override
-        public final boolean isInitialized() {
-          return true;
-        }
-
-        @java.lang.Override
-        public Builder mergeFrom(
-            com.google.protobuf.CodedInputStream input,
-            com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-            throws java.io.IOException {
-          org.tensorflow.proto.framework.CostGraphDef.Node.InputInfo parsedMessage = null;
-          try {
-            parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-          } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-            parsedMessage = (org.tensorflow.proto.framework.CostGraphDef.Node.InputInfo) e.getUnfinishedMessage();
-            throw e.unwrapIOException();
-          } finally {
-            if (parsedMessage != null) {
-              mergeFrom(parsedMessage);
-            }
-          }
-          return this;
-        }
-
-        private int precedingNode_ ;
-        /**
-         * <code>int32 preceding_node = 1;</code>
-         */
-        public int getPrecedingNode() {
-          return precedingNode_;
-        }
-        /**
-         * <code>int32 preceding_node = 1;</code>
-         */
-        public Builder setPrecedingNode(int value) {
-          
-          precedingNode_ = value;
-          onChanged();
-          return this;
-        }
-        /**
-         * <code>int32 preceding_node = 1;</code>
-         */
-        public Builder clearPrecedingNode() {
-          
-          precedingNode_ = 0;
-          onChanged();
-          return this;
-        }
-
-        private int precedingPort_ ;
-        /**
-         * <code>int32 preceding_port = 2;</code>
-         */
-        public int getPrecedingPort() {
-          return precedingPort_;
-        }
-        /**
-         * <code>int32 preceding_port = 2;</code>
-         */
-        public Builder setPrecedingPort(int value) {
-          
-          precedingPort_ = value;
-          onChanged();
-          return this;
-        }
-        /**
-         * <code>int32 preceding_port = 2;</code>
-         */
-        public Builder clearPrecedingPort() {
-          
-          precedingPort_ = 0;
-          onChanged();
-          return this;
-        }
-        @java.lang.Override
-        public final Builder setUnknownFields(
-            final com.google.protobuf.UnknownFieldSet unknownFields) {
-          return super.setUnknownFields(unknownFields);
-        }
-
-        @java.lang.Override
-        public final Builder mergeUnknownFields(
-            final com.google.protobuf.UnknownFieldSet unknownFields) {
-          return super.mergeUnknownFields(unknownFields);
-        }
-
-
-        // @@protoc_insertion_point(builder_scope:tensorflow.CostGraphDef.Node.InputInfo)
-      }
-
-      // @@protoc_insertion_point(class_scope:tensorflow.CostGraphDef.Node.InputInfo)
-      private static final org.tensorflow.proto.framework.CostGraphDef.Node.InputInfo DEFAULT_INSTANCE;
-      static {
-        DEFAULT_INSTANCE = new org.tensorflow.proto.framework.CostGraphDef.Node.InputInfo();
-      }
-
-      public static org.tensorflow.proto.framework.CostGraphDef.Node.InputInfo getDefaultInstance() {
-        return DEFAULT_INSTANCE;
-      }
-
-      private static final com.google.protobuf.Parser<InputInfo>
-          PARSER = new com.google.protobuf.AbstractParser<InputInfo>() {
-        @java.lang.Override
-        public InputInfo parsePartialFrom(
-            com.google.protobuf.CodedInputStream input,
-            com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-            throws com.google.protobuf.InvalidProtocolBufferException {
-          return new InputInfo(input, extensionRegistry);
-        }
-      };
-
-      public static com.google.protobuf.Parser<InputInfo> parser() {
-        return PARSER;
-      }
-
-      @java.lang.Override
-      public com.google.protobuf.Parser<InputInfo> getParserForType() {
-        return PARSER;
-      }
-
-      @java.lang.Override
-      public org.tensorflow.proto.framework.CostGraphDef.Node.InputInfo getDefaultInstanceForType() {
-        return DEFAULT_INSTANCE;
-      }
-
-    }
-
-    public interface OutputInfoOrBuilder extends
-        // @@protoc_insertion_point(interface_extends:tensorflow.CostGraphDef.Node.OutputInfo)
-        com.google.protobuf.MessageOrBuilder {
-
-      /**
-       * <code>int64 size = 1;</code>
-       */
-      long getSize();
-
-      /**
-       * <pre>
-       * If &gt;= 0, the output is an alias of an input. Note that an alias input
-       * may itself be an alias. The algorithm will therefore need to follow
-       * those pointers.
-       * </pre>
-       *
-       * <code>int64 alias_input_port = 2;</code>
-       */
-      long getAliasInputPort();
-
-      /**
-       * <code>.tensorflow.TensorShapeProto shape = 3;</code>
-       */
-      boolean hasShape();
-      /**
-       * <code>.tensorflow.TensorShapeProto shape = 3;</code>
-       */
-      org.tensorflow.proto.framework.TensorShapeProto getShape();
-      /**
-       * <code>.tensorflow.TensorShapeProto shape = 3;</code>
-       */
-      org.tensorflow.proto.framework.TensorShapeProtoOrBuilder getShapeOrBuilder();
-
-      /**
-       * <code>.tensorflow.DataType dtype = 4;</code>
-       */
-      int getDtypeValue();
-      /**
-       * <code>.tensorflow.DataType dtype = 4;</code>
-       */
-      org.tensorflow.proto.framework.DataType getDtype();
-    }
-    /**
-     * <pre>
-     * Outputs of this node.
-     * </pre>
-     *
-     * Protobuf type {@code tensorflow.CostGraphDef.Node.OutputInfo}
-     */
-    public  static final class OutputInfo extends
-        com.google.protobuf.GeneratedMessageV3 implements
-        // @@protoc_insertion_point(message_implements:tensorflow.CostGraphDef.Node.OutputInfo)
-        OutputInfoOrBuilder {
-    private static final long serialVersionUID = 0L;
-      // Use OutputInfo.newBuilder() to construct.
-      private OutputInfo(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-        super(builder);
-      }
-      private OutputInfo() {
-        dtype_ = 0;
-      }
-
-      @java.lang.Override
-      @SuppressWarnings({"unused"})
-      protected java.lang.Object newInstance(
-          UnusedPrivateParameter unused) {
-        return new OutputInfo();
-      }
-
-      @java.lang.Override
-      public final com.google.protobuf.UnknownFieldSet
-      getUnknownFields() {
-        return this.unknownFields;
-      }
-      private OutputInfo(
-          com.google.protobuf.CodedInputStream input,
-          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-          throws com.google.protobuf.InvalidProtocolBufferException {
-        this();
-        if (extensionRegistry == null) {
-          throw new java.lang.NullPointerException();
-        }
-        com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-            com.google.protobuf.UnknownFieldSet.newBuilder();
-        try {
-          boolean done = false;
-          while (!done) {
-            int tag = input.readTag();
-            switch (tag) {
-              case 0:
-                done = true;
-                break;
-              case 8: {
-
-                size_ = input.readInt64();
-                break;
-              }
-              case 16: {
-
-                aliasInputPort_ = input.readInt64();
-                break;
-              }
-              case 26: {
-                org.tensorflow.proto.framework.TensorShapeProto.Builder subBuilder = null;
-                if (shape_ != null) {
-                  subBuilder = shape_.toBuilder();
-                }
-                shape_ = input.readMessage(org.tensorflow.proto.framework.TensorShapeProto.parser(), extensionRegistry);
-                if (subBuilder != null) {
-                  subBuilder.mergeFrom(shape_);
-                  shape_ = subBuilder.buildPartial();
-                }
-
-                break;
-              }
-              case 32: {
-                int rawValue = input.readEnum();
-
-                dtype_ = rawValue;
-                break;
-              }
-              default: {
-                if (!parseUnknownField(
-                    input, unknownFields, extensionRegistry, tag)) {
-                  done = true;
-                }
-                break;
-              }
-            }
-          }
-        } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-          throw e.setUnfinishedMessage(this);
-        } catch (java.io.IOException e) {
-          throw new com.google.protobuf.InvalidProtocolBufferException(
-              e).setUnfinishedMessage(this);
-        } finally {
-          this.unknownFields = unknownFields.build();
-          makeExtensionsImmutable();
-        }
-      }
-      public static final com.google.protobuf.Descriptors.Descriptor
-          getDescriptor() {
-        return org.tensorflow.proto.framework.CostGraphProtos.internal_static_tensorflow_CostGraphDef_Node_OutputInfo_descriptor;
-      }
-
-      @java.lang.Override
-      protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-          internalGetFieldAccessorTable() {
-        return org.tensorflow.proto.framework.CostGraphProtos.internal_static_tensorflow_CostGraphDef_Node_OutputInfo_fieldAccessorTable
-            .ensureFieldAccessorsInitialized(
-                org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfo.class, org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfo.Builder.class);
-      }
-
-      public static final int SIZE_FIELD_NUMBER = 1;
-      private long size_;
-      /**
-       * <code>int64 size = 1;</code>
-       */
-      public long getSize() {
-        return size_;
-      }
-
-      public static final int ALIAS_INPUT_PORT_FIELD_NUMBER = 2;
-      private long aliasInputPort_;
-      /**
-       * <pre>
-       * If &gt;= 0, the output is an alias of an input. Note that an alias input
-       * may itself be an alias. The algorithm will therefore need to follow
-       * those pointers.
-       * </pre>
-       *
-       * <code>int64 alias_input_port = 2;</code>
-       */
-      public long getAliasInputPort() {
-        return aliasInputPort_;
-      }
-
-      public static final int SHAPE_FIELD_NUMBER = 3;
-      private org.tensorflow.proto.framework.TensorShapeProto shape_;
-      /**
-       * <code>.tensorflow.TensorShapeProto shape = 3;</code>
-       */
-      public boolean hasShape() {
-        return shape_ != null;
-      }
-      /**
-       * <code>.tensorflow.TensorShapeProto shape = 3;</code>
-       */
-      public org.tensorflow.proto.framework.TensorShapeProto getShape() {
-        return shape_ == null ? org.tensorflow.proto.framework.TensorShapeProto.getDefaultInstance() : shape_;
-      }
-      /**
-       * <code>.tensorflow.TensorShapeProto shape = 3;</code>
-       */
-      public org.tensorflow.proto.framework.TensorShapeProtoOrBuilder getShapeOrBuilder() {
-        return getShape();
-      }
-
-      public static final int DTYPE_FIELD_NUMBER = 4;
-      private int dtype_;
-      /**
-       * <code>.tensorflow.DataType dtype = 4;</code>
-       */
-      public int getDtypeValue() {
-        return dtype_;
-      }
-      /**
-       * <code>.tensorflow.DataType dtype = 4;</code>
-       */
-      public org.tensorflow.proto.framework.DataType getDtype() {
-        @SuppressWarnings("deprecation")
-        org.tensorflow.proto.framework.DataType result = org.tensorflow.proto.framework.DataType.valueOf(dtype_);
-        return result == null ? org.tensorflow.proto.framework.DataType.UNRECOGNIZED : result;
-      }
-
-      private byte memoizedIsInitialized = -1;
-      @java.lang.Override
-      public final boolean isInitialized() {
-        byte isInitialized = memoizedIsInitialized;
-        if (isInitialized == 1) return true;
-        if (isInitialized == 0) return false;
-
-        memoizedIsInitialized = 1;
-        return true;
-      }
-
-      @java.lang.Override
-      public void writeTo(com.google.protobuf.CodedOutputStream output)
-                          throws java.io.IOException {
-        if (size_ != 0L) {
-          output.writeInt64(1, size_);
-        }
-        if (aliasInputPort_ != 0L) {
-          output.writeInt64(2, aliasInputPort_);
-        }
-        if (shape_ != null) {
-          output.writeMessage(3, getShape());
-        }
-        if (dtype_ != org.tensorflow.proto.framework.DataType.DT_INVALID.getNumber()) {
-          output.writeEnum(4, dtype_);
-        }
-        unknownFields.writeTo(output);
-      }
-
-      @java.lang.Override
-      public int getSerializedSize() {
-        int size = memoizedSize;
-        if (size != -1) return size;
-
-        size = 0;
-        if (size_ != 0L) {
-          size += com.google.protobuf.CodedOutputStream
-            .computeInt64Size(1, size_);
-        }
-        if (aliasInputPort_ != 0L) {
-          size += com.google.protobuf.CodedOutputStream
-            .computeInt64Size(2, aliasInputPort_);
-        }
-        if (shape_ != null) {
-          size += com.google.protobuf.CodedOutputStream
-            .computeMessageSize(3, getShape());
-        }
-        if (dtype_ != org.tensorflow.proto.framework.DataType.DT_INVALID.getNumber()) {
-          size += com.google.protobuf.CodedOutputStream
-            .computeEnumSize(4, dtype_);
-        }
-        size += unknownFields.getSerializedSize();
-        memoizedSize = size;
-        return size;
-      }
-
-      @java.lang.Override
-      public boolean equals(final java.lang.Object obj) {
-        if (obj == this) {
-         return true;
-        }
-        if (!(obj instanceof org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfo)) {
-          return super.equals(obj);
-        }
-        org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfo other = (org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfo) obj;
-
-        if (getSize()
-            != other.getSize()) return false;
-        if (getAliasInputPort()
-            != other.getAliasInputPort()) return false;
-        if (hasShape() != other.hasShape()) return false;
-        if (hasShape()) {
-          if (!getShape()
-              .equals(other.getShape())) return false;
-        }
-        if (dtype_ != other.dtype_) return false;
-        if (!unknownFields.equals(other.unknownFields)) return false;
-        return true;
-      }
-
-      @java.lang.Override
-      public int hashCode() {
-        if (memoizedHashCode != 0) {
-          return memoizedHashCode;
-        }
-        int hash = 41;
-        hash = (19 * hash) + getDescriptor().hashCode();
-        hash = (37 * hash) + SIZE_FIELD_NUMBER;
-        hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-            getSize());
-        hash = (37 * hash) + ALIAS_INPUT_PORT_FIELD_NUMBER;
-        hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-            getAliasInputPort());
-        if (hasShape()) {
-          hash = (37 * hash) + SHAPE_FIELD_NUMBER;
-          hash = (53 * hash) + getShape().hashCode();
-        }
-        hash = (37 * hash) + DTYPE_FIELD_NUMBER;
-        hash = (53 * hash) + dtype_;
-        hash = (29 * hash) + unknownFields.hashCode();
-        memoizedHashCode = hash;
-        return hash;
-      }
-
-      public static org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfo parseFrom(
-          java.nio.ByteBuffer data)
-          throws com.google.protobuf.InvalidProtocolBufferException {
-        return PARSER.parseFrom(data);
-      }
-      public static org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfo parseFrom(
-          java.nio.ByteBuffer data,
-          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-          throws com.google.protobuf.InvalidProtocolBufferException {
-        return PARSER.parseFrom(data, extensionRegistry);
-      }
-      public static org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfo parseFrom(
-          com.google.protobuf.ByteString data)
-          throws com.google.protobuf.InvalidProtocolBufferException {
-        return PARSER.parseFrom(data);
-      }
-      public static org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfo parseFrom(
-          com.google.protobuf.ByteString data,
-          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-          throws com.google.protobuf.InvalidProtocolBufferException {
-        return PARSER.parseFrom(data, extensionRegistry);
-      }
-      public static org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfo parseFrom(byte[] data)
-          throws com.google.protobuf.InvalidProtocolBufferException {
-        return PARSER.parseFrom(data);
-      }
-      public static org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfo parseFrom(
-          byte[] data,
-          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-          throws com.google.protobuf.InvalidProtocolBufferException {
-        return PARSER.parseFrom(data, extensionRegistry);
-      }
-      public static org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfo parseFrom(java.io.InputStream input)
-          throws java.io.IOException {
-        return com.google.protobuf.GeneratedMessageV3
-            .parseWithIOException(PARSER, input);
-      }
-      public static org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfo parseFrom(
-          java.io.InputStream input,
-          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-          throws java.io.IOException {
-        return com.google.protobuf.GeneratedMessageV3
-            .parseWithIOException(PARSER, input, extensionRegistry);
-      }
-      public static org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfo parseDelimitedFrom(java.io.InputStream input)
-          throws java.io.IOException {
-        return com.google.protobuf.GeneratedMessageV3
-            .parseDelimitedWithIOException(PARSER, input);
-      }
-      public static org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfo parseDelimitedFrom(
-          java.io.InputStream input,
-          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-          throws java.io.IOException {
-        return com.google.protobuf.GeneratedMessageV3
-            .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-      }
-      public static org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfo parseFrom(
-          com.google.protobuf.CodedInputStream input)
-          throws java.io.IOException {
-        return com.google.protobuf.GeneratedMessageV3
-            .parseWithIOException(PARSER, input);
-      }
-      public static org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfo parseFrom(
-          com.google.protobuf.CodedInputStream input,
-          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-          throws java.io.IOException {
-        return com.google.protobuf.GeneratedMessageV3
-            .parseWithIOException(PARSER, input, extensionRegistry);
-      }
-
-      @java.lang.Override
-      public Builder newBuilderForType() { return newBuilder(); }
-      public static Builder newBuilder() {
-        return DEFAULT_INSTANCE.toBuilder();
-      }
-      public static Builder newBuilder(org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfo prototype) {
-        return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-      }
-      @java.lang.Override
-      public Builder toBuilder() {
-        return this == DEFAULT_INSTANCE
-            ? new Builder() : new Builder().mergeFrom(this);
-      }
-
-      @java.lang.Override
-      protected Builder newBuilderForType(
-          com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-        Builder builder = new Builder(parent);
-        return builder;
-      }
-      /**
-       * <pre>
-       * Outputs of this node.
-       * </pre>
-       *
-       * Protobuf type {@code tensorflow.CostGraphDef.Node.OutputInfo}
-       */
-      public static final class Builder extends
-          com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-          // @@protoc_insertion_point(builder_implements:tensorflow.CostGraphDef.Node.OutputInfo)
-          org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfoOrBuilder {
-        public static final com.google.protobuf.Descriptors.Descriptor
-            getDescriptor() {
-          return org.tensorflow.proto.framework.CostGraphProtos.internal_static_tensorflow_CostGraphDef_Node_OutputInfo_descriptor;
-        }
-
-        @java.lang.Override
-        protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-            internalGetFieldAccessorTable() {
-          return org.tensorflow.proto.framework.CostGraphProtos.internal_static_tensorflow_CostGraphDef_Node_OutputInfo_fieldAccessorTable
-              .ensureFieldAccessorsInitialized(
-                  org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfo.class, org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfo.Builder.class);
-        }
-
-        // Construct using org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfo.newBuilder()
-        private Builder() {
-          maybeForceBuilderInitialization();
-        }
-
-        private Builder(
-            com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-          super(parent);
-          maybeForceBuilderInitialization();
-        }
-        private void maybeForceBuilderInitialization() {
-          if (com.google.protobuf.GeneratedMessageV3
-                  .alwaysUseFieldBuilders) {
-          }
-        }
-        @java.lang.Override
-        public Builder clear() {
-          super.clear();
-          size_ = 0L;
-
-          aliasInputPort_ = 0L;
-
-          if (shapeBuilder_ == null) {
-            shape_ = null;
-          } else {
-            shape_ = null;
-            shapeBuilder_ = null;
-          }
-          dtype_ = 0;
-
-          return this;
-        }
-
-        @java.lang.Override
-        public com.google.protobuf.Descriptors.Descriptor
-            getDescriptorForType() {
-          return org.tensorflow.proto.framework.CostGraphProtos.internal_static_tensorflow_CostGraphDef_Node_OutputInfo_descriptor;
-        }
-
-        @java.lang.Override
-        public org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfo getDefaultInstanceForType() {
-          return org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfo.getDefaultInstance();
-        }
-
-        @java.lang.Override
-        public org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfo build() {
-          org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfo result = buildPartial();
-          if (!result.isInitialized()) {
-            throw newUninitializedMessageException(result);
-          }
-          return result;
-        }
-
-        @java.lang.Override
-        public org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfo buildPartial() {
-          org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfo result = new org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfo(this);
-          result.size_ = size_;
-          result.aliasInputPort_ = aliasInputPort_;
-          if (shapeBuilder_ == null) {
-            result.shape_ = shape_;
-          } else {
-            result.shape_ = shapeBuilder_.build();
-          }
-          result.dtype_ = dtype_;
-          onBuilt();
-          return result;
-        }
-
-        @java.lang.Override
-        public Builder clone() {
-          return super.clone();
-        }
-        @java.lang.Override
-        public Builder setField(
-            com.google.protobuf.Descriptors.FieldDescriptor field,
-            java.lang.Object value) {
-          return super.setField(field, value);
-        }
-        @java.lang.Override
-        public Builder clearField(
-            com.google.protobuf.Descriptors.FieldDescriptor field) {
-          return super.clearField(field);
-        }
-        @java.lang.Override
-        public Builder clearOneof(
-            com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-          return super.clearOneof(oneof);
-        }
-        @java.lang.Override
-        public Builder setRepeatedField(
-            com.google.protobuf.Descriptors.FieldDescriptor field,
-            int index, java.lang.Object value) {
-          return super.setRepeatedField(field, index, value);
-        }
-        @java.lang.Override
-        public Builder addRepeatedField(
-            com.google.protobuf.Descriptors.FieldDescriptor field,
-            java.lang.Object value) {
-          return super.addRepeatedField(field, value);
-        }
-        @java.lang.Override
-        public Builder mergeFrom(com.google.protobuf.Message other) {
-          if (other instanceof org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfo) {
-            return mergeFrom((org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfo)other);
-          } else {
-            super.mergeFrom(other);
-            return this;
-          }
-        }
-
-        public Builder mergeFrom(org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfo other) {
-          if (other == org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfo.getDefaultInstance()) return this;
-          if (other.getSize() != 0L) {
-            setSize(other.getSize());
-          }
-          if (other.getAliasInputPort() != 0L) {
-            setAliasInputPort(other.getAliasInputPort());
-          }
-          if (other.hasShape()) {
-            mergeShape(other.getShape());
-          }
-          if (other.dtype_ != 0) {
-            setDtypeValue(other.getDtypeValue());
-          }
-          this.mergeUnknownFields(other.unknownFields);
-          onChanged();
-          return this;
-        }
-
-        @java.lang.Override
-        public final boolean isInitialized() {
-          return true;
-        }
-
-        @java.lang.Override
-        public Builder mergeFrom(
-            com.google.protobuf.CodedInputStream input,
-            com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-            throws java.io.IOException {
-          org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfo parsedMessage = null;
-          try {
-            parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-          } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-            parsedMessage = (org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfo) e.getUnfinishedMessage();
-            throw e.unwrapIOException();
-          } finally {
-            if (parsedMessage != null) {
-              mergeFrom(parsedMessage);
-            }
-          }
-          return this;
-        }
-
-        private long size_ ;
-        /**
-         * <code>int64 size = 1;</code>
-         */
-        public long getSize() {
-          return size_;
-        }
-        /**
-         * <code>int64 size = 1;</code>
-         */
-        public Builder setSize(long value) {
-          
-          size_ = value;
-          onChanged();
-          return this;
-        }
-        /**
-         * <code>int64 size = 1;</code>
-         */
-        public Builder clearSize() {
-          
-          size_ = 0L;
-          onChanged();
-          return this;
-        }
-
-        private long aliasInputPort_ ;
-        /**
-         * <pre>
-         * If &gt;= 0, the output is an alias of an input. Note that an alias input
-         * may itself be an alias. The algorithm will therefore need to follow
-         * those pointers.
-         * </pre>
-         *
-         * <code>int64 alias_input_port = 2;</code>
-         */
-        public long getAliasInputPort() {
-          return aliasInputPort_;
-        }
-        /**
-         * <pre>
-         * If &gt;= 0, the output is an alias of an input. Note that an alias input
-         * may itself be an alias. The algorithm will therefore need to follow
-         * those pointers.
-         * </pre>
-         *
-         * <code>int64 alias_input_port = 2;</code>
-         */
-        public Builder setAliasInputPort(long value) {
-          
-          aliasInputPort_ = value;
-          onChanged();
-          return this;
-        }
-        /**
-         * <pre>
-         * If &gt;= 0, the output is an alias of an input. Note that an alias input
-         * may itself be an alias. The algorithm will therefore need to follow
-         * those pointers.
-         * </pre>
-         *
-         * <code>int64 alias_input_port = 2;</code>
-         */
-        public Builder clearAliasInputPort() {
-          
-          aliasInputPort_ = 0L;
-          onChanged();
-          return this;
-        }
-
-        private org.tensorflow.proto.framework.TensorShapeProto shape_;
-        private com.google.protobuf.SingleFieldBuilderV3<
-            org.tensorflow.proto.framework.TensorShapeProto, org.tensorflow.proto.framework.TensorShapeProto.Builder, org.tensorflow.proto.framework.TensorShapeProtoOrBuilder> shapeBuilder_;
-        /**
-         * <code>.tensorflow.TensorShapeProto shape = 3;</code>
-         */
-        public boolean hasShape() {
-          return shapeBuilder_ != null || shape_ != null;
-        }
-        /**
-         * <code>.tensorflow.TensorShapeProto shape = 3;</code>
-         */
-        public org.tensorflow.proto.framework.TensorShapeProto getShape() {
-          if (shapeBuilder_ == null) {
-            return shape_ == null ? org.tensorflow.proto.framework.TensorShapeProto.getDefaultInstance() : shape_;
-          } else {
-            return shapeBuilder_.getMessage();
-          }
-        }
-        /**
-         * <code>.tensorflow.TensorShapeProto shape = 3;</code>
-         */
-        public Builder setShape(org.tensorflow.proto.framework.TensorShapeProto value) {
-          if (shapeBuilder_ == null) {
-            if (value == null) {
-              throw new NullPointerException();
-            }
-            shape_ = value;
-            onChanged();
-          } else {
-            shapeBuilder_.setMessage(value);
-          }
-
-          return this;
-        }
-        /**
-         * <code>.tensorflow.TensorShapeProto shape = 3;</code>
-         */
-        public Builder setShape(
-            org.tensorflow.proto.framework.TensorShapeProto.Builder builderForValue) {
-          if (shapeBuilder_ == null) {
-            shape_ = builderForValue.build();
-            onChanged();
-          } else {
-            shapeBuilder_.setMessage(builderForValue.build());
-          }
-
-          return this;
-        }
-        /**
-         * <code>.tensorflow.TensorShapeProto shape = 3;</code>
-         */
-        public Builder mergeShape(org.tensorflow.proto.framework.TensorShapeProto value) {
-          if (shapeBuilder_ == null) {
-            if (shape_ != null) {
-              shape_ =
-                org.tensorflow.proto.framework.TensorShapeProto.newBuilder(shape_).mergeFrom(value).buildPartial();
-            } else {
-              shape_ = value;
-            }
-            onChanged();
-          } else {
-            shapeBuilder_.mergeFrom(value);
-          }
-
-          return this;
-        }
-        /**
-         * <code>.tensorflow.TensorShapeProto shape = 3;</code>
-         */
-        public Builder clearShape() {
-          if (shapeBuilder_ == null) {
-            shape_ = null;
-            onChanged();
-          } else {
-            shape_ = null;
-            shapeBuilder_ = null;
-          }
-
-          return this;
-        }
-        /**
-         * <code>.tensorflow.TensorShapeProto shape = 3;</code>
-         */
-        public org.tensorflow.proto.framework.TensorShapeProto.Builder getShapeBuilder() {
-          
-          onChanged();
-          return getShapeFieldBuilder().getBuilder();
-        }
-        /**
-         * <code>.tensorflow.TensorShapeProto shape = 3;</code>
-         */
-        public org.tensorflow.proto.framework.TensorShapeProtoOrBuilder getShapeOrBuilder() {
-          if (shapeBuilder_ != null) {
-            return shapeBuilder_.getMessageOrBuilder();
-          } else {
-            return shape_ == null ?
-                org.tensorflow.proto.framework.TensorShapeProto.getDefaultInstance() : shape_;
-          }
-        }
-        /**
-         * <code>.tensorflow.TensorShapeProto shape = 3;</code>
-         */
-        private com.google.protobuf.SingleFieldBuilderV3<
-            org.tensorflow.proto.framework.TensorShapeProto, org.tensorflow.proto.framework.TensorShapeProto.Builder, org.tensorflow.proto.framework.TensorShapeProtoOrBuilder> 
-            getShapeFieldBuilder() {
-          if (shapeBuilder_ == null) {
-            shapeBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-                org.tensorflow.proto.framework.TensorShapeProto, org.tensorflow.proto.framework.TensorShapeProto.Builder, org.tensorflow.proto.framework.TensorShapeProtoOrBuilder>(
-                    getShape(),
-                    getParentForChildren(),
-                    isClean());
-            shape_ = null;
-          }
-          return shapeBuilder_;
-        }
-
-        private int dtype_ = 0;
-        /**
-         * <code>.tensorflow.DataType dtype = 4;</code>
-         */
-        public int getDtypeValue() {
-          return dtype_;
-        }
-        /**
-         * <code>.tensorflow.DataType dtype = 4;</code>
-         */
-        public Builder setDtypeValue(int value) {
-          dtype_ = value;
-          onChanged();
-          return this;
-        }
-        /**
-         * <code>.tensorflow.DataType dtype = 4;</code>
-         */
-        public org.tensorflow.proto.framework.DataType getDtype() {
-          @SuppressWarnings("deprecation")
-          org.tensorflow.proto.framework.DataType result = org.tensorflow.proto.framework.DataType.valueOf(dtype_);
-          return result == null ? org.tensorflow.proto.framework.DataType.UNRECOGNIZED : result;
-        }
-        /**
-         * <code>.tensorflow.DataType dtype = 4;</code>
-         */
-        public Builder setDtype(org.tensorflow.proto.framework.DataType value) {
-          if (value == null) {
-            throw new NullPointerException();
-          }
-          
-          dtype_ = value.getNumber();
-          onChanged();
-          return this;
-        }
-        /**
-         * <code>.tensorflow.DataType dtype = 4;</code>
-         */
-        public Builder clearDtype() {
-          
-          dtype_ = 0;
-          onChanged();
-          return this;
-        }
-        @java.lang.Override
-        public final Builder setUnknownFields(
-            final com.google.protobuf.UnknownFieldSet unknownFields) {
-          return super.setUnknownFields(unknownFields);
-        }
-
-        @java.lang.Override
-        public final Builder mergeUnknownFields(
-            final com.google.protobuf.UnknownFieldSet unknownFields) {
-          return super.mergeUnknownFields(unknownFields);
-        }
-
-
-        // @@protoc_insertion_point(builder_scope:tensorflow.CostGraphDef.Node.OutputInfo)
-      }
-
-      // @@protoc_insertion_point(class_scope:tensorflow.CostGraphDef.Node.OutputInfo)
-      private static final org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfo DEFAULT_INSTANCE;
-      static {
-        DEFAULT_INSTANCE = new org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfo();
-      }
-
-      public static org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfo getDefaultInstance() {
-        return DEFAULT_INSTANCE;
-      }
-
-      private static final com.google.protobuf.Parser<OutputInfo>
-          PARSER = new com.google.protobuf.AbstractParser<OutputInfo>() {
-        @java.lang.Override
-        public OutputInfo parsePartialFrom(
-            com.google.protobuf.CodedInputStream input,
-            com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-            throws com.google.protobuf.InvalidProtocolBufferException {
-          return new OutputInfo(input, extensionRegistry);
-        }
-      };
-
-      public static com.google.protobuf.Parser<OutputInfo> parser() {
-        return PARSER;
-      }
-
-      @java.lang.Override
-      public com.google.protobuf.Parser<OutputInfo> getParserForType() {
-        return PARSER;
-      }
-
-      @java.lang.Override
-      public org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfo getDefaultInstanceForType() {
-        return DEFAULT_INSTANCE;
-      }
-
-    }
-
-    public static final int NAME_FIELD_NUMBER = 1;
-    private volatile java.lang.Object name_;
-    /**
-     * <pre>
-     * The name of the node. Names are globally unique.
-     * </pre>
-     *
-     * <code>string name = 1;</code>
-     */
-    public java.lang.String getName() {
-      java.lang.Object ref = name_;
-      if (ref instanceof java.lang.String) {
-        return (java.lang.String) ref;
-      } else {
-        com.google.protobuf.ByteString bs = 
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        name_ = s;
-        return s;
-      }
-    }
-    /**
-     * <pre>
-     * The name of the node. Names are globally unique.
-     * </pre>
-     *
-     * <code>string name = 1;</code>
-     */
-    public com.google.protobuf.ByteString
-        getNameBytes() {
-      java.lang.Object ref = name_;
-      if (ref instanceof java.lang.String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        name_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-
-    public static final int DEVICE_FIELD_NUMBER = 2;
-    private volatile java.lang.Object device_;
-    /**
-     * <pre>
-     * The device of the node. Can be empty if the node is mapped to the
-     * default partition or partitioning hasn't been run yet.
-     * </pre>
-     *
-     * <code>string device = 2;</code>
-     */
-    public java.lang.String getDevice() {
-      java.lang.Object ref = device_;
-      if (ref instanceof java.lang.String) {
-        return (java.lang.String) ref;
-      } else {
-        com.google.protobuf.ByteString bs = 
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        device_ = s;
-        return s;
-      }
-    }
-    /**
-     * <pre>
-     * The device of the node. Can be empty if the node is mapped to the
-     * default partition or partitioning hasn't been run yet.
-     * </pre>
-     *
-     * <code>string device = 2;</code>
-     */
-    public com.google.protobuf.ByteString
-        getDeviceBytes() {
-      java.lang.Object ref = device_;
-      if (ref instanceof java.lang.String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        device_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-
-    public static final int ID_FIELD_NUMBER = 3;
-    private int id_;
-    /**
-     * <pre>
-     * The id of the node. Node ids are only unique inside a partition.
-     * </pre>
-     *
-     * <code>int32 id = 3;</code>
-     */
-    public int getId() {
-      return id_;
-    }
-
-    public static final int INPUT_INFO_FIELD_NUMBER = 4;
-    private java.util.List<org.tensorflow.proto.framework.CostGraphDef.Node.InputInfo> inputInfo_;
-    /**
-     * <code>repeated .tensorflow.CostGraphDef.Node.InputInfo input_info = 4;</code>
-     */
-    public java.util.List<org.tensorflow.proto.framework.CostGraphDef.Node.InputInfo> getInputInfoList() {
-      return inputInfo_;
-    }
-    /**
-     * <code>repeated .tensorflow.CostGraphDef.Node.InputInfo input_info = 4;</code>
-     */
-    public java.util.List<? extends org.tensorflow.proto.framework.CostGraphDef.Node.InputInfoOrBuilder> 
-        getInputInfoOrBuilderList() {
-      return inputInfo_;
-    }
-    /**
-     * <code>repeated .tensorflow.CostGraphDef.Node.InputInfo input_info = 4;</code>
-     */
-    public int getInputInfoCount() {
-      return inputInfo_.size();
-    }
-    /**
-     * <code>repeated .tensorflow.CostGraphDef.Node.InputInfo input_info = 4;</code>
-     */
-    public org.tensorflow.proto.framework.CostGraphDef.Node.InputInfo getInputInfo(int index) {
-      return inputInfo_.get(index);
-    }
-    /**
-     * <code>repeated .tensorflow.CostGraphDef.Node.InputInfo input_info = 4;</code>
-     */
-    public org.tensorflow.proto.framework.CostGraphDef.Node.InputInfoOrBuilder getInputInfoOrBuilder(
-        int index) {
-      return inputInfo_.get(index);
-    }
-
-    public static final int OUTPUT_INFO_FIELD_NUMBER = 5;
-    private java.util.List<org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfo> outputInfo_;
-    /**
-     * <code>repeated .tensorflow.CostGraphDef.Node.OutputInfo output_info = 5;</code>
-     */
-    public java.util.List<org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfo> getOutputInfoList() {
-      return outputInfo_;
-    }
-    /**
-     * <code>repeated .tensorflow.CostGraphDef.Node.OutputInfo output_info = 5;</code>
-     */
-    public java.util.List<? extends org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfoOrBuilder> 
-        getOutputInfoOrBuilderList() {
-      return outputInfo_;
-    }
-    /**
-     * <code>repeated .tensorflow.CostGraphDef.Node.OutputInfo output_info = 5;</code>
-     */
-    public int getOutputInfoCount() {
-      return outputInfo_.size();
-    }
-    /**
-     * <code>repeated .tensorflow.CostGraphDef.Node.OutputInfo output_info = 5;</code>
-     */
-    public org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfo getOutputInfo(int index) {
-      return outputInfo_.get(index);
-    }
-    /**
-     * <code>repeated .tensorflow.CostGraphDef.Node.OutputInfo output_info = 5;</code>
-     */
-    public org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfoOrBuilder getOutputInfoOrBuilder(
-        int index) {
-      return outputInfo_.get(index);
-    }
-
-    public static final int TEMPORARY_MEMORY_SIZE_FIELD_NUMBER = 6;
-    private long temporaryMemorySize_;
-    /**
-     * <pre>
-     * Temporary memory used by this node.
-     * </pre>
-     *
-     * <code>int64 temporary_memory_size = 6;</code>
-     */
-    public long getTemporaryMemorySize() {
-      return temporaryMemorySize_;
-    }
-
-    public static final int PERSISTENT_MEMORY_SIZE_FIELD_NUMBER = 12;
-    private long persistentMemorySize_;
-    /**
-     * <pre>
-     * Persistent memory used by this node.
-     * </pre>
-     *
-     * <code>int64 persistent_memory_size = 12;</code>
-     */
-    public long getPersistentMemorySize() {
-      return persistentMemorySize_;
-    }
-
-    public static final int HOST_TEMP_MEMORY_SIZE_FIELD_NUMBER = 10;
-    private long hostTempMemorySize_;
-    /**
-     * <code>int64 host_temp_memory_size = 10 [deprecated = true];</code>
-     */
-    @java.lang.Deprecated public long getHostTempMemorySize() {
-      return hostTempMemorySize_;
-    }
-
-    public static final int DEVICE_TEMP_MEMORY_SIZE_FIELD_NUMBER = 11;
-    private long deviceTempMemorySize_;
-    /**
-     * <code>int64 device_temp_memory_size = 11 [deprecated = true];</code>
-     */
-    @java.lang.Deprecated public long getDeviceTempMemorySize() {
-      return deviceTempMemorySize_;
-    }
-
-    public static final int DEVICE_PERSISTENT_MEMORY_SIZE_FIELD_NUMBER = 16;
-    private long devicePersistentMemorySize_;
-    /**
-     * <code>int64 device_persistent_memory_size = 16 [deprecated = true];</code>
-     */
-    @java.lang.Deprecated public long getDevicePersistentMemorySize() {
-      return devicePersistentMemorySize_;
-    }
-
-    public static final int COMPUTE_COST_FIELD_NUMBER = 9;
-    private long computeCost_;
-    /**
-     * <pre>
-     * Estimate of the computational cost of this node, in microseconds.
-     * </pre>
-     *
-     * <code>int64 compute_cost = 9;</code>
-     */
-    public long getComputeCost() {
-      return computeCost_;
-    }
-
-    public static final int COMPUTE_TIME_FIELD_NUMBER = 14;
-    private long computeTime_;
-    /**
-     * <pre>
-     * Analytical estimate of the computational cost of this node, in
-     * microseconds.
-     * </pre>
-     *
-     * <code>int64 compute_time = 14;</code>
-     */
-    public long getComputeTime() {
-      return computeTime_;
-    }
-
-    public static final int MEMORY_TIME_FIELD_NUMBER = 15;
-    private long memoryTime_;
-    /**
-     * <pre>
-     * Analytical estimate of the memory access cost of this node, in
-     * microseconds.
-     * </pre>
-     *
-     * <code>int64 memory_time = 15;</code>
-     */
-    public long getMemoryTime() {
-      return memoryTime_;
-    }
-
-    public static final int IS_FINAL_FIELD_NUMBER = 7;
-    private boolean isFinal_;
-    /**
-     * <pre>
-     * If true, the output is permanent: it can't be discarded, because this
-     * node is part of the "final output". Nodes may depend on final nodes.
-     * </pre>
-     *
-     * <code>bool is_final = 7;</code>
-     */
-    public boolean getIsFinal() {
-      return isFinal_;
-    }
-
-    public static final int CONTROL_INPUT_FIELD_NUMBER = 8;
-    private com.google.protobuf.Internal.IntList controlInput_;
-    /**
-     * <pre>
-     * Ids of the control inputs for this node.
-     * </pre>
-     *
-     * <code>repeated int32 control_input = 8;</code>
-     */
-    public java.util.List<java.lang.Integer>
-        getControlInputList() {
-      return controlInput_;
-    }
-    /**
-     * <pre>
-     * Ids of the control inputs for this node.
-     * </pre>
-     *
-     * <code>repeated int32 control_input = 8;</code>
-     */
-    public int getControlInputCount() {
-      return controlInput_.size();
-    }
-    /**
-     * <pre>
-     * Ids of the control inputs for this node.
-     * </pre>
-     *
-     * <code>repeated int32 control_input = 8;</code>
-     */
-    public int getControlInput(int index) {
-      return controlInput_.getInt(index);
-    }
-    private int controlInputMemoizedSerializedSize = -1;
-
-    public static final int INACCURATE_FIELD_NUMBER = 17;
-    private boolean inaccurate_;
-    /**
-     * <pre>
-     * Are the costs inaccurate?
-     * </pre>
-     *
-     * <code>bool inaccurate = 17;</code>
-     */
-    public boolean getInaccurate() {
-      return inaccurate_;
-    }
-
-    private byte memoizedIsInitialized = -1;
-    @java.lang.Override
-    public final boolean isInitialized() {
-      byte isInitialized = memoizedIsInitialized;
-      if (isInitialized == 1) return true;
-      if (isInitialized == 0) return false;
-
-      memoizedIsInitialized = 1;
-      return true;
-    }
-
-    @java.lang.Override
-    public void writeTo(com.google.protobuf.CodedOutputStream output)
-                        throws java.io.IOException {
-      getSerializedSize();
-      if (!getNameBytes().isEmpty()) {
-        com.google.protobuf.GeneratedMessageV3.writeString(output, 1, name_);
-      }
-      if (!getDeviceBytes().isEmpty()) {
-        com.google.protobuf.GeneratedMessageV3.writeString(output, 2, device_);
-      }
-      if (id_ != 0) {
-        output.writeInt32(3, id_);
-      }
-      for (int i = 0; i < inputInfo_.size(); i++) {
-        output.writeMessage(4, inputInfo_.get(i));
-      }
-      for (int i = 0; i < outputInfo_.size(); i++) {
-        output.writeMessage(5, outputInfo_.get(i));
-      }
-      if (temporaryMemorySize_ != 0L) {
-        output.writeInt64(6, temporaryMemorySize_);
-      }
-      if (isFinal_ != false) {
-        output.writeBool(7, isFinal_);
-      }
-      if (getControlInputList().size() > 0) {
-        output.writeUInt32NoTag(66);
-        output.writeUInt32NoTag(controlInputMemoizedSerializedSize);
-      }
-      for (int i = 0; i < controlInput_.size(); i++) {
-        output.writeInt32NoTag(controlInput_.getInt(i));
-      }
-      if (computeCost_ != 0L) {
-        output.writeInt64(9, computeCost_);
-      }
-      if (hostTempMemorySize_ != 0L) {
-        output.writeInt64(10, hostTempMemorySize_);
-      }
-      if (deviceTempMemorySize_ != 0L) {
-        output.writeInt64(11, deviceTempMemorySize_);
-      }
-      if (persistentMemorySize_ != 0L) {
-        output.writeInt64(12, persistentMemorySize_);
-      }
-      if (computeTime_ != 0L) {
-        output.writeInt64(14, computeTime_);
-      }
-      if (memoryTime_ != 0L) {
-        output.writeInt64(15, memoryTime_);
-      }
-      if (devicePersistentMemorySize_ != 0L) {
-        output.writeInt64(16, devicePersistentMemorySize_);
-      }
-      if (inaccurate_ != false) {
-        output.writeBool(17, inaccurate_);
-      }
-      unknownFields.writeTo(output);
-    }
-
-    @java.lang.Override
-    public int getSerializedSize() {
-      int size = memoizedSize;
-      if (size != -1) return size;
-
-      size = 0;
-      if (!getNameBytes().isEmpty()) {
-        size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, name_);
-      }
-      if (!getDeviceBytes().isEmpty()) {
-        size += com.google.protobuf.GeneratedMessageV3.computeStringSize(2, device_);
-      }
-      if (id_ != 0) {
-        size += com.google.protobuf.CodedOutputStream
-          .computeInt32Size(3, id_);
-      }
-      for (int i = 0; i < inputInfo_.size(); i++) {
-        size += com.google.protobuf.CodedOutputStream
-          .computeMessageSize(4, inputInfo_.get(i));
-      }
-      for (int i = 0; i < outputInfo_.size(); i++) {
-        size += com.google.protobuf.CodedOutputStream
-          .computeMessageSize(5, outputInfo_.get(i));
-      }
-      if (temporaryMemorySize_ != 0L) {
-        size += com.google.protobuf.CodedOutputStream
-          .computeInt64Size(6, temporaryMemorySize_);
-      }
-      if (isFinal_ != false) {
-        size += com.google.protobuf.CodedOutputStream
-          .computeBoolSize(7, isFinal_);
-      }
-      {
-        int dataSize = 0;
-        for (int i = 0; i < controlInput_.size(); i++) {
-          dataSize += com.google.protobuf.CodedOutputStream
-            .computeInt32SizeNoTag(controlInput_.getInt(i));
-        }
-        size += dataSize;
-        if (!getControlInputList().isEmpty()) {
-          size += 1;
-          size += com.google.protobuf.CodedOutputStream
-              .computeInt32SizeNoTag(dataSize);
-        }
-        controlInputMemoizedSerializedSize = dataSize;
-      }
-      if (computeCost_ != 0L) {
-        size += com.google.protobuf.CodedOutputStream
-          .computeInt64Size(9, computeCost_);
-      }
-      if (hostTempMemorySize_ != 0L) {
-        size += com.google.protobuf.CodedOutputStream
-          .computeInt64Size(10, hostTempMemorySize_);
-      }
-      if (deviceTempMemorySize_ != 0L) {
-        size += com.google.protobuf.CodedOutputStream
-          .computeInt64Size(11, deviceTempMemorySize_);
-      }
-      if (persistentMemorySize_ != 0L) {
-        size += com.google.protobuf.CodedOutputStream
-          .computeInt64Size(12, persistentMemorySize_);
-      }
-      if (computeTime_ != 0L) {
-        size += com.google.protobuf.CodedOutputStream
-          .computeInt64Size(14, computeTime_);
-      }
-      if (memoryTime_ != 0L) {
-        size += com.google.protobuf.CodedOutputStream
-          .computeInt64Size(15, memoryTime_);
-      }
-      if (devicePersistentMemorySize_ != 0L) {
-        size += com.google.protobuf.CodedOutputStream
-          .computeInt64Size(16, devicePersistentMemorySize_);
-      }
-      if (inaccurate_ != false) {
-        size += com.google.protobuf.CodedOutputStream
-          .computeBoolSize(17, inaccurate_);
-      }
-      size += unknownFields.getSerializedSize();
-      memoizedSize = size;
-      return size;
-    }
-
-    @java.lang.Override
-    public boolean equals(final java.lang.Object obj) {
-      if (obj == this) {
-       return true;
-      }
-      if (!(obj instanceof org.tensorflow.proto.framework.CostGraphDef.Node)) {
-        return super.equals(obj);
-      }
-      org.tensorflow.proto.framework.CostGraphDef.Node other = (org.tensorflow.proto.framework.CostGraphDef.Node) obj;
-
-      if (!getName()
-          .equals(other.getName())) return false;
-      if (!getDevice()
-          .equals(other.getDevice())) return false;
-      if (getId()
-          != other.getId()) return false;
-      if (!getInputInfoList()
-          .equals(other.getInputInfoList())) return false;
-      if (!getOutputInfoList()
-          .equals(other.getOutputInfoList())) return false;
-      if (getTemporaryMemorySize()
-          != other.getTemporaryMemorySize()) return false;
-      if (getPersistentMemorySize()
-          != other.getPersistentMemorySize()) return false;
-      if (getHostTempMemorySize()
-          != other.getHostTempMemorySize()) return false;
-      if (getDeviceTempMemorySize()
-          != other.getDeviceTempMemorySize()) return false;
-      if (getDevicePersistentMemorySize()
-          != other.getDevicePersistentMemorySize()) return false;
-      if (getComputeCost()
-          != other.getComputeCost()) return false;
-      if (getComputeTime()
-          != other.getComputeTime()) return false;
-      if (getMemoryTime()
-          != other.getMemoryTime()) return false;
-      if (getIsFinal()
-          != other.getIsFinal()) return false;
-      if (!getControlInputList()
-          .equals(other.getControlInputList())) return false;
-      if (getInaccurate()
-          != other.getInaccurate()) return false;
-      if (!unknownFields.equals(other.unknownFields)) return false;
-      return true;
-    }
-
-    @java.lang.Override
-    public int hashCode() {
-      if (memoizedHashCode != 0) {
-        return memoizedHashCode;
-      }
-      int hash = 41;
-      hash = (19 * hash) + getDescriptor().hashCode();
-      hash = (37 * hash) + NAME_FIELD_NUMBER;
-      hash = (53 * hash) + getName().hashCode();
-      hash = (37 * hash) + DEVICE_FIELD_NUMBER;
-      hash = (53 * hash) + getDevice().hashCode();
-      hash = (37 * hash) + ID_FIELD_NUMBER;
-      hash = (53 * hash) + getId();
-      if (getInputInfoCount() > 0) {
-        hash = (37 * hash) + INPUT_INFO_FIELD_NUMBER;
-        hash = (53 * hash) + getInputInfoList().hashCode();
-      }
-      if (getOutputInfoCount() > 0) {
-        hash = (37 * hash) + OUTPUT_INFO_FIELD_NUMBER;
-        hash = (53 * hash) + getOutputInfoList().hashCode();
-      }
-      hash = (37 * hash) + TEMPORARY_MEMORY_SIZE_FIELD_NUMBER;
-      hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-          getTemporaryMemorySize());
-      hash = (37 * hash) + PERSISTENT_MEMORY_SIZE_FIELD_NUMBER;
-      hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-          getPersistentMemorySize());
-      hash = (37 * hash) + HOST_TEMP_MEMORY_SIZE_FIELD_NUMBER;
-      hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-          getHostTempMemorySize());
-      hash = (37 * hash) + DEVICE_TEMP_MEMORY_SIZE_FIELD_NUMBER;
-      hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-          getDeviceTempMemorySize());
-      hash = (37 * hash) + DEVICE_PERSISTENT_MEMORY_SIZE_FIELD_NUMBER;
-      hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-          getDevicePersistentMemorySize());
-      hash = (37 * hash) + COMPUTE_COST_FIELD_NUMBER;
-      hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-          getComputeCost());
-      hash = (37 * hash) + COMPUTE_TIME_FIELD_NUMBER;
-      hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-          getComputeTime());
-      hash = (37 * hash) + MEMORY_TIME_FIELD_NUMBER;
-      hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-          getMemoryTime());
-      hash = (37 * hash) + IS_FINAL_FIELD_NUMBER;
-      hash = (53 * hash) + com.google.protobuf.Internal.hashBoolean(
-          getIsFinal());
-      if (getControlInputCount() > 0) {
-        hash = (37 * hash) + CONTROL_INPUT_FIELD_NUMBER;
-        hash = (53 * hash) + getControlInputList().hashCode();
-      }
-      hash = (37 * hash) + INACCURATE_FIELD_NUMBER;
-      hash = (53 * hash) + com.google.protobuf.Internal.hashBoolean(
-          getInaccurate());
-      hash = (29 * hash) + unknownFields.hashCode();
-      memoizedHashCode = hash;
-      return hash;
-    }
-
-    public static org.tensorflow.proto.framework.CostGraphDef.Node parseFrom(
-        java.nio.ByteBuffer data)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data);
-    }
-    public static org.tensorflow.proto.framework.CostGraphDef.Node parseFrom(
-        java.nio.ByteBuffer data,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.CostGraphDef.Node parseFrom(
-        com.google.protobuf.ByteString data)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data);
-    }
-    public static org.tensorflow.proto.framework.CostGraphDef.Node parseFrom(
-        com.google.protobuf.ByteString data,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.CostGraphDef.Node parseFrom(byte[] data)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data);
-    }
-    public static org.tensorflow.proto.framework.CostGraphDef.Node parseFrom(
-        byte[] data,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.CostGraphDef.Node parseFrom(java.io.InputStream input)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input);
-    }
-    public static org.tensorflow.proto.framework.CostGraphDef.Node parseFrom(
-        java.io.InputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.CostGraphDef.Node parseDelimitedFrom(java.io.InputStream input)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseDelimitedWithIOException(PARSER, input);
-    }
-    public static org.tensorflow.proto.framework.CostGraphDef.Node parseDelimitedFrom(
-        java.io.InputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.CostGraphDef.Node parseFrom(
-        com.google.protobuf.CodedInputStream input)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input);
-    }
-    public static org.tensorflow.proto.framework.CostGraphDef.Node parseFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input, extensionRegistry);
-    }
-
-    @java.lang.Override
-    public Builder newBuilderForType() { return newBuilder(); }
-    public static Builder newBuilder() {
-      return DEFAULT_INSTANCE.toBuilder();
-    }
-    public static Builder newBuilder(org.tensorflow.proto.framework.CostGraphDef.Node prototype) {
-      return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-    }
-    @java.lang.Override
-    public Builder toBuilder() {
-      return this == DEFAULT_INSTANCE
-          ? new Builder() : new Builder().mergeFrom(this);
-    }
-
-    @java.lang.Override
-    protected Builder newBuilderForType(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      Builder builder = new Builder(parent);
-      return builder;
-    }
-    /**
-     * Protobuf type {@code tensorflow.CostGraphDef.Node}
-     */
-    public static final class Builder extends
-        com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-        // @@protoc_insertion_point(builder_implements:tensorflow.CostGraphDef.Node)
-        org.tensorflow.proto.framework.CostGraphDef.NodeOrBuilder {
-      public static final com.google.protobuf.Descriptors.Descriptor
-          getDescriptor() {
-        return org.tensorflow.proto.framework.CostGraphProtos.internal_static_tensorflow_CostGraphDef_Node_descriptor;
-      }
-
-      @java.lang.Override
-      protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-          internalGetFieldAccessorTable() {
-        return org.tensorflow.proto.framework.CostGraphProtos.internal_static_tensorflow_CostGraphDef_Node_fieldAccessorTable
-            .ensureFieldAccessorsInitialized(
-                org.tensorflow.proto.framework.CostGraphDef.Node.class, org.tensorflow.proto.framework.CostGraphDef.Node.Builder.class);
-      }
-
-      // Construct using org.tensorflow.proto.framework.CostGraphDef.Node.newBuilder()
-      private Builder() {
-        maybeForceBuilderInitialization();
-      }
-
-      private Builder(
-          com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-        super(parent);
-        maybeForceBuilderInitialization();
-      }
-      private void maybeForceBuilderInitialization() {
-        if (com.google.protobuf.GeneratedMessageV3
-                .alwaysUseFieldBuilders) {
-          getInputInfoFieldBuilder();
-          getOutputInfoFieldBuilder();
-        }
-      }
-      @java.lang.Override
-      public Builder clear() {
-        super.clear();
-        name_ = "";
-
-        device_ = "";
-
-        id_ = 0;
-
-        if (inputInfoBuilder_ == null) {
-          inputInfo_ = java.util.Collections.emptyList();
-          bitField0_ = (bitField0_ & ~0x00000001);
-        } else {
-          inputInfoBuilder_.clear();
-        }
-        if (outputInfoBuilder_ == null) {
-          outputInfo_ = java.util.Collections.emptyList();
-          bitField0_ = (bitField0_ & ~0x00000002);
-        } else {
-          outputInfoBuilder_.clear();
-        }
-        temporaryMemorySize_ = 0L;
-
-        persistentMemorySize_ = 0L;
-
-        hostTempMemorySize_ = 0L;
-
-        deviceTempMemorySize_ = 0L;
-
-        devicePersistentMemorySize_ = 0L;
-
-        computeCost_ = 0L;
-
-        computeTime_ = 0L;
-
-        memoryTime_ = 0L;
-
-        isFinal_ = false;
-
-        controlInput_ = emptyIntList();
-        bitField0_ = (bitField0_ & ~0x00000004);
-        inaccurate_ = false;
-
-        return this;
-      }
-
-      @java.lang.Override
-      public com.google.protobuf.Descriptors.Descriptor
-          getDescriptorForType() {
-        return org.tensorflow.proto.framework.CostGraphProtos.internal_static_tensorflow_CostGraphDef_Node_descriptor;
-      }
-
-      @java.lang.Override
-      public org.tensorflow.proto.framework.CostGraphDef.Node getDefaultInstanceForType() {
-        return org.tensorflow.proto.framework.CostGraphDef.Node.getDefaultInstance();
-      }
-
-      @java.lang.Override
-      public org.tensorflow.proto.framework.CostGraphDef.Node build() {
-        org.tensorflow.proto.framework.CostGraphDef.Node result = buildPartial();
-        if (!result.isInitialized()) {
-          throw newUninitializedMessageException(result);
-        }
-        return result;
-      }
-
-      @java.lang.Override
-      public org.tensorflow.proto.framework.CostGraphDef.Node buildPartial() {
-        org.tensorflow.proto.framework.CostGraphDef.Node result = new org.tensorflow.proto.framework.CostGraphDef.Node(this);
-        int from_bitField0_ = bitField0_;
-        result.name_ = name_;
-        result.device_ = device_;
-        result.id_ = id_;
-        if (inputInfoBuilder_ == null) {
-          if (((bitField0_ & 0x00000001) != 0)) {
-            inputInfo_ = java.util.Collections.unmodifiableList(inputInfo_);
-            bitField0_ = (bitField0_ & ~0x00000001);
-          }
-          result.inputInfo_ = inputInfo_;
-        } else {
-          result.inputInfo_ = inputInfoBuilder_.build();
-        }
-        if (outputInfoBuilder_ == null) {
-          if (((bitField0_ & 0x00000002) != 0)) {
-            outputInfo_ = java.util.Collections.unmodifiableList(outputInfo_);
-            bitField0_ = (bitField0_ & ~0x00000002);
-          }
-          result.outputInfo_ = outputInfo_;
-        } else {
-          result.outputInfo_ = outputInfoBuilder_.build();
-        }
-        result.temporaryMemorySize_ = temporaryMemorySize_;
-        result.persistentMemorySize_ = persistentMemorySize_;
-        result.hostTempMemorySize_ = hostTempMemorySize_;
-        result.deviceTempMemorySize_ = deviceTempMemorySize_;
-        result.devicePersistentMemorySize_ = devicePersistentMemorySize_;
-        result.computeCost_ = computeCost_;
-        result.computeTime_ = computeTime_;
-        result.memoryTime_ = memoryTime_;
-        result.isFinal_ = isFinal_;
-        if (((bitField0_ & 0x00000004) != 0)) {
-          controlInput_.makeImmutable();
-          bitField0_ = (bitField0_ & ~0x00000004);
-        }
-        result.controlInput_ = controlInput_;
-        result.inaccurate_ = inaccurate_;
-        onBuilt();
-        return result;
-      }
-
-      @java.lang.Override
-      public Builder clone() {
-        return super.clone();
-      }
-      @java.lang.Override
-      public Builder setField(
-          com.google.protobuf.Descriptors.FieldDescriptor field,
-          java.lang.Object value) {
-        return super.setField(field, value);
-      }
-      @java.lang.Override
-      public Builder clearField(
-          com.google.protobuf.Descriptors.FieldDescriptor field) {
-        return super.clearField(field);
-      }
-      @java.lang.Override
-      public Builder clearOneof(
-          com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-        return super.clearOneof(oneof);
-      }
-      @java.lang.Override
-      public Builder setRepeatedField(
-          com.google.protobuf.Descriptors.FieldDescriptor field,
-          int index, java.lang.Object value) {
-        return super.setRepeatedField(field, index, value);
-      }
-      @java.lang.Override
-      public Builder addRepeatedField(
-          com.google.protobuf.Descriptors.FieldDescriptor field,
-          java.lang.Object value) {
-        return super.addRepeatedField(field, value);
-      }
-      @java.lang.Override
-      public Builder mergeFrom(com.google.protobuf.Message other) {
-        if (other instanceof org.tensorflow.proto.framework.CostGraphDef.Node) {
-          return mergeFrom((org.tensorflow.proto.framework.CostGraphDef.Node)other);
-        } else {
-          super.mergeFrom(other);
-          return this;
-        }
-      }
-
-      public Builder mergeFrom(org.tensorflow.proto.framework.CostGraphDef.Node other) {
-        if (other == org.tensorflow.proto.framework.CostGraphDef.Node.getDefaultInstance()) return this;
-        if (!other.getName().isEmpty()) {
-          name_ = other.name_;
-          onChanged();
-        }
-        if (!other.getDevice().isEmpty()) {
-          device_ = other.device_;
-          onChanged();
-        }
-        if (other.getId() != 0) {
-          setId(other.getId());
-        }
-        if (inputInfoBuilder_ == null) {
-          if (!other.inputInfo_.isEmpty()) {
-            if (inputInfo_.isEmpty()) {
-              inputInfo_ = other.inputInfo_;
-              bitField0_ = (bitField0_ & ~0x00000001);
-            } else {
-              ensureInputInfoIsMutable();
-              inputInfo_.addAll(other.inputInfo_);
-            }
-            onChanged();
-          }
-        } else {
-          if (!other.inputInfo_.isEmpty()) {
-            if (inputInfoBuilder_.isEmpty()) {
-              inputInfoBuilder_.dispose();
-              inputInfoBuilder_ = null;
-              inputInfo_ = other.inputInfo_;
-              bitField0_ = (bitField0_ & ~0x00000001);
-              inputInfoBuilder_ = 
-                com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
-                   getInputInfoFieldBuilder() : null;
-            } else {
-              inputInfoBuilder_.addAllMessages(other.inputInfo_);
-            }
-          }
-        }
-        if (outputInfoBuilder_ == null) {
-          if (!other.outputInfo_.isEmpty()) {
-            if (outputInfo_.isEmpty()) {
-              outputInfo_ = other.outputInfo_;
-              bitField0_ = (bitField0_ & ~0x00000002);
-            } else {
-              ensureOutputInfoIsMutable();
-              outputInfo_.addAll(other.outputInfo_);
-            }
-            onChanged();
-          }
-        } else {
-          if (!other.outputInfo_.isEmpty()) {
-            if (outputInfoBuilder_.isEmpty()) {
-              outputInfoBuilder_.dispose();
-              outputInfoBuilder_ = null;
-              outputInfo_ = other.outputInfo_;
-              bitField0_ = (bitField0_ & ~0x00000002);
-              outputInfoBuilder_ = 
-                com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
-                   getOutputInfoFieldBuilder() : null;
-            } else {
-              outputInfoBuilder_.addAllMessages(other.outputInfo_);
-            }
-          }
-        }
-        if (other.getTemporaryMemorySize() != 0L) {
-          setTemporaryMemorySize(other.getTemporaryMemorySize());
-        }
-        if (other.getPersistentMemorySize() != 0L) {
-          setPersistentMemorySize(other.getPersistentMemorySize());
-        }
-        if (other.getHostTempMemorySize() != 0L) {
-          setHostTempMemorySize(other.getHostTempMemorySize());
-        }
-        if (other.getDeviceTempMemorySize() != 0L) {
-          setDeviceTempMemorySize(other.getDeviceTempMemorySize());
-        }
-        if (other.getDevicePersistentMemorySize() != 0L) {
-          setDevicePersistentMemorySize(other.getDevicePersistentMemorySize());
-        }
-        if (other.getComputeCost() != 0L) {
-          setComputeCost(other.getComputeCost());
-        }
-        if (other.getComputeTime() != 0L) {
-          setComputeTime(other.getComputeTime());
-        }
-        if (other.getMemoryTime() != 0L) {
-          setMemoryTime(other.getMemoryTime());
-        }
-        if (other.getIsFinal() != false) {
-          setIsFinal(other.getIsFinal());
-        }
-        if (!other.controlInput_.isEmpty()) {
-          if (controlInput_.isEmpty()) {
-            controlInput_ = other.controlInput_;
-            bitField0_ = (bitField0_ & ~0x00000004);
-          } else {
-            ensureControlInputIsMutable();
-            controlInput_.addAll(other.controlInput_);
-          }
-          onChanged();
-        }
-        if (other.getInaccurate() != false) {
-          setInaccurate(other.getInaccurate());
-        }
-        this.mergeUnknownFields(other.unknownFields);
-        onChanged();
-        return this;
-      }
-
-      @java.lang.Override
-      public final boolean isInitialized() {
-        return true;
-      }
-
-      @java.lang.Override
-      public Builder mergeFrom(
-          com.google.protobuf.CodedInputStream input,
-          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-          throws java.io.IOException {
-        org.tensorflow.proto.framework.CostGraphDef.Node parsedMessage = null;
-        try {
-          parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-        } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-          parsedMessage = (org.tensorflow.proto.framework.CostGraphDef.Node) e.getUnfinishedMessage();
-          throw e.unwrapIOException();
-        } finally {
-          if (parsedMessage != null) {
-            mergeFrom(parsedMessage);
-          }
-        }
-        return this;
-      }
-      private int bitField0_;
-
-      private java.lang.Object name_ = "";
-      /**
-       * <pre>
-       * The name of the node. Names are globally unique.
-       * </pre>
-       *
-       * <code>string name = 1;</code>
-       */
-      public java.lang.String getName() {
-        java.lang.Object ref = name_;
-        if (!(ref instanceof java.lang.String)) {
-          com.google.protobuf.ByteString bs =
-              (com.google.protobuf.ByteString) ref;
-          java.lang.String s = bs.toStringUtf8();
-          name_ = s;
-          return s;
-        } else {
-          return (java.lang.String) ref;
-        }
-      }
-      /**
-       * <pre>
-       * The name of the node. Names are globally unique.
-       * </pre>
-       *
-       * <code>string name = 1;</code>
-       */
-      public com.google.protobuf.ByteString
-          getNameBytes() {
-        java.lang.Object ref = name_;
-        if (ref instanceof String) {
-          com.google.protobuf.ByteString b = 
-              com.google.protobuf.ByteString.copyFromUtf8(
-                  (java.lang.String) ref);
-          name_ = b;
-          return b;
-        } else {
-          return (com.google.protobuf.ByteString) ref;
-        }
-      }
-      /**
-       * <pre>
-       * The name of the node. Names are globally unique.
-       * </pre>
-       *
-       * <code>string name = 1;</code>
-       */
-      public Builder setName(
-          java.lang.String value) {
-        if (value == null) {
-    throw new NullPointerException();
-  }
-  
-        name_ = value;
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * The name of the node. Names are globally unique.
-       * </pre>
-       *
-       * <code>string name = 1;</code>
-       */
-      public Builder clearName() {
-        
-        name_ = getDefaultInstance().getName();
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * The name of the node. Names are globally unique.
-       * </pre>
-       *
-       * <code>string name = 1;</code>
-       */
-      public Builder setNameBytes(
-          com.google.protobuf.ByteString value) {
-        if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-        
-        name_ = value;
-        onChanged();
-        return this;
-      }
-
-      private java.lang.Object device_ = "";
-      /**
-       * <pre>
-       * The device of the node. Can be empty if the node is mapped to the
-       * default partition or partitioning hasn't been run yet.
-       * </pre>
-       *
-       * <code>string device = 2;</code>
-       */
-      public java.lang.String getDevice() {
-        java.lang.Object ref = device_;
-        if (!(ref instanceof java.lang.String)) {
-          com.google.protobuf.ByteString bs =
-              (com.google.protobuf.ByteString) ref;
-          java.lang.String s = bs.toStringUtf8();
-          device_ = s;
-          return s;
-        } else {
-          return (java.lang.String) ref;
-        }
-      }
-      /**
-       * <pre>
-       * The device of the node. Can be empty if the node is mapped to the
-       * default partition or partitioning hasn't been run yet.
-       * </pre>
-       *
-       * <code>string device = 2;</code>
-       */
-      public com.google.protobuf.ByteString
-          getDeviceBytes() {
-        java.lang.Object ref = device_;
-        if (ref instanceof String) {
-          com.google.protobuf.ByteString b = 
-              com.google.protobuf.ByteString.copyFromUtf8(
-                  (java.lang.String) ref);
-          device_ = b;
-          return b;
-        } else {
-          return (com.google.protobuf.ByteString) ref;
-        }
-      }
-      /**
-       * <pre>
-       * The device of the node. Can be empty if the node is mapped to the
-       * default partition or partitioning hasn't been run yet.
-       * </pre>
-       *
-       * <code>string device = 2;</code>
-       */
-      public Builder setDevice(
-          java.lang.String value) {
-        if (value == null) {
-    throw new NullPointerException();
-  }
-  
-        device_ = value;
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * The device of the node. Can be empty if the node is mapped to the
-       * default partition or partitioning hasn't been run yet.
-       * </pre>
-       *
-       * <code>string device = 2;</code>
-       */
-      public Builder clearDevice() {
-        
-        device_ = getDefaultInstance().getDevice();
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * The device of the node. Can be empty if the node is mapped to the
-       * default partition or partitioning hasn't been run yet.
-       * </pre>
-       *
-       * <code>string device = 2;</code>
-       */
-      public Builder setDeviceBytes(
-          com.google.protobuf.ByteString value) {
-        if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-        
-        device_ = value;
-        onChanged();
-        return this;
-      }
-
-      private int id_ ;
-      /**
-       * <pre>
-       * The id of the node. Node ids are only unique inside a partition.
-       * </pre>
-       *
-       * <code>int32 id = 3;</code>
-       */
-      public int getId() {
-        return id_;
-      }
-      /**
-       * <pre>
-       * The id of the node. Node ids are only unique inside a partition.
-       * </pre>
-       *
-       * <code>int32 id = 3;</code>
-       */
-      public Builder setId(int value) {
-        
-        id_ = value;
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * The id of the node. Node ids are only unique inside a partition.
-       * </pre>
-       *
-       * <code>int32 id = 3;</code>
-       */
-      public Builder clearId() {
-        
-        id_ = 0;
-        onChanged();
-        return this;
-      }
-
-      private java.util.List<org.tensorflow.proto.framework.CostGraphDef.Node.InputInfo> inputInfo_ =
-        java.util.Collections.emptyList();
-      private void ensureInputInfoIsMutable() {
-        if (!((bitField0_ & 0x00000001) != 0)) {
-          inputInfo_ = new java.util.ArrayList<org.tensorflow.proto.framework.CostGraphDef.Node.InputInfo>(inputInfo_);
-          bitField0_ |= 0x00000001;
-         }
-      }
-
-      private com.google.protobuf.RepeatedFieldBuilderV3<
-          org.tensorflow.proto.framework.CostGraphDef.Node.InputInfo, org.tensorflow.proto.framework.CostGraphDef.Node.InputInfo.Builder, org.tensorflow.proto.framework.CostGraphDef.Node.InputInfoOrBuilder> inputInfoBuilder_;
-
-      /**
-       * <code>repeated .tensorflow.CostGraphDef.Node.InputInfo input_info = 4;</code>
-       */
-      public java.util.List<org.tensorflow.proto.framework.CostGraphDef.Node.InputInfo> getInputInfoList() {
-        if (inputInfoBuilder_ == null) {
-          return java.util.Collections.unmodifiableList(inputInfo_);
-        } else {
-          return inputInfoBuilder_.getMessageList();
-        }
-      }
-      /**
-       * <code>repeated .tensorflow.CostGraphDef.Node.InputInfo input_info = 4;</code>
-       */
-      public int getInputInfoCount() {
-        if (inputInfoBuilder_ == null) {
-          return inputInfo_.size();
-        } else {
-          return inputInfoBuilder_.getCount();
-        }
-      }
-      /**
-       * <code>repeated .tensorflow.CostGraphDef.Node.InputInfo input_info = 4;</code>
-       */
-      public org.tensorflow.proto.framework.CostGraphDef.Node.InputInfo getInputInfo(int index) {
-        if (inputInfoBuilder_ == null) {
-          return inputInfo_.get(index);
-        } else {
-          return inputInfoBuilder_.getMessage(index);
-        }
-      }
-      /**
-       * <code>repeated .tensorflow.CostGraphDef.Node.InputInfo input_info = 4;</code>
-       */
-      public Builder setInputInfo(
-          int index, org.tensorflow.proto.framework.CostGraphDef.Node.InputInfo value) {
-        if (inputInfoBuilder_ == null) {
-          if (value == null) {
-            throw new NullPointerException();
-          }
-          ensureInputInfoIsMutable();
-          inputInfo_.set(index, value);
-          onChanged();
-        } else {
-          inputInfoBuilder_.setMessage(index, value);
-        }
-        return this;
-      }
-      /**
-       * <code>repeated .tensorflow.CostGraphDef.Node.InputInfo input_info = 4;</code>
-       */
-      public Builder setInputInfo(
-          int index, org.tensorflow.proto.framework.CostGraphDef.Node.InputInfo.Builder builderForValue) {
-        if (inputInfoBuilder_ == null) {
-          ensureInputInfoIsMutable();
-          inputInfo_.set(index, builderForValue.build());
-          onChanged();
-        } else {
-          inputInfoBuilder_.setMessage(index, builderForValue.build());
-        }
-        return this;
-      }
-      /**
-       * <code>repeated .tensorflow.CostGraphDef.Node.InputInfo input_info = 4;</code>
-       */
-      public Builder addInputInfo(org.tensorflow.proto.framework.CostGraphDef.Node.InputInfo value) {
-        if (inputInfoBuilder_ == null) {
-          if (value == null) {
-            throw new NullPointerException();
-          }
-          ensureInputInfoIsMutable();
-          inputInfo_.add(value);
-          onChanged();
-        } else {
-          inputInfoBuilder_.addMessage(value);
-        }
-        return this;
-      }
-      /**
-       * <code>repeated .tensorflow.CostGraphDef.Node.InputInfo input_info = 4;</code>
-       */
-      public Builder addInputInfo(
-          int index, org.tensorflow.proto.framework.CostGraphDef.Node.InputInfo value) {
-        if (inputInfoBuilder_ == null) {
-          if (value == null) {
-            throw new NullPointerException();
-          }
-          ensureInputInfoIsMutable();
-          inputInfo_.add(index, value);
-          onChanged();
-        } else {
-          inputInfoBuilder_.addMessage(index, value);
-        }
-        return this;
-      }
-      /**
-       * <code>repeated .tensorflow.CostGraphDef.Node.InputInfo input_info = 4;</code>
-       */
-      public Builder addInputInfo(
-          org.tensorflow.proto.framework.CostGraphDef.Node.InputInfo.Builder builderForValue) {
-        if (inputInfoBuilder_ == null) {
-          ensureInputInfoIsMutable();
-          inputInfo_.add(builderForValue.build());
-          onChanged();
-        } else {
-          inputInfoBuilder_.addMessage(builderForValue.build());
-        }
-        return this;
-      }
-      /**
-       * <code>repeated .tensorflow.CostGraphDef.Node.InputInfo input_info = 4;</code>
-       */
-      public Builder addInputInfo(
-          int index, org.tensorflow.proto.framework.CostGraphDef.Node.InputInfo.Builder builderForValue) {
-        if (inputInfoBuilder_ == null) {
-          ensureInputInfoIsMutable();
-          inputInfo_.add(index, builderForValue.build());
-          onChanged();
-        } else {
-          inputInfoBuilder_.addMessage(index, builderForValue.build());
-        }
-        return this;
-      }
-      /**
-       * <code>repeated .tensorflow.CostGraphDef.Node.InputInfo input_info = 4;</code>
-       */
-      public Builder addAllInputInfo(
-          java.lang.Iterable<? extends org.tensorflow.proto.framework.CostGraphDef.Node.InputInfo> values) {
-        if (inputInfoBuilder_ == null) {
-          ensureInputInfoIsMutable();
-          com.google.protobuf.AbstractMessageLite.Builder.addAll(
-              values, inputInfo_);
-          onChanged();
-        } else {
-          inputInfoBuilder_.addAllMessages(values);
-        }
-        return this;
-      }
-      /**
-       * <code>repeated .tensorflow.CostGraphDef.Node.InputInfo input_info = 4;</code>
-       */
-      public Builder clearInputInfo() {
-        if (inputInfoBuilder_ == null) {
-          inputInfo_ = java.util.Collections.emptyList();
-          bitField0_ = (bitField0_ & ~0x00000001);
-          onChanged();
-        } else {
-          inputInfoBuilder_.clear();
-        }
-        return this;
-      }
-      /**
-       * <code>repeated .tensorflow.CostGraphDef.Node.InputInfo input_info = 4;</code>
-       */
-      public Builder removeInputInfo(int index) {
-        if (inputInfoBuilder_ == null) {
-          ensureInputInfoIsMutable();
-          inputInfo_.remove(index);
-          onChanged();
-        } else {
-          inputInfoBuilder_.remove(index);
-        }
-        return this;
-      }
-      /**
-       * <code>repeated .tensorflow.CostGraphDef.Node.InputInfo input_info = 4;</code>
-       */
-      public org.tensorflow.proto.framework.CostGraphDef.Node.InputInfo.Builder getInputInfoBuilder(
-          int index) {
-        return getInputInfoFieldBuilder().getBuilder(index);
-      }
-      /**
-       * <code>repeated .tensorflow.CostGraphDef.Node.InputInfo input_info = 4;</code>
-       */
-      public org.tensorflow.proto.framework.CostGraphDef.Node.InputInfoOrBuilder getInputInfoOrBuilder(
-          int index) {
-        if (inputInfoBuilder_ == null) {
-          return inputInfo_.get(index);  } else {
-          return inputInfoBuilder_.getMessageOrBuilder(index);
-        }
-      }
-      /**
-       * <code>repeated .tensorflow.CostGraphDef.Node.InputInfo input_info = 4;</code>
-       */
-      public java.util.List<? extends org.tensorflow.proto.framework.CostGraphDef.Node.InputInfoOrBuilder> 
-           getInputInfoOrBuilderList() {
-        if (inputInfoBuilder_ != null) {
-          return inputInfoBuilder_.getMessageOrBuilderList();
-        } else {
-          return java.util.Collections.unmodifiableList(inputInfo_);
-        }
-      }
-      /**
-       * <code>repeated .tensorflow.CostGraphDef.Node.InputInfo input_info = 4;</code>
-       */
-      public org.tensorflow.proto.framework.CostGraphDef.Node.InputInfo.Builder addInputInfoBuilder() {
-        return getInputInfoFieldBuilder().addBuilder(
-            org.tensorflow.proto.framework.CostGraphDef.Node.InputInfo.getDefaultInstance());
-      }
-      /**
-       * <code>repeated .tensorflow.CostGraphDef.Node.InputInfo input_info = 4;</code>
-       */
-      public org.tensorflow.proto.framework.CostGraphDef.Node.InputInfo.Builder addInputInfoBuilder(
-          int index) {
-        return getInputInfoFieldBuilder().addBuilder(
-            index, org.tensorflow.proto.framework.CostGraphDef.Node.InputInfo.getDefaultInstance());
-      }
-      /**
-       * <code>repeated .tensorflow.CostGraphDef.Node.InputInfo input_info = 4;</code>
-       */
-      public java.util.List<org.tensorflow.proto.framework.CostGraphDef.Node.InputInfo.Builder> 
-           getInputInfoBuilderList() {
-        return getInputInfoFieldBuilder().getBuilderList();
-      }
-      private com.google.protobuf.RepeatedFieldBuilderV3<
-          org.tensorflow.proto.framework.CostGraphDef.Node.InputInfo, org.tensorflow.proto.framework.CostGraphDef.Node.InputInfo.Builder, org.tensorflow.proto.framework.CostGraphDef.Node.InputInfoOrBuilder> 
-          getInputInfoFieldBuilder() {
-        if (inputInfoBuilder_ == null) {
-          inputInfoBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
-              org.tensorflow.proto.framework.CostGraphDef.Node.InputInfo, org.tensorflow.proto.framework.CostGraphDef.Node.InputInfo.Builder, org.tensorflow.proto.framework.CostGraphDef.Node.InputInfoOrBuilder>(
-                  inputInfo_,
-                  ((bitField0_ & 0x00000001) != 0),
-                  getParentForChildren(),
-                  isClean());
-          inputInfo_ = null;
-        }
-        return inputInfoBuilder_;
-      }
-
-      private java.util.List<org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfo> outputInfo_ =
-        java.util.Collections.emptyList();
-      private void ensureOutputInfoIsMutable() {
-        if (!((bitField0_ & 0x00000002) != 0)) {
-          outputInfo_ = new java.util.ArrayList<org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfo>(outputInfo_);
-          bitField0_ |= 0x00000002;
-         }
-      }
-
-      private com.google.protobuf.RepeatedFieldBuilderV3<
-          org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfo, org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfo.Builder, org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfoOrBuilder> outputInfoBuilder_;
-
-      /**
-       * <code>repeated .tensorflow.CostGraphDef.Node.OutputInfo output_info = 5;</code>
-       */
-      public java.util.List<org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfo> getOutputInfoList() {
-        if (outputInfoBuilder_ == null) {
-          return java.util.Collections.unmodifiableList(outputInfo_);
-        } else {
-          return outputInfoBuilder_.getMessageList();
-        }
-      }
-      /**
-       * <code>repeated .tensorflow.CostGraphDef.Node.OutputInfo output_info = 5;</code>
-       */
-      public int getOutputInfoCount() {
-        if (outputInfoBuilder_ == null) {
-          return outputInfo_.size();
-        } else {
-          return outputInfoBuilder_.getCount();
-        }
-      }
-      /**
-       * <code>repeated .tensorflow.CostGraphDef.Node.OutputInfo output_info = 5;</code>
-       */
-      public org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfo getOutputInfo(int index) {
-        if (outputInfoBuilder_ == null) {
-          return outputInfo_.get(index);
-        } else {
-          return outputInfoBuilder_.getMessage(index);
-        }
-      }
-      /**
-       * <code>repeated .tensorflow.CostGraphDef.Node.OutputInfo output_info = 5;</code>
-       */
-      public Builder setOutputInfo(
-          int index, org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfo value) {
-        if (outputInfoBuilder_ == null) {
-          if (value == null) {
-            throw new NullPointerException();
-          }
-          ensureOutputInfoIsMutable();
-          outputInfo_.set(index, value);
-          onChanged();
-        } else {
-          outputInfoBuilder_.setMessage(index, value);
-        }
-        return this;
-      }
-      /**
-       * <code>repeated .tensorflow.CostGraphDef.Node.OutputInfo output_info = 5;</code>
-       */
-      public Builder setOutputInfo(
-          int index, org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfo.Builder builderForValue) {
-        if (outputInfoBuilder_ == null) {
-          ensureOutputInfoIsMutable();
-          outputInfo_.set(index, builderForValue.build());
-          onChanged();
-        } else {
-          outputInfoBuilder_.setMessage(index, builderForValue.build());
-        }
-        return this;
-      }
-      /**
-       * <code>repeated .tensorflow.CostGraphDef.Node.OutputInfo output_info = 5;</code>
-       */
-      public Builder addOutputInfo(org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfo value) {
-        if (outputInfoBuilder_ == null) {
-          if (value == null) {
-            throw new NullPointerException();
-          }
-          ensureOutputInfoIsMutable();
-          outputInfo_.add(value);
-          onChanged();
-        } else {
-          outputInfoBuilder_.addMessage(value);
-        }
-        return this;
-      }
-      /**
-       * <code>repeated .tensorflow.CostGraphDef.Node.OutputInfo output_info = 5;</code>
-       */
-      public Builder addOutputInfo(
-          int index, org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfo value) {
-        if (outputInfoBuilder_ == null) {
-          if (value == null) {
-            throw new NullPointerException();
-          }
-          ensureOutputInfoIsMutable();
-          outputInfo_.add(index, value);
-          onChanged();
-        } else {
-          outputInfoBuilder_.addMessage(index, value);
-        }
-        return this;
-      }
-      /**
-       * <code>repeated .tensorflow.CostGraphDef.Node.OutputInfo output_info = 5;</code>
-       */
-      public Builder addOutputInfo(
-          org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfo.Builder builderForValue) {
-        if (outputInfoBuilder_ == null) {
-          ensureOutputInfoIsMutable();
-          outputInfo_.add(builderForValue.build());
-          onChanged();
-        } else {
-          outputInfoBuilder_.addMessage(builderForValue.build());
-        }
-        return this;
-      }
-      /**
-       * <code>repeated .tensorflow.CostGraphDef.Node.OutputInfo output_info = 5;</code>
-       */
-      public Builder addOutputInfo(
-          int index, org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfo.Builder builderForValue) {
-        if (outputInfoBuilder_ == null) {
-          ensureOutputInfoIsMutable();
-          outputInfo_.add(index, builderForValue.build());
-          onChanged();
-        } else {
-          outputInfoBuilder_.addMessage(index, builderForValue.build());
-        }
-        return this;
-      }
-      /**
-       * <code>repeated .tensorflow.CostGraphDef.Node.OutputInfo output_info = 5;</code>
-       */
-      public Builder addAllOutputInfo(
-          java.lang.Iterable<? extends org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfo> values) {
-        if (outputInfoBuilder_ == null) {
-          ensureOutputInfoIsMutable();
-          com.google.protobuf.AbstractMessageLite.Builder.addAll(
-              values, outputInfo_);
-          onChanged();
-        } else {
-          outputInfoBuilder_.addAllMessages(values);
-        }
-        return this;
-      }
-      /**
-       * <code>repeated .tensorflow.CostGraphDef.Node.OutputInfo output_info = 5;</code>
-       */
-      public Builder clearOutputInfo() {
-        if (outputInfoBuilder_ == null) {
-          outputInfo_ = java.util.Collections.emptyList();
-          bitField0_ = (bitField0_ & ~0x00000002);
-          onChanged();
-        } else {
-          outputInfoBuilder_.clear();
-        }
-        return this;
-      }
-      /**
-       * <code>repeated .tensorflow.CostGraphDef.Node.OutputInfo output_info = 5;</code>
-       */
-      public Builder removeOutputInfo(int index) {
-        if (outputInfoBuilder_ == null) {
-          ensureOutputInfoIsMutable();
-          outputInfo_.remove(index);
-          onChanged();
-        } else {
-          outputInfoBuilder_.remove(index);
-        }
-        return this;
-      }
-      /**
-       * <code>repeated .tensorflow.CostGraphDef.Node.OutputInfo output_info = 5;</code>
-       */
-      public org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfo.Builder getOutputInfoBuilder(
-          int index) {
-        return getOutputInfoFieldBuilder().getBuilder(index);
-      }
-      /**
-       * <code>repeated .tensorflow.CostGraphDef.Node.OutputInfo output_info = 5;</code>
-       */
-      public org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfoOrBuilder getOutputInfoOrBuilder(
-          int index) {
-        if (outputInfoBuilder_ == null) {
-          return outputInfo_.get(index);  } else {
-          return outputInfoBuilder_.getMessageOrBuilder(index);
-        }
-      }
-      /**
-       * <code>repeated .tensorflow.CostGraphDef.Node.OutputInfo output_info = 5;</code>
-       */
-      public java.util.List<? extends org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfoOrBuilder> 
-           getOutputInfoOrBuilderList() {
-        if (outputInfoBuilder_ != null) {
-          return outputInfoBuilder_.getMessageOrBuilderList();
-        } else {
-          return java.util.Collections.unmodifiableList(outputInfo_);
-        }
-      }
-      /**
-       * <code>repeated .tensorflow.CostGraphDef.Node.OutputInfo output_info = 5;</code>
-       */
-      public org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfo.Builder addOutputInfoBuilder() {
-        return getOutputInfoFieldBuilder().addBuilder(
-            org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfo.getDefaultInstance());
-      }
-      /**
-       * <code>repeated .tensorflow.CostGraphDef.Node.OutputInfo output_info = 5;</code>
-       */
-      public org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfo.Builder addOutputInfoBuilder(
-          int index) {
-        return getOutputInfoFieldBuilder().addBuilder(
-            index, org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfo.getDefaultInstance());
-      }
-      /**
-       * <code>repeated .tensorflow.CostGraphDef.Node.OutputInfo output_info = 5;</code>
-       */
-      public java.util.List<org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfo.Builder> 
-           getOutputInfoBuilderList() {
-        return getOutputInfoFieldBuilder().getBuilderList();
-      }
-      private com.google.protobuf.RepeatedFieldBuilderV3<
-          org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfo, org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfo.Builder, org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfoOrBuilder> 
-          getOutputInfoFieldBuilder() {
-        if (outputInfoBuilder_ == null) {
-          outputInfoBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
-              org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfo, org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfo.Builder, org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfoOrBuilder>(
-                  outputInfo_,
-                  ((bitField0_ & 0x00000002) != 0),
-                  getParentForChildren(),
-                  isClean());
-          outputInfo_ = null;
-        }
-        return outputInfoBuilder_;
-      }
-
-      private long temporaryMemorySize_ ;
-      /**
-       * <pre>
-       * Temporary memory used by this node.
-       * </pre>
-       *
-       * <code>int64 temporary_memory_size = 6;</code>
-       */
-      public long getTemporaryMemorySize() {
-        return temporaryMemorySize_;
-      }
-      /**
-       * <pre>
-       * Temporary memory used by this node.
-       * </pre>
-       *
-       * <code>int64 temporary_memory_size = 6;</code>
-       */
-      public Builder setTemporaryMemorySize(long value) {
-        
-        temporaryMemorySize_ = value;
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * Temporary memory used by this node.
-       * </pre>
-       *
-       * <code>int64 temporary_memory_size = 6;</code>
-       */
-      public Builder clearTemporaryMemorySize() {
-        
-        temporaryMemorySize_ = 0L;
-        onChanged();
-        return this;
-      }
-
-      private long persistentMemorySize_ ;
-      /**
-       * <pre>
-       * Persistent memory used by this node.
-       * </pre>
-       *
-       * <code>int64 persistent_memory_size = 12;</code>
-       */
-      public long getPersistentMemorySize() {
-        return persistentMemorySize_;
-      }
-      /**
-       * <pre>
-       * Persistent memory used by this node.
-       * </pre>
-       *
-       * <code>int64 persistent_memory_size = 12;</code>
-       */
-      public Builder setPersistentMemorySize(long value) {
-        
-        persistentMemorySize_ = value;
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * Persistent memory used by this node.
-       * </pre>
-       *
-       * <code>int64 persistent_memory_size = 12;</code>
-       */
-      public Builder clearPersistentMemorySize() {
-        
-        persistentMemorySize_ = 0L;
-        onChanged();
-        return this;
-      }
-
-      private long hostTempMemorySize_ ;
-      /**
-       * <code>int64 host_temp_memory_size = 10 [deprecated = true];</code>
-       */
-      @java.lang.Deprecated public long getHostTempMemorySize() {
-        return hostTempMemorySize_;
-      }
-      /**
-       * <code>int64 host_temp_memory_size = 10 [deprecated = true];</code>
-       */
-      @java.lang.Deprecated public Builder setHostTempMemorySize(long value) {
-        
-        hostTempMemorySize_ = value;
-        onChanged();
-        return this;
-      }
-      /**
-       * <code>int64 host_temp_memory_size = 10 [deprecated = true];</code>
-       */
-      @java.lang.Deprecated public Builder clearHostTempMemorySize() {
-        
-        hostTempMemorySize_ = 0L;
-        onChanged();
-        return this;
-      }
-
-      private long deviceTempMemorySize_ ;
-      /**
-       * <code>int64 device_temp_memory_size = 11 [deprecated = true];</code>
-       */
-      @java.lang.Deprecated public long getDeviceTempMemorySize() {
-        return deviceTempMemorySize_;
-      }
-      /**
-       * <code>int64 device_temp_memory_size = 11 [deprecated = true];</code>
-       */
-      @java.lang.Deprecated public Builder setDeviceTempMemorySize(long value) {
-        
-        deviceTempMemorySize_ = value;
-        onChanged();
-        return this;
-      }
-      /**
-       * <code>int64 device_temp_memory_size = 11 [deprecated = true];</code>
-       */
-      @java.lang.Deprecated public Builder clearDeviceTempMemorySize() {
-        
-        deviceTempMemorySize_ = 0L;
-        onChanged();
-        return this;
-      }
-
-      private long devicePersistentMemorySize_ ;
-      /**
-       * <code>int64 device_persistent_memory_size = 16 [deprecated = true];</code>
-       */
-      @java.lang.Deprecated public long getDevicePersistentMemorySize() {
-        return devicePersistentMemorySize_;
-      }
-      /**
-       * <code>int64 device_persistent_memory_size = 16 [deprecated = true];</code>
-       */
-      @java.lang.Deprecated public Builder setDevicePersistentMemorySize(long value) {
-        
-        devicePersistentMemorySize_ = value;
-        onChanged();
-        return this;
-      }
-      /**
-       * <code>int64 device_persistent_memory_size = 16 [deprecated = true];</code>
-       */
-      @java.lang.Deprecated public Builder clearDevicePersistentMemorySize() {
-        
-        devicePersistentMemorySize_ = 0L;
-        onChanged();
-        return this;
-      }
-
-      private long computeCost_ ;
-      /**
-       * <pre>
-       * Estimate of the computational cost of this node, in microseconds.
-       * </pre>
-       *
-       * <code>int64 compute_cost = 9;</code>
-       */
-      public long getComputeCost() {
-        return computeCost_;
-      }
-      /**
-       * <pre>
-       * Estimate of the computational cost of this node, in microseconds.
-       * </pre>
-       *
-       * <code>int64 compute_cost = 9;</code>
-       */
-      public Builder setComputeCost(long value) {
-        
-        computeCost_ = value;
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * Estimate of the computational cost of this node, in microseconds.
-       * </pre>
-       *
-       * <code>int64 compute_cost = 9;</code>
-       */
-      public Builder clearComputeCost() {
-        
-        computeCost_ = 0L;
-        onChanged();
-        return this;
-      }
-
-      private long computeTime_ ;
-      /**
-       * <pre>
-       * Analytical estimate of the computational cost of this node, in
-       * microseconds.
-       * </pre>
-       *
-       * <code>int64 compute_time = 14;</code>
-       */
-      public long getComputeTime() {
-        return computeTime_;
-      }
-      /**
-       * <pre>
-       * Analytical estimate of the computational cost of this node, in
-       * microseconds.
-       * </pre>
-       *
-       * <code>int64 compute_time = 14;</code>
-       */
-      public Builder setComputeTime(long value) {
-        
-        computeTime_ = value;
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * Analytical estimate of the computational cost of this node, in
-       * microseconds.
-       * </pre>
-       *
-       * <code>int64 compute_time = 14;</code>
-       */
-      public Builder clearComputeTime() {
-        
-        computeTime_ = 0L;
-        onChanged();
-        return this;
-      }
-
-      private long memoryTime_ ;
-      /**
-       * <pre>
-       * Analytical estimate of the memory access cost of this node, in
-       * microseconds.
-       * </pre>
-       *
-       * <code>int64 memory_time = 15;</code>
-       */
-      public long getMemoryTime() {
-        return memoryTime_;
-      }
-      /**
-       * <pre>
-       * Analytical estimate of the memory access cost of this node, in
-       * microseconds.
-       * </pre>
-       *
-       * <code>int64 memory_time = 15;</code>
-       */
-      public Builder setMemoryTime(long value) {
-        
-        memoryTime_ = value;
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * Analytical estimate of the memory access cost of this node, in
-       * microseconds.
-       * </pre>
-       *
-       * <code>int64 memory_time = 15;</code>
-       */
-      public Builder clearMemoryTime() {
-        
-        memoryTime_ = 0L;
-        onChanged();
-        return this;
-      }
-
-      private boolean isFinal_ ;
-      /**
-       * <pre>
-       * If true, the output is permanent: it can't be discarded, because this
-       * node is part of the "final output". Nodes may depend on final nodes.
-       * </pre>
-       *
-       * <code>bool is_final = 7;</code>
-       */
-      public boolean getIsFinal() {
-        return isFinal_;
-      }
-      /**
-       * <pre>
-       * If true, the output is permanent: it can't be discarded, because this
-       * node is part of the "final output". Nodes may depend on final nodes.
-       * </pre>
-       *
-       * <code>bool is_final = 7;</code>
-       */
-      public Builder setIsFinal(boolean value) {
-        
-        isFinal_ = value;
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * If true, the output is permanent: it can't be discarded, because this
-       * node is part of the "final output". Nodes may depend on final nodes.
-       * </pre>
-       *
-       * <code>bool is_final = 7;</code>
-       */
-      public Builder clearIsFinal() {
-        
-        isFinal_ = false;
-        onChanged();
-        return this;
-      }
-
-      private com.google.protobuf.Internal.IntList controlInput_ = emptyIntList();
-      private void ensureControlInputIsMutable() {
-        if (!((bitField0_ & 0x00000004) != 0)) {
-          controlInput_ = mutableCopy(controlInput_);
-          bitField0_ |= 0x00000004;
-         }
-      }
-      /**
-       * <pre>
-       * Ids of the control inputs for this node.
-       * </pre>
-       *
-       * <code>repeated int32 control_input = 8;</code>
-       */
-      public java.util.List<java.lang.Integer>
-          getControlInputList() {
-        return ((bitField0_ & 0x00000004) != 0) ?
-                 java.util.Collections.unmodifiableList(controlInput_) : controlInput_;
-      }
-      /**
-       * <pre>
-       * Ids of the control inputs for this node.
-       * </pre>
-       *
-       * <code>repeated int32 control_input = 8;</code>
-       */
-      public int getControlInputCount() {
-        return controlInput_.size();
-      }
-      /**
-       * <pre>
-       * Ids of the control inputs for this node.
-       * </pre>
-       *
-       * <code>repeated int32 control_input = 8;</code>
-       */
-      public int getControlInput(int index) {
-        return controlInput_.getInt(index);
-      }
-      /**
-       * <pre>
-       * Ids of the control inputs for this node.
-       * </pre>
-       *
-       * <code>repeated int32 control_input = 8;</code>
-       */
-      public Builder setControlInput(
-          int index, int value) {
-        ensureControlInputIsMutable();
-        controlInput_.setInt(index, value);
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * Ids of the control inputs for this node.
-       * </pre>
-       *
-       * <code>repeated int32 control_input = 8;</code>
-       */
-      public Builder addControlInput(int value) {
-        ensureControlInputIsMutable();
-        controlInput_.addInt(value);
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * Ids of the control inputs for this node.
-       * </pre>
-       *
-       * <code>repeated int32 control_input = 8;</code>
-       */
-      public Builder addAllControlInput(
-          java.lang.Iterable<? extends java.lang.Integer> values) {
-        ensureControlInputIsMutable();
-        com.google.protobuf.AbstractMessageLite.Builder.addAll(
-            values, controlInput_);
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * Ids of the control inputs for this node.
-       * </pre>
-       *
-       * <code>repeated int32 control_input = 8;</code>
-       */
-      public Builder clearControlInput() {
-        controlInput_ = emptyIntList();
-        bitField0_ = (bitField0_ & ~0x00000004);
-        onChanged();
-        return this;
-      }
-
-      private boolean inaccurate_ ;
-      /**
-       * <pre>
-       * Are the costs inaccurate?
-       * </pre>
-       *
-       * <code>bool inaccurate = 17;</code>
-       */
-      public boolean getInaccurate() {
-        return inaccurate_;
-      }
-      /**
-       * <pre>
-       * Are the costs inaccurate?
-       * </pre>
-       *
-       * <code>bool inaccurate = 17;</code>
-       */
-      public Builder setInaccurate(boolean value) {
-        
-        inaccurate_ = value;
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * Are the costs inaccurate?
-       * </pre>
-       *
-       * <code>bool inaccurate = 17;</code>
-       */
-      public Builder clearInaccurate() {
-        
-        inaccurate_ = false;
-        onChanged();
-        return this;
-      }
-      @java.lang.Override
-      public final Builder setUnknownFields(
-          final com.google.protobuf.UnknownFieldSet unknownFields) {
-        return super.setUnknownFields(unknownFields);
-      }
-
-      @java.lang.Override
-      public final Builder mergeUnknownFields(
-          final com.google.protobuf.UnknownFieldSet unknownFields) {
-        return super.mergeUnknownFields(unknownFields);
-      }
-
-
-      // @@protoc_insertion_point(builder_scope:tensorflow.CostGraphDef.Node)
-    }
-
-    // @@protoc_insertion_point(class_scope:tensorflow.CostGraphDef.Node)
-    private static final org.tensorflow.proto.framework.CostGraphDef.Node DEFAULT_INSTANCE;
-    static {
-      DEFAULT_INSTANCE = new org.tensorflow.proto.framework.CostGraphDef.Node();
-    }
-
-    public static org.tensorflow.proto.framework.CostGraphDef.Node getDefaultInstance() {
-      return DEFAULT_INSTANCE;
-    }
-
-    private static final com.google.protobuf.Parser<Node>
-        PARSER = new com.google.protobuf.AbstractParser<Node>() {
-      @java.lang.Override
-      public Node parsePartialFrom(
-          com.google.protobuf.CodedInputStream input,
-          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-          throws com.google.protobuf.InvalidProtocolBufferException {
-        return new Node(input, extensionRegistry);
-      }
-    };
-
-    public static com.google.protobuf.Parser<Node> parser() {
-      return PARSER;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Parser<Node> getParserForType() {
-      return PARSER;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.CostGraphDef.Node getDefaultInstanceForType() {
-      return DEFAULT_INSTANCE;
-    }
-
-  }
-
-  public interface AggregatedCostOrBuilder extends
-      // @@protoc_insertion_point(interface_extends:tensorflow.CostGraphDef.AggregatedCost)
-      com.google.protobuf.MessageOrBuilder {
-
-    /**
-     * <pre>
-     * Aggregated cost value.
-     * </pre>
-     *
-     * <code>float cost = 1;</code>
-     */
-    float getCost();
-
-    /**
-     * <pre>
-     * Aggregated cost dimension (e.g. 'memory', 'compute', 'network').
-     * </pre>
-     *
-     * <code>string dimension = 2;</code>
-     */
-    java.lang.String getDimension();
-    /**
-     * <pre>
-     * Aggregated cost dimension (e.g. 'memory', 'compute', 'network').
-     * </pre>
-     *
-     * <code>string dimension = 2;</code>
-     */
-    com.google.protobuf.ByteString
-        getDimensionBytes();
-  }
-  /**
-   * <pre>
-   * Total cost of this graph, typically used for balancing decisions.
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.CostGraphDef.AggregatedCost}
-   */
-  public  static final class AggregatedCost extends
-      com.google.protobuf.GeneratedMessageV3 implements
-      // @@protoc_insertion_point(message_implements:tensorflow.CostGraphDef.AggregatedCost)
-      AggregatedCostOrBuilder {
-  private static final long serialVersionUID = 0L;
-    // Use AggregatedCost.newBuilder() to construct.
-    private AggregatedCost(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-      super(builder);
-    }
-    private AggregatedCost() {
-      dimension_ = "";
-    }
-
-    @java.lang.Override
-    @SuppressWarnings({"unused"})
-    protected java.lang.Object newInstance(
-        UnusedPrivateParameter unused) {
-      return new AggregatedCost();
-    }
-
-    @java.lang.Override
-    public final com.google.protobuf.UnknownFieldSet
-    getUnknownFields() {
-      return this.unknownFields;
-    }
-    private AggregatedCost(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      this();
-      if (extensionRegistry == null) {
-        throw new java.lang.NullPointerException();
-      }
-      com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-          com.google.protobuf.UnknownFieldSet.newBuilder();
-      try {
-        boolean done = false;
-        while (!done) {
-          int tag = input.readTag();
-          switch (tag) {
-            case 0:
-              done = true;
-              break;
-            case 13: {
-
-              cost_ = input.readFloat();
-              break;
-            }
-            case 18: {
-              java.lang.String s = input.readStringRequireUtf8();
-
-              dimension_ = s;
-              break;
-            }
-            default: {
-              if (!parseUnknownField(
-                  input, unknownFields, extensionRegistry, tag)) {
-                done = true;
-              }
-              break;
-            }
-          }
-        }
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        throw e.setUnfinishedMessage(this);
-      } catch (java.io.IOException e) {
-        throw new com.google.protobuf.InvalidProtocolBufferException(
-            e).setUnfinishedMessage(this);
-      } finally {
-        this.unknownFields = unknownFields.build();
-        makeExtensionsImmutable();
-      }
-    }
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.CostGraphProtos.internal_static_tensorflow_CostGraphDef_AggregatedCost_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.CostGraphProtos.internal_static_tensorflow_CostGraphDef_AggregatedCost_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.CostGraphDef.AggregatedCost.class, org.tensorflow.proto.framework.CostGraphDef.AggregatedCost.Builder.class);
-    }
-
-    public static final int COST_FIELD_NUMBER = 1;
-    private float cost_;
-    /**
-     * <pre>
-     * Aggregated cost value.
-     * </pre>
-     *
-     * <code>float cost = 1;</code>
-     */
-    public float getCost() {
-      return cost_;
-    }
-
-    public static final int DIMENSION_FIELD_NUMBER = 2;
-    private volatile java.lang.Object dimension_;
-    /**
-     * <pre>
-     * Aggregated cost dimension (e.g. 'memory', 'compute', 'network').
-     * </pre>
-     *
-     * <code>string dimension = 2;</code>
-     */
-    public java.lang.String getDimension() {
-      java.lang.Object ref = dimension_;
-      if (ref instanceof java.lang.String) {
-        return (java.lang.String) ref;
-      } else {
-        com.google.protobuf.ByteString bs = 
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        dimension_ = s;
-        return s;
-      }
-    }
-    /**
-     * <pre>
-     * Aggregated cost dimension (e.g. 'memory', 'compute', 'network').
-     * </pre>
-     *
-     * <code>string dimension = 2;</code>
-     */
-    public com.google.protobuf.ByteString
-        getDimensionBytes() {
-      java.lang.Object ref = dimension_;
-      if (ref instanceof java.lang.String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        dimension_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-
-    private byte memoizedIsInitialized = -1;
-    @java.lang.Override
-    public final boolean isInitialized() {
-      byte isInitialized = memoizedIsInitialized;
-      if (isInitialized == 1) return true;
-      if (isInitialized == 0) return false;
-
-      memoizedIsInitialized = 1;
-      return true;
-    }
-
-    @java.lang.Override
-    public void writeTo(com.google.protobuf.CodedOutputStream output)
-                        throws java.io.IOException {
-      if (cost_ != 0F) {
-        output.writeFloat(1, cost_);
-      }
-      if (!getDimensionBytes().isEmpty()) {
-        com.google.protobuf.GeneratedMessageV3.writeString(output, 2, dimension_);
-      }
-      unknownFields.writeTo(output);
-    }
-
-    @java.lang.Override
-    public int getSerializedSize() {
-      int size = memoizedSize;
-      if (size != -1) return size;
-
-      size = 0;
-      if (cost_ != 0F) {
-        size += com.google.protobuf.CodedOutputStream
-          .computeFloatSize(1, cost_);
-      }
-      if (!getDimensionBytes().isEmpty()) {
-        size += com.google.protobuf.GeneratedMessageV3.computeStringSize(2, dimension_);
-      }
-      size += unknownFields.getSerializedSize();
-      memoizedSize = size;
-      return size;
-    }
-
-    @java.lang.Override
-    public boolean equals(final java.lang.Object obj) {
-      if (obj == this) {
-       return true;
-      }
-      if (!(obj instanceof org.tensorflow.proto.framework.CostGraphDef.AggregatedCost)) {
-        return super.equals(obj);
-      }
-      org.tensorflow.proto.framework.CostGraphDef.AggregatedCost other = (org.tensorflow.proto.framework.CostGraphDef.AggregatedCost) obj;
-
-      if (java.lang.Float.floatToIntBits(getCost())
-          != java.lang.Float.floatToIntBits(
-              other.getCost())) return false;
-      if (!getDimension()
-          .equals(other.getDimension())) return false;
-      if (!unknownFields.equals(other.unknownFields)) return false;
-      return true;
-    }
-
-    @java.lang.Override
-    public int hashCode() {
-      if (memoizedHashCode != 0) {
-        return memoizedHashCode;
-      }
-      int hash = 41;
-      hash = (19 * hash) + getDescriptor().hashCode();
-      hash = (37 * hash) + COST_FIELD_NUMBER;
-      hash = (53 * hash) + java.lang.Float.floatToIntBits(
-          getCost());
-      hash = (37 * hash) + DIMENSION_FIELD_NUMBER;
-      hash = (53 * hash) + getDimension().hashCode();
-      hash = (29 * hash) + unknownFields.hashCode();
-      memoizedHashCode = hash;
-      return hash;
-    }
-
-    public static org.tensorflow.proto.framework.CostGraphDef.AggregatedCost parseFrom(
-        java.nio.ByteBuffer data)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data);
-    }
-    public static org.tensorflow.proto.framework.CostGraphDef.AggregatedCost parseFrom(
-        java.nio.ByteBuffer data,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.CostGraphDef.AggregatedCost parseFrom(
-        com.google.protobuf.ByteString data)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data);
-    }
-    public static org.tensorflow.proto.framework.CostGraphDef.AggregatedCost parseFrom(
-        com.google.protobuf.ByteString data,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.CostGraphDef.AggregatedCost parseFrom(byte[] data)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data);
-    }
-    public static org.tensorflow.proto.framework.CostGraphDef.AggregatedCost parseFrom(
-        byte[] data,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.CostGraphDef.AggregatedCost parseFrom(java.io.InputStream input)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input);
-    }
-    public static org.tensorflow.proto.framework.CostGraphDef.AggregatedCost parseFrom(
-        java.io.InputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.CostGraphDef.AggregatedCost parseDelimitedFrom(java.io.InputStream input)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseDelimitedWithIOException(PARSER, input);
-    }
-    public static org.tensorflow.proto.framework.CostGraphDef.AggregatedCost parseDelimitedFrom(
-        java.io.InputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.CostGraphDef.AggregatedCost parseFrom(
-        com.google.protobuf.CodedInputStream input)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input);
-    }
-    public static org.tensorflow.proto.framework.CostGraphDef.AggregatedCost parseFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input, extensionRegistry);
-    }
-
-    @java.lang.Override
-    public Builder newBuilderForType() { return newBuilder(); }
-    public static Builder newBuilder() {
-      return DEFAULT_INSTANCE.toBuilder();
-    }
-    public static Builder newBuilder(org.tensorflow.proto.framework.CostGraphDef.AggregatedCost prototype) {
-      return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-    }
-    @java.lang.Override
-    public Builder toBuilder() {
-      return this == DEFAULT_INSTANCE
-          ? new Builder() : new Builder().mergeFrom(this);
-    }
-
-    @java.lang.Override
-    protected Builder newBuilderForType(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      Builder builder = new Builder(parent);
-      return builder;
-    }
-    /**
-     * <pre>
-     * Total cost of this graph, typically used for balancing decisions.
-     * </pre>
-     *
-     * Protobuf type {@code tensorflow.CostGraphDef.AggregatedCost}
-     */
-    public static final class Builder extends
-        com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-        // @@protoc_insertion_point(builder_implements:tensorflow.CostGraphDef.AggregatedCost)
-        org.tensorflow.proto.framework.CostGraphDef.AggregatedCostOrBuilder {
-      public static final com.google.protobuf.Descriptors.Descriptor
-          getDescriptor() {
-        return org.tensorflow.proto.framework.CostGraphProtos.internal_static_tensorflow_CostGraphDef_AggregatedCost_descriptor;
-      }
-
-      @java.lang.Override
-      protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-          internalGetFieldAccessorTable() {
-        return org.tensorflow.proto.framework.CostGraphProtos.internal_static_tensorflow_CostGraphDef_AggregatedCost_fieldAccessorTable
-            .ensureFieldAccessorsInitialized(
-                org.tensorflow.proto.framework.CostGraphDef.AggregatedCost.class, org.tensorflow.proto.framework.CostGraphDef.AggregatedCost.Builder.class);
-      }
-
-      // Construct using org.tensorflow.proto.framework.CostGraphDef.AggregatedCost.newBuilder()
-      private Builder() {
-        maybeForceBuilderInitialization();
-      }
-
-      private Builder(
-          com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-        super(parent);
-        maybeForceBuilderInitialization();
-      }
-      private void maybeForceBuilderInitialization() {
-        if (com.google.protobuf.GeneratedMessageV3
-                .alwaysUseFieldBuilders) {
-        }
-      }
-      @java.lang.Override
-      public Builder clear() {
-        super.clear();
-        cost_ = 0F;
-
-        dimension_ = "";
-
-        return this;
-      }
-
-      @java.lang.Override
-      public com.google.protobuf.Descriptors.Descriptor
-          getDescriptorForType() {
-        return org.tensorflow.proto.framework.CostGraphProtos.internal_static_tensorflow_CostGraphDef_AggregatedCost_descriptor;
-      }
-
-      @java.lang.Override
-      public org.tensorflow.proto.framework.CostGraphDef.AggregatedCost getDefaultInstanceForType() {
-        return org.tensorflow.proto.framework.CostGraphDef.AggregatedCost.getDefaultInstance();
-      }
-
-      @java.lang.Override
-      public org.tensorflow.proto.framework.CostGraphDef.AggregatedCost build() {
-        org.tensorflow.proto.framework.CostGraphDef.AggregatedCost result = buildPartial();
-        if (!result.isInitialized()) {
-          throw newUninitializedMessageException(result);
-        }
-        return result;
-      }
-
-      @java.lang.Override
-      public org.tensorflow.proto.framework.CostGraphDef.AggregatedCost buildPartial() {
-        org.tensorflow.proto.framework.CostGraphDef.AggregatedCost result = new org.tensorflow.proto.framework.CostGraphDef.AggregatedCost(this);
-        result.cost_ = cost_;
-        result.dimension_ = dimension_;
-        onBuilt();
-        return result;
-      }
-
-      @java.lang.Override
-      public Builder clone() {
-        return super.clone();
-      }
-      @java.lang.Override
-      public Builder setField(
-          com.google.protobuf.Descriptors.FieldDescriptor field,
-          java.lang.Object value) {
-        return super.setField(field, value);
-      }
-      @java.lang.Override
-      public Builder clearField(
-          com.google.protobuf.Descriptors.FieldDescriptor field) {
-        return super.clearField(field);
-      }
-      @java.lang.Override
-      public Builder clearOneof(
-          com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-        return super.clearOneof(oneof);
-      }
-      @java.lang.Override
-      public Builder setRepeatedField(
-          com.google.protobuf.Descriptors.FieldDescriptor field,
-          int index, java.lang.Object value) {
-        return super.setRepeatedField(field, index, value);
-      }
-      @java.lang.Override
-      public Builder addRepeatedField(
-          com.google.protobuf.Descriptors.FieldDescriptor field,
-          java.lang.Object value) {
-        return super.addRepeatedField(field, value);
-      }
-      @java.lang.Override
-      public Builder mergeFrom(com.google.protobuf.Message other) {
-        if (other instanceof org.tensorflow.proto.framework.CostGraphDef.AggregatedCost) {
-          return mergeFrom((org.tensorflow.proto.framework.CostGraphDef.AggregatedCost)other);
-        } else {
-          super.mergeFrom(other);
-          return this;
-        }
-      }
-
-      public Builder mergeFrom(org.tensorflow.proto.framework.CostGraphDef.AggregatedCost other) {
-        if (other == org.tensorflow.proto.framework.CostGraphDef.AggregatedCost.getDefaultInstance()) return this;
-        if (other.getCost() != 0F) {
-          setCost(other.getCost());
-        }
-        if (!other.getDimension().isEmpty()) {
-          dimension_ = other.dimension_;
-          onChanged();
-        }
-        this.mergeUnknownFields(other.unknownFields);
-        onChanged();
-        return this;
-      }
-
-      @java.lang.Override
-      public final boolean isInitialized() {
-        return true;
-      }
-
-      @java.lang.Override
-      public Builder mergeFrom(
-          com.google.protobuf.CodedInputStream input,
-          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-          throws java.io.IOException {
-        org.tensorflow.proto.framework.CostGraphDef.AggregatedCost parsedMessage = null;
-        try {
-          parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-        } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-          parsedMessage = (org.tensorflow.proto.framework.CostGraphDef.AggregatedCost) e.getUnfinishedMessage();
-          throw e.unwrapIOException();
-        } finally {
-          if (parsedMessage != null) {
-            mergeFrom(parsedMessage);
-          }
-        }
-        return this;
-      }
-
-      private float cost_ ;
-      /**
-       * <pre>
-       * Aggregated cost value.
-       * </pre>
-       *
-       * <code>float cost = 1;</code>
-       */
-      public float getCost() {
-        return cost_;
-      }
-      /**
-       * <pre>
-       * Aggregated cost value.
-       * </pre>
-       *
-       * <code>float cost = 1;</code>
-       */
-      public Builder setCost(float value) {
-        
-        cost_ = value;
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * Aggregated cost value.
-       * </pre>
-       *
-       * <code>float cost = 1;</code>
-       */
-      public Builder clearCost() {
-        
-        cost_ = 0F;
-        onChanged();
-        return this;
-      }
-
-      private java.lang.Object dimension_ = "";
-      /**
-       * <pre>
-       * Aggregated cost dimension (e.g. 'memory', 'compute', 'network').
-       * </pre>
-       *
-       * <code>string dimension = 2;</code>
-       */
-      public java.lang.String getDimension() {
-        java.lang.Object ref = dimension_;
-        if (!(ref instanceof java.lang.String)) {
-          com.google.protobuf.ByteString bs =
-              (com.google.protobuf.ByteString) ref;
-          java.lang.String s = bs.toStringUtf8();
-          dimension_ = s;
-          return s;
-        } else {
-          return (java.lang.String) ref;
-        }
-      }
-      /**
-       * <pre>
-       * Aggregated cost dimension (e.g. 'memory', 'compute', 'network').
-       * </pre>
-       *
-       * <code>string dimension = 2;</code>
-       */
-      public com.google.protobuf.ByteString
-          getDimensionBytes() {
-        java.lang.Object ref = dimension_;
-        if (ref instanceof String) {
-          com.google.protobuf.ByteString b = 
-              com.google.protobuf.ByteString.copyFromUtf8(
-                  (java.lang.String) ref);
-          dimension_ = b;
-          return b;
-        } else {
-          return (com.google.protobuf.ByteString) ref;
-        }
-      }
-      /**
-       * <pre>
-       * Aggregated cost dimension (e.g. 'memory', 'compute', 'network').
-       * </pre>
-       *
-       * <code>string dimension = 2;</code>
-       */
-      public Builder setDimension(
-          java.lang.String value) {
-        if (value == null) {
-    throw new NullPointerException();
-  }
-  
-        dimension_ = value;
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * Aggregated cost dimension (e.g. 'memory', 'compute', 'network').
-       * </pre>
-       *
-       * <code>string dimension = 2;</code>
-       */
-      public Builder clearDimension() {
-        
-        dimension_ = getDefaultInstance().getDimension();
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * Aggregated cost dimension (e.g. 'memory', 'compute', 'network').
-       * </pre>
-       *
-       * <code>string dimension = 2;</code>
-       */
-      public Builder setDimensionBytes(
-          com.google.protobuf.ByteString value) {
-        if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-        
-        dimension_ = value;
-        onChanged();
-        return this;
-      }
-      @java.lang.Override
-      public final Builder setUnknownFields(
-          final com.google.protobuf.UnknownFieldSet unknownFields) {
-        return super.setUnknownFields(unknownFields);
-      }
-
-      @java.lang.Override
-      public final Builder mergeUnknownFields(
-          final com.google.protobuf.UnknownFieldSet unknownFields) {
-        return super.mergeUnknownFields(unknownFields);
-      }
-
-
-      // @@protoc_insertion_point(builder_scope:tensorflow.CostGraphDef.AggregatedCost)
-    }
-
-    // @@protoc_insertion_point(class_scope:tensorflow.CostGraphDef.AggregatedCost)
-    private static final org.tensorflow.proto.framework.CostGraphDef.AggregatedCost DEFAULT_INSTANCE;
-    static {
-      DEFAULT_INSTANCE = new org.tensorflow.proto.framework.CostGraphDef.AggregatedCost();
-    }
-
-    public static org.tensorflow.proto.framework.CostGraphDef.AggregatedCost getDefaultInstance() {
-      return DEFAULT_INSTANCE;
-    }
-
-    private static final com.google.protobuf.Parser<AggregatedCost>
-        PARSER = new com.google.protobuf.AbstractParser<AggregatedCost>() {
-      @java.lang.Override
-      public AggregatedCost parsePartialFrom(
-          com.google.protobuf.CodedInputStream input,
-          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-          throws com.google.protobuf.InvalidProtocolBufferException {
-        return new AggregatedCost(input, extensionRegistry);
-      }
-    };
-
-    public static com.google.protobuf.Parser<AggregatedCost> parser() {
-      return PARSER;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Parser<AggregatedCost> getParserForType() {
-      return PARSER;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.CostGraphDef.AggregatedCost getDefaultInstanceForType() {
-      return DEFAULT_INSTANCE;
-    }
-
-  }
-
-  public static final int NODE_FIELD_NUMBER = 1;
-  private java.util.List<org.tensorflow.proto.framework.CostGraphDef.Node> node_;
-  /**
-   * <code>repeated .tensorflow.CostGraphDef.Node node = 1;</code>
-   */
-  public java.util.List<org.tensorflow.proto.framework.CostGraphDef.Node> getNodeList() {
-    return node_;
-  }
-  /**
-   * <code>repeated .tensorflow.CostGraphDef.Node node = 1;</code>
-   */
-  public java.util.List<? extends org.tensorflow.proto.framework.CostGraphDef.NodeOrBuilder> 
-      getNodeOrBuilderList() {
-    return node_;
-  }
-  /**
-   * <code>repeated .tensorflow.CostGraphDef.Node node = 1;</code>
-   */
-  public int getNodeCount() {
-    return node_.size();
-  }
-  /**
-   * <code>repeated .tensorflow.CostGraphDef.Node node = 1;</code>
-   */
-  public org.tensorflow.proto.framework.CostGraphDef.Node getNode(int index) {
-    return node_.get(index);
-  }
-  /**
-   * <code>repeated .tensorflow.CostGraphDef.Node node = 1;</code>
-   */
-  public org.tensorflow.proto.framework.CostGraphDef.NodeOrBuilder getNodeOrBuilder(
-      int index) {
-    return node_.get(index);
-  }
-
-  public static final int COST_FIELD_NUMBER = 2;
-  private java.util.List<org.tensorflow.proto.framework.CostGraphDef.AggregatedCost> cost_;
-  /**
-   * <code>repeated .tensorflow.CostGraphDef.AggregatedCost cost = 2;</code>
-   */
-  public java.util.List<org.tensorflow.proto.framework.CostGraphDef.AggregatedCost> getCostList() {
-    return cost_;
-  }
-  /**
-   * <code>repeated .tensorflow.CostGraphDef.AggregatedCost cost = 2;</code>
-   */
-  public java.util.List<? extends org.tensorflow.proto.framework.CostGraphDef.AggregatedCostOrBuilder> 
-      getCostOrBuilderList() {
-    return cost_;
-  }
-  /**
-   * <code>repeated .tensorflow.CostGraphDef.AggregatedCost cost = 2;</code>
-   */
-  public int getCostCount() {
-    return cost_.size();
-  }
-  /**
-   * <code>repeated .tensorflow.CostGraphDef.AggregatedCost cost = 2;</code>
-   */
-  public org.tensorflow.proto.framework.CostGraphDef.AggregatedCost getCost(int index) {
-    return cost_.get(index);
-  }
-  /**
-   * <code>repeated .tensorflow.CostGraphDef.AggregatedCost cost = 2;</code>
-   */
-  public org.tensorflow.proto.framework.CostGraphDef.AggregatedCostOrBuilder getCostOrBuilder(
-      int index) {
-    return cost_.get(index);
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    for (int i = 0; i < node_.size(); i++) {
-      output.writeMessage(1, node_.get(i));
-    }
-    for (int i = 0; i < cost_.size(); i++) {
-      output.writeMessage(2, cost_.get(i));
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    for (int i = 0; i < node_.size(); i++) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(1, node_.get(i));
-    }
-    for (int i = 0; i < cost_.size(); i++) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(2, cost_.get(i));
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.framework.CostGraphDef)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.framework.CostGraphDef other = (org.tensorflow.proto.framework.CostGraphDef) obj;
-
-    if (!getNodeList()
-        .equals(other.getNodeList())) return false;
-    if (!getCostList()
-        .equals(other.getCostList())) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    if (getNodeCount() > 0) {
-      hash = (37 * hash) + NODE_FIELD_NUMBER;
-      hash = (53 * hash) + getNodeList().hashCode();
-    }
-    if (getCostCount() > 0) {
-      hash = (37 * hash) + COST_FIELD_NUMBER;
-      hash = (53 * hash) + getCostList().hashCode();
-    }
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.framework.CostGraphDef parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.CostGraphDef parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.CostGraphDef parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.CostGraphDef parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.CostGraphDef parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.CostGraphDef parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.CostGraphDef parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.CostGraphDef parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.CostGraphDef parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.CostGraphDef parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.CostGraphDef parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.CostGraphDef parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.framework.CostGraphDef prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * Protobuf type {@code tensorflow.CostGraphDef}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.CostGraphDef)
-      org.tensorflow.proto.framework.CostGraphDefOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.CostGraphProtos.internal_static_tensorflow_CostGraphDef_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.CostGraphProtos.internal_static_tensorflow_CostGraphDef_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.CostGraphDef.class, org.tensorflow.proto.framework.CostGraphDef.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.framework.CostGraphDef.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-        getNodeFieldBuilder();
-        getCostFieldBuilder();
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      if (nodeBuilder_ == null) {
-        node_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000001);
-      } else {
-        nodeBuilder_.clear();
-      }
-      if (costBuilder_ == null) {
-        cost_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000002);
-      } else {
-        costBuilder_.clear();
-      }
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.framework.CostGraphProtos.internal_static_tensorflow_CostGraphDef_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.CostGraphDef getDefaultInstanceForType() {
-      return org.tensorflow.proto.framework.CostGraphDef.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.CostGraphDef build() {
-      org.tensorflow.proto.framework.CostGraphDef result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.CostGraphDef buildPartial() {
-      org.tensorflow.proto.framework.CostGraphDef result = new org.tensorflow.proto.framework.CostGraphDef(this);
-      int from_bitField0_ = bitField0_;
-      if (nodeBuilder_ == null) {
-        if (((bitField0_ & 0x00000001) != 0)) {
-          node_ = java.util.Collections.unmodifiableList(node_);
-          bitField0_ = (bitField0_ & ~0x00000001);
-        }
-        result.node_ = node_;
-      } else {
-        result.node_ = nodeBuilder_.build();
-      }
-      if (costBuilder_ == null) {
-        if (((bitField0_ & 0x00000002) != 0)) {
-          cost_ = java.util.Collections.unmodifiableList(cost_);
-          bitField0_ = (bitField0_ & ~0x00000002);
-        }
-        result.cost_ = cost_;
-      } else {
-        result.cost_ = costBuilder_.build();
-      }
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.framework.CostGraphDef) {
-        return mergeFrom((org.tensorflow.proto.framework.CostGraphDef)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.framework.CostGraphDef other) {
-      if (other == org.tensorflow.proto.framework.CostGraphDef.getDefaultInstance()) return this;
-      if (nodeBuilder_ == null) {
-        if (!other.node_.isEmpty()) {
-          if (node_.isEmpty()) {
-            node_ = other.node_;
-            bitField0_ = (bitField0_ & ~0x00000001);
-          } else {
-            ensureNodeIsMutable();
-            node_.addAll(other.node_);
-          }
-          onChanged();
-        }
-      } else {
-        if (!other.node_.isEmpty()) {
-          if (nodeBuilder_.isEmpty()) {
-            nodeBuilder_.dispose();
-            nodeBuilder_ = null;
-            node_ = other.node_;
-            bitField0_ = (bitField0_ & ~0x00000001);
-            nodeBuilder_ = 
-              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
-                 getNodeFieldBuilder() : null;
-          } else {
-            nodeBuilder_.addAllMessages(other.node_);
-          }
-        }
-      }
-      if (costBuilder_ == null) {
-        if (!other.cost_.isEmpty()) {
-          if (cost_.isEmpty()) {
-            cost_ = other.cost_;
-            bitField0_ = (bitField0_ & ~0x00000002);
-          } else {
-            ensureCostIsMutable();
-            cost_.addAll(other.cost_);
-          }
-          onChanged();
-        }
-      } else {
-        if (!other.cost_.isEmpty()) {
-          if (costBuilder_.isEmpty()) {
-            costBuilder_.dispose();
-            costBuilder_ = null;
-            cost_ = other.cost_;
-            bitField0_ = (bitField0_ & ~0x00000002);
-            costBuilder_ = 
-              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
-                 getCostFieldBuilder() : null;
-          } else {
-            costBuilder_.addAllMessages(other.cost_);
-          }
-        }
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.framework.CostGraphDef parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.framework.CostGraphDef) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-    private int bitField0_;
-
-    private java.util.List<org.tensorflow.proto.framework.CostGraphDef.Node> node_ =
-      java.util.Collections.emptyList();
-    private void ensureNodeIsMutable() {
-      if (!((bitField0_ & 0x00000001) != 0)) {
-        node_ = new java.util.ArrayList<org.tensorflow.proto.framework.CostGraphDef.Node>(node_);
-        bitField0_ |= 0x00000001;
-       }
-    }
-
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.framework.CostGraphDef.Node, org.tensorflow.proto.framework.CostGraphDef.Node.Builder, org.tensorflow.proto.framework.CostGraphDef.NodeOrBuilder> nodeBuilder_;
-
-    /**
-     * <code>repeated .tensorflow.CostGraphDef.Node node = 1;</code>
-     */
-    public java.util.List<org.tensorflow.proto.framework.CostGraphDef.Node> getNodeList() {
-      if (nodeBuilder_ == null) {
-        return java.util.Collections.unmodifiableList(node_);
-      } else {
-        return nodeBuilder_.getMessageList();
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.CostGraphDef.Node node = 1;</code>
-     */
-    public int getNodeCount() {
-      if (nodeBuilder_ == null) {
-        return node_.size();
-      } else {
-        return nodeBuilder_.getCount();
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.CostGraphDef.Node node = 1;</code>
-     */
-    public org.tensorflow.proto.framework.CostGraphDef.Node getNode(int index) {
-      if (nodeBuilder_ == null) {
-        return node_.get(index);
-      } else {
-        return nodeBuilder_.getMessage(index);
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.CostGraphDef.Node node = 1;</code>
-     */
-    public Builder setNode(
-        int index, org.tensorflow.proto.framework.CostGraphDef.Node value) {
-      if (nodeBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureNodeIsMutable();
-        node_.set(index, value);
-        onChanged();
-      } else {
-        nodeBuilder_.setMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.CostGraphDef.Node node = 1;</code>
-     */
-    public Builder setNode(
-        int index, org.tensorflow.proto.framework.CostGraphDef.Node.Builder builderForValue) {
-      if (nodeBuilder_ == null) {
-        ensureNodeIsMutable();
-        node_.set(index, builderForValue.build());
-        onChanged();
-      } else {
-        nodeBuilder_.setMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.CostGraphDef.Node node = 1;</code>
-     */
-    public Builder addNode(org.tensorflow.proto.framework.CostGraphDef.Node value) {
-      if (nodeBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureNodeIsMutable();
-        node_.add(value);
-        onChanged();
-      } else {
-        nodeBuilder_.addMessage(value);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.CostGraphDef.Node node = 1;</code>
-     */
-    public Builder addNode(
-        int index, org.tensorflow.proto.framework.CostGraphDef.Node value) {
-      if (nodeBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureNodeIsMutable();
-        node_.add(index, value);
-        onChanged();
-      } else {
-        nodeBuilder_.addMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.CostGraphDef.Node node = 1;</code>
-     */
-    public Builder addNode(
-        org.tensorflow.proto.framework.CostGraphDef.Node.Builder builderForValue) {
-      if (nodeBuilder_ == null) {
-        ensureNodeIsMutable();
-        node_.add(builderForValue.build());
-        onChanged();
-      } else {
-        nodeBuilder_.addMessage(builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.CostGraphDef.Node node = 1;</code>
-     */
-    public Builder addNode(
-        int index, org.tensorflow.proto.framework.CostGraphDef.Node.Builder builderForValue) {
-      if (nodeBuilder_ == null) {
-        ensureNodeIsMutable();
-        node_.add(index, builderForValue.build());
-        onChanged();
-      } else {
-        nodeBuilder_.addMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.CostGraphDef.Node node = 1;</code>
-     */
-    public Builder addAllNode(
-        java.lang.Iterable<? extends org.tensorflow.proto.framework.CostGraphDef.Node> values) {
-      if (nodeBuilder_ == null) {
-        ensureNodeIsMutable();
-        com.google.protobuf.AbstractMessageLite.Builder.addAll(
-            values, node_);
-        onChanged();
-      } else {
-        nodeBuilder_.addAllMessages(values);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.CostGraphDef.Node node = 1;</code>
-     */
-    public Builder clearNode() {
-      if (nodeBuilder_ == null) {
-        node_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000001);
-        onChanged();
-      } else {
-        nodeBuilder_.clear();
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.CostGraphDef.Node node = 1;</code>
-     */
-    public Builder removeNode(int index) {
-      if (nodeBuilder_ == null) {
-        ensureNodeIsMutable();
-        node_.remove(index);
-        onChanged();
-      } else {
-        nodeBuilder_.remove(index);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.CostGraphDef.Node node = 1;</code>
-     */
-    public org.tensorflow.proto.framework.CostGraphDef.Node.Builder getNodeBuilder(
-        int index) {
-      return getNodeFieldBuilder().getBuilder(index);
-    }
-    /**
-     * <code>repeated .tensorflow.CostGraphDef.Node node = 1;</code>
-     */
-    public org.tensorflow.proto.framework.CostGraphDef.NodeOrBuilder getNodeOrBuilder(
-        int index) {
-      if (nodeBuilder_ == null) {
-        return node_.get(index);  } else {
-        return nodeBuilder_.getMessageOrBuilder(index);
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.CostGraphDef.Node node = 1;</code>
-     */
-    public java.util.List<? extends org.tensorflow.proto.framework.CostGraphDef.NodeOrBuilder> 
-         getNodeOrBuilderList() {
-      if (nodeBuilder_ != null) {
-        return nodeBuilder_.getMessageOrBuilderList();
-      } else {
-        return java.util.Collections.unmodifiableList(node_);
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.CostGraphDef.Node node = 1;</code>
-     */
-    public org.tensorflow.proto.framework.CostGraphDef.Node.Builder addNodeBuilder() {
-      return getNodeFieldBuilder().addBuilder(
-          org.tensorflow.proto.framework.CostGraphDef.Node.getDefaultInstance());
-    }
-    /**
-     * <code>repeated .tensorflow.CostGraphDef.Node node = 1;</code>
-     */
-    public org.tensorflow.proto.framework.CostGraphDef.Node.Builder addNodeBuilder(
-        int index) {
-      return getNodeFieldBuilder().addBuilder(
-          index, org.tensorflow.proto.framework.CostGraphDef.Node.getDefaultInstance());
-    }
-    /**
-     * <code>repeated .tensorflow.CostGraphDef.Node node = 1;</code>
-     */
-    public java.util.List<org.tensorflow.proto.framework.CostGraphDef.Node.Builder> 
-         getNodeBuilderList() {
-      return getNodeFieldBuilder().getBuilderList();
-    }
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.framework.CostGraphDef.Node, org.tensorflow.proto.framework.CostGraphDef.Node.Builder, org.tensorflow.proto.framework.CostGraphDef.NodeOrBuilder> 
-        getNodeFieldBuilder() {
-      if (nodeBuilder_ == null) {
-        nodeBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
-            org.tensorflow.proto.framework.CostGraphDef.Node, org.tensorflow.proto.framework.CostGraphDef.Node.Builder, org.tensorflow.proto.framework.CostGraphDef.NodeOrBuilder>(
-                node_,
-                ((bitField0_ & 0x00000001) != 0),
-                getParentForChildren(),
-                isClean());
-        node_ = null;
-      }
-      return nodeBuilder_;
-    }
-
-    private java.util.List<org.tensorflow.proto.framework.CostGraphDef.AggregatedCost> cost_ =
-      java.util.Collections.emptyList();
-    private void ensureCostIsMutable() {
-      if (!((bitField0_ & 0x00000002) != 0)) {
-        cost_ = new java.util.ArrayList<org.tensorflow.proto.framework.CostGraphDef.AggregatedCost>(cost_);
-        bitField0_ |= 0x00000002;
-       }
-    }
-
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.framework.CostGraphDef.AggregatedCost, org.tensorflow.proto.framework.CostGraphDef.AggregatedCost.Builder, org.tensorflow.proto.framework.CostGraphDef.AggregatedCostOrBuilder> costBuilder_;
-
-    /**
-     * <code>repeated .tensorflow.CostGraphDef.AggregatedCost cost = 2;</code>
-     */
-    public java.util.List<org.tensorflow.proto.framework.CostGraphDef.AggregatedCost> getCostList() {
-      if (costBuilder_ == null) {
-        return java.util.Collections.unmodifiableList(cost_);
-      } else {
-        return costBuilder_.getMessageList();
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.CostGraphDef.AggregatedCost cost = 2;</code>
-     */
-    public int getCostCount() {
-      if (costBuilder_ == null) {
-        return cost_.size();
-      } else {
-        return costBuilder_.getCount();
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.CostGraphDef.AggregatedCost cost = 2;</code>
-     */
-    public org.tensorflow.proto.framework.CostGraphDef.AggregatedCost getCost(int index) {
-      if (costBuilder_ == null) {
-        return cost_.get(index);
-      } else {
-        return costBuilder_.getMessage(index);
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.CostGraphDef.AggregatedCost cost = 2;</code>
-     */
-    public Builder setCost(
-        int index, org.tensorflow.proto.framework.CostGraphDef.AggregatedCost value) {
-      if (costBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureCostIsMutable();
-        cost_.set(index, value);
-        onChanged();
-      } else {
-        costBuilder_.setMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.CostGraphDef.AggregatedCost cost = 2;</code>
-     */
-    public Builder setCost(
-        int index, org.tensorflow.proto.framework.CostGraphDef.AggregatedCost.Builder builderForValue) {
-      if (costBuilder_ == null) {
-        ensureCostIsMutable();
-        cost_.set(index, builderForValue.build());
-        onChanged();
-      } else {
-        costBuilder_.setMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.CostGraphDef.AggregatedCost cost = 2;</code>
-     */
-    public Builder addCost(org.tensorflow.proto.framework.CostGraphDef.AggregatedCost value) {
-      if (costBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureCostIsMutable();
-        cost_.add(value);
-        onChanged();
-      } else {
-        costBuilder_.addMessage(value);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.CostGraphDef.AggregatedCost cost = 2;</code>
-     */
-    public Builder addCost(
-        int index, org.tensorflow.proto.framework.CostGraphDef.AggregatedCost value) {
-      if (costBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureCostIsMutable();
-        cost_.add(index, value);
-        onChanged();
-      } else {
-        costBuilder_.addMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.CostGraphDef.AggregatedCost cost = 2;</code>
-     */
-    public Builder addCost(
-        org.tensorflow.proto.framework.CostGraphDef.AggregatedCost.Builder builderForValue) {
-      if (costBuilder_ == null) {
-        ensureCostIsMutable();
-        cost_.add(builderForValue.build());
-        onChanged();
-      } else {
-        costBuilder_.addMessage(builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.CostGraphDef.AggregatedCost cost = 2;</code>
-     */
-    public Builder addCost(
-        int index, org.tensorflow.proto.framework.CostGraphDef.AggregatedCost.Builder builderForValue) {
-      if (costBuilder_ == null) {
-        ensureCostIsMutable();
-        cost_.add(index, builderForValue.build());
-        onChanged();
-      } else {
-        costBuilder_.addMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.CostGraphDef.AggregatedCost cost = 2;</code>
-     */
-    public Builder addAllCost(
-        java.lang.Iterable<? extends org.tensorflow.proto.framework.CostGraphDef.AggregatedCost> values) {
-      if (costBuilder_ == null) {
-        ensureCostIsMutable();
-        com.google.protobuf.AbstractMessageLite.Builder.addAll(
-            values, cost_);
-        onChanged();
-      } else {
-        costBuilder_.addAllMessages(values);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.CostGraphDef.AggregatedCost cost = 2;</code>
-     */
-    public Builder clearCost() {
-      if (costBuilder_ == null) {
-        cost_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000002);
-        onChanged();
-      } else {
-        costBuilder_.clear();
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.CostGraphDef.AggregatedCost cost = 2;</code>
-     */
-    public Builder removeCost(int index) {
-      if (costBuilder_ == null) {
-        ensureCostIsMutable();
-        cost_.remove(index);
-        onChanged();
-      } else {
-        costBuilder_.remove(index);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.CostGraphDef.AggregatedCost cost = 2;</code>
-     */
-    public org.tensorflow.proto.framework.CostGraphDef.AggregatedCost.Builder getCostBuilder(
-        int index) {
-      return getCostFieldBuilder().getBuilder(index);
-    }
-    /**
-     * <code>repeated .tensorflow.CostGraphDef.AggregatedCost cost = 2;</code>
-     */
-    public org.tensorflow.proto.framework.CostGraphDef.AggregatedCostOrBuilder getCostOrBuilder(
-        int index) {
-      if (costBuilder_ == null) {
-        return cost_.get(index);  } else {
-        return costBuilder_.getMessageOrBuilder(index);
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.CostGraphDef.AggregatedCost cost = 2;</code>
-     */
-    public java.util.List<? extends org.tensorflow.proto.framework.CostGraphDef.AggregatedCostOrBuilder> 
-         getCostOrBuilderList() {
-      if (costBuilder_ != null) {
-        return costBuilder_.getMessageOrBuilderList();
-      } else {
-        return java.util.Collections.unmodifiableList(cost_);
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.CostGraphDef.AggregatedCost cost = 2;</code>
-     */
-    public org.tensorflow.proto.framework.CostGraphDef.AggregatedCost.Builder addCostBuilder() {
-      return getCostFieldBuilder().addBuilder(
-          org.tensorflow.proto.framework.CostGraphDef.AggregatedCost.getDefaultInstance());
-    }
-    /**
-     * <code>repeated .tensorflow.CostGraphDef.AggregatedCost cost = 2;</code>
-     */
-    public org.tensorflow.proto.framework.CostGraphDef.AggregatedCost.Builder addCostBuilder(
-        int index) {
-      return getCostFieldBuilder().addBuilder(
-          index, org.tensorflow.proto.framework.CostGraphDef.AggregatedCost.getDefaultInstance());
-    }
-    /**
-     * <code>repeated .tensorflow.CostGraphDef.AggregatedCost cost = 2;</code>
-     */
-    public java.util.List<org.tensorflow.proto.framework.CostGraphDef.AggregatedCost.Builder> 
-         getCostBuilderList() {
-      return getCostFieldBuilder().getBuilderList();
-    }
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.framework.CostGraphDef.AggregatedCost, org.tensorflow.proto.framework.CostGraphDef.AggregatedCost.Builder, org.tensorflow.proto.framework.CostGraphDef.AggregatedCostOrBuilder> 
-        getCostFieldBuilder() {
-      if (costBuilder_ == null) {
-        costBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
-            org.tensorflow.proto.framework.CostGraphDef.AggregatedCost, org.tensorflow.proto.framework.CostGraphDef.AggregatedCost.Builder, org.tensorflow.proto.framework.CostGraphDef.AggregatedCostOrBuilder>(
-                cost_,
-                ((bitField0_ & 0x00000002) != 0),
-                getParentForChildren(),
-                isClean());
-        cost_ = null;
-      }
-      return costBuilder_;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.CostGraphDef)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.CostGraphDef)
-  private static final org.tensorflow.proto.framework.CostGraphDef DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.CostGraphDef();
-  }
-
-  public static org.tensorflow.proto.framework.CostGraphDef getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<CostGraphDef>
-      PARSER = new com.google.protobuf.AbstractParser<CostGraphDef>() {
-    @java.lang.Override
-    public CostGraphDef parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new CostGraphDef(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<CostGraphDef> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<CostGraphDef> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.framework.CostGraphDef getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/CostGraphDefOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/CostGraphDefOrBuilder.java
deleted file mode 100644
index 7d8046a5e27..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/CostGraphDefOrBuilder.java
+++ /dev/null
@@ -1,57 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/framework/cost_graph.proto
-
-package org.tensorflow.proto.framework;
-
-public interface CostGraphDefOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.CostGraphDef)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <code>repeated .tensorflow.CostGraphDef.Node node = 1;</code>
-   */
-  java.util.List<org.tensorflow.proto.framework.CostGraphDef.Node> 
-      getNodeList();
-  /**
-   * <code>repeated .tensorflow.CostGraphDef.Node node = 1;</code>
-   */
-  org.tensorflow.proto.framework.CostGraphDef.Node getNode(int index);
-  /**
-   * <code>repeated .tensorflow.CostGraphDef.Node node = 1;</code>
-   */
-  int getNodeCount();
-  /**
-   * <code>repeated .tensorflow.CostGraphDef.Node node = 1;</code>
-   */
-  java.util.List<? extends org.tensorflow.proto.framework.CostGraphDef.NodeOrBuilder> 
-      getNodeOrBuilderList();
-  /**
-   * <code>repeated .tensorflow.CostGraphDef.Node node = 1;</code>
-   */
-  org.tensorflow.proto.framework.CostGraphDef.NodeOrBuilder getNodeOrBuilder(
-      int index);
-
-  /**
-   * <code>repeated .tensorflow.CostGraphDef.AggregatedCost cost = 2;</code>
-   */
-  java.util.List<org.tensorflow.proto.framework.CostGraphDef.AggregatedCost> 
-      getCostList();
-  /**
-   * <code>repeated .tensorflow.CostGraphDef.AggregatedCost cost = 2;</code>
-   */
-  org.tensorflow.proto.framework.CostGraphDef.AggregatedCost getCost(int index);
-  /**
-   * <code>repeated .tensorflow.CostGraphDef.AggregatedCost cost = 2;</code>
-   */
-  int getCostCount();
-  /**
-   * <code>repeated .tensorflow.CostGraphDef.AggregatedCost cost = 2;</code>
-   */
-  java.util.List<? extends org.tensorflow.proto.framework.CostGraphDef.AggregatedCostOrBuilder> 
-      getCostOrBuilderList();
-  /**
-   * <code>repeated .tensorflow.CostGraphDef.AggregatedCost cost = 2;</code>
-   */
-  org.tensorflow.proto.framework.CostGraphDef.AggregatedCostOrBuilder getCostOrBuilder(
-      int index);
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/CostGraphProtos.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/CostGraphProtos.java
deleted file mode 100644
index b115a24c302..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/CostGraphProtos.java
+++ /dev/null
@@ -1,122 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/framework/cost_graph.proto
-
-package org.tensorflow.proto.framework;
-
-public final class CostGraphProtos {
-  private CostGraphProtos() {}
-  public static void registerAllExtensions(
-      com.google.protobuf.ExtensionRegistryLite registry) {
-  }
-
-  public static void registerAllExtensions(
-      com.google.protobuf.ExtensionRegistry registry) {
-    registerAllExtensions(
-        (com.google.protobuf.ExtensionRegistryLite) registry);
-  }
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_CostGraphDef_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_CostGraphDef_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_CostGraphDef_Node_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_CostGraphDef_Node_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_CostGraphDef_Node_InputInfo_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_CostGraphDef_Node_InputInfo_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_CostGraphDef_Node_OutputInfo_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_CostGraphDef_Node_OutputInfo_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_CostGraphDef_AggregatedCost_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_CostGraphDef_AggregatedCost_fieldAccessorTable;
-
-  public static com.google.protobuf.Descriptors.FileDescriptor
-      getDescriptor() {
-    return descriptor;
-  }
-  private static  com.google.protobuf.Descriptors.FileDescriptor
-      descriptor;
-  static {
-    java.lang.String[] descriptorData = {
-      "\n*tensorflow/core/framework/cost_graph.p" +
-      "roto\022\ntensorflow\032,tensorflow/core/framew" +
-      "ork/tensor_shape.proto\032%tensorflow/core/" +
-      "framework/types.proto\"\312\006\n\014CostGraphDef\022+" +
-      "\n\004node\030\001 \003(\0132\035.tensorflow.CostGraphDef.N" +
-      "ode\0225\n\004cost\030\002 \003(\0132\'.tensorflow.CostGraph" +
-      "Def.AggregatedCost\032\242\005\n\004Node\022\014\n\004name\030\001 \001(" +
-      "\t\022\016\n\006device\030\002 \001(\t\022\n\n\002id\030\003 \001(\005\022;\n\ninput_i" +
-      "nfo\030\004 \003(\0132\'.tensorflow.CostGraphDef.Node" +
-      ".InputInfo\022=\n\013output_info\030\005 \003(\0132(.tensor" +
-      "flow.CostGraphDef.Node.OutputInfo\022\035\n\025tem" +
-      "porary_memory_size\030\006 \001(\003\022\036\n\026persistent_m" +
-      "emory_size\030\014 \001(\003\022!\n\025host_temp_memory_siz" +
-      "e\030\n \001(\003B\002\030\001\022#\n\027device_temp_memory_size\030\013" +
-      " \001(\003B\002\030\001\022)\n\035device_persistent_memory_siz" +
-      "e\030\020 \001(\003B\002\030\001\022\024\n\014compute_cost\030\t \001(\003\022\024\n\014com" +
-      "pute_time\030\016 \001(\003\022\023\n\013memory_time\030\017 \001(\003\022\020\n\010" +
-      "is_final\030\007 \001(\010\022\025\n\rcontrol_input\030\010 \003(\005\022\022\n" +
-      "\ninaccurate\030\021 \001(\010\032;\n\tInputInfo\022\026\n\016preced" +
-      "ing_node\030\001 \001(\005\022\026\n\016preceding_port\030\002 \001(\005\032\206" +
-      "\001\n\nOutputInfo\022\014\n\004size\030\001 \001(\003\022\030\n\020alias_inp" +
-      "ut_port\030\002 \001(\003\022+\n\005shape\030\003 \001(\0132\034.tensorflo" +
-      "w.TensorShapeProto\022#\n\005dtype\030\004 \001(\0162\024.tens" +
-      "orflow.DataType\0321\n\016AggregatedCost\022\014\n\004cos" +
-      "t\030\001 \001(\002\022\021\n\tdimension\030\002 \001(\tB\211\001\n\036org.tenso" +
-      "rflow.proto.frameworkB\017CostGraphProtosP\001" +
-      "ZQgithub.com/tensorflow/tensorflow/tenso" +
-      "rflow/go/core/framework/cost_graph_go_pr" +
-      "oto\370\001\001b\006proto3"
-    };
-    descriptor = com.google.protobuf.Descriptors.FileDescriptor
-      .internalBuildGeneratedFileFrom(descriptorData,
-        new com.google.protobuf.Descriptors.FileDescriptor[] {
-          org.tensorflow.proto.framework.TensorShapeProtos.getDescriptor(),
-          org.tensorflow.proto.framework.TypesProtos.getDescriptor(),
-        });
-    internal_static_tensorflow_CostGraphDef_descriptor =
-      getDescriptor().getMessageTypes().get(0);
-    internal_static_tensorflow_CostGraphDef_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_CostGraphDef_descriptor,
-        new java.lang.String[] { "Node", "Cost", });
-    internal_static_tensorflow_CostGraphDef_Node_descriptor =
-      internal_static_tensorflow_CostGraphDef_descriptor.getNestedTypes().get(0);
-    internal_static_tensorflow_CostGraphDef_Node_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_CostGraphDef_Node_descriptor,
-        new java.lang.String[] { "Name", "Device", "Id", "InputInfo", "OutputInfo", "TemporaryMemorySize", "PersistentMemorySize", "HostTempMemorySize", "DeviceTempMemorySize", "DevicePersistentMemorySize", "ComputeCost", "ComputeTime", "MemoryTime", "IsFinal", "ControlInput", "Inaccurate", });
-    internal_static_tensorflow_CostGraphDef_Node_InputInfo_descriptor =
-      internal_static_tensorflow_CostGraphDef_Node_descriptor.getNestedTypes().get(0);
-    internal_static_tensorflow_CostGraphDef_Node_InputInfo_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_CostGraphDef_Node_InputInfo_descriptor,
-        new java.lang.String[] { "PrecedingNode", "PrecedingPort", });
-    internal_static_tensorflow_CostGraphDef_Node_OutputInfo_descriptor =
-      internal_static_tensorflow_CostGraphDef_Node_descriptor.getNestedTypes().get(1);
-    internal_static_tensorflow_CostGraphDef_Node_OutputInfo_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_CostGraphDef_Node_OutputInfo_descriptor,
-        new java.lang.String[] { "Size", "AliasInputPort", "Shape", "Dtype", });
-    internal_static_tensorflow_CostGraphDef_AggregatedCost_descriptor =
-      internal_static_tensorflow_CostGraphDef_descriptor.getNestedTypes().get(1);
-    internal_static_tensorflow_CostGraphDef_AggregatedCost_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_CostGraphDef_AggregatedCost_descriptor,
-        new java.lang.String[] { "Cost", "Dimension", });
-    org.tensorflow.proto.framework.TensorShapeProtos.getDescriptor();
-    org.tensorflow.proto.framework.TypesProtos.getDescriptor();
-  }
-
-  // @@protoc_insertion_point(outer_class_scope)
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DataClass.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DataClass.java
deleted file mode 100644
index 895366e817c..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DataClass.java
+++ /dev/null
@@ -1,167 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/framework/summary.proto
-
-package org.tensorflow.proto.framework;
-
-/**
- * Protobuf enum {@code tensorflow.DataClass}
- */
-public enum DataClass
-    implements com.google.protobuf.ProtocolMessageEnum {
-  /**
-   * <pre>
-   * Unknown data class, used (implicitly) for legacy data. Will not be
-   * processed by data ingestion pipelines.
-   * </pre>
-   *
-   * <code>DATA_CLASS_UNKNOWN = 0;</code>
-   */
-  DATA_CLASS_UNKNOWN(0),
-  /**
-   * <pre>
-   * Scalar time series. Each `Value` for the corresponding tag must have
-   * `tensor` set to a rank-0 tensor of floating-point dtype, which will be
-   * converted to float64.
-   * </pre>
-   *
-   * <code>DATA_CLASS_SCALAR = 1;</code>
-   */
-  DATA_CLASS_SCALAR(1),
-  /**
-   * <pre>
-   * Tensor time series. Each `Value` for the corresponding tag must have
-   * `tensor` set. The tensor value is arbitrary, but should be small to
-   * accommodate direct storage in database backends: an upper bound of a few
-   * kilobytes is a reasonable rule of thumb.
-   * </pre>
-   *
-   * <code>DATA_CLASS_TENSOR = 2;</code>
-   */
-  DATA_CLASS_TENSOR(2),
-  /**
-   * <pre>
-   * Blob sequence time series. Each `Value` for the corresponding tag must
-   * have `tensor` set to a rank-1 tensor of bytestring dtype.
-   * </pre>
-   *
-   * <code>DATA_CLASS_BLOB_SEQUENCE = 3;</code>
-   */
-  DATA_CLASS_BLOB_SEQUENCE(3),
-  UNRECOGNIZED(-1),
-  ;
-
-  /**
-   * <pre>
-   * Unknown data class, used (implicitly) for legacy data. Will not be
-   * processed by data ingestion pipelines.
-   * </pre>
-   *
-   * <code>DATA_CLASS_UNKNOWN = 0;</code>
-   */
-  public static final int DATA_CLASS_UNKNOWN_VALUE = 0;
-  /**
-   * <pre>
-   * Scalar time series. Each `Value` for the corresponding tag must have
-   * `tensor` set to a rank-0 tensor of floating-point dtype, which will be
-   * converted to float64.
-   * </pre>
-   *
-   * <code>DATA_CLASS_SCALAR = 1;</code>
-   */
-  public static final int DATA_CLASS_SCALAR_VALUE = 1;
-  /**
-   * <pre>
-   * Tensor time series. Each `Value` for the corresponding tag must have
-   * `tensor` set. The tensor value is arbitrary, but should be small to
-   * accommodate direct storage in database backends: an upper bound of a few
-   * kilobytes is a reasonable rule of thumb.
-   * </pre>
-   *
-   * <code>DATA_CLASS_TENSOR = 2;</code>
-   */
-  public static final int DATA_CLASS_TENSOR_VALUE = 2;
-  /**
-   * <pre>
-   * Blob sequence time series. Each `Value` for the corresponding tag must
-   * have `tensor` set to a rank-1 tensor of bytestring dtype.
-   * </pre>
-   *
-   * <code>DATA_CLASS_BLOB_SEQUENCE = 3;</code>
-   */
-  public static final int DATA_CLASS_BLOB_SEQUENCE_VALUE = 3;
-
-
-  public final int getNumber() {
-    if (this == UNRECOGNIZED) {
-      throw new java.lang.IllegalArgumentException(
-          "Can't get the number of an unknown enum value.");
-    }
-    return value;
-  }
-
-  /**
-   * @deprecated Use {@link #forNumber(int)} instead.
-   */
-  @java.lang.Deprecated
-  public static DataClass valueOf(int value) {
-    return forNumber(value);
-  }
-
-  public static DataClass forNumber(int value) {
-    switch (value) {
-      case 0: return DATA_CLASS_UNKNOWN;
-      case 1: return DATA_CLASS_SCALAR;
-      case 2: return DATA_CLASS_TENSOR;
-      case 3: return DATA_CLASS_BLOB_SEQUENCE;
-      default: return null;
-    }
-  }
-
-  public static com.google.protobuf.Internal.EnumLiteMap<DataClass>
-      internalGetValueMap() {
-    return internalValueMap;
-  }
-  private static final com.google.protobuf.Internal.EnumLiteMap<
-      DataClass> internalValueMap =
-        new com.google.protobuf.Internal.EnumLiteMap<DataClass>() {
-          public DataClass findValueByNumber(int number) {
-            return DataClass.forNumber(number);
-          }
-        };
-
-  public final com.google.protobuf.Descriptors.EnumValueDescriptor
-      getValueDescriptor() {
-    return getDescriptor().getValues().get(ordinal());
-  }
-  public final com.google.protobuf.Descriptors.EnumDescriptor
-      getDescriptorForType() {
-    return getDescriptor();
-  }
-  public static final com.google.protobuf.Descriptors.EnumDescriptor
-      getDescriptor() {
-    return org.tensorflow.proto.framework.SummaryProtos.getDescriptor().getEnumTypes().get(0);
-  }
-
-  private static final DataClass[] VALUES = values();
-
-  public static DataClass valueOf(
-      com.google.protobuf.Descriptors.EnumValueDescriptor desc) {
-    if (desc.getType() != getDescriptor()) {
-      throw new java.lang.IllegalArgumentException(
-        "EnumValueDescriptor is not for this type.");
-    }
-    if (desc.getIndex() == -1) {
-      return UNRECOGNIZED;
-    }
-    return VALUES[desc.getIndex()];
-  }
-
-  private final int value;
-
-  private DataClass(int value) {
-    this.value = value;
-  }
-
-  // @@protoc_insertion_point(enum_scope:tensorflow.DataClass)
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DataType.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DataType.java
deleted file mode 100644
index 6b836c8353a..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DataType.java
+++ /dev/null
@@ -1,615 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/framework/types.proto
-
-package org.tensorflow.proto.framework;
-
-/**
- * <pre>
- * (== suppress_warning documentation-presence ==)
- * LINT.IfChange
- * </pre>
- *
- * Protobuf enum {@code tensorflow.DataType}
- */
-public enum DataType
-    implements com.google.protobuf.ProtocolMessageEnum {
-  /**
-   * <pre>
-   * Not a legal value for DataType.  Used to indicate a DataType field
-   * has not been set.
-   * </pre>
-   *
-   * <code>DT_INVALID = 0;</code>
-   */
-  DT_INVALID(0),
-  /**
-   * <pre>
-   * Data types that all computation devices are expected to be
-   * capable to support.
-   * </pre>
-   *
-   * <code>DT_FLOAT = 1;</code>
-   */
-  DT_FLOAT(1),
-  /**
-   * <code>DT_DOUBLE = 2;</code>
-   */
-  DT_DOUBLE(2),
-  /**
-   * <code>DT_INT32 = 3;</code>
-   */
-  DT_INT32(3),
-  /**
-   * <code>DT_UINT8 = 4;</code>
-   */
-  DT_UINT8(4),
-  /**
-   * <code>DT_INT16 = 5;</code>
-   */
-  DT_INT16(5),
-  /**
-   * <code>DT_INT8 = 6;</code>
-   */
-  DT_INT8(6),
-  /**
-   * <code>DT_STRING = 7;</code>
-   */
-  DT_STRING(7),
-  /**
-   * <pre>
-   * Single-precision complex
-   * </pre>
-   *
-   * <code>DT_COMPLEX64 = 8;</code>
-   */
-  DT_COMPLEX64(8),
-  /**
-   * <code>DT_INT64 = 9;</code>
-   */
-  DT_INT64(9),
-  /**
-   * <code>DT_BOOL = 10;</code>
-   */
-  DT_BOOL(10),
-  /**
-   * <pre>
-   * Quantized int8
-   * </pre>
-   *
-   * <code>DT_QINT8 = 11;</code>
-   */
-  DT_QINT8(11),
-  /**
-   * <pre>
-   * Quantized uint8
-   * </pre>
-   *
-   * <code>DT_QUINT8 = 12;</code>
-   */
-  DT_QUINT8(12),
-  /**
-   * <pre>
-   * Quantized int32
-   * </pre>
-   *
-   * <code>DT_QINT32 = 13;</code>
-   */
-  DT_QINT32(13),
-  /**
-   * <pre>
-   * Float32 truncated to 16 bits.  Only for cast ops.
-   * </pre>
-   *
-   * <code>DT_BFLOAT16 = 14;</code>
-   */
-  DT_BFLOAT16(14),
-  /**
-   * <pre>
-   * Quantized int16
-   * </pre>
-   *
-   * <code>DT_QINT16 = 15;</code>
-   */
-  DT_QINT16(15),
-  /**
-   * <pre>
-   * Quantized uint16
-   * </pre>
-   *
-   * <code>DT_QUINT16 = 16;</code>
-   */
-  DT_QUINT16(16),
-  /**
-   * <code>DT_UINT16 = 17;</code>
-   */
-  DT_UINT16(17),
-  /**
-   * <pre>
-   * Double-precision complex
-   * </pre>
-   *
-   * <code>DT_COMPLEX128 = 18;</code>
-   */
-  DT_COMPLEX128(18),
-  /**
-   * <code>DT_HALF = 19;</code>
-   */
-  DT_HALF(19),
-  /**
-   * <code>DT_RESOURCE = 20;</code>
-   */
-  DT_RESOURCE(20),
-  /**
-   * <pre>
-   * Arbitrary C++ data types
-   * </pre>
-   *
-   * <code>DT_VARIANT = 21;</code>
-   */
-  DT_VARIANT(21),
-  /**
-   * <code>DT_UINT32 = 22;</code>
-   */
-  DT_UINT32(22),
-  /**
-   * <code>DT_UINT64 = 23;</code>
-   */
-  DT_UINT64(23),
-  /**
-   * <pre>
-   * Do not use!  These are only for parameters.  Every enum above
-   * should have a corresponding value below (verified by types_test).
-   * </pre>
-   *
-   * <code>DT_FLOAT_REF = 101;</code>
-   */
-  DT_FLOAT_REF(101),
-  /**
-   * <code>DT_DOUBLE_REF = 102;</code>
-   */
-  DT_DOUBLE_REF(102),
-  /**
-   * <code>DT_INT32_REF = 103;</code>
-   */
-  DT_INT32_REF(103),
-  /**
-   * <code>DT_UINT8_REF = 104;</code>
-   */
-  DT_UINT8_REF(104),
-  /**
-   * <code>DT_INT16_REF = 105;</code>
-   */
-  DT_INT16_REF(105),
-  /**
-   * <code>DT_INT8_REF = 106;</code>
-   */
-  DT_INT8_REF(106),
-  /**
-   * <code>DT_STRING_REF = 107;</code>
-   */
-  DT_STRING_REF(107),
-  /**
-   * <code>DT_COMPLEX64_REF = 108;</code>
-   */
-  DT_COMPLEX64_REF(108),
-  /**
-   * <code>DT_INT64_REF = 109;</code>
-   */
-  DT_INT64_REF(109),
-  /**
-   * <code>DT_BOOL_REF = 110;</code>
-   */
-  DT_BOOL_REF(110),
-  /**
-   * <code>DT_QINT8_REF = 111;</code>
-   */
-  DT_QINT8_REF(111),
-  /**
-   * <code>DT_QUINT8_REF = 112;</code>
-   */
-  DT_QUINT8_REF(112),
-  /**
-   * <code>DT_QINT32_REF = 113;</code>
-   */
-  DT_QINT32_REF(113),
-  /**
-   * <code>DT_BFLOAT16_REF = 114;</code>
-   */
-  DT_BFLOAT16_REF(114),
-  /**
-   * <code>DT_QINT16_REF = 115;</code>
-   */
-  DT_QINT16_REF(115),
-  /**
-   * <code>DT_QUINT16_REF = 116;</code>
-   */
-  DT_QUINT16_REF(116),
-  /**
-   * <code>DT_UINT16_REF = 117;</code>
-   */
-  DT_UINT16_REF(117),
-  /**
-   * <code>DT_COMPLEX128_REF = 118;</code>
-   */
-  DT_COMPLEX128_REF(118),
-  /**
-   * <code>DT_HALF_REF = 119;</code>
-   */
-  DT_HALF_REF(119),
-  /**
-   * <code>DT_RESOURCE_REF = 120;</code>
-   */
-  DT_RESOURCE_REF(120),
-  /**
-   * <code>DT_VARIANT_REF = 121;</code>
-   */
-  DT_VARIANT_REF(121),
-  /**
-   * <code>DT_UINT32_REF = 122;</code>
-   */
-  DT_UINT32_REF(122),
-  /**
-   * <code>DT_UINT64_REF = 123;</code>
-   */
-  DT_UINT64_REF(123),
-  UNRECOGNIZED(-1),
-  ;
-
-  /**
-   * <pre>
-   * Not a legal value for DataType.  Used to indicate a DataType field
-   * has not been set.
-   * </pre>
-   *
-   * <code>DT_INVALID = 0;</code>
-   */
-  public static final int DT_INVALID_VALUE = 0;
-  /**
-   * <pre>
-   * Data types that all computation devices are expected to be
-   * capable to support.
-   * </pre>
-   *
-   * <code>DT_FLOAT = 1;</code>
-   */
-  public static final int DT_FLOAT_VALUE = 1;
-  /**
-   * <code>DT_DOUBLE = 2;</code>
-   */
-  public static final int DT_DOUBLE_VALUE = 2;
-  /**
-   * <code>DT_INT32 = 3;</code>
-   */
-  public static final int DT_INT32_VALUE = 3;
-  /**
-   * <code>DT_UINT8 = 4;</code>
-   */
-  public static final int DT_UINT8_VALUE = 4;
-  /**
-   * <code>DT_INT16 = 5;</code>
-   */
-  public static final int DT_INT16_VALUE = 5;
-  /**
-   * <code>DT_INT8 = 6;</code>
-   */
-  public static final int DT_INT8_VALUE = 6;
-  /**
-   * <code>DT_STRING = 7;</code>
-   */
-  public static final int DT_STRING_VALUE = 7;
-  /**
-   * <pre>
-   * Single-precision complex
-   * </pre>
-   *
-   * <code>DT_COMPLEX64 = 8;</code>
-   */
-  public static final int DT_COMPLEX64_VALUE = 8;
-  /**
-   * <code>DT_INT64 = 9;</code>
-   */
-  public static final int DT_INT64_VALUE = 9;
-  /**
-   * <code>DT_BOOL = 10;</code>
-   */
-  public static final int DT_BOOL_VALUE = 10;
-  /**
-   * <pre>
-   * Quantized int8
-   * </pre>
-   *
-   * <code>DT_QINT8 = 11;</code>
-   */
-  public static final int DT_QINT8_VALUE = 11;
-  /**
-   * <pre>
-   * Quantized uint8
-   * </pre>
-   *
-   * <code>DT_QUINT8 = 12;</code>
-   */
-  public static final int DT_QUINT8_VALUE = 12;
-  /**
-   * <pre>
-   * Quantized int32
-   * </pre>
-   *
-   * <code>DT_QINT32 = 13;</code>
-   */
-  public static final int DT_QINT32_VALUE = 13;
-  /**
-   * <pre>
-   * Float32 truncated to 16 bits.  Only for cast ops.
-   * </pre>
-   *
-   * <code>DT_BFLOAT16 = 14;</code>
-   */
-  public static final int DT_BFLOAT16_VALUE = 14;
-  /**
-   * <pre>
-   * Quantized int16
-   * </pre>
-   *
-   * <code>DT_QINT16 = 15;</code>
-   */
-  public static final int DT_QINT16_VALUE = 15;
-  /**
-   * <pre>
-   * Quantized uint16
-   * </pre>
-   *
-   * <code>DT_QUINT16 = 16;</code>
-   */
-  public static final int DT_QUINT16_VALUE = 16;
-  /**
-   * <code>DT_UINT16 = 17;</code>
-   */
-  public static final int DT_UINT16_VALUE = 17;
-  /**
-   * <pre>
-   * Double-precision complex
-   * </pre>
-   *
-   * <code>DT_COMPLEX128 = 18;</code>
-   */
-  public static final int DT_COMPLEX128_VALUE = 18;
-  /**
-   * <code>DT_HALF = 19;</code>
-   */
-  public static final int DT_HALF_VALUE = 19;
-  /**
-   * <code>DT_RESOURCE = 20;</code>
-   */
-  public static final int DT_RESOURCE_VALUE = 20;
-  /**
-   * <pre>
-   * Arbitrary C++ data types
-   * </pre>
-   *
-   * <code>DT_VARIANT = 21;</code>
-   */
-  public static final int DT_VARIANT_VALUE = 21;
-  /**
-   * <code>DT_UINT32 = 22;</code>
-   */
-  public static final int DT_UINT32_VALUE = 22;
-  /**
-   * <code>DT_UINT64 = 23;</code>
-   */
-  public static final int DT_UINT64_VALUE = 23;
-  /**
-   * <pre>
-   * Do not use!  These are only for parameters.  Every enum above
-   * should have a corresponding value below (verified by types_test).
-   * </pre>
-   *
-   * <code>DT_FLOAT_REF = 101;</code>
-   */
-  public static final int DT_FLOAT_REF_VALUE = 101;
-  /**
-   * <code>DT_DOUBLE_REF = 102;</code>
-   */
-  public static final int DT_DOUBLE_REF_VALUE = 102;
-  /**
-   * <code>DT_INT32_REF = 103;</code>
-   */
-  public static final int DT_INT32_REF_VALUE = 103;
-  /**
-   * <code>DT_UINT8_REF = 104;</code>
-   */
-  public static final int DT_UINT8_REF_VALUE = 104;
-  /**
-   * <code>DT_INT16_REF = 105;</code>
-   */
-  public static final int DT_INT16_REF_VALUE = 105;
-  /**
-   * <code>DT_INT8_REF = 106;</code>
-   */
-  public static final int DT_INT8_REF_VALUE = 106;
-  /**
-   * <code>DT_STRING_REF = 107;</code>
-   */
-  public static final int DT_STRING_REF_VALUE = 107;
-  /**
-   * <code>DT_COMPLEX64_REF = 108;</code>
-   */
-  public static final int DT_COMPLEX64_REF_VALUE = 108;
-  /**
-   * <code>DT_INT64_REF = 109;</code>
-   */
-  public static final int DT_INT64_REF_VALUE = 109;
-  /**
-   * <code>DT_BOOL_REF = 110;</code>
-   */
-  public static final int DT_BOOL_REF_VALUE = 110;
-  /**
-   * <code>DT_QINT8_REF = 111;</code>
-   */
-  public static final int DT_QINT8_REF_VALUE = 111;
-  /**
-   * <code>DT_QUINT8_REF = 112;</code>
-   */
-  public static final int DT_QUINT8_REF_VALUE = 112;
-  /**
-   * <code>DT_QINT32_REF = 113;</code>
-   */
-  public static final int DT_QINT32_REF_VALUE = 113;
-  /**
-   * <code>DT_BFLOAT16_REF = 114;</code>
-   */
-  public static final int DT_BFLOAT16_REF_VALUE = 114;
-  /**
-   * <code>DT_QINT16_REF = 115;</code>
-   */
-  public static final int DT_QINT16_REF_VALUE = 115;
-  /**
-   * <code>DT_QUINT16_REF = 116;</code>
-   */
-  public static final int DT_QUINT16_REF_VALUE = 116;
-  /**
-   * <code>DT_UINT16_REF = 117;</code>
-   */
-  public static final int DT_UINT16_REF_VALUE = 117;
-  /**
-   * <code>DT_COMPLEX128_REF = 118;</code>
-   */
-  public static final int DT_COMPLEX128_REF_VALUE = 118;
-  /**
-   * <code>DT_HALF_REF = 119;</code>
-   */
-  public static final int DT_HALF_REF_VALUE = 119;
-  /**
-   * <code>DT_RESOURCE_REF = 120;</code>
-   */
-  public static final int DT_RESOURCE_REF_VALUE = 120;
-  /**
-   * <code>DT_VARIANT_REF = 121;</code>
-   */
-  public static final int DT_VARIANT_REF_VALUE = 121;
-  /**
-   * <code>DT_UINT32_REF = 122;</code>
-   */
-  public static final int DT_UINT32_REF_VALUE = 122;
-  /**
-   * <code>DT_UINT64_REF = 123;</code>
-   */
-  public static final int DT_UINT64_REF_VALUE = 123;
-
-
-  public final int getNumber() {
-    if (this == UNRECOGNIZED) {
-      throw new java.lang.IllegalArgumentException(
-          "Can't get the number of an unknown enum value.");
-    }
-    return value;
-  }
-
-  /**
-   * @deprecated Use {@link #forNumber(int)} instead.
-   */
-  @java.lang.Deprecated
-  public static DataType valueOf(int value) {
-    return forNumber(value);
-  }
-
-  public static DataType forNumber(int value) {
-    switch (value) {
-      case 0: return DT_INVALID;
-      case 1: return DT_FLOAT;
-      case 2: return DT_DOUBLE;
-      case 3: return DT_INT32;
-      case 4: return DT_UINT8;
-      case 5: return DT_INT16;
-      case 6: return DT_INT8;
-      case 7: return DT_STRING;
-      case 8: return DT_COMPLEX64;
-      case 9: return DT_INT64;
-      case 10: return DT_BOOL;
-      case 11: return DT_QINT8;
-      case 12: return DT_QUINT8;
-      case 13: return DT_QINT32;
-      case 14: return DT_BFLOAT16;
-      case 15: return DT_QINT16;
-      case 16: return DT_QUINT16;
-      case 17: return DT_UINT16;
-      case 18: return DT_COMPLEX128;
-      case 19: return DT_HALF;
-      case 20: return DT_RESOURCE;
-      case 21: return DT_VARIANT;
-      case 22: return DT_UINT32;
-      case 23: return DT_UINT64;
-      case 101: return DT_FLOAT_REF;
-      case 102: return DT_DOUBLE_REF;
-      case 103: return DT_INT32_REF;
-      case 104: return DT_UINT8_REF;
-      case 105: return DT_INT16_REF;
-      case 106: return DT_INT8_REF;
-      case 107: return DT_STRING_REF;
-      case 108: return DT_COMPLEX64_REF;
-      case 109: return DT_INT64_REF;
-      case 110: return DT_BOOL_REF;
-      case 111: return DT_QINT8_REF;
-      case 112: return DT_QUINT8_REF;
-      case 113: return DT_QINT32_REF;
-      case 114: return DT_BFLOAT16_REF;
-      case 115: return DT_QINT16_REF;
-      case 116: return DT_QUINT16_REF;
-      case 117: return DT_UINT16_REF;
-      case 118: return DT_COMPLEX128_REF;
-      case 119: return DT_HALF_REF;
-      case 120: return DT_RESOURCE_REF;
-      case 121: return DT_VARIANT_REF;
-      case 122: return DT_UINT32_REF;
-      case 123: return DT_UINT64_REF;
-      default: return null;
-    }
-  }
-
-  public static com.google.protobuf.Internal.EnumLiteMap<DataType>
-      internalGetValueMap() {
-    return internalValueMap;
-  }
-  private static final com.google.protobuf.Internal.EnumLiteMap<
-      DataType> internalValueMap =
-        new com.google.protobuf.Internal.EnumLiteMap<DataType>() {
-          public DataType findValueByNumber(int number) {
-            return DataType.forNumber(number);
-          }
-        };
-
-  public final com.google.protobuf.Descriptors.EnumValueDescriptor
-      getValueDescriptor() {
-    return getDescriptor().getValues().get(ordinal());
-  }
-  public final com.google.protobuf.Descriptors.EnumDescriptor
-      getDescriptorForType() {
-    return getDescriptor();
-  }
-  public static final com.google.protobuf.Descriptors.EnumDescriptor
-      getDescriptor() {
-    return org.tensorflow.proto.framework.TypesProtos.getDescriptor().getEnumTypes().get(0);
-  }
-
-  private static final DataType[] VALUES = values();
-
-  public static DataType valueOf(
-      com.google.protobuf.Descriptors.EnumValueDescriptor desc) {
-    if (desc.getType() != getDescriptor()) {
-      throw new java.lang.IllegalArgumentException(
-        "EnumValueDescriptor is not for this type.");
-    }
-    if (desc.getIndex() == -1) {
-      return UNRECOGNIZED;
-    }
-    return VALUES[desc.getIndex()];
-  }
-
-  private final int value;
-
-  private DataType(int value) {
-    this.value = value;
-  }
-
-  // @@protoc_insertion_point(enum_scope:tensorflow.DataType)
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DebugOptions.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DebugOptions.java
deleted file mode 100644
index 73b4e10af4e..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DebugOptions.java
+++ /dev/null
@@ -1,1033 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/debug.proto
-
-package org.tensorflow.proto.framework;
-
-/**
- * <pre>
- * Options for initializing DebuggerState in TensorFlow Debugger (tfdbg).
- * </pre>
- *
- * Protobuf type {@code tensorflow.DebugOptions}
- */
-public  final class DebugOptions extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.DebugOptions)
-    DebugOptionsOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use DebugOptions.newBuilder() to construct.
-  private DebugOptions(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private DebugOptions() {
-    debugTensorWatchOpts_ = java.util.Collections.emptyList();
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new DebugOptions();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private DebugOptions(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    int mutable_bitField0_ = 0;
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 34: {
-            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
-              debugTensorWatchOpts_ = new java.util.ArrayList<org.tensorflow.proto.framework.DebugTensorWatch>();
-              mutable_bitField0_ |= 0x00000001;
-            }
-            debugTensorWatchOpts_.add(
-                input.readMessage(org.tensorflow.proto.framework.DebugTensorWatch.parser(), extensionRegistry));
-            break;
-          }
-          case 80: {
-
-            globalStep_ = input.readInt64();
-            break;
-          }
-          case 88: {
-
-            resetDiskByteUsage_ = input.readBool();
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      if (((mutable_bitField0_ & 0x00000001) != 0)) {
-        debugTensorWatchOpts_ = java.util.Collections.unmodifiableList(debugTensorWatchOpts_);
-      }
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.framework.DebugProtos.internal_static_tensorflow_DebugOptions_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.framework.DebugProtos.internal_static_tensorflow_DebugOptions_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.framework.DebugOptions.class, org.tensorflow.proto.framework.DebugOptions.Builder.class);
-  }
-
-  public static final int DEBUG_TENSOR_WATCH_OPTS_FIELD_NUMBER = 4;
-  private java.util.List<org.tensorflow.proto.framework.DebugTensorWatch> debugTensorWatchOpts_;
-  /**
-   * <pre>
-   * Debugging options
-   * </pre>
-   *
-   * <code>repeated .tensorflow.DebugTensorWatch debug_tensor_watch_opts = 4;</code>
-   */
-  public java.util.List<org.tensorflow.proto.framework.DebugTensorWatch> getDebugTensorWatchOptsList() {
-    return debugTensorWatchOpts_;
-  }
-  /**
-   * <pre>
-   * Debugging options
-   * </pre>
-   *
-   * <code>repeated .tensorflow.DebugTensorWatch debug_tensor_watch_opts = 4;</code>
-   */
-  public java.util.List<? extends org.tensorflow.proto.framework.DebugTensorWatchOrBuilder> 
-      getDebugTensorWatchOptsOrBuilderList() {
-    return debugTensorWatchOpts_;
-  }
-  /**
-   * <pre>
-   * Debugging options
-   * </pre>
-   *
-   * <code>repeated .tensorflow.DebugTensorWatch debug_tensor_watch_opts = 4;</code>
-   */
-  public int getDebugTensorWatchOptsCount() {
-    return debugTensorWatchOpts_.size();
-  }
-  /**
-   * <pre>
-   * Debugging options
-   * </pre>
-   *
-   * <code>repeated .tensorflow.DebugTensorWatch debug_tensor_watch_opts = 4;</code>
-   */
-  public org.tensorflow.proto.framework.DebugTensorWatch getDebugTensorWatchOpts(int index) {
-    return debugTensorWatchOpts_.get(index);
-  }
-  /**
-   * <pre>
-   * Debugging options
-   * </pre>
-   *
-   * <code>repeated .tensorflow.DebugTensorWatch debug_tensor_watch_opts = 4;</code>
-   */
-  public org.tensorflow.proto.framework.DebugTensorWatchOrBuilder getDebugTensorWatchOptsOrBuilder(
-      int index) {
-    return debugTensorWatchOpts_.get(index);
-  }
-
-  public static final int GLOBAL_STEP_FIELD_NUMBER = 10;
-  private long globalStep_;
-  /**
-   * <pre>
-   * Caller-specified global step count.
-   * Note that this is distinct from the session run count and the executor
-   * step count.
-   * </pre>
-   *
-   * <code>int64 global_step = 10;</code>
-   */
-  public long getGlobalStep() {
-    return globalStep_;
-  }
-
-  public static final int RESET_DISK_BYTE_USAGE_FIELD_NUMBER = 11;
-  private boolean resetDiskByteUsage_;
-  /**
-   * <pre>
-   * Whether the total disk usage of tfdbg is to be reset to zero
-   * in this Session.run call. This is used by wrappers and hooks
-   * such as the local CLI ones to indicate that the dumped tensors
-   * are cleaned up from the disk after each Session.run.
-   * </pre>
-   *
-   * <code>bool reset_disk_byte_usage = 11;</code>
-   */
-  public boolean getResetDiskByteUsage() {
-    return resetDiskByteUsage_;
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    for (int i = 0; i < debugTensorWatchOpts_.size(); i++) {
-      output.writeMessage(4, debugTensorWatchOpts_.get(i));
-    }
-    if (globalStep_ != 0L) {
-      output.writeInt64(10, globalStep_);
-    }
-    if (resetDiskByteUsage_ != false) {
-      output.writeBool(11, resetDiskByteUsage_);
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    for (int i = 0; i < debugTensorWatchOpts_.size(); i++) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(4, debugTensorWatchOpts_.get(i));
-    }
-    if (globalStep_ != 0L) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt64Size(10, globalStep_);
-    }
-    if (resetDiskByteUsage_ != false) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeBoolSize(11, resetDiskByteUsage_);
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.framework.DebugOptions)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.framework.DebugOptions other = (org.tensorflow.proto.framework.DebugOptions) obj;
-
-    if (!getDebugTensorWatchOptsList()
-        .equals(other.getDebugTensorWatchOptsList())) return false;
-    if (getGlobalStep()
-        != other.getGlobalStep()) return false;
-    if (getResetDiskByteUsage()
-        != other.getResetDiskByteUsage()) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    if (getDebugTensorWatchOptsCount() > 0) {
-      hash = (37 * hash) + DEBUG_TENSOR_WATCH_OPTS_FIELD_NUMBER;
-      hash = (53 * hash) + getDebugTensorWatchOptsList().hashCode();
-    }
-    hash = (37 * hash) + GLOBAL_STEP_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-        getGlobalStep());
-    hash = (37 * hash) + RESET_DISK_BYTE_USAGE_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashBoolean(
-        getResetDiskByteUsage());
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.framework.DebugOptions parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.DebugOptions parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.DebugOptions parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.DebugOptions parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.DebugOptions parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.DebugOptions parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.DebugOptions parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.DebugOptions parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.DebugOptions parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.DebugOptions parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.DebugOptions parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.DebugOptions parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.framework.DebugOptions prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * <pre>
-   * Options for initializing DebuggerState in TensorFlow Debugger (tfdbg).
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.DebugOptions}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.DebugOptions)
-      org.tensorflow.proto.framework.DebugOptionsOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.DebugProtos.internal_static_tensorflow_DebugOptions_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.DebugProtos.internal_static_tensorflow_DebugOptions_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.DebugOptions.class, org.tensorflow.proto.framework.DebugOptions.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.framework.DebugOptions.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-        getDebugTensorWatchOptsFieldBuilder();
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      if (debugTensorWatchOptsBuilder_ == null) {
-        debugTensorWatchOpts_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000001);
-      } else {
-        debugTensorWatchOptsBuilder_.clear();
-      }
-      globalStep_ = 0L;
-
-      resetDiskByteUsage_ = false;
-
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.framework.DebugProtos.internal_static_tensorflow_DebugOptions_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.DebugOptions getDefaultInstanceForType() {
-      return org.tensorflow.proto.framework.DebugOptions.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.DebugOptions build() {
-      org.tensorflow.proto.framework.DebugOptions result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.DebugOptions buildPartial() {
-      org.tensorflow.proto.framework.DebugOptions result = new org.tensorflow.proto.framework.DebugOptions(this);
-      int from_bitField0_ = bitField0_;
-      if (debugTensorWatchOptsBuilder_ == null) {
-        if (((bitField0_ & 0x00000001) != 0)) {
-          debugTensorWatchOpts_ = java.util.Collections.unmodifiableList(debugTensorWatchOpts_);
-          bitField0_ = (bitField0_ & ~0x00000001);
-        }
-        result.debugTensorWatchOpts_ = debugTensorWatchOpts_;
-      } else {
-        result.debugTensorWatchOpts_ = debugTensorWatchOptsBuilder_.build();
-      }
-      result.globalStep_ = globalStep_;
-      result.resetDiskByteUsage_ = resetDiskByteUsage_;
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.framework.DebugOptions) {
-        return mergeFrom((org.tensorflow.proto.framework.DebugOptions)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.framework.DebugOptions other) {
-      if (other == org.tensorflow.proto.framework.DebugOptions.getDefaultInstance()) return this;
-      if (debugTensorWatchOptsBuilder_ == null) {
-        if (!other.debugTensorWatchOpts_.isEmpty()) {
-          if (debugTensorWatchOpts_.isEmpty()) {
-            debugTensorWatchOpts_ = other.debugTensorWatchOpts_;
-            bitField0_ = (bitField0_ & ~0x00000001);
-          } else {
-            ensureDebugTensorWatchOptsIsMutable();
-            debugTensorWatchOpts_.addAll(other.debugTensorWatchOpts_);
-          }
-          onChanged();
-        }
-      } else {
-        if (!other.debugTensorWatchOpts_.isEmpty()) {
-          if (debugTensorWatchOptsBuilder_.isEmpty()) {
-            debugTensorWatchOptsBuilder_.dispose();
-            debugTensorWatchOptsBuilder_ = null;
-            debugTensorWatchOpts_ = other.debugTensorWatchOpts_;
-            bitField0_ = (bitField0_ & ~0x00000001);
-            debugTensorWatchOptsBuilder_ = 
-              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
-                 getDebugTensorWatchOptsFieldBuilder() : null;
-          } else {
-            debugTensorWatchOptsBuilder_.addAllMessages(other.debugTensorWatchOpts_);
-          }
-        }
-      }
-      if (other.getGlobalStep() != 0L) {
-        setGlobalStep(other.getGlobalStep());
-      }
-      if (other.getResetDiskByteUsage() != false) {
-        setResetDiskByteUsage(other.getResetDiskByteUsage());
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.framework.DebugOptions parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.framework.DebugOptions) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-    private int bitField0_;
-
-    private java.util.List<org.tensorflow.proto.framework.DebugTensorWatch> debugTensorWatchOpts_ =
-      java.util.Collections.emptyList();
-    private void ensureDebugTensorWatchOptsIsMutable() {
-      if (!((bitField0_ & 0x00000001) != 0)) {
-        debugTensorWatchOpts_ = new java.util.ArrayList<org.tensorflow.proto.framework.DebugTensorWatch>(debugTensorWatchOpts_);
-        bitField0_ |= 0x00000001;
-       }
-    }
-
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.framework.DebugTensorWatch, org.tensorflow.proto.framework.DebugTensorWatch.Builder, org.tensorflow.proto.framework.DebugTensorWatchOrBuilder> debugTensorWatchOptsBuilder_;
-
-    /**
-     * <pre>
-     * Debugging options
-     * </pre>
-     *
-     * <code>repeated .tensorflow.DebugTensorWatch debug_tensor_watch_opts = 4;</code>
-     */
-    public java.util.List<org.tensorflow.proto.framework.DebugTensorWatch> getDebugTensorWatchOptsList() {
-      if (debugTensorWatchOptsBuilder_ == null) {
-        return java.util.Collections.unmodifiableList(debugTensorWatchOpts_);
-      } else {
-        return debugTensorWatchOptsBuilder_.getMessageList();
-      }
-    }
-    /**
-     * <pre>
-     * Debugging options
-     * </pre>
-     *
-     * <code>repeated .tensorflow.DebugTensorWatch debug_tensor_watch_opts = 4;</code>
-     */
-    public int getDebugTensorWatchOptsCount() {
-      if (debugTensorWatchOptsBuilder_ == null) {
-        return debugTensorWatchOpts_.size();
-      } else {
-        return debugTensorWatchOptsBuilder_.getCount();
-      }
-    }
-    /**
-     * <pre>
-     * Debugging options
-     * </pre>
-     *
-     * <code>repeated .tensorflow.DebugTensorWatch debug_tensor_watch_opts = 4;</code>
-     */
-    public org.tensorflow.proto.framework.DebugTensorWatch getDebugTensorWatchOpts(int index) {
-      if (debugTensorWatchOptsBuilder_ == null) {
-        return debugTensorWatchOpts_.get(index);
-      } else {
-        return debugTensorWatchOptsBuilder_.getMessage(index);
-      }
-    }
-    /**
-     * <pre>
-     * Debugging options
-     * </pre>
-     *
-     * <code>repeated .tensorflow.DebugTensorWatch debug_tensor_watch_opts = 4;</code>
-     */
-    public Builder setDebugTensorWatchOpts(
-        int index, org.tensorflow.proto.framework.DebugTensorWatch value) {
-      if (debugTensorWatchOptsBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureDebugTensorWatchOptsIsMutable();
-        debugTensorWatchOpts_.set(index, value);
-        onChanged();
-      } else {
-        debugTensorWatchOptsBuilder_.setMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Debugging options
-     * </pre>
-     *
-     * <code>repeated .tensorflow.DebugTensorWatch debug_tensor_watch_opts = 4;</code>
-     */
-    public Builder setDebugTensorWatchOpts(
-        int index, org.tensorflow.proto.framework.DebugTensorWatch.Builder builderForValue) {
-      if (debugTensorWatchOptsBuilder_ == null) {
-        ensureDebugTensorWatchOptsIsMutable();
-        debugTensorWatchOpts_.set(index, builderForValue.build());
-        onChanged();
-      } else {
-        debugTensorWatchOptsBuilder_.setMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Debugging options
-     * </pre>
-     *
-     * <code>repeated .tensorflow.DebugTensorWatch debug_tensor_watch_opts = 4;</code>
-     */
-    public Builder addDebugTensorWatchOpts(org.tensorflow.proto.framework.DebugTensorWatch value) {
-      if (debugTensorWatchOptsBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureDebugTensorWatchOptsIsMutable();
-        debugTensorWatchOpts_.add(value);
-        onChanged();
-      } else {
-        debugTensorWatchOptsBuilder_.addMessage(value);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Debugging options
-     * </pre>
-     *
-     * <code>repeated .tensorflow.DebugTensorWatch debug_tensor_watch_opts = 4;</code>
-     */
-    public Builder addDebugTensorWatchOpts(
-        int index, org.tensorflow.proto.framework.DebugTensorWatch value) {
-      if (debugTensorWatchOptsBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureDebugTensorWatchOptsIsMutable();
-        debugTensorWatchOpts_.add(index, value);
-        onChanged();
-      } else {
-        debugTensorWatchOptsBuilder_.addMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Debugging options
-     * </pre>
-     *
-     * <code>repeated .tensorflow.DebugTensorWatch debug_tensor_watch_opts = 4;</code>
-     */
-    public Builder addDebugTensorWatchOpts(
-        org.tensorflow.proto.framework.DebugTensorWatch.Builder builderForValue) {
-      if (debugTensorWatchOptsBuilder_ == null) {
-        ensureDebugTensorWatchOptsIsMutable();
-        debugTensorWatchOpts_.add(builderForValue.build());
-        onChanged();
-      } else {
-        debugTensorWatchOptsBuilder_.addMessage(builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Debugging options
-     * </pre>
-     *
-     * <code>repeated .tensorflow.DebugTensorWatch debug_tensor_watch_opts = 4;</code>
-     */
-    public Builder addDebugTensorWatchOpts(
-        int index, org.tensorflow.proto.framework.DebugTensorWatch.Builder builderForValue) {
-      if (debugTensorWatchOptsBuilder_ == null) {
-        ensureDebugTensorWatchOptsIsMutable();
-        debugTensorWatchOpts_.add(index, builderForValue.build());
-        onChanged();
-      } else {
-        debugTensorWatchOptsBuilder_.addMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Debugging options
-     * </pre>
-     *
-     * <code>repeated .tensorflow.DebugTensorWatch debug_tensor_watch_opts = 4;</code>
-     */
-    public Builder addAllDebugTensorWatchOpts(
-        java.lang.Iterable<? extends org.tensorflow.proto.framework.DebugTensorWatch> values) {
-      if (debugTensorWatchOptsBuilder_ == null) {
-        ensureDebugTensorWatchOptsIsMutable();
-        com.google.protobuf.AbstractMessageLite.Builder.addAll(
-            values, debugTensorWatchOpts_);
-        onChanged();
-      } else {
-        debugTensorWatchOptsBuilder_.addAllMessages(values);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Debugging options
-     * </pre>
-     *
-     * <code>repeated .tensorflow.DebugTensorWatch debug_tensor_watch_opts = 4;</code>
-     */
-    public Builder clearDebugTensorWatchOpts() {
-      if (debugTensorWatchOptsBuilder_ == null) {
-        debugTensorWatchOpts_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000001);
-        onChanged();
-      } else {
-        debugTensorWatchOptsBuilder_.clear();
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Debugging options
-     * </pre>
-     *
-     * <code>repeated .tensorflow.DebugTensorWatch debug_tensor_watch_opts = 4;</code>
-     */
-    public Builder removeDebugTensorWatchOpts(int index) {
-      if (debugTensorWatchOptsBuilder_ == null) {
-        ensureDebugTensorWatchOptsIsMutable();
-        debugTensorWatchOpts_.remove(index);
-        onChanged();
-      } else {
-        debugTensorWatchOptsBuilder_.remove(index);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Debugging options
-     * </pre>
-     *
-     * <code>repeated .tensorflow.DebugTensorWatch debug_tensor_watch_opts = 4;</code>
-     */
-    public org.tensorflow.proto.framework.DebugTensorWatch.Builder getDebugTensorWatchOptsBuilder(
-        int index) {
-      return getDebugTensorWatchOptsFieldBuilder().getBuilder(index);
-    }
-    /**
-     * <pre>
-     * Debugging options
-     * </pre>
-     *
-     * <code>repeated .tensorflow.DebugTensorWatch debug_tensor_watch_opts = 4;</code>
-     */
-    public org.tensorflow.proto.framework.DebugTensorWatchOrBuilder getDebugTensorWatchOptsOrBuilder(
-        int index) {
-      if (debugTensorWatchOptsBuilder_ == null) {
-        return debugTensorWatchOpts_.get(index);  } else {
-        return debugTensorWatchOptsBuilder_.getMessageOrBuilder(index);
-      }
-    }
-    /**
-     * <pre>
-     * Debugging options
-     * </pre>
-     *
-     * <code>repeated .tensorflow.DebugTensorWatch debug_tensor_watch_opts = 4;</code>
-     */
-    public java.util.List<? extends org.tensorflow.proto.framework.DebugTensorWatchOrBuilder> 
-         getDebugTensorWatchOptsOrBuilderList() {
-      if (debugTensorWatchOptsBuilder_ != null) {
-        return debugTensorWatchOptsBuilder_.getMessageOrBuilderList();
-      } else {
-        return java.util.Collections.unmodifiableList(debugTensorWatchOpts_);
-      }
-    }
-    /**
-     * <pre>
-     * Debugging options
-     * </pre>
-     *
-     * <code>repeated .tensorflow.DebugTensorWatch debug_tensor_watch_opts = 4;</code>
-     */
-    public org.tensorflow.proto.framework.DebugTensorWatch.Builder addDebugTensorWatchOptsBuilder() {
-      return getDebugTensorWatchOptsFieldBuilder().addBuilder(
-          org.tensorflow.proto.framework.DebugTensorWatch.getDefaultInstance());
-    }
-    /**
-     * <pre>
-     * Debugging options
-     * </pre>
-     *
-     * <code>repeated .tensorflow.DebugTensorWatch debug_tensor_watch_opts = 4;</code>
-     */
-    public org.tensorflow.proto.framework.DebugTensorWatch.Builder addDebugTensorWatchOptsBuilder(
-        int index) {
-      return getDebugTensorWatchOptsFieldBuilder().addBuilder(
-          index, org.tensorflow.proto.framework.DebugTensorWatch.getDefaultInstance());
-    }
-    /**
-     * <pre>
-     * Debugging options
-     * </pre>
-     *
-     * <code>repeated .tensorflow.DebugTensorWatch debug_tensor_watch_opts = 4;</code>
-     */
-    public java.util.List<org.tensorflow.proto.framework.DebugTensorWatch.Builder> 
-         getDebugTensorWatchOptsBuilderList() {
-      return getDebugTensorWatchOptsFieldBuilder().getBuilderList();
-    }
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.framework.DebugTensorWatch, org.tensorflow.proto.framework.DebugTensorWatch.Builder, org.tensorflow.proto.framework.DebugTensorWatchOrBuilder> 
-        getDebugTensorWatchOptsFieldBuilder() {
-      if (debugTensorWatchOptsBuilder_ == null) {
-        debugTensorWatchOptsBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
-            org.tensorflow.proto.framework.DebugTensorWatch, org.tensorflow.proto.framework.DebugTensorWatch.Builder, org.tensorflow.proto.framework.DebugTensorWatchOrBuilder>(
-                debugTensorWatchOpts_,
-                ((bitField0_ & 0x00000001) != 0),
-                getParentForChildren(),
-                isClean());
-        debugTensorWatchOpts_ = null;
-      }
-      return debugTensorWatchOptsBuilder_;
-    }
-
-    private long globalStep_ ;
-    /**
-     * <pre>
-     * Caller-specified global step count.
-     * Note that this is distinct from the session run count and the executor
-     * step count.
-     * </pre>
-     *
-     * <code>int64 global_step = 10;</code>
-     */
-    public long getGlobalStep() {
-      return globalStep_;
-    }
-    /**
-     * <pre>
-     * Caller-specified global step count.
-     * Note that this is distinct from the session run count and the executor
-     * step count.
-     * </pre>
-     *
-     * <code>int64 global_step = 10;</code>
-     */
-    public Builder setGlobalStep(long value) {
-      
-      globalStep_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Caller-specified global step count.
-     * Note that this is distinct from the session run count and the executor
-     * step count.
-     * </pre>
-     *
-     * <code>int64 global_step = 10;</code>
-     */
-    public Builder clearGlobalStep() {
-      
-      globalStep_ = 0L;
-      onChanged();
-      return this;
-    }
-
-    private boolean resetDiskByteUsage_ ;
-    /**
-     * <pre>
-     * Whether the total disk usage of tfdbg is to be reset to zero
-     * in this Session.run call. This is used by wrappers and hooks
-     * such as the local CLI ones to indicate that the dumped tensors
-     * are cleaned up from the disk after each Session.run.
-     * </pre>
-     *
-     * <code>bool reset_disk_byte_usage = 11;</code>
-     */
-    public boolean getResetDiskByteUsage() {
-      return resetDiskByteUsage_;
-    }
-    /**
-     * <pre>
-     * Whether the total disk usage of tfdbg is to be reset to zero
-     * in this Session.run call. This is used by wrappers and hooks
-     * such as the local CLI ones to indicate that the dumped tensors
-     * are cleaned up from the disk after each Session.run.
-     * </pre>
-     *
-     * <code>bool reset_disk_byte_usage = 11;</code>
-     */
-    public Builder setResetDiskByteUsage(boolean value) {
-      
-      resetDiskByteUsage_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Whether the total disk usage of tfdbg is to be reset to zero
-     * in this Session.run call. This is used by wrappers and hooks
-     * such as the local CLI ones to indicate that the dumped tensors
-     * are cleaned up from the disk after each Session.run.
-     * </pre>
-     *
-     * <code>bool reset_disk_byte_usage = 11;</code>
-     */
-    public Builder clearResetDiskByteUsage() {
-      
-      resetDiskByteUsage_ = false;
-      onChanged();
-      return this;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.DebugOptions)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.DebugOptions)
-  private static final org.tensorflow.proto.framework.DebugOptions DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.DebugOptions();
-  }
-
-  public static org.tensorflow.proto.framework.DebugOptions getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<DebugOptions>
-      PARSER = new com.google.protobuf.AbstractParser<DebugOptions>() {
-    @java.lang.Override
-    public DebugOptions parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new DebugOptions(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<DebugOptions> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<DebugOptions> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.framework.DebugOptions getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DebugOptionsOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DebugOptionsOrBuilder.java
deleted file mode 100644
index 65f0407737a..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DebugOptionsOrBuilder.java
+++ /dev/null
@@ -1,76 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/debug.proto
-
-package org.tensorflow.proto.framework;
-
-public interface DebugOptionsOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.DebugOptions)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <pre>
-   * Debugging options
-   * </pre>
-   *
-   * <code>repeated .tensorflow.DebugTensorWatch debug_tensor_watch_opts = 4;</code>
-   */
-  java.util.List<org.tensorflow.proto.framework.DebugTensorWatch> 
-      getDebugTensorWatchOptsList();
-  /**
-   * <pre>
-   * Debugging options
-   * </pre>
-   *
-   * <code>repeated .tensorflow.DebugTensorWatch debug_tensor_watch_opts = 4;</code>
-   */
-  org.tensorflow.proto.framework.DebugTensorWatch getDebugTensorWatchOpts(int index);
-  /**
-   * <pre>
-   * Debugging options
-   * </pre>
-   *
-   * <code>repeated .tensorflow.DebugTensorWatch debug_tensor_watch_opts = 4;</code>
-   */
-  int getDebugTensorWatchOptsCount();
-  /**
-   * <pre>
-   * Debugging options
-   * </pre>
-   *
-   * <code>repeated .tensorflow.DebugTensorWatch debug_tensor_watch_opts = 4;</code>
-   */
-  java.util.List<? extends org.tensorflow.proto.framework.DebugTensorWatchOrBuilder> 
-      getDebugTensorWatchOptsOrBuilderList();
-  /**
-   * <pre>
-   * Debugging options
-   * </pre>
-   *
-   * <code>repeated .tensorflow.DebugTensorWatch debug_tensor_watch_opts = 4;</code>
-   */
-  org.tensorflow.proto.framework.DebugTensorWatchOrBuilder getDebugTensorWatchOptsOrBuilder(
-      int index);
-
-  /**
-   * <pre>
-   * Caller-specified global step count.
-   * Note that this is distinct from the session run count and the executor
-   * step count.
-   * </pre>
-   *
-   * <code>int64 global_step = 10;</code>
-   */
-  long getGlobalStep();
-
-  /**
-   * <pre>
-   * Whether the total disk usage of tfdbg is to be reset to zero
-   * in this Session.run call. This is used by wrappers and hooks
-   * such as the local CLI ones to indicate that the dumped tensors
-   * are cleaned up from the disk after each Session.run.
-   * </pre>
-   *
-   * <code>bool reset_disk_byte_usage = 11;</code>
-   */
-  boolean getResetDiskByteUsage();
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DebugProtos.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DebugProtos.java
deleted file mode 100644
index edc853e9376..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DebugProtos.java
+++ /dev/null
@@ -1,95 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/debug.proto
-
-package org.tensorflow.proto.framework;
-
-public final class DebugProtos {
-  private DebugProtos() {}
-  public static void registerAllExtensions(
-      com.google.protobuf.ExtensionRegistryLite registry) {
-  }
-
-  public static void registerAllExtensions(
-      com.google.protobuf.ExtensionRegistry registry) {
-    registerAllExtensions(
-        (com.google.protobuf.ExtensionRegistryLite) registry);
-  }
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_DebugTensorWatch_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_DebugTensorWatch_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_DebugOptions_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_DebugOptions_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_DebuggedSourceFile_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_DebuggedSourceFile_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_DebuggedSourceFiles_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_DebuggedSourceFiles_fieldAccessorTable;
-
-  public static com.google.protobuf.Descriptors.FileDescriptor
-      getDescriptor() {
-    return descriptor;
-  }
-  private static  com.google.protobuf.Descriptors.FileDescriptor
-      descriptor;
-  static {
-    java.lang.String[] descriptorData = {
-      "\n$tensorflow/core/protobuf/debug.proto\022\n" +
-      "tensorflow\"\216\001\n\020DebugTensorWatch\022\021\n\tnode_" +
-      "name\030\001 \001(\t\022\023\n\013output_slot\030\002 \001(\005\022\021\n\tdebug" +
-      "_ops\030\003 \003(\t\022\022\n\ndebug_urls\030\004 \003(\t\022+\n#tolera" +
-      "te_debug_op_creation_failures\030\005 \001(\010\"\201\001\n\014" +
-      "DebugOptions\022=\n\027debug_tensor_watch_opts\030" +
-      "\004 \003(\0132\034.tensorflow.DebugTensorWatch\022\023\n\013g" +
-      "lobal_step\030\n \001(\003\022\035\n\025reset_disk_byte_usag" +
-      "e\030\013 \001(\010\"j\n\022DebuggedSourceFile\022\014\n\004host\030\001 " +
-      "\001(\t\022\021\n\tfile_path\030\002 \001(\t\022\025\n\rlast_modified\030" +
-      "\003 \001(\003\022\r\n\005bytes\030\004 \001(\003\022\r\n\005lines\030\005 \003(\t\"K\n\023D" +
-      "ebuggedSourceFiles\0224\n\014source_files\030\001 \003(\013" +
-      "2\036.tensorflow.DebuggedSourceFileB|\n\036org." +
-      "tensorflow.proto.frameworkB\013DebugProtosP" +
-      "\001ZHgithub.com/tensorflow/tensorflow/tens" +
-      "orflow/go/core/core_protos_go_proto\370\001\001b\006" +
-      "proto3"
-    };
-    descriptor = com.google.protobuf.Descriptors.FileDescriptor
-      .internalBuildGeneratedFileFrom(descriptorData,
-        new com.google.protobuf.Descriptors.FileDescriptor[] {
-        });
-    internal_static_tensorflow_DebugTensorWatch_descriptor =
-      getDescriptor().getMessageTypes().get(0);
-    internal_static_tensorflow_DebugTensorWatch_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_DebugTensorWatch_descriptor,
-        new java.lang.String[] { "NodeName", "OutputSlot", "DebugOps", "DebugUrls", "TolerateDebugOpCreationFailures", });
-    internal_static_tensorflow_DebugOptions_descriptor =
-      getDescriptor().getMessageTypes().get(1);
-    internal_static_tensorflow_DebugOptions_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_DebugOptions_descriptor,
-        new java.lang.String[] { "DebugTensorWatchOpts", "GlobalStep", "ResetDiskByteUsage", });
-    internal_static_tensorflow_DebuggedSourceFile_descriptor =
-      getDescriptor().getMessageTypes().get(2);
-    internal_static_tensorflow_DebuggedSourceFile_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_DebuggedSourceFile_descriptor,
-        new java.lang.String[] { "Host", "FilePath", "LastModified", "Bytes", "Lines", });
-    internal_static_tensorflow_DebuggedSourceFiles_descriptor =
-      getDescriptor().getMessageTypes().get(3);
-    internal_static_tensorflow_DebuggedSourceFiles_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_DebuggedSourceFiles_descriptor,
-        new java.lang.String[] { "SourceFiles", });
-  }
-
-  // @@protoc_insertion_point(outer_class_scope)
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DebugTensorWatch.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DebugTensorWatch.java
deleted file mode 100644
index 6536f3402fc..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DebugTensorWatch.java
+++ /dev/null
@@ -1,1444 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/debug.proto
-
-package org.tensorflow.proto.framework;
-
-/**
- * <pre>
- * Option for watching a node in TensorFlow Debugger (tfdbg).
- * </pre>
- *
- * Protobuf type {@code tensorflow.DebugTensorWatch}
- */
-public  final class DebugTensorWatch extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.DebugTensorWatch)
-    DebugTensorWatchOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use DebugTensorWatch.newBuilder() to construct.
-  private DebugTensorWatch(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private DebugTensorWatch() {
-    nodeName_ = "";
-    debugOps_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-    debugUrls_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new DebugTensorWatch();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private DebugTensorWatch(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    int mutable_bitField0_ = 0;
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 10: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            nodeName_ = s;
-            break;
-          }
-          case 16: {
-
-            outputSlot_ = input.readInt32();
-            break;
-          }
-          case 26: {
-            java.lang.String s = input.readStringRequireUtf8();
-            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
-              debugOps_ = new com.google.protobuf.LazyStringArrayList();
-              mutable_bitField0_ |= 0x00000001;
-            }
-            debugOps_.add(s);
-            break;
-          }
-          case 34: {
-            java.lang.String s = input.readStringRequireUtf8();
-            if (!((mutable_bitField0_ & 0x00000002) != 0)) {
-              debugUrls_ = new com.google.protobuf.LazyStringArrayList();
-              mutable_bitField0_ |= 0x00000002;
-            }
-            debugUrls_.add(s);
-            break;
-          }
-          case 40: {
-
-            tolerateDebugOpCreationFailures_ = input.readBool();
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      if (((mutable_bitField0_ & 0x00000001) != 0)) {
-        debugOps_ = debugOps_.getUnmodifiableView();
-      }
-      if (((mutable_bitField0_ & 0x00000002) != 0)) {
-        debugUrls_ = debugUrls_.getUnmodifiableView();
-      }
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.framework.DebugProtos.internal_static_tensorflow_DebugTensorWatch_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.framework.DebugProtos.internal_static_tensorflow_DebugTensorWatch_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.framework.DebugTensorWatch.class, org.tensorflow.proto.framework.DebugTensorWatch.Builder.class);
-  }
-
-  public static final int NODE_NAME_FIELD_NUMBER = 1;
-  private volatile java.lang.Object nodeName_;
-  /**
-   * <pre>
-   * Name of the node to watch.
-   * Use "*" for wildcard. But note: currently, regex is not supported in
-   * general.
-   * </pre>
-   *
-   * <code>string node_name = 1;</code>
-   */
-  public java.lang.String getNodeName() {
-    java.lang.Object ref = nodeName_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      nodeName_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * Name of the node to watch.
-   * Use "*" for wildcard. But note: currently, regex is not supported in
-   * general.
-   * </pre>
-   *
-   * <code>string node_name = 1;</code>
-   */
-  public com.google.protobuf.ByteString
-      getNodeNameBytes() {
-    java.lang.Object ref = nodeName_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      nodeName_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int OUTPUT_SLOT_FIELD_NUMBER = 2;
-  private int outputSlot_;
-  /**
-   * <pre>
-   * Output slot to watch.
-   * The semantics of output_slot == -1 is that all outputs of the node
-   * will be watched (i.e., a wildcard).
-   * Other negative values of output_slot are invalid and will lead to
-   * errors currently.
-   * </pre>
-   *
-   * <code>int32 output_slot = 2;</code>
-   */
-  public int getOutputSlot() {
-    return outputSlot_;
-  }
-
-  public static final int DEBUG_OPS_FIELD_NUMBER = 3;
-  private com.google.protobuf.LazyStringList debugOps_;
-  /**
-   * <pre>
-   * Name(s) of the debugging op(s).
-   * One or more than one probes on a tensor.
-   * e.g., {"DebugIdentity", "DebugNanCount"}
-   * </pre>
-   *
-   * <code>repeated string debug_ops = 3;</code>
-   */
-  public com.google.protobuf.ProtocolStringList
-      getDebugOpsList() {
-    return debugOps_;
-  }
-  /**
-   * <pre>
-   * Name(s) of the debugging op(s).
-   * One or more than one probes on a tensor.
-   * e.g., {"DebugIdentity", "DebugNanCount"}
-   * </pre>
-   *
-   * <code>repeated string debug_ops = 3;</code>
-   */
-  public int getDebugOpsCount() {
-    return debugOps_.size();
-  }
-  /**
-   * <pre>
-   * Name(s) of the debugging op(s).
-   * One or more than one probes on a tensor.
-   * e.g., {"DebugIdentity", "DebugNanCount"}
-   * </pre>
-   *
-   * <code>repeated string debug_ops = 3;</code>
-   */
-  public java.lang.String getDebugOps(int index) {
-    return debugOps_.get(index);
-  }
-  /**
-   * <pre>
-   * Name(s) of the debugging op(s).
-   * One or more than one probes on a tensor.
-   * e.g., {"DebugIdentity", "DebugNanCount"}
-   * </pre>
-   *
-   * <code>repeated string debug_ops = 3;</code>
-   */
-  public com.google.protobuf.ByteString
-      getDebugOpsBytes(int index) {
-    return debugOps_.getByteString(index);
-  }
-
-  public static final int DEBUG_URLS_FIELD_NUMBER = 4;
-  private com.google.protobuf.LazyStringList debugUrls_;
-  /**
-   * <pre>
-   * URL(s) for debug targets(s).
-   * Supported URL formats are:
-   *   - file:///foo/tfdbg_dump: Writes out Event content to file
-   *     /foo/tfdbg_dump.  Assumes all directories can be created if they don't
-   *     already exist.
-   *   - grpc://localhost:11011: Sends an RPC request to an EventListener
-   *     service running at localhost:11011 with the event.
-   *   - memcbk:///event_key: Routes tensors to clients using the
-   *     callback registered with the DebugCallbackRegistry for event_key.
-   * Each debug op listed in debug_ops will publish its output tensor (debug
-   * signal) to all URLs in debug_urls.
-   * N.B. Session::Run() supports concurrent invocations of the same inputs
-   * (feed keys), outputs and target nodes. If such concurrent invocations
-   * are to be debugged, the callers of Session::Run() must use distinct
-   * debug_urls to make sure that the streamed or dumped events do not overlap
-   * among the invocations.
-   * TODO(cais): More visible documentation of this in g3docs.
-   * </pre>
-   *
-   * <code>repeated string debug_urls = 4;</code>
-   */
-  public com.google.protobuf.ProtocolStringList
-      getDebugUrlsList() {
-    return debugUrls_;
-  }
-  /**
-   * <pre>
-   * URL(s) for debug targets(s).
-   * Supported URL formats are:
-   *   - file:///foo/tfdbg_dump: Writes out Event content to file
-   *     /foo/tfdbg_dump.  Assumes all directories can be created if they don't
-   *     already exist.
-   *   - grpc://localhost:11011: Sends an RPC request to an EventListener
-   *     service running at localhost:11011 with the event.
-   *   - memcbk:///event_key: Routes tensors to clients using the
-   *     callback registered with the DebugCallbackRegistry for event_key.
-   * Each debug op listed in debug_ops will publish its output tensor (debug
-   * signal) to all URLs in debug_urls.
-   * N.B. Session::Run() supports concurrent invocations of the same inputs
-   * (feed keys), outputs and target nodes. If such concurrent invocations
-   * are to be debugged, the callers of Session::Run() must use distinct
-   * debug_urls to make sure that the streamed or dumped events do not overlap
-   * among the invocations.
-   * TODO(cais): More visible documentation of this in g3docs.
-   * </pre>
-   *
-   * <code>repeated string debug_urls = 4;</code>
-   */
-  public int getDebugUrlsCount() {
-    return debugUrls_.size();
-  }
-  /**
-   * <pre>
-   * URL(s) for debug targets(s).
-   * Supported URL formats are:
-   *   - file:///foo/tfdbg_dump: Writes out Event content to file
-   *     /foo/tfdbg_dump.  Assumes all directories can be created if they don't
-   *     already exist.
-   *   - grpc://localhost:11011: Sends an RPC request to an EventListener
-   *     service running at localhost:11011 with the event.
-   *   - memcbk:///event_key: Routes tensors to clients using the
-   *     callback registered with the DebugCallbackRegistry for event_key.
-   * Each debug op listed in debug_ops will publish its output tensor (debug
-   * signal) to all URLs in debug_urls.
-   * N.B. Session::Run() supports concurrent invocations of the same inputs
-   * (feed keys), outputs and target nodes. If such concurrent invocations
-   * are to be debugged, the callers of Session::Run() must use distinct
-   * debug_urls to make sure that the streamed or dumped events do not overlap
-   * among the invocations.
-   * TODO(cais): More visible documentation of this in g3docs.
-   * </pre>
-   *
-   * <code>repeated string debug_urls = 4;</code>
-   */
-  public java.lang.String getDebugUrls(int index) {
-    return debugUrls_.get(index);
-  }
-  /**
-   * <pre>
-   * URL(s) for debug targets(s).
-   * Supported URL formats are:
-   *   - file:///foo/tfdbg_dump: Writes out Event content to file
-   *     /foo/tfdbg_dump.  Assumes all directories can be created if they don't
-   *     already exist.
-   *   - grpc://localhost:11011: Sends an RPC request to an EventListener
-   *     service running at localhost:11011 with the event.
-   *   - memcbk:///event_key: Routes tensors to clients using the
-   *     callback registered with the DebugCallbackRegistry for event_key.
-   * Each debug op listed in debug_ops will publish its output tensor (debug
-   * signal) to all URLs in debug_urls.
-   * N.B. Session::Run() supports concurrent invocations of the same inputs
-   * (feed keys), outputs and target nodes. If such concurrent invocations
-   * are to be debugged, the callers of Session::Run() must use distinct
-   * debug_urls to make sure that the streamed or dumped events do not overlap
-   * among the invocations.
-   * TODO(cais): More visible documentation of this in g3docs.
-   * </pre>
-   *
-   * <code>repeated string debug_urls = 4;</code>
-   */
-  public com.google.protobuf.ByteString
-      getDebugUrlsBytes(int index) {
-    return debugUrls_.getByteString(index);
-  }
-
-  public static final int TOLERATE_DEBUG_OP_CREATION_FAILURES_FIELD_NUMBER = 5;
-  private boolean tolerateDebugOpCreationFailures_;
-  /**
-   * <pre>
-   * Do not error out if debug op creation fails (e.g., due to dtype
-   * incompatibility). Instead, just log the failure.
-   * </pre>
-   *
-   * <code>bool tolerate_debug_op_creation_failures = 5;</code>
-   */
-  public boolean getTolerateDebugOpCreationFailures() {
-    return tolerateDebugOpCreationFailures_;
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    if (!getNodeNameBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 1, nodeName_);
-    }
-    if (outputSlot_ != 0) {
-      output.writeInt32(2, outputSlot_);
-    }
-    for (int i = 0; i < debugOps_.size(); i++) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 3, debugOps_.getRaw(i));
-    }
-    for (int i = 0; i < debugUrls_.size(); i++) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 4, debugUrls_.getRaw(i));
-    }
-    if (tolerateDebugOpCreationFailures_ != false) {
-      output.writeBool(5, tolerateDebugOpCreationFailures_);
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (!getNodeNameBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, nodeName_);
-    }
-    if (outputSlot_ != 0) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt32Size(2, outputSlot_);
-    }
-    {
-      int dataSize = 0;
-      for (int i = 0; i < debugOps_.size(); i++) {
-        dataSize += computeStringSizeNoTag(debugOps_.getRaw(i));
-      }
-      size += dataSize;
-      size += 1 * getDebugOpsList().size();
-    }
-    {
-      int dataSize = 0;
-      for (int i = 0; i < debugUrls_.size(); i++) {
-        dataSize += computeStringSizeNoTag(debugUrls_.getRaw(i));
-      }
-      size += dataSize;
-      size += 1 * getDebugUrlsList().size();
-    }
-    if (tolerateDebugOpCreationFailures_ != false) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeBoolSize(5, tolerateDebugOpCreationFailures_);
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.framework.DebugTensorWatch)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.framework.DebugTensorWatch other = (org.tensorflow.proto.framework.DebugTensorWatch) obj;
-
-    if (!getNodeName()
-        .equals(other.getNodeName())) return false;
-    if (getOutputSlot()
-        != other.getOutputSlot()) return false;
-    if (!getDebugOpsList()
-        .equals(other.getDebugOpsList())) return false;
-    if (!getDebugUrlsList()
-        .equals(other.getDebugUrlsList())) return false;
-    if (getTolerateDebugOpCreationFailures()
-        != other.getTolerateDebugOpCreationFailures()) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    hash = (37 * hash) + NODE_NAME_FIELD_NUMBER;
-    hash = (53 * hash) + getNodeName().hashCode();
-    hash = (37 * hash) + OUTPUT_SLOT_FIELD_NUMBER;
-    hash = (53 * hash) + getOutputSlot();
-    if (getDebugOpsCount() > 0) {
-      hash = (37 * hash) + DEBUG_OPS_FIELD_NUMBER;
-      hash = (53 * hash) + getDebugOpsList().hashCode();
-    }
-    if (getDebugUrlsCount() > 0) {
-      hash = (37 * hash) + DEBUG_URLS_FIELD_NUMBER;
-      hash = (53 * hash) + getDebugUrlsList().hashCode();
-    }
-    hash = (37 * hash) + TOLERATE_DEBUG_OP_CREATION_FAILURES_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashBoolean(
-        getTolerateDebugOpCreationFailures());
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.framework.DebugTensorWatch parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.DebugTensorWatch parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.DebugTensorWatch parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.DebugTensorWatch parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.DebugTensorWatch parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.DebugTensorWatch parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.DebugTensorWatch parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.DebugTensorWatch parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.DebugTensorWatch parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.DebugTensorWatch parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.DebugTensorWatch parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.DebugTensorWatch parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.framework.DebugTensorWatch prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * <pre>
-   * Option for watching a node in TensorFlow Debugger (tfdbg).
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.DebugTensorWatch}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.DebugTensorWatch)
-      org.tensorflow.proto.framework.DebugTensorWatchOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.DebugProtos.internal_static_tensorflow_DebugTensorWatch_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.DebugProtos.internal_static_tensorflow_DebugTensorWatch_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.DebugTensorWatch.class, org.tensorflow.proto.framework.DebugTensorWatch.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.framework.DebugTensorWatch.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      nodeName_ = "";
-
-      outputSlot_ = 0;
-
-      debugOps_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-      bitField0_ = (bitField0_ & ~0x00000001);
-      debugUrls_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-      bitField0_ = (bitField0_ & ~0x00000002);
-      tolerateDebugOpCreationFailures_ = false;
-
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.framework.DebugProtos.internal_static_tensorflow_DebugTensorWatch_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.DebugTensorWatch getDefaultInstanceForType() {
-      return org.tensorflow.proto.framework.DebugTensorWatch.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.DebugTensorWatch build() {
-      org.tensorflow.proto.framework.DebugTensorWatch result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.DebugTensorWatch buildPartial() {
-      org.tensorflow.proto.framework.DebugTensorWatch result = new org.tensorflow.proto.framework.DebugTensorWatch(this);
-      int from_bitField0_ = bitField0_;
-      result.nodeName_ = nodeName_;
-      result.outputSlot_ = outputSlot_;
-      if (((bitField0_ & 0x00000001) != 0)) {
-        debugOps_ = debugOps_.getUnmodifiableView();
-        bitField0_ = (bitField0_ & ~0x00000001);
-      }
-      result.debugOps_ = debugOps_;
-      if (((bitField0_ & 0x00000002) != 0)) {
-        debugUrls_ = debugUrls_.getUnmodifiableView();
-        bitField0_ = (bitField0_ & ~0x00000002);
-      }
-      result.debugUrls_ = debugUrls_;
-      result.tolerateDebugOpCreationFailures_ = tolerateDebugOpCreationFailures_;
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.framework.DebugTensorWatch) {
-        return mergeFrom((org.tensorflow.proto.framework.DebugTensorWatch)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.framework.DebugTensorWatch other) {
-      if (other == org.tensorflow.proto.framework.DebugTensorWatch.getDefaultInstance()) return this;
-      if (!other.getNodeName().isEmpty()) {
-        nodeName_ = other.nodeName_;
-        onChanged();
-      }
-      if (other.getOutputSlot() != 0) {
-        setOutputSlot(other.getOutputSlot());
-      }
-      if (!other.debugOps_.isEmpty()) {
-        if (debugOps_.isEmpty()) {
-          debugOps_ = other.debugOps_;
-          bitField0_ = (bitField0_ & ~0x00000001);
-        } else {
-          ensureDebugOpsIsMutable();
-          debugOps_.addAll(other.debugOps_);
-        }
-        onChanged();
-      }
-      if (!other.debugUrls_.isEmpty()) {
-        if (debugUrls_.isEmpty()) {
-          debugUrls_ = other.debugUrls_;
-          bitField0_ = (bitField0_ & ~0x00000002);
-        } else {
-          ensureDebugUrlsIsMutable();
-          debugUrls_.addAll(other.debugUrls_);
-        }
-        onChanged();
-      }
-      if (other.getTolerateDebugOpCreationFailures() != false) {
-        setTolerateDebugOpCreationFailures(other.getTolerateDebugOpCreationFailures());
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.framework.DebugTensorWatch parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.framework.DebugTensorWatch) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-    private int bitField0_;
-
-    private java.lang.Object nodeName_ = "";
-    /**
-     * <pre>
-     * Name of the node to watch.
-     * Use "*" for wildcard. But note: currently, regex is not supported in
-     * general.
-     * </pre>
-     *
-     * <code>string node_name = 1;</code>
-     */
-    public java.lang.String getNodeName() {
-      java.lang.Object ref = nodeName_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        nodeName_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Name of the node to watch.
-     * Use "*" for wildcard. But note: currently, regex is not supported in
-     * general.
-     * </pre>
-     *
-     * <code>string node_name = 1;</code>
-     */
-    public com.google.protobuf.ByteString
-        getNodeNameBytes() {
-      java.lang.Object ref = nodeName_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        nodeName_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Name of the node to watch.
-     * Use "*" for wildcard. But note: currently, regex is not supported in
-     * general.
-     * </pre>
-     *
-     * <code>string node_name = 1;</code>
-     */
-    public Builder setNodeName(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      nodeName_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Name of the node to watch.
-     * Use "*" for wildcard. But note: currently, regex is not supported in
-     * general.
-     * </pre>
-     *
-     * <code>string node_name = 1;</code>
-     */
-    public Builder clearNodeName() {
-      
-      nodeName_ = getDefaultInstance().getNodeName();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Name of the node to watch.
-     * Use "*" for wildcard. But note: currently, regex is not supported in
-     * general.
-     * </pre>
-     *
-     * <code>string node_name = 1;</code>
-     */
-    public Builder setNodeNameBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      nodeName_ = value;
-      onChanged();
-      return this;
-    }
-
-    private int outputSlot_ ;
-    /**
-     * <pre>
-     * Output slot to watch.
-     * The semantics of output_slot == -1 is that all outputs of the node
-     * will be watched (i.e., a wildcard).
-     * Other negative values of output_slot are invalid and will lead to
-     * errors currently.
-     * </pre>
-     *
-     * <code>int32 output_slot = 2;</code>
-     */
-    public int getOutputSlot() {
-      return outputSlot_;
-    }
-    /**
-     * <pre>
-     * Output slot to watch.
-     * The semantics of output_slot == -1 is that all outputs of the node
-     * will be watched (i.e., a wildcard).
-     * Other negative values of output_slot are invalid and will lead to
-     * errors currently.
-     * </pre>
-     *
-     * <code>int32 output_slot = 2;</code>
-     */
-    public Builder setOutputSlot(int value) {
-      
-      outputSlot_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Output slot to watch.
-     * The semantics of output_slot == -1 is that all outputs of the node
-     * will be watched (i.e., a wildcard).
-     * Other negative values of output_slot are invalid and will lead to
-     * errors currently.
-     * </pre>
-     *
-     * <code>int32 output_slot = 2;</code>
-     */
-    public Builder clearOutputSlot() {
-      
-      outputSlot_ = 0;
-      onChanged();
-      return this;
-    }
-
-    private com.google.protobuf.LazyStringList debugOps_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-    private void ensureDebugOpsIsMutable() {
-      if (!((bitField0_ & 0x00000001) != 0)) {
-        debugOps_ = new com.google.protobuf.LazyStringArrayList(debugOps_);
-        bitField0_ |= 0x00000001;
-       }
-    }
-    /**
-     * <pre>
-     * Name(s) of the debugging op(s).
-     * One or more than one probes on a tensor.
-     * e.g., {"DebugIdentity", "DebugNanCount"}
-     * </pre>
-     *
-     * <code>repeated string debug_ops = 3;</code>
-     */
-    public com.google.protobuf.ProtocolStringList
-        getDebugOpsList() {
-      return debugOps_.getUnmodifiableView();
-    }
-    /**
-     * <pre>
-     * Name(s) of the debugging op(s).
-     * One or more than one probes on a tensor.
-     * e.g., {"DebugIdentity", "DebugNanCount"}
-     * </pre>
-     *
-     * <code>repeated string debug_ops = 3;</code>
-     */
-    public int getDebugOpsCount() {
-      return debugOps_.size();
-    }
-    /**
-     * <pre>
-     * Name(s) of the debugging op(s).
-     * One or more than one probes on a tensor.
-     * e.g., {"DebugIdentity", "DebugNanCount"}
-     * </pre>
-     *
-     * <code>repeated string debug_ops = 3;</code>
-     */
-    public java.lang.String getDebugOps(int index) {
-      return debugOps_.get(index);
-    }
-    /**
-     * <pre>
-     * Name(s) of the debugging op(s).
-     * One or more than one probes on a tensor.
-     * e.g., {"DebugIdentity", "DebugNanCount"}
-     * </pre>
-     *
-     * <code>repeated string debug_ops = 3;</code>
-     */
-    public com.google.protobuf.ByteString
-        getDebugOpsBytes(int index) {
-      return debugOps_.getByteString(index);
-    }
-    /**
-     * <pre>
-     * Name(s) of the debugging op(s).
-     * One or more than one probes on a tensor.
-     * e.g., {"DebugIdentity", "DebugNanCount"}
-     * </pre>
-     *
-     * <code>repeated string debug_ops = 3;</code>
-     */
-    public Builder setDebugOps(
-        int index, java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  ensureDebugOpsIsMutable();
-      debugOps_.set(index, value);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Name(s) of the debugging op(s).
-     * One or more than one probes on a tensor.
-     * e.g., {"DebugIdentity", "DebugNanCount"}
-     * </pre>
-     *
-     * <code>repeated string debug_ops = 3;</code>
-     */
-    public Builder addDebugOps(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  ensureDebugOpsIsMutable();
-      debugOps_.add(value);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Name(s) of the debugging op(s).
-     * One or more than one probes on a tensor.
-     * e.g., {"DebugIdentity", "DebugNanCount"}
-     * </pre>
-     *
-     * <code>repeated string debug_ops = 3;</code>
-     */
-    public Builder addAllDebugOps(
-        java.lang.Iterable<java.lang.String> values) {
-      ensureDebugOpsIsMutable();
-      com.google.protobuf.AbstractMessageLite.Builder.addAll(
-          values, debugOps_);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Name(s) of the debugging op(s).
-     * One or more than one probes on a tensor.
-     * e.g., {"DebugIdentity", "DebugNanCount"}
-     * </pre>
-     *
-     * <code>repeated string debug_ops = 3;</code>
-     */
-    public Builder clearDebugOps() {
-      debugOps_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-      bitField0_ = (bitField0_ & ~0x00000001);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Name(s) of the debugging op(s).
-     * One or more than one probes on a tensor.
-     * e.g., {"DebugIdentity", "DebugNanCount"}
-     * </pre>
-     *
-     * <code>repeated string debug_ops = 3;</code>
-     */
-    public Builder addDebugOpsBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      ensureDebugOpsIsMutable();
-      debugOps_.add(value);
-      onChanged();
-      return this;
-    }
-
-    private com.google.protobuf.LazyStringList debugUrls_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-    private void ensureDebugUrlsIsMutable() {
-      if (!((bitField0_ & 0x00000002) != 0)) {
-        debugUrls_ = new com.google.protobuf.LazyStringArrayList(debugUrls_);
-        bitField0_ |= 0x00000002;
-       }
-    }
-    /**
-     * <pre>
-     * URL(s) for debug targets(s).
-     * Supported URL formats are:
-     *   - file:///foo/tfdbg_dump: Writes out Event content to file
-     *     /foo/tfdbg_dump.  Assumes all directories can be created if they don't
-     *     already exist.
-     *   - grpc://localhost:11011: Sends an RPC request to an EventListener
-     *     service running at localhost:11011 with the event.
-     *   - memcbk:///event_key: Routes tensors to clients using the
-     *     callback registered with the DebugCallbackRegistry for event_key.
-     * Each debug op listed in debug_ops will publish its output tensor (debug
-     * signal) to all URLs in debug_urls.
-     * N.B. Session::Run() supports concurrent invocations of the same inputs
-     * (feed keys), outputs and target nodes. If such concurrent invocations
-     * are to be debugged, the callers of Session::Run() must use distinct
-     * debug_urls to make sure that the streamed or dumped events do not overlap
-     * among the invocations.
-     * TODO(cais): More visible documentation of this in g3docs.
-     * </pre>
-     *
-     * <code>repeated string debug_urls = 4;</code>
-     */
-    public com.google.protobuf.ProtocolStringList
-        getDebugUrlsList() {
-      return debugUrls_.getUnmodifiableView();
-    }
-    /**
-     * <pre>
-     * URL(s) for debug targets(s).
-     * Supported URL formats are:
-     *   - file:///foo/tfdbg_dump: Writes out Event content to file
-     *     /foo/tfdbg_dump.  Assumes all directories can be created if they don't
-     *     already exist.
-     *   - grpc://localhost:11011: Sends an RPC request to an EventListener
-     *     service running at localhost:11011 with the event.
-     *   - memcbk:///event_key: Routes tensors to clients using the
-     *     callback registered with the DebugCallbackRegistry for event_key.
-     * Each debug op listed in debug_ops will publish its output tensor (debug
-     * signal) to all URLs in debug_urls.
-     * N.B. Session::Run() supports concurrent invocations of the same inputs
-     * (feed keys), outputs and target nodes. If such concurrent invocations
-     * are to be debugged, the callers of Session::Run() must use distinct
-     * debug_urls to make sure that the streamed or dumped events do not overlap
-     * among the invocations.
-     * TODO(cais): More visible documentation of this in g3docs.
-     * </pre>
-     *
-     * <code>repeated string debug_urls = 4;</code>
-     */
-    public int getDebugUrlsCount() {
-      return debugUrls_.size();
-    }
-    /**
-     * <pre>
-     * URL(s) for debug targets(s).
-     * Supported URL formats are:
-     *   - file:///foo/tfdbg_dump: Writes out Event content to file
-     *     /foo/tfdbg_dump.  Assumes all directories can be created if they don't
-     *     already exist.
-     *   - grpc://localhost:11011: Sends an RPC request to an EventListener
-     *     service running at localhost:11011 with the event.
-     *   - memcbk:///event_key: Routes tensors to clients using the
-     *     callback registered with the DebugCallbackRegistry for event_key.
-     * Each debug op listed in debug_ops will publish its output tensor (debug
-     * signal) to all URLs in debug_urls.
-     * N.B. Session::Run() supports concurrent invocations of the same inputs
-     * (feed keys), outputs and target nodes. If such concurrent invocations
-     * are to be debugged, the callers of Session::Run() must use distinct
-     * debug_urls to make sure that the streamed or dumped events do not overlap
-     * among the invocations.
-     * TODO(cais): More visible documentation of this in g3docs.
-     * </pre>
-     *
-     * <code>repeated string debug_urls = 4;</code>
-     */
-    public java.lang.String getDebugUrls(int index) {
-      return debugUrls_.get(index);
-    }
-    /**
-     * <pre>
-     * URL(s) for debug targets(s).
-     * Supported URL formats are:
-     *   - file:///foo/tfdbg_dump: Writes out Event content to file
-     *     /foo/tfdbg_dump.  Assumes all directories can be created if they don't
-     *     already exist.
-     *   - grpc://localhost:11011: Sends an RPC request to an EventListener
-     *     service running at localhost:11011 with the event.
-     *   - memcbk:///event_key: Routes tensors to clients using the
-     *     callback registered with the DebugCallbackRegistry for event_key.
-     * Each debug op listed in debug_ops will publish its output tensor (debug
-     * signal) to all URLs in debug_urls.
-     * N.B. Session::Run() supports concurrent invocations of the same inputs
-     * (feed keys), outputs and target nodes. If such concurrent invocations
-     * are to be debugged, the callers of Session::Run() must use distinct
-     * debug_urls to make sure that the streamed or dumped events do not overlap
-     * among the invocations.
-     * TODO(cais): More visible documentation of this in g3docs.
-     * </pre>
-     *
-     * <code>repeated string debug_urls = 4;</code>
-     */
-    public com.google.protobuf.ByteString
-        getDebugUrlsBytes(int index) {
-      return debugUrls_.getByteString(index);
-    }
-    /**
-     * <pre>
-     * URL(s) for debug targets(s).
-     * Supported URL formats are:
-     *   - file:///foo/tfdbg_dump: Writes out Event content to file
-     *     /foo/tfdbg_dump.  Assumes all directories can be created if they don't
-     *     already exist.
-     *   - grpc://localhost:11011: Sends an RPC request to an EventListener
-     *     service running at localhost:11011 with the event.
-     *   - memcbk:///event_key: Routes tensors to clients using the
-     *     callback registered with the DebugCallbackRegistry for event_key.
-     * Each debug op listed in debug_ops will publish its output tensor (debug
-     * signal) to all URLs in debug_urls.
-     * N.B. Session::Run() supports concurrent invocations of the same inputs
-     * (feed keys), outputs and target nodes. If such concurrent invocations
-     * are to be debugged, the callers of Session::Run() must use distinct
-     * debug_urls to make sure that the streamed or dumped events do not overlap
-     * among the invocations.
-     * TODO(cais): More visible documentation of this in g3docs.
-     * </pre>
-     *
-     * <code>repeated string debug_urls = 4;</code>
-     */
-    public Builder setDebugUrls(
-        int index, java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  ensureDebugUrlsIsMutable();
-      debugUrls_.set(index, value);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * URL(s) for debug targets(s).
-     * Supported URL formats are:
-     *   - file:///foo/tfdbg_dump: Writes out Event content to file
-     *     /foo/tfdbg_dump.  Assumes all directories can be created if they don't
-     *     already exist.
-     *   - grpc://localhost:11011: Sends an RPC request to an EventListener
-     *     service running at localhost:11011 with the event.
-     *   - memcbk:///event_key: Routes tensors to clients using the
-     *     callback registered with the DebugCallbackRegistry for event_key.
-     * Each debug op listed in debug_ops will publish its output tensor (debug
-     * signal) to all URLs in debug_urls.
-     * N.B. Session::Run() supports concurrent invocations of the same inputs
-     * (feed keys), outputs and target nodes. If such concurrent invocations
-     * are to be debugged, the callers of Session::Run() must use distinct
-     * debug_urls to make sure that the streamed or dumped events do not overlap
-     * among the invocations.
-     * TODO(cais): More visible documentation of this in g3docs.
-     * </pre>
-     *
-     * <code>repeated string debug_urls = 4;</code>
-     */
-    public Builder addDebugUrls(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  ensureDebugUrlsIsMutable();
-      debugUrls_.add(value);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * URL(s) for debug targets(s).
-     * Supported URL formats are:
-     *   - file:///foo/tfdbg_dump: Writes out Event content to file
-     *     /foo/tfdbg_dump.  Assumes all directories can be created if they don't
-     *     already exist.
-     *   - grpc://localhost:11011: Sends an RPC request to an EventListener
-     *     service running at localhost:11011 with the event.
-     *   - memcbk:///event_key: Routes tensors to clients using the
-     *     callback registered with the DebugCallbackRegistry for event_key.
-     * Each debug op listed in debug_ops will publish its output tensor (debug
-     * signal) to all URLs in debug_urls.
-     * N.B. Session::Run() supports concurrent invocations of the same inputs
-     * (feed keys), outputs and target nodes. If such concurrent invocations
-     * are to be debugged, the callers of Session::Run() must use distinct
-     * debug_urls to make sure that the streamed or dumped events do not overlap
-     * among the invocations.
-     * TODO(cais): More visible documentation of this in g3docs.
-     * </pre>
-     *
-     * <code>repeated string debug_urls = 4;</code>
-     */
-    public Builder addAllDebugUrls(
-        java.lang.Iterable<java.lang.String> values) {
-      ensureDebugUrlsIsMutable();
-      com.google.protobuf.AbstractMessageLite.Builder.addAll(
-          values, debugUrls_);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * URL(s) for debug targets(s).
-     * Supported URL formats are:
-     *   - file:///foo/tfdbg_dump: Writes out Event content to file
-     *     /foo/tfdbg_dump.  Assumes all directories can be created if they don't
-     *     already exist.
-     *   - grpc://localhost:11011: Sends an RPC request to an EventListener
-     *     service running at localhost:11011 with the event.
-     *   - memcbk:///event_key: Routes tensors to clients using the
-     *     callback registered with the DebugCallbackRegistry for event_key.
-     * Each debug op listed in debug_ops will publish its output tensor (debug
-     * signal) to all URLs in debug_urls.
-     * N.B. Session::Run() supports concurrent invocations of the same inputs
-     * (feed keys), outputs and target nodes. If such concurrent invocations
-     * are to be debugged, the callers of Session::Run() must use distinct
-     * debug_urls to make sure that the streamed or dumped events do not overlap
-     * among the invocations.
-     * TODO(cais): More visible documentation of this in g3docs.
-     * </pre>
-     *
-     * <code>repeated string debug_urls = 4;</code>
-     */
-    public Builder clearDebugUrls() {
-      debugUrls_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-      bitField0_ = (bitField0_ & ~0x00000002);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * URL(s) for debug targets(s).
-     * Supported URL formats are:
-     *   - file:///foo/tfdbg_dump: Writes out Event content to file
-     *     /foo/tfdbg_dump.  Assumes all directories can be created if they don't
-     *     already exist.
-     *   - grpc://localhost:11011: Sends an RPC request to an EventListener
-     *     service running at localhost:11011 with the event.
-     *   - memcbk:///event_key: Routes tensors to clients using the
-     *     callback registered with the DebugCallbackRegistry for event_key.
-     * Each debug op listed in debug_ops will publish its output tensor (debug
-     * signal) to all URLs in debug_urls.
-     * N.B. Session::Run() supports concurrent invocations of the same inputs
-     * (feed keys), outputs and target nodes. If such concurrent invocations
-     * are to be debugged, the callers of Session::Run() must use distinct
-     * debug_urls to make sure that the streamed or dumped events do not overlap
-     * among the invocations.
-     * TODO(cais): More visible documentation of this in g3docs.
-     * </pre>
-     *
-     * <code>repeated string debug_urls = 4;</code>
-     */
-    public Builder addDebugUrlsBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      ensureDebugUrlsIsMutable();
-      debugUrls_.add(value);
-      onChanged();
-      return this;
-    }
-
-    private boolean tolerateDebugOpCreationFailures_ ;
-    /**
-     * <pre>
-     * Do not error out if debug op creation fails (e.g., due to dtype
-     * incompatibility). Instead, just log the failure.
-     * </pre>
-     *
-     * <code>bool tolerate_debug_op_creation_failures = 5;</code>
-     */
-    public boolean getTolerateDebugOpCreationFailures() {
-      return tolerateDebugOpCreationFailures_;
-    }
-    /**
-     * <pre>
-     * Do not error out if debug op creation fails (e.g., due to dtype
-     * incompatibility). Instead, just log the failure.
-     * </pre>
-     *
-     * <code>bool tolerate_debug_op_creation_failures = 5;</code>
-     */
-    public Builder setTolerateDebugOpCreationFailures(boolean value) {
-      
-      tolerateDebugOpCreationFailures_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Do not error out if debug op creation fails (e.g., due to dtype
-     * incompatibility). Instead, just log the failure.
-     * </pre>
-     *
-     * <code>bool tolerate_debug_op_creation_failures = 5;</code>
-     */
-    public Builder clearTolerateDebugOpCreationFailures() {
-      
-      tolerateDebugOpCreationFailures_ = false;
-      onChanged();
-      return this;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.DebugTensorWatch)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.DebugTensorWatch)
-  private static final org.tensorflow.proto.framework.DebugTensorWatch DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.DebugTensorWatch();
-  }
-
-  public static org.tensorflow.proto.framework.DebugTensorWatch getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<DebugTensorWatch>
-      PARSER = new com.google.protobuf.AbstractParser<DebugTensorWatch>() {
-    @java.lang.Override
-    public DebugTensorWatch parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new DebugTensorWatch(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<DebugTensorWatch> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<DebugTensorWatch> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.framework.DebugTensorWatch getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DebugTensorWatchOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DebugTensorWatchOrBuilder.java
deleted file mode 100644
index cce3536c5d1..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DebugTensorWatchOrBuilder.java
+++ /dev/null
@@ -1,196 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/debug.proto
-
-package org.tensorflow.proto.framework;
-
-public interface DebugTensorWatchOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.DebugTensorWatch)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <pre>
-   * Name of the node to watch.
-   * Use "*" for wildcard. But note: currently, regex is not supported in
-   * general.
-   * </pre>
-   *
-   * <code>string node_name = 1;</code>
-   */
-  java.lang.String getNodeName();
-  /**
-   * <pre>
-   * Name of the node to watch.
-   * Use "*" for wildcard. But note: currently, regex is not supported in
-   * general.
-   * </pre>
-   *
-   * <code>string node_name = 1;</code>
-   */
-  com.google.protobuf.ByteString
-      getNodeNameBytes();
-
-  /**
-   * <pre>
-   * Output slot to watch.
-   * The semantics of output_slot == -1 is that all outputs of the node
-   * will be watched (i.e., a wildcard).
-   * Other negative values of output_slot are invalid and will lead to
-   * errors currently.
-   * </pre>
-   *
-   * <code>int32 output_slot = 2;</code>
-   */
-  int getOutputSlot();
-
-  /**
-   * <pre>
-   * Name(s) of the debugging op(s).
-   * One or more than one probes on a tensor.
-   * e.g., {"DebugIdentity", "DebugNanCount"}
-   * </pre>
-   *
-   * <code>repeated string debug_ops = 3;</code>
-   */
-  java.util.List<java.lang.String>
-      getDebugOpsList();
-  /**
-   * <pre>
-   * Name(s) of the debugging op(s).
-   * One or more than one probes on a tensor.
-   * e.g., {"DebugIdentity", "DebugNanCount"}
-   * </pre>
-   *
-   * <code>repeated string debug_ops = 3;</code>
-   */
-  int getDebugOpsCount();
-  /**
-   * <pre>
-   * Name(s) of the debugging op(s).
-   * One or more than one probes on a tensor.
-   * e.g., {"DebugIdentity", "DebugNanCount"}
-   * </pre>
-   *
-   * <code>repeated string debug_ops = 3;</code>
-   */
-  java.lang.String getDebugOps(int index);
-  /**
-   * <pre>
-   * Name(s) of the debugging op(s).
-   * One or more than one probes on a tensor.
-   * e.g., {"DebugIdentity", "DebugNanCount"}
-   * </pre>
-   *
-   * <code>repeated string debug_ops = 3;</code>
-   */
-  com.google.protobuf.ByteString
-      getDebugOpsBytes(int index);
-
-  /**
-   * <pre>
-   * URL(s) for debug targets(s).
-   * Supported URL formats are:
-   *   - file:///foo/tfdbg_dump: Writes out Event content to file
-   *     /foo/tfdbg_dump.  Assumes all directories can be created if they don't
-   *     already exist.
-   *   - grpc://localhost:11011: Sends an RPC request to an EventListener
-   *     service running at localhost:11011 with the event.
-   *   - memcbk:///event_key: Routes tensors to clients using the
-   *     callback registered with the DebugCallbackRegistry for event_key.
-   * Each debug op listed in debug_ops will publish its output tensor (debug
-   * signal) to all URLs in debug_urls.
-   * N.B. Session::Run() supports concurrent invocations of the same inputs
-   * (feed keys), outputs and target nodes. If such concurrent invocations
-   * are to be debugged, the callers of Session::Run() must use distinct
-   * debug_urls to make sure that the streamed or dumped events do not overlap
-   * among the invocations.
-   * TODO(cais): More visible documentation of this in g3docs.
-   * </pre>
-   *
-   * <code>repeated string debug_urls = 4;</code>
-   */
-  java.util.List<java.lang.String>
-      getDebugUrlsList();
-  /**
-   * <pre>
-   * URL(s) for debug targets(s).
-   * Supported URL formats are:
-   *   - file:///foo/tfdbg_dump: Writes out Event content to file
-   *     /foo/tfdbg_dump.  Assumes all directories can be created if they don't
-   *     already exist.
-   *   - grpc://localhost:11011: Sends an RPC request to an EventListener
-   *     service running at localhost:11011 with the event.
-   *   - memcbk:///event_key: Routes tensors to clients using the
-   *     callback registered with the DebugCallbackRegistry for event_key.
-   * Each debug op listed in debug_ops will publish its output tensor (debug
-   * signal) to all URLs in debug_urls.
-   * N.B. Session::Run() supports concurrent invocations of the same inputs
-   * (feed keys), outputs and target nodes. If such concurrent invocations
-   * are to be debugged, the callers of Session::Run() must use distinct
-   * debug_urls to make sure that the streamed or dumped events do not overlap
-   * among the invocations.
-   * TODO(cais): More visible documentation of this in g3docs.
-   * </pre>
-   *
-   * <code>repeated string debug_urls = 4;</code>
-   */
-  int getDebugUrlsCount();
-  /**
-   * <pre>
-   * URL(s) for debug targets(s).
-   * Supported URL formats are:
-   *   - file:///foo/tfdbg_dump: Writes out Event content to file
-   *     /foo/tfdbg_dump.  Assumes all directories can be created if they don't
-   *     already exist.
-   *   - grpc://localhost:11011: Sends an RPC request to an EventListener
-   *     service running at localhost:11011 with the event.
-   *   - memcbk:///event_key: Routes tensors to clients using the
-   *     callback registered with the DebugCallbackRegistry for event_key.
-   * Each debug op listed in debug_ops will publish its output tensor (debug
-   * signal) to all URLs in debug_urls.
-   * N.B. Session::Run() supports concurrent invocations of the same inputs
-   * (feed keys), outputs and target nodes. If such concurrent invocations
-   * are to be debugged, the callers of Session::Run() must use distinct
-   * debug_urls to make sure that the streamed or dumped events do not overlap
-   * among the invocations.
-   * TODO(cais): More visible documentation of this in g3docs.
-   * </pre>
-   *
-   * <code>repeated string debug_urls = 4;</code>
-   */
-  java.lang.String getDebugUrls(int index);
-  /**
-   * <pre>
-   * URL(s) for debug targets(s).
-   * Supported URL formats are:
-   *   - file:///foo/tfdbg_dump: Writes out Event content to file
-   *     /foo/tfdbg_dump.  Assumes all directories can be created if they don't
-   *     already exist.
-   *   - grpc://localhost:11011: Sends an RPC request to an EventListener
-   *     service running at localhost:11011 with the event.
-   *   - memcbk:///event_key: Routes tensors to clients using the
-   *     callback registered with the DebugCallbackRegistry for event_key.
-   * Each debug op listed in debug_ops will publish its output tensor (debug
-   * signal) to all URLs in debug_urls.
-   * N.B. Session::Run() supports concurrent invocations of the same inputs
-   * (feed keys), outputs and target nodes. If such concurrent invocations
-   * are to be debugged, the callers of Session::Run() must use distinct
-   * debug_urls to make sure that the streamed or dumped events do not overlap
-   * among the invocations.
-   * TODO(cais): More visible documentation of this in g3docs.
-   * </pre>
-   *
-   * <code>repeated string debug_urls = 4;</code>
-   */
-  com.google.protobuf.ByteString
-      getDebugUrlsBytes(int index);
-
-  /**
-   * <pre>
-   * Do not error out if debug op creation fails (e.g., due to dtype
-   * incompatibility). Instead, just log the failure.
-   * </pre>
-   *
-   * <code>bool tolerate_debug_op_creation_failures = 5;</code>
-   */
-  boolean getTolerateDebugOpCreationFailures();
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DebuggedSourceFile.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DebuggedSourceFile.java
deleted file mode 100644
index d0752d9022b..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DebuggedSourceFile.java
+++ /dev/null
@@ -1,1102 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/debug.proto
-
-package org.tensorflow.proto.framework;
-
-/**
- * Protobuf type {@code tensorflow.DebuggedSourceFile}
- */
-public  final class DebuggedSourceFile extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.DebuggedSourceFile)
-    DebuggedSourceFileOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use DebuggedSourceFile.newBuilder() to construct.
-  private DebuggedSourceFile(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private DebuggedSourceFile() {
-    host_ = "";
-    filePath_ = "";
-    lines_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new DebuggedSourceFile();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private DebuggedSourceFile(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    int mutable_bitField0_ = 0;
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 10: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            host_ = s;
-            break;
-          }
-          case 18: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            filePath_ = s;
-            break;
-          }
-          case 24: {
-
-            lastModified_ = input.readInt64();
-            break;
-          }
-          case 32: {
-
-            bytes_ = input.readInt64();
-            break;
-          }
-          case 42: {
-            java.lang.String s = input.readStringRequireUtf8();
-            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
-              lines_ = new com.google.protobuf.LazyStringArrayList();
-              mutable_bitField0_ |= 0x00000001;
-            }
-            lines_.add(s);
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      if (((mutable_bitField0_ & 0x00000001) != 0)) {
-        lines_ = lines_.getUnmodifiableView();
-      }
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.framework.DebugProtos.internal_static_tensorflow_DebuggedSourceFile_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.framework.DebugProtos.internal_static_tensorflow_DebuggedSourceFile_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.framework.DebuggedSourceFile.class, org.tensorflow.proto.framework.DebuggedSourceFile.Builder.class);
-  }
-
-  public static final int HOST_FIELD_NUMBER = 1;
-  private volatile java.lang.Object host_;
-  /**
-   * <pre>
-   * The host name on which a source code file is located.
-   * </pre>
-   *
-   * <code>string host = 1;</code>
-   */
-  public java.lang.String getHost() {
-    java.lang.Object ref = host_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      host_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * The host name on which a source code file is located.
-   * </pre>
-   *
-   * <code>string host = 1;</code>
-   */
-  public com.google.protobuf.ByteString
-      getHostBytes() {
-    java.lang.Object ref = host_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      host_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int FILE_PATH_FIELD_NUMBER = 2;
-  private volatile java.lang.Object filePath_;
-  /**
-   * <pre>
-   * Path to the source code file.
-   * </pre>
-   *
-   * <code>string file_path = 2;</code>
-   */
-  public java.lang.String getFilePath() {
-    java.lang.Object ref = filePath_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      filePath_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * Path to the source code file.
-   * </pre>
-   *
-   * <code>string file_path = 2;</code>
-   */
-  public com.google.protobuf.ByteString
-      getFilePathBytes() {
-    java.lang.Object ref = filePath_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      filePath_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int LAST_MODIFIED_FIELD_NUMBER = 3;
-  private long lastModified_;
-  /**
-   * <pre>
-   * The timestamp at which the source code file is last modified.
-   * </pre>
-   *
-   * <code>int64 last_modified = 3;</code>
-   */
-  public long getLastModified() {
-    return lastModified_;
-  }
-
-  public static final int BYTES_FIELD_NUMBER = 4;
-  private long bytes_;
-  /**
-   * <pre>
-   * Byte size of the file.
-   * </pre>
-   *
-   * <code>int64 bytes = 4;</code>
-   */
-  public long getBytes() {
-    return bytes_;
-  }
-
-  public static final int LINES_FIELD_NUMBER = 5;
-  private com.google.protobuf.LazyStringList lines_;
-  /**
-   * <pre>
-   * Line-by-line content of the source code file.
-   * </pre>
-   *
-   * <code>repeated string lines = 5;</code>
-   */
-  public com.google.protobuf.ProtocolStringList
-      getLinesList() {
-    return lines_;
-  }
-  /**
-   * <pre>
-   * Line-by-line content of the source code file.
-   * </pre>
-   *
-   * <code>repeated string lines = 5;</code>
-   */
-  public int getLinesCount() {
-    return lines_.size();
-  }
-  /**
-   * <pre>
-   * Line-by-line content of the source code file.
-   * </pre>
-   *
-   * <code>repeated string lines = 5;</code>
-   */
-  public java.lang.String getLines(int index) {
-    return lines_.get(index);
-  }
-  /**
-   * <pre>
-   * Line-by-line content of the source code file.
-   * </pre>
-   *
-   * <code>repeated string lines = 5;</code>
-   */
-  public com.google.protobuf.ByteString
-      getLinesBytes(int index) {
-    return lines_.getByteString(index);
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    if (!getHostBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 1, host_);
-    }
-    if (!getFilePathBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 2, filePath_);
-    }
-    if (lastModified_ != 0L) {
-      output.writeInt64(3, lastModified_);
-    }
-    if (bytes_ != 0L) {
-      output.writeInt64(4, bytes_);
-    }
-    for (int i = 0; i < lines_.size(); i++) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 5, lines_.getRaw(i));
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (!getHostBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, host_);
-    }
-    if (!getFilePathBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(2, filePath_);
-    }
-    if (lastModified_ != 0L) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt64Size(3, lastModified_);
-    }
-    if (bytes_ != 0L) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt64Size(4, bytes_);
-    }
-    {
-      int dataSize = 0;
-      for (int i = 0; i < lines_.size(); i++) {
-        dataSize += computeStringSizeNoTag(lines_.getRaw(i));
-      }
-      size += dataSize;
-      size += 1 * getLinesList().size();
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.framework.DebuggedSourceFile)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.framework.DebuggedSourceFile other = (org.tensorflow.proto.framework.DebuggedSourceFile) obj;
-
-    if (!getHost()
-        .equals(other.getHost())) return false;
-    if (!getFilePath()
-        .equals(other.getFilePath())) return false;
-    if (getLastModified()
-        != other.getLastModified()) return false;
-    if (getBytes()
-        != other.getBytes()) return false;
-    if (!getLinesList()
-        .equals(other.getLinesList())) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    hash = (37 * hash) + HOST_FIELD_NUMBER;
-    hash = (53 * hash) + getHost().hashCode();
-    hash = (37 * hash) + FILE_PATH_FIELD_NUMBER;
-    hash = (53 * hash) + getFilePath().hashCode();
-    hash = (37 * hash) + LAST_MODIFIED_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-        getLastModified());
-    hash = (37 * hash) + BYTES_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-        getBytes());
-    if (getLinesCount() > 0) {
-      hash = (37 * hash) + LINES_FIELD_NUMBER;
-      hash = (53 * hash) + getLinesList().hashCode();
-    }
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.framework.DebuggedSourceFile parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.DebuggedSourceFile parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.DebuggedSourceFile parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.DebuggedSourceFile parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.DebuggedSourceFile parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.DebuggedSourceFile parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.DebuggedSourceFile parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.DebuggedSourceFile parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.DebuggedSourceFile parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.DebuggedSourceFile parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.DebuggedSourceFile parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.DebuggedSourceFile parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.framework.DebuggedSourceFile prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * Protobuf type {@code tensorflow.DebuggedSourceFile}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.DebuggedSourceFile)
-      org.tensorflow.proto.framework.DebuggedSourceFileOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.DebugProtos.internal_static_tensorflow_DebuggedSourceFile_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.DebugProtos.internal_static_tensorflow_DebuggedSourceFile_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.DebuggedSourceFile.class, org.tensorflow.proto.framework.DebuggedSourceFile.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.framework.DebuggedSourceFile.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      host_ = "";
-
-      filePath_ = "";
-
-      lastModified_ = 0L;
-
-      bytes_ = 0L;
-
-      lines_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-      bitField0_ = (bitField0_ & ~0x00000001);
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.framework.DebugProtos.internal_static_tensorflow_DebuggedSourceFile_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.DebuggedSourceFile getDefaultInstanceForType() {
-      return org.tensorflow.proto.framework.DebuggedSourceFile.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.DebuggedSourceFile build() {
-      org.tensorflow.proto.framework.DebuggedSourceFile result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.DebuggedSourceFile buildPartial() {
-      org.tensorflow.proto.framework.DebuggedSourceFile result = new org.tensorflow.proto.framework.DebuggedSourceFile(this);
-      int from_bitField0_ = bitField0_;
-      result.host_ = host_;
-      result.filePath_ = filePath_;
-      result.lastModified_ = lastModified_;
-      result.bytes_ = bytes_;
-      if (((bitField0_ & 0x00000001) != 0)) {
-        lines_ = lines_.getUnmodifiableView();
-        bitField0_ = (bitField0_ & ~0x00000001);
-      }
-      result.lines_ = lines_;
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.framework.DebuggedSourceFile) {
-        return mergeFrom((org.tensorflow.proto.framework.DebuggedSourceFile)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.framework.DebuggedSourceFile other) {
-      if (other == org.tensorflow.proto.framework.DebuggedSourceFile.getDefaultInstance()) return this;
-      if (!other.getHost().isEmpty()) {
-        host_ = other.host_;
-        onChanged();
-      }
-      if (!other.getFilePath().isEmpty()) {
-        filePath_ = other.filePath_;
-        onChanged();
-      }
-      if (other.getLastModified() != 0L) {
-        setLastModified(other.getLastModified());
-      }
-      if (other.getBytes() != 0L) {
-        setBytes(other.getBytes());
-      }
-      if (!other.lines_.isEmpty()) {
-        if (lines_.isEmpty()) {
-          lines_ = other.lines_;
-          bitField0_ = (bitField0_ & ~0x00000001);
-        } else {
-          ensureLinesIsMutable();
-          lines_.addAll(other.lines_);
-        }
-        onChanged();
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.framework.DebuggedSourceFile parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.framework.DebuggedSourceFile) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-    private int bitField0_;
-
-    private java.lang.Object host_ = "";
-    /**
-     * <pre>
-     * The host name on which a source code file is located.
-     * </pre>
-     *
-     * <code>string host = 1;</code>
-     */
-    public java.lang.String getHost() {
-      java.lang.Object ref = host_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        host_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * The host name on which a source code file is located.
-     * </pre>
-     *
-     * <code>string host = 1;</code>
-     */
-    public com.google.protobuf.ByteString
-        getHostBytes() {
-      java.lang.Object ref = host_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        host_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * The host name on which a source code file is located.
-     * </pre>
-     *
-     * <code>string host = 1;</code>
-     */
-    public Builder setHost(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      host_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * The host name on which a source code file is located.
-     * </pre>
-     *
-     * <code>string host = 1;</code>
-     */
-    public Builder clearHost() {
-      
-      host_ = getDefaultInstance().getHost();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * The host name on which a source code file is located.
-     * </pre>
-     *
-     * <code>string host = 1;</code>
-     */
-    public Builder setHostBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      host_ = value;
-      onChanged();
-      return this;
-    }
-
-    private java.lang.Object filePath_ = "";
-    /**
-     * <pre>
-     * Path to the source code file.
-     * </pre>
-     *
-     * <code>string file_path = 2;</code>
-     */
-    public java.lang.String getFilePath() {
-      java.lang.Object ref = filePath_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        filePath_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Path to the source code file.
-     * </pre>
-     *
-     * <code>string file_path = 2;</code>
-     */
-    public com.google.protobuf.ByteString
-        getFilePathBytes() {
-      java.lang.Object ref = filePath_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        filePath_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Path to the source code file.
-     * </pre>
-     *
-     * <code>string file_path = 2;</code>
-     */
-    public Builder setFilePath(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      filePath_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Path to the source code file.
-     * </pre>
-     *
-     * <code>string file_path = 2;</code>
-     */
-    public Builder clearFilePath() {
-      
-      filePath_ = getDefaultInstance().getFilePath();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Path to the source code file.
-     * </pre>
-     *
-     * <code>string file_path = 2;</code>
-     */
-    public Builder setFilePathBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      filePath_ = value;
-      onChanged();
-      return this;
-    }
-
-    private long lastModified_ ;
-    /**
-     * <pre>
-     * The timestamp at which the source code file is last modified.
-     * </pre>
-     *
-     * <code>int64 last_modified = 3;</code>
-     */
-    public long getLastModified() {
-      return lastModified_;
-    }
-    /**
-     * <pre>
-     * The timestamp at which the source code file is last modified.
-     * </pre>
-     *
-     * <code>int64 last_modified = 3;</code>
-     */
-    public Builder setLastModified(long value) {
-      
-      lastModified_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * The timestamp at which the source code file is last modified.
-     * </pre>
-     *
-     * <code>int64 last_modified = 3;</code>
-     */
-    public Builder clearLastModified() {
-      
-      lastModified_ = 0L;
-      onChanged();
-      return this;
-    }
-
-    private long bytes_ ;
-    /**
-     * <pre>
-     * Byte size of the file.
-     * </pre>
-     *
-     * <code>int64 bytes = 4;</code>
-     */
-    public long getBytes() {
-      return bytes_;
-    }
-    /**
-     * <pre>
-     * Byte size of the file.
-     * </pre>
-     *
-     * <code>int64 bytes = 4;</code>
-     */
-    public Builder setBytes(long value) {
-      
-      bytes_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Byte size of the file.
-     * </pre>
-     *
-     * <code>int64 bytes = 4;</code>
-     */
-    public Builder clearBytes() {
-      
-      bytes_ = 0L;
-      onChanged();
-      return this;
-    }
-
-    private com.google.protobuf.LazyStringList lines_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-    private void ensureLinesIsMutable() {
-      if (!((bitField0_ & 0x00000001) != 0)) {
-        lines_ = new com.google.protobuf.LazyStringArrayList(lines_);
-        bitField0_ |= 0x00000001;
-       }
-    }
-    /**
-     * <pre>
-     * Line-by-line content of the source code file.
-     * </pre>
-     *
-     * <code>repeated string lines = 5;</code>
-     */
-    public com.google.protobuf.ProtocolStringList
-        getLinesList() {
-      return lines_.getUnmodifiableView();
-    }
-    /**
-     * <pre>
-     * Line-by-line content of the source code file.
-     * </pre>
-     *
-     * <code>repeated string lines = 5;</code>
-     */
-    public int getLinesCount() {
-      return lines_.size();
-    }
-    /**
-     * <pre>
-     * Line-by-line content of the source code file.
-     * </pre>
-     *
-     * <code>repeated string lines = 5;</code>
-     */
-    public java.lang.String getLines(int index) {
-      return lines_.get(index);
-    }
-    /**
-     * <pre>
-     * Line-by-line content of the source code file.
-     * </pre>
-     *
-     * <code>repeated string lines = 5;</code>
-     */
-    public com.google.protobuf.ByteString
-        getLinesBytes(int index) {
-      return lines_.getByteString(index);
-    }
-    /**
-     * <pre>
-     * Line-by-line content of the source code file.
-     * </pre>
-     *
-     * <code>repeated string lines = 5;</code>
-     */
-    public Builder setLines(
-        int index, java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  ensureLinesIsMutable();
-      lines_.set(index, value);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Line-by-line content of the source code file.
-     * </pre>
-     *
-     * <code>repeated string lines = 5;</code>
-     */
-    public Builder addLines(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  ensureLinesIsMutable();
-      lines_.add(value);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Line-by-line content of the source code file.
-     * </pre>
-     *
-     * <code>repeated string lines = 5;</code>
-     */
-    public Builder addAllLines(
-        java.lang.Iterable<java.lang.String> values) {
-      ensureLinesIsMutable();
-      com.google.protobuf.AbstractMessageLite.Builder.addAll(
-          values, lines_);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Line-by-line content of the source code file.
-     * </pre>
-     *
-     * <code>repeated string lines = 5;</code>
-     */
-    public Builder clearLines() {
-      lines_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-      bitField0_ = (bitField0_ & ~0x00000001);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Line-by-line content of the source code file.
-     * </pre>
-     *
-     * <code>repeated string lines = 5;</code>
-     */
-    public Builder addLinesBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      ensureLinesIsMutable();
-      lines_.add(value);
-      onChanged();
-      return this;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.DebuggedSourceFile)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.DebuggedSourceFile)
-  private static final org.tensorflow.proto.framework.DebuggedSourceFile DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.DebuggedSourceFile();
-  }
-
-  public static org.tensorflow.proto.framework.DebuggedSourceFile getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<DebuggedSourceFile>
-      PARSER = new com.google.protobuf.AbstractParser<DebuggedSourceFile>() {
-    @java.lang.Override
-    public DebuggedSourceFile parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new DebuggedSourceFile(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<DebuggedSourceFile> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<DebuggedSourceFile> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.framework.DebuggedSourceFile getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DebuggedSourceFileOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DebuggedSourceFileOrBuilder.java
deleted file mode 100644
index 1a855dc8a07..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DebuggedSourceFileOrBuilder.java
+++ /dev/null
@@ -1,98 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/debug.proto
-
-package org.tensorflow.proto.framework;
-
-public interface DebuggedSourceFileOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.DebuggedSourceFile)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <pre>
-   * The host name on which a source code file is located.
-   * </pre>
-   *
-   * <code>string host = 1;</code>
-   */
-  java.lang.String getHost();
-  /**
-   * <pre>
-   * The host name on which a source code file is located.
-   * </pre>
-   *
-   * <code>string host = 1;</code>
-   */
-  com.google.protobuf.ByteString
-      getHostBytes();
-
-  /**
-   * <pre>
-   * Path to the source code file.
-   * </pre>
-   *
-   * <code>string file_path = 2;</code>
-   */
-  java.lang.String getFilePath();
-  /**
-   * <pre>
-   * Path to the source code file.
-   * </pre>
-   *
-   * <code>string file_path = 2;</code>
-   */
-  com.google.protobuf.ByteString
-      getFilePathBytes();
-
-  /**
-   * <pre>
-   * The timestamp at which the source code file is last modified.
-   * </pre>
-   *
-   * <code>int64 last_modified = 3;</code>
-   */
-  long getLastModified();
-
-  /**
-   * <pre>
-   * Byte size of the file.
-   * </pre>
-   *
-   * <code>int64 bytes = 4;</code>
-   */
-  long getBytes();
-
-  /**
-   * <pre>
-   * Line-by-line content of the source code file.
-   * </pre>
-   *
-   * <code>repeated string lines = 5;</code>
-   */
-  java.util.List<java.lang.String>
-      getLinesList();
-  /**
-   * <pre>
-   * Line-by-line content of the source code file.
-   * </pre>
-   *
-   * <code>repeated string lines = 5;</code>
-   */
-  int getLinesCount();
-  /**
-   * <pre>
-   * Line-by-line content of the source code file.
-   * </pre>
-   *
-   * <code>repeated string lines = 5;</code>
-   */
-  java.lang.String getLines(int index);
-  /**
-   * <pre>
-   * Line-by-line content of the source code file.
-   * </pre>
-   *
-   * <code>repeated string lines = 5;</code>
-   */
-  com.google.protobuf.ByteString
-      getLinesBytes(int index);
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DebuggedSourceFiles.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DebuggedSourceFiles.java
deleted file mode 100644
index c5d014b4bb6..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DebuggedSourceFiles.java
+++ /dev/null
@@ -1,857 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/debug.proto
-
-package org.tensorflow.proto.framework;
-
-/**
- * Protobuf type {@code tensorflow.DebuggedSourceFiles}
- */
-public  final class DebuggedSourceFiles extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.DebuggedSourceFiles)
-    DebuggedSourceFilesOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use DebuggedSourceFiles.newBuilder() to construct.
-  private DebuggedSourceFiles(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private DebuggedSourceFiles() {
-    sourceFiles_ = java.util.Collections.emptyList();
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new DebuggedSourceFiles();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private DebuggedSourceFiles(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    int mutable_bitField0_ = 0;
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 10: {
-            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
-              sourceFiles_ = new java.util.ArrayList<org.tensorflow.proto.framework.DebuggedSourceFile>();
-              mutable_bitField0_ |= 0x00000001;
-            }
-            sourceFiles_.add(
-                input.readMessage(org.tensorflow.proto.framework.DebuggedSourceFile.parser(), extensionRegistry));
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      if (((mutable_bitField0_ & 0x00000001) != 0)) {
-        sourceFiles_ = java.util.Collections.unmodifiableList(sourceFiles_);
-      }
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.framework.DebugProtos.internal_static_tensorflow_DebuggedSourceFiles_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.framework.DebugProtos.internal_static_tensorflow_DebuggedSourceFiles_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.framework.DebuggedSourceFiles.class, org.tensorflow.proto.framework.DebuggedSourceFiles.Builder.class);
-  }
-
-  public static final int SOURCE_FILES_FIELD_NUMBER = 1;
-  private java.util.List<org.tensorflow.proto.framework.DebuggedSourceFile> sourceFiles_;
-  /**
-   * <pre>
-   * A collection of source code files.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.DebuggedSourceFile source_files = 1;</code>
-   */
-  public java.util.List<org.tensorflow.proto.framework.DebuggedSourceFile> getSourceFilesList() {
-    return sourceFiles_;
-  }
-  /**
-   * <pre>
-   * A collection of source code files.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.DebuggedSourceFile source_files = 1;</code>
-   */
-  public java.util.List<? extends org.tensorflow.proto.framework.DebuggedSourceFileOrBuilder> 
-      getSourceFilesOrBuilderList() {
-    return sourceFiles_;
-  }
-  /**
-   * <pre>
-   * A collection of source code files.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.DebuggedSourceFile source_files = 1;</code>
-   */
-  public int getSourceFilesCount() {
-    return sourceFiles_.size();
-  }
-  /**
-   * <pre>
-   * A collection of source code files.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.DebuggedSourceFile source_files = 1;</code>
-   */
-  public org.tensorflow.proto.framework.DebuggedSourceFile getSourceFiles(int index) {
-    return sourceFiles_.get(index);
-  }
-  /**
-   * <pre>
-   * A collection of source code files.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.DebuggedSourceFile source_files = 1;</code>
-   */
-  public org.tensorflow.proto.framework.DebuggedSourceFileOrBuilder getSourceFilesOrBuilder(
-      int index) {
-    return sourceFiles_.get(index);
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    for (int i = 0; i < sourceFiles_.size(); i++) {
-      output.writeMessage(1, sourceFiles_.get(i));
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    for (int i = 0; i < sourceFiles_.size(); i++) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(1, sourceFiles_.get(i));
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.framework.DebuggedSourceFiles)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.framework.DebuggedSourceFiles other = (org.tensorflow.proto.framework.DebuggedSourceFiles) obj;
-
-    if (!getSourceFilesList()
-        .equals(other.getSourceFilesList())) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    if (getSourceFilesCount() > 0) {
-      hash = (37 * hash) + SOURCE_FILES_FIELD_NUMBER;
-      hash = (53 * hash) + getSourceFilesList().hashCode();
-    }
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.framework.DebuggedSourceFiles parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.DebuggedSourceFiles parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.DebuggedSourceFiles parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.DebuggedSourceFiles parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.DebuggedSourceFiles parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.DebuggedSourceFiles parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.DebuggedSourceFiles parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.DebuggedSourceFiles parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.DebuggedSourceFiles parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.DebuggedSourceFiles parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.DebuggedSourceFiles parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.DebuggedSourceFiles parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.framework.DebuggedSourceFiles prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * Protobuf type {@code tensorflow.DebuggedSourceFiles}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.DebuggedSourceFiles)
-      org.tensorflow.proto.framework.DebuggedSourceFilesOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.DebugProtos.internal_static_tensorflow_DebuggedSourceFiles_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.DebugProtos.internal_static_tensorflow_DebuggedSourceFiles_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.DebuggedSourceFiles.class, org.tensorflow.proto.framework.DebuggedSourceFiles.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.framework.DebuggedSourceFiles.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-        getSourceFilesFieldBuilder();
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      if (sourceFilesBuilder_ == null) {
-        sourceFiles_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000001);
-      } else {
-        sourceFilesBuilder_.clear();
-      }
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.framework.DebugProtos.internal_static_tensorflow_DebuggedSourceFiles_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.DebuggedSourceFiles getDefaultInstanceForType() {
-      return org.tensorflow.proto.framework.DebuggedSourceFiles.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.DebuggedSourceFiles build() {
-      org.tensorflow.proto.framework.DebuggedSourceFiles result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.DebuggedSourceFiles buildPartial() {
-      org.tensorflow.proto.framework.DebuggedSourceFiles result = new org.tensorflow.proto.framework.DebuggedSourceFiles(this);
-      int from_bitField0_ = bitField0_;
-      if (sourceFilesBuilder_ == null) {
-        if (((bitField0_ & 0x00000001) != 0)) {
-          sourceFiles_ = java.util.Collections.unmodifiableList(sourceFiles_);
-          bitField0_ = (bitField0_ & ~0x00000001);
-        }
-        result.sourceFiles_ = sourceFiles_;
-      } else {
-        result.sourceFiles_ = sourceFilesBuilder_.build();
-      }
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.framework.DebuggedSourceFiles) {
-        return mergeFrom((org.tensorflow.proto.framework.DebuggedSourceFiles)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.framework.DebuggedSourceFiles other) {
-      if (other == org.tensorflow.proto.framework.DebuggedSourceFiles.getDefaultInstance()) return this;
-      if (sourceFilesBuilder_ == null) {
-        if (!other.sourceFiles_.isEmpty()) {
-          if (sourceFiles_.isEmpty()) {
-            sourceFiles_ = other.sourceFiles_;
-            bitField0_ = (bitField0_ & ~0x00000001);
-          } else {
-            ensureSourceFilesIsMutable();
-            sourceFiles_.addAll(other.sourceFiles_);
-          }
-          onChanged();
-        }
-      } else {
-        if (!other.sourceFiles_.isEmpty()) {
-          if (sourceFilesBuilder_.isEmpty()) {
-            sourceFilesBuilder_.dispose();
-            sourceFilesBuilder_ = null;
-            sourceFiles_ = other.sourceFiles_;
-            bitField0_ = (bitField0_ & ~0x00000001);
-            sourceFilesBuilder_ = 
-              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
-                 getSourceFilesFieldBuilder() : null;
-          } else {
-            sourceFilesBuilder_.addAllMessages(other.sourceFiles_);
-          }
-        }
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.framework.DebuggedSourceFiles parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.framework.DebuggedSourceFiles) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-    private int bitField0_;
-
-    private java.util.List<org.tensorflow.proto.framework.DebuggedSourceFile> sourceFiles_ =
-      java.util.Collections.emptyList();
-    private void ensureSourceFilesIsMutable() {
-      if (!((bitField0_ & 0x00000001) != 0)) {
-        sourceFiles_ = new java.util.ArrayList<org.tensorflow.proto.framework.DebuggedSourceFile>(sourceFiles_);
-        bitField0_ |= 0x00000001;
-       }
-    }
-
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.framework.DebuggedSourceFile, org.tensorflow.proto.framework.DebuggedSourceFile.Builder, org.tensorflow.proto.framework.DebuggedSourceFileOrBuilder> sourceFilesBuilder_;
-
-    /**
-     * <pre>
-     * A collection of source code files.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.DebuggedSourceFile source_files = 1;</code>
-     */
-    public java.util.List<org.tensorflow.proto.framework.DebuggedSourceFile> getSourceFilesList() {
-      if (sourceFilesBuilder_ == null) {
-        return java.util.Collections.unmodifiableList(sourceFiles_);
-      } else {
-        return sourceFilesBuilder_.getMessageList();
-      }
-    }
-    /**
-     * <pre>
-     * A collection of source code files.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.DebuggedSourceFile source_files = 1;</code>
-     */
-    public int getSourceFilesCount() {
-      if (sourceFilesBuilder_ == null) {
-        return sourceFiles_.size();
-      } else {
-        return sourceFilesBuilder_.getCount();
-      }
-    }
-    /**
-     * <pre>
-     * A collection of source code files.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.DebuggedSourceFile source_files = 1;</code>
-     */
-    public org.tensorflow.proto.framework.DebuggedSourceFile getSourceFiles(int index) {
-      if (sourceFilesBuilder_ == null) {
-        return sourceFiles_.get(index);
-      } else {
-        return sourceFilesBuilder_.getMessage(index);
-      }
-    }
-    /**
-     * <pre>
-     * A collection of source code files.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.DebuggedSourceFile source_files = 1;</code>
-     */
-    public Builder setSourceFiles(
-        int index, org.tensorflow.proto.framework.DebuggedSourceFile value) {
-      if (sourceFilesBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureSourceFilesIsMutable();
-        sourceFiles_.set(index, value);
-        onChanged();
-      } else {
-        sourceFilesBuilder_.setMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * A collection of source code files.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.DebuggedSourceFile source_files = 1;</code>
-     */
-    public Builder setSourceFiles(
-        int index, org.tensorflow.proto.framework.DebuggedSourceFile.Builder builderForValue) {
-      if (sourceFilesBuilder_ == null) {
-        ensureSourceFilesIsMutable();
-        sourceFiles_.set(index, builderForValue.build());
-        onChanged();
-      } else {
-        sourceFilesBuilder_.setMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * A collection of source code files.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.DebuggedSourceFile source_files = 1;</code>
-     */
-    public Builder addSourceFiles(org.tensorflow.proto.framework.DebuggedSourceFile value) {
-      if (sourceFilesBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureSourceFilesIsMutable();
-        sourceFiles_.add(value);
-        onChanged();
-      } else {
-        sourceFilesBuilder_.addMessage(value);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * A collection of source code files.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.DebuggedSourceFile source_files = 1;</code>
-     */
-    public Builder addSourceFiles(
-        int index, org.tensorflow.proto.framework.DebuggedSourceFile value) {
-      if (sourceFilesBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureSourceFilesIsMutable();
-        sourceFiles_.add(index, value);
-        onChanged();
-      } else {
-        sourceFilesBuilder_.addMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * A collection of source code files.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.DebuggedSourceFile source_files = 1;</code>
-     */
-    public Builder addSourceFiles(
-        org.tensorflow.proto.framework.DebuggedSourceFile.Builder builderForValue) {
-      if (sourceFilesBuilder_ == null) {
-        ensureSourceFilesIsMutable();
-        sourceFiles_.add(builderForValue.build());
-        onChanged();
-      } else {
-        sourceFilesBuilder_.addMessage(builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * A collection of source code files.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.DebuggedSourceFile source_files = 1;</code>
-     */
-    public Builder addSourceFiles(
-        int index, org.tensorflow.proto.framework.DebuggedSourceFile.Builder builderForValue) {
-      if (sourceFilesBuilder_ == null) {
-        ensureSourceFilesIsMutable();
-        sourceFiles_.add(index, builderForValue.build());
-        onChanged();
-      } else {
-        sourceFilesBuilder_.addMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * A collection of source code files.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.DebuggedSourceFile source_files = 1;</code>
-     */
-    public Builder addAllSourceFiles(
-        java.lang.Iterable<? extends org.tensorflow.proto.framework.DebuggedSourceFile> values) {
-      if (sourceFilesBuilder_ == null) {
-        ensureSourceFilesIsMutable();
-        com.google.protobuf.AbstractMessageLite.Builder.addAll(
-            values, sourceFiles_);
-        onChanged();
-      } else {
-        sourceFilesBuilder_.addAllMessages(values);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * A collection of source code files.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.DebuggedSourceFile source_files = 1;</code>
-     */
-    public Builder clearSourceFiles() {
-      if (sourceFilesBuilder_ == null) {
-        sourceFiles_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000001);
-        onChanged();
-      } else {
-        sourceFilesBuilder_.clear();
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * A collection of source code files.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.DebuggedSourceFile source_files = 1;</code>
-     */
-    public Builder removeSourceFiles(int index) {
-      if (sourceFilesBuilder_ == null) {
-        ensureSourceFilesIsMutable();
-        sourceFiles_.remove(index);
-        onChanged();
-      } else {
-        sourceFilesBuilder_.remove(index);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * A collection of source code files.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.DebuggedSourceFile source_files = 1;</code>
-     */
-    public org.tensorflow.proto.framework.DebuggedSourceFile.Builder getSourceFilesBuilder(
-        int index) {
-      return getSourceFilesFieldBuilder().getBuilder(index);
-    }
-    /**
-     * <pre>
-     * A collection of source code files.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.DebuggedSourceFile source_files = 1;</code>
-     */
-    public org.tensorflow.proto.framework.DebuggedSourceFileOrBuilder getSourceFilesOrBuilder(
-        int index) {
-      if (sourceFilesBuilder_ == null) {
-        return sourceFiles_.get(index);  } else {
-        return sourceFilesBuilder_.getMessageOrBuilder(index);
-      }
-    }
-    /**
-     * <pre>
-     * A collection of source code files.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.DebuggedSourceFile source_files = 1;</code>
-     */
-    public java.util.List<? extends org.tensorflow.proto.framework.DebuggedSourceFileOrBuilder> 
-         getSourceFilesOrBuilderList() {
-      if (sourceFilesBuilder_ != null) {
-        return sourceFilesBuilder_.getMessageOrBuilderList();
-      } else {
-        return java.util.Collections.unmodifiableList(sourceFiles_);
-      }
-    }
-    /**
-     * <pre>
-     * A collection of source code files.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.DebuggedSourceFile source_files = 1;</code>
-     */
-    public org.tensorflow.proto.framework.DebuggedSourceFile.Builder addSourceFilesBuilder() {
-      return getSourceFilesFieldBuilder().addBuilder(
-          org.tensorflow.proto.framework.DebuggedSourceFile.getDefaultInstance());
-    }
-    /**
-     * <pre>
-     * A collection of source code files.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.DebuggedSourceFile source_files = 1;</code>
-     */
-    public org.tensorflow.proto.framework.DebuggedSourceFile.Builder addSourceFilesBuilder(
-        int index) {
-      return getSourceFilesFieldBuilder().addBuilder(
-          index, org.tensorflow.proto.framework.DebuggedSourceFile.getDefaultInstance());
-    }
-    /**
-     * <pre>
-     * A collection of source code files.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.DebuggedSourceFile source_files = 1;</code>
-     */
-    public java.util.List<org.tensorflow.proto.framework.DebuggedSourceFile.Builder> 
-         getSourceFilesBuilderList() {
-      return getSourceFilesFieldBuilder().getBuilderList();
-    }
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.framework.DebuggedSourceFile, org.tensorflow.proto.framework.DebuggedSourceFile.Builder, org.tensorflow.proto.framework.DebuggedSourceFileOrBuilder> 
-        getSourceFilesFieldBuilder() {
-      if (sourceFilesBuilder_ == null) {
-        sourceFilesBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
-            org.tensorflow.proto.framework.DebuggedSourceFile, org.tensorflow.proto.framework.DebuggedSourceFile.Builder, org.tensorflow.proto.framework.DebuggedSourceFileOrBuilder>(
-                sourceFiles_,
-                ((bitField0_ & 0x00000001) != 0),
-                getParentForChildren(),
-                isClean());
-        sourceFiles_ = null;
-      }
-      return sourceFilesBuilder_;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.DebuggedSourceFiles)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.DebuggedSourceFiles)
-  private static final org.tensorflow.proto.framework.DebuggedSourceFiles DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.DebuggedSourceFiles();
-  }
-
-  public static org.tensorflow.proto.framework.DebuggedSourceFiles getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<DebuggedSourceFiles>
-      PARSER = new com.google.protobuf.AbstractParser<DebuggedSourceFiles>() {
-    @java.lang.Override
-    public DebuggedSourceFiles parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new DebuggedSourceFiles(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<DebuggedSourceFiles> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<DebuggedSourceFiles> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.framework.DebuggedSourceFiles getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DebuggedSourceFilesOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DebuggedSourceFilesOrBuilder.java
deleted file mode 100644
index 3afc8e4f78f..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DebuggedSourceFilesOrBuilder.java
+++ /dev/null
@@ -1,53 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/debug.proto
-
-package org.tensorflow.proto.framework;
-
-public interface DebuggedSourceFilesOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.DebuggedSourceFiles)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <pre>
-   * A collection of source code files.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.DebuggedSourceFile source_files = 1;</code>
-   */
-  java.util.List<org.tensorflow.proto.framework.DebuggedSourceFile> 
-      getSourceFilesList();
-  /**
-   * <pre>
-   * A collection of source code files.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.DebuggedSourceFile source_files = 1;</code>
-   */
-  org.tensorflow.proto.framework.DebuggedSourceFile getSourceFiles(int index);
-  /**
-   * <pre>
-   * A collection of source code files.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.DebuggedSourceFile source_files = 1;</code>
-   */
-  int getSourceFilesCount();
-  /**
-   * <pre>
-   * A collection of source code files.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.DebuggedSourceFile source_files = 1;</code>
-   */
-  java.util.List<? extends org.tensorflow.proto.framework.DebuggedSourceFileOrBuilder> 
-      getSourceFilesOrBuilderList();
-  /**
-   * <pre>
-   * A collection of source code files.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.DebuggedSourceFile source_files = 1;</code>
-   */
-  org.tensorflow.proto.framework.DebuggedSourceFileOrBuilder getSourceFilesOrBuilder(
-      int index);
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/Device.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/Device.java
deleted file mode 100644
index 59507a00680..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/Device.java
+++ /dev/null
@@ -1,977 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/trace_events.proto
-
-package org.tensorflow.proto.framework;
-
-/**
- * <pre>
- * A 'device' is a physical entity in the system and is comprised of several
- * resources.
- * </pre>
- *
- * Protobuf type {@code tensorflow.profiler.Device}
- */
-public  final class Device extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.profiler.Device)
-    DeviceOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use Device.newBuilder() to construct.
-  private Device(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private Device() {
-    name_ = "";
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new Device();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private Device(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    int mutable_bitField0_ = 0;
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 10: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            name_ = s;
-            break;
-          }
-          case 16: {
-
-            deviceId_ = input.readUInt32();
-            break;
-          }
-          case 26: {
-            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
-              resources_ = com.google.protobuf.MapField.newMapField(
-                  ResourcesDefaultEntryHolder.defaultEntry);
-              mutable_bitField0_ |= 0x00000001;
-            }
-            com.google.protobuf.MapEntry<java.lang.Integer, org.tensorflow.proto.framework.Resource>
-            resources__ = input.readMessage(
-                ResourcesDefaultEntryHolder.defaultEntry.getParserForType(), extensionRegistry);
-            resources_.getMutableMap().put(
-                resources__.getKey(), resources__.getValue());
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.framework.TraceEventsProtos.internal_static_tensorflow_profiler_Device_descriptor;
-  }
-
-  @SuppressWarnings({"rawtypes"})
-  @java.lang.Override
-  protected com.google.protobuf.MapField internalGetMapField(
-      int number) {
-    switch (number) {
-      case 3:
-        return internalGetResources();
-      default:
-        throw new RuntimeException(
-            "Invalid map field number: " + number);
-    }
-  }
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.framework.TraceEventsProtos.internal_static_tensorflow_profiler_Device_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.framework.Device.class, org.tensorflow.proto.framework.Device.Builder.class);
-  }
-
-  public static final int NAME_FIELD_NUMBER = 1;
-  private volatile java.lang.Object name_;
-  /**
-   * <pre>
-   * The name of the device.
-   * </pre>
-   *
-   * <code>string name = 1;</code>
-   */
-  public java.lang.String getName() {
-    java.lang.Object ref = name_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      name_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * The name of the device.
-   * </pre>
-   *
-   * <code>string name = 1;</code>
-   */
-  public com.google.protobuf.ByteString
-      getNameBytes() {
-    java.lang.Object ref = name_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      name_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int DEVICE_ID_FIELD_NUMBER = 2;
-  private int deviceId_;
-  /**
-   * <pre>
-   * The id of this device, unique in a single trace.
-   * </pre>
-   *
-   * <code>uint32 device_id = 2;</code>
-   */
-  public int getDeviceId() {
-    return deviceId_;
-  }
-
-  public static final int RESOURCES_FIELD_NUMBER = 3;
-  private static final class ResourcesDefaultEntryHolder {
-    static final com.google.protobuf.MapEntry<
-        java.lang.Integer, org.tensorflow.proto.framework.Resource> defaultEntry =
-            com.google.protobuf.MapEntry
-            .<java.lang.Integer, org.tensorflow.proto.framework.Resource>newDefaultInstance(
-                org.tensorflow.proto.framework.TraceEventsProtos.internal_static_tensorflow_profiler_Device_ResourcesEntry_descriptor, 
-                com.google.protobuf.WireFormat.FieldType.UINT32,
-                0,
-                com.google.protobuf.WireFormat.FieldType.MESSAGE,
-                org.tensorflow.proto.framework.Resource.getDefaultInstance());
-  }
-  private com.google.protobuf.MapField<
-      java.lang.Integer, org.tensorflow.proto.framework.Resource> resources_;
-  private com.google.protobuf.MapField<java.lang.Integer, org.tensorflow.proto.framework.Resource>
-  internalGetResources() {
-    if (resources_ == null) {
-      return com.google.protobuf.MapField.emptyMapField(
-          ResourcesDefaultEntryHolder.defaultEntry);
-    }
-    return resources_;
-  }
-
-  public int getResourcesCount() {
-    return internalGetResources().getMap().size();
-  }
-  /**
-   * <pre>
-   * The resources on this device, keyed by resource_id;
-   * </pre>
-   *
-   * <code>map&lt;uint32, .tensorflow.profiler.Resource&gt; resources = 3;</code>
-   */
-
-  public boolean containsResources(
-      int key) {
-    
-    return internalGetResources().getMap().containsKey(key);
-  }
-  /**
-   * Use {@link #getResourcesMap()} instead.
-   */
-  @java.lang.Deprecated
-  public java.util.Map<java.lang.Integer, org.tensorflow.proto.framework.Resource> getResources() {
-    return getResourcesMap();
-  }
-  /**
-   * <pre>
-   * The resources on this device, keyed by resource_id;
-   * </pre>
-   *
-   * <code>map&lt;uint32, .tensorflow.profiler.Resource&gt; resources = 3;</code>
-   */
-
-  public java.util.Map<java.lang.Integer, org.tensorflow.proto.framework.Resource> getResourcesMap() {
-    return internalGetResources().getMap();
-  }
-  /**
-   * <pre>
-   * The resources on this device, keyed by resource_id;
-   * </pre>
-   *
-   * <code>map&lt;uint32, .tensorflow.profiler.Resource&gt; resources = 3;</code>
-   */
-
-  public org.tensorflow.proto.framework.Resource getResourcesOrDefault(
-      int key,
-      org.tensorflow.proto.framework.Resource defaultValue) {
-    
-    java.util.Map<java.lang.Integer, org.tensorflow.proto.framework.Resource> map =
-        internalGetResources().getMap();
-    return map.containsKey(key) ? map.get(key) : defaultValue;
-  }
-  /**
-   * <pre>
-   * The resources on this device, keyed by resource_id;
-   * </pre>
-   *
-   * <code>map&lt;uint32, .tensorflow.profiler.Resource&gt; resources = 3;</code>
-   */
-
-  public org.tensorflow.proto.framework.Resource getResourcesOrThrow(
-      int key) {
-    
-    java.util.Map<java.lang.Integer, org.tensorflow.proto.framework.Resource> map =
-        internalGetResources().getMap();
-    if (!map.containsKey(key)) {
-      throw new java.lang.IllegalArgumentException();
-    }
-    return map.get(key);
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    if (!getNameBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 1, name_);
-    }
-    if (deviceId_ != 0) {
-      output.writeUInt32(2, deviceId_);
-    }
-    com.google.protobuf.GeneratedMessageV3
-      .serializeIntegerMapTo(
-        output,
-        internalGetResources(),
-        ResourcesDefaultEntryHolder.defaultEntry,
-        3);
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (!getNameBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, name_);
-    }
-    if (deviceId_ != 0) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeUInt32Size(2, deviceId_);
-    }
-    for (java.util.Map.Entry<java.lang.Integer, org.tensorflow.proto.framework.Resource> entry
-         : internalGetResources().getMap().entrySet()) {
-      com.google.protobuf.MapEntry<java.lang.Integer, org.tensorflow.proto.framework.Resource>
-      resources__ = ResourcesDefaultEntryHolder.defaultEntry.newBuilderForType()
-          .setKey(entry.getKey())
-          .setValue(entry.getValue())
-          .build();
-      size += com.google.protobuf.CodedOutputStream
-          .computeMessageSize(3, resources__);
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.framework.Device)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.framework.Device other = (org.tensorflow.proto.framework.Device) obj;
-
-    if (!getName()
-        .equals(other.getName())) return false;
-    if (getDeviceId()
-        != other.getDeviceId()) return false;
-    if (!internalGetResources().equals(
-        other.internalGetResources())) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    hash = (37 * hash) + NAME_FIELD_NUMBER;
-    hash = (53 * hash) + getName().hashCode();
-    hash = (37 * hash) + DEVICE_ID_FIELD_NUMBER;
-    hash = (53 * hash) + getDeviceId();
-    if (!internalGetResources().getMap().isEmpty()) {
-      hash = (37 * hash) + RESOURCES_FIELD_NUMBER;
-      hash = (53 * hash) + internalGetResources().hashCode();
-    }
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.framework.Device parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.Device parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.Device parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.Device parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.Device parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.Device parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.Device parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.Device parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.Device parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.Device parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.Device parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.Device parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.framework.Device prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * <pre>
-   * A 'device' is a physical entity in the system and is comprised of several
-   * resources.
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.profiler.Device}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.profiler.Device)
-      org.tensorflow.proto.framework.DeviceOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.TraceEventsProtos.internal_static_tensorflow_profiler_Device_descriptor;
-    }
-
-    @SuppressWarnings({"rawtypes"})
-    protected com.google.protobuf.MapField internalGetMapField(
-        int number) {
-      switch (number) {
-        case 3:
-          return internalGetResources();
-        default:
-          throw new RuntimeException(
-              "Invalid map field number: " + number);
-      }
-    }
-    @SuppressWarnings({"rawtypes"})
-    protected com.google.protobuf.MapField internalGetMutableMapField(
-        int number) {
-      switch (number) {
-        case 3:
-          return internalGetMutableResources();
-        default:
-          throw new RuntimeException(
-              "Invalid map field number: " + number);
-      }
-    }
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.TraceEventsProtos.internal_static_tensorflow_profiler_Device_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.Device.class, org.tensorflow.proto.framework.Device.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.framework.Device.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      name_ = "";
-
-      deviceId_ = 0;
-
-      internalGetMutableResources().clear();
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.framework.TraceEventsProtos.internal_static_tensorflow_profiler_Device_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.Device getDefaultInstanceForType() {
-      return org.tensorflow.proto.framework.Device.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.Device build() {
-      org.tensorflow.proto.framework.Device result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.Device buildPartial() {
-      org.tensorflow.proto.framework.Device result = new org.tensorflow.proto.framework.Device(this);
-      int from_bitField0_ = bitField0_;
-      result.name_ = name_;
-      result.deviceId_ = deviceId_;
-      result.resources_ = internalGetResources();
-      result.resources_.makeImmutable();
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.framework.Device) {
-        return mergeFrom((org.tensorflow.proto.framework.Device)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.framework.Device other) {
-      if (other == org.tensorflow.proto.framework.Device.getDefaultInstance()) return this;
-      if (!other.getName().isEmpty()) {
-        name_ = other.name_;
-        onChanged();
-      }
-      if (other.getDeviceId() != 0) {
-        setDeviceId(other.getDeviceId());
-      }
-      internalGetMutableResources().mergeFrom(
-          other.internalGetResources());
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.framework.Device parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.framework.Device) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-    private int bitField0_;
-
-    private java.lang.Object name_ = "";
-    /**
-     * <pre>
-     * The name of the device.
-     * </pre>
-     *
-     * <code>string name = 1;</code>
-     */
-    public java.lang.String getName() {
-      java.lang.Object ref = name_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        name_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * The name of the device.
-     * </pre>
-     *
-     * <code>string name = 1;</code>
-     */
-    public com.google.protobuf.ByteString
-        getNameBytes() {
-      java.lang.Object ref = name_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        name_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * The name of the device.
-     * </pre>
-     *
-     * <code>string name = 1;</code>
-     */
-    public Builder setName(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      name_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * The name of the device.
-     * </pre>
-     *
-     * <code>string name = 1;</code>
-     */
-    public Builder clearName() {
-      
-      name_ = getDefaultInstance().getName();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * The name of the device.
-     * </pre>
-     *
-     * <code>string name = 1;</code>
-     */
-    public Builder setNameBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      name_ = value;
-      onChanged();
-      return this;
-    }
-
-    private int deviceId_ ;
-    /**
-     * <pre>
-     * The id of this device, unique in a single trace.
-     * </pre>
-     *
-     * <code>uint32 device_id = 2;</code>
-     */
-    public int getDeviceId() {
-      return deviceId_;
-    }
-    /**
-     * <pre>
-     * The id of this device, unique in a single trace.
-     * </pre>
-     *
-     * <code>uint32 device_id = 2;</code>
-     */
-    public Builder setDeviceId(int value) {
-      
-      deviceId_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * The id of this device, unique in a single trace.
-     * </pre>
-     *
-     * <code>uint32 device_id = 2;</code>
-     */
-    public Builder clearDeviceId() {
-      
-      deviceId_ = 0;
-      onChanged();
-      return this;
-    }
-
-    private com.google.protobuf.MapField<
-        java.lang.Integer, org.tensorflow.proto.framework.Resource> resources_;
-    private com.google.protobuf.MapField<java.lang.Integer, org.tensorflow.proto.framework.Resource>
-    internalGetResources() {
-      if (resources_ == null) {
-        return com.google.protobuf.MapField.emptyMapField(
-            ResourcesDefaultEntryHolder.defaultEntry);
-      }
-      return resources_;
-    }
-    private com.google.protobuf.MapField<java.lang.Integer, org.tensorflow.proto.framework.Resource>
-    internalGetMutableResources() {
-      onChanged();;
-      if (resources_ == null) {
-        resources_ = com.google.protobuf.MapField.newMapField(
-            ResourcesDefaultEntryHolder.defaultEntry);
-      }
-      if (!resources_.isMutable()) {
-        resources_ = resources_.copy();
-      }
-      return resources_;
-    }
-
-    public int getResourcesCount() {
-      return internalGetResources().getMap().size();
-    }
-    /**
-     * <pre>
-     * The resources on this device, keyed by resource_id;
-     * </pre>
-     *
-     * <code>map&lt;uint32, .tensorflow.profiler.Resource&gt; resources = 3;</code>
-     */
-
-    public boolean containsResources(
-        int key) {
-      
-      return internalGetResources().getMap().containsKey(key);
-    }
-    /**
-     * Use {@link #getResourcesMap()} instead.
-     */
-    @java.lang.Deprecated
-    public java.util.Map<java.lang.Integer, org.tensorflow.proto.framework.Resource> getResources() {
-      return getResourcesMap();
-    }
-    /**
-     * <pre>
-     * The resources on this device, keyed by resource_id;
-     * </pre>
-     *
-     * <code>map&lt;uint32, .tensorflow.profiler.Resource&gt; resources = 3;</code>
-     */
-
-    public java.util.Map<java.lang.Integer, org.tensorflow.proto.framework.Resource> getResourcesMap() {
-      return internalGetResources().getMap();
-    }
-    /**
-     * <pre>
-     * The resources on this device, keyed by resource_id;
-     * </pre>
-     *
-     * <code>map&lt;uint32, .tensorflow.profiler.Resource&gt; resources = 3;</code>
-     */
-
-    public org.tensorflow.proto.framework.Resource getResourcesOrDefault(
-        int key,
-        org.tensorflow.proto.framework.Resource defaultValue) {
-      
-      java.util.Map<java.lang.Integer, org.tensorflow.proto.framework.Resource> map =
-          internalGetResources().getMap();
-      return map.containsKey(key) ? map.get(key) : defaultValue;
-    }
-    /**
-     * <pre>
-     * The resources on this device, keyed by resource_id;
-     * </pre>
-     *
-     * <code>map&lt;uint32, .tensorflow.profiler.Resource&gt; resources = 3;</code>
-     */
-
-    public org.tensorflow.proto.framework.Resource getResourcesOrThrow(
-        int key) {
-      
-      java.util.Map<java.lang.Integer, org.tensorflow.proto.framework.Resource> map =
-          internalGetResources().getMap();
-      if (!map.containsKey(key)) {
-        throw new java.lang.IllegalArgumentException();
-      }
-      return map.get(key);
-    }
-
-    public Builder clearResources() {
-      internalGetMutableResources().getMutableMap()
-          .clear();
-      return this;
-    }
-    /**
-     * <pre>
-     * The resources on this device, keyed by resource_id;
-     * </pre>
-     *
-     * <code>map&lt;uint32, .tensorflow.profiler.Resource&gt; resources = 3;</code>
-     */
-
-    public Builder removeResources(
-        int key) {
-      
-      internalGetMutableResources().getMutableMap()
-          .remove(key);
-      return this;
-    }
-    /**
-     * Use alternate mutation accessors instead.
-     */
-    @java.lang.Deprecated
-    public java.util.Map<java.lang.Integer, org.tensorflow.proto.framework.Resource>
-    getMutableResources() {
-      return internalGetMutableResources().getMutableMap();
-    }
-    /**
-     * <pre>
-     * The resources on this device, keyed by resource_id;
-     * </pre>
-     *
-     * <code>map&lt;uint32, .tensorflow.profiler.Resource&gt; resources = 3;</code>
-     */
-    public Builder putResources(
-        int key,
-        org.tensorflow.proto.framework.Resource value) {
-      
-      if (value == null) { throw new java.lang.NullPointerException(); }
-      internalGetMutableResources().getMutableMap()
-          .put(key, value);
-      return this;
-    }
-    /**
-     * <pre>
-     * The resources on this device, keyed by resource_id;
-     * </pre>
-     *
-     * <code>map&lt;uint32, .tensorflow.profiler.Resource&gt; resources = 3;</code>
-     */
-
-    public Builder putAllResources(
-        java.util.Map<java.lang.Integer, org.tensorflow.proto.framework.Resource> values) {
-      internalGetMutableResources().getMutableMap()
-          .putAll(values);
-      return this;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.profiler.Device)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.profiler.Device)
-  private static final org.tensorflow.proto.framework.Device DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.Device();
-  }
-
-  public static org.tensorflow.proto.framework.Device getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<Device>
-      PARSER = new com.google.protobuf.AbstractParser<Device>() {
-    @java.lang.Override
-    public Device parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new Device(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<Device> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<Device> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.framework.Device getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DeviceAttributes.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DeviceAttributes.java
deleted file mode 100644
index 3da166dfbfd..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DeviceAttributes.java
+++ /dev/null
@@ -1,1277 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/framework/device_attributes.proto
-
-package org.tensorflow.proto.framework;
-
-/**
- * Protobuf type {@code tensorflow.DeviceAttributes}
- */
-public  final class DeviceAttributes extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.DeviceAttributes)
-    DeviceAttributesOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use DeviceAttributes.newBuilder() to construct.
-  private DeviceAttributes(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private DeviceAttributes() {
-    name_ = "";
-    deviceType_ = "";
-    physicalDeviceDesc_ = "";
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new DeviceAttributes();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private DeviceAttributes(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 10: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            name_ = s;
-            break;
-          }
-          case 18: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            deviceType_ = s;
-            break;
-          }
-          case 32: {
-
-            memoryLimit_ = input.readInt64();
-            break;
-          }
-          case 42: {
-            org.tensorflow.proto.framework.DeviceLocality.Builder subBuilder = null;
-            if (locality_ != null) {
-              subBuilder = locality_.toBuilder();
-            }
-            locality_ = input.readMessage(org.tensorflow.proto.framework.DeviceLocality.parser(), extensionRegistry);
-            if (subBuilder != null) {
-              subBuilder.mergeFrom(locality_);
-              locality_ = subBuilder.buildPartial();
-            }
-
-            break;
-          }
-          case 49: {
-
-            incarnation_ = input.readFixed64();
-            break;
-          }
-          case 58: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            physicalDeviceDesc_ = s;
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.framework.DeviceAttributesProtos.internal_static_tensorflow_DeviceAttributes_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.framework.DeviceAttributesProtos.internal_static_tensorflow_DeviceAttributes_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.framework.DeviceAttributes.class, org.tensorflow.proto.framework.DeviceAttributes.Builder.class);
-  }
-
-  public static final int NAME_FIELD_NUMBER = 1;
-  private volatile java.lang.Object name_;
-  /**
-   * <pre>
-   * Fully specified name of the device within a cluster.
-   * </pre>
-   *
-   * <code>string name = 1;</code>
-   */
-  public java.lang.String getName() {
-    java.lang.Object ref = name_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      name_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * Fully specified name of the device within a cluster.
-   * </pre>
-   *
-   * <code>string name = 1;</code>
-   */
-  public com.google.protobuf.ByteString
-      getNameBytes() {
-    java.lang.Object ref = name_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      name_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int DEVICE_TYPE_FIELD_NUMBER = 2;
-  private volatile java.lang.Object deviceType_;
-  /**
-   * <pre>
-   * String representation of device_type.
-   * </pre>
-   *
-   * <code>string device_type = 2;</code>
-   */
-  public java.lang.String getDeviceType() {
-    java.lang.Object ref = deviceType_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      deviceType_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * String representation of device_type.
-   * </pre>
-   *
-   * <code>string device_type = 2;</code>
-   */
-  public com.google.protobuf.ByteString
-      getDeviceTypeBytes() {
-    java.lang.Object ref = deviceType_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      deviceType_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int MEMORY_LIMIT_FIELD_NUMBER = 4;
-  private long memoryLimit_;
-  /**
-   * <pre>
-   * Memory capacity of device in bytes.
-   * </pre>
-   *
-   * <code>int64 memory_limit = 4;</code>
-   */
-  public long getMemoryLimit() {
-    return memoryLimit_;
-  }
-
-  public static final int LOCALITY_FIELD_NUMBER = 5;
-  private org.tensorflow.proto.framework.DeviceLocality locality_;
-  /**
-   * <pre>
-   * Platform-specific data about device that may be useful
-   * for supporting efficient data transfers.
-   * </pre>
-   *
-   * <code>.tensorflow.DeviceLocality locality = 5;</code>
-   */
-  public boolean hasLocality() {
-    return locality_ != null;
-  }
-  /**
-   * <pre>
-   * Platform-specific data about device that may be useful
-   * for supporting efficient data transfers.
-   * </pre>
-   *
-   * <code>.tensorflow.DeviceLocality locality = 5;</code>
-   */
-  public org.tensorflow.proto.framework.DeviceLocality getLocality() {
-    return locality_ == null ? org.tensorflow.proto.framework.DeviceLocality.getDefaultInstance() : locality_;
-  }
-  /**
-   * <pre>
-   * Platform-specific data about device that may be useful
-   * for supporting efficient data transfers.
-   * </pre>
-   *
-   * <code>.tensorflow.DeviceLocality locality = 5;</code>
-   */
-  public org.tensorflow.proto.framework.DeviceLocalityOrBuilder getLocalityOrBuilder() {
-    return getLocality();
-  }
-
-  public static final int INCARNATION_FIELD_NUMBER = 6;
-  private long incarnation_;
-  /**
-   * <pre>
-   * A device is assigned a global unique number each time it is
-   * initialized. "incarnation" should never be 0.
-   * </pre>
-   *
-   * <code>fixed64 incarnation = 6;</code>
-   */
-  public long getIncarnation() {
-    return incarnation_;
-  }
-
-  public static final int PHYSICAL_DEVICE_DESC_FIELD_NUMBER = 7;
-  private volatile java.lang.Object physicalDeviceDesc_;
-  /**
-   * <pre>
-   * String representation of the physical device that this device maps to.
-   * </pre>
-   *
-   * <code>string physical_device_desc = 7;</code>
-   */
-  public java.lang.String getPhysicalDeviceDesc() {
-    java.lang.Object ref = physicalDeviceDesc_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      physicalDeviceDesc_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * String representation of the physical device that this device maps to.
-   * </pre>
-   *
-   * <code>string physical_device_desc = 7;</code>
-   */
-  public com.google.protobuf.ByteString
-      getPhysicalDeviceDescBytes() {
-    java.lang.Object ref = physicalDeviceDesc_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      physicalDeviceDesc_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    if (!getNameBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 1, name_);
-    }
-    if (!getDeviceTypeBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 2, deviceType_);
-    }
-    if (memoryLimit_ != 0L) {
-      output.writeInt64(4, memoryLimit_);
-    }
-    if (locality_ != null) {
-      output.writeMessage(5, getLocality());
-    }
-    if (incarnation_ != 0L) {
-      output.writeFixed64(6, incarnation_);
-    }
-    if (!getPhysicalDeviceDescBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 7, physicalDeviceDesc_);
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (!getNameBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, name_);
-    }
-    if (!getDeviceTypeBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(2, deviceType_);
-    }
-    if (memoryLimit_ != 0L) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt64Size(4, memoryLimit_);
-    }
-    if (locality_ != null) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(5, getLocality());
-    }
-    if (incarnation_ != 0L) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeFixed64Size(6, incarnation_);
-    }
-    if (!getPhysicalDeviceDescBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(7, physicalDeviceDesc_);
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.framework.DeviceAttributes)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.framework.DeviceAttributes other = (org.tensorflow.proto.framework.DeviceAttributes) obj;
-
-    if (!getName()
-        .equals(other.getName())) return false;
-    if (!getDeviceType()
-        .equals(other.getDeviceType())) return false;
-    if (getMemoryLimit()
-        != other.getMemoryLimit()) return false;
-    if (hasLocality() != other.hasLocality()) return false;
-    if (hasLocality()) {
-      if (!getLocality()
-          .equals(other.getLocality())) return false;
-    }
-    if (getIncarnation()
-        != other.getIncarnation()) return false;
-    if (!getPhysicalDeviceDesc()
-        .equals(other.getPhysicalDeviceDesc())) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    hash = (37 * hash) + NAME_FIELD_NUMBER;
-    hash = (53 * hash) + getName().hashCode();
-    hash = (37 * hash) + DEVICE_TYPE_FIELD_NUMBER;
-    hash = (53 * hash) + getDeviceType().hashCode();
-    hash = (37 * hash) + MEMORY_LIMIT_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-        getMemoryLimit());
-    if (hasLocality()) {
-      hash = (37 * hash) + LOCALITY_FIELD_NUMBER;
-      hash = (53 * hash) + getLocality().hashCode();
-    }
-    hash = (37 * hash) + INCARNATION_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-        getIncarnation());
-    hash = (37 * hash) + PHYSICAL_DEVICE_DESC_FIELD_NUMBER;
-    hash = (53 * hash) + getPhysicalDeviceDesc().hashCode();
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.framework.DeviceAttributes parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.DeviceAttributes parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.DeviceAttributes parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.DeviceAttributes parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.DeviceAttributes parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.DeviceAttributes parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.DeviceAttributes parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.DeviceAttributes parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.DeviceAttributes parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.DeviceAttributes parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.DeviceAttributes parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.DeviceAttributes parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.framework.DeviceAttributes prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * Protobuf type {@code tensorflow.DeviceAttributes}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.DeviceAttributes)
-      org.tensorflow.proto.framework.DeviceAttributesOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.DeviceAttributesProtos.internal_static_tensorflow_DeviceAttributes_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.DeviceAttributesProtos.internal_static_tensorflow_DeviceAttributes_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.DeviceAttributes.class, org.tensorflow.proto.framework.DeviceAttributes.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.framework.DeviceAttributes.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      name_ = "";
-
-      deviceType_ = "";
-
-      memoryLimit_ = 0L;
-
-      if (localityBuilder_ == null) {
-        locality_ = null;
-      } else {
-        locality_ = null;
-        localityBuilder_ = null;
-      }
-      incarnation_ = 0L;
-
-      physicalDeviceDesc_ = "";
-
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.framework.DeviceAttributesProtos.internal_static_tensorflow_DeviceAttributes_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.DeviceAttributes getDefaultInstanceForType() {
-      return org.tensorflow.proto.framework.DeviceAttributes.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.DeviceAttributes build() {
-      org.tensorflow.proto.framework.DeviceAttributes result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.DeviceAttributes buildPartial() {
-      org.tensorflow.proto.framework.DeviceAttributes result = new org.tensorflow.proto.framework.DeviceAttributes(this);
-      result.name_ = name_;
-      result.deviceType_ = deviceType_;
-      result.memoryLimit_ = memoryLimit_;
-      if (localityBuilder_ == null) {
-        result.locality_ = locality_;
-      } else {
-        result.locality_ = localityBuilder_.build();
-      }
-      result.incarnation_ = incarnation_;
-      result.physicalDeviceDesc_ = physicalDeviceDesc_;
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.framework.DeviceAttributes) {
-        return mergeFrom((org.tensorflow.proto.framework.DeviceAttributes)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.framework.DeviceAttributes other) {
-      if (other == org.tensorflow.proto.framework.DeviceAttributes.getDefaultInstance()) return this;
-      if (!other.getName().isEmpty()) {
-        name_ = other.name_;
-        onChanged();
-      }
-      if (!other.getDeviceType().isEmpty()) {
-        deviceType_ = other.deviceType_;
-        onChanged();
-      }
-      if (other.getMemoryLimit() != 0L) {
-        setMemoryLimit(other.getMemoryLimit());
-      }
-      if (other.hasLocality()) {
-        mergeLocality(other.getLocality());
-      }
-      if (other.getIncarnation() != 0L) {
-        setIncarnation(other.getIncarnation());
-      }
-      if (!other.getPhysicalDeviceDesc().isEmpty()) {
-        physicalDeviceDesc_ = other.physicalDeviceDesc_;
-        onChanged();
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.framework.DeviceAttributes parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.framework.DeviceAttributes) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-
-    private java.lang.Object name_ = "";
-    /**
-     * <pre>
-     * Fully specified name of the device within a cluster.
-     * </pre>
-     *
-     * <code>string name = 1;</code>
-     */
-    public java.lang.String getName() {
-      java.lang.Object ref = name_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        name_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Fully specified name of the device within a cluster.
-     * </pre>
-     *
-     * <code>string name = 1;</code>
-     */
-    public com.google.protobuf.ByteString
-        getNameBytes() {
-      java.lang.Object ref = name_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        name_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Fully specified name of the device within a cluster.
-     * </pre>
-     *
-     * <code>string name = 1;</code>
-     */
-    public Builder setName(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      name_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Fully specified name of the device within a cluster.
-     * </pre>
-     *
-     * <code>string name = 1;</code>
-     */
-    public Builder clearName() {
-      
-      name_ = getDefaultInstance().getName();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Fully specified name of the device within a cluster.
-     * </pre>
-     *
-     * <code>string name = 1;</code>
-     */
-    public Builder setNameBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      name_ = value;
-      onChanged();
-      return this;
-    }
-
-    private java.lang.Object deviceType_ = "";
-    /**
-     * <pre>
-     * String representation of device_type.
-     * </pre>
-     *
-     * <code>string device_type = 2;</code>
-     */
-    public java.lang.String getDeviceType() {
-      java.lang.Object ref = deviceType_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        deviceType_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * String representation of device_type.
-     * </pre>
-     *
-     * <code>string device_type = 2;</code>
-     */
-    public com.google.protobuf.ByteString
-        getDeviceTypeBytes() {
-      java.lang.Object ref = deviceType_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        deviceType_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * String representation of device_type.
-     * </pre>
-     *
-     * <code>string device_type = 2;</code>
-     */
-    public Builder setDeviceType(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      deviceType_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * String representation of device_type.
-     * </pre>
-     *
-     * <code>string device_type = 2;</code>
-     */
-    public Builder clearDeviceType() {
-      
-      deviceType_ = getDefaultInstance().getDeviceType();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * String representation of device_type.
-     * </pre>
-     *
-     * <code>string device_type = 2;</code>
-     */
-    public Builder setDeviceTypeBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      deviceType_ = value;
-      onChanged();
-      return this;
-    }
-
-    private long memoryLimit_ ;
-    /**
-     * <pre>
-     * Memory capacity of device in bytes.
-     * </pre>
-     *
-     * <code>int64 memory_limit = 4;</code>
-     */
-    public long getMemoryLimit() {
-      return memoryLimit_;
-    }
-    /**
-     * <pre>
-     * Memory capacity of device in bytes.
-     * </pre>
-     *
-     * <code>int64 memory_limit = 4;</code>
-     */
-    public Builder setMemoryLimit(long value) {
-      
-      memoryLimit_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Memory capacity of device in bytes.
-     * </pre>
-     *
-     * <code>int64 memory_limit = 4;</code>
-     */
-    public Builder clearMemoryLimit() {
-      
-      memoryLimit_ = 0L;
-      onChanged();
-      return this;
-    }
-
-    private org.tensorflow.proto.framework.DeviceLocality locality_;
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.DeviceLocality, org.tensorflow.proto.framework.DeviceLocality.Builder, org.tensorflow.proto.framework.DeviceLocalityOrBuilder> localityBuilder_;
-    /**
-     * <pre>
-     * Platform-specific data about device that may be useful
-     * for supporting efficient data transfers.
-     * </pre>
-     *
-     * <code>.tensorflow.DeviceLocality locality = 5;</code>
-     */
-    public boolean hasLocality() {
-      return localityBuilder_ != null || locality_ != null;
-    }
-    /**
-     * <pre>
-     * Platform-specific data about device that may be useful
-     * for supporting efficient data transfers.
-     * </pre>
-     *
-     * <code>.tensorflow.DeviceLocality locality = 5;</code>
-     */
-    public org.tensorflow.proto.framework.DeviceLocality getLocality() {
-      if (localityBuilder_ == null) {
-        return locality_ == null ? org.tensorflow.proto.framework.DeviceLocality.getDefaultInstance() : locality_;
-      } else {
-        return localityBuilder_.getMessage();
-      }
-    }
-    /**
-     * <pre>
-     * Platform-specific data about device that may be useful
-     * for supporting efficient data transfers.
-     * </pre>
-     *
-     * <code>.tensorflow.DeviceLocality locality = 5;</code>
-     */
-    public Builder setLocality(org.tensorflow.proto.framework.DeviceLocality value) {
-      if (localityBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        locality_ = value;
-        onChanged();
-      } else {
-        localityBuilder_.setMessage(value);
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Platform-specific data about device that may be useful
-     * for supporting efficient data transfers.
-     * </pre>
-     *
-     * <code>.tensorflow.DeviceLocality locality = 5;</code>
-     */
-    public Builder setLocality(
-        org.tensorflow.proto.framework.DeviceLocality.Builder builderForValue) {
-      if (localityBuilder_ == null) {
-        locality_ = builderForValue.build();
-        onChanged();
-      } else {
-        localityBuilder_.setMessage(builderForValue.build());
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Platform-specific data about device that may be useful
-     * for supporting efficient data transfers.
-     * </pre>
-     *
-     * <code>.tensorflow.DeviceLocality locality = 5;</code>
-     */
-    public Builder mergeLocality(org.tensorflow.proto.framework.DeviceLocality value) {
-      if (localityBuilder_ == null) {
-        if (locality_ != null) {
-          locality_ =
-            org.tensorflow.proto.framework.DeviceLocality.newBuilder(locality_).mergeFrom(value).buildPartial();
-        } else {
-          locality_ = value;
-        }
-        onChanged();
-      } else {
-        localityBuilder_.mergeFrom(value);
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Platform-specific data about device that may be useful
-     * for supporting efficient data transfers.
-     * </pre>
-     *
-     * <code>.tensorflow.DeviceLocality locality = 5;</code>
-     */
-    public Builder clearLocality() {
-      if (localityBuilder_ == null) {
-        locality_ = null;
-        onChanged();
-      } else {
-        locality_ = null;
-        localityBuilder_ = null;
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Platform-specific data about device that may be useful
-     * for supporting efficient data transfers.
-     * </pre>
-     *
-     * <code>.tensorflow.DeviceLocality locality = 5;</code>
-     */
-    public org.tensorflow.proto.framework.DeviceLocality.Builder getLocalityBuilder() {
-      
-      onChanged();
-      return getLocalityFieldBuilder().getBuilder();
-    }
-    /**
-     * <pre>
-     * Platform-specific data about device that may be useful
-     * for supporting efficient data transfers.
-     * </pre>
-     *
-     * <code>.tensorflow.DeviceLocality locality = 5;</code>
-     */
-    public org.tensorflow.proto.framework.DeviceLocalityOrBuilder getLocalityOrBuilder() {
-      if (localityBuilder_ != null) {
-        return localityBuilder_.getMessageOrBuilder();
-      } else {
-        return locality_ == null ?
-            org.tensorflow.proto.framework.DeviceLocality.getDefaultInstance() : locality_;
-      }
-    }
-    /**
-     * <pre>
-     * Platform-specific data about device that may be useful
-     * for supporting efficient data transfers.
-     * </pre>
-     *
-     * <code>.tensorflow.DeviceLocality locality = 5;</code>
-     */
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.DeviceLocality, org.tensorflow.proto.framework.DeviceLocality.Builder, org.tensorflow.proto.framework.DeviceLocalityOrBuilder> 
-        getLocalityFieldBuilder() {
-      if (localityBuilder_ == null) {
-        localityBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-            org.tensorflow.proto.framework.DeviceLocality, org.tensorflow.proto.framework.DeviceLocality.Builder, org.tensorflow.proto.framework.DeviceLocalityOrBuilder>(
-                getLocality(),
-                getParentForChildren(),
-                isClean());
-        locality_ = null;
-      }
-      return localityBuilder_;
-    }
-
-    private long incarnation_ ;
-    /**
-     * <pre>
-     * A device is assigned a global unique number each time it is
-     * initialized. "incarnation" should never be 0.
-     * </pre>
-     *
-     * <code>fixed64 incarnation = 6;</code>
-     */
-    public long getIncarnation() {
-      return incarnation_;
-    }
-    /**
-     * <pre>
-     * A device is assigned a global unique number each time it is
-     * initialized. "incarnation" should never be 0.
-     * </pre>
-     *
-     * <code>fixed64 incarnation = 6;</code>
-     */
-    public Builder setIncarnation(long value) {
-      
-      incarnation_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * A device is assigned a global unique number each time it is
-     * initialized. "incarnation" should never be 0.
-     * </pre>
-     *
-     * <code>fixed64 incarnation = 6;</code>
-     */
-    public Builder clearIncarnation() {
-      
-      incarnation_ = 0L;
-      onChanged();
-      return this;
-    }
-
-    private java.lang.Object physicalDeviceDesc_ = "";
-    /**
-     * <pre>
-     * String representation of the physical device that this device maps to.
-     * </pre>
-     *
-     * <code>string physical_device_desc = 7;</code>
-     */
-    public java.lang.String getPhysicalDeviceDesc() {
-      java.lang.Object ref = physicalDeviceDesc_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        physicalDeviceDesc_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * String representation of the physical device that this device maps to.
-     * </pre>
-     *
-     * <code>string physical_device_desc = 7;</code>
-     */
-    public com.google.protobuf.ByteString
-        getPhysicalDeviceDescBytes() {
-      java.lang.Object ref = physicalDeviceDesc_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        physicalDeviceDesc_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * String representation of the physical device that this device maps to.
-     * </pre>
-     *
-     * <code>string physical_device_desc = 7;</code>
-     */
-    public Builder setPhysicalDeviceDesc(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      physicalDeviceDesc_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * String representation of the physical device that this device maps to.
-     * </pre>
-     *
-     * <code>string physical_device_desc = 7;</code>
-     */
-    public Builder clearPhysicalDeviceDesc() {
-      
-      physicalDeviceDesc_ = getDefaultInstance().getPhysicalDeviceDesc();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * String representation of the physical device that this device maps to.
-     * </pre>
-     *
-     * <code>string physical_device_desc = 7;</code>
-     */
-    public Builder setPhysicalDeviceDescBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      physicalDeviceDesc_ = value;
-      onChanged();
-      return this;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.DeviceAttributes)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.DeviceAttributes)
-  private static final org.tensorflow.proto.framework.DeviceAttributes DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.DeviceAttributes();
-  }
-
-  public static org.tensorflow.proto.framework.DeviceAttributes getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<DeviceAttributes>
-      PARSER = new com.google.protobuf.AbstractParser<DeviceAttributes>() {
-    @java.lang.Override
-    public DeviceAttributes parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new DeviceAttributes(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<DeviceAttributes> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<DeviceAttributes> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.framework.DeviceAttributes getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DeviceAttributesOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DeviceAttributesOrBuilder.java
deleted file mode 100644
index bb8c806b325..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DeviceAttributesOrBuilder.java
+++ /dev/null
@@ -1,110 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/framework/device_attributes.proto
-
-package org.tensorflow.proto.framework;
-
-public interface DeviceAttributesOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.DeviceAttributes)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <pre>
-   * Fully specified name of the device within a cluster.
-   * </pre>
-   *
-   * <code>string name = 1;</code>
-   */
-  java.lang.String getName();
-  /**
-   * <pre>
-   * Fully specified name of the device within a cluster.
-   * </pre>
-   *
-   * <code>string name = 1;</code>
-   */
-  com.google.protobuf.ByteString
-      getNameBytes();
-
-  /**
-   * <pre>
-   * String representation of device_type.
-   * </pre>
-   *
-   * <code>string device_type = 2;</code>
-   */
-  java.lang.String getDeviceType();
-  /**
-   * <pre>
-   * String representation of device_type.
-   * </pre>
-   *
-   * <code>string device_type = 2;</code>
-   */
-  com.google.protobuf.ByteString
-      getDeviceTypeBytes();
-
-  /**
-   * <pre>
-   * Memory capacity of device in bytes.
-   * </pre>
-   *
-   * <code>int64 memory_limit = 4;</code>
-   */
-  long getMemoryLimit();
-
-  /**
-   * <pre>
-   * Platform-specific data about device that may be useful
-   * for supporting efficient data transfers.
-   * </pre>
-   *
-   * <code>.tensorflow.DeviceLocality locality = 5;</code>
-   */
-  boolean hasLocality();
-  /**
-   * <pre>
-   * Platform-specific data about device that may be useful
-   * for supporting efficient data transfers.
-   * </pre>
-   *
-   * <code>.tensorflow.DeviceLocality locality = 5;</code>
-   */
-  org.tensorflow.proto.framework.DeviceLocality getLocality();
-  /**
-   * <pre>
-   * Platform-specific data about device that may be useful
-   * for supporting efficient data transfers.
-   * </pre>
-   *
-   * <code>.tensorflow.DeviceLocality locality = 5;</code>
-   */
-  org.tensorflow.proto.framework.DeviceLocalityOrBuilder getLocalityOrBuilder();
-
-  /**
-   * <pre>
-   * A device is assigned a global unique number each time it is
-   * initialized. "incarnation" should never be 0.
-   * </pre>
-   *
-   * <code>fixed64 incarnation = 6;</code>
-   */
-  long getIncarnation();
-
-  /**
-   * <pre>
-   * String representation of the physical device that this device maps to.
-   * </pre>
-   *
-   * <code>string physical_device_desc = 7;</code>
-   */
-  java.lang.String getPhysicalDeviceDesc();
-  /**
-   * <pre>
-   * String representation of the physical device that this device maps to.
-   * </pre>
-   *
-   * <code>string physical_device_desc = 7;</code>
-   */
-  com.google.protobuf.ByteString
-      getPhysicalDeviceDescBytes();
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DeviceAttributesProtos.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DeviceAttributesProtos.java
deleted file mode 100644
index f9e2dac8a67..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DeviceAttributesProtos.java
+++ /dev/null
@@ -1,94 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/framework/device_attributes.proto
-
-package org.tensorflow.proto.framework;
-
-public final class DeviceAttributesProtos {
-  private DeviceAttributesProtos() {}
-  public static void registerAllExtensions(
-      com.google.protobuf.ExtensionRegistryLite registry) {
-  }
-
-  public static void registerAllExtensions(
-      com.google.protobuf.ExtensionRegistry registry) {
-    registerAllExtensions(
-        (com.google.protobuf.ExtensionRegistryLite) registry);
-  }
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_InterconnectLink_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_InterconnectLink_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_LocalLinks_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_LocalLinks_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_DeviceLocality_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_DeviceLocality_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_DeviceAttributes_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_DeviceAttributes_fieldAccessorTable;
-
-  public static com.google.protobuf.Descriptors.FileDescriptor
-      getDescriptor() {
-    return descriptor;
-  }
-  private static  com.google.protobuf.Descriptors.FileDescriptor
-      descriptor;
-  static {
-    java.lang.String[] descriptorData = {
-      "\n1tensorflow/core/framework/device_attri" +
-      "butes.proto\022\ntensorflow\"E\n\020InterconnectL" +
-      "ink\022\021\n\tdevice_id\030\001 \001(\005\022\014\n\004type\030\002 \001(\t\022\020\n\010" +
-      "strength\030\003 \001(\005\"8\n\nLocalLinks\022*\n\004link\030\001 \003" +
-      "(\0132\034.tensorflow.InterconnectLink\"Z\n\016Devi" +
-      "ceLocality\022\016\n\006bus_id\030\001 \001(\005\022\021\n\tnuma_node\030" +
-      "\002 \001(\005\022%\n\005links\030\003 \001(\0132\026.tensorflow.LocalL" +
-      "inks\"\254\001\n\020DeviceAttributes\022\014\n\004name\030\001 \001(\t\022" +
-      "\023\n\013device_type\030\002 \001(\t\022\024\n\014memory_limit\030\004 \001" +
-      "(\003\022,\n\010locality\030\005 \001(\0132\032.tensorflow.Device" +
-      "Locality\022\023\n\013incarnation\030\006 \001(\006\022\034\n\024physica" +
-      "l_device_desc\030\007 \001(\tB\227\001\n\036org.tensorflow.p" +
-      "roto.frameworkB\026DeviceAttributesProtosP\001" +
-      "ZXgithub.com/tensorflow/tensorflow/tenso" +
-      "rflow/go/core/framework/device_attribute" +
-      "s_go_proto\370\001\001b\006proto3"
-    };
-    descriptor = com.google.protobuf.Descriptors.FileDescriptor
-      .internalBuildGeneratedFileFrom(descriptorData,
-        new com.google.protobuf.Descriptors.FileDescriptor[] {
-        });
-    internal_static_tensorflow_InterconnectLink_descriptor =
-      getDescriptor().getMessageTypes().get(0);
-    internal_static_tensorflow_InterconnectLink_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_InterconnectLink_descriptor,
-        new java.lang.String[] { "DeviceId", "Type", "Strength", });
-    internal_static_tensorflow_LocalLinks_descriptor =
-      getDescriptor().getMessageTypes().get(1);
-    internal_static_tensorflow_LocalLinks_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_LocalLinks_descriptor,
-        new java.lang.String[] { "Link", });
-    internal_static_tensorflow_DeviceLocality_descriptor =
-      getDescriptor().getMessageTypes().get(2);
-    internal_static_tensorflow_DeviceLocality_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_DeviceLocality_descriptor,
-        new java.lang.String[] { "BusId", "NumaNode", "Links", });
-    internal_static_tensorflow_DeviceAttributes_descriptor =
-      getDescriptor().getMessageTypes().get(3);
-    internal_static_tensorflow_DeviceAttributes_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_DeviceAttributes_descriptor,
-        new java.lang.String[] { "Name", "DeviceType", "MemoryLimit", "Locality", "Incarnation", "PhysicalDeviceDesc", });
-  }
-
-  // @@protoc_insertion_point(outer_class_scope)
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DeviceLocality.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DeviceLocality.java
deleted file mode 100644
index 25d5784a00d..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DeviceLocality.java
+++ /dev/null
@@ -1,798 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/framework/device_attributes.proto
-
-package org.tensorflow.proto.framework;
-
-/**
- * Protobuf type {@code tensorflow.DeviceLocality}
- */
-public  final class DeviceLocality extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.DeviceLocality)
-    DeviceLocalityOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use DeviceLocality.newBuilder() to construct.
-  private DeviceLocality(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private DeviceLocality() {
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new DeviceLocality();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private DeviceLocality(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 8: {
-
-            busId_ = input.readInt32();
-            break;
-          }
-          case 16: {
-
-            numaNode_ = input.readInt32();
-            break;
-          }
-          case 26: {
-            org.tensorflow.proto.framework.LocalLinks.Builder subBuilder = null;
-            if (links_ != null) {
-              subBuilder = links_.toBuilder();
-            }
-            links_ = input.readMessage(org.tensorflow.proto.framework.LocalLinks.parser(), extensionRegistry);
-            if (subBuilder != null) {
-              subBuilder.mergeFrom(links_);
-              links_ = subBuilder.buildPartial();
-            }
-
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.framework.DeviceAttributesProtos.internal_static_tensorflow_DeviceLocality_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.framework.DeviceAttributesProtos.internal_static_tensorflow_DeviceLocality_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.framework.DeviceLocality.class, org.tensorflow.proto.framework.DeviceLocality.Builder.class);
-  }
-
-  public static final int BUS_ID_FIELD_NUMBER = 1;
-  private int busId_;
-  /**
-   * <pre>
-   * Optional bus locality of device.  Default value of 0 means
-   * no specific locality.  Specific localities are indexed from 1.
-   * </pre>
-   *
-   * <code>int32 bus_id = 1;</code>
-   */
-  public int getBusId() {
-    return busId_;
-  }
-
-  public static final int NUMA_NODE_FIELD_NUMBER = 2;
-  private int numaNode_;
-  /**
-   * <pre>
-   * Optional NUMA locality of device.
-   * </pre>
-   *
-   * <code>int32 numa_node = 2;</code>
-   */
-  public int getNumaNode() {
-    return numaNode_;
-  }
-
-  public static final int LINKS_FIELD_NUMBER = 3;
-  private org.tensorflow.proto.framework.LocalLinks links_;
-  /**
-   * <pre>
-   * Optional local interconnect links to other devices.
-   * </pre>
-   *
-   * <code>.tensorflow.LocalLinks links = 3;</code>
-   */
-  public boolean hasLinks() {
-    return links_ != null;
-  }
-  /**
-   * <pre>
-   * Optional local interconnect links to other devices.
-   * </pre>
-   *
-   * <code>.tensorflow.LocalLinks links = 3;</code>
-   */
-  public org.tensorflow.proto.framework.LocalLinks getLinks() {
-    return links_ == null ? org.tensorflow.proto.framework.LocalLinks.getDefaultInstance() : links_;
-  }
-  /**
-   * <pre>
-   * Optional local interconnect links to other devices.
-   * </pre>
-   *
-   * <code>.tensorflow.LocalLinks links = 3;</code>
-   */
-  public org.tensorflow.proto.framework.LocalLinksOrBuilder getLinksOrBuilder() {
-    return getLinks();
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    if (busId_ != 0) {
-      output.writeInt32(1, busId_);
-    }
-    if (numaNode_ != 0) {
-      output.writeInt32(2, numaNode_);
-    }
-    if (links_ != null) {
-      output.writeMessage(3, getLinks());
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (busId_ != 0) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt32Size(1, busId_);
-    }
-    if (numaNode_ != 0) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt32Size(2, numaNode_);
-    }
-    if (links_ != null) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(3, getLinks());
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.framework.DeviceLocality)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.framework.DeviceLocality other = (org.tensorflow.proto.framework.DeviceLocality) obj;
-
-    if (getBusId()
-        != other.getBusId()) return false;
-    if (getNumaNode()
-        != other.getNumaNode()) return false;
-    if (hasLinks() != other.hasLinks()) return false;
-    if (hasLinks()) {
-      if (!getLinks()
-          .equals(other.getLinks())) return false;
-    }
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    hash = (37 * hash) + BUS_ID_FIELD_NUMBER;
-    hash = (53 * hash) + getBusId();
-    hash = (37 * hash) + NUMA_NODE_FIELD_NUMBER;
-    hash = (53 * hash) + getNumaNode();
-    if (hasLinks()) {
-      hash = (37 * hash) + LINKS_FIELD_NUMBER;
-      hash = (53 * hash) + getLinks().hashCode();
-    }
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.framework.DeviceLocality parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.DeviceLocality parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.DeviceLocality parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.DeviceLocality parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.DeviceLocality parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.DeviceLocality parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.DeviceLocality parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.DeviceLocality parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.DeviceLocality parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.DeviceLocality parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.DeviceLocality parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.DeviceLocality parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.framework.DeviceLocality prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * Protobuf type {@code tensorflow.DeviceLocality}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.DeviceLocality)
-      org.tensorflow.proto.framework.DeviceLocalityOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.DeviceAttributesProtos.internal_static_tensorflow_DeviceLocality_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.DeviceAttributesProtos.internal_static_tensorflow_DeviceLocality_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.DeviceLocality.class, org.tensorflow.proto.framework.DeviceLocality.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.framework.DeviceLocality.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      busId_ = 0;
-
-      numaNode_ = 0;
-
-      if (linksBuilder_ == null) {
-        links_ = null;
-      } else {
-        links_ = null;
-        linksBuilder_ = null;
-      }
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.framework.DeviceAttributesProtos.internal_static_tensorflow_DeviceLocality_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.DeviceLocality getDefaultInstanceForType() {
-      return org.tensorflow.proto.framework.DeviceLocality.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.DeviceLocality build() {
-      org.tensorflow.proto.framework.DeviceLocality result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.DeviceLocality buildPartial() {
-      org.tensorflow.proto.framework.DeviceLocality result = new org.tensorflow.proto.framework.DeviceLocality(this);
-      result.busId_ = busId_;
-      result.numaNode_ = numaNode_;
-      if (linksBuilder_ == null) {
-        result.links_ = links_;
-      } else {
-        result.links_ = linksBuilder_.build();
-      }
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.framework.DeviceLocality) {
-        return mergeFrom((org.tensorflow.proto.framework.DeviceLocality)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.framework.DeviceLocality other) {
-      if (other == org.tensorflow.proto.framework.DeviceLocality.getDefaultInstance()) return this;
-      if (other.getBusId() != 0) {
-        setBusId(other.getBusId());
-      }
-      if (other.getNumaNode() != 0) {
-        setNumaNode(other.getNumaNode());
-      }
-      if (other.hasLinks()) {
-        mergeLinks(other.getLinks());
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.framework.DeviceLocality parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.framework.DeviceLocality) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-
-    private int busId_ ;
-    /**
-     * <pre>
-     * Optional bus locality of device.  Default value of 0 means
-     * no specific locality.  Specific localities are indexed from 1.
-     * </pre>
-     *
-     * <code>int32 bus_id = 1;</code>
-     */
-    public int getBusId() {
-      return busId_;
-    }
-    /**
-     * <pre>
-     * Optional bus locality of device.  Default value of 0 means
-     * no specific locality.  Specific localities are indexed from 1.
-     * </pre>
-     *
-     * <code>int32 bus_id = 1;</code>
-     */
-    public Builder setBusId(int value) {
-      
-      busId_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Optional bus locality of device.  Default value of 0 means
-     * no specific locality.  Specific localities are indexed from 1.
-     * </pre>
-     *
-     * <code>int32 bus_id = 1;</code>
-     */
-    public Builder clearBusId() {
-      
-      busId_ = 0;
-      onChanged();
-      return this;
-    }
-
-    private int numaNode_ ;
-    /**
-     * <pre>
-     * Optional NUMA locality of device.
-     * </pre>
-     *
-     * <code>int32 numa_node = 2;</code>
-     */
-    public int getNumaNode() {
-      return numaNode_;
-    }
-    /**
-     * <pre>
-     * Optional NUMA locality of device.
-     * </pre>
-     *
-     * <code>int32 numa_node = 2;</code>
-     */
-    public Builder setNumaNode(int value) {
-      
-      numaNode_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Optional NUMA locality of device.
-     * </pre>
-     *
-     * <code>int32 numa_node = 2;</code>
-     */
-    public Builder clearNumaNode() {
-      
-      numaNode_ = 0;
-      onChanged();
-      return this;
-    }
-
-    private org.tensorflow.proto.framework.LocalLinks links_;
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.LocalLinks, org.tensorflow.proto.framework.LocalLinks.Builder, org.tensorflow.proto.framework.LocalLinksOrBuilder> linksBuilder_;
-    /**
-     * <pre>
-     * Optional local interconnect links to other devices.
-     * </pre>
-     *
-     * <code>.tensorflow.LocalLinks links = 3;</code>
-     */
-    public boolean hasLinks() {
-      return linksBuilder_ != null || links_ != null;
-    }
-    /**
-     * <pre>
-     * Optional local interconnect links to other devices.
-     * </pre>
-     *
-     * <code>.tensorflow.LocalLinks links = 3;</code>
-     */
-    public org.tensorflow.proto.framework.LocalLinks getLinks() {
-      if (linksBuilder_ == null) {
-        return links_ == null ? org.tensorflow.proto.framework.LocalLinks.getDefaultInstance() : links_;
-      } else {
-        return linksBuilder_.getMessage();
-      }
-    }
-    /**
-     * <pre>
-     * Optional local interconnect links to other devices.
-     * </pre>
-     *
-     * <code>.tensorflow.LocalLinks links = 3;</code>
-     */
-    public Builder setLinks(org.tensorflow.proto.framework.LocalLinks value) {
-      if (linksBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        links_ = value;
-        onChanged();
-      } else {
-        linksBuilder_.setMessage(value);
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Optional local interconnect links to other devices.
-     * </pre>
-     *
-     * <code>.tensorflow.LocalLinks links = 3;</code>
-     */
-    public Builder setLinks(
-        org.tensorflow.proto.framework.LocalLinks.Builder builderForValue) {
-      if (linksBuilder_ == null) {
-        links_ = builderForValue.build();
-        onChanged();
-      } else {
-        linksBuilder_.setMessage(builderForValue.build());
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Optional local interconnect links to other devices.
-     * </pre>
-     *
-     * <code>.tensorflow.LocalLinks links = 3;</code>
-     */
-    public Builder mergeLinks(org.tensorflow.proto.framework.LocalLinks value) {
-      if (linksBuilder_ == null) {
-        if (links_ != null) {
-          links_ =
-            org.tensorflow.proto.framework.LocalLinks.newBuilder(links_).mergeFrom(value).buildPartial();
-        } else {
-          links_ = value;
-        }
-        onChanged();
-      } else {
-        linksBuilder_.mergeFrom(value);
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Optional local interconnect links to other devices.
-     * </pre>
-     *
-     * <code>.tensorflow.LocalLinks links = 3;</code>
-     */
-    public Builder clearLinks() {
-      if (linksBuilder_ == null) {
-        links_ = null;
-        onChanged();
-      } else {
-        links_ = null;
-        linksBuilder_ = null;
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Optional local interconnect links to other devices.
-     * </pre>
-     *
-     * <code>.tensorflow.LocalLinks links = 3;</code>
-     */
-    public org.tensorflow.proto.framework.LocalLinks.Builder getLinksBuilder() {
-      
-      onChanged();
-      return getLinksFieldBuilder().getBuilder();
-    }
-    /**
-     * <pre>
-     * Optional local interconnect links to other devices.
-     * </pre>
-     *
-     * <code>.tensorflow.LocalLinks links = 3;</code>
-     */
-    public org.tensorflow.proto.framework.LocalLinksOrBuilder getLinksOrBuilder() {
-      if (linksBuilder_ != null) {
-        return linksBuilder_.getMessageOrBuilder();
-      } else {
-        return links_ == null ?
-            org.tensorflow.proto.framework.LocalLinks.getDefaultInstance() : links_;
-      }
-    }
-    /**
-     * <pre>
-     * Optional local interconnect links to other devices.
-     * </pre>
-     *
-     * <code>.tensorflow.LocalLinks links = 3;</code>
-     */
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.LocalLinks, org.tensorflow.proto.framework.LocalLinks.Builder, org.tensorflow.proto.framework.LocalLinksOrBuilder> 
-        getLinksFieldBuilder() {
-      if (linksBuilder_ == null) {
-        linksBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-            org.tensorflow.proto.framework.LocalLinks, org.tensorflow.proto.framework.LocalLinks.Builder, org.tensorflow.proto.framework.LocalLinksOrBuilder>(
-                getLinks(),
-                getParentForChildren(),
-                isClean());
-        links_ = null;
-      }
-      return linksBuilder_;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.DeviceLocality)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.DeviceLocality)
-  private static final org.tensorflow.proto.framework.DeviceLocality DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.DeviceLocality();
-  }
-
-  public static org.tensorflow.proto.framework.DeviceLocality getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<DeviceLocality>
-      PARSER = new com.google.protobuf.AbstractParser<DeviceLocality>() {
-    @java.lang.Override
-    public DeviceLocality parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new DeviceLocality(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<DeviceLocality> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<DeviceLocality> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.framework.DeviceLocality getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DeviceLocalityOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DeviceLocalityOrBuilder.java
deleted file mode 100644
index cfb7b2c5287..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DeviceLocalityOrBuilder.java
+++ /dev/null
@@ -1,53 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/framework/device_attributes.proto
-
-package org.tensorflow.proto.framework;
-
-public interface DeviceLocalityOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.DeviceLocality)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <pre>
-   * Optional bus locality of device.  Default value of 0 means
-   * no specific locality.  Specific localities are indexed from 1.
-   * </pre>
-   *
-   * <code>int32 bus_id = 1;</code>
-   */
-  int getBusId();
-
-  /**
-   * <pre>
-   * Optional NUMA locality of device.
-   * </pre>
-   *
-   * <code>int32 numa_node = 2;</code>
-   */
-  int getNumaNode();
-
-  /**
-   * <pre>
-   * Optional local interconnect links to other devices.
-   * </pre>
-   *
-   * <code>.tensorflow.LocalLinks links = 3;</code>
-   */
-  boolean hasLinks();
-  /**
-   * <pre>
-   * Optional local interconnect links to other devices.
-   * </pre>
-   *
-   * <code>.tensorflow.LocalLinks links = 3;</code>
-   */
-  org.tensorflow.proto.framework.LocalLinks getLinks();
-  /**
-   * <pre>
-   * Optional local interconnect links to other devices.
-   * </pre>
-   *
-   * <code>.tensorflow.LocalLinks links = 3;</code>
-   */
-  org.tensorflow.proto.framework.LocalLinksOrBuilder getLinksOrBuilder();
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DeviceOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DeviceOrBuilder.java
deleted file mode 100644
index 3ce193459ff..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DeviceOrBuilder.java
+++ /dev/null
@@ -1,90 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/trace_events.proto
-
-package org.tensorflow.proto.framework;
-
-public interface DeviceOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.profiler.Device)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <pre>
-   * The name of the device.
-   * </pre>
-   *
-   * <code>string name = 1;</code>
-   */
-  java.lang.String getName();
-  /**
-   * <pre>
-   * The name of the device.
-   * </pre>
-   *
-   * <code>string name = 1;</code>
-   */
-  com.google.protobuf.ByteString
-      getNameBytes();
-
-  /**
-   * <pre>
-   * The id of this device, unique in a single trace.
-   * </pre>
-   *
-   * <code>uint32 device_id = 2;</code>
-   */
-  int getDeviceId();
-
-  /**
-   * <pre>
-   * The resources on this device, keyed by resource_id;
-   * </pre>
-   *
-   * <code>map&lt;uint32, .tensorflow.profiler.Resource&gt; resources = 3;</code>
-   */
-  int getResourcesCount();
-  /**
-   * <pre>
-   * The resources on this device, keyed by resource_id;
-   * </pre>
-   *
-   * <code>map&lt;uint32, .tensorflow.profiler.Resource&gt; resources = 3;</code>
-   */
-  boolean containsResources(
-      int key);
-  /**
-   * Use {@link #getResourcesMap()} instead.
-   */
-  @java.lang.Deprecated
-  java.util.Map<java.lang.Integer, org.tensorflow.proto.framework.Resource>
-  getResources();
-  /**
-   * <pre>
-   * The resources on this device, keyed by resource_id;
-   * </pre>
-   *
-   * <code>map&lt;uint32, .tensorflow.profiler.Resource&gt; resources = 3;</code>
-   */
-  java.util.Map<java.lang.Integer, org.tensorflow.proto.framework.Resource>
-  getResourcesMap();
-  /**
-   * <pre>
-   * The resources on this device, keyed by resource_id;
-   * </pre>
-   *
-   * <code>map&lt;uint32, .tensorflow.profiler.Resource&gt; resources = 3;</code>
-   */
-
-  org.tensorflow.proto.framework.Resource getResourcesOrDefault(
-      int key,
-      org.tensorflow.proto.framework.Resource defaultValue);
-  /**
-   * <pre>
-   * The resources on this device, keyed by resource_id;
-   * </pre>
-   *
-   * <code>map&lt;uint32, .tensorflow.profiler.Resource&gt; resources = 3;</code>
-   */
-
-  org.tensorflow.proto.framework.Resource getResourcesOrThrow(
-      int key);
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DeviceProperties.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DeviceProperties.java
deleted file mode 100644
index a6e16b28d1f..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DeviceProperties.java
+++ /dev/null
@@ -1,1885 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/device_properties.proto
-
-package org.tensorflow.proto.framework;
-
-/**
- * Protobuf type {@code tensorflow.DeviceProperties}
- */
-public  final class DeviceProperties extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.DeviceProperties)
-    DevicePropertiesOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use DeviceProperties.newBuilder() to construct.
-  private DeviceProperties(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private DeviceProperties() {
-    type_ = "";
-    vendor_ = "";
-    model_ = "";
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new DeviceProperties();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private DeviceProperties(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    int mutable_bitField0_ = 0;
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 10: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            type_ = s;
-            break;
-          }
-          case 18: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            vendor_ = s;
-            break;
-          }
-          case 26: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            model_ = s;
-            break;
-          }
-          case 32: {
-
-            frequency_ = input.readInt64();
-            break;
-          }
-          case 40: {
-
-            numCores_ = input.readInt64();
-            break;
-          }
-          case 50: {
-            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
-              environment_ = com.google.protobuf.MapField.newMapField(
-                  EnvironmentDefaultEntryHolder.defaultEntry);
-              mutable_bitField0_ |= 0x00000001;
-            }
-            com.google.protobuf.MapEntry<java.lang.String, java.lang.String>
-            environment__ = input.readMessage(
-                EnvironmentDefaultEntryHolder.defaultEntry.getParserForType(), extensionRegistry);
-            environment_.getMutableMap().put(
-                environment__.getKey(), environment__.getValue());
-            break;
-          }
-          case 56: {
-
-            numRegisters_ = input.readInt64();
-            break;
-          }
-          case 64: {
-
-            l1CacheSize_ = input.readInt64();
-            break;
-          }
-          case 72: {
-
-            l2CacheSize_ = input.readInt64();
-            break;
-          }
-          case 80: {
-
-            l3CacheSize_ = input.readInt64();
-            break;
-          }
-          case 88: {
-
-            sharedMemorySizePerMultiprocessor_ = input.readInt64();
-            break;
-          }
-          case 96: {
-
-            memorySize_ = input.readInt64();
-            break;
-          }
-          case 104: {
-
-            bandwidth_ = input.readInt64();
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.framework.DevicePropertiesProtos.internal_static_tensorflow_DeviceProperties_descriptor;
-  }
-
-  @SuppressWarnings({"rawtypes"})
-  @java.lang.Override
-  protected com.google.protobuf.MapField internalGetMapField(
-      int number) {
-    switch (number) {
-      case 6:
-        return internalGetEnvironment();
-      default:
-        throw new RuntimeException(
-            "Invalid map field number: " + number);
-    }
-  }
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.framework.DevicePropertiesProtos.internal_static_tensorflow_DeviceProperties_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.framework.DeviceProperties.class, org.tensorflow.proto.framework.DeviceProperties.Builder.class);
-  }
-
-  public static final int TYPE_FIELD_NUMBER = 1;
-  private volatile java.lang.Object type_;
-  /**
-   * <pre>
-   * Device type (CPU, GPU, ...)
-   * </pre>
-   *
-   * <code>string type = 1;</code>
-   */
-  public java.lang.String getType() {
-    java.lang.Object ref = type_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      type_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * Device type (CPU, GPU, ...)
-   * </pre>
-   *
-   * <code>string type = 1;</code>
-   */
-  public com.google.protobuf.ByteString
-      getTypeBytes() {
-    java.lang.Object ref = type_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      type_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int VENDOR_FIELD_NUMBER = 2;
-  private volatile java.lang.Object vendor_;
-  /**
-   * <pre>
-   * Vendor (Intel, nvidia, ...)
-   * </pre>
-   *
-   * <code>string vendor = 2;</code>
-   */
-  public java.lang.String getVendor() {
-    java.lang.Object ref = vendor_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      vendor_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * Vendor (Intel, nvidia, ...)
-   * </pre>
-   *
-   * <code>string vendor = 2;</code>
-   */
-  public com.google.protobuf.ByteString
-      getVendorBytes() {
-    java.lang.Object ref = vendor_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      vendor_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int MODEL_FIELD_NUMBER = 3;
-  private volatile java.lang.Object model_;
-  /**
-   * <pre>
-   * Model (Haswell, K40, ...)
-   * </pre>
-   *
-   * <code>string model = 3;</code>
-   */
-  public java.lang.String getModel() {
-    java.lang.Object ref = model_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      model_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * Model (Haswell, K40, ...)
-   * </pre>
-   *
-   * <code>string model = 3;</code>
-   */
-  public com.google.protobuf.ByteString
-      getModelBytes() {
-    java.lang.Object ref = model_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      model_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int FREQUENCY_FIELD_NUMBER = 4;
-  private long frequency_;
-  /**
-   * <pre>
-   * Core Frequency in Mhz
-   * </pre>
-   *
-   * <code>int64 frequency = 4;</code>
-   */
-  public long getFrequency() {
-    return frequency_;
-  }
-
-  public static final int NUM_CORES_FIELD_NUMBER = 5;
-  private long numCores_;
-  /**
-   * <pre>
-   * Number of cores
-   * </pre>
-   *
-   * <code>int64 num_cores = 5;</code>
-   */
-  public long getNumCores() {
-    return numCores_;
-  }
-
-  public static final int ENVIRONMENT_FIELD_NUMBER = 6;
-  private static final class EnvironmentDefaultEntryHolder {
-    static final com.google.protobuf.MapEntry<
-        java.lang.String, java.lang.String> defaultEntry =
-            com.google.protobuf.MapEntry
-            .<java.lang.String, java.lang.String>newDefaultInstance(
-                org.tensorflow.proto.framework.DevicePropertiesProtos.internal_static_tensorflow_DeviceProperties_EnvironmentEntry_descriptor, 
-                com.google.protobuf.WireFormat.FieldType.STRING,
-                "",
-                com.google.protobuf.WireFormat.FieldType.STRING,
-                "");
-  }
-  private com.google.protobuf.MapField<
-      java.lang.String, java.lang.String> environment_;
-  private com.google.protobuf.MapField<java.lang.String, java.lang.String>
-  internalGetEnvironment() {
-    if (environment_ == null) {
-      return com.google.protobuf.MapField.emptyMapField(
-          EnvironmentDefaultEntryHolder.defaultEntry);
-    }
-    return environment_;
-  }
-
-  public int getEnvironmentCount() {
-    return internalGetEnvironment().getMap().size();
-  }
-  /**
-   * <pre>
-   * Version of the tools and libraries used with this device (e.g. gcc 4.9,
-   * cudnn 5.1)
-   * </pre>
-   *
-   * <code>map&lt;string, string&gt; environment = 6;</code>
-   */
-
-  public boolean containsEnvironment(
-      java.lang.String key) {
-    if (key == null) { throw new java.lang.NullPointerException(); }
-    return internalGetEnvironment().getMap().containsKey(key);
-  }
-  /**
-   * Use {@link #getEnvironmentMap()} instead.
-   */
-  @java.lang.Deprecated
-  public java.util.Map<java.lang.String, java.lang.String> getEnvironment() {
-    return getEnvironmentMap();
-  }
-  /**
-   * <pre>
-   * Version of the tools and libraries used with this device (e.g. gcc 4.9,
-   * cudnn 5.1)
-   * </pre>
-   *
-   * <code>map&lt;string, string&gt; environment = 6;</code>
-   */
-
-  public java.util.Map<java.lang.String, java.lang.String> getEnvironmentMap() {
-    return internalGetEnvironment().getMap();
-  }
-  /**
-   * <pre>
-   * Version of the tools and libraries used with this device (e.g. gcc 4.9,
-   * cudnn 5.1)
-   * </pre>
-   *
-   * <code>map&lt;string, string&gt; environment = 6;</code>
-   */
-
-  public java.lang.String getEnvironmentOrDefault(
-      java.lang.String key,
-      java.lang.String defaultValue) {
-    if (key == null) { throw new java.lang.NullPointerException(); }
-    java.util.Map<java.lang.String, java.lang.String> map =
-        internalGetEnvironment().getMap();
-    return map.containsKey(key) ? map.get(key) : defaultValue;
-  }
-  /**
-   * <pre>
-   * Version of the tools and libraries used with this device (e.g. gcc 4.9,
-   * cudnn 5.1)
-   * </pre>
-   *
-   * <code>map&lt;string, string&gt; environment = 6;</code>
-   */
-
-  public java.lang.String getEnvironmentOrThrow(
-      java.lang.String key) {
-    if (key == null) { throw new java.lang.NullPointerException(); }
-    java.util.Map<java.lang.String, java.lang.String> map =
-        internalGetEnvironment().getMap();
-    if (!map.containsKey(key)) {
-      throw new java.lang.IllegalArgumentException();
-    }
-    return map.get(key);
-  }
-
-  public static final int NUM_REGISTERS_FIELD_NUMBER = 7;
-  private long numRegisters_;
-  /**
-   * <pre>
-   * Number of registers per core.
-   * </pre>
-   *
-   * <code>int64 num_registers = 7;</code>
-   */
-  public long getNumRegisters() {
-    return numRegisters_;
-  }
-
-  public static final int L1_CACHE_SIZE_FIELD_NUMBER = 8;
-  private long l1CacheSize_;
-  /**
-   * <pre>
-   * L1 cache size in bytes
-   * </pre>
-   *
-   * <code>int64 l1_cache_size = 8;</code>
-   */
-  public long getL1CacheSize() {
-    return l1CacheSize_;
-  }
-
-  public static final int L2_CACHE_SIZE_FIELD_NUMBER = 9;
-  private long l2CacheSize_;
-  /**
-   * <pre>
-   * L2 cache size in bytes
-   * </pre>
-   *
-   * <code>int64 l2_cache_size = 9;</code>
-   */
-  public long getL2CacheSize() {
-    return l2CacheSize_;
-  }
-
-  public static final int L3_CACHE_SIZE_FIELD_NUMBER = 10;
-  private long l3CacheSize_;
-  /**
-   * <pre>
-   * L3 cache size in bytes
-   * </pre>
-   *
-   * <code>int64 l3_cache_size = 10;</code>
-   */
-  public long getL3CacheSize() {
-    return l3CacheSize_;
-  }
-
-  public static final int SHARED_MEMORY_SIZE_PER_MULTIPROCESSOR_FIELD_NUMBER = 11;
-  private long sharedMemorySizePerMultiprocessor_;
-  /**
-   * <pre>
-   * Shared memory size per multiprocessor in bytes. This field is
-   * applicable to GPUs only.
-   * </pre>
-   *
-   * <code>int64 shared_memory_size_per_multiprocessor = 11;</code>
-   */
-  public long getSharedMemorySizePerMultiprocessor() {
-    return sharedMemorySizePerMultiprocessor_;
-  }
-
-  public static final int MEMORY_SIZE_FIELD_NUMBER = 12;
-  private long memorySize_;
-  /**
-   * <pre>
-   * Memory size in bytes
-   * </pre>
-   *
-   * <code>int64 memory_size = 12;</code>
-   */
-  public long getMemorySize() {
-    return memorySize_;
-  }
-
-  public static final int BANDWIDTH_FIELD_NUMBER = 13;
-  private long bandwidth_;
-  /**
-   * <pre>
-   * Memory bandwidth in KB/s
-   * </pre>
-   *
-   * <code>int64 bandwidth = 13;</code>
-   */
-  public long getBandwidth() {
-    return bandwidth_;
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    if (!getTypeBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 1, type_);
-    }
-    if (!getVendorBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 2, vendor_);
-    }
-    if (!getModelBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 3, model_);
-    }
-    if (frequency_ != 0L) {
-      output.writeInt64(4, frequency_);
-    }
-    if (numCores_ != 0L) {
-      output.writeInt64(5, numCores_);
-    }
-    com.google.protobuf.GeneratedMessageV3
-      .serializeStringMapTo(
-        output,
-        internalGetEnvironment(),
-        EnvironmentDefaultEntryHolder.defaultEntry,
-        6);
-    if (numRegisters_ != 0L) {
-      output.writeInt64(7, numRegisters_);
-    }
-    if (l1CacheSize_ != 0L) {
-      output.writeInt64(8, l1CacheSize_);
-    }
-    if (l2CacheSize_ != 0L) {
-      output.writeInt64(9, l2CacheSize_);
-    }
-    if (l3CacheSize_ != 0L) {
-      output.writeInt64(10, l3CacheSize_);
-    }
-    if (sharedMemorySizePerMultiprocessor_ != 0L) {
-      output.writeInt64(11, sharedMemorySizePerMultiprocessor_);
-    }
-    if (memorySize_ != 0L) {
-      output.writeInt64(12, memorySize_);
-    }
-    if (bandwidth_ != 0L) {
-      output.writeInt64(13, bandwidth_);
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (!getTypeBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, type_);
-    }
-    if (!getVendorBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(2, vendor_);
-    }
-    if (!getModelBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(3, model_);
-    }
-    if (frequency_ != 0L) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt64Size(4, frequency_);
-    }
-    if (numCores_ != 0L) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt64Size(5, numCores_);
-    }
-    for (java.util.Map.Entry<java.lang.String, java.lang.String> entry
-         : internalGetEnvironment().getMap().entrySet()) {
-      com.google.protobuf.MapEntry<java.lang.String, java.lang.String>
-      environment__ = EnvironmentDefaultEntryHolder.defaultEntry.newBuilderForType()
-          .setKey(entry.getKey())
-          .setValue(entry.getValue())
-          .build();
-      size += com.google.protobuf.CodedOutputStream
-          .computeMessageSize(6, environment__);
-    }
-    if (numRegisters_ != 0L) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt64Size(7, numRegisters_);
-    }
-    if (l1CacheSize_ != 0L) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt64Size(8, l1CacheSize_);
-    }
-    if (l2CacheSize_ != 0L) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt64Size(9, l2CacheSize_);
-    }
-    if (l3CacheSize_ != 0L) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt64Size(10, l3CacheSize_);
-    }
-    if (sharedMemorySizePerMultiprocessor_ != 0L) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt64Size(11, sharedMemorySizePerMultiprocessor_);
-    }
-    if (memorySize_ != 0L) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt64Size(12, memorySize_);
-    }
-    if (bandwidth_ != 0L) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt64Size(13, bandwidth_);
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.framework.DeviceProperties)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.framework.DeviceProperties other = (org.tensorflow.proto.framework.DeviceProperties) obj;
-
-    if (!getType()
-        .equals(other.getType())) return false;
-    if (!getVendor()
-        .equals(other.getVendor())) return false;
-    if (!getModel()
-        .equals(other.getModel())) return false;
-    if (getFrequency()
-        != other.getFrequency()) return false;
-    if (getNumCores()
-        != other.getNumCores()) return false;
-    if (!internalGetEnvironment().equals(
-        other.internalGetEnvironment())) return false;
-    if (getNumRegisters()
-        != other.getNumRegisters()) return false;
-    if (getL1CacheSize()
-        != other.getL1CacheSize()) return false;
-    if (getL2CacheSize()
-        != other.getL2CacheSize()) return false;
-    if (getL3CacheSize()
-        != other.getL3CacheSize()) return false;
-    if (getSharedMemorySizePerMultiprocessor()
-        != other.getSharedMemorySizePerMultiprocessor()) return false;
-    if (getMemorySize()
-        != other.getMemorySize()) return false;
-    if (getBandwidth()
-        != other.getBandwidth()) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    hash = (37 * hash) + TYPE_FIELD_NUMBER;
-    hash = (53 * hash) + getType().hashCode();
-    hash = (37 * hash) + VENDOR_FIELD_NUMBER;
-    hash = (53 * hash) + getVendor().hashCode();
-    hash = (37 * hash) + MODEL_FIELD_NUMBER;
-    hash = (53 * hash) + getModel().hashCode();
-    hash = (37 * hash) + FREQUENCY_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-        getFrequency());
-    hash = (37 * hash) + NUM_CORES_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-        getNumCores());
-    if (!internalGetEnvironment().getMap().isEmpty()) {
-      hash = (37 * hash) + ENVIRONMENT_FIELD_NUMBER;
-      hash = (53 * hash) + internalGetEnvironment().hashCode();
-    }
-    hash = (37 * hash) + NUM_REGISTERS_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-        getNumRegisters());
-    hash = (37 * hash) + L1_CACHE_SIZE_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-        getL1CacheSize());
-    hash = (37 * hash) + L2_CACHE_SIZE_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-        getL2CacheSize());
-    hash = (37 * hash) + L3_CACHE_SIZE_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-        getL3CacheSize());
-    hash = (37 * hash) + SHARED_MEMORY_SIZE_PER_MULTIPROCESSOR_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-        getSharedMemorySizePerMultiprocessor());
-    hash = (37 * hash) + MEMORY_SIZE_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-        getMemorySize());
-    hash = (37 * hash) + BANDWIDTH_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-        getBandwidth());
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.framework.DeviceProperties parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.DeviceProperties parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.DeviceProperties parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.DeviceProperties parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.DeviceProperties parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.DeviceProperties parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.DeviceProperties parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.DeviceProperties parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.DeviceProperties parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.DeviceProperties parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.DeviceProperties parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.DeviceProperties parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.framework.DeviceProperties prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * Protobuf type {@code tensorflow.DeviceProperties}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.DeviceProperties)
-      org.tensorflow.proto.framework.DevicePropertiesOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.DevicePropertiesProtos.internal_static_tensorflow_DeviceProperties_descriptor;
-    }
-
-    @SuppressWarnings({"rawtypes"})
-    protected com.google.protobuf.MapField internalGetMapField(
-        int number) {
-      switch (number) {
-        case 6:
-          return internalGetEnvironment();
-        default:
-          throw new RuntimeException(
-              "Invalid map field number: " + number);
-      }
-    }
-    @SuppressWarnings({"rawtypes"})
-    protected com.google.protobuf.MapField internalGetMutableMapField(
-        int number) {
-      switch (number) {
-        case 6:
-          return internalGetMutableEnvironment();
-        default:
-          throw new RuntimeException(
-              "Invalid map field number: " + number);
-      }
-    }
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.DevicePropertiesProtos.internal_static_tensorflow_DeviceProperties_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.DeviceProperties.class, org.tensorflow.proto.framework.DeviceProperties.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.framework.DeviceProperties.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      type_ = "";
-
-      vendor_ = "";
-
-      model_ = "";
-
-      frequency_ = 0L;
-
-      numCores_ = 0L;
-
-      internalGetMutableEnvironment().clear();
-      numRegisters_ = 0L;
-
-      l1CacheSize_ = 0L;
-
-      l2CacheSize_ = 0L;
-
-      l3CacheSize_ = 0L;
-
-      sharedMemorySizePerMultiprocessor_ = 0L;
-
-      memorySize_ = 0L;
-
-      bandwidth_ = 0L;
-
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.framework.DevicePropertiesProtos.internal_static_tensorflow_DeviceProperties_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.DeviceProperties getDefaultInstanceForType() {
-      return org.tensorflow.proto.framework.DeviceProperties.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.DeviceProperties build() {
-      org.tensorflow.proto.framework.DeviceProperties result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.DeviceProperties buildPartial() {
-      org.tensorflow.proto.framework.DeviceProperties result = new org.tensorflow.proto.framework.DeviceProperties(this);
-      int from_bitField0_ = bitField0_;
-      result.type_ = type_;
-      result.vendor_ = vendor_;
-      result.model_ = model_;
-      result.frequency_ = frequency_;
-      result.numCores_ = numCores_;
-      result.environment_ = internalGetEnvironment();
-      result.environment_.makeImmutable();
-      result.numRegisters_ = numRegisters_;
-      result.l1CacheSize_ = l1CacheSize_;
-      result.l2CacheSize_ = l2CacheSize_;
-      result.l3CacheSize_ = l3CacheSize_;
-      result.sharedMemorySizePerMultiprocessor_ = sharedMemorySizePerMultiprocessor_;
-      result.memorySize_ = memorySize_;
-      result.bandwidth_ = bandwidth_;
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.framework.DeviceProperties) {
-        return mergeFrom((org.tensorflow.proto.framework.DeviceProperties)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.framework.DeviceProperties other) {
-      if (other == org.tensorflow.proto.framework.DeviceProperties.getDefaultInstance()) return this;
-      if (!other.getType().isEmpty()) {
-        type_ = other.type_;
-        onChanged();
-      }
-      if (!other.getVendor().isEmpty()) {
-        vendor_ = other.vendor_;
-        onChanged();
-      }
-      if (!other.getModel().isEmpty()) {
-        model_ = other.model_;
-        onChanged();
-      }
-      if (other.getFrequency() != 0L) {
-        setFrequency(other.getFrequency());
-      }
-      if (other.getNumCores() != 0L) {
-        setNumCores(other.getNumCores());
-      }
-      internalGetMutableEnvironment().mergeFrom(
-          other.internalGetEnvironment());
-      if (other.getNumRegisters() != 0L) {
-        setNumRegisters(other.getNumRegisters());
-      }
-      if (other.getL1CacheSize() != 0L) {
-        setL1CacheSize(other.getL1CacheSize());
-      }
-      if (other.getL2CacheSize() != 0L) {
-        setL2CacheSize(other.getL2CacheSize());
-      }
-      if (other.getL3CacheSize() != 0L) {
-        setL3CacheSize(other.getL3CacheSize());
-      }
-      if (other.getSharedMemorySizePerMultiprocessor() != 0L) {
-        setSharedMemorySizePerMultiprocessor(other.getSharedMemorySizePerMultiprocessor());
-      }
-      if (other.getMemorySize() != 0L) {
-        setMemorySize(other.getMemorySize());
-      }
-      if (other.getBandwidth() != 0L) {
-        setBandwidth(other.getBandwidth());
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.framework.DeviceProperties parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.framework.DeviceProperties) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-    private int bitField0_;
-
-    private java.lang.Object type_ = "";
-    /**
-     * <pre>
-     * Device type (CPU, GPU, ...)
-     * </pre>
-     *
-     * <code>string type = 1;</code>
-     */
-    public java.lang.String getType() {
-      java.lang.Object ref = type_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        type_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Device type (CPU, GPU, ...)
-     * </pre>
-     *
-     * <code>string type = 1;</code>
-     */
-    public com.google.protobuf.ByteString
-        getTypeBytes() {
-      java.lang.Object ref = type_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        type_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Device type (CPU, GPU, ...)
-     * </pre>
-     *
-     * <code>string type = 1;</code>
-     */
-    public Builder setType(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      type_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Device type (CPU, GPU, ...)
-     * </pre>
-     *
-     * <code>string type = 1;</code>
-     */
-    public Builder clearType() {
-      
-      type_ = getDefaultInstance().getType();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Device type (CPU, GPU, ...)
-     * </pre>
-     *
-     * <code>string type = 1;</code>
-     */
-    public Builder setTypeBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      type_ = value;
-      onChanged();
-      return this;
-    }
-
-    private java.lang.Object vendor_ = "";
-    /**
-     * <pre>
-     * Vendor (Intel, nvidia, ...)
-     * </pre>
-     *
-     * <code>string vendor = 2;</code>
-     */
-    public java.lang.String getVendor() {
-      java.lang.Object ref = vendor_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        vendor_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Vendor (Intel, nvidia, ...)
-     * </pre>
-     *
-     * <code>string vendor = 2;</code>
-     */
-    public com.google.protobuf.ByteString
-        getVendorBytes() {
-      java.lang.Object ref = vendor_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        vendor_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Vendor (Intel, nvidia, ...)
-     * </pre>
-     *
-     * <code>string vendor = 2;</code>
-     */
-    public Builder setVendor(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      vendor_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Vendor (Intel, nvidia, ...)
-     * </pre>
-     *
-     * <code>string vendor = 2;</code>
-     */
-    public Builder clearVendor() {
-      
-      vendor_ = getDefaultInstance().getVendor();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Vendor (Intel, nvidia, ...)
-     * </pre>
-     *
-     * <code>string vendor = 2;</code>
-     */
-    public Builder setVendorBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      vendor_ = value;
-      onChanged();
-      return this;
-    }
-
-    private java.lang.Object model_ = "";
-    /**
-     * <pre>
-     * Model (Haswell, K40, ...)
-     * </pre>
-     *
-     * <code>string model = 3;</code>
-     */
-    public java.lang.String getModel() {
-      java.lang.Object ref = model_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        model_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Model (Haswell, K40, ...)
-     * </pre>
-     *
-     * <code>string model = 3;</code>
-     */
-    public com.google.protobuf.ByteString
-        getModelBytes() {
-      java.lang.Object ref = model_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        model_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Model (Haswell, K40, ...)
-     * </pre>
-     *
-     * <code>string model = 3;</code>
-     */
-    public Builder setModel(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      model_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Model (Haswell, K40, ...)
-     * </pre>
-     *
-     * <code>string model = 3;</code>
-     */
-    public Builder clearModel() {
-      
-      model_ = getDefaultInstance().getModel();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Model (Haswell, K40, ...)
-     * </pre>
-     *
-     * <code>string model = 3;</code>
-     */
-    public Builder setModelBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      model_ = value;
-      onChanged();
-      return this;
-    }
-
-    private long frequency_ ;
-    /**
-     * <pre>
-     * Core Frequency in Mhz
-     * </pre>
-     *
-     * <code>int64 frequency = 4;</code>
-     */
-    public long getFrequency() {
-      return frequency_;
-    }
-    /**
-     * <pre>
-     * Core Frequency in Mhz
-     * </pre>
-     *
-     * <code>int64 frequency = 4;</code>
-     */
-    public Builder setFrequency(long value) {
-      
-      frequency_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Core Frequency in Mhz
-     * </pre>
-     *
-     * <code>int64 frequency = 4;</code>
-     */
-    public Builder clearFrequency() {
-      
-      frequency_ = 0L;
-      onChanged();
-      return this;
-    }
-
-    private long numCores_ ;
-    /**
-     * <pre>
-     * Number of cores
-     * </pre>
-     *
-     * <code>int64 num_cores = 5;</code>
-     */
-    public long getNumCores() {
-      return numCores_;
-    }
-    /**
-     * <pre>
-     * Number of cores
-     * </pre>
-     *
-     * <code>int64 num_cores = 5;</code>
-     */
-    public Builder setNumCores(long value) {
-      
-      numCores_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Number of cores
-     * </pre>
-     *
-     * <code>int64 num_cores = 5;</code>
-     */
-    public Builder clearNumCores() {
-      
-      numCores_ = 0L;
-      onChanged();
-      return this;
-    }
-
-    private com.google.protobuf.MapField<
-        java.lang.String, java.lang.String> environment_;
-    private com.google.protobuf.MapField<java.lang.String, java.lang.String>
-    internalGetEnvironment() {
-      if (environment_ == null) {
-        return com.google.protobuf.MapField.emptyMapField(
-            EnvironmentDefaultEntryHolder.defaultEntry);
-      }
-      return environment_;
-    }
-    private com.google.protobuf.MapField<java.lang.String, java.lang.String>
-    internalGetMutableEnvironment() {
-      onChanged();;
-      if (environment_ == null) {
-        environment_ = com.google.protobuf.MapField.newMapField(
-            EnvironmentDefaultEntryHolder.defaultEntry);
-      }
-      if (!environment_.isMutable()) {
-        environment_ = environment_.copy();
-      }
-      return environment_;
-    }
-
-    public int getEnvironmentCount() {
-      return internalGetEnvironment().getMap().size();
-    }
-    /**
-     * <pre>
-     * Version of the tools and libraries used with this device (e.g. gcc 4.9,
-     * cudnn 5.1)
-     * </pre>
-     *
-     * <code>map&lt;string, string&gt; environment = 6;</code>
-     */
-
-    public boolean containsEnvironment(
-        java.lang.String key) {
-      if (key == null) { throw new java.lang.NullPointerException(); }
-      return internalGetEnvironment().getMap().containsKey(key);
-    }
-    /**
-     * Use {@link #getEnvironmentMap()} instead.
-     */
-    @java.lang.Deprecated
-    public java.util.Map<java.lang.String, java.lang.String> getEnvironment() {
-      return getEnvironmentMap();
-    }
-    /**
-     * <pre>
-     * Version of the tools and libraries used with this device (e.g. gcc 4.9,
-     * cudnn 5.1)
-     * </pre>
-     *
-     * <code>map&lt;string, string&gt; environment = 6;</code>
-     */
-
-    public java.util.Map<java.lang.String, java.lang.String> getEnvironmentMap() {
-      return internalGetEnvironment().getMap();
-    }
-    /**
-     * <pre>
-     * Version of the tools and libraries used with this device (e.g. gcc 4.9,
-     * cudnn 5.1)
-     * </pre>
-     *
-     * <code>map&lt;string, string&gt; environment = 6;</code>
-     */
-
-    public java.lang.String getEnvironmentOrDefault(
-        java.lang.String key,
-        java.lang.String defaultValue) {
-      if (key == null) { throw new java.lang.NullPointerException(); }
-      java.util.Map<java.lang.String, java.lang.String> map =
-          internalGetEnvironment().getMap();
-      return map.containsKey(key) ? map.get(key) : defaultValue;
-    }
-    /**
-     * <pre>
-     * Version of the tools and libraries used with this device (e.g. gcc 4.9,
-     * cudnn 5.1)
-     * </pre>
-     *
-     * <code>map&lt;string, string&gt; environment = 6;</code>
-     */
-
-    public java.lang.String getEnvironmentOrThrow(
-        java.lang.String key) {
-      if (key == null) { throw new java.lang.NullPointerException(); }
-      java.util.Map<java.lang.String, java.lang.String> map =
-          internalGetEnvironment().getMap();
-      if (!map.containsKey(key)) {
-        throw new java.lang.IllegalArgumentException();
-      }
-      return map.get(key);
-    }
-
-    public Builder clearEnvironment() {
-      internalGetMutableEnvironment().getMutableMap()
-          .clear();
-      return this;
-    }
-    /**
-     * <pre>
-     * Version of the tools and libraries used with this device (e.g. gcc 4.9,
-     * cudnn 5.1)
-     * </pre>
-     *
-     * <code>map&lt;string, string&gt; environment = 6;</code>
-     */
-
-    public Builder removeEnvironment(
-        java.lang.String key) {
-      if (key == null) { throw new java.lang.NullPointerException(); }
-      internalGetMutableEnvironment().getMutableMap()
-          .remove(key);
-      return this;
-    }
-    /**
-     * Use alternate mutation accessors instead.
-     */
-    @java.lang.Deprecated
-    public java.util.Map<java.lang.String, java.lang.String>
-    getMutableEnvironment() {
-      return internalGetMutableEnvironment().getMutableMap();
-    }
-    /**
-     * <pre>
-     * Version of the tools and libraries used with this device (e.g. gcc 4.9,
-     * cudnn 5.1)
-     * </pre>
-     *
-     * <code>map&lt;string, string&gt; environment = 6;</code>
-     */
-    public Builder putEnvironment(
-        java.lang.String key,
-        java.lang.String value) {
-      if (key == null) { throw new java.lang.NullPointerException(); }
-      if (value == null) { throw new java.lang.NullPointerException(); }
-      internalGetMutableEnvironment().getMutableMap()
-          .put(key, value);
-      return this;
-    }
-    /**
-     * <pre>
-     * Version of the tools and libraries used with this device (e.g. gcc 4.9,
-     * cudnn 5.1)
-     * </pre>
-     *
-     * <code>map&lt;string, string&gt; environment = 6;</code>
-     */
-
-    public Builder putAllEnvironment(
-        java.util.Map<java.lang.String, java.lang.String> values) {
-      internalGetMutableEnvironment().getMutableMap()
-          .putAll(values);
-      return this;
-    }
-
-    private long numRegisters_ ;
-    /**
-     * <pre>
-     * Number of registers per core.
-     * </pre>
-     *
-     * <code>int64 num_registers = 7;</code>
-     */
-    public long getNumRegisters() {
-      return numRegisters_;
-    }
-    /**
-     * <pre>
-     * Number of registers per core.
-     * </pre>
-     *
-     * <code>int64 num_registers = 7;</code>
-     */
-    public Builder setNumRegisters(long value) {
-      
-      numRegisters_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Number of registers per core.
-     * </pre>
-     *
-     * <code>int64 num_registers = 7;</code>
-     */
-    public Builder clearNumRegisters() {
-      
-      numRegisters_ = 0L;
-      onChanged();
-      return this;
-    }
-
-    private long l1CacheSize_ ;
-    /**
-     * <pre>
-     * L1 cache size in bytes
-     * </pre>
-     *
-     * <code>int64 l1_cache_size = 8;</code>
-     */
-    public long getL1CacheSize() {
-      return l1CacheSize_;
-    }
-    /**
-     * <pre>
-     * L1 cache size in bytes
-     * </pre>
-     *
-     * <code>int64 l1_cache_size = 8;</code>
-     */
-    public Builder setL1CacheSize(long value) {
-      
-      l1CacheSize_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * L1 cache size in bytes
-     * </pre>
-     *
-     * <code>int64 l1_cache_size = 8;</code>
-     */
-    public Builder clearL1CacheSize() {
-      
-      l1CacheSize_ = 0L;
-      onChanged();
-      return this;
-    }
-
-    private long l2CacheSize_ ;
-    /**
-     * <pre>
-     * L2 cache size in bytes
-     * </pre>
-     *
-     * <code>int64 l2_cache_size = 9;</code>
-     */
-    public long getL2CacheSize() {
-      return l2CacheSize_;
-    }
-    /**
-     * <pre>
-     * L2 cache size in bytes
-     * </pre>
-     *
-     * <code>int64 l2_cache_size = 9;</code>
-     */
-    public Builder setL2CacheSize(long value) {
-      
-      l2CacheSize_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * L2 cache size in bytes
-     * </pre>
-     *
-     * <code>int64 l2_cache_size = 9;</code>
-     */
-    public Builder clearL2CacheSize() {
-      
-      l2CacheSize_ = 0L;
-      onChanged();
-      return this;
-    }
-
-    private long l3CacheSize_ ;
-    /**
-     * <pre>
-     * L3 cache size in bytes
-     * </pre>
-     *
-     * <code>int64 l3_cache_size = 10;</code>
-     */
-    public long getL3CacheSize() {
-      return l3CacheSize_;
-    }
-    /**
-     * <pre>
-     * L3 cache size in bytes
-     * </pre>
-     *
-     * <code>int64 l3_cache_size = 10;</code>
-     */
-    public Builder setL3CacheSize(long value) {
-      
-      l3CacheSize_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * L3 cache size in bytes
-     * </pre>
-     *
-     * <code>int64 l3_cache_size = 10;</code>
-     */
-    public Builder clearL3CacheSize() {
-      
-      l3CacheSize_ = 0L;
-      onChanged();
-      return this;
-    }
-
-    private long sharedMemorySizePerMultiprocessor_ ;
-    /**
-     * <pre>
-     * Shared memory size per multiprocessor in bytes. This field is
-     * applicable to GPUs only.
-     * </pre>
-     *
-     * <code>int64 shared_memory_size_per_multiprocessor = 11;</code>
-     */
-    public long getSharedMemorySizePerMultiprocessor() {
-      return sharedMemorySizePerMultiprocessor_;
-    }
-    /**
-     * <pre>
-     * Shared memory size per multiprocessor in bytes. This field is
-     * applicable to GPUs only.
-     * </pre>
-     *
-     * <code>int64 shared_memory_size_per_multiprocessor = 11;</code>
-     */
-    public Builder setSharedMemorySizePerMultiprocessor(long value) {
-      
-      sharedMemorySizePerMultiprocessor_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Shared memory size per multiprocessor in bytes. This field is
-     * applicable to GPUs only.
-     * </pre>
-     *
-     * <code>int64 shared_memory_size_per_multiprocessor = 11;</code>
-     */
-    public Builder clearSharedMemorySizePerMultiprocessor() {
-      
-      sharedMemorySizePerMultiprocessor_ = 0L;
-      onChanged();
-      return this;
-    }
-
-    private long memorySize_ ;
-    /**
-     * <pre>
-     * Memory size in bytes
-     * </pre>
-     *
-     * <code>int64 memory_size = 12;</code>
-     */
-    public long getMemorySize() {
-      return memorySize_;
-    }
-    /**
-     * <pre>
-     * Memory size in bytes
-     * </pre>
-     *
-     * <code>int64 memory_size = 12;</code>
-     */
-    public Builder setMemorySize(long value) {
-      
-      memorySize_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Memory size in bytes
-     * </pre>
-     *
-     * <code>int64 memory_size = 12;</code>
-     */
-    public Builder clearMemorySize() {
-      
-      memorySize_ = 0L;
-      onChanged();
-      return this;
-    }
-
-    private long bandwidth_ ;
-    /**
-     * <pre>
-     * Memory bandwidth in KB/s
-     * </pre>
-     *
-     * <code>int64 bandwidth = 13;</code>
-     */
-    public long getBandwidth() {
-      return bandwidth_;
-    }
-    /**
-     * <pre>
-     * Memory bandwidth in KB/s
-     * </pre>
-     *
-     * <code>int64 bandwidth = 13;</code>
-     */
-    public Builder setBandwidth(long value) {
-      
-      bandwidth_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Memory bandwidth in KB/s
-     * </pre>
-     *
-     * <code>int64 bandwidth = 13;</code>
-     */
-    public Builder clearBandwidth() {
-      
-      bandwidth_ = 0L;
-      onChanged();
-      return this;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.DeviceProperties)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.DeviceProperties)
-  private static final org.tensorflow.proto.framework.DeviceProperties DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.DeviceProperties();
-  }
-
-  public static org.tensorflow.proto.framework.DeviceProperties getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<DeviceProperties>
-      PARSER = new com.google.protobuf.AbstractParser<DeviceProperties>() {
-    @java.lang.Override
-    public DeviceProperties parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new DeviceProperties(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<DeviceProperties> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<DeviceProperties> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.framework.DeviceProperties getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DevicePropertiesOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DevicePropertiesOrBuilder.java
deleted file mode 100644
index 675627a60fa..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DevicePropertiesOrBuilder.java
+++ /dev/null
@@ -1,204 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/device_properties.proto
-
-package org.tensorflow.proto.framework;
-
-public interface DevicePropertiesOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.DeviceProperties)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <pre>
-   * Device type (CPU, GPU, ...)
-   * </pre>
-   *
-   * <code>string type = 1;</code>
-   */
-  java.lang.String getType();
-  /**
-   * <pre>
-   * Device type (CPU, GPU, ...)
-   * </pre>
-   *
-   * <code>string type = 1;</code>
-   */
-  com.google.protobuf.ByteString
-      getTypeBytes();
-
-  /**
-   * <pre>
-   * Vendor (Intel, nvidia, ...)
-   * </pre>
-   *
-   * <code>string vendor = 2;</code>
-   */
-  java.lang.String getVendor();
-  /**
-   * <pre>
-   * Vendor (Intel, nvidia, ...)
-   * </pre>
-   *
-   * <code>string vendor = 2;</code>
-   */
-  com.google.protobuf.ByteString
-      getVendorBytes();
-
-  /**
-   * <pre>
-   * Model (Haswell, K40, ...)
-   * </pre>
-   *
-   * <code>string model = 3;</code>
-   */
-  java.lang.String getModel();
-  /**
-   * <pre>
-   * Model (Haswell, K40, ...)
-   * </pre>
-   *
-   * <code>string model = 3;</code>
-   */
-  com.google.protobuf.ByteString
-      getModelBytes();
-
-  /**
-   * <pre>
-   * Core Frequency in Mhz
-   * </pre>
-   *
-   * <code>int64 frequency = 4;</code>
-   */
-  long getFrequency();
-
-  /**
-   * <pre>
-   * Number of cores
-   * </pre>
-   *
-   * <code>int64 num_cores = 5;</code>
-   */
-  long getNumCores();
-
-  /**
-   * <pre>
-   * Version of the tools and libraries used with this device (e.g. gcc 4.9,
-   * cudnn 5.1)
-   * </pre>
-   *
-   * <code>map&lt;string, string&gt; environment = 6;</code>
-   */
-  int getEnvironmentCount();
-  /**
-   * <pre>
-   * Version of the tools and libraries used with this device (e.g. gcc 4.9,
-   * cudnn 5.1)
-   * </pre>
-   *
-   * <code>map&lt;string, string&gt; environment = 6;</code>
-   */
-  boolean containsEnvironment(
-      java.lang.String key);
-  /**
-   * Use {@link #getEnvironmentMap()} instead.
-   */
-  @java.lang.Deprecated
-  java.util.Map<java.lang.String, java.lang.String>
-  getEnvironment();
-  /**
-   * <pre>
-   * Version of the tools and libraries used with this device (e.g. gcc 4.9,
-   * cudnn 5.1)
-   * </pre>
-   *
-   * <code>map&lt;string, string&gt; environment = 6;</code>
-   */
-  java.util.Map<java.lang.String, java.lang.String>
-  getEnvironmentMap();
-  /**
-   * <pre>
-   * Version of the tools and libraries used with this device (e.g. gcc 4.9,
-   * cudnn 5.1)
-   * </pre>
-   *
-   * <code>map&lt;string, string&gt; environment = 6;</code>
-   */
-
-  java.lang.String getEnvironmentOrDefault(
-      java.lang.String key,
-      java.lang.String defaultValue);
-  /**
-   * <pre>
-   * Version of the tools and libraries used with this device (e.g. gcc 4.9,
-   * cudnn 5.1)
-   * </pre>
-   *
-   * <code>map&lt;string, string&gt; environment = 6;</code>
-   */
-
-  java.lang.String getEnvironmentOrThrow(
-      java.lang.String key);
-
-  /**
-   * <pre>
-   * Number of registers per core.
-   * </pre>
-   *
-   * <code>int64 num_registers = 7;</code>
-   */
-  long getNumRegisters();
-
-  /**
-   * <pre>
-   * L1 cache size in bytes
-   * </pre>
-   *
-   * <code>int64 l1_cache_size = 8;</code>
-   */
-  long getL1CacheSize();
-
-  /**
-   * <pre>
-   * L2 cache size in bytes
-   * </pre>
-   *
-   * <code>int64 l2_cache_size = 9;</code>
-   */
-  long getL2CacheSize();
-
-  /**
-   * <pre>
-   * L3 cache size in bytes
-   * </pre>
-   *
-   * <code>int64 l3_cache_size = 10;</code>
-   */
-  long getL3CacheSize();
-
-  /**
-   * <pre>
-   * Shared memory size per multiprocessor in bytes. This field is
-   * applicable to GPUs only.
-   * </pre>
-   *
-   * <code>int64 shared_memory_size_per_multiprocessor = 11;</code>
-   */
-  long getSharedMemorySizePerMultiprocessor();
-
-  /**
-   * <pre>
-   * Memory size in bytes
-   * </pre>
-   *
-   * <code>int64 memory_size = 12;</code>
-   */
-  long getMemorySize();
-
-  /**
-   * <pre>
-   * Memory bandwidth in KB/s
-   * </pre>
-   *
-   * <code>int64 bandwidth = 13;</code>
-   */
-  long getBandwidth();
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DevicePropertiesProtos.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DevicePropertiesProtos.java
deleted file mode 100644
index 7f6f979fca2..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DevicePropertiesProtos.java
+++ /dev/null
@@ -1,85 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/device_properties.proto
-
-package org.tensorflow.proto.framework;
-
-public final class DevicePropertiesProtos {
-  private DevicePropertiesProtos() {}
-  public static void registerAllExtensions(
-      com.google.protobuf.ExtensionRegistryLite registry) {
-  }
-
-  public static void registerAllExtensions(
-      com.google.protobuf.ExtensionRegistry registry) {
-    registerAllExtensions(
-        (com.google.protobuf.ExtensionRegistryLite) registry);
-  }
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_DeviceProperties_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_DeviceProperties_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_DeviceProperties_EnvironmentEntry_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_DeviceProperties_EnvironmentEntry_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_NamedDevice_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_NamedDevice_fieldAccessorTable;
-
-  public static com.google.protobuf.Descriptors.FileDescriptor
-      getDescriptor() {
-    return descriptor;
-  }
-  private static  com.google.protobuf.Descriptors.FileDescriptor
-      descriptor;
-  static {
-    java.lang.String[] descriptorData = {
-      "\n0tensorflow/core/protobuf/device_proper" +
-      "ties.proto\022\ntensorflow\"\220\003\n\020DevicePropert" +
-      "ies\022\014\n\004type\030\001 \001(\t\022\016\n\006vendor\030\002 \001(\t\022\r\n\005mod" +
-      "el\030\003 \001(\t\022\021\n\tfrequency\030\004 \001(\003\022\021\n\tnum_cores" +
-      "\030\005 \001(\003\022B\n\013environment\030\006 \003(\0132-.tensorflow" +
-      ".DeviceProperties.EnvironmentEntry\022\025\n\rnu" +
-      "m_registers\030\007 \001(\003\022\025\n\rl1_cache_size\030\010 \001(\003" +
-      "\022\025\n\rl2_cache_size\030\t \001(\003\022\025\n\rl3_cache_size" +
-      "\030\n \001(\003\022-\n%shared_memory_size_per_multipr" +
-      "ocessor\030\013 \001(\003\022\023\n\013memory_size\030\014 \001(\003\022\021\n\tba" +
-      "ndwidth\030\r \001(\003\0322\n\020EnvironmentEntry\022\013\n\003key" +
-      "\030\001 \001(\t\022\r\n\005value\030\002 \001(\t:\0028\001\"M\n\013NamedDevice" +
-      "\022\014\n\004name\030\001 \001(\t\0220\n\nproperties\030\002 \001(\0132\034.ten" +
-      "sorflow.DevicePropertiesB\207\001\n\036org.tensorf" +
-      "low.proto.frameworkB\026DevicePropertiesPro" +
-      "tosP\001ZHgithub.com/tensorflow/tensorflow/" +
-      "tensorflow/go/core/core_protos_go_proto\370" +
-      "\001\001b\006proto3"
-    };
-    descriptor = com.google.protobuf.Descriptors.FileDescriptor
-      .internalBuildGeneratedFileFrom(descriptorData,
-        new com.google.protobuf.Descriptors.FileDescriptor[] {
-        });
-    internal_static_tensorflow_DeviceProperties_descriptor =
-      getDescriptor().getMessageTypes().get(0);
-    internal_static_tensorflow_DeviceProperties_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_DeviceProperties_descriptor,
-        new java.lang.String[] { "Type", "Vendor", "Model", "Frequency", "NumCores", "Environment", "NumRegisters", "L1CacheSize", "L2CacheSize", "L3CacheSize", "SharedMemorySizePerMultiprocessor", "MemorySize", "Bandwidth", });
-    internal_static_tensorflow_DeviceProperties_EnvironmentEntry_descriptor =
-      internal_static_tensorflow_DeviceProperties_descriptor.getNestedTypes().get(0);
-    internal_static_tensorflow_DeviceProperties_EnvironmentEntry_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_DeviceProperties_EnvironmentEntry_descriptor,
-        new java.lang.String[] { "Key", "Value", });
-    internal_static_tensorflow_NamedDevice_descriptor =
-      getDescriptor().getMessageTypes().get(1);
-    internal_static_tensorflow_NamedDevice_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_NamedDevice_descriptor,
-        new java.lang.String[] { "Name", "Properties", });
-  }
-
-  // @@protoc_insertion_point(outer_class_scope)
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DeviceStepStats.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DeviceStepStats.java
deleted file mode 100644
index 8da65e2b3d5..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DeviceStepStats.java
+++ /dev/null
@@ -1,1209 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/framework/step_stats.proto
-
-package org.tensorflow.proto.framework;
-
-/**
- * Protobuf type {@code tensorflow.DeviceStepStats}
- */
-public  final class DeviceStepStats extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.DeviceStepStats)
-    DeviceStepStatsOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use DeviceStepStats.newBuilder() to construct.
-  private DeviceStepStats(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private DeviceStepStats() {
-    device_ = "";
-    nodeStats_ = java.util.Collections.emptyList();
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new DeviceStepStats();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private DeviceStepStats(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    int mutable_bitField0_ = 0;
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 10: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            device_ = s;
-            break;
-          }
-          case 18: {
-            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
-              nodeStats_ = new java.util.ArrayList<org.tensorflow.proto.framework.NodeExecStats>();
-              mutable_bitField0_ |= 0x00000001;
-            }
-            nodeStats_.add(
-                input.readMessage(org.tensorflow.proto.framework.NodeExecStats.parser(), extensionRegistry));
-            break;
-          }
-          case 26: {
-            if (!((mutable_bitField0_ & 0x00000002) != 0)) {
-              threadNames_ = com.google.protobuf.MapField.newMapField(
-                  ThreadNamesDefaultEntryHolder.defaultEntry);
-              mutable_bitField0_ |= 0x00000002;
-            }
-            com.google.protobuf.MapEntry<java.lang.Integer, java.lang.String>
-            threadNames__ = input.readMessage(
-                ThreadNamesDefaultEntryHolder.defaultEntry.getParserForType(), extensionRegistry);
-            threadNames_.getMutableMap().put(
-                threadNames__.getKey(), threadNames__.getValue());
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      if (((mutable_bitField0_ & 0x00000001) != 0)) {
-        nodeStats_ = java.util.Collections.unmodifiableList(nodeStats_);
-      }
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.framework.StepStatsProtos.internal_static_tensorflow_DeviceStepStats_descriptor;
-  }
-
-  @SuppressWarnings({"rawtypes"})
-  @java.lang.Override
-  protected com.google.protobuf.MapField internalGetMapField(
-      int number) {
-    switch (number) {
-      case 3:
-        return internalGetThreadNames();
-      default:
-        throw new RuntimeException(
-            "Invalid map field number: " + number);
-    }
-  }
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.framework.StepStatsProtos.internal_static_tensorflow_DeviceStepStats_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.framework.DeviceStepStats.class, org.tensorflow.proto.framework.DeviceStepStats.Builder.class);
-  }
-
-  public static final int DEVICE_FIELD_NUMBER = 1;
-  private volatile java.lang.Object device_;
-  /**
-   * <code>string device = 1;</code>
-   */
-  public java.lang.String getDevice() {
-    java.lang.Object ref = device_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      device_ = s;
-      return s;
-    }
-  }
-  /**
-   * <code>string device = 1;</code>
-   */
-  public com.google.protobuf.ByteString
-      getDeviceBytes() {
-    java.lang.Object ref = device_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      device_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int NODE_STATS_FIELD_NUMBER = 2;
-  private java.util.List<org.tensorflow.proto.framework.NodeExecStats> nodeStats_;
-  /**
-   * <code>repeated .tensorflow.NodeExecStats node_stats = 2;</code>
-   */
-  public java.util.List<org.tensorflow.proto.framework.NodeExecStats> getNodeStatsList() {
-    return nodeStats_;
-  }
-  /**
-   * <code>repeated .tensorflow.NodeExecStats node_stats = 2;</code>
-   */
-  public java.util.List<? extends org.tensorflow.proto.framework.NodeExecStatsOrBuilder> 
-      getNodeStatsOrBuilderList() {
-    return nodeStats_;
-  }
-  /**
-   * <code>repeated .tensorflow.NodeExecStats node_stats = 2;</code>
-   */
-  public int getNodeStatsCount() {
-    return nodeStats_.size();
-  }
-  /**
-   * <code>repeated .tensorflow.NodeExecStats node_stats = 2;</code>
-   */
-  public org.tensorflow.proto.framework.NodeExecStats getNodeStats(int index) {
-    return nodeStats_.get(index);
-  }
-  /**
-   * <code>repeated .tensorflow.NodeExecStats node_stats = 2;</code>
-   */
-  public org.tensorflow.proto.framework.NodeExecStatsOrBuilder getNodeStatsOrBuilder(
-      int index) {
-    return nodeStats_.get(index);
-  }
-
-  public static final int THREAD_NAMES_FIELD_NUMBER = 3;
-  private static final class ThreadNamesDefaultEntryHolder {
-    static final com.google.protobuf.MapEntry<
-        java.lang.Integer, java.lang.String> defaultEntry =
-            com.google.protobuf.MapEntry
-            .<java.lang.Integer, java.lang.String>newDefaultInstance(
-                org.tensorflow.proto.framework.StepStatsProtos.internal_static_tensorflow_DeviceStepStats_ThreadNamesEntry_descriptor, 
-                com.google.protobuf.WireFormat.FieldType.UINT32,
-                0,
-                com.google.protobuf.WireFormat.FieldType.STRING,
-                "");
-  }
-  private com.google.protobuf.MapField<
-      java.lang.Integer, java.lang.String> threadNames_;
-  private com.google.protobuf.MapField<java.lang.Integer, java.lang.String>
-  internalGetThreadNames() {
-    if (threadNames_ == null) {
-      return com.google.protobuf.MapField.emptyMapField(
-          ThreadNamesDefaultEntryHolder.defaultEntry);
-    }
-    return threadNames_;
-  }
-
-  public int getThreadNamesCount() {
-    return internalGetThreadNames().getMap().size();
-  }
-  /**
-   * <pre>
-   * Its key is thread id.
-   * </pre>
-   *
-   * <code>map&lt;uint32, string&gt; thread_names = 3;</code>
-   */
-
-  public boolean containsThreadNames(
-      int key) {
-    
-    return internalGetThreadNames().getMap().containsKey(key);
-  }
-  /**
-   * Use {@link #getThreadNamesMap()} instead.
-   */
-  @java.lang.Deprecated
-  public java.util.Map<java.lang.Integer, java.lang.String> getThreadNames() {
-    return getThreadNamesMap();
-  }
-  /**
-   * <pre>
-   * Its key is thread id.
-   * </pre>
-   *
-   * <code>map&lt;uint32, string&gt; thread_names = 3;</code>
-   */
-
-  public java.util.Map<java.lang.Integer, java.lang.String> getThreadNamesMap() {
-    return internalGetThreadNames().getMap();
-  }
-  /**
-   * <pre>
-   * Its key is thread id.
-   * </pre>
-   *
-   * <code>map&lt;uint32, string&gt; thread_names = 3;</code>
-   */
-
-  public java.lang.String getThreadNamesOrDefault(
-      int key,
-      java.lang.String defaultValue) {
-    
-    java.util.Map<java.lang.Integer, java.lang.String> map =
-        internalGetThreadNames().getMap();
-    return map.containsKey(key) ? map.get(key) : defaultValue;
-  }
-  /**
-   * <pre>
-   * Its key is thread id.
-   * </pre>
-   *
-   * <code>map&lt;uint32, string&gt; thread_names = 3;</code>
-   */
-
-  public java.lang.String getThreadNamesOrThrow(
-      int key) {
-    
-    java.util.Map<java.lang.Integer, java.lang.String> map =
-        internalGetThreadNames().getMap();
-    if (!map.containsKey(key)) {
-      throw new java.lang.IllegalArgumentException();
-    }
-    return map.get(key);
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    if (!getDeviceBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 1, device_);
-    }
-    for (int i = 0; i < nodeStats_.size(); i++) {
-      output.writeMessage(2, nodeStats_.get(i));
-    }
-    com.google.protobuf.GeneratedMessageV3
-      .serializeIntegerMapTo(
-        output,
-        internalGetThreadNames(),
-        ThreadNamesDefaultEntryHolder.defaultEntry,
-        3);
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (!getDeviceBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, device_);
-    }
-    for (int i = 0; i < nodeStats_.size(); i++) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(2, nodeStats_.get(i));
-    }
-    for (java.util.Map.Entry<java.lang.Integer, java.lang.String> entry
-         : internalGetThreadNames().getMap().entrySet()) {
-      com.google.protobuf.MapEntry<java.lang.Integer, java.lang.String>
-      threadNames__ = ThreadNamesDefaultEntryHolder.defaultEntry.newBuilderForType()
-          .setKey(entry.getKey())
-          .setValue(entry.getValue())
-          .build();
-      size += com.google.protobuf.CodedOutputStream
-          .computeMessageSize(3, threadNames__);
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.framework.DeviceStepStats)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.framework.DeviceStepStats other = (org.tensorflow.proto.framework.DeviceStepStats) obj;
-
-    if (!getDevice()
-        .equals(other.getDevice())) return false;
-    if (!getNodeStatsList()
-        .equals(other.getNodeStatsList())) return false;
-    if (!internalGetThreadNames().equals(
-        other.internalGetThreadNames())) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    hash = (37 * hash) + DEVICE_FIELD_NUMBER;
-    hash = (53 * hash) + getDevice().hashCode();
-    if (getNodeStatsCount() > 0) {
-      hash = (37 * hash) + NODE_STATS_FIELD_NUMBER;
-      hash = (53 * hash) + getNodeStatsList().hashCode();
-    }
-    if (!internalGetThreadNames().getMap().isEmpty()) {
-      hash = (37 * hash) + THREAD_NAMES_FIELD_NUMBER;
-      hash = (53 * hash) + internalGetThreadNames().hashCode();
-    }
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.framework.DeviceStepStats parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.DeviceStepStats parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.DeviceStepStats parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.DeviceStepStats parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.DeviceStepStats parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.DeviceStepStats parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.DeviceStepStats parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.DeviceStepStats parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.DeviceStepStats parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.DeviceStepStats parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.DeviceStepStats parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.DeviceStepStats parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.framework.DeviceStepStats prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * Protobuf type {@code tensorflow.DeviceStepStats}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.DeviceStepStats)
-      org.tensorflow.proto.framework.DeviceStepStatsOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.StepStatsProtos.internal_static_tensorflow_DeviceStepStats_descriptor;
-    }
-
-    @SuppressWarnings({"rawtypes"})
-    protected com.google.protobuf.MapField internalGetMapField(
-        int number) {
-      switch (number) {
-        case 3:
-          return internalGetThreadNames();
-        default:
-          throw new RuntimeException(
-              "Invalid map field number: " + number);
-      }
-    }
-    @SuppressWarnings({"rawtypes"})
-    protected com.google.protobuf.MapField internalGetMutableMapField(
-        int number) {
-      switch (number) {
-        case 3:
-          return internalGetMutableThreadNames();
-        default:
-          throw new RuntimeException(
-              "Invalid map field number: " + number);
-      }
-    }
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.StepStatsProtos.internal_static_tensorflow_DeviceStepStats_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.DeviceStepStats.class, org.tensorflow.proto.framework.DeviceStepStats.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.framework.DeviceStepStats.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-        getNodeStatsFieldBuilder();
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      device_ = "";
-
-      if (nodeStatsBuilder_ == null) {
-        nodeStats_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000001);
-      } else {
-        nodeStatsBuilder_.clear();
-      }
-      internalGetMutableThreadNames().clear();
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.framework.StepStatsProtos.internal_static_tensorflow_DeviceStepStats_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.DeviceStepStats getDefaultInstanceForType() {
-      return org.tensorflow.proto.framework.DeviceStepStats.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.DeviceStepStats build() {
-      org.tensorflow.proto.framework.DeviceStepStats result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.DeviceStepStats buildPartial() {
-      org.tensorflow.proto.framework.DeviceStepStats result = new org.tensorflow.proto.framework.DeviceStepStats(this);
-      int from_bitField0_ = bitField0_;
-      result.device_ = device_;
-      if (nodeStatsBuilder_ == null) {
-        if (((bitField0_ & 0x00000001) != 0)) {
-          nodeStats_ = java.util.Collections.unmodifiableList(nodeStats_);
-          bitField0_ = (bitField0_ & ~0x00000001);
-        }
-        result.nodeStats_ = nodeStats_;
-      } else {
-        result.nodeStats_ = nodeStatsBuilder_.build();
-      }
-      result.threadNames_ = internalGetThreadNames();
-      result.threadNames_.makeImmutable();
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.framework.DeviceStepStats) {
-        return mergeFrom((org.tensorflow.proto.framework.DeviceStepStats)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.framework.DeviceStepStats other) {
-      if (other == org.tensorflow.proto.framework.DeviceStepStats.getDefaultInstance()) return this;
-      if (!other.getDevice().isEmpty()) {
-        device_ = other.device_;
-        onChanged();
-      }
-      if (nodeStatsBuilder_ == null) {
-        if (!other.nodeStats_.isEmpty()) {
-          if (nodeStats_.isEmpty()) {
-            nodeStats_ = other.nodeStats_;
-            bitField0_ = (bitField0_ & ~0x00000001);
-          } else {
-            ensureNodeStatsIsMutable();
-            nodeStats_.addAll(other.nodeStats_);
-          }
-          onChanged();
-        }
-      } else {
-        if (!other.nodeStats_.isEmpty()) {
-          if (nodeStatsBuilder_.isEmpty()) {
-            nodeStatsBuilder_.dispose();
-            nodeStatsBuilder_ = null;
-            nodeStats_ = other.nodeStats_;
-            bitField0_ = (bitField0_ & ~0x00000001);
-            nodeStatsBuilder_ = 
-              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
-                 getNodeStatsFieldBuilder() : null;
-          } else {
-            nodeStatsBuilder_.addAllMessages(other.nodeStats_);
-          }
-        }
-      }
-      internalGetMutableThreadNames().mergeFrom(
-          other.internalGetThreadNames());
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.framework.DeviceStepStats parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.framework.DeviceStepStats) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-    private int bitField0_;
-
-    private java.lang.Object device_ = "";
-    /**
-     * <code>string device = 1;</code>
-     */
-    public java.lang.String getDevice() {
-      java.lang.Object ref = device_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        device_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <code>string device = 1;</code>
-     */
-    public com.google.protobuf.ByteString
-        getDeviceBytes() {
-      java.lang.Object ref = device_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        device_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <code>string device = 1;</code>
-     */
-    public Builder setDevice(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      device_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>string device = 1;</code>
-     */
-    public Builder clearDevice() {
-      
-      device_ = getDefaultInstance().getDevice();
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>string device = 1;</code>
-     */
-    public Builder setDeviceBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      device_ = value;
-      onChanged();
-      return this;
-    }
-
-    private java.util.List<org.tensorflow.proto.framework.NodeExecStats> nodeStats_ =
-      java.util.Collections.emptyList();
-    private void ensureNodeStatsIsMutable() {
-      if (!((bitField0_ & 0x00000001) != 0)) {
-        nodeStats_ = new java.util.ArrayList<org.tensorflow.proto.framework.NodeExecStats>(nodeStats_);
-        bitField0_ |= 0x00000001;
-       }
-    }
-
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.framework.NodeExecStats, org.tensorflow.proto.framework.NodeExecStats.Builder, org.tensorflow.proto.framework.NodeExecStatsOrBuilder> nodeStatsBuilder_;
-
-    /**
-     * <code>repeated .tensorflow.NodeExecStats node_stats = 2;</code>
-     */
-    public java.util.List<org.tensorflow.proto.framework.NodeExecStats> getNodeStatsList() {
-      if (nodeStatsBuilder_ == null) {
-        return java.util.Collections.unmodifiableList(nodeStats_);
-      } else {
-        return nodeStatsBuilder_.getMessageList();
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.NodeExecStats node_stats = 2;</code>
-     */
-    public int getNodeStatsCount() {
-      if (nodeStatsBuilder_ == null) {
-        return nodeStats_.size();
-      } else {
-        return nodeStatsBuilder_.getCount();
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.NodeExecStats node_stats = 2;</code>
-     */
-    public org.tensorflow.proto.framework.NodeExecStats getNodeStats(int index) {
-      if (nodeStatsBuilder_ == null) {
-        return nodeStats_.get(index);
-      } else {
-        return nodeStatsBuilder_.getMessage(index);
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.NodeExecStats node_stats = 2;</code>
-     */
-    public Builder setNodeStats(
-        int index, org.tensorflow.proto.framework.NodeExecStats value) {
-      if (nodeStatsBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureNodeStatsIsMutable();
-        nodeStats_.set(index, value);
-        onChanged();
-      } else {
-        nodeStatsBuilder_.setMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.NodeExecStats node_stats = 2;</code>
-     */
-    public Builder setNodeStats(
-        int index, org.tensorflow.proto.framework.NodeExecStats.Builder builderForValue) {
-      if (nodeStatsBuilder_ == null) {
-        ensureNodeStatsIsMutable();
-        nodeStats_.set(index, builderForValue.build());
-        onChanged();
-      } else {
-        nodeStatsBuilder_.setMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.NodeExecStats node_stats = 2;</code>
-     */
-    public Builder addNodeStats(org.tensorflow.proto.framework.NodeExecStats value) {
-      if (nodeStatsBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureNodeStatsIsMutable();
-        nodeStats_.add(value);
-        onChanged();
-      } else {
-        nodeStatsBuilder_.addMessage(value);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.NodeExecStats node_stats = 2;</code>
-     */
-    public Builder addNodeStats(
-        int index, org.tensorflow.proto.framework.NodeExecStats value) {
-      if (nodeStatsBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureNodeStatsIsMutable();
-        nodeStats_.add(index, value);
-        onChanged();
-      } else {
-        nodeStatsBuilder_.addMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.NodeExecStats node_stats = 2;</code>
-     */
-    public Builder addNodeStats(
-        org.tensorflow.proto.framework.NodeExecStats.Builder builderForValue) {
-      if (nodeStatsBuilder_ == null) {
-        ensureNodeStatsIsMutable();
-        nodeStats_.add(builderForValue.build());
-        onChanged();
-      } else {
-        nodeStatsBuilder_.addMessage(builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.NodeExecStats node_stats = 2;</code>
-     */
-    public Builder addNodeStats(
-        int index, org.tensorflow.proto.framework.NodeExecStats.Builder builderForValue) {
-      if (nodeStatsBuilder_ == null) {
-        ensureNodeStatsIsMutable();
-        nodeStats_.add(index, builderForValue.build());
-        onChanged();
-      } else {
-        nodeStatsBuilder_.addMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.NodeExecStats node_stats = 2;</code>
-     */
-    public Builder addAllNodeStats(
-        java.lang.Iterable<? extends org.tensorflow.proto.framework.NodeExecStats> values) {
-      if (nodeStatsBuilder_ == null) {
-        ensureNodeStatsIsMutable();
-        com.google.protobuf.AbstractMessageLite.Builder.addAll(
-            values, nodeStats_);
-        onChanged();
-      } else {
-        nodeStatsBuilder_.addAllMessages(values);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.NodeExecStats node_stats = 2;</code>
-     */
-    public Builder clearNodeStats() {
-      if (nodeStatsBuilder_ == null) {
-        nodeStats_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000001);
-        onChanged();
-      } else {
-        nodeStatsBuilder_.clear();
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.NodeExecStats node_stats = 2;</code>
-     */
-    public Builder removeNodeStats(int index) {
-      if (nodeStatsBuilder_ == null) {
-        ensureNodeStatsIsMutable();
-        nodeStats_.remove(index);
-        onChanged();
-      } else {
-        nodeStatsBuilder_.remove(index);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.NodeExecStats node_stats = 2;</code>
-     */
-    public org.tensorflow.proto.framework.NodeExecStats.Builder getNodeStatsBuilder(
-        int index) {
-      return getNodeStatsFieldBuilder().getBuilder(index);
-    }
-    /**
-     * <code>repeated .tensorflow.NodeExecStats node_stats = 2;</code>
-     */
-    public org.tensorflow.proto.framework.NodeExecStatsOrBuilder getNodeStatsOrBuilder(
-        int index) {
-      if (nodeStatsBuilder_ == null) {
-        return nodeStats_.get(index);  } else {
-        return nodeStatsBuilder_.getMessageOrBuilder(index);
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.NodeExecStats node_stats = 2;</code>
-     */
-    public java.util.List<? extends org.tensorflow.proto.framework.NodeExecStatsOrBuilder> 
-         getNodeStatsOrBuilderList() {
-      if (nodeStatsBuilder_ != null) {
-        return nodeStatsBuilder_.getMessageOrBuilderList();
-      } else {
-        return java.util.Collections.unmodifiableList(nodeStats_);
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.NodeExecStats node_stats = 2;</code>
-     */
-    public org.tensorflow.proto.framework.NodeExecStats.Builder addNodeStatsBuilder() {
-      return getNodeStatsFieldBuilder().addBuilder(
-          org.tensorflow.proto.framework.NodeExecStats.getDefaultInstance());
-    }
-    /**
-     * <code>repeated .tensorflow.NodeExecStats node_stats = 2;</code>
-     */
-    public org.tensorflow.proto.framework.NodeExecStats.Builder addNodeStatsBuilder(
-        int index) {
-      return getNodeStatsFieldBuilder().addBuilder(
-          index, org.tensorflow.proto.framework.NodeExecStats.getDefaultInstance());
-    }
-    /**
-     * <code>repeated .tensorflow.NodeExecStats node_stats = 2;</code>
-     */
-    public java.util.List<org.tensorflow.proto.framework.NodeExecStats.Builder> 
-         getNodeStatsBuilderList() {
-      return getNodeStatsFieldBuilder().getBuilderList();
-    }
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.framework.NodeExecStats, org.tensorflow.proto.framework.NodeExecStats.Builder, org.tensorflow.proto.framework.NodeExecStatsOrBuilder> 
-        getNodeStatsFieldBuilder() {
-      if (nodeStatsBuilder_ == null) {
-        nodeStatsBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
-            org.tensorflow.proto.framework.NodeExecStats, org.tensorflow.proto.framework.NodeExecStats.Builder, org.tensorflow.proto.framework.NodeExecStatsOrBuilder>(
-                nodeStats_,
-                ((bitField0_ & 0x00000001) != 0),
-                getParentForChildren(),
-                isClean());
-        nodeStats_ = null;
-      }
-      return nodeStatsBuilder_;
-    }
-
-    private com.google.protobuf.MapField<
-        java.lang.Integer, java.lang.String> threadNames_;
-    private com.google.protobuf.MapField<java.lang.Integer, java.lang.String>
-    internalGetThreadNames() {
-      if (threadNames_ == null) {
-        return com.google.protobuf.MapField.emptyMapField(
-            ThreadNamesDefaultEntryHolder.defaultEntry);
-      }
-      return threadNames_;
-    }
-    private com.google.protobuf.MapField<java.lang.Integer, java.lang.String>
-    internalGetMutableThreadNames() {
-      onChanged();;
-      if (threadNames_ == null) {
-        threadNames_ = com.google.protobuf.MapField.newMapField(
-            ThreadNamesDefaultEntryHolder.defaultEntry);
-      }
-      if (!threadNames_.isMutable()) {
-        threadNames_ = threadNames_.copy();
-      }
-      return threadNames_;
-    }
-
-    public int getThreadNamesCount() {
-      return internalGetThreadNames().getMap().size();
-    }
-    /**
-     * <pre>
-     * Its key is thread id.
-     * </pre>
-     *
-     * <code>map&lt;uint32, string&gt; thread_names = 3;</code>
-     */
-
-    public boolean containsThreadNames(
-        int key) {
-      
-      return internalGetThreadNames().getMap().containsKey(key);
-    }
-    /**
-     * Use {@link #getThreadNamesMap()} instead.
-     */
-    @java.lang.Deprecated
-    public java.util.Map<java.lang.Integer, java.lang.String> getThreadNames() {
-      return getThreadNamesMap();
-    }
-    /**
-     * <pre>
-     * Its key is thread id.
-     * </pre>
-     *
-     * <code>map&lt;uint32, string&gt; thread_names = 3;</code>
-     */
-
-    public java.util.Map<java.lang.Integer, java.lang.String> getThreadNamesMap() {
-      return internalGetThreadNames().getMap();
-    }
-    /**
-     * <pre>
-     * Its key is thread id.
-     * </pre>
-     *
-     * <code>map&lt;uint32, string&gt; thread_names = 3;</code>
-     */
-
-    public java.lang.String getThreadNamesOrDefault(
-        int key,
-        java.lang.String defaultValue) {
-      
-      java.util.Map<java.lang.Integer, java.lang.String> map =
-          internalGetThreadNames().getMap();
-      return map.containsKey(key) ? map.get(key) : defaultValue;
-    }
-    /**
-     * <pre>
-     * Its key is thread id.
-     * </pre>
-     *
-     * <code>map&lt;uint32, string&gt; thread_names = 3;</code>
-     */
-
-    public java.lang.String getThreadNamesOrThrow(
-        int key) {
-      
-      java.util.Map<java.lang.Integer, java.lang.String> map =
-          internalGetThreadNames().getMap();
-      if (!map.containsKey(key)) {
-        throw new java.lang.IllegalArgumentException();
-      }
-      return map.get(key);
-    }
-
-    public Builder clearThreadNames() {
-      internalGetMutableThreadNames().getMutableMap()
-          .clear();
-      return this;
-    }
-    /**
-     * <pre>
-     * Its key is thread id.
-     * </pre>
-     *
-     * <code>map&lt;uint32, string&gt; thread_names = 3;</code>
-     */
-
-    public Builder removeThreadNames(
-        int key) {
-      
-      internalGetMutableThreadNames().getMutableMap()
-          .remove(key);
-      return this;
-    }
-    /**
-     * Use alternate mutation accessors instead.
-     */
-    @java.lang.Deprecated
-    public java.util.Map<java.lang.Integer, java.lang.String>
-    getMutableThreadNames() {
-      return internalGetMutableThreadNames().getMutableMap();
-    }
-    /**
-     * <pre>
-     * Its key is thread id.
-     * </pre>
-     *
-     * <code>map&lt;uint32, string&gt; thread_names = 3;</code>
-     */
-    public Builder putThreadNames(
-        int key,
-        java.lang.String value) {
-      
-      if (value == null) { throw new java.lang.NullPointerException(); }
-      internalGetMutableThreadNames().getMutableMap()
-          .put(key, value);
-      return this;
-    }
-    /**
-     * <pre>
-     * Its key is thread id.
-     * </pre>
-     *
-     * <code>map&lt;uint32, string&gt; thread_names = 3;</code>
-     */
-
-    public Builder putAllThreadNames(
-        java.util.Map<java.lang.Integer, java.lang.String> values) {
-      internalGetMutableThreadNames().getMutableMap()
-          .putAll(values);
-      return this;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.DeviceStepStats)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.DeviceStepStats)
-  private static final org.tensorflow.proto.framework.DeviceStepStats DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.DeviceStepStats();
-  }
-
-  public static org.tensorflow.proto.framework.DeviceStepStats getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<DeviceStepStats>
-      PARSER = new com.google.protobuf.AbstractParser<DeviceStepStats>() {
-    @java.lang.Override
-    public DeviceStepStats parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new DeviceStepStats(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<DeviceStepStats> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<DeviceStepStats> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.framework.DeviceStepStats getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DeviceStepStatsOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DeviceStepStatsOrBuilder.java
deleted file mode 100644
index d53ab25f2e0..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DeviceStepStatsOrBuilder.java
+++ /dev/null
@@ -1,97 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/framework/step_stats.proto
-
-package org.tensorflow.proto.framework;
-
-public interface DeviceStepStatsOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.DeviceStepStats)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <code>string device = 1;</code>
-   */
-  java.lang.String getDevice();
-  /**
-   * <code>string device = 1;</code>
-   */
-  com.google.protobuf.ByteString
-      getDeviceBytes();
-
-  /**
-   * <code>repeated .tensorflow.NodeExecStats node_stats = 2;</code>
-   */
-  java.util.List<org.tensorflow.proto.framework.NodeExecStats> 
-      getNodeStatsList();
-  /**
-   * <code>repeated .tensorflow.NodeExecStats node_stats = 2;</code>
-   */
-  org.tensorflow.proto.framework.NodeExecStats getNodeStats(int index);
-  /**
-   * <code>repeated .tensorflow.NodeExecStats node_stats = 2;</code>
-   */
-  int getNodeStatsCount();
-  /**
-   * <code>repeated .tensorflow.NodeExecStats node_stats = 2;</code>
-   */
-  java.util.List<? extends org.tensorflow.proto.framework.NodeExecStatsOrBuilder> 
-      getNodeStatsOrBuilderList();
-  /**
-   * <code>repeated .tensorflow.NodeExecStats node_stats = 2;</code>
-   */
-  org.tensorflow.proto.framework.NodeExecStatsOrBuilder getNodeStatsOrBuilder(
-      int index);
-
-  /**
-   * <pre>
-   * Its key is thread id.
-   * </pre>
-   *
-   * <code>map&lt;uint32, string&gt; thread_names = 3;</code>
-   */
-  int getThreadNamesCount();
-  /**
-   * <pre>
-   * Its key is thread id.
-   * </pre>
-   *
-   * <code>map&lt;uint32, string&gt; thread_names = 3;</code>
-   */
-  boolean containsThreadNames(
-      int key);
-  /**
-   * Use {@link #getThreadNamesMap()} instead.
-   */
-  @java.lang.Deprecated
-  java.util.Map<java.lang.Integer, java.lang.String>
-  getThreadNames();
-  /**
-   * <pre>
-   * Its key is thread id.
-   * </pre>
-   *
-   * <code>map&lt;uint32, string&gt; thread_names = 3;</code>
-   */
-  java.util.Map<java.lang.Integer, java.lang.String>
-  getThreadNamesMap();
-  /**
-   * <pre>
-   * Its key is thread id.
-   * </pre>
-   *
-   * <code>map&lt;uint32, string&gt; thread_names = 3;</code>
-   */
-
-  java.lang.String getThreadNamesOrDefault(
-      int key,
-      java.lang.String defaultValue);
-  /**
-   * <pre>
-   * Its key is thread id.
-   * </pre>
-   *
-   * <code>map&lt;uint32, string&gt; thread_names = 3;</code>
-   */
-
-  java.lang.String getThreadNamesOrThrow(
-      int key);
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DictValue.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DictValue.java
deleted file mode 100644
index e8e6321372f..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DictValue.java
+++ /dev/null
@@ -1,705 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/struct.proto
-
-package org.tensorflow.proto.framework;
-
-/**
- * <pre>
- * Represents a Python dict keyed by `str`.
- * The comment on Unicode from Value.string_value applies analogously.
- * </pre>
- *
- * Protobuf type {@code tensorflow.DictValue}
- */
-public  final class DictValue extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.DictValue)
-    DictValueOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use DictValue.newBuilder() to construct.
-  private DictValue(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private DictValue() {
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new DictValue();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private DictValue(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    int mutable_bitField0_ = 0;
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 10: {
-            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
-              fields_ = com.google.protobuf.MapField.newMapField(
-                  FieldsDefaultEntryHolder.defaultEntry);
-              mutable_bitField0_ |= 0x00000001;
-            }
-            com.google.protobuf.MapEntry<java.lang.String, org.tensorflow.proto.framework.StructuredValue>
-            fields__ = input.readMessage(
-                FieldsDefaultEntryHolder.defaultEntry.getParserForType(), extensionRegistry);
-            fields_.getMutableMap().put(
-                fields__.getKey(), fields__.getValue());
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.framework.StructProtos.internal_static_tensorflow_DictValue_descriptor;
-  }
-
-  @SuppressWarnings({"rawtypes"})
-  @java.lang.Override
-  protected com.google.protobuf.MapField internalGetMapField(
-      int number) {
-    switch (number) {
-      case 1:
-        return internalGetFields();
-      default:
-        throw new RuntimeException(
-            "Invalid map field number: " + number);
-    }
-  }
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.framework.StructProtos.internal_static_tensorflow_DictValue_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.framework.DictValue.class, org.tensorflow.proto.framework.DictValue.Builder.class);
-  }
-
-  public static final int FIELDS_FIELD_NUMBER = 1;
-  private static final class FieldsDefaultEntryHolder {
-    static final com.google.protobuf.MapEntry<
-        java.lang.String, org.tensorflow.proto.framework.StructuredValue> defaultEntry =
-            com.google.protobuf.MapEntry
-            .<java.lang.String, org.tensorflow.proto.framework.StructuredValue>newDefaultInstance(
-                org.tensorflow.proto.framework.StructProtos.internal_static_tensorflow_DictValue_FieldsEntry_descriptor, 
-                com.google.protobuf.WireFormat.FieldType.STRING,
-                "",
-                com.google.protobuf.WireFormat.FieldType.MESSAGE,
-                org.tensorflow.proto.framework.StructuredValue.getDefaultInstance());
-  }
-  private com.google.protobuf.MapField<
-      java.lang.String, org.tensorflow.proto.framework.StructuredValue> fields_;
-  private com.google.protobuf.MapField<java.lang.String, org.tensorflow.proto.framework.StructuredValue>
-  internalGetFields() {
-    if (fields_ == null) {
-      return com.google.protobuf.MapField.emptyMapField(
-          FieldsDefaultEntryHolder.defaultEntry);
-    }
-    return fields_;
-  }
-
-  public int getFieldsCount() {
-    return internalGetFields().getMap().size();
-  }
-  /**
-   * <code>map&lt;string, .tensorflow.StructuredValue&gt; fields = 1;</code>
-   */
-
-  public boolean containsFields(
-      java.lang.String key) {
-    if (key == null) { throw new java.lang.NullPointerException(); }
-    return internalGetFields().getMap().containsKey(key);
-  }
-  /**
-   * Use {@link #getFieldsMap()} instead.
-   */
-  @java.lang.Deprecated
-  public java.util.Map<java.lang.String, org.tensorflow.proto.framework.StructuredValue> getFields() {
-    return getFieldsMap();
-  }
-  /**
-   * <code>map&lt;string, .tensorflow.StructuredValue&gt; fields = 1;</code>
-   */
-
-  public java.util.Map<java.lang.String, org.tensorflow.proto.framework.StructuredValue> getFieldsMap() {
-    return internalGetFields().getMap();
-  }
-  /**
-   * <code>map&lt;string, .tensorflow.StructuredValue&gt; fields = 1;</code>
-   */
-
-  public org.tensorflow.proto.framework.StructuredValue getFieldsOrDefault(
-      java.lang.String key,
-      org.tensorflow.proto.framework.StructuredValue defaultValue) {
-    if (key == null) { throw new java.lang.NullPointerException(); }
-    java.util.Map<java.lang.String, org.tensorflow.proto.framework.StructuredValue> map =
-        internalGetFields().getMap();
-    return map.containsKey(key) ? map.get(key) : defaultValue;
-  }
-  /**
-   * <code>map&lt;string, .tensorflow.StructuredValue&gt; fields = 1;</code>
-   */
-
-  public org.tensorflow.proto.framework.StructuredValue getFieldsOrThrow(
-      java.lang.String key) {
-    if (key == null) { throw new java.lang.NullPointerException(); }
-    java.util.Map<java.lang.String, org.tensorflow.proto.framework.StructuredValue> map =
-        internalGetFields().getMap();
-    if (!map.containsKey(key)) {
-      throw new java.lang.IllegalArgumentException();
-    }
-    return map.get(key);
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    com.google.protobuf.GeneratedMessageV3
-      .serializeStringMapTo(
-        output,
-        internalGetFields(),
-        FieldsDefaultEntryHolder.defaultEntry,
-        1);
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    for (java.util.Map.Entry<java.lang.String, org.tensorflow.proto.framework.StructuredValue> entry
-         : internalGetFields().getMap().entrySet()) {
-      com.google.protobuf.MapEntry<java.lang.String, org.tensorflow.proto.framework.StructuredValue>
-      fields__ = FieldsDefaultEntryHolder.defaultEntry.newBuilderForType()
-          .setKey(entry.getKey())
-          .setValue(entry.getValue())
-          .build();
-      size += com.google.protobuf.CodedOutputStream
-          .computeMessageSize(1, fields__);
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.framework.DictValue)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.framework.DictValue other = (org.tensorflow.proto.framework.DictValue) obj;
-
-    if (!internalGetFields().equals(
-        other.internalGetFields())) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    if (!internalGetFields().getMap().isEmpty()) {
-      hash = (37 * hash) + FIELDS_FIELD_NUMBER;
-      hash = (53 * hash) + internalGetFields().hashCode();
-    }
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.framework.DictValue parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.DictValue parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.DictValue parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.DictValue parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.DictValue parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.DictValue parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.DictValue parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.DictValue parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.DictValue parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.DictValue parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.DictValue parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.DictValue parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.framework.DictValue prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * <pre>
-   * Represents a Python dict keyed by `str`.
-   * The comment on Unicode from Value.string_value applies analogously.
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.DictValue}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.DictValue)
-      org.tensorflow.proto.framework.DictValueOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.StructProtos.internal_static_tensorflow_DictValue_descriptor;
-    }
-
-    @SuppressWarnings({"rawtypes"})
-    protected com.google.protobuf.MapField internalGetMapField(
-        int number) {
-      switch (number) {
-        case 1:
-          return internalGetFields();
-        default:
-          throw new RuntimeException(
-              "Invalid map field number: " + number);
-      }
-    }
-    @SuppressWarnings({"rawtypes"})
-    protected com.google.protobuf.MapField internalGetMutableMapField(
-        int number) {
-      switch (number) {
-        case 1:
-          return internalGetMutableFields();
-        default:
-          throw new RuntimeException(
-              "Invalid map field number: " + number);
-      }
-    }
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.StructProtos.internal_static_tensorflow_DictValue_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.DictValue.class, org.tensorflow.proto.framework.DictValue.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.framework.DictValue.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      internalGetMutableFields().clear();
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.framework.StructProtos.internal_static_tensorflow_DictValue_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.DictValue getDefaultInstanceForType() {
-      return org.tensorflow.proto.framework.DictValue.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.DictValue build() {
-      org.tensorflow.proto.framework.DictValue result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.DictValue buildPartial() {
-      org.tensorflow.proto.framework.DictValue result = new org.tensorflow.proto.framework.DictValue(this);
-      int from_bitField0_ = bitField0_;
-      result.fields_ = internalGetFields();
-      result.fields_.makeImmutable();
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.framework.DictValue) {
-        return mergeFrom((org.tensorflow.proto.framework.DictValue)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.framework.DictValue other) {
-      if (other == org.tensorflow.proto.framework.DictValue.getDefaultInstance()) return this;
-      internalGetMutableFields().mergeFrom(
-          other.internalGetFields());
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.framework.DictValue parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.framework.DictValue) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-    private int bitField0_;
-
-    private com.google.protobuf.MapField<
-        java.lang.String, org.tensorflow.proto.framework.StructuredValue> fields_;
-    private com.google.protobuf.MapField<java.lang.String, org.tensorflow.proto.framework.StructuredValue>
-    internalGetFields() {
-      if (fields_ == null) {
-        return com.google.protobuf.MapField.emptyMapField(
-            FieldsDefaultEntryHolder.defaultEntry);
-      }
-      return fields_;
-    }
-    private com.google.protobuf.MapField<java.lang.String, org.tensorflow.proto.framework.StructuredValue>
-    internalGetMutableFields() {
-      onChanged();;
-      if (fields_ == null) {
-        fields_ = com.google.protobuf.MapField.newMapField(
-            FieldsDefaultEntryHolder.defaultEntry);
-      }
-      if (!fields_.isMutable()) {
-        fields_ = fields_.copy();
-      }
-      return fields_;
-    }
-
-    public int getFieldsCount() {
-      return internalGetFields().getMap().size();
-    }
-    /**
-     * <code>map&lt;string, .tensorflow.StructuredValue&gt; fields = 1;</code>
-     */
-
-    public boolean containsFields(
-        java.lang.String key) {
-      if (key == null) { throw new java.lang.NullPointerException(); }
-      return internalGetFields().getMap().containsKey(key);
-    }
-    /**
-     * Use {@link #getFieldsMap()} instead.
-     */
-    @java.lang.Deprecated
-    public java.util.Map<java.lang.String, org.tensorflow.proto.framework.StructuredValue> getFields() {
-      return getFieldsMap();
-    }
-    /**
-     * <code>map&lt;string, .tensorflow.StructuredValue&gt; fields = 1;</code>
-     */
-
-    public java.util.Map<java.lang.String, org.tensorflow.proto.framework.StructuredValue> getFieldsMap() {
-      return internalGetFields().getMap();
-    }
-    /**
-     * <code>map&lt;string, .tensorflow.StructuredValue&gt; fields = 1;</code>
-     */
-
-    public org.tensorflow.proto.framework.StructuredValue getFieldsOrDefault(
-        java.lang.String key,
-        org.tensorflow.proto.framework.StructuredValue defaultValue) {
-      if (key == null) { throw new java.lang.NullPointerException(); }
-      java.util.Map<java.lang.String, org.tensorflow.proto.framework.StructuredValue> map =
-          internalGetFields().getMap();
-      return map.containsKey(key) ? map.get(key) : defaultValue;
-    }
-    /**
-     * <code>map&lt;string, .tensorflow.StructuredValue&gt; fields = 1;</code>
-     */
-
-    public org.tensorflow.proto.framework.StructuredValue getFieldsOrThrow(
-        java.lang.String key) {
-      if (key == null) { throw new java.lang.NullPointerException(); }
-      java.util.Map<java.lang.String, org.tensorflow.proto.framework.StructuredValue> map =
-          internalGetFields().getMap();
-      if (!map.containsKey(key)) {
-        throw new java.lang.IllegalArgumentException();
-      }
-      return map.get(key);
-    }
-
-    public Builder clearFields() {
-      internalGetMutableFields().getMutableMap()
-          .clear();
-      return this;
-    }
-    /**
-     * <code>map&lt;string, .tensorflow.StructuredValue&gt; fields = 1;</code>
-     */
-
-    public Builder removeFields(
-        java.lang.String key) {
-      if (key == null) { throw new java.lang.NullPointerException(); }
-      internalGetMutableFields().getMutableMap()
-          .remove(key);
-      return this;
-    }
-    /**
-     * Use alternate mutation accessors instead.
-     */
-    @java.lang.Deprecated
-    public java.util.Map<java.lang.String, org.tensorflow.proto.framework.StructuredValue>
-    getMutableFields() {
-      return internalGetMutableFields().getMutableMap();
-    }
-    /**
-     * <code>map&lt;string, .tensorflow.StructuredValue&gt; fields = 1;</code>
-     */
-    public Builder putFields(
-        java.lang.String key,
-        org.tensorflow.proto.framework.StructuredValue value) {
-      if (key == null) { throw new java.lang.NullPointerException(); }
-      if (value == null) { throw new java.lang.NullPointerException(); }
-      internalGetMutableFields().getMutableMap()
-          .put(key, value);
-      return this;
-    }
-    /**
-     * <code>map&lt;string, .tensorflow.StructuredValue&gt; fields = 1;</code>
-     */
-
-    public Builder putAllFields(
-        java.util.Map<java.lang.String, org.tensorflow.proto.framework.StructuredValue> values) {
-      internalGetMutableFields().getMutableMap()
-          .putAll(values);
-      return this;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.DictValue)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.DictValue)
-  private static final org.tensorflow.proto.framework.DictValue DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.DictValue();
-  }
-
-  public static org.tensorflow.proto.framework.DictValue getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<DictValue>
-      PARSER = new com.google.protobuf.AbstractParser<DictValue>() {
-    @java.lang.Override
-    public DictValue parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new DictValue(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<DictValue> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<DictValue> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.framework.DictValue getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DictValueOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DictValueOrBuilder.java
deleted file mode 100644
index 527fdffc935..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DictValueOrBuilder.java
+++ /dev/null
@@ -1,43 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/struct.proto
-
-package org.tensorflow.proto.framework;
-
-public interface DictValueOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.DictValue)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <code>map&lt;string, .tensorflow.StructuredValue&gt; fields = 1;</code>
-   */
-  int getFieldsCount();
-  /**
-   * <code>map&lt;string, .tensorflow.StructuredValue&gt; fields = 1;</code>
-   */
-  boolean containsFields(
-      java.lang.String key);
-  /**
-   * Use {@link #getFieldsMap()} instead.
-   */
-  @java.lang.Deprecated
-  java.util.Map<java.lang.String, org.tensorflow.proto.framework.StructuredValue>
-  getFields();
-  /**
-   * <code>map&lt;string, .tensorflow.StructuredValue&gt; fields = 1;</code>
-   */
-  java.util.Map<java.lang.String, org.tensorflow.proto.framework.StructuredValue>
-  getFieldsMap();
-  /**
-   * <code>map&lt;string, .tensorflow.StructuredValue&gt; fields = 1;</code>
-   */
-
-  org.tensorflow.proto.framework.StructuredValue getFieldsOrDefault(
-      java.lang.String key,
-      org.tensorflow.proto.framework.StructuredValue defaultValue);
-  /**
-   * <code>map&lt;string, .tensorflow.StructuredValue&gt; fields = 1;</code>
-   */
-
-  org.tensorflow.proto.framework.StructuredValue getFieldsOrThrow(
-      java.lang.String key);
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ErrorCodes.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ErrorCodes.java
deleted file mode 100644
index 113b8465373..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ErrorCodes.java
+++ /dev/null
@@ -1,40 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/lib/core/error_codes.proto
-
-package org.tensorflow.proto.framework;
-
-public final class ErrorCodes {
-  private ErrorCodes() {}
-  public static void registerAllExtensions(
-      com.google.protobuf.ExtensionRegistryLite registry) {
-  }
-
-  public static void registerAllExtensions(
-      com.google.protobuf.ExtensionRegistry registry) {
-    registerAllExtensions(
-        (com.google.protobuf.ExtensionRegistryLite) registry);
-  }
-
-  public static com.google.protobuf.Descriptors.FileDescriptor
-      getDescriptor() {
-    return descriptor;
-  }
-  private static  com.google.protobuf.Descriptors.FileDescriptor
-      descriptor;
-  static {
-    java.lang.String[] descriptorData = {
-      "\n*tensorflow/core/lib/core/error_codes.p" +
-      "roto\032*tensorflow/core/protobuf/error_cod" +
-      "es.protoB \n\036org.tensorflow.proto.framewo" +
-      "rkP\000b\006proto3"
-    };
-    descriptor = com.google.protobuf.Descriptors.FileDescriptor
-      .internalBuildGeneratedFileFrom(descriptorData,
-        new com.google.protobuf.Descriptors.FileDescriptor[] {
-          org.tensorflow.proto.framework.ErrorCodesProtos.getDescriptor(),
-        });
-    org.tensorflow.proto.framework.ErrorCodesProtos.getDescriptor();
-  }
-
-  // @@protoc_insertion_point(outer_class_scope)
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ErrorCodesProtos.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ErrorCodesProtos.java
deleted file mode 100644
index 9e424017f74..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ErrorCodesProtos.java
+++ /dev/null
@@ -1,49 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/error_codes.proto
-
-package org.tensorflow.proto.framework;
-
-public final class ErrorCodesProtos {
-  private ErrorCodesProtos() {}
-  public static void registerAllExtensions(
-      com.google.protobuf.ExtensionRegistryLite registry) {
-  }
-
-  public static void registerAllExtensions(
-      com.google.protobuf.ExtensionRegistry registry) {
-    registerAllExtensions(
-        (com.google.protobuf.ExtensionRegistryLite) registry);
-  }
-
-  public static com.google.protobuf.Descriptors.FileDescriptor
-      getDescriptor() {
-    return descriptor;
-  }
-  private static  com.google.protobuf.Descriptors.FileDescriptor
-      descriptor;
-  static {
-    java.lang.String[] descriptorData = {
-      "\n*tensorflow/core/protobuf/error_codes.p" +
-      "roto\022\020tensorflow.error*\204\003\n\004Code\022\006\n\002OK\020\000\022" +
-      "\r\n\tCANCELLED\020\001\022\013\n\007UNKNOWN\020\002\022\024\n\020INVALID_A" +
-      "RGUMENT\020\003\022\025\n\021DEADLINE_EXCEEDED\020\004\022\r\n\tNOT_" +
-      "FOUND\020\005\022\022\n\016ALREADY_EXISTS\020\006\022\025\n\021PERMISSIO" +
-      "N_DENIED\020\007\022\023\n\017UNAUTHENTICATED\020\020\022\026\n\022RESOU" +
-      "RCE_EXHAUSTED\020\010\022\027\n\023FAILED_PRECONDITION\020\t" +
-      "\022\013\n\007ABORTED\020\n\022\020\n\014OUT_OF_RANGE\020\013\022\021\n\rUNIMP" +
-      "LEMENTED\020\014\022\014\n\010INTERNAL\020\r\022\017\n\013UNAVAILABLE\020" +
-      "\016\022\r\n\tDATA_LOSS\020\017\022K\nGDO_NOT_USE_RESERVED_" +
-      "FOR_FUTURE_EXPANSION_USE_DEFAULT_IN_SWIT" +
-      "CH_INSTEAD_\020\024B\201\001\n\036org.tensorflow.proto.f" +
-      "rameworkB\020ErrorCodesProtosP\001ZHgithub.com" +
-      "/tensorflow/tensorflow/tensorflow/go/cor" +
-      "e/core_protos_go_proto\370\001\001b\006proto3"
-    };
-    descriptor = com.google.protobuf.Descriptors.FileDescriptor
-      .internalBuildGeneratedFileFrom(descriptorData,
-        new com.google.protobuf.Descriptors.FileDescriptor[] {
-        });
-  }
-
-  // @@protoc_insertion_point(outer_class_scope)
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/FunctionDef.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/FunctionDef.java
deleted file mode 100644
index 3437857d610..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/FunctionDef.java
+++ /dev/null
@@ -1,3415 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/framework/function.proto
-
-package org.tensorflow.proto.framework;
-
-/**
- * <pre>
- * A function can be instantiated when the runtime can bind every attr
- * with a value. When a GraphDef has a call to a function, it must
- * have binding for every attr defined in the signature.
- * TODO(zhifengc):
- *   * device spec, etc.
- * </pre>
- *
- * Protobuf type {@code tensorflow.FunctionDef}
- */
-public  final class FunctionDef extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.FunctionDef)
-    FunctionDefOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use FunctionDef.newBuilder() to construct.
-  private FunctionDef(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private FunctionDef() {
-    nodeDef_ = java.util.Collections.emptyList();
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new FunctionDef();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private FunctionDef(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    int mutable_bitField0_ = 0;
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 10: {
-            org.tensorflow.proto.framework.OpDef.Builder subBuilder = null;
-            if (signature_ != null) {
-              subBuilder = signature_.toBuilder();
-            }
-            signature_ = input.readMessage(org.tensorflow.proto.framework.OpDef.parser(), extensionRegistry);
-            if (subBuilder != null) {
-              subBuilder.mergeFrom(signature_);
-              signature_ = subBuilder.buildPartial();
-            }
-
-            break;
-          }
-          case 26: {
-            if (!((mutable_bitField0_ & 0x00000008) != 0)) {
-              nodeDef_ = new java.util.ArrayList<org.tensorflow.proto.framework.NodeDef>();
-              mutable_bitField0_ |= 0x00000008;
-            }
-            nodeDef_.add(
-                input.readMessage(org.tensorflow.proto.framework.NodeDef.parser(), extensionRegistry));
-            break;
-          }
-          case 34: {
-            if (!((mutable_bitField0_ & 0x00000010) != 0)) {
-              ret_ = com.google.protobuf.MapField.newMapField(
-                  RetDefaultEntryHolder.defaultEntry);
-              mutable_bitField0_ |= 0x00000010;
-            }
-            com.google.protobuf.MapEntry<java.lang.String, java.lang.String>
-            ret__ = input.readMessage(
-                RetDefaultEntryHolder.defaultEntry.getParserForType(), extensionRegistry);
-            ret_.getMutableMap().put(
-                ret__.getKey(), ret__.getValue());
-            break;
-          }
-          case 42: {
-            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
-              attr_ = com.google.protobuf.MapField.newMapField(
-                  AttrDefaultEntryHolder.defaultEntry);
-              mutable_bitField0_ |= 0x00000001;
-            }
-            com.google.protobuf.MapEntry<java.lang.String, org.tensorflow.proto.framework.AttrValue>
-            attr__ = input.readMessage(
-                AttrDefaultEntryHolder.defaultEntry.getParserForType(), extensionRegistry);
-            attr_.getMutableMap().put(
-                attr__.getKey(), attr__.getValue());
-            break;
-          }
-          case 50: {
-            if (!((mutable_bitField0_ & 0x00000020) != 0)) {
-              controlRet_ = com.google.protobuf.MapField.newMapField(
-                  ControlRetDefaultEntryHolder.defaultEntry);
-              mutable_bitField0_ |= 0x00000020;
-            }
-            com.google.protobuf.MapEntry<java.lang.String, java.lang.String>
-            controlRet__ = input.readMessage(
-                ControlRetDefaultEntryHolder.defaultEntry.getParserForType(), extensionRegistry);
-            controlRet_.getMutableMap().put(
-                controlRet__.getKey(), controlRet__.getValue());
-            break;
-          }
-          case 58: {
-            if (!((mutable_bitField0_ & 0x00000002) != 0)) {
-              argAttr_ = com.google.protobuf.MapField.newMapField(
-                  ArgAttrDefaultEntryHolder.defaultEntry);
-              mutable_bitField0_ |= 0x00000002;
-            }
-            com.google.protobuf.MapEntry<java.lang.Integer, org.tensorflow.proto.framework.FunctionDef.ArgAttrs>
-            argAttr__ = input.readMessage(
-                ArgAttrDefaultEntryHolder.defaultEntry.getParserForType(), extensionRegistry);
-            argAttr_.getMutableMap().put(
-                argAttr__.getKey(), argAttr__.getValue());
-            break;
-          }
-          case 66: {
-            if (!((mutable_bitField0_ & 0x00000004) != 0)) {
-              resourceArgUniqueId_ = com.google.protobuf.MapField.newMapField(
-                  ResourceArgUniqueIdDefaultEntryHolder.defaultEntry);
-              mutable_bitField0_ |= 0x00000004;
-            }
-            com.google.protobuf.MapEntry<java.lang.Integer, java.lang.Integer>
-            resourceArgUniqueId__ = input.readMessage(
-                ResourceArgUniqueIdDefaultEntryHolder.defaultEntry.getParserForType(), extensionRegistry);
-            resourceArgUniqueId_.getMutableMap().put(
-                resourceArgUniqueId__.getKey(), resourceArgUniqueId__.getValue());
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      if (((mutable_bitField0_ & 0x00000008) != 0)) {
-        nodeDef_ = java.util.Collections.unmodifiableList(nodeDef_);
-      }
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.framework.FunctionProtos.internal_static_tensorflow_FunctionDef_descriptor;
-  }
-
-  @SuppressWarnings({"rawtypes"})
-  @java.lang.Override
-  protected com.google.protobuf.MapField internalGetMapField(
-      int number) {
-    switch (number) {
-      case 5:
-        return internalGetAttr();
-      case 7:
-        return internalGetArgAttr();
-      case 8:
-        return internalGetResourceArgUniqueId();
-      case 4:
-        return internalGetRet();
-      case 6:
-        return internalGetControlRet();
-      default:
-        throw new RuntimeException(
-            "Invalid map field number: " + number);
-    }
-  }
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.framework.FunctionProtos.internal_static_tensorflow_FunctionDef_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.framework.FunctionDef.class, org.tensorflow.proto.framework.FunctionDef.Builder.class);
-  }
-
-  public interface ArgAttrsOrBuilder extends
-      // @@protoc_insertion_point(interface_extends:tensorflow.FunctionDef.ArgAttrs)
-      com.google.protobuf.MessageOrBuilder {
-
-    /**
-     * <code>map&lt;string, .tensorflow.AttrValue&gt; attr = 1;</code>
-     */
-    int getAttrCount();
-    /**
-     * <code>map&lt;string, .tensorflow.AttrValue&gt; attr = 1;</code>
-     */
-    boolean containsAttr(
-        java.lang.String key);
-    /**
-     * Use {@link #getAttrMap()} instead.
-     */
-    @java.lang.Deprecated
-    java.util.Map<java.lang.String, org.tensorflow.proto.framework.AttrValue>
-    getAttr();
-    /**
-     * <code>map&lt;string, .tensorflow.AttrValue&gt; attr = 1;</code>
-     */
-    java.util.Map<java.lang.String, org.tensorflow.proto.framework.AttrValue>
-    getAttrMap();
-    /**
-     * <code>map&lt;string, .tensorflow.AttrValue&gt; attr = 1;</code>
-     */
-
-    org.tensorflow.proto.framework.AttrValue getAttrOrDefault(
-        java.lang.String key,
-        org.tensorflow.proto.framework.AttrValue defaultValue);
-    /**
-     * <code>map&lt;string, .tensorflow.AttrValue&gt; attr = 1;</code>
-     */
-
-    org.tensorflow.proto.framework.AttrValue getAttrOrThrow(
-        java.lang.String key);
-  }
-  /**
-   * <pre>
-   * Attributes for function arguments. These attributes are the same set of
-   * valid attributes as to _Arg nodes.
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.FunctionDef.ArgAttrs}
-   */
-  public  static final class ArgAttrs extends
-      com.google.protobuf.GeneratedMessageV3 implements
-      // @@protoc_insertion_point(message_implements:tensorflow.FunctionDef.ArgAttrs)
-      ArgAttrsOrBuilder {
-  private static final long serialVersionUID = 0L;
-    // Use ArgAttrs.newBuilder() to construct.
-    private ArgAttrs(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-      super(builder);
-    }
-    private ArgAttrs() {
-    }
-
-    @java.lang.Override
-    @SuppressWarnings({"unused"})
-    protected java.lang.Object newInstance(
-        UnusedPrivateParameter unused) {
-      return new ArgAttrs();
-    }
-
-    @java.lang.Override
-    public final com.google.protobuf.UnknownFieldSet
-    getUnknownFields() {
-      return this.unknownFields;
-    }
-    private ArgAttrs(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      this();
-      if (extensionRegistry == null) {
-        throw new java.lang.NullPointerException();
-      }
-      int mutable_bitField0_ = 0;
-      com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-          com.google.protobuf.UnknownFieldSet.newBuilder();
-      try {
-        boolean done = false;
-        while (!done) {
-          int tag = input.readTag();
-          switch (tag) {
-            case 0:
-              done = true;
-              break;
-            case 10: {
-              if (!((mutable_bitField0_ & 0x00000001) != 0)) {
-                attr_ = com.google.protobuf.MapField.newMapField(
-                    AttrDefaultEntryHolder.defaultEntry);
-                mutable_bitField0_ |= 0x00000001;
-              }
-              com.google.protobuf.MapEntry<java.lang.String, org.tensorflow.proto.framework.AttrValue>
-              attr__ = input.readMessage(
-                  AttrDefaultEntryHolder.defaultEntry.getParserForType(), extensionRegistry);
-              attr_.getMutableMap().put(
-                  attr__.getKey(), attr__.getValue());
-              break;
-            }
-            default: {
-              if (!parseUnknownField(
-                  input, unknownFields, extensionRegistry, tag)) {
-                done = true;
-              }
-              break;
-            }
-          }
-        }
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        throw e.setUnfinishedMessage(this);
-      } catch (java.io.IOException e) {
-        throw new com.google.protobuf.InvalidProtocolBufferException(
-            e).setUnfinishedMessage(this);
-      } finally {
-        this.unknownFields = unknownFields.build();
-        makeExtensionsImmutable();
-      }
-    }
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.FunctionProtos.internal_static_tensorflow_FunctionDef_ArgAttrs_descriptor;
-    }
-
-    @SuppressWarnings({"rawtypes"})
-    @java.lang.Override
-    protected com.google.protobuf.MapField internalGetMapField(
-        int number) {
-      switch (number) {
-        case 1:
-          return internalGetAttr();
-        default:
-          throw new RuntimeException(
-              "Invalid map field number: " + number);
-      }
-    }
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.FunctionProtos.internal_static_tensorflow_FunctionDef_ArgAttrs_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.FunctionDef.ArgAttrs.class, org.tensorflow.proto.framework.FunctionDef.ArgAttrs.Builder.class);
-    }
-
-    public static final int ATTR_FIELD_NUMBER = 1;
-    private static final class AttrDefaultEntryHolder {
-      static final com.google.protobuf.MapEntry<
-          java.lang.String, org.tensorflow.proto.framework.AttrValue> defaultEntry =
-              com.google.protobuf.MapEntry
-              .<java.lang.String, org.tensorflow.proto.framework.AttrValue>newDefaultInstance(
-                  org.tensorflow.proto.framework.FunctionProtos.internal_static_tensorflow_FunctionDef_ArgAttrs_AttrEntry_descriptor, 
-                  com.google.protobuf.WireFormat.FieldType.STRING,
-                  "",
-                  com.google.protobuf.WireFormat.FieldType.MESSAGE,
-                  org.tensorflow.proto.framework.AttrValue.getDefaultInstance());
-    }
-    private com.google.protobuf.MapField<
-        java.lang.String, org.tensorflow.proto.framework.AttrValue> attr_;
-    private com.google.protobuf.MapField<java.lang.String, org.tensorflow.proto.framework.AttrValue>
-    internalGetAttr() {
-      if (attr_ == null) {
-        return com.google.protobuf.MapField.emptyMapField(
-            AttrDefaultEntryHolder.defaultEntry);
-      }
-      return attr_;
-    }
-
-    public int getAttrCount() {
-      return internalGetAttr().getMap().size();
-    }
-    /**
-     * <code>map&lt;string, .tensorflow.AttrValue&gt; attr = 1;</code>
-     */
-
-    public boolean containsAttr(
-        java.lang.String key) {
-      if (key == null) { throw new java.lang.NullPointerException(); }
-      return internalGetAttr().getMap().containsKey(key);
-    }
-    /**
-     * Use {@link #getAttrMap()} instead.
-     */
-    @java.lang.Deprecated
-    public java.util.Map<java.lang.String, org.tensorflow.proto.framework.AttrValue> getAttr() {
-      return getAttrMap();
-    }
-    /**
-     * <code>map&lt;string, .tensorflow.AttrValue&gt; attr = 1;</code>
-     */
-
-    public java.util.Map<java.lang.String, org.tensorflow.proto.framework.AttrValue> getAttrMap() {
-      return internalGetAttr().getMap();
-    }
-    /**
-     * <code>map&lt;string, .tensorflow.AttrValue&gt; attr = 1;</code>
-     */
-
-    public org.tensorflow.proto.framework.AttrValue getAttrOrDefault(
-        java.lang.String key,
-        org.tensorflow.proto.framework.AttrValue defaultValue) {
-      if (key == null) { throw new java.lang.NullPointerException(); }
-      java.util.Map<java.lang.String, org.tensorflow.proto.framework.AttrValue> map =
-          internalGetAttr().getMap();
-      return map.containsKey(key) ? map.get(key) : defaultValue;
-    }
-    /**
-     * <code>map&lt;string, .tensorflow.AttrValue&gt; attr = 1;</code>
-     */
-
-    public org.tensorflow.proto.framework.AttrValue getAttrOrThrow(
-        java.lang.String key) {
-      if (key == null) { throw new java.lang.NullPointerException(); }
-      java.util.Map<java.lang.String, org.tensorflow.proto.framework.AttrValue> map =
-          internalGetAttr().getMap();
-      if (!map.containsKey(key)) {
-        throw new java.lang.IllegalArgumentException();
-      }
-      return map.get(key);
-    }
-
-    private byte memoizedIsInitialized = -1;
-    @java.lang.Override
-    public final boolean isInitialized() {
-      byte isInitialized = memoizedIsInitialized;
-      if (isInitialized == 1) return true;
-      if (isInitialized == 0) return false;
-
-      memoizedIsInitialized = 1;
-      return true;
-    }
-
-    @java.lang.Override
-    public void writeTo(com.google.protobuf.CodedOutputStream output)
-                        throws java.io.IOException {
-      com.google.protobuf.GeneratedMessageV3
-        .serializeStringMapTo(
-          output,
-          internalGetAttr(),
-          AttrDefaultEntryHolder.defaultEntry,
-          1);
-      unknownFields.writeTo(output);
-    }
-
-    @java.lang.Override
-    public int getSerializedSize() {
-      int size = memoizedSize;
-      if (size != -1) return size;
-
-      size = 0;
-      for (java.util.Map.Entry<java.lang.String, org.tensorflow.proto.framework.AttrValue> entry
-           : internalGetAttr().getMap().entrySet()) {
-        com.google.protobuf.MapEntry<java.lang.String, org.tensorflow.proto.framework.AttrValue>
-        attr__ = AttrDefaultEntryHolder.defaultEntry.newBuilderForType()
-            .setKey(entry.getKey())
-            .setValue(entry.getValue())
-            .build();
-        size += com.google.protobuf.CodedOutputStream
-            .computeMessageSize(1, attr__);
-      }
-      size += unknownFields.getSerializedSize();
-      memoizedSize = size;
-      return size;
-    }
-
-    @java.lang.Override
-    public boolean equals(final java.lang.Object obj) {
-      if (obj == this) {
-       return true;
-      }
-      if (!(obj instanceof org.tensorflow.proto.framework.FunctionDef.ArgAttrs)) {
-        return super.equals(obj);
-      }
-      org.tensorflow.proto.framework.FunctionDef.ArgAttrs other = (org.tensorflow.proto.framework.FunctionDef.ArgAttrs) obj;
-
-      if (!internalGetAttr().equals(
-          other.internalGetAttr())) return false;
-      if (!unknownFields.equals(other.unknownFields)) return false;
-      return true;
-    }
-
-    @java.lang.Override
-    public int hashCode() {
-      if (memoizedHashCode != 0) {
-        return memoizedHashCode;
-      }
-      int hash = 41;
-      hash = (19 * hash) + getDescriptor().hashCode();
-      if (!internalGetAttr().getMap().isEmpty()) {
-        hash = (37 * hash) + ATTR_FIELD_NUMBER;
-        hash = (53 * hash) + internalGetAttr().hashCode();
-      }
-      hash = (29 * hash) + unknownFields.hashCode();
-      memoizedHashCode = hash;
-      return hash;
-    }
-
-    public static org.tensorflow.proto.framework.FunctionDef.ArgAttrs parseFrom(
-        java.nio.ByteBuffer data)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data);
-    }
-    public static org.tensorflow.proto.framework.FunctionDef.ArgAttrs parseFrom(
-        java.nio.ByteBuffer data,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.FunctionDef.ArgAttrs parseFrom(
-        com.google.protobuf.ByteString data)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data);
-    }
-    public static org.tensorflow.proto.framework.FunctionDef.ArgAttrs parseFrom(
-        com.google.protobuf.ByteString data,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.FunctionDef.ArgAttrs parseFrom(byte[] data)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data);
-    }
-    public static org.tensorflow.proto.framework.FunctionDef.ArgAttrs parseFrom(
-        byte[] data,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.FunctionDef.ArgAttrs parseFrom(java.io.InputStream input)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input);
-    }
-    public static org.tensorflow.proto.framework.FunctionDef.ArgAttrs parseFrom(
-        java.io.InputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.FunctionDef.ArgAttrs parseDelimitedFrom(java.io.InputStream input)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseDelimitedWithIOException(PARSER, input);
-    }
-    public static org.tensorflow.proto.framework.FunctionDef.ArgAttrs parseDelimitedFrom(
-        java.io.InputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.FunctionDef.ArgAttrs parseFrom(
-        com.google.protobuf.CodedInputStream input)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input);
-    }
-    public static org.tensorflow.proto.framework.FunctionDef.ArgAttrs parseFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input, extensionRegistry);
-    }
-
-    @java.lang.Override
-    public Builder newBuilderForType() { return newBuilder(); }
-    public static Builder newBuilder() {
-      return DEFAULT_INSTANCE.toBuilder();
-    }
-    public static Builder newBuilder(org.tensorflow.proto.framework.FunctionDef.ArgAttrs prototype) {
-      return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-    }
-    @java.lang.Override
-    public Builder toBuilder() {
-      return this == DEFAULT_INSTANCE
-          ? new Builder() : new Builder().mergeFrom(this);
-    }
-
-    @java.lang.Override
-    protected Builder newBuilderForType(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      Builder builder = new Builder(parent);
-      return builder;
-    }
-    /**
-     * <pre>
-     * Attributes for function arguments. These attributes are the same set of
-     * valid attributes as to _Arg nodes.
-     * </pre>
-     *
-     * Protobuf type {@code tensorflow.FunctionDef.ArgAttrs}
-     */
-    public static final class Builder extends
-        com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-        // @@protoc_insertion_point(builder_implements:tensorflow.FunctionDef.ArgAttrs)
-        org.tensorflow.proto.framework.FunctionDef.ArgAttrsOrBuilder {
-      public static final com.google.protobuf.Descriptors.Descriptor
-          getDescriptor() {
-        return org.tensorflow.proto.framework.FunctionProtos.internal_static_tensorflow_FunctionDef_ArgAttrs_descriptor;
-      }
-
-      @SuppressWarnings({"rawtypes"})
-      protected com.google.protobuf.MapField internalGetMapField(
-          int number) {
-        switch (number) {
-          case 1:
-            return internalGetAttr();
-          default:
-            throw new RuntimeException(
-                "Invalid map field number: " + number);
-        }
-      }
-      @SuppressWarnings({"rawtypes"})
-      protected com.google.protobuf.MapField internalGetMutableMapField(
-          int number) {
-        switch (number) {
-          case 1:
-            return internalGetMutableAttr();
-          default:
-            throw new RuntimeException(
-                "Invalid map field number: " + number);
-        }
-      }
-      @java.lang.Override
-      protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-          internalGetFieldAccessorTable() {
-        return org.tensorflow.proto.framework.FunctionProtos.internal_static_tensorflow_FunctionDef_ArgAttrs_fieldAccessorTable
-            .ensureFieldAccessorsInitialized(
-                org.tensorflow.proto.framework.FunctionDef.ArgAttrs.class, org.tensorflow.proto.framework.FunctionDef.ArgAttrs.Builder.class);
-      }
-
-      // Construct using org.tensorflow.proto.framework.FunctionDef.ArgAttrs.newBuilder()
-      private Builder() {
-        maybeForceBuilderInitialization();
-      }
-
-      private Builder(
-          com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-        super(parent);
-        maybeForceBuilderInitialization();
-      }
-      private void maybeForceBuilderInitialization() {
-        if (com.google.protobuf.GeneratedMessageV3
-                .alwaysUseFieldBuilders) {
-        }
-      }
-      @java.lang.Override
-      public Builder clear() {
-        super.clear();
-        internalGetMutableAttr().clear();
-        return this;
-      }
-
-      @java.lang.Override
-      public com.google.protobuf.Descriptors.Descriptor
-          getDescriptorForType() {
-        return org.tensorflow.proto.framework.FunctionProtos.internal_static_tensorflow_FunctionDef_ArgAttrs_descriptor;
-      }
-
-      @java.lang.Override
-      public org.tensorflow.proto.framework.FunctionDef.ArgAttrs getDefaultInstanceForType() {
-        return org.tensorflow.proto.framework.FunctionDef.ArgAttrs.getDefaultInstance();
-      }
-
-      @java.lang.Override
-      public org.tensorflow.proto.framework.FunctionDef.ArgAttrs build() {
-        org.tensorflow.proto.framework.FunctionDef.ArgAttrs result = buildPartial();
-        if (!result.isInitialized()) {
-          throw newUninitializedMessageException(result);
-        }
-        return result;
-      }
-
-      @java.lang.Override
-      public org.tensorflow.proto.framework.FunctionDef.ArgAttrs buildPartial() {
-        org.tensorflow.proto.framework.FunctionDef.ArgAttrs result = new org.tensorflow.proto.framework.FunctionDef.ArgAttrs(this);
-        int from_bitField0_ = bitField0_;
-        result.attr_ = internalGetAttr();
-        result.attr_.makeImmutable();
-        onBuilt();
-        return result;
-      }
-
-      @java.lang.Override
-      public Builder clone() {
-        return super.clone();
-      }
-      @java.lang.Override
-      public Builder setField(
-          com.google.protobuf.Descriptors.FieldDescriptor field,
-          java.lang.Object value) {
-        return super.setField(field, value);
-      }
-      @java.lang.Override
-      public Builder clearField(
-          com.google.protobuf.Descriptors.FieldDescriptor field) {
-        return super.clearField(field);
-      }
-      @java.lang.Override
-      public Builder clearOneof(
-          com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-        return super.clearOneof(oneof);
-      }
-      @java.lang.Override
-      public Builder setRepeatedField(
-          com.google.protobuf.Descriptors.FieldDescriptor field,
-          int index, java.lang.Object value) {
-        return super.setRepeatedField(field, index, value);
-      }
-      @java.lang.Override
-      public Builder addRepeatedField(
-          com.google.protobuf.Descriptors.FieldDescriptor field,
-          java.lang.Object value) {
-        return super.addRepeatedField(field, value);
-      }
-      @java.lang.Override
-      public Builder mergeFrom(com.google.protobuf.Message other) {
-        if (other instanceof org.tensorflow.proto.framework.FunctionDef.ArgAttrs) {
-          return mergeFrom((org.tensorflow.proto.framework.FunctionDef.ArgAttrs)other);
-        } else {
-          super.mergeFrom(other);
-          return this;
-        }
-      }
-
-      public Builder mergeFrom(org.tensorflow.proto.framework.FunctionDef.ArgAttrs other) {
-        if (other == org.tensorflow.proto.framework.FunctionDef.ArgAttrs.getDefaultInstance()) return this;
-        internalGetMutableAttr().mergeFrom(
-            other.internalGetAttr());
-        this.mergeUnknownFields(other.unknownFields);
-        onChanged();
-        return this;
-      }
-
-      @java.lang.Override
-      public final boolean isInitialized() {
-        return true;
-      }
-
-      @java.lang.Override
-      public Builder mergeFrom(
-          com.google.protobuf.CodedInputStream input,
-          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-          throws java.io.IOException {
-        org.tensorflow.proto.framework.FunctionDef.ArgAttrs parsedMessage = null;
-        try {
-          parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-        } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-          parsedMessage = (org.tensorflow.proto.framework.FunctionDef.ArgAttrs) e.getUnfinishedMessage();
-          throw e.unwrapIOException();
-        } finally {
-          if (parsedMessage != null) {
-            mergeFrom(parsedMessage);
-          }
-        }
-        return this;
-      }
-      private int bitField0_;
-
-      private com.google.protobuf.MapField<
-          java.lang.String, org.tensorflow.proto.framework.AttrValue> attr_;
-      private com.google.protobuf.MapField<java.lang.String, org.tensorflow.proto.framework.AttrValue>
-      internalGetAttr() {
-        if (attr_ == null) {
-          return com.google.protobuf.MapField.emptyMapField(
-              AttrDefaultEntryHolder.defaultEntry);
-        }
-        return attr_;
-      }
-      private com.google.protobuf.MapField<java.lang.String, org.tensorflow.proto.framework.AttrValue>
-      internalGetMutableAttr() {
-        onChanged();;
-        if (attr_ == null) {
-          attr_ = com.google.protobuf.MapField.newMapField(
-              AttrDefaultEntryHolder.defaultEntry);
-        }
-        if (!attr_.isMutable()) {
-          attr_ = attr_.copy();
-        }
-        return attr_;
-      }
-
-      public int getAttrCount() {
-        return internalGetAttr().getMap().size();
-      }
-      /**
-       * <code>map&lt;string, .tensorflow.AttrValue&gt; attr = 1;</code>
-       */
-
-      public boolean containsAttr(
-          java.lang.String key) {
-        if (key == null) { throw new java.lang.NullPointerException(); }
-        return internalGetAttr().getMap().containsKey(key);
-      }
-      /**
-       * Use {@link #getAttrMap()} instead.
-       */
-      @java.lang.Deprecated
-      public java.util.Map<java.lang.String, org.tensorflow.proto.framework.AttrValue> getAttr() {
-        return getAttrMap();
-      }
-      /**
-       * <code>map&lt;string, .tensorflow.AttrValue&gt; attr = 1;</code>
-       */
-
-      public java.util.Map<java.lang.String, org.tensorflow.proto.framework.AttrValue> getAttrMap() {
-        return internalGetAttr().getMap();
-      }
-      /**
-       * <code>map&lt;string, .tensorflow.AttrValue&gt; attr = 1;</code>
-       */
-
-      public org.tensorflow.proto.framework.AttrValue getAttrOrDefault(
-          java.lang.String key,
-          org.tensorflow.proto.framework.AttrValue defaultValue) {
-        if (key == null) { throw new java.lang.NullPointerException(); }
-        java.util.Map<java.lang.String, org.tensorflow.proto.framework.AttrValue> map =
-            internalGetAttr().getMap();
-        return map.containsKey(key) ? map.get(key) : defaultValue;
-      }
-      /**
-       * <code>map&lt;string, .tensorflow.AttrValue&gt; attr = 1;</code>
-       */
-
-      public org.tensorflow.proto.framework.AttrValue getAttrOrThrow(
-          java.lang.String key) {
-        if (key == null) { throw new java.lang.NullPointerException(); }
-        java.util.Map<java.lang.String, org.tensorflow.proto.framework.AttrValue> map =
-            internalGetAttr().getMap();
-        if (!map.containsKey(key)) {
-          throw new java.lang.IllegalArgumentException();
-        }
-        return map.get(key);
-      }
-
-      public Builder clearAttr() {
-        internalGetMutableAttr().getMutableMap()
-            .clear();
-        return this;
-      }
-      /**
-       * <code>map&lt;string, .tensorflow.AttrValue&gt; attr = 1;</code>
-       */
-
-      public Builder removeAttr(
-          java.lang.String key) {
-        if (key == null) { throw new java.lang.NullPointerException(); }
-        internalGetMutableAttr().getMutableMap()
-            .remove(key);
-        return this;
-      }
-      /**
-       * Use alternate mutation accessors instead.
-       */
-      @java.lang.Deprecated
-      public java.util.Map<java.lang.String, org.tensorflow.proto.framework.AttrValue>
-      getMutableAttr() {
-        return internalGetMutableAttr().getMutableMap();
-      }
-      /**
-       * <code>map&lt;string, .tensorflow.AttrValue&gt; attr = 1;</code>
-       */
-      public Builder putAttr(
-          java.lang.String key,
-          org.tensorflow.proto.framework.AttrValue value) {
-        if (key == null) { throw new java.lang.NullPointerException(); }
-        if (value == null) { throw new java.lang.NullPointerException(); }
-        internalGetMutableAttr().getMutableMap()
-            .put(key, value);
-        return this;
-      }
-      /**
-       * <code>map&lt;string, .tensorflow.AttrValue&gt; attr = 1;</code>
-       */
-
-      public Builder putAllAttr(
-          java.util.Map<java.lang.String, org.tensorflow.proto.framework.AttrValue> values) {
-        internalGetMutableAttr().getMutableMap()
-            .putAll(values);
-        return this;
-      }
-      @java.lang.Override
-      public final Builder setUnknownFields(
-          final com.google.protobuf.UnknownFieldSet unknownFields) {
-        return super.setUnknownFields(unknownFields);
-      }
-
-      @java.lang.Override
-      public final Builder mergeUnknownFields(
-          final com.google.protobuf.UnknownFieldSet unknownFields) {
-        return super.mergeUnknownFields(unknownFields);
-      }
-
-
-      // @@protoc_insertion_point(builder_scope:tensorflow.FunctionDef.ArgAttrs)
-    }
-
-    // @@protoc_insertion_point(class_scope:tensorflow.FunctionDef.ArgAttrs)
-    private static final org.tensorflow.proto.framework.FunctionDef.ArgAttrs DEFAULT_INSTANCE;
-    static {
-      DEFAULT_INSTANCE = new org.tensorflow.proto.framework.FunctionDef.ArgAttrs();
-    }
-
-    public static org.tensorflow.proto.framework.FunctionDef.ArgAttrs getDefaultInstance() {
-      return DEFAULT_INSTANCE;
-    }
-
-    private static final com.google.protobuf.Parser<ArgAttrs>
-        PARSER = new com.google.protobuf.AbstractParser<ArgAttrs>() {
-      @java.lang.Override
-      public ArgAttrs parsePartialFrom(
-          com.google.protobuf.CodedInputStream input,
-          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-          throws com.google.protobuf.InvalidProtocolBufferException {
-        return new ArgAttrs(input, extensionRegistry);
-      }
-    };
-
-    public static com.google.protobuf.Parser<ArgAttrs> parser() {
-      return PARSER;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Parser<ArgAttrs> getParserForType() {
-      return PARSER;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.FunctionDef.ArgAttrs getDefaultInstanceForType() {
-      return DEFAULT_INSTANCE;
-    }
-
-  }
-
-  public static final int SIGNATURE_FIELD_NUMBER = 1;
-  private org.tensorflow.proto.framework.OpDef signature_;
-  /**
-   * <pre>
-   * The definition of the function's name, arguments, return values,
-   * attrs etc.
-   * </pre>
-   *
-   * <code>.tensorflow.OpDef signature = 1;</code>
-   */
-  public boolean hasSignature() {
-    return signature_ != null;
-  }
-  /**
-   * <pre>
-   * The definition of the function's name, arguments, return values,
-   * attrs etc.
-   * </pre>
-   *
-   * <code>.tensorflow.OpDef signature = 1;</code>
-   */
-  public org.tensorflow.proto.framework.OpDef getSignature() {
-    return signature_ == null ? org.tensorflow.proto.framework.OpDef.getDefaultInstance() : signature_;
-  }
-  /**
-   * <pre>
-   * The definition of the function's name, arguments, return values,
-   * attrs etc.
-   * </pre>
-   *
-   * <code>.tensorflow.OpDef signature = 1;</code>
-   */
-  public org.tensorflow.proto.framework.OpDefOrBuilder getSignatureOrBuilder() {
-    return getSignature();
-  }
-
-  public static final int ATTR_FIELD_NUMBER = 5;
-  private static final class AttrDefaultEntryHolder {
-    static final com.google.protobuf.MapEntry<
-        java.lang.String, org.tensorflow.proto.framework.AttrValue> defaultEntry =
-            com.google.protobuf.MapEntry
-            .<java.lang.String, org.tensorflow.proto.framework.AttrValue>newDefaultInstance(
-                org.tensorflow.proto.framework.FunctionProtos.internal_static_tensorflow_FunctionDef_AttrEntry_descriptor, 
-                com.google.protobuf.WireFormat.FieldType.STRING,
-                "",
-                com.google.protobuf.WireFormat.FieldType.MESSAGE,
-                org.tensorflow.proto.framework.AttrValue.getDefaultInstance());
-  }
-  private com.google.protobuf.MapField<
-      java.lang.String, org.tensorflow.proto.framework.AttrValue> attr_;
-  private com.google.protobuf.MapField<java.lang.String, org.tensorflow.proto.framework.AttrValue>
-  internalGetAttr() {
-    if (attr_ == null) {
-      return com.google.protobuf.MapField.emptyMapField(
-          AttrDefaultEntryHolder.defaultEntry);
-    }
-    return attr_;
-  }
-
-  public int getAttrCount() {
-    return internalGetAttr().getMap().size();
-  }
-  /**
-   * <pre>
-   * Attributes specific to this function definition.
-   * </pre>
-   *
-   * <code>map&lt;string, .tensorflow.AttrValue&gt; attr = 5;</code>
-   */
-
-  public boolean containsAttr(
-      java.lang.String key) {
-    if (key == null) { throw new java.lang.NullPointerException(); }
-    return internalGetAttr().getMap().containsKey(key);
-  }
-  /**
-   * Use {@link #getAttrMap()} instead.
-   */
-  @java.lang.Deprecated
-  public java.util.Map<java.lang.String, org.tensorflow.proto.framework.AttrValue> getAttr() {
-    return getAttrMap();
-  }
-  /**
-   * <pre>
-   * Attributes specific to this function definition.
-   * </pre>
-   *
-   * <code>map&lt;string, .tensorflow.AttrValue&gt; attr = 5;</code>
-   */
-
-  public java.util.Map<java.lang.String, org.tensorflow.proto.framework.AttrValue> getAttrMap() {
-    return internalGetAttr().getMap();
-  }
-  /**
-   * <pre>
-   * Attributes specific to this function definition.
-   * </pre>
-   *
-   * <code>map&lt;string, .tensorflow.AttrValue&gt; attr = 5;</code>
-   */
-
-  public org.tensorflow.proto.framework.AttrValue getAttrOrDefault(
-      java.lang.String key,
-      org.tensorflow.proto.framework.AttrValue defaultValue) {
-    if (key == null) { throw new java.lang.NullPointerException(); }
-    java.util.Map<java.lang.String, org.tensorflow.proto.framework.AttrValue> map =
-        internalGetAttr().getMap();
-    return map.containsKey(key) ? map.get(key) : defaultValue;
-  }
-  /**
-   * <pre>
-   * Attributes specific to this function definition.
-   * </pre>
-   *
-   * <code>map&lt;string, .tensorflow.AttrValue&gt; attr = 5;</code>
-   */
-
-  public org.tensorflow.proto.framework.AttrValue getAttrOrThrow(
-      java.lang.String key) {
-    if (key == null) { throw new java.lang.NullPointerException(); }
-    java.util.Map<java.lang.String, org.tensorflow.proto.framework.AttrValue> map =
-        internalGetAttr().getMap();
-    if (!map.containsKey(key)) {
-      throw new java.lang.IllegalArgumentException();
-    }
-    return map.get(key);
-  }
-
-  public static final int ARG_ATTR_FIELD_NUMBER = 7;
-  private static final class ArgAttrDefaultEntryHolder {
-    static final com.google.protobuf.MapEntry<
-        java.lang.Integer, org.tensorflow.proto.framework.FunctionDef.ArgAttrs> defaultEntry =
-            com.google.protobuf.MapEntry
-            .<java.lang.Integer, org.tensorflow.proto.framework.FunctionDef.ArgAttrs>newDefaultInstance(
-                org.tensorflow.proto.framework.FunctionProtos.internal_static_tensorflow_FunctionDef_ArgAttrEntry_descriptor, 
-                com.google.protobuf.WireFormat.FieldType.UINT32,
-                0,
-                com.google.protobuf.WireFormat.FieldType.MESSAGE,
-                org.tensorflow.proto.framework.FunctionDef.ArgAttrs.getDefaultInstance());
-  }
-  private com.google.protobuf.MapField<
-      java.lang.Integer, org.tensorflow.proto.framework.FunctionDef.ArgAttrs> argAttr_;
-  private com.google.protobuf.MapField<java.lang.Integer, org.tensorflow.proto.framework.FunctionDef.ArgAttrs>
-  internalGetArgAttr() {
-    if (argAttr_ == null) {
-      return com.google.protobuf.MapField.emptyMapField(
-          ArgAttrDefaultEntryHolder.defaultEntry);
-    }
-    return argAttr_;
-  }
-
-  public int getArgAttrCount() {
-    return internalGetArgAttr().getMap().size();
-  }
-  /**
-   * <code>map&lt;uint32, .tensorflow.FunctionDef.ArgAttrs&gt; arg_attr = 7;</code>
-   */
-
-  public boolean containsArgAttr(
-      int key) {
-    
-    return internalGetArgAttr().getMap().containsKey(key);
-  }
-  /**
-   * Use {@link #getArgAttrMap()} instead.
-   */
-  @java.lang.Deprecated
-  public java.util.Map<java.lang.Integer, org.tensorflow.proto.framework.FunctionDef.ArgAttrs> getArgAttr() {
-    return getArgAttrMap();
-  }
-  /**
-   * <code>map&lt;uint32, .tensorflow.FunctionDef.ArgAttrs&gt; arg_attr = 7;</code>
-   */
-
-  public java.util.Map<java.lang.Integer, org.tensorflow.proto.framework.FunctionDef.ArgAttrs> getArgAttrMap() {
-    return internalGetArgAttr().getMap();
-  }
-  /**
-   * <code>map&lt;uint32, .tensorflow.FunctionDef.ArgAttrs&gt; arg_attr = 7;</code>
-   */
-
-  public org.tensorflow.proto.framework.FunctionDef.ArgAttrs getArgAttrOrDefault(
-      int key,
-      org.tensorflow.proto.framework.FunctionDef.ArgAttrs defaultValue) {
-    
-    java.util.Map<java.lang.Integer, org.tensorflow.proto.framework.FunctionDef.ArgAttrs> map =
-        internalGetArgAttr().getMap();
-    return map.containsKey(key) ? map.get(key) : defaultValue;
-  }
-  /**
-   * <code>map&lt;uint32, .tensorflow.FunctionDef.ArgAttrs&gt; arg_attr = 7;</code>
-   */
-
-  public org.tensorflow.proto.framework.FunctionDef.ArgAttrs getArgAttrOrThrow(
-      int key) {
-    
-    java.util.Map<java.lang.Integer, org.tensorflow.proto.framework.FunctionDef.ArgAttrs> map =
-        internalGetArgAttr().getMap();
-    if (!map.containsKey(key)) {
-      throw new java.lang.IllegalArgumentException();
-    }
-    return map.get(key);
-  }
-
-  public static final int RESOURCE_ARG_UNIQUE_ID_FIELD_NUMBER = 8;
-  private static final class ResourceArgUniqueIdDefaultEntryHolder {
-    static final com.google.protobuf.MapEntry<
-        java.lang.Integer, java.lang.Integer> defaultEntry =
-            com.google.protobuf.MapEntry
-            .<java.lang.Integer, java.lang.Integer>newDefaultInstance(
-                org.tensorflow.proto.framework.FunctionProtos.internal_static_tensorflow_FunctionDef_ResourceArgUniqueIdEntry_descriptor, 
-                com.google.protobuf.WireFormat.FieldType.UINT32,
-                0,
-                com.google.protobuf.WireFormat.FieldType.UINT32,
-                0);
-  }
-  private com.google.protobuf.MapField<
-      java.lang.Integer, java.lang.Integer> resourceArgUniqueId_;
-  private com.google.protobuf.MapField<java.lang.Integer, java.lang.Integer>
-  internalGetResourceArgUniqueId() {
-    if (resourceArgUniqueId_ == null) {
-      return com.google.protobuf.MapField.emptyMapField(
-          ResourceArgUniqueIdDefaultEntryHolder.defaultEntry);
-    }
-    return resourceArgUniqueId_;
-  }
-
-  public int getResourceArgUniqueIdCount() {
-    return internalGetResourceArgUniqueId().getMap().size();
-  }
-  /**
-   * <pre>
-   * Unique IDs for each resource argument, used to track aliasing resources. If
-   * Argument A and Argument B alias each other, then
-   * resource_arg_unique_ids[A.index] == resource_arg_unique_ids[B.index].
-   * If this field is empty, none of the arguments could alias; otherwise, every
-   * resource argument should have an entry in this field.
-   * When instantiated, the unique IDs will be attached to the _Arg nodes'
-   * "_resource_arg_unique_id" attribute.
-   * </pre>
-   *
-   * <code>map&lt;uint32, uint32&gt; resource_arg_unique_id = 8;</code>
-   */
-
-  public boolean containsResourceArgUniqueId(
-      int key) {
-    
-    return internalGetResourceArgUniqueId().getMap().containsKey(key);
-  }
-  /**
-   * Use {@link #getResourceArgUniqueIdMap()} instead.
-   */
-  @java.lang.Deprecated
-  public java.util.Map<java.lang.Integer, java.lang.Integer> getResourceArgUniqueId() {
-    return getResourceArgUniqueIdMap();
-  }
-  /**
-   * <pre>
-   * Unique IDs for each resource argument, used to track aliasing resources. If
-   * Argument A and Argument B alias each other, then
-   * resource_arg_unique_ids[A.index] == resource_arg_unique_ids[B.index].
-   * If this field is empty, none of the arguments could alias; otherwise, every
-   * resource argument should have an entry in this field.
-   * When instantiated, the unique IDs will be attached to the _Arg nodes'
-   * "_resource_arg_unique_id" attribute.
-   * </pre>
-   *
-   * <code>map&lt;uint32, uint32&gt; resource_arg_unique_id = 8;</code>
-   */
-
-  public java.util.Map<java.lang.Integer, java.lang.Integer> getResourceArgUniqueIdMap() {
-    return internalGetResourceArgUniqueId().getMap();
-  }
-  /**
-   * <pre>
-   * Unique IDs for each resource argument, used to track aliasing resources. If
-   * Argument A and Argument B alias each other, then
-   * resource_arg_unique_ids[A.index] == resource_arg_unique_ids[B.index].
-   * If this field is empty, none of the arguments could alias; otherwise, every
-   * resource argument should have an entry in this field.
-   * When instantiated, the unique IDs will be attached to the _Arg nodes'
-   * "_resource_arg_unique_id" attribute.
-   * </pre>
-   *
-   * <code>map&lt;uint32, uint32&gt; resource_arg_unique_id = 8;</code>
-   */
-
-  public int getResourceArgUniqueIdOrDefault(
-      int key,
-      int defaultValue) {
-    
-    java.util.Map<java.lang.Integer, java.lang.Integer> map =
-        internalGetResourceArgUniqueId().getMap();
-    return map.containsKey(key) ? map.get(key) : defaultValue;
-  }
-  /**
-   * <pre>
-   * Unique IDs for each resource argument, used to track aliasing resources. If
-   * Argument A and Argument B alias each other, then
-   * resource_arg_unique_ids[A.index] == resource_arg_unique_ids[B.index].
-   * If this field is empty, none of the arguments could alias; otherwise, every
-   * resource argument should have an entry in this field.
-   * When instantiated, the unique IDs will be attached to the _Arg nodes'
-   * "_resource_arg_unique_id" attribute.
-   * </pre>
-   *
-   * <code>map&lt;uint32, uint32&gt; resource_arg_unique_id = 8;</code>
-   */
-
-  public int getResourceArgUniqueIdOrThrow(
-      int key) {
-    
-    java.util.Map<java.lang.Integer, java.lang.Integer> map =
-        internalGetResourceArgUniqueId().getMap();
-    if (!map.containsKey(key)) {
-      throw new java.lang.IllegalArgumentException();
-    }
-    return map.get(key);
-  }
-
-  public static final int NODE_DEF_FIELD_NUMBER = 3;
-  private java.util.List<org.tensorflow.proto.framework.NodeDef> nodeDef_;
-  /**
-   * <pre>
-   * By convention, "op" in node_def is resolved by consulting with a
-   * user-defined library first. If not resolved, "func" is assumed to
-   * be a builtin op.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.NodeDef node_def = 3;</code>
-   */
-  public java.util.List<org.tensorflow.proto.framework.NodeDef> getNodeDefList() {
-    return nodeDef_;
-  }
-  /**
-   * <pre>
-   * By convention, "op" in node_def is resolved by consulting with a
-   * user-defined library first. If not resolved, "func" is assumed to
-   * be a builtin op.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.NodeDef node_def = 3;</code>
-   */
-  public java.util.List<? extends org.tensorflow.proto.framework.NodeDefOrBuilder> 
-      getNodeDefOrBuilderList() {
-    return nodeDef_;
-  }
-  /**
-   * <pre>
-   * By convention, "op" in node_def is resolved by consulting with a
-   * user-defined library first. If not resolved, "func" is assumed to
-   * be a builtin op.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.NodeDef node_def = 3;</code>
-   */
-  public int getNodeDefCount() {
-    return nodeDef_.size();
-  }
-  /**
-   * <pre>
-   * By convention, "op" in node_def is resolved by consulting with a
-   * user-defined library first. If not resolved, "func" is assumed to
-   * be a builtin op.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.NodeDef node_def = 3;</code>
-   */
-  public org.tensorflow.proto.framework.NodeDef getNodeDef(int index) {
-    return nodeDef_.get(index);
-  }
-  /**
-   * <pre>
-   * By convention, "op" in node_def is resolved by consulting with a
-   * user-defined library first. If not resolved, "func" is assumed to
-   * be a builtin op.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.NodeDef node_def = 3;</code>
-   */
-  public org.tensorflow.proto.framework.NodeDefOrBuilder getNodeDefOrBuilder(
-      int index) {
-    return nodeDef_.get(index);
-  }
-
-  public static final int RET_FIELD_NUMBER = 4;
-  private static final class RetDefaultEntryHolder {
-    static final com.google.protobuf.MapEntry<
-        java.lang.String, java.lang.String> defaultEntry =
-            com.google.protobuf.MapEntry
-            .<java.lang.String, java.lang.String>newDefaultInstance(
-                org.tensorflow.proto.framework.FunctionProtos.internal_static_tensorflow_FunctionDef_RetEntry_descriptor, 
-                com.google.protobuf.WireFormat.FieldType.STRING,
-                "",
-                com.google.protobuf.WireFormat.FieldType.STRING,
-                "");
-  }
-  private com.google.protobuf.MapField<
-      java.lang.String, java.lang.String> ret_;
-  private com.google.protobuf.MapField<java.lang.String, java.lang.String>
-  internalGetRet() {
-    if (ret_ == null) {
-      return com.google.protobuf.MapField.emptyMapField(
-          RetDefaultEntryHolder.defaultEntry);
-    }
-    return ret_;
-  }
-
-  public int getRetCount() {
-    return internalGetRet().getMap().size();
-  }
-  /**
-   * <pre>
-   * A mapping from the output arg names from `signature` to the
-   * outputs from `node_def` that should be returned by the function.
-   * </pre>
-   *
-   * <code>map&lt;string, string&gt; ret = 4;</code>
-   */
-
-  public boolean containsRet(
-      java.lang.String key) {
-    if (key == null) { throw new java.lang.NullPointerException(); }
-    return internalGetRet().getMap().containsKey(key);
-  }
-  /**
-   * Use {@link #getRetMap()} instead.
-   */
-  @java.lang.Deprecated
-  public java.util.Map<java.lang.String, java.lang.String> getRet() {
-    return getRetMap();
-  }
-  /**
-   * <pre>
-   * A mapping from the output arg names from `signature` to the
-   * outputs from `node_def` that should be returned by the function.
-   * </pre>
-   *
-   * <code>map&lt;string, string&gt; ret = 4;</code>
-   */
-
-  public java.util.Map<java.lang.String, java.lang.String> getRetMap() {
-    return internalGetRet().getMap();
-  }
-  /**
-   * <pre>
-   * A mapping from the output arg names from `signature` to the
-   * outputs from `node_def` that should be returned by the function.
-   * </pre>
-   *
-   * <code>map&lt;string, string&gt; ret = 4;</code>
-   */
-
-  public java.lang.String getRetOrDefault(
-      java.lang.String key,
-      java.lang.String defaultValue) {
-    if (key == null) { throw new java.lang.NullPointerException(); }
-    java.util.Map<java.lang.String, java.lang.String> map =
-        internalGetRet().getMap();
-    return map.containsKey(key) ? map.get(key) : defaultValue;
-  }
-  /**
-   * <pre>
-   * A mapping from the output arg names from `signature` to the
-   * outputs from `node_def` that should be returned by the function.
-   * </pre>
-   *
-   * <code>map&lt;string, string&gt; ret = 4;</code>
-   */
-
-  public java.lang.String getRetOrThrow(
-      java.lang.String key) {
-    if (key == null) { throw new java.lang.NullPointerException(); }
-    java.util.Map<java.lang.String, java.lang.String> map =
-        internalGetRet().getMap();
-    if (!map.containsKey(key)) {
-      throw new java.lang.IllegalArgumentException();
-    }
-    return map.get(key);
-  }
-
-  public static final int CONTROL_RET_FIELD_NUMBER = 6;
-  private static final class ControlRetDefaultEntryHolder {
-    static final com.google.protobuf.MapEntry<
-        java.lang.String, java.lang.String> defaultEntry =
-            com.google.protobuf.MapEntry
-            .<java.lang.String, java.lang.String>newDefaultInstance(
-                org.tensorflow.proto.framework.FunctionProtos.internal_static_tensorflow_FunctionDef_ControlRetEntry_descriptor, 
-                com.google.protobuf.WireFormat.FieldType.STRING,
-                "",
-                com.google.protobuf.WireFormat.FieldType.STRING,
-                "");
-  }
-  private com.google.protobuf.MapField<
-      java.lang.String, java.lang.String> controlRet_;
-  private com.google.protobuf.MapField<java.lang.String, java.lang.String>
-  internalGetControlRet() {
-    if (controlRet_ == null) {
-      return com.google.protobuf.MapField.emptyMapField(
-          ControlRetDefaultEntryHolder.defaultEntry);
-    }
-    return controlRet_;
-  }
-
-  public int getControlRetCount() {
-    return internalGetControlRet().getMap().size();
-  }
-  /**
-   * <pre>
-   * A mapping from control output names from `signature` to node names in
-   * `node_def` which should be control outputs of this function.
-   * </pre>
-   *
-   * <code>map&lt;string, string&gt; control_ret = 6;</code>
-   */
-
-  public boolean containsControlRet(
-      java.lang.String key) {
-    if (key == null) { throw new java.lang.NullPointerException(); }
-    return internalGetControlRet().getMap().containsKey(key);
-  }
-  /**
-   * Use {@link #getControlRetMap()} instead.
-   */
-  @java.lang.Deprecated
-  public java.util.Map<java.lang.String, java.lang.String> getControlRet() {
-    return getControlRetMap();
-  }
-  /**
-   * <pre>
-   * A mapping from control output names from `signature` to node names in
-   * `node_def` which should be control outputs of this function.
-   * </pre>
-   *
-   * <code>map&lt;string, string&gt; control_ret = 6;</code>
-   */
-
-  public java.util.Map<java.lang.String, java.lang.String> getControlRetMap() {
-    return internalGetControlRet().getMap();
-  }
-  /**
-   * <pre>
-   * A mapping from control output names from `signature` to node names in
-   * `node_def` which should be control outputs of this function.
-   * </pre>
-   *
-   * <code>map&lt;string, string&gt; control_ret = 6;</code>
-   */
-
-  public java.lang.String getControlRetOrDefault(
-      java.lang.String key,
-      java.lang.String defaultValue) {
-    if (key == null) { throw new java.lang.NullPointerException(); }
-    java.util.Map<java.lang.String, java.lang.String> map =
-        internalGetControlRet().getMap();
-    return map.containsKey(key) ? map.get(key) : defaultValue;
-  }
-  /**
-   * <pre>
-   * A mapping from control output names from `signature` to node names in
-   * `node_def` which should be control outputs of this function.
-   * </pre>
-   *
-   * <code>map&lt;string, string&gt; control_ret = 6;</code>
-   */
-
-  public java.lang.String getControlRetOrThrow(
-      java.lang.String key) {
-    if (key == null) { throw new java.lang.NullPointerException(); }
-    java.util.Map<java.lang.String, java.lang.String> map =
-        internalGetControlRet().getMap();
-    if (!map.containsKey(key)) {
-      throw new java.lang.IllegalArgumentException();
-    }
-    return map.get(key);
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    if (signature_ != null) {
-      output.writeMessage(1, getSignature());
-    }
-    for (int i = 0; i < nodeDef_.size(); i++) {
-      output.writeMessage(3, nodeDef_.get(i));
-    }
-    com.google.protobuf.GeneratedMessageV3
-      .serializeStringMapTo(
-        output,
-        internalGetRet(),
-        RetDefaultEntryHolder.defaultEntry,
-        4);
-    com.google.protobuf.GeneratedMessageV3
-      .serializeStringMapTo(
-        output,
-        internalGetAttr(),
-        AttrDefaultEntryHolder.defaultEntry,
-        5);
-    com.google.protobuf.GeneratedMessageV3
-      .serializeStringMapTo(
-        output,
-        internalGetControlRet(),
-        ControlRetDefaultEntryHolder.defaultEntry,
-        6);
-    com.google.protobuf.GeneratedMessageV3
-      .serializeIntegerMapTo(
-        output,
-        internalGetArgAttr(),
-        ArgAttrDefaultEntryHolder.defaultEntry,
-        7);
-    com.google.protobuf.GeneratedMessageV3
-      .serializeIntegerMapTo(
-        output,
-        internalGetResourceArgUniqueId(),
-        ResourceArgUniqueIdDefaultEntryHolder.defaultEntry,
-        8);
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (signature_ != null) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(1, getSignature());
-    }
-    for (int i = 0; i < nodeDef_.size(); i++) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(3, nodeDef_.get(i));
-    }
-    for (java.util.Map.Entry<java.lang.String, java.lang.String> entry
-         : internalGetRet().getMap().entrySet()) {
-      com.google.protobuf.MapEntry<java.lang.String, java.lang.String>
-      ret__ = RetDefaultEntryHolder.defaultEntry.newBuilderForType()
-          .setKey(entry.getKey())
-          .setValue(entry.getValue())
-          .build();
-      size += com.google.protobuf.CodedOutputStream
-          .computeMessageSize(4, ret__);
-    }
-    for (java.util.Map.Entry<java.lang.String, org.tensorflow.proto.framework.AttrValue> entry
-         : internalGetAttr().getMap().entrySet()) {
-      com.google.protobuf.MapEntry<java.lang.String, org.tensorflow.proto.framework.AttrValue>
-      attr__ = AttrDefaultEntryHolder.defaultEntry.newBuilderForType()
-          .setKey(entry.getKey())
-          .setValue(entry.getValue())
-          .build();
-      size += com.google.protobuf.CodedOutputStream
-          .computeMessageSize(5, attr__);
-    }
-    for (java.util.Map.Entry<java.lang.String, java.lang.String> entry
-         : internalGetControlRet().getMap().entrySet()) {
-      com.google.protobuf.MapEntry<java.lang.String, java.lang.String>
-      controlRet__ = ControlRetDefaultEntryHolder.defaultEntry.newBuilderForType()
-          .setKey(entry.getKey())
-          .setValue(entry.getValue())
-          .build();
-      size += com.google.protobuf.CodedOutputStream
-          .computeMessageSize(6, controlRet__);
-    }
-    for (java.util.Map.Entry<java.lang.Integer, org.tensorflow.proto.framework.FunctionDef.ArgAttrs> entry
-         : internalGetArgAttr().getMap().entrySet()) {
-      com.google.protobuf.MapEntry<java.lang.Integer, org.tensorflow.proto.framework.FunctionDef.ArgAttrs>
-      argAttr__ = ArgAttrDefaultEntryHolder.defaultEntry.newBuilderForType()
-          .setKey(entry.getKey())
-          .setValue(entry.getValue())
-          .build();
-      size += com.google.protobuf.CodedOutputStream
-          .computeMessageSize(7, argAttr__);
-    }
-    for (java.util.Map.Entry<java.lang.Integer, java.lang.Integer> entry
-         : internalGetResourceArgUniqueId().getMap().entrySet()) {
-      com.google.protobuf.MapEntry<java.lang.Integer, java.lang.Integer>
-      resourceArgUniqueId__ = ResourceArgUniqueIdDefaultEntryHolder.defaultEntry.newBuilderForType()
-          .setKey(entry.getKey())
-          .setValue(entry.getValue())
-          .build();
-      size += com.google.protobuf.CodedOutputStream
-          .computeMessageSize(8, resourceArgUniqueId__);
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.framework.FunctionDef)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.framework.FunctionDef other = (org.tensorflow.proto.framework.FunctionDef) obj;
-
-    if (hasSignature() != other.hasSignature()) return false;
-    if (hasSignature()) {
-      if (!getSignature()
-          .equals(other.getSignature())) return false;
-    }
-    if (!internalGetAttr().equals(
-        other.internalGetAttr())) return false;
-    if (!internalGetArgAttr().equals(
-        other.internalGetArgAttr())) return false;
-    if (!internalGetResourceArgUniqueId().equals(
-        other.internalGetResourceArgUniqueId())) return false;
-    if (!getNodeDefList()
-        .equals(other.getNodeDefList())) return false;
-    if (!internalGetRet().equals(
-        other.internalGetRet())) return false;
-    if (!internalGetControlRet().equals(
-        other.internalGetControlRet())) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    if (hasSignature()) {
-      hash = (37 * hash) + SIGNATURE_FIELD_NUMBER;
-      hash = (53 * hash) + getSignature().hashCode();
-    }
-    if (!internalGetAttr().getMap().isEmpty()) {
-      hash = (37 * hash) + ATTR_FIELD_NUMBER;
-      hash = (53 * hash) + internalGetAttr().hashCode();
-    }
-    if (!internalGetArgAttr().getMap().isEmpty()) {
-      hash = (37 * hash) + ARG_ATTR_FIELD_NUMBER;
-      hash = (53 * hash) + internalGetArgAttr().hashCode();
-    }
-    if (!internalGetResourceArgUniqueId().getMap().isEmpty()) {
-      hash = (37 * hash) + RESOURCE_ARG_UNIQUE_ID_FIELD_NUMBER;
-      hash = (53 * hash) + internalGetResourceArgUniqueId().hashCode();
-    }
-    if (getNodeDefCount() > 0) {
-      hash = (37 * hash) + NODE_DEF_FIELD_NUMBER;
-      hash = (53 * hash) + getNodeDefList().hashCode();
-    }
-    if (!internalGetRet().getMap().isEmpty()) {
-      hash = (37 * hash) + RET_FIELD_NUMBER;
-      hash = (53 * hash) + internalGetRet().hashCode();
-    }
-    if (!internalGetControlRet().getMap().isEmpty()) {
-      hash = (37 * hash) + CONTROL_RET_FIELD_NUMBER;
-      hash = (53 * hash) + internalGetControlRet().hashCode();
-    }
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.framework.FunctionDef parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.FunctionDef parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.FunctionDef parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.FunctionDef parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.FunctionDef parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.FunctionDef parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.FunctionDef parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.FunctionDef parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.FunctionDef parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.FunctionDef parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.FunctionDef parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.FunctionDef parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.framework.FunctionDef prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * <pre>
-   * A function can be instantiated when the runtime can bind every attr
-   * with a value. When a GraphDef has a call to a function, it must
-   * have binding for every attr defined in the signature.
-   * TODO(zhifengc):
-   *   * device spec, etc.
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.FunctionDef}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.FunctionDef)
-      org.tensorflow.proto.framework.FunctionDefOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.FunctionProtos.internal_static_tensorflow_FunctionDef_descriptor;
-    }
-
-    @SuppressWarnings({"rawtypes"})
-    protected com.google.protobuf.MapField internalGetMapField(
-        int number) {
-      switch (number) {
-        case 5:
-          return internalGetAttr();
-        case 7:
-          return internalGetArgAttr();
-        case 8:
-          return internalGetResourceArgUniqueId();
-        case 4:
-          return internalGetRet();
-        case 6:
-          return internalGetControlRet();
-        default:
-          throw new RuntimeException(
-              "Invalid map field number: " + number);
-      }
-    }
-    @SuppressWarnings({"rawtypes"})
-    protected com.google.protobuf.MapField internalGetMutableMapField(
-        int number) {
-      switch (number) {
-        case 5:
-          return internalGetMutableAttr();
-        case 7:
-          return internalGetMutableArgAttr();
-        case 8:
-          return internalGetMutableResourceArgUniqueId();
-        case 4:
-          return internalGetMutableRet();
-        case 6:
-          return internalGetMutableControlRet();
-        default:
-          throw new RuntimeException(
-              "Invalid map field number: " + number);
-      }
-    }
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.FunctionProtos.internal_static_tensorflow_FunctionDef_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.FunctionDef.class, org.tensorflow.proto.framework.FunctionDef.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.framework.FunctionDef.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-        getNodeDefFieldBuilder();
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      if (signatureBuilder_ == null) {
-        signature_ = null;
-      } else {
-        signature_ = null;
-        signatureBuilder_ = null;
-      }
-      internalGetMutableAttr().clear();
-      internalGetMutableArgAttr().clear();
-      internalGetMutableResourceArgUniqueId().clear();
-      if (nodeDefBuilder_ == null) {
-        nodeDef_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000008);
-      } else {
-        nodeDefBuilder_.clear();
-      }
-      internalGetMutableRet().clear();
-      internalGetMutableControlRet().clear();
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.framework.FunctionProtos.internal_static_tensorflow_FunctionDef_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.FunctionDef getDefaultInstanceForType() {
-      return org.tensorflow.proto.framework.FunctionDef.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.FunctionDef build() {
-      org.tensorflow.proto.framework.FunctionDef result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.FunctionDef buildPartial() {
-      org.tensorflow.proto.framework.FunctionDef result = new org.tensorflow.proto.framework.FunctionDef(this);
-      int from_bitField0_ = bitField0_;
-      if (signatureBuilder_ == null) {
-        result.signature_ = signature_;
-      } else {
-        result.signature_ = signatureBuilder_.build();
-      }
-      result.attr_ = internalGetAttr();
-      result.attr_.makeImmutable();
-      result.argAttr_ = internalGetArgAttr();
-      result.argAttr_.makeImmutable();
-      result.resourceArgUniqueId_ = internalGetResourceArgUniqueId();
-      result.resourceArgUniqueId_.makeImmutable();
-      if (nodeDefBuilder_ == null) {
-        if (((bitField0_ & 0x00000008) != 0)) {
-          nodeDef_ = java.util.Collections.unmodifiableList(nodeDef_);
-          bitField0_ = (bitField0_ & ~0x00000008);
-        }
-        result.nodeDef_ = nodeDef_;
-      } else {
-        result.nodeDef_ = nodeDefBuilder_.build();
-      }
-      result.ret_ = internalGetRet();
-      result.ret_.makeImmutable();
-      result.controlRet_ = internalGetControlRet();
-      result.controlRet_.makeImmutable();
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.framework.FunctionDef) {
-        return mergeFrom((org.tensorflow.proto.framework.FunctionDef)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.framework.FunctionDef other) {
-      if (other == org.tensorflow.proto.framework.FunctionDef.getDefaultInstance()) return this;
-      if (other.hasSignature()) {
-        mergeSignature(other.getSignature());
-      }
-      internalGetMutableAttr().mergeFrom(
-          other.internalGetAttr());
-      internalGetMutableArgAttr().mergeFrom(
-          other.internalGetArgAttr());
-      internalGetMutableResourceArgUniqueId().mergeFrom(
-          other.internalGetResourceArgUniqueId());
-      if (nodeDefBuilder_ == null) {
-        if (!other.nodeDef_.isEmpty()) {
-          if (nodeDef_.isEmpty()) {
-            nodeDef_ = other.nodeDef_;
-            bitField0_ = (bitField0_ & ~0x00000008);
-          } else {
-            ensureNodeDefIsMutable();
-            nodeDef_.addAll(other.nodeDef_);
-          }
-          onChanged();
-        }
-      } else {
-        if (!other.nodeDef_.isEmpty()) {
-          if (nodeDefBuilder_.isEmpty()) {
-            nodeDefBuilder_.dispose();
-            nodeDefBuilder_ = null;
-            nodeDef_ = other.nodeDef_;
-            bitField0_ = (bitField0_ & ~0x00000008);
-            nodeDefBuilder_ = 
-              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
-                 getNodeDefFieldBuilder() : null;
-          } else {
-            nodeDefBuilder_.addAllMessages(other.nodeDef_);
-          }
-        }
-      }
-      internalGetMutableRet().mergeFrom(
-          other.internalGetRet());
-      internalGetMutableControlRet().mergeFrom(
-          other.internalGetControlRet());
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.framework.FunctionDef parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.framework.FunctionDef) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-    private int bitField0_;
-
-    private org.tensorflow.proto.framework.OpDef signature_;
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.OpDef, org.tensorflow.proto.framework.OpDef.Builder, org.tensorflow.proto.framework.OpDefOrBuilder> signatureBuilder_;
-    /**
-     * <pre>
-     * The definition of the function's name, arguments, return values,
-     * attrs etc.
-     * </pre>
-     *
-     * <code>.tensorflow.OpDef signature = 1;</code>
-     */
-    public boolean hasSignature() {
-      return signatureBuilder_ != null || signature_ != null;
-    }
-    /**
-     * <pre>
-     * The definition of the function's name, arguments, return values,
-     * attrs etc.
-     * </pre>
-     *
-     * <code>.tensorflow.OpDef signature = 1;</code>
-     */
-    public org.tensorflow.proto.framework.OpDef getSignature() {
-      if (signatureBuilder_ == null) {
-        return signature_ == null ? org.tensorflow.proto.framework.OpDef.getDefaultInstance() : signature_;
-      } else {
-        return signatureBuilder_.getMessage();
-      }
-    }
-    /**
-     * <pre>
-     * The definition of the function's name, arguments, return values,
-     * attrs etc.
-     * </pre>
-     *
-     * <code>.tensorflow.OpDef signature = 1;</code>
-     */
-    public Builder setSignature(org.tensorflow.proto.framework.OpDef value) {
-      if (signatureBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        signature_ = value;
-        onChanged();
-      } else {
-        signatureBuilder_.setMessage(value);
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * The definition of the function's name, arguments, return values,
-     * attrs etc.
-     * </pre>
-     *
-     * <code>.tensorflow.OpDef signature = 1;</code>
-     */
-    public Builder setSignature(
-        org.tensorflow.proto.framework.OpDef.Builder builderForValue) {
-      if (signatureBuilder_ == null) {
-        signature_ = builderForValue.build();
-        onChanged();
-      } else {
-        signatureBuilder_.setMessage(builderForValue.build());
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * The definition of the function's name, arguments, return values,
-     * attrs etc.
-     * </pre>
-     *
-     * <code>.tensorflow.OpDef signature = 1;</code>
-     */
-    public Builder mergeSignature(org.tensorflow.proto.framework.OpDef value) {
-      if (signatureBuilder_ == null) {
-        if (signature_ != null) {
-          signature_ =
-            org.tensorflow.proto.framework.OpDef.newBuilder(signature_).mergeFrom(value).buildPartial();
-        } else {
-          signature_ = value;
-        }
-        onChanged();
-      } else {
-        signatureBuilder_.mergeFrom(value);
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * The definition of the function's name, arguments, return values,
-     * attrs etc.
-     * </pre>
-     *
-     * <code>.tensorflow.OpDef signature = 1;</code>
-     */
-    public Builder clearSignature() {
-      if (signatureBuilder_ == null) {
-        signature_ = null;
-        onChanged();
-      } else {
-        signature_ = null;
-        signatureBuilder_ = null;
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * The definition of the function's name, arguments, return values,
-     * attrs etc.
-     * </pre>
-     *
-     * <code>.tensorflow.OpDef signature = 1;</code>
-     */
-    public org.tensorflow.proto.framework.OpDef.Builder getSignatureBuilder() {
-      
-      onChanged();
-      return getSignatureFieldBuilder().getBuilder();
-    }
-    /**
-     * <pre>
-     * The definition of the function's name, arguments, return values,
-     * attrs etc.
-     * </pre>
-     *
-     * <code>.tensorflow.OpDef signature = 1;</code>
-     */
-    public org.tensorflow.proto.framework.OpDefOrBuilder getSignatureOrBuilder() {
-      if (signatureBuilder_ != null) {
-        return signatureBuilder_.getMessageOrBuilder();
-      } else {
-        return signature_ == null ?
-            org.tensorflow.proto.framework.OpDef.getDefaultInstance() : signature_;
-      }
-    }
-    /**
-     * <pre>
-     * The definition of the function's name, arguments, return values,
-     * attrs etc.
-     * </pre>
-     *
-     * <code>.tensorflow.OpDef signature = 1;</code>
-     */
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.OpDef, org.tensorflow.proto.framework.OpDef.Builder, org.tensorflow.proto.framework.OpDefOrBuilder> 
-        getSignatureFieldBuilder() {
-      if (signatureBuilder_ == null) {
-        signatureBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-            org.tensorflow.proto.framework.OpDef, org.tensorflow.proto.framework.OpDef.Builder, org.tensorflow.proto.framework.OpDefOrBuilder>(
-                getSignature(),
-                getParentForChildren(),
-                isClean());
-        signature_ = null;
-      }
-      return signatureBuilder_;
-    }
-
-    private com.google.protobuf.MapField<
-        java.lang.String, org.tensorflow.proto.framework.AttrValue> attr_;
-    private com.google.protobuf.MapField<java.lang.String, org.tensorflow.proto.framework.AttrValue>
-    internalGetAttr() {
-      if (attr_ == null) {
-        return com.google.protobuf.MapField.emptyMapField(
-            AttrDefaultEntryHolder.defaultEntry);
-      }
-      return attr_;
-    }
-    private com.google.protobuf.MapField<java.lang.String, org.tensorflow.proto.framework.AttrValue>
-    internalGetMutableAttr() {
-      onChanged();;
-      if (attr_ == null) {
-        attr_ = com.google.protobuf.MapField.newMapField(
-            AttrDefaultEntryHolder.defaultEntry);
-      }
-      if (!attr_.isMutable()) {
-        attr_ = attr_.copy();
-      }
-      return attr_;
-    }
-
-    public int getAttrCount() {
-      return internalGetAttr().getMap().size();
-    }
-    /**
-     * <pre>
-     * Attributes specific to this function definition.
-     * </pre>
-     *
-     * <code>map&lt;string, .tensorflow.AttrValue&gt; attr = 5;</code>
-     */
-
-    public boolean containsAttr(
-        java.lang.String key) {
-      if (key == null) { throw new java.lang.NullPointerException(); }
-      return internalGetAttr().getMap().containsKey(key);
-    }
-    /**
-     * Use {@link #getAttrMap()} instead.
-     */
-    @java.lang.Deprecated
-    public java.util.Map<java.lang.String, org.tensorflow.proto.framework.AttrValue> getAttr() {
-      return getAttrMap();
-    }
-    /**
-     * <pre>
-     * Attributes specific to this function definition.
-     * </pre>
-     *
-     * <code>map&lt;string, .tensorflow.AttrValue&gt; attr = 5;</code>
-     */
-
-    public java.util.Map<java.lang.String, org.tensorflow.proto.framework.AttrValue> getAttrMap() {
-      return internalGetAttr().getMap();
-    }
-    /**
-     * <pre>
-     * Attributes specific to this function definition.
-     * </pre>
-     *
-     * <code>map&lt;string, .tensorflow.AttrValue&gt; attr = 5;</code>
-     */
-
-    public org.tensorflow.proto.framework.AttrValue getAttrOrDefault(
-        java.lang.String key,
-        org.tensorflow.proto.framework.AttrValue defaultValue) {
-      if (key == null) { throw new java.lang.NullPointerException(); }
-      java.util.Map<java.lang.String, org.tensorflow.proto.framework.AttrValue> map =
-          internalGetAttr().getMap();
-      return map.containsKey(key) ? map.get(key) : defaultValue;
-    }
-    /**
-     * <pre>
-     * Attributes specific to this function definition.
-     * </pre>
-     *
-     * <code>map&lt;string, .tensorflow.AttrValue&gt; attr = 5;</code>
-     */
-
-    public org.tensorflow.proto.framework.AttrValue getAttrOrThrow(
-        java.lang.String key) {
-      if (key == null) { throw new java.lang.NullPointerException(); }
-      java.util.Map<java.lang.String, org.tensorflow.proto.framework.AttrValue> map =
-          internalGetAttr().getMap();
-      if (!map.containsKey(key)) {
-        throw new java.lang.IllegalArgumentException();
-      }
-      return map.get(key);
-    }
-
-    public Builder clearAttr() {
-      internalGetMutableAttr().getMutableMap()
-          .clear();
-      return this;
-    }
-    /**
-     * <pre>
-     * Attributes specific to this function definition.
-     * </pre>
-     *
-     * <code>map&lt;string, .tensorflow.AttrValue&gt; attr = 5;</code>
-     */
-
-    public Builder removeAttr(
-        java.lang.String key) {
-      if (key == null) { throw new java.lang.NullPointerException(); }
-      internalGetMutableAttr().getMutableMap()
-          .remove(key);
-      return this;
-    }
-    /**
-     * Use alternate mutation accessors instead.
-     */
-    @java.lang.Deprecated
-    public java.util.Map<java.lang.String, org.tensorflow.proto.framework.AttrValue>
-    getMutableAttr() {
-      return internalGetMutableAttr().getMutableMap();
-    }
-    /**
-     * <pre>
-     * Attributes specific to this function definition.
-     * </pre>
-     *
-     * <code>map&lt;string, .tensorflow.AttrValue&gt; attr = 5;</code>
-     */
-    public Builder putAttr(
-        java.lang.String key,
-        org.tensorflow.proto.framework.AttrValue value) {
-      if (key == null) { throw new java.lang.NullPointerException(); }
-      if (value == null) { throw new java.lang.NullPointerException(); }
-      internalGetMutableAttr().getMutableMap()
-          .put(key, value);
-      return this;
-    }
-    /**
-     * <pre>
-     * Attributes specific to this function definition.
-     * </pre>
-     *
-     * <code>map&lt;string, .tensorflow.AttrValue&gt; attr = 5;</code>
-     */
-
-    public Builder putAllAttr(
-        java.util.Map<java.lang.String, org.tensorflow.proto.framework.AttrValue> values) {
-      internalGetMutableAttr().getMutableMap()
-          .putAll(values);
-      return this;
-    }
-
-    private com.google.protobuf.MapField<
-        java.lang.Integer, org.tensorflow.proto.framework.FunctionDef.ArgAttrs> argAttr_;
-    private com.google.protobuf.MapField<java.lang.Integer, org.tensorflow.proto.framework.FunctionDef.ArgAttrs>
-    internalGetArgAttr() {
-      if (argAttr_ == null) {
-        return com.google.protobuf.MapField.emptyMapField(
-            ArgAttrDefaultEntryHolder.defaultEntry);
-      }
-      return argAttr_;
-    }
-    private com.google.protobuf.MapField<java.lang.Integer, org.tensorflow.proto.framework.FunctionDef.ArgAttrs>
-    internalGetMutableArgAttr() {
-      onChanged();;
-      if (argAttr_ == null) {
-        argAttr_ = com.google.protobuf.MapField.newMapField(
-            ArgAttrDefaultEntryHolder.defaultEntry);
-      }
-      if (!argAttr_.isMutable()) {
-        argAttr_ = argAttr_.copy();
-      }
-      return argAttr_;
-    }
-
-    public int getArgAttrCount() {
-      return internalGetArgAttr().getMap().size();
-    }
-    /**
-     * <code>map&lt;uint32, .tensorflow.FunctionDef.ArgAttrs&gt; arg_attr = 7;</code>
-     */
-
-    public boolean containsArgAttr(
-        int key) {
-      
-      return internalGetArgAttr().getMap().containsKey(key);
-    }
-    /**
-     * Use {@link #getArgAttrMap()} instead.
-     */
-    @java.lang.Deprecated
-    public java.util.Map<java.lang.Integer, org.tensorflow.proto.framework.FunctionDef.ArgAttrs> getArgAttr() {
-      return getArgAttrMap();
-    }
-    /**
-     * <code>map&lt;uint32, .tensorflow.FunctionDef.ArgAttrs&gt; arg_attr = 7;</code>
-     */
-
-    public java.util.Map<java.lang.Integer, org.tensorflow.proto.framework.FunctionDef.ArgAttrs> getArgAttrMap() {
-      return internalGetArgAttr().getMap();
-    }
-    /**
-     * <code>map&lt;uint32, .tensorflow.FunctionDef.ArgAttrs&gt; arg_attr = 7;</code>
-     */
-
-    public org.tensorflow.proto.framework.FunctionDef.ArgAttrs getArgAttrOrDefault(
-        int key,
-        org.tensorflow.proto.framework.FunctionDef.ArgAttrs defaultValue) {
-      
-      java.util.Map<java.lang.Integer, org.tensorflow.proto.framework.FunctionDef.ArgAttrs> map =
-          internalGetArgAttr().getMap();
-      return map.containsKey(key) ? map.get(key) : defaultValue;
-    }
-    /**
-     * <code>map&lt;uint32, .tensorflow.FunctionDef.ArgAttrs&gt; arg_attr = 7;</code>
-     */
-
-    public org.tensorflow.proto.framework.FunctionDef.ArgAttrs getArgAttrOrThrow(
-        int key) {
-      
-      java.util.Map<java.lang.Integer, org.tensorflow.proto.framework.FunctionDef.ArgAttrs> map =
-          internalGetArgAttr().getMap();
-      if (!map.containsKey(key)) {
-        throw new java.lang.IllegalArgumentException();
-      }
-      return map.get(key);
-    }
-
-    public Builder clearArgAttr() {
-      internalGetMutableArgAttr().getMutableMap()
-          .clear();
-      return this;
-    }
-    /**
-     * <code>map&lt;uint32, .tensorflow.FunctionDef.ArgAttrs&gt; arg_attr = 7;</code>
-     */
-
-    public Builder removeArgAttr(
-        int key) {
-      
-      internalGetMutableArgAttr().getMutableMap()
-          .remove(key);
-      return this;
-    }
-    /**
-     * Use alternate mutation accessors instead.
-     */
-    @java.lang.Deprecated
-    public java.util.Map<java.lang.Integer, org.tensorflow.proto.framework.FunctionDef.ArgAttrs>
-    getMutableArgAttr() {
-      return internalGetMutableArgAttr().getMutableMap();
-    }
-    /**
-     * <code>map&lt;uint32, .tensorflow.FunctionDef.ArgAttrs&gt; arg_attr = 7;</code>
-     */
-    public Builder putArgAttr(
-        int key,
-        org.tensorflow.proto.framework.FunctionDef.ArgAttrs value) {
-      
-      if (value == null) { throw new java.lang.NullPointerException(); }
-      internalGetMutableArgAttr().getMutableMap()
-          .put(key, value);
-      return this;
-    }
-    /**
-     * <code>map&lt;uint32, .tensorflow.FunctionDef.ArgAttrs&gt; arg_attr = 7;</code>
-     */
-
-    public Builder putAllArgAttr(
-        java.util.Map<java.lang.Integer, org.tensorflow.proto.framework.FunctionDef.ArgAttrs> values) {
-      internalGetMutableArgAttr().getMutableMap()
-          .putAll(values);
-      return this;
-    }
-
-    private com.google.protobuf.MapField<
-        java.lang.Integer, java.lang.Integer> resourceArgUniqueId_;
-    private com.google.protobuf.MapField<java.lang.Integer, java.lang.Integer>
-    internalGetResourceArgUniqueId() {
-      if (resourceArgUniqueId_ == null) {
-        return com.google.protobuf.MapField.emptyMapField(
-            ResourceArgUniqueIdDefaultEntryHolder.defaultEntry);
-      }
-      return resourceArgUniqueId_;
-    }
-    private com.google.protobuf.MapField<java.lang.Integer, java.lang.Integer>
-    internalGetMutableResourceArgUniqueId() {
-      onChanged();;
-      if (resourceArgUniqueId_ == null) {
-        resourceArgUniqueId_ = com.google.protobuf.MapField.newMapField(
-            ResourceArgUniqueIdDefaultEntryHolder.defaultEntry);
-      }
-      if (!resourceArgUniqueId_.isMutable()) {
-        resourceArgUniqueId_ = resourceArgUniqueId_.copy();
-      }
-      return resourceArgUniqueId_;
-    }
-
-    public int getResourceArgUniqueIdCount() {
-      return internalGetResourceArgUniqueId().getMap().size();
-    }
-    /**
-     * <pre>
-     * Unique IDs for each resource argument, used to track aliasing resources. If
-     * Argument A and Argument B alias each other, then
-     * resource_arg_unique_ids[A.index] == resource_arg_unique_ids[B.index].
-     * If this field is empty, none of the arguments could alias; otherwise, every
-     * resource argument should have an entry in this field.
-     * When instantiated, the unique IDs will be attached to the _Arg nodes'
-     * "_resource_arg_unique_id" attribute.
-     * </pre>
-     *
-     * <code>map&lt;uint32, uint32&gt; resource_arg_unique_id = 8;</code>
-     */
-
-    public boolean containsResourceArgUniqueId(
-        int key) {
-      
-      return internalGetResourceArgUniqueId().getMap().containsKey(key);
-    }
-    /**
-     * Use {@link #getResourceArgUniqueIdMap()} instead.
-     */
-    @java.lang.Deprecated
-    public java.util.Map<java.lang.Integer, java.lang.Integer> getResourceArgUniqueId() {
-      return getResourceArgUniqueIdMap();
-    }
-    /**
-     * <pre>
-     * Unique IDs for each resource argument, used to track aliasing resources. If
-     * Argument A and Argument B alias each other, then
-     * resource_arg_unique_ids[A.index] == resource_arg_unique_ids[B.index].
-     * If this field is empty, none of the arguments could alias; otherwise, every
-     * resource argument should have an entry in this field.
-     * When instantiated, the unique IDs will be attached to the _Arg nodes'
-     * "_resource_arg_unique_id" attribute.
-     * </pre>
-     *
-     * <code>map&lt;uint32, uint32&gt; resource_arg_unique_id = 8;</code>
-     */
-
-    public java.util.Map<java.lang.Integer, java.lang.Integer> getResourceArgUniqueIdMap() {
-      return internalGetResourceArgUniqueId().getMap();
-    }
-    /**
-     * <pre>
-     * Unique IDs for each resource argument, used to track aliasing resources. If
-     * Argument A and Argument B alias each other, then
-     * resource_arg_unique_ids[A.index] == resource_arg_unique_ids[B.index].
-     * If this field is empty, none of the arguments could alias; otherwise, every
-     * resource argument should have an entry in this field.
-     * When instantiated, the unique IDs will be attached to the _Arg nodes'
-     * "_resource_arg_unique_id" attribute.
-     * </pre>
-     *
-     * <code>map&lt;uint32, uint32&gt; resource_arg_unique_id = 8;</code>
-     */
-
-    public int getResourceArgUniqueIdOrDefault(
-        int key,
-        int defaultValue) {
-      
-      java.util.Map<java.lang.Integer, java.lang.Integer> map =
-          internalGetResourceArgUniqueId().getMap();
-      return map.containsKey(key) ? map.get(key) : defaultValue;
-    }
-    /**
-     * <pre>
-     * Unique IDs for each resource argument, used to track aliasing resources. If
-     * Argument A and Argument B alias each other, then
-     * resource_arg_unique_ids[A.index] == resource_arg_unique_ids[B.index].
-     * If this field is empty, none of the arguments could alias; otherwise, every
-     * resource argument should have an entry in this field.
-     * When instantiated, the unique IDs will be attached to the _Arg nodes'
-     * "_resource_arg_unique_id" attribute.
-     * </pre>
-     *
-     * <code>map&lt;uint32, uint32&gt; resource_arg_unique_id = 8;</code>
-     */
-
-    public int getResourceArgUniqueIdOrThrow(
-        int key) {
-      
-      java.util.Map<java.lang.Integer, java.lang.Integer> map =
-          internalGetResourceArgUniqueId().getMap();
-      if (!map.containsKey(key)) {
-        throw new java.lang.IllegalArgumentException();
-      }
-      return map.get(key);
-    }
-
-    public Builder clearResourceArgUniqueId() {
-      internalGetMutableResourceArgUniqueId().getMutableMap()
-          .clear();
-      return this;
-    }
-    /**
-     * <pre>
-     * Unique IDs for each resource argument, used to track aliasing resources. If
-     * Argument A and Argument B alias each other, then
-     * resource_arg_unique_ids[A.index] == resource_arg_unique_ids[B.index].
-     * If this field is empty, none of the arguments could alias; otherwise, every
-     * resource argument should have an entry in this field.
-     * When instantiated, the unique IDs will be attached to the _Arg nodes'
-     * "_resource_arg_unique_id" attribute.
-     * </pre>
-     *
-     * <code>map&lt;uint32, uint32&gt; resource_arg_unique_id = 8;</code>
-     */
-
-    public Builder removeResourceArgUniqueId(
-        int key) {
-      
-      internalGetMutableResourceArgUniqueId().getMutableMap()
-          .remove(key);
-      return this;
-    }
-    /**
-     * Use alternate mutation accessors instead.
-     */
-    @java.lang.Deprecated
-    public java.util.Map<java.lang.Integer, java.lang.Integer>
-    getMutableResourceArgUniqueId() {
-      return internalGetMutableResourceArgUniqueId().getMutableMap();
-    }
-    /**
-     * <pre>
-     * Unique IDs for each resource argument, used to track aliasing resources. If
-     * Argument A and Argument B alias each other, then
-     * resource_arg_unique_ids[A.index] == resource_arg_unique_ids[B.index].
-     * If this field is empty, none of the arguments could alias; otherwise, every
-     * resource argument should have an entry in this field.
-     * When instantiated, the unique IDs will be attached to the _Arg nodes'
-     * "_resource_arg_unique_id" attribute.
-     * </pre>
-     *
-     * <code>map&lt;uint32, uint32&gt; resource_arg_unique_id = 8;</code>
-     */
-    public Builder putResourceArgUniqueId(
-        int key,
-        int value) {
-      
-      
-      internalGetMutableResourceArgUniqueId().getMutableMap()
-          .put(key, value);
-      return this;
-    }
-    /**
-     * <pre>
-     * Unique IDs for each resource argument, used to track aliasing resources. If
-     * Argument A and Argument B alias each other, then
-     * resource_arg_unique_ids[A.index] == resource_arg_unique_ids[B.index].
-     * If this field is empty, none of the arguments could alias; otherwise, every
-     * resource argument should have an entry in this field.
-     * When instantiated, the unique IDs will be attached to the _Arg nodes'
-     * "_resource_arg_unique_id" attribute.
-     * </pre>
-     *
-     * <code>map&lt;uint32, uint32&gt; resource_arg_unique_id = 8;</code>
-     */
-
-    public Builder putAllResourceArgUniqueId(
-        java.util.Map<java.lang.Integer, java.lang.Integer> values) {
-      internalGetMutableResourceArgUniqueId().getMutableMap()
-          .putAll(values);
-      return this;
-    }
-
-    private java.util.List<org.tensorflow.proto.framework.NodeDef> nodeDef_ =
-      java.util.Collections.emptyList();
-    private void ensureNodeDefIsMutable() {
-      if (!((bitField0_ & 0x00000008) != 0)) {
-        nodeDef_ = new java.util.ArrayList<org.tensorflow.proto.framework.NodeDef>(nodeDef_);
-        bitField0_ |= 0x00000008;
-       }
-    }
-
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.framework.NodeDef, org.tensorflow.proto.framework.NodeDef.Builder, org.tensorflow.proto.framework.NodeDefOrBuilder> nodeDefBuilder_;
-
-    /**
-     * <pre>
-     * By convention, "op" in node_def is resolved by consulting with a
-     * user-defined library first. If not resolved, "func" is assumed to
-     * be a builtin op.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.NodeDef node_def = 3;</code>
-     */
-    public java.util.List<org.tensorflow.proto.framework.NodeDef> getNodeDefList() {
-      if (nodeDefBuilder_ == null) {
-        return java.util.Collections.unmodifiableList(nodeDef_);
-      } else {
-        return nodeDefBuilder_.getMessageList();
-      }
-    }
-    /**
-     * <pre>
-     * By convention, "op" in node_def is resolved by consulting with a
-     * user-defined library first. If not resolved, "func" is assumed to
-     * be a builtin op.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.NodeDef node_def = 3;</code>
-     */
-    public int getNodeDefCount() {
-      if (nodeDefBuilder_ == null) {
-        return nodeDef_.size();
-      } else {
-        return nodeDefBuilder_.getCount();
-      }
-    }
-    /**
-     * <pre>
-     * By convention, "op" in node_def is resolved by consulting with a
-     * user-defined library first. If not resolved, "func" is assumed to
-     * be a builtin op.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.NodeDef node_def = 3;</code>
-     */
-    public org.tensorflow.proto.framework.NodeDef getNodeDef(int index) {
-      if (nodeDefBuilder_ == null) {
-        return nodeDef_.get(index);
-      } else {
-        return nodeDefBuilder_.getMessage(index);
-      }
-    }
-    /**
-     * <pre>
-     * By convention, "op" in node_def is resolved by consulting with a
-     * user-defined library first. If not resolved, "func" is assumed to
-     * be a builtin op.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.NodeDef node_def = 3;</code>
-     */
-    public Builder setNodeDef(
-        int index, org.tensorflow.proto.framework.NodeDef value) {
-      if (nodeDefBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureNodeDefIsMutable();
-        nodeDef_.set(index, value);
-        onChanged();
-      } else {
-        nodeDefBuilder_.setMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * By convention, "op" in node_def is resolved by consulting with a
-     * user-defined library first. If not resolved, "func" is assumed to
-     * be a builtin op.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.NodeDef node_def = 3;</code>
-     */
-    public Builder setNodeDef(
-        int index, org.tensorflow.proto.framework.NodeDef.Builder builderForValue) {
-      if (nodeDefBuilder_ == null) {
-        ensureNodeDefIsMutable();
-        nodeDef_.set(index, builderForValue.build());
-        onChanged();
-      } else {
-        nodeDefBuilder_.setMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * By convention, "op" in node_def is resolved by consulting with a
-     * user-defined library first. If not resolved, "func" is assumed to
-     * be a builtin op.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.NodeDef node_def = 3;</code>
-     */
-    public Builder addNodeDef(org.tensorflow.proto.framework.NodeDef value) {
-      if (nodeDefBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureNodeDefIsMutable();
-        nodeDef_.add(value);
-        onChanged();
-      } else {
-        nodeDefBuilder_.addMessage(value);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * By convention, "op" in node_def is resolved by consulting with a
-     * user-defined library first. If not resolved, "func" is assumed to
-     * be a builtin op.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.NodeDef node_def = 3;</code>
-     */
-    public Builder addNodeDef(
-        int index, org.tensorflow.proto.framework.NodeDef value) {
-      if (nodeDefBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureNodeDefIsMutable();
-        nodeDef_.add(index, value);
-        onChanged();
-      } else {
-        nodeDefBuilder_.addMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * By convention, "op" in node_def is resolved by consulting with a
-     * user-defined library first. If not resolved, "func" is assumed to
-     * be a builtin op.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.NodeDef node_def = 3;</code>
-     */
-    public Builder addNodeDef(
-        org.tensorflow.proto.framework.NodeDef.Builder builderForValue) {
-      if (nodeDefBuilder_ == null) {
-        ensureNodeDefIsMutable();
-        nodeDef_.add(builderForValue.build());
-        onChanged();
-      } else {
-        nodeDefBuilder_.addMessage(builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * By convention, "op" in node_def is resolved by consulting with a
-     * user-defined library first. If not resolved, "func" is assumed to
-     * be a builtin op.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.NodeDef node_def = 3;</code>
-     */
-    public Builder addNodeDef(
-        int index, org.tensorflow.proto.framework.NodeDef.Builder builderForValue) {
-      if (nodeDefBuilder_ == null) {
-        ensureNodeDefIsMutable();
-        nodeDef_.add(index, builderForValue.build());
-        onChanged();
-      } else {
-        nodeDefBuilder_.addMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * By convention, "op" in node_def is resolved by consulting with a
-     * user-defined library first. If not resolved, "func" is assumed to
-     * be a builtin op.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.NodeDef node_def = 3;</code>
-     */
-    public Builder addAllNodeDef(
-        java.lang.Iterable<? extends org.tensorflow.proto.framework.NodeDef> values) {
-      if (nodeDefBuilder_ == null) {
-        ensureNodeDefIsMutable();
-        com.google.protobuf.AbstractMessageLite.Builder.addAll(
-            values, nodeDef_);
-        onChanged();
-      } else {
-        nodeDefBuilder_.addAllMessages(values);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * By convention, "op" in node_def is resolved by consulting with a
-     * user-defined library first. If not resolved, "func" is assumed to
-     * be a builtin op.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.NodeDef node_def = 3;</code>
-     */
-    public Builder clearNodeDef() {
-      if (nodeDefBuilder_ == null) {
-        nodeDef_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000008);
-        onChanged();
-      } else {
-        nodeDefBuilder_.clear();
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * By convention, "op" in node_def is resolved by consulting with a
-     * user-defined library first. If not resolved, "func" is assumed to
-     * be a builtin op.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.NodeDef node_def = 3;</code>
-     */
-    public Builder removeNodeDef(int index) {
-      if (nodeDefBuilder_ == null) {
-        ensureNodeDefIsMutable();
-        nodeDef_.remove(index);
-        onChanged();
-      } else {
-        nodeDefBuilder_.remove(index);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * By convention, "op" in node_def is resolved by consulting with a
-     * user-defined library first. If not resolved, "func" is assumed to
-     * be a builtin op.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.NodeDef node_def = 3;</code>
-     */
-    public org.tensorflow.proto.framework.NodeDef.Builder getNodeDefBuilder(
-        int index) {
-      return getNodeDefFieldBuilder().getBuilder(index);
-    }
-    /**
-     * <pre>
-     * By convention, "op" in node_def is resolved by consulting with a
-     * user-defined library first. If not resolved, "func" is assumed to
-     * be a builtin op.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.NodeDef node_def = 3;</code>
-     */
-    public org.tensorflow.proto.framework.NodeDefOrBuilder getNodeDefOrBuilder(
-        int index) {
-      if (nodeDefBuilder_ == null) {
-        return nodeDef_.get(index);  } else {
-        return nodeDefBuilder_.getMessageOrBuilder(index);
-      }
-    }
-    /**
-     * <pre>
-     * By convention, "op" in node_def is resolved by consulting with a
-     * user-defined library first. If not resolved, "func" is assumed to
-     * be a builtin op.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.NodeDef node_def = 3;</code>
-     */
-    public java.util.List<? extends org.tensorflow.proto.framework.NodeDefOrBuilder> 
-         getNodeDefOrBuilderList() {
-      if (nodeDefBuilder_ != null) {
-        return nodeDefBuilder_.getMessageOrBuilderList();
-      } else {
-        return java.util.Collections.unmodifiableList(nodeDef_);
-      }
-    }
-    /**
-     * <pre>
-     * By convention, "op" in node_def is resolved by consulting with a
-     * user-defined library first. If not resolved, "func" is assumed to
-     * be a builtin op.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.NodeDef node_def = 3;</code>
-     */
-    public org.tensorflow.proto.framework.NodeDef.Builder addNodeDefBuilder() {
-      return getNodeDefFieldBuilder().addBuilder(
-          org.tensorflow.proto.framework.NodeDef.getDefaultInstance());
-    }
-    /**
-     * <pre>
-     * By convention, "op" in node_def is resolved by consulting with a
-     * user-defined library first. If not resolved, "func" is assumed to
-     * be a builtin op.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.NodeDef node_def = 3;</code>
-     */
-    public org.tensorflow.proto.framework.NodeDef.Builder addNodeDefBuilder(
-        int index) {
-      return getNodeDefFieldBuilder().addBuilder(
-          index, org.tensorflow.proto.framework.NodeDef.getDefaultInstance());
-    }
-    /**
-     * <pre>
-     * By convention, "op" in node_def is resolved by consulting with a
-     * user-defined library first. If not resolved, "func" is assumed to
-     * be a builtin op.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.NodeDef node_def = 3;</code>
-     */
-    public java.util.List<org.tensorflow.proto.framework.NodeDef.Builder> 
-         getNodeDefBuilderList() {
-      return getNodeDefFieldBuilder().getBuilderList();
-    }
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.framework.NodeDef, org.tensorflow.proto.framework.NodeDef.Builder, org.tensorflow.proto.framework.NodeDefOrBuilder> 
-        getNodeDefFieldBuilder() {
-      if (nodeDefBuilder_ == null) {
-        nodeDefBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
-            org.tensorflow.proto.framework.NodeDef, org.tensorflow.proto.framework.NodeDef.Builder, org.tensorflow.proto.framework.NodeDefOrBuilder>(
-                nodeDef_,
-                ((bitField0_ & 0x00000008) != 0),
-                getParentForChildren(),
-                isClean());
-        nodeDef_ = null;
-      }
-      return nodeDefBuilder_;
-    }
-
-    private com.google.protobuf.MapField<
-        java.lang.String, java.lang.String> ret_;
-    private com.google.protobuf.MapField<java.lang.String, java.lang.String>
-    internalGetRet() {
-      if (ret_ == null) {
-        return com.google.protobuf.MapField.emptyMapField(
-            RetDefaultEntryHolder.defaultEntry);
-      }
-      return ret_;
-    }
-    private com.google.protobuf.MapField<java.lang.String, java.lang.String>
-    internalGetMutableRet() {
-      onChanged();;
-      if (ret_ == null) {
-        ret_ = com.google.protobuf.MapField.newMapField(
-            RetDefaultEntryHolder.defaultEntry);
-      }
-      if (!ret_.isMutable()) {
-        ret_ = ret_.copy();
-      }
-      return ret_;
-    }
-
-    public int getRetCount() {
-      return internalGetRet().getMap().size();
-    }
-    /**
-     * <pre>
-     * A mapping from the output arg names from `signature` to the
-     * outputs from `node_def` that should be returned by the function.
-     * </pre>
-     *
-     * <code>map&lt;string, string&gt; ret = 4;</code>
-     */
-
-    public boolean containsRet(
-        java.lang.String key) {
-      if (key == null) { throw new java.lang.NullPointerException(); }
-      return internalGetRet().getMap().containsKey(key);
-    }
-    /**
-     * Use {@link #getRetMap()} instead.
-     */
-    @java.lang.Deprecated
-    public java.util.Map<java.lang.String, java.lang.String> getRet() {
-      return getRetMap();
-    }
-    /**
-     * <pre>
-     * A mapping from the output arg names from `signature` to the
-     * outputs from `node_def` that should be returned by the function.
-     * </pre>
-     *
-     * <code>map&lt;string, string&gt; ret = 4;</code>
-     */
-
-    public java.util.Map<java.lang.String, java.lang.String> getRetMap() {
-      return internalGetRet().getMap();
-    }
-    /**
-     * <pre>
-     * A mapping from the output arg names from `signature` to the
-     * outputs from `node_def` that should be returned by the function.
-     * </pre>
-     *
-     * <code>map&lt;string, string&gt; ret = 4;</code>
-     */
-
-    public java.lang.String getRetOrDefault(
-        java.lang.String key,
-        java.lang.String defaultValue) {
-      if (key == null) { throw new java.lang.NullPointerException(); }
-      java.util.Map<java.lang.String, java.lang.String> map =
-          internalGetRet().getMap();
-      return map.containsKey(key) ? map.get(key) : defaultValue;
-    }
-    /**
-     * <pre>
-     * A mapping from the output arg names from `signature` to the
-     * outputs from `node_def` that should be returned by the function.
-     * </pre>
-     *
-     * <code>map&lt;string, string&gt; ret = 4;</code>
-     */
-
-    public java.lang.String getRetOrThrow(
-        java.lang.String key) {
-      if (key == null) { throw new java.lang.NullPointerException(); }
-      java.util.Map<java.lang.String, java.lang.String> map =
-          internalGetRet().getMap();
-      if (!map.containsKey(key)) {
-        throw new java.lang.IllegalArgumentException();
-      }
-      return map.get(key);
-    }
-
-    public Builder clearRet() {
-      internalGetMutableRet().getMutableMap()
-          .clear();
-      return this;
-    }
-    /**
-     * <pre>
-     * A mapping from the output arg names from `signature` to the
-     * outputs from `node_def` that should be returned by the function.
-     * </pre>
-     *
-     * <code>map&lt;string, string&gt; ret = 4;</code>
-     */
-
-    public Builder removeRet(
-        java.lang.String key) {
-      if (key == null) { throw new java.lang.NullPointerException(); }
-      internalGetMutableRet().getMutableMap()
-          .remove(key);
-      return this;
-    }
-    /**
-     * Use alternate mutation accessors instead.
-     */
-    @java.lang.Deprecated
-    public java.util.Map<java.lang.String, java.lang.String>
-    getMutableRet() {
-      return internalGetMutableRet().getMutableMap();
-    }
-    /**
-     * <pre>
-     * A mapping from the output arg names from `signature` to the
-     * outputs from `node_def` that should be returned by the function.
-     * </pre>
-     *
-     * <code>map&lt;string, string&gt; ret = 4;</code>
-     */
-    public Builder putRet(
-        java.lang.String key,
-        java.lang.String value) {
-      if (key == null) { throw new java.lang.NullPointerException(); }
-      if (value == null) { throw new java.lang.NullPointerException(); }
-      internalGetMutableRet().getMutableMap()
-          .put(key, value);
-      return this;
-    }
-    /**
-     * <pre>
-     * A mapping from the output arg names from `signature` to the
-     * outputs from `node_def` that should be returned by the function.
-     * </pre>
-     *
-     * <code>map&lt;string, string&gt; ret = 4;</code>
-     */
-
-    public Builder putAllRet(
-        java.util.Map<java.lang.String, java.lang.String> values) {
-      internalGetMutableRet().getMutableMap()
-          .putAll(values);
-      return this;
-    }
-
-    private com.google.protobuf.MapField<
-        java.lang.String, java.lang.String> controlRet_;
-    private com.google.protobuf.MapField<java.lang.String, java.lang.String>
-    internalGetControlRet() {
-      if (controlRet_ == null) {
-        return com.google.protobuf.MapField.emptyMapField(
-            ControlRetDefaultEntryHolder.defaultEntry);
-      }
-      return controlRet_;
-    }
-    private com.google.protobuf.MapField<java.lang.String, java.lang.String>
-    internalGetMutableControlRet() {
-      onChanged();;
-      if (controlRet_ == null) {
-        controlRet_ = com.google.protobuf.MapField.newMapField(
-            ControlRetDefaultEntryHolder.defaultEntry);
-      }
-      if (!controlRet_.isMutable()) {
-        controlRet_ = controlRet_.copy();
-      }
-      return controlRet_;
-    }
-
-    public int getControlRetCount() {
-      return internalGetControlRet().getMap().size();
-    }
-    /**
-     * <pre>
-     * A mapping from control output names from `signature` to node names in
-     * `node_def` which should be control outputs of this function.
-     * </pre>
-     *
-     * <code>map&lt;string, string&gt; control_ret = 6;</code>
-     */
-
-    public boolean containsControlRet(
-        java.lang.String key) {
-      if (key == null) { throw new java.lang.NullPointerException(); }
-      return internalGetControlRet().getMap().containsKey(key);
-    }
-    /**
-     * Use {@link #getControlRetMap()} instead.
-     */
-    @java.lang.Deprecated
-    public java.util.Map<java.lang.String, java.lang.String> getControlRet() {
-      return getControlRetMap();
-    }
-    /**
-     * <pre>
-     * A mapping from control output names from `signature` to node names in
-     * `node_def` which should be control outputs of this function.
-     * </pre>
-     *
-     * <code>map&lt;string, string&gt; control_ret = 6;</code>
-     */
-
-    public java.util.Map<java.lang.String, java.lang.String> getControlRetMap() {
-      return internalGetControlRet().getMap();
-    }
-    /**
-     * <pre>
-     * A mapping from control output names from `signature` to node names in
-     * `node_def` which should be control outputs of this function.
-     * </pre>
-     *
-     * <code>map&lt;string, string&gt; control_ret = 6;</code>
-     */
-
-    public java.lang.String getControlRetOrDefault(
-        java.lang.String key,
-        java.lang.String defaultValue) {
-      if (key == null) { throw new java.lang.NullPointerException(); }
-      java.util.Map<java.lang.String, java.lang.String> map =
-          internalGetControlRet().getMap();
-      return map.containsKey(key) ? map.get(key) : defaultValue;
-    }
-    /**
-     * <pre>
-     * A mapping from control output names from `signature` to node names in
-     * `node_def` which should be control outputs of this function.
-     * </pre>
-     *
-     * <code>map&lt;string, string&gt; control_ret = 6;</code>
-     */
-
-    public java.lang.String getControlRetOrThrow(
-        java.lang.String key) {
-      if (key == null) { throw new java.lang.NullPointerException(); }
-      java.util.Map<java.lang.String, java.lang.String> map =
-          internalGetControlRet().getMap();
-      if (!map.containsKey(key)) {
-        throw new java.lang.IllegalArgumentException();
-      }
-      return map.get(key);
-    }
-
-    public Builder clearControlRet() {
-      internalGetMutableControlRet().getMutableMap()
-          .clear();
-      return this;
-    }
-    /**
-     * <pre>
-     * A mapping from control output names from `signature` to node names in
-     * `node_def` which should be control outputs of this function.
-     * </pre>
-     *
-     * <code>map&lt;string, string&gt; control_ret = 6;</code>
-     */
-
-    public Builder removeControlRet(
-        java.lang.String key) {
-      if (key == null) { throw new java.lang.NullPointerException(); }
-      internalGetMutableControlRet().getMutableMap()
-          .remove(key);
-      return this;
-    }
-    /**
-     * Use alternate mutation accessors instead.
-     */
-    @java.lang.Deprecated
-    public java.util.Map<java.lang.String, java.lang.String>
-    getMutableControlRet() {
-      return internalGetMutableControlRet().getMutableMap();
-    }
-    /**
-     * <pre>
-     * A mapping from control output names from `signature` to node names in
-     * `node_def` which should be control outputs of this function.
-     * </pre>
-     *
-     * <code>map&lt;string, string&gt; control_ret = 6;</code>
-     */
-    public Builder putControlRet(
-        java.lang.String key,
-        java.lang.String value) {
-      if (key == null) { throw new java.lang.NullPointerException(); }
-      if (value == null) { throw new java.lang.NullPointerException(); }
-      internalGetMutableControlRet().getMutableMap()
-          .put(key, value);
-      return this;
-    }
-    /**
-     * <pre>
-     * A mapping from control output names from `signature` to node names in
-     * `node_def` which should be control outputs of this function.
-     * </pre>
-     *
-     * <code>map&lt;string, string&gt; control_ret = 6;</code>
-     */
-
-    public Builder putAllControlRet(
-        java.util.Map<java.lang.String, java.lang.String> values) {
-      internalGetMutableControlRet().getMutableMap()
-          .putAll(values);
-      return this;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.FunctionDef)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.FunctionDef)
-  private static final org.tensorflow.proto.framework.FunctionDef DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.FunctionDef();
-  }
-
-  public static org.tensorflow.proto.framework.FunctionDef getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<FunctionDef>
-      PARSER = new com.google.protobuf.AbstractParser<FunctionDef>() {
-    @java.lang.Override
-    public FunctionDef parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new FunctionDef(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<FunctionDef> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<FunctionDef> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.framework.FunctionDef getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/FunctionDefLibrary.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/FunctionDefLibrary.java
deleted file mode 100644
index a4ad715f17c..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/FunctionDefLibrary.java
+++ /dev/null
@@ -1,1116 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/framework/function.proto
-
-package org.tensorflow.proto.framework;
-
-/**
- * <pre>
- * A library is a set of named functions.
- * </pre>
- *
- * Protobuf type {@code tensorflow.FunctionDefLibrary}
- */
-public  final class FunctionDefLibrary extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.FunctionDefLibrary)
-    FunctionDefLibraryOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use FunctionDefLibrary.newBuilder() to construct.
-  private FunctionDefLibrary(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private FunctionDefLibrary() {
-    function_ = java.util.Collections.emptyList();
-    gradient_ = java.util.Collections.emptyList();
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new FunctionDefLibrary();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private FunctionDefLibrary(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    int mutable_bitField0_ = 0;
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 10: {
-            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
-              function_ = new java.util.ArrayList<org.tensorflow.proto.framework.FunctionDef>();
-              mutable_bitField0_ |= 0x00000001;
-            }
-            function_.add(
-                input.readMessage(org.tensorflow.proto.framework.FunctionDef.parser(), extensionRegistry));
-            break;
-          }
-          case 18: {
-            if (!((mutable_bitField0_ & 0x00000002) != 0)) {
-              gradient_ = new java.util.ArrayList<org.tensorflow.proto.framework.GradientDef>();
-              mutable_bitField0_ |= 0x00000002;
-            }
-            gradient_.add(
-                input.readMessage(org.tensorflow.proto.framework.GradientDef.parser(), extensionRegistry));
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      if (((mutable_bitField0_ & 0x00000001) != 0)) {
-        function_ = java.util.Collections.unmodifiableList(function_);
-      }
-      if (((mutable_bitField0_ & 0x00000002) != 0)) {
-        gradient_ = java.util.Collections.unmodifiableList(gradient_);
-      }
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.framework.FunctionProtos.internal_static_tensorflow_FunctionDefLibrary_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.framework.FunctionProtos.internal_static_tensorflow_FunctionDefLibrary_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.framework.FunctionDefLibrary.class, org.tensorflow.proto.framework.FunctionDefLibrary.Builder.class);
-  }
-
-  public static final int FUNCTION_FIELD_NUMBER = 1;
-  private java.util.List<org.tensorflow.proto.framework.FunctionDef> function_;
-  /**
-   * <code>repeated .tensorflow.FunctionDef function = 1;</code>
-   */
-  public java.util.List<org.tensorflow.proto.framework.FunctionDef> getFunctionList() {
-    return function_;
-  }
-  /**
-   * <code>repeated .tensorflow.FunctionDef function = 1;</code>
-   */
-  public java.util.List<? extends org.tensorflow.proto.framework.FunctionDefOrBuilder> 
-      getFunctionOrBuilderList() {
-    return function_;
-  }
-  /**
-   * <code>repeated .tensorflow.FunctionDef function = 1;</code>
-   */
-  public int getFunctionCount() {
-    return function_.size();
-  }
-  /**
-   * <code>repeated .tensorflow.FunctionDef function = 1;</code>
-   */
-  public org.tensorflow.proto.framework.FunctionDef getFunction(int index) {
-    return function_.get(index);
-  }
-  /**
-   * <code>repeated .tensorflow.FunctionDef function = 1;</code>
-   */
-  public org.tensorflow.proto.framework.FunctionDefOrBuilder getFunctionOrBuilder(
-      int index) {
-    return function_.get(index);
-  }
-
-  public static final int GRADIENT_FIELD_NUMBER = 2;
-  private java.util.List<org.tensorflow.proto.framework.GradientDef> gradient_;
-  /**
-   * <code>repeated .tensorflow.GradientDef gradient = 2;</code>
-   */
-  public java.util.List<org.tensorflow.proto.framework.GradientDef> getGradientList() {
-    return gradient_;
-  }
-  /**
-   * <code>repeated .tensorflow.GradientDef gradient = 2;</code>
-   */
-  public java.util.List<? extends org.tensorflow.proto.framework.GradientDefOrBuilder> 
-      getGradientOrBuilderList() {
-    return gradient_;
-  }
-  /**
-   * <code>repeated .tensorflow.GradientDef gradient = 2;</code>
-   */
-  public int getGradientCount() {
-    return gradient_.size();
-  }
-  /**
-   * <code>repeated .tensorflow.GradientDef gradient = 2;</code>
-   */
-  public org.tensorflow.proto.framework.GradientDef getGradient(int index) {
-    return gradient_.get(index);
-  }
-  /**
-   * <code>repeated .tensorflow.GradientDef gradient = 2;</code>
-   */
-  public org.tensorflow.proto.framework.GradientDefOrBuilder getGradientOrBuilder(
-      int index) {
-    return gradient_.get(index);
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    for (int i = 0; i < function_.size(); i++) {
-      output.writeMessage(1, function_.get(i));
-    }
-    for (int i = 0; i < gradient_.size(); i++) {
-      output.writeMessage(2, gradient_.get(i));
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    for (int i = 0; i < function_.size(); i++) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(1, function_.get(i));
-    }
-    for (int i = 0; i < gradient_.size(); i++) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(2, gradient_.get(i));
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.framework.FunctionDefLibrary)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.framework.FunctionDefLibrary other = (org.tensorflow.proto.framework.FunctionDefLibrary) obj;
-
-    if (!getFunctionList()
-        .equals(other.getFunctionList())) return false;
-    if (!getGradientList()
-        .equals(other.getGradientList())) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    if (getFunctionCount() > 0) {
-      hash = (37 * hash) + FUNCTION_FIELD_NUMBER;
-      hash = (53 * hash) + getFunctionList().hashCode();
-    }
-    if (getGradientCount() > 0) {
-      hash = (37 * hash) + GRADIENT_FIELD_NUMBER;
-      hash = (53 * hash) + getGradientList().hashCode();
-    }
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.framework.FunctionDefLibrary parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.FunctionDefLibrary parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.FunctionDefLibrary parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.FunctionDefLibrary parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.FunctionDefLibrary parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.FunctionDefLibrary parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.FunctionDefLibrary parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.FunctionDefLibrary parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.FunctionDefLibrary parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.FunctionDefLibrary parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.FunctionDefLibrary parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.FunctionDefLibrary parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.framework.FunctionDefLibrary prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * <pre>
-   * A library is a set of named functions.
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.FunctionDefLibrary}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.FunctionDefLibrary)
-      org.tensorflow.proto.framework.FunctionDefLibraryOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.FunctionProtos.internal_static_tensorflow_FunctionDefLibrary_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.FunctionProtos.internal_static_tensorflow_FunctionDefLibrary_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.FunctionDefLibrary.class, org.tensorflow.proto.framework.FunctionDefLibrary.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.framework.FunctionDefLibrary.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-        getFunctionFieldBuilder();
-        getGradientFieldBuilder();
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      if (functionBuilder_ == null) {
-        function_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000001);
-      } else {
-        functionBuilder_.clear();
-      }
-      if (gradientBuilder_ == null) {
-        gradient_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000002);
-      } else {
-        gradientBuilder_.clear();
-      }
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.framework.FunctionProtos.internal_static_tensorflow_FunctionDefLibrary_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.FunctionDefLibrary getDefaultInstanceForType() {
-      return org.tensorflow.proto.framework.FunctionDefLibrary.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.FunctionDefLibrary build() {
-      org.tensorflow.proto.framework.FunctionDefLibrary result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.FunctionDefLibrary buildPartial() {
-      org.tensorflow.proto.framework.FunctionDefLibrary result = new org.tensorflow.proto.framework.FunctionDefLibrary(this);
-      int from_bitField0_ = bitField0_;
-      if (functionBuilder_ == null) {
-        if (((bitField0_ & 0x00000001) != 0)) {
-          function_ = java.util.Collections.unmodifiableList(function_);
-          bitField0_ = (bitField0_ & ~0x00000001);
-        }
-        result.function_ = function_;
-      } else {
-        result.function_ = functionBuilder_.build();
-      }
-      if (gradientBuilder_ == null) {
-        if (((bitField0_ & 0x00000002) != 0)) {
-          gradient_ = java.util.Collections.unmodifiableList(gradient_);
-          bitField0_ = (bitField0_ & ~0x00000002);
-        }
-        result.gradient_ = gradient_;
-      } else {
-        result.gradient_ = gradientBuilder_.build();
-      }
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.framework.FunctionDefLibrary) {
-        return mergeFrom((org.tensorflow.proto.framework.FunctionDefLibrary)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.framework.FunctionDefLibrary other) {
-      if (other == org.tensorflow.proto.framework.FunctionDefLibrary.getDefaultInstance()) return this;
-      if (functionBuilder_ == null) {
-        if (!other.function_.isEmpty()) {
-          if (function_.isEmpty()) {
-            function_ = other.function_;
-            bitField0_ = (bitField0_ & ~0x00000001);
-          } else {
-            ensureFunctionIsMutable();
-            function_.addAll(other.function_);
-          }
-          onChanged();
-        }
-      } else {
-        if (!other.function_.isEmpty()) {
-          if (functionBuilder_.isEmpty()) {
-            functionBuilder_.dispose();
-            functionBuilder_ = null;
-            function_ = other.function_;
-            bitField0_ = (bitField0_ & ~0x00000001);
-            functionBuilder_ = 
-              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
-                 getFunctionFieldBuilder() : null;
-          } else {
-            functionBuilder_.addAllMessages(other.function_);
-          }
-        }
-      }
-      if (gradientBuilder_ == null) {
-        if (!other.gradient_.isEmpty()) {
-          if (gradient_.isEmpty()) {
-            gradient_ = other.gradient_;
-            bitField0_ = (bitField0_ & ~0x00000002);
-          } else {
-            ensureGradientIsMutable();
-            gradient_.addAll(other.gradient_);
-          }
-          onChanged();
-        }
-      } else {
-        if (!other.gradient_.isEmpty()) {
-          if (gradientBuilder_.isEmpty()) {
-            gradientBuilder_.dispose();
-            gradientBuilder_ = null;
-            gradient_ = other.gradient_;
-            bitField0_ = (bitField0_ & ~0x00000002);
-            gradientBuilder_ = 
-              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
-                 getGradientFieldBuilder() : null;
-          } else {
-            gradientBuilder_.addAllMessages(other.gradient_);
-          }
-        }
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.framework.FunctionDefLibrary parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.framework.FunctionDefLibrary) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-    private int bitField0_;
-
-    private java.util.List<org.tensorflow.proto.framework.FunctionDef> function_ =
-      java.util.Collections.emptyList();
-    private void ensureFunctionIsMutable() {
-      if (!((bitField0_ & 0x00000001) != 0)) {
-        function_ = new java.util.ArrayList<org.tensorflow.proto.framework.FunctionDef>(function_);
-        bitField0_ |= 0x00000001;
-       }
-    }
-
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.framework.FunctionDef, org.tensorflow.proto.framework.FunctionDef.Builder, org.tensorflow.proto.framework.FunctionDefOrBuilder> functionBuilder_;
-
-    /**
-     * <code>repeated .tensorflow.FunctionDef function = 1;</code>
-     */
-    public java.util.List<org.tensorflow.proto.framework.FunctionDef> getFunctionList() {
-      if (functionBuilder_ == null) {
-        return java.util.Collections.unmodifiableList(function_);
-      } else {
-        return functionBuilder_.getMessageList();
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.FunctionDef function = 1;</code>
-     */
-    public int getFunctionCount() {
-      if (functionBuilder_ == null) {
-        return function_.size();
-      } else {
-        return functionBuilder_.getCount();
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.FunctionDef function = 1;</code>
-     */
-    public org.tensorflow.proto.framework.FunctionDef getFunction(int index) {
-      if (functionBuilder_ == null) {
-        return function_.get(index);
-      } else {
-        return functionBuilder_.getMessage(index);
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.FunctionDef function = 1;</code>
-     */
-    public Builder setFunction(
-        int index, org.tensorflow.proto.framework.FunctionDef value) {
-      if (functionBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureFunctionIsMutable();
-        function_.set(index, value);
-        onChanged();
-      } else {
-        functionBuilder_.setMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.FunctionDef function = 1;</code>
-     */
-    public Builder setFunction(
-        int index, org.tensorflow.proto.framework.FunctionDef.Builder builderForValue) {
-      if (functionBuilder_ == null) {
-        ensureFunctionIsMutable();
-        function_.set(index, builderForValue.build());
-        onChanged();
-      } else {
-        functionBuilder_.setMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.FunctionDef function = 1;</code>
-     */
-    public Builder addFunction(org.tensorflow.proto.framework.FunctionDef value) {
-      if (functionBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureFunctionIsMutable();
-        function_.add(value);
-        onChanged();
-      } else {
-        functionBuilder_.addMessage(value);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.FunctionDef function = 1;</code>
-     */
-    public Builder addFunction(
-        int index, org.tensorflow.proto.framework.FunctionDef value) {
-      if (functionBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureFunctionIsMutable();
-        function_.add(index, value);
-        onChanged();
-      } else {
-        functionBuilder_.addMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.FunctionDef function = 1;</code>
-     */
-    public Builder addFunction(
-        org.tensorflow.proto.framework.FunctionDef.Builder builderForValue) {
-      if (functionBuilder_ == null) {
-        ensureFunctionIsMutable();
-        function_.add(builderForValue.build());
-        onChanged();
-      } else {
-        functionBuilder_.addMessage(builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.FunctionDef function = 1;</code>
-     */
-    public Builder addFunction(
-        int index, org.tensorflow.proto.framework.FunctionDef.Builder builderForValue) {
-      if (functionBuilder_ == null) {
-        ensureFunctionIsMutable();
-        function_.add(index, builderForValue.build());
-        onChanged();
-      } else {
-        functionBuilder_.addMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.FunctionDef function = 1;</code>
-     */
-    public Builder addAllFunction(
-        java.lang.Iterable<? extends org.tensorflow.proto.framework.FunctionDef> values) {
-      if (functionBuilder_ == null) {
-        ensureFunctionIsMutable();
-        com.google.protobuf.AbstractMessageLite.Builder.addAll(
-            values, function_);
-        onChanged();
-      } else {
-        functionBuilder_.addAllMessages(values);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.FunctionDef function = 1;</code>
-     */
-    public Builder clearFunction() {
-      if (functionBuilder_ == null) {
-        function_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000001);
-        onChanged();
-      } else {
-        functionBuilder_.clear();
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.FunctionDef function = 1;</code>
-     */
-    public Builder removeFunction(int index) {
-      if (functionBuilder_ == null) {
-        ensureFunctionIsMutable();
-        function_.remove(index);
-        onChanged();
-      } else {
-        functionBuilder_.remove(index);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.FunctionDef function = 1;</code>
-     */
-    public org.tensorflow.proto.framework.FunctionDef.Builder getFunctionBuilder(
-        int index) {
-      return getFunctionFieldBuilder().getBuilder(index);
-    }
-    /**
-     * <code>repeated .tensorflow.FunctionDef function = 1;</code>
-     */
-    public org.tensorflow.proto.framework.FunctionDefOrBuilder getFunctionOrBuilder(
-        int index) {
-      if (functionBuilder_ == null) {
-        return function_.get(index);  } else {
-        return functionBuilder_.getMessageOrBuilder(index);
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.FunctionDef function = 1;</code>
-     */
-    public java.util.List<? extends org.tensorflow.proto.framework.FunctionDefOrBuilder> 
-         getFunctionOrBuilderList() {
-      if (functionBuilder_ != null) {
-        return functionBuilder_.getMessageOrBuilderList();
-      } else {
-        return java.util.Collections.unmodifiableList(function_);
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.FunctionDef function = 1;</code>
-     */
-    public org.tensorflow.proto.framework.FunctionDef.Builder addFunctionBuilder() {
-      return getFunctionFieldBuilder().addBuilder(
-          org.tensorflow.proto.framework.FunctionDef.getDefaultInstance());
-    }
-    /**
-     * <code>repeated .tensorflow.FunctionDef function = 1;</code>
-     */
-    public org.tensorflow.proto.framework.FunctionDef.Builder addFunctionBuilder(
-        int index) {
-      return getFunctionFieldBuilder().addBuilder(
-          index, org.tensorflow.proto.framework.FunctionDef.getDefaultInstance());
-    }
-    /**
-     * <code>repeated .tensorflow.FunctionDef function = 1;</code>
-     */
-    public java.util.List<org.tensorflow.proto.framework.FunctionDef.Builder> 
-         getFunctionBuilderList() {
-      return getFunctionFieldBuilder().getBuilderList();
-    }
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.framework.FunctionDef, org.tensorflow.proto.framework.FunctionDef.Builder, org.tensorflow.proto.framework.FunctionDefOrBuilder> 
-        getFunctionFieldBuilder() {
-      if (functionBuilder_ == null) {
-        functionBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
-            org.tensorflow.proto.framework.FunctionDef, org.tensorflow.proto.framework.FunctionDef.Builder, org.tensorflow.proto.framework.FunctionDefOrBuilder>(
-                function_,
-                ((bitField0_ & 0x00000001) != 0),
-                getParentForChildren(),
-                isClean());
-        function_ = null;
-      }
-      return functionBuilder_;
-    }
-
-    private java.util.List<org.tensorflow.proto.framework.GradientDef> gradient_ =
-      java.util.Collections.emptyList();
-    private void ensureGradientIsMutable() {
-      if (!((bitField0_ & 0x00000002) != 0)) {
-        gradient_ = new java.util.ArrayList<org.tensorflow.proto.framework.GradientDef>(gradient_);
-        bitField0_ |= 0x00000002;
-       }
-    }
-
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.framework.GradientDef, org.tensorflow.proto.framework.GradientDef.Builder, org.tensorflow.proto.framework.GradientDefOrBuilder> gradientBuilder_;
-
-    /**
-     * <code>repeated .tensorflow.GradientDef gradient = 2;</code>
-     */
-    public java.util.List<org.tensorflow.proto.framework.GradientDef> getGradientList() {
-      if (gradientBuilder_ == null) {
-        return java.util.Collections.unmodifiableList(gradient_);
-      } else {
-        return gradientBuilder_.getMessageList();
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.GradientDef gradient = 2;</code>
-     */
-    public int getGradientCount() {
-      if (gradientBuilder_ == null) {
-        return gradient_.size();
-      } else {
-        return gradientBuilder_.getCount();
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.GradientDef gradient = 2;</code>
-     */
-    public org.tensorflow.proto.framework.GradientDef getGradient(int index) {
-      if (gradientBuilder_ == null) {
-        return gradient_.get(index);
-      } else {
-        return gradientBuilder_.getMessage(index);
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.GradientDef gradient = 2;</code>
-     */
-    public Builder setGradient(
-        int index, org.tensorflow.proto.framework.GradientDef value) {
-      if (gradientBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureGradientIsMutable();
-        gradient_.set(index, value);
-        onChanged();
-      } else {
-        gradientBuilder_.setMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.GradientDef gradient = 2;</code>
-     */
-    public Builder setGradient(
-        int index, org.tensorflow.proto.framework.GradientDef.Builder builderForValue) {
-      if (gradientBuilder_ == null) {
-        ensureGradientIsMutable();
-        gradient_.set(index, builderForValue.build());
-        onChanged();
-      } else {
-        gradientBuilder_.setMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.GradientDef gradient = 2;</code>
-     */
-    public Builder addGradient(org.tensorflow.proto.framework.GradientDef value) {
-      if (gradientBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureGradientIsMutable();
-        gradient_.add(value);
-        onChanged();
-      } else {
-        gradientBuilder_.addMessage(value);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.GradientDef gradient = 2;</code>
-     */
-    public Builder addGradient(
-        int index, org.tensorflow.proto.framework.GradientDef value) {
-      if (gradientBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureGradientIsMutable();
-        gradient_.add(index, value);
-        onChanged();
-      } else {
-        gradientBuilder_.addMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.GradientDef gradient = 2;</code>
-     */
-    public Builder addGradient(
-        org.tensorflow.proto.framework.GradientDef.Builder builderForValue) {
-      if (gradientBuilder_ == null) {
-        ensureGradientIsMutable();
-        gradient_.add(builderForValue.build());
-        onChanged();
-      } else {
-        gradientBuilder_.addMessage(builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.GradientDef gradient = 2;</code>
-     */
-    public Builder addGradient(
-        int index, org.tensorflow.proto.framework.GradientDef.Builder builderForValue) {
-      if (gradientBuilder_ == null) {
-        ensureGradientIsMutable();
-        gradient_.add(index, builderForValue.build());
-        onChanged();
-      } else {
-        gradientBuilder_.addMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.GradientDef gradient = 2;</code>
-     */
-    public Builder addAllGradient(
-        java.lang.Iterable<? extends org.tensorflow.proto.framework.GradientDef> values) {
-      if (gradientBuilder_ == null) {
-        ensureGradientIsMutable();
-        com.google.protobuf.AbstractMessageLite.Builder.addAll(
-            values, gradient_);
-        onChanged();
-      } else {
-        gradientBuilder_.addAllMessages(values);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.GradientDef gradient = 2;</code>
-     */
-    public Builder clearGradient() {
-      if (gradientBuilder_ == null) {
-        gradient_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000002);
-        onChanged();
-      } else {
-        gradientBuilder_.clear();
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.GradientDef gradient = 2;</code>
-     */
-    public Builder removeGradient(int index) {
-      if (gradientBuilder_ == null) {
-        ensureGradientIsMutable();
-        gradient_.remove(index);
-        onChanged();
-      } else {
-        gradientBuilder_.remove(index);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.GradientDef gradient = 2;</code>
-     */
-    public org.tensorflow.proto.framework.GradientDef.Builder getGradientBuilder(
-        int index) {
-      return getGradientFieldBuilder().getBuilder(index);
-    }
-    /**
-     * <code>repeated .tensorflow.GradientDef gradient = 2;</code>
-     */
-    public org.tensorflow.proto.framework.GradientDefOrBuilder getGradientOrBuilder(
-        int index) {
-      if (gradientBuilder_ == null) {
-        return gradient_.get(index);  } else {
-        return gradientBuilder_.getMessageOrBuilder(index);
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.GradientDef gradient = 2;</code>
-     */
-    public java.util.List<? extends org.tensorflow.proto.framework.GradientDefOrBuilder> 
-         getGradientOrBuilderList() {
-      if (gradientBuilder_ != null) {
-        return gradientBuilder_.getMessageOrBuilderList();
-      } else {
-        return java.util.Collections.unmodifiableList(gradient_);
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.GradientDef gradient = 2;</code>
-     */
-    public org.tensorflow.proto.framework.GradientDef.Builder addGradientBuilder() {
-      return getGradientFieldBuilder().addBuilder(
-          org.tensorflow.proto.framework.GradientDef.getDefaultInstance());
-    }
-    /**
-     * <code>repeated .tensorflow.GradientDef gradient = 2;</code>
-     */
-    public org.tensorflow.proto.framework.GradientDef.Builder addGradientBuilder(
-        int index) {
-      return getGradientFieldBuilder().addBuilder(
-          index, org.tensorflow.proto.framework.GradientDef.getDefaultInstance());
-    }
-    /**
-     * <code>repeated .tensorflow.GradientDef gradient = 2;</code>
-     */
-    public java.util.List<org.tensorflow.proto.framework.GradientDef.Builder> 
-         getGradientBuilderList() {
-      return getGradientFieldBuilder().getBuilderList();
-    }
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.framework.GradientDef, org.tensorflow.proto.framework.GradientDef.Builder, org.tensorflow.proto.framework.GradientDefOrBuilder> 
-        getGradientFieldBuilder() {
-      if (gradientBuilder_ == null) {
-        gradientBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
-            org.tensorflow.proto.framework.GradientDef, org.tensorflow.proto.framework.GradientDef.Builder, org.tensorflow.proto.framework.GradientDefOrBuilder>(
-                gradient_,
-                ((bitField0_ & 0x00000002) != 0),
-                getParentForChildren(),
-                isClean());
-        gradient_ = null;
-      }
-      return gradientBuilder_;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.FunctionDefLibrary)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.FunctionDefLibrary)
-  private static final org.tensorflow.proto.framework.FunctionDefLibrary DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.FunctionDefLibrary();
-  }
-
-  public static org.tensorflow.proto.framework.FunctionDefLibrary getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<FunctionDefLibrary>
-      PARSER = new com.google.protobuf.AbstractParser<FunctionDefLibrary>() {
-    @java.lang.Override
-    public FunctionDefLibrary parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new FunctionDefLibrary(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<FunctionDefLibrary> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<FunctionDefLibrary> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.framework.FunctionDefLibrary getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/FunctionDefLibraryOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/FunctionDefLibraryOrBuilder.java
deleted file mode 100644
index cfb61e977ba..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/FunctionDefLibraryOrBuilder.java
+++ /dev/null
@@ -1,57 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/framework/function.proto
-
-package org.tensorflow.proto.framework;
-
-public interface FunctionDefLibraryOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.FunctionDefLibrary)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <code>repeated .tensorflow.FunctionDef function = 1;</code>
-   */
-  java.util.List<org.tensorflow.proto.framework.FunctionDef> 
-      getFunctionList();
-  /**
-   * <code>repeated .tensorflow.FunctionDef function = 1;</code>
-   */
-  org.tensorflow.proto.framework.FunctionDef getFunction(int index);
-  /**
-   * <code>repeated .tensorflow.FunctionDef function = 1;</code>
-   */
-  int getFunctionCount();
-  /**
-   * <code>repeated .tensorflow.FunctionDef function = 1;</code>
-   */
-  java.util.List<? extends org.tensorflow.proto.framework.FunctionDefOrBuilder> 
-      getFunctionOrBuilderList();
-  /**
-   * <code>repeated .tensorflow.FunctionDef function = 1;</code>
-   */
-  org.tensorflow.proto.framework.FunctionDefOrBuilder getFunctionOrBuilder(
-      int index);
-
-  /**
-   * <code>repeated .tensorflow.GradientDef gradient = 2;</code>
-   */
-  java.util.List<org.tensorflow.proto.framework.GradientDef> 
-      getGradientList();
-  /**
-   * <code>repeated .tensorflow.GradientDef gradient = 2;</code>
-   */
-  org.tensorflow.proto.framework.GradientDef getGradient(int index);
-  /**
-   * <code>repeated .tensorflow.GradientDef gradient = 2;</code>
-   */
-  int getGradientCount();
-  /**
-   * <code>repeated .tensorflow.GradientDef gradient = 2;</code>
-   */
-  java.util.List<? extends org.tensorflow.proto.framework.GradientDefOrBuilder> 
-      getGradientOrBuilderList();
-  /**
-   * <code>repeated .tensorflow.GradientDef gradient = 2;</code>
-   */
-  org.tensorflow.proto.framework.GradientDefOrBuilder getGradientOrBuilder(
-      int index);
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/FunctionDefOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/FunctionDefOrBuilder.java
deleted file mode 100644
index ebae0875f50..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/FunctionDefOrBuilder.java
+++ /dev/null
@@ -1,381 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/framework/function.proto
-
-package org.tensorflow.proto.framework;
-
-public interface FunctionDefOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.FunctionDef)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <pre>
-   * The definition of the function's name, arguments, return values,
-   * attrs etc.
-   * </pre>
-   *
-   * <code>.tensorflow.OpDef signature = 1;</code>
-   */
-  boolean hasSignature();
-  /**
-   * <pre>
-   * The definition of the function's name, arguments, return values,
-   * attrs etc.
-   * </pre>
-   *
-   * <code>.tensorflow.OpDef signature = 1;</code>
-   */
-  org.tensorflow.proto.framework.OpDef getSignature();
-  /**
-   * <pre>
-   * The definition of the function's name, arguments, return values,
-   * attrs etc.
-   * </pre>
-   *
-   * <code>.tensorflow.OpDef signature = 1;</code>
-   */
-  org.tensorflow.proto.framework.OpDefOrBuilder getSignatureOrBuilder();
-
-  /**
-   * <pre>
-   * Attributes specific to this function definition.
-   * </pre>
-   *
-   * <code>map&lt;string, .tensorflow.AttrValue&gt; attr = 5;</code>
-   */
-  int getAttrCount();
-  /**
-   * <pre>
-   * Attributes specific to this function definition.
-   * </pre>
-   *
-   * <code>map&lt;string, .tensorflow.AttrValue&gt; attr = 5;</code>
-   */
-  boolean containsAttr(
-      java.lang.String key);
-  /**
-   * Use {@link #getAttrMap()} instead.
-   */
-  @java.lang.Deprecated
-  java.util.Map<java.lang.String, org.tensorflow.proto.framework.AttrValue>
-  getAttr();
-  /**
-   * <pre>
-   * Attributes specific to this function definition.
-   * </pre>
-   *
-   * <code>map&lt;string, .tensorflow.AttrValue&gt; attr = 5;</code>
-   */
-  java.util.Map<java.lang.String, org.tensorflow.proto.framework.AttrValue>
-  getAttrMap();
-  /**
-   * <pre>
-   * Attributes specific to this function definition.
-   * </pre>
-   *
-   * <code>map&lt;string, .tensorflow.AttrValue&gt; attr = 5;</code>
-   */
-
-  org.tensorflow.proto.framework.AttrValue getAttrOrDefault(
-      java.lang.String key,
-      org.tensorflow.proto.framework.AttrValue defaultValue);
-  /**
-   * <pre>
-   * Attributes specific to this function definition.
-   * </pre>
-   *
-   * <code>map&lt;string, .tensorflow.AttrValue&gt; attr = 5;</code>
-   */
-
-  org.tensorflow.proto.framework.AttrValue getAttrOrThrow(
-      java.lang.String key);
-
-  /**
-   * <code>map&lt;uint32, .tensorflow.FunctionDef.ArgAttrs&gt; arg_attr = 7;</code>
-   */
-  int getArgAttrCount();
-  /**
-   * <code>map&lt;uint32, .tensorflow.FunctionDef.ArgAttrs&gt; arg_attr = 7;</code>
-   */
-  boolean containsArgAttr(
-      int key);
-  /**
-   * Use {@link #getArgAttrMap()} instead.
-   */
-  @java.lang.Deprecated
-  java.util.Map<java.lang.Integer, org.tensorflow.proto.framework.FunctionDef.ArgAttrs>
-  getArgAttr();
-  /**
-   * <code>map&lt;uint32, .tensorflow.FunctionDef.ArgAttrs&gt; arg_attr = 7;</code>
-   */
-  java.util.Map<java.lang.Integer, org.tensorflow.proto.framework.FunctionDef.ArgAttrs>
-  getArgAttrMap();
-  /**
-   * <code>map&lt;uint32, .tensorflow.FunctionDef.ArgAttrs&gt; arg_attr = 7;</code>
-   */
-
-  org.tensorflow.proto.framework.FunctionDef.ArgAttrs getArgAttrOrDefault(
-      int key,
-      org.tensorflow.proto.framework.FunctionDef.ArgAttrs defaultValue);
-  /**
-   * <code>map&lt;uint32, .tensorflow.FunctionDef.ArgAttrs&gt; arg_attr = 7;</code>
-   */
-
-  org.tensorflow.proto.framework.FunctionDef.ArgAttrs getArgAttrOrThrow(
-      int key);
-
-  /**
-   * <pre>
-   * Unique IDs for each resource argument, used to track aliasing resources. If
-   * Argument A and Argument B alias each other, then
-   * resource_arg_unique_ids[A.index] == resource_arg_unique_ids[B.index].
-   * If this field is empty, none of the arguments could alias; otherwise, every
-   * resource argument should have an entry in this field.
-   * When instantiated, the unique IDs will be attached to the _Arg nodes'
-   * "_resource_arg_unique_id" attribute.
-   * </pre>
-   *
-   * <code>map&lt;uint32, uint32&gt; resource_arg_unique_id = 8;</code>
-   */
-  int getResourceArgUniqueIdCount();
-  /**
-   * <pre>
-   * Unique IDs for each resource argument, used to track aliasing resources. If
-   * Argument A and Argument B alias each other, then
-   * resource_arg_unique_ids[A.index] == resource_arg_unique_ids[B.index].
-   * If this field is empty, none of the arguments could alias; otherwise, every
-   * resource argument should have an entry in this field.
-   * When instantiated, the unique IDs will be attached to the _Arg nodes'
-   * "_resource_arg_unique_id" attribute.
-   * </pre>
-   *
-   * <code>map&lt;uint32, uint32&gt; resource_arg_unique_id = 8;</code>
-   */
-  boolean containsResourceArgUniqueId(
-      int key);
-  /**
-   * Use {@link #getResourceArgUniqueIdMap()} instead.
-   */
-  @java.lang.Deprecated
-  java.util.Map<java.lang.Integer, java.lang.Integer>
-  getResourceArgUniqueId();
-  /**
-   * <pre>
-   * Unique IDs for each resource argument, used to track aliasing resources. If
-   * Argument A and Argument B alias each other, then
-   * resource_arg_unique_ids[A.index] == resource_arg_unique_ids[B.index].
-   * If this field is empty, none of the arguments could alias; otherwise, every
-   * resource argument should have an entry in this field.
-   * When instantiated, the unique IDs will be attached to the _Arg nodes'
-   * "_resource_arg_unique_id" attribute.
-   * </pre>
-   *
-   * <code>map&lt;uint32, uint32&gt; resource_arg_unique_id = 8;</code>
-   */
-  java.util.Map<java.lang.Integer, java.lang.Integer>
-  getResourceArgUniqueIdMap();
-  /**
-   * <pre>
-   * Unique IDs for each resource argument, used to track aliasing resources. If
-   * Argument A and Argument B alias each other, then
-   * resource_arg_unique_ids[A.index] == resource_arg_unique_ids[B.index].
-   * If this field is empty, none of the arguments could alias; otherwise, every
-   * resource argument should have an entry in this field.
-   * When instantiated, the unique IDs will be attached to the _Arg nodes'
-   * "_resource_arg_unique_id" attribute.
-   * </pre>
-   *
-   * <code>map&lt;uint32, uint32&gt; resource_arg_unique_id = 8;</code>
-   */
-
-  int getResourceArgUniqueIdOrDefault(
-      int key,
-      int defaultValue);
-  /**
-   * <pre>
-   * Unique IDs for each resource argument, used to track aliasing resources. If
-   * Argument A and Argument B alias each other, then
-   * resource_arg_unique_ids[A.index] == resource_arg_unique_ids[B.index].
-   * If this field is empty, none of the arguments could alias; otherwise, every
-   * resource argument should have an entry in this field.
-   * When instantiated, the unique IDs will be attached to the _Arg nodes'
-   * "_resource_arg_unique_id" attribute.
-   * </pre>
-   *
-   * <code>map&lt;uint32, uint32&gt; resource_arg_unique_id = 8;</code>
-   */
-
-  int getResourceArgUniqueIdOrThrow(
-      int key);
-
-  /**
-   * <pre>
-   * By convention, "op" in node_def is resolved by consulting with a
-   * user-defined library first. If not resolved, "func" is assumed to
-   * be a builtin op.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.NodeDef node_def = 3;</code>
-   */
-  java.util.List<org.tensorflow.proto.framework.NodeDef> 
-      getNodeDefList();
-  /**
-   * <pre>
-   * By convention, "op" in node_def is resolved by consulting with a
-   * user-defined library first. If not resolved, "func" is assumed to
-   * be a builtin op.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.NodeDef node_def = 3;</code>
-   */
-  org.tensorflow.proto.framework.NodeDef getNodeDef(int index);
-  /**
-   * <pre>
-   * By convention, "op" in node_def is resolved by consulting with a
-   * user-defined library first. If not resolved, "func" is assumed to
-   * be a builtin op.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.NodeDef node_def = 3;</code>
-   */
-  int getNodeDefCount();
-  /**
-   * <pre>
-   * By convention, "op" in node_def is resolved by consulting with a
-   * user-defined library first. If not resolved, "func" is assumed to
-   * be a builtin op.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.NodeDef node_def = 3;</code>
-   */
-  java.util.List<? extends org.tensorflow.proto.framework.NodeDefOrBuilder> 
-      getNodeDefOrBuilderList();
-  /**
-   * <pre>
-   * By convention, "op" in node_def is resolved by consulting with a
-   * user-defined library first. If not resolved, "func" is assumed to
-   * be a builtin op.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.NodeDef node_def = 3;</code>
-   */
-  org.tensorflow.proto.framework.NodeDefOrBuilder getNodeDefOrBuilder(
-      int index);
-
-  /**
-   * <pre>
-   * A mapping from the output arg names from `signature` to the
-   * outputs from `node_def` that should be returned by the function.
-   * </pre>
-   *
-   * <code>map&lt;string, string&gt; ret = 4;</code>
-   */
-  int getRetCount();
-  /**
-   * <pre>
-   * A mapping from the output arg names from `signature` to the
-   * outputs from `node_def` that should be returned by the function.
-   * </pre>
-   *
-   * <code>map&lt;string, string&gt; ret = 4;</code>
-   */
-  boolean containsRet(
-      java.lang.String key);
-  /**
-   * Use {@link #getRetMap()} instead.
-   */
-  @java.lang.Deprecated
-  java.util.Map<java.lang.String, java.lang.String>
-  getRet();
-  /**
-   * <pre>
-   * A mapping from the output arg names from `signature` to the
-   * outputs from `node_def` that should be returned by the function.
-   * </pre>
-   *
-   * <code>map&lt;string, string&gt; ret = 4;</code>
-   */
-  java.util.Map<java.lang.String, java.lang.String>
-  getRetMap();
-  /**
-   * <pre>
-   * A mapping from the output arg names from `signature` to the
-   * outputs from `node_def` that should be returned by the function.
-   * </pre>
-   *
-   * <code>map&lt;string, string&gt; ret = 4;</code>
-   */
-
-  java.lang.String getRetOrDefault(
-      java.lang.String key,
-      java.lang.String defaultValue);
-  /**
-   * <pre>
-   * A mapping from the output arg names from `signature` to the
-   * outputs from `node_def` that should be returned by the function.
-   * </pre>
-   *
-   * <code>map&lt;string, string&gt; ret = 4;</code>
-   */
-
-  java.lang.String getRetOrThrow(
-      java.lang.String key);
-
-  /**
-   * <pre>
-   * A mapping from control output names from `signature` to node names in
-   * `node_def` which should be control outputs of this function.
-   * </pre>
-   *
-   * <code>map&lt;string, string&gt; control_ret = 6;</code>
-   */
-  int getControlRetCount();
-  /**
-   * <pre>
-   * A mapping from control output names from `signature` to node names in
-   * `node_def` which should be control outputs of this function.
-   * </pre>
-   *
-   * <code>map&lt;string, string&gt; control_ret = 6;</code>
-   */
-  boolean containsControlRet(
-      java.lang.String key);
-  /**
-   * Use {@link #getControlRetMap()} instead.
-   */
-  @java.lang.Deprecated
-  java.util.Map<java.lang.String, java.lang.String>
-  getControlRet();
-  /**
-   * <pre>
-   * A mapping from control output names from `signature` to node names in
-   * `node_def` which should be control outputs of this function.
-   * </pre>
-   *
-   * <code>map&lt;string, string&gt; control_ret = 6;</code>
-   */
-  java.util.Map<java.lang.String, java.lang.String>
-  getControlRetMap();
-  /**
-   * <pre>
-   * A mapping from control output names from `signature` to node names in
-   * `node_def` which should be control outputs of this function.
-   * </pre>
-   *
-   * <code>map&lt;string, string&gt; control_ret = 6;</code>
-   */
-
-  java.lang.String getControlRetOrDefault(
-      java.lang.String key,
-      java.lang.String defaultValue);
-  /**
-   * <pre>
-   * A mapping from control output names from `signature` to node names in
-   * `node_def` which should be control outputs of this function.
-   * </pre>
-   *
-   * <code>map&lt;string, string&gt; control_ret = 6;</code>
-   */
-
-  java.lang.String getControlRetOrThrow(
-      java.lang.String key);
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/FunctionProtos.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/FunctionProtos.java
deleted file mode 100644
index e33df603ca8..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/FunctionProtos.java
+++ /dev/null
@@ -1,184 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/framework/function.proto
-
-package org.tensorflow.proto.framework;
-
-public final class FunctionProtos {
-  private FunctionProtos() {}
-  public static void registerAllExtensions(
-      com.google.protobuf.ExtensionRegistryLite registry) {
-  }
-
-  public static void registerAllExtensions(
-      com.google.protobuf.ExtensionRegistry registry) {
-    registerAllExtensions(
-        (com.google.protobuf.ExtensionRegistryLite) registry);
-  }
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_FunctionDefLibrary_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_FunctionDefLibrary_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_FunctionDef_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_FunctionDef_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_FunctionDef_AttrEntry_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_FunctionDef_AttrEntry_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_FunctionDef_ArgAttrs_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_FunctionDef_ArgAttrs_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_FunctionDef_ArgAttrs_AttrEntry_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_FunctionDef_ArgAttrs_AttrEntry_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_FunctionDef_ArgAttrEntry_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_FunctionDef_ArgAttrEntry_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_FunctionDef_ResourceArgUniqueIdEntry_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_FunctionDef_ResourceArgUniqueIdEntry_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_FunctionDef_RetEntry_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_FunctionDef_RetEntry_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_FunctionDef_ControlRetEntry_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_FunctionDef_ControlRetEntry_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_GradientDef_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_GradientDef_fieldAccessorTable;
-
-  public static com.google.protobuf.Descriptors.FileDescriptor
-      getDescriptor() {
-    return descriptor;
-  }
-  private static  com.google.protobuf.Descriptors.FileDescriptor
-      descriptor;
-  static {
-    java.lang.String[] descriptorData = {
-      "\n(tensorflow/core/framework/function.pro" +
-      "to\022\ntensorflow\032*tensorflow/core/framewor" +
-      "k/attr_value.proto\032(tensorflow/core/fram" +
-      "ework/node_def.proto\032&tensorflow/core/fr" +
-      "amework/op_def.proto\"j\n\022FunctionDefLibra" +
-      "ry\022)\n\010function\030\001 \003(\0132\027.tensorflow.Functi" +
-      "onDef\022)\n\010gradient\030\002 \003(\0132\027.tensorflow.Gra" +
-      "dientDef\"\304\006\n\013FunctionDef\022$\n\tsignature\030\001 " +
-      "\001(\0132\021.tensorflow.OpDef\022/\n\004attr\030\005 \003(\0132!.t" +
-      "ensorflow.FunctionDef.AttrEntry\0226\n\010arg_a" +
-      "ttr\030\007 \003(\0132$.tensorflow.FunctionDef.ArgAt" +
-      "trEntry\022P\n\026resource_arg_unique_id\030\010 \003(\0132" +
-      "0.tensorflow.FunctionDef.ResourceArgUniq" +
-      "ueIdEntry\022%\n\010node_def\030\003 \003(\0132\023.tensorflow" +
-      ".NodeDef\022-\n\003ret\030\004 \003(\0132 .tensorflow.Funct" +
-      "ionDef.RetEntry\022<\n\013control_ret\030\006 \003(\0132\'.t" +
-      "ensorflow.FunctionDef.ControlRetEntry\032B\n" +
-      "\tAttrEntry\022\013\n\003key\030\001 \001(\t\022$\n\005value\030\002 \001(\0132\025" +
-      ".tensorflow.AttrValue:\0028\001\032\210\001\n\010ArgAttrs\0228" +
-      "\n\004attr\030\001 \003(\0132*.tensorflow.FunctionDef.Ar" +
-      "gAttrs.AttrEntry\032B\n\tAttrEntry\022\013\n\003key\030\001 \001" +
-      "(\t\022$\n\005value\030\002 \001(\0132\025.tensorflow.AttrValue" +
-      ":\0028\001\032P\n\014ArgAttrEntry\022\013\n\003key\030\001 \001(\r\022/\n\005val" +
-      "ue\030\002 \001(\0132 .tensorflow.FunctionDef.ArgAtt" +
-      "rs:\0028\001\032:\n\030ResourceArgUniqueIdEntry\022\013\n\003ke" +
-      "y\030\001 \001(\r\022\r\n\005value\030\002 \001(\r:\0028\001\032*\n\010RetEntry\022\013" +
-      "\n\003key\030\001 \001(\t\022\r\n\005value\030\002 \001(\t:\0028\001\0321\n\017Contro" +
-      "lRetEntry\022\013\n\003key\030\001 \001(\t\022\r\n\005value\030\002 \001(\t:\0028" +
-      "\001J\004\010\002\020\003\";\n\013GradientDef\022\025\n\rfunction_name\030" +
-      "\001 \001(\t\022\025\n\rgradient_func\030\002 \001(\tB\206\001\n\036org.ten" +
-      "sorflow.proto.frameworkB\016FunctionProtosP" +
-      "\001ZOgithub.com/tensorflow/tensorflow/tens" +
-      "orflow/go/core/framework/function_go_pro" +
-      "to\370\001\001b\006proto3"
-    };
-    descriptor = com.google.protobuf.Descriptors.FileDescriptor
-      .internalBuildGeneratedFileFrom(descriptorData,
-        new com.google.protobuf.Descriptors.FileDescriptor[] {
-          org.tensorflow.proto.framework.AttrValueProtos.getDescriptor(),
-          org.tensorflow.proto.framework.NodeProto.getDescriptor(),
-          org.tensorflow.proto.framework.OpDefProtos.getDescriptor(),
-        });
-    internal_static_tensorflow_FunctionDefLibrary_descriptor =
-      getDescriptor().getMessageTypes().get(0);
-    internal_static_tensorflow_FunctionDefLibrary_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_FunctionDefLibrary_descriptor,
-        new java.lang.String[] { "Function", "Gradient", });
-    internal_static_tensorflow_FunctionDef_descriptor =
-      getDescriptor().getMessageTypes().get(1);
-    internal_static_tensorflow_FunctionDef_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_FunctionDef_descriptor,
-        new java.lang.String[] { "Signature", "Attr", "ArgAttr", "ResourceArgUniqueId", "NodeDef", "Ret", "ControlRet", });
-    internal_static_tensorflow_FunctionDef_AttrEntry_descriptor =
-      internal_static_tensorflow_FunctionDef_descriptor.getNestedTypes().get(0);
-    internal_static_tensorflow_FunctionDef_AttrEntry_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_FunctionDef_AttrEntry_descriptor,
-        new java.lang.String[] { "Key", "Value", });
-    internal_static_tensorflow_FunctionDef_ArgAttrs_descriptor =
-      internal_static_tensorflow_FunctionDef_descriptor.getNestedTypes().get(1);
-    internal_static_tensorflow_FunctionDef_ArgAttrs_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_FunctionDef_ArgAttrs_descriptor,
-        new java.lang.String[] { "Attr", });
-    internal_static_tensorflow_FunctionDef_ArgAttrs_AttrEntry_descriptor =
-      internal_static_tensorflow_FunctionDef_ArgAttrs_descriptor.getNestedTypes().get(0);
-    internal_static_tensorflow_FunctionDef_ArgAttrs_AttrEntry_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_FunctionDef_ArgAttrs_AttrEntry_descriptor,
-        new java.lang.String[] { "Key", "Value", });
-    internal_static_tensorflow_FunctionDef_ArgAttrEntry_descriptor =
-      internal_static_tensorflow_FunctionDef_descriptor.getNestedTypes().get(2);
-    internal_static_tensorflow_FunctionDef_ArgAttrEntry_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_FunctionDef_ArgAttrEntry_descriptor,
-        new java.lang.String[] { "Key", "Value", });
-    internal_static_tensorflow_FunctionDef_ResourceArgUniqueIdEntry_descriptor =
-      internal_static_tensorflow_FunctionDef_descriptor.getNestedTypes().get(3);
-    internal_static_tensorflow_FunctionDef_ResourceArgUniqueIdEntry_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_FunctionDef_ResourceArgUniqueIdEntry_descriptor,
-        new java.lang.String[] { "Key", "Value", });
-    internal_static_tensorflow_FunctionDef_RetEntry_descriptor =
-      internal_static_tensorflow_FunctionDef_descriptor.getNestedTypes().get(4);
-    internal_static_tensorflow_FunctionDef_RetEntry_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_FunctionDef_RetEntry_descriptor,
-        new java.lang.String[] { "Key", "Value", });
-    internal_static_tensorflow_FunctionDef_ControlRetEntry_descriptor =
-      internal_static_tensorflow_FunctionDef_descriptor.getNestedTypes().get(5);
-    internal_static_tensorflow_FunctionDef_ControlRetEntry_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_FunctionDef_ControlRetEntry_descriptor,
-        new java.lang.String[] { "Key", "Value", });
-    internal_static_tensorflow_GradientDef_descriptor =
-      getDescriptor().getMessageTypes().get(2);
-    internal_static_tensorflow_GradientDef_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_GradientDef_descriptor,
-        new java.lang.String[] { "FunctionName", "GradientFunc", });
-    org.tensorflow.proto.framework.AttrValueProtos.getDescriptor();
-    org.tensorflow.proto.framework.NodeProto.getDescriptor();
-    org.tensorflow.proto.framework.OpDefProtos.getDescriptor();
-  }
-
-  // @@protoc_insertion_point(outer_class_scope)
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/FunctionSpec.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/FunctionSpec.java
deleted file mode 100644
index 0feaaf41951..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/FunctionSpec.java
+++ /dev/null
@@ -1,961 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/saved_object_graph.proto
-
-package org.tensorflow.proto.framework;
-
-/**
- * <pre>
- * Represents `FunctionSpec` used in `Function`. This represents a
- * function that has been wrapped as a TensorFlow `Function`.
- * </pre>
- *
- * Protobuf type {@code tensorflow.FunctionSpec}
- */
-public  final class FunctionSpec extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.FunctionSpec)
-    FunctionSpecOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use FunctionSpec.newBuilder() to construct.
-  private FunctionSpec(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private FunctionSpec() {
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new FunctionSpec();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private FunctionSpec(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 10: {
-            org.tensorflow.proto.framework.StructuredValue.Builder subBuilder = null;
-            if (fullargspec_ != null) {
-              subBuilder = fullargspec_.toBuilder();
-            }
-            fullargspec_ = input.readMessage(org.tensorflow.proto.framework.StructuredValue.parser(), extensionRegistry);
-            if (subBuilder != null) {
-              subBuilder.mergeFrom(fullargspec_);
-              fullargspec_ = subBuilder.buildPartial();
-            }
-
-            break;
-          }
-          case 16: {
-
-            isMethod_ = input.readBool();
-            break;
-          }
-          case 42: {
-            org.tensorflow.proto.framework.StructuredValue.Builder subBuilder = null;
-            if (inputSignature_ != null) {
-              subBuilder = inputSignature_.toBuilder();
-            }
-            inputSignature_ = input.readMessage(org.tensorflow.proto.framework.StructuredValue.parser(), extensionRegistry);
-            if (subBuilder != null) {
-              subBuilder.mergeFrom(inputSignature_);
-              inputSignature_ = subBuilder.buildPartial();
-            }
-
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.framework.SavedObjectGraphProtos.internal_static_tensorflow_FunctionSpec_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.framework.SavedObjectGraphProtos.internal_static_tensorflow_FunctionSpec_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.framework.FunctionSpec.class, org.tensorflow.proto.framework.FunctionSpec.Builder.class);
-  }
-
-  public static final int FULLARGSPEC_FIELD_NUMBER = 1;
-  private org.tensorflow.proto.framework.StructuredValue fullargspec_;
-  /**
-   * <pre>
-   * Full arg spec from inspect.getfullargspec().
-   * </pre>
-   *
-   * <code>.tensorflow.StructuredValue fullargspec = 1;</code>
-   */
-  public boolean hasFullargspec() {
-    return fullargspec_ != null;
-  }
-  /**
-   * <pre>
-   * Full arg spec from inspect.getfullargspec().
-   * </pre>
-   *
-   * <code>.tensorflow.StructuredValue fullargspec = 1;</code>
-   */
-  public org.tensorflow.proto.framework.StructuredValue getFullargspec() {
-    return fullargspec_ == null ? org.tensorflow.proto.framework.StructuredValue.getDefaultInstance() : fullargspec_;
-  }
-  /**
-   * <pre>
-   * Full arg spec from inspect.getfullargspec().
-   * </pre>
-   *
-   * <code>.tensorflow.StructuredValue fullargspec = 1;</code>
-   */
-  public org.tensorflow.proto.framework.StructuredValueOrBuilder getFullargspecOrBuilder() {
-    return getFullargspec();
-  }
-
-  public static final int IS_METHOD_FIELD_NUMBER = 2;
-  private boolean isMethod_;
-  /**
-   * <pre>
-   * Whether this represents a class method.
-   * </pre>
-   *
-   * <code>bool is_method = 2;</code>
-   */
-  public boolean getIsMethod() {
-    return isMethod_;
-  }
-
-  public static final int INPUT_SIGNATURE_FIELD_NUMBER = 5;
-  private org.tensorflow.proto.framework.StructuredValue inputSignature_;
-  /**
-   * <pre>
-   * The input signature, if specified.
-   * </pre>
-   *
-   * <code>.tensorflow.StructuredValue input_signature = 5;</code>
-   */
-  public boolean hasInputSignature() {
-    return inputSignature_ != null;
-  }
-  /**
-   * <pre>
-   * The input signature, if specified.
-   * </pre>
-   *
-   * <code>.tensorflow.StructuredValue input_signature = 5;</code>
-   */
-  public org.tensorflow.proto.framework.StructuredValue getInputSignature() {
-    return inputSignature_ == null ? org.tensorflow.proto.framework.StructuredValue.getDefaultInstance() : inputSignature_;
-  }
-  /**
-   * <pre>
-   * The input signature, if specified.
-   * </pre>
-   *
-   * <code>.tensorflow.StructuredValue input_signature = 5;</code>
-   */
-  public org.tensorflow.proto.framework.StructuredValueOrBuilder getInputSignatureOrBuilder() {
-    return getInputSignature();
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    if (fullargspec_ != null) {
-      output.writeMessage(1, getFullargspec());
-    }
-    if (isMethod_ != false) {
-      output.writeBool(2, isMethod_);
-    }
-    if (inputSignature_ != null) {
-      output.writeMessage(5, getInputSignature());
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (fullargspec_ != null) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(1, getFullargspec());
-    }
-    if (isMethod_ != false) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeBoolSize(2, isMethod_);
-    }
-    if (inputSignature_ != null) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(5, getInputSignature());
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.framework.FunctionSpec)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.framework.FunctionSpec other = (org.tensorflow.proto.framework.FunctionSpec) obj;
-
-    if (hasFullargspec() != other.hasFullargspec()) return false;
-    if (hasFullargspec()) {
-      if (!getFullargspec()
-          .equals(other.getFullargspec())) return false;
-    }
-    if (getIsMethod()
-        != other.getIsMethod()) return false;
-    if (hasInputSignature() != other.hasInputSignature()) return false;
-    if (hasInputSignature()) {
-      if (!getInputSignature()
-          .equals(other.getInputSignature())) return false;
-    }
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    if (hasFullargspec()) {
-      hash = (37 * hash) + FULLARGSPEC_FIELD_NUMBER;
-      hash = (53 * hash) + getFullargspec().hashCode();
-    }
-    hash = (37 * hash) + IS_METHOD_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashBoolean(
-        getIsMethod());
-    if (hasInputSignature()) {
-      hash = (37 * hash) + INPUT_SIGNATURE_FIELD_NUMBER;
-      hash = (53 * hash) + getInputSignature().hashCode();
-    }
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.framework.FunctionSpec parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.FunctionSpec parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.FunctionSpec parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.FunctionSpec parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.FunctionSpec parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.FunctionSpec parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.FunctionSpec parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.FunctionSpec parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.FunctionSpec parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.FunctionSpec parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.FunctionSpec parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.FunctionSpec parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.framework.FunctionSpec prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * <pre>
-   * Represents `FunctionSpec` used in `Function`. This represents a
-   * function that has been wrapped as a TensorFlow `Function`.
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.FunctionSpec}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.FunctionSpec)
-      org.tensorflow.proto.framework.FunctionSpecOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.SavedObjectGraphProtos.internal_static_tensorflow_FunctionSpec_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.SavedObjectGraphProtos.internal_static_tensorflow_FunctionSpec_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.FunctionSpec.class, org.tensorflow.proto.framework.FunctionSpec.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.framework.FunctionSpec.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      if (fullargspecBuilder_ == null) {
-        fullargspec_ = null;
-      } else {
-        fullargspec_ = null;
-        fullargspecBuilder_ = null;
-      }
-      isMethod_ = false;
-
-      if (inputSignatureBuilder_ == null) {
-        inputSignature_ = null;
-      } else {
-        inputSignature_ = null;
-        inputSignatureBuilder_ = null;
-      }
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.framework.SavedObjectGraphProtos.internal_static_tensorflow_FunctionSpec_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.FunctionSpec getDefaultInstanceForType() {
-      return org.tensorflow.proto.framework.FunctionSpec.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.FunctionSpec build() {
-      org.tensorflow.proto.framework.FunctionSpec result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.FunctionSpec buildPartial() {
-      org.tensorflow.proto.framework.FunctionSpec result = new org.tensorflow.proto.framework.FunctionSpec(this);
-      if (fullargspecBuilder_ == null) {
-        result.fullargspec_ = fullargspec_;
-      } else {
-        result.fullargspec_ = fullargspecBuilder_.build();
-      }
-      result.isMethod_ = isMethod_;
-      if (inputSignatureBuilder_ == null) {
-        result.inputSignature_ = inputSignature_;
-      } else {
-        result.inputSignature_ = inputSignatureBuilder_.build();
-      }
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.framework.FunctionSpec) {
-        return mergeFrom((org.tensorflow.proto.framework.FunctionSpec)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.framework.FunctionSpec other) {
-      if (other == org.tensorflow.proto.framework.FunctionSpec.getDefaultInstance()) return this;
-      if (other.hasFullargspec()) {
-        mergeFullargspec(other.getFullargspec());
-      }
-      if (other.getIsMethod() != false) {
-        setIsMethod(other.getIsMethod());
-      }
-      if (other.hasInputSignature()) {
-        mergeInputSignature(other.getInputSignature());
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.framework.FunctionSpec parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.framework.FunctionSpec) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-
-    private org.tensorflow.proto.framework.StructuredValue fullargspec_;
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.StructuredValue, org.tensorflow.proto.framework.StructuredValue.Builder, org.tensorflow.proto.framework.StructuredValueOrBuilder> fullargspecBuilder_;
-    /**
-     * <pre>
-     * Full arg spec from inspect.getfullargspec().
-     * </pre>
-     *
-     * <code>.tensorflow.StructuredValue fullargspec = 1;</code>
-     */
-    public boolean hasFullargspec() {
-      return fullargspecBuilder_ != null || fullargspec_ != null;
-    }
-    /**
-     * <pre>
-     * Full arg spec from inspect.getfullargspec().
-     * </pre>
-     *
-     * <code>.tensorflow.StructuredValue fullargspec = 1;</code>
-     */
-    public org.tensorflow.proto.framework.StructuredValue getFullargspec() {
-      if (fullargspecBuilder_ == null) {
-        return fullargspec_ == null ? org.tensorflow.proto.framework.StructuredValue.getDefaultInstance() : fullargspec_;
-      } else {
-        return fullargspecBuilder_.getMessage();
-      }
-    }
-    /**
-     * <pre>
-     * Full arg spec from inspect.getfullargspec().
-     * </pre>
-     *
-     * <code>.tensorflow.StructuredValue fullargspec = 1;</code>
-     */
-    public Builder setFullargspec(org.tensorflow.proto.framework.StructuredValue value) {
-      if (fullargspecBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        fullargspec_ = value;
-        onChanged();
-      } else {
-        fullargspecBuilder_.setMessage(value);
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Full arg spec from inspect.getfullargspec().
-     * </pre>
-     *
-     * <code>.tensorflow.StructuredValue fullargspec = 1;</code>
-     */
-    public Builder setFullargspec(
-        org.tensorflow.proto.framework.StructuredValue.Builder builderForValue) {
-      if (fullargspecBuilder_ == null) {
-        fullargspec_ = builderForValue.build();
-        onChanged();
-      } else {
-        fullargspecBuilder_.setMessage(builderForValue.build());
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Full arg spec from inspect.getfullargspec().
-     * </pre>
-     *
-     * <code>.tensorflow.StructuredValue fullargspec = 1;</code>
-     */
-    public Builder mergeFullargspec(org.tensorflow.proto.framework.StructuredValue value) {
-      if (fullargspecBuilder_ == null) {
-        if (fullargspec_ != null) {
-          fullargspec_ =
-            org.tensorflow.proto.framework.StructuredValue.newBuilder(fullargspec_).mergeFrom(value).buildPartial();
-        } else {
-          fullargspec_ = value;
-        }
-        onChanged();
-      } else {
-        fullargspecBuilder_.mergeFrom(value);
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Full arg spec from inspect.getfullargspec().
-     * </pre>
-     *
-     * <code>.tensorflow.StructuredValue fullargspec = 1;</code>
-     */
-    public Builder clearFullargspec() {
-      if (fullargspecBuilder_ == null) {
-        fullargspec_ = null;
-        onChanged();
-      } else {
-        fullargspec_ = null;
-        fullargspecBuilder_ = null;
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Full arg spec from inspect.getfullargspec().
-     * </pre>
-     *
-     * <code>.tensorflow.StructuredValue fullargspec = 1;</code>
-     */
-    public org.tensorflow.proto.framework.StructuredValue.Builder getFullargspecBuilder() {
-      
-      onChanged();
-      return getFullargspecFieldBuilder().getBuilder();
-    }
-    /**
-     * <pre>
-     * Full arg spec from inspect.getfullargspec().
-     * </pre>
-     *
-     * <code>.tensorflow.StructuredValue fullargspec = 1;</code>
-     */
-    public org.tensorflow.proto.framework.StructuredValueOrBuilder getFullargspecOrBuilder() {
-      if (fullargspecBuilder_ != null) {
-        return fullargspecBuilder_.getMessageOrBuilder();
-      } else {
-        return fullargspec_ == null ?
-            org.tensorflow.proto.framework.StructuredValue.getDefaultInstance() : fullargspec_;
-      }
-    }
-    /**
-     * <pre>
-     * Full arg spec from inspect.getfullargspec().
-     * </pre>
-     *
-     * <code>.tensorflow.StructuredValue fullargspec = 1;</code>
-     */
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.StructuredValue, org.tensorflow.proto.framework.StructuredValue.Builder, org.tensorflow.proto.framework.StructuredValueOrBuilder> 
-        getFullargspecFieldBuilder() {
-      if (fullargspecBuilder_ == null) {
-        fullargspecBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-            org.tensorflow.proto.framework.StructuredValue, org.tensorflow.proto.framework.StructuredValue.Builder, org.tensorflow.proto.framework.StructuredValueOrBuilder>(
-                getFullargspec(),
-                getParentForChildren(),
-                isClean());
-        fullargspec_ = null;
-      }
-      return fullargspecBuilder_;
-    }
-
-    private boolean isMethod_ ;
-    /**
-     * <pre>
-     * Whether this represents a class method.
-     * </pre>
-     *
-     * <code>bool is_method = 2;</code>
-     */
-    public boolean getIsMethod() {
-      return isMethod_;
-    }
-    /**
-     * <pre>
-     * Whether this represents a class method.
-     * </pre>
-     *
-     * <code>bool is_method = 2;</code>
-     */
-    public Builder setIsMethod(boolean value) {
-      
-      isMethod_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Whether this represents a class method.
-     * </pre>
-     *
-     * <code>bool is_method = 2;</code>
-     */
-    public Builder clearIsMethod() {
-      
-      isMethod_ = false;
-      onChanged();
-      return this;
-    }
-
-    private org.tensorflow.proto.framework.StructuredValue inputSignature_;
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.StructuredValue, org.tensorflow.proto.framework.StructuredValue.Builder, org.tensorflow.proto.framework.StructuredValueOrBuilder> inputSignatureBuilder_;
-    /**
-     * <pre>
-     * The input signature, if specified.
-     * </pre>
-     *
-     * <code>.tensorflow.StructuredValue input_signature = 5;</code>
-     */
-    public boolean hasInputSignature() {
-      return inputSignatureBuilder_ != null || inputSignature_ != null;
-    }
-    /**
-     * <pre>
-     * The input signature, if specified.
-     * </pre>
-     *
-     * <code>.tensorflow.StructuredValue input_signature = 5;</code>
-     */
-    public org.tensorflow.proto.framework.StructuredValue getInputSignature() {
-      if (inputSignatureBuilder_ == null) {
-        return inputSignature_ == null ? org.tensorflow.proto.framework.StructuredValue.getDefaultInstance() : inputSignature_;
-      } else {
-        return inputSignatureBuilder_.getMessage();
-      }
-    }
-    /**
-     * <pre>
-     * The input signature, if specified.
-     * </pre>
-     *
-     * <code>.tensorflow.StructuredValue input_signature = 5;</code>
-     */
-    public Builder setInputSignature(org.tensorflow.proto.framework.StructuredValue value) {
-      if (inputSignatureBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        inputSignature_ = value;
-        onChanged();
-      } else {
-        inputSignatureBuilder_.setMessage(value);
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * The input signature, if specified.
-     * </pre>
-     *
-     * <code>.tensorflow.StructuredValue input_signature = 5;</code>
-     */
-    public Builder setInputSignature(
-        org.tensorflow.proto.framework.StructuredValue.Builder builderForValue) {
-      if (inputSignatureBuilder_ == null) {
-        inputSignature_ = builderForValue.build();
-        onChanged();
-      } else {
-        inputSignatureBuilder_.setMessage(builderForValue.build());
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * The input signature, if specified.
-     * </pre>
-     *
-     * <code>.tensorflow.StructuredValue input_signature = 5;</code>
-     */
-    public Builder mergeInputSignature(org.tensorflow.proto.framework.StructuredValue value) {
-      if (inputSignatureBuilder_ == null) {
-        if (inputSignature_ != null) {
-          inputSignature_ =
-            org.tensorflow.proto.framework.StructuredValue.newBuilder(inputSignature_).mergeFrom(value).buildPartial();
-        } else {
-          inputSignature_ = value;
-        }
-        onChanged();
-      } else {
-        inputSignatureBuilder_.mergeFrom(value);
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * The input signature, if specified.
-     * </pre>
-     *
-     * <code>.tensorflow.StructuredValue input_signature = 5;</code>
-     */
-    public Builder clearInputSignature() {
-      if (inputSignatureBuilder_ == null) {
-        inputSignature_ = null;
-        onChanged();
-      } else {
-        inputSignature_ = null;
-        inputSignatureBuilder_ = null;
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * The input signature, if specified.
-     * </pre>
-     *
-     * <code>.tensorflow.StructuredValue input_signature = 5;</code>
-     */
-    public org.tensorflow.proto.framework.StructuredValue.Builder getInputSignatureBuilder() {
-      
-      onChanged();
-      return getInputSignatureFieldBuilder().getBuilder();
-    }
-    /**
-     * <pre>
-     * The input signature, if specified.
-     * </pre>
-     *
-     * <code>.tensorflow.StructuredValue input_signature = 5;</code>
-     */
-    public org.tensorflow.proto.framework.StructuredValueOrBuilder getInputSignatureOrBuilder() {
-      if (inputSignatureBuilder_ != null) {
-        return inputSignatureBuilder_.getMessageOrBuilder();
-      } else {
-        return inputSignature_ == null ?
-            org.tensorflow.proto.framework.StructuredValue.getDefaultInstance() : inputSignature_;
-      }
-    }
-    /**
-     * <pre>
-     * The input signature, if specified.
-     * </pre>
-     *
-     * <code>.tensorflow.StructuredValue input_signature = 5;</code>
-     */
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.StructuredValue, org.tensorflow.proto.framework.StructuredValue.Builder, org.tensorflow.proto.framework.StructuredValueOrBuilder> 
-        getInputSignatureFieldBuilder() {
-      if (inputSignatureBuilder_ == null) {
-        inputSignatureBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-            org.tensorflow.proto.framework.StructuredValue, org.tensorflow.proto.framework.StructuredValue.Builder, org.tensorflow.proto.framework.StructuredValueOrBuilder>(
-                getInputSignature(),
-                getParentForChildren(),
-                isClean());
-        inputSignature_ = null;
-      }
-      return inputSignatureBuilder_;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.FunctionSpec)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.FunctionSpec)
-  private static final org.tensorflow.proto.framework.FunctionSpec DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.FunctionSpec();
-  }
-
-  public static org.tensorflow.proto.framework.FunctionSpec getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<FunctionSpec>
-      PARSER = new com.google.protobuf.AbstractParser<FunctionSpec>() {
-    @java.lang.Override
-    public FunctionSpec parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new FunctionSpec(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<FunctionSpec> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<FunctionSpec> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.framework.FunctionSpec getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/FunctionSpecOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/FunctionSpecOrBuilder.java
deleted file mode 100644
index 8c200ad6323..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/FunctionSpecOrBuilder.java
+++ /dev/null
@@ -1,68 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/saved_object_graph.proto
-
-package org.tensorflow.proto.framework;
-
-public interface FunctionSpecOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.FunctionSpec)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <pre>
-   * Full arg spec from inspect.getfullargspec().
-   * </pre>
-   *
-   * <code>.tensorflow.StructuredValue fullargspec = 1;</code>
-   */
-  boolean hasFullargspec();
-  /**
-   * <pre>
-   * Full arg spec from inspect.getfullargspec().
-   * </pre>
-   *
-   * <code>.tensorflow.StructuredValue fullargspec = 1;</code>
-   */
-  org.tensorflow.proto.framework.StructuredValue getFullargspec();
-  /**
-   * <pre>
-   * Full arg spec from inspect.getfullargspec().
-   * </pre>
-   *
-   * <code>.tensorflow.StructuredValue fullargspec = 1;</code>
-   */
-  org.tensorflow.proto.framework.StructuredValueOrBuilder getFullargspecOrBuilder();
-
-  /**
-   * <pre>
-   * Whether this represents a class method.
-   * </pre>
-   *
-   * <code>bool is_method = 2;</code>
-   */
-  boolean getIsMethod();
-
-  /**
-   * <pre>
-   * The input signature, if specified.
-   * </pre>
-   *
-   * <code>.tensorflow.StructuredValue input_signature = 5;</code>
-   */
-  boolean hasInputSignature();
-  /**
-   * <pre>
-   * The input signature, if specified.
-   * </pre>
-   *
-   * <code>.tensorflow.StructuredValue input_signature = 5;</code>
-   */
-  org.tensorflow.proto.framework.StructuredValue getInputSignature();
-  /**
-   * <pre>
-   * The input signature, if specified.
-   * </pre>
-   *
-   * <code>.tensorflow.StructuredValue input_signature = 5;</code>
-   */
-  org.tensorflow.proto.framework.StructuredValueOrBuilder getInputSignatureOrBuilder();
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GPUOptions.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GPUOptions.java
deleted file mode 100644
index 44b8a1c43e3..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GPUOptions.java
+++ /dev/null
@@ -1,5061 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/config.proto
-
-package org.tensorflow.proto.framework;
-
-/**
- * Protobuf type {@code tensorflow.GPUOptions}
- */
-public  final class GPUOptions extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.GPUOptions)
-    GPUOptionsOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use GPUOptions.newBuilder() to construct.
-  private GPUOptions(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private GPUOptions() {
-    allocatorType_ = "";
-    visibleDeviceList_ = "";
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new GPUOptions();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private GPUOptions(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 9: {
-
-            perProcessGpuMemoryFraction_ = input.readDouble();
-            break;
-          }
-          case 18: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            allocatorType_ = s;
-            break;
-          }
-          case 24: {
-
-            deferredDeletionBytes_ = input.readInt64();
-            break;
-          }
-          case 32: {
-
-            allowGrowth_ = input.readBool();
-            break;
-          }
-          case 42: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            visibleDeviceList_ = s;
-            break;
-          }
-          case 48: {
-
-            pollingActiveDelayUsecs_ = input.readInt32();
-            break;
-          }
-          case 56: {
-
-            pollingInactiveDelayMsecs_ = input.readInt32();
-            break;
-          }
-          case 64: {
-
-            forceGpuCompatible_ = input.readBool();
-            break;
-          }
-          case 74: {
-            org.tensorflow.proto.framework.GPUOptions.Experimental.Builder subBuilder = null;
-            if (experimental_ != null) {
-              subBuilder = experimental_.toBuilder();
-            }
-            experimental_ = input.readMessage(org.tensorflow.proto.framework.GPUOptions.Experimental.parser(), extensionRegistry);
-            if (subBuilder != null) {
-              subBuilder.mergeFrom(experimental_);
-              experimental_ = subBuilder.buildPartial();
-            }
-
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_GPUOptions_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_GPUOptions_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.framework.GPUOptions.class, org.tensorflow.proto.framework.GPUOptions.Builder.class);
-  }
-
-  public interface ExperimentalOrBuilder extends
-      // @@protoc_insertion_point(interface_extends:tensorflow.GPUOptions.Experimental)
-      com.google.protobuf.MessageOrBuilder {
-
-    /**
-     * <pre>
-     * The multi virtual device settings. If empty (not set), it will create
-     * single virtual device on each visible GPU, according to the settings
-     * in "visible_device_list" above. Otherwise, the number of elements in the
-     * list must be the same as the number of visible GPUs (after
-     * "visible_device_list" filtering if it is set), and the string represented
-     * device names (e.g. /device:GPU:&lt;id&gt;) will refer to the virtual
-     * devices and have the &lt;id&gt; field assigned sequentially starting from 0,
-     * according to the order they appear in this list and the "memory_limit"
-     * list inside each element. For example,
-     *   visible_device_list = "1,0"
-     *   virtual_devices { memory_limit: 1GB memory_limit: 2GB }
-     *   virtual_devices {}
-     * will create three virtual devices as:
-     *   /device:GPU:0 -&gt; visible GPU 1 with 1GB memory
-     *   /device:GPU:1 -&gt; visible GPU 1 with 2GB memory
-     *   /device:GPU:2 -&gt; visible GPU 0 with all available memory
-     * NOTE:
-     * 1. It's invalid to set both this and "per_process_gpu_memory_fraction"
-     *    at the same time.
-     * 2. Currently this setting is per-process, not per-session. Using
-     *    different settings in different sessions within same process will
-     *    result in undefined behavior.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.GPUOptions.Experimental.VirtualDevices virtual_devices = 1;</code>
-     */
-    java.util.List<org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevices> 
-        getVirtualDevicesList();
-    /**
-     * <pre>
-     * The multi virtual device settings. If empty (not set), it will create
-     * single virtual device on each visible GPU, according to the settings
-     * in "visible_device_list" above. Otherwise, the number of elements in the
-     * list must be the same as the number of visible GPUs (after
-     * "visible_device_list" filtering if it is set), and the string represented
-     * device names (e.g. /device:GPU:&lt;id&gt;) will refer to the virtual
-     * devices and have the &lt;id&gt; field assigned sequentially starting from 0,
-     * according to the order they appear in this list and the "memory_limit"
-     * list inside each element. For example,
-     *   visible_device_list = "1,0"
-     *   virtual_devices { memory_limit: 1GB memory_limit: 2GB }
-     *   virtual_devices {}
-     * will create three virtual devices as:
-     *   /device:GPU:0 -&gt; visible GPU 1 with 1GB memory
-     *   /device:GPU:1 -&gt; visible GPU 1 with 2GB memory
-     *   /device:GPU:2 -&gt; visible GPU 0 with all available memory
-     * NOTE:
-     * 1. It's invalid to set both this and "per_process_gpu_memory_fraction"
-     *    at the same time.
-     * 2. Currently this setting is per-process, not per-session. Using
-     *    different settings in different sessions within same process will
-     *    result in undefined behavior.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.GPUOptions.Experimental.VirtualDevices virtual_devices = 1;</code>
-     */
-    org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevices getVirtualDevices(int index);
-    /**
-     * <pre>
-     * The multi virtual device settings. If empty (not set), it will create
-     * single virtual device on each visible GPU, according to the settings
-     * in "visible_device_list" above. Otherwise, the number of elements in the
-     * list must be the same as the number of visible GPUs (after
-     * "visible_device_list" filtering if it is set), and the string represented
-     * device names (e.g. /device:GPU:&lt;id&gt;) will refer to the virtual
-     * devices and have the &lt;id&gt; field assigned sequentially starting from 0,
-     * according to the order they appear in this list and the "memory_limit"
-     * list inside each element. For example,
-     *   visible_device_list = "1,0"
-     *   virtual_devices { memory_limit: 1GB memory_limit: 2GB }
-     *   virtual_devices {}
-     * will create three virtual devices as:
-     *   /device:GPU:0 -&gt; visible GPU 1 with 1GB memory
-     *   /device:GPU:1 -&gt; visible GPU 1 with 2GB memory
-     *   /device:GPU:2 -&gt; visible GPU 0 with all available memory
-     * NOTE:
-     * 1. It's invalid to set both this and "per_process_gpu_memory_fraction"
-     *    at the same time.
-     * 2. Currently this setting is per-process, not per-session. Using
-     *    different settings in different sessions within same process will
-     *    result in undefined behavior.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.GPUOptions.Experimental.VirtualDevices virtual_devices = 1;</code>
-     */
-    int getVirtualDevicesCount();
-    /**
-     * <pre>
-     * The multi virtual device settings. If empty (not set), it will create
-     * single virtual device on each visible GPU, according to the settings
-     * in "visible_device_list" above. Otherwise, the number of elements in the
-     * list must be the same as the number of visible GPUs (after
-     * "visible_device_list" filtering if it is set), and the string represented
-     * device names (e.g. /device:GPU:&lt;id&gt;) will refer to the virtual
-     * devices and have the &lt;id&gt; field assigned sequentially starting from 0,
-     * according to the order they appear in this list and the "memory_limit"
-     * list inside each element. For example,
-     *   visible_device_list = "1,0"
-     *   virtual_devices { memory_limit: 1GB memory_limit: 2GB }
-     *   virtual_devices {}
-     * will create three virtual devices as:
-     *   /device:GPU:0 -&gt; visible GPU 1 with 1GB memory
-     *   /device:GPU:1 -&gt; visible GPU 1 with 2GB memory
-     *   /device:GPU:2 -&gt; visible GPU 0 with all available memory
-     * NOTE:
-     * 1. It's invalid to set both this and "per_process_gpu_memory_fraction"
-     *    at the same time.
-     * 2. Currently this setting is per-process, not per-session. Using
-     *    different settings in different sessions within same process will
-     *    result in undefined behavior.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.GPUOptions.Experimental.VirtualDevices virtual_devices = 1;</code>
-     */
-    java.util.List<? extends org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevicesOrBuilder> 
-        getVirtualDevicesOrBuilderList();
-    /**
-     * <pre>
-     * The multi virtual device settings. If empty (not set), it will create
-     * single virtual device on each visible GPU, according to the settings
-     * in "visible_device_list" above. Otherwise, the number of elements in the
-     * list must be the same as the number of visible GPUs (after
-     * "visible_device_list" filtering if it is set), and the string represented
-     * device names (e.g. /device:GPU:&lt;id&gt;) will refer to the virtual
-     * devices and have the &lt;id&gt; field assigned sequentially starting from 0,
-     * according to the order they appear in this list and the "memory_limit"
-     * list inside each element. For example,
-     *   visible_device_list = "1,0"
-     *   virtual_devices { memory_limit: 1GB memory_limit: 2GB }
-     *   virtual_devices {}
-     * will create three virtual devices as:
-     *   /device:GPU:0 -&gt; visible GPU 1 with 1GB memory
-     *   /device:GPU:1 -&gt; visible GPU 1 with 2GB memory
-     *   /device:GPU:2 -&gt; visible GPU 0 with all available memory
-     * NOTE:
-     * 1. It's invalid to set both this and "per_process_gpu_memory_fraction"
-     *    at the same time.
-     * 2. Currently this setting is per-process, not per-session. Using
-     *    different settings in different sessions within same process will
-     *    result in undefined behavior.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.GPUOptions.Experimental.VirtualDevices virtual_devices = 1;</code>
-     */
-    org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevicesOrBuilder getVirtualDevicesOrBuilder(
-        int index);
-
-    /**
-     * <pre>
-     * If true, uses CUDA unified memory for memory allocations. If
-     * per_process_gpu_memory_fraction option is greater than 1.0, then unified
-     * memory is used regardless of the value for this field. See comments for
-     * per_process_gpu_memory_fraction field for more details and requirements
-     * of the unified memory. This option is useful to oversubscribe memory if
-     * multiple processes are sharing a single GPU while individually using less
-     * than 1.0 per process memory fraction.
-     * </pre>
-     *
-     * <code>bool use_unified_memory = 2;</code>
-     */
-    boolean getUseUnifiedMemory();
-
-    /**
-     * <pre>
-     * If &gt; 1, the number of device-to-device copy streams to create
-     * for each GPUDevice.  Default value is 0, which is automatically
-     * converted to 1.
-     * </pre>
-     *
-     * <code>int32 num_dev_to_dev_copy_streams = 3;</code>
-     */
-    int getNumDevToDevCopyStreams();
-
-    /**
-     * <pre>
-     * If non-empty, defines a good GPU ring order on a single worker based on
-     * device interconnect.  This assumes that all workers have the same GPU
-     * topology.  Specify as a comma-separated string, e.g. "3,2,1,0,7,6,5,4".
-     * This ring order is used by the RingReducer implementation of
-     * CollectiveReduce, and serves as an override to automatic ring order
-     * generation in OrderTaskDeviceMap() during CollectiveParam resolution.
-     * </pre>
-     *
-     * <code>string collective_ring_order = 4;</code>
-     */
-    java.lang.String getCollectiveRingOrder();
-    /**
-     * <pre>
-     * If non-empty, defines a good GPU ring order on a single worker based on
-     * device interconnect.  This assumes that all workers have the same GPU
-     * topology.  Specify as a comma-separated string, e.g. "3,2,1,0,7,6,5,4".
-     * This ring order is used by the RingReducer implementation of
-     * CollectiveReduce, and serves as an override to automatic ring order
-     * generation in OrderTaskDeviceMap() during CollectiveParam resolution.
-     * </pre>
-     *
-     * <code>string collective_ring_order = 4;</code>
-     */
-    com.google.protobuf.ByteString
-        getCollectiveRingOrderBytes();
-
-    /**
-     * <pre>
-     * If true then extra work is done by GPUDevice and GPUBFCAllocator to
-     * keep track of when GPU memory is freed and when kernels actually
-     * complete so that we can know when a nominally free memory chunk
-     * is really not subject to pending use.
-     * </pre>
-     *
-     * <code>bool timestamped_allocator = 5;</code>
-     */
-    boolean getTimestampedAllocator();
-
-    /**
-     * <pre>
-     * Parameters for GPUKernelTracker.  By default no kernel tracking is done.
-     * Note that timestamped_allocator is only effective if some tracking is
-     * specified.
-     * If kernel_tracker_max_interval = n &gt; 0, then a tracking event
-     * is inserted after every n kernels without an event.
-     * </pre>
-     *
-     * <code>int32 kernel_tracker_max_interval = 7;</code>
-     */
-    int getKernelTrackerMaxInterval();
-
-    /**
-     * <pre>
-     * If kernel_tracker_max_bytes = n &gt; 0, then a tracking event is
-     * inserted after every series of kernels allocating a sum of
-     * memory &gt;= n.  If one kernel allocates b * n bytes, then one
-     * event will be inserted after it, but it will count as b against
-     * the pending limit.
-     * </pre>
-     *
-     * <code>int32 kernel_tracker_max_bytes = 8;</code>
-     */
-    int getKernelTrackerMaxBytes();
-
-    /**
-     * <pre>
-     * If kernel_tracker_max_pending &gt; 0 then no more than this many
-     * tracking events can be outstanding at a time.  An attempt to
-     * launch an additional kernel will stall until an event
-     * completes.
-     * </pre>
-     *
-     * <code>int32 kernel_tracker_max_pending = 9;</code>
-     */
-    int getKernelTrackerMaxPending();
-  }
-  /**
-   * Protobuf type {@code tensorflow.GPUOptions.Experimental}
-   */
-  public  static final class Experimental extends
-      com.google.protobuf.GeneratedMessageV3 implements
-      // @@protoc_insertion_point(message_implements:tensorflow.GPUOptions.Experimental)
-      ExperimentalOrBuilder {
-  private static final long serialVersionUID = 0L;
-    // Use Experimental.newBuilder() to construct.
-    private Experimental(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-      super(builder);
-    }
-    private Experimental() {
-      virtualDevices_ = java.util.Collections.emptyList();
-      collectiveRingOrder_ = "";
-    }
-
-    @java.lang.Override
-    @SuppressWarnings({"unused"})
-    protected java.lang.Object newInstance(
-        UnusedPrivateParameter unused) {
-      return new Experimental();
-    }
-
-    @java.lang.Override
-    public final com.google.protobuf.UnknownFieldSet
-    getUnknownFields() {
-      return this.unknownFields;
-    }
-    private Experimental(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      this();
-      if (extensionRegistry == null) {
-        throw new java.lang.NullPointerException();
-      }
-      int mutable_bitField0_ = 0;
-      com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-          com.google.protobuf.UnknownFieldSet.newBuilder();
-      try {
-        boolean done = false;
-        while (!done) {
-          int tag = input.readTag();
-          switch (tag) {
-            case 0:
-              done = true;
-              break;
-            case 10: {
-              if (!((mutable_bitField0_ & 0x00000001) != 0)) {
-                virtualDevices_ = new java.util.ArrayList<org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevices>();
-                mutable_bitField0_ |= 0x00000001;
-              }
-              virtualDevices_.add(
-                  input.readMessage(org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevices.parser(), extensionRegistry));
-              break;
-            }
-            case 16: {
-
-              useUnifiedMemory_ = input.readBool();
-              break;
-            }
-            case 24: {
-
-              numDevToDevCopyStreams_ = input.readInt32();
-              break;
-            }
-            case 34: {
-              java.lang.String s = input.readStringRequireUtf8();
-
-              collectiveRingOrder_ = s;
-              break;
-            }
-            case 40: {
-
-              timestampedAllocator_ = input.readBool();
-              break;
-            }
-            case 56: {
-
-              kernelTrackerMaxInterval_ = input.readInt32();
-              break;
-            }
-            case 64: {
-
-              kernelTrackerMaxBytes_ = input.readInt32();
-              break;
-            }
-            case 72: {
-
-              kernelTrackerMaxPending_ = input.readInt32();
-              break;
-            }
-            default: {
-              if (!parseUnknownField(
-                  input, unknownFields, extensionRegistry, tag)) {
-                done = true;
-              }
-              break;
-            }
-          }
-        }
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        throw e.setUnfinishedMessage(this);
-      } catch (java.io.IOException e) {
-        throw new com.google.protobuf.InvalidProtocolBufferException(
-            e).setUnfinishedMessage(this);
-      } finally {
-        if (((mutable_bitField0_ & 0x00000001) != 0)) {
-          virtualDevices_ = java.util.Collections.unmodifiableList(virtualDevices_);
-        }
-        this.unknownFields = unknownFields.build();
-        makeExtensionsImmutable();
-      }
-    }
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_GPUOptions_Experimental_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_GPUOptions_Experimental_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.GPUOptions.Experimental.class, org.tensorflow.proto.framework.GPUOptions.Experimental.Builder.class);
-    }
-
-    public interface VirtualDevicesOrBuilder extends
-        // @@protoc_insertion_point(interface_extends:tensorflow.GPUOptions.Experimental.VirtualDevices)
-        com.google.protobuf.MessageOrBuilder {
-
-      /**
-       * <pre>
-       * Per "virtual" device memory limit, in MB. The number of elements in
-       * the list is the number of virtual devices to create on the
-       * corresponding visible GPU (see "virtual_devices" below).
-       * If empty, it will create single virtual device taking all available
-       * memory from the device.
-       * For the concept of "visible" and "virtual" GPU, see the comments for
-       * "visible_device_list" above for more information.
-       * </pre>
-       *
-       * <code>repeated float memory_limit_mb = 1;</code>
-       */
-      java.util.List<java.lang.Float> getMemoryLimitMbList();
-      /**
-       * <pre>
-       * Per "virtual" device memory limit, in MB. The number of elements in
-       * the list is the number of virtual devices to create on the
-       * corresponding visible GPU (see "virtual_devices" below).
-       * If empty, it will create single virtual device taking all available
-       * memory from the device.
-       * For the concept of "visible" and "virtual" GPU, see the comments for
-       * "visible_device_list" above for more information.
-       * </pre>
-       *
-       * <code>repeated float memory_limit_mb = 1;</code>
-       */
-      int getMemoryLimitMbCount();
-      /**
-       * <pre>
-       * Per "virtual" device memory limit, in MB. The number of elements in
-       * the list is the number of virtual devices to create on the
-       * corresponding visible GPU (see "virtual_devices" below).
-       * If empty, it will create single virtual device taking all available
-       * memory from the device.
-       * For the concept of "visible" and "virtual" GPU, see the comments for
-       * "visible_device_list" above for more information.
-       * </pre>
-       *
-       * <code>repeated float memory_limit_mb = 1;</code>
-       */
-      float getMemoryLimitMb(int index);
-
-      /**
-       * <pre>
-       * Priority values to use with the virtual devices. Use the cuda function
-       * cudaDeviceGetStreamPriorityRange to query for valid range of values for
-       * priority.
-       * On a P4000 GPU with cuda 10.1, the priority range reported was 0 for
-       * least priority and -1 for greatest priority.
-       * If this field is not specified, then the virtual devices will be
-       * created with the default. If this field has values set, then the size
-       * of this must match with the above memory_limit_mb.
-       * </pre>
-       *
-       * <code>repeated int32 priority = 2;</code>
-       */
-      java.util.List<java.lang.Integer> getPriorityList();
-      /**
-       * <pre>
-       * Priority values to use with the virtual devices. Use the cuda function
-       * cudaDeviceGetStreamPriorityRange to query for valid range of values for
-       * priority.
-       * On a P4000 GPU with cuda 10.1, the priority range reported was 0 for
-       * least priority and -1 for greatest priority.
-       * If this field is not specified, then the virtual devices will be
-       * created with the default. If this field has values set, then the size
-       * of this must match with the above memory_limit_mb.
-       * </pre>
-       *
-       * <code>repeated int32 priority = 2;</code>
-       */
-      int getPriorityCount();
-      /**
-       * <pre>
-       * Priority values to use with the virtual devices. Use the cuda function
-       * cudaDeviceGetStreamPriorityRange to query for valid range of values for
-       * priority.
-       * On a P4000 GPU with cuda 10.1, the priority range reported was 0 for
-       * least priority and -1 for greatest priority.
-       * If this field is not specified, then the virtual devices will be
-       * created with the default. If this field has values set, then the size
-       * of this must match with the above memory_limit_mb.
-       * </pre>
-       *
-       * <code>repeated int32 priority = 2;</code>
-       */
-      int getPriority(int index);
-    }
-    /**
-     * <pre>
-     * Configuration for breaking down a visible GPU into multiple "virtual"
-     * devices.
-     * </pre>
-     *
-     * Protobuf type {@code tensorflow.GPUOptions.Experimental.VirtualDevices}
-     */
-    public  static final class VirtualDevices extends
-        com.google.protobuf.GeneratedMessageV3 implements
-        // @@protoc_insertion_point(message_implements:tensorflow.GPUOptions.Experimental.VirtualDevices)
-        VirtualDevicesOrBuilder {
-    private static final long serialVersionUID = 0L;
-      // Use VirtualDevices.newBuilder() to construct.
-      private VirtualDevices(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-        super(builder);
-      }
-      private VirtualDevices() {
-        memoryLimitMb_ = emptyFloatList();
-        priority_ = emptyIntList();
-      }
-
-      @java.lang.Override
-      @SuppressWarnings({"unused"})
-      protected java.lang.Object newInstance(
-          UnusedPrivateParameter unused) {
-        return new VirtualDevices();
-      }
-
-      @java.lang.Override
-      public final com.google.protobuf.UnknownFieldSet
-      getUnknownFields() {
-        return this.unknownFields;
-      }
-      private VirtualDevices(
-          com.google.protobuf.CodedInputStream input,
-          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-          throws com.google.protobuf.InvalidProtocolBufferException {
-        this();
-        if (extensionRegistry == null) {
-          throw new java.lang.NullPointerException();
-        }
-        int mutable_bitField0_ = 0;
-        com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-            com.google.protobuf.UnknownFieldSet.newBuilder();
-        try {
-          boolean done = false;
-          while (!done) {
-            int tag = input.readTag();
-            switch (tag) {
-              case 0:
-                done = true;
-                break;
-              case 13: {
-                if (!((mutable_bitField0_ & 0x00000001) != 0)) {
-                  memoryLimitMb_ = newFloatList();
-                  mutable_bitField0_ |= 0x00000001;
-                }
-                memoryLimitMb_.addFloat(input.readFloat());
-                break;
-              }
-              case 10: {
-                int length = input.readRawVarint32();
-                int limit = input.pushLimit(length);
-                if (!((mutable_bitField0_ & 0x00000001) != 0) && input.getBytesUntilLimit() > 0) {
-                  memoryLimitMb_ = newFloatList();
-                  mutable_bitField0_ |= 0x00000001;
-                }
-                while (input.getBytesUntilLimit() > 0) {
-                  memoryLimitMb_.addFloat(input.readFloat());
-                }
-                input.popLimit(limit);
-                break;
-              }
-              case 16: {
-                if (!((mutable_bitField0_ & 0x00000002) != 0)) {
-                  priority_ = newIntList();
-                  mutable_bitField0_ |= 0x00000002;
-                }
-                priority_.addInt(input.readInt32());
-                break;
-              }
-              case 18: {
-                int length = input.readRawVarint32();
-                int limit = input.pushLimit(length);
-                if (!((mutable_bitField0_ & 0x00000002) != 0) && input.getBytesUntilLimit() > 0) {
-                  priority_ = newIntList();
-                  mutable_bitField0_ |= 0x00000002;
-                }
-                while (input.getBytesUntilLimit() > 0) {
-                  priority_.addInt(input.readInt32());
-                }
-                input.popLimit(limit);
-                break;
-              }
-              default: {
-                if (!parseUnknownField(
-                    input, unknownFields, extensionRegistry, tag)) {
-                  done = true;
-                }
-                break;
-              }
-            }
-          }
-        } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-          throw e.setUnfinishedMessage(this);
-        } catch (java.io.IOException e) {
-          throw new com.google.protobuf.InvalidProtocolBufferException(
-              e).setUnfinishedMessage(this);
-        } finally {
-          if (((mutable_bitField0_ & 0x00000001) != 0)) {
-            memoryLimitMb_.makeImmutable(); // C
-          }
-          if (((mutable_bitField0_ & 0x00000002) != 0)) {
-            priority_.makeImmutable(); // C
-          }
-          this.unknownFields = unknownFields.build();
-          makeExtensionsImmutable();
-        }
-      }
-      public static final com.google.protobuf.Descriptors.Descriptor
-          getDescriptor() {
-        return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_GPUOptions_Experimental_VirtualDevices_descriptor;
-      }
-
-      @java.lang.Override
-      protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-          internalGetFieldAccessorTable() {
-        return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_GPUOptions_Experimental_VirtualDevices_fieldAccessorTable
-            .ensureFieldAccessorsInitialized(
-                org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevices.class, org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevices.Builder.class);
-      }
-
-      public static final int MEMORY_LIMIT_MB_FIELD_NUMBER = 1;
-      private com.google.protobuf.Internal.FloatList memoryLimitMb_;
-      /**
-       * <pre>
-       * Per "virtual" device memory limit, in MB. The number of elements in
-       * the list is the number of virtual devices to create on the
-       * corresponding visible GPU (see "virtual_devices" below).
-       * If empty, it will create single virtual device taking all available
-       * memory from the device.
-       * For the concept of "visible" and "virtual" GPU, see the comments for
-       * "visible_device_list" above for more information.
-       * </pre>
-       *
-       * <code>repeated float memory_limit_mb = 1;</code>
-       */
-      public java.util.List<java.lang.Float>
-          getMemoryLimitMbList() {
-        return memoryLimitMb_;
-      }
-      /**
-       * <pre>
-       * Per "virtual" device memory limit, in MB. The number of elements in
-       * the list is the number of virtual devices to create on the
-       * corresponding visible GPU (see "virtual_devices" below).
-       * If empty, it will create single virtual device taking all available
-       * memory from the device.
-       * For the concept of "visible" and "virtual" GPU, see the comments for
-       * "visible_device_list" above for more information.
-       * </pre>
-       *
-       * <code>repeated float memory_limit_mb = 1;</code>
-       */
-      public int getMemoryLimitMbCount() {
-        return memoryLimitMb_.size();
-      }
-      /**
-       * <pre>
-       * Per "virtual" device memory limit, in MB. The number of elements in
-       * the list is the number of virtual devices to create on the
-       * corresponding visible GPU (see "virtual_devices" below).
-       * If empty, it will create single virtual device taking all available
-       * memory from the device.
-       * For the concept of "visible" and "virtual" GPU, see the comments for
-       * "visible_device_list" above for more information.
-       * </pre>
-       *
-       * <code>repeated float memory_limit_mb = 1;</code>
-       */
-      public float getMemoryLimitMb(int index) {
-        return memoryLimitMb_.getFloat(index);
-      }
-      private int memoryLimitMbMemoizedSerializedSize = -1;
-
-      public static final int PRIORITY_FIELD_NUMBER = 2;
-      private com.google.protobuf.Internal.IntList priority_;
-      /**
-       * <pre>
-       * Priority values to use with the virtual devices. Use the cuda function
-       * cudaDeviceGetStreamPriorityRange to query for valid range of values for
-       * priority.
-       * On a P4000 GPU with cuda 10.1, the priority range reported was 0 for
-       * least priority and -1 for greatest priority.
-       * If this field is not specified, then the virtual devices will be
-       * created with the default. If this field has values set, then the size
-       * of this must match with the above memory_limit_mb.
-       * </pre>
-       *
-       * <code>repeated int32 priority = 2;</code>
-       */
-      public java.util.List<java.lang.Integer>
-          getPriorityList() {
-        return priority_;
-      }
-      /**
-       * <pre>
-       * Priority values to use with the virtual devices. Use the cuda function
-       * cudaDeviceGetStreamPriorityRange to query for valid range of values for
-       * priority.
-       * On a P4000 GPU with cuda 10.1, the priority range reported was 0 for
-       * least priority and -1 for greatest priority.
-       * If this field is not specified, then the virtual devices will be
-       * created with the default. If this field has values set, then the size
-       * of this must match with the above memory_limit_mb.
-       * </pre>
-       *
-       * <code>repeated int32 priority = 2;</code>
-       */
-      public int getPriorityCount() {
-        return priority_.size();
-      }
-      /**
-       * <pre>
-       * Priority values to use with the virtual devices. Use the cuda function
-       * cudaDeviceGetStreamPriorityRange to query for valid range of values for
-       * priority.
-       * On a P4000 GPU with cuda 10.1, the priority range reported was 0 for
-       * least priority and -1 for greatest priority.
-       * If this field is not specified, then the virtual devices will be
-       * created with the default. If this field has values set, then the size
-       * of this must match with the above memory_limit_mb.
-       * </pre>
-       *
-       * <code>repeated int32 priority = 2;</code>
-       */
-      public int getPriority(int index) {
-        return priority_.getInt(index);
-      }
-      private int priorityMemoizedSerializedSize = -1;
-
-      private byte memoizedIsInitialized = -1;
-      @java.lang.Override
-      public final boolean isInitialized() {
-        byte isInitialized = memoizedIsInitialized;
-        if (isInitialized == 1) return true;
-        if (isInitialized == 0) return false;
-
-        memoizedIsInitialized = 1;
-        return true;
-      }
-
-      @java.lang.Override
-      public void writeTo(com.google.protobuf.CodedOutputStream output)
-                          throws java.io.IOException {
-        getSerializedSize();
-        if (getMemoryLimitMbList().size() > 0) {
-          output.writeUInt32NoTag(10);
-          output.writeUInt32NoTag(memoryLimitMbMemoizedSerializedSize);
-        }
-        for (int i = 0; i < memoryLimitMb_.size(); i++) {
-          output.writeFloatNoTag(memoryLimitMb_.getFloat(i));
-        }
-        if (getPriorityList().size() > 0) {
-          output.writeUInt32NoTag(18);
-          output.writeUInt32NoTag(priorityMemoizedSerializedSize);
-        }
-        for (int i = 0; i < priority_.size(); i++) {
-          output.writeInt32NoTag(priority_.getInt(i));
-        }
-        unknownFields.writeTo(output);
-      }
-
-      @java.lang.Override
-      public int getSerializedSize() {
-        int size = memoizedSize;
-        if (size != -1) return size;
-
-        size = 0;
-        {
-          int dataSize = 0;
-          dataSize = 4 * getMemoryLimitMbList().size();
-          size += dataSize;
-          if (!getMemoryLimitMbList().isEmpty()) {
-            size += 1;
-            size += com.google.protobuf.CodedOutputStream
-                .computeInt32SizeNoTag(dataSize);
-          }
-          memoryLimitMbMemoizedSerializedSize = dataSize;
-        }
-        {
-          int dataSize = 0;
-          for (int i = 0; i < priority_.size(); i++) {
-            dataSize += com.google.protobuf.CodedOutputStream
-              .computeInt32SizeNoTag(priority_.getInt(i));
-          }
-          size += dataSize;
-          if (!getPriorityList().isEmpty()) {
-            size += 1;
-            size += com.google.protobuf.CodedOutputStream
-                .computeInt32SizeNoTag(dataSize);
-          }
-          priorityMemoizedSerializedSize = dataSize;
-        }
-        size += unknownFields.getSerializedSize();
-        memoizedSize = size;
-        return size;
-      }
-
-      @java.lang.Override
-      public boolean equals(final java.lang.Object obj) {
-        if (obj == this) {
-         return true;
-        }
-        if (!(obj instanceof org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevices)) {
-          return super.equals(obj);
-        }
-        org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevices other = (org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevices) obj;
-
-        if (!getMemoryLimitMbList()
-            .equals(other.getMemoryLimitMbList())) return false;
-        if (!getPriorityList()
-            .equals(other.getPriorityList())) return false;
-        if (!unknownFields.equals(other.unknownFields)) return false;
-        return true;
-      }
-
-      @java.lang.Override
-      public int hashCode() {
-        if (memoizedHashCode != 0) {
-          return memoizedHashCode;
-        }
-        int hash = 41;
-        hash = (19 * hash) + getDescriptor().hashCode();
-        if (getMemoryLimitMbCount() > 0) {
-          hash = (37 * hash) + MEMORY_LIMIT_MB_FIELD_NUMBER;
-          hash = (53 * hash) + getMemoryLimitMbList().hashCode();
-        }
-        if (getPriorityCount() > 0) {
-          hash = (37 * hash) + PRIORITY_FIELD_NUMBER;
-          hash = (53 * hash) + getPriorityList().hashCode();
-        }
-        hash = (29 * hash) + unknownFields.hashCode();
-        memoizedHashCode = hash;
-        return hash;
-      }
-
-      public static org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevices parseFrom(
-          java.nio.ByteBuffer data)
-          throws com.google.protobuf.InvalidProtocolBufferException {
-        return PARSER.parseFrom(data);
-      }
-      public static org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevices parseFrom(
-          java.nio.ByteBuffer data,
-          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-          throws com.google.protobuf.InvalidProtocolBufferException {
-        return PARSER.parseFrom(data, extensionRegistry);
-      }
-      public static org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevices parseFrom(
-          com.google.protobuf.ByteString data)
-          throws com.google.protobuf.InvalidProtocolBufferException {
-        return PARSER.parseFrom(data);
-      }
-      public static org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevices parseFrom(
-          com.google.protobuf.ByteString data,
-          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-          throws com.google.protobuf.InvalidProtocolBufferException {
-        return PARSER.parseFrom(data, extensionRegistry);
-      }
-      public static org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevices parseFrom(byte[] data)
-          throws com.google.protobuf.InvalidProtocolBufferException {
-        return PARSER.parseFrom(data);
-      }
-      public static org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevices parseFrom(
-          byte[] data,
-          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-          throws com.google.protobuf.InvalidProtocolBufferException {
-        return PARSER.parseFrom(data, extensionRegistry);
-      }
-      public static org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevices parseFrom(java.io.InputStream input)
-          throws java.io.IOException {
-        return com.google.protobuf.GeneratedMessageV3
-            .parseWithIOException(PARSER, input);
-      }
-      public static org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevices parseFrom(
-          java.io.InputStream input,
-          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-          throws java.io.IOException {
-        return com.google.protobuf.GeneratedMessageV3
-            .parseWithIOException(PARSER, input, extensionRegistry);
-      }
-      public static org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevices parseDelimitedFrom(java.io.InputStream input)
-          throws java.io.IOException {
-        return com.google.protobuf.GeneratedMessageV3
-            .parseDelimitedWithIOException(PARSER, input);
-      }
-      public static org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevices parseDelimitedFrom(
-          java.io.InputStream input,
-          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-          throws java.io.IOException {
-        return com.google.protobuf.GeneratedMessageV3
-            .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-      }
-      public static org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevices parseFrom(
-          com.google.protobuf.CodedInputStream input)
-          throws java.io.IOException {
-        return com.google.protobuf.GeneratedMessageV3
-            .parseWithIOException(PARSER, input);
-      }
-      public static org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevices parseFrom(
-          com.google.protobuf.CodedInputStream input,
-          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-          throws java.io.IOException {
-        return com.google.protobuf.GeneratedMessageV3
-            .parseWithIOException(PARSER, input, extensionRegistry);
-      }
-
-      @java.lang.Override
-      public Builder newBuilderForType() { return newBuilder(); }
-      public static Builder newBuilder() {
-        return DEFAULT_INSTANCE.toBuilder();
-      }
-      public static Builder newBuilder(org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevices prototype) {
-        return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-      }
-      @java.lang.Override
-      public Builder toBuilder() {
-        return this == DEFAULT_INSTANCE
-            ? new Builder() : new Builder().mergeFrom(this);
-      }
-
-      @java.lang.Override
-      protected Builder newBuilderForType(
-          com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-        Builder builder = new Builder(parent);
-        return builder;
-      }
-      /**
-       * <pre>
-       * Configuration for breaking down a visible GPU into multiple "virtual"
-       * devices.
-       * </pre>
-       *
-       * Protobuf type {@code tensorflow.GPUOptions.Experimental.VirtualDevices}
-       */
-      public static final class Builder extends
-          com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-          // @@protoc_insertion_point(builder_implements:tensorflow.GPUOptions.Experimental.VirtualDevices)
-          org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevicesOrBuilder {
-        public static final com.google.protobuf.Descriptors.Descriptor
-            getDescriptor() {
-          return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_GPUOptions_Experimental_VirtualDevices_descriptor;
-        }
-
-        @java.lang.Override
-        protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-            internalGetFieldAccessorTable() {
-          return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_GPUOptions_Experimental_VirtualDevices_fieldAccessorTable
-              .ensureFieldAccessorsInitialized(
-                  org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevices.class, org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevices.Builder.class);
-        }
-
-        // Construct using org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevices.newBuilder()
-        private Builder() {
-          maybeForceBuilderInitialization();
-        }
-
-        private Builder(
-            com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-          super(parent);
-          maybeForceBuilderInitialization();
-        }
-        private void maybeForceBuilderInitialization() {
-          if (com.google.protobuf.GeneratedMessageV3
-                  .alwaysUseFieldBuilders) {
-          }
-        }
-        @java.lang.Override
-        public Builder clear() {
-          super.clear();
-          memoryLimitMb_ = emptyFloatList();
-          bitField0_ = (bitField0_ & ~0x00000001);
-          priority_ = emptyIntList();
-          bitField0_ = (bitField0_ & ~0x00000002);
-          return this;
-        }
-
-        @java.lang.Override
-        public com.google.protobuf.Descriptors.Descriptor
-            getDescriptorForType() {
-          return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_GPUOptions_Experimental_VirtualDevices_descriptor;
-        }
-
-        @java.lang.Override
-        public org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevices getDefaultInstanceForType() {
-          return org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevices.getDefaultInstance();
-        }
-
-        @java.lang.Override
-        public org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevices build() {
-          org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevices result = buildPartial();
-          if (!result.isInitialized()) {
-            throw newUninitializedMessageException(result);
-          }
-          return result;
-        }
-
-        @java.lang.Override
-        public org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevices buildPartial() {
-          org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevices result = new org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevices(this);
-          int from_bitField0_ = bitField0_;
-          if (((bitField0_ & 0x00000001) != 0)) {
-            memoryLimitMb_.makeImmutable();
-            bitField0_ = (bitField0_ & ~0x00000001);
-          }
-          result.memoryLimitMb_ = memoryLimitMb_;
-          if (((bitField0_ & 0x00000002) != 0)) {
-            priority_.makeImmutable();
-            bitField0_ = (bitField0_ & ~0x00000002);
-          }
-          result.priority_ = priority_;
-          onBuilt();
-          return result;
-        }
-
-        @java.lang.Override
-        public Builder clone() {
-          return super.clone();
-        }
-        @java.lang.Override
-        public Builder setField(
-            com.google.protobuf.Descriptors.FieldDescriptor field,
-            java.lang.Object value) {
-          return super.setField(field, value);
-        }
-        @java.lang.Override
-        public Builder clearField(
-            com.google.protobuf.Descriptors.FieldDescriptor field) {
-          return super.clearField(field);
-        }
-        @java.lang.Override
-        public Builder clearOneof(
-            com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-          return super.clearOneof(oneof);
-        }
-        @java.lang.Override
-        public Builder setRepeatedField(
-            com.google.protobuf.Descriptors.FieldDescriptor field,
-            int index, java.lang.Object value) {
-          return super.setRepeatedField(field, index, value);
-        }
-        @java.lang.Override
-        public Builder addRepeatedField(
-            com.google.protobuf.Descriptors.FieldDescriptor field,
-            java.lang.Object value) {
-          return super.addRepeatedField(field, value);
-        }
-        @java.lang.Override
-        public Builder mergeFrom(com.google.protobuf.Message other) {
-          if (other instanceof org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevices) {
-            return mergeFrom((org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevices)other);
-          } else {
-            super.mergeFrom(other);
-            return this;
-          }
-        }
-
-        public Builder mergeFrom(org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevices other) {
-          if (other == org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevices.getDefaultInstance()) return this;
-          if (!other.memoryLimitMb_.isEmpty()) {
-            if (memoryLimitMb_.isEmpty()) {
-              memoryLimitMb_ = other.memoryLimitMb_;
-              bitField0_ = (bitField0_ & ~0x00000001);
-            } else {
-              ensureMemoryLimitMbIsMutable();
-              memoryLimitMb_.addAll(other.memoryLimitMb_);
-            }
-            onChanged();
-          }
-          if (!other.priority_.isEmpty()) {
-            if (priority_.isEmpty()) {
-              priority_ = other.priority_;
-              bitField0_ = (bitField0_ & ~0x00000002);
-            } else {
-              ensurePriorityIsMutable();
-              priority_.addAll(other.priority_);
-            }
-            onChanged();
-          }
-          this.mergeUnknownFields(other.unknownFields);
-          onChanged();
-          return this;
-        }
-
-        @java.lang.Override
-        public final boolean isInitialized() {
-          return true;
-        }
-
-        @java.lang.Override
-        public Builder mergeFrom(
-            com.google.protobuf.CodedInputStream input,
-            com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-            throws java.io.IOException {
-          org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevices parsedMessage = null;
-          try {
-            parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-          } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-            parsedMessage = (org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevices) e.getUnfinishedMessage();
-            throw e.unwrapIOException();
-          } finally {
-            if (parsedMessage != null) {
-              mergeFrom(parsedMessage);
-            }
-          }
-          return this;
-        }
-        private int bitField0_;
-
-        private com.google.protobuf.Internal.FloatList memoryLimitMb_ = emptyFloatList();
-        private void ensureMemoryLimitMbIsMutable() {
-          if (!((bitField0_ & 0x00000001) != 0)) {
-            memoryLimitMb_ = mutableCopy(memoryLimitMb_);
-            bitField0_ |= 0x00000001;
-           }
-        }
-        /**
-         * <pre>
-         * Per "virtual" device memory limit, in MB. The number of elements in
-         * the list is the number of virtual devices to create on the
-         * corresponding visible GPU (see "virtual_devices" below).
-         * If empty, it will create single virtual device taking all available
-         * memory from the device.
-         * For the concept of "visible" and "virtual" GPU, see the comments for
-         * "visible_device_list" above for more information.
-         * </pre>
-         *
-         * <code>repeated float memory_limit_mb = 1;</code>
-         */
-        public java.util.List<java.lang.Float>
-            getMemoryLimitMbList() {
-          return ((bitField0_ & 0x00000001) != 0) ?
-                   java.util.Collections.unmodifiableList(memoryLimitMb_) : memoryLimitMb_;
-        }
-        /**
-         * <pre>
-         * Per "virtual" device memory limit, in MB. The number of elements in
-         * the list is the number of virtual devices to create on the
-         * corresponding visible GPU (see "virtual_devices" below).
-         * If empty, it will create single virtual device taking all available
-         * memory from the device.
-         * For the concept of "visible" and "virtual" GPU, see the comments for
-         * "visible_device_list" above for more information.
-         * </pre>
-         *
-         * <code>repeated float memory_limit_mb = 1;</code>
-         */
-        public int getMemoryLimitMbCount() {
-          return memoryLimitMb_.size();
-        }
-        /**
-         * <pre>
-         * Per "virtual" device memory limit, in MB. The number of elements in
-         * the list is the number of virtual devices to create on the
-         * corresponding visible GPU (see "virtual_devices" below).
-         * If empty, it will create single virtual device taking all available
-         * memory from the device.
-         * For the concept of "visible" and "virtual" GPU, see the comments for
-         * "visible_device_list" above for more information.
-         * </pre>
-         *
-         * <code>repeated float memory_limit_mb = 1;</code>
-         */
-        public float getMemoryLimitMb(int index) {
-          return memoryLimitMb_.getFloat(index);
-        }
-        /**
-         * <pre>
-         * Per "virtual" device memory limit, in MB. The number of elements in
-         * the list is the number of virtual devices to create on the
-         * corresponding visible GPU (see "virtual_devices" below).
-         * If empty, it will create single virtual device taking all available
-         * memory from the device.
-         * For the concept of "visible" and "virtual" GPU, see the comments for
-         * "visible_device_list" above for more information.
-         * </pre>
-         *
-         * <code>repeated float memory_limit_mb = 1;</code>
-         */
-        public Builder setMemoryLimitMb(
-            int index, float value) {
-          ensureMemoryLimitMbIsMutable();
-          memoryLimitMb_.setFloat(index, value);
-          onChanged();
-          return this;
-        }
-        /**
-         * <pre>
-         * Per "virtual" device memory limit, in MB. The number of elements in
-         * the list is the number of virtual devices to create on the
-         * corresponding visible GPU (see "virtual_devices" below).
-         * If empty, it will create single virtual device taking all available
-         * memory from the device.
-         * For the concept of "visible" and "virtual" GPU, see the comments for
-         * "visible_device_list" above for more information.
-         * </pre>
-         *
-         * <code>repeated float memory_limit_mb = 1;</code>
-         */
-        public Builder addMemoryLimitMb(float value) {
-          ensureMemoryLimitMbIsMutable();
-          memoryLimitMb_.addFloat(value);
-          onChanged();
-          return this;
-        }
-        /**
-         * <pre>
-         * Per "virtual" device memory limit, in MB. The number of elements in
-         * the list is the number of virtual devices to create on the
-         * corresponding visible GPU (see "virtual_devices" below).
-         * If empty, it will create single virtual device taking all available
-         * memory from the device.
-         * For the concept of "visible" and "virtual" GPU, see the comments for
-         * "visible_device_list" above for more information.
-         * </pre>
-         *
-         * <code>repeated float memory_limit_mb = 1;</code>
-         */
-        public Builder addAllMemoryLimitMb(
-            java.lang.Iterable<? extends java.lang.Float> values) {
-          ensureMemoryLimitMbIsMutable();
-          com.google.protobuf.AbstractMessageLite.Builder.addAll(
-              values, memoryLimitMb_);
-          onChanged();
-          return this;
-        }
-        /**
-         * <pre>
-         * Per "virtual" device memory limit, in MB. The number of elements in
-         * the list is the number of virtual devices to create on the
-         * corresponding visible GPU (see "virtual_devices" below).
-         * If empty, it will create single virtual device taking all available
-         * memory from the device.
-         * For the concept of "visible" and "virtual" GPU, see the comments for
-         * "visible_device_list" above for more information.
-         * </pre>
-         *
-         * <code>repeated float memory_limit_mb = 1;</code>
-         */
-        public Builder clearMemoryLimitMb() {
-          memoryLimitMb_ = emptyFloatList();
-          bitField0_ = (bitField0_ & ~0x00000001);
-          onChanged();
-          return this;
-        }
-
-        private com.google.protobuf.Internal.IntList priority_ = emptyIntList();
-        private void ensurePriorityIsMutable() {
-          if (!((bitField0_ & 0x00000002) != 0)) {
-            priority_ = mutableCopy(priority_);
-            bitField0_ |= 0x00000002;
-           }
-        }
-        /**
-         * <pre>
-         * Priority values to use with the virtual devices. Use the cuda function
-         * cudaDeviceGetStreamPriorityRange to query for valid range of values for
-         * priority.
-         * On a P4000 GPU with cuda 10.1, the priority range reported was 0 for
-         * least priority and -1 for greatest priority.
-         * If this field is not specified, then the virtual devices will be
-         * created with the default. If this field has values set, then the size
-         * of this must match with the above memory_limit_mb.
-         * </pre>
-         *
-         * <code>repeated int32 priority = 2;</code>
-         */
-        public java.util.List<java.lang.Integer>
-            getPriorityList() {
-          return ((bitField0_ & 0x00000002) != 0) ?
-                   java.util.Collections.unmodifiableList(priority_) : priority_;
-        }
-        /**
-         * <pre>
-         * Priority values to use with the virtual devices. Use the cuda function
-         * cudaDeviceGetStreamPriorityRange to query for valid range of values for
-         * priority.
-         * On a P4000 GPU with cuda 10.1, the priority range reported was 0 for
-         * least priority and -1 for greatest priority.
-         * If this field is not specified, then the virtual devices will be
-         * created with the default. If this field has values set, then the size
-         * of this must match with the above memory_limit_mb.
-         * </pre>
-         *
-         * <code>repeated int32 priority = 2;</code>
-         */
-        public int getPriorityCount() {
-          return priority_.size();
-        }
-        /**
-         * <pre>
-         * Priority values to use with the virtual devices. Use the cuda function
-         * cudaDeviceGetStreamPriorityRange to query for valid range of values for
-         * priority.
-         * On a P4000 GPU with cuda 10.1, the priority range reported was 0 for
-         * least priority and -1 for greatest priority.
-         * If this field is not specified, then the virtual devices will be
-         * created with the default. If this field has values set, then the size
-         * of this must match with the above memory_limit_mb.
-         * </pre>
-         *
-         * <code>repeated int32 priority = 2;</code>
-         */
-        public int getPriority(int index) {
-          return priority_.getInt(index);
-        }
-        /**
-         * <pre>
-         * Priority values to use with the virtual devices. Use the cuda function
-         * cudaDeviceGetStreamPriorityRange to query for valid range of values for
-         * priority.
-         * On a P4000 GPU with cuda 10.1, the priority range reported was 0 for
-         * least priority and -1 for greatest priority.
-         * If this field is not specified, then the virtual devices will be
-         * created with the default. If this field has values set, then the size
-         * of this must match with the above memory_limit_mb.
-         * </pre>
-         *
-         * <code>repeated int32 priority = 2;</code>
-         */
-        public Builder setPriority(
-            int index, int value) {
-          ensurePriorityIsMutable();
-          priority_.setInt(index, value);
-          onChanged();
-          return this;
-        }
-        /**
-         * <pre>
-         * Priority values to use with the virtual devices. Use the cuda function
-         * cudaDeviceGetStreamPriorityRange to query for valid range of values for
-         * priority.
-         * On a P4000 GPU with cuda 10.1, the priority range reported was 0 for
-         * least priority and -1 for greatest priority.
-         * If this field is not specified, then the virtual devices will be
-         * created with the default. If this field has values set, then the size
-         * of this must match with the above memory_limit_mb.
-         * </pre>
-         *
-         * <code>repeated int32 priority = 2;</code>
-         */
-        public Builder addPriority(int value) {
-          ensurePriorityIsMutable();
-          priority_.addInt(value);
-          onChanged();
-          return this;
-        }
-        /**
-         * <pre>
-         * Priority values to use with the virtual devices. Use the cuda function
-         * cudaDeviceGetStreamPriorityRange to query for valid range of values for
-         * priority.
-         * On a P4000 GPU with cuda 10.1, the priority range reported was 0 for
-         * least priority and -1 for greatest priority.
-         * If this field is not specified, then the virtual devices will be
-         * created with the default. If this field has values set, then the size
-         * of this must match with the above memory_limit_mb.
-         * </pre>
-         *
-         * <code>repeated int32 priority = 2;</code>
-         */
-        public Builder addAllPriority(
-            java.lang.Iterable<? extends java.lang.Integer> values) {
-          ensurePriorityIsMutable();
-          com.google.protobuf.AbstractMessageLite.Builder.addAll(
-              values, priority_);
-          onChanged();
-          return this;
-        }
-        /**
-         * <pre>
-         * Priority values to use with the virtual devices. Use the cuda function
-         * cudaDeviceGetStreamPriorityRange to query for valid range of values for
-         * priority.
-         * On a P4000 GPU with cuda 10.1, the priority range reported was 0 for
-         * least priority and -1 for greatest priority.
-         * If this field is not specified, then the virtual devices will be
-         * created with the default. If this field has values set, then the size
-         * of this must match with the above memory_limit_mb.
-         * </pre>
-         *
-         * <code>repeated int32 priority = 2;</code>
-         */
-        public Builder clearPriority() {
-          priority_ = emptyIntList();
-          bitField0_ = (bitField0_ & ~0x00000002);
-          onChanged();
-          return this;
-        }
-        @java.lang.Override
-        public final Builder setUnknownFields(
-            final com.google.protobuf.UnknownFieldSet unknownFields) {
-          return super.setUnknownFields(unknownFields);
-        }
-
-        @java.lang.Override
-        public final Builder mergeUnknownFields(
-            final com.google.protobuf.UnknownFieldSet unknownFields) {
-          return super.mergeUnknownFields(unknownFields);
-        }
-
-
-        // @@protoc_insertion_point(builder_scope:tensorflow.GPUOptions.Experimental.VirtualDevices)
-      }
-
-      // @@protoc_insertion_point(class_scope:tensorflow.GPUOptions.Experimental.VirtualDevices)
-      private static final org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevices DEFAULT_INSTANCE;
-      static {
-        DEFAULT_INSTANCE = new org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevices();
-      }
-
-      public static org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevices getDefaultInstance() {
-        return DEFAULT_INSTANCE;
-      }
-
-      private static final com.google.protobuf.Parser<VirtualDevices>
-          PARSER = new com.google.protobuf.AbstractParser<VirtualDevices>() {
-        @java.lang.Override
-        public VirtualDevices parsePartialFrom(
-            com.google.protobuf.CodedInputStream input,
-            com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-            throws com.google.protobuf.InvalidProtocolBufferException {
-          return new VirtualDevices(input, extensionRegistry);
-        }
-      };
-
-      public static com.google.protobuf.Parser<VirtualDevices> parser() {
-        return PARSER;
-      }
-
-      @java.lang.Override
-      public com.google.protobuf.Parser<VirtualDevices> getParserForType() {
-        return PARSER;
-      }
-
-      @java.lang.Override
-      public org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevices getDefaultInstanceForType() {
-        return DEFAULT_INSTANCE;
-      }
-
-    }
-
-    public static final int VIRTUAL_DEVICES_FIELD_NUMBER = 1;
-    private java.util.List<org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevices> virtualDevices_;
-    /**
-     * <pre>
-     * The multi virtual device settings. If empty (not set), it will create
-     * single virtual device on each visible GPU, according to the settings
-     * in "visible_device_list" above. Otherwise, the number of elements in the
-     * list must be the same as the number of visible GPUs (after
-     * "visible_device_list" filtering if it is set), and the string represented
-     * device names (e.g. /device:GPU:&lt;id&gt;) will refer to the virtual
-     * devices and have the &lt;id&gt; field assigned sequentially starting from 0,
-     * according to the order they appear in this list and the "memory_limit"
-     * list inside each element. For example,
-     *   visible_device_list = "1,0"
-     *   virtual_devices { memory_limit: 1GB memory_limit: 2GB }
-     *   virtual_devices {}
-     * will create three virtual devices as:
-     *   /device:GPU:0 -&gt; visible GPU 1 with 1GB memory
-     *   /device:GPU:1 -&gt; visible GPU 1 with 2GB memory
-     *   /device:GPU:2 -&gt; visible GPU 0 with all available memory
-     * NOTE:
-     * 1. It's invalid to set both this and "per_process_gpu_memory_fraction"
-     *    at the same time.
-     * 2. Currently this setting is per-process, not per-session. Using
-     *    different settings in different sessions within same process will
-     *    result in undefined behavior.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.GPUOptions.Experimental.VirtualDevices virtual_devices = 1;</code>
-     */
-    public java.util.List<org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevices> getVirtualDevicesList() {
-      return virtualDevices_;
-    }
-    /**
-     * <pre>
-     * The multi virtual device settings. If empty (not set), it will create
-     * single virtual device on each visible GPU, according to the settings
-     * in "visible_device_list" above. Otherwise, the number of elements in the
-     * list must be the same as the number of visible GPUs (after
-     * "visible_device_list" filtering if it is set), and the string represented
-     * device names (e.g. /device:GPU:&lt;id&gt;) will refer to the virtual
-     * devices and have the &lt;id&gt; field assigned sequentially starting from 0,
-     * according to the order they appear in this list and the "memory_limit"
-     * list inside each element. For example,
-     *   visible_device_list = "1,0"
-     *   virtual_devices { memory_limit: 1GB memory_limit: 2GB }
-     *   virtual_devices {}
-     * will create three virtual devices as:
-     *   /device:GPU:0 -&gt; visible GPU 1 with 1GB memory
-     *   /device:GPU:1 -&gt; visible GPU 1 with 2GB memory
-     *   /device:GPU:2 -&gt; visible GPU 0 with all available memory
-     * NOTE:
-     * 1. It's invalid to set both this and "per_process_gpu_memory_fraction"
-     *    at the same time.
-     * 2. Currently this setting is per-process, not per-session. Using
-     *    different settings in different sessions within same process will
-     *    result in undefined behavior.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.GPUOptions.Experimental.VirtualDevices virtual_devices = 1;</code>
-     */
-    public java.util.List<? extends org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevicesOrBuilder> 
-        getVirtualDevicesOrBuilderList() {
-      return virtualDevices_;
-    }
-    /**
-     * <pre>
-     * The multi virtual device settings. If empty (not set), it will create
-     * single virtual device on each visible GPU, according to the settings
-     * in "visible_device_list" above. Otherwise, the number of elements in the
-     * list must be the same as the number of visible GPUs (after
-     * "visible_device_list" filtering if it is set), and the string represented
-     * device names (e.g. /device:GPU:&lt;id&gt;) will refer to the virtual
-     * devices and have the &lt;id&gt; field assigned sequentially starting from 0,
-     * according to the order they appear in this list and the "memory_limit"
-     * list inside each element. For example,
-     *   visible_device_list = "1,0"
-     *   virtual_devices { memory_limit: 1GB memory_limit: 2GB }
-     *   virtual_devices {}
-     * will create three virtual devices as:
-     *   /device:GPU:0 -&gt; visible GPU 1 with 1GB memory
-     *   /device:GPU:1 -&gt; visible GPU 1 with 2GB memory
-     *   /device:GPU:2 -&gt; visible GPU 0 with all available memory
-     * NOTE:
-     * 1. It's invalid to set both this and "per_process_gpu_memory_fraction"
-     *    at the same time.
-     * 2. Currently this setting is per-process, not per-session. Using
-     *    different settings in different sessions within same process will
-     *    result in undefined behavior.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.GPUOptions.Experimental.VirtualDevices virtual_devices = 1;</code>
-     */
-    public int getVirtualDevicesCount() {
-      return virtualDevices_.size();
-    }
-    /**
-     * <pre>
-     * The multi virtual device settings. If empty (not set), it will create
-     * single virtual device on each visible GPU, according to the settings
-     * in "visible_device_list" above. Otherwise, the number of elements in the
-     * list must be the same as the number of visible GPUs (after
-     * "visible_device_list" filtering if it is set), and the string represented
-     * device names (e.g. /device:GPU:&lt;id&gt;) will refer to the virtual
-     * devices and have the &lt;id&gt; field assigned sequentially starting from 0,
-     * according to the order they appear in this list and the "memory_limit"
-     * list inside each element. For example,
-     *   visible_device_list = "1,0"
-     *   virtual_devices { memory_limit: 1GB memory_limit: 2GB }
-     *   virtual_devices {}
-     * will create three virtual devices as:
-     *   /device:GPU:0 -&gt; visible GPU 1 with 1GB memory
-     *   /device:GPU:1 -&gt; visible GPU 1 with 2GB memory
-     *   /device:GPU:2 -&gt; visible GPU 0 with all available memory
-     * NOTE:
-     * 1. It's invalid to set both this and "per_process_gpu_memory_fraction"
-     *    at the same time.
-     * 2. Currently this setting is per-process, not per-session. Using
-     *    different settings in different sessions within same process will
-     *    result in undefined behavior.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.GPUOptions.Experimental.VirtualDevices virtual_devices = 1;</code>
-     */
-    public org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevices getVirtualDevices(int index) {
-      return virtualDevices_.get(index);
-    }
-    /**
-     * <pre>
-     * The multi virtual device settings. If empty (not set), it will create
-     * single virtual device on each visible GPU, according to the settings
-     * in "visible_device_list" above. Otherwise, the number of elements in the
-     * list must be the same as the number of visible GPUs (after
-     * "visible_device_list" filtering if it is set), and the string represented
-     * device names (e.g. /device:GPU:&lt;id&gt;) will refer to the virtual
-     * devices and have the &lt;id&gt; field assigned sequentially starting from 0,
-     * according to the order they appear in this list and the "memory_limit"
-     * list inside each element. For example,
-     *   visible_device_list = "1,0"
-     *   virtual_devices { memory_limit: 1GB memory_limit: 2GB }
-     *   virtual_devices {}
-     * will create three virtual devices as:
-     *   /device:GPU:0 -&gt; visible GPU 1 with 1GB memory
-     *   /device:GPU:1 -&gt; visible GPU 1 with 2GB memory
-     *   /device:GPU:2 -&gt; visible GPU 0 with all available memory
-     * NOTE:
-     * 1. It's invalid to set both this and "per_process_gpu_memory_fraction"
-     *    at the same time.
-     * 2. Currently this setting is per-process, not per-session. Using
-     *    different settings in different sessions within same process will
-     *    result in undefined behavior.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.GPUOptions.Experimental.VirtualDevices virtual_devices = 1;</code>
-     */
-    public org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevicesOrBuilder getVirtualDevicesOrBuilder(
-        int index) {
-      return virtualDevices_.get(index);
-    }
-
-    public static final int USE_UNIFIED_MEMORY_FIELD_NUMBER = 2;
-    private boolean useUnifiedMemory_;
-    /**
-     * <pre>
-     * If true, uses CUDA unified memory for memory allocations. If
-     * per_process_gpu_memory_fraction option is greater than 1.0, then unified
-     * memory is used regardless of the value for this field. See comments for
-     * per_process_gpu_memory_fraction field for more details and requirements
-     * of the unified memory. This option is useful to oversubscribe memory if
-     * multiple processes are sharing a single GPU while individually using less
-     * than 1.0 per process memory fraction.
-     * </pre>
-     *
-     * <code>bool use_unified_memory = 2;</code>
-     */
-    public boolean getUseUnifiedMemory() {
-      return useUnifiedMemory_;
-    }
-
-    public static final int NUM_DEV_TO_DEV_COPY_STREAMS_FIELD_NUMBER = 3;
-    private int numDevToDevCopyStreams_;
-    /**
-     * <pre>
-     * If &gt; 1, the number of device-to-device copy streams to create
-     * for each GPUDevice.  Default value is 0, which is automatically
-     * converted to 1.
-     * </pre>
-     *
-     * <code>int32 num_dev_to_dev_copy_streams = 3;</code>
-     */
-    public int getNumDevToDevCopyStreams() {
-      return numDevToDevCopyStreams_;
-    }
-
-    public static final int COLLECTIVE_RING_ORDER_FIELD_NUMBER = 4;
-    private volatile java.lang.Object collectiveRingOrder_;
-    /**
-     * <pre>
-     * If non-empty, defines a good GPU ring order on a single worker based on
-     * device interconnect.  This assumes that all workers have the same GPU
-     * topology.  Specify as a comma-separated string, e.g. "3,2,1,0,7,6,5,4".
-     * This ring order is used by the RingReducer implementation of
-     * CollectiveReduce, and serves as an override to automatic ring order
-     * generation in OrderTaskDeviceMap() during CollectiveParam resolution.
-     * </pre>
-     *
-     * <code>string collective_ring_order = 4;</code>
-     */
-    public java.lang.String getCollectiveRingOrder() {
-      java.lang.Object ref = collectiveRingOrder_;
-      if (ref instanceof java.lang.String) {
-        return (java.lang.String) ref;
-      } else {
-        com.google.protobuf.ByteString bs = 
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        collectiveRingOrder_ = s;
-        return s;
-      }
-    }
-    /**
-     * <pre>
-     * If non-empty, defines a good GPU ring order on a single worker based on
-     * device interconnect.  This assumes that all workers have the same GPU
-     * topology.  Specify as a comma-separated string, e.g. "3,2,1,0,7,6,5,4".
-     * This ring order is used by the RingReducer implementation of
-     * CollectiveReduce, and serves as an override to automatic ring order
-     * generation in OrderTaskDeviceMap() during CollectiveParam resolution.
-     * </pre>
-     *
-     * <code>string collective_ring_order = 4;</code>
-     */
-    public com.google.protobuf.ByteString
-        getCollectiveRingOrderBytes() {
-      java.lang.Object ref = collectiveRingOrder_;
-      if (ref instanceof java.lang.String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        collectiveRingOrder_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-
-    public static final int TIMESTAMPED_ALLOCATOR_FIELD_NUMBER = 5;
-    private boolean timestampedAllocator_;
-    /**
-     * <pre>
-     * If true then extra work is done by GPUDevice and GPUBFCAllocator to
-     * keep track of when GPU memory is freed and when kernels actually
-     * complete so that we can know when a nominally free memory chunk
-     * is really not subject to pending use.
-     * </pre>
-     *
-     * <code>bool timestamped_allocator = 5;</code>
-     */
-    public boolean getTimestampedAllocator() {
-      return timestampedAllocator_;
-    }
-
-    public static final int KERNEL_TRACKER_MAX_INTERVAL_FIELD_NUMBER = 7;
-    private int kernelTrackerMaxInterval_;
-    /**
-     * <pre>
-     * Parameters for GPUKernelTracker.  By default no kernel tracking is done.
-     * Note that timestamped_allocator is only effective if some tracking is
-     * specified.
-     * If kernel_tracker_max_interval = n &gt; 0, then a tracking event
-     * is inserted after every n kernels without an event.
-     * </pre>
-     *
-     * <code>int32 kernel_tracker_max_interval = 7;</code>
-     */
-    public int getKernelTrackerMaxInterval() {
-      return kernelTrackerMaxInterval_;
-    }
-
-    public static final int KERNEL_TRACKER_MAX_BYTES_FIELD_NUMBER = 8;
-    private int kernelTrackerMaxBytes_;
-    /**
-     * <pre>
-     * If kernel_tracker_max_bytes = n &gt; 0, then a tracking event is
-     * inserted after every series of kernels allocating a sum of
-     * memory &gt;= n.  If one kernel allocates b * n bytes, then one
-     * event will be inserted after it, but it will count as b against
-     * the pending limit.
-     * </pre>
-     *
-     * <code>int32 kernel_tracker_max_bytes = 8;</code>
-     */
-    public int getKernelTrackerMaxBytes() {
-      return kernelTrackerMaxBytes_;
-    }
-
-    public static final int KERNEL_TRACKER_MAX_PENDING_FIELD_NUMBER = 9;
-    private int kernelTrackerMaxPending_;
-    /**
-     * <pre>
-     * If kernel_tracker_max_pending &gt; 0 then no more than this many
-     * tracking events can be outstanding at a time.  An attempt to
-     * launch an additional kernel will stall until an event
-     * completes.
-     * </pre>
-     *
-     * <code>int32 kernel_tracker_max_pending = 9;</code>
-     */
-    public int getKernelTrackerMaxPending() {
-      return kernelTrackerMaxPending_;
-    }
-
-    private byte memoizedIsInitialized = -1;
-    @java.lang.Override
-    public final boolean isInitialized() {
-      byte isInitialized = memoizedIsInitialized;
-      if (isInitialized == 1) return true;
-      if (isInitialized == 0) return false;
-
-      memoizedIsInitialized = 1;
-      return true;
-    }
-
-    @java.lang.Override
-    public void writeTo(com.google.protobuf.CodedOutputStream output)
-                        throws java.io.IOException {
-      for (int i = 0; i < virtualDevices_.size(); i++) {
-        output.writeMessage(1, virtualDevices_.get(i));
-      }
-      if (useUnifiedMemory_ != false) {
-        output.writeBool(2, useUnifiedMemory_);
-      }
-      if (numDevToDevCopyStreams_ != 0) {
-        output.writeInt32(3, numDevToDevCopyStreams_);
-      }
-      if (!getCollectiveRingOrderBytes().isEmpty()) {
-        com.google.protobuf.GeneratedMessageV3.writeString(output, 4, collectiveRingOrder_);
-      }
-      if (timestampedAllocator_ != false) {
-        output.writeBool(5, timestampedAllocator_);
-      }
-      if (kernelTrackerMaxInterval_ != 0) {
-        output.writeInt32(7, kernelTrackerMaxInterval_);
-      }
-      if (kernelTrackerMaxBytes_ != 0) {
-        output.writeInt32(8, kernelTrackerMaxBytes_);
-      }
-      if (kernelTrackerMaxPending_ != 0) {
-        output.writeInt32(9, kernelTrackerMaxPending_);
-      }
-      unknownFields.writeTo(output);
-    }
-
-    @java.lang.Override
-    public int getSerializedSize() {
-      int size = memoizedSize;
-      if (size != -1) return size;
-
-      size = 0;
-      for (int i = 0; i < virtualDevices_.size(); i++) {
-        size += com.google.protobuf.CodedOutputStream
-          .computeMessageSize(1, virtualDevices_.get(i));
-      }
-      if (useUnifiedMemory_ != false) {
-        size += com.google.protobuf.CodedOutputStream
-          .computeBoolSize(2, useUnifiedMemory_);
-      }
-      if (numDevToDevCopyStreams_ != 0) {
-        size += com.google.protobuf.CodedOutputStream
-          .computeInt32Size(3, numDevToDevCopyStreams_);
-      }
-      if (!getCollectiveRingOrderBytes().isEmpty()) {
-        size += com.google.protobuf.GeneratedMessageV3.computeStringSize(4, collectiveRingOrder_);
-      }
-      if (timestampedAllocator_ != false) {
-        size += com.google.protobuf.CodedOutputStream
-          .computeBoolSize(5, timestampedAllocator_);
-      }
-      if (kernelTrackerMaxInterval_ != 0) {
-        size += com.google.protobuf.CodedOutputStream
-          .computeInt32Size(7, kernelTrackerMaxInterval_);
-      }
-      if (kernelTrackerMaxBytes_ != 0) {
-        size += com.google.protobuf.CodedOutputStream
-          .computeInt32Size(8, kernelTrackerMaxBytes_);
-      }
-      if (kernelTrackerMaxPending_ != 0) {
-        size += com.google.protobuf.CodedOutputStream
-          .computeInt32Size(9, kernelTrackerMaxPending_);
-      }
-      size += unknownFields.getSerializedSize();
-      memoizedSize = size;
-      return size;
-    }
-
-    @java.lang.Override
-    public boolean equals(final java.lang.Object obj) {
-      if (obj == this) {
-       return true;
-      }
-      if (!(obj instanceof org.tensorflow.proto.framework.GPUOptions.Experimental)) {
-        return super.equals(obj);
-      }
-      org.tensorflow.proto.framework.GPUOptions.Experimental other = (org.tensorflow.proto.framework.GPUOptions.Experimental) obj;
-
-      if (!getVirtualDevicesList()
-          .equals(other.getVirtualDevicesList())) return false;
-      if (getUseUnifiedMemory()
-          != other.getUseUnifiedMemory()) return false;
-      if (getNumDevToDevCopyStreams()
-          != other.getNumDevToDevCopyStreams()) return false;
-      if (!getCollectiveRingOrder()
-          .equals(other.getCollectiveRingOrder())) return false;
-      if (getTimestampedAllocator()
-          != other.getTimestampedAllocator()) return false;
-      if (getKernelTrackerMaxInterval()
-          != other.getKernelTrackerMaxInterval()) return false;
-      if (getKernelTrackerMaxBytes()
-          != other.getKernelTrackerMaxBytes()) return false;
-      if (getKernelTrackerMaxPending()
-          != other.getKernelTrackerMaxPending()) return false;
-      if (!unknownFields.equals(other.unknownFields)) return false;
-      return true;
-    }
-
-    @java.lang.Override
-    public int hashCode() {
-      if (memoizedHashCode != 0) {
-        return memoizedHashCode;
-      }
-      int hash = 41;
-      hash = (19 * hash) + getDescriptor().hashCode();
-      if (getVirtualDevicesCount() > 0) {
-        hash = (37 * hash) + VIRTUAL_DEVICES_FIELD_NUMBER;
-        hash = (53 * hash) + getVirtualDevicesList().hashCode();
-      }
-      hash = (37 * hash) + USE_UNIFIED_MEMORY_FIELD_NUMBER;
-      hash = (53 * hash) + com.google.protobuf.Internal.hashBoolean(
-          getUseUnifiedMemory());
-      hash = (37 * hash) + NUM_DEV_TO_DEV_COPY_STREAMS_FIELD_NUMBER;
-      hash = (53 * hash) + getNumDevToDevCopyStreams();
-      hash = (37 * hash) + COLLECTIVE_RING_ORDER_FIELD_NUMBER;
-      hash = (53 * hash) + getCollectiveRingOrder().hashCode();
-      hash = (37 * hash) + TIMESTAMPED_ALLOCATOR_FIELD_NUMBER;
-      hash = (53 * hash) + com.google.protobuf.Internal.hashBoolean(
-          getTimestampedAllocator());
-      hash = (37 * hash) + KERNEL_TRACKER_MAX_INTERVAL_FIELD_NUMBER;
-      hash = (53 * hash) + getKernelTrackerMaxInterval();
-      hash = (37 * hash) + KERNEL_TRACKER_MAX_BYTES_FIELD_NUMBER;
-      hash = (53 * hash) + getKernelTrackerMaxBytes();
-      hash = (37 * hash) + KERNEL_TRACKER_MAX_PENDING_FIELD_NUMBER;
-      hash = (53 * hash) + getKernelTrackerMaxPending();
-      hash = (29 * hash) + unknownFields.hashCode();
-      memoizedHashCode = hash;
-      return hash;
-    }
-
-    public static org.tensorflow.proto.framework.GPUOptions.Experimental parseFrom(
-        java.nio.ByteBuffer data)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data);
-    }
-    public static org.tensorflow.proto.framework.GPUOptions.Experimental parseFrom(
-        java.nio.ByteBuffer data,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.GPUOptions.Experimental parseFrom(
-        com.google.protobuf.ByteString data)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data);
-    }
-    public static org.tensorflow.proto.framework.GPUOptions.Experimental parseFrom(
-        com.google.protobuf.ByteString data,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.GPUOptions.Experimental parseFrom(byte[] data)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data);
-    }
-    public static org.tensorflow.proto.framework.GPUOptions.Experimental parseFrom(
-        byte[] data,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.GPUOptions.Experimental parseFrom(java.io.InputStream input)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input);
-    }
-    public static org.tensorflow.proto.framework.GPUOptions.Experimental parseFrom(
-        java.io.InputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.GPUOptions.Experimental parseDelimitedFrom(java.io.InputStream input)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseDelimitedWithIOException(PARSER, input);
-    }
-    public static org.tensorflow.proto.framework.GPUOptions.Experimental parseDelimitedFrom(
-        java.io.InputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.GPUOptions.Experimental parseFrom(
-        com.google.protobuf.CodedInputStream input)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input);
-    }
-    public static org.tensorflow.proto.framework.GPUOptions.Experimental parseFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input, extensionRegistry);
-    }
-
-    @java.lang.Override
-    public Builder newBuilderForType() { return newBuilder(); }
-    public static Builder newBuilder() {
-      return DEFAULT_INSTANCE.toBuilder();
-    }
-    public static Builder newBuilder(org.tensorflow.proto.framework.GPUOptions.Experimental prototype) {
-      return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-    }
-    @java.lang.Override
-    public Builder toBuilder() {
-      return this == DEFAULT_INSTANCE
-          ? new Builder() : new Builder().mergeFrom(this);
-    }
-
-    @java.lang.Override
-    protected Builder newBuilderForType(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      Builder builder = new Builder(parent);
-      return builder;
-    }
-    /**
-     * Protobuf type {@code tensorflow.GPUOptions.Experimental}
-     */
-    public static final class Builder extends
-        com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-        // @@protoc_insertion_point(builder_implements:tensorflow.GPUOptions.Experimental)
-        org.tensorflow.proto.framework.GPUOptions.ExperimentalOrBuilder {
-      public static final com.google.protobuf.Descriptors.Descriptor
-          getDescriptor() {
-        return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_GPUOptions_Experimental_descriptor;
-      }
-
-      @java.lang.Override
-      protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-          internalGetFieldAccessorTable() {
-        return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_GPUOptions_Experimental_fieldAccessorTable
-            .ensureFieldAccessorsInitialized(
-                org.tensorflow.proto.framework.GPUOptions.Experimental.class, org.tensorflow.proto.framework.GPUOptions.Experimental.Builder.class);
-      }
-
-      // Construct using org.tensorflow.proto.framework.GPUOptions.Experimental.newBuilder()
-      private Builder() {
-        maybeForceBuilderInitialization();
-      }
-
-      private Builder(
-          com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-        super(parent);
-        maybeForceBuilderInitialization();
-      }
-      private void maybeForceBuilderInitialization() {
-        if (com.google.protobuf.GeneratedMessageV3
-                .alwaysUseFieldBuilders) {
-          getVirtualDevicesFieldBuilder();
-        }
-      }
-      @java.lang.Override
-      public Builder clear() {
-        super.clear();
-        if (virtualDevicesBuilder_ == null) {
-          virtualDevices_ = java.util.Collections.emptyList();
-          bitField0_ = (bitField0_ & ~0x00000001);
-        } else {
-          virtualDevicesBuilder_.clear();
-        }
-        useUnifiedMemory_ = false;
-
-        numDevToDevCopyStreams_ = 0;
-
-        collectiveRingOrder_ = "";
-
-        timestampedAllocator_ = false;
-
-        kernelTrackerMaxInterval_ = 0;
-
-        kernelTrackerMaxBytes_ = 0;
-
-        kernelTrackerMaxPending_ = 0;
-
-        return this;
-      }
-
-      @java.lang.Override
-      public com.google.protobuf.Descriptors.Descriptor
-          getDescriptorForType() {
-        return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_GPUOptions_Experimental_descriptor;
-      }
-
-      @java.lang.Override
-      public org.tensorflow.proto.framework.GPUOptions.Experimental getDefaultInstanceForType() {
-        return org.tensorflow.proto.framework.GPUOptions.Experimental.getDefaultInstance();
-      }
-
-      @java.lang.Override
-      public org.tensorflow.proto.framework.GPUOptions.Experimental build() {
-        org.tensorflow.proto.framework.GPUOptions.Experimental result = buildPartial();
-        if (!result.isInitialized()) {
-          throw newUninitializedMessageException(result);
-        }
-        return result;
-      }
-
-      @java.lang.Override
-      public org.tensorflow.proto.framework.GPUOptions.Experimental buildPartial() {
-        org.tensorflow.proto.framework.GPUOptions.Experimental result = new org.tensorflow.proto.framework.GPUOptions.Experimental(this);
-        int from_bitField0_ = bitField0_;
-        if (virtualDevicesBuilder_ == null) {
-          if (((bitField0_ & 0x00000001) != 0)) {
-            virtualDevices_ = java.util.Collections.unmodifiableList(virtualDevices_);
-            bitField0_ = (bitField0_ & ~0x00000001);
-          }
-          result.virtualDevices_ = virtualDevices_;
-        } else {
-          result.virtualDevices_ = virtualDevicesBuilder_.build();
-        }
-        result.useUnifiedMemory_ = useUnifiedMemory_;
-        result.numDevToDevCopyStreams_ = numDevToDevCopyStreams_;
-        result.collectiveRingOrder_ = collectiveRingOrder_;
-        result.timestampedAllocator_ = timestampedAllocator_;
-        result.kernelTrackerMaxInterval_ = kernelTrackerMaxInterval_;
-        result.kernelTrackerMaxBytes_ = kernelTrackerMaxBytes_;
-        result.kernelTrackerMaxPending_ = kernelTrackerMaxPending_;
-        onBuilt();
-        return result;
-      }
-
-      @java.lang.Override
-      public Builder clone() {
-        return super.clone();
-      }
-      @java.lang.Override
-      public Builder setField(
-          com.google.protobuf.Descriptors.FieldDescriptor field,
-          java.lang.Object value) {
-        return super.setField(field, value);
-      }
-      @java.lang.Override
-      public Builder clearField(
-          com.google.protobuf.Descriptors.FieldDescriptor field) {
-        return super.clearField(field);
-      }
-      @java.lang.Override
-      public Builder clearOneof(
-          com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-        return super.clearOneof(oneof);
-      }
-      @java.lang.Override
-      public Builder setRepeatedField(
-          com.google.protobuf.Descriptors.FieldDescriptor field,
-          int index, java.lang.Object value) {
-        return super.setRepeatedField(field, index, value);
-      }
-      @java.lang.Override
-      public Builder addRepeatedField(
-          com.google.protobuf.Descriptors.FieldDescriptor field,
-          java.lang.Object value) {
-        return super.addRepeatedField(field, value);
-      }
-      @java.lang.Override
-      public Builder mergeFrom(com.google.protobuf.Message other) {
-        if (other instanceof org.tensorflow.proto.framework.GPUOptions.Experimental) {
-          return mergeFrom((org.tensorflow.proto.framework.GPUOptions.Experimental)other);
-        } else {
-          super.mergeFrom(other);
-          return this;
-        }
-      }
-
-      public Builder mergeFrom(org.tensorflow.proto.framework.GPUOptions.Experimental other) {
-        if (other == org.tensorflow.proto.framework.GPUOptions.Experimental.getDefaultInstance()) return this;
-        if (virtualDevicesBuilder_ == null) {
-          if (!other.virtualDevices_.isEmpty()) {
-            if (virtualDevices_.isEmpty()) {
-              virtualDevices_ = other.virtualDevices_;
-              bitField0_ = (bitField0_ & ~0x00000001);
-            } else {
-              ensureVirtualDevicesIsMutable();
-              virtualDevices_.addAll(other.virtualDevices_);
-            }
-            onChanged();
-          }
-        } else {
-          if (!other.virtualDevices_.isEmpty()) {
-            if (virtualDevicesBuilder_.isEmpty()) {
-              virtualDevicesBuilder_.dispose();
-              virtualDevicesBuilder_ = null;
-              virtualDevices_ = other.virtualDevices_;
-              bitField0_ = (bitField0_ & ~0x00000001);
-              virtualDevicesBuilder_ = 
-                com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
-                   getVirtualDevicesFieldBuilder() : null;
-            } else {
-              virtualDevicesBuilder_.addAllMessages(other.virtualDevices_);
-            }
-          }
-        }
-        if (other.getUseUnifiedMemory() != false) {
-          setUseUnifiedMemory(other.getUseUnifiedMemory());
-        }
-        if (other.getNumDevToDevCopyStreams() != 0) {
-          setNumDevToDevCopyStreams(other.getNumDevToDevCopyStreams());
-        }
-        if (!other.getCollectiveRingOrder().isEmpty()) {
-          collectiveRingOrder_ = other.collectiveRingOrder_;
-          onChanged();
-        }
-        if (other.getTimestampedAllocator() != false) {
-          setTimestampedAllocator(other.getTimestampedAllocator());
-        }
-        if (other.getKernelTrackerMaxInterval() != 0) {
-          setKernelTrackerMaxInterval(other.getKernelTrackerMaxInterval());
-        }
-        if (other.getKernelTrackerMaxBytes() != 0) {
-          setKernelTrackerMaxBytes(other.getKernelTrackerMaxBytes());
-        }
-        if (other.getKernelTrackerMaxPending() != 0) {
-          setKernelTrackerMaxPending(other.getKernelTrackerMaxPending());
-        }
-        this.mergeUnknownFields(other.unknownFields);
-        onChanged();
-        return this;
-      }
-
-      @java.lang.Override
-      public final boolean isInitialized() {
-        return true;
-      }
-
-      @java.lang.Override
-      public Builder mergeFrom(
-          com.google.protobuf.CodedInputStream input,
-          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-          throws java.io.IOException {
-        org.tensorflow.proto.framework.GPUOptions.Experimental parsedMessage = null;
-        try {
-          parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-        } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-          parsedMessage = (org.tensorflow.proto.framework.GPUOptions.Experimental) e.getUnfinishedMessage();
-          throw e.unwrapIOException();
-        } finally {
-          if (parsedMessage != null) {
-            mergeFrom(parsedMessage);
-          }
-        }
-        return this;
-      }
-      private int bitField0_;
-
-      private java.util.List<org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevices> virtualDevices_ =
-        java.util.Collections.emptyList();
-      private void ensureVirtualDevicesIsMutable() {
-        if (!((bitField0_ & 0x00000001) != 0)) {
-          virtualDevices_ = new java.util.ArrayList<org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevices>(virtualDevices_);
-          bitField0_ |= 0x00000001;
-         }
-      }
-
-      private com.google.protobuf.RepeatedFieldBuilderV3<
-          org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevices, org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevices.Builder, org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevicesOrBuilder> virtualDevicesBuilder_;
-
-      /**
-       * <pre>
-       * The multi virtual device settings. If empty (not set), it will create
-       * single virtual device on each visible GPU, according to the settings
-       * in "visible_device_list" above. Otherwise, the number of elements in the
-       * list must be the same as the number of visible GPUs (after
-       * "visible_device_list" filtering if it is set), and the string represented
-       * device names (e.g. /device:GPU:&lt;id&gt;) will refer to the virtual
-       * devices and have the &lt;id&gt; field assigned sequentially starting from 0,
-       * according to the order they appear in this list and the "memory_limit"
-       * list inside each element. For example,
-       *   visible_device_list = "1,0"
-       *   virtual_devices { memory_limit: 1GB memory_limit: 2GB }
-       *   virtual_devices {}
-       * will create three virtual devices as:
-       *   /device:GPU:0 -&gt; visible GPU 1 with 1GB memory
-       *   /device:GPU:1 -&gt; visible GPU 1 with 2GB memory
-       *   /device:GPU:2 -&gt; visible GPU 0 with all available memory
-       * NOTE:
-       * 1. It's invalid to set both this and "per_process_gpu_memory_fraction"
-       *    at the same time.
-       * 2. Currently this setting is per-process, not per-session. Using
-       *    different settings in different sessions within same process will
-       *    result in undefined behavior.
-       * </pre>
-       *
-       * <code>repeated .tensorflow.GPUOptions.Experimental.VirtualDevices virtual_devices = 1;</code>
-       */
-      public java.util.List<org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevices> getVirtualDevicesList() {
-        if (virtualDevicesBuilder_ == null) {
-          return java.util.Collections.unmodifiableList(virtualDevices_);
-        } else {
-          return virtualDevicesBuilder_.getMessageList();
-        }
-      }
-      /**
-       * <pre>
-       * The multi virtual device settings. If empty (not set), it will create
-       * single virtual device on each visible GPU, according to the settings
-       * in "visible_device_list" above. Otherwise, the number of elements in the
-       * list must be the same as the number of visible GPUs (after
-       * "visible_device_list" filtering if it is set), and the string represented
-       * device names (e.g. /device:GPU:&lt;id&gt;) will refer to the virtual
-       * devices and have the &lt;id&gt; field assigned sequentially starting from 0,
-       * according to the order they appear in this list and the "memory_limit"
-       * list inside each element. For example,
-       *   visible_device_list = "1,0"
-       *   virtual_devices { memory_limit: 1GB memory_limit: 2GB }
-       *   virtual_devices {}
-       * will create three virtual devices as:
-       *   /device:GPU:0 -&gt; visible GPU 1 with 1GB memory
-       *   /device:GPU:1 -&gt; visible GPU 1 with 2GB memory
-       *   /device:GPU:2 -&gt; visible GPU 0 with all available memory
-       * NOTE:
-       * 1. It's invalid to set both this and "per_process_gpu_memory_fraction"
-       *    at the same time.
-       * 2. Currently this setting is per-process, not per-session. Using
-       *    different settings in different sessions within same process will
-       *    result in undefined behavior.
-       * </pre>
-       *
-       * <code>repeated .tensorflow.GPUOptions.Experimental.VirtualDevices virtual_devices = 1;</code>
-       */
-      public int getVirtualDevicesCount() {
-        if (virtualDevicesBuilder_ == null) {
-          return virtualDevices_.size();
-        } else {
-          return virtualDevicesBuilder_.getCount();
-        }
-      }
-      /**
-       * <pre>
-       * The multi virtual device settings. If empty (not set), it will create
-       * single virtual device on each visible GPU, according to the settings
-       * in "visible_device_list" above. Otherwise, the number of elements in the
-       * list must be the same as the number of visible GPUs (after
-       * "visible_device_list" filtering if it is set), and the string represented
-       * device names (e.g. /device:GPU:&lt;id&gt;) will refer to the virtual
-       * devices and have the &lt;id&gt; field assigned sequentially starting from 0,
-       * according to the order they appear in this list and the "memory_limit"
-       * list inside each element. For example,
-       *   visible_device_list = "1,0"
-       *   virtual_devices { memory_limit: 1GB memory_limit: 2GB }
-       *   virtual_devices {}
-       * will create three virtual devices as:
-       *   /device:GPU:0 -&gt; visible GPU 1 with 1GB memory
-       *   /device:GPU:1 -&gt; visible GPU 1 with 2GB memory
-       *   /device:GPU:2 -&gt; visible GPU 0 with all available memory
-       * NOTE:
-       * 1. It's invalid to set both this and "per_process_gpu_memory_fraction"
-       *    at the same time.
-       * 2. Currently this setting is per-process, not per-session. Using
-       *    different settings in different sessions within same process will
-       *    result in undefined behavior.
-       * </pre>
-       *
-       * <code>repeated .tensorflow.GPUOptions.Experimental.VirtualDevices virtual_devices = 1;</code>
-       */
-      public org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevices getVirtualDevices(int index) {
-        if (virtualDevicesBuilder_ == null) {
-          return virtualDevices_.get(index);
-        } else {
-          return virtualDevicesBuilder_.getMessage(index);
-        }
-      }
-      /**
-       * <pre>
-       * The multi virtual device settings. If empty (not set), it will create
-       * single virtual device on each visible GPU, according to the settings
-       * in "visible_device_list" above. Otherwise, the number of elements in the
-       * list must be the same as the number of visible GPUs (after
-       * "visible_device_list" filtering if it is set), and the string represented
-       * device names (e.g. /device:GPU:&lt;id&gt;) will refer to the virtual
-       * devices and have the &lt;id&gt; field assigned sequentially starting from 0,
-       * according to the order they appear in this list and the "memory_limit"
-       * list inside each element. For example,
-       *   visible_device_list = "1,0"
-       *   virtual_devices { memory_limit: 1GB memory_limit: 2GB }
-       *   virtual_devices {}
-       * will create three virtual devices as:
-       *   /device:GPU:0 -&gt; visible GPU 1 with 1GB memory
-       *   /device:GPU:1 -&gt; visible GPU 1 with 2GB memory
-       *   /device:GPU:2 -&gt; visible GPU 0 with all available memory
-       * NOTE:
-       * 1. It's invalid to set both this and "per_process_gpu_memory_fraction"
-       *    at the same time.
-       * 2. Currently this setting is per-process, not per-session. Using
-       *    different settings in different sessions within same process will
-       *    result in undefined behavior.
-       * </pre>
-       *
-       * <code>repeated .tensorflow.GPUOptions.Experimental.VirtualDevices virtual_devices = 1;</code>
-       */
-      public Builder setVirtualDevices(
-          int index, org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevices value) {
-        if (virtualDevicesBuilder_ == null) {
-          if (value == null) {
-            throw new NullPointerException();
-          }
-          ensureVirtualDevicesIsMutable();
-          virtualDevices_.set(index, value);
-          onChanged();
-        } else {
-          virtualDevicesBuilder_.setMessage(index, value);
-        }
-        return this;
-      }
-      /**
-       * <pre>
-       * The multi virtual device settings. If empty (not set), it will create
-       * single virtual device on each visible GPU, according to the settings
-       * in "visible_device_list" above. Otherwise, the number of elements in the
-       * list must be the same as the number of visible GPUs (after
-       * "visible_device_list" filtering if it is set), and the string represented
-       * device names (e.g. /device:GPU:&lt;id&gt;) will refer to the virtual
-       * devices and have the &lt;id&gt; field assigned sequentially starting from 0,
-       * according to the order they appear in this list and the "memory_limit"
-       * list inside each element. For example,
-       *   visible_device_list = "1,0"
-       *   virtual_devices { memory_limit: 1GB memory_limit: 2GB }
-       *   virtual_devices {}
-       * will create three virtual devices as:
-       *   /device:GPU:0 -&gt; visible GPU 1 with 1GB memory
-       *   /device:GPU:1 -&gt; visible GPU 1 with 2GB memory
-       *   /device:GPU:2 -&gt; visible GPU 0 with all available memory
-       * NOTE:
-       * 1. It's invalid to set both this and "per_process_gpu_memory_fraction"
-       *    at the same time.
-       * 2. Currently this setting is per-process, not per-session. Using
-       *    different settings in different sessions within same process will
-       *    result in undefined behavior.
-       * </pre>
-       *
-       * <code>repeated .tensorflow.GPUOptions.Experimental.VirtualDevices virtual_devices = 1;</code>
-       */
-      public Builder setVirtualDevices(
-          int index, org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevices.Builder builderForValue) {
-        if (virtualDevicesBuilder_ == null) {
-          ensureVirtualDevicesIsMutable();
-          virtualDevices_.set(index, builderForValue.build());
-          onChanged();
-        } else {
-          virtualDevicesBuilder_.setMessage(index, builderForValue.build());
-        }
-        return this;
-      }
-      /**
-       * <pre>
-       * The multi virtual device settings. If empty (not set), it will create
-       * single virtual device on each visible GPU, according to the settings
-       * in "visible_device_list" above. Otherwise, the number of elements in the
-       * list must be the same as the number of visible GPUs (after
-       * "visible_device_list" filtering if it is set), and the string represented
-       * device names (e.g. /device:GPU:&lt;id&gt;) will refer to the virtual
-       * devices and have the &lt;id&gt; field assigned sequentially starting from 0,
-       * according to the order they appear in this list and the "memory_limit"
-       * list inside each element. For example,
-       *   visible_device_list = "1,0"
-       *   virtual_devices { memory_limit: 1GB memory_limit: 2GB }
-       *   virtual_devices {}
-       * will create three virtual devices as:
-       *   /device:GPU:0 -&gt; visible GPU 1 with 1GB memory
-       *   /device:GPU:1 -&gt; visible GPU 1 with 2GB memory
-       *   /device:GPU:2 -&gt; visible GPU 0 with all available memory
-       * NOTE:
-       * 1. It's invalid to set both this and "per_process_gpu_memory_fraction"
-       *    at the same time.
-       * 2. Currently this setting is per-process, not per-session. Using
-       *    different settings in different sessions within same process will
-       *    result in undefined behavior.
-       * </pre>
-       *
-       * <code>repeated .tensorflow.GPUOptions.Experimental.VirtualDevices virtual_devices = 1;</code>
-       */
-      public Builder addVirtualDevices(org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevices value) {
-        if (virtualDevicesBuilder_ == null) {
-          if (value == null) {
-            throw new NullPointerException();
-          }
-          ensureVirtualDevicesIsMutable();
-          virtualDevices_.add(value);
-          onChanged();
-        } else {
-          virtualDevicesBuilder_.addMessage(value);
-        }
-        return this;
-      }
-      /**
-       * <pre>
-       * The multi virtual device settings. If empty (not set), it will create
-       * single virtual device on each visible GPU, according to the settings
-       * in "visible_device_list" above. Otherwise, the number of elements in the
-       * list must be the same as the number of visible GPUs (after
-       * "visible_device_list" filtering if it is set), and the string represented
-       * device names (e.g. /device:GPU:&lt;id&gt;) will refer to the virtual
-       * devices and have the &lt;id&gt; field assigned sequentially starting from 0,
-       * according to the order they appear in this list and the "memory_limit"
-       * list inside each element. For example,
-       *   visible_device_list = "1,0"
-       *   virtual_devices { memory_limit: 1GB memory_limit: 2GB }
-       *   virtual_devices {}
-       * will create three virtual devices as:
-       *   /device:GPU:0 -&gt; visible GPU 1 with 1GB memory
-       *   /device:GPU:1 -&gt; visible GPU 1 with 2GB memory
-       *   /device:GPU:2 -&gt; visible GPU 0 with all available memory
-       * NOTE:
-       * 1. It's invalid to set both this and "per_process_gpu_memory_fraction"
-       *    at the same time.
-       * 2. Currently this setting is per-process, not per-session. Using
-       *    different settings in different sessions within same process will
-       *    result in undefined behavior.
-       * </pre>
-       *
-       * <code>repeated .tensorflow.GPUOptions.Experimental.VirtualDevices virtual_devices = 1;</code>
-       */
-      public Builder addVirtualDevices(
-          int index, org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevices value) {
-        if (virtualDevicesBuilder_ == null) {
-          if (value == null) {
-            throw new NullPointerException();
-          }
-          ensureVirtualDevicesIsMutable();
-          virtualDevices_.add(index, value);
-          onChanged();
-        } else {
-          virtualDevicesBuilder_.addMessage(index, value);
-        }
-        return this;
-      }
-      /**
-       * <pre>
-       * The multi virtual device settings. If empty (not set), it will create
-       * single virtual device on each visible GPU, according to the settings
-       * in "visible_device_list" above. Otherwise, the number of elements in the
-       * list must be the same as the number of visible GPUs (after
-       * "visible_device_list" filtering if it is set), and the string represented
-       * device names (e.g. /device:GPU:&lt;id&gt;) will refer to the virtual
-       * devices and have the &lt;id&gt; field assigned sequentially starting from 0,
-       * according to the order they appear in this list and the "memory_limit"
-       * list inside each element. For example,
-       *   visible_device_list = "1,0"
-       *   virtual_devices { memory_limit: 1GB memory_limit: 2GB }
-       *   virtual_devices {}
-       * will create three virtual devices as:
-       *   /device:GPU:0 -&gt; visible GPU 1 with 1GB memory
-       *   /device:GPU:1 -&gt; visible GPU 1 with 2GB memory
-       *   /device:GPU:2 -&gt; visible GPU 0 with all available memory
-       * NOTE:
-       * 1. It's invalid to set both this and "per_process_gpu_memory_fraction"
-       *    at the same time.
-       * 2. Currently this setting is per-process, not per-session. Using
-       *    different settings in different sessions within same process will
-       *    result in undefined behavior.
-       * </pre>
-       *
-       * <code>repeated .tensorflow.GPUOptions.Experimental.VirtualDevices virtual_devices = 1;</code>
-       */
-      public Builder addVirtualDevices(
-          org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevices.Builder builderForValue) {
-        if (virtualDevicesBuilder_ == null) {
-          ensureVirtualDevicesIsMutable();
-          virtualDevices_.add(builderForValue.build());
-          onChanged();
-        } else {
-          virtualDevicesBuilder_.addMessage(builderForValue.build());
-        }
-        return this;
-      }
-      /**
-       * <pre>
-       * The multi virtual device settings. If empty (not set), it will create
-       * single virtual device on each visible GPU, according to the settings
-       * in "visible_device_list" above. Otherwise, the number of elements in the
-       * list must be the same as the number of visible GPUs (after
-       * "visible_device_list" filtering if it is set), and the string represented
-       * device names (e.g. /device:GPU:&lt;id&gt;) will refer to the virtual
-       * devices and have the &lt;id&gt; field assigned sequentially starting from 0,
-       * according to the order they appear in this list and the "memory_limit"
-       * list inside each element. For example,
-       *   visible_device_list = "1,0"
-       *   virtual_devices { memory_limit: 1GB memory_limit: 2GB }
-       *   virtual_devices {}
-       * will create three virtual devices as:
-       *   /device:GPU:0 -&gt; visible GPU 1 with 1GB memory
-       *   /device:GPU:1 -&gt; visible GPU 1 with 2GB memory
-       *   /device:GPU:2 -&gt; visible GPU 0 with all available memory
-       * NOTE:
-       * 1. It's invalid to set both this and "per_process_gpu_memory_fraction"
-       *    at the same time.
-       * 2. Currently this setting is per-process, not per-session. Using
-       *    different settings in different sessions within same process will
-       *    result in undefined behavior.
-       * </pre>
-       *
-       * <code>repeated .tensorflow.GPUOptions.Experimental.VirtualDevices virtual_devices = 1;</code>
-       */
-      public Builder addVirtualDevices(
-          int index, org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevices.Builder builderForValue) {
-        if (virtualDevicesBuilder_ == null) {
-          ensureVirtualDevicesIsMutable();
-          virtualDevices_.add(index, builderForValue.build());
-          onChanged();
-        } else {
-          virtualDevicesBuilder_.addMessage(index, builderForValue.build());
-        }
-        return this;
-      }
-      /**
-       * <pre>
-       * The multi virtual device settings. If empty (not set), it will create
-       * single virtual device on each visible GPU, according to the settings
-       * in "visible_device_list" above. Otherwise, the number of elements in the
-       * list must be the same as the number of visible GPUs (after
-       * "visible_device_list" filtering if it is set), and the string represented
-       * device names (e.g. /device:GPU:&lt;id&gt;) will refer to the virtual
-       * devices and have the &lt;id&gt; field assigned sequentially starting from 0,
-       * according to the order they appear in this list and the "memory_limit"
-       * list inside each element. For example,
-       *   visible_device_list = "1,0"
-       *   virtual_devices { memory_limit: 1GB memory_limit: 2GB }
-       *   virtual_devices {}
-       * will create three virtual devices as:
-       *   /device:GPU:0 -&gt; visible GPU 1 with 1GB memory
-       *   /device:GPU:1 -&gt; visible GPU 1 with 2GB memory
-       *   /device:GPU:2 -&gt; visible GPU 0 with all available memory
-       * NOTE:
-       * 1. It's invalid to set both this and "per_process_gpu_memory_fraction"
-       *    at the same time.
-       * 2. Currently this setting is per-process, not per-session. Using
-       *    different settings in different sessions within same process will
-       *    result in undefined behavior.
-       * </pre>
-       *
-       * <code>repeated .tensorflow.GPUOptions.Experimental.VirtualDevices virtual_devices = 1;</code>
-       */
-      public Builder addAllVirtualDevices(
-          java.lang.Iterable<? extends org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevices> values) {
-        if (virtualDevicesBuilder_ == null) {
-          ensureVirtualDevicesIsMutable();
-          com.google.protobuf.AbstractMessageLite.Builder.addAll(
-              values, virtualDevices_);
-          onChanged();
-        } else {
-          virtualDevicesBuilder_.addAllMessages(values);
-        }
-        return this;
-      }
-      /**
-       * <pre>
-       * The multi virtual device settings. If empty (not set), it will create
-       * single virtual device on each visible GPU, according to the settings
-       * in "visible_device_list" above. Otherwise, the number of elements in the
-       * list must be the same as the number of visible GPUs (after
-       * "visible_device_list" filtering if it is set), and the string represented
-       * device names (e.g. /device:GPU:&lt;id&gt;) will refer to the virtual
-       * devices and have the &lt;id&gt; field assigned sequentially starting from 0,
-       * according to the order they appear in this list and the "memory_limit"
-       * list inside each element. For example,
-       *   visible_device_list = "1,0"
-       *   virtual_devices { memory_limit: 1GB memory_limit: 2GB }
-       *   virtual_devices {}
-       * will create three virtual devices as:
-       *   /device:GPU:0 -&gt; visible GPU 1 with 1GB memory
-       *   /device:GPU:1 -&gt; visible GPU 1 with 2GB memory
-       *   /device:GPU:2 -&gt; visible GPU 0 with all available memory
-       * NOTE:
-       * 1. It's invalid to set both this and "per_process_gpu_memory_fraction"
-       *    at the same time.
-       * 2. Currently this setting is per-process, not per-session. Using
-       *    different settings in different sessions within same process will
-       *    result in undefined behavior.
-       * </pre>
-       *
-       * <code>repeated .tensorflow.GPUOptions.Experimental.VirtualDevices virtual_devices = 1;</code>
-       */
-      public Builder clearVirtualDevices() {
-        if (virtualDevicesBuilder_ == null) {
-          virtualDevices_ = java.util.Collections.emptyList();
-          bitField0_ = (bitField0_ & ~0x00000001);
-          onChanged();
-        } else {
-          virtualDevicesBuilder_.clear();
-        }
-        return this;
-      }
-      /**
-       * <pre>
-       * The multi virtual device settings. If empty (not set), it will create
-       * single virtual device on each visible GPU, according to the settings
-       * in "visible_device_list" above. Otherwise, the number of elements in the
-       * list must be the same as the number of visible GPUs (after
-       * "visible_device_list" filtering if it is set), and the string represented
-       * device names (e.g. /device:GPU:&lt;id&gt;) will refer to the virtual
-       * devices and have the &lt;id&gt; field assigned sequentially starting from 0,
-       * according to the order they appear in this list and the "memory_limit"
-       * list inside each element. For example,
-       *   visible_device_list = "1,0"
-       *   virtual_devices { memory_limit: 1GB memory_limit: 2GB }
-       *   virtual_devices {}
-       * will create three virtual devices as:
-       *   /device:GPU:0 -&gt; visible GPU 1 with 1GB memory
-       *   /device:GPU:1 -&gt; visible GPU 1 with 2GB memory
-       *   /device:GPU:2 -&gt; visible GPU 0 with all available memory
-       * NOTE:
-       * 1. It's invalid to set both this and "per_process_gpu_memory_fraction"
-       *    at the same time.
-       * 2. Currently this setting is per-process, not per-session. Using
-       *    different settings in different sessions within same process will
-       *    result in undefined behavior.
-       * </pre>
-       *
-       * <code>repeated .tensorflow.GPUOptions.Experimental.VirtualDevices virtual_devices = 1;</code>
-       */
-      public Builder removeVirtualDevices(int index) {
-        if (virtualDevicesBuilder_ == null) {
-          ensureVirtualDevicesIsMutable();
-          virtualDevices_.remove(index);
-          onChanged();
-        } else {
-          virtualDevicesBuilder_.remove(index);
-        }
-        return this;
-      }
-      /**
-       * <pre>
-       * The multi virtual device settings. If empty (not set), it will create
-       * single virtual device on each visible GPU, according to the settings
-       * in "visible_device_list" above. Otherwise, the number of elements in the
-       * list must be the same as the number of visible GPUs (after
-       * "visible_device_list" filtering if it is set), and the string represented
-       * device names (e.g. /device:GPU:&lt;id&gt;) will refer to the virtual
-       * devices and have the &lt;id&gt; field assigned sequentially starting from 0,
-       * according to the order they appear in this list and the "memory_limit"
-       * list inside each element. For example,
-       *   visible_device_list = "1,0"
-       *   virtual_devices { memory_limit: 1GB memory_limit: 2GB }
-       *   virtual_devices {}
-       * will create three virtual devices as:
-       *   /device:GPU:0 -&gt; visible GPU 1 with 1GB memory
-       *   /device:GPU:1 -&gt; visible GPU 1 with 2GB memory
-       *   /device:GPU:2 -&gt; visible GPU 0 with all available memory
-       * NOTE:
-       * 1. It's invalid to set both this and "per_process_gpu_memory_fraction"
-       *    at the same time.
-       * 2. Currently this setting is per-process, not per-session. Using
-       *    different settings in different sessions within same process will
-       *    result in undefined behavior.
-       * </pre>
-       *
-       * <code>repeated .tensorflow.GPUOptions.Experimental.VirtualDevices virtual_devices = 1;</code>
-       */
-      public org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevices.Builder getVirtualDevicesBuilder(
-          int index) {
-        return getVirtualDevicesFieldBuilder().getBuilder(index);
-      }
-      /**
-       * <pre>
-       * The multi virtual device settings. If empty (not set), it will create
-       * single virtual device on each visible GPU, according to the settings
-       * in "visible_device_list" above. Otherwise, the number of elements in the
-       * list must be the same as the number of visible GPUs (after
-       * "visible_device_list" filtering if it is set), and the string represented
-       * device names (e.g. /device:GPU:&lt;id&gt;) will refer to the virtual
-       * devices and have the &lt;id&gt; field assigned sequentially starting from 0,
-       * according to the order they appear in this list and the "memory_limit"
-       * list inside each element. For example,
-       *   visible_device_list = "1,0"
-       *   virtual_devices { memory_limit: 1GB memory_limit: 2GB }
-       *   virtual_devices {}
-       * will create three virtual devices as:
-       *   /device:GPU:0 -&gt; visible GPU 1 with 1GB memory
-       *   /device:GPU:1 -&gt; visible GPU 1 with 2GB memory
-       *   /device:GPU:2 -&gt; visible GPU 0 with all available memory
-       * NOTE:
-       * 1. It's invalid to set both this and "per_process_gpu_memory_fraction"
-       *    at the same time.
-       * 2. Currently this setting is per-process, not per-session. Using
-       *    different settings in different sessions within same process will
-       *    result in undefined behavior.
-       * </pre>
-       *
-       * <code>repeated .tensorflow.GPUOptions.Experimental.VirtualDevices virtual_devices = 1;</code>
-       */
-      public org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevicesOrBuilder getVirtualDevicesOrBuilder(
-          int index) {
-        if (virtualDevicesBuilder_ == null) {
-          return virtualDevices_.get(index);  } else {
-          return virtualDevicesBuilder_.getMessageOrBuilder(index);
-        }
-      }
-      /**
-       * <pre>
-       * The multi virtual device settings. If empty (not set), it will create
-       * single virtual device on each visible GPU, according to the settings
-       * in "visible_device_list" above. Otherwise, the number of elements in the
-       * list must be the same as the number of visible GPUs (after
-       * "visible_device_list" filtering if it is set), and the string represented
-       * device names (e.g. /device:GPU:&lt;id&gt;) will refer to the virtual
-       * devices and have the &lt;id&gt; field assigned sequentially starting from 0,
-       * according to the order they appear in this list and the "memory_limit"
-       * list inside each element. For example,
-       *   visible_device_list = "1,0"
-       *   virtual_devices { memory_limit: 1GB memory_limit: 2GB }
-       *   virtual_devices {}
-       * will create three virtual devices as:
-       *   /device:GPU:0 -&gt; visible GPU 1 with 1GB memory
-       *   /device:GPU:1 -&gt; visible GPU 1 with 2GB memory
-       *   /device:GPU:2 -&gt; visible GPU 0 with all available memory
-       * NOTE:
-       * 1. It's invalid to set both this and "per_process_gpu_memory_fraction"
-       *    at the same time.
-       * 2. Currently this setting is per-process, not per-session. Using
-       *    different settings in different sessions within same process will
-       *    result in undefined behavior.
-       * </pre>
-       *
-       * <code>repeated .tensorflow.GPUOptions.Experimental.VirtualDevices virtual_devices = 1;</code>
-       */
-      public java.util.List<? extends org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevicesOrBuilder> 
-           getVirtualDevicesOrBuilderList() {
-        if (virtualDevicesBuilder_ != null) {
-          return virtualDevicesBuilder_.getMessageOrBuilderList();
-        } else {
-          return java.util.Collections.unmodifiableList(virtualDevices_);
-        }
-      }
-      /**
-       * <pre>
-       * The multi virtual device settings. If empty (not set), it will create
-       * single virtual device on each visible GPU, according to the settings
-       * in "visible_device_list" above. Otherwise, the number of elements in the
-       * list must be the same as the number of visible GPUs (after
-       * "visible_device_list" filtering if it is set), and the string represented
-       * device names (e.g. /device:GPU:&lt;id&gt;) will refer to the virtual
-       * devices and have the &lt;id&gt; field assigned sequentially starting from 0,
-       * according to the order they appear in this list and the "memory_limit"
-       * list inside each element. For example,
-       *   visible_device_list = "1,0"
-       *   virtual_devices { memory_limit: 1GB memory_limit: 2GB }
-       *   virtual_devices {}
-       * will create three virtual devices as:
-       *   /device:GPU:0 -&gt; visible GPU 1 with 1GB memory
-       *   /device:GPU:1 -&gt; visible GPU 1 with 2GB memory
-       *   /device:GPU:2 -&gt; visible GPU 0 with all available memory
-       * NOTE:
-       * 1. It's invalid to set both this and "per_process_gpu_memory_fraction"
-       *    at the same time.
-       * 2. Currently this setting is per-process, not per-session. Using
-       *    different settings in different sessions within same process will
-       *    result in undefined behavior.
-       * </pre>
-       *
-       * <code>repeated .tensorflow.GPUOptions.Experimental.VirtualDevices virtual_devices = 1;</code>
-       */
-      public org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevices.Builder addVirtualDevicesBuilder() {
-        return getVirtualDevicesFieldBuilder().addBuilder(
-            org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevices.getDefaultInstance());
-      }
-      /**
-       * <pre>
-       * The multi virtual device settings. If empty (not set), it will create
-       * single virtual device on each visible GPU, according to the settings
-       * in "visible_device_list" above. Otherwise, the number of elements in the
-       * list must be the same as the number of visible GPUs (after
-       * "visible_device_list" filtering if it is set), and the string represented
-       * device names (e.g. /device:GPU:&lt;id&gt;) will refer to the virtual
-       * devices and have the &lt;id&gt; field assigned sequentially starting from 0,
-       * according to the order they appear in this list and the "memory_limit"
-       * list inside each element. For example,
-       *   visible_device_list = "1,0"
-       *   virtual_devices { memory_limit: 1GB memory_limit: 2GB }
-       *   virtual_devices {}
-       * will create three virtual devices as:
-       *   /device:GPU:0 -&gt; visible GPU 1 with 1GB memory
-       *   /device:GPU:1 -&gt; visible GPU 1 with 2GB memory
-       *   /device:GPU:2 -&gt; visible GPU 0 with all available memory
-       * NOTE:
-       * 1. It's invalid to set both this and "per_process_gpu_memory_fraction"
-       *    at the same time.
-       * 2. Currently this setting is per-process, not per-session. Using
-       *    different settings in different sessions within same process will
-       *    result in undefined behavior.
-       * </pre>
-       *
-       * <code>repeated .tensorflow.GPUOptions.Experimental.VirtualDevices virtual_devices = 1;</code>
-       */
-      public org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevices.Builder addVirtualDevicesBuilder(
-          int index) {
-        return getVirtualDevicesFieldBuilder().addBuilder(
-            index, org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevices.getDefaultInstance());
-      }
-      /**
-       * <pre>
-       * The multi virtual device settings. If empty (not set), it will create
-       * single virtual device on each visible GPU, according to the settings
-       * in "visible_device_list" above. Otherwise, the number of elements in the
-       * list must be the same as the number of visible GPUs (after
-       * "visible_device_list" filtering if it is set), and the string represented
-       * device names (e.g. /device:GPU:&lt;id&gt;) will refer to the virtual
-       * devices and have the &lt;id&gt; field assigned sequentially starting from 0,
-       * according to the order they appear in this list and the "memory_limit"
-       * list inside each element. For example,
-       *   visible_device_list = "1,0"
-       *   virtual_devices { memory_limit: 1GB memory_limit: 2GB }
-       *   virtual_devices {}
-       * will create three virtual devices as:
-       *   /device:GPU:0 -&gt; visible GPU 1 with 1GB memory
-       *   /device:GPU:1 -&gt; visible GPU 1 with 2GB memory
-       *   /device:GPU:2 -&gt; visible GPU 0 with all available memory
-       * NOTE:
-       * 1. It's invalid to set both this and "per_process_gpu_memory_fraction"
-       *    at the same time.
-       * 2. Currently this setting is per-process, not per-session. Using
-       *    different settings in different sessions within same process will
-       *    result in undefined behavior.
-       * </pre>
-       *
-       * <code>repeated .tensorflow.GPUOptions.Experimental.VirtualDevices virtual_devices = 1;</code>
-       */
-      public java.util.List<org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevices.Builder> 
-           getVirtualDevicesBuilderList() {
-        return getVirtualDevicesFieldBuilder().getBuilderList();
-      }
-      private com.google.protobuf.RepeatedFieldBuilderV3<
-          org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevices, org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevices.Builder, org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevicesOrBuilder> 
-          getVirtualDevicesFieldBuilder() {
-        if (virtualDevicesBuilder_ == null) {
-          virtualDevicesBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
-              org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevices, org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevices.Builder, org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevicesOrBuilder>(
-                  virtualDevices_,
-                  ((bitField0_ & 0x00000001) != 0),
-                  getParentForChildren(),
-                  isClean());
-          virtualDevices_ = null;
-        }
-        return virtualDevicesBuilder_;
-      }
-
-      private boolean useUnifiedMemory_ ;
-      /**
-       * <pre>
-       * If true, uses CUDA unified memory for memory allocations. If
-       * per_process_gpu_memory_fraction option is greater than 1.0, then unified
-       * memory is used regardless of the value for this field. See comments for
-       * per_process_gpu_memory_fraction field for more details and requirements
-       * of the unified memory. This option is useful to oversubscribe memory if
-       * multiple processes are sharing a single GPU while individually using less
-       * than 1.0 per process memory fraction.
-       * </pre>
-       *
-       * <code>bool use_unified_memory = 2;</code>
-       */
-      public boolean getUseUnifiedMemory() {
-        return useUnifiedMemory_;
-      }
-      /**
-       * <pre>
-       * If true, uses CUDA unified memory for memory allocations. If
-       * per_process_gpu_memory_fraction option is greater than 1.0, then unified
-       * memory is used regardless of the value for this field. See comments for
-       * per_process_gpu_memory_fraction field for more details and requirements
-       * of the unified memory. This option is useful to oversubscribe memory if
-       * multiple processes are sharing a single GPU while individually using less
-       * than 1.0 per process memory fraction.
-       * </pre>
-       *
-       * <code>bool use_unified_memory = 2;</code>
-       */
-      public Builder setUseUnifiedMemory(boolean value) {
-        
-        useUnifiedMemory_ = value;
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * If true, uses CUDA unified memory for memory allocations. If
-       * per_process_gpu_memory_fraction option is greater than 1.0, then unified
-       * memory is used regardless of the value for this field. See comments for
-       * per_process_gpu_memory_fraction field for more details and requirements
-       * of the unified memory. This option is useful to oversubscribe memory if
-       * multiple processes are sharing a single GPU while individually using less
-       * than 1.0 per process memory fraction.
-       * </pre>
-       *
-       * <code>bool use_unified_memory = 2;</code>
-       */
-      public Builder clearUseUnifiedMemory() {
-        
-        useUnifiedMemory_ = false;
-        onChanged();
-        return this;
-      }
-
-      private int numDevToDevCopyStreams_ ;
-      /**
-       * <pre>
-       * If &gt; 1, the number of device-to-device copy streams to create
-       * for each GPUDevice.  Default value is 0, which is automatically
-       * converted to 1.
-       * </pre>
-       *
-       * <code>int32 num_dev_to_dev_copy_streams = 3;</code>
-       */
-      public int getNumDevToDevCopyStreams() {
-        return numDevToDevCopyStreams_;
-      }
-      /**
-       * <pre>
-       * If &gt; 1, the number of device-to-device copy streams to create
-       * for each GPUDevice.  Default value is 0, which is automatically
-       * converted to 1.
-       * </pre>
-       *
-       * <code>int32 num_dev_to_dev_copy_streams = 3;</code>
-       */
-      public Builder setNumDevToDevCopyStreams(int value) {
-        
-        numDevToDevCopyStreams_ = value;
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * If &gt; 1, the number of device-to-device copy streams to create
-       * for each GPUDevice.  Default value is 0, which is automatically
-       * converted to 1.
-       * </pre>
-       *
-       * <code>int32 num_dev_to_dev_copy_streams = 3;</code>
-       */
-      public Builder clearNumDevToDevCopyStreams() {
-        
-        numDevToDevCopyStreams_ = 0;
-        onChanged();
-        return this;
-      }
-
-      private java.lang.Object collectiveRingOrder_ = "";
-      /**
-       * <pre>
-       * If non-empty, defines a good GPU ring order on a single worker based on
-       * device interconnect.  This assumes that all workers have the same GPU
-       * topology.  Specify as a comma-separated string, e.g. "3,2,1,0,7,6,5,4".
-       * This ring order is used by the RingReducer implementation of
-       * CollectiveReduce, and serves as an override to automatic ring order
-       * generation in OrderTaskDeviceMap() during CollectiveParam resolution.
-       * </pre>
-       *
-       * <code>string collective_ring_order = 4;</code>
-       */
-      public java.lang.String getCollectiveRingOrder() {
-        java.lang.Object ref = collectiveRingOrder_;
-        if (!(ref instanceof java.lang.String)) {
-          com.google.protobuf.ByteString bs =
-              (com.google.protobuf.ByteString) ref;
-          java.lang.String s = bs.toStringUtf8();
-          collectiveRingOrder_ = s;
-          return s;
-        } else {
-          return (java.lang.String) ref;
-        }
-      }
-      /**
-       * <pre>
-       * If non-empty, defines a good GPU ring order on a single worker based on
-       * device interconnect.  This assumes that all workers have the same GPU
-       * topology.  Specify as a comma-separated string, e.g. "3,2,1,0,7,6,5,4".
-       * This ring order is used by the RingReducer implementation of
-       * CollectiveReduce, and serves as an override to automatic ring order
-       * generation in OrderTaskDeviceMap() during CollectiveParam resolution.
-       * </pre>
-       *
-       * <code>string collective_ring_order = 4;</code>
-       */
-      public com.google.protobuf.ByteString
-          getCollectiveRingOrderBytes() {
-        java.lang.Object ref = collectiveRingOrder_;
-        if (ref instanceof String) {
-          com.google.protobuf.ByteString b = 
-              com.google.protobuf.ByteString.copyFromUtf8(
-                  (java.lang.String) ref);
-          collectiveRingOrder_ = b;
-          return b;
-        } else {
-          return (com.google.protobuf.ByteString) ref;
-        }
-      }
-      /**
-       * <pre>
-       * If non-empty, defines a good GPU ring order on a single worker based on
-       * device interconnect.  This assumes that all workers have the same GPU
-       * topology.  Specify as a comma-separated string, e.g. "3,2,1,0,7,6,5,4".
-       * This ring order is used by the RingReducer implementation of
-       * CollectiveReduce, and serves as an override to automatic ring order
-       * generation in OrderTaskDeviceMap() during CollectiveParam resolution.
-       * </pre>
-       *
-       * <code>string collective_ring_order = 4;</code>
-       */
-      public Builder setCollectiveRingOrder(
-          java.lang.String value) {
-        if (value == null) {
-    throw new NullPointerException();
-  }
-  
-        collectiveRingOrder_ = value;
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * If non-empty, defines a good GPU ring order on a single worker based on
-       * device interconnect.  This assumes that all workers have the same GPU
-       * topology.  Specify as a comma-separated string, e.g. "3,2,1,0,7,6,5,4".
-       * This ring order is used by the RingReducer implementation of
-       * CollectiveReduce, and serves as an override to automatic ring order
-       * generation in OrderTaskDeviceMap() during CollectiveParam resolution.
-       * </pre>
-       *
-       * <code>string collective_ring_order = 4;</code>
-       */
-      public Builder clearCollectiveRingOrder() {
-        
-        collectiveRingOrder_ = getDefaultInstance().getCollectiveRingOrder();
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * If non-empty, defines a good GPU ring order on a single worker based on
-       * device interconnect.  This assumes that all workers have the same GPU
-       * topology.  Specify as a comma-separated string, e.g. "3,2,1,0,7,6,5,4".
-       * This ring order is used by the RingReducer implementation of
-       * CollectiveReduce, and serves as an override to automatic ring order
-       * generation in OrderTaskDeviceMap() during CollectiveParam resolution.
-       * </pre>
-       *
-       * <code>string collective_ring_order = 4;</code>
-       */
-      public Builder setCollectiveRingOrderBytes(
-          com.google.protobuf.ByteString value) {
-        if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-        
-        collectiveRingOrder_ = value;
-        onChanged();
-        return this;
-      }
-
-      private boolean timestampedAllocator_ ;
-      /**
-       * <pre>
-       * If true then extra work is done by GPUDevice and GPUBFCAllocator to
-       * keep track of when GPU memory is freed and when kernels actually
-       * complete so that we can know when a nominally free memory chunk
-       * is really not subject to pending use.
-       * </pre>
-       *
-       * <code>bool timestamped_allocator = 5;</code>
-       */
-      public boolean getTimestampedAllocator() {
-        return timestampedAllocator_;
-      }
-      /**
-       * <pre>
-       * If true then extra work is done by GPUDevice and GPUBFCAllocator to
-       * keep track of when GPU memory is freed and when kernels actually
-       * complete so that we can know when a nominally free memory chunk
-       * is really not subject to pending use.
-       * </pre>
-       *
-       * <code>bool timestamped_allocator = 5;</code>
-       */
-      public Builder setTimestampedAllocator(boolean value) {
-        
-        timestampedAllocator_ = value;
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * If true then extra work is done by GPUDevice and GPUBFCAllocator to
-       * keep track of when GPU memory is freed and when kernels actually
-       * complete so that we can know when a nominally free memory chunk
-       * is really not subject to pending use.
-       * </pre>
-       *
-       * <code>bool timestamped_allocator = 5;</code>
-       */
-      public Builder clearTimestampedAllocator() {
-        
-        timestampedAllocator_ = false;
-        onChanged();
-        return this;
-      }
-
-      private int kernelTrackerMaxInterval_ ;
-      /**
-       * <pre>
-       * Parameters for GPUKernelTracker.  By default no kernel tracking is done.
-       * Note that timestamped_allocator is only effective if some tracking is
-       * specified.
-       * If kernel_tracker_max_interval = n &gt; 0, then a tracking event
-       * is inserted after every n kernels without an event.
-       * </pre>
-       *
-       * <code>int32 kernel_tracker_max_interval = 7;</code>
-       */
-      public int getKernelTrackerMaxInterval() {
-        return kernelTrackerMaxInterval_;
-      }
-      /**
-       * <pre>
-       * Parameters for GPUKernelTracker.  By default no kernel tracking is done.
-       * Note that timestamped_allocator is only effective if some tracking is
-       * specified.
-       * If kernel_tracker_max_interval = n &gt; 0, then a tracking event
-       * is inserted after every n kernels without an event.
-       * </pre>
-       *
-       * <code>int32 kernel_tracker_max_interval = 7;</code>
-       */
-      public Builder setKernelTrackerMaxInterval(int value) {
-        
-        kernelTrackerMaxInterval_ = value;
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * Parameters for GPUKernelTracker.  By default no kernel tracking is done.
-       * Note that timestamped_allocator is only effective if some tracking is
-       * specified.
-       * If kernel_tracker_max_interval = n &gt; 0, then a tracking event
-       * is inserted after every n kernels without an event.
-       * </pre>
-       *
-       * <code>int32 kernel_tracker_max_interval = 7;</code>
-       */
-      public Builder clearKernelTrackerMaxInterval() {
-        
-        kernelTrackerMaxInterval_ = 0;
-        onChanged();
-        return this;
-      }
-
-      private int kernelTrackerMaxBytes_ ;
-      /**
-       * <pre>
-       * If kernel_tracker_max_bytes = n &gt; 0, then a tracking event is
-       * inserted after every series of kernels allocating a sum of
-       * memory &gt;= n.  If one kernel allocates b * n bytes, then one
-       * event will be inserted after it, but it will count as b against
-       * the pending limit.
-       * </pre>
-       *
-       * <code>int32 kernel_tracker_max_bytes = 8;</code>
-       */
-      public int getKernelTrackerMaxBytes() {
-        return kernelTrackerMaxBytes_;
-      }
-      /**
-       * <pre>
-       * If kernel_tracker_max_bytes = n &gt; 0, then a tracking event is
-       * inserted after every series of kernels allocating a sum of
-       * memory &gt;= n.  If one kernel allocates b * n bytes, then one
-       * event will be inserted after it, but it will count as b against
-       * the pending limit.
-       * </pre>
-       *
-       * <code>int32 kernel_tracker_max_bytes = 8;</code>
-       */
-      public Builder setKernelTrackerMaxBytes(int value) {
-        
-        kernelTrackerMaxBytes_ = value;
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * If kernel_tracker_max_bytes = n &gt; 0, then a tracking event is
-       * inserted after every series of kernels allocating a sum of
-       * memory &gt;= n.  If one kernel allocates b * n bytes, then one
-       * event will be inserted after it, but it will count as b against
-       * the pending limit.
-       * </pre>
-       *
-       * <code>int32 kernel_tracker_max_bytes = 8;</code>
-       */
-      public Builder clearKernelTrackerMaxBytes() {
-        
-        kernelTrackerMaxBytes_ = 0;
-        onChanged();
-        return this;
-      }
-
-      private int kernelTrackerMaxPending_ ;
-      /**
-       * <pre>
-       * If kernel_tracker_max_pending &gt; 0 then no more than this many
-       * tracking events can be outstanding at a time.  An attempt to
-       * launch an additional kernel will stall until an event
-       * completes.
-       * </pre>
-       *
-       * <code>int32 kernel_tracker_max_pending = 9;</code>
-       */
-      public int getKernelTrackerMaxPending() {
-        return kernelTrackerMaxPending_;
-      }
-      /**
-       * <pre>
-       * If kernel_tracker_max_pending &gt; 0 then no more than this many
-       * tracking events can be outstanding at a time.  An attempt to
-       * launch an additional kernel will stall until an event
-       * completes.
-       * </pre>
-       *
-       * <code>int32 kernel_tracker_max_pending = 9;</code>
-       */
-      public Builder setKernelTrackerMaxPending(int value) {
-        
-        kernelTrackerMaxPending_ = value;
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * If kernel_tracker_max_pending &gt; 0 then no more than this many
-       * tracking events can be outstanding at a time.  An attempt to
-       * launch an additional kernel will stall until an event
-       * completes.
-       * </pre>
-       *
-       * <code>int32 kernel_tracker_max_pending = 9;</code>
-       */
-      public Builder clearKernelTrackerMaxPending() {
-        
-        kernelTrackerMaxPending_ = 0;
-        onChanged();
-        return this;
-      }
-      @java.lang.Override
-      public final Builder setUnknownFields(
-          final com.google.protobuf.UnknownFieldSet unknownFields) {
-        return super.setUnknownFields(unknownFields);
-      }
-
-      @java.lang.Override
-      public final Builder mergeUnknownFields(
-          final com.google.protobuf.UnknownFieldSet unknownFields) {
-        return super.mergeUnknownFields(unknownFields);
-      }
-
-
-      // @@protoc_insertion_point(builder_scope:tensorflow.GPUOptions.Experimental)
-    }
-
-    // @@protoc_insertion_point(class_scope:tensorflow.GPUOptions.Experimental)
-    private static final org.tensorflow.proto.framework.GPUOptions.Experimental DEFAULT_INSTANCE;
-    static {
-      DEFAULT_INSTANCE = new org.tensorflow.proto.framework.GPUOptions.Experimental();
-    }
-
-    public static org.tensorflow.proto.framework.GPUOptions.Experimental getDefaultInstance() {
-      return DEFAULT_INSTANCE;
-    }
-
-    private static final com.google.protobuf.Parser<Experimental>
-        PARSER = new com.google.protobuf.AbstractParser<Experimental>() {
-      @java.lang.Override
-      public Experimental parsePartialFrom(
-          com.google.protobuf.CodedInputStream input,
-          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-          throws com.google.protobuf.InvalidProtocolBufferException {
-        return new Experimental(input, extensionRegistry);
-      }
-    };
-
-    public static com.google.protobuf.Parser<Experimental> parser() {
-      return PARSER;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Parser<Experimental> getParserForType() {
-      return PARSER;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.GPUOptions.Experimental getDefaultInstanceForType() {
-      return DEFAULT_INSTANCE;
-    }
-
-  }
-
-  public static final int PER_PROCESS_GPU_MEMORY_FRACTION_FIELD_NUMBER = 1;
-  private double perProcessGpuMemoryFraction_;
-  /**
-   * <pre>
-   * Fraction of the available GPU memory to allocate for each process.
-   * 1 means to allocate all of the GPU memory, 0.5 means the process
-   * allocates up to ~50% of the available GPU memory.
-   * GPU memory is pre-allocated unless the allow_growth option is enabled.
-   * If greater than 1.0, uses CUDA unified memory to potentially oversubscribe
-   * the amount of memory available on the GPU device by using host memory as a
-   * swap space. Accessing memory not available on the device will be
-   * significantly slower as that would require memory transfer between the host
-   * and the device. Options to reduce the memory requirement should be
-   * considered before enabling this option as this may come with a negative
-   * performance impact. Oversubscription using the unified memory requires
-   * Pascal class or newer GPUs and it is currently only supported on the Linux
-   * operating system. See
-   * https://docs.nvidia.com/cuda/cuda-c-programming-guide/index.html#um-requirements
-   * for the detailed requirements.
-   * </pre>
-   *
-   * <code>double per_process_gpu_memory_fraction = 1;</code>
-   */
-  public double getPerProcessGpuMemoryFraction() {
-    return perProcessGpuMemoryFraction_;
-  }
-
-  public static final int ALLOW_GROWTH_FIELD_NUMBER = 4;
-  private boolean allowGrowth_;
-  /**
-   * <pre>
-   * If true, the allocator does not pre-allocate the entire specified
-   * GPU memory region, instead starting small and growing as needed.
-   * </pre>
-   *
-   * <code>bool allow_growth = 4;</code>
-   */
-  public boolean getAllowGrowth() {
-    return allowGrowth_;
-  }
-
-  public static final int ALLOCATOR_TYPE_FIELD_NUMBER = 2;
-  private volatile java.lang.Object allocatorType_;
-  /**
-   * <pre>
-   * The type of GPU allocation strategy to use.
-   * Allowed values:
-   * "": The empty string (default) uses a system-chosen default
-   *     which may change over time.
-   * "BFC": A "Best-fit with coalescing" algorithm, simplified from a
-   *        version of dlmalloc.
-   * </pre>
-   *
-   * <code>string allocator_type = 2;</code>
-   */
-  public java.lang.String getAllocatorType() {
-    java.lang.Object ref = allocatorType_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      allocatorType_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * The type of GPU allocation strategy to use.
-   * Allowed values:
-   * "": The empty string (default) uses a system-chosen default
-   *     which may change over time.
-   * "BFC": A "Best-fit with coalescing" algorithm, simplified from a
-   *        version of dlmalloc.
-   * </pre>
-   *
-   * <code>string allocator_type = 2;</code>
-   */
-  public com.google.protobuf.ByteString
-      getAllocatorTypeBytes() {
-    java.lang.Object ref = allocatorType_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      allocatorType_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int DEFERRED_DELETION_BYTES_FIELD_NUMBER = 3;
-  private long deferredDeletionBytes_;
-  /**
-   * <pre>
-   * Delay deletion of up to this many bytes to reduce the number of
-   * interactions with gpu driver code.  If 0, the system chooses
-   * a reasonable default (several MBs).
-   * </pre>
-   *
-   * <code>int64 deferred_deletion_bytes = 3;</code>
-   */
-  public long getDeferredDeletionBytes() {
-    return deferredDeletionBytes_;
-  }
-
-  public static final int VISIBLE_DEVICE_LIST_FIELD_NUMBER = 5;
-  private volatile java.lang.Object visibleDeviceList_;
-  /**
-   * <pre>
-   * A comma-separated list of GPU ids that determines the 'visible'
-   * to 'virtual' mapping of GPU devices.  For example, if TensorFlow
-   * can see 8 GPU devices in the process, and one wanted to map
-   * visible GPU devices 5 and 3 as "/device:GPU:0", and "/device:GPU:1",
-   * then one would specify this field as "5,3".  This field is similar in
-   * spirit to the CUDA_VISIBLE_DEVICES environment variable, except
-   * it applies to the visible GPU devices in the process.
-   * NOTE:
-   * 1. The GPU driver provides the process with the visible GPUs
-   *    in an order which is not guaranteed to have any correlation to
-   *    the *physical* GPU id in the machine.  This field is used for
-   *    remapping "visible" to "virtual", which means this operates only
-   *    after the process starts.  Users are required to use vendor
-   *    specific mechanisms (e.g., CUDA_VISIBLE_DEVICES) to control the
-   *    physical to visible device mapping prior to invoking TensorFlow.
-   * 2. In the code, the ids in this list are also called "platform GPU id"s,
-   *    and the 'virtual' ids of GPU devices (i.e. the ids in the device
-   *    name "/device:GPU:&lt;id&gt;") are also called "TF GPU id"s. Please
-   *    refer to third_party/tensorflow/core/common_runtime/gpu/gpu_id.h
-   *    for more information.
-   * </pre>
-   *
-   * <code>string visible_device_list = 5;</code>
-   */
-  public java.lang.String getVisibleDeviceList() {
-    java.lang.Object ref = visibleDeviceList_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      visibleDeviceList_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * A comma-separated list of GPU ids that determines the 'visible'
-   * to 'virtual' mapping of GPU devices.  For example, if TensorFlow
-   * can see 8 GPU devices in the process, and one wanted to map
-   * visible GPU devices 5 and 3 as "/device:GPU:0", and "/device:GPU:1",
-   * then one would specify this field as "5,3".  This field is similar in
-   * spirit to the CUDA_VISIBLE_DEVICES environment variable, except
-   * it applies to the visible GPU devices in the process.
-   * NOTE:
-   * 1. The GPU driver provides the process with the visible GPUs
-   *    in an order which is not guaranteed to have any correlation to
-   *    the *physical* GPU id in the machine.  This field is used for
-   *    remapping "visible" to "virtual", which means this operates only
-   *    after the process starts.  Users are required to use vendor
-   *    specific mechanisms (e.g., CUDA_VISIBLE_DEVICES) to control the
-   *    physical to visible device mapping prior to invoking TensorFlow.
-   * 2. In the code, the ids in this list are also called "platform GPU id"s,
-   *    and the 'virtual' ids of GPU devices (i.e. the ids in the device
-   *    name "/device:GPU:&lt;id&gt;") are also called "TF GPU id"s. Please
-   *    refer to third_party/tensorflow/core/common_runtime/gpu/gpu_id.h
-   *    for more information.
-   * </pre>
-   *
-   * <code>string visible_device_list = 5;</code>
-   */
-  public com.google.protobuf.ByteString
-      getVisibleDeviceListBytes() {
-    java.lang.Object ref = visibleDeviceList_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      visibleDeviceList_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int POLLING_ACTIVE_DELAY_USECS_FIELD_NUMBER = 6;
-  private int pollingActiveDelayUsecs_;
-  /**
-   * <pre>
-   * In the event polling loop sleep this many microseconds between
-   * PollEvents calls, when the queue is not empty.  If value is not
-   * set or set to 0, gets set to a non-zero default.
-   * </pre>
-   *
-   * <code>int32 polling_active_delay_usecs = 6;</code>
-   */
-  public int getPollingActiveDelayUsecs() {
-    return pollingActiveDelayUsecs_;
-  }
-
-  public static final int POLLING_INACTIVE_DELAY_MSECS_FIELD_NUMBER = 7;
-  private int pollingInactiveDelayMsecs_;
-  /**
-   * <pre>
-   * This field is deprecated and ignored.
-   * </pre>
-   *
-   * <code>int32 polling_inactive_delay_msecs = 7;</code>
-   */
-  public int getPollingInactiveDelayMsecs() {
-    return pollingInactiveDelayMsecs_;
-  }
-
-  public static final int FORCE_GPU_COMPATIBLE_FIELD_NUMBER = 8;
-  private boolean forceGpuCompatible_;
-  /**
-   * <pre>
-   * Force all tensors to be gpu_compatible. On a GPU-enabled TensorFlow,
-   * enabling this option forces all CPU tensors to be allocated with Cuda
-   * pinned memory. Normally, TensorFlow will infer which tensors should be
-   * allocated as the pinned memory. But in case where the inference is
-   * incomplete, this option can significantly speed up the cross-device memory
-   * copy performance as long as it fits the memory.
-   * Note that this option is not something that should be
-   * enabled by default for unknown or very large models, since all Cuda pinned
-   * memory is unpageable, having too much pinned memory might negatively impact
-   * the overall host system performance.
-   * </pre>
-   *
-   * <code>bool force_gpu_compatible = 8;</code>
-   */
-  public boolean getForceGpuCompatible() {
-    return forceGpuCompatible_;
-  }
-
-  public static final int EXPERIMENTAL_FIELD_NUMBER = 9;
-  private org.tensorflow.proto.framework.GPUOptions.Experimental experimental_;
-  /**
-   * <pre>
-   * Everything inside experimental is subject to change and is not subject
-   * to API stability guarantees in
-   * https://www.tensorflow.org/guide/version_compat.
-   * </pre>
-   *
-   * <code>.tensorflow.GPUOptions.Experimental experimental = 9;</code>
-   */
-  public boolean hasExperimental() {
-    return experimental_ != null;
-  }
-  /**
-   * <pre>
-   * Everything inside experimental is subject to change and is not subject
-   * to API stability guarantees in
-   * https://www.tensorflow.org/guide/version_compat.
-   * </pre>
-   *
-   * <code>.tensorflow.GPUOptions.Experimental experimental = 9;</code>
-   */
-  public org.tensorflow.proto.framework.GPUOptions.Experimental getExperimental() {
-    return experimental_ == null ? org.tensorflow.proto.framework.GPUOptions.Experimental.getDefaultInstance() : experimental_;
-  }
-  /**
-   * <pre>
-   * Everything inside experimental is subject to change and is not subject
-   * to API stability guarantees in
-   * https://www.tensorflow.org/guide/version_compat.
-   * </pre>
-   *
-   * <code>.tensorflow.GPUOptions.Experimental experimental = 9;</code>
-   */
-  public org.tensorflow.proto.framework.GPUOptions.ExperimentalOrBuilder getExperimentalOrBuilder() {
-    return getExperimental();
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    if (perProcessGpuMemoryFraction_ != 0D) {
-      output.writeDouble(1, perProcessGpuMemoryFraction_);
-    }
-    if (!getAllocatorTypeBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 2, allocatorType_);
-    }
-    if (deferredDeletionBytes_ != 0L) {
-      output.writeInt64(3, deferredDeletionBytes_);
-    }
-    if (allowGrowth_ != false) {
-      output.writeBool(4, allowGrowth_);
-    }
-    if (!getVisibleDeviceListBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 5, visibleDeviceList_);
-    }
-    if (pollingActiveDelayUsecs_ != 0) {
-      output.writeInt32(6, pollingActiveDelayUsecs_);
-    }
-    if (pollingInactiveDelayMsecs_ != 0) {
-      output.writeInt32(7, pollingInactiveDelayMsecs_);
-    }
-    if (forceGpuCompatible_ != false) {
-      output.writeBool(8, forceGpuCompatible_);
-    }
-    if (experimental_ != null) {
-      output.writeMessage(9, getExperimental());
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (perProcessGpuMemoryFraction_ != 0D) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeDoubleSize(1, perProcessGpuMemoryFraction_);
-    }
-    if (!getAllocatorTypeBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(2, allocatorType_);
-    }
-    if (deferredDeletionBytes_ != 0L) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt64Size(3, deferredDeletionBytes_);
-    }
-    if (allowGrowth_ != false) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeBoolSize(4, allowGrowth_);
-    }
-    if (!getVisibleDeviceListBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(5, visibleDeviceList_);
-    }
-    if (pollingActiveDelayUsecs_ != 0) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt32Size(6, pollingActiveDelayUsecs_);
-    }
-    if (pollingInactiveDelayMsecs_ != 0) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt32Size(7, pollingInactiveDelayMsecs_);
-    }
-    if (forceGpuCompatible_ != false) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeBoolSize(8, forceGpuCompatible_);
-    }
-    if (experimental_ != null) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(9, getExperimental());
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.framework.GPUOptions)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.framework.GPUOptions other = (org.tensorflow.proto.framework.GPUOptions) obj;
-
-    if (java.lang.Double.doubleToLongBits(getPerProcessGpuMemoryFraction())
-        != java.lang.Double.doubleToLongBits(
-            other.getPerProcessGpuMemoryFraction())) return false;
-    if (getAllowGrowth()
-        != other.getAllowGrowth()) return false;
-    if (!getAllocatorType()
-        .equals(other.getAllocatorType())) return false;
-    if (getDeferredDeletionBytes()
-        != other.getDeferredDeletionBytes()) return false;
-    if (!getVisibleDeviceList()
-        .equals(other.getVisibleDeviceList())) return false;
-    if (getPollingActiveDelayUsecs()
-        != other.getPollingActiveDelayUsecs()) return false;
-    if (getPollingInactiveDelayMsecs()
-        != other.getPollingInactiveDelayMsecs()) return false;
-    if (getForceGpuCompatible()
-        != other.getForceGpuCompatible()) return false;
-    if (hasExperimental() != other.hasExperimental()) return false;
-    if (hasExperimental()) {
-      if (!getExperimental()
-          .equals(other.getExperimental())) return false;
-    }
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    hash = (37 * hash) + PER_PROCESS_GPU_MEMORY_FRACTION_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-        java.lang.Double.doubleToLongBits(getPerProcessGpuMemoryFraction()));
-    hash = (37 * hash) + ALLOW_GROWTH_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashBoolean(
-        getAllowGrowth());
-    hash = (37 * hash) + ALLOCATOR_TYPE_FIELD_NUMBER;
-    hash = (53 * hash) + getAllocatorType().hashCode();
-    hash = (37 * hash) + DEFERRED_DELETION_BYTES_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-        getDeferredDeletionBytes());
-    hash = (37 * hash) + VISIBLE_DEVICE_LIST_FIELD_NUMBER;
-    hash = (53 * hash) + getVisibleDeviceList().hashCode();
-    hash = (37 * hash) + POLLING_ACTIVE_DELAY_USECS_FIELD_NUMBER;
-    hash = (53 * hash) + getPollingActiveDelayUsecs();
-    hash = (37 * hash) + POLLING_INACTIVE_DELAY_MSECS_FIELD_NUMBER;
-    hash = (53 * hash) + getPollingInactiveDelayMsecs();
-    hash = (37 * hash) + FORCE_GPU_COMPATIBLE_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashBoolean(
-        getForceGpuCompatible());
-    if (hasExperimental()) {
-      hash = (37 * hash) + EXPERIMENTAL_FIELD_NUMBER;
-      hash = (53 * hash) + getExperimental().hashCode();
-    }
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.framework.GPUOptions parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.GPUOptions parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.GPUOptions parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.GPUOptions parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.GPUOptions parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.GPUOptions parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.GPUOptions parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.GPUOptions parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.GPUOptions parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.GPUOptions parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.GPUOptions parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.GPUOptions parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.framework.GPUOptions prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * Protobuf type {@code tensorflow.GPUOptions}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.GPUOptions)
-      org.tensorflow.proto.framework.GPUOptionsOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_GPUOptions_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_GPUOptions_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.GPUOptions.class, org.tensorflow.proto.framework.GPUOptions.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.framework.GPUOptions.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      perProcessGpuMemoryFraction_ = 0D;
-
-      allowGrowth_ = false;
-
-      allocatorType_ = "";
-
-      deferredDeletionBytes_ = 0L;
-
-      visibleDeviceList_ = "";
-
-      pollingActiveDelayUsecs_ = 0;
-
-      pollingInactiveDelayMsecs_ = 0;
-
-      forceGpuCompatible_ = false;
-
-      if (experimentalBuilder_ == null) {
-        experimental_ = null;
-      } else {
-        experimental_ = null;
-        experimentalBuilder_ = null;
-      }
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_GPUOptions_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.GPUOptions getDefaultInstanceForType() {
-      return org.tensorflow.proto.framework.GPUOptions.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.GPUOptions build() {
-      org.tensorflow.proto.framework.GPUOptions result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.GPUOptions buildPartial() {
-      org.tensorflow.proto.framework.GPUOptions result = new org.tensorflow.proto.framework.GPUOptions(this);
-      result.perProcessGpuMemoryFraction_ = perProcessGpuMemoryFraction_;
-      result.allowGrowth_ = allowGrowth_;
-      result.allocatorType_ = allocatorType_;
-      result.deferredDeletionBytes_ = deferredDeletionBytes_;
-      result.visibleDeviceList_ = visibleDeviceList_;
-      result.pollingActiveDelayUsecs_ = pollingActiveDelayUsecs_;
-      result.pollingInactiveDelayMsecs_ = pollingInactiveDelayMsecs_;
-      result.forceGpuCompatible_ = forceGpuCompatible_;
-      if (experimentalBuilder_ == null) {
-        result.experimental_ = experimental_;
-      } else {
-        result.experimental_ = experimentalBuilder_.build();
-      }
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.framework.GPUOptions) {
-        return mergeFrom((org.tensorflow.proto.framework.GPUOptions)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.framework.GPUOptions other) {
-      if (other == org.tensorflow.proto.framework.GPUOptions.getDefaultInstance()) return this;
-      if (other.getPerProcessGpuMemoryFraction() != 0D) {
-        setPerProcessGpuMemoryFraction(other.getPerProcessGpuMemoryFraction());
-      }
-      if (other.getAllowGrowth() != false) {
-        setAllowGrowth(other.getAllowGrowth());
-      }
-      if (!other.getAllocatorType().isEmpty()) {
-        allocatorType_ = other.allocatorType_;
-        onChanged();
-      }
-      if (other.getDeferredDeletionBytes() != 0L) {
-        setDeferredDeletionBytes(other.getDeferredDeletionBytes());
-      }
-      if (!other.getVisibleDeviceList().isEmpty()) {
-        visibleDeviceList_ = other.visibleDeviceList_;
-        onChanged();
-      }
-      if (other.getPollingActiveDelayUsecs() != 0) {
-        setPollingActiveDelayUsecs(other.getPollingActiveDelayUsecs());
-      }
-      if (other.getPollingInactiveDelayMsecs() != 0) {
-        setPollingInactiveDelayMsecs(other.getPollingInactiveDelayMsecs());
-      }
-      if (other.getForceGpuCompatible() != false) {
-        setForceGpuCompatible(other.getForceGpuCompatible());
-      }
-      if (other.hasExperimental()) {
-        mergeExperimental(other.getExperimental());
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.framework.GPUOptions parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.framework.GPUOptions) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-
-    private double perProcessGpuMemoryFraction_ ;
-    /**
-     * <pre>
-     * Fraction of the available GPU memory to allocate for each process.
-     * 1 means to allocate all of the GPU memory, 0.5 means the process
-     * allocates up to ~50% of the available GPU memory.
-     * GPU memory is pre-allocated unless the allow_growth option is enabled.
-     * If greater than 1.0, uses CUDA unified memory to potentially oversubscribe
-     * the amount of memory available on the GPU device by using host memory as a
-     * swap space. Accessing memory not available on the device will be
-     * significantly slower as that would require memory transfer between the host
-     * and the device. Options to reduce the memory requirement should be
-     * considered before enabling this option as this may come with a negative
-     * performance impact. Oversubscription using the unified memory requires
-     * Pascal class or newer GPUs and it is currently only supported on the Linux
-     * operating system. See
-     * https://docs.nvidia.com/cuda/cuda-c-programming-guide/index.html#um-requirements
-     * for the detailed requirements.
-     * </pre>
-     *
-     * <code>double per_process_gpu_memory_fraction = 1;</code>
-     */
-    public double getPerProcessGpuMemoryFraction() {
-      return perProcessGpuMemoryFraction_;
-    }
-    /**
-     * <pre>
-     * Fraction of the available GPU memory to allocate for each process.
-     * 1 means to allocate all of the GPU memory, 0.5 means the process
-     * allocates up to ~50% of the available GPU memory.
-     * GPU memory is pre-allocated unless the allow_growth option is enabled.
-     * If greater than 1.0, uses CUDA unified memory to potentially oversubscribe
-     * the amount of memory available on the GPU device by using host memory as a
-     * swap space. Accessing memory not available on the device will be
-     * significantly slower as that would require memory transfer between the host
-     * and the device. Options to reduce the memory requirement should be
-     * considered before enabling this option as this may come with a negative
-     * performance impact. Oversubscription using the unified memory requires
-     * Pascal class or newer GPUs and it is currently only supported on the Linux
-     * operating system. See
-     * https://docs.nvidia.com/cuda/cuda-c-programming-guide/index.html#um-requirements
-     * for the detailed requirements.
-     * </pre>
-     *
-     * <code>double per_process_gpu_memory_fraction = 1;</code>
-     */
-    public Builder setPerProcessGpuMemoryFraction(double value) {
-      
-      perProcessGpuMemoryFraction_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Fraction of the available GPU memory to allocate for each process.
-     * 1 means to allocate all of the GPU memory, 0.5 means the process
-     * allocates up to ~50% of the available GPU memory.
-     * GPU memory is pre-allocated unless the allow_growth option is enabled.
-     * If greater than 1.0, uses CUDA unified memory to potentially oversubscribe
-     * the amount of memory available on the GPU device by using host memory as a
-     * swap space. Accessing memory not available on the device will be
-     * significantly slower as that would require memory transfer between the host
-     * and the device. Options to reduce the memory requirement should be
-     * considered before enabling this option as this may come with a negative
-     * performance impact. Oversubscription using the unified memory requires
-     * Pascal class or newer GPUs and it is currently only supported on the Linux
-     * operating system. See
-     * https://docs.nvidia.com/cuda/cuda-c-programming-guide/index.html#um-requirements
-     * for the detailed requirements.
-     * </pre>
-     *
-     * <code>double per_process_gpu_memory_fraction = 1;</code>
-     */
-    public Builder clearPerProcessGpuMemoryFraction() {
-      
-      perProcessGpuMemoryFraction_ = 0D;
-      onChanged();
-      return this;
-    }
-
-    private boolean allowGrowth_ ;
-    /**
-     * <pre>
-     * If true, the allocator does not pre-allocate the entire specified
-     * GPU memory region, instead starting small and growing as needed.
-     * </pre>
-     *
-     * <code>bool allow_growth = 4;</code>
-     */
-    public boolean getAllowGrowth() {
-      return allowGrowth_;
-    }
-    /**
-     * <pre>
-     * If true, the allocator does not pre-allocate the entire specified
-     * GPU memory region, instead starting small and growing as needed.
-     * </pre>
-     *
-     * <code>bool allow_growth = 4;</code>
-     */
-    public Builder setAllowGrowth(boolean value) {
-      
-      allowGrowth_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * If true, the allocator does not pre-allocate the entire specified
-     * GPU memory region, instead starting small and growing as needed.
-     * </pre>
-     *
-     * <code>bool allow_growth = 4;</code>
-     */
-    public Builder clearAllowGrowth() {
-      
-      allowGrowth_ = false;
-      onChanged();
-      return this;
-    }
-
-    private java.lang.Object allocatorType_ = "";
-    /**
-     * <pre>
-     * The type of GPU allocation strategy to use.
-     * Allowed values:
-     * "": The empty string (default) uses a system-chosen default
-     *     which may change over time.
-     * "BFC": A "Best-fit with coalescing" algorithm, simplified from a
-     *        version of dlmalloc.
-     * </pre>
-     *
-     * <code>string allocator_type = 2;</code>
-     */
-    public java.lang.String getAllocatorType() {
-      java.lang.Object ref = allocatorType_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        allocatorType_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * The type of GPU allocation strategy to use.
-     * Allowed values:
-     * "": The empty string (default) uses a system-chosen default
-     *     which may change over time.
-     * "BFC": A "Best-fit with coalescing" algorithm, simplified from a
-     *        version of dlmalloc.
-     * </pre>
-     *
-     * <code>string allocator_type = 2;</code>
-     */
-    public com.google.protobuf.ByteString
-        getAllocatorTypeBytes() {
-      java.lang.Object ref = allocatorType_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        allocatorType_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * The type of GPU allocation strategy to use.
-     * Allowed values:
-     * "": The empty string (default) uses a system-chosen default
-     *     which may change over time.
-     * "BFC": A "Best-fit with coalescing" algorithm, simplified from a
-     *        version of dlmalloc.
-     * </pre>
-     *
-     * <code>string allocator_type = 2;</code>
-     */
-    public Builder setAllocatorType(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      allocatorType_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * The type of GPU allocation strategy to use.
-     * Allowed values:
-     * "": The empty string (default) uses a system-chosen default
-     *     which may change over time.
-     * "BFC": A "Best-fit with coalescing" algorithm, simplified from a
-     *        version of dlmalloc.
-     * </pre>
-     *
-     * <code>string allocator_type = 2;</code>
-     */
-    public Builder clearAllocatorType() {
-      
-      allocatorType_ = getDefaultInstance().getAllocatorType();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * The type of GPU allocation strategy to use.
-     * Allowed values:
-     * "": The empty string (default) uses a system-chosen default
-     *     which may change over time.
-     * "BFC": A "Best-fit with coalescing" algorithm, simplified from a
-     *        version of dlmalloc.
-     * </pre>
-     *
-     * <code>string allocator_type = 2;</code>
-     */
-    public Builder setAllocatorTypeBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      allocatorType_ = value;
-      onChanged();
-      return this;
-    }
-
-    private long deferredDeletionBytes_ ;
-    /**
-     * <pre>
-     * Delay deletion of up to this many bytes to reduce the number of
-     * interactions with gpu driver code.  If 0, the system chooses
-     * a reasonable default (several MBs).
-     * </pre>
-     *
-     * <code>int64 deferred_deletion_bytes = 3;</code>
-     */
-    public long getDeferredDeletionBytes() {
-      return deferredDeletionBytes_;
-    }
-    /**
-     * <pre>
-     * Delay deletion of up to this many bytes to reduce the number of
-     * interactions with gpu driver code.  If 0, the system chooses
-     * a reasonable default (several MBs).
-     * </pre>
-     *
-     * <code>int64 deferred_deletion_bytes = 3;</code>
-     */
-    public Builder setDeferredDeletionBytes(long value) {
-      
-      deferredDeletionBytes_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Delay deletion of up to this many bytes to reduce the number of
-     * interactions with gpu driver code.  If 0, the system chooses
-     * a reasonable default (several MBs).
-     * </pre>
-     *
-     * <code>int64 deferred_deletion_bytes = 3;</code>
-     */
-    public Builder clearDeferredDeletionBytes() {
-      
-      deferredDeletionBytes_ = 0L;
-      onChanged();
-      return this;
-    }
-
-    private java.lang.Object visibleDeviceList_ = "";
-    /**
-     * <pre>
-     * A comma-separated list of GPU ids that determines the 'visible'
-     * to 'virtual' mapping of GPU devices.  For example, if TensorFlow
-     * can see 8 GPU devices in the process, and one wanted to map
-     * visible GPU devices 5 and 3 as "/device:GPU:0", and "/device:GPU:1",
-     * then one would specify this field as "5,3".  This field is similar in
-     * spirit to the CUDA_VISIBLE_DEVICES environment variable, except
-     * it applies to the visible GPU devices in the process.
-     * NOTE:
-     * 1. The GPU driver provides the process with the visible GPUs
-     *    in an order which is not guaranteed to have any correlation to
-     *    the *physical* GPU id in the machine.  This field is used for
-     *    remapping "visible" to "virtual", which means this operates only
-     *    after the process starts.  Users are required to use vendor
-     *    specific mechanisms (e.g., CUDA_VISIBLE_DEVICES) to control the
-     *    physical to visible device mapping prior to invoking TensorFlow.
-     * 2. In the code, the ids in this list are also called "platform GPU id"s,
-     *    and the 'virtual' ids of GPU devices (i.e. the ids in the device
-     *    name "/device:GPU:&lt;id&gt;") are also called "TF GPU id"s. Please
-     *    refer to third_party/tensorflow/core/common_runtime/gpu/gpu_id.h
-     *    for more information.
-     * </pre>
-     *
-     * <code>string visible_device_list = 5;</code>
-     */
-    public java.lang.String getVisibleDeviceList() {
-      java.lang.Object ref = visibleDeviceList_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        visibleDeviceList_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * A comma-separated list of GPU ids that determines the 'visible'
-     * to 'virtual' mapping of GPU devices.  For example, if TensorFlow
-     * can see 8 GPU devices in the process, and one wanted to map
-     * visible GPU devices 5 and 3 as "/device:GPU:0", and "/device:GPU:1",
-     * then one would specify this field as "5,3".  This field is similar in
-     * spirit to the CUDA_VISIBLE_DEVICES environment variable, except
-     * it applies to the visible GPU devices in the process.
-     * NOTE:
-     * 1. The GPU driver provides the process with the visible GPUs
-     *    in an order which is not guaranteed to have any correlation to
-     *    the *physical* GPU id in the machine.  This field is used for
-     *    remapping "visible" to "virtual", which means this operates only
-     *    after the process starts.  Users are required to use vendor
-     *    specific mechanisms (e.g., CUDA_VISIBLE_DEVICES) to control the
-     *    physical to visible device mapping prior to invoking TensorFlow.
-     * 2. In the code, the ids in this list are also called "platform GPU id"s,
-     *    and the 'virtual' ids of GPU devices (i.e. the ids in the device
-     *    name "/device:GPU:&lt;id&gt;") are also called "TF GPU id"s. Please
-     *    refer to third_party/tensorflow/core/common_runtime/gpu/gpu_id.h
-     *    for more information.
-     * </pre>
-     *
-     * <code>string visible_device_list = 5;</code>
-     */
-    public com.google.protobuf.ByteString
-        getVisibleDeviceListBytes() {
-      java.lang.Object ref = visibleDeviceList_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        visibleDeviceList_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * A comma-separated list of GPU ids that determines the 'visible'
-     * to 'virtual' mapping of GPU devices.  For example, if TensorFlow
-     * can see 8 GPU devices in the process, and one wanted to map
-     * visible GPU devices 5 and 3 as "/device:GPU:0", and "/device:GPU:1",
-     * then one would specify this field as "5,3".  This field is similar in
-     * spirit to the CUDA_VISIBLE_DEVICES environment variable, except
-     * it applies to the visible GPU devices in the process.
-     * NOTE:
-     * 1. The GPU driver provides the process with the visible GPUs
-     *    in an order which is not guaranteed to have any correlation to
-     *    the *physical* GPU id in the machine.  This field is used for
-     *    remapping "visible" to "virtual", which means this operates only
-     *    after the process starts.  Users are required to use vendor
-     *    specific mechanisms (e.g., CUDA_VISIBLE_DEVICES) to control the
-     *    physical to visible device mapping prior to invoking TensorFlow.
-     * 2. In the code, the ids in this list are also called "platform GPU id"s,
-     *    and the 'virtual' ids of GPU devices (i.e. the ids in the device
-     *    name "/device:GPU:&lt;id&gt;") are also called "TF GPU id"s. Please
-     *    refer to third_party/tensorflow/core/common_runtime/gpu/gpu_id.h
-     *    for more information.
-     * </pre>
-     *
-     * <code>string visible_device_list = 5;</code>
-     */
-    public Builder setVisibleDeviceList(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      visibleDeviceList_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * A comma-separated list of GPU ids that determines the 'visible'
-     * to 'virtual' mapping of GPU devices.  For example, if TensorFlow
-     * can see 8 GPU devices in the process, and one wanted to map
-     * visible GPU devices 5 and 3 as "/device:GPU:0", and "/device:GPU:1",
-     * then one would specify this field as "5,3".  This field is similar in
-     * spirit to the CUDA_VISIBLE_DEVICES environment variable, except
-     * it applies to the visible GPU devices in the process.
-     * NOTE:
-     * 1. The GPU driver provides the process with the visible GPUs
-     *    in an order which is not guaranteed to have any correlation to
-     *    the *physical* GPU id in the machine.  This field is used for
-     *    remapping "visible" to "virtual", which means this operates only
-     *    after the process starts.  Users are required to use vendor
-     *    specific mechanisms (e.g., CUDA_VISIBLE_DEVICES) to control the
-     *    physical to visible device mapping prior to invoking TensorFlow.
-     * 2. In the code, the ids in this list are also called "platform GPU id"s,
-     *    and the 'virtual' ids of GPU devices (i.e. the ids in the device
-     *    name "/device:GPU:&lt;id&gt;") are also called "TF GPU id"s. Please
-     *    refer to third_party/tensorflow/core/common_runtime/gpu/gpu_id.h
-     *    for more information.
-     * </pre>
-     *
-     * <code>string visible_device_list = 5;</code>
-     */
-    public Builder clearVisibleDeviceList() {
-      
-      visibleDeviceList_ = getDefaultInstance().getVisibleDeviceList();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * A comma-separated list of GPU ids that determines the 'visible'
-     * to 'virtual' mapping of GPU devices.  For example, if TensorFlow
-     * can see 8 GPU devices in the process, and one wanted to map
-     * visible GPU devices 5 and 3 as "/device:GPU:0", and "/device:GPU:1",
-     * then one would specify this field as "5,3".  This field is similar in
-     * spirit to the CUDA_VISIBLE_DEVICES environment variable, except
-     * it applies to the visible GPU devices in the process.
-     * NOTE:
-     * 1. The GPU driver provides the process with the visible GPUs
-     *    in an order which is not guaranteed to have any correlation to
-     *    the *physical* GPU id in the machine.  This field is used for
-     *    remapping "visible" to "virtual", which means this operates only
-     *    after the process starts.  Users are required to use vendor
-     *    specific mechanisms (e.g., CUDA_VISIBLE_DEVICES) to control the
-     *    physical to visible device mapping prior to invoking TensorFlow.
-     * 2. In the code, the ids in this list are also called "platform GPU id"s,
-     *    and the 'virtual' ids of GPU devices (i.e. the ids in the device
-     *    name "/device:GPU:&lt;id&gt;") are also called "TF GPU id"s. Please
-     *    refer to third_party/tensorflow/core/common_runtime/gpu/gpu_id.h
-     *    for more information.
-     * </pre>
-     *
-     * <code>string visible_device_list = 5;</code>
-     */
-    public Builder setVisibleDeviceListBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      visibleDeviceList_ = value;
-      onChanged();
-      return this;
-    }
-
-    private int pollingActiveDelayUsecs_ ;
-    /**
-     * <pre>
-     * In the event polling loop sleep this many microseconds between
-     * PollEvents calls, when the queue is not empty.  If value is not
-     * set or set to 0, gets set to a non-zero default.
-     * </pre>
-     *
-     * <code>int32 polling_active_delay_usecs = 6;</code>
-     */
-    public int getPollingActiveDelayUsecs() {
-      return pollingActiveDelayUsecs_;
-    }
-    /**
-     * <pre>
-     * In the event polling loop sleep this many microseconds between
-     * PollEvents calls, when the queue is not empty.  If value is not
-     * set or set to 0, gets set to a non-zero default.
-     * </pre>
-     *
-     * <code>int32 polling_active_delay_usecs = 6;</code>
-     */
-    public Builder setPollingActiveDelayUsecs(int value) {
-      
-      pollingActiveDelayUsecs_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * In the event polling loop sleep this many microseconds between
-     * PollEvents calls, when the queue is not empty.  If value is not
-     * set or set to 0, gets set to a non-zero default.
-     * </pre>
-     *
-     * <code>int32 polling_active_delay_usecs = 6;</code>
-     */
-    public Builder clearPollingActiveDelayUsecs() {
-      
-      pollingActiveDelayUsecs_ = 0;
-      onChanged();
-      return this;
-    }
-
-    private int pollingInactiveDelayMsecs_ ;
-    /**
-     * <pre>
-     * This field is deprecated and ignored.
-     * </pre>
-     *
-     * <code>int32 polling_inactive_delay_msecs = 7;</code>
-     */
-    public int getPollingInactiveDelayMsecs() {
-      return pollingInactiveDelayMsecs_;
-    }
-    /**
-     * <pre>
-     * This field is deprecated and ignored.
-     * </pre>
-     *
-     * <code>int32 polling_inactive_delay_msecs = 7;</code>
-     */
-    public Builder setPollingInactiveDelayMsecs(int value) {
-      
-      pollingInactiveDelayMsecs_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * This field is deprecated and ignored.
-     * </pre>
-     *
-     * <code>int32 polling_inactive_delay_msecs = 7;</code>
-     */
-    public Builder clearPollingInactiveDelayMsecs() {
-      
-      pollingInactiveDelayMsecs_ = 0;
-      onChanged();
-      return this;
-    }
-
-    private boolean forceGpuCompatible_ ;
-    /**
-     * <pre>
-     * Force all tensors to be gpu_compatible. On a GPU-enabled TensorFlow,
-     * enabling this option forces all CPU tensors to be allocated with Cuda
-     * pinned memory. Normally, TensorFlow will infer which tensors should be
-     * allocated as the pinned memory. But in case where the inference is
-     * incomplete, this option can significantly speed up the cross-device memory
-     * copy performance as long as it fits the memory.
-     * Note that this option is not something that should be
-     * enabled by default for unknown or very large models, since all Cuda pinned
-     * memory is unpageable, having too much pinned memory might negatively impact
-     * the overall host system performance.
-     * </pre>
-     *
-     * <code>bool force_gpu_compatible = 8;</code>
-     */
-    public boolean getForceGpuCompatible() {
-      return forceGpuCompatible_;
-    }
-    /**
-     * <pre>
-     * Force all tensors to be gpu_compatible. On a GPU-enabled TensorFlow,
-     * enabling this option forces all CPU tensors to be allocated with Cuda
-     * pinned memory. Normally, TensorFlow will infer which tensors should be
-     * allocated as the pinned memory. But in case where the inference is
-     * incomplete, this option can significantly speed up the cross-device memory
-     * copy performance as long as it fits the memory.
-     * Note that this option is not something that should be
-     * enabled by default for unknown or very large models, since all Cuda pinned
-     * memory is unpageable, having too much pinned memory might negatively impact
-     * the overall host system performance.
-     * </pre>
-     *
-     * <code>bool force_gpu_compatible = 8;</code>
-     */
-    public Builder setForceGpuCompatible(boolean value) {
-      
-      forceGpuCompatible_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Force all tensors to be gpu_compatible. On a GPU-enabled TensorFlow,
-     * enabling this option forces all CPU tensors to be allocated with Cuda
-     * pinned memory. Normally, TensorFlow will infer which tensors should be
-     * allocated as the pinned memory. But in case where the inference is
-     * incomplete, this option can significantly speed up the cross-device memory
-     * copy performance as long as it fits the memory.
-     * Note that this option is not something that should be
-     * enabled by default for unknown or very large models, since all Cuda pinned
-     * memory is unpageable, having too much pinned memory might negatively impact
-     * the overall host system performance.
-     * </pre>
-     *
-     * <code>bool force_gpu_compatible = 8;</code>
-     */
-    public Builder clearForceGpuCompatible() {
-      
-      forceGpuCompatible_ = false;
-      onChanged();
-      return this;
-    }
-
-    private org.tensorflow.proto.framework.GPUOptions.Experimental experimental_;
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.GPUOptions.Experimental, org.tensorflow.proto.framework.GPUOptions.Experimental.Builder, org.tensorflow.proto.framework.GPUOptions.ExperimentalOrBuilder> experimentalBuilder_;
-    /**
-     * <pre>
-     * Everything inside experimental is subject to change and is not subject
-     * to API stability guarantees in
-     * https://www.tensorflow.org/guide/version_compat.
-     * </pre>
-     *
-     * <code>.tensorflow.GPUOptions.Experimental experimental = 9;</code>
-     */
-    public boolean hasExperimental() {
-      return experimentalBuilder_ != null || experimental_ != null;
-    }
-    /**
-     * <pre>
-     * Everything inside experimental is subject to change and is not subject
-     * to API stability guarantees in
-     * https://www.tensorflow.org/guide/version_compat.
-     * </pre>
-     *
-     * <code>.tensorflow.GPUOptions.Experimental experimental = 9;</code>
-     */
-    public org.tensorflow.proto.framework.GPUOptions.Experimental getExperimental() {
-      if (experimentalBuilder_ == null) {
-        return experimental_ == null ? org.tensorflow.proto.framework.GPUOptions.Experimental.getDefaultInstance() : experimental_;
-      } else {
-        return experimentalBuilder_.getMessage();
-      }
-    }
-    /**
-     * <pre>
-     * Everything inside experimental is subject to change and is not subject
-     * to API stability guarantees in
-     * https://www.tensorflow.org/guide/version_compat.
-     * </pre>
-     *
-     * <code>.tensorflow.GPUOptions.Experimental experimental = 9;</code>
-     */
-    public Builder setExperimental(org.tensorflow.proto.framework.GPUOptions.Experimental value) {
-      if (experimentalBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        experimental_ = value;
-        onChanged();
-      } else {
-        experimentalBuilder_.setMessage(value);
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Everything inside experimental is subject to change and is not subject
-     * to API stability guarantees in
-     * https://www.tensorflow.org/guide/version_compat.
-     * </pre>
-     *
-     * <code>.tensorflow.GPUOptions.Experimental experimental = 9;</code>
-     */
-    public Builder setExperimental(
-        org.tensorflow.proto.framework.GPUOptions.Experimental.Builder builderForValue) {
-      if (experimentalBuilder_ == null) {
-        experimental_ = builderForValue.build();
-        onChanged();
-      } else {
-        experimentalBuilder_.setMessage(builderForValue.build());
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Everything inside experimental is subject to change and is not subject
-     * to API stability guarantees in
-     * https://www.tensorflow.org/guide/version_compat.
-     * </pre>
-     *
-     * <code>.tensorflow.GPUOptions.Experimental experimental = 9;</code>
-     */
-    public Builder mergeExperimental(org.tensorflow.proto.framework.GPUOptions.Experimental value) {
-      if (experimentalBuilder_ == null) {
-        if (experimental_ != null) {
-          experimental_ =
-            org.tensorflow.proto.framework.GPUOptions.Experimental.newBuilder(experimental_).mergeFrom(value).buildPartial();
-        } else {
-          experimental_ = value;
-        }
-        onChanged();
-      } else {
-        experimentalBuilder_.mergeFrom(value);
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Everything inside experimental is subject to change and is not subject
-     * to API stability guarantees in
-     * https://www.tensorflow.org/guide/version_compat.
-     * </pre>
-     *
-     * <code>.tensorflow.GPUOptions.Experimental experimental = 9;</code>
-     */
-    public Builder clearExperimental() {
-      if (experimentalBuilder_ == null) {
-        experimental_ = null;
-        onChanged();
-      } else {
-        experimental_ = null;
-        experimentalBuilder_ = null;
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Everything inside experimental is subject to change and is not subject
-     * to API stability guarantees in
-     * https://www.tensorflow.org/guide/version_compat.
-     * </pre>
-     *
-     * <code>.tensorflow.GPUOptions.Experimental experimental = 9;</code>
-     */
-    public org.tensorflow.proto.framework.GPUOptions.Experimental.Builder getExperimentalBuilder() {
-      
-      onChanged();
-      return getExperimentalFieldBuilder().getBuilder();
-    }
-    /**
-     * <pre>
-     * Everything inside experimental is subject to change and is not subject
-     * to API stability guarantees in
-     * https://www.tensorflow.org/guide/version_compat.
-     * </pre>
-     *
-     * <code>.tensorflow.GPUOptions.Experimental experimental = 9;</code>
-     */
-    public org.tensorflow.proto.framework.GPUOptions.ExperimentalOrBuilder getExperimentalOrBuilder() {
-      if (experimentalBuilder_ != null) {
-        return experimentalBuilder_.getMessageOrBuilder();
-      } else {
-        return experimental_ == null ?
-            org.tensorflow.proto.framework.GPUOptions.Experimental.getDefaultInstance() : experimental_;
-      }
-    }
-    /**
-     * <pre>
-     * Everything inside experimental is subject to change and is not subject
-     * to API stability guarantees in
-     * https://www.tensorflow.org/guide/version_compat.
-     * </pre>
-     *
-     * <code>.tensorflow.GPUOptions.Experimental experimental = 9;</code>
-     */
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.GPUOptions.Experimental, org.tensorflow.proto.framework.GPUOptions.Experimental.Builder, org.tensorflow.proto.framework.GPUOptions.ExperimentalOrBuilder> 
-        getExperimentalFieldBuilder() {
-      if (experimentalBuilder_ == null) {
-        experimentalBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-            org.tensorflow.proto.framework.GPUOptions.Experimental, org.tensorflow.proto.framework.GPUOptions.Experimental.Builder, org.tensorflow.proto.framework.GPUOptions.ExperimentalOrBuilder>(
-                getExperimental(),
-                getParentForChildren(),
-                isClean());
-        experimental_ = null;
-      }
-      return experimentalBuilder_;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.GPUOptions)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.GPUOptions)
-  private static final org.tensorflow.proto.framework.GPUOptions DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.GPUOptions();
-  }
-
-  public static org.tensorflow.proto.framework.GPUOptions getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<GPUOptions>
-      PARSER = new com.google.protobuf.AbstractParser<GPUOptions>() {
-    @java.lang.Override
-    public GPUOptions parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new GPUOptions(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<GPUOptions> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<GPUOptions> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.framework.GPUOptions getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GPUOptionsOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GPUOptionsOrBuilder.java
deleted file mode 100644
index 6f11472d49a..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GPUOptionsOrBuilder.java
+++ /dev/null
@@ -1,206 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/config.proto
-
-package org.tensorflow.proto.framework;
-
-public interface GPUOptionsOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.GPUOptions)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <pre>
-   * Fraction of the available GPU memory to allocate for each process.
-   * 1 means to allocate all of the GPU memory, 0.5 means the process
-   * allocates up to ~50% of the available GPU memory.
-   * GPU memory is pre-allocated unless the allow_growth option is enabled.
-   * If greater than 1.0, uses CUDA unified memory to potentially oversubscribe
-   * the amount of memory available on the GPU device by using host memory as a
-   * swap space. Accessing memory not available on the device will be
-   * significantly slower as that would require memory transfer between the host
-   * and the device. Options to reduce the memory requirement should be
-   * considered before enabling this option as this may come with a negative
-   * performance impact. Oversubscription using the unified memory requires
-   * Pascal class or newer GPUs and it is currently only supported on the Linux
-   * operating system. See
-   * https://docs.nvidia.com/cuda/cuda-c-programming-guide/index.html#um-requirements
-   * for the detailed requirements.
-   * </pre>
-   *
-   * <code>double per_process_gpu_memory_fraction = 1;</code>
-   */
-  double getPerProcessGpuMemoryFraction();
-
-  /**
-   * <pre>
-   * If true, the allocator does not pre-allocate the entire specified
-   * GPU memory region, instead starting small and growing as needed.
-   * </pre>
-   *
-   * <code>bool allow_growth = 4;</code>
-   */
-  boolean getAllowGrowth();
-
-  /**
-   * <pre>
-   * The type of GPU allocation strategy to use.
-   * Allowed values:
-   * "": The empty string (default) uses a system-chosen default
-   *     which may change over time.
-   * "BFC": A "Best-fit with coalescing" algorithm, simplified from a
-   *        version of dlmalloc.
-   * </pre>
-   *
-   * <code>string allocator_type = 2;</code>
-   */
-  java.lang.String getAllocatorType();
-  /**
-   * <pre>
-   * The type of GPU allocation strategy to use.
-   * Allowed values:
-   * "": The empty string (default) uses a system-chosen default
-   *     which may change over time.
-   * "BFC": A "Best-fit with coalescing" algorithm, simplified from a
-   *        version of dlmalloc.
-   * </pre>
-   *
-   * <code>string allocator_type = 2;</code>
-   */
-  com.google.protobuf.ByteString
-      getAllocatorTypeBytes();
-
-  /**
-   * <pre>
-   * Delay deletion of up to this many bytes to reduce the number of
-   * interactions with gpu driver code.  If 0, the system chooses
-   * a reasonable default (several MBs).
-   * </pre>
-   *
-   * <code>int64 deferred_deletion_bytes = 3;</code>
-   */
-  long getDeferredDeletionBytes();
-
-  /**
-   * <pre>
-   * A comma-separated list of GPU ids that determines the 'visible'
-   * to 'virtual' mapping of GPU devices.  For example, if TensorFlow
-   * can see 8 GPU devices in the process, and one wanted to map
-   * visible GPU devices 5 and 3 as "/device:GPU:0", and "/device:GPU:1",
-   * then one would specify this field as "5,3".  This field is similar in
-   * spirit to the CUDA_VISIBLE_DEVICES environment variable, except
-   * it applies to the visible GPU devices in the process.
-   * NOTE:
-   * 1. The GPU driver provides the process with the visible GPUs
-   *    in an order which is not guaranteed to have any correlation to
-   *    the *physical* GPU id in the machine.  This field is used for
-   *    remapping "visible" to "virtual", which means this operates only
-   *    after the process starts.  Users are required to use vendor
-   *    specific mechanisms (e.g., CUDA_VISIBLE_DEVICES) to control the
-   *    physical to visible device mapping prior to invoking TensorFlow.
-   * 2. In the code, the ids in this list are also called "platform GPU id"s,
-   *    and the 'virtual' ids of GPU devices (i.e. the ids in the device
-   *    name "/device:GPU:&lt;id&gt;") are also called "TF GPU id"s. Please
-   *    refer to third_party/tensorflow/core/common_runtime/gpu/gpu_id.h
-   *    for more information.
-   * </pre>
-   *
-   * <code>string visible_device_list = 5;</code>
-   */
-  java.lang.String getVisibleDeviceList();
-  /**
-   * <pre>
-   * A comma-separated list of GPU ids that determines the 'visible'
-   * to 'virtual' mapping of GPU devices.  For example, if TensorFlow
-   * can see 8 GPU devices in the process, and one wanted to map
-   * visible GPU devices 5 and 3 as "/device:GPU:0", and "/device:GPU:1",
-   * then one would specify this field as "5,3".  This field is similar in
-   * spirit to the CUDA_VISIBLE_DEVICES environment variable, except
-   * it applies to the visible GPU devices in the process.
-   * NOTE:
-   * 1. The GPU driver provides the process with the visible GPUs
-   *    in an order which is not guaranteed to have any correlation to
-   *    the *physical* GPU id in the machine.  This field is used for
-   *    remapping "visible" to "virtual", which means this operates only
-   *    after the process starts.  Users are required to use vendor
-   *    specific mechanisms (e.g., CUDA_VISIBLE_DEVICES) to control the
-   *    physical to visible device mapping prior to invoking TensorFlow.
-   * 2. In the code, the ids in this list are also called "platform GPU id"s,
-   *    and the 'virtual' ids of GPU devices (i.e. the ids in the device
-   *    name "/device:GPU:&lt;id&gt;") are also called "TF GPU id"s. Please
-   *    refer to third_party/tensorflow/core/common_runtime/gpu/gpu_id.h
-   *    for more information.
-   * </pre>
-   *
-   * <code>string visible_device_list = 5;</code>
-   */
-  com.google.protobuf.ByteString
-      getVisibleDeviceListBytes();
-
-  /**
-   * <pre>
-   * In the event polling loop sleep this many microseconds between
-   * PollEvents calls, when the queue is not empty.  If value is not
-   * set or set to 0, gets set to a non-zero default.
-   * </pre>
-   *
-   * <code>int32 polling_active_delay_usecs = 6;</code>
-   */
-  int getPollingActiveDelayUsecs();
-
-  /**
-   * <pre>
-   * This field is deprecated and ignored.
-   * </pre>
-   *
-   * <code>int32 polling_inactive_delay_msecs = 7;</code>
-   */
-  int getPollingInactiveDelayMsecs();
-
-  /**
-   * <pre>
-   * Force all tensors to be gpu_compatible. On a GPU-enabled TensorFlow,
-   * enabling this option forces all CPU tensors to be allocated with Cuda
-   * pinned memory. Normally, TensorFlow will infer which tensors should be
-   * allocated as the pinned memory. But in case where the inference is
-   * incomplete, this option can significantly speed up the cross-device memory
-   * copy performance as long as it fits the memory.
-   * Note that this option is not something that should be
-   * enabled by default for unknown or very large models, since all Cuda pinned
-   * memory is unpageable, having too much pinned memory might negatively impact
-   * the overall host system performance.
-   * </pre>
-   *
-   * <code>bool force_gpu_compatible = 8;</code>
-   */
-  boolean getForceGpuCompatible();
-
-  /**
-   * <pre>
-   * Everything inside experimental is subject to change and is not subject
-   * to API stability guarantees in
-   * https://www.tensorflow.org/guide/version_compat.
-   * </pre>
-   *
-   * <code>.tensorflow.GPUOptions.Experimental experimental = 9;</code>
-   */
-  boolean hasExperimental();
-  /**
-   * <pre>
-   * Everything inside experimental is subject to change and is not subject
-   * to API stability guarantees in
-   * https://www.tensorflow.org/guide/version_compat.
-   * </pre>
-   *
-   * <code>.tensorflow.GPUOptions.Experimental experimental = 9;</code>
-   */
-  org.tensorflow.proto.framework.GPUOptions.Experimental getExperimental();
-  /**
-   * <pre>
-   * Everything inside experimental is subject to change and is not subject
-   * to API stability guarantees in
-   * https://www.tensorflow.org/guide/version_compat.
-   * </pre>
-   *
-   * <code>.tensorflow.GPUOptions.Experimental experimental = 9;</code>
-   */
-  org.tensorflow.proto.framework.GPUOptions.ExperimentalOrBuilder getExperimentalOrBuilder();
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GradientDef.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GradientDef.java
deleted file mode 100644
index fe1126cb2ef..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GradientDef.java
+++ /dev/null
@@ -1,765 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/framework/function.proto
-
-package org.tensorflow.proto.framework;
-
-/**
- * <pre>
- * GradientDef defines the gradient function of a function defined in
- * a function library.
- * A gradient function g (specified by gradient_func) for a function f
- * (specified by function_name) must follow the following:
- * The function 'f' must be a numerical function which takes N inputs
- * and produces M outputs. Its gradient function 'g', which is a
- * function taking N + M inputs and produces N outputs.
- * I.e. if we have
- *    (y1, y2, ..., y_M) = f(x1, x2, ..., x_N),
- * then, g is
- *    (dL/dx1, dL/dx2, ..., dL/dx_N) = g(x1, x2, ..., x_N,
- *                                      dL/dy1, dL/dy2, ..., dL/dy_M),
- * where L is a scalar-value function of (x1, x2, ..., xN) (e.g., the
- * loss function). dL/dx_i is the partial derivative of L with respect
- * to x_i.
- * </pre>
- *
- * Protobuf type {@code tensorflow.GradientDef}
- */
-public  final class GradientDef extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.GradientDef)
-    GradientDefOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use GradientDef.newBuilder() to construct.
-  private GradientDef(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private GradientDef() {
-    functionName_ = "";
-    gradientFunc_ = "";
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new GradientDef();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private GradientDef(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 10: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            functionName_ = s;
-            break;
-          }
-          case 18: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            gradientFunc_ = s;
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.framework.FunctionProtos.internal_static_tensorflow_GradientDef_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.framework.FunctionProtos.internal_static_tensorflow_GradientDef_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.framework.GradientDef.class, org.tensorflow.proto.framework.GradientDef.Builder.class);
-  }
-
-  public static final int FUNCTION_NAME_FIELD_NUMBER = 1;
-  private volatile java.lang.Object functionName_;
-  /**
-   * <pre>
-   * The function name.
-   * </pre>
-   *
-   * <code>string function_name = 1;</code>
-   */
-  public java.lang.String getFunctionName() {
-    java.lang.Object ref = functionName_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      functionName_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * The function name.
-   * </pre>
-   *
-   * <code>string function_name = 1;</code>
-   */
-  public com.google.protobuf.ByteString
-      getFunctionNameBytes() {
-    java.lang.Object ref = functionName_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      functionName_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int GRADIENT_FUNC_FIELD_NUMBER = 2;
-  private volatile java.lang.Object gradientFunc_;
-  /**
-   * <pre>
-   * The gradient function's name.
-   * </pre>
-   *
-   * <code>string gradient_func = 2;</code>
-   */
-  public java.lang.String getGradientFunc() {
-    java.lang.Object ref = gradientFunc_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      gradientFunc_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * The gradient function's name.
-   * </pre>
-   *
-   * <code>string gradient_func = 2;</code>
-   */
-  public com.google.protobuf.ByteString
-      getGradientFuncBytes() {
-    java.lang.Object ref = gradientFunc_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      gradientFunc_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    if (!getFunctionNameBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 1, functionName_);
-    }
-    if (!getGradientFuncBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 2, gradientFunc_);
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (!getFunctionNameBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, functionName_);
-    }
-    if (!getGradientFuncBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(2, gradientFunc_);
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.framework.GradientDef)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.framework.GradientDef other = (org.tensorflow.proto.framework.GradientDef) obj;
-
-    if (!getFunctionName()
-        .equals(other.getFunctionName())) return false;
-    if (!getGradientFunc()
-        .equals(other.getGradientFunc())) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    hash = (37 * hash) + FUNCTION_NAME_FIELD_NUMBER;
-    hash = (53 * hash) + getFunctionName().hashCode();
-    hash = (37 * hash) + GRADIENT_FUNC_FIELD_NUMBER;
-    hash = (53 * hash) + getGradientFunc().hashCode();
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.framework.GradientDef parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.GradientDef parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.GradientDef parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.GradientDef parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.GradientDef parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.GradientDef parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.GradientDef parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.GradientDef parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.GradientDef parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.GradientDef parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.GradientDef parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.GradientDef parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.framework.GradientDef prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * <pre>
-   * GradientDef defines the gradient function of a function defined in
-   * a function library.
-   * A gradient function g (specified by gradient_func) for a function f
-   * (specified by function_name) must follow the following:
-   * The function 'f' must be a numerical function which takes N inputs
-   * and produces M outputs. Its gradient function 'g', which is a
-   * function taking N + M inputs and produces N outputs.
-   * I.e. if we have
-   *    (y1, y2, ..., y_M) = f(x1, x2, ..., x_N),
-   * then, g is
-   *    (dL/dx1, dL/dx2, ..., dL/dx_N) = g(x1, x2, ..., x_N,
-   *                                      dL/dy1, dL/dy2, ..., dL/dy_M),
-   * where L is a scalar-value function of (x1, x2, ..., xN) (e.g., the
-   * loss function). dL/dx_i is the partial derivative of L with respect
-   * to x_i.
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.GradientDef}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.GradientDef)
-      org.tensorflow.proto.framework.GradientDefOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.FunctionProtos.internal_static_tensorflow_GradientDef_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.FunctionProtos.internal_static_tensorflow_GradientDef_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.GradientDef.class, org.tensorflow.proto.framework.GradientDef.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.framework.GradientDef.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      functionName_ = "";
-
-      gradientFunc_ = "";
-
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.framework.FunctionProtos.internal_static_tensorflow_GradientDef_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.GradientDef getDefaultInstanceForType() {
-      return org.tensorflow.proto.framework.GradientDef.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.GradientDef build() {
-      org.tensorflow.proto.framework.GradientDef result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.GradientDef buildPartial() {
-      org.tensorflow.proto.framework.GradientDef result = new org.tensorflow.proto.framework.GradientDef(this);
-      result.functionName_ = functionName_;
-      result.gradientFunc_ = gradientFunc_;
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.framework.GradientDef) {
-        return mergeFrom((org.tensorflow.proto.framework.GradientDef)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.framework.GradientDef other) {
-      if (other == org.tensorflow.proto.framework.GradientDef.getDefaultInstance()) return this;
-      if (!other.getFunctionName().isEmpty()) {
-        functionName_ = other.functionName_;
-        onChanged();
-      }
-      if (!other.getGradientFunc().isEmpty()) {
-        gradientFunc_ = other.gradientFunc_;
-        onChanged();
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.framework.GradientDef parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.framework.GradientDef) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-
-    private java.lang.Object functionName_ = "";
-    /**
-     * <pre>
-     * The function name.
-     * </pre>
-     *
-     * <code>string function_name = 1;</code>
-     */
-    public java.lang.String getFunctionName() {
-      java.lang.Object ref = functionName_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        functionName_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * The function name.
-     * </pre>
-     *
-     * <code>string function_name = 1;</code>
-     */
-    public com.google.protobuf.ByteString
-        getFunctionNameBytes() {
-      java.lang.Object ref = functionName_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        functionName_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * The function name.
-     * </pre>
-     *
-     * <code>string function_name = 1;</code>
-     */
-    public Builder setFunctionName(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      functionName_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * The function name.
-     * </pre>
-     *
-     * <code>string function_name = 1;</code>
-     */
-    public Builder clearFunctionName() {
-      
-      functionName_ = getDefaultInstance().getFunctionName();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * The function name.
-     * </pre>
-     *
-     * <code>string function_name = 1;</code>
-     */
-    public Builder setFunctionNameBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      functionName_ = value;
-      onChanged();
-      return this;
-    }
-
-    private java.lang.Object gradientFunc_ = "";
-    /**
-     * <pre>
-     * The gradient function's name.
-     * </pre>
-     *
-     * <code>string gradient_func = 2;</code>
-     */
-    public java.lang.String getGradientFunc() {
-      java.lang.Object ref = gradientFunc_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        gradientFunc_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * The gradient function's name.
-     * </pre>
-     *
-     * <code>string gradient_func = 2;</code>
-     */
-    public com.google.protobuf.ByteString
-        getGradientFuncBytes() {
-      java.lang.Object ref = gradientFunc_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        gradientFunc_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * The gradient function's name.
-     * </pre>
-     *
-     * <code>string gradient_func = 2;</code>
-     */
-    public Builder setGradientFunc(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      gradientFunc_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * The gradient function's name.
-     * </pre>
-     *
-     * <code>string gradient_func = 2;</code>
-     */
-    public Builder clearGradientFunc() {
-      
-      gradientFunc_ = getDefaultInstance().getGradientFunc();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * The gradient function's name.
-     * </pre>
-     *
-     * <code>string gradient_func = 2;</code>
-     */
-    public Builder setGradientFuncBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      gradientFunc_ = value;
-      onChanged();
-      return this;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.GradientDef)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.GradientDef)
-  private static final org.tensorflow.proto.framework.GradientDef DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.GradientDef();
-  }
-
-  public static org.tensorflow.proto.framework.GradientDef getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<GradientDef>
-      PARSER = new com.google.protobuf.AbstractParser<GradientDef>() {
-    @java.lang.Override
-    public GradientDef parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new GradientDef(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<GradientDef> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<GradientDef> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.framework.GradientDef getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GradientDefOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GradientDefOrBuilder.java
deleted file mode 100644
index 9437941de35..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GradientDefOrBuilder.java
+++ /dev/null
@@ -1,45 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/framework/function.proto
-
-package org.tensorflow.proto.framework;
-
-public interface GradientDefOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.GradientDef)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <pre>
-   * The function name.
-   * </pre>
-   *
-   * <code>string function_name = 1;</code>
-   */
-  java.lang.String getFunctionName();
-  /**
-   * <pre>
-   * The function name.
-   * </pre>
-   *
-   * <code>string function_name = 1;</code>
-   */
-  com.google.protobuf.ByteString
-      getFunctionNameBytes();
-
-  /**
-   * <pre>
-   * The gradient function's name.
-   * </pre>
-   *
-   * <code>string gradient_func = 2;</code>
-   */
-  java.lang.String getGradientFunc();
-  /**
-   * <pre>
-   * The gradient function's name.
-   * </pre>
-   *
-   * <code>string gradient_func = 2;</code>
-   */
-  com.google.protobuf.ByteString
-      getGradientFuncBytes();
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphDebugInfo.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphDebugInfo.java
deleted file mode 100644
index 165c76f3ce7..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphDebugInfo.java
+++ /dev/null
@@ -1,3004 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/graph_debug_info.proto
-
-package org.tensorflow.proto.framework;
-
-/**
- * Protobuf type {@code tensorflow.GraphDebugInfo}
- */
-public  final class GraphDebugInfo extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.GraphDebugInfo)
-    GraphDebugInfoOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use GraphDebugInfo.newBuilder() to construct.
-  private GraphDebugInfo(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private GraphDebugInfo() {
-    files_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new GraphDebugInfo();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private GraphDebugInfo(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    int mutable_bitField0_ = 0;
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 10: {
-            java.lang.String s = input.readStringRequireUtf8();
-            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
-              files_ = new com.google.protobuf.LazyStringArrayList();
-              mutable_bitField0_ |= 0x00000001;
-            }
-            files_.add(s);
-            break;
-          }
-          case 18: {
-            if (!((mutable_bitField0_ & 0x00000002) != 0)) {
-              traces_ = com.google.protobuf.MapField.newMapField(
-                  TracesDefaultEntryHolder.defaultEntry);
-              mutable_bitField0_ |= 0x00000002;
-            }
-            com.google.protobuf.MapEntry<java.lang.String, org.tensorflow.proto.framework.GraphDebugInfo.StackTrace>
-            traces__ = input.readMessage(
-                TracesDefaultEntryHolder.defaultEntry.getParserForType(), extensionRegistry);
-            traces_.getMutableMap().put(
-                traces__.getKey(), traces__.getValue());
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      if (((mutable_bitField0_ & 0x00000001) != 0)) {
-        files_ = files_.getUnmodifiableView();
-      }
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.framework.GraphDebugInfoProtos.internal_static_tensorflow_GraphDebugInfo_descriptor;
-  }
-
-  @SuppressWarnings({"rawtypes"})
-  @java.lang.Override
-  protected com.google.protobuf.MapField internalGetMapField(
-      int number) {
-    switch (number) {
-      case 2:
-        return internalGetTraces();
-      default:
-        throw new RuntimeException(
-            "Invalid map field number: " + number);
-    }
-  }
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.framework.GraphDebugInfoProtos.internal_static_tensorflow_GraphDebugInfo_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.framework.GraphDebugInfo.class, org.tensorflow.proto.framework.GraphDebugInfo.Builder.class);
-  }
-
-  public interface FileLineColOrBuilder extends
-      // @@protoc_insertion_point(interface_extends:tensorflow.GraphDebugInfo.FileLineCol)
-      com.google.protobuf.MessageOrBuilder {
-
-    /**
-     * <pre>
-     * File name index, which can be used to retrieve the file name string from
-     * `files`. The value should be between 0 and (len(files)-1)
-     * </pre>
-     *
-     * <code>int32 file_index = 1;</code>
-     */
-    int getFileIndex();
-
-    /**
-     * <pre>
-     * Line number in the file.
-     * </pre>
-     *
-     * <code>int32 line = 2;</code>
-     */
-    int getLine();
-
-    /**
-     * <pre>
-     * Col number in the file line.
-     * </pre>
-     *
-     * <code>int32 col = 3;</code>
-     */
-    int getCol();
-
-    /**
-     * <pre>
-     * Name of function contains the file line.
-     * </pre>
-     *
-     * <code>string func = 4;</code>
-     */
-    java.lang.String getFunc();
-    /**
-     * <pre>
-     * Name of function contains the file line.
-     * </pre>
-     *
-     * <code>string func = 4;</code>
-     */
-    com.google.protobuf.ByteString
-        getFuncBytes();
-
-    /**
-     * <pre>
-     * Source code contained in this file line.
-     * </pre>
-     *
-     * <code>string code = 5;</code>
-     */
-    java.lang.String getCode();
-    /**
-     * <pre>
-     * Source code contained in this file line.
-     * </pre>
-     *
-     * <code>string code = 5;</code>
-     */
-    com.google.protobuf.ByteString
-        getCodeBytes();
-  }
-  /**
-   * <pre>
-   * This represents a file/line location in the source code.
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.GraphDebugInfo.FileLineCol}
-   */
-  public  static final class FileLineCol extends
-      com.google.protobuf.GeneratedMessageV3 implements
-      // @@protoc_insertion_point(message_implements:tensorflow.GraphDebugInfo.FileLineCol)
-      FileLineColOrBuilder {
-  private static final long serialVersionUID = 0L;
-    // Use FileLineCol.newBuilder() to construct.
-    private FileLineCol(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-      super(builder);
-    }
-    private FileLineCol() {
-      func_ = "";
-      code_ = "";
-    }
-
-    @java.lang.Override
-    @SuppressWarnings({"unused"})
-    protected java.lang.Object newInstance(
-        UnusedPrivateParameter unused) {
-      return new FileLineCol();
-    }
-
-    @java.lang.Override
-    public final com.google.protobuf.UnknownFieldSet
-    getUnknownFields() {
-      return this.unknownFields;
-    }
-    private FileLineCol(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      this();
-      if (extensionRegistry == null) {
-        throw new java.lang.NullPointerException();
-      }
-      com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-          com.google.protobuf.UnknownFieldSet.newBuilder();
-      try {
-        boolean done = false;
-        while (!done) {
-          int tag = input.readTag();
-          switch (tag) {
-            case 0:
-              done = true;
-              break;
-            case 8: {
-
-              fileIndex_ = input.readInt32();
-              break;
-            }
-            case 16: {
-
-              line_ = input.readInt32();
-              break;
-            }
-            case 24: {
-
-              col_ = input.readInt32();
-              break;
-            }
-            case 34: {
-              java.lang.String s = input.readStringRequireUtf8();
-
-              func_ = s;
-              break;
-            }
-            case 42: {
-              java.lang.String s = input.readStringRequireUtf8();
-
-              code_ = s;
-              break;
-            }
-            default: {
-              if (!parseUnknownField(
-                  input, unknownFields, extensionRegistry, tag)) {
-                done = true;
-              }
-              break;
-            }
-          }
-        }
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        throw e.setUnfinishedMessage(this);
-      } catch (java.io.IOException e) {
-        throw new com.google.protobuf.InvalidProtocolBufferException(
-            e).setUnfinishedMessage(this);
-      } finally {
-        this.unknownFields = unknownFields.build();
-        makeExtensionsImmutable();
-      }
-    }
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.GraphDebugInfoProtos.internal_static_tensorflow_GraphDebugInfo_FileLineCol_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.GraphDebugInfoProtos.internal_static_tensorflow_GraphDebugInfo_FileLineCol_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol.class, org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol.Builder.class);
-    }
-
-    public static final int FILE_INDEX_FIELD_NUMBER = 1;
-    private int fileIndex_;
-    /**
-     * <pre>
-     * File name index, which can be used to retrieve the file name string from
-     * `files`. The value should be between 0 and (len(files)-1)
-     * </pre>
-     *
-     * <code>int32 file_index = 1;</code>
-     */
-    public int getFileIndex() {
-      return fileIndex_;
-    }
-
-    public static final int LINE_FIELD_NUMBER = 2;
-    private int line_;
-    /**
-     * <pre>
-     * Line number in the file.
-     * </pre>
-     *
-     * <code>int32 line = 2;</code>
-     */
-    public int getLine() {
-      return line_;
-    }
-
-    public static final int COL_FIELD_NUMBER = 3;
-    private int col_;
-    /**
-     * <pre>
-     * Col number in the file line.
-     * </pre>
-     *
-     * <code>int32 col = 3;</code>
-     */
-    public int getCol() {
-      return col_;
-    }
-
-    public static final int FUNC_FIELD_NUMBER = 4;
-    private volatile java.lang.Object func_;
-    /**
-     * <pre>
-     * Name of function contains the file line.
-     * </pre>
-     *
-     * <code>string func = 4;</code>
-     */
-    public java.lang.String getFunc() {
-      java.lang.Object ref = func_;
-      if (ref instanceof java.lang.String) {
-        return (java.lang.String) ref;
-      } else {
-        com.google.protobuf.ByteString bs = 
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        func_ = s;
-        return s;
-      }
-    }
-    /**
-     * <pre>
-     * Name of function contains the file line.
-     * </pre>
-     *
-     * <code>string func = 4;</code>
-     */
-    public com.google.protobuf.ByteString
-        getFuncBytes() {
-      java.lang.Object ref = func_;
-      if (ref instanceof java.lang.String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        func_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-
-    public static final int CODE_FIELD_NUMBER = 5;
-    private volatile java.lang.Object code_;
-    /**
-     * <pre>
-     * Source code contained in this file line.
-     * </pre>
-     *
-     * <code>string code = 5;</code>
-     */
-    public java.lang.String getCode() {
-      java.lang.Object ref = code_;
-      if (ref instanceof java.lang.String) {
-        return (java.lang.String) ref;
-      } else {
-        com.google.protobuf.ByteString bs = 
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        code_ = s;
-        return s;
-      }
-    }
-    /**
-     * <pre>
-     * Source code contained in this file line.
-     * </pre>
-     *
-     * <code>string code = 5;</code>
-     */
-    public com.google.protobuf.ByteString
-        getCodeBytes() {
-      java.lang.Object ref = code_;
-      if (ref instanceof java.lang.String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        code_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-
-    private byte memoizedIsInitialized = -1;
-    @java.lang.Override
-    public final boolean isInitialized() {
-      byte isInitialized = memoizedIsInitialized;
-      if (isInitialized == 1) return true;
-      if (isInitialized == 0) return false;
-
-      memoizedIsInitialized = 1;
-      return true;
-    }
-
-    @java.lang.Override
-    public void writeTo(com.google.protobuf.CodedOutputStream output)
-                        throws java.io.IOException {
-      if (fileIndex_ != 0) {
-        output.writeInt32(1, fileIndex_);
-      }
-      if (line_ != 0) {
-        output.writeInt32(2, line_);
-      }
-      if (col_ != 0) {
-        output.writeInt32(3, col_);
-      }
-      if (!getFuncBytes().isEmpty()) {
-        com.google.protobuf.GeneratedMessageV3.writeString(output, 4, func_);
-      }
-      if (!getCodeBytes().isEmpty()) {
-        com.google.protobuf.GeneratedMessageV3.writeString(output, 5, code_);
-      }
-      unknownFields.writeTo(output);
-    }
-
-    @java.lang.Override
-    public int getSerializedSize() {
-      int size = memoizedSize;
-      if (size != -1) return size;
-
-      size = 0;
-      if (fileIndex_ != 0) {
-        size += com.google.protobuf.CodedOutputStream
-          .computeInt32Size(1, fileIndex_);
-      }
-      if (line_ != 0) {
-        size += com.google.protobuf.CodedOutputStream
-          .computeInt32Size(2, line_);
-      }
-      if (col_ != 0) {
-        size += com.google.protobuf.CodedOutputStream
-          .computeInt32Size(3, col_);
-      }
-      if (!getFuncBytes().isEmpty()) {
-        size += com.google.protobuf.GeneratedMessageV3.computeStringSize(4, func_);
-      }
-      if (!getCodeBytes().isEmpty()) {
-        size += com.google.protobuf.GeneratedMessageV3.computeStringSize(5, code_);
-      }
-      size += unknownFields.getSerializedSize();
-      memoizedSize = size;
-      return size;
-    }
-
-    @java.lang.Override
-    public boolean equals(final java.lang.Object obj) {
-      if (obj == this) {
-       return true;
-      }
-      if (!(obj instanceof org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol)) {
-        return super.equals(obj);
-      }
-      org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol other = (org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol) obj;
-
-      if (getFileIndex()
-          != other.getFileIndex()) return false;
-      if (getLine()
-          != other.getLine()) return false;
-      if (getCol()
-          != other.getCol()) return false;
-      if (!getFunc()
-          .equals(other.getFunc())) return false;
-      if (!getCode()
-          .equals(other.getCode())) return false;
-      if (!unknownFields.equals(other.unknownFields)) return false;
-      return true;
-    }
-
-    @java.lang.Override
-    public int hashCode() {
-      if (memoizedHashCode != 0) {
-        return memoizedHashCode;
-      }
-      int hash = 41;
-      hash = (19 * hash) + getDescriptor().hashCode();
-      hash = (37 * hash) + FILE_INDEX_FIELD_NUMBER;
-      hash = (53 * hash) + getFileIndex();
-      hash = (37 * hash) + LINE_FIELD_NUMBER;
-      hash = (53 * hash) + getLine();
-      hash = (37 * hash) + COL_FIELD_NUMBER;
-      hash = (53 * hash) + getCol();
-      hash = (37 * hash) + FUNC_FIELD_NUMBER;
-      hash = (53 * hash) + getFunc().hashCode();
-      hash = (37 * hash) + CODE_FIELD_NUMBER;
-      hash = (53 * hash) + getCode().hashCode();
-      hash = (29 * hash) + unknownFields.hashCode();
-      memoizedHashCode = hash;
-      return hash;
-    }
-
-    public static org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol parseFrom(
-        java.nio.ByteBuffer data)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data);
-    }
-    public static org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol parseFrom(
-        java.nio.ByteBuffer data,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol parseFrom(
-        com.google.protobuf.ByteString data)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data);
-    }
-    public static org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol parseFrom(
-        com.google.protobuf.ByteString data,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol parseFrom(byte[] data)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data);
-    }
-    public static org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol parseFrom(
-        byte[] data,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol parseFrom(java.io.InputStream input)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input);
-    }
-    public static org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol parseFrom(
-        java.io.InputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol parseDelimitedFrom(java.io.InputStream input)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseDelimitedWithIOException(PARSER, input);
-    }
-    public static org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol parseDelimitedFrom(
-        java.io.InputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol parseFrom(
-        com.google.protobuf.CodedInputStream input)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input);
-    }
-    public static org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol parseFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input, extensionRegistry);
-    }
-
-    @java.lang.Override
-    public Builder newBuilderForType() { return newBuilder(); }
-    public static Builder newBuilder() {
-      return DEFAULT_INSTANCE.toBuilder();
-    }
-    public static Builder newBuilder(org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol prototype) {
-      return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-    }
-    @java.lang.Override
-    public Builder toBuilder() {
-      return this == DEFAULT_INSTANCE
-          ? new Builder() : new Builder().mergeFrom(this);
-    }
-
-    @java.lang.Override
-    protected Builder newBuilderForType(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      Builder builder = new Builder(parent);
-      return builder;
-    }
-    /**
-     * <pre>
-     * This represents a file/line location in the source code.
-     * </pre>
-     *
-     * Protobuf type {@code tensorflow.GraphDebugInfo.FileLineCol}
-     */
-    public static final class Builder extends
-        com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-        // @@protoc_insertion_point(builder_implements:tensorflow.GraphDebugInfo.FileLineCol)
-        org.tensorflow.proto.framework.GraphDebugInfo.FileLineColOrBuilder {
-      public static final com.google.protobuf.Descriptors.Descriptor
-          getDescriptor() {
-        return org.tensorflow.proto.framework.GraphDebugInfoProtos.internal_static_tensorflow_GraphDebugInfo_FileLineCol_descriptor;
-      }
-
-      @java.lang.Override
-      protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-          internalGetFieldAccessorTable() {
-        return org.tensorflow.proto.framework.GraphDebugInfoProtos.internal_static_tensorflow_GraphDebugInfo_FileLineCol_fieldAccessorTable
-            .ensureFieldAccessorsInitialized(
-                org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol.class, org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol.Builder.class);
-      }
-
-      // Construct using org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol.newBuilder()
-      private Builder() {
-        maybeForceBuilderInitialization();
-      }
-
-      private Builder(
-          com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-        super(parent);
-        maybeForceBuilderInitialization();
-      }
-      private void maybeForceBuilderInitialization() {
-        if (com.google.protobuf.GeneratedMessageV3
-                .alwaysUseFieldBuilders) {
-        }
-      }
-      @java.lang.Override
-      public Builder clear() {
-        super.clear();
-        fileIndex_ = 0;
-
-        line_ = 0;
-
-        col_ = 0;
-
-        func_ = "";
-
-        code_ = "";
-
-        return this;
-      }
-
-      @java.lang.Override
-      public com.google.protobuf.Descriptors.Descriptor
-          getDescriptorForType() {
-        return org.tensorflow.proto.framework.GraphDebugInfoProtos.internal_static_tensorflow_GraphDebugInfo_FileLineCol_descriptor;
-      }
-
-      @java.lang.Override
-      public org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol getDefaultInstanceForType() {
-        return org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol.getDefaultInstance();
-      }
-
-      @java.lang.Override
-      public org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol build() {
-        org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol result = buildPartial();
-        if (!result.isInitialized()) {
-          throw newUninitializedMessageException(result);
-        }
-        return result;
-      }
-
-      @java.lang.Override
-      public org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol buildPartial() {
-        org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol result = new org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol(this);
-        result.fileIndex_ = fileIndex_;
-        result.line_ = line_;
-        result.col_ = col_;
-        result.func_ = func_;
-        result.code_ = code_;
-        onBuilt();
-        return result;
-      }
-
-      @java.lang.Override
-      public Builder clone() {
-        return super.clone();
-      }
-      @java.lang.Override
-      public Builder setField(
-          com.google.protobuf.Descriptors.FieldDescriptor field,
-          java.lang.Object value) {
-        return super.setField(field, value);
-      }
-      @java.lang.Override
-      public Builder clearField(
-          com.google.protobuf.Descriptors.FieldDescriptor field) {
-        return super.clearField(field);
-      }
-      @java.lang.Override
-      public Builder clearOneof(
-          com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-        return super.clearOneof(oneof);
-      }
-      @java.lang.Override
-      public Builder setRepeatedField(
-          com.google.protobuf.Descriptors.FieldDescriptor field,
-          int index, java.lang.Object value) {
-        return super.setRepeatedField(field, index, value);
-      }
-      @java.lang.Override
-      public Builder addRepeatedField(
-          com.google.protobuf.Descriptors.FieldDescriptor field,
-          java.lang.Object value) {
-        return super.addRepeatedField(field, value);
-      }
-      @java.lang.Override
-      public Builder mergeFrom(com.google.protobuf.Message other) {
-        if (other instanceof org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol) {
-          return mergeFrom((org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol)other);
-        } else {
-          super.mergeFrom(other);
-          return this;
-        }
-      }
-
-      public Builder mergeFrom(org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol other) {
-        if (other == org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol.getDefaultInstance()) return this;
-        if (other.getFileIndex() != 0) {
-          setFileIndex(other.getFileIndex());
-        }
-        if (other.getLine() != 0) {
-          setLine(other.getLine());
-        }
-        if (other.getCol() != 0) {
-          setCol(other.getCol());
-        }
-        if (!other.getFunc().isEmpty()) {
-          func_ = other.func_;
-          onChanged();
-        }
-        if (!other.getCode().isEmpty()) {
-          code_ = other.code_;
-          onChanged();
-        }
-        this.mergeUnknownFields(other.unknownFields);
-        onChanged();
-        return this;
-      }
-
-      @java.lang.Override
-      public final boolean isInitialized() {
-        return true;
-      }
-
-      @java.lang.Override
-      public Builder mergeFrom(
-          com.google.protobuf.CodedInputStream input,
-          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-          throws java.io.IOException {
-        org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol parsedMessage = null;
-        try {
-          parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-        } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-          parsedMessage = (org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol) e.getUnfinishedMessage();
-          throw e.unwrapIOException();
-        } finally {
-          if (parsedMessage != null) {
-            mergeFrom(parsedMessage);
-          }
-        }
-        return this;
-      }
-
-      private int fileIndex_ ;
-      /**
-       * <pre>
-       * File name index, which can be used to retrieve the file name string from
-       * `files`. The value should be between 0 and (len(files)-1)
-       * </pre>
-       *
-       * <code>int32 file_index = 1;</code>
-       */
-      public int getFileIndex() {
-        return fileIndex_;
-      }
-      /**
-       * <pre>
-       * File name index, which can be used to retrieve the file name string from
-       * `files`. The value should be between 0 and (len(files)-1)
-       * </pre>
-       *
-       * <code>int32 file_index = 1;</code>
-       */
-      public Builder setFileIndex(int value) {
-        
-        fileIndex_ = value;
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * File name index, which can be used to retrieve the file name string from
-       * `files`. The value should be between 0 and (len(files)-1)
-       * </pre>
-       *
-       * <code>int32 file_index = 1;</code>
-       */
-      public Builder clearFileIndex() {
-        
-        fileIndex_ = 0;
-        onChanged();
-        return this;
-      }
-
-      private int line_ ;
-      /**
-       * <pre>
-       * Line number in the file.
-       * </pre>
-       *
-       * <code>int32 line = 2;</code>
-       */
-      public int getLine() {
-        return line_;
-      }
-      /**
-       * <pre>
-       * Line number in the file.
-       * </pre>
-       *
-       * <code>int32 line = 2;</code>
-       */
-      public Builder setLine(int value) {
-        
-        line_ = value;
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * Line number in the file.
-       * </pre>
-       *
-       * <code>int32 line = 2;</code>
-       */
-      public Builder clearLine() {
-        
-        line_ = 0;
-        onChanged();
-        return this;
-      }
-
-      private int col_ ;
-      /**
-       * <pre>
-       * Col number in the file line.
-       * </pre>
-       *
-       * <code>int32 col = 3;</code>
-       */
-      public int getCol() {
-        return col_;
-      }
-      /**
-       * <pre>
-       * Col number in the file line.
-       * </pre>
-       *
-       * <code>int32 col = 3;</code>
-       */
-      public Builder setCol(int value) {
-        
-        col_ = value;
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * Col number in the file line.
-       * </pre>
-       *
-       * <code>int32 col = 3;</code>
-       */
-      public Builder clearCol() {
-        
-        col_ = 0;
-        onChanged();
-        return this;
-      }
-
-      private java.lang.Object func_ = "";
-      /**
-       * <pre>
-       * Name of function contains the file line.
-       * </pre>
-       *
-       * <code>string func = 4;</code>
-       */
-      public java.lang.String getFunc() {
-        java.lang.Object ref = func_;
-        if (!(ref instanceof java.lang.String)) {
-          com.google.protobuf.ByteString bs =
-              (com.google.protobuf.ByteString) ref;
-          java.lang.String s = bs.toStringUtf8();
-          func_ = s;
-          return s;
-        } else {
-          return (java.lang.String) ref;
-        }
-      }
-      /**
-       * <pre>
-       * Name of function contains the file line.
-       * </pre>
-       *
-       * <code>string func = 4;</code>
-       */
-      public com.google.protobuf.ByteString
-          getFuncBytes() {
-        java.lang.Object ref = func_;
-        if (ref instanceof String) {
-          com.google.protobuf.ByteString b = 
-              com.google.protobuf.ByteString.copyFromUtf8(
-                  (java.lang.String) ref);
-          func_ = b;
-          return b;
-        } else {
-          return (com.google.protobuf.ByteString) ref;
-        }
-      }
-      /**
-       * <pre>
-       * Name of function contains the file line.
-       * </pre>
-       *
-       * <code>string func = 4;</code>
-       */
-      public Builder setFunc(
-          java.lang.String value) {
-        if (value == null) {
-    throw new NullPointerException();
-  }
-  
-        func_ = value;
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * Name of function contains the file line.
-       * </pre>
-       *
-       * <code>string func = 4;</code>
-       */
-      public Builder clearFunc() {
-        
-        func_ = getDefaultInstance().getFunc();
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * Name of function contains the file line.
-       * </pre>
-       *
-       * <code>string func = 4;</code>
-       */
-      public Builder setFuncBytes(
-          com.google.protobuf.ByteString value) {
-        if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-        
-        func_ = value;
-        onChanged();
-        return this;
-      }
-
-      private java.lang.Object code_ = "";
-      /**
-       * <pre>
-       * Source code contained in this file line.
-       * </pre>
-       *
-       * <code>string code = 5;</code>
-       */
-      public java.lang.String getCode() {
-        java.lang.Object ref = code_;
-        if (!(ref instanceof java.lang.String)) {
-          com.google.protobuf.ByteString bs =
-              (com.google.protobuf.ByteString) ref;
-          java.lang.String s = bs.toStringUtf8();
-          code_ = s;
-          return s;
-        } else {
-          return (java.lang.String) ref;
-        }
-      }
-      /**
-       * <pre>
-       * Source code contained in this file line.
-       * </pre>
-       *
-       * <code>string code = 5;</code>
-       */
-      public com.google.protobuf.ByteString
-          getCodeBytes() {
-        java.lang.Object ref = code_;
-        if (ref instanceof String) {
-          com.google.protobuf.ByteString b = 
-              com.google.protobuf.ByteString.copyFromUtf8(
-                  (java.lang.String) ref);
-          code_ = b;
-          return b;
-        } else {
-          return (com.google.protobuf.ByteString) ref;
-        }
-      }
-      /**
-       * <pre>
-       * Source code contained in this file line.
-       * </pre>
-       *
-       * <code>string code = 5;</code>
-       */
-      public Builder setCode(
-          java.lang.String value) {
-        if (value == null) {
-    throw new NullPointerException();
-  }
-  
-        code_ = value;
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * Source code contained in this file line.
-       * </pre>
-       *
-       * <code>string code = 5;</code>
-       */
-      public Builder clearCode() {
-        
-        code_ = getDefaultInstance().getCode();
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * Source code contained in this file line.
-       * </pre>
-       *
-       * <code>string code = 5;</code>
-       */
-      public Builder setCodeBytes(
-          com.google.protobuf.ByteString value) {
-        if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-        
-        code_ = value;
-        onChanged();
-        return this;
-      }
-      @java.lang.Override
-      public final Builder setUnknownFields(
-          final com.google.protobuf.UnknownFieldSet unknownFields) {
-        return super.setUnknownFields(unknownFields);
-      }
-
-      @java.lang.Override
-      public final Builder mergeUnknownFields(
-          final com.google.protobuf.UnknownFieldSet unknownFields) {
-        return super.mergeUnknownFields(unknownFields);
-      }
-
-
-      // @@protoc_insertion_point(builder_scope:tensorflow.GraphDebugInfo.FileLineCol)
-    }
-
-    // @@protoc_insertion_point(class_scope:tensorflow.GraphDebugInfo.FileLineCol)
-    private static final org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol DEFAULT_INSTANCE;
-    static {
-      DEFAULT_INSTANCE = new org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol();
-    }
-
-    public static org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol getDefaultInstance() {
-      return DEFAULT_INSTANCE;
-    }
-
-    private static final com.google.protobuf.Parser<FileLineCol>
-        PARSER = new com.google.protobuf.AbstractParser<FileLineCol>() {
-      @java.lang.Override
-      public FileLineCol parsePartialFrom(
-          com.google.protobuf.CodedInputStream input,
-          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-          throws com.google.protobuf.InvalidProtocolBufferException {
-        return new FileLineCol(input, extensionRegistry);
-      }
-    };
-
-    public static com.google.protobuf.Parser<FileLineCol> parser() {
-      return PARSER;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Parser<FileLineCol> getParserForType() {
-      return PARSER;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol getDefaultInstanceForType() {
-      return DEFAULT_INSTANCE;
-    }
-
-  }
-
-  public interface StackTraceOrBuilder extends
-      // @@protoc_insertion_point(interface_extends:tensorflow.GraphDebugInfo.StackTrace)
-      com.google.protobuf.MessageOrBuilder {
-
-    /**
-     * <pre>
-     * Each line in the stack trace.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.GraphDebugInfo.FileLineCol file_line_cols = 1;</code>
-     */
-    java.util.List<org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol> 
-        getFileLineColsList();
-    /**
-     * <pre>
-     * Each line in the stack trace.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.GraphDebugInfo.FileLineCol file_line_cols = 1;</code>
-     */
-    org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol getFileLineCols(int index);
-    /**
-     * <pre>
-     * Each line in the stack trace.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.GraphDebugInfo.FileLineCol file_line_cols = 1;</code>
-     */
-    int getFileLineColsCount();
-    /**
-     * <pre>
-     * Each line in the stack trace.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.GraphDebugInfo.FileLineCol file_line_cols = 1;</code>
-     */
-    java.util.List<? extends org.tensorflow.proto.framework.GraphDebugInfo.FileLineColOrBuilder> 
-        getFileLineColsOrBuilderList();
-    /**
-     * <pre>
-     * Each line in the stack trace.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.GraphDebugInfo.FileLineCol file_line_cols = 1;</code>
-     */
-    org.tensorflow.proto.framework.GraphDebugInfo.FileLineColOrBuilder getFileLineColsOrBuilder(
-        int index);
-  }
-  /**
-   * <pre>
-   * This represents a stack trace which is a ordered list of `FileLineCol`.
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.GraphDebugInfo.StackTrace}
-   */
-  public  static final class StackTrace extends
-      com.google.protobuf.GeneratedMessageV3 implements
-      // @@protoc_insertion_point(message_implements:tensorflow.GraphDebugInfo.StackTrace)
-      StackTraceOrBuilder {
-  private static final long serialVersionUID = 0L;
-    // Use StackTrace.newBuilder() to construct.
-    private StackTrace(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-      super(builder);
-    }
-    private StackTrace() {
-      fileLineCols_ = java.util.Collections.emptyList();
-    }
-
-    @java.lang.Override
-    @SuppressWarnings({"unused"})
-    protected java.lang.Object newInstance(
-        UnusedPrivateParameter unused) {
-      return new StackTrace();
-    }
-
-    @java.lang.Override
-    public final com.google.protobuf.UnknownFieldSet
-    getUnknownFields() {
-      return this.unknownFields;
-    }
-    private StackTrace(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      this();
-      if (extensionRegistry == null) {
-        throw new java.lang.NullPointerException();
-      }
-      int mutable_bitField0_ = 0;
-      com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-          com.google.protobuf.UnknownFieldSet.newBuilder();
-      try {
-        boolean done = false;
-        while (!done) {
-          int tag = input.readTag();
-          switch (tag) {
-            case 0:
-              done = true;
-              break;
-            case 10: {
-              if (!((mutable_bitField0_ & 0x00000001) != 0)) {
-                fileLineCols_ = new java.util.ArrayList<org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol>();
-                mutable_bitField0_ |= 0x00000001;
-              }
-              fileLineCols_.add(
-                  input.readMessage(org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol.parser(), extensionRegistry));
-              break;
-            }
-            default: {
-              if (!parseUnknownField(
-                  input, unknownFields, extensionRegistry, tag)) {
-                done = true;
-              }
-              break;
-            }
-          }
-        }
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        throw e.setUnfinishedMessage(this);
-      } catch (java.io.IOException e) {
-        throw new com.google.protobuf.InvalidProtocolBufferException(
-            e).setUnfinishedMessage(this);
-      } finally {
-        if (((mutable_bitField0_ & 0x00000001) != 0)) {
-          fileLineCols_ = java.util.Collections.unmodifiableList(fileLineCols_);
-        }
-        this.unknownFields = unknownFields.build();
-        makeExtensionsImmutable();
-      }
-    }
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.GraphDebugInfoProtos.internal_static_tensorflow_GraphDebugInfo_StackTrace_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.GraphDebugInfoProtos.internal_static_tensorflow_GraphDebugInfo_StackTrace_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.GraphDebugInfo.StackTrace.class, org.tensorflow.proto.framework.GraphDebugInfo.StackTrace.Builder.class);
-    }
-
-    public static final int FILE_LINE_COLS_FIELD_NUMBER = 1;
-    private java.util.List<org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol> fileLineCols_;
-    /**
-     * <pre>
-     * Each line in the stack trace.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.GraphDebugInfo.FileLineCol file_line_cols = 1;</code>
-     */
-    public java.util.List<org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol> getFileLineColsList() {
-      return fileLineCols_;
-    }
-    /**
-     * <pre>
-     * Each line in the stack trace.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.GraphDebugInfo.FileLineCol file_line_cols = 1;</code>
-     */
-    public java.util.List<? extends org.tensorflow.proto.framework.GraphDebugInfo.FileLineColOrBuilder> 
-        getFileLineColsOrBuilderList() {
-      return fileLineCols_;
-    }
-    /**
-     * <pre>
-     * Each line in the stack trace.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.GraphDebugInfo.FileLineCol file_line_cols = 1;</code>
-     */
-    public int getFileLineColsCount() {
-      return fileLineCols_.size();
-    }
-    /**
-     * <pre>
-     * Each line in the stack trace.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.GraphDebugInfo.FileLineCol file_line_cols = 1;</code>
-     */
-    public org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol getFileLineCols(int index) {
-      return fileLineCols_.get(index);
-    }
-    /**
-     * <pre>
-     * Each line in the stack trace.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.GraphDebugInfo.FileLineCol file_line_cols = 1;</code>
-     */
-    public org.tensorflow.proto.framework.GraphDebugInfo.FileLineColOrBuilder getFileLineColsOrBuilder(
-        int index) {
-      return fileLineCols_.get(index);
-    }
-
-    private byte memoizedIsInitialized = -1;
-    @java.lang.Override
-    public final boolean isInitialized() {
-      byte isInitialized = memoizedIsInitialized;
-      if (isInitialized == 1) return true;
-      if (isInitialized == 0) return false;
-
-      memoizedIsInitialized = 1;
-      return true;
-    }
-
-    @java.lang.Override
-    public void writeTo(com.google.protobuf.CodedOutputStream output)
-                        throws java.io.IOException {
-      for (int i = 0; i < fileLineCols_.size(); i++) {
-        output.writeMessage(1, fileLineCols_.get(i));
-      }
-      unknownFields.writeTo(output);
-    }
-
-    @java.lang.Override
-    public int getSerializedSize() {
-      int size = memoizedSize;
-      if (size != -1) return size;
-
-      size = 0;
-      for (int i = 0; i < fileLineCols_.size(); i++) {
-        size += com.google.protobuf.CodedOutputStream
-          .computeMessageSize(1, fileLineCols_.get(i));
-      }
-      size += unknownFields.getSerializedSize();
-      memoizedSize = size;
-      return size;
-    }
-
-    @java.lang.Override
-    public boolean equals(final java.lang.Object obj) {
-      if (obj == this) {
-       return true;
-      }
-      if (!(obj instanceof org.tensorflow.proto.framework.GraphDebugInfo.StackTrace)) {
-        return super.equals(obj);
-      }
-      org.tensorflow.proto.framework.GraphDebugInfo.StackTrace other = (org.tensorflow.proto.framework.GraphDebugInfo.StackTrace) obj;
-
-      if (!getFileLineColsList()
-          .equals(other.getFileLineColsList())) return false;
-      if (!unknownFields.equals(other.unknownFields)) return false;
-      return true;
-    }
-
-    @java.lang.Override
-    public int hashCode() {
-      if (memoizedHashCode != 0) {
-        return memoizedHashCode;
-      }
-      int hash = 41;
-      hash = (19 * hash) + getDescriptor().hashCode();
-      if (getFileLineColsCount() > 0) {
-        hash = (37 * hash) + FILE_LINE_COLS_FIELD_NUMBER;
-        hash = (53 * hash) + getFileLineColsList().hashCode();
-      }
-      hash = (29 * hash) + unknownFields.hashCode();
-      memoizedHashCode = hash;
-      return hash;
-    }
-
-    public static org.tensorflow.proto.framework.GraphDebugInfo.StackTrace parseFrom(
-        java.nio.ByteBuffer data)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data);
-    }
-    public static org.tensorflow.proto.framework.GraphDebugInfo.StackTrace parseFrom(
-        java.nio.ByteBuffer data,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.GraphDebugInfo.StackTrace parseFrom(
-        com.google.protobuf.ByteString data)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data);
-    }
-    public static org.tensorflow.proto.framework.GraphDebugInfo.StackTrace parseFrom(
-        com.google.protobuf.ByteString data,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.GraphDebugInfo.StackTrace parseFrom(byte[] data)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data);
-    }
-    public static org.tensorflow.proto.framework.GraphDebugInfo.StackTrace parseFrom(
-        byte[] data,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.GraphDebugInfo.StackTrace parseFrom(java.io.InputStream input)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input);
-    }
-    public static org.tensorflow.proto.framework.GraphDebugInfo.StackTrace parseFrom(
-        java.io.InputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.GraphDebugInfo.StackTrace parseDelimitedFrom(java.io.InputStream input)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseDelimitedWithIOException(PARSER, input);
-    }
-    public static org.tensorflow.proto.framework.GraphDebugInfo.StackTrace parseDelimitedFrom(
-        java.io.InputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.GraphDebugInfo.StackTrace parseFrom(
-        com.google.protobuf.CodedInputStream input)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input);
-    }
-    public static org.tensorflow.proto.framework.GraphDebugInfo.StackTrace parseFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input, extensionRegistry);
-    }
-
-    @java.lang.Override
-    public Builder newBuilderForType() { return newBuilder(); }
-    public static Builder newBuilder() {
-      return DEFAULT_INSTANCE.toBuilder();
-    }
-    public static Builder newBuilder(org.tensorflow.proto.framework.GraphDebugInfo.StackTrace prototype) {
-      return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-    }
-    @java.lang.Override
-    public Builder toBuilder() {
-      return this == DEFAULT_INSTANCE
-          ? new Builder() : new Builder().mergeFrom(this);
-    }
-
-    @java.lang.Override
-    protected Builder newBuilderForType(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      Builder builder = new Builder(parent);
-      return builder;
-    }
-    /**
-     * <pre>
-     * This represents a stack trace which is a ordered list of `FileLineCol`.
-     * </pre>
-     *
-     * Protobuf type {@code tensorflow.GraphDebugInfo.StackTrace}
-     */
-    public static final class Builder extends
-        com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-        // @@protoc_insertion_point(builder_implements:tensorflow.GraphDebugInfo.StackTrace)
-        org.tensorflow.proto.framework.GraphDebugInfo.StackTraceOrBuilder {
-      public static final com.google.protobuf.Descriptors.Descriptor
-          getDescriptor() {
-        return org.tensorflow.proto.framework.GraphDebugInfoProtos.internal_static_tensorflow_GraphDebugInfo_StackTrace_descriptor;
-      }
-
-      @java.lang.Override
-      protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-          internalGetFieldAccessorTable() {
-        return org.tensorflow.proto.framework.GraphDebugInfoProtos.internal_static_tensorflow_GraphDebugInfo_StackTrace_fieldAccessorTable
-            .ensureFieldAccessorsInitialized(
-                org.tensorflow.proto.framework.GraphDebugInfo.StackTrace.class, org.tensorflow.proto.framework.GraphDebugInfo.StackTrace.Builder.class);
-      }
-
-      // Construct using org.tensorflow.proto.framework.GraphDebugInfo.StackTrace.newBuilder()
-      private Builder() {
-        maybeForceBuilderInitialization();
-      }
-
-      private Builder(
-          com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-        super(parent);
-        maybeForceBuilderInitialization();
-      }
-      private void maybeForceBuilderInitialization() {
-        if (com.google.protobuf.GeneratedMessageV3
-                .alwaysUseFieldBuilders) {
-          getFileLineColsFieldBuilder();
-        }
-      }
-      @java.lang.Override
-      public Builder clear() {
-        super.clear();
-        if (fileLineColsBuilder_ == null) {
-          fileLineCols_ = java.util.Collections.emptyList();
-          bitField0_ = (bitField0_ & ~0x00000001);
-        } else {
-          fileLineColsBuilder_.clear();
-        }
-        return this;
-      }
-
-      @java.lang.Override
-      public com.google.protobuf.Descriptors.Descriptor
-          getDescriptorForType() {
-        return org.tensorflow.proto.framework.GraphDebugInfoProtos.internal_static_tensorflow_GraphDebugInfo_StackTrace_descriptor;
-      }
-
-      @java.lang.Override
-      public org.tensorflow.proto.framework.GraphDebugInfo.StackTrace getDefaultInstanceForType() {
-        return org.tensorflow.proto.framework.GraphDebugInfo.StackTrace.getDefaultInstance();
-      }
-
-      @java.lang.Override
-      public org.tensorflow.proto.framework.GraphDebugInfo.StackTrace build() {
-        org.tensorflow.proto.framework.GraphDebugInfo.StackTrace result = buildPartial();
-        if (!result.isInitialized()) {
-          throw newUninitializedMessageException(result);
-        }
-        return result;
-      }
-
-      @java.lang.Override
-      public org.tensorflow.proto.framework.GraphDebugInfo.StackTrace buildPartial() {
-        org.tensorflow.proto.framework.GraphDebugInfo.StackTrace result = new org.tensorflow.proto.framework.GraphDebugInfo.StackTrace(this);
-        int from_bitField0_ = bitField0_;
-        if (fileLineColsBuilder_ == null) {
-          if (((bitField0_ & 0x00000001) != 0)) {
-            fileLineCols_ = java.util.Collections.unmodifiableList(fileLineCols_);
-            bitField0_ = (bitField0_ & ~0x00000001);
-          }
-          result.fileLineCols_ = fileLineCols_;
-        } else {
-          result.fileLineCols_ = fileLineColsBuilder_.build();
-        }
-        onBuilt();
-        return result;
-      }
-
-      @java.lang.Override
-      public Builder clone() {
-        return super.clone();
-      }
-      @java.lang.Override
-      public Builder setField(
-          com.google.protobuf.Descriptors.FieldDescriptor field,
-          java.lang.Object value) {
-        return super.setField(field, value);
-      }
-      @java.lang.Override
-      public Builder clearField(
-          com.google.protobuf.Descriptors.FieldDescriptor field) {
-        return super.clearField(field);
-      }
-      @java.lang.Override
-      public Builder clearOneof(
-          com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-        return super.clearOneof(oneof);
-      }
-      @java.lang.Override
-      public Builder setRepeatedField(
-          com.google.protobuf.Descriptors.FieldDescriptor field,
-          int index, java.lang.Object value) {
-        return super.setRepeatedField(field, index, value);
-      }
-      @java.lang.Override
-      public Builder addRepeatedField(
-          com.google.protobuf.Descriptors.FieldDescriptor field,
-          java.lang.Object value) {
-        return super.addRepeatedField(field, value);
-      }
-      @java.lang.Override
-      public Builder mergeFrom(com.google.protobuf.Message other) {
-        if (other instanceof org.tensorflow.proto.framework.GraphDebugInfo.StackTrace) {
-          return mergeFrom((org.tensorflow.proto.framework.GraphDebugInfo.StackTrace)other);
-        } else {
-          super.mergeFrom(other);
-          return this;
-        }
-      }
-
-      public Builder mergeFrom(org.tensorflow.proto.framework.GraphDebugInfo.StackTrace other) {
-        if (other == org.tensorflow.proto.framework.GraphDebugInfo.StackTrace.getDefaultInstance()) return this;
-        if (fileLineColsBuilder_ == null) {
-          if (!other.fileLineCols_.isEmpty()) {
-            if (fileLineCols_.isEmpty()) {
-              fileLineCols_ = other.fileLineCols_;
-              bitField0_ = (bitField0_ & ~0x00000001);
-            } else {
-              ensureFileLineColsIsMutable();
-              fileLineCols_.addAll(other.fileLineCols_);
-            }
-            onChanged();
-          }
-        } else {
-          if (!other.fileLineCols_.isEmpty()) {
-            if (fileLineColsBuilder_.isEmpty()) {
-              fileLineColsBuilder_.dispose();
-              fileLineColsBuilder_ = null;
-              fileLineCols_ = other.fileLineCols_;
-              bitField0_ = (bitField0_ & ~0x00000001);
-              fileLineColsBuilder_ = 
-                com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
-                   getFileLineColsFieldBuilder() : null;
-            } else {
-              fileLineColsBuilder_.addAllMessages(other.fileLineCols_);
-            }
-          }
-        }
-        this.mergeUnknownFields(other.unknownFields);
-        onChanged();
-        return this;
-      }
-
-      @java.lang.Override
-      public final boolean isInitialized() {
-        return true;
-      }
-
-      @java.lang.Override
-      public Builder mergeFrom(
-          com.google.protobuf.CodedInputStream input,
-          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-          throws java.io.IOException {
-        org.tensorflow.proto.framework.GraphDebugInfo.StackTrace parsedMessage = null;
-        try {
-          parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-        } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-          parsedMessage = (org.tensorflow.proto.framework.GraphDebugInfo.StackTrace) e.getUnfinishedMessage();
-          throw e.unwrapIOException();
-        } finally {
-          if (parsedMessage != null) {
-            mergeFrom(parsedMessage);
-          }
-        }
-        return this;
-      }
-      private int bitField0_;
-
-      private java.util.List<org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol> fileLineCols_ =
-        java.util.Collections.emptyList();
-      private void ensureFileLineColsIsMutable() {
-        if (!((bitField0_ & 0x00000001) != 0)) {
-          fileLineCols_ = new java.util.ArrayList<org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol>(fileLineCols_);
-          bitField0_ |= 0x00000001;
-         }
-      }
-
-      private com.google.protobuf.RepeatedFieldBuilderV3<
-          org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol, org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol.Builder, org.tensorflow.proto.framework.GraphDebugInfo.FileLineColOrBuilder> fileLineColsBuilder_;
-
-      /**
-       * <pre>
-       * Each line in the stack trace.
-       * </pre>
-       *
-       * <code>repeated .tensorflow.GraphDebugInfo.FileLineCol file_line_cols = 1;</code>
-       */
-      public java.util.List<org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol> getFileLineColsList() {
-        if (fileLineColsBuilder_ == null) {
-          return java.util.Collections.unmodifiableList(fileLineCols_);
-        } else {
-          return fileLineColsBuilder_.getMessageList();
-        }
-      }
-      /**
-       * <pre>
-       * Each line in the stack trace.
-       * </pre>
-       *
-       * <code>repeated .tensorflow.GraphDebugInfo.FileLineCol file_line_cols = 1;</code>
-       */
-      public int getFileLineColsCount() {
-        if (fileLineColsBuilder_ == null) {
-          return fileLineCols_.size();
-        } else {
-          return fileLineColsBuilder_.getCount();
-        }
-      }
-      /**
-       * <pre>
-       * Each line in the stack trace.
-       * </pre>
-       *
-       * <code>repeated .tensorflow.GraphDebugInfo.FileLineCol file_line_cols = 1;</code>
-       */
-      public org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol getFileLineCols(int index) {
-        if (fileLineColsBuilder_ == null) {
-          return fileLineCols_.get(index);
-        } else {
-          return fileLineColsBuilder_.getMessage(index);
-        }
-      }
-      /**
-       * <pre>
-       * Each line in the stack trace.
-       * </pre>
-       *
-       * <code>repeated .tensorflow.GraphDebugInfo.FileLineCol file_line_cols = 1;</code>
-       */
-      public Builder setFileLineCols(
-          int index, org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol value) {
-        if (fileLineColsBuilder_ == null) {
-          if (value == null) {
-            throw new NullPointerException();
-          }
-          ensureFileLineColsIsMutable();
-          fileLineCols_.set(index, value);
-          onChanged();
-        } else {
-          fileLineColsBuilder_.setMessage(index, value);
-        }
-        return this;
-      }
-      /**
-       * <pre>
-       * Each line in the stack trace.
-       * </pre>
-       *
-       * <code>repeated .tensorflow.GraphDebugInfo.FileLineCol file_line_cols = 1;</code>
-       */
-      public Builder setFileLineCols(
-          int index, org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol.Builder builderForValue) {
-        if (fileLineColsBuilder_ == null) {
-          ensureFileLineColsIsMutable();
-          fileLineCols_.set(index, builderForValue.build());
-          onChanged();
-        } else {
-          fileLineColsBuilder_.setMessage(index, builderForValue.build());
-        }
-        return this;
-      }
-      /**
-       * <pre>
-       * Each line in the stack trace.
-       * </pre>
-       *
-       * <code>repeated .tensorflow.GraphDebugInfo.FileLineCol file_line_cols = 1;</code>
-       */
-      public Builder addFileLineCols(org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol value) {
-        if (fileLineColsBuilder_ == null) {
-          if (value == null) {
-            throw new NullPointerException();
-          }
-          ensureFileLineColsIsMutable();
-          fileLineCols_.add(value);
-          onChanged();
-        } else {
-          fileLineColsBuilder_.addMessage(value);
-        }
-        return this;
-      }
-      /**
-       * <pre>
-       * Each line in the stack trace.
-       * </pre>
-       *
-       * <code>repeated .tensorflow.GraphDebugInfo.FileLineCol file_line_cols = 1;</code>
-       */
-      public Builder addFileLineCols(
-          int index, org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol value) {
-        if (fileLineColsBuilder_ == null) {
-          if (value == null) {
-            throw new NullPointerException();
-          }
-          ensureFileLineColsIsMutable();
-          fileLineCols_.add(index, value);
-          onChanged();
-        } else {
-          fileLineColsBuilder_.addMessage(index, value);
-        }
-        return this;
-      }
-      /**
-       * <pre>
-       * Each line in the stack trace.
-       * </pre>
-       *
-       * <code>repeated .tensorflow.GraphDebugInfo.FileLineCol file_line_cols = 1;</code>
-       */
-      public Builder addFileLineCols(
-          org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol.Builder builderForValue) {
-        if (fileLineColsBuilder_ == null) {
-          ensureFileLineColsIsMutable();
-          fileLineCols_.add(builderForValue.build());
-          onChanged();
-        } else {
-          fileLineColsBuilder_.addMessage(builderForValue.build());
-        }
-        return this;
-      }
-      /**
-       * <pre>
-       * Each line in the stack trace.
-       * </pre>
-       *
-       * <code>repeated .tensorflow.GraphDebugInfo.FileLineCol file_line_cols = 1;</code>
-       */
-      public Builder addFileLineCols(
-          int index, org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol.Builder builderForValue) {
-        if (fileLineColsBuilder_ == null) {
-          ensureFileLineColsIsMutable();
-          fileLineCols_.add(index, builderForValue.build());
-          onChanged();
-        } else {
-          fileLineColsBuilder_.addMessage(index, builderForValue.build());
-        }
-        return this;
-      }
-      /**
-       * <pre>
-       * Each line in the stack trace.
-       * </pre>
-       *
-       * <code>repeated .tensorflow.GraphDebugInfo.FileLineCol file_line_cols = 1;</code>
-       */
-      public Builder addAllFileLineCols(
-          java.lang.Iterable<? extends org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol> values) {
-        if (fileLineColsBuilder_ == null) {
-          ensureFileLineColsIsMutable();
-          com.google.protobuf.AbstractMessageLite.Builder.addAll(
-              values, fileLineCols_);
-          onChanged();
-        } else {
-          fileLineColsBuilder_.addAllMessages(values);
-        }
-        return this;
-      }
-      /**
-       * <pre>
-       * Each line in the stack trace.
-       * </pre>
-       *
-       * <code>repeated .tensorflow.GraphDebugInfo.FileLineCol file_line_cols = 1;</code>
-       */
-      public Builder clearFileLineCols() {
-        if (fileLineColsBuilder_ == null) {
-          fileLineCols_ = java.util.Collections.emptyList();
-          bitField0_ = (bitField0_ & ~0x00000001);
-          onChanged();
-        } else {
-          fileLineColsBuilder_.clear();
-        }
-        return this;
-      }
-      /**
-       * <pre>
-       * Each line in the stack trace.
-       * </pre>
-       *
-       * <code>repeated .tensorflow.GraphDebugInfo.FileLineCol file_line_cols = 1;</code>
-       */
-      public Builder removeFileLineCols(int index) {
-        if (fileLineColsBuilder_ == null) {
-          ensureFileLineColsIsMutable();
-          fileLineCols_.remove(index);
-          onChanged();
-        } else {
-          fileLineColsBuilder_.remove(index);
-        }
-        return this;
-      }
-      /**
-       * <pre>
-       * Each line in the stack trace.
-       * </pre>
-       *
-       * <code>repeated .tensorflow.GraphDebugInfo.FileLineCol file_line_cols = 1;</code>
-       */
-      public org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol.Builder getFileLineColsBuilder(
-          int index) {
-        return getFileLineColsFieldBuilder().getBuilder(index);
-      }
-      /**
-       * <pre>
-       * Each line in the stack trace.
-       * </pre>
-       *
-       * <code>repeated .tensorflow.GraphDebugInfo.FileLineCol file_line_cols = 1;</code>
-       */
-      public org.tensorflow.proto.framework.GraphDebugInfo.FileLineColOrBuilder getFileLineColsOrBuilder(
-          int index) {
-        if (fileLineColsBuilder_ == null) {
-          return fileLineCols_.get(index);  } else {
-          return fileLineColsBuilder_.getMessageOrBuilder(index);
-        }
-      }
-      /**
-       * <pre>
-       * Each line in the stack trace.
-       * </pre>
-       *
-       * <code>repeated .tensorflow.GraphDebugInfo.FileLineCol file_line_cols = 1;</code>
-       */
-      public java.util.List<? extends org.tensorflow.proto.framework.GraphDebugInfo.FileLineColOrBuilder> 
-           getFileLineColsOrBuilderList() {
-        if (fileLineColsBuilder_ != null) {
-          return fileLineColsBuilder_.getMessageOrBuilderList();
-        } else {
-          return java.util.Collections.unmodifiableList(fileLineCols_);
-        }
-      }
-      /**
-       * <pre>
-       * Each line in the stack trace.
-       * </pre>
-       *
-       * <code>repeated .tensorflow.GraphDebugInfo.FileLineCol file_line_cols = 1;</code>
-       */
-      public org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol.Builder addFileLineColsBuilder() {
-        return getFileLineColsFieldBuilder().addBuilder(
-            org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol.getDefaultInstance());
-      }
-      /**
-       * <pre>
-       * Each line in the stack trace.
-       * </pre>
-       *
-       * <code>repeated .tensorflow.GraphDebugInfo.FileLineCol file_line_cols = 1;</code>
-       */
-      public org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol.Builder addFileLineColsBuilder(
-          int index) {
-        return getFileLineColsFieldBuilder().addBuilder(
-            index, org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol.getDefaultInstance());
-      }
-      /**
-       * <pre>
-       * Each line in the stack trace.
-       * </pre>
-       *
-       * <code>repeated .tensorflow.GraphDebugInfo.FileLineCol file_line_cols = 1;</code>
-       */
-      public java.util.List<org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol.Builder> 
-           getFileLineColsBuilderList() {
-        return getFileLineColsFieldBuilder().getBuilderList();
-      }
-      private com.google.protobuf.RepeatedFieldBuilderV3<
-          org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol, org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol.Builder, org.tensorflow.proto.framework.GraphDebugInfo.FileLineColOrBuilder> 
-          getFileLineColsFieldBuilder() {
-        if (fileLineColsBuilder_ == null) {
-          fileLineColsBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
-              org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol, org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol.Builder, org.tensorflow.proto.framework.GraphDebugInfo.FileLineColOrBuilder>(
-                  fileLineCols_,
-                  ((bitField0_ & 0x00000001) != 0),
-                  getParentForChildren(),
-                  isClean());
-          fileLineCols_ = null;
-        }
-        return fileLineColsBuilder_;
-      }
-      @java.lang.Override
-      public final Builder setUnknownFields(
-          final com.google.protobuf.UnknownFieldSet unknownFields) {
-        return super.setUnknownFields(unknownFields);
-      }
-
-      @java.lang.Override
-      public final Builder mergeUnknownFields(
-          final com.google.protobuf.UnknownFieldSet unknownFields) {
-        return super.mergeUnknownFields(unknownFields);
-      }
-
-
-      // @@protoc_insertion_point(builder_scope:tensorflow.GraphDebugInfo.StackTrace)
-    }
-
-    // @@protoc_insertion_point(class_scope:tensorflow.GraphDebugInfo.StackTrace)
-    private static final org.tensorflow.proto.framework.GraphDebugInfo.StackTrace DEFAULT_INSTANCE;
-    static {
-      DEFAULT_INSTANCE = new org.tensorflow.proto.framework.GraphDebugInfo.StackTrace();
-    }
-
-    public static org.tensorflow.proto.framework.GraphDebugInfo.StackTrace getDefaultInstance() {
-      return DEFAULT_INSTANCE;
-    }
-
-    private static final com.google.protobuf.Parser<StackTrace>
-        PARSER = new com.google.protobuf.AbstractParser<StackTrace>() {
-      @java.lang.Override
-      public StackTrace parsePartialFrom(
-          com.google.protobuf.CodedInputStream input,
-          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-          throws com.google.protobuf.InvalidProtocolBufferException {
-        return new StackTrace(input, extensionRegistry);
-      }
-    };
-
-    public static com.google.protobuf.Parser<StackTrace> parser() {
-      return PARSER;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Parser<StackTrace> getParserForType() {
-      return PARSER;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.GraphDebugInfo.StackTrace getDefaultInstanceForType() {
-      return DEFAULT_INSTANCE;
-    }
-
-  }
-
-  public static final int FILES_FIELD_NUMBER = 1;
-  private com.google.protobuf.LazyStringList files_;
-  /**
-   * <pre>
-   * This stores all the source code file names and can be indexed by the
-   * `file_index`.
-   * </pre>
-   *
-   * <code>repeated string files = 1;</code>
-   */
-  public com.google.protobuf.ProtocolStringList
-      getFilesList() {
-    return files_;
-  }
-  /**
-   * <pre>
-   * This stores all the source code file names and can be indexed by the
-   * `file_index`.
-   * </pre>
-   *
-   * <code>repeated string files = 1;</code>
-   */
-  public int getFilesCount() {
-    return files_.size();
-  }
-  /**
-   * <pre>
-   * This stores all the source code file names and can be indexed by the
-   * `file_index`.
-   * </pre>
-   *
-   * <code>repeated string files = 1;</code>
-   */
-  public java.lang.String getFiles(int index) {
-    return files_.get(index);
-  }
-  /**
-   * <pre>
-   * This stores all the source code file names and can be indexed by the
-   * `file_index`.
-   * </pre>
-   *
-   * <code>repeated string files = 1;</code>
-   */
-  public com.google.protobuf.ByteString
-      getFilesBytes(int index) {
-    return files_.getByteString(index);
-  }
-
-  public static final int TRACES_FIELD_NUMBER = 2;
-  private static final class TracesDefaultEntryHolder {
-    static final com.google.protobuf.MapEntry<
-        java.lang.String, org.tensorflow.proto.framework.GraphDebugInfo.StackTrace> defaultEntry =
-            com.google.protobuf.MapEntry
-            .<java.lang.String, org.tensorflow.proto.framework.GraphDebugInfo.StackTrace>newDefaultInstance(
-                org.tensorflow.proto.framework.GraphDebugInfoProtos.internal_static_tensorflow_GraphDebugInfo_TracesEntry_descriptor, 
-                com.google.protobuf.WireFormat.FieldType.STRING,
-                "",
-                com.google.protobuf.WireFormat.FieldType.MESSAGE,
-                org.tensorflow.proto.framework.GraphDebugInfo.StackTrace.getDefaultInstance());
-  }
-  private com.google.protobuf.MapField<
-      java.lang.String, org.tensorflow.proto.framework.GraphDebugInfo.StackTrace> traces_;
-  private com.google.protobuf.MapField<java.lang.String, org.tensorflow.proto.framework.GraphDebugInfo.StackTrace>
-  internalGetTraces() {
-    if (traces_ == null) {
-      return com.google.protobuf.MapField.emptyMapField(
-          TracesDefaultEntryHolder.defaultEntry);
-    }
-    return traces_;
-  }
-
-  public int getTracesCount() {
-    return internalGetTraces().getMap().size();
-  }
-  /**
-   * <pre>
-   * This maps a node name to a stack trace in the source code.
-   * The map key is a mangling of the containing function and op name with
-   * syntax:
-   *   op.name '&#64;' func_name
-   * For ops in the top-level graph, the func_name is the empty string.
-   * Note that op names are restricted to a small number of characters which
-   * exclude '&#64;', making it impossible to collide keys of this form. Function
-   * names accept a much wider set of characters.
-   * It would be preferable to avoid mangling and use a tuple key of (op.name,
-   * func_name), but this is not supported with protocol buffers.
-   * </pre>
-   *
-   * <code>map&lt;string, .tensorflow.GraphDebugInfo.StackTrace&gt; traces = 2;</code>
-   */
-
-  public boolean containsTraces(
-      java.lang.String key) {
-    if (key == null) { throw new java.lang.NullPointerException(); }
-    return internalGetTraces().getMap().containsKey(key);
-  }
-  /**
-   * Use {@link #getTracesMap()} instead.
-   */
-  @java.lang.Deprecated
-  public java.util.Map<java.lang.String, org.tensorflow.proto.framework.GraphDebugInfo.StackTrace> getTraces() {
-    return getTracesMap();
-  }
-  /**
-   * <pre>
-   * This maps a node name to a stack trace in the source code.
-   * The map key is a mangling of the containing function and op name with
-   * syntax:
-   *   op.name '&#64;' func_name
-   * For ops in the top-level graph, the func_name is the empty string.
-   * Note that op names are restricted to a small number of characters which
-   * exclude '&#64;', making it impossible to collide keys of this form. Function
-   * names accept a much wider set of characters.
-   * It would be preferable to avoid mangling and use a tuple key of (op.name,
-   * func_name), but this is not supported with protocol buffers.
-   * </pre>
-   *
-   * <code>map&lt;string, .tensorflow.GraphDebugInfo.StackTrace&gt; traces = 2;</code>
-   */
-
-  public java.util.Map<java.lang.String, org.tensorflow.proto.framework.GraphDebugInfo.StackTrace> getTracesMap() {
-    return internalGetTraces().getMap();
-  }
-  /**
-   * <pre>
-   * This maps a node name to a stack trace in the source code.
-   * The map key is a mangling of the containing function and op name with
-   * syntax:
-   *   op.name '&#64;' func_name
-   * For ops in the top-level graph, the func_name is the empty string.
-   * Note that op names are restricted to a small number of characters which
-   * exclude '&#64;', making it impossible to collide keys of this form. Function
-   * names accept a much wider set of characters.
-   * It would be preferable to avoid mangling and use a tuple key of (op.name,
-   * func_name), but this is not supported with protocol buffers.
-   * </pre>
-   *
-   * <code>map&lt;string, .tensorflow.GraphDebugInfo.StackTrace&gt; traces = 2;</code>
-   */
-
-  public org.tensorflow.proto.framework.GraphDebugInfo.StackTrace getTracesOrDefault(
-      java.lang.String key,
-      org.tensorflow.proto.framework.GraphDebugInfo.StackTrace defaultValue) {
-    if (key == null) { throw new java.lang.NullPointerException(); }
-    java.util.Map<java.lang.String, org.tensorflow.proto.framework.GraphDebugInfo.StackTrace> map =
-        internalGetTraces().getMap();
-    return map.containsKey(key) ? map.get(key) : defaultValue;
-  }
-  /**
-   * <pre>
-   * This maps a node name to a stack trace in the source code.
-   * The map key is a mangling of the containing function and op name with
-   * syntax:
-   *   op.name '&#64;' func_name
-   * For ops in the top-level graph, the func_name is the empty string.
-   * Note that op names are restricted to a small number of characters which
-   * exclude '&#64;', making it impossible to collide keys of this form. Function
-   * names accept a much wider set of characters.
-   * It would be preferable to avoid mangling and use a tuple key of (op.name,
-   * func_name), but this is not supported with protocol buffers.
-   * </pre>
-   *
-   * <code>map&lt;string, .tensorflow.GraphDebugInfo.StackTrace&gt; traces = 2;</code>
-   */
-
-  public org.tensorflow.proto.framework.GraphDebugInfo.StackTrace getTracesOrThrow(
-      java.lang.String key) {
-    if (key == null) { throw new java.lang.NullPointerException(); }
-    java.util.Map<java.lang.String, org.tensorflow.proto.framework.GraphDebugInfo.StackTrace> map =
-        internalGetTraces().getMap();
-    if (!map.containsKey(key)) {
-      throw new java.lang.IllegalArgumentException();
-    }
-    return map.get(key);
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    for (int i = 0; i < files_.size(); i++) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 1, files_.getRaw(i));
-    }
-    com.google.protobuf.GeneratedMessageV3
-      .serializeStringMapTo(
-        output,
-        internalGetTraces(),
-        TracesDefaultEntryHolder.defaultEntry,
-        2);
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    {
-      int dataSize = 0;
-      for (int i = 0; i < files_.size(); i++) {
-        dataSize += computeStringSizeNoTag(files_.getRaw(i));
-      }
-      size += dataSize;
-      size += 1 * getFilesList().size();
-    }
-    for (java.util.Map.Entry<java.lang.String, org.tensorflow.proto.framework.GraphDebugInfo.StackTrace> entry
-         : internalGetTraces().getMap().entrySet()) {
-      com.google.protobuf.MapEntry<java.lang.String, org.tensorflow.proto.framework.GraphDebugInfo.StackTrace>
-      traces__ = TracesDefaultEntryHolder.defaultEntry.newBuilderForType()
-          .setKey(entry.getKey())
-          .setValue(entry.getValue())
-          .build();
-      size += com.google.protobuf.CodedOutputStream
-          .computeMessageSize(2, traces__);
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.framework.GraphDebugInfo)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.framework.GraphDebugInfo other = (org.tensorflow.proto.framework.GraphDebugInfo) obj;
-
-    if (!getFilesList()
-        .equals(other.getFilesList())) return false;
-    if (!internalGetTraces().equals(
-        other.internalGetTraces())) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    if (getFilesCount() > 0) {
-      hash = (37 * hash) + FILES_FIELD_NUMBER;
-      hash = (53 * hash) + getFilesList().hashCode();
-    }
-    if (!internalGetTraces().getMap().isEmpty()) {
-      hash = (37 * hash) + TRACES_FIELD_NUMBER;
-      hash = (53 * hash) + internalGetTraces().hashCode();
-    }
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.framework.GraphDebugInfo parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.GraphDebugInfo parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.GraphDebugInfo parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.GraphDebugInfo parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.GraphDebugInfo parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.GraphDebugInfo parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.GraphDebugInfo parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.GraphDebugInfo parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.GraphDebugInfo parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.GraphDebugInfo parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.GraphDebugInfo parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.GraphDebugInfo parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.framework.GraphDebugInfo prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * Protobuf type {@code tensorflow.GraphDebugInfo}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.GraphDebugInfo)
-      org.tensorflow.proto.framework.GraphDebugInfoOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.GraphDebugInfoProtos.internal_static_tensorflow_GraphDebugInfo_descriptor;
-    }
-
-    @SuppressWarnings({"rawtypes"})
-    protected com.google.protobuf.MapField internalGetMapField(
-        int number) {
-      switch (number) {
-        case 2:
-          return internalGetTraces();
-        default:
-          throw new RuntimeException(
-              "Invalid map field number: " + number);
-      }
-    }
-    @SuppressWarnings({"rawtypes"})
-    protected com.google.protobuf.MapField internalGetMutableMapField(
-        int number) {
-      switch (number) {
-        case 2:
-          return internalGetMutableTraces();
-        default:
-          throw new RuntimeException(
-              "Invalid map field number: " + number);
-      }
-    }
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.GraphDebugInfoProtos.internal_static_tensorflow_GraphDebugInfo_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.GraphDebugInfo.class, org.tensorflow.proto.framework.GraphDebugInfo.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.framework.GraphDebugInfo.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      files_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-      bitField0_ = (bitField0_ & ~0x00000001);
-      internalGetMutableTraces().clear();
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.framework.GraphDebugInfoProtos.internal_static_tensorflow_GraphDebugInfo_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.GraphDebugInfo getDefaultInstanceForType() {
-      return org.tensorflow.proto.framework.GraphDebugInfo.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.GraphDebugInfo build() {
-      org.tensorflow.proto.framework.GraphDebugInfo result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.GraphDebugInfo buildPartial() {
-      org.tensorflow.proto.framework.GraphDebugInfo result = new org.tensorflow.proto.framework.GraphDebugInfo(this);
-      int from_bitField0_ = bitField0_;
-      if (((bitField0_ & 0x00000001) != 0)) {
-        files_ = files_.getUnmodifiableView();
-        bitField0_ = (bitField0_ & ~0x00000001);
-      }
-      result.files_ = files_;
-      result.traces_ = internalGetTraces();
-      result.traces_.makeImmutable();
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.framework.GraphDebugInfo) {
-        return mergeFrom((org.tensorflow.proto.framework.GraphDebugInfo)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.framework.GraphDebugInfo other) {
-      if (other == org.tensorflow.proto.framework.GraphDebugInfo.getDefaultInstance()) return this;
-      if (!other.files_.isEmpty()) {
-        if (files_.isEmpty()) {
-          files_ = other.files_;
-          bitField0_ = (bitField0_ & ~0x00000001);
-        } else {
-          ensureFilesIsMutable();
-          files_.addAll(other.files_);
-        }
-        onChanged();
-      }
-      internalGetMutableTraces().mergeFrom(
-          other.internalGetTraces());
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.framework.GraphDebugInfo parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.framework.GraphDebugInfo) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-    private int bitField0_;
-
-    private com.google.protobuf.LazyStringList files_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-    private void ensureFilesIsMutable() {
-      if (!((bitField0_ & 0x00000001) != 0)) {
-        files_ = new com.google.protobuf.LazyStringArrayList(files_);
-        bitField0_ |= 0x00000001;
-       }
-    }
-    /**
-     * <pre>
-     * This stores all the source code file names and can be indexed by the
-     * `file_index`.
-     * </pre>
-     *
-     * <code>repeated string files = 1;</code>
-     */
-    public com.google.protobuf.ProtocolStringList
-        getFilesList() {
-      return files_.getUnmodifiableView();
-    }
-    /**
-     * <pre>
-     * This stores all the source code file names and can be indexed by the
-     * `file_index`.
-     * </pre>
-     *
-     * <code>repeated string files = 1;</code>
-     */
-    public int getFilesCount() {
-      return files_.size();
-    }
-    /**
-     * <pre>
-     * This stores all the source code file names and can be indexed by the
-     * `file_index`.
-     * </pre>
-     *
-     * <code>repeated string files = 1;</code>
-     */
-    public java.lang.String getFiles(int index) {
-      return files_.get(index);
-    }
-    /**
-     * <pre>
-     * This stores all the source code file names and can be indexed by the
-     * `file_index`.
-     * </pre>
-     *
-     * <code>repeated string files = 1;</code>
-     */
-    public com.google.protobuf.ByteString
-        getFilesBytes(int index) {
-      return files_.getByteString(index);
-    }
-    /**
-     * <pre>
-     * This stores all the source code file names and can be indexed by the
-     * `file_index`.
-     * </pre>
-     *
-     * <code>repeated string files = 1;</code>
-     */
-    public Builder setFiles(
-        int index, java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  ensureFilesIsMutable();
-      files_.set(index, value);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * This stores all the source code file names and can be indexed by the
-     * `file_index`.
-     * </pre>
-     *
-     * <code>repeated string files = 1;</code>
-     */
-    public Builder addFiles(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  ensureFilesIsMutable();
-      files_.add(value);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * This stores all the source code file names and can be indexed by the
-     * `file_index`.
-     * </pre>
-     *
-     * <code>repeated string files = 1;</code>
-     */
-    public Builder addAllFiles(
-        java.lang.Iterable<java.lang.String> values) {
-      ensureFilesIsMutable();
-      com.google.protobuf.AbstractMessageLite.Builder.addAll(
-          values, files_);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * This stores all the source code file names and can be indexed by the
-     * `file_index`.
-     * </pre>
-     *
-     * <code>repeated string files = 1;</code>
-     */
-    public Builder clearFiles() {
-      files_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-      bitField0_ = (bitField0_ & ~0x00000001);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * This stores all the source code file names and can be indexed by the
-     * `file_index`.
-     * </pre>
-     *
-     * <code>repeated string files = 1;</code>
-     */
-    public Builder addFilesBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      ensureFilesIsMutable();
-      files_.add(value);
-      onChanged();
-      return this;
-    }
-
-    private com.google.protobuf.MapField<
-        java.lang.String, org.tensorflow.proto.framework.GraphDebugInfo.StackTrace> traces_;
-    private com.google.protobuf.MapField<java.lang.String, org.tensorflow.proto.framework.GraphDebugInfo.StackTrace>
-    internalGetTraces() {
-      if (traces_ == null) {
-        return com.google.protobuf.MapField.emptyMapField(
-            TracesDefaultEntryHolder.defaultEntry);
-      }
-      return traces_;
-    }
-    private com.google.protobuf.MapField<java.lang.String, org.tensorflow.proto.framework.GraphDebugInfo.StackTrace>
-    internalGetMutableTraces() {
-      onChanged();;
-      if (traces_ == null) {
-        traces_ = com.google.protobuf.MapField.newMapField(
-            TracesDefaultEntryHolder.defaultEntry);
-      }
-      if (!traces_.isMutable()) {
-        traces_ = traces_.copy();
-      }
-      return traces_;
-    }
-
-    public int getTracesCount() {
-      return internalGetTraces().getMap().size();
-    }
-    /**
-     * <pre>
-     * This maps a node name to a stack trace in the source code.
-     * The map key is a mangling of the containing function and op name with
-     * syntax:
-     *   op.name '&#64;' func_name
-     * For ops in the top-level graph, the func_name is the empty string.
-     * Note that op names are restricted to a small number of characters which
-     * exclude '&#64;', making it impossible to collide keys of this form. Function
-     * names accept a much wider set of characters.
-     * It would be preferable to avoid mangling and use a tuple key of (op.name,
-     * func_name), but this is not supported with protocol buffers.
-     * </pre>
-     *
-     * <code>map&lt;string, .tensorflow.GraphDebugInfo.StackTrace&gt; traces = 2;</code>
-     */
-
-    public boolean containsTraces(
-        java.lang.String key) {
-      if (key == null) { throw new java.lang.NullPointerException(); }
-      return internalGetTraces().getMap().containsKey(key);
-    }
-    /**
-     * Use {@link #getTracesMap()} instead.
-     */
-    @java.lang.Deprecated
-    public java.util.Map<java.lang.String, org.tensorflow.proto.framework.GraphDebugInfo.StackTrace> getTraces() {
-      return getTracesMap();
-    }
-    /**
-     * <pre>
-     * This maps a node name to a stack trace in the source code.
-     * The map key is a mangling of the containing function and op name with
-     * syntax:
-     *   op.name '&#64;' func_name
-     * For ops in the top-level graph, the func_name is the empty string.
-     * Note that op names are restricted to a small number of characters which
-     * exclude '&#64;', making it impossible to collide keys of this form. Function
-     * names accept a much wider set of characters.
-     * It would be preferable to avoid mangling and use a tuple key of (op.name,
-     * func_name), but this is not supported with protocol buffers.
-     * </pre>
-     *
-     * <code>map&lt;string, .tensorflow.GraphDebugInfo.StackTrace&gt; traces = 2;</code>
-     */
-
-    public java.util.Map<java.lang.String, org.tensorflow.proto.framework.GraphDebugInfo.StackTrace> getTracesMap() {
-      return internalGetTraces().getMap();
-    }
-    /**
-     * <pre>
-     * This maps a node name to a stack trace in the source code.
-     * The map key is a mangling of the containing function and op name with
-     * syntax:
-     *   op.name '&#64;' func_name
-     * For ops in the top-level graph, the func_name is the empty string.
-     * Note that op names are restricted to a small number of characters which
-     * exclude '&#64;', making it impossible to collide keys of this form. Function
-     * names accept a much wider set of characters.
-     * It would be preferable to avoid mangling and use a tuple key of (op.name,
-     * func_name), but this is not supported with protocol buffers.
-     * </pre>
-     *
-     * <code>map&lt;string, .tensorflow.GraphDebugInfo.StackTrace&gt; traces = 2;</code>
-     */
-
-    public org.tensorflow.proto.framework.GraphDebugInfo.StackTrace getTracesOrDefault(
-        java.lang.String key,
-        org.tensorflow.proto.framework.GraphDebugInfo.StackTrace defaultValue) {
-      if (key == null) { throw new java.lang.NullPointerException(); }
-      java.util.Map<java.lang.String, org.tensorflow.proto.framework.GraphDebugInfo.StackTrace> map =
-          internalGetTraces().getMap();
-      return map.containsKey(key) ? map.get(key) : defaultValue;
-    }
-    /**
-     * <pre>
-     * This maps a node name to a stack trace in the source code.
-     * The map key is a mangling of the containing function and op name with
-     * syntax:
-     *   op.name '&#64;' func_name
-     * For ops in the top-level graph, the func_name is the empty string.
-     * Note that op names are restricted to a small number of characters which
-     * exclude '&#64;', making it impossible to collide keys of this form. Function
-     * names accept a much wider set of characters.
-     * It would be preferable to avoid mangling and use a tuple key of (op.name,
-     * func_name), but this is not supported with protocol buffers.
-     * </pre>
-     *
-     * <code>map&lt;string, .tensorflow.GraphDebugInfo.StackTrace&gt; traces = 2;</code>
-     */
-
-    public org.tensorflow.proto.framework.GraphDebugInfo.StackTrace getTracesOrThrow(
-        java.lang.String key) {
-      if (key == null) { throw new java.lang.NullPointerException(); }
-      java.util.Map<java.lang.String, org.tensorflow.proto.framework.GraphDebugInfo.StackTrace> map =
-          internalGetTraces().getMap();
-      if (!map.containsKey(key)) {
-        throw new java.lang.IllegalArgumentException();
-      }
-      return map.get(key);
-    }
-
-    public Builder clearTraces() {
-      internalGetMutableTraces().getMutableMap()
-          .clear();
-      return this;
-    }
-    /**
-     * <pre>
-     * This maps a node name to a stack trace in the source code.
-     * The map key is a mangling of the containing function and op name with
-     * syntax:
-     *   op.name '&#64;' func_name
-     * For ops in the top-level graph, the func_name is the empty string.
-     * Note that op names are restricted to a small number of characters which
-     * exclude '&#64;', making it impossible to collide keys of this form. Function
-     * names accept a much wider set of characters.
-     * It would be preferable to avoid mangling and use a tuple key of (op.name,
-     * func_name), but this is not supported with protocol buffers.
-     * </pre>
-     *
-     * <code>map&lt;string, .tensorflow.GraphDebugInfo.StackTrace&gt; traces = 2;</code>
-     */
-
-    public Builder removeTraces(
-        java.lang.String key) {
-      if (key == null) { throw new java.lang.NullPointerException(); }
-      internalGetMutableTraces().getMutableMap()
-          .remove(key);
-      return this;
-    }
-    /**
-     * Use alternate mutation accessors instead.
-     */
-    @java.lang.Deprecated
-    public java.util.Map<java.lang.String, org.tensorflow.proto.framework.GraphDebugInfo.StackTrace>
-    getMutableTraces() {
-      return internalGetMutableTraces().getMutableMap();
-    }
-    /**
-     * <pre>
-     * This maps a node name to a stack trace in the source code.
-     * The map key is a mangling of the containing function and op name with
-     * syntax:
-     *   op.name '&#64;' func_name
-     * For ops in the top-level graph, the func_name is the empty string.
-     * Note that op names are restricted to a small number of characters which
-     * exclude '&#64;', making it impossible to collide keys of this form. Function
-     * names accept a much wider set of characters.
-     * It would be preferable to avoid mangling and use a tuple key of (op.name,
-     * func_name), but this is not supported with protocol buffers.
-     * </pre>
-     *
-     * <code>map&lt;string, .tensorflow.GraphDebugInfo.StackTrace&gt; traces = 2;</code>
-     */
-    public Builder putTraces(
-        java.lang.String key,
-        org.tensorflow.proto.framework.GraphDebugInfo.StackTrace value) {
-      if (key == null) { throw new java.lang.NullPointerException(); }
-      if (value == null) { throw new java.lang.NullPointerException(); }
-      internalGetMutableTraces().getMutableMap()
-          .put(key, value);
-      return this;
-    }
-    /**
-     * <pre>
-     * This maps a node name to a stack trace in the source code.
-     * The map key is a mangling of the containing function and op name with
-     * syntax:
-     *   op.name '&#64;' func_name
-     * For ops in the top-level graph, the func_name is the empty string.
-     * Note that op names are restricted to a small number of characters which
-     * exclude '&#64;', making it impossible to collide keys of this form. Function
-     * names accept a much wider set of characters.
-     * It would be preferable to avoid mangling and use a tuple key of (op.name,
-     * func_name), but this is not supported with protocol buffers.
-     * </pre>
-     *
-     * <code>map&lt;string, .tensorflow.GraphDebugInfo.StackTrace&gt; traces = 2;</code>
-     */
-
-    public Builder putAllTraces(
-        java.util.Map<java.lang.String, org.tensorflow.proto.framework.GraphDebugInfo.StackTrace> values) {
-      internalGetMutableTraces().getMutableMap()
-          .putAll(values);
-      return this;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.GraphDebugInfo)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.GraphDebugInfo)
-  private static final org.tensorflow.proto.framework.GraphDebugInfo DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.GraphDebugInfo();
-  }
-
-  public static org.tensorflow.proto.framework.GraphDebugInfo getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<GraphDebugInfo>
-      PARSER = new com.google.protobuf.AbstractParser<GraphDebugInfo>() {
-    @java.lang.Override
-    public GraphDebugInfo parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new GraphDebugInfo(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<GraphDebugInfo> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<GraphDebugInfo> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.framework.GraphDebugInfo getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphDebugInfoOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphDebugInfoOrBuilder.java
deleted file mode 100644
index 86c48de7b70..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphDebugInfoOrBuilder.java
+++ /dev/null
@@ -1,147 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/graph_debug_info.proto
-
-package org.tensorflow.proto.framework;
-
-public interface GraphDebugInfoOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.GraphDebugInfo)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <pre>
-   * This stores all the source code file names and can be indexed by the
-   * `file_index`.
-   * </pre>
-   *
-   * <code>repeated string files = 1;</code>
-   */
-  java.util.List<java.lang.String>
-      getFilesList();
-  /**
-   * <pre>
-   * This stores all the source code file names and can be indexed by the
-   * `file_index`.
-   * </pre>
-   *
-   * <code>repeated string files = 1;</code>
-   */
-  int getFilesCount();
-  /**
-   * <pre>
-   * This stores all the source code file names and can be indexed by the
-   * `file_index`.
-   * </pre>
-   *
-   * <code>repeated string files = 1;</code>
-   */
-  java.lang.String getFiles(int index);
-  /**
-   * <pre>
-   * This stores all the source code file names and can be indexed by the
-   * `file_index`.
-   * </pre>
-   *
-   * <code>repeated string files = 1;</code>
-   */
-  com.google.protobuf.ByteString
-      getFilesBytes(int index);
-
-  /**
-   * <pre>
-   * This maps a node name to a stack trace in the source code.
-   * The map key is a mangling of the containing function and op name with
-   * syntax:
-   *   op.name '&#64;' func_name
-   * For ops in the top-level graph, the func_name is the empty string.
-   * Note that op names are restricted to a small number of characters which
-   * exclude '&#64;', making it impossible to collide keys of this form. Function
-   * names accept a much wider set of characters.
-   * It would be preferable to avoid mangling and use a tuple key of (op.name,
-   * func_name), but this is not supported with protocol buffers.
-   * </pre>
-   *
-   * <code>map&lt;string, .tensorflow.GraphDebugInfo.StackTrace&gt; traces = 2;</code>
-   */
-  int getTracesCount();
-  /**
-   * <pre>
-   * This maps a node name to a stack trace in the source code.
-   * The map key is a mangling of the containing function and op name with
-   * syntax:
-   *   op.name '&#64;' func_name
-   * For ops in the top-level graph, the func_name is the empty string.
-   * Note that op names are restricted to a small number of characters which
-   * exclude '&#64;', making it impossible to collide keys of this form. Function
-   * names accept a much wider set of characters.
-   * It would be preferable to avoid mangling and use a tuple key of (op.name,
-   * func_name), but this is not supported with protocol buffers.
-   * </pre>
-   *
-   * <code>map&lt;string, .tensorflow.GraphDebugInfo.StackTrace&gt; traces = 2;</code>
-   */
-  boolean containsTraces(
-      java.lang.String key);
-  /**
-   * Use {@link #getTracesMap()} instead.
-   */
-  @java.lang.Deprecated
-  java.util.Map<java.lang.String, org.tensorflow.proto.framework.GraphDebugInfo.StackTrace>
-  getTraces();
-  /**
-   * <pre>
-   * This maps a node name to a stack trace in the source code.
-   * The map key is a mangling of the containing function and op name with
-   * syntax:
-   *   op.name '&#64;' func_name
-   * For ops in the top-level graph, the func_name is the empty string.
-   * Note that op names are restricted to a small number of characters which
-   * exclude '&#64;', making it impossible to collide keys of this form. Function
-   * names accept a much wider set of characters.
-   * It would be preferable to avoid mangling and use a tuple key of (op.name,
-   * func_name), but this is not supported with protocol buffers.
-   * </pre>
-   *
-   * <code>map&lt;string, .tensorflow.GraphDebugInfo.StackTrace&gt; traces = 2;</code>
-   */
-  java.util.Map<java.lang.String, org.tensorflow.proto.framework.GraphDebugInfo.StackTrace>
-  getTracesMap();
-  /**
-   * <pre>
-   * This maps a node name to a stack trace in the source code.
-   * The map key is a mangling of the containing function and op name with
-   * syntax:
-   *   op.name '&#64;' func_name
-   * For ops in the top-level graph, the func_name is the empty string.
-   * Note that op names are restricted to a small number of characters which
-   * exclude '&#64;', making it impossible to collide keys of this form. Function
-   * names accept a much wider set of characters.
-   * It would be preferable to avoid mangling and use a tuple key of (op.name,
-   * func_name), but this is not supported with protocol buffers.
-   * </pre>
-   *
-   * <code>map&lt;string, .tensorflow.GraphDebugInfo.StackTrace&gt; traces = 2;</code>
-   */
-
-  org.tensorflow.proto.framework.GraphDebugInfo.StackTrace getTracesOrDefault(
-      java.lang.String key,
-      org.tensorflow.proto.framework.GraphDebugInfo.StackTrace defaultValue);
-  /**
-   * <pre>
-   * This maps a node name to a stack trace in the source code.
-   * The map key is a mangling of the containing function and op name with
-   * syntax:
-   *   op.name '&#64;' func_name
-   * For ops in the top-level graph, the func_name is the empty string.
-   * Note that op names are restricted to a small number of characters which
-   * exclude '&#64;', making it impossible to collide keys of this form. Function
-   * names accept a much wider set of characters.
-   * It would be preferable to avoid mangling and use a tuple key of (op.name,
-   * func_name), but this is not supported with protocol buffers.
-   * </pre>
-   *
-   * <code>map&lt;string, .tensorflow.GraphDebugInfo.StackTrace&gt; traces = 2;</code>
-   */
-
-  org.tensorflow.proto.framework.GraphDebugInfo.StackTrace getTracesOrThrow(
-      java.lang.String key);
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphDebugInfoProtos.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphDebugInfoProtos.java
deleted file mode 100644
index c206eb2d9cf..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphDebugInfoProtos.java
+++ /dev/null
@@ -1,92 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/graph_debug_info.proto
-
-package org.tensorflow.proto.framework;
-
-public final class GraphDebugInfoProtos {
-  private GraphDebugInfoProtos() {}
-  public static void registerAllExtensions(
-      com.google.protobuf.ExtensionRegistryLite registry) {
-  }
-
-  public static void registerAllExtensions(
-      com.google.protobuf.ExtensionRegistry registry) {
-    registerAllExtensions(
-        (com.google.protobuf.ExtensionRegistryLite) registry);
-  }
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_GraphDebugInfo_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_GraphDebugInfo_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_GraphDebugInfo_FileLineCol_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_GraphDebugInfo_FileLineCol_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_GraphDebugInfo_StackTrace_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_GraphDebugInfo_StackTrace_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_GraphDebugInfo_TracesEntry_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_GraphDebugInfo_TracesEntry_fieldAccessorTable;
-
-  public static com.google.protobuf.Descriptors.FileDescriptor
-      getDescriptor() {
-    return descriptor;
-  }
-  private static  com.google.protobuf.Descriptors.FileDescriptor
-      descriptor;
-  static {
-    java.lang.String[] descriptorData = {
-      "\n/tensorflow/core/protobuf/graph_debug_i" +
-      "nfo.proto\022\ntensorflow\"\325\002\n\016GraphDebugInfo" +
-      "\022\r\n\005files\030\001 \003(\t\0226\n\006traces\030\002 \003(\0132&.tensor" +
-      "flow.GraphDebugInfo.TracesEntry\032X\n\013FileL" +
-      "ineCol\022\022\n\nfile_index\030\001 \001(\005\022\014\n\004line\030\002 \001(\005" +
-      "\022\013\n\003col\030\003 \001(\005\022\014\n\004func\030\004 \001(\t\022\014\n\004code\030\005 \001(" +
-      "\t\032L\n\nStackTrace\022>\n\016file_line_cols\030\001 \003(\0132" +
-      "&.tensorflow.GraphDebugInfo.FileLineCol\032" +
-      "T\n\013TracesEntry\022\013\n\003key\030\001 \001(\t\0224\n\005value\030\002 \001" +
-      "(\0132%.tensorflow.GraphDebugInfo.StackTrac" +
-      "e:\0028\001B\205\001\n\036org.tensorflow.proto.framework" +
-      "B\024GraphDebugInfoProtosP\001ZHgithub.com/ten" +
-      "sorflow/tensorflow/tensorflow/go/core/co" +
-      "re_protos_go_proto\370\001\001b\006proto3"
-    };
-    descriptor = com.google.protobuf.Descriptors.FileDescriptor
-      .internalBuildGeneratedFileFrom(descriptorData,
-        new com.google.protobuf.Descriptors.FileDescriptor[] {
-        });
-    internal_static_tensorflow_GraphDebugInfo_descriptor =
-      getDescriptor().getMessageTypes().get(0);
-    internal_static_tensorflow_GraphDebugInfo_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_GraphDebugInfo_descriptor,
-        new java.lang.String[] { "Files", "Traces", });
-    internal_static_tensorflow_GraphDebugInfo_FileLineCol_descriptor =
-      internal_static_tensorflow_GraphDebugInfo_descriptor.getNestedTypes().get(0);
-    internal_static_tensorflow_GraphDebugInfo_FileLineCol_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_GraphDebugInfo_FileLineCol_descriptor,
-        new java.lang.String[] { "FileIndex", "Line", "Col", "Func", "Code", });
-    internal_static_tensorflow_GraphDebugInfo_StackTrace_descriptor =
-      internal_static_tensorflow_GraphDebugInfo_descriptor.getNestedTypes().get(1);
-    internal_static_tensorflow_GraphDebugInfo_StackTrace_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_GraphDebugInfo_StackTrace_descriptor,
-        new java.lang.String[] { "FileLineCols", });
-    internal_static_tensorflow_GraphDebugInfo_TracesEntry_descriptor =
-      internal_static_tensorflow_GraphDebugInfo_descriptor.getNestedTypes().get(2);
-    internal_static_tensorflow_GraphDebugInfo_TracesEntry_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_GraphDebugInfo_TracesEntry_descriptor,
-        new java.lang.String[] { "Key", "Value", });
-  }
-
-  // @@protoc_insertion_point(outer_class_scope)
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphDef.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphDef.java
deleted file mode 100644
index 18e75b5e768..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphDef.java
+++ /dev/null
@@ -1,1588 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/framework/graph.proto
-
-package org.tensorflow.proto.framework;
-
-/**
- * <pre>
- * Represents the graph of operations
- * </pre>
- *
- * Protobuf type {@code tensorflow.GraphDef}
- */
-public  final class GraphDef extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.GraphDef)
-    GraphDefOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use GraphDef.newBuilder() to construct.
-  private GraphDef(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private GraphDef() {
-    node_ = java.util.Collections.emptyList();
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new GraphDef();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private GraphDef(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    int mutable_bitField0_ = 0;
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 10: {
-            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
-              node_ = new java.util.ArrayList<org.tensorflow.proto.framework.NodeDef>();
-              mutable_bitField0_ |= 0x00000001;
-            }
-            node_.add(
-                input.readMessage(org.tensorflow.proto.framework.NodeDef.parser(), extensionRegistry));
-            break;
-          }
-          case 18: {
-            org.tensorflow.proto.framework.FunctionDefLibrary.Builder subBuilder = null;
-            if (library_ != null) {
-              subBuilder = library_.toBuilder();
-            }
-            library_ = input.readMessage(org.tensorflow.proto.framework.FunctionDefLibrary.parser(), extensionRegistry);
-            if (subBuilder != null) {
-              subBuilder.mergeFrom(library_);
-              library_ = subBuilder.buildPartial();
-            }
-
-            break;
-          }
-          case 24: {
-
-            version_ = input.readInt32();
-            break;
-          }
-          case 34: {
-            org.tensorflow.proto.framework.VersionDef.Builder subBuilder = null;
-            if (versions_ != null) {
-              subBuilder = versions_.toBuilder();
-            }
-            versions_ = input.readMessage(org.tensorflow.proto.framework.VersionDef.parser(), extensionRegistry);
-            if (subBuilder != null) {
-              subBuilder.mergeFrom(versions_);
-              versions_ = subBuilder.buildPartial();
-            }
-
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      if (((mutable_bitField0_ & 0x00000001) != 0)) {
-        node_ = java.util.Collections.unmodifiableList(node_);
-      }
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.framework.GraphProtos.internal_static_tensorflow_GraphDef_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.framework.GraphProtos.internal_static_tensorflow_GraphDef_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.framework.GraphDef.class, org.tensorflow.proto.framework.GraphDef.Builder.class);
-  }
-
-  public static final int NODE_FIELD_NUMBER = 1;
-  private java.util.List<org.tensorflow.proto.framework.NodeDef> node_;
-  /**
-   * <code>repeated .tensorflow.NodeDef node = 1;</code>
-   */
-  public java.util.List<org.tensorflow.proto.framework.NodeDef> getNodeList() {
-    return node_;
-  }
-  /**
-   * <code>repeated .tensorflow.NodeDef node = 1;</code>
-   */
-  public java.util.List<? extends org.tensorflow.proto.framework.NodeDefOrBuilder> 
-      getNodeOrBuilderList() {
-    return node_;
-  }
-  /**
-   * <code>repeated .tensorflow.NodeDef node = 1;</code>
-   */
-  public int getNodeCount() {
-    return node_.size();
-  }
-  /**
-   * <code>repeated .tensorflow.NodeDef node = 1;</code>
-   */
-  public org.tensorflow.proto.framework.NodeDef getNode(int index) {
-    return node_.get(index);
-  }
-  /**
-   * <code>repeated .tensorflow.NodeDef node = 1;</code>
-   */
-  public org.tensorflow.proto.framework.NodeDefOrBuilder getNodeOrBuilder(
-      int index) {
-    return node_.get(index);
-  }
-
-  public static final int VERSIONS_FIELD_NUMBER = 4;
-  private org.tensorflow.proto.framework.VersionDef versions_;
-  /**
-   * <pre>
-   * Compatibility versions of the graph.  See core/public/version.h for version
-   * history.  The GraphDef version is distinct from the TensorFlow version, and
-   * each release of TensorFlow will support a range of GraphDef versions.
-   * </pre>
-   *
-   * <code>.tensorflow.VersionDef versions = 4;</code>
-   */
-  public boolean hasVersions() {
-    return versions_ != null;
-  }
-  /**
-   * <pre>
-   * Compatibility versions of the graph.  See core/public/version.h for version
-   * history.  The GraphDef version is distinct from the TensorFlow version, and
-   * each release of TensorFlow will support a range of GraphDef versions.
-   * </pre>
-   *
-   * <code>.tensorflow.VersionDef versions = 4;</code>
-   */
-  public org.tensorflow.proto.framework.VersionDef getVersions() {
-    return versions_ == null ? org.tensorflow.proto.framework.VersionDef.getDefaultInstance() : versions_;
-  }
-  /**
-   * <pre>
-   * Compatibility versions of the graph.  See core/public/version.h for version
-   * history.  The GraphDef version is distinct from the TensorFlow version, and
-   * each release of TensorFlow will support a range of GraphDef versions.
-   * </pre>
-   *
-   * <code>.tensorflow.VersionDef versions = 4;</code>
-   */
-  public org.tensorflow.proto.framework.VersionDefOrBuilder getVersionsOrBuilder() {
-    return getVersions();
-  }
-
-  public static final int VERSION_FIELD_NUMBER = 3;
-  private int version_;
-  /**
-   * <pre>
-   * Deprecated single version field; use versions above instead.  Since all
-   * GraphDef changes before "versions" was introduced were forward
-   * compatible, this field is entirely ignored.
-   * </pre>
-   *
-   * <code>int32 version = 3 [deprecated = true];</code>
-   */
-  @java.lang.Deprecated public int getVersion() {
-    return version_;
-  }
-
-  public static final int LIBRARY_FIELD_NUMBER = 2;
-  private org.tensorflow.proto.framework.FunctionDefLibrary library_;
-  /**
-   * <pre>
-   * EXPERIMENTAL. DO NOT USE OR DEPEND ON THIS YET.
-   * "library" provides user-defined functions.
-   * Naming:
-   *   * library.function.name are in a flat namespace.
-   *     NOTE: We may need to change it to be hierarchical to support
-   *     different orgs. E.g.,
-   *     { "/google/nn", { ... }},
-   *     { "/google/vision", { ... }}
-   *     { "/org_foo/module_bar", { ... }}
-   *     map&lt;string, FunctionDefLib&gt; named_lib;
-   *   * If node[i].op is the name of one function in "library",
-   *     node[i] is deemed as a function call. Otherwise, node[i].op
-   *     must be a primitive operation supported by the runtime.
-   * Function call semantics:
-   *   * The callee may start execution as soon as some of its inputs
-   *     are ready. The caller may want to use Tuple() mechanism to
-   *     ensure all inputs are ready in the same time.
-   *   * The consumer of return values may start executing as soon as
-   *     the return values the consumer depends on are ready.  The
-   *     consumer may want to use Tuple() mechanism to ensure the
-   *     consumer does not start until all return values of the callee
-   *     function are ready.
-   * </pre>
-   *
-   * <code>.tensorflow.FunctionDefLibrary library = 2;</code>
-   */
-  public boolean hasLibrary() {
-    return library_ != null;
-  }
-  /**
-   * <pre>
-   * EXPERIMENTAL. DO NOT USE OR DEPEND ON THIS YET.
-   * "library" provides user-defined functions.
-   * Naming:
-   *   * library.function.name are in a flat namespace.
-   *     NOTE: We may need to change it to be hierarchical to support
-   *     different orgs. E.g.,
-   *     { "/google/nn", { ... }},
-   *     { "/google/vision", { ... }}
-   *     { "/org_foo/module_bar", { ... }}
-   *     map&lt;string, FunctionDefLib&gt; named_lib;
-   *   * If node[i].op is the name of one function in "library",
-   *     node[i] is deemed as a function call. Otherwise, node[i].op
-   *     must be a primitive operation supported by the runtime.
-   * Function call semantics:
-   *   * The callee may start execution as soon as some of its inputs
-   *     are ready. The caller may want to use Tuple() mechanism to
-   *     ensure all inputs are ready in the same time.
-   *   * The consumer of return values may start executing as soon as
-   *     the return values the consumer depends on are ready.  The
-   *     consumer may want to use Tuple() mechanism to ensure the
-   *     consumer does not start until all return values of the callee
-   *     function are ready.
-   * </pre>
-   *
-   * <code>.tensorflow.FunctionDefLibrary library = 2;</code>
-   */
-  public org.tensorflow.proto.framework.FunctionDefLibrary getLibrary() {
-    return library_ == null ? org.tensorflow.proto.framework.FunctionDefLibrary.getDefaultInstance() : library_;
-  }
-  /**
-   * <pre>
-   * EXPERIMENTAL. DO NOT USE OR DEPEND ON THIS YET.
-   * "library" provides user-defined functions.
-   * Naming:
-   *   * library.function.name are in a flat namespace.
-   *     NOTE: We may need to change it to be hierarchical to support
-   *     different orgs. E.g.,
-   *     { "/google/nn", { ... }},
-   *     { "/google/vision", { ... }}
-   *     { "/org_foo/module_bar", { ... }}
-   *     map&lt;string, FunctionDefLib&gt; named_lib;
-   *   * If node[i].op is the name of one function in "library",
-   *     node[i] is deemed as a function call. Otherwise, node[i].op
-   *     must be a primitive operation supported by the runtime.
-   * Function call semantics:
-   *   * The callee may start execution as soon as some of its inputs
-   *     are ready. The caller may want to use Tuple() mechanism to
-   *     ensure all inputs are ready in the same time.
-   *   * The consumer of return values may start executing as soon as
-   *     the return values the consumer depends on are ready.  The
-   *     consumer may want to use Tuple() mechanism to ensure the
-   *     consumer does not start until all return values of the callee
-   *     function are ready.
-   * </pre>
-   *
-   * <code>.tensorflow.FunctionDefLibrary library = 2;</code>
-   */
-  public org.tensorflow.proto.framework.FunctionDefLibraryOrBuilder getLibraryOrBuilder() {
-    return getLibrary();
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    for (int i = 0; i < node_.size(); i++) {
-      output.writeMessage(1, node_.get(i));
-    }
-    if (library_ != null) {
-      output.writeMessage(2, getLibrary());
-    }
-    if (version_ != 0) {
-      output.writeInt32(3, version_);
-    }
-    if (versions_ != null) {
-      output.writeMessage(4, getVersions());
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    for (int i = 0; i < node_.size(); i++) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(1, node_.get(i));
-    }
-    if (library_ != null) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(2, getLibrary());
-    }
-    if (version_ != 0) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt32Size(3, version_);
-    }
-    if (versions_ != null) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(4, getVersions());
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.framework.GraphDef)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.framework.GraphDef other = (org.tensorflow.proto.framework.GraphDef) obj;
-
-    if (!getNodeList()
-        .equals(other.getNodeList())) return false;
-    if (hasVersions() != other.hasVersions()) return false;
-    if (hasVersions()) {
-      if (!getVersions()
-          .equals(other.getVersions())) return false;
-    }
-    if (getVersion()
-        != other.getVersion()) return false;
-    if (hasLibrary() != other.hasLibrary()) return false;
-    if (hasLibrary()) {
-      if (!getLibrary()
-          .equals(other.getLibrary())) return false;
-    }
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    if (getNodeCount() > 0) {
-      hash = (37 * hash) + NODE_FIELD_NUMBER;
-      hash = (53 * hash) + getNodeList().hashCode();
-    }
-    if (hasVersions()) {
-      hash = (37 * hash) + VERSIONS_FIELD_NUMBER;
-      hash = (53 * hash) + getVersions().hashCode();
-    }
-    hash = (37 * hash) + VERSION_FIELD_NUMBER;
-    hash = (53 * hash) + getVersion();
-    if (hasLibrary()) {
-      hash = (37 * hash) + LIBRARY_FIELD_NUMBER;
-      hash = (53 * hash) + getLibrary().hashCode();
-    }
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.framework.GraphDef parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.GraphDef parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.GraphDef parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.GraphDef parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.GraphDef parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.GraphDef parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.GraphDef parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.GraphDef parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.GraphDef parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.GraphDef parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.GraphDef parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.GraphDef parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.framework.GraphDef prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * <pre>
-   * Represents the graph of operations
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.GraphDef}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.GraphDef)
-      org.tensorflow.proto.framework.GraphDefOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.GraphProtos.internal_static_tensorflow_GraphDef_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.GraphProtos.internal_static_tensorflow_GraphDef_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.GraphDef.class, org.tensorflow.proto.framework.GraphDef.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.framework.GraphDef.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-        getNodeFieldBuilder();
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      if (nodeBuilder_ == null) {
-        node_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000001);
-      } else {
-        nodeBuilder_.clear();
-      }
-      if (versionsBuilder_ == null) {
-        versions_ = null;
-      } else {
-        versions_ = null;
-        versionsBuilder_ = null;
-      }
-      version_ = 0;
-
-      if (libraryBuilder_ == null) {
-        library_ = null;
-      } else {
-        library_ = null;
-        libraryBuilder_ = null;
-      }
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.framework.GraphProtos.internal_static_tensorflow_GraphDef_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.GraphDef getDefaultInstanceForType() {
-      return org.tensorflow.proto.framework.GraphDef.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.GraphDef build() {
-      org.tensorflow.proto.framework.GraphDef result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.GraphDef buildPartial() {
-      org.tensorflow.proto.framework.GraphDef result = new org.tensorflow.proto.framework.GraphDef(this);
-      int from_bitField0_ = bitField0_;
-      if (nodeBuilder_ == null) {
-        if (((bitField0_ & 0x00000001) != 0)) {
-          node_ = java.util.Collections.unmodifiableList(node_);
-          bitField0_ = (bitField0_ & ~0x00000001);
-        }
-        result.node_ = node_;
-      } else {
-        result.node_ = nodeBuilder_.build();
-      }
-      if (versionsBuilder_ == null) {
-        result.versions_ = versions_;
-      } else {
-        result.versions_ = versionsBuilder_.build();
-      }
-      result.version_ = version_;
-      if (libraryBuilder_ == null) {
-        result.library_ = library_;
-      } else {
-        result.library_ = libraryBuilder_.build();
-      }
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.framework.GraphDef) {
-        return mergeFrom((org.tensorflow.proto.framework.GraphDef)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.framework.GraphDef other) {
-      if (other == org.tensorflow.proto.framework.GraphDef.getDefaultInstance()) return this;
-      if (nodeBuilder_ == null) {
-        if (!other.node_.isEmpty()) {
-          if (node_.isEmpty()) {
-            node_ = other.node_;
-            bitField0_ = (bitField0_ & ~0x00000001);
-          } else {
-            ensureNodeIsMutable();
-            node_.addAll(other.node_);
-          }
-          onChanged();
-        }
-      } else {
-        if (!other.node_.isEmpty()) {
-          if (nodeBuilder_.isEmpty()) {
-            nodeBuilder_.dispose();
-            nodeBuilder_ = null;
-            node_ = other.node_;
-            bitField0_ = (bitField0_ & ~0x00000001);
-            nodeBuilder_ = 
-              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
-                 getNodeFieldBuilder() : null;
-          } else {
-            nodeBuilder_.addAllMessages(other.node_);
-          }
-        }
-      }
-      if (other.hasVersions()) {
-        mergeVersions(other.getVersions());
-      }
-      if (other.getVersion() != 0) {
-        setVersion(other.getVersion());
-      }
-      if (other.hasLibrary()) {
-        mergeLibrary(other.getLibrary());
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.framework.GraphDef parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.framework.GraphDef) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-    private int bitField0_;
-
-    private java.util.List<org.tensorflow.proto.framework.NodeDef> node_ =
-      java.util.Collections.emptyList();
-    private void ensureNodeIsMutable() {
-      if (!((bitField0_ & 0x00000001) != 0)) {
-        node_ = new java.util.ArrayList<org.tensorflow.proto.framework.NodeDef>(node_);
-        bitField0_ |= 0x00000001;
-       }
-    }
-
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.framework.NodeDef, org.tensorflow.proto.framework.NodeDef.Builder, org.tensorflow.proto.framework.NodeDefOrBuilder> nodeBuilder_;
-
-    /**
-     * <code>repeated .tensorflow.NodeDef node = 1;</code>
-     */
-    public java.util.List<org.tensorflow.proto.framework.NodeDef> getNodeList() {
-      if (nodeBuilder_ == null) {
-        return java.util.Collections.unmodifiableList(node_);
-      } else {
-        return nodeBuilder_.getMessageList();
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.NodeDef node = 1;</code>
-     */
-    public int getNodeCount() {
-      if (nodeBuilder_ == null) {
-        return node_.size();
-      } else {
-        return nodeBuilder_.getCount();
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.NodeDef node = 1;</code>
-     */
-    public org.tensorflow.proto.framework.NodeDef getNode(int index) {
-      if (nodeBuilder_ == null) {
-        return node_.get(index);
-      } else {
-        return nodeBuilder_.getMessage(index);
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.NodeDef node = 1;</code>
-     */
-    public Builder setNode(
-        int index, org.tensorflow.proto.framework.NodeDef value) {
-      if (nodeBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureNodeIsMutable();
-        node_.set(index, value);
-        onChanged();
-      } else {
-        nodeBuilder_.setMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.NodeDef node = 1;</code>
-     */
-    public Builder setNode(
-        int index, org.tensorflow.proto.framework.NodeDef.Builder builderForValue) {
-      if (nodeBuilder_ == null) {
-        ensureNodeIsMutable();
-        node_.set(index, builderForValue.build());
-        onChanged();
-      } else {
-        nodeBuilder_.setMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.NodeDef node = 1;</code>
-     */
-    public Builder addNode(org.tensorflow.proto.framework.NodeDef value) {
-      if (nodeBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureNodeIsMutable();
-        node_.add(value);
-        onChanged();
-      } else {
-        nodeBuilder_.addMessage(value);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.NodeDef node = 1;</code>
-     */
-    public Builder addNode(
-        int index, org.tensorflow.proto.framework.NodeDef value) {
-      if (nodeBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureNodeIsMutable();
-        node_.add(index, value);
-        onChanged();
-      } else {
-        nodeBuilder_.addMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.NodeDef node = 1;</code>
-     */
-    public Builder addNode(
-        org.tensorflow.proto.framework.NodeDef.Builder builderForValue) {
-      if (nodeBuilder_ == null) {
-        ensureNodeIsMutable();
-        node_.add(builderForValue.build());
-        onChanged();
-      } else {
-        nodeBuilder_.addMessage(builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.NodeDef node = 1;</code>
-     */
-    public Builder addNode(
-        int index, org.tensorflow.proto.framework.NodeDef.Builder builderForValue) {
-      if (nodeBuilder_ == null) {
-        ensureNodeIsMutable();
-        node_.add(index, builderForValue.build());
-        onChanged();
-      } else {
-        nodeBuilder_.addMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.NodeDef node = 1;</code>
-     */
-    public Builder addAllNode(
-        java.lang.Iterable<? extends org.tensorflow.proto.framework.NodeDef> values) {
-      if (nodeBuilder_ == null) {
-        ensureNodeIsMutable();
-        com.google.protobuf.AbstractMessageLite.Builder.addAll(
-            values, node_);
-        onChanged();
-      } else {
-        nodeBuilder_.addAllMessages(values);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.NodeDef node = 1;</code>
-     */
-    public Builder clearNode() {
-      if (nodeBuilder_ == null) {
-        node_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000001);
-        onChanged();
-      } else {
-        nodeBuilder_.clear();
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.NodeDef node = 1;</code>
-     */
-    public Builder removeNode(int index) {
-      if (nodeBuilder_ == null) {
-        ensureNodeIsMutable();
-        node_.remove(index);
-        onChanged();
-      } else {
-        nodeBuilder_.remove(index);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.NodeDef node = 1;</code>
-     */
-    public org.tensorflow.proto.framework.NodeDef.Builder getNodeBuilder(
-        int index) {
-      return getNodeFieldBuilder().getBuilder(index);
-    }
-    /**
-     * <code>repeated .tensorflow.NodeDef node = 1;</code>
-     */
-    public org.tensorflow.proto.framework.NodeDefOrBuilder getNodeOrBuilder(
-        int index) {
-      if (nodeBuilder_ == null) {
-        return node_.get(index);  } else {
-        return nodeBuilder_.getMessageOrBuilder(index);
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.NodeDef node = 1;</code>
-     */
-    public java.util.List<? extends org.tensorflow.proto.framework.NodeDefOrBuilder> 
-         getNodeOrBuilderList() {
-      if (nodeBuilder_ != null) {
-        return nodeBuilder_.getMessageOrBuilderList();
-      } else {
-        return java.util.Collections.unmodifiableList(node_);
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.NodeDef node = 1;</code>
-     */
-    public org.tensorflow.proto.framework.NodeDef.Builder addNodeBuilder() {
-      return getNodeFieldBuilder().addBuilder(
-          org.tensorflow.proto.framework.NodeDef.getDefaultInstance());
-    }
-    /**
-     * <code>repeated .tensorflow.NodeDef node = 1;</code>
-     */
-    public org.tensorflow.proto.framework.NodeDef.Builder addNodeBuilder(
-        int index) {
-      return getNodeFieldBuilder().addBuilder(
-          index, org.tensorflow.proto.framework.NodeDef.getDefaultInstance());
-    }
-    /**
-     * <code>repeated .tensorflow.NodeDef node = 1;</code>
-     */
-    public java.util.List<org.tensorflow.proto.framework.NodeDef.Builder> 
-         getNodeBuilderList() {
-      return getNodeFieldBuilder().getBuilderList();
-    }
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.framework.NodeDef, org.tensorflow.proto.framework.NodeDef.Builder, org.tensorflow.proto.framework.NodeDefOrBuilder> 
-        getNodeFieldBuilder() {
-      if (nodeBuilder_ == null) {
-        nodeBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
-            org.tensorflow.proto.framework.NodeDef, org.tensorflow.proto.framework.NodeDef.Builder, org.tensorflow.proto.framework.NodeDefOrBuilder>(
-                node_,
-                ((bitField0_ & 0x00000001) != 0),
-                getParentForChildren(),
-                isClean());
-        node_ = null;
-      }
-      return nodeBuilder_;
-    }
-
-    private org.tensorflow.proto.framework.VersionDef versions_;
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.VersionDef, org.tensorflow.proto.framework.VersionDef.Builder, org.tensorflow.proto.framework.VersionDefOrBuilder> versionsBuilder_;
-    /**
-     * <pre>
-     * Compatibility versions of the graph.  See core/public/version.h for version
-     * history.  The GraphDef version is distinct from the TensorFlow version, and
-     * each release of TensorFlow will support a range of GraphDef versions.
-     * </pre>
-     *
-     * <code>.tensorflow.VersionDef versions = 4;</code>
-     */
-    public boolean hasVersions() {
-      return versionsBuilder_ != null || versions_ != null;
-    }
-    /**
-     * <pre>
-     * Compatibility versions of the graph.  See core/public/version.h for version
-     * history.  The GraphDef version is distinct from the TensorFlow version, and
-     * each release of TensorFlow will support a range of GraphDef versions.
-     * </pre>
-     *
-     * <code>.tensorflow.VersionDef versions = 4;</code>
-     */
-    public org.tensorflow.proto.framework.VersionDef getVersions() {
-      if (versionsBuilder_ == null) {
-        return versions_ == null ? org.tensorflow.proto.framework.VersionDef.getDefaultInstance() : versions_;
-      } else {
-        return versionsBuilder_.getMessage();
-      }
-    }
-    /**
-     * <pre>
-     * Compatibility versions of the graph.  See core/public/version.h for version
-     * history.  The GraphDef version is distinct from the TensorFlow version, and
-     * each release of TensorFlow will support a range of GraphDef versions.
-     * </pre>
-     *
-     * <code>.tensorflow.VersionDef versions = 4;</code>
-     */
-    public Builder setVersions(org.tensorflow.proto.framework.VersionDef value) {
-      if (versionsBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        versions_ = value;
-        onChanged();
-      } else {
-        versionsBuilder_.setMessage(value);
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Compatibility versions of the graph.  See core/public/version.h for version
-     * history.  The GraphDef version is distinct from the TensorFlow version, and
-     * each release of TensorFlow will support a range of GraphDef versions.
-     * </pre>
-     *
-     * <code>.tensorflow.VersionDef versions = 4;</code>
-     */
-    public Builder setVersions(
-        org.tensorflow.proto.framework.VersionDef.Builder builderForValue) {
-      if (versionsBuilder_ == null) {
-        versions_ = builderForValue.build();
-        onChanged();
-      } else {
-        versionsBuilder_.setMessage(builderForValue.build());
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Compatibility versions of the graph.  See core/public/version.h for version
-     * history.  The GraphDef version is distinct from the TensorFlow version, and
-     * each release of TensorFlow will support a range of GraphDef versions.
-     * </pre>
-     *
-     * <code>.tensorflow.VersionDef versions = 4;</code>
-     */
-    public Builder mergeVersions(org.tensorflow.proto.framework.VersionDef value) {
-      if (versionsBuilder_ == null) {
-        if (versions_ != null) {
-          versions_ =
-            org.tensorflow.proto.framework.VersionDef.newBuilder(versions_).mergeFrom(value).buildPartial();
-        } else {
-          versions_ = value;
-        }
-        onChanged();
-      } else {
-        versionsBuilder_.mergeFrom(value);
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Compatibility versions of the graph.  See core/public/version.h for version
-     * history.  The GraphDef version is distinct from the TensorFlow version, and
-     * each release of TensorFlow will support a range of GraphDef versions.
-     * </pre>
-     *
-     * <code>.tensorflow.VersionDef versions = 4;</code>
-     */
-    public Builder clearVersions() {
-      if (versionsBuilder_ == null) {
-        versions_ = null;
-        onChanged();
-      } else {
-        versions_ = null;
-        versionsBuilder_ = null;
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Compatibility versions of the graph.  See core/public/version.h for version
-     * history.  The GraphDef version is distinct from the TensorFlow version, and
-     * each release of TensorFlow will support a range of GraphDef versions.
-     * </pre>
-     *
-     * <code>.tensorflow.VersionDef versions = 4;</code>
-     */
-    public org.tensorflow.proto.framework.VersionDef.Builder getVersionsBuilder() {
-      
-      onChanged();
-      return getVersionsFieldBuilder().getBuilder();
-    }
-    /**
-     * <pre>
-     * Compatibility versions of the graph.  See core/public/version.h for version
-     * history.  The GraphDef version is distinct from the TensorFlow version, and
-     * each release of TensorFlow will support a range of GraphDef versions.
-     * </pre>
-     *
-     * <code>.tensorflow.VersionDef versions = 4;</code>
-     */
-    public org.tensorflow.proto.framework.VersionDefOrBuilder getVersionsOrBuilder() {
-      if (versionsBuilder_ != null) {
-        return versionsBuilder_.getMessageOrBuilder();
-      } else {
-        return versions_ == null ?
-            org.tensorflow.proto.framework.VersionDef.getDefaultInstance() : versions_;
-      }
-    }
-    /**
-     * <pre>
-     * Compatibility versions of the graph.  See core/public/version.h for version
-     * history.  The GraphDef version is distinct from the TensorFlow version, and
-     * each release of TensorFlow will support a range of GraphDef versions.
-     * </pre>
-     *
-     * <code>.tensorflow.VersionDef versions = 4;</code>
-     */
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.VersionDef, org.tensorflow.proto.framework.VersionDef.Builder, org.tensorflow.proto.framework.VersionDefOrBuilder> 
-        getVersionsFieldBuilder() {
-      if (versionsBuilder_ == null) {
-        versionsBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-            org.tensorflow.proto.framework.VersionDef, org.tensorflow.proto.framework.VersionDef.Builder, org.tensorflow.proto.framework.VersionDefOrBuilder>(
-                getVersions(),
-                getParentForChildren(),
-                isClean());
-        versions_ = null;
-      }
-      return versionsBuilder_;
-    }
-
-    private int version_ ;
-    /**
-     * <pre>
-     * Deprecated single version field; use versions above instead.  Since all
-     * GraphDef changes before "versions" was introduced were forward
-     * compatible, this field is entirely ignored.
-     * </pre>
-     *
-     * <code>int32 version = 3 [deprecated = true];</code>
-     */
-    @java.lang.Deprecated public int getVersion() {
-      return version_;
-    }
-    /**
-     * <pre>
-     * Deprecated single version field; use versions above instead.  Since all
-     * GraphDef changes before "versions" was introduced were forward
-     * compatible, this field is entirely ignored.
-     * </pre>
-     *
-     * <code>int32 version = 3 [deprecated = true];</code>
-     */
-    @java.lang.Deprecated public Builder setVersion(int value) {
-      
-      version_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Deprecated single version field; use versions above instead.  Since all
-     * GraphDef changes before "versions" was introduced were forward
-     * compatible, this field is entirely ignored.
-     * </pre>
-     *
-     * <code>int32 version = 3 [deprecated = true];</code>
-     */
-    @java.lang.Deprecated public Builder clearVersion() {
-      
-      version_ = 0;
-      onChanged();
-      return this;
-    }
-
-    private org.tensorflow.proto.framework.FunctionDefLibrary library_;
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.FunctionDefLibrary, org.tensorflow.proto.framework.FunctionDefLibrary.Builder, org.tensorflow.proto.framework.FunctionDefLibraryOrBuilder> libraryBuilder_;
-    /**
-     * <pre>
-     * EXPERIMENTAL. DO NOT USE OR DEPEND ON THIS YET.
-     * "library" provides user-defined functions.
-     * Naming:
-     *   * library.function.name are in a flat namespace.
-     *     NOTE: We may need to change it to be hierarchical to support
-     *     different orgs. E.g.,
-     *     { "/google/nn", { ... }},
-     *     { "/google/vision", { ... }}
-     *     { "/org_foo/module_bar", { ... }}
-     *     map&lt;string, FunctionDefLib&gt; named_lib;
-     *   * If node[i].op is the name of one function in "library",
-     *     node[i] is deemed as a function call. Otherwise, node[i].op
-     *     must be a primitive operation supported by the runtime.
-     * Function call semantics:
-     *   * The callee may start execution as soon as some of its inputs
-     *     are ready. The caller may want to use Tuple() mechanism to
-     *     ensure all inputs are ready in the same time.
-     *   * The consumer of return values may start executing as soon as
-     *     the return values the consumer depends on are ready.  The
-     *     consumer may want to use Tuple() mechanism to ensure the
-     *     consumer does not start until all return values of the callee
-     *     function are ready.
-     * </pre>
-     *
-     * <code>.tensorflow.FunctionDefLibrary library = 2;</code>
-     */
-    public boolean hasLibrary() {
-      return libraryBuilder_ != null || library_ != null;
-    }
-    /**
-     * <pre>
-     * EXPERIMENTAL. DO NOT USE OR DEPEND ON THIS YET.
-     * "library" provides user-defined functions.
-     * Naming:
-     *   * library.function.name are in a flat namespace.
-     *     NOTE: We may need to change it to be hierarchical to support
-     *     different orgs. E.g.,
-     *     { "/google/nn", { ... }},
-     *     { "/google/vision", { ... }}
-     *     { "/org_foo/module_bar", { ... }}
-     *     map&lt;string, FunctionDefLib&gt; named_lib;
-     *   * If node[i].op is the name of one function in "library",
-     *     node[i] is deemed as a function call. Otherwise, node[i].op
-     *     must be a primitive operation supported by the runtime.
-     * Function call semantics:
-     *   * The callee may start execution as soon as some of its inputs
-     *     are ready. The caller may want to use Tuple() mechanism to
-     *     ensure all inputs are ready in the same time.
-     *   * The consumer of return values may start executing as soon as
-     *     the return values the consumer depends on are ready.  The
-     *     consumer may want to use Tuple() mechanism to ensure the
-     *     consumer does not start until all return values of the callee
-     *     function are ready.
-     * </pre>
-     *
-     * <code>.tensorflow.FunctionDefLibrary library = 2;</code>
-     */
-    public org.tensorflow.proto.framework.FunctionDefLibrary getLibrary() {
-      if (libraryBuilder_ == null) {
-        return library_ == null ? org.tensorflow.proto.framework.FunctionDefLibrary.getDefaultInstance() : library_;
-      } else {
-        return libraryBuilder_.getMessage();
-      }
-    }
-    /**
-     * <pre>
-     * EXPERIMENTAL. DO NOT USE OR DEPEND ON THIS YET.
-     * "library" provides user-defined functions.
-     * Naming:
-     *   * library.function.name are in a flat namespace.
-     *     NOTE: We may need to change it to be hierarchical to support
-     *     different orgs. E.g.,
-     *     { "/google/nn", { ... }},
-     *     { "/google/vision", { ... }}
-     *     { "/org_foo/module_bar", { ... }}
-     *     map&lt;string, FunctionDefLib&gt; named_lib;
-     *   * If node[i].op is the name of one function in "library",
-     *     node[i] is deemed as a function call. Otherwise, node[i].op
-     *     must be a primitive operation supported by the runtime.
-     * Function call semantics:
-     *   * The callee may start execution as soon as some of its inputs
-     *     are ready. The caller may want to use Tuple() mechanism to
-     *     ensure all inputs are ready in the same time.
-     *   * The consumer of return values may start executing as soon as
-     *     the return values the consumer depends on are ready.  The
-     *     consumer may want to use Tuple() mechanism to ensure the
-     *     consumer does not start until all return values of the callee
-     *     function are ready.
-     * </pre>
-     *
-     * <code>.tensorflow.FunctionDefLibrary library = 2;</code>
-     */
-    public Builder setLibrary(org.tensorflow.proto.framework.FunctionDefLibrary value) {
-      if (libraryBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        library_ = value;
-        onChanged();
-      } else {
-        libraryBuilder_.setMessage(value);
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * EXPERIMENTAL. DO NOT USE OR DEPEND ON THIS YET.
-     * "library" provides user-defined functions.
-     * Naming:
-     *   * library.function.name are in a flat namespace.
-     *     NOTE: We may need to change it to be hierarchical to support
-     *     different orgs. E.g.,
-     *     { "/google/nn", { ... }},
-     *     { "/google/vision", { ... }}
-     *     { "/org_foo/module_bar", { ... }}
-     *     map&lt;string, FunctionDefLib&gt; named_lib;
-     *   * If node[i].op is the name of one function in "library",
-     *     node[i] is deemed as a function call. Otherwise, node[i].op
-     *     must be a primitive operation supported by the runtime.
-     * Function call semantics:
-     *   * The callee may start execution as soon as some of its inputs
-     *     are ready. The caller may want to use Tuple() mechanism to
-     *     ensure all inputs are ready in the same time.
-     *   * The consumer of return values may start executing as soon as
-     *     the return values the consumer depends on are ready.  The
-     *     consumer may want to use Tuple() mechanism to ensure the
-     *     consumer does not start until all return values of the callee
-     *     function are ready.
-     * </pre>
-     *
-     * <code>.tensorflow.FunctionDefLibrary library = 2;</code>
-     */
-    public Builder setLibrary(
-        org.tensorflow.proto.framework.FunctionDefLibrary.Builder builderForValue) {
-      if (libraryBuilder_ == null) {
-        library_ = builderForValue.build();
-        onChanged();
-      } else {
-        libraryBuilder_.setMessage(builderForValue.build());
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * EXPERIMENTAL. DO NOT USE OR DEPEND ON THIS YET.
-     * "library" provides user-defined functions.
-     * Naming:
-     *   * library.function.name are in a flat namespace.
-     *     NOTE: We may need to change it to be hierarchical to support
-     *     different orgs. E.g.,
-     *     { "/google/nn", { ... }},
-     *     { "/google/vision", { ... }}
-     *     { "/org_foo/module_bar", { ... }}
-     *     map&lt;string, FunctionDefLib&gt; named_lib;
-     *   * If node[i].op is the name of one function in "library",
-     *     node[i] is deemed as a function call. Otherwise, node[i].op
-     *     must be a primitive operation supported by the runtime.
-     * Function call semantics:
-     *   * The callee may start execution as soon as some of its inputs
-     *     are ready. The caller may want to use Tuple() mechanism to
-     *     ensure all inputs are ready in the same time.
-     *   * The consumer of return values may start executing as soon as
-     *     the return values the consumer depends on are ready.  The
-     *     consumer may want to use Tuple() mechanism to ensure the
-     *     consumer does not start until all return values of the callee
-     *     function are ready.
-     * </pre>
-     *
-     * <code>.tensorflow.FunctionDefLibrary library = 2;</code>
-     */
-    public Builder mergeLibrary(org.tensorflow.proto.framework.FunctionDefLibrary value) {
-      if (libraryBuilder_ == null) {
-        if (library_ != null) {
-          library_ =
-            org.tensorflow.proto.framework.FunctionDefLibrary.newBuilder(library_).mergeFrom(value).buildPartial();
-        } else {
-          library_ = value;
-        }
-        onChanged();
-      } else {
-        libraryBuilder_.mergeFrom(value);
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * EXPERIMENTAL. DO NOT USE OR DEPEND ON THIS YET.
-     * "library" provides user-defined functions.
-     * Naming:
-     *   * library.function.name are in a flat namespace.
-     *     NOTE: We may need to change it to be hierarchical to support
-     *     different orgs. E.g.,
-     *     { "/google/nn", { ... }},
-     *     { "/google/vision", { ... }}
-     *     { "/org_foo/module_bar", { ... }}
-     *     map&lt;string, FunctionDefLib&gt; named_lib;
-     *   * If node[i].op is the name of one function in "library",
-     *     node[i] is deemed as a function call. Otherwise, node[i].op
-     *     must be a primitive operation supported by the runtime.
-     * Function call semantics:
-     *   * The callee may start execution as soon as some of its inputs
-     *     are ready. The caller may want to use Tuple() mechanism to
-     *     ensure all inputs are ready in the same time.
-     *   * The consumer of return values may start executing as soon as
-     *     the return values the consumer depends on are ready.  The
-     *     consumer may want to use Tuple() mechanism to ensure the
-     *     consumer does not start until all return values of the callee
-     *     function are ready.
-     * </pre>
-     *
-     * <code>.tensorflow.FunctionDefLibrary library = 2;</code>
-     */
-    public Builder clearLibrary() {
-      if (libraryBuilder_ == null) {
-        library_ = null;
-        onChanged();
-      } else {
-        library_ = null;
-        libraryBuilder_ = null;
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * EXPERIMENTAL. DO NOT USE OR DEPEND ON THIS YET.
-     * "library" provides user-defined functions.
-     * Naming:
-     *   * library.function.name are in a flat namespace.
-     *     NOTE: We may need to change it to be hierarchical to support
-     *     different orgs. E.g.,
-     *     { "/google/nn", { ... }},
-     *     { "/google/vision", { ... }}
-     *     { "/org_foo/module_bar", { ... }}
-     *     map&lt;string, FunctionDefLib&gt; named_lib;
-     *   * If node[i].op is the name of one function in "library",
-     *     node[i] is deemed as a function call. Otherwise, node[i].op
-     *     must be a primitive operation supported by the runtime.
-     * Function call semantics:
-     *   * The callee may start execution as soon as some of its inputs
-     *     are ready. The caller may want to use Tuple() mechanism to
-     *     ensure all inputs are ready in the same time.
-     *   * The consumer of return values may start executing as soon as
-     *     the return values the consumer depends on are ready.  The
-     *     consumer may want to use Tuple() mechanism to ensure the
-     *     consumer does not start until all return values of the callee
-     *     function are ready.
-     * </pre>
-     *
-     * <code>.tensorflow.FunctionDefLibrary library = 2;</code>
-     */
-    public org.tensorflow.proto.framework.FunctionDefLibrary.Builder getLibraryBuilder() {
-      
-      onChanged();
-      return getLibraryFieldBuilder().getBuilder();
-    }
-    /**
-     * <pre>
-     * EXPERIMENTAL. DO NOT USE OR DEPEND ON THIS YET.
-     * "library" provides user-defined functions.
-     * Naming:
-     *   * library.function.name are in a flat namespace.
-     *     NOTE: We may need to change it to be hierarchical to support
-     *     different orgs. E.g.,
-     *     { "/google/nn", { ... }},
-     *     { "/google/vision", { ... }}
-     *     { "/org_foo/module_bar", { ... }}
-     *     map&lt;string, FunctionDefLib&gt; named_lib;
-     *   * If node[i].op is the name of one function in "library",
-     *     node[i] is deemed as a function call. Otherwise, node[i].op
-     *     must be a primitive operation supported by the runtime.
-     * Function call semantics:
-     *   * The callee may start execution as soon as some of its inputs
-     *     are ready. The caller may want to use Tuple() mechanism to
-     *     ensure all inputs are ready in the same time.
-     *   * The consumer of return values may start executing as soon as
-     *     the return values the consumer depends on are ready.  The
-     *     consumer may want to use Tuple() mechanism to ensure the
-     *     consumer does not start until all return values of the callee
-     *     function are ready.
-     * </pre>
-     *
-     * <code>.tensorflow.FunctionDefLibrary library = 2;</code>
-     */
-    public org.tensorflow.proto.framework.FunctionDefLibraryOrBuilder getLibraryOrBuilder() {
-      if (libraryBuilder_ != null) {
-        return libraryBuilder_.getMessageOrBuilder();
-      } else {
-        return library_ == null ?
-            org.tensorflow.proto.framework.FunctionDefLibrary.getDefaultInstance() : library_;
-      }
-    }
-    /**
-     * <pre>
-     * EXPERIMENTAL. DO NOT USE OR DEPEND ON THIS YET.
-     * "library" provides user-defined functions.
-     * Naming:
-     *   * library.function.name are in a flat namespace.
-     *     NOTE: We may need to change it to be hierarchical to support
-     *     different orgs. E.g.,
-     *     { "/google/nn", { ... }},
-     *     { "/google/vision", { ... }}
-     *     { "/org_foo/module_bar", { ... }}
-     *     map&lt;string, FunctionDefLib&gt; named_lib;
-     *   * If node[i].op is the name of one function in "library",
-     *     node[i] is deemed as a function call. Otherwise, node[i].op
-     *     must be a primitive operation supported by the runtime.
-     * Function call semantics:
-     *   * The callee may start execution as soon as some of its inputs
-     *     are ready. The caller may want to use Tuple() mechanism to
-     *     ensure all inputs are ready in the same time.
-     *   * The consumer of return values may start executing as soon as
-     *     the return values the consumer depends on are ready.  The
-     *     consumer may want to use Tuple() mechanism to ensure the
-     *     consumer does not start until all return values of the callee
-     *     function are ready.
-     * </pre>
-     *
-     * <code>.tensorflow.FunctionDefLibrary library = 2;</code>
-     */
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.FunctionDefLibrary, org.tensorflow.proto.framework.FunctionDefLibrary.Builder, org.tensorflow.proto.framework.FunctionDefLibraryOrBuilder> 
-        getLibraryFieldBuilder() {
-      if (libraryBuilder_ == null) {
-        libraryBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-            org.tensorflow.proto.framework.FunctionDefLibrary, org.tensorflow.proto.framework.FunctionDefLibrary.Builder, org.tensorflow.proto.framework.FunctionDefLibraryOrBuilder>(
-                getLibrary(),
-                getParentForChildren(),
-                isClean());
-        library_ = null;
-      }
-      return libraryBuilder_;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.GraphDef)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.GraphDef)
-  private static final org.tensorflow.proto.framework.GraphDef DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.GraphDef();
-  }
-
-  public static org.tensorflow.proto.framework.GraphDef getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<GraphDef>
-      PARSER = new com.google.protobuf.AbstractParser<GraphDef>() {
-    @java.lang.Override
-    public GraphDef parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new GraphDef(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<GraphDef> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<GraphDef> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.framework.GraphDef getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphDefOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphDefOrBuilder.java
deleted file mode 100644
index 0e4ac4dca6e..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphDefOrBuilder.java
+++ /dev/null
@@ -1,163 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/framework/graph.proto
-
-package org.tensorflow.proto.framework;
-
-public interface GraphDefOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.GraphDef)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <code>repeated .tensorflow.NodeDef node = 1;</code>
-   */
-  java.util.List<org.tensorflow.proto.framework.NodeDef> 
-      getNodeList();
-  /**
-   * <code>repeated .tensorflow.NodeDef node = 1;</code>
-   */
-  org.tensorflow.proto.framework.NodeDef getNode(int index);
-  /**
-   * <code>repeated .tensorflow.NodeDef node = 1;</code>
-   */
-  int getNodeCount();
-  /**
-   * <code>repeated .tensorflow.NodeDef node = 1;</code>
-   */
-  java.util.List<? extends org.tensorflow.proto.framework.NodeDefOrBuilder> 
-      getNodeOrBuilderList();
-  /**
-   * <code>repeated .tensorflow.NodeDef node = 1;</code>
-   */
-  org.tensorflow.proto.framework.NodeDefOrBuilder getNodeOrBuilder(
-      int index);
-
-  /**
-   * <pre>
-   * Compatibility versions of the graph.  See core/public/version.h for version
-   * history.  The GraphDef version is distinct from the TensorFlow version, and
-   * each release of TensorFlow will support a range of GraphDef versions.
-   * </pre>
-   *
-   * <code>.tensorflow.VersionDef versions = 4;</code>
-   */
-  boolean hasVersions();
-  /**
-   * <pre>
-   * Compatibility versions of the graph.  See core/public/version.h for version
-   * history.  The GraphDef version is distinct from the TensorFlow version, and
-   * each release of TensorFlow will support a range of GraphDef versions.
-   * </pre>
-   *
-   * <code>.tensorflow.VersionDef versions = 4;</code>
-   */
-  org.tensorflow.proto.framework.VersionDef getVersions();
-  /**
-   * <pre>
-   * Compatibility versions of the graph.  See core/public/version.h for version
-   * history.  The GraphDef version is distinct from the TensorFlow version, and
-   * each release of TensorFlow will support a range of GraphDef versions.
-   * </pre>
-   *
-   * <code>.tensorflow.VersionDef versions = 4;</code>
-   */
-  org.tensorflow.proto.framework.VersionDefOrBuilder getVersionsOrBuilder();
-
-  /**
-   * <pre>
-   * Deprecated single version field; use versions above instead.  Since all
-   * GraphDef changes before "versions" was introduced were forward
-   * compatible, this field is entirely ignored.
-   * </pre>
-   *
-   * <code>int32 version = 3 [deprecated = true];</code>
-   */
-  @java.lang.Deprecated int getVersion();
-
-  /**
-   * <pre>
-   * EXPERIMENTAL. DO NOT USE OR DEPEND ON THIS YET.
-   * "library" provides user-defined functions.
-   * Naming:
-   *   * library.function.name are in a flat namespace.
-   *     NOTE: We may need to change it to be hierarchical to support
-   *     different orgs. E.g.,
-   *     { "/google/nn", { ... }},
-   *     { "/google/vision", { ... }}
-   *     { "/org_foo/module_bar", { ... }}
-   *     map&lt;string, FunctionDefLib&gt; named_lib;
-   *   * If node[i].op is the name of one function in "library",
-   *     node[i] is deemed as a function call. Otherwise, node[i].op
-   *     must be a primitive operation supported by the runtime.
-   * Function call semantics:
-   *   * The callee may start execution as soon as some of its inputs
-   *     are ready. The caller may want to use Tuple() mechanism to
-   *     ensure all inputs are ready in the same time.
-   *   * The consumer of return values may start executing as soon as
-   *     the return values the consumer depends on are ready.  The
-   *     consumer may want to use Tuple() mechanism to ensure the
-   *     consumer does not start until all return values of the callee
-   *     function are ready.
-   * </pre>
-   *
-   * <code>.tensorflow.FunctionDefLibrary library = 2;</code>
-   */
-  boolean hasLibrary();
-  /**
-   * <pre>
-   * EXPERIMENTAL. DO NOT USE OR DEPEND ON THIS YET.
-   * "library" provides user-defined functions.
-   * Naming:
-   *   * library.function.name are in a flat namespace.
-   *     NOTE: We may need to change it to be hierarchical to support
-   *     different orgs. E.g.,
-   *     { "/google/nn", { ... }},
-   *     { "/google/vision", { ... }}
-   *     { "/org_foo/module_bar", { ... }}
-   *     map&lt;string, FunctionDefLib&gt; named_lib;
-   *   * If node[i].op is the name of one function in "library",
-   *     node[i] is deemed as a function call. Otherwise, node[i].op
-   *     must be a primitive operation supported by the runtime.
-   * Function call semantics:
-   *   * The callee may start execution as soon as some of its inputs
-   *     are ready. The caller may want to use Tuple() mechanism to
-   *     ensure all inputs are ready in the same time.
-   *   * The consumer of return values may start executing as soon as
-   *     the return values the consumer depends on are ready.  The
-   *     consumer may want to use Tuple() mechanism to ensure the
-   *     consumer does not start until all return values of the callee
-   *     function are ready.
-   * </pre>
-   *
-   * <code>.tensorflow.FunctionDefLibrary library = 2;</code>
-   */
-  org.tensorflow.proto.framework.FunctionDefLibrary getLibrary();
-  /**
-   * <pre>
-   * EXPERIMENTAL. DO NOT USE OR DEPEND ON THIS YET.
-   * "library" provides user-defined functions.
-   * Naming:
-   *   * library.function.name are in a flat namespace.
-   *     NOTE: We may need to change it to be hierarchical to support
-   *     different orgs. E.g.,
-   *     { "/google/nn", { ... }},
-   *     { "/google/vision", { ... }}
-   *     { "/org_foo/module_bar", { ... }}
-   *     map&lt;string, FunctionDefLib&gt; named_lib;
-   *   * If node[i].op is the name of one function in "library",
-   *     node[i] is deemed as a function call. Otherwise, node[i].op
-   *     must be a primitive operation supported by the runtime.
-   * Function call semantics:
-   *   * The callee may start execution as soon as some of its inputs
-   *     are ready. The caller may want to use Tuple() mechanism to
-   *     ensure all inputs are ready in the same time.
-   *   * The consumer of return values may start executing as soon as
-   *     the return values the consumer depends on are ready.  The
-   *     consumer may want to use Tuple() mechanism to ensure the
-   *     consumer does not start until all return values of the callee
-   *     function are ready.
-   * </pre>
-   *
-   * <code>.tensorflow.FunctionDefLibrary library = 2;</code>
-   */
-  org.tensorflow.proto.framework.FunctionDefLibraryOrBuilder getLibraryOrBuilder();
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphOptions.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphOptions.java
deleted file mode 100644
index 5916d82ca1c..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphOptions.java
+++ /dev/null
@@ -1,1462 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/config.proto
-
-package org.tensorflow.proto.framework;
-
-/**
- * Protobuf type {@code tensorflow.GraphOptions}
- */
-public  final class GraphOptions extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.GraphOptions)
-    GraphOptionsOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use GraphOptions.newBuilder() to construct.
-  private GraphOptions(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private GraphOptions() {
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new GraphOptions();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private GraphOptions(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 16: {
-
-            enableRecvScheduling_ = input.readBool();
-            break;
-          }
-          case 26: {
-            org.tensorflow.proto.framework.OptimizerOptions.Builder subBuilder = null;
-            if (optimizerOptions_ != null) {
-              subBuilder = optimizerOptions_.toBuilder();
-            }
-            optimizerOptions_ = input.readMessage(org.tensorflow.proto.framework.OptimizerOptions.parser(), extensionRegistry);
-            if (subBuilder != null) {
-              subBuilder.mergeFrom(optimizerOptions_);
-              optimizerOptions_ = subBuilder.buildPartial();
-            }
-
-            break;
-          }
-          case 32: {
-
-            buildCostModel_ = input.readInt64();
-            break;
-          }
-          case 40: {
-
-            inferShapes_ = input.readBool();
-            break;
-          }
-          case 48: {
-
-            placePrunedGraph_ = input.readBool();
-            break;
-          }
-          case 56: {
-
-            enableBfloat16Sendrecv_ = input.readBool();
-            break;
-          }
-          case 64: {
-
-            timelineStep_ = input.readInt32();
-            break;
-          }
-          case 72: {
-
-            buildCostModelAfter_ = input.readInt64();
-            break;
-          }
-          case 82: {
-            org.tensorflow.proto.framework.RewriterConfig.Builder subBuilder = null;
-            if (rewriteOptions_ != null) {
-              subBuilder = rewriteOptions_.toBuilder();
-            }
-            rewriteOptions_ = input.readMessage(org.tensorflow.proto.framework.RewriterConfig.parser(), extensionRegistry);
-            if (subBuilder != null) {
-              subBuilder.mergeFrom(rewriteOptions_);
-              rewriteOptions_ = subBuilder.buildPartial();
-            }
-
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_GraphOptions_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_GraphOptions_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.framework.GraphOptions.class, org.tensorflow.proto.framework.GraphOptions.Builder.class);
-  }
-
-  public static final int ENABLE_RECV_SCHEDULING_FIELD_NUMBER = 2;
-  private boolean enableRecvScheduling_;
-  /**
-   * <pre>
-   * If true, use control flow to schedule the activation of Recv nodes.
-   * (Currently ignored.)
-   * </pre>
-   *
-   * <code>bool enable_recv_scheduling = 2;</code>
-   */
-  public boolean getEnableRecvScheduling() {
-    return enableRecvScheduling_;
-  }
-
-  public static final int OPTIMIZER_OPTIONS_FIELD_NUMBER = 3;
-  private org.tensorflow.proto.framework.OptimizerOptions optimizerOptions_;
-  /**
-   * <pre>
-   * Options controlling how graph is optimized.
-   * </pre>
-   *
-   * <code>.tensorflow.OptimizerOptions optimizer_options = 3;</code>
-   */
-  public boolean hasOptimizerOptions() {
-    return optimizerOptions_ != null;
-  }
-  /**
-   * <pre>
-   * Options controlling how graph is optimized.
-   * </pre>
-   *
-   * <code>.tensorflow.OptimizerOptions optimizer_options = 3;</code>
-   */
-  public org.tensorflow.proto.framework.OptimizerOptions getOptimizerOptions() {
-    return optimizerOptions_ == null ? org.tensorflow.proto.framework.OptimizerOptions.getDefaultInstance() : optimizerOptions_;
-  }
-  /**
-   * <pre>
-   * Options controlling how graph is optimized.
-   * </pre>
-   *
-   * <code>.tensorflow.OptimizerOptions optimizer_options = 3;</code>
-   */
-  public org.tensorflow.proto.framework.OptimizerOptionsOrBuilder getOptimizerOptionsOrBuilder() {
-    return getOptimizerOptions();
-  }
-
-  public static final int BUILD_COST_MODEL_FIELD_NUMBER = 4;
-  private long buildCostModel_;
-  /**
-   * <pre>
-   * The number of steps to run before returning a cost model detailing
-   * the memory usage and performance of each node of the graph. 0 means
-   * no cost model.
-   * </pre>
-   *
-   * <code>int64 build_cost_model = 4;</code>
-   */
-  public long getBuildCostModel() {
-    return buildCostModel_;
-  }
-
-  public static final int BUILD_COST_MODEL_AFTER_FIELD_NUMBER = 9;
-  private long buildCostModelAfter_;
-  /**
-   * <pre>
-   * The number of steps to skip before collecting statistics for the
-   * cost model.
-   * </pre>
-   *
-   * <code>int64 build_cost_model_after = 9;</code>
-   */
-  public long getBuildCostModelAfter() {
-    return buildCostModelAfter_;
-  }
-
-  public static final int INFER_SHAPES_FIELD_NUMBER = 5;
-  private boolean inferShapes_;
-  /**
-   * <pre>
-   * Annotate each Node with Op output shape data, to the extent it can
-   * be statically inferred.
-   * </pre>
-   *
-   * <code>bool infer_shapes = 5;</code>
-   */
-  public boolean getInferShapes() {
-    return inferShapes_;
-  }
-
-  public static final int PLACE_PRUNED_GRAPH_FIELD_NUMBER = 6;
-  private boolean placePrunedGraph_;
-  /**
-   * <pre>
-   * Only place the subgraphs that are run, rather than the entire graph.
-   * This is useful for interactive graph building, where one might
-   * produce graphs that cannot be placed during the debugging
-   * process.  In particular, it allows the client to continue work in
-   * a session after adding a node to a graph whose placement
-   * constraints are unsatisfiable.
-   * </pre>
-   *
-   * <code>bool place_pruned_graph = 6;</code>
-   */
-  public boolean getPlacePrunedGraph() {
-    return placePrunedGraph_;
-  }
-
-  public static final int ENABLE_BFLOAT16_SENDRECV_FIELD_NUMBER = 7;
-  private boolean enableBfloat16Sendrecv_;
-  /**
-   * <pre>
-   * If true, transfer float values between processes as bfloat16.
-   * </pre>
-   *
-   * <code>bool enable_bfloat16_sendrecv = 7;</code>
-   */
-  public boolean getEnableBfloat16Sendrecv() {
-    return enableBfloat16Sendrecv_;
-  }
-
-  public static final int TIMELINE_STEP_FIELD_NUMBER = 8;
-  private int timelineStep_;
-  /**
-   * <pre>
-   * If &gt; 0, record a timeline every this many steps.
-   * EXPERIMENTAL: This currently has no effect in MasterSession.
-   * </pre>
-   *
-   * <code>int32 timeline_step = 8;</code>
-   */
-  public int getTimelineStep() {
-    return timelineStep_;
-  }
-
-  public static final int REWRITE_OPTIONS_FIELD_NUMBER = 10;
-  private org.tensorflow.proto.framework.RewriterConfig rewriteOptions_;
-  /**
-   * <pre>
-   * Options that control the type and amount of graph rewriting.
-   * Not currently configurable via the public Python API (i.e. there is no API
-   * stability guarantee if you import RewriterConfig explicitly).
-   * </pre>
-   *
-   * <code>.tensorflow.RewriterConfig rewrite_options = 10;</code>
-   */
-  public boolean hasRewriteOptions() {
-    return rewriteOptions_ != null;
-  }
-  /**
-   * <pre>
-   * Options that control the type and amount of graph rewriting.
-   * Not currently configurable via the public Python API (i.e. there is no API
-   * stability guarantee if you import RewriterConfig explicitly).
-   * </pre>
-   *
-   * <code>.tensorflow.RewriterConfig rewrite_options = 10;</code>
-   */
-  public org.tensorflow.proto.framework.RewriterConfig getRewriteOptions() {
-    return rewriteOptions_ == null ? org.tensorflow.proto.framework.RewriterConfig.getDefaultInstance() : rewriteOptions_;
-  }
-  /**
-   * <pre>
-   * Options that control the type and amount of graph rewriting.
-   * Not currently configurable via the public Python API (i.e. there is no API
-   * stability guarantee if you import RewriterConfig explicitly).
-   * </pre>
-   *
-   * <code>.tensorflow.RewriterConfig rewrite_options = 10;</code>
-   */
-  public org.tensorflow.proto.framework.RewriterConfigOrBuilder getRewriteOptionsOrBuilder() {
-    return getRewriteOptions();
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    if (enableRecvScheduling_ != false) {
-      output.writeBool(2, enableRecvScheduling_);
-    }
-    if (optimizerOptions_ != null) {
-      output.writeMessage(3, getOptimizerOptions());
-    }
-    if (buildCostModel_ != 0L) {
-      output.writeInt64(4, buildCostModel_);
-    }
-    if (inferShapes_ != false) {
-      output.writeBool(5, inferShapes_);
-    }
-    if (placePrunedGraph_ != false) {
-      output.writeBool(6, placePrunedGraph_);
-    }
-    if (enableBfloat16Sendrecv_ != false) {
-      output.writeBool(7, enableBfloat16Sendrecv_);
-    }
-    if (timelineStep_ != 0) {
-      output.writeInt32(8, timelineStep_);
-    }
-    if (buildCostModelAfter_ != 0L) {
-      output.writeInt64(9, buildCostModelAfter_);
-    }
-    if (rewriteOptions_ != null) {
-      output.writeMessage(10, getRewriteOptions());
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (enableRecvScheduling_ != false) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeBoolSize(2, enableRecvScheduling_);
-    }
-    if (optimizerOptions_ != null) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(3, getOptimizerOptions());
-    }
-    if (buildCostModel_ != 0L) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt64Size(4, buildCostModel_);
-    }
-    if (inferShapes_ != false) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeBoolSize(5, inferShapes_);
-    }
-    if (placePrunedGraph_ != false) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeBoolSize(6, placePrunedGraph_);
-    }
-    if (enableBfloat16Sendrecv_ != false) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeBoolSize(7, enableBfloat16Sendrecv_);
-    }
-    if (timelineStep_ != 0) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt32Size(8, timelineStep_);
-    }
-    if (buildCostModelAfter_ != 0L) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt64Size(9, buildCostModelAfter_);
-    }
-    if (rewriteOptions_ != null) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(10, getRewriteOptions());
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.framework.GraphOptions)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.framework.GraphOptions other = (org.tensorflow.proto.framework.GraphOptions) obj;
-
-    if (getEnableRecvScheduling()
-        != other.getEnableRecvScheduling()) return false;
-    if (hasOptimizerOptions() != other.hasOptimizerOptions()) return false;
-    if (hasOptimizerOptions()) {
-      if (!getOptimizerOptions()
-          .equals(other.getOptimizerOptions())) return false;
-    }
-    if (getBuildCostModel()
-        != other.getBuildCostModel()) return false;
-    if (getBuildCostModelAfter()
-        != other.getBuildCostModelAfter()) return false;
-    if (getInferShapes()
-        != other.getInferShapes()) return false;
-    if (getPlacePrunedGraph()
-        != other.getPlacePrunedGraph()) return false;
-    if (getEnableBfloat16Sendrecv()
-        != other.getEnableBfloat16Sendrecv()) return false;
-    if (getTimelineStep()
-        != other.getTimelineStep()) return false;
-    if (hasRewriteOptions() != other.hasRewriteOptions()) return false;
-    if (hasRewriteOptions()) {
-      if (!getRewriteOptions()
-          .equals(other.getRewriteOptions())) return false;
-    }
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    hash = (37 * hash) + ENABLE_RECV_SCHEDULING_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashBoolean(
-        getEnableRecvScheduling());
-    if (hasOptimizerOptions()) {
-      hash = (37 * hash) + OPTIMIZER_OPTIONS_FIELD_NUMBER;
-      hash = (53 * hash) + getOptimizerOptions().hashCode();
-    }
-    hash = (37 * hash) + BUILD_COST_MODEL_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-        getBuildCostModel());
-    hash = (37 * hash) + BUILD_COST_MODEL_AFTER_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-        getBuildCostModelAfter());
-    hash = (37 * hash) + INFER_SHAPES_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashBoolean(
-        getInferShapes());
-    hash = (37 * hash) + PLACE_PRUNED_GRAPH_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashBoolean(
-        getPlacePrunedGraph());
-    hash = (37 * hash) + ENABLE_BFLOAT16_SENDRECV_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashBoolean(
-        getEnableBfloat16Sendrecv());
-    hash = (37 * hash) + TIMELINE_STEP_FIELD_NUMBER;
-    hash = (53 * hash) + getTimelineStep();
-    if (hasRewriteOptions()) {
-      hash = (37 * hash) + REWRITE_OPTIONS_FIELD_NUMBER;
-      hash = (53 * hash) + getRewriteOptions().hashCode();
-    }
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.framework.GraphOptions parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.GraphOptions parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.GraphOptions parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.GraphOptions parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.GraphOptions parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.GraphOptions parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.GraphOptions parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.GraphOptions parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.GraphOptions parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.GraphOptions parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.GraphOptions parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.GraphOptions parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.framework.GraphOptions prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * Protobuf type {@code tensorflow.GraphOptions}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.GraphOptions)
-      org.tensorflow.proto.framework.GraphOptionsOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_GraphOptions_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_GraphOptions_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.GraphOptions.class, org.tensorflow.proto.framework.GraphOptions.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.framework.GraphOptions.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      enableRecvScheduling_ = false;
-
-      if (optimizerOptionsBuilder_ == null) {
-        optimizerOptions_ = null;
-      } else {
-        optimizerOptions_ = null;
-        optimizerOptionsBuilder_ = null;
-      }
-      buildCostModel_ = 0L;
-
-      buildCostModelAfter_ = 0L;
-
-      inferShapes_ = false;
-
-      placePrunedGraph_ = false;
-
-      enableBfloat16Sendrecv_ = false;
-
-      timelineStep_ = 0;
-
-      if (rewriteOptionsBuilder_ == null) {
-        rewriteOptions_ = null;
-      } else {
-        rewriteOptions_ = null;
-        rewriteOptionsBuilder_ = null;
-      }
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_GraphOptions_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.GraphOptions getDefaultInstanceForType() {
-      return org.tensorflow.proto.framework.GraphOptions.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.GraphOptions build() {
-      org.tensorflow.proto.framework.GraphOptions result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.GraphOptions buildPartial() {
-      org.tensorflow.proto.framework.GraphOptions result = new org.tensorflow.proto.framework.GraphOptions(this);
-      result.enableRecvScheduling_ = enableRecvScheduling_;
-      if (optimizerOptionsBuilder_ == null) {
-        result.optimizerOptions_ = optimizerOptions_;
-      } else {
-        result.optimizerOptions_ = optimizerOptionsBuilder_.build();
-      }
-      result.buildCostModel_ = buildCostModel_;
-      result.buildCostModelAfter_ = buildCostModelAfter_;
-      result.inferShapes_ = inferShapes_;
-      result.placePrunedGraph_ = placePrunedGraph_;
-      result.enableBfloat16Sendrecv_ = enableBfloat16Sendrecv_;
-      result.timelineStep_ = timelineStep_;
-      if (rewriteOptionsBuilder_ == null) {
-        result.rewriteOptions_ = rewriteOptions_;
-      } else {
-        result.rewriteOptions_ = rewriteOptionsBuilder_.build();
-      }
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.framework.GraphOptions) {
-        return mergeFrom((org.tensorflow.proto.framework.GraphOptions)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.framework.GraphOptions other) {
-      if (other == org.tensorflow.proto.framework.GraphOptions.getDefaultInstance()) return this;
-      if (other.getEnableRecvScheduling() != false) {
-        setEnableRecvScheduling(other.getEnableRecvScheduling());
-      }
-      if (other.hasOptimizerOptions()) {
-        mergeOptimizerOptions(other.getOptimizerOptions());
-      }
-      if (other.getBuildCostModel() != 0L) {
-        setBuildCostModel(other.getBuildCostModel());
-      }
-      if (other.getBuildCostModelAfter() != 0L) {
-        setBuildCostModelAfter(other.getBuildCostModelAfter());
-      }
-      if (other.getInferShapes() != false) {
-        setInferShapes(other.getInferShapes());
-      }
-      if (other.getPlacePrunedGraph() != false) {
-        setPlacePrunedGraph(other.getPlacePrunedGraph());
-      }
-      if (other.getEnableBfloat16Sendrecv() != false) {
-        setEnableBfloat16Sendrecv(other.getEnableBfloat16Sendrecv());
-      }
-      if (other.getTimelineStep() != 0) {
-        setTimelineStep(other.getTimelineStep());
-      }
-      if (other.hasRewriteOptions()) {
-        mergeRewriteOptions(other.getRewriteOptions());
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.framework.GraphOptions parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.framework.GraphOptions) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-
-    private boolean enableRecvScheduling_ ;
-    /**
-     * <pre>
-     * If true, use control flow to schedule the activation of Recv nodes.
-     * (Currently ignored.)
-     * </pre>
-     *
-     * <code>bool enable_recv_scheduling = 2;</code>
-     */
-    public boolean getEnableRecvScheduling() {
-      return enableRecvScheduling_;
-    }
-    /**
-     * <pre>
-     * If true, use control flow to schedule the activation of Recv nodes.
-     * (Currently ignored.)
-     * </pre>
-     *
-     * <code>bool enable_recv_scheduling = 2;</code>
-     */
-    public Builder setEnableRecvScheduling(boolean value) {
-      
-      enableRecvScheduling_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * If true, use control flow to schedule the activation of Recv nodes.
-     * (Currently ignored.)
-     * </pre>
-     *
-     * <code>bool enable_recv_scheduling = 2;</code>
-     */
-    public Builder clearEnableRecvScheduling() {
-      
-      enableRecvScheduling_ = false;
-      onChanged();
-      return this;
-    }
-
-    private org.tensorflow.proto.framework.OptimizerOptions optimizerOptions_;
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.OptimizerOptions, org.tensorflow.proto.framework.OptimizerOptions.Builder, org.tensorflow.proto.framework.OptimizerOptionsOrBuilder> optimizerOptionsBuilder_;
-    /**
-     * <pre>
-     * Options controlling how graph is optimized.
-     * </pre>
-     *
-     * <code>.tensorflow.OptimizerOptions optimizer_options = 3;</code>
-     */
-    public boolean hasOptimizerOptions() {
-      return optimizerOptionsBuilder_ != null || optimizerOptions_ != null;
-    }
-    /**
-     * <pre>
-     * Options controlling how graph is optimized.
-     * </pre>
-     *
-     * <code>.tensorflow.OptimizerOptions optimizer_options = 3;</code>
-     */
-    public org.tensorflow.proto.framework.OptimizerOptions getOptimizerOptions() {
-      if (optimizerOptionsBuilder_ == null) {
-        return optimizerOptions_ == null ? org.tensorflow.proto.framework.OptimizerOptions.getDefaultInstance() : optimizerOptions_;
-      } else {
-        return optimizerOptionsBuilder_.getMessage();
-      }
-    }
-    /**
-     * <pre>
-     * Options controlling how graph is optimized.
-     * </pre>
-     *
-     * <code>.tensorflow.OptimizerOptions optimizer_options = 3;</code>
-     */
-    public Builder setOptimizerOptions(org.tensorflow.proto.framework.OptimizerOptions value) {
-      if (optimizerOptionsBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        optimizerOptions_ = value;
-        onChanged();
-      } else {
-        optimizerOptionsBuilder_.setMessage(value);
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Options controlling how graph is optimized.
-     * </pre>
-     *
-     * <code>.tensorflow.OptimizerOptions optimizer_options = 3;</code>
-     */
-    public Builder setOptimizerOptions(
-        org.tensorflow.proto.framework.OptimizerOptions.Builder builderForValue) {
-      if (optimizerOptionsBuilder_ == null) {
-        optimizerOptions_ = builderForValue.build();
-        onChanged();
-      } else {
-        optimizerOptionsBuilder_.setMessage(builderForValue.build());
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Options controlling how graph is optimized.
-     * </pre>
-     *
-     * <code>.tensorflow.OptimizerOptions optimizer_options = 3;</code>
-     */
-    public Builder mergeOptimizerOptions(org.tensorflow.proto.framework.OptimizerOptions value) {
-      if (optimizerOptionsBuilder_ == null) {
-        if (optimizerOptions_ != null) {
-          optimizerOptions_ =
-            org.tensorflow.proto.framework.OptimizerOptions.newBuilder(optimizerOptions_).mergeFrom(value).buildPartial();
-        } else {
-          optimizerOptions_ = value;
-        }
-        onChanged();
-      } else {
-        optimizerOptionsBuilder_.mergeFrom(value);
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Options controlling how graph is optimized.
-     * </pre>
-     *
-     * <code>.tensorflow.OptimizerOptions optimizer_options = 3;</code>
-     */
-    public Builder clearOptimizerOptions() {
-      if (optimizerOptionsBuilder_ == null) {
-        optimizerOptions_ = null;
-        onChanged();
-      } else {
-        optimizerOptions_ = null;
-        optimizerOptionsBuilder_ = null;
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Options controlling how graph is optimized.
-     * </pre>
-     *
-     * <code>.tensorflow.OptimizerOptions optimizer_options = 3;</code>
-     */
-    public org.tensorflow.proto.framework.OptimizerOptions.Builder getOptimizerOptionsBuilder() {
-      
-      onChanged();
-      return getOptimizerOptionsFieldBuilder().getBuilder();
-    }
-    /**
-     * <pre>
-     * Options controlling how graph is optimized.
-     * </pre>
-     *
-     * <code>.tensorflow.OptimizerOptions optimizer_options = 3;</code>
-     */
-    public org.tensorflow.proto.framework.OptimizerOptionsOrBuilder getOptimizerOptionsOrBuilder() {
-      if (optimizerOptionsBuilder_ != null) {
-        return optimizerOptionsBuilder_.getMessageOrBuilder();
-      } else {
-        return optimizerOptions_ == null ?
-            org.tensorflow.proto.framework.OptimizerOptions.getDefaultInstance() : optimizerOptions_;
-      }
-    }
-    /**
-     * <pre>
-     * Options controlling how graph is optimized.
-     * </pre>
-     *
-     * <code>.tensorflow.OptimizerOptions optimizer_options = 3;</code>
-     */
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.OptimizerOptions, org.tensorflow.proto.framework.OptimizerOptions.Builder, org.tensorflow.proto.framework.OptimizerOptionsOrBuilder> 
-        getOptimizerOptionsFieldBuilder() {
-      if (optimizerOptionsBuilder_ == null) {
-        optimizerOptionsBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-            org.tensorflow.proto.framework.OptimizerOptions, org.tensorflow.proto.framework.OptimizerOptions.Builder, org.tensorflow.proto.framework.OptimizerOptionsOrBuilder>(
-                getOptimizerOptions(),
-                getParentForChildren(),
-                isClean());
-        optimizerOptions_ = null;
-      }
-      return optimizerOptionsBuilder_;
-    }
-
-    private long buildCostModel_ ;
-    /**
-     * <pre>
-     * The number of steps to run before returning a cost model detailing
-     * the memory usage and performance of each node of the graph. 0 means
-     * no cost model.
-     * </pre>
-     *
-     * <code>int64 build_cost_model = 4;</code>
-     */
-    public long getBuildCostModel() {
-      return buildCostModel_;
-    }
-    /**
-     * <pre>
-     * The number of steps to run before returning a cost model detailing
-     * the memory usage and performance of each node of the graph. 0 means
-     * no cost model.
-     * </pre>
-     *
-     * <code>int64 build_cost_model = 4;</code>
-     */
-    public Builder setBuildCostModel(long value) {
-      
-      buildCostModel_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * The number of steps to run before returning a cost model detailing
-     * the memory usage and performance of each node of the graph. 0 means
-     * no cost model.
-     * </pre>
-     *
-     * <code>int64 build_cost_model = 4;</code>
-     */
-    public Builder clearBuildCostModel() {
-      
-      buildCostModel_ = 0L;
-      onChanged();
-      return this;
-    }
-
-    private long buildCostModelAfter_ ;
-    /**
-     * <pre>
-     * The number of steps to skip before collecting statistics for the
-     * cost model.
-     * </pre>
-     *
-     * <code>int64 build_cost_model_after = 9;</code>
-     */
-    public long getBuildCostModelAfter() {
-      return buildCostModelAfter_;
-    }
-    /**
-     * <pre>
-     * The number of steps to skip before collecting statistics for the
-     * cost model.
-     * </pre>
-     *
-     * <code>int64 build_cost_model_after = 9;</code>
-     */
-    public Builder setBuildCostModelAfter(long value) {
-      
-      buildCostModelAfter_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * The number of steps to skip before collecting statistics for the
-     * cost model.
-     * </pre>
-     *
-     * <code>int64 build_cost_model_after = 9;</code>
-     */
-    public Builder clearBuildCostModelAfter() {
-      
-      buildCostModelAfter_ = 0L;
-      onChanged();
-      return this;
-    }
-
-    private boolean inferShapes_ ;
-    /**
-     * <pre>
-     * Annotate each Node with Op output shape data, to the extent it can
-     * be statically inferred.
-     * </pre>
-     *
-     * <code>bool infer_shapes = 5;</code>
-     */
-    public boolean getInferShapes() {
-      return inferShapes_;
-    }
-    /**
-     * <pre>
-     * Annotate each Node with Op output shape data, to the extent it can
-     * be statically inferred.
-     * </pre>
-     *
-     * <code>bool infer_shapes = 5;</code>
-     */
-    public Builder setInferShapes(boolean value) {
-      
-      inferShapes_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Annotate each Node with Op output shape data, to the extent it can
-     * be statically inferred.
-     * </pre>
-     *
-     * <code>bool infer_shapes = 5;</code>
-     */
-    public Builder clearInferShapes() {
-      
-      inferShapes_ = false;
-      onChanged();
-      return this;
-    }
-
-    private boolean placePrunedGraph_ ;
-    /**
-     * <pre>
-     * Only place the subgraphs that are run, rather than the entire graph.
-     * This is useful for interactive graph building, where one might
-     * produce graphs that cannot be placed during the debugging
-     * process.  In particular, it allows the client to continue work in
-     * a session after adding a node to a graph whose placement
-     * constraints are unsatisfiable.
-     * </pre>
-     *
-     * <code>bool place_pruned_graph = 6;</code>
-     */
-    public boolean getPlacePrunedGraph() {
-      return placePrunedGraph_;
-    }
-    /**
-     * <pre>
-     * Only place the subgraphs that are run, rather than the entire graph.
-     * This is useful for interactive graph building, where one might
-     * produce graphs that cannot be placed during the debugging
-     * process.  In particular, it allows the client to continue work in
-     * a session after adding a node to a graph whose placement
-     * constraints are unsatisfiable.
-     * </pre>
-     *
-     * <code>bool place_pruned_graph = 6;</code>
-     */
-    public Builder setPlacePrunedGraph(boolean value) {
-      
-      placePrunedGraph_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Only place the subgraphs that are run, rather than the entire graph.
-     * This is useful for interactive graph building, where one might
-     * produce graphs that cannot be placed during the debugging
-     * process.  In particular, it allows the client to continue work in
-     * a session after adding a node to a graph whose placement
-     * constraints are unsatisfiable.
-     * </pre>
-     *
-     * <code>bool place_pruned_graph = 6;</code>
-     */
-    public Builder clearPlacePrunedGraph() {
-      
-      placePrunedGraph_ = false;
-      onChanged();
-      return this;
-    }
-
-    private boolean enableBfloat16Sendrecv_ ;
-    /**
-     * <pre>
-     * If true, transfer float values between processes as bfloat16.
-     * </pre>
-     *
-     * <code>bool enable_bfloat16_sendrecv = 7;</code>
-     */
-    public boolean getEnableBfloat16Sendrecv() {
-      return enableBfloat16Sendrecv_;
-    }
-    /**
-     * <pre>
-     * If true, transfer float values between processes as bfloat16.
-     * </pre>
-     *
-     * <code>bool enable_bfloat16_sendrecv = 7;</code>
-     */
-    public Builder setEnableBfloat16Sendrecv(boolean value) {
-      
-      enableBfloat16Sendrecv_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * If true, transfer float values between processes as bfloat16.
-     * </pre>
-     *
-     * <code>bool enable_bfloat16_sendrecv = 7;</code>
-     */
-    public Builder clearEnableBfloat16Sendrecv() {
-      
-      enableBfloat16Sendrecv_ = false;
-      onChanged();
-      return this;
-    }
-
-    private int timelineStep_ ;
-    /**
-     * <pre>
-     * If &gt; 0, record a timeline every this many steps.
-     * EXPERIMENTAL: This currently has no effect in MasterSession.
-     * </pre>
-     *
-     * <code>int32 timeline_step = 8;</code>
-     */
-    public int getTimelineStep() {
-      return timelineStep_;
-    }
-    /**
-     * <pre>
-     * If &gt; 0, record a timeline every this many steps.
-     * EXPERIMENTAL: This currently has no effect in MasterSession.
-     * </pre>
-     *
-     * <code>int32 timeline_step = 8;</code>
-     */
-    public Builder setTimelineStep(int value) {
-      
-      timelineStep_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * If &gt; 0, record a timeline every this many steps.
-     * EXPERIMENTAL: This currently has no effect in MasterSession.
-     * </pre>
-     *
-     * <code>int32 timeline_step = 8;</code>
-     */
-    public Builder clearTimelineStep() {
-      
-      timelineStep_ = 0;
-      onChanged();
-      return this;
-    }
-
-    private org.tensorflow.proto.framework.RewriterConfig rewriteOptions_;
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.RewriterConfig, org.tensorflow.proto.framework.RewriterConfig.Builder, org.tensorflow.proto.framework.RewriterConfigOrBuilder> rewriteOptionsBuilder_;
-    /**
-     * <pre>
-     * Options that control the type and amount of graph rewriting.
-     * Not currently configurable via the public Python API (i.e. there is no API
-     * stability guarantee if you import RewriterConfig explicitly).
-     * </pre>
-     *
-     * <code>.tensorflow.RewriterConfig rewrite_options = 10;</code>
-     */
-    public boolean hasRewriteOptions() {
-      return rewriteOptionsBuilder_ != null || rewriteOptions_ != null;
-    }
-    /**
-     * <pre>
-     * Options that control the type and amount of graph rewriting.
-     * Not currently configurable via the public Python API (i.e. there is no API
-     * stability guarantee if you import RewriterConfig explicitly).
-     * </pre>
-     *
-     * <code>.tensorflow.RewriterConfig rewrite_options = 10;</code>
-     */
-    public org.tensorflow.proto.framework.RewriterConfig getRewriteOptions() {
-      if (rewriteOptionsBuilder_ == null) {
-        return rewriteOptions_ == null ? org.tensorflow.proto.framework.RewriterConfig.getDefaultInstance() : rewriteOptions_;
-      } else {
-        return rewriteOptionsBuilder_.getMessage();
-      }
-    }
-    /**
-     * <pre>
-     * Options that control the type and amount of graph rewriting.
-     * Not currently configurable via the public Python API (i.e. there is no API
-     * stability guarantee if you import RewriterConfig explicitly).
-     * </pre>
-     *
-     * <code>.tensorflow.RewriterConfig rewrite_options = 10;</code>
-     */
-    public Builder setRewriteOptions(org.tensorflow.proto.framework.RewriterConfig value) {
-      if (rewriteOptionsBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        rewriteOptions_ = value;
-        onChanged();
-      } else {
-        rewriteOptionsBuilder_.setMessage(value);
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Options that control the type and amount of graph rewriting.
-     * Not currently configurable via the public Python API (i.e. there is no API
-     * stability guarantee if you import RewriterConfig explicitly).
-     * </pre>
-     *
-     * <code>.tensorflow.RewriterConfig rewrite_options = 10;</code>
-     */
-    public Builder setRewriteOptions(
-        org.tensorflow.proto.framework.RewriterConfig.Builder builderForValue) {
-      if (rewriteOptionsBuilder_ == null) {
-        rewriteOptions_ = builderForValue.build();
-        onChanged();
-      } else {
-        rewriteOptionsBuilder_.setMessage(builderForValue.build());
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Options that control the type and amount of graph rewriting.
-     * Not currently configurable via the public Python API (i.e. there is no API
-     * stability guarantee if you import RewriterConfig explicitly).
-     * </pre>
-     *
-     * <code>.tensorflow.RewriterConfig rewrite_options = 10;</code>
-     */
-    public Builder mergeRewriteOptions(org.tensorflow.proto.framework.RewriterConfig value) {
-      if (rewriteOptionsBuilder_ == null) {
-        if (rewriteOptions_ != null) {
-          rewriteOptions_ =
-            org.tensorflow.proto.framework.RewriterConfig.newBuilder(rewriteOptions_).mergeFrom(value).buildPartial();
-        } else {
-          rewriteOptions_ = value;
-        }
-        onChanged();
-      } else {
-        rewriteOptionsBuilder_.mergeFrom(value);
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Options that control the type and amount of graph rewriting.
-     * Not currently configurable via the public Python API (i.e. there is no API
-     * stability guarantee if you import RewriterConfig explicitly).
-     * </pre>
-     *
-     * <code>.tensorflow.RewriterConfig rewrite_options = 10;</code>
-     */
-    public Builder clearRewriteOptions() {
-      if (rewriteOptionsBuilder_ == null) {
-        rewriteOptions_ = null;
-        onChanged();
-      } else {
-        rewriteOptions_ = null;
-        rewriteOptionsBuilder_ = null;
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Options that control the type and amount of graph rewriting.
-     * Not currently configurable via the public Python API (i.e. there is no API
-     * stability guarantee if you import RewriterConfig explicitly).
-     * </pre>
-     *
-     * <code>.tensorflow.RewriterConfig rewrite_options = 10;</code>
-     */
-    public org.tensorflow.proto.framework.RewriterConfig.Builder getRewriteOptionsBuilder() {
-      
-      onChanged();
-      return getRewriteOptionsFieldBuilder().getBuilder();
-    }
-    /**
-     * <pre>
-     * Options that control the type and amount of graph rewriting.
-     * Not currently configurable via the public Python API (i.e. there is no API
-     * stability guarantee if you import RewriterConfig explicitly).
-     * </pre>
-     *
-     * <code>.tensorflow.RewriterConfig rewrite_options = 10;</code>
-     */
-    public org.tensorflow.proto.framework.RewriterConfigOrBuilder getRewriteOptionsOrBuilder() {
-      if (rewriteOptionsBuilder_ != null) {
-        return rewriteOptionsBuilder_.getMessageOrBuilder();
-      } else {
-        return rewriteOptions_ == null ?
-            org.tensorflow.proto.framework.RewriterConfig.getDefaultInstance() : rewriteOptions_;
-      }
-    }
-    /**
-     * <pre>
-     * Options that control the type and amount of graph rewriting.
-     * Not currently configurable via the public Python API (i.e. there is no API
-     * stability guarantee if you import RewriterConfig explicitly).
-     * </pre>
-     *
-     * <code>.tensorflow.RewriterConfig rewrite_options = 10;</code>
-     */
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.RewriterConfig, org.tensorflow.proto.framework.RewriterConfig.Builder, org.tensorflow.proto.framework.RewriterConfigOrBuilder> 
-        getRewriteOptionsFieldBuilder() {
-      if (rewriteOptionsBuilder_ == null) {
-        rewriteOptionsBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-            org.tensorflow.proto.framework.RewriterConfig, org.tensorflow.proto.framework.RewriterConfig.Builder, org.tensorflow.proto.framework.RewriterConfigOrBuilder>(
-                getRewriteOptions(),
-                getParentForChildren(),
-                isClean());
-        rewriteOptions_ = null;
-      }
-      return rewriteOptionsBuilder_;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.GraphOptions)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.GraphOptions)
-  private static final org.tensorflow.proto.framework.GraphOptions DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.GraphOptions();
-  }
-
-  public static org.tensorflow.proto.framework.GraphOptions getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<GraphOptions>
-      PARSER = new com.google.protobuf.AbstractParser<GraphOptions>() {
-    @java.lang.Override
-    public GraphOptions parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new GraphOptions(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<GraphOptions> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<GraphOptions> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.framework.GraphOptions getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphOptionsOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphOptionsOrBuilder.java
deleted file mode 100644
index bc373a1f956..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphOptionsOrBuilder.java
+++ /dev/null
@@ -1,139 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/config.proto
-
-package org.tensorflow.proto.framework;
-
-public interface GraphOptionsOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.GraphOptions)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <pre>
-   * If true, use control flow to schedule the activation of Recv nodes.
-   * (Currently ignored.)
-   * </pre>
-   *
-   * <code>bool enable_recv_scheduling = 2;</code>
-   */
-  boolean getEnableRecvScheduling();
-
-  /**
-   * <pre>
-   * Options controlling how graph is optimized.
-   * </pre>
-   *
-   * <code>.tensorflow.OptimizerOptions optimizer_options = 3;</code>
-   */
-  boolean hasOptimizerOptions();
-  /**
-   * <pre>
-   * Options controlling how graph is optimized.
-   * </pre>
-   *
-   * <code>.tensorflow.OptimizerOptions optimizer_options = 3;</code>
-   */
-  org.tensorflow.proto.framework.OptimizerOptions getOptimizerOptions();
-  /**
-   * <pre>
-   * Options controlling how graph is optimized.
-   * </pre>
-   *
-   * <code>.tensorflow.OptimizerOptions optimizer_options = 3;</code>
-   */
-  org.tensorflow.proto.framework.OptimizerOptionsOrBuilder getOptimizerOptionsOrBuilder();
-
-  /**
-   * <pre>
-   * The number of steps to run before returning a cost model detailing
-   * the memory usage and performance of each node of the graph. 0 means
-   * no cost model.
-   * </pre>
-   *
-   * <code>int64 build_cost_model = 4;</code>
-   */
-  long getBuildCostModel();
-
-  /**
-   * <pre>
-   * The number of steps to skip before collecting statistics for the
-   * cost model.
-   * </pre>
-   *
-   * <code>int64 build_cost_model_after = 9;</code>
-   */
-  long getBuildCostModelAfter();
-
-  /**
-   * <pre>
-   * Annotate each Node with Op output shape data, to the extent it can
-   * be statically inferred.
-   * </pre>
-   *
-   * <code>bool infer_shapes = 5;</code>
-   */
-  boolean getInferShapes();
-
-  /**
-   * <pre>
-   * Only place the subgraphs that are run, rather than the entire graph.
-   * This is useful for interactive graph building, where one might
-   * produce graphs that cannot be placed during the debugging
-   * process.  In particular, it allows the client to continue work in
-   * a session after adding a node to a graph whose placement
-   * constraints are unsatisfiable.
-   * </pre>
-   *
-   * <code>bool place_pruned_graph = 6;</code>
-   */
-  boolean getPlacePrunedGraph();
-
-  /**
-   * <pre>
-   * If true, transfer float values between processes as bfloat16.
-   * </pre>
-   *
-   * <code>bool enable_bfloat16_sendrecv = 7;</code>
-   */
-  boolean getEnableBfloat16Sendrecv();
-
-  /**
-   * <pre>
-   * If &gt; 0, record a timeline every this many steps.
-   * EXPERIMENTAL: This currently has no effect in MasterSession.
-   * </pre>
-   *
-   * <code>int32 timeline_step = 8;</code>
-   */
-  int getTimelineStep();
-
-  /**
-   * <pre>
-   * Options that control the type and amount of graph rewriting.
-   * Not currently configurable via the public Python API (i.e. there is no API
-   * stability guarantee if you import RewriterConfig explicitly).
-   * </pre>
-   *
-   * <code>.tensorflow.RewriterConfig rewrite_options = 10;</code>
-   */
-  boolean hasRewriteOptions();
-  /**
-   * <pre>
-   * Options that control the type and amount of graph rewriting.
-   * Not currently configurable via the public Python API (i.e. there is no API
-   * stability guarantee if you import RewriterConfig explicitly).
-   * </pre>
-   *
-   * <code>.tensorflow.RewriterConfig rewrite_options = 10;</code>
-   */
-  org.tensorflow.proto.framework.RewriterConfig getRewriteOptions();
-  /**
-   * <pre>
-   * Options that control the type and amount of graph rewriting.
-   * Not currently configurable via the public Python API (i.e. there is no API
-   * stability guarantee if you import RewriterConfig explicitly).
-   * </pre>
-   *
-   * <code>.tensorflow.RewriterConfig rewrite_options = 10;</code>
-   */
-  org.tensorflow.proto.framework.RewriterConfigOrBuilder getRewriteOptionsOrBuilder();
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphProtos.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphProtos.java
deleted file mode 100644
index 43eba4ee9bd..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphProtos.java
+++ /dev/null
@@ -1,63 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/framework/graph.proto
-
-package org.tensorflow.proto.framework;
-
-public final class GraphProtos {
-  private GraphProtos() {}
-  public static void registerAllExtensions(
-      com.google.protobuf.ExtensionRegistryLite registry) {
-  }
-
-  public static void registerAllExtensions(
-      com.google.protobuf.ExtensionRegistry registry) {
-    registerAllExtensions(
-        (com.google.protobuf.ExtensionRegistryLite) registry);
-  }
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_GraphDef_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_GraphDef_fieldAccessorTable;
-
-  public static com.google.protobuf.Descriptors.FileDescriptor
-      getDescriptor() {
-    return descriptor;
-  }
-  private static  com.google.protobuf.Descriptors.FileDescriptor
-      descriptor;
-  static {
-    java.lang.String[] descriptorData = {
-      "\n%tensorflow/core/framework/graph.proto\022" +
-      "\ntensorflow\032(tensorflow/core/framework/f" +
-      "unction.proto\032(tensorflow/core/framework" +
-      "/node_def.proto\032(tensorflow/core/framewo" +
-      "rk/versions.proto\"\235\001\n\010GraphDef\022!\n\004node\030\001" +
-      " \003(\0132\023.tensorflow.NodeDef\022(\n\010versions\030\004 " +
-      "\001(\0132\026.tensorflow.VersionDef\022\023\n\007version\030\003" +
-      " \001(\005B\002\030\001\022/\n\007library\030\002 \001(\0132\036.tensorflow.F" +
-      "unctionDefLibraryB\200\001\n\036org.tensorflow.pro" +
-      "to.frameworkB\013GraphProtosP\001ZLgithub.com/" +
-      "tensorflow/tensorflow/tensorflow/go/core" +
-      "/framework/graph_go_proto\370\001\001b\006proto3"
-    };
-    descriptor = com.google.protobuf.Descriptors.FileDescriptor
-      .internalBuildGeneratedFileFrom(descriptorData,
-        new com.google.protobuf.Descriptors.FileDescriptor[] {
-          org.tensorflow.proto.framework.FunctionProtos.getDescriptor(),
-          org.tensorflow.proto.framework.NodeProto.getDescriptor(),
-          org.tensorflow.proto.framework.VersionsProtos.getDescriptor(),
-        });
-    internal_static_tensorflow_GraphDef_descriptor =
-      getDescriptor().getMessageTypes().get(0);
-    internal_static_tensorflow_GraphDef_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_GraphDef_descriptor,
-        new java.lang.String[] { "Node", "Versions", "Version", "Library", });
-    org.tensorflow.proto.framework.FunctionProtos.getDescriptor();
-    org.tensorflow.proto.framework.NodeProto.getDescriptor();
-    org.tensorflow.proto.framework.VersionsProtos.getDescriptor();
-  }
-
-  // @@protoc_insertion_point(outer_class_scope)
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphTransferConstNodeInfo.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphTransferConstNodeInfo.java
deleted file mode 100644
index a1564a2090a..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphTransferConstNodeInfo.java
+++ /dev/null
@@ -1,912 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/framework/graph_transfer_info.proto
-
-package org.tensorflow.proto.framework;
-
-/**
- * Protobuf type {@code tensorflow.GraphTransferConstNodeInfo}
- */
-public  final class GraphTransferConstNodeInfo extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.GraphTransferConstNodeInfo)
-    GraphTransferConstNodeInfoOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use GraphTransferConstNodeInfo.newBuilder() to construct.
-  private GraphTransferConstNodeInfo(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private GraphTransferConstNodeInfo() {
-    name_ = "";
-    shape_ = emptyLongList();
-    data_ = com.google.protobuf.ByteString.EMPTY;
-    dtype_ = 0;
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new GraphTransferConstNodeInfo();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private GraphTransferConstNodeInfo(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    int mutable_bitField0_ = 0;
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 10: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            name_ = s;
-            break;
-          }
-          case 16: {
-
-            nodeId_ = input.readInt32();
-            break;
-          }
-          case 24: {
-            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
-              shape_ = newLongList();
-              mutable_bitField0_ |= 0x00000001;
-            }
-            shape_.addLong(input.readInt64());
-            break;
-          }
-          case 26: {
-            int length = input.readRawVarint32();
-            int limit = input.pushLimit(length);
-            if (!((mutable_bitField0_ & 0x00000001) != 0) && input.getBytesUntilLimit() > 0) {
-              shape_ = newLongList();
-              mutable_bitField0_ |= 0x00000001;
-            }
-            while (input.getBytesUntilLimit() > 0) {
-              shape_.addLong(input.readInt64());
-            }
-            input.popLimit(limit);
-            break;
-          }
-          case 34: {
-
-            data_ = input.readBytes();
-            break;
-          }
-          case 40: {
-            int rawValue = input.readEnum();
-
-            dtype_ = rawValue;
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      if (((mutable_bitField0_ & 0x00000001) != 0)) {
-        shape_.makeImmutable(); // C
-      }
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.framework.GraphTransferInfoProto.internal_static_tensorflow_GraphTransferConstNodeInfo_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.framework.GraphTransferInfoProto.internal_static_tensorflow_GraphTransferConstNodeInfo_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.framework.GraphTransferConstNodeInfo.class, org.tensorflow.proto.framework.GraphTransferConstNodeInfo.Builder.class);
-  }
-
-  public static final int NAME_FIELD_NUMBER = 1;
-  private volatile java.lang.Object name_;
-  /**
-   * <code>string name = 1;</code>
-   */
-  public java.lang.String getName() {
-    java.lang.Object ref = name_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      name_ = s;
-      return s;
-    }
-  }
-  /**
-   * <code>string name = 1;</code>
-   */
-  public com.google.protobuf.ByteString
-      getNameBytes() {
-    java.lang.Object ref = name_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      name_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int NODE_ID_FIELD_NUMBER = 2;
-  private int nodeId_;
-  /**
-   * <code>int32 node_id = 2;</code>
-   */
-  public int getNodeId() {
-    return nodeId_;
-  }
-
-  public static final int SHAPE_FIELD_NUMBER = 3;
-  private com.google.protobuf.Internal.LongList shape_;
-  /**
-   * <code>repeated int64 shape = 3;</code>
-   */
-  public java.util.List<java.lang.Long>
-      getShapeList() {
-    return shape_;
-  }
-  /**
-   * <code>repeated int64 shape = 3;</code>
-   */
-  public int getShapeCount() {
-    return shape_.size();
-  }
-  /**
-   * <code>repeated int64 shape = 3;</code>
-   */
-  public long getShape(int index) {
-    return shape_.getLong(index);
-  }
-  private int shapeMemoizedSerializedSize = -1;
-
-  public static final int DATA_FIELD_NUMBER = 4;
-  private com.google.protobuf.ByteString data_;
-  /**
-   * <code>bytes data = 4;</code>
-   */
-  public com.google.protobuf.ByteString getData() {
-    return data_;
-  }
-
-  public static final int DTYPE_FIELD_NUMBER = 5;
-  private int dtype_;
-  /**
-   * <code>.tensorflow.DataType dtype = 5;</code>
-   */
-  public int getDtypeValue() {
-    return dtype_;
-  }
-  /**
-   * <code>.tensorflow.DataType dtype = 5;</code>
-   */
-  public org.tensorflow.proto.framework.DataType getDtype() {
-    @SuppressWarnings("deprecation")
-    org.tensorflow.proto.framework.DataType result = org.tensorflow.proto.framework.DataType.valueOf(dtype_);
-    return result == null ? org.tensorflow.proto.framework.DataType.UNRECOGNIZED : result;
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    getSerializedSize();
-    if (!getNameBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 1, name_);
-    }
-    if (nodeId_ != 0) {
-      output.writeInt32(2, nodeId_);
-    }
-    if (getShapeList().size() > 0) {
-      output.writeUInt32NoTag(26);
-      output.writeUInt32NoTag(shapeMemoizedSerializedSize);
-    }
-    for (int i = 0; i < shape_.size(); i++) {
-      output.writeInt64NoTag(shape_.getLong(i));
-    }
-    if (!data_.isEmpty()) {
-      output.writeBytes(4, data_);
-    }
-    if (dtype_ != org.tensorflow.proto.framework.DataType.DT_INVALID.getNumber()) {
-      output.writeEnum(5, dtype_);
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (!getNameBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, name_);
-    }
-    if (nodeId_ != 0) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt32Size(2, nodeId_);
-    }
-    {
-      int dataSize = 0;
-      for (int i = 0; i < shape_.size(); i++) {
-        dataSize += com.google.protobuf.CodedOutputStream
-          .computeInt64SizeNoTag(shape_.getLong(i));
-      }
-      size += dataSize;
-      if (!getShapeList().isEmpty()) {
-        size += 1;
-        size += com.google.protobuf.CodedOutputStream
-            .computeInt32SizeNoTag(dataSize);
-      }
-      shapeMemoizedSerializedSize = dataSize;
-    }
-    if (!data_.isEmpty()) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeBytesSize(4, data_);
-    }
-    if (dtype_ != org.tensorflow.proto.framework.DataType.DT_INVALID.getNumber()) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeEnumSize(5, dtype_);
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.framework.GraphTransferConstNodeInfo)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.framework.GraphTransferConstNodeInfo other = (org.tensorflow.proto.framework.GraphTransferConstNodeInfo) obj;
-
-    if (!getName()
-        .equals(other.getName())) return false;
-    if (getNodeId()
-        != other.getNodeId()) return false;
-    if (!getShapeList()
-        .equals(other.getShapeList())) return false;
-    if (!getData()
-        .equals(other.getData())) return false;
-    if (dtype_ != other.dtype_) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    hash = (37 * hash) + NAME_FIELD_NUMBER;
-    hash = (53 * hash) + getName().hashCode();
-    hash = (37 * hash) + NODE_ID_FIELD_NUMBER;
-    hash = (53 * hash) + getNodeId();
-    if (getShapeCount() > 0) {
-      hash = (37 * hash) + SHAPE_FIELD_NUMBER;
-      hash = (53 * hash) + getShapeList().hashCode();
-    }
-    hash = (37 * hash) + DATA_FIELD_NUMBER;
-    hash = (53 * hash) + getData().hashCode();
-    hash = (37 * hash) + DTYPE_FIELD_NUMBER;
-    hash = (53 * hash) + dtype_;
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.framework.GraphTransferConstNodeInfo parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.GraphTransferConstNodeInfo parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.GraphTransferConstNodeInfo parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.GraphTransferConstNodeInfo parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.GraphTransferConstNodeInfo parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.GraphTransferConstNodeInfo parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.GraphTransferConstNodeInfo parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.GraphTransferConstNodeInfo parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.GraphTransferConstNodeInfo parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.GraphTransferConstNodeInfo parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.GraphTransferConstNodeInfo parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.GraphTransferConstNodeInfo parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.framework.GraphTransferConstNodeInfo prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * Protobuf type {@code tensorflow.GraphTransferConstNodeInfo}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.GraphTransferConstNodeInfo)
-      org.tensorflow.proto.framework.GraphTransferConstNodeInfoOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.GraphTransferInfoProto.internal_static_tensorflow_GraphTransferConstNodeInfo_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.GraphTransferInfoProto.internal_static_tensorflow_GraphTransferConstNodeInfo_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.GraphTransferConstNodeInfo.class, org.tensorflow.proto.framework.GraphTransferConstNodeInfo.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.framework.GraphTransferConstNodeInfo.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      name_ = "";
-
-      nodeId_ = 0;
-
-      shape_ = emptyLongList();
-      bitField0_ = (bitField0_ & ~0x00000001);
-      data_ = com.google.protobuf.ByteString.EMPTY;
-
-      dtype_ = 0;
-
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.framework.GraphTransferInfoProto.internal_static_tensorflow_GraphTransferConstNodeInfo_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.GraphTransferConstNodeInfo getDefaultInstanceForType() {
-      return org.tensorflow.proto.framework.GraphTransferConstNodeInfo.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.GraphTransferConstNodeInfo build() {
-      org.tensorflow.proto.framework.GraphTransferConstNodeInfo result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.GraphTransferConstNodeInfo buildPartial() {
-      org.tensorflow.proto.framework.GraphTransferConstNodeInfo result = new org.tensorflow.proto.framework.GraphTransferConstNodeInfo(this);
-      int from_bitField0_ = bitField0_;
-      result.name_ = name_;
-      result.nodeId_ = nodeId_;
-      if (((bitField0_ & 0x00000001) != 0)) {
-        shape_.makeImmutable();
-        bitField0_ = (bitField0_ & ~0x00000001);
-      }
-      result.shape_ = shape_;
-      result.data_ = data_;
-      result.dtype_ = dtype_;
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.framework.GraphTransferConstNodeInfo) {
-        return mergeFrom((org.tensorflow.proto.framework.GraphTransferConstNodeInfo)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.framework.GraphTransferConstNodeInfo other) {
-      if (other == org.tensorflow.proto.framework.GraphTransferConstNodeInfo.getDefaultInstance()) return this;
-      if (!other.getName().isEmpty()) {
-        name_ = other.name_;
-        onChanged();
-      }
-      if (other.getNodeId() != 0) {
-        setNodeId(other.getNodeId());
-      }
-      if (!other.shape_.isEmpty()) {
-        if (shape_.isEmpty()) {
-          shape_ = other.shape_;
-          bitField0_ = (bitField0_ & ~0x00000001);
-        } else {
-          ensureShapeIsMutable();
-          shape_.addAll(other.shape_);
-        }
-        onChanged();
-      }
-      if (other.getData() != com.google.protobuf.ByteString.EMPTY) {
-        setData(other.getData());
-      }
-      if (other.dtype_ != 0) {
-        setDtypeValue(other.getDtypeValue());
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.framework.GraphTransferConstNodeInfo parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.framework.GraphTransferConstNodeInfo) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-    private int bitField0_;
-
-    private java.lang.Object name_ = "";
-    /**
-     * <code>string name = 1;</code>
-     */
-    public java.lang.String getName() {
-      java.lang.Object ref = name_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        name_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <code>string name = 1;</code>
-     */
-    public com.google.protobuf.ByteString
-        getNameBytes() {
-      java.lang.Object ref = name_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        name_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <code>string name = 1;</code>
-     */
-    public Builder setName(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      name_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>string name = 1;</code>
-     */
-    public Builder clearName() {
-      
-      name_ = getDefaultInstance().getName();
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>string name = 1;</code>
-     */
-    public Builder setNameBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      name_ = value;
-      onChanged();
-      return this;
-    }
-
-    private int nodeId_ ;
-    /**
-     * <code>int32 node_id = 2;</code>
-     */
-    public int getNodeId() {
-      return nodeId_;
-    }
-    /**
-     * <code>int32 node_id = 2;</code>
-     */
-    public Builder setNodeId(int value) {
-      
-      nodeId_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>int32 node_id = 2;</code>
-     */
-    public Builder clearNodeId() {
-      
-      nodeId_ = 0;
-      onChanged();
-      return this;
-    }
-
-    private com.google.protobuf.Internal.LongList shape_ = emptyLongList();
-    private void ensureShapeIsMutable() {
-      if (!((bitField0_ & 0x00000001) != 0)) {
-        shape_ = mutableCopy(shape_);
-        bitField0_ |= 0x00000001;
-       }
-    }
-    /**
-     * <code>repeated int64 shape = 3;</code>
-     */
-    public java.util.List<java.lang.Long>
-        getShapeList() {
-      return ((bitField0_ & 0x00000001) != 0) ?
-               java.util.Collections.unmodifiableList(shape_) : shape_;
-    }
-    /**
-     * <code>repeated int64 shape = 3;</code>
-     */
-    public int getShapeCount() {
-      return shape_.size();
-    }
-    /**
-     * <code>repeated int64 shape = 3;</code>
-     */
-    public long getShape(int index) {
-      return shape_.getLong(index);
-    }
-    /**
-     * <code>repeated int64 shape = 3;</code>
-     */
-    public Builder setShape(
-        int index, long value) {
-      ensureShapeIsMutable();
-      shape_.setLong(index, value);
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>repeated int64 shape = 3;</code>
-     */
-    public Builder addShape(long value) {
-      ensureShapeIsMutable();
-      shape_.addLong(value);
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>repeated int64 shape = 3;</code>
-     */
-    public Builder addAllShape(
-        java.lang.Iterable<? extends java.lang.Long> values) {
-      ensureShapeIsMutable();
-      com.google.protobuf.AbstractMessageLite.Builder.addAll(
-          values, shape_);
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>repeated int64 shape = 3;</code>
-     */
-    public Builder clearShape() {
-      shape_ = emptyLongList();
-      bitField0_ = (bitField0_ & ~0x00000001);
-      onChanged();
-      return this;
-    }
-
-    private com.google.protobuf.ByteString data_ = com.google.protobuf.ByteString.EMPTY;
-    /**
-     * <code>bytes data = 4;</code>
-     */
-    public com.google.protobuf.ByteString getData() {
-      return data_;
-    }
-    /**
-     * <code>bytes data = 4;</code>
-     */
-    public Builder setData(com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      data_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>bytes data = 4;</code>
-     */
-    public Builder clearData() {
-      
-      data_ = getDefaultInstance().getData();
-      onChanged();
-      return this;
-    }
-
-    private int dtype_ = 0;
-    /**
-     * <code>.tensorflow.DataType dtype = 5;</code>
-     */
-    public int getDtypeValue() {
-      return dtype_;
-    }
-    /**
-     * <code>.tensorflow.DataType dtype = 5;</code>
-     */
-    public Builder setDtypeValue(int value) {
-      dtype_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>.tensorflow.DataType dtype = 5;</code>
-     */
-    public org.tensorflow.proto.framework.DataType getDtype() {
-      @SuppressWarnings("deprecation")
-      org.tensorflow.proto.framework.DataType result = org.tensorflow.proto.framework.DataType.valueOf(dtype_);
-      return result == null ? org.tensorflow.proto.framework.DataType.UNRECOGNIZED : result;
-    }
-    /**
-     * <code>.tensorflow.DataType dtype = 5;</code>
-     */
-    public Builder setDtype(org.tensorflow.proto.framework.DataType value) {
-      if (value == null) {
-        throw new NullPointerException();
-      }
-      
-      dtype_ = value.getNumber();
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>.tensorflow.DataType dtype = 5;</code>
-     */
-    public Builder clearDtype() {
-      
-      dtype_ = 0;
-      onChanged();
-      return this;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.GraphTransferConstNodeInfo)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.GraphTransferConstNodeInfo)
-  private static final org.tensorflow.proto.framework.GraphTransferConstNodeInfo DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.GraphTransferConstNodeInfo();
-  }
-
-  public static org.tensorflow.proto.framework.GraphTransferConstNodeInfo getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<GraphTransferConstNodeInfo>
-      PARSER = new com.google.protobuf.AbstractParser<GraphTransferConstNodeInfo>() {
-    @java.lang.Override
-    public GraphTransferConstNodeInfo parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new GraphTransferConstNodeInfo(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<GraphTransferConstNodeInfo> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<GraphTransferConstNodeInfo> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.framework.GraphTransferConstNodeInfo getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphTransferConstNodeInfoOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphTransferConstNodeInfoOrBuilder.java
deleted file mode 100644
index ec0b9e60f90..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphTransferConstNodeInfoOrBuilder.java
+++ /dev/null
@@ -1,51 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/framework/graph_transfer_info.proto
-
-package org.tensorflow.proto.framework;
-
-public interface GraphTransferConstNodeInfoOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.GraphTransferConstNodeInfo)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <code>string name = 1;</code>
-   */
-  java.lang.String getName();
-  /**
-   * <code>string name = 1;</code>
-   */
-  com.google.protobuf.ByteString
-      getNameBytes();
-
-  /**
-   * <code>int32 node_id = 2;</code>
-   */
-  int getNodeId();
-
-  /**
-   * <code>repeated int64 shape = 3;</code>
-   */
-  java.util.List<java.lang.Long> getShapeList();
-  /**
-   * <code>repeated int64 shape = 3;</code>
-   */
-  int getShapeCount();
-  /**
-   * <code>repeated int64 shape = 3;</code>
-   */
-  long getShape(int index);
-
-  /**
-   * <code>bytes data = 4;</code>
-   */
-  com.google.protobuf.ByteString getData();
-
-  /**
-   * <code>.tensorflow.DataType dtype = 5;</code>
-   */
-  int getDtypeValue();
-  /**
-   * <code>.tensorflow.DataType dtype = 5;</code>
-   */
-  org.tensorflow.proto.framework.DataType getDtype();
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphTransferGraphInputNodeInfo.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphTransferGraphInputNodeInfo.java
deleted file mode 100644
index b4d12c201b5..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphTransferGraphInputNodeInfo.java
+++ /dev/null
@@ -1,794 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/framework/graph_transfer_info.proto
-
-package org.tensorflow.proto.framework;
-
-/**
- * Protobuf type {@code tensorflow.GraphTransferGraphInputNodeInfo}
- */
-public  final class GraphTransferGraphInputNodeInfo extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.GraphTransferGraphInputNodeInfo)
-    GraphTransferGraphInputNodeInfoOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use GraphTransferGraphInputNodeInfo.newBuilder() to construct.
-  private GraphTransferGraphInputNodeInfo(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private GraphTransferGraphInputNodeInfo() {
-    name_ = "";
-    shape_ = emptyLongList();
-    dtype_ = 0;
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new GraphTransferGraphInputNodeInfo();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private GraphTransferGraphInputNodeInfo(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    int mutable_bitField0_ = 0;
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 10: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            name_ = s;
-            break;
-          }
-          case 16: {
-            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
-              shape_ = newLongList();
-              mutable_bitField0_ |= 0x00000001;
-            }
-            shape_.addLong(input.readInt64());
-            break;
-          }
-          case 18: {
-            int length = input.readRawVarint32();
-            int limit = input.pushLimit(length);
-            if (!((mutable_bitField0_ & 0x00000001) != 0) && input.getBytesUntilLimit() > 0) {
-              shape_ = newLongList();
-              mutable_bitField0_ |= 0x00000001;
-            }
-            while (input.getBytesUntilLimit() > 0) {
-              shape_.addLong(input.readInt64());
-            }
-            input.popLimit(limit);
-            break;
-          }
-          case 24: {
-            int rawValue = input.readEnum();
-
-            dtype_ = rawValue;
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      if (((mutable_bitField0_ & 0x00000001) != 0)) {
-        shape_.makeImmutable(); // C
-      }
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.framework.GraphTransferInfoProto.internal_static_tensorflow_GraphTransferGraphInputNodeInfo_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.framework.GraphTransferInfoProto.internal_static_tensorflow_GraphTransferGraphInputNodeInfo_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfo.class, org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfo.Builder.class);
-  }
-
-  public static final int NAME_FIELD_NUMBER = 1;
-  private volatile java.lang.Object name_;
-  /**
-   * <code>string name = 1;</code>
-   */
-  public java.lang.String getName() {
-    java.lang.Object ref = name_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      name_ = s;
-      return s;
-    }
-  }
-  /**
-   * <code>string name = 1;</code>
-   */
-  public com.google.protobuf.ByteString
-      getNameBytes() {
-    java.lang.Object ref = name_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      name_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int SHAPE_FIELD_NUMBER = 2;
-  private com.google.protobuf.Internal.LongList shape_;
-  /**
-   * <code>repeated int64 shape = 2;</code>
-   */
-  public java.util.List<java.lang.Long>
-      getShapeList() {
-    return shape_;
-  }
-  /**
-   * <code>repeated int64 shape = 2;</code>
-   */
-  public int getShapeCount() {
-    return shape_.size();
-  }
-  /**
-   * <code>repeated int64 shape = 2;</code>
-   */
-  public long getShape(int index) {
-    return shape_.getLong(index);
-  }
-  private int shapeMemoizedSerializedSize = -1;
-
-  public static final int DTYPE_FIELD_NUMBER = 3;
-  private int dtype_;
-  /**
-   * <code>.tensorflow.DataType dtype = 3;</code>
-   */
-  public int getDtypeValue() {
-    return dtype_;
-  }
-  /**
-   * <code>.tensorflow.DataType dtype = 3;</code>
-   */
-  public org.tensorflow.proto.framework.DataType getDtype() {
-    @SuppressWarnings("deprecation")
-    org.tensorflow.proto.framework.DataType result = org.tensorflow.proto.framework.DataType.valueOf(dtype_);
-    return result == null ? org.tensorflow.proto.framework.DataType.UNRECOGNIZED : result;
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    getSerializedSize();
-    if (!getNameBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 1, name_);
-    }
-    if (getShapeList().size() > 0) {
-      output.writeUInt32NoTag(18);
-      output.writeUInt32NoTag(shapeMemoizedSerializedSize);
-    }
-    for (int i = 0; i < shape_.size(); i++) {
-      output.writeInt64NoTag(shape_.getLong(i));
-    }
-    if (dtype_ != org.tensorflow.proto.framework.DataType.DT_INVALID.getNumber()) {
-      output.writeEnum(3, dtype_);
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (!getNameBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, name_);
-    }
-    {
-      int dataSize = 0;
-      for (int i = 0; i < shape_.size(); i++) {
-        dataSize += com.google.protobuf.CodedOutputStream
-          .computeInt64SizeNoTag(shape_.getLong(i));
-      }
-      size += dataSize;
-      if (!getShapeList().isEmpty()) {
-        size += 1;
-        size += com.google.protobuf.CodedOutputStream
-            .computeInt32SizeNoTag(dataSize);
-      }
-      shapeMemoizedSerializedSize = dataSize;
-    }
-    if (dtype_ != org.tensorflow.proto.framework.DataType.DT_INVALID.getNumber()) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeEnumSize(3, dtype_);
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfo)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfo other = (org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfo) obj;
-
-    if (!getName()
-        .equals(other.getName())) return false;
-    if (!getShapeList()
-        .equals(other.getShapeList())) return false;
-    if (dtype_ != other.dtype_) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    hash = (37 * hash) + NAME_FIELD_NUMBER;
-    hash = (53 * hash) + getName().hashCode();
-    if (getShapeCount() > 0) {
-      hash = (37 * hash) + SHAPE_FIELD_NUMBER;
-      hash = (53 * hash) + getShapeList().hashCode();
-    }
-    hash = (37 * hash) + DTYPE_FIELD_NUMBER;
-    hash = (53 * hash) + dtype_;
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfo parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfo parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfo parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfo parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfo parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfo parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfo parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfo parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfo parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfo parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfo parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfo parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfo prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * Protobuf type {@code tensorflow.GraphTransferGraphInputNodeInfo}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.GraphTransferGraphInputNodeInfo)
-      org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfoOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.GraphTransferInfoProto.internal_static_tensorflow_GraphTransferGraphInputNodeInfo_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.GraphTransferInfoProto.internal_static_tensorflow_GraphTransferGraphInputNodeInfo_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfo.class, org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfo.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfo.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      name_ = "";
-
-      shape_ = emptyLongList();
-      bitField0_ = (bitField0_ & ~0x00000001);
-      dtype_ = 0;
-
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.framework.GraphTransferInfoProto.internal_static_tensorflow_GraphTransferGraphInputNodeInfo_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfo getDefaultInstanceForType() {
-      return org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfo.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfo build() {
-      org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfo result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfo buildPartial() {
-      org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfo result = new org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfo(this);
-      int from_bitField0_ = bitField0_;
-      result.name_ = name_;
-      if (((bitField0_ & 0x00000001) != 0)) {
-        shape_.makeImmutable();
-        bitField0_ = (bitField0_ & ~0x00000001);
-      }
-      result.shape_ = shape_;
-      result.dtype_ = dtype_;
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfo) {
-        return mergeFrom((org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfo)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfo other) {
-      if (other == org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfo.getDefaultInstance()) return this;
-      if (!other.getName().isEmpty()) {
-        name_ = other.name_;
-        onChanged();
-      }
-      if (!other.shape_.isEmpty()) {
-        if (shape_.isEmpty()) {
-          shape_ = other.shape_;
-          bitField0_ = (bitField0_ & ~0x00000001);
-        } else {
-          ensureShapeIsMutable();
-          shape_.addAll(other.shape_);
-        }
-        onChanged();
-      }
-      if (other.dtype_ != 0) {
-        setDtypeValue(other.getDtypeValue());
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfo parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfo) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-    private int bitField0_;
-
-    private java.lang.Object name_ = "";
-    /**
-     * <code>string name = 1;</code>
-     */
-    public java.lang.String getName() {
-      java.lang.Object ref = name_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        name_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <code>string name = 1;</code>
-     */
-    public com.google.protobuf.ByteString
-        getNameBytes() {
-      java.lang.Object ref = name_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        name_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <code>string name = 1;</code>
-     */
-    public Builder setName(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      name_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>string name = 1;</code>
-     */
-    public Builder clearName() {
-      
-      name_ = getDefaultInstance().getName();
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>string name = 1;</code>
-     */
-    public Builder setNameBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      name_ = value;
-      onChanged();
-      return this;
-    }
-
-    private com.google.protobuf.Internal.LongList shape_ = emptyLongList();
-    private void ensureShapeIsMutable() {
-      if (!((bitField0_ & 0x00000001) != 0)) {
-        shape_ = mutableCopy(shape_);
-        bitField0_ |= 0x00000001;
-       }
-    }
-    /**
-     * <code>repeated int64 shape = 2;</code>
-     */
-    public java.util.List<java.lang.Long>
-        getShapeList() {
-      return ((bitField0_ & 0x00000001) != 0) ?
-               java.util.Collections.unmodifiableList(shape_) : shape_;
-    }
-    /**
-     * <code>repeated int64 shape = 2;</code>
-     */
-    public int getShapeCount() {
-      return shape_.size();
-    }
-    /**
-     * <code>repeated int64 shape = 2;</code>
-     */
-    public long getShape(int index) {
-      return shape_.getLong(index);
-    }
-    /**
-     * <code>repeated int64 shape = 2;</code>
-     */
-    public Builder setShape(
-        int index, long value) {
-      ensureShapeIsMutable();
-      shape_.setLong(index, value);
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>repeated int64 shape = 2;</code>
-     */
-    public Builder addShape(long value) {
-      ensureShapeIsMutable();
-      shape_.addLong(value);
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>repeated int64 shape = 2;</code>
-     */
-    public Builder addAllShape(
-        java.lang.Iterable<? extends java.lang.Long> values) {
-      ensureShapeIsMutable();
-      com.google.protobuf.AbstractMessageLite.Builder.addAll(
-          values, shape_);
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>repeated int64 shape = 2;</code>
-     */
-    public Builder clearShape() {
-      shape_ = emptyLongList();
-      bitField0_ = (bitField0_ & ~0x00000001);
-      onChanged();
-      return this;
-    }
-
-    private int dtype_ = 0;
-    /**
-     * <code>.tensorflow.DataType dtype = 3;</code>
-     */
-    public int getDtypeValue() {
-      return dtype_;
-    }
-    /**
-     * <code>.tensorflow.DataType dtype = 3;</code>
-     */
-    public Builder setDtypeValue(int value) {
-      dtype_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>.tensorflow.DataType dtype = 3;</code>
-     */
-    public org.tensorflow.proto.framework.DataType getDtype() {
-      @SuppressWarnings("deprecation")
-      org.tensorflow.proto.framework.DataType result = org.tensorflow.proto.framework.DataType.valueOf(dtype_);
-      return result == null ? org.tensorflow.proto.framework.DataType.UNRECOGNIZED : result;
-    }
-    /**
-     * <code>.tensorflow.DataType dtype = 3;</code>
-     */
-    public Builder setDtype(org.tensorflow.proto.framework.DataType value) {
-      if (value == null) {
-        throw new NullPointerException();
-      }
-      
-      dtype_ = value.getNumber();
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>.tensorflow.DataType dtype = 3;</code>
-     */
-    public Builder clearDtype() {
-      
-      dtype_ = 0;
-      onChanged();
-      return this;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.GraphTransferGraphInputNodeInfo)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.GraphTransferGraphInputNodeInfo)
-  private static final org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfo DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfo();
-  }
-
-  public static org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfo getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<GraphTransferGraphInputNodeInfo>
-      PARSER = new com.google.protobuf.AbstractParser<GraphTransferGraphInputNodeInfo>() {
-    @java.lang.Override
-    public GraphTransferGraphInputNodeInfo parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new GraphTransferGraphInputNodeInfo(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<GraphTransferGraphInputNodeInfo> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<GraphTransferGraphInputNodeInfo> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfo getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphTransferGraphInputNodeInfoOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphTransferGraphInputNodeInfoOrBuilder.java
deleted file mode 100644
index c121353b111..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphTransferGraphInputNodeInfoOrBuilder.java
+++ /dev/null
@@ -1,41 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/framework/graph_transfer_info.proto
-
-package org.tensorflow.proto.framework;
-
-public interface GraphTransferGraphInputNodeInfoOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.GraphTransferGraphInputNodeInfo)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <code>string name = 1;</code>
-   */
-  java.lang.String getName();
-  /**
-   * <code>string name = 1;</code>
-   */
-  com.google.protobuf.ByteString
-      getNameBytes();
-
-  /**
-   * <code>repeated int64 shape = 2;</code>
-   */
-  java.util.List<java.lang.Long> getShapeList();
-  /**
-   * <code>repeated int64 shape = 2;</code>
-   */
-  int getShapeCount();
-  /**
-   * <code>repeated int64 shape = 2;</code>
-   */
-  long getShape(int index);
-
-  /**
-   * <code>.tensorflow.DataType dtype = 3;</code>
-   */
-  int getDtypeValue();
-  /**
-   * <code>.tensorflow.DataType dtype = 3;</code>
-   */
-  org.tensorflow.proto.framework.DataType getDtype();
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphTransferGraphOutputNodeInfo.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphTransferGraphOutputNodeInfo.java
deleted file mode 100644
index 78bf270776e..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphTransferGraphOutputNodeInfo.java
+++ /dev/null
@@ -1,794 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/framework/graph_transfer_info.proto
-
-package org.tensorflow.proto.framework;
-
-/**
- * Protobuf type {@code tensorflow.GraphTransferGraphOutputNodeInfo}
- */
-public  final class GraphTransferGraphOutputNodeInfo extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.GraphTransferGraphOutputNodeInfo)
-    GraphTransferGraphOutputNodeInfoOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use GraphTransferGraphOutputNodeInfo.newBuilder() to construct.
-  private GraphTransferGraphOutputNodeInfo(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private GraphTransferGraphOutputNodeInfo() {
-    name_ = "";
-    shape_ = emptyLongList();
-    dtype_ = 0;
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new GraphTransferGraphOutputNodeInfo();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private GraphTransferGraphOutputNodeInfo(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    int mutable_bitField0_ = 0;
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 10: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            name_ = s;
-            break;
-          }
-          case 16: {
-            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
-              shape_ = newLongList();
-              mutable_bitField0_ |= 0x00000001;
-            }
-            shape_.addLong(input.readInt64());
-            break;
-          }
-          case 18: {
-            int length = input.readRawVarint32();
-            int limit = input.pushLimit(length);
-            if (!((mutable_bitField0_ & 0x00000001) != 0) && input.getBytesUntilLimit() > 0) {
-              shape_ = newLongList();
-              mutable_bitField0_ |= 0x00000001;
-            }
-            while (input.getBytesUntilLimit() > 0) {
-              shape_.addLong(input.readInt64());
-            }
-            input.popLimit(limit);
-            break;
-          }
-          case 24: {
-            int rawValue = input.readEnum();
-
-            dtype_ = rawValue;
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      if (((mutable_bitField0_ & 0x00000001) != 0)) {
-        shape_.makeImmutable(); // C
-      }
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.framework.GraphTransferInfoProto.internal_static_tensorflow_GraphTransferGraphOutputNodeInfo_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.framework.GraphTransferInfoProto.internal_static_tensorflow_GraphTransferGraphOutputNodeInfo_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfo.class, org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfo.Builder.class);
-  }
-
-  public static final int NAME_FIELD_NUMBER = 1;
-  private volatile java.lang.Object name_;
-  /**
-   * <code>string name = 1;</code>
-   */
-  public java.lang.String getName() {
-    java.lang.Object ref = name_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      name_ = s;
-      return s;
-    }
-  }
-  /**
-   * <code>string name = 1;</code>
-   */
-  public com.google.protobuf.ByteString
-      getNameBytes() {
-    java.lang.Object ref = name_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      name_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int SHAPE_FIELD_NUMBER = 2;
-  private com.google.protobuf.Internal.LongList shape_;
-  /**
-   * <code>repeated int64 shape = 2;</code>
-   */
-  public java.util.List<java.lang.Long>
-      getShapeList() {
-    return shape_;
-  }
-  /**
-   * <code>repeated int64 shape = 2;</code>
-   */
-  public int getShapeCount() {
-    return shape_.size();
-  }
-  /**
-   * <code>repeated int64 shape = 2;</code>
-   */
-  public long getShape(int index) {
-    return shape_.getLong(index);
-  }
-  private int shapeMemoizedSerializedSize = -1;
-
-  public static final int DTYPE_FIELD_NUMBER = 3;
-  private int dtype_;
-  /**
-   * <code>.tensorflow.DataType dtype = 3;</code>
-   */
-  public int getDtypeValue() {
-    return dtype_;
-  }
-  /**
-   * <code>.tensorflow.DataType dtype = 3;</code>
-   */
-  public org.tensorflow.proto.framework.DataType getDtype() {
-    @SuppressWarnings("deprecation")
-    org.tensorflow.proto.framework.DataType result = org.tensorflow.proto.framework.DataType.valueOf(dtype_);
-    return result == null ? org.tensorflow.proto.framework.DataType.UNRECOGNIZED : result;
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    getSerializedSize();
-    if (!getNameBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 1, name_);
-    }
-    if (getShapeList().size() > 0) {
-      output.writeUInt32NoTag(18);
-      output.writeUInt32NoTag(shapeMemoizedSerializedSize);
-    }
-    for (int i = 0; i < shape_.size(); i++) {
-      output.writeInt64NoTag(shape_.getLong(i));
-    }
-    if (dtype_ != org.tensorflow.proto.framework.DataType.DT_INVALID.getNumber()) {
-      output.writeEnum(3, dtype_);
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (!getNameBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, name_);
-    }
-    {
-      int dataSize = 0;
-      for (int i = 0; i < shape_.size(); i++) {
-        dataSize += com.google.protobuf.CodedOutputStream
-          .computeInt64SizeNoTag(shape_.getLong(i));
-      }
-      size += dataSize;
-      if (!getShapeList().isEmpty()) {
-        size += 1;
-        size += com.google.protobuf.CodedOutputStream
-            .computeInt32SizeNoTag(dataSize);
-      }
-      shapeMemoizedSerializedSize = dataSize;
-    }
-    if (dtype_ != org.tensorflow.proto.framework.DataType.DT_INVALID.getNumber()) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeEnumSize(3, dtype_);
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfo)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfo other = (org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfo) obj;
-
-    if (!getName()
-        .equals(other.getName())) return false;
-    if (!getShapeList()
-        .equals(other.getShapeList())) return false;
-    if (dtype_ != other.dtype_) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    hash = (37 * hash) + NAME_FIELD_NUMBER;
-    hash = (53 * hash) + getName().hashCode();
-    if (getShapeCount() > 0) {
-      hash = (37 * hash) + SHAPE_FIELD_NUMBER;
-      hash = (53 * hash) + getShapeList().hashCode();
-    }
-    hash = (37 * hash) + DTYPE_FIELD_NUMBER;
-    hash = (53 * hash) + dtype_;
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfo parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfo parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfo parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfo parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfo parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfo parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfo parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfo parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfo parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfo parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfo parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfo parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfo prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * Protobuf type {@code tensorflow.GraphTransferGraphOutputNodeInfo}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.GraphTransferGraphOutputNodeInfo)
-      org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfoOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.GraphTransferInfoProto.internal_static_tensorflow_GraphTransferGraphOutputNodeInfo_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.GraphTransferInfoProto.internal_static_tensorflow_GraphTransferGraphOutputNodeInfo_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfo.class, org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfo.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfo.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      name_ = "";
-
-      shape_ = emptyLongList();
-      bitField0_ = (bitField0_ & ~0x00000001);
-      dtype_ = 0;
-
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.framework.GraphTransferInfoProto.internal_static_tensorflow_GraphTransferGraphOutputNodeInfo_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfo getDefaultInstanceForType() {
-      return org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfo.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfo build() {
-      org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfo result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfo buildPartial() {
-      org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfo result = new org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfo(this);
-      int from_bitField0_ = bitField0_;
-      result.name_ = name_;
-      if (((bitField0_ & 0x00000001) != 0)) {
-        shape_.makeImmutable();
-        bitField0_ = (bitField0_ & ~0x00000001);
-      }
-      result.shape_ = shape_;
-      result.dtype_ = dtype_;
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfo) {
-        return mergeFrom((org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfo)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfo other) {
-      if (other == org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfo.getDefaultInstance()) return this;
-      if (!other.getName().isEmpty()) {
-        name_ = other.name_;
-        onChanged();
-      }
-      if (!other.shape_.isEmpty()) {
-        if (shape_.isEmpty()) {
-          shape_ = other.shape_;
-          bitField0_ = (bitField0_ & ~0x00000001);
-        } else {
-          ensureShapeIsMutable();
-          shape_.addAll(other.shape_);
-        }
-        onChanged();
-      }
-      if (other.dtype_ != 0) {
-        setDtypeValue(other.getDtypeValue());
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfo parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfo) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-    private int bitField0_;
-
-    private java.lang.Object name_ = "";
-    /**
-     * <code>string name = 1;</code>
-     */
-    public java.lang.String getName() {
-      java.lang.Object ref = name_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        name_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <code>string name = 1;</code>
-     */
-    public com.google.protobuf.ByteString
-        getNameBytes() {
-      java.lang.Object ref = name_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        name_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <code>string name = 1;</code>
-     */
-    public Builder setName(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      name_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>string name = 1;</code>
-     */
-    public Builder clearName() {
-      
-      name_ = getDefaultInstance().getName();
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>string name = 1;</code>
-     */
-    public Builder setNameBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      name_ = value;
-      onChanged();
-      return this;
-    }
-
-    private com.google.protobuf.Internal.LongList shape_ = emptyLongList();
-    private void ensureShapeIsMutable() {
-      if (!((bitField0_ & 0x00000001) != 0)) {
-        shape_ = mutableCopy(shape_);
-        bitField0_ |= 0x00000001;
-       }
-    }
-    /**
-     * <code>repeated int64 shape = 2;</code>
-     */
-    public java.util.List<java.lang.Long>
-        getShapeList() {
-      return ((bitField0_ & 0x00000001) != 0) ?
-               java.util.Collections.unmodifiableList(shape_) : shape_;
-    }
-    /**
-     * <code>repeated int64 shape = 2;</code>
-     */
-    public int getShapeCount() {
-      return shape_.size();
-    }
-    /**
-     * <code>repeated int64 shape = 2;</code>
-     */
-    public long getShape(int index) {
-      return shape_.getLong(index);
-    }
-    /**
-     * <code>repeated int64 shape = 2;</code>
-     */
-    public Builder setShape(
-        int index, long value) {
-      ensureShapeIsMutable();
-      shape_.setLong(index, value);
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>repeated int64 shape = 2;</code>
-     */
-    public Builder addShape(long value) {
-      ensureShapeIsMutable();
-      shape_.addLong(value);
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>repeated int64 shape = 2;</code>
-     */
-    public Builder addAllShape(
-        java.lang.Iterable<? extends java.lang.Long> values) {
-      ensureShapeIsMutable();
-      com.google.protobuf.AbstractMessageLite.Builder.addAll(
-          values, shape_);
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>repeated int64 shape = 2;</code>
-     */
-    public Builder clearShape() {
-      shape_ = emptyLongList();
-      bitField0_ = (bitField0_ & ~0x00000001);
-      onChanged();
-      return this;
-    }
-
-    private int dtype_ = 0;
-    /**
-     * <code>.tensorflow.DataType dtype = 3;</code>
-     */
-    public int getDtypeValue() {
-      return dtype_;
-    }
-    /**
-     * <code>.tensorflow.DataType dtype = 3;</code>
-     */
-    public Builder setDtypeValue(int value) {
-      dtype_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>.tensorflow.DataType dtype = 3;</code>
-     */
-    public org.tensorflow.proto.framework.DataType getDtype() {
-      @SuppressWarnings("deprecation")
-      org.tensorflow.proto.framework.DataType result = org.tensorflow.proto.framework.DataType.valueOf(dtype_);
-      return result == null ? org.tensorflow.proto.framework.DataType.UNRECOGNIZED : result;
-    }
-    /**
-     * <code>.tensorflow.DataType dtype = 3;</code>
-     */
-    public Builder setDtype(org.tensorflow.proto.framework.DataType value) {
-      if (value == null) {
-        throw new NullPointerException();
-      }
-      
-      dtype_ = value.getNumber();
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>.tensorflow.DataType dtype = 3;</code>
-     */
-    public Builder clearDtype() {
-      
-      dtype_ = 0;
-      onChanged();
-      return this;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.GraphTransferGraphOutputNodeInfo)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.GraphTransferGraphOutputNodeInfo)
-  private static final org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfo DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfo();
-  }
-
-  public static org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfo getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<GraphTransferGraphOutputNodeInfo>
-      PARSER = new com.google.protobuf.AbstractParser<GraphTransferGraphOutputNodeInfo>() {
-    @java.lang.Override
-    public GraphTransferGraphOutputNodeInfo parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new GraphTransferGraphOutputNodeInfo(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<GraphTransferGraphOutputNodeInfo> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<GraphTransferGraphOutputNodeInfo> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfo getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphTransferGraphOutputNodeInfoOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphTransferGraphOutputNodeInfoOrBuilder.java
deleted file mode 100644
index 3fcf3f3ae7e..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphTransferGraphOutputNodeInfoOrBuilder.java
+++ /dev/null
@@ -1,41 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/framework/graph_transfer_info.proto
-
-package org.tensorflow.proto.framework;
-
-public interface GraphTransferGraphOutputNodeInfoOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.GraphTransferGraphOutputNodeInfo)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <code>string name = 1;</code>
-   */
-  java.lang.String getName();
-  /**
-   * <code>string name = 1;</code>
-   */
-  com.google.protobuf.ByteString
-      getNameBytes();
-
-  /**
-   * <code>repeated int64 shape = 2;</code>
-   */
-  java.util.List<java.lang.Long> getShapeList();
-  /**
-   * <code>repeated int64 shape = 2;</code>
-   */
-  int getShapeCount();
-  /**
-   * <code>repeated int64 shape = 2;</code>
-   */
-  long getShape(int index);
-
-  /**
-   * <code>.tensorflow.DataType dtype = 3;</code>
-   */
-  int getDtypeValue();
-  /**
-   * <code>.tensorflow.DataType dtype = 3;</code>
-   */
-  org.tensorflow.proto.framework.DataType getDtype();
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphTransferInfo.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphTransferInfo.java
deleted file mode 100644
index f9d53390a8b..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphTransferInfo.java
+++ /dev/null
@@ -1,2795 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/framework/graph_transfer_info.proto
-
-package org.tensorflow.proto.framework;
-
-/**
- * <pre>
- * Protocol buffer representing a handle to a tensorflow resource. Handles are
- * not valid across executions, but can be serialized back and forth from within
- * a single run.
- * </pre>
- *
- * Protobuf type {@code tensorflow.GraphTransferInfo}
- */
-public  final class GraphTransferInfo extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.GraphTransferInfo)
-    GraphTransferInfoOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use GraphTransferInfo.newBuilder() to construct.
-  private GraphTransferInfo(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private GraphTransferInfo() {
-    nodeInfo_ = java.util.Collections.emptyList();
-    constNodeInfo_ = java.util.Collections.emptyList();
-    nodeInputInfo_ = java.util.Collections.emptyList();
-    nodeOutputInfo_ = java.util.Collections.emptyList();
-    graphInputNodeInfo_ = java.util.Collections.emptyList();
-    graphOutputNodeInfo_ = java.util.Collections.emptyList();
-    destination_ = 0;
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new GraphTransferInfo();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private GraphTransferInfo(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    int mutable_bitField0_ = 0;
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 10: {
-            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
-              nodeInfo_ = new java.util.ArrayList<org.tensorflow.proto.framework.GraphTransferNodeInfo>();
-              mutable_bitField0_ |= 0x00000001;
-            }
-            nodeInfo_.add(
-                input.readMessage(org.tensorflow.proto.framework.GraphTransferNodeInfo.parser(), extensionRegistry));
-            break;
-          }
-          case 18: {
-            if (!((mutable_bitField0_ & 0x00000002) != 0)) {
-              constNodeInfo_ = new java.util.ArrayList<org.tensorflow.proto.framework.GraphTransferConstNodeInfo>();
-              mutable_bitField0_ |= 0x00000002;
-            }
-            constNodeInfo_.add(
-                input.readMessage(org.tensorflow.proto.framework.GraphTransferConstNodeInfo.parser(), extensionRegistry));
-            break;
-          }
-          case 26: {
-            if (!((mutable_bitField0_ & 0x00000004) != 0)) {
-              nodeInputInfo_ = new java.util.ArrayList<org.tensorflow.proto.framework.GraphTransferNodeInputInfo>();
-              mutable_bitField0_ |= 0x00000004;
-            }
-            nodeInputInfo_.add(
-                input.readMessage(org.tensorflow.proto.framework.GraphTransferNodeInputInfo.parser(), extensionRegistry));
-            break;
-          }
-          case 34: {
-            if (!((mutable_bitField0_ & 0x00000008) != 0)) {
-              nodeOutputInfo_ = new java.util.ArrayList<org.tensorflow.proto.framework.GraphTransferNodeOutputInfo>();
-              mutable_bitField0_ |= 0x00000008;
-            }
-            nodeOutputInfo_.add(
-                input.readMessage(org.tensorflow.proto.framework.GraphTransferNodeOutputInfo.parser(), extensionRegistry));
-            break;
-          }
-          case 42: {
-            if (!((mutable_bitField0_ & 0x00000010) != 0)) {
-              graphInputNodeInfo_ = new java.util.ArrayList<org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfo>();
-              mutable_bitField0_ |= 0x00000010;
-            }
-            graphInputNodeInfo_.add(
-                input.readMessage(org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfo.parser(), extensionRegistry));
-            break;
-          }
-          case 50: {
-            if (!((mutable_bitField0_ & 0x00000020) != 0)) {
-              graphOutputNodeInfo_ = new java.util.ArrayList<org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfo>();
-              mutable_bitField0_ |= 0x00000020;
-            }
-            graphOutputNodeInfo_.add(
-                input.readMessage(org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfo.parser(), extensionRegistry));
-            break;
-          }
-          case 56: {
-            int rawValue = input.readEnum();
-
-            destination_ = rawValue;
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      if (((mutable_bitField0_ & 0x00000001) != 0)) {
-        nodeInfo_ = java.util.Collections.unmodifiableList(nodeInfo_);
-      }
-      if (((mutable_bitField0_ & 0x00000002) != 0)) {
-        constNodeInfo_ = java.util.Collections.unmodifiableList(constNodeInfo_);
-      }
-      if (((mutable_bitField0_ & 0x00000004) != 0)) {
-        nodeInputInfo_ = java.util.Collections.unmodifiableList(nodeInputInfo_);
-      }
-      if (((mutable_bitField0_ & 0x00000008) != 0)) {
-        nodeOutputInfo_ = java.util.Collections.unmodifiableList(nodeOutputInfo_);
-      }
-      if (((mutable_bitField0_ & 0x00000010) != 0)) {
-        graphInputNodeInfo_ = java.util.Collections.unmodifiableList(graphInputNodeInfo_);
-      }
-      if (((mutable_bitField0_ & 0x00000020) != 0)) {
-        graphOutputNodeInfo_ = java.util.Collections.unmodifiableList(graphOutputNodeInfo_);
-      }
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.framework.GraphTransferInfoProto.internal_static_tensorflow_GraphTransferInfo_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.framework.GraphTransferInfoProto.internal_static_tensorflow_GraphTransferInfo_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.framework.GraphTransferInfo.class, org.tensorflow.proto.framework.GraphTransferInfo.Builder.class);
-  }
-
-  /**
-   * Protobuf enum {@code tensorflow.GraphTransferInfo.Destination}
-   */
-  public enum Destination
-      implements com.google.protobuf.ProtocolMessageEnum {
-    /**
-     * <code>NOP = 0;</code>
-     */
-    NOP(0),
-    /**
-     * <code>HEXAGON = 1;</code>
-     */
-    HEXAGON(1),
-    UNRECOGNIZED(-1),
-    ;
-
-    /**
-     * <code>NOP = 0;</code>
-     */
-    public static final int NOP_VALUE = 0;
-    /**
-     * <code>HEXAGON = 1;</code>
-     */
-    public static final int HEXAGON_VALUE = 1;
-
-
-    public final int getNumber() {
-      if (this == UNRECOGNIZED) {
-        throw new java.lang.IllegalArgumentException(
-            "Can't get the number of an unknown enum value.");
-      }
-      return value;
-    }
-
-    /**
-     * @deprecated Use {@link #forNumber(int)} instead.
-     */
-    @java.lang.Deprecated
-    public static Destination valueOf(int value) {
-      return forNumber(value);
-    }
-
-    public static Destination forNumber(int value) {
-      switch (value) {
-        case 0: return NOP;
-        case 1: return HEXAGON;
-        default: return null;
-      }
-    }
-
-    public static com.google.protobuf.Internal.EnumLiteMap<Destination>
-        internalGetValueMap() {
-      return internalValueMap;
-    }
-    private static final com.google.protobuf.Internal.EnumLiteMap<
-        Destination> internalValueMap =
-          new com.google.protobuf.Internal.EnumLiteMap<Destination>() {
-            public Destination findValueByNumber(int number) {
-              return Destination.forNumber(number);
-            }
-          };
-
-    public final com.google.protobuf.Descriptors.EnumValueDescriptor
-        getValueDescriptor() {
-      return getDescriptor().getValues().get(ordinal());
-    }
-    public final com.google.protobuf.Descriptors.EnumDescriptor
-        getDescriptorForType() {
-      return getDescriptor();
-    }
-    public static final com.google.protobuf.Descriptors.EnumDescriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.GraphTransferInfo.getDescriptor().getEnumTypes().get(0);
-    }
-
-    private static final Destination[] VALUES = values();
-
-    public static Destination valueOf(
-        com.google.protobuf.Descriptors.EnumValueDescriptor desc) {
-      if (desc.getType() != getDescriptor()) {
-        throw new java.lang.IllegalArgumentException(
-          "EnumValueDescriptor is not for this type.");
-      }
-      if (desc.getIndex() == -1) {
-        return UNRECOGNIZED;
-      }
-      return VALUES[desc.getIndex()];
-    }
-
-    private final int value;
-
-    private Destination(int value) {
-      this.value = value;
-    }
-
-    // @@protoc_insertion_point(enum_scope:tensorflow.GraphTransferInfo.Destination)
-  }
-
-  public static final int NODE_INFO_FIELD_NUMBER = 1;
-  private java.util.List<org.tensorflow.proto.framework.GraphTransferNodeInfo> nodeInfo_;
-  /**
-   * <code>repeated .tensorflow.GraphTransferNodeInfo node_info = 1;</code>
-   */
-  public java.util.List<org.tensorflow.proto.framework.GraphTransferNodeInfo> getNodeInfoList() {
-    return nodeInfo_;
-  }
-  /**
-   * <code>repeated .tensorflow.GraphTransferNodeInfo node_info = 1;</code>
-   */
-  public java.util.List<? extends org.tensorflow.proto.framework.GraphTransferNodeInfoOrBuilder> 
-      getNodeInfoOrBuilderList() {
-    return nodeInfo_;
-  }
-  /**
-   * <code>repeated .tensorflow.GraphTransferNodeInfo node_info = 1;</code>
-   */
-  public int getNodeInfoCount() {
-    return nodeInfo_.size();
-  }
-  /**
-   * <code>repeated .tensorflow.GraphTransferNodeInfo node_info = 1;</code>
-   */
-  public org.tensorflow.proto.framework.GraphTransferNodeInfo getNodeInfo(int index) {
-    return nodeInfo_.get(index);
-  }
-  /**
-   * <code>repeated .tensorflow.GraphTransferNodeInfo node_info = 1;</code>
-   */
-  public org.tensorflow.proto.framework.GraphTransferNodeInfoOrBuilder getNodeInfoOrBuilder(
-      int index) {
-    return nodeInfo_.get(index);
-  }
-
-  public static final int CONST_NODE_INFO_FIELD_NUMBER = 2;
-  private java.util.List<org.tensorflow.proto.framework.GraphTransferConstNodeInfo> constNodeInfo_;
-  /**
-   * <code>repeated .tensorflow.GraphTransferConstNodeInfo const_node_info = 2;</code>
-   */
-  public java.util.List<org.tensorflow.proto.framework.GraphTransferConstNodeInfo> getConstNodeInfoList() {
-    return constNodeInfo_;
-  }
-  /**
-   * <code>repeated .tensorflow.GraphTransferConstNodeInfo const_node_info = 2;</code>
-   */
-  public java.util.List<? extends org.tensorflow.proto.framework.GraphTransferConstNodeInfoOrBuilder> 
-      getConstNodeInfoOrBuilderList() {
-    return constNodeInfo_;
-  }
-  /**
-   * <code>repeated .tensorflow.GraphTransferConstNodeInfo const_node_info = 2;</code>
-   */
-  public int getConstNodeInfoCount() {
-    return constNodeInfo_.size();
-  }
-  /**
-   * <code>repeated .tensorflow.GraphTransferConstNodeInfo const_node_info = 2;</code>
-   */
-  public org.tensorflow.proto.framework.GraphTransferConstNodeInfo getConstNodeInfo(int index) {
-    return constNodeInfo_.get(index);
-  }
-  /**
-   * <code>repeated .tensorflow.GraphTransferConstNodeInfo const_node_info = 2;</code>
-   */
-  public org.tensorflow.proto.framework.GraphTransferConstNodeInfoOrBuilder getConstNodeInfoOrBuilder(
-      int index) {
-    return constNodeInfo_.get(index);
-  }
-
-  public static final int NODE_INPUT_INFO_FIELD_NUMBER = 3;
-  private java.util.List<org.tensorflow.proto.framework.GraphTransferNodeInputInfo> nodeInputInfo_;
-  /**
-   * <code>repeated .tensorflow.GraphTransferNodeInputInfo node_input_info = 3;</code>
-   */
-  public java.util.List<org.tensorflow.proto.framework.GraphTransferNodeInputInfo> getNodeInputInfoList() {
-    return nodeInputInfo_;
-  }
-  /**
-   * <code>repeated .tensorflow.GraphTransferNodeInputInfo node_input_info = 3;</code>
-   */
-  public java.util.List<? extends org.tensorflow.proto.framework.GraphTransferNodeInputInfoOrBuilder> 
-      getNodeInputInfoOrBuilderList() {
-    return nodeInputInfo_;
-  }
-  /**
-   * <code>repeated .tensorflow.GraphTransferNodeInputInfo node_input_info = 3;</code>
-   */
-  public int getNodeInputInfoCount() {
-    return nodeInputInfo_.size();
-  }
-  /**
-   * <code>repeated .tensorflow.GraphTransferNodeInputInfo node_input_info = 3;</code>
-   */
-  public org.tensorflow.proto.framework.GraphTransferNodeInputInfo getNodeInputInfo(int index) {
-    return nodeInputInfo_.get(index);
-  }
-  /**
-   * <code>repeated .tensorflow.GraphTransferNodeInputInfo node_input_info = 3;</code>
-   */
-  public org.tensorflow.proto.framework.GraphTransferNodeInputInfoOrBuilder getNodeInputInfoOrBuilder(
-      int index) {
-    return nodeInputInfo_.get(index);
-  }
-
-  public static final int NODE_OUTPUT_INFO_FIELD_NUMBER = 4;
-  private java.util.List<org.tensorflow.proto.framework.GraphTransferNodeOutputInfo> nodeOutputInfo_;
-  /**
-   * <code>repeated .tensorflow.GraphTransferNodeOutputInfo node_output_info = 4;</code>
-   */
-  public java.util.List<org.tensorflow.proto.framework.GraphTransferNodeOutputInfo> getNodeOutputInfoList() {
-    return nodeOutputInfo_;
-  }
-  /**
-   * <code>repeated .tensorflow.GraphTransferNodeOutputInfo node_output_info = 4;</code>
-   */
-  public java.util.List<? extends org.tensorflow.proto.framework.GraphTransferNodeOutputInfoOrBuilder> 
-      getNodeOutputInfoOrBuilderList() {
-    return nodeOutputInfo_;
-  }
-  /**
-   * <code>repeated .tensorflow.GraphTransferNodeOutputInfo node_output_info = 4;</code>
-   */
-  public int getNodeOutputInfoCount() {
-    return nodeOutputInfo_.size();
-  }
-  /**
-   * <code>repeated .tensorflow.GraphTransferNodeOutputInfo node_output_info = 4;</code>
-   */
-  public org.tensorflow.proto.framework.GraphTransferNodeOutputInfo getNodeOutputInfo(int index) {
-    return nodeOutputInfo_.get(index);
-  }
-  /**
-   * <code>repeated .tensorflow.GraphTransferNodeOutputInfo node_output_info = 4;</code>
-   */
-  public org.tensorflow.proto.framework.GraphTransferNodeOutputInfoOrBuilder getNodeOutputInfoOrBuilder(
-      int index) {
-    return nodeOutputInfo_.get(index);
-  }
-
-  public static final int GRAPH_INPUT_NODE_INFO_FIELD_NUMBER = 5;
-  private java.util.List<org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfo> graphInputNodeInfo_;
-  /**
-   * <pre>
-   * Input Node parameters of transferred graph
-   * </pre>
-   *
-   * <code>repeated .tensorflow.GraphTransferGraphInputNodeInfo graph_input_node_info = 5;</code>
-   */
-  public java.util.List<org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfo> getGraphInputNodeInfoList() {
-    return graphInputNodeInfo_;
-  }
-  /**
-   * <pre>
-   * Input Node parameters of transferred graph
-   * </pre>
-   *
-   * <code>repeated .tensorflow.GraphTransferGraphInputNodeInfo graph_input_node_info = 5;</code>
-   */
-  public java.util.List<? extends org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfoOrBuilder> 
-      getGraphInputNodeInfoOrBuilderList() {
-    return graphInputNodeInfo_;
-  }
-  /**
-   * <pre>
-   * Input Node parameters of transferred graph
-   * </pre>
-   *
-   * <code>repeated .tensorflow.GraphTransferGraphInputNodeInfo graph_input_node_info = 5;</code>
-   */
-  public int getGraphInputNodeInfoCount() {
-    return graphInputNodeInfo_.size();
-  }
-  /**
-   * <pre>
-   * Input Node parameters of transferred graph
-   * </pre>
-   *
-   * <code>repeated .tensorflow.GraphTransferGraphInputNodeInfo graph_input_node_info = 5;</code>
-   */
-  public org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfo getGraphInputNodeInfo(int index) {
-    return graphInputNodeInfo_.get(index);
-  }
-  /**
-   * <pre>
-   * Input Node parameters of transferred graph
-   * </pre>
-   *
-   * <code>repeated .tensorflow.GraphTransferGraphInputNodeInfo graph_input_node_info = 5;</code>
-   */
-  public org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfoOrBuilder getGraphInputNodeInfoOrBuilder(
-      int index) {
-    return graphInputNodeInfo_.get(index);
-  }
-
-  public static final int GRAPH_OUTPUT_NODE_INFO_FIELD_NUMBER = 6;
-  private java.util.List<org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfo> graphOutputNodeInfo_;
-  /**
-   * <code>repeated .tensorflow.GraphTransferGraphOutputNodeInfo graph_output_node_info = 6;</code>
-   */
-  public java.util.List<org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfo> getGraphOutputNodeInfoList() {
-    return graphOutputNodeInfo_;
-  }
-  /**
-   * <code>repeated .tensorflow.GraphTransferGraphOutputNodeInfo graph_output_node_info = 6;</code>
-   */
-  public java.util.List<? extends org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfoOrBuilder> 
-      getGraphOutputNodeInfoOrBuilderList() {
-    return graphOutputNodeInfo_;
-  }
-  /**
-   * <code>repeated .tensorflow.GraphTransferGraphOutputNodeInfo graph_output_node_info = 6;</code>
-   */
-  public int getGraphOutputNodeInfoCount() {
-    return graphOutputNodeInfo_.size();
-  }
-  /**
-   * <code>repeated .tensorflow.GraphTransferGraphOutputNodeInfo graph_output_node_info = 6;</code>
-   */
-  public org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfo getGraphOutputNodeInfo(int index) {
-    return graphOutputNodeInfo_.get(index);
-  }
-  /**
-   * <code>repeated .tensorflow.GraphTransferGraphOutputNodeInfo graph_output_node_info = 6;</code>
-   */
-  public org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfoOrBuilder getGraphOutputNodeInfoOrBuilder(
-      int index) {
-    return graphOutputNodeInfo_.get(index);
-  }
-
-  public static final int DESTINATION_FIELD_NUMBER = 7;
-  private int destination_;
-  /**
-   * <pre>
-   * Destination of graph transfer
-   * </pre>
-   *
-   * <code>.tensorflow.GraphTransferInfo.Destination destination = 7;</code>
-   */
-  public int getDestinationValue() {
-    return destination_;
-  }
-  /**
-   * <pre>
-   * Destination of graph transfer
-   * </pre>
-   *
-   * <code>.tensorflow.GraphTransferInfo.Destination destination = 7;</code>
-   */
-  public org.tensorflow.proto.framework.GraphTransferInfo.Destination getDestination() {
-    @SuppressWarnings("deprecation")
-    org.tensorflow.proto.framework.GraphTransferInfo.Destination result = org.tensorflow.proto.framework.GraphTransferInfo.Destination.valueOf(destination_);
-    return result == null ? org.tensorflow.proto.framework.GraphTransferInfo.Destination.UNRECOGNIZED : result;
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    for (int i = 0; i < nodeInfo_.size(); i++) {
-      output.writeMessage(1, nodeInfo_.get(i));
-    }
-    for (int i = 0; i < constNodeInfo_.size(); i++) {
-      output.writeMessage(2, constNodeInfo_.get(i));
-    }
-    for (int i = 0; i < nodeInputInfo_.size(); i++) {
-      output.writeMessage(3, nodeInputInfo_.get(i));
-    }
-    for (int i = 0; i < nodeOutputInfo_.size(); i++) {
-      output.writeMessage(4, nodeOutputInfo_.get(i));
-    }
-    for (int i = 0; i < graphInputNodeInfo_.size(); i++) {
-      output.writeMessage(5, graphInputNodeInfo_.get(i));
-    }
-    for (int i = 0; i < graphOutputNodeInfo_.size(); i++) {
-      output.writeMessage(6, graphOutputNodeInfo_.get(i));
-    }
-    if (destination_ != org.tensorflow.proto.framework.GraphTransferInfo.Destination.NOP.getNumber()) {
-      output.writeEnum(7, destination_);
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    for (int i = 0; i < nodeInfo_.size(); i++) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(1, nodeInfo_.get(i));
-    }
-    for (int i = 0; i < constNodeInfo_.size(); i++) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(2, constNodeInfo_.get(i));
-    }
-    for (int i = 0; i < nodeInputInfo_.size(); i++) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(3, nodeInputInfo_.get(i));
-    }
-    for (int i = 0; i < nodeOutputInfo_.size(); i++) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(4, nodeOutputInfo_.get(i));
-    }
-    for (int i = 0; i < graphInputNodeInfo_.size(); i++) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(5, graphInputNodeInfo_.get(i));
-    }
-    for (int i = 0; i < graphOutputNodeInfo_.size(); i++) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(6, graphOutputNodeInfo_.get(i));
-    }
-    if (destination_ != org.tensorflow.proto.framework.GraphTransferInfo.Destination.NOP.getNumber()) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeEnumSize(7, destination_);
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.framework.GraphTransferInfo)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.framework.GraphTransferInfo other = (org.tensorflow.proto.framework.GraphTransferInfo) obj;
-
-    if (!getNodeInfoList()
-        .equals(other.getNodeInfoList())) return false;
-    if (!getConstNodeInfoList()
-        .equals(other.getConstNodeInfoList())) return false;
-    if (!getNodeInputInfoList()
-        .equals(other.getNodeInputInfoList())) return false;
-    if (!getNodeOutputInfoList()
-        .equals(other.getNodeOutputInfoList())) return false;
-    if (!getGraphInputNodeInfoList()
-        .equals(other.getGraphInputNodeInfoList())) return false;
-    if (!getGraphOutputNodeInfoList()
-        .equals(other.getGraphOutputNodeInfoList())) return false;
-    if (destination_ != other.destination_) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    if (getNodeInfoCount() > 0) {
-      hash = (37 * hash) + NODE_INFO_FIELD_NUMBER;
-      hash = (53 * hash) + getNodeInfoList().hashCode();
-    }
-    if (getConstNodeInfoCount() > 0) {
-      hash = (37 * hash) + CONST_NODE_INFO_FIELD_NUMBER;
-      hash = (53 * hash) + getConstNodeInfoList().hashCode();
-    }
-    if (getNodeInputInfoCount() > 0) {
-      hash = (37 * hash) + NODE_INPUT_INFO_FIELD_NUMBER;
-      hash = (53 * hash) + getNodeInputInfoList().hashCode();
-    }
-    if (getNodeOutputInfoCount() > 0) {
-      hash = (37 * hash) + NODE_OUTPUT_INFO_FIELD_NUMBER;
-      hash = (53 * hash) + getNodeOutputInfoList().hashCode();
-    }
-    if (getGraphInputNodeInfoCount() > 0) {
-      hash = (37 * hash) + GRAPH_INPUT_NODE_INFO_FIELD_NUMBER;
-      hash = (53 * hash) + getGraphInputNodeInfoList().hashCode();
-    }
-    if (getGraphOutputNodeInfoCount() > 0) {
-      hash = (37 * hash) + GRAPH_OUTPUT_NODE_INFO_FIELD_NUMBER;
-      hash = (53 * hash) + getGraphOutputNodeInfoList().hashCode();
-    }
-    hash = (37 * hash) + DESTINATION_FIELD_NUMBER;
-    hash = (53 * hash) + destination_;
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.framework.GraphTransferInfo parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.GraphTransferInfo parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.GraphTransferInfo parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.GraphTransferInfo parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.GraphTransferInfo parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.GraphTransferInfo parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.GraphTransferInfo parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.GraphTransferInfo parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.GraphTransferInfo parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.GraphTransferInfo parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.GraphTransferInfo parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.GraphTransferInfo parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.framework.GraphTransferInfo prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * <pre>
-   * Protocol buffer representing a handle to a tensorflow resource. Handles are
-   * not valid across executions, but can be serialized back and forth from within
-   * a single run.
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.GraphTransferInfo}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.GraphTransferInfo)
-      org.tensorflow.proto.framework.GraphTransferInfoOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.GraphTransferInfoProto.internal_static_tensorflow_GraphTransferInfo_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.GraphTransferInfoProto.internal_static_tensorflow_GraphTransferInfo_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.GraphTransferInfo.class, org.tensorflow.proto.framework.GraphTransferInfo.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.framework.GraphTransferInfo.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-        getNodeInfoFieldBuilder();
-        getConstNodeInfoFieldBuilder();
-        getNodeInputInfoFieldBuilder();
-        getNodeOutputInfoFieldBuilder();
-        getGraphInputNodeInfoFieldBuilder();
-        getGraphOutputNodeInfoFieldBuilder();
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      if (nodeInfoBuilder_ == null) {
-        nodeInfo_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000001);
-      } else {
-        nodeInfoBuilder_.clear();
-      }
-      if (constNodeInfoBuilder_ == null) {
-        constNodeInfo_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000002);
-      } else {
-        constNodeInfoBuilder_.clear();
-      }
-      if (nodeInputInfoBuilder_ == null) {
-        nodeInputInfo_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000004);
-      } else {
-        nodeInputInfoBuilder_.clear();
-      }
-      if (nodeOutputInfoBuilder_ == null) {
-        nodeOutputInfo_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000008);
-      } else {
-        nodeOutputInfoBuilder_.clear();
-      }
-      if (graphInputNodeInfoBuilder_ == null) {
-        graphInputNodeInfo_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000010);
-      } else {
-        graphInputNodeInfoBuilder_.clear();
-      }
-      if (graphOutputNodeInfoBuilder_ == null) {
-        graphOutputNodeInfo_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000020);
-      } else {
-        graphOutputNodeInfoBuilder_.clear();
-      }
-      destination_ = 0;
-
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.framework.GraphTransferInfoProto.internal_static_tensorflow_GraphTransferInfo_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.GraphTransferInfo getDefaultInstanceForType() {
-      return org.tensorflow.proto.framework.GraphTransferInfo.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.GraphTransferInfo build() {
-      org.tensorflow.proto.framework.GraphTransferInfo result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.GraphTransferInfo buildPartial() {
-      org.tensorflow.proto.framework.GraphTransferInfo result = new org.tensorflow.proto.framework.GraphTransferInfo(this);
-      int from_bitField0_ = bitField0_;
-      if (nodeInfoBuilder_ == null) {
-        if (((bitField0_ & 0x00000001) != 0)) {
-          nodeInfo_ = java.util.Collections.unmodifiableList(nodeInfo_);
-          bitField0_ = (bitField0_ & ~0x00000001);
-        }
-        result.nodeInfo_ = nodeInfo_;
-      } else {
-        result.nodeInfo_ = nodeInfoBuilder_.build();
-      }
-      if (constNodeInfoBuilder_ == null) {
-        if (((bitField0_ & 0x00000002) != 0)) {
-          constNodeInfo_ = java.util.Collections.unmodifiableList(constNodeInfo_);
-          bitField0_ = (bitField0_ & ~0x00000002);
-        }
-        result.constNodeInfo_ = constNodeInfo_;
-      } else {
-        result.constNodeInfo_ = constNodeInfoBuilder_.build();
-      }
-      if (nodeInputInfoBuilder_ == null) {
-        if (((bitField0_ & 0x00000004) != 0)) {
-          nodeInputInfo_ = java.util.Collections.unmodifiableList(nodeInputInfo_);
-          bitField0_ = (bitField0_ & ~0x00000004);
-        }
-        result.nodeInputInfo_ = nodeInputInfo_;
-      } else {
-        result.nodeInputInfo_ = nodeInputInfoBuilder_.build();
-      }
-      if (nodeOutputInfoBuilder_ == null) {
-        if (((bitField0_ & 0x00000008) != 0)) {
-          nodeOutputInfo_ = java.util.Collections.unmodifiableList(nodeOutputInfo_);
-          bitField0_ = (bitField0_ & ~0x00000008);
-        }
-        result.nodeOutputInfo_ = nodeOutputInfo_;
-      } else {
-        result.nodeOutputInfo_ = nodeOutputInfoBuilder_.build();
-      }
-      if (graphInputNodeInfoBuilder_ == null) {
-        if (((bitField0_ & 0x00000010) != 0)) {
-          graphInputNodeInfo_ = java.util.Collections.unmodifiableList(graphInputNodeInfo_);
-          bitField0_ = (bitField0_ & ~0x00000010);
-        }
-        result.graphInputNodeInfo_ = graphInputNodeInfo_;
-      } else {
-        result.graphInputNodeInfo_ = graphInputNodeInfoBuilder_.build();
-      }
-      if (graphOutputNodeInfoBuilder_ == null) {
-        if (((bitField0_ & 0x00000020) != 0)) {
-          graphOutputNodeInfo_ = java.util.Collections.unmodifiableList(graphOutputNodeInfo_);
-          bitField0_ = (bitField0_ & ~0x00000020);
-        }
-        result.graphOutputNodeInfo_ = graphOutputNodeInfo_;
-      } else {
-        result.graphOutputNodeInfo_ = graphOutputNodeInfoBuilder_.build();
-      }
-      result.destination_ = destination_;
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.framework.GraphTransferInfo) {
-        return mergeFrom((org.tensorflow.proto.framework.GraphTransferInfo)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.framework.GraphTransferInfo other) {
-      if (other == org.tensorflow.proto.framework.GraphTransferInfo.getDefaultInstance()) return this;
-      if (nodeInfoBuilder_ == null) {
-        if (!other.nodeInfo_.isEmpty()) {
-          if (nodeInfo_.isEmpty()) {
-            nodeInfo_ = other.nodeInfo_;
-            bitField0_ = (bitField0_ & ~0x00000001);
-          } else {
-            ensureNodeInfoIsMutable();
-            nodeInfo_.addAll(other.nodeInfo_);
-          }
-          onChanged();
-        }
-      } else {
-        if (!other.nodeInfo_.isEmpty()) {
-          if (nodeInfoBuilder_.isEmpty()) {
-            nodeInfoBuilder_.dispose();
-            nodeInfoBuilder_ = null;
-            nodeInfo_ = other.nodeInfo_;
-            bitField0_ = (bitField0_ & ~0x00000001);
-            nodeInfoBuilder_ = 
-              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
-                 getNodeInfoFieldBuilder() : null;
-          } else {
-            nodeInfoBuilder_.addAllMessages(other.nodeInfo_);
-          }
-        }
-      }
-      if (constNodeInfoBuilder_ == null) {
-        if (!other.constNodeInfo_.isEmpty()) {
-          if (constNodeInfo_.isEmpty()) {
-            constNodeInfo_ = other.constNodeInfo_;
-            bitField0_ = (bitField0_ & ~0x00000002);
-          } else {
-            ensureConstNodeInfoIsMutable();
-            constNodeInfo_.addAll(other.constNodeInfo_);
-          }
-          onChanged();
-        }
-      } else {
-        if (!other.constNodeInfo_.isEmpty()) {
-          if (constNodeInfoBuilder_.isEmpty()) {
-            constNodeInfoBuilder_.dispose();
-            constNodeInfoBuilder_ = null;
-            constNodeInfo_ = other.constNodeInfo_;
-            bitField0_ = (bitField0_ & ~0x00000002);
-            constNodeInfoBuilder_ = 
-              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
-                 getConstNodeInfoFieldBuilder() : null;
-          } else {
-            constNodeInfoBuilder_.addAllMessages(other.constNodeInfo_);
-          }
-        }
-      }
-      if (nodeInputInfoBuilder_ == null) {
-        if (!other.nodeInputInfo_.isEmpty()) {
-          if (nodeInputInfo_.isEmpty()) {
-            nodeInputInfo_ = other.nodeInputInfo_;
-            bitField0_ = (bitField0_ & ~0x00000004);
-          } else {
-            ensureNodeInputInfoIsMutable();
-            nodeInputInfo_.addAll(other.nodeInputInfo_);
-          }
-          onChanged();
-        }
-      } else {
-        if (!other.nodeInputInfo_.isEmpty()) {
-          if (nodeInputInfoBuilder_.isEmpty()) {
-            nodeInputInfoBuilder_.dispose();
-            nodeInputInfoBuilder_ = null;
-            nodeInputInfo_ = other.nodeInputInfo_;
-            bitField0_ = (bitField0_ & ~0x00000004);
-            nodeInputInfoBuilder_ = 
-              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
-                 getNodeInputInfoFieldBuilder() : null;
-          } else {
-            nodeInputInfoBuilder_.addAllMessages(other.nodeInputInfo_);
-          }
-        }
-      }
-      if (nodeOutputInfoBuilder_ == null) {
-        if (!other.nodeOutputInfo_.isEmpty()) {
-          if (nodeOutputInfo_.isEmpty()) {
-            nodeOutputInfo_ = other.nodeOutputInfo_;
-            bitField0_ = (bitField0_ & ~0x00000008);
-          } else {
-            ensureNodeOutputInfoIsMutable();
-            nodeOutputInfo_.addAll(other.nodeOutputInfo_);
-          }
-          onChanged();
-        }
-      } else {
-        if (!other.nodeOutputInfo_.isEmpty()) {
-          if (nodeOutputInfoBuilder_.isEmpty()) {
-            nodeOutputInfoBuilder_.dispose();
-            nodeOutputInfoBuilder_ = null;
-            nodeOutputInfo_ = other.nodeOutputInfo_;
-            bitField0_ = (bitField0_ & ~0x00000008);
-            nodeOutputInfoBuilder_ = 
-              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
-                 getNodeOutputInfoFieldBuilder() : null;
-          } else {
-            nodeOutputInfoBuilder_.addAllMessages(other.nodeOutputInfo_);
-          }
-        }
-      }
-      if (graphInputNodeInfoBuilder_ == null) {
-        if (!other.graphInputNodeInfo_.isEmpty()) {
-          if (graphInputNodeInfo_.isEmpty()) {
-            graphInputNodeInfo_ = other.graphInputNodeInfo_;
-            bitField0_ = (bitField0_ & ~0x00000010);
-          } else {
-            ensureGraphInputNodeInfoIsMutable();
-            graphInputNodeInfo_.addAll(other.graphInputNodeInfo_);
-          }
-          onChanged();
-        }
-      } else {
-        if (!other.graphInputNodeInfo_.isEmpty()) {
-          if (graphInputNodeInfoBuilder_.isEmpty()) {
-            graphInputNodeInfoBuilder_.dispose();
-            graphInputNodeInfoBuilder_ = null;
-            graphInputNodeInfo_ = other.graphInputNodeInfo_;
-            bitField0_ = (bitField0_ & ~0x00000010);
-            graphInputNodeInfoBuilder_ = 
-              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
-                 getGraphInputNodeInfoFieldBuilder() : null;
-          } else {
-            graphInputNodeInfoBuilder_.addAllMessages(other.graphInputNodeInfo_);
-          }
-        }
-      }
-      if (graphOutputNodeInfoBuilder_ == null) {
-        if (!other.graphOutputNodeInfo_.isEmpty()) {
-          if (graphOutputNodeInfo_.isEmpty()) {
-            graphOutputNodeInfo_ = other.graphOutputNodeInfo_;
-            bitField0_ = (bitField0_ & ~0x00000020);
-          } else {
-            ensureGraphOutputNodeInfoIsMutable();
-            graphOutputNodeInfo_.addAll(other.graphOutputNodeInfo_);
-          }
-          onChanged();
-        }
-      } else {
-        if (!other.graphOutputNodeInfo_.isEmpty()) {
-          if (graphOutputNodeInfoBuilder_.isEmpty()) {
-            graphOutputNodeInfoBuilder_.dispose();
-            graphOutputNodeInfoBuilder_ = null;
-            graphOutputNodeInfo_ = other.graphOutputNodeInfo_;
-            bitField0_ = (bitField0_ & ~0x00000020);
-            graphOutputNodeInfoBuilder_ = 
-              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
-                 getGraphOutputNodeInfoFieldBuilder() : null;
-          } else {
-            graphOutputNodeInfoBuilder_.addAllMessages(other.graphOutputNodeInfo_);
-          }
-        }
-      }
-      if (other.destination_ != 0) {
-        setDestinationValue(other.getDestinationValue());
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.framework.GraphTransferInfo parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.framework.GraphTransferInfo) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-    private int bitField0_;
-
-    private java.util.List<org.tensorflow.proto.framework.GraphTransferNodeInfo> nodeInfo_ =
-      java.util.Collections.emptyList();
-    private void ensureNodeInfoIsMutable() {
-      if (!((bitField0_ & 0x00000001) != 0)) {
-        nodeInfo_ = new java.util.ArrayList<org.tensorflow.proto.framework.GraphTransferNodeInfo>(nodeInfo_);
-        bitField0_ |= 0x00000001;
-       }
-    }
-
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.framework.GraphTransferNodeInfo, org.tensorflow.proto.framework.GraphTransferNodeInfo.Builder, org.tensorflow.proto.framework.GraphTransferNodeInfoOrBuilder> nodeInfoBuilder_;
-
-    /**
-     * <code>repeated .tensorflow.GraphTransferNodeInfo node_info = 1;</code>
-     */
-    public java.util.List<org.tensorflow.proto.framework.GraphTransferNodeInfo> getNodeInfoList() {
-      if (nodeInfoBuilder_ == null) {
-        return java.util.Collections.unmodifiableList(nodeInfo_);
-      } else {
-        return nodeInfoBuilder_.getMessageList();
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.GraphTransferNodeInfo node_info = 1;</code>
-     */
-    public int getNodeInfoCount() {
-      if (nodeInfoBuilder_ == null) {
-        return nodeInfo_.size();
-      } else {
-        return nodeInfoBuilder_.getCount();
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.GraphTransferNodeInfo node_info = 1;</code>
-     */
-    public org.tensorflow.proto.framework.GraphTransferNodeInfo getNodeInfo(int index) {
-      if (nodeInfoBuilder_ == null) {
-        return nodeInfo_.get(index);
-      } else {
-        return nodeInfoBuilder_.getMessage(index);
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.GraphTransferNodeInfo node_info = 1;</code>
-     */
-    public Builder setNodeInfo(
-        int index, org.tensorflow.proto.framework.GraphTransferNodeInfo value) {
-      if (nodeInfoBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureNodeInfoIsMutable();
-        nodeInfo_.set(index, value);
-        onChanged();
-      } else {
-        nodeInfoBuilder_.setMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.GraphTransferNodeInfo node_info = 1;</code>
-     */
-    public Builder setNodeInfo(
-        int index, org.tensorflow.proto.framework.GraphTransferNodeInfo.Builder builderForValue) {
-      if (nodeInfoBuilder_ == null) {
-        ensureNodeInfoIsMutable();
-        nodeInfo_.set(index, builderForValue.build());
-        onChanged();
-      } else {
-        nodeInfoBuilder_.setMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.GraphTransferNodeInfo node_info = 1;</code>
-     */
-    public Builder addNodeInfo(org.tensorflow.proto.framework.GraphTransferNodeInfo value) {
-      if (nodeInfoBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureNodeInfoIsMutable();
-        nodeInfo_.add(value);
-        onChanged();
-      } else {
-        nodeInfoBuilder_.addMessage(value);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.GraphTransferNodeInfo node_info = 1;</code>
-     */
-    public Builder addNodeInfo(
-        int index, org.tensorflow.proto.framework.GraphTransferNodeInfo value) {
-      if (nodeInfoBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureNodeInfoIsMutable();
-        nodeInfo_.add(index, value);
-        onChanged();
-      } else {
-        nodeInfoBuilder_.addMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.GraphTransferNodeInfo node_info = 1;</code>
-     */
-    public Builder addNodeInfo(
-        org.tensorflow.proto.framework.GraphTransferNodeInfo.Builder builderForValue) {
-      if (nodeInfoBuilder_ == null) {
-        ensureNodeInfoIsMutable();
-        nodeInfo_.add(builderForValue.build());
-        onChanged();
-      } else {
-        nodeInfoBuilder_.addMessage(builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.GraphTransferNodeInfo node_info = 1;</code>
-     */
-    public Builder addNodeInfo(
-        int index, org.tensorflow.proto.framework.GraphTransferNodeInfo.Builder builderForValue) {
-      if (nodeInfoBuilder_ == null) {
-        ensureNodeInfoIsMutable();
-        nodeInfo_.add(index, builderForValue.build());
-        onChanged();
-      } else {
-        nodeInfoBuilder_.addMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.GraphTransferNodeInfo node_info = 1;</code>
-     */
-    public Builder addAllNodeInfo(
-        java.lang.Iterable<? extends org.tensorflow.proto.framework.GraphTransferNodeInfo> values) {
-      if (nodeInfoBuilder_ == null) {
-        ensureNodeInfoIsMutable();
-        com.google.protobuf.AbstractMessageLite.Builder.addAll(
-            values, nodeInfo_);
-        onChanged();
-      } else {
-        nodeInfoBuilder_.addAllMessages(values);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.GraphTransferNodeInfo node_info = 1;</code>
-     */
-    public Builder clearNodeInfo() {
-      if (nodeInfoBuilder_ == null) {
-        nodeInfo_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000001);
-        onChanged();
-      } else {
-        nodeInfoBuilder_.clear();
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.GraphTransferNodeInfo node_info = 1;</code>
-     */
-    public Builder removeNodeInfo(int index) {
-      if (nodeInfoBuilder_ == null) {
-        ensureNodeInfoIsMutable();
-        nodeInfo_.remove(index);
-        onChanged();
-      } else {
-        nodeInfoBuilder_.remove(index);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.GraphTransferNodeInfo node_info = 1;</code>
-     */
-    public org.tensorflow.proto.framework.GraphTransferNodeInfo.Builder getNodeInfoBuilder(
-        int index) {
-      return getNodeInfoFieldBuilder().getBuilder(index);
-    }
-    /**
-     * <code>repeated .tensorflow.GraphTransferNodeInfo node_info = 1;</code>
-     */
-    public org.tensorflow.proto.framework.GraphTransferNodeInfoOrBuilder getNodeInfoOrBuilder(
-        int index) {
-      if (nodeInfoBuilder_ == null) {
-        return nodeInfo_.get(index);  } else {
-        return nodeInfoBuilder_.getMessageOrBuilder(index);
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.GraphTransferNodeInfo node_info = 1;</code>
-     */
-    public java.util.List<? extends org.tensorflow.proto.framework.GraphTransferNodeInfoOrBuilder> 
-         getNodeInfoOrBuilderList() {
-      if (nodeInfoBuilder_ != null) {
-        return nodeInfoBuilder_.getMessageOrBuilderList();
-      } else {
-        return java.util.Collections.unmodifiableList(nodeInfo_);
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.GraphTransferNodeInfo node_info = 1;</code>
-     */
-    public org.tensorflow.proto.framework.GraphTransferNodeInfo.Builder addNodeInfoBuilder() {
-      return getNodeInfoFieldBuilder().addBuilder(
-          org.tensorflow.proto.framework.GraphTransferNodeInfo.getDefaultInstance());
-    }
-    /**
-     * <code>repeated .tensorflow.GraphTransferNodeInfo node_info = 1;</code>
-     */
-    public org.tensorflow.proto.framework.GraphTransferNodeInfo.Builder addNodeInfoBuilder(
-        int index) {
-      return getNodeInfoFieldBuilder().addBuilder(
-          index, org.tensorflow.proto.framework.GraphTransferNodeInfo.getDefaultInstance());
-    }
-    /**
-     * <code>repeated .tensorflow.GraphTransferNodeInfo node_info = 1;</code>
-     */
-    public java.util.List<org.tensorflow.proto.framework.GraphTransferNodeInfo.Builder> 
-         getNodeInfoBuilderList() {
-      return getNodeInfoFieldBuilder().getBuilderList();
-    }
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.framework.GraphTransferNodeInfo, org.tensorflow.proto.framework.GraphTransferNodeInfo.Builder, org.tensorflow.proto.framework.GraphTransferNodeInfoOrBuilder> 
-        getNodeInfoFieldBuilder() {
-      if (nodeInfoBuilder_ == null) {
-        nodeInfoBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
-            org.tensorflow.proto.framework.GraphTransferNodeInfo, org.tensorflow.proto.framework.GraphTransferNodeInfo.Builder, org.tensorflow.proto.framework.GraphTransferNodeInfoOrBuilder>(
-                nodeInfo_,
-                ((bitField0_ & 0x00000001) != 0),
-                getParentForChildren(),
-                isClean());
-        nodeInfo_ = null;
-      }
-      return nodeInfoBuilder_;
-    }
-
-    private java.util.List<org.tensorflow.proto.framework.GraphTransferConstNodeInfo> constNodeInfo_ =
-      java.util.Collections.emptyList();
-    private void ensureConstNodeInfoIsMutable() {
-      if (!((bitField0_ & 0x00000002) != 0)) {
-        constNodeInfo_ = new java.util.ArrayList<org.tensorflow.proto.framework.GraphTransferConstNodeInfo>(constNodeInfo_);
-        bitField0_ |= 0x00000002;
-       }
-    }
-
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.framework.GraphTransferConstNodeInfo, org.tensorflow.proto.framework.GraphTransferConstNodeInfo.Builder, org.tensorflow.proto.framework.GraphTransferConstNodeInfoOrBuilder> constNodeInfoBuilder_;
-
-    /**
-     * <code>repeated .tensorflow.GraphTransferConstNodeInfo const_node_info = 2;</code>
-     */
-    public java.util.List<org.tensorflow.proto.framework.GraphTransferConstNodeInfo> getConstNodeInfoList() {
-      if (constNodeInfoBuilder_ == null) {
-        return java.util.Collections.unmodifiableList(constNodeInfo_);
-      } else {
-        return constNodeInfoBuilder_.getMessageList();
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.GraphTransferConstNodeInfo const_node_info = 2;</code>
-     */
-    public int getConstNodeInfoCount() {
-      if (constNodeInfoBuilder_ == null) {
-        return constNodeInfo_.size();
-      } else {
-        return constNodeInfoBuilder_.getCount();
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.GraphTransferConstNodeInfo const_node_info = 2;</code>
-     */
-    public org.tensorflow.proto.framework.GraphTransferConstNodeInfo getConstNodeInfo(int index) {
-      if (constNodeInfoBuilder_ == null) {
-        return constNodeInfo_.get(index);
-      } else {
-        return constNodeInfoBuilder_.getMessage(index);
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.GraphTransferConstNodeInfo const_node_info = 2;</code>
-     */
-    public Builder setConstNodeInfo(
-        int index, org.tensorflow.proto.framework.GraphTransferConstNodeInfo value) {
-      if (constNodeInfoBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureConstNodeInfoIsMutable();
-        constNodeInfo_.set(index, value);
-        onChanged();
-      } else {
-        constNodeInfoBuilder_.setMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.GraphTransferConstNodeInfo const_node_info = 2;</code>
-     */
-    public Builder setConstNodeInfo(
-        int index, org.tensorflow.proto.framework.GraphTransferConstNodeInfo.Builder builderForValue) {
-      if (constNodeInfoBuilder_ == null) {
-        ensureConstNodeInfoIsMutable();
-        constNodeInfo_.set(index, builderForValue.build());
-        onChanged();
-      } else {
-        constNodeInfoBuilder_.setMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.GraphTransferConstNodeInfo const_node_info = 2;</code>
-     */
-    public Builder addConstNodeInfo(org.tensorflow.proto.framework.GraphTransferConstNodeInfo value) {
-      if (constNodeInfoBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureConstNodeInfoIsMutable();
-        constNodeInfo_.add(value);
-        onChanged();
-      } else {
-        constNodeInfoBuilder_.addMessage(value);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.GraphTransferConstNodeInfo const_node_info = 2;</code>
-     */
-    public Builder addConstNodeInfo(
-        int index, org.tensorflow.proto.framework.GraphTransferConstNodeInfo value) {
-      if (constNodeInfoBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureConstNodeInfoIsMutable();
-        constNodeInfo_.add(index, value);
-        onChanged();
-      } else {
-        constNodeInfoBuilder_.addMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.GraphTransferConstNodeInfo const_node_info = 2;</code>
-     */
-    public Builder addConstNodeInfo(
-        org.tensorflow.proto.framework.GraphTransferConstNodeInfo.Builder builderForValue) {
-      if (constNodeInfoBuilder_ == null) {
-        ensureConstNodeInfoIsMutable();
-        constNodeInfo_.add(builderForValue.build());
-        onChanged();
-      } else {
-        constNodeInfoBuilder_.addMessage(builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.GraphTransferConstNodeInfo const_node_info = 2;</code>
-     */
-    public Builder addConstNodeInfo(
-        int index, org.tensorflow.proto.framework.GraphTransferConstNodeInfo.Builder builderForValue) {
-      if (constNodeInfoBuilder_ == null) {
-        ensureConstNodeInfoIsMutable();
-        constNodeInfo_.add(index, builderForValue.build());
-        onChanged();
-      } else {
-        constNodeInfoBuilder_.addMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.GraphTransferConstNodeInfo const_node_info = 2;</code>
-     */
-    public Builder addAllConstNodeInfo(
-        java.lang.Iterable<? extends org.tensorflow.proto.framework.GraphTransferConstNodeInfo> values) {
-      if (constNodeInfoBuilder_ == null) {
-        ensureConstNodeInfoIsMutable();
-        com.google.protobuf.AbstractMessageLite.Builder.addAll(
-            values, constNodeInfo_);
-        onChanged();
-      } else {
-        constNodeInfoBuilder_.addAllMessages(values);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.GraphTransferConstNodeInfo const_node_info = 2;</code>
-     */
-    public Builder clearConstNodeInfo() {
-      if (constNodeInfoBuilder_ == null) {
-        constNodeInfo_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000002);
-        onChanged();
-      } else {
-        constNodeInfoBuilder_.clear();
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.GraphTransferConstNodeInfo const_node_info = 2;</code>
-     */
-    public Builder removeConstNodeInfo(int index) {
-      if (constNodeInfoBuilder_ == null) {
-        ensureConstNodeInfoIsMutable();
-        constNodeInfo_.remove(index);
-        onChanged();
-      } else {
-        constNodeInfoBuilder_.remove(index);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.GraphTransferConstNodeInfo const_node_info = 2;</code>
-     */
-    public org.tensorflow.proto.framework.GraphTransferConstNodeInfo.Builder getConstNodeInfoBuilder(
-        int index) {
-      return getConstNodeInfoFieldBuilder().getBuilder(index);
-    }
-    /**
-     * <code>repeated .tensorflow.GraphTransferConstNodeInfo const_node_info = 2;</code>
-     */
-    public org.tensorflow.proto.framework.GraphTransferConstNodeInfoOrBuilder getConstNodeInfoOrBuilder(
-        int index) {
-      if (constNodeInfoBuilder_ == null) {
-        return constNodeInfo_.get(index);  } else {
-        return constNodeInfoBuilder_.getMessageOrBuilder(index);
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.GraphTransferConstNodeInfo const_node_info = 2;</code>
-     */
-    public java.util.List<? extends org.tensorflow.proto.framework.GraphTransferConstNodeInfoOrBuilder> 
-         getConstNodeInfoOrBuilderList() {
-      if (constNodeInfoBuilder_ != null) {
-        return constNodeInfoBuilder_.getMessageOrBuilderList();
-      } else {
-        return java.util.Collections.unmodifiableList(constNodeInfo_);
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.GraphTransferConstNodeInfo const_node_info = 2;</code>
-     */
-    public org.tensorflow.proto.framework.GraphTransferConstNodeInfo.Builder addConstNodeInfoBuilder() {
-      return getConstNodeInfoFieldBuilder().addBuilder(
-          org.tensorflow.proto.framework.GraphTransferConstNodeInfo.getDefaultInstance());
-    }
-    /**
-     * <code>repeated .tensorflow.GraphTransferConstNodeInfo const_node_info = 2;</code>
-     */
-    public org.tensorflow.proto.framework.GraphTransferConstNodeInfo.Builder addConstNodeInfoBuilder(
-        int index) {
-      return getConstNodeInfoFieldBuilder().addBuilder(
-          index, org.tensorflow.proto.framework.GraphTransferConstNodeInfo.getDefaultInstance());
-    }
-    /**
-     * <code>repeated .tensorflow.GraphTransferConstNodeInfo const_node_info = 2;</code>
-     */
-    public java.util.List<org.tensorflow.proto.framework.GraphTransferConstNodeInfo.Builder> 
-         getConstNodeInfoBuilderList() {
-      return getConstNodeInfoFieldBuilder().getBuilderList();
-    }
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.framework.GraphTransferConstNodeInfo, org.tensorflow.proto.framework.GraphTransferConstNodeInfo.Builder, org.tensorflow.proto.framework.GraphTransferConstNodeInfoOrBuilder> 
-        getConstNodeInfoFieldBuilder() {
-      if (constNodeInfoBuilder_ == null) {
-        constNodeInfoBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
-            org.tensorflow.proto.framework.GraphTransferConstNodeInfo, org.tensorflow.proto.framework.GraphTransferConstNodeInfo.Builder, org.tensorflow.proto.framework.GraphTransferConstNodeInfoOrBuilder>(
-                constNodeInfo_,
-                ((bitField0_ & 0x00000002) != 0),
-                getParentForChildren(),
-                isClean());
-        constNodeInfo_ = null;
-      }
-      return constNodeInfoBuilder_;
-    }
-
-    private java.util.List<org.tensorflow.proto.framework.GraphTransferNodeInputInfo> nodeInputInfo_ =
-      java.util.Collections.emptyList();
-    private void ensureNodeInputInfoIsMutable() {
-      if (!((bitField0_ & 0x00000004) != 0)) {
-        nodeInputInfo_ = new java.util.ArrayList<org.tensorflow.proto.framework.GraphTransferNodeInputInfo>(nodeInputInfo_);
-        bitField0_ |= 0x00000004;
-       }
-    }
-
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.framework.GraphTransferNodeInputInfo, org.tensorflow.proto.framework.GraphTransferNodeInputInfo.Builder, org.tensorflow.proto.framework.GraphTransferNodeInputInfoOrBuilder> nodeInputInfoBuilder_;
-
-    /**
-     * <code>repeated .tensorflow.GraphTransferNodeInputInfo node_input_info = 3;</code>
-     */
-    public java.util.List<org.tensorflow.proto.framework.GraphTransferNodeInputInfo> getNodeInputInfoList() {
-      if (nodeInputInfoBuilder_ == null) {
-        return java.util.Collections.unmodifiableList(nodeInputInfo_);
-      } else {
-        return nodeInputInfoBuilder_.getMessageList();
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.GraphTransferNodeInputInfo node_input_info = 3;</code>
-     */
-    public int getNodeInputInfoCount() {
-      if (nodeInputInfoBuilder_ == null) {
-        return nodeInputInfo_.size();
-      } else {
-        return nodeInputInfoBuilder_.getCount();
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.GraphTransferNodeInputInfo node_input_info = 3;</code>
-     */
-    public org.tensorflow.proto.framework.GraphTransferNodeInputInfo getNodeInputInfo(int index) {
-      if (nodeInputInfoBuilder_ == null) {
-        return nodeInputInfo_.get(index);
-      } else {
-        return nodeInputInfoBuilder_.getMessage(index);
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.GraphTransferNodeInputInfo node_input_info = 3;</code>
-     */
-    public Builder setNodeInputInfo(
-        int index, org.tensorflow.proto.framework.GraphTransferNodeInputInfo value) {
-      if (nodeInputInfoBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureNodeInputInfoIsMutable();
-        nodeInputInfo_.set(index, value);
-        onChanged();
-      } else {
-        nodeInputInfoBuilder_.setMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.GraphTransferNodeInputInfo node_input_info = 3;</code>
-     */
-    public Builder setNodeInputInfo(
-        int index, org.tensorflow.proto.framework.GraphTransferNodeInputInfo.Builder builderForValue) {
-      if (nodeInputInfoBuilder_ == null) {
-        ensureNodeInputInfoIsMutable();
-        nodeInputInfo_.set(index, builderForValue.build());
-        onChanged();
-      } else {
-        nodeInputInfoBuilder_.setMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.GraphTransferNodeInputInfo node_input_info = 3;</code>
-     */
-    public Builder addNodeInputInfo(org.tensorflow.proto.framework.GraphTransferNodeInputInfo value) {
-      if (nodeInputInfoBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureNodeInputInfoIsMutable();
-        nodeInputInfo_.add(value);
-        onChanged();
-      } else {
-        nodeInputInfoBuilder_.addMessage(value);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.GraphTransferNodeInputInfo node_input_info = 3;</code>
-     */
-    public Builder addNodeInputInfo(
-        int index, org.tensorflow.proto.framework.GraphTransferNodeInputInfo value) {
-      if (nodeInputInfoBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureNodeInputInfoIsMutable();
-        nodeInputInfo_.add(index, value);
-        onChanged();
-      } else {
-        nodeInputInfoBuilder_.addMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.GraphTransferNodeInputInfo node_input_info = 3;</code>
-     */
-    public Builder addNodeInputInfo(
-        org.tensorflow.proto.framework.GraphTransferNodeInputInfo.Builder builderForValue) {
-      if (nodeInputInfoBuilder_ == null) {
-        ensureNodeInputInfoIsMutable();
-        nodeInputInfo_.add(builderForValue.build());
-        onChanged();
-      } else {
-        nodeInputInfoBuilder_.addMessage(builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.GraphTransferNodeInputInfo node_input_info = 3;</code>
-     */
-    public Builder addNodeInputInfo(
-        int index, org.tensorflow.proto.framework.GraphTransferNodeInputInfo.Builder builderForValue) {
-      if (nodeInputInfoBuilder_ == null) {
-        ensureNodeInputInfoIsMutable();
-        nodeInputInfo_.add(index, builderForValue.build());
-        onChanged();
-      } else {
-        nodeInputInfoBuilder_.addMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.GraphTransferNodeInputInfo node_input_info = 3;</code>
-     */
-    public Builder addAllNodeInputInfo(
-        java.lang.Iterable<? extends org.tensorflow.proto.framework.GraphTransferNodeInputInfo> values) {
-      if (nodeInputInfoBuilder_ == null) {
-        ensureNodeInputInfoIsMutable();
-        com.google.protobuf.AbstractMessageLite.Builder.addAll(
-            values, nodeInputInfo_);
-        onChanged();
-      } else {
-        nodeInputInfoBuilder_.addAllMessages(values);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.GraphTransferNodeInputInfo node_input_info = 3;</code>
-     */
-    public Builder clearNodeInputInfo() {
-      if (nodeInputInfoBuilder_ == null) {
-        nodeInputInfo_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000004);
-        onChanged();
-      } else {
-        nodeInputInfoBuilder_.clear();
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.GraphTransferNodeInputInfo node_input_info = 3;</code>
-     */
-    public Builder removeNodeInputInfo(int index) {
-      if (nodeInputInfoBuilder_ == null) {
-        ensureNodeInputInfoIsMutable();
-        nodeInputInfo_.remove(index);
-        onChanged();
-      } else {
-        nodeInputInfoBuilder_.remove(index);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.GraphTransferNodeInputInfo node_input_info = 3;</code>
-     */
-    public org.tensorflow.proto.framework.GraphTransferNodeInputInfo.Builder getNodeInputInfoBuilder(
-        int index) {
-      return getNodeInputInfoFieldBuilder().getBuilder(index);
-    }
-    /**
-     * <code>repeated .tensorflow.GraphTransferNodeInputInfo node_input_info = 3;</code>
-     */
-    public org.tensorflow.proto.framework.GraphTransferNodeInputInfoOrBuilder getNodeInputInfoOrBuilder(
-        int index) {
-      if (nodeInputInfoBuilder_ == null) {
-        return nodeInputInfo_.get(index);  } else {
-        return nodeInputInfoBuilder_.getMessageOrBuilder(index);
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.GraphTransferNodeInputInfo node_input_info = 3;</code>
-     */
-    public java.util.List<? extends org.tensorflow.proto.framework.GraphTransferNodeInputInfoOrBuilder> 
-         getNodeInputInfoOrBuilderList() {
-      if (nodeInputInfoBuilder_ != null) {
-        return nodeInputInfoBuilder_.getMessageOrBuilderList();
-      } else {
-        return java.util.Collections.unmodifiableList(nodeInputInfo_);
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.GraphTransferNodeInputInfo node_input_info = 3;</code>
-     */
-    public org.tensorflow.proto.framework.GraphTransferNodeInputInfo.Builder addNodeInputInfoBuilder() {
-      return getNodeInputInfoFieldBuilder().addBuilder(
-          org.tensorflow.proto.framework.GraphTransferNodeInputInfo.getDefaultInstance());
-    }
-    /**
-     * <code>repeated .tensorflow.GraphTransferNodeInputInfo node_input_info = 3;</code>
-     */
-    public org.tensorflow.proto.framework.GraphTransferNodeInputInfo.Builder addNodeInputInfoBuilder(
-        int index) {
-      return getNodeInputInfoFieldBuilder().addBuilder(
-          index, org.tensorflow.proto.framework.GraphTransferNodeInputInfo.getDefaultInstance());
-    }
-    /**
-     * <code>repeated .tensorflow.GraphTransferNodeInputInfo node_input_info = 3;</code>
-     */
-    public java.util.List<org.tensorflow.proto.framework.GraphTransferNodeInputInfo.Builder> 
-         getNodeInputInfoBuilderList() {
-      return getNodeInputInfoFieldBuilder().getBuilderList();
-    }
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.framework.GraphTransferNodeInputInfo, org.tensorflow.proto.framework.GraphTransferNodeInputInfo.Builder, org.tensorflow.proto.framework.GraphTransferNodeInputInfoOrBuilder> 
-        getNodeInputInfoFieldBuilder() {
-      if (nodeInputInfoBuilder_ == null) {
-        nodeInputInfoBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
-            org.tensorflow.proto.framework.GraphTransferNodeInputInfo, org.tensorflow.proto.framework.GraphTransferNodeInputInfo.Builder, org.tensorflow.proto.framework.GraphTransferNodeInputInfoOrBuilder>(
-                nodeInputInfo_,
-                ((bitField0_ & 0x00000004) != 0),
-                getParentForChildren(),
-                isClean());
-        nodeInputInfo_ = null;
-      }
-      return nodeInputInfoBuilder_;
-    }
-
-    private java.util.List<org.tensorflow.proto.framework.GraphTransferNodeOutputInfo> nodeOutputInfo_ =
-      java.util.Collections.emptyList();
-    private void ensureNodeOutputInfoIsMutable() {
-      if (!((bitField0_ & 0x00000008) != 0)) {
-        nodeOutputInfo_ = new java.util.ArrayList<org.tensorflow.proto.framework.GraphTransferNodeOutputInfo>(nodeOutputInfo_);
-        bitField0_ |= 0x00000008;
-       }
-    }
-
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.framework.GraphTransferNodeOutputInfo, org.tensorflow.proto.framework.GraphTransferNodeOutputInfo.Builder, org.tensorflow.proto.framework.GraphTransferNodeOutputInfoOrBuilder> nodeOutputInfoBuilder_;
-
-    /**
-     * <code>repeated .tensorflow.GraphTransferNodeOutputInfo node_output_info = 4;</code>
-     */
-    public java.util.List<org.tensorflow.proto.framework.GraphTransferNodeOutputInfo> getNodeOutputInfoList() {
-      if (nodeOutputInfoBuilder_ == null) {
-        return java.util.Collections.unmodifiableList(nodeOutputInfo_);
-      } else {
-        return nodeOutputInfoBuilder_.getMessageList();
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.GraphTransferNodeOutputInfo node_output_info = 4;</code>
-     */
-    public int getNodeOutputInfoCount() {
-      if (nodeOutputInfoBuilder_ == null) {
-        return nodeOutputInfo_.size();
-      } else {
-        return nodeOutputInfoBuilder_.getCount();
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.GraphTransferNodeOutputInfo node_output_info = 4;</code>
-     */
-    public org.tensorflow.proto.framework.GraphTransferNodeOutputInfo getNodeOutputInfo(int index) {
-      if (nodeOutputInfoBuilder_ == null) {
-        return nodeOutputInfo_.get(index);
-      } else {
-        return nodeOutputInfoBuilder_.getMessage(index);
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.GraphTransferNodeOutputInfo node_output_info = 4;</code>
-     */
-    public Builder setNodeOutputInfo(
-        int index, org.tensorflow.proto.framework.GraphTransferNodeOutputInfo value) {
-      if (nodeOutputInfoBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureNodeOutputInfoIsMutable();
-        nodeOutputInfo_.set(index, value);
-        onChanged();
-      } else {
-        nodeOutputInfoBuilder_.setMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.GraphTransferNodeOutputInfo node_output_info = 4;</code>
-     */
-    public Builder setNodeOutputInfo(
-        int index, org.tensorflow.proto.framework.GraphTransferNodeOutputInfo.Builder builderForValue) {
-      if (nodeOutputInfoBuilder_ == null) {
-        ensureNodeOutputInfoIsMutable();
-        nodeOutputInfo_.set(index, builderForValue.build());
-        onChanged();
-      } else {
-        nodeOutputInfoBuilder_.setMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.GraphTransferNodeOutputInfo node_output_info = 4;</code>
-     */
-    public Builder addNodeOutputInfo(org.tensorflow.proto.framework.GraphTransferNodeOutputInfo value) {
-      if (nodeOutputInfoBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureNodeOutputInfoIsMutable();
-        nodeOutputInfo_.add(value);
-        onChanged();
-      } else {
-        nodeOutputInfoBuilder_.addMessage(value);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.GraphTransferNodeOutputInfo node_output_info = 4;</code>
-     */
-    public Builder addNodeOutputInfo(
-        int index, org.tensorflow.proto.framework.GraphTransferNodeOutputInfo value) {
-      if (nodeOutputInfoBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureNodeOutputInfoIsMutable();
-        nodeOutputInfo_.add(index, value);
-        onChanged();
-      } else {
-        nodeOutputInfoBuilder_.addMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.GraphTransferNodeOutputInfo node_output_info = 4;</code>
-     */
-    public Builder addNodeOutputInfo(
-        org.tensorflow.proto.framework.GraphTransferNodeOutputInfo.Builder builderForValue) {
-      if (nodeOutputInfoBuilder_ == null) {
-        ensureNodeOutputInfoIsMutable();
-        nodeOutputInfo_.add(builderForValue.build());
-        onChanged();
-      } else {
-        nodeOutputInfoBuilder_.addMessage(builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.GraphTransferNodeOutputInfo node_output_info = 4;</code>
-     */
-    public Builder addNodeOutputInfo(
-        int index, org.tensorflow.proto.framework.GraphTransferNodeOutputInfo.Builder builderForValue) {
-      if (nodeOutputInfoBuilder_ == null) {
-        ensureNodeOutputInfoIsMutable();
-        nodeOutputInfo_.add(index, builderForValue.build());
-        onChanged();
-      } else {
-        nodeOutputInfoBuilder_.addMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.GraphTransferNodeOutputInfo node_output_info = 4;</code>
-     */
-    public Builder addAllNodeOutputInfo(
-        java.lang.Iterable<? extends org.tensorflow.proto.framework.GraphTransferNodeOutputInfo> values) {
-      if (nodeOutputInfoBuilder_ == null) {
-        ensureNodeOutputInfoIsMutable();
-        com.google.protobuf.AbstractMessageLite.Builder.addAll(
-            values, nodeOutputInfo_);
-        onChanged();
-      } else {
-        nodeOutputInfoBuilder_.addAllMessages(values);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.GraphTransferNodeOutputInfo node_output_info = 4;</code>
-     */
-    public Builder clearNodeOutputInfo() {
-      if (nodeOutputInfoBuilder_ == null) {
-        nodeOutputInfo_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000008);
-        onChanged();
-      } else {
-        nodeOutputInfoBuilder_.clear();
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.GraphTransferNodeOutputInfo node_output_info = 4;</code>
-     */
-    public Builder removeNodeOutputInfo(int index) {
-      if (nodeOutputInfoBuilder_ == null) {
-        ensureNodeOutputInfoIsMutable();
-        nodeOutputInfo_.remove(index);
-        onChanged();
-      } else {
-        nodeOutputInfoBuilder_.remove(index);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.GraphTransferNodeOutputInfo node_output_info = 4;</code>
-     */
-    public org.tensorflow.proto.framework.GraphTransferNodeOutputInfo.Builder getNodeOutputInfoBuilder(
-        int index) {
-      return getNodeOutputInfoFieldBuilder().getBuilder(index);
-    }
-    /**
-     * <code>repeated .tensorflow.GraphTransferNodeOutputInfo node_output_info = 4;</code>
-     */
-    public org.tensorflow.proto.framework.GraphTransferNodeOutputInfoOrBuilder getNodeOutputInfoOrBuilder(
-        int index) {
-      if (nodeOutputInfoBuilder_ == null) {
-        return nodeOutputInfo_.get(index);  } else {
-        return nodeOutputInfoBuilder_.getMessageOrBuilder(index);
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.GraphTransferNodeOutputInfo node_output_info = 4;</code>
-     */
-    public java.util.List<? extends org.tensorflow.proto.framework.GraphTransferNodeOutputInfoOrBuilder> 
-         getNodeOutputInfoOrBuilderList() {
-      if (nodeOutputInfoBuilder_ != null) {
-        return nodeOutputInfoBuilder_.getMessageOrBuilderList();
-      } else {
-        return java.util.Collections.unmodifiableList(nodeOutputInfo_);
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.GraphTransferNodeOutputInfo node_output_info = 4;</code>
-     */
-    public org.tensorflow.proto.framework.GraphTransferNodeOutputInfo.Builder addNodeOutputInfoBuilder() {
-      return getNodeOutputInfoFieldBuilder().addBuilder(
-          org.tensorflow.proto.framework.GraphTransferNodeOutputInfo.getDefaultInstance());
-    }
-    /**
-     * <code>repeated .tensorflow.GraphTransferNodeOutputInfo node_output_info = 4;</code>
-     */
-    public org.tensorflow.proto.framework.GraphTransferNodeOutputInfo.Builder addNodeOutputInfoBuilder(
-        int index) {
-      return getNodeOutputInfoFieldBuilder().addBuilder(
-          index, org.tensorflow.proto.framework.GraphTransferNodeOutputInfo.getDefaultInstance());
-    }
-    /**
-     * <code>repeated .tensorflow.GraphTransferNodeOutputInfo node_output_info = 4;</code>
-     */
-    public java.util.List<org.tensorflow.proto.framework.GraphTransferNodeOutputInfo.Builder> 
-         getNodeOutputInfoBuilderList() {
-      return getNodeOutputInfoFieldBuilder().getBuilderList();
-    }
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.framework.GraphTransferNodeOutputInfo, org.tensorflow.proto.framework.GraphTransferNodeOutputInfo.Builder, org.tensorflow.proto.framework.GraphTransferNodeOutputInfoOrBuilder> 
-        getNodeOutputInfoFieldBuilder() {
-      if (nodeOutputInfoBuilder_ == null) {
-        nodeOutputInfoBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
-            org.tensorflow.proto.framework.GraphTransferNodeOutputInfo, org.tensorflow.proto.framework.GraphTransferNodeOutputInfo.Builder, org.tensorflow.proto.framework.GraphTransferNodeOutputInfoOrBuilder>(
-                nodeOutputInfo_,
-                ((bitField0_ & 0x00000008) != 0),
-                getParentForChildren(),
-                isClean());
-        nodeOutputInfo_ = null;
-      }
-      return nodeOutputInfoBuilder_;
-    }
-
-    private java.util.List<org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfo> graphInputNodeInfo_ =
-      java.util.Collections.emptyList();
-    private void ensureGraphInputNodeInfoIsMutable() {
-      if (!((bitField0_ & 0x00000010) != 0)) {
-        graphInputNodeInfo_ = new java.util.ArrayList<org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfo>(graphInputNodeInfo_);
-        bitField0_ |= 0x00000010;
-       }
-    }
-
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfo, org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfo.Builder, org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfoOrBuilder> graphInputNodeInfoBuilder_;
-
-    /**
-     * <pre>
-     * Input Node parameters of transferred graph
-     * </pre>
-     *
-     * <code>repeated .tensorflow.GraphTransferGraphInputNodeInfo graph_input_node_info = 5;</code>
-     */
-    public java.util.List<org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfo> getGraphInputNodeInfoList() {
-      if (graphInputNodeInfoBuilder_ == null) {
-        return java.util.Collections.unmodifiableList(graphInputNodeInfo_);
-      } else {
-        return graphInputNodeInfoBuilder_.getMessageList();
-      }
-    }
-    /**
-     * <pre>
-     * Input Node parameters of transferred graph
-     * </pre>
-     *
-     * <code>repeated .tensorflow.GraphTransferGraphInputNodeInfo graph_input_node_info = 5;</code>
-     */
-    public int getGraphInputNodeInfoCount() {
-      if (graphInputNodeInfoBuilder_ == null) {
-        return graphInputNodeInfo_.size();
-      } else {
-        return graphInputNodeInfoBuilder_.getCount();
-      }
-    }
-    /**
-     * <pre>
-     * Input Node parameters of transferred graph
-     * </pre>
-     *
-     * <code>repeated .tensorflow.GraphTransferGraphInputNodeInfo graph_input_node_info = 5;</code>
-     */
-    public org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfo getGraphInputNodeInfo(int index) {
-      if (graphInputNodeInfoBuilder_ == null) {
-        return graphInputNodeInfo_.get(index);
-      } else {
-        return graphInputNodeInfoBuilder_.getMessage(index);
-      }
-    }
-    /**
-     * <pre>
-     * Input Node parameters of transferred graph
-     * </pre>
-     *
-     * <code>repeated .tensorflow.GraphTransferGraphInputNodeInfo graph_input_node_info = 5;</code>
-     */
-    public Builder setGraphInputNodeInfo(
-        int index, org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfo value) {
-      if (graphInputNodeInfoBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureGraphInputNodeInfoIsMutable();
-        graphInputNodeInfo_.set(index, value);
-        onChanged();
-      } else {
-        graphInputNodeInfoBuilder_.setMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Input Node parameters of transferred graph
-     * </pre>
-     *
-     * <code>repeated .tensorflow.GraphTransferGraphInputNodeInfo graph_input_node_info = 5;</code>
-     */
-    public Builder setGraphInputNodeInfo(
-        int index, org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfo.Builder builderForValue) {
-      if (graphInputNodeInfoBuilder_ == null) {
-        ensureGraphInputNodeInfoIsMutable();
-        graphInputNodeInfo_.set(index, builderForValue.build());
-        onChanged();
-      } else {
-        graphInputNodeInfoBuilder_.setMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Input Node parameters of transferred graph
-     * </pre>
-     *
-     * <code>repeated .tensorflow.GraphTransferGraphInputNodeInfo graph_input_node_info = 5;</code>
-     */
-    public Builder addGraphInputNodeInfo(org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfo value) {
-      if (graphInputNodeInfoBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureGraphInputNodeInfoIsMutable();
-        graphInputNodeInfo_.add(value);
-        onChanged();
-      } else {
-        graphInputNodeInfoBuilder_.addMessage(value);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Input Node parameters of transferred graph
-     * </pre>
-     *
-     * <code>repeated .tensorflow.GraphTransferGraphInputNodeInfo graph_input_node_info = 5;</code>
-     */
-    public Builder addGraphInputNodeInfo(
-        int index, org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfo value) {
-      if (graphInputNodeInfoBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureGraphInputNodeInfoIsMutable();
-        graphInputNodeInfo_.add(index, value);
-        onChanged();
-      } else {
-        graphInputNodeInfoBuilder_.addMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Input Node parameters of transferred graph
-     * </pre>
-     *
-     * <code>repeated .tensorflow.GraphTransferGraphInputNodeInfo graph_input_node_info = 5;</code>
-     */
-    public Builder addGraphInputNodeInfo(
-        org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfo.Builder builderForValue) {
-      if (graphInputNodeInfoBuilder_ == null) {
-        ensureGraphInputNodeInfoIsMutable();
-        graphInputNodeInfo_.add(builderForValue.build());
-        onChanged();
-      } else {
-        graphInputNodeInfoBuilder_.addMessage(builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Input Node parameters of transferred graph
-     * </pre>
-     *
-     * <code>repeated .tensorflow.GraphTransferGraphInputNodeInfo graph_input_node_info = 5;</code>
-     */
-    public Builder addGraphInputNodeInfo(
-        int index, org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfo.Builder builderForValue) {
-      if (graphInputNodeInfoBuilder_ == null) {
-        ensureGraphInputNodeInfoIsMutable();
-        graphInputNodeInfo_.add(index, builderForValue.build());
-        onChanged();
-      } else {
-        graphInputNodeInfoBuilder_.addMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Input Node parameters of transferred graph
-     * </pre>
-     *
-     * <code>repeated .tensorflow.GraphTransferGraphInputNodeInfo graph_input_node_info = 5;</code>
-     */
-    public Builder addAllGraphInputNodeInfo(
-        java.lang.Iterable<? extends org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfo> values) {
-      if (graphInputNodeInfoBuilder_ == null) {
-        ensureGraphInputNodeInfoIsMutable();
-        com.google.protobuf.AbstractMessageLite.Builder.addAll(
-            values, graphInputNodeInfo_);
-        onChanged();
-      } else {
-        graphInputNodeInfoBuilder_.addAllMessages(values);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Input Node parameters of transferred graph
-     * </pre>
-     *
-     * <code>repeated .tensorflow.GraphTransferGraphInputNodeInfo graph_input_node_info = 5;</code>
-     */
-    public Builder clearGraphInputNodeInfo() {
-      if (graphInputNodeInfoBuilder_ == null) {
-        graphInputNodeInfo_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000010);
-        onChanged();
-      } else {
-        graphInputNodeInfoBuilder_.clear();
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Input Node parameters of transferred graph
-     * </pre>
-     *
-     * <code>repeated .tensorflow.GraphTransferGraphInputNodeInfo graph_input_node_info = 5;</code>
-     */
-    public Builder removeGraphInputNodeInfo(int index) {
-      if (graphInputNodeInfoBuilder_ == null) {
-        ensureGraphInputNodeInfoIsMutable();
-        graphInputNodeInfo_.remove(index);
-        onChanged();
-      } else {
-        graphInputNodeInfoBuilder_.remove(index);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Input Node parameters of transferred graph
-     * </pre>
-     *
-     * <code>repeated .tensorflow.GraphTransferGraphInputNodeInfo graph_input_node_info = 5;</code>
-     */
-    public org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfo.Builder getGraphInputNodeInfoBuilder(
-        int index) {
-      return getGraphInputNodeInfoFieldBuilder().getBuilder(index);
-    }
-    /**
-     * <pre>
-     * Input Node parameters of transferred graph
-     * </pre>
-     *
-     * <code>repeated .tensorflow.GraphTransferGraphInputNodeInfo graph_input_node_info = 5;</code>
-     */
-    public org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfoOrBuilder getGraphInputNodeInfoOrBuilder(
-        int index) {
-      if (graphInputNodeInfoBuilder_ == null) {
-        return graphInputNodeInfo_.get(index);  } else {
-        return graphInputNodeInfoBuilder_.getMessageOrBuilder(index);
-      }
-    }
-    /**
-     * <pre>
-     * Input Node parameters of transferred graph
-     * </pre>
-     *
-     * <code>repeated .tensorflow.GraphTransferGraphInputNodeInfo graph_input_node_info = 5;</code>
-     */
-    public java.util.List<? extends org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfoOrBuilder> 
-         getGraphInputNodeInfoOrBuilderList() {
-      if (graphInputNodeInfoBuilder_ != null) {
-        return graphInputNodeInfoBuilder_.getMessageOrBuilderList();
-      } else {
-        return java.util.Collections.unmodifiableList(graphInputNodeInfo_);
-      }
-    }
-    /**
-     * <pre>
-     * Input Node parameters of transferred graph
-     * </pre>
-     *
-     * <code>repeated .tensorflow.GraphTransferGraphInputNodeInfo graph_input_node_info = 5;</code>
-     */
-    public org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfo.Builder addGraphInputNodeInfoBuilder() {
-      return getGraphInputNodeInfoFieldBuilder().addBuilder(
-          org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfo.getDefaultInstance());
-    }
-    /**
-     * <pre>
-     * Input Node parameters of transferred graph
-     * </pre>
-     *
-     * <code>repeated .tensorflow.GraphTransferGraphInputNodeInfo graph_input_node_info = 5;</code>
-     */
-    public org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfo.Builder addGraphInputNodeInfoBuilder(
-        int index) {
-      return getGraphInputNodeInfoFieldBuilder().addBuilder(
-          index, org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfo.getDefaultInstance());
-    }
-    /**
-     * <pre>
-     * Input Node parameters of transferred graph
-     * </pre>
-     *
-     * <code>repeated .tensorflow.GraphTransferGraphInputNodeInfo graph_input_node_info = 5;</code>
-     */
-    public java.util.List<org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfo.Builder> 
-         getGraphInputNodeInfoBuilderList() {
-      return getGraphInputNodeInfoFieldBuilder().getBuilderList();
-    }
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfo, org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfo.Builder, org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfoOrBuilder> 
-        getGraphInputNodeInfoFieldBuilder() {
-      if (graphInputNodeInfoBuilder_ == null) {
-        graphInputNodeInfoBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
-            org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfo, org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfo.Builder, org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfoOrBuilder>(
-                graphInputNodeInfo_,
-                ((bitField0_ & 0x00000010) != 0),
-                getParentForChildren(),
-                isClean());
-        graphInputNodeInfo_ = null;
-      }
-      return graphInputNodeInfoBuilder_;
-    }
-
-    private java.util.List<org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfo> graphOutputNodeInfo_ =
-      java.util.Collections.emptyList();
-    private void ensureGraphOutputNodeInfoIsMutable() {
-      if (!((bitField0_ & 0x00000020) != 0)) {
-        graphOutputNodeInfo_ = new java.util.ArrayList<org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfo>(graphOutputNodeInfo_);
-        bitField0_ |= 0x00000020;
-       }
-    }
-
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfo, org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfo.Builder, org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfoOrBuilder> graphOutputNodeInfoBuilder_;
-
-    /**
-     * <code>repeated .tensorflow.GraphTransferGraphOutputNodeInfo graph_output_node_info = 6;</code>
-     */
-    public java.util.List<org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfo> getGraphOutputNodeInfoList() {
-      if (graphOutputNodeInfoBuilder_ == null) {
-        return java.util.Collections.unmodifiableList(graphOutputNodeInfo_);
-      } else {
-        return graphOutputNodeInfoBuilder_.getMessageList();
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.GraphTransferGraphOutputNodeInfo graph_output_node_info = 6;</code>
-     */
-    public int getGraphOutputNodeInfoCount() {
-      if (graphOutputNodeInfoBuilder_ == null) {
-        return graphOutputNodeInfo_.size();
-      } else {
-        return graphOutputNodeInfoBuilder_.getCount();
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.GraphTransferGraphOutputNodeInfo graph_output_node_info = 6;</code>
-     */
-    public org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfo getGraphOutputNodeInfo(int index) {
-      if (graphOutputNodeInfoBuilder_ == null) {
-        return graphOutputNodeInfo_.get(index);
-      } else {
-        return graphOutputNodeInfoBuilder_.getMessage(index);
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.GraphTransferGraphOutputNodeInfo graph_output_node_info = 6;</code>
-     */
-    public Builder setGraphOutputNodeInfo(
-        int index, org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfo value) {
-      if (graphOutputNodeInfoBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureGraphOutputNodeInfoIsMutable();
-        graphOutputNodeInfo_.set(index, value);
-        onChanged();
-      } else {
-        graphOutputNodeInfoBuilder_.setMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.GraphTransferGraphOutputNodeInfo graph_output_node_info = 6;</code>
-     */
-    public Builder setGraphOutputNodeInfo(
-        int index, org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfo.Builder builderForValue) {
-      if (graphOutputNodeInfoBuilder_ == null) {
-        ensureGraphOutputNodeInfoIsMutable();
-        graphOutputNodeInfo_.set(index, builderForValue.build());
-        onChanged();
-      } else {
-        graphOutputNodeInfoBuilder_.setMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.GraphTransferGraphOutputNodeInfo graph_output_node_info = 6;</code>
-     */
-    public Builder addGraphOutputNodeInfo(org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfo value) {
-      if (graphOutputNodeInfoBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureGraphOutputNodeInfoIsMutable();
-        graphOutputNodeInfo_.add(value);
-        onChanged();
-      } else {
-        graphOutputNodeInfoBuilder_.addMessage(value);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.GraphTransferGraphOutputNodeInfo graph_output_node_info = 6;</code>
-     */
-    public Builder addGraphOutputNodeInfo(
-        int index, org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfo value) {
-      if (graphOutputNodeInfoBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureGraphOutputNodeInfoIsMutable();
-        graphOutputNodeInfo_.add(index, value);
-        onChanged();
-      } else {
-        graphOutputNodeInfoBuilder_.addMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.GraphTransferGraphOutputNodeInfo graph_output_node_info = 6;</code>
-     */
-    public Builder addGraphOutputNodeInfo(
-        org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfo.Builder builderForValue) {
-      if (graphOutputNodeInfoBuilder_ == null) {
-        ensureGraphOutputNodeInfoIsMutable();
-        graphOutputNodeInfo_.add(builderForValue.build());
-        onChanged();
-      } else {
-        graphOutputNodeInfoBuilder_.addMessage(builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.GraphTransferGraphOutputNodeInfo graph_output_node_info = 6;</code>
-     */
-    public Builder addGraphOutputNodeInfo(
-        int index, org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfo.Builder builderForValue) {
-      if (graphOutputNodeInfoBuilder_ == null) {
-        ensureGraphOutputNodeInfoIsMutable();
-        graphOutputNodeInfo_.add(index, builderForValue.build());
-        onChanged();
-      } else {
-        graphOutputNodeInfoBuilder_.addMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.GraphTransferGraphOutputNodeInfo graph_output_node_info = 6;</code>
-     */
-    public Builder addAllGraphOutputNodeInfo(
-        java.lang.Iterable<? extends org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfo> values) {
-      if (graphOutputNodeInfoBuilder_ == null) {
-        ensureGraphOutputNodeInfoIsMutable();
-        com.google.protobuf.AbstractMessageLite.Builder.addAll(
-            values, graphOutputNodeInfo_);
-        onChanged();
-      } else {
-        graphOutputNodeInfoBuilder_.addAllMessages(values);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.GraphTransferGraphOutputNodeInfo graph_output_node_info = 6;</code>
-     */
-    public Builder clearGraphOutputNodeInfo() {
-      if (graphOutputNodeInfoBuilder_ == null) {
-        graphOutputNodeInfo_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000020);
-        onChanged();
-      } else {
-        graphOutputNodeInfoBuilder_.clear();
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.GraphTransferGraphOutputNodeInfo graph_output_node_info = 6;</code>
-     */
-    public Builder removeGraphOutputNodeInfo(int index) {
-      if (graphOutputNodeInfoBuilder_ == null) {
-        ensureGraphOutputNodeInfoIsMutable();
-        graphOutputNodeInfo_.remove(index);
-        onChanged();
-      } else {
-        graphOutputNodeInfoBuilder_.remove(index);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.GraphTransferGraphOutputNodeInfo graph_output_node_info = 6;</code>
-     */
-    public org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfo.Builder getGraphOutputNodeInfoBuilder(
-        int index) {
-      return getGraphOutputNodeInfoFieldBuilder().getBuilder(index);
-    }
-    /**
-     * <code>repeated .tensorflow.GraphTransferGraphOutputNodeInfo graph_output_node_info = 6;</code>
-     */
-    public org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfoOrBuilder getGraphOutputNodeInfoOrBuilder(
-        int index) {
-      if (graphOutputNodeInfoBuilder_ == null) {
-        return graphOutputNodeInfo_.get(index);  } else {
-        return graphOutputNodeInfoBuilder_.getMessageOrBuilder(index);
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.GraphTransferGraphOutputNodeInfo graph_output_node_info = 6;</code>
-     */
-    public java.util.List<? extends org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfoOrBuilder> 
-         getGraphOutputNodeInfoOrBuilderList() {
-      if (graphOutputNodeInfoBuilder_ != null) {
-        return graphOutputNodeInfoBuilder_.getMessageOrBuilderList();
-      } else {
-        return java.util.Collections.unmodifiableList(graphOutputNodeInfo_);
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.GraphTransferGraphOutputNodeInfo graph_output_node_info = 6;</code>
-     */
-    public org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfo.Builder addGraphOutputNodeInfoBuilder() {
-      return getGraphOutputNodeInfoFieldBuilder().addBuilder(
-          org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfo.getDefaultInstance());
-    }
-    /**
-     * <code>repeated .tensorflow.GraphTransferGraphOutputNodeInfo graph_output_node_info = 6;</code>
-     */
-    public org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfo.Builder addGraphOutputNodeInfoBuilder(
-        int index) {
-      return getGraphOutputNodeInfoFieldBuilder().addBuilder(
-          index, org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfo.getDefaultInstance());
-    }
-    /**
-     * <code>repeated .tensorflow.GraphTransferGraphOutputNodeInfo graph_output_node_info = 6;</code>
-     */
-    public java.util.List<org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfo.Builder> 
-         getGraphOutputNodeInfoBuilderList() {
-      return getGraphOutputNodeInfoFieldBuilder().getBuilderList();
-    }
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfo, org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfo.Builder, org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfoOrBuilder> 
-        getGraphOutputNodeInfoFieldBuilder() {
-      if (graphOutputNodeInfoBuilder_ == null) {
-        graphOutputNodeInfoBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
-            org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfo, org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfo.Builder, org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfoOrBuilder>(
-                graphOutputNodeInfo_,
-                ((bitField0_ & 0x00000020) != 0),
-                getParentForChildren(),
-                isClean());
-        graphOutputNodeInfo_ = null;
-      }
-      return graphOutputNodeInfoBuilder_;
-    }
-
-    private int destination_ = 0;
-    /**
-     * <pre>
-     * Destination of graph transfer
-     * </pre>
-     *
-     * <code>.tensorflow.GraphTransferInfo.Destination destination = 7;</code>
-     */
-    public int getDestinationValue() {
-      return destination_;
-    }
-    /**
-     * <pre>
-     * Destination of graph transfer
-     * </pre>
-     *
-     * <code>.tensorflow.GraphTransferInfo.Destination destination = 7;</code>
-     */
-    public Builder setDestinationValue(int value) {
-      destination_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Destination of graph transfer
-     * </pre>
-     *
-     * <code>.tensorflow.GraphTransferInfo.Destination destination = 7;</code>
-     */
-    public org.tensorflow.proto.framework.GraphTransferInfo.Destination getDestination() {
-      @SuppressWarnings("deprecation")
-      org.tensorflow.proto.framework.GraphTransferInfo.Destination result = org.tensorflow.proto.framework.GraphTransferInfo.Destination.valueOf(destination_);
-      return result == null ? org.tensorflow.proto.framework.GraphTransferInfo.Destination.UNRECOGNIZED : result;
-    }
-    /**
-     * <pre>
-     * Destination of graph transfer
-     * </pre>
-     *
-     * <code>.tensorflow.GraphTransferInfo.Destination destination = 7;</code>
-     */
-    public Builder setDestination(org.tensorflow.proto.framework.GraphTransferInfo.Destination value) {
-      if (value == null) {
-        throw new NullPointerException();
-      }
-      
-      destination_ = value.getNumber();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Destination of graph transfer
-     * </pre>
-     *
-     * <code>.tensorflow.GraphTransferInfo.Destination destination = 7;</code>
-     */
-    public Builder clearDestination() {
-      
-      destination_ = 0;
-      onChanged();
-      return this;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.GraphTransferInfo)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.GraphTransferInfo)
-  private static final org.tensorflow.proto.framework.GraphTransferInfo DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.GraphTransferInfo();
-  }
-
-  public static org.tensorflow.proto.framework.GraphTransferInfo getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<GraphTransferInfo>
-      PARSER = new com.google.protobuf.AbstractParser<GraphTransferInfo>() {
-    @java.lang.Override
-    public GraphTransferInfo parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new GraphTransferInfo(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<GraphTransferInfo> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<GraphTransferInfo> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.framework.GraphTransferInfo getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphTransferInfoOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphTransferInfoOrBuilder.java
deleted file mode 100644
index d999e1d9d48..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphTransferInfoOrBuilder.java
+++ /dev/null
@@ -1,190 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/framework/graph_transfer_info.proto
-
-package org.tensorflow.proto.framework;
-
-public interface GraphTransferInfoOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.GraphTransferInfo)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <code>repeated .tensorflow.GraphTransferNodeInfo node_info = 1;</code>
-   */
-  java.util.List<org.tensorflow.proto.framework.GraphTransferNodeInfo> 
-      getNodeInfoList();
-  /**
-   * <code>repeated .tensorflow.GraphTransferNodeInfo node_info = 1;</code>
-   */
-  org.tensorflow.proto.framework.GraphTransferNodeInfo getNodeInfo(int index);
-  /**
-   * <code>repeated .tensorflow.GraphTransferNodeInfo node_info = 1;</code>
-   */
-  int getNodeInfoCount();
-  /**
-   * <code>repeated .tensorflow.GraphTransferNodeInfo node_info = 1;</code>
-   */
-  java.util.List<? extends org.tensorflow.proto.framework.GraphTransferNodeInfoOrBuilder> 
-      getNodeInfoOrBuilderList();
-  /**
-   * <code>repeated .tensorflow.GraphTransferNodeInfo node_info = 1;</code>
-   */
-  org.tensorflow.proto.framework.GraphTransferNodeInfoOrBuilder getNodeInfoOrBuilder(
-      int index);
-
-  /**
-   * <code>repeated .tensorflow.GraphTransferConstNodeInfo const_node_info = 2;</code>
-   */
-  java.util.List<org.tensorflow.proto.framework.GraphTransferConstNodeInfo> 
-      getConstNodeInfoList();
-  /**
-   * <code>repeated .tensorflow.GraphTransferConstNodeInfo const_node_info = 2;</code>
-   */
-  org.tensorflow.proto.framework.GraphTransferConstNodeInfo getConstNodeInfo(int index);
-  /**
-   * <code>repeated .tensorflow.GraphTransferConstNodeInfo const_node_info = 2;</code>
-   */
-  int getConstNodeInfoCount();
-  /**
-   * <code>repeated .tensorflow.GraphTransferConstNodeInfo const_node_info = 2;</code>
-   */
-  java.util.List<? extends org.tensorflow.proto.framework.GraphTransferConstNodeInfoOrBuilder> 
-      getConstNodeInfoOrBuilderList();
-  /**
-   * <code>repeated .tensorflow.GraphTransferConstNodeInfo const_node_info = 2;</code>
-   */
-  org.tensorflow.proto.framework.GraphTransferConstNodeInfoOrBuilder getConstNodeInfoOrBuilder(
-      int index);
-
-  /**
-   * <code>repeated .tensorflow.GraphTransferNodeInputInfo node_input_info = 3;</code>
-   */
-  java.util.List<org.tensorflow.proto.framework.GraphTransferNodeInputInfo> 
-      getNodeInputInfoList();
-  /**
-   * <code>repeated .tensorflow.GraphTransferNodeInputInfo node_input_info = 3;</code>
-   */
-  org.tensorflow.proto.framework.GraphTransferNodeInputInfo getNodeInputInfo(int index);
-  /**
-   * <code>repeated .tensorflow.GraphTransferNodeInputInfo node_input_info = 3;</code>
-   */
-  int getNodeInputInfoCount();
-  /**
-   * <code>repeated .tensorflow.GraphTransferNodeInputInfo node_input_info = 3;</code>
-   */
-  java.util.List<? extends org.tensorflow.proto.framework.GraphTransferNodeInputInfoOrBuilder> 
-      getNodeInputInfoOrBuilderList();
-  /**
-   * <code>repeated .tensorflow.GraphTransferNodeInputInfo node_input_info = 3;</code>
-   */
-  org.tensorflow.proto.framework.GraphTransferNodeInputInfoOrBuilder getNodeInputInfoOrBuilder(
-      int index);
-
-  /**
-   * <code>repeated .tensorflow.GraphTransferNodeOutputInfo node_output_info = 4;</code>
-   */
-  java.util.List<org.tensorflow.proto.framework.GraphTransferNodeOutputInfo> 
-      getNodeOutputInfoList();
-  /**
-   * <code>repeated .tensorflow.GraphTransferNodeOutputInfo node_output_info = 4;</code>
-   */
-  org.tensorflow.proto.framework.GraphTransferNodeOutputInfo getNodeOutputInfo(int index);
-  /**
-   * <code>repeated .tensorflow.GraphTransferNodeOutputInfo node_output_info = 4;</code>
-   */
-  int getNodeOutputInfoCount();
-  /**
-   * <code>repeated .tensorflow.GraphTransferNodeOutputInfo node_output_info = 4;</code>
-   */
-  java.util.List<? extends org.tensorflow.proto.framework.GraphTransferNodeOutputInfoOrBuilder> 
-      getNodeOutputInfoOrBuilderList();
-  /**
-   * <code>repeated .tensorflow.GraphTransferNodeOutputInfo node_output_info = 4;</code>
-   */
-  org.tensorflow.proto.framework.GraphTransferNodeOutputInfoOrBuilder getNodeOutputInfoOrBuilder(
-      int index);
-
-  /**
-   * <pre>
-   * Input Node parameters of transferred graph
-   * </pre>
-   *
-   * <code>repeated .tensorflow.GraphTransferGraphInputNodeInfo graph_input_node_info = 5;</code>
-   */
-  java.util.List<org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfo> 
-      getGraphInputNodeInfoList();
-  /**
-   * <pre>
-   * Input Node parameters of transferred graph
-   * </pre>
-   *
-   * <code>repeated .tensorflow.GraphTransferGraphInputNodeInfo graph_input_node_info = 5;</code>
-   */
-  org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfo getGraphInputNodeInfo(int index);
-  /**
-   * <pre>
-   * Input Node parameters of transferred graph
-   * </pre>
-   *
-   * <code>repeated .tensorflow.GraphTransferGraphInputNodeInfo graph_input_node_info = 5;</code>
-   */
-  int getGraphInputNodeInfoCount();
-  /**
-   * <pre>
-   * Input Node parameters of transferred graph
-   * </pre>
-   *
-   * <code>repeated .tensorflow.GraphTransferGraphInputNodeInfo graph_input_node_info = 5;</code>
-   */
-  java.util.List<? extends org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfoOrBuilder> 
-      getGraphInputNodeInfoOrBuilderList();
-  /**
-   * <pre>
-   * Input Node parameters of transferred graph
-   * </pre>
-   *
-   * <code>repeated .tensorflow.GraphTransferGraphInputNodeInfo graph_input_node_info = 5;</code>
-   */
-  org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfoOrBuilder getGraphInputNodeInfoOrBuilder(
-      int index);
-
-  /**
-   * <code>repeated .tensorflow.GraphTransferGraphOutputNodeInfo graph_output_node_info = 6;</code>
-   */
-  java.util.List<org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfo> 
-      getGraphOutputNodeInfoList();
-  /**
-   * <code>repeated .tensorflow.GraphTransferGraphOutputNodeInfo graph_output_node_info = 6;</code>
-   */
-  org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfo getGraphOutputNodeInfo(int index);
-  /**
-   * <code>repeated .tensorflow.GraphTransferGraphOutputNodeInfo graph_output_node_info = 6;</code>
-   */
-  int getGraphOutputNodeInfoCount();
-  /**
-   * <code>repeated .tensorflow.GraphTransferGraphOutputNodeInfo graph_output_node_info = 6;</code>
-   */
-  java.util.List<? extends org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfoOrBuilder> 
-      getGraphOutputNodeInfoOrBuilderList();
-  /**
-   * <code>repeated .tensorflow.GraphTransferGraphOutputNodeInfo graph_output_node_info = 6;</code>
-   */
-  org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfoOrBuilder getGraphOutputNodeInfoOrBuilder(
-      int index);
-
-  /**
-   * <pre>
-   * Destination of graph transfer
-   * </pre>
-   *
-   * <code>.tensorflow.GraphTransferInfo.Destination destination = 7;</code>
-   */
-  int getDestinationValue();
-  /**
-   * <pre>
-   * Destination of graph transfer
-   * </pre>
-   *
-   * <code>.tensorflow.GraphTransferInfo.Destination destination = 7;</code>
-   */
-  org.tensorflow.proto.framework.GraphTransferInfo.Destination getDestination();
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphTransferInfoProto.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphTransferInfoProto.java
deleted file mode 100644
index 750526c8897..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphTransferInfoProto.java
+++ /dev/null
@@ -1,163 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/framework/graph_transfer_info.proto
-
-package org.tensorflow.proto.framework;
-
-public final class GraphTransferInfoProto {
-  private GraphTransferInfoProto() {}
-  public static void registerAllExtensions(
-      com.google.protobuf.ExtensionRegistryLite registry) {
-  }
-
-  public static void registerAllExtensions(
-      com.google.protobuf.ExtensionRegistry registry) {
-    registerAllExtensions(
-        (com.google.protobuf.ExtensionRegistryLite) registry);
-  }
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_GraphTransferNodeInput_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_GraphTransferNodeInput_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_GraphTransferNodeInfo_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_GraphTransferNodeInfo_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_GraphTransferConstNodeInfo_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_GraphTransferConstNodeInfo_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_GraphTransferNodeInputInfo_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_GraphTransferNodeInputInfo_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_GraphTransferNodeOutputInfo_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_GraphTransferNodeOutputInfo_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_GraphTransferGraphInputNodeInfo_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_GraphTransferGraphInputNodeInfo_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_GraphTransferGraphOutputNodeInfo_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_GraphTransferGraphOutputNodeInfo_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_GraphTransferInfo_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_GraphTransferInfo_fieldAccessorTable;
-
-  public static com.google.protobuf.Descriptors.FileDescriptor
-      getDescriptor() {
-    return descriptor;
-  }
-  private static  com.google.protobuf.Descriptors.FileDescriptor
-      descriptor;
-  static {
-    java.lang.String[] descriptorData = {
-      "\n3tensorflow/core/framework/graph_transf" +
-      "er_info.proto\022\ntensorflow\032%tensorflow/co" +
-      "re/framework/types.proto\">\n\026GraphTransfe" +
-      "rNodeInput\022\017\n\007node_id\030\001 \001(\005\022\023\n\013output_po" +
-      "rt\030\002 \001(\005\"\233\001\n\025GraphTransferNodeInfo\022\014\n\004na" +
-      "me\030\001 \001(\t\022\017\n\007node_id\030\002 \001(\005\022\021\n\ttype_name\030\003" +
-      " \001(\t\022\021\n\tsoc_op_id\030\004 \001(\005\022\022\n\npadding_id\030\005 " +
-      "\001(\005\022\023\n\013input_count\030\006 \001(\005\022\024\n\014output_count" +
-      "\030\007 \001(\005\"}\n\032GraphTransferConstNodeInfo\022\014\n\004" +
-      "name\030\001 \001(\t\022\017\n\007node_id\030\002 \001(\005\022\r\n\005shape\030\003 \003" +
-      "(\003\022\014\n\004data\030\004 \001(\014\022#\n\005dtype\030\005 \001(\0162\024.tensor" +
-      "flow.DataType\"e\n\032GraphTransferNodeInputI" +
-      "nfo\022\017\n\007node_id\030\001 \001(\005\0226\n\nnode_input\030\002 \003(\013" +
-      "2\".tensorflow.GraphTransferNodeInput\"E\n\033" +
-      "GraphTransferNodeOutputInfo\022\017\n\007node_id\030\001" +
-      " \001(\005\022\025\n\rmax_byte_size\030\002 \003(\005\"c\n\037GraphTran" +
-      "sferGraphInputNodeInfo\022\014\n\004name\030\001 \001(\t\022\r\n\005" +
-      "shape\030\002 \003(\003\022#\n\005dtype\030\003 \001(\0162\024.tensorflow." +
-      "DataType\"d\n GraphTransferGraphOutputNode" +
-      "Info\022\014\n\004name\030\001 \001(\t\022\r\n\005shape\030\002 \003(\003\022#\n\005dty" +
-      "pe\030\003 \001(\0162\024.tensorflow.DataType\"\215\004\n\021Graph" +
-      "TransferInfo\0224\n\tnode_info\030\001 \003(\0132!.tensor" +
-      "flow.GraphTransferNodeInfo\022?\n\017const_node" +
-      "_info\030\002 \003(\0132&.tensorflow.GraphTransferCo" +
-      "nstNodeInfo\022?\n\017node_input_info\030\003 \003(\0132&.t" +
-      "ensorflow.GraphTransferNodeInputInfo\022A\n\020" +
-      "node_output_info\030\004 \003(\0132\'.tensorflow.Grap" +
-      "hTransferNodeOutputInfo\022J\n\025graph_input_n" +
-      "ode_info\030\005 \003(\0132+.tensorflow.GraphTransfe" +
-      "rGraphInputNodeInfo\022L\n\026graph_output_node" +
-      "_info\030\006 \003(\0132,.tensorflow.GraphTransferGr" +
-      "aphOutputNodeInfo\022>\n\013destination\030\007 \001(\0162)" +
-      ".tensorflow.GraphTransferInfo.Destinatio" +
-      "n\"#\n\013Destination\022\007\n\003NOP\020\000\022\013\n\007HEXAGON\020\001B\231" +
-      "\001\n\036org.tensorflow.proto.frameworkB\026Graph" +
-      "TransferInfoProtoP\001ZZgithub.com/tensorfl" +
-      "ow/tensorflow/tensorflow/go/core/framewo" +
-      "rk/graph_transfer_info_go_proto\370\001\001b\006prot" +
-      "o3"
-    };
-    descriptor = com.google.protobuf.Descriptors.FileDescriptor
-      .internalBuildGeneratedFileFrom(descriptorData,
-        new com.google.protobuf.Descriptors.FileDescriptor[] {
-          org.tensorflow.proto.framework.TypesProtos.getDescriptor(),
-        });
-    internal_static_tensorflow_GraphTransferNodeInput_descriptor =
-      getDescriptor().getMessageTypes().get(0);
-    internal_static_tensorflow_GraphTransferNodeInput_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_GraphTransferNodeInput_descriptor,
-        new java.lang.String[] { "NodeId", "OutputPort", });
-    internal_static_tensorflow_GraphTransferNodeInfo_descriptor =
-      getDescriptor().getMessageTypes().get(1);
-    internal_static_tensorflow_GraphTransferNodeInfo_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_GraphTransferNodeInfo_descriptor,
-        new java.lang.String[] { "Name", "NodeId", "TypeName", "SocOpId", "PaddingId", "InputCount", "OutputCount", });
-    internal_static_tensorflow_GraphTransferConstNodeInfo_descriptor =
-      getDescriptor().getMessageTypes().get(2);
-    internal_static_tensorflow_GraphTransferConstNodeInfo_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_GraphTransferConstNodeInfo_descriptor,
-        new java.lang.String[] { "Name", "NodeId", "Shape", "Data", "Dtype", });
-    internal_static_tensorflow_GraphTransferNodeInputInfo_descriptor =
-      getDescriptor().getMessageTypes().get(3);
-    internal_static_tensorflow_GraphTransferNodeInputInfo_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_GraphTransferNodeInputInfo_descriptor,
-        new java.lang.String[] { "NodeId", "NodeInput", });
-    internal_static_tensorflow_GraphTransferNodeOutputInfo_descriptor =
-      getDescriptor().getMessageTypes().get(4);
-    internal_static_tensorflow_GraphTransferNodeOutputInfo_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_GraphTransferNodeOutputInfo_descriptor,
-        new java.lang.String[] { "NodeId", "MaxByteSize", });
-    internal_static_tensorflow_GraphTransferGraphInputNodeInfo_descriptor =
-      getDescriptor().getMessageTypes().get(5);
-    internal_static_tensorflow_GraphTransferGraphInputNodeInfo_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_GraphTransferGraphInputNodeInfo_descriptor,
-        new java.lang.String[] { "Name", "Shape", "Dtype", });
-    internal_static_tensorflow_GraphTransferGraphOutputNodeInfo_descriptor =
-      getDescriptor().getMessageTypes().get(6);
-    internal_static_tensorflow_GraphTransferGraphOutputNodeInfo_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_GraphTransferGraphOutputNodeInfo_descriptor,
-        new java.lang.String[] { "Name", "Shape", "Dtype", });
-    internal_static_tensorflow_GraphTransferInfo_descriptor =
-      getDescriptor().getMessageTypes().get(7);
-    internal_static_tensorflow_GraphTransferInfo_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_GraphTransferInfo_descriptor,
-        new java.lang.String[] { "NodeInfo", "ConstNodeInfo", "NodeInputInfo", "NodeOutputInfo", "GraphInputNodeInfo", "GraphOutputNodeInfo", "Destination", });
-    org.tensorflow.proto.framework.TypesProtos.getDescriptor();
-  }
-
-  // @@protoc_insertion_point(outer_class_scope)
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphTransferNodeInfo.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphTransferNodeInfo.java
deleted file mode 100644
index d144359008c..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphTransferNodeInfo.java
+++ /dev/null
@@ -1,958 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/framework/graph_transfer_info.proto
-
-package org.tensorflow.proto.framework;
-
-/**
- * Protobuf type {@code tensorflow.GraphTransferNodeInfo}
- */
-public  final class GraphTransferNodeInfo extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.GraphTransferNodeInfo)
-    GraphTransferNodeInfoOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use GraphTransferNodeInfo.newBuilder() to construct.
-  private GraphTransferNodeInfo(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private GraphTransferNodeInfo() {
-    name_ = "";
-    typeName_ = "";
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new GraphTransferNodeInfo();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private GraphTransferNodeInfo(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 10: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            name_ = s;
-            break;
-          }
-          case 16: {
-
-            nodeId_ = input.readInt32();
-            break;
-          }
-          case 26: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            typeName_ = s;
-            break;
-          }
-          case 32: {
-
-            socOpId_ = input.readInt32();
-            break;
-          }
-          case 40: {
-
-            paddingId_ = input.readInt32();
-            break;
-          }
-          case 48: {
-
-            inputCount_ = input.readInt32();
-            break;
-          }
-          case 56: {
-
-            outputCount_ = input.readInt32();
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.framework.GraphTransferInfoProto.internal_static_tensorflow_GraphTransferNodeInfo_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.framework.GraphTransferInfoProto.internal_static_tensorflow_GraphTransferNodeInfo_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.framework.GraphTransferNodeInfo.class, org.tensorflow.proto.framework.GraphTransferNodeInfo.Builder.class);
-  }
-
-  public static final int NAME_FIELD_NUMBER = 1;
-  private volatile java.lang.Object name_;
-  /**
-   * <code>string name = 1;</code>
-   */
-  public java.lang.String getName() {
-    java.lang.Object ref = name_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      name_ = s;
-      return s;
-    }
-  }
-  /**
-   * <code>string name = 1;</code>
-   */
-  public com.google.protobuf.ByteString
-      getNameBytes() {
-    java.lang.Object ref = name_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      name_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int NODE_ID_FIELD_NUMBER = 2;
-  private int nodeId_;
-  /**
-   * <code>int32 node_id = 2;</code>
-   */
-  public int getNodeId() {
-    return nodeId_;
-  }
-
-  public static final int TYPE_NAME_FIELD_NUMBER = 3;
-  private volatile java.lang.Object typeName_;
-  /**
-   * <code>string type_name = 3;</code>
-   */
-  public java.lang.String getTypeName() {
-    java.lang.Object ref = typeName_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      typeName_ = s;
-      return s;
-    }
-  }
-  /**
-   * <code>string type_name = 3;</code>
-   */
-  public com.google.protobuf.ByteString
-      getTypeNameBytes() {
-    java.lang.Object ref = typeName_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      typeName_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int SOC_OP_ID_FIELD_NUMBER = 4;
-  private int socOpId_;
-  /**
-   * <code>int32 soc_op_id = 4;</code>
-   */
-  public int getSocOpId() {
-    return socOpId_;
-  }
-
-  public static final int PADDING_ID_FIELD_NUMBER = 5;
-  private int paddingId_;
-  /**
-   * <code>int32 padding_id = 5;</code>
-   */
-  public int getPaddingId() {
-    return paddingId_;
-  }
-
-  public static final int INPUT_COUNT_FIELD_NUMBER = 6;
-  private int inputCount_;
-  /**
-   * <code>int32 input_count = 6;</code>
-   */
-  public int getInputCount() {
-    return inputCount_;
-  }
-
-  public static final int OUTPUT_COUNT_FIELD_NUMBER = 7;
-  private int outputCount_;
-  /**
-   * <code>int32 output_count = 7;</code>
-   */
-  public int getOutputCount() {
-    return outputCount_;
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    if (!getNameBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 1, name_);
-    }
-    if (nodeId_ != 0) {
-      output.writeInt32(2, nodeId_);
-    }
-    if (!getTypeNameBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 3, typeName_);
-    }
-    if (socOpId_ != 0) {
-      output.writeInt32(4, socOpId_);
-    }
-    if (paddingId_ != 0) {
-      output.writeInt32(5, paddingId_);
-    }
-    if (inputCount_ != 0) {
-      output.writeInt32(6, inputCount_);
-    }
-    if (outputCount_ != 0) {
-      output.writeInt32(7, outputCount_);
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (!getNameBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, name_);
-    }
-    if (nodeId_ != 0) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt32Size(2, nodeId_);
-    }
-    if (!getTypeNameBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(3, typeName_);
-    }
-    if (socOpId_ != 0) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt32Size(4, socOpId_);
-    }
-    if (paddingId_ != 0) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt32Size(5, paddingId_);
-    }
-    if (inputCount_ != 0) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt32Size(6, inputCount_);
-    }
-    if (outputCount_ != 0) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt32Size(7, outputCount_);
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.framework.GraphTransferNodeInfo)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.framework.GraphTransferNodeInfo other = (org.tensorflow.proto.framework.GraphTransferNodeInfo) obj;
-
-    if (!getName()
-        .equals(other.getName())) return false;
-    if (getNodeId()
-        != other.getNodeId()) return false;
-    if (!getTypeName()
-        .equals(other.getTypeName())) return false;
-    if (getSocOpId()
-        != other.getSocOpId()) return false;
-    if (getPaddingId()
-        != other.getPaddingId()) return false;
-    if (getInputCount()
-        != other.getInputCount()) return false;
-    if (getOutputCount()
-        != other.getOutputCount()) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    hash = (37 * hash) + NAME_FIELD_NUMBER;
-    hash = (53 * hash) + getName().hashCode();
-    hash = (37 * hash) + NODE_ID_FIELD_NUMBER;
-    hash = (53 * hash) + getNodeId();
-    hash = (37 * hash) + TYPE_NAME_FIELD_NUMBER;
-    hash = (53 * hash) + getTypeName().hashCode();
-    hash = (37 * hash) + SOC_OP_ID_FIELD_NUMBER;
-    hash = (53 * hash) + getSocOpId();
-    hash = (37 * hash) + PADDING_ID_FIELD_NUMBER;
-    hash = (53 * hash) + getPaddingId();
-    hash = (37 * hash) + INPUT_COUNT_FIELD_NUMBER;
-    hash = (53 * hash) + getInputCount();
-    hash = (37 * hash) + OUTPUT_COUNT_FIELD_NUMBER;
-    hash = (53 * hash) + getOutputCount();
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.framework.GraphTransferNodeInfo parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.GraphTransferNodeInfo parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.GraphTransferNodeInfo parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.GraphTransferNodeInfo parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.GraphTransferNodeInfo parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.GraphTransferNodeInfo parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.GraphTransferNodeInfo parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.GraphTransferNodeInfo parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.GraphTransferNodeInfo parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.GraphTransferNodeInfo parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.GraphTransferNodeInfo parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.GraphTransferNodeInfo parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.framework.GraphTransferNodeInfo prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * Protobuf type {@code tensorflow.GraphTransferNodeInfo}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.GraphTransferNodeInfo)
-      org.tensorflow.proto.framework.GraphTransferNodeInfoOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.GraphTransferInfoProto.internal_static_tensorflow_GraphTransferNodeInfo_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.GraphTransferInfoProto.internal_static_tensorflow_GraphTransferNodeInfo_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.GraphTransferNodeInfo.class, org.tensorflow.proto.framework.GraphTransferNodeInfo.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.framework.GraphTransferNodeInfo.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      name_ = "";
-
-      nodeId_ = 0;
-
-      typeName_ = "";
-
-      socOpId_ = 0;
-
-      paddingId_ = 0;
-
-      inputCount_ = 0;
-
-      outputCount_ = 0;
-
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.framework.GraphTransferInfoProto.internal_static_tensorflow_GraphTransferNodeInfo_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.GraphTransferNodeInfo getDefaultInstanceForType() {
-      return org.tensorflow.proto.framework.GraphTransferNodeInfo.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.GraphTransferNodeInfo build() {
-      org.tensorflow.proto.framework.GraphTransferNodeInfo result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.GraphTransferNodeInfo buildPartial() {
-      org.tensorflow.proto.framework.GraphTransferNodeInfo result = new org.tensorflow.proto.framework.GraphTransferNodeInfo(this);
-      result.name_ = name_;
-      result.nodeId_ = nodeId_;
-      result.typeName_ = typeName_;
-      result.socOpId_ = socOpId_;
-      result.paddingId_ = paddingId_;
-      result.inputCount_ = inputCount_;
-      result.outputCount_ = outputCount_;
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.framework.GraphTransferNodeInfo) {
-        return mergeFrom((org.tensorflow.proto.framework.GraphTransferNodeInfo)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.framework.GraphTransferNodeInfo other) {
-      if (other == org.tensorflow.proto.framework.GraphTransferNodeInfo.getDefaultInstance()) return this;
-      if (!other.getName().isEmpty()) {
-        name_ = other.name_;
-        onChanged();
-      }
-      if (other.getNodeId() != 0) {
-        setNodeId(other.getNodeId());
-      }
-      if (!other.getTypeName().isEmpty()) {
-        typeName_ = other.typeName_;
-        onChanged();
-      }
-      if (other.getSocOpId() != 0) {
-        setSocOpId(other.getSocOpId());
-      }
-      if (other.getPaddingId() != 0) {
-        setPaddingId(other.getPaddingId());
-      }
-      if (other.getInputCount() != 0) {
-        setInputCount(other.getInputCount());
-      }
-      if (other.getOutputCount() != 0) {
-        setOutputCount(other.getOutputCount());
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.framework.GraphTransferNodeInfo parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.framework.GraphTransferNodeInfo) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-
-    private java.lang.Object name_ = "";
-    /**
-     * <code>string name = 1;</code>
-     */
-    public java.lang.String getName() {
-      java.lang.Object ref = name_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        name_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <code>string name = 1;</code>
-     */
-    public com.google.protobuf.ByteString
-        getNameBytes() {
-      java.lang.Object ref = name_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        name_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <code>string name = 1;</code>
-     */
-    public Builder setName(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      name_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>string name = 1;</code>
-     */
-    public Builder clearName() {
-      
-      name_ = getDefaultInstance().getName();
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>string name = 1;</code>
-     */
-    public Builder setNameBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      name_ = value;
-      onChanged();
-      return this;
-    }
-
-    private int nodeId_ ;
-    /**
-     * <code>int32 node_id = 2;</code>
-     */
-    public int getNodeId() {
-      return nodeId_;
-    }
-    /**
-     * <code>int32 node_id = 2;</code>
-     */
-    public Builder setNodeId(int value) {
-      
-      nodeId_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>int32 node_id = 2;</code>
-     */
-    public Builder clearNodeId() {
-      
-      nodeId_ = 0;
-      onChanged();
-      return this;
-    }
-
-    private java.lang.Object typeName_ = "";
-    /**
-     * <code>string type_name = 3;</code>
-     */
-    public java.lang.String getTypeName() {
-      java.lang.Object ref = typeName_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        typeName_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <code>string type_name = 3;</code>
-     */
-    public com.google.protobuf.ByteString
-        getTypeNameBytes() {
-      java.lang.Object ref = typeName_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        typeName_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <code>string type_name = 3;</code>
-     */
-    public Builder setTypeName(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      typeName_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>string type_name = 3;</code>
-     */
-    public Builder clearTypeName() {
-      
-      typeName_ = getDefaultInstance().getTypeName();
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>string type_name = 3;</code>
-     */
-    public Builder setTypeNameBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      typeName_ = value;
-      onChanged();
-      return this;
-    }
-
-    private int socOpId_ ;
-    /**
-     * <code>int32 soc_op_id = 4;</code>
-     */
-    public int getSocOpId() {
-      return socOpId_;
-    }
-    /**
-     * <code>int32 soc_op_id = 4;</code>
-     */
-    public Builder setSocOpId(int value) {
-      
-      socOpId_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>int32 soc_op_id = 4;</code>
-     */
-    public Builder clearSocOpId() {
-      
-      socOpId_ = 0;
-      onChanged();
-      return this;
-    }
-
-    private int paddingId_ ;
-    /**
-     * <code>int32 padding_id = 5;</code>
-     */
-    public int getPaddingId() {
-      return paddingId_;
-    }
-    /**
-     * <code>int32 padding_id = 5;</code>
-     */
-    public Builder setPaddingId(int value) {
-      
-      paddingId_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>int32 padding_id = 5;</code>
-     */
-    public Builder clearPaddingId() {
-      
-      paddingId_ = 0;
-      onChanged();
-      return this;
-    }
-
-    private int inputCount_ ;
-    /**
-     * <code>int32 input_count = 6;</code>
-     */
-    public int getInputCount() {
-      return inputCount_;
-    }
-    /**
-     * <code>int32 input_count = 6;</code>
-     */
-    public Builder setInputCount(int value) {
-      
-      inputCount_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>int32 input_count = 6;</code>
-     */
-    public Builder clearInputCount() {
-      
-      inputCount_ = 0;
-      onChanged();
-      return this;
-    }
-
-    private int outputCount_ ;
-    /**
-     * <code>int32 output_count = 7;</code>
-     */
-    public int getOutputCount() {
-      return outputCount_;
-    }
-    /**
-     * <code>int32 output_count = 7;</code>
-     */
-    public Builder setOutputCount(int value) {
-      
-      outputCount_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>int32 output_count = 7;</code>
-     */
-    public Builder clearOutputCount() {
-      
-      outputCount_ = 0;
-      onChanged();
-      return this;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.GraphTransferNodeInfo)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.GraphTransferNodeInfo)
-  private static final org.tensorflow.proto.framework.GraphTransferNodeInfo DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.GraphTransferNodeInfo();
-  }
-
-  public static org.tensorflow.proto.framework.GraphTransferNodeInfo getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<GraphTransferNodeInfo>
-      PARSER = new com.google.protobuf.AbstractParser<GraphTransferNodeInfo>() {
-    @java.lang.Override
-    public GraphTransferNodeInfo parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new GraphTransferNodeInfo(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<GraphTransferNodeInfo> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<GraphTransferNodeInfo> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.framework.GraphTransferNodeInfo getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphTransferNodeInfoOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphTransferNodeInfoOrBuilder.java
deleted file mode 100644
index 98656c6ee2c..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphTransferNodeInfoOrBuilder.java
+++ /dev/null
@@ -1,54 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/framework/graph_transfer_info.proto
-
-package org.tensorflow.proto.framework;
-
-public interface GraphTransferNodeInfoOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.GraphTransferNodeInfo)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <code>string name = 1;</code>
-   */
-  java.lang.String getName();
-  /**
-   * <code>string name = 1;</code>
-   */
-  com.google.protobuf.ByteString
-      getNameBytes();
-
-  /**
-   * <code>int32 node_id = 2;</code>
-   */
-  int getNodeId();
-
-  /**
-   * <code>string type_name = 3;</code>
-   */
-  java.lang.String getTypeName();
-  /**
-   * <code>string type_name = 3;</code>
-   */
-  com.google.protobuf.ByteString
-      getTypeNameBytes();
-
-  /**
-   * <code>int32 soc_op_id = 4;</code>
-   */
-  int getSocOpId();
-
-  /**
-   * <code>int32 padding_id = 5;</code>
-   */
-  int getPaddingId();
-
-  /**
-   * <code>int32 input_count = 6;</code>
-   */
-  int getInputCount();
-
-  /**
-   * <code>int32 output_count = 7;</code>
-   */
-  int getOutputCount();
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphTransferNodeInput.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphTransferNodeInput.java
deleted file mode 100644
index 77ad6b87dc8..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphTransferNodeInput.java
+++ /dev/null
@@ -1,533 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/framework/graph_transfer_info.proto
-
-package org.tensorflow.proto.framework;
-
-/**
- * Protobuf type {@code tensorflow.GraphTransferNodeInput}
- */
-public  final class GraphTransferNodeInput extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.GraphTransferNodeInput)
-    GraphTransferNodeInputOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use GraphTransferNodeInput.newBuilder() to construct.
-  private GraphTransferNodeInput(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private GraphTransferNodeInput() {
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new GraphTransferNodeInput();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private GraphTransferNodeInput(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 8: {
-
-            nodeId_ = input.readInt32();
-            break;
-          }
-          case 16: {
-
-            outputPort_ = input.readInt32();
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.framework.GraphTransferInfoProto.internal_static_tensorflow_GraphTransferNodeInput_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.framework.GraphTransferInfoProto.internal_static_tensorflow_GraphTransferNodeInput_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.framework.GraphTransferNodeInput.class, org.tensorflow.proto.framework.GraphTransferNodeInput.Builder.class);
-  }
-
-  public static final int NODE_ID_FIELD_NUMBER = 1;
-  private int nodeId_;
-  /**
-   * <code>int32 node_id = 1;</code>
-   */
-  public int getNodeId() {
-    return nodeId_;
-  }
-
-  public static final int OUTPUT_PORT_FIELD_NUMBER = 2;
-  private int outputPort_;
-  /**
-   * <code>int32 output_port = 2;</code>
-   */
-  public int getOutputPort() {
-    return outputPort_;
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    if (nodeId_ != 0) {
-      output.writeInt32(1, nodeId_);
-    }
-    if (outputPort_ != 0) {
-      output.writeInt32(2, outputPort_);
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (nodeId_ != 0) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt32Size(1, nodeId_);
-    }
-    if (outputPort_ != 0) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt32Size(2, outputPort_);
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.framework.GraphTransferNodeInput)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.framework.GraphTransferNodeInput other = (org.tensorflow.proto.framework.GraphTransferNodeInput) obj;
-
-    if (getNodeId()
-        != other.getNodeId()) return false;
-    if (getOutputPort()
-        != other.getOutputPort()) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    hash = (37 * hash) + NODE_ID_FIELD_NUMBER;
-    hash = (53 * hash) + getNodeId();
-    hash = (37 * hash) + OUTPUT_PORT_FIELD_NUMBER;
-    hash = (53 * hash) + getOutputPort();
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.framework.GraphTransferNodeInput parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.GraphTransferNodeInput parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.GraphTransferNodeInput parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.GraphTransferNodeInput parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.GraphTransferNodeInput parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.GraphTransferNodeInput parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.GraphTransferNodeInput parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.GraphTransferNodeInput parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.GraphTransferNodeInput parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.GraphTransferNodeInput parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.GraphTransferNodeInput parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.GraphTransferNodeInput parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.framework.GraphTransferNodeInput prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * Protobuf type {@code tensorflow.GraphTransferNodeInput}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.GraphTransferNodeInput)
-      org.tensorflow.proto.framework.GraphTransferNodeInputOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.GraphTransferInfoProto.internal_static_tensorflow_GraphTransferNodeInput_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.GraphTransferInfoProto.internal_static_tensorflow_GraphTransferNodeInput_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.GraphTransferNodeInput.class, org.tensorflow.proto.framework.GraphTransferNodeInput.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.framework.GraphTransferNodeInput.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      nodeId_ = 0;
-
-      outputPort_ = 0;
-
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.framework.GraphTransferInfoProto.internal_static_tensorflow_GraphTransferNodeInput_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.GraphTransferNodeInput getDefaultInstanceForType() {
-      return org.tensorflow.proto.framework.GraphTransferNodeInput.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.GraphTransferNodeInput build() {
-      org.tensorflow.proto.framework.GraphTransferNodeInput result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.GraphTransferNodeInput buildPartial() {
-      org.tensorflow.proto.framework.GraphTransferNodeInput result = new org.tensorflow.proto.framework.GraphTransferNodeInput(this);
-      result.nodeId_ = nodeId_;
-      result.outputPort_ = outputPort_;
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.framework.GraphTransferNodeInput) {
-        return mergeFrom((org.tensorflow.proto.framework.GraphTransferNodeInput)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.framework.GraphTransferNodeInput other) {
-      if (other == org.tensorflow.proto.framework.GraphTransferNodeInput.getDefaultInstance()) return this;
-      if (other.getNodeId() != 0) {
-        setNodeId(other.getNodeId());
-      }
-      if (other.getOutputPort() != 0) {
-        setOutputPort(other.getOutputPort());
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.framework.GraphTransferNodeInput parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.framework.GraphTransferNodeInput) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-
-    private int nodeId_ ;
-    /**
-     * <code>int32 node_id = 1;</code>
-     */
-    public int getNodeId() {
-      return nodeId_;
-    }
-    /**
-     * <code>int32 node_id = 1;</code>
-     */
-    public Builder setNodeId(int value) {
-      
-      nodeId_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>int32 node_id = 1;</code>
-     */
-    public Builder clearNodeId() {
-      
-      nodeId_ = 0;
-      onChanged();
-      return this;
-    }
-
-    private int outputPort_ ;
-    /**
-     * <code>int32 output_port = 2;</code>
-     */
-    public int getOutputPort() {
-      return outputPort_;
-    }
-    /**
-     * <code>int32 output_port = 2;</code>
-     */
-    public Builder setOutputPort(int value) {
-      
-      outputPort_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>int32 output_port = 2;</code>
-     */
-    public Builder clearOutputPort() {
-      
-      outputPort_ = 0;
-      onChanged();
-      return this;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.GraphTransferNodeInput)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.GraphTransferNodeInput)
-  private static final org.tensorflow.proto.framework.GraphTransferNodeInput DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.GraphTransferNodeInput();
-  }
-
-  public static org.tensorflow.proto.framework.GraphTransferNodeInput getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<GraphTransferNodeInput>
-      PARSER = new com.google.protobuf.AbstractParser<GraphTransferNodeInput>() {
-    @java.lang.Override
-    public GraphTransferNodeInput parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new GraphTransferNodeInput(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<GraphTransferNodeInput> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<GraphTransferNodeInput> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.framework.GraphTransferNodeInput getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphTransferNodeInputInfo.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphTransferNodeInputInfo.java
deleted file mode 100644
index 71d0552eacd..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphTransferNodeInputInfo.java
+++ /dev/null
@@ -1,822 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/framework/graph_transfer_info.proto
-
-package org.tensorflow.proto.framework;
-
-/**
- * Protobuf type {@code tensorflow.GraphTransferNodeInputInfo}
- */
-public  final class GraphTransferNodeInputInfo extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.GraphTransferNodeInputInfo)
-    GraphTransferNodeInputInfoOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use GraphTransferNodeInputInfo.newBuilder() to construct.
-  private GraphTransferNodeInputInfo(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private GraphTransferNodeInputInfo() {
-    nodeInput_ = java.util.Collections.emptyList();
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new GraphTransferNodeInputInfo();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private GraphTransferNodeInputInfo(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    int mutable_bitField0_ = 0;
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 8: {
-
-            nodeId_ = input.readInt32();
-            break;
-          }
-          case 18: {
-            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
-              nodeInput_ = new java.util.ArrayList<org.tensorflow.proto.framework.GraphTransferNodeInput>();
-              mutable_bitField0_ |= 0x00000001;
-            }
-            nodeInput_.add(
-                input.readMessage(org.tensorflow.proto.framework.GraphTransferNodeInput.parser(), extensionRegistry));
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      if (((mutable_bitField0_ & 0x00000001) != 0)) {
-        nodeInput_ = java.util.Collections.unmodifiableList(nodeInput_);
-      }
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.framework.GraphTransferInfoProto.internal_static_tensorflow_GraphTransferNodeInputInfo_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.framework.GraphTransferInfoProto.internal_static_tensorflow_GraphTransferNodeInputInfo_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.framework.GraphTransferNodeInputInfo.class, org.tensorflow.proto.framework.GraphTransferNodeInputInfo.Builder.class);
-  }
-
-  public static final int NODE_ID_FIELD_NUMBER = 1;
-  private int nodeId_;
-  /**
-   * <code>int32 node_id = 1;</code>
-   */
-  public int getNodeId() {
-    return nodeId_;
-  }
-
-  public static final int NODE_INPUT_FIELD_NUMBER = 2;
-  private java.util.List<org.tensorflow.proto.framework.GraphTransferNodeInput> nodeInput_;
-  /**
-   * <code>repeated .tensorflow.GraphTransferNodeInput node_input = 2;</code>
-   */
-  public java.util.List<org.tensorflow.proto.framework.GraphTransferNodeInput> getNodeInputList() {
-    return nodeInput_;
-  }
-  /**
-   * <code>repeated .tensorflow.GraphTransferNodeInput node_input = 2;</code>
-   */
-  public java.util.List<? extends org.tensorflow.proto.framework.GraphTransferNodeInputOrBuilder> 
-      getNodeInputOrBuilderList() {
-    return nodeInput_;
-  }
-  /**
-   * <code>repeated .tensorflow.GraphTransferNodeInput node_input = 2;</code>
-   */
-  public int getNodeInputCount() {
-    return nodeInput_.size();
-  }
-  /**
-   * <code>repeated .tensorflow.GraphTransferNodeInput node_input = 2;</code>
-   */
-  public org.tensorflow.proto.framework.GraphTransferNodeInput getNodeInput(int index) {
-    return nodeInput_.get(index);
-  }
-  /**
-   * <code>repeated .tensorflow.GraphTransferNodeInput node_input = 2;</code>
-   */
-  public org.tensorflow.proto.framework.GraphTransferNodeInputOrBuilder getNodeInputOrBuilder(
-      int index) {
-    return nodeInput_.get(index);
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    if (nodeId_ != 0) {
-      output.writeInt32(1, nodeId_);
-    }
-    for (int i = 0; i < nodeInput_.size(); i++) {
-      output.writeMessage(2, nodeInput_.get(i));
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (nodeId_ != 0) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt32Size(1, nodeId_);
-    }
-    for (int i = 0; i < nodeInput_.size(); i++) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(2, nodeInput_.get(i));
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.framework.GraphTransferNodeInputInfo)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.framework.GraphTransferNodeInputInfo other = (org.tensorflow.proto.framework.GraphTransferNodeInputInfo) obj;
-
-    if (getNodeId()
-        != other.getNodeId()) return false;
-    if (!getNodeInputList()
-        .equals(other.getNodeInputList())) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    hash = (37 * hash) + NODE_ID_FIELD_NUMBER;
-    hash = (53 * hash) + getNodeId();
-    if (getNodeInputCount() > 0) {
-      hash = (37 * hash) + NODE_INPUT_FIELD_NUMBER;
-      hash = (53 * hash) + getNodeInputList().hashCode();
-    }
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.framework.GraphTransferNodeInputInfo parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.GraphTransferNodeInputInfo parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.GraphTransferNodeInputInfo parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.GraphTransferNodeInputInfo parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.GraphTransferNodeInputInfo parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.GraphTransferNodeInputInfo parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.GraphTransferNodeInputInfo parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.GraphTransferNodeInputInfo parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.GraphTransferNodeInputInfo parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.GraphTransferNodeInputInfo parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.GraphTransferNodeInputInfo parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.GraphTransferNodeInputInfo parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.framework.GraphTransferNodeInputInfo prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * Protobuf type {@code tensorflow.GraphTransferNodeInputInfo}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.GraphTransferNodeInputInfo)
-      org.tensorflow.proto.framework.GraphTransferNodeInputInfoOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.GraphTransferInfoProto.internal_static_tensorflow_GraphTransferNodeInputInfo_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.GraphTransferInfoProto.internal_static_tensorflow_GraphTransferNodeInputInfo_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.GraphTransferNodeInputInfo.class, org.tensorflow.proto.framework.GraphTransferNodeInputInfo.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.framework.GraphTransferNodeInputInfo.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-        getNodeInputFieldBuilder();
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      nodeId_ = 0;
-
-      if (nodeInputBuilder_ == null) {
-        nodeInput_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000001);
-      } else {
-        nodeInputBuilder_.clear();
-      }
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.framework.GraphTransferInfoProto.internal_static_tensorflow_GraphTransferNodeInputInfo_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.GraphTransferNodeInputInfo getDefaultInstanceForType() {
-      return org.tensorflow.proto.framework.GraphTransferNodeInputInfo.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.GraphTransferNodeInputInfo build() {
-      org.tensorflow.proto.framework.GraphTransferNodeInputInfo result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.GraphTransferNodeInputInfo buildPartial() {
-      org.tensorflow.proto.framework.GraphTransferNodeInputInfo result = new org.tensorflow.proto.framework.GraphTransferNodeInputInfo(this);
-      int from_bitField0_ = bitField0_;
-      result.nodeId_ = nodeId_;
-      if (nodeInputBuilder_ == null) {
-        if (((bitField0_ & 0x00000001) != 0)) {
-          nodeInput_ = java.util.Collections.unmodifiableList(nodeInput_);
-          bitField0_ = (bitField0_ & ~0x00000001);
-        }
-        result.nodeInput_ = nodeInput_;
-      } else {
-        result.nodeInput_ = nodeInputBuilder_.build();
-      }
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.framework.GraphTransferNodeInputInfo) {
-        return mergeFrom((org.tensorflow.proto.framework.GraphTransferNodeInputInfo)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.framework.GraphTransferNodeInputInfo other) {
-      if (other == org.tensorflow.proto.framework.GraphTransferNodeInputInfo.getDefaultInstance()) return this;
-      if (other.getNodeId() != 0) {
-        setNodeId(other.getNodeId());
-      }
-      if (nodeInputBuilder_ == null) {
-        if (!other.nodeInput_.isEmpty()) {
-          if (nodeInput_.isEmpty()) {
-            nodeInput_ = other.nodeInput_;
-            bitField0_ = (bitField0_ & ~0x00000001);
-          } else {
-            ensureNodeInputIsMutable();
-            nodeInput_.addAll(other.nodeInput_);
-          }
-          onChanged();
-        }
-      } else {
-        if (!other.nodeInput_.isEmpty()) {
-          if (nodeInputBuilder_.isEmpty()) {
-            nodeInputBuilder_.dispose();
-            nodeInputBuilder_ = null;
-            nodeInput_ = other.nodeInput_;
-            bitField0_ = (bitField0_ & ~0x00000001);
-            nodeInputBuilder_ = 
-              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
-                 getNodeInputFieldBuilder() : null;
-          } else {
-            nodeInputBuilder_.addAllMessages(other.nodeInput_);
-          }
-        }
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.framework.GraphTransferNodeInputInfo parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.framework.GraphTransferNodeInputInfo) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-    private int bitField0_;
-
-    private int nodeId_ ;
-    /**
-     * <code>int32 node_id = 1;</code>
-     */
-    public int getNodeId() {
-      return nodeId_;
-    }
-    /**
-     * <code>int32 node_id = 1;</code>
-     */
-    public Builder setNodeId(int value) {
-      
-      nodeId_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>int32 node_id = 1;</code>
-     */
-    public Builder clearNodeId() {
-      
-      nodeId_ = 0;
-      onChanged();
-      return this;
-    }
-
-    private java.util.List<org.tensorflow.proto.framework.GraphTransferNodeInput> nodeInput_ =
-      java.util.Collections.emptyList();
-    private void ensureNodeInputIsMutable() {
-      if (!((bitField0_ & 0x00000001) != 0)) {
-        nodeInput_ = new java.util.ArrayList<org.tensorflow.proto.framework.GraphTransferNodeInput>(nodeInput_);
-        bitField0_ |= 0x00000001;
-       }
-    }
-
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.framework.GraphTransferNodeInput, org.tensorflow.proto.framework.GraphTransferNodeInput.Builder, org.tensorflow.proto.framework.GraphTransferNodeInputOrBuilder> nodeInputBuilder_;
-
-    /**
-     * <code>repeated .tensorflow.GraphTransferNodeInput node_input = 2;</code>
-     */
-    public java.util.List<org.tensorflow.proto.framework.GraphTransferNodeInput> getNodeInputList() {
-      if (nodeInputBuilder_ == null) {
-        return java.util.Collections.unmodifiableList(nodeInput_);
-      } else {
-        return nodeInputBuilder_.getMessageList();
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.GraphTransferNodeInput node_input = 2;</code>
-     */
-    public int getNodeInputCount() {
-      if (nodeInputBuilder_ == null) {
-        return nodeInput_.size();
-      } else {
-        return nodeInputBuilder_.getCount();
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.GraphTransferNodeInput node_input = 2;</code>
-     */
-    public org.tensorflow.proto.framework.GraphTransferNodeInput getNodeInput(int index) {
-      if (nodeInputBuilder_ == null) {
-        return nodeInput_.get(index);
-      } else {
-        return nodeInputBuilder_.getMessage(index);
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.GraphTransferNodeInput node_input = 2;</code>
-     */
-    public Builder setNodeInput(
-        int index, org.tensorflow.proto.framework.GraphTransferNodeInput value) {
-      if (nodeInputBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureNodeInputIsMutable();
-        nodeInput_.set(index, value);
-        onChanged();
-      } else {
-        nodeInputBuilder_.setMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.GraphTransferNodeInput node_input = 2;</code>
-     */
-    public Builder setNodeInput(
-        int index, org.tensorflow.proto.framework.GraphTransferNodeInput.Builder builderForValue) {
-      if (nodeInputBuilder_ == null) {
-        ensureNodeInputIsMutable();
-        nodeInput_.set(index, builderForValue.build());
-        onChanged();
-      } else {
-        nodeInputBuilder_.setMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.GraphTransferNodeInput node_input = 2;</code>
-     */
-    public Builder addNodeInput(org.tensorflow.proto.framework.GraphTransferNodeInput value) {
-      if (nodeInputBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureNodeInputIsMutable();
-        nodeInput_.add(value);
-        onChanged();
-      } else {
-        nodeInputBuilder_.addMessage(value);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.GraphTransferNodeInput node_input = 2;</code>
-     */
-    public Builder addNodeInput(
-        int index, org.tensorflow.proto.framework.GraphTransferNodeInput value) {
-      if (nodeInputBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureNodeInputIsMutable();
-        nodeInput_.add(index, value);
-        onChanged();
-      } else {
-        nodeInputBuilder_.addMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.GraphTransferNodeInput node_input = 2;</code>
-     */
-    public Builder addNodeInput(
-        org.tensorflow.proto.framework.GraphTransferNodeInput.Builder builderForValue) {
-      if (nodeInputBuilder_ == null) {
-        ensureNodeInputIsMutable();
-        nodeInput_.add(builderForValue.build());
-        onChanged();
-      } else {
-        nodeInputBuilder_.addMessage(builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.GraphTransferNodeInput node_input = 2;</code>
-     */
-    public Builder addNodeInput(
-        int index, org.tensorflow.proto.framework.GraphTransferNodeInput.Builder builderForValue) {
-      if (nodeInputBuilder_ == null) {
-        ensureNodeInputIsMutable();
-        nodeInput_.add(index, builderForValue.build());
-        onChanged();
-      } else {
-        nodeInputBuilder_.addMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.GraphTransferNodeInput node_input = 2;</code>
-     */
-    public Builder addAllNodeInput(
-        java.lang.Iterable<? extends org.tensorflow.proto.framework.GraphTransferNodeInput> values) {
-      if (nodeInputBuilder_ == null) {
-        ensureNodeInputIsMutable();
-        com.google.protobuf.AbstractMessageLite.Builder.addAll(
-            values, nodeInput_);
-        onChanged();
-      } else {
-        nodeInputBuilder_.addAllMessages(values);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.GraphTransferNodeInput node_input = 2;</code>
-     */
-    public Builder clearNodeInput() {
-      if (nodeInputBuilder_ == null) {
-        nodeInput_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000001);
-        onChanged();
-      } else {
-        nodeInputBuilder_.clear();
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.GraphTransferNodeInput node_input = 2;</code>
-     */
-    public Builder removeNodeInput(int index) {
-      if (nodeInputBuilder_ == null) {
-        ensureNodeInputIsMutable();
-        nodeInput_.remove(index);
-        onChanged();
-      } else {
-        nodeInputBuilder_.remove(index);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.GraphTransferNodeInput node_input = 2;</code>
-     */
-    public org.tensorflow.proto.framework.GraphTransferNodeInput.Builder getNodeInputBuilder(
-        int index) {
-      return getNodeInputFieldBuilder().getBuilder(index);
-    }
-    /**
-     * <code>repeated .tensorflow.GraphTransferNodeInput node_input = 2;</code>
-     */
-    public org.tensorflow.proto.framework.GraphTransferNodeInputOrBuilder getNodeInputOrBuilder(
-        int index) {
-      if (nodeInputBuilder_ == null) {
-        return nodeInput_.get(index);  } else {
-        return nodeInputBuilder_.getMessageOrBuilder(index);
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.GraphTransferNodeInput node_input = 2;</code>
-     */
-    public java.util.List<? extends org.tensorflow.proto.framework.GraphTransferNodeInputOrBuilder> 
-         getNodeInputOrBuilderList() {
-      if (nodeInputBuilder_ != null) {
-        return nodeInputBuilder_.getMessageOrBuilderList();
-      } else {
-        return java.util.Collections.unmodifiableList(nodeInput_);
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.GraphTransferNodeInput node_input = 2;</code>
-     */
-    public org.tensorflow.proto.framework.GraphTransferNodeInput.Builder addNodeInputBuilder() {
-      return getNodeInputFieldBuilder().addBuilder(
-          org.tensorflow.proto.framework.GraphTransferNodeInput.getDefaultInstance());
-    }
-    /**
-     * <code>repeated .tensorflow.GraphTransferNodeInput node_input = 2;</code>
-     */
-    public org.tensorflow.proto.framework.GraphTransferNodeInput.Builder addNodeInputBuilder(
-        int index) {
-      return getNodeInputFieldBuilder().addBuilder(
-          index, org.tensorflow.proto.framework.GraphTransferNodeInput.getDefaultInstance());
-    }
-    /**
-     * <code>repeated .tensorflow.GraphTransferNodeInput node_input = 2;</code>
-     */
-    public java.util.List<org.tensorflow.proto.framework.GraphTransferNodeInput.Builder> 
-         getNodeInputBuilderList() {
-      return getNodeInputFieldBuilder().getBuilderList();
-    }
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.framework.GraphTransferNodeInput, org.tensorflow.proto.framework.GraphTransferNodeInput.Builder, org.tensorflow.proto.framework.GraphTransferNodeInputOrBuilder> 
-        getNodeInputFieldBuilder() {
-      if (nodeInputBuilder_ == null) {
-        nodeInputBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
-            org.tensorflow.proto.framework.GraphTransferNodeInput, org.tensorflow.proto.framework.GraphTransferNodeInput.Builder, org.tensorflow.proto.framework.GraphTransferNodeInputOrBuilder>(
-                nodeInput_,
-                ((bitField0_ & 0x00000001) != 0),
-                getParentForChildren(),
-                isClean());
-        nodeInput_ = null;
-      }
-      return nodeInputBuilder_;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.GraphTransferNodeInputInfo)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.GraphTransferNodeInputInfo)
-  private static final org.tensorflow.proto.framework.GraphTransferNodeInputInfo DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.GraphTransferNodeInputInfo();
-  }
-
-  public static org.tensorflow.proto.framework.GraphTransferNodeInputInfo getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<GraphTransferNodeInputInfo>
-      PARSER = new com.google.protobuf.AbstractParser<GraphTransferNodeInputInfo>() {
-    @java.lang.Override
-    public GraphTransferNodeInputInfo parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new GraphTransferNodeInputInfo(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<GraphTransferNodeInputInfo> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<GraphTransferNodeInputInfo> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.framework.GraphTransferNodeInputInfo getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphTransferNodeInputInfoOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphTransferNodeInputInfoOrBuilder.java
deleted file mode 100644
index 552a182d67b..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphTransferNodeInputInfoOrBuilder.java
+++ /dev/null
@@ -1,38 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/framework/graph_transfer_info.proto
-
-package org.tensorflow.proto.framework;
-
-public interface GraphTransferNodeInputInfoOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.GraphTransferNodeInputInfo)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <code>int32 node_id = 1;</code>
-   */
-  int getNodeId();
-
-  /**
-   * <code>repeated .tensorflow.GraphTransferNodeInput node_input = 2;</code>
-   */
-  java.util.List<org.tensorflow.proto.framework.GraphTransferNodeInput> 
-      getNodeInputList();
-  /**
-   * <code>repeated .tensorflow.GraphTransferNodeInput node_input = 2;</code>
-   */
-  org.tensorflow.proto.framework.GraphTransferNodeInput getNodeInput(int index);
-  /**
-   * <code>repeated .tensorflow.GraphTransferNodeInput node_input = 2;</code>
-   */
-  int getNodeInputCount();
-  /**
-   * <code>repeated .tensorflow.GraphTransferNodeInput node_input = 2;</code>
-   */
-  java.util.List<? extends org.tensorflow.proto.framework.GraphTransferNodeInputOrBuilder> 
-      getNodeInputOrBuilderList();
-  /**
-   * <code>repeated .tensorflow.GraphTransferNodeInput node_input = 2;</code>
-   */
-  org.tensorflow.proto.framework.GraphTransferNodeInputOrBuilder getNodeInputOrBuilder(
-      int index);
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphTransferNodeInputOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphTransferNodeInputOrBuilder.java
deleted file mode 100644
index 971ffed8443..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphTransferNodeInputOrBuilder.java
+++ /dev/null
@@ -1,19 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/framework/graph_transfer_info.proto
-
-package org.tensorflow.proto.framework;
-
-public interface GraphTransferNodeInputOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.GraphTransferNodeInput)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <code>int32 node_id = 1;</code>
-   */
-  int getNodeId();
-
-  /**
-   * <code>int32 output_port = 2;</code>
-   */
-  int getOutputPort();
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphTransferNodeOutputInfo.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphTransferNodeOutputInfo.java
deleted file mode 100644
index c39062bebdb..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphTransferNodeOutputInfo.java
+++ /dev/null
@@ -1,639 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/framework/graph_transfer_info.proto
-
-package org.tensorflow.proto.framework;
-
-/**
- * Protobuf type {@code tensorflow.GraphTransferNodeOutputInfo}
- */
-public  final class GraphTransferNodeOutputInfo extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.GraphTransferNodeOutputInfo)
-    GraphTransferNodeOutputInfoOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use GraphTransferNodeOutputInfo.newBuilder() to construct.
-  private GraphTransferNodeOutputInfo(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private GraphTransferNodeOutputInfo() {
-    maxByteSize_ = emptyIntList();
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new GraphTransferNodeOutputInfo();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private GraphTransferNodeOutputInfo(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    int mutable_bitField0_ = 0;
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 8: {
-
-            nodeId_ = input.readInt32();
-            break;
-          }
-          case 16: {
-            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
-              maxByteSize_ = newIntList();
-              mutable_bitField0_ |= 0x00000001;
-            }
-            maxByteSize_.addInt(input.readInt32());
-            break;
-          }
-          case 18: {
-            int length = input.readRawVarint32();
-            int limit = input.pushLimit(length);
-            if (!((mutable_bitField0_ & 0x00000001) != 0) && input.getBytesUntilLimit() > 0) {
-              maxByteSize_ = newIntList();
-              mutable_bitField0_ |= 0x00000001;
-            }
-            while (input.getBytesUntilLimit() > 0) {
-              maxByteSize_.addInt(input.readInt32());
-            }
-            input.popLimit(limit);
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      if (((mutable_bitField0_ & 0x00000001) != 0)) {
-        maxByteSize_.makeImmutable(); // C
-      }
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.framework.GraphTransferInfoProto.internal_static_tensorflow_GraphTransferNodeOutputInfo_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.framework.GraphTransferInfoProto.internal_static_tensorflow_GraphTransferNodeOutputInfo_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.framework.GraphTransferNodeOutputInfo.class, org.tensorflow.proto.framework.GraphTransferNodeOutputInfo.Builder.class);
-  }
-
-  public static final int NODE_ID_FIELD_NUMBER = 1;
-  private int nodeId_;
-  /**
-   * <code>int32 node_id = 1;</code>
-   */
-  public int getNodeId() {
-    return nodeId_;
-  }
-
-  public static final int MAX_BYTE_SIZE_FIELD_NUMBER = 2;
-  private com.google.protobuf.Internal.IntList maxByteSize_;
-  /**
-   * <code>repeated int32 max_byte_size = 2;</code>
-   */
-  public java.util.List<java.lang.Integer>
-      getMaxByteSizeList() {
-    return maxByteSize_;
-  }
-  /**
-   * <code>repeated int32 max_byte_size = 2;</code>
-   */
-  public int getMaxByteSizeCount() {
-    return maxByteSize_.size();
-  }
-  /**
-   * <code>repeated int32 max_byte_size = 2;</code>
-   */
-  public int getMaxByteSize(int index) {
-    return maxByteSize_.getInt(index);
-  }
-  private int maxByteSizeMemoizedSerializedSize = -1;
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    getSerializedSize();
-    if (nodeId_ != 0) {
-      output.writeInt32(1, nodeId_);
-    }
-    if (getMaxByteSizeList().size() > 0) {
-      output.writeUInt32NoTag(18);
-      output.writeUInt32NoTag(maxByteSizeMemoizedSerializedSize);
-    }
-    for (int i = 0; i < maxByteSize_.size(); i++) {
-      output.writeInt32NoTag(maxByteSize_.getInt(i));
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (nodeId_ != 0) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt32Size(1, nodeId_);
-    }
-    {
-      int dataSize = 0;
-      for (int i = 0; i < maxByteSize_.size(); i++) {
-        dataSize += com.google.protobuf.CodedOutputStream
-          .computeInt32SizeNoTag(maxByteSize_.getInt(i));
-      }
-      size += dataSize;
-      if (!getMaxByteSizeList().isEmpty()) {
-        size += 1;
-        size += com.google.protobuf.CodedOutputStream
-            .computeInt32SizeNoTag(dataSize);
-      }
-      maxByteSizeMemoizedSerializedSize = dataSize;
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.framework.GraphTransferNodeOutputInfo)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.framework.GraphTransferNodeOutputInfo other = (org.tensorflow.proto.framework.GraphTransferNodeOutputInfo) obj;
-
-    if (getNodeId()
-        != other.getNodeId()) return false;
-    if (!getMaxByteSizeList()
-        .equals(other.getMaxByteSizeList())) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    hash = (37 * hash) + NODE_ID_FIELD_NUMBER;
-    hash = (53 * hash) + getNodeId();
-    if (getMaxByteSizeCount() > 0) {
-      hash = (37 * hash) + MAX_BYTE_SIZE_FIELD_NUMBER;
-      hash = (53 * hash) + getMaxByteSizeList().hashCode();
-    }
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.framework.GraphTransferNodeOutputInfo parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.GraphTransferNodeOutputInfo parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.GraphTransferNodeOutputInfo parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.GraphTransferNodeOutputInfo parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.GraphTransferNodeOutputInfo parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.GraphTransferNodeOutputInfo parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.GraphTransferNodeOutputInfo parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.GraphTransferNodeOutputInfo parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.GraphTransferNodeOutputInfo parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.GraphTransferNodeOutputInfo parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.GraphTransferNodeOutputInfo parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.GraphTransferNodeOutputInfo parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.framework.GraphTransferNodeOutputInfo prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * Protobuf type {@code tensorflow.GraphTransferNodeOutputInfo}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.GraphTransferNodeOutputInfo)
-      org.tensorflow.proto.framework.GraphTransferNodeOutputInfoOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.GraphTransferInfoProto.internal_static_tensorflow_GraphTransferNodeOutputInfo_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.GraphTransferInfoProto.internal_static_tensorflow_GraphTransferNodeOutputInfo_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.GraphTransferNodeOutputInfo.class, org.tensorflow.proto.framework.GraphTransferNodeOutputInfo.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.framework.GraphTransferNodeOutputInfo.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      nodeId_ = 0;
-
-      maxByteSize_ = emptyIntList();
-      bitField0_ = (bitField0_ & ~0x00000001);
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.framework.GraphTransferInfoProto.internal_static_tensorflow_GraphTransferNodeOutputInfo_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.GraphTransferNodeOutputInfo getDefaultInstanceForType() {
-      return org.tensorflow.proto.framework.GraphTransferNodeOutputInfo.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.GraphTransferNodeOutputInfo build() {
-      org.tensorflow.proto.framework.GraphTransferNodeOutputInfo result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.GraphTransferNodeOutputInfo buildPartial() {
-      org.tensorflow.proto.framework.GraphTransferNodeOutputInfo result = new org.tensorflow.proto.framework.GraphTransferNodeOutputInfo(this);
-      int from_bitField0_ = bitField0_;
-      result.nodeId_ = nodeId_;
-      if (((bitField0_ & 0x00000001) != 0)) {
-        maxByteSize_.makeImmutable();
-        bitField0_ = (bitField0_ & ~0x00000001);
-      }
-      result.maxByteSize_ = maxByteSize_;
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.framework.GraphTransferNodeOutputInfo) {
-        return mergeFrom((org.tensorflow.proto.framework.GraphTransferNodeOutputInfo)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.framework.GraphTransferNodeOutputInfo other) {
-      if (other == org.tensorflow.proto.framework.GraphTransferNodeOutputInfo.getDefaultInstance()) return this;
-      if (other.getNodeId() != 0) {
-        setNodeId(other.getNodeId());
-      }
-      if (!other.maxByteSize_.isEmpty()) {
-        if (maxByteSize_.isEmpty()) {
-          maxByteSize_ = other.maxByteSize_;
-          bitField0_ = (bitField0_ & ~0x00000001);
-        } else {
-          ensureMaxByteSizeIsMutable();
-          maxByteSize_.addAll(other.maxByteSize_);
-        }
-        onChanged();
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.framework.GraphTransferNodeOutputInfo parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.framework.GraphTransferNodeOutputInfo) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-    private int bitField0_;
-
-    private int nodeId_ ;
-    /**
-     * <code>int32 node_id = 1;</code>
-     */
-    public int getNodeId() {
-      return nodeId_;
-    }
-    /**
-     * <code>int32 node_id = 1;</code>
-     */
-    public Builder setNodeId(int value) {
-      
-      nodeId_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>int32 node_id = 1;</code>
-     */
-    public Builder clearNodeId() {
-      
-      nodeId_ = 0;
-      onChanged();
-      return this;
-    }
-
-    private com.google.protobuf.Internal.IntList maxByteSize_ = emptyIntList();
-    private void ensureMaxByteSizeIsMutable() {
-      if (!((bitField0_ & 0x00000001) != 0)) {
-        maxByteSize_ = mutableCopy(maxByteSize_);
-        bitField0_ |= 0x00000001;
-       }
-    }
-    /**
-     * <code>repeated int32 max_byte_size = 2;</code>
-     */
-    public java.util.List<java.lang.Integer>
-        getMaxByteSizeList() {
-      return ((bitField0_ & 0x00000001) != 0) ?
-               java.util.Collections.unmodifiableList(maxByteSize_) : maxByteSize_;
-    }
-    /**
-     * <code>repeated int32 max_byte_size = 2;</code>
-     */
-    public int getMaxByteSizeCount() {
-      return maxByteSize_.size();
-    }
-    /**
-     * <code>repeated int32 max_byte_size = 2;</code>
-     */
-    public int getMaxByteSize(int index) {
-      return maxByteSize_.getInt(index);
-    }
-    /**
-     * <code>repeated int32 max_byte_size = 2;</code>
-     */
-    public Builder setMaxByteSize(
-        int index, int value) {
-      ensureMaxByteSizeIsMutable();
-      maxByteSize_.setInt(index, value);
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>repeated int32 max_byte_size = 2;</code>
-     */
-    public Builder addMaxByteSize(int value) {
-      ensureMaxByteSizeIsMutable();
-      maxByteSize_.addInt(value);
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>repeated int32 max_byte_size = 2;</code>
-     */
-    public Builder addAllMaxByteSize(
-        java.lang.Iterable<? extends java.lang.Integer> values) {
-      ensureMaxByteSizeIsMutable();
-      com.google.protobuf.AbstractMessageLite.Builder.addAll(
-          values, maxByteSize_);
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>repeated int32 max_byte_size = 2;</code>
-     */
-    public Builder clearMaxByteSize() {
-      maxByteSize_ = emptyIntList();
-      bitField0_ = (bitField0_ & ~0x00000001);
-      onChanged();
-      return this;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.GraphTransferNodeOutputInfo)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.GraphTransferNodeOutputInfo)
-  private static final org.tensorflow.proto.framework.GraphTransferNodeOutputInfo DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.GraphTransferNodeOutputInfo();
-  }
-
-  public static org.tensorflow.proto.framework.GraphTransferNodeOutputInfo getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<GraphTransferNodeOutputInfo>
-      PARSER = new com.google.protobuf.AbstractParser<GraphTransferNodeOutputInfo>() {
-    @java.lang.Override
-    public GraphTransferNodeOutputInfo parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new GraphTransferNodeOutputInfo(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<GraphTransferNodeOutputInfo> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<GraphTransferNodeOutputInfo> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.framework.GraphTransferNodeOutputInfo getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphTransferNodeOutputInfoOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphTransferNodeOutputInfoOrBuilder.java
deleted file mode 100644
index 7721bc2dc6a..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphTransferNodeOutputInfoOrBuilder.java
+++ /dev/null
@@ -1,27 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/framework/graph_transfer_info.proto
-
-package org.tensorflow.proto.framework;
-
-public interface GraphTransferNodeOutputInfoOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.GraphTransferNodeOutputInfo)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <code>int32 node_id = 1;</code>
-   */
-  int getNodeId();
-
-  /**
-   * <code>repeated int32 max_byte_size = 2;</code>
-   */
-  java.util.List<java.lang.Integer> getMaxByteSizeList();
-  /**
-   * <code>repeated int32 max_byte_size = 2;</code>
-   */
-  int getMaxByteSizeCount();
-  /**
-   * <code>repeated int32 max_byte_size = 2;</code>
-   */
-  int getMaxByteSize(int index);
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/HistogramProto.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/HistogramProto.java
deleted file mode 100644
index e5e19354670..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/HistogramProto.java
+++ /dev/null
@@ -1,1120 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/framework/summary.proto
-
-package org.tensorflow.proto.framework;
-
-/**
- * <pre>
- * Serialization format for histogram module in
- * core/lib/histogram/histogram.h
- * </pre>
- *
- * Protobuf type {@code tensorflow.HistogramProto}
- */
-public  final class HistogramProto extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.HistogramProto)
-    HistogramProtoOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use HistogramProto.newBuilder() to construct.
-  private HistogramProto(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private HistogramProto() {
-    bucketLimit_ = emptyDoubleList();
-    bucket_ = emptyDoubleList();
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new HistogramProto();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private HistogramProto(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    int mutable_bitField0_ = 0;
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 9: {
-
-            min_ = input.readDouble();
-            break;
-          }
-          case 17: {
-
-            max_ = input.readDouble();
-            break;
-          }
-          case 25: {
-
-            num_ = input.readDouble();
-            break;
-          }
-          case 33: {
-
-            sum_ = input.readDouble();
-            break;
-          }
-          case 41: {
-
-            sumSquares_ = input.readDouble();
-            break;
-          }
-          case 49: {
-            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
-              bucketLimit_ = newDoubleList();
-              mutable_bitField0_ |= 0x00000001;
-            }
-            bucketLimit_.addDouble(input.readDouble());
-            break;
-          }
-          case 50: {
-            int length = input.readRawVarint32();
-            int limit = input.pushLimit(length);
-            if (!((mutable_bitField0_ & 0x00000001) != 0) && input.getBytesUntilLimit() > 0) {
-              bucketLimit_ = newDoubleList();
-              mutable_bitField0_ |= 0x00000001;
-            }
-            while (input.getBytesUntilLimit() > 0) {
-              bucketLimit_.addDouble(input.readDouble());
-            }
-            input.popLimit(limit);
-            break;
-          }
-          case 57: {
-            if (!((mutable_bitField0_ & 0x00000002) != 0)) {
-              bucket_ = newDoubleList();
-              mutable_bitField0_ |= 0x00000002;
-            }
-            bucket_.addDouble(input.readDouble());
-            break;
-          }
-          case 58: {
-            int length = input.readRawVarint32();
-            int limit = input.pushLimit(length);
-            if (!((mutable_bitField0_ & 0x00000002) != 0) && input.getBytesUntilLimit() > 0) {
-              bucket_ = newDoubleList();
-              mutable_bitField0_ |= 0x00000002;
-            }
-            while (input.getBytesUntilLimit() > 0) {
-              bucket_.addDouble(input.readDouble());
-            }
-            input.popLimit(limit);
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      if (((mutable_bitField0_ & 0x00000001) != 0)) {
-        bucketLimit_.makeImmutable(); // C
-      }
-      if (((mutable_bitField0_ & 0x00000002) != 0)) {
-        bucket_.makeImmutable(); // C
-      }
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.framework.SummaryProtos.internal_static_tensorflow_HistogramProto_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.framework.SummaryProtos.internal_static_tensorflow_HistogramProto_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.framework.HistogramProto.class, org.tensorflow.proto.framework.HistogramProto.Builder.class);
-  }
-
-  public static final int MIN_FIELD_NUMBER = 1;
-  private double min_;
-  /**
-   * <code>double min = 1;</code>
-   */
-  public double getMin() {
-    return min_;
-  }
-
-  public static final int MAX_FIELD_NUMBER = 2;
-  private double max_;
-  /**
-   * <code>double max = 2;</code>
-   */
-  public double getMax() {
-    return max_;
-  }
-
-  public static final int NUM_FIELD_NUMBER = 3;
-  private double num_;
-  /**
-   * <code>double num = 3;</code>
-   */
-  public double getNum() {
-    return num_;
-  }
-
-  public static final int SUM_FIELD_NUMBER = 4;
-  private double sum_;
-  /**
-   * <code>double sum = 4;</code>
-   */
-  public double getSum() {
-    return sum_;
-  }
-
-  public static final int SUM_SQUARES_FIELD_NUMBER = 5;
-  private double sumSquares_;
-  /**
-   * <code>double sum_squares = 5;</code>
-   */
-  public double getSumSquares() {
-    return sumSquares_;
-  }
-
-  public static final int BUCKET_LIMIT_FIELD_NUMBER = 6;
-  private com.google.protobuf.Internal.DoubleList bucketLimit_;
-  /**
-   * <pre>
-   * Parallel arrays encoding the bucket boundaries and the bucket values.
-   * bucket(i) is the count for the bucket i.  The range for
-   * a bucket is:
-   *   i == 0:  -DBL_MAX .. bucket_limit(0)
-   *   i != 0:  bucket_limit(i-1) .. bucket_limit(i)
-   * </pre>
-   *
-   * <code>repeated double bucket_limit = 6 [packed = true];</code>
-   */
-  public java.util.List<java.lang.Double>
-      getBucketLimitList() {
-    return bucketLimit_;
-  }
-  /**
-   * <pre>
-   * Parallel arrays encoding the bucket boundaries and the bucket values.
-   * bucket(i) is the count for the bucket i.  The range for
-   * a bucket is:
-   *   i == 0:  -DBL_MAX .. bucket_limit(0)
-   *   i != 0:  bucket_limit(i-1) .. bucket_limit(i)
-   * </pre>
-   *
-   * <code>repeated double bucket_limit = 6 [packed = true];</code>
-   */
-  public int getBucketLimitCount() {
-    return bucketLimit_.size();
-  }
-  /**
-   * <pre>
-   * Parallel arrays encoding the bucket boundaries and the bucket values.
-   * bucket(i) is the count for the bucket i.  The range for
-   * a bucket is:
-   *   i == 0:  -DBL_MAX .. bucket_limit(0)
-   *   i != 0:  bucket_limit(i-1) .. bucket_limit(i)
-   * </pre>
-   *
-   * <code>repeated double bucket_limit = 6 [packed = true];</code>
-   */
-  public double getBucketLimit(int index) {
-    return bucketLimit_.getDouble(index);
-  }
-  private int bucketLimitMemoizedSerializedSize = -1;
-
-  public static final int BUCKET_FIELD_NUMBER = 7;
-  private com.google.protobuf.Internal.DoubleList bucket_;
-  /**
-   * <code>repeated double bucket = 7 [packed = true];</code>
-   */
-  public java.util.List<java.lang.Double>
-      getBucketList() {
-    return bucket_;
-  }
-  /**
-   * <code>repeated double bucket = 7 [packed = true];</code>
-   */
-  public int getBucketCount() {
-    return bucket_.size();
-  }
-  /**
-   * <code>repeated double bucket = 7 [packed = true];</code>
-   */
-  public double getBucket(int index) {
-    return bucket_.getDouble(index);
-  }
-  private int bucketMemoizedSerializedSize = -1;
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    getSerializedSize();
-    if (min_ != 0D) {
-      output.writeDouble(1, min_);
-    }
-    if (max_ != 0D) {
-      output.writeDouble(2, max_);
-    }
-    if (num_ != 0D) {
-      output.writeDouble(3, num_);
-    }
-    if (sum_ != 0D) {
-      output.writeDouble(4, sum_);
-    }
-    if (sumSquares_ != 0D) {
-      output.writeDouble(5, sumSquares_);
-    }
-    if (getBucketLimitList().size() > 0) {
-      output.writeUInt32NoTag(50);
-      output.writeUInt32NoTag(bucketLimitMemoizedSerializedSize);
-    }
-    for (int i = 0; i < bucketLimit_.size(); i++) {
-      output.writeDoubleNoTag(bucketLimit_.getDouble(i));
-    }
-    if (getBucketList().size() > 0) {
-      output.writeUInt32NoTag(58);
-      output.writeUInt32NoTag(bucketMemoizedSerializedSize);
-    }
-    for (int i = 0; i < bucket_.size(); i++) {
-      output.writeDoubleNoTag(bucket_.getDouble(i));
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (min_ != 0D) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeDoubleSize(1, min_);
-    }
-    if (max_ != 0D) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeDoubleSize(2, max_);
-    }
-    if (num_ != 0D) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeDoubleSize(3, num_);
-    }
-    if (sum_ != 0D) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeDoubleSize(4, sum_);
-    }
-    if (sumSquares_ != 0D) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeDoubleSize(5, sumSquares_);
-    }
-    {
-      int dataSize = 0;
-      dataSize = 8 * getBucketLimitList().size();
-      size += dataSize;
-      if (!getBucketLimitList().isEmpty()) {
-        size += 1;
-        size += com.google.protobuf.CodedOutputStream
-            .computeInt32SizeNoTag(dataSize);
-      }
-      bucketLimitMemoizedSerializedSize = dataSize;
-    }
-    {
-      int dataSize = 0;
-      dataSize = 8 * getBucketList().size();
-      size += dataSize;
-      if (!getBucketList().isEmpty()) {
-        size += 1;
-        size += com.google.protobuf.CodedOutputStream
-            .computeInt32SizeNoTag(dataSize);
-      }
-      bucketMemoizedSerializedSize = dataSize;
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.framework.HistogramProto)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.framework.HistogramProto other = (org.tensorflow.proto.framework.HistogramProto) obj;
-
-    if (java.lang.Double.doubleToLongBits(getMin())
-        != java.lang.Double.doubleToLongBits(
-            other.getMin())) return false;
-    if (java.lang.Double.doubleToLongBits(getMax())
-        != java.lang.Double.doubleToLongBits(
-            other.getMax())) return false;
-    if (java.lang.Double.doubleToLongBits(getNum())
-        != java.lang.Double.doubleToLongBits(
-            other.getNum())) return false;
-    if (java.lang.Double.doubleToLongBits(getSum())
-        != java.lang.Double.doubleToLongBits(
-            other.getSum())) return false;
-    if (java.lang.Double.doubleToLongBits(getSumSquares())
-        != java.lang.Double.doubleToLongBits(
-            other.getSumSquares())) return false;
-    if (!getBucketLimitList()
-        .equals(other.getBucketLimitList())) return false;
-    if (!getBucketList()
-        .equals(other.getBucketList())) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    hash = (37 * hash) + MIN_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-        java.lang.Double.doubleToLongBits(getMin()));
-    hash = (37 * hash) + MAX_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-        java.lang.Double.doubleToLongBits(getMax()));
-    hash = (37 * hash) + NUM_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-        java.lang.Double.doubleToLongBits(getNum()));
-    hash = (37 * hash) + SUM_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-        java.lang.Double.doubleToLongBits(getSum()));
-    hash = (37 * hash) + SUM_SQUARES_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-        java.lang.Double.doubleToLongBits(getSumSquares()));
-    if (getBucketLimitCount() > 0) {
-      hash = (37 * hash) + BUCKET_LIMIT_FIELD_NUMBER;
-      hash = (53 * hash) + getBucketLimitList().hashCode();
-    }
-    if (getBucketCount() > 0) {
-      hash = (37 * hash) + BUCKET_FIELD_NUMBER;
-      hash = (53 * hash) + getBucketList().hashCode();
-    }
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.framework.HistogramProto parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.HistogramProto parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.HistogramProto parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.HistogramProto parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.HistogramProto parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.HistogramProto parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.HistogramProto parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.HistogramProto parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.HistogramProto parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.HistogramProto parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.HistogramProto parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.HistogramProto parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.framework.HistogramProto prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * <pre>
-   * Serialization format for histogram module in
-   * core/lib/histogram/histogram.h
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.HistogramProto}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.HistogramProto)
-      org.tensorflow.proto.framework.HistogramProtoOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.SummaryProtos.internal_static_tensorflow_HistogramProto_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.SummaryProtos.internal_static_tensorflow_HistogramProto_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.HistogramProto.class, org.tensorflow.proto.framework.HistogramProto.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.framework.HistogramProto.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      min_ = 0D;
-
-      max_ = 0D;
-
-      num_ = 0D;
-
-      sum_ = 0D;
-
-      sumSquares_ = 0D;
-
-      bucketLimit_ = emptyDoubleList();
-      bitField0_ = (bitField0_ & ~0x00000001);
-      bucket_ = emptyDoubleList();
-      bitField0_ = (bitField0_ & ~0x00000002);
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.framework.SummaryProtos.internal_static_tensorflow_HistogramProto_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.HistogramProto getDefaultInstanceForType() {
-      return org.tensorflow.proto.framework.HistogramProto.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.HistogramProto build() {
-      org.tensorflow.proto.framework.HistogramProto result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.HistogramProto buildPartial() {
-      org.tensorflow.proto.framework.HistogramProto result = new org.tensorflow.proto.framework.HistogramProto(this);
-      int from_bitField0_ = bitField0_;
-      result.min_ = min_;
-      result.max_ = max_;
-      result.num_ = num_;
-      result.sum_ = sum_;
-      result.sumSquares_ = sumSquares_;
-      if (((bitField0_ & 0x00000001) != 0)) {
-        bucketLimit_.makeImmutable();
-        bitField0_ = (bitField0_ & ~0x00000001);
-      }
-      result.bucketLimit_ = bucketLimit_;
-      if (((bitField0_ & 0x00000002) != 0)) {
-        bucket_.makeImmutable();
-        bitField0_ = (bitField0_ & ~0x00000002);
-      }
-      result.bucket_ = bucket_;
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.framework.HistogramProto) {
-        return mergeFrom((org.tensorflow.proto.framework.HistogramProto)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.framework.HistogramProto other) {
-      if (other == org.tensorflow.proto.framework.HistogramProto.getDefaultInstance()) return this;
-      if (other.getMin() != 0D) {
-        setMin(other.getMin());
-      }
-      if (other.getMax() != 0D) {
-        setMax(other.getMax());
-      }
-      if (other.getNum() != 0D) {
-        setNum(other.getNum());
-      }
-      if (other.getSum() != 0D) {
-        setSum(other.getSum());
-      }
-      if (other.getSumSquares() != 0D) {
-        setSumSquares(other.getSumSquares());
-      }
-      if (!other.bucketLimit_.isEmpty()) {
-        if (bucketLimit_.isEmpty()) {
-          bucketLimit_ = other.bucketLimit_;
-          bitField0_ = (bitField0_ & ~0x00000001);
-        } else {
-          ensureBucketLimitIsMutable();
-          bucketLimit_.addAll(other.bucketLimit_);
-        }
-        onChanged();
-      }
-      if (!other.bucket_.isEmpty()) {
-        if (bucket_.isEmpty()) {
-          bucket_ = other.bucket_;
-          bitField0_ = (bitField0_ & ~0x00000002);
-        } else {
-          ensureBucketIsMutable();
-          bucket_.addAll(other.bucket_);
-        }
-        onChanged();
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.framework.HistogramProto parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.framework.HistogramProto) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-    private int bitField0_;
-
-    private double min_ ;
-    /**
-     * <code>double min = 1;</code>
-     */
-    public double getMin() {
-      return min_;
-    }
-    /**
-     * <code>double min = 1;</code>
-     */
-    public Builder setMin(double value) {
-      
-      min_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>double min = 1;</code>
-     */
-    public Builder clearMin() {
-      
-      min_ = 0D;
-      onChanged();
-      return this;
-    }
-
-    private double max_ ;
-    /**
-     * <code>double max = 2;</code>
-     */
-    public double getMax() {
-      return max_;
-    }
-    /**
-     * <code>double max = 2;</code>
-     */
-    public Builder setMax(double value) {
-      
-      max_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>double max = 2;</code>
-     */
-    public Builder clearMax() {
-      
-      max_ = 0D;
-      onChanged();
-      return this;
-    }
-
-    private double num_ ;
-    /**
-     * <code>double num = 3;</code>
-     */
-    public double getNum() {
-      return num_;
-    }
-    /**
-     * <code>double num = 3;</code>
-     */
-    public Builder setNum(double value) {
-      
-      num_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>double num = 3;</code>
-     */
-    public Builder clearNum() {
-      
-      num_ = 0D;
-      onChanged();
-      return this;
-    }
-
-    private double sum_ ;
-    /**
-     * <code>double sum = 4;</code>
-     */
-    public double getSum() {
-      return sum_;
-    }
-    /**
-     * <code>double sum = 4;</code>
-     */
-    public Builder setSum(double value) {
-      
-      sum_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>double sum = 4;</code>
-     */
-    public Builder clearSum() {
-      
-      sum_ = 0D;
-      onChanged();
-      return this;
-    }
-
-    private double sumSquares_ ;
-    /**
-     * <code>double sum_squares = 5;</code>
-     */
-    public double getSumSquares() {
-      return sumSquares_;
-    }
-    /**
-     * <code>double sum_squares = 5;</code>
-     */
-    public Builder setSumSquares(double value) {
-      
-      sumSquares_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>double sum_squares = 5;</code>
-     */
-    public Builder clearSumSquares() {
-      
-      sumSquares_ = 0D;
-      onChanged();
-      return this;
-    }
-
-    private com.google.protobuf.Internal.DoubleList bucketLimit_ = emptyDoubleList();
-    private void ensureBucketLimitIsMutable() {
-      if (!((bitField0_ & 0x00000001) != 0)) {
-        bucketLimit_ = mutableCopy(bucketLimit_);
-        bitField0_ |= 0x00000001;
-       }
-    }
-    /**
-     * <pre>
-     * Parallel arrays encoding the bucket boundaries and the bucket values.
-     * bucket(i) is the count for the bucket i.  The range for
-     * a bucket is:
-     *   i == 0:  -DBL_MAX .. bucket_limit(0)
-     *   i != 0:  bucket_limit(i-1) .. bucket_limit(i)
-     * </pre>
-     *
-     * <code>repeated double bucket_limit = 6 [packed = true];</code>
-     */
-    public java.util.List<java.lang.Double>
-        getBucketLimitList() {
-      return ((bitField0_ & 0x00000001) != 0) ?
-               java.util.Collections.unmodifiableList(bucketLimit_) : bucketLimit_;
-    }
-    /**
-     * <pre>
-     * Parallel arrays encoding the bucket boundaries and the bucket values.
-     * bucket(i) is the count for the bucket i.  The range for
-     * a bucket is:
-     *   i == 0:  -DBL_MAX .. bucket_limit(0)
-     *   i != 0:  bucket_limit(i-1) .. bucket_limit(i)
-     * </pre>
-     *
-     * <code>repeated double bucket_limit = 6 [packed = true];</code>
-     */
-    public int getBucketLimitCount() {
-      return bucketLimit_.size();
-    }
-    /**
-     * <pre>
-     * Parallel arrays encoding the bucket boundaries and the bucket values.
-     * bucket(i) is the count for the bucket i.  The range for
-     * a bucket is:
-     *   i == 0:  -DBL_MAX .. bucket_limit(0)
-     *   i != 0:  bucket_limit(i-1) .. bucket_limit(i)
-     * </pre>
-     *
-     * <code>repeated double bucket_limit = 6 [packed = true];</code>
-     */
-    public double getBucketLimit(int index) {
-      return bucketLimit_.getDouble(index);
-    }
-    /**
-     * <pre>
-     * Parallel arrays encoding the bucket boundaries and the bucket values.
-     * bucket(i) is the count for the bucket i.  The range for
-     * a bucket is:
-     *   i == 0:  -DBL_MAX .. bucket_limit(0)
-     *   i != 0:  bucket_limit(i-1) .. bucket_limit(i)
-     * </pre>
-     *
-     * <code>repeated double bucket_limit = 6 [packed = true];</code>
-     */
-    public Builder setBucketLimit(
-        int index, double value) {
-      ensureBucketLimitIsMutable();
-      bucketLimit_.setDouble(index, value);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Parallel arrays encoding the bucket boundaries and the bucket values.
-     * bucket(i) is the count for the bucket i.  The range for
-     * a bucket is:
-     *   i == 0:  -DBL_MAX .. bucket_limit(0)
-     *   i != 0:  bucket_limit(i-1) .. bucket_limit(i)
-     * </pre>
-     *
-     * <code>repeated double bucket_limit = 6 [packed = true];</code>
-     */
-    public Builder addBucketLimit(double value) {
-      ensureBucketLimitIsMutable();
-      bucketLimit_.addDouble(value);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Parallel arrays encoding the bucket boundaries and the bucket values.
-     * bucket(i) is the count for the bucket i.  The range for
-     * a bucket is:
-     *   i == 0:  -DBL_MAX .. bucket_limit(0)
-     *   i != 0:  bucket_limit(i-1) .. bucket_limit(i)
-     * </pre>
-     *
-     * <code>repeated double bucket_limit = 6 [packed = true];</code>
-     */
-    public Builder addAllBucketLimit(
-        java.lang.Iterable<? extends java.lang.Double> values) {
-      ensureBucketLimitIsMutable();
-      com.google.protobuf.AbstractMessageLite.Builder.addAll(
-          values, bucketLimit_);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Parallel arrays encoding the bucket boundaries and the bucket values.
-     * bucket(i) is the count for the bucket i.  The range for
-     * a bucket is:
-     *   i == 0:  -DBL_MAX .. bucket_limit(0)
-     *   i != 0:  bucket_limit(i-1) .. bucket_limit(i)
-     * </pre>
-     *
-     * <code>repeated double bucket_limit = 6 [packed = true];</code>
-     */
-    public Builder clearBucketLimit() {
-      bucketLimit_ = emptyDoubleList();
-      bitField0_ = (bitField0_ & ~0x00000001);
-      onChanged();
-      return this;
-    }
-
-    private com.google.protobuf.Internal.DoubleList bucket_ = emptyDoubleList();
-    private void ensureBucketIsMutable() {
-      if (!((bitField0_ & 0x00000002) != 0)) {
-        bucket_ = mutableCopy(bucket_);
-        bitField0_ |= 0x00000002;
-       }
-    }
-    /**
-     * <code>repeated double bucket = 7 [packed = true];</code>
-     */
-    public java.util.List<java.lang.Double>
-        getBucketList() {
-      return ((bitField0_ & 0x00000002) != 0) ?
-               java.util.Collections.unmodifiableList(bucket_) : bucket_;
-    }
-    /**
-     * <code>repeated double bucket = 7 [packed = true];</code>
-     */
-    public int getBucketCount() {
-      return bucket_.size();
-    }
-    /**
-     * <code>repeated double bucket = 7 [packed = true];</code>
-     */
-    public double getBucket(int index) {
-      return bucket_.getDouble(index);
-    }
-    /**
-     * <code>repeated double bucket = 7 [packed = true];</code>
-     */
-    public Builder setBucket(
-        int index, double value) {
-      ensureBucketIsMutable();
-      bucket_.setDouble(index, value);
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>repeated double bucket = 7 [packed = true];</code>
-     */
-    public Builder addBucket(double value) {
-      ensureBucketIsMutable();
-      bucket_.addDouble(value);
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>repeated double bucket = 7 [packed = true];</code>
-     */
-    public Builder addAllBucket(
-        java.lang.Iterable<? extends java.lang.Double> values) {
-      ensureBucketIsMutable();
-      com.google.protobuf.AbstractMessageLite.Builder.addAll(
-          values, bucket_);
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>repeated double bucket = 7 [packed = true];</code>
-     */
-    public Builder clearBucket() {
-      bucket_ = emptyDoubleList();
-      bitField0_ = (bitField0_ & ~0x00000002);
-      onChanged();
-      return this;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.HistogramProto)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.HistogramProto)
-  private static final org.tensorflow.proto.framework.HistogramProto DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.HistogramProto();
-  }
-
-  public static org.tensorflow.proto.framework.HistogramProto getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<HistogramProto>
-      PARSER = new com.google.protobuf.AbstractParser<HistogramProto>() {
-    @java.lang.Override
-    public HistogramProto parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new HistogramProto(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<HistogramProto> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<HistogramProto> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.framework.HistogramProto getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/HistogramProtoOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/HistogramProtoOrBuilder.java
deleted file mode 100644
index d76afe24d19..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/HistogramProtoOrBuilder.java
+++ /dev/null
@@ -1,84 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/framework/summary.proto
-
-package org.tensorflow.proto.framework;
-
-public interface HistogramProtoOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.HistogramProto)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <code>double min = 1;</code>
-   */
-  double getMin();
-
-  /**
-   * <code>double max = 2;</code>
-   */
-  double getMax();
-
-  /**
-   * <code>double num = 3;</code>
-   */
-  double getNum();
-
-  /**
-   * <code>double sum = 4;</code>
-   */
-  double getSum();
-
-  /**
-   * <code>double sum_squares = 5;</code>
-   */
-  double getSumSquares();
-
-  /**
-   * <pre>
-   * Parallel arrays encoding the bucket boundaries and the bucket values.
-   * bucket(i) is the count for the bucket i.  The range for
-   * a bucket is:
-   *   i == 0:  -DBL_MAX .. bucket_limit(0)
-   *   i != 0:  bucket_limit(i-1) .. bucket_limit(i)
-   * </pre>
-   *
-   * <code>repeated double bucket_limit = 6 [packed = true];</code>
-   */
-  java.util.List<java.lang.Double> getBucketLimitList();
-  /**
-   * <pre>
-   * Parallel arrays encoding the bucket boundaries and the bucket values.
-   * bucket(i) is the count for the bucket i.  The range for
-   * a bucket is:
-   *   i == 0:  -DBL_MAX .. bucket_limit(0)
-   *   i != 0:  bucket_limit(i-1) .. bucket_limit(i)
-   * </pre>
-   *
-   * <code>repeated double bucket_limit = 6 [packed = true];</code>
-   */
-  int getBucketLimitCount();
-  /**
-   * <pre>
-   * Parallel arrays encoding the bucket boundaries and the bucket values.
-   * bucket(i) is the count for the bucket i.  The range for
-   * a bucket is:
-   *   i == 0:  -DBL_MAX .. bucket_limit(0)
-   *   i != 0:  bucket_limit(i-1) .. bucket_limit(i)
-   * </pre>
-   *
-   * <code>repeated double bucket_limit = 6 [packed = true];</code>
-   */
-  double getBucketLimit(int index);
-
-  /**
-   * <code>repeated double bucket = 7 [packed = true];</code>
-   */
-  java.util.List<java.lang.Double> getBucketList();
-  /**
-   * <code>repeated double bucket = 7 [packed = true];</code>
-   */
-  int getBucketCount();
-  /**
-   * <code>repeated double bucket = 7 [packed = true];</code>
-   */
-  double getBucket(int index);
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/InterconnectLink.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/InterconnectLink.java
deleted file mode 100644
index 89fbd7a8dec..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/InterconnectLink.java
+++ /dev/null
@@ -1,660 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/framework/device_attributes.proto
-
-package org.tensorflow.proto.framework;
-
-/**
- * Protobuf type {@code tensorflow.InterconnectLink}
- */
-public  final class InterconnectLink extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.InterconnectLink)
-    InterconnectLinkOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use InterconnectLink.newBuilder() to construct.
-  private InterconnectLink(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private InterconnectLink() {
-    type_ = "";
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new InterconnectLink();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private InterconnectLink(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 8: {
-
-            deviceId_ = input.readInt32();
-            break;
-          }
-          case 18: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            type_ = s;
-            break;
-          }
-          case 24: {
-
-            strength_ = input.readInt32();
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.framework.DeviceAttributesProtos.internal_static_tensorflow_InterconnectLink_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.framework.DeviceAttributesProtos.internal_static_tensorflow_InterconnectLink_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.framework.InterconnectLink.class, org.tensorflow.proto.framework.InterconnectLink.Builder.class);
-  }
-
-  public static final int DEVICE_ID_FIELD_NUMBER = 1;
-  private int deviceId_;
-  /**
-   * <code>int32 device_id = 1;</code>
-   */
-  public int getDeviceId() {
-    return deviceId_;
-  }
-
-  public static final int TYPE_FIELD_NUMBER = 2;
-  private volatile java.lang.Object type_;
-  /**
-   * <code>string type = 2;</code>
-   */
-  public java.lang.String getType() {
-    java.lang.Object ref = type_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      type_ = s;
-      return s;
-    }
-  }
-  /**
-   * <code>string type = 2;</code>
-   */
-  public com.google.protobuf.ByteString
-      getTypeBytes() {
-    java.lang.Object ref = type_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      type_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int STRENGTH_FIELD_NUMBER = 3;
-  private int strength_;
-  /**
-   * <code>int32 strength = 3;</code>
-   */
-  public int getStrength() {
-    return strength_;
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    if (deviceId_ != 0) {
-      output.writeInt32(1, deviceId_);
-    }
-    if (!getTypeBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 2, type_);
-    }
-    if (strength_ != 0) {
-      output.writeInt32(3, strength_);
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (deviceId_ != 0) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt32Size(1, deviceId_);
-    }
-    if (!getTypeBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(2, type_);
-    }
-    if (strength_ != 0) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt32Size(3, strength_);
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.framework.InterconnectLink)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.framework.InterconnectLink other = (org.tensorflow.proto.framework.InterconnectLink) obj;
-
-    if (getDeviceId()
-        != other.getDeviceId()) return false;
-    if (!getType()
-        .equals(other.getType())) return false;
-    if (getStrength()
-        != other.getStrength()) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    hash = (37 * hash) + DEVICE_ID_FIELD_NUMBER;
-    hash = (53 * hash) + getDeviceId();
-    hash = (37 * hash) + TYPE_FIELD_NUMBER;
-    hash = (53 * hash) + getType().hashCode();
-    hash = (37 * hash) + STRENGTH_FIELD_NUMBER;
-    hash = (53 * hash) + getStrength();
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.framework.InterconnectLink parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.InterconnectLink parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.InterconnectLink parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.InterconnectLink parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.InterconnectLink parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.InterconnectLink parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.InterconnectLink parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.InterconnectLink parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.InterconnectLink parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.InterconnectLink parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.InterconnectLink parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.InterconnectLink parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.framework.InterconnectLink prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * Protobuf type {@code tensorflow.InterconnectLink}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.InterconnectLink)
-      org.tensorflow.proto.framework.InterconnectLinkOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.DeviceAttributesProtos.internal_static_tensorflow_InterconnectLink_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.DeviceAttributesProtos.internal_static_tensorflow_InterconnectLink_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.InterconnectLink.class, org.tensorflow.proto.framework.InterconnectLink.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.framework.InterconnectLink.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      deviceId_ = 0;
-
-      type_ = "";
-
-      strength_ = 0;
-
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.framework.DeviceAttributesProtos.internal_static_tensorflow_InterconnectLink_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.InterconnectLink getDefaultInstanceForType() {
-      return org.tensorflow.proto.framework.InterconnectLink.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.InterconnectLink build() {
-      org.tensorflow.proto.framework.InterconnectLink result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.InterconnectLink buildPartial() {
-      org.tensorflow.proto.framework.InterconnectLink result = new org.tensorflow.proto.framework.InterconnectLink(this);
-      result.deviceId_ = deviceId_;
-      result.type_ = type_;
-      result.strength_ = strength_;
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.framework.InterconnectLink) {
-        return mergeFrom((org.tensorflow.proto.framework.InterconnectLink)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.framework.InterconnectLink other) {
-      if (other == org.tensorflow.proto.framework.InterconnectLink.getDefaultInstance()) return this;
-      if (other.getDeviceId() != 0) {
-        setDeviceId(other.getDeviceId());
-      }
-      if (!other.getType().isEmpty()) {
-        type_ = other.type_;
-        onChanged();
-      }
-      if (other.getStrength() != 0) {
-        setStrength(other.getStrength());
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.framework.InterconnectLink parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.framework.InterconnectLink) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-
-    private int deviceId_ ;
-    /**
-     * <code>int32 device_id = 1;</code>
-     */
-    public int getDeviceId() {
-      return deviceId_;
-    }
-    /**
-     * <code>int32 device_id = 1;</code>
-     */
-    public Builder setDeviceId(int value) {
-      
-      deviceId_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>int32 device_id = 1;</code>
-     */
-    public Builder clearDeviceId() {
-      
-      deviceId_ = 0;
-      onChanged();
-      return this;
-    }
-
-    private java.lang.Object type_ = "";
-    /**
-     * <code>string type = 2;</code>
-     */
-    public java.lang.String getType() {
-      java.lang.Object ref = type_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        type_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <code>string type = 2;</code>
-     */
-    public com.google.protobuf.ByteString
-        getTypeBytes() {
-      java.lang.Object ref = type_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        type_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <code>string type = 2;</code>
-     */
-    public Builder setType(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      type_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>string type = 2;</code>
-     */
-    public Builder clearType() {
-      
-      type_ = getDefaultInstance().getType();
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>string type = 2;</code>
-     */
-    public Builder setTypeBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      type_ = value;
-      onChanged();
-      return this;
-    }
-
-    private int strength_ ;
-    /**
-     * <code>int32 strength = 3;</code>
-     */
-    public int getStrength() {
-      return strength_;
-    }
-    /**
-     * <code>int32 strength = 3;</code>
-     */
-    public Builder setStrength(int value) {
-      
-      strength_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>int32 strength = 3;</code>
-     */
-    public Builder clearStrength() {
-      
-      strength_ = 0;
-      onChanged();
-      return this;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.InterconnectLink)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.InterconnectLink)
-  private static final org.tensorflow.proto.framework.InterconnectLink DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.InterconnectLink();
-  }
-
-  public static org.tensorflow.proto.framework.InterconnectLink getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<InterconnectLink>
-      PARSER = new com.google.protobuf.AbstractParser<InterconnectLink>() {
-    @java.lang.Override
-    public InterconnectLink parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new InterconnectLink(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<InterconnectLink> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<InterconnectLink> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.framework.InterconnectLink getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/InterconnectLinkOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/InterconnectLinkOrBuilder.java
deleted file mode 100644
index 61316cfa8cc..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/InterconnectLinkOrBuilder.java
+++ /dev/null
@@ -1,29 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/framework/device_attributes.proto
-
-package org.tensorflow.proto.framework;
-
-public interface InterconnectLinkOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.InterconnectLink)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <code>int32 device_id = 1;</code>
-   */
-  int getDeviceId();
-
-  /**
-   * <code>string type = 2;</code>
-   */
-  java.lang.String getType();
-  /**
-   * <code>string type = 2;</code>
-   */
-  com.google.protobuf.ByteString
-      getTypeBytes();
-
-  /**
-   * <code>int32 strength = 3;</code>
-   */
-  int getStrength();
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/KernelDef.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/KernelDef.java
deleted file mode 100644
index 76637537c9e..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/KernelDef.java
+++ /dev/null
@@ -1,2420 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/framework/kernel_def.proto
-
-package org.tensorflow.proto.framework;
-
-/**
- * Protobuf type {@code tensorflow.KernelDef}
- */
-public  final class KernelDef extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.KernelDef)
-    KernelDefOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use KernelDef.newBuilder() to construct.
-  private KernelDef(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private KernelDef() {
-    op_ = "";
-    deviceType_ = "";
-    constraint_ = java.util.Collections.emptyList();
-    hostMemoryArg_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-    label_ = "";
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new KernelDef();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private KernelDef(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    int mutable_bitField0_ = 0;
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 10: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            op_ = s;
-            break;
-          }
-          case 18: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            deviceType_ = s;
-            break;
-          }
-          case 26: {
-            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
-              constraint_ = new java.util.ArrayList<org.tensorflow.proto.framework.KernelDef.AttrConstraint>();
-              mutable_bitField0_ |= 0x00000001;
-            }
-            constraint_.add(
-                input.readMessage(org.tensorflow.proto.framework.KernelDef.AttrConstraint.parser(), extensionRegistry));
-            break;
-          }
-          case 34: {
-            java.lang.String s = input.readStringRequireUtf8();
-            if (!((mutable_bitField0_ & 0x00000002) != 0)) {
-              hostMemoryArg_ = new com.google.protobuf.LazyStringArrayList();
-              mutable_bitField0_ |= 0x00000002;
-            }
-            hostMemoryArg_.add(s);
-            break;
-          }
-          case 42: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            label_ = s;
-            break;
-          }
-          case 48: {
-
-            priority_ = input.readInt32();
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      if (((mutable_bitField0_ & 0x00000001) != 0)) {
-        constraint_ = java.util.Collections.unmodifiableList(constraint_);
-      }
-      if (((mutable_bitField0_ & 0x00000002) != 0)) {
-        hostMemoryArg_ = hostMemoryArg_.getUnmodifiableView();
-      }
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.framework.KernelDefProtos.internal_static_tensorflow_KernelDef_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.framework.KernelDefProtos.internal_static_tensorflow_KernelDef_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.framework.KernelDef.class, org.tensorflow.proto.framework.KernelDef.Builder.class);
-  }
-
-  public interface AttrConstraintOrBuilder extends
-      // @@protoc_insertion_point(interface_extends:tensorflow.KernelDef.AttrConstraint)
-      com.google.protobuf.MessageOrBuilder {
-
-    /**
-     * <pre>
-     * Name of an attr from the Op.
-     * </pre>
-     *
-     * <code>string name = 1;</code>
-     */
-    java.lang.String getName();
-    /**
-     * <pre>
-     * Name of an attr from the Op.
-     * </pre>
-     *
-     * <code>string name = 1;</code>
-     */
-    com.google.protobuf.ByteString
-        getNameBytes();
-
-    /**
-     * <pre>
-     * A list of values that this kernel supports for this attr.
-     * Like OpDef.AttrDef.allowed_values, except for kernels instead of Ops.
-     * </pre>
-     *
-     * <code>.tensorflow.AttrValue allowed_values = 2;</code>
-     */
-    boolean hasAllowedValues();
-    /**
-     * <pre>
-     * A list of values that this kernel supports for this attr.
-     * Like OpDef.AttrDef.allowed_values, except for kernels instead of Ops.
-     * </pre>
-     *
-     * <code>.tensorflow.AttrValue allowed_values = 2;</code>
-     */
-    org.tensorflow.proto.framework.AttrValue getAllowedValues();
-    /**
-     * <pre>
-     * A list of values that this kernel supports for this attr.
-     * Like OpDef.AttrDef.allowed_values, except for kernels instead of Ops.
-     * </pre>
-     *
-     * <code>.tensorflow.AttrValue allowed_values = 2;</code>
-     */
-    org.tensorflow.proto.framework.AttrValueOrBuilder getAllowedValuesOrBuilder();
-  }
-  /**
-   * Protobuf type {@code tensorflow.KernelDef.AttrConstraint}
-   */
-  public  static final class AttrConstraint extends
-      com.google.protobuf.GeneratedMessageV3 implements
-      // @@protoc_insertion_point(message_implements:tensorflow.KernelDef.AttrConstraint)
-      AttrConstraintOrBuilder {
-  private static final long serialVersionUID = 0L;
-    // Use AttrConstraint.newBuilder() to construct.
-    private AttrConstraint(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-      super(builder);
-    }
-    private AttrConstraint() {
-      name_ = "";
-    }
-
-    @java.lang.Override
-    @SuppressWarnings({"unused"})
-    protected java.lang.Object newInstance(
-        UnusedPrivateParameter unused) {
-      return new AttrConstraint();
-    }
-
-    @java.lang.Override
-    public final com.google.protobuf.UnknownFieldSet
-    getUnknownFields() {
-      return this.unknownFields;
-    }
-    private AttrConstraint(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      this();
-      if (extensionRegistry == null) {
-        throw new java.lang.NullPointerException();
-      }
-      com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-          com.google.protobuf.UnknownFieldSet.newBuilder();
-      try {
-        boolean done = false;
-        while (!done) {
-          int tag = input.readTag();
-          switch (tag) {
-            case 0:
-              done = true;
-              break;
-            case 10: {
-              java.lang.String s = input.readStringRequireUtf8();
-
-              name_ = s;
-              break;
-            }
-            case 18: {
-              org.tensorflow.proto.framework.AttrValue.Builder subBuilder = null;
-              if (allowedValues_ != null) {
-                subBuilder = allowedValues_.toBuilder();
-              }
-              allowedValues_ = input.readMessage(org.tensorflow.proto.framework.AttrValue.parser(), extensionRegistry);
-              if (subBuilder != null) {
-                subBuilder.mergeFrom(allowedValues_);
-                allowedValues_ = subBuilder.buildPartial();
-              }
-
-              break;
-            }
-            default: {
-              if (!parseUnknownField(
-                  input, unknownFields, extensionRegistry, tag)) {
-                done = true;
-              }
-              break;
-            }
-          }
-        }
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        throw e.setUnfinishedMessage(this);
-      } catch (java.io.IOException e) {
-        throw new com.google.protobuf.InvalidProtocolBufferException(
-            e).setUnfinishedMessage(this);
-      } finally {
-        this.unknownFields = unknownFields.build();
-        makeExtensionsImmutable();
-      }
-    }
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.KernelDefProtos.internal_static_tensorflow_KernelDef_AttrConstraint_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.KernelDefProtos.internal_static_tensorflow_KernelDef_AttrConstraint_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.KernelDef.AttrConstraint.class, org.tensorflow.proto.framework.KernelDef.AttrConstraint.Builder.class);
-    }
-
-    public static final int NAME_FIELD_NUMBER = 1;
-    private volatile java.lang.Object name_;
-    /**
-     * <pre>
-     * Name of an attr from the Op.
-     * </pre>
-     *
-     * <code>string name = 1;</code>
-     */
-    public java.lang.String getName() {
-      java.lang.Object ref = name_;
-      if (ref instanceof java.lang.String) {
-        return (java.lang.String) ref;
-      } else {
-        com.google.protobuf.ByteString bs = 
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        name_ = s;
-        return s;
-      }
-    }
-    /**
-     * <pre>
-     * Name of an attr from the Op.
-     * </pre>
-     *
-     * <code>string name = 1;</code>
-     */
-    public com.google.protobuf.ByteString
-        getNameBytes() {
-      java.lang.Object ref = name_;
-      if (ref instanceof java.lang.String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        name_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-
-    public static final int ALLOWED_VALUES_FIELD_NUMBER = 2;
-    private org.tensorflow.proto.framework.AttrValue allowedValues_;
-    /**
-     * <pre>
-     * A list of values that this kernel supports for this attr.
-     * Like OpDef.AttrDef.allowed_values, except for kernels instead of Ops.
-     * </pre>
-     *
-     * <code>.tensorflow.AttrValue allowed_values = 2;</code>
-     */
-    public boolean hasAllowedValues() {
-      return allowedValues_ != null;
-    }
-    /**
-     * <pre>
-     * A list of values that this kernel supports for this attr.
-     * Like OpDef.AttrDef.allowed_values, except for kernels instead of Ops.
-     * </pre>
-     *
-     * <code>.tensorflow.AttrValue allowed_values = 2;</code>
-     */
-    public org.tensorflow.proto.framework.AttrValue getAllowedValues() {
-      return allowedValues_ == null ? org.tensorflow.proto.framework.AttrValue.getDefaultInstance() : allowedValues_;
-    }
-    /**
-     * <pre>
-     * A list of values that this kernel supports for this attr.
-     * Like OpDef.AttrDef.allowed_values, except for kernels instead of Ops.
-     * </pre>
-     *
-     * <code>.tensorflow.AttrValue allowed_values = 2;</code>
-     */
-    public org.tensorflow.proto.framework.AttrValueOrBuilder getAllowedValuesOrBuilder() {
-      return getAllowedValues();
-    }
-
-    private byte memoizedIsInitialized = -1;
-    @java.lang.Override
-    public final boolean isInitialized() {
-      byte isInitialized = memoizedIsInitialized;
-      if (isInitialized == 1) return true;
-      if (isInitialized == 0) return false;
-
-      memoizedIsInitialized = 1;
-      return true;
-    }
-
-    @java.lang.Override
-    public void writeTo(com.google.protobuf.CodedOutputStream output)
-                        throws java.io.IOException {
-      if (!getNameBytes().isEmpty()) {
-        com.google.protobuf.GeneratedMessageV3.writeString(output, 1, name_);
-      }
-      if (allowedValues_ != null) {
-        output.writeMessage(2, getAllowedValues());
-      }
-      unknownFields.writeTo(output);
-    }
-
-    @java.lang.Override
-    public int getSerializedSize() {
-      int size = memoizedSize;
-      if (size != -1) return size;
-
-      size = 0;
-      if (!getNameBytes().isEmpty()) {
-        size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, name_);
-      }
-      if (allowedValues_ != null) {
-        size += com.google.protobuf.CodedOutputStream
-          .computeMessageSize(2, getAllowedValues());
-      }
-      size += unknownFields.getSerializedSize();
-      memoizedSize = size;
-      return size;
-    }
-
-    @java.lang.Override
-    public boolean equals(final java.lang.Object obj) {
-      if (obj == this) {
-       return true;
-      }
-      if (!(obj instanceof org.tensorflow.proto.framework.KernelDef.AttrConstraint)) {
-        return super.equals(obj);
-      }
-      org.tensorflow.proto.framework.KernelDef.AttrConstraint other = (org.tensorflow.proto.framework.KernelDef.AttrConstraint) obj;
-
-      if (!getName()
-          .equals(other.getName())) return false;
-      if (hasAllowedValues() != other.hasAllowedValues()) return false;
-      if (hasAllowedValues()) {
-        if (!getAllowedValues()
-            .equals(other.getAllowedValues())) return false;
-      }
-      if (!unknownFields.equals(other.unknownFields)) return false;
-      return true;
-    }
-
-    @java.lang.Override
-    public int hashCode() {
-      if (memoizedHashCode != 0) {
-        return memoizedHashCode;
-      }
-      int hash = 41;
-      hash = (19 * hash) + getDescriptor().hashCode();
-      hash = (37 * hash) + NAME_FIELD_NUMBER;
-      hash = (53 * hash) + getName().hashCode();
-      if (hasAllowedValues()) {
-        hash = (37 * hash) + ALLOWED_VALUES_FIELD_NUMBER;
-        hash = (53 * hash) + getAllowedValues().hashCode();
-      }
-      hash = (29 * hash) + unknownFields.hashCode();
-      memoizedHashCode = hash;
-      return hash;
-    }
-
-    public static org.tensorflow.proto.framework.KernelDef.AttrConstraint parseFrom(
-        java.nio.ByteBuffer data)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data);
-    }
-    public static org.tensorflow.proto.framework.KernelDef.AttrConstraint parseFrom(
-        java.nio.ByteBuffer data,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.KernelDef.AttrConstraint parseFrom(
-        com.google.protobuf.ByteString data)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data);
-    }
-    public static org.tensorflow.proto.framework.KernelDef.AttrConstraint parseFrom(
-        com.google.protobuf.ByteString data,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.KernelDef.AttrConstraint parseFrom(byte[] data)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data);
-    }
-    public static org.tensorflow.proto.framework.KernelDef.AttrConstraint parseFrom(
-        byte[] data,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.KernelDef.AttrConstraint parseFrom(java.io.InputStream input)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input);
-    }
-    public static org.tensorflow.proto.framework.KernelDef.AttrConstraint parseFrom(
-        java.io.InputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.KernelDef.AttrConstraint parseDelimitedFrom(java.io.InputStream input)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseDelimitedWithIOException(PARSER, input);
-    }
-    public static org.tensorflow.proto.framework.KernelDef.AttrConstraint parseDelimitedFrom(
-        java.io.InputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.KernelDef.AttrConstraint parseFrom(
-        com.google.protobuf.CodedInputStream input)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input);
-    }
-    public static org.tensorflow.proto.framework.KernelDef.AttrConstraint parseFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input, extensionRegistry);
-    }
-
-    @java.lang.Override
-    public Builder newBuilderForType() { return newBuilder(); }
-    public static Builder newBuilder() {
-      return DEFAULT_INSTANCE.toBuilder();
-    }
-    public static Builder newBuilder(org.tensorflow.proto.framework.KernelDef.AttrConstraint prototype) {
-      return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-    }
-    @java.lang.Override
-    public Builder toBuilder() {
-      return this == DEFAULT_INSTANCE
-          ? new Builder() : new Builder().mergeFrom(this);
-    }
-
-    @java.lang.Override
-    protected Builder newBuilderForType(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      Builder builder = new Builder(parent);
-      return builder;
-    }
-    /**
-     * Protobuf type {@code tensorflow.KernelDef.AttrConstraint}
-     */
-    public static final class Builder extends
-        com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-        // @@protoc_insertion_point(builder_implements:tensorflow.KernelDef.AttrConstraint)
-        org.tensorflow.proto.framework.KernelDef.AttrConstraintOrBuilder {
-      public static final com.google.protobuf.Descriptors.Descriptor
-          getDescriptor() {
-        return org.tensorflow.proto.framework.KernelDefProtos.internal_static_tensorflow_KernelDef_AttrConstraint_descriptor;
-      }
-
-      @java.lang.Override
-      protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-          internalGetFieldAccessorTable() {
-        return org.tensorflow.proto.framework.KernelDefProtos.internal_static_tensorflow_KernelDef_AttrConstraint_fieldAccessorTable
-            .ensureFieldAccessorsInitialized(
-                org.tensorflow.proto.framework.KernelDef.AttrConstraint.class, org.tensorflow.proto.framework.KernelDef.AttrConstraint.Builder.class);
-      }
-
-      // Construct using org.tensorflow.proto.framework.KernelDef.AttrConstraint.newBuilder()
-      private Builder() {
-        maybeForceBuilderInitialization();
-      }
-
-      private Builder(
-          com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-        super(parent);
-        maybeForceBuilderInitialization();
-      }
-      private void maybeForceBuilderInitialization() {
-        if (com.google.protobuf.GeneratedMessageV3
-                .alwaysUseFieldBuilders) {
-        }
-      }
-      @java.lang.Override
-      public Builder clear() {
-        super.clear();
-        name_ = "";
-
-        if (allowedValuesBuilder_ == null) {
-          allowedValues_ = null;
-        } else {
-          allowedValues_ = null;
-          allowedValuesBuilder_ = null;
-        }
-        return this;
-      }
-
-      @java.lang.Override
-      public com.google.protobuf.Descriptors.Descriptor
-          getDescriptorForType() {
-        return org.tensorflow.proto.framework.KernelDefProtos.internal_static_tensorflow_KernelDef_AttrConstraint_descriptor;
-      }
-
-      @java.lang.Override
-      public org.tensorflow.proto.framework.KernelDef.AttrConstraint getDefaultInstanceForType() {
-        return org.tensorflow.proto.framework.KernelDef.AttrConstraint.getDefaultInstance();
-      }
-
-      @java.lang.Override
-      public org.tensorflow.proto.framework.KernelDef.AttrConstraint build() {
-        org.tensorflow.proto.framework.KernelDef.AttrConstraint result = buildPartial();
-        if (!result.isInitialized()) {
-          throw newUninitializedMessageException(result);
-        }
-        return result;
-      }
-
-      @java.lang.Override
-      public org.tensorflow.proto.framework.KernelDef.AttrConstraint buildPartial() {
-        org.tensorflow.proto.framework.KernelDef.AttrConstraint result = new org.tensorflow.proto.framework.KernelDef.AttrConstraint(this);
-        result.name_ = name_;
-        if (allowedValuesBuilder_ == null) {
-          result.allowedValues_ = allowedValues_;
-        } else {
-          result.allowedValues_ = allowedValuesBuilder_.build();
-        }
-        onBuilt();
-        return result;
-      }
-
-      @java.lang.Override
-      public Builder clone() {
-        return super.clone();
-      }
-      @java.lang.Override
-      public Builder setField(
-          com.google.protobuf.Descriptors.FieldDescriptor field,
-          java.lang.Object value) {
-        return super.setField(field, value);
-      }
-      @java.lang.Override
-      public Builder clearField(
-          com.google.protobuf.Descriptors.FieldDescriptor field) {
-        return super.clearField(field);
-      }
-      @java.lang.Override
-      public Builder clearOneof(
-          com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-        return super.clearOneof(oneof);
-      }
-      @java.lang.Override
-      public Builder setRepeatedField(
-          com.google.protobuf.Descriptors.FieldDescriptor field,
-          int index, java.lang.Object value) {
-        return super.setRepeatedField(field, index, value);
-      }
-      @java.lang.Override
-      public Builder addRepeatedField(
-          com.google.protobuf.Descriptors.FieldDescriptor field,
-          java.lang.Object value) {
-        return super.addRepeatedField(field, value);
-      }
-      @java.lang.Override
-      public Builder mergeFrom(com.google.protobuf.Message other) {
-        if (other instanceof org.tensorflow.proto.framework.KernelDef.AttrConstraint) {
-          return mergeFrom((org.tensorflow.proto.framework.KernelDef.AttrConstraint)other);
-        } else {
-          super.mergeFrom(other);
-          return this;
-        }
-      }
-
-      public Builder mergeFrom(org.tensorflow.proto.framework.KernelDef.AttrConstraint other) {
-        if (other == org.tensorflow.proto.framework.KernelDef.AttrConstraint.getDefaultInstance()) return this;
-        if (!other.getName().isEmpty()) {
-          name_ = other.name_;
-          onChanged();
-        }
-        if (other.hasAllowedValues()) {
-          mergeAllowedValues(other.getAllowedValues());
-        }
-        this.mergeUnknownFields(other.unknownFields);
-        onChanged();
-        return this;
-      }
-
-      @java.lang.Override
-      public final boolean isInitialized() {
-        return true;
-      }
-
-      @java.lang.Override
-      public Builder mergeFrom(
-          com.google.protobuf.CodedInputStream input,
-          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-          throws java.io.IOException {
-        org.tensorflow.proto.framework.KernelDef.AttrConstraint parsedMessage = null;
-        try {
-          parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-        } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-          parsedMessage = (org.tensorflow.proto.framework.KernelDef.AttrConstraint) e.getUnfinishedMessage();
-          throw e.unwrapIOException();
-        } finally {
-          if (parsedMessage != null) {
-            mergeFrom(parsedMessage);
-          }
-        }
-        return this;
-      }
-
-      private java.lang.Object name_ = "";
-      /**
-       * <pre>
-       * Name of an attr from the Op.
-       * </pre>
-       *
-       * <code>string name = 1;</code>
-       */
-      public java.lang.String getName() {
-        java.lang.Object ref = name_;
-        if (!(ref instanceof java.lang.String)) {
-          com.google.protobuf.ByteString bs =
-              (com.google.protobuf.ByteString) ref;
-          java.lang.String s = bs.toStringUtf8();
-          name_ = s;
-          return s;
-        } else {
-          return (java.lang.String) ref;
-        }
-      }
-      /**
-       * <pre>
-       * Name of an attr from the Op.
-       * </pre>
-       *
-       * <code>string name = 1;</code>
-       */
-      public com.google.protobuf.ByteString
-          getNameBytes() {
-        java.lang.Object ref = name_;
-        if (ref instanceof String) {
-          com.google.protobuf.ByteString b = 
-              com.google.protobuf.ByteString.copyFromUtf8(
-                  (java.lang.String) ref);
-          name_ = b;
-          return b;
-        } else {
-          return (com.google.protobuf.ByteString) ref;
-        }
-      }
-      /**
-       * <pre>
-       * Name of an attr from the Op.
-       * </pre>
-       *
-       * <code>string name = 1;</code>
-       */
-      public Builder setName(
-          java.lang.String value) {
-        if (value == null) {
-    throw new NullPointerException();
-  }
-  
-        name_ = value;
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * Name of an attr from the Op.
-       * </pre>
-       *
-       * <code>string name = 1;</code>
-       */
-      public Builder clearName() {
-        
-        name_ = getDefaultInstance().getName();
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * Name of an attr from the Op.
-       * </pre>
-       *
-       * <code>string name = 1;</code>
-       */
-      public Builder setNameBytes(
-          com.google.protobuf.ByteString value) {
-        if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-        
-        name_ = value;
-        onChanged();
-        return this;
-      }
-
-      private org.tensorflow.proto.framework.AttrValue allowedValues_;
-      private com.google.protobuf.SingleFieldBuilderV3<
-          org.tensorflow.proto.framework.AttrValue, org.tensorflow.proto.framework.AttrValue.Builder, org.tensorflow.proto.framework.AttrValueOrBuilder> allowedValuesBuilder_;
-      /**
-       * <pre>
-       * A list of values that this kernel supports for this attr.
-       * Like OpDef.AttrDef.allowed_values, except for kernels instead of Ops.
-       * </pre>
-       *
-       * <code>.tensorflow.AttrValue allowed_values = 2;</code>
-       */
-      public boolean hasAllowedValues() {
-        return allowedValuesBuilder_ != null || allowedValues_ != null;
-      }
-      /**
-       * <pre>
-       * A list of values that this kernel supports for this attr.
-       * Like OpDef.AttrDef.allowed_values, except for kernels instead of Ops.
-       * </pre>
-       *
-       * <code>.tensorflow.AttrValue allowed_values = 2;</code>
-       */
-      public org.tensorflow.proto.framework.AttrValue getAllowedValues() {
-        if (allowedValuesBuilder_ == null) {
-          return allowedValues_ == null ? org.tensorflow.proto.framework.AttrValue.getDefaultInstance() : allowedValues_;
-        } else {
-          return allowedValuesBuilder_.getMessage();
-        }
-      }
-      /**
-       * <pre>
-       * A list of values that this kernel supports for this attr.
-       * Like OpDef.AttrDef.allowed_values, except for kernels instead of Ops.
-       * </pre>
-       *
-       * <code>.tensorflow.AttrValue allowed_values = 2;</code>
-       */
-      public Builder setAllowedValues(org.tensorflow.proto.framework.AttrValue value) {
-        if (allowedValuesBuilder_ == null) {
-          if (value == null) {
-            throw new NullPointerException();
-          }
-          allowedValues_ = value;
-          onChanged();
-        } else {
-          allowedValuesBuilder_.setMessage(value);
-        }
-
-        return this;
-      }
-      /**
-       * <pre>
-       * A list of values that this kernel supports for this attr.
-       * Like OpDef.AttrDef.allowed_values, except for kernels instead of Ops.
-       * </pre>
-       *
-       * <code>.tensorflow.AttrValue allowed_values = 2;</code>
-       */
-      public Builder setAllowedValues(
-          org.tensorflow.proto.framework.AttrValue.Builder builderForValue) {
-        if (allowedValuesBuilder_ == null) {
-          allowedValues_ = builderForValue.build();
-          onChanged();
-        } else {
-          allowedValuesBuilder_.setMessage(builderForValue.build());
-        }
-
-        return this;
-      }
-      /**
-       * <pre>
-       * A list of values that this kernel supports for this attr.
-       * Like OpDef.AttrDef.allowed_values, except for kernels instead of Ops.
-       * </pre>
-       *
-       * <code>.tensorflow.AttrValue allowed_values = 2;</code>
-       */
-      public Builder mergeAllowedValues(org.tensorflow.proto.framework.AttrValue value) {
-        if (allowedValuesBuilder_ == null) {
-          if (allowedValues_ != null) {
-            allowedValues_ =
-              org.tensorflow.proto.framework.AttrValue.newBuilder(allowedValues_).mergeFrom(value).buildPartial();
-          } else {
-            allowedValues_ = value;
-          }
-          onChanged();
-        } else {
-          allowedValuesBuilder_.mergeFrom(value);
-        }
-
-        return this;
-      }
-      /**
-       * <pre>
-       * A list of values that this kernel supports for this attr.
-       * Like OpDef.AttrDef.allowed_values, except for kernels instead of Ops.
-       * </pre>
-       *
-       * <code>.tensorflow.AttrValue allowed_values = 2;</code>
-       */
-      public Builder clearAllowedValues() {
-        if (allowedValuesBuilder_ == null) {
-          allowedValues_ = null;
-          onChanged();
-        } else {
-          allowedValues_ = null;
-          allowedValuesBuilder_ = null;
-        }
-
-        return this;
-      }
-      /**
-       * <pre>
-       * A list of values that this kernel supports for this attr.
-       * Like OpDef.AttrDef.allowed_values, except for kernels instead of Ops.
-       * </pre>
-       *
-       * <code>.tensorflow.AttrValue allowed_values = 2;</code>
-       */
-      public org.tensorflow.proto.framework.AttrValue.Builder getAllowedValuesBuilder() {
-        
-        onChanged();
-        return getAllowedValuesFieldBuilder().getBuilder();
-      }
-      /**
-       * <pre>
-       * A list of values that this kernel supports for this attr.
-       * Like OpDef.AttrDef.allowed_values, except for kernels instead of Ops.
-       * </pre>
-       *
-       * <code>.tensorflow.AttrValue allowed_values = 2;</code>
-       */
-      public org.tensorflow.proto.framework.AttrValueOrBuilder getAllowedValuesOrBuilder() {
-        if (allowedValuesBuilder_ != null) {
-          return allowedValuesBuilder_.getMessageOrBuilder();
-        } else {
-          return allowedValues_ == null ?
-              org.tensorflow.proto.framework.AttrValue.getDefaultInstance() : allowedValues_;
-        }
-      }
-      /**
-       * <pre>
-       * A list of values that this kernel supports for this attr.
-       * Like OpDef.AttrDef.allowed_values, except for kernels instead of Ops.
-       * </pre>
-       *
-       * <code>.tensorflow.AttrValue allowed_values = 2;</code>
-       */
-      private com.google.protobuf.SingleFieldBuilderV3<
-          org.tensorflow.proto.framework.AttrValue, org.tensorflow.proto.framework.AttrValue.Builder, org.tensorflow.proto.framework.AttrValueOrBuilder> 
-          getAllowedValuesFieldBuilder() {
-        if (allowedValuesBuilder_ == null) {
-          allowedValuesBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-              org.tensorflow.proto.framework.AttrValue, org.tensorflow.proto.framework.AttrValue.Builder, org.tensorflow.proto.framework.AttrValueOrBuilder>(
-                  getAllowedValues(),
-                  getParentForChildren(),
-                  isClean());
-          allowedValues_ = null;
-        }
-        return allowedValuesBuilder_;
-      }
-      @java.lang.Override
-      public final Builder setUnknownFields(
-          final com.google.protobuf.UnknownFieldSet unknownFields) {
-        return super.setUnknownFields(unknownFields);
-      }
-
-      @java.lang.Override
-      public final Builder mergeUnknownFields(
-          final com.google.protobuf.UnknownFieldSet unknownFields) {
-        return super.mergeUnknownFields(unknownFields);
-      }
-
-
-      // @@protoc_insertion_point(builder_scope:tensorflow.KernelDef.AttrConstraint)
-    }
-
-    // @@protoc_insertion_point(class_scope:tensorflow.KernelDef.AttrConstraint)
-    private static final org.tensorflow.proto.framework.KernelDef.AttrConstraint DEFAULT_INSTANCE;
-    static {
-      DEFAULT_INSTANCE = new org.tensorflow.proto.framework.KernelDef.AttrConstraint();
-    }
-
-    public static org.tensorflow.proto.framework.KernelDef.AttrConstraint getDefaultInstance() {
-      return DEFAULT_INSTANCE;
-    }
-
-    private static final com.google.protobuf.Parser<AttrConstraint>
-        PARSER = new com.google.protobuf.AbstractParser<AttrConstraint>() {
-      @java.lang.Override
-      public AttrConstraint parsePartialFrom(
-          com.google.protobuf.CodedInputStream input,
-          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-          throws com.google.protobuf.InvalidProtocolBufferException {
-        return new AttrConstraint(input, extensionRegistry);
-      }
-    };
-
-    public static com.google.protobuf.Parser<AttrConstraint> parser() {
-      return PARSER;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Parser<AttrConstraint> getParserForType() {
-      return PARSER;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.KernelDef.AttrConstraint getDefaultInstanceForType() {
-      return DEFAULT_INSTANCE;
-    }
-
-  }
-
-  public static final int OP_FIELD_NUMBER = 1;
-  private volatile java.lang.Object op_;
-  /**
-   * <pre>
-   * Must match the name of an Op.
-   * </pre>
-   *
-   * <code>string op = 1;</code>
-   */
-  public java.lang.String getOp() {
-    java.lang.Object ref = op_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      op_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * Must match the name of an Op.
-   * </pre>
-   *
-   * <code>string op = 1;</code>
-   */
-  public com.google.protobuf.ByteString
-      getOpBytes() {
-    java.lang.Object ref = op_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      op_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int DEVICE_TYPE_FIELD_NUMBER = 2;
-  private volatile java.lang.Object deviceType_;
-  /**
-   * <pre>
-   * Type of device this kernel runs on.
-   * </pre>
-   *
-   * <code>string device_type = 2;</code>
-   */
-  public java.lang.String getDeviceType() {
-    java.lang.Object ref = deviceType_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      deviceType_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * Type of device this kernel runs on.
-   * </pre>
-   *
-   * <code>string device_type = 2;</code>
-   */
-  public com.google.protobuf.ByteString
-      getDeviceTypeBytes() {
-    java.lang.Object ref = deviceType_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      deviceType_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int CONSTRAINT_FIELD_NUMBER = 3;
-  private java.util.List<org.tensorflow.proto.framework.KernelDef.AttrConstraint> constraint_;
-  /**
-   * <code>repeated .tensorflow.KernelDef.AttrConstraint constraint = 3;</code>
-   */
-  public java.util.List<org.tensorflow.proto.framework.KernelDef.AttrConstraint> getConstraintList() {
-    return constraint_;
-  }
-  /**
-   * <code>repeated .tensorflow.KernelDef.AttrConstraint constraint = 3;</code>
-   */
-  public java.util.List<? extends org.tensorflow.proto.framework.KernelDef.AttrConstraintOrBuilder> 
-      getConstraintOrBuilderList() {
-    return constraint_;
-  }
-  /**
-   * <code>repeated .tensorflow.KernelDef.AttrConstraint constraint = 3;</code>
-   */
-  public int getConstraintCount() {
-    return constraint_.size();
-  }
-  /**
-   * <code>repeated .tensorflow.KernelDef.AttrConstraint constraint = 3;</code>
-   */
-  public org.tensorflow.proto.framework.KernelDef.AttrConstraint getConstraint(int index) {
-    return constraint_.get(index);
-  }
-  /**
-   * <code>repeated .tensorflow.KernelDef.AttrConstraint constraint = 3;</code>
-   */
-  public org.tensorflow.proto.framework.KernelDef.AttrConstraintOrBuilder getConstraintOrBuilder(
-      int index) {
-    return constraint_.get(index);
-  }
-
-  public static final int HOST_MEMORY_ARG_FIELD_NUMBER = 4;
-  private com.google.protobuf.LazyStringList hostMemoryArg_;
-  /**
-   * <pre>
-   * Names of the Op's input_/output_args that reside in host memory
-   * instead of device memory.
-   * </pre>
-   *
-   * <code>repeated string host_memory_arg = 4;</code>
-   */
-  public com.google.protobuf.ProtocolStringList
-      getHostMemoryArgList() {
-    return hostMemoryArg_;
-  }
-  /**
-   * <pre>
-   * Names of the Op's input_/output_args that reside in host memory
-   * instead of device memory.
-   * </pre>
-   *
-   * <code>repeated string host_memory_arg = 4;</code>
-   */
-  public int getHostMemoryArgCount() {
-    return hostMemoryArg_.size();
-  }
-  /**
-   * <pre>
-   * Names of the Op's input_/output_args that reside in host memory
-   * instead of device memory.
-   * </pre>
-   *
-   * <code>repeated string host_memory_arg = 4;</code>
-   */
-  public java.lang.String getHostMemoryArg(int index) {
-    return hostMemoryArg_.get(index);
-  }
-  /**
-   * <pre>
-   * Names of the Op's input_/output_args that reside in host memory
-   * instead of device memory.
-   * </pre>
-   *
-   * <code>repeated string host_memory_arg = 4;</code>
-   */
-  public com.google.protobuf.ByteString
-      getHostMemoryArgBytes(int index) {
-    return hostMemoryArg_.getByteString(index);
-  }
-
-  public static final int LABEL_FIELD_NUMBER = 5;
-  private volatile java.lang.Object label_;
-  /**
-   * <pre>
-   * This allows experimental kernels to be registered for an op that
-   * won't be used unless the user specifies a "_kernel" attr with
-   * value matching this.
-   * </pre>
-   *
-   * <code>string label = 5;</code>
-   */
-  public java.lang.String getLabel() {
-    java.lang.Object ref = label_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      label_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * This allows experimental kernels to be registered for an op that
-   * won't be used unless the user specifies a "_kernel" attr with
-   * value matching this.
-   * </pre>
-   *
-   * <code>string label = 5;</code>
-   */
-  public com.google.protobuf.ByteString
-      getLabelBytes() {
-    java.lang.Object ref = label_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      label_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int PRIORITY_FIELD_NUMBER = 6;
-  private int priority_;
-  /**
-   * <pre>
-   * Prioritization of kernel amongst different devices. By default we assume
-   * priority is 0. The higher the priority the better. By default (i.e. if
-   * this is not set), we prefer GPU kernels over CPU.
-   * </pre>
-   *
-   * <code>int32 priority = 6;</code>
-   */
-  public int getPriority() {
-    return priority_;
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    if (!getOpBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 1, op_);
-    }
-    if (!getDeviceTypeBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 2, deviceType_);
-    }
-    for (int i = 0; i < constraint_.size(); i++) {
-      output.writeMessage(3, constraint_.get(i));
-    }
-    for (int i = 0; i < hostMemoryArg_.size(); i++) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 4, hostMemoryArg_.getRaw(i));
-    }
-    if (!getLabelBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 5, label_);
-    }
-    if (priority_ != 0) {
-      output.writeInt32(6, priority_);
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (!getOpBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, op_);
-    }
-    if (!getDeviceTypeBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(2, deviceType_);
-    }
-    for (int i = 0; i < constraint_.size(); i++) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(3, constraint_.get(i));
-    }
-    {
-      int dataSize = 0;
-      for (int i = 0; i < hostMemoryArg_.size(); i++) {
-        dataSize += computeStringSizeNoTag(hostMemoryArg_.getRaw(i));
-      }
-      size += dataSize;
-      size += 1 * getHostMemoryArgList().size();
-    }
-    if (!getLabelBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(5, label_);
-    }
-    if (priority_ != 0) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt32Size(6, priority_);
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.framework.KernelDef)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.framework.KernelDef other = (org.tensorflow.proto.framework.KernelDef) obj;
-
-    if (!getOp()
-        .equals(other.getOp())) return false;
-    if (!getDeviceType()
-        .equals(other.getDeviceType())) return false;
-    if (!getConstraintList()
-        .equals(other.getConstraintList())) return false;
-    if (!getHostMemoryArgList()
-        .equals(other.getHostMemoryArgList())) return false;
-    if (!getLabel()
-        .equals(other.getLabel())) return false;
-    if (getPriority()
-        != other.getPriority()) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    hash = (37 * hash) + OP_FIELD_NUMBER;
-    hash = (53 * hash) + getOp().hashCode();
-    hash = (37 * hash) + DEVICE_TYPE_FIELD_NUMBER;
-    hash = (53 * hash) + getDeviceType().hashCode();
-    if (getConstraintCount() > 0) {
-      hash = (37 * hash) + CONSTRAINT_FIELD_NUMBER;
-      hash = (53 * hash) + getConstraintList().hashCode();
-    }
-    if (getHostMemoryArgCount() > 0) {
-      hash = (37 * hash) + HOST_MEMORY_ARG_FIELD_NUMBER;
-      hash = (53 * hash) + getHostMemoryArgList().hashCode();
-    }
-    hash = (37 * hash) + LABEL_FIELD_NUMBER;
-    hash = (53 * hash) + getLabel().hashCode();
-    hash = (37 * hash) + PRIORITY_FIELD_NUMBER;
-    hash = (53 * hash) + getPriority();
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.framework.KernelDef parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.KernelDef parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.KernelDef parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.KernelDef parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.KernelDef parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.KernelDef parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.KernelDef parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.KernelDef parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.KernelDef parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.KernelDef parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.KernelDef parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.KernelDef parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.framework.KernelDef prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * Protobuf type {@code tensorflow.KernelDef}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.KernelDef)
-      org.tensorflow.proto.framework.KernelDefOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.KernelDefProtos.internal_static_tensorflow_KernelDef_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.KernelDefProtos.internal_static_tensorflow_KernelDef_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.KernelDef.class, org.tensorflow.proto.framework.KernelDef.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.framework.KernelDef.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-        getConstraintFieldBuilder();
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      op_ = "";
-
-      deviceType_ = "";
-
-      if (constraintBuilder_ == null) {
-        constraint_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000001);
-      } else {
-        constraintBuilder_.clear();
-      }
-      hostMemoryArg_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-      bitField0_ = (bitField0_ & ~0x00000002);
-      label_ = "";
-
-      priority_ = 0;
-
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.framework.KernelDefProtos.internal_static_tensorflow_KernelDef_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.KernelDef getDefaultInstanceForType() {
-      return org.tensorflow.proto.framework.KernelDef.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.KernelDef build() {
-      org.tensorflow.proto.framework.KernelDef result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.KernelDef buildPartial() {
-      org.tensorflow.proto.framework.KernelDef result = new org.tensorflow.proto.framework.KernelDef(this);
-      int from_bitField0_ = bitField0_;
-      result.op_ = op_;
-      result.deviceType_ = deviceType_;
-      if (constraintBuilder_ == null) {
-        if (((bitField0_ & 0x00000001) != 0)) {
-          constraint_ = java.util.Collections.unmodifiableList(constraint_);
-          bitField0_ = (bitField0_ & ~0x00000001);
-        }
-        result.constraint_ = constraint_;
-      } else {
-        result.constraint_ = constraintBuilder_.build();
-      }
-      if (((bitField0_ & 0x00000002) != 0)) {
-        hostMemoryArg_ = hostMemoryArg_.getUnmodifiableView();
-        bitField0_ = (bitField0_ & ~0x00000002);
-      }
-      result.hostMemoryArg_ = hostMemoryArg_;
-      result.label_ = label_;
-      result.priority_ = priority_;
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.framework.KernelDef) {
-        return mergeFrom((org.tensorflow.proto.framework.KernelDef)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.framework.KernelDef other) {
-      if (other == org.tensorflow.proto.framework.KernelDef.getDefaultInstance()) return this;
-      if (!other.getOp().isEmpty()) {
-        op_ = other.op_;
-        onChanged();
-      }
-      if (!other.getDeviceType().isEmpty()) {
-        deviceType_ = other.deviceType_;
-        onChanged();
-      }
-      if (constraintBuilder_ == null) {
-        if (!other.constraint_.isEmpty()) {
-          if (constraint_.isEmpty()) {
-            constraint_ = other.constraint_;
-            bitField0_ = (bitField0_ & ~0x00000001);
-          } else {
-            ensureConstraintIsMutable();
-            constraint_.addAll(other.constraint_);
-          }
-          onChanged();
-        }
-      } else {
-        if (!other.constraint_.isEmpty()) {
-          if (constraintBuilder_.isEmpty()) {
-            constraintBuilder_.dispose();
-            constraintBuilder_ = null;
-            constraint_ = other.constraint_;
-            bitField0_ = (bitField0_ & ~0x00000001);
-            constraintBuilder_ = 
-              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
-                 getConstraintFieldBuilder() : null;
-          } else {
-            constraintBuilder_.addAllMessages(other.constraint_);
-          }
-        }
-      }
-      if (!other.hostMemoryArg_.isEmpty()) {
-        if (hostMemoryArg_.isEmpty()) {
-          hostMemoryArg_ = other.hostMemoryArg_;
-          bitField0_ = (bitField0_ & ~0x00000002);
-        } else {
-          ensureHostMemoryArgIsMutable();
-          hostMemoryArg_.addAll(other.hostMemoryArg_);
-        }
-        onChanged();
-      }
-      if (!other.getLabel().isEmpty()) {
-        label_ = other.label_;
-        onChanged();
-      }
-      if (other.getPriority() != 0) {
-        setPriority(other.getPriority());
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.framework.KernelDef parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.framework.KernelDef) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-    private int bitField0_;
-
-    private java.lang.Object op_ = "";
-    /**
-     * <pre>
-     * Must match the name of an Op.
-     * </pre>
-     *
-     * <code>string op = 1;</code>
-     */
-    public java.lang.String getOp() {
-      java.lang.Object ref = op_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        op_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Must match the name of an Op.
-     * </pre>
-     *
-     * <code>string op = 1;</code>
-     */
-    public com.google.protobuf.ByteString
-        getOpBytes() {
-      java.lang.Object ref = op_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        op_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Must match the name of an Op.
-     * </pre>
-     *
-     * <code>string op = 1;</code>
-     */
-    public Builder setOp(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      op_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Must match the name of an Op.
-     * </pre>
-     *
-     * <code>string op = 1;</code>
-     */
-    public Builder clearOp() {
-      
-      op_ = getDefaultInstance().getOp();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Must match the name of an Op.
-     * </pre>
-     *
-     * <code>string op = 1;</code>
-     */
-    public Builder setOpBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      op_ = value;
-      onChanged();
-      return this;
-    }
-
-    private java.lang.Object deviceType_ = "";
-    /**
-     * <pre>
-     * Type of device this kernel runs on.
-     * </pre>
-     *
-     * <code>string device_type = 2;</code>
-     */
-    public java.lang.String getDeviceType() {
-      java.lang.Object ref = deviceType_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        deviceType_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Type of device this kernel runs on.
-     * </pre>
-     *
-     * <code>string device_type = 2;</code>
-     */
-    public com.google.protobuf.ByteString
-        getDeviceTypeBytes() {
-      java.lang.Object ref = deviceType_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        deviceType_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Type of device this kernel runs on.
-     * </pre>
-     *
-     * <code>string device_type = 2;</code>
-     */
-    public Builder setDeviceType(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      deviceType_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Type of device this kernel runs on.
-     * </pre>
-     *
-     * <code>string device_type = 2;</code>
-     */
-    public Builder clearDeviceType() {
-      
-      deviceType_ = getDefaultInstance().getDeviceType();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Type of device this kernel runs on.
-     * </pre>
-     *
-     * <code>string device_type = 2;</code>
-     */
-    public Builder setDeviceTypeBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      deviceType_ = value;
-      onChanged();
-      return this;
-    }
-
-    private java.util.List<org.tensorflow.proto.framework.KernelDef.AttrConstraint> constraint_ =
-      java.util.Collections.emptyList();
-    private void ensureConstraintIsMutable() {
-      if (!((bitField0_ & 0x00000001) != 0)) {
-        constraint_ = new java.util.ArrayList<org.tensorflow.proto.framework.KernelDef.AttrConstraint>(constraint_);
-        bitField0_ |= 0x00000001;
-       }
-    }
-
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.framework.KernelDef.AttrConstraint, org.tensorflow.proto.framework.KernelDef.AttrConstraint.Builder, org.tensorflow.proto.framework.KernelDef.AttrConstraintOrBuilder> constraintBuilder_;
-
-    /**
-     * <code>repeated .tensorflow.KernelDef.AttrConstraint constraint = 3;</code>
-     */
-    public java.util.List<org.tensorflow.proto.framework.KernelDef.AttrConstraint> getConstraintList() {
-      if (constraintBuilder_ == null) {
-        return java.util.Collections.unmodifiableList(constraint_);
-      } else {
-        return constraintBuilder_.getMessageList();
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.KernelDef.AttrConstraint constraint = 3;</code>
-     */
-    public int getConstraintCount() {
-      if (constraintBuilder_ == null) {
-        return constraint_.size();
-      } else {
-        return constraintBuilder_.getCount();
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.KernelDef.AttrConstraint constraint = 3;</code>
-     */
-    public org.tensorflow.proto.framework.KernelDef.AttrConstraint getConstraint(int index) {
-      if (constraintBuilder_ == null) {
-        return constraint_.get(index);
-      } else {
-        return constraintBuilder_.getMessage(index);
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.KernelDef.AttrConstraint constraint = 3;</code>
-     */
-    public Builder setConstraint(
-        int index, org.tensorflow.proto.framework.KernelDef.AttrConstraint value) {
-      if (constraintBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureConstraintIsMutable();
-        constraint_.set(index, value);
-        onChanged();
-      } else {
-        constraintBuilder_.setMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.KernelDef.AttrConstraint constraint = 3;</code>
-     */
-    public Builder setConstraint(
-        int index, org.tensorflow.proto.framework.KernelDef.AttrConstraint.Builder builderForValue) {
-      if (constraintBuilder_ == null) {
-        ensureConstraintIsMutable();
-        constraint_.set(index, builderForValue.build());
-        onChanged();
-      } else {
-        constraintBuilder_.setMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.KernelDef.AttrConstraint constraint = 3;</code>
-     */
-    public Builder addConstraint(org.tensorflow.proto.framework.KernelDef.AttrConstraint value) {
-      if (constraintBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureConstraintIsMutable();
-        constraint_.add(value);
-        onChanged();
-      } else {
-        constraintBuilder_.addMessage(value);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.KernelDef.AttrConstraint constraint = 3;</code>
-     */
-    public Builder addConstraint(
-        int index, org.tensorflow.proto.framework.KernelDef.AttrConstraint value) {
-      if (constraintBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureConstraintIsMutable();
-        constraint_.add(index, value);
-        onChanged();
-      } else {
-        constraintBuilder_.addMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.KernelDef.AttrConstraint constraint = 3;</code>
-     */
-    public Builder addConstraint(
-        org.tensorflow.proto.framework.KernelDef.AttrConstraint.Builder builderForValue) {
-      if (constraintBuilder_ == null) {
-        ensureConstraintIsMutable();
-        constraint_.add(builderForValue.build());
-        onChanged();
-      } else {
-        constraintBuilder_.addMessage(builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.KernelDef.AttrConstraint constraint = 3;</code>
-     */
-    public Builder addConstraint(
-        int index, org.tensorflow.proto.framework.KernelDef.AttrConstraint.Builder builderForValue) {
-      if (constraintBuilder_ == null) {
-        ensureConstraintIsMutable();
-        constraint_.add(index, builderForValue.build());
-        onChanged();
-      } else {
-        constraintBuilder_.addMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.KernelDef.AttrConstraint constraint = 3;</code>
-     */
-    public Builder addAllConstraint(
-        java.lang.Iterable<? extends org.tensorflow.proto.framework.KernelDef.AttrConstraint> values) {
-      if (constraintBuilder_ == null) {
-        ensureConstraintIsMutable();
-        com.google.protobuf.AbstractMessageLite.Builder.addAll(
-            values, constraint_);
-        onChanged();
-      } else {
-        constraintBuilder_.addAllMessages(values);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.KernelDef.AttrConstraint constraint = 3;</code>
-     */
-    public Builder clearConstraint() {
-      if (constraintBuilder_ == null) {
-        constraint_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000001);
-        onChanged();
-      } else {
-        constraintBuilder_.clear();
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.KernelDef.AttrConstraint constraint = 3;</code>
-     */
-    public Builder removeConstraint(int index) {
-      if (constraintBuilder_ == null) {
-        ensureConstraintIsMutable();
-        constraint_.remove(index);
-        onChanged();
-      } else {
-        constraintBuilder_.remove(index);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.KernelDef.AttrConstraint constraint = 3;</code>
-     */
-    public org.tensorflow.proto.framework.KernelDef.AttrConstraint.Builder getConstraintBuilder(
-        int index) {
-      return getConstraintFieldBuilder().getBuilder(index);
-    }
-    /**
-     * <code>repeated .tensorflow.KernelDef.AttrConstraint constraint = 3;</code>
-     */
-    public org.tensorflow.proto.framework.KernelDef.AttrConstraintOrBuilder getConstraintOrBuilder(
-        int index) {
-      if (constraintBuilder_ == null) {
-        return constraint_.get(index);  } else {
-        return constraintBuilder_.getMessageOrBuilder(index);
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.KernelDef.AttrConstraint constraint = 3;</code>
-     */
-    public java.util.List<? extends org.tensorflow.proto.framework.KernelDef.AttrConstraintOrBuilder> 
-         getConstraintOrBuilderList() {
-      if (constraintBuilder_ != null) {
-        return constraintBuilder_.getMessageOrBuilderList();
-      } else {
-        return java.util.Collections.unmodifiableList(constraint_);
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.KernelDef.AttrConstraint constraint = 3;</code>
-     */
-    public org.tensorflow.proto.framework.KernelDef.AttrConstraint.Builder addConstraintBuilder() {
-      return getConstraintFieldBuilder().addBuilder(
-          org.tensorflow.proto.framework.KernelDef.AttrConstraint.getDefaultInstance());
-    }
-    /**
-     * <code>repeated .tensorflow.KernelDef.AttrConstraint constraint = 3;</code>
-     */
-    public org.tensorflow.proto.framework.KernelDef.AttrConstraint.Builder addConstraintBuilder(
-        int index) {
-      return getConstraintFieldBuilder().addBuilder(
-          index, org.tensorflow.proto.framework.KernelDef.AttrConstraint.getDefaultInstance());
-    }
-    /**
-     * <code>repeated .tensorflow.KernelDef.AttrConstraint constraint = 3;</code>
-     */
-    public java.util.List<org.tensorflow.proto.framework.KernelDef.AttrConstraint.Builder> 
-         getConstraintBuilderList() {
-      return getConstraintFieldBuilder().getBuilderList();
-    }
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.framework.KernelDef.AttrConstraint, org.tensorflow.proto.framework.KernelDef.AttrConstraint.Builder, org.tensorflow.proto.framework.KernelDef.AttrConstraintOrBuilder> 
-        getConstraintFieldBuilder() {
-      if (constraintBuilder_ == null) {
-        constraintBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
-            org.tensorflow.proto.framework.KernelDef.AttrConstraint, org.tensorflow.proto.framework.KernelDef.AttrConstraint.Builder, org.tensorflow.proto.framework.KernelDef.AttrConstraintOrBuilder>(
-                constraint_,
-                ((bitField0_ & 0x00000001) != 0),
-                getParentForChildren(),
-                isClean());
-        constraint_ = null;
-      }
-      return constraintBuilder_;
-    }
-
-    private com.google.protobuf.LazyStringList hostMemoryArg_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-    private void ensureHostMemoryArgIsMutable() {
-      if (!((bitField0_ & 0x00000002) != 0)) {
-        hostMemoryArg_ = new com.google.protobuf.LazyStringArrayList(hostMemoryArg_);
-        bitField0_ |= 0x00000002;
-       }
-    }
-    /**
-     * <pre>
-     * Names of the Op's input_/output_args that reside in host memory
-     * instead of device memory.
-     * </pre>
-     *
-     * <code>repeated string host_memory_arg = 4;</code>
-     */
-    public com.google.protobuf.ProtocolStringList
-        getHostMemoryArgList() {
-      return hostMemoryArg_.getUnmodifiableView();
-    }
-    /**
-     * <pre>
-     * Names of the Op's input_/output_args that reside in host memory
-     * instead of device memory.
-     * </pre>
-     *
-     * <code>repeated string host_memory_arg = 4;</code>
-     */
-    public int getHostMemoryArgCount() {
-      return hostMemoryArg_.size();
-    }
-    /**
-     * <pre>
-     * Names of the Op's input_/output_args that reside in host memory
-     * instead of device memory.
-     * </pre>
-     *
-     * <code>repeated string host_memory_arg = 4;</code>
-     */
-    public java.lang.String getHostMemoryArg(int index) {
-      return hostMemoryArg_.get(index);
-    }
-    /**
-     * <pre>
-     * Names of the Op's input_/output_args that reside in host memory
-     * instead of device memory.
-     * </pre>
-     *
-     * <code>repeated string host_memory_arg = 4;</code>
-     */
-    public com.google.protobuf.ByteString
-        getHostMemoryArgBytes(int index) {
-      return hostMemoryArg_.getByteString(index);
-    }
-    /**
-     * <pre>
-     * Names of the Op's input_/output_args that reside in host memory
-     * instead of device memory.
-     * </pre>
-     *
-     * <code>repeated string host_memory_arg = 4;</code>
-     */
-    public Builder setHostMemoryArg(
-        int index, java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  ensureHostMemoryArgIsMutable();
-      hostMemoryArg_.set(index, value);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Names of the Op's input_/output_args that reside in host memory
-     * instead of device memory.
-     * </pre>
-     *
-     * <code>repeated string host_memory_arg = 4;</code>
-     */
-    public Builder addHostMemoryArg(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  ensureHostMemoryArgIsMutable();
-      hostMemoryArg_.add(value);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Names of the Op's input_/output_args that reside in host memory
-     * instead of device memory.
-     * </pre>
-     *
-     * <code>repeated string host_memory_arg = 4;</code>
-     */
-    public Builder addAllHostMemoryArg(
-        java.lang.Iterable<java.lang.String> values) {
-      ensureHostMemoryArgIsMutable();
-      com.google.protobuf.AbstractMessageLite.Builder.addAll(
-          values, hostMemoryArg_);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Names of the Op's input_/output_args that reside in host memory
-     * instead of device memory.
-     * </pre>
-     *
-     * <code>repeated string host_memory_arg = 4;</code>
-     */
-    public Builder clearHostMemoryArg() {
-      hostMemoryArg_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-      bitField0_ = (bitField0_ & ~0x00000002);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Names of the Op's input_/output_args that reside in host memory
-     * instead of device memory.
-     * </pre>
-     *
-     * <code>repeated string host_memory_arg = 4;</code>
-     */
-    public Builder addHostMemoryArgBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      ensureHostMemoryArgIsMutable();
-      hostMemoryArg_.add(value);
-      onChanged();
-      return this;
-    }
-
-    private java.lang.Object label_ = "";
-    /**
-     * <pre>
-     * This allows experimental kernels to be registered for an op that
-     * won't be used unless the user specifies a "_kernel" attr with
-     * value matching this.
-     * </pre>
-     *
-     * <code>string label = 5;</code>
-     */
-    public java.lang.String getLabel() {
-      java.lang.Object ref = label_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        label_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * This allows experimental kernels to be registered for an op that
-     * won't be used unless the user specifies a "_kernel" attr with
-     * value matching this.
-     * </pre>
-     *
-     * <code>string label = 5;</code>
-     */
-    public com.google.protobuf.ByteString
-        getLabelBytes() {
-      java.lang.Object ref = label_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        label_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * This allows experimental kernels to be registered for an op that
-     * won't be used unless the user specifies a "_kernel" attr with
-     * value matching this.
-     * </pre>
-     *
-     * <code>string label = 5;</code>
-     */
-    public Builder setLabel(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      label_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * This allows experimental kernels to be registered for an op that
-     * won't be used unless the user specifies a "_kernel" attr with
-     * value matching this.
-     * </pre>
-     *
-     * <code>string label = 5;</code>
-     */
-    public Builder clearLabel() {
-      
-      label_ = getDefaultInstance().getLabel();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * This allows experimental kernels to be registered for an op that
-     * won't be used unless the user specifies a "_kernel" attr with
-     * value matching this.
-     * </pre>
-     *
-     * <code>string label = 5;</code>
-     */
-    public Builder setLabelBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      label_ = value;
-      onChanged();
-      return this;
-    }
-
-    private int priority_ ;
-    /**
-     * <pre>
-     * Prioritization of kernel amongst different devices. By default we assume
-     * priority is 0. The higher the priority the better. By default (i.e. if
-     * this is not set), we prefer GPU kernels over CPU.
-     * </pre>
-     *
-     * <code>int32 priority = 6;</code>
-     */
-    public int getPriority() {
-      return priority_;
-    }
-    /**
-     * <pre>
-     * Prioritization of kernel amongst different devices. By default we assume
-     * priority is 0. The higher the priority the better. By default (i.e. if
-     * this is not set), we prefer GPU kernels over CPU.
-     * </pre>
-     *
-     * <code>int32 priority = 6;</code>
-     */
-    public Builder setPriority(int value) {
-      
-      priority_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Prioritization of kernel amongst different devices. By default we assume
-     * priority is 0. The higher the priority the better. By default (i.e. if
-     * this is not set), we prefer GPU kernels over CPU.
-     * </pre>
-     *
-     * <code>int32 priority = 6;</code>
-     */
-    public Builder clearPriority() {
-      
-      priority_ = 0;
-      onChanged();
-      return this;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.KernelDef)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.KernelDef)
-  private static final org.tensorflow.proto.framework.KernelDef DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.KernelDef();
-  }
-
-  public static org.tensorflow.proto.framework.KernelDef getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<KernelDef>
-      PARSER = new com.google.protobuf.AbstractParser<KernelDef>() {
-    @java.lang.Override
-    public KernelDef parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new KernelDef(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<KernelDef> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<KernelDef> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.framework.KernelDef getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/KernelDefOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/KernelDefOrBuilder.java
deleted file mode 100644
index 2d62f81acc3..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/KernelDefOrBuilder.java
+++ /dev/null
@@ -1,141 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/framework/kernel_def.proto
-
-package org.tensorflow.proto.framework;
-
-public interface KernelDefOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.KernelDef)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <pre>
-   * Must match the name of an Op.
-   * </pre>
-   *
-   * <code>string op = 1;</code>
-   */
-  java.lang.String getOp();
-  /**
-   * <pre>
-   * Must match the name of an Op.
-   * </pre>
-   *
-   * <code>string op = 1;</code>
-   */
-  com.google.protobuf.ByteString
-      getOpBytes();
-
-  /**
-   * <pre>
-   * Type of device this kernel runs on.
-   * </pre>
-   *
-   * <code>string device_type = 2;</code>
-   */
-  java.lang.String getDeviceType();
-  /**
-   * <pre>
-   * Type of device this kernel runs on.
-   * </pre>
-   *
-   * <code>string device_type = 2;</code>
-   */
-  com.google.protobuf.ByteString
-      getDeviceTypeBytes();
-
-  /**
-   * <code>repeated .tensorflow.KernelDef.AttrConstraint constraint = 3;</code>
-   */
-  java.util.List<org.tensorflow.proto.framework.KernelDef.AttrConstraint> 
-      getConstraintList();
-  /**
-   * <code>repeated .tensorflow.KernelDef.AttrConstraint constraint = 3;</code>
-   */
-  org.tensorflow.proto.framework.KernelDef.AttrConstraint getConstraint(int index);
-  /**
-   * <code>repeated .tensorflow.KernelDef.AttrConstraint constraint = 3;</code>
-   */
-  int getConstraintCount();
-  /**
-   * <code>repeated .tensorflow.KernelDef.AttrConstraint constraint = 3;</code>
-   */
-  java.util.List<? extends org.tensorflow.proto.framework.KernelDef.AttrConstraintOrBuilder> 
-      getConstraintOrBuilderList();
-  /**
-   * <code>repeated .tensorflow.KernelDef.AttrConstraint constraint = 3;</code>
-   */
-  org.tensorflow.proto.framework.KernelDef.AttrConstraintOrBuilder getConstraintOrBuilder(
-      int index);
-
-  /**
-   * <pre>
-   * Names of the Op's input_/output_args that reside in host memory
-   * instead of device memory.
-   * </pre>
-   *
-   * <code>repeated string host_memory_arg = 4;</code>
-   */
-  java.util.List<java.lang.String>
-      getHostMemoryArgList();
-  /**
-   * <pre>
-   * Names of the Op's input_/output_args that reside in host memory
-   * instead of device memory.
-   * </pre>
-   *
-   * <code>repeated string host_memory_arg = 4;</code>
-   */
-  int getHostMemoryArgCount();
-  /**
-   * <pre>
-   * Names of the Op's input_/output_args that reside in host memory
-   * instead of device memory.
-   * </pre>
-   *
-   * <code>repeated string host_memory_arg = 4;</code>
-   */
-  java.lang.String getHostMemoryArg(int index);
-  /**
-   * <pre>
-   * Names of the Op's input_/output_args that reside in host memory
-   * instead of device memory.
-   * </pre>
-   *
-   * <code>repeated string host_memory_arg = 4;</code>
-   */
-  com.google.protobuf.ByteString
-      getHostMemoryArgBytes(int index);
-
-  /**
-   * <pre>
-   * This allows experimental kernels to be registered for an op that
-   * won't be used unless the user specifies a "_kernel" attr with
-   * value matching this.
-   * </pre>
-   *
-   * <code>string label = 5;</code>
-   */
-  java.lang.String getLabel();
-  /**
-   * <pre>
-   * This allows experimental kernels to be registered for an op that
-   * won't be used unless the user specifies a "_kernel" attr with
-   * value matching this.
-   * </pre>
-   *
-   * <code>string label = 5;</code>
-   */
-  com.google.protobuf.ByteString
-      getLabelBytes();
-
-  /**
-   * <pre>
-   * Prioritization of kernel amongst different devices. By default we assume
-   * priority is 0. The higher the priority the better. By default (i.e. if
-   * this is not set), we prefer GPU kernels over CPU.
-   * </pre>
-   *
-   * <code>int32 priority = 6;</code>
-   */
-  int getPriority();
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/KernelDefProtos.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/KernelDefProtos.java
deleted file mode 100644
index 50b235b600f..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/KernelDefProtos.java
+++ /dev/null
@@ -1,83 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/framework/kernel_def.proto
-
-package org.tensorflow.proto.framework;
-
-public final class KernelDefProtos {
-  private KernelDefProtos() {}
-  public static void registerAllExtensions(
-      com.google.protobuf.ExtensionRegistryLite registry) {
-  }
-
-  public static void registerAllExtensions(
-      com.google.protobuf.ExtensionRegistry registry) {
-    registerAllExtensions(
-        (com.google.protobuf.ExtensionRegistryLite) registry);
-  }
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_KernelDef_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_KernelDef_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_KernelDef_AttrConstraint_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_KernelDef_AttrConstraint_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_KernelList_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_KernelList_fieldAccessorTable;
-
-  public static com.google.protobuf.Descriptors.FileDescriptor
-      getDescriptor() {
-    return descriptor;
-  }
-  private static  com.google.protobuf.Descriptors.FileDescriptor
-      descriptor;
-  static {
-    java.lang.String[] descriptorData = {
-      "\n*tensorflow/core/framework/kernel_def.p" +
-      "roto\022\ntensorflow\032*tensorflow/core/framew" +
-      "ork/attr_value.proto\"\357\001\n\tKernelDef\022\n\n\002op" +
-      "\030\001 \001(\t\022\023\n\013device_type\030\002 \001(\t\0228\n\nconstrain" +
-      "t\030\003 \003(\0132$.tensorflow.KernelDef.AttrConst" +
-      "raint\022\027\n\017host_memory_arg\030\004 \003(\t\022\r\n\005label\030" +
-      "\005 \001(\t\022\020\n\010priority\030\006 \001(\005\032M\n\016AttrConstrain" +
-      "t\022\014\n\004name\030\001 \001(\t\022-\n\016allowed_values\030\002 \001(\0132" +
-      "\025.tensorflow.AttrValue\"3\n\nKernelList\022%\n\006" +
-      "kernel\030\001 \003(\0132\025.tensorflow.KernelDefB\211\001\n\036" +
-      "org.tensorflow.proto.frameworkB\017KernelDe" +
-      "fProtosP\001ZQgithub.com/tensorflow/tensorf" +
-      "low/tensorflow/go/core/framework/kernel_" +
-      "def_go_proto\370\001\001b\006proto3"
-    };
-    descriptor = com.google.protobuf.Descriptors.FileDescriptor
-      .internalBuildGeneratedFileFrom(descriptorData,
-        new com.google.protobuf.Descriptors.FileDescriptor[] {
-          org.tensorflow.proto.framework.AttrValueProtos.getDescriptor(),
-        });
-    internal_static_tensorflow_KernelDef_descriptor =
-      getDescriptor().getMessageTypes().get(0);
-    internal_static_tensorflow_KernelDef_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_KernelDef_descriptor,
-        new java.lang.String[] { "Op", "DeviceType", "Constraint", "HostMemoryArg", "Label", "Priority", });
-    internal_static_tensorflow_KernelDef_AttrConstraint_descriptor =
-      internal_static_tensorflow_KernelDef_descriptor.getNestedTypes().get(0);
-    internal_static_tensorflow_KernelDef_AttrConstraint_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_KernelDef_AttrConstraint_descriptor,
-        new java.lang.String[] { "Name", "AllowedValues", });
-    internal_static_tensorflow_KernelList_descriptor =
-      getDescriptor().getMessageTypes().get(1);
-    internal_static_tensorflow_KernelList_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_KernelList_descriptor,
-        new java.lang.String[] { "Kernel", });
-    org.tensorflow.proto.framework.AttrValueProtos.getDescriptor();
-  }
-
-  // @@protoc_insertion_point(outer_class_scope)
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/KernelList.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/KernelList.java
deleted file mode 100644
index 06c6e5a22f2..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/KernelList.java
+++ /dev/null
@@ -1,773 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/framework/kernel_def.proto
-
-package org.tensorflow.proto.framework;
-
-/**
- * <pre>
- * A collection of KernelDefs
- * </pre>
- *
- * Protobuf type {@code tensorflow.KernelList}
- */
-public  final class KernelList extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.KernelList)
-    KernelListOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use KernelList.newBuilder() to construct.
-  private KernelList(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private KernelList() {
-    kernel_ = java.util.Collections.emptyList();
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new KernelList();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private KernelList(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    int mutable_bitField0_ = 0;
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 10: {
-            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
-              kernel_ = new java.util.ArrayList<org.tensorflow.proto.framework.KernelDef>();
-              mutable_bitField0_ |= 0x00000001;
-            }
-            kernel_.add(
-                input.readMessage(org.tensorflow.proto.framework.KernelDef.parser(), extensionRegistry));
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      if (((mutable_bitField0_ & 0x00000001) != 0)) {
-        kernel_ = java.util.Collections.unmodifiableList(kernel_);
-      }
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.framework.KernelDefProtos.internal_static_tensorflow_KernelList_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.framework.KernelDefProtos.internal_static_tensorflow_KernelList_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.framework.KernelList.class, org.tensorflow.proto.framework.KernelList.Builder.class);
-  }
-
-  public static final int KERNEL_FIELD_NUMBER = 1;
-  private java.util.List<org.tensorflow.proto.framework.KernelDef> kernel_;
-  /**
-   * <code>repeated .tensorflow.KernelDef kernel = 1;</code>
-   */
-  public java.util.List<org.tensorflow.proto.framework.KernelDef> getKernelList() {
-    return kernel_;
-  }
-  /**
-   * <code>repeated .tensorflow.KernelDef kernel = 1;</code>
-   */
-  public java.util.List<? extends org.tensorflow.proto.framework.KernelDefOrBuilder> 
-      getKernelOrBuilderList() {
-    return kernel_;
-  }
-  /**
-   * <code>repeated .tensorflow.KernelDef kernel = 1;</code>
-   */
-  public int getKernelCount() {
-    return kernel_.size();
-  }
-  /**
-   * <code>repeated .tensorflow.KernelDef kernel = 1;</code>
-   */
-  public org.tensorflow.proto.framework.KernelDef getKernel(int index) {
-    return kernel_.get(index);
-  }
-  /**
-   * <code>repeated .tensorflow.KernelDef kernel = 1;</code>
-   */
-  public org.tensorflow.proto.framework.KernelDefOrBuilder getKernelOrBuilder(
-      int index) {
-    return kernel_.get(index);
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    for (int i = 0; i < kernel_.size(); i++) {
-      output.writeMessage(1, kernel_.get(i));
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    for (int i = 0; i < kernel_.size(); i++) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(1, kernel_.get(i));
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.framework.KernelList)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.framework.KernelList other = (org.tensorflow.proto.framework.KernelList) obj;
-
-    if (!getKernelList()
-        .equals(other.getKernelList())) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    if (getKernelCount() > 0) {
-      hash = (37 * hash) + KERNEL_FIELD_NUMBER;
-      hash = (53 * hash) + getKernelList().hashCode();
-    }
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.framework.KernelList parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.KernelList parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.KernelList parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.KernelList parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.KernelList parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.KernelList parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.KernelList parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.KernelList parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.KernelList parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.KernelList parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.KernelList parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.KernelList parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.framework.KernelList prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * <pre>
-   * A collection of KernelDefs
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.KernelList}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.KernelList)
-      org.tensorflow.proto.framework.KernelListOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.KernelDefProtos.internal_static_tensorflow_KernelList_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.KernelDefProtos.internal_static_tensorflow_KernelList_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.KernelList.class, org.tensorflow.proto.framework.KernelList.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.framework.KernelList.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-        getKernelFieldBuilder();
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      if (kernelBuilder_ == null) {
-        kernel_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000001);
-      } else {
-        kernelBuilder_.clear();
-      }
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.framework.KernelDefProtos.internal_static_tensorflow_KernelList_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.KernelList getDefaultInstanceForType() {
-      return org.tensorflow.proto.framework.KernelList.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.KernelList build() {
-      org.tensorflow.proto.framework.KernelList result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.KernelList buildPartial() {
-      org.tensorflow.proto.framework.KernelList result = new org.tensorflow.proto.framework.KernelList(this);
-      int from_bitField0_ = bitField0_;
-      if (kernelBuilder_ == null) {
-        if (((bitField0_ & 0x00000001) != 0)) {
-          kernel_ = java.util.Collections.unmodifiableList(kernel_);
-          bitField0_ = (bitField0_ & ~0x00000001);
-        }
-        result.kernel_ = kernel_;
-      } else {
-        result.kernel_ = kernelBuilder_.build();
-      }
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.framework.KernelList) {
-        return mergeFrom((org.tensorflow.proto.framework.KernelList)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.framework.KernelList other) {
-      if (other == org.tensorflow.proto.framework.KernelList.getDefaultInstance()) return this;
-      if (kernelBuilder_ == null) {
-        if (!other.kernel_.isEmpty()) {
-          if (kernel_.isEmpty()) {
-            kernel_ = other.kernel_;
-            bitField0_ = (bitField0_ & ~0x00000001);
-          } else {
-            ensureKernelIsMutable();
-            kernel_.addAll(other.kernel_);
-          }
-          onChanged();
-        }
-      } else {
-        if (!other.kernel_.isEmpty()) {
-          if (kernelBuilder_.isEmpty()) {
-            kernelBuilder_.dispose();
-            kernelBuilder_ = null;
-            kernel_ = other.kernel_;
-            bitField0_ = (bitField0_ & ~0x00000001);
-            kernelBuilder_ = 
-              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
-                 getKernelFieldBuilder() : null;
-          } else {
-            kernelBuilder_.addAllMessages(other.kernel_);
-          }
-        }
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.framework.KernelList parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.framework.KernelList) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-    private int bitField0_;
-
-    private java.util.List<org.tensorflow.proto.framework.KernelDef> kernel_ =
-      java.util.Collections.emptyList();
-    private void ensureKernelIsMutable() {
-      if (!((bitField0_ & 0x00000001) != 0)) {
-        kernel_ = new java.util.ArrayList<org.tensorflow.proto.framework.KernelDef>(kernel_);
-        bitField0_ |= 0x00000001;
-       }
-    }
-
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.framework.KernelDef, org.tensorflow.proto.framework.KernelDef.Builder, org.tensorflow.proto.framework.KernelDefOrBuilder> kernelBuilder_;
-
-    /**
-     * <code>repeated .tensorflow.KernelDef kernel = 1;</code>
-     */
-    public java.util.List<org.tensorflow.proto.framework.KernelDef> getKernelList() {
-      if (kernelBuilder_ == null) {
-        return java.util.Collections.unmodifiableList(kernel_);
-      } else {
-        return kernelBuilder_.getMessageList();
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.KernelDef kernel = 1;</code>
-     */
-    public int getKernelCount() {
-      if (kernelBuilder_ == null) {
-        return kernel_.size();
-      } else {
-        return kernelBuilder_.getCount();
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.KernelDef kernel = 1;</code>
-     */
-    public org.tensorflow.proto.framework.KernelDef getKernel(int index) {
-      if (kernelBuilder_ == null) {
-        return kernel_.get(index);
-      } else {
-        return kernelBuilder_.getMessage(index);
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.KernelDef kernel = 1;</code>
-     */
-    public Builder setKernel(
-        int index, org.tensorflow.proto.framework.KernelDef value) {
-      if (kernelBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureKernelIsMutable();
-        kernel_.set(index, value);
-        onChanged();
-      } else {
-        kernelBuilder_.setMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.KernelDef kernel = 1;</code>
-     */
-    public Builder setKernel(
-        int index, org.tensorflow.proto.framework.KernelDef.Builder builderForValue) {
-      if (kernelBuilder_ == null) {
-        ensureKernelIsMutable();
-        kernel_.set(index, builderForValue.build());
-        onChanged();
-      } else {
-        kernelBuilder_.setMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.KernelDef kernel = 1;</code>
-     */
-    public Builder addKernel(org.tensorflow.proto.framework.KernelDef value) {
-      if (kernelBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureKernelIsMutable();
-        kernel_.add(value);
-        onChanged();
-      } else {
-        kernelBuilder_.addMessage(value);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.KernelDef kernel = 1;</code>
-     */
-    public Builder addKernel(
-        int index, org.tensorflow.proto.framework.KernelDef value) {
-      if (kernelBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureKernelIsMutable();
-        kernel_.add(index, value);
-        onChanged();
-      } else {
-        kernelBuilder_.addMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.KernelDef kernel = 1;</code>
-     */
-    public Builder addKernel(
-        org.tensorflow.proto.framework.KernelDef.Builder builderForValue) {
-      if (kernelBuilder_ == null) {
-        ensureKernelIsMutable();
-        kernel_.add(builderForValue.build());
-        onChanged();
-      } else {
-        kernelBuilder_.addMessage(builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.KernelDef kernel = 1;</code>
-     */
-    public Builder addKernel(
-        int index, org.tensorflow.proto.framework.KernelDef.Builder builderForValue) {
-      if (kernelBuilder_ == null) {
-        ensureKernelIsMutable();
-        kernel_.add(index, builderForValue.build());
-        onChanged();
-      } else {
-        kernelBuilder_.addMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.KernelDef kernel = 1;</code>
-     */
-    public Builder addAllKernel(
-        java.lang.Iterable<? extends org.tensorflow.proto.framework.KernelDef> values) {
-      if (kernelBuilder_ == null) {
-        ensureKernelIsMutable();
-        com.google.protobuf.AbstractMessageLite.Builder.addAll(
-            values, kernel_);
-        onChanged();
-      } else {
-        kernelBuilder_.addAllMessages(values);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.KernelDef kernel = 1;</code>
-     */
-    public Builder clearKernel() {
-      if (kernelBuilder_ == null) {
-        kernel_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000001);
-        onChanged();
-      } else {
-        kernelBuilder_.clear();
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.KernelDef kernel = 1;</code>
-     */
-    public Builder removeKernel(int index) {
-      if (kernelBuilder_ == null) {
-        ensureKernelIsMutable();
-        kernel_.remove(index);
-        onChanged();
-      } else {
-        kernelBuilder_.remove(index);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.KernelDef kernel = 1;</code>
-     */
-    public org.tensorflow.proto.framework.KernelDef.Builder getKernelBuilder(
-        int index) {
-      return getKernelFieldBuilder().getBuilder(index);
-    }
-    /**
-     * <code>repeated .tensorflow.KernelDef kernel = 1;</code>
-     */
-    public org.tensorflow.proto.framework.KernelDefOrBuilder getKernelOrBuilder(
-        int index) {
-      if (kernelBuilder_ == null) {
-        return kernel_.get(index);  } else {
-        return kernelBuilder_.getMessageOrBuilder(index);
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.KernelDef kernel = 1;</code>
-     */
-    public java.util.List<? extends org.tensorflow.proto.framework.KernelDefOrBuilder> 
-         getKernelOrBuilderList() {
-      if (kernelBuilder_ != null) {
-        return kernelBuilder_.getMessageOrBuilderList();
-      } else {
-        return java.util.Collections.unmodifiableList(kernel_);
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.KernelDef kernel = 1;</code>
-     */
-    public org.tensorflow.proto.framework.KernelDef.Builder addKernelBuilder() {
-      return getKernelFieldBuilder().addBuilder(
-          org.tensorflow.proto.framework.KernelDef.getDefaultInstance());
-    }
-    /**
-     * <code>repeated .tensorflow.KernelDef kernel = 1;</code>
-     */
-    public org.tensorflow.proto.framework.KernelDef.Builder addKernelBuilder(
-        int index) {
-      return getKernelFieldBuilder().addBuilder(
-          index, org.tensorflow.proto.framework.KernelDef.getDefaultInstance());
-    }
-    /**
-     * <code>repeated .tensorflow.KernelDef kernel = 1;</code>
-     */
-    public java.util.List<org.tensorflow.proto.framework.KernelDef.Builder> 
-         getKernelBuilderList() {
-      return getKernelFieldBuilder().getBuilderList();
-    }
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.framework.KernelDef, org.tensorflow.proto.framework.KernelDef.Builder, org.tensorflow.proto.framework.KernelDefOrBuilder> 
-        getKernelFieldBuilder() {
-      if (kernelBuilder_ == null) {
-        kernelBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
-            org.tensorflow.proto.framework.KernelDef, org.tensorflow.proto.framework.KernelDef.Builder, org.tensorflow.proto.framework.KernelDefOrBuilder>(
-                kernel_,
-                ((bitField0_ & 0x00000001) != 0),
-                getParentForChildren(),
-                isClean());
-        kernel_ = null;
-      }
-      return kernelBuilder_;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.KernelList)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.KernelList)
-  private static final org.tensorflow.proto.framework.KernelList DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.KernelList();
-  }
-
-  public static org.tensorflow.proto.framework.KernelList getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<KernelList>
-      PARSER = new com.google.protobuf.AbstractParser<KernelList>() {
-    @java.lang.Override
-    public KernelList parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new KernelList(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<KernelList> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<KernelList> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.framework.KernelList getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/KernelListOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/KernelListOrBuilder.java
deleted file mode 100644
index 31c95b5de9d..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/KernelListOrBuilder.java
+++ /dev/null
@@ -1,33 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/framework/kernel_def.proto
-
-package org.tensorflow.proto.framework;
-
-public interface KernelListOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.KernelList)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <code>repeated .tensorflow.KernelDef kernel = 1;</code>
-   */
-  java.util.List<org.tensorflow.proto.framework.KernelDef> 
-      getKernelList();
-  /**
-   * <code>repeated .tensorflow.KernelDef kernel = 1;</code>
-   */
-  org.tensorflow.proto.framework.KernelDef getKernel(int index);
-  /**
-   * <code>repeated .tensorflow.KernelDef kernel = 1;</code>
-   */
-  int getKernelCount();
-  /**
-   * <code>repeated .tensorflow.KernelDef kernel = 1;</code>
-   */
-  java.util.List<? extends org.tensorflow.proto.framework.KernelDefOrBuilder> 
-      getKernelOrBuilderList();
-  /**
-   * <code>repeated .tensorflow.KernelDef kernel = 1;</code>
-   */
-  org.tensorflow.proto.framework.KernelDefOrBuilder getKernelOrBuilder(
-      int index);
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ListValue.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ListValue.java
deleted file mode 100644
index 56101ea237c..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ListValue.java
+++ /dev/null
@@ -1,773 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/struct.proto
-
-package org.tensorflow.proto.framework;
-
-/**
- * <pre>
- * Represents a Python list.
- * </pre>
- *
- * Protobuf type {@code tensorflow.ListValue}
- */
-public  final class ListValue extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.ListValue)
-    ListValueOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use ListValue.newBuilder() to construct.
-  private ListValue(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private ListValue() {
-    values_ = java.util.Collections.emptyList();
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new ListValue();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private ListValue(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    int mutable_bitField0_ = 0;
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 10: {
-            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
-              values_ = new java.util.ArrayList<org.tensorflow.proto.framework.StructuredValue>();
-              mutable_bitField0_ |= 0x00000001;
-            }
-            values_.add(
-                input.readMessage(org.tensorflow.proto.framework.StructuredValue.parser(), extensionRegistry));
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      if (((mutable_bitField0_ & 0x00000001) != 0)) {
-        values_ = java.util.Collections.unmodifiableList(values_);
-      }
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.framework.StructProtos.internal_static_tensorflow_ListValue_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.framework.StructProtos.internal_static_tensorflow_ListValue_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.framework.ListValue.class, org.tensorflow.proto.framework.ListValue.Builder.class);
-  }
-
-  public static final int VALUES_FIELD_NUMBER = 1;
-  private java.util.List<org.tensorflow.proto.framework.StructuredValue> values_;
-  /**
-   * <code>repeated .tensorflow.StructuredValue values = 1;</code>
-   */
-  public java.util.List<org.tensorflow.proto.framework.StructuredValue> getValuesList() {
-    return values_;
-  }
-  /**
-   * <code>repeated .tensorflow.StructuredValue values = 1;</code>
-   */
-  public java.util.List<? extends org.tensorflow.proto.framework.StructuredValueOrBuilder> 
-      getValuesOrBuilderList() {
-    return values_;
-  }
-  /**
-   * <code>repeated .tensorflow.StructuredValue values = 1;</code>
-   */
-  public int getValuesCount() {
-    return values_.size();
-  }
-  /**
-   * <code>repeated .tensorflow.StructuredValue values = 1;</code>
-   */
-  public org.tensorflow.proto.framework.StructuredValue getValues(int index) {
-    return values_.get(index);
-  }
-  /**
-   * <code>repeated .tensorflow.StructuredValue values = 1;</code>
-   */
-  public org.tensorflow.proto.framework.StructuredValueOrBuilder getValuesOrBuilder(
-      int index) {
-    return values_.get(index);
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    for (int i = 0; i < values_.size(); i++) {
-      output.writeMessage(1, values_.get(i));
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    for (int i = 0; i < values_.size(); i++) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(1, values_.get(i));
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.framework.ListValue)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.framework.ListValue other = (org.tensorflow.proto.framework.ListValue) obj;
-
-    if (!getValuesList()
-        .equals(other.getValuesList())) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    if (getValuesCount() > 0) {
-      hash = (37 * hash) + VALUES_FIELD_NUMBER;
-      hash = (53 * hash) + getValuesList().hashCode();
-    }
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.framework.ListValue parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.ListValue parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.ListValue parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.ListValue parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.ListValue parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.ListValue parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.ListValue parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.ListValue parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.ListValue parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.ListValue parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.ListValue parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.ListValue parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.framework.ListValue prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * <pre>
-   * Represents a Python list.
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.ListValue}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.ListValue)
-      org.tensorflow.proto.framework.ListValueOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.StructProtos.internal_static_tensorflow_ListValue_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.StructProtos.internal_static_tensorflow_ListValue_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.ListValue.class, org.tensorflow.proto.framework.ListValue.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.framework.ListValue.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-        getValuesFieldBuilder();
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      if (valuesBuilder_ == null) {
-        values_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000001);
-      } else {
-        valuesBuilder_.clear();
-      }
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.framework.StructProtos.internal_static_tensorflow_ListValue_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.ListValue getDefaultInstanceForType() {
-      return org.tensorflow.proto.framework.ListValue.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.ListValue build() {
-      org.tensorflow.proto.framework.ListValue result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.ListValue buildPartial() {
-      org.tensorflow.proto.framework.ListValue result = new org.tensorflow.proto.framework.ListValue(this);
-      int from_bitField0_ = bitField0_;
-      if (valuesBuilder_ == null) {
-        if (((bitField0_ & 0x00000001) != 0)) {
-          values_ = java.util.Collections.unmodifiableList(values_);
-          bitField0_ = (bitField0_ & ~0x00000001);
-        }
-        result.values_ = values_;
-      } else {
-        result.values_ = valuesBuilder_.build();
-      }
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.framework.ListValue) {
-        return mergeFrom((org.tensorflow.proto.framework.ListValue)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.framework.ListValue other) {
-      if (other == org.tensorflow.proto.framework.ListValue.getDefaultInstance()) return this;
-      if (valuesBuilder_ == null) {
-        if (!other.values_.isEmpty()) {
-          if (values_.isEmpty()) {
-            values_ = other.values_;
-            bitField0_ = (bitField0_ & ~0x00000001);
-          } else {
-            ensureValuesIsMutable();
-            values_.addAll(other.values_);
-          }
-          onChanged();
-        }
-      } else {
-        if (!other.values_.isEmpty()) {
-          if (valuesBuilder_.isEmpty()) {
-            valuesBuilder_.dispose();
-            valuesBuilder_ = null;
-            values_ = other.values_;
-            bitField0_ = (bitField0_ & ~0x00000001);
-            valuesBuilder_ = 
-              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
-                 getValuesFieldBuilder() : null;
-          } else {
-            valuesBuilder_.addAllMessages(other.values_);
-          }
-        }
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.framework.ListValue parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.framework.ListValue) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-    private int bitField0_;
-
-    private java.util.List<org.tensorflow.proto.framework.StructuredValue> values_ =
-      java.util.Collections.emptyList();
-    private void ensureValuesIsMutable() {
-      if (!((bitField0_ & 0x00000001) != 0)) {
-        values_ = new java.util.ArrayList<org.tensorflow.proto.framework.StructuredValue>(values_);
-        bitField0_ |= 0x00000001;
-       }
-    }
-
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.framework.StructuredValue, org.tensorflow.proto.framework.StructuredValue.Builder, org.tensorflow.proto.framework.StructuredValueOrBuilder> valuesBuilder_;
-
-    /**
-     * <code>repeated .tensorflow.StructuredValue values = 1;</code>
-     */
-    public java.util.List<org.tensorflow.proto.framework.StructuredValue> getValuesList() {
-      if (valuesBuilder_ == null) {
-        return java.util.Collections.unmodifiableList(values_);
-      } else {
-        return valuesBuilder_.getMessageList();
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.StructuredValue values = 1;</code>
-     */
-    public int getValuesCount() {
-      if (valuesBuilder_ == null) {
-        return values_.size();
-      } else {
-        return valuesBuilder_.getCount();
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.StructuredValue values = 1;</code>
-     */
-    public org.tensorflow.proto.framework.StructuredValue getValues(int index) {
-      if (valuesBuilder_ == null) {
-        return values_.get(index);
-      } else {
-        return valuesBuilder_.getMessage(index);
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.StructuredValue values = 1;</code>
-     */
-    public Builder setValues(
-        int index, org.tensorflow.proto.framework.StructuredValue value) {
-      if (valuesBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureValuesIsMutable();
-        values_.set(index, value);
-        onChanged();
-      } else {
-        valuesBuilder_.setMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.StructuredValue values = 1;</code>
-     */
-    public Builder setValues(
-        int index, org.tensorflow.proto.framework.StructuredValue.Builder builderForValue) {
-      if (valuesBuilder_ == null) {
-        ensureValuesIsMutable();
-        values_.set(index, builderForValue.build());
-        onChanged();
-      } else {
-        valuesBuilder_.setMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.StructuredValue values = 1;</code>
-     */
-    public Builder addValues(org.tensorflow.proto.framework.StructuredValue value) {
-      if (valuesBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureValuesIsMutable();
-        values_.add(value);
-        onChanged();
-      } else {
-        valuesBuilder_.addMessage(value);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.StructuredValue values = 1;</code>
-     */
-    public Builder addValues(
-        int index, org.tensorflow.proto.framework.StructuredValue value) {
-      if (valuesBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureValuesIsMutable();
-        values_.add(index, value);
-        onChanged();
-      } else {
-        valuesBuilder_.addMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.StructuredValue values = 1;</code>
-     */
-    public Builder addValues(
-        org.tensorflow.proto.framework.StructuredValue.Builder builderForValue) {
-      if (valuesBuilder_ == null) {
-        ensureValuesIsMutable();
-        values_.add(builderForValue.build());
-        onChanged();
-      } else {
-        valuesBuilder_.addMessage(builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.StructuredValue values = 1;</code>
-     */
-    public Builder addValues(
-        int index, org.tensorflow.proto.framework.StructuredValue.Builder builderForValue) {
-      if (valuesBuilder_ == null) {
-        ensureValuesIsMutable();
-        values_.add(index, builderForValue.build());
-        onChanged();
-      } else {
-        valuesBuilder_.addMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.StructuredValue values = 1;</code>
-     */
-    public Builder addAllValues(
-        java.lang.Iterable<? extends org.tensorflow.proto.framework.StructuredValue> values) {
-      if (valuesBuilder_ == null) {
-        ensureValuesIsMutable();
-        com.google.protobuf.AbstractMessageLite.Builder.addAll(
-            values, values_);
-        onChanged();
-      } else {
-        valuesBuilder_.addAllMessages(values);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.StructuredValue values = 1;</code>
-     */
-    public Builder clearValues() {
-      if (valuesBuilder_ == null) {
-        values_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000001);
-        onChanged();
-      } else {
-        valuesBuilder_.clear();
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.StructuredValue values = 1;</code>
-     */
-    public Builder removeValues(int index) {
-      if (valuesBuilder_ == null) {
-        ensureValuesIsMutable();
-        values_.remove(index);
-        onChanged();
-      } else {
-        valuesBuilder_.remove(index);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.StructuredValue values = 1;</code>
-     */
-    public org.tensorflow.proto.framework.StructuredValue.Builder getValuesBuilder(
-        int index) {
-      return getValuesFieldBuilder().getBuilder(index);
-    }
-    /**
-     * <code>repeated .tensorflow.StructuredValue values = 1;</code>
-     */
-    public org.tensorflow.proto.framework.StructuredValueOrBuilder getValuesOrBuilder(
-        int index) {
-      if (valuesBuilder_ == null) {
-        return values_.get(index);  } else {
-        return valuesBuilder_.getMessageOrBuilder(index);
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.StructuredValue values = 1;</code>
-     */
-    public java.util.List<? extends org.tensorflow.proto.framework.StructuredValueOrBuilder> 
-         getValuesOrBuilderList() {
-      if (valuesBuilder_ != null) {
-        return valuesBuilder_.getMessageOrBuilderList();
-      } else {
-        return java.util.Collections.unmodifiableList(values_);
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.StructuredValue values = 1;</code>
-     */
-    public org.tensorflow.proto.framework.StructuredValue.Builder addValuesBuilder() {
-      return getValuesFieldBuilder().addBuilder(
-          org.tensorflow.proto.framework.StructuredValue.getDefaultInstance());
-    }
-    /**
-     * <code>repeated .tensorflow.StructuredValue values = 1;</code>
-     */
-    public org.tensorflow.proto.framework.StructuredValue.Builder addValuesBuilder(
-        int index) {
-      return getValuesFieldBuilder().addBuilder(
-          index, org.tensorflow.proto.framework.StructuredValue.getDefaultInstance());
-    }
-    /**
-     * <code>repeated .tensorflow.StructuredValue values = 1;</code>
-     */
-    public java.util.List<org.tensorflow.proto.framework.StructuredValue.Builder> 
-         getValuesBuilderList() {
-      return getValuesFieldBuilder().getBuilderList();
-    }
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.framework.StructuredValue, org.tensorflow.proto.framework.StructuredValue.Builder, org.tensorflow.proto.framework.StructuredValueOrBuilder> 
-        getValuesFieldBuilder() {
-      if (valuesBuilder_ == null) {
-        valuesBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
-            org.tensorflow.proto.framework.StructuredValue, org.tensorflow.proto.framework.StructuredValue.Builder, org.tensorflow.proto.framework.StructuredValueOrBuilder>(
-                values_,
-                ((bitField0_ & 0x00000001) != 0),
-                getParentForChildren(),
-                isClean());
-        values_ = null;
-      }
-      return valuesBuilder_;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.ListValue)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.ListValue)
-  private static final org.tensorflow.proto.framework.ListValue DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.ListValue();
-  }
-
-  public static org.tensorflow.proto.framework.ListValue getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<ListValue>
-      PARSER = new com.google.protobuf.AbstractParser<ListValue>() {
-    @java.lang.Override
-    public ListValue parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new ListValue(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<ListValue> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<ListValue> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.framework.ListValue getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ListValueOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ListValueOrBuilder.java
deleted file mode 100644
index e4b2d7076d9..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ListValueOrBuilder.java
+++ /dev/null
@@ -1,33 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/struct.proto
-
-package org.tensorflow.proto.framework;
-
-public interface ListValueOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.ListValue)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <code>repeated .tensorflow.StructuredValue values = 1;</code>
-   */
-  java.util.List<org.tensorflow.proto.framework.StructuredValue> 
-      getValuesList();
-  /**
-   * <code>repeated .tensorflow.StructuredValue values = 1;</code>
-   */
-  org.tensorflow.proto.framework.StructuredValue getValues(int index);
-  /**
-   * <code>repeated .tensorflow.StructuredValue values = 1;</code>
-   */
-  int getValuesCount();
-  /**
-   * <code>repeated .tensorflow.StructuredValue values = 1;</code>
-   */
-  java.util.List<? extends org.tensorflow.proto.framework.StructuredValueOrBuilder> 
-      getValuesOrBuilderList();
-  /**
-   * <code>repeated .tensorflow.StructuredValue values = 1;</code>
-   */
-  org.tensorflow.proto.framework.StructuredValueOrBuilder getValuesOrBuilder(
-      int index);
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/LocalLinks.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/LocalLinks.java
deleted file mode 100644
index 2219a69600d..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/LocalLinks.java
+++ /dev/null
@@ -1,765 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/framework/device_attributes.proto
-
-package org.tensorflow.proto.framework;
-
-/**
- * Protobuf type {@code tensorflow.LocalLinks}
- */
-public  final class LocalLinks extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.LocalLinks)
-    LocalLinksOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use LocalLinks.newBuilder() to construct.
-  private LocalLinks(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private LocalLinks() {
-    link_ = java.util.Collections.emptyList();
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new LocalLinks();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private LocalLinks(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    int mutable_bitField0_ = 0;
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 10: {
-            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
-              link_ = new java.util.ArrayList<org.tensorflow.proto.framework.InterconnectLink>();
-              mutable_bitField0_ |= 0x00000001;
-            }
-            link_.add(
-                input.readMessage(org.tensorflow.proto.framework.InterconnectLink.parser(), extensionRegistry));
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      if (((mutable_bitField0_ & 0x00000001) != 0)) {
-        link_ = java.util.Collections.unmodifiableList(link_);
-      }
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.framework.DeviceAttributesProtos.internal_static_tensorflow_LocalLinks_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.framework.DeviceAttributesProtos.internal_static_tensorflow_LocalLinks_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.framework.LocalLinks.class, org.tensorflow.proto.framework.LocalLinks.Builder.class);
-  }
-
-  public static final int LINK_FIELD_NUMBER = 1;
-  private java.util.List<org.tensorflow.proto.framework.InterconnectLink> link_;
-  /**
-   * <code>repeated .tensorflow.InterconnectLink link = 1;</code>
-   */
-  public java.util.List<org.tensorflow.proto.framework.InterconnectLink> getLinkList() {
-    return link_;
-  }
-  /**
-   * <code>repeated .tensorflow.InterconnectLink link = 1;</code>
-   */
-  public java.util.List<? extends org.tensorflow.proto.framework.InterconnectLinkOrBuilder> 
-      getLinkOrBuilderList() {
-    return link_;
-  }
-  /**
-   * <code>repeated .tensorflow.InterconnectLink link = 1;</code>
-   */
-  public int getLinkCount() {
-    return link_.size();
-  }
-  /**
-   * <code>repeated .tensorflow.InterconnectLink link = 1;</code>
-   */
-  public org.tensorflow.proto.framework.InterconnectLink getLink(int index) {
-    return link_.get(index);
-  }
-  /**
-   * <code>repeated .tensorflow.InterconnectLink link = 1;</code>
-   */
-  public org.tensorflow.proto.framework.InterconnectLinkOrBuilder getLinkOrBuilder(
-      int index) {
-    return link_.get(index);
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    for (int i = 0; i < link_.size(); i++) {
-      output.writeMessage(1, link_.get(i));
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    for (int i = 0; i < link_.size(); i++) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(1, link_.get(i));
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.framework.LocalLinks)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.framework.LocalLinks other = (org.tensorflow.proto.framework.LocalLinks) obj;
-
-    if (!getLinkList()
-        .equals(other.getLinkList())) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    if (getLinkCount() > 0) {
-      hash = (37 * hash) + LINK_FIELD_NUMBER;
-      hash = (53 * hash) + getLinkList().hashCode();
-    }
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.framework.LocalLinks parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.LocalLinks parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.LocalLinks parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.LocalLinks parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.LocalLinks parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.LocalLinks parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.LocalLinks parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.LocalLinks parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.LocalLinks parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.LocalLinks parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.LocalLinks parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.LocalLinks parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.framework.LocalLinks prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * Protobuf type {@code tensorflow.LocalLinks}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.LocalLinks)
-      org.tensorflow.proto.framework.LocalLinksOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.DeviceAttributesProtos.internal_static_tensorflow_LocalLinks_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.DeviceAttributesProtos.internal_static_tensorflow_LocalLinks_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.LocalLinks.class, org.tensorflow.proto.framework.LocalLinks.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.framework.LocalLinks.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-        getLinkFieldBuilder();
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      if (linkBuilder_ == null) {
-        link_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000001);
-      } else {
-        linkBuilder_.clear();
-      }
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.framework.DeviceAttributesProtos.internal_static_tensorflow_LocalLinks_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.LocalLinks getDefaultInstanceForType() {
-      return org.tensorflow.proto.framework.LocalLinks.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.LocalLinks build() {
-      org.tensorflow.proto.framework.LocalLinks result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.LocalLinks buildPartial() {
-      org.tensorflow.proto.framework.LocalLinks result = new org.tensorflow.proto.framework.LocalLinks(this);
-      int from_bitField0_ = bitField0_;
-      if (linkBuilder_ == null) {
-        if (((bitField0_ & 0x00000001) != 0)) {
-          link_ = java.util.Collections.unmodifiableList(link_);
-          bitField0_ = (bitField0_ & ~0x00000001);
-        }
-        result.link_ = link_;
-      } else {
-        result.link_ = linkBuilder_.build();
-      }
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.framework.LocalLinks) {
-        return mergeFrom((org.tensorflow.proto.framework.LocalLinks)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.framework.LocalLinks other) {
-      if (other == org.tensorflow.proto.framework.LocalLinks.getDefaultInstance()) return this;
-      if (linkBuilder_ == null) {
-        if (!other.link_.isEmpty()) {
-          if (link_.isEmpty()) {
-            link_ = other.link_;
-            bitField0_ = (bitField0_ & ~0x00000001);
-          } else {
-            ensureLinkIsMutable();
-            link_.addAll(other.link_);
-          }
-          onChanged();
-        }
-      } else {
-        if (!other.link_.isEmpty()) {
-          if (linkBuilder_.isEmpty()) {
-            linkBuilder_.dispose();
-            linkBuilder_ = null;
-            link_ = other.link_;
-            bitField0_ = (bitField0_ & ~0x00000001);
-            linkBuilder_ = 
-              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
-                 getLinkFieldBuilder() : null;
-          } else {
-            linkBuilder_.addAllMessages(other.link_);
-          }
-        }
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.framework.LocalLinks parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.framework.LocalLinks) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-    private int bitField0_;
-
-    private java.util.List<org.tensorflow.proto.framework.InterconnectLink> link_ =
-      java.util.Collections.emptyList();
-    private void ensureLinkIsMutable() {
-      if (!((bitField0_ & 0x00000001) != 0)) {
-        link_ = new java.util.ArrayList<org.tensorflow.proto.framework.InterconnectLink>(link_);
-        bitField0_ |= 0x00000001;
-       }
-    }
-
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.framework.InterconnectLink, org.tensorflow.proto.framework.InterconnectLink.Builder, org.tensorflow.proto.framework.InterconnectLinkOrBuilder> linkBuilder_;
-
-    /**
-     * <code>repeated .tensorflow.InterconnectLink link = 1;</code>
-     */
-    public java.util.List<org.tensorflow.proto.framework.InterconnectLink> getLinkList() {
-      if (linkBuilder_ == null) {
-        return java.util.Collections.unmodifiableList(link_);
-      } else {
-        return linkBuilder_.getMessageList();
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.InterconnectLink link = 1;</code>
-     */
-    public int getLinkCount() {
-      if (linkBuilder_ == null) {
-        return link_.size();
-      } else {
-        return linkBuilder_.getCount();
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.InterconnectLink link = 1;</code>
-     */
-    public org.tensorflow.proto.framework.InterconnectLink getLink(int index) {
-      if (linkBuilder_ == null) {
-        return link_.get(index);
-      } else {
-        return linkBuilder_.getMessage(index);
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.InterconnectLink link = 1;</code>
-     */
-    public Builder setLink(
-        int index, org.tensorflow.proto.framework.InterconnectLink value) {
-      if (linkBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureLinkIsMutable();
-        link_.set(index, value);
-        onChanged();
-      } else {
-        linkBuilder_.setMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.InterconnectLink link = 1;</code>
-     */
-    public Builder setLink(
-        int index, org.tensorflow.proto.framework.InterconnectLink.Builder builderForValue) {
-      if (linkBuilder_ == null) {
-        ensureLinkIsMutable();
-        link_.set(index, builderForValue.build());
-        onChanged();
-      } else {
-        linkBuilder_.setMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.InterconnectLink link = 1;</code>
-     */
-    public Builder addLink(org.tensorflow.proto.framework.InterconnectLink value) {
-      if (linkBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureLinkIsMutable();
-        link_.add(value);
-        onChanged();
-      } else {
-        linkBuilder_.addMessage(value);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.InterconnectLink link = 1;</code>
-     */
-    public Builder addLink(
-        int index, org.tensorflow.proto.framework.InterconnectLink value) {
-      if (linkBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureLinkIsMutable();
-        link_.add(index, value);
-        onChanged();
-      } else {
-        linkBuilder_.addMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.InterconnectLink link = 1;</code>
-     */
-    public Builder addLink(
-        org.tensorflow.proto.framework.InterconnectLink.Builder builderForValue) {
-      if (linkBuilder_ == null) {
-        ensureLinkIsMutable();
-        link_.add(builderForValue.build());
-        onChanged();
-      } else {
-        linkBuilder_.addMessage(builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.InterconnectLink link = 1;</code>
-     */
-    public Builder addLink(
-        int index, org.tensorflow.proto.framework.InterconnectLink.Builder builderForValue) {
-      if (linkBuilder_ == null) {
-        ensureLinkIsMutable();
-        link_.add(index, builderForValue.build());
-        onChanged();
-      } else {
-        linkBuilder_.addMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.InterconnectLink link = 1;</code>
-     */
-    public Builder addAllLink(
-        java.lang.Iterable<? extends org.tensorflow.proto.framework.InterconnectLink> values) {
-      if (linkBuilder_ == null) {
-        ensureLinkIsMutable();
-        com.google.protobuf.AbstractMessageLite.Builder.addAll(
-            values, link_);
-        onChanged();
-      } else {
-        linkBuilder_.addAllMessages(values);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.InterconnectLink link = 1;</code>
-     */
-    public Builder clearLink() {
-      if (linkBuilder_ == null) {
-        link_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000001);
-        onChanged();
-      } else {
-        linkBuilder_.clear();
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.InterconnectLink link = 1;</code>
-     */
-    public Builder removeLink(int index) {
-      if (linkBuilder_ == null) {
-        ensureLinkIsMutable();
-        link_.remove(index);
-        onChanged();
-      } else {
-        linkBuilder_.remove(index);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.InterconnectLink link = 1;</code>
-     */
-    public org.tensorflow.proto.framework.InterconnectLink.Builder getLinkBuilder(
-        int index) {
-      return getLinkFieldBuilder().getBuilder(index);
-    }
-    /**
-     * <code>repeated .tensorflow.InterconnectLink link = 1;</code>
-     */
-    public org.tensorflow.proto.framework.InterconnectLinkOrBuilder getLinkOrBuilder(
-        int index) {
-      if (linkBuilder_ == null) {
-        return link_.get(index);  } else {
-        return linkBuilder_.getMessageOrBuilder(index);
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.InterconnectLink link = 1;</code>
-     */
-    public java.util.List<? extends org.tensorflow.proto.framework.InterconnectLinkOrBuilder> 
-         getLinkOrBuilderList() {
-      if (linkBuilder_ != null) {
-        return linkBuilder_.getMessageOrBuilderList();
-      } else {
-        return java.util.Collections.unmodifiableList(link_);
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.InterconnectLink link = 1;</code>
-     */
-    public org.tensorflow.proto.framework.InterconnectLink.Builder addLinkBuilder() {
-      return getLinkFieldBuilder().addBuilder(
-          org.tensorflow.proto.framework.InterconnectLink.getDefaultInstance());
-    }
-    /**
-     * <code>repeated .tensorflow.InterconnectLink link = 1;</code>
-     */
-    public org.tensorflow.proto.framework.InterconnectLink.Builder addLinkBuilder(
-        int index) {
-      return getLinkFieldBuilder().addBuilder(
-          index, org.tensorflow.proto.framework.InterconnectLink.getDefaultInstance());
-    }
-    /**
-     * <code>repeated .tensorflow.InterconnectLink link = 1;</code>
-     */
-    public java.util.List<org.tensorflow.proto.framework.InterconnectLink.Builder> 
-         getLinkBuilderList() {
-      return getLinkFieldBuilder().getBuilderList();
-    }
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.framework.InterconnectLink, org.tensorflow.proto.framework.InterconnectLink.Builder, org.tensorflow.proto.framework.InterconnectLinkOrBuilder> 
-        getLinkFieldBuilder() {
-      if (linkBuilder_ == null) {
-        linkBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
-            org.tensorflow.proto.framework.InterconnectLink, org.tensorflow.proto.framework.InterconnectLink.Builder, org.tensorflow.proto.framework.InterconnectLinkOrBuilder>(
-                link_,
-                ((bitField0_ & 0x00000001) != 0),
-                getParentForChildren(),
-                isClean());
-        link_ = null;
-      }
-      return linkBuilder_;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.LocalLinks)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.LocalLinks)
-  private static final org.tensorflow.proto.framework.LocalLinks DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.LocalLinks();
-  }
-
-  public static org.tensorflow.proto.framework.LocalLinks getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<LocalLinks>
-      PARSER = new com.google.protobuf.AbstractParser<LocalLinks>() {
-    @java.lang.Override
-    public LocalLinks parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new LocalLinks(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<LocalLinks> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<LocalLinks> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.framework.LocalLinks getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/LocalLinksOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/LocalLinksOrBuilder.java
deleted file mode 100644
index e1cd5854d3b..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/LocalLinksOrBuilder.java
+++ /dev/null
@@ -1,33 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/framework/device_attributes.proto
-
-package org.tensorflow.proto.framework;
-
-public interface LocalLinksOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.LocalLinks)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <code>repeated .tensorflow.InterconnectLink link = 1;</code>
-   */
-  java.util.List<org.tensorflow.proto.framework.InterconnectLink> 
-      getLinkList();
-  /**
-   * <code>repeated .tensorflow.InterconnectLink link = 1;</code>
-   */
-  org.tensorflow.proto.framework.InterconnectLink getLink(int index);
-  /**
-   * <code>repeated .tensorflow.InterconnectLink link = 1;</code>
-   */
-  int getLinkCount();
-  /**
-   * <code>repeated .tensorflow.InterconnectLink link = 1;</code>
-   */
-  java.util.List<? extends org.tensorflow.proto.framework.InterconnectLinkOrBuilder> 
-      getLinkOrBuilderList();
-  /**
-   * <code>repeated .tensorflow.InterconnectLink link = 1;</code>
-   */
-  org.tensorflow.proto.framework.InterconnectLinkOrBuilder getLinkOrBuilder(
-      int index);
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/LogMemoryProtos.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/LogMemoryProtos.java
deleted file mode 100644
index 4cef19ad6fe..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/LogMemoryProtos.java
+++ /dev/null
@@ -1,125 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/framework/log_memory.proto
-
-package org.tensorflow.proto.framework;
-
-public final class LogMemoryProtos {
-  private LogMemoryProtos() {}
-  public static void registerAllExtensions(
-      com.google.protobuf.ExtensionRegistryLite registry) {
-  }
-
-  public static void registerAllExtensions(
-      com.google.protobuf.ExtensionRegistry registry) {
-    registerAllExtensions(
-        (com.google.protobuf.ExtensionRegistryLite) registry);
-  }
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_MemoryLogStep_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_MemoryLogStep_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_MemoryLogTensorAllocation_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_MemoryLogTensorAllocation_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_MemoryLogTensorDeallocation_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_MemoryLogTensorDeallocation_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_MemoryLogTensorOutput_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_MemoryLogTensorOutput_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_MemoryLogRawAllocation_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_MemoryLogRawAllocation_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_MemoryLogRawDeallocation_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_MemoryLogRawDeallocation_fieldAccessorTable;
-
-  public static com.google.protobuf.Descriptors.FileDescriptor
-      getDescriptor() {
-    return descriptor;
-  }
-  private static  com.google.protobuf.Descriptors.FileDescriptor
-      descriptor;
-  static {
-    java.lang.String[] descriptorData = {
-      "\n*tensorflow/core/framework/log_memory.p" +
-      "roto\022\ntensorflow\0322tensorflow/core/framew" +
-      "ork/tensor_description.proto\"0\n\rMemoryLo" +
-      "gStep\022\017\n\007step_id\030\001 \001(\003\022\016\n\006handle\030\002 \001(\t\"p" +
-      "\n\031MemoryLogTensorAllocation\022\017\n\007step_id\030\001" +
-      " \001(\003\022\023\n\013kernel_name\030\002 \001(\t\022-\n\006tensor\030\003 \001(" +
-      "\0132\035.tensorflow.TensorDescription\"L\n\033Memo" +
-      "ryLogTensorDeallocation\022\025\n\rallocation_id" +
-      "\030\001 \001(\003\022\026\n\016allocator_name\030\002 \001(\t\"{\n\025Memory" +
-      "LogTensorOutput\022\017\n\007step_id\030\001 \001(\003\022\023\n\013kern" +
-      "el_name\030\002 \001(\t\022\r\n\005index\030\003 \001(\005\022-\n\006tensor\030\004" +
-      " \001(\0132\035.tensorflow.TensorDescription\"\213\001\n\026" +
-      "MemoryLogRawAllocation\022\017\n\007step_id\030\001 \001(\003\022" +
-      "\021\n\toperation\030\002 \001(\t\022\021\n\tnum_bytes\030\003 \001(\003\022\013\n" +
-      "\003ptr\030\004 \001(\004\022\025\n\rallocation_id\030\005 \001(\003\022\026\n\016all" +
-      "ocator_name\030\006 \001(\t\"\177\n\030MemoryLogRawDealloc" +
-      "ation\022\017\n\007step_id\030\001 \001(\003\022\021\n\toperation\030\002 \001(" +
-      "\t\022\025\n\rallocation_id\030\003 \001(\003\022\026\n\016allocator_na" +
-      "me\030\004 \001(\t\022\020\n\010deferred\030\005 \001(\010B\211\001\n\036org.tenso" +
-      "rflow.proto.frameworkB\017LogMemoryProtosP\001" +
-      "ZQgithub.com/tensorflow/tensorflow/tenso" +
-      "rflow/go/core/framework/log_memory_go_pr" +
-      "oto\370\001\001b\006proto3"
-    };
-    descriptor = com.google.protobuf.Descriptors.FileDescriptor
-      .internalBuildGeneratedFileFrom(descriptorData,
-        new com.google.protobuf.Descriptors.FileDescriptor[] {
-          org.tensorflow.proto.framework.TensorDescriptionProtos.getDescriptor(),
-        });
-    internal_static_tensorflow_MemoryLogStep_descriptor =
-      getDescriptor().getMessageTypes().get(0);
-    internal_static_tensorflow_MemoryLogStep_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_MemoryLogStep_descriptor,
-        new java.lang.String[] { "StepId", "Handle", });
-    internal_static_tensorflow_MemoryLogTensorAllocation_descriptor =
-      getDescriptor().getMessageTypes().get(1);
-    internal_static_tensorflow_MemoryLogTensorAllocation_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_MemoryLogTensorAllocation_descriptor,
-        new java.lang.String[] { "StepId", "KernelName", "Tensor", });
-    internal_static_tensorflow_MemoryLogTensorDeallocation_descriptor =
-      getDescriptor().getMessageTypes().get(2);
-    internal_static_tensorflow_MemoryLogTensorDeallocation_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_MemoryLogTensorDeallocation_descriptor,
-        new java.lang.String[] { "AllocationId", "AllocatorName", });
-    internal_static_tensorflow_MemoryLogTensorOutput_descriptor =
-      getDescriptor().getMessageTypes().get(3);
-    internal_static_tensorflow_MemoryLogTensorOutput_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_MemoryLogTensorOutput_descriptor,
-        new java.lang.String[] { "StepId", "KernelName", "Index", "Tensor", });
-    internal_static_tensorflow_MemoryLogRawAllocation_descriptor =
-      getDescriptor().getMessageTypes().get(4);
-    internal_static_tensorflow_MemoryLogRawAllocation_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_MemoryLogRawAllocation_descriptor,
-        new java.lang.String[] { "StepId", "Operation", "NumBytes", "Ptr", "AllocationId", "AllocatorName", });
-    internal_static_tensorflow_MemoryLogRawDeallocation_descriptor =
-      getDescriptor().getMessageTypes().get(5);
-    internal_static_tensorflow_MemoryLogRawDeallocation_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_MemoryLogRawDeallocation_descriptor,
-        new java.lang.String[] { "StepId", "Operation", "AllocationId", "AllocatorName", "Deferred", });
-    org.tensorflow.proto.framework.TensorDescriptionProtos.getDescriptor();
-  }
-
-  // @@protoc_insertion_point(outer_class_scope)
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/MemoryLogRawAllocation.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/MemoryLogRawAllocation.java
deleted file mode 100644
index db40403eba4..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/MemoryLogRawAllocation.java
+++ /dev/null
@@ -1,1029 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/framework/log_memory.proto
-
-package org.tensorflow.proto.framework;
-
-/**
- * Protobuf type {@code tensorflow.MemoryLogRawAllocation}
- */
-public  final class MemoryLogRawAllocation extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.MemoryLogRawAllocation)
-    MemoryLogRawAllocationOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use MemoryLogRawAllocation.newBuilder() to construct.
-  private MemoryLogRawAllocation(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private MemoryLogRawAllocation() {
-    operation_ = "";
-    allocatorName_ = "";
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new MemoryLogRawAllocation();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private MemoryLogRawAllocation(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 8: {
-
-            stepId_ = input.readInt64();
-            break;
-          }
-          case 18: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            operation_ = s;
-            break;
-          }
-          case 24: {
-
-            numBytes_ = input.readInt64();
-            break;
-          }
-          case 32: {
-
-            ptr_ = input.readUInt64();
-            break;
-          }
-          case 40: {
-
-            allocationId_ = input.readInt64();
-            break;
-          }
-          case 50: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            allocatorName_ = s;
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.framework.LogMemoryProtos.internal_static_tensorflow_MemoryLogRawAllocation_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.framework.LogMemoryProtos.internal_static_tensorflow_MemoryLogRawAllocation_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.framework.MemoryLogRawAllocation.class, org.tensorflow.proto.framework.MemoryLogRawAllocation.Builder.class);
-  }
-
-  public static final int STEP_ID_FIELD_NUMBER = 1;
-  private long stepId_;
-  /**
-   * <pre>
-   * Process-unique step id.
-   * </pre>
-   *
-   * <code>int64 step_id = 1;</code>
-   */
-  public long getStepId() {
-    return stepId_;
-  }
-
-  public static final int OPERATION_FIELD_NUMBER = 2;
-  private volatile java.lang.Object operation_;
-  /**
-   * <pre>
-   * Name of the operation making the allocation.
-   * </pre>
-   *
-   * <code>string operation = 2;</code>
-   */
-  public java.lang.String getOperation() {
-    java.lang.Object ref = operation_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      operation_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * Name of the operation making the allocation.
-   * </pre>
-   *
-   * <code>string operation = 2;</code>
-   */
-  public com.google.protobuf.ByteString
-      getOperationBytes() {
-    java.lang.Object ref = operation_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      operation_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int NUM_BYTES_FIELD_NUMBER = 3;
-  private long numBytes_;
-  /**
-   * <pre>
-   * Number of bytes in the allocation.
-   * </pre>
-   *
-   * <code>int64 num_bytes = 3;</code>
-   */
-  public long getNumBytes() {
-    return numBytes_;
-  }
-
-  public static final int PTR_FIELD_NUMBER = 4;
-  private long ptr_;
-  /**
-   * <pre>
-   * Address of the allocation.
-   * </pre>
-   *
-   * <code>uint64 ptr = 4;</code>
-   */
-  public long getPtr() {
-    return ptr_;
-  }
-
-  public static final int ALLOCATION_ID_FIELD_NUMBER = 5;
-  private long allocationId_;
-  /**
-   * <pre>
-   * Id of the tensor buffer being allocated, used to match to a
-   * corresponding deallocation.
-   * </pre>
-   *
-   * <code>int64 allocation_id = 5;</code>
-   */
-  public long getAllocationId() {
-    return allocationId_;
-  }
-
-  public static final int ALLOCATOR_NAME_FIELD_NUMBER = 6;
-  private volatile java.lang.Object allocatorName_;
-  /**
-   * <pre>
-   * Name of the allocator used.
-   * </pre>
-   *
-   * <code>string allocator_name = 6;</code>
-   */
-  public java.lang.String getAllocatorName() {
-    java.lang.Object ref = allocatorName_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      allocatorName_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * Name of the allocator used.
-   * </pre>
-   *
-   * <code>string allocator_name = 6;</code>
-   */
-  public com.google.protobuf.ByteString
-      getAllocatorNameBytes() {
-    java.lang.Object ref = allocatorName_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      allocatorName_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    if (stepId_ != 0L) {
-      output.writeInt64(1, stepId_);
-    }
-    if (!getOperationBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 2, operation_);
-    }
-    if (numBytes_ != 0L) {
-      output.writeInt64(3, numBytes_);
-    }
-    if (ptr_ != 0L) {
-      output.writeUInt64(4, ptr_);
-    }
-    if (allocationId_ != 0L) {
-      output.writeInt64(5, allocationId_);
-    }
-    if (!getAllocatorNameBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 6, allocatorName_);
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (stepId_ != 0L) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt64Size(1, stepId_);
-    }
-    if (!getOperationBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(2, operation_);
-    }
-    if (numBytes_ != 0L) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt64Size(3, numBytes_);
-    }
-    if (ptr_ != 0L) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeUInt64Size(4, ptr_);
-    }
-    if (allocationId_ != 0L) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt64Size(5, allocationId_);
-    }
-    if (!getAllocatorNameBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(6, allocatorName_);
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.framework.MemoryLogRawAllocation)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.framework.MemoryLogRawAllocation other = (org.tensorflow.proto.framework.MemoryLogRawAllocation) obj;
-
-    if (getStepId()
-        != other.getStepId()) return false;
-    if (!getOperation()
-        .equals(other.getOperation())) return false;
-    if (getNumBytes()
-        != other.getNumBytes()) return false;
-    if (getPtr()
-        != other.getPtr()) return false;
-    if (getAllocationId()
-        != other.getAllocationId()) return false;
-    if (!getAllocatorName()
-        .equals(other.getAllocatorName())) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    hash = (37 * hash) + STEP_ID_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-        getStepId());
-    hash = (37 * hash) + OPERATION_FIELD_NUMBER;
-    hash = (53 * hash) + getOperation().hashCode();
-    hash = (37 * hash) + NUM_BYTES_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-        getNumBytes());
-    hash = (37 * hash) + PTR_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-        getPtr());
-    hash = (37 * hash) + ALLOCATION_ID_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-        getAllocationId());
-    hash = (37 * hash) + ALLOCATOR_NAME_FIELD_NUMBER;
-    hash = (53 * hash) + getAllocatorName().hashCode();
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.framework.MemoryLogRawAllocation parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.MemoryLogRawAllocation parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.MemoryLogRawAllocation parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.MemoryLogRawAllocation parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.MemoryLogRawAllocation parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.MemoryLogRawAllocation parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.MemoryLogRawAllocation parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.MemoryLogRawAllocation parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.MemoryLogRawAllocation parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.MemoryLogRawAllocation parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.MemoryLogRawAllocation parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.MemoryLogRawAllocation parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.framework.MemoryLogRawAllocation prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * Protobuf type {@code tensorflow.MemoryLogRawAllocation}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.MemoryLogRawAllocation)
-      org.tensorflow.proto.framework.MemoryLogRawAllocationOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.LogMemoryProtos.internal_static_tensorflow_MemoryLogRawAllocation_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.LogMemoryProtos.internal_static_tensorflow_MemoryLogRawAllocation_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.MemoryLogRawAllocation.class, org.tensorflow.proto.framework.MemoryLogRawAllocation.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.framework.MemoryLogRawAllocation.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      stepId_ = 0L;
-
-      operation_ = "";
-
-      numBytes_ = 0L;
-
-      ptr_ = 0L;
-
-      allocationId_ = 0L;
-
-      allocatorName_ = "";
-
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.framework.LogMemoryProtos.internal_static_tensorflow_MemoryLogRawAllocation_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.MemoryLogRawAllocation getDefaultInstanceForType() {
-      return org.tensorflow.proto.framework.MemoryLogRawAllocation.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.MemoryLogRawAllocation build() {
-      org.tensorflow.proto.framework.MemoryLogRawAllocation result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.MemoryLogRawAllocation buildPartial() {
-      org.tensorflow.proto.framework.MemoryLogRawAllocation result = new org.tensorflow.proto.framework.MemoryLogRawAllocation(this);
-      result.stepId_ = stepId_;
-      result.operation_ = operation_;
-      result.numBytes_ = numBytes_;
-      result.ptr_ = ptr_;
-      result.allocationId_ = allocationId_;
-      result.allocatorName_ = allocatorName_;
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.framework.MemoryLogRawAllocation) {
-        return mergeFrom((org.tensorflow.proto.framework.MemoryLogRawAllocation)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.framework.MemoryLogRawAllocation other) {
-      if (other == org.tensorflow.proto.framework.MemoryLogRawAllocation.getDefaultInstance()) return this;
-      if (other.getStepId() != 0L) {
-        setStepId(other.getStepId());
-      }
-      if (!other.getOperation().isEmpty()) {
-        operation_ = other.operation_;
-        onChanged();
-      }
-      if (other.getNumBytes() != 0L) {
-        setNumBytes(other.getNumBytes());
-      }
-      if (other.getPtr() != 0L) {
-        setPtr(other.getPtr());
-      }
-      if (other.getAllocationId() != 0L) {
-        setAllocationId(other.getAllocationId());
-      }
-      if (!other.getAllocatorName().isEmpty()) {
-        allocatorName_ = other.allocatorName_;
-        onChanged();
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.framework.MemoryLogRawAllocation parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.framework.MemoryLogRawAllocation) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-
-    private long stepId_ ;
-    /**
-     * <pre>
-     * Process-unique step id.
-     * </pre>
-     *
-     * <code>int64 step_id = 1;</code>
-     */
-    public long getStepId() {
-      return stepId_;
-    }
-    /**
-     * <pre>
-     * Process-unique step id.
-     * </pre>
-     *
-     * <code>int64 step_id = 1;</code>
-     */
-    public Builder setStepId(long value) {
-      
-      stepId_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Process-unique step id.
-     * </pre>
-     *
-     * <code>int64 step_id = 1;</code>
-     */
-    public Builder clearStepId() {
-      
-      stepId_ = 0L;
-      onChanged();
-      return this;
-    }
-
-    private java.lang.Object operation_ = "";
-    /**
-     * <pre>
-     * Name of the operation making the allocation.
-     * </pre>
-     *
-     * <code>string operation = 2;</code>
-     */
-    public java.lang.String getOperation() {
-      java.lang.Object ref = operation_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        operation_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Name of the operation making the allocation.
-     * </pre>
-     *
-     * <code>string operation = 2;</code>
-     */
-    public com.google.protobuf.ByteString
-        getOperationBytes() {
-      java.lang.Object ref = operation_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        operation_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Name of the operation making the allocation.
-     * </pre>
-     *
-     * <code>string operation = 2;</code>
-     */
-    public Builder setOperation(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      operation_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Name of the operation making the allocation.
-     * </pre>
-     *
-     * <code>string operation = 2;</code>
-     */
-    public Builder clearOperation() {
-      
-      operation_ = getDefaultInstance().getOperation();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Name of the operation making the allocation.
-     * </pre>
-     *
-     * <code>string operation = 2;</code>
-     */
-    public Builder setOperationBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      operation_ = value;
-      onChanged();
-      return this;
-    }
-
-    private long numBytes_ ;
-    /**
-     * <pre>
-     * Number of bytes in the allocation.
-     * </pre>
-     *
-     * <code>int64 num_bytes = 3;</code>
-     */
-    public long getNumBytes() {
-      return numBytes_;
-    }
-    /**
-     * <pre>
-     * Number of bytes in the allocation.
-     * </pre>
-     *
-     * <code>int64 num_bytes = 3;</code>
-     */
-    public Builder setNumBytes(long value) {
-      
-      numBytes_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Number of bytes in the allocation.
-     * </pre>
-     *
-     * <code>int64 num_bytes = 3;</code>
-     */
-    public Builder clearNumBytes() {
-      
-      numBytes_ = 0L;
-      onChanged();
-      return this;
-    }
-
-    private long ptr_ ;
-    /**
-     * <pre>
-     * Address of the allocation.
-     * </pre>
-     *
-     * <code>uint64 ptr = 4;</code>
-     */
-    public long getPtr() {
-      return ptr_;
-    }
-    /**
-     * <pre>
-     * Address of the allocation.
-     * </pre>
-     *
-     * <code>uint64 ptr = 4;</code>
-     */
-    public Builder setPtr(long value) {
-      
-      ptr_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Address of the allocation.
-     * </pre>
-     *
-     * <code>uint64 ptr = 4;</code>
-     */
-    public Builder clearPtr() {
-      
-      ptr_ = 0L;
-      onChanged();
-      return this;
-    }
-
-    private long allocationId_ ;
-    /**
-     * <pre>
-     * Id of the tensor buffer being allocated, used to match to a
-     * corresponding deallocation.
-     * </pre>
-     *
-     * <code>int64 allocation_id = 5;</code>
-     */
-    public long getAllocationId() {
-      return allocationId_;
-    }
-    /**
-     * <pre>
-     * Id of the tensor buffer being allocated, used to match to a
-     * corresponding deallocation.
-     * </pre>
-     *
-     * <code>int64 allocation_id = 5;</code>
-     */
-    public Builder setAllocationId(long value) {
-      
-      allocationId_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Id of the tensor buffer being allocated, used to match to a
-     * corresponding deallocation.
-     * </pre>
-     *
-     * <code>int64 allocation_id = 5;</code>
-     */
-    public Builder clearAllocationId() {
-      
-      allocationId_ = 0L;
-      onChanged();
-      return this;
-    }
-
-    private java.lang.Object allocatorName_ = "";
-    /**
-     * <pre>
-     * Name of the allocator used.
-     * </pre>
-     *
-     * <code>string allocator_name = 6;</code>
-     */
-    public java.lang.String getAllocatorName() {
-      java.lang.Object ref = allocatorName_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        allocatorName_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Name of the allocator used.
-     * </pre>
-     *
-     * <code>string allocator_name = 6;</code>
-     */
-    public com.google.protobuf.ByteString
-        getAllocatorNameBytes() {
-      java.lang.Object ref = allocatorName_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        allocatorName_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Name of the allocator used.
-     * </pre>
-     *
-     * <code>string allocator_name = 6;</code>
-     */
-    public Builder setAllocatorName(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      allocatorName_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Name of the allocator used.
-     * </pre>
-     *
-     * <code>string allocator_name = 6;</code>
-     */
-    public Builder clearAllocatorName() {
-      
-      allocatorName_ = getDefaultInstance().getAllocatorName();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Name of the allocator used.
-     * </pre>
-     *
-     * <code>string allocator_name = 6;</code>
-     */
-    public Builder setAllocatorNameBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      allocatorName_ = value;
-      onChanged();
-      return this;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.MemoryLogRawAllocation)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.MemoryLogRawAllocation)
-  private static final org.tensorflow.proto.framework.MemoryLogRawAllocation DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.MemoryLogRawAllocation();
-  }
-
-  public static org.tensorflow.proto.framework.MemoryLogRawAllocation getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<MemoryLogRawAllocation>
-      PARSER = new com.google.protobuf.AbstractParser<MemoryLogRawAllocation>() {
-    @java.lang.Override
-    public MemoryLogRawAllocation parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new MemoryLogRawAllocation(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<MemoryLogRawAllocation> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<MemoryLogRawAllocation> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.framework.MemoryLogRawAllocation getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/MemoryLogRawAllocationOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/MemoryLogRawAllocationOrBuilder.java
deleted file mode 100644
index d8cabb6c9d4..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/MemoryLogRawAllocationOrBuilder.java
+++ /dev/null
@@ -1,82 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/framework/log_memory.proto
-
-package org.tensorflow.proto.framework;
-
-public interface MemoryLogRawAllocationOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.MemoryLogRawAllocation)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <pre>
-   * Process-unique step id.
-   * </pre>
-   *
-   * <code>int64 step_id = 1;</code>
-   */
-  long getStepId();
-
-  /**
-   * <pre>
-   * Name of the operation making the allocation.
-   * </pre>
-   *
-   * <code>string operation = 2;</code>
-   */
-  java.lang.String getOperation();
-  /**
-   * <pre>
-   * Name of the operation making the allocation.
-   * </pre>
-   *
-   * <code>string operation = 2;</code>
-   */
-  com.google.protobuf.ByteString
-      getOperationBytes();
-
-  /**
-   * <pre>
-   * Number of bytes in the allocation.
-   * </pre>
-   *
-   * <code>int64 num_bytes = 3;</code>
-   */
-  long getNumBytes();
-
-  /**
-   * <pre>
-   * Address of the allocation.
-   * </pre>
-   *
-   * <code>uint64 ptr = 4;</code>
-   */
-  long getPtr();
-
-  /**
-   * <pre>
-   * Id of the tensor buffer being allocated, used to match to a
-   * corresponding deallocation.
-   * </pre>
-   *
-   * <code>int64 allocation_id = 5;</code>
-   */
-  long getAllocationId();
-
-  /**
-   * <pre>
-   * Name of the allocator used.
-   * </pre>
-   *
-   * <code>string allocator_name = 6;</code>
-   */
-  java.lang.String getAllocatorName();
-  /**
-   * <pre>
-   * Name of the allocator used.
-   * </pre>
-   *
-   * <code>string allocator_name = 6;</code>
-   */
-  com.google.protobuf.ByteString
-      getAllocatorNameBytes();
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/MemoryLogRawDeallocation.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/MemoryLogRawDeallocation.java
deleted file mode 100644
index ae8c55fcf31..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/MemoryLogRawDeallocation.java
+++ /dev/null
@@ -1,959 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/framework/log_memory.proto
-
-package org.tensorflow.proto.framework;
-
-/**
- * Protobuf type {@code tensorflow.MemoryLogRawDeallocation}
- */
-public  final class MemoryLogRawDeallocation extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.MemoryLogRawDeallocation)
-    MemoryLogRawDeallocationOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use MemoryLogRawDeallocation.newBuilder() to construct.
-  private MemoryLogRawDeallocation(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private MemoryLogRawDeallocation() {
-    operation_ = "";
-    allocatorName_ = "";
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new MemoryLogRawDeallocation();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private MemoryLogRawDeallocation(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 8: {
-
-            stepId_ = input.readInt64();
-            break;
-          }
-          case 18: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            operation_ = s;
-            break;
-          }
-          case 24: {
-
-            allocationId_ = input.readInt64();
-            break;
-          }
-          case 34: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            allocatorName_ = s;
-            break;
-          }
-          case 40: {
-
-            deferred_ = input.readBool();
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.framework.LogMemoryProtos.internal_static_tensorflow_MemoryLogRawDeallocation_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.framework.LogMemoryProtos.internal_static_tensorflow_MemoryLogRawDeallocation_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.framework.MemoryLogRawDeallocation.class, org.tensorflow.proto.framework.MemoryLogRawDeallocation.Builder.class);
-  }
-
-  public static final int STEP_ID_FIELD_NUMBER = 1;
-  private long stepId_;
-  /**
-   * <pre>
-   * Process-unique step id.
-   * </pre>
-   *
-   * <code>int64 step_id = 1;</code>
-   */
-  public long getStepId() {
-    return stepId_;
-  }
-
-  public static final int OPERATION_FIELD_NUMBER = 2;
-  private volatile java.lang.Object operation_;
-  /**
-   * <pre>
-   * Name of the operation making the deallocation.
-   * </pre>
-   *
-   * <code>string operation = 2;</code>
-   */
-  public java.lang.String getOperation() {
-    java.lang.Object ref = operation_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      operation_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * Name of the operation making the deallocation.
-   * </pre>
-   *
-   * <code>string operation = 2;</code>
-   */
-  public com.google.protobuf.ByteString
-      getOperationBytes() {
-    java.lang.Object ref = operation_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      operation_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int ALLOCATION_ID_FIELD_NUMBER = 3;
-  private long allocationId_;
-  /**
-   * <pre>
-   * Id of the tensor buffer being deallocated, used to match to a
-   * corresponding allocation.
-   * </pre>
-   *
-   * <code>int64 allocation_id = 3;</code>
-   */
-  public long getAllocationId() {
-    return allocationId_;
-  }
-
-  public static final int ALLOCATOR_NAME_FIELD_NUMBER = 4;
-  private volatile java.lang.Object allocatorName_;
-  /**
-   * <pre>
-   * Name of the allocator used.
-   * </pre>
-   *
-   * <code>string allocator_name = 4;</code>
-   */
-  public java.lang.String getAllocatorName() {
-    java.lang.Object ref = allocatorName_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      allocatorName_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * Name of the allocator used.
-   * </pre>
-   *
-   * <code>string allocator_name = 4;</code>
-   */
-  public com.google.protobuf.ByteString
-      getAllocatorNameBytes() {
-    java.lang.Object ref = allocatorName_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      allocatorName_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int DEFERRED_FIELD_NUMBER = 5;
-  private boolean deferred_;
-  /**
-   * <pre>
-   * True if the deallocation is queued and will be performed later,
-   * e.g. for GPU lazy freeing of buffers.
-   * </pre>
-   *
-   * <code>bool deferred = 5;</code>
-   */
-  public boolean getDeferred() {
-    return deferred_;
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    if (stepId_ != 0L) {
-      output.writeInt64(1, stepId_);
-    }
-    if (!getOperationBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 2, operation_);
-    }
-    if (allocationId_ != 0L) {
-      output.writeInt64(3, allocationId_);
-    }
-    if (!getAllocatorNameBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 4, allocatorName_);
-    }
-    if (deferred_ != false) {
-      output.writeBool(5, deferred_);
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (stepId_ != 0L) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt64Size(1, stepId_);
-    }
-    if (!getOperationBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(2, operation_);
-    }
-    if (allocationId_ != 0L) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt64Size(3, allocationId_);
-    }
-    if (!getAllocatorNameBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(4, allocatorName_);
-    }
-    if (deferred_ != false) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeBoolSize(5, deferred_);
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.framework.MemoryLogRawDeallocation)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.framework.MemoryLogRawDeallocation other = (org.tensorflow.proto.framework.MemoryLogRawDeallocation) obj;
-
-    if (getStepId()
-        != other.getStepId()) return false;
-    if (!getOperation()
-        .equals(other.getOperation())) return false;
-    if (getAllocationId()
-        != other.getAllocationId()) return false;
-    if (!getAllocatorName()
-        .equals(other.getAllocatorName())) return false;
-    if (getDeferred()
-        != other.getDeferred()) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    hash = (37 * hash) + STEP_ID_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-        getStepId());
-    hash = (37 * hash) + OPERATION_FIELD_NUMBER;
-    hash = (53 * hash) + getOperation().hashCode();
-    hash = (37 * hash) + ALLOCATION_ID_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-        getAllocationId());
-    hash = (37 * hash) + ALLOCATOR_NAME_FIELD_NUMBER;
-    hash = (53 * hash) + getAllocatorName().hashCode();
-    hash = (37 * hash) + DEFERRED_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashBoolean(
-        getDeferred());
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.framework.MemoryLogRawDeallocation parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.MemoryLogRawDeallocation parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.MemoryLogRawDeallocation parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.MemoryLogRawDeallocation parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.MemoryLogRawDeallocation parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.MemoryLogRawDeallocation parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.MemoryLogRawDeallocation parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.MemoryLogRawDeallocation parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.MemoryLogRawDeallocation parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.MemoryLogRawDeallocation parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.MemoryLogRawDeallocation parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.MemoryLogRawDeallocation parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.framework.MemoryLogRawDeallocation prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * Protobuf type {@code tensorflow.MemoryLogRawDeallocation}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.MemoryLogRawDeallocation)
-      org.tensorflow.proto.framework.MemoryLogRawDeallocationOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.LogMemoryProtos.internal_static_tensorflow_MemoryLogRawDeallocation_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.LogMemoryProtos.internal_static_tensorflow_MemoryLogRawDeallocation_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.MemoryLogRawDeallocation.class, org.tensorflow.proto.framework.MemoryLogRawDeallocation.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.framework.MemoryLogRawDeallocation.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      stepId_ = 0L;
-
-      operation_ = "";
-
-      allocationId_ = 0L;
-
-      allocatorName_ = "";
-
-      deferred_ = false;
-
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.framework.LogMemoryProtos.internal_static_tensorflow_MemoryLogRawDeallocation_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.MemoryLogRawDeallocation getDefaultInstanceForType() {
-      return org.tensorflow.proto.framework.MemoryLogRawDeallocation.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.MemoryLogRawDeallocation build() {
-      org.tensorflow.proto.framework.MemoryLogRawDeallocation result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.MemoryLogRawDeallocation buildPartial() {
-      org.tensorflow.proto.framework.MemoryLogRawDeallocation result = new org.tensorflow.proto.framework.MemoryLogRawDeallocation(this);
-      result.stepId_ = stepId_;
-      result.operation_ = operation_;
-      result.allocationId_ = allocationId_;
-      result.allocatorName_ = allocatorName_;
-      result.deferred_ = deferred_;
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.framework.MemoryLogRawDeallocation) {
-        return mergeFrom((org.tensorflow.proto.framework.MemoryLogRawDeallocation)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.framework.MemoryLogRawDeallocation other) {
-      if (other == org.tensorflow.proto.framework.MemoryLogRawDeallocation.getDefaultInstance()) return this;
-      if (other.getStepId() != 0L) {
-        setStepId(other.getStepId());
-      }
-      if (!other.getOperation().isEmpty()) {
-        operation_ = other.operation_;
-        onChanged();
-      }
-      if (other.getAllocationId() != 0L) {
-        setAllocationId(other.getAllocationId());
-      }
-      if (!other.getAllocatorName().isEmpty()) {
-        allocatorName_ = other.allocatorName_;
-        onChanged();
-      }
-      if (other.getDeferred() != false) {
-        setDeferred(other.getDeferred());
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.framework.MemoryLogRawDeallocation parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.framework.MemoryLogRawDeallocation) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-
-    private long stepId_ ;
-    /**
-     * <pre>
-     * Process-unique step id.
-     * </pre>
-     *
-     * <code>int64 step_id = 1;</code>
-     */
-    public long getStepId() {
-      return stepId_;
-    }
-    /**
-     * <pre>
-     * Process-unique step id.
-     * </pre>
-     *
-     * <code>int64 step_id = 1;</code>
-     */
-    public Builder setStepId(long value) {
-      
-      stepId_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Process-unique step id.
-     * </pre>
-     *
-     * <code>int64 step_id = 1;</code>
-     */
-    public Builder clearStepId() {
-      
-      stepId_ = 0L;
-      onChanged();
-      return this;
-    }
-
-    private java.lang.Object operation_ = "";
-    /**
-     * <pre>
-     * Name of the operation making the deallocation.
-     * </pre>
-     *
-     * <code>string operation = 2;</code>
-     */
-    public java.lang.String getOperation() {
-      java.lang.Object ref = operation_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        operation_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Name of the operation making the deallocation.
-     * </pre>
-     *
-     * <code>string operation = 2;</code>
-     */
-    public com.google.protobuf.ByteString
-        getOperationBytes() {
-      java.lang.Object ref = operation_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        operation_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Name of the operation making the deallocation.
-     * </pre>
-     *
-     * <code>string operation = 2;</code>
-     */
-    public Builder setOperation(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      operation_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Name of the operation making the deallocation.
-     * </pre>
-     *
-     * <code>string operation = 2;</code>
-     */
-    public Builder clearOperation() {
-      
-      operation_ = getDefaultInstance().getOperation();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Name of the operation making the deallocation.
-     * </pre>
-     *
-     * <code>string operation = 2;</code>
-     */
-    public Builder setOperationBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      operation_ = value;
-      onChanged();
-      return this;
-    }
-
-    private long allocationId_ ;
-    /**
-     * <pre>
-     * Id of the tensor buffer being deallocated, used to match to a
-     * corresponding allocation.
-     * </pre>
-     *
-     * <code>int64 allocation_id = 3;</code>
-     */
-    public long getAllocationId() {
-      return allocationId_;
-    }
-    /**
-     * <pre>
-     * Id of the tensor buffer being deallocated, used to match to a
-     * corresponding allocation.
-     * </pre>
-     *
-     * <code>int64 allocation_id = 3;</code>
-     */
-    public Builder setAllocationId(long value) {
-      
-      allocationId_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Id of the tensor buffer being deallocated, used to match to a
-     * corresponding allocation.
-     * </pre>
-     *
-     * <code>int64 allocation_id = 3;</code>
-     */
-    public Builder clearAllocationId() {
-      
-      allocationId_ = 0L;
-      onChanged();
-      return this;
-    }
-
-    private java.lang.Object allocatorName_ = "";
-    /**
-     * <pre>
-     * Name of the allocator used.
-     * </pre>
-     *
-     * <code>string allocator_name = 4;</code>
-     */
-    public java.lang.String getAllocatorName() {
-      java.lang.Object ref = allocatorName_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        allocatorName_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Name of the allocator used.
-     * </pre>
-     *
-     * <code>string allocator_name = 4;</code>
-     */
-    public com.google.protobuf.ByteString
-        getAllocatorNameBytes() {
-      java.lang.Object ref = allocatorName_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        allocatorName_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Name of the allocator used.
-     * </pre>
-     *
-     * <code>string allocator_name = 4;</code>
-     */
-    public Builder setAllocatorName(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      allocatorName_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Name of the allocator used.
-     * </pre>
-     *
-     * <code>string allocator_name = 4;</code>
-     */
-    public Builder clearAllocatorName() {
-      
-      allocatorName_ = getDefaultInstance().getAllocatorName();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Name of the allocator used.
-     * </pre>
-     *
-     * <code>string allocator_name = 4;</code>
-     */
-    public Builder setAllocatorNameBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      allocatorName_ = value;
-      onChanged();
-      return this;
-    }
-
-    private boolean deferred_ ;
-    /**
-     * <pre>
-     * True if the deallocation is queued and will be performed later,
-     * e.g. for GPU lazy freeing of buffers.
-     * </pre>
-     *
-     * <code>bool deferred = 5;</code>
-     */
-    public boolean getDeferred() {
-      return deferred_;
-    }
-    /**
-     * <pre>
-     * True if the deallocation is queued and will be performed later,
-     * e.g. for GPU lazy freeing of buffers.
-     * </pre>
-     *
-     * <code>bool deferred = 5;</code>
-     */
-    public Builder setDeferred(boolean value) {
-      
-      deferred_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * True if the deallocation is queued and will be performed later,
-     * e.g. for GPU lazy freeing of buffers.
-     * </pre>
-     *
-     * <code>bool deferred = 5;</code>
-     */
-    public Builder clearDeferred() {
-      
-      deferred_ = false;
-      onChanged();
-      return this;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.MemoryLogRawDeallocation)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.MemoryLogRawDeallocation)
-  private static final org.tensorflow.proto.framework.MemoryLogRawDeallocation DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.MemoryLogRawDeallocation();
-  }
-
-  public static org.tensorflow.proto.framework.MemoryLogRawDeallocation getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<MemoryLogRawDeallocation>
-      PARSER = new com.google.protobuf.AbstractParser<MemoryLogRawDeallocation>() {
-    @java.lang.Override
-    public MemoryLogRawDeallocation parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new MemoryLogRawDeallocation(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<MemoryLogRawDeallocation> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<MemoryLogRawDeallocation> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.framework.MemoryLogRawDeallocation getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/MemoryLogRawDeallocationOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/MemoryLogRawDeallocationOrBuilder.java
deleted file mode 100644
index e8f66fff55c..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/MemoryLogRawDeallocationOrBuilder.java
+++ /dev/null
@@ -1,74 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/framework/log_memory.proto
-
-package org.tensorflow.proto.framework;
-
-public interface MemoryLogRawDeallocationOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.MemoryLogRawDeallocation)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <pre>
-   * Process-unique step id.
-   * </pre>
-   *
-   * <code>int64 step_id = 1;</code>
-   */
-  long getStepId();
-
-  /**
-   * <pre>
-   * Name of the operation making the deallocation.
-   * </pre>
-   *
-   * <code>string operation = 2;</code>
-   */
-  java.lang.String getOperation();
-  /**
-   * <pre>
-   * Name of the operation making the deallocation.
-   * </pre>
-   *
-   * <code>string operation = 2;</code>
-   */
-  com.google.protobuf.ByteString
-      getOperationBytes();
-
-  /**
-   * <pre>
-   * Id of the tensor buffer being deallocated, used to match to a
-   * corresponding allocation.
-   * </pre>
-   *
-   * <code>int64 allocation_id = 3;</code>
-   */
-  long getAllocationId();
-
-  /**
-   * <pre>
-   * Name of the allocator used.
-   * </pre>
-   *
-   * <code>string allocator_name = 4;</code>
-   */
-  java.lang.String getAllocatorName();
-  /**
-   * <pre>
-   * Name of the allocator used.
-   * </pre>
-   *
-   * <code>string allocator_name = 4;</code>
-   */
-  com.google.protobuf.ByteString
-      getAllocatorNameBytes();
-
-  /**
-   * <pre>
-   * True if the deallocation is queued and will be performed later,
-   * e.g. for GPU lazy freeing of buffers.
-   * </pre>
-   *
-   * <code>bool deferred = 5;</code>
-   */
-  boolean getDeferred();
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/MemoryLogStep.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/MemoryLogStep.java
deleted file mode 100644
index 73d400a36e0..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/MemoryLogStep.java
+++ /dev/null
@@ -1,648 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/framework/log_memory.proto
-
-package org.tensorflow.proto.framework;
-
-/**
- * Protobuf type {@code tensorflow.MemoryLogStep}
- */
-public  final class MemoryLogStep extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.MemoryLogStep)
-    MemoryLogStepOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use MemoryLogStep.newBuilder() to construct.
-  private MemoryLogStep(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private MemoryLogStep() {
-    handle_ = "";
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new MemoryLogStep();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private MemoryLogStep(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 8: {
-
-            stepId_ = input.readInt64();
-            break;
-          }
-          case 18: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            handle_ = s;
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.framework.LogMemoryProtos.internal_static_tensorflow_MemoryLogStep_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.framework.LogMemoryProtos.internal_static_tensorflow_MemoryLogStep_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.framework.MemoryLogStep.class, org.tensorflow.proto.framework.MemoryLogStep.Builder.class);
-  }
-
-  public static final int STEP_ID_FIELD_NUMBER = 1;
-  private long stepId_;
-  /**
-   * <pre>
-   * Process-unique step id.
-   * </pre>
-   *
-   * <code>int64 step_id = 1;</code>
-   */
-  public long getStepId() {
-    return stepId_;
-  }
-
-  public static final int HANDLE_FIELD_NUMBER = 2;
-  private volatile java.lang.Object handle_;
-  /**
-   * <pre>
-   * Handle describing the feeds and fetches of the step.
-   * </pre>
-   *
-   * <code>string handle = 2;</code>
-   */
-  public java.lang.String getHandle() {
-    java.lang.Object ref = handle_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      handle_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * Handle describing the feeds and fetches of the step.
-   * </pre>
-   *
-   * <code>string handle = 2;</code>
-   */
-  public com.google.protobuf.ByteString
-      getHandleBytes() {
-    java.lang.Object ref = handle_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      handle_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    if (stepId_ != 0L) {
-      output.writeInt64(1, stepId_);
-    }
-    if (!getHandleBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 2, handle_);
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (stepId_ != 0L) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt64Size(1, stepId_);
-    }
-    if (!getHandleBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(2, handle_);
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.framework.MemoryLogStep)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.framework.MemoryLogStep other = (org.tensorflow.proto.framework.MemoryLogStep) obj;
-
-    if (getStepId()
-        != other.getStepId()) return false;
-    if (!getHandle()
-        .equals(other.getHandle())) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    hash = (37 * hash) + STEP_ID_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-        getStepId());
-    hash = (37 * hash) + HANDLE_FIELD_NUMBER;
-    hash = (53 * hash) + getHandle().hashCode();
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.framework.MemoryLogStep parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.MemoryLogStep parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.MemoryLogStep parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.MemoryLogStep parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.MemoryLogStep parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.MemoryLogStep parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.MemoryLogStep parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.MemoryLogStep parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.MemoryLogStep parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.MemoryLogStep parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.MemoryLogStep parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.MemoryLogStep parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.framework.MemoryLogStep prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * Protobuf type {@code tensorflow.MemoryLogStep}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.MemoryLogStep)
-      org.tensorflow.proto.framework.MemoryLogStepOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.LogMemoryProtos.internal_static_tensorflow_MemoryLogStep_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.LogMemoryProtos.internal_static_tensorflow_MemoryLogStep_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.MemoryLogStep.class, org.tensorflow.proto.framework.MemoryLogStep.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.framework.MemoryLogStep.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      stepId_ = 0L;
-
-      handle_ = "";
-
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.framework.LogMemoryProtos.internal_static_tensorflow_MemoryLogStep_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.MemoryLogStep getDefaultInstanceForType() {
-      return org.tensorflow.proto.framework.MemoryLogStep.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.MemoryLogStep build() {
-      org.tensorflow.proto.framework.MemoryLogStep result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.MemoryLogStep buildPartial() {
-      org.tensorflow.proto.framework.MemoryLogStep result = new org.tensorflow.proto.framework.MemoryLogStep(this);
-      result.stepId_ = stepId_;
-      result.handle_ = handle_;
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.framework.MemoryLogStep) {
-        return mergeFrom((org.tensorflow.proto.framework.MemoryLogStep)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.framework.MemoryLogStep other) {
-      if (other == org.tensorflow.proto.framework.MemoryLogStep.getDefaultInstance()) return this;
-      if (other.getStepId() != 0L) {
-        setStepId(other.getStepId());
-      }
-      if (!other.getHandle().isEmpty()) {
-        handle_ = other.handle_;
-        onChanged();
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.framework.MemoryLogStep parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.framework.MemoryLogStep) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-
-    private long stepId_ ;
-    /**
-     * <pre>
-     * Process-unique step id.
-     * </pre>
-     *
-     * <code>int64 step_id = 1;</code>
-     */
-    public long getStepId() {
-      return stepId_;
-    }
-    /**
-     * <pre>
-     * Process-unique step id.
-     * </pre>
-     *
-     * <code>int64 step_id = 1;</code>
-     */
-    public Builder setStepId(long value) {
-      
-      stepId_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Process-unique step id.
-     * </pre>
-     *
-     * <code>int64 step_id = 1;</code>
-     */
-    public Builder clearStepId() {
-      
-      stepId_ = 0L;
-      onChanged();
-      return this;
-    }
-
-    private java.lang.Object handle_ = "";
-    /**
-     * <pre>
-     * Handle describing the feeds and fetches of the step.
-     * </pre>
-     *
-     * <code>string handle = 2;</code>
-     */
-    public java.lang.String getHandle() {
-      java.lang.Object ref = handle_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        handle_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Handle describing the feeds and fetches of the step.
-     * </pre>
-     *
-     * <code>string handle = 2;</code>
-     */
-    public com.google.protobuf.ByteString
-        getHandleBytes() {
-      java.lang.Object ref = handle_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        handle_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Handle describing the feeds and fetches of the step.
-     * </pre>
-     *
-     * <code>string handle = 2;</code>
-     */
-    public Builder setHandle(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      handle_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Handle describing the feeds and fetches of the step.
-     * </pre>
-     *
-     * <code>string handle = 2;</code>
-     */
-    public Builder clearHandle() {
-      
-      handle_ = getDefaultInstance().getHandle();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Handle describing the feeds and fetches of the step.
-     * </pre>
-     *
-     * <code>string handle = 2;</code>
-     */
-    public Builder setHandleBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      handle_ = value;
-      onChanged();
-      return this;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.MemoryLogStep)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.MemoryLogStep)
-  private static final org.tensorflow.proto.framework.MemoryLogStep DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.MemoryLogStep();
-  }
-
-  public static org.tensorflow.proto.framework.MemoryLogStep getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<MemoryLogStep>
-      PARSER = new com.google.protobuf.AbstractParser<MemoryLogStep>() {
-    @java.lang.Override
-    public MemoryLogStep parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new MemoryLogStep(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<MemoryLogStep> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<MemoryLogStep> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.framework.MemoryLogStep getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/MemoryLogStepOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/MemoryLogStepOrBuilder.java
deleted file mode 100644
index 3e8a13645c1..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/MemoryLogStepOrBuilder.java
+++ /dev/null
@@ -1,36 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/framework/log_memory.proto
-
-package org.tensorflow.proto.framework;
-
-public interface MemoryLogStepOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.MemoryLogStep)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <pre>
-   * Process-unique step id.
-   * </pre>
-   *
-   * <code>int64 step_id = 1;</code>
-   */
-  long getStepId();
-
-  /**
-   * <pre>
-   * Handle describing the feeds and fetches of the step.
-   * </pre>
-   *
-   * <code>string handle = 2;</code>
-   */
-  java.lang.String getHandle();
-  /**
-   * <pre>
-   * Handle describing the feeds and fetches of the step.
-   * </pre>
-   *
-   * <code>string handle = 2;</code>
-   */
-  com.google.protobuf.ByteString
-      getHandleBytes();
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/MemoryLogTensorAllocation.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/MemoryLogTensorAllocation.java
deleted file mode 100644
index bf59aacf6ed..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/MemoryLogTensorAllocation.java
+++ /dev/null
@@ -1,884 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/framework/log_memory.proto
-
-package org.tensorflow.proto.framework;
-
-/**
- * Protobuf type {@code tensorflow.MemoryLogTensorAllocation}
- */
-public  final class MemoryLogTensorAllocation extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.MemoryLogTensorAllocation)
-    MemoryLogTensorAllocationOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use MemoryLogTensorAllocation.newBuilder() to construct.
-  private MemoryLogTensorAllocation(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private MemoryLogTensorAllocation() {
-    kernelName_ = "";
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new MemoryLogTensorAllocation();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private MemoryLogTensorAllocation(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 8: {
-
-            stepId_ = input.readInt64();
-            break;
-          }
-          case 18: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            kernelName_ = s;
-            break;
-          }
-          case 26: {
-            org.tensorflow.proto.framework.TensorDescription.Builder subBuilder = null;
-            if (tensor_ != null) {
-              subBuilder = tensor_.toBuilder();
-            }
-            tensor_ = input.readMessage(org.tensorflow.proto.framework.TensorDescription.parser(), extensionRegistry);
-            if (subBuilder != null) {
-              subBuilder.mergeFrom(tensor_);
-              tensor_ = subBuilder.buildPartial();
-            }
-
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.framework.LogMemoryProtos.internal_static_tensorflow_MemoryLogTensorAllocation_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.framework.LogMemoryProtos.internal_static_tensorflow_MemoryLogTensorAllocation_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.framework.MemoryLogTensorAllocation.class, org.tensorflow.proto.framework.MemoryLogTensorAllocation.Builder.class);
-  }
-
-  public static final int STEP_ID_FIELD_NUMBER = 1;
-  private long stepId_;
-  /**
-   * <pre>
-   * Process-unique step id.
-   * </pre>
-   *
-   * <code>int64 step_id = 1;</code>
-   */
-  public long getStepId() {
-    return stepId_;
-  }
-
-  public static final int KERNEL_NAME_FIELD_NUMBER = 2;
-  private volatile java.lang.Object kernelName_;
-  /**
-   * <pre>
-   * Name of the kernel making the allocation as set in GraphDef,
-   * e.g., "affine2/weights/Assign".
-   * </pre>
-   *
-   * <code>string kernel_name = 2;</code>
-   */
-  public java.lang.String getKernelName() {
-    java.lang.Object ref = kernelName_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      kernelName_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * Name of the kernel making the allocation as set in GraphDef,
-   * e.g., "affine2/weights/Assign".
-   * </pre>
-   *
-   * <code>string kernel_name = 2;</code>
-   */
-  public com.google.protobuf.ByteString
-      getKernelNameBytes() {
-    java.lang.Object ref = kernelName_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      kernelName_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int TENSOR_FIELD_NUMBER = 3;
-  private org.tensorflow.proto.framework.TensorDescription tensor_;
-  /**
-   * <pre>
-   * Allocated tensor details.
-   * </pre>
-   *
-   * <code>.tensorflow.TensorDescription tensor = 3;</code>
-   */
-  public boolean hasTensor() {
-    return tensor_ != null;
-  }
-  /**
-   * <pre>
-   * Allocated tensor details.
-   * </pre>
-   *
-   * <code>.tensorflow.TensorDescription tensor = 3;</code>
-   */
-  public org.tensorflow.proto.framework.TensorDescription getTensor() {
-    return tensor_ == null ? org.tensorflow.proto.framework.TensorDescription.getDefaultInstance() : tensor_;
-  }
-  /**
-   * <pre>
-   * Allocated tensor details.
-   * </pre>
-   *
-   * <code>.tensorflow.TensorDescription tensor = 3;</code>
-   */
-  public org.tensorflow.proto.framework.TensorDescriptionOrBuilder getTensorOrBuilder() {
-    return getTensor();
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    if (stepId_ != 0L) {
-      output.writeInt64(1, stepId_);
-    }
-    if (!getKernelNameBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 2, kernelName_);
-    }
-    if (tensor_ != null) {
-      output.writeMessage(3, getTensor());
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (stepId_ != 0L) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt64Size(1, stepId_);
-    }
-    if (!getKernelNameBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(2, kernelName_);
-    }
-    if (tensor_ != null) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(3, getTensor());
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.framework.MemoryLogTensorAllocation)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.framework.MemoryLogTensorAllocation other = (org.tensorflow.proto.framework.MemoryLogTensorAllocation) obj;
-
-    if (getStepId()
-        != other.getStepId()) return false;
-    if (!getKernelName()
-        .equals(other.getKernelName())) return false;
-    if (hasTensor() != other.hasTensor()) return false;
-    if (hasTensor()) {
-      if (!getTensor()
-          .equals(other.getTensor())) return false;
-    }
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    hash = (37 * hash) + STEP_ID_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-        getStepId());
-    hash = (37 * hash) + KERNEL_NAME_FIELD_NUMBER;
-    hash = (53 * hash) + getKernelName().hashCode();
-    if (hasTensor()) {
-      hash = (37 * hash) + TENSOR_FIELD_NUMBER;
-      hash = (53 * hash) + getTensor().hashCode();
-    }
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.framework.MemoryLogTensorAllocation parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.MemoryLogTensorAllocation parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.MemoryLogTensorAllocation parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.MemoryLogTensorAllocation parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.MemoryLogTensorAllocation parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.MemoryLogTensorAllocation parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.MemoryLogTensorAllocation parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.MemoryLogTensorAllocation parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.MemoryLogTensorAllocation parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.MemoryLogTensorAllocation parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.MemoryLogTensorAllocation parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.MemoryLogTensorAllocation parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.framework.MemoryLogTensorAllocation prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * Protobuf type {@code tensorflow.MemoryLogTensorAllocation}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.MemoryLogTensorAllocation)
-      org.tensorflow.proto.framework.MemoryLogTensorAllocationOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.LogMemoryProtos.internal_static_tensorflow_MemoryLogTensorAllocation_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.LogMemoryProtos.internal_static_tensorflow_MemoryLogTensorAllocation_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.MemoryLogTensorAllocation.class, org.tensorflow.proto.framework.MemoryLogTensorAllocation.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.framework.MemoryLogTensorAllocation.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      stepId_ = 0L;
-
-      kernelName_ = "";
-
-      if (tensorBuilder_ == null) {
-        tensor_ = null;
-      } else {
-        tensor_ = null;
-        tensorBuilder_ = null;
-      }
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.framework.LogMemoryProtos.internal_static_tensorflow_MemoryLogTensorAllocation_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.MemoryLogTensorAllocation getDefaultInstanceForType() {
-      return org.tensorflow.proto.framework.MemoryLogTensorAllocation.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.MemoryLogTensorAllocation build() {
-      org.tensorflow.proto.framework.MemoryLogTensorAllocation result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.MemoryLogTensorAllocation buildPartial() {
-      org.tensorflow.proto.framework.MemoryLogTensorAllocation result = new org.tensorflow.proto.framework.MemoryLogTensorAllocation(this);
-      result.stepId_ = stepId_;
-      result.kernelName_ = kernelName_;
-      if (tensorBuilder_ == null) {
-        result.tensor_ = tensor_;
-      } else {
-        result.tensor_ = tensorBuilder_.build();
-      }
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.framework.MemoryLogTensorAllocation) {
-        return mergeFrom((org.tensorflow.proto.framework.MemoryLogTensorAllocation)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.framework.MemoryLogTensorAllocation other) {
-      if (other == org.tensorflow.proto.framework.MemoryLogTensorAllocation.getDefaultInstance()) return this;
-      if (other.getStepId() != 0L) {
-        setStepId(other.getStepId());
-      }
-      if (!other.getKernelName().isEmpty()) {
-        kernelName_ = other.kernelName_;
-        onChanged();
-      }
-      if (other.hasTensor()) {
-        mergeTensor(other.getTensor());
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.framework.MemoryLogTensorAllocation parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.framework.MemoryLogTensorAllocation) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-
-    private long stepId_ ;
-    /**
-     * <pre>
-     * Process-unique step id.
-     * </pre>
-     *
-     * <code>int64 step_id = 1;</code>
-     */
-    public long getStepId() {
-      return stepId_;
-    }
-    /**
-     * <pre>
-     * Process-unique step id.
-     * </pre>
-     *
-     * <code>int64 step_id = 1;</code>
-     */
-    public Builder setStepId(long value) {
-      
-      stepId_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Process-unique step id.
-     * </pre>
-     *
-     * <code>int64 step_id = 1;</code>
-     */
-    public Builder clearStepId() {
-      
-      stepId_ = 0L;
-      onChanged();
-      return this;
-    }
-
-    private java.lang.Object kernelName_ = "";
-    /**
-     * <pre>
-     * Name of the kernel making the allocation as set in GraphDef,
-     * e.g., "affine2/weights/Assign".
-     * </pre>
-     *
-     * <code>string kernel_name = 2;</code>
-     */
-    public java.lang.String getKernelName() {
-      java.lang.Object ref = kernelName_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        kernelName_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Name of the kernel making the allocation as set in GraphDef,
-     * e.g., "affine2/weights/Assign".
-     * </pre>
-     *
-     * <code>string kernel_name = 2;</code>
-     */
-    public com.google.protobuf.ByteString
-        getKernelNameBytes() {
-      java.lang.Object ref = kernelName_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        kernelName_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Name of the kernel making the allocation as set in GraphDef,
-     * e.g., "affine2/weights/Assign".
-     * </pre>
-     *
-     * <code>string kernel_name = 2;</code>
-     */
-    public Builder setKernelName(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      kernelName_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Name of the kernel making the allocation as set in GraphDef,
-     * e.g., "affine2/weights/Assign".
-     * </pre>
-     *
-     * <code>string kernel_name = 2;</code>
-     */
-    public Builder clearKernelName() {
-      
-      kernelName_ = getDefaultInstance().getKernelName();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Name of the kernel making the allocation as set in GraphDef,
-     * e.g., "affine2/weights/Assign".
-     * </pre>
-     *
-     * <code>string kernel_name = 2;</code>
-     */
-    public Builder setKernelNameBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      kernelName_ = value;
-      onChanged();
-      return this;
-    }
-
-    private org.tensorflow.proto.framework.TensorDescription tensor_;
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.TensorDescription, org.tensorflow.proto.framework.TensorDescription.Builder, org.tensorflow.proto.framework.TensorDescriptionOrBuilder> tensorBuilder_;
-    /**
-     * <pre>
-     * Allocated tensor details.
-     * </pre>
-     *
-     * <code>.tensorflow.TensorDescription tensor = 3;</code>
-     */
-    public boolean hasTensor() {
-      return tensorBuilder_ != null || tensor_ != null;
-    }
-    /**
-     * <pre>
-     * Allocated tensor details.
-     * </pre>
-     *
-     * <code>.tensorflow.TensorDescription tensor = 3;</code>
-     */
-    public org.tensorflow.proto.framework.TensorDescription getTensor() {
-      if (tensorBuilder_ == null) {
-        return tensor_ == null ? org.tensorflow.proto.framework.TensorDescription.getDefaultInstance() : tensor_;
-      } else {
-        return tensorBuilder_.getMessage();
-      }
-    }
-    /**
-     * <pre>
-     * Allocated tensor details.
-     * </pre>
-     *
-     * <code>.tensorflow.TensorDescription tensor = 3;</code>
-     */
-    public Builder setTensor(org.tensorflow.proto.framework.TensorDescription value) {
-      if (tensorBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        tensor_ = value;
-        onChanged();
-      } else {
-        tensorBuilder_.setMessage(value);
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Allocated tensor details.
-     * </pre>
-     *
-     * <code>.tensorflow.TensorDescription tensor = 3;</code>
-     */
-    public Builder setTensor(
-        org.tensorflow.proto.framework.TensorDescription.Builder builderForValue) {
-      if (tensorBuilder_ == null) {
-        tensor_ = builderForValue.build();
-        onChanged();
-      } else {
-        tensorBuilder_.setMessage(builderForValue.build());
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Allocated tensor details.
-     * </pre>
-     *
-     * <code>.tensorflow.TensorDescription tensor = 3;</code>
-     */
-    public Builder mergeTensor(org.tensorflow.proto.framework.TensorDescription value) {
-      if (tensorBuilder_ == null) {
-        if (tensor_ != null) {
-          tensor_ =
-            org.tensorflow.proto.framework.TensorDescription.newBuilder(tensor_).mergeFrom(value).buildPartial();
-        } else {
-          tensor_ = value;
-        }
-        onChanged();
-      } else {
-        tensorBuilder_.mergeFrom(value);
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Allocated tensor details.
-     * </pre>
-     *
-     * <code>.tensorflow.TensorDescription tensor = 3;</code>
-     */
-    public Builder clearTensor() {
-      if (tensorBuilder_ == null) {
-        tensor_ = null;
-        onChanged();
-      } else {
-        tensor_ = null;
-        tensorBuilder_ = null;
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Allocated tensor details.
-     * </pre>
-     *
-     * <code>.tensorflow.TensorDescription tensor = 3;</code>
-     */
-    public org.tensorflow.proto.framework.TensorDescription.Builder getTensorBuilder() {
-      
-      onChanged();
-      return getTensorFieldBuilder().getBuilder();
-    }
-    /**
-     * <pre>
-     * Allocated tensor details.
-     * </pre>
-     *
-     * <code>.tensorflow.TensorDescription tensor = 3;</code>
-     */
-    public org.tensorflow.proto.framework.TensorDescriptionOrBuilder getTensorOrBuilder() {
-      if (tensorBuilder_ != null) {
-        return tensorBuilder_.getMessageOrBuilder();
-      } else {
-        return tensor_ == null ?
-            org.tensorflow.proto.framework.TensorDescription.getDefaultInstance() : tensor_;
-      }
-    }
-    /**
-     * <pre>
-     * Allocated tensor details.
-     * </pre>
-     *
-     * <code>.tensorflow.TensorDescription tensor = 3;</code>
-     */
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.TensorDescription, org.tensorflow.proto.framework.TensorDescription.Builder, org.tensorflow.proto.framework.TensorDescriptionOrBuilder> 
-        getTensorFieldBuilder() {
-      if (tensorBuilder_ == null) {
-        tensorBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-            org.tensorflow.proto.framework.TensorDescription, org.tensorflow.proto.framework.TensorDescription.Builder, org.tensorflow.proto.framework.TensorDescriptionOrBuilder>(
-                getTensor(),
-                getParentForChildren(),
-                isClean());
-        tensor_ = null;
-      }
-      return tensorBuilder_;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.MemoryLogTensorAllocation)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.MemoryLogTensorAllocation)
-  private static final org.tensorflow.proto.framework.MemoryLogTensorAllocation DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.MemoryLogTensorAllocation();
-  }
-
-  public static org.tensorflow.proto.framework.MemoryLogTensorAllocation getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<MemoryLogTensorAllocation>
-      PARSER = new com.google.protobuf.AbstractParser<MemoryLogTensorAllocation>() {
-    @java.lang.Override
-    public MemoryLogTensorAllocation parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new MemoryLogTensorAllocation(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<MemoryLogTensorAllocation> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<MemoryLogTensorAllocation> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.framework.MemoryLogTensorAllocation getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/MemoryLogTensorAllocationOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/MemoryLogTensorAllocationOrBuilder.java
deleted file mode 100644
index 1b3a3732149..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/MemoryLogTensorAllocationOrBuilder.java
+++ /dev/null
@@ -1,63 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/framework/log_memory.proto
-
-package org.tensorflow.proto.framework;
-
-public interface MemoryLogTensorAllocationOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.MemoryLogTensorAllocation)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <pre>
-   * Process-unique step id.
-   * </pre>
-   *
-   * <code>int64 step_id = 1;</code>
-   */
-  long getStepId();
-
-  /**
-   * <pre>
-   * Name of the kernel making the allocation as set in GraphDef,
-   * e.g., "affine2/weights/Assign".
-   * </pre>
-   *
-   * <code>string kernel_name = 2;</code>
-   */
-  java.lang.String getKernelName();
-  /**
-   * <pre>
-   * Name of the kernel making the allocation as set in GraphDef,
-   * e.g., "affine2/weights/Assign".
-   * </pre>
-   *
-   * <code>string kernel_name = 2;</code>
-   */
-  com.google.protobuf.ByteString
-      getKernelNameBytes();
-
-  /**
-   * <pre>
-   * Allocated tensor details.
-   * </pre>
-   *
-   * <code>.tensorflow.TensorDescription tensor = 3;</code>
-   */
-  boolean hasTensor();
-  /**
-   * <pre>
-   * Allocated tensor details.
-   * </pre>
-   *
-   * <code>.tensorflow.TensorDescription tensor = 3;</code>
-   */
-  org.tensorflow.proto.framework.TensorDescription getTensor();
-  /**
-   * <pre>
-   * Allocated tensor details.
-   * </pre>
-   *
-   * <code>.tensorflow.TensorDescription tensor = 3;</code>
-   */
-  org.tensorflow.proto.framework.TensorDescriptionOrBuilder getTensorOrBuilder();
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/MemoryLogTensorDeallocation.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/MemoryLogTensorDeallocation.java
deleted file mode 100644
index f0061b53c50..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/MemoryLogTensorDeallocation.java
+++ /dev/null
@@ -1,652 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/framework/log_memory.proto
-
-package org.tensorflow.proto.framework;
-
-/**
- * Protobuf type {@code tensorflow.MemoryLogTensorDeallocation}
- */
-public  final class MemoryLogTensorDeallocation extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.MemoryLogTensorDeallocation)
-    MemoryLogTensorDeallocationOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use MemoryLogTensorDeallocation.newBuilder() to construct.
-  private MemoryLogTensorDeallocation(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private MemoryLogTensorDeallocation() {
-    allocatorName_ = "";
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new MemoryLogTensorDeallocation();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private MemoryLogTensorDeallocation(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 8: {
-
-            allocationId_ = input.readInt64();
-            break;
-          }
-          case 18: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            allocatorName_ = s;
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.framework.LogMemoryProtos.internal_static_tensorflow_MemoryLogTensorDeallocation_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.framework.LogMemoryProtos.internal_static_tensorflow_MemoryLogTensorDeallocation_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.framework.MemoryLogTensorDeallocation.class, org.tensorflow.proto.framework.MemoryLogTensorDeallocation.Builder.class);
-  }
-
-  public static final int ALLOCATION_ID_FIELD_NUMBER = 1;
-  private long allocationId_;
-  /**
-   * <pre>
-   * Id of the tensor buffer being deallocated, used to match to a
-   * corresponding allocation.
-   * </pre>
-   *
-   * <code>int64 allocation_id = 1;</code>
-   */
-  public long getAllocationId() {
-    return allocationId_;
-  }
-
-  public static final int ALLOCATOR_NAME_FIELD_NUMBER = 2;
-  private volatile java.lang.Object allocatorName_;
-  /**
-   * <pre>
-   * Name of the allocator used.
-   * </pre>
-   *
-   * <code>string allocator_name = 2;</code>
-   */
-  public java.lang.String getAllocatorName() {
-    java.lang.Object ref = allocatorName_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      allocatorName_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * Name of the allocator used.
-   * </pre>
-   *
-   * <code>string allocator_name = 2;</code>
-   */
-  public com.google.protobuf.ByteString
-      getAllocatorNameBytes() {
-    java.lang.Object ref = allocatorName_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      allocatorName_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    if (allocationId_ != 0L) {
-      output.writeInt64(1, allocationId_);
-    }
-    if (!getAllocatorNameBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 2, allocatorName_);
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (allocationId_ != 0L) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt64Size(1, allocationId_);
-    }
-    if (!getAllocatorNameBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(2, allocatorName_);
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.framework.MemoryLogTensorDeallocation)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.framework.MemoryLogTensorDeallocation other = (org.tensorflow.proto.framework.MemoryLogTensorDeallocation) obj;
-
-    if (getAllocationId()
-        != other.getAllocationId()) return false;
-    if (!getAllocatorName()
-        .equals(other.getAllocatorName())) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    hash = (37 * hash) + ALLOCATION_ID_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-        getAllocationId());
-    hash = (37 * hash) + ALLOCATOR_NAME_FIELD_NUMBER;
-    hash = (53 * hash) + getAllocatorName().hashCode();
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.framework.MemoryLogTensorDeallocation parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.MemoryLogTensorDeallocation parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.MemoryLogTensorDeallocation parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.MemoryLogTensorDeallocation parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.MemoryLogTensorDeallocation parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.MemoryLogTensorDeallocation parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.MemoryLogTensorDeallocation parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.MemoryLogTensorDeallocation parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.MemoryLogTensorDeallocation parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.MemoryLogTensorDeallocation parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.MemoryLogTensorDeallocation parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.MemoryLogTensorDeallocation parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.framework.MemoryLogTensorDeallocation prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * Protobuf type {@code tensorflow.MemoryLogTensorDeallocation}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.MemoryLogTensorDeallocation)
-      org.tensorflow.proto.framework.MemoryLogTensorDeallocationOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.LogMemoryProtos.internal_static_tensorflow_MemoryLogTensorDeallocation_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.LogMemoryProtos.internal_static_tensorflow_MemoryLogTensorDeallocation_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.MemoryLogTensorDeallocation.class, org.tensorflow.proto.framework.MemoryLogTensorDeallocation.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.framework.MemoryLogTensorDeallocation.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      allocationId_ = 0L;
-
-      allocatorName_ = "";
-
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.framework.LogMemoryProtos.internal_static_tensorflow_MemoryLogTensorDeallocation_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.MemoryLogTensorDeallocation getDefaultInstanceForType() {
-      return org.tensorflow.proto.framework.MemoryLogTensorDeallocation.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.MemoryLogTensorDeallocation build() {
-      org.tensorflow.proto.framework.MemoryLogTensorDeallocation result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.MemoryLogTensorDeallocation buildPartial() {
-      org.tensorflow.proto.framework.MemoryLogTensorDeallocation result = new org.tensorflow.proto.framework.MemoryLogTensorDeallocation(this);
-      result.allocationId_ = allocationId_;
-      result.allocatorName_ = allocatorName_;
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.framework.MemoryLogTensorDeallocation) {
-        return mergeFrom((org.tensorflow.proto.framework.MemoryLogTensorDeallocation)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.framework.MemoryLogTensorDeallocation other) {
-      if (other == org.tensorflow.proto.framework.MemoryLogTensorDeallocation.getDefaultInstance()) return this;
-      if (other.getAllocationId() != 0L) {
-        setAllocationId(other.getAllocationId());
-      }
-      if (!other.getAllocatorName().isEmpty()) {
-        allocatorName_ = other.allocatorName_;
-        onChanged();
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.framework.MemoryLogTensorDeallocation parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.framework.MemoryLogTensorDeallocation) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-
-    private long allocationId_ ;
-    /**
-     * <pre>
-     * Id of the tensor buffer being deallocated, used to match to a
-     * corresponding allocation.
-     * </pre>
-     *
-     * <code>int64 allocation_id = 1;</code>
-     */
-    public long getAllocationId() {
-      return allocationId_;
-    }
-    /**
-     * <pre>
-     * Id of the tensor buffer being deallocated, used to match to a
-     * corresponding allocation.
-     * </pre>
-     *
-     * <code>int64 allocation_id = 1;</code>
-     */
-    public Builder setAllocationId(long value) {
-      
-      allocationId_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Id of the tensor buffer being deallocated, used to match to a
-     * corresponding allocation.
-     * </pre>
-     *
-     * <code>int64 allocation_id = 1;</code>
-     */
-    public Builder clearAllocationId() {
-      
-      allocationId_ = 0L;
-      onChanged();
-      return this;
-    }
-
-    private java.lang.Object allocatorName_ = "";
-    /**
-     * <pre>
-     * Name of the allocator used.
-     * </pre>
-     *
-     * <code>string allocator_name = 2;</code>
-     */
-    public java.lang.String getAllocatorName() {
-      java.lang.Object ref = allocatorName_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        allocatorName_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Name of the allocator used.
-     * </pre>
-     *
-     * <code>string allocator_name = 2;</code>
-     */
-    public com.google.protobuf.ByteString
-        getAllocatorNameBytes() {
-      java.lang.Object ref = allocatorName_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        allocatorName_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Name of the allocator used.
-     * </pre>
-     *
-     * <code>string allocator_name = 2;</code>
-     */
-    public Builder setAllocatorName(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      allocatorName_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Name of the allocator used.
-     * </pre>
-     *
-     * <code>string allocator_name = 2;</code>
-     */
-    public Builder clearAllocatorName() {
-      
-      allocatorName_ = getDefaultInstance().getAllocatorName();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Name of the allocator used.
-     * </pre>
-     *
-     * <code>string allocator_name = 2;</code>
-     */
-    public Builder setAllocatorNameBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      allocatorName_ = value;
-      onChanged();
-      return this;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.MemoryLogTensorDeallocation)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.MemoryLogTensorDeallocation)
-  private static final org.tensorflow.proto.framework.MemoryLogTensorDeallocation DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.MemoryLogTensorDeallocation();
-  }
-
-  public static org.tensorflow.proto.framework.MemoryLogTensorDeallocation getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<MemoryLogTensorDeallocation>
-      PARSER = new com.google.protobuf.AbstractParser<MemoryLogTensorDeallocation>() {
-    @java.lang.Override
-    public MemoryLogTensorDeallocation parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new MemoryLogTensorDeallocation(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<MemoryLogTensorDeallocation> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<MemoryLogTensorDeallocation> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.framework.MemoryLogTensorDeallocation getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/MemoryLogTensorDeallocationOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/MemoryLogTensorDeallocationOrBuilder.java
deleted file mode 100644
index 7d45248a17a..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/MemoryLogTensorDeallocationOrBuilder.java
+++ /dev/null
@@ -1,37 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/framework/log_memory.proto
-
-package org.tensorflow.proto.framework;
-
-public interface MemoryLogTensorDeallocationOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.MemoryLogTensorDeallocation)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <pre>
-   * Id of the tensor buffer being deallocated, used to match to a
-   * corresponding allocation.
-   * </pre>
-   *
-   * <code>int64 allocation_id = 1;</code>
-   */
-  long getAllocationId();
-
-  /**
-   * <pre>
-   * Name of the allocator used.
-   * </pre>
-   *
-   * <code>string allocator_name = 2;</code>
-   */
-  java.lang.String getAllocatorName();
-  /**
-   * <pre>
-   * Name of the allocator used.
-   * </pre>
-   *
-   * <code>string allocator_name = 2;</code>
-   */
-  com.google.protobuf.ByteString
-      getAllocatorNameBytes();
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/MemoryLogTensorOutput.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/MemoryLogTensorOutput.java
deleted file mode 100644
index a8aaf7dc7d9..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/MemoryLogTensorOutput.java
+++ /dev/null
@@ -1,957 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/framework/log_memory.proto
-
-package org.tensorflow.proto.framework;
-
-/**
- * Protobuf type {@code tensorflow.MemoryLogTensorOutput}
- */
-public  final class MemoryLogTensorOutput extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.MemoryLogTensorOutput)
-    MemoryLogTensorOutputOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use MemoryLogTensorOutput.newBuilder() to construct.
-  private MemoryLogTensorOutput(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private MemoryLogTensorOutput() {
-    kernelName_ = "";
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new MemoryLogTensorOutput();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private MemoryLogTensorOutput(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 8: {
-
-            stepId_ = input.readInt64();
-            break;
-          }
-          case 18: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            kernelName_ = s;
-            break;
-          }
-          case 24: {
-
-            index_ = input.readInt32();
-            break;
-          }
-          case 34: {
-            org.tensorflow.proto.framework.TensorDescription.Builder subBuilder = null;
-            if (tensor_ != null) {
-              subBuilder = tensor_.toBuilder();
-            }
-            tensor_ = input.readMessage(org.tensorflow.proto.framework.TensorDescription.parser(), extensionRegistry);
-            if (subBuilder != null) {
-              subBuilder.mergeFrom(tensor_);
-              tensor_ = subBuilder.buildPartial();
-            }
-
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.framework.LogMemoryProtos.internal_static_tensorflow_MemoryLogTensorOutput_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.framework.LogMemoryProtos.internal_static_tensorflow_MemoryLogTensorOutput_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.framework.MemoryLogTensorOutput.class, org.tensorflow.proto.framework.MemoryLogTensorOutput.Builder.class);
-  }
-
-  public static final int STEP_ID_FIELD_NUMBER = 1;
-  private long stepId_;
-  /**
-   * <pre>
-   * Process-unique step id.
-   * </pre>
-   *
-   * <code>int64 step_id = 1;</code>
-   */
-  public long getStepId() {
-    return stepId_;
-  }
-
-  public static final int KERNEL_NAME_FIELD_NUMBER = 2;
-  private volatile java.lang.Object kernelName_;
-  /**
-   * <pre>
-   * Name of the kernel producing an output as set in GraphDef, e.g.,
-   * "affine2/weights/Assign".
-   * </pre>
-   *
-   * <code>string kernel_name = 2;</code>
-   */
-  public java.lang.String getKernelName() {
-    java.lang.Object ref = kernelName_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      kernelName_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * Name of the kernel producing an output as set in GraphDef, e.g.,
-   * "affine2/weights/Assign".
-   * </pre>
-   *
-   * <code>string kernel_name = 2;</code>
-   */
-  public com.google.protobuf.ByteString
-      getKernelNameBytes() {
-    java.lang.Object ref = kernelName_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      kernelName_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int INDEX_FIELD_NUMBER = 3;
-  private int index_;
-  /**
-   * <pre>
-   * Index of the output being set.
-   * </pre>
-   *
-   * <code>int32 index = 3;</code>
-   */
-  public int getIndex() {
-    return index_;
-  }
-
-  public static final int TENSOR_FIELD_NUMBER = 4;
-  private org.tensorflow.proto.framework.TensorDescription tensor_;
-  /**
-   * <pre>
-   * Output tensor details.
-   * </pre>
-   *
-   * <code>.tensorflow.TensorDescription tensor = 4;</code>
-   */
-  public boolean hasTensor() {
-    return tensor_ != null;
-  }
-  /**
-   * <pre>
-   * Output tensor details.
-   * </pre>
-   *
-   * <code>.tensorflow.TensorDescription tensor = 4;</code>
-   */
-  public org.tensorflow.proto.framework.TensorDescription getTensor() {
-    return tensor_ == null ? org.tensorflow.proto.framework.TensorDescription.getDefaultInstance() : tensor_;
-  }
-  /**
-   * <pre>
-   * Output tensor details.
-   * </pre>
-   *
-   * <code>.tensorflow.TensorDescription tensor = 4;</code>
-   */
-  public org.tensorflow.proto.framework.TensorDescriptionOrBuilder getTensorOrBuilder() {
-    return getTensor();
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    if (stepId_ != 0L) {
-      output.writeInt64(1, stepId_);
-    }
-    if (!getKernelNameBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 2, kernelName_);
-    }
-    if (index_ != 0) {
-      output.writeInt32(3, index_);
-    }
-    if (tensor_ != null) {
-      output.writeMessage(4, getTensor());
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (stepId_ != 0L) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt64Size(1, stepId_);
-    }
-    if (!getKernelNameBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(2, kernelName_);
-    }
-    if (index_ != 0) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt32Size(3, index_);
-    }
-    if (tensor_ != null) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(4, getTensor());
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.framework.MemoryLogTensorOutput)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.framework.MemoryLogTensorOutput other = (org.tensorflow.proto.framework.MemoryLogTensorOutput) obj;
-
-    if (getStepId()
-        != other.getStepId()) return false;
-    if (!getKernelName()
-        .equals(other.getKernelName())) return false;
-    if (getIndex()
-        != other.getIndex()) return false;
-    if (hasTensor() != other.hasTensor()) return false;
-    if (hasTensor()) {
-      if (!getTensor()
-          .equals(other.getTensor())) return false;
-    }
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    hash = (37 * hash) + STEP_ID_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-        getStepId());
-    hash = (37 * hash) + KERNEL_NAME_FIELD_NUMBER;
-    hash = (53 * hash) + getKernelName().hashCode();
-    hash = (37 * hash) + INDEX_FIELD_NUMBER;
-    hash = (53 * hash) + getIndex();
-    if (hasTensor()) {
-      hash = (37 * hash) + TENSOR_FIELD_NUMBER;
-      hash = (53 * hash) + getTensor().hashCode();
-    }
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.framework.MemoryLogTensorOutput parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.MemoryLogTensorOutput parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.MemoryLogTensorOutput parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.MemoryLogTensorOutput parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.MemoryLogTensorOutput parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.MemoryLogTensorOutput parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.MemoryLogTensorOutput parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.MemoryLogTensorOutput parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.MemoryLogTensorOutput parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.MemoryLogTensorOutput parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.MemoryLogTensorOutput parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.MemoryLogTensorOutput parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.framework.MemoryLogTensorOutput prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * Protobuf type {@code tensorflow.MemoryLogTensorOutput}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.MemoryLogTensorOutput)
-      org.tensorflow.proto.framework.MemoryLogTensorOutputOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.LogMemoryProtos.internal_static_tensorflow_MemoryLogTensorOutput_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.LogMemoryProtos.internal_static_tensorflow_MemoryLogTensorOutput_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.MemoryLogTensorOutput.class, org.tensorflow.proto.framework.MemoryLogTensorOutput.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.framework.MemoryLogTensorOutput.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      stepId_ = 0L;
-
-      kernelName_ = "";
-
-      index_ = 0;
-
-      if (tensorBuilder_ == null) {
-        tensor_ = null;
-      } else {
-        tensor_ = null;
-        tensorBuilder_ = null;
-      }
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.framework.LogMemoryProtos.internal_static_tensorflow_MemoryLogTensorOutput_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.MemoryLogTensorOutput getDefaultInstanceForType() {
-      return org.tensorflow.proto.framework.MemoryLogTensorOutput.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.MemoryLogTensorOutput build() {
-      org.tensorflow.proto.framework.MemoryLogTensorOutput result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.MemoryLogTensorOutput buildPartial() {
-      org.tensorflow.proto.framework.MemoryLogTensorOutput result = new org.tensorflow.proto.framework.MemoryLogTensorOutput(this);
-      result.stepId_ = stepId_;
-      result.kernelName_ = kernelName_;
-      result.index_ = index_;
-      if (tensorBuilder_ == null) {
-        result.tensor_ = tensor_;
-      } else {
-        result.tensor_ = tensorBuilder_.build();
-      }
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.framework.MemoryLogTensorOutput) {
-        return mergeFrom((org.tensorflow.proto.framework.MemoryLogTensorOutput)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.framework.MemoryLogTensorOutput other) {
-      if (other == org.tensorflow.proto.framework.MemoryLogTensorOutput.getDefaultInstance()) return this;
-      if (other.getStepId() != 0L) {
-        setStepId(other.getStepId());
-      }
-      if (!other.getKernelName().isEmpty()) {
-        kernelName_ = other.kernelName_;
-        onChanged();
-      }
-      if (other.getIndex() != 0) {
-        setIndex(other.getIndex());
-      }
-      if (other.hasTensor()) {
-        mergeTensor(other.getTensor());
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.framework.MemoryLogTensorOutput parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.framework.MemoryLogTensorOutput) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-
-    private long stepId_ ;
-    /**
-     * <pre>
-     * Process-unique step id.
-     * </pre>
-     *
-     * <code>int64 step_id = 1;</code>
-     */
-    public long getStepId() {
-      return stepId_;
-    }
-    /**
-     * <pre>
-     * Process-unique step id.
-     * </pre>
-     *
-     * <code>int64 step_id = 1;</code>
-     */
-    public Builder setStepId(long value) {
-      
-      stepId_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Process-unique step id.
-     * </pre>
-     *
-     * <code>int64 step_id = 1;</code>
-     */
-    public Builder clearStepId() {
-      
-      stepId_ = 0L;
-      onChanged();
-      return this;
-    }
-
-    private java.lang.Object kernelName_ = "";
-    /**
-     * <pre>
-     * Name of the kernel producing an output as set in GraphDef, e.g.,
-     * "affine2/weights/Assign".
-     * </pre>
-     *
-     * <code>string kernel_name = 2;</code>
-     */
-    public java.lang.String getKernelName() {
-      java.lang.Object ref = kernelName_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        kernelName_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Name of the kernel producing an output as set in GraphDef, e.g.,
-     * "affine2/weights/Assign".
-     * </pre>
-     *
-     * <code>string kernel_name = 2;</code>
-     */
-    public com.google.protobuf.ByteString
-        getKernelNameBytes() {
-      java.lang.Object ref = kernelName_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        kernelName_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Name of the kernel producing an output as set in GraphDef, e.g.,
-     * "affine2/weights/Assign".
-     * </pre>
-     *
-     * <code>string kernel_name = 2;</code>
-     */
-    public Builder setKernelName(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      kernelName_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Name of the kernel producing an output as set in GraphDef, e.g.,
-     * "affine2/weights/Assign".
-     * </pre>
-     *
-     * <code>string kernel_name = 2;</code>
-     */
-    public Builder clearKernelName() {
-      
-      kernelName_ = getDefaultInstance().getKernelName();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Name of the kernel producing an output as set in GraphDef, e.g.,
-     * "affine2/weights/Assign".
-     * </pre>
-     *
-     * <code>string kernel_name = 2;</code>
-     */
-    public Builder setKernelNameBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      kernelName_ = value;
-      onChanged();
-      return this;
-    }
-
-    private int index_ ;
-    /**
-     * <pre>
-     * Index of the output being set.
-     * </pre>
-     *
-     * <code>int32 index = 3;</code>
-     */
-    public int getIndex() {
-      return index_;
-    }
-    /**
-     * <pre>
-     * Index of the output being set.
-     * </pre>
-     *
-     * <code>int32 index = 3;</code>
-     */
-    public Builder setIndex(int value) {
-      
-      index_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Index of the output being set.
-     * </pre>
-     *
-     * <code>int32 index = 3;</code>
-     */
-    public Builder clearIndex() {
-      
-      index_ = 0;
-      onChanged();
-      return this;
-    }
-
-    private org.tensorflow.proto.framework.TensorDescription tensor_;
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.TensorDescription, org.tensorflow.proto.framework.TensorDescription.Builder, org.tensorflow.proto.framework.TensorDescriptionOrBuilder> tensorBuilder_;
-    /**
-     * <pre>
-     * Output tensor details.
-     * </pre>
-     *
-     * <code>.tensorflow.TensorDescription tensor = 4;</code>
-     */
-    public boolean hasTensor() {
-      return tensorBuilder_ != null || tensor_ != null;
-    }
-    /**
-     * <pre>
-     * Output tensor details.
-     * </pre>
-     *
-     * <code>.tensorflow.TensorDescription tensor = 4;</code>
-     */
-    public org.tensorflow.proto.framework.TensorDescription getTensor() {
-      if (tensorBuilder_ == null) {
-        return tensor_ == null ? org.tensorflow.proto.framework.TensorDescription.getDefaultInstance() : tensor_;
-      } else {
-        return tensorBuilder_.getMessage();
-      }
-    }
-    /**
-     * <pre>
-     * Output tensor details.
-     * </pre>
-     *
-     * <code>.tensorflow.TensorDescription tensor = 4;</code>
-     */
-    public Builder setTensor(org.tensorflow.proto.framework.TensorDescription value) {
-      if (tensorBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        tensor_ = value;
-        onChanged();
-      } else {
-        tensorBuilder_.setMessage(value);
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Output tensor details.
-     * </pre>
-     *
-     * <code>.tensorflow.TensorDescription tensor = 4;</code>
-     */
-    public Builder setTensor(
-        org.tensorflow.proto.framework.TensorDescription.Builder builderForValue) {
-      if (tensorBuilder_ == null) {
-        tensor_ = builderForValue.build();
-        onChanged();
-      } else {
-        tensorBuilder_.setMessage(builderForValue.build());
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Output tensor details.
-     * </pre>
-     *
-     * <code>.tensorflow.TensorDescription tensor = 4;</code>
-     */
-    public Builder mergeTensor(org.tensorflow.proto.framework.TensorDescription value) {
-      if (tensorBuilder_ == null) {
-        if (tensor_ != null) {
-          tensor_ =
-            org.tensorflow.proto.framework.TensorDescription.newBuilder(tensor_).mergeFrom(value).buildPartial();
-        } else {
-          tensor_ = value;
-        }
-        onChanged();
-      } else {
-        tensorBuilder_.mergeFrom(value);
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Output tensor details.
-     * </pre>
-     *
-     * <code>.tensorflow.TensorDescription tensor = 4;</code>
-     */
-    public Builder clearTensor() {
-      if (tensorBuilder_ == null) {
-        tensor_ = null;
-        onChanged();
-      } else {
-        tensor_ = null;
-        tensorBuilder_ = null;
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Output tensor details.
-     * </pre>
-     *
-     * <code>.tensorflow.TensorDescription tensor = 4;</code>
-     */
-    public org.tensorflow.proto.framework.TensorDescription.Builder getTensorBuilder() {
-      
-      onChanged();
-      return getTensorFieldBuilder().getBuilder();
-    }
-    /**
-     * <pre>
-     * Output tensor details.
-     * </pre>
-     *
-     * <code>.tensorflow.TensorDescription tensor = 4;</code>
-     */
-    public org.tensorflow.proto.framework.TensorDescriptionOrBuilder getTensorOrBuilder() {
-      if (tensorBuilder_ != null) {
-        return tensorBuilder_.getMessageOrBuilder();
-      } else {
-        return tensor_ == null ?
-            org.tensorflow.proto.framework.TensorDescription.getDefaultInstance() : tensor_;
-      }
-    }
-    /**
-     * <pre>
-     * Output tensor details.
-     * </pre>
-     *
-     * <code>.tensorflow.TensorDescription tensor = 4;</code>
-     */
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.TensorDescription, org.tensorflow.proto.framework.TensorDescription.Builder, org.tensorflow.proto.framework.TensorDescriptionOrBuilder> 
-        getTensorFieldBuilder() {
-      if (tensorBuilder_ == null) {
-        tensorBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-            org.tensorflow.proto.framework.TensorDescription, org.tensorflow.proto.framework.TensorDescription.Builder, org.tensorflow.proto.framework.TensorDescriptionOrBuilder>(
-                getTensor(),
-                getParentForChildren(),
-                isClean());
-        tensor_ = null;
-      }
-      return tensorBuilder_;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.MemoryLogTensorOutput)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.MemoryLogTensorOutput)
-  private static final org.tensorflow.proto.framework.MemoryLogTensorOutput DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.MemoryLogTensorOutput();
-  }
-
-  public static org.tensorflow.proto.framework.MemoryLogTensorOutput getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<MemoryLogTensorOutput>
-      PARSER = new com.google.protobuf.AbstractParser<MemoryLogTensorOutput>() {
-    @java.lang.Override
-    public MemoryLogTensorOutput parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new MemoryLogTensorOutput(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<MemoryLogTensorOutput> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<MemoryLogTensorOutput> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.framework.MemoryLogTensorOutput getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/MemoryLogTensorOutputOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/MemoryLogTensorOutputOrBuilder.java
deleted file mode 100644
index b9275e22f7e..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/MemoryLogTensorOutputOrBuilder.java
+++ /dev/null
@@ -1,72 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/framework/log_memory.proto
-
-package org.tensorflow.proto.framework;
-
-public interface MemoryLogTensorOutputOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.MemoryLogTensorOutput)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <pre>
-   * Process-unique step id.
-   * </pre>
-   *
-   * <code>int64 step_id = 1;</code>
-   */
-  long getStepId();
-
-  /**
-   * <pre>
-   * Name of the kernel producing an output as set in GraphDef, e.g.,
-   * "affine2/weights/Assign".
-   * </pre>
-   *
-   * <code>string kernel_name = 2;</code>
-   */
-  java.lang.String getKernelName();
-  /**
-   * <pre>
-   * Name of the kernel producing an output as set in GraphDef, e.g.,
-   * "affine2/weights/Assign".
-   * </pre>
-   *
-   * <code>string kernel_name = 2;</code>
-   */
-  com.google.protobuf.ByteString
-      getKernelNameBytes();
-
-  /**
-   * <pre>
-   * Index of the output being set.
-   * </pre>
-   *
-   * <code>int32 index = 3;</code>
-   */
-  int getIndex();
-
-  /**
-   * <pre>
-   * Output tensor details.
-   * </pre>
-   *
-   * <code>.tensorflow.TensorDescription tensor = 4;</code>
-   */
-  boolean hasTensor();
-  /**
-   * <pre>
-   * Output tensor details.
-   * </pre>
-   *
-   * <code>.tensorflow.TensorDescription tensor = 4;</code>
-   */
-  org.tensorflow.proto.framework.TensorDescription getTensor();
-  /**
-   * <pre>
-   * Output tensor details.
-   * </pre>
-   *
-   * <code>.tensorflow.TensorDescription tensor = 4;</code>
-   */
-  org.tensorflow.proto.framework.TensorDescriptionOrBuilder getTensorOrBuilder();
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/MemoryStats.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/MemoryStats.java
deleted file mode 100644
index e351d0be2b0..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/MemoryStats.java
+++ /dev/null
@@ -1,981 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/framework/step_stats.proto
-
-package org.tensorflow.proto.framework;
-
-/**
- * <pre>
- * For memory tracking.
- * </pre>
- *
- * Protobuf type {@code tensorflow.MemoryStats}
- */
-public  final class MemoryStats extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.MemoryStats)
-    MemoryStatsOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use MemoryStats.newBuilder() to construct.
-  private MemoryStats(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private MemoryStats() {
-    persistentTensorAllocIds_ = emptyLongList();
-    devicePersistentTensorAllocIds_ = emptyLongList();
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new MemoryStats();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private MemoryStats(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    int mutable_bitField0_ = 0;
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 8: {
-
-            tempMemorySize_ = input.readInt64();
-            break;
-          }
-          case 16: {
-
-            deviceTempMemorySize_ = input.readInt64();
-            break;
-          }
-          case 24: {
-
-            persistentMemorySize_ = input.readInt64();
-            break;
-          }
-          case 32: {
-
-            devicePersistentMemorySize_ = input.readInt64();
-            break;
-          }
-          case 40: {
-            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
-              persistentTensorAllocIds_ = newLongList();
-              mutable_bitField0_ |= 0x00000001;
-            }
-            persistentTensorAllocIds_.addLong(input.readInt64());
-            break;
-          }
-          case 42: {
-            int length = input.readRawVarint32();
-            int limit = input.pushLimit(length);
-            if (!((mutable_bitField0_ & 0x00000001) != 0) && input.getBytesUntilLimit() > 0) {
-              persistentTensorAllocIds_ = newLongList();
-              mutable_bitField0_ |= 0x00000001;
-            }
-            while (input.getBytesUntilLimit() > 0) {
-              persistentTensorAllocIds_.addLong(input.readInt64());
-            }
-            input.popLimit(limit);
-            break;
-          }
-          case 48: {
-            if (!((mutable_bitField0_ & 0x00000002) != 0)) {
-              devicePersistentTensorAllocIds_ = newLongList();
-              mutable_bitField0_ |= 0x00000002;
-            }
-            devicePersistentTensorAllocIds_.addLong(input.readInt64());
-            break;
-          }
-          case 50: {
-            int length = input.readRawVarint32();
-            int limit = input.pushLimit(length);
-            if (!((mutable_bitField0_ & 0x00000002) != 0) && input.getBytesUntilLimit() > 0) {
-              devicePersistentTensorAllocIds_ = newLongList();
-              mutable_bitField0_ |= 0x00000002;
-            }
-            while (input.getBytesUntilLimit() > 0) {
-              devicePersistentTensorAllocIds_.addLong(input.readInt64());
-            }
-            input.popLimit(limit);
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      if (((mutable_bitField0_ & 0x00000001) != 0)) {
-        persistentTensorAllocIds_.makeImmutable(); // C
-      }
-      if (((mutable_bitField0_ & 0x00000002) != 0)) {
-        devicePersistentTensorAllocIds_.makeImmutable(); // C
-      }
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.framework.StepStatsProtos.internal_static_tensorflow_MemoryStats_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.framework.StepStatsProtos.internal_static_tensorflow_MemoryStats_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.framework.MemoryStats.class, org.tensorflow.proto.framework.MemoryStats.Builder.class);
-  }
-
-  public static final int TEMP_MEMORY_SIZE_FIELD_NUMBER = 1;
-  private long tempMemorySize_;
-  /**
-   * <code>int64 temp_memory_size = 1;</code>
-   */
-  public long getTempMemorySize() {
-    return tempMemorySize_;
-  }
-
-  public static final int PERSISTENT_MEMORY_SIZE_FIELD_NUMBER = 3;
-  private long persistentMemorySize_;
-  /**
-   * <code>int64 persistent_memory_size = 3;</code>
-   */
-  public long getPersistentMemorySize() {
-    return persistentMemorySize_;
-  }
-
-  public static final int PERSISTENT_TENSOR_ALLOC_IDS_FIELD_NUMBER = 5;
-  private com.google.protobuf.Internal.LongList persistentTensorAllocIds_;
-  /**
-   * <code>repeated int64 persistent_tensor_alloc_ids = 5;</code>
-   */
-  public java.util.List<java.lang.Long>
-      getPersistentTensorAllocIdsList() {
-    return persistentTensorAllocIds_;
-  }
-  /**
-   * <code>repeated int64 persistent_tensor_alloc_ids = 5;</code>
-   */
-  public int getPersistentTensorAllocIdsCount() {
-    return persistentTensorAllocIds_.size();
-  }
-  /**
-   * <code>repeated int64 persistent_tensor_alloc_ids = 5;</code>
-   */
-  public long getPersistentTensorAllocIds(int index) {
-    return persistentTensorAllocIds_.getLong(index);
-  }
-  private int persistentTensorAllocIdsMemoizedSerializedSize = -1;
-
-  public static final int DEVICE_TEMP_MEMORY_SIZE_FIELD_NUMBER = 2;
-  private long deviceTempMemorySize_;
-  /**
-   * <code>int64 device_temp_memory_size = 2 [deprecated = true];</code>
-   */
-  @java.lang.Deprecated public long getDeviceTempMemorySize() {
-    return deviceTempMemorySize_;
-  }
-
-  public static final int DEVICE_PERSISTENT_MEMORY_SIZE_FIELD_NUMBER = 4;
-  private long devicePersistentMemorySize_;
-  /**
-   * <code>int64 device_persistent_memory_size = 4 [deprecated = true];</code>
-   */
-  @java.lang.Deprecated public long getDevicePersistentMemorySize() {
-    return devicePersistentMemorySize_;
-  }
-
-  public static final int DEVICE_PERSISTENT_TENSOR_ALLOC_IDS_FIELD_NUMBER = 6;
-  private com.google.protobuf.Internal.LongList devicePersistentTensorAllocIds_;
-  /**
-   * <code>repeated int64 device_persistent_tensor_alloc_ids = 6 [deprecated = true];</code>
-   */
-  @java.lang.Deprecated public java.util.List<java.lang.Long>
-      getDevicePersistentTensorAllocIdsList() {
-    return devicePersistentTensorAllocIds_;
-  }
-  /**
-   * <code>repeated int64 device_persistent_tensor_alloc_ids = 6 [deprecated = true];</code>
-   */
-  @java.lang.Deprecated public int getDevicePersistentTensorAllocIdsCount() {
-    return devicePersistentTensorAllocIds_.size();
-  }
-  /**
-   * <code>repeated int64 device_persistent_tensor_alloc_ids = 6 [deprecated = true];</code>
-   */
-  @java.lang.Deprecated public long getDevicePersistentTensorAllocIds(int index) {
-    return devicePersistentTensorAllocIds_.getLong(index);
-  }
-  private int devicePersistentTensorAllocIdsMemoizedSerializedSize = -1;
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    getSerializedSize();
-    if (tempMemorySize_ != 0L) {
-      output.writeInt64(1, tempMemorySize_);
-    }
-    if (deviceTempMemorySize_ != 0L) {
-      output.writeInt64(2, deviceTempMemorySize_);
-    }
-    if (persistentMemorySize_ != 0L) {
-      output.writeInt64(3, persistentMemorySize_);
-    }
-    if (devicePersistentMemorySize_ != 0L) {
-      output.writeInt64(4, devicePersistentMemorySize_);
-    }
-    if (getPersistentTensorAllocIdsList().size() > 0) {
-      output.writeUInt32NoTag(42);
-      output.writeUInt32NoTag(persistentTensorAllocIdsMemoizedSerializedSize);
-    }
-    for (int i = 0; i < persistentTensorAllocIds_.size(); i++) {
-      output.writeInt64NoTag(persistentTensorAllocIds_.getLong(i));
-    }
-    if (getDevicePersistentTensorAllocIdsList().size() > 0) {
-      output.writeUInt32NoTag(50);
-      output.writeUInt32NoTag(devicePersistentTensorAllocIdsMemoizedSerializedSize);
-    }
-    for (int i = 0; i < devicePersistentTensorAllocIds_.size(); i++) {
-      output.writeInt64NoTag(devicePersistentTensorAllocIds_.getLong(i));
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (tempMemorySize_ != 0L) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt64Size(1, tempMemorySize_);
-    }
-    if (deviceTempMemorySize_ != 0L) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt64Size(2, deviceTempMemorySize_);
-    }
-    if (persistentMemorySize_ != 0L) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt64Size(3, persistentMemorySize_);
-    }
-    if (devicePersistentMemorySize_ != 0L) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt64Size(4, devicePersistentMemorySize_);
-    }
-    {
-      int dataSize = 0;
-      for (int i = 0; i < persistentTensorAllocIds_.size(); i++) {
-        dataSize += com.google.protobuf.CodedOutputStream
-          .computeInt64SizeNoTag(persistentTensorAllocIds_.getLong(i));
-      }
-      size += dataSize;
-      if (!getPersistentTensorAllocIdsList().isEmpty()) {
-        size += 1;
-        size += com.google.protobuf.CodedOutputStream
-            .computeInt32SizeNoTag(dataSize);
-      }
-      persistentTensorAllocIdsMemoizedSerializedSize = dataSize;
-    }
-    {
-      int dataSize = 0;
-      for (int i = 0; i < devicePersistentTensorAllocIds_.size(); i++) {
-        dataSize += com.google.protobuf.CodedOutputStream
-          .computeInt64SizeNoTag(devicePersistentTensorAllocIds_.getLong(i));
-      }
-      size += dataSize;
-      if (!getDevicePersistentTensorAllocIdsList().isEmpty()) {
-        size += 1;
-        size += com.google.protobuf.CodedOutputStream
-            .computeInt32SizeNoTag(dataSize);
-      }
-      devicePersistentTensorAllocIdsMemoizedSerializedSize = dataSize;
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.framework.MemoryStats)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.framework.MemoryStats other = (org.tensorflow.proto.framework.MemoryStats) obj;
-
-    if (getTempMemorySize()
-        != other.getTempMemorySize()) return false;
-    if (getPersistentMemorySize()
-        != other.getPersistentMemorySize()) return false;
-    if (!getPersistentTensorAllocIdsList()
-        .equals(other.getPersistentTensorAllocIdsList())) return false;
-    if (getDeviceTempMemorySize()
-        != other.getDeviceTempMemorySize()) return false;
-    if (getDevicePersistentMemorySize()
-        != other.getDevicePersistentMemorySize()) return false;
-    if (!getDevicePersistentTensorAllocIdsList()
-        .equals(other.getDevicePersistentTensorAllocIdsList())) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    hash = (37 * hash) + TEMP_MEMORY_SIZE_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-        getTempMemorySize());
-    hash = (37 * hash) + PERSISTENT_MEMORY_SIZE_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-        getPersistentMemorySize());
-    if (getPersistentTensorAllocIdsCount() > 0) {
-      hash = (37 * hash) + PERSISTENT_TENSOR_ALLOC_IDS_FIELD_NUMBER;
-      hash = (53 * hash) + getPersistentTensorAllocIdsList().hashCode();
-    }
-    hash = (37 * hash) + DEVICE_TEMP_MEMORY_SIZE_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-        getDeviceTempMemorySize());
-    hash = (37 * hash) + DEVICE_PERSISTENT_MEMORY_SIZE_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-        getDevicePersistentMemorySize());
-    if (getDevicePersistentTensorAllocIdsCount() > 0) {
-      hash = (37 * hash) + DEVICE_PERSISTENT_TENSOR_ALLOC_IDS_FIELD_NUMBER;
-      hash = (53 * hash) + getDevicePersistentTensorAllocIdsList().hashCode();
-    }
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.framework.MemoryStats parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.MemoryStats parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.MemoryStats parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.MemoryStats parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.MemoryStats parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.MemoryStats parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.MemoryStats parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.MemoryStats parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.MemoryStats parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.MemoryStats parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.MemoryStats parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.MemoryStats parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.framework.MemoryStats prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * <pre>
-   * For memory tracking.
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.MemoryStats}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.MemoryStats)
-      org.tensorflow.proto.framework.MemoryStatsOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.StepStatsProtos.internal_static_tensorflow_MemoryStats_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.StepStatsProtos.internal_static_tensorflow_MemoryStats_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.MemoryStats.class, org.tensorflow.proto.framework.MemoryStats.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.framework.MemoryStats.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      tempMemorySize_ = 0L;
-
-      persistentMemorySize_ = 0L;
-
-      persistentTensorAllocIds_ = emptyLongList();
-      bitField0_ = (bitField0_ & ~0x00000001);
-      deviceTempMemorySize_ = 0L;
-
-      devicePersistentMemorySize_ = 0L;
-
-      devicePersistentTensorAllocIds_ = emptyLongList();
-      bitField0_ = (bitField0_ & ~0x00000002);
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.framework.StepStatsProtos.internal_static_tensorflow_MemoryStats_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.MemoryStats getDefaultInstanceForType() {
-      return org.tensorflow.proto.framework.MemoryStats.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.MemoryStats build() {
-      org.tensorflow.proto.framework.MemoryStats result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.MemoryStats buildPartial() {
-      org.tensorflow.proto.framework.MemoryStats result = new org.tensorflow.proto.framework.MemoryStats(this);
-      int from_bitField0_ = bitField0_;
-      result.tempMemorySize_ = tempMemorySize_;
-      result.persistentMemorySize_ = persistentMemorySize_;
-      if (((bitField0_ & 0x00000001) != 0)) {
-        persistentTensorAllocIds_.makeImmutable();
-        bitField0_ = (bitField0_ & ~0x00000001);
-      }
-      result.persistentTensorAllocIds_ = persistentTensorAllocIds_;
-      result.deviceTempMemorySize_ = deviceTempMemorySize_;
-      result.devicePersistentMemorySize_ = devicePersistentMemorySize_;
-      if (((bitField0_ & 0x00000002) != 0)) {
-        devicePersistentTensorAllocIds_.makeImmutable();
-        bitField0_ = (bitField0_ & ~0x00000002);
-      }
-      result.devicePersistentTensorAllocIds_ = devicePersistentTensorAllocIds_;
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.framework.MemoryStats) {
-        return mergeFrom((org.tensorflow.proto.framework.MemoryStats)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.framework.MemoryStats other) {
-      if (other == org.tensorflow.proto.framework.MemoryStats.getDefaultInstance()) return this;
-      if (other.getTempMemorySize() != 0L) {
-        setTempMemorySize(other.getTempMemorySize());
-      }
-      if (other.getPersistentMemorySize() != 0L) {
-        setPersistentMemorySize(other.getPersistentMemorySize());
-      }
-      if (!other.persistentTensorAllocIds_.isEmpty()) {
-        if (persistentTensorAllocIds_.isEmpty()) {
-          persistentTensorAllocIds_ = other.persistentTensorAllocIds_;
-          bitField0_ = (bitField0_ & ~0x00000001);
-        } else {
-          ensurePersistentTensorAllocIdsIsMutable();
-          persistentTensorAllocIds_.addAll(other.persistentTensorAllocIds_);
-        }
-        onChanged();
-      }
-      if (other.getDeviceTempMemorySize() != 0L) {
-        setDeviceTempMemorySize(other.getDeviceTempMemorySize());
-      }
-      if (other.getDevicePersistentMemorySize() != 0L) {
-        setDevicePersistentMemorySize(other.getDevicePersistentMemorySize());
-      }
-      if (!other.devicePersistentTensorAllocIds_.isEmpty()) {
-        if (devicePersistentTensorAllocIds_.isEmpty()) {
-          devicePersistentTensorAllocIds_ = other.devicePersistentTensorAllocIds_;
-          bitField0_ = (bitField0_ & ~0x00000002);
-        } else {
-          ensureDevicePersistentTensorAllocIdsIsMutable();
-          devicePersistentTensorAllocIds_.addAll(other.devicePersistentTensorAllocIds_);
-        }
-        onChanged();
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.framework.MemoryStats parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.framework.MemoryStats) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-    private int bitField0_;
-
-    private long tempMemorySize_ ;
-    /**
-     * <code>int64 temp_memory_size = 1;</code>
-     */
-    public long getTempMemorySize() {
-      return tempMemorySize_;
-    }
-    /**
-     * <code>int64 temp_memory_size = 1;</code>
-     */
-    public Builder setTempMemorySize(long value) {
-      
-      tempMemorySize_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>int64 temp_memory_size = 1;</code>
-     */
-    public Builder clearTempMemorySize() {
-      
-      tempMemorySize_ = 0L;
-      onChanged();
-      return this;
-    }
-
-    private long persistentMemorySize_ ;
-    /**
-     * <code>int64 persistent_memory_size = 3;</code>
-     */
-    public long getPersistentMemorySize() {
-      return persistentMemorySize_;
-    }
-    /**
-     * <code>int64 persistent_memory_size = 3;</code>
-     */
-    public Builder setPersistentMemorySize(long value) {
-      
-      persistentMemorySize_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>int64 persistent_memory_size = 3;</code>
-     */
-    public Builder clearPersistentMemorySize() {
-      
-      persistentMemorySize_ = 0L;
-      onChanged();
-      return this;
-    }
-
-    private com.google.protobuf.Internal.LongList persistentTensorAllocIds_ = emptyLongList();
-    private void ensurePersistentTensorAllocIdsIsMutable() {
-      if (!((bitField0_ & 0x00000001) != 0)) {
-        persistentTensorAllocIds_ = mutableCopy(persistentTensorAllocIds_);
-        bitField0_ |= 0x00000001;
-       }
-    }
-    /**
-     * <code>repeated int64 persistent_tensor_alloc_ids = 5;</code>
-     */
-    public java.util.List<java.lang.Long>
-        getPersistentTensorAllocIdsList() {
-      return ((bitField0_ & 0x00000001) != 0) ?
-               java.util.Collections.unmodifiableList(persistentTensorAllocIds_) : persistentTensorAllocIds_;
-    }
-    /**
-     * <code>repeated int64 persistent_tensor_alloc_ids = 5;</code>
-     */
-    public int getPersistentTensorAllocIdsCount() {
-      return persistentTensorAllocIds_.size();
-    }
-    /**
-     * <code>repeated int64 persistent_tensor_alloc_ids = 5;</code>
-     */
-    public long getPersistentTensorAllocIds(int index) {
-      return persistentTensorAllocIds_.getLong(index);
-    }
-    /**
-     * <code>repeated int64 persistent_tensor_alloc_ids = 5;</code>
-     */
-    public Builder setPersistentTensorAllocIds(
-        int index, long value) {
-      ensurePersistentTensorAllocIdsIsMutable();
-      persistentTensorAllocIds_.setLong(index, value);
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>repeated int64 persistent_tensor_alloc_ids = 5;</code>
-     */
-    public Builder addPersistentTensorAllocIds(long value) {
-      ensurePersistentTensorAllocIdsIsMutable();
-      persistentTensorAllocIds_.addLong(value);
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>repeated int64 persistent_tensor_alloc_ids = 5;</code>
-     */
-    public Builder addAllPersistentTensorAllocIds(
-        java.lang.Iterable<? extends java.lang.Long> values) {
-      ensurePersistentTensorAllocIdsIsMutable();
-      com.google.protobuf.AbstractMessageLite.Builder.addAll(
-          values, persistentTensorAllocIds_);
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>repeated int64 persistent_tensor_alloc_ids = 5;</code>
-     */
-    public Builder clearPersistentTensorAllocIds() {
-      persistentTensorAllocIds_ = emptyLongList();
-      bitField0_ = (bitField0_ & ~0x00000001);
-      onChanged();
-      return this;
-    }
-
-    private long deviceTempMemorySize_ ;
-    /**
-     * <code>int64 device_temp_memory_size = 2 [deprecated = true];</code>
-     */
-    @java.lang.Deprecated public long getDeviceTempMemorySize() {
-      return deviceTempMemorySize_;
-    }
-    /**
-     * <code>int64 device_temp_memory_size = 2 [deprecated = true];</code>
-     */
-    @java.lang.Deprecated public Builder setDeviceTempMemorySize(long value) {
-      
-      deviceTempMemorySize_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>int64 device_temp_memory_size = 2 [deprecated = true];</code>
-     */
-    @java.lang.Deprecated public Builder clearDeviceTempMemorySize() {
-      
-      deviceTempMemorySize_ = 0L;
-      onChanged();
-      return this;
-    }
-
-    private long devicePersistentMemorySize_ ;
-    /**
-     * <code>int64 device_persistent_memory_size = 4 [deprecated = true];</code>
-     */
-    @java.lang.Deprecated public long getDevicePersistentMemorySize() {
-      return devicePersistentMemorySize_;
-    }
-    /**
-     * <code>int64 device_persistent_memory_size = 4 [deprecated = true];</code>
-     */
-    @java.lang.Deprecated public Builder setDevicePersistentMemorySize(long value) {
-      
-      devicePersistentMemorySize_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>int64 device_persistent_memory_size = 4 [deprecated = true];</code>
-     */
-    @java.lang.Deprecated public Builder clearDevicePersistentMemorySize() {
-      
-      devicePersistentMemorySize_ = 0L;
-      onChanged();
-      return this;
-    }
-
-    private com.google.protobuf.Internal.LongList devicePersistentTensorAllocIds_ = emptyLongList();
-    private void ensureDevicePersistentTensorAllocIdsIsMutable() {
-      if (!((bitField0_ & 0x00000002) != 0)) {
-        devicePersistentTensorAllocIds_ = mutableCopy(devicePersistentTensorAllocIds_);
-        bitField0_ |= 0x00000002;
-       }
-    }
-    /**
-     * <code>repeated int64 device_persistent_tensor_alloc_ids = 6 [deprecated = true];</code>
-     */
-    @java.lang.Deprecated public java.util.List<java.lang.Long>
-        getDevicePersistentTensorAllocIdsList() {
-      return ((bitField0_ & 0x00000002) != 0) ?
-               java.util.Collections.unmodifiableList(devicePersistentTensorAllocIds_) : devicePersistentTensorAllocIds_;
-    }
-    /**
-     * <code>repeated int64 device_persistent_tensor_alloc_ids = 6 [deprecated = true];</code>
-     */
-    @java.lang.Deprecated public int getDevicePersistentTensorAllocIdsCount() {
-      return devicePersistentTensorAllocIds_.size();
-    }
-    /**
-     * <code>repeated int64 device_persistent_tensor_alloc_ids = 6 [deprecated = true];</code>
-     */
-    @java.lang.Deprecated public long getDevicePersistentTensorAllocIds(int index) {
-      return devicePersistentTensorAllocIds_.getLong(index);
-    }
-    /**
-     * <code>repeated int64 device_persistent_tensor_alloc_ids = 6 [deprecated = true];</code>
-     */
-    @java.lang.Deprecated public Builder setDevicePersistentTensorAllocIds(
-        int index, long value) {
-      ensureDevicePersistentTensorAllocIdsIsMutable();
-      devicePersistentTensorAllocIds_.setLong(index, value);
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>repeated int64 device_persistent_tensor_alloc_ids = 6 [deprecated = true];</code>
-     */
-    @java.lang.Deprecated public Builder addDevicePersistentTensorAllocIds(long value) {
-      ensureDevicePersistentTensorAllocIdsIsMutable();
-      devicePersistentTensorAllocIds_.addLong(value);
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>repeated int64 device_persistent_tensor_alloc_ids = 6 [deprecated = true];</code>
-     */
-    @java.lang.Deprecated public Builder addAllDevicePersistentTensorAllocIds(
-        java.lang.Iterable<? extends java.lang.Long> values) {
-      ensureDevicePersistentTensorAllocIdsIsMutable();
-      com.google.protobuf.AbstractMessageLite.Builder.addAll(
-          values, devicePersistentTensorAllocIds_);
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>repeated int64 device_persistent_tensor_alloc_ids = 6 [deprecated = true];</code>
-     */
-    @java.lang.Deprecated public Builder clearDevicePersistentTensorAllocIds() {
-      devicePersistentTensorAllocIds_ = emptyLongList();
-      bitField0_ = (bitField0_ & ~0x00000002);
-      onChanged();
-      return this;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.MemoryStats)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.MemoryStats)
-  private static final org.tensorflow.proto.framework.MemoryStats DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.MemoryStats();
-  }
-
-  public static org.tensorflow.proto.framework.MemoryStats getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<MemoryStats>
-      PARSER = new com.google.protobuf.AbstractParser<MemoryStats>() {
-    @java.lang.Override
-    public MemoryStats parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new MemoryStats(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<MemoryStats> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<MemoryStats> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.framework.MemoryStats getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/MemoryStatsOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/MemoryStatsOrBuilder.java
deleted file mode 100644
index db3c896352c..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/MemoryStatsOrBuilder.java
+++ /dev/null
@@ -1,55 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/framework/step_stats.proto
-
-package org.tensorflow.proto.framework;
-
-public interface MemoryStatsOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.MemoryStats)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <code>int64 temp_memory_size = 1;</code>
-   */
-  long getTempMemorySize();
-
-  /**
-   * <code>int64 persistent_memory_size = 3;</code>
-   */
-  long getPersistentMemorySize();
-
-  /**
-   * <code>repeated int64 persistent_tensor_alloc_ids = 5;</code>
-   */
-  java.util.List<java.lang.Long> getPersistentTensorAllocIdsList();
-  /**
-   * <code>repeated int64 persistent_tensor_alloc_ids = 5;</code>
-   */
-  int getPersistentTensorAllocIdsCount();
-  /**
-   * <code>repeated int64 persistent_tensor_alloc_ids = 5;</code>
-   */
-  long getPersistentTensorAllocIds(int index);
-
-  /**
-   * <code>int64 device_temp_memory_size = 2 [deprecated = true];</code>
-   */
-  @java.lang.Deprecated long getDeviceTempMemorySize();
-
-  /**
-   * <code>int64 device_persistent_memory_size = 4 [deprecated = true];</code>
-   */
-  @java.lang.Deprecated long getDevicePersistentMemorySize();
-
-  /**
-   * <code>repeated int64 device_persistent_tensor_alloc_ids = 6 [deprecated = true];</code>
-   */
-  @java.lang.Deprecated java.util.List<java.lang.Long> getDevicePersistentTensorAllocIdsList();
-  /**
-   * <code>repeated int64 device_persistent_tensor_alloc_ids = 6 [deprecated = true];</code>
-   */
-  @java.lang.Deprecated int getDevicePersistentTensorAllocIdsCount();
-  /**
-   * <code>repeated int64 device_persistent_tensor_alloc_ids = 6 [deprecated = true];</code>
-   */
-  @java.lang.Deprecated long getDevicePersistentTensorAllocIds(int index);
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/MetaGraphDef.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/MetaGraphDef.java
deleted file mode 100644
index b071cf84bf8..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/MetaGraphDef.java
+++ /dev/null
@@ -1,4702 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/meta_graph.proto
-
-package org.tensorflow.proto.framework;
-
-/**
- * <pre>
- * NOTE: This protocol buffer is evolving, and will go through revisions in the
- * coming months.
- * Protocol buffer containing the following which are necessary to restart
- * training, run inference. It can be used to serialize/de-serialize memory
- * objects necessary for running computation in a graph when crossing the
- * process boundary. It can be used for long term storage of graphs,
- * cross-language execution of graphs, etc.
- *   MetaInfoDef
- *   GraphDef
- *   SaverDef
- *   CollectionDef
- *   TensorInfo
- *   SignatureDef
- * </pre>
- *
- * Protobuf type {@code tensorflow.MetaGraphDef}
- */
-public  final class MetaGraphDef extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.MetaGraphDef)
-    MetaGraphDefOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use MetaGraphDef.newBuilder() to construct.
-  private MetaGraphDef(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private MetaGraphDef() {
-    assetFileDef_ = java.util.Collections.emptyList();
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new MetaGraphDef();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private MetaGraphDef(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    int mutable_bitField0_ = 0;
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 10: {
-            org.tensorflow.proto.framework.MetaGraphDef.MetaInfoDef.Builder subBuilder = null;
-            if (metaInfoDef_ != null) {
-              subBuilder = metaInfoDef_.toBuilder();
-            }
-            metaInfoDef_ = input.readMessage(org.tensorflow.proto.framework.MetaGraphDef.MetaInfoDef.parser(), extensionRegistry);
-            if (subBuilder != null) {
-              subBuilder.mergeFrom(metaInfoDef_);
-              metaInfoDef_ = subBuilder.buildPartial();
-            }
-
-            break;
-          }
-          case 18: {
-            org.tensorflow.proto.framework.GraphDef.Builder subBuilder = null;
-            if (graphDef_ != null) {
-              subBuilder = graphDef_.toBuilder();
-            }
-            graphDef_ = input.readMessage(org.tensorflow.proto.framework.GraphDef.parser(), extensionRegistry);
-            if (subBuilder != null) {
-              subBuilder.mergeFrom(graphDef_);
-              graphDef_ = subBuilder.buildPartial();
-            }
-
-            break;
-          }
-          case 26: {
-            org.tensorflow.proto.util.SaverDef.Builder subBuilder = null;
-            if (saverDef_ != null) {
-              subBuilder = saverDef_.toBuilder();
-            }
-            saverDef_ = input.readMessage(org.tensorflow.proto.util.SaverDef.parser(), extensionRegistry);
-            if (subBuilder != null) {
-              subBuilder.mergeFrom(saverDef_);
-              saverDef_ = subBuilder.buildPartial();
-            }
-
-            break;
-          }
-          case 34: {
-            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
-              collectionDef_ = com.google.protobuf.MapField.newMapField(
-                  CollectionDefDefaultEntryHolder.defaultEntry);
-              mutable_bitField0_ |= 0x00000001;
-            }
-            com.google.protobuf.MapEntry<java.lang.String, org.tensorflow.proto.framework.CollectionDef>
-            collectionDef__ = input.readMessage(
-                CollectionDefDefaultEntryHolder.defaultEntry.getParserForType(), extensionRegistry);
-            collectionDef_.getMutableMap().put(
-                collectionDef__.getKey(), collectionDef__.getValue());
-            break;
-          }
-          case 42: {
-            if (!((mutable_bitField0_ & 0x00000002) != 0)) {
-              signatureDef_ = com.google.protobuf.MapField.newMapField(
-                  SignatureDefDefaultEntryHolder.defaultEntry);
-              mutable_bitField0_ |= 0x00000002;
-            }
-            com.google.protobuf.MapEntry<java.lang.String, org.tensorflow.proto.framework.SignatureDef>
-            signatureDef__ = input.readMessage(
-                SignatureDefDefaultEntryHolder.defaultEntry.getParserForType(), extensionRegistry);
-            signatureDef_.getMutableMap().put(
-                signatureDef__.getKey(), signatureDef__.getValue());
-            break;
-          }
-          case 50: {
-            if (!((mutable_bitField0_ & 0x00000004) != 0)) {
-              assetFileDef_ = new java.util.ArrayList<org.tensorflow.proto.framework.AssetFileDef>();
-              mutable_bitField0_ |= 0x00000004;
-            }
-            assetFileDef_.add(
-                input.readMessage(org.tensorflow.proto.framework.AssetFileDef.parser(), extensionRegistry));
-            break;
-          }
-          case 58: {
-            org.tensorflow.proto.framework.SavedObjectGraph.Builder subBuilder = null;
-            if (objectGraphDef_ != null) {
-              subBuilder = objectGraphDef_.toBuilder();
-            }
-            objectGraphDef_ = input.readMessage(org.tensorflow.proto.framework.SavedObjectGraph.parser(), extensionRegistry);
-            if (subBuilder != null) {
-              subBuilder.mergeFrom(objectGraphDef_);
-              objectGraphDef_ = subBuilder.buildPartial();
-            }
-
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      if (((mutable_bitField0_ & 0x00000004) != 0)) {
-        assetFileDef_ = java.util.Collections.unmodifiableList(assetFileDef_);
-      }
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.framework.MetaGraphProtos.internal_static_tensorflow_MetaGraphDef_descriptor;
-  }
-
-  @SuppressWarnings({"rawtypes"})
-  @java.lang.Override
-  protected com.google.protobuf.MapField internalGetMapField(
-      int number) {
-    switch (number) {
-      case 4:
-        return internalGetCollectionDef();
-      case 5:
-        return internalGetSignatureDef();
-      default:
-        throw new RuntimeException(
-            "Invalid map field number: " + number);
-    }
-  }
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.framework.MetaGraphProtos.internal_static_tensorflow_MetaGraphDef_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.framework.MetaGraphDef.class, org.tensorflow.proto.framework.MetaGraphDef.Builder.class);
-  }
-
-  public interface MetaInfoDefOrBuilder extends
-      // @@protoc_insertion_point(interface_extends:tensorflow.MetaGraphDef.MetaInfoDef)
-      com.google.protobuf.MessageOrBuilder {
-
-    /**
-     * <pre>
-     * User specified Version string. Can be the name of the model and revision,
-     * steps this model has been trained to, etc.
-     * </pre>
-     *
-     * <code>string meta_graph_version = 1;</code>
-     */
-    java.lang.String getMetaGraphVersion();
-    /**
-     * <pre>
-     * User specified Version string. Can be the name of the model and revision,
-     * steps this model has been trained to, etc.
-     * </pre>
-     *
-     * <code>string meta_graph_version = 1;</code>
-     */
-    com.google.protobuf.ByteString
-        getMetaGraphVersionBytes();
-
-    /**
-     * <pre>
-     * A copy of the OpDefs used by the producer of this graph_def.
-     * Descriptions and Ops not used in graph_def are stripped out.
-     * </pre>
-     *
-     * <code>.tensorflow.OpList stripped_op_list = 2;</code>
-     */
-    boolean hasStrippedOpList();
-    /**
-     * <pre>
-     * A copy of the OpDefs used by the producer of this graph_def.
-     * Descriptions and Ops not used in graph_def are stripped out.
-     * </pre>
-     *
-     * <code>.tensorflow.OpList stripped_op_list = 2;</code>
-     */
-    org.tensorflow.proto.framework.OpList getStrippedOpList();
-    /**
-     * <pre>
-     * A copy of the OpDefs used by the producer of this graph_def.
-     * Descriptions and Ops not used in graph_def are stripped out.
-     * </pre>
-     *
-     * <code>.tensorflow.OpList stripped_op_list = 2;</code>
-     */
-    org.tensorflow.proto.framework.OpListOrBuilder getStrippedOpListOrBuilder();
-
-    /**
-     * <pre>
-     * A serialized protobuf. Can be the time this meta graph is created, or
-     * modified, or name of the model.
-     * </pre>
-     *
-     * <code>.google.protobuf.Any any_info = 3;</code>
-     */
-    boolean hasAnyInfo();
-    /**
-     * <pre>
-     * A serialized protobuf. Can be the time this meta graph is created, or
-     * modified, or name of the model.
-     * </pre>
-     *
-     * <code>.google.protobuf.Any any_info = 3;</code>
-     */
-    com.google.protobuf.Any getAnyInfo();
-    /**
-     * <pre>
-     * A serialized protobuf. Can be the time this meta graph is created, or
-     * modified, or name of the model.
-     * </pre>
-     *
-     * <code>.google.protobuf.Any any_info = 3;</code>
-     */
-    com.google.protobuf.AnyOrBuilder getAnyInfoOrBuilder();
-
-    /**
-     * <pre>
-     * User supplied tag(s) on the meta_graph and included graph_def.
-     * MetaGraphDefs should be tagged with their capabilities or use-cases.
-     * Examples: "train", "serve", "gpu", "tpu", etc.
-     * These tags enable loaders to access the MetaGraph(s) appropriate for a
-     * specific use-case or runtime environment.
-     * </pre>
-     *
-     * <code>repeated string tags = 4;</code>
-     */
-    java.util.List<java.lang.String>
-        getTagsList();
-    /**
-     * <pre>
-     * User supplied tag(s) on the meta_graph and included graph_def.
-     * MetaGraphDefs should be tagged with their capabilities or use-cases.
-     * Examples: "train", "serve", "gpu", "tpu", etc.
-     * These tags enable loaders to access the MetaGraph(s) appropriate for a
-     * specific use-case or runtime environment.
-     * </pre>
-     *
-     * <code>repeated string tags = 4;</code>
-     */
-    int getTagsCount();
-    /**
-     * <pre>
-     * User supplied tag(s) on the meta_graph and included graph_def.
-     * MetaGraphDefs should be tagged with their capabilities or use-cases.
-     * Examples: "train", "serve", "gpu", "tpu", etc.
-     * These tags enable loaders to access the MetaGraph(s) appropriate for a
-     * specific use-case or runtime environment.
-     * </pre>
-     *
-     * <code>repeated string tags = 4;</code>
-     */
-    java.lang.String getTags(int index);
-    /**
-     * <pre>
-     * User supplied tag(s) on the meta_graph and included graph_def.
-     * MetaGraphDefs should be tagged with their capabilities or use-cases.
-     * Examples: "train", "serve", "gpu", "tpu", etc.
-     * These tags enable loaders to access the MetaGraph(s) appropriate for a
-     * specific use-case or runtime environment.
-     * </pre>
-     *
-     * <code>repeated string tags = 4;</code>
-     */
-    com.google.protobuf.ByteString
-        getTagsBytes(int index);
-
-    /**
-     * <pre>
-     * The __version__ string of the tensorflow build used to write this graph.
-     * This will be populated by the framework, which will overwrite any user
-     * supplied value.
-     * </pre>
-     *
-     * <code>string tensorflow_version = 5;</code>
-     */
-    java.lang.String getTensorflowVersion();
-    /**
-     * <pre>
-     * The __version__ string of the tensorflow build used to write this graph.
-     * This will be populated by the framework, which will overwrite any user
-     * supplied value.
-     * </pre>
-     *
-     * <code>string tensorflow_version = 5;</code>
-     */
-    com.google.protobuf.ByteString
-        getTensorflowVersionBytes();
-
-    /**
-     * <pre>
-     * The __git_version__ string of the tensorflow build used to write this
-     * graph. This will be populated by the framework, which will overwrite any
-     * user supplied value.
-     * </pre>
-     *
-     * <code>string tensorflow_git_version = 6;</code>
-     */
-    java.lang.String getTensorflowGitVersion();
-    /**
-     * <pre>
-     * The __git_version__ string of the tensorflow build used to write this
-     * graph. This will be populated by the framework, which will overwrite any
-     * user supplied value.
-     * </pre>
-     *
-     * <code>string tensorflow_git_version = 6;</code>
-     */
-    com.google.protobuf.ByteString
-        getTensorflowGitVersionBytes();
-
-    /**
-     * <pre>
-     * A flag to denote whether default-valued attrs have been stripped from
-     * the nodes in this graph_def.
-     * </pre>
-     *
-     * <code>bool stripped_default_attrs = 7;</code>
-     */
-    boolean getStrippedDefaultAttrs();
-
-    /**
-     * <pre>
-     * FunctionDef name to aliases mapping.
-     * </pre>
-     *
-     * <code>map&lt;string, string&gt; function_aliases = 8;</code>
-     */
-    int getFunctionAliasesCount();
-    /**
-     * <pre>
-     * FunctionDef name to aliases mapping.
-     * </pre>
-     *
-     * <code>map&lt;string, string&gt; function_aliases = 8;</code>
-     */
-    boolean containsFunctionAliases(
-        java.lang.String key);
-    /**
-     * Use {@link #getFunctionAliasesMap()} instead.
-     */
-    @java.lang.Deprecated
-    java.util.Map<java.lang.String, java.lang.String>
-    getFunctionAliases();
-    /**
-     * <pre>
-     * FunctionDef name to aliases mapping.
-     * </pre>
-     *
-     * <code>map&lt;string, string&gt; function_aliases = 8;</code>
-     */
-    java.util.Map<java.lang.String, java.lang.String>
-    getFunctionAliasesMap();
-    /**
-     * <pre>
-     * FunctionDef name to aliases mapping.
-     * </pre>
-     *
-     * <code>map&lt;string, string&gt; function_aliases = 8;</code>
-     */
-
-    java.lang.String getFunctionAliasesOrDefault(
-        java.lang.String key,
-        java.lang.String defaultValue);
-    /**
-     * <pre>
-     * FunctionDef name to aliases mapping.
-     * </pre>
-     *
-     * <code>map&lt;string, string&gt; function_aliases = 8;</code>
-     */
-
-    java.lang.String getFunctionAliasesOrThrow(
-        java.lang.String key);
-  }
-  /**
-   * <pre>
-   * Meta information regarding the graph to be exported.  To be used by users
-   * of this protocol buffer to encode information regarding their meta graph.
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.MetaGraphDef.MetaInfoDef}
-   */
-  public  static final class MetaInfoDef extends
-      com.google.protobuf.GeneratedMessageV3 implements
-      // @@protoc_insertion_point(message_implements:tensorflow.MetaGraphDef.MetaInfoDef)
-      MetaInfoDefOrBuilder {
-  private static final long serialVersionUID = 0L;
-    // Use MetaInfoDef.newBuilder() to construct.
-    private MetaInfoDef(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-      super(builder);
-    }
-    private MetaInfoDef() {
-      metaGraphVersion_ = "";
-      tags_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-      tensorflowVersion_ = "";
-      tensorflowGitVersion_ = "";
-    }
-
-    @java.lang.Override
-    @SuppressWarnings({"unused"})
-    protected java.lang.Object newInstance(
-        UnusedPrivateParameter unused) {
-      return new MetaInfoDef();
-    }
-
-    @java.lang.Override
-    public final com.google.protobuf.UnknownFieldSet
-    getUnknownFields() {
-      return this.unknownFields;
-    }
-    private MetaInfoDef(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      this();
-      if (extensionRegistry == null) {
-        throw new java.lang.NullPointerException();
-      }
-      int mutable_bitField0_ = 0;
-      com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-          com.google.protobuf.UnknownFieldSet.newBuilder();
-      try {
-        boolean done = false;
-        while (!done) {
-          int tag = input.readTag();
-          switch (tag) {
-            case 0:
-              done = true;
-              break;
-            case 10: {
-              java.lang.String s = input.readStringRequireUtf8();
-
-              metaGraphVersion_ = s;
-              break;
-            }
-            case 18: {
-              org.tensorflow.proto.framework.OpList.Builder subBuilder = null;
-              if (strippedOpList_ != null) {
-                subBuilder = strippedOpList_.toBuilder();
-              }
-              strippedOpList_ = input.readMessage(org.tensorflow.proto.framework.OpList.parser(), extensionRegistry);
-              if (subBuilder != null) {
-                subBuilder.mergeFrom(strippedOpList_);
-                strippedOpList_ = subBuilder.buildPartial();
-              }
-
-              break;
-            }
-            case 26: {
-              com.google.protobuf.Any.Builder subBuilder = null;
-              if (anyInfo_ != null) {
-                subBuilder = anyInfo_.toBuilder();
-              }
-              anyInfo_ = input.readMessage(com.google.protobuf.Any.parser(), extensionRegistry);
-              if (subBuilder != null) {
-                subBuilder.mergeFrom(anyInfo_);
-                anyInfo_ = subBuilder.buildPartial();
-              }
-
-              break;
-            }
-            case 34: {
-              java.lang.String s = input.readStringRequireUtf8();
-              if (!((mutable_bitField0_ & 0x00000001) != 0)) {
-                tags_ = new com.google.protobuf.LazyStringArrayList();
-                mutable_bitField0_ |= 0x00000001;
-              }
-              tags_.add(s);
-              break;
-            }
-            case 42: {
-              java.lang.String s = input.readStringRequireUtf8();
-
-              tensorflowVersion_ = s;
-              break;
-            }
-            case 50: {
-              java.lang.String s = input.readStringRequireUtf8();
-
-              tensorflowGitVersion_ = s;
-              break;
-            }
-            case 56: {
-
-              strippedDefaultAttrs_ = input.readBool();
-              break;
-            }
-            case 66: {
-              if (!((mutable_bitField0_ & 0x00000002) != 0)) {
-                functionAliases_ = com.google.protobuf.MapField.newMapField(
-                    FunctionAliasesDefaultEntryHolder.defaultEntry);
-                mutable_bitField0_ |= 0x00000002;
-              }
-              com.google.protobuf.MapEntry<java.lang.String, java.lang.String>
-              functionAliases__ = input.readMessage(
-                  FunctionAliasesDefaultEntryHolder.defaultEntry.getParserForType(), extensionRegistry);
-              functionAliases_.getMutableMap().put(
-                  functionAliases__.getKey(), functionAliases__.getValue());
-              break;
-            }
-            default: {
-              if (!parseUnknownField(
-                  input, unknownFields, extensionRegistry, tag)) {
-                done = true;
-              }
-              break;
-            }
-          }
-        }
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        throw e.setUnfinishedMessage(this);
-      } catch (java.io.IOException e) {
-        throw new com.google.protobuf.InvalidProtocolBufferException(
-            e).setUnfinishedMessage(this);
-      } finally {
-        if (((mutable_bitField0_ & 0x00000001) != 0)) {
-          tags_ = tags_.getUnmodifiableView();
-        }
-        this.unknownFields = unknownFields.build();
-        makeExtensionsImmutable();
-      }
-    }
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.MetaGraphProtos.internal_static_tensorflow_MetaGraphDef_MetaInfoDef_descriptor;
-    }
-
-    @SuppressWarnings({"rawtypes"})
-    @java.lang.Override
-    protected com.google.protobuf.MapField internalGetMapField(
-        int number) {
-      switch (number) {
-        case 8:
-          return internalGetFunctionAliases();
-        default:
-          throw new RuntimeException(
-              "Invalid map field number: " + number);
-      }
-    }
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.MetaGraphProtos.internal_static_tensorflow_MetaGraphDef_MetaInfoDef_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.MetaGraphDef.MetaInfoDef.class, org.tensorflow.proto.framework.MetaGraphDef.MetaInfoDef.Builder.class);
-    }
-
-    public static final int META_GRAPH_VERSION_FIELD_NUMBER = 1;
-    private volatile java.lang.Object metaGraphVersion_;
-    /**
-     * <pre>
-     * User specified Version string. Can be the name of the model and revision,
-     * steps this model has been trained to, etc.
-     * </pre>
-     *
-     * <code>string meta_graph_version = 1;</code>
-     */
-    public java.lang.String getMetaGraphVersion() {
-      java.lang.Object ref = metaGraphVersion_;
-      if (ref instanceof java.lang.String) {
-        return (java.lang.String) ref;
-      } else {
-        com.google.protobuf.ByteString bs = 
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        metaGraphVersion_ = s;
-        return s;
-      }
-    }
-    /**
-     * <pre>
-     * User specified Version string. Can be the name of the model and revision,
-     * steps this model has been trained to, etc.
-     * </pre>
-     *
-     * <code>string meta_graph_version = 1;</code>
-     */
-    public com.google.protobuf.ByteString
-        getMetaGraphVersionBytes() {
-      java.lang.Object ref = metaGraphVersion_;
-      if (ref instanceof java.lang.String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        metaGraphVersion_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-
-    public static final int STRIPPED_OP_LIST_FIELD_NUMBER = 2;
-    private org.tensorflow.proto.framework.OpList strippedOpList_;
-    /**
-     * <pre>
-     * A copy of the OpDefs used by the producer of this graph_def.
-     * Descriptions and Ops not used in graph_def are stripped out.
-     * </pre>
-     *
-     * <code>.tensorflow.OpList stripped_op_list = 2;</code>
-     */
-    public boolean hasStrippedOpList() {
-      return strippedOpList_ != null;
-    }
-    /**
-     * <pre>
-     * A copy of the OpDefs used by the producer of this graph_def.
-     * Descriptions and Ops not used in graph_def are stripped out.
-     * </pre>
-     *
-     * <code>.tensorflow.OpList stripped_op_list = 2;</code>
-     */
-    public org.tensorflow.proto.framework.OpList getStrippedOpList() {
-      return strippedOpList_ == null ? org.tensorflow.proto.framework.OpList.getDefaultInstance() : strippedOpList_;
-    }
-    /**
-     * <pre>
-     * A copy of the OpDefs used by the producer of this graph_def.
-     * Descriptions and Ops not used in graph_def are stripped out.
-     * </pre>
-     *
-     * <code>.tensorflow.OpList stripped_op_list = 2;</code>
-     */
-    public org.tensorflow.proto.framework.OpListOrBuilder getStrippedOpListOrBuilder() {
-      return getStrippedOpList();
-    }
-
-    public static final int ANY_INFO_FIELD_NUMBER = 3;
-    private com.google.protobuf.Any anyInfo_;
-    /**
-     * <pre>
-     * A serialized protobuf. Can be the time this meta graph is created, or
-     * modified, or name of the model.
-     * </pre>
-     *
-     * <code>.google.protobuf.Any any_info = 3;</code>
-     */
-    public boolean hasAnyInfo() {
-      return anyInfo_ != null;
-    }
-    /**
-     * <pre>
-     * A serialized protobuf. Can be the time this meta graph is created, or
-     * modified, or name of the model.
-     * </pre>
-     *
-     * <code>.google.protobuf.Any any_info = 3;</code>
-     */
-    public com.google.protobuf.Any getAnyInfo() {
-      return anyInfo_ == null ? com.google.protobuf.Any.getDefaultInstance() : anyInfo_;
-    }
-    /**
-     * <pre>
-     * A serialized protobuf. Can be the time this meta graph is created, or
-     * modified, or name of the model.
-     * </pre>
-     *
-     * <code>.google.protobuf.Any any_info = 3;</code>
-     */
-    public com.google.protobuf.AnyOrBuilder getAnyInfoOrBuilder() {
-      return getAnyInfo();
-    }
-
-    public static final int TAGS_FIELD_NUMBER = 4;
-    private com.google.protobuf.LazyStringList tags_;
-    /**
-     * <pre>
-     * User supplied tag(s) on the meta_graph and included graph_def.
-     * MetaGraphDefs should be tagged with their capabilities or use-cases.
-     * Examples: "train", "serve", "gpu", "tpu", etc.
-     * These tags enable loaders to access the MetaGraph(s) appropriate for a
-     * specific use-case or runtime environment.
-     * </pre>
-     *
-     * <code>repeated string tags = 4;</code>
-     */
-    public com.google.protobuf.ProtocolStringList
-        getTagsList() {
-      return tags_;
-    }
-    /**
-     * <pre>
-     * User supplied tag(s) on the meta_graph and included graph_def.
-     * MetaGraphDefs should be tagged with their capabilities or use-cases.
-     * Examples: "train", "serve", "gpu", "tpu", etc.
-     * These tags enable loaders to access the MetaGraph(s) appropriate for a
-     * specific use-case or runtime environment.
-     * </pre>
-     *
-     * <code>repeated string tags = 4;</code>
-     */
-    public int getTagsCount() {
-      return tags_.size();
-    }
-    /**
-     * <pre>
-     * User supplied tag(s) on the meta_graph and included graph_def.
-     * MetaGraphDefs should be tagged with their capabilities or use-cases.
-     * Examples: "train", "serve", "gpu", "tpu", etc.
-     * These tags enable loaders to access the MetaGraph(s) appropriate for a
-     * specific use-case or runtime environment.
-     * </pre>
-     *
-     * <code>repeated string tags = 4;</code>
-     */
-    public java.lang.String getTags(int index) {
-      return tags_.get(index);
-    }
-    /**
-     * <pre>
-     * User supplied tag(s) on the meta_graph and included graph_def.
-     * MetaGraphDefs should be tagged with their capabilities or use-cases.
-     * Examples: "train", "serve", "gpu", "tpu", etc.
-     * These tags enable loaders to access the MetaGraph(s) appropriate for a
-     * specific use-case or runtime environment.
-     * </pre>
-     *
-     * <code>repeated string tags = 4;</code>
-     */
-    public com.google.protobuf.ByteString
-        getTagsBytes(int index) {
-      return tags_.getByteString(index);
-    }
-
-    public static final int TENSORFLOW_VERSION_FIELD_NUMBER = 5;
-    private volatile java.lang.Object tensorflowVersion_;
-    /**
-     * <pre>
-     * The __version__ string of the tensorflow build used to write this graph.
-     * This will be populated by the framework, which will overwrite any user
-     * supplied value.
-     * </pre>
-     *
-     * <code>string tensorflow_version = 5;</code>
-     */
-    public java.lang.String getTensorflowVersion() {
-      java.lang.Object ref = tensorflowVersion_;
-      if (ref instanceof java.lang.String) {
-        return (java.lang.String) ref;
-      } else {
-        com.google.protobuf.ByteString bs = 
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        tensorflowVersion_ = s;
-        return s;
-      }
-    }
-    /**
-     * <pre>
-     * The __version__ string of the tensorflow build used to write this graph.
-     * This will be populated by the framework, which will overwrite any user
-     * supplied value.
-     * </pre>
-     *
-     * <code>string tensorflow_version = 5;</code>
-     */
-    public com.google.protobuf.ByteString
-        getTensorflowVersionBytes() {
-      java.lang.Object ref = tensorflowVersion_;
-      if (ref instanceof java.lang.String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        tensorflowVersion_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-
-    public static final int TENSORFLOW_GIT_VERSION_FIELD_NUMBER = 6;
-    private volatile java.lang.Object tensorflowGitVersion_;
-    /**
-     * <pre>
-     * The __git_version__ string of the tensorflow build used to write this
-     * graph. This will be populated by the framework, which will overwrite any
-     * user supplied value.
-     * </pre>
-     *
-     * <code>string tensorflow_git_version = 6;</code>
-     */
-    public java.lang.String getTensorflowGitVersion() {
-      java.lang.Object ref = tensorflowGitVersion_;
-      if (ref instanceof java.lang.String) {
-        return (java.lang.String) ref;
-      } else {
-        com.google.protobuf.ByteString bs = 
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        tensorflowGitVersion_ = s;
-        return s;
-      }
-    }
-    /**
-     * <pre>
-     * The __git_version__ string of the tensorflow build used to write this
-     * graph. This will be populated by the framework, which will overwrite any
-     * user supplied value.
-     * </pre>
-     *
-     * <code>string tensorflow_git_version = 6;</code>
-     */
-    public com.google.protobuf.ByteString
-        getTensorflowGitVersionBytes() {
-      java.lang.Object ref = tensorflowGitVersion_;
-      if (ref instanceof java.lang.String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        tensorflowGitVersion_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-
-    public static final int STRIPPED_DEFAULT_ATTRS_FIELD_NUMBER = 7;
-    private boolean strippedDefaultAttrs_;
-    /**
-     * <pre>
-     * A flag to denote whether default-valued attrs have been stripped from
-     * the nodes in this graph_def.
-     * </pre>
-     *
-     * <code>bool stripped_default_attrs = 7;</code>
-     */
-    public boolean getStrippedDefaultAttrs() {
-      return strippedDefaultAttrs_;
-    }
-
-    public static final int FUNCTION_ALIASES_FIELD_NUMBER = 8;
-    private static final class FunctionAliasesDefaultEntryHolder {
-      static final com.google.protobuf.MapEntry<
-          java.lang.String, java.lang.String> defaultEntry =
-              com.google.protobuf.MapEntry
-              .<java.lang.String, java.lang.String>newDefaultInstance(
-                  org.tensorflow.proto.framework.MetaGraphProtos.internal_static_tensorflow_MetaGraphDef_MetaInfoDef_FunctionAliasesEntry_descriptor, 
-                  com.google.protobuf.WireFormat.FieldType.STRING,
-                  "",
-                  com.google.protobuf.WireFormat.FieldType.STRING,
-                  "");
-    }
-    private com.google.protobuf.MapField<
-        java.lang.String, java.lang.String> functionAliases_;
-    private com.google.protobuf.MapField<java.lang.String, java.lang.String>
-    internalGetFunctionAliases() {
-      if (functionAliases_ == null) {
-        return com.google.protobuf.MapField.emptyMapField(
-            FunctionAliasesDefaultEntryHolder.defaultEntry);
-      }
-      return functionAliases_;
-    }
-
-    public int getFunctionAliasesCount() {
-      return internalGetFunctionAliases().getMap().size();
-    }
-    /**
-     * <pre>
-     * FunctionDef name to aliases mapping.
-     * </pre>
-     *
-     * <code>map&lt;string, string&gt; function_aliases = 8;</code>
-     */
-
-    public boolean containsFunctionAliases(
-        java.lang.String key) {
-      if (key == null) { throw new java.lang.NullPointerException(); }
-      return internalGetFunctionAliases().getMap().containsKey(key);
-    }
-    /**
-     * Use {@link #getFunctionAliasesMap()} instead.
-     */
-    @java.lang.Deprecated
-    public java.util.Map<java.lang.String, java.lang.String> getFunctionAliases() {
-      return getFunctionAliasesMap();
-    }
-    /**
-     * <pre>
-     * FunctionDef name to aliases mapping.
-     * </pre>
-     *
-     * <code>map&lt;string, string&gt; function_aliases = 8;</code>
-     */
-
-    public java.util.Map<java.lang.String, java.lang.String> getFunctionAliasesMap() {
-      return internalGetFunctionAliases().getMap();
-    }
-    /**
-     * <pre>
-     * FunctionDef name to aliases mapping.
-     * </pre>
-     *
-     * <code>map&lt;string, string&gt; function_aliases = 8;</code>
-     */
-
-    public java.lang.String getFunctionAliasesOrDefault(
-        java.lang.String key,
-        java.lang.String defaultValue) {
-      if (key == null) { throw new java.lang.NullPointerException(); }
-      java.util.Map<java.lang.String, java.lang.String> map =
-          internalGetFunctionAliases().getMap();
-      return map.containsKey(key) ? map.get(key) : defaultValue;
-    }
-    /**
-     * <pre>
-     * FunctionDef name to aliases mapping.
-     * </pre>
-     *
-     * <code>map&lt;string, string&gt; function_aliases = 8;</code>
-     */
-
-    public java.lang.String getFunctionAliasesOrThrow(
-        java.lang.String key) {
-      if (key == null) { throw new java.lang.NullPointerException(); }
-      java.util.Map<java.lang.String, java.lang.String> map =
-          internalGetFunctionAliases().getMap();
-      if (!map.containsKey(key)) {
-        throw new java.lang.IllegalArgumentException();
-      }
-      return map.get(key);
-    }
-
-    private byte memoizedIsInitialized = -1;
-    @java.lang.Override
-    public final boolean isInitialized() {
-      byte isInitialized = memoizedIsInitialized;
-      if (isInitialized == 1) return true;
-      if (isInitialized == 0) return false;
-
-      memoizedIsInitialized = 1;
-      return true;
-    }
-
-    @java.lang.Override
-    public void writeTo(com.google.protobuf.CodedOutputStream output)
-                        throws java.io.IOException {
-      if (!getMetaGraphVersionBytes().isEmpty()) {
-        com.google.protobuf.GeneratedMessageV3.writeString(output, 1, metaGraphVersion_);
-      }
-      if (strippedOpList_ != null) {
-        output.writeMessage(2, getStrippedOpList());
-      }
-      if (anyInfo_ != null) {
-        output.writeMessage(3, getAnyInfo());
-      }
-      for (int i = 0; i < tags_.size(); i++) {
-        com.google.protobuf.GeneratedMessageV3.writeString(output, 4, tags_.getRaw(i));
-      }
-      if (!getTensorflowVersionBytes().isEmpty()) {
-        com.google.protobuf.GeneratedMessageV3.writeString(output, 5, tensorflowVersion_);
-      }
-      if (!getTensorflowGitVersionBytes().isEmpty()) {
-        com.google.protobuf.GeneratedMessageV3.writeString(output, 6, tensorflowGitVersion_);
-      }
-      if (strippedDefaultAttrs_ != false) {
-        output.writeBool(7, strippedDefaultAttrs_);
-      }
-      com.google.protobuf.GeneratedMessageV3
-        .serializeStringMapTo(
-          output,
-          internalGetFunctionAliases(),
-          FunctionAliasesDefaultEntryHolder.defaultEntry,
-          8);
-      unknownFields.writeTo(output);
-    }
-
-    @java.lang.Override
-    public int getSerializedSize() {
-      int size = memoizedSize;
-      if (size != -1) return size;
-
-      size = 0;
-      if (!getMetaGraphVersionBytes().isEmpty()) {
-        size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, metaGraphVersion_);
-      }
-      if (strippedOpList_ != null) {
-        size += com.google.protobuf.CodedOutputStream
-          .computeMessageSize(2, getStrippedOpList());
-      }
-      if (anyInfo_ != null) {
-        size += com.google.protobuf.CodedOutputStream
-          .computeMessageSize(3, getAnyInfo());
-      }
-      {
-        int dataSize = 0;
-        for (int i = 0; i < tags_.size(); i++) {
-          dataSize += computeStringSizeNoTag(tags_.getRaw(i));
-        }
-        size += dataSize;
-        size += 1 * getTagsList().size();
-      }
-      if (!getTensorflowVersionBytes().isEmpty()) {
-        size += com.google.protobuf.GeneratedMessageV3.computeStringSize(5, tensorflowVersion_);
-      }
-      if (!getTensorflowGitVersionBytes().isEmpty()) {
-        size += com.google.protobuf.GeneratedMessageV3.computeStringSize(6, tensorflowGitVersion_);
-      }
-      if (strippedDefaultAttrs_ != false) {
-        size += com.google.protobuf.CodedOutputStream
-          .computeBoolSize(7, strippedDefaultAttrs_);
-      }
-      for (java.util.Map.Entry<java.lang.String, java.lang.String> entry
-           : internalGetFunctionAliases().getMap().entrySet()) {
-        com.google.protobuf.MapEntry<java.lang.String, java.lang.String>
-        functionAliases__ = FunctionAliasesDefaultEntryHolder.defaultEntry.newBuilderForType()
-            .setKey(entry.getKey())
-            .setValue(entry.getValue())
-            .build();
-        size += com.google.protobuf.CodedOutputStream
-            .computeMessageSize(8, functionAliases__);
-      }
-      size += unknownFields.getSerializedSize();
-      memoizedSize = size;
-      return size;
-    }
-
-    @java.lang.Override
-    public boolean equals(final java.lang.Object obj) {
-      if (obj == this) {
-       return true;
-      }
-      if (!(obj instanceof org.tensorflow.proto.framework.MetaGraphDef.MetaInfoDef)) {
-        return super.equals(obj);
-      }
-      org.tensorflow.proto.framework.MetaGraphDef.MetaInfoDef other = (org.tensorflow.proto.framework.MetaGraphDef.MetaInfoDef) obj;
-
-      if (!getMetaGraphVersion()
-          .equals(other.getMetaGraphVersion())) return false;
-      if (hasStrippedOpList() != other.hasStrippedOpList()) return false;
-      if (hasStrippedOpList()) {
-        if (!getStrippedOpList()
-            .equals(other.getStrippedOpList())) return false;
-      }
-      if (hasAnyInfo() != other.hasAnyInfo()) return false;
-      if (hasAnyInfo()) {
-        if (!getAnyInfo()
-            .equals(other.getAnyInfo())) return false;
-      }
-      if (!getTagsList()
-          .equals(other.getTagsList())) return false;
-      if (!getTensorflowVersion()
-          .equals(other.getTensorflowVersion())) return false;
-      if (!getTensorflowGitVersion()
-          .equals(other.getTensorflowGitVersion())) return false;
-      if (getStrippedDefaultAttrs()
-          != other.getStrippedDefaultAttrs()) return false;
-      if (!internalGetFunctionAliases().equals(
-          other.internalGetFunctionAliases())) return false;
-      if (!unknownFields.equals(other.unknownFields)) return false;
-      return true;
-    }
-
-    @java.lang.Override
-    public int hashCode() {
-      if (memoizedHashCode != 0) {
-        return memoizedHashCode;
-      }
-      int hash = 41;
-      hash = (19 * hash) + getDescriptor().hashCode();
-      hash = (37 * hash) + META_GRAPH_VERSION_FIELD_NUMBER;
-      hash = (53 * hash) + getMetaGraphVersion().hashCode();
-      if (hasStrippedOpList()) {
-        hash = (37 * hash) + STRIPPED_OP_LIST_FIELD_NUMBER;
-        hash = (53 * hash) + getStrippedOpList().hashCode();
-      }
-      if (hasAnyInfo()) {
-        hash = (37 * hash) + ANY_INFO_FIELD_NUMBER;
-        hash = (53 * hash) + getAnyInfo().hashCode();
-      }
-      if (getTagsCount() > 0) {
-        hash = (37 * hash) + TAGS_FIELD_NUMBER;
-        hash = (53 * hash) + getTagsList().hashCode();
-      }
-      hash = (37 * hash) + TENSORFLOW_VERSION_FIELD_NUMBER;
-      hash = (53 * hash) + getTensorflowVersion().hashCode();
-      hash = (37 * hash) + TENSORFLOW_GIT_VERSION_FIELD_NUMBER;
-      hash = (53 * hash) + getTensorflowGitVersion().hashCode();
-      hash = (37 * hash) + STRIPPED_DEFAULT_ATTRS_FIELD_NUMBER;
-      hash = (53 * hash) + com.google.protobuf.Internal.hashBoolean(
-          getStrippedDefaultAttrs());
-      if (!internalGetFunctionAliases().getMap().isEmpty()) {
-        hash = (37 * hash) + FUNCTION_ALIASES_FIELD_NUMBER;
-        hash = (53 * hash) + internalGetFunctionAliases().hashCode();
-      }
-      hash = (29 * hash) + unknownFields.hashCode();
-      memoizedHashCode = hash;
-      return hash;
-    }
-
-    public static org.tensorflow.proto.framework.MetaGraphDef.MetaInfoDef parseFrom(
-        java.nio.ByteBuffer data)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data);
-    }
-    public static org.tensorflow.proto.framework.MetaGraphDef.MetaInfoDef parseFrom(
-        java.nio.ByteBuffer data,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.MetaGraphDef.MetaInfoDef parseFrom(
-        com.google.protobuf.ByteString data)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data);
-    }
-    public static org.tensorflow.proto.framework.MetaGraphDef.MetaInfoDef parseFrom(
-        com.google.protobuf.ByteString data,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.MetaGraphDef.MetaInfoDef parseFrom(byte[] data)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data);
-    }
-    public static org.tensorflow.proto.framework.MetaGraphDef.MetaInfoDef parseFrom(
-        byte[] data,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.MetaGraphDef.MetaInfoDef parseFrom(java.io.InputStream input)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input);
-    }
-    public static org.tensorflow.proto.framework.MetaGraphDef.MetaInfoDef parseFrom(
-        java.io.InputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.MetaGraphDef.MetaInfoDef parseDelimitedFrom(java.io.InputStream input)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseDelimitedWithIOException(PARSER, input);
-    }
-    public static org.tensorflow.proto.framework.MetaGraphDef.MetaInfoDef parseDelimitedFrom(
-        java.io.InputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.MetaGraphDef.MetaInfoDef parseFrom(
-        com.google.protobuf.CodedInputStream input)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input);
-    }
-    public static org.tensorflow.proto.framework.MetaGraphDef.MetaInfoDef parseFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input, extensionRegistry);
-    }
-
-    @java.lang.Override
-    public Builder newBuilderForType() { return newBuilder(); }
-    public static Builder newBuilder() {
-      return DEFAULT_INSTANCE.toBuilder();
-    }
-    public static Builder newBuilder(org.tensorflow.proto.framework.MetaGraphDef.MetaInfoDef prototype) {
-      return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-    }
-    @java.lang.Override
-    public Builder toBuilder() {
-      return this == DEFAULT_INSTANCE
-          ? new Builder() : new Builder().mergeFrom(this);
-    }
-
-    @java.lang.Override
-    protected Builder newBuilderForType(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      Builder builder = new Builder(parent);
-      return builder;
-    }
-    /**
-     * <pre>
-     * Meta information regarding the graph to be exported.  To be used by users
-     * of this protocol buffer to encode information regarding their meta graph.
-     * </pre>
-     *
-     * Protobuf type {@code tensorflow.MetaGraphDef.MetaInfoDef}
-     */
-    public static final class Builder extends
-        com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-        // @@protoc_insertion_point(builder_implements:tensorflow.MetaGraphDef.MetaInfoDef)
-        org.tensorflow.proto.framework.MetaGraphDef.MetaInfoDefOrBuilder {
-      public static final com.google.protobuf.Descriptors.Descriptor
-          getDescriptor() {
-        return org.tensorflow.proto.framework.MetaGraphProtos.internal_static_tensorflow_MetaGraphDef_MetaInfoDef_descriptor;
-      }
-
-      @SuppressWarnings({"rawtypes"})
-      protected com.google.protobuf.MapField internalGetMapField(
-          int number) {
-        switch (number) {
-          case 8:
-            return internalGetFunctionAliases();
-          default:
-            throw new RuntimeException(
-                "Invalid map field number: " + number);
-        }
-      }
-      @SuppressWarnings({"rawtypes"})
-      protected com.google.protobuf.MapField internalGetMutableMapField(
-          int number) {
-        switch (number) {
-          case 8:
-            return internalGetMutableFunctionAliases();
-          default:
-            throw new RuntimeException(
-                "Invalid map field number: " + number);
-        }
-      }
-      @java.lang.Override
-      protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-          internalGetFieldAccessorTable() {
-        return org.tensorflow.proto.framework.MetaGraphProtos.internal_static_tensorflow_MetaGraphDef_MetaInfoDef_fieldAccessorTable
-            .ensureFieldAccessorsInitialized(
-                org.tensorflow.proto.framework.MetaGraphDef.MetaInfoDef.class, org.tensorflow.proto.framework.MetaGraphDef.MetaInfoDef.Builder.class);
-      }
-
-      // Construct using org.tensorflow.proto.framework.MetaGraphDef.MetaInfoDef.newBuilder()
-      private Builder() {
-        maybeForceBuilderInitialization();
-      }
-
-      private Builder(
-          com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-        super(parent);
-        maybeForceBuilderInitialization();
-      }
-      private void maybeForceBuilderInitialization() {
-        if (com.google.protobuf.GeneratedMessageV3
-                .alwaysUseFieldBuilders) {
-        }
-      }
-      @java.lang.Override
-      public Builder clear() {
-        super.clear();
-        metaGraphVersion_ = "";
-
-        if (strippedOpListBuilder_ == null) {
-          strippedOpList_ = null;
-        } else {
-          strippedOpList_ = null;
-          strippedOpListBuilder_ = null;
-        }
-        if (anyInfoBuilder_ == null) {
-          anyInfo_ = null;
-        } else {
-          anyInfo_ = null;
-          anyInfoBuilder_ = null;
-        }
-        tags_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-        bitField0_ = (bitField0_ & ~0x00000001);
-        tensorflowVersion_ = "";
-
-        tensorflowGitVersion_ = "";
-
-        strippedDefaultAttrs_ = false;
-
-        internalGetMutableFunctionAliases().clear();
-        return this;
-      }
-
-      @java.lang.Override
-      public com.google.protobuf.Descriptors.Descriptor
-          getDescriptorForType() {
-        return org.tensorflow.proto.framework.MetaGraphProtos.internal_static_tensorflow_MetaGraphDef_MetaInfoDef_descriptor;
-      }
-
-      @java.lang.Override
-      public org.tensorflow.proto.framework.MetaGraphDef.MetaInfoDef getDefaultInstanceForType() {
-        return org.tensorflow.proto.framework.MetaGraphDef.MetaInfoDef.getDefaultInstance();
-      }
-
-      @java.lang.Override
-      public org.tensorflow.proto.framework.MetaGraphDef.MetaInfoDef build() {
-        org.tensorflow.proto.framework.MetaGraphDef.MetaInfoDef result = buildPartial();
-        if (!result.isInitialized()) {
-          throw newUninitializedMessageException(result);
-        }
-        return result;
-      }
-
-      @java.lang.Override
-      public org.tensorflow.proto.framework.MetaGraphDef.MetaInfoDef buildPartial() {
-        org.tensorflow.proto.framework.MetaGraphDef.MetaInfoDef result = new org.tensorflow.proto.framework.MetaGraphDef.MetaInfoDef(this);
-        int from_bitField0_ = bitField0_;
-        result.metaGraphVersion_ = metaGraphVersion_;
-        if (strippedOpListBuilder_ == null) {
-          result.strippedOpList_ = strippedOpList_;
-        } else {
-          result.strippedOpList_ = strippedOpListBuilder_.build();
-        }
-        if (anyInfoBuilder_ == null) {
-          result.anyInfo_ = anyInfo_;
-        } else {
-          result.anyInfo_ = anyInfoBuilder_.build();
-        }
-        if (((bitField0_ & 0x00000001) != 0)) {
-          tags_ = tags_.getUnmodifiableView();
-          bitField0_ = (bitField0_ & ~0x00000001);
-        }
-        result.tags_ = tags_;
-        result.tensorflowVersion_ = tensorflowVersion_;
-        result.tensorflowGitVersion_ = tensorflowGitVersion_;
-        result.strippedDefaultAttrs_ = strippedDefaultAttrs_;
-        result.functionAliases_ = internalGetFunctionAliases();
-        result.functionAliases_.makeImmutable();
-        onBuilt();
-        return result;
-      }
-
-      @java.lang.Override
-      public Builder clone() {
-        return super.clone();
-      }
-      @java.lang.Override
-      public Builder setField(
-          com.google.protobuf.Descriptors.FieldDescriptor field,
-          java.lang.Object value) {
-        return super.setField(field, value);
-      }
-      @java.lang.Override
-      public Builder clearField(
-          com.google.protobuf.Descriptors.FieldDescriptor field) {
-        return super.clearField(field);
-      }
-      @java.lang.Override
-      public Builder clearOneof(
-          com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-        return super.clearOneof(oneof);
-      }
-      @java.lang.Override
-      public Builder setRepeatedField(
-          com.google.protobuf.Descriptors.FieldDescriptor field,
-          int index, java.lang.Object value) {
-        return super.setRepeatedField(field, index, value);
-      }
-      @java.lang.Override
-      public Builder addRepeatedField(
-          com.google.protobuf.Descriptors.FieldDescriptor field,
-          java.lang.Object value) {
-        return super.addRepeatedField(field, value);
-      }
-      @java.lang.Override
-      public Builder mergeFrom(com.google.protobuf.Message other) {
-        if (other instanceof org.tensorflow.proto.framework.MetaGraphDef.MetaInfoDef) {
-          return mergeFrom((org.tensorflow.proto.framework.MetaGraphDef.MetaInfoDef)other);
-        } else {
-          super.mergeFrom(other);
-          return this;
-        }
-      }
-
-      public Builder mergeFrom(org.tensorflow.proto.framework.MetaGraphDef.MetaInfoDef other) {
-        if (other == org.tensorflow.proto.framework.MetaGraphDef.MetaInfoDef.getDefaultInstance()) return this;
-        if (!other.getMetaGraphVersion().isEmpty()) {
-          metaGraphVersion_ = other.metaGraphVersion_;
-          onChanged();
-        }
-        if (other.hasStrippedOpList()) {
-          mergeStrippedOpList(other.getStrippedOpList());
-        }
-        if (other.hasAnyInfo()) {
-          mergeAnyInfo(other.getAnyInfo());
-        }
-        if (!other.tags_.isEmpty()) {
-          if (tags_.isEmpty()) {
-            tags_ = other.tags_;
-            bitField0_ = (bitField0_ & ~0x00000001);
-          } else {
-            ensureTagsIsMutable();
-            tags_.addAll(other.tags_);
-          }
-          onChanged();
-        }
-        if (!other.getTensorflowVersion().isEmpty()) {
-          tensorflowVersion_ = other.tensorflowVersion_;
-          onChanged();
-        }
-        if (!other.getTensorflowGitVersion().isEmpty()) {
-          tensorflowGitVersion_ = other.tensorflowGitVersion_;
-          onChanged();
-        }
-        if (other.getStrippedDefaultAttrs() != false) {
-          setStrippedDefaultAttrs(other.getStrippedDefaultAttrs());
-        }
-        internalGetMutableFunctionAliases().mergeFrom(
-            other.internalGetFunctionAliases());
-        this.mergeUnknownFields(other.unknownFields);
-        onChanged();
-        return this;
-      }
-
-      @java.lang.Override
-      public final boolean isInitialized() {
-        return true;
-      }
-
-      @java.lang.Override
-      public Builder mergeFrom(
-          com.google.protobuf.CodedInputStream input,
-          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-          throws java.io.IOException {
-        org.tensorflow.proto.framework.MetaGraphDef.MetaInfoDef parsedMessage = null;
-        try {
-          parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-        } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-          parsedMessage = (org.tensorflow.proto.framework.MetaGraphDef.MetaInfoDef) e.getUnfinishedMessage();
-          throw e.unwrapIOException();
-        } finally {
-          if (parsedMessage != null) {
-            mergeFrom(parsedMessage);
-          }
-        }
-        return this;
-      }
-      private int bitField0_;
-
-      private java.lang.Object metaGraphVersion_ = "";
-      /**
-       * <pre>
-       * User specified Version string. Can be the name of the model and revision,
-       * steps this model has been trained to, etc.
-       * </pre>
-       *
-       * <code>string meta_graph_version = 1;</code>
-       */
-      public java.lang.String getMetaGraphVersion() {
-        java.lang.Object ref = metaGraphVersion_;
-        if (!(ref instanceof java.lang.String)) {
-          com.google.protobuf.ByteString bs =
-              (com.google.protobuf.ByteString) ref;
-          java.lang.String s = bs.toStringUtf8();
-          metaGraphVersion_ = s;
-          return s;
-        } else {
-          return (java.lang.String) ref;
-        }
-      }
-      /**
-       * <pre>
-       * User specified Version string. Can be the name of the model and revision,
-       * steps this model has been trained to, etc.
-       * </pre>
-       *
-       * <code>string meta_graph_version = 1;</code>
-       */
-      public com.google.protobuf.ByteString
-          getMetaGraphVersionBytes() {
-        java.lang.Object ref = metaGraphVersion_;
-        if (ref instanceof String) {
-          com.google.protobuf.ByteString b = 
-              com.google.protobuf.ByteString.copyFromUtf8(
-                  (java.lang.String) ref);
-          metaGraphVersion_ = b;
-          return b;
-        } else {
-          return (com.google.protobuf.ByteString) ref;
-        }
-      }
-      /**
-       * <pre>
-       * User specified Version string. Can be the name of the model and revision,
-       * steps this model has been trained to, etc.
-       * </pre>
-       *
-       * <code>string meta_graph_version = 1;</code>
-       */
-      public Builder setMetaGraphVersion(
-          java.lang.String value) {
-        if (value == null) {
-    throw new NullPointerException();
-  }
-  
-        metaGraphVersion_ = value;
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * User specified Version string. Can be the name of the model and revision,
-       * steps this model has been trained to, etc.
-       * </pre>
-       *
-       * <code>string meta_graph_version = 1;</code>
-       */
-      public Builder clearMetaGraphVersion() {
-        
-        metaGraphVersion_ = getDefaultInstance().getMetaGraphVersion();
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * User specified Version string. Can be the name of the model and revision,
-       * steps this model has been trained to, etc.
-       * </pre>
-       *
-       * <code>string meta_graph_version = 1;</code>
-       */
-      public Builder setMetaGraphVersionBytes(
-          com.google.protobuf.ByteString value) {
-        if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-        
-        metaGraphVersion_ = value;
-        onChanged();
-        return this;
-      }
-
-      private org.tensorflow.proto.framework.OpList strippedOpList_;
-      private com.google.protobuf.SingleFieldBuilderV3<
-          org.tensorflow.proto.framework.OpList, org.tensorflow.proto.framework.OpList.Builder, org.tensorflow.proto.framework.OpListOrBuilder> strippedOpListBuilder_;
-      /**
-       * <pre>
-       * A copy of the OpDefs used by the producer of this graph_def.
-       * Descriptions and Ops not used in graph_def are stripped out.
-       * </pre>
-       *
-       * <code>.tensorflow.OpList stripped_op_list = 2;</code>
-       */
-      public boolean hasStrippedOpList() {
-        return strippedOpListBuilder_ != null || strippedOpList_ != null;
-      }
-      /**
-       * <pre>
-       * A copy of the OpDefs used by the producer of this graph_def.
-       * Descriptions and Ops not used in graph_def are stripped out.
-       * </pre>
-       *
-       * <code>.tensorflow.OpList stripped_op_list = 2;</code>
-       */
-      public org.tensorflow.proto.framework.OpList getStrippedOpList() {
-        if (strippedOpListBuilder_ == null) {
-          return strippedOpList_ == null ? org.tensorflow.proto.framework.OpList.getDefaultInstance() : strippedOpList_;
-        } else {
-          return strippedOpListBuilder_.getMessage();
-        }
-      }
-      /**
-       * <pre>
-       * A copy of the OpDefs used by the producer of this graph_def.
-       * Descriptions and Ops not used in graph_def are stripped out.
-       * </pre>
-       *
-       * <code>.tensorflow.OpList stripped_op_list = 2;</code>
-       */
-      public Builder setStrippedOpList(org.tensorflow.proto.framework.OpList value) {
-        if (strippedOpListBuilder_ == null) {
-          if (value == null) {
-            throw new NullPointerException();
-          }
-          strippedOpList_ = value;
-          onChanged();
-        } else {
-          strippedOpListBuilder_.setMessage(value);
-        }
-
-        return this;
-      }
-      /**
-       * <pre>
-       * A copy of the OpDefs used by the producer of this graph_def.
-       * Descriptions and Ops not used in graph_def are stripped out.
-       * </pre>
-       *
-       * <code>.tensorflow.OpList stripped_op_list = 2;</code>
-       */
-      public Builder setStrippedOpList(
-          org.tensorflow.proto.framework.OpList.Builder builderForValue) {
-        if (strippedOpListBuilder_ == null) {
-          strippedOpList_ = builderForValue.build();
-          onChanged();
-        } else {
-          strippedOpListBuilder_.setMessage(builderForValue.build());
-        }
-
-        return this;
-      }
-      /**
-       * <pre>
-       * A copy of the OpDefs used by the producer of this graph_def.
-       * Descriptions and Ops not used in graph_def are stripped out.
-       * </pre>
-       *
-       * <code>.tensorflow.OpList stripped_op_list = 2;</code>
-       */
-      public Builder mergeStrippedOpList(org.tensorflow.proto.framework.OpList value) {
-        if (strippedOpListBuilder_ == null) {
-          if (strippedOpList_ != null) {
-            strippedOpList_ =
-              org.tensorflow.proto.framework.OpList.newBuilder(strippedOpList_).mergeFrom(value).buildPartial();
-          } else {
-            strippedOpList_ = value;
-          }
-          onChanged();
-        } else {
-          strippedOpListBuilder_.mergeFrom(value);
-        }
-
-        return this;
-      }
-      /**
-       * <pre>
-       * A copy of the OpDefs used by the producer of this graph_def.
-       * Descriptions and Ops not used in graph_def are stripped out.
-       * </pre>
-       *
-       * <code>.tensorflow.OpList stripped_op_list = 2;</code>
-       */
-      public Builder clearStrippedOpList() {
-        if (strippedOpListBuilder_ == null) {
-          strippedOpList_ = null;
-          onChanged();
-        } else {
-          strippedOpList_ = null;
-          strippedOpListBuilder_ = null;
-        }
-
-        return this;
-      }
-      /**
-       * <pre>
-       * A copy of the OpDefs used by the producer of this graph_def.
-       * Descriptions and Ops not used in graph_def are stripped out.
-       * </pre>
-       *
-       * <code>.tensorflow.OpList stripped_op_list = 2;</code>
-       */
-      public org.tensorflow.proto.framework.OpList.Builder getStrippedOpListBuilder() {
-        
-        onChanged();
-        return getStrippedOpListFieldBuilder().getBuilder();
-      }
-      /**
-       * <pre>
-       * A copy of the OpDefs used by the producer of this graph_def.
-       * Descriptions and Ops not used in graph_def are stripped out.
-       * </pre>
-       *
-       * <code>.tensorflow.OpList stripped_op_list = 2;</code>
-       */
-      public org.tensorflow.proto.framework.OpListOrBuilder getStrippedOpListOrBuilder() {
-        if (strippedOpListBuilder_ != null) {
-          return strippedOpListBuilder_.getMessageOrBuilder();
-        } else {
-          return strippedOpList_ == null ?
-              org.tensorflow.proto.framework.OpList.getDefaultInstance() : strippedOpList_;
-        }
-      }
-      /**
-       * <pre>
-       * A copy of the OpDefs used by the producer of this graph_def.
-       * Descriptions and Ops not used in graph_def are stripped out.
-       * </pre>
-       *
-       * <code>.tensorflow.OpList stripped_op_list = 2;</code>
-       */
-      private com.google.protobuf.SingleFieldBuilderV3<
-          org.tensorflow.proto.framework.OpList, org.tensorflow.proto.framework.OpList.Builder, org.tensorflow.proto.framework.OpListOrBuilder> 
-          getStrippedOpListFieldBuilder() {
-        if (strippedOpListBuilder_ == null) {
-          strippedOpListBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-              org.tensorflow.proto.framework.OpList, org.tensorflow.proto.framework.OpList.Builder, org.tensorflow.proto.framework.OpListOrBuilder>(
-                  getStrippedOpList(),
-                  getParentForChildren(),
-                  isClean());
-          strippedOpList_ = null;
-        }
-        return strippedOpListBuilder_;
-      }
-
-      private com.google.protobuf.Any anyInfo_;
-      private com.google.protobuf.SingleFieldBuilderV3<
-          com.google.protobuf.Any, com.google.protobuf.Any.Builder, com.google.protobuf.AnyOrBuilder> anyInfoBuilder_;
-      /**
-       * <pre>
-       * A serialized protobuf. Can be the time this meta graph is created, or
-       * modified, or name of the model.
-       * </pre>
-       *
-       * <code>.google.protobuf.Any any_info = 3;</code>
-       */
-      public boolean hasAnyInfo() {
-        return anyInfoBuilder_ != null || anyInfo_ != null;
-      }
-      /**
-       * <pre>
-       * A serialized protobuf. Can be the time this meta graph is created, or
-       * modified, or name of the model.
-       * </pre>
-       *
-       * <code>.google.protobuf.Any any_info = 3;</code>
-       */
-      public com.google.protobuf.Any getAnyInfo() {
-        if (anyInfoBuilder_ == null) {
-          return anyInfo_ == null ? com.google.protobuf.Any.getDefaultInstance() : anyInfo_;
-        } else {
-          return anyInfoBuilder_.getMessage();
-        }
-      }
-      /**
-       * <pre>
-       * A serialized protobuf. Can be the time this meta graph is created, or
-       * modified, or name of the model.
-       * </pre>
-       *
-       * <code>.google.protobuf.Any any_info = 3;</code>
-       */
-      public Builder setAnyInfo(com.google.protobuf.Any value) {
-        if (anyInfoBuilder_ == null) {
-          if (value == null) {
-            throw new NullPointerException();
-          }
-          anyInfo_ = value;
-          onChanged();
-        } else {
-          anyInfoBuilder_.setMessage(value);
-        }
-
-        return this;
-      }
-      /**
-       * <pre>
-       * A serialized protobuf. Can be the time this meta graph is created, or
-       * modified, or name of the model.
-       * </pre>
-       *
-       * <code>.google.protobuf.Any any_info = 3;</code>
-       */
-      public Builder setAnyInfo(
-          com.google.protobuf.Any.Builder builderForValue) {
-        if (anyInfoBuilder_ == null) {
-          anyInfo_ = builderForValue.build();
-          onChanged();
-        } else {
-          anyInfoBuilder_.setMessage(builderForValue.build());
-        }
-
-        return this;
-      }
-      /**
-       * <pre>
-       * A serialized protobuf. Can be the time this meta graph is created, or
-       * modified, or name of the model.
-       * </pre>
-       *
-       * <code>.google.protobuf.Any any_info = 3;</code>
-       */
-      public Builder mergeAnyInfo(com.google.protobuf.Any value) {
-        if (anyInfoBuilder_ == null) {
-          if (anyInfo_ != null) {
-            anyInfo_ =
-              com.google.protobuf.Any.newBuilder(anyInfo_).mergeFrom(value).buildPartial();
-          } else {
-            anyInfo_ = value;
-          }
-          onChanged();
-        } else {
-          anyInfoBuilder_.mergeFrom(value);
-        }
-
-        return this;
-      }
-      /**
-       * <pre>
-       * A serialized protobuf. Can be the time this meta graph is created, or
-       * modified, or name of the model.
-       * </pre>
-       *
-       * <code>.google.protobuf.Any any_info = 3;</code>
-       */
-      public Builder clearAnyInfo() {
-        if (anyInfoBuilder_ == null) {
-          anyInfo_ = null;
-          onChanged();
-        } else {
-          anyInfo_ = null;
-          anyInfoBuilder_ = null;
-        }
-
-        return this;
-      }
-      /**
-       * <pre>
-       * A serialized protobuf. Can be the time this meta graph is created, or
-       * modified, or name of the model.
-       * </pre>
-       *
-       * <code>.google.protobuf.Any any_info = 3;</code>
-       */
-      public com.google.protobuf.Any.Builder getAnyInfoBuilder() {
-        
-        onChanged();
-        return getAnyInfoFieldBuilder().getBuilder();
-      }
-      /**
-       * <pre>
-       * A serialized protobuf. Can be the time this meta graph is created, or
-       * modified, or name of the model.
-       * </pre>
-       *
-       * <code>.google.protobuf.Any any_info = 3;</code>
-       */
-      public com.google.protobuf.AnyOrBuilder getAnyInfoOrBuilder() {
-        if (anyInfoBuilder_ != null) {
-          return anyInfoBuilder_.getMessageOrBuilder();
-        } else {
-          return anyInfo_ == null ?
-              com.google.protobuf.Any.getDefaultInstance() : anyInfo_;
-        }
-      }
-      /**
-       * <pre>
-       * A serialized protobuf. Can be the time this meta graph is created, or
-       * modified, or name of the model.
-       * </pre>
-       *
-       * <code>.google.protobuf.Any any_info = 3;</code>
-       */
-      private com.google.protobuf.SingleFieldBuilderV3<
-          com.google.protobuf.Any, com.google.protobuf.Any.Builder, com.google.protobuf.AnyOrBuilder> 
-          getAnyInfoFieldBuilder() {
-        if (anyInfoBuilder_ == null) {
-          anyInfoBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-              com.google.protobuf.Any, com.google.protobuf.Any.Builder, com.google.protobuf.AnyOrBuilder>(
-                  getAnyInfo(),
-                  getParentForChildren(),
-                  isClean());
-          anyInfo_ = null;
-        }
-        return anyInfoBuilder_;
-      }
-
-      private com.google.protobuf.LazyStringList tags_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-      private void ensureTagsIsMutable() {
-        if (!((bitField0_ & 0x00000001) != 0)) {
-          tags_ = new com.google.protobuf.LazyStringArrayList(tags_);
-          bitField0_ |= 0x00000001;
-         }
-      }
-      /**
-       * <pre>
-       * User supplied tag(s) on the meta_graph and included graph_def.
-       * MetaGraphDefs should be tagged with their capabilities or use-cases.
-       * Examples: "train", "serve", "gpu", "tpu", etc.
-       * These tags enable loaders to access the MetaGraph(s) appropriate for a
-       * specific use-case or runtime environment.
-       * </pre>
-       *
-       * <code>repeated string tags = 4;</code>
-       */
-      public com.google.protobuf.ProtocolStringList
-          getTagsList() {
-        return tags_.getUnmodifiableView();
-      }
-      /**
-       * <pre>
-       * User supplied tag(s) on the meta_graph and included graph_def.
-       * MetaGraphDefs should be tagged with their capabilities or use-cases.
-       * Examples: "train", "serve", "gpu", "tpu", etc.
-       * These tags enable loaders to access the MetaGraph(s) appropriate for a
-       * specific use-case or runtime environment.
-       * </pre>
-       *
-       * <code>repeated string tags = 4;</code>
-       */
-      public int getTagsCount() {
-        return tags_.size();
-      }
-      /**
-       * <pre>
-       * User supplied tag(s) on the meta_graph and included graph_def.
-       * MetaGraphDefs should be tagged with their capabilities or use-cases.
-       * Examples: "train", "serve", "gpu", "tpu", etc.
-       * These tags enable loaders to access the MetaGraph(s) appropriate for a
-       * specific use-case or runtime environment.
-       * </pre>
-       *
-       * <code>repeated string tags = 4;</code>
-       */
-      public java.lang.String getTags(int index) {
-        return tags_.get(index);
-      }
-      /**
-       * <pre>
-       * User supplied tag(s) on the meta_graph and included graph_def.
-       * MetaGraphDefs should be tagged with their capabilities or use-cases.
-       * Examples: "train", "serve", "gpu", "tpu", etc.
-       * These tags enable loaders to access the MetaGraph(s) appropriate for a
-       * specific use-case or runtime environment.
-       * </pre>
-       *
-       * <code>repeated string tags = 4;</code>
-       */
-      public com.google.protobuf.ByteString
-          getTagsBytes(int index) {
-        return tags_.getByteString(index);
-      }
-      /**
-       * <pre>
-       * User supplied tag(s) on the meta_graph and included graph_def.
-       * MetaGraphDefs should be tagged with their capabilities or use-cases.
-       * Examples: "train", "serve", "gpu", "tpu", etc.
-       * These tags enable loaders to access the MetaGraph(s) appropriate for a
-       * specific use-case or runtime environment.
-       * </pre>
-       *
-       * <code>repeated string tags = 4;</code>
-       */
-      public Builder setTags(
-          int index, java.lang.String value) {
-        if (value == null) {
-    throw new NullPointerException();
-  }
-  ensureTagsIsMutable();
-        tags_.set(index, value);
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * User supplied tag(s) on the meta_graph and included graph_def.
-       * MetaGraphDefs should be tagged with their capabilities or use-cases.
-       * Examples: "train", "serve", "gpu", "tpu", etc.
-       * These tags enable loaders to access the MetaGraph(s) appropriate for a
-       * specific use-case or runtime environment.
-       * </pre>
-       *
-       * <code>repeated string tags = 4;</code>
-       */
-      public Builder addTags(
-          java.lang.String value) {
-        if (value == null) {
-    throw new NullPointerException();
-  }
-  ensureTagsIsMutable();
-        tags_.add(value);
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * User supplied tag(s) on the meta_graph and included graph_def.
-       * MetaGraphDefs should be tagged with their capabilities or use-cases.
-       * Examples: "train", "serve", "gpu", "tpu", etc.
-       * These tags enable loaders to access the MetaGraph(s) appropriate for a
-       * specific use-case or runtime environment.
-       * </pre>
-       *
-       * <code>repeated string tags = 4;</code>
-       */
-      public Builder addAllTags(
-          java.lang.Iterable<java.lang.String> values) {
-        ensureTagsIsMutable();
-        com.google.protobuf.AbstractMessageLite.Builder.addAll(
-            values, tags_);
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * User supplied tag(s) on the meta_graph and included graph_def.
-       * MetaGraphDefs should be tagged with their capabilities or use-cases.
-       * Examples: "train", "serve", "gpu", "tpu", etc.
-       * These tags enable loaders to access the MetaGraph(s) appropriate for a
-       * specific use-case or runtime environment.
-       * </pre>
-       *
-       * <code>repeated string tags = 4;</code>
-       */
-      public Builder clearTags() {
-        tags_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-        bitField0_ = (bitField0_ & ~0x00000001);
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * User supplied tag(s) on the meta_graph and included graph_def.
-       * MetaGraphDefs should be tagged with their capabilities or use-cases.
-       * Examples: "train", "serve", "gpu", "tpu", etc.
-       * These tags enable loaders to access the MetaGraph(s) appropriate for a
-       * specific use-case or runtime environment.
-       * </pre>
-       *
-       * <code>repeated string tags = 4;</code>
-       */
-      public Builder addTagsBytes(
-          com.google.protobuf.ByteString value) {
-        if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-        ensureTagsIsMutable();
-        tags_.add(value);
-        onChanged();
-        return this;
-      }
-
-      private java.lang.Object tensorflowVersion_ = "";
-      /**
-       * <pre>
-       * The __version__ string of the tensorflow build used to write this graph.
-       * This will be populated by the framework, which will overwrite any user
-       * supplied value.
-       * </pre>
-       *
-       * <code>string tensorflow_version = 5;</code>
-       */
-      public java.lang.String getTensorflowVersion() {
-        java.lang.Object ref = tensorflowVersion_;
-        if (!(ref instanceof java.lang.String)) {
-          com.google.protobuf.ByteString bs =
-              (com.google.protobuf.ByteString) ref;
-          java.lang.String s = bs.toStringUtf8();
-          tensorflowVersion_ = s;
-          return s;
-        } else {
-          return (java.lang.String) ref;
-        }
-      }
-      /**
-       * <pre>
-       * The __version__ string of the tensorflow build used to write this graph.
-       * This will be populated by the framework, which will overwrite any user
-       * supplied value.
-       * </pre>
-       *
-       * <code>string tensorflow_version = 5;</code>
-       */
-      public com.google.protobuf.ByteString
-          getTensorflowVersionBytes() {
-        java.lang.Object ref = tensorflowVersion_;
-        if (ref instanceof String) {
-          com.google.protobuf.ByteString b = 
-              com.google.protobuf.ByteString.copyFromUtf8(
-                  (java.lang.String) ref);
-          tensorflowVersion_ = b;
-          return b;
-        } else {
-          return (com.google.protobuf.ByteString) ref;
-        }
-      }
-      /**
-       * <pre>
-       * The __version__ string of the tensorflow build used to write this graph.
-       * This will be populated by the framework, which will overwrite any user
-       * supplied value.
-       * </pre>
-       *
-       * <code>string tensorflow_version = 5;</code>
-       */
-      public Builder setTensorflowVersion(
-          java.lang.String value) {
-        if (value == null) {
-    throw new NullPointerException();
-  }
-  
-        tensorflowVersion_ = value;
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * The __version__ string of the tensorflow build used to write this graph.
-       * This will be populated by the framework, which will overwrite any user
-       * supplied value.
-       * </pre>
-       *
-       * <code>string tensorflow_version = 5;</code>
-       */
-      public Builder clearTensorflowVersion() {
-        
-        tensorflowVersion_ = getDefaultInstance().getTensorflowVersion();
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * The __version__ string of the tensorflow build used to write this graph.
-       * This will be populated by the framework, which will overwrite any user
-       * supplied value.
-       * </pre>
-       *
-       * <code>string tensorflow_version = 5;</code>
-       */
-      public Builder setTensorflowVersionBytes(
-          com.google.protobuf.ByteString value) {
-        if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-        
-        tensorflowVersion_ = value;
-        onChanged();
-        return this;
-      }
-
-      private java.lang.Object tensorflowGitVersion_ = "";
-      /**
-       * <pre>
-       * The __git_version__ string of the tensorflow build used to write this
-       * graph. This will be populated by the framework, which will overwrite any
-       * user supplied value.
-       * </pre>
-       *
-       * <code>string tensorflow_git_version = 6;</code>
-       */
-      public java.lang.String getTensorflowGitVersion() {
-        java.lang.Object ref = tensorflowGitVersion_;
-        if (!(ref instanceof java.lang.String)) {
-          com.google.protobuf.ByteString bs =
-              (com.google.protobuf.ByteString) ref;
-          java.lang.String s = bs.toStringUtf8();
-          tensorflowGitVersion_ = s;
-          return s;
-        } else {
-          return (java.lang.String) ref;
-        }
-      }
-      /**
-       * <pre>
-       * The __git_version__ string of the tensorflow build used to write this
-       * graph. This will be populated by the framework, which will overwrite any
-       * user supplied value.
-       * </pre>
-       *
-       * <code>string tensorflow_git_version = 6;</code>
-       */
-      public com.google.protobuf.ByteString
-          getTensorflowGitVersionBytes() {
-        java.lang.Object ref = tensorflowGitVersion_;
-        if (ref instanceof String) {
-          com.google.protobuf.ByteString b = 
-              com.google.protobuf.ByteString.copyFromUtf8(
-                  (java.lang.String) ref);
-          tensorflowGitVersion_ = b;
-          return b;
-        } else {
-          return (com.google.protobuf.ByteString) ref;
-        }
-      }
-      /**
-       * <pre>
-       * The __git_version__ string of the tensorflow build used to write this
-       * graph. This will be populated by the framework, which will overwrite any
-       * user supplied value.
-       * </pre>
-       *
-       * <code>string tensorflow_git_version = 6;</code>
-       */
-      public Builder setTensorflowGitVersion(
-          java.lang.String value) {
-        if (value == null) {
-    throw new NullPointerException();
-  }
-  
-        tensorflowGitVersion_ = value;
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * The __git_version__ string of the tensorflow build used to write this
-       * graph. This will be populated by the framework, which will overwrite any
-       * user supplied value.
-       * </pre>
-       *
-       * <code>string tensorflow_git_version = 6;</code>
-       */
-      public Builder clearTensorflowGitVersion() {
-        
-        tensorflowGitVersion_ = getDefaultInstance().getTensorflowGitVersion();
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * The __git_version__ string of the tensorflow build used to write this
-       * graph. This will be populated by the framework, which will overwrite any
-       * user supplied value.
-       * </pre>
-       *
-       * <code>string tensorflow_git_version = 6;</code>
-       */
-      public Builder setTensorflowGitVersionBytes(
-          com.google.protobuf.ByteString value) {
-        if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-        
-        tensorflowGitVersion_ = value;
-        onChanged();
-        return this;
-      }
-
-      private boolean strippedDefaultAttrs_ ;
-      /**
-       * <pre>
-       * A flag to denote whether default-valued attrs have been stripped from
-       * the nodes in this graph_def.
-       * </pre>
-       *
-       * <code>bool stripped_default_attrs = 7;</code>
-       */
-      public boolean getStrippedDefaultAttrs() {
-        return strippedDefaultAttrs_;
-      }
-      /**
-       * <pre>
-       * A flag to denote whether default-valued attrs have been stripped from
-       * the nodes in this graph_def.
-       * </pre>
-       *
-       * <code>bool stripped_default_attrs = 7;</code>
-       */
-      public Builder setStrippedDefaultAttrs(boolean value) {
-        
-        strippedDefaultAttrs_ = value;
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * A flag to denote whether default-valued attrs have been stripped from
-       * the nodes in this graph_def.
-       * </pre>
-       *
-       * <code>bool stripped_default_attrs = 7;</code>
-       */
-      public Builder clearStrippedDefaultAttrs() {
-        
-        strippedDefaultAttrs_ = false;
-        onChanged();
-        return this;
-      }
-
-      private com.google.protobuf.MapField<
-          java.lang.String, java.lang.String> functionAliases_;
-      private com.google.protobuf.MapField<java.lang.String, java.lang.String>
-      internalGetFunctionAliases() {
-        if (functionAliases_ == null) {
-          return com.google.protobuf.MapField.emptyMapField(
-              FunctionAliasesDefaultEntryHolder.defaultEntry);
-        }
-        return functionAliases_;
-      }
-      private com.google.protobuf.MapField<java.lang.String, java.lang.String>
-      internalGetMutableFunctionAliases() {
-        onChanged();;
-        if (functionAliases_ == null) {
-          functionAliases_ = com.google.protobuf.MapField.newMapField(
-              FunctionAliasesDefaultEntryHolder.defaultEntry);
-        }
-        if (!functionAliases_.isMutable()) {
-          functionAliases_ = functionAliases_.copy();
-        }
-        return functionAliases_;
-      }
-
-      public int getFunctionAliasesCount() {
-        return internalGetFunctionAliases().getMap().size();
-      }
-      /**
-       * <pre>
-       * FunctionDef name to aliases mapping.
-       * </pre>
-       *
-       * <code>map&lt;string, string&gt; function_aliases = 8;</code>
-       */
-
-      public boolean containsFunctionAliases(
-          java.lang.String key) {
-        if (key == null) { throw new java.lang.NullPointerException(); }
-        return internalGetFunctionAliases().getMap().containsKey(key);
-      }
-      /**
-       * Use {@link #getFunctionAliasesMap()} instead.
-       */
-      @java.lang.Deprecated
-      public java.util.Map<java.lang.String, java.lang.String> getFunctionAliases() {
-        return getFunctionAliasesMap();
-      }
-      /**
-       * <pre>
-       * FunctionDef name to aliases mapping.
-       * </pre>
-       *
-       * <code>map&lt;string, string&gt; function_aliases = 8;</code>
-       */
-
-      public java.util.Map<java.lang.String, java.lang.String> getFunctionAliasesMap() {
-        return internalGetFunctionAliases().getMap();
-      }
-      /**
-       * <pre>
-       * FunctionDef name to aliases mapping.
-       * </pre>
-       *
-       * <code>map&lt;string, string&gt; function_aliases = 8;</code>
-       */
-
-      public java.lang.String getFunctionAliasesOrDefault(
-          java.lang.String key,
-          java.lang.String defaultValue) {
-        if (key == null) { throw new java.lang.NullPointerException(); }
-        java.util.Map<java.lang.String, java.lang.String> map =
-            internalGetFunctionAliases().getMap();
-        return map.containsKey(key) ? map.get(key) : defaultValue;
-      }
-      /**
-       * <pre>
-       * FunctionDef name to aliases mapping.
-       * </pre>
-       *
-       * <code>map&lt;string, string&gt; function_aliases = 8;</code>
-       */
-
-      public java.lang.String getFunctionAliasesOrThrow(
-          java.lang.String key) {
-        if (key == null) { throw new java.lang.NullPointerException(); }
-        java.util.Map<java.lang.String, java.lang.String> map =
-            internalGetFunctionAliases().getMap();
-        if (!map.containsKey(key)) {
-          throw new java.lang.IllegalArgumentException();
-        }
-        return map.get(key);
-      }
-
-      public Builder clearFunctionAliases() {
-        internalGetMutableFunctionAliases().getMutableMap()
-            .clear();
-        return this;
-      }
-      /**
-       * <pre>
-       * FunctionDef name to aliases mapping.
-       * </pre>
-       *
-       * <code>map&lt;string, string&gt; function_aliases = 8;</code>
-       */
-
-      public Builder removeFunctionAliases(
-          java.lang.String key) {
-        if (key == null) { throw new java.lang.NullPointerException(); }
-        internalGetMutableFunctionAliases().getMutableMap()
-            .remove(key);
-        return this;
-      }
-      /**
-       * Use alternate mutation accessors instead.
-       */
-      @java.lang.Deprecated
-      public java.util.Map<java.lang.String, java.lang.String>
-      getMutableFunctionAliases() {
-        return internalGetMutableFunctionAliases().getMutableMap();
-      }
-      /**
-       * <pre>
-       * FunctionDef name to aliases mapping.
-       * </pre>
-       *
-       * <code>map&lt;string, string&gt; function_aliases = 8;</code>
-       */
-      public Builder putFunctionAliases(
-          java.lang.String key,
-          java.lang.String value) {
-        if (key == null) { throw new java.lang.NullPointerException(); }
-        if (value == null) { throw new java.lang.NullPointerException(); }
-        internalGetMutableFunctionAliases().getMutableMap()
-            .put(key, value);
-        return this;
-      }
-      /**
-       * <pre>
-       * FunctionDef name to aliases mapping.
-       * </pre>
-       *
-       * <code>map&lt;string, string&gt; function_aliases = 8;</code>
-       */
-
-      public Builder putAllFunctionAliases(
-          java.util.Map<java.lang.String, java.lang.String> values) {
-        internalGetMutableFunctionAliases().getMutableMap()
-            .putAll(values);
-        return this;
-      }
-      @java.lang.Override
-      public final Builder setUnknownFields(
-          final com.google.protobuf.UnknownFieldSet unknownFields) {
-        return super.setUnknownFields(unknownFields);
-      }
-
-      @java.lang.Override
-      public final Builder mergeUnknownFields(
-          final com.google.protobuf.UnknownFieldSet unknownFields) {
-        return super.mergeUnknownFields(unknownFields);
-      }
-
-
-      // @@protoc_insertion_point(builder_scope:tensorflow.MetaGraphDef.MetaInfoDef)
-    }
-
-    // @@protoc_insertion_point(class_scope:tensorflow.MetaGraphDef.MetaInfoDef)
-    private static final org.tensorflow.proto.framework.MetaGraphDef.MetaInfoDef DEFAULT_INSTANCE;
-    static {
-      DEFAULT_INSTANCE = new org.tensorflow.proto.framework.MetaGraphDef.MetaInfoDef();
-    }
-
-    public static org.tensorflow.proto.framework.MetaGraphDef.MetaInfoDef getDefaultInstance() {
-      return DEFAULT_INSTANCE;
-    }
-
-    private static final com.google.protobuf.Parser<MetaInfoDef>
-        PARSER = new com.google.protobuf.AbstractParser<MetaInfoDef>() {
-      @java.lang.Override
-      public MetaInfoDef parsePartialFrom(
-          com.google.protobuf.CodedInputStream input,
-          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-          throws com.google.protobuf.InvalidProtocolBufferException {
-        return new MetaInfoDef(input, extensionRegistry);
-      }
-    };
-
-    public static com.google.protobuf.Parser<MetaInfoDef> parser() {
-      return PARSER;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Parser<MetaInfoDef> getParserForType() {
-      return PARSER;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.MetaGraphDef.MetaInfoDef getDefaultInstanceForType() {
-      return DEFAULT_INSTANCE;
-    }
-
-  }
-
-  public static final int META_INFO_DEF_FIELD_NUMBER = 1;
-  private org.tensorflow.proto.framework.MetaGraphDef.MetaInfoDef metaInfoDef_;
-  /**
-   * <code>.tensorflow.MetaGraphDef.MetaInfoDef meta_info_def = 1;</code>
-   */
-  public boolean hasMetaInfoDef() {
-    return metaInfoDef_ != null;
-  }
-  /**
-   * <code>.tensorflow.MetaGraphDef.MetaInfoDef meta_info_def = 1;</code>
-   */
-  public org.tensorflow.proto.framework.MetaGraphDef.MetaInfoDef getMetaInfoDef() {
-    return metaInfoDef_ == null ? org.tensorflow.proto.framework.MetaGraphDef.MetaInfoDef.getDefaultInstance() : metaInfoDef_;
-  }
-  /**
-   * <code>.tensorflow.MetaGraphDef.MetaInfoDef meta_info_def = 1;</code>
-   */
-  public org.tensorflow.proto.framework.MetaGraphDef.MetaInfoDefOrBuilder getMetaInfoDefOrBuilder() {
-    return getMetaInfoDef();
-  }
-
-  public static final int GRAPH_DEF_FIELD_NUMBER = 2;
-  private org.tensorflow.proto.framework.GraphDef graphDef_;
-  /**
-   * <pre>
-   * GraphDef.
-   * </pre>
-   *
-   * <code>.tensorflow.GraphDef graph_def = 2;</code>
-   */
-  public boolean hasGraphDef() {
-    return graphDef_ != null;
-  }
-  /**
-   * <pre>
-   * GraphDef.
-   * </pre>
-   *
-   * <code>.tensorflow.GraphDef graph_def = 2;</code>
-   */
-  public org.tensorflow.proto.framework.GraphDef getGraphDef() {
-    return graphDef_ == null ? org.tensorflow.proto.framework.GraphDef.getDefaultInstance() : graphDef_;
-  }
-  /**
-   * <pre>
-   * GraphDef.
-   * </pre>
-   *
-   * <code>.tensorflow.GraphDef graph_def = 2;</code>
-   */
-  public org.tensorflow.proto.framework.GraphDefOrBuilder getGraphDefOrBuilder() {
-    return getGraphDef();
-  }
-
-  public static final int SAVER_DEF_FIELD_NUMBER = 3;
-  private org.tensorflow.proto.util.SaverDef saverDef_;
-  /**
-   * <pre>
-   * SaverDef.
-   * </pre>
-   *
-   * <code>.tensorflow.SaverDef saver_def = 3;</code>
-   */
-  public boolean hasSaverDef() {
-    return saverDef_ != null;
-  }
-  /**
-   * <pre>
-   * SaverDef.
-   * </pre>
-   *
-   * <code>.tensorflow.SaverDef saver_def = 3;</code>
-   */
-  public org.tensorflow.proto.util.SaverDef getSaverDef() {
-    return saverDef_ == null ? org.tensorflow.proto.util.SaverDef.getDefaultInstance() : saverDef_;
-  }
-  /**
-   * <pre>
-   * SaverDef.
-   * </pre>
-   *
-   * <code>.tensorflow.SaverDef saver_def = 3;</code>
-   */
-  public org.tensorflow.proto.util.SaverDefOrBuilder getSaverDefOrBuilder() {
-    return getSaverDef();
-  }
-
-  public static final int COLLECTION_DEF_FIELD_NUMBER = 4;
-  private static final class CollectionDefDefaultEntryHolder {
-    static final com.google.protobuf.MapEntry<
-        java.lang.String, org.tensorflow.proto.framework.CollectionDef> defaultEntry =
-            com.google.protobuf.MapEntry
-            .<java.lang.String, org.tensorflow.proto.framework.CollectionDef>newDefaultInstance(
-                org.tensorflow.proto.framework.MetaGraphProtos.internal_static_tensorflow_MetaGraphDef_CollectionDefEntry_descriptor, 
-                com.google.protobuf.WireFormat.FieldType.STRING,
-                "",
-                com.google.protobuf.WireFormat.FieldType.MESSAGE,
-                org.tensorflow.proto.framework.CollectionDef.getDefaultInstance());
-  }
-  private com.google.protobuf.MapField<
-      java.lang.String, org.tensorflow.proto.framework.CollectionDef> collectionDef_;
-  private com.google.protobuf.MapField<java.lang.String, org.tensorflow.proto.framework.CollectionDef>
-  internalGetCollectionDef() {
-    if (collectionDef_ == null) {
-      return com.google.protobuf.MapField.emptyMapField(
-          CollectionDefDefaultEntryHolder.defaultEntry);
-    }
-    return collectionDef_;
-  }
-
-  public int getCollectionDefCount() {
-    return internalGetCollectionDef().getMap().size();
-  }
-  /**
-   * <pre>
-   * collection_def: Map from collection name to collections.
-   * See CollectionDef section for details.
-   * </pre>
-   *
-   * <code>map&lt;string, .tensorflow.CollectionDef&gt; collection_def = 4;</code>
-   */
-
-  public boolean containsCollectionDef(
-      java.lang.String key) {
-    if (key == null) { throw new java.lang.NullPointerException(); }
-    return internalGetCollectionDef().getMap().containsKey(key);
-  }
-  /**
-   * Use {@link #getCollectionDefMap()} instead.
-   */
-  @java.lang.Deprecated
-  public java.util.Map<java.lang.String, org.tensorflow.proto.framework.CollectionDef> getCollectionDef() {
-    return getCollectionDefMap();
-  }
-  /**
-   * <pre>
-   * collection_def: Map from collection name to collections.
-   * See CollectionDef section for details.
-   * </pre>
-   *
-   * <code>map&lt;string, .tensorflow.CollectionDef&gt; collection_def = 4;</code>
-   */
-
-  public java.util.Map<java.lang.String, org.tensorflow.proto.framework.CollectionDef> getCollectionDefMap() {
-    return internalGetCollectionDef().getMap();
-  }
-  /**
-   * <pre>
-   * collection_def: Map from collection name to collections.
-   * See CollectionDef section for details.
-   * </pre>
-   *
-   * <code>map&lt;string, .tensorflow.CollectionDef&gt; collection_def = 4;</code>
-   */
-
-  public org.tensorflow.proto.framework.CollectionDef getCollectionDefOrDefault(
-      java.lang.String key,
-      org.tensorflow.proto.framework.CollectionDef defaultValue) {
-    if (key == null) { throw new java.lang.NullPointerException(); }
-    java.util.Map<java.lang.String, org.tensorflow.proto.framework.CollectionDef> map =
-        internalGetCollectionDef().getMap();
-    return map.containsKey(key) ? map.get(key) : defaultValue;
-  }
-  /**
-   * <pre>
-   * collection_def: Map from collection name to collections.
-   * See CollectionDef section for details.
-   * </pre>
-   *
-   * <code>map&lt;string, .tensorflow.CollectionDef&gt; collection_def = 4;</code>
-   */
-
-  public org.tensorflow.proto.framework.CollectionDef getCollectionDefOrThrow(
-      java.lang.String key) {
-    if (key == null) { throw new java.lang.NullPointerException(); }
-    java.util.Map<java.lang.String, org.tensorflow.proto.framework.CollectionDef> map =
-        internalGetCollectionDef().getMap();
-    if (!map.containsKey(key)) {
-      throw new java.lang.IllegalArgumentException();
-    }
-    return map.get(key);
-  }
-
-  public static final int SIGNATURE_DEF_FIELD_NUMBER = 5;
-  private static final class SignatureDefDefaultEntryHolder {
-    static final com.google.protobuf.MapEntry<
-        java.lang.String, org.tensorflow.proto.framework.SignatureDef> defaultEntry =
-            com.google.protobuf.MapEntry
-            .<java.lang.String, org.tensorflow.proto.framework.SignatureDef>newDefaultInstance(
-                org.tensorflow.proto.framework.MetaGraphProtos.internal_static_tensorflow_MetaGraphDef_SignatureDefEntry_descriptor, 
-                com.google.protobuf.WireFormat.FieldType.STRING,
-                "",
-                com.google.protobuf.WireFormat.FieldType.MESSAGE,
-                org.tensorflow.proto.framework.SignatureDef.getDefaultInstance());
-  }
-  private com.google.protobuf.MapField<
-      java.lang.String, org.tensorflow.proto.framework.SignatureDef> signatureDef_;
-  private com.google.protobuf.MapField<java.lang.String, org.tensorflow.proto.framework.SignatureDef>
-  internalGetSignatureDef() {
-    if (signatureDef_ == null) {
-      return com.google.protobuf.MapField.emptyMapField(
-          SignatureDefDefaultEntryHolder.defaultEntry);
-    }
-    return signatureDef_;
-  }
-
-  public int getSignatureDefCount() {
-    return internalGetSignatureDef().getMap().size();
-  }
-  /**
-   * <pre>
-   * signature_def: Map from user supplied key for a signature to a single
-   * SignatureDef.
-   * </pre>
-   *
-   * <code>map&lt;string, .tensorflow.SignatureDef&gt; signature_def = 5;</code>
-   */
-
-  public boolean containsSignatureDef(
-      java.lang.String key) {
-    if (key == null) { throw new java.lang.NullPointerException(); }
-    return internalGetSignatureDef().getMap().containsKey(key);
-  }
-  /**
-   * Use {@link #getSignatureDefMap()} instead.
-   */
-  @java.lang.Deprecated
-  public java.util.Map<java.lang.String, org.tensorflow.proto.framework.SignatureDef> getSignatureDef() {
-    return getSignatureDefMap();
-  }
-  /**
-   * <pre>
-   * signature_def: Map from user supplied key for a signature to a single
-   * SignatureDef.
-   * </pre>
-   *
-   * <code>map&lt;string, .tensorflow.SignatureDef&gt; signature_def = 5;</code>
-   */
-
-  public java.util.Map<java.lang.String, org.tensorflow.proto.framework.SignatureDef> getSignatureDefMap() {
-    return internalGetSignatureDef().getMap();
-  }
-  /**
-   * <pre>
-   * signature_def: Map from user supplied key for a signature to a single
-   * SignatureDef.
-   * </pre>
-   *
-   * <code>map&lt;string, .tensorflow.SignatureDef&gt; signature_def = 5;</code>
-   */
-
-  public org.tensorflow.proto.framework.SignatureDef getSignatureDefOrDefault(
-      java.lang.String key,
-      org.tensorflow.proto.framework.SignatureDef defaultValue) {
-    if (key == null) { throw new java.lang.NullPointerException(); }
-    java.util.Map<java.lang.String, org.tensorflow.proto.framework.SignatureDef> map =
-        internalGetSignatureDef().getMap();
-    return map.containsKey(key) ? map.get(key) : defaultValue;
-  }
-  /**
-   * <pre>
-   * signature_def: Map from user supplied key for a signature to a single
-   * SignatureDef.
-   * </pre>
-   *
-   * <code>map&lt;string, .tensorflow.SignatureDef&gt; signature_def = 5;</code>
-   */
-
-  public org.tensorflow.proto.framework.SignatureDef getSignatureDefOrThrow(
-      java.lang.String key) {
-    if (key == null) { throw new java.lang.NullPointerException(); }
-    java.util.Map<java.lang.String, org.tensorflow.proto.framework.SignatureDef> map =
-        internalGetSignatureDef().getMap();
-    if (!map.containsKey(key)) {
-      throw new java.lang.IllegalArgumentException();
-    }
-    return map.get(key);
-  }
-
-  public static final int ASSET_FILE_DEF_FIELD_NUMBER = 6;
-  private java.util.List<org.tensorflow.proto.framework.AssetFileDef> assetFileDef_;
-  /**
-   * <pre>
-   * Asset file def to be used with the defined graph.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.AssetFileDef asset_file_def = 6;</code>
-   */
-  public java.util.List<org.tensorflow.proto.framework.AssetFileDef> getAssetFileDefList() {
-    return assetFileDef_;
-  }
-  /**
-   * <pre>
-   * Asset file def to be used with the defined graph.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.AssetFileDef asset_file_def = 6;</code>
-   */
-  public java.util.List<? extends org.tensorflow.proto.framework.AssetFileDefOrBuilder> 
-      getAssetFileDefOrBuilderList() {
-    return assetFileDef_;
-  }
-  /**
-   * <pre>
-   * Asset file def to be used with the defined graph.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.AssetFileDef asset_file_def = 6;</code>
-   */
-  public int getAssetFileDefCount() {
-    return assetFileDef_.size();
-  }
-  /**
-   * <pre>
-   * Asset file def to be used with the defined graph.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.AssetFileDef asset_file_def = 6;</code>
-   */
-  public org.tensorflow.proto.framework.AssetFileDef getAssetFileDef(int index) {
-    return assetFileDef_.get(index);
-  }
-  /**
-   * <pre>
-   * Asset file def to be used with the defined graph.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.AssetFileDef asset_file_def = 6;</code>
-   */
-  public org.tensorflow.proto.framework.AssetFileDefOrBuilder getAssetFileDefOrBuilder(
-      int index) {
-    return assetFileDef_.get(index);
-  }
-
-  public static final int OBJECT_GRAPH_DEF_FIELD_NUMBER = 7;
-  private org.tensorflow.proto.framework.SavedObjectGraph objectGraphDef_;
-  /**
-   * <pre>
-   * Extra information about the structure of functions and stateful objects.
-   * </pre>
-   *
-   * <code>.tensorflow.SavedObjectGraph object_graph_def = 7;</code>
-   */
-  public boolean hasObjectGraphDef() {
-    return objectGraphDef_ != null;
-  }
-  /**
-   * <pre>
-   * Extra information about the structure of functions and stateful objects.
-   * </pre>
-   *
-   * <code>.tensorflow.SavedObjectGraph object_graph_def = 7;</code>
-   */
-  public org.tensorflow.proto.framework.SavedObjectGraph getObjectGraphDef() {
-    return objectGraphDef_ == null ? org.tensorflow.proto.framework.SavedObjectGraph.getDefaultInstance() : objectGraphDef_;
-  }
-  /**
-   * <pre>
-   * Extra information about the structure of functions and stateful objects.
-   * </pre>
-   *
-   * <code>.tensorflow.SavedObjectGraph object_graph_def = 7;</code>
-   */
-  public org.tensorflow.proto.framework.SavedObjectGraphOrBuilder getObjectGraphDefOrBuilder() {
-    return getObjectGraphDef();
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    if (metaInfoDef_ != null) {
-      output.writeMessage(1, getMetaInfoDef());
-    }
-    if (graphDef_ != null) {
-      output.writeMessage(2, getGraphDef());
-    }
-    if (saverDef_ != null) {
-      output.writeMessage(3, getSaverDef());
-    }
-    com.google.protobuf.GeneratedMessageV3
-      .serializeStringMapTo(
-        output,
-        internalGetCollectionDef(),
-        CollectionDefDefaultEntryHolder.defaultEntry,
-        4);
-    com.google.protobuf.GeneratedMessageV3
-      .serializeStringMapTo(
-        output,
-        internalGetSignatureDef(),
-        SignatureDefDefaultEntryHolder.defaultEntry,
-        5);
-    for (int i = 0; i < assetFileDef_.size(); i++) {
-      output.writeMessage(6, assetFileDef_.get(i));
-    }
-    if (objectGraphDef_ != null) {
-      output.writeMessage(7, getObjectGraphDef());
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (metaInfoDef_ != null) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(1, getMetaInfoDef());
-    }
-    if (graphDef_ != null) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(2, getGraphDef());
-    }
-    if (saverDef_ != null) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(3, getSaverDef());
-    }
-    for (java.util.Map.Entry<java.lang.String, org.tensorflow.proto.framework.CollectionDef> entry
-         : internalGetCollectionDef().getMap().entrySet()) {
-      com.google.protobuf.MapEntry<java.lang.String, org.tensorflow.proto.framework.CollectionDef>
-      collectionDef__ = CollectionDefDefaultEntryHolder.defaultEntry.newBuilderForType()
-          .setKey(entry.getKey())
-          .setValue(entry.getValue())
-          .build();
-      size += com.google.protobuf.CodedOutputStream
-          .computeMessageSize(4, collectionDef__);
-    }
-    for (java.util.Map.Entry<java.lang.String, org.tensorflow.proto.framework.SignatureDef> entry
-         : internalGetSignatureDef().getMap().entrySet()) {
-      com.google.protobuf.MapEntry<java.lang.String, org.tensorflow.proto.framework.SignatureDef>
-      signatureDef__ = SignatureDefDefaultEntryHolder.defaultEntry.newBuilderForType()
-          .setKey(entry.getKey())
-          .setValue(entry.getValue())
-          .build();
-      size += com.google.protobuf.CodedOutputStream
-          .computeMessageSize(5, signatureDef__);
-    }
-    for (int i = 0; i < assetFileDef_.size(); i++) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(6, assetFileDef_.get(i));
-    }
-    if (objectGraphDef_ != null) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(7, getObjectGraphDef());
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.framework.MetaGraphDef)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.framework.MetaGraphDef other = (org.tensorflow.proto.framework.MetaGraphDef) obj;
-
-    if (hasMetaInfoDef() != other.hasMetaInfoDef()) return false;
-    if (hasMetaInfoDef()) {
-      if (!getMetaInfoDef()
-          .equals(other.getMetaInfoDef())) return false;
-    }
-    if (hasGraphDef() != other.hasGraphDef()) return false;
-    if (hasGraphDef()) {
-      if (!getGraphDef()
-          .equals(other.getGraphDef())) return false;
-    }
-    if (hasSaverDef() != other.hasSaverDef()) return false;
-    if (hasSaverDef()) {
-      if (!getSaverDef()
-          .equals(other.getSaverDef())) return false;
-    }
-    if (!internalGetCollectionDef().equals(
-        other.internalGetCollectionDef())) return false;
-    if (!internalGetSignatureDef().equals(
-        other.internalGetSignatureDef())) return false;
-    if (!getAssetFileDefList()
-        .equals(other.getAssetFileDefList())) return false;
-    if (hasObjectGraphDef() != other.hasObjectGraphDef()) return false;
-    if (hasObjectGraphDef()) {
-      if (!getObjectGraphDef()
-          .equals(other.getObjectGraphDef())) return false;
-    }
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    if (hasMetaInfoDef()) {
-      hash = (37 * hash) + META_INFO_DEF_FIELD_NUMBER;
-      hash = (53 * hash) + getMetaInfoDef().hashCode();
-    }
-    if (hasGraphDef()) {
-      hash = (37 * hash) + GRAPH_DEF_FIELD_NUMBER;
-      hash = (53 * hash) + getGraphDef().hashCode();
-    }
-    if (hasSaverDef()) {
-      hash = (37 * hash) + SAVER_DEF_FIELD_NUMBER;
-      hash = (53 * hash) + getSaverDef().hashCode();
-    }
-    if (!internalGetCollectionDef().getMap().isEmpty()) {
-      hash = (37 * hash) + COLLECTION_DEF_FIELD_NUMBER;
-      hash = (53 * hash) + internalGetCollectionDef().hashCode();
-    }
-    if (!internalGetSignatureDef().getMap().isEmpty()) {
-      hash = (37 * hash) + SIGNATURE_DEF_FIELD_NUMBER;
-      hash = (53 * hash) + internalGetSignatureDef().hashCode();
-    }
-    if (getAssetFileDefCount() > 0) {
-      hash = (37 * hash) + ASSET_FILE_DEF_FIELD_NUMBER;
-      hash = (53 * hash) + getAssetFileDefList().hashCode();
-    }
-    if (hasObjectGraphDef()) {
-      hash = (37 * hash) + OBJECT_GRAPH_DEF_FIELD_NUMBER;
-      hash = (53 * hash) + getObjectGraphDef().hashCode();
-    }
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.framework.MetaGraphDef parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.MetaGraphDef parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.MetaGraphDef parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.MetaGraphDef parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.MetaGraphDef parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.MetaGraphDef parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.MetaGraphDef parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.MetaGraphDef parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.MetaGraphDef parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.MetaGraphDef parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.MetaGraphDef parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.MetaGraphDef parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.framework.MetaGraphDef prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * <pre>
-   * NOTE: This protocol buffer is evolving, and will go through revisions in the
-   * coming months.
-   * Protocol buffer containing the following which are necessary to restart
-   * training, run inference. It can be used to serialize/de-serialize memory
-   * objects necessary for running computation in a graph when crossing the
-   * process boundary. It can be used for long term storage of graphs,
-   * cross-language execution of graphs, etc.
-   *   MetaInfoDef
-   *   GraphDef
-   *   SaverDef
-   *   CollectionDef
-   *   TensorInfo
-   *   SignatureDef
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.MetaGraphDef}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.MetaGraphDef)
-      org.tensorflow.proto.framework.MetaGraphDefOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.MetaGraphProtos.internal_static_tensorflow_MetaGraphDef_descriptor;
-    }
-
-    @SuppressWarnings({"rawtypes"})
-    protected com.google.protobuf.MapField internalGetMapField(
-        int number) {
-      switch (number) {
-        case 4:
-          return internalGetCollectionDef();
-        case 5:
-          return internalGetSignatureDef();
-        default:
-          throw new RuntimeException(
-              "Invalid map field number: " + number);
-      }
-    }
-    @SuppressWarnings({"rawtypes"})
-    protected com.google.protobuf.MapField internalGetMutableMapField(
-        int number) {
-      switch (number) {
-        case 4:
-          return internalGetMutableCollectionDef();
-        case 5:
-          return internalGetMutableSignatureDef();
-        default:
-          throw new RuntimeException(
-              "Invalid map field number: " + number);
-      }
-    }
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.MetaGraphProtos.internal_static_tensorflow_MetaGraphDef_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.MetaGraphDef.class, org.tensorflow.proto.framework.MetaGraphDef.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.framework.MetaGraphDef.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-        getAssetFileDefFieldBuilder();
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      if (metaInfoDefBuilder_ == null) {
-        metaInfoDef_ = null;
-      } else {
-        metaInfoDef_ = null;
-        metaInfoDefBuilder_ = null;
-      }
-      if (graphDefBuilder_ == null) {
-        graphDef_ = null;
-      } else {
-        graphDef_ = null;
-        graphDefBuilder_ = null;
-      }
-      if (saverDefBuilder_ == null) {
-        saverDef_ = null;
-      } else {
-        saverDef_ = null;
-        saverDefBuilder_ = null;
-      }
-      internalGetMutableCollectionDef().clear();
-      internalGetMutableSignatureDef().clear();
-      if (assetFileDefBuilder_ == null) {
-        assetFileDef_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000004);
-      } else {
-        assetFileDefBuilder_.clear();
-      }
-      if (objectGraphDefBuilder_ == null) {
-        objectGraphDef_ = null;
-      } else {
-        objectGraphDef_ = null;
-        objectGraphDefBuilder_ = null;
-      }
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.framework.MetaGraphProtos.internal_static_tensorflow_MetaGraphDef_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.MetaGraphDef getDefaultInstanceForType() {
-      return org.tensorflow.proto.framework.MetaGraphDef.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.MetaGraphDef build() {
-      org.tensorflow.proto.framework.MetaGraphDef result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.MetaGraphDef buildPartial() {
-      org.tensorflow.proto.framework.MetaGraphDef result = new org.tensorflow.proto.framework.MetaGraphDef(this);
-      int from_bitField0_ = bitField0_;
-      if (metaInfoDefBuilder_ == null) {
-        result.metaInfoDef_ = metaInfoDef_;
-      } else {
-        result.metaInfoDef_ = metaInfoDefBuilder_.build();
-      }
-      if (graphDefBuilder_ == null) {
-        result.graphDef_ = graphDef_;
-      } else {
-        result.graphDef_ = graphDefBuilder_.build();
-      }
-      if (saverDefBuilder_ == null) {
-        result.saverDef_ = saverDef_;
-      } else {
-        result.saverDef_ = saverDefBuilder_.build();
-      }
-      result.collectionDef_ = internalGetCollectionDef();
-      result.collectionDef_.makeImmutable();
-      result.signatureDef_ = internalGetSignatureDef();
-      result.signatureDef_.makeImmutable();
-      if (assetFileDefBuilder_ == null) {
-        if (((bitField0_ & 0x00000004) != 0)) {
-          assetFileDef_ = java.util.Collections.unmodifiableList(assetFileDef_);
-          bitField0_ = (bitField0_ & ~0x00000004);
-        }
-        result.assetFileDef_ = assetFileDef_;
-      } else {
-        result.assetFileDef_ = assetFileDefBuilder_.build();
-      }
-      if (objectGraphDefBuilder_ == null) {
-        result.objectGraphDef_ = objectGraphDef_;
-      } else {
-        result.objectGraphDef_ = objectGraphDefBuilder_.build();
-      }
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.framework.MetaGraphDef) {
-        return mergeFrom((org.tensorflow.proto.framework.MetaGraphDef)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.framework.MetaGraphDef other) {
-      if (other == org.tensorflow.proto.framework.MetaGraphDef.getDefaultInstance()) return this;
-      if (other.hasMetaInfoDef()) {
-        mergeMetaInfoDef(other.getMetaInfoDef());
-      }
-      if (other.hasGraphDef()) {
-        mergeGraphDef(other.getGraphDef());
-      }
-      if (other.hasSaverDef()) {
-        mergeSaverDef(other.getSaverDef());
-      }
-      internalGetMutableCollectionDef().mergeFrom(
-          other.internalGetCollectionDef());
-      internalGetMutableSignatureDef().mergeFrom(
-          other.internalGetSignatureDef());
-      if (assetFileDefBuilder_ == null) {
-        if (!other.assetFileDef_.isEmpty()) {
-          if (assetFileDef_.isEmpty()) {
-            assetFileDef_ = other.assetFileDef_;
-            bitField0_ = (bitField0_ & ~0x00000004);
-          } else {
-            ensureAssetFileDefIsMutable();
-            assetFileDef_.addAll(other.assetFileDef_);
-          }
-          onChanged();
-        }
-      } else {
-        if (!other.assetFileDef_.isEmpty()) {
-          if (assetFileDefBuilder_.isEmpty()) {
-            assetFileDefBuilder_.dispose();
-            assetFileDefBuilder_ = null;
-            assetFileDef_ = other.assetFileDef_;
-            bitField0_ = (bitField0_ & ~0x00000004);
-            assetFileDefBuilder_ = 
-              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
-                 getAssetFileDefFieldBuilder() : null;
-          } else {
-            assetFileDefBuilder_.addAllMessages(other.assetFileDef_);
-          }
-        }
-      }
-      if (other.hasObjectGraphDef()) {
-        mergeObjectGraphDef(other.getObjectGraphDef());
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.framework.MetaGraphDef parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.framework.MetaGraphDef) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-    private int bitField0_;
-
-    private org.tensorflow.proto.framework.MetaGraphDef.MetaInfoDef metaInfoDef_;
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.MetaGraphDef.MetaInfoDef, org.tensorflow.proto.framework.MetaGraphDef.MetaInfoDef.Builder, org.tensorflow.proto.framework.MetaGraphDef.MetaInfoDefOrBuilder> metaInfoDefBuilder_;
-    /**
-     * <code>.tensorflow.MetaGraphDef.MetaInfoDef meta_info_def = 1;</code>
-     */
-    public boolean hasMetaInfoDef() {
-      return metaInfoDefBuilder_ != null || metaInfoDef_ != null;
-    }
-    /**
-     * <code>.tensorflow.MetaGraphDef.MetaInfoDef meta_info_def = 1;</code>
-     */
-    public org.tensorflow.proto.framework.MetaGraphDef.MetaInfoDef getMetaInfoDef() {
-      if (metaInfoDefBuilder_ == null) {
-        return metaInfoDef_ == null ? org.tensorflow.proto.framework.MetaGraphDef.MetaInfoDef.getDefaultInstance() : metaInfoDef_;
-      } else {
-        return metaInfoDefBuilder_.getMessage();
-      }
-    }
-    /**
-     * <code>.tensorflow.MetaGraphDef.MetaInfoDef meta_info_def = 1;</code>
-     */
-    public Builder setMetaInfoDef(org.tensorflow.proto.framework.MetaGraphDef.MetaInfoDef value) {
-      if (metaInfoDefBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        metaInfoDef_ = value;
-        onChanged();
-      } else {
-        metaInfoDefBuilder_.setMessage(value);
-      }
-
-      return this;
-    }
-    /**
-     * <code>.tensorflow.MetaGraphDef.MetaInfoDef meta_info_def = 1;</code>
-     */
-    public Builder setMetaInfoDef(
-        org.tensorflow.proto.framework.MetaGraphDef.MetaInfoDef.Builder builderForValue) {
-      if (metaInfoDefBuilder_ == null) {
-        metaInfoDef_ = builderForValue.build();
-        onChanged();
-      } else {
-        metaInfoDefBuilder_.setMessage(builderForValue.build());
-      }
-
-      return this;
-    }
-    /**
-     * <code>.tensorflow.MetaGraphDef.MetaInfoDef meta_info_def = 1;</code>
-     */
-    public Builder mergeMetaInfoDef(org.tensorflow.proto.framework.MetaGraphDef.MetaInfoDef value) {
-      if (metaInfoDefBuilder_ == null) {
-        if (metaInfoDef_ != null) {
-          metaInfoDef_ =
-            org.tensorflow.proto.framework.MetaGraphDef.MetaInfoDef.newBuilder(metaInfoDef_).mergeFrom(value).buildPartial();
-        } else {
-          metaInfoDef_ = value;
-        }
-        onChanged();
-      } else {
-        metaInfoDefBuilder_.mergeFrom(value);
-      }
-
-      return this;
-    }
-    /**
-     * <code>.tensorflow.MetaGraphDef.MetaInfoDef meta_info_def = 1;</code>
-     */
-    public Builder clearMetaInfoDef() {
-      if (metaInfoDefBuilder_ == null) {
-        metaInfoDef_ = null;
-        onChanged();
-      } else {
-        metaInfoDef_ = null;
-        metaInfoDefBuilder_ = null;
-      }
-
-      return this;
-    }
-    /**
-     * <code>.tensorflow.MetaGraphDef.MetaInfoDef meta_info_def = 1;</code>
-     */
-    public org.tensorflow.proto.framework.MetaGraphDef.MetaInfoDef.Builder getMetaInfoDefBuilder() {
-      
-      onChanged();
-      return getMetaInfoDefFieldBuilder().getBuilder();
-    }
-    /**
-     * <code>.tensorflow.MetaGraphDef.MetaInfoDef meta_info_def = 1;</code>
-     */
-    public org.tensorflow.proto.framework.MetaGraphDef.MetaInfoDefOrBuilder getMetaInfoDefOrBuilder() {
-      if (metaInfoDefBuilder_ != null) {
-        return metaInfoDefBuilder_.getMessageOrBuilder();
-      } else {
-        return metaInfoDef_ == null ?
-            org.tensorflow.proto.framework.MetaGraphDef.MetaInfoDef.getDefaultInstance() : metaInfoDef_;
-      }
-    }
-    /**
-     * <code>.tensorflow.MetaGraphDef.MetaInfoDef meta_info_def = 1;</code>
-     */
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.MetaGraphDef.MetaInfoDef, org.tensorflow.proto.framework.MetaGraphDef.MetaInfoDef.Builder, org.tensorflow.proto.framework.MetaGraphDef.MetaInfoDefOrBuilder> 
-        getMetaInfoDefFieldBuilder() {
-      if (metaInfoDefBuilder_ == null) {
-        metaInfoDefBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-            org.tensorflow.proto.framework.MetaGraphDef.MetaInfoDef, org.tensorflow.proto.framework.MetaGraphDef.MetaInfoDef.Builder, org.tensorflow.proto.framework.MetaGraphDef.MetaInfoDefOrBuilder>(
-                getMetaInfoDef(),
-                getParentForChildren(),
-                isClean());
-        metaInfoDef_ = null;
-      }
-      return metaInfoDefBuilder_;
-    }
-
-    private org.tensorflow.proto.framework.GraphDef graphDef_;
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.GraphDef, org.tensorflow.proto.framework.GraphDef.Builder, org.tensorflow.proto.framework.GraphDefOrBuilder> graphDefBuilder_;
-    /**
-     * <pre>
-     * GraphDef.
-     * </pre>
-     *
-     * <code>.tensorflow.GraphDef graph_def = 2;</code>
-     */
-    public boolean hasGraphDef() {
-      return graphDefBuilder_ != null || graphDef_ != null;
-    }
-    /**
-     * <pre>
-     * GraphDef.
-     * </pre>
-     *
-     * <code>.tensorflow.GraphDef graph_def = 2;</code>
-     */
-    public org.tensorflow.proto.framework.GraphDef getGraphDef() {
-      if (graphDefBuilder_ == null) {
-        return graphDef_ == null ? org.tensorflow.proto.framework.GraphDef.getDefaultInstance() : graphDef_;
-      } else {
-        return graphDefBuilder_.getMessage();
-      }
-    }
-    /**
-     * <pre>
-     * GraphDef.
-     * </pre>
-     *
-     * <code>.tensorflow.GraphDef graph_def = 2;</code>
-     */
-    public Builder setGraphDef(org.tensorflow.proto.framework.GraphDef value) {
-      if (graphDefBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        graphDef_ = value;
-        onChanged();
-      } else {
-        graphDefBuilder_.setMessage(value);
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * GraphDef.
-     * </pre>
-     *
-     * <code>.tensorflow.GraphDef graph_def = 2;</code>
-     */
-    public Builder setGraphDef(
-        org.tensorflow.proto.framework.GraphDef.Builder builderForValue) {
-      if (graphDefBuilder_ == null) {
-        graphDef_ = builderForValue.build();
-        onChanged();
-      } else {
-        graphDefBuilder_.setMessage(builderForValue.build());
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * GraphDef.
-     * </pre>
-     *
-     * <code>.tensorflow.GraphDef graph_def = 2;</code>
-     */
-    public Builder mergeGraphDef(org.tensorflow.proto.framework.GraphDef value) {
-      if (graphDefBuilder_ == null) {
-        if (graphDef_ != null) {
-          graphDef_ =
-            org.tensorflow.proto.framework.GraphDef.newBuilder(graphDef_).mergeFrom(value).buildPartial();
-        } else {
-          graphDef_ = value;
-        }
-        onChanged();
-      } else {
-        graphDefBuilder_.mergeFrom(value);
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * GraphDef.
-     * </pre>
-     *
-     * <code>.tensorflow.GraphDef graph_def = 2;</code>
-     */
-    public Builder clearGraphDef() {
-      if (graphDefBuilder_ == null) {
-        graphDef_ = null;
-        onChanged();
-      } else {
-        graphDef_ = null;
-        graphDefBuilder_ = null;
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * GraphDef.
-     * </pre>
-     *
-     * <code>.tensorflow.GraphDef graph_def = 2;</code>
-     */
-    public org.tensorflow.proto.framework.GraphDef.Builder getGraphDefBuilder() {
-      
-      onChanged();
-      return getGraphDefFieldBuilder().getBuilder();
-    }
-    /**
-     * <pre>
-     * GraphDef.
-     * </pre>
-     *
-     * <code>.tensorflow.GraphDef graph_def = 2;</code>
-     */
-    public org.tensorflow.proto.framework.GraphDefOrBuilder getGraphDefOrBuilder() {
-      if (graphDefBuilder_ != null) {
-        return graphDefBuilder_.getMessageOrBuilder();
-      } else {
-        return graphDef_ == null ?
-            org.tensorflow.proto.framework.GraphDef.getDefaultInstance() : graphDef_;
-      }
-    }
-    /**
-     * <pre>
-     * GraphDef.
-     * </pre>
-     *
-     * <code>.tensorflow.GraphDef graph_def = 2;</code>
-     */
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.GraphDef, org.tensorflow.proto.framework.GraphDef.Builder, org.tensorflow.proto.framework.GraphDefOrBuilder> 
-        getGraphDefFieldBuilder() {
-      if (graphDefBuilder_ == null) {
-        graphDefBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-            org.tensorflow.proto.framework.GraphDef, org.tensorflow.proto.framework.GraphDef.Builder, org.tensorflow.proto.framework.GraphDefOrBuilder>(
-                getGraphDef(),
-                getParentForChildren(),
-                isClean());
-        graphDef_ = null;
-      }
-      return graphDefBuilder_;
-    }
-
-    private org.tensorflow.proto.util.SaverDef saverDef_;
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.util.SaverDef, org.tensorflow.proto.util.SaverDef.Builder, org.tensorflow.proto.util.SaverDefOrBuilder> saverDefBuilder_;
-    /**
-     * <pre>
-     * SaverDef.
-     * </pre>
-     *
-     * <code>.tensorflow.SaverDef saver_def = 3;</code>
-     */
-    public boolean hasSaverDef() {
-      return saverDefBuilder_ != null || saverDef_ != null;
-    }
-    /**
-     * <pre>
-     * SaverDef.
-     * </pre>
-     *
-     * <code>.tensorflow.SaverDef saver_def = 3;</code>
-     */
-    public org.tensorflow.proto.util.SaverDef getSaverDef() {
-      if (saverDefBuilder_ == null) {
-        return saverDef_ == null ? org.tensorflow.proto.util.SaverDef.getDefaultInstance() : saverDef_;
-      } else {
-        return saverDefBuilder_.getMessage();
-      }
-    }
-    /**
-     * <pre>
-     * SaverDef.
-     * </pre>
-     *
-     * <code>.tensorflow.SaverDef saver_def = 3;</code>
-     */
-    public Builder setSaverDef(org.tensorflow.proto.util.SaverDef value) {
-      if (saverDefBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        saverDef_ = value;
-        onChanged();
-      } else {
-        saverDefBuilder_.setMessage(value);
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * SaverDef.
-     * </pre>
-     *
-     * <code>.tensorflow.SaverDef saver_def = 3;</code>
-     */
-    public Builder setSaverDef(
-        org.tensorflow.proto.util.SaverDef.Builder builderForValue) {
-      if (saverDefBuilder_ == null) {
-        saverDef_ = builderForValue.build();
-        onChanged();
-      } else {
-        saverDefBuilder_.setMessage(builderForValue.build());
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * SaverDef.
-     * </pre>
-     *
-     * <code>.tensorflow.SaverDef saver_def = 3;</code>
-     */
-    public Builder mergeSaverDef(org.tensorflow.proto.util.SaverDef value) {
-      if (saverDefBuilder_ == null) {
-        if (saverDef_ != null) {
-          saverDef_ =
-            org.tensorflow.proto.util.SaverDef.newBuilder(saverDef_).mergeFrom(value).buildPartial();
-        } else {
-          saverDef_ = value;
-        }
-        onChanged();
-      } else {
-        saverDefBuilder_.mergeFrom(value);
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * SaverDef.
-     * </pre>
-     *
-     * <code>.tensorflow.SaverDef saver_def = 3;</code>
-     */
-    public Builder clearSaverDef() {
-      if (saverDefBuilder_ == null) {
-        saverDef_ = null;
-        onChanged();
-      } else {
-        saverDef_ = null;
-        saverDefBuilder_ = null;
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * SaverDef.
-     * </pre>
-     *
-     * <code>.tensorflow.SaverDef saver_def = 3;</code>
-     */
-    public org.tensorflow.proto.util.SaverDef.Builder getSaverDefBuilder() {
-      
-      onChanged();
-      return getSaverDefFieldBuilder().getBuilder();
-    }
-    /**
-     * <pre>
-     * SaverDef.
-     * </pre>
-     *
-     * <code>.tensorflow.SaverDef saver_def = 3;</code>
-     */
-    public org.tensorflow.proto.util.SaverDefOrBuilder getSaverDefOrBuilder() {
-      if (saverDefBuilder_ != null) {
-        return saverDefBuilder_.getMessageOrBuilder();
-      } else {
-        return saverDef_ == null ?
-            org.tensorflow.proto.util.SaverDef.getDefaultInstance() : saverDef_;
-      }
-    }
-    /**
-     * <pre>
-     * SaverDef.
-     * </pre>
-     *
-     * <code>.tensorflow.SaverDef saver_def = 3;</code>
-     */
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.util.SaverDef, org.tensorflow.proto.util.SaverDef.Builder, org.tensorflow.proto.util.SaverDefOrBuilder> 
-        getSaverDefFieldBuilder() {
-      if (saverDefBuilder_ == null) {
-        saverDefBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-            org.tensorflow.proto.util.SaverDef, org.tensorflow.proto.util.SaverDef.Builder, org.tensorflow.proto.util.SaverDefOrBuilder>(
-                getSaverDef(),
-                getParentForChildren(),
-                isClean());
-        saverDef_ = null;
-      }
-      return saverDefBuilder_;
-    }
-
-    private com.google.protobuf.MapField<
-        java.lang.String, org.tensorflow.proto.framework.CollectionDef> collectionDef_;
-    private com.google.protobuf.MapField<java.lang.String, org.tensorflow.proto.framework.CollectionDef>
-    internalGetCollectionDef() {
-      if (collectionDef_ == null) {
-        return com.google.protobuf.MapField.emptyMapField(
-            CollectionDefDefaultEntryHolder.defaultEntry);
-      }
-      return collectionDef_;
-    }
-    private com.google.protobuf.MapField<java.lang.String, org.tensorflow.proto.framework.CollectionDef>
-    internalGetMutableCollectionDef() {
-      onChanged();;
-      if (collectionDef_ == null) {
-        collectionDef_ = com.google.protobuf.MapField.newMapField(
-            CollectionDefDefaultEntryHolder.defaultEntry);
-      }
-      if (!collectionDef_.isMutable()) {
-        collectionDef_ = collectionDef_.copy();
-      }
-      return collectionDef_;
-    }
-
-    public int getCollectionDefCount() {
-      return internalGetCollectionDef().getMap().size();
-    }
-    /**
-     * <pre>
-     * collection_def: Map from collection name to collections.
-     * See CollectionDef section for details.
-     * </pre>
-     *
-     * <code>map&lt;string, .tensorflow.CollectionDef&gt; collection_def = 4;</code>
-     */
-
-    public boolean containsCollectionDef(
-        java.lang.String key) {
-      if (key == null) { throw new java.lang.NullPointerException(); }
-      return internalGetCollectionDef().getMap().containsKey(key);
-    }
-    /**
-     * Use {@link #getCollectionDefMap()} instead.
-     */
-    @java.lang.Deprecated
-    public java.util.Map<java.lang.String, org.tensorflow.proto.framework.CollectionDef> getCollectionDef() {
-      return getCollectionDefMap();
-    }
-    /**
-     * <pre>
-     * collection_def: Map from collection name to collections.
-     * See CollectionDef section for details.
-     * </pre>
-     *
-     * <code>map&lt;string, .tensorflow.CollectionDef&gt; collection_def = 4;</code>
-     */
-
-    public java.util.Map<java.lang.String, org.tensorflow.proto.framework.CollectionDef> getCollectionDefMap() {
-      return internalGetCollectionDef().getMap();
-    }
-    /**
-     * <pre>
-     * collection_def: Map from collection name to collections.
-     * See CollectionDef section for details.
-     * </pre>
-     *
-     * <code>map&lt;string, .tensorflow.CollectionDef&gt; collection_def = 4;</code>
-     */
-
-    public org.tensorflow.proto.framework.CollectionDef getCollectionDefOrDefault(
-        java.lang.String key,
-        org.tensorflow.proto.framework.CollectionDef defaultValue) {
-      if (key == null) { throw new java.lang.NullPointerException(); }
-      java.util.Map<java.lang.String, org.tensorflow.proto.framework.CollectionDef> map =
-          internalGetCollectionDef().getMap();
-      return map.containsKey(key) ? map.get(key) : defaultValue;
-    }
-    /**
-     * <pre>
-     * collection_def: Map from collection name to collections.
-     * See CollectionDef section for details.
-     * </pre>
-     *
-     * <code>map&lt;string, .tensorflow.CollectionDef&gt; collection_def = 4;</code>
-     */
-
-    public org.tensorflow.proto.framework.CollectionDef getCollectionDefOrThrow(
-        java.lang.String key) {
-      if (key == null) { throw new java.lang.NullPointerException(); }
-      java.util.Map<java.lang.String, org.tensorflow.proto.framework.CollectionDef> map =
-          internalGetCollectionDef().getMap();
-      if (!map.containsKey(key)) {
-        throw new java.lang.IllegalArgumentException();
-      }
-      return map.get(key);
-    }
-
-    public Builder clearCollectionDef() {
-      internalGetMutableCollectionDef().getMutableMap()
-          .clear();
-      return this;
-    }
-    /**
-     * <pre>
-     * collection_def: Map from collection name to collections.
-     * See CollectionDef section for details.
-     * </pre>
-     *
-     * <code>map&lt;string, .tensorflow.CollectionDef&gt; collection_def = 4;</code>
-     */
-
-    public Builder removeCollectionDef(
-        java.lang.String key) {
-      if (key == null) { throw new java.lang.NullPointerException(); }
-      internalGetMutableCollectionDef().getMutableMap()
-          .remove(key);
-      return this;
-    }
-    /**
-     * Use alternate mutation accessors instead.
-     */
-    @java.lang.Deprecated
-    public java.util.Map<java.lang.String, org.tensorflow.proto.framework.CollectionDef>
-    getMutableCollectionDef() {
-      return internalGetMutableCollectionDef().getMutableMap();
-    }
-    /**
-     * <pre>
-     * collection_def: Map from collection name to collections.
-     * See CollectionDef section for details.
-     * </pre>
-     *
-     * <code>map&lt;string, .tensorflow.CollectionDef&gt; collection_def = 4;</code>
-     */
-    public Builder putCollectionDef(
-        java.lang.String key,
-        org.tensorflow.proto.framework.CollectionDef value) {
-      if (key == null) { throw new java.lang.NullPointerException(); }
-      if (value == null) { throw new java.lang.NullPointerException(); }
-      internalGetMutableCollectionDef().getMutableMap()
-          .put(key, value);
-      return this;
-    }
-    /**
-     * <pre>
-     * collection_def: Map from collection name to collections.
-     * See CollectionDef section for details.
-     * </pre>
-     *
-     * <code>map&lt;string, .tensorflow.CollectionDef&gt; collection_def = 4;</code>
-     */
-
-    public Builder putAllCollectionDef(
-        java.util.Map<java.lang.String, org.tensorflow.proto.framework.CollectionDef> values) {
-      internalGetMutableCollectionDef().getMutableMap()
-          .putAll(values);
-      return this;
-    }
-
-    private com.google.protobuf.MapField<
-        java.lang.String, org.tensorflow.proto.framework.SignatureDef> signatureDef_;
-    private com.google.protobuf.MapField<java.lang.String, org.tensorflow.proto.framework.SignatureDef>
-    internalGetSignatureDef() {
-      if (signatureDef_ == null) {
-        return com.google.protobuf.MapField.emptyMapField(
-            SignatureDefDefaultEntryHolder.defaultEntry);
-      }
-      return signatureDef_;
-    }
-    private com.google.protobuf.MapField<java.lang.String, org.tensorflow.proto.framework.SignatureDef>
-    internalGetMutableSignatureDef() {
-      onChanged();;
-      if (signatureDef_ == null) {
-        signatureDef_ = com.google.protobuf.MapField.newMapField(
-            SignatureDefDefaultEntryHolder.defaultEntry);
-      }
-      if (!signatureDef_.isMutable()) {
-        signatureDef_ = signatureDef_.copy();
-      }
-      return signatureDef_;
-    }
-
-    public int getSignatureDefCount() {
-      return internalGetSignatureDef().getMap().size();
-    }
-    /**
-     * <pre>
-     * signature_def: Map from user supplied key for a signature to a single
-     * SignatureDef.
-     * </pre>
-     *
-     * <code>map&lt;string, .tensorflow.SignatureDef&gt; signature_def = 5;</code>
-     */
-
-    public boolean containsSignatureDef(
-        java.lang.String key) {
-      if (key == null) { throw new java.lang.NullPointerException(); }
-      return internalGetSignatureDef().getMap().containsKey(key);
-    }
-    /**
-     * Use {@link #getSignatureDefMap()} instead.
-     */
-    @java.lang.Deprecated
-    public java.util.Map<java.lang.String, org.tensorflow.proto.framework.SignatureDef> getSignatureDef() {
-      return getSignatureDefMap();
-    }
-    /**
-     * <pre>
-     * signature_def: Map from user supplied key for a signature to a single
-     * SignatureDef.
-     * </pre>
-     *
-     * <code>map&lt;string, .tensorflow.SignatureDef&gt; signature_def = 5;</code>
-     */
-
-    public java.util.Map<java.lang.String, org.tensorflow.proto.framework.SignatureDef> getSignatureDefMap() {
-      return internalGetSignatureDef().getMap();
-    }
-    /**
-     * <pre>
-     * signature_def: Map from user supplied key for a signature to a single
-     * SignatureDef.
-     * </pre>
-     *
-     * <code>map&lt;string, .tensorflow.SignatureDef&gt; signature_def = 5;</code>
-     */
-
-    public org.tensorflow.proto.framework.SignatureDef getSignatureDefOrDefault(
-        java.lang.String key,
-        org.tensorflow.proto.framework.SignatureDef defaultValue) {
-      if (key == null) { throw new java.lang.NullPointerException(); }
-      java.util.Map<java.lang.String, org.tensorflow.proto.framework.SignatureDef> map =
-          internalGetSignatureDef().getMap();
-      return map.containsKey(key) ? map.get(key) : defaultValue;
-    }
-    /**
-     * <pre>
-     * signature_def: Map from user supplied key for a signature to a single
-     * SignatureDef.
-     * </pre>
-     *
-     * <code>map&lt;string, .tensorflow.SignatureDef&gt; signature_def = 5;</code>
-     */
-
-    public org.tensorflow.proto.framework.SignatureDef getSignatureDefOrThrow(
-        java.lang.String key) {
-      if (key == null) { throw new java.lang.NullPointerException(); }
-      java.util.Map<java.lang.String, org.tensorflow.proto.framework.SignatureDef> map =
-          internalGetSignatureDef().getMap();
-      if (!map.containsKey(key)) {
-        throw new java.lang.IllegalArgumentException();
-      }
-      return map.get(key);
-    }
-
-    public Builder clearSignatureDef() {
-      internalGetMutableSignatureDef().getMutableMap()
-          .clear();
-      return this;
-    }
-    /**
-     * <pre>
-     * signature_def: Map from user supplied key for a signature to a single
-     * SignatureDef.
-     * </pre>
-     *
-     * <code>map&lt;string, .tensorflow.SignatureDef&gt; signature_def = 5;</code>
-     */
-
-    public Builder removeSignatureDef(
-        java.lang.String key) {
-      if (key == null) { throw new java.lang.NullPointerException(); }
-      internalGetMutableSignatureDef().getMutableMap()
-          .remove(key);
-      return this;
-    }
-    /**
-     * Use alternate mutation accessors instead.
-     */
-    @java.lang.Deprecated
-    public java.util.Map<java.lang.String, org.tensorflow.proto.framework.SignatureDef>
-    getMutableSignatureDef() {
-      return internalGetMutableSignatureDef().getMutableMap();
-    }
-    /**
-     * <pre>
-     * signature_def: Map from user supplied key for a signature to a single
-     * SignatureDef.
-     * </pre>
-     *
-     * <code>map&lt;string, .tensorflow.SignatureDef&gt; signature_def = 5;</code>
-     */
-    public Builder putSignatureDef(
-        java.lang.String key,
-        org.tensorflow.proto.framework.SignatureDef value) {
-      if (key == null) { throw new java.lang.NullPointerException(); }
-      if (value == null) { throw new java.lang.NullPointerException(); }
-      internalGetMutableSignatureDef().getMutableMap()
-          .put(key, value);
-      return this;
-    }
-    /**
-     * <pre>
-     * signature_def: Map from user supplied key for a signature to a single
-     * SignatureDef.
-     * </pre>
-     *
-     * <code>map&lt;string, .tensorflow.SignatureDef&gt; signature_def = 5;</code>
-     */
-
-    public Builder putAllSignatureDef(
-        java.util.Map<java.lang.String, org.tensorflow.proto.framework.SignatureDef> values) {
-      internalGetMutableSignatureDef().getMutableMap()
-          .putAll(values);
-      return this;
-    }
-
-    private java.util.List<org.tensorflow.proto.framework.AssetFileDef> assetFileDef_ =
-      java.util.Collections.emptyList();
-    private void ensureAssetFileDefIsMutable() {
-      if (!((bitField0_ & 0x00000004) != 0)) {
-        assetFileDef_ = new java.util.ArrayList<org.tensorflow.proto.framework.AssetFileDef>(assetFileDef_);
-        bitField0_ |= 0x00000004;
-       }
-    }
-
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.framework.AssetFileDef, org.tensorflow.proto.framework.AssetFileDef.Builder, org.tensorflow.proto.framework.AssetFileDefOrBuilder> assetFileDefBuilder_;
-
-    /**
-     * <pre>
-     * Asset file def to be used with the defined graph.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.AssetFileDef asset_file_def = 6;</code>
-     */
-    public java.util.List<org.tensorflow.proto.framework.AssetFileDef> getAssetFileDefList() {
-      if (assetFileDefBuilder_ == null) {
-        return java.util.Collections.unmodifiableList(assetFileDef_);
-      } else {
-        return assetFileDefBuilder_.getMessageList();
-      }
-    }
-    /**
-     * <pre>
-     * Asset file def to be used with the defined graph.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.AssetFileDef asset_file_def = 6;</code>
-     */
-    public int getAssetFileDefCount() {
-      if (assetFileDefBuilder_ == null) {
-        return assetFileDef_.size();
-      } else {
-        return assetFileDefBuilder_.getCount();
-      }
-    }
-    /**
-     * <pre>
-     * Asset file def to be used with the defined graph.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.AssetFileDef asset_file_def = 6;</code>
-     */
-    public org.tensorflow.proto.framework.AssetFileDef getAssetFileDef(int index) {
-      if (assetFileDefBuilder_ == null) {
-        return assetFileDef_.get(index);
-      } else {
-        return assetFileDefBuilder_.getMessage(index);
-      }
-    }
-    /**
-     * <pre>
-     * Asset file def to be used with the defined graph.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.AssetFileDef asset_file_def = 6;</code>
-     */
-    public Builder setAssetFileDef(
-        int index, org.tensorflow.proto.framework.AssetFileDef value) {
-      if (assetFileDefBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureAssetFileDefIsMutable();
-        assetFileDef_.set(index, value);
-        onChanged();
-      } else {
-        assetFileDefBuilder_.setMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Asset file def to be used with the defined graph.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.AssetFileDef asset_file_def = 6;</code>
-     */
-    public Builder setAssetFileDef(
-        int index, org.tensorflow.proto.framework.AssetFileDef.Builder builderForValue) {
-      if (assetFileDefBuilder_ == null) {
-        ensureAssetFileDefIsMutable();
-        assetFileDef_.set(index, builderForValue.build());
-        onChanged();
-      } else {
-        assetFileDefBuilder_.setMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Asset file def to be used with the defined graph.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.AssetFileDef asset_file_def = 6;</code>
-     */
-    public Builder addAssetFileDef(org.tensorflow.proto.framework.AssetFileDef value) {
-      if (assetFileDefBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureAssetFileDefIsMutable();
-        assetFileDef_.add(value);
-        onChanged();
-      } else {
-        assetFileDefBuilder_.addMessage(value);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Asset file def to be used with the defined graph.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.AssetFileDef asset_file_def = 6;</code>
-     */
-    public Builder addAssetFileDef(
-        int index, org.tensorflow.proto.framework.AssetFileDef value) {
-      if (assetFileDefBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureAssetFileDefIsMutable();
-        assetFileDef_.add(index, value);
-        onChanged();
-      } else {
-        assetFileDefBuilder_.addMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Asset file def to be used with the defined graph.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.AssetFileDef asset_file_def = 6;</code>
-     */
-    public Builder addAssetFileDef(
-        org.tensorflow.proto.framework.AssetFileDef.Builder builderForValue) {
-      if (assetFileDefBuilder_ == null) {
-        ensureAssetFileDefIsMutable();
-        assetFileDef_.add(builderForValue.build());
-        onChanged();
-      } else {
-        assetFileDefBuilder_.addMessage(builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Asset file def to be used with the defined graph.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.AssetFileDef asset_file_def = 6;</code>
-     */
-    public Builder addAssetFileDef(
-        int index, org.tensorflow.proto.framework.AssetFileDef.Builder builderForValue) {
-      if (assetFileDefBuilder_ == null) {
-        ensureAssetFileDefIsMutable();
-        assetFileDef_.add(index, builderForValue.build());
-        onChanged();
-      } else {
-        assetFileDefBuilder_.addMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Asset file def to be used with the defined graph.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.AssetFileDef asset_file_def = 6;</code>
-     */
-    public Builder addAllAssetFileDef(
-        java.lang.Iterable<? extends org.tensorflow.proto.framework.AssetFileDef> values) {
-      if (assetFileDefBuilder_ == null) {
-        ensureAssetFileDefIsMutable();
-        com.google.protobuf.AbstractMessageLite.Builder.addAll(
-            values, assetFileDef_);
-        onChanged();
-      } else {
-        assetFileDefBuilder_.addAllMessages(values);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Asset file def to be used with the defined graph.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.AssetFileDef asset_file_def = 6;</code>
-     */
-    public Builder clearAssetFileDef() {
-      if (assetFileDefBuilder_ == null) {
-        assetFileDef_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000004);
-        onChanged();
-      } else {
-        assetFileDefBuilder_.clear();
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Asset file def to be used with the defined graph.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.AssetFileDef asset_file_def = 6;</code>
-     */
-    public Builder removeAssetFileDef(int index) {
-      if (assetFileDefBuilder_ == null) {
-        ensureAssetFileDefIsMutable();
-        assetFileDef_.remove(index);
-        onChanged();
-      } else {
-        assetFileDefBuilder_.remove(index);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Asset file def to be used with the defined graph.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.AssetFileDef asset_file_def = 6;</code>
-     */
-    public org.tensorflow.proto.framework.AssetFileDef.Builder getAssetFileDefBuilder(
-        int index) {
-      return getAssetFileDefFieldBuilder().getBuilder(index);
-    }
-    /**
-     * <pre>
-     * Asset file def to be used with the defined graph.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.AssetFileDef asset_file_def = 6;</code>
-     */
-    public org.tensorflow.proto.framework.AssetFileDefOrBuilder getAssetFileDefOrBuilder(
-        int index) {
-      if (assetFileDefBuilder_ == null) {
-        return assetFileDef_.get(index);  } else {
-        return assetFileDefBuilder_.getMessageOrBuilder(index);
-      }
-    }
-    /**
-     * <pre>
-     * Asset file def to be used with the defined graph.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.AssetFileDef asset_file_def = 6;</code>
-     */
-    public java.util.List<? extends org.tensorflow.proto.framework.AssetFileDefOrBuilder> 
-         getAssetFileDefOrBuilderList() {
-      if (assetFileDefBuilder_ != null) {
-        return assetFileDefBuilder_.getMessageOrBuilderList();
-      } else {
-        return java.util.Collections.unmodifiableList(assetFileDef_);
-      }
-    }
-    /**
-     * <pre>
-     * Asset file def to be used with the defined graph.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.AssetFileDef asset_file_def = 6;</code>
-     */
-    public org.tensorflow.proto.framework.AssetFileDef.Builder addAssetFileDefBuilder() {
-      return getAssetFileDefFieldBuilder().addBuilder(
-          org.tensorflow.proto.framework.AssetFileDef.getDefaultInstance());
-    }
-    /**
-     * <pre>
-     * Asset file def to be used with the defined graph.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.AssetFileDef asset_file_def = 6;</code>
-     */
-    public org.tensorflow.proto.framework.AssetFileDef.Builder addAssetFileDefBuilder(
-        int index) {
-      return getAssetFileDefFieldBuilder().addBuilder(
-          index, org.tensorflow.proto.framework.AssetFileDef.getDefaultInstance());
-    }
-    /**
-     * <pre>
-     * Asset file def to be used with the defined graph.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.AssetFileDef asset_file_def = 6;</code>
-     */
-    public java.util.List<org.tensorflow.proto.framework.AssetFileDef.Builder> 
-         getAssetFileDefBuilderList() {
-      return getAssetFileDefFieldBuilder().getBuilderList();
-    }
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.framework.AssetFileDef, org.tensorflow.proto.framework.AssetFileDef.Builder, org.tensorflow.proto.framework.AssetFileDefOrBuilder> 
-        getAssetFileDefFieldBuilder() {
-      if (assetFileDefBuilder_ == null) {
-        assetFileDefBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
-            org.tensorflow.proto.framework.AssetFileDef, org.tensorflow.proto.framework.AssetFileDef.Builder, org.tensorflow.proto.framework.AssetFileDefOrBuilder>(
-                assetFileDef_,
-                ((bitField0_ & 0x00000004) != 0),
-                getParentForChildren(),
-                isClean());
-        assetFileDef_ = null;
-      }
-      return assetFileDefBuilder_;
-    }
-
-    private org.tensorflow.proto.framework.SavedObjectGraph objectGraphDef_;
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.SavedObjectGraph, org.tensorflow.proto.framework.SavedObjectGraph.Builder, org.tensorflow.proto.framework.SavedObjectGraphOrBuilder> objectGraphDefBuilder_;
-    /**
-     * <pre>
-     * Extra information about the structure of functions and stateful objects.
-     * </pre>
-     *
-     * <code>.tensorflow.SavedObjectGraph object_graph_def = 7;</code>
-     */
-    public boolean hasObjectGraphDef() {
-      return objectGraphDefBuilder_ != null || objectGraphDef_ != null;
-    }
-    /**
-     * <pre>
-     * Extra information about the structure of functions and stateful objects.
-     * </pre>
-     *
-     * <code>.tensorflow.SavedObjectGraph object_graph_def = 7;</code>
-     */
-    public org.tensorflow.proto.framework.SavedObjectGraph getObjectGraphDef() {
-      if (objectGraphDefBuilder_ == null) {
-        return objectGraphDef_ == null ? org.tensorflow.proto.framework.SavedObjectGraph.getDefaultInstance() : objectGraphDef_;
-      } else {
-        return objectGraphDefBuilder_.getMessage();
-      }
-    }
-    /**
-     * <pre>
-     * Extra information about the structure of functions and stateful objects.
-     * </pre>
-     *
-     * <code>.tensorflow.SavedObjectGraph object_graph_def = 7;</code>
-     */
-    public Builder setObjectGraphDef(org.tensorflow.proto.framework.SavedObjectGraph value) {
-      if (objectGraphDefBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        objectGraphDef_ = value;
-        onChanged();
-      } else {
-        objectGraphDefBuilder_.setMessage(value);
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Extra information about the structure of functions and stateful objects.
-     * </pre>
-     *
-     * <code>.tensorflow.SavedObjectGraph object_graph_def = 7;</code>
-     */
-    public Builder setObjectGraphDef(
-        org.tensorflow.proto.framework.SavedObjectGraph.Builder builderForValue) {
-      if (objectGraphDefBuilder_ == null) {
-        objectGraphDef_ = builderForValue.build();
-        onChanged();
-      } else {
-        objectGraphDefBuilder_.setMessage(builderForValue.build());
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Extra information about the structure of functions and stateful objects.
-     * </pre>
-     *
-     * <code>.tensorflow.SavedObjectGraph object_graph_def = 7;</code>
-     */
-    public Builder mergeObjectGraphDef(org.tensorflow.proto.framework.SavedObjectGraph value) {
-      if (objectGraphDefBuilder_ == null) {
-        if (objectGraphDef_ != null) {
-          objectGraphDef_ =
-            org.tensorflow.proto.framework.SavedObjectGraph.newBuilder(objectGraphDef_).mergeFrom(value).buildPartial();
-        } else {
-          objectGraphDef_ = value;
-        }
-        onChanged();
-      } else {
-        objectGraphDefBuilder_.mergeFrom(value);
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Extra information about the structure of functions and stateful objects.
-     * </pre>
-     *
-     * <code>.tensorflow.SavedObjectGraph object_graph_def = 7;</code>
-     */
-    public Builder clearObjectGraphDef() {
-      if (objectGraphDefBuilder_ == null) {
-        objectGraphDef_ = null;
-        onChanged();
-      } else {
-        objectGraphDef_ = null;
-        objectGraphDefBuilder_ = null;
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Extra information about the structure of functions and stateful objects.
-     * </pre>
-     *
-     * <code>.tensorflow.SavedObjectGraph object_graph_def = 7;</code>
-     */
-    public org.tensorflow.proto.framework.SavedObjectGraph.Builder getObjectGraphDefBuilder() {
-      
-      onChanged();
-      return getObjectGraphDefFieldBuilder().getBuilder();
-    }
-    /**
-     * <pre>
-     * Extra information about the structure of functions and stateful objects.
-     * </pre>
-     *
-     * <code>.tensorflow.SavedObjectGraph object_graph_def = 7;</code>
-     */
-    public org.tensorflow.proto.framework.SavedObjectGraphOrBuilder getObjectGraphDefOrBuilder() {
-      if (objectGraphDefBuilder_ != null) {
-        return objectGraphDefBuilder_.getMessageOrBuilder();
-      } else {
-        return objectGraphDef_ == null ?
-            org.tensorflow.proto.framework.SavedObjectGraph.getDefaultInstance() : objectGraphDef_;
-      }
-    }
-    /**
-     * <pre>
-     * Extra information about the structure of functions and stateful objects.
-     * </pre>
-     *
-     * <code>.tensorflow.SavedObjectGraph object_graph_def = 7;</code>
-     */
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.SavedObjectGraph, org.tensorflow.proto.framework.SavedObjectGraph.Builder, org.tensorflow.proto.framework.SavedObjectGraphOrBuilder> 
-        getObjectGraphDefFieldBuilder() {
-      if (objectGraphDefBuilder_ == null) {
-        objectGraphDefBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-            org.tensorflow.proto.framework.SavedObjectGraph, org.tensorflow.proto.framework.SavedObjectGraph.Builder, org.tensorflow.proto.framework.SavedObjectGraphOrBuilder>(
-                getObjectGraphDef(),
-                getParentForChildren(),
-                isClean());
-        objectGraphDef_ = null;
-      }
-      return objectGraphDefBuilder_;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.MetaGraphDef)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.MetaGraphDef)
-  private static final org.tensorflow.proto.framework.MetaGraphDef DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.MetaGraphDef();
-  }
-
-  public static org.tensorflow.proto.framework.MetaGraphDef getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<MetaGraphDef>
-      PARSER = new com.google.protobuf.AbstractParser<MetaGraphDef>() {
-    @java.lang.Override
-    public MetaGraphDef parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new MetaGraphDef(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<MetaGraphDef> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<MetaGraphDef> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.framework.MetaGraphDef getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/MetaGraphDefOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/MetaGraphDefOrBuilder.java
deleted file mode 100644
index ca9cf06f97b..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/MetaGraphDefOrBuilder.java
+++ /dev/null
@@ -1,259 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/meta_graph.proto
-
-package org.tensorflow.proto.framework;
-
-public interface MetaGraphDefOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.MetaGraphDef)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <code>.tensorflow.MetaGraphDef.MetaInfoDef meta_info_def = 1;</code>
-   */
-  boolean hasMetaInfoDef();
-  /**
-   * <code>.tensorflow.MetaGraphDef.MetaInfoDef meta_info_def = 1;</code>
-   */
-  org.tensorflow.proto.framework.MetaGraphDef.MetaInfoDef getMetaInfoDef();
-  /**
-   * <code>.tensorflow.MetaGraphDef.MetaInfoDef meta_info_def = 1;</code>
-   */
-  org.tensorflow.proto.framework.MetaGraphDef.MetaInfoDefOrBuilder getMetaInfoDefOrBuilder();
-
-  /**
-   * <pre>
-   * GraphDef.
-   * </pre>
-   *
-   * <code>.tensorflow.GraphDef graph_def = 2;</code>
-   */
-  boolean hasGraphDef();
-  /**
-   * <pre>
-   * GraphDef.
-   * </pre>
-   *
-   * <code>.tensorflow.GraphDef graph_def = 2;</code>
-   */
-  org.tensorflow.proto.framework.GraphDef getGraphDef();
-  /**
-   * <pre>
-   * GraphDef.
-   * </pre>
-   *
-   * <code>.tensorflow.GraphDef graph_def = 2;</code>
-   */
-  org.tensorflow.proto.framework.GraphDefOrBuilder getGraphDefOrBuilder();
-
-  /**
-   * <pre>
-   * SaverDef.
-   * </pre>
-   *
-   * <code>.tensorflow.SaverDef saver_def = 3;</code>
-   */
-  boolean hasSaverDef();
-  /**
-   * <pre>
-   * SaverDef.
-   * </pre>
-   *
-   * <code>.tensorflow.SaverDef saver_def = 3;</code>
-   */
-  org.tensorflow.proto.util.SaverDef getSaverDef();
-  /**
-   * <pre>
-   * SaverDef.
-   * </pre>
-   *
-   * <code>.tensorflow.SaverDef saver_def = 3;</code>
-   */
-  org.tensorflow.proto.util.SaverDefOrBuilder getSaverDefOrBuilder();
-
-  /**
-   * <pre>
-   * collection_def: Map from collection name to collections.
-   * See CollectionDef section for details.
-   * </pre>
-   *
-   * <code>map&lt;string, .tensorflow.CollectionDef&gt; collection_def = 4;</code>
-   */
-  int getCollectionDefCount();
-  /**
-   * <pre>
-   * collection_def: Map from collection name to collections.
-   * See CollectionDef section for details.
-   * </pre>
-   *
-   * <code>map&lt;string, .tensorflow.CollectionDef&gt; collection_def = 4;</code>
-   */
-  boolean containsCollectionDef(
-      java.lang.String key);
-  /**
-   * Use {@link #getCollectionDefMap()} instead.
-   */
-  @java.lang.Deprecated
-  java.util.Map<java.lang.String, org.tensorflow.proto.framework.CollectionDef>
-  getCollectionDef();
-  /**
-   * <pre>
-   * collection_def: Map from collection name to collections.
-   * See CollectionDef section for details.
-   * </pre>
-   *
-   * <code>map&lt;string, .tensorflow.CollectionDef&gt; collection_def = 4;</code>
-   */
-  java.util.Map<java.lang.String, org.tensorflow.proto.framework.CollectionDef>
-  getCollectionDefMap();
-  /**
-   * <pre>
-   * collection_def: Map from collection name to collections.
-   * See CollectionDef section for details.
-   * </pre>
-   *
-   * <code>map&lt;string, .tensorflow.CollectionDef&gt; collection_def = 4;</code>
-   */
-
-  org.tensorflow.proto.framework.CollectionDef getCollectionDefOrDefault(
-      java.lang.String key,
-      org.tensorflow.proto.framework.CollectionDef defaultValue);
-  /**
-   * <pre>
-   * collection_def: Map from collection name to collections.
-   * See CollectionDef section for details.
-   * </pre>
-   *
-   * <code>map&lt;string, .tensorflow.CollectionDef&gt; collection_def = 4;</code>
-   */
-
-  org.tensorflow.proto.framework.CollectionDef getCollectionDefOrThrow(
-      java.lang.String key);
-
-  /**
-   * <pre>
-   * signature_def: Map from user supplied key for a signature to a single
-   * SignatureDef.
-   * </pre>
-   *
-   * <code>map&lt;string, .tensorflow.SignatureDef&gt; signature_def = 5;</code>
-   */
-  int getSignatureDefCount();
-  /**
-   * <pre>
-   * signature_def: Map from user supplied key for a signature to a single
-   * SignatureDef.
-   * </pre>
-   *
-   * <code>map&lt;string, .tensorflow.SignatureDef&gt; signature_def = 5;</code>
-   */
-  boolean containsSignatureDef(
-      java.lang.String key);
-  /**
-   * Use {@link #getSignatureDefMap()} instead.
-   */
-  @java.lang.Deprecated
-  java.util.Map<java.lang.String, org.tensorflow.proto.framework.SignatureDef>
-  getSignatureDef();
-  /**
-   * <pre>
-   * signature_def: Map from user supplied key for a signature to a single
-   * SignatureDef.
-   * </pre>
-   *
-   * <code>map&lt;string, .tensorflow.SignatureDef&gt; signature_def = 5;</code>
-   */
-  java.util.Map<java.lang.String, org.tensorflow.proto.framework.SignatureDef>
-  getSignatureDefMap();
-  /**
-   * <pre>
-   * signature_def: Map from user supplied key for a signature to a single
-   * SignatureDef.
-   * </pre>
-   *
-   * <code>map&lt;string, .tensorflow.SignatureDef&gt; signature_def = 5;</code>
-   */
-
-  org.tensorflow.proto.framework.SignatureDef getSignatureDefOrDefault(
-      java.lang.String key,
-      org.tensorflow.proto.framework.SignatureDef defaultValue);
-  /**
-   * <pre>
-   * signature_def: Map from user supplied key for a signature to a single
-   * SignatureDef.
-   * </pre>
-   *
-   * <code>map&lt;string, .tensorflow.SignatureDef&gt; signature_def = 5;</code>
-   */
-
-  org.tensorflow.proto.framework.SignatureDef getSignatureDefOrThrow(
-      java.lang.String key);
-
-  /**
-   * <pre>
-   * Asset file def to be used with the defined graph.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.AssetFileDef asset_file_def = 6;</code>
-   */
-  java.util.List<org.tensorflow.proto.framework.AssetFileDef> 
-      getAssetFileDefList();
-  /**
-   * <pre>
-   * Asset file def to be used with the defined graph.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.AssetFileDef asset_file_def = 6;</code>
-   */
-  org.tensorflow.proto.framework.AssetFileDef getAssetFileDef(int index);
-  /**
-   * <pre>
-   * Asset file def to be used with the defined graph.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.AssetFileDef asset_file_def = 6;</code>
-   */
-  int getAssetFileDefCount();
-  /**
-   * <pre>
-   * Asset file def to be used with the defined graph.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.AssetFileDef asset_file_def = 6;</code>
-   */
-  java.util.List<? extends org.tensorflow.proto.framework.AssetFileDefOrBuilder> 
-      getAssetFileDefOrBuilderList();
-  /**
-   * <pre>
-   * Asset file def to be used with the defined graph.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.AssetFileDef asset_file_def = 6;</code>
-   */
-  org.tensorflow.proto.framework.AssetFileDefOrBuilder getAssetFileDefOrBuilder(
-      int index);
-
-  /**
-   * <pre>
-   * Extra information about the structure of functions and stateful objects.
-   * </pre>
-   *
-   * <code>.tensorflow.SavedObjectGraph object_graph_def = 7;</code>
-   */
-  boolean hasObjectGraphDef();
-  /**
-   * <pre>
-   * Extra information about the structure of functions and stateful objects.
-   * </pre>
-   *
-   * <code>.tensorflow.SavedObjectGraph object_graph_def = 7;</code>
-   */
-  org.tensorflow.proto.framework.SavedObjectGraph getObjectGraphDef();
-  /**
-   * <pre>
-   * Extra information about the structure of functions and stateful objects.
-   * </pre>
-   *
-   * <code>.tensorflow.SavedObjectGraph object_graph_def = 7;</code>
-   */
-  org.tensorflow.proto.framework.SavedObjectGraphOrBuilder getObjectGraphDefOrBuilder();
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/MetaGraphProtos.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/MetaGraphProtos.java
deleted file mode 100644
index cd79fd2f986..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/MetaGraphProtos.java
+++ /dev/null
@@ -1,318 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/meta_graph.proto
-
-package org.tensorflow.proto.framework;
-
-public final class MetaGraphProtos {
-  private MetaGraphProtos() {}
-  public static void registerAllExtensions(
-      com.google.protobuf.ExtensionRegistryLite registry) {
-  }
-
-  public static void registerAllExtensions(
-      com.google.protobuf.ExtensionRegistry registry) {
-    registerAllExtensions(
-        (com.google.protobuf.ExtensionRegistryLite) registry);
-  }
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_MetaGraphDef_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_MetaGraphDef_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_MetaGraphDef_MetaInfoDef_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_MetaGraphDef_MetaInfoDef_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_MetaGraphDef_MetaInfoDef_FunctionAliasesEntry_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_MetaGraphDef_MetaInfoDef_FunctionAliasesEntry_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_MetaGraphDef_CollectionDefEntry_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_MetaGraphDef_CollectionDefEntry_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_MetaGraphDef_SignatureDefEntry_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_MetaGraphDef_SignatureDefEntry_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_CollectionDef_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_CollectionDef_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_CollectionDef_NodeList_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_CollectionDef_NodeList_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_CollectionDef_BytesList_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_CollectionDef_BytesList_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_CollectionDef_Int64List_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_CollectionDef_Int64List_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_CollectionDef_FloatList_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_CollectionDef_FloatList_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_CollectionDef_AnyList_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_CollectionDef_AnyList_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_TensorInfo_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_TensorInfo_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_TensorInfo_CooSparse_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_TensorInfo_CooSparse_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_TensorInfo_CompositeTensor_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_TensorInfo_CompositeTensor_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_SignatureDef_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_SignatureDef_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_SignatureDef_InputsEntry_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_SignatureDef_InputsEntry_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_SignatureDef_OutputsEntry_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_SignatureDef_OutputsEntry_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_AssetFileDef_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_AssetFileDef_fieldAccessorTable;
-
-  public static com.google.protobuf.Descriptors.FileDescriptor
-      getDescriptor() {
-    return descriptor;
-  }
-  private static  com.google.protobuf.Descriptors.FileDescriptor
-      descriptor;
-  static {
-    java.lang.String[] descriptorData = {
-      "\n)tensorflow/core/protobuf/meta_graph.pr" +
-      "oto\022\ntensorflow\032\031google/protobuf/any.pro" +
-      "to\032%tensorflow/core/framework/graph.prot" +
-      "o\032&tensorflow/core/framework/op_def.prot" +
-      "o\032,tensorflow/core/framework/tensor_shap" +
-      "e.proto\032%tensorflow/core/framework/types" +
-      ".proto\0321tensorflow/core/protobuf/saved_o" +
-      "bject_graph.proto\032$tensorflow/core/proto" +
-      "buf/saver.proto\032%tensorflow/core/protobu" +
-      "f/struct.proto\"\250\007\n\014MetaGraphDef\022;\n\rmeta_" +
-      "info_def\030\001 \001(\0132$.tensorflow.MetaGraphDef" +
-      ".MetaInfoDef\022\'\n\tgraph_def\030\002 \001(\0132\024.tensor" +
-      "flow.GraphDef\022\'\n\tsaver_def\030\003 \001(\0132\024.tenso" +
-      "rflow.SaverDef\022C\n\016collection_def\030\004 \003(\0132+" +
-      ".tensorflow.MetaGraphDef.CollectionDefEn" +
-      "try\022A\n\rsignature_def\030\005 \003(\0132*.tensorflow." +
-      "MetaGraphDef.SignatureDefEntry\0220\n\016asset_" +
-      "file_def\030\006 \003(\0132\030.tensorflow.AssetFileDef" +
-      "\0226\n\020object_graph_def\030\007 \001(\0132\034.tensorflow." +
-      "SavedObjectGraph\032\366\002\n\013MetaInfoDef\022\032\n\022meta" +
-      "_graph_version\030\001 \001(\t\022,\n\020stripped_op_list" +
-      "\030\002 \001(\0132\022.tensorflow.OpList\022&\n\010any_info\030\003" +
-      " \001(\0132\024.google.protobuf.Any\022\014\n\004tags\030\004 \003(\t" +
-      "\022\032\n\022tensorflow_version\030\005 \001(\t\022\036\n\026tensorfl" +
-      "ow_git_version\030\006 \001(\t\022\036\n\026stripped_default" +
-      "_attrs\030\007 \001(\010\022S\n\020function_aliases\030\010 \003(\01329" +
-      ".tensorflow.MetaGraphDef.MetaInfoDef.Fun" +
-      "ctionAliasesEntry\0326\n\024FunctionAliasesEntr" +
-      "y\022\013\n\003key\030\001 \001(\t\022\r\n\005value\030\002 \001(\t:\0028\001\032O\n\022Col" +
-      "lectionDefEntry\022\013\n\003key\030\001 \001(\t\022(\n\005value\030\002 " +
-      "\001(\0132\031.tensorflow.CollectionDef:\0028\001\032M\n\021Si" +
-      "gnatureDefEntry\022\013\n\003key\030\001 \001(\t\022\'\n\005value\030\002 " +
-      "\001(\0132\030.tensorflow.SignatureDef:\0028\001\"\337\003\n\rCo" +
-      "llectionDef\0227\n\tnode_list\030\001 \001(\0132\".tensorf" +
-      "low.CollectionDef.NodeListH\000\0229\n\nbytes_li" +
-      "st\030\002 \001(\0132#.tensorflow.CollectionDef.Byte" +
-      "sListH\000\0229\n\nint64_list\030\003 \001(\0132#.tensorflow" +
-      ".CollectionDef.Int64ListH\000\0229\n\nfloat_list" +
-      "\030\004 \001(\0132#.tensorflow.CollectionDef.FloatL" +
-      "istH\000\0225\n\010any_list\030\005 \001(\0132!.tensorflow.Col" +
-      "lectionDef.AnyListH\000\032\031\n\010NodeList\022\r\n\005valu" +
-      "e\030\001 \003(\t\032\032\n\tBytesList\022\r\n\005value\030\001 \003(\014\032\036\n\tI" +
-      "nt64List\022\021\n\005value\030\001 \003(\003B\002\020\001\032\036\n\tFloatList" +
-      "\022\021\n\005value\030\001 \003(\002B\002\020\001\032.\n\007AnyList\022#\n\005value\030" +
-      "\001 \003(\0132\024.google.protobuf.AnyB\006\n\004kind\"\321\003\n\n" +
-      "TensorInfo\022\016\n\004name\030\001 \001(\tH\000\0226\n\ncoo_sparse" +
-      "\030\004 \001(\0132 .tensorflow.TensorInfo.CooSparse" +
-      "H\000\022B\n\020composite_tensor\030\005 \001(\0132&.tensorflo" +
-      "w.TensorInfo.CompositeTensorH\000\022#\n\005dtype\030" +
-      "\002 \001(\0162\024.tensorflow.DataType\0222\n\014tensor_sh" +
-      "ape\030\003 \001(\0132\034.tensorflow.TensorShapeProto\032" +
-      "e\n\tCooSparse\022\032\n\022values_tensor_name\030\001 \001(\t" +
-      "\022\033\n\023indices_tensor_name\030\002 \001(\t\022\037\n\027dense_s" +
-      "hape_tensor_name\030\003 \001(\t\032k\n\017CompositeTenso" +
-      "r\022,\n\ttype_spec\030\001 \001(\0132\031.tensorflow.TypeSp" +
-      "ecProto\022*\n\ncomponents\030\002 \003(\0132\026.tensorflow" +
-      ".TensorInfoB\n\n\010encoding\"\240\002\n\014SignatureDef" +
-      "\0224\n\006inputs\030\001 \003(\0132$.tensorflow.SignatureD" +
-      "ef.InputsEntry\0226\n\007outputs\030\002 \003(\0132%.tensor" +
-      "flow.SignatureDef.OutputsEntry\022\023\n\013method" +
-      "_name\030\003 \001(\t\032E\n\013InputsEntry\022\013\n\003key\030\001 \001(\t\022" +
-      "%\n\005value\030\002 \001(\0132\026.tensorflow.TensorInfo:\002" +
-      "8\001\032F\n\014OutputsEntry\022\013\n\003key\030\001 \001(\t\022%\n\005value" +
-      "\030\002 \001(\0132\026.tensorflow.TensorInfo:\0028\001\"M\n\014As" +
-      "setFileDef\022+\n\013tensor_info\030\001 \001(\0132\026.tensor" +
-      "flow.TensorInfo\022\020\n\010filename\030\002 \001(\tB\200\001\n\036or" +
-      "g.tensorflow.proto.frameworkB\017MetaGraphP" +
-      "rotosP\001ZHgithub.com/tensorflow/tensorflo" +
-      "w/tensorflow/go/core/core_protos_go_prot" +
-      "o\370\001\001b\006proto3"
-    };
-    descriptor = com.google.protobuf.Descriptors.FileDescriptor
-      .internalBuildGeneratedFileFrom(descriptorData,
-        new com.google.protobuf.Descriptors.FileDescriptor[] {
-          com.google.protobuf.AnyProto.getDescriptor(),
-          org.tensorflow.proto.framework.GraphProtos.getDescriptor(),
-          org.tensorflow.proto.framework.OpDefProtos.getDescriptor(),
-          org.tensorflow.proto.framework.TensorShapeProtos.getDescriptor(),
-          org.tensorflow.proto.framework.TypesProtos.getDescriptor(),
-          org.tensorflow.proto.framework.SavedObjectGraphProtos.getDescriptor(),
-          org.tensorflow.proto.util.SaverProtos.getDescriptor(),
-          org.tensorflow.proto.framework.StructProtos.getDescriptor(),
-        });
-    internal_static_tensorflow_MetaGraphDef_descriptor =
-      getDescriptor().getMessageTypes().get(0);
-    internal_static_tensorflow_MetaGraphDef_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_MetaGraphDef_descriptor,
-        new java.lang.String[] { "MetaInfoDef", "GraphDef", "SaverDef", "CollectionDef", "SignatureDef", "AssetFileDef", "ObjectGraphDef", });
-    internal_static_tensorflow_MetaGraphDef_MetaInfoDef_descriptor =
-      internal_static_tensorflow_MetaGraphDef_descriptor.getNestedTypes().get(0);
-    internal_static_tensorflow_MetaGraphDef_MetaInfoDef_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_MetaGraphDef_MetaInfoDef_descriptor,
-        new java.lang.String[] { "MetaGraphVersion", "StrippedOpList", "AnyInfo", "Tags", "TensorflowVersion", "TensorflowGitVersion", "StrippedDefaultAttrs", "FunctionAliases", });
-    internal_static_tensorflow_MetaGraphDef_MetaInfoDef_FunctionAliasesEntry_descriptor =
-      internal_static_tensorflow_MetaGraphDef_MetaInfoDef_descriptor.getNestedTypes().get(0);
-    internal_static_tensorflow_MetaGraphDef_MetaInfoDef_FunctionAliasesEntry_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_MetaGraphDef_MetaInfoDef_FunctionAliasesEntry_descriptor,
-        new java.lang.String[] { "Key", "Value", });
-    internal_static_tensorflow_MetaGraphDef_CollectionDefEntry_descriptor =
-      internal_static_tensorflow_MetaGraphDef_descriptor.getNestedTypes().get(1);
-    internal_static_tensorflow_MetaGraphDef_CollectionDefEntry_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_MetaGraphDef_CollectionDefEntry_descriptor,
-        new java.lang.String[] { "Key", "Value", });
-    internal_static_tensorflow_MetaGraphDef_SignatureDefEntry_descriptor =
-      internal_static_tensorflow_MetaGraphDef_descriptor.getNestedTypes().get(2);
-    internal_static_tensorflow_MetaGraphDef_SignatureDefEntry_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_MetaGraphDef_SignatureDefEntry_descriptor,
-        new java.lang.String[] { "Key", "Value", });
-    internal_static_tensorflow_CollectionDef_descriptor =
-      getDescriptor().getMessageTypes().get(1);
-    internal_static_tensorflow_CollectionDef_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_CollectionDef_descriptor,
-        new java.lang.String[] { "NodeList", "BytesList", "Int64List", "FloatList", "AnyList", "Kind", });
-    internal_static_tensorflow_CollectionDef_NodeList_descriptor =
-      internal_static_tensorflow_CollectionDef_descriptor.getNestedTypes().get(0);
-    internal_static_tensorflow_CollectionDef_NodeList_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_CollectionDef_NodeList_descriptor,
-        new java.lang.String[] { "Value", });
-    internal_static_tensorflow_CollectionDef_BytesList_descriptor =
-      internal_static_tensorflow_CollectionDef_descriptor.getNestedTypes().get(1);
-    internal_static_tensorflow_CollectionDef_BytesList_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_CollectionDef_BytesList_descriptor,
-        new java.lang.String[] { "Value", });
-    internal_static_tensorflow_CollectionDef_Int64List_descriptor =
-      internal_static_tensorflow_CollectionDef_descriptor.getNestedTypes().get(2);
-    internal_static_tensorflow_CollectionDef_Int64List_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_CollectionDef_Int64List_descriptor,
-        new java.lang.String[] { "Value", });
-    internal_static_tensorflow_CollectionDef_FloatList_descriptor =
-      internal_static_tensorflow_CollectionDef_descriptor.getNestedTypes().get(3);
-    internal_static_tensorflow_CollectionDef_FloatList_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_CollectionDef_FloatList_descriptor,
-        new java.lang.String[] { "Value", });
-    internal_static_tensorflow_CollectionDef_AnyList_descriptor =
-      internal_static_tensorflow_CollectionDef_descriptor.getNestedTypes().get(4);
-    internal_static_tensorflow_CollectionDef_AnyList_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_CollectionDef_AnyList_descriptor,
-        new java.lang.String[] { "Value", });
-    internal_static_tensorflow_TensorInfo_descriptor =
-      getDescriptor().getMessageTypes().get(2);
-    internal_static_tensorflow_TensorInfo_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_TensorInfo_descriptor,
-        new java.lang.String[] { "Name", "CooSparse", "CompositeTensor", "Dtype", "TensorShape", "Encoding", });
-    internal_static_tensorflow_TensorInfo_CooSparse_descriptor =
-      internal_static_tensorflow_TensorInfo_descriptor.getNestedTypes().get(0);
-    internal_static_tensorflow_TensorInfo_CooSparse_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_TensorInfo_CooSparse_descriptor,
-        new java.lang.String[] { "ValuesTensorName", "IndicesTensorName", "DenseShapeTensorName", });
-    internal_static_tensorflow_TensorInfo_CompositeTensor_descriptor =
-      internal_static_tensorflow_TensorInfo_descriptor.getNestedTypes().get(1);
-    internal_static_tensorflow_TensorInfo_CompositeTensor_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_TensorInfo_CompositeTensor_descriptor,
-        new java.lang.String[] { "TypeSpec", "Components", });
-    internal_static_tensorflow_SignatureDef_descriptor =
-      getDescriptor().getMessageTypes().get(3);
-    internal_static_tensorflow_SignatureDef_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_SignatureDef_descriptor,
-        new java.lang.String[] { "Inputs", "Outputs", "MethodName", });
-    internal_static_tensorflow_SignatureDef_InputsEntry_descriptor =
-      internal_static_tensorflow_SignatureDef_descriptor.getNestedTypes().get(0);
-    internal_static_tensorflow_SignatureDef_InputsEntry_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_SignatureDef_InputsEntry_descriptor,
-        new java.lang.String[] { "Key", "Value", });
-    internal_static_tensorflow_SignatureDef_OutputsEntry_descriptor =
-      internal_static_tensorflow_SignatureDef_descriptor.getNestedTypes().get(1);
-    internal_static_tensorflow_SignatureDef_OutputsEntry_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_SignatureDef_OutputsEntry_descriptor,
-        new java.lang.String[] { "Key", "Value", });
-    internal_static_tensorflow_AssetFileDef_descriptor =
-      getDescriptor().getMessageTypes().get(4);
-    internal_static_tensorflow_AssetFileDef_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_AssetFileDef_descriptor,
-        new java.lang.String[] { "TensorInfo", "Filename", });
-    com.google.protobuf.AnyProto.getDescriptor();
-    org.tensorflow.proto.framework.GraphProtos.getDescriptor();
-    org.tensorflow.proto.framework.OpDefProtos.getDescriptor();
-    org.tensorflow.proto.framework.TensorShapeProtos.getDescriptor();
-    org.tensorflow.proto.framework.TypesProtos.getDescriptor();
-    org.tensorflow.proto.framework.SavedObjectGraphProtos.getDescriptor();
-    org.tensorflow.proto.util.SaverProtos.getDescriptor();
-    org.tensorflow.proto.framework.StructProtos.getDescriptor();
-  }
-
-  // @@protoc_insertion_point(outer_class_scope)
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NameAttrList.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NameAttrList.java
deleted file mode 100644
index a17b31f8aa1..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NameAttrList.java
+++ /dev/null
@@ -1,832 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/framework/attr_value.proto
-
-package org.tensorflow.proto.framework;
-
-/**
- * <pre>
- * A list of attr names and their values. The whole list is attached
- * with a string name.  E.g., MatMul[T=float].
- * </pre>
- *
- * Protobuf type {@code tensorflow.NameAttrList}
- */
-public  final class NameAttrList extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.NameAttrList)
-    NameAttrListOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use NameAttrList.newBuilder() to construct.
-  private NameAttrList(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private NameAttrList() {
-    name_ = "";
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new NameAttrList();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private NameAttrList(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    int mutable_bitField0_ = 0;
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 10: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            name_ = s;
-            break;
-          }
-          case 18: {
-            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
-              attr_ = com.google.protobuf.MapField.newMapField(
-                  AttrDefaultEntryHolder.defaultEntry);
-              mutable_bitField0_ |= 0x00000001;
-            }
-            com.google.protobuf.MapEntry<java.lang.String, org.tensorflow.proto.framework.AttrValue>
-            attr__ = input.readMessage(
-                AttrDefaultEntryHolder.defaultEntry.getParserForType(), extensionRegistry);
-            attr_.getMutableMap().put(
-                attr__.getKey(), attr__.getValue());
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.framework.AttrValueProtos.internal_static_tensorflow_NameAttrList_descriptor;
-  }
-
-  @SuppressWarnings({"rawtypes"})
-  @java.lang.Override
-  protected com.google.protobuf.MapField internalGetMapField(
-      int number) {
-    switch (number) {
-      case 2:
-        return internalGetAttr();
-      default:
-        throw new RuntimeException(
-            "Invalid map field number: " + number);
-    }
-  }
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.framework.AttrValueProtos.internal_static_tensorflow_NameAttrList_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.framework.NameAttrList.class, org.tensorflow.proto.framework.NameAttrList.Builder.class);
-  }
-
-  public static final int NAME_FIELD_NUMBER = 1;
-  private volatile java.lang.Object name_;
-  /**
-   * <code>string name = 1;</code>
-   */
-  public java.lang.String getName() {
-    java.lang.Object ref = name_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      name_ = s;
-      return s;
-    }
-  }
-  /**
-   * <code>string name = 1;</code>
-   */
-  public com.google.protobuf.ByteString
-      getNameBytes() {
-    java.lang.Object ref = name_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      name_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int ATTR_FIELD_NUMBER = 2;
-  private static final class AttrDefaultEntryHolder {
-    static final com.google.protobuf.MapEntry<
-        java.lang.String, org.tensorflow.proto.framework.AttrValue> defaultEntry =
-            com.google.protobuf.MapEntry
-            .<java.lang.String, org.tensorflow.proto.framework.AttrValue>newDefaultInstance(
-                org.tensorflow.proto.framework.AttrValueProtos.internal_static_tensorflow_NameAttrList_AttrEntry_descriptor, 
-                com.google.protobuf.WireFormat.FieldType.STRING,
-                "",
-                com.google.protobuf.WireFormat.FieldType.MESSAGE,
-                org.tensorflow.proto.framework.AttrValue.getDefaultInstance());
-  }
-  private com.google.protobuf.MapField<
-      java.lang.String, org.tensorflow.proto.framework.AttrValue> attr_;
-  private com.google.protobuf.MapField<java.lang.String, org.tensorflow.proto.framework.AttrValue>
-  internalGetAttr() {
-    if (attr_ == null) {
-      return com.google.protobuf.MapField.emptyMapField(
-          AttrDefaultEntryHolder.defaultEntry);
-    }
-    return attr_;
-  }
-
-  public int getAttrCount() {
-    return internalGetAttr().getMap().size();
-  }
-  /**
-   * <code>map&lt;string, .tensorflow.AttrValue&gt; attr = 2;</code>
-   */
-
-  public boolean containsAttr(
-      java.lang.String key) {
-    if (key == null) { throw new java.lang.NullPointerException(); }
-    return internalGetAttr().getMap().containsKey(key);
-  }
-  /**
-   * Use {@link #getAttrMap()} instead.
-   */
-  @java.lang.Deprecated
-  public java.util.Map<java.lang.String, org.tensorflow.proto.framework.AttrValue> getAttr() {
-    return getAttrMap();
-  }
-  /**
-   * <code>map&lt;string, .tensorflow.AttrValue&gt; attr = 2;</code>
-   */
-
-  public java.util.Map<java.lang.String, org.tensorflow.proto.framework.AttrValue> getAttrMap() {
-    return internalGetAttr().getMap();
-  }
-  /**
-   * <code>map&lt;string, .tensorflow.AttrValue&gt; attr = 2;</code>
-   */
-
-  public org.tensorflow.proto.framework.AttrValue getAttrOrDefault(
-      java.lang.String key,
-      org.tensorflow.proto.framework.AttrValue defaultValue) {
-    if (key == null) { throw new java.lang.NullPointerException(); }
-    java.util.Map<java.lang.String, org.tensorflow.proto.framework.AttrValue> map =
-        internalGetAttr().getMap();
-    return map.containsKey(key) ? map.get(key) : defaultValue;
-  }
-  /**
-   * <code>map&lt;string, .tensorflow.AttrValue&gt; attr = 2;</code>
-   */
-
-  public org.tensorflow.proto.framework.AttrValue getAttrOrThrow(
-      java.lang.String key) {
-    if (key == null) { throw new java.lang.NullPointerException(); }
-    java.util.Map<java.lang.String, org.tensorflow.proto.framework.AttrValue> map =
-        internalGetAttr().getMap();
-    if (!map.containsKey(key)) {
-      throw new java.lang.IllegalArgumentException();
-    }
-    return map.get(key);
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    if (!getNameBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 1, name_);
-    }
-    com.google.protobuf.GeneratedMessageV3
-      .serializeStringMapTo(
-        output,
-        internalGetAttr(),
-        AttrDefaultEntryHolder.defaultEntry,
-        2);
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (!getNameBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, name_);
-    }
-    for (java.util.Map.Entry<java.lang.String, org.tensorflow.proto.framework.AttrValue> entry
-         : internalGetAttr().getMap().entrySet()) {
-      com.google.protobuf.MapEntry<java.lang.String, org.tensorflow.proto.framework.AttrValue>
-      attr__ = AttrDefaultEntryHolder.defaultEntry.newBuilderForType()
-          .setKey(entry.getKey())
-          .setValue(entry.getValue())
-          .build();
-      size += com.google.protobuf.CodedOutputStream
-          .computeMessageSize(2, attr__);
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.framework.NameAttrList)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.framework.NameAttrList other = (org.tensorflow.proto.framework.NameAttrList) obj;
-
-    if (!getName()
-        .equals(other.getName())) return false;
-    if (!internalGetAttr().equals(
-        other.internalGetAttr())) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    hash = (37 * hash) + NAME_FIELD_NUMBER;
-    hash = (53 * hash) + getName().hashCode();
-    if (!internalGetAttr().getMap().isEmpty()) {
-      hash = (37 * hash) + ATTR_FIELD_NUMBER;
-      hash = (53 * hash) + internalGetAttr().hashCode();
-    }
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.framework.NameAttrList parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.NameAttrList parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.NameAttrList parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.NameAttrList parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.NameAttrList parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.NameAttrList parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.NameAttrList parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.NameAttrList parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.NameAttrList parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.NameAttrList parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.NameAttrList parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.NameAttrList parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.framework.NameAttrList prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * <pre>
-   * A list of attr names and their values. The whole list is attached
-   * with a string name.  E.g., MatMul[T=float].
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.NameAttrList}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.NameAttrList)
-      org.tensorflow.proto.framework.NameAttrListOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.AttrValueProtos.internal_static_tensorflow_NameAttrList_descriptor;
-    }
-
-    @SuppressWarnings({"rawtypes"})
-    protected com.google.protobuf.MapField internalGetMapField(
-        int number) {
-      switch (number) {
-        case 2:
-          return internalGetAttr();
-        default:
-          throw new RuntimeException(
-              "Invalid map field number: " + number);
-      }
-    }
-    @SuppressWarnings({"rawtypes"})
-    protected com.google.protobuf.MapField internalGetMutableMapField(
-        int number) {
-      switch (number) {
-        case 2:
-          return internalGetMutableAttr();
-        default:
-          throw new RuntimeException(
-              "Invalid map field number: " + number);
-      }
-    }
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.AttrValueProtos.internal_static_tensorflow_NameAttrList_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.NameAttrList.class, org.tensorflow.proto.framework.NameAttrList.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.framework.NameAttrList.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      name_ = "";
-
-      internalGetMutableAttr().clear();
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.framework.AttrValueProtos.internal_static_tensorflow_NameAttrList_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.NameAttrList getDefaultInstanceForType() {
-      return org.tensorflow.proto.framework.NameAttrList.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.NameAttrList build() {
-      org.tensorflow.proto.framework.NameAttrList result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.NameAttrList buildPartial() {
-      org.tensorflow.proto.framework.NameAttrList result = new org.tensorflow.proto.framework.NameAttrList(this);
-      int from_bitField0_ = bitField0_;
-      result.name_ = name_;
-      result.attr_ = internalGetAttr();
-      result.attr_.makeImmutable();
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.framework.NameAttrList) {
-        return mergeFrom((org.tensorflow.proto.framework.NameAttrList)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.framework.NameAttrList other) {
-      if (other == org.tensorflow.proto.framework.NameAttrList.getDefaultInstance()) return this;
-      if (!other.getName().isEmpty()) {
-        name_ = other.name_;
-        onChanged();
-      }
-      internalGetMutableAttr().mergeFrom(
-          other.internalGetAttr());
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.framework.NameAttrList parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.framework.NameAttrList) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-    private int bitField0_;
-
-    private java.lang.Object name_ = "";
-    /**
-     * <code>string name = 1;</code>
-     */
-    public java.lang.String getName() {
-      java.lang.Object ref = name_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        name_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <code>string name = 1;</code>
-     */
-    public com.google.protobuf.ByteString
-        getNameBytes() {
-      java.lang.Object ref = name_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        name_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <code>string name = 1;</code>
-     */
-    public Builder setName(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      name_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>string name = 1;</code>
-     */
-    public Builder clearName() {
-      
-      name_ = getDefaultInstance().getName();
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>string name = 1;</code>
-     */
-    public Builder setNameBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      name_ = value;
-      onChanged();
-      return this;
-    }
-
-    private com.google.protobuf.MapField<
-        java.lang.String, org.tensorflow.proto.framework.AttrValue> attr_;
-    private com.google.protobuf.MapField<java.lang.String, org.tensorflow.proto.framework.AttrValue>
-    internalGetAttr() {
-      if (attr_ == null) {
-        return com.google.protobuf.MapField.emptyMapField(
-            AttrDefaultEntryHolder.defaultEntry);
-      }
-      return attr_;
-    }
-    private com.google.protobuf.MapField<java.lang.String, org.tensorflow.proto.framework.AttrValue>
-    internalGetMutableAttr() {
-      onChanged();;
-      if (attr_ == null) {
-        attr_ = com.google.protobuf.MapField.newMapField(
-            AttrDefaultEntryHolder.defaultEntry);
-      }
-      if (!attr_.isMutable()) {
-        attr_ = attr_.copy();
-      }
-      return attr_;
-    }
-
-    public int getAttrCount() {
-      return internalGetAttr().getMap().size();
-    }
-    /**
-     * <code>map&lt;string, .tensorflow.AttrValue&gt; attr = 2;</code>
-     */
-
-    public boolean containsAttr(
-        java.lang.String key) {
-      if (key == null) { throw new java.lang.NullPointerException(); }
-      return internalGetAttr().getMap().containsKey(key);
-    }
-    /**
-     * Use {@link #getAttrMap()} instead.
-     */
-    @java.lang.Deprecated
-    public java.util.Map<java.lang.String, org.tensorflow.proto.framework.AttrValue> getAttr() {
-      return getAttrMap();
-    }
-    /**
-     * <code>map&lt;string, .tensorflow.AttrValue&gt; attr = 2;</code>
-     */
-
-    public java.util.Map<java.lang.String, org.tensorflow.proto.framework.AttrValue> getAttrMap() {
-      return internalGetAttr().getMap();
-    }
-    /**
-     * <code>map&lt;string, .tensorflow.AttrValue&gt; attr = 2;</code>
-     */
-
-    public org.tensorflow.proto.framework.AttrValue getAttrOrDefault(
-        java.lang.String key,
-        org.tensorflow.proto.framework.AttrValue defaultValue) {
-      if (key == null) { throw new java.lang.NullPointerException(); }
-      java.util.Map<java.lang.String, org.tensorflow.proto.framework.AttrValue> map =
-          internalGetAttr().getMap();
-      return map.containsKey(key) ? map.get(key) : defaultValue;
-    }
-    /**
-     * <code>map&lt;string, .tensorflow.AttrValue&gt; attr = 2;</code>
-     */
-
-    public org.tensorflow.proto.framework.AttrValue getAttrOrThrow(
-        java.lang.String key) {
-      if (key == null) { throw new java.lang.NullPointerException(); }
-      java.util.Map<java.lang.String, org.tensorflow.proto.framework.AttrValue> map =
-          internalGetAttr().getMap();
-      if (!map.containsKey(key)) {
-        throw new java.lang.IllegalArgumentException();
-      }
-      return map.get(key);
-    }
-
-    public Builder clearAttr() {
-      internalGetMutableAttr().getMutableMap()
-          .clear();
-      return this;
-    }
-    /**
-     * <code>map&lt;string, .tensorflow.AttrValue&gt; attr = 2;</code>
-     */
-
-    public Builder removeAttr(
-        java.lang.String key) {
-      if (key == null) { throw new java.lang.NullPointerException(); }
-      internalGetMutableAttr().getMutableMap()
-          .remove(key);
-      return this;
-    }
-    /**
-     * Use alternate mutation accessors instead.
-     */
-    @java.lang.Deprecated
-    public java.util.Map<java.lang.String, org.tensorflow.proto.framework.AttrValue>
-    getMutableAttr() {
-      return internalGetMutableAttr().getMutableMap();
-    }
-    /**
-     * <code>map&lt;string, .tensorflow.AttrValue&gt; attr = 2;</code>
-     */
-    public Builder putAttr(
-        java.lang.String key,
-        org.tensorflow.proto.framework.AttrValue value) {
-      if (key == null) { throw new java.lang.NullPointerException(); }
-      if (value == null) { throw new java.lang.NullPointerException(); }
-      internalGetMutableAttr().getMutableMap()
-          .put(key, value);
-      return this;
-    }
-    /**
-     * <code>map&lt;string, .tensorflow.AttrValue&gt; attr = 2;</code>
-     */
-
-    public Builder putAllAttr(
-        java.util.Map<java.lang.String, org.tensorflow.proto.framework.AttrValue> values) {
-      internalGetMutableAttr().getMutableMap()
-          .putAll(values);
-      return this;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.NameAttrList)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.NameAttrList)
-  private static final org.tensorflow.proto.framework.NameAttrList DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.NameAttrList();
-  }
-
-  public static org.tensorflow.proto.framework.NameAttrList getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<NameAttrList>
-      PARSER = new com.google.protobuf.AbstractParser<NameAttrList>() {
-    @java.lang.Override
-    public NameAttrList parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new NameAttrList(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<NameAttrList> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<NameAttrList> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.framework.NameAttrList getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NameAttrListOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NameAttrListOrBuilder.java
deleted file mode 100644
index 293b4278408..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NameAttrListOrBuilder.java
+++ /dev/null
@@ -1,53 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/framework/attr_value.proto
-
-package org.tensorflow.proto.framework;
-
-public interface NameAttrListOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.NameAttrList)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <code>string name = 1;</code>
-   */
-  java.lang.String getName();
-  /**
-   * <code>string name = 1;</code>
-   */
-  com.google.protobuf.ByteString
-      getNameBytes();
-
-  /**
-   * <code>map&lt;string, .tensorflow.AttrValue&gt; attr = 2;</code>
-   */
-  int getAttrCount();
-  /**
-   * <code>map&lt;string, .tensorflow.AttrValue&gt; attr = 2;</code>
-   */
-  boolean containsAttr(
-      java.lang.String key);
-  /**
-   * Use {@link #getAttrMap()} instead.
-   */
-  @java.lang.Deprecated
-  java.util.Map<java.lang.String, org.tensorflow.proto.framework.AttrValue>
-  getAttr();
-  /**
-   * <code>map&lt;string, .tensorflow.AttrValue&gt; attr = 2;</code>
-   */
-  java.util.Map<java.lang.String, org.tensorflow.proto.framework.AttrValue>
-  getAttrMap();
-  /**
-   * <code>map&lt;string, .tensorflow.AttrValue&gt; attr = 2;</code>
-   */
-
-  org.tensorflow.proto.framework.AttrValue getAttrOrDefault(
-      java.lang.String key,
-      org.tensorflow.proto.framework.AttrValue defaultValue);
-  /**
-   * <code>map&lt;string, .tensorflow.AttrValue&gt; attr = 2;</code>
-   */
-
-  org.tensorflow.proto.framework.AttrValue getAttrOrThrow(
-      java.lang.String key);
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NamedDevice.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NamedDevice.java
deleted file mode 100644
index 02a1dc9f6c7..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NamedDevice.java
+++ /dev/null
@@ -1,727 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/device_properties.proto
-
-package org.tensorflow.proto.framework;
-
-/**
- * Protobuf type {@code tensorflow.NamedDevice}
- */
-public  final class NamedDevice extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.NamedDevice)
-    NamedDeviceOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use NamedDevice.newBuilder() to construct.
-  private NamedDevice(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private NamedDevice() {
-    name_ = "";
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new NamedDevice();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private NamedDevice(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 10: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            name_ = s;
-            break;
-          }
-          case 18: {
-            org.tensorflow.proto.framework.DeviceProperties.Builder subBuilder = null;
-            if (properties_ != null) {
-              subBuilder = properties_.toBuilder();
-            }
-            properties_ = input.readMessage(org.tensorflow.proto.framework.DeviceProperties.parser(), extensionRegistry);
-            if (subBuilder != null) {
-              subBuilder.mergeFrom(properties_);
-              properties_ = subBuilder.buildPartial();
-            }
-
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.framework.DevicePropertiesProtos.internal_static_tensorflow_NamedDevice_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.framework.DevicePropertiesProtos.internal_static_tensorflow_NamedDevice_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.framework.NamedDevice.class, org.tensorflow.proto.framework.NamedDevice.Builder.class);
-  }
-
-  public static final int NAME_FIELD_NUMBER = 1;
-  private volatile java.lang.Object name_;
-  /**
-   * <code>string name = 1;</code>
-   */
-  public java.lang.String getName() {
-    java.lang.Object ref = name_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      name_ = s;
-      return s;
-    }
-  }
-  /**
-   * <code>string name = 1;</code>
-   */
-  public com.google.protobuf.ByteString
-      getNameBytes() {
-    java.lang.Object ref = name_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      name_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int PROPERTIES_FIELD_NUMBER = 2;
-  private org.tensorflow.proto.framework.DeviceProperties properties_;
-  /**
-   * <code>.tensorflow.DeviceProperties properties = 2;</code>
-   */
-  public boolean hasProperties() {
-    return properties_ != null;
-  }
-  /**
-   * <code>.tensorflow.DeviceProperties properties = 2;</code>
-   */
-  public org.tensorflow.proto.framework.DeviceProperties getProperties() {
-    return properties_ == null ? org.tensorflow.proto.framework.DeviceProperties.getDefaultInstance() : properties_;
-  }
-  /**
-   * <code>.tensorflow.DeviceProperties properties = 2;</code>
-   */
-  public org.tensorflow.proto.framework.DevicePropertiesOrBuilder getPropertiesOrBuilder() {
-    return getProperties();
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    if (!getNameBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 1, name_);
-    }
-    if (properties_ != null) {
-      output.writeMessage(2, getProperties());
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (!getNameBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, name_);
-    }
-    if (properties_ != null) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(2, getProperties());
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.framework.NamedDevice)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.framework.NamedDevice other = (org.tensorflow.proto.framework.NamedDevice) obj;
-
-    if (!getName()
-        .equals(other.getName())) return false;
-    if (hasProperties() != other.hasProperties()) return false;
-    if (hasProperties()) {
-      if (!getProperties()
-          .equals(other.getProperties())) return false;
-    }
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    hash = (37 * hash) + NAME_FIELD_NUMBER;
-    hash = (53 * hash) + getName().hashCode();
-    if (hasProperties()) {
-      hash = (37 * hash) + PROPERTIES_FIELD_NUMBER;
-      hash = (53 * hash) + getProperties().hashCode();
-    }
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.framework.NamedDevice parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.NamedDevice parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.NamedDevice parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.NamedDevice parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.NamedDevice parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.NamedDevice parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.NamedDevice parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.NamedDevice parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.NamedDevice parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.NamedDevice parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.NamedDevice parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.NamedDevice parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.framework.NamedDevice prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * Protobuf type {@code tensorflow.NamedDevice}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.NamedDevice)
-      org.tensorflow.proto.framework.NamedDeviceOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.DevicePropertiesProtos.internal_static_tensorflow_NamedDevice_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.DevicePropertiesProtos.internal_static_tensorflow_NamedDevice_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.NamedDevice.class, org.tensorflow.proto.framework.NamedDevice.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.framework.NamedDevice.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      name_ = "";
-
-      if (propertiesBuilder_ == null) {
-        properties_ = null;
-      } else {
-        properties_ = null;
-        propertiesBuilder_ = null;
-      }
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.framework.DevicePropertiesProtos.internal_static_tensorflow_NamedDevice_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.NamedDevice getDefaultInstanceForType() {
-      return org.tensorflow.proto.framework.NamedDevice.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.NamedDevice build() {
-      org.tensorflow.proto.framework.NamedDevice result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.NamedDevice buildPartial() {
-      org.tensorflow.proto.framework.NamedDevice result = new org.tensorflow.proto.framework.NamedDevice(this);
-      result.name_ = name_;
-      if (propertiesBuilder_ == null) {
-        result.properties_ = properties_;
-      } else {
-        result.properties_ = propertiesBuilder_.build();
-      }
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.framework.NamedDevice) {
-        return mergeFrom((org.tensorflow.proto.framework.NamedDevice)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.framework.NamedDevice other) {
-      if (other == org.tensorflow.proto.framework.NamedDevice.getDefaultInstance()) return this;
-      if (!other.getName().isEmpty()) {
-        name_ = other.name_;
-        onChanged();
-      }
-      if (other.hasProperties()) {
-        mergeProperties(other.getProperties());
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.framework.NamedDevice parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.framework.NamedDevice) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-
-    private java.lang.Object name_ = "";
-    /**
-     * <code>string name = 1;</code>
-     */
-    public java.lang.String getName() {
-      java.lang.Object ref = name_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        name_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <code>string name = 1;</code>
-     */
-    public com.google.protobuf.ByteString
-        getNameBytes() {
-      java.lang.Object ref = name_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        name_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <code>string name = 1;</code>
-     */
-    public Builder setName(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      name_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>string name = 1;</code>
-     */
-    public Builder clearName() {
-      
-      name_ = getDefaultInstance().getName();
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>string name = 1;</code>
-     */
-    public Builder setNameBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      name_ = value;
-      onChanged();
-      return this;
-    }
-
-    private org.tensorflow.proto.framework.DeviceProperties properties_;
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.DeviceProperties, org.tensorflow.proto.framework.DeviceProperties.Builder, org.tensorflow.proto.framework.DevicePropertiesOrBuilder> propertiesBuilder_;
-    /**
-     * <code>.tensorflow.DeviceProperties properties = 2;</code>
-     */
-    public boolean hasProperties() {
-      return propertiesBuilder_ != null || properties_ != null;
-    }
-    /**
-     * <code>.tensorflow.DeviceProperties properties = 2;</code>
-     */
-    public org.tensorflow.proto.framework.DeviceProperties getProperties() {
-      if (propertiesBuilder_ == null) {
-        return properties_ == null ? org.tensorflow.proto.framework.DeviceProperties.getDefaultInstance() : properties_;
-      } else {
-        return propertiesBuilder_.getMessage();
-      }
-    }
-    /**
-     * <code>.tensorflow.DeviceProperties properties = 2;</code>
-     */
-    public Builder setProperties(org.tensorflow.proto.framework.DeviceProperties value) {
-      if (propertiesBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        properties_ = value;
-        onChanged();
-      } else {
-        propertiesBuilder_.setMessage(value);
-      }
-
-      return this;
-    }
-    /**
-     * <code>.tensorflow.DeviceProperties properties = 2;</code>
-     */
-    public Builder setProperties(
-        org.tensorflow.proto.framework.DeviceProperties.Builder builderForValue) {
-      if (propertiesBuilder_ == null) {
-        properties_ = builderForValue.build();
-        onChanged();
-      } else {
-        propertiesBuilder_.setMessage(builderForValue.build());
-      }
-
-      return this;
-    }
-    /**
-     * <code>.tensorflow.DeviceProperties properties = 2;</code>
-     */
-    public Builder mergeProperties(org.tensorflow.proto.framework.DeviceProperties value) {
-      if (propertiesBuilder_ == null) {
-        if (properties_ != null) {
-          properties_ =
-            org.tensorflow.proto.framework.DeviceProperties.newBuilder(properties_).mergeFrom(value).buildPartial();
-        } else {
-          properties_ = value;
-        }
-        onChanged();
-      } else {
-        propertiesBuilder_.mergeFrom(value);
-      }
-
-      return this;
-    }
-    /**
-     * <code>.tensorflow.DeviceProperties properties = 2;</code>
-     */
-    public Builder clearProperties() {
-      if (propertiesBuilder_ == null) {
-        properties_ = null;
-        onChanged();
-      } else {
-        properties_ = null;
-        propertiesBuilder_ = null;
-      }
-
-      return this;
-    }
-    /**
-     * <code>.tensorflow.DeviceProperties properties = 2;</code>
-     */
-    public org.tensorflow.proto.framework.DeviceProperties.Builder getPropertiesBuilder() {
-      
-      onChanged();
-      return getPropertiesFieldBuilder().getBuilder();
-    }
-    /**
-     * <code>.tensorflow.DeviceProperties properties = 2;</code>
-     */
-    public org.tensorflow.proto.framework.DevicePropertiesOrBuilder getPropertiesOrBuilder() {
-      if (propertiesBuilder_ != null) {
-        return propertiesBuilder_.getMessageOrBuilder();
-      } else {
-        return properties_ == null ?
-            org.tensorflow.proto.framework.DeviceProperties.getDefaultInstance() : properties_;
-      }
-    }
-    /**
-     * <code>.tensorflow.DeviceProperties properties = 2;</code>
-     */
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.DeviceProperties, org.tensorflow.proto.framework.DeviceProperties.Builder, org.tensorflow.proto.framework.DevicePropertiesOrBuilder> 
-        getPropertiesFieldBuilder() {
-      if (propertiesBuilder_ == null) {
-        propertiesBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-            org.tensorflow.proto.framework.DeviceProperties, org.tensorflow.proto.framework.DeviceProperties.Builder, org.tensorflow.proto.framework.DevicePropertiesOrBuilder>(
-                getProperties(),
-                getParentForChildren(),
-                isClean());
-        properties_ = null;
-      }
-      return propertiesBuilder_;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.NamedDevice)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.NamedDevice)
-  private static final org.tensorflow.proto.framework.NamedDevice DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.NamedDevice();
-  }
-
-  public static org.tensorflow.proto.framework.NamedDevice getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<NamedDevice>
-      PARSER = new com.google.protobuf.AbstractParser<NamedDevice>() {
-    @java.lang.Override
-    public NamedDevice parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new NamedDevice(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<NamedDevice> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<NamedDevice> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.framework.NamedDevice getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NamedDeviceOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NamedDeviceOrBuilder.java
deleted file mode 100644
index 52eed19dd38..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NamedDeviceOrBuilder.java
+++ /dev/null
@@ -1,32 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/device_properties.proto
-
-package org.tensorflow.proto.framework;
-
-public interface NamedDeviceOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.NamedDevice)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <code>string name = 1;</code>
-   */
-  java.lang.String getName();
-  /**
-   * <code>string name = 1;</code>
-   */
-  com.google.protobuf.ByteString
-      getNameBytes();
-
-  /**
-   * <code>.tensorflow.DeviceProperties properties = 2;</code>
-   */
-  boolean hasProperties();
-  /**
-   * <code>.tensorflow.DeviceProperties properties = 2;</code>
-   */
-  org.tensorflow.proto.framework.DeviceProperties getProperties();
-  /**
-   * <code>.tensorflow.DeviceProperties properties = 2;</code>
-   */
-  org.tensorflow.proto.framework.DevicePropertiesOrBuilder getPropertiesOrBuilder();
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NamedTensorProto.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NamedTensorProto.java
deleted file mode 100644
index c5b5514e2e3..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NamedTensorProto.java
+++ /dev/null
@@ -1,859 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/named_tensor.proto
-
-package org.tensorflow.proto.framework;
-
-/**
- * <pre>
- * A pair of tensor name and tensor values.
- * </pre>
- *
- * Protobuf type {@code tensorflow.NamedTensorProto}
- */
-public  final class NamedTensorProto extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.NamedTensorProto)
-    NamedTensorProtoOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use NamedTensorProto.newBuilder() to construct.
-  private NamedTensorProto(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private NamedTensorProto() {
-    name_ = "";
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new NamedTensorProto();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private NamedTensorProto(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 10: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            name_ = s;
-            break;
-          }
-          case 18: {
-            org.tensorflow.proto.framework.TensorProto.Builder subBuilder = null;
-            if (tensor_ != null) {
-              subBuilder = tensor_.toBuilder();
-            }
-            tensor_ = input.readMessage(org.tensorflow.proto.framework.TensorProto.parser(), extensionRegistry);
-            if (subBuilder != null) {
-              subBuilder.mergeFrom(tensor_);
-              tensor_ = subBuilder.buildPartial();
-            }
-
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.framework.NamedTensorProtos.internal_static_tensorflow_NamedTensorProto_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.framework.NamedTensorProtos.internal_static_tensorflow_NamedTensorProto_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.framework.NamedTensorProto.class, org.tensorflow.proto.framework.NamedTensorProto.Builder.class);
-  }
-
-  public static final int NAME_FIELD_NUMBER = 1;
-  private volatile java.lang.Object name_;
-  /**
-   * <pre>
-   * Name of the tensor.
-   * </pre>
-   *
-   * <code>string name = 1;</code>
-   */
-  public java.lang.String getName() {
-    java.lang.Object ref = name_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      name_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * Name of the tensor.
-   * </pre>
-   *
-   * <code>string name = 1;</code>
-   */
-  public com.google.protobuf.ByteString
-      getNameBytes() {
-    java.lang.Object ref = name_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      name_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int TENSOR_FIELD_NUMBER = 2;
-  private org.tensorflow.proto.framework.TensorProto tensor_;
-  /**
-   * <pre>
-   * The client can populate a TensorProto using a tensorflow::Tensor`, or
-   * directly using the protobuf field accessors.
-   * The client specifies whether the returned tensor values should be
-   * filled tensor fields (float_val, int_val, etc.) or encoded in a
-   * compact form in tensor.tensor_content.
-   * </pre>
-   *
-   * <code>.tensorflow.TensorProto tensor = 2;</code>
-   */
-  public boolean hasTensor() {
-    return tensor_ != null;
-  }
-  /**
-   * <pre>
-   * The client can populate a TensorProto using a tensorflow::Tensor`, or
-   * directly using the protobuf field accessors.
-   * The client specifies whether the returned tensor values should be
-   * filled tensor fields (float_val, int_val, etc.) or encoded in a
-   * compact form in tensor.tensor_content.
-   * </pre>
-   *
-   * <code>.tensorflow.TensorProto tensor = 2;</code>
-   */
-  public org.tensorflow.proto.framework.TensorProto getTensor() {
-    return tensor_ == null ? org.tensorflow.proto.framework.TensorProto.getDefaultInstance() : tensor_;
-  }
-  /**
-   * <pre>
-   * The client can populate a TensorProto using a tensorflow::Tensor`, or
-   * directly using the protobuf field accessors.
-   * The client specifies whether the returned tensor values should be
-   * filled tensor fields (float_val, int_val, etc.) or encoded in a
-   * compact form in tensor.tensor_content.
-   * </pre>
-   *
-   * <code>.tensorflow.TensorProto tensor = 2;</code>
-   */
-  public org.tensorflow.proto.framework.TensorProtoOrBuilder getTensorOrBuilder() {
-    return getTensor();
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    if (!getNameBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 1, name_);
-    }
-    if (tensor_ != null) {
-      output.writeMessage(2, getTensor());
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (!getNameBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, name_);
-    }
-    if (tensor_ != null) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(2, getTensor());
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.framework.NamedTensorProto)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.framework.NamedTensorProto other = (org.tensorflow.proto.framework.NamedTensorProto) obj;
-
-    if (!getName()
-        .equals(other.getName())) return false;
-    if (hasTensor() != other.hasTensor()) return false;
-    if (hasTensor()) {
-      if (!getTensor()
-          .equals(other.getTensor())) return false;
-    }
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    hash = (37 * hash) + NAME_FIELD_NUMBER;
-    hash = (53 * hash) + getName().hashCode();
-    if (hasTensor()) {
-      hash = (37 * hash) + TENSOR_FIELD_NUMBER;
-      hash = (53 * hash) + getTensor().hashCode();
-    }
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.framework.NamedTensorProto parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.NamedTensorProto parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.NamedTensorProto parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.NamedTensorProto parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.NamedTensorProto parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.NamedTensorProto parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.NamedTensorProto parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.NamedTensorProto parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.NamedTensorProto parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.NamedTensorProto parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.NamedTensorProto parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.NamedTensorProto parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.framework.NamedTensorProto prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * <pre>
-   * A pair of tensor name and tensor values.
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.NamedTensorProto}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.NamedTensorProto)
-      org.tensorflow.proto.framework.NamedTensorProtoOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.NamedTensorProtos.internal_static_tensorflow_NamedTensorProto_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.NamedTensorProtos.internal_static_tensorflow_NamedTensorProto_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.NamedTensorProto.class, org.tensorflow.proto.framework.NamedTensorProto.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.framework.NamedTensorProto.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      name_ = "";
-
-      if (tensorBuilder_ == null) {
-        tensor_ = null;
-      } else {
-        tensor_ = null;
-        tensorBuilder_ = null;
-      }
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.framework.NamedTensorProtos.internal_static_tensorflow_NamedTensorProto_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.NamedTensorProto getDefaultInstanceForType() {
-      return org.tensorflow.proto.framework.NamedTensorProto.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.NamedTensorProto build() {
-      org.tensorflow.proto.framework.NamedTensorProto result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.NamedTensorProto buildPartial() {
-      org.tensorflow.proto.framework.NamedTensorProto result = new org.tensorflow.proto.framework.NamedTensorProto(this);
-      result.name_ = name_;
-      if (tensorBuilder_ == null) {
-        result.tensor_ = tensor_;
-      } else {
-        result.tensor_ = tensorBuilder_.build();
-      }
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.framework.NamedTensorProto) {
-        return mergeFrom((org.tensorflow.proto.framework.NamedTensorProto)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.framework.NamedTensorProto other) {
-      if (other == org.tensorflow.proto.framework.NamedTensorProto.getDefaultInstance()) return this;
-      if (!other.getName().isEmpty()) {
-        name_ = other.name_;
-        onChanged();
-      }
-      if (other.hasTensor()) {
-        mergeTensor(other.getTensor());
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.framework.NamedTensorProto parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.framework.NamedTensorProto) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-
-    private java.lang.Object name_ = "";
-    /**
-     * <pre>
-     * Name of the tensor.
-     * </pre>
-     *
-     * <code>string name = 1;</code>
-     */
-    public java.lang.String getName() {
-      java.lang.Object ref = name_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        name_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Name of the tensor.
-     * </pre>
-     *
-     * <code>string name = 1;</code>
-     */
-    public com.google.protobuf.ByteString
-        getNameBytes() {
-      java.lang.Object ref = name_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        name_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Name of the tensor.
-     * </pre>
-     *
-     * <code>string name = 1;</code>
-     */
-    public Builder setName(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      name_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Name of the tensor.
-     * </pre>
-     *
-     * <code>string name = 1;</code>
-     */
-    public Builder clearName() {
-      
-      name_ = getDefaultInstance().getName();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Name of the tensor.
-     * </pre>
-     *
-     * <code>string name = 1;</code>
-     */
-    public Builder setNameBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      name_ = value;
-      onChanged();
-      return this;
-    }
-
-    private org.tensorflow.proto.framework.TensorProto tensor_;
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.TensorProto, org.tensorflow.proto.framework.TensorProto.Builder, org.tensorflow.proto.framework.TensorProtoOrBuilder> tensorBuilder_;
-    /**
-     * <pre>
-     * The client can populate a TensorProto using a tensorflow::Tensor`, or
-     * directly using the protobuf field accessors.
-     * The client specifies whether the returned tensor values should be
-     * filled tensor fields (float_val, int_val, etc.) or encoded in a
-     * compact form in tensor.tensor_content.
-     * </pre>
-     *
-     * <code>.tensorflow.TensorProto tensor = 2;</code>
-     */
-    public boolean hasTensor() {
-      return tensorBuilder_ != null || tensor_ != null;
-    }
-    /**
-     * <pre>
-     * The client can populate a TensorProto using a tensorflow::Tensor`, or
-     * directly using the protobuf field accessors.
-     * The client specifies whether the returned tensor values should be
-     * filled tensor fields (float_val, int_val, etc.) or encoded in a
-     * compact form in tensor.tensor_content.
-     * </pre>
-     *
-     * <code>.tensorflow.TensorProto tensor = 2;</code>
-     */
-    public org.tensorflow.proto.framework.TensorProto getTensor() {
-      if (tensorBuilder_ == null) {
-        return tensor_ == null ? org.tensorflow.proto.framework.TensorProto.getDefaultInstance() : tensor_;
-      } else {
-        return tensorBuilder_.getMessage();
-      }
-    }
-    /**
-     * <pre>
-     * The client can populate a TensorProto using a tensorflow::Tensor`, or
-     * directly using the protobuf field accessors.
-     * The client specifies whether the returned tensor values should be
-     * filled tensor fields (float_val, int_val, etc.) or encoded in a
-     * compact form in tensor.tensor_content.
-     * </pre>
-     *
-     * <code>.tensorflow.TensorProto tensor = 2;</code>
-     */
-    public Builder setTensor(org.tensorflow.proto.framework.TensorProto value) {
-      if (tensorBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        tensor_ = value;
-        onChanged();
-      } else {
-        tensorBuilder_.setMessage(value);
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * The client can populate a TensorProto using a tensorflow::Tensor`, or
-     * directly using the protobuf field accessors.
-     * The client specifies whether the returned tensor values should be
-     * filled tensor fields (float_val, int_val, etc.) or encoded in a
-     * compact form in tensor.tensor_content.
-     * </pre>
-     *
-     * <code>.tensorflow.TensorProto tensor = 2;</code>
-     */
-    public Builder setTensor(
-        org.tensorflow.proto.framework.TensorProto.Builder builderForValue) {
-      if (tensorBuilder_ == null) {
-        tensor_ = builderForValue.build();
-        onChanged();
-      } else {
-        tensorBuilder_.setMessage(builderForValue.build());
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * The client can populate a TensorProto using a tensorflow::Tensor`, or
-     * directly using the protobuf field accessors.
-     * The client specifies whether the returned tensor values should be
-     * filled tensor fields (float_val, int_val, etc.) or encoded in a
-     * compact form in tensor.tensor_content.
-     * </pre>
-     *
-     * <code>.tensorflow.TensorProto tensor = 2;</code>
-     */
-    public Builder mergeTensor(org.tensorflow.proto.framework.TensorProto value) {
-      if (tensorBuilder_ == null) {
-        if (tensor_ != null) {
-          tensor_ =
-            org.tensorflow.proto.framework.TensorProto.newBuilder(tensor_).mergeFrom(value).buildPartial();
-        } else {
-          tensor_ = value;
-        }
-        onChanged();
-      } else {
-        tensorBuilder_.mergeFrom(value);
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * The client can populate a TensorProto using a tensorflow::Tensor`, or
-     * directly using the protobuf field accessors.
-     * The client specifies whether the returned tensor values should be
-     * filled tensor fields (float_val, int_val, etc.) or encoded in a
-     * compact form in tensor.tensor_content.
-     * </pre>
-     *
-     * <code>.tensorflow.TensorProto tensor = 2;</code>
-     */
-    public Builder clearTensor() {
-      if (tensorBuilder_ == null) {
-        tensor_ = null;
-        onChanged();
-      } else {
-        tensor_ = null;
-        tensorBuilder_ = null;
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * The client can populate a TensorProto using a tensorflow::Tensor`, or
-     * directly using the protobuf field accessors.
-     * The client specifies whether the returned tensor values should be
-     * filled tensor fields (float_val, int_val, etc.) or encoded in a
-     * compact form in tensor.tensor_content.
-     * </pre>
-     *
-     * <code>.tensorflow.TensorProto tensor = 2;</code>
-     */
-    public org.tensorflow.proto.framework.TensorProto.Builder getTensorBuilder() {
-      
-      onChanged();
-      return getTensorFieldBuilder().getBuilder();
-    }
-    /**
-     * <pre>
-     * The client can populate a TensorProto using a tensorflow::Tensor`, or
-     * directly using the protobuf field accessors.
-     * The client specifies whether the returned tensor values should be
-     * filled tensor fields (float_val, int_val, etc.) or encoded in a
-     * compact form in tensor.tensor_content.
-     * </pre>
-     *
-     * <code>.tensorflow.TensorProto tensor = 2;</code>
-     */
-    public org.tensorflow.proto.framework.TensorProtoOrBuilder getTensorOrBuilder() {
-      if (tensorBuilder_ != null) {
-        return tensorBuilder_.getMessageOrBuilder();
-      } else {
-        return tensor_ == null ?
-            org.tensorflow.proto.framework.TensorProto.getDefaultInstance() : tensor_;
-      }
-    }
-    /**
-     * <pre>
-     * The client can populate a TensorProto using a tensorflow::Tensor`, or
-     * directly using the protobuf field accessors.
-     * The client specifies whether the returned tensor values should be
-     * filled tensor fields (float_val, int_val, etc.) or encoded in a
-     * compact form in tensor.tensor_content.
-     * </pre>
-     *
-     * <code>.tensorflow.TensorProto tensor = 2;</code>
-     */
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.TensorProto, org.tensorflow.proto.framework.TensorProto.Builder, org.tensorflow.proto.framework.TensorProtoOrBuilder> 
-        getTensorFieldBuilder() {
-      if (tensorBuilder_ == null) {
-        tensorBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-            org.tensorflow.proto.framework.TensorProto, org.tensorflow.proto.framework.TensorProto.Builder, org.tensorflow.proto.framework.TensorProtoOrBuilder>(
-                getTensor(),
-                getParentForChildren(),
-                isClean());
-        tensor_ = null;
-      }
-      return tensorBuilder_;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.NamedTensorProto)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.NamedTensorProto)
-  private static final org.tensorflow.proto.framework.NamedTensorProto DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.NamedTensorProto();
-  }
-
-  public static org.tensorflow.proto.framework.NamedTensorProto getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<NamedTensorProto>
-      PARSER = new com.google.protobuf.AbstractParser<NamedTensorProto>() {
-    @java.lang.Override
-    public NamedTensorProto parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new NamedTensorProto(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<NamedTensorProto> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<NamedTensorProto> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.framework.NamedTensorProto getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NamedTensorProtoOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NamedTensorProtoOrBuilder.java
deleted file mode 100644
index 41b758b2f91..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NamedTensorProtoOrBuilder.java
+++ /dev/null
@@ -1,64 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/named_tensor.proto
-
-package org.tensorflow.proto.framework;
-
-public interface NamedTensorProtoOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.NamedTensorProto)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <pre>
-   * Name of the tensor.
-   * </pre>
-   *
-   * <code>string name = 1;</code>
-   */
-  java.lang.String getName();
-  /**
-   * <pre>
-   * Name of the tensor.
-   * </pre>
-   *
-   * <code>string name = 1;</code>
-   */
-  com.google.protobuf.ByteString
-      getNameBytes();
-
-  /**
-   * <pre>
-   * The client can populate a TensorProto using a tensorflow::Tensor`, or
-   * directly using the protobuf field accessors.
-   * The client specifies whether the returned tensor values should be
-   * filled tensor fields (float_val, int_val, etc.) or encoded in a
-   * compact form in tensor.tensor_content.
-   * </pre>
-   *
-   * <code>.tensorflow.TensorProto tensor = 2;</code>
-   */
-  boolean hasTensor();
-  /**
-   * <pre>
-   * The client can populate a TensorProto using a tensorflow::Tensor`, or
-   * directly using the protobuf field accessors.
-   * The client specifies whether the returned tensor values should be
-   * filled tensor fields (float_val, int_val, etc.) or encoded in a
-   * compact form in tensor.tensor_content.
-   * </pre>
-   *
-   * <code>.tensorflow.TensorProto tensor = 2;</code>
-   */
-  org.tensorflow.proto.framework.TensorProto getTensor();
-  /**
-   * <pre>
-   * The client can populate a TensorProto using a tensorflow::Tensor`, or
-   * directly using the protobuf field accessors.
-   * The client specifies whether the returned tensor values should be
-   * filled tensor fields (float_val, int_val, etc.) or encoded in a
-   * compact form in tensor.tensor_content.
-   * </pre>
-   *
-   * <code>.tensorflow.TensorProto tensor = 2;</code>
-   */
-  org.tensorflow.proto.framework.TensorProtoOrBuilder getTensorOrBuilder();
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NamedTensorProtos.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NamedTensorProtos.java
deleted file mode 100644
index 98685b73c33..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NamedTensorProtos.java
+++ /dev/null
@@ -1,55 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/named_tensor.proto
-
-package org.tensorflow.proto.framework;
-
-public final class NamedTensorProtos {
-  private NamedTensorProtos() {}
-  public static void registerAllExtensions(
-      com.google.protobuf.ExtensionRegistryLite registry) {
-  }
-
-  public static void registerAllExtensions(
-      com.google.protobuf.ExtensionRegistry registry) {
-    registerAllExtensions(
-        (com.google.protobuf.ExtensionRegistryLite) registry);
-  }
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_NamedTensorProto_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_NamedTensorProto_fieldAccessorTable;
-
-  public static com.google.protobuf.Descriptors.FileDescriptor
-      getDescriptor() {
-    return descriptor;
-  }
-  private static  com.google.protobuf.Descriptors.FileDescriptor
-      descriptor;
-  static {
-    java.lang.String[] descriptorData = {
-      "\n+tensorflow/core/protobuf/named_tensor." +
-      "proto\022\ntensorflow\032&tensorflow/core/frame" +
-      "work/tensor.proto\"I\n\020NamedTensorProto\022\014\n" +
-      "\004name\030\001 \001(\t\022\'\n\006tensor\030\002 \001(\0132\027.tensorflow" +
-      ".TensorProtoB\202\001\n\036org.tensorflow.proto.fr" +
-      "ameworkB\021NamedTensorProtosP\001ZHgithub.com" +
-      "/tensorflow/tensorflow/tensorflow/go/cor" +
-      "e/core_protos_go_proto\370\001\001b\006proto3"
-    };
-    descriptor = com.google.protobuf.Descriptors.FileDescriptor
-      .internalBuildGeneratedFileFrom(descriptorData,
-        new com.google.protobuf.Descriptors.FileDescriptor[] {
-          org.tensorflow.proto.framework.TensorProtos.getDescriptor(),
-        });
-    internal_static_tensorflow_NamedTensorProto_descriptor =
-      getDescriptor().getMessageTypes().get(0);
-    internal_static_tensorflow_NamedTensorProto_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_NamedTensorProto_descriptor,
-        new java.lang.String[] { "Name", "Tensor", });
-    org.tensorflow.proto.framework.TensorProtos.getDescriptor();
-  }
-
-  // @@protoc_insertion_point(outer_class_scope)
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NamedTupleValue.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NamedTupleValue.java
deleted file mode 100644
index e1adfe3a508..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NamedTupleValue.java
+++ /dev/null
@@ -1,900 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/struct.proto
-
-package org.tensorflow.proto.framework;
-
-/**
- * <pre>
- * Represents Python's namedtuple.
- * </pre>
- *
- * Protobuf type {@code tensorflow.NamedTupleValue}
- */
-public  final class NamedTupleValue extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.NamedTupleValue)
-    NamedTupleValueOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use NamedTupleValue.newBuilder() to construct.
-  private NamedTupleValue(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private NamedTupleValue() {
-    name_ = "";
-    values_ = java.util.Collections.emptyList();
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new NamedTupleValue();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private NamedTupleValue(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    int mutable_bitField0_ = 0;
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 10: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            name_ = s;
-            break;
-          }
-          case 18: {
-            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
-              values_ = new java.util.ArrayList<org.tensorflow.proto.framework.PairValue>();
-              mutable_bitField0_ |= 0x00000001;
-            }
-            values_.add(
-                input.readMessage(org.tensorflow.proto.framework.PairValue.parser(), extensionRegistry));
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      if (((mutable_bitField0_ & 0x00000001) != 0)) {
-        values_ = java.util.Collections.unmodifiableList(values_);
-      }
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.framework.StructProtos.internal_static_tensorflow_NamedTupleValue_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.framework.StructProtos.internal_static_tensorflow_NamedTupleValue_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.framework.NamedTupleValue.class, org.tensorflow.proto.framework.NamedTupleValue.Builder.class);
-  }
-
-  public static final int NAME_FIELD_NUMBER = 1;
-  private volatile java.lang.Object name_;
-  /**
-   * <code>string name = 1;</code>
-   */
-  public java.lang.String getName() {
-    java.lang.Object ref = name_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      name_ = s;
-      return s;
-    }
-  }
-  /**
-   * <code>string name = 1;</code>
-   */
-  public com.google.protobuf.ByteString
-      getNameBytes() {
-    java.lang.Object ref = name_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      name_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int VALUES_FIELD_NUMBER = 2;
-  private java.util.List<org.tensorflow.proto.framework.PairValue> values_;
-  /**
-   * <code>repeated .tensorflow.PairValue values = 2;</code>
-   */
-  public java.util.List<org.tensorflow.proto.framework.PairValue> getValuesList() {
-    return values_;
-  }
-  /**
-   * <code>repeated .tensorflow.PairValue values = 2;</code>
-   */
-  public java.util.List<? extends org.tensorflow.proto.framework.PairValueOrBuilder> 
-      getValuesOrBuilderList() {
-    return values_;
-  }
-  /**
-   * <code>repeated .tensorflow.PairValue values = 2;</code>
-   */
-  public int getValuesCount() {
-    return values_.size();
-  }
-  /**
-   * <code>repeated .tensorflow.PairValue values = 2;</code>
-   */
-  public org.tensorflow.proto.framework.PairValue getValues(int index) {
-    return values_.get(index);
-  }
-  /**
-   * <code>repeated .tensorflow.PairValue values = 2;</code>
-   */
-  public org.tensorflow.proto.framework.PairValueOrBuilder getValuesOrBuilder(
-      int index) {
-    return values_.get(index);
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    if (!getNameBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 1, name_);
-    }
-    for (int i = 0; i < values_.size(); i++) {
-      output.writeMessage(2, values_.get(i));
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (!getNameBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, name_);
-    }
-    for (int i = 0; i < values_.size(); i++) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(2, values_.get(i));
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.framework.NamedTupleValue)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.framework.NamedTupleValue other = (org.tensorflow.proto.framework.NamedTupleValue) obj;
-
-    if (!getName()
-        .equals(other.getName())) return false;
-    if (!getValuesList()
-        .equals(other.getValuesList())) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    hash = (37 * hash) + NAME_FIELD_NUMBER;
-    hash = (53 * hash) + getName().hashCode();
-    if (getValuesCount() > 0) {
-      hash = (37 * hash) + VALUES_FIELD_NUMBER;
-      hash = (53 * hash) + getValuesList().hashCode();
-    }
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.framework.NamedTupleValue parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.NamedTupleValue parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.NamedTupleValue parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.NamedTupleValue parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.NamedTupleValue parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.NamedTupleValue parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.NamedTupleValue parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.NamedTupleValue parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.NamedTupleValue parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.NamedTupleValue parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.NamedTupleValue parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.NamedTupleValue parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.framework.NamedTupleValue prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * <pre>
-   * Represents Python's namedtuple.
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.NamedTupleValue}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.NamedTupleValue)
-      org.tensorflow.proto.framework.NamedTupleValueOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.StructProtos.internal_static_tensorflow_NamedTupleValue_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.StructProtos.internal_static_tensorflow_NamedTupleValue_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.NamedTupleValue.class, org.tensorflow.proto.framework.NamedTupleValue.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.framework.NamedTupleValue.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-        getValuesFieldBuilder();
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      name_ = "";
-
-      if (valuesBuilder_ == null) {
-        values_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000001);
-      } else {
-        valuesBuilder_.clear();
-      }
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.framework.StructProtos.internal_static_tensorflow_NamedTupleValue_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.NamedTupleValue getDefaultInstanceForType() {
-      return org.tensorflow.proto.framework.NamedTupleValue.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.NamedTupleValue build() {
-      org.tensorflow.proto.framework.NamedTupleValue result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.NamedTupleValue buildPartial() {
-      org.tensorflow.proto.framework.NamedTupleValue result = new org.tensorflow.proto.framework.NamedTupleValue(this);
-      int from_bitField0_ = bitField0_;
-      result.name_ = name_;
-      if (valuesBuilder_ == null) {
-        if (((bitField0_ & 0x00000001) != 0)) {
-          values_ = java.util.Collections.unmodifiableList(values_);
-          bitField0_ = (bitField0_ & ~0x00000001);
-        }
-        result.values_ = values_;
-      } else {
-        result.values_ = valuesBuilder_.build();
-      }
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.framework.NamedTupleValue) {
-        return mergeFrom((org.tensorflow.proto.framework.NamedTupleValue)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.framework.NamedTupleValue other) {
-      if (other == org.tensorflow.proto.framework.NamedTupleValue.getDefaultInstance()) return this;
-      if (!other.getName().isEmpty()) {
-        name_ = other.name_;
-        onChanged();
-      }
-      if (valuesBuilder_ == null) {
-        if (!other.values_.isEmpty()) {
-          if (values_.isEmpty()) {
-            values_ = other.values_;
-            bitField0_ = (bitField0_ & ~0x00000001);
-          } else {
-            ensureValuesIsMutable();
-            values_.addAll(other.values_);
-          }
-          onChanged();
-        }
-      } else {
-        if (!other.values_.isEmpty()) {
-          if (valuesBuilder_.isEmpty()) {
-            valuesBuilder_.dispose();
-            valuesBuilder_ = null;
-            values_ = other.values_;
-            bitField0_ = (bitField0_ & ~0x00000001);
-            valuesBuilder_ = 
-              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
-                 getValuesFieldBuilder() : null;
-          } else {
-            valuesBuilder_.addAllMessages(other.values_);
-          }
-        }
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.framework.NamedTupleValue parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.framework.NamedTupleValue) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-    private int bitField0_;
-
-    private java.lang.Object name_ = "";
-    /**
-     * <code>string name = 1;</code>
-     */
-    public java.lang.String getName() {
-      java.lang.Object ref = name_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        name_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <code>string name = 1;</code>
-     */
-    public com.google.protobuf.ByteString
-        getNameBytes() {
-      java.lang.Object ref = name_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        name_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <code>string name = 1;</code>
-     */
-    public Builder setName(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      name_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>string name = 1;</code>
-     */
-    public Builder clearName() {
-      
-      name_ = getDefaultInstance().getName();
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>string name = 1;</code>
-     */
-    public Builder setNameBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      name_ = value;
-      onChanged();
-      return this;
-    }
-
-    private java.util.List<org.tensorflow.proto.framework.PairValue> values_ =
-      java.util.Collections.emptyList();
-    private void ensureValuesIsMutable() {
-      if (!((bitField0_ & 0x00000001) != 0)) {
-        values_ = new java.util.ArrayList<org.tensorflow.proto.framework.PairValue>(values_);
-        bitField0_ |= 0x00000001;
-       }
-    }
-
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.framework.PairValue, org.tensorflow.proto.framework.PairValue.Builder, org.tensorflow.proto.framework.PairValueOrBuilder> valuesBuilder_;
-
-    /**
-     * <code>repeated .tensorflow.PairValue values = 2;</code>
-     */
-    public java.util.List<org.tensorflow.proto.framework.PairValue> getValuesList() {
-      if (valuesBuilder_ == null) {
-        return java.util.Collections.unmodifiableList(values_);
-      } else {
-        return valuesBuilder_.getMessageList();
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.PairValue values = 2;</code>
-     */
-    public int getValuesCount() {
-      if (valuesBuilder_ == null) {
-        return values_.size();
-      } else {
-        return valuesBuilder_.getCount();
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.PairValue values = 2;</code>
-     */
-    public org.tensorflow.proto.framework.PairValue getValues(int index) {
-      if (valuesBuilder_ == null) {
-        return values_.get(index);
-      } else {
-        return valuesBuilder_.getMessage(index);
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.PairValue values = 2;</code>
-     */
-    public Builder setValues(
-        int index, org.tensorflow.proto.framework.PairValue value) {
-      if (valuesBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureValuesIsMutable();
-        values_.set(index, value);
-        onChanged();
-      } else {
-        valuesBuilder_.setMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.PairValue values = 2;</code>
-     */
-    public Builder setValues(
-        int index, org.tensorflow.proto.framework.PairValue.Builder builderForValue) {
-      if (valuesBuilder_ == null) {
-        ensureValuesIsMutable();
-        values_.set(index, builderForValue.build());
-        onChanged();
-      } else {
-        valuesBuilder_.setMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.PairValue values = 2;</code>
-     */
-    public Builder addValues(org.tensorflow.proto.framework.PairValue value) {
-      if (valuesBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureValuesIsMutable();
-        values_.add(value);
-        onChanged();
-      } else {
-        valuesBuilder_.addMessage(value);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.PairValue values = 2;</code>
-     */
-    public Builder addValues(
-        int index, org.tensorflow.proto.framework.PairValue value) {
-      if (valuesBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureValuesIsMutable();
-        values_.add(index, value);
-        onChanged();
-      } else {
-        valuesBuilder_.addMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.PairValue values = 2;</code>
-     */
-    public Builder addValues(
-        org.tensorflow.proto.framework.PairValue.Builder builderForValue) {
-      if (valuesBuilder_ == null) {
-        ensureValuesIsMutable();
-        values_.add(builderForValue.build());
-        onChanged();
-      } else {
-        valuesBuilder_.addMessage(builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.PairValue values = 2;</code>
-     */
-    public Builder addValues(
-        int index, org.tensorflow.proto.framework.PairValue.Builder builderForValue) {
-      if (valuesBuilder_ == null) {
-        ensureValuesIsMutable();
-        values_.add(index, builderForValue.build());
-        onChanged();
-      } else {
-        valuesBuilder_.addMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.PairValue values = 2;</code>
-     */
-    public Builder addAllValues(
-        java.lang.Iterable<? extends org.tensorflow.proto.framework.PairValue> values) {
-      if (valuesBuilder_ == null) {
-        ensureValuesIsMutable();
-        com.google.protobuf.AbstractMessageLite.Builder.addAll(
-            values, values_);
-        onChanged();
-      } else {
-        valuesBuilder_.addAllMessages(values);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.PairValue values = 2;</code>
-     */
-    public Builder clearValues() {
-      if (valuesBuilder_ == null) {
-        values_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000001);
-        onChanged();
-      } else {
-        valuesBuilder_.clear();
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.PairValue values = 2;</code>
-     */
-    public Builder removeValues(int index) {
-      if (valuesBuilder_ == null) {
-        ensureValuesIsMutable();
-        values_.remove(index);
-        onChanged();
-      } else {
-        valuesBuilder_.remove(index);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.PairValue values = 2;</code>
-     */
-    public org.tensorflow.proto.framework.PairValue.Builder getValuesBuilder(
-        int index) {
-      return getValuesFieldBuilder().getBuilder(index);
-    }
-    /**
-     * <code>repeated .tensorflow.PairValue values = 2;</code>
-     */
-    public org.tensorflow.proto.framework.PairValueOrBuilder getValuesOrBuilder(
-        int index) {
-      if (valuesBuilder_ == null) {
-        return values_.get(index);  } else {
-        return valuesBuilder_.getMessageOrBuilder(index);
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.PairValue values = 2;</code>
-     */
-    public java.util.List<? extends org.tensorflow.proto.framework.PairValueOrBuilder> 
-         getValuesOrBuilderList() {
-      if (valuesBuilder_ != null) {
-        return valuesBuilder_.getMessageOrBuilderList();
-      } else {
-        return java.util.Collections.unmodifiableList(values_);
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.PairValue values = 2;</code>
-     */
-    public org.tensorflow.proto.framework.PairValue.Builder addValuesBuilder() {
-      return getValuesFieldBuilder().addBuilder(
-          org.tensorflow.proto.framework.PairValue.getDefaultInstance());
-    }
-    /**
-     * <code>repeated .tensorflow.PairValue values = 2;</code>
-     */
-    public org.tensorflow.proto.framework.PairValue.Builder addValuesBuilder(
-        int index) {
-      return getValuesFieldBuilder().addBuilder(
-          index, org.tensorflow.proto.framework.PairValue.getDefaultInstance());
-    }
-    /**
-     * <code>repeated .tensorflow.PairValue values = 2;</code>
-     */
-    public java.util.List<org.tensorflow.proto.framework.PairValue.Builder> 
-         getValuesBuilderList() {
-      return getValuesFieldBuilder().getBuilderList();
-    }
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.framework.PairValue, org.tensorflow.proto.framework.PairValue.Builder, org.tensorflow.proto.framework.PairValueOrBuilder> 
-        getValuesFieldBuilder() {
-      if (valuesBuilder_ == null) {
-        valuesBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
-            org.tensorflow.proto.framework.PairValue, org.tensorflow.proto.framework.PairValue.Builder, org.tensorflow.proto.framework.PairValueOrBuilder>(
-                values_,
-                ((bitField0_ & 0x00000001) != 0),
-                getParentForChildren(),
-                isClean());
-        values_ = null;
-      }
-      return valuesBuilder_;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.NamedTupleValue)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.NamedTupleValue)
-  private static final org.tensorflow.proto.framework.NamedTupleValue DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.NamedTupleValue();
-  }
-
-  public static org.tensorflow.proto.framework.NamedTupleValue getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<NamedTupleValue>
-      PARSER = new com.google.protobuf.AbstractParser<NamedTupleValue>() {
-    @java.lang.Override
-    public NamedTupleValue parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new NamedTupleValue(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<NamedTupleValue> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<NamedTupleValue> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.framework.NamedTupleValue getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NamedTupleValueOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NamedTupleValueOrBuilder.java
deleted file mode 100644
index 6829f4aa2ab..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NamedTupleValueOrBuilder.java
+++ /dev/null
@@ -1,43 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/struct.proto
-
-package org.tensorflow.proto.framework;
-
-public interface NamedTupleValueOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.NamedTupleValue)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <code>string name = 1;</code>
-   */
-  java.lang.String getName();
-  /**
-   * <code>string name = 1;</code>
-   */
-  com.google.protobuf.ByteString
-      getNameBytes();
-
-  /**
-   * <code>repeated .tensorflow.PairValue values = 2;</code>
-   */
-  java.util.List<org.tensorflow.proto.framework.PairValue> 
-      getValuesList();
-  /**
-   * <code>repeated .tensorflow.PairValue values = 2;</code>
-   */
-  org.tensorflow.proto.framework.PairValue getValues(int index);
-  /**
-   * <code>repeated .tensorflow.PairValue values = 2;</code>
-   */
-  int getValuesCount();
-  /**
-   * <code>repeated .tensorflow.PairValue values = 2;</code>
-   */
-  java.util.List<? extends org.tensorflow.proto.framework.PairValueOrBuilder> 
-      getValuesOrBuilderList();
-  /**
-   * <code>repeated .tensorflow.PairValue values = 2;</code>
-   */
-  org.tensorflow.proto.framework.PairValueOrBuilder getValuesOrBuilder(
-      int index);
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NodeDef.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NodeDef.java
deleted file mode 100644
index 9bfd5c5c1f3..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NodeDef.java
+++ /dev/null
@@ -1,3076 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/framework/node_def.proto
-
-package org.tensorflow.proto.framework;
-
-/**
- * Protobuf type {@code tensorflow.NodeDef}
- */
-public  final class NodeDef extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.NodeDef)
-    NodeDefOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use NodeDef.newBuilder() to construct.
-  private NodeDef(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private NodeDef() {
-    name_ = "";
-    op_ = "";
-    input_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-    device_ = "";
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new NodeDef();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private NodeDef(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    int mutable_bitField0_ = 0;
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 10: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            name_ = s;
-            break;
-          }
-          case 18: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            op_ = s;
-            break;
-          }
-          case 26: {
-            java.lang.String s = input.readStringRequireUtf8();
-            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
-              input_ = new com.google.protobuf.LazyStringArrayList();
-              mutable_bitField0_ |= 0x00000001;
-            }
-            input_.add(s);
-            break;
-          }
-          case 34: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            device_ = s;
-            break;
-          }
-          case 42: {
-            if (!((mutable_bitField0_ & 0x00000002) != 0)) {
-              attr_ = com.google.protobuf.MapField.newMapField(
-                  AttrDefaultEntryHolder.defaultEntry);
-              mutable_bitField0_ |= 0x00000002;
-            }
-            com.google.protobuf.MapEntry<java.lang.String, org.tensorflow.proto.framework.AttrValue>
-            attr__ = input.readMessage(
-                AttrDefaultEntryHolder.defaultEntry.getParserForType(), extensionRegistry);
-            attr_.getMutableMap().put(
-                attr__.getKey(), attr__.getValue());
-            break;
-          }
-          case 50: {
-            org.tensorflow.proto.framework.NodeDef.ExperimentalDebugInfo.Builder subBuilder = null;
-            if (experimentalDebugInfo_ != null) {
-              subBuilder = experimentalDebugInfo_.toBuilder();
-            }
-            experimentalDebugInfo_ = input.readMessage(org.tensorflow.proto.framework.NodeDef.ExperimentalDebugInfo.parser(), extensionRegistry);
-            if (subBuilder != null) {
-              subBuilder.mergeFrom(experimentalDebugInfo_);
-              experimentalDebugInfo_ = subBuilder.buildPartial();
-            }
-
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      if (((mutable_bitField0_ & 0x00000001) != 0)) {
-        input_ = input_.getUnmodifiableView();
-      }
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.framework.NodeProto.internal_static_tensorflow_NodeDef_descriptor;
-  }
-
-  @SuppressWarnings({"rawtypes"})
-  @java.lang.Override
-  protected com.google.protobuf.MapField internalGetMapField(
-      int number) {
-    switch (number) {
-      case 5:
-        return internalGetAttr();
-      default:
-        throw new RuntimeException(
-            "Invalid map field number: " + number);
-    }
-  }
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.framework.NodeProto.internal_static_tensorflow_NodeDef_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.framework.NodeDef.class, org.tensorflow.proto.framework.NodeDef.Builder.class);
-  }
-
-  public interface ExperimentalDebugInfoOrBuilder extends
-      // @@protoc_insertion_point(interface_extends:tensorflow.NodeDef.ExperimentalDebugInfo)
-      com.google.protobuf.MessageOrBuilder {
-
-    /**
-     * <pre>
-     * Opaque string inserted into error messages created by the runtime.
-     * This is intended to store the list of names of the nodes from the
-     * original graph that this node was derived. For example if this node, say
-     * C, was result of a fusion of 2 nodes A and B, then 'original_node' would
-     * be {A, B}. This information can be used to map errors originating at the
-     * current node to some top level source code.
-     * </pre>
-     *
-     * <code>repeated string original_node_names = 1;</code>
-     */
-    java.util.List<java.lang.String>
-        getOriginalNodeNamesList();
-    /**
-     * <pre>
-     * Opaque string inserted into error messages created by the runtime.
-     * This is intended to store the list of names of the nodes from the
-     * original graph that this node was derived. For example if this node, say
-     * C, was result of a fusion of 2 nodes A and B, then 'original_node' would
-     * be {A, B}. This information can be used to map errors originating at the
-     * current node to some top level source code.
-     * </pre>
-     *
-     * <code>repeated string original_node_names = 1;</code>
-     */
-    int getOriginalNodeNamesCount();
-    /**
-     * <pre>
-     * Opaque string inserted into error messages created by the runtime.
-     * This is intended to store the list of names of the nodes from the
-     * original graph that this node was derived. For example if this node, say
-     * C, was result of a fusion of 2 nodes A and B, then 'original_node' would
-     * be {A, B}. This information can be used to map errors originating at the
-     * current node to some top level source code.
-     * </pre>
-     *
-     * <code>repeated string original_node_names = 1;</code>
-     */
-    java.lang.String getOriginalNodeNames(int index);
-    /**
-     * <pre>
-     * Opaque string inserted into error messages created by the runtime.
-     * This is intended to store the list of names of the nodes from the
-     * original graph that this node was derived. For example if this node, say
-     * C, was result of a fusion of 2 nodes A and B, then 'original_node' would
-     * be {A, B}. This information can be used to map errors originating at the
-     * current node to some top level source code.
-     * </pre>
-     *
-     * <code>repeated string original_node_names = 1;</code>
-     */
-    com.google.protobuf.ByteString
-        getOriginalNodeNamesBytes(int index);
-
-    /**
-     * <pre>
-     * This is intended to store the list of names of the functions from the
-     * original graph that this node was derived. For example if this node, say
-     * C, was result of a fusion of node A in function FA and node B in function
-     * FB, then `original_funcs` would be {FA, FB}. If the node is in the top
-     * level graph, the `original_func` is empty. This information, with the
-     * `original_node_names` can be used to map errors originating at the
-     * current ndoe to some top level source code.
-     * </pre>
-     *
-     * <code>repeated string original_func_names = 2;</code>
-     */
-    java.util.List<java.lang.String>
-        getOriginalFuncNamesList();
-    /**
-     * <pre>
-     * This is intended to store the list of names of the functions from the
-     * original graph that this node was derived. For example if this node, say
-     * C, was result of a fusion of node A in function FA and node B in function
-     * FB, then `original_funcs` would be {FA, FB}. If the node is in the top
-     * level graph, the `original_func` is empty. This information, with the
-     * `original_node_names` can be used to map errors originating at the
-     * current ndoe to some top level source code.
-     * </pre>
-     *
-     * <code>repeated string original_func_names = 2;</code>
-     */
-    int getOriginalFuncNamesCount();
-    /**
-     * <pre>
-     * This is intended to store the list of names of the functions from the
-     * original graph that this node was derived. For example if this node, say
-     * C, was result of a fusion of node A in function FA and node B in function
-     * FB, then `original_funcs` would be {FA, FB}. If the node is in the top
-     * level graph, the `original_func` is empty. This information, with the
-     * `original_node_names` can be used to map errors originating at the
-     * current ndoe to some top level source code.
-     * </pre>
-     *
-     * <code>repeated string original_func_names = 2;</code>
-     */
-    java.lang.String getOriginalFuncNames(int index);
-    /**
-     * <pre>
-     * This is intended to store the list of names of the functions from the
-     * original graph that this node was derived. For example if this node, say
-     * C, was result of a fusion of node A in function FA and node B in function
-     * FB, then `original_funcs` would be {FA, FB}. If the node is in the top
-     * level graph, the `original_func` is empty. This information, with the
-     * `original_node_names` can be used to map errors originating at the
-     * current ndoe to some top level source code.
-     * </pre>
-     *
-     * <code>repeated string original_func_names = 2;</code>
-     */
-    com.google.protobuf.ByteString
-        getOriginalFuncNamesBytes(int index);
-  }
-  /**
-   * Protobuf type {@code tensorflow.NodeDef.ExperimentalDebugInfo}
-   */
-  public  static final class ExperimentalDebugInfo extends
-      com.google.protobuf.GeneratedMessageV3 implements
-      // @@protoc_insertion_point(message_implements:tensorflow.NodeDef.ExperimentalDebugInfo)
-      ExperimentalDebugInfoOrBuilder {
-  private static final long serialVersionUID = 0L;
-    // Use ExperimentalDebugInfo.newBuilder() to construct.
-    private ExperimentalDebugInfo(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-      super(builder);
-    }
-    private ExperimentalDebugInfo() {
-      originalNodeNames_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-      originalFuncNames_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-    }
-
-    @java.lang.Override
-    @SuppressWarnings({"unused"})
-    protected java.lang.Object newInstance(
-        UnusedPrivateParameter unused) {
-      return new ExperimentalDebugInfo();
-    }
-
-    @java.lang.Override
-    public final com.google.protobuf.UnknownFieldSet
-    getUnknownFields() {
-      return this.unknownFields;
-    }
-    private ExperimentalDebugInfo(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      this();
-      if (extensionRegistry == null) {
-        throw new java.lang.NullPointerException();
-      }
-      int mutable_bitField0_ = 0;
-      com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-          com.google.protobuf.UnknownFieldSet.newBuilder();
-      try {
-        boolean done = false;
-        while (!done) {
-          int tag = input.readTag();
-          switch (tag) {
-            case 0:
-              done = true;
-              break;
-            case 10: {
-              java.lang.String s = input.readStringRequireUtf8();
-              if (!((mutable_bitField0_ & 0x00000001) != 0)) {
-                originalNodeNames_ = new com.google.protobuf.LazyStringArrayList();
-                mutable_bitField0_ |= 0x00000001;
-              }
-              originalNodeNames_.add(s);
-              break;
-            }
-            case 18: {
-              java.lang.String s = input.readStringRequireUtf8();
-              if (!((mutable_bitField0_ & 0x00000002) != 0)) {
-                originalFuncNames_ = new com.google.protobuf.LazyStringArrayList();
-                mutable_bitField0_ |= 0x00000002;
-              }
-              originalFuncNames_.add(s);
-              break;
-            }
-            default: {
-              if (!parseUnknownField(
-                  input, unknownFields, extensionRegistry, tag)) {
-                done = true;
-              }
-              break;
-            }
-          }
-        }
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        throw e.setUnfinishedMessage(this);
-      } catch (java.io.IOException e) {
-        throw new com.google.protobuf.InvalidProtocolBufferException(
-            e).setUnfinishedMessage(this);
-      } finally {
-        if (((mutable_bitField0_ & 0x00000001) != 0)) {
-          originalNodeNames_ = originalNodeNames_.getUnmodifiableView();
-        }
-        if (((mutable_bitField0_ & 0x00000002) != 0)) {
-          originalFuncNames_ = originalFuncNames_.getUnmodifiableView();
-        }
-        this.unknownFields = unknownFields.build();
-        makeExtensionsImmutable();
-      }
-    }
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.NodeProto.internal_static_tensorflow_NodeDef_ExperimentalDebugInfo_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.NodeProto.internal_static_tensorflow_NodeDef_ExperimentalDebugInfo_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.NodeDef.ExperimentalDebugInfo.class, org.tensorflow.proto.framework.NodeDef.ExperimentalDebugInfo.Builder.class);
-    }
-
-    public static final int ORIGINAL_NODE_NAMES_FIELD_NUMBER = 1;
-    private com.google.protobuf.LazyStringList originalNodeNames_;
-    /**
-     * <pre>
-     * Opaque string inserted into error messages created by the runtime.
-     * This is intended to store the list of names of the nodes from the
-     * original graph that this node was derived. For example if this node, say
-     * C, was result of a fusion of 2 nodes A and B, then 'original_node' would
-     * be {A, B}. This information can be used to map errors originating at the
-     * current node to some top level source code.
-     * </pre>
-     *
-     * <code>repeated string original_node_names = 1;</code>
-     */
-    public com.google.protobuf.ProtocolStringList
-        getOriginalNodeNamesList() {
-      return originalNodeNames_;
-    }
-    /**
-     * <pre>
-     * Opaque string inserted into error messages created by the runtime.
-     * This is intended to store the list of names of the nodes from the
-     * original graph that this node was derived. For example if this node, say
-     * C, was result of a fusion of 2 nodes A and B, then 'original_node' would
-     * be {A, B}. This information can be used to map errors originating at the
-     * current node to some top level source code.
-     * </pre>
-     *
-     * <code>repeated string original_node_names = 1;</code>
-     */
-    public int getOriginalNodeNamesCount() {
-      return originalNodeNames_.size();
-    }
-    /**
-     * <pre>
-     * Opaque string inserted into error messages created by the runtime.
-     * This is intended to store the list of names of the nodes from the
-     * original graph that this node was derived. For example if this node, say
-     * C, was result of a fusion of 2 nodes A and B, then 'original_node' would
-     * be {A, B}. This information can be used to map errors originating at the
-     * current node to some top level source code.
-     * </pre>
-     *
-     * <code>repeated string original_node_names = 1;</code>
-     */
-    public java.lang.String getOriginalNodeNames(int index) {
-      return originalNodeNames_.get(index);
-    }
-    /**
-     * <pre>
-     * Opaque string inserted into error messages created by the runtime.
-     * This is intended to store the list of names of the nodes from the
-     * original graph that this node was derived. For example if this node, say
-     * C, was result of a fusion of 2 nodes A and B, then 'original_node' would
-     * be {A, B}. This information can be used to map errors originating at the
-     * current node to some top level source code.
-     * </pre>
-     *
-     * <code>repeated string original_node_names = 1;</code>
-     */
-    public com.google.protobuf.ByteString
-        getOriginalNodeNamesBytes(int index) {
-      return originalNodeNames_.getByteString(index);
-    }
-
-    public static final int ORIGINAL_FUNC_NAMES_FIELD_NUMBER = 2;
-    private com.google.protobuf.LazyStringList originalFuncNames_;
-    /**
-     * <pre>
-     * This is intended to store the list of names of the functions from the
-     * original graph that this node was derived. For example if this node, say
-     * C, was result of a fusion of node A in function FA and node B in function
-     * FB, then `original_funcs` would be {FA, FB}. If the node is in the top
-     * level graph, the `original_func` is empty. This information, with the
-     * `original_node_names` can be used to map errors originating at the
-     * current ndoe to some top level source code.
-     * </pre>
-     *
-     * <code>repeated string original_func_names = 2;</code>
-     */
-    public com.google.protobuf.ProtocolStringList
-        getOriginalFuncNamesList() {
-      return originalFuncNames_;
-    }
-    /**
-     * <pre>
-     * This is intended to store the list of names of the functions from the
-     * original graph that this node was derived. For example if this node, say
-     * C, was result of a fusion of node A in function FA and node B in function
-     * FB, then `original_funcs` would be {FA, FB}. If the node is in the top
-     * level graph, the `original_func` is empty. This information, with the
-     * `original_node_names` can be used to map errors originating at the
-     * current ndoe to some top level source code.
-     * </pre>
-     *
-     * <code>repeated string original_func_names = 2;</code>
-     */
-    public int getOriginalFuncNamesCount() {
-      return originalFuncNames_.size();
-    }
-    /**
-     * <pre>
-     * This is intended to store the list of names of the functions from the
-     * original graph that this node was derived. For example if this node, say
-     * C, was result of a fusion of node A in function FA and node B in function
-     * FB, then `original_funcs` would be {FA, FB}. If the node is in the top
-     * level graph, the `original_func` is empty. This information, with the
-     * `original_node_names` can be used to map errors originating at the
-     * current ndoe to some top level source code.
-     * </pre>
-     *
-     * <code>repeated string original_func_names = 2;</code>
-     */
-    public java.lang.String getOriginalFuncNames(int index) {
-      return originalFuncNames_.get(index);
-    }
-    /**
-     * <pre>
-     * This is intended to store the list of names of the functions from the
-     * original graph that this node was derived. For example if this node, say
-     * C, was result of a fusion of node A in function FA and node B in function
-     * FB, then `original_funcs` would be {FA, FB}. If the node is in the top
-     * level graph, the `original_func` is empty. This information, with the
-     * `original_node_names` can be used to map errors originating at the
-     * current ndoe to some top level source code.
-     * </pre>
-     *
-     * <code>repeated string original_func_names = 2;</code>
-     */
-    public com.google.protobuf.ByteString
-        getOriginalFuncNamesBytes(int index) {
-      return originalFuncNames_.getByteString(index);
-    }
-
-    private byte memoizedIsInitialized = -1;
-    @java.lang.Override
-    public final boolean isInitialized() {
-      byte isInitialized = memoizedIsInitialized;
-      if (isInitialized == 1) return true;
-      if (isInitialized == 0) return false;
-
-      memoizedIsInitialized = 1;
-      return true;
-    }
-
-    @java.lang.Override
-    public void writeTo(com.google.protobuf.CodedOutputStream output)
-                        throws java.io.IOException {
-      for (int i = 0; i < originalNodeNames_.size(); i++) {
-        com.google.protobuf.GeneratedMessageV3.writeString(output, 1, originalNodeNames_.getRaw(i));
-      }
-      for (int i = 0; i < originalFuncNames_.size(); i++) {
-        com.google.protobuf.GeneratedMessageV3.writeString(output, 2, originalFuncNames_.getRaw(i));
-      }
-      unknownFields.writeTo(output);
-    }
-
-    @java.lang.Override
-    public int getSerializedSize() {
-      int size = memoizedSize;
-      if (size != -1) return size;
-
-      size = 0;
-      {
-        int dataSize = 0;
-        for (int i = 0; i < originalNodeNames_.size(); i++) {
-          dataSize += computeStringSizeNoTag(originalNodeNames_.getRaw(i));
-        }
-        size += dataSize;
-        size += 1 * getOriginalNodeNamesList().size();
-      }
-      {
-        int dataSize = 0;
-        for (int i = 0; i < originalFuncNames_.size(); i++) {
-          dataSize += computeStringSizeNoTag(originalFuncNames_.getRaw(i));
-        }
-        size += dataSize;
-        size += 1 * getOriginalFuncNamesList().size();
-      }
-      size += unknownFields.getSerializedSize();
-      memoizedSize = size;
-      return size;
-    }
-
-    @java.lang.Override
-    public boolean equals(final java.lang.Object obj) {
-      if (obj == this) {
-       return true;
-      }
-      if (!(obj instanceof org.tensorflow.proto.framework.NodeDef.ExperimentalDebugInfo)) {
-        return super.equals(obj);
-      }
-      org.tensorflow.proto.framework.NodeDef.ExperimentalDebugInfo other = (org.tensorflow.proto.framework.NodeDef.ExperimentalDebugInfo) obj;
-
-      if (!getOriginalNodeNamesList()
-          .equals(other.getOriginalNodeNamesList())) return false;
-      if (!getOriginalFuncNamesList()
-          .equals(other.getOriginalFuncNamesList())) return false;
-      if (!unknownFields.equals(other.unknownFields)) return false;
-      return true;
-    }
-
-    @java.lang.Override
-    public int hashCode() {
-      if (memoizedHashCode != 0) {
-        return memoizedHashCode;
-      }
-      int hash = 41;
-      hash = (19 * hash) + getDescriptor().hashCode();
-      if (getOriginalNodeNamesCount() > 0) {
-        hash = (37 * hash) + ORIGINAL_NODE_NAMES_FIELD_NUMBER;
-        hash = (53 * hash) + getOriginalNodeNamesList().hashCode();
-      }
-      if (getOriginalFuncNamesCount() > 0) {
-        hash = (37 * hash) + ORIGINAL_FUNC_NAMES_FIELD_NUMBER;
-        hash = (53 * hash) + getOriginalFuncNamesList().hashCode();
-      }
-      hash = (29 * hash) + unknownFields.hashCode();
-      memoizedHashCode = hash;
-      return hash;
-    }
-
-    public static org.tensorflow.proto.framework.NodeDef.ExperimentalDebugInfo parseFrom(
-        java.nio.ByteBuffer data)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data);
-    }
-    public static org.tensorflow.proto.framework.NodeDef.ExperimentalDebugInfo parseFrom(
-        java.nio.ByteBuffer data,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.NodeDef.ExperimentalDebugInfo parseFrom(
-        com.google.protobuf.ByteString data)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data);
-    }
-    public static org.tensorflow.proto.framework.NodeDef.ExperimentalDebugInfo parseFrom(
-        com.google.protobuf.ByteString data,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.NodeDef.ExperimentalDebugInfo parseFrom(byte[] data)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data);
-    }
-    public static org.tensorflow.proto.framework.NodeDef.ExperimentalDebugInfo parseFrom(
-        byte[] data,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.NodeDef.ExperimentalDebugInfo parseFrom(java.io.InputStream input)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input);
-    }
-    public static org.tensorflow.proto.framework.NodeDef.ExperimentalDebugInfo parseFrom(
-        java.io.InputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.NodeDef.ExperimentalDebugInfo parseDelimitedFrom(java.io.InputStream input)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseDelimitedWithIOException(PARSER, input);
-    }
-    public static org.tensorflow.proto.framework.NodeDef.ExperimentalDebugInfo parseDelimitedFrom(
-        java.io.InputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.NodeDef.ExperimentalDebugInfo parseFrom(
-        com.google.protobuf.CodedInputStream input)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input);
-    }
-    public static org.tensorflow.proto.framework.NodeDef.ExperimentalDebugInfo parseFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input, extensionRegistry);
-    }
-
-    @java.lang.Override
-    public Builder newBuilderForType() { return newBuilder(); }
-    public static Builder newBuilder() {
-      return DEFAULT_INSTANCE.toBuilder();
-    }
-    public static Builder newBuilder(org.tensorflow.proto.framework.NodeDef.ExperimentalDebugInfo prototype) {
-      return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-    }
-    @java.lang.Override
-    public Builder toBuilder() {
-      return this == DEFAULT_INSTANCE
-          ? new Builder() : new Builder().mergeFrom(this);
-    }
-
-    @java.lang.Override
-    protected Builder newBuilderForType(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      Builder builder = new Builder(parent);
-      return builder;
-    }
-    /**
-     * Protobuf type {@code tensorflow.NodeDef.ExperimentalDebugInfo}
-     */
-    public static final class Builder extends
-        com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-        // @@protoc_insertion_point(builder_implements:tensorflow.NodeDef.ExperimentalDebugInfo)
-        org.tensorflow.proto.framework.NodeDef.ExperimentalDebugInfoOrBuilder {
-      public static final com.google.protobuf.Descriptors.Descriptor
-          getDescriptor() {
-        return org.tensorflow.proto.framework.NodeProto.internal_static_tensorflow_NodeDef_ExperimentalDebugInfo_descriptor;
-      }
-
-      @java.lang.Override
-      protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-          internalGetFieldAccessorTable() {
-        return org.tensorflow.proto.framework.NodeProto.internal_static_tensorflow_NodeDef_ExperimentalDebugInfo_fieldAccessorTable
-            .ensureFieldAccessorsInitialized(
-                org.tensorflow.proto.framework.NodeDef.ExperimentalDebugInfo.class, org.tensorflow.proto.framework.NodeDef.ExperimentalDebugInfo.Builder.class);
-      }
-
-      // Construct using org.tensorflow.proto.framework.NodeDef.ExperimentalDebugInfo.newBuilder()
-      private Builder() {
-        maybeForceBuilderInitialization();
-      }
-
-      private Builder(
-          com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-        super(parent);
-        maybeForceBuilderInitialization();
-      }
-      private void maybeForceBuilderInitialization() {
-        if (com.google.protobuf.GeneratedMessageV3
-                .alwaysUseFieldBuilders) {
-        }
-      }
-      @java.lang.Override
-      public Builder clear() {
-        super.clear();
-        originalNodeNames_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-        bitField0_ = (bitField0_ & ~0x00000001);
-        originalFuncNames_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-        bitField0_ = (bitField0_ & ~0x00000002);
-        return this;
-      }
-
-      @java.lang.Override
-      public com.google.protobuf.Descriptors.Descriptor
-          getDescriptorForType() {
-        return org.tensorflow.proto.framework.NodeProto.internal_static_tensorflow_NodeDef_ExperimentalDebugInfo_descriptor;
-      }
-
-      @java.lang.Override
-      public org.tensorflow.proto.framework.NodeDef.ExperimentalDebugInfo getDefaultInstanceForType() {
-        return org.tensorflow.proto.framework.NodeDef.ExperimentalDebugInfo.getDefaultInstance();
-      }
-
-      @java.lang.Override
-      public org.tensorflow.proto.framework.NodeDef.ExperimentalDebugInfo build() {
-        org.tensorflow.proto.framework.NodeDef.ExperimentalDebugInfo result = buildPartial();
-        if (!result.isInitialized()) {
-          throw newUninitializedMessageException(result);
-        }
-        return result;
-      }
-
-      @java.lang.Override
-      public org.tensorflow.proto.framework.NodeDef.ExperimentalDebugInfo buildPartial() {
-        org.tensorflow.proto.framework.NodeDef.ExperimentalDebugInfo result = new org.tensorflow.proto.framework.NodeDef.ExperimentalDebugInfo(this);
-        int from_bitField0_ = bitField0_;
-        if (((bitField0_ & 0x00000001) != 0)) {
-          originalNodeNames_ = originalNodeNames_.getUnmodifiableView();
-          bitField0_ = (bitField0_ & ~0x00000001);
-        }
-        result.originalNodeNames_ = originalNodeNames_;
-        if (((bitField0_ & 0x00000002) != 0)) {
-          originalFuncNames_ = originalFuncNames_.getUnmodifiableView();
-          bitField0_ = (bitField0_ & ~0x00000002);
-        }
-        result.originalFuncNames_ = originalFuncNames_;
-        onBuilt();
-        return result;
-      }
-
-      @java.lang.Override
-      public Builder clone() {
-        return super.clone();
-      }
-      @java.lang.Override
-      public Builder setField(
-          com.google.protobuf.Descriptors.FieldDescriptor field,
-          java.lang.Object value) {
-        return super.setField(field, value);
-      }
-      @java.lang.Override
-      public Builder clearField(
-          com.google.protobuf.Descriptors.FieldDescriptor field) {
-        return super.clearField(field);
-      }
-      @java.lang.Override
-      public Builder clearOneof(
-          com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-        return super.clearOneof(oneof);
-      }
-      @java.lang.Override
-      public Builder setRepeatedField(
-          com.google.protobuf.Descriptors.FieldDescriptor field,
-          int index, java.lang.Object value) {
-        return super.setRepeatedField(field, index, value);
-      }
-      @java.lang.Override
-      public Builder addRepeatedField(
-          com.google.protobuf.Descriptors.FieldDescriptor field,
-          java.lang.Object value) {
-        return super.addRepeatedField(field, value);
-      }
-      @java.lang.Override
-      public Builder mergeFrom(com.google.protobuf.Message other) {
-        if (other instanceof org.tensorflow.proto.framework.NodeDef.ExperimentalDebugInfo) {
-          return mergeFrom((org.tensorflow.proto.framework.NodeDef.ExperimentalDebugInfo)other);
-        } else {
-          super.mergeFrom(other);
-          return this;
-        }
-      }
-
-      public Builder mergeFrom(org.tensorflow.proto.framework.NodeDef.ExperimentalDebugInfo other) {
-        if (other == org.tensorflow.proto.framework.NodeDef.ExperimentalDebugInfo.getDefaultInstance()) return this;
-        if (!other.originalNodeNames_.isEmpty()) {
-          if (originalNodeNames_.isEmpty()) {
-            originalNodeNames_ = other.originalNodeNames_;
-            bitField0_ = (bitField0_ & ~0x00000001);
-          } else {
-            ensureOriginalNodeNamesIsMutable();
-            originalNodeNames_.addAll(other.originalNodeNames_);
-          }
-          onChanged();
-        }
-        if (!other.originalFuncNames_.isEmpty()) {
-          if (originalFuncNames_.isEmpty()) {
-            originalFuncNames_ = other.originalFuncNames_;
-            bitField0_ = (bitField0_ & ~0x00000002);
-          } else {
-            ensureOriginalFuncNamesIsMutable();
-            originalFuncNames_.addAll(other.originalFuncNames_);
-          }
-          onChanged();
-        }
-        this.mergeUnknownFields(other.unknownFields);
-        onChanged();
-        return this;
-      }
-
-      @java.lang.Override
-      public final boolean isInitialized() {
-        return true;
-      }
-
-      @java.lang.Override
-      public Builder mergeFrom(
-          com.google.protobuf.CodedInputStream input,
-          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-          throws java.io.IOException {
-        org.tensorflow.proto.framework.NodeDef.ExperimentalDebugInfo parsedMessage = null;
-        try {
-          parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-        } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-          parsedMessage = (org.tensorflow.proto.framework.NodeDef.ExperimentalDebugInfo) e.getUnfinishedMessage();
-          throw e.unwrapIOException();
-        } finally {
-          if (parsedMessage != null) {
-            mergeFrom(parsedMessage);
-          }
-        }
-        return this;
-      }
-      private int bitField0_;
-
-      private com.google.protobuf.LazyStringList originalNodeNames_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-      private void ensureOriginalNodeNamesIsMutable() {
-        if (!((bitField0_ & 0x00000001) != 0)) {
-          originalNodeNames_ = new com.google.protobuf.LazyStringArrayList(originalNodeNames_);
-          bitField0_ |= 0x00000001;
-         }
-      }
-      /**
-       * <pre>
-       * Opaque string inserted into error messages created by the runtime.
-       * This is intended to store the list of names of the nodes from the
-       * original graph that this node was derived. For example if this node, say
-       * C, was result of a fusion of 2 nodes A and B, then 'original_node' would
-       * be {A, B}. This information can be used to map errors originating at the
-       * current node to some top level source code.
-       * </pre>
-       *
-       * <code>repeated string original_node_names = 1;</code>
-       */
-      public com.google.protobuf.ProtocolStringList
-          getOriginalNodeNamesList() {
-        return originalNodeNames_.getUnmodifiableView();
-      }
-      /**
-       * <pre>
-       * Opaque string inserted into error messages created by the runtime.
-       * This is intended to store the list of names of the nodes from the
-       * original graph that this node was derived. For example if this node, say
-       * C, was result of a fusion of 2 nodes A and B, then 'original_node' would
-       * be {A, B}. This information can be used to map errors originating at the
-       * current node to some top level source code.
-       * </pre>
-       *
-       * <code>repeated string original_node_names = 1;</code>
-       */
-      public int getOriginalNodeNamesCount() {
-        return originalNodeNames_.size();
-      }
-      /**
-       * <pre>
-       * Opaque string inserted into error messages created by the runtime.
-       * This is intended to store the list of names of the nodes from the
-       * original graph that this node was derived. For example if this node, say
-       * C, was result of a fusion of 2 nodes A and B, then 'original_node' would
-       * be {A, B}. This information can be used to map errors originating at the
-       * current node to some top level source code.
-       * </pre>
-       *
-       * <code>repeated string original_node_names = 1;</code>
-       */
-      public java.lang.String getOriginalNodeNames(int index) {
-        return originalNodeNames_.get(index);
-      }
-      /**
-       * <pre>
-       * Opaque string inserted into error messages created by the runtime.
-       * This is intended to store the list of names of the nodes from the
-       * original graph that this node was derived. For example if this node, say
-       * C, was result of a fusion of 2 nodes A and B, then 'original_node' would
-       * be {A, B}. This information can be used to map errors originating at the
-       * current node to some top level source code.
-       * </pre>
-       *
-       * <code>repeated string original_node_names = 1;</code>
-       */
-      public com.google.protobuf.ByteString
-          getOriginalNodeNamesBytes(int index) {
-        return originalNodeNames_.getByteString(index);
-      }
-      /**
-       * <pre>
-       * Opaque string inserted into error messages created by the runtime.
-       * This is intended to store the list of names of the nodes from the
-       * original graph that this node was derived. For example if this node, say
-       * C, was result of a fusion of 2 nodes A and B, then 'original_node' would
-       * be {A, B}. This information can be used to map errors originating at the
-       * current node to some top level source code.
-       * </pre>
-       *
-       * <code>repeated string original_node_names = 1;</code>
-       */
-      public Builder setOriginalNodeNames(
-          int index, java.lang.String value) {
-        if (value == null) {
-    throw new NullPointerException();
-  }
-  ensureOriginalNodeNamesIsMutable();
-        originalNodeNames_.set(index, value);
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * Opaque string inserted into error messages created by the runtime.
-       * This is intended to store the list of names of the nodes from the
-       * original graph that this node was derived. For example if this node, say
-       * C, was result of a fusion of 2 nodes A and B, then 'original_node' would
-       * be {A, B}. This information can be used to map errors originating at the
-       * current node to some top level source code.
-       * </pre>
-       *
-       * <code>repeated string original_node_names = 1;</code>
-       */
-      public Builder addOriginalNodeNames(
-          java.lang.String value) {
-        if (value == null) {
-    throw new NullPointerException();
-  }
-  ensureOriginalNodeNamesIsMutable();
-        originalNodeNames_.add(value);
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * Opaque string inserted into error messages created by the runtime.
-       * This is intended to store the list of names of the nodes from the
-       * original graph that this node was derived. For example if this node, say
-       * C, was result of a fusion of 2 nodes A and B, then 'original_node' would
-       * be {A, B}. This information can be used to map errors originating at the
-       * current node to some top level source code.
-       * </pre>
-       *
-       * <code>repeated string original_node_names = 1;</code>
-       */
-      public Builder addAllOriginalNodeNames(
-          java.lang.Iterable<java.lang.String> values) {
-        ensureOriginalNodeNamesIsMutable();
-        com.google.protobuf.AbstractMessageLite.Builder.addAll(
-            values, originalNodeNames_);
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * Opaque string inserted into error messages created by the runtime.
-       * This is intended to store the list of names of the nodes from the
-       * original graph that this node was derived. For example if this node, say
-       * C, was result of a fusion of 2 nodes A and B, then 'original_node' would
-       * be {A, B}. This information can be used to map errors originating at the
-       * current node to some top level source code.
-       * </pre>
-       *
-       * <code>repeated string original_node_names = 1;</code>
-       */
-      public Builder clearOriginalNodeNames() {
-        originalNodeNames_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-        bitField0_ = (bitField0_ & ~0x00000001);
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * Opaque string inserted into error messages created by the runtime.
-       * This is intended to store the list of names of the nodes from the
-       * original graph that this node was derived. For example if this node, say
-       * C, was result of a fusion of 2 nodes A and B, then 'original_node' would
-       * be {A, B}. This information can be used to map errors originating at the
-       * current node to some top level source code.
-       * </pre>
-       *
-       * <code>repeated string original_node_names = 1;</code>
-       */
-      public Builder addOriginalNodeNamesBytes(
-          com.google.protobuf.ByteString value) {
-        if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-        ensureOriginalNodeNamesIsMutable();
-        originalNodeNames_.add(value);
-        onChanged();
-        return this;
-      }
-
-      private com.google.protobuf.LazyStringList originalFuncNames_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-      private void ensureOriginalFuncNamesIsMutable() {
-        if (!((bitField0_ & 0x00000002) != 0)) {
-          originalFuncNames_ = new com.google.protobuf.LazyStringArrayList(originalFuncNames_);
-          bitField0_ |= 0x00000002;
-         }
-      }
-      /**
-       * <pre>
-       * This is intended to store the list of names of the functions from the
-       * original graph that this node was derived. For example if this node, say
-       * C, was result of a fusion of node A in function FA and node B in function
-       * FB, then `original_funcs` would be {FA, FB}. If the node is in the top
-       * level graph, the `original_func` is empty. This information, with the
-       * `original_node_names` can be used to map errors originating at the
-       * current ndoe to some top level source code.
-       * </pre>
-       *
-       * <code>repeated string original_func_names = 2;</code>
-       */
-      public com.google.protobuf.ProtocolStringList
-          getOriginalFuncNamesList() {
-        return originalFuncNames_.getUnmodifiableView();
-      }
-      /**
-       * <pre>
-       * This is intended to store the list of names of the functions from the
-       * original graph that this node was derived. For example if this node, say
-       * C, was result of a fusion of node A in function FA and node B in function
-       * FB, then `original_funcs` would be {FA, FB}. If the node is in the top
-       * level graph, the `original_func` is empty. This information, with the
-       * `original_node_names` can be used to map errors originating at the
-       * current ndoe to some top level source code.
-       * </pre>
-       *
-       * <code>repeated string original_func_names = 2;</code>
-       */
-      public int getOriginalFuncNamesCount() {
-        return originalFuncNames_.size();
-      }
-      /**
-       * <pre>
-       * This is intended to store the list of names of the functions from the
-       * original graph that this node was derived. For example if this node, say
-       * C, was result of a fusion of node A in function FA and node B in function
-       * FB, then `original_funcs` would be {FA, FB}. If the node is in the top
-       * level graph, the `original_func` is empty. This information, with the
-       * `original_node_names` can be used to map errors originating at the
-       * current ndoe to some top level source code.
-       * </pre>
-       *
-       * <code>repeated string original_func_names = 2;</code>
-       */
-      public java.lang.String getOriginalFuncNames(int index) {
-        return originalFuncNames_.get(index);
-      }
-      /**
-       * <pre>
-       * This is intended to store the list of names of the functions from the
-       * original graph that this node was derived. For example if this node, say
-       * C, was result of a fusion of node A in function FA and node B in function
-       * FB, then `original_funcs` would be {FA, FB}. If the node is in the top
-       * level graph, the `original_func` is empty. This information, with the
-       * `original_node_names` can be used to map errors originating at the
-       * current ndoe to some top level source code.
-       * </pre>
-       *
-       * <code>repeated string original_func_names = 2;</code>
-       */
-      public com.google.protobuf.ByteString
-          getOriginalFuncNamesBytes(int index) {
-        return originalFuncNames_.getByteString(index);
-      }
-      /**
-       * <pre>
-       * This is intended to store the list of names of the functions from the
-       * original graph that this node was derived. For example if this node, say
-       * C, was result of a fusion of node A in function FA and node B in function
-       * FB, then `original_funcs` would be {FA, FB}. If the node is in the top
-       * level graph, the `original_func` is empty. This information, with the
-       * `original_node_names` can be used to map errors originating at the
-       * current ndoe to some top level source code.
-       * </pre>
-       *
-       * <code>repeated string original_func_names = 2;</code>
-       */
-      public Builder setOriginalFuncNames(
-          int index, java.lang.String value) {
-        if (value == null) {
-    throw new NullPointerException();
-  }
-  ensureOriginalFuncNamesIsMutable();
-        originalFuncNames_.set(index, value);
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * This is intended to store the list of names of the functions from the
-       * original graph that this node was derived. For example if this node, say
-       * C, was result of a fusion of node A in function FA and node B in function
-       * FB, then `original_funcs` would be {FA, FB}. If the node is in the top
-       * level graph, the `original_func` is empty. This information, with the
-       * `original_node_names` can be used to map errors originating at the
-       * current ndoe to some top level source code.
-       * </pre>
-       *
-       * <code>repeated string original_func_names = 2;</code>
-       */
-      public Builder addOriginalFuncNames(
-          java.lang.String value) {
-        if (value == null) {
-    throw new NullPointerException();
-  }
-  ensureOriginalFuncNamesIsMutable();
-        originalFuncNames_.add(value);
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * This is intended to store the list of names of the functions from the
-       * original graph that this node was derived. For example if this node, say
-       * C, was result of a fusion of node A in function FA and node B in function
-       * FB, then `original_funcs` would be {FA, FB}. If the node is in the top
-       * level graph, the `original_func` is empty. This information, with the
-       * `original_node_names` can be used to map errors originating at the
-       * current ndoe to some top level source code.
-       * </pre>
-       *
-       * <code>repeated string original_func_names = 2;</code>
-       */
-      public Builder addAllOriginalFuncNames(
-          java.lang.Iterable<java.lang.String> values) {
-        ensureOriginalFuncNamesIsMutable();
-        com.google.protobuf.AbstractMessageLite.Builder.addAll(
-            values, originalFuncNames_);
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * This is intended to store the list of names of the functions from the
-       * original graph that this node was derived. For example if this node, say
-       * C, was result of a fusion of node A in function FA and node B in function
-       * FB, then `original_funcs` would be {FA, FB}. If the node is in the top
-       * level graph, the `original_func` is empty. This information, with the
-       * `original_node_names` can be used to map errors originating at the
-       * current ndoe to some top level source code.
-       * </pre>
-       *
-       * <code>repeated string original_func_names = 2;</code>
-       */
-      public Builder clearOriginalFuncNames() {
-        originalFuncNames_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-        bitField0_ = (bitField0_ & ~0x00000002);
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * This is intended to store the list of names of the functions from the
-       * original graph that this node was derived. For example if this node, say
-       * C, was result of a fusion of node A in function FA and node B in function
-       * FB, then `original_funcs` would be {FA, FB}. If the node is in the top
-       * level graph, the `original_func` is empty. This information, with the
-       * `original_node_names` can be used to map errors originating at the
-       * current ndoe to some top level source code.
-       * </pre>
-       *
-       * <code>repeated string original_func_names = 2;</code>
-       */
-      public Builder addOriginalFuncNamesBytes(
-          com.google.protobuf.ByteString value) {
-        if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-        ensureOriginalFuncNamesIsMutable();
-        originalFuncNames_.add(value);
-        onChanged();
-        return this;
-      }
-      @java.lang.Override
-      public final Builder setUnknownFields(
-          final com.google.protobuf.UnknownFieldSet unknownFields) {
-        return super.setUnknownFields(unknownFields);
-      }
-
-      @java.lang.Override
-      public final Builder mergeUnknownFields(
-          final com.google.protobuf.UnknownFieldSet unknownFields) {
-        return super.mergeUnknownFields(unknownFields);
-      }
-
-
-      // @@protoc_insertion_point(builder_scope:tensorflow.NodeDef.ExperimentalDebugInfo)
-    }
-
-    // @@protoc_insertion_point(class_scope:tensorflow.NodeDef.ExperimentalDebugInfo)
-    private static final org.tensorflow.proto.framework.NodeDef.ExperimentalDebugInfo DEFAULT_INSTANCE;
-    static {
-      DEFAULT_INSTANCE = new org.tensorflow.proto.framework.NodeDef.ExperimentalDebugInfo();
-    }
-
-    public static org.tensorflow.proto.framework.NodeDef.ExperimentalDebugInfo getDefaultInstance() {
-      return DEFAULT_INSTANCE;
-    }
-
-    private static final com.google.protobuf.Parser<ExperimentalDebugInfo>
-        PARSER = new com.google.protobuf.AbstractParser<ExperimentalDebugInfo>() {
-      @java.lang.Override
-      public ExperimentalDebugInfo parsePartialFrom(
-          com.google.protobuf.CodedInputStream input,
-          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-          throws com.google.protobuf.InvalidProtocolBufferException {
-        return new ExperimentalDebugInfo(input, extensionRegistry);
-      }
-    };
-
-    public static com.google.protobuf.Parser<ExperimentalDebugInfo> parser() {
-      return PARSER;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Parser<ExperimentalDebugInfo> getParserForType() {
-      return PARSER;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.NodeDef.ExperimentalDebugInfo getDefaultInstanceForType() {
-      return DEFAULT_INSTANCE;
-    }
-
-  }
-
-  public static final int NAME_FIELD_NUMBER = 1;
-  private volatile java.lang.Object name_;
-  /**
-   * <pre>
-   * The name given to this operator. Used for naming inputs,
-   * logging, visualization, etc.  Unique within a single GraphDef.
-   * Must match the regexp "[A-Za-z0-9.][A-Za-z0-9_&gt;./]*".
-   * </pre>
-   *
-   * <code>string name = 1;</code>
-   */
-  public java.lang.String getName() {
-    java.lang.Object ref = name_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      name_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * The name given to this operator. Used for naming inputs,
-   * logging, visualization, etc.  Unique within a single GraphDef.
-   * Must match the regexp "[A-Za-z0-9.][A-Za-z0-9_&gt;./]*".
-   * </pre>
-   *
-   * <code>string name = 1;</code>
-   */
-  public com.google.protobuf.ByteString
-      getNameBytes() {
-    java.lang.Object ref = name_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      name_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int OP_FIELD_NUMBER = 2;
-  private volatile java.lang.Object op_;
-  /**
-   * <pre>
-   * The operation name.  There may be custom parameters in attrs.
-   * Op names starting with an underscore are reserved for internal use.
-   * </pre>
-   *
-   * <code>string op = 2;</code>
-   */
-  public java.lang.String getOp() {
-    java.lang.Object ref = op_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      op_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * The operation name.  There may be custom parameters in attrs.
-   * Op names starting with an underscore are reserved for internal use.
-   * </pre>
-   *
-   * <code>string op = 2;</code>
-   */
-  public com.google.protobuf.ByteString
-      getOpBytes() {
-    java.lang.Object ref = op_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      op_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int INPUT_FIELD_NUMBER = 3;
-  private com.google.protobuf.LazyStringList input_;
-  /**
-   * <pre>
-   * Each input is "node:src_output" with "node" being a string name and
-   * "src_output" indicating which output tensor to use from "node". If
-   * "src_output" is 0 the ":0" suffix can be omitted.  Regular inputs
-   * may optionally be followed by control inputs that have the format
-   * "^node".
-   * </pre>
-   *
-   * <code>repeated string input = 3;</code>
-   */
-  public com.google.protobuf.ProtocolStringList
-      getInputList() {
-    return input_;
-  }
-  /**
-   * <pre>
-   * Each input is "node:src_output" with "node" being a string name and
-   * "src_output" indicating which output tensor to use from "node". If
-   * "src_output" is 0 the ":0" suffix can be omitted.  Regular inputs
-   * may optionally be followed by control inputs that have the format
-   * "^node".
-   * </pre>
-   *
-   * <code>repeated string input = 3;</code>
-   */
-  public int getInputCount() {
-    return input_.size();
-  }
-  /**
-   * <pre>
-   * Each input is "node:src_output" with "node" being a string name and
-   * "src_output" indicating which output tensor to use from "node". If
-   * "src_output" is 0 the ":0" suffix can be omitted.  Regular inputs
-   * may optionally be followed by control inputs that have the format
-   * "^node".
-   * </pre>
-   *
-   * <code>repeated string input = 3;</code>
-   */
-  public java.lang.String getInput(int index) {
-    return input_.get(index);
-  }
-  /**
-   * <pre>
-   * Each input is "node:src_output" with "node" being a string name and
-   * "src_output" indicating which output tensor to use from "node". If
-   * "src_output" is 0 the ":0" suffix can be omitted.  Regular inputs
-   * may optionally be followed by control inputs that have the format
-   * "^node".
-   * </pre>
-   *
-   * <code>repeated string input = 3;</code>
-   */
-  public com.google.protobuf.ByteString
-      getInputBytes(int index) {
-    return input_.getByteString(index);
-  }
-
-  public static final int DEVICE_FIELD_NUMBER = 4;
-  private volatile java.lang.Object device_;
-  /**
-   * <pre>
-   * A (possibly partial) specification for the device on which this
-   * node should be placed.
-   * The expected syntax for this string is as follows:
-   * DEVICE_SPEC ::= PARTIAL_SPEC
-   * PARTIAL_SPEC ::= ("/" CONSTRAINT) *
-   * CONSTRAINT ::= ("job:" JOB_NAME)
-   *              | ("replica:" [1-9][0-9]*)
-   *              | ("task:" [1-9][0-9]*)
-   *              | ("device:" [A-Za-z]* ":" ([1-9][0-9]* | "*") )
-   * Valid values for this string include:
-   * * "/job:worker/replica:0/task:1/device:GPU:3"  (full specification)
-   * * "/job:worker/device:GPU:3"                   (partial specification)
-   * * ""                                    (no specification)
-   * If the constraints do not resolve to a single device (or if this
-   * field is empty or not present), the runtime will attempt to
-   * choose a device automatically.
-   * </pre>
-   *
-   * <code>string device = 4;</code>
-   */
-  public java.lang.String getDevice() {
-    java.lang.Object ref = device_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      device_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * A (possibly partial) specification for the device on which this
-   * node should be placed.
-   * The expected syntax for this string is as follows:
-   * DEVICE_SPEC ::= PARTIAL_SPEC
-   * PARTIAL_SPEC ::= ("/" CONSTRAINT) *
-   * CONSTRAINT ::= ("job:" JOB_NAME)
-   *              | ("replica:" [1-9][0-9]*)
-   *              | ("task:" [1-9][0-9]*)
-   *              | ("device:" [A-Za-z]* ":" ([1-9][0-9]* | "*") )
-   * Valid values for this string include:
-   * * "/job:worker/replica:0/task:1/device:GPU:3"  (full specification)
-   * * "/job:worker/device:GPU:3"                   (partial specification)
-   * * ""                                    (no specification)
-   * If the constraints do not resolve to a single device (or if this
-   * field is empty or not present), the runtime will attempt to
-   * choose a device automatically.
-   * </pre>
-   *
-   * <code>string device = 4;</code>
-   */
-  public com.google.protobuf.ByteString
-      getDeviceBytes() {
-    java.lang.Object ref = device_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      device_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int ATTR_FIELD_NUMBER = 5;
-  private static final class AttrDefaultEntryHolder {
-    static final com.google.protobuf.MapEntry<
-        java.lang.String, org.tensorflow.proto.framework.AttrValue> defaultEntry =
-            com.google.protobuf.MapEntry
-            .<java.lang.String, org.tensorflow.proto.framework.AttrValue>newDefaultInstance(
-                org.tensorflow.proto.framework.NodeProto.internal_static_tensorflow_NodeDef_AttrEntry_descriptor, 
-                com.google.protobuf.WireFormat.FieldType.STRING,
-                "",
-                com.google.protobuf.WireFormat.FieldType.MESSAGE,
-                org.tensorflow.proto.framework.AttrValue.getDefaultInstance());
-  }
-  private com.google.protobuf.MapField<
-      java.lang.String, org.tensorflow.proto.framework.AttrValue> attr_;
-  private com.google.protobuf.MapField<java.lang.String, org.tensorflow.proto.framework.AttrValue>
-  internalGetAttr() {
-    if (attr_ == null) {
-      return com.google.protobuf.MapField.emptyMapField(
-          AttrDefaultEntryHolder.defaultEntry);
-    }
-    return attr_;
-  }
-
-  public int getAttrCount() {
-    return internalGetAttr().getMap().size();
-  }
-  /**
-   * <pre>
-   * Operation-specific graph-construction-time configuration.
-   * Note that this should include all attrs defined in the
-   * corresponding OpDef, including those with a value matching
-   * the default -- this allows the default to change and makes
-   * NodeDefs easier to interpret on their own.  However, if
-   * an attr with a default is not specified in this list, the
-   * default will be used.
-   * The "names" (keys) must match the regexp "[a-z][a-z0-9_]+" (and
-   * one of the names from the corresponding OpDef's attr field).
-   * The values must have a type matching the corresponding OpDef
-   * attr's type field.
-   * TODO(josh11b): Add some examples here showing best practices.
-   * </pre>
-   *
-   * <code>map&lt;string, .tensorflow.AttrValue&gt; attr = 5;</code>
-   */
-
-  public boolean containsAttr(
-      java.lang.String key) {
-    if (key == null) { throw new java.lang.NullPointerException(); }
-    return internalGetAttr().getMap().containsKey(key);
-  }
-  /**
-   * Use {@link #getAttrMap()} instead.
-   */
-  @java.lang.Deprecated
-  public java.util.Map<java.lang.String, org.tensorflow.proto.framework.AttrValue> getAttr() {
-    return getAttrMap();
-  }
-  /**
-   * <pre>
-   * Operation-specific graph-construction-time configuration.
-   * Note that this should include all attrs defined in the
-   * corresponding OpDef, including those with a value matching
-   * the default -- this allows the default to change and makes
-   * NodeDefs easier to interpret on their own.  However, if
-   * an attr with a default is not specified in this list, the
-   * default will be used.
-   * The "names" (keys) must match the regexp "[a-z][a-z0-9_]+" (and
-   * one of the names from the corresponding OpDef's attr field).
-   * The values must have a type matching the corresponding OpDef
-   * attr's type field.
-   * TODO(josh11b): Add some examples here showing best practices.
-   * </pre>
-   *
-   * <code>map&lt;string, .tensorflow.AttrValue&gt; attr = 5;</code>
-   */
-
-  public java.util.Map<java.lang.String, org.tensorflow.proto.framework.AttrValue> getAttrMap() {
-    return internalGetAttr().getMap();
-  }
-  /**
-   * <pre>
-   * Operation-specific graph-construction-time configuration.
-   * Note that this should include all attrs defined in the
-   * corresponding OpDef, including those with a value matching
-   * the default -- this allows the default to change and makes
-   * NodeDefs easier to interpret on their own.  However, if
-   * an attr with a default is not specified in this list, the
-   * default will be used.
-   * The "names" (keys) must match the regexp "[a-z][a-z0-9_]+" (and
-   * one of the names from the corresponding OpDef's attr field).
-   * The values must have a type matching the corresponding OpDef
-   * attr's type field.
-   * TODO(josh11b): Add some examples here showing best practices.
-   * </pre>
-   *
-   * <code>map&lt;string, .tensorflow.AttrValue&gt; attr = 5;</code>
-   */
-
-  public org.tensorflow.proto.framework.AttrValue getAttrOrDefault(
-      java.lang.String key,
-      org.tensorflow.proto.framework.AttrValue defaultValue) {
-    if (key == null) { throw new java.lang.NullPointerException(); }
-    java.util.Map<java.lang.String, org.tensorflow.proto.framework.AttrValue> map =
-        internalGetAttr().getMap();
-    return map.containsKey(key) ? map.get(key) : defaultValue;
-  }
-  /**
-   * <pre>
-   * Operation-specific graph-construction-time configuration.
-   * Note that this should include all attrs defined in the
-   * corresponding OpDef, including those with a value matching
-   * the default -- this allows the default to change and makes
-   * NodeDefs easier to interpret on their own.  However, if
-   * an attr with a default is not specified in this list, the
-   * default will be used.
-   * The "names" (keys) must match the regexp "[a-z][a-z0-9_]+" (and
-   * one of the names from the corresponding OpDef's attr field).
-   * The values must have a type matching the corresponding OpDef
-   * attr's type field.
-   * TODO(josh11b): Add some examples here showing best practices.
-   * </pre>
-   *
-   * <code>map&lt;string, .tensorflow.AttrValue&gt; attr = 5;</code>
-   */
-
-  public org.tensorflow.proto.framework.AttrValue getAttrOrThrow(
-      java.lang.String key) {
-    if (key == null) { throw new java.lang.NullPointerException(); }
-    java.util.Map<java.lang.String, org.tensorflow.proto.framework.AttrValue> map =
-        internalGetAttr().getMap();
-    if (!map.containsKey(key)) {
-      throw new java.lang.IllegalArgumentException();
-    }
-    return map.get(key);
-  }
-
-  public static final int EXPERIMENTAL_DEBUG_INFO_FIELD_NUMBER = 6;
-  private org.tensorflow.proto.framework.NodeDef.ExperimentalDebugInfo experimentalDebugInfo_;
-  /**
-   * <pre>
-   * This stores debug information associated with the node.
-   * </pre>
-   *
-   * <code>.tensorflow.NodeDef.ExperimentalDebugInfo experimental_debug_info = 6;</code>
-   */
-  public boolean hasExperimentalDebugInfo() {
-    return experimentalDebugInfo_ != null;
-  }
-  /**
-   * <pre>
-   * This stores debug information associated with the node.
-   * </pre>
-   *
-   * <code>.tensorflow.NodeDef.ExperimentalDebugInfo experimental_debug_info = 6;</code>
-   */
-  public org.tensorflow.proto.framework.NodeDef.ExperimentalDebugInfo getExperimentalDebugInfo() {
-    return experimentalDebugInfo_ == null ? org.tensorflow.proto.framework.NodeDef.ExperimentalDebugInfo.getDefaultInstance() : experimentalDebugInfo_;
-  }
-  /**
-   * <pre>
-   * This stores debug information associated with the node.
-   * </pre>
-   *
-   * <code>.tensorflow.NodeDef.ExperimentalDebugInfo experimental_debug_info = 6;</code>
-   */
-  public org.tensorflow.proto.framework.NodeDef.ExperimentalDebugInfoOrBuilder getExperimentalDebugInfoOrBuilder() {
-    return getExperimentalDebugInfo();
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    if (!getNameBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 1, name_);
-    }
-    if (!getOpBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 2, op_);
-    }
-    for (int i = 0; i < input_.size(); i++) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 3, input_.getRaw(i));
-    }
-    if (!getDeviceBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 4, device_);
-    }
-    com.google.protobuf.GeneratedMessageV3
-      .serializeStringMapTo(
-        output,
-        internalGetAttr(),
-        AttrDefaultEntryHolder.defaultEntry,
-        5);
-    if (experimentalDebugInfo_ != null) {
-      output.writeMessage(6, getExperimentalDebugInfo());
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (!getNameBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, name_);
-    }
-    if (!getOpBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(2, op_);
-    }
-    {
-      int dataSize = 0;
-      for (int i = 0; i < input_.size(); i++) {
-        dataSize += computeStringSizeNoTag(input_.getRaw(i));
-      }
-      size += dataSize;
-      size += 1 * getInputList().size();
-    }
-    if (!getDeviceBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(4, device_);
-    }
-    for (java.util.Map.Entry<java.lang.String, org.tensorflow.proto.framework.AttrValue> entry
-         : internalGetAttr().getMap().entrySet()) {
-      com.google.protobuf.MapEntry<java.lang.String, org.tensorflow.proto.framework.AttrValue>
-      attr__ = AttrDefaultEntryHolder.defaultEntry.newBuilderForType()
-          .setKey(entry.getKey())
-          .setValue(entry.getValue())
-          .build();
-      size += com.google.protobuf.CodedOutputStream
-          .computeMessageSize(5, attr__);
-    }
-    if (experimentalDebugInfo_ != null) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(6, getExperimentalDebugInfo());
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.framework.NodeDef)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.framework.NodeDef other = (org.tensorflow.proto.framework.NodeDef) obj;
-
-    if (!getName()
-        .equals(other.getName())) return false;
-    if (!getOp()
-        .equals(other.getOp())) return false;
-    if (!getInputList()
-        .equals(other.getInputList())) return false;
-    if (!getDevice()
-        .equals(other.getDevice())) return false;
-    if (!internalGetAttr().equals(
-        other.internalGetAttr())) return false;
-    if (hasExperimentalDebugInfo() != other.hasExperimentalDebugInfo()) return false;
-    if (hasExperimentalDebugInfo()) {
-      if (!getExperimentalDebugInfo()
-          .equals(other.getExperimentalDebugInfo())) return false;
-    }
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    hash = (37 * hash) + NAME_FIELD_NUMBER;
-    hash = (53 * hash) + getName().hashCode();
-    hash = (37 * hash) + OP_FIELD_NUMBER;
-    hash = (53 * hash) + getOp().hashCode();
-    if (getInputCount() > 0) {
-      hash = (37 * hash) + INPUT_FIELD_NUMBER;
-      hash = (53 * hash) + getInputList().hashCode();
-    }
-    hash = (37 * hash) + DEVICE_FIELD_NUMBER;
-    hash = (53 * hash) + getDevice().hashCode();
-    if (!internalGetAttr().getMap().isEmpty()) {
-      hash = (37 * hash) + ATTR_FIELD_NUMBER;
-      hash = (53 * hash) + internalGetAttr().hashCode();
-    }
-    if (hasExperimentalDebugInfo()) {
-      hash = (37 * hash) + EXPERIMENTAL_DEBUG_INFO_FIELD_NUMBER;
-      hash = (53 * hash) + getExperimentalDebugInfo().hashCode();
-    }
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.framework.NodeDef parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.NodeDef parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.NodeDef parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.NodeDef parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.NodeDef parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.NodeDef parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.NodeDef parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.NodeDef parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.NodeDef parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.NodeDef parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.NodeDef parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.NodeDef parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.framework.NodeDef prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * Protobuf type {@code tensorflow.NodeDef}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.NodeDef)
-      org.tensorflow.proto.framework.NodeDefOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.NodeProto.internal_static_tensorflow_NodeDef_descriptor;
-    }
-
-    @SuppressWarnings({"rawtypes"})
-    protected com.google.protobuf.MapField internalGetMapField(
-        int number) {
-      switch (number) {
-        case 5:
-          return internalGetAttr();
-        default:
-          throw new RuntimeException(
-              "Invalid map field number: " + number);
-      }
-    }
-    @SuppressWarnings({"rawtypes"})
-    protected com.google.protobuf.MapField internalGetMutableMapField(
-        int number) {
-      switch (number) {
-        case 5:
-          return internalGetMutableAttr();
-        default:
-          throw new RuntimeException(
-              "Invalid map field number: " + number);
-      }
-    }
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.NodeProto.internal_static_tensorflow_NodeDef_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.NodeDef.class, org.tensorflow.proto.framework.NodeDef.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.framework.NodeDef.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      name_ = "";
-
-      op_ = "";
-
-      input_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-      bitField0_ = (bitField0_ & ~0x00000001);
-      device_ = "";
-
-      internalGetMutableAttr().clear();
-      if (experimentalDebugInfoBuilder_ == null) {
-        experimentalDebugInfo_ = null;
-      } else {
-        experimentalDebugInfo_ = null;
-        experimentalDebugInfoBuilder_ = null;
-      }
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.framework.NodeProto.internal_static_tensorflow_NodeDef_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.NodeDef getDefaultInstanceForType() {
-      return org.tensorflow.proto.framework.NodeDef.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.NodeDef build() {
-      org.tensorflow.proto.framework.NodeDef result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.NodeDef buildPartial() {
-      org.tensorflow.proto.framework.NodeDef result = new org.tensorflow.proto.framework.NodeDef(this);
-      int from_bitField0_ = bitField0_;
-      result.name_ = name_;
-      result.op_ = op_;
-      if (((bitField0_ & 0x00000001) != 0)) {
-        input_ = input_.getUnmodifiableView();
-        bitField0_ = (bitField0_ & ~0x00000001);
-      }
-      result.input_ = input_;
-      result.device_ = device_;
-      result.attr_ = internalGetAttr();
-      result.attr_.makeImmutable();
-      if (experimentalDebugInfoBuilder_ == null) {
-        result.experimentalDebugInfo_ = experimentalDebugInfo_;
-      } else {
-        result.experimentalDebugInfo_ = experimentalDebugInfoBuilder_.build();
-      }
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.framework.NodeDef) {
-        return mergeFrom((org.tensorflow.proto.framework.NodeDef)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.framework.NodeDef other) {
-      if (other == org.tensorflow.proto.framework.NodeDef.getDefaultInstance()) return this;
-      if (!other.getName().isEmpty()) {
-        name_ = other.name_;
-        onChanged();
-      }
-      if (!other.getOp().isEmpty()) {
-        op_ = other.op_;
-        onChanged();
-      }
-      if (!other.input_.isEmpty()) {
-        if (input_.isEmpty()) {
-          input_ = other.input_;
-          bitField0_ = (bitField0_ & ~0x00000001);
-        } else {
-          ensureInputIsMutable();
-          input_.addAll(other.input_);
-        }
-        onChanged();
-      }
-      if (!other.getDevice().isEmpty()) {
-        device_ = other.device_;
-        onChanged();
-      }
-      internalGetMutableAttr().mergeFrom(
-          other.internalGetAttr());
-      if (other.hasExperimentalDebugInfo()) {
-        mergeExperimentalDebugInfo(other.getExperimentalDebugInfo());
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.framework.NodeDef parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.framework.NodeDef) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-    private int bitField0_;
-
-    private java.lang.Object name_ = "";
-    /**
-     * <pre>
-     * The name given to this operator. Used for naming inputs,
-     * logging, visualization, etc.  Unique within a single GraphDef.
-     * Must match the regexp "[A-Za-z0-9.][A-Za-z0-9_&gt;./]*".
-     * </pre>
-     *
-     * <code>string name = 1;</code>
-     */
-    public java.lang.String getName() {
-      java.lang.Object ref = name_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        name_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * The name given to this operator. Used for naming inputs,
-     * logging, visualization, etc.  Unique within a single GraphDef.
-     * Must match the regexp "[A-Za-z0-9.][A-Za-z0-9_&gt;./]*".
-     * </pre>
-     *
-     * <code>string name = 1;</code>
-     */
-    public com.google.protobuf.ByteString
-        getNameBytes() {
-      java.lang.Object ref = name_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        name_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * The name given to this operator. Used for naming inputs,
-     * logging, visualization, etc.  Unique within a single GraphDef.
-     * Must match the regexp "[A-Za-z0-9.][A-Za-z0-9_&gt;./]*".
-     * </pre>
-     *
-     * <code>string name = 1;</code>
-     */
-    public Builder setName(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      name_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * The name given to this operator. Used for naming inputs,
-     * logging, visualization, etc.  Unique within a single GraphDef.
-     * Must match the regexp "[A-Za-z0-9.][A-Za-z0-9_&gt;./]*".
-     * </pre>
-     *
-     * <code>string name = 1;</code>
-     */
-    public Builder clearName() {
-      
-      name_ = getDefaultInstance().getName();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * The name given to this operator. Used for naming inputs,
-     * logging, visualization, etc.  Unique within a single GraphDef.
-     * Must match the regexp "[A-Za-z0-9.][A-Za-z0-9_&gt;./]*".
-     * </pre>
-     *
-     * <code>string name = 1;</code>
-     */
-    public Builder setNameBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      name_ = value;
-      onChanged();
-      return this;
-    }
-
-    private java.lang.Object op_ = "";
-    /**
-     * <pre>
-     * The operation name.  There may be custom parameters in attrs.
-     * Op names starting with an underscore are reserved for internal use.
-     * </pre>
-     *
-     * <code>string op = 2;</code>
-     */
-    public java.lang.String getOp() {
-      java.lang.Object ref = op_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        op_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * The operation name.  There may be custom parameters in attrs.
-     * Op names starting with an underscore are reserved for internal use.
-     * </pre>
-     *
-     * <code>string op = 2;</code>
-     */
-    public com.google.protobuf.ByteString
-        getOpBytes() {
-      java.lang.Object ref = op_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        op_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * The operation name.  There may be custom parameters in attrs.
-     * Op names starting with an underscore are reserved for internal use.
-     * </pre>
-     *
-     * <code>string op = 2;</code>
-     */
-    public Builder setOp(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      op_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * The operation name.  There may be custom parameters in attrs.
-     * Op names starting with an underscore are reserved for internal use.
-     * </pre>
-     *
-     * <code>string op = 2;</code>
-     */
-    public Builder clearOp() {
-      
-      op_ = getDefaultInstance().getOp();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * The operation name.  There may be custom parameters in attrs.
-     * Op names starting with an underscore are reserved for internal use.
-     * </pre>
-     *
-     * <code>string op = 2;</code>
-     */
-    public Builder setOpBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      op_ = value;
-      onChanged();
-      return this;
-    }
-
-    private com.google.protobuf.LazyStringList input_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-    private void ensureInputIsMutable() {
-      if (!((bitField0_ & 0x00000001) != 0)) {
-        input_ = new com.google.protobuf.LazyStringArrayList(input_);
-        bitField0_ |= 0x00000001;
-       }
-    }
-    /**
-     * <pre>
-     * Each input is "node:src_output" with "node" being a string name and
-     * "src_output" indicating which output tensor to use from "node". If
-     * "src_output" is 0 the ":0" suffix can be omitted.  Regular inputs
-     * may optionally be followed by control inputs that have the format
-     * "^node".
-     * </pre>
-     *
-     * <code>repeated string input = 3;</code>
-     */
-    public com.google.protobuf.ProtocolStringList
-        getInputList() {
-      return input_.getUnmodifiableView();
-    }
-    /**
-     * <pre>
-     * Each input is "node:src_output" with "node" being a string name and
-     * "src_output" indicating which output tensor to use from "node". If
-     * "src_output" is 0 the ":0" suffix can be omitted.  Regular inputs
-     * may optionally be followed by control inputs that have the format
-     * "^node".
-     * </pre>
-     *
-     * <code>repeated string input = 3;</code>
-     */
-    public int getInputCount() {
-      return input_.size();
-    }
-    /**
-     * <pre>
-     * Each input is "node:src_output" with "node" being a string name and
-     * "src_output" indicating which output tensor to use from "node". If
-     * "src_output" is 0 the ":0" suffix can be omitted.  Regular inputs
-     * may optionally be followed by control inputs that have the format
-     * "^node".
-     * </pre>
-     *
-     * <code>repeated string input = 3;</code>
-     */
-    public java.lang.String getInput(int index) {
-      return input_.get(index);
-    }
-    /**
-     * <pre>
-     * Each input is "node:src_output" with "node" being a string name and
-     * "src_output" indicating which output tensor to use from "node". If
-     * "src_output" is 0 the ":0" suffix can be omitted.  Regular inputs
-     * may optionally be followed by control inputs that have the format
-     * "^node".
-     * </pre>
-     *
-     * <code>repeated string input = 3;</code>
-     */
-    public com.google.protobuf.ByteString
-        getInputBytes(int index) {
-      return input_.getByteString(index);
-    }
-    /**
-     * <pre>
-     * Each input is "node:src_output" with "node" being a string name and
-     * "src_output" indicating which output tensor to use from "node". If
-     * "src_output" is 0 the ":0" suffix can be omitted.  Regular inputs
-     * may optionally be followed by control inputs that have the format
-     * "^node".
-     * </pre>
-     *
-     * <code>repeated string input = 3;</code>
-     */
-    public Builder setInput(
-        int index, java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  ensureInputIsMutable();
-      input_.set(index, value);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Each input is "node:src_output" with "node" being a string name and
-     * "src_output" indicating which output tensor to use from "node". If
-     * "src_output" is 0 the ":0" suffix can be omitted.  Regular inputs
-     * may optionally be followed by control inputs that have the format
-     * "^node".
-     * </pre>
-     *
-     * <code>repeated string input = 3;</code>
-     */
-    public Builder addInput(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  ensureInputIsMutable();
-      input_.add(value);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Each input is "node:src_output" with "node" being a string name and
-     * "src_output" indicating which output tensor to use from "node". If
-     * "src_output" is 0 the ":0" suffix can be omitted.  Regular inputs
-     * may optionally be followed by control inputs that have the format
-     * "^node".
-     * </pre>
-     *
-     * <code>repeated string input = 3;</code>
-     */
-    public Builder addAllInput(
-        java.lang.Iterable<java.lang.String> values) {
-      ensureInputIsMutable();
-      com.google.protobuf.AbstractMessageLite.Builder.addAll(
-          values, input_);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Each input is "node:src_output" with "node" being a string name and
-     * "src_output" indicating which output tensor to use from "node". If
-     * "src_output" is 0 the ":0" suffix can be omitted.  Regular inputs
-     * may optionally be followed by control inputs that have the format
-     * "^node".
-     * </pre>
-     *
-     * <code>repeated string input = 3;</code>
-     */
-    public Builder clearInput() {
-      input_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-      bitField0_ = (bitField0_ & ~0x00000001);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Each input is "node:src_output" with "node" being a string name and
-     * "src_output" indicating which output tensor to use from "node". If
-     * "src_output" is 0 the ":0" suffix can be omitted.  Regular inputs
-     * may optionally be followed by control inputs that have the format
-     * "^node".
-     * </pre>
-     *
-     * <code>repeated string input = 3;</code>
-     */
-    public Builder addInputBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      ensureInputIsMutable();
-      input_.add(value);
-      onChanged();
-      return this;
-    }
-
-    private java.lang.Object device_ = "";
-    /**
-     * <pre>
-     * A (possibly partial) specification for the device on which this
-     * node should be placed.
-     * The expected syntax for this string is as follows:
-     * DEVICE_SPEC ::= PARTIAL_SPEC
-     * PARTIAL_SPEC ::= ("/" CONSTRAINT) *
-     * CONSTRAINT ::= ("job:" JOB_NAME)
-     *              | ("replica:" [1-9][0-9]*)
-     *              | ("task:" [1-9][0-9]*)
-     *              | ("device:" [A-Za-z]* ":" ([1-9][0-9]* | "*") )
-     * Valid values for this string include:
-     * * "/job:worker/replica:0/task:1/device:GPU:3"  (full specification)
-     * * "/job:worker/device:GPU:3"                   (partial specification)
-     * * ""                                    (no specification)
-     * If the constraints do not resolve to a single device (or if this
-     * field is empty or not present), the runtime will attempt to
-     * choose a device automatically.
-     * </pre>
-     *
-     * <code>string device = 4;</code>
-     */
-    public java.lang.String getDevice() {
-      java.lang.Object ref = device_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        device_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * A (possibly partial) specification for the device on which this
-     * node should be placed.
-     * The expected syntax for this string is as follows:
-     * DEVICE_SPEC ::= PARTIAL_SPEC
-     * PARTIAL_SPEC ::= ("/" CONSTRAINT) *
-     * CONSTRAINT ::= ("job:" JOB_NAME)
-     *              | ("replica:" [1-9][0-9]*)
-     *              | ("task:" [1-9][0-9]*)
-     *              | ("device:" [A-Za-z]* ":" ([1-9][0-9]* | "*") )
-     * Valid values for this string include:
-     * * "/job:worker/replica:0/task:1/device:GPU:3"  (full specification)
-     * * "/job:worker/device:GPU:3"                   (partial specification)
-     * * ""                                    (no specification)
-     * If the constraints do not resolve to a single device (or if this
-     * field is empty or not present), the runtime will attempt to
-     * choose a device automatically.
-     * </pre>
-     *
-     * <code>string device = 4;</code>
-     */
-    public com.google.protobuf.ByteString
-        getDeviceBytes() {
-      java.lang.Object ref = device_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        device_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * A (possibly partial) specification for the device on which this
-     * node should be placed.
-     * The expected syntax for this string is as follows:
-     * DEVICE_SPEC ::= PARTIAL_SPEC
-     * PARTIAL_SPEC ::= ("/" CONSTRAINT) *
-     * CONSTRAINT ::= ("job:" JOB_NAME)
-     *              | ("replica:" [1-9][0-9]*)
-     *              | ("task:" [1-9][0-9]*)
-     *              | ("device:" [A-Za-z]* ":" ([1-9][0-9]* | "*") )
-     * Valid values for this string include:
-     * * "/job:worker/replica:0/task:1/device:GPU:3"  (full specification)
-     * * "/job:worker/device:GPU:3"                   (partial specification)
-     * * ""                                    (no specification)
-     * If the constraints do not resolve to a single device (or if this
-     * field is empty or not present), the runtime will attempt to
-     * choose a device automatically.
-     * </pre>
-     *
-     * <code>string device = 4;</code>
-     */
-    public Builder setDevice(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      device_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * A (possibly partial) specification for the device on which this
-     * node should be placed.
-     * The expected syntax for this string is as follows:
-     * DEVICE_SPEC ::= PARTIAL_SPEC
-     * PARTIAL_SPEC ::= ("/" CONSTRAINT) *
-     * CONSTRAINT ::= ("job:" JOB_NAME)
-     *              | ("replica:" [1-9][0-9]*)
-     *              | ("task:" [1-9][0-9]*)
-     *              | ("device:" [A-Za-z]* ":" ([1-9][0-9]* | "*") )
-     * Valid values for this string include:
-     * * "/job:worker/replica:0/task:1/device:GPU:3"  (full specification)
-     * * "/job:worker/device:GPU:3"                   (partial specification)
-     * * ""                                    (no specification)
-     * If the constraints do not resolve to a single device (or if this
-     * field is empty or not present), the runtime will attempt to
-     * choose a device automatically.
-     * </pre>
-     *
-     * <code>string device = 4;</code>
-     */
-    public Builder clearDevice() {
-      
-      device_ = getDefaultInstance().getDevice();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * A (possibly partial) specification for the device on which this
-     * node should be placed.
-     * The expected syntax for this string is as follows:
-     * DEVICE_SPEC ::= PARTIAL_SPEC
-     * PARTIAL_SPEC ::= ("/" CONSTRAINT) *
-     * CONSTRAINT ::= ("job:" JOB_NAME)
-     *              | ("replica:" [1-9][0-9]*)
-     *              | ("task:" [1-9][0-9]*)
-     *              | ("device:" [A-Za-z]* ":" ([1-9][0-9]* | "*") )
-     * Valid values for this string include:
-     * * "/job:worker/replica:0/task:1/device:GPU:3"  (full specification)
-     * * "/job:worker/device:GPU:3"                   (partial specification)
-     * * ""                                    (no specification)
-     * If the constraints do not resolve to a single device (or if this
-     * field is empty or not present), the runtime will attempt to
-     * choose a device automatically.
-     * </pre>
-     *
-     * <code>string device = 4;</code>
-     */
-    public Builder setDeviceBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      device_ = value;
-      onChanged();
-      return this;
-    }
-
-    private com.google.protobuf.MapField<
-        java.lang.String, org.tensorflow.proto.framework.AttrValue> attr_;
-    private com.google.protobuf.MapField<java.lang.String, org.tensorflow.proto.framework.AttrValue>
-    internalGetAttr() {
-      if (attr_ == null) {
-        return com.google.protobuf.MapField.emptyMapField(
-            AttrDefaultEntryHolder.defaultEntry);
-      }
-      return attr_;
-    }
-    private com.google.protobuf.MapField<java.lang.String, org.tensorflow.proto.framework.AttrValue>
-    internalGetMutableAttr() {
-      onChanged();;
-      if (attr_ == null) {
-        attr_ = com.google.protobuf.MapField.newMapField(
-            AttrDefaultEntryHolder.defaultEntry);
-      }
-      if (!attr_.isMutable()) {
-        attr_ = attr_.copy();
-      }
-      return attr_;
-    }
-
-    public int getAttrCount() {
-      return internalGetAttr().getMap().size();
-    }
-    /**
-     * <pre>
-     * Operation-specific graph-construction-time configuration.
-     * Note that this should include all attrs defined in the
-     * corresponding OpDef, including those with a value matching
-     * the default -- this allows the default to change and makes
-     * NodeDefs easier to interpret on their own.  However, if
-     * an attr with a default is not specified in this list, the
-     * default will be used.
-     * The "names" (keys) must match the regexp "[a-z][a-z0-9_]+" (and
-     * one of the names from the corresponding OpDef's attr field).
-     * The values must have a type matching the corresponding OpDef
-     * attr's type field.
-     * TODO(josh11b): Add some examples here showing best practices.
-     * </pre>
-     *
-     * <code>map&lt;string, .tensorflow.AttrValue&gt; attr = 5;</code>
-     */
-
-    public boolean containsAttr(
-        java.lang.String key) {
-      if (key == null) { throw new java.lang.NullPointerException(); }
-      return internalGetAttr().getMap().containsKey(key);
-    }
-    /**
-     * Use {@link #getAttrMap()} instead.
-     */
-    @java.lang.Deprecated
-    public java.util.Map<java.lang.String, org.tensorflow.proto.framework.AttrValue> getAttr() {
-      return getAttrMap();
-    }
-    /**
-     * <pre>
-     * Operation-specific graph-construction-time configuration.
-     * Note that this should include all attrs defined in the
-     * corresponding OpDef, including those with a value matching
-     * the default -- this allows the default to change and makes
-     * NodeDefs easier to interpret on their own.  However, if
-     * an attr with a default is not specified in this list, the
-     * default will be used.
-     * The "names" (keys) must match the regexp "[a-z][a-z0-9_]+" (and
-     * one of the names from the corresponding OpDef's attr field).
-     * The values must have a type matching the corresponding OpDef
-     * attr's type field.
-     * TODO(josh11b): Add some examples here showing best practices.
-     * </pre>
-     *
-     * <code>map&lt;string, .tensorflow.AttrValue&gt; attr = 5;</code>
-     */
-
-    public java.util.Map<java.lang.String, org.tensorflow.proto.framework.AttrValue> getAttrMap() {
-      return internalGetAttr().getMap();
-    }
-    /**
-     * <pre>
-     * Operation-specific graph-construction-time configuration.
-     * Note that this should include all attrs defined in the
-     * corresponding OpDef, including those with a value matching
-     * the default -- this allows the default to change and makes
-     * NodeDefs easier to interpret on their own.  However, if
-     * an attr with a default is not specified in this list, the
-     * default will be used.
-     * The "names" (keys) must match the regexp "[a-z][a-z0-9_]+" (and
-     * one of the names from the corresponding OpDef's attr field).
-     * The values must have a type matching the corresponding OpDef
-     * attr's type field.
-     * TODO(josh11b): Add some examples here showing best practices.
-     * </pre>
-     *
-     * <code>map&lt;string, .tensorflow.AttrValue&gt; attr = 5;</code>
-     */
-
-    public org.tensorflow.proto.framework.AttrValue getAttrOrDefault(
-        java.lang.String key,
-        org.tensorflow.proto.framework.AttrValue defaultValue) {
-      if (key == null) { throw new java.lang.NullPointerException(); }
-      java.util.Map<java.lang.String, org.tensorflow.proto.framework.AttrValue> map =
-          internalGetAttr().getMap();
-      return map.containsKey(key) ? map.get(key) : defaultValue;
-    }
-    /**
-     * <pre>
-     * Operation-specific graph-construction-time configuration.
-     * Note that this should include all attrs defined in the
-     * corresponding OpDef, including those with a value matching
-     * the default -- this allows the default to change and makes
-     * NodeDefs easier to interpret on their own.  However, if
-     * an attr with a default is not specified in this list, the
-     * default will be used.
-     * The "names" (keys) must match the regexp "[a-z][a-z0-9_]+" (and
-     * one of the names from the corresponding OpDef's attr field).
-     * The values must have a type matching the corresponding OpDef
-     * attr's type field.
-     * TODO(josh11b): Add some examples here showing best practices.
-     * </pre>
-     *
-     * <code>map&lt;string, .tensorflow.AttrValue&gt; attr = 5;</code>
-     */
-
-    public org.tensorflow.proto.framework.AttrValue getAttrOrThrow(
-        java.lang.String key) {
-      if (key == null) { throw new java.lang.NullPointerException(); }
-      java.util.Map<java.lang.String, org.tensorflow.proto.framework.AttrValue> map =
-          internalGetAttr().getMap();
-      if (!map.containsKey(key)) {
-        throw new java.lang.IllegalArgumentException();
-      }
-      return map.get(key);
-    }
-
-    public Builder clearAttr() {
-      internalGetMutableAttr().getMutableMap()
-          .clear();
-      return this;
-    }
-    /**
-     * <pre>
-     * Operation-specific graph-construction-time configuration.
-     * Note that this should include all attrs defined in the
-     * corresponding OpDef, including those with a value matching
-     * the default -- this allows the default to change and makes
-     * NodeDefs easier to interpret on their own.  However, if
-     * an attr with a default is not specified in this list, the
-     * default will be used.
-     * The "names" (keys) must match the regexp "[a-z][a-z0-9_]+" (and
-     * one of the names from the corresponding OpDef's attr field).
-     * The values must have a type matching the corresponding OpDef
-     * attr's type field.
-     * TODO(josh11b): Add some examples here showing best practices.
-     * </pre>
-     *
-     * <code>map&lt;string, .tensorflow.AttrValue&gt; attr = 5;</code>
-     */
-
-    public Builder removeAttr(
-        java.lang.String key) {
-      if (key == null) { throw new java.lang.NullPointerException(); }
-      internalGetMutableAttr().getMutableMap()
-          .remove(key);
-      return this;
-    }
-    /**
-     * Use alternate mutation accessors instead.
-     */
-    @java.lang.Deprecated
-    public java.util.Map<java.lang.String, org.tensorflow.proto.framework.AttrValue>
-    getMutableAttr() {
-      return internalGetMutableAttr().getMutableMap();
-    }
-    /**
-     * <pre>
-     * Operation-specific graph-construction-time configuration.
-     * Note that this should include all attrs defined in the
-     * corresponding OpDef, including those with a value matching
-     * the default -- this allows the default to change and makes
-     * NodeDefs easier to interpret on their own.  However, if
-     * an attr with a default is not specified in this list, the
-     * default will be used.
-     * The "names" (keys) must match the regexp "[a-z][a-z0-9_]+" (and
-     * one of the names from the corresponding OpDef's attr field).
-     * The values must have a type matching the corresponding OpDef
-     * attr's type field.
-     * TODO(josh11b): Add some examples here showing best practices.
-     * </pre>
-     *
-     * <code>map&lt;string, .tensorflow.AttrValue&gt; attr = 5;</code>
-     */
-    public Builder putAttr(
-        java.lang.String key,
-        org.tensorflow.proto.framework.AttrValue value) {
-      if (key == null) { throw new java.lang.NullPointerException(); }
-      if (value == null) { throw new java.lang.NullPointerException(); }
-      internalGetMutableAttr().getMutableMap()
-          .put(key, value);
-      return this;
-    }
-    /**
-     * <pre>
-     * Operation-specific graph-construction-time configuration.
-     * Note that this should include all attrs defined in the
-     * corresponding OpDef, including those with a value matching
-     * the default -- this allows the default to change and makes
-     * NodeDefs easier to interpret on their own.  However, if
-     * an attr with a default is not specified in this list, the
-     * default will be used.
-     * The "names" (keys) must match the regexp "[a-z][a-z0-9_]+" (and
-     * one of the names from the corresponding OpDef's attr field).
-     * The values must have a type matching the corresponding OpDef
-     * attr's type field.
-     * TODO(josh11b): Add some examples here showing best practices.
-     * </pre>
-     *
-     * <code>map&lt;string, .tensorflow.AttrValue&gt; attr = 5;</code>
-     */
-
-    public Builder putAllAttr(
-        java.util.Map<java.lang.String, org.tensorflow.proto.framework.AttrValue> values) {
-      internalGetMutableAttr().getMutableMap()
-          .putAll(values);
-      return this;
-    }
-
-    private org.tensorflow.proto.framework.NodeDef.ExperimentalDebugInfo experimentalDebugInfo_;
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.NodeDef.ExperimentalDebugInfo, org.tensorflow.proto.framework.NodeDef.ExperimentalDebugInfo.Builder, org.tensorflow.proto.framework.NodeDef.ExperimentalDebugInfoOrBuilder> experimentalDebugInfoBuilder_;
-    /**
-     * <pre>
-     * This stores debug information associated with the node.
-     * </pre>
-     *
-     * <code>.tensorflow.NodeDef.ExperimentalDebugInfo experimental_debug_info = 6;</code>
-     */
-    public boolean hasExperimentalDebugInfo() {
-      return experimentalDebugInfoBuilder_ != null || experimentalDebugInfo_ != null;
-    }
-    /**
-     * <pre>
-     * This stores debug information associated with the node.
-     * </pre>
-     *
-     * <code>.tensorflow.NodeDef.ExperimentalDebugInfo experimental_debug_info = 6;</code>
-     */
-    public org.tensorflow.proto.framework.NodeDef.ExperimentalDebugInfo getExperimentalDebugInfo() {
-      if (experimentalDebugInfoBuilder_ == null) {
-        return experimentalDebugInfo_ == null ? org.tensorflow.proto.framework.NodeDef.ExperimentalDebugInfo.getDefaultInstance() : experimentalDebugInfo_;
-      } else {
-        return experimentalDebugInfoBuilder_.getMessage();
-      }
-    }
-    /**
-     * <pre>
-     * This stores debug information associated with the node.
-     * </pre>
-     *
-     * <code>.tensorflow.NodeDef.ExperimentalDebugInfo experimental_debug_info = 6;</code>
-     */
-    public Builder setExperimentalDebugInfo(org.tensorflow.proto.framework.NodeDef.ExperimentalDebugInfo value) {
-      if (experimentalDebugInfoBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        experimentalDebugInfo_ = value;
-        onChanged();
-      } else {
-        experimentalDebugInfoBuilder_.setMessage(value);
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * This stores debug information associated with the node.
-     * </pre>
-     *
-     * <code>.tensorflow.NodeDef.ExperimentalDebugInfo experimental_debug_info = 6;</code>
-     */
-    public Builder setExperimentalDebugInfo(
-        org.tensorflow.proto.framework.NodeDef.ExperimentalDebugInfo.Builder builderForValue) {
-      if (experimentalDebugInfoBuilder_ == null) {
-        experimentalDebugInfo_ = builderForValue.build();
-        onChanged();
-      } else {
-        experimentalDebugInfoBuilder_.setMessage(builderForValue.build());
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * This stores debug information associated with the node.
-     * </pre>
-     *
-     * <code>.tensorflow.NodeDef.ExperimentalDebugInfo experimental_debug_info = 6;</code>
-     */
-    public Builder mergeExperimentalDebugInfo(org.tensorflow.proto.framework.NodeDef.ExperimentalDebugInfo value) {
-      if (experimentalDebugInfoBuilder_ == null) {
-        if (experimentalDebugInfo_ != null) {
-          experimentalDebugInfo_ =
-            org.tensorflow.proto.framework.NodeDef.ExperimentalDebugInfo.newBuilder(experimentalDebugInfo_).mergeFrom(value).buildPartial();
-        } else {
-          experimentalDebugInfo_ = value;
-        }
-        onChanged();
-      } else {
-        experimentalDebugInfoBuilder_.mergeFrom(value);
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * This stores debug information associated with the node.
-     * </pre>
-     *
-     * <code>.tensorflow.NodeDef.ExperimentalDebugInfo experimental_debug_info = 6;</code>
-     */
-    public Builder clearExperimentalDebugInfo() {
-      if (experimentalDebugInfoBuilder_ == null) {
-        experimentalDebugInfo_ = null;
-        onChanged();
-      } else {
-        experimentalDebugInfo_ = null;
-        experimentalDebugInfoBuilder_ = null;
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * This stores debug information associated with the node.
-     * </pre>
-     *
-     * <code>.tensorflow.NodeDef.ExperimentalDebugInfo experimental_debug_info = 6;</code>
-     */
-    public org.tensorflow.proto.framework.NodeDef.ExperimentalDebugInfo.Builder getExperimentalDebugInfoBuilder() {
-      
-      onChanged();
-      return getExperimentalDebugInfoFieldBuilder().getBuilder();
-    }
-    /**
-     * <pre>
-     * This stores debug information associated with the node.
-     * </pre>
-     *
-     * <code>.tensorflow.NodeDef.ExperimentalDebugInfo experimental_debug_info = 6;</code>
-     */
-    public org.tensorflow.proto.framework.NodeDef.ExperimentalDebugInfoOrBuilder getExperimentalDebugInfoOrBuilder() {
-      if (experimentalDebugInfoBuilder_ != null) {
-        return experimentalDebugInfoBuilder_.getMessageOrBuilder();
-      } else {
-        return experimentalDebugInfo_ == null ?
-            org.tensorflow.proto.framework.NodeDef.ExperimentalDebugInfo.getDefaultInstance() : experimentalDebugInfo_;
-      }
-    }
-    /**
-     * <pre>
-     * This stores debug information associated with the node.
-     * </pre>
-     *
-     * <code>.tensorflow.NodeDef.ExperimentalDebugInfo experimental_debug_info = 6;</code>
-     */
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.NodeDef.ExperimentalDebugInfo, org.tensorflow.proto.framework.NodeDef.ExperimentalDebugInfo.Builder, org.tensorflow.proto.framework.NodeDef.ExperimentalDebugInfoOrBuilder> 
-        getExperimentalDebugInfoFieldBuilder() {
-      if (experimentalDebugInfoBuilder_ == null) {
-        experimentalDebugInfoBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-            org.tensorflow.proto.framework.NodeDef.ExperimentalDebugInfo, org.tensorflow.proto.framework.NodeDef.ExperimentalDebugInfo.Builder, org.tensorflow.proto.framework.NodeDef.ExperimentalDebugInfoOrBuilder>(
-                getExperimentalDebugInfo(),
-                getParentForChildren(),
-                isClean());
-        experimentalDebugInfo_ = null;
-      }
-      return experimentalDebugInfoBuilder_;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.NodeDef)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.NodeDef)
-  private static final org.tensorflow.proto.framework.NodeDef DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.NodeDef();
-  }
-
-  public static org.tensorflow.proto.framework.NodeDef getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<NodeDef>
-      PARSER = new com.google.protobuf.AbstractParser<NodeDef>() {
-    @java.lang.Override
-    public NodeDef parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new NodeDef(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<NodeDef> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<NodeDef> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.framework.NodeDef getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NodeDefOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NodeDefOrBuilder.java
deleted file mode 100644
index 43971913d97..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NodeDefOrBuilder.java
+++ /dev/null
@@ -1,284 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/framework/node_def.proto
-
-package org.tensorflow.proto.framework;
-
-public interface NodeDefOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.NodeDef)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <pre>
-   * The name given to this operator. Used for naming inputs,
-   * logging, visualization, etc.  Unique within a single GraphDef.
-   * Must match the regexp "[A-Za-z0-9.][A-Za-z0-9_&gt;./]*".
-   * </pre>
-   *
-   * <code>string name = 1;</code>
-   */
-  java.lang.String getName();
-  /**
-   * <pre>
-   * The name given to this operator. Used for naming inputs,
-   * logging, visualization, etc.  Unique within a single GraphDef.
-   * Must match the regexp "[A-Za-z0-9.][A-Za-z0-9_&gt;./]*".
-   * </pre>
-   *
-   * <code>string name = 1;</code>
-   */
-  com.google.protobuf.ByteString
-      getNameBytes();
-
-  /**
-   * <pre>
-   * The operation name.  There may be custom parameters in attrs.
-   * Op names starting with an underscore are reserved for internal use.
-   * </pre>
-   *
-   * <code>string op = 2;</code>
-   */
-  java.lang.String getOp();
-  /**
-   * <pre>
-   * The operation name.  There may be custom parameters in attrs.
-   * Op names starting with an underscore are reserved for internal use.
-   * </pre>
-   *
-   * <code>string op = 2;</code>
-   */
-  com.google.protobuf.ByteString
-      getOpBytes();
-
-  /**
-   * <pre>
-   * Each input is "node:src_output" with "node" being a string name and
-   * "src_output" indicating which output tensor to use from "node". If
-   * "src_output" is 0 the ":0" suffix can be omitted.  Regular inputs
-   * may optionally be followed by control inputs that have the format
-   * "^node".
-   * </pre>
-   *
-   * <code>repeated string input = 3;</code>
-   */
-  java.util.List<java.lang.String>
-      getInputList();
-  /**
-   * <pre>
-   * Each input is "node:src_output" with "node" being a string name and
-   * "src_output" indicating which output tensor to use from "node". If
-   * "src_output" is 0 the ":0" suffix can be omitted.  Regular inputs
-   * may optionally be followed by control inputs that have the format
-   * "^node".
-   * </pre>
-   *
-   * <code>repeated string input = 3;</code>
-   */
-  int getInputCount();
-  /**
-   * <pre>
-   * Each input is "node:src_output" with "node" being a string name and
-   * "src_output" indicating which output tensor to use from "node". If
-   * "src_output" is 0 the ":0" suffix can be omitted.  Regular inputs
-   * may optionally be followed by control inputs that have the format
-   * "^node".
-   * </pre>
-   *
-   * <code>repeated string input = 3;</code>
-   */
-  java.lang.String getInput(int index);
-  /**
-   * <pre>
-   * Each input is "node:src_output" with "node" being a string name and
-   * "src_output" indicating which output tensor to use from "node". If
-   * "src_output" is 0 the ":0" suffix can be omitted.  Regular inputs
-   * may optionally be followed by control inputs that have the format
-   * "^node".
-   * </pre>
-   *
-   * <code>repeated string input = 3;</code>
-   */
-  com.google.protobuf.ByteString
-      getInputBytes(int index);
-
-  /**
-   * <pre>
-   * A (possibly partial) specification for the device on which this
-   * node should be placed.
-   * The expected syntax for this string is as follows:
-   * DEVICE_SPEC ::= PARTIAL_SPEC
-   * PARTIAL_SPEC ::= ("/" CONSTRAINT) *
-   * CONSTRAINT ::= ("job:" JOB_NAME)
-   *              | ("replica:" [1-9][0-9]*)
-   *              | ("task:" [1-9][0-9]*)
-   *              | ("device:" [A-Za-z]* ":" ([1-9][0-9]* | "*") )
-   * Valid values for this string include:
-   * * "/job:worker/replica:0/task:1/device:GPU:3"  (full specification)
-   * * "/job:worker/device:GPU:3"                   (partial specification)
-   * * ""                                    (no specification)
-   * If the constraints do not resolve to a single device (or if this
-   * field is empty or not present), the runtime will attempt to
-   * choose a device automatically.
-   * </pre>
-   *
-   * <code>string device = 4;</code>
-   */
-  java.lang.String getDevice();
-  /**
-   * <pre>
-   * A (possibly partial) specification for the device on which this
-   * node should be placed.
-   * The expected syntax for this string is as follows:
-   * DEVICE_SPEC ::= PARTIAL_SPEC
-   * PARTIAL_SPEC ::= ("/" CONSTRAINT) *
-   * CONSTRAINT ::= ("job:" JOB_NAME)
-   *              | ("replica:" [1-9][0-9]*)
-   *              | ("task:" [1-9][0-9]*)
-   *              | ("device:" [A-Za-z]* ":" ([1-9][0-9]* | "*") )
-   * Valid values for this string include:
-   * * "/job:worker/replica:0/task:1/device:GPU:3"  (full specification)
-   * * "/job:worker/device:GPU:3"                   (partial specification)
-   * * ""                                    (no specification)
-   * If the constraints do not resolve to a single device (or if this
-   * field is empty or not present), the runtime will attempt to
-   * choose a device automatically.
-   * </pre>
-   *
-   * <code>string device = 4;</code>
-   */
-  com.google.protobuf.ByteString
-      getDeviceBytes();
-
-  /**
-   * <pre>
-   * Operation-specific graph-construction-time configuration.
-   * Note that this should include all attrs defined in the
-   * corresponding OpDef, including those with a value matching
-   * the default -- this allows the default to change and makes
-   * NodeDefs easier to interpret on their own.  However, if
-   * an attr with a default is not specified in this list, the
-   * default will be used.
-   * The "names" (keys) must match the regexp "[a-z][a-z0-9_]+" (and
-   * one of the names from the corresponding OpDef's attr field).
-   * The values must have a type matching the corresponding OpDef
-   * attr's type field.
-   * TODO(josh11b): Add some examples here showing best practices.
-   * </pre>
-   *
-   * <code>map&lt;string, .tensorflow.AttrValue&gt; attr = 5;</code>
-   */
-  int getAttrCount();
-  /**
-   * <pre>
-   * Operation-specific graph-construction-time configuration.
-   * Note that this should include all attrs defined in the
-   * corresponding OpDef, including those with a value matching
-   * the default -- this allows the default to change and makes
-   * NodeDefs easier to interpret on their own.  However, if
-   * an attr with a default is not specified in this list, the
-   * default will be used.
-   * The "names" (keys) must match the regexp "[a-z][a-z0-9_]+" (and
-   * one of the names from the corresponding OpDef's attr field).
-   * The values must have a type matching the corresponding OpDef
-   * attr's type field.
-   * TODO(josh11b): Add some examples here showing best practices.
-   * </pre>
-   *
-   * <code>map&lt;string, .tensorflow.AttrValue&gt; attr = 5;</code>
-   */
-  boolean containsAttr(
-      java.lang.String key);
-  /**
-   * Use {@link #getAttrMap()} instead.
-   */
-  @java.lang.Deprecated
-  java.util.Map<java.lang.String, org.tensorflow.proto.framework.AttrValue>
-  getAttr();
-  /**
-   * <pre>
-   * Operation-specific graph-construction-time configuration.
-   * Note that this should include all attrs defined in the
-   * corresponding OpDef, including those with a value matching
-   * the default -- this allows the default to change and makes
-   * NodeDefs easier to interpret on their own.  However, if
-   * an attr with a default is not specified in this list, the
-   * default will be used.
-   * The "names" (keys) must match the regexp "[a-z][a-z0-9_]+" (and
-   * one of the names from the corresponding OpDef's attr field).
-   * The values must have a type matching the corresponding OpDef
-   * attr's type field.
-   * TODO(josh11b): Add some examples here showing best practices.
-   * </pre>
-   *
-   * <code>map&lt;string, .tensorflow.AttrValue&gt; attr = 5;</code>
-   */
-  java.util.Map<java.lang.String, org.tensorflow.proto.framework.AttrValue>
-  getAttrMap();
-  /**
-   * <pre>
-   * Operation-specific graph-construction-time configuration.
-   * Note that this should include all attrs defined in the
-   * corresponding OpDef, including those with a value matching
-   * the default -- this allows the default to change and makes
-   * NodeDefs easier to interpret on their own.  However, if
-   * an attr with a default is not specified in this list, the
-   * default will be used.
-   * The "names" (keys) must match the regexp "[a-z][a-z0-9_]+" (and
-   * one of the names from the corresponding OpDef's attr field).
-   * The values must have a type matching the corresponding OpDef
-   * attr's type field.
-   * TODO(josh11b): Add some examples here showing best practices.
-   * </pre>
-   *
-   * <code>map&lt;string, .tensorflow.AttrValue&gt; attr = 5;</code>
-   */
-
-  org.tensorflow.proto.framework.AttrValue getAttrOrDefault(
-      java.lang.String key,
-      org.tensorflow.proto.framework.AttrValue defaultValue);
-  /**
-   * <pre>
-   * Operation-specific graph-construction-time configuration.
-   * Note that this should include all attrs defined in the
-   * corresponding OpDef, including those with a value matching
-   * the default -- this allows the default to change and makes
-   * NodeDefs easier to interpret on their own.  However, if
-   * an attr with a default is not specified in this list, the
-   * default will be used.
-   * The "names" (keys) must match the regexp "[a-z][a-z0-9_]+" (and
-   * one of the names from the corresponding OpDef's attr field).
-   * The values must have a type matching the corresponding OpDef
-   * attr's type field.
-   * TODO(josh11b): Add some examples here showing best practices.
-   * </pre>
-   *
-   * <code>map&lt;string, .tensorflow.AttrValue&gt; attr = 5;</code>
-   */
-
-  org.tensorflow.proto.framework.AttrValue getAttrOrThrow(
-      java.lang.String key);
-
-  /**
-   * <pre>
-   * This stores debug information associated with the node.
-   * </pre>
-   *
-   * <code>.tensorflow.NodeDef.ExperimentalDebugInfo experimental_debug_info = 6;</code>
-   */
-  boolean hasExperimentalDebugInfo();
-  /**
-   * <pre>
-   * This stores debug information associated with the node.
-   * </pre>
-   *
-   * <code>.tensorflow.NodeDef.ExperimentalDebugInfo experimental_debug_info = 6;</code>
-   */
-  org.tensorflow.proto.framework.NodeDef.ExperimentalDebugInfo getExperimentalDebugInfo();
-  /**
-   * <pre>
-   * This stores debug information associated with the node.
-   * </pre>
-   *
-   * <code>.tensorflow.NodeDef.ExperimentalDebugInfo experimental_debug_info = 6;</code>
-   */
-  org.tensorflow.proto.framework.NodeDef.ExperimentalDebugInfoOrBuilder getExperimentalDebugInfoOrBuilder();
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NodeExecStats.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NodeExecStats.java
deleted file mode 100644
index 93e1e3df32c..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NodeExecStats.java
+++ /dev/null
@@ -1,2580 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/framework/step_stats.proto
-
-package org.tensorflow.proto.framework;
-
-/**
- * <pre>
- * Time/size stats recorded for a single execution of a graph node.
- * </pre>
- *
- * Protobuf type {@code tensorflow.NodeExecStats}
- */
-public  final class NodeExecStats extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.NodeExecStats)
-    NodeExecStatsOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use NodeExecStats.newBuilder() to construct.
-  private NodeExecStats(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private NodeExecStats() {
-    nodeName_ = "";
-    memory_ = java.util.Collections.emptyList();
-    output_ = java.util.Collections.emptyList();
-    timelineLabel_ = "";
-    referencedTensor_ = java.util.Collections.emptyList();
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new NodeExecStats();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private NodeExecStats(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    int mutable_bitField0_ = 0;
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 10: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            nodeName_ = s;
-            break;
-          }
-          case 16: {
-
-            allStartMicros_ = input.readInt64();
-            break;
-          }
-          case 24: {
-
-            opStartRelMicros_ = input.readInt64();
-            break;
-          }
-          case 32: {
-
-            opEndRelMicros_ = input.readInt64();
-            break;
-          }
-          case 40: {
-
-            allEndRelMicros_ = input.readInt64();
-            break;
-          }
-          case 50: {
-            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
-              memory_ = new java.util.ArrayList<org.tensorflow.proto.framework.AllocatorMemoryUsed>();
-              mutable_bitField0_ |= 0x00000001;
-            }
-            memory_.add(
-                input.readMessage(org.tensorflow.proto.framework.AllocatorMemoryUsed.parser(), extensionRegistry));
-            break;
-          }
-          case 58: {
-            if (!((mutable_bitField0_ & 0x00000002) != 0)) {
-              output_ = new java.util.ArrayList<org.tensorflow.proto.framework.NodeOutput>();
-              mutable_bitField0_ |= 0x00000002;
-            }
-            output_.add(
-                input.readMessage(org.tensorflow.proto.framework.NodeOutput.parser(), extensionRegistry));
-            break;
-          }
-          case 66: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            timelineLabel_ = s;
-            break;
-          }
-          case 72: {
-
-            scheduledMicros_ = input.readInt64();
-            break;
-          }
-          case 80: {
-
-            threadId_ = input.readUInt32();
-            break;
-          }
-          case 90: {
-            if (!((mutable_bitField0_ & 0x00000004) != 0)) {
-              referencedTensor_ = new java.util.ArrayList<org.tensorflow.proto.framework.AllocationDescription>();
-              mutable_bitField0_ |= 0x00000004;
-            }
-            referencedTensor_.add(
-                input.readMessage(org.tensorflow.proto.framework.AllocationDescription.parser(), extensionRegistry));
-            break;
-          }
-          case 98: {
-            org.tensorflow.proto.framework.MemoryStats.Builder subBuilder = null;
-            if (memoryStats_ != null) {
-              subBuilder = memoryStats_.toBuilder();
-            }
-            memoryStats_ = input.readMessage(org.tensorflow.proto.framework.MemoryStats.parser(), extensionRegistry);
-            if (subBuilder != null) {
-              subBuilder.mergeFrom(memoryStats_);
-              memoryStats_ = subBuilder.buildPartial();
-            }
-
-            break;
-          }
-          case 104: {
-
-            allStartNanos_ = input.readInt64();
-            break;
-          }
-          case 112: {
-
-            opStartRelNanos_ = input.readInt64();
-            break;
-          }
-          case 120: {
-
-            opEndRelNanos_ = input.readInt64();
-            break;
-          }
-          case 128: {
-
-            allEndRelNanos_ = input.readInt64();
-            break;
-          }
-          case 136: {
-
-            scheduledNanos_ = input.readInt64();
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      if (((mutable_bitField0_ & 0x00000001) != 0)) {
-        memory_ = java.util.Collections.unmodifiableList(memory_);
-      }
-      if (((mutable_bitField0_ & 0x00000002) != 0)) {
-        output_ = java.util.Collections.unmodifiableList(output_);
-      }
-      if (((mutable_bitField0_ & 0x00000004) != 0)) {
-        referencedTensor_ = java.util.Collections.unmodifiableList(referencedTensor_);
-      }
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.framework.StepStatsProtos.internal_static_tensorflow_NodeExecStats_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.framework.StepStatsProtos.internal_static_tensorflow_NodeExecStats_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.framework.NodeExecStats.class, org.tensorflow.proto.framework.NodeExecStats.Builder.class);
-  }
-
-  public static final int NODE_NAME_FIELD_NUMBER = 1;
-  private volatile java.lang.Object nodeName_;
-  /**
-   * <pre>
-   * TODO(tucker): Use some more compact form of node identity than
-   * the full string name.  Either all processes should agree on a
-   * global id (cost_id?) for each node, or we should use a hash of
-   * the name.
-   * </pre>
-   *
-   * <code>string node_name = 1;</code>
-   */
-  public java.lang.String getNodeName() {
-    java.lang.Object ref = nodeName_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      nodeName_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * TODO(tucker): Use some more compact form of node identity than
-   * the full string name.  Either all processes should agree on a
-   * global id (cost_id?) for each node, or we should use a hash of
-   * the name.
-   * </pre>
-   *
-   * <code>string node_name = 1;</code>
-   */
-  public com.google.protobuf.ByteString
-      getNodeNameBytes() {
-    java.lang.Object ref = nodeName_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      nodeName_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int ALL_START_MICROS_FIELD_NUMBER = 2;
-  private long allStartMicros_;
-  /**
-   * <code>int64 all_start_micros = 2;</code>
-   */
-  public long getAllStartMicros() {
-    return allStartMicros_;
-  }
-
-  public static final int OP_START_REL_MICROS_FIELD_NUMBER = 3;
-  private long opStartRelMicros_;
-  /**
-   * <code>int64 op_start_rel_micros = 3;</code>
-   */
-  public long getOpStartRelMicros() {
-    return opStartRelMicros_;
-  }
-
-  public static final int OP_END_REL_MICROS_FIELD_NUMBER = 4;
-  private long opEndRelMicros_;
-  /**
-   * <code>int64 op_end_rel_micros = 4;</code>
-   */
-  public long getOpEndRelMicros() {
-    return opEndRelMicros_;
-  }
-
-  public static final int ALL_END_REL_MICROS_FIELD_NUMBER = 5;
-  private long allEndRelMicros_;
-  /**
-   * <code>int64 all_end_rel_micros = 5;</code>
-   */
-  public long getAllEndRelMicros() {
-    return allEndRelMicros_;
-  }
-
-  public static final int MEMORY_FIELD_NUMBER = 6;
-  private java.util.List<org.tensorflow.proto.framework.AllocatorMemoryUsed> memory_;
-  /**
-   * <code>repeated .tensorflow.AllocatorMemoryUsed memory = 6;</code>
-   */
-  public java.util.List<org.tensorflow.proto.framework.AllocatorMemoryUsed> getMemoryList() {
-    return memory_;
-  }
-  /**
-   * <code>repeated .tensorflow.AllocatorMemoryUsed memory = 6;</code>
-   */
-  public java.util.List<? extends org.tensorflow.proto.framework.AllocatorMemoryUsedOrBuilder> 
-      getMemoryOrBuilderList() {
-    return memory_;
-  }
-  /**
-   * <code>repeated .tensorflow.AllocatorMemoryUsed memory = 6;</code>
-   */
-  public int getMemoryCount() {
-    return memory_.size();
-  }
-  /**
-   * <code>repeated .tensorflow.AllocatorMemoryUsed memory = 6;</code>
-   */
-  public org.tensorflow.proto.framework.AllocatorMemoryUsed getMemory(int index) {
-    return memory_.get(index);
-  }
-  /**
-   * <code>repeated .tensorflow.AllocatorMemoryUsed memory = 6;</code>
-   */
-  public org.tensorflow.proto.framework.AllocatorMemoryUsedOrBuilder getMemoryOrBuilder(
-      int index) {
-    return memory_.get(index);
-  }
-
-  public static final int OUTPUT_FIELD_NUMBER = 7;
-  private java.util.List<org.tensorflow.proto.framework.NodeOutput> output_;
-  /**
-   * <code>repeated .tensorflow.NodeOutput output = 7;</code>
-   */
-  public java.util.List<org.tensorflow.proto.framework.NodeOutput> getOutputList() {
-    return output_;
-  }
-  /**
-   * <code>repeated .tensorflow.NodeOutput output = 7;</code>
-   */
-  public java.util.List<? extends org.tensorflow.proto.framework.NodeOutputOrBuilder> 
-      getOutputOrBuilderList() {
-    return output_;
-  }
-  /**
-   * <code>repeated .tensorflow.NodeOutput output = 7;</code>
-   */
-  public int getOutputCount() {
-    return output_.size();
-  }
-  /**
-   * <code>repeated .tensorflow.NodeOutput output = 7;</code>
-   */
-  public org.tensorflow.proto.framework.NodeOutput getOutput(int index) {
-    return output_.get(index);
-  }
-  /**
-   * <code>repeated .tensorflow.NodeOutput output = 7;</code>
-   */
-  public org.tensorflow.proto.framework.NodeOutputOrBuilder getOutputOrBuilder(
-      int index) {
-    return output_.get(index);
-  }
-
-  public static final int TIMELINE_LABEL_FIELD_NUMBER = 8;
-  private volatile java.lang.Object timelineLabel_;
-  /**
-   * <code>string timeline_label = 8;</code>
-   */
-  public java.lang.String getTimelineLabel() {
-    java.lang.Object ref = timelineLabel_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      timelineLabel_ = s;
-      return s;
-    }
-  }
-  /**
-   * <code>string timeline_label = 8;</code>
-   */
-  public com.google.protobuf.ByteString
-      getTimelineLabelBytes() {
-    java.lang.Object ref = timelineLabel_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      timelineLabel_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int SCHEDULED_MICROS_FIELD_NUMBER = 9;
-  private long scheduledMicros_;
-  /**
-   * <code>int64 scheduled_micros = 9;</code>
-   */
-  public long getScheduledMicros() {
-    return scheduledMicros_;
-  }
-
-  public static final int THREAD_ID_FIELD_NUMBER = 10;
-  private int threadId_;
-  /**
-   * <code>uint32 thread_id = 10;</code>
-   */
-  public int getThreadId() {
-    return threadId_;
-  }
-
-  public static final int REFERENCED_TENSOR_FIELD_NUMBER = 11;
-  private java.util.List<org.tensorflow.proto.framework.AllocationDescription> referencedTensor_;
-  /**
-   * <code>repeated .tensorflow.AllocationDescription referenced_tensor = 11;</code>
-   */
-  public java.util.List<org.tensorflow.proto.framework.AllocationDescription> getReferencedTensorList() {
-    return referencedTensor_;
-  }
-  /**
-   * <code>repeated .tensorflow.AllocationDescription referenced_tensor = 11;</code>
-   */
-  public java.util.List<? extends org.tensorflow.proto.framework.AllocationDescriptionOrBuilder> 
-      getReferencedTensorOrBuilderList() {
-    return referencedTensor_;
-  }
-  /**
-   * <code>repeated .tensorflow.AllocationDescription referenced_tensor = 11;</code>
-   */
-  public int getReferencedTensorCount() {
-    return referencedTensor_.size();
-  }
-  /**
-   * <code>repeated .tensorflow.AllocationDescription referenced_tensor = 11;</code>
-   */
-  public org.tensorflow.proto.framework.AllocationDescription getReferencedTensor(int index) {
-    return referencedTensor_.get(index);
-  }
-  /**
-   * <code>repeated .tensorflow.AllocationDescription referenced_tensor = 11;</code>
-   */
-  public org.tensorflow.proto.framework.AllocationDescriptionOrBuilder getReferencedTensorOrBuilder(
-      int index) {
-    return referencedTensor_.get(index);
-  }
-
-  public static final int MEMORY_STATS_FIELD_NUMBER = 12;
-  private org.tensorflow.proto.framework.MemoryStats memoryStats_;
-  /**
-   * <code>.tensorflow.MemoryStats memory_stats = 12;</code>
-   */
-  public boolean hasMemoryStats() {
-    return memoryStats_ != null;
-  }
-  /**
-   * <code>.tensorflow.MemoryStats memory_stats = 12;</code>
-   */
-  public org.tensorflow.proto.framework.MemoryStats getMemoryStats() {
-    return memoryStats_ == null ? org.tensorflow.proto.framework.MemoryStats.getDefaultInstance() : memoryStats_;
-  }
-  /**
-   * <code>.tensorflow.MemoryStats memory_stats = 12;</code>
-   */
-  public org.tensorflow.proto.framework.MemoryStatsOrBuilder getMemoryStatsOrBuilder() {
-    return getMemoryStats();
-  }
-
-  public static final int ALL_START_NANOS_FIELD_NUMBER = 13;
-  private long allStartNanos_;
-  /**
-   * <code>int64 all_start_nanos = 13;</code>
-   */
-  public long getAllStartNanos() {
-    return allStartNanos_;
-  }
-
-  public static final int OP_START_REL_NANOS_FIELD_NUMBER = 14;
-  private long opStartRelNanos_;
-  /**
-   * <code>int64 op_start_rel_nanos = 14;</code>
-   */
-  public long getOpStartRelNanos() {
-    return opStartRelNanos_;
-  }
-
-  public static final int OP_END_REL_NANOS_FIELD_NUMBER = 15;
-  private long opEndRelNanos_;
-  /**
-   * <code>int64 op_end_rel_nanos = 15;</code>
-   */
-  public long getOpEndRelNanos() {
-    return opEndRelNanos_;
-  }
-
-  public static final int ALL_END_REL_NANOS_FIELD_NUMBER = 16;
-  private long allEndRelNanos_;
-  /**
-   * <code>int64 all_end_rel_nanos = 16;</code>
-   */
-  public long getAllEndRelNanos() {
-    return allEndRelNanos_;
-  }
-
-  public static final int SCHEDULED_NANOS_FIELD_NUMBER = 17;
-  private long scheduledNanos_;
-  /**
-   * <code>int64 scheduled_nanos = 17;</code>
-   */
-  public long getScheduledNanos() {
-    return scheduledNanos_;
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    if (!getNodeNameBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 1, nodeName_);
-    }
-    if (allStartMicros_ != 0L) {
-      output.writeInt64(2, allStartMicros_);
-    }
-    if (opStartRelMicros_ != 0L) {
-      output.writeInt64(3, opStartRelMicros_);
-    }
-    if (opEndRelMicros_ != 0L) {
-      output.writeInt64(4, opEndRelMicros_);
-    }
-    if (allEndRelMicros_ != 0L) {
-      output.writeInt64(5, allEndRelMicros_);
-    }
-    for (int i = 0; i < memory_.size(); i++) {
-      output.writeMessage(6, memory_.get(i));
-    }
-    for (int i = 0; i < output_.size(); i++) {
-      output.writeMessage(7, output_.get(i));
-    }
-    if (!getTimelineLabelBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 8, timelineLabel_);
-    }
-    if (scheduledMicros_ != 0L) {
-      output.writeInt64(9, scheduledMicros_);
-    }
-    if (threadId_ != 0) {
-      output.writeUInt32(10, threadId_);
-    }
-    for (int i = 0; i < referencedTensor_.size(); i++) {
-      output.writeMessage(11, referencedTensor_.get(i));
-    }
-    if (memoryStats_ != null) {
-      output.writeMessage(12, getMemoryStats());
-    }
-    if (allStartNanos_ != 0L) {
-      output.writeInt64(13, allStartNanos_);
-    }
-    if (opStartRelNanos_ != 0L) {
-      output.writeInt64(14, opStartRelNanos_);
-    }
-    if (opEndRelNanos_ != 0L) {
-      output.writeInt64(15, opEndRelNanos_);
-    }
-    if (allEndRelNanos_ != 0L) {
-      output.writeInt64(16, allEndRelNanos_);
-    }
-    if (scheduledNanos_ != 0L) {
-      output.writeInt64(17, scheduledNanos_);
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (!getNodeNameBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, nodeName_);
-    }
-    if (allStartMicros_ != 0L) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt64Size(2, allStartMicros_);
-    }
-    if (opStartRelMicros_ != 0L) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt64Size(3, opStartRelMicros_);
-    }
-    if (opEndRelMicros_ != 0L) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt64Size(4, opEndRelMicros_);
-    }
-    if (allEndRelMicros_ != 0L) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt64Size(5, allEndRelMicros_);
-    }
-    for (int i = 0; i < memory_.size(); i++) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(6, memory_.get(i));
-    }
-    for (int i = 0; i < output_.size(); i++) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(7, output_.get(i));
-    }
-    if (!getTimelineLabelBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(8, timelineLabel_);
-    }
-    if (scheduledMicros_ != 0L) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt64Size(9, scheduledMicros_);
-    }
-    if (threadId_ != 0) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeUInt32Size(10, threadId_);
-    }
-    for (int i = 0; i < referencedTensor_.size(); i++) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(11, referencedTensor_.get(i));
-    }
-    if (memoryStats_ != null) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(12, getMemoryStats());
-    }
-    if (allStartNanos_ != 0L) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt64Size(13, allStartNanos_);
-    }
-    if (opStartRelNanos_ != 0L) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt64Size(14, opStartRelNanos_);
-    }
-    if (opEndRelNanos_ != 0L) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt64Size(15, opEndRelNanos_);
-    }
-    if (allEndRelNanos_ != 0L) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt64Size(16, allEndRelNanos_);
-    }
-    if (scheduledNanos_ != 0L) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt64Size(17, scheduledNanos_);
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.framework.NodeExecStats)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.framework.NodeExecStats other = (org.tensorflow.proto.framework.NodeExecStats) obj;
-
-    if (!getNodeName()
-        .equals(other.getNodeName())) return false;
-    if (getAllStartMicros()
-        != other.getAllStartMicros()) return false;
-    if (getOpStartRelMicros()
-        != other.getOpStartRelMicros()) return false;
-    if (getOpEndRelMicros()
-        != other.getOpEndRelMicros()) return false;
-    if (getAllEndRelMicros()
-        != other.getAllEndRelMicros()) return false;
-    if (!getMemoryList()
-        .equals(other.getMemoryList())) return false;
-    if (!getOutputList()
-        .equals(other.getOutputList())) return false;
-    if (!getTimelineLabel()
-        .equals(other.getTimelineLabel())) return false;
-    if (getScheduledMicros()
-        != other.getScheduledMicros()) return false;
-    if (getThreadId()
-        != other.getThreadId()) return false;
-    if (!getReferencedTensorList()
-        .equals(other.getReferencedTensorList())) return false;
-    if (hasMemoryStats() != other.hasMemoryStats()) return false;
-    if (hasMemoryStats()) {
-      if (!getMemoryStats()
-          .equals(other.getMemoryStats())) return false;
-    }
-    if (getAllStartNanos()
-        != other.getAllStartNanos()) return false;
-    if (getOpStartRelNanos()
-        != other.getOpStartRelNanos()) return false;
-    if (getOpEndRelNanos()
-        != other.getOpEndRelNanos()) return false;
-    if (getAllEndRelNanos()
-        != other.getAllEndRelNanos()) return false;
-    if (getScheduledNanos()
-        != other.getScheduledNanos()) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    hash = (37 * hash) + NODE_NAME_FIELD_NUMBER;
-    hash = (53 * hash) + getNodeName().hashCode();
-    hash = (37 * hash) + ALL_START_MICROS_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-        getAllStartMicros());
-    hash = (37 * hash) + OP_START_REL_MICROS_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-        getOpStartRelMicros());
-    hash = (37 * hash) + OP_END_REL_MICROS_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-        getOpEndRelMicros());
-    hash = (37 * hash) + ALL_END_REL_MICROS_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-        getAllEndRelMicros());
-    if (getMemoryCount() > 0) {
-      hash = (37 * hash) + MEMORY_FIELD_NUMBER;
-      hash = (53 * hash) + getMemoryList().hashCode();
-    }
-    if (getOutputCount() > 0) {
-      hash = (37 * hash) + OUTPUT_FIELD_NUMBER;
-      hash = (53 * hash) + getOutputList().hashCode();
-    }
-    hash = (37 * hash) + TIMELINE_LABEL_FIELD_NUMBER;
-    hash = (53 * hash) + getTimelineLabel().hashCode();
-    hash = (37 * hash) + SCHEDULED_MICROS_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-        getScheduledMicros());
-    hash = (37 * hash) + THREAD_ID_FIELD_NUMBER;
-    hash = (53 * hash) + getThreadId();
-    if (getReferencedTensorCount() > 0) {
-      hash = (37 * hash) + REFERENCED_TENSOR_FIELD_NUMBER;
-      hash = (53 * hash) + getReferencedTensorList().hashCode();
-    }
-    if (hasMemoryStats()) {
-      hash = (37 * hash) + MEMORY_STATS_FIELD_NUMBER;
-      hash = (53 * hash) + getMemoryStats().hashCode();
-    }
-    hash = (37 * hash) + ALL_START_NANOS_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-        getAllStartNanos());
-    hash = (37 * hash) + OP_START_REL_NANOS_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-        getOpStartRelNanos());
-    hash = (37 * hash) + OP_END_REL_NANOS_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-        getOpEndRelNanos());
-    hash = (37 * hash) + ALL_END_REL_NANOS_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-        getAllEndRelNanos());
-    hash = (37 * hash) + SCHEDULED_NANOS_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-        getScheduledNanos());
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.framework.NodeExecStats parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.NodeExecStats parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.NodeExecStats parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.NodeExecStats parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.NodeExecStats parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.NodeExecStats parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.NodeExecStats parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.NodeExecStats parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.NodeExecStats parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.NodeExecStats parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.NodeExecStats parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.NodeExecStats parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.framework.NodeExecStats prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * <pre>
-   * Time/size stats recorded for a single execution of a graph node.
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.NodeExecStats}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.NodeExecStats)
-      org.tensorflow.proto.framework.NodeExecStatsOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.StepStatsProtos.internal_static_tensorflow_NodeExecStats_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.StepStatsProtos.internal_static_tensorflow_NodeExecStats_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.NodeExecStats.class, org.tensorflow.proto.framework.NodeExecStats.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.framework.NodeExecStats.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-        getMemoryFieldBuilder();
-        getOutputFieldBuilder();
-        getReferencedTensorFieldBuilder();
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      nodeName_ = "";
-
-      allStartMicros_ = 0L;
-
-      opStartRelMicros_ = 0L;
-
-      opEndRelMicros_ = 0L;
-
-      allEndRelMicros_ = 0L;
-
-      if (memoryBuilder_ == null) {
-        memory_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000001);
-      } else {
-        memoryBuilder_.clear();
-      }
-      if (outputBuilder_ == null) {
-        output_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000002);
-      } else {
-        outputBuilder_.clear();
-      }
-      timelineLabel_ = "";
-
-      scheduledMicros_ = 0L;
-
-      threadId_ = 0;
-
-      if (referencedTensorBuilder_ == null) {
-        referencedTensor_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000004);
-      } else {
-        referencedTensorBuilder_.clear();
-      }
-      if (memoryStatsBuilder_ == null) {
-        memoryStats_ = null;
-      } else {
-        memoryStats_ = null;
-        memoryStatsBuilder_ = null;
-      }
-      allStartNanos_ = 0L;
-
-      opStartRelNanos_ = 0L;
-
-      opEndRelNanos_ = 0L;
-
-      allEndRelNanos_ = 0L;
-
-      scheduledNanos_ = 0L;
-
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.framework.StepStatsProtos.internal_static_tensorflow_NodeExecStats_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.NodeExecStats getDefaultInstanceForType() {
-      return org.tensorflow.proto.framework.NodeExecStats.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.NodeExecStats build() {
-      org.tensorflow.proto.framework.NodeExecStats result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.NodeExecStats buildPartial() {
-      org.tensorflow.proto.framework.NodeExecStats result = new org.tensorflow.proto.framework.NodeExecStats(this);
-      int from_bitField0_ = bitField0_;
-      result.nodeName_ = nodeName_;
-      result.allStartMicros_ = allStartMicros_;
-      result.opStartRelMicros_ = opStartRelMicros_;
-      result.opEndRelMicros_ = opEndRelMicros_;
-      result.allEndRelMicros_ = allEndRelMicros_;
-      if (memoryBuilder_ == null) {
-        if (((bitField0_ & 0x00000001) != 0)) {
-          memory_ = java.util.Collections.unmodifiableList(memory_);
-          bitField0_ = (bitField0_ & ~0x00000001);
-        }
-        result.memory_ = memory_;
-      } else {
-        result.memory_ = memoryBuilder_.build();
-      }
-      if (outputBuilder_ == null) {
-        if (((bitField0_ & 0x00000002) != 0)) {
-          output_ = java.util.Collections.unmodifiableList(output_);
-          bitField0_ = (bitField0_ & ~0x00000002);
-        }
-        result.output_ = output_;
-      } else {
-        result.output_ = outputBuilder_.build();
-      }
-      result.timelineLabel_ = timelineLabel_;
-      result.scheduledMicros_ = scheduledMicros_;
-      result.threadId_ = threadId_;
-      if (referencedTensorBuilder_ == null) {
-        if (((bitField0_ & 0x00000004) != 0)) {
-          referencedTensor_ = java.util.Collections.unmodifiableList(referencedTensor_);
-          bitField0_ = (bitField0_ & ~0x00000004);
-        }
-        result.referencedTensor_ = referencedTensor_;
-      } else {
-        result.referencedTensor_ = referencedTensorBuilder_.build();
-      }
-      if (memoryStatsBuilder_ == null) {
-        result.memoryStats_ = memoryStats_;
-      } else {
-        result.memoryStats_ = memoryStatsBuilder_.build();
-      }
-      result.allStartNanos_ = allStartNanos_;
-      result.opStartRelNanos_ = opStartRelNanos_;
-      result.opEndRelNanos_ = opEndRelNanos_;
-      result.allEndRelNanos_ = allEndRelNanos_;
-      result.scheduledNanos_ = scheduledNanos_;
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.framework.NodeExecStats) {
-        return mergeFrom((org.tensorflow.proto.framework.NodeExecStats)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.framework.NodeExecStats other) {
-      if (other == org.tensorflow.proto.framework.NodeExecStats.getDefaultInstance()) return this;
-      if (!other.getNodeName().isEmpty()) {
-        nodeName_ = other.nodeName_;
-        onChanged();
-      }
-      if (other.getAllStartMicros() != 0L) {
-        setAllStartMicros(other.getAllStartMicros());
-      }
-      if (other.getOpStartRelMicros() != 0L) {
-        setOpStartRelMicros(other.getOpStartRelMicros());
-      }
-      if (other.getOpEndRelMicros() != 0L) {
-        setOpEndRelMicros(other.getOpEndRelMicros());
-      }
-      if (other.getAllEndRelMicros() != 0L) {
-        setAllEndRelMicros(other.getAllEndRelMicros());
-      }
-      if (memoryBuilder_ == null) {
-        if (!other.memory_.isEmpty()) {
-          if (memory_.isEmpty()) {
-            memory_ = other.memory_;
-            bitField0_ = (bitField0_ & ~0x00000001);
-          } else {
-            ensureMemoryIsMutable();
-            memory_.addAll(other.memory_);
-          }
-          onChanged();
-        }
-      } else {
-        if (!other.memory_.isEmpty()) {
-          if (memoryBuilder_.isEmpty()) {
-            memoryBuilder_.dispose();
-            memoryBuilder_ = null;
-            memory_ = other.memory_;
-            bitField0_ = (bitField0_ & ~0x00000001);
-            memoryBuilder_ = 
-              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
-                 getMemoryFieldBuilder() : null;
-          } else {
-            memoryBuilder_.addAllMessages(other.memory_);
-          }
-        }
-      }
-      if (outputBuilder_ == null) {
-        if (!other.output_.isEmpty()) {
-          if (output_.isEmpty()) {
-            output_ = other.output_;
-            bitField0_ = (bitField0_ & ~0x00000002);
-          } else {
-            ensureOutputIsMutable();
-            output_.addAll(other.output_);
-          }
-          onChanged();
-        }
-      } else {
-        if (!other.output_.isEmpty()) {
-          if (outputBuilder_.isEmpty()) {
-            outputBuilder_.dispose();
-            outputBuilder_ = null;
-            output_ = other.output_;
-            bitField0_ = (bitField0_ & ~0x00000002);
-            outputBuilder_ = 
-              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
-                 getOutputFieldBuilder() : null;
-          } else {
-            outputBuilder_.addAllMessages(other.output_);
-          }
-        }
-      }
-      if (!other.getTimelineLabel().isEmpty()) {
-        timelineLabel_ = other.timelineLabel_;
-        onChanged();
-      }
-      if (other.getScheduledMicros() != 0L) {
-        setScheduledMicros(other.getScheduledMicros());
-      }
-      if (other.getThreadId() != 0) {
-        setThreadId(other.getThreadId());
-      }
-      if (referencedTensorBuilder_ == null) {
-        if (!other.referencedTensor_.isEmpty()) {
-          if (referencedTensor_.isEmpty()) {
-            referencedTensor_ = other.referencedTensor_;
-            bitField0_ = (bitField0_ & ~0x00000004);
-          } else {
-            ensureReferencedTensorIsMutable();
-            referencedTensor_.addAll(other.referencedTensor_);
-          }
-          onChanged();
-        }
-      } else {
-        if (!other.referencedTensor_.isEmpty()) {
-          if (referencedTensorBuilder_.isEmpty()) {
-            referencedTensorBuilder_.dispose();
-            referencedTensorBuilder_ = null;
-            referencedTensor_ = other.referencedTensor_;
-            bitField0_ = (bitField0_ & ~0x00000004);
-            referencedTensorBuilder_ = 
-              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
-                 getReferencedTensorFieldBuilder() : null;
-          } else {
-            referencedTensorBuilder_.addAllMessages(other.referencedTensor_);
-          }
-        }
-      }
-      if (other.hasMemoryStats()) {
-        mergeMemoryStats(other.getMemoryStats());
-      }
-      if (other.getAllStartNanos() != 0L) {
-        setAllStartNanos(other.getAllStartNanos());
-      }
-      if (other.getOpStartRelNanos() != 0L) {
-        setOpStartRelNanos(other.getOpStartRelNanos());
-      }
-      if (other.getOpEndRelNanos() != 0L) {
-        setOpEndRelNanos(other.getOpEndRelNanos());
-      }
-      if (other.getAllEndRelNanos() != 0L) {
-        setAllEndRelNanos(other.getAllEndRelNanos());
-      }
-      if (other.getScheduledNanos() != 0L) {
-        setScheduledNanos(other.getScheduledNanos());
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.framework.NodeExecStats parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.framework.NodeExecStats) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-    private int bitField0_;
-
-    private java.lang.Object nodeName_ = "";
-    /**
-     * <pre>
-     * TODO(tucker): Use some more compact form of node identity than
-     * the full string name.  Either all processes should agree on a
-     * global id (cost_id?) for each node, or we should use a hash of
-     * the name.
-     * </pre>
-     *
-     * <code>string node_name = 1;</code>
-     */
-    public java.lang.String getNodeName() {
-      java.lang.Object ref = nodeName_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        nodeName_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * TODO(tucker): Use some more compact form of node identity than
-     * the full string name.  Either all processes should agree on a
-     * global id (cost_id?) for each node, or we should use a hash of
-     * the name.
-     * </pre>
-     *
-     * <code>string node_name = 1;</code>
-     */
-    public com.google.protobuf.ByteString
-        getNodeNameBytes() {
-      java.lang.Object ref = nodeName_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        nodeName_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * TODO(tucker): Use some more compact form of node identity than
-     * the full string name.  Either all processes should agree on a
-     * global id (cost_id?) for each node, or we should use a hash of
-     * the name.
-     * </pre>
-     *
-     * <code>string node_name = 1;</code>
-     */
-    public Builder setNodeName(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      nodeName_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * TODO(tucker): Use some more compact form of node identity than
-     * the full string name.  Either all processes should agree on a
-     * global id (cost_id?) for each node, or we should use a hash of
-     * the name.
-     * </pre>
-     *
-     * <code>string node_name = 1;</code>
-     */
-    public Builder clearNodeName() {
-      
-      nodeName_ = getDefaultInstance().getNodeName();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * TODO(tucker): Use some more compact form of node identity than
-     * the full string name.  Either all processes should agree on a
-     * global id (cost_id?) for each node, or we should use a hash of
-     * the name.
-     * </pre>
-     *
-     * <code>string node_name = 1;</code>
-     */
-    public Builder setNodeNameBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      nodeName_ = value;
-      onChanged();
-      return this;
-    }
-
-    private long allStartMicros_ ;
-    /**
-     * <code>int64 all_start_micros = 2;</code>
-     */
-    public long getAllStartMicros() {
-      return allStartMicros_;
-    }
-    /**
-     * <code>int64 all_start_micros = 2;</code>
-     */
-    public Builder setAllStartMicros(long value) {
-      
-      allStartMicros_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>int64 all_start_micros = 2;</code>
-     */
-    public Builder clearAllStartMicros() {
-      
-      allStartMicros_ = 0L;
-      onChanged();
-      return this;
-    }
-
-    private long opStartRelMicros_ ;
-    /**
-     * <code>int64 op_start_rel_micros = 3;</code>
-     */
-    public long getOpStartRelMicros() {
-      return opStartRelMicros_;
-    }
-    /**
-     * <code>int64 op_start_rel_micros = 3;</code>
-     */
-    public Builder setOpStartRelMicros(long value) {
-      
-      opStartRelMicros_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>int64 op_start_rel_micros = 3;</code>
-     */
-    public Builder clearOpStartRelMicros() {
-      
-      opStartRelMicros_ = 0L;
-      onChanged();
-      return this;
-    }
-
-    private long opEndRelMicros_ ;
-    /**
-     * <code>int64 op_end_rel_micros = 4;</code>
-     */
-    public long getOpEndRelMicros() {
-      return opEndRelMicros_;
-    }
-    /**
-     * <code>int64 op_end_rel_micros = 4;</code>
-     */
-    public Builder setOpEndRelMicros(long value) {
-      
-      opEndRelMicros_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>int64 op_end_rel_micros = 4;</code>
-     */
-    public Builder clearOpEndRelMicros() {
-      
-      opEndRelMicros_ = 0L;
-      onChanged();
-      return this;
-    }
-
-    private long allEndRelMicros_ ;
-    /**
-     * <code>int64 all_end_rel_micros = 5;</code>
-     */
-    public long getAllEndRelMicros() {
-      return allEndRelMicros_;
-    }
-    /**
-     * <code>int64 all_end_rel_micros = 5;</code>
-     */
-    public Builder setAllEndRelMicros(long value) {
-      
-      allEndRelMicros_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>int64 all_end_rel_micros = 5;</code>
-     */
-    public Builder clearAllEndRelMicros() {
-      
-      allEndRelMicros_ = 0L;
-      onChanged();
-      return this;
-    }
-
-    private java.util.List<org.tensorflow.proto.framework.AllocatorMemoryUsed> memory_ =
-      java.util.Collections.emptyList();
-    private void ensureMemoryIsMutable() {
-      if (!((bitField0_ & 0x00000001) != 0)) {
-        memory_ = new java.util.ArrayList<org.tensorflow.proto.framework.AllocatorMemoryUsed>(memory_);
-        bitField0_ |= 0x00000001;
-       }
-    }
-
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.framework.AllocatorMemoryUsed, org.tensorflow.proto.framework.AllocatorMemoryUsed.Builder, org.tensorflow.proto.framework.AllocatorMemoryUsedOrBuilder> memoryBuilder_;
-
-    /**
-     * <code>repeated .tensorflow.AllocatorMemoryUsed memory = 6;</code>
-     */
-    public java.util.List<org.tensorflow.proto.framework.AllocatorMemoryUsed> getMemoryList() {
-      if (memoryBuilder_ == null) {
-        return java.util.Collections.unmodifiableList(memory_);
-      } else {
-        return memoryBuilder_.getMessageList();
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.AllocatorMemoryUsed memory = 6;</code>
-     */
-    public int getMemoryCount() {
-      if (memoryBuilder_ == null) {
-        return memory_.size();
-      } else {
-        return memoryBuilder_.getCount();
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.AllocatorMemoryUsed memory = 6;</code>
-     */
-    public org.tensorflow.proto.framework.AllocatorMemoryUsed getMemory(int index) {
-      if (memoryBuilder_ == null) {
-        return memory_.get(index);
-      } else {
-        return memoryBuilder_.getMessage(index);
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.AllocatorMemoryUsed memory = 6;</code>
-     */
-    public Builder setMemory(
-        int index, org.tensorflow.proto.framework.AllocatorMemoryUsed value) {
-      if (memoryBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureMemoryIsMutable();
-        memory_.set(index, value);
-        onChanged();
-      } else {
-        memoryBuilder_.setMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.AllocatorMemoryUsed memory = 6;</code>
-     */
-    public Builder setMemory(
-        int index, org.tensorflow.proto.framework.AllocatorMemoryUsed.Builder builderForValue) {
-      if (memoryBuilder_ == null) {
-        ensureMemoryIsMutable();
-        memory_.set(index, builderForValue.build());
-        onChanged();
-      } else {
-        memoryBuilder_.setMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.AllocatorMemoryUsed memory = 6;</code>
-     */
-    public Builder addMemory(org.tensorflow.proto.framework.AllocatorMemoryUsed value) {
-      if (memoryBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureMemoryIsMutable();
-        memory_.add(value);
-        onChanged();
-      } else {
-        memoryBuilder_.addMessage(value);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.AllocatorMemoryUsed memory = 6;</code>
-     */
-    public Builder addMemory(
-        int index, org.tensorflow.proto.framework.AllocatorMemoryUsed value) {
-      if (memoryBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureMemoryIsMutable();
-        memory_.add(index, value);
-        onChanged();
-      } else {
-        memoryBuilder_.addMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.AllocatorMemoryUsed memory = 6;</code>
-     */
-    public Builder addMemory(
-        org.tensorflow.proto.framework.AllocatorMemoryUsed.Builder builderForValue) {
-      if (memoryBuilder_ == null) {
-        ensureMemoryIsMutable();
-        memory_.add(builderForValue.build());
-        onChanged();
-      } else {
-        memoryBuilder_.addMessage(builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.AllocatorMemoryUsed memory = 6;</code>
-     */
-    public Builder addMemory(
-        int index, org.tensorflow.proto.framework.AllocatorMemoryUsed.Builder builderForValue) {
-      if (memoryBuilder_ == null) {
-        ensureMemoryIsMutable();
-        memory_.add(index, builderForValue.build());
-        onChanged();
-      } else {
-        memoryBuilder_.addMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.AllocatorMemoryUsed memory = 6;</code>
-     */
-    public Builder addAllMemory(
-        java.lang.Iterable<? extends org.tensorflow.proto.framework.AllocatorMemoryUsed> values) {
-      if (memoryBuilder_ == null) {
-        ensureMemoryIsMutable();
-        com.google.protobuf.AbstractMessageLite.Builder.addAll(
-            values, memory_);
-        onChanged();
-      } else {
-        memoryBuilder_.addAllMessages(values);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.AllocatorMemoryUsed memory = 6;</code>
-     */
-    public Builder clearMemory() {
-      if (memoryBuilder_ == null) {
-        memory_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000001);
-        onChanged();
-      } else {
-        memoryBuilder_.clear();
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.AllocatorMemoryUsed memory = 6;</code>
-     */
-    public Builder removeMemory(int index) {
-      if (memoryBuilder_ == null) {
-        ensureMemoryIsMutable();
-        memory_.remove(index);
-        onChanged();
-      } else {
-        memoryBuilder_.remove(index);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.AllocatorMemoryUsed memory = 6;</code>
-     */
-    public org.tensorflow.proto.framework.AllocatorMemoryUsed.Builder getMemoryBuilder(
-        int index) {
-      return getMemoryFieldBuilder().getBuilder(index);
-    }
-    /**
-     * <code>repeated .tensorflow.AllocatorMemoryUsed memory = 6;</code>
-     */
-    public org.tensorflow.proto.framework.AllocatorMemoryUsedOrBuilder getMemoryOrBuilder(
-        int index) {
-      if (memoryBuilder_ == null) {
-        return memory_.get(index);  } else {
-        return memoryBuilder_.getMessageOrBuilder(index);
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.AllocatorMemoryUsed memory = 6;</code>
-     */
-    public java.util.List<? extends org.tensorflow.proto.framework.AllocatorMemoryUsedOrBuilder> 
-         getMemoryOrBuilderList() {
-      if (memoryBuilder_ != null) {
-        return memoryBuilder_.getMessageOrBuilderList();
-      } else {
-        return java.util.Collections.unmodifiableList(memory_);
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.AllocatorMemoryUsed memory = 6;</code>
-     */
-    public org.tensorflow.proto.framework.AllocatorMemoryUsed.Builder addMemoryBuilder() {
-      return getMemoryFieldBuilder().addBuilder(
-          org.tensorflow.proto.framework.AllocatorMemoryUsed.getDefaultInstance());
-    }
-    /**
-     * <code>repeated .tensorflow.AllocatorMemoryUsed memory = 6;</code>
-     */
-    public org.tensorflow.proto.framework.AllocatorMemoryUsed.Builder addMemoryBuilder(
-        int index) {
-      return getMemoryFieldBuilder().addBuilder(
-          index, org.tensorflow.proto.framework.AllocatorMemoryUsed.getDefaultInstance());
-    }
-    /**
-     * <code>repeated .tensorflow.AllocatorMemoryUsed memory = 6;</code>
-     */
-    public java.util.List<org.tensorflow.proto.framework.AllocatorMemoryUsed.Builder> 
-         getMemoryBuilderList() {
-      return getMemoryFieldBuilder().getBuilderList();
-    }
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.framework.AllocatorMemoryUsed, org.tensorflow.proto.framework.AllocatorMemoryUsed.Builder, org.tensorflow.proto.framework.AllocatorMemoryUsedOrBuilder> 
-        getMemoryFieldBuilder() {
-      if (memoryBuilder_ == null) {
-        memoryBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
-            org.tensorflow.proto.framework.AllocatorMemoryUsed, org.tensorflow.proto.framework.AllocatorMemoryUsed.Builder, org.tensorflow.proto.framework.AllocatorMemoryUsedOrBuilder>(
-                memory_,
-                ((bitField0_ & 0x00000001) != 0),
-                getParentForChildren(),
-                isClean());
-        memory_ = null;
-      }
-      return memoryBuilder_;
-    }
-
-    private java.util.List<org.tensorflow.proto.framework.NodeOutput> output_ =
-      java.util.Collections.emptyList();
-    private void ensureOutputIsMutable() {
-      if (!((bitField0_ & 0x00000002) != 0)) {
-        output_ = new java.util.ArrayList<org.tensorflow.proto.framework.NodeOutput>(output_);
-        bitField0_ |= 0x00000002;
-       }
-    }
-
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.framework.NodeOutput, org.tensorflow.proto.framework.NodeOutput.Builder, org.tensorflow.proto.framework.NodeOutputOrBuilder> outputBuilder_;
-
-    /**
-     * <code>repeated .tensorflow.NodeOutput output = 7;</code>
-     */
-    public java.util.List<org.tensorflow.proto.framework.NodeOutput> getOutputList() {
-      if (outputBuilder_ == null) {
-        return java.util.Collections.unmodifiableList(output_);
-      } else {
-        return outputBuilder_.getMessageList();
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.NodeOutput output = 7;</code>
-     */
-    public int getOutputCount() {
-      if (outputBuilder_ == null) {
-        return output_.size();
-      } else {
-        return outputBuilder_.getCount();
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.NodeOutput output = 7;</code>
-     */
-    public org.tensorflow.proto.framework.NodeOutput getOutput(int index) {
-      if (outputBuilder_ == null) {
-        return output_.get(index);
-      } else {
-        return outputBuilder_.getMessage(index);
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.NodeOutput output = 7;</code>
-     */
-    public Builder setOutput(
-        int index, org.tensorflow.proto.framework.NodeOutput value) {
-      if (outputBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureOutputIsMutable();
-        output_.set(index, value);
-        onChanged();
-      } else {
-        outputBuilder_.setMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.NodeOutput output = 7;</code>
-     */
-    public Builder setOutput(
-        int index, org.tensorflow.proto.framework.NodeOutput.Builder builderForValue) {
-      if (outputBuilder_ == null) {
-        ensureOutputIsMutable();
-        output_.set(index, builderForValue.build());
-        onChanged();
-      } else {
-        outputBuilder_.setMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.NodeOutput output = 7;</code>
-     */
-    public Builder addOutput(org.tensorflow.proto.framework.NodeOutput value) {
-      if (outputBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureOutputIsMutable();
-        output_.add(value);
-        onChanged();
-      } else {
-        outputBuilder_.addMessage(value);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.NodeOutput output = 7;</code>
-     */
-    public Builder addOutput(
-        int index, org.tensorflow.proto.framework.NodeOutput value) {
-      if (outputBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureOutputIsMutable();
-        output_.add(index, value);
-        onChanged();
-      } else {
-        outputBuilder_.addMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.NodeOutput output = 7;</code>
-     */
-    public Builder addOutput(
-        org.tensorflow.proto.framework.NodeOutput.Builder builderForValue) {
-      if (outputBuilder_ == null) {
-        ensureOutputIsMutable();
-        output_.add(builderForValue.build());
-        onChanged();
-      } else {
-        outputBuilder_.addMessage(builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.NodeOutput output = 7;</code>
-     */
-    public Builder addOutput(
-        int index, org.tensorflow.proto.framework.NodeOutput.Builder builderForValue) {
-      if (outputBuilder_ == null) {
-        ensureOutputIsMutable();
-        output_.add(index, builderForValue.build());
-        onChanged();
-      } else {
-        outputBuilder_.addMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.NodeOutput output = 7;</code>
-     */
-    public Builder addAllOutput(
-        java.lang.Iterable<? extends org.tensorflow.proto.framework.NodeOutput> values) {
-      if (outputBuilder_ == null) {
-        ensureOutputIsMutable();
-        com.google.protobuf.AbstractMessageLite.Builder.addAll(
-            values, output_);
-        onChanged();
-      } else {
-        outputBuilder_.addAllMessages(values);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.NodeOutput output = 7;</code>
-     */
-    public Builder clearOutput() {
-      if (outputBuilder_ == null) {
-        output_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000002);
-        onChanged();
-      } else {
-        outputBuilder_.clear();
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.NodeOutput output = 7;</code>
-     */
-    public Builder removeOutput(int index) {
-      if (outputBuilder_ == null) {
-        ensureOutputIsMutable();
-        output_.remove(index);
-        onChanged();
-      } else {
-        outputBuilder_.remove(index);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.NodeOutput output = 7;</code>
-     */
-    public org.tensorflow.proto.framework.NodeOutput.Builder getOutputBuilder(
-        int index) {
-      return getOutputFieldBuilder().getBuilder(index);
-    }
-    /**
-     * <code>repeated .tensorflow.NodeOutput output = 7;</code>
-     */
-    public org.tensorflow.proto.framework.NodeOutputOrBuilder getOutputOrBuilder(
-        int index) {
-      if (outputBuilder_ == null) {
-        return output_.get(index);  } else {
-        return outputBuilder_.getMessageOrBuilder(index);
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.NodeOutput output = 7;</code>
-     */
-    public java.util.List<? extends org.tensorflow.proto.framework.NodeOutputOrBuilder> 
-         getOutputOrBuilderList() {
-      if (outputBuilder_ != null) {
-        return outputBuilder_.getMessageOrBuilderList();
-      } else {
-        return java.util.Collections.unmodifiableList(output_);
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.NodeOutput output = 7;</code>
-     */
-    public org.tensorflow.proto.framework.NodeOutput.Builder addOutputBuilder() {
-      return getOutputFieldBuilder().addBuilder(
-          org.tensorflow.proto.framework.NodeOutput.getDefaultInstance());
-    }
-    /**
-     * <code>repeated .tensorflow.NodeOutput output = 7;</code>
-     */
-    public org.tensorflow.proto.framework.NodeOutput.Builder addOutputBuilder(
-        int index) {
-      return getOutputFieldBuilder().addBuilder(
-          index, org.tensorflow.proto.framework.NodeOutput.getDefaultInstance());
-    }
-    /**
-     * <code>repeated .tensorflow.NodeOutput output = 7;</code>
-     */
-    public java.util.List<org.tensorflow.proto.framework.NodeOutput.Builder> 
-         getOutputBuilderList() {
-      return getOutputFieldBuilder().getBuilderList();
-    }
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.framework.NodeOutput, org.tensorflow.proto.framework.NodeOutput.Builder, org.tensorflow.proto.framework.NodeOutputOrBuilder> 
-        getOutputFieldBuilder() {
-      if (outputBuilder_ == null) {
-        outputBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
-            org.tensorflow.proto.framework.NodeOutput, org.tensorflow.proto.framework.NodeOutput.Builder, org.tensorflow.proto.framework.NodeOutputOrBuilder>(
-                output_,
-                ((bitField0_ & 0x00000002) != 0),
-                getParentForChildren(),
-                isClean());
-        output_ = null;
-      }
-      return outputBuilder_;
-    }
-
-    private java.lang.Object timelineLabel_ = "";
-    /**
-     * <code>string timeline_label = 8;</code>
-     */
-    public java.lang.String getTimelineLabel() {
-      java.lang.Object ref = timelineLabel_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        timelineLabel_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <code>string timeline_label = 8;</code>
-     */
-    public com.google.protobuf.ByteString
-        getTimelineLabelBytes() {
-      java.lang.Object ref = timelineLabel_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        timelineLabel_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <code>string timeline_label = 8;</code>
-     */
-    public Builder setTimelineLabel(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      timelineLabel_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>string timeline_label = 8;</code>
-     */
-    public Builder clearTimelineLabel() {
-      
-      timelineLabel_ = getDefaultInstance().getTimelineLabel();
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>string timeline_label = 8;</code>
-     */
-    public Builder setTimelineLabelBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      timelineLabel_ = value;
-      onChanged();
-      return this;
-    }
-
-    private long scheduledMicros_ ;
-    /**
-     * <code>int64 scheduled_micros = 9;</code>
-     */
-    public long getScheduledMicros() {
-      return scheduledMicros_;
-    }
-    /**
-     * <code>int64 scheduled_micros = 9;</code>
-     */
-    public Builder setScheduledMicros(long value) {
-      
-      scheduledMicros_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>int64 scheduled_micros = 9;</code>
-     */
-    public Builder clearScheduledMicros() {
-      
-      scheduledMicros_ = 0L;
-      onChanged();
-      return this;
-    }
-
-    private int threadId_ ;
-    /**
-     * <code>uint32 thread_id = 10;</code>
-     */
-    public int getThreadId() {
-      return threadId_;
-    }
-    /**
-     * <code>uint32 thread_id = 10;</code>
-     */
-    public Builder setThreadId(int value) {
-      
-      threadId_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>uint32 thread_id = 10;</code>
-     */
-    public Builder clearThreadId() {
-      
-      threadId_ = 0;
-      onChanged();
-      return this;
-    }
-
-    private java.util.List<org.tensorflow.proto.framework.AllocationDescription> referencedTensor_ =
-      java.util.Collections.emptyList();
-    private void ensureReferencedTensorIsMutable() {
-      if (!((bitField0_ & 0x00000004) != 0)) {
-        referencedTensor_ = new java.util.ArrayList<org.tensorflow.proto.framework.AllocationDescription>(referencedTensor_);
-        bitField0_ |= 0x00000004;
-       }
-    }
-
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.framework.AllocationDescription, org.tensorflow.proto.framework.AllocationDescription.Builder, org.tensorflow.proto.framework.AllocationDescriptionOrBuilder> referencedTensorBuilder_;
-
-    /**
-     * <code>repeated .tensorflow.AllocationDescription referenced_tensor = 11;</code>
-     */
-    public java.util.List<org.tensorflow.proto.framework.AllocationDescription> getReferencedTensorList() {
-      if (referencedTensorBuilder_ == null) {
-        return java.util.Collections.unmodifiableList(referencedTensor_);
-      } else {
-        return referencedTensorBuilder_.getMessageList();
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.AllocationDescription referenced_tensor = 11;</code>
-     */
-    public int getReferencedTensorCount() {
-      if (referencedTensorBuilder_ == null) {
-        return referencedTensor_.size();
-      } else {
-        return referencedTensorBuilder_.getCount();
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.AllocationDescription referenced_tensor = 11;</code>
-     */
-    public org.tensorflow.proto.framework.AllocationDescription getReferencedTensor(int index) {
-      if (referencedTensorBuilder_ == null) {
-        return referencedTensor_.get(index);
-      } else {
-        return referencedTensorBuilder_.getMessage(index);
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.AllocationDescription referenced_tensor = 11;</code>
-     */
-    public Builder setReferencedTensor(
-        int index, org.tensorflow.proto.framework.AllocationDescription value) {
-      if (referencedTensorBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureReferencedTensorIsMutable();
-        referencedTensor_.set(index, value);
-        onChanged();
-      } else {
-        referencedTensorBuilder_.setMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.AllocationDescription referenced_tensor = 11;</code>
-     */
-    public Builder setReferencedTensor(
-        int index, org.tensorflow.proto.framework.AllocationDescription.Builder builderForValue) {
-      if (referencedTensorBuilder_ == null) {
-        ensureReferencedTensorIsMutable();
-        referencedTensor_.set(index, builderForValue.build());
-        onChanged();
-      } else {
-        referencedTensorBuilder_.setMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.AllocationDescription referenced_tensor = 11;</code>
-     */
-    public Builder addReferencedTensor(org.tensorflow.proto.framework.AllocationDescription value) {
-      if (referencedTensorBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureReferencedTensorIsMutable();
-        referencedTensor_.add(value);
-        onChanged();
-      } else {
-        referencedTensorBuilder_.addMessage(value);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.AllocationDescription referenced_tensor = 11;</code>
-     */
-    public Builder addReferencedTensor(
-        int index, org.tensorflow.proto.framework.AllocationDescription value) {
-      if (referencedTensorBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureReferencedTensorIsMutable();
-        referencedTensor_.add(index, value);
-        onChanged();
-      } else {
-        referencedTensorBuilder_.addMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.AllocationDescription referenced_tensor = 11;</code>
-     */
-    public Builder addReferencedTensor(
-        org.tensorflow.proto.framework.AllocationDescription.Builder builderForValue) {
-      if (referencedTensorBuilder_ == null) {
-        ensureReferencedTensorIsMutable();
-        referencedTensor_.add(builderForValue.build());
-        onChanged();
-      } else {
-        referencedTensorBuilder_.addMessage(builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.AllocationDescription referenced_tensor = 11;</code>
-     */
-    public Builder addReferencedTensor(
-        int index, org.tensorflow.proto.framework.AllocationDescription.Builder builderForValue) {
-      if (referencedTensorBuilder_ == null) {
-        ensureReferencedTensorIsMutable();
-        referencedTensor_.add(index, builderForValue.build());
-        onChanged();
-      } else {
-        referencedTensorBuilder_.addMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.AllocationDescription referenced_tensor = 11;</code>
-     */
-    public Builder addAllReferencedTensor(
-        java.lang.Iterable<? extends org.tensorflow.proto.framework.AllocationDescription> values) {
-      if (referencedTensorBuilder_ == null) {
-        ensureReferencedTensorIsMutable();
-        com.google.protobuf.AbstractMessageLite.Builder.addAll(
-            values, referencedTensor_);
-        onChanged();
-      } else {
-        referencedTensorBuilder_.addAllMessages(values);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.AllocationDescription referenced_tensor = 11;</code>
-     */
-    public Builder clearReferencedTensor() {
-      if (referencedTensorBuilder_ == null) {
-        referencedTensor_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000004);
-        onChanged();
-      } else {
-        referencedTensorBuilder_.clear();
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.AllocationDescription referenced_tensor = 11;</code>
-     */
-    public Builder removeReferencedTensor(int index) {
-      if (referencedTensorBuilder_ == null) {
-        ensureReferencedTensorIsMutable();
-        referencedTensor_.remove(index);
-        onChanged();
-      } else {
-        referencedTensorBuilder_.remove(index);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.AllocationDescription referenced_tensor = 11;</code>
-     */
-    public org.tensorflow.proto.framework.AllocationDescription.Builder getReferencedTensorBuilder(
-        int index) {
-      return getReferencedTensorFieldBuilder().getBuilder(index);
-    }
-    /**
-     * <code>repeated .tensorflow.AllocationDescription referenced_tensor = 11;</code>
-     */
-    public org.tensorflow.proto.framework.AllocationDescriptionOrBuilder getReferencedTensorOrBuilder(
-        int index) {
-      if (referencedTensorBuilder_ == null) {
-        return referencedTensor_.get(index);  } else {
-        return referencedTensorBuilder_.getMessageOrBuilder(index);
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.AllocationDescription referenced_tensor = 11;</code>
-     */
-    public java.util.List<? extends org.tensorflow.proto.framework.AllocationDescriptionOrBuilder> 
-         getReferencedTensorOrBuilderList() {
-      if (referencedTensorBuilder_ != null) {
-        return referencedTensorBuilder_.getMessageOrBuilderList();
-      } else {
-        return java.util.Collections.unmodifiableList(referencedTensor_);
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.AllocationDescription referenced_tensor = 11;</code>
-     */
-    public org.tensorflow.proto.framework.AllocationDescription.Builder addReferencedTensorBuilder() {
-      return getReferencedTensorFieldBuilder().addBuilder(
-          org.tensorflow.proto.framework.AllocationDescription.getDefaultInstance());
-    }
-    /**
-     * <code>repeated .tensorflow.AllocationDescription referenced_tensor = 11;</code>
-     */
-    public org.tensorflow.proto.framework.AllocationDescription.Builder addReferencedTensorBuilder(
-        int index) {
-      return getReferencedTensorFieldBuilder().addBuilder(
-          index, org.tensorflow.proto.framework.AllocationDescription.getDefaultInstance());
-    }
-    /**
-     * <code>repeated .tensorflow.AllocationDescription referenced_tensor = 11;</code>
-     */
-    public java.util.List<org.tensorflow.proto.framework.AllocationDescription.Builder> 
-         getReferencedTensorBuilderList() {
-      return getReferencedTensorFieldBuilder().getBuilderList();
-    }
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.framework.AllocationDescription, org.tensorflow.proto.framework.AllocationDescription.Builder, org.tensorflow.proto.framework.AllocationDescriptionOrBuilder> 
-        getReferencedTensorFieldBuilder() {
-      if (referencedTensorBuilder_ == null) {
-        referencedTensorBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
-            org.tensorflow.proto.framework.AllocationDescription, org.tensorflow.proto.framework.AllocationDescription.Builder, org.tensorflow.proto.framework.AllocationDescriptionOrBuilder>(
-                referencedTensor_,
-                ((bitField0_ & 0x00000004) != 0),
-                getParentForChildren(),
-                isClean());
-        referencedTensor_ = null;
-      }
-      return referencedTensorBuilder_;
-    }
-
-    private org.tensorflow.proto.framework.MemoryStats memoryStats_;
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.MemoryStats, org.tensorflow.proto.framework.MemoryStats.Builder, org.tensorflow.proto.framework.MemoryStatsOrBuilder> memoryStatsBuilder_;
-    /**
-     * <code>.tensorflow.MemoryStats memory_stats = 12;</code>
-     */
-    public boolean hasMemoryStats() {
-      return memoryStatsBuilder_ != null || memoryStats_ != null;
-    }
-    /**
-     * <code>.tensorflow.MemoryStats memory_stats = 12;</code>
-     */
-    public org.tensorflow.proto.framework.MemoryStats getMemoryStats() {
-      if (memoryStatsBuilder_ == null) {
-        return memoryStats_ == null ? org.tensorflow.proto.framework.MemoryStats.getDefaultInstance() : memoryStats_;
-      } else {
-        return memoryStatsBuilder_.getMessage();
-      }
-    }
-    /**
-     * <code>.tensorflow.MemoryStats memory_stats = 12;</code>
-     */
-    public Builder setMemoryStats(org.tensorflow.proto.framework.MemoryStats value) {
-      if (memoryStatsBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        memoryStats_ = value;
-        onChanged();
-      } else {
-        memoryStatsBuilder_.setMessage(value);
-      }
-
-      return this;
-    }
-    /**
-     * <code>.tensorflow.MemoryStats memory_stats = 12;</code>
-     */
-    public Builder setMemoryStats(
-        org.tensorflow.proto.framework.MemoryStats.Builder builderForValue) {
-      if (memoryStatsBuilder_ == null) {
-        memoryStats_ = builderForValue.build();
-        onChanged();
-      } else {
-        memoryStatsBuilder_.setMessage(builderForValue.build());
-      }
-
-      return this;
-    }
-    /**
-     * <code>.tensorflow.MemoryStats memory_stats = 12;</code>
-     */
-    public Builder mergeMemoryStats(org.tensorflow.proto.framework.MemoryStats value) {
-      if (memoryStatsBuilder_ == null) {
-        if (memoryStats_ != null) {
-          memoryStats_ =
-            org.tensorflow.proto.framework.MemoryStats.newBuilder(memoryStats_).mergeFrom(value).buildPartial();
-        } else {
-          memoryStats_ = value;
-        }
-        onChanged();
-      } else {
-        memoryStatsBuilder_.mergeFrom(value);
-      }
-
-      return this;
-    }
-    /**
-     * <code>.tensorflow.MemoryStats memory_stats = 12;</code>
-     */
-    public Builder clearMemoryStats() {
-      if (memoryStatsBuilder_ == null) {
-        memoryStats_ = null;
-        onChanged();
-      } else {
-        memoryStats_ = null;
-        memoryStatsBuilder_ = null;
-      }
-
-      return this;
-    }
-    /**
-     * <code>.tensorflow.MemoryStats memory_stats = 12;</code>
-     */
-    public org.tensorflow.proto.framework.MemoryStats.Builder getMemoryStatsBuilder() {
-      
-      onChanged();
-      return getMemoryStatsFieldBuilder().getBuilder();
-    }
-    /**
-     * <code>.tensorflow.MemoryStats memory_stats = 12;</code>
-     */
-    public org.tensorflow.proto.framework.MemoryStatsOrBuilder getMemoryStatsOrBuilder() {
-      if (memoryStatsBuilder_ != null) {
-        return memoryStatsBuilder_.getMessageOrBuilder();
-      } else {
-        return memoryStats_ == null ?
-            org.tensorflow.proto.framework.MemoryStats.getDefaultInstance() : memoryStats_;
-      }
-    }
-    /**
-     * <code>.tensorflow.MemoryStats memory_stats = 12;</code>
-     */
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.MemoryStats, org.tensorflow.proto.framework.MemoryStats.Builder, org.tensorflow.proto.framework.MemoryStatsOrBuilder> 
-        getMemoryStatsFieldBuilder() {
-      if (memoryStatsBuilder_ == null) {
-        memoryStatsBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-            org.tensorflow.proto.framework.MemoryStats, org.tensorflow.proto.framework.MemoryStats.Builder, org.tensorflow.proto.framework.MemoryStatsOrBuilder>(
-                getMemoryStats(),
-                getParentForChildren(),
-                isClean());
-        memoryStats_ = null;
-      }
-      return memoryStatsBuilder_;
-    }
-
-    private long allStartNanos_ ;
-    /**
-     * <code>int64 all_start_nanos = 13;</code>
-     */
-    public long getAllStartNanos() {
-      return allStartNanos_;
-    }
-    /**
-     * <code>int64 all_start_nanos = 13;</code>
-     */
-    public Builder setAllStartNanos(long value) {
-      
-      allStartNanos_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>int64 all_start_nanos = 13;</code>
-     */
-    public Builder clearAllStartNanos() {
-      
-      allStartNanos_ = 0L;
-      onChanged();
-      return this;
-    }
-
-    private long opStartRelNanos_ ;
-    /**
-     * <code>int64 op_start_rel_nanos = 14;</code>
-     */
-    public long getOpStartRelNanos() {
-      return opStartRelNanos_;
-    }
-    /**
-     * <code>int64 op_start_rel_nanos = 14;</code>
-     */
-    public Builder setOpStartRelNanos(long value) {
-      
-      opStartRelNanos_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>int64 op_start_rel_nanos = 14;</code>
-     */
-    public Builder clearOpStartRelNanos() {
-      
-      opStartRelNanos_ = 0L;
-      onChanged();
-      return this;
-    }
-
-    private long opEndRelNanos_ ;
-    /**
-     * <code>int64 op_end_rel_nanos = 15;</code>
-     */
-    public long getOpEndRelNanos() {
-      return opEndRelNanos_;
-    }
-    /**
-     * <code>int64 op_end_rel_nanos = 15;</code>
-     */
-    public Builder setOpEndRelNanos(long value) {
-      
-      opEndRelNanos_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>int64 op_end_rel_nanos = 15;</code>
-     */
-    public Builder clearOpEndRelNanos() {
-      
-      opEndRelNanos_ = 0L;
-      onChanged();
-      return this;
-    }
-
-    private long allEndRelNanos_ ;
-    /**
-     * <code>int64 all_end_rel_nanos = 16;</code>
-     */
-    public long getAllEndRelNanos() {
-      return allEndRelNanos_;
-    }
-    /**
-     * <code>int64 all_end_rel_nanos = 16;</code>
-     */
-    public Builder setAllEndRelNanos(long value) {
-      
-      allEndRelNanos_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>int64 all_end_rel_nanos = 16;</code>
-     */
-    public Builder clearAllEndRelNanos() {
-      
-      allEndRelNanos_ = 0L;
-      onChanged();
-      return this;
-    }
-
-    private long scheduledNanos_ ;
-    /**
-     * <code>int64 scheduled_nanos = 17;</code>
-     */
-    public long getScheduledNanos() {
-      return scheduledNanos_;
-    }
-    /**
-     * <code>int64 scheduled_nanos = 17;</code>
-     */
-    public Builder setScheduledNanos(long value) {
-      
-      scheduledNanos_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>int64 scheduled_nanos = 17;</code>
-     */
-    public Builder clearScheduledNanos() {
-      
-      scheduledNanos_ = 0L;
-      onChanged();
-      return this;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.NodeExecStats)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.NodeExecStats)
-  private static final org.tensorflow.proto.framework.NodeExecStats DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.NodeExecStats();
-  }
-
-  public static org.tensorflow.proto.framework.NodeExecStats getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<NodeExecStats>
-      PARSER = new com.google.protobuf.AbstractParser<NodeExecStats>() {
-    @java.lang.Override
-    public NodeExecStats parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new NodeExecStats(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<NodeExecStats> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<NodeExecStats> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.framework.NodeExecStats getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NodeExecStatsOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NodeExecStatsOrBuilder.java
deleted file mode 100644
index 60be2e49270..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NodeExecStatsOrBuilder.java
+++ /dev/null
@@ -1,183 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/framework/step_stats.proto
-
-package org.tensorflow.proto.framework;
-
-public interface NodeExecStatsOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.NodeExecStats)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <pre>
-   * TODO(tucker): Use some more compact form of node identity than
-   * the full string name.  Either all processes should agree on a
-   * global id (cost_id?) for each node, or we should use a hash of
-   * the name.
-   * </pre>
-   *
-   * <code>string node_name = 1;</code>
-   */
-  java.lang.String getNodeName();
-  /**
-   * <pre>
-   * TODO(tucker): Use some more compact form of node identity than
-   * the full string name.  Either all processes should agree on a
-   * global id (cost_id?) for each node, or we should use a hash of
-   * the name.
-   * </pre>
-   *
-   * <code>string node_name = 1;</code>
-   */
-  com.google.protobuf.ByteString
-      getNodeNameBytes();
-
-  /**
-   * <code>int64 all_start_micros = 2;</code>
-   */
-  long getAllStartMicros();
-
-  /**
-   * <code>int64 op_start_rel_micros = 3;</code>
-   */
-  long getOpStartRelMicros();
-
-  /**
-   * <code>int64 op_end_rel_micros = 4;</code>
-   */
-  long getOpEndRelMicros();
-
-  /**
-   * <code>int64 all_end_rel_micros = 5;</code>
-   */
-  long getAllEndRelMicros();
-
-  /**
-   * <code>repeated .tensorflow.AllocatorMemoryUsed memory = 6;</code>
-   */
-  java.util.List<org.tensorflow.proto.framework.AllocatorMemoryUsed> 
-      getMemoryList();
-  /**
-   * <code>repeated .tensorflow.AllocatorMemoryUsed memory = 6;</code>
-   */
-  org.tensorflow.proto.framework.AllocatorMemoryUsed getMemory(int index);
-  /**
-   * <code>repeated .tensorflow.AllocatorMemoryUsed memory = 6;</code>
-   */
-  int getMemoryCount();
-  /**
-   * <code>repeated .tensorflow.AllocatorMemoryUsed memory = 6;</code>
-   */
-  java.util.List<? extends org.tensorflow.proto.framework.AllocatorMemoryUsedOrBuilder> 
-      getMemoryOrBuilderList();
-  /**
-   * <code>repeated .tensorflow.AllocatorMemoryUsed memory = 6;</code>
-   */
-  org.tensorflow.proto.framework.AllocatorMemoryUsedOrBuilder getMemoryOrBuilder(
-      int index);
-
-  /**
-   * <code>repeated .tensorflow.NodeOutput output = 7;</code>
-   */
-  java.util.List<org.tensorflow.proto.framework.NodeOutput> 
-      getOutputList();
-  /**
-   * <code>repeated .tensorflow.NodeOutput output = 7;</code>
-   */
-  org.tensorflow.proto.framework.NodeOutput getOutput(int index);
-  /**
-   * <code>repeated .tensorflow.NodeOutput output = 7;</code>
-   */
-  int getOutputCount();
-  /**
-   * <code>repeated .tensorflow.NodeOutput output = 7;</code>
-   */
-  java.util.List<? extends org.tensorflow.proto.framework.NodeOutputOrBuilder> 
-      getOutputOrBuilderList();
-  /**
-   * <code>repeated .tensorflow.NodeOutput output = 7;</code>
-   */
-  org.tensorflow.proto.framework.NodeOutputOrBuilder getOutputOrBuilder(
-      int index);
-
-  /**
-   * <code>string timeline_label = 8;</code>
-   */
-  java.lang.String getTimelineLabel();
-  /**
-   * <code>string timeline_label = 8;</code>
-   */
-  com.google.protobuf.ByteString
-      getTimelineLabelBytes();
-
-  /**
-   * <code>int64 scheduled_micros = 9;</code>
-   */
-  long getScheduledMicros();
-
-  /**
-   * <code>uint32 thread_id = 10;</code>
-   */
-  int getThreadId();
-
-  /**
-   * <code>repeated .tensorflow.AllocationDescription referenced_tensor = 11;</code>
-   */
-  java.util.List<org.tensorflow.proto.framework.AllocationDescription> 
-      getReferencedTensorList();
-  /**
-   * <code>repeated .tensorflow.AllocationDescription referenced_tensor = 11;</code>
-   */
-  org.tensorflow.proto.framework.AllocationDescription getReferencedTensor(int index);
-  /**
-   * <code>repeated .tensorflow.AllocationDescription referenced_tensor = 11;</code>
-   */
-  int getReferencedTensorCount();
-  /**
-   * <code>repeated .tensorflow.AllocationDescription referenced_tensor = 11;</code>
-   */
-  java.util.List<? extends org.tensorflow.proto.framework.AllocationDescriptionOrBuilder> 
-      getReferencedTensorOrBuilderList();
-  /**
-   * <code>repeated .tensorflow.AllocationDescription referenced_tensor = 11;</code>
-   */
-  org.tensorflow.proto.framework.AllocationDescriptionOrBuilder getReferencedTensorOrBuilder(
-      int index);
-
-  /**
-   * <code>.tensorflow.MemoryStats memory_stats = 12;</code>
-   */
-  boolean hasMemoryStats();
-  /**
-   * <code>.tensorflow.MemoryStats memory_stats = 12;</code>
-   */
-  org.tensorflow.proto.framework.MemoryStats getMemoryStats();
-  /**
-   * <code>.tensorflow.MemoryStats memory_stats = 12;</code>
-   */
-  org.tensorflow.proto.framework.MemoryStatsOrBuilder getMemoryStatsOrBuilder();
-
-  /**
-   * <code>int64 all_start_nanos = 13;</code>
-   */
-  long getAllStartNanos();
-
-  /**
-   * <code>int64 op_start_rel_nanos = 14;</code>
-   */
-  long getOpStartRelNanos();
-
-  /**
-   * <code>int64 op_end_rel_nanos = 15;</code>
-   */
-  long getOpEndRelNanos();
-
-  /**
-   * <code>int64 all_end_rel_nanos = 16;</code>
-   */
-  long getAllEndRelNanos();
-
-  /**
-   * <code>int64 scheduled_nanos = 17;</code>
-   */
-  long getScheduledNanos();
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NodeOutput.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NodeOutput.java
deleted file mode 100644
index c6e38aaec46..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NodeOutput.java
+++ /dev/null
@@ -1,665 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/framework/step_stats.proto
-
-package org.tensorflow.proto.framework;
-
-/**
- * <pre>
- * Output sizes recorded for a single execution of a graph node.
- * </pre>
- *
- * Protobuf type {@code tensorflow.NodeOutput}
- */
-public  final class NodeOutput extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.NodeOutput)
-    NodeOutputOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use NodeOutput.newBuilder() to construct.
-  private NodeOutput(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private NodeOutput() {
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new NodeOutput();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private NodeOutput(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 8: {
-
-            slot_ = input.readInt32();
-            break;
-          }
-          case 26: {
-            org.tensorflow.proto.framework.TensorDescription.Builder subBuilder = null;
-            if (tensorDescription_ != null) {
-              subBuilder = tensorDescription_.toBuilder();
-            }
-            tensorDescription_ = input.readMessage(org.tensorflow.proto.framework.TensorDescription.parser(), extensionRegistry);
-            if (subBuilder != null) {
-              subBuilder.mergeFrom(tensorDescription_);
-              tensorDescription_ = subBuilder.buildPartial();
-            }
-
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.framework.StepStatsProtos.internal_static_tensorflow_NodeOutput_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.framework.StepStatsProtos.internal_static_tensorflow_NodeOutput_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.framework.NodeOutput.class, org.tensorflow.proto.framework.NodeOutput.Builder.class);
-  }
-
-  public static final int SLOT_FIELD_NUMBER = 1;
-  private int slot_;
-  /**
-   * <code>int32 slot = 1;</code>
-   */
-  public int getSlot() {
-    return slot_;
-  }
-
-  public static final int TENSOR_DESCRIPTION_FIELD_NUMBER = 3;
-  private org.tensorflow.proto.framework.TensorDescription tensorDescription_;
-  /**
-   * <code>.tensorflow.TensorDescription tensor_description = 3;</code>
-   */
-  public boolean hasTensorDescription() {
-    return tensorDescription_ != null;
-  }
-  /**
-   * <code>.tensorflow.TensorDescription tensor_description = 3;</code>
-   */
-  public org.tensorflow.proto.framework.TensorDescription getTensorDescription() {
-    return tensorDescription_ == null ? org.tensorflow.proto.framework.TensorDescription.getDefaultInstance() : tensorDescription_;
-  }
-  /**
-   * <code>.tensorflow.TensorDescription tensor_description = 3;</code>
-   */
-  public org.tensorflow.proto.framework.TensorDescriptionOrBuilder getTensorDescriptionOrBuilder() {
-    return getTensorDescription();
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    if (slot_ != 0) {
-      output.writeInt32(1, slot_);
-    }
-    if (tensorDescription_ != null) {
-      output.writeMessage(3, getTensorDescription());
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (slot_ != 0) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt32Size(1, slot_);
-    }
-    if (tensorDescription_ != null) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(3, getTensorDescription());
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.framework.NodeOutput)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.framework.NodeOutput other = (org.tensorflow.proto.framework.NodeOutput) obj;
-
-    if (getSlot()
-        != other.getSlot()) return false;
-    if (hasTensorDescription() != other.hasTensorDescription()) return false;
-    if (hasTensorDescription()) {
-      if (!getTensorDescription()
-          .equals(other.getTensorDescription())) return false;
-    }
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    hash = (37 * hash) + SLOT_FIELD_NUMBER;
-    hash = (53 * hash) + getSlot();
-    if (hasTensorDescription()) {
-      hash = (37 * hash) + TENSOR_DESCRIPTION_FIELD_NUMBER;
-      hash = (53 * hash) + getTensorDescription().hashCode();
-    }
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.framework.NodeOutput parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.NodeOutput parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.NodeOutput parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.NodeOutput parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.NodeOutput parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.NodeOutput parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.NodeOutput parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.NodeOutput parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.NodeOutput parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.NodeOutput parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.NodeOutput parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.NodeOutput parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.framework.NodeOutput prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * <pre>
-   * Output sizes recorded for a single execution of a graph node.
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.NodeOutput}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.NodeOutput)
-      org.tensorflow.proto.framework.NodeOutputOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.StepStatsProtos.internal_static_tensorflow_NodeOutput_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.StepStatsProtos.internal_static_tensorflow_NodeOutput_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.NodeOutput.class, org.tensorflow.proto.framework.NodeOutput.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.framework.NodeOutput.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      slot_ = 0;
-
-      if (tensorDescriptionBuilder_ == null) {
-        tensorDescription_ = null;
-      } else {
-        tensorDescription_ = null;
-        tensorDescriptionBuilder_ = null;
-      }
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.framework.StepStatsProtos.internal_static_tensorflow_NodeOutput_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.NodeOutput getDefaultInstanceForType() {
-      return org.tensorflow.proto.framework.NodeOutput.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.NodeOutput build() {
-      org.tensorflow.proto.framework.NodeOutput result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.NodeOutput buildPartial() {
-      org.tensorflow.proto.framework.NodeOutput result = new org.tensorflow.proto.framework.NodeOutput(this);
-      result.slot_ = slot_;
-      if (tensorDescriptionBuilder_ == null) {
-        result.tensorDescription_ = tensorDescription_;
-      } else {
-        result.tensorDescription_ = tensorDescriptionBuilder_.build();
-      }
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.framework.NodeOutput) {
-        return mergeFrom((org.tensorflow.proto.framework.NodeOutput)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.framework.NodeOutput other) {
-      if (other == org.tensorflow.proto.framework.NodeOutput.getDefaultInstance()) return this;
-      if (other.getSlot() != 0) {
-        setSlot(other.getSlot());
-      }
-      if (other.hasTensorDescription()) {
-        mergeTensorDescription(other.getTensorDescription());
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.framework.NodeOutput parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.framework.NodeOutput) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-
-    private int slot_ ;
-    /**
-     * <code>int32 slot = 1;</code>
-     */
-    public int getSlot() {
-      return slot_;
-    }
-    /**
-     * <code>int32 slot = 1;</code>
-     */
-    public Builder setSlot(int value) {
-      
-      slot_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>int32 slot = 1;</code>
-     */
-    public Builder clearSlot() {
-      
-      slot_ = 0;
-      onChanged();
-      return this;
-    }
-
-    private org.tensorflow.proto.framework.TensorDescription tensorDescription_;
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.TensorDescription, org.tensorflow.proto.framework.TensorDescription.Builder, org.tensorflow.proto.framework.TensorDescriptionOrBuilder> tensorDescriptionBuilder_;
-    /**
-     * <code>.tensorflow.TensorDescription tensor_description = 3;</code>
-     */
-    public boolean hasTensorDescription() {
-      return tensorDescriptionBuilder_ != null || tensorDescription_ != null;
-    }
-    /**
-     * <code>.tensorflow.TensorDescription tensor_description = 3;</code>
-     */
-    public org.tensorflow.proto.framework.TensorDescription getTensorDescription() {
-      if (tensorDescriptionBuilder_ == null) {
-        return tensorDescription_ == null ? org.tensorflow.proto.framework.TensorDescription.getDefaultInstance() : tensorDescription_;
-      } else {
-        return tensorDescriptionBuilder_.getMessage();
-      }
-    }
-    /**
-     * <code>.tensorflow.TensorDescription tensor_description = 3;</code>
-     */
-    public Builder setTensorDescription(org.tensorflow.proto.framework.TensorDescription value) {
-      if (tensorDescriptionBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        tensorDescription_ = value;
-        onChanged();
-      } else {
-        tensorDescriptionBuilder_.setMessage(value);
-      }
-
-      return this;
-    }
-    /**
-     * <code>.tensorflow.TensorDescription tensor_description = 3;</code>
-     */
-    public Builder setTensorDescription(
-        org.tensorflow.proto.framework.TensorDescription.Builder builderForValue) {
-      if (tensorDescriptionBuilder_ == null) {
-        tensorDescription_ = builderForValue.build();
-        onChanged();
-      } else {
-        tensorDescriptionBuilder_.setMessage(builderForValue.build());
-      }
-
-      return this;
-    }
-    /**
-     * <code>.tensorflow.TensorDescription tensor_description = 3;</code>
-     */
-    public Builder mergeTensorDescription(org.tensorflow.proto.framework.TensorDescription value) {
-      if (tensorDescriptionBuilder_ == null) {
-        if (tensorDescription_ != null) {
-          tensorDescription_ =
-            org.tensorflow.proto.framework.TensorDescription.newBuilder(tensorDescription_).mergeFrom(value).buildPartial();
-        } else {
-          tensorDescription_ = value;
-        }
-        onChanged();
-      } else {
-        tensorDescriptionBuilder_.mergeFrom(value);
-      }
-
-      return this;
-    }
-    /**
-     * <code>.tensorflow.TensorDescription tensor_description = 3;</code>
-     */
-    public Builder clearTensorDescription() {
-      if (tensorDescriptionBuilder_ == null) {
-        tensorDescription_ = null;
-        onChanged();
-      } else {
-        tensorDescription_ = null;
-        tensorDescriptionBuilder_ = null;
-      }
-
-      return this;
-    }
-    /**
-     * <code>.tensorflow.TensorDescription tensor_description = 3;</code>
-     */
-    public org.tensorflow.proto.framework.TensorDescription.Builder getTensorDescriptionBuilder() {
-      
-      onChanged();
-      return getTensorDescriptionFieldBuilder().getBuilder();
-    }
-    /**
-     * <code>.tensorflow.TensorDescription tensor_description = 3;</code>
-     */
-    public org.tensorflow.proto.framework.TensorDescriptionOrBuilder getTensorDescriptionOrBuilder() {
-      if (tensorDescriptionBuilder_ != null) {
-        return tensorDescriptionBuilder_.getMessageOrBuilder();
-      } else {
-        return tensorDescription_ == null ?
-            org.tensorflow.proto.framework.TensorDescription.getDefaultInstance() : tensorDescription_;
-      }
-    }
-    /**
-     * <code>.tensorflow.TensorDescription tensor_description = 3;</code>
-     */
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.TensorDescription, org.tensorflow.proto.framework.TensorDescription.Builder, org.tensorflow.proto.framework.TensorDescriptionOrBuilder> 
-        getTensorDescriptionFieldBuilder() {
-      if (tensorDescriptionBuilder_ == null) {
-        tensorDescriptionBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-            org.tensorflow.proto.framework.TensorDescription, org.tensorflow.proto.framework.TensorDescription.Builder, org.tensorflow.proto.framework.TensorDescriptionOrBuilder>(
-                getTensorDescription(),
-                getParentForChildren(),
-                isClean());
-        tensorDescription_ = null;
-      }
-      return tensorDescriptionBuilder_;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.NodeOutput)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.NodeOutput)
-  private static final org.tensorflow.proto.framework.NodeOutput DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.NodeOutput();
-  }
-
-  public static org.tensorflow.proto.framework.NodeOutput getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<NodeOutput>
-      PARSER = new com.google.protobuf.AbstractParser<NodeOutput>() {
-    @java.lang.Override
-    public NodeOutput parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new NodeOutput(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<NodeOutput> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<NodeOutput> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.framework.NodeOutput getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NodeOutputOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NodeOutputOrBuilder.java
deleted file mode 100644
index 940f8defed7..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NodeOutputOrBuilder.java
+++ /dev/null
@@ -1,27 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/framework/step_stats.proto
-
-package org.tensorflow.proto.framework;
-
-public interface NodeOutputOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.NodeOutput)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <code>int32 slot = 1;</code>
-   */
-  int getSlot();
-
-  /**
-   * <code>.tensorflow.TensorDescription tensor_description = 3;</code>
-   */
-  boolean hasTensorDescription();
-  /**
-   * <code>.tensorflow.TensorDescription tensor_description = 3;</code>
-   */
-  org.tensorflow.proto.framework.TensorDescription getTensorDescription();
-  /**
-   * <code>.tensorflow.TensorDescription tensor_description = 3;</code>
-   */
-  org.tensorflow.proto.framework.TensorDescriptionOrBuilder getTensorDescriptionOrBuilder();
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NodeProto.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NodeProto.java
deleted file mode 100644
index ef4be3729ea..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NodeProto.java
+++ /dev/null
@@ -1,84 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/framework/node_def.proto
-
-package org.tensorflow.proto.framework;
-
-public final class NodeProto {
-  private NodeProto() {}
-  public static void registerAllExtensions(
-      com.google.protobuf.ExtensionRegistryLite registry) {
-  }
-
-  public static void registerAllExtensions(
-      com.google.protobuf.ExtensionRegistry registry) {
-    registerAllExtensions(
-        (com.google.protobuf.ExtensionRegistryLite) registry);
-  }
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_NodeDef_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_NodeDef_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_NodeDef_AttrEntry_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_NodeDef_AttrEntry_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_NodeDef_ExperimentalDebugInfo_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_NodeDef_ExperimentalDebugInfo_fieldAccessorTable;
-
-  public static com.google.protobuf.Descriptors.FileDescriptor
-      getDescriptor() {
-    return descriptor;
-  }
-  private static  com.google.protobuf.Descriptors.FileDescriptor
-      descriptor;
-  static {
-    java.lang.String[] descriptorData = {
-      "\n(tensorflow/core/framework/node_def.pro" +
-      "to\022\ntensorflow\032*tensorflow/core/framewor" +
-      "k/attr_value.proto\"\322\002\n\007NodeDef\022\014\n\004name\030\001" +
-      " \001(\t\022\n\n\002op\030\002 \001(\t\022\r\n\005input\030\003 \003(\t\022\016\n\006devic" +
-      "e\030\004 \001(\t\022+\n\004attr\030\005 \003(\0132\035.tensorflow.NodeD" +
-      "ef.AttrEntry\022J\n\027experimental_debug_info\030" +
-      "\006 \001(\0132).tensorflow.NodeDef.ExperimentalD" +
-      "ebugInfo\032B\n\tAttrEntry\022\013\n\003key\030\001 \001(\t\022$\n\005va" +
-      "lue\030\002 \001(\0132\025.tensorflow.AttrValue:\0028\001\032Q\n\025" +
-      "ExperimentalDebugInfo\022\033\n\023original_node_n" +
-      "ames\030\001 \003(\t\022\033\n\023original_func_names\030\002 \003(\tB" +
-      "\201\001\n\036org.tensorflow.proto.frameworkB\tNode" +
-      "ProtoP\001ZOgithub.com/tensorflow/tensorflo" +
-      "w/tensorflow/go/core/framework/node_def_" +
-      "go_proto\370\001\001b\006proto3"
-    };
-    descriptor = com.google.protobuf.Descriptors.FileDescriptor
-      .internalBuildGeneratedFileFrom(descriptorData,
-        new com.google.protobuf.Descriptors.FileDescriptor[] {
-          org.tensorflow.proto.framework.AttrValueProtos.getDescriptor(),
-        });
-    internal_static_tensorflow_NodeDef_descriptor =
-      getDescriptor().getMessageTypes().get(0);
-    internal_static_tensorflow_NodeDef_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_NodeDef_descriptor,
-        new java.lang.String[] { "Name", "Op", "Input", "Device", "Attr", "ExperimentalDebugInfo", });
-    internal_static_tensorflow_NodeDef_AttrEntry_descriptor =
-      internal_static_tensorflow_NodeDef_descriptor.getNestedTypes().get(0);
-    internal_static_tensorflow_NodeDef_AttrEntry_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_NodeDef_AttrEntry_descriptor,
-        new java.lang.String[] { "Key", "Value", });
-    internal_static_tensorflow_NodeDef_ExperimentalDebugInfo_descriptor =
-      internal_static_tensorflow_NodeDef_descriptor.getNestedTypes().get(1);
-    internal_static_tensorflow_NodeDef_ExperimentalDebugInfo_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_NodeDef_ExperimentalDebugInfo_descriptor,
-        new java.lang.String[] { "OriginalNodeNames", "OriginalFuncNames", });
-    org.tensorflow.proto.framework.AttrValueProtos.getDescriptor();
-  }
-
-  // @@protoc_insertion_point(outer_class_scope)
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NoneValue.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NoneValue.java
deleted file mode 100644
index 7aaba25f2d2..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NoneValue.java
+++ /dev/null
@@ -1,427 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/struct.proto
-
-package org.tensorflow.proto.framework;
-
-/**
- * <pre>
- * Represents None.
- * </pre>
- *
- * Protobuf type {@code tensorflow.NoneValue}
- */
-public  final class NoneValue extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.NoneValue)
-    NoneValueOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use NoneValue.newBuilder() to construct.
-  private NoneValue(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private NoneValue() {
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new NoneValue();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private NoneValue(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.framework.StructProtos.internal_static_tensorflow_NoneValue_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.framework.StructProtos.internal_static_tensorflow_NoneValue_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.framework.NoneValue.class, org.tensorflow.proto.framework.NoneValue.Builder.class);
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.framework.NoneValue)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.framework.NoneValue other = (org.tensorflow.proto.framework.NoneValue) obj;
-
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.framework.NoneValue parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.NoneValue parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.NoneValue parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.NoneValue parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.NoneValue parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.NoneValue parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.NoneValue parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.NoneValue parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.NoneValue parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.NoneValue parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.NoneValue parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.NoneValue parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.framework.NoneValue prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * <pre>
-   * Represents None.
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.NoneValue}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.NoneValue)
-      org.tensorflow.proto.framework.NoneValueOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.StructProtos.internal_static_tensorflow_NoneValue_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.StructProtos.internal_static_tensorflow_NoneValue_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.NoneValue.class, org.tensorflow.proto.framework.NoneValue.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.framework.NoneValue.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.framework.StructProtos.internal_static_tensorflow_NoneValue_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.NoneValue getDefaultInstanceForType() {
-      return org.tensorflow.proto.framework.NoneValue.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.NoneValue build() {
-      org.tensorflow.proto.framework.NoneValue result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.NoneValue buildPartial() {
-      org.tensorflow.proto.framework.NoneValue result = new org.tensorflow.proto.framework.NoneValue(this);
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.framework.NoneValue) {
-        return mergeFrom((org.tensorflow.proto.framework.NoneValue)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.framework.NoneValue other) {
-      if (other == org.tensorflow.proto.framework.NoneValue.getDefaultInstance()) return this;
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.framework.NoneValue parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.framework.NoneValue) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.NoneValue)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.NoneValue)
-  private static final org.tensorflow.proto.framework.NoneValue DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.NoneValue();
-  }
-
-  public static org.tensorflow.proto.framework.NoneValue getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<NoneValue>
-      PARSER = new com.google.protobuf.AbstractParser<NoneValue>() {
-    @java.lang.Override
-    public NoneValue parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new NoneValue(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<NoneValue> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<NoneValue> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.framework.NoneValue getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NoneValueOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NoneValueOrBuilder.java
deleted file mode 100644
index 3d720d0c269..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NoneValueOrBuilder.java
+++ /dev/null
@@ -1,9 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/struct.proto
-
-package org.tensorflow.proto.framework;
-
-public interface NoneValueOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.NoneValue)
-    com.google.protobuf.MessageOrBuilder {
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/OpDef.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/OpDef.java
deleted file mode 100644
index 80839f48dcc..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/OpDef.java
+++ /dev/null
@@ -1,6284 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/framework/op_def.proto
-
-package org.tensorflow.proto.framework;
-
-/**
- * <pre>
- * Defines an operation. A NodeDef in a GraphDef specifies an Op by
- * using the "op" field which should match the name of a OpDef.
- * LINT.IfChange
- * </pre>
- *
- * Protobuf type {@code tensorflow.OpDef}
- */
-public  final class OpDef extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.OpDef)
-    OpDefOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use OpDef.newBuilder() to construct.
-  private OpDef(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private OpDef() {
-    name_ = "";
-    inputArg_ = java.util.Collections.emptyList();
-    outputArg_ = java.util.Collections.emptyList();
-    controlOutput_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-    attr_ = java.util.Collections.emptyList();
-    summary_ = "";
-    description_ = "";
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new OpDef();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private OpDef(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    int mutable_bitField0_ = 0;
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 10: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            name_ = s;
-            break;
-          }
-          case 18: {
-            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
-              inputArg_ = new java.util.ArrayList<org.tensorflow.proto.framework.OpDef.ArgDef>();
-              mutable_bitField0_ |= 0x00000001;
-            }
-            inputArg_.add(
-                input.readMessage(org.tensorflow.proto.framework.OpDef.ArgDef.parser(), extensionRegistry));
-            break;
-          }
-          case 26: {
-            if (!((mutable_bitField0_ & 0x00000002) != 0)) {
-              outputArg_ = new java.util.ArrayList<org.tensorflow.proto.framework.OpDef.ArgDef>();
-              mutable_bitField0_ |= 0x00000002;
-            }
-            outputArg_.add(
-                input.readMessage(org.tensorflow.proto.framework.OpDef.ArgDef.parser(), extensionRegistry));
-            break;
-          }
-          case 34: {
-            if (!((mutable_bitField0_ & 0x00000008) != 0)) {
-              attr_ = new java.util.ArrayList<org.tensorflow.proto.framework.OpDef.AttrDef>();
-              mutable_bitField0_ |= 0x00000008;
-            }
-            attr_.add(
-                input.readMessage(org.tensorflow.proto.framework.OpDef.AttrDef.parser(), extensionRegistry));
-            break;
-          }
-          case 42: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            summary_ = s;
-            break;
-          }
-          case 50: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            description_ = s;
-            break;
-          }
-          case 66: {
-            org.tensorflow.proto.framework.OpDeprecation.Builder subBuilder = null;
-            if (deprecation_ != null) {
-              subBuilder = deprecation_.toBuilder();
-            }
-            deprecation_ = input.readMessage(org.tensorflow.proto.framework.OpDeprecation.parser(), extensionRegistry);
-            if (subBuilder != null) {
-              subBuilder.mergeFrom(deprecation_);
-              deprecation_ = subBuilder.buildPartial();
-            }
-
-            break;
-          }
-          case 128: {
-
-            isAggregate_ = input.readBool();
-            break;
-          }
-          case 136: {
-
-            isStateful_ = input.readBool();
-            break;
-          }
-          case 144: {
-
-            isCommutative_ = input.readBool();
-            break;
-          }
-          case 152: {
-
-            allowsUninitializedInput_ = input.readBool();
-            break;
-          }
-          case 162: {
-            java.lang.String s = input.readStringRequireUtf8();
-            if (!((mutable_bitField0_ & 0x00000004) != 0)) {
-              controlOutput_ = new com.google.protobuf.LazyStringArrayList();
-              mutable_bitField0_ |= 0x00000004;
-            }
-            controlOutput_.add(s);
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      if (((mutable_bitField0_ & 0x00000001) != 0)) {
-        inputArg_ = java.util.Collections.unmodifiableList(inputArg_);
-      }
-      if (((mutable_bitField0_ & 0x00000002) != 0)) {
-        outputArg_ = java.util.Collections.unmodifiableList(outputArg_);
-      }
-      if (((mutable_bitField0_ & 0x00000008) != 0)) {
-        attr_ = java.util.Collections.unmodifiableList(attr_);
-      }
-      if (((mutable_bitField0_ & 0x00000004) != 0)) {
-        controlOutput_ = controlOutput_.getUnmodifiableView();
-      }
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.framework.OpDefProtos.internal_static_tensorflow_OpDef_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.framework.OpDefProtos.internal_static_tensorflow_OpDef_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.framework.OpDef.class, org.tensorflow.proto.framework.OpDef.Builder.class);
-  }
-
-  public interface ArgDefOrBuilder extends
-      // @@protoc_insertion_point(interface_extends:tensorflow.OpDef.ArgDef)
-      com.google.protobuf.MessageOrBuilder {
-
-    /**
-     * <pre>
-     * Name for the input/output.  Should match the regexp "[a-z][a-z0-9_]*".
-     * </pre>
-     *
-     * <code>string name = 1;</code>
-     */
-    java.lang.String getName();
-    /**
-     * <pre>
-     * Name for the input/output.  Should match the regexp "[a-z][a-z0-9_]*".
-     * </pre>
-     *
-     * <code>string name = 1;</code>
-     */
-    com.google.protobuf.ByteString
-        getNameBytes();
-
-    /**
-     * <pre>
-     * Human readable description.
-     * </pre>
-     *
-     * <code>string description = 2;</code>
-     */
-    java.lang.String getDescription();
-    /**
-     * <pre>
-     * Human readable description.
-     * </pre>
-     *
-     * <code>string description = 2;</code>
-     */
-    com.google.protobuf.ByteString
-        getDescriptionBytes();
-
-    /**
-     * <pre>
-     * Describes the type of one or more tensors that are accepted/produced
-     * by this input/output arg.  The only legal combinations are:
-     * * For a single tensor: either the "type" field is set or the
-     *   "type_attr" field is set to the name of an attr with type "type".
-     * * For a sequence of tensors with the same type: the "number_attr"
-     *   field will be set to the name of an attr with type "int", and
-     *   either the "type" or "type_attr" field will be set as for
-     *   single tensors.
-     * * For a sequence of tensors, the "type_list_attr" field will be set
-     *   to the name of an attr with type "list(type)".
-     * </pre>
-     *
-     * <code>.tensorflow.DataType type = 3;</code>
-     */
-    int getTypeValue();
-    /**
-     * <pre>
-     * Describes the type of one or more tensors that are accepted/produced
-     * by this input/output arg.  The only legal combinations are:
-     * * For a single tensor: either the "type" field is set or the
-     *   "type_attr" field is set to the name of an attr with type "type".
-     * * For a sequence of tensors with the same type: the "number_attr"
-     *   field will be set to the name of an attr with type "int", and
-     *   either the "type" or "type_attr" field will be set as for
-     *   single tensors.
-     * * For a sequence of tensors, the "type_list_attr" field will be set
-     *   to the name of an attr with type "list(type)".
-     * </pre>
-     *
-     * <code>.tensorflow.DataType type = 3;</code>
-     */
-    org.tensorflow.proto.framework.DataType getType();
-
-    /**
-     * <pre>
-     * if specified, attr must have type "type"
-     * </pre>
-     *
-     * <code>string type_attr = 4;</code>
-     */
-    java.lang.String getTypeAttr();
-    /**
-     * <pre>
-     * if specified, attr must have type "type"
-     * </pre>
-     *
-     * <code>string type_attr = 4;</code>
-     */
-    com.google.protobuf.ByteString
-        getTypeAttrBytes();
-
-    /**
-     * <pre>
-     * if specified, attr must have type "int"
-     * </pre>
-     *
-     * <code>string number_attr = 5;</code>
-     */
-    java.lang.String getNumberAttr();
-    /**
-     * <pre>
-     * if specified, attr must have type "int"
-     * </pre>
-     *
-     * <code>string number_attr = 5;</code>
-     */
-    com.google.protobuf.ByteString
-        getNumberAttrBytes();
-
-    /**
-     * <pre>
-     * If specified, attr must have type "list(type)", and none of
-     * type, type_attr, and number_attr may be specified.
-     * </pre>
-     *
-     * <code>string type_list_attr = 6;</code>
-     */
-    java.lang.String getTypeListAttr();
-    /**
-     * <pre>
-     * If specified, attr must have type "list(type)", and none of
-     * type, type_attr, and number_attr may be specified.
-     * </pre>
-     *
-     * <code>string type_list_attr = 6;</code>
-     */
-    com.google.protobuf.ByteString
-        getTypeListAttrBytes();
-
-    /**
-     * <pre>
-     * For inputs: if true, the inputs are required to be refs.
-     *   By default, inputs can be either refs or non-refs.
-     * For outputs: if true, outputs are refs, otherwise they are not.
-     * </pre>
-     *
-     * <code>bool is_ref = 16;</code>
-     */
-    boolean getIsRef();
-  }
-  /**
-   * <pre>
-   * For describing inputs and outputs.
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.OpDef.ArgDef}
-   */
-  public  static final class ArgDef extends
-      com.google.protobuf.GeneratedMessageV3 implements
-      // @@protoc_insertion_point(message_implements:tensorflow.OpDef.ArgDef)
-      ArgDefOrBuilder {
-  private static final long serialVersionUID = 0L;
-    // Use ArgDef.newBuilder() to construct.
-    private ArgDef(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-      super(builder);
-    }
-    private ArgDef() {
-      name_ = "";
-      description_ = "";
-      type_ = 0;
-      typeAttr_ = "";
-      numberAttr_ = "";
-      typeListAttr_ = "";
-    }
-
-    @java.lang.Override
-    @SuppressWarnings({"unused"})
-    protected java.lang.Object newInstance(
-        UnusedPrivateParameter unused) {
-      return new ArgDef();
-    }
-
-    @java.lang.Override
-    public final com.google.protobuf.UnknownFieldSet
-    getUnknownFields() {
-      return this.unknownFields;
-    }
-    private ArgDef(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      this();
-      if (extensionRegistry == null) {
-        throw new java.lang.NullPointerException();
-      }
-      com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-          com.google.protobuf.UnknownFieldSet.newBuilder();
-      try {
-        boolean done = false;
-        while (!done) {
-          int tag = input.readTag();
-          switch (tag) {
-            case 0:
-              done = true;
-              break;
-            case 10: {
-              java.lang.String s = input.readStringRequireUtf8();
-
-              name_ = s;
-              break;
-            }
-            case 18: {
-              java.lang.String s = input.readStringRequireUtf8();
-
-              description_ = s;
-              break;
-            }
-            case 24: {
-              int rawValue = input.readEnum();
-
-              type_ = rawValue;
-              break;
-            }
-            case 34: {
-              java.lang.String s = input.readStringRequireUtf8();
-
-              typeAttr_ = s;
-              break;
-            }
-            case 42: {
-              java.lang.String s = input.readStringRequireUtf8();
-
-              numberAttr_ = s;
-              break;
-            }
-            case 50: {
-              java.lang.String s = input.readStringRequireUtf8();
-
-              typeListAttr_ = s;
-              break;
-            }
-            case 128: {
-
-              isRef_ = input.readBool();
-              break;
-            }
-            default: {
-              if (!parseUnknownField(
-                  input, unknownFields, extensionRegistry, tag)) {
-                done = true;
-              }
-              break;
-            }
-          }
-        }
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        throw e.setUnfinishedMessage(this);
-      } catch (java.io.IOException e) {
-        throw new com.google.protobuf.InvalidProtocolBufferException(
-            e).setUnfinishedMessage(this);
-      } finally {
-        this.unknownFields = unknownFields.build();
-        makeExtensionsImmutable();
-      }
-    }
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.OpDefProtos.internal_static_tensorflow_OpDef_ArgDef_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.OpDefProtos.internal_static_tensorflow_OpDef_ArgDef_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.OpDef.ArgDef.class, org.tensorflow.proto.framework.OpDef.ArgDef.Builder.class);
-    }
-
-    public static final int NAME_FIELD_NUMBER = 1;
-    private volatile java.lang.Object name_;
-    /**
-     * <pre>
-     * Name for the input/output.  Should match the regexp "[a-z][a-z0-9_]*".
-     * </pre>
-     *
-     * <code>string name = 1;</code>
-     */
-    public java.lang.String getName() {
-      java.lang.Object ref = name_;
-      if (ref instanceof java.lang.String) {
-        return (java.lang.String) ref;
-      } else {
-        com.google.protobuf.ByteString bs = 
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        name_ = s;
-        return s;
-      }
-    }
-    /**
-     * <pre>
-     * Name for the input/output.  Should match the regexp "[a-z][a-z0-9_]*".
-     * </pre>
-     *
-     * <code>string name = 1;</code>
-     */
-    public com.google.protobuf.ByteString
-        getNameBytes() {
-      java.lang.Object ref = name_;
-      if (ref instanceof java.lang.String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        name_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-
-    public static final int DESCRIPTION_FIELD_NUMBER = 2;
-    private volatile java.lang.Object description_;
-    /**
-     * <pre>
-     * Human readable description.
-     * </pre>
-     *
-     * <code>string description = 2;</code>
-     */
-    public java.lang.String getDescription() {
-      java.lang.Object ref = description_;
-      if (ref instanceof java.lang.String) {
-        return (java.lang.String) ref;
-      } else {
-        com.google.protobuf.ByteString bs = 
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        description_ = s;
-        return s;
-      }
-    }
-    /**
-     * <pre>
-     * Human readable description.
-     * </pre>
-     *
-     * <code>string description = 2;</code>
-     */
-    public com.google.protobuf.ByteString
-        getDescriptionBytes() {
-      java.lang.Object ref = description_;
-      if (ref instanceof java.lang.String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        description_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-
-    public static final int TYPE_FIELD_NUMBER = 3;
-    private int type_;
-    /**
-     * <pre>
-     * Describes the type of one or more tensors that are accepted/produced
-     * by this input/output arg.  The only legal combinations are:
-     * * For a single tensor: either the "type" field is set or the
-     *   "type_attr" field is set to the name of an attr with type "type".
-     * * For a sequence of tensors with the same type: the "number_attr"
-     *   field will be set to the name of an attr with type "int", and
-     *   either the "type" or "type_attr" field will be set as for
-     *   single tensors.
-     * * For a sequence of tensors, the "type_list_attr" field will be set
-     *   to the name of an attr with type "list(type)".
-     * </pre>
-     *
-     * <code>.tensorflow.DataType type = 3;</code>
-     */
-    public int getTypeValue() {
-      return type_;
-    }
-    /**
-     * <pre>
-     * Describes the type of one or more tensors that are accepted/produced
-     * by this input/output arg.  The only legal combinations are:
-     * * For a single tensor: either the "type" field is set or the
-     *   "type_attr" field is set to the name of an attr with type "type".
-     * * For a sequence of tensors with the same type: the "number_attr"
-     *   field will be set to the name of an attr with type "int", and
-     *   either the "type" or "type_attr" field will be set as for
-     *   single tensors.
-     * * For a sequence of tensors, the "type_list_attr" field will be set
-     *   to the name of an attr with type "list(type)".
-     * </pre>
-     *
-     * <code>.tensorflow.DataType type = 3;</code>
-     */
-    public org.tensorflow.proto.framework.DataType getType() {
-      @SuppressWarnings("deprecation")
-      org.tensorflow.proto.framework.DataType result = org.tensorflow.proto.framework.DataType.valueOf(type_);
-      return result == null ? org.tensorflow.proto.framework.DataType.UNRECOGNIZED : result;
-    }
-
-    public static final int TYPE_ATTR_FIELD_NUMBER = 4;
-    private volatile java.lang.Object typeAttr_;
-    /**
-     * <pre>
-     * if specified, attr must have type "type"
-     * </pre>
-     *
-     * <code>string type_attr = 4;</code>
-     */
-    public java.lang.String getTypeAttr() {
-      java.lang.Object ref = typeAttr_;
-      if (ref instanceof java.lang.String) {
-        return (java.lang.String) ref;
-      } else {
-        com.google.protobuf.ByteString bs = 
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        typeAttr_ = s;
-        return s;
-      }
-    }
-    /**
-     * <pre>
-     * if specified, attr must have type "type"
-     * </pre>
-     *
-     * <code>string type_attr = 4;</code>
-     */
-    public com.google.protobuf.ByteString
-        getTypeAttrBytes() {
-      java.lang.Object ref = typeAttr_;
-      if (ref instanceof java.lang.String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        typeAttr_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-
-    public static final int NUMBER_ATTR_FIELD_NUMBER = 5;
-    private volatile java.lang.Object numberAttr_;
-    /**
-     * <pre>
-     * if specified, attr must have type "int"
-     * </pre>
-     *
-     * <code>string number_attr = 5;</code>
-     */
-    public java.lang.String getNumberAttr() {
-      java.lang.Object ref = numberAttr_;
-      if (ref instanceof java.lang.String) {
-        return (java.lang.String) ref;
-      } else {
-        com.google.protobuf.ByteString bs = 
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        numberAttr_ = s;
-        return s;
-      }
-    }
-    /**
-     * <pre>
-     * if specified, attr must have type "int"
-     * </pre>
-     *
-     * <code>string number_attr = 5;</code>
-     */
-    public com.google.protobuf.ByteString
-        getNumberAttrBytes() {
-      java.lang.Object ref = numberAttr_;
-      if (ref instanceof java.lang.String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        numberAttr_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-
-    public static final int TYPE_LIST_ATTR_FIELD_NUMBER = 6;
-    private volatile java.lang.Object typeListAttr_;
-    /**
-     * <pre>
-     * If specified, attr must have type "list(type)", and none of
-     * type, type_attr, and number_attr may be specified.
-     * </pre>
-     *
-     * <code>string type_list_attr = 6;</code>
-     */
-    public java.lang.String getTypeListAttr() {
-      java.lang.Object ref = typeListAttr_;
-      if (ref instanceof java.lang.String) {
-        return (java.lang.String) ref;
-      } else {
-        com.google.protobuf.ByteString bs = 
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        typeListAttr_ = s;
-        return s;
-      }
-    }
-    /**
-     * <pre>
-     * If specified, attr must have type "list(type)", and none of
-     * type, type_attr, and number_attr may be specified.
-     * </pre>
-     *
-     * <code>string type_list_attr = 6;</code>
-     */
-    public com.google.protobuf.ByteString
-        getTypeListAttrBytes() {
-      java.lang.Object ref = typeListAttr_;
-      if (ref instanceof java.lang.String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        typeListAttr_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-
-    public static final int IS_REF_FIELD_NUMBER = 16;
-    private boolean isRef_;
-    /**
-     * <pre>
-     * For inputs: if true, the inputs are required to be refs.
-     *   By default, inputs can be either refs or non-refs.
-     * For outputs: if true, outputs are refs, otherwise they are not.
-     * </pre>
-     *
-     * <code>bool is_ref = 16;</code>
-     */
-    public boolean getIsRef() {
-      return isRef_;
-    }
-
-    private byte memoizedIsInitialized = -1;
-    @java.lang.Override
-    public final boolean isInitialized() {
-      byte isInitialized = memoizedIsInitialized;
-      if (isInitialized == 1) return true;
-      if (isInitialized == 0) return false;
-
-      memoizedIsInitialized = 1;
-      return true;
-    }
-
-    @java.lang.Override
-    public void writeTo(com.google.protobuf.CodedOutputStream output)
-                        throws java.io.IOException {
-      if (!getNameBytes().isEmpty()) {
-        com.google.protobuf.GeneratedMessageV3.writeString(output, 1, name_);
-      }
-      if (!getDescriptionBytes().isEmpty()) {
-        com.google.protobuf.GeneratedMessageV3.writeString(output, 2, description_);
-      }
-      if (type_ != org.tensorflow.proto.framework.DataType.DT_INVALID.getNumber()) {
-        output.writeEnum(3, type_);
-      }
-      if (!getTypeAttrBytes().isEmpty()) {
-        com.google.protobuf.GeneratedMessageV3.writeString(output, 4, typeAttr_);
-      }
-      if (!getNumberAttrBytes().isEmpty()) {
-        com.google.protobuf.GeneratedMessageV3.writeString(output, 5, numberAttr_);
-      }
-      if (!getTypeListAttrBytes().isEmpty()) {
-        com.google.protobuf.GeneratedMessageV3.writeString(output, 6, typeListAttr_);
-      }
-      if (isRef_ != false) {
-        output.writeBool(16, isRef_);
-      }
-      unknownFields.writeTo(output);
-    }
-
-    @java.lang.Override
-    public int getSerializedSize() {
-      int size = memoizedSize;
-      if (size != -1) return size;
-
-      size = 0;
-      if (!getNameBytes().isEmpty()) {
-        size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, name_);
-      }
-      if (!getDescriptionBytes().isEmpty()) {
-        size += com.google.protobuf.GeneratedMessageV3.computeStringSize(2, description_);
-      }
-      if (type_ != org.tensorflow.proto.framework.DataType.DT_INVALID.getNumber()) {
-        size += com.google.protobuf.CodedOutputStream
-          .computeEnumSize(3, type_);
-      }
-      if (!getTypeAttrBytes().isEmpty()) {
-        size += com.google.protobuf.GeneratedMessageV3.computeStringSize(4, typeAttr_);
-      }
-      if (!getNumberAttrBytes().isEmpty()) {
-        size += com.google.protobuf.GeneratedMessageV3.computeStringSize(5, numberAttr_);
-      }
-      if (!getTypeListAttrBytes().isEmpty()) {
-        size += com.google.protobuf.GeneratedMessageV3.computeStringSize(6, typeListAttr_);
-      }
-      if (isRef_ != false) {
-        size += com.google.protobuf.CodedOutputStream
-          .computeBoolSize(16, isRef_);
-      }
-      size += unknownFields.getSerializedSize();
-      memoizedSize = size;
-      return size;
-    }
-
-    @java.lang.Override
-    public boolean equals(final java.lang.Object obj) {
-      if (obj == this) {
-       return true;
-      }
-      if (!(obj instanceof org.tensorflow.proto.framework.OpDef.ArgDef)) {
-        return super.equals(obj);
-      }
-      org.tensorflow.proto.framework.OpDef.ArgDef other = (org.tensorflow.proto.framework.OpDef.ArgDef) obj;
-
-      if (!getName()
-          .equals(other.getName())) return false;
-      if (!getDescription()
-          .equals(other.getDescription())) return false;
-      if (type_ != other.type_) return false;
-      if (!getTypeAttr()
-          .equals(other.getTypeAttr())) return false;
-      if (!getNumberAttr()
-          .equals(other.getNumberAttr())) return false;
-      if (!getTypeListAttr()
-          .equals(other.getTypeListAttr())) return false;
-      if (getIsRef()
-          != other.getIsRef()) return false;
-      if (!unknownFields.equals(other.unknownFields)) return false;
-      return true;
-    }
-
-    @java.lang.Override
-    public int hashCode() {
-      if (memoizedHashCode != 0) {
-        return memoizedHashCode;
-      }
-      int hash = 41;
-      hash = (19 * hash) + getDescriptor().hashCode();
-      hash = (37 * hash) + NAME_FIELD_NUMBER;
-      hash = (53 * hash) + getName().hashCode();
-      hash = (37 * hash) + DESCRIPTION_FIELD_NUMBER;
-      hash = (53 * hash) + getDescription().hashCode();
-      hash = (37 * hash) + TYPE_FIELD_NUMBER;
-      hash = (53 * hash) + type_;
-      hash = (37 * hash) + TYPE_ATTR_FIELD_NUMBER;
-      hash = (53 * hash) + getTypeAttr().hashCode();
-      hash = (37 * hash) + NUMBER_ATTR_FIELD_NUMBER;
-      hash = (53 * hash) + getNumberAttr().hashCode();
-      hash = (37 * hash) + TYPE_LIST_ATTR_FIELD_NUMBER;
-      hash = (53 * hash) + getTypeListAttr().hashCode();
-      hash = (37 * hash) + IS_REF_FIELD_NUMBER;
-      hash = (53 * hash) + com.google.protobuf.Internal.hashBoolean(
-          getIsRef());
-      hash = (29 * hash) + unknownFields.hashCode();
-      memoizedHashCode = hash;
-      return hash;
-    }
-
-    public static org.tensorflow.proto.framework.OpDef.ArgDef parseFrom(
-        java.nio.ByteBuffer data)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data);
-    }
-    public static org.tensorflow.proto.framework.OpDef.ArgDef parseFrom(
-        java.nio.ByteBuffer data,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.OpDef.ArgDef parseFrom(
-        com.google.protobuf.ByteString data)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data);
-    }
-    public static org.tensorflow.proto.framework.OpDef.ArgDef parseFrom(
-        com.google.protobuf.ByteString data,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.OpDef.ArgDef parseFrom(byte[] data)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data);
-    }
-    public static org.tensorflow.proto.framework.OpDef.ArgDef parseFrom(
-        byte[] data,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.OpDef.ArgDef parseFrom(java.io.InputStream input)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input);
-    }
-    public static org.tensorflow.proto.framework.OpDef.ArgDef parseFrom(
-        java.io.InputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.OpDef.ArgDef parseDelimitedFrom(java.io.InputStream input)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseDelimitedWithIOException(PARSER, input);
-    }
-    public static org.tensorflow.proto.framework.OpDef.ArgDef parseDelimitedFrom(
-        java.io.InputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.OpDef.ArgDef parseFrom(
-        com.google.protobuf.CodedInputStream input)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input);
-    }
-    public static org.tensorflow.proto.framework.OpDef.ArgDef parseFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input, extensionRegistry);
-    }
-
-    @java.lang.Override
-    public Builder newBuilderForType() { return newBuilder(); }
-    public static Builder newBuilder() {
-      return DEFAULT_INSTANCE.toBuilder();
-    }
-    public static Builder newBuilder(org.tensorflow.proto.framework.OpDef.ArgDef prototype) {
-      return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-    }
-    @java.lang.Override
-    public Builder toBuilder() {
-      return this == DEFAULT_INSTANCE
-          ? new Builder() : new Builder().mergeFrom(this);
-    }
-
-    @java.lang.Override
-    protected Builder newBuilderForType(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      Builder builder = new Builder(parent);
-      return builder;
-    }
-    /**
-     * <pre>
-     * For describing inputs and outputs.
-     * </pre>
-     *
-     * Protobuf type {@code tensorflow.OpDef.ArgDef}
-     */
-    public static final class Builder extends
-        com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-        // @@protoc_insertion_point(builder_implements:tensorflow.OpDef.ArgDef)
-        org.tensorflow.proto.framework.OpDef.ArgDefOrBuilder {
-      public static final com.google.protobuf.Descriptors.Descriptor
-          getDescriptor() {
-        return org.tensorflow.proto.framework.OpDefProtos.internal_static_tensorflow_OpDef_ArgDef_descriptor;
-      }
-
-      @java.lang.Override
-      protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-          internalGetFieldAccessorTable() {
-        return org.tensorflow.proto.framework.OpDefProtos.internal_static_tensorflow_OpDef_ArgDef_fieldAccessorTable
-            .ensureFieldAccessorsInitialized(
-                org.tensorflow.proto.framework.OpDef.ArgDef.class, org.tensorflow.proto.framework.OpDef.ArgDef.Builder.class);
-      }
-
-      // Construct using org.tensorflow.proto.framework.OpDef.ArgDef.newBuilder()
-      private Builder() {
-        maybeForceBuilderInitialization();
-      }
-
-      private Builder(
-          com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-        super(parent);
-        maybeForceBuilderInitialization();
-      }
-      private void maybeForceBuilderInitialization() {
-        if (com.google.protobuf.GeneratedMessageV3
-                .alwaysUseFieldBuilders) {
-        }
-      }
-      @java.lang.Override
-      public Builder clear() {
-        super.clear();
-        name_ = "";
-
-        description_ = "";
-
-        type_ = 0;
-
-        typeAttr_ = "";
-
-        numberAttr_ = "";
-
-        typeListAttr_ = "";
-
-        isRef_ = false;
-
-        return this;
-      }
-
-      @java.lang.Override
-      public com.google.protobuf.Descriptors.Descriptor
-          getDescriptorForType() {
-        return org.tensorflow.proto.framework.OpDefProtos.internal_static_tensorflow_OpDef_ArgDef_descriptor;
-      }
-
-      @java.lang.Override
-      public org.tensorflow.proto.framework.OpDef.ArgDef getDefaultInstanceForType() {
-        return org.tensorflow.proto.framework.OpDef.ArgDef.getDefaultInstance();
-      }
-
-      @java.lang.Override
-      public org.tensorflow.proto.framework.OpDef.ArgDef build() {
-        org.tensorflow.proto.framework.OpDef.ArgDef result = buildPartial();
-        if (!result.isInitialized()) {
-          throw newUninitializedMessageException(result);
-        }
-        return result;
-      }
-
-      @java.lang.Override
-      public org.tensorflow.proto.framework.OpDef.ArgDef buildPartial() {
-        org.tensorflow.proto.framework.OpDef.ArgDef result = new org.tensorflow.proto.framework.OpDef.ArgDef(this);
-        result.name_ = name_;
-        result.description_ = description_;
-        result.type_ = type_;
-        result.typeAttr_ = typeAttr_;
-        result.numberAttr_ = numberAttr_;
-        result.typeListAttr_ = typeListAttr_;
-        result.isRef_ = isRef_;
-        onBuilt();
-        return result;
-      }
-
-      @java.lang.Override
-      public Builder clone() {
-        return super.clone();
-      }
-      @java.lang.Override
-      public Builder setField(
-          com.google.protobuf.Descriptors.FieldDescriptor field,
-          java.lang.Object value) {
-        return super.setField(field, value);
-      }
-      @java.lang.Override
-      public Builder clearField(
-          com.google.protobuf.Descriptors.FieldDescriptor field) {
-        return super.clearField(field);
-      }
-      @java.lang.Override
-      public Builder clearOneof(
-          com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-        return super.clearOneof(oneof);
-      }
-      @java.lang.Override
-      public Builder setRepeatedField(
-          com.google.protobuf.Descriptors.FieldDescriptor field,
-          int index, java.lang.Object value) {
-        return super.setRepeatedField(field, index, value);
-      }
-      @java.lang.Override
-      public Builder addRepeatedField(
-          com.google.protobuf.Descriptors.FieldDescriptor field,
-          java.lang.Object value) {
-        return super.addRepeatedField(field, value);
-      }
-      @java.lang.Override
-      public Builder mergeFrom(com.google.protobuf.Message other) {
-        if (other instanceof org.tensorflow.proto.framework.OpDef.ArgDef) {
-          return mergeFrom((org.tensorflow.proto.framework.OpDef.ArgDef)other);
-        } else {
-          super.mergeFrom(other);
-          return this;
-        }
-      }
-
-      public Builder mergeFrom(org.tensorflow.proto.framework.OpDef.ArgDef other) {
-        if (other == org.tensorflow.proto.framework.OpDef.ArgDef.getDefaultInstance()) return this;
-        if (!other.getName().isEmpty()) {
-          name_ = other.name_;
-          onChanged();
-        }
-        if (!other.getDescription().isEmpty()) {
-          description_ = other.description_;
-          onChanged();
-        }
-        if (other.type_ != 0) {
-          setTypeValue(other.getTypeValue());
-        }
-        if (!other.getTypeAttr().isEmpty()) {
-          typeAttr_ = other.typeAttr_;
-          onChanged();
-        }
-        if (!other.getNumberAttr().isEmpty()) {
-          numberAttr_ = other.numberAttr_;
-          onChanged();
-        }
-        if (!other.getTypeListAttr().isEmpty()) {
-          typeListAttr_ = other.typeListAttr_;
-          onChanged();
-        }
-        if (other.getIsRef() != false) {
-          setIsRef(other.getIsRef());
-        }
-        this.mergeUnknownFields(other.unknownFields);
-        onChanged();
-        return this;
-      }
-
-      @java.lang.Override
-      public final boolean isInitialized() {
-        return true;
-      }
-
-      @java.lang.Override
-      public Builder mergeFrom(
-          com.google.protobuf.CodedInputStream input,
-          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-          throws java.io.IOException {
-        org.tensorflow.proto.framework.OpDef.ArgDef parsedMessage = null;
-        try {
-          parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-        } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-          parsedMessage = (org.tensorflow.proto.framework.OpDef.ArgDef) e.getUnfinishedMessage();
-          throw e.unwrapIOException();
-        } finally {
-          if (parsedMessage != null) {
-            mergeFrom(parsedMessage);
-          }
-        }
-        return this;
-      }
-
-      private java.lang.Object name_ = "";
-      /**
-       * <pre>
-       * Name for the input/output.  Should match the regexp "[a-z][a-z0-9_]*".
-       * </pre>
-       *
-       * <code>string name = 1;</code>
-       */
-      public java.lang.String getName() {
-        java.lang.Object ref = name_;
-        if (!(ref instanceof java.lang.String)) {
-          com.google.protobuf.ByteString bs =
-              (com.google.protobuf.ByteString) ref;
-          java.lang.String s = bs.toStringUtf8();
-          name_ = s;
-          return s;
-        } else {
-          return (java.lang.String) ref;
-        }
-      }
-      /**
-       * <pre>
-       * Name for the input/output.  Should match the regexp "[a-z][a-z0-9_]*".
-       * </pre>
-       *
-       * <code>string name = 1;</code>
-       */
-      public com.google.protobuf.ByteString
-          getNameBytes() {
-        java.lang.Object ref = name_;
-        if (ref instanceof String) {
-          com.google.protobuf.ByteString b = 
-              com.google.protobuf.ByteString.copyFromUtf8(
-                  (java.lang.String) ref);
-          name_ = b;
-          return b;
-        } else {
-          return (com.google.protobuf.ByteString) ref;
-        }
-      }
-      /**
-       * <pre>
-       * Name for the input/output.  Should match the regexp "[a-z][a-z0-9_]*".
-       * </pre>
-       *
-       * <code>string name = 1;</code>
-       */
-      public Builder setName(
-          java.lang.String value) {
-        if (value == null) {
-    throw new NullPointerException();
-  }
-  
-        name_ = value;
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * Name for the input/output.  Should match the regexp "[a-z][a-z0-9_]*".
-       * </pre>
-       *
-       * <code>string name = 1;</code>
-       */
-      public Builder clearName() {
-        
-        name_ = getDefaultInstance().getName();
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * Name for the input/output.  Should match the regexp "[a-z][a-z0-9_]*".
-       * </pre>
-       *
-       * <code>string name = 1;</code>
-       */
-      public Builder setNameBytes(
-          com.google.protobuf.ByteString value) {
-        if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-        
-        name_ = value;
-        onChanged();
-        return this;
-      }
-
-      private java.lang.Object description_ = "";
-      /**
-       * <pre>
-       * Human readable description.
-       * </pre>
-       *
-       * <code>string description = 2;</code>
-       */
-      public java.lang.String getDescription() {
-        java.lang.Object ref = description_;
-        if (!(ref instanceof java.lang.String)) {
-          com.google.protobuf.ByteString bs =
-              (com.google.protobuf.ByteString) ref;
-          java.lang.String s = bs.toStringUtf8();
-          description_ = s;
-          return s;
-        } else {
-          return (java.lang.String) ref;
-        }
-      }
-      /**
-       * <pre>
-       * Human readable description.
-       * </pre>
-       *
-       * <code>string description = 2;</code>
-       */
-      public com.google.protobuf.ByteString
-          getDescriptionBytes() {
-        java.lang.Object ref = description_;
-        if (ref instanceof String) {
-          com.google.protobuf.ByteString b = 
-              com.google.protobuf.ByteString.copyFromUtf8(
-                  (java.lang.String) ref);
-          description_ = b;
-          return b;
-        } else {
-          return (com.google.protobuf.ByteString) ref;
-        }
-      }
-      /**
-       * <pre>
-       * Human readable description.
-       * </pre>
-       *
-       * <code>string description = 2;</code>
-       */
-      public Builder setDescription(
-          java.lang.String value) {
-        if (value == null) {
-    throw new NullPointerException();
-  }
-  
-        description_ = value;
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * Human readable description.
-       * </pre>
-       *
-       * <code>string description = 2;</code>
-       */
-      public Builder clearDescription() {
-        
-        description_ = getDefaultInstance().getDescription();
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * Human readable description.
-       * </pre>
-       *
-       * <code>string description = 2;</code>
-       */
-      public Builder setDescriptionBytes(
-          com.google.protobuf.ByteString value) {
-        if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-        
-        description_ = value;
-        onChanged();
-        return this;
-      }
-
-      private int type_ = 0;
-      /**
-       * <pre>
-       * Describes the type of one or more tensors that are accepted/produced
-       * by this input/output arg.  The only legal combinations are:
-       * * For a single tensor: either the "type" field is set or the
-       *   "type_attr" field is set to the name of an attr with type "type".
-       * * For a sequence of tensors with the same type: the "number_attr"
-       *   field will be set to the name of an attr with type "int", and
-       *   either the "type" or "type_attr" field will be set as for
-       *   single tensors.
-       * * For a sequence of tensors, the "type_list_attr" field will be set
-       *   to the name of an attr with type "list(type)".
-       * </pre>
-       *
-       * <code>.tensorflow.DataType type = 3;</code>
-       */
-      public int getTypeValue() {
-        return type_;
-      }
-      /**
-       * <pre>
-       * Describes the type of one or more tensors that are accepted/produced
-       * by this input/output arg.  The only legal combinations are:
-       * * For a single tensor: either the "type" field is set or the
-       *   "type_attr" field is set to the name of an attr with type "type".
-       * * For a sequence of tensors with the same type: the "number_attr"
-       *   field will be set to the name of an attr with type "int", and
-       *   either the "type" or "type_attr" field will be set as for
-       *   single tensors.
-       * * For a sequence of tensors, the "type_list_attr" field will be set
-       *   to the name of an attr with type "list(type)".
-       * </pre>
-       *
-       * <code>.tensorflow.DataType type = 3;</code>
-       */
-      public Builder setTypeValue(int value) {
-        type_ = value;
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * Describes the type of one or more tensors that are accepted/produced
-       * by this input/output arg.  The only legal combinations are:
-       * * For a single tensor: either the "type" field is set or the
-       *   "type_attr" field is set to the name of an attr with type "type".
-       * * For a sequence of tensors with the same type: the "number_attr"
-       *   field will be set to the name of an attr with type "int", and
-       *   either the "type" or "type_attr" field will be set as for
-       *   single tensors.
-       * * For a sequence of tensors, the "type_list_attr" field will be set
-       *   to the name of an attr with type "list(type)".
-       * </pre>
-       *
-       * <code>.tensorflow.DataType type = 3;</code>
-       */
-      public org.tensorflow.proto.framework.DataType getType() {
-        @SuppressWarnings("deprecation")
-        org.tensorflow.proto.framework.DataType result = org.tensorflow.proto.framework.DataType.valueOf(type_);
-        return result == null ? org.tensorflow.proto.framework.DataType.UNRECOGNIZED : result;
-      }
-      /**
-       * <pre>
-       * Describes the type of one or more tensors that are accepted/produced
-       * by this input/output arg.  The only legal combinations are:
-       * * For a single tensor: either the "type" field is set or the
-       *   "type_attr" field is set to the name of an attr with type "type".
-       * * For a sequence of tensors with the same type: the "number_attr"
-       *   field will be set to the name of an attr with type "int", and
-       *   either the "type" or "type_attr" field will be set as for
-       *   single tensors.
-       * * For a sequence of tensors, the "type_list_attr" field will be set
-       *   to the name of an attr with type "list(type)".
-       * </pre>
-       *
-       * <code>.tensorflow.DataType type = 3;</code>
-       */
-      public Builder setType(org.tensorflow.proto.framework.DataType value) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        
-        type_ = value.getNumber();
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * Describes the type of one or more tensors that are accepted/produced
-       * by this input/output arg.  The only legal combinations are:
-       * * For a single tensor: either the "type" field is set or the
-       *   "type_attr" field is set to the name of an attr with type "type".
-       * * For a sequence of tensors with the same type: the "number_attr"
-       *   field will be set to the name of an attr with type "int", and
-       *   either the "type" or "type_attr" field will be set as for
-       *   single tensors.
-       * * For a sequence of tensors, the "type_list_attr" field will be set
-       *   to the name of an attr with type "list(type)".
-       * </pre>
-       *
-       * <code>.tensorflow.DataType type = 3;</code>
-       */
-      public Builder clearType() {
-        
-        type_ = 0;
-        onChanged();
-        return this;
-      }
-
-      private java.lang.Object typeAttr_ = "";
-      /**
-       * <pre>
-       * if specified, attr must have type "type"
-       * </pre>
-       *
-       * <code>string type_attr = 4;</code>
-       */
-      public java.lang.String getTypeAttr() {
-        java.lang.Object ref = typeAttr_;
-        if (!(ref instanceof java.lang.String)) {
-          com.google.protobuf.ByteString bs =
-              (com.google.protobuf.ByteString) ref;
-          java.lang.String s = bs.toStringUtf8();
-          typeAttr_ = s;
-          return s;
-        } else {
-          return (java.lang.String) ref;
-        }
-      }
-      /**
-       * <pre>
-       * if specified, attr must have type "type"
-       * </pre>
-       *
-       * <code>string type_attr = 4;</code>
-       */
-      public com.google.protobuf.ByteString
-          getTypeAttrBytes() {
-        java.lang.Object ref = typeAttr_;
-        if (ref instanceof String) {
-          com.google.protobuf.ByteString b = 
-              com.google.protobuf.ByteString.copyFromUtf8(
-                  (java.lang.String) ref);
-          typeAttr_ = b;
-          return b;
-        } else {
-          return (com.google.protobuf.ByteString) ref;
-        }
-      }
-      /**
-       * <pre>
-       * if specified, attr must have type "type"
-       * </pre>
-       *
-       * <code>string type_attr = 4;</code>
-       */
-      public Builder setTypeAttr(
-          java.lang.String value) {
-        if (value == null) {
-    throw new NullPointerException();
-  }
-  
-        typeAttr_ = value;
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * if specified, attr must have type "type"
-       * </pre>
-       *
-       * <code>string type_attr = 4;</code>
-       */
-      public Builder clearTypeAttr() {
-        
-        typeAttr_ = getDefaultInstance().getTypeAttr();
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * if specified, attr must have type "type"
-       * </pre>
-       *
-       * <code>string type_attr = 4;</code>
-       */
-      public Builder setTypeAttrBytes(
-          com.google.protobuf.ByteString value) {
-        if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-        
-        typeAttr_ = value;
-        onChanged();
-        return this;
-      }
-
-      private java.lang.Object numberAttr_ = "";
-      /**
-       * <pre>
-       * if specified, attr must have type "int"
-       * </pre>
-       *
-       * <code>string number_attr = 5;</code>
-       */
-      public java.lang.String getNumberAttr() {
-        java.lang.Object ref = numberAttr_;
-        if (!(ref instanceof java.lang.String)) {
-          com.google.protobuf.ByteString bs =
-              (com.google.protobuf.ByteString) ref;
-          java.lang.String s = bs.toStringUtf8();
-          numberAttr_ = s;
-          return s;
-        } else {
-          return (java.lang.String) ref;
-        }
-      }
-      /**
-       * <pre>
-       * if specified, attr must have type "int"
-       * </pre>
-       *
-       * <code>string number_attr = 5;</code>
-       */
-      public com.google.protobuf.ByteString
-          getNumberAttrBytes() {
-        java.lang.Object ref = numberAttr_;
-        if (ref instanceof String) {
-          com.google.protobuf.ByteString b = 
-              com.google.protobuf.ByteString.copyFromUtf8(
-                  (java.lang.String) ref);
-          numberAttr_ = b;
-          return b;
-        } else {
-          return (com.google.protobuf.ByteString) ref;
-        }
-      }
-      /**
-       * <pre>
-       * if specified, attr must have type "int"
-       * </pre>
-       *
-       * <code>string number_attr = 5;</code>
-       */
-      public Builder setNumberAttr(
-          java.lang.String value) {
-        if (value == null) {
-    throw new NullPointerException();
-  }
-  
-        numberAttr_ = value;
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * if specified, attr must have type "int"
-       * </pre>
-       *
-       * <code>string number_attr = 5;</code>
-       */
-      public Builder clearNumberAttr() {
-        
-        numberAttr_ = getDefaultInstance().getNumberAttr();
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * if specified, attr must have type "int"
-       * </pre>
-       *
-       * <code>string number_attr = 5;</code>
-       */
-      public Builder setNumberAttrBytes(
-          com.google.protobuf.ByteString value) {
-        if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-        
-        numberAttr_ = value;
-        onChanged();
-        return this;
-      }
-
-      private java.lang.Object typeListAttr_ = "";
-      /**
-       * <pre>
-       * If specified, attr must have type "list(type)", and none of
-       * type, type_attr, and number_attr may be specified.
-       * </pre>
-       *
-       * <code>string type_list_attr = 6;</code>
-       */
-      public java.lang.String getTypeListAttr() {
-        java.lang.Object ref = typeListAttr_;
-        if (!(ref instanceof java.lang.String)) {
-          com.google.protobuf.ByteString bs =
-              (com.google.protobuf.ByteString) ref;
-          java.lang.String s = bs.toStringUtf8();
-          typeListAttr_ = s;
-          return s;
-        } else {
-          return (java.lang.String) ref;
-        }
-      }
-      /**
-       * <pre>
-       * If specified, attr must have type "list(type)", and none of
-       * type, type_attr, and number_attr may be specified.
-       * </pre>
-       *
-       * <code>string type_list_attr = 6;</code>
-       */
-      public com.google.protobuf.ByteString
-          getTypeListAttrBytes() {
-        java.lang.Object ref = typeListAttr_;
-        if (ref instanceof String) {
-          com.google.protobuf.ByteString b = 
-              com.google.protobuf.ByteString.copyFromUtf8(
-                  (java.lang.String) ref);
-          typeListAttr_ = b;
-          return b;
-        } else {
-          return (com.google.protobuf.ByteString) ref;
-        }
-      }
-      /**
-       * <pre>
-       * If specified, attr must have type "list(type)", and none of
-       * type, type_attr, and number_attr may be specified.
-       * </pre>
-       *
-       * <code>string type_list_attr = 6;</code>
-       */
-      public Builder setTypeListAttr(
-          java.lang.String value) {
-        if (value == null) {
-    throw new NullPointerException();
-  }
-  
-        typeListAttr_ = value;
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * If specified, attr must have type "list(type)", and none of
-       * type, type_attr, and number_attr may be specified.
-       * </pre>
-       *
-       * <code>string type_list_attr = 6;</code>
-       */
-      public Builder clearTypeListAttr() {
-        
-        typeListAttr_ = getDefaultInstance().getTypeListAttr();
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * If specified, attr must have type "list(type)", and none of
-       * type, type_attr, and number_attr may be specified.
-       * </pre>
-       *
-       * <code>string type_list_attr = 6;</code>
-       */
-      public Builder setTypeListAttrBytes(
-          com.google.protobuf.ByteString value) {
-        if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-        
-        typeListAttr_ = value;
-        onChanged();
-        return this;
-      }
-
-      private boolean isRef_ ;
-      /**
-       * <pre>
-       * For inputs: if true, the inputs are required to be refs.
-       *   By default, inputs can be either refs or non-refs.
-       * For outputs: if true, outputs are refs, otherwise they are not.
-       * </pre>
-       *
-       * <code>bool is_ref = 16;</code>
-       */
-      public boolean getIsRef() {
-        return isRef_;
-      }
-      /**
-       * <pre>
-       * For inputs: if true, the inputs are required to be refs.
-       *   By default, inputs can be either refs or non-refs.
-       * For outputs: if true, outputs are refs, otherwise they are not.
-       * </pre>
-       *
-       * <code>bool is_ref = 16;</code>
-       */
-      public Builder setIsRef(boolean value) {
-        
-        isRef_ = value;
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * For inputs: if true, the inputs are required to be refs.
-       *   By default, inputs can be either refs or non-refs.
-       * For outputs: if true, outputs are refs, otherwise they are not.
-       * </pre>
-       *
-       * <code>bool is_ref = 16;</code>
-       */
-      public Builder clearIsRef() {
-        
-        isRef_ = false;
-        onChanged();
-        return this;
-      }
-      @java.lang.Override
-      public final Builder setUnknownFields(
-          final com.google.protobuf.UnknownFieldSet unknownFields) {
-        return super.setUnknownFields(unknownFields);
-      }
-
-      @java.lang.Override
-      public final Builder mergeUnknownFields(
-          final com.google.protobuf.UnknownFieldSet unknownFields) {
-        return super.mergeUnknownFields(unknownFields);
-      }
-
-
-      // @@protoc_insertion_point(builder_scope:tensorflow.OpDef.ArgDef)
-    }
-
-    // @@protoc_insertion_point(class_scope:tensorflow.OpDef.ArgDef)
-    private static final org.tensorflow.proto.framework.OpDef.ArgDef DEFAULT_INSTANCE;
-    static {
-      DEFAULT_INSTANCE = new org.tensorflow.proto.framework.OpDef.ArgDef();
-    }
-
-    public static org.tensorflow.proto.framework.OpDef.ArgDef getDefaultInstance() {
-      return DEFAULT_INSTANCE;
-    }
-
-    private static final com.google.protobuf.Parser<ArgDef>
-        PARSER = new com.google.protobuf.AbstractParser<ArgDef>() {
-      @java.lang.Override
-      public ArgDef parsePartialFrom(
-          com.google.protobuf.CodedInputStream input,
-          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-          throws com.google.protobuf.InvalidProtocolBufferException {
-        return new ArgDef(input, extensionRegistry);
-      }
-    };
-
-    public static com.google.protobuf.Parser<ArgDef> parser() {
-      return PARSER;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Parser<ArgDef> getParserForType() {
-      return PARSER;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.OpDef.ArgDef getDefaultInstanceForType() {
-      return DEFAULT_INSTANCE;
-    }
-
-  }
-
-  public interface AttrDefOrBuilder extends
-      // @@protoc_insertion_point(interface_extends:tensorflow.OpDef.AttrDef)
-      com.google.protobuf.MessageOrBuilder {
-
-    /**
-     * <pre>
-     * A descriptive name for the argument.  May be used, e.g. by the
-     * Python client, as a keyword argument name, and so should match
-     * the regexp "[a-z][a-z0-9_]+".
-     * </pre>
-     *
-     * <code>string name = 1;</code>
-     */
-    java.lang.String getName();
-    /**
-     * <pre>
-     * A descriptive name for the argument.  May be used, e.g. by the
-     * Python client, as a keyword argument name, and so should match
-     * the regexp "[a-z][a-z0-9_]+".
-     * </pre>
-     *
-     * <code>string name = 1;</code>
-     */
-    com.google.protobuf.ByteString
-        getNameBytes();
-
-    /**
-     * <pre>
-     * One of the type names from attr_value.proto ("string", "list(string)",
-     * "int", etc.).
-     * </pre>
-     *
-     * <code>string type = 2;</code>
-     */
-    java.lang.String getType();
-    /**
-     * <pre>
-     * One of the type names from attr_value.proto ("string", "list(string)",
-     * "int", etc.).
-     * </pre>
-     *
-     * <code>string type = 2;</code>
-     */
-    com.google.protobuf.ByteString
-        getTypeBytes();
-
-    /**
-     * <pre>
-     * A reasonable default for this attribute if the user does not supply
-     * a value.  If not specified, the user must supply a value.
-     * </pre>
-     *
-     * <code>.tensorflow.AttrValue default_value = 3;</code>
-     */
-    boolean hasDefaultValue();
-    /**
-     * <pre>
-     * A reasonable default for this attribute if the user does not supply
-     * a value.  If not specified, the user must supply a value.
-     * </pre>
-     *
-     * <code>.tensorflow.AttrValue default_value = 3;</code>
-     */
-    org.tensorflow.proto.framework.AttrValue getDefaultValue();
-    /**
-     * <pre>
-     * A reasonable default for this attribute if the user does not supply
-     * a value.  If not specified, the user must supply a value.
-     * </pre>
-     *
-     * <code>.tensorflow.AttrValue default_value = 3;</code>
-     */
-    org.tensorflow.proto.framework.AttrValueOrBuilder getDefaultValueOrBuilder();
-
-    /**
-     * <pre>
-     * Human-readable description.
-     * </pre>
-     *
-     * <code>string description = 4;</code>
-     */
-    java.lang.String getDescription();
-    /**
-     * <pre>
-     * Human-readable description.
-     * </pre>
-     *
-     * <code>string description = 4;</code>
-     */
-    com.google.protobuf.ByteString
-        getDescriptionBytes();
-
-    /**
-     * <pre>
-     * For type == "int", this is a minimum value.  For "list(___)"
-     * types, this is the minimum length.
-     * </pre>
-     *
-     * <code>bool has_minimum = 5;</code>
-     */
-    boolean getHasMinimum();
-
-    /**
-     * <code>int64 minimum = 6;</code>
-     */
-    long getMinimum();
-
-    /**
-     * <pre>
-     * The set of allowed values.  Has type that is the "list" version
-     * of the "type" field above (uses the "list" field of AttrValue).
-     * If type == "type" or "list(type)" above, then the "type" field
-     * of "allowed_values.list" has the set of allowed DataTypes.
-     * If type == "string" or "list(string)", then the "s" field of
-     * "allowed_values.list" has the set of allowed strings.
-     * </pre>
-     *
-     * <code>.tensorflow.AttrValue allowed_values = 7;</code>
-     */
-    boolean hasAllowedValues();
-    /**
-     * <pre>
-     * The set of allowed values.  Has type that is the "list" version
-     * of the "type" field above (uses the "list" field of AttrValue).
-     * If type == "type" or "list(type)" above, then the "type" field
-     * of "allowed_values.list" has the set of allowed DataTypes.
-     * If type == "string" or "list(string)", then the "s" field of
-     * "allowed_values.list" has the set of allowed strings.
-     * </pre>
-     *
-     * <code>.tensorflow.AttrValue allowed_values = 7;</code>
-     */
-    org.tensorflow.proto.framework.AttrValue getAllowedValues();
-    /**
-     * <pre>
-     * The set of allowed values.  Has type that is the "list" version
-     * of the "type" field above (uses the "list" field of AttrValue).
-     * If type == "type" or "list(type)" above, then the "type" field
-     * of "allowed_values.list" has the set of allowed DataTypes.
-     * If type == "string" or "list(string)", then the "s" field of
-     * "allowed_values.list" has the set of allowed strings.
-     * </pre>
-     *
-     * <code>.tensorflow.AttrValue allowed_values = 7;</code>
-     */
-    org.tensorflow.proto.framework.AttrValueOrBuilder getAllowedValuesOrBuilder();
-  }
-  /**
-   * <pre>
-   * Description of the graph-construction-time configuration of this
-   * Op.  That is to say, this describes the attr fields that will
-   * be specified in the NodeDef.
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.OpDef.AttrDef}
-   */
-  public  static final class AttrDef extends
-      com.google.protobuf.GeneratedMessageV3 implements
-      // @@protoc_insertion_point(message_implements:tensorflow.OpDef.AttrDef)
-      AttrDefOrBuilder {
-  private static final long serialVersionUID = 0L;
-    // Use AttrDef.newBuilder() to construct.
-    private AttrDef(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-      super(builder);
-    }
-    private AttrDef() {
-      name_ = "";
-      type_ = "";
-      description_ = "";
-    }
-
-    @java.lang.Override
-    @SuppressWarnings({"unused"})
-    protected java.lang.Object newInstance(
-        UnusedPrivateParameter unused) {
-      return new AttrDef();
-    }
-
-    @java.lang.Override
-    public final com.google.protobuf.UnknownFieldSet
-    getUnknownFields() {
-      return this.unknownFields;
-    }
-    private AttrDef(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      this();
-      if (extensionRegistry == null) {
-        throw new java.lang.NullPointerException();
-      }
-      com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-          com.google.protobuf.UnknownFieldSet.newBuilder();
-      try {
-        boolean done = false;
-        while (!done) {
-          int tag = input.readTag();
-          switch (tag) {
-            case 0:
-              done = true;
-              break;
-            case 10: {
-              java.lang.String s = input.readStringRequireUtf8();
-
-              name_ = s;
-              break;
-            }
-            case 18: {
-              java.lang.String s = input.readStringRequireUtf8();
-
-              type_ = s;
-              break;
-            }
-            case 26: {
-              org.tensorflow.proto.framework.AttrValue.Builder subBuilder = null;
-              if (defaultValue_ != null) {
-                subBuilder = defaultValue_.toBuilder();
-              }
-              defaultValue_ = input.readMessage(org.tensorflow.proto.framework.AttrValue.parser(), extensionRegistry);
-              if (subBuilder != null) {
-                subBuilder.mergeFrom(defaultValue_);
-                defaultValue_ = subBuilder.buildPartial();
-              }
-
-              break;
-            }
-            case 34: {
-              java.lang.String s = input.readStringRequireUtf8();
-
-              description_ = s;
-              break;
-            }
-            case 40: {
-
-              hasMinimum_ = input.readBool();
-              break;
-            }
-            case 48: {
-
-              minimum_ = input.readInt64();
-              break;
-            }
-            case 58: {
-              org.tensorflow.proto.framework.AttrValue.Builder subBuilder = null;
-              if (allowedValues_ != null) {
-                subBuilder = allowedValues_.toBuilder();
-              }
-              allowedValues_ = input.readMessage(org.tensorflow.proto.framework.AttrValue.parser(), extensionRegistry);
-              if (subBuilder != null) {
-                subBuilder.mergeFrom(allowedValues_);
-                allowedValues_ = subBuilder.buildPartial();
-              }
-
-              break;
-            }
-            default: {
-              if (!parseUnknownField(
-                  input, unknownFields, extensionRegistry, tag)) {
-                done = true;
-              }
-              break;
-            }
-          }
-        }
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        throw e.setUnfinishedMessage(this);
-      } catch (java.io.IOException e) {
-        throw new com.google.protobuf.InvalidProtocolBufferException(
-            e).setUnfinishedMessage(this);
-      } finally {
-        this.unknownFields = unknownFields.build();
-        makeExtensionsImmutable();
-      }
-    }
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.OpDefProtos.internal_static_tensorflow_OpDef_AttrDef_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.OpDefProtos.internal_static_tensorflow_OpDef_AttrDef_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.OpDef.AttrDef.class, org.tensorflow.proto.framework.OpDef.AttrDef.Builder.class);
-    }
-
-    public static final int NAME_FIELD_NUMBER = 1;
-    private volatile java.lang.Object name_;
-    /**
-     * <pre>
-     * A descriptive name for the argument.  May be used, e.g. by the
-     * Python client, as a keyword argument name, and so should match
-     * the regexp "[a-z][a-z0-9_]+".
-     * </pre>
-     *
-     * <code>string name = 1;</code>
-     */
-    public java.lang.String getName() {
-      java.lang.Object ref = name_;
-      if (ref instanceof java.lang.String) {
-        return (java.lang.String) ref;
-      } else {
-        com.google.protobuf.ByteString bs = 
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        name_ = s;
-        return s;
-      }
-    }
-    /**
-     * <pre>
-     * A descriptive name for the argument.  May be used, e.g. by the
-     * Python client, as a keyword argument name, and so should match
-     * the regexp "[a-z][a-z0-9_]+".
-     * </pre>
-     *
-     * <code>string name = 1;</code>
-     */
-    public com.google.protobuf.ByteString
-        getNameBytes() {
-      java.lang.Object ref = name_;
-      if (ref instanceof java.lang.String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        name_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-
-    public static final int TYPE_FIELD_NUMBER = 2;
-    private volatile java.lang.Object type_;
-    /**
-     * <pre>
-     * One of the type names from attr_value.proto ("string", "list(string)",
-     * "int", etc.).
-     * </pre>
-     *
-     * <code>string type = 2;</code>
-     */
-    public java.lang.String getType() {
-      java.lang.Object ref = type_;
-      if (ref instanceof java.lang.String) {
-        return (java.lang.String) ref;
-      } else {
-        com.google.protobuf.ByteString bs = 
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        type_ = s;
-        return s;
-      }
-    }
-    /**
-     * <pre>
-     * One of the type names from attr_value.proto ("string", "list(string)",
-     * "int", etc.).
-     * </pre>
-     *
-     * <code>string type = 2;</code>
-     */
-    public com.google.protobuf.ByteString
-        getTypeBytes() {
-      java.lang.Object ref = type_;
-      if (ref instanceof java.lang.String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        type_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-
-    public static final int DEFAULT_VALUE_FIELD_NUMBER = 3;
-    private org.tensorflow.proto.framework.AttrValue defaultValue_;
-    /**
-     * <pre>
-     * A reasonable default for this attribute if the user does not supply
-     * a value.  If not specified, the user must supply a value.
-     * </pre>
-     *
-     * <code>.tensorflow.AttrValue default_value = 3;</code>
-     */
-    public boolean hasDefaultValue() {
-      return defaultValue_ != null;
-    }
-    /**
-     * <pre>
-     * A reasonable default for this attribute if the user does not supply
-     * a value.  If not specified, the user must supply a value.
-     * </pre>
-     *
-     * <code>.tensorflow.AttrValue default_value = 3;</code>
-     */
-    public org.tensorflow.proto.framework.AttrValue getDefaultValue() {
-      return defaultValue_ == null ? org.tensorflow.proto.framework.AttrValue.getDefaultInstance() : defaultValue_;
-    }
-    /**
-     * <pre>
-     * A reasonable default for this attribute if the user does not supply
-     * a value.  If not specified, the user must supply a value.
-     * </pre>
-     *
-     * <code>.tensorflow.AttrValue default_value = 3;</code>
-     */
-    public org.tensorflow.proto.framework.AttrValueOrBuilder getDefaultValueOrBuilder() {
-      return getDefaultValue();
-    }
-
-    public static final int DESCRIPTION_FIELD_NUMBER = 4;
-    private volatile java.lang.Object description_;
-    /**
-     * <pre>
-     * Human-readable description.
-     * </pre>
-     *
-     * <code>string description = 4;</code>
-     */
-    public java.lang.String getDescription() {
-      java.lang.Object ref = description_;
-      if (ref instanceof java.lang.String) {
-        return (java.lang.String) ref;
-      } else {
-        com.google.protobuf.ByteString bs = 
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        description_ = s;
-        return s;
-      }
-    }
-    /**
-     * <pre>
-     * Human-readable description.
-     * </pre>
-     *
-     * <code>string description = 4;</code>
-     */
-    public com.google.protobuf.ByteString
-        getDescriptionBytes() {
-      java.lang.Object ref = description_;
-      if (ref instanceof java.lang.String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        description_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-
-    public static final int HAS_MINIMUM_FIELD_NUMBER = 5;
-    private boolean hasMinimum_;
-    /**
-     * <pre>
-     * For type == "int", this is a minimum value.  For "list(___)"
-     * types, this is the minimum length.
-     * </pre>
-     *
-     * <code>bool has_minimum = 5;</code>
-     */
-    public boolean getHasMinimum() {
-      return hasMinimum_;
-    }
-
-    public static final int MINIMUM_FIELD_NUMBER = 6;
-    private long minimum_;
-    /**
-     * <code>int64 minimum = 6;</code>
-     */
-    public long getMinimum() {
-      return minimum_;
-    }
-
-    public static final int ALLOWED_VALUES_FIELD_NUMBER = 7;
-    private org.tensorflow.proto.framework.AttrValue allowedValues_;
-    /**
-     * <pre>
-     * The set of allowed values.  Has type that is the "list" version
-     * of the "type" field above (uses the "list" field of AttrValue).
-     * If type == "type" or "list(type)" above, then the "type" field
-     * of "allowed_values.list" has the set of allowed DataTypes.
-     * If type == "string" or "list(string)", then the "s" field of
-     * "allowed_values.list" has the set of allowed strings.
-     * </pre>
-     *
-     * <code>.tensorflow.AttrValue allowed_values = 7;</code>
-     */
-    public boolean hasAllowedValues() {
-      return allowedValues_ != null;
-    }
-    /**
-     * <pre>
-     * The set of allowed values.  Has type that is the "list" version
-     * of the "type" field above (uses the "list" field of AttrValue).
-     * If type == "type" or "list(type)" above, then the "type" field
-     * of "allowed_values.list" has the set of allowed DataTypes.
-     * If type == "string" or "list(string)", then the "s" field of
-     * "allowed_values.list" has the set of allowed strings.
-     * </pre>
-     *
-     * <code>.tensorflow.AttrValue allowed_values = 7;</code>
-     */
-    public org.tensorflow.proto.framework.AttrValue getAllowedValues() {
-      return allowedValues_ == null ? org.tensorflow.proto.framework.AttrValue.getDefaultInstance() : allowedValues_;
-    }
-    /**
-     * <pre>
-     * The set of allowed values.  Has type that is the "list" version
-     * of the "type" field above (uses the "list" field of AttrValue).
-     * If type == "type" or "list(type)" above, then the "type" field
-     * of "allowed_values.list" has the set of allowed DataTypes.
-     * If type == "string" or "list(string)", then the "s" field of
-     * "allowed_values.list" has the set of allowed strings.
-     * </pre>
-     *
-     * <code>.tensorflow.AttrValue allowed_values = 7;</code>
-     */
-    public org.tensorflow.proto.framework.AttrValueOrBuilder getAllowedValuesOrBuilder() {
-      return getAllowedValues();
-    }
-
-    private byte memoizedIsInitialized = -1;
-    @java.lang.Override
-    public final boolean isInitialized() {
-      byte isInitialized = memoizedIsInitialized;
-      if (isInitialized == 1) return true;
-      if (isInitialized == 0) return false;
-
-      memoizedIsInitialized = 1;
-      return true;
-    }
-
-    @java.lang.Override
-    public void writeTo(com.google.protobuf.CodedOutputStream output)
-                        throws java.io.IOException {
-      if (!getNameBytes().isEmpty()) {
-        com.google.protobuf.GeneratedMessageV3.writeString(output, 1, name_);
-      }
-      if (!getTypeBytes().isEmpty()) {
-        com.google.protobuf.GeneratedMessageV3.writeString(output, 2, type_);
-      }
-      if (defaultValue_ != null) {
-        output.writeMessage(3, getDefaultValue());
-      }
-      if (!getDescriptionBytes().isEmpty()) {
-        com.google.protobuf.GeneratedMessageV3.writeString(output, 4, description_);
-      }
-      if (hasMinimum_ != false) {
-        output.writeBool(5, hasMinimum_);
-      }
-      if (minimum_ != 0L) {
-        output.writeInt64(6, minimum_);
-      }
-      if (allowedValues_ != null) {
-        output.writeMessage(7, getAllowedValues());
-      }
-      unknownFields.writeTo(output);
-    }
-
-    @java.lang.Override
-    public int getSerializedSize() {
-      int size = memoizedSize;
-      if (size != -1) return size;
-
-      size = 0;
-      if (!getNameBytes().isEmpty()) {
-        size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, name_);
-      }
-      if (!getTypeBytes().isEmpty()) {
-        size += com.google.protobuf.GeneratedMessageV3.computeStringSize(2, type_);
-      }
-      if (defaultValue_ != null) {
-        size += com.google.protobuf.CodedOutputStream
-          .computeMessageSize(3, getDefaultValue());
-      }
-      if (!getDescriptionBytes().isEmpty()) {
-        size += com.google.protobuf.GeneratedMessageV3.computeStringSize(4, description_);
-      }
-      if (hasMinimum_ != false) {
-        size += com.google.protobuf.CodedOutputStream
-          .computeBoolSize(5, hasMinimum_);
-      }
-      if (minimum_ != 0L) {
-        size += com.google.protobuf.CodedOutputStream
-          .computeInt64Size(6, minimum_);
-      }
-      if (allowedValues_ != null) {
-        size += com.google.protobuf.CodedOutputStream
-          .computeMessageSize(7, getAllowedValues());
-      }
-      size += unknownFields.getSerializedSize();
-      memoizedSize = size;
-      return size;
-    }
-
-    @java.lang.Override
-    public boolean equals(final java.lang.Object obj) {
-      if (obj == this) {
-       return true;
-      }
-      if (!(obj instanceof org.tensorflow.proto.framework.OpDef.AttrDef)) {
-        return super.equals(obj);
-      }
-      org.tensorflow.proto.framework.OpDef.AttrDef other = (org.tensorflow.proto.framework.OpDef.AttrDef) obj;
-
-      if (!getName()
-          .equals(other.getName())) return false;
-      if (!getType()
-          .equals(other.getType())) return false;
-      if (hasDefaultValue() != other.hasDefaultValue()) return false;
-      if (hasDefaultValue()) {
-        if (!getDefaultValue()
-            .equals(other.getDefaultValue())) return false;
-      }
-      if (!getDescription()
-          .equals(other.getDescription())) return false;
-      if (getHasMinimum()
-          != other.getHasMinimum()) return false;
-      if (getMinimum()
-          != other.getMinimum()) return false;
-      if (hasAllowedValues() != other.hasAllowedValues()) return false;
-      if (hasAllowedValues()) {
-        if (!getAllowedValues()
-            .equals(other.getAllowedValues())) return false;
-      }
-      if (!unknownFields.equals(other.unknownFields)) return false;
-      return true;
-    }
-
-    @java.lang.Override
-    public int hashCode() {
-      if (memoizedHashCode != 0) {
-        return memoizedHashCode;
-      }
-      int hash = 41;
-      hash = (19 * hash) + getDescriptor().hashCode();
-      hash = (37 * hash) + NAME_FIELD_NUMBER;
-      hash = (53 * hash) + getName().hashCode();
-      hash = (37 * hash) + TYPE_FIELD_NUMBER;
-      hash = (53 * hash) + getType().hashCode();
-      if (hasDefaultValue()) {
-        hash = (37 * hash) + DEFAULT_VALUE_FIELD_NUMBER;
-        hash = (53 * hash) + getDefaultValue().hashCode();
-      }
-      hash = (37 * hash) + DESCRIPTION_FIELD_NUMBER;
-      hash = (53 * hash) + getDescription().hashCode();
-      hash = (37 * hash) + HAS_MINIMUM_FIELD_NUMBER;
-      hash = (53 * hash) + com.google.protobuf.Internal.hashBoolean(
-          getHasMinimum());
-      hash = (37 * hash) + MINIMUM_FIELD_NUMBER;
-      hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-          getMinimum());
-      if (hasAllowedValues()) {
-        hash = (37 * hash) + ALLOWED_VALUES_FIELD_NUMBER;
-        hash = (53 * hash) + getAllowedValues().hashCode();
-      }
-      hash = (29 * hash) + unknownFields.hashCode();
-      memoizedHashCode = hash;
-      return hash;
-    }
-
-    public static org.tensorflow.proto.framework.OpDef.AttrDef parseFrom(
-        java.nio.ByteBuffer data)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data);
-    }
-    public static org.tensorflow.proto.framework.OpDef.AttrDef parseFrom(
-        java.nio.ByteBuffer data,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.OpDef.AttrDef parseFrom(
-        com.google.protobuf.ByteString data)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data);
-    }
-    public static org.tensorflow.proto.framework.OpDef.AttrDef parseFrom(
-        com.google.protobuf.ByteString data,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.OpDef.AttrDef parseFrom(byte[] data)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data);
-    }
-    public static org.tensorflow.proto.framework.OpDef.AttrDef parseFrom(
-        byte[] data,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.OpDef.AttrDef parseFrom(java.io.InputStream input)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input);
-    }
-    public static org.tensorflow.proto.framework.OpDef.AttrDef parseFrom(
-        java.io.InputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.OpDef.AttrDef parseDelimitedFrom(java.io.InputStream input)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseDelimitedWithIOException(PARSER, input);
-    }
-    public static org.tensorflow.proto.framework.OpDef.AttrDef parseDelimitedFrom(
-        java.io.InputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.OpDef.AttrDef parseFrom(
-        com.google.protobuf.CodedInputStream input)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input);
-    }
-    public static org.tensorflow.proto.framework.OpDef.AttrDef parseFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input, extensionRegistry);
-    }
-
-    @java.lang.Override
-    public Builder newBuilderForType() { return newBuilder(); }
-    public static Builder newBuilder() {
-      return DEFAULT_INSTANCE.toBuilder();
-    }
-    public static Builder newBuilder(org.tensorflow.proto.framework.OpDef.AttrDef prototype) {
-      return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-    }
-    @java.lang.Override
-    public Builder toBuilder() {
-      return this == DEFAULT_INSTANCE
-          ? new Builder() : new Builder().mergeFrom(this);
-    }
-
-    @java.lang.Override
-    protected Builder newBuilderForType(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      Builder builder = new Builder(parent);
-      return builder;
-    }
-    /**
-     * <pre>
-     * Description of the graph-construction-time configuration of this
-     * Op.  That is to say, this describes the attr fields that will
-     * be specified in the NodeDef.
-     * </pre>
-     *
-     * Protobuf type {@code tensorflow.OpDef.AttrDef}
-     */
-    public static final class Builder extends
-        com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-        // @@protoc_insertion_point(builder_implements:tensorflow.OpDef.AttrDef)
-        org.tensorflow.proto.framework.OpDef.AttrDefOrBuilder {
-      public static final com.google.protobuf.Descriptors.Descriptor
-          getDescriptor() {
-        return org.tensorflow.proto.framework.OpDefProtos.internal_static_tensorflow_OpDef_AttrDef_descriptor;
-      }
-
-      @java.lang.Override
-      protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-          internalGetFieldAccessorTable() {
-        return org.tensorflow.proto.framework.OpDefProtos.internal_static_tensorflow_OpDef_AttrDef_fieldAccessorTable
-            .ensureFieldAccessorsInitialized(
-                org.tensorflow.proto.framework.OpDef.AttrDef.class, org.tensorflow.proto.framework.OpDef.AttrDef.Builder.class);
-      }
-
-      // Construct using org.tensorflow.proto.framework.OpDef.AttrDef.newBuilder()
-      private Builder() {
-        maybeForceBuilderInitialization();
-      }
-
-      private Builder(
-          com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-        super(parent);
-        maybeForceBuilderInitialization();
-      }
-      private void maybeForceBuilderInitialization() {
-        if (com.google.protobuf.GeneratedMessageV3
-                .alwaysUseFieldBuilders) {
-        }
-      }
-      @java.lang.Override
-      public Builder clear() {
-        super.clear();
-        name_ = "";
-
-        type_ = "";
-
-        if (defaultValueBuilder_ == null) {
-          defaultValue_ = null;
-        } else {
-          defaultValue_ = null;
-          defaultValueBuilder_ = null;
-        }
-        description_ = "";
-
-        hasMinimum_ = false;
-
-        minimum_ = 0L;
-
-        if (allowedValuesBuilder_ == null) {
-          allowedValues_ = null;
-        } else {
-          allowedValues_ = null;
-          allowedValuesBuilder_ = null;
-        }
-        return this;
-      }
-
-      @java.lang.Override
-      public com.google.protobuf.Descriptors.Descriptor
-          getDescriptorForType() {
-        return org.tensorflow.proto.framework.OpDefProtos.internal_static_tensorflow_OpDef_AttrDef_descriptor;
-      }
-
-      @java.lang.Override
-      public org.tensorflow.proto.framework.OpDef.AttrDef getDefaultInstanceForType() {
-        return org.tensorflow.proto.framework.OpDef.AttrDef.getDefaultInstance();
-      }
-
-      @java.lang.Override
-      public org.tensorflow.proto.framework.OpDef.AttrDef build() {
-        org.tensorflow.proto.framework.OpDef.AttrDef result = buildPartial();
-        if (!result.isInitialized()) {
-          throw newUninitializedMessageException(result);
-        }
-        return result;
-      }
-
-      @java.lang.Override
-      public org.tensorflow.proto.framework.OpDef.AttrDef buildPartial() {
-        org.tensorflow.proto.framework.OpDef.AttrDef result = new org.tensorflow.proto.framework.OpDef.AttrDef(this);
-        result.name_ = name_;
-        result.type_ = type_;
-        if (defaultValueBuilder_ == null) {
-          result.defaultValue_ = defaultValue_;
-        } else {
-          result.defaultValue_ = defaultValueBuilder_.build();
-        }
-        result.description_ = description_;
-        result.hasMinimum_ = hasMinimum_;
-        result.minimum_ = minimum_;
-        if (allowedValuesBuilder_ == null) {
-          result.allowedValues_ = allowedValues_;
-        } else {
-          result.allowedValues_ = allowedValuesBuilder_.build();
-        }
-        onBuilt();
-        return result;
-      }
-
-      @java.lang.Override
-      public Builder clone() {
-        return super.clone();
-      }
-      @java.lang.Override
-      public Builder setField(
-          com.google.protobuf.Descriptors.FieldDescriptor field,
-          java.lang.Object value) {
-        return super.setField(field, value);
-      }
-      @java.lang.Override
-      public Builder clearField(
-          com.google.protobuf.Descriptors.FieldDescriptor field) {
-        return super.clearField(field);
-      }
-      @java.lang.Override
-      public Builder clearOneof(
-          com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-        return super.clearOneof(oneof);
-      }
-      @java.lang.Override
-      public Builder setRepeatedField(
-          com.google.protobuf.Descriptors.FieldDescriptor field,
-          int index, java.lang.Object value) {
-        return super.setRepeatedField(field, index, value);
-      }
-      @java.lang.Override
-      public Builder addRepeatedField(
-          com.google.protobuf.Descriptors.FieldDescriptor field,
-          java.lang.Object value) {
-        return super.addRepeatedField(field, value);
-      }
-      @java.lang.Override
-      public Builder mergeFrom(com.google.protobuf.Message other) {
-        if (other instanceof org.tensorflow.proto.framework.OpDef.AttrDef) {
-          return mergeFrom((org.tensorflow.proto.framework.OpDef.AttrDef)other);
-        } else {
-          super.mergeFrom(other);
-          return this;
-        }
-      }
-
-      public Builder mergeFrom(org.tensorflow.proto.framework.OpDef.AttrDef other) {
-        if (other == org.tensorflow.proto.framework.OpDef.AttrDef.getDefaultInstance()) return this;
-        if (!other.getName().isEmpty()) {
-          name_ = other.name_;
-          onChanged();
-        }
-        if (!other.getType().isEmpty()) {
-          type_ = other.type_;
-          onChanged();
-        }
-        if (other.hasDefaultValue()) {
-          mergeDefaultValue(other.getDefaultValue());
-        }
-        if (!other.getDescription().isEmpty()) {
-          description_ = other.description_;
-          onChanged();
-        }
-        if (other.getHasMinimum() != false) {
-          setHasMinimum(other.getHasMinimum());
-        }
-        if (other.getMinimum() != 0L) {
-          setMinimum(other.getMinimum());
-        }
-        if (other.hasAllowedValues()) {
-          mergeAllowedValues(other.getAllowedValues());
-        }
-        this.mergeUnknownFields(other.unknownFields);
-        onChanged();
-        return this;
-      }
-
-      @java.lang.Override
-      public final boolean isInitialized() {
-        return true;
-      }
-
-      @java.lang.Override
-      public Builder mergeFrom(
-          com.google.protobuf.CodedInputStream input,
-          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-          throws java.io.IOException {
-        org.tensorflow.proto.framework.OpDef.AttrDef parsedMessage = null;
-        try {
-          parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-        } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-          parsedMessage = (org.tensorflow.proto.framework.OpDef.AttrDef) e.getUnfinishedMessage();
-          throw e.unwrapIOException();
-        } finally {
-          if (parsedMessage != null) {
-            mergeFrom(parsedMessage);
-          }
-        }
-        return this;
-      }
-
-      private java.lang.Object name_ = "";
-      /**
-       * <pre>
-       * A descriptive name for the argument.  May be used, e.g. by the
-       * Python client, as a keyword argument name, and so should match
-       * the regexp "[a-z][a-z0-9_]+".
-       * </pre>
-       *
-       * <code>string name = 1;</code>
-       */
-      public java.lang.String getName() {
-        java.lang.Object ref = name_;
-        if (!(ref instanceof java.lang.String)) {
-          com.google.protobuf.ByteString bs =
-              (com.google.protobuf.ByteString) ref;
-          java.lang.String s = bs.toStringUtf8();
-          name_ = s;
-          return s;
-        } else {
-          return (java.lang.String) ref;
-        }
-      }
-      /**
-       * <pre>
-       * A descriptive name for the argument.  May be used, e.g. by the
-       * Python client, as a keyword argument name, and so should match
-       * the regexp "[a-z][a-z0-9_]+".
-       * </pre>
-       *
-       * <code>string name = 1;</code>
-       */
-      public com.google.protobuf.ByteString
-          getNameBytes() {
-        java.lang.Object ref = name_;
-        if (ref instanceof String) {
-          com.google.protobuf.ByteString b = 
-              com.google.protobuf.ByteString.copyFromUtf8(
-                  (java.lang.String) ref);
-          name_ = b;
-          return b;
-        } else {
-          return (com.google.protobuf.ByteString) ref;
-        }
-      }
-      /**
-       * <pre>
-       * A descriptive name for the argument.  May be used, e.g. by the
-       * Python client, as a keyword argument name, and so should match
-       * the regexp "[a-z][a-z0-9_]+".
-       * </pre>
-       *
-       * <code>string name = 1;</code>
-       */
-      public Builder setName(
-          java.lang.String value) {
-        if (value == null) {
-    throw new NullPointerException();
-  }
-  
-        name_ = value;
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * A descriptive name for the argument.  May be used, e.g. by the
-       * Python client, as a keyword argument name, and so should match
-       * the regexp "[a-z][a-z0-9_]+".
-       * </pre>
-       *
-       * <code>string name = 1;</code>
-       */
-      public Builder clearName() {
-        
-        name_ = getDefaultInstance().getName();
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * A descriptive name for the argument.  May be used, e.g. by the
-       * Python client, as a keyword argument name, and so should match
-       * the regexp "[a-z][a-z0-9_]+".
-       * </pre>
-       *
-       * <code>string name = 1;</code>
-       */
-      public Builder setNameBytes(
-          com.google.protobuf.ByteString value) {
-        if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-        
-        name_ = value;
-        onChanged();
-        return this;
-      }
-
-      private java.lang.Object type_ = "";
-      /**
-       * <pre>
-       * One of the type names from attr_value.proto ("string", "list(string)",
-       * "int", etc.).
-       * </pre>
-       *
-       * <code>string type = 2;</code>
-       */
-      public java.lang.String getType() {
-        java.lang.Object ref = type_;
-        if (!(ref instanceof java.lang.String)) {
-          com.google.protobuf.ByteString bs =
-              (com.google.protobuf.ByteString) ref;
-          java.lang.String s = bs.toStringUtf8();
-          type_ = s;
-          return s;
-        } else {
-          return (java.lang.String) ref;
-        }
-      }
-      /**
-       * <pre>
-       * One of the type names from attr_value.proto ("string", "list(string)",
-       * "int", etc.).
-       * </pre>
-       *
-       * <code>string type = 2;</code>
-       */
-      public com.google.protobuf.ByteString
-          getTypeBytes() {
-        java.lang.Object ref = type_;
-        if (ref instanceof String) {
-          com.google.protobuf.ByteString b = 
-              com.google.protobuf.ByteString.copyFromUtf8(
-                  (java.lang.String) ref);
-          type_ = b;
-          return b;
-        } else {
-          return (com.google.protobuf.ByteString) ref;
-        }
-      }
-      /**
-       * <pre>
-       * One of the type names from attr_value.proto ("string", "list(string)",
-       * "int", etc.).
-       * </pre>
-       *
-       * <code>string type = 2;</code>
-       */
-      public Builder setType(
-          java.lang.String value) {
-        if (value == null) {
-    throw new NullPointerException();
-  }
-  
-        type_ = value;
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * One of the type names from attr_value.proto ("string", "list(string)",
-       * "int", etc.).
-       * </pre>
-       *
-       * <code>string type = 2;</code>
-       */
-      public Builder clearType() {
-        
-        type_ = getDefaultInstance().getType();
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * One of the type names from attr_value.proto ("string", "list(string)",
-       * "int", etc.).
-       * </pre>
-       *
-       * <code>string type = 2;</code>
-       */
-      public Builder setTypeBytes(
-          com.google.protobuf.ByteString value) {
-        if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-        
-        type_ = value;
-        onChanged();
-        return this;
-      }
-
-      private org.tensorflow.proto.framework.AttrValue defaultValue_;
-      private com.google.protobuf.SingleFieldBuilderV3<
-          org.tensorflow.proto.framework.AttrValue, org.tensorflow.proto.framework.AttrValue.Builder, org.tensorflow.proto.framework.AttrValueOrBuilder> defaultValueBuilder_;
-      /**
-       * <pre>
-       * A reasonable default for this attribute if the user does not supply
-       * a value.  If not specified, the user must supply a value.
-       * </pre>
-       *
-       * <code>.tensorflow.AttrValue default_value = 3;</code>
-       */
-      public boolean hasDefaultValue() {
-        return defaultValueBuilder_ != null || defaultValue_ != null;
-      }
-      /**
-       * <pre>
-       * A reasonable default for this attribute if the user does not supply
-       * a value.  If not specified, the user must supply a value.
-       * </pre>
-       *
-       * <code>.tensorflow.AttrValue default_value = 3;</code>
-       */
-      public org.tensorflow.proto.framework.AttrValue getDefaultValue() {
-        if (defaultValueBuilder_ == null) {
-          return defaultValue_ == null ? org.tensorflow.proto.framework.AttrValue.getDefaultInstance() : defaultValue_;
-        } else {
-          return defaultValueBuilder_.getMessage();
-        }
-      }
-      /**
-       * <pre>
-       * A reasonable default for this attribute if the user does not supply
-       * a value.  If not specified, the user must supply a value.
-       * </pre>
-       *
-       * <code>.tensorflow.AttrValue default_value = 3;</code>
-       */
-      public Builder setDefaultValue(org.tensorflow.proto.framework.AttrValue value) {
-        if (defaultValueBuilder_ == null) {
-          if (value == null) {
-            throw new NullPointerException();
-          }
-          defaultValue_ = value;
-          onChanged();
-        } else {
-          defaultValueBuilder_.setMessage(value);
-        }
-
-        return this;
-      }
-      /**
-       * <pre>
-       * A reasonable default for this attribute if the user does not supply
-       * a value.  If not specified, the user must supply a value.
-       * </pre>
-       *
-       * <code>.tensorflow.AttrValue default_value = 3;</code>
-       */
-      public Builder setDefaultValue(
-          org.tensorflow.proto.framework.AttrValue.Builder builderForValue) {
-        if (defaultValueBuilder_ == null) {
-          defaultValue_ = builderForValue.build();
-          onChanged();
-        } else {
-          defaultValueBuilder_.setMessage(builderForValue.build());
-        }
-
-        return this;
-      }
-      /**
-       * <pre>
-       * A reasonable default for this attribute if the user does not supply
-       * a value.  If not specified, the user must supply a value.
-       * </pre>
-       *
-       * <code>.tensorflow.AttrValue default_value = 3;</code>
-       */
-      public Builder mergeDefaultValue(org.tensorflow.proto.framework.AttrValue value) {
-        if (defaultValueBuilder_ == null) {
-          if (defaultValue_ != null) {
-            defaultValue_ =
-              org.tensorflow.proto.framework.AttrValue.newBuilder(defaultValue_).mergeFrom(value).buildPartial();
-          } else {
-            defaultValue_ = value;
-          }
-          onChanged();
-        } else {
-          defaultValueBuilder_.mergeFrom(value);
-        }
-
-        return this;
-      }
-      /**
-       * <pre>
-       * A reasonable default for this attribute if the user does not supply
-       * a value.  If not specified, the user must supply a value.
-       * </pre>
-       *
-       * <code>.tensorflow.AttrValue default_value = 3;</code>
-       */
-      public Builder clearDefaultValue() {
-        if (defaultValueBuilder_ == null) {
-          defaultValue_ = null;
-          onChanged();
-        } else {
-          defaultValue_ = null;
-          defaultValueBuilder_ = null;
-        }
-
-        return this;
-      }
-      /**
-       * <pre>
-       * A reasonable default for this attribute if the user does not supply
-       * a value.  If not specified, the user must supply a value.
-       * </pre>
-       *
-       * <code>.tensorflow.AttrValue default_value = 3;</code>
-       */
-      public org.tensorflow.proto.framework.AttrValue.Builder getDefaultValueBuilder() {
-        
-        onChanged();
-        return getDefaultValueFieldBuilder().getBuilder();
-      }
-      /**
-       * <pre>
-       * A reasonable default for this attribute if the user does not supply
-       * a value.  If not specified, the user must supply a value.
-       * </pre>
-       *
-       * <code>.tensorflow.AttrValue default_value = 3;</code>
-       */
-      public org.tensorflow.proto.framework.AttrValueOrBuilder getDefaultValueOrBuilder() {
-        if (defaultValueBuilder_ != null) {
-          return defaultValueBuilder_.getMessageOrBuilder();
-        } else {
-          return defaultValue_ == null ?
-              org.tensorflow.proto.framework.AttrValue.getDefaultInstance() : defaultValue_;
-        }
-      }
-      /**
-       * <pre>
-       * A reasonable default for this attribute if the user does not supply
-       * a value.  If not specified, the user must supply a value.
-       * </pre>
-       *
-       * <code>.tensorflow.AttrValue default_value = 3;</code>
-       */
-      private com.google.protobuf.SingleFieldBuilderV3<
-          org.tensorflow.proto.framework.AttrValue, org.tensorflow.proto.framework.AttrValue.Builder, org.tensorflow.proto.framework.AttrValueOrBuilder> 
-          getDefaultValueFieldBuilder() {
-        if (defaultValueBuilder_ == null) {
-          defaultValueBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-              org.tensorflow.proto.framework.AttrValue, org.tensorflow.proto.framework.AttrValue.Builder, org.tensorflow.proto.framework.AttrValueOrBuilder>(
-                  getDefaultValue(),
-                  getParentForChildren(),
-                  isClean());
-          defaultValue_ = null;
-        }
-        return defaultValueBuilder_;
-      }
-
-      private java.lang.Object description_ = "";
-      /**
-       * <pre>
-       * Human-readable description.
-       * </pre>
-       *
-       * <code>string description = 4;</code>
-       */
-      public java.lang.String getDescription() {
-        java.lang.Object ref = description_;
-        if (!(ref instanceof java.lang.String)) {
-          com.google.protobuf.ByteString bs =
-              (com.google.protobuf.ByteString) ref;
-          java.lang.String s = bs.toStringUtf8();
-          description_ = s;
-          return s;
-        } else {
-          return (java.lang.String) ref;
-        }
-      }
-      /**
-       * <pre>
-       * Human-readable description.
-       * </pre>
-       *
-       * <code>string description = 4;</code>
-       */
-      public com.google.protobuf.ByteString
-          getDescriptionBytes() {
-        java.lang.Object ref = description_;
-        if (ref instanceof String) {
-          com.google.protobuf.ByteString b = 
-              com.google.protobuf.ByteString.copyFromUtf8(
-                  (java.lang.String) ref);
-          description_ = b;
-          return b;
-        } else {
-          return (com.google.protobuf.ByteString) ref;
-        }
-      }
-      /**
-       * <pre>
-       * Human-readable description.
-       * </pre>
-       *
-       * <code>string description = 4;</code>
-       */
-      public Builder setDescription(
-          java.lang.String value) {
-        if (value == null) {
-    throw new NullPointerException();
-  }
-  
-        description_ = value;
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * Human-readable description.
-       * </pre>
-       *
-       * <code>string description = 4;</code>
-       */
-      public Builder clearDescription() {
-        
-        description_ = getDefaultInstance().getDescription();
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * Human-readable description.
-       * </pre>
-       *
-       * <code>string description = 4;</code>
-       */
-      public Builder setDescriptionBytes(
-          com.google.protobuf.ByteString value) {
-        if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-        
-        description_ = value;
-        onChanged();
-        return this;
-      }
-
-      private boolean hasMinimum_ ;
-      /**
-       * <pre>
-       * For type == "int", this is a minimum value.  For "list(___)"
-       * types, this is the minimum length.
-       * </pre>
-       *
-       * <code>bool has_minimum = 5;</code>
-       */
-      public boolean getHasMinimum() {
-        return hasMinimum_;
-      }
-      /**
-       * <pre>
-       * For type == "int", this is a minimum value.  For "list(___)"
-       * types, this is the minimum length.
-       * </pre>
-       *
-       * <code>bool has_minimum = 5;</code>
-       */
-      public Builder setHasMinimum(boolean value) {
-        
-        hasMinimum_ = value;
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * For type == "int", this is a minimum value.  For "list(___)"
-       * types, this is the minimum length.
-       * </pre>
-       *
-       * <code>bool has_minimum = 5;</code>
-       */
-      public Builder clearHasMinimum() {
-        
-        hasMinimum_ = false;
-        onChanged();
-        return this;
-      }
-
-      private long minimum_ ;
-      /**
-       * <code>int64 minimum = 6;</code>
-       */
-      public long getMinimum() {
-        return minimum_;
-      }
-      /**
-       * <code>int64 minimum = 6;</code>
-       */
-      public Builder setMinimum(long value) {
-        
-        minimum_ = value;
-        onChanged();
-        return this;
-      }
-      /**
-       * <code>int64 minimum = 6;</code>
-       */
-      public Builder clearMinimum() {
-        
-        minimum_ = 0L;
-        onChanged();
-        return this;
-      }
-
-      private org.tensorflow.proto.framework.AttrValue allowedValues_;
-      private com.google.protobuf.SingleFieldBuilderV3<
-          org.tensorflow.proto.framework.AttrValue, org.tensorflow.proto.framework.AttrValue.Builder, org.tensorflow.proto.framework.AttrValueOrBuilder> allowedValuesBuilder_;
-      /**
-       * <pre>
-       * The set of allowed values.  Has type that is the "list" version
-       * of the "type" field above (uses the "list" field of AttrValue).
-       * If type == "type" or "list(type)" above, then the "type" field
-       * of "allowed_values.list" has the set of allowed DataTypes.
-       * If type == "string" or "list(string)", then the "s" field of
-       * "allowed_values.list" has the set of allowed strings.
-       * </pre>
-       *
-       * <code>.tensorflow.AttrValue allowed_values = 7;</code>
-       */
-      public boolean hasAllowedValues() {
-        return allowedValuesBuilder_ != null || allowedValues_ != null;
-      }
-      /**
-       * <pre>
-       * The set of allowed values.  Has type that is the "list" version
-       * of the "type" field above (uses the "list" field of AttrValue).
-       * If type == "type" or "list(type)" above, then the "type" field
-       * of "allowed_values.list" has the set of allowed DataTypes.
-       * If type == "string" or "list(string)", then the "s" field of
-       * "allowed_values.list" has the set of allowed strings.
-       * </pre>
-       *
-       * <code>.tensorflow.AttrValue allowed_values = 7;</code>
-       */
-      public org.tensorflow.proto.framework.AttrValue getAllowedValues() {
-        if (allowedValuesBuilder_ == null) {
-          return allowedValues_ == null ? org.tensorflow.proto.framework.AttrValue.getDefaultInstance() : allowedValues_;
-        } else {
-          return allowedValuesBuilder_.getMessage();
-        }
-      }
-      /**
-       * <pre>
-       * The set of allowed values.  Has type that is the "list" version
-       * of the "type" field above (uses the "list" field of AttrValue).
-       * If type == "type" or "list(type)" above, then the "type" field
-       * of "allowed_values.list" has the set of allowed DataTypes.
-       * If type == "string" or "list(string)", then the "s" field of
-       * "allowed_values.list" has the set of allowed strings.
-       * </pre>
-       *
-       * <code>.tensorflow.AttrValue allowed_values = 7;</code>
-       */
-      public Builder setAllowedValues(org.tensorflow.proto.framework.AttrValue value) {
-        if (allowedValuesBuilder_ == null) {
-          if (value == null) {
-            throw new NullPointerException();
-          }
-          allowedValues_ = value;
-          onChanged();
-        } else {
-          allowedValuesBuilder_.setMessage(value);
-        }
-
-        return this;
-      }
-      /**
-       * <pre>
-       * The set of allowed values.  Has type that is the "list" version
-       * of the "type" field above (uses the "list" field of AttrValue).
-       * If type == "type" or "list(type)" above, then the "type" field
-       * of "allowed_values.list" has the set of allowed DataTypes.
-       * If type == "string" or "list(string)", then the "s" field of
-       * "allowed_values.list" has the set of allowed strings.
-       * </pre>
-       *
-       * <code>.tensorflow.AttrValue allowed_values = 7;</code>
-       */
-      public Builder setAllowedValues(
-          org.tensorflow.proto.framework.AttrValue.Builder builderForValue) {
-        if (allowedValuesBuilder_ == null) {
-          allowedValues_ = builderForValue.build();
-          onChanged();
-        } else {
-          allowedValuesBuilder_.setMessage(builderForValue.build());
-        }
-
-        return this;
-      }
-      /**
-       * <pre>
-       * The set of allowed values.  Has type that is the "list" version
-       * of the "type" field above (uses the "list" field of AttrValue).
-       * If type == "type" or "list(type)" above, then the "type" field
-       * of "allowed_values.list" has the set of allowed DataTypes.
-       * If type == "string" or "list(string)", then the "s" field of
-       * "allowed_values.list" has the set of allowed strings.
-       * </pre>
-       *
-       * <code>.tensorflow.AttrValue allowed_values = 7;</code>
-       */
-      public Builder mergeAllowedValues(org.tensorflow.proto.framework.AttrValue value) {
-        if (allowedValuesBuilder_ == null) {
-          if (allowedValues_ != null) {
-            allowedValues_ =
-              org.tensorflow.proto.framework.AttrValue.newBuilder(allowedValues_).mergeFrom(value).buildPartial();
-          } else {
-            allowedValues_ = value;
-          }
-          onChanged();
-        } else {
-          allowedValuesBuilder_.mergeFrom(value);
-        }
-
-        return this;
-      }
-      /**
-       * <pre>
-       * The set of allowed values.  Has type that is the "list" version
-       * of the "type" field above (uses the "list" field of AttrValue).
-       * If type == "type" or "list(type)" above, then the "type" field
-       * of "allowed_values.list" has the set of allowed DataTypes.
-       * If type == "string" or "list(string)", then the "s" field of
-       * "allowed_values.list" has the set of allowed strings.
-       * </pre>
-       *
-       * <code>.tensorflow.AttrValue allowed_values = 7;</code>
-       */
-      public Builder clearAllowedValues() {
-        if (allowedValuesBuilder_ == null) {
-          allowedValues_ = null;
-          onChanged();
-        } else {
-          allowedValues_ = null;
-          allowedValuesBuilder_ = null;
-        }
-
-        return this;
-      }
-      /**
-       * <pre>
-       * The set of allowed values.  Has type that is the "list" version
-       * of the "type" field above (uses the "list" field of AttrValue).
-       * If type == "type" or "list(type)" above, then the "type" field
-       * of "allowed_values.list" has the set of allowed DataTypes.
-       * If type == "string" or "list(string)", then the "s" field of
-       * "allowed_values.list" has the set of allowed strings.
-       * </pre>
-       *
-       * <code>.tensorflow.AttrValue allowed_values = 7;</code>
-       */
-      public org.tensorflow.proto.framework.AttrValue.Builder getAllowedValuesBuilder() {
-        
-        onChanged();
-        return getAllowedValuesFieldBuilder().getBuilder();
-      }
-      /**
-       * <pre>
-       * The set of allowed values.  Has type that is the "list" version
-       * of the "type" field above (uses the "list" field of AttrValue).
-       * If type == "type" or "list(type)" above, then the "type" field
-       * of "allowed_values.list" has the set of allowed DataTypes.
-       * If type == "string" or "list(string)", then the "s" field of
-       * "allowed_values.list" has the set of allowed strings.
-       * </pre>
-       *
-       * <code>.tensorflow.AttrValue allowed_values = 7;</code>
-       */
-      public org.tensorflow.proto.framework.AttrValueOrBuilder getAllowedValuesOrBuilder() {
-        if (allowedValuesBuilder_ != null) {
-          return allowedValuesBuilder_.getMessageOrBuilder();
-        } else {
-          return allowedValues_ == null ?
-              org.tensorflow.proto.framework.AttrValue.getDefaultInstance() : allowedValues_;
-        }
-      }
-      /**
-       * <pre>
-       * The set of allowed values.  Has type that is the "list" version
-       * of the "type" field above (uses the "list" field of AttrValue).
-       * If type == "type" or "list(type)" above, then the "type" field
-       * of "allowed_values.list" has the set of allowed DataTypes.
-       * If type == "string" or "list(string)", then the "s" field of
-       * "allowed_values.list" has the set of allowed strings.
-       * </pre>
-       *
-       * <code>.tensorflow.AttrValue allowed_values = 7;</code>
-       */
-      private com.google.protobuf.SingleFieldBuilderV3<
-          org.tensorflow.proto.framework.AttrValue, org.tensorflow.proto.framework.AttrValue.Builder, org.tensorflow.proto.framework.AttrValueOrBuilder> 
-          getAllowedValuesFieldBuilder() {
-        if (allowedValuesBuilder_ == null) {
-          allowedValuesBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-              org.tensorflow.proto.framework.AttrValue, org.tensorflow.proto.framework.AttrValue.Builder, org.tensorflow.proto.framework.AttrValueOrBuilder>(
-                  getAllowedValues(),
-                  getParentForChildren(),
-                  isClean());
-          allowedValues_ = null;
-        }
-        return allowedValuesBuilder_;
-      }
-      @java.lang.Override
-      public final Builder setUnknownFields(
-          final com.google.protobuf.UnknownFieldSet unknownFields) {
-        return super.setUnknownFields(unknownFields);
-      }
-
-      @java.lang.Override
-      public final Builder mergeUnknownFields(
-          final com.google.protobuf.UnknownFieldSet unknownFields) {
-        return super.mergeUnknownFields(unknownFields);
-      }
-
-
-      // @@protoc_insertion_point(builder_scope:tensorflow.OpDef.AttrDef)
-    }
-
-    // @@protoc_insertion_point(class_scope:tensorflow.OpDef.AttrDef)
-    private static final org.tensorflow.proto.framework.OpDef.AttrDef DEFAULT_INSTANCE;
-    static {
-      DEFAULT_INSTANCE = new org.tensorflow.proto.framework.OpDef.AttrDef();
-    }
-
-    public static org.tensorflow.proto.framework.OpDef.AttrDef getDefaultInstance() {
-      return DEFAULT_INSTANCE;
-    }
-
-    private static final com.google.protobuf.Parser<AttrDef>
-        PARSER = new com.google.protobuf.AbstractParser<AttrDef>() {
-      @java.lang.Override
-      public AttrDef parsePartialFrom(
-          com.google.protobuf.CodedInputStream input,
-          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-          throws com.google.protobuf.InvalidProtocolBufferException {
-        return new AttrDef(input, extensionRegistry);
-      }
-    };
-
-    public static com.google.protobuf.Parser<AttrDef> parser() {
-      return PARSER;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Parser<AttrDef> getParserForType() {
-      return PARSER;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.OpDef.AttrDef getDefaultInstanceForType() {
-      return DEFAULT_INSTANCE;
-    }
-
-  }
-
-  public static final int NAME_FIELD_NUMBER = 1;
-  private volatile java.lang.Object name_;
-  /**
-   * <pre>
-   * Op names starting with an underscore are reserved for internal use.
-   * Names should be CamelCase and match the regexp "[A-Z][a-zA-Z0-9&gt;_]*".
-   * </pre>
-   *
-   * <code>string name = 1;</code>
-   */
-  public java.lang.String getName() {
-    java.lang.Object ref = name_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      name_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * Op names starting with an underscore are reserved for internal use.
-   * Names should be CamelCase and match the regexp "[A-Z][a-zA-Z0-9&gt;_]*".
-   * </pre>
-   *
-   * <code>string name = 1;</code>
-   */
-  public com.google.protobuf.ByteString
-      getNameBytes() {
-    java.lang.Object ref = name_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      name_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int INPUT_ARG_FIELD_NUMBER = 2;
-  private java.util.List<org.tensorflow.proto.framework.OpDef.ArgDef> inputArg_;
-  /**
-   * <pre>
-   * Description of the input(s).
-   * </pre>
-   *
-   * <code>repeated .tensorflow.OpDef.ArgDef input_arg = 2;</code>
-   */
-  public java.util.List<org.tensorflow.proto.framework.OpDef.ArgDef> getInputArgList() {
-    return inputArg_;
-  }
-  /**
-   * <pre>
-   * Description of the input(s).
-   * </pre>
-   *
-   * <code>repeated .tensorflow.OpDef.ArgDef input_arg = 2;</code>
-   */
-  public java.util.List<? extends org.tensorflow.proto.framework.OpDef.ArgDefOrBuilder> 
-      getInputArgOrBuilderList() {
-    return inputArg_;
-  }
-  /**
-   * <pre>
-   * Description of the input(s).
-   * </pre>
-   *
-   * <code>repeated .tensorflow.OpDef.ArgDef input_arg = 2;</code>
-   */
-  public int getInputArgCount() {
-    return inputArg_.size();
-  }
-  /**
-   * <pre>
-   * Description of the input(s).
-   * </pre>
-   *
-   * <code>repeated .tensorflow.OpDef.ArgDef input_arg = 2;</code>
-   */
-  public org.tensorflow.proto.framework.OpDef.ArgDef getInputArg(int index) {
-    return inputArg_.get(index);
-  }
-  /**
-   * <pre>
-   * Description of the input(s).
-   * </pre>
-   *
-   * <code>repeated .tensorflow.OpDef.ArgDef input_arg = 2;</code>
-   */
-  public org.tensorflow.proto.framework.OpDef.ArgDefOrBuilder getInputArgOrBuilder(
-      int index) {
-    return inputArg_.get(index);
-  }
-
-  public static final int OUTPUT_ARG_FIELD_NUMBER = 3;
-  private java.util.List<org.tensorflow.proto.framework.OpDef.ArgDef> outputArg_;
-  /**
-   * <pre>
-   * Description of the output(s).
-   * </pre>
-   *
-   * <code>repeated .tensorflow.OpDef.ArgDef output_arg = 3;</code>
-   */
-  public java.util.List<org.tensorflow.proto.framework.OpDef.ArgDef> getOutputArgList() {
-    return outputArg_;
-  }
-  /**
-   * <pre>
-   * Description of the output(s).
-   * </pre>
-   *
-   * <code>repeated .tensorflow.OpDef.ArgDef output_arg = 3;</code>
-   */
-  public java.util.List<? extends org.tensorflow.proto.framework.OpDef.ArgDefOrBuilder> 
-      getOutputArgOrBuilderList() {
-    return outputArg_;
-  }
-  /**
-   * <pre>
-   * Description of the output(s).
-   * </pre>
-   *
-   * <code>repeated .tensorflow.OpDef.ArgDef output_arg = 3;</code>
-   */
-  public int getOutputArgCount() {
-    return outputArg_.size();
-  }
-  /**
-   * <pre>
-   * Description of the output(s).
-   * </pre>
-   *
-   * <code>repeated .tensorflow.OpDef.ArgDef output_arg = 3;</code>
-   */
-  public org.tensorflow.proto.framework.OpDef.ArgDef getOutputArg(int index) {
-    return outputArg_.get(index);
-  }
-  /**
-   * <pre>
-   * Description of the output(s).
-   * </pre>
-   *
-   * <code>repeated .tensorflow.OpDef.ArgDef output_arg = 3;</code>
-   */
-  public org.tensorflow.proto.framework.OpDef.ArgDefOrBuilder getOutputArgOrBuilder(
-      int index) {
-    return outputArg_.get(index);
-  }
-
-  public static final int CONTROL_OUTPUT_FIELD_NUMBER = 20;
-  private com.google.protobuf.LazyStringList controlOutput_;
-  /**
-   * <pre>
-   * Named control outputs for this operation. Useful only for composite
-   * operations (i.e. functions) which want to name different control outputs.
-   * </pre>
-   *
-   * <code>repeated string control_output = 20;</code>
-   */
-  public com.google.protobuf.ProtocolStringList
-      getControlOutputList() {
-    return controlOutput_;
-  }
-  /**
-   * <pre>
-   * Named control outputs for this operation. Useful only for composite
-   * operations (i.e. functions) which want to name different control outputs.
-   * </pre>
-   *
-   * <code>repeated string control_output = 20;</code>
-   */
-  public int getControlOutputCount() {
-    return controlOutput_.size();
-  }
-  /**
-   * <pre>
-   * Named control outputs for this operation. Useful only for composite
-   * operations (i.e. functions) which want to name different control outputs.
-   * </pre>
-   *
-   * <code>repeated string control_output = 20;</code>
-   */
-  public java.lang.String getControlOutput(int index) {
-    return controlOutput_.get(index);
-  }
-  /**
-   * <pre>
-   * Named control outputs for this operation. Useful only for composite
-   * operations (i.e. functions) which want to name different control outputs.
-   * </pre>
-   *
-   * <code>repeated string control_output = 20;</code>
-   */
-  public com.google.protobuf.ByteString
-      getControlOutputBytes(int index) {
-    return controlOutput_.getByteString(index);
-  }
-
-  public static final int ATTR_FIELD_NUMBER = 4;
-  private java.util.List<org.tensorflow.proto.framework.OpDef.AttrDef> attr_;
-  /**
-   * <code>repeated .tensorflow.OpDef.AttrDef attr = 4;</code>
-   */
-  public java.util.List<org.tensorflow.proto.framework.OpDef.AttrDef> getAttrList() {
-    return attr_;
-  }
-  /**
-   * <code>repeated .tensorflow.OpDef.AttrDef attr = 4;</code>
-   */
-  public java.util.List<? extends org.tensorflow.proto.framework.OpDef.AttrDefOrBuilder> 
-      getAttrOrBuilderList() {
-    return attr_;
-  }
-  /**
-   * <code>repeated .tensorflow.OpDef.AttrDef attr = 4;</code>
-   */
-  public int getAttrCount() {
-    return attr_.size();
-  }
-  /**
-   * <code>repeated .tensorflow.OpDef.AttrDef attr = 4;</code>
-   */
-  public org.tensorflow.proto.framework.OpDef.AttrDef getAttr(int index) {
-    return attr_.get(index);
-  }
-  /**
-   * <code>repeated .tensorflow.OpDef.AttrDef attr = 4;</code>
-   */
-  public org.tensorflow.proto.framework.OpDef.AttrDefOrBuilder getAttrOrBuilder(
-      int index) {
-    return attr_.get(index);
-  }
-
-  public static final int DEPRECATION_FIELD_NUMBER = 8;
-  private org.tensorflow.proto.framework.OpDeprecation deprecation_;
-  /**
-   * <pre>
-   * Optional deprecation based on GraphDef versions.
-   * </pre>
-   *
-   * <code>.tensorflow.OpDeprecation deprecation = 8;</code>
-   */
-  public boolean hasDeprecation() {
-    return deprecation_ != null;
-  }
-  /**
-   * <pre>
-   * Optional deprecation based on GraphDef versions.
-   * </pre>
-   *
-   * <code>.tensorflow.OpDeprecation deprecation = 8;</code>
-   */
-  public org.tensorflow.proto.framework.OpDeprecation getDeprecation() {
-    return deprecation_ == null ? org.tensorflow.proto.framework.OpDeprecation.getDefaultInstance() : deprecation_;
-  }
-  /**
-   * <pre>
-   * Optional deprecation based on GraphDef versions.
-   * </pre>
-   *
-   * <code>.tensorflow.OpDeprecation deprecation = 8;</code>
-   */
-  public org.tensorflow.proto.framework.OpDeprecationOrBuilder getDeprecationOrBuilder() {
-    return getDeprecation();
-  }
-
-  public static final int SUMMARY_FIELD_NUMBER = 5;
-  private volatile java.lang.Object summary_;
-  /**
-   * <pre>
-   * One-line human-readable description of what the Op does.
-   * </pre>
-   *
-   * <code>string summary = 5;</code>
-   */
-  public java.lang.String getSummary() {
-    java.lang.Object ref = summary_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      summary_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * One-line human-readable description of what the Op does.
-   * </pre>
-   *
-   * <code>string summary = 5;</code>
-   */
-  public com.google.protobuf.ByteString
-      getSummaryBytes() {
-    java.lang.Object ref = summary_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      summary_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int DESCRIPTION_FIELD_NUMBER = 6;
-  private volatile java.lang.Object description_;
-  /**
-   * <pre>
-   * Additional, longer human-readable description of what the Op does.
-   * </pre>
-   *
-   * <code>string description = 6;</code>
-   */
-  public java.lang.String getDescription() {
-    java.lang.Object ref = description_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      description_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * Additional, longer human-readable description of what the Op does.
-   * </pre>
-   *
-   * <code>string description = 6;</code>
-   */
-  public com.google.protobuf.ByteString
-      getDescriptionBytes() {
-    java.lang.Object ref = description_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      description_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int IS_COMMUTATIVE_FIELD_NUMBER = 18;
-  private boolean isCommutative_;
-  /**
-   * <pre>
-   * True if the operation is commutative ("op(a,b) == op(b,a)" for all inputs)
-   * </pre>
-   *
-   * <code>bool is_commutative = 18;</code>
-   */
-  public boolean getIsCommutative() {
-    return isCommutative_;
-  }
-
-  public static final int IS_AGGREGATE_FIELD_NUMBER = 16;
-  private boolean isAggregate_;
-  /**
-   * <pre>
-   * If is_aggregate is true, then this operation accepts N &gt;= 2
-   * inputs and produces 1 output all of the same type.  Should be
-   * associative and commutative, and produce output with the same
-   * shape as the input.  The optimizer may replace an aggregate op
-   * taking input from multiple devices with a tree of aggregate ops
-   * that aggregate locally within each device (and possibly within
-   * groups of nearby devices) before communicating.
-   * TODO(josh11b): Implement that optimization.
-   * </pre>
-   *
-   * <code>bool is_aggregate = 16;</code>
-   */
-  public boolean getIsAggregate() {
-    return isAggregate_;
-  }
-
-  public static final int IS_STATEFUL_FIELD_NUMBER = 17;
-  private boolean isStateful_;
-  /**
-   * <pre>
-   * Ops are marked as stateful if their behavior depends on some state beyond
-   * their input tensors (e.g. variable reading op) or if they have
-   * a side-effect (e.g. printing or asserting ops). Equivalently, stateless ops
-   * must always produce the same output for the same input and have
-   * no side-effects.
-   * By default Ops may be moved between devices.  Stateful ops should
-   * either not be moved, or should only be moved if that state can also
-   * be moved (e.g. via some sort of save / restore).
-   * Stateful ops are guaranteed to never be optimized away by Common
-   * Subexpression Elimination (CSE).
-   * </pre>
-   *
-   * <code>bool is_stateful = 17;</code>
-   */
-  public boolean getIsStateful() {
-    return isStateful_;
-  }
-
-  public static final int ALLOWS_UNINITIALIZED_INPUT_FIELD_NUMBER = 19;
-  private boolean allowsUninitializedInput_;
-  /**
-   * <pre>
-   * By default, all inputs to an Op must be initialized Tensors.  Ops
-   * that may initialize tensors for the first time should set this
-   * field to true, to allow the Op to take an uninitialized Tensor as
-   * input.
-   * </pre>
-   *
-   * <code>bool allows_uninitialized_input = 19;</code>
-   */
-  public boolean getAllowsUninitializedInput() {
-    return allowsUninitializedInput_;
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    if (!getNameBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 1, name_);
-    }
-    for (int i = 0; i < inputArg_.size(); i++) {
-      output.writeMessage(2, inputArg_.get(i));
-    }
-    for (int i = 0; i < outputArg_.size(); i++) {
-      output.writeMessage(3, outputArg_.get(i));
-    }
-    for (int i = 0; i < attr_.size(); i++) {
-      output.writeMessage(4, attr_.get(i));
-    }
-    if (!getSummaryBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 5, summary_);
-    }
-    if (!getDescriptionBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 6, description_);
-    }
-    if (deprecation_ != null) {
-      output.writeMessage(8, getDeprecation());
-    }
-    if (isAggregate_ != false) {
-      output.writeBool(16, isAggregate_);
-    }
-    if (isStateful_ != false) {
-      output.writeBool(17, isStateful_);
-    }
-    if (isCommutative_ != false) {
-      output.writeBool(18, isCommutative_);
-    }
-    if (allowsUninitializedInput_ != false) {
-      output.writeBool(19, allowsUninitializedInput_);
-    }
-    for (int i = 0; i < controlOutput_.size(); i++) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 20, controlOutput_.getRaw(i));
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (!getNameBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, name_);
-    }
-    for (int i = 0; i < inputArg_.size(); i++) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(2, inputArg_.get(i));
-    }
-    for (int i = 0; i < outputArg_.size(); i++) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(3, outputArg_.get(i));
-    }
-    for (int i = 0; i < attr_.size(); i++) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(4, attr_.get(i));
-    }
-    if (!getSummaryBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(5, summary_);
-    }
-    if (!getDescriptionBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(6, description_);
-    }
-    if (deprecation_ != null) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(8, getDeprecation());
-    }
-    if (isAggregate_ != false) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeBoolSize(16, isAggregate_);
-    }
-    if (isStateful_ != false) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeBoolSize(17, isStateful_);
-    }
-    if (isCommutative_ != false) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeBoolSize(18, isCommutative_);
-    }
-    if (allowsUninitializedInput_ != false) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeBoolSize(19, allowsUninitializedInput_);
-    }
-    {
-      int dataSize = 0;
-      for (int i = 0; i < controlOutput_.size(); i++) {
-        dataSize += computeStringSizeNoTag(controlOutput_.getRaw(i));
-      }
-      size += dataSize;
-      size += 2 * getControlOutputList().size();
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.framework.OpDef)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.framework.OpDef other = (org.tensorflow.proto.framework.OpDef) obj;
-
-    if (!getName()
-        .equals(other.getName())) return false;
-    if (!getInputArgList()
-        .equals(other.getInputArgList())) return false;
-    if (!getOutputArgList()
-        .equals(other.getOutputArgList())) return false;
-    if (!getControlOutputList()
-        .equals(other.getControlOutputList())) return false;
-    if (!getAttrList()
-        .equals(other.getAttrList())) return false;
-    if (hasDeprecation() != other.hasDeprecation()) return false;
-    if (hasDeprecation()) {
-      if (!getDeprecation()
-          .equals(other.getDeprecation())) return false;
-    }
-    if (!getSummary()
-        .equals(other.getSummary())) return false;
-    if (!getDescription()
-        .equals(other.getDescription())) return false;
-    if (getIsCommutative()
-        != other.getIsCommutative()) return false;
-    if (getIsAggregate()
-        != other.getIsAggregate()) return false;
-    if (getIsStateful()
-        != other.getIsStateful()) return false;
-    if (getAllowsUninitializedInput()
-        != other.getAllowsUninitializedInput()) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    hash = (37 * hash) + NAME_FIELD_NUMBER;
-    hash = (53 * hash) + getName().hashCode();
-    if (getInputArgCount() > 0) {
-      hash = (37 * hash) + INPUT_ARG_FIELD_NUMBER;
-      hash = (53 * hash) + getInputArgList().hashCode();
-    }
-    if (getOutputArgCount() > 0) {
-      hash = (37 * hash) + OUTPUT_ARG_FIELD_NUMBER;
-      hash = (53 * hash) + getOutputArgList().hashCode();
-    }
-    if (getControlOutputCount() > 0) {
-      hash = (37 * hash) + CONTROL_OUTPUT_FIELD_NUMBER;
-      hash = (53 * hash) + getControlOutputList().hashCode();
-    }
-    if (getAttrCount() > 0) {
-      hash = (37 * hash) + ATTR_FIELD_NUMBER;
-      hash = (53 * hash) + getAttrList().hashCode();
-    }
-    if (hasDeprecation()) {
-      hash = (37 * hash) + DEPRECATION_FIELD_NUMBER;
-      hash = (53 * hash) + getDeprecation().hashCode();
-    }
-    hash = (37 * hash) + SUMMARY_FIELD_NUMBER;
-    hash = (53 * hash) + getSummary().hashCode();
-    hash = (37 * hash) + DESCRIPTION_FIELD_NUMBER;
-    hash = (53 * hash) + getDescription().hashCode();
-    hash = (37 * hash) + IS_COMMUTATIVE_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashBoolean(
-        getIsCommutative());
-    hash = (37 * hash) + IS_AGGREGATE_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashBoolean(
-        getIsAggregate());
-    hash = (37 * hash) + IS_STATEFUL_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashBoolean(
-        getIsStateful());
-    hash = (37 * hash) + ALLOWS_UNINITIALIZED_INPUT_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashBoolean(
-        getAllowsUninitializedInput());
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.framework.OpDef parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.OpDef parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.OpDef parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.OpDef parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.OpDef parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.OpDef parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.OpDef parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.OpDef parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.OpDef parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.OpDef parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.OpDef parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.OpDef parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.framework.OpDef prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * <pre>
-   * Defines an operation. A NodeDef in a GraphDef specifies an Op by
-   * using the "op" field which should match the name of a OpDef.
-   * LINT.IfChange
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.OpDef}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.OpDef)
-      org.tensorflow.proto.framework.OpDefOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.OpDefProtos.internal_static_tensorflow_OpDef_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.OpDefProtos.internal_static_tensorflow_OpDef_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.OpDef.class, org.tensorflow.proto.framework.OpDef.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.framework.OpDef.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-        getInputArgFieldBuilder();
-        getOutputArgFieldBuilder();
-        getAttrFieldBuilder();
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      name_ = "";
-
-      if (inputArgBuilder_ == null) {
-        inputArg_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000001);
-      } else {
-        inputArgBuilder_.clear();
-      }
-      if (outputArgBuilder_ == null) {
-        outputArg_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000002);
-      } else {
-        outputArgBuilder_.clear();
-      }
-      controlOutput_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-      bitField0_ = (bitField0_ & ~0x00000004);
-      if (attrBuilder_ == null) {
-        attr_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000008);
-      } else {
-        attrBuilder_.clear();
-      }
-      if (deprecationBuilder_ == null) {
-        deprecation_ = null;
-      } else {
-        deprecation_ = null;
-        deprecationBuilder_ = null;
-      }
-      summary_ = "";
-
-      description_ = "";
-
-      isCommutative_ = false;
-
-      isAggregate_ = false;
-
-      isStateful_ = false;
-
-      allowsUninitializedInput_ = false;
-
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.framework.OpDefProtos.internal_static_tensorflow_OpDef_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.OpDef getDefaultInstanceForType() {
-      return org.tensorflow.proto.framework.OpDef.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.OpDef build() {
-      org.tensorflow.proto.framework.OpDef result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.OpDef buildPartial() {
-      org.tensorflow.proto.framework.OpDef result = new org.tensorflow.proto.framework.OpDef(this);
-      int from_bitField0_ = bitField0_;
-      result.name_ = name_;
-      if (inputArgBuilder_ == null) {
-        if (((bitField0_ & 0x00000001) != 0)) {
-          inputArg_ = java.util.Collections.unmodifiableList(inputArg_);
-          bitField0_ = (bitField0_ & ~0x00000001);
-        }
-        result.inputArg_ = inputArg_;
-      } else {
-        result.inputArg_ = inputArgBuilder_.build();
-      }
-      if (outputArgBuilder_ == null) {
-        if (((bitField0_ & 0x00000002) != 0)) {
-          outputArg_ = java.util.Collections.unmodifiableList(outputArg_);
-          bitField0_ = (bitField0_ & ~0x00000002);
-        }
-        result.outputArg_ = outputArg_;
-      } else {
-        result.outputArg_ = outputArgBuilder_.build();
-      }
-      if (((bitField0_ & 0x00000004) != 0)) {
-        controlOutput_ = controlOutput_.getUnmodifiableView();
-        bitField0_ = (bitField0_ & ~0x00000004);
-      }
-      result.controlOutput_ = controlOutput_;
-      if (attrBuilder_ == null) {
-        if (((bitField0_ & 0x00000008) != 0)) {
-          attr_ = java.util.Collections.unmodifiableList(attr_);
-          bitField0_ = (bitField0_ & ~0x00000008);
-        }
-        result.attr_ = attr_;
-      } else {
-        result.attr_ = attrBuilder_.build();
-      }
-      if (deprecationBuilder_ == null) {
-        result.deprecation_ = deprecation_;
-      } else {
-        result.deprecation_ = deprecationBuilder_.build();
-      }
-      result.summary_ = summary_;
-      result.description_ = description_;
-      result.isCommutative_ = isCommutative_;
-      result.isAggregate_ = isAggregate_;
-      result.isStateful_ = isStateful_;
-      result.allowsUninitializedInput_ = allowsUninitializedInput_;
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.framework.OpDef) {
-        return mergeFrom((org.tensorflow.proto.framework.OpDef)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.framework.OpDef other) {
-      if (other == org.tensorflow.proto.framework.OpDef.getDefaultInstance()) return this;
-      if (!other.getName().isEmpty()) {
-        name_ = other.name_;
-        onChanged();
-      }
-      if (inputArgBuilder_ == null) {
-        if (!other.inputArg_.isEmpty()) {
-          if (inputArg_.isEmpty()) {
-            inputArg_ = other.inputArg_;
-            bitField0_ = (bitField0_ & ~0x00000001);
-          } else {
-            ensureInputArgIsMutable();
-            inputArg_.addAll(other.inputArg_);
-          }
-          onChanged();
-        }
-      } else {
-        if (!other.inputArg_.isEmpty()) {
-          if (inputArgBuilder_.isEmpty()) {
-            inputArgBuilder_.dispose();
-            inputArgBuilder_ = null;
-            inputArg_ = other.inputArg_;
-            bitField0_ = (bitField0_ & ~0x00000001);
-            inputArgBuilder_ = 
-              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
-                 getInputArgFieldBuilder() : null;
-          } else {
-            inputArgBuilder_.addAllMessages(other.inputArg_);
-          }
-        }
-      }
-      if (outputArgBuilder_ == null) {
-        if (!other.outputArg_.isEmpty()) {
-          if (outputArg_.isEmpty()) {
-            outputArg_ = other.outputArg_;
-            bitField0_ = (bitField0_ & ~0x00000002);
-          } else {
-            ensureOutputArgIsMutable();
-            outputArg_.addAll(other.outputArg_);
-          }
-          onChanged();
-        }
-      } else {
-        if (!other.outputArg_.isEmpty()) {
-          if (outputArgBuilder_.isEmpty()) {
-            outputArgBuilder_.dispose();
-            outputArgBuilder_ = null;
-            outputArg_ = other.outputArg_;
-            bitField0_ = (bitField0_ & ~0x00000002);
-            outputArgBuilder_ = 
-              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
-                 getOutputArgFieldBuilder() : null;
-          } else {
-            outputArgBuilder_.addAllMessages(other.outputArg_);
-          }
-        }
-      }
-      if (!other.controlOutput_.isEmpty()) {
-        if (controlOutput_.isEmpty()) {
-          controlOutput_ = other.controlOutput_;
-          bitField0_ = (bitField0_ & ~0x00000004);
-        } else {
-          ensureControlOutputIsMutable();
-          controlOutput_.addAll(other.controlOutput_);
-        }
-        onChanged();
-      }
-      if (attrBuilder_ == null) {
-        if (!other.attr_.isEmpty()) {
-          if (attr_.isEmpty()) {
-            attr_ = other.attr_;
-            bitField0_ = (bitField0_ & ~0x00000008);
-          } else {
-            ensureAttrIsMutable();
-            attr_.addAll(other.attr_);
-          }
-          onChanged();
-        }
-      } else {
-        if (!other.attr_.isEmpty()) {
-          if (attrBuilder_.isEmpty()) {
-            attrBuilder_.dispose();
-            attrBuilder_ = null;
-            attr_ = other.attr_;
-            bitField0_ = (bitField0_ & ~0x00000008);
-            attrBuilder_ = 
-              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
-                 getAttrFieldBuilder() : null;
-          } else {
-            attrBuilder_.addAllMessages(other.attr_);
-          }
-        }
-      }
-      if (other.hasDeprecation()) {
-        mergeDeprecation(other.getDeprecation());
-      }
-      if (!other.getSummary().isEmpty()) {
-        summary_ = other.summary_;
-        onChanged();
-      }
-      if (!other.getDescription().isEmpty()) {
-        description_ = other.description_;
-        onChanged();
-      }
-      if (other.getIsCommutative() != false) {
-        setIsCommutative(other.getIsCommutative());
-      }
-      if (other.getIsAggregate() != false) {
-        setIsAggregate(other.getIsAggregate());
-      }
-      if (other.getIsStateful() != false) {
-        setIsStateful(other.getIsStateful());
-      }
-      if (other.getAllowsUninitializedInput() != false) {
-        setAllowsUninitializedInput(other.getAllowsUninitializedInput());
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.framework.OpDef parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.framework.OpDef) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-    private int bitField0_;
-
-    private java.lang.Object name_ = "";
-    /**
-     * <pre>
-     * Op names starting with an underscore are reserved for internal use.
-     * Names should be CamelCase and match the regexp "[A-Z][a-zA-Z0-9&gt;_]*".
-     * </pre>
-     *
-     * <code>string name = 1;</code>
-     */
-    public java.lang.String getName() {
-      java.lang.Object ref = name_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        name_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Op names starting with an underscore are reserved for internal use.
-     * Names should be CamelCase and match the regexp "[A-Z][a-zA-Z0-9&gt;_]*".
-     * </pre>
-     *
-     * <code>string name = 1;</code>
-     */
-    public com.google.protobuf.ByteString
-        getNameBytes() {
-      java.lang.Object ref = name_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        name_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Op names starting with an underscore are reserved for internal use.
-     * Names should be CamelCase and match the regexp "[A-Z][a-zA-Z0-9&gt;_]*".
-     * </pre>
-     *
-     * <code>string name = 1;</code>
-     */
-    public Builder setName(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      name_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Op names starting with an underscore are reserved for internal use.
-     * Names should be CamelCase and match the regexp "[A-Z][a-zA-Z0-9&gt;_]*".
-     * </pre>
-     *
-     * <code>string name = 1;</code>
-     */
-    public Builder clearName() {
-      
-      name_ = getDefaultInstance().getName();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Op names starting with an underscore are reserved for internal use.
-     * Names should be CamelCase and match the regexp "[A-Z][a-zA-Z0-9&gt;_]*".
-     * </pre>
-     *
-     * <code>string name = 1;</code>
-     */
-    public Builder setNameBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      name_ = value;
-      onChanged();
-      return this;
-    }
-
-    private java.util.List<org.tensorflow.proto.framework.OpDef.ArgDef> inputArg_ =
-      java.util.Collections.emptyList();
-    private void ensureInputArgIsMutable() {
-      if (!((bitField0_ & 0x00000001) != 0)) {
-        inputArg_ = new java.util.ArrayList<org.tensorflow.proto.framework.OpDef.ArgDef>(inputArg_);
-        bitField0_ |= 0x00000001;
-       }
-    }
-
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.framework.OpDef.ArgDef, org.tensorflow.proto.framework.OpDef.ArgDef.Builder, org.tensorflow.proto.framework.OpDef.ArgDefOrBuilder> inputArgBuilder_;
-
-    /**
-     * <pre>
-     * Description of the input(s).
-     * </pre>
-     *
-     * <code>repeated .tensorflow.OpDef.ArgDef input_arg = 2;</code>
-     */
-    public java.util.List<org.tensorflow.proto.framework.OpDef.ArgDef> getInputArgList() {
-      if (inputArgBuilder_ == null) {
-        return java.util.Collections.unmodifiableList(inputArg_);
-      } else {
-        return inputArgBuilder_.getMessageList();
-      }
-    }
-    /**
-     * <pre>
-     * Description of the input(s).
-     * </pre>
-     *
-     * <code>repeated .tensorflow.OpDef.ArgDef input_arg = 2;</code>
-     */
-    public int getInputArgCount() {
-      if (inputArgBuilder_ == null) {
-        return inputArg_.size();
-      } else {
-        return inputArgBuilder_.getCount();
-      }
-    }
-    /**
-     * <pre>
-     * Description of the input(s).
-     * </pre>
-     *
-     * <code>repeated .tensorflow.OpDef.ArgDef input_arg = 2;</code>
-     */
-    public org.tensorflow.proto.framework.OpDef.ArgDef getInputArg(int index) {
-      if (inputArgBuilder_ == null) {
-        return inputArg_.get(index);
-      } else {
-        return inputArgBuilder_.getMessage(index);
-      }
-    }
-    /**
-     * <pre>
-     * Description of the input(s).
-     * </pre>
-     *
-     * <code>repeated .tensorflow.OpDef.ArgDef input_arg = 2;</code>
-     */
-    public Builder setInputArg(
-        int index, org.tensorflow.proto.framework.OpDef.ArgDef value) {
-      if (inputArgBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureInputArgIsMutable();
-        inputArg_.set(index, value);
-        onChanged();
-      } else {
-        inputArgBuilder_.setMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Description of the input(s).
-     * </pre>
-     *
-     * <code>repeated .tensorflow.OpDef.ArgDef input_arg = 2;</code>
-     */
-    public Builder setInputArg(
-        int index, org.tensorflow.proto.framework.OpDef.ArgDef.Builder builderForValue) {
-      if (inputArgBuilder_ == null) {
-        ensureInputArgIsMutable();
-        inputArg_.set(index, builderForValue.build());
-        onChanged();
-      } else {
-        inputArgBuilder_.setMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Description of the input(s).
-     * </pre>
-     *
-     * <code>repeated .tensorflow.OpDef.ArgDef input_arg = 2;</code>
-     */
-    public Builder addInputArg(org.tensorflow.proto.framework.OpDef.ArgDef value) {
-      if (inputArgBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureInputArgIsMutable();
-        inputArg_.add(value);
-        onChanged();
-      } else {
-        inputArgBuilder_.addMessage(value);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Description of the input(s).
-     * </pre>
-     *
-     * <code>repeated .tensorflow.OpDef.ArgDef input_arg = 2;</code>
-     */
-    public Builder addInputArg(
-        int index, org.tensorflow.proto.framework.OpDef.ArgDef value) {
-      if (inputArgBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureInputArgIsMutable();
-        inputArg_.add(index, value);
-        onChanged();
-      } else {
-        inputArgBuilder_.addMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Description of the input(s).
-     * </pre>
-     *
-     * <code>repeated .tensorflow.OpDef.ArgDef input_arg = 2;</code>
-     */
-    public Builder addInputArg(
-        org.tensorflow.proto.framework.OpDef.ArgDef.Builder builderForValue) {
-      if (inputArgBuilder_ == null) {
-        ensureInputArgIsMutable();
-        inputArg_.add(builderForValue.build());
-        onChanged();
-      } else {
-        inputArgBuilder_.addMessage(builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Description of the input(s).
-     * </pre>
-     *
-     * <code>repeated .tensorflow.OpDef.ArgDef input_arg = 2;</code>
-     */
-    public Builder addInputArg(
-        int index, org.tensorflow.proto.framework.OpDef.ArgDef.Builder builderForValue) {
-      if (inputArgBuilder_ == null) {
-        ensureInputArgIsMutable();
-        inputArg_.add(index, builderForValue.build());
-        onChanged();
-      } else {
-        inputArgBuilder_.addMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Description of the input(s).
-     * </pre>
-     *
-     * <code>repeated .tensorflow.OpDef.ArgDef input_arg = 2;</code>
-     */
-    public Builder addAllInputArg(
-        java.lang.Iterable<? extends org.tensorflow.proto.framework.OpDef.ArgDef> values) {
-      if (inputArgBuilder_ == null) {
-        ensureInputArgIsMutable();
-        com.google.protobuf.AbstractMessageLite.Builder.addAll(
-            values, inputArg_);
-        onChanged();
-      } else {
-        inputArgBuilder_.addAllMessages(values);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Description of the input(s).
-     * </pre>
-     *
-     * <code>repeated .tensorflow.OpDef.ArgDef input_arg = 2;</code>
-     */
-    public Builder clearInputArg() {
-      if (inputArgBuilder_ == null) {
-        inputArg_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000001);
-        onChanged();
-      } else {
-        inputArgBuilder_.clear();
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Description of the input(s).
-     * </pre>
-     *
-     * <code>repeated .tensorflow.OpDef.ArgDef input_arg = 2;</code>
-     */
-    public Builder removeInputArg(int index) {
-      if (inputArgBuilder_ == null) {
-        ensureInputArgIsMutable();
-        inputArg_.remove(index);
-        onChanged();
-      } else {
-        inputArgBuilder_.remove(index);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Description of the input(s).
-     * </pre>
-     *
-     * <code>repeated .tensorflow.OpDef.ArgDef input_arg = 2;</code>
-     */
-    public org.tensorflow.proto.framework.OpDef.ArgDef.Builder getInputArgBuilder(
-        int index) {
-      return getInputArgFieldBuilder().getBuilder(index);
-    }
-    /**
-     * <pre>
-     * Description of the input(s).
-     * </pre>
-     *
-     * <code>repeated .tensorflow.OpDef.ArgDef input_arg = 2;</code>
-     */
-    public org.tensorflow.proto.framework.OpDef.ArgDefOrBuilder getInputArgOrBuilder(
-        int index) {
-      if (inputArgBuilder_ == null) {
-        return inputArg_.get(index);  } else {
-        return inputArgBuilder_.getMessageOrBuilder(index);
-      }
-    }
-    /**
-     * <pre>
-     * Description of the input(s).
-     * </pre>
-     *
-     * <code>repeated .tensorflow.OpDef.ArgDef input_arg = 2;</code>
-     */
-    public java.util.List<? extends org.tensorflow.proto.framework.OpDef.ArgDefOrBuilder> 
-         getInputArgOrBuilderList() {
-      if (inputArgBuilder_ != null) {
-        return inputArgBuilder_.getMessageOrBuilderList();
-      } else {
-        return java.util.Collections.unmodifiableList(inputArg_);
-      }
-    }
-    /**
-     * <pre>
-     * Description of the input(s).
-     * </pre>
-     *
-     * <code>repeated .tensorflow.OpDef.ArgDef input_arg = 2;</code>
-     */
-    public org.tensorflow.proto.framework.OpDef.ArgDef.Builder addInputArgBuilder() {
-      return getInputArgFieldBuilder().addBuilder(
-          org.tensorflow.proto.framework.OpDef.ArgDef.getDefaultInstance());
-    }
-    /**
-     * <pre>
-     * Description of the input(s).
-     * </pre>
-     *
-     * <code>repeated .tensorflow.OpDef.ArgDef input_arg = 2;</code>
-     */
-    public org.tensorflow.proto.framework.OpDef.ArgDef.Builder addInputArgBuilder(
-        int index) {
-      return getInputArgFieldBuilder().addBuilder(
-          index, org.tensorflow.proto.framework.OpDef.ArgDef.getDefaultInstance());
-    }
-    /**
-     * <pre>
-     * Description of the input(s).
-     * </pre>
-     *
-     * <code>repeated .tensorflow.OpDef.ArgDef input_arg = 2;</code>
-     */
-    public java.util.List<org.tensorflow.proto.framework.OpDef.ArgDef.Builder> 
-         getInputArgBuilderList() {
-      return getInputArgFieldBuilder().getBuilderList();
-    }
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.framework.OpDef.ArgDef, org.tensorflow.proto.framework.OpDef.ArgDef.Builder, org.tensorflow.proto.framework.OpDef.ArgDefOrBuilder> 
-        getInputArgFieldBuilder() {
-      if (inputArgBuilder_ == null) {
-        inputArgBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
-            org.tensorflow.proto.framework.OpDef.ArgDef, org.tensorflow.proto.framework.OpDef.ArgDef.Builder, org.tensorflow.proto.framework.OpDef.ArgDefOrBuilder>(
-                inputArg_,
-                ((bitField0_ & 0x00000001) != 0),
-                getParentForChildren(),
-                isClean());
-        inputArg_ = null;
-      }
-      return inputArgBuilder_;
-    }
-
-    private java.util.List<org.tensorflow.proto.framework.OpDef.ArgDef> outputArg_ =
-      java.util.Collections.emptyList();
-    private void ensureOutputArgIsMutable() {
-      if (!((bitField0_ & 0x00000002) != 0)) {
-        outputArg_ = new java.util.ArrayList<org.tensorflow.proto.framework.OpDef.ArgDef>(outputArg_);
-        bitField0_ |= 0x00000002;
-       }
-    }
-
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.framework.OpDef.ArgDef, org.tensorflow.proto.framework.OpDef.ArgDef.Builder, org.tensorflow.proto.framework.OpDef.ArgDefOrBuilder> outputArgBuilder_;
-
-    /**
-     * <pre>
-     * Description of the output(s).
-     * </pre>
-     *
-     * <code>repeated .tensorflow.OpDef.ArgDef output_arg = 3;</code>
-     */
-    public java.util.List<org.tensorflow.proto.framework.OpDef.ArgDef> getOutputArgList() {
-      if (outputArgBuilder_ == null) {
-        return java.util.Collections.unmodifiableList(outputArg_);
-      } else {
-        return outputArgBuilder_.getMessageList();
-      }
-    }
-    /**
-     * <pre>
-     * Description of the output(s).
-     * </pre>
-     *
-     * <code>repeated .tensorflow.OpDef.ArgDef output_arg = 3;</code>
-     */
-    public int getOutputArgCount() {
-      if (outputArgBuilder_ == null) {
-        return outputArg_.size();
-      } else {
-        return outputArgBuilder_.getCount();
-      }
-    }
-    /**
-     * <pre>
-     * Description of the output(s).
-     * </pre>
-     *
-     * <code>repeated .tensorflow.OpDef.ArgDef output_arg = 3;</code>
-     */
-    public org.tensorflow.proto.framework.OpDef.ArgDef getOutputArg(int index) {
-      if (outputArgBuilder_ == null) {
-        return outputArg_.get(index);
-      } else {
-        return outputArgBuilder_.getMessage(index);
-      }
-    }
-    /**
-     * <pre>
-     * Description of the output(s).
-     * </pre>
-     *
-     * <code>repeated .tensorflow.OpDef.ArgDef output_arg = 3;</code>
-     */
-    public Builder setOutputArg(
-        int index, org.tensorflow.proto.framework.OpDef.ArgDef value) {
-      if (outputArgBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureOutputArgIsMutable();
-        outputArg_.set(index, value);
-        onChanged();
-      } else {
-        outputArgBuilder_.setMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Description of the output(s).
-     * </pre>
-     *
-     * <code>repeated .tensorflow.OpDef.ArgDef output_arg = 3;</code>
-     */
-    public Builder setOutputArg(
-        int index, org.tensorflow.proto.framework.OpDef.ArgDef.Builder builderForValue) {
-      if (outputArgBuilder_ == null) {
-        ensureOutputArgIsMutable();
-        outputArg_.set(index, builderForValue.build());
-        onChanged();
-      } else {
-        outputArgBuilder_.setMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Description of the output(s).
-     * </pre>
-     *
-     * <code>repeated .tensorflow.OpDef.ArgDef output_arg = 3;</code>
-     */
-    public Builder addOutputArg(org.tensorflow.proto.framework.OpDef.ArgDef value) {
-      if (outputArgBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureOutputArgIsMutable();
-        outputArg_.add(value);
-        onChanged();
-      } else {
-        outputArgBuilder_.addMessage(value);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Description of the output(s).
-     * </pre>
-     *
-     * <code>repeated .tensorflow.OpDef.ArgDef output_arg = 3;</code>
-     */
-    public Builder addOutputArg(
-        int index, org.tensorflow.proto.framework.OpDef.ArgDef value) {
-      if (outputArgBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureOutputArgIsMutable();
-        outputArg_.add(index, value);
-        onChanged();
-      } else {
-        outputArgBuilder_.addMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Description of the output(s).
-     * </pre>
-     *
-     * <code>repeated .tensorflow.OpDef.ArgDef output_arg = 3;</code>
-     */
-    public Builder addOutputArg(
-        org.tensorflow.proto.framework.OpDef.ArgDef.Builder builderForValue) {
-      if (outputArgBuilder_ == null) {
-        ensureOutputArgIsMutable();
-        outputArg_.add(builderForValue.build());
-        onChanged();
-      } else {
-        outputArgBuilder_.addMessage(builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Description of the output(s).
-     * </pre>
-     *
-     * <code>repeated .tensorflow.OpDef.ArgDef output_arg = 3;</code>
-     */
-    public Builder addOutputArg(
-        int index, org.tensorflow.proto.framework.OpDef.ArgDef.Builder builderForValue) {
-      if (outputArgBuilder_ == null) {
-        ensureOutputArgIsMutable();
-        outputArg_.add(index, builderForValue.build());
-        onChanged();
-      } else {
-        outputArgBuilder_.addMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Description of the output(s).
-     * </pre>
-     *
-     * <code>repeated .tensorflow.OpDef.ArgDef output_arg = 3;</code>
-     */
-    public Builder addAllOutputArg(
-        java.lang.Iterable<? extends org.tensorflow.proto.framework.OpDef.ArgDef> values) {
-      if (outputArgBuilder_ == null) {
-        ensureOutputArgIsMutable();
-        com.google.protobuf.AbstractMessageLite.Builder.addAll(
-            values, outputArg_);
-        onChanged();
-      } else {
-        outputArgBuilder_.addAllMessages(values);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Description of the output(s).
-     * </pre>
-     *
-     * <code>repeated .tensorflow.OpDef.ArgDef output_arg = 3;</code>
-     */
-    public Builder clearOutputArg() {
-      if (outputArgBuilder_ == null) {
-        outputArg_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000002);
-        onChanged();
-      } else {
-        outputArgBuilder_.clear();
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Description of the output(s).
-     * </pre>
-     *
-     * <code>repeated .tensorflow.OpDef.ArgDef output_arg = 3;</code>
-     */
-    public Builder removeOutputArg(int index) {
-      if (outputArgBuilder_ == null) {
-        ensureOutputArgIsMutable();
-        outputArg_.remove(index);
-        onChanged();
-      } else {
-        outputArgBuilder_.remove(index);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Description of the output(s).
-     * </pre>
-     *
-     * <code>repeated .tensorflow.OpDef.ArgDef output_arg = 3;</code>
-     */
-    public org.tensorflow.proto.framework.OpDef.ArgDef.Builder getOutputArgBuilder(
-        int index) {
-      return getOutputArgFieldBuilder().getBuilder(index);
-    }
-    /**
-     * <pre>
-     * Description of the output(s).
-     * </pre>
-     *
-     * <code>repeated .tensorflow.OpDef.ArgDef output_arg = 3;</code>
-     */
-    public org.tensorflow.proto.framework.OpDef.ArgDefOrBuilder getOutputArgOrBuilder(
-        int index) {
-      if (outputArgBuilder_ == null) {
-        return outputArg_.get(index);  } else {
-        return outputArgBuilder_.getMessageOrBuilder(index);
-      }
-    }
-    /**
-     * <pre>
-     * Description of the output(s).
-     * </pre>
-     *
-     * <code>repeated .tensorflow.OpDef.ArgDef output_arg = 3;</code>
-     */
-    public java.util.List<? extends org.tensorflow.proto.framework.OpDef.ArgDefOrBuilder> 
-         getOutputArgOrBuilderList() {
-      if (outputArgBuilder_ != null) {
-        return outputArgBuilder_.getMessageOrBuilderList();
-      } else {
-        return java.util.Collections.unmodifiableList(outputArg_);
-      }
-    }
-    /**
-     * <pre>
-     * Description of the output(s).
-     * </pre>
-     *
-     * <code>repeated .tensorflow.OpDef.ArgDef output_arg = 3;</code>
-     */
-    public org.tensorflow.proto.framework.OpDef.ArgDef.Builder addOutputArgBuilder() {
-      return getOutputArgFieldBuilder().addBuilder(
-          org.tensorflow.proto.framework.OpDef.ArgDef.getDefaultInstance());
-    }
-    /**
-     * <pre>
-     * Description of the output(s).
-     * </pre>
-     *
-     * <code>repeated .tensorflow.OpDef.ArgDef output_arg = 3;</code>
-     */
-    public org.tensorflow.proto.framework.OpDef.ArgDef.Builder addOutputArgBuilder(
-        int index) {
-      return getOutputArgFieldBuilder().addBuilder(
-          index, org.tensorflow.proto.framework.OpDef.ArgDef.getDefaultInstance());
-    }
-    /**
-     * <pre>
-     * Description of the output(s).
-     * </pre>
-     *
-     * <code>repeated .tensorflow.OpDef.ArgDef output_arg = 3;</code>
-     */
-    public java.util.List<org.tensorflow.proto.framework.OpDef.ArgDef.Builder> 
-         getOutputArgBuilderList() {
-      return getOutputArgFieldBuilder().getBuilderList();
-    }
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.framework.OpDef.ArgDef, org.tensorflow.proto.framework.OpDef.ArgDef.Builder, org.tensorflow.proto.framework.OpDef.ArgDefOrBuilder> 
-        getOutputArgFieldBuilder() {
-      if (outputArgBuilder_ == null) {
-        outputArgBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
-            org.tensorflow.proto.framework.OpDef.ArgDef, org.tensorflow.proto.framework.OpDef.ArgDef.Builder, org.tensorflow.proto.framework.OpDef.ArgDefOrBuilder>(
-                outputArg_,
-                ((bitField0_ & 0x00000002) != 0),
-                getParentForChildren(),
-                isClean());
-        outputArg_ = null;
-      }
-      return outputArgBuilder_;
-    }
-
-    private com.google.protobuf.LazyStringList controlOutput_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-    private void ensureControlOutputIsMutable() {
-      if (!((bitField0_ & 0x00000004) != 0)) {
-        controlOutput_ = new com.google.protobuf.LazyStringArrayList(controlOutput_);
-        bitField0_ |= 0x00000004;
-       }
-    }
-    /**
-     * <pre>
-     * Named control outputs for this operation. Useful only for composite
-     * operations (i.e. functions) which want to name different control outputs.
-     * </pre>
-     *
-     * <code>repeated string control_output = 20;</code>
-     */
-    public com.google.protobuf.ProtocolStringList
-        getControlOutputList() {
-      return controlOutput_.getUnmodifiableView();
-    }
-    /**
-     * <pre>
-     * Named control outputs for this operation. Useful only for composite
-     * operations (i.e. functions) which want to name different control outputs.
-     * </pre>
-     *
-     * <code>repeated string control_output = 20;</code>
-     */
-    public int getControlOutputCount() {
-      return controlOutput_.size();
-    }
-    /**
-     * <pre>
-     * Named control outputs for this operation. Useful only for composite
-     * operations (i.e. functions) which want to name different control outputs.
-     * </pre>
-     *
-     * <code>repeated string control_output = 20;</code>
-     */
-    public java.lang.String getControlOutput(int index) {
-      return controlOutput_.get(index);
-    }
-    /**
-     * <pre>
-     * Named control outputs for this operation. Useful only for composite
-     * operations (i.e. functions) which want to name different control outputs.
-     * </pre>
-     *
-     * <code>repeated string control_output = 20;</code>
-     */
-    public com.google.protobuf.ByteString
-        getControlOutputBytes(int index) {
-      return controlOutput_.getByteString(index);
-    }
-    /**
-     * <pre>
-     * Named control outputs for this operation. Useful only for composite
-     * operations (i.e. functions) which want to name different control outputs.
-     * </pre>
-     *
-     * <code>repeated string control_output = 20;</code>
-     */
-    public Builder setControlOutput(
-        int index, java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  ensureControlOutputIsMutable();
-      controlOutput_.set(index, value);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Named control outputs for this operation. Useful only for composite
-     * operations (i.e. functions) which want to name different control outputs.
-     * </pre>
-     *
-     * <code>repeated string control_output = 20;</code>
-     */
-    public Builder addControlOutput(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  ensureControlOutputIsMutable();
-      controlOutput_.add(value);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Named control outputs for this operation. Useful only for composite
-     * operations (i.e. functions) which want to name different control outputs.
-     * </pre>
-     *
-     * <code>repeated string control_output = 20;</code>
-     */
-    public Builder addAllControlOutput(
-        java.lang.Iterable<java.lang.String> values) {
-      ensureControlOutputIsMutable();
-      com.google.protobuf.AbstractMessageLite.Builder.addAll(
-          values, controlOutput_);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Named control outputs for this operation. Useful only for composite
-     * operations (i.e. functions) which want to name different control outputs.
-     * </pre>
-     *
-     * <code>repeated string control_output = 20;</code>
-     */
-    public Builder clearControlOutput() {
-      controlOutput_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-      bitField0_ = (bitField0_ & ~0x00000004);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Named control outputs for this operation. Useful only for composite
-     * operations (i.e. functions) which want to name different control outputs.
-     * </pre>
-     *
-     * <code>repeated string control_output = 20;</code>
-     */
-    public Builder addControlOutputBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      ensureControlOutputIsMutable();
-      controlOutput_.add(value);
-      onChanged();
-      return this;
-    }
-
-    private java.util.List<org.tensorflow.proto.framework.OpDef.AttrDef> attr_ =
-      java.util.Collections.emptyList();
-    private void ensureAttrIsMutable() {
-      if (!((bitField0_ & 0x00000008) != 0)) {
-        attr_ = new java.util.ArrayList<org.tensorflow.proto.framework.OpDef.AttrDef>(attr_);
-        bitField0_ |= 0x00000008;
-       }
-    }
-
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.framework.OpDef.AttrDef, org.tensorflow.proto.framework.OpDef.AttrDef.Builder, org.tensorflow.proto.framework.OpDef.AttrDefOrBuilder> attrBuilder_;
-
-    /**
-     * <code>repeated .tensorflow.OpDef.AttrDef attr = 4;</code>
-     */
-    public java.util.List<org.tensorflow.proto.framework.OpDef.AttrDef> getAttrList() {
-      if (attrBuilder_ == null) {
-        return java.util.Collections.unmodifiableList(attr_);
-      } else {
-        return attrBuilder_.getMessageList();
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.OpDef.AttrDef attr = 4;</code>
-     */
-    public int getAttrCount() {
-      if (attrBuilder_ == null) {
-        return attr_.size();
-      } else {
-        return attrBuilder_.getCount();
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.OpDef.AttrDef attr = 4;</code>
-     */
-    public org.tensorflow.proto.framework.OpDef.AttrDef getAttr(int index) {
-      if (attrBuilder_ == null) {
-        return attr_.get(index);
-      } else {
-        return attrBuilder_.getMessage(index);
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.OpDef.AttrDef attr = 4;</code>
-     */
-    public Builder setAttr(
-        int index, org.tensorflow.proto.framework.OpDef.AttrDef value) {
-      if (attrBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureAttrIsMutable();
-        attr_.set(index, value);
-        onChanged();
-      } else {
-        attrBuilder_.setMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.OpDef.AttrDef attr = 4;</code>
-     */
-    public Builder setAttr(
-        int index, org.tensorflow.proto.framework.OpDef.AttrDef.Builder builderForValue) {
-      if (attrBuilder_ == null) {
-        ensureAttrIsMutable();
-        attr_.set(index, builderForValue.build());
-        onChanged();
-      } else {
-        attrBuilder_.setMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.OpDef.AttrDef attr = 4;</code>
-     */
-    public Builder addAttr(org.tensorflow.proto.framework.OpDef.AttrDef value) {
-      if (attrBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureAttrIsMutable();
-        attr_.add(value);
-        onChanged();
-      } else {
-        attrBuilder_.addMessage(value);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.OpDef.AttrDef attr = 4;</code>
-     */
-    public Builder addAttr(
-        int index, org.tensorflow.proto.framework.OpDef.AttrDef value) {
-      if (attrBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureAttrIsMutable();
-        attr_.add(index, value);
-        onChanged();
-      } else {
-        attrBuilder_.addMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.OpDef.AttrDef attr = 4;</code>
-     */
-    public Builder addAttr(
-        org.tensorflow.proto.framework.OpDef.AttrDef.Builder builderForValue) {
-      if (attrBuilder_ == null) {
-        ensureAttrIsMutable();
-        attr_.add(builderForValue.build());
-        onChanged();
-      } else {
-        attrBuilder_.addMessage(builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.OpDef.AttrDef attr = 4;</code>
-     */
-    public Builder addAttr(
-        int index, org.tensorflow.proto.framework.OpDef.AttrDef.Builder builderForValue) {
-      if (attrBuilder_ == null) {
-        ensureAttrIsMutable();
-        attr_.add(index, builderForValue.build());
-        onChanged();
-      } else {
-        attrBuilder_.addMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.OpDef.AttrDef attr = 4;</code>
-     */
-    public Builder addAllAttr(
-        java.lang.Iterable<? extends org.tensorflow.proto.framework.OpDef.AttrDef> values) {
-      if (attrBuilder_ == null) {
-        ensureAttrIsMutable();
-        com.google.protobuf.AbstractMessageLite.Builder.addAll(
-            values, attr_);
-        onChanged();
-      } else {
-        attrBuilder_.addAllMessages(values);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.OpDef.AttrDef attr = 4;</code>
-     */
-    public Builder clearAttr() {
-      if (attrBuilder_ == null) {
-        attr_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000008);
-        onChanged();
-      } else {
-        attrBuilder_.clear();
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.OpDef.AttrDef attr = 4;</code>
-     */
-    public Builder removeAttr(int index) {
-      if (attrBuilder_ == null) {
-        ensureAttrIsMutable();
-        attr_.remove(index);
-        onChanged();
-      } else {
-        attrBuilder_.remove(index);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.OpDef.AttrDef attr = 4;</code>
-     */
-    public org.tensorflow.proto.framework.OpDef.AttrDef.Builder getAttrBuilder(
-        int index) {
-      return getAttrFieldBuilder().getBuilder(index);
-    }
-    /**
-     * <code>repeated .tensorflow.OpDef.AttrDef attr = 4;</code>
-     */
-    public org.tensorflow.proto.framework.OpDef.AttrDefOrBuilder getAttrOrBuilder(
-        int index) {
-      if (attrBuilder_ == null) {
-        return attr_.get(index);  } else {
-        return attrBuilder_.getMessageOrBuilder(index);
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.OpDef.AttrDef attr = 4;</code>
-     */
-    public java.util.List<? extends org.tensorflow.proto.framework.OpDef.AttrDefOrBuilder> 
-         getAttrOrBuilderList() {
-      if (attrBuilder_ != null) {
-        return attrBuilder_.getMessageOrBuilderList();
-      } else {
-        return java.util.Collections.unmodifiableList(attr_);
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.OpDef.AttrDef attr = 4;</code>
-     */
-    public org.tensorflow.proto.framework.OpDef.AttrDef.Builder addAttrBuilder() {
-      return getAttrFieldBuilder().addBuilder(
-          org.tensorflow.proto.framework.OpDef.AttrDef.getDefaultInstance());
-    }
-    /**
-     * <code>repeated .tensorflow.OpDef.AttrDef attr = 4;</code>
-     */
-    public org.tensorflow.proto.framework.OpDef.AttrDef.Builder addAttrBuilder(
-        int index) {
-      return getAttrFieldBuilder().addBuilder(
-          index, org.tensorflow.proto.framework.OpDef.AttrDef.getDefaultInstance());
-    }
-    /**
-     * <code>repeated .tensorflow.OpDef.AttrDef attr = 4;</code>
-     */
-    public java.util.List<org.tensorflow.proto.framework.OpDef.AttrDef.Builder> 
-         getAttrBuilderList() {
-      return getAttrFieldBuilder().getBuilderList();
-    }
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.framework.OpDef.AttrDef, org.tensorflow.proto.framework.OpDef.AttrDef.Builder, org.tensorflow.proto.framework.OpDef.AttrDefOrBuilder> 
-        getAttrFieldBuilder() {
-      if (attrBuilder_ == null) {
-        attrBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
-            org.tensorflow.proto.framework.OpDef.AttrDef, org.tensorflow.proto.framework.OpDef.AttrDef.Builder, org.tensorflow.proto.framework.OpDef.AttrDefOrBuilder>(
-                attr_,
-                ((bitField0_ & 0x00000008) != 0),
-                getParentForChildren(),
-                isClean());
-        attr_ = null;
-      }
-      return attrBuilder_;
-    }
-
-    private org.tensorflow.proto.framework.OpDeprecation deprecation_;
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.OpDeprecation, org.tensorflow.proto.framework.OpDeprecation.Builder, org.tensorflow.proto.framework.OpDeprecationOrBuilder> deprecationBuilder_;
-    /**
-     * <pre>
-     * Optional deprecation based on GraphDef versions.
-     * </pre>
-     *
-     * <code>.tensorflow.OpDeprecation deprecation = 8;</code>
-     */
-    public boolean hasDeprecation() {
-      return deprecationBuilder_ != null || deprecation_ != null;
-    }
-    /**
-     * <pre>
-     * Optional deprecation based on GraphDef versions.
-     * </pre>
-     *
-     * <code>.tensorflow.OpDeprecation deprecation = 8;</code>
-     */
-    public org.tensorflow.proto.framework.OpDeprecation getDeprecation() {
-      if (deprecationBuilder_ == null) {
-        return deprecation_ == null ? org.tensorflow.proto.framework.OpDeprecation.getDefaultInstance() : deprecation_;
-      } else {
-        return deprecationBuilder_.getMessage();
-      }
-    }
-    /**
-     * <pre>
-     * Optional deprecation based on GraphDef versions.
-     * </pre>
-     *
-     * <code>.tensorflow.OpDeprecation deprecation = 8;</code>
-     */
-    public Builder setDeprecation(org.tensorflow.proto.framework.OpDeprecation value) {
-      if (deprecationBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        deprecation_ = value;
-        onChanged();
-      } else {
-        deprecationBuilder_.setMessage(value);
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Optional deprecation based on GraphDef versions.
-     * </pre>
-     *
-     * <code>.tensorflow.OpDeprecation deprecation = 8;</code>
-     */
-    public Builder setDeprecation(
-        org.tensorflow.proto.framework.OpDeprecation.Builder builderForValue) {
-      if (deprecationBuilder_ == null) {
-        deprecation_ = builderForValue.build();
-        onChanged();
-      } else {
-        deprecationBuilder_.setMessage(builderForValue.build());
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Optional deprecation based on GraphDef versions.
-     * </pre>
-     *
-     * <code>.tensorflow.OpDeprecation deprecation = 8;</code>
-     */
-    public Builder mergeDeprecation(org.tensorflow.proto.framework.OpDeprecation value) {
-      if (deprecationBuilder_ == null) {
-        if (deprecation_ != null) {
-          deprecation_ =
-            org.tensorflow.proto.framework.OpDeprecation.newBuilder(deprecation_).mergeFrom(value).buildPartial();
-        } else {
-          deprecation_ = value;
-        }
-        onChanged();
-      } else {
-        deprecationBuilder_.mergeFrom(value);
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Optional deprecation based on GraphDef versions.
-     * </pre>
-     *
-     * <code>.tensorflow.OpDeprecation deprecation = 8;</code>
-     */
-    public Builder clearDeprecation() {
-      if (deprecationBuilder_ == null) {
-        deprecation_ = null;
-        onChanged();
-      } else {
-        deprecation_ = null;
-        deprecationBuilder_ = null;
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Optional deprecation based on GraphDef versions.
-     * </pre>
-     *
-     * <code>.tensorflow.OpDeprecation deprecation = 8;</code>
-     */
-    public org.tensorflow.proto.framework.OpDeprecation.Builder getDeprecationBuilder() {
-      
-      onChanged();
-      return getDeprecationFieldBuilder().getBuilder();
-    }
-    /**
-     * <pre>
-     * Optional deprecation based on GraphDef versions.
-     * </pre>
-     *
-     * <code>.tensorflow.OpDeprecation deprecation = 8;</code>
-     */
-    public org.tensorflow.proto.framework.OpDeprecationOrBuilder getDeprecationOrBuilder() {
-      if (deprecationBuilder_ != null) {
-        return deprecationBuilder_.getMessageOrBuilder();
-      } else {
-        return deprecation_ == null ?
-            org.tensorflow.proto.framework.OpDeprecation.getDefaultInstance() : deprecation_;
-      }
-    }
-    /**
-     * <pre>
-     * Optional deprecation based on GraphDef versions.
-     * </pre>
-     *
-     * <code>.tensorflow.OpDeprecation deprecation = 8;</code>
-     */
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.OpDeprecation, org.tensorflow.proto.framework.OpDeprecation.Builder, org.tensorflow.proto.framework.OpDeprecationOrBuilder> 
-        getDeprecationFieldBuilder() {
-      if (deprecationBuilder_ == null) {
-        deprecationBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-            org.tensorflow.proto.framework.OpDeprecation, org.tensorflow.proto.framework.OpDeprecation.Builder, org.tensorflow.proto.framework.OpDeprecationOrBuilder>(
-                getDeprecation(),
-                getParentForChildren(),
-                isClean());
-        deprecation_ = null;
-      }
-      return deprecationBuilder_;
-    }
-
-    private java.lang.Object summary_ = "";
-    /**
-     * <pre>
-     * One-line human-readable description of what the Op does.
-     * </pre>
-     *
-     * <code>string summary = 5;</code>
-     */
-    public java.lang.String getSummary() {
-      java.lang.Object ref = summary_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        summary_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * One-line human-readable description of what the Op does.
-     * </pre>
-     *
-     * <code>string summary = 5;</code>
-     */
-    public com.google.protobuf.ByteString
-        getSummaryBytes() {
-      java.lang.Object ref = summary_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        summary_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * One-line human-readable description of what the Op does.
-     * </pre>
-     *
-     * <code>string summary = 5;</code>
-     */
-    public Builder setSummary(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      summary_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * One-line human-readable description of what the Op does.
-     * </pre>
-     *
-     * <code>string summary = 5;</code>
-     */
-    public Builder clearSummary() {
-      
-      summary_ = getDefaultInstance().getSummary();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * One-line human-readable description of what the Op does.
-     * </pre>
-     *
-     * <code>string summary = 5;</code>
-     */
-    public Builder setSummaryBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      summary_ = value;
-      onChanged();
-      return this;
-    }
-
-    private java.lang.Object description_ = "";
-    /**
-     * <pre>
-     * Additional, longer human-readable description of what the Op does.
-     * </pre>
-     *
-     * <code>string description = 6;</code>
-     */
-    public java.lang.String getDescription() {
-      java.lang.Object ref = description_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        description_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Additional, longer human-readable description of what the Op does.
-     * </pre>
-     *
-     * <code>string description = 6;</code>
-     */
-    public com.google.protobuf.ByteString
-        getDescriptionBytes() {
-      java.lang.Object ref = description_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        description_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Additional, longer human-readable description of what the Op does.
-     * </pre>
-     *
-     * <code>string description = 6;</code>
-     */
-    public Builder setDescription(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      description_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Additional, longer human-readable description of what the Op does.
-     * </pre>
-     *
-     * <code>string description = 6;</code>
-     */
-    public Builder clearDescription() {
-      
-      description_ = getDefaultInstance().getDescription();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Additional, longer human-readable description of what the Op does.
-     * </pre>
-     *
-     * <code>string description = 6;</code>
-     */
-    public Builder setDescriptionBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      description_ = value;
-      onChanged();
-      return this;
-    }
-
-    private boolean isCommutative_ ;
-    /**
-     * <pre>
-     * True if the operation is commutative ("op(a,b) == op(b,a)" for all inputs)
-     * </pre>
-     *
-     * <code>bool is_commutative = 18;</code>
-     */
-    public boolean getIsCommutative() {
-      return isCommutative_;
-    }
-    /**
-     * <pre>
-     * True if the operation is commutative ("op(a,b) == op(b,a)" for all inputs)
-     * </pre>
-     *
-     * <code>bool is_commutative = 18;</code>
-     */
-    public Builder setIsCommutative(boolean value) {
-      
-      isCommutative_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * True if the operation is commutative ("op(a,b) == op(b,a)" for all inputs)
-     * </pre>
-     *
-     * <code>bool is_commutative = 18;</code>
-     */
-    public Builder clearIsCommutative() {
-      
-      isCommutative_ = false;
-      onChanged();
-      return this;
-    }
-
-    private boolean isAggregate_ ;
-    /**
-     * <pre>
-     * If is_aggregate is true, then this operation accepts N &gt;= 2
-     * inputs and produces 1 output all of the same type.  Should be
-     * associative and commutative, and produce output with the same
-     * shape as the input.  The optimizer may replace an aggregate op
-     * taking input from multiple devices with a tree of aggregate ops
-     * that aggregate locally within each device (and possibly within
-     * groups of nearby devices) before communicating.
-     * TODO(josh11b): Implement that optimization.
-     * </pre>
-     *
-     * <code>bool is_aggregate = 16;</code>
-     */
-    public boolean getIsAggregate() {
-      return isAggregate_;
-    }
-    /**
-     * <pre>
-     * If is_aggregate is true, then this operation accepts N &gt;= 2
-     * inputs and produces 1 output all of the same type.  Should be
-     * associative and commutative, and produce output with the same
-     * shape as the input.  The optimizer may replace an aggregate op
-     * taking input from multiple devices with a tree of aggregate ops
-     * that aggregate locally within each device (and possibly within
-     * groups of nearby devices) before communicating.
-     * TODO(josh11b): Implement that optimization.
-     * </pre>
-     *
-     * <code>bool is_aggregate = 16;</code>
-     */
-    public Builder setIsAggregate(boolean value) {
-      
-      isAggregate_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * If is_aggregate is true, then this operation accepts N &gt;= 2
-     * inputs and produces 1 output all of the same type.  Should be
-     * associative and commutative, and produce output with the same
-     * shape as the input.  The optimizer may replace an aggregate op
-     * taking input from multiple devices with a tree of aggregate ops
-     * that aggregate locally within each device (and possibly within
-     * groups of nearby devices) before communicating.
-     * TODO(josh11b): Implement that optimization.
-     * </pre>
-     *
-     * <code>bool is_aggregate = 16;</code>
-     */
-    public Builder clearIsAggregate() {
-      
-      isAggregate_ = false;
-      onChanged();
-      return this;
-    }
-
-    private boolean isStateful_ ;
-    /**
-     * <pre>
-     * Ops are marked as stateful if their behavior depends on some state beyond
-     * their input tensors (e.g. variable reading op) or if they have
-     * a side-effect (e.g. printing or asserting ops). Equivalently, stateless ops
-     * must always produce the same output for the same input and have
-     * no side-effects.
-     * By default Ops may be moved between devices.  Stateful ops should
-     * either not be moved, or should only be moved if that state can also
-     * be moved (e.g. via some sort of save / restore).
-     * Stateful ops are guaranteed to never be optimized away by Common
-     * Subexpression Elimination (CSE).
-     * </pre>
-     *
-     * <code>bool is_stateful = 17;</code>
-     */
-    public boolean getIsStateful() {
-      return isStateful_;
-    }
-    /**
-     * <pre>
-     * Ops are marked as stateful if their behavior depends on some state beyond
-     * their input tensors (e.g. variable reading op) or if they have
-     * a side-effect (e.g. printing or asserting ops). Equivalently, stateless ops
-     * must always produce the same output for the same input and have
-     * no side-effects.
-     * By default Ops may be moved between devices.  Stateful ops should
-     * either not be moved, or should only be moved if that state can also
-     * be moved (e.g. via some sort of save / restore).
-     * Stateful ops are guaranteed to never be optimized away by Common
-     * Subexpression Elimination (CSE).
-     * </pre>
-     *
-     * <code>bool is_stateful = 17;</code>
-     */
-    public Builder setIsStateful(boolean value) {
-      
-      isStateful_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Ops are marked as stateful if their behavior depends on some state beyond
-     * their input tensors (e.g. variable reading op) or if they have
-     * a side-effect (e.g. printing or asserting ops). Equivalently, stateless ops
-     * must always produce the same output for the same input and have
-     * no side-effects.
-     * By default Ops may be moved between devices.  Stateful ops should
-     * either not be moved, or should only be moved if that state can also
-     * be moved (e.g. via some sort of save / restore).
-     * Stateful ops are guaranteed to never be optimized away by Common
-     * Subexpression Elimination (CSE).
-     * </pre>
-     *
-     * <code>bool is_stateful = 17;</code>
-     */
-    public Builder clearIsStateful() {
-      
-      isStateful_ = false;
-      onChanged();
-      return this;
-    }
-
-    private boolean allowsUninitializedInput_ ;
-    /**
-     * <pre>
-     * By default, all inputs to an Op must be initialized Tensors.  Ops
-     * that may initialize tensors for the first time should set this
-     * field to true, to allow the Op to take an uninitialized Tensor as
-     * input.
-     * </pre>
-     *
-     * <code>bool allows_uninitialized_input = 19;</code>
-     */
-    public boolean getAllowsUninitializedInput() {
-      return allowsUninitializedInput_;
-    }
-    /**
-     * <pre>
-     * By default, all inputs to an Op must be initialized Tensors.  Ops
-     * that may initialize tensors for the first time should set this
-     * field to true, to allow the Op to take an uninitialized Tensor as
-     * input.
-     * </pre>
-     *
-     * <code>bool allows_uninitialized_input = 19;</code>
-     */
-    public Builder setAllowsUninitializedInput(boolean value) {
-      
-      allowsUninitializedInput_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * By default, all inputs to an Op must be initialized Tensors.  Ops
-     * that may initialize tensors for the first time should set this
-     * field to true, to allow the Op to take an uninitialized Tensor as
-     * input.
-     * </pre>
-     *
-     * <code>bool allows_uninitialized_input = 19;</code>
-     */
-    public Builder clearAllowsUninitializedInput() {
-      
-      allowsUninitializedInput_ = false;
-      onChanged();
-      return this;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.OpDef)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.OpDef)
-  private static final org.tensorflow.proto.framework.OpDef DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.OpDef();
-  }
-
-  public static org.tensorflow.proto.framework.OpDef getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<OpDef>
-      PARSER = new com.google.protobuf.AbstractParser<OpDef>() {
-    @java.lang.Override
-    public OpDef parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new OpDef(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<OpDef> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<OpDef> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.framework.OpDef getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/OpDefOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/OpDefOrBuilder.java
deleted file mode 100644
index 6ac64c0ed97..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/OpDefOrBuilder.java
+++ /dev/null
@@ -1,296 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/framework/op_def.proto
-
-package org.tensorflow.proto.framework;
-
-public interface OpDefOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.OpDef)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <pre>
-   * Op names starting with an underscore are reserved for internal use.
-   * Names should be CamelCase and match the regexp "[A-Z][a-zA-Z0-9&gt;_]*".
-   * </pre>
-   *
-   * <code>string name = 1;</code>
-   */
-  java.lang.String getName();
-  /**
-   * <pre>
-   * Op names starting with an underscore are reserved for internal use.
-   * Names should be CamelCase and match the regexp "[A-Z][a-zA-Z0-9&gt;_]*".
-   * </pre>
-   *
-   * <code>string name = 1;</code>
-   */
-  com.google.protobuf.ByteString
-      getNameBytes();
-
-  /**
-   * <pre>
-   * Description of the input(s).
-   * </pre>
-   *
-   * <code>repeated .tensorflow.OpDef.ArgDef input_arg = 2;</code>
-   */
-  java.util.List<org.tensorflow.proto.framework.OpDef.ArgDef> 
-      getInputArgList();
-  /**
-   * <pre>
-   * Description of the input(s).
-   * </pre>
-   *
-   * <code>repeated .tensorflow.OpDef.ArgDef input_arg = 2;</code>
-   */
-  org.tensorflow.proto.framework.OpDef.ArgDef getInputArg(int index);
-  /**
-   * <pre>
-   * Description of the input(s).
-   * </pre>
-   *
-   * <code>repeated .tensorflow.OpDef.ArgDef input_arg = 2;</code>
-   */
-  int getInputArgCount();
-  /**
-   * <pre>
-   * Description of the input(s).
-   * </pre>
-   *
-   * <code>repeated .tensorflow.OpDef.ArgDef input_arg = 2;</code>
-   */
-  java.util.List<? extends org.tensorflow.proto.framework.OpDef.ArgDefOrBuilder> 
-      getInputArgOrBuilderList();
-  /**
-   * <pre>
-   * Description of the input(s).
-   * </pre>
-   *
-   * <code>repeated .tensorflow.OpDef.ArgDef input_arg = 2;</code>
-   */
-  org.tensorflow.proto.framework.OpDef.ArgDefOrBuilder getInputArgOrBuilder(
-      int index);
-
-  /**
-   * <pre>
-   * Description of the output(s).
-   * </pre>
-   *
-   * <code>repeated .tensorflow.OpDef.ArgDef output_arg = 3;</code>
-   */
-  java.util.List<org.tensorflow.proto.framework.OpDef.ArgDef> 
-      getOutputArgList();
-  /**
-   * <pre>
-   * Description of the output(s).
-   * </pre>
-   *
-   * <code>repeated .tensorflow.OpDef.ArgDef output_arg = 3;</code>
-   */
-  org.tensorflow.proto.framework.OpDef.ArgDef getOutputArg(int index);
-  /**
-   * <pre>
-   * Description of the output(s).
-   * </pre>
-   *
-   * <code>repeated .tensorflow.OpDef.ArgDef output_arg = 3;</code>
-   */
-  int getOutputArgCount();
-  /**
-   * <pre>
-   * Description of the output(s).
-   * </pre>
-   *
-   * <code>repeated .tensorflow.OpDef.ArgDef output_arg = 3;</code>
-   */
-  java.util.List<? extends org.tensorflow.proto.framework.OpDef.ArgDefOrBuilder> 
-      getOutputArgOrBuilderList();
-  /**
-   * <pre>
-   * Description of the output(s).
-   * </pre>
-   *
-   * <code>repeated .tensorflow.OpDef.ArgDef output_arg = 3;</code>
-   */
-  org.tensorflow.proto.framework.OpDef.ArgDefOrBuilder getOutputArgOrBuilder(
-      int index);
-
-  /**
-   * <pre>
-   * Named control outputs for this operation. Useful only for composite
-   * operations (i.e. functions) which want to name different control outputs.
-   * </pre>
-   *
-   * <code>repeated string control_output = 20;</code>
-   */
-  java.util.List<java.lang.String>
-      getControlOutputList();
-  /**
-   * <pre>
-   * Named control outputs for this operation. Useful only for composite
-   * operations (i.e. functions) which want to name different control outputs.
-   * </pre>
-   *
-   * <code>repeated string control_output = 20;</code>
-   */
-  int getControlOutputCount();
-  /**
-   * <pre>
-   * Named control outputs for this operation. Useful only for composite
-   * operations (i.e. functions) which want to name different control outputs.
-   * </pre>
-   *
-   * <code>repeated string control_output = 20;</code>
-   */
-  java.lang.String getControlOutput(int index);
-  /**
-   * <pre>
-   * Named control outputs for this operation. Useful only for composite
-   * operations (i.e. functions) which want to name different control outputs.
-   * </pre>
-   *
-   * <code>repeated string control_output = 20;</code>
-   */
-  com.google.protobuf.ByteString
-      getControlOutputBytes(int index);
-
-  /**
-   * <code>repeated .tensorflow.OpDef.AttrDef attr = 4;</code>
-   */
-  java.util.List<org.tensorflow.proto.framework.OpDef.AttrDef> 
-      getAttrList();
-  /**
-   * <code>repeated .tensorflow.OpDef.AttrDef attr = 4;</code>
-   */
-  org.tensorflow.proto.framework.OpDef.AttrDef getAttr(int index);
-  /**
-   * <code>repeated .tensorflow.OpDef.AttrDef attr = 4;</code>
-   */
-  int getAttrCount();
-  /**
-   * <code>repeated .tensorflow.OpDef.AttrDef attr = 4;</code>
-   */
-  java.util.List<? extends org.tensorflow.proto.framework.OpDef.AttrDefOrBuilder> 
-      getAttrOrBuilderList();
-  /**
-   * <code>repeated .tensorflow.OpDef.AttrDef attr = 4;</code>
-   */
-  org.tensorflow.proto.framework.OpDef.AttrDefOrBuilder getAttrOrBuilder(
-      int index);
-
-  /**
-   * <pre>
-   * Optional deprecation based on GraphDef versions.
-   * </pre>
-   *
-   * <code>.tensorflow.OpDeprecation deprecation = 8;</code>
-   */
-  boolean hasDeprecation();
-  /**
-   * <pre>
-   * Optional deprecation based on GraphDef versions.
-   * </pre>
-   *
-   * <code>.tensorflow.OpDeprecation deprecation = 8;</code>
-   */
-  org.tensorflow.proto.framework.OpDeprecation getDeprecation();
-  /**
-   * <pre>
-   * Optional deprecation based on GraphDef versions.
-   * </pre>
-   *
-   * <code>.tensorflow.OpDeprecation deprecation = 8;</code>
-   */
-  org.tensorflow.proto.framework.OpDeprecationOrBuilder getDeprecationOrBuilder();
-
-  /**
-   * <pre>
-   * One-line human-readable description of what the Op does.
-   * </pre>
-   *
-   * <code>string summary = 5;</code>
-   */
-  java.lang.String getSummary();
-  /**
-   * <pre>
-   * One-line human-readable description of what the Op does.
-   * </pre>
-   *
-   * <code>string summary = 5;</code>
-   */
-  com.google.protobuf.ByteString
-      getSummaryBytes();
-
-  /**
-   * <pre>
-   * Additional, longer human-readable description of what the Op does.
-   * </pre>
-   *
-   * <code>string description = 6;</code>
-   */
-  java.lang.String getDescription();
-  /**
-   * <pre>
-   * Additional, longer human-readable description of what the Op does.
-   * </pre>
-   *
-   * <code>string description = 6;</code>
-   */
-  com.google.protobuf.ByteString
-      getDescriptionBytes();
-
-  /**
-   * <pre>
-   * True if the operation is commutative ("op(a,b) == op(b,a)" for all inputs)
-   * </pre>
-   *
-   * <code>bool is_commutative = 18;</code>
-   */
-  boolean getIsCommutative();
-
-  /**
-   * <pre>
-   * If is_aggregate is true, then this operation accepts N &gt;= 2
-   * inputs and produces 1 output all of the same type.  Should be
-   * associative and commutative, and produce output with the same
-   * shape as the input.  The optimizer may replace an aggregate op
-   * taking input from multiple devices with a tree of aggregate ops
-   * that aggregate locally within each device (and possibly within
-   * groups of nearby devices) before communicating.
-   * TODO(josh11b): Implement that optimization.
-   * </pre>
-   *
-   * <code>bool is_aggregate = 16;</code>
-   */
-  boolean getIsAggregate();
-
-  /**
-   * <pre>
-   * Ops are marked as stateful if their behavior depends on some state beyond
-   * their input tensors (e.g. variable reading op) or if they have
-   * a side-effect (e.g. printing or asserting ops). Equivalently, stateless ops
-   * must always produce the same output for the same input and have
-   * no side-effects.
-   * By default Ops may be moved between devices.  Stateful ops should
-   * either not be moved, or should only be moved if that state can also
-   * be moved (e.g. via some sort of save / restore).
-   * Stateful ops are guaranteed to never be optimized away by Common
-   * Subexpression Elimination (CSE).
-   * </pre>
-   *
-   * <code>bool is_stateful = 17;</code>
-   */
-  boolean getIsStateful();
-
-  /**
-   * <pre>
-   * By default, all inputs to an Op must be initialized Tensors.  Ops
-   * that may initialize tensors for the first time should set this
-   * field to true, to allow the Op to take an uninitialized Tensor as
-   * input.
-   * </pre>
-   *
-   * <code>bool allows_uninitialized_input = 19;</code>
-   */
-  boolean getAllowsUninitializedInput();
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/OpDefProtos.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/OpDefProtos.java
deleted file mode 100644
index 8beef161f83..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/OpDefProtos.java
+++ /dev/null
@@ -1,121 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/framework/op_def.proto
-
-package org.tensorflow.proto.framework;
-
-public final class OpDefProtos {
-  private OpDefProtos() {}
-  public static void registerAllExtensions(
-      com.google.protobuf.ExtensionRegistryLite registry) {
-  }
-
-  public static void registerAllExtensions(
-      com.google.protobuf.ExtensionRegistry registry) {
-    registerAllExtensions(
-        (com.google.protobuf.ExtensionRegistryLite) registry);
-  }
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_OpDef_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_OpDef_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_OpDef_ArgDef_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_OpDef_ArgDef_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_OpDef_AttrDef_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_OpDef_AttrDef_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_OpDeprecation_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_OpDeprecation_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_OpList_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_OpList_fieldAccessorTable;
-
-  public static com.google.protobuf.Descriptors.FileDescriptor
-      getDescriptor() {
-    return descriptor;
-  }
-  private static  com.google.protobuf.Descriptors.FileDescriptor
-      descriptor;
-  static {
-    java.lang.String[] descriptorData = {
-      "\n&tensorflow/core/framework/op_def.proto" +
-      "\022\ntensorflow\032*tensorflow/core/framework/" +
-      "attr_value.proto\032%tensorflow/core/framew" +
-      "ork/types.proto\"\320\005\n\005OpDef\022\014\n\004name\030\001 \001(\t\022" +
-      "+\n\tinput_arg\030\002 \003(\0132\030.tensorflow.OpDef.Ar" +
-      "gDef\022,\n\noutput_arg\030\003 \003(\0132\030.tensorflow.Op" +
-      "Def.ArgDef\022\026\n\016control_output\030\024 \003(\t\022\'\n\004at" +
-      "tr\030\004 \003(\0132\031.tensorflow.OpDef.AttrDef\022.\n\013d" +
-      "eprecation\030\010 \001(\0132\031.tensorflow.OpDeprecat" +
-      "ion\022\017\n\007summary\030\005 \001(\t\022\023\n\013description\030\006 \001(" +
-      "\t\022\026\n\016is_commutative\030\022 \001(\010\022\024\n\014is_aggregat" +
-      "e\030\020 \001(\010\022\023\n\013is_stateful\030\021 \001(\010\022\"\n\032allows_u" +
-      "ninitialized_input\030\023 \001(\010\032\237\001\n\006ArgDef\022\014\n\004n" +
-      "ame\030\001 \001(\t\022\023\n\013description\030\002 \001(\t\022\"\n\004type\030\003" +
-      " \001(\0162\024.tensorflow.DataType\022\021\n\ttype_attr\030" +
-      "\004 \001(\t\022\023\n\013number_attr\030\005 \001(\t\022\026\n\016type_list_" +
-      "attr\030\006 \001(\t\022\016\n\006is_ref\030\020 \001(\010\032\275\001\n\007AttrDef\022\014" +
-      "\n\004name\030\001 \001(\t\022\014\n\004type\030\002 \001(\t\022,\n\rdefault_va" +
-      "lue\030\003 \001(\0132\025.tensorflow.AttrValue\022\023\n\013desc" +
-      "ription\030\004 \001(\t\022\023\n\013has_minimum\030\005 \001(\010\022\017\n\007mi" +
-      "nimum\030\006 \001(\003\022-\n\016allowed_values\030\007 \001(\0132\025.te" +
-      "nsorflow.AttrValue\"5\n\rOpDeprecation\022\017\n\007v" +
-      "ersion\030\001 \001(\005\022\023\n\013explanation\030\002 \001(\t\"\'\n\006OpL" +
-      "ist\022\035\n\002op\030\001 \003(\0132\021.tensorflow.OpDefB\201\001\n\036o" +
-      "rg.tensorflow.proto.frameworkB\013OpDefProt" +
-      "osP\001ZMgithub.com/tensorflow/tensorflow/t" +
-      "ensorflow/go/core/framework/op_def_go_pr" +
-      "oto\370\001\001b\006proto3"
-    };
-    descriptor = com.google.protobuf.Descriptors.FileDescriptor
-      .internalBuildGeneratedFileFrom(descriptorData,
-        new com.google.protobuf.Descriptors.FileDescriptor[] {
-          org.tensorflow.proto.framework.AttrValueProtos.getDescriptor(),
-          org.tensorflow.proto.framework.TypesProtos.getDescriptor(),
-        });
-    internal_static_tensorflow_OpDef_descriptor =
-      getDescriptor().getMessageTypes().get(0);
-    internal_static_tensorflow_OpDef_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_OpDef_descriptor,
-        new java.lang.String[] { "Name", "InputArg", "OutputArg", "ControlOutput", "Attr", "Deprecation", "Summary", "Description", "IsCommutative", "IsAggregate", "IsStateful", "AllowsUninitializedInput", });
-    internal_static_tensorflow_OpDef_ArgDef_descriptor =
-      internal_static_tensorflow_OpDef_descriptor.getNestedTypes().get(0);
-    internal_static_tensorflow_OpDef_ArgDef_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_OpDef_ArgDef_descriptor,
-        new java.lang.String[] { "Name", "Description", "Type", "TypeAttr", "NumberAttr", "TypeListAttr", "IsRef", });
-    internal_static_tensorflow_OpDef_AttrDef_descriptor =
-      internal_static_tensorflow_OpDef_descriptor.getNestedTypes().get(1);
-    internal_static_tensorflow_OpDef_AttrDef_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_OpDef_AttrDef_descriptor,
-        new java.lang.String[] { "Name", "Type", "DefaultValue", "Description", "HasMinimum", "Minimum", "AllowedValues", });
-    internal_static_tensorflow_OpDeprecation_descriptor =
-      getDescriptor().getMessageTypes().get(1);
-    internal_static_tensorflow_OpDeprecation_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_OpDeprecation_descriptor,
-        new java.lang.String[] { "Version", "Explanation", });
-    internal_static_tensorflow_OpList_descriptor =
-      getDescriptor().getMessageTypes().get(2);
-    internal_static_tensorflow_OpList_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_OpList_descriptor,
-        new java.lang.String[] { "Op", });
-    org.tensorflow.proto.framework.AttrValueProtos.getDescriptor();
-    org.tensorflow.proto.framework.TypesProtos.getDescriptor();
-  }
-
-  // @@protoc_insertion_point(outer_class_scope)
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/OpDeprecation.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/OpDeprecation.java
deleted file mode 100644
index f33a34c9052..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/OpDeprecation.java
+++ /dev/null
@@ -1,655 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/framework/op_def.proto
-
-package org.tensorflow.proto.framework;
-
-/**
- * <pre>
- * Information about version-dependent deprecation of an op
- * </pre>
- *
- * Protobuf type {@code tensorflow.OpDeprecation}
- */
-public  final class OpDeprecation extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.OpDeprecation)
-    OpDeprecationOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use OpDeprecation.newBuilder() to construct.
-  private OpDeprecation(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private OpDeprecation() {
-    explanation_ = "";
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new OpDeprecation();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private OpDeprecation(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 8: {
-
-            version_ = input.readInt32();
-            break;
-          }
-          case 18: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            explanation_ = s;
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.framework.OpDefProtos.internal_static_tensorflow_OpDeprecation_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.framework.OpDefProtos.internal_static_tensorflow_OpDeprecation_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.framework.OpDeprecation.class, org.tensorflow.proto.framework.OpDeprecation.Builder.class);
-  }
-
-  public static final int VERSION_FIELD_NUMBER = 1;
-  private int version_;
-  /**
-   * <pre>
-   * First GraphDef version at which the op is disallowed.
-   * </pre>
-   *
-   * <code>int32 version = 1;</code>
-   */
-  public int getVersion() {
-    return version_;
-  }
-
-  public static final int EXPLANATION_FIELD_NUMBER = 2;
-  private volatile java.lang.Object explanation_;
-  /**
-   * <pre>
-   * Explanation of why it was deprecated and what to use instead.
-   * </pre>
-   *
-   * <code>string explanation = 2;</code>
-   */
-  public java.lang.String getExplanation() {
-    java.lang.Object ref = explanation_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      explanation_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * Explanation of why it was deprecated and what to use instead.
-   * </pre>
-   *
-   * <code>string explanation = 2;</code>
-   */
-  public com.google.protobuf.ByteString
-      getExplanationBytes() {
-    java.lang.Object ref = explanation_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      explanation_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    if (version_ != 0) {
-      output.writeInt32(1, version_);
-    }
-    if (!getExplanationBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 2, explanation_);
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (version_ != 0) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt32Size(1, version_);
-    }
-    if (!getExplanationBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(2, explanation_);
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.framework.OpDeprecation)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.framework.OpDeprecation other = (org.tensorflow.proto.framework.OpDeprecation) obj;
-
-    if (getVersion()
-        != other.getVersion()) return false;
-    if (!getExplanation()
-        .equals(other.getExplanation())) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    hash = (37 * hash) + VERSION_FIELD_NUMBER;
-    hash = (53 * hash) + getVersion();
-    hash = (37 * hash) + EXPLANATION_FIELD_NUMBER;
-    hash = (53 * hash) + getExplanation().hashCode();
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.framework.OpDeprecation parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.OpDeprecation parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.OpDeprecation parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.OpDeprecation parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.OpDeprecation parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.OpDeprecation parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.OpDeprecation parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.OpDeprecation parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.OpDeprecation parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.OpDeprecation parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.OpDeprecation parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.OpDeprecation parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.framework.OpDeprecation prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * <pre>
-   * Information about version-dependent deprecation of an op
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.OpDeprecation}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.OpDeprecation)
-      org.tensorflow.proto.framework.OpDeprecationOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.OpDefProtos.internal_static_tensorflow_OpDeprecation_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.OpDefProtos.internal_static_tensorflow_OpDeprecation_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.OpDeprecation.class, org.tensorflow.proto.framework.OpDeprecation.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.framework.OpDeprecation.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      version_ = 0;
-
-      explanation_ = "";
-
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.framework.OpDefProtos.internal_static_tensorflow_OpDeprecation_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.OpDeprecation getDefaultInstanceForType() {
-      return org.tensorflow.proto.framework.OpDeprecation.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.OpDeprecation build() {
-      org.tensorflow.proto.framework.OpDeprecation result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.OpDeprecation buildPartial() {
-      org.tensorflow.proto.framework.OpDeprecation result = new org.tensorflow.proto.framework.OpDeprecation(this);
-      result.version_ = version_;
-      result.explanation_ = explanation_;
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.framework.OpDeprecation) {
-        return mergeFrom((org.tensorflow.proto.framework.OpDeprecation)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.framework.OpDeprecation other) {
-      if (other == org.tensorflow.proto.framework.OpDeprecation.getDefaultInstance()) return this;
-      if (other.getVersion() != 0) {
-        setVersion(other.getVersion());
-      }
-      if (!other.getExplanation().isEmpty()) {
-        explanation_ = other.explanation_;
-        onChanged();
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.framework.OpDeprecation parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.framework.OpDeprecation) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-
-    private int version_ ;
-    /**
-     * <pre>
-     * First GraphDef version at which the op is disallowed.
-     * </pre>
-     *
-     * <code>int32 version = 1;</code>
-     */
-    public int getVersion() {
-      return version_;
-    }
-    /**
-     * <pre>
-     * First GraphDef version at which the op is disallowed.
-     * </pre>
-     *
-     * <code>int32 version = 1;</code>
-     */
-    public Builder setVersion(int value) {
-      
-      version_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * First GraphDef version at which the op is disallowed.
-     * </pre>
-     *
-     * <code>int32 version = 1;</code>
-     */
-    public Builder clearVersion() {
-      
-      version_ = 0;
-      onChanged();
-      return this;
-    }
-
-    private java.lang.Object explanation_ = "";
-    /**
-     * <pre>
-     * Explanation of why it was deprecated and what to use instead.
-     * </pre>
-     *
-     * <code>string explanation = 2;</code>
-     */
-    public java.lang.String getExplanation() {
-      java.lang.Object ref = explanation_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        explanation_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Explanation of why it was deprecated and what to use instead.
-     * </pre>
-     *
-     * <code>string explanation = 2;</code>
-     */
-    public com.google.protobuf.ByteString
-        getExplanationBytes() {
-      java.lang.Object ref = explanation_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        explanation_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Explanation of why it was deprecated and what to use instead.
-     * </pre>
-     *
-     * <code>string explanation = 2;</code>
-     */
-    public Builder setExplanation(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      explanation_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Explanation of why it was deprecated and what to use instead.
-     * </pre>
-     *
-     * <code>string explanation = 2;</code>
-     */
-    public Builder clearExplanation() {
-      
-      explanation_ = getDefaultInstance().getExplanation();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Explanation of why it was deprecated and what to use instead.
-     * </pre>
-     *
-     * <code>string explanation = 2;</code>
-     */
-    public Builder setExplanationBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      explanation_ = value;
-      onChanged();
-      return this;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.OpDeprecation)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.OpDeprecation)
-  private static final org.tensorflow.proto.framework.OpDeprecation DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.OpDeprecation();
-  }
-
-  public static org.tensorflow.proto.framework.OpDeprecation getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<OpDeprecation>
-      PARSER = new com.google.protobuf.AbstractParser<OpDeprecation>() {
-    @java.lang.Override
-    public OpDeprecation parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new OpDeprecation(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<OpDeprecation> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<OpDeprecation> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.framework.OpDeprecation getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/OpDeprecationOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/OpDeprecationOrBuilder.java
deleted file mode 100644
index 3248db05508..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/OpDeprecationOrBuilder.java
+++ /dev/null
@@ -1,36 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/framework/op_def.proto
-
-package org.tensorflow.proto.framework;
-
-public interface OpDeprecationOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.OpDeprecation)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <pre>
-   * First GraphDef version at which the op is disallowed.
-   * </pre>
-   *
-   * <code>int32 version = 1;</code>
-   */
-  int getVersion();
-
-  /**
-   * <pre>
-   * Explanation of why it was deprecated and what to use instead.
-   * </pre>
-   *
-   * <code>string explanation = 2;</code>
-   */
-  java.lang.String getExplanation();
-  /**
-   * <pre>
-   * Explanation of why it was deprecated and what to use instead.
-   * </pre>
-   *
-   * <code>string explanation = 2;</code>
-   */
-  com.google.protobuf.ByteString
-      getExplanationBytes();
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/OpList.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/OpList.java
deleted file mode 100644
index 1be81195b29..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/OpList.java
+++ /dev/null
@@ -1,773 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/framework/op_def.proto
-
-package org.tensorflow.proto.framework;
-
-/**
- * <pre>
- * A collection of OpDefs
- * </pre>
- *
- * Protobuf type {@code tensorflow.OpList}
- */
-public  final class OpList extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.OpList)
-    OpListOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use OpList.newBuilder() to construct.
-  private OpList(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private OpList() {
-    op_ = java.util.Collections.emptyList();
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new OpList();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private OpList(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    int mutable_bitField0_ = 0;
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 10: {
-            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
-              op_ = new java.util.ArrayList<org.tensorflow.proto.framework.OpDef>();
-              mutable_bitField0_ |= 0x00000001;
-            }
-            op_.add(
-                input.readMessage(org.tensorflow.proto.framework.OpDef.parser(), extensionRegistry));
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      if (((mutable_bitField0_ & 0x00000001) != 0)) {
-        op_ = java.util.Collections.unmodifiableList(op_);
-      }
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.framework.OpDefProtos.internal_static_tensorflow_OpList_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.framework.OpDefProtos.internal_static_tensorflow_OpList_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.framework.OpList.class, org.tensorflow.proto.framework.OpList.Builder.class);
-  }
-
-  public static final int OP_FIELD_NUMBER = 1;
-  private java.util.List<org.tensorflow.proto.framework.OpDef> op_;
-  /**
-   * <code>repeated .tensorflow.OpDef op = 1;</code>
-   */
-  public java.util.List<org.tensorflow.proto.framework.OpDef> getOpList() {
-    return op_;
-  }
-  /**
-   * <code>repeated .tensorflow.OpDef op = 1;</code>
-   */
-  public java.util.List<? extends org.tensorflow.proto.framework.OpDefOrBuilder> 
-      getOpOrBuilderList() {
-    return op_;
-  }
-  /**
-   * <code>repeated .tensorflow.OpDef op = 1;</code>
-   */
-  public int getOpCount() {
-    return op_.size();
-  }
-  /**
-   * <code>repeated .tensorflow.OpDef op = 1;</code>
-   */
-  public org.tensorflow.proto.framework.OpDef getOp(int index) {
-    return op_.get(index);
-  }
-  /**
-   * <code>repeated .tensorflow.OpDef op = 1;</code>
-   */
-  public org.tensorflow.proto.framework.OpDefOrBuilder getOpOrBuilder(
-      int index) {
-    return op_.get(index);
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    for (int i = 0; i < op_.size(); i++) {
-      output.writeMessage(1, op_.get(i));
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    for (int i = 0; i < op_.size(); i++) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(1, op_.get(i));
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.framework.OpList)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.framework.OpList other = (org.tensorflow.proto.framework.OpList) obj;
-
-    if (!getOpList()
-        .equals(other.getOpList())) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    if (getOpCount() > 0) {
-      hash = (37 * hash) + OP_FIELD_NUMBER;
-      hash = (53 * hash) + getOpList().hashCode();
-    }
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.framework.OpList parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.OpList parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.OpList parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.OpList parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.OpList parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.OpList parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.OpList parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.OpList parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.OpList parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.OpList parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.OpList parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.OpList parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.framework.OpList prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * <pre>
-   * A collection of OpDefs
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.OpList}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.OpList)
-      org.tensorflow.proto.framework.OpListOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.OpDefProtos.internal_static_tensorflow_OpList_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.OpDefProtos.internal_static_tensorflow_OpList_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.OpList.class, org.tensorflow.proto.framework.OpList.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.framework.OpList.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-        getOpFieldBuilder();
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      if (opBuilder_ == null) {
-        op_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000001);
-      } else {
-        opBuilder_.clear();
-      }
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.framework.OpDefProtos.internal_static_tensorflow_OpList_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.OpList getDefaultInstanceForType() {
-      return org.tensorflow.proto.framework.OpList.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.OpList build() {
-      org.tensorflow.proto.framework.OpList result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.OpList buildPartial() {
-      org.tensorflow.proto.framework.OpList result = new org.tensorflow.proto.framework.OpList(this);
-      int from_bitField0_ = bitField0_;
-      if (opBuilder_ == null) {
-        if (((bitField0_ & 0x00000001) != 0)) {
-          op_ = java.util.Collections.unmodifiableList(op_);
-          bitField0_ = (bitField0_ & ~0x00000001);
-        }
-        result.op_ = op_;
-      } else {
-        result.op_ = opBuilder_.build();
-      }
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.framework.OpList) {
-        return mergeFrom((org.tensorflow.proto.framework.OpList)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.framework.OpList other) {
-      if (other == org.tensorflow.proto.framework.OpList.getDefaultInstance()) return this;
-      if (opBuilder_ == null) {
-        if (!other.op_.isEmpty()) {
-          if (op_.isEmpty()) {
-            op_ = other.op_;
-            bitField0_ = (bitField0_ & ~0x00000001);
-          } else {
-            ensureOpIsMutable();
-            op_.addAll(other.op_);
-          }
-          onChanged();
-        }
-      } else {
-        if (!other.op_.isEmpty()) {
-          if (opBuilder_.isEmpty()) {
-            opBuilder_.dispose();
-            opBuilder_ = null;
-            op_ = other.op_;
-            bitField0_ = (bitField0_ & ~0x00000001);
-            opBuilder_ = 
-              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
-                 getOpFieldBuilder() : null;
-          } else {
-            opBuilder_.addAllMessages(other.op_);
-          }
-        }
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.framework.OpList parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.framework.OpList) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-    private int bitField0_;
-
-    private java.util.List<org.tensorflow.proto.framework.OpDef> op_ =
-      java.util.Collections.emptyList();
-    private void ensureOpIsMutable() {
-      if (!((bitField0_ & 0x00000001) != 0)) {
-        op_ = new java.util.ArrayList<org.tensorflow.proto.framework.OpDef>(op_);
-        bitField0_ |= 0x00000001;
-       }
-    }
-
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.framework.OpDef, org.tensorflow.proto.framework.OpDef.Builder, org.tensorflow.proto.framework.OpDefOrBuilder> opBuilder_;
-
-    /**
-     * <code>repeated .tensorflow.OpDef op = 1;</code>
-     */
-    public java.util.List<org.tensorflow.proto.framework.OpDef> getOpList() {
-      if (opBuilder_ == null) {
-        return java.util.Collections.unmodifiableList(op_);
-      } else {
-        return opBuilder_.getMessageList();
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.OpDef op = 1;</code>
-     */
-    public int getOpCount() {
-      if (opBuilder_ == null) {
-        return op_.size();
-      } else {
-        return opBuilder_.getCount();
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.OpDef op = 1;</code>
-     */
-    public org.tensorflow.proto.framework.OpDef getOp(int index) {
-      if (opBuilder_ == null) {
-        return op_.get(index);
-      } else {
-        return opBuilder_.getMessage(index);
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.OpDef op = 1;</code>
-     */
-    public Builder setOp(
-        int index, org.tensorflow.proto.framework.OpDef value) {
-      if (opBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureOpIsMutable();
-        op_.set(index, value);
-        onChanged();
-      } else {
-        opBuilder_.setMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.OpDef op = 1;</code>
-     */
-    public Builder setOp(
-        int index, org.tensorflow.proto.framework.OpDef.Builder builderForValue) {
-      if (opBuilder_ == null) {
-        ensureOpIsMutable();
-        op_.set(index, builderForValue.build());
-        onChanged();
-      } else {
-        opBuilder_.setMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.OpDef op = 1;</code>
-     */
-    public Builder addOp(org.tensorflow.proto.framework.OpDef value) {
-      if (opBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureOpIsMutable();
-        op_.add(value);
-        onChanged();
-      } else {
-        opBuilder_.addMessage(value);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.OpDef op = 1;</code>
-     */
-    public Builder addOp(
-        int index, org.tensorflow.proto.framework.OpDef value) {
-      if (opBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureOpIsMutable();
-        op_.add(index, value);
-        onChanged();
-      } else {
-        opBuilder_.addMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.OpDef op = 1;</code>
-     */
-    public Builder addOp(
-        org.tensorflow.proto.framework.OpDef.Builder builderForValue) {
-      if (opBuilder_ == null) {
-        ensureOpIsMutable();
-        op_.add(builderForValue.build());
-        onChanged();
-      } else {
-        opBuilder_.addMessage(builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.OpDef op = 1;</code>
-     */
-    public Builder addOp(
-        int index, org.tensorflow.proto.framework.OpDef.Builder builderForValue) {
-      if (opBuilder_ == null) {
-        ensureOpIsMutable();
-        op_.add(index, builderForValue.build());
-        onChanged();
-      } else {
-        opBuilder_.addMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.OpDef op = 1;</code>
-     */
-    public Builder addAllOp(
-        java.lang.Iterable<? extends org.tensorflow.proto.framework.OpDef> values) {
-      if (opBuilder_ == null) {
-        ensureOpIsMutable();
-        com.google.protobuf.AbstractMessageLite.Builder.addAll(
-            values, op_);
-        onChanged();
-      } else {
-        opBuilder_.addAllMessages(values);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.OpDef op = 1;</code>
-     */
-    public Builder clearOp() {
-      if (opBuilder_ == null) {
-        op_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000001);
-        onChanged();
-      } else {
-        opBuilder_.clear();
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.OpDef op = 1;</code>
-     */
-    public Builder removeOp(int index) {
-      if (opBuilder_ == null) {
-        ensureOpIsMutable();
-        op_.remove(index);
-        onChanged();
-      } else {
-        opBuilder_.remove(index);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.OpDef op = 1;</code>
-     */
-    public org.tensorflow.proto.framework.OpDef.Builder getOpBuilder(
-        int index) {
-      return getOpFieldBuilder().getBuilder(index);
-    }
-    /**
-     * <code>repeated .tensorflow.OpDef op = 1;</code>
-     */
-    public org.tensorflow.proto.framework.OpDefOrBuilder getOpOrBuilder(
-        int index) {
-      if (opBuilder_ == null) {
-        return op_.get(index);  } else {
-        return opBuilder_.getMessageOrBuilder(index);
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.OpDef op = 1;</code>
-     */
-    public java.util.List<? extends org.tensorflow.proto.framework.OpDefOrBuilder> 
-         getOpOrBuilderList() {
-      if (opBuilder_ != null) {
-        return opBuilder_.getMessageOrBuilderList();
-      } else {
-        return java.util.Collections.unmodifiableList(op_);
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.OpDef op = 1;</code>
-     */
-    public org.tensorflow.proto.framework.OpDef.Builder addOpBuilder() {
-      return getOpFieldBuilder().addBuilder(
-          org.tensorflow.proto.framework.OpDef.getDefaultInstance());
-    }
-    /**
-     * <code>repeated .tensorflow.OpDef op = 1;</code>
-     */
-    public org.tensorflow.proto.framework.OpDef.Builder addOpBuilder(
-        int index) {
-      return getOpFieldBuilder().addBuilder(
-          index, org.tensorflow.proto.framework.OpDef.getDefaultInstance());
-    }
-    /**
-     * <code>repeated .tensorflow.OpDef op = 1;</code>
-     */
-    public java.util.List<org.tensorflow.proto.framework.OpDef.Builder> 
-         getOpBuilderList() {
-      return getOpFieldBuilder().getBuilderList();
-    }
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.framework.OpDef, org.tensorflow.proto.framework.OpDef.Builder, org.tensorflow.proto.framework.OpDefOrBuilder> 
-        getOpFieldBuilder() {
-      if (opBuilder_ == null) {
-        opBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
-            org.tensorflow.proto.framework.OpDef, org.tensorflow.proto.framework.OpDef.Builder, org.tensorflow.proto.framework.OpDefOrBuilder>(
-                op_,
-                ((bitField0_ & 0x00000001) != 0),
-                getParentForChildren(),
-                isClean());
-        op_ = null;
-      }
-      return opBuilder_;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.OpList)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.OpList)
-  private static final org.tensorflow.proto.framework.OpList DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.OpList();
-  }
-
-  public static org.tensorflow.proto.framework.OpList getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<OpList>
-      PARSER = new com.google.protobuf.AbstractParser<OpList>() {
-    @java.lang.Override
-    public OpList parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new OpList(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<OpList> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<OpList> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.framework.OpList getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/OpListOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/OpListOrBuilder.java
deleted file mode 100644
index a3fc1fa9856..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/OpListOrBuilder.java
+++ /dev/null
@@ -1,33 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/framework/op_def.proto
-
-package org.tensorflow.proto.framework;
-
-public interface OpListOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.OpList)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <code>repeated .tensorflow.OpDef op = 1;</code>
-   */
-  java.util.List<org.tensorflow.proto.framework.OpDef> 
-      getOpList();
-  /**
-   * <code>repeated .tensorflow.OpDef op = 1;</code>
-   */
-  org.tensorflow.proto.framework.OpDef getOp(int index);
-  /**
-   * <code>repeated .tensorflow.OpDef op = 1;</code>
-   */
-  int getOpCount();
-  /**
-   * <code>repeated .tensorflow.OpDef op = 1;</code>
-   */
-  java.util.List<? extends org.tensorflow.proto.framework.OpDefOrBuilder> 
-      getOpOrBuilderList();
-  /**
-   * <code>repeated .tensorflow.OpDef op = 1;</code>
-   */
-  org.tensorflow.proto.framework.OpDefOrBuilder getOpOrBuilder(
-      int index);
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/OptimizerOptions.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/OptimizerOptions.java
deleted file mode 100644
index 2afa5a96067..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/OptimizerOptions.java
+++ /dev/null
@@ -1,1210 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/config.proto
-
-package org.tensorflow.proto.framework;
-
-/**
- * <pre>
- * Options passed to the graph optimizer
- * </pre>
- *
- * Protobuf type {@code tensorflow.OptimizerOptions}
- */
-public  final class OptimizerOptions extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.OptimizerOptions)
-    OptimizerOptionsOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use OptimizerOptions.newBuilder() to construct.
-  private OptimizerOptions(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private OptimizerOptions() {
-    optLevel_ = 0;
-    globalJitLevel_ = 0;
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new OptimizerOptions();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private OptimizerOptions(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 8: {
-
-            doCommonSubexpressionElimination_ = input.readBool();
-            break;
-          }
-          case 16: {
-
-            doConstantFolding_ = input.readBool();
-            break;
-          }
-          case 24: {
-            int rawValue = input.readEnum();
-
-            optLevel_ = rawValue;
-            break;
-          }
-          case 32: {
-
-            doFunctionInlining_ = input.readBool();
-            break;
-          }
-          case 40: {
-            int rawValue = input.readEnum();
-
-            globalJitLevel_ = rawValue;
-            break;
-          }
-          case 48: {
-
-            maxFoldedConstantInBytes_ = input.readInt64();
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_OptimizerOptions_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_OptimizerOptions_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.framework.OptimizerOptions.class, org.tensorflow.proto.framework.OptimizerOptions.Builder.class);
-  }
-
-  /**
-   * <pre>
-   * Optimization level
-   * </pre>
-   *
-   * Protobuf enum {@code tensorflow.OptimizerOptions.Level}
-   */
-  public enum Level
-      implements com.google.protobuf.ProtocolMessageEnum {
-    /**
-     * <pre>
-     * L1 is the default level.
-     * Optimization performed at L1 :
-     * 1. Common subexpression elimination
-     * 2. Constant folding
-     * </pre>
-     *
-     * <code>L1 = 0;</code>
-     */
-    L1(0),
-    /**
-     * <pre>
-     * No optimizations
-     * </pre>
-     *
-     * <code>L0 = -1;</code>
-     */
-    L0(-1),
-    UNRECOGNIZED(-1),
-    ;
-
-    /**
-     * <pre>
-     * L1 is the default level.
-     * Optimization performed at L1 :
-     * 1. Common subexpression elimination
-     * 2. Constant folding
-     * </pre>
-     *
-     * <code>L1 = 0;</code>
-     */
-    public static final int L1_VALUE = 0;
-    /**
-     * <pre>
-     * No optimizations
-     * </pre>
-     *
-     * <code>L0 = -1;</code>
-     */
-    public static final int L0_VALUE = -1;
-
-
-    public final int getNumber() {
-      if (this == UNRECOGNIZED) {
-        throw new java.lang.IllegalArgumentException(
-            "Can't get the number of an unknown enum value.");
-      }
-      return value;
-    }
-
-    /**
-     * @deprecated Use {@link #forNumber(int)} instead.
-     */
-    @java.lang.Deprecated
-    public static Level valueOf(int value) {
-      return forNumber(value);
-    }
-
-    public static Level forNumber(int value) {
-      switch (value) {
-        case 0: return L1;
-        case -1: return L0;
-        default: return null;
-      }
-    }
-
-    public static com.google.protobuf.Internal.EnumLiteMap<Level>
-        internalGetValueMap() {
-      return internalValueMap;
-    }
-    private static final com.google.protobuf.Internal.EnumLiteMap<
-        Level> internalValueMap =
-          new com.google.protobuf.Internal.EnumLiteMap<Level>() {
-            public Level findValueByNumber(int number) {
-              return Level.forNumber(number);
-            }
-          };
-
-    public final com.google.protobuf.Descriptors.EnumValueDescriptor
-        getValueDescriptor() {
-      return getDescriptor().getValues().get(ordinal());
-    }
-    public final com.google.protobuf.Descriptors.EnumDescriptor
-        getDescriptorForType() {
-      return getDescriptor();
-    }
-    public static final com.google.protobuf.Descriptors.EnumDescriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.OptimizerOptions.getDescriptor().getEnumTypes().get(0);
-    }
-
-    private static final Level[] VALUES = values();
-
-    public static Level valueOf(
-        com.google.protobuf.Descriptors.EnumValueDescriptor desc) {
-      if (desc.getType() != getDescriptor()) {
-        throw new java.lang.IllegalArgumentException(
-          "EnumValueDescriptor is not for this type.");
-      }
-      if (desc.getIndex() == -1) {
-        return UNRECOGNIZED;
-      }
-      return VALUES[desc.getIndex()];
-    }
-
-    private final int value;
-
-    private Level(int value) {
-      this.value = value;
-    }
-
-    // @@protoc_insertion_point(enum_scope:tensorflow.OptimizerOptions.Level)
-  }
-
-  /**
-   * <pre>
-   * Control the use of the compiler/jit.  Experimental.
-   * </pre>
-   *
-   * Protobuf enum {@code tensorflow.OptimizerOptions.GlobalJitLevel}
-   */
-  public enum GlobalJitLevel
-      implements com.google.protobuf.ProtocolMessageEnum {
-    /**
-     * <pre>
-     * Default setting ("off" now, but later expected to be "on")
-     * </pre>
-     *
-     * <code>DEFAULT = 0;</code>
-     */
-    DEFAULT(0),
-    /**
-     * <code>OFF = -1;</code>
-     */
-    OFF(-1),
-    /**
-     * <pre>
-     * The following settings turn on compilation, with higher values being
-     * more aggressive.  Higher values may reduce opportunities for parallelism
-     * and may use more memory.  (At present, there is no distinction, but this
-     * is expected to change.)
-     * </pre>
-     *
-     * <code>ON_1 = 1;</code>
-     */
-    ON_1(1),
-    /**
-     * <code>ON_2 = 2;</code>
-     */
-    ON_2(2),
-    UNRECOGNIZED(-1),
-    ;
-
-    /**
-     * <pre>
-     * Default setting ("off" now, but later expected to be "on")
-     * </pre>
-     *
-     * <code>DEFAULT = 0;</code>
-     */
-    public static final int DEFAULT_VALUE = 0;
-    /**
-     * <code>OFF = -1;</code>
-     */
-    public static final int OFF_VALUE = -1;
-    /**
-     * <pre>
-     * The following settings turn on compilation, with higher values being
-     * more aggressive.  Higher values may reduce opportunities for parallelism
-     * and may use more memory.  (At present, there is no distinction, but this
-     * is expected to change.)
-     * </pre>
-     *
-     * <code>ON_1 = 1;</code>
-     */
-    public static final int ON_1_VALUE = 1;
-    /**
-     * <code>ON_2 = 2;</code>
-     */
-    public static final int ON_2_VALUE = 2;
-
-
-    public final int getNumber() {
-      if (this == UNRECOGNIZED) {
-        throw new java.lang.IllegalArgumentException(
-            "Can't get the number of an unknown enum value.");
-      }
-      return value;
-    }
-
-    /**
-     * @deprecated Use {@link #forNumber(int)} instead.
-     */
-    @java.lang.Deprecated
-    public static GlobalJitLevel valueOf(int value) {
-      return forNumber(value);
-    }
-
-    public static GlobalJitLevel forNumber(int value) {
-      switch (value) {
-        case 0: return DEFAULT;
-        case -1: return OFF;
-        case 1: return ON_1;
-        case 2: return ON_2;
-        default: return null;
-      }
-    }
-
-    public static com.google.protobuf.Internal.EnumLiteMap<GlobalJitLevel>
-        internalGetValueMap() {
-      return internalValueMap;
-    }
-    private static final com.google.protobuf.Internal.EnumLiteMap<
-        GlobalJitLevel> internalValueMap =
-          new com.google.protobuf.Internal.EnumLiteMap<GlobalJitLevel>() {
-            public GlobalJitLevel findValueByNumber(int number) {
-              return GlobalJitLevel.forNumber(number);
-            }
-          };
-
-    public final com.google.protobuf.Descriptors.EnumValueDescriptor
-        getValueDescriptor() {
-      return getDescriptor().getValues().get(ordinal());
-    }
-    public final com.google.protobuf.Descriptors.EnumDescriptor
-        getDescriptorForType() {
-      return getDescriptor();
-    }
-    public static final com.google.protobuf.Descriptors.EnumDescriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.OptimizerOptions.getDescriptor().getEnumTypes().get(1);
-    }
-
-    private static final GlobalJitLevel[] VALUES = values();
-
-    public static GlobalJitLevel valueOf(
-        com.google.protobuf.Descriptors.EnumValueDescriptor desc) {
-      if (desc.getType() != getDescriptor()) {
-        throw new java.lang.IllegalArgumentException(
-          "EnumValueDescriptor is not for this type.");
-      }
-      if (desc.getIndex() == -1) {
-        return UNRECOGNIZED;
-      }
-      return VALUES[desc.getIndex()];
-    }
-
-    private final int value;
-
-    private GlobalJitLevel(int value) {
-      this.value = value;
-    }
-
-    // @@protoc_insertion_point(enum_scope:tensorflow.OptimizerOptions.GlobalJitLevel)
-  }
-
-  public static final int DO_COMMON_SUBEXPRESSION_ELIMINATION_FIELD_NUMBER = 1;
-  private boolean doCommonSubexpressionElimination_;
-  /**
-   * <pre>
-   * If true, optimize the graph using common subexpression elimination.
-   * </pre>
-   *
-   * <code>bool do_common_subexpression_elimination = 1;</code>
-   */
-  public boolean getDoCommonSubexpressionElimination() {
-    return doCommonSubexpressionElimination_;
-  }
-
-  public static final int DO_CONSTANT_FOLDING_FIELD_NUMBER = 2;
-  private boolean doConstantFolding_;
-  /**
-   * <pre>
-   * If true, perform constant folding optimization on the graph.
-   * </pre>
-   *
-   * <code>bool do_constant_folding = 2;</code>
-   */
-  public boolean getDoConstantFolding() {
-    return doConstantFolding_;
-  }
-
-  public static final int MAX_FOLDED_CONSTANT_IN_BYTES_FIELD_NUMBER = 6;
-  private long maxFoldedConstantInBytes_;
-  /**
-   * <pre>
-   * Constant folding optimization replaces tensors whose values can be
-   * predetermined, with constant nodes. To avoid inserting too large constants,
-   * the size of each constant created can be limited. If this value is zero, a
-   * default limit of 10 MiB will be applied. If constant folding optimization
-   * is disabled, this value is ignored.
-   * </pre>
-   *
-   * <code>int64 max_folded_constant_in_bytes = 6;</code>
-   */
-  public long getMaxFoldedConstantInBytes() {
-    return maxFoldedConstantInBytes_;
-  }
-
-  public static final int DO_FUNCTION_INLINING_FIELD_NUMBER = 4;
-  private boolean doFunctionInlining_;
-  /**
-   * <pre>
-   * If true, perform function inlining on the graph.
-   * </pre>
-   *
-   * <code>bool do_function_inlining = 4;</code>
-   */
-  public boolean getDoFunctionInlining() {
-    return doFunctionInlining_;
-  }
-
-  public static final int OPT_LEVEL_FIELD_NUMBER = 3;
-  private int optLevel_;
-  /**
-   * <pre>
-   * Overall optimization level. The actual optimizations applied will be the
-   * logical OR of the flags that this level implies and any flags already set.
-   * </pre>
-   *
-   * <code>.tensorflow.OptimizerOptions.Level opt_level = 3;</code>
-   */
-  public int getOptLevelValue() {
-    return optLevel_;
-  }
-  /**
-   * <pre>
-   * Overall optimization level. The actual optimizations applied will be the
-   * logical OR of the flags that this level implies and any flags already set.
-   * </pre>
-   *
-   * <code>.tensorflow.OptimizerOptions.Level opt_level = 3;</code>
-   */
-  public org.tensorflow.proto.framework.OptimizerOptions.Level getOptLevel() {
-    @SuppressWarnings("deprecation")
-    org.tensorflow.proto.framework.OptimizerOptions.Level result = org.tensorflow.proto.framework.OptimizerOptions.Level.valueOf(optLevel_);
-    return result == null ? org.tensorflow.proto.framework.OptimizerOptions.Level.UNRECOGNIZED : result;
-  }
-
-  public static final int GLOBAL_JIT_LEVEL_FIELD_NUMBER = 5;
-  private int globalJitLevel_;
-  /**
-   * <code>.tensorflow.OptimizerOptions.GlobalJitLevel global_jit_level = 5;</code>
-   */
-  public int getGlobalJitLevelValue() {
-    return globalJitLevel_;
-  }
-  /**
-   * <code>.tensorflow.OptimizerOptions.GlobalJitLevel global_jit_level = 5;</code>
-   */
-  public org.tensorflow.proto.framework.OptimizerOptions.GlobalJitLevel getGlobalJitLevel() {
-    @SuppressWarnings("deprecation")
-    org.tensorflow.proto.framework.OptimizerOptions.GlobalJitLevel result = org.tensorflow.proto.framework.OptimizerOptions.GlobalJitLevel.valueOf(globalJitLevel_);
-    return result == null ? org.tensorflow.proto.framework.OptimizerOptions.GlobalJitLevel.UNRECOGNIZED : result;
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    if (doCommonSubexpressionElimination_ != false) {
-      output.writeBool(1, doCommonSubexpressionElimination_);
-    }
-    if (doConstantFolding_ != false) {
-      output.writeBool(2, doConstantFolding_);
-    }
-    if (optLevel_ != org.tensorflow.proto.framework.OptimizerOptions.Level.L1.getNumber()) {
-      output.writeEnum(3, optLevel_);
-    }
-    if (doFunctionInlining_ != false) {
-      output.writeBool(4, doFunctionInlining_);
-    }
-    if (globalJitLevel_ != org.tensorflow.proto.framework.OptimizerOptions.GlobalJitLevel.DEFAULT.getNumber()) {
-      output.writeEnum(5, globalJitLevel_);
-    }
-    if (maxFoldedConstantInBytes_ != 0L) {
-      output.writeInt64(6, maxFoldedConstantInBytes_);
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (doCommonSubexpressionElimination_ != false) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeBoolSize(1, doCommonSubexpressionElimination_);
-    }
-    if (doConstantFolding_ != false) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeBoolSize(2, doConstantFolding_);
-    }
-    if (optLevel_ != org.tensorflow.proto.framework.OptimizerOptions.Level.L1.getNumber()) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeEnumSize(3, optLevel_);
-    }
-    if (doFunctionInlining_ != false) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeBoolSize(4, doFunctionInlining_);
-    }
-    if (globalJitLevel_ != org.tensorflow.proto.framework.OptimizerOptions.GlobalJitLevel.DEFAULT.getNumber()) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeEnumSize(5, globalJitLevel_);
-    }
-    if (maxFoldedConstantInBytes_ != 0L) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt64Size(6, maxFoldedConstantInBytes_);
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.framework.OptimizerOptions)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.framework.OptimizerOptions other = (org.tensorflow.proto.framework.OptimizerOptions) obj;
-
-    if (getDoCommonSubexpressionElimination()
-        != other.getDoCommonSubexpressionElimination()) return false;
-    if (getDoConstantFolding()
-        != other.getDoConstantFolding()) return false;
-    if (getMaxFoldedConstantInBytes()
-        != other.getMaxFoldedConstantInBytes()) return false;
-    if (getDoFunctionInlining()
-        != other.getDoFunctionInlining()) return false;
-    if (optLevel_ != other.optLevel_) return false;
-    if (globalJitLevel_ != other.globalJitLevel_) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    hash = (37 * hash) + DO_COMMON_SUBEXPRESSION_ELIMINATION_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashBoolean(
-        getDoCommonSubexpressionElimination());
-    hash = (37 * hash) + DO_CONSTANT_FOLDING_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashBoolean(
-        getDoConstantFolding());
-    hash = (37 * hash) + MAX_FOLDED_CONSTANT_IN_BYTES_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-        getMaxFoldedConstantInBytes());
-    hash = (37 * hash) + DO_FUNCTION_INLINING_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashBoolean(
-        getDoFunctionInlining());
-    hash = (37 * hash) + OPT_LEVEL_FIELD_NUMBER;
-    hash = (53 * hash) + optLevel_;
-    hash = (37 * hash) + GLOBAL_JIT_LEVEL_FIELD_NUMBER;
-    hash = (53 * hash) + globalJitLevel_;
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.framework.OptimizerOptions parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.OptimizerOptions parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.OptimizerOptions parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.OptimizerOptions parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.OptimizerOptions parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.OptimizerOptions parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.OptimizerOptions parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.OptimizerOptions parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.OptimizerOptions parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.OptimizerOptions parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.OptimizerOptions parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.OptimizerOptions parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.framework.OptimizerOptions prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * <pre>
-   * Options passed to the graph optimizer
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.OptimizerOptions}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.OptimizerOptions)
-      org.tensorflow.proto.framework.OptimizerOptionsOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_OptimizerOptions_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_OptimizerOptions_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.OptimizerOptions.class, org.tensorflow.proto.framework.OptimizerOptions.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.framework.OptimizerOptions.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      doCommonSubexpressionElimination_ = false;
-
-      doConstantFolding_ = false;
-
-      maxFoldedConstantInBytes_ = 0L;
-
-      doFunctionInlining_ = false;
-
-      optLevel_ = 0;
-
-      globalJitLevel_ = 0;
-
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_OptimizerOptions_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.OptimizerOptions getDefaultInstanceForType() {
-      return org.tensorflow.proto.framework.OptimizerOptions.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.OptimizerOptions build() {
-      org.tensorflow.proto.framework.OptimizerOptions result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.OptimizerOptions buildPartial() {
-      org.tensorflow.proto.framework.OptimizerOptions result = new org.tensorflow.proto.framework.OptimizerOptions(this);
-      result.doCommonSubexpressionElimination_ = doCommonSubexpressionElimination_;
-      result.doConstantFolding_ = doConstantFolding_;
-      result.maxFoldedConstantInBytes_ = maxFoldedConstantInBytes_;
-      result.doFunctionInlining_ = doFunctionInlining_;
-      result.optLevel_ = optLevel_;
-      result.globalJitLevel_ = globalJitLevel_;
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.framework.OptimizerOptions) {
-        return mergeFrom((org.tensorflow.proto.framework.OptimizerOptions)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.framework.OptimizerOptions other) {
-      if (other == org.tensorflow.proto.framework.OptimizerOptions.getDefaultInstance()) return this;
-      if (other.getDoCommonSubexpressionElimination() != false) {
-        setDoCommonSubexpressionElimination(other.getDoCommonSubexpressionElimination());
-      }
-      if (other.getDoConstantFolding() != false) {
-        setDoConstantFolding(other.getDoConstantFolding());
-      }
-      if (other.getMaxFoldedConstantInBytes() != 0L) {
-        setMaxFoldedConstantInBytes(other.getMaxFoldedConstantInBytes());
-      }
-      if (other.getDoFunctionInlining() != false) {
-        setDoFunctionInlining(other.getDoFunctionInlining());
-      }
-      if (other.optLevel_ != 0) {
-        setOptLevelValue(other.getOptLevelValue());
-      }
-      if (other.globalJitLevel_ != 0) {
-        setGlobalJitLevelValue(other.getGlobalJitLevelValue());
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.framework.OptimizerOptions parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.framework.OptimizerOptions) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-
-    private boolean doCommonSubexpressionElimination_ ;
-    /**
-     * <pre>
-     * If true, optimize the graph using common subexpression elimination.
-     * </pre>
-     *
-     * <code>bool do_common_subexpression_elimination = 1;</code>
-     */
-    public boolean getDoCommonSubexpressionElimination() {
-      return doCommonSubexpressionElimination_;
-    }
-    /**
-     * <pre>
-     * If true, optimize the graph using common subexpression elimination.
-     * </pre>
-     *
-     * <code>bool do_common_subexpression_elimination = 1;</code>
-     */
-    public Builder setDoCommonSubexpressionElimination(boolean value) {
-      
-      doCommonSubexpressionElimination_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * If true, optimize the graph using common subexpression elimination.
-     * </pre>
-     *
-     * <code>bool do_common_subexpression_elimination = 1;</code>
-     */
-    public Builder clearDoCommonSubexpressionElimination() {
-      
-      doCommonSubexpressionElimination_ = false;
-      onChanged();
-      return this;
-    }
-
-    private boolean doConstantFolding_ ;
-    /**
-     * <pre>
-     * If true, perform constant folding optimization on the graph.
-     * </pre>
-     *
-     * <code>bool do_constant_folding = 2;</code>
-     */
-    public boolean getDoConstantFolding() {
-      return doConstantFolding_;
-    }
-    /**
-     * <pre>
-     * If true, perform constant folding optimization on the graph.
-     * </pre>
-     *
-     * <code>bool do_constant_folding = 2;</code>
-     */
-    public Builder setDoConstantFolding(boolean value) {
-      
-      doConstantFolding_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * If true, perform constant folding optimization on the graph.
-     * </pre>
-     *
-     * <code>bool do_constant_folding = 2;</code>
-     */
-    public Builder clearDoConstantFolding() {
-      
-      doConstantFolding_ = false;
-      onChanged();
-      return this;
-    }
-
-    private long maxFoldedConstantInBytes_ ;
-    /**
-     * <pre>
-     * Constant folding optimization replaces tensors whose values can be
-     * predetermined, with constant nodes. To avoid inserting too large constants,
-     * the size of each constant created can be limited. If this value is zero, a
-     * default limit of 10 MiB will be applied. If constant folding optimization
-     * is disabled, this value is ignored.
-     * </pre>
-     *
-     * <code>int64 max_folded_constant_in_bytes = 6;</code>
-     */
-    public long getMaxFoldedConstantInBytes() {
-      return maxFoldedConstantInBytes_;
-    }
-    /**
-     * <pre>
-     * Constant folding optimization replaces tensors whose values can be
-     * predetermined, with constant nodes. To avoid inserting too large constants,
-     * the size of each constant created can be limited. If this value is zero, a
-     * default limit of 10 MiB will be applied. If constant folding optimization
-     * is disabled, this value is ignored.
-     * </pre>
-     *
-     * <code>int64 max_folded_constant_in_bytes = 6;</code>
-     */
-    public Builder setMaxFoldedConstantInBytes(long value) {
-      
-      maxFoldedConstantInBytes_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Constant folding optimization replaces tensors whose values can be
-     * predetermined, with constant nodes. To avoid inserting too large constants,
-     * the size of each constant created can be limited. If this value is zero, a
-     * default limit of 10 MiB will be applied. If constant folding optimization
-     * is disabled, this value is ignored.
-     * </pre>
-     *
-     * <code>int64 max_folded_constant_in_bytes = 6;</code>
-     */
-    public Builder clearMaxFoldedConstantInBytes() {
-      
-      maxFoldedConstantInBytes_ = 0L;
-      onChanged();
-      return this;
-    }
-
-    private boolean doFunctionInlining_ ;
-    /**
-     * <pre>
-     * If true, perform function inlining on the graph.
-     * </pre>
-     *
-     * <code>bool do_function_inlining = 4;</code>
-     */
-    public boolean getDoFunctionInlining() {
-      return doFunctionInlining_;
-    }
-    /**
-     * <pre>
-     * If true, perform function inlining on the graph.
-     * </pre>
-     *
-     * <code>bool do_function_inlining = 4;</code>
-     */
-    public Builder setDoFunctionInlining(boolean value) {
-      
-      doFunctionInlining_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * If true, perform function inlining on the graph.
-     * </pre>
-     *
-     * <code>bool do_function_inlining = 4;</code>
-     */
-    public Builder clearDoFunctionInlining() {
-      
-      doFunctionInlining_ = false;
-      onChanged();
-      return this;
-    }
-
-    private int optLevel_ = 0;
-    /**
-     * <pre>
-     * Overall optimization level. The actual optimizations applied will be the
-     * logical OR of the flags that this level implies and any flags already set.
-     * </pre>
-     *
-     * <code>.tensorflow.OptimizerOptions.Level opt_level = 3;</code>
-     */
-    public int getOptLevelValue() {
-      return optLevel_;
-    }
-    /**
-     * <pre>
-     * Overall optimization level. The actual optimizations applied will be the
-     * logical OR of the flags that this level implies and any flags already set.
-     * </pre>
-     *
-     * <code>.tensorflow.OptimizerOptions.Level opt_level = 3;</code>
-     */
-    public Builder setOptLevelValue(int value) {
-      optLevel_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Overall optimization level. The actual optimizations applied will be the
-     * logical OR of the flags that this level implies and any flags already set.
-     * </pre>
-     *
-     * <code>.tensorflow.OptimizerOptions.Level opt_level = 3;</code>
-     */
-    public org.tensorflow.proto.framework.OptimizerOptions.Level getOptLevel() {
-      @SuppressWarnings("deprecation")
-      org.tensorflow.proto.framework.OptimizerOptions.Level result = org.tensorflow.proto.framework.OptimizerOptions.Level.valueOf(optLevel_);
-      return result == null ? org.tensorflow.proto.framework.OptimizerOptions.Level.UNRECOGNIZED : result;
-    }
-    /**
-     * <pre>
-     * Overall optimization level. The actual optimizations applied will be the
-     * logical OR of the flags that this level implies and any flags already set.
-     * </pre>
-     *
-     * <code>.tensorflow.OptimizerOptions.Level opt_level = 3;</code>
-     */
-    public Builder setOptLevel(org.tensorflow.proto.framework.OptimizerOptions.Level value) {
-      if (value == null) {
-        throw new NullPointerException();
-      }
-      
-      optLevel_ = value.getNumber();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Overall optimization level. The actual optimizations applied will be the
-     * logical OR of the flags that this level implies and any flags already set.
-     * </pre>
-     *
-     * <code>.tensorflow.OptimizerOptions.Level opt_level = 3;</code>
-     */
-    public Builder clearOptLevel() {
-      
-      optLevel_ = 0;
-      onChanged();
-      return this;
-    }
-
-    private int globalJitLevel_ = 0;
-    /**
-     * <code>.tensorflow.OptimizerOptions.GlobalJitLevel global_jit_level = 5;</code>
-     */
-    public int getGlobalJitLevelValue() {
-      return globalJitLevel_;
-    }
-    /**
-     * <code>.tensorflow.OptimizerOptions.GlobalJitLevel global_jit_level = 5;</code>
-     */
-    public Builder setGlobalJitLevelValue(int value) {
-      globalJitLevel_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>.tensorflow.OptimizerOptions.GlobalJitLevel global_jit_level = 5;</code>
-     */
-    public org.tensorflow.proto.framework.OptimizerOptions.GlobalJitLevel getGlobalJitLevel() {
-      @SuppressWarnings("deprecation")
-      org.tensorflow.proto.framework.OptimizerOptions.GlobalJitLevel result = org.tensorflow.proto.framework.OptimizerOptions.GlobalJitLevel.valueOf(globalJitLevel_);
-      return result == null ? org.tensorflow.proto.framework.OptimizerOptions.GlobalJitLevel.UNRECOGNIZED : result;
-    }
-    /**
-     * <code>.tensorflow.OptimizerOptions.GlobalJitLevel global_jit_level = 5;</code>
-     */
-    public Builder setGlobalJitLevel(org.tensorflow.proto.framework.OptimizerOptions.GlobalJitLevel value) {
-      if (value == null) {
-        throw new NullPointerException();
-      }
-      
-      globalJitLevel_ = value.getNumber();
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>.tensorflow.OptimizerOptions.GlobalJitLevel global_jit_level = 5;</code>
-     */
-    public Builder clearGlobalJitLevel() {
-      
-      globalJitLevel_ = 0;
-      onChanged();
-      return this;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.OptimizerOptions)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.OptimizerOptions)
-  private static final org.tensorflow.proto.framework.OptimizerOptions DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.OptimizerOptions();
-  }
-
-  public static org.tensorflow.proto.framework.OptimizerOptions getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<OptimizerOptions>
-      PARSER = new com.google.protobuf.AbstractParser<OptimizerOptions>() {
-    @java.lang.Override
-    public OptimizerOptions parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new OptimizerOptions(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<OptimizerOptions> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<OptimizerOptions> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.framework.OptimizerOptions getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/OptimizerOptionsOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/OptimizerOptionsOrBuilder.java
deleted file mode 100644
index 255ace49e05..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/OptimizerOptionsOrBuilder.java
+++ /dev/null
@@ -1,77 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/config.proto
-
-package org.tensorflow.proto.framework;
-
-public interface OptimizerOptionsOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.OptimizerOptions)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <pre>
-   * If true, optimize the graph using common subexpression elimination.
-   * </pre>
-   *
-   * <code>bool do_common_subexpression_elimination = 1;</code>
-   */
-  boolean getDoCommonSubexpressionElimination();
-
-  /**
-   * <pre>
-   * If true, perform constant folding optimization on the graph.
-   * </pre>
-   *
-   * <code>bool do_constant_folding = 2;</code>
-   */
-  boolean getDoConstantFolding();
-
-  /**
-   * <pre>
-   * Constant folding optimization replaces tensors whose values can be
-   * predetermined, with constant nodes. To avoid inserting too large constants,
-   * the size of each constant created can be limited. If this value is zero, a
-   * default limit of 10 MiB will be applied. If constant folding optimization
-   * is disabled, this value is ignored.
-   * </pre>
-   *
-   * <code>int64 max_folded_constant_in_bytes = 6;</code>
-   */
-  long getMaxFoldedConstantInBytes();
-
-  /**
-   * <pre>
-   * If true, perform function inlining on the graph.
-   * </pre>
-   *
-   * <code>bool do_function_inlining = 4;</code>
-   */
-  boolean getDoFunctionInlining();
-
-  /**
-   * <pre>
-   * Overall optimization level. The actual optimizations applied will be the
-   * logical OR of the flags that this level implies and any flags already set.
-   * </pre>
-   *
-   * <code>.tensorflow.OptimizerOptions.Level opt_level = 3;</code>
-   */
-  int getOptLevelValue();
-  /**
-   * <pre>
-   * Overall optimization level. The actual optimizations applied will be the
-   * logical OR of the flags that this level implies and any flags already set.
-   * </pre>
-   *
-   * <code>.tensorflow.OptimizerOptions.Level opt_level = 3;</code>
-   */
-  org.tensorflow.proto.framework.OptimizerOptions.Level getOptLevel();
-
-  /**
-   * <code>.tensorflow.OptimizerOptions.GlobalJitLevel global_jit_level = 5;</code>
-   */
-  int getGlobalJitLevelValue();
-  /**
-   * <code>.tensorflow.OptimizerOptions.GlobalJitLevel global_jit_level = 5;</code>
-   */
-  org.tensorflow.proto.framework.OptimizerOptions.GlobalJitLevel getGlobalJitLevel();
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/PairValue.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/PairValue.java
deleted file mode 100644
index f9972971b17..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/PairValue.java
+++ /dev/null
@@ -1,735 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/struct.proto
-
-package org.tensorflow.proto.framework;
-
-/**
- * <pre>
- * Represents a (key, value) pair.
- * </pre>
- *
- * Protobuf type {@code tensorflow.PairValue}
- */
-public  final class PairValue extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.PairValue)
-    PairValueOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use PairValue.newBuilder() to construct.
-  private PairValue(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private PairValue() {
-    key_ = "";
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new PairValue();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private PairValue(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 10: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            key_ = s;
-            break;
-          }
-          case 18: {
-            org.tensorflow.proto.framework.StructuredValue.Builder subBuilder = null;
-            if (value_ != null) {
-              subBuilder = value_.toBuilder();
-            }
-            value_ = input.readMessage(org.tensorflow.proto.framework.StructuredValue.parser(), extensionRegistry);
-            if (subBuilder != null) {
-              subBuilder.mergeFrom(value_);
-              value_ = subBuilder.buildPartial();
-            }
-
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.framework.StructProtos.internal_static_tensorflow_PairValue_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.framework.StructProtos.internal_static_tensorflow_PairValue_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.framework.PairValue.class, org.tensorflow.proto.framework.PairValue.Builder.class);
-  }
-
-  public static final int KEY_FIELD_NUMBER = 1;
-  private volatile java.lang.Object key_;
-  /**
-   * <code>string key = 1;</code>
-   */
-  public java.lang.String getKey() {
-    java.lang.Object ref = key_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      key_ = s;
-      return s;
-    }
-  }
-  /**
-   * <code>string key = 1;</code>
-   */
-  public com.google.protobuf.ByteString
-      getKeyBytes() {
-    java.lang.Object ref = key_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      key_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int VALUE_FIELD_NUMBER = 2;
-  private org.tensorflow.proto.framework.StructuredValue value_;
-  /**
-   * <code>.tensorflow.StructuredValue value = 2;</code>
-   */
-  public boolean hasValue() {
-    return value_ != null;
-  }
-  /**
-   * <code>.tensorflow.StructuredValue value = 2;</code>
-   */
-  public org.tensorflow.proto.framework.StructuredValue getValue() {
-    return value_ == null ? org.tensorflow.proto.framework.StructuredValue.getDefaultInstance() : value_;
-  }
-  /**
-   * <code>.tensorflow.StructuredValue value = 2;</code>
-   */
-  public org.tensorflow.proto.framework.StructuredValueOrBuilder getValueOrBuilder() {
-    return getValue();
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    if (!getKeyBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 1, key_);
-    }
-    if (value_ != null) {
-      output.writeMessage(2, getValue());
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (!getKeyBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, key_);
-    }
-    if (value_ != null) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(2, getValue());
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.framework.PairValue)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.framework.PairValue other = (org.tensorflow.proto.framework.PairValue) obj;
-
-    if (!getKey()
-        .equals(other.getKey())) return false;
-    if (hasValue() != other.hasValue()) return false;
-    if (hasValue()) {
-      if (!getValue()
-          .equals(other.getValue())) return false;
-    }
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    hash = (37 * hash) + KEY_FIELD_NUMBER;
-    hash = (53 * hash) + getKey().hashCode();
-    if (hasValue()) {
-      hash = (37 * hash) + VALUE_FIELD_NUMBER;
-      hash = (53 * hash) + getValue().hashCode();
-    }
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.framework.PairValue parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.PairValue parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.PairValue parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.PairValue parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.PairValue parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.PairValue parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.PairValue parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.PairValue parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.PairValue parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.PairValue parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.PairValue parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.PairValue parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.framework.PairValue prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * <pre>
-   * Represents a (key, value) pair.
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.PairValue}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.PairValue)
-      org.tensorflow.proto.framework.PairValueOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.StructProtos.internal_static_tensorflow_PairValue_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.StructProtos.internal_static_tensorflow_PairValue_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.PairValue.class, org.tensorflow.proto.framework.PairValue.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.framework.PairValue.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      key_ = "";
-
-      if (valueBuilder_ == null) {
-        value_ = null;
-      } else {
-        value_ = null;
-        valueBuilder_ = null;
-      }
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.framework.StructProtos.internal_static_tensorflow_PairValue_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.PairValue getDefaultInstanceForType() {
-      return org.tensorflow.proto.framework.PairValue.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.PairValue build() {
-      org.tensorflow.proto.framework.PairValue result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.PairValue buildPartial() {
-      org.tensorflow.proto.framework.PairValue result = new org.tensorflow.proto.framework.PairValue(this);
-      result.key_ = key_;
-      if (valueBuilder_ == null) {
-        result.value_ = value_;
-      } else {
-        result.value_ = valueBuilder_.build();
-      }
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.framework.PairValue) {
-        return mergeFrom((org.tensorflow.proto.framework.PairValue)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.framework.PairValue other) {
-      if (other == org.tensorflow.proto.framework.PairValue.getDefaultInstance()) return this;
-      if (!other.getKey().isEmpty()) {
-        key_ = other.key_;
-        onChanged();
-      }
-      if (other.hasValue()) {
-        mergeValue(other.getValue());
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.framework.PairValue parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.framework.PairValue) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-
-    private java.lang.Object key_ = "";
-    /**
-     * <code>string key = 1;</code>
-     */
-    public java.lang.String getKey() {
-      java.lang.Object ref = key_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        key_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <code>string key = 1;</code>
-     */
-    public com.google.protobuf.ByteString
-        getKeyBytes() {
-      java.lang.Object ref = key_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        key_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <code>string key = 1;</code>
-     */
-    public Builder setKey(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      key_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>string key = 1;</code>
-     */
-    public Builder clearKey() {
-      
-      key_ = getDefaultInstance().getKey();
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>string key = 1;</code>
-     */
-    public Builder setKeyBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      key_ = value;
-      onChanged();
-      return this;
-    }
-
-    private org.tensorflow.proto.framework.StructuredValue value_;
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.StructuredValue, org.tensorflow.proto.framework.StructuredValue.Builder, org.tensorflow.proto.framework.StructuredValueOrBuilder> valueBuilder_;
-    /**
-     * <code>.tensorflow.StructuredValue value = 2;</code>
-     */
-    public boolean hasValue() {
-      return valueBuilder_ != null || value_ != null;
-    }
-    /**
-     * <code>.tensorflow.StructuredValue value = 2;</code>
-     */
-    public org.tensorflow.proto.framework.StructuredValue getValue() {
-      if (valueBuilder_ == null) {
-        return value_ == null ? org.tensorflow.proto.framework.StructuredValue.getDefaultInstance() : value_;
-      } else {
-        return valueBuilder_.getMessage();
-      }
-    }
-    /**
-     * <code>.tensorflow.StructuredValue value = 2;</code>
-     */
-    public Builder setValue(org.tensorflow.proto.framework.StructuredValue value) {
-      if (valueBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        value_ = value;
-        onChanged();
-      } else {
-        valueBuilder_.setMessage(value);
-      }
-
-      return this;
-    }
-    /**
-     * <code>.tensorflow.StructuredValue value = 2;</code>
-     */
-    public Builder setValue(
-        org.tensorflow.proto.framework.StructuredValue.Builder builderForValue) {
-      if (valueBuilder_ == null) {
-        value_ = builderForValue.build();
-        onChanged();
-      } else {
-        valueBuilder_.setMessage(builderForValue.build());
-      }
-
-      return this;
-    }
-    /**
-     * <code>.tensorflow.StructuredValue value = 2;</code>
-     */
-    public Builder mergeValue(org.tensorflow.proto.framework.StructuredValue value) {
-      if (valueBuilder_ == null) {
-        if (value_ != null) {
-          value_ =
-            org.tensorflow.proto.framework.StructuredValue.newBuilder(value_).mergeFrom(value).buildPartial();
-        } else {
-          value_ = value;
-        }
-        onChanged();
-      } else {
-        valueBuilder_.mergeFrom(value);
-      }
-
-      return this;
-    }
-    /**
-     * <code>.tensorflow.StructuredValue value = 2;</code>
-     */
-    public Builder clearValue() {
-      if (valueBuilder_ == null) {
-        value_ = null;
-        onChanged();
-      } else {
-        value_ = null;
-        valueBuilder_ = null;
-      }
-
-      return this;
-    }
-    /**
-     * <code>.tensorflow.StructuredValue value = 2;</code>
-     */
-    public org.tensorflow.proto.framework.StructuredValue.Builder getValueBuilder() {
-      
-      onChanged();
-      return getValueFieldBuilder().getBuilder();
-    }
-    /**
-     * <code>.tensorflow.StructuredValue value = 2;</code>
-     */
-    public org.tensorflow.proto.framework.StructuredValueOrBuilder getValueOrBuilder() {
-      if (valueBuilder_ != null) {
-        return valueBuilder_.getMessageOrBuilder();
-      } else {
-        return value_ == null ?
-            org.tensorflow.proto.framework.StructuredValue.getDefaultInstance() : value_;
-      }
-    }
-    /**
-     * <code>.tensorflow.StructuredValue value = 2;</code>
-     */
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.StructuredValue, org.tensorflow.proto.framework.StructuredValue.Builder, org.tensorflow.proto.framework.StructuredValueOrBuilder> 
-        getValueFieldBuilder() {
-      if (valueBuilder_ == null) {
-        valueBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-            org.tensorflow.proto.framework.StructuredValue, org.tensorflow.proto.framework.StructuredValue.Builder, org.tensorflow.proto.framework.StructuredValueOrBuilder>(
-                getValue(),
-                getParentForChildren(),
-                isClean());
-        value_ = null;
-      }
-      return valueBuilder_;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.PairValue)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.PairValue)
-  private static final org.tensorflow.proto.framework.PairValue DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.PairValue();
-  }
-
-  public static org.tensorflow.proto.framework.PairValue getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<PairValue>
-      PARSER = new com.google.protobuf.AbstractParser<PairValue>() {
-    @java.lang.Override
-    public PairValue parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new PairValue(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<PairValue> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<PairValue> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.framework.PairValue getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/PairValueOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/PairValueOrBuilder.java
deleted file mode 100644
index 0e35d82c1af..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/PairValueOrBuilder.java
+++ /dev/null
@@ -1,32 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/struct.proto
-
-package org.tensorflow.proto.framework;
-
-public interface PairValueOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.PairValue)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <code>string key = 1;</code>
-   */
-  java.lang.String getKey();
-  /**
-   * <code>string key = 1;</code>
-   */
-  com.google.protobuf.ByteString
-      getKeyBytes();
-
-  /**
-   * <code>.tensorflow.StructuredValue value = 2;</code>
-   */
-  boolean hasValue();
-  /**
-   * <code>.tensorflow.StructuredValue value = 2;</code>
-   */
-  org.tensorflow.proto.framework.StructuredValue getValue();
-  /**
-   * <code>.tensorflow.StructuredValue value = 2;</code>
-   */
-  org.tensorflow.proto.framework.StructuredValueOrBuilder getValueOrBuilder();
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/QueueRunnerDef.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/QueueRunnerDef.java
deleted file mode 100644
index 8ca69c8f7f2..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/QueueRunnerDef.java
+++ /dev/null
@@ -1,1435 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/queue_runner.proto
-
-package org.tensorflow.proto.framework;
-
-/**
- * <pre>
- * Protocol buffer representing a QueueRunner.
- * </pre>
- *
- * Protobuf type {@code tensorflow.QueueRunnerDef}
- */
-public  final class QueueRunnerDef extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.QueueRunnerDef)
-    QueueRunnerDefOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use QueueRunnerDef.newBuilder() to construct.
-  private QueueRunnerDef(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private QueueRunnerDef() {
-    queueName_ = "";
-    enqueueOpName_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-    closeOpName_ = "";
-    cancelOpName_ = "";
-    queueClosedExceptionTypes_ = java.util.Collections.emptyList();
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new QueueRunnerDef();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private QueueRunnerDef(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    int mutable_bitField0_ = 0;
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 10: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            queueName_ = s;
-            break;
-          }
-          case 18: {
-            java.lang.String s = input.readStringRequireUtf8();
-            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
-              enqueueOpName_ = new com.google.protobuf.LazyStringArrayList();
-              mutable_bitField0_ |= 0x00000001;
-            }
-            enqueueOpName_.add(s);
-            break;
-          }
-          case 26: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            closeOpName_ = s;
-            break;
-          }
-          case 34: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            cancelOpName_ = s;
-            break;
-          }
-          case 40: {
-            int rawValue = input.readEnum();
-            if (!((mutable_bitField0_ & 0x00000002) != 0)) {
-              queueClosedExceptionTypes_ = new java.util.ArrayList<java.lang.Integer>();
-              mutable_bitField0_ |= 0x00000002;
-            }
-            queueClosedExceptionTypes_.add(rawValue);
-            break;
-          }
-          case 42: {
-            int length = input.readRawVarint32();
-            int oldLimit = input.pushLimit(length);
-            while(input.getBytesUntilLimit() > 0) {
-              int rawValue = input.readEnum();
-              if (!((mutable_bitField0_ & 0x00000002) != 0)) {
-                queueClosedExceptionTypes_ = new java.util.ArrayList<java.lang.Integer>();
-                mutable_bitField0_ |= 0x00000002;
-              }
-              queueClosedExceptionTypes_.add(rawValue);
-            }
-            input.popLimit(oldLimit);
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      if (((mutable_bitField0_ & 0x00000001) != 0)) {
-        enqueueOpName_ = enqueueOpName_.getUnmodifiableView();
-      }
-      if (((mutable_bitField0_ & 0x00000002) != 0)) {
-        queueClosedExceptionTypes_ = java.util.Collections.unmodifiableList(queueClosedExceptionTypes_);
-      }
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.framework.QueueRunnerProtos.internal_static_tensorflow_QueueRunnerDef_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.framework.QueueRunnerProtos.internal_static_tensorflow_QueueRunnerDef_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.framework.QueueRunnerDef.class, org.tensorflow.proto.framework.QueueRunnerDef.Builder.class);
-  }
-
-  public static final int QUEUE_NAME_FIELD_NUMBER = 1;
-  private volatile java.lang.Object queueName_;
-  /**
-   * <pre>
-   * Queue name.
-   * </pre>
-   *
-   * <code>string queue_name = 1;</code>
-   */
-  public java.lang.String getQueueName() {
-    java.lang.Object ref = queueName_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      queueName_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * Queue name.
-   * </pre>
-   *
-   * <code>string queue_name = 1;</code>
-   */
-  public com.google.protobuf.ByteString
-      getQueueNameBytes() {
-    java.lang.Object ref = queueName_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      queueName_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int ENQUEUE_OP_NAME_FIELD_NUMBER = 2;
-  private com.google.protobuf.LazyStringList enqueueOpName_;
-  /**
-   * <pre>
-   * A list of enqueue operations.
-   * </pre>
-   *
-   * <code>repeated string enqueue_op_name = 2;</code>
-   */
-  public com.google.protobuf.ProtocolStringList
-      getEnqueueOpNameList() {
-    return enqueueOpName_;
-  }
-  /**
-   * <pre>
-   * A list of enqueue operations.
-   * </pre>
-   *
-   * <code>repeated string enqueue_op_name = 2;</code>
-   */
-  public int getEnqueueOpNameCount() {
-    return enqueueOpName_.size();
-  }
-  /**
-   * <pre>
-   * A list of enqueue operations.
-   * </pre>
-   *
-   * <code>repeated string enqueue_op_name = 2;</code>
-   */
-  public java.lang.String getEnqueueOpName(int index) {
-    return enqueueOpName_.get(index);
-  }
-  /**
-   * <pre>
-   * A list of enqueue operations.
-   * </pre>
-   *
-   * <code>repeated string enqueue_op_name = 2;</code>
-   */
-  public com.google.protobuf.ByteString
-      getEnqueueOpNameBytes(int index) {
-    return enqueueOpName_.getByteString(index);
-  }
-
-  public static final int CLOSE_OP_NAME_FIELD_NUMBER = 3;
-  private volatile java.lang.Object closeOpName_;
-  /**
-   * <pre>
-   * The operation to run to close the queue.
-   * </pre>
-   *
-   * <code>string close_op_name = 3;</code>
-   */
-  public java.lang.String getCloseOpName() {
-    java.lang.Object ref = closeOpName_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      closeOpName_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * The operation to run to close the queue.
-   * </pre>
-   *
-   * <code>string close_op_name = 3;</code>
-   */
-  public com.google.protobuf.ByteString
-      getCloseOpNameBytes() {
-    java.lang.Object ref = closeOpName_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      closeOpName_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int CANCEL_OP_NAME_FIELD_NUMBER = 4;
-  private volatile java.lang.Object cancelOpName_;
-  /**
-   * <pre>
-   * The operation to run to cancel the queue.
-   * </pre>
-   *
-   * <code>string cancel_op_name = 4;</code>
-   */
-  public java.lang.String getCancelOpName() {
-    java.lang.Object ref = cancelOpName_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      cancelOpName_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * The operation to run to cancel the queue.
-   * </pre>
-   *
-   * <code>string cancel_op_name = 4;</code>
-   */
-  public com.google.protobuf.ByteString
-      getCancelOpNameBytes() {
-    java.lang.Object ref = cancelOpName_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      cancelOpName_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int QUEUE_CLOSED_EXCEPTION_TYPES_FIELD_NUMBER = 5;
-  private java.util.List<java.lang.Integer> queueClosedExceptionTypes_;
-  private static final com.google.protobuf.Internal.ListAdapter.Converter<
-      java.lang.Integer, org.tensorflow.proto.framework.Code> queueClosedExceptionTypes_converter_ =
-          new com.google.protobuf.Internal.ListAdapter.Converter<
-              java.lang.Integer, org.tensorflow.proto.framework.Code>() {
-            public org.tensorflow.proto.framework.Code convert(java.lang.Integer from) {
-              @SuppressWarnings("deprecation")
-              org.tensorflow.proto.framework.Code result = org.tensorflow.proto.framework.Code.valueOf(from);
-              return result == null ? org.tensorflow.proto.framework.Code.UNRECOGNIZED : result;
-            }
-          };
-  /**
-   * <pre>
-   * A list of exception types considered to signal a safely closed queue
-   * if raised during enqueue operations.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.error.Code queue_closed_exception_types = 5;</code>
-   */
-  public java.util.List<org.tensorflow.proto.framework.Code> getQueueClosedExceptionTypesList() {
-    return new com.google.protobuf.Internal.ListAdapter<
-        java.lang.Integer, org.tensorflow.proto.framework.Code>(queueClosedExceptionTypes_, queueClosedExceptionTypes_converter_);
-  }
-  /**
-   * <pre>
-   * A list of exception types considered to signal a safely closed queue
-   * if raised during enqueue operations.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.error.Code queue_closed_exception_types = 5;</code>
-   */
-  public int getQueueClosedExceptionTypesCount() {
-    return queueClosedExceptionTypes_.size();
-  }
-  /**
-   * <pre>
-   * A list of exception types considered to signal a safely closed queue
-   * if raised during enqueue operations.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.error.Code queue_closed_exception_types = 5;</code>
-   */
-  public org.tensorflow.proto.framework.Code getQueueClosedExceptionTypes(int index) {
-    return queueClosedExceptionTypes_converter_.convert(queueClosedExceptionTypes_.get(index));
-  }
-  /**
-   * <pre>
-   * A list of exception types considered to signal a safely closed queue
-   * if raised during enqueue operations.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.error.Code queue_closed_exception_types = 5;</code>
-   */
-  public java.util.List<java.lang.Integer>
-  getQueueClosedExceptionTypesValueList() {
-    return queueClosedExceptionTypes_;
-  }
-  /**
-   * <pre>
-   * A list of exception types considered to signal a safely closed queue
-   * if raised during enqueue operations.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.error.Code queue_closed_exception_types = 5;</code>
-   */
-  public int getQueueClosedExceptionTypesValue(int index) {
-    return queueClosedExceptionTypes_.get(index);
-  }
-  private int queueClosedExceptionTypesMemoizedSerializedSize;
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    getSerializedSize();
-    if (!getQueueNameBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 1, queueName_);
-    }
-    for (int i = 0; i < enqueueOpName_.size(); i++) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 2, enqueueOpName_.getRaw(i));
-    }
-    if (!getCloseOpNameBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 3, closeOpName_);
-    }
-    if (!getCancelOpNameBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 4, cancelOpName_);
-    }
-    if (getQueueClosedExceptionTypesList().size() > 0) {
-      output.writeUInt32NoTag(42);
-      output.writeUInt32NoTag(queueClosedExceptionTypesMemoizedSerializedSize);
-    }
-    for (int i = 0; i < queueClosedExceptionTypes_.size(); i++) {
-      output.writeEnumNoTag(queueClosedExceptionTypes_.get(i));
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (!getQueueNameBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, queueName_);
-    }
-    {
-      int dataSize = 0;
-      for (int i = 0; i < enqueueOpName_.size(); i++) {
-        dataSize += computeStringSizeNoTag(enqueueOpName_.getRaw(i));
-      }
-      size += dataSize;
-      size += 1 * getEnqueueOpNameList().size();
-    }
-    if (!getCloseOpNameBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(3, closeOpName_);
-    }
-    if (!getCancelOpNameBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(4, cancelOpName_);
-    }
-    {
-      int dataSize = 0;
-      for (int i = 0; i < queueClosedExceptionTypes_.size(); i++) {
-        dataSize += com.google.protobuf.CodedOutputStream
-          .computeEnumSizeNoTag(queueClosedExceptionTypes_.get(i));
-      }
-      size += dataSize;
-      if (!getQueueClosedExceptionTypesList().isEmpty()) {  size += 1;
-        size += com.google.protobuf.CodedOutputStream
-          .computeUInt32SizeNoTag(dataSize);
-      }queueClosedExceptionTypesMemoizedSerializedSize = dataSize;
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.framework.QueueRunnerDef)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.framework.QueueRunnerDef other = (org.tensorflow.proto.framework.QueueRunnerDef) obj;
-
-    if (!getQueueName()
-        .equals(other.getQueueName())) return false;
-    if (!getEnqueueOpNameList()
-        .equals(other.getEnqueueOpNameList())) return false;
-    if (!getCloseOpName()
-        .equals(other.getCloseOpName())) return false;
-    if (!getCancelOpName()
-        .equals(other.getCancelOpName())) return false;
-    if (!queueClosedExceptionTypes_.equals(other.queueClosedExceptionTypes_)) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    hash = (37 * hash) + QUEUE_NAME_FIELD_NUMBER;
-    hash = (53 * hash) + getQueueName().hashCode();
-    if (getEnqueueOpNameCount() > 0) {
-      hash = (37 * hash) + ENQUEUE_OP_NAME_FIELD_NUMBER;
-      hash = (53 * hash) + getEnqueueOpNameList().hashCode();
-    }
-    hash = (37 * hash) + CLOSE_OP_NAME_FIELD_NUMBER;
-    hash = (53 * hash) + getCloseOpName().hashCode();
-    hash = (37 * hash) + CANCEL_OP_NAME_FIELD_NUMBER;
-    hash = (53 * hash) + getCancelOpName().hashCode();
-    if (getQueueClosedExceptionTypesCount() > 0) {
-      hash = (37 * hash) + QUEUE_CLOSED_EXCEPTION_TYPES_FIELD_NUMBER;
-      hash = (53 * hash) + queueClosedExceptionTypes_.hashCode();
-    }
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.framework.QueueRunnerDef parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.QueueRunnerDef parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.QueueRunnerDef parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.QueueRunnerDef parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.QueueRunnerDef parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.QueueRunnerDef parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.QueueRunnerDef parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.QueueRunnerDef parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.QueueRunnerDef parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.QueueRunnerDef parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.QueueRunnerDef parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.QueueRunnerDef parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.framework.QueueRunnerDef prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * <pre>
-   * Protocol buffer representing a QueueRunner.
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.QueueRunnerDef}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.QueueRunnerDef)
-      org.tensorflow.proto.framework.QueueRunnerDefOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.QueueRunnerProtos.internal_static_tensorflow_QueueRunnerDef_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.QueueRunnerProtos.internal_static_tensorflow_QueueRunnerDef_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.QueueRunnerDef.class, org.tensorflow.proto.framework.QueueRunnerDef.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.framework.QueueRunnerDef.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      queueName_ = "";
-
-      enqueueOpName_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-      bitField0_ = (bitField0_ & ~0x00000001);
-      closeOpName_ = "";
-
-      cancelOpName_ = "";
-
-      queueClosedExceptionTypes_ = java.util.Collections.emptyList();
-      bitField0_ = (bitField0_ & ~0x00000002);
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.framework.QueueRunnerProtos.internal_static_tensorflow_QueueRunnerDef_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.QueueRunnerDef getDefaultInstanceForType() {
-      return org.tensorflow.proto.framework.QueueRunnerDef.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.QueueRunnerDef build() {
-      org.tensorflow.proto.framework.QueueRunnerDef result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.QueueRunnerDef buildPartial() {
-      org.tensorflow.proto.framework.QueueRunnerDef result = new org.tensorflow.proto.framework.QueueRunnerDef(this);
-      int from_bitField0_ = bitField0_;
-      result.queueName_ = queueName_;
-      if (((bitField0_ & 0x00000001) != 0)) {
-        enqueueOpName_ = enqueueOpName_.getUnmodifiableView();
-        bitField0_ = (bitField0_ & ~0x00000001);
-      }
-      result.enqueueOpName_ = enqueueOpName_;
-      result.closeOpName_ = closeOpName_;
-      result.cancelOpName_ = cancelOpName_;
-      if (((bitField0_ & 0x00000002) != 0)) {
-        queueClosedExceptionTypes_ = java.util.Collections.unmodifiableList(queueClosedExceptionTypes_);
-        bitField0_ = (bitField0_ & ~0x00000002);
-      }
-      result.queueClosedExceptionTypes_ = queueClosedExceptionTypes_;
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.framework.QueueRunnerDef) {
-        return mergeFrom((org.tensorflow.proto.framework.QueueRunnerDef)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.framework.QueueRunnerDef other) {
-      if (other == org.tensorflow.proto.framework.QueueRunnerDef.getDefaultInstance()) return this;
-      if (!other.getQueueName().isEmpty()) {
-        queueName_ = other.queueName_;
-        onChanged();
-      }
-      if (!other.enqueueOpName_.isEmpty()) {
-        if (enqueueOpName_.isEmpty()) {
-          enqueueOpName_ = other.enqueueOpName_;
-          bitField0_ = (bitField0_ & ~0x00000001);
-        } else {
-          ensureEnqueueOpNameIsMutable();
-          enqueueOpName_.addAll(other.enqueueOpName_);
-        }
-        onChanged();
-      }
-      if (!other.getCloseOpName().isEmpty()) {
-        closeOpName_ = other.closeOpName_;
-        onChanged();
-      }
-      if (!other.getCancelOpName().isEmpty()) {
-        cancelOpName_ = other.cancelOpName_;
-        onChanged();
-      }
-      if (!other.queueClosedExceptionTypes_.isEmpty()) {
-        if (queueClosedExceptionTypes_.isEmpty()) {
-          queueClosedExceptionTypes_ = other.queueClosedExceptionTypes_;
-          bitField0_ = (bitField0_ & ~0x00000002);
-        } else {
-          ensureQueueClosedExceptionTypesIsMutable();
-          queueClosedExceptionTypes_.addAll(other.queueClosedExceptionTypes_);
-        }
-        onChanged();
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.framework.QueueRunnerDef parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.framework.QueueRunnerDef) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-    private int bitField0_;
-
-    private java.lang.Object queueName_ = "";
-    /**
-     * <pre>
-     * Queue name.
-     * </pre>
-     *
-     * <code>string queue_name = 1;</code>
-     */
-    public java.lang.String getQueueName() {
-      java.lang.Object ref = queueName_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        queueName_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Queue name.
-     * </pre>
-     *
-     * <code>string queue_name = 1;</code>
-     */
-    public com.google.protobuf.ByteString
-        getQueueNameBytes() {
-      java.lang.Object ref = queueName_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        queueName_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Queue name.
-     * </pre>
-     *
-     * <code>string queue_name = 1;</code>
-     */
-    public Builder setQueueName(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      queueName_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Queue name.
-     * </pre>
-     *
-     * <code>string queue_name = 1;</code>
-     */
-    public Builder clearQueueName() {
-      
-      queueName_ = getDefaultInstance().getQueueName();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Queue name.
-     * </pre>
-     *
-     * <code>string queue_name = 1;</code>
-     */
-    public Builder setQueueNameBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      queueName_ = value;
-      onChanged();
-      return this;
-    }
-
-    private com.google.protobuf.LazyStringList enqueueOpName_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-    private void ensureEnqueueOpNameIsMutable() {
-      if (!((bitField0_ & 0x00000001) != 0)) {
-        enqueueOpName_ = new com.google.protobuf.LazyStringArrayList(enqueueOpName_);
-        bitField0_ |= 0x00000001;
-       }
-    }
-    /**
-     * <pre>
-     * A list of enqueue operations.
-     * </pre>
-     *
-     * <code>repeated string enqueue_op_name = 2;</code>
-     */
-    public com.google.protobuf.ProtocolStringList
-        getEnqueueOpNameList() {
-      return enqueueOpName_.getUnmodifiableView();
-    }
-    /**
-     * <pre>
-     * A list of enqueue operations.
-     * </pre>
-     *
-     * <code>repeated string enqueue_op_name = 2;</code>
-     */
-    public int getEnqueueOpNameCount() {
-      return enqueueOpName_.size();
-    }
-    /**
-     * <pre>
-     * A list of enqueue operations.
-     * </pre>
-     *
-     * <code>repeated string enqueue_op_name = 2;</code>
-     */
-    public java.lang.String getEnqueueOpName(int index) {
-      return enqueueOpName_.get(index);
-    }
-    /**
-     * <pre>
-     * A list of enqueue operations.
-     * </pre>
-     *
-     * <code>repeated string enqueue_op_name = 2;</code>
-     */
-    public com.google.protobuf.ByteString
-        getEnqueueOpNameBytes(int index) {
-      return enqueueOpName_.getByteString(index);
-    }
-    /**
-     * <pre>
-     * A list of enqueue operations.
-     * </pre>
-     *
-     * <code>repeated string enqueue_op_name = 2;</code>
-     */
-    public Builder setEnqueueOpName(
-        int index, java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  ensureEnqueueOpNameIsMutable();
-      enqueueOpName_.set(index, value);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * A list of enqueue operations.
-     * </pre>
-     *
-     * <code>repeated string enqueue_op_name = 2;</code>
-     */
-    public Builder addEnqueueOpName(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  ensureEnqueueOpNameIsMutable();
-      enqueueOpName_.add(value);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * A list of enqueue operations.
-     * </pre>
-     *
-     * <code>repeated string enqueue_op_name = 2;</code>
-     */
-    public Builder addAllEnqueueOpName(
-        java.lang.Iterable<java.lang.String> values) {
-      ensureEnqueueOpNameIsMutable();
-      com.google.protobuf.AbstractMessageLite.Builder.addAll(
-          values, enqueueOpName_);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * A list of enqueue operations.
-     * </pre>
-     *
-     * <code>repeated string enqueue_op_name = 2;</code>
-     */
-    public Builder clearEnqueueOpName() {
-      enqueueOpName_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-      bitField0_ = (bitField0_ & ~0x00000001);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * A list of enqueue operations.
-     * </pre>
-     *
-     * <code>repeated string enqueue_op_name = 2;</code>
-     */
-    public Builder addEnqueueOpNameBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      ensureEnqueueOpNameIsMutable();
-      enqueueOpName_.add(value);
-      onChanged();
-      return this;
-    }
-
-    private java.lang.Object closeOpName_ = "";
-    /**
-     * <pre>
-     * The operation to run to close the queue.
-     * </pre>
-     *
-     * <code>string close_op_name = 3;</code>
-     */
-    public java.lang.String getCloseOpName() {
-      java.lang.Object ref = closeOpName_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        closeOpName_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * The operation to run to close the queue.
-     * </pre>
-     *
-     * <code>string close_op_name = 3;</code>
-     */
-    public com.google.protobuf.ByteString
-        getCloseOpNameBytes() {
-      java.lang.Object ref = closeOpName_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        closeOpName_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * The operation to run to close the queue.
-     * </pre>
-     *
-     * <code>string close_op_name = 3;</code>
-     */
-    public Builder setCloseOpName(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      closeOpName_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * The operation to run to close the queue.
-     * </pre>
-     *
-     * <code>string close_op_name = 3;</code>
-     */
-    public Builder clearCloseOpName() {
-      
-      closeOpName_ = getDefaultInstance().getCloseOpName();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * The operation to run to close the queue.
-     * </pre>
-     *
-     * <code>string close_op_name = 3;</code>
-     */
-    public Builder setCloseOpNameBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      closeOpName_ = value;
-      onChanged();
-      return this;
-    }
-
-    private java.lang.Object cancelOpName_ = "";
-    /**
-     * <pre>
-     * The operation to run to cancel the queue.
-     * </pre>
-     *
-     * <code>string cancel_op_name = 4;</code>
-     */
-    public java.lang.String getCancelOpName() {
-      java.lang.Object ref = cancelOpName_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        cancelOpName_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * The operation to run to cancel the queue.
-     * </pre>
-     *
-     * <code>string cancel_op_name = 4;</code>
-     */
-    public com.google.protobuf.ByteString
-        getCancelOpNameBytes() {
-      java.lang.Object ref = cancelOpName_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        cancelOpName_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * The operation to run to cancel the queue.
-     * </pre>
-     *
-     * <code>string cancel_op_name = 4;</code>
-     */
-    public Builder setCancelOpName(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      cancelOpName_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * The operation to run to cancel the queue.
-     * </pre>
-     *
-     * <code>string cancel_op_name = 4;</code>
-     */
-    public Builder clearCancelOpName() {
-      
-      cancelOpName_ = getDefaultInstance().getCancelOpName();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * The operation to run to cancel the queue.
-     * </pre>
-     *
-     * <code>string cancel_op_name = 4;</code>
-     */
-    public Builder setCancelOpNameBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      cancelOpName_ = value;
-      onChanged();
-      return this;
-    }
-
-    private java.util.List<java.lang.Integer> queueClosedExceptionTypes_ =
-      java.util.Collections.emptyList();
-    private void ensureQueueClosedExceptionTypesIsMutable() {
-      if (!((bitField0_ & 0x00000002) != 0)) {
-        queueClosedExceptionTypes_ = new java.util.ArrayList<java.lang.Integer>(queueClosedExceptionTypes_);
-        bitField0_ |= 0x00000002;
-      }
-    }
-    /**
-     * <pre>
-     * A list of exception types considered to signal a safely closed queue
-     * if raised during enqueue operations.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.error.Code queue_closed_exception_types = 5;</code>
-     */
-    public java.util.List<org.tensorflow.proto.framework.Code> getQueueClosedExceptionTypesList() {
-      return new com.google.protobuf.Internal.ListAdapter<
-          java.lang.Integer, org.tensorflow.proto.framework.Code>(queueClosedExceptionTypes_, queueClosedExceptionTypes_converter_);
-    }
-    /**
-     * <pre>
-     * A list of exception types considered to signal a safely closed queue
-     * if raised during enqueue operations.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.error.Code queue_closed_exception_types = 5;</code>
-     */
-    public int getQueueClosedExceptionTypesCount() {
-      return queueClosedExceptionTypes_.size();
-    }
-    /**
-     * <pre>
-     * A list of exception types considered to signal a safely closed queue
-     * if raised during enqueue operations.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.error.Code queue_closed_exception_types = 5;</code>
-     */
-    public org.tensorflow.proto.framework.Code getQueueClosedExceptionTypes(int index) {
-      return queueClosedExceptionTypes_converter_.convert(queueClosedExceptionTypes_.get(index));
-    }
-    /**
-     * <pre>
-     * A list of exception types considered to signal a safely closed queue
-     * if raised during enqueue operations.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.error.Code queue_closed_exception_types = 5;</code>
-     */
-    public Builder setQueueClosedExceptionTypes(
-        int index, org.tensorflow.proto.framework.Code value) {
-      if (value == null) {
-        throw new NullPointerException();
-      }
-      ensureQueueClosedExceptionTypesIsMutable();
-      queueClosedExceptionTypes_.set(index, value.getNumber());
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * A list of exception types considered to signal a safely closed queue
-     * if raised during enqueue operations.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.error.Code queue_closed_exception_types = 5;</code>
-     */
-    public Builder addQueueClosedExceptionTypes(org.tensorflow.proto.framework.Code value) {
-      if (value == null) {
-        throw new NullPointerException();
-      }
-      ensureQueueClosedExceptionTypesIsMutable();
-      queueClosedExceptionTypes_.add(value.getNumber());
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * A list of exception types considered to signal a safely closed queue
-     * if raised during enqueue operations.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.error.Code queue_closed_exception_types = 5;</code>
-     */
-    public Builder addAllQueueClosedExceptionTypes(
-        java.lang.Iterable<? extends org.tensorflow.proto.framework.Code> values) {
-      ensureQueueClosedExceptionTypesIsMutable();
-      for (org.tensorflow.proto.framework.Code value : values) {
-        queueClosedExceptionTypes_.add(value.getNumber());
-      }
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * A list of exception types considered to signal a safely closed queue
-     * if raised during enqueue operations.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.error.Code queue_closed_exception_types = 5;</code>
-     */
-    public Builder clearQueueClosedExceptionTypes() {
-      queueClosedExceptionTypes_ = java.util.Collections.emptyList();
-      bitField0_ = (bitField0_ & ~0x00000002);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * A list of exception types considered to signal a safely closed queue
-     * if raised during enqueue operations.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.error.Code queue_closed_exception_types = 5;</code>
-     */
-    public java.util.List<java.lang.Integer>
-    getQueueClosedExceptionTypesValueList() {
-      return java.util.Collections.unmodifiableList(queueClosedExceptionTypes_);
-    }
-    /**
-     * <pre>
-     * A list of exception types considered to signal a safely closed queue
-     * if raised during enqueue operations.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.error.Code queue_closed_exception_types = 5;</code>
-     */
-    public int getQueueClosedExceptionTypesValue(int index) {
-      return queueClosedExceptionTypes_.get(index);
-    }
-    /**
-     * <pre>
-     * A list of exception types considered to signal a safely closed queue
-     * if raised during enqueue operations.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.error.Code queue_closed_exception_types = 5;</code>
-     */
-    public Builder setQueueClosedExceptionTypesValue(
-        int index, int value) {
-      ensureQueueClosedExceptionTypesIsMutable();
-      queueClosedExceptionTypes_.set(index, value);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * A list of exception types considered to signal a safely closed queue
-     * if raised during enqueue operations.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.error.Code queue_closed_exception_types = 5;</code>
-     */
-    public Builder addQueueClosedExceptionTypesValue(int value) {
-      ensureQueueClosedExceptionTypesIsMutable();
-      queueClosedExceptionTypes_.add(value);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * A list of exception types considered to signal a safely closed queue
-     * if raised during enqueue operations.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.error.Code queue_closed_exception_types = 5;</code>
-     */
-    public Builder addAllQueueClosedExceptionTypesValue(
-        java.lang.Iterable<java.lang.Integer> values) {
-      ensureQueueClosedExceptionTypesIsMutable();
-      for (int value : values) {
-        queueClosedExceptionTypes_.add(value);
-      }
-      onChanged();
-      return this;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.QueueRunnerDef)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.QueueRunnerDef)
-  private static final org.tensorflow.proto.framework.QueueRunnerDef DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.QueueRunnerDef();
-  }
-
-  public static org.tensorflow.proto.framework.QueueRunnerDef getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<QueueRunnerDef>
-      PARSER = new com.google.protobuf.AbstractParser<QueueRunnerDef>() {
-    @java.lang.Override
-    public QueueRunnerDef parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new QueueRunnerDef(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<QueueRunnerDef> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<QueueRunnerDef> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.framework.QueueRunnerDef getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/QueueRunnerDefOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/QueueRunnerDefOrBuilder.java
deleted file mode 100644
index 61a20e767a3..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/QueueRunnerDefOrBuilder.java
+++ /dev/null
@@ -1,145 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/queue_runner.proto
-
-package org.tensorflow.proto.framework;
-
-public interface QueueRunnerDefOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.QueueRunnerDef)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <pre>
-   * Queue name.
-   * </pre>
-   *
-   * <code>string queue_name = 1;</code>
-   */
-  java.lang.String getQueueName();
-  /**
-   * <pre>
-   * Queue name.
-   * </pre>
-   *
-   * <code>string queue_name = 1;</code>
-   */
-  com.google.protobuf.ByteString
-      getQueueNameBytes();
-
-  /**
-   * <pre>
-   * A list of enqueue operations.
-   * </pre>
-   *
-   * <code>repeated string enqueue_op_name = 2;</code>
-   */
-  java.util.List<java.lang.String>
-      getEnqueueOpNameList();
-  /**
-   * <pre>
-   * A list of enqueue operations.
-   * </pre>
-   *
-   * <code>repeated string enqueue_op_name = 2;</code>
-   */
-  int getEnqueueOpNameCount();
-  /**
-   * <pre>
-   * A list of enqueue operations.
-   * </pre>
-   *
-   * <code>repeated string enqueue_op_name = 2;</code>
-   */
-  java.lang.String getEnqueueOpName(int index);
-  /**
-   * <pre>
-   * A list of enqueue operations.
-   * </pre>
-   *
-   * <code>repeated string enqueue_op_name = 2;</code>
-   */
-  com.google.protobuf.ByteString
-      getEnqueueOpNameBytes(int index);
-
-  /**
-   * <pre>
-   * The operation to run to close the queue.
-   * </pre>
-   *
-   * <code>string close_op_name = 3;</code>
-   */
-  java.lang.String getCloseOpName();
-  /**
-   * <pre>
-   * The operation to run to close the queue.
-   * </pre>
-   *
-   * <code>string close_op_name = 3;</code>
-   */
-  com.google.protobuf.ByteString
-      getCloseOpNameBytes();
-
-  /**
-   * <pre>
-   * The operation to run to cancel the queue.
-   * </pre>
-   *
-   * <code>string cancel_op_name = 4;</code>
-   */
-  java.lang.String getCancelOpName();
-  /**
-   * <pre>
-   * The operation to run to cancel the queue.
-   * </pre>
-   *
-   * <code>string cancel_op_name = 4;</code>
-   */
-  com.google.protobuf.ByteString
-      getCancelOpNameBytes();
-
-  /**
-   * <pre>
-   * A list of exception types considered to signal a safely closed queue
-   * if raised during enqueue operations.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.error.Code queue_closed_exception_types = 5;</code>
-   */
-  java.util.List<org.tensorflow.proto.framework.Code> getQueueClosedExceptionTypesList();
-  /**
-   * <pre>
-   * A list of exception types considered to signal a safely closed queue
-   * if raised during enqueue operations.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.error.Code queue_closed_exception_types = 5;</code>
-   */
-  int getQueueClosedExceptionTypesCount();
-  /**
-   * <pre>
-   * A list of exception types considered to signal a safely closed queue
-   * if raised during enqueue operations.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.error.Code queue_closed_exception_types = 5;</code>
-   */
-  org.tensorflow.proto.framework.Code getQueueClosedExceptionTypes(int index);
-  /**
-   * <pre>
-   * A list of exception types considered to signal a safely closed queue
-   * if raised during enqueue operations.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.error.Code queue_closed_exception_types = 5;</code>
-   */
-  java.util.List<java.lang.Integer>
-  getQueueClosedExceptionTypesValueList();
-  /**
-   * <pre>
-   * A list of exception types considered to signal a safely closed queue
-   * if raised during enqueue operations.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.error.Code queue_closed_exception_types = 5;</code>
-   */
-  int getQueueClosedExceptionTypesValue(int index);
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/QueueRunnerProtos.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/QueueRunnerProtos.java
deleted file mode 100644
index 37a45c16673..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/QueueRunnerProtos.java
+++ /dev/null
@@ -1,58 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/queue_runner.proto
-
-package org.tensorflow.proto.framework;
-
-public final class QueueRunnerProtos {
-  private QueueRunnerProtos() {}
-  public static void registerAllExtensions(
-      com.google.protobuf.ExtensionRegistryLite registry) {
-  }
-
-  public static void registerAllExtensions(
-      com.google.protobuf.ExtensionRegistry registry) {
-    registerAllExtensions(
-        (com.google.protobuf.ExtensionRegistryLite) registry);
-  }
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_QueueRunnerDef_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_QueueRunnerDef_fieldAccessorTable;
-
-  public static com.google.protobuf.Descriptors.FileDescriptor
-      getDescriptor() {
-    return descriptor;
-  }
-  private static  com.google.protobuf.Descriptors.FileDescriptor
-      descriptor;
-  static {
-    java.lang.String[] descriptorData = {
-      "\n+tensorflow/core/protobuf/queue_runner." +
-      "proto\022\ntensorflow\032*tensorflow/core/proto" +
-      "buf/error_codes.proto\"\252\001\n\016QueueRunnerDef" +
-      "\022\022\n\nqueue_name\030\001 \001(\t\022\027\n\017enqueue_op_name\030" +
-      "\002 \003(\t\022\025\n\rclose_op_name\030\003 \001(\t\022\026\n\016cancel_o" +
-      "p_name\030\004 \001(\t\022<\n\034queue_closed_exception_t" +
-      "ypes\030\005 \003(\0162\026.tensorflow.error.CodeB\202\001\n\036o" +
-      "rg.tensorflow.proto.frameworkB\021QueueRunn" +
-      "erProtosP\001ZHgithub.com/tensorflow/tensor" +
-      "flow/tensorflow/go/core/core_protos_go_p" +
-      "roto\370\001\001b\006proto3"
-    };
-    descriptor = com.google.protobuf.Descriptors.FileDescriptor
-      .internalBuildGeneratedFileFrom(descriptorData,
-        new com.google.protobuf.Descriptors.FileDescriptor[] {
-          org.tensorflow.proto.framework.ErrorCodesProtos.getDescriptor(),
-        });
-    internal_static_tensorflow_QueueRunnerDef_descriptor =
-      getDescriptor().getMessageTypes().get(0);
-    internal_static_tensorflow_QueueRunnerDef_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_QueueRunnerDef_descriptor,
-        new java.lang.String[] { "QueueName", "EnqueueOpName", "CloseOpName", "CancelOpName", "QueueClosedExceptionTypes", });
-    org.tensorflow.proto.framework.ErrorCodesProtos.getDescriptor();
-  }
-
-  // @@protoc_insertion_point(outer_class_scope)
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/RPCOptions.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/RPCOptions.java
deleted file mode 100644
index 6c1a717876d..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/RPCOptions.java
+++ /dev/null
@@ -1,905 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/config.proto
-
-package org.tensorflow.proto.framework;
-
-/**
- * Protobuf type {@code tensorflow.RPCOptions}
- */
-public  final class RPCOptions extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.RPCOptions)
-    RPCOptionsOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use RPCOptions.newBuilder() to construct.
-  private RPCOptions(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private RPCOptions() {
-    compressionAlgorithm_ = "";
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new RPCOptions();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private RPCOptions(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 8: {
-
-            useRpcForInprocessMaster_ = input.readBool();
-            break;
-          }
-          case 18: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            compressionAlgorithm_ = s;
-            break;
-          }
-          case 24: {
-
-            compressionLevel_ = input.readInt32();
-            break;
-          }
-          case 32: {
-
-            cacheRpcResponse_ = input.readBool();
-            break;
-          }
-          case 40: {
-
-            disableSessionConnectionSharing_ = input.readBool();
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_RPCOptions_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_RPCOptions_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.framework.RPCOptions.class, org.tensorflow.proto.framework.RPCOptions.Builder.class);
-  }
-
-  public static final int USE_RPC_FOR_INPROCESS_MASTER_FIELD_NUMBER = 1;
-  private boolean useRpcForInprocessMaster_;
-  /**
-   * <pre>
-   * If true, always use RPC to contact the session target.
-   * If false (the default option), TensorFlow may use an optimized
-   * transport for client-master communication that avoids the RPC
-   * stack. This option is primarily for used testing the RPC stack.
-   * </pre>
-   *
-   * <code>bool use_rpc_for_inprocess_master = 1;</code>
-   */
-  public boolean getUseRpcForInprocessMaster() {
-    return useRpcForInprocessMaster_;
-  }
-
-  public static final int COMPRESSION_ALGORITHM_FIELD_NUMBER = 2;
-  private volatile java.lang.Object compressionAlgorithm_;
-  /**
-   * <pre>
-   * The compression algorithm to be used. One of "deflate", "gzip".
-   * </pre>
-   *
-   * <code>string compression_algorithm = 2;</code>
-   */
-  public java.lang.String getCompressionAlgorithm() {
-    java.lang.Object ref = compressionAlgorithm_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      compressionAlgorithm_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * The compression algorithm to be used. One of "deflate", "gzip".
-   * </pre>
-   *
-   * <code>string compression_algorithm = 2;</code>
-   */
-  public com.google.protobuf.ByteString
-      getCompressionAlgorithmBytes() {
-    java.lang.Object ref = compressionAlgorithm_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      compressionAlgorithm_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int COMPRESSION_LEVEL_FIELD_NUMBER = 3;
-  private int compressionLevel_;
-  /**
-   * <pre>
-   * If compression_algorithm is set, the compression level to be used.
-   * From 0 (no compression), up to 3.
-   * </pre>
-   *
-   * <code>int32 compression_level = 3;</code>
-   */
-  public int getCompressionLevel() {
-    return compressionLevel_;
-  }
-
-  public static final int CACHE_RPC_RESPONSE_FIELD_NUMBER = 4;
-  private boolean cacheRpcResponse_;
-  /**
-   * <pre>
-   * Setting cache_rpc_response to true will enable sender side caching of
-   * response for RecvTensorAsync and RecvBufAsync to allow receiver to retry
-   * requests . This is only necessary when the network fabric is experiencing a
-   * significant error rate.  Without it we'll fail a step on an network error,
-   * while with it we'll be able to complete long steps (like complex
-   * initializations) in the face of some network errors during RecvTensor.
-   * </pre>
-   *
-   * <code>bool cache_rpc_response = 4;</code>
-   */
-  public boolean getCacheRpcResponse() {
-    return cacheRpcResponse_;
-  }
-
-  public static final int DISABLE_SESSION_CONNECTION_SHARING_FIELD_NUMBER = 5;
-  private boolean disableSessionConnectionSharing_;
-  /**
-   * <pre>
-   * Disables TCP connection sharing when opening a new RPC channel.
-   * </pre>
-   *
-   * <code>bool disable_session_connection_sharing = 5;</code>
-   */
-  public boolean getDisableSessionConnectionSharing() {
-    return disableSessionConnectionSharing_;
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    if (useRpcForInprocessMaster_ != false) {
-      output.writeBool(1, useRpcForInprocessMaster_);
-    }
-    if (!getCompressionAlgorithmBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 2, compressionAlgorithm_);
-    }
-    if (compressionLevel_ != 0) {
-      output.writeInt32(3, compressionLevel_);
-    }
-    if (cacheRpcResponse_ != false) {
-      output.writeBool(4, cacheRpcResponse_);
-    }
-    if (disableSessionConnectionSharing_ != false) {
-      output.writeBool(5, disableSessionConnectionSharing_);
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (useRpcForInprocessMaster_ != false) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeBoolSize(1, useRpcForInprocessMaster_);
-    }
-    if (!getCompressionAlgorithmBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(2, compressionAlgorithm_);
-    }
-    if (compressionLevel_ != 0) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt32Size(3, compressionLevel_);
-    }
-    if (cacheRpcResponse_ != false) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeBoolSize(4, cacheRpcResponse_);
-    }
-    if (disableSessionConnectionSharing_ != false) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeBoolSize(5, disableSessionConnectionSharing_);
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.framework.RPCOptions)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.framework.RPCOptions other = (org.tensorflow.proto.framework.RPCOptions) obj;
-
-    if (getUseRpcForInprocessMaster()
-        != other.getUseRpcForInprocessMaster()) return false;
-    if (!getCompressionAlgorithm()
-        .equals(other.getCompressionAlgorithm())) return false;
-    if (getCompressionLevel()
-        != other.getCompressionLevel()) return false;
-    if (getCacheRpcResponse()
-        != other.getCacheRpcResponse()) return false;
-    if (getDisableSessionConnectionSharing()
-        != other.getDisableSessionConnectionSharing()) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    hash = (37 * hash) + USE_RPC_FOR_INPROCESS_MASTER_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashBoolean(
-        getUseRpcForInprocessMaster());
-    hash = (37 * hash) + COMPRESSION_ALGORITHM_FIELD_NUMBER;
-    hash = (53 * hash) + getCompressionAlgorithm().hashCode();
-    hash = (37 * hash) + COMPRESSION_LEVEL_FIELD_NUMBER;
-    hash = (53 * hash) + getCompressionLevel();
-    hash = (37 * hash) + CACHE_RPC_RESPONSE_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashBoolean(
-        getCacheRpcResponse());
-    hash = (37 * hash) + DISABLE_SESSION_CONNECTION_SHARING_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashBoolean(
-        getDisableSessionConnectionSharing());
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.framework.RPCOptions parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.RPCOptions parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.RPCOptions parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.RPCOptions parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.RPCOptions parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.RPCOptions parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.RPCOptions parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.RPCOptions parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.RPCOptions parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.RPCOptions parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.RPCOptions parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.RPCOptions parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.framework.RPCOptions prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * Protobuf type {@code tensorflow.RPCOptions}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.RPCOptions)
-      org.tensorflow.proto.framework.RPCOptionsOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_RPCOptions_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_RPCOptions_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.RPCOptions.class, org.tensorflow.proto.framework.RPCOptions.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.framework.RPCOptions.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      useRpcForInprocessMaster_ = false;
-
-      compressionAlgorithm_ = "";
-
-      compressionLevel_ = 0;
-
-      cacheRpcResponse_ = false;
-
-      disableSessionConnectionSharing_ = false;
-
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_RPCOptions_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.RPCOptions getDefaultInstanceForType() {
-      return org.tensorflow.proto.framework.RPCOptions.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.RPCOptions build() {
-      org.tensorflow.proto.framework.RPCOptions result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.RPCOptions buildPartial() {
-      org.tensorflow.proto.framework.RPCOptions result = new org.tensorflow.proto.framework.RPCOptions(this);
-      result.useRpcForInprocessMaster_ = useRpcForInprocessMaster_;
-      result.compressionAlgorithm_ = compressionAlgorithm_;
-      result.compressionLevel_ = compressionLevel_;
-      result.cacheRpcResponse_ = cacheRpcResponse_;
-      result.disableSessionConnectionSharing_ = disableSessionConnectionSharing_;
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.framework.RPCOptions) {
-        return mergeFrom((org.tensorflow.proto.framework.RPCOptions)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.framework.RPCOptions other) {
-      if (other == org.tensorflow.proto.framework.RPCOptions.getDefaultInstance()) return this;
-      if (other.getUseRpcForInprocessMaster() != false) {
-        setUseRpcForInprocessMaster(other.getUseRpcForInprocessMaster());
-      }
-      if (!other.getCompressionAlgorithm().isEmpty()) {
-        compressionAlgorithm_ = other.compressionAlgorithm_;
-        onChanged();
-      }
-      if (other.getCompressionLevel() != 0) {
-        setCompressionLevel(other.getCompressionLevel());
-      }
-      if (other.getCacheRpcResponse() != false) {
-        setCacheRpcResponse(other.getCacheRpcResponse());
-      }
-      if (other.getDisableSessionConnectionSharing() != false) {
-        setDisableSessionConnectionSharing(other.getDisableSessionConnectionSharing());
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.framework.RPCOptions parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.framework.RPCOptions) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-
-    private boolean useRpcForInprocessMaster_ ;
-    /**
-     * <pre>
-     * If true, always use RPC to contact the session target.
-     * If false (the default option), TensorFlow may use an optimized
-     * transport for client-master communication that avoids the RPC
-     * stack. This option is primarily for used testing the RPC stack.
-     * </pre>
-     *
-     * <code>bool use_rpc_for_inprocess_master = 1;</code>
-     */
-    public boolean getUseRpcForInprocessMaster() {
-      return useRpcForInprocessMaster_;
-    }
-    /**
-     * <pre>
-     * If true, always use RPC to contact the session target.
-     * If false (the default option), TensorFlow may use an optimized
-     * transport for client-master communication that avoids the RPC
-     * stack. This option is primarily for used testing the RPC stack.
-     * </pre>
-     *
-     * <code>bool use_rpc_for_inprocess_master = 1;</code>
-     */
-    public Builder setUseRpcForInprocessMaster(boolean value) {
-      
-      useRpcForInprocessMaster_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * If true, always use RPC to contact the session target.
-     * If false (the default option), TensorFlow may use an optimized
-     * transport for client-master communication that avoids the RPC
-     * stack. This option is primarily for used testing the RPC stack.
-     * </pre>
-     *
-     * <code>bool use_rpc_for_inprocess_master = 1;</code>
-     */
-    public Builder clearUseRpcForInprocessMaster() {
-      
-      useRpcForInprocessMaster_ = false;
-      onChanged();
-      return this;
-    }
-
-    private java.lang.Object compressionAlgorithm_ = "";
-    /**
-     * <pre>
-     * The compression algorithm to be used. One of "deflate", "gzip".
-     * </pre>
-     *
-     * <code>string compression_algorithm = 2;</code>
-     */
-    public java.lang.String getCompressionAlgorithm() {
-      java.lang.Object ref = compressionAlgorithm_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        compressionAlgorithm_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * The compression algorithm to be used. One of "deflate", "gzip".
-     * </pre>
-     *
-     * <code>string compression_algorithm = 2;</code>
-     */
-    public com.google.protobuf.ByteString
-        getCompressionAlgorithmBytes() {
-      java.lang.Object ref = compressionAlgorithm_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        compressionAlgorithm_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * The compression algorithm to be used. One of "deflate", "gzip".
-     * </pre>
-     *
-     * <code>string compression_algorithm = 2;</code>
-     */
-    public Builder setCompressionAlgorithm(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      compressionAlgorithm_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * The compression algorithm to be used. One of "deflate", "gzip".
-     * </pre>
-     *
-     * <code>string compression_algorithm = 2;</code>
-     */
-    public Builder clearCompressionAlgorithm() {
-      
-      compressionAlgorithm_ = getDefaultInstance().getCompressionAlgorithm();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * The compression algorithm to be used. One of "deflate", "gzip".
-     * </pre>
-     *
-     * <code>string compression_algorithm = 2;</code>
-     */
-    public Builder setCompressionAlgorithmBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      compressionAlgorithm_ = value;
-      onChanged();
-      return this;
-    }
-
-    private int compressionLevel_ ;
-    /**
-     * <pre>
-     * If compression_algorithm is set, the compression level to be used.
-     * From 0 (no compression), up to 3.
-     * </pre>
-     *
-     * <code>int32 compression_level = 3;</code>
-     */
-    public int getCompressionLevel() {
-      return compressionLevel_;
-    }
-    /**
-     * <pre>
-     * If compression_algorithm is set, the compression level to be used.
-     * From 0 (no compression), up to 3.
-     * </pre>
-     *
-     * <code>int32 compression_level = 3;</code>
-     */
-    public Builder setCompressionLevel(int value) {
-      
-      compressionLevel_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * If compression_algorithm is set, the compression level to be used.
-     * From 0 (no compression), up to 3.
-     * </pre>
-     *
-     * <code>int32 compression_level = 3;</code>
-     */
-    public Builder clearCompressionLevel() {
-      
-      compressionLevel_ = 0;
-      onChanged();
-      return this;
-    }
-
-    private boolean cacheRpcResponse_ ;
-    /**
-     * <pre>
-     * Setting cache_rpc_response to true will enable sender side caching of
-     * response for RecvTensorAsync and RecvBufAsync to allow receiver to retry
-     * requests . This is only necessary when the network fabric is experiencing a
-     * significant error rate.  Without it we'll fail a step on an network error,
-     * while with it we'll be able to complete long steps (like complex
-     * initializations) in the face of some network errors during RecvTensor.
-     * </pre>
-     *
-     * <code>bool cache_rpc_response = 4;</code>
-     */
-    public boolean getCacheRpcResponse() {
-      return cacheRpcResponse_;
-    }
-    /**
-     * <pre>
-     * Setting cache_rpc_response to true will enable sender side caching of
-     * response for RecvTensorAsync and RecvBufAsync to allow receiver to retry
-     * requests . This is only necessary when the network fabric is experiencing a
-     * significant error rate.  Without it we'll fail a step on an network error,
-     * while with it we'll be able to complete long steps (like complex
-     * initializations) in the face of some network errors during RecvTensor.
-     * </pre>
-     *
-     * <code>bool cache_rpc_response = 4;</code>
-     */
-    public Builder setCacheRpcResponse(boolean value) {
-      
-      cacheRpcResponse_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Setting cache_rpc_response to true will enable sender side caching of
-     * response for RecvTensorAsync and RecvBufAsync to allow receiver to retry
-     * requests . This is only necessary when the network fabric is experiencing a
-     * significant error rate.  Without it we'll fail a step on an network error,
-     * while with it we'll be able to complete long steps (like complex
-     * initializations) in the face of some network errors during RecvTensor.
-     * </pre>
-     *
-     * <code>bool cache_rpc_response = 4;</code>
-     */
-    public Builder clearCacheRpcResponse() {
-      
-      cacheRpcResponse_ = false;
-      onChanged();
-      return this;
-    }
-
-    private boolean disableSessionConnectionSharing_ ;
-    /**
-     * <pre>
-     * Disables TCP connection sharing when opening a new RPC channel.
-     * </pre>
-     *
-     * <code>bool disable_session_connection_sharing = 5;</code>
-     */
-    public boolean getDisableSessionConnectionSharing() {
-      return disableSessionConnectionSharing_;
-    }
-    /**
-     * <pre>
-     * Disables TCP connection sharing when opening a new RPC channel.
-     * </pre>
-     *
-     * <code>bool disable_session_connection_sharing = 5;</code>
-     */
-    public Builder setDisableSessionConnectionSharing(boolean value) {
-      
-      disableSessionConnectionSharing_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Disables TCP connection sharing when opening a new RPC channel.
-     * </pre>
-     *
-     * <code>bool disable_session_connection_sharing = 5;</code>
-     */
-    public Builder clearDisableSessionConnectionSharing() {
-      
-      disableSessionConnectionSharing_ = false;
-      onChanged();
-      return this;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.RPCOptions)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.RPCOptions)
-  private static final org.tensorflow.proto.framework.RPCOptions DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.RPCOptions();
-  }
-
-  public static org.tensorflow.proto.framework.RPCOptions getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<RPCOptions>
-      PARSER = new com.google.protobuf.AbstractParser<RPCOptions>() {
-    @java.lang.Override
-    public RPCOptions parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new RPCOptions(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<RPCOptions> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<RPCOptions> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.framework.RPCOptions getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/RPCOptionsOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/RPCOptionsOrBuilder.java
deleted file mode 100644
index e7178532c17..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/RPCOptionsOrBuilder.java
+++ /dev/null
@@ -1,72 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/config.proto
-
-package org.tensorflow.proto.framework;
-
-public interface RPCOptionsOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.RPCOptions)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <pre>
-   * If true, always use RPC to contact the session target.
-   * If false (the default option), TensorFlow may use an optimized
-   * transport for client-master communication that avoids the RPC
-   * stack. This option is primarily for used testing the RPC stack.
-   * </pre>
-   *
-   * <code>bool use_rpc_for_inprocess_master = 1;</code>
-   */
-  boolean getUseRpcForInprocessMaster();
-
-  /**
-   * <pre>
-   * The compression algorithm to be used. One of "deflate", "gzip".
-   * </pre>
-   *
-   * <code>string compression_algorithm = 2;</code>
-   */
-  java.lang.String getCompressionAlgorithm();
-  /**
-   * <pre>
-   * The compression algorithm to be used. One of "deflate", "gzip".
-   * </pre>
-   *
-   * <code>string compression_algorithm = 2;</code>
-   */
-  com.google.protobuf.ByteString
-      getCompressionAlgorithmBytes();
-
-  /**
-   * <pre>
-   * If compression_algorithm is set, the compression level to be used.
-   * From 0 (no compression), up to 3.
-   * </pre>
-   *
-   * <code>int32 compression_level = 3;</code>
-   */
-  int getCompressionLevel();
-
-  /**
-   * <pre>
-   * Setting cache_rpc_response to true will enable sender side caching of
-   * response for RecvTensorAsync and RecvBufAsync to allow receiver to retry
-   * requests . This is only necessary when the network fabric is experiencing a
-   * significant error rate.  Without it we'll fail a step on an network error,
-   * while with it we'll be able to complete long steps (like complex
-   * initializations) in the face of some network errors during RecvTensor.
-   * </pre>
-   *
-   * <code>bool cache_rpc_response = 4;</code>
-   */
-  boolean getCacheRpcResponse();
-
-  /**
-   * <pre>
-   * Disables TCP connection sharing when opening a new RPC channel.
-   * </pre>
-   *
-   * <code>bool disable_session_connection_sharing = 5;</code>
-   */
-  boolean getDisableSessionConnectionSharing();
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ReaderBaseProtos.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ReaderBaseProtos.java
deleted file mode 100644
index 3c4e600440b..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ReaderBaseProtos.java
+++ /dev/null
@@ -1,54 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/framework/reader_base.proto
-
-package org.tensorflow.proto.framework;
-
-public final class ReaderBaseProtos {
-  private ReaderBaseProtos() {}
-  public static void registerAllExtensions(
-      com.google.protobuf.ExtensionRegistryLite registry) {
-  }
-
-  public static void registerAllExtensions(
-      com.google.protobuf.ExtensionRegistry registry) {
-    registerAllExtensions(
-        (com.google.protobuf.ExtensionRegistryLite) registry);
-  }
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_ReaderBaseState_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_ReaderBaseState_fieldAccessorTable;
-
-  public static com.google.protobuf.Descriptors.FileDescriptor
-      getDescriptor() {
-    return descriptor;
-  }
-  private static  com.google.protobuf.Descriptors.FileDescriptor
-      descriptor;
-  static {
-    java.lang.String[] descriptorData = {
-      "\n+tensorflow/core/framework/reader_base." +
-      "proto\022\ntensorflow\"r\n\017ReaderBaseState\022\024\n\014" +
-      "work_started\030\001 \001(\003\022\025\n\rwork_finished\030\002 \001(" +
-      "\003\022\034\n\024num_records_produced\030\003 \001(\003\022\024\n\014curre" +
-      "nt_work\030\004 \001(\014B\213\001\n\036org.tensorflow.proto.f" +
-      "rameworkB\020ReaderBaseProtosP\001ZRgithub.com" +
-      "/tensorflow/tensorflow/tensorflow/go/cor" +
-      "e/framework/reader_base_go_proto\370\001\001b\006pro" +
-      "to3"
-    };
-    descriptor = com.google.protobuf.Descriptors.FileDescriptor
-      .internalBuildGeneratedFileFrom(descriptorData,
-        new com.google.protobuf.Descriptors.FileDescriptor[] {
-        });
-    internal_static_tensorflow_ReaderBaseState_descriptor =
-      getDescriptor().getMessageTypes().get(0);
-    internal_static_tensorflow_ReaderBaseState_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_ReaderBaseState_descriptor,
-        new java.lang.String[] { "WorkStarted", "WorkFinished", "NumRecordsProduced", "CurrentWork", });
-  }
-
-  // @@protoc_insertion_point(outer_class_scope)
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ReaderBaseState.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ReaderBaseState.java
deleted file mode 100644
index 4602bd786a5..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ReaderBaseState.java
+++ /dev/null
@@ -1,664 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/framework/reader_base.proto
-
-package org.tensorflow.proto.framework;
-
-/**
- * <pre>
- * For serializing and restoring the state of ReaderBase, see
- * reader_base.h for details.
- * </pre>
- *
- * Protobuf type {@code tensorflow.ReaderBaseState}
- */
-public  final class ReaderBaseState extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.ReaderBaseState)
-    ReaderBaseStateOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use ReaderBaseState.newBuilder() to construct.
-  private ReaderBaseState(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private ReaderBaseState() {
-    currentWork_ = com.google.protobuf.ByteString.EMPTY;
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new ReaderBaseState();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private ReaderBaseState(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 8: {
-
-            workStarted_ = input.readInt64();
-            break;
-          }
-          case 16: {
-
-            workFinished_ = input.readInt64();
-            break;
-          }
-          case 24: {
-
-            numRecordsProduced_ = input.readInt64();
-            break;
-          }
-          case 34: {
-
-            currentWork_ = input.readBytes();
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.framework.ReaderBaseProtos.internal_static_tensorflow_ReaderBaseState_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.framework.ReaderBaseProtos.internal_static_tensorflow_ReaderBaseState_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.framework.ReaderBaseState.class, org.tensorflow.proto.framework.ReaderBaseState.Builder.class);
-  }
-
-  public static final int WORK_STARTED_FIELD_NUMBER = 1;
-  private long workStarted_;
-  /**
-   * <code>int64 work_started = 1;</code>
-   */
-  public long getWorkStarted() {
-    return workStarted_;
-  }
-
-  public static final int WORK_FINISHED_FIELD_NUMBER = 2;
-  private long workFinished_;
-  /**
-   * <code>int64 work_finished = 2;</code>
-   */
-  public long getWorkFinished() {
-    return workFinished_;
-  }
-
-  public static final int NUM_RECORDS_PRODUCED_FIELD_NUMBER = 3;
-  private long numRecordsProduced_;
-  /**
-   * <code>int64 num_records_produced = 3;</code>
-   */
-  public long getNumRecordsProduced() {
-    return numRecordsProduced_;
-  }
-
-  public static final int CURRENT_WORK_FIELD_NUMBER = 4;
-  private com.google.protobuf.ByteString currentWork_;
-  /**
-   * <code>bytes current_work = 4;</code>
-   */
-  public com.google.protobuf.ByteString getCurrentWork() {
-    return currentWork_;
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    if (workStarted_ != 0L) {
-      output.writeInt64(1, workStarted_);
-    }
-    if (workFinished_ != 0L) {
-      output.writeInt64(2, workFinished_);
-    }
-    if (numRecordsProduced_ != 0L) {
-      output.writeInt64(3, numRecordsProduced_);
-    }
-    if (!currentWork_.isEmpty()) {
-      output.writeBytes(4, currentWork_);
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (workStarted_ != 0L) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt64Size(1, workStarted_);
-    }
-    if (workFinished_ != 0L) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt64Size(2, workFinished_);
-    }
-    if (numRecordsProduced_ != 0L) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt64Size(3, numRecordsProduced_);
-    }
-    if (!currentWork_.isEmpty()) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeBytesSize(4, currentWork_);
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.framework.ReaderBaseState)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.framework.ReaderBaseState other = (org.tensorflow.proto.framework.ReaderBaseState) obj;
-
-    if (getWorkStarted()
-        != other.getWorkStarted()) return false;
-    if (getWorkFinished()
-        != other.getWorkFinished()) return false;
-    if (getNumRecordsProduced()
-        != other.getNumRecordsProduced()) return false;
-    if (!getCurrentWork()
-        .equals(other.getCurrentWork())) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    hash = (37 * hash) + WORK_STARTED_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-        getWorkStarted());
-    hash = (37 * hash) + WORK_FINISHED_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-        getWorkFinished());
-    hash = (37 * hash) + NUM_RECORDS_PRODUCED_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-        getNumRecordsProduced());
-    hash = (37 * hash) + CURRENT_WORK_FIELD_NUMBER;
-    hash = (53 * hash) + getCurrentWork().hashCode();
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.framework.ReaderBaseState parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.ReaderBaseState parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.ReaderBaseState parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.ReaderBaseState parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.ReaderBaseState parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.ReaderBaseState parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.ReaderBaseState parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.ReaderBaseState parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.ReaderBaseState parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.ReaderBaseState parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.ReaderBaseState parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.ReaderBaseState parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.framework.ReaderBaseState prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * <pre>
-   * For serializing and restoring the state of ReaderBase, see
-   * reader_base.h for details.
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.ReaderBaseState}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.ReaderBaseState)
-      org.tensorflow.proto.framework.ReaderBaseStateOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.ReaderBaseProtos.internal_static_tensorflow_ReaderBaseState_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.ReaderBaseProtos.internal_static_tensorflow_ReaderBaseState_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.ReaderBaseState.class, org.tensorflow.proto.framework.ReaderBaseState.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.framework.ReaderBaseState.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      workStarted_ = 0L;
-
-      workFinished_ = 0L;
-
-      numRecordsProduced_ = 0L;
-
-      currentWork_ = com.google.protobuf.ByteString.EMPTY;
-
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.framework.ReaderBaseProtos.internal_static_tensorflow_ReaderBaseState_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.ReaderBaseState getDefaultInstanceForType() {
-      return org.tensorflow.proto.framework.ReaderBaseState.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.ReaderBaseState build() {
-      org.tensorflow.proto.framework.ReaderBaseState result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.ReaderBaseState buildPartial() {
-      org.tensorflow.proto.framework.ReaderBaseState result = new org.tensorflow.proto.framework.ReaderBaseState(this);
-      result.workStarted_ = workStarted_;
-      result.workFinished_ = workFinished_;
-      result.numRecordsProduced_ = numRecordsProduced_;
-      result.currentWork_ = currentWork_;
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.framework.ReaderBaseState) {
-        return mergeFrom((org.tensorflow.proto.framework.ReaderBaseState)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.framework.ReaderBaseState other) {
-      if (other == org.tensorflow.proto.framework.ReaderBaseState.getDefaultInstance()) return this;
-      if (other.getWorkStarted() != 0L) {
-        setWorkStarted(other.getWorkStarted());
-      }
-      if (other.getWorkFinished() != 0L) {
-        setWorkFinished(other.getWorkFinished());
-      }
-      if (other.getNumRecordsProduced() != 0L) {
-        setNumRecordsProduced(other.getNumRecordsProduced());
-      }
-      if (other.getCurrentWork() != com.google.protobuf.ByteString.EMPTY) {
-        setCurrentWork(other.getCurrentWork());
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.framework.ReaderBaseState parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.framework.ReaderBaseState) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-
-    private long workStarted_ ;
-    /**
-     * <code>int64 work_started = 1;</code>
-     */
-    public long getWorkStarted() {
-      return workStarted_;
-    }
-    /**
-     * <code>int64 work_started = 1;</code>
-     */
-    public Builder setWorkStarted(long value) {
-      
-      workStarted_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>int64 work_started = 1;</code>
-     */
-    public Builder clearWorkStarted() {
-      
-      workStarted_ = 0L;
-      onChanged();
-      return this;
-    }
-
-    private long workFinished_ ;
-    /**
-     * <code>int64 work_finished = 2;</code>
-     */
-    public long getWorkFinished() {
-      return workFinished_;
-    }
-    /**
-     * <code>int64 work_finished = 2;</code>
-     */
-    public Builder setWorkFinished(long value) {
-      
-      workFinished_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>int64 work_finished = 2;</code>
-     */
-    public Builder clearWorkFinished() {
-      
-      workFinished_ = 0L;
-      onChanged();
-      return this;
-    }
-
-    private long numRecordsProduced_ ;
-    /**
-     * <code>int64 num_records_produced = 3;</code>
-     */
-    public long getNumRecordsProduced() {
-      return numRecordsProduced_;
-    }
-    /**
-     * <code>int64 num_records_produced = 3;</code>
-     */
-    public Builder setNumRecordsProduced(long value) {
-      
-      numRecordsProduced_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>int64 num_records_produced = 3;</code>
-     */
-    public Builder clearNumRecordsProduced() {
-      
-      numRecordsProduced_ = 0L;
-      onChanged();
-      return this;
-    }
-
-    private com.google.protobuf.ByteString currentWork_ = com.google.protobuf.ByteString.EMPTY;
-    /**
-     * <code>bytes current_work = 4;</code>
-     */
-    public com.google.protobuf.ByteString getCurrentWork() {
-      return currentWork_;
-    }
-    /**
-     * <code>bytes current_work = 4;</code>
-     */
-    public Builder setCurrentWork(com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      currentWork_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>bytes current_work = 4;</code>
-     */
-    public Builder clearCurrentWork() {
-      
-      currentWork_ = getDefaultInstance().getCurrentWork();
-      onChanged();
-      return this;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.ReaderBaseState)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.ReaderBaseState)
-  private static final org.tensorflow.proto.framework.ReaderBaseState DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.ReaderBaseState();
-  }
-
-  public static org.tensorflow.proto.framework.ReaderBaseState getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<ReaderBaseState>
-      PARSER = new com.google.protobuf.AbstractParser<ReaderBaseState>() {
-    @java.lang.Override
-    public ReaderBaseState parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new ReaderBaseState(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<ReaderBaseState> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<ReaderBaseState> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.framework.ReaderBaseState getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ReaderBaseStateOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ReaderBaseStateOrBuilder.java
deleted file mode 100644
index 72f0c6c0922..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ReaderBaseStateOrBuilder.java
+++ /dev/null
@@ -1,29 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/framework/reader_base.proto
-
-package org.tensorflow.proto.framework;
-
-public interface ReaderBaseStateOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.ReaderBaseState)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <code>int64 work_started = 1;</code>
-   */
-  long getWorkStarted();
-
-  /**
-   * <code>int64 work_finished = 2;</code>
-   */
-  long getWorkFinished();
-
-  /**
-   * <code>int64 num_records_produced = 3;</code>
-   */
-  long getNumRecordsProduced();
-
-  /**
-   * <code>bytes current_work = 4;</code>
-   */
-  com.google.protobuf.ByteString getCurrentWork();
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/RemoteFusedGraphExecuteInfo.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/RemoteFusedGraphExecuteInfo.java
deleted file mode 100644
index bc2e6840de2..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/RemoteFusedGraphExecuteInfo.java
+++ /dev/null
@@ -1,3007 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/framework/remote_fused_graph_execute_info.proto
-
-package org.tensorflow.proto.framework;
-
-/**
- * <pre>
- * Protocol buffer representing a handle to a tensorflow resource. Handles are
- * not valid across executions, but can be serialized back and forth from within
- * a single run.
- * </pre>
- *
- * Protobuf type {@code tensorflow.RemoteFusedGraphExecuteInfo}
- */
-public  final class RemoteFusedGraphExecuteInfo extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.RemoteFusedGraphExecuteInfo)
-    RemoteFusedGraphExecuteInfoOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use RemoteFusedGraphExecuteInfo.newBuilder() to construct.
-  private RemoteFusedGraphExecuteInfo(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private RemoteFusedGraphExecuteInfo() {
-    graphInputNodeName_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-    graphOutputNodeName_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-    executorName_ = "";
-    serializedExecutorParameters_ = com.google.protobuf.ByteString.EMPTY;
-    defaultGraphInputTensorShape_ = java.util.Collections.emptyList();
-    defaultGraphOutputTensorShape_ = java.util.Collections.emptyList();
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new RemoteFusedGraphExecuteInfo();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private RemoteFusedGraphExecuteInfo(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    int mutable_bitField0_ = 0;
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 10: {
-            org.tensorflow.proto.framework.GraphDef.Builder subBuilder = null;
-            if (remoteGraph_ != null) {
-              subBuilder = remoteGraph_.toBuilder();
-            }
-            remoteGraph_ = input.readMessage(org.tensorflow.proto.framework.GraphDef.parser(), extensionRegistry);
-            if (subBuilder != null) {
-              subBuilder.mergeFrom(remoteGraph_);
-              remoteGraph_ = subBuilder.buildPartial();
-            }
-
-            break;
-          }
-          case 18: {
-            java.lang.String s = input.readStringRequireUtf8();
-            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
-              graphInputNodeName_ = new com.google.protobuf.LazyStringArrayList();
-              mutable_bitField0_ |= 0x00000001;
-            }
-            graphInputNodeName_.add(s);
-            break;
-          }
-          case 26: {
-            java.lang.String s = input.readStringRequireUtf8();
-            if (!((mutable_bitField0_ & 0x00000002) != 0)) {
-              graphOutputNodeName_ = new com.google.protobuf.LazyStringArrayList();
-              mutable_bitField0_ |= 0x00000002;
-            }
-            graphOutputNodeName_.add(s);
-            break;
-          }
-          case 34: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            executorName_ = s;
-            break;
-          }
-          case 42: {
-
-            serializedExecutorParameters_ = input.readBytes();
-            break;
-          }
-          case 50: {
-            if (!((mutable_bitField0_ & 0x00000004) != 0)) {
-              defaultGraphInputTensorShape_ = new java.util.ArrayList<org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto>();
-              mutable_bitField0_ |= 0x00000004;
-            }
-            defaultGraphInputTensorShape_.add(
-                input.readMessage(org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto.parser(), extensionRegistry));
-            break;
-          }
-          case 58: {
-            if (!((mutable_bitField0_ & 0x00000008) != 0)) {
-              defaultGraphOutputTensorShape_ = new java.util.ArrayList<org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto>();
-              mutable_bitField0_ |= 0x00000008;
-            }
-            defaultGraphOutputTensorShape_.add(
-                input.readMessage(org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto.parser(), extensionRegistry));
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      if (((mutable_bitField0_ & 0x00000001) != 0)) {
-        graphInputNodeName_ = graphInputNodeName_.getUnmodifiableView();
-      }
-      if (((mutable_bitField0_ & 0x00000002) != 0)) {
-        graphOutputNodeName_ = graphOutputNodeName_.getUnmodifiableView();
-      }
-      if (((mutable_bitField0_ & 0x00000004) != 0)) {
-        defaultGraphInputTensorShape_ = java.util.Collections.unmodifiableList(defaultGraphInputTensorShape_);
-      }
-      if (((mutable_bitField0_ & 0x00000008) != 0)) {
-        defaultGraphOutputTensorShape_ = java.util.Collections.unmodifiableList(defaultGraphOutputTensorShape_);
-      }
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfoProto.internal_static_tensorflow_RemoteFusedGraphExecuteInfo_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfoProto.internal_static_tensorflow_RemoteFusedGraphExecuteInfo_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.class, org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.Builder.class);
-  }
-
-  public interface TensorShapeTypeProtoOrBuilder extends
-      // @@protoc_insertion_point(interface_extends:tensorflow.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto)
-      com.google.protobuf.MessageOrBuilder {
-
-    /**
-     * <code>.tensorflow.DataType dtype = 1;</code>
-     */
-    int getDtypeValue();
-    /**
-     * <code>.tensorflow.DataType dtype = 1;</code>
-     */
-    org.tensorflow.proto.framework.DataType getDtype();
-
-    /**
-     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-     */
-    boolean hasShape();
-    /**
-     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-     */
-    org.tensorflow.proto.framework.TensorShapeProto getShape();
-    /**
-     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-     */
-    org.tensorflow.proto.framework.TensorShapeProtoOrBuilder getShapeOrBuilder();
-  }
-  /**
-   * Protobuf type {@code tensorflow.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto}
-   */
-  public  static final class TensorShapeTypeProto extends
-      com.google.protobuf.GeneratedMessageV3 implements
-      // @@protoc_insertion_point(message_implements:tensorflow.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto)
-      TensorShapeTypeProtoOrBuilder {
-  private static final long serialVersionUID = 0L;
-    // Use TensorShapeTypeProto.newBuilder() to construct.
-    private TensorShapeTypeProto(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-      super(builder);
-    }
-    private TensorShapeTypeProto() {
-      dtype_ = 0;
-    }
-
-    @java.lang.Override
-    @SuppressWarnings({"unused"})
-    protected java.lang.Object newInstance(
-        UnusedPrivateParameter unused) {
-      return new TensorShapeTypeProto();
-    }
-
-    @java.lang.Override
-    public final com.google.protobuf.UnknownFieldSet
-    getUnknownFields() {
-      return this.unknownFields;
-    }
-    private TensorShapeTypeProto(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      this();
-      if (extensionRegistry == null) {
-        throw new java.lang.NullPointerException();
-      }
-      com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-          com.google.protobuf.UnknownFieldSet.newBuilder();
-      try {
-        boolean done = false;
-        while (!done) {
-          int tag = input.readTag();
-          switch (tag) {
-            case 0:
-              done = true;
-              break;
-            case 8: {
-              int rawValue = input.readEnum();
-
-              dtype_ = rawValue;
-              break;
-            }
-            case 18: {
-              org.tensorflow.proto.framework.TensorShapeProto.Builder subBuilder = null;
-              if (shape_ != null) {
-                subBuilder = shape_.toBuilder();
-              }
-              shape_ = input.readMessage(org.tensorflow.proto.framework.TensorShapeProto.parser(), extensionRegistry);
-              if (subBuilder != null) {
-                subBuilder.mergeFrom(shape_);
-                shape_ = subBuilder.buildPartial();
-              }
-
-              break;
-            }
-            default: {
-              if (!parseUnknownField(
-                  input, unknownFields, extensionRegistry, tag)) {
-                done = true;
-              }
-              break;
-            }
-          }
-        }
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        throw e.setUnfinishedMessage(this);
-      } catch (java.io.IOException e) {
-        throw new com.google.protobuf.InvalidProtocolBufferException(
-            e).setUnfinishedMessage(this);
-      } finally {
-        this.unknownFields = unknownFields.build();
-        makeExtensionsImmutable();
-      }
-    }
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfoProto.internal_static_tensorflow_RemoteFusedGraphExecuteInfo_TensorShapeTypeProto_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfoProto.internal_static_tensorflow_RemoteFusedGraphExecuteInfo_TensorShapeTypeProto_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto.class, org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto.Builder.class);
-    }
-
-    public static final int DTYPE_FIELD_NUMBER = 1;
-    private int dtype_;
-    /**
-     * <code>.tensorflow.DataType dtype = 1;</code>
-     */
-    public int getDtypeValue() {
-      return dtype_;
-    }
-    /**
-     * <code>.tensorflow.DataType dtype = 1;</code>
-     */
-    public org.tensorflow.proto.framework.DataType getDtype() {
-      @SuppressWarnings("deprecation")
-      org.tensorflow.proto.framework.DataType result = org.tensorflow.proto.framework.DataType.valueOf(dtype_);
-      return result == null ? org.tensorflow.proto.framework.DataType.UNRECOGNIZED : result;
-    }
-
-    public static final int SHAPE_FIELD_NUMBER = 2;
-    private org.tensorflow.proto.framework.TensorShapeProto shape_;
-    /**
-     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-     */
-    public boolean hasShape() {
-      return shape_ != null;
-    }
-    /**
-     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-     */
-    public org.tensorflow.proto.framework.TensorShapeProto getShape() {
-      return shape_ == null ? org.tensorflow.proto.framework.TensorShapeProto.getDefaultInstance() : shape_;
-    }
-    /**
-     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-     */
-    public org.tensorflow.proto.framework.TensorShapeProtoOrBuilder getShapeOrBuilder() {
-      return getShape();
-    }
-
-    private byte memoizedIsInitialized = -1;
-    @java.lang.Override
-    public final boolean isInitialized() {
-      byte isInitialized = memoizedIsInitialized;
-      if (isInitialized == 1) return true;
-      if (isInitialized == 0) return false;
-
-      memoizedIsInitialized = 1;
-      return true;
-    }
-
-    @java.lang.Override
-    public void writeTo(com.google.protobuf.CodedOutputStream output)
-                        throws java.io.IOException {
-      if (dtype_ != org.tensorflow.proto.framework.DataType.DT_INVALID.getNumber()) {
-        output.writeEnum(1, dtype_);
-      }
-      if (shape_ != null) {
-        output.writeMessage(2, getShape());
-      }
-      unknownFields.writeTo(output);
-    }
-
-    @java.lang.Override
-    public int getSerializedSize() {
-      int size = memoizedSize;
-      if (size != -1) return size;
-
-      size = 0;
-      if (dtype_ != org.tensorflow.proto.framework.DataType.DT_INVALID.getNumber()) {
-        size += com.google.protobuf.CodedOutputStream
-          .computeEnumSize(1, dtype_);
-      }
-      if (shape_ != null) {
-        size += com.google.protobuf.CodedOutputStream
-          .computeMessageSize(2, getShape());
-      }
-      size += unknownFields.getSerializedSize();
-      memoizedSize = size;
-      return size;
-    }
-
-    @java.lang.Override
-    public boolean equals(final java.lang.Object obj) {
-      if (obj == this) {
-       return true;
-      }
-      if (!(obj instanceof org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto)) {
-        return super.equals(obj);
-      }
-      org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto other = (org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto) obj;
-
-      if (dtype_ != other.dtype_) return false;
-      if (hasShape() != other.hasShape()) return false;
-      if (hasShape()) {
-        if (!getShape()
-            .equals(other.getShape())) return false;
-      }
-      if (!unknownFields.equals(other.unknownFields)) return false;
-      return true;
-    }
-
-    @java.lang.Override
-    public int hashCode() {
-      if (memoizedHashCode != 0) {
-        return memoizedHashCode;
-      }
-      int hash = 41;
-      hash = (19 * hash) + getDescriptor().hashCode();
-      hash = (37 * hash) + DTYPE_FIELD_NUMBER;
-      hash = (53 * hash) + dtype_;
-      if (hasShape()) {
-        hash = (37 * hash) + SHAPE_FIELD_NUMBER;
-        hash = (53 * hash) + getShape().hashCode();
-      }
-      hash = (29 * hash) + unknownFields.hashCode();
-      memoizedHashCode = hash;
-      return hash;
-    }
-
-    public static org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto parseFrom(
-        java.nio.ByteBuffer data)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data);
-    }
-    public static org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto parseFrom(
-        java.nio.ByteBuffer data,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto parseFrom(
-        com.google.protobuf.ByteString data)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data);
-    }
-    public static org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto parseFrom(
-        com.google.protobuf.ByteString data,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto parseFrom(byte[] data)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data);
-    }
-    public static org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto parseFrom(
-        byte[] data,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto parseFrom(java.io.InputStream input)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input);
-    }
-    public static org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto parseFrom(
-        java.io.InputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto parseDelimitedFrom(java.io.InputStream input)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseDelimitedWithIOException(PARSER, input);
-    }
-    public static org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto parseDelimitedFrom(
-        java.io.InputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto parseFrom(
-        com.google.protobuf.CodedInputStream input)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input);
-    }
-    public static org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto parseFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input, extensionRegistry);
-    }
-
-    @java.lang.Override
-    public Builder newBuilderForType() { return newBuilder(); }
-    public static Builder newBuilder() {
-      return DEFAULT_INSTANCE.toBuilder();
-    }
-    public static Builder newBuilder(org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto prototype) {
-      return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-    }
-    @java.lang.Override
-    public Builder toBuilder() {
-      return this == DEFAULT_INSTANCE
-          ? new Builder() : new Builder().mergeFrom(this);
-    }
-
-    @java.lang.Override
-    protected Builder newBuilderForType(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      Builder builder = new Builder(parent);
-      return builder;
-    }
-    /**
-     * Protobuf type {@code tensorflow.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto}
-     */
-    public static final class Builder extends
-        com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-        // @@protoc_insertion_point(builder_implements:tensorflow.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto)
-        org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProtoOrBuilder {
-      public static final com.google.protobuf.Descriptors.Descriptor
-          getDescriptor() {
-        return org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfoProto.internal_static_tensorflow_RemoteFusedGraphExecuteInfo_TensorShapeTypeProto_descriptor;
-      }
-
-      @java.lang.Override
-      protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-          internalGetFieldAccessorTable() {
-        return org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfoProto.internal_static_tensorflow_RemoteFusedGraphExecuteInfo_TensorShapeTypeProto_fieldAccessorTable
-            .ensureFieldAccessorsInitialized(
-                org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto.class, org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto.Builder.class);
-      }
-
-      // Construct using org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto.newBuilder()
-      private Builder() {
-        maybeForceBuilderInitialization();
-      }
-
-      private Builder(
-          com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-        super(parent);
-        maybeForceBuilderInitialization();
-      }
-      private void maybeForceBuilderInitialization() {
-        if (com.google.protobuf.GeneratedMessageV3
-                .alwaysUseFieldBuilders) {
-        }
-      }
-      @java.lang.Override
-      public Builder clear() {
-        super.clear();
-        dtype_ = 0;
-
-        if (shapeBuilder_ == null) {
-          shape_ = null;
-        } else {
-          shape_ = null;
-          shapeBuilder_ = null;
-        }
-        return this;
-      }
-
-      @java.lang.Override
-      public com.google.protobuf.Descriptors.Descriptor
-          getDescriptorForType() {
-        return org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfoProto.internal_static_tensorflow_RemoteFusedGraphExecuteInfo_TensorShapeTypeProto_descriptor;
-      }
-
-      @java.lang.Override
-      public org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto getDefaultInstanceForType() {
-        return org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto.getDefaultInstance();
-      }
-
-      @java.lang.Override
-      public org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto build() {
-        org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto result = buildPartial();
-        if (!result.isInitialized()) {
-          throw newUninitializedMessageException(result);
-        }
-        return result;
-      }
-
-      @java.lang.Override
-      public org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto buildPartial() {
-        org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto result = new org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto(this);
-        result.dtype_ = dtype_;
-        if (shapeBuilder_ == null) {
-          result.shape_ = shape_;
-        } else {
-          result.shape_ = shapeBuilder_.build();
-        }
-        onBuilt();
-        return result;
-      }
-
-      @java.lang.Override
-      public Builder clone() {
-        return super.clone();
-      }
-      @java.lang.Override
-      public Builder setField(
-          com.google.protobuf.Descriptors.FieldDescriptor field,
-          java.lang.Object value) {
-        return super.setField(field, value);
-      }
-      @java.lang.Override
-      public Builder clearField(
-          com.google.protobuf.Descriptors.FieldDescriptor field) {
-        return super.clearField(field);
-      }
-      @java.lang.Override
-      public Builder clearOneof(
-          com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-        return super.clearOneof(oneof);
-      }
-      @java.lang.Override
-      public Builder setRepeatedField(
-          com.google.protobuf.Descriptors.FieldDescriptor field,
-          int index, java.lang.Object value) {
-        return super.setRepeatedField(field, index, value);
-      }
-      @java.lang.Override
-      public Builder addRepeatedField(
-          com.google.protobuf.Descriptors.FieldDescriptor field,
-          java.lang.Object value) {
-        return super.addRepeatedField(field, value);
-      }
-      @java.lang.Override
-      public Builder mergeFrom(com.google.protobuf.Message other) {
-        if (other instanceof org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto) {
-          return mergeFrom((org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto)other);
-        } else {
-          super.mergeFrom(other);
-          return this;
-        }
-      }
-
-      public Builder mergeFrom(org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto other) {
-        if (other == org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto.getDefaultInstance()) return this;
-        if (other.dtype_ != 0) {
-          setDtypeValue(other.getDtypeValue());
-        }
-        if (other.hasShape()) {
-          mergeShape(other.getShape());
-        }
-        this.mergeUnknownFields(other.unknownFields);
-        onChanged();
-        return this;
-      }
-
-      @java.lang.Override
-      public final boolean isInitialized() {
-        return true;
-      }
-
-      @java.lang.Override
-      public Builder mergeFrom(
-          com.google.protobuf.CodedInputStream input,
-          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-          throws java.io.IOException {
-        org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto parsedMessage = null;
-        try {
-          parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-        } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-          parsedMessage = (org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto) e.getUnfinishedMessage();
-          throw e.unwrapIOException();
-        } finally {
-          if (parsedMessage != null) {
-            mergeFrom(parsedMessage);
-          }
-        }
-        return this;
-      }
-
-      private int dtype_ = 0;
-      /**
-       * <code>.tensorflow.DataType dtype = 1;</code>
-       */
-      public int getDtypeValue() {
-        return dtype_;
-      }
-      /**
-       * <code>.tensorflow.DataType dtype = 1;</code>
-       */
-      public Builder setDtypeValue(int value) {
-        dtype_ = value;
-        onChanged();
-        return this;
-      }
-      /**
-       * <code>.tensorflow.DataType dtype = 1;</code>
-       */
-      public org.tensorflow.proto.framework.DataType getDtype() {
-        @SuppressWarnings("deprecation")
-        org.tensorflow.proto.framework.DataType result = org.tensorflow.proto.framework.DataType.valueOf(dtype_);
-        return result == null ? org.tensorflow.proto.framework.DataType.UNRECOGNIZED : result;
-      }
-      /**
-       * <code>.tensorflow.DataType dtype = 1;</code>
-       */
-      public Builder setDtype(org.tensorflow.proto.framework.DataType value) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        
-        dtype_ = value.getNumber();
-        onChanged();
-        return this;
-      }
-      /**
-       * <code>.tensorflow.DataType dtype = 1;</code>
-       */
-      public Builder clearDtype() {
-        
-        dtype_ = 0;
-        onChanged();
-        return this;
-      }
-
-      private org.tensorflow.proto.framework.TensorShapeProto shape_;
-      private com.google.protobuf.SingleFieldBuilderV3<
-          org.tensorflow.proto.framework.TensorShapeProto, org.tensorflow.proto.framework.TensorShapeProto.Builder, org.tensorflow.proto.framework.TensorShapeProtoOrBuilder> shapeBuilder_;
-      /**
-       * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-       */
-      public boolean hasShape() {
-        return shapeBuilder_ != null || shape_ != null;
-      }
-      /**
-       * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-       */
-      public org.tensorflow.proto.framework.TensorShapeProto getShape() {
-        if (shapeBuilder_ == null) {
-          return shape_ == null ? org.tensorflow.proto.framework.TensorShapeProto.getDefaultInstance() : shape_;
-        } else {
-          return shapeBuilder_.getMessage();
-        }
-      }
-      /**
-       * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-       */
-      public Builder setShape(org.tensorflow.proto.framework.TensorShapeProto value) {
-        if (shapeBuilder_ == null) {
-          if (value == null) {
-            throw new NullPointerException();
-          }
-          shape_ = value;
-          onChanged();
-        } else {
-          shapeBuilder_.setMessage(value);
-        }
-
-        return this;
-      }
-      /**
-       * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-       */
-      public Builder setShape(
-          org.tensorflow.proto.framework.TensorShapeProto.Builder builderForValue) {
-        if (shapeBuilder_ == null) {
-          shape_ = builderForValue.build();
-          onChanged();
-        } else {
-          shapeBuilder_.setMessage(builderForValue.build());
-        }
-
-        return this;
-      }
-      /**
-       * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-       */
-      public Builder mergeShape(org.tensorflow.proto.framework.TensorShapeProto value) {
-        if (shapeBuilder_ == null) {
-          if (shape_ != null) {
-            shape_ =
-              org.tensorflow.proto.framework.TensorShapeProto.newBuilder(shape_).mergeFrom(value).buildPartial();
-          } else {
-            shape_ = value;
-          }
-          onChanged();
-        } else {
-          shapeBuilder_.mergeFrom(value);
-        }
-
-        return this;
-      }
-      /**
-       * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-       */
-      public Builder clearShape() {
-        if (shapeBuilder_ == null) {
-          shape_ = null;
-          onChanged();
-        } else {
-          shape_ = null;
-          shapeBuilder_ = null;
-        }
-
-        return this;
-      }
-      /**
-       * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-       */
-      public org.tensorflow.proto.framework.TensorShapeProto.Builder getShapeBuilder() {
-        
-        onChanged();
-        return getShapeFieldBuilder().getBuilder();
-      }
-      /**
-       * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-       */
-      public org.tensorflow.proto.framework.TensorShapeProtoOrBuilder getShapeOrBuilder() {
-        if (shapeBuilder_ != null) {
-          return shapeBuilder_.getMessageOrBuilder();
-        } else {
-          return shape_ == null ?
-              org.tensorflow.proto.framework.TensorShapeProto.getDefaultInstance() : shape_;
-        }
-      }
-      /**
-       * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-       */
-      private com.google.protobuf.SingleFieldBuilderV3<
-          org.tensorflow.proto.framework.TensorShapeProto, org.tensorflow.proto.framework.TensorShapeProto.Builder, org.tensorflow.proto.framework.TensorShapeProtoOrBuilder> 
-          getShapeFieldBuilder() {
-        if (shapeBuilder_ == null) {
-          shapeBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-              org.tensorflow.proto.framework.TensorShapeProto, org.tensorflow.proto.framework.TensorShapeProto.Builder, org.tensorflow.proto.framework.TensorShapeProtoOrBuilder>(
-                  getShape(),
-                  getParentForChildren(),
-                  isClean());
-          shape_ = null;
-        }
-        return shapeBuilder_;
-      }
-      @java.lang.Override
-      public final Builder setUnknownFields(
-          final com.google.protobuf.UnknownFieldSet unknownFields) {
-        return super.setUnknownFields(unknownFields);
-      }
-
-      @java.lang.Override
-      public final Builder mergeUnknownFields(
-          final com.google.protobuf.UnknownFieldSet unknownFields) {
-        return super.mergeUnknownFields(unknownFields);
-      }
-
-
-      // @@protoc_insertion_point(builder_scope:tensorflow.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto)
-    }
-
-    // @@protoc_insertion_point(class_scope:tensorflow.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto)
-    private static final org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto DEFAULT_INSTANCE;
-    static {
-      DEFAULT_INSTANCE = new org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto();
-    }
-
-    public static org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto getDefaultInstance() {
-      return DEFAULT_INSTANCE;
-    }
-
-    private static final com.google.protobuf.Parser<TensorShapeTypeProto>
-        PARSER = new com.google.protobuf.AbstractParser<TensorShapeTypeProto>() {
-      @java.lang.Override
-      public TensorShapeTypeProto parsePartialFrom(
-          com.google.protobuf.CodedInputStream input,
-          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-          throws com.google.protobuf.InvalidProtocolBufferException {
-        return new TensorShapeTypeProto(input, extensionRegistry);
-      }
-    };
-
-    public static com.google.protobuf.Parser<TensorShapeTypeProto> parser() {
-      return PARSER;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Parser<TensorShapeTypeProto> getParserForType() {
-      return PARSER;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto getDefaultInstanceForType() {
-      return DEFAULT_INSTANCE;
-    }
-
-  }
-
-  public static final int REMOTE_GRAPH_FIELD_NUMBER = 1;
-  private org.tensorflow.proto.framework.GraphDef remoteGraph_;
-  /**
-   * <pre>
-   * Definition of remote graph
-   * </pre>
-   *
-   * <code>.tensorflow.GraphDef remote_graph = 1;</code>
-   */
-  public boolean hasRemoteGraph() {
-    return remoteGraph_ != null;
-  }
-  /**
-   * <pre>
-   * Definition of remote graph
-   * </pre>
-   *
-   * <code>.tensorflow.GraphDef remote_graph = 1;</code>
-   */
-  public org.tensorflow.proto.framework.GraphDef getRemoteGraph() {
-    return remoteGraph_ == null ? org.tensorflow.proto.framework.GraphDef.getDefaultInstance() : remoteGraph_;
-  }
-  /**
-   * <pre>
-   * Definition of remote graph
-   * </pre>
-   *
-   * <code>.tensorflow.GraphDef remote_graph = 1;</code>
-   */
-  public org.tensorflow.proto.framework.GraphDefOrBuilder getRemoteGraphOrBuilder() {
-    return getRemoteGraph();
-  }
-
-  public static final int GRAPH_INPUT_NODE_NAME_FIELD_NUMBER = 2;
-  private com.google.protobuf.LazyStringList graphInputNodeName_;
-  /**
-   * <pre>
-   * Remote fused graph input node name
-   * </pre>
-   *
-   * <code>repeated string graph_input_node_name = 2;</code>
-   */
-  public com.google.protobuf.ProtocolStringList
-      getGraphInputNodeNameList() {
-    return graphInputNodeName_;
-  }
-  /**
-   * <pre>
-   * Remote fused graph input node name
-   * </pre>
-   *
-   * <code>repeated string graph_input_node_name = 2;</code>
-   */
-  public int getGraphInputNodeNameCount() {
-    return graphInputNodeName_.size();
-  }
-  /**
-   * <pre>
-   * Remote fused graph input node name
-   * </pre>
-   *
-   * <code>repeated string graph_input_node_name = 2;</code>
-   */
-  public java.lang.String getGraphInputNodeName(int index) {
-    return graphInputNodeName_.get(index);
-  }
-  /**
-   * <pre>
-   * Remote fused graph input node name
-   * </pre>
-   *
-   * <code>repeated string graph_input_node_name = 2;</code>
-   */
-  public com.google.protobuf.ByteString
-      getGraphInputNodeNameBytes(int index) {
-    return graphInputNodeName_.getByteString(index);
-  }
-
-  public static final int GRAPH_OUTPUT_NODE_NAME_FIELD_NUMBER = 3;
-  private com.google.protobuf.LazyStringList graphOutputNodeName_;
-  /**
-   * <pre>
-   * Remote fused graph output node name
-   * </pre>
-   *
-   * <code>repeated string graph_output_node_name = 3;</code>
-   */
-  public com.google.protobuf.ProtocolStringList
-      getGraphOutputNodeNameList() {
-    return graphOutputNodeName_;
-  }
-  /**
-   * <pre>
-   * Remote fused graph output node name
-   * </pre>
-   *
-   * <code>repeated string graph_output_node_name = 3;</code>
-   */
-  public int getGraphOutputNodeNameCount() {
-    return graphOutputNodeName_.size();
-  }
-  /**
-   * <pre>
-   * Remote fused graph output node name
-   * </pre>
-   *
-   * <code>repeated string graph_output_node_name = 3;</code>
-   */
-  public java.lang.String getGraphOutputNodeName(int index) {
-    return graphOutputNodeName_.get(index);
-  }
-  /**
-   * <pre>
-   * Remote fused graph output node name
-   * </pre>
-   *
-   * <code>repeated string graph_output_node_name = 3;</code>
-   */
-  public com.google.protobuf.ByteString
-      getGraphOutputNodeNameBytes(int index) {
-    return graphOutputNodeName_.getByteString(index);
-  }
-
-  public static final int EXECUTOR_NAME_FIELD_NUMBER = 4;
-  private volatile java.lang.Object executorName_;
-  /**
-   * <pre>
-   * Executor's name
-   * </pre>
-   *
-   * <code>string executor_name = 4;</code>
-   */
-  public java.lang.String getExecutorName() {
-    java.lang.Object ref = executorName_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      executorName_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * Executor's name
-   * </pre>
-   *
-   * <code>string executor_name = 4;</code>
-   */
-  public com.google.protobuf.ByteString
-      getExecutorNameBytes() {
-    java.lang.Object ref = executorName_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      executorName_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int SERIALIZED_EXECUTOR_PARAMETERS_FIELD_NUMBER = 5;
-  private com.google.protobuf.ByteString serializedExecutorParameters_;
-  /**
-   * <pre>
-   * Optional: Parameters given to the executor
-   * </pre>
-   *
-   * <code>bytes serialized_executor_parameters = 5;</code>
-   */
-  public com.google.protobuf.ByteString getSerializedExecutorParameters() {
-    return serializedExecutorParameters_;
-  }
-
-  public static final int DEFAULT_GRAPH_INPUT_TENSOR_SHAPE_FIELD_NUMBER = 6;
-  private java.util.List<org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto> defaultGraphInputTensorShape_;
-  /**
-   * <pre>
-   * Optional: Default graph input tensor shape used to allocate memory
-   * before executing op
-   * </pre>
-   *
-   * <code>repeated .tensorflow.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto default_graph_input_tensor_shape = 6;</code>
-   */
-  public java.util.List<org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto> getDefaultGraphInputTensorShapeList() {
-    return defaultGraphInputTensorShape_;
-  }
-  /**
-   * <pre>
-   * Optional: Default graph input tensor shape used to allocate memory
-   * before executing op
-   * </pre>
-   *
-   * <code>repeated .tensorflow.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto default_graph_input_tensor_shape = 6;</code>
-   */
-  public java.util.List<? extends org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProtoOrBuilder> 
-      getDefaultGraphInputTensorShapeOrBuilderList() {
-    return defaultGraphInputTensorShape_;
-  }
-  /**
-   * <pre>
-   * Optional: Default graph input tensor shape used to allocate memory
-   * before executing op
-   * </pre>
-   *
-   * <code>repeated .tensorflow.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto default_graph_input_tensor_shape = 6;</code>
-   */
-  public int getDefaultGraphInputTensorShapeCount() {
-    return defaultGraphInputTensorShape_.size();
-  }
-  /**
-   * <pre>
-   * Optional: Default graph input tensor shape used to allocate memory
-   * before executing op
-   * </pre>
-   *
-   * <code>repeated .tensorflow.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto default_graph_input_tensor_shape = 6;</code>
-   */
-  public org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto getDefaultGraphInputTensorShape(int index) {
-    return defaultGraphInputTensorShape_.get(index);
-  }
-  /**
-   * <pre>
-   * Optional: Default graph input tensor shape used to allocate memory
-   * before executing op
-   * </pre>
-   *
-   * <code>repeated .tensorflow.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto default_graph_input_tensor_shape = 6;</code>
-   */
-  public org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProtoOrBuilder getDefaultGraphInputTensorShapeOrBuilder(
-      int index) {
-    return defaultGraphInputTensorShape_.get(index);
-  }
-
-  public static final int DEFAULT_GRAPH_OUTPUT_TENSOR_SHAPE_FIELD_NUMBER = 7;
-  private java.util.List<org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto> defaultGraphOutputTensorShape_;
-  /**
-   * <pre>
-   * Optional: Default graph input tensor shape used to allocate memory
-   * before executing op
-   * TODO(satok): Remote output tensor shape once shape information is stored
-   * in NodeDef
-   * </pre>
-   *
-   * <code>repeated .tensorflow.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto default_graph_output_tensor_shape = 7;</code>
-   */
-  public java.util.List<org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto> getDefaultGraphOutputTensorShapeList() {
-    return defaultGraphOutputTensorShape_;
-  }
-  /**
-   * <pre>
-   * Optional: Default graph input tensor shape used to allocate memory
-   * before executing op
-   * TODO(satok): Remote output tensor shape once shape information is stored
-   * in NodeDef
-   * </pre>
-   *
-   * <code>repeated .tensorflow.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto default_graph_output_tensor_shape = 7;</code>
-   */
-  public java.util.List<? extends org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProtoOrBuilder> 
-      getDefaultGraphOutputTensorShapeOrBuilderList() {
-    return defaultGraphOutputTensorShape_;
-  }
-  /**
-   * <pre>
-   * Optional: Default graph input tensor shape used to allocate memory
-   * before executing op
-   * TODO(satok): Remote output tensor shape once shape information is stored
-   * in NodeDef
-   * </pre>
-   *
-   * <code>repeated .tensorflow.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto default_graph_output_tensor_shape = 7;</code>
-   */
-  public int getDefaultGraphOutputTensorShapeCount() {
-    return defaultGraphOutputTensorShape_.size();
-  }
-  /**
-   * <pre>
-   * Optional: Default graph input tensor shape used to allocate memory
-   * before executing op
-   * TODO(satok): Remote output tensor shape once shape information is stored
-   * in NodeDef
-   * </pre>
-   *
-   * <code>repeated .tensorflow.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto default_graph_output_tensor_shape = 7;</code>
-   */
-  public org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto getDefaultGraphOutputTensorShape(int index) {
-    return defaultGraphOutputTensorShape_.get(index);
-  }
-  /**
-   * <pre>
-   * Optional: Default graph input tensor shape used to allocate memory
-   * before executing op
-   * TODO(satok): Remote output tensor shape once shape information is stored
-   * in NodeDef
-   * </pre>
-   *
-   * <code>repeated .tensorflow.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto default_graph_output_tensor_shape = 7;</code>
-   */
-  public org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProtoOrBuilder getDefaultGraphOutputTensorShapeOrBuilder(
-      int index) {
-    return defaultGraphOutputTensorShape_.get(index);
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    if (remoteGraph_ != null) {
-      output.writeMessage(1, getRemoteGraph());
-    }
-    for (int i = 0; i < graphInputNodeName_.size(); i++) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 2, graphInputNodeName_.getRaw(i));
-    }
-    for (int i = 0; i < graphOutputNodeName_.size(); i++) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 3, graphOutputNodeName_.getRaw(i));
-    }
-    if (!getExecutorNameBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 4, executorName_);
-    }
-    if (!serializedExecutorParameters_.isEmpty()) {
-      output.writeBytes(5, serializedExecutorParameters_);
-    }
-    for (int i = 0; i < defaultGraphInputTensorShape_.size(); i++) {
-      output.writeMessage(6, defaultGraphInputTensorShape_.get(i));
-    }
-    for (int i = 0; i < defaultGraphOutputTensorShape_.size(); i++) {
-      output.writeMessage(7, defaultGraphOutputTensorShape_.get(i));
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (remoteGraph_ != null) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(1, getRemoteGraph());
-    }
-    {
-      int dataSize = 0;
-      for (int i = 0; i < graphInputNodeName_.size(); i++) {
-        dataSize += computeStringSizeNoTag(graphInputNodeName_.getRaw(i));
-      }
-      size += dataSize;
-      size += 1 * getGraphInputNodeNameList().size();
-    }
-    {
-      int dataSize = 0;
-      for (int i = 0; i < graphOutputNodeName_.size(); i++) {
-        dataSize += computeStringSizeNoTag(graphOutputNodeName_.getRaw(i));
-      }
-      size += dataSize;
-      size += 1 * getGraphOutputNodeNameList().size();
-    }
-    if (!getExecutorNameBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(4, executorName_);
-    }
-    if (!serializedExecutorParameters_.isEmpty()) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeBytesSize(5, serializedExecutorParameters_);
-    }
-    for (int i = 0; i < defaultGraphInputTensorShape_.size(); i++) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(6, defaultGraphInputTensorShape_.get(i));
-    }
-    for (int i = 0; i < defaultGraphOutputTensorShape_.size(); i++) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(7, defaultGraphOutputTensorShape_.get(i));
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo other = (org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo) obj;
-
-    if (hasRemoteGraph() != other.hasRemoteGraph()) return false;
-    if (hasRemoteGraph()) {
-      if (!getRemoteGraph()
-          .equals(other.getRemoteGraph())) return false;
-    }
-    if (!getGraphInputNodeNameList()
-        .equals(other.getGraphInputNodeNameList())) return false;
-    if (!getGraphOutputNodeNameList()
-        .equals(other.getGraphOutputNodeNameList())) return false;
-    if (!getExecutorName()
-        .equals(other.getExecutorName())) return false;
-    if (!getSerializedExecutorParameters()
-        .equals(other.getSerializedExecutorParameters())) return false;
-    if (!getDefaultGraphInputTensorShapeList()
-        .equals(other.getDefaultGraphInputTensorShapeList())) return false;
-    if (!getDefaultGraphOutputTensorShapeList()
-        .equals(other.getDefaultGraphOutputTensorShapeList())) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    if (hasRemoteGraph()) {
-      hash = (37 * hash) + REMOTE_GRAPH_FIELD_NUMBER;
-      hash = (53 * hash) + getRemoteGraph().hashCode();
-    }
-    if (getGraphInputNodeNameCount() > 0) {
-      hash = (37 * hash) + GRAPH_INPUT_NODE_NAME_FIELD_NUMBER;
-      hash = (53 * hash) + getGraphInputNodeNameList().hashCode();
-    }
-    if (getGraphOutputNodeNameCount() > 0) {
-      hash = (37 * hash) + GRAPH_OUTPUT_NODE_NAME_FIELD_NUMBER;
-      hash = (53 * hash) + getGraphOutputNodeNameList().hashCode();
-    }
-    hash = (37 * hash) + EXECUTOR_NAME_FIELD_NUMBER;
-    hash = (53 * hash) + getExecutorName().hashCode();
-    hash = (37 * hash) + SERIALIZED_EXECUTOR_PARAMETERS_FIELD_NUMBER;
-    hash = (53 * hash) + getSerializedExecutorParameters().hashCode();
-    if (getDefaultGraphInputTensorShapeCount() > 0) {
-      hash = (37 * hash) + DEFAULT_GRAPH_INPUT_TENSOR_SHAPE_FIELD_NUMBER;
-      hash = (53 * hash) + getDefaultGraphInputTensorShapeList().hashCode();
-    }
-    if (getDefaultGraphOutputTensorShapeCount() > 0) {
-      hash = (37 * hash) + DEFAULT_GRAPH_OUTPUT_TENSOR_SHAPE_FIELD_NUMBER;
-      hash = (53 * hash) + getDefaultGraphOutputTensorShapeList().hashCode();
-    }
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * <pre>
-   * Protocol buffer representing a handle to a tensorflow resource. Handles are
-   * not valid across executions, but can be serialized back and forth from within
-   * a single run.
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.RemoteFusedGraphExecuteInfo}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.RemoteFusedGraphExecuteInfo)
-      org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfoOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfoProto.internal_static_tensorflow_RemoteFusedGraphExecuteInfo_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfoProto.internal_static_tensorflow_RemoteFusedGraphExecuteInfo_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.class, org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-        getDefaultGraphInputTensorShapeFieldBuilder();
-        getDefaultGraphOutputTensorShapeFieldBuilder();
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      if (remoteGraphBuilder_ == null) {
-        remoteGraph_ = null;
-      } else {
-        remoteGraph_ = null;
-        remoteGraphBuilder_ = null;
-      }
-      graphInputNodeName_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-      bitField0_ = (bitField0_ & ~0x00000001);
-      graphOutputNodeName_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-      bitField0_ = (bitField0_ & ~0x00000002);
-      executorName_ = "";
-
-      serializedExecutorParameters_ = com.google.protobuf.ByteString.EMPTY;
-
-      if (defaultGraphInputTensorShapeBuilder_ == null) {
-        defaultGraphInputTensorShape_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000004);
-      } else {
-        defaultGraphInputTensorShapeBuilder_.clear();
-      }
-      if (defaultGraphOutputTensorShapeBuilder_ == null) {
-        defaultGraphOutputTensorShape_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000008);
-      } else {
-        defaultGraphOutputTensorShapeBuilder_.clear();
-      }
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfoProto.internal_static_tensorflow_RemoteFusedGraphExecuteInfo_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo getDefaultInstanceForType() {
-      return org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo build() {
-      org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo buildPartial() {
-      org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo result = new org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo(this);
-      int from_bitField0_ = bitField0_;
-      if (remoteGraphBuilder_ == null) {
-        result.remoteGraph_ = remoteGraph_;
-      } else {
-        result.remoteGraph_ = remoteGraphBuilder_.build();
-      }
-      if (((bitField0_ & 0x00000001) != 0)) {
-        graphInputNodeName_ = graphInputNodeName_.getUnmodifiableView();
-        bitField0_ = (bitField0_ & ~0x00000001);
-      }
-      result.graphInputNodeName_ = graphInputNodeName_;
-      if (((bitField0_ & 0x00000002) != 0)) {
-        graphOutputNodeName_ = graphOutputNodeName_.getUnmodifiableView();
-        bitField0_ = (bitField0_ & ~0x00000002);
-      }
-      result.graphOutputNodeName_ = graphOutputNodeName_;
-      result.executorName_ = executorName_;
-      result.serializedExecutorParameters_ = serializedExecutorParameters_;
-      if (defaultGraphInputTensorShapeBuilder_ == null) {
-        if (((bitField0_ & 0x00000004) != 0)) {
-          defaultGraphInputTensorShape_ = java.util.Collections.unmodifiableList(defaultGraphInputTensorShape_);
-          bitField0_ = (bitField0_ & ~0x00000004);
-        }
-        result.defaultGraphInputTensorShape_ = defaultGraphInputTensorShape_;
-      } else {
-        result.defaultGraphInputTensorShape_ = defaultGraphInputTensorShapeBuilder_.build();
-      }
-      if (defaultGraphOutputTensorShapeBuilder_ == null) {
-        if (((bitField0_ & 0x00000008) != 0)) {
-          defaultGraphOutputTensorShape_ = java.util.Collections.unmodifiableList(defaultGraphOutputTensorShape_);
-          bitField0_ = (bitField0_ & ~0x00000008);
-        }
-        result.defaultGraphOutputTensorShape_ = defaultGraphOutputTensorShape_;
-      } else {
-        result.defaultGraphOutputTensorShape_ = defaultGraphOutputTensorShapeBuilder_.build();
-      }
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo) {
-        return mergeFrom((org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo other) {
-      if (other == org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.getDefaultInstance()) return this;
-      if (other.hasRemoteGraph()) {
-        mergeRemoteGraph(other.getRemoteGraph());
-      }
-      if (!other.graphInputNodeName_.isEmpty()) {
-        if (graphInputNodeName_.isEmpty()) {
-          graphInputNodeName_ = other.graphInputNodeName_;
-          bitField0_ = (bitField0_ & ~0x00000001);
-        } else {
-          ensureGraphInputNodeNameIsMutable();
-          graphInputNodeName_.addAll(other.graphInputNodeName_);
-        }
-        onChanged();
-      }
-      if (!other.graphOutputNodeName_.isEmpty()) {
-        if (graphOutputNodeName_.isEmpty()) {
-          graphOutputNodeName_ = other.graphOutputNodeName_;
-          bitField0_ = (bitField0_ & ~0x00000002);
-        } else {
-          ensureGraphOutputNodeNameIsMutable();
-          graphOutputNodeName_.addAll(other.graphOutputNodeName_);
-        }
-        onChanged();
-      }
-      if (!other.getExecutorName().isEmpty()) {
-        executorName_ = other.executorName_;
-        onChanged();
-      }
-      if (other.getSerializedExecutorParameters() != com.google.protobuf.ByteString.EMPTY) {
-        setSerializedExecutorParameters(other.getSerializedExecutorParameters());
-      }
-      if (defaultGraphInputTensorShapeBuilder_ == null) {
-        if (!other.defaultGraphInputTensorShape_.isEmpty()) {
-          if (defaultGraphInputTensorShape_.isEmpty()) {
-            defaultGraphInputTensorShape_ = other.defaultGraphInputTensorShape_;
-            bitField0_ = (bitField0_ & ~0x00000004);
-          } else {
-            ensureDefaultGraphInputTensorShapeIsMutable();
-            defaultGraphInputTensorShape_.addAll(other.defaultGraphInputTensorShape_);
-          }
-          onChanged();
-        }
-      } else {
-        if (!other.defaultGraphInputTensorShape_.isEmpty()) {
-          if (defaultGraphInputTensorShapeBuilder_.isEmpty()) {
-            defaultGraphInputTensorShapeBuilder_.dispose();
-            defaultGraphInputTensorShapeBuilder_ = null;
-            defaultGraphInputTensorShape_ = other.defaultGraphInputTensorShape_;
-            bitField0_ = (bitField0_ & ~0x00000004);
-            defaultGraphInputTensorShapeBuilder_ = 
-              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
-                 getDefaultGraphInputTensorShapeFieldBuilder() : null;
-          } else {
-            defaultGraphInputTensorShapeBuilder_.addAllMessages(other.defaultGraphInputTensorShape_);
-          }
-        }
-      }
-      if (defaultGraphOutputTensorShapeBuilder_ == null) {
-        if (!other.defaultGraphOutputTensorShape_.isEmpty()) {
-          if (defaultGraphOutputTensorShape_.isEmpty()) {
-            defaultGraphOutputTensorShape_ = other.defaultGraphOutputTensorShape_;
-            bitField0_ = (bitField0_ & ~0x00000008);
-          } else {
-            ensureDefaultGraphOutputTensorShapeIsMutable();
-            defaultGraphOutputTensorShape_.addAll(other.defaultGraphOutputTensorShape_);
-          }
-          onChanged();
-        }
-      } else {
-        if (!other.defaultGraphOutputTensorShape_.isEmpty()) {
-          if (defaultGraphOutputTensorShapeBuilder_.isEmpty()) {
-            defaultGraphOutputTensorShapeBuilder_.dispose();
-            defaultGraphOutputTensorShapeBuilder_ = null;
-            defaultGraphOutputTensorShape_ = other.defaultGraphOutputTensorShape_;
-            bitField0_ = (bitField0_ & ~0x00000008);
-            defaultGraphOutputTensorShapeBuilder_ = 
-              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
-                 getDefaultGraphOutputTensorShapeFieldBuilder() : null;
-          } else {
-            defaultGraphOutputTensorShapeBuilder_.addAllMessages(other.defaultGraphOutputTensorShape_);
-          }
-        }
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-    private int bitField0_;
-
-    private org.tensorflow.proto.framework.GraphDef remoteGraph_;
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.GraphDef, org.tensorflow.proto.framework.GraphDef.Builder, org.tensorflow.proto.framework.GraphDefOrBuilder> remoteGraphBuilder_;
-    /**
-     * <pre>
-     * Definition of remote graph
-     * </pre>
-     *
-     * <code>.tensorflow.GraphDef remote_graph = 1;</code>
-     */
-    public boolean hasRemoteGraph() {
-      return remoteGraphBuilder_ != null || remoteGraph_ != null;
-    }
-    /**
-     * <pre>
-     * Definition of remote graph
-     * </pre>
-     *
-     * <code>.tensorflow.GraphDef remote_graph = 1;</code>
-     */
-    public org.tensorflow.proto.framework.GraphDef getRemoteGraph() {
-      if (remoteGraphBuilder_ == null) {
-        return remoteGraph_ == null ? org.tensorflow.proto.framework.GraphDef.getDefaultInstance() : remoteGraph_;
-      } else {
-        return remoteGraphBuilder_.getMessage();
-      }
-    }
-    /**
-     * <pre>
-     * Definition of remote graph
-     * </pre>
-     *
-     * <code>.tensorflow.GraphDef remote_graph = 1;</code>
-     */
-    public Builder setRemoteGraph(org.tensorflow.proto.framework.GraphDef value) {
-      if (remoteGraphBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        remoteGraph_ = value;
-        onChanged();
-      } else {
-        remoteGraphBuilder_.setMessage(value);
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Definition of remote graph
-     * </pre>
-     *
-     * <code>.tensorflow.GraphDef remote_graph = 1;</code>
-     */
-    public Builder setRemoteGraph(
-        org.tensorflow.proto.framework.GraphDef.Builder builderForValue) {
-      if (remoteGraphBuilder_ == null) {
-        remoteGraph_ = builderForValue.build();
-        onChanged();
-      } else {
-        remoteGraphBuilder_.setMessage(builderForValue.build());
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Definition of remote graph
-     * </pre>
-     *
-     * <code>.tensorflow.GraphDef remote_graph = 1;</code>
-     */
-    public Builder mergeRemoteGraph(org.tensorflow.proto.framework.GraphDef value) {
-      if (remoteGraphBuilder_ == null) {
-        if (remoteGraph_ != null) {
-          remoteGraph_ =
-            org.tensorflow.proto.framework.GraphDef.newBuilder(remoteGraph_).mergeFrom(value).buildPartial();
-        } else {
-          remoteGraph_ = value;
-        }
-        onChanged();
-      } else {
-        remoteGraphBuilder_.mergeFrom(value);
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Definition of remote graph
-     * </pre>
-     *
-     * <code>.tensorflow.GraphDef remote_graph = 1;</code>
-     */
-    public Builder clearRemoteGraph() {
-      if (remoteGraphBuilder_ == null) {
-        remoteGraph_ = null;
-        onChanged();
-      } else {
-        remoteGraph_ = null;
-        remoteGraphBuilder_ = null;
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Definition of remote graph
-     * </pre>
-     *
-     * <code>.tensorflow.GraphDef remote_graph = 1;</code>
-     */
-    public org.tensorflow.proto.framework.GraphDef.Builder getRemoteGraphBuilder() {
-      
-      onChanged();
-      return getRemoteGraphFieldBuilder().getBuilder();
-    }
-    /**
-     * <pre>
-     * Definition of remote graph
-     * </pre>
-     *
-     * <code>.tensorflow.GraphDef remote_graph = 1;</code>
-     */
-    public org.tensorflow.proto.framework.GraphDefOrBuilder getRemoteGraphOrBuilder() {
-      if (remoteGraphBuilder_ != null) {
-        return remoteGraphBuilder_.getMessageOrBuilder();
-      } else {
-        return remoteGraph_ == null ?
-            org.tensorflow.proto.framework.GraphDef.getDefaultInstance() : remoteGraph_;
-      }
-    }
-    /**
-     * <pre>
-     * Definition of remote graph
-     * </pre>
-     *
-     * <code>.tensorflow.GraphDef remote_graph = 1;</code>
-     */
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.GraphDef, org.tensorflow.proto.framework.GraphDef.Builder, org.tensorflow.proto.framework.GraphDefOrBuilder> 
-        getRemoteGraphFieldBuilder() {
-      if (remoteGraphBuilder_ == null) {
-        remoteGraphBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-            org.tensorflow.proto.framework.GraphDef, org.tensorflow.proto.framework.GraphDef.Builder, org.tensorflow.proto.framework.GraphDefOrBuilder>(
-                getRemoteGraph(),
-                getParentForChildren(),
-                isClean());
-        remoteGraph_ = null;
-      }
-      return remoteGraphBuilder_;
-    }
-
-    private com.google.protobuf.LazyStringList graphInputNodeName_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-    private void ensureGraphInputNodeNameIsMutable() {
-      if (!((bitField0_ & 0x00000001) != 0)) {
-        graphInputNodeName_ = new com.google.protobuf.LazyStringArrayList(graphInputNodeName_);
-        bitField0_ |= 0x00000001;
-       }
-    }
-    /**
-     * <pre>
-     * Remote fused graph input node name
-     * </pre>
-     *
-     * <code>repeated string graph_input_node_name = 2;</code>
-     */
-    public com.google.protobuf.ProtocolStringList
-        getGraphInputNodeNameList() {
-      return graphInputNodeName_.getUnmodifiableView();
-    }
-    /**
-     * <pre>
-     * Remote fused graph input node name
-     * </pre>
-     *
-     * <code>repeated string graph_input_node_name = 2;</code>
-     */
-    public int getGraphInputNodeNameCount() {
-      return graphInputNodeName_.size();
-    }
-    /**
-     * <pre>
-     * Remote fused graph input node name
-     * </pre>
-     *
-     * <code>repeated string graph_input_node_name = 2;</code>
-     */
-    public java.lang.String getGraphInputNodeName(int index) {
-      return graphInputNodeName_.get(index);
-    }
-    /**
-     * <pre>
-     * Remote fused graph input node name
-     * </pre>
-     *
-     * <code>repeated string graph_input_node_name = 2;</code>
-     */
-    public com.google.protobuf.ByteString
-        getGraphInputNodeNameBytes(int index) {
-      return graphInputNodeName_.getByteString(index);
-    }
-    /**
-     * <pre>
-     * Remote fused graph input node name
-     * </pre>
-     *
-     * <code>repeated string graph_input_node_name = 2;</code>
-     */
-    public Builder setGraphInputNodeName(
-        int index, java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  ensureGraphInputNodeNameIsMutable();
-      graphInputNodeName_.set(index, value);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Remote fused graph input node name
-     * </pre>
-     *
-     * <code>repeated string graph_input_node_name = 2;</code>
-     */
-    public Builder addGraphInputNodeName(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  ensureGraphInputNodeNameIsMutable();
-      graphInputNodeName_.add(value);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Remote fused graph input node name
-     * </pre>
-     *
-     * <code>repeated string graph_input_node_name = 2;</code>
-     */
-    public Builder addAllGraphInputNodeName(
-        java.lang.Iterable<java.lang.String> values) {
-      ensureGraphInputNodeNameIsMutable();
-      com.google.protobuf.AbstractMessageLite.Builder.addAll(
-          values, graphInputNodeName_);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Remote fused graph input node name
-     * </pre>
-     *
-     * <code>repeated string graph_input_node_name = 2;</code>
-     */
-    public Builder clearGraphInputNodeName() {
-      graphInputNodeName_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-      bitField0_ = (bitField0_ & ~0x00000001);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Remote fused graph input node name
-     * </pre>
-     *
-     * <code>repeated string graph_input_node_name = 2;</code>
-     */
-    public Builder addGraphInputNodeNameBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      ensureGraphInputNodeNameIsMutable();
-      graphInputNodeName_.add(value);
-      onChanged();
-      return this;
-    }
-
-    private com.google.protobuf.LazyStringList graphOutputNodeName_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-    private void ensureGraphOutputNodeNameIsMutable() {
-      if (!((bitField0_ & 0x00000002) != 0)) {
-        graphOutputNodeName_ = new com.google.protobuf.LazyStringArrayList(graphOutputNodeName_);
-        bitField0_ |= 0x00000002;
-       }
-    }
-    /**
-     * <pre>
-     * Remote fused graph output node name
-     * </pre>
-     *
-     * <code>repeated string graph_output_node_name = 3;</code>
-     */
-    public com.google.protobuf.ProtocolStringList
-        getGraphOutputNodeNameList() {
-      return graphOutputNodeName_.getUnmodifiableView();
-    }
-    /**
-     * <pre>
-     * Remote fused graph output node name
-     * </pre>
-     *
-     * <code>repeated string graph_output_node_name = 3;</code>
-     */
-    public int getGraphOutputNodeNameCount() {
-      return graphOutputNodeName_.size();
-    }
-    /**
-     * <pre>
-     * Remote fused graph output node name
-     * </pre>
-     *
-     * <code>repeated string graph_output_node_name = 3;</code>
-     */
-    public java.lang.String getGraphOutputNodeName(int index) {
-      return graphOutputNodeName_.get(index);
-    }
-    /**
-     * <pre>
-     * Remote fused graph output node name
-     * </pre>
-     *
-     * <code>repeated string graph_output_node_name = 3;</code>
-     */
-    public com.google.protobuf.ByteString
-        getGraphOutputNodeNameBytes(int index) {
-      return graphOutputNodeName_.getByteString(index);
-    }
-    /**
-     * <pre>
-     * Remote fused graph output node name
-     * </pre>
-     *
-     * <code>repeated string graph_output_node_name = 3;</code>
-     */
-    public Builder setGraphOutputNodeName(
-        int index, java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  ensureGraphOutputNodeNameIsMutable();
-      graphOutputNodeName_.set(index, value);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Remote fused graph output node name
-     * </pre>
-     *
-     * <code>repeated string graph_output_node_name = 3;</code>
-     */
-    public Builder addGraphOutputNodeName(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  ensureGraphOutputNodeNameIsMutable();
-      graphOutputNodeName_.add(value);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Remote fused graph output node name
-     * </pre>
-     *
-     * <code>repeated string graph_output_node_name = 3;</code>
-     */
-    public Builder addAllGraphOutputNodeName(
-        java.lang.Iterable<java.lang.String> values) {
-      ensureGraphOutputNodeNameIsMutable();
-      com.google.protobuf.AbstractMessageLite.Builder.addAll(
-          values, graphOutputNodeName_);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Remote fused graph output node name
-     * </pre>
-     *
-     * <code>repeated string graph_output_node_name = 3;</code>
-     */
-    public Builder clearGraphOutputNodeName() {
-      graphOutputNodeName_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-      bitField0_ = (bitField0_ & ~0x00000002);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Remote fused graph output node name
-     * </pre>
-     *
-     * <code>repeated string graph_output_node_name = 3;</code>
-     */
-    public Builder addGraphOutputNodeNameBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      ensureGraphOutputNodeNameIsMutable();
-      graphOutputNodeName_.add(value);
-      onChanged();
-      return this;
-    }
-
-    private java.lang.Object executorName_ = "";
-    /**
-     * <pre>
-     * Executor's name
-     * </pre>
-     *
-     * <code>string executor_name = 4;</code>
-     */
-    public java.lang.String getExecutorName() {
-      java.lang.Object ref = executorName_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        executorName_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Executor's name
-     * </pre>
-     *
-     * <code>string executor_name = 4;</code>
-     */
-    public com.google.protobuf.ByteString
-        getExecutorNameBytes() {
-      java.lang.Object ref = executorName_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        executorName_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Executor's name
-     * </pre>
-     *
-     * <code>string executor_name = 4;</code>
-     */
-    public Builder setExecutorName(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      executorName_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Executor's name
-     * </pre>
-     *
-     * <code>string executor_name = 4;</code>
-     */
-    public Builder clearExecutorName() {
-      
-      executorName_ = getDefaultInstance().getExecutorName();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Executor's name
-     * </pre>
-     *
-     * <code>string executor_name = 4;</code>
-     */
-    public Builder setExecutorNameBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      executorName_ = value;
-      onChanged();
-      return this;
-    }
-
-    private com.google.protobuf.ByteString serializedExecutorParameters_ = com.google.protobuf.ByteString.EMPTY;
-    /**
-     * <pre>
-     * Optional: Parameters given to the executor
-     * </pre>
-     *
-     * <code>bytes serialized_executor_parameters = 5;</code>
-     */
-    public com.google.protobuf.ByteString getSerializedExecutorParameters() {
-      return serializedExecutorParameters_;
-    }
-    /**
-     * <pre>
-     * Optional: Parameters given to the executor
-     * </pre>
-     *
-     * <code>bytes serialized_executor_parameters = 5;</code>
-     */
-    public Builder setSerializedExecutorParameters(com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      serializedExecutorParameters_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Optional: Parameters given to the executor
-     * </pre>
-     *
-     * <code>bytes serialized_executor_parameters = 5;</code>
-     */
-    public Builder clearSerializedExecutorParameters() {
-      
-      serializedExecutorParameters_ = getDefaultInstance().getSerializedExecutorParameters();
-      onChanged();
-      return this;
-    }
-
-    private java.util.List<org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto> defaultGraphInputTensorShape_ =
-      java.util.Collections.emptyList();
-    private void ensureDefaultGraphInputTensorShapeIsMutable() {
-      if (!((bitField0_ & 0x00000004) != 0)) {
-        defaultGraphInputTensorShape_ = new java.util.ArrayList<org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto>(defaultGraphInputTensorShape_);
-        bitField0_ |= 0x00000004;
-       }
-    }
-
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto, org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto.Builder, org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProtoOrBuilder> defaultGraphInputTensorShapeBuilder_;
-
-    /**
-     * <pre>
-     * Optional: Default graph input tensor shape used to allocate memory
-     * before executing op
-     * </pre>
-     *
-     * <code>repeated .tensorflow.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto default_graph_input_tensor_shape = 6;</code>
-     */
-    public java.util.List<org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto> getDefaultGraphInputTensorShapeList() {
-      if (defaultGraphInputTensorShapeBuilder_ == null) {
-        return java.util.Collections.unmodifiableList(defaultGraphInputTensorShape_);
-      } else {
-        return defaultGraphInputTensorShapeBuilder_.getMessageList();
-      }
-    }
-    /**
-     * <pre>
-     * Optional: Default graph input tensor shape used to allocate memory
-     * before executing op
-     * </pre>
-     *
-     * <code>repeated .tensorflow.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto default_graph_input_tensor_shape = 6;</code>
-     */
-    public int getDefaultGraphInputTensorShapeCount() {
-      if (defaultGraphInputTensorShapeBuilder_ == null) {
-        return defaultGraphInputTensorShape_.size();
-      } else {
-        return defaultGraphInputTensorShapeBuilder_.getCount();
-      }
-    }
-    /**
-     * <pre>
-     * Optional: Default graph input tensor shape used to allocate memory
-     * before executing op
-     * </pre>
-     *
-     * <code>repeated .tensorflow.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto default_graph_input_tensor_shape = 6;</code>
-     */
-    public org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto getDefaultGraphInputTensorShape(int index) {
-      if (defaultGraphInputTensorShapeBuilder_ == null) {
-        return defaultGraphInputTensorShape_.get(index);
-      } else {
-        return defaultGraphInputTensorShapeBuilder_.getMessage(index);
-      }
-    }
-    /**
-     * <pre>
-     * Optional: Default graph input tensor shape used to allocate memory
-     * before executing op
-     * </pre>
-     *
-     * <code>repeated .tensorflow.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto default_graph_input_tensor_shape = 6;</code>
-     */
-    public Builder setDefaultGraphInputTensorShape(
-        int index, org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto value) {
-      if (defaultGraphInputTensorShapeBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureDefaultGraphInputTensorShapeIsMutable();
-        defaultGraphInputTensorShape_.set(index, value);
-        onChanged();
-      } else {
-        defaultGraphInputTensorShapeBuilder_.setMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Optional: Default graph input tensor shape used to allocate memory
-     * before executing op
-     * </pre>
-     *
-     * <code>repeated .tensorflow.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto default_graph_input_tensor_shape = 6;</code>
-     */
-    public Builder setDefaultGraphInputTensorShape(
-        int index, org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto.Builder builderForValue) {
-      if (defaultGraphInputTensorShapeBuilder_ == null) {
-        ensureDefaultGraphInputTensorShapeIsMutable();
-        defaultGraphInputTensorShape_.set(index, builderForValue.build());
-        onChanged();
-      } else {
-        defaultGraphInputTensorShapeBuilder_.setMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Optional: Default graph input tensor shape used to allocate memory
-     * before executing op
-     * </pre>
-     *
-     * <code>repeated .tensorflow.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto default_graph_input_tensor_shape = 6;</code>
-     */
-    public Builder addDefaultGraphInputTensorShape(org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto value) {
-      if (defaultGraphInputTensorShapeBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureDefaultGraphInputTensorShapeIsMutable();
-        defaultGraphInputTensorShape_.add(value);
-        onChanged();
-      } else {
-        defaultGraphInputTensorShapeBuilder_.addMessage(value);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Optional: Default graph input tensor shape used to allocate memory
-     * before executing op
-     * </pre>
-     *
-     * <code>repeated .tensorflow.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto default_graph_input_tensor_shape = 6;</code>
-     */
-    public Builder addDefaultGraphInputTensorShape(
-        int index, org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto value) {
-      if (defaultGraphInputTensorShapeBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureDefaultGraphInputTensorShapeIsMutable();
-        defaultGraphInputTensorShape_.add(index, value);
-        onChanged();
-      } else {
-        defaultGraphInputTensorShapeBuilder_.addMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Optional: Default graph input tensor shape used to allocate memory
-     * before executing op
-     * </pre>
-     *
-     * <code>repeated .tensorflow.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto default_graph_input_tensor_shape = 6;</code>
-     */
-    public Builder addDefaultGraphInputTensorShape(
-        org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto.Builder builderForValue) {
-      if (defaultGraphInputTensorShapeBuilder_ == null) {
-        ensureDefaultGraphInputTensorShapeIsMutable();
-        defaultGraphInputTensorShape_.add(builderForValue.build());
-        onChanged();
-      } else {
-        defaultGraphInputTensorShapeBuilder_.addMessage(builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Optional: Default graph input tensor shape used to allocate memory
-     * before executing op
-     * </pre>
-     *
-     * <code>repeated .tensorflow.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto default_graph_input_tensor_shape = 6;</code>
-     */
-    public Builder addDefaultGraphInputTensorShape(
-        int index, org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto.Builder builderForValue) {
-      if (defaultGraphInputTensorShapeBuilder_ == null) {
-        ensureDefaultGraphInputTensorShapeIsMutable();
-        defaultGraphInputTensorShape_.add(index, builderForValue.build());
-        onChanged();
-      } else {
-        defaultGraphInputTensorShapeBuilder_.addMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Optional: Default graph input tensor shape used to allocate memory
-     * before executing op
-     * </pre>
-     *
-     * <code>repeated .tensorflow.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto default_graph_input_tensor_shape = 6;</code>
-     */
-    public Builder addAllDefaultGraphInputTensorShape(
-        java.lang.Iterable<? extends org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto> values) {
-      if (defaultGraphInputTensorShapeBuilder_ == null) {
-        ensureDefaultGraphInputTensorShapeIsMutable();
-        com.google.protobuf.AbstractMessageLite.Builder.addAll(
-            values, defaultGraphInputTensorShape_);
-        onChanged();
-      } else {
-        defaultGraphInputTensorShapeBuilder_.addAllMessages(values);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Optional: Default graph input tensor shape used to allocate memory
-     * before executing op
-     * </pre>
-     *
-     * <code>repeated .tensorflow.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto default_graph_input_tensor_shape = 6;</code>
-     */
-    public Builder clearDefaultGraphInputTensorShape() {
-      if (defaultGraphInputTensorShapeBuilder_ == null) {
-        defaultGraphInputTensorShape_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000004);
-        onChanged();
-      } else {
-        defaultGraphInputTensorShapeBuilder_.clear();
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Optional: Default graph input tensor shape used to allocate memory
-     * before executing op
-     * </pre>
-     *
-     * <code>repeated .tensorflow.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto default_graph_input_tensor_shape = 6;</code>
-     */
-    public Builder removeDefaultGraphInputTensorShape(int index) {
-      if (defaultGraphInputTensorShapeBuilder_ == null) {
-        ensureDefaultGraphInputTensorShapeIsMutable();
-        defaultGraphInputTensorShape_.remove(index);
-        onChanged();
-      } else {
-        defaultGraphInputTensorShapeBuilder_.remove(index);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Optional: Default graph input tensor shape used to allocate memory
-     * before executing op
-     * </pre>
-     *
-     * <code>repeated .tensorflow.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto default_graph_input_tensor_shape = 6;</code>
-     */
-    public org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto.Builder getDefaultGraphInputTensorShapeBuilder(
-        int index) {
-      return getDefaultGraphInputTensorShapeFieldBuilder().getBuilder(index);
-    }
-    /**
-     * <pre>
-     * Optional: Default graph input tensor shape used to allocate memory
-     * before executing op
-     * </pre>
-     *
-     * <code>repeated .tensorflow.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto default_graph_input_tensor_shape = 6;</code>
-     */
-    public org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProtoOrBuilder getDefaultGraphInputTensorShapeOrBuilder(
-        int index) {
-      if (defaultGraphInputTensorShapeBuilder_ == null) {
-        return defaultGraphInputTensorShape_.get(index);  } else {
-        return defaultGraphInputTensorShapeBuilder_.getMessageOrBuilder(index);
-      }
-    }
-    /**
-     * <pre>
-     * Optional: Default graph input tensor shape used to allocate memory
-     * before executing op
-     * </pre>
-     *
-     * <code>repeated .tensorflow.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto default_graph_input_tensor_shape = 6;</code>
-     */
-    public java.util.List<? extends org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProtoOrBuilder> 
-         getDefaultGraphInputTensorShapeOrBuilderList() {
-      if (defaultGraphInputTensorShapeBuilder_ != null) {
-        return defaultGraphInputTensorShapeBuilder_.getMessageOrBuilderList();
-      } else {
-        return java.util.Collections.unmodifiableList(defaultGraphInputTensorShape_);
-      }
-    }
-    /**
-     * <pre>
-     * Optional: Default graph input tensor shape used to allocate memory
-     * before executing op
-     * </pre>
-     *
-     * <code>repeated .tensorflow.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto default_graph_input_tensor_shape = 6;</code>
-     */
-    public org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto.Builder addDefaultGraphInputTensorShapeBuilder() {
-      return getDefaultGraphInputTensorShapeFieldBuilder().addBuilder(
-          org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto.getDefaultInstance());
-    }
-    /**
-     * <pre>
-     * Optional: Default graph input tensor shape used to allocate memory
-     * before executing op
-     * </pre>
-     *
-     * <code>repeated .tensorflow.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto default_graph_input_tensor_shape = 6;</code>
-     */
-    public org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto.Builder addDefaultGraphInputTensorShapeBuilder(
-        int index) {
-      return getDefaultGraphInputTensorShapeFieldBuilder().addBuilder(
-          index, org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto.getDefaultInstance());
-    }
-    /**
-     * <pre>
-     * Optional: Default graph input tensor shape used to allocate memory
-     * before executing op
-     * </pre>
-     *
-     * <code>repeated .tensorflow.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto default_graph_input_tensor_shape = 6;</code>
-     */
-    public java.util.List<org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto.Builder> 
-         getDefaultGraphInputTensorShapeBuilderList() {
-      return getDefaultGraphInputTensorShapeFieldBuilder().getBuilderList();
-    }
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto, org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto.Builder, org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProtoOrBuilder> 
-        getDefaultGraphInputTensorShapeFieldBuilder() {
-      if (defaultGraphInputTensorShapeBuilder_ == null) {
-        defaultGraphInputTensorShapeBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
-            org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto, org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto.Builder, org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProtoOrBuilder>(
-                defaultGraphInputTensorShape_,
-                ((bitField0_ & 0x00000004) != 0),
-                getParentForChildren(),
-                isClean());
-        defaultGraphInputTensorShape_ = null;
-      }
-      return defaultGraphInputTensorShapeBuilder_;
-    }
-
-    private java.util.List<org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto> defaultGraphOutputTensorShape_ =
-      java.util.Collections.emptyList();
-    private void ensureDefaultGraphOutputTensorShapeIsMutable() {
-      if (!((bitField0_ & 0x00000008) != 0)) {
-        defaultGraphOutputTensorShape_ = new java.util.ArrayList<org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto>(defaultGraphOutputTensorShape_);
-        bitField0_ |= 0x00000008;
-       }
-    }
-
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto, org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto.Builder, org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProtoOrBuilder> defaultGraphOutputTensorShapeBuilder_;
-
-    /**
-     * <pre>
-     * Optional: Default graph input tensor shape used to allocate memory
-     * before executing op
-     * TODO(satok): Remote output tensor shape once shape information is stored
-     * in NodeDef
-     * </pre>
-     *
-     * <code>repeated .tensorflow.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto default_graph_output_tensor_shape = 7;</code>
-     */
-    public java.util.List<org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto> getDefaultGraphOutputTensorShapeList() {
-      if (defaultGraphOutputTensorShapeBuilder_ == null) {
-        return java.util.Collections.unmodifiableList(defaultGraphOutputTensorShape_);
-      } else {
-        return defaultGraphOutputTensorShapeBuilder_.getMessageList();
-      }
-    }
-    /**
-     * <pre>
-     * Optional: Default graph input tensor shape used to allocate memory
-     * before executing op
-     * TODO(satok): Remote output tensor shape once shape information is stored
-     * in NodeDef
-     * </pre>
-     *
-     * <code>repeated .tensorflow.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto default_graph_output_tensor_shape = 7;</code>
-     */
-    public int getDefaultGraphOutputTensorShapeCount() {
-      if (defaultGraphOutputTensorShapeBuilder_ == null) {
-        return defaultGraphOutputTensorShape_.size();
-      } else {
-        return defaultGraphOutputTensorShapeBuilder_.getCount();
-      }
-    }
-    /**
-     * <pre>
-     * Optional: Default graph input tensor shape used to allocate memory
-     * before executing op
-     * TODO(satok): Remote output tensor shape once shape information is stored
-     * in NodeDef
-     * </pre>
-     *
-     * <code>repeated .tensorflow.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto default_graph_output_tensor_shape = 7;</code>
-     */
-    public org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto getDefaultGraphOutputTensorShape(int index) {
-      if (defaultGraphOutputTensorShapeBuilder_ == null) {
-        return defaultGraphOutputTensorShape_.get(index);
-      } else {
-        return defaultGraphOutputTensorShapeBuilder_.getMessage(index);
-      }
-    }
-    /**
-     * <pre>
-     * Optional: Default graph input tensor shape used to allocate memory
-     * before executing op
-     * TODO(satok): Remote output tensor shape once shape information is stored
-     * in NodeDef
-     * </pre>
-     *
-     * <code>repeated .tensorflow.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto default_graph_output_tensor_shape = 7;</code>
-     */
-    public Builder setDefaultGraphOutputTensorShape(
-        int index, org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto value) {
-      if (defaultGraphOutputTensorShapeBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureDefaultGraphOutputTensorShapeIsMutable();
-        defaultGraphOutputTensorShape_.set(index, value);
-        onChanged();
-      } else {
-        defaultGraphOutputTensorShapeBuilder_.setMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Optional: Default graph input tensor shape used to allocate memory
-     * before executing op
-     * TODO(satok): Remote output tensor shape once shape information is stored
-     * in NodeDef
-     * </pre>
-     *
-     * <code>repeated .tensorflow.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto default_graph_output_tensor_shape = 7;</code>
-     */
-    public Builder setDefaultGraphOutputTensorShape(
-        int index, org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto.Builder builderForValue) {
-      if (defaultGraphOutputTensorShapeBuilder_ == null) {
-        ensureDefaultGraphOutputTensorShapeIsMutable();
-        defaultGraphOutputTensorShape_.set(index, builderForValue.build());
-        onChanged();
-      } else {
-        defaultGraphOutputTensorShapeBuilder_.setMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Optional: Default graph input tensor shape used to allocate memory
-     * before executing op
-     * TODO(satok): Remote output tensor shape once shape information is stored
-     * in NodeDef
-     * </pre>
-     *
-     * <code>repeated .tensorflow.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto default_graph_output_tensor_shape = 7;</code>
-     */
-    public Builder addDefaultGraphOutputTensorShape(org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto value) {
-      if (defaultGraphOutputTensorShapeBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureDefaultGraphOutputTensorShapeIsMutable();
-        defaultGraphOutputTensorShape_.add(value);
-        onChanged();
-      } else {
-        defaultGraphOutputTensorShapeBuilder_.addMessage(value);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Optional: Default graph input tensor shape used to allocate memory
-     * before executing op
-     * TODO(satok): Remote output tensor shape once shape information is stored
-     * in NodeDef
-     * </pre>
-     *
-     * <code>repeated .tensorflow.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto default_graph_output_tensor_shape = 7;</code>
-     */
-    public Builder addDefaultGraphOutputTensorShape(
-        int index, org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto value) {
-      if (defaultGraphOutputTensorShapeBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureDefaultGraphOutputTensorShapeIsMutable();
-        defaultGraphOutputTensorShape_.add(index, value);
-        onChanged();
-      } else {
-        defaultGraphOutputTensorShapeBuilder_.addMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Optional: Default graph input tensor shape used to allocate memory
-     * before executing op
-     * TODO(satok): Remote output tensor shape once shape information is stored
-     * in NodeDef
-     * </pre>
-     *
-     * <code>repeated .tensorflow.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto default_graph_output_tensor_shape = 7;</code>
-     */
-    public Builder addDefaultGraphOutputTensorShape(
-        org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto.Builder builderForValue) {
-      if (defaultGraphOutputTensorShapeBuilder_ == null) {
-        ensureDefaultGraphOutputTensorShapeIsMutable();
-        defaultGraphOutputTensorShape_.add(builderForValue.build());
-        onChanged();
-      } else {
-        defaultGraphOutputTensorShapeBuilder_.addMessage(builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Optional: Default graph input tensor shape used to allocate memory
-     * before executing op
-     * TODO(satok): Remote output tensor shape once shape information is stored
-     * in NodeDef
-     * </pre>
-     *
-     * <code>repeated .tensorflow.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto default_graph_output_tensor_shape = 7;</code>
-     */
-    public Builder addDefaultGraphOutputTensorShape(
-        int index, org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto.Builder builderForValue) {
-      if (defaultGraphOutputTensorShapeBuilder_ == null) {
-        ensureDefaultGraphOutputTensorShapeIsMutable();
-        defaultGraphOutputTensorShape_.add(index, builderForValue.build());
-        onChanged();
-      } else {
-        defaultGraphOutputTensorShapeBuilder_.addMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Optional: Default graph input tensor shape used to allocate memory
-     * before executing op
-     * TODO(satok): Remote output tensor shape once shape information is stored
-     * in NodeDef
-     * </pre>
-     *
-     * <code>repeated .tensorflow.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto default_graph_output_tensor_shape = 7;</code>
-     */
-    public Builder addAllDefaultGraphOutputTensorShape(
-        java.lang.Iterable<? extends org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto> values) {
-      if (defaultGraphOutputTensorShapeBuilder_ == null) {
-        ensureDefaultGraphOutputTensorShapeIsMutable();
-        com.google.protobuf.AbstractMessageLite.Builder.addAll(
-            values, defaultGraphOutputTensorShape_);
-        onChanged();
-      } else {
-        defaultGraphOutputTensorShapeBuilder_.addAllMessages(values);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Optional: Default graph input tensor shape used to allocate memory
-     * before executing op
-     * TODO(satok): Remote output tensor shape once shape information is stored
-     * in NodeDef
-     * </pre>
-     *
-     * <code>repeated .tensorflow.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto default_graph_output_tensor_shape = 7;</code>
-     */
-    public Builder clearDefaultGraphOutputTensorShape() {
-      if (defaultGraphOutputTensorShapeBuilder_ == null) {
-        defaultGraphOutputTensorShape_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000008);
-        onChanged();
-      } else {
-        defaultGraphOutputTensorShapeBuilder_.clear();
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Optional: Default graph input tensor shape used to allocate memory
-     * before executing op
-     * TODO(satok): Remote output tensor shape once shape information is stored
-     * in NodeDef
-     * </pre>
-     *
-     * <code>repeated .tensorflow.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto default_graph_output_tensor_shape = 7;</code>
-     */
-    public Builder removeDefaultGraphOutputTensorShape(int index) {
-      if (defaultGraphOutputTensorShapeBuilder_ == null) {
-        ensureDefaultGraphOutputTensorShapeIsMutable();
-        defaultGraphOutputTensorShape_.remove(index);
-        onChanged();
-      } else {
-        defaultGraphOutputTensorShapeBuilder_.remove(index);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Optional: Default graph input tensor shape used to allocate memory
-     * before executing op
-     * TODO(satok): Remote output tensor shape once shape information is stored
-     * in NodeDef
-     * </pre>
-     *
-     * <code>repeated .tensorflow.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto default_graph_output_tensor_shape = 7;</code>
-     */
-    public org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto.Builder getDefaultGraphOutputTensorShapeBuilder(
-        int index) {
-      return getDefaultGraphOutputTensorShapeFieldBuilder().getBuilder(index);
-    }
-    /**
-     * <pre>
-     * Optional: Default graph input tensor shape used to allocate memory
-     * before executing op
-     * TODO(satok): Remote output tensor shape once shape information is stored
-     * in NodeDef
-     * </pre>
-     *
-     * <code>repeated .tensorflow.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto default_graph_output_tensor_shape = 7;</code>
-     */
-    public org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProtoOrBuilder getDefaultGraphOutputTensorShapeOrBuilder(
-        int index) {
-      if (defaultGraphOutputTensorShapeBuilder_ == null) {
-        return defaultGraphOutputTensorShape_.get(index);  } else {
-        return defaultGraphOutputTensorShapeBuilder_.getMessageOrBuilder(index);
-      }
-    }
-    /**
-     * <pre>
-     * Optional: Default graph input tensor shape used to allocate memory
-     * before executing op
-     * TODO(satok): Remote output tensor shape once shape information is stored
-     * in NodeDef
-     * </pre>
-     *
-     * <code>repeated .tensorflow.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto default_graph_output_tensor_shape = 7;</code>
-     */
-    public java.util.List<? extends org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProtoOrBuilder> 
-         getDefaultGraphOutputTensorShapeOrBuilderList() {
-      if (defaultGraphOutputTensorShapeBuilder_ != null) {
-        return defaultGraphOutputTensorShapeBuilder_.getMessageOrBuilderList();
-      } else {
-        return java.util.Collections.unmodifiableList(defaultGraphOutputTensorShape_);
-      }
-    }
-    /**
-     * <pre>
-     * Optional: Default graph input tensor shape used to allocate memory
-     * before executing op
-     * TODO(satok): Remote output tensor shape once shape information is stored
-     * in NodeDef
-     * </pre>
-     *
-     * <code>repeated .tensorflow.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto default_graph_output_tensor_shape = 7;</code>
-     */
-    public org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto.Builder addDefaultGraphOutputTensorShapeBuilder() {
-      return getDefaultGraphOutputTensorShapeFieldBuilder().addBuilder(
-          org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto.getDefaultInstance());
-    }
-    /**
-     * <pre>
-     * Optional: Default graph input tensor shape used to allocate memory
-     * before executing op
-     * TODO(satok): Remote output tensor shape once shape information is stored
-     * in NodeDef
-     * </pre>
-     *
-     * <code>repeated .tensorflow.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto default_graph_output_tensor_shape = 7;</code>
-     */
-    public org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto.Builder addDefaultGraphOutputTensorShapeBuilder(
-        int index) {
-      return getDefaultGraphOutputTensorShapeFieldBuilder().addBuilder(
-          index, org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto.getDefaultInstance());
-    }
-    /**
-     * <pre>
-     * Optional: Default graph input tensor shape used to allocate memory
-     * before executing op
-     * TODO(satok): Remote output tensor shape once shape information is stored
-     * in NodeDef
-     * </pre>
-     *
-     * <code>repeated .tensorflow.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto default_graph_output_tensor_shape = 7;</code>
-     */
-    public java.util.List<org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto.Builder> 
-         getDefaultGraphOutputTensorShapeBuilderList() {
-      return getDefaultGraphOutputTensorShapeFieldBuilder().getBuilderList();
-    }
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto, org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto.Builder, org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProtoOrBuilder> 
-        getDefaultGraphOutputTensorShapeFieldBuilder() {
-      if (defaultGraphOutputTensorShapeBuilder_ == null) {
-        defaultGraphOutputTensorShapeBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
-            org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto, org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto.Builder, org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProtoOrBuilder>(
-                defaultGraphOutputTensorShape_,
-                ((bitField0_ & 0x00000008) != 0),
-                getParentForChildren(),
-                isClean());
-        defaultGraphOutputTensorShape_ = null;
-      }
-      return defaultGraphOutputTensorShapeBuilder_;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.RemoteFusedGraphExecuteInfo)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.RemoteFusedGraphExecuteInfo)
-  private static final org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo();
-  }
-
-  public static org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<RemoteFusedGraphExecuteInfo>
-      PARSER = new com.google.protobuf.AbstractParser<RemoteFusedGraphExecuteInfo>() {
-    @java.lang.Override
-    public RemoteFusedGraphExecuteInfo parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new RemoteFusedGraphExecuteInfo(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<RemoteFusedGraphExecuteInfo> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<RemoteFusedGraphExecuteInfo> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/RemoteFusedGraphExecuteInfoOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/RemoteFusedGraphExecuteInfoOrBuilder.java
deleted file mode 100644
index 6e14ad8a2e0..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/RemoteFusedGraphExecuteInfoOrBuilder.java
+++ /dev/null
@@ -1,239 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/framework/remote_fused_graph_execute_info.proto
-
-package org.tensorflow.proto.framework;
-
-public interface RemoteFusedGraphExecuteInfoOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.RemoteFusedGraphExecuteInfo)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <pre>
-   * Definition of remote graph
-   * </pre>
-   *
-   * <code>.tensorflow.GraphDef remote_graph = 1;</code>
-   */
-  boolean hasRemoteGraph();
-  /**
-   * <pre>
-   * Definition of remote graph
-   * </pre>
-   *
-   * <code>.tensorflow.GraphDef remote_graph = 1;</code>
-   */
-  org.tensorflow.proto.framework.GraphDef getRemoteGraph();
-  /**
-   * <pre>
-   * Definition of remote graph
-   * </pre>
-   *
-   * <code>.tensorflow.GraphDef remote_graph = 1;</code>
-   */
-  org.tensorflow.proto.framework.GraphDefOrBuilder getRemoteGraphOrBuilder();
-
-  /**
-   * <pre>
-   * Remote fused graph input node name
-   * </pre>
-   *
-   * <code>repeated string graph_input_node_name = 2;</code>
-   */
-  java.util.List<java.lang.String>
-      getGraphInputNodeNameList();
-  /**
-   * <pre>
-   * Remote fused graph input node name
-   * </pre>
-   *
-   * <code>repeated string graph_input_node_name = 2;</code>
-   */
-  int getGraphInputNodeNameCount();
-  /**
-   * <pre>
-   * Remote fused graph input node name
-   * </pre>
-   *
-   * <code>repeated string graph_input_node_name = 2;</code>
-   */
-  java.lang.String getGraphInputNodeName(int index);
-  /**
-   * <pre>
-   * Remote fused graph input node name
-   * </pre>
-   *
-   * <code>repeated string graph_input_node_name = 2;</code>
-   */
-  com.google.protobuf.ByteString
-      getGraphInputNodeNameBytes(int index);
-
-  /**
-   * <pre>
-   * Remote fused graph output node name
-   * </pre>
-   *
-   * <code>repeated string graph_output_node_name = 3;</code>
-   */
-  java.util.List<java.lang.String>
-      getGraphOutputNodeNameList();
-  /**
-   * <pre>
-   * Remote fused graph output node name
-   * </pre>
-   *
-   * <code>repeated string graph_output_node_name = 3;</code>
-   */
-  int getGraphOutputNodeNameCount();
-  /**
-   * <pre>
-   * Remote fused graph output node name
-   * </pre>
-   *
-   * <code>repeated string graph_output_node_name = 3;</code>
-   */
-  java.lang.String getGraphOutputNodeName(int index);
-  /**
-   * <pre>
-   * Remote fused graph output node name
-   * </pre>
-   *
-   * <code>repeated string graph_output_node_name = 3;</code>
-   */
-  com.google.protobuf.ByteString
-      getGraphOutputNodeNameBytes(int index);
-
-  /**
-   * <pre>
-   * Executor's name
-   * </pre>
-   *
-   * <code>string executor_name = 4;</code>
-   */
-  java.lang.String getExecutorName();
-  /**
-   * <pre>
-   * Executor's name
-   * </pre>
-   *
-   * <code>string executor_name = 4;</code>
-   */
-  com.google.protobuf.ByteString
-      getExecutorNameBytes();
-
-  /**
-   * <pre>
-   * Optional: Parameters given to the executor
-   * </pre>
-   *
-   * <code>bytes serialized_executor_parameters = 5;</code>
-   */
-  com.google.protobuf.ByteString getSerializedExecutorParameters();
-
-  /**
-   * <pre>
-   * Optional: Default graph input tensor shape used to allocate memory
-   * before executing op
-   * </pre>
-   *
-   * <code>repeated .tensorflow.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto default_graph_input_tensor_shape = 6;</code>
-   */
-  java.util.List<org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto> 
-      getDefaultGraphInputTensorShapeList();
-  /**
-   * <pre>
-   * Optional: Default graph input tensor shape used to allocate memory
-   * before executing op
-   * </pre>
-   *
-   * <code>repeated .tensorflow.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto default_graph_input_tensor_shape = 6;</code>
-   */
-  org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto getDefaultGraphInputTensorShape(int index);
-  /**
-   * <pre>
-   * Optional: Default graph input tensor shape used to allocate memory
-   * before executing op
-   * </pre>
-   *
-   * <code>repeated .tensorflow.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto default_graph_input_tensor_shape = 6;</code>
-   */
-  int getDefaultGraphInputTensorShapeCount();
-  /**
-   * <pre>
-   * Optional: Default graph input tensor shape used to allocate memory
-   * before executing op
-   * </pre>
-   *
-   * <code>repeated .tensorflow.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto default_graph_input_tensor_shape = 6;</code>
-   */
-  java.util.List<? extends org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProtoOrBuilder> 
-      getDefaultGraphInputTensorShapeOrBuilderList();
-  /**
-   * <pre>
-   * Optional: Default graph input tensor shape used to allocate memory
-   * before executing op
-   * </pre>
-   *
-   * <code>repeated .tensorflow.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto default_graph_input_tensor_shape = 6;</code>
-   */
-  org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProtoOrBuilder getDefaultGraphInputTensorShapeOrBuilder(
-      int index);
-
-  /**
-   * <pre>
-   * Optional: Default graph input tensor shape used to allocate memory
-   * before executing op
-   * TODO(satok): Remote output tensor shape once shape information is stored
-   * in NodeDef
-   * </pre>
-   *
-   * <code>repeated .tensorflow.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto default_graph_output_tensor_shape = 7;</code>
-   */
-  java.util.List<org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto> 
-      getDefaultGraphOutputTensorShapeList();
-  /**
-   * <pre>
-   * Optional: Default graph input tensor shape used to allocate memory
-   * before executing op
-   * TODO(satok): Remote output tensor shape once shape information is stored
-   * in NodeDef
-   * </pre>
-   *
-   * <code>repeated .tensorflow.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto default_graph_output_tensor_shape = 7;</code>
-   */
-  org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto getDefaultGraphOutputTensorShape(int index);
-  /**
-   * <pre>
-   * Optional: Default graph input tensor shape used to allocate memory
-   * before executing op
-   * TODO(satok): Remote output tensor shape once shape information is stored
-   * in NodeDef
-   * </pre>
-   *
-   * <code>repeated .tensorflow.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto default_graph_output_tensor_shape = 7;</code>
-   */
-  int getDefaultGraphOutputTensorShapeCount();
-  /**
-   * <pre>
-   * Optional: Default graph input tensor shape used to allocate memory
-   * before executing op
-   * TODO(satok): Remote output tensor shape once shape information is stored
-   * in NodeDef
-   * </pre>
-   *
-   * <code>repeated .tensorflow.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto default_graph_output_tensor_shape = 7;</code>
-   */
-  java.util.List<? extends org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProtoOrBuilder> 
-      getDefaultGraphOutputTensorShapeOrBuilderList();
-  /**
-   * <pre>
-   * Optional: Default graph input tensor shape used to allocate memory
-   * before executing op
-   * TODO(satok): Remote output tensor shape once shape information is stored
-   * in NodeDef
-   * </pre>
-   *
-   * <code>repeated .tensorflow.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto default_graph_output_tensor_shape = 7;</code>
-   */
-  org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProtoOrBuilder getDefaultGraphOutputTensorShapeOrBuilder(
-      int index);
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/RemoteFusedGraphExecuteInfoProto.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/RemoteFusedGraphExecuteInfoProto.java
deleted file mode 100644
index 95aeddaf025..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/RemoteFusedGraphExecuteInfoProto.java
+++ /dev/null
@@ -1,85 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/framework/remote_fused_graph_execute_info.proto
-
-package org.tensorflow.proto.framework;
-
-public final class RemoteFusedGraphExecuteInfoProto {
-  private RemoteFusedGraphExecuteInfoProto() {}
-  public static void registerAllExtensions(
-      com.google.protobuf.ExtensionRegistryLite registry) {
-  }
-
-  public static void registerAllExtensions(
-      com.google.protobuf.ExtensionRegistry registry) {
-    registerAllExtensions(
-        (com.google.protobuf.ExtensionRegistryLite) registry);
-  }
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_RemoteFusedGraphExecuteInfo_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_RemoteFusedGraphExecuteInfo_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_RemoteFusedGraphExecuteInfo_TensorShapeTypeProto_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_RemoteFusedGraphExecuteInfo_TensorShapeTypeProto_fieldAccessorTable;
-
-  public static com.google.protobuf.Descriptors.FileDescriptor
-      getDescriptor() {
-    return descriptor;
-  }
-  private static  com.google.protobuf.Descriptors.FileDescriptor
-      descriptor;
-  static {
-    java.lang.String[] descriptorData = {
-      "\n?tensorflow/core/framework/remote_fused" +
-      "_graph_execute_info.proto\022\ntensorflow\032%t" +
-      "ensorflow/core/framework/graph.proto\032,te" +
-      "nsorflow/core/framework/tensor_shape.pro" +
-      "to\032%tensorflow/core/framework/types.prot" +
-      "o\"\202\004\n\033RemoteFusedGraphExecuteInfo\022*\n\014rem" +
-      "ote_graph\030\001 \001(\0132\024.tensorflow.GraphDef\022\035\n" +
-      "\025graph_input_node_name\030\002 \003(\t\022\036\n\026graph_ou" +
-      "tput_node_name\030\003 \003(\t\022\025\n\rexecutor_name\030\004 " +
-      "\001(\t\022&\n\036serialized_executor_parameters\030\005 " +
-      "\001(\014\022f\n default_graph_input_tensor_shape\030" +
-      "\006 \003(\0132<.tensorflow.RemoteFusedGraphExecu" +
-      "teInfo.TensorShapeTypeProto\022g\n!default_g" +
-      "raph_output_tensor_shape\030\007 \003(\0132<.tensorf" +
-      "low.RemoteFusedGraphExecuteInfo.TensorSh" +
-      "apeTypeProto\032h\n\024TensorShapeTypeProto\022#\n\005" +
-      "dtype\030\001 \001(\0162\024.tensorflow.DataType\022+\n\005sha" +
-      "pe\030\002 \001(\0132\034.tensorflow.TensorShapeProtoB\257" +
-      "\001\n\036org.tensorflow.proto.frameworkB Remot" +
-      "eFusedGraphExecuteInfoProtoP\001Zfgithub.co" +
-      "m/tensorflow/tensorflow/tensorflow/go/co" +
-      "re/framework/remote_fused_graph_execute_" +
-      "info_go_proto\370\001\001b\006proto3"
-    };
-    descriptor = com.google.protobuf.Descriptors.FileDescriptor
-      .internalBuildGeneratedFileFrom(descriptorData,
-        new com.google.protobuf.Descriptors.FileDescriptor[] {
-          org.tensorflow.proto.framework.GraphProtos.getDescriptor(),
-          org.tensorflow.proto.framework.TensorShapeProtos.getDescriptor(),
-          org.tensorflow.proto.framework.TypesProtos.getDescriptor(),
-        });
-    internal_static_tensorflow_RemoteFusedGraphExecuteInfo_descriptor =
-      getDescriptor().getMessageTypes().get(0);
-    internal_static_tensorflow_RemoteFusedGraphExecuteInfo_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_RemoteFusedGraphExecuteInfo_descriptor,
-        new java.lang.String[] { "RemoteGraph", "GraphInputNodeName", "GraphOutputNodeName", "ExecutorName", "SerializedExecutorParameters", "DefaultGraphInputTensorShape", "DefaultGraphOutputTensorShape", });
-    internal_static_tensorflow_RemoteFusedGraphExecuteInfo_TensorShapeTypeProto_descriptor =
-      internal_static_tensorflow_RemoteFusedGraphExecuteInfo_descriptor.getNestedTypes().get(0);
-    internal_static_tensorflow_RemoteFusedGraphExecuteInfo_TensorShapeTypeProto_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_RemoteFusedGraphExecuteInfo_TensorShapeTypeProto_descriptor,
-        new java.lang.String[] { "Dtype", "Shape", });
-    org.tensorflow.proto.framework.GraphProtos.getDescriptor();
-    org.tensorflow.proto.framework.TensorShapeProtos.getDescriptor();
-    org.tensorflow.proto.framework.TypesProtos.getDescriptor();
-  }
-
-  // @@protoc_insertion_point(outer_class_scope)
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/RemoteTensorHandle.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/RemoteTensorHandle.java
deleted file mode 100644
index 85a3d489f33..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/RemoteTensorHandle.java
+++ /dev/null
@@ -1,1441 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/remote_tensor_handle.proto
-
-package org.tensorflow.proto.framework;
-
-/**
- * Protobuf type {@code tensorflow.eager.RemoteTensorHandle}
- */
-public  final class RemoteTensorHandle extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.eager.RemoteTensorHandle)
-    RemoteTensorHandleOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use RemoteTensorHandle.newBuilder() to construct.
-  private RemoteTensorHandle(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private RemoteTensorHandle() {
-    device_ = "";
-    opDevice_ = "";
-    dtype_ = 0;
-    resourceDtypesAndShapes_ = java.util.Collections.emptyList();
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new RemoteTensorHandle();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private RemoteTensorHandle(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    int mutable_bitField0_ = 0;
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 8: {
-
-            opId_ = input.readInt64();
-            break;
-          }
-          case 16: {
-
-            outputNum_ = input.readInt32();
-            break;
-          }
-          case 26: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            device_ = s;
-            break;
-          }
-          case 34: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            opDevice_ = s;
-            break;
-          }
-          case 40: {
-            int rawValue = input.readEnum();
-
-            dtype_ = rawValue;
-            break;
-          }
-          case 50: {
-            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
-              resourceDtypesAndShapes_ = new java.util.ArrayList<org.tensorflow.proto.framework.ResourceDtypeAndShape>();
-              mutable_bitField0_ |= 0x00000001;
-            }
-            resourceDtypesAndShapes_.add(
-                input.readMessage(org.tensorflow.proto.framework.ResourceDtypeAndShape.parser(), extensionRegistry));
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      if (((mutable_bitField0_ & 0x00000001) != 0)) {
-        resourceDtypesAndShapes_ = java.util.Collections.unmodifiableList(resourceDtypesAndShapes_);
-      }
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.framework.RemoteTensorHandleProtos.internal_static_tensorflow_eager_RemoteTensorHandle_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.framework.RemoteTensorHandleProtos.internal_static_tensorflow_eager_RemoteTensorHandle_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.framework.RemoteTensorHandle.class, org.tensorflow.proto.framework.RemoteTensorHandle.Builder.class);
-  }
-
-  public static final int OP_ID_FIELD_NUMBER = 1;
-  private long opId_;
-  /**
-   * <pre>
-   * The ID of the operation that produced this tensor.
-   * </pre>
-   *
-   * <code>int64 op_id = 1;</code>
-   */
-  public long getOpId() {
-    return opId_;
-  }
-
-  public static final int OUTPUT_NUM_FIELD_NUMBER = 2;
-  private int outputNum_;
-  /**
-   * <pre>
-   * The index into the outputs of the operation that produced this tensor.
-   * </pre>
-   *
-   * <code>int32 output_num = 2;</code>
-   */
-  public int getOutputNum() {
-    return outputNum_;
-  }
-
-  public static final int DEVICE_FIELD_NUMBER = 3;
-  private volatile java.lang.Object device_;
-  /**
-   * <pre>
-   * Device where the tensor is located. Cannot be empty.
-   * For multi-device functions, it's the default device passed to placer.
-   * </pre>
-   *
-   * <code>string device = 3;</code>
-   */
-  public java.lang.String getDevice() {
-    java.lang.Object ref = device_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      device_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * Device where the tensor is located. Cannot be empty.
-   * For multi-device functions, it's the default device passed to placer.
-   * </pre>
-   *
-   * <code>string device = 3;</code>
-   */
-  public com.google.protobuf.ByteString
-      getDeviceBytes() {
-    java.lang.Object ref = device_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      device_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int OP_DEVICE_FIELD_NUMBER = 4;
-  private volatile java.lang.Object opDevice_;
-  /**
-   * <pre>
-   * Device of the operation producing this tensor. Can be empty if the
-   * operation producing this tensor is a multi-device function.
-   * </pre>
-   *
-   * <code>string op_device = 4;</code>
-   */
-  public java.lang.String getOpDevice() {
-    java.lang.Object ref = opDevice_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      opDevice_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * Device of the operation producing this tensor. Can be empty if the
-   * operation producing this tensor is a multi-device function.
-   * </pre>
-   *
-   * <code>string op_device = 4;</code>
-   */
-  public com.google.protobuf.ByteString
-      getOpDeviceBytes() {
-    java.lang.Object ref = opDevice_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      opDevice_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int DTYPE_FIELD_NUMBER = 5;
-  private int dtype_;
-  /**
-   * <pre>
-   * Tensor type.
-   * </pre>
-   *
-   * <code>.tensorflow.DataType dtype = 5;</code>
-   */
-  public int getDtypeValue() {
-    return dtype_;
-  }
-  /**
-   * <pre>
-   * Tensor type.
-   * </pre>
-   *
-   * <code>.tensorflow.DataType dtype = 5;</code>
-   */
-  public org.tensorflow.proto.framework.DataType getDtype() {
-    @SuppressWarnings("deprecation")
-    org.tensorflow.proto.framework.DataType result = org.tensorflow.proto.framework.DataType.valueOf(dtype_);
-    return result == null ? org.tensorflow.proto.framework.DataType.UNRECOGNIZED : result;
-  }
-
-  public static final int RESOURCE_DTYPES_AND_SHAPES_FIELD_NUMBER = 6;
-  private java.util.List<org.tensorflow.proto.framework.ResourceDtypeAndShape> resourceDtypesAndShapes_;
-  /**
-   * <pre>
-   * Optional data types and shapes of a remote resource variable.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.eager.ResourceDtypeAndShape resource_dtypes_and_shapes = 6;</code>
-   */
-  public java.util.List<org.tensorflow.proto.framework.ResourceDtypeAndShape> getResourceDtypesAndShapesList() {
-    return resourceDtypesAndShapes_;
-  }
-  /**
-   * <pre>
-   * Optional data types and shapes of a remote resource variable.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.eager.ResourceDtypeAndShape resource_dtypes_and_shapes = 6;</code>
-   */
-  public java.util.List<? extends org.tensorflow.proto.framework.ResourceDtypeAndShapeOrBuilder> 
-      getResourceDtypesAndShapesOrBuilderList() {
-    return resourceDtypesAndShapes_;
-  }
-  /**
-   * <pre>
-   * Optional data types and shapes of a remote resource variable.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.eager.ResourceDtypeAndShape resource_dtypes_and_shapes = 6;</code>
-   */
-  public int getResourceDtypesAndShapesCount() {
-    return resourceDtypesAndShapes_.size();
-  }
-  /**
-   * <pre>
-   * Optional data types and shapes of a remote resource variable.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.eager.ResourceDtypeAndShape resource_dtypes_and_shapes = 6;</code>
-   */
-  public org.tensorflow.proto.framework.ResourceDtypeAndShape getResourceDtypesAndShapes(int index) {
-    return resourceDtypesAndShapes_.get(index);
-  }
-  /**
-   * <pre>
-   * Optional data types and shapes of a remote resource variable.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.eager.ResourceDtypeAndShape resource_dtypes_and_shapes = 6;</code>
-   */
-  public org.tensorflow.proto.framework.ResourceDtypeAndShapeOrBuilder getResourceDtypesAndShapesOrBuilder(
-      int index) {
-    return resourceDtypesAndShapes_.get(index);
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    if (opId_ != 0L) {
-      output.writeInt64(1, opId_);
-    }
-    if (outputNum_ != 0) {
-      output.writeInt32(2, outputNum_);
-    }
-    if (!getDeviceBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 3, device_);
-    }
-    if (!getOpDeviceBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 4, opDevice_);
-    }
-    if (dtype_ != org.tensorflow.proto.framework.DataType.DT_INVALID.getNumber()) {
-      output.writeEnum(5, dtype_);
-    }
-    for (int i = 0; i < resourceDtypesAndShapes_.size(); i++) {
-      output.writeMessage(6, resourceDtypesAndShapes_.get(i));
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (opId_ != 0L) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt64Size(1, opId_);
-    }
-    if (outputNum_ != 0) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt32Size(2, outputNum_);
-    }
-    if (!getDeviceBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(3, device_);
-    }
-    if (!getOpDeviceBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(4, opDevice_);
-    }
-    if (dtype_ != org.tensorflow.proto.framework.DataType.DT_INVALID.getNumber()) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeEnumSize(5, dtype_);
-    }
-    for (int i = 0; i < resourceDtypesAndShapes_.size(); i++) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(6, resourceDtypesAndShapes_.get(i));
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.framework.RemoteTensorHandle)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.framework.RemoteTensorHandle other = (org.tensorflow.proto.framework.RemoteTensorHandle) obj;
-
-    if (getOpId()
-        != other.getOpId()) return false;
-    if (getOutputNum()
-        != other.getOutputNum()) return false;
-    if (!getDevice()
-        .equals(other.getDevice())) return false;
-    if (!getOpDevice()
-        .equals(other.getOpDevice())) return false;
-    if (dtype_ != other.dtype_) return false;
-    if (!getResourceDtypesAndShapesList()
-        .equals(other.getResourceDtypesAndShapesList())) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    hash = (37 * hash) + OP_ID_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-        getOpId());
-    hash = (37 * hash) + OUTPUT_NUM_FIELD_NUMBER;
-    hash = (53 * hash) + getOutputNum();
-    hash = (37 * hash) + DEVICE_FIELD_NUMBER;
-    hash = (53 * hash) + getDevice().hashCode();
-    hash = (37 * hash) + OP_DEVICE_FIELD_NUMBER;
-    hash = (53 * hash) + getOpDevice().hashCode();
-    hash = (37 * hash) + DTYPE_FIELD_NUMBER;
-    hash = (53 * hash) + dtype_;
-    if (getResourceDtypesAndShapesCount() > 0) {
-      hash = (37 * hash) + RESOURCE_DTYPES_AND_SHAPES_FIELD_NUMBER;
-      hash = (53 * hash) + getResourceDtypesAndShapesList().hashCode();
-    }
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.framework.RemoteTensorHandle parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.RemoteTensorHandle parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.RemoteTensorHandle parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.RemoteTensorHandle parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.RemoteTensorHandle parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.RemoteTensorHandle parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.RemoteTensorHandle parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.RemoteTensorHandle parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.RemoteTensorHandle parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.RemoteTensorHandle parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.RemoteTensorHandle parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.RemoteTensorHandle parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.framework.RemoteTensorHandle prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * Protobuf type {@code tensorflow.eager.RemoteTensorHandle}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.eager.RemoteTensorHandle)
-      org.tensorflow.proto.framework.RemoteTensorHandleOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.RemoteTensorHandleProtos.internal_static_tensorflow_eager_RemoteTensorHandle_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.RemoteTensorHandleProtos.internal_static_tensorflow_eager_RemoteTensorHandle_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.RemoteTensorHandle.class, org.tensorflow.proto.framework.RemoteTensorHandle.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.framework.RemoteTensorHandle.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-        getResourceDtypesAndShapesFieldBuilder();
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      opId_ = 0L;
-
-      outputNum_ = 0;
-
-      device_ = "";
-
-      opDevice_ = "";
-
-      dtype_ = 0;
-
-      if (resourceDtypesAndShapesBuilder_ == null) {
-        resourceDtypesAndShapes_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000001);
-      } else {
-        resourceDtypesAndShapesBuilder_.clear();
-      }
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.framework.RemoteTensorHandleProtos.internal_static_tensorflow_eager_RemoteTensorHandle_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.RemoteTensorHandle getDefaultInstanceForType() {
-      return org.tensorflow.proto.framework.RemoteTensorHandle.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.RemoteTensorHandle build() {
-      org.tensorflow.proto.framework.RemoteTensorHandle result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.RemoteTensorHandle buildPartial() {
-      org.tensorflow.proto.framework.RemoteTensorHandle result = new org.tensorflow.proto.framework.RemoteTensorHandle(this);
-      int from_bitField0_ = bitField0_;
-      result.opId_ = opId_;
-      result.outputNum_ = outputNum_;
-      result.device_ = device_;
-      result.opDevice_ = opDevice_;
-      result.dtype_ = dtype_;
-      if (resourceDtypesAndShapesBuilder_ == null) {
-        if (((bitField0_ & 0x00000001) != 0)) {
-          resourceDtypesAndShapes_ = java.util.Collections.unmodifiableList(resourceDtypesAndShapes_);
-          bitField0_ = (bitField0_ & ~0x00000001);
-        }
-        result.resourceDtypesAndShapes_ = resourceDtypesAndShapes_;
-      } else {
-        result.resourceDtypesAndShapes_ = resourceDtypesAndShapesBuilder_.build();
-      }
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.framework.RemoteTensorHandle) {
-        return mergeFrom((org.tensorflow.proto.framework.RemoteTensorHandle)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.framework.RemoteTensorHandle other) {
-      if (other == org.tensorflow.proto.framework.RemoteTensorHandle.getDefaultInstance()) return this;
-      if (other.getOpId() != 0L) {
-        setOpId(other.getOpId());
-      }
-      if (other.getOutputNum() != 0) {
-        setOutputNum(other.getOutputNum());
-      }
-      if (!other.getDevice().isEmpty()) {
-        device_ = other.device_;
-        onChanged();
-      }
-      if (!other.getOpDevice().isEmpty()) {
-        opDevice_ = other.opDevice_;
-        onChanged();
-      }
-      if (other.dtype_ != 0) {
-        setDtypeValue(other.getDtypeValue());
-      }
-      if (resourceDtypesAndShapesBuilder_ == null) {
-        if (!other.resourceDtypesAndShapes_.isEmpty()) {
-          if (resourceDtypesAndShapes_.isEmpty()) {
-            resourceDtypesAndShapes_ = other.resourceDtypesAndShapes_;
-            bitField0_ = (bitField0_ & ~0x00000001);
-          } else {
-            ensureResourceDtypesAndShapesIsMutable();
-            resourceDtypesAndShapes_.addAll(other.resourceDtypesAndShapes_);
-          }
-          onChanged();
-        }
-      } else {
-        if (!other.resourceDtypesAndShapes_.isEmpty()) {
-          if (resourceDtypesAndShapesBuilder_.isEmpty()) {
-            resourceDtypesAndShapesBuilder_.dispose();
-            resourceDtypesAndShapesBuilder_ = null;
-            resourceDtypesAndShapes_ = other.resourceDtypesAndShapes_;
-            bitField0_ = (bitField0_ & ~0x00000001);
-            resourceDtypesAndShapesBuilder_ = 
-              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
-                 getResourceDtypesAndShapesFieldBuilder() : null;
-          } else {
-            resourceDtypesAndShapesBuilder_.addAllMessages(other.resourceDtypesAndShapes_);
-          }
-        }
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.framework.RemoteTensorHandle parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.framework.RemoteTensorHandle) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-    private int bitField0_;
-
-    private long opId_ ;
-    /**
-     * <pre>
-     * The ID of the operation that produced this tensor.
-     * </pre>
-     *
-     * <code>int64 op_id = 1;</code>
-     */
-    public long getOpId() {
-      return opId_;
-    }
-    /**
-     * <pre>
-     * The ID of the operation that produced this tensor.
-     * </pre>
-     *
-     * <code>int64 op_id = 1;</code>
-     */
-    public Builder setOpId(long value) {
-      
-      opId_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * The ID of the operation that produced this tensor.
-     * </pre>
-     *
-     * <code>int64 op_id = 1;</code>
-     */
-    public Builder clearOpId() {
-      
-      opId_ = 0L;
-      onChanged();
-      return this;
-    }
-
-    private int outputNum_ ;
-    /**
-     * <pre>
-     * The index into the outputs of the operation that produced this tensor.
-     * </pre>
-     *
-     * <code>int32 output_num = 2;</code>
-     */
-    public int getOutputNum() {
-      return outputNum_;
-    }
-    /**
-     * <pre>
-     * The index into the outputs of the operation that produced this tensor.
-     * </pre>
-     *
-     * <code>int32 output_num = 2;</code>
-     */
-    public Builder setOutputNum(int value) {
-      
-      outputNum_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * The index into the outputs of the operation that produced this tensor.
-     * </pre>
-     *
-     * <code>int32 output_num = 2;</code>
-     */
-    public Builder clearOutputNum() {
-      
-      outputNum_ = 0;
-      onChanged();
-      return this;
-    }
-
-    private java.lang.Object device_ = "";
-    /**
-     * <pre>
-     * Device where the tensor is located. Cannot be empty.
-     * For multi-device functions, it's the default device passed to placer.
-     * </pre>
-     *
-     * <code>string device = 3;</code>
-     */
-    public java.lang.String getDevice() {
-      java.lang.Object ref = device_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        device_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Device where the tensor is located. Cannot be empty.
-     * For multi-device functions, it's the default device passed to placer.
-     * </pre>
-     *
-     * <code>string device = 3;</code>
-     */
-    public com.google.protobuf.ByteString
-        getDeviceBytes() {
-      java.lang.Object ref = device_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        device_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Device where the tensor is located. Cannot be empty.
-     * For multi-device functions, it's the default device passed to placer.
-     * </pre>
-     *
-     * <code>string device = 3;</code>
-     */
-    public Builder setDevice(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      device_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Device where the tensor is located. Cannot be empty.
-     * For multi-device functions, it's the default device passed to placer.
-     * </pre>
-     *
-     * <code>string device = 3;</code>
-     */
-    public Builder clearDevice() {
-      
-      device_ = getDefaultInstance().getDevice();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Device where the tensor is located. Cannot be empty.
-     * For multi-device functions, it's the default device passed to placer.
-     * </pre>
-     *
-     * <code>string device = 3;</code>
-     */
-    public Builder setDeviceBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      device_ = value;
-      onChanged();
-      return this;
-    }
-
-    private java.lang.Object opDevice_ = "";
-    /**
-     * <pre>
-     * Device of the operation producing this tensor. Can be empty if the
-     * operation producing this tensor is a multi-device function.
-     * </pre>
-     *
-     * <code>string op_device = 4;</code>
-     */
-    public java.lang.String getOpDevice() {
-      java.lang.Object ref = opDevice_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        opDevice_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Device of the operation producing this tensor. Can be empty if the
-     * operation producing this tensor is a multi-device function.
-     * </pre>
-     *
-     * <code>string op_device = 4;</code>
-     */
-    public com.google.protobuf.ByteString
-        getOpDeviceBytes() {
-      java.lang.Object ref = opDevice_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        opDevice_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Device of the operation producing this tensor. Can be empty if the
-     * operation producing this tensor is a multi-device function.
-     * </pre>
-     *
-     * <code>string op_device = 4;</code>
-     */
-    public Builder setOpDevice(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      opDevice_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Device of the operation producing this tensor. Can be empty if the
-     * operation producing this tensor is a multi-device function.
-     * </pre>
-     *
-     * <code>string op_device = 4;</code>
-     */
-    public Builder clearOpDevice() {
-      
-      opDevice_ = getDefaultInstance().getOpDevice();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Device of the operation producing this tensor. Can be empty if the
-     * operation producing this tensor is a multi-device function.
-     * </pre>
-     *
-     * <code>string op_device = 4;</code>
-     */
-    public Builder setOpDeviceBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      opDevice_ = value;
-      onChanged();
-      return this;
-    }
-
-    private int dtype_ = 0;
-    /**
-     * <pre>
-     * Tensor type.
-     * </pre>
-     *
-     * <code>.tensorflow.DataType dtype = 5;</code>
-     */
-    public int getDtypeValue() {
-      return dtype_;
-    }
-    /**
-     * <pre>
-     * Tensor type.
-     * </pre>
-     *
-     * <code>.tensorflow.DataType dtype = 5;</code>
-     */
-    public Builder setDtypeValue(int value) {
-      dtype_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Tensor type.
-     * </pre>
-     *
-     * <code>.tensorflow.DataType dtype = 5;</code>
-     */
-    public org.tensorflow.proto.framework.DataType getDtype() {
-      @SuppressWarnings("deprecation")
-      org.tensorflow.proto.framework.DataType result = org.tensorflow.proto.framework.DataType.valueOf(dtype_);
-      return result == null ? org.tensorflow.proto.framework.DataType.UNRECOGNIZED : result;
-    }
-    /**
-     * <pre>
-     * Tensor type.
-     * </pre>
-     *
-     * <code>.tensorflow.DataType dtype = 5;</code>
-     */
-    public Builder setDtype(org.tensorflow.proto.framework.DataType value) {
-      if (value == null) {
-        throw new NullPointerException();
-      }
-      
-      dtype_ = value.getNumber();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Tensor type.
-     * </pre>
-     *
-     * <code>.tensorflow.DataType dtype = 5;</code>
-     */
-    public Builder clearDtype() {
-      
-      dtype_ = 0;
-      onChanged();
-      return this;
-    }
-
-    private java.util.List<org.tensorflow.proto.framework.ResourceDtypeAndShape> resourceDtypesAndShapes_ =
-      java.util.Collections.emptyList();
-    private void ensureResourceDtypesAndShapesIsMutable() {
-      if (!((bitField0_ & 0x00000001) != 0)) {
-        resourceDtypesAndShapes_ = new java.util.ArrayList<org.tensorflow.proto.framework.ResourceDtypeAndShape>(resourceDtypesAndShapes_);
-        bitField0_ |= 0x00000001;
-       }
-    }
-
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.framework.ResourceDtypeAndShape, org.tensorflow.proto.framework.ResourceDtypeAndShape.Builder, org.tensorflow.proto.framework.ResourceDtypeAndShapeOrBuilder> resourceDtypesAndShapesBuilder_;
-
-    /**
-     * <pre>
-     * Optional data types and shapes of a remote resource variable.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.eager.ResourceDtypeAndShape resource_dtypes_and_shapes = 6;</code>
-     */
-    public java.util.List<org.tensorflow.proto.framework.ResourceDtypeAndShape> getResourceDtypesAndShapesList() {
-      if (resourceDtypesAndShapesBuilder_ == null) {
-        return java.util.Collections.unmodifiableList(resourceDtypesAndShapes_);
-      } else {
-        return resourceDtypesAndShapesBuilder_.getMessageList();
-      }
-    }
-    /**
-     * <pre>
-     * Optional data types and shapes of a remote resource variable.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.eager.ResourceDtypeAndShape resource_dtypes_and_shapes = 6;</code>
-     */
-    public int getResourceDtypesAndShapesCount() {
-      if (resourceDtypesAndShapesBuilder_ == null) {
-        return resourceDtypesAndShapes_.size();
-      } else {
-        return resourceDtypesAndShapesBuilder_.getCount();
-      }
-    }
-    /**
-     * <pre>
-     * Optional data types and shapes of a remote resource variable.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.eager.ResourceDtypeAndShape resource_dtypes_and_shapes = 6;</code>
-     */
-    public org.tensorflow.proto.framework.ResourceDtypeAndShape getResourceDtypesAndShapes(int index) {
-      if (resourceDtypesAndShapesBuilder_ == null) {
-        return resourceDtypesAndShapes_.get(index);
-      } else {
-        return resourceDtypesAndShapesBuilder_.getMessage(index);
-      }
-    }
-    /**
-     * <pre>
-     * Optional data types and shapes of a remote resource variable.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.eager.ResourceDtypeAndShape resource_dtypes_and_shapes = 6;</code>
-     */
-    public Builder setResourceDtypesAndShapes(
-        int index, org.tensorflow.proto.framework.ResourceDtypeAndShape value) {
-      if (resourceDtypesAndShapesBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureResourceDtypesAndShapesIsMutable();
-        resourceDtypesAndShapes_.set(index, value);
-        onChanged();
-      } else {
-        resourceDtypesAndShapesBuilder_.setMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Optional data types and shapes of a remote resource variable.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.eager.ResourceDtypeAndShape resource_dtypes_and_shapes = 6;</code>
-     */
-    public Builder setResourceDtypesAndShapes(
-        int index, org.tensorflow.proto.framework.ResourceDtypeAndShape.Builder builderForValue) {
-      if (resourceDtypesAndShapesBuilder_ == null) {
-        ensureResourceDtypesAndShapesIsMutable();
-        resourceDtypesAndShapes_.set(index, builderForValue.build());
-        onChanged();
-      } else {
-        resourceDtypesAndShapesBuilder_.setMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Optional data types and shapes of a remote resource variable.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.eager.ResourceDtypeAndShape resource_dtypes_and_shapes = 6;</code>
-     */
-    public Builder addResourceDtypesAndShapes(org.tensorflow.proto.framework.ResourceDtypeAndShape value) {
-      if (resourceDtypesAndShapesBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureResourceDtypesAndShapesIsMutable();
-        resourceDtypesAndShapes_.add(value);
-        onChanged();
-      } else {
-        resourceDtypesAndShapesBuilder_.addMessage(value);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Optional data types and shapes of a remote resource variable.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.eager.ResourceDtypeAndShape resource_dtypes_and_shapes = 6;</code>
-     */
-    public Builder addResourceDtypesAndShapes(
-        int index, org.tensorflow.proto.framework.ResourceDtypeAndShape value) {
-      if (resourceDtypesAndShapesBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureResourceDtypesAndShapesIsMutable();
-        resourceDtypesAndShapes_.add(index, value);
-        onChanged();
-      } else {
-        resourceDtypesAndShapesBuilder_.addMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Optional data types and shapes of a remote resource variable.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.eager.ResourceDtypeAndShape resource_dtypes_and_shapes = 6;</code>
-     */
-    public Builder addResourceDtypesAndShapes(
-        org.tensorflow.proto.framework.ResourceDtypeAndShape.Builder builderForValue) {
-      if (resourceDtypesAndShapesBuilder_ == null) {
-        ensureResourceDtypesAndShapesIsMutable();
-        resourceDtypesAndShapes_.add(builderForValue.build());
-        onChanged();
-      } else {
-        resourceDtypesAndShapesBuilder_.addMessage(builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Optional data types and shapes of a remote resource variable.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.eager.ResourceDtypeAndShape resource_dtypes_and_shapes = 6;</code>
-     */
-    public Builder addResourceDtypesAndShapes(
-        int index, org.tensorflow.proto.framework.ResourceDtypeAndShape.Builder builderForValue) {
-      if (resourceDtypesAndShapesBuilder_ == null) {
-        ensureResourceDtypesAndShapesIsMutable();
-        resourceDtypesAndShapes_.add(index, builderForValue.build());
-        onChanged();
-      } else {
-        resourceDtypesAndShapesBuilder_.addMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Optional data types and shapes of a remote resource variable.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.eager.ResourceDtypeAndShape resource_dtypes_and_shapes = 6;</code>
-     */
-    public Builder addAllResourceDtypesAndShapes(
-        java.lang.Iterable<? extends org.tensorflow.proto.framework.ResourceDtypeAndShape> values) {
-      if (resourceDtypesAndShapesBuilder_ == null) {
-        ensureResourceDtypesAndShapesIsMutable();
-        com.google.protobuf.AbstractMessageLite.Builder.addAll(
-            values, resourceDtypesAndShapes_);
-        onChanged();
-      } else {
-        resourceDtypesAndShapesBuilder_.addAllMessages(values);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Optional data types and shapes of a remote resource variable.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.eager.ResourceDtypeAndShape resource_dtypes_and_shapes = 6;</code>
-     */
-    public Builder clearResourceDtypesAndShapes() {
-      if (resourceDtypesAndShapesBuilder_ == null) {
-        resourceDtypesAndShapes_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000001);
-        onChanged();
-      } else {
-        resourceDtypesAndShapesBuilder_.clear();
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Optional data types and shapes of a remote resource variable.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.eager.ResourceDtypeAndShape resource_dtypes_and_shapes = 6;</code>
-     */
-    public Builder removeResourceDtypesAndShapes(int index) {
-      if (resourceDtypesAndShapesBuilder_ == null) {
-        ensureResourceDtypesAndShapesIsMutable();
-        resourceDtypesAndShapes_.remove(index);
-        onChanged();
-      } else {
-        resourceDtypesAndShapesBuilder_.remove(index);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Optional data types and shapes of a remote resource variable.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.eager.ResourceDtypeAndShape resource_dtypes_and_shapes = 6;</code>
-     */
-    public org.tensorflow.proto.framework.ResourceDtypeAndShape.Builder getResourceDtypesAndShapesBuilder(
-        int index) {
-      return getResourceDtypesAndShapesFieldBuilder().getBuilder(index);
-    }
-    /**
-     * <pre>
-     * Optional data types and shapes of a remote resource variable.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.eager.ResourceDtypeAndShape resource_dtypes_and_shapes = 6;</code>
-     */
-    public org.tensorflow.proto.framework.ResourceDtypeAndShapeOrBuilder getResourceDtypesAndShapesOrBuilder(
-        int index) {
-      if (resourceDtypesAndShapesBuilder_ == null) {
-        return resourceDtypesAndShapes_.get(index);  } else {
-        return resourceDtypesAndShapesBuilder_.getMessageOrBuilder(index);
-      }
-    }
-    /**
-     * <pre>
-     * Optional data types and shapes of a remote resource variable.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.eager.ResourceDtypeAndShape resource_dtypes_and_shapes = 6;</code>
-     */
-    public java.util.List<? extends org.tensorflow.proto.framework.ResourceDtypeAndShapeOrBuilder> 
-         getResourceDtypesAndShapesOrBuilderList() {
-      if (resourceDtypesAndShapesBuilder_ != null) {
-        return resourceDtypesAndShapesBuilder_.getMessageOrBuilderList();
-      } else {
-        return java.util.Collections.unmodifiableList(resourceDtypesAndShapes_);
-      }
-    }
-    /**
-     * <pre>
-     * Optional data types and shapes of a remote resource variable.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.eager.ResourceDtypeAndShape resource_dtypes_and_shapes = 6;</code>
-     */
-    public org.tensorflow.proto.framework.ResourceDtypeAndShape.Builder addResourceDtypesAndShapesBuilder() {
-      return getResourceDtypesAndShapesFieldBuilder().addBuilder(
-          org.tensorflow.proto.framework.ResourceDtypeAndShape.getDefaultInstance());
-    }
-    /**
-     * <pre>
-     * Optional data types and shapes of a remote resource variable.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.eager.ResourceDtypeAndShape resource_dtypes_and_shapes = 6;</code>
-     */
-    public org.tensorflow.proto.framework.ResourceDtypeAndShape.Builder addResourceDtypesAndShapesBuilder(
-        int index) {
-      return getResourceDtypesAndShapesFieldBuilder().addBuilder(
-          index, org.tensorflow.proto.framework.ResourceDtypeAndShape.getDefaultInstance());
-    }
-    /**
-     * <pre>
-     * Optional data types and shapes of a remote resource variable.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.eager.ResourceDtypeAndShape resource_dtypes_and_shapes = 6;</code>
-     */
-    public java.util.List<org.tensorflow.proto.framework.ResourceDtypeAndShape.Builder> 
-         getResourceDtypesAndShapesBuilderList() {
-      return getResourceDtypesAndShapesFieldBuilder().getBuilderList();
-    }
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.framework.ResourceDtypeAndShape, org.tensorflow.proto.framework.ResourceDtypeAndShape.Builder, org.tensorflow.proto.framework.ResourceDtypeAndShapeOrBuilder> 
-        getResourceDtypesAndShapesFieldBuilder() {
-      if (resourceDtypesAndShapesBuilder_ == null) {
-        resourceDtypesAndShapesBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
-            org.tensorflow.proto.framework.ResourceDtypeAndShape, org.tensorflow.proto.framework.ResourceDtypeAndShape.Builder, org.tensorflow.proto.framework.ResourceDtypeAndShapeOrBuilder>(
-                resourceDtypesAndShapes_,
-                ((bitField0_ & 0x00000001) != 0),
-                getParentForChildren(),
-                isClean());
-        resourceDtypesAndShapes_ = null;
-      }
-      return resourceDtypesAndShapesBuilder_;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.eager.RemoteTensorHandle)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.eager.RemoteTensorHandle)
-  private static final org.tensorflow.proto.framework.RemoteTensorHandle DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.RemoteTensorHandle();
-  }
-
-  public static org.tensorflow.proto.framework.RemoteTensorHandle getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<RemoteTensorHandle>
-      PARSER = new com.google.protobuf.AbstractParser<RemoteTensorHandle>() {
-    @java.lang.Override
-    public RemoteTensorHandle parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new RemoteTensorHandle(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<RemoteTensorHandle> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<RemoteTensorHandle> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.framework.RemoteTensorHandle getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/RemoteTensorHandleOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/RemoteTensorHandleOrBuilder.java
deleted file mode 100644
index e9f1098f494..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/RemoteTensorHandleOrBuilder.java
+++ /dev/null
@@ -1,128 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/remote_tensor_handle.proto
-
-package org.tensorflow.proto.framework;
-
-public interface RemoteTensorHandleOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.eager.RemoteTensorHandle)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <pre>
-   * The ID of the operation that produced this tensor.
-   * </pre>
-   *
-   * <code>int64 op_id = 1;</code>
-   */
-  long getOpId();
-
-  /**
-   * <pre>
-   * The index into the outputs of the operation that produced this tensor.
-   * </pre>
-   *
-   * <code>int32 output_num = 2;</code>
-   */
-  int getOutputNum();
-
-  /**
-   * <pre>
-   * Device where the tensor is located. Cannot be empty.
-   * For multi-device functions, it's the default device passed to placer.
-   * </pre>
-   *
-   * <code>string device = 3;</code>
-   */
-  java.lang.String getDevice();
-  /**
-   * <pre>
-   * Device where the tensor is located. Cannot be empty.
-   * For multi-device functions, it's the default device passed to placer.
-   * </pre>
-   *
-   * <code>string device = 3;</code>
-   */
-  com.google.protobuf.ByteString
-      getDeviceBytes();
-
-  /**
-   * <pre>
-   * Device of the operation producing this tensor. Can be empty if the
-   * operation producing this tensor is a multi-device function.
-   * </pre>
-   *
-   * <code>string op_device = 4;</code>
-   */
-  java.lang.String getOpDevice();
-  /**
-   * <pre>
-   * Device of the operation producing this tensor. Can be empty if the
-   * operation producing this tensor is a multi-device function.
-   * </pre>
-   *
-   * <code>string op_device = 4;</code>
-   */
-  com.google.protobuf.ByteString
-      getOpDeviceBytes();
-
-  /**
-   * <pre>
-   * Tensor type.
-   * </pre>
-   *
-   * <code>.tensorflow.DataType dtype = 5;</code>
-   */
-  int getDtypeValue();
-  /**
-   * <pre>
-   * Tensor type.
-   * </pre>
-   *
-   * <code>.tensorflow.DataType dtype = 5;</code>
-   */
-  org.tensorflow.proto.framework.DataType getDtype();
-
-  /**
-   * <pre>
-   * Optional data types and shapes of a remote resource variable.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.eager.ResourceDtypeAndShape resource_dtypes_and_shapes = 6;</code>
-   */
-  java.util.List<org.tensorflow.proto.framework.ResourceDtypeAndShape> 
-      getResourceDtypesAndShapesList();
-  /**
-   * <pre>
-   * Optional data types and shapes of a remote resource variable.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.eager.ResourceDtypeAndShape resource_dtypes_and_shapes = 6;</code>
-   */
-  org.tensorflow.proto.framework.ResourceDtypeAndShape getResourceDtypesAndShapes(int index);
-  /**
-   * <pre>
-   * Optional data types and shapes of a remote resource variable.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.eager.ResourceDtypeAndShape resource_dtypes_and_shapes = 6;</code>
-   */
-  int getResourceDtypesAndShapesCount();
-  /**
-   * <pre>
-   * Optional data types and shapes of a remote resource variable.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.eager.ResourceDtypeAndShape resource_dtypes_and_shapes = 6;</code>
-   */
-  java.util.List<? extends org.tensorflow.proto.framework.ResourceDtypeAndShapeOrBuilder> 
-      getResourceDtypesAndShapesOrBuilderList();
-  /**
-   * <pre>
-   * Optional data types and shapes of a remote resource variable.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.eager.ResourceDtypeAndShape resource_dtypes_and_shapes = 6;</code>
-   */
-  org.tensorflow.proto.framework.ResourceDtypeAndShapeOrBuilder getResourceDtypesAndShapesOrBuilder(
-      int index);
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/RemoteTensorHandleProtos.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/RemoteTensorHandleProtos.java
deleted file mode 100644
index 82e7ea5bc2c..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/RemoteTensorHandleProtos.java
+++ /dev/null
@@ -1,76 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/remote_tensor_handle.proto
-
-package org.tensorflow.proto.framework;
-
-public final class RemoteTensorHandleProtos {
-  private RemoteTensorHandleProtos() {}
-  public static void registerAllExtensions(
-      com.google.protobuf.ExtensionRegistryLite registry) {
-  }
-
-  public static void registerAllExtensions(
-      com.google.protobuf.ExtensionRegistry registry) {
-    registerAllExtensions(
-        (com.google.protobuf.ExtensionRegistryLite) registry);
-  }
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_eager_ResourceDtypeAndShape_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_eager_ResourceDtypeAndShape_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_eager_RemoteTensorHandle_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_eager_RemoteTensorHandle_fieldAccessorTable;
-
-  public static com.google.protobuf.Descriptors.FileDescriptor
-      getDescriptor() {
-    return descriptor;
-  }
-  private static  com.google.protobuf.Descriptors.FileDescriptor
-      descriptor;
-  static {
-    java.lang.String[] descriptorData = {
-      "\n3tensorflow/core/protobuf/remote_tensor" +
-      "_handle.proto\022\020tensorflow.eager\032,tensorf" +
-      "low/core/framework/tensor_shape.proto\032%t" +
-      "ensorflow/core/framework/types.proto\"i\n\025" +
-      "ResourceDtypeAndShape\022#\n\005dtype\030\001 \001(\0162\024.t" +
-      "ensorflow.DataType\022+\n\005shape\030\002 \001(\0132\034.tens" +
-      "orflow.TensorShapeProto\"\314\001\n\022RemoteTensor" +
-      "Handle\022\r\n\005op_id\030\001 \001(\003\022\022\n\noutput_num\030\002 \001(" +
-      "\005\022\016\n\006device\030\003 \001(\t\022\021\n\top_device\030\004 \001(\t\022#\n\005" +
-      "dtype\030\005 \001(\0162\024.tensorflow.DataType\022K\n\032res" +
-      "ource_dtypes_and_shapes\030\006 \003(\0132\'.tensorfl" +
-      "ow.eager.ResourceDtypeAndShapeB\211\001\n\036org.t" +
-      "ensorflow.proto.frameworkB\030RemoteTensorH" +
-      "andleProtosP\001ZHgithub.com/tensorflow/ten" +
-      "sorflow/tensorflow/go/core/core_protos_g" +
-      "o_proto\370\001\001b\006proto3"
-    };
-    descriptor = com.google.protobuf.Descriptors.FileDescriptor
-      .internalBuildGeneratedFileFrom(descriptorData,
-        new com.google.protobuf.Descriptors.FileDescriptor[] {
-          org.tensorflow.proto.framework.TensorShapeProtos.getDescriptor(),
-          org.tensorflow.proto.framework.TypesProtos.getDescriptor(),
-        });
-    internal_static_tensorflow_eager_ResourceDtypeAndShape_descriptor =
-      getDescriptor().getMessageTypes().get(0);
-    internal_static_tensorflow_eager_ResourceDtypeAndShape_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_eager_ResourceDtypeAndShape_descriptor,
-        new java.lang.String[] { "Dtype", "Shape", });
-    internal_static_tensorflow_eager_RemoteTensorHandle_descriptor =
-      getDescriptor().getMessageTypes().get(1);
-    internal_static_tensorflow_eager_RemoteTensorHandle_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_eager_RemoteTensorHandle_descriptor,
-        new java.lang.String[] { "OpId", "OutputNum", "Device", "OpDevice", "Dtype", "ResourceDtypesAndShapes", });
-    org.tensorflow.proto.framework.TensorShapeProtos.getDescriptor();
-    org.tensorflow.proto.framework.TypesProtos.getDescriptor();
-  }
-
-  // @@protoc_insertion_point(outer_class_scope)
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/Resource.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/Resource.java
deleted file mode 100644
index 7743aecf24e..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/Resource.java
+++ /dev/null
@@ -1,659 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/trace_events.proto
-
-package org.tensorflow.proto.framework;
-
-/**
- * <pre>
- * A 'resource' generally is a specific computation component on a device. These
- * can range from threads on CPUs to specific arithmetic units on hardware
- * devices.
- * </pre>
- *
- * Protobuf type {@code tensorflow.profiler.Resource}
- */
-public  final class Resource extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.profiler.Resource)
-    ResourceOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use Resource.newBuilder() to construct.
-  private Resource(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private Resource() {
-    name_ = "";
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new Resource();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private Resource(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 10: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            name_ = s;
-            break;
-          }
-          case 16: {
-
-            resourceId_ = input.readUInt32();
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.framework.TraceEventsProtos.internal_static_tensorflow_profiler_Resource_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.framework.TraceEventsProtos.internal_static_tensorflow_profiler_Resource_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.framework.Resource.class, org.tensorflow.proto.framework.Resource.Builder.class);
-  }
-
-  public static final int NAME_FIELD_NUMBER = 1;
-  private volatile java.lang.Object name_;
-  /**
-   * <pre>
-   * The name of the resource.
-   * </pre>
-   *
-   * <code>string name = 1;</code>
-   */
-  public java.lang.String getName() {
-    java.lang.Object ref = name_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      name_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * The name of the resource.
-   * </pre>
-   *
-   * <code>string name = 1;</code>
-   */
-  public com.google.protobuf.ByteString
-      getNameBytes() {
-    java.lang.Object ref = name_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      name_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int RESOURCE_ID_FIELD_NUMBER = 2;
-  private int resourceId_;
-  /**
-   * <pre>
-   * The id of the resource. Unique within a device.
-   * </pre>
-   *
-   * <code>uint32 resource_id = 2;</code>
-   */
-  public int getResourceId() {
-    return resourceId_;
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    if (!getNameBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 1, name_);
-    }
-    if (resourceId_ != 0) {
-      output.writeUInt32(2, resourceId_);
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (!getNameBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, name_);
-    }
-    if (resourceId_ != 0) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeUInt32Size(2, resourceId_);
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.framework.Resource)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.framework.Resource other = (org.tensorflow.proto.framework.Resource) obj;
-
-    if (!getName()
-        .equals(other.getName())) return false;
-    if (getResourceId()
-        != other.getResourceId()) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    hash = (37 * hash) + NAME_FIELD_NUMBER;
-    hash = (53 * hash) + getName().hashCode();
-    hash = (37 * hash) + RESOURCE_ID_FIELD_NUMBER;
-    hash = (53 * hash) + getResourceId();
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.framework.Resource parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.Resource parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.Resource parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.Resource parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.Resource parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.Resource parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.Resource parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.Resource parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.Resource parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.Resource parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.Resource parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.Resource parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.framework.Resource prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * <pre>
-   * A 'resource' generally is a specific computation component on a device. These
-   * can range from threads on CPUs to specific arithmetic units on hardware
-   * devices.
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.profiler.Resource}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.profiler.Resource)
-      org.tensorflow.proto.framework.ResourceOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.TraceEventsProtos.internal_static_tensorflow_profiler_Resource_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.TraceEventsProtos.internal_static_tensorflow_profiler_Resource_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.Resource.class, org.tensorflow.proto.framework.Resource.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.framework.Resource.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      name_ = "";
-
-      resourceId_ = 0;
-
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.framework.TraceEventsProtos.internal_static_tensorflow_profiler_Resource_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.Resource getDefaultInstanceForType() {
-      return org.tensorflow.proto.framework.Resource.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.Resource build() {
-      org.tensorflow.proto.framework.Resource result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.Resource buildPartial() {
-      org.tensorflow.proto.framework.Resource result = new org.tensorflow.proto.framework.Resource(this);
-      result.name_ = name_;
-      result.resourceId_ = resourceId_;
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.framework.Resource) {
-        return mergeFrom((org.tensorflow.proto.framework.Resource)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.framework.Resource other) {
-      if (other == org.tensorflow.proto.framework.Resource.getDefaultInstance()) return this;
-      if (!other.getName().isEmpty()) {
-        name_ = other.name_;
-        onChanged();
-      }
-      if (other.getResourceId() != 0) {
-        setResourceId(other.getResourceId());
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.framework.Resource parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.framework.Resource) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-
-    private java.lang.Object name_ = "";
-    /**
-     * <pre>
-     * The name of the resource.
-     * </pre>
-     *
-     * <code>string name = 1;</code>
-     */
-    public java.lang.String getName() {
-      java.lang.Object ref = name_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        name_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * The name of the resource.
-     * </pre>
-     *
-     * <code>string name = 1;</code>
-     */
-    public com.google.protobuf.ByteString
-        getNameBytes() {
-      java.lang.Object ref = name_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        name_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * The name of the resource.
-     * </pre>
-     *
-     * <code>string name = 1;</code>
-     */
-    public Builder setName(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      name_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * The name of the resource.
-     * </pre>
-     *
-     * <code>string name = 1;</code>
-     */
-    public Builder clearName() {
-      
-      name_ = getDefaultInstance().getName();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * The name of the resource.
-     * </pre>
-     *
-     * <code>string name = 1;</code>
-     */
-    public Builder setNameBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      name_ = value;
-      onChanged();
-      return this;
-    }
-
-    private int resourceId_ ;
-    /**
-     * <pre>
-     * The id of the resource. Unique within a device.
-     * </pre>
-     *
-     * <code>uint32 resource_id = 2;</code>
-     */
-    public int getResourceId() {
-      return resourceId_;
-    }
-    /**
-     * <pre>
-     * The id of the resource. Unique within a device.
-     * </pre>
-     *
-     * <code>uint32 resource_id = 2;</code>
-     */
-    public Builder setResourceId(int value) {
-      
-      resourceId_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * The id of the resource. Unique within a device.
-     * </pre>
-     *
-     * <code>uint32 resource_id = 2;</code>
-     */
-    public Builder clearResourceId() {
-      
-      resourceId_ = 0;
-      onChanged();
-      return this;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.profiler.Resource)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.profiler.Resource)
-  private static final org.tensorflow.proto.framework.Resource DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.Resource();
-  }
-
-  public static org.tensorflow.proto.framework.Resource getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<Resource>
-      PARSER = new com.google.protobuf.AbstractParser<Resource>() {
-    @java.lang.Override
-    public Resource parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new Resource(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<Resource> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<Resource> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.framework.Resource getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ResourceDtypeAndShape.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ResourceDtypeAndShape.java
deleted file mode 100644
index 2586a7e626a..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ResourceDtypeAndShape.java
+++ /dev/null
@@ -1,685 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/remote_tensor_handle.proto
-
-package org.tensorflow.proto.framework;
-
-/**
- * Protobuf type {@code tensorflow.eager.ResourceDtypeAndShape}
- */
-public  final class ResourceDtypeAndShape extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.eager.ResourceDtypeAndShape)
-    ResourceDtypeAndShapeOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use ResourceDtypeAndShape.newBuilder() to construct.
-  private ResourceDtypeAndShape(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private ResourceDtypeAndShape() {
-    dtype_ = 0;
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new ResourceDtypeAndShape();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private ResourceDtypeAndShape(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 8: {
-            int rawValue = input.readEnum();
-
-            dtype_ = rawValue;
-            break;
-          }
-          case 18: {
-            org.tensorflow.proto.framework.TensorShapeProto.Builder subBuilder = null;
-            if (shape_ != null) {
-              subBuilder = shape_.toBuilder();
-            }
-            shape_ = input.readMessage(org.tensorflow.proto.framework.TensorShapeProto.parser(), extensionRegistry);
-            if (subBuilder != null) {
-              subBuilder.mergeFrom(shape_);
-              shape_ = subBuilder.buildPartial();
-            }
-
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.framework.RemoteTensorHandleProtos.internal_static_tensorflow_eager_ResourceDtypeAndShape_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.framework.RemoteTensorHandleProtos.internal_static_tensorflow_eager_ResourceDtypeAndShape_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.framework.ResourceDtypeAndShape.class, org.tensorflow.proto.framework.ResourceDtypeAndShape.Builder.class);
-  }
-
-  public static final int DTYPE_FIELD_NUMBER = 1;
-  private int dtype_;
-  /**
-   * <code>.tensorflow.DataType dtype = 1;</code>
-   */
-  public int getDtypeValue() {
-    return dtype_;
-  }
-  /**
-   * <code>.tensorflow.DataType dtype = 1;</code>
-   */
-  public org.tensorflow.proto.framework.DataType getDtype() {
-    @SuppressWarnings("deprecation")
-    org.tensorflow.proto.framework.DataType result = org.tensorflow.proto.framework.DataType.valueOf(dtype_);
-    return result == null ? org.tensorflow.proto.framework.DataType.UNRECOGNIZED : result;
-  }
-
-  public static final int SHAPE_FIELD_NUMBER = 2;
-  private org.tensorflow.proto.framework.TensorShapeProto shape_;
-  /**
-   * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-   */
-  public boolean hasShape() {
-    return shape_ != null;
-  }
-  /**
-   * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-   */
-  public org.tensorflow.proto.framework.TensorShapeProto getShape() {
-    return shape_ == null ? org.tensorflow.proto.framework.TensorShapeProto.getDefaultInstance() : shape_;
-  }
-  /**
-   * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-   */
-  public org.tensorflow.proto.framework.TensorShapeProtoOrBuilder getShapeOrBuilder() {
-    return getShape();
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    if (dtype_ != org.tensorflow.proto.framework.DataType.DT_INVALID.getNumber()) {
-      output.writeEnum(1, dtype_);
-    }
-    if (shape_ != null) {
-      output.writeMessage(2, getShape());
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (dtype_ != org.tensorflow.proto.framework.DataType.DT_INVALID.getNumber()) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeEnumSize(1, dtype_);
-    }
-    if (shape_ != null) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(2, getShape());
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.framework.ResourceDtypeAndShape)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.framework.ResourceDtypeAndShape other = (org.tensorflow.proto.framework.ResourceDtypeAndShape) obj;
-
-    if (dtype_ != other.dtype_) return false;
-    if (hasShape() != other.hasShape()) return false;
-    if (hasShape()) {
-      if (!getShape()
-          .equals(other.getShape())) return false;
-    }
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    hash = (37 * hash) + DTYPE_FIELD_NUMBER;
-    hash = (53 * hash) + dtype_;
-    if (hasShape()) {
-      hash = (37 * hash) + SHAPE_FIELD_NUMBER;
-      hash = (53 * hash) + getShape().hashCode();
-    }
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.framework.ResourceDtypeAndShape parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.ResourceDtypeAndShape parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.ResourceDtypeAndShape parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.ResourceDtypeAndShape parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.ResourceDtypeAndShape parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.ResourceDtypeAndShape parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.ResourceDtypeAndShape parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.ResourceDtypeAndShape parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.ResourceDtypeAndShape parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.ResourceDtypeAndShape parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.ResourceDtypeAndShape parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.ResourceDtypeAndShape parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.framework.ResourceDtypeAndShape prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * Protobuf type {@code tensorflow.eager.ResourceDtypeAndShape}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.eager.ResourceDtypeAndShape)
-      org.tensorflow.proto.framework.ResourceDtypeAndShapeOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.RemoteTensorHandleProtos.internal_static_tensorflow_eager_ResourceDtypeAndShape_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.RemoteTensorHandleProtos.internal_static_tensorflow_eager_ResourceDtypeAndShape_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.ResourceDtypeAndShape.class, org.tensorflow.proto.framework.ResourceDtypeAndShape.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.framework.ResourceDtypeAndShape.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      dtype_ = 0;
-
-      if (shapeBuilder_ == null) {
-        shape_ = null;
-      } else {
-        shape_ = null;
-        shapeBuilder_ = null;
-      }
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.framework.RemoteTensorHandleProtos.internal_static_tensorflow_eager_ResourceDtypeAndShape_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.ResourceDtypeAndShape getDefaultInstanceForType() {
-      return org.tensorflow.proto.framework.ResourceDtypeAndShape.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.ResourceDtypeAndShape build() {
-      org.tensorflow.proto.framework.ResourceDtypeAndShape result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.ResourceDtypeAndShape buildPartial() {
-      org.tensorflow.proto.framework.ResourceDtypeAndShape result = new org.tensorflow.proto.framework.ResourceDtypeAndShape(this);
-      result.dtype_ = dtype_;
-      if (shapeBuilder_ == null) {
-        result.shape_ = shape_;
-      } else {
-        result.shape_ = shapeBuilder_.build();
-      }
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.framework.ResourceDtypeAndShape) {
-        return mergeFrom((org.tensorflow.proto.framework.ResourceDtypeAndShape)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.framework.ResourceDtypeAndShape other) {
-      if (other == org.tensorflow.proto.framework.ResourceDtypeAndShape.getDefaultInstance()) return this;
-      if (other.dtype_ != 0) {
-        setDtypeValue(other.getDtypeValue());
-      }
-      if (other.hasShape()) {
-        mergeShape(other.getShape());
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.framework.ResourceDtypeAndShape parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.framework.ResourceDtypeAndShape) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-
-    private int dtype_ = 0;
-    /**
-     * <code>.tensorflow.DataType dtype = 1;</code>
-     */
-    public int getDtypeValue() {
-      return dtype_;
-    }
-    /**
-     * <code>.tensorflow.DataType dtype = 1;</code>
-     */
-    public Builder setDtypeValue(int value) {
-      dtype_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>.tensorflow.DataType dtype = 1;</code>
-     */
-    public org.tensorflow.proto.framework.DataType getDtype() {
-      @SuppressWarnings("deprecation")
-      org.tensorflow.proto.framework.DataType result = org.tensorflow.proto.framework.DataType.valueOf(dtype_);
-      return result == null ? org.tensorflow.proto.framework.DataType.UNRECOGNIZED : result;
-    }
-    /**
-     * <code>.tensorflow.DataType dtype = 1;</code>
-     */
-    public Builder setDtype(org.tensorflow.proto.framework.DataType value) {
-      if (value == null) {
-        throw new NullPointerException();
-      }
-      
-      dtype_ = value.getNumber();
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>.tensorflow.DataType dtype = 1;</code>
-     */
-    public Builder clearDtype() {
-      
-      dtype_ = 0;
-      onChanged();
-      return this;
-    }
-
-    private org.tensorflow.proto.framework.TensorShapeProto shape_;
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.TensorShapeProto, org.tensorflow.proto.framework.TensorShapeProto.Builder, org.tensorflow.proto.framework.TensorShapeProtoOrBuilder> shapeBuilder_;
-    /**
-     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-     */
-    public boolean hasShape() {
-      return shapeBuilder_ != null || shape_ != null;
-    }
-    /**
-     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-     */
-    public org.tensorflow.proto.framework.TensorShapeProto getShape() {
-      if (shapeBuilder_ == null) {
-        return shape_ == null ? org.tensorflow.proto.framework.TensorShapeProto.getDefaultInstance() : shape_;
-      } else {
-        return shapeBuilder_.getMessage();
-      }
-    }
-    /**
-     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-     */
-    public Builder setShape(org.tensorflow.proto.framework.TensorShapeProto value) {
-      if (shapeBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        shape_ = value;
-        onChanged();
-      } else {
-        shapeBuilder_.setMessage(value);
-      }
-
-      return this;
-    }
-    /**
-     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-     */
-    public Builder setShape(
-        org.tensorflow.proto.framework.TensorShapeProto.Builder builderForValue) {
-      if (shapeBuilder_ == null) {
-        shape_ = builderForValue.build();
-        onChanged();
-      } else {
-        shapeBuilder_.setMessage(builderForValue.build());
-      }
-
-      return this;
-    }
-    /**
-     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-     */
-    public Builder mergeShape(org.tensorflow.proto.framework.TensorShapeProto value) {
-      if (shapeBuilder_ == null) {
-        if (shape_ != null) {
-          shape_ =
-            org.tensorflow.proto.framework.TensorShapeProto.newBuilder(shape_).mergeFrom(value).buildPartial();
-        } else {
-          shape_ = value;
-        }
-        onChanged();
-      } else {
-        shapeBuilder_.mergeFrom(value);
-      }
-
-      return this;
-    }
-    /**
-     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-     */
-    public Builder clearShape() {
-      if (shapeBuilder_ == null) {
-        shape_ = null;
-        onChanged();
-      } else {
-        shape_ = null;
-        shapeBuilder_ = null;
-      }
-
-      return this;
-    }
-    /**
-     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-     */
-    public org.tensorflow.proto.framework.TensorShapeProto.Builder getShapeBuilder() {
-      
-      onChanged();
-      return getShapeFieldBuilder().getBuilder();
-    }
-    /**
-     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-     */
-    public org.tensorflow.proto.framework.TensorShapeProtoOrBuilder getShapeOrBuilder() {
-      if (shapeBuilder_ != null) {
-        return shapeBuilder_.getMessageOrBuilder();
-      } else {
-        return shape_ == null ?
-            org.tensorflow.proto.framework.TensorShapeProto.getDefaultInstance() : shape_;
-      }
-    }
-    /**
-     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-     */
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.TensorShapeProto, org.tensorflow.proto.framework.TensorShapeProto.Builder, org.tensorflow.proto.framework.TensorShapeProtoOrBuilder> 
-        getShapeFieldBuilder() {
-      if (shapeBuilder_ == null) {
-        shapeBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-            org.tensorflow.proto.framework.TensorShapeProto, org.tensorflow.proto.framework.TensorShapeProto.Builder, org.tensorflow.proto.framework.TensorShapeProtoOrBuilder>(
-                getShape(),
-                getParentForChildren(),
-                isClean());
-        shape_ = null;
-      }
-      return shapeBuilder_;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.eager.ResourceDtypeAndShape)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.eager.ResourceDtypeAndShape)
-  private static final org.tensorflow.proto.framework.ResourceDtypeAndShape DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.ResourceDtypeAndShape();
-  }
-
-  public static org.tensorflow.proto.framework.ResourceDtypeAndShape getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<ResourceDtypeAndShape>
-      PARSER = new com.google.protobuf.AbstractParser<ResourceDtypeAndShape>() {
-    @java.lang.Override
-    public ResourceDtypeAndShape parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new ResourceDtypeAndShape(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<ResourceDtypeAndShape> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<ResourceDtypeAndShape> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.framework.ResourceDtypeAndShape getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ResourceDtypeAndShapeOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ResourceDtypeAndShapeOrBuilder.java
deleted file mode 100644
index 67ab4da1b8d..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ResourceDtypeAndShapeOrBuilder.java
+++ /dev/null
@@ -1,31 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/remote_tensor_handle.proto
-
-package org.tensorflow.proto.framework;
-
-public interface ResourceDtypeAndShapeOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.eager.ResourceDtypeAndShape)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <code>.tensorflow.DataType dtype = 1;</code>
-   */
-  int getDtypeValue();
-  /**
-   * <code>.tensorflow.DataType dtype = 1;</code>
-   */
-  org.tensorflow.proto.framework.DataType getDtype();
-
-  /**
-   * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-   */
-  boolean hasShape();
-  /**
-   * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-   */
-  org.tensorflow.proto.framework.TensorShapeProto getShape();
-  /**
-   * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-   */
-  org.tensorflow.proto.framework.TensorShapeProtoOrBuilder getShapeOrBuilder();
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ResourceHandle.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ResourceHandle.java
deleted file mode 100644
index 5a96505f0eb..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ResourceHandle.java
+++ /dev/null
@@ -1,75 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/framework/resource_handle.proto
-
-package org.tensorflow.proto.framework;
-
-public final class ResourceHandle {
-  private ResourceHandle() {}
-  public static void registerAllExtensions(
-      com.google.protobuf.ExtensionRegistryLite registry) {
-  }
-
-  public static void registerAllExtensions(
-      com.google.protobuf.ExtensionRegistry registry) {
-    registerAllExtensions(
-        (com.google.protobuf.ExtensionRegistryLite) registry);
-  }
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_ResourceHandleProto_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_ResourceHandleProto_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_ResourceHandleProto_DtypeAndShape_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_ResourceHandleProto_DtypeAndShape_fieldAccessorTable;
-
-  public static com.google.protobuf.Descriptors.FileDescriptor
-      getDescriptor() {
-    return descriptor;
-  }
-  private static  com.google.protobuf.Descriptors.FileDescriptor
-      descriptor;
-  static {
-    java.lang.String[] descriptorData = {
-      "\n/tensorflow/core/framework/resource_han" +
-      "dle.proto\022\ntensorflow\032,tensorflow/core/f" +
-      "ramework/tensor_shape.proto\032%tensorflow/" +
-      "core/framework/types.proto\"\245\002\n\023ResourceH" +
-      "andleProto\022\016\n\006device\030\001 \001(\t\022\021\n\tcontainer\030" +
-      "\002 \001(\t\022\014\n\004name\030\003 \001(\t\022\021\n\thash_code\030\004 \001(\004\022\027" +
-      "\n\017maybe_type_name\030\005 \001(\t\022H\n\021dtypes_and_sh" +
-      "apes\030\006 \003(\0132-.tensorflow.ResourceHandlePr" +
-      "oto.DtypeAndShape\032a\n\rDtypeAndShape\022#\n\005dt" +
-      "ype\030\001 \001(\0162\024.tensorflow.DataType\022+\n\005shape" +
-      "\030\002 \001(\0132\034.tensorflow.TensorShapeProtoJ\004\010\007" +
-      "\020\010B\215\001\n\036org.tensorflow.proto.frameworkB\016R" +
-      "esourceHandleP\001ZVgithub.com/tensorflow/t" +
-      "ensorflow/tensorflow/go/core/framework/r" +
-      "esource_handle_go_proto\370\001\001b\006proto3"
-    };
-    descriptor = com.google.protobuf.Descriptors.FileDescriptor
-      .internalBuildGeneratedFileFrom(descriptorData,
-        new com.google.protobuf.Descriptors.FileDescriptor[] {
-          org.tensorflow.proto.framework.TensorShapeProtos.getDescriptor(),
-          org.tensorflow.proto.framework.TypesProtos.getDescriptor(),
-        });
-    internal_static_tensorflow_ResourceHandleProto_descriptor =
-      getDescriptor().getMessageTypes().get(0);
-    internal_static_tensorflow_ResourceHandleProto_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_ResourceHandleProto_descriptor,
-        new java.lang.String[] { "Device", "Container", "Name", "HashCode", "MaybeTypeName", "DtypesAndShapes", });
-    internal_static_tensorflow_ResourceHandleProto_DtypeAndShape_descriptor =
-      internal_static_tensorflow_ResourceHandleProto_descriptor.getNestedTypes().get(0);
-    internal_static_tensorflow_ResourceHandleProto_DtypeAndShape_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_ResourceHandleProto_DtypeAndShape_descriptor,
-        new java.lang.String[] { "Dtype", "Shape", });
-    org.tensorflow.proto.framework.TensorShapeProtos.getDescriptor();
-    org.tensorflow.proto.framework.TypesProtos.getDescriptor();
-  }
-
-  // @@protoc_insertion_point(outer_class_scope)
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ResourceHandleProto.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ResourceHandleProto.java
deleted file mode 100644
index 3d31206b794..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ResourceHandleProto.java
+++ /dev/null
@@ -1,2288 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/framework/resource_handle.proto
-
-package org.tensorflow.proto.framework;
-
-/**
- * <pre>
- * Protocol buffer representing a handle to a tensorflow resource. Handles are
- * not valid across executions, but can be serialized back and forth from within
- * a single run.
- * </pre>
- *
- * Protobuf type {@code tensorflow.ResourceHandleProto}
- */
-public  final class ResourceHandleProto extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.ResourceHandleProto)
-    ResourceHandleProtoOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use ResourceHandleProto.newBuilder() to construct.
-  private ResourceHandleProto(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private ResourceHandleProto() {
-    device_ = "";
-    container_ = "";
-    name_ = "";
-    maybeTypeName_ = "";
-    dtypesAndShapes_ = java.util.Collections.emptyList();
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new ResourceHandleProto();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private ResourceHandleProto(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    int mutable_bitField0_ = 0;
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 10: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            device_ = s;
-            break;
-          }
-          case 18: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            container_ = s;
-            break;
-          }
-          case 26: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            name_ = s;
-            break;
-          }
-          case 32: {
-
-            hashCode_ = input.readUInt64();
-            break;
-          }
-          case 42: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            maybeTypeName_ = s;
-            break;
-          }
-          case 50: {
-            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
-              dtypesAndShapes_ = new java.util.ArrayList<org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShape>();
-              mutable_bitField0_ |= 0x00000001;
-            }
-            dtypesAndShapes_.add(
-                input.readMessage(org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShape.parser(), extensionRegistry));
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      if (((mutable_bitField0_ & 0x00000001) != 0)) {
-        dtypesAndShapes_ = java.util.Collections.unmodifiableList(dtypesAndShapes_);
-      }
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.framework.ResourceHandle.internal_static_tensorflow_ResourceHandleProto_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.framework.ResourceHandle.internal_static_tensorflow_ResourceHandleProto_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.framework.ResourceHandleProto.class, org.tensorflow.proto.framework.ResourceHandleProto.Builder.class);
-  }
-
-  public interface DtypeAndShapeOrBuilder extends
-      // @@protoc_insertion_point(interface_extends:tensorflow.ResourceHandleProto.DtypeAndShape)
-      com.google.protobuf.MessageOrBuilder {
-
-    /**
-     * <code>.tensorflow.DataType dtype = 1;</code>
-     */
-    int getDtypeValue();
-    /**
-     * <code>.tensorflow.DataType dtype = 1;</code>
-     */
-    org.tensorflow.proto.framework.DataType getDtype();
-
-    /**
-     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-     */
-    boolean hasShape();
-    /**
-     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-     */
-    org.tensorflow.proto.framework.TensorShapeProto getShape();
-    /**
-     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-     */
-    org.tensorflow.proto.framework.TensorShapeProtoOrBuilder getShapeOrBuilder();
-  }
-  /**
-   * <pre>
-   * Protocol buffer representing a pair of (data type, tensor shape).
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.ResourceHandleProto.DtypeAndShape}
-   */
-  public  static final class DtypeAndShape extends
-      com.google.protobuf.GeneratedMessageV3 implements
-      // @@protoc_insertion_point(message_implements:tensorflow.ResourceHandleProto.DtypeAndShape)
-      DtypeAndShapeOrBuilder {
-  private static final long serialVersionUID = 0L;
-    // Use DtypeAndShape.newBuilder() to construct.
-    private DtypeAndShape(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-      super(builder);
-    }
-    private DtypeAndShape() {
-      dtype_ = 0;
-    }
-
-    @java.lang.Override
-    @SuppressWarnings({"unused"})
-    protected java.lang.Object newInstance(
-        UnusedPrivateParameter unused) {
-      return new DtypeAndShape();
-    }
-
-    @java.lang.Override
-    public final com.google.protobuf.UnknownFieldSet
-    getUnknownFields() {
-      return this.unknownFields;
-    }
-    private DtypeAndShape(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      this();
-      if (extensionRegistry == null) {
-        throw new java.lang.NullPointerException();
-      }
-      com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-          com.google.protobuf.UnknownFieldSet.newBuilder();
-      try {
-        boolean done = false;
-        while (!done) {
-          int tag = input.readTag();
-          switch (tag) {
-            case 0:
-              done = true;
-              break;
-            case 8: {
-              int rawValue = input.readEnum();
-
-              dtype_ = rawValue;
-              break;
-            }
-            case 18: {
-              org.tensorflow.proto.framework.TensorShapeProto.Builder subBuilder = null;
-              if (shape_ != null) {
-                subBuilder = shape_.toBuilder();
-              }
-              shape_ = input.readMessage(org.tensorflow.proto.framework.TensorShapeProto.parser(), extensionRegistry);
-              if (subBuilder != null) {
-                subBuilder.mergeFrom(shape_);
-                shape_ = subBuilder.buildPartial();
-              }
-
-              break;
-            }
-            default: {
-              if (!parseUnknownField(
-                  input, unknownFields, extensionRegistry, tag)) {
-                done = true;
-              }
-              break;
-            }
-          }
-        }
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        throw e.setUnfinishedMessage(this);
-      } catch (java.io.IOException e) {
-        throw new com.google.protobuf.InvalidProtocolBufferException(
-            e).setUnfinishedMessage(this);
-      } finally {
-        this.unknownFields = unknownFields.build();
-        makeExtensionsImmutable();
-      }
-    }
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.ResourceHandle.internal_static_tensorflow_ResourceHandleProto_DtypeAndShape_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.ResourceHandle.internal_static_tensorflow_ResourceHandleProto_DtypeAndShape_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShape.class, org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShape.Builder.class);
-    }
-
-    public static final int DTYPE_FIELD_NUMBER = 1;
-    private int dtype_;
-    /**
-     * <code>.tensorflow.DataType dtype = 1;</code>
-     */
-    public int getDtypeValue() {
-      return dtype_;
-    }
-    /**
-     * <code>.tensorflow.DataType dtype = 1;</code>
-     */
-    public org.tensorflow.proto.framework.DataType getDtype() {
-      @SuppressWarnings("deprecation")
-      org.tensorflow.proto.framework.DataType result = org.tensorflow.proto.framework.DataType.valueOf(dtype_);
-      return result == null ? org.tensorflow.proto.framework.DataType.UNRECOGNIZED : result;
-    }
-
-    public static final int SHAPE_FIELD_NUMBER = 2;
-    private org.tensorflow.proto.framework.TensorShapeProto shape_;
-    /**
-     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-     */
-    public boolean hasShape() {
-      return shape_ != null;
-    }
-    /**
-     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-     */
-    public org.tensorflow.proto.framework.TensorShapeProto getShape() {
-      return shape_ == null ? org.tensorflow.proto.framework.TensorShapeProto.getDefaultInstance() : shape_;
-    }
-    /**
-     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-     */
-    public org.tensorflow.proto.framework.TensorShapeProtoOrBuilder getShapeOrBuilder() {
-      return getShape();
-    }
-
-    private byte memoizedIsInitialized = -1;
-    @java.lang.Override
-    public final boolean isInitialized() {
-      byte isInitialized = memoizedIsInitialized;
-      if (isInitialized == 1) return true;
-      if (isInitialized == 0) return false;
-
-      memoizedIsInitialized = 1;
-      return true;
-    }
-
-    @java.lang.Override
-    public void writeTo(com.google.protobuf.CodedOutputStream output)
-                        throws java.io.IOException {
-      if (dtype_ != org.tensorflow.proto.framework.DataType.DT_INVALID.getNumber()) {
-        output.writeEnum(1, dtype_);
-      }
-      if (shape_ != null) {
-        output.writeMessage(2, getShape());
-      }
-      unknownFields.writeTo(output);
-    }
-
-    @java.lang.Override
-    public int getSerializedSize() {
-      int size = memoizedSize;
-      if (size != -1) return size;
-
-      size = 0;
-      if (dtype_ != org.tensorflow.proto.framework.DataType.DT_INVALID.getNumber()) {
-        size += com.google.protobuf.CodedOutputStream
-          .computeEnumSize(1, dtype_);
-      }
-      if (shape_ != null) {
-        size += com.google.protobuf.CodedOutputStream
-          .computeMessageSize(2, getShape());
-      }
-      size += unknownFields.getSerializedSize();
-      memoizedSize = size;
-      return size;
-    }
-
-    @java.lang.Override
-    public boolean equals(final java.lang.Object obj) {
-      if (obj == this) {
-       return true;
-      }
-      if (!(obj instanceof org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShape)) {
-        return super.equals(obj);
-      }
-      org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShape other = (org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShape) obj;
-
-      if (dtype_ != other.dtype_) return false;
-      if (hasShape() != other.hasShape()) return false;
-      if (hasShape()) {
-        if (!getShape()
-            .equals(other.getShape())) return false;
-      }
-      if (!unknownFields.equals(other.unknownFields)) return false;
-      return true;
-    }
-
-    @java.lang.Override
-    public int hashCode() {
-      if (memoizedHashCode != 0) {
-        return memoizedHashCode;
-      }
-      int hash = 41;
-      hash = (19 * hash) + getDescriptor().hashCode();
-      hash = (37 * hash) + DTYPE_FIELD_NUMBER;
-      hash = (53 * hash) + dtype_;
-      if (hasShape()) {
-        hash = (37 * hash) + SHAPE_FIELD_NUMBER;
-        hash = (53 * hash) + getShape().hashCode();
-      }
-      hash = (29 * hash) + unknownFields.hashCode();
-      memoizedHashCode = hash;
-      return hash;
-    }
-
-    public static org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShape parseFrom(
-        java.nio.ByteBuffer data)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data);
-    }
-    public static org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShape parseFrom(
-        java.nio.ByteBuffer data,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShape parseFrom(
-        com.google.protobuf.ByteString data)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data);
-    }
-    public static org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShape parseFrom(
-        com.google.protobuf.ByteString data,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShape parseFrom(byte[] data)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data);
-    }
-    public static org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShape parseFrom(
-        byte[] data,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShape parseFrom(java.io.InputStream input)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input);
-    }
-    public static org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShape parseFrom(
-        java.io.InputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShape parseDelimitedFrom(java.io.InputStream input)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseDelimitedWithIOException(PARSER, input);
-    }
-    public static org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShape parseDelimitedFrom(
-        java.io.InputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShape parseFrom(
-        com.google.protobuf.CodedInputStream input)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input);
-    }
-    public static org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShape parseFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input, extensionRegistry);
-    }
-
-    @java.lang.Override
-    public Builder newBuilderForType() { return newBuilder(); }
-    public static Builder newBuilder() {
-      return DEFAULT_INSTANCE.toBuilder();
-    }
-    public static Builder newBuilder(org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShape prototype) {
-      return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-    }
-    @java.lang.Override
-    public Builder toBuilder() {
-      return this == DEFAULT_INSTANCE
-          ? new Builder() : new Builder().mergeFrom(this);
-    }
-
-    @java.lang.Override
-    protected Builder newBuilderForType(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      Builder builder = new Builder(parent);
-      return builder;
-    }
-    /**
-     * <pre>
-     * Protocol buffer representing a pair of (data type, tensor shape).
-     * </pre>
-     *
-     * Protobuf type {@code tensorflow.ResourceHandleProto.DtypeAndShape}
-     */
-    public static final class Builder extends
-        com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-        // @@protoc_insertion_point(builder_implements:tensorflow.ResourceHandleProto.DtypeAndShape)
-        org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShapeOrBuilder {
-      public static final com.google.protobuf.Descriptors.Descriptor
-          getDescriptor() {
-        return org.tensorflow.proto.framework.ResourceHandle.internal_static_tensorflow_ResourceHandleProto_DtypeAndShape_descriptor;
-      }
-
-      @java.lang.Override
-      protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-          internalGetFieldAccessorTable() {
-        return org.tensorflow.proto.framework.ResourceHandle.internal_static_tensorflow_ResourceHandleProto_DtypeAndShape_fieldAccessorTable
-            .ensureFieldAccessorsInitialized(
-                org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShape.class, org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShape.Builder.class);
-      }
-
-      // Construct using org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShape.newBuilder()
-      private Builder() {
-        maybeForceBuilderInitialization();
-      }
-
-      private Builder(
-          com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-        super(parent);
-        maybeForceBuilderInitialization();
-      }
-      private void maybeForceBuilderInitialization() {
-        if (com.google.protobuf.GeneratedMessageV3
-                .alwaysUseFieldBuilders) {
-        }
-      }
-      @java.lang.Override
-      public Builder clear() {
-        super.clear();
-        dtype_ = 0;
-
-        if (shapeBuilder_ == null) {
-          shape_ = null;
-        } else {
-          shape_ = null;
-          shapeBuilder_ = null;
-        }
-        return this;
-      }
-
-      @java.lang.Override
-      public com.google.protobuf.Descriptors.Descriptor
-          getDescriptorForType() {
-        return org.tensorflow.proto.framework.ResourceHandle.internal_static_tensorflow_ResourceHandleProto_DtypeAndShape_descriptor;
-      }
-
-      @java.lang.Override
-      public org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShape getDefaultInstanceForType() {
-        return org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShape.getDefaultInstance();
-      }
-
-      @java.lang.Override
-      public org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShape build() {
-        org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShape result = buildPartial();
-        if (!result.isInitialized()) {
-          throw newUninitializedMessageException(result);
-        }
-        return result;
-      }
-
-      @java.lang.Override
-      public org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShape buildPartial() {
-        org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShape result = new org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShape(this);
-        result.dtype_ = dtype_;
-        if (shapeBuilder_ == null) {
-          result.shape_ = shape_;
-        } else {
-          result.shape_ = shapeBuilder_.build();
-        }
-        onBuilt();
-        return result;
-      }
-
-      @java.lang.Override
-      public Builder clone() {
-        return super.clone();
-      }
-      @java.lang.Override
-      public Builder setField(
-          com.google.protobuf.Descriptors.FieldDescriptor field,
-          java.lang.Object value) {
-        return super.setField(field, value);
-      }
-      @java.lang.Override
-      public Builder clearField(
-          com.google.protobuf.Descriptors.FieldDescriptor field) {
-        return super.clearField(field);
-      }
-      @java.lang.Override
-      public Builder clearOneof(
-          com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-        return super.clearOneof(oneof);
-      }
-      @java.lang.Override
-      public Builder setRepeatedField(
-          com.google.protobuf.Descriptors.FieldDescriptor field,
-          int index, java.lang.Object value) {
-        return super.setRepeatedField(field, index, value);
-      }
-      @java.lang.Override
-      public Builder addRepeatedField(
-          com.google.protobuf.Descriptors.FieldDescriptor field,
-          java.lang.Object value) {
-        return super.addRepeatedField(field, value);
-      }
-      @java.lang.Override
-      public Builder mergeFrom(com.google.protobuf.Message other) {
-        if (other instanceof org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShape) {
-          return mergeFrom((org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShape)other);
-        } else {
-          super.mergeFrom(other);
-          return this;
-        }
-      }
-
-      public Builder mergeFrom(org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShape other) {
-        if (other == org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShape.getDefaultInstance()) return this;
-        if (other.dtype_ != 0) {
-          setDtypeValue(other.getDtypeValue());
-        }
-        if (other.hasShape()) {
-          mergeShape(other.getShape());
-        }
-        this.mergeUnknownFields(other.unknownFields);
-        onChanged();
-        return this;
-      }
-
-      @java.lang.Override
-      public final boolean isInitialized() {
-        return true;
-      }
-
-      @java.lang.Override
-      public Builder mergeFrom(
-          com.google.protobuf.CodedInputStream input,
-          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-          throws java.io.IOException {
-        org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShape parsedMessage = null;
-        try {
-          parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-        } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-          parsedMessage = (org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShape) e.getUnfinishedMessage();
-          throw e.unwrapIOException();
-        } finally {
-          if (parsedMessage != null) {
-            mergeFrom(parsedMessage);
-          }
-        }
-        return this;
-      }
-
-      private int dtype_ = 0;
-      /**
-       * <code>.tensorflow.DataType dtype = 1;</code>
-       */
-      public int getDtypeValue() {
-        return dtype_;
-      }
-      /**
-       * <code>.tensorflow.DataType dtype = 1;</code>
-       */
-      public Builder setDtypeValue(int value) {
-        dtype_ = value;
-        onChanged();
-        return this;
-      }
-      /**
-       * <code>.tensorflow.DataType dtype = 1;</code>
-       */
-      public org.tensorflow.proto.framework.DataType getDtype() {
-        @SuppressWarnings("deprecation")
-        org.tensorflow.proto.framework.DataType result = org.tensorflow.proto.framework.DataType.valueOf(dtype_);
-        return result == null ? org.tensorflow.proto.framework.DataType.UNRECOGNIZED : result;
-      }
-      /**
-       * <code>.tensorflow.DataType dtype = 1;</code>
-       */
-      public Builder setDtype(org.tensorflow.proto.framework.DataType value) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        
-        dtype_ = value.getNumber();
-        onChanged();
-        return this;
-      }
-      /**
-       * <code>.tensorflow.DataType dtype = 1;</code>
-       */
-      public Builder clearDtype() {
-        
-        dtype_ = 0;
-        onChanged();
-        return this;
-      }
-
-      private org.tensorflow.proto.framework.TensorShapeProto shape_;
-      private com.google.protobuf.SingleFieldBuilderV3<
-          org.tensorflow.proto.framework.TensorShapeProto, org.tensorflow.proto.framework.TensorShapeProto.Builder, org.tensorflow.proto.framework.TensorShapeProtoOrBuilder> shapeBuilder_;
-      /**
-       * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-       */
-      public boolean hasShape() {
-        return shapeBuilder_ != null || shape_ != null;
-      }
-      /**
-       * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-       */
-      public org.tensorflow.proto.framework.TensorShapeProto getShape() {
-        if (shapeBuilder_ == null) {
-          return shape_ == null ? org.tensorflow.proto.framework.TensorShapeProto.getDefaultInstance() : shape_;
-        } else {
-          return shapeBuilder_.getMessage();
-        }
-      }
-      /**
-       * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-       */
-      public Builder setShape(org.tensorflow.proto.framework.TensorShapeProto value) {
-        if (shapeBuilder_ == null) {
-          if (value == null) {
-            throw new NullPointerException();
-          }
-          shape_ = value;
-          onChanged();
-        } else {
-          shapeBuilder_.setMessage(value);
-        }
-
-        return this;
-      }
-      /**
-       * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-       */
-      public Builder setShape(
-          org.tensorflow.proto.framework.TensorShapeProto.Builder builderForValue) {
-        if (shapeBuilder_ == null) {
-          shape_ = builderForValue.build();
-          onChanged();
-        } else {
-          shapeBuilder_.setMessage(builderForValue.build());
-        }
-
-        return this;
-      }
-      /**
-       * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-       */
-      public Builder mergeShape(org.tensorflow.proto.framework.TensorShapeProto value) {
-        if (shapeBuilder_ == null) {
-          if (shape_ != null) {
-            shape_ =
-              org.tensorflow.proto.framework.TensorShapeProto.newBuilder(shape_).mergeFrom(value).buildPartial();
-          } else {
-            shape_ = value;
-          }
-          onChanged();
-        } else {
-          shapeBuilder_.mergeFrom(value);
-        }
-
-        return this;
-      }
-      /**
-       * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-       */
-      public Builder clearShape() {
-        if (shapeBuilder_ == null) {
-          shape_ = null;
-          onChanged();
-        } else {
-          shape_ = null;
-          shapeBuilder_ = null;
-        }
-
-        return this;
-      }
-      /**
-       * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-       */
-      public org.tensorflow.proto.framework.TensorShapeProto.Builder getShapeBuilder() {
-        
-        onChanged();
-        return getShapeFieldBuilder().getBuilder();
-      }
-      /**
-       * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-       */
-      public org.tensorflow.proto.framework.TensorShapeProtoOrBuilder getShapeOrBuilder() {
-        if (shapeBuilder_ != null) {
-          return shapeBuilder_.getMessageOrBuilder();
-        } else {
-          return shape_ == null ?
-              org.tensorflow.proto.framework.TensorShapeProto.getDefaultInstance() : shape_;
-        }
-      }
-      /**
-       * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-       */
-      private com.google.protobuf.SingleFieldBuilderV3<
-          org.tensorflow.proto.framework.TensorShapeProto, org.tensorflow.proto.framework.TensorShapeProto.Builder, org.tensorflow.proto.framework.TensorShapeProtoOrBuilder> 
-          getShapeFieldBuilder() {
-        if (shapeBuilder_ == null) {
-          shapeBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-              org.tensorflow.proto.framework.TensorShapeProto, org.tensorflow.proto.framework.TensorShapeProto.Builder, org.tensorflow.proto.framework.TensorShapeProtoOrBuilder>(
-                  getShape(),
-                  getParentForChildren(),
-                  isClean());
-          shape_ = null;
-        }
-        return shapeBuilder_;
-      }
-      @java.lang.Override
-      public final Builder setUnknownFields(
-          final com.google.protobuf.UnknownFieldSet unknownFields) {
-        return super.setUnknownFields(unknownFields);
-      }
-
-      @java.lang.Override
-      public final Builder mergeUnknownFields(
-          final com.google.protobuf.UnknownFieldSet unknownFields) {
-        return super.mergeUnknownFields(unknownFields);
-      }
-
-
-      // @@protoc_insertion_point(builder_scope:tensorflow.ResourceHandleProto.DtypeAndShape)
-    }
-
-    // @@protoc_insertion_point(class_scope:tensorflow.ResourceHandleProto.DtypeAndShape)
-    private static final org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShape DEFAULT_INSTANCE;
-    static {
-      DEFAULT_INSTANCE = new org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShape();
-    }
-
-    public static org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShape getDefaultInstance() {
-      return DEFAULT_INSTANCE;
-    }
-
-    private static final com.google.protobuf.Parser<DtypeAndShape>
-        PARSER = new com.google.protobuf.AbstractParser<DtypeAndShape>() {
-      @java.lang.Override
-      public DtypeAndShape parsePartialFrom(
-          com.google.protobuf.CodedInputStream input,
-          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-          throws com.google.protobuf.InvalidProtocolBufferException {
-        return new DtypeAndShape(input, extensionRegistry);
-      }
-    };
-
-    public static com.google.protobuf.Parser<DtypeAndShape> parser() {
-      return PARSER;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Parser<DtypeAndShape> getParserForType() {
-      return PARSER;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShape getDefaultInstanceForType() {
-      return DEFAULT_INSTANCE;
-    }
-
-  }
-
-  public static final int DEVICE_FIELD_NUMBER = 1;
-  private volatile java.lang.Object device_;
-  /**
-   * <pre>
-   * Unique name for the device containing the resource.
-   * </pre>
-   *
-   * <code>string device = 1;</code>
-   */
-  public java.lang.String getDevice() {
-    java.lang.Object ref = device_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      device_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * Unique name for the device containing the resource.
-   * </pre>
-   *
-   * <code>string device = 1;</code>
-   */
-  public com.google.protobuf.ByteString
-      getDeviceBytes() {
-    java.lang.Object ref = device_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      device_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int CONTAINER_FIELD_NUMBER = 2;
-  private volatile java.lang.Object container_;
-  /**
-   * <pre>
-   * Container in which this resource is placed.
-   * </pre>
-   *
-   * <code>string container = 2;</code>
-   */
-  public java.lang.String getContainer() {
-    java.lang.Object ref = container_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      container_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * Container in which this resource is placed.
-   * </pre>
-   *
-   * <code>string container = 2;</code>
-   */
-  public com.google.protobuf.ByteString
-      getContainerBytes() {
-    java.lang.Object ref = container_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      container_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int NAME_FIELD_NUMBER = 3;
-  private volatile java.lang.Object name_;
-  /**
-   * <pre>
-   * Unique name of this resource.
-   * </pre>
-   *
-   * <code>string name = 3;</code>
-   */
-  public java.lang.String getName() {
-    java.lang.Object ref = name_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      name_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * Unique name of this resource.
-   * </pre>
-   *
-   * <code>string name = 3;</code>
-   */
-  public com.google.protobuf.ByteString
-      getNameBytes() {
-    java.lang.Object ref = name_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      name_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int HASH_CODE_FIELD_NUMBER = 4;
-  private long hashCode_;
-  /**
-   * <pre>
-   * Hash code for the type of the resource. Is only valid in the same device
-   * and in the same execution.
-   * </pre>
-   *
-   * <code>uint64 hash_code = 4;</code>
-   */
-  public long getHashCode() {
-    return hashCode_;
-  }
-
-  public static final int MAYBE_TYPE_NAME_FIELD_NUMBER = 5;
-  private volatile java.lang.Object maybeTypeName_;
-  /**
-   * <pre>
-   * For debug-only, the name of the type pointed to by this handle, if
-   * available.
-   * </pre>
-   *
-   * <code>string maybe_type_name = 5;</code>
-   */
-  public java.lang.String getMaybeTypeName() {
-    java.lang.Object ref = maybeTypeName_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      maybeTypeName_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * For debug-only, the name of the type pointed to by this handle, if
-   * available.
-   * </pre>
-   *
-   * <code>string maybe_type_name = 5;</code>
-   */
-  public com.google.protobuf.ByteString
-      getMaybeTypeNameBytes() {
-    java.lang.Object ref = maybeTypeName_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      maybeTypeName_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int DTYPES_AND_SHAPES_FIELD_NUMBER = 6;
-  private java.util.List<org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShape> dtypesAndShapes_;
-  /**
-   * <pre>
-   * Data types and shapes for the underlying resource.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.ResourceHandleProto.DtypeAndShape dtypes_and_shapes = 6;</code>
-   */
-  public java.util.List<org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShape> getDtypesAndShapesList() {
-    return dtypesAndShapes_;
-  }
-  /**
-   * <pre>
-   * Data types and shapes for the underlying resource.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.ResourceHandleProto.DtypeAndShape dtypes_and_shapes = 6;</code>
-   */
-  public java.util.List<? extends org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShapeOrBuilder> 
-      getDtypesAndShapesOrBuilderList() {
-    return dtypesAndShapes_;
-  }
-  /**
-   * <pre>
-   * Data types and shapes for the underlying resource.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.ResourceHandleProto.DtypeAndShape dtypes_and_shapes = 6;</code>
-   */
-  public int getDtypesAndShapesCount() {
-    return dtypesAndShapes_.size();
-  }
-  /**
-   * <pre>
-   * Data types and shapes for the underlying resource.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.ResourceHandleProto.DtypeAndShape dtypes_and_shapes = 6;</code>
-   */
-  public org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShape getDtypesAndShapes(int index) {
-    return dtypesAndShapes_.get(index);
-  }
-  /**
-   * <pre>
-   * Data types and shapes for the underlying resource.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.ResourceHandleProto.DtypeAndShape dtypes_and_shapes = 6;</code>
-   */
-  public org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShapeOrBuilder getDtypesAndShapesOrBuilder(
-      int index) {
-    return dtypesAndShapes_.get(index);
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    if (!getDeviceBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 1, device_);
-    }
-    if (!getContainerBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 2, container_);
-    }
-    if (!getNameBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 3, name_);
-    }
-    if (hashCode_ != 0L) {
-      output.writeUInt64(4, hashCode_);
-    }
-    if (!getMaybeTypeNameBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 5, maybeTypeName_);
-    }
-    for (int i = 0; i < dtypesAndShapes_.size(); i++) {
-      output.writeMessage(6, dtypesAndShapes_.get(i));
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (!getDeviceBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, device_);
-    }
-    if (!getContainerBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(2, container_);
-    }
-    if (!getNameBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(3, name_);
-    }
-    if (hashCode_ != 0L) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeUInt64Size(4, hashCode_);
-    }
-    if (!getMaybeTypeNameBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(5, maybeTypeName_);
-    }
-    for (int i = 0; i < dtypesAndShapes_.size(); i++) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(6, dtypesAndShapes_.get(i));
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.framework.ResourceHandleProto)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.framework.ResourceHandleProto other = (org.tensorflow.proto.framework.ResourceHandleProto) obj;
-
-    if (!getDevice()
-        .equals(other.getDevice())) return false;
-    if (!getContainer()
-        .equals(other.getContainer())) return false;
-    if (!getName()
-        .equals(other.getName())) return false;
-    if (getHashCode()
-        != other.getHashCode()) return false;
-    if (!getMaybeTypeName()
-        .equals(other.getMaybeTypeName())) return false;
-    if (!getDtypesAndShapesList()
-        .equals(other.getDtypesAndShapesList())) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    hash = (37 * hash) + DEVICE_FIELD_NUMBER;
-    hash = (53 * hash) + getDevice().hashCode();
-    hash = (37 * hash) + CONTAINER_FIELD_NUMBER;
-    hash = (53 * hash) + getContainer().hashCode();
-    hash = (37 * hash) + NAME_FIELD_NUMBER;
-    hash = (53 * hash) + getName().hashCode();
-    hash = (37 * hash) + HASH_CODE_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-        getHashCode());
-    hash = (37 * hash) + MAYBE_TYPE_NAME_FIELD_NUMBER;
-    hash = (53 * hash) + getMaybeTypeName().hashCode();
-    if (getDtypesAndShapesCount() > 0) {
-      hash = (37 * hash) + DTYPES_AND_SHAPES_FIELD_NUMBER;
-      hash = (53 * hash) + getDtypesAndShapesList().hashCode();
-    }
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.framework.ResourceHandleProto parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.ResourceHandleProto parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.ResourceHandleProto parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.ResourceHandleProto parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.ResourceHandleProto parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.ResourceHandleProto parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.ResourceHandleProto parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.ResourceHandleProto parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.ResourceHandleProto parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.ResourceHandleProto parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.ResourceHandleProto parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.ResourceHandleProto parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.framework.ResourceHandleProto prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * <pre>
-   * Protocol buffer representing a handle to a tensorflow resource. Handles are
-   * not valid across executions, but can be serialized back and forth from within
-   * a single run.
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.ResourceHandleProto}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.ResourceHandleProto)
-      org.tensorflow.proto.framework.ResourceHandleProtoOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.ResourceHandle.internal_static_tensorflow_ResourceHandleProto_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.ResourceHandle.internal_static_tensorflow_ResourceHandleProto_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.ResourceHandleProto.class, org.tensorflow.proto.framework.ResourceHandleProto.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.framework.ResourceHandleProto.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-        getDtypesAndShapesFieldBuilder();
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      device_ = "";
-
-      container_ = "";
-
-      name_ = "";
-
-      hashCode_ = 0L;
-
-      maybeTypeName_ = "";
-
-      if (dtypesAndShapesBuilder_ == null) {
-        dtypesAndShapes_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000001);
-      } else {
-        dtypesAndShapesBuilder_.clear();
-      }
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.framework.ResourceHandle.internal_static_tensorflow_ResourceHandleProto_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.ResourceHandleProto getDefaultInstanceForType() {
-      return org.tensorflow.proto.framework.ResourceHandleProto.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.ResourceHandleProto build() {
-      org.tensorflow.proto.framework.ResourceHandleProto result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.ResourceHandleProto buildPartial() {
-      org.tensorflow.proto.framework.ResourceHandleProto result = new org.tensorflow.proto.framework.ResourceHandleProto(this);
-      int from_bitField0_ = bitField0_;
-      result.device_ = device_;
-      result.container_ = container_;
-      result.name_ = name_;
-      result.hashCode_ = hashCode_;
-      result.maybeTypeName_ = maybeTypeName_;
-      if (dtypesAndShapesBuilder_ == null) {
-        if (((bitField0_ & 0x00000001) != 0)) {
-          dtypesAndShapes_ = java.util.Collections.unmodifiableList(dtypesAndShapes_);
-          bitField0_ = (bitField0_ & ~0x00000001);
-        }
-        result.dtypesAndShapes_ = dtypesAndShapes_;
-      } else {
-        result.dtypesAndShapes_ = dtypesAndShapesBuilder_.build();
-      }
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.framework.ResourceHandleProto) {
-        return mergeFrom((org.tensorflow.proto.framework.ResourceHandleProto)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.framework.ResourceHandleProto other) {
-      if (other == org.tensorflow.proto.framework.ResourceHandleProto.getDefaultInstance()) return this;
-      if (!other.getDevice().isEmpty()) {
-        device_ = other.device_;
-        onChanged();
-      }
-      if (!other.getContainer().isEmpty()) {
-        container_ = other.container_;
-        onChanged();
-      }
-      if (!other.getName().isEmpty()) {
-        name_ = other.name_;
-        onChanged();
-      }
-      if (other.getHashCode() != 0L) {
-        setHashCode(other.getHashCode());
-      }
-      if (!other.getMaybeTypeName().isEmpty()) {
-        maybeTypeName_ = other.maybeTypeName_;
-        onChanged();
-      }
-      if (dtypesAndShapesBuilder_ == null) {
-        if (!other.dtypesAndShapes_.isEmpty()) {
-          if (dtypesAndShapes_.isEmpty()) {
-            dtypesAndShapes_ = other.dtypesAndShapes_;
-            bitField0_ = (bitField0_ & ~0x00000001);
-          } else {
-            ensureDtypesAndShapesIsMutable();
-            dtypesAndShapes_.addAll(other.dtypesAndShapes_);
-          }
-          onChanged();
-        }
-      } else {
-        if (!other.dtypesAndShapes_.isEmpty()) {
-          if (dtypesAndShapesBuilder_.isEmpty()) {
-            dtypesAndShapesBuilder_.dispose();
-            dtypesAndShapesBuilder_ = null;
-            dtypesAndShapes_ = other.dtypesAndShapes_;
-            bitField0_ = (bitField0_ & ~0x00000001);
-            dtypesAndShapesBuilder_ = 
-              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
-                 getDtypesAndShapesFieldBuilder() : null;
-          } else {
-            dtypesAndShapesBuilder_.addAllMessages(other.dtypesAndShapes_);
-          }
-        }
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.framework.ResourceHandleProto parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.framework.ResourceHandleProto) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-    private int bitField0_;
-
-    private java.lang.Object device_ = "";
-    /**
-     * <pre>
-     * Unique name for the device containing the resource.
-     * </pre>
-     *
-     * <code>string device = 1;</code>
-     */
-    public java.lang.String getDevice() {
-      java.lang.Object ref = device_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        device_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Unique name for the device containing the resource.
-     * </pre>
-     *
-     * <code>string device = 1;</code>
-     */
-    public com.google.protobuf.ByteString
-        getDeviceBytes() {
-      java.lang.Object ref = device_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        device_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Unique name for the device containing the resource.
-     * </pre>
-     *
-     * <code>string device = 1;</code>
-     */
-    public Builder setDevice(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      device_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Unique name for the device containing the resource.
-     * </pre>
-     *
-     * <code>string device = 1;</code>
-     */
-    public Builder clearDevice() {
-      
-      device_ = getDefaultInstance().getDevice();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Unique name for the device containing the resource.
-     * </pre>
-     *
-     * <code>string device = 1;</code>
-     */
-    public Builder setDeviceBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      device_ = value;
-      onChanged();
-      return this;
-    }
-
-    private java.lang.Object container_ = "";
-    /**
-     * <pre>
-     * Container in which this resource is placed.
-     * </pre>
-     *
-     * <code>string container = 2;</code>
-     */
-    public java.lang.String getContainer() {
-      java.lang.Object ref = container_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        container_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Container in which this resource is placed.
-     * </pre>
-     *
-     * <code>string container = 2;</code>
-     */
-    public com.google.protobuf.ByteString
-        getContainerBytes() {
-      java.lang.Object ref = container_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        container_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Container in which this resource is placed.
-     * </pre>
-     *
-     * <code>string container = 2;</code>
-     */
-    public Builder setContainer(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      container_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Container in which this resource is placed.
-     * </pre>
-     *
-     * <code>string container = 2;</code>
-     */
-    public Builder clearContainer() {
-      
-      container_ = getDefaultInstance().getContainer();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Container in which this resource is placed.
-     * </pre>
-     *
-     * <code>string container = 2;</code>
-     */
-    public Builder setContainerBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      container_ = value;
-      onChanged();
-      return this;
-    }
-
-    private java.lang.Object name_ = "";
-    /**
-     * <pre>
-     * Unique name of this resource.
-     * </pre>
-     *
-     * <code>string name = 3;</code>
-     */
-    public java.lang.String getName() {
-      java.lang.Object ref = name_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        name_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Unique name of this resource.
-     * </pre>
-     *
-     * <code>string name = 3;</code>
-     */
-    public com.google.protobuf.ByteString
-        getNameBytes() {
-      java.lang.Object ref = name_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        name_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Unique name of this resource.
-     * </pre>
-     *
-     * <code>string name = 3;</code>
-     */
-    public Builder setName(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      name_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Unique name of this resource.
-     * </pre>
-     *
-     * <code>string name = 3;</code>
-     */
-    public Builder clearName() {
-      
-      name_ = getDefaultInstance().getName();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Unique name of this resource.
-     * </pre>
-     *
-     * <code>string name = 3;</code>
-     */
-    public Builder setNameBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      name_ = value;
-      onChanged();
-      return this;
-    }
-
-    private long hashCode_ ;
-    /**
-     * <pre>
-     * Hash code for the type of the resource. Is only valid in the same device
-     * and in the same execution.
-     * </pre>
-     *
-     * <code>uint64 hash_code = 4;</code>
-     */
-    public long getHashCode() {
-      return hashCode_;
-    }
-    /**
-     * <pre>
-     * Hash code for the type of the resource. Is only valid in the same device
-     * and in the same execution.
-     * </pre>
-     *
-     * <code>uint64 hash_code = 4;</code>
-     */
-    public Builder setHashCode(long value) {
-      
-      hashCode_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Hash code for the type of the resource. Is only valid in the same device
-     * and in the same execution.
-     * </pre>
-     *
-     * <code>uint64 hash_code = 4;</code>
-     */
-    public Builder clearHashCode() {
-      
-      hashCode_ = 0L;
-      onChanged();
-      return this;
-    }
-
-    private java.lang.Object maybeTypeName_ = "";
-    /**
-     * <pre>
-     * For debug-only, the name of the type pointed to by this handle, if
-     * available.
-     * </pre>
-     *
-     * <code>string maybe_type_name = 5;</code>
-     */
-    public java.lang.String getMaybeTypeName() {
-      java.lang.Object ref = maybeTypeName_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        maybeTypeName_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * For debug-only, the name of the type pointed to by this handle, if
-     * available.
-     * </pre>
-     *
-     * <code>string maybe_type_name = 5;</code>
-     */
-    public com.google.protobuf.ByteString
-        getMaybeTypeNameBytes() {
-      java.lang.Object ref = maybeTypeName_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        maybeTypeName_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * For debug-only, the name of the type pointed to by this handle, if
-     * available.
-     * </pre>
-     *
-     * <code>string maybe_type_name = 5;</code>
-     */
-    public Builder setMaybeTypeName(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      maybeTypeName_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * For debug-only, the name of the type pointed to by this handle, if
-     * available.
-     * </pre>
-     *
-     * <code>string maybe_type_name = 5;</code>
-     */
-    public Builder clearMaybeTypeName() {
-      
-      maybeTypeName_ = getDefaultInstance().getMaybeTypeName();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * For debug-only, the name of the type pointed to by this handle, if
-     * available.
-     * </pre>
-     *
-     * <code>string maybe_type_name = 5;</code>
-     */
-    public Builder setMaybeTypeNameBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      maybeTypeName_ = value;
-      onChanged();
-      return this;
-    }
-
-    private java.util.List<org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShape> dtypesAndShapes_ =
-      java.util.Collections.emptyList();
-    private void ensureDtypesAndShapesIsMutable() {
-      if (!((bitField0_ & 0x00000001) != 0)) {
-        dtypesAndShapes_ = new java.util.ArrayList<org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShape>(dtypesAndShapes_);
-        bitField0_ |= 0x00000001;
-       }
-    }
-
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShape, org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShape.Builder, org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShapeOrBuilder> dtypesAndShapesBuilder_;
-
-    /**
-     * <pre>
-     * Data types and shapes for the underlying resource.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.ResourceHandleProto.DtypeAndShape dtypes_and_shapes = 6;</code>
-     */
-    public java.util.List<org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShape> getDtypesAndShapesList() {
-      if (dtypesAndShapesBuilder_ == null) {
-        return java.util.Collections.unmodifiableList(dtypesAndShapes_);
-      } else {
-        return dtypesAndShapesBuilder_.getMessageList();
-      }
-    }
-    /**
-     * <pre>
-     * Data types and shapes for the underlying resource.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.ResourceHandleProto.DtypeAndShape dtypes_and_shapes = 6;</code>
-     */
-    public int getDtypesAndShapesCount() {
-      if (dtypesAndShapesBuilder_ == null) {
-        return dtypesAndShapes_.size();
-      } else {
-        return dtypesAndShapesBuilder_.getCount();
-      }
-    }
-    /**
-     * <pre>
-     * Data types and shapes for the underlying resource.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.ResourceHandleProto.DtypeAndShape dtypes_and_shapes = 6;</code>
-     */
-    public org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShape getDtypesAndShapes(int index) {
-      if (dtypesAndShapesBuilder_ == null) {
-        return dtypesAndShapes_.get(index);
-      } else {
-        return dtypesAndShapesBuilder_.getMessage(index);
-      }
-    }
-    /**
-     * <pre>
-     * Data types and shapes for the underlying resource.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.ResourceHandleProto.DtypeAndShape dtypes_and_shapes = 6;</code>
-     */
-    public Builder setDtypesAndShapes(
-        int index, org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShape value) {
-      if (dtypesAndShapesBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureDtypesAndShapesIsMutable();
-        dtypesAndShapes_.set(index, value);
-        onChanged();
-      } else {
-        dtypesAndShapesBuilder_.setMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Data types and shapes for the underlying resource.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.ResourceHandleProto.DtypeAndShape dtypes_and_shapes = 6;</code>
-     */
-    public Builder setDtypesAndShapes(
-        int index, org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShape.Builder builderForValue) {
-      if (dtypesAndShapesBuilder_ == null) {
-        ensureDtypesAndShapesIsMutable();
-        dtypesAndShapes_.set(index, builderForValue.build());
-        onChanged();
-      } else {
-        dtypesAndShapesBuilder_.setMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Data types and shapes for the underlying resource.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.ResourceHandleProto.DtypeAndShape dtypes_and_shapes = 6;</code>
-     */
-    public Builder addDtypesAndShapes(org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShape value) {
-      if (dtypesAndShapesBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureDtypesAndShapesIsMutable();
-        dtypesAndShapes_.add(value);
-        onChanged();
-      } else {
-        dtypesAndShapesBuilder_.addMessage(value);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Data types and shapes for the underlying resource.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.ResourceHandleProto.DtypeAndShape dtypes_and_shapes = 6;</code>
-     */
-    public Builder addDtypesAndShapes(
-        int index, org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShape value) {
-      if (dtypesAndShapesBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureDtypesAndShapesIsMutable();
-        dtypesAndShapes_.add(index, value);
-        onChanged();
-      } else {
-        dtypesAndShapesBuilder_.addMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Data types and shapes for the underlying resource.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.ResourceHandleProto.DtypeAndShape dtypes_and_shapes = 6;</code>
-     */
-    public Builder addDtypesAndShapes(
-        org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShape.Builder builderForValue) {
-      if (dtypesAndShapesBuilder_ == null) {
-        ensureDtypesAndShapesIsMutable();
-        dtypesAndShapes_.add(builderForValue.build());
-        onChanged();
-      } else {
-        dtypesAndShapesBuilder_.addMessage(builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Data types and shapes for the underlying resource.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.ResourceHandleProto.DtypeAndShape dtypes_and_shapes = 6;</code>
-     */
-    public Builder addDtypesAndShapes(
-        int index, org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShape.Builder builderForValue) {
-      if (dtypesAndShapesBuilder_ == null) {
-        ensureDtypesAndShapesIsMutable();
-        dtypesAndShapes_.add(index, builderForValue.build());
-        onChanged();
-      } else {
-        dtypesAndShapesBuilder_.addMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Data types and shapes for the underlying resource.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.ResourceHandleProto.DtypeAndShape dtypes_and_shapes = 6;</code>
-     */
-    public Builder addAllDtypesAndShapes(
-        java.lang.Iterable<? extends org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShape> values) {
-      if (dtypesAndShapesBuilder_ == null) {
-        ensureDtypesAndShapesIsMutable();
-        com.google.protobuf.AbstractMessageLite.Builder.addAll(
-            values, dtypesAndShapes_);
-        onChanged();
-      } else {
-        dtypesAndShapesBuilder_.addAllMessages(values);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Data types and shapes for the underlying resource.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.ResourceHandleProto.DtypeAndShape dtypes_and_shapes = 6;</code>
-     */
-    public Builder clearDtypesAndShapes() {
-      if (dtypesAndShapesBuilder_ == null) {
-        dtypesAndShapes_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000001);
-        onChanged();
-      } else {
-        dtypesAndShapesBuilder_.clear();
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Data types and shapes for the underlying resource.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.ResourceHandleProto.DtypeAndShape dtypes_and_shapes = 6;</code>
-     */
-    public Builder removeDtypesAndShapes(int index) {
-      if (dtypesAndShapesBuilder_ == null) {
-        ensureDtypesAndShapesIsMutable();
-        dtypesAndShapes_.remove(index);
-        onChanged();
-      } else {
-        dtypesAndShapesBuilder_.remove(index);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Data types and shapes for the underlying resource.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.ResourceHandleProto.DtypeAndShape dtypes_and_shapes = 6;</code>
-     */
-    public org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShape.Builder getDtypesAndShapesBuilder(
-        int index) {
-      return getDtypesAndShapesFieldBuilder().getBuilder(index);
-    }
-    /**
-     * <pre>
-     * Data types and shapes for the underlying resource.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.ResourceHandleProto.DtypeAndShape dtypes_and_shapes = 6;</code>
-     */
-    public org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShapeOrBuilder getDtypesAndShapesOrBuilder(
-        int index) {
-      if (dtypesAndShapesBuilder_ == null) {
-        return dtypesAndShapes_.get(index);  } else {
-        return dtypesAndShapesBuilder_.getMessageOrBuilder(index);
-      }
-    }
-    /**
-     * <pre>
-     * Data types and shapes for the underlying resource.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.ResourceHandleProto.DtypeAndShape dtypes_and_shapes = 6;</code>
-     */
-    public java.util.List<? extends org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShapeOrBuilder> 
-         getDtypesAndShapesOrBuilderList() {
-      if (dtypesAndShapesBuilder_ != null) {
-        return dtypesAndShapesBuilder_.getMessageOrBuilderList();
-      } else {
-        return java.util.Collections.unmodifiableList(dtypesAndShapes_);
-      }
-    }
-    /**
-     * <pre>
-     * Data types and shapes for the underlying resource.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.ResourceHandleProto.DtypeAndShape dtypes_and_shapes = 6;</code>
-     */
-    public org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShape.Builder addDtypesAndShapesBuilder() {
-      return getDtypesAndShapesFieldBuilder().addBuilder(
-          org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShape.getDefaultInstance());
-    }
-    /**
-     * <pre>
-     * Data types and shapes for the underlying resource.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.ResourceHandleProto.DtypeAndShape dtypes_and_shapes = 6;</code>
-     */
-    public org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShape.Builder addDtypesAndShapesBuilder(
-        int index) {
-      return getDtypesAndShapesFieldBuilder().addBuilder(
-          index, org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShape.getDefaultInstance());
-    }
-    /**
-     * <pre>
-     * Data types and shapes for the underlying resource.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.ResourceHandleProto.DtypeAndShape dtypes_and_shapes = 6;</code>
-     */
-    public java.util.List<org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShape.Builder> 
-         getDtypesAndShapesBuilderList() {
-      return getDtypesAndShapesFieldBuilder().getBuilderList();
-    }
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShape, org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShape.Builder, org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShapeOrBuilder> 
-        getDtypesAndShapesFieldBuilder() {
-      if (dtypesAndShapesBuilder_ == null) {
-        dtypesAndShapesBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
-            org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShape, org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShape.Builder, org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShapeOrBuilder>(
-                dtypesAndShapes_,
-                ((bitField0_ & 0x00000001) != 0),
-                getParentForChildren(),
-                isClean());
-        dtypesAndShapes_ = null;
-      }
-      return dtypesAndShapesBuilder_;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.ResourceHandleProto)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.ResourceHandleProto)
-  private static final org.tensorflow.proto.framework.ResourceHandleProto DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.ResourceHandleProto();
-  }
-
-  public static org.tensorflow.proto.framework.ResourceHandleProto getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<ResourceHandleProto>
-      PARSER = new com.google.protobuf.AbstractParser<ResourceHandleProto>() {
-    @java.lang.Override
-    public ResourceHandleProto parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new ResourceHandleProto(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<ResourceHandleProto> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<ResourceHandleProto> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.framework.ResourceHandleProto getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ResourceHandleProtoOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ResourceHandleProtoOrBuilder.java
deleted file mode 100644
index 461b2de1dba..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ResourceHandleProtoOrBuilder.java
+++ /dev/null
@@ -1,137 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/framework/resource_handle.proto
-
-package org.tensorflow.proto.framework;
-
-public interface ResourceHandleProtoOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.ResourceHandleProto)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <pre>
-   * Unique name for the device containing the resource.
-   * </pre>
-   *
-   * <code>string device = 1;</code>
-   */
-  java.lang.String getDevice();
-  /**
-   * <pre>
-   * Unique name for the device containing the resource.
-   * </pre>
-   *
-   * <code>string device = 1;</code>
-   */
-  com.google.protobuf.ByteString
-      getDeviceBytes();
-
-  /**
-   * <pre>
-   * Container in which this resource is placed.
-   * </pre>
-   *
-   * <code>string container = 2;</code>
-   */
-  java.lang.String getContainer();
-  /**
-   * <pre>
-   * Container in which this resource is placed.
-   * </pre>
-   *
-   * <code>string container = 2;</code>
-   */
-  com.google.protobuf.ByteString
-      getContainerBytes();
-
-  /**
-   * <pre>
-   * Unique name of this resource.
-   * </pre>
-   *
-   * <code>string name = 3;</code>
-   */
-  java.lang.String getName();
-  /**
-   * <pre>
-   * Unique name of this resource.
-   * </pre>
-   *
-   * <code>string name = 3;</code>
-   */
-  com.google.protobuf.ByteString
-      getNameBytes();
-
-  /**
-   * <pre>
-   * Hash code for the type of the resource. Is only valid in the same device
-   * and in the same execution.
-   * </pre>
-   *
-   * <code>uint64 hash_code = 4;</code>
-   */
-  long getHashCode();
-
-  /**
-   * <pre>
-   * For debug-only, the name of the type pointed to by this handle, if
-   * available.
-   * </pre>
-   *
-   * <code>string maybe_type_name = 5;</code>
-   */
-  java.lang.String getMaybeTypeName();
-  /**
-   * <pre>
-   * For debug-only, the name of the type pointed to by this handle, if
-   * available.
-   * </pre>
-   *
-   * <code>string maybe_type_name = 5;</code>
-   */
-  com.google.protobuf.ByteString
-      getMaybeTypeNameBytes();
-
-  /**
-   * <pre>
-   * Data types and shapes for the underlying resource.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.ResourceHandleProto.DtypeAndShape dtypes_and_shapes = 6;</code>
-   */
-  java.util.List<org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShape> 
-      getDtypesAndShapesList();
-  /**
-   * <pre>
-   * Data types and shapes for the underlying resource.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.ResourceHandleProto.DtypeAndShape dtypes_and_shapes = 6;</code>
-   */
-  org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShape getDtypesAndShapes(int index);
-  /**
-   * <pre>
-   * Data types and shapes for the underlying resource.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.ResourceHandleProto.DtypeAndShape dtypes_and_shapes = 6;</code>
-   */
-  int getDtypesAndShapesCount();
-  /**
-   * <pre>
-   * Data types and shapes for the underlying resource.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.ResourceHandleProto.DtypeAndShape dtypes_and_shapes = 6;</code>
-   */
-  java.util.List<? extends org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShapeOrBuilder> 
-      getDtypesAndShapesOrBuilderList();
-  /**
-   * <pre>
-   * Data types and shapes for the underlying resource.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.ResourceHandleProto.DtypeAndShape dtypes_and_shapes = 6;</code>
-   */
-  org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShapeOrBuilder getDtypesAndShapesOrBuilder(
-      int index);
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ResourceOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ResourceOrBuilder.java
deleted file mode 100644
index f6bc0c0d190..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ResourceOrBuilder.java
+++ /dev/null
@@ -1,36 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/trace_events.proto
-
-package org.tensorflow.proto.framework;
-
-public interface ResourceOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.profiler.Resource)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <pre>
-   * The name of the resource.
-   * </pre>
-   *
-   * <code>string name = 1;</code>
-   */
-  java.lang.String getName();
-  /**
-   * <pre>
-   * The name of the resource.
-   * </pre>
-   *
-   * <code>string name = 1;</code>
-   */
-  com.google.protobuf.ByteString
-      getNameBytes();
-
-  /**
-   * <pre>
-   * The id of the resource. Unique within a device.
-   * </pre>
-   *
-   * <code>uint32 resource_id = 2;</code>
-   */
-  int getResourceId();
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/RewriterConfig.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/RewriterConfig.java
deleted file mode 100644
index fe1571856ee..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/RewriterConfig.java
+++ /dev/null
@@ -1,6064 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/rewriter_config.proto
-
-package org.tensorflow.proto.framework;
-
-/**
- * <pre>
- * Graph rewriting is experimental and subject to change, not covered by any
- * API stability guarantees.
- * </pre>
- *
- * Protobuf type {@code tensorflow.RewriterConfig}
- */
-public  final class RewriterConfig extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.RewriterConfig)
-    RewriterConfigOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use RewriterConfig.newBuilder() to construct.
-  private RewriterConfig(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private RewriterConfig() {
-    layoutOptimizer_ = 0;
-    constantFolding_ = 0;
-    shapeOptimization_ = 0;
-    remapping_ = 0;
-    commonSubgraphElimination_ = 0;
-    arithmeticOptimization_ = 0;
-    dependencyOptimization_ = 0;
-    loopOptimization_ = 0;
-    functionOptimization_ = 0;
-    debugStripper_ = 0;
-    scopedAllocatorOptimization_ = 0;
-    pinToHostOptimization_ = 0;
-    implementationSelector_ = 0;
-    autoMixedPrecision_ = 0;
-    autoMixedPrecisionMkl_ = 0;
-    metaOptimizerIterations_ = 0;
-    memoryOptimization_ = 0;
-    memoryOptimizerTargetNodeNameScope_ = "";
-    optimizers_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-    customOptimizers_ = java.util.Collections.emptyList();
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new RewriterConfig();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private RewriterConfig(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    int mutable_bitField0_ = 0;
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 8: {
-            int rawValue = input.readEnum();
-
-            layoutOptimizer_ = rawValue;
-            break;
-          }
-          case 16: {
-
-            disableModelPruning_ = input.readBool();
-            break;
-          }
-          case 24: {
-            int rawValue = input.readEnum();
-
-            constantFolding_ = rawValue;
-            break;
-          }
-          case 32: {
-            int rawValue = input.readEnum();
-
-            memoryOptimization_ = rawValue;
-            break;
-          }
-          case 42: {
-            org.tensorflow.proto.framework.AutoParallelOptions.Builder subBuilder = null;
-            if (autoParallel_ != null) {
-              subBuilder = autoParallel_.toBuilder();
-            }
-            autoParallel_ = input.readMessage(org.tensorflow.proto.framework.AutoParallelOptions.parser(), extensionRegistry);
-            if (subBuilder != null) {
-              subBuilder.mergeFrom(autoParallel_);
-              autoParallel_ = subBuilder.buildPartial();
-            }
-
-            break;
-          }
-          case 50: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            memoryOptimizerTargetNodeNameScope_ = s;
-            break;
-          }
-          case 56: {
-            int rawValue = input.readEnum();
-
-            arithmeticOptimization_ = rawValue;
-            break;
-          }
-          case 64: {
-            int rawValue = input.readEnum();
-
-            dependencyOptimization_ = rawValue;
-            break;
-          }
-          case 72: {
-            int rawValue = input.readEnum();
-
-            loopOptimization_ = rawValue;
-            break;
-          }
-          case 80: {
-            int rawValue = input.readEnum();
-
-            functionOptimization_ = rawValue;
-            break;
-          }
-          case 88: {
-            int rawValue = input.readEnum();
-
-            debugStripper_ = rawValue;
-            break;
-          }
-          case 96: {
-            int rawValue = input.readEnum();
-
-            metaOptimizerIterations_ = rawValue;
-            break;
-          }
-          case 104: {
-            int rawValue = input.readEnum();
-
-            shapeOptimization_ = rawValue;
-            break;
-          }
-          case 112: {
-            int rawValue = input.readEnum();
-
-            remapping_ = rawValue;
-            break;
-          }
-          case 120: {
-            int rawValue = input.readEnum();
-
-            scopedAllocatorOptimization_ = rawValue;
-            break;
-          }
-          case 130: {
-            org.tensorflow.proto.framework.ScopedAllocatorOptions.Builder subBuilder = null;
-            if (scopedAllocatorOpts_ != null) {
-              subBuilder = scopedAllocatorOpts_.toBuilder();
-            }
-            scopedAllocatorOpts_ = input.readMessage(org.tensorflow.proto.framework.ScopedAllocatorOptions.parser(), extensionRegistry);
-            if (subBuilder != null) {
-              subBuilder.mergeFrom(scopedAllocatorOpts_);
-              scopedAllocatorOpts_ = subBuilder.buildPartial();
-            }
-
-            break;
-          }
-          case 136: {
-
-            minGraphNodes_ = input.readInt32();
-            break;
-          }
-          case 144: {
-            int rawValue = input.readEnum();
-
-            pinToHostOptimization_ = rawValue;
-            break;
-          }
-          case 152: {
-
-            disableMetaOptimizer_ = input.readBool();
-            break;
-          }
-          case 160: {
-
-            metaOptimizerTimeoutMs_ = input.readInt64();
-            break;
-          }
-          case 168: {
-
-            failOnOptimizerErrors_ = input.readBool();
-            break;
-          }
-          case 176: {
-            int rawValue = input.readEnum();
-
-            implementationSelector_ = rawValue;
-            break;
-          }
-          case 184: {
-            int rawValue = input.readEnum();
-
-            autoMixedPrecision_ = rawValue;
-            break;
-          }
-          case 192: {
-            int rawValue = input.readEnum();
-
-            commonSubgraphElimination_ = rawValue;
-            break;
-          }
-          case 200: {
-            int rawValue = input.readEnum();
-
-            autoMixedPrecisionMkl_ = rawValue;
-            break;
-          }
-          case 802: {
-            java.lang.String s = input.readStringRequireUtf8();
-            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
-              optimizers_ = new com.google.protobuf.LazyStringArrayList();
-              mutable_bitField0_ |= 0x00000001;
-            }
-            optimizers_.add(s);
-            break;
-          }
-          case 1602: {
-            if (!((mutable_bitField0_ & 0x00000002) != 0)) {
-              customOptimizers_ = new java.util.ArrayList<org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizer>();
-              mutable_bitField0_ |= 0x00000002;
-            }
-            customOptimizers_.add(
-                input.readMessage(org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizer.parser(), extensionRegistry));
-            break;
-          }
-          case 2402: {
-            org.tensorflow.proto.framework.VerifierConfig.Builder subBuilder = null;
-            if (interOptimizerVerifierConfig_ != null) {
-              subBuilder = interOptimizerVerifierConfig_.toBuilder();
-            }
-            interOptimizerVerifierConfig_ = input.readMessage(org.tensorflow.proto.framework.VerifierConfig.parser(), extensionRegistry);
-            if (subBuilder != null) {
-              subBuilder.mergeFrom(interOptimizerVerifierConfig_);
-              interOptimizerVerifierConfig_ = subBuilder.buildPartial();
-            }
-
-            break;
-          }
-          case 2410: {
-            org.tensorflow.proto.framework.VerifierConfig.Builder subBuilder = null;
-            if (postOptimizationVerifierConfig_ != null) {
-              subBuilder = postOptimizationVerifierConfig_.toBuilder();
-            }
-            postOptimizationVerifierConfig_ = input.readMessage(org.tensorflow.proto.framework.VerifierConfig.parser(), extensionRegistry);
-            if (subBuilder != null) {
-              subBuilder.mergeFrom(postOptimizationVerifierConfig_);
-              postOptimizationVerifierConfig_ = subBuilder.buildPartial();
-            }
-
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      if (((mutable_bitField0_ & 0x00000001) != 0)) {
-        optimizers_ = optimizers_.getUnmodifiableView();
-      }
-      if (((mutable_bitField0_ & 0x00000002) != 0)) {
-        customOptimizers_ = java.util.Collections.unmodifiableList(customOptimizers_);
-      }
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.framework.RewriterConfigProtos.internal_static_tensorflow_RewriterConfig_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.framework.RewriterConfigProtos.internal_static_tensorflow_RewriterConfig_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.framework.RewriterConfig.class, org.tensorflow.proto.framework.RewriterConfig.Builder.class);
-  }
-
-  /**
-   * Protobuf enum {@code tensorflow.RewriterConfig.Toggle}
-   */
-  public enum Toggle
-      implements com.google.protobuf.ProtocolMessageEnum {
-    /**
-     * <code>DEFAULT = 0;</code>
-     */
-    DEFAULT(0),
-    /**
-     * <code>ON = 1;</code>
-     */
-    ON(1),
-    /**
-     * <code>OFF = 2;</code>
-     */
-    OFF(2),
-    /**
-     * <pre>
-     * Enable some aggressive optimizations that use assumptions that TF graphs
-     * may break. For example, assume the shape of a placeholder matches its
-     * actual feed.
-     * </pre>
-     *
-     * <code>AGGRESSIVE = 3;</code>
-     */
-    AGGRESSIVE(3),
-    UNRECOGNIZED(-1),
-    ;
-
-    /**
-     * <code>DEFAULT = 0;</code>
-     */
-    public static final int DEFAULT_VALUE = 0;
-    /**
-     * <code>ON = 1;</code>
-     */
-    public static final int ON_VALUE = 1;
-    /**
-     * <code>OFF = 2;</code>
-     */
-    public static final int OFF_VALUE = 2;
-    /**
-     * <pre>
-     * Enable some aggressive optimizations that use assumptions that TF graphs
-     * may break. For example, assume the shape of a placeholder matches its
-     * actual feed.
-     * </pre>
-     *
-     * <code>AGGRESSIVE = 3;</code>
-     */
-    public static final int AGGRESSIVE_VALUE = 3;
-
-
-    public final int getNumber() {
-      if (this == UNRECOGNIZED) {
-        throw new java.lang.IllegalArgumentException(
-            "Can't get the number of an unknown enum value.");
-      }
-      return value;
-    }
-
-    /**
-     * @deprecated Use {@link #forNumber(int)} instead.
-     */
-    @java.lang.Deprecated
-    public static Toggle valueOf(int value) {
-      return forNumber(value);
-    }
-
-    public static Toggle forNumber(int value) {
-      switch (value) {
-        case 0: return DEFAULT;
-        case 1: return ON;
-        case 2: return OFF;
-        case 3: return AGGRESSIVE;
-        default: return null;
-      }
-    }
-
-    public static com.google.protobuf.Internal.EnumLiteMap<Toggle>
-        internalGetValueMap() {
-      return internalValueMap;
-    }
-    private static final com.google.protobuf.Internal.EnumLiteMap<
-        Toggle> internalValueMap =
-          new com.google.protobuf.Internal.EnumLiteMap<Toggle>() {
-            public Toggle findValueByNumber(int number) {
-              return Toggle.forNumber(number);
-            }
-          };
-
-    public final com.google.protobuf.Descriptors.EnumValueDescriptor
-        getValueDescriptor() {
-      return getDescriptor().getValues().get(ordinal());
-    }
-    public final com.google.protobuf.Descriptors.EnumDescriptor
-        getDescriptorForType() {
-      return getDescriptor();
-    }
-    public static final com.google.protobuf.Descriptors.EnumDescriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.RewriterConfig.getDescriptor().getEnumTypes().get(0);
-    }
-
-    private static final Toggle[] VALUES = values();
-
-    public static Toggle valueOf(
-        com.google.protobuf.Descriptors.EnumValueDescriptor desc) {
-      if (desc.getType() != getDescriptor()) {
-        throw new java.lang.IllegalArgumentException(
-          "EnumValueDescriptor is not for this type.");
-      }
-      if (desc.getIndex() == -1) {
-        return UNRECOGNIZED;
-      }
-      return VALUES[desc.getIndex()];
-    }
-
-    private final int value;
-
-    private Toggle(int value) {
-      this.value = value;
-    }
-
-    // @@protoc_insertion_point(enum_scope:tensorflow.RewriterConfig.Toggle)
-  }
-
-  /**
-   * <pre>
-   * Enum controlling the number of times to run optimizers. The default is to
-   * run them twice.
-   * </pre>
-   *
-   * Protobuf enum {@code tensorflow.RewriterConfig.NumIterationsType}
-   */
-  public enum NumIterationsType
-      implements com.google.protobuf.ProtocolMessageEnum {
-    /**
-     * <code>DEFAULT_NUM_ITERS = 0;</code>
-     */
-    DEFAULT_NUM_ITERS(0),
-    /**
-     * <code>ONE = 1;</code>
-     */
-    ONE(1),
-    /**
-     * <code>TWO = 2;</code>
-     */
-    TWO(2),
-    UNRECOGNIZED(-1),
-    ;
-
-    /**
-     * <code>DEFAULT_NUM_ITERS = 0;</code>
-     */
-    public static final int DEFAULT_NUM_ITERS_VALUE = 0;
-    /**
-     * <code>ONE = 1;</code>
-     */
-    public static final int ONE_VALUE = 1;
-    /**
-     * <code>TWO = 2;</code>
-     */
-    public static final int TWO_VALUE = 2;
-
-
-    public final int getNumber() {
-      if (this == UNRECOGNIZED) {
-        throw new java.lang.IllegalArgumentException(
-            "Can't get the number of an unknown enum value.");
-      }
-      return value;
-    }
-
-    /**
-     * @deprecated Use {@link #forNumber(int)} instead.
-     */
-    @java.lang.Deprecated
-    public static NumIterationsType valueOf(int value) {
-      return forNumber(value);
-    }
-
-    public static NumIterationsType forNumber(int value) {
-      switch (value) {
-        case 0: return DEFAULT_NUM_ITERS;
-        case 1: return ONE;
-        case 2: return TWO;
-        default: return null;
-      }
-    }
-
-    public static com.google.protobuf.Internal.EnumLiteMap<NumIterationsType>
-        internalGetValueMap() {
-      return internalValueMap;
-    }
-    private static final com.google.protobuf.Internal.EnumLiteMap<
-        NumIterationsType> internalValueMap =
-          new com.google.protobuf.Internal.EnumLiteMap<NumIterationsType>() {
-            public NumIterationsType findValueByNumber(int number) {
-              return NumIterationsType.forNumber(number);
-            }
-          };
-
-    public final com.google.protobuf.Descriptors.EnumValueDescriptor
-        getValueDescriptor() {
-      return getDescriptor().getValues().get(ordinal());
-    }
-    public final com.google.protobuf.Descriptors.EnumDescriptor
-        getDescriptorForType() {
-      return getDescriptor();
-    }
-    public static final com.google.protobuf.Descriptors.EnumDescriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.RewriterConfig.getDescriptor().getEnumTypes().get(1);
-    }
-
-    private static final NumIterationsType[] VALUES = values();
-
-    public static NumIterationsType valueOf(
-        com.google.protobuf.Descriptors.EnumValueDescriptor desc) {
-      if (desc.getType() != getDescriptor()) {
-        throw new java.lang.IllegalArgumentException(
-          "EnumValueDescriptor is not for this type.");
-      }
-      if (desc.getIndex() == -1) {
-        return UNRECOGNIZED;
-      }
-      return VALUES[desc.getIndex()];
-    }
-
-    private final int value;
-
-    private NumIterationsType(int value) {
-      this.value = value;
-    }
-
-    // @@protoc_insertion_point(enum_scope:tensorflow.RewriterConfig.NumIterationsType)
-  }
-
-  /**
-   * Protobuf enum {@code tensorflow.RewriterConfig.MemOptType}
-   */
-  public enum MemOptType
-      implements com.google.protobuf.ProtocolMessageEnum {
-    /**
-     * <pre>
-     * The default setting (SCHEDULING and SWAPPING HEURISTICS only)
-     * </pre>
-     *
-     * <code>DEFAULT_MEM_OPT = 0;</code>
-     */
-    DEFAULT_MEM_OPT(0),
-    /**
-     * <pre>
-     * Disabled in the meta-optimizer.
-     * </pre>
-     *
-     * <code>NO_MEM_OPT = 1;</code>
-     */
-    NO_MEM_OPT(1),
-    /**
-     * <pre>
-     * Driven by manual op-level annotations.
-     * </pre>
-     *
-     * <code>MANUAL = 2;</code>
-     */
-    MANUAL(2),
-    /**
-     * <pre>
-     * Swapping heuristic will move a tensor from the GPU to the CPU and move
-     * it back when needed to reduce peak memory usage.
-     * </pre>
-     *
-     * <code>SWAPPING_HEURISTICS = 4;</code>
-     */
-    SWAPPING_HEURISTICS(4),
-    /**
-     * <pre>
-     * Recomputation heuristics will recompute ops (such as Relu activation)
-     * during backprop instead of storing them, reducing peak memory usage.
-     * </pre>
-     *
-     * <code>RECOMPUTATION_HEURISTICS = 5;</code>
-     */
-    RECOMPUTATION_HEURISTICS(5),
-    /**
-     * <pre>
-     * Scheduling will split big ops such as AddN and try to enforce a schedule
-     * of the new computations that decreases peak memory usage.
-     * </pre>
-     *
-     * <code>SCHEDULING_HEURISTICS = 6;</code>
-     */
-    SCHEDULING_HEURISTICS(6),
-    /**
-     * <pre>
-     * Use any combination of swapping and recomputation heuristics.
-     * </pre>
-     *
-     * <code>HEURISTICS = 3;</code>
-     */
-    HEURISTICS(3),
-    UNRECOGNIZED(-1),
-    ;
-
-    /**
-     * <pre>
-     * The default setting (SCHEDULING and SWAPPING HEURISTICS only)
-     * </pre>
-     *
-     * <code>DEFAULT_MEM_OPT = 0;</code>
-     */
-    public static final int DEFAULT_MEM_OPT_VALUE = 0;
-    /**
-     * <pre>
-     * Disabled in the meta-optimizer.
-     * </pre>
-     *
-     * <code>NO_MEM_OPT = 1;</code>
-     */
-    public static final int NO_MEM_OPT_VALUE = 1;
-    /**
-     * <pre>
-     * Driven by manual op-level annotations.
-     * </pre>
-     *
-     * <code>MANUAL = 2;</code>
-     */
-    public static final int MANUAL_VALUE = 2;
-    /**
-     * <pre>
-     * Swapping heuristic will move a tensor from the GPU to the CPU and move
-     * it back when needed to reduce peak memory usage.
-     * </pre>
-     *
-     * <code>SWAPPING_HEURISTICS = 4;</code>
-     */
-    public static final int SWAPPING_HEURISTICS_VALUE = 4;
-    /**
-     * <pre>
-     * Recomputation heuristics will recompute ops (such as Relu activation)
-     * during backprop instead of storing them, reducing peak memory usage.
-     * </pre>
-     *
-     * <code>RECOMPUTATION_HEURISTICS = 5;</code>
-     */
-    public static final int RECOMPUTATION_HEURISTICS_VALUE = 5;
-    /**
-     * <pre>
-     * Scheduling will split big ops such as AddN and try to enforce a schedule
-     * of the new computations that decreases peak memory usage.
-     * </pre>
-     *
-     * <code>SCHEDULING_HEURISTICS = 6;</code>
-     */
-    public static final int SCHEDULING_HEURISTICS_VALUE = 6;
-    /**
-     * <pre>
-     * Use any combination of swapping and recomputation heuristics.
-     * </pre>
-     *
-     * <code>HEURISTICS = 3;</code>
-     */
-    public static final int HEURISTICS_VALUE = 3;
-
-
-    public final int getNumber() {
-      if (this == UNRECOGNIZED) {
-        throw new java.lang.IllegalArgumentException(
-            "Can't get the number of an unknown enum value.");
-      }
-      return value;
-    }
-
-    /**
-     * @deprecated Use {@link #forNumber(int)} instead.
-     */
-    @java.lang.Deprecated
-    public static MemOptType valueOf(int value) {
-      return forNumber(value);
-    }
-
-    public static MemOptType forNumber(int value) {
-      switch (value) {
-        case 0: return DEFAULT_MEM_OPT;
-        case 1: return NO_MEM_OPT;
-        case 2: return MANUAL;
-        case 4: return SWAPPING_HEURISTICS;
-        case 5: return RECOMPUTATION_HEURISTICS;
-        case 6: return SCHEDULING_HEURISTICS;
-        case 3: return HEURISTICS;
-        default: return null;
-      }
-    }
-
-    public static com.google.protobuf.Internal.EnumLiteMap<MemOptType>
-        internalGetValueMap() {
-      return internalValueMap;
-    }
-    private static final com.google.protobuf.Internal.EnumLiteMap<
-        MemOptType> internalValueMap =
-          new com.google.protobuf.Internal.EnumLiteMap<MemOptType>() {
-            public MemOptType findValueByNumber(int number) {
-              return MemOptType.forNumber(number);
-            }
-          };
-
-    public final com.google.protobuf.Descriptors.EnumValueDescriptor
-        getValueDescriptor() {
-      return getDescriptor().getValues().get(ordinal());
-    }
-    public final com.google.protobuf.Descriptors.EnumDescriptor
-        getDescriptorForType() {
-      return getDescriptor();
-    }
-    public static final com.google.protobuf.Descriptors.EnumDescriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.RewriterConfig.getDescriptor().getEnumTypes().get(2);
-    }
-
-    private static final MemOptType[] VALUES = values();
-
-    public static MemOptType valueOf(
-        com.google.protobuf.Descriptors.EnumValueDescriptor desc) {
-      if (desc.getType() != getDescriptor()) {
-        throw new java.lang.IllegalArgumentException(
-          "EnumValueDescriptor is not for this type.");
-      }
-      if (desc.getIndex() == -1) {
-        return UNRECOGNIZED;
-      }
-      return VALUES[desc.getIndex()];
-    }
-
-    private final int value;
-
-    private MemOptType(int value) {
-      this.value = value;
-    }
-
-    // @@protoc_insertion_point(enum_scope:tensorflow.RewriterConfig.MemOptType)
-  }
-
-  public interface CustomGraphOptimizerOrBuilder extends
-      // @@protoc_insertion_point(interface_extends:tensorflow.RewriterConfig.CustomGraphOptimizer)
-      com.google.protobuf.MessageOrBuilder {
-
-    /**
-     * <code>string name = 1;</code>
-     */
-    java.lang.String getName();
-    /**
-     * <code>string name = 1;</code>
-     */
-    com.google.protobuf.ByteString
-        getNameBytes();
-
-    /**
-     * <code>map&lt;string, .tensorflow.AttrValue&gt; parameter_map = 2;</code>
-     */
-    int getParameterMapCount();
-    /**
-     * <code>map&lt;string, .tensorflow.AttrValue&gt; parameter_map = 2;</code>
-     */
-    boolean containsParameterMap(
-        java.lang.String key);
-    /**
-     * Use {@link #getParameterMapMap()} instead.
-     */
-    @java.lang.Deprecated
-    java.util.Map<java.lang.String, org.tensorflow.proto.framework.AttrValue>
-    getParameterMap();
-    /**
-     * <code>map&lt;string, .tensorflow.AttrValue&gt; parameter_map = 2;</code>
-     */
-    java.util.Map<java.lang.String, org.tensorflow.proto.framework.AttrValue>
-    getParameterMapMap();
-    /**
-     * <code>map&lt;string, .tensorflow.AttrValue&gt; parameter_map = 2;</code>
-     */
-
-    org.tensorflow.proto.framework.AttrValue getParameterMapOrDefault(
-        java.lang.String key,
-        org.tensorflow.proto.framework.AttrValue defaultValue);
-    /**
-     * <code>map&lt;string, .tensorflow.AttrValue&gt; parameter_map = 2;</code>
-     */
-
-    org.tensorflow.proto.framework.AttrValue getParameterMapOrThrow(
-        java.lang.String key);
-  }
-  /**
-   * <pre>
-   * Message to describe custom graph optimizer and its parameters
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.RewriterConfig.CustomGraphOptimizer}
-   */
-  public  static final class CustomGraphOptimizer extends
-      com.google.protobuf.GeneratedMessageV3 implements
-      // @@protoc_insertion_point(message_implements:tensorflow.RewriterConfig.CustomGraphOptimizer)
-      CustomGraphOptimizerOrBuilder {
-  private static final long serialVersionUID = 0L;
-    // Use CustomGraphOptimizer.newBuilder() to construct.
-    private CustomGraphOptimizer(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-      super(builder);
-    }
-    private CustomGraphOptimizer() {
-      name_ = "";
-    }
-
-    @java.lang.Override
-    @SuppressWarnings({"unused"})
-    protected java.lang.Object newInstance(
-        UnusedPrivateParameter unused) {
-      return new CustomGraphOptimizer();
-    }
-
-    @java.lang.Override
-    public final com.google.protobuf.UnknownFieldSet
-    getUnknownFields() {
-      return this.unknownFields;
-    }
-    private CustomGraphOptimizer(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      this();
-      if (extensionRegistry == null) {
-        throw new java.lang.NullPointerException();
-      }
-      int mutable_bitField0_ = 0;
-      com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-          com.google.protobuf.UnknownFieldSet.newBuilder();
-      try {
-        boolean done = false;
-        while (!done) {
-          int tag = input.readTag();
-          switch (tag) {
-            case 0:
-              done = true;
-              break;
-            case 10: {
-              java.lang.String s = input.readStringRequireUtf8();
-
-              name_ = s;
-              break;
-            }
-            case 18: {
-              if (!((mutable_bitField0_ & 0x00000001) != 0)) {
-                parameterMap_ = com.google.protobuf.MapField.newMapField(
-                    ParameterMapDefaultEntryHolder.defaultEntry);
-                mutable_bitField0_ |= 0x00000001;
-              }
-              com.google.protobuf.MapEntry<java.lang.String, org.tensorflow.proto.framework.AttrValue>
-              parameterMap__ = input.readMessage(
-                  ParameterMapDefaultEntryHolder.defaultEntry.getParserForType(), extensionRegistry);
-              parameterMap_.getMutableMap().put(
-                  parameterMap__.getKey(), parameterMap__.getValue());
-              break;
-            }
-            default: {
-              if (!parseUnknownField(
-                  input, unknownFields, extensionRegistry, tag)) {
-                done = true;
-              }
-              break;
-            }
-          }
-        }
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        throw e.setUnfinishedMessage(this);
-      } catch (java.io.IOException e) {
-        throw new com.google.protobuf.InvalidProtocolBufferException(
-            e).setUnfinishedMessage(this);
-      } finally {
-        this.unknownFields = unknownFields.build();
-        makeExtensionsImmutable();
-      }
-    }
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.RewriterConfigProtos.internal_static_tensorflow_RewriterConfig_CustomGraphOptimizer_descriptor;
-    }
-
-    @SuppressWarnings({"rawtypes"})
-    @java.lang.Override
-    protected com.google.protobuf.MapField internalGetMapField(
-        int number) {
-      switch (number) {
-        case 2:
-          return internalGetParameterMap();
-        default:
-          throw new RuntimeException(
-              "Invalid map field number: " + number);
-      }
-    }
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.RewriterConfigProtos.internal_static_tensorflow_RewriterConfig_CustomGraphOptimizer_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizer.class, org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizer.Builder.class);
-    }
-
-    public static final int NAME_FIELD_NUMBER = 1;
-    private volatile java.lang.Object name_;
-    /**
-     * <code>string name = 1;</code>
-     */
-    public java.lang.String getName() {
-      java.lang.Object ref = name_;
-      if (ref instanceof java.lang.String) {
-        return (java.lang.String) ref;
-      } else {
-        com.google.protobuf.ByteString bs = 
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        name_ = s;
-        return s;
-      }
-    }
-    /**
-     * <code>string name = 1;</code>
-     */
-    public com.google.protobuf.ByteString
-        getNameBytes() {
-      java.lang.Object ref = name_;
-      if (ref instanceof java.lang.String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        name_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-
-    public static final int PARAMETER_MAP_FIELD_NUMBER = 2;
-    private static final class ParameterMapDefaultEntryHolder {
-      static final com.google.protobuf.MapEntry<
-          java.lang.String, org.tensorflow.proto.framework.AttrValue> defaultEntry =
-              com.google.protobuf.MapEntry
-              .<java.lang.String, org.tensorflow.proto.framework.AttrValue>newDefaultInstance(
-                  org.tensorflow.proto.framework.RewriterConfigProtos.internal_static_tensorflow_RewriterConfig_CustomGraphOptimizer_ParameterMapEntry_descriptor, 
-                  com.google.protobuf.WireFormat.FieldType.STRING,
-                  "",
-                  com.google.protobuf.WireFormat.FieldType.MESSAGE,
-                  org.tensorflow.proto.framework.AttrValue.getDefaultInstance());
-    }
-    private com.google.protobuf.MapField<
-        java.lang.String, org.tensorflow.proto.framework.AttrValue> parameterMap_;
-    private com.google.protobuf.MapField<java.lang.String, org.tensorflow.proto.framework.AttrValue>
-    internalGetParameterMap() {
-      if (parameterMap_ == null) {
-        return com.google.protobuf.MapField.emptyMapField(
-            ParameterMapDefaultEntryHolder.defaultEntry);
-      }
-      return parameterMap_;
-    }
-
-    public int getParameterMapCount() {
-      return internalGetParameterMap().getMap().size();
-    }
-    /**
-     * <code>map&lt;string, .tensorflow.AttrValue&gt; parameter_map = 2;</code>
-     */
-
-    public boolean containsParameterMap(
-        java.lang.String key) {
-      if (key == null) { throw new java.lang.NullPointerException(); }
-      return internalGetParameterMap().getMap().containsKey(key);
-    }
-    /**
-     * Use {@link #getParameterMapMap()} instead.
-     */
-    @java.lang.Deprecated
-    public java.util.Map<java.lang.String, org.tensorflow.proto.framework.AttrValue> getParameterMap() {
-      return getParameterMapMap();
-    }
-    /**
-     * <code>map&lt;string, .tensorflow.AttrValue&gt; parameter_map = 2;</code>
-     */
-
-    public java.util.Map<java.lang.String, org.tensorflow.proto.framework.AttrValue> getParameterMapMap() {
-      return internalGetParameterMap().getMap();
-    }
-    /**
-     * <code>map&lt;string, .tensorflow.AttrValue&gt; parameter_map = 2;</code>
-     */
-
-    public org.tensorflow.proto.framework.AttrValue getParameterMapOrDefault(
-        java.lang.String key,
-        org.tensorflow.proto.framework.AttrValue defaultValue) {
-      if (key == null) { throw new java.lang.NullPointerException(); }
-      java.util.Map<java.lang.String, org.tensorflow.proto.framework.AttrValue> map =
-          internalGetParameterMap().getMap();
-      return map.containsKey(key) ? map.get(key) : defaultValue;
-    }
-    /**
-     * <code>map&lt;string, .tensorflow.AttrValue&gt; parameter_map = 2;</code>
-     */
-
-    public org.tensorflow.proto.framework.AttrValue getParameterMapOrThrow(
-        java.lang.String key) {
-      if (key == null) { throw new java.lang.NullPointerException(); }
-      java.util.Map<java.lang.String, org.tensorflow.proto.framework.AttrValue> map =
-          internalGetParameterMap().getMap();
-      if (!map.containsKey(key)) {
-        throw new java.lang.IllegalArgumentException();
-      }
-      return map.get(key);
-    }
-
-    private byte memoizedIsInitialized = -1;
-    @java.lang.Override
-    public final boolean isInitialized() {
-      byte isInitialized = memoizedIsInitialized;
-      if (isInitialized == 1) return true;
-      if (isInitialized == 0) return false;
-
-      memoizedIsInitialized = 1;
-      return true;
-    }
-
-    @java.lang.Override
-    public void writeTo(com.google.protobuf.CodedOutputStream output)
-                        throws java.io.IOException {
-      if (!getNameBytes().isEmpty()) {
-        com.google.protobuf.GeneratedMessageV3.writeString(output, 1, name_);
-      }
-      com.google.protobuf.GeneratedMessageV3
-        .serializeStringMapTo(
-          output,
-          internalGetParameterMap(),
-          ParameterMapDefaultEntryHolder.defaultEntry,
-          2);
-      unknownFields.writeTo(output);
-    }
-
-    @java.lang.Override
-    public int getSerializedSize() {
-      int size = memoizedSize;
-      if (size != -1) return size;
-
-      size = 0;
-      if (!getNameBytes().isEmpty()) {
-        size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, name_);
-      }
-      for (java.util.Map.Entry<java.lang.String, org.tensorflow.proto.framework.AttrValue> entry
-           : internalGetParameterMap().getMap().entrySet()) {
-        com.google.protobuf.MapEntry<java.lang.String, org.tensorflow.proto.framework.AttrValue>
-        parameterMap__ = ParameterMapDefaultEntryHolder.defaultEntry.newBuilderForType()
-            .setKey(entry.getKey())
-            .setValue(entry.getValue())
-            .build();
-        size += com.google.protobuf.CodedOutputStream
-            .computeMessageSize(2, parameterMap__);
-      }
-      size += unknownFields.getSerializedSize();
-      memoizedSize = size;
-      return size;
-    }
-
-    @java.lang.Override
-    public boolean equals(final java.lang.Object obj) {
-      if (obj == this) {
-       return true;
-      }
-      if (!(obj instanceof org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizer)) {
-        return super.equals(obj);
-      }
-      org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizer other = (org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizer) obj;
-
-      if (!getName()
-          .equals(other.getName())) return false;
-      if (!internalGetParameterMap().equals(
-          other.internalGetParameterMap())) return false;
-      if (!unknownFields.equals(other.unknownFields)) return false;
-      return true;
-    }
-
-    @java.lang.Override
-    public int hashCode() {
-      if (memoizedHashCode != 0) {
-        return memoizedHashCode;
-      }
-      int hash = 41;
-      hash = (19 * hash) + getDescriptor().hashCode();
-      hash = (37 * hash) + NAME_FIELD_NUMBER;
-      hash = (53 * hash) + getName().hashCode();
-      if (!internalGetParameterMap().getMap().isEmpty()) {
-        hash = (37 * hash) + PARAMETER_MAP_FIELD_NUMBER;
-        hash = (53 * hash) + internalGetParameterMap().hashCode();
-      }
-      hash = (29 * hash) + unknownFields.hashCode();
-      memoizedHashCode = hash;
-      return hash;
-    }
-
-    public static org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizer parseFrom(
-        java.nio.ByteBuffer data)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data);
-    }
-    public static org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizer parseFrom(
-        java.nio.ByteBuffer data,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizer parseFrom(
-        com.google.protobuf.ByteString data)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data);
-    }
-    public static org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizer parseFrom(
-        com.google.protobuf.ByteString data,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizer parseFrom(byte[] data)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data);
-    }
-    public static org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizer parseFrom(
-        byte[] data,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizer parseFrom(java.io.InputStream input)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input);
-    }
-    public static org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizer parseFrom(
-        java.io.InputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizer parseDelimitedFrom(java.io.InputStream input)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseDelimitedWithIOException(PARSER, input);
-    }
-    public static org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizer parseDelimitedFrom(
-        java.io.InputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizer parseFrom(
-        com.google.protobuf.CodedInputStream input)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input);
-    }
-    public static org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizer parseFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input, extensionRegistry);
-    }
-
-    @java.lang.Override
-    public Builder newBuilderForType() { return newBuilder(); }
-    public static Builder newBuilder() {
-      return DEFAULT_INSTANCE.toBuilder();
-    }
-    public static Builder newBuilder(org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizer prototype) {
-      return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-    }
-    @java.lang.Override
-    public Builder toBuilder() {
-      return this == DEFAULT_INSTANCE
-          ? new Builder() : new Builder().mergeFrom(this);
-    }
-
-    @java.lang.Override
-    protected Builder newBuilderForType(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      Builder builder = new Builder(parent);
-      return builder;
-    }
-    /**
-     * <pre>
-     * Message to describe custom graph optimizer and its parameters
-     * </pre>
-     *
-     * Protobuf type {@code tensorflow.RewriterConfig.CustomGraphOptimizer}
-     */
-    public static final class Builder extends
-        com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-        // @@protoc_insertion_point(builder_implements:tensorflow.RewriterConfig.CustomGraphOptimizer)
-        org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizerOrBuilder {
-      public static final com.google.protobuf.Descriptors.Descriptor
-          getDescriptor() {
-        return org.tensorflow.proto.framework.RewriterConfigProtos.internal_static_tensorflow_RewriterConfig_CustomGraphOptimizer_descriptor;
-      }
-
-      @SuppressWarnings({"rawtypes"})
-      protected com.google.protobuf.MapField internalGetMapField(
-          int number) {
-        switch (number) {
-          case 2:
-            return internalGetParameterMap();
-          default:
-            throw new RuntimeException(
-                "Invalid map field number: " + number);
-        }
-      }
-      @SuppressWarnings({"rawtypes"})
-      protected com.google.protobuf.MapField internalGetMutableMapField(
-          int number) {
-        switch (number) {
-          case 2:
-            return internalGetMutableParameterMap();
-          default:
-            throw new RuntimeException(
-                "Invalid map field number: " + number);
-        }
-      }
-      @java.lang.Override
-      protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-          internalGetFieldAccessorTable() {
-        return org.tensorflow.proto.framework.RewriterConfigProtos.internal_static_tensorflow_RewriterConfig_CustomGraphOptimizer_fieldAccessorTable
-            .ensureFieldAccessorsInitialized(
-                org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizer.class, org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizer.Builder.class);
-      }
-
-      // Construct using org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizer.newBuilder()
-      private Builder() {
-        maybeForceBuilderInitialization();
-      }
-
-      private Builder(
-          com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-        super(parent);
-        maybeForceBuilderInitialization();
-      }
-      private void maybeForceBuilderInitialization() {
-        if (com.google.protobuf.GeneratedMessageV3
-                .alwaysUseFieldBuilders) {
-        }
-      }
-      @java.lang.Override
-      public Builder clear() {
-        super.clear();
-        name_ = "";
-
-        internalGetMutableParameterMap().clear();
-        return this;
-      }
-
-      @java.lang.Override
-      public com.google.protobuf.Descriptors.Descriptor
-          getDescriptorForType() {
-        return org.tensorflow.proto.framework.RewriterConfigProtos.internal_static_tensorflow_RewriterConfig_CustomGraphOptimizer_descriptor;
-      }
-
-      @java.lang.Override
-      public org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizer getDefaultInstanceForType() {
-        return org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizer.getDefaultInstance();
-      }
-
-      @java.lang.Override
-      public org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizer build() {
-        org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizer result = buildPartial();
-        if (!result.isInitialized()) {
-          throw newUninitializedMessageException(result);
-        }
-        return result;
-      }
-
-      @java.lang.Override
-      public org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizer buildPartial() {
-        org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizer result = new org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizer(this);
-        int from_bitField0_ = bitField0_;
-        result.name_ = name_;
-        result.parameterMap_ = internalGetParameterMap();
-        result.parameterMap_.makeImmutable();
-        onBuilt();
-        return result;
-      }
-
-      @java.lang.Override
-      public Builder clone() {
-        return super.clone();
-      }
-      @java.lang.Override
-      public Builder setField(
-          com.google.protobuf.Descriptors.FieldDescriptor field,
-          java.lang.Object value) {
-        return super.setField(field, value);
-      }
-      @java.lang.Override
-      public Builder clearField(
-          com.google.protobuf.Descriptors.FieldDescriptor field) {
-        return super.clearField(field);
-      }
-      @java.lang.Override
-      public Builder clearOneof(
-          com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-        return super.clearOneof(oneof);
-      }
-      @java.lang.Override
-      public Builder setRepeatedField(
-          com.google.protobuf.Descriptors.FieldDescriptor field,
-          int index, java.lang.Object value) {
-        return super.setRepeatedField(field, index, value);
-      }
-      @java.lang.Override
-      public Builder addRepeatedField(
-          com.google.protobuf.Descriptors.FieldDescriptor field,
-          java.lang.Object value) {
-        return super.addRepeatedField(field, value);
-      }
-      @java.lang.Override
-      public Builder mergeFrom(com.google.protobuf.Message other) {
-        if (other instanceof org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizer) {
-          return mergeFrom((org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizer)other);
-        } else {
-          super.mergeFrom(other);
-          return this;
-        }
-      }
-
-      public Builder mergeFrom(org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizer other) {
-        if (other == org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizer.getDefaultInstance()) return this;
-        if (!other.getName().isEmpty()) {
-          name_ = other.name_;
-          onChanged();
-        }
-        internalGetMutableParameterMap().mergeFrom(
-            other.internalGetParameterMap());
-        this.mergeUnknownFields(other.unknownFields);
-        onChanged();
-        return this;
-      }
-
-      @java.lang.Override
-      public final boolean isInitialized() {
-        return true;
-      }
-
-      @java.lang.Override
-      public Builder mergeFrom(
-          com.google.protobuf.CodedInputStream input,
-          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-          throws java.io.IOException {
-        org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizer parsedMessage = null;
-        try {
-          parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-        } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-          parsedMessage = (org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizer) e.getUnfinishedMessage();
-          throw e.unwrapIOException();
-        } finally {
-          if (parsedMessage != null) {
-            mergeFrom(parsedMessage);
-          }
-        }
-        return this;
-      }
-      private int bitField0_;
-
-      private java.lang.Object name_ = "";
-      /**
-       * <code>string name = 1;</code>
-       */
-      public java.lang.String getName() {
-        java.lang.Object ref = name_;
-        if (!(ref instanceof java.lang.String)) {
-          com.google.protobuf.ByteString bs =
-              (com.google.protobuf.ByteString) ref;
-          java.lang.String s = bs.toStringUtf8();
-          name_ = s;
-          return s;
-        } else {
-          return (java.lang.String) ref;
-        }
-      }
-      /**
-       * <code>string name = 1;</code>
-       */
-      public com.google.protobuf.ByteString
-          getNameBytes() {
-        java.lang.Object ref = name_;
-        if (ref instanceof String) {
-          com.google.protobuf.ByteString b = 
-              com.google.protobuf.ByteString.copyFromUtf8(
-                  (java.lang.String) ref);
-          name_ = b;
-          return b;
-        } else {
-          return (com.google.protobuf.ByteString) ref;
-        }
-      }
-      /**
-       * <code>string name = 1;</code>
-       */
-      public Builder setName(
-          java.lang.String value) {
-        if (value == null) {
-    throw new NullPointerException();
-  }
-  
-        name_ = value;
-        onChanged();
-        return this;
-      }
-      /**
-       * <code>string name = 1;</code>
-       */
-      public Builder clearName() {
-        
-        name_ = getDefaultInstance().getName();
-        onChanged();
-        return this;
-      }
-      /**
-       * <code>string name = 1;</code>
-       */
-      public Builder setNameBytes(
-          com.google.protobuf.ByteString value) {
-        if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-        
-        name_ = value;
-        onChanged();
-        return this;
-      }
-
-      private com.google.protobuf.MapField<
-          java.lang.String, org.tensorflow.proto.framework.AttrValue> parameterMap_;
-      private com.google.protobuf.MapField<java.lang.String, org.tensorflow.proto.framework.AttrValue>
-      internalGetParameterMap() {
-        if (parameterMap_ == null) {
-          return com.google.protobuf.MapField.emptyMapField(
-              ParameterMapDefaultEntryHolder.defaultEntry);
-        }
-        return parameterMap_;
-      }
-      private com.google.protobuf.MapField<java.lang.String, org.tensorflow.proto.framework.AttrValue>
-      internalGetMutableParameterMap() {
-        onChanged();;
-        if (parameterMap_ == null) {
-          parameterMap_ = com.google.protobuf.MapField.newMapField(
-              ParameterMapDefaultEntryHolder.defaultEntry);
-        }
-        if (!parameterMap_.isMutable()) {
-          parameterMap_ = parameterMap_.copy();
-        }
-        return parameterMap_;
-      }
-
-      public int getParameterMapCount() {
-        return internalGetParameterMap().getMap().size();
-      }
-      /**
-       * <code>map&lt;string, .tensorflow.AttrValue&gt; parameter_map = 2;</code>
-       */
-
-      public boolean containsParameterMap(
-          java.lang.String key) {
-        if (key == null) { throw new java.lang.NullPointerException(); }
-        return internalGetParameterMap().getMap().containsKey(key);
-      }
-      /**
-       * Use {@link #getParameterMapMap()} instead.
-       */
-      @java.lang.Deprecated
-      public java.util.Map<java.lang.String, org.tensorflow.proto.framework.AttrValue> getParameterMap() {
-        return getParameterMapMap();
-      }
-      /**
-       * <code>map&lt;string, .tensorflow.AttrValue&gt; parameter_map = 2;</code>
-       */
-
-      public java.util.Map<java.lang.String, org.tensorflow.proto.framework.AttrValue> getParameterMapMap() {
-        return internalGetParameterMap().getMap();
-      }
-      /**
-       * <code>map&lt;string, .tensorflow.AttrValue&gt; parameter_map = 2;</code>
-       */
-
-      public org.tensorflow.proto.framework.AttrValue getParameterMapOrDefault(
-          java.lang.String key,
-          org.tensorflow.proto.framework.AttrValue defaultValue) {
-        if (key == null) { throw new java.lang.NullPointerException(); }
-        java.util.Map<java.lang.String, org.tensorflow.proto.framework.AttrValue> map =
-            internalGetParameterMap().getMap();
-        return map.containsKey(key) ? map.get(key) : defaultValue;
-      }
-      /**
-       * <code>map&lt;string, .tensorflow.AttrValue&gt; parameter_map = 2;</code>
-       */
-
-      public org.tensorflow.proto.framework.AttrValue getParameterMapOrThrow(
-          java.lang.String key) {
-        if (key == null) { throw new java.lang.NullPointerException(); }
-        java.util.Map<java.lang.String, org.tensorflow.proto.framework.AttrValue> map =
-            internalGetParameterMap().getMap();
-        if (!map.containsKey(key)) {
-          throw new java.lang.IllegalArgumentException();
-        }
-        return map.get(key);
-      }
-
-      public Builder clearParameterMap() {
-        internalGetMutableParameterMap().getMutableMap()
-            .clear();
-        return this;
-      }
-      /**
-       * <code>map&lt;string, .tensorflow.AttrValue&gt; parameter_map = 2;</code>
-       */
-
-      public Builder removeParameterMap(
-          java.lang.String key) {
-        if (key == null) { throw new java.lang.NullPointerException(); }
-        internalGetMutableParameterMap().getMutableMap()
-            .remove(key);
-        return this;
-      }
-      /**
-       * Use alternate mutation accessors instead.
-       */
-      @java.lang.Deprecated
-      public java.util.Map<java.lang.String, org.tensorflow.proto.framework.AttrValue>
-      getMutableParameterMap() {
-        return internalGetMutableParameterMap().getMutableMap();
-      }
-      /**
-       * <code>map&lt;string, .tensorflow.AttrValue&gt; parameter_map = 2;</code>
-       */
-      public Builder putParameterMap(
-          java.lang.String key,
-          org.tensorflow.proto.framework.AttrValue value) {
-        if (key == null) { throw new java.lang.NullPointerException(); }
-        if (value == null) { throw new java.lang.NullPointerException(); }
-        internalGetMutableParameterMap().getMutableMap()
-            .put(key, value);
-        return this;
-      }
-      /**
-       * <code>map&lt;string, .tensorflow.AttrValue&gt; parameter_map = 2;</code>
-       */
-
-      public Builder putAllParameterMap(
-          java.util.Map<java.lang.String, org.tensorflow.proto.framework.AttrValue> values) {
-        internalGetMutableParameterMap().getMutableMap()
-            .putAll(values);
-        return this;
-      }
-      @java.lang.Override
-      public final Builder setUnknownFields(
-          final com.google.protobuf.UnknownFieldSet unknownFields) {
-        return super.setUnknownFields(unknownFields);
-      }
-
-      @java.lang.Override
-      public final Builder mergeUnknownFields(
-          final com.google.protobuf.UnknownFieldSet unknownFields) {
-        return super.mergeUnknownFields(unknownFields);
-      }
-
-
-      // @@protoc_insertion_point(builder_scope:tensorflow.RewriterConfig.CustomGraphOptimizer)
-    }
-
-    // @@protoc_insertion_point(class_scope:tensorflow.RewriterConfig.CustomGraphOptimizer)
-    private static final org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizer DEFAULT_INSTANCE;
-    static {
-      DEFAULT_INSTANCE = new org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizer();
-    }
-
-    public static org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizer getDefaultInstance() {
-      return DEFAULT_INSTANCE;
-    }
-
-    private static final com.google.protobuf.Parser<CustomGraphOptimizer>
-        PARSER = new com.google.protobuf.AbstractParser<CustomGraphOptimizer>() {
-      @java.lang.Override
-      public CustomGraphOptimizer parsePartialFrom(
-          com.google.protobuf.CodedInputStream input,
-          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-          throws com.google.protobuf.InvalidProtocolBufferException {
-        return new CustomGraphOptimizer(input, extensionRegistry);
-      }
-    };
-
-    public static com.google.protobuf.Parser<CustomGraphOptimizer> parser() {
-      return PARSER;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Parser<CustomGraphOptimizer> getParserForType() {
-      return PARSER;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizer getDefaultInstanceForType() {
-      return DEFAULT_INSTANCE;
-    }
-
-  }
-
-  public static final int LAYOUT_OPTIMIZER_FIELD_NUMBER = 1;
-  private int layoutOptimizer_;
-  /**
-   * <pre>
-   * Optimize tensor layouts (default is ON)
-   * e.g. This will try to use NCHW layout on GPU which is faster.
-   * </pre>
-   *
-   * <code>.tensorflow.RewriterConfig.Toggle layout_optimizer = 1;</code>
-   */
-  public int getLayoutOptimizerValue() {
-    return layoutOptimizer_;
-  }
-  /**
-   * <pre>
-   * Optimize tensor layouts (default is ON)
-   * e.g. This will try to use NCHW layout on GPU which is faster.
-   * </pre>
-   *
-   * <code>.tensorflow.RewriterConfig.Toggle layout_optimizer = 1;</code>
-   */
-  public org.tensorflow.proto.framework.RewriterConfig.Toggle getLayoutOptimizer() {
-    @SuppressWarnings("deprecation")
-    org.tensorflow.proto.framework.RewriterConfig.Toggle result = org.tensorflow.proto.framework.RewriterConfig.Toggle.valueOf(layoutOptimizer_);
-    return result == null ? org.tensorflow.proto.framework.RewriterConfig.Toggle.UNRECOGNIZED : result;
-  }
-
-  public static final int CONSTANT_FOLDING_FIELD_NUMBER = 3;
-  private int constantFolding_;
-  /**
-   * <pre>
-   * Fold constants (default is ON)
-   * Statically infer the value of tensors when possible, and materialize the
-   * result using constants.
-   * </pre>
-   *
-   * <code>.tensorflow.RewriterConfig.Toggle constant_folding = 3;</code>
-   */
-  public int getConstantFoldingValue() {
-    return constantFolding_;
-  }
-  /**
-   * <pre>
-   * Fold constants (default is ON)
-   * Statically infer the value of tensors when possible, and materialize the
-   * result using constants.
-   * </pre>
-   *
-   * <code>.tensorflow.RewriterConfig.Toggle constant_folding = 3;</code>
-   */
-  public org.tensorflow.proto.framework.RewriterConfig.Toggle getConstantFolding() {
-    @SuppressWarnings("deprecation")
-    org.tensorflow.proto.framework.RewriterConfig.Toggle result = org.tensorflow.proto.framework.RewriterConfig.Toggle.valueOf(constantFolding_);
-    return result == null ? org.tensorflow.proto.framework.RewriterConfig.Toggle.UNRECOGNIZED : result;
-  }
-
-  public static final int SHAPE_OPTIMIZATION_FIELD_NUMBER = 13;
-  private int shapeOptimization_;
-  /**
-   * <pre>
-   * Shape optimizations (default is ON)
-   * Simplify computations made on shapes.
-   * </pre>
-   *
-   * <code>.tensorflow.RewriterConfig.Toggle shape_optimization = 13;</code>
-   */
-  public int getShapeOptimizationValue() {
-    return shapeOptimization_;
-  }
-  /**
-   * <pre>
-   * Shape optimizations (default is ON)
-   * Simplify computations made on shapes.
-   * </pre>
-   *
-   * <code>.tensorflow.RewriterConfig.Toggle shape_optimization = 13;</code>
-   */
-  public org.tensorflow.proto.framework.RewriterConfig.Toggle getShapeOptimization() {
-    @SuppressWarnings("deprecation")
-    org.tensorflow.proto.framework.RewriterConfig.Toggle result = org.tensorflow.proto.framework.RewriterConfig.Toggle.valueOf(shapeOptimization_);
-    return result == null ? org.tensorflow.proto.framework.RewriterConfig.Toggle.UNRECOGNIZED : result;
-  }
-
-  public static final int REMAPPING_FIELD_NUMBER = 14;
-  private int remapping_;
-  /**
-   * <pre>
-   * Remapping (default is ON)
-   * Remap subgraphs onto more efficient implementations.
-   * </pre>
-   *
-   * <code>.tensorflow.RewriterConfig.Toggle remapping = 14;</code>
-   */
-  public int getRemappingValue() {
-    return remapping_;
-  }
-  /**
-   * <pre>
-   * Remapping (default is ON)
-   * Remap subgraphs onto more efficient implementations.
-   * </pre>
-   *
-   * <code>.tensorflow.RewriterConfig.Toggle remapping = 14;</code>
-   */
-  public org.tensorflow.proto.framework.RewriterConfig.Toggle getRemapping() {
-    @SuppressWarnings("deprecation")
-    org.tensorflow.proto.framework.RewriterConfig.Toggle result = org.tensorflow.proto.framework.RewriterConfig.Toggle.valueOf(remapping_);
-    return result == null ? org.tensorflow.proto.framework.RewriterConfig.Toggle.UNRECOGNIZED : result;
-  }
-
-  public static final int COMMON_SUBGRAPH_ELIMINATION_FIELD_NUMBER = 24;
-  private int commonSubgraphElimination_;
-  /**
-   * <pre>
-   * Common subgraph elimination (default is ON)
-   * e.g. Simplify arithmetic ops; merge ops with same value (like constants).
-   * </pre>
-   *
-   * <code>.tensorflow.RewriterConfig.Toggle common_subgraph_elimination = 24;</code>
-   */
-  public int getCommonSubgraphEliminationValue() {
-    return commonSubgraphElimination_;
-  }
-  /**
-   * <pre>
-   * Common subgraph elimination (default is ON)
-   * e.g. Simplify arithmetic ops; merge ops with same value (like constants).
-   * </pre>
-   *
-   * <code>.tensorflow.RewriterConfig.Toggle common_subgraph_elimination = 24;</code>
-   */
-  public org.tensorflow.proto.framework.RewriterConfig.Toggle getCommonSubgraphElimination() {
-    @SuppressWarnings("deprecation")
-    org.tensorflow.proto.framework.RewriterConfig.Toggle result = org.tensorflow.proto.framework.RewriterConfig.Toggle.valueOf(commonSubgraphElimination_);
-    return result == null ? org.tensorflow.proto.framework.RewriterConfig.Toggle.UNRECOGNIZED : result;
-  }
-
-  public static final int ARITHMETIC_OPTIMIZATION_FIELD_NUMBER = 7;
-  private int arithmeticOptimization_;
-  /**
-   * <pre>
-   * Arithmetic optimizations (default is ON)
-   * e.g. Simplify arithmetic ops; merge ops with same value (like constants).
-   * </pre>
-   *
-   * <code>.tensorflow.RewriterConfig.Toggle arithmetic_optimization = 7;</code>
-   */
-  public int getArithmeticOptimizationValue() {
-    return arithmeticOptimization_;
-  }
-  /**
-   * <pre>
-   * Arithmetic optimizations (default is ON)
-   * e.g. Simplify arithmetic ops; merge ops with same value (like constants).
-   * </pre>
-   *
-   * <code>.tensorflow.RewriterConfig.Toggle arithmetic_optimization = 7;</code>
-   */
-  public org.tensorflow.proto.framework.RewriterConfig.Toggle getArithmeticOptimization() {
-    @SuppressWarnings("deprecation")
-    org.tensorflow.proto.framework.RewriterConfig.Toggle result = org.tensorflow.proto.framework.RewriterConfig.Toggle.valueOf(arithmeticOptimization_);
-    return result == null ? org.tensorflow.proto.framework.RewriterConfig.Toggle.UNRECOGNIZED : result;
-  }
-
-  public static final int DEPENDENCY_OPTIMIZATION_FIELD_NUMBER = 8;
-  private int dependencyOptimization_;
-  /**
-   * <pre>
-   * Control dependency optimizations (default is ON).
-   * Remove redundant control dependencies, which may enable other optimization.
-   * </pre>
-   *
-   * <code>.tensorflow.RewriterConfig.Toggle dependency_optimization = 8;</code>
-   */
-  public int getDependencyOptimizationValue() {
-    return dependencyOptimization_;
-  }
-  /**
-   * <pre>
-   * Control dependency optimizations (default is ON).
-   * Remove redundant control dependencies, which may enable other optimization.
-   * </pre>
-   *
-   * <code>.tensorflow.RewriterConfig.Toggle dependency_optimization = 8;</code>
-   */
-  public org.tensorflow.proto.framework.RewriterConfig.Toggle getDependencyOptimization() {
-    @SuppressWarnings("deprecation")
-    org.tensorflow.proto.framework.RewriterConfig.Toggle result = org.tensorflow.proto.framework.RewriterConfig.Toggle.valueOf(dependencyOptimization_);
-    return result == null ? org.tensorflow.proto.framework.RewriterConfig.Toggle.UNRECOGNIZED : result;
-  }
-
-  public static final int LOOP_OPTIMIZATION_FIELD_NUMBER = 9;
-  private int loopOptimization_;
-  /**
-   * <pre>
-   * Loop optimizations (default is ON).
-   * </pre>
-   *
-   * <code>.tensorflow.RewriterConfig.Toggle loop_optimization = 9;</code>
-   */
-  public int getLoopOptimizationValue() {
-    return loopOptimization_;
-  }
-  /**
-   * <pre>
-   * Loop optimizations (default is ON).
-   * </pre>
-   *
-   * <code>.tensorflow.RewriterConfig.Toggle loop_optimization = 9;</code>
-   */
-  public org.tensorflow.proto.framework.RewriterConfig.Toggle getLoopOptimization() {
-    @SuppressWarnings("deprecation")
-    org.tensorflow.proto.framework.RewriterConfig.Toggle result = org.tensorflow.proto.framework.RewriterConfig.Toggle.valueOf(loopOptimization_);
-    return result == null ? org.tensorflow.proto.framework.RewriterConfig.Toggle.UNRECOGNIZED : result;
-  }
-
-  public static final int FUNCTION_OPTIMIZATION_FIELD_NUMBER = 10;
-  private int functionOptimization_;
-  /**
-   * <pre>
-   * Function optimizations (default is ON).
-   * </pre>
-   *
-   * <code>.tensorflow.RewriterConfig.Toggle function_optimization = 10;</code>
-   */
-  public int getFunctionOptimizationValue() {
-    return functionOptimization_;
-  }
-  /**
-   * <pre>
-   * Function optimizations (default is ON).
-   * </pre>
-   *
-   * <code>.tensorflow.RewriterConfig.Toggle function_optimization = 10;</code>
-   */
-  public org.tensorflow.proto.framework.RewriterConfig.Toggle getFunctionOptimization() {
-    @SuppressWarnings("deprecation")
-    org.tensorflow.proto.framework.RewriterConfig.Toggle result = org.tensorflow.proto.framework.RewriterConfig.Toggle.valueOf(functionOptimization_);
-    return result == null ? org.tensorflow.proto.framework.RewriterConfig.Toggle.UNRECOGNIZED : result;
-  }
-
-  public static final int DEBUG_STRIPPER_FIELD_NUMBER = 11;
-  private int debugStripper_;
-  /**
-   * <pre>
-   * Strips debug-related nodes from the graph (off by default).
-   * </pre>
-   *
-   * <code>.tensorflow.RewriterConfig.Toggle debug_stripper = 11;</code>
-   */
-  public int getDebugStripperValue() {
-    return debugStripper_;
-  }
-  /**
-   * <pre>
-   * Strips debug-related nodes from the graph (off by default).
-   * </pre>
-   *
-   * <code>.tensorflow.RewriterConfig.Toggle debug_stripper = 11;</code>
-   */
-  public org.tensorflow.proto.framework.RewriterConfig.Toggle getDebugStripper() {
-    @SuppressWarnings("deprecation")
-    org.tensorflow.proto.framework.RewriterConfig.Toggle result = org.tensorflow.proto.framework.RewriterConfig.Toggle.valueOf(debugStripper_);
-    return result == null ? org.tensorflow.proto.framework.RewriterConfig.Toggle.UNRECOGNIZED : result;
-  }
-
-  public static final int DISABLE_MODEL_PRUNING_FIELD_NUMBER = 2;
-  private boolean disableModelPruning_;
-  /**
-   * <pre>
-   * If true, don't remove unnecessary ops from the graph
-   * </pre>
-   *
-   * <code>bool disable_model_pruning = 2;</code>
-   */
-  public boolean getDisableModelPruning() {
-    return disableModelPruning_;
-  }
-
-  public static final int SCOPED_ALLOCATOR_OPTIMIZATION_FIELD_NUMBER = 15;
-  private int scopedAllocatorOptimization_;
-  /**
-   * <pre>
-   * Try to allocate some independent Op outputs contiguously in order to
-   * merge or eliminate downstream Ops (off by default).
-   * </pre>
-   *
-   * <code>.tensorflow.RewriterConfig.Toggle scoped_allocator_optimization = 15;</code>
-   */
-  public int getScopedAllocatorOptimizationValue() {
-    return scopedAllocatorOptimization_;
-  }
-  /**
-   * <pre>
-   * Try to allocate some independent Op outputs contiguously in order to
-   * merge or eliminate downstream Ops (off by default).
-   * </pre>
-   *
-   * <code>.tensorflow.RewriterConfig.Toggle scoped_allocator_optimization = 15;</code>
-   */
-  public org.tensorflow.proto.framework.RewriterConfig.Toggle getScopedAllocatorOptimization() {
-    @SuppressWarnings("deprecation")
-    org.tensorflow.proto.framework.RewriterConfig.Toggle result = org.tensorflow.proto.framework.RewriterConfig.Toggle.valueOf(scopedAllocatorOptimization_);
-    return result == null ? org.tensorflow.proto.framework.RewriterConfig.Toggle.UNRECOGNIZED : result;
-  }
-
-  public static final int PIN_TO_HOST_OPTIMIZATION_FIELD_NUMBER = 18;
-  private int pinToHostOptimization_;
-  /**
-   * <pre>
-   * Force small ops onto the CPU (default is OFF).
-   * </pre>
-   *
-   * <code>.tensorflow.RewriterConfig.Toggle pin_to_host_optimization = 18;</code>
-   */
-  public int getPinToHostOptimizationValue() {
-    return pinToHostOptimization_;
-  }
-  /**
-   * <pre>
-   * Force small ops onto the CPU (default is OFF).
-   * </pre>
-   *
-   * <code>.tensorflow.RewriterConfig.Toggle pin_to_host_optimization = 18;</code>
-   */
-  public org.tensorflow.proto.framework.RewriterConfig.Toggle getPinToHostOptimization() {
-    @SuppressWarnings("deprecation")
-    org.tensorflow.proto.framework.RewriterConfig.Toggle result = org.tensorflow.proto.framework.RewriterConfig.Toggle.valueOf(pinToHostOptimization_);
-    return result == null ? org.tensorflow.proto.framework.RewriterConfig.Toggle.UNRECOGNIZED : result;
-  }
-
-  public static final int IMPLEMENTATION_SELECTOR_FIELD_NUMBER = 22;
-  private int implementationSelector_;
-  /**
-   * <pre>
-   * Enable the swap of kernel implementations based on the device placement
-   * (default is ON).
-   * </pre>
-   *
-   * <code>.tensorflow.RewriterConfig.Toggle implementation_selector = 22;</code>
-   */
-  public int getImplementationSelectorValue() {
-    return implementationSelector_;
-  }
-  /**
-   * <pre>
-   * Enable the swap of kernel implementations based on the device placement
-   * (default is ON).
-   * </pre>
-   *
-   * <code>.tensorflow.RewriterConfig.Toggle implementation_selector = 22;</code>
-   */
-  public org.tensorflow.proto.framework.RewriterConfig.Toggle getImplementationSelector() {
-    @SuppressWarnings("deprecation")
-    org.tensorflow.proto.framework.RewriterConfig.Toggle result = org.tensorflow.proto.framework.RewriterConfig.Toggle.valueOf(implementationSelector_);
-    return result == null ? org.tensorflow.proto.framework.RewriterConfig.Toggle.UNRECOGNIZED : result;
-  }
-
-  public static final int AUTO_MIXED_PRECISION_FIELD_NUMBER = 23;
-  private int autoMixedPrecision_;
-  /**
-   * <pre>
-   * Optimize data types for CUDA (default is OFF).
-   * This will try to use float16 on GPU which is faster.
-   * Note that this can change the numerical stability of the graph and may
-   * require the use of loss scaling to maintain model convergence.
-   * </pre>
-   *
-   * <code>.tensorflow.RewriterConfig.Toggle auto_mixed_precision = 23;</code>
-   */
-  public int getAutoMixedPrecisionValue() {
-    return autoMixedPrecision_;
-  }
-  /**
-   * <pre>
-   * Optimize data types for CUDA (default is OFF).
-   * This will try to use float16 on GPU which is faster.
-   * Note that this can change the numerical stability of the graph and may
-   * require the use of loss scaling to maintain model convergence.
-   * </pre>
-   *
-   * <code>.tensorflow.RewriterConfig.Toggle auto_mixed_precision = 23;</code>
-   */
-  public org.tensorflow.proto.framework.RewriterConfig.Toggle getAutoMixedPrecision() {
-    @SuppressWarnings("deprecation")
-    org.tensorflow.proto.framework.RewriterConfig.Toggle result = org.tensorflow.proto.framework.RewriterConfig.Toggle.valueOf(autoMixedPrecision_);
-    return result == null ? org.tensorflow.proto.framework.RewriterConfig.Toggle.UNRECOGNIZED : result;
-  }
-
-  public static final int AUTO_MIXED_PRECISION_MKL_FIELD_NUMBER = 25;
-  private int autoMixedPrecisionMkl_;
-  /**
-   * <pre>
-   * Optimize data types for MKL (default is OFF).
-   * This will try to use bfloat16 on CPUs, which is faster.
-   * Note that this can change the numerical stability of the graph.
-   * </pre>
-   *
-   * <code>.tensorflow.RewriterConfig.Toggle auto_mixed_precision_mkl = 25;</code>
-   */
-  public int getAutoMixedPrecisionMklValue() {
-    return autoMixedPrecisionMkl_;
-  }
-  /**
-   * <pre>
-   * Optimize data types for MKL (default is OFF).
-   * This will try to use bfloat16 on CPUs, which is faster.
-   * Note that this can change the numerical stability of the graph.
-   * </pre>
-   *
-   * <code>.tensorflow.RewriterConfig.Toggle auto_mixed_precision_mkl = 25;</code>
-   */
-  public org.tensorflow.proto.framework.RewriterConfig.Toggle getAutoMixedPrecisionMkl() {
-    @SuppressWarnings("deprecation")
-    org.tensorflow.proto.framework.RewriterConfig.Toggle result = org.tensorflow.proto.framework.RewriterConfig.Toggle.valueOf(autoMixedPrecisionMkl_);
-    return result == null ? org.tensorflow.proto.framework.RewriterConfig.Toggle.UNRECOGNIZED : result;
-  }
-
-  public static final int DISABLE_META_OPTIMIZER_FIELD_NUMBER = 19;
-  private boolean disableMetaOptimizer_;
-  /**
-   * <pre>
-   * Disable the entire meta optimizer (off by default).
-   * </pre>
-   *
-   * <code>bool disable_meta_optimizer = 19;</code>
-   */
-  public boolean getDisableMetaOptimizer() {
-    return disableMetaOptimizer_;
-  }
-
-  public static final int META_OPTIMIZER_ITERATIONS_FIELD_NUMBER = 12;
-  private int metaOptimizerIterations_;
-  /**
-   * <pre>
-   * Controls how many times we run the optimizers in meta optimizer (default
-   * is once).
-   * </pre>
-   *
-   * <code>.tensorflow.RewriterConfig.NumIterationsType meta_optimizer_iterations = 12;</code>
-   */
-  public int getMetaOptimizerIterationsValue() {
-    return metaOptimizerIterations_;
-  }
-  /**
-   * <pre>
-   * Controls how many times we run the optimizers in meta optimizer (default
-   * is once).
-   * </pre>
-   *
-   * <code>.tensorflow.RewriterConfig.NumIterationsType meta_optimizer_iterations = 12;</code>
-   */
-  public org.tensorflow.proto.framework.RewriterConfig.NumIterationsType getMetaOptimizerIterations() {
-    @SuppressWarnings("deprecation")
-    org.tensorflow.proto.framework.RewriterConfig.NumIterationsType result = org.tensorflow.proto.framework.RewriterConfig.NumIterationsType.valueOf(metaOptimizerIterations_);
-    return result == null ? org.tensorflow.proto.framework.RewriterConfig.NumIterationsType.UNRECOGNIZED : result;
-  }
-
-  public static final int MIN_GRAPH_NODES_FIELD_NUMBER = 17;
-  private int minGraphNodes_;
-  /**
-   * <pre>
-   * The minimum number of nodes in a graph to optimizer. For smaller graphs,
-   * optimization is skipped.
-   * 0 means the system picks an appropriate number.
-   * &lt; 0 means do not skip optimization.
-   * </pre>
-   *
-   * <code>int32 min_graph_nodes = 17;</code>
-   */
-  public int getMinGraphNodes() {
-    return minGraphNodes_;
-  }
-
-  public static final int MEMORY_OPTIMIZATION_FIELD_NUMBER = 4;
-  private int memoryOptimization_;
-  /**
-   * <pre>
-   * Configures memory optimization passes through the meta-optimizer. Has no
-   * effect on manually requested memory optimization passes in the optimizers
-   * field.
-   * </pre>
-   *
-   * <code>.tensorflow.RewriterConfig.MemOptType memory_optimization = 4;</code>
-   */
-  public int getMemoryOptimizationValue() {
-    return memoryOptimization_;
-  }
-  /**
-   * <pre>
-   * Configures memory optimization passes through the meta-optimizer. Has no
-   * effect on manually requested memory optimization passes in the optimizers
-   * field.
-   * </pre>
-   *
-   * <code>.tensorflow.RewriterConfig.MemOptType memory_optimization = 4;</code>
-   */
-  public org.tensorflow.proto.framework.RewriterConfig.MemOptType getMemoryOptimization() {
-    @SuppressWarnings("deprecation")
-    org.tensorflow.proto.framework.RewriterConfig.MemOptType result = org.tensorflow.proto.framework.RewriterConfig.MemOptType.valueOf(memoryOptimization_);
-    return result == null ? org.tensorflow.proto.framework.RewriterConfig.MemOptType.UNRECOGNIZED : result;
-  }
-
-  public static final int MEMORY_OPTIMIZER_TARGET_NODE_NAME_SCOPE_FIELD_NUMBER = 6;
-  private volatile java.lang.Object memoryOptimizerTargetNodeNameScope_;
-  /**
-   * <pre>
-   * A node name scope for node names which are valid outputs of recomputations.
-   * Inputs to nodes that match this scope may be recomputed (subject either to
-   * manual annotation of those input nodes or to manual annotation and
-   * heuristics depending on memory_optimization), but the nodes themselves will
-   * not be recomputed. This matches any sub-scopes as well, meaning the scope
-   * can appear not just as a top-level scope. For example, if the value is
-   * "gradients/", the default, it will match node name "gradients/foo",
-   * "foo/gradients/bar", but not "foo_gradients/"
-   * </pre>
-   *
-   * <code>string memory_optimizer_target_node_name_scope = 6;</code>
-   */
-  public java.lang.String getMemoryOptimizerTargetNodeNameScope() {
-    java.lang.Object ref = memoryOptimizerTargetNodeNameScope_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      memoryOptimizerTargetNodeNameScope_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * A node name scope for node names which are valid outputs of recomputations.
-   * Inputs to nodes that match this scope may be recomputed (subject either to
-   * manual annotation of those input nodes or to manual annotation and
-   * heuristics depending on memory_optimization), but the nodes themselves will
-   * not be recomputed. This matches any sub-scopes as well, meaning the scope
-   * can appear not just as a top-level scope. For example, if the value is
-   * "gradients/", the default, it will match node name "gradients/foo",
-   * "foo/gradients/bar", but not "foo_gradients/"
-   * </pre>
-   *
-   * <code>string memory_optimizer_target_node_name_scope = 6;</code>
-   */
-  public com.google.protobuf.ByteString
-      getMemoryOptimizerTargetNodeNameScopeBytes() {
-    java.lang.Object ref = memoryOptimizerTargetNodeNameScope_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      memoryOptimizerTargetNodeNameScope_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int META_OPTIMIZER_TIMEOUT_MS_FIELD_NUMBER = 20;
-  private long metaOptimizerTimeoutMs_;
-  /**
-   * <pre>
-   * Maximum number of milliseconds to spend optimizing a single graph before
-   * timing out. If equal to 0 the system picks a default (currently 5 minutes).
-   * If less than 0 the optimizer will never time out.
-   * </pre>
-   *
-   * <code>int64 meta_optimizer_timeout_ms = 20;</code>
-   */
-  public long getMetaOptimizerTimeoutMs() {
-    return metaOptimizerTimeoutMs_;
-  }
-
-  public static final int AUTO_PARALLEL_FIELD_NUMBER = 5;
-  private org.tensorflow.proto.framework.AutoParallelOptions autoParallel_;
-  /**
-   * <pre>
-   * Configures AutoParallel optimization passes either through the
-   * meta-optimizer or when manually specified through the optimizers field.
-   * </pre>
-   *
-   * <code>.tensorflow.AutoParallelOptions auto_parallel = 5;</code>
-   */
-  public boolean hasAutoParallel() {
-    return autoParallel_ != null;
-  }
-  /**
-   * <pre>
-   * Configures AutoParallel optimization passes either through the
-   * meta-optimizer or when manually specified through the optimizers field.
-   * </pre>
-   *
-   * <code>.tensorflow.AutoParallelOptions auto_parallel = 5;</code>
-   */
-  public org.tensorflow.proto.framework.AutoParallelOptions getAutoParallel() {
-    return autoParallel_ == null ? org.tensorflow.proto.framework.AutoParallelOptions.getDefaultInstance() : autoParallel_;
-  }
-  /**
-   * <pre>
-   * Configures AutoParallel optimization passes either through the
-   * meta-optimizer or when manually specified through the optimizers field.
-   * </pre>
-   *
-   * <code>.tensorflow.AutoParallelOptions auto_parallel = 5;</code>
-   */
-  public org.tensorflow.proto.framework.AutoParallelOptionsOrBuilder getAutoParallelOrBuilder() {
-    return getAutoParallel();
-  }
-
-  public static final int FAIL_ON_OPTIMIZER_ERRORS_FIELD_NUMBER = 21;
-  private boolean failOnOptimizerErrors_;
-  /**
-   * <pre>
-   * If true, any optimization pass failing will cause the MetaOptimizer to
-   * stop with an error. By default - or when set to false, failing passes are
-   * skipped silently.
-   * </pre>
-   *
-   * <code>bool fail_on_optimizer_errors = 21;</code>
-   */
-  public boolean getFailOnOptimizerErrors() {
-    return failOnOptimizerErrors_;
-  }
-
-  public static final int SCOPED_ALLOCATOR_OPTS_FIELD_NUMBER = 16;
-  private org.tensorflow.proto.framework.ScopedAllocatorOptions scopedAllocatorOpts_;
-  /**
-   * <code>.tensorflow.ScopedAllocatorOptions scoped_allocator_opts = 16;</code>
-   */
-  public boolean hasScopedAllocatorOpts() {
-    return scopedAllocatorOpts_ != null;
-  }
-  /**
-   * <code>.tensorflow.ScopedAllocatorOptions scoped_allocator_opts = 16;</code>
-   */
-  public org.tensorflow.proto.framework.ScopedAllocatorOptions getScopedAllocatorOpts() {
-    return scopedAllocatorOpts_ == null ? org.tensorflow.proto.framework.ScopedAllocatorOptions.getDefaultInstance() : scopedAllocatorOpts_;
-  }
-  /**
-   * <code>.tensorflow.ScopedAllocatorOptions scoped_allocator_opts = 16;</code>
-   */
-  public org.tensorflow.proto.framework.ScopedAllocatorOptionsOrBuilder getScopedAllocatorOptsOrBuilder() {
-    return getScopedAllocatorOpts();
-  }
-
-  public static final int OPTIMIZERS_FIELD_NUMBER = 100;
-  private com.google.protobuf.LazyStringList optimizers_;
-  /**
-   * <pre>
-   * If non-empty, will use this as an alternative way to specify a list of
-   * optimizations to turn on and the order of the optimizations (replacing the
-   * meta-optimizer).
-   * Of the RewriterConfig options, only the AutoParallel configuration options
-   * (the auto_parallel field) apply to manually requested optimization passes
-   * ("autoparallel"). Memory optimization passes ("memory") invoked here are
-   * not configurable (in contrast to memory optimization passes through the
-   * meta-optimizer) and act only on manual op annotations.
-   * Custom optimizers (see custom_optimizers) that are not part of this
-   * schedule will be run after - in the order that they were specified.
-   * </pre>
-   *
-   * <code>repeated string optimizers = 100;</code>
-   */
-  public com.google.protobuf.ProtocolStringList
-      getOptimizersList() {
-    return optimizers_;
-  }
-  /**
-   * <pre>
-   * If non-empty, will use this as an alternative way to specify a list of
-   * optimizations to turn on and the order of the optimizations (replacing the
-   * meta-optimizer).
-   * Of the RewriterConfig options, only the AutoParallel configuration options
-   * (the auto_parallel field) apply to manually requested optimization passes
-   * ("autoparallel"). Memory optimization passes ("memory") invoked here are
-   * not configurable (in contrast to memory optimization passes through the
-   * meta-optimizer) and act only on manual op annotations.
-   * Custom optimizers (see custom_optimizers) that are not part of this
-   * schedule will be run after - in the order that they were specified.
-   * </pre>
-   *
-   * <code>repeated string optimizers = 100;</code>
-   */
-  public int getOptimizersCount() {
-    return optimizers_.size();
-  }
-  /**
-   * <pre>
-   * If non-empty, will use this as an alternative way to specify a list of
-   * optimizations to turn on and the order of the optimizations (replacing the
-   * meta-optimizer).
-   * Of the RewriterConfig options, only the AutoParallel configuration options
-   * (the auto_parallel field) apply to manually requested optimization passes
-   * ("autoparallel"). Memory optimization passes ("memory") invoked here are
-   * not configurable (in contrast to memory optimization passes through the
-   * meta-optimizer) and act only on manual op annotations.
-   * Custom optimizers (see custom_optimizers) that are not part of this
-   * schedule will be run after - in the order that they were specified.
-   * </pre>
-   *
-   * <code>repeated string optimizers = 100;</code>
-   */
-  public java.lang.String getOptimizers(int index) {
-    return optimizers_.get(index);
-  }
-  /**
-   * <pre>
-   * If non-empty, will use this as an alternative way to specify a list of
-   * optimizations to turn on and the order of the optimizations (replacing the
-   * meta-optimizer).
-   * Of the RewriterConfig options, only the AutoParallel configuration options
-   * (the auto_parallel field) apply to manually requested optimization passes
-   * ("autoparallel"). Memory optimization passes ("memory") invoked here are
-   * not configurable (in contrast to memory optimization passes through the
-   * meta-optimizer) and act only on manual op annotations.
-   * Custom optimizers (see custom_optimizers) that are not part of this
-   * schedule will be run after - in the order that they were specified.
-   * </pre>
-   *
-   * <code>repeated string optimizers = 100;</code>
-   */
-  public com.google.protobuf.ByteString
-      getOptimizersBytes(int index) {
-    return optimizers_.getByteString(index);
-  }
-
-  public static final int CUSTOM_OPTIMIZERS_FIELD_NUMBER = 200;
-  private java.util.List<org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizer> customOptimizers_;
-  /**
-   * <pre>
-   * list of CustomGraphOptimizers to apply.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.RewriterConfig.CustomGraphOptimizer custom_optimizers = 200;</code>
-   */
-  public java.util.List<org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizer> getCustomOptimizersList() {
-    return customOptimizers_;
-  }
-  /**
-   * <pre>
-   * list of CustomGraphOptimizers to apply.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.RewriterConfig.CustomGraphOptimizer custom_optimizers = 200;</code>
-   */
-  public java.util.List<? extends org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizerOrBuilder> 
-      getCustomOptimizersOrBuilderList() {
-    return customOptimizers_;
-  }
-  /**
-   * <pre>
-   * list of CustomGraphOptimizers to apply.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.RewriterConfig.CustomGraphOptimizer custom_optimizers = 200;</code>
-   */
-  public int getCustomOptimizersCount() {
-    return customOptimizers_.size();
-  }
-  /**
-   * <pre>
-   * list of CustomGraphOptimizers to apply.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.RewriterConfig.CustomGraphOptimizer custom_optimizers = 200;</code>
-   */
-  public org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizer getCustomOptimizers(int index) {
-    return customOptimizers_.get(index);
-  }
-  /**
-   * <pre>
-   * list of CustomGraphOptimizers to apply.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.RewriterConfig.CustomGraphOptimizer custom_optimizers = 200;</code>
-   */
-  public org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizerOrBuilder getCustomOptimizersOrBuilder(
-      int index) {
-    return customOptimizers_.get(index);
-  }
-
-  public static final int INTER_OPTIMIZER_VERIFIER_CONFIG_FIELD_NUMBER = 300;
-  private org.tensorflow.proto.framework.VerifierConfig interOptimizerVerifierConfig_;
-  /**
-   * <pre>
-   * VerifierConfig specifying the verifiers to be run after every optimizer.
-   * </pre>
-   *
-   * <code>.tensorflow.VerifierConfig inter_optimizer_verifier_config = 300;</code>
-   */
-  public boolean hasInterOptimizerVerifierConfig() {
-    return interOptimizerVerifierConfig_ != null;
-  }
-  /**
-   * <pre>
-   * VerifierConfig specifying the verifiers to be run after every optimizer.
-   * </pre>
-   *
-   * <code>.tensorflow.VerifierConfig inter_optimizer_verifier_config = 300;</code>
-   */
-  public org.tensorflow.proto.framework.VerifierConfig getInterOptimizerVerifierConfig() {
-    return interOptimizerVerifierConfig_ == null ? org.tensorflow.proto.framework.VerifierConfig.getDefaultInstance() : interOptimizerVerifierConfig_;
-  }
-  /**
-   * <pre>
-   * VerifierConfig specifying the verifiers to be run after every optimizer.
-   * </pre>
-   *
-   * <code>.tensorflow.VerifierConfig inter_optimizer_verifier_config = 300;</code>
-   */
-  public org.tensorflow.proto.framework.VerifierConfigOrBuilder getInterOptimizerVerifierConfigOrBuilder() {
-    return getInterOptimizerVerifierConfig();
-  }
-
-  public static final int POST_OPTIMIZATION_VERIFIER_CONFIG_FIELD_NUMBER = 301;
-  private org.tensorflow.proto.framework.VerifierConfig postOptimizationVerifierConfig_;
-  /**
-   * <pre>
-   * VerifierConfig specifying the verifiers to be run at the end, after all
-   * optimizers have run.
-   * </pre>
-   *
-   * <code>.tensorflow.VerifierConfig post_optimization_verifier_config = 301;</code>
-   */
-  public boolean hasPostOptimizationVerifierConfig() {
-    return postOptimizationVerifierConfig_ != null;
-  }
-  /**
-   * <pre>
-   * VerifierConfig specifying the verifiers to be run at the end, after all
-   * optimizers have run.
-   * </pre>
-   *
-   * <code>.tensorflow.VerifierConfig post_optimization_verifier_config = 301;</code>
-   */
-  public org.tensorflow.proto.framework.VerifierConfig getPostOptimizationVerifierConfig() {
-    return postOptimizationVerifierConfig_ == null ? org.tensorflow.proto.framework.VerifierConfig.getDefaultInstance() : postOptimizationVerifierConfig_;
-  }
-  /**
-   * <pre>
-   * VerifierConfig specifying the verifiers to be run at the end, after all
-   * optimizers have run.
-   * </pre>
-   *
-   * <code>.tensorflow.VerifierConfig post_optimization_verifier_config = 301;</code>
-   */
-  public org.tensorflow.proto.framework.VerifierConfigOrBuilder getPostOptimizationVerifierConfigOrBuilder() {
-    return getPostOptimizationVerifierConfig();
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    if (layoutOptimizer_ != org.tensorflow.proto.framework.RewriterConfig.Toggle.DEFAULT.getNumber()) {
-      output.writeEnum(1, layoutOptimizer_);
-    }
-    if (disableModelPruning_ != false) {
-      output.writeBool(2, disableModelPruning_);
-    }
-    if (constantFolding_ != org.tensorflow.proto.framework.RewriterConfig.Toggle.DEFAULT.getNumber()) {
-      output.writeEnum(3, constantFolding_);
-    }
-    if (memoryOptimization_ != org.tensorflow.proto.framework.RewriterConfig.MemOptType.DEFAULT_MEM_OPT.getNumber()) {
-      output.writeEnum(4, memoryOptimization_);
-    }
-    if (autoParallel_ != null) {
-      output.writeMessage(5, getAutoParallel());
-    }
-    if (!getMemoryOptimizerTargetNodeNameScopeBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 6, memoryOptimizerTargetNodeNameScope_);
-    }
-    if (arithmeticOptimization_ != org.tensorflow.proto.framework.RewriterConfig.Toggle.DEFAULT.getNumber()) {
-      output.writeEnum(7, arithmeticOptimization_);
-    }
-    if (dependencyOptimization_ != org.tensorflow.proto.framework.RewriterConfig.Toggle.DEFAULT.getNumber()) {
-      output.writeEnum(8, dependencyOptimization_);
-    }
-    if (loopOptimization_ != org.tensorflow.proto.framework.RewriterConfig.Toggle.DEFAULT.getNumber()) {
-      output.writeEnum(9, loopOptimization_);
-    }
-    if (functionOptimization_ != org.tensorflow.proto.framework.RewriterConfig.Toggle.DEFAULT.getNumber()) {
-      output.writeEnum(10, functionOptimization_);
-    }
-    if (debugStripper_ != org.tensorflow.proto.framework.RewriterConfig.Toggle.DEFAULT.getNumber()) {
-      output.writeEnum(11, debugStripper_);
-    }
-    if (metaOptimizerIterations_ != org.tensorflow.proto.framework.RewriterConfig.NumIterationsType.DEFAULT_NUM_ITERS.getNumber()) {
-      output.writeEnum(12, metaOptimizerIterations_);
-    }
-    if (shapeOptimization_ != org.tensorflow.proto.framework.RewriterConfig.Toggle.DEFAULT.getNumber()) {
-      output.writeEnum(13, shapeOptimization_);
-    }
-    if (remapping_ != org.tensorflow.proto.framework.RewriterConfig.Toggle.DEFAULT.getNumber()) {
-      output.writeEnum(14, remapping_);
-    }
-    if (scopedAllocatorOptimization_ != org.tensorflow.proto.framework.RewriterConfig.Toggle.DEFAULT.getNumber()) {
-      output.writeEnum(15, scopedAllocatorOptimization_);
-    }
-    if (scopedAllocatorOpts_ != null) {
-      output.writeMessage(16, getScopedAllocatorOpts());
-    }
-    if (minGraphNodes_ != 0) {
-      output.writeInt32(17, minGraphNodes_);
-    }
-    if (pinToHostOptimization_ != org.tensorflow.proto.framework.RewriterConfig.Toggle.DEFAULT.getNumber()) {
-      output.writeEnum(18, pinToHostOptimization_);
-    }
-    if (disableMetaOptimizer_ != false) {
-      output.writeBool(19, disableMetaOptimizer_);
-    }
-    if (metaOptimizerTimeoutMs_ != 0L) {
-      output.writeInt64(20, metaOptimizerTimeoutMs_);
-    }
-    if (failOnOptimizerErrors_ != false) {
-      output.writeBool(21, failOnOptimizerErrors_);
-    }
-    if (implementationSelector_ != org.tensorflow.proto.framework.RewriterConfig.Toggle.DEFAULT.getNumber()) {
-      output.writeEnum(22, implementationSelector_);
-    }
-    if (autoMixedPrecision_ != org.tensorflow.proto.framework.RewriterConfig.Toggle.DEFAULT.getNumber()) {
-      output.writeEnum(23, autoMixedPrecision_);
-    }
-    if (commonSubgraphElimination_ != org.tensorflow.proto.framework.RewriterConfig.Toggle.DEFAULT.getNumber()) {
-      output.writeEnum(24, commonSubgraphElimination_);
-    }
-    if (autoMixedPrecisionMkl_ != org.tensorflow.proto.framework.RewriterConfig.Toggle.DEFAULT.getNumber()) {
-      output.writeEnum(25, autoMixedPrecisionMkl_);
-    }
-    for (int i = 0; i < optimizers_.size(); i++) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 100, optimizers_.getRaw(i));
-    }
-    for (int i = 0; i < customOptimizers_.size(); i++) {
-      output.writeMessage(200, customOptimizers_.get(i));
-    }
-    if (interOptimizerVerifierConfig_ != null) {
-      output.writeMessage(300, getInterOptimizerVerifierConfig());
-    }
-    if (postOptimizationVerifierConfig_ != null) {
-      output.writeMessage(301, getPostOptimizationVerifierConfig());
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (layoutOptimizer_ != org.tensorflow.proto.framework.RewriterConfig.Toggle.DEFAULT.getNumber()) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeEnumSize(1, layoutOptimizer_);
-    }
-    if (disableModelPruning_ != false) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeBoolSize(2, disableModelPruning_);
-    }
-    if (constantFolding_ != org.tensorflow.proto.framework.RewriterConfig.Toggle.DEFAULT.getNumber()) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeEnumSize(3, constantFolding_);
-    }
-    if (memoryOptimization_ != org.tensorflow.proto.framework.RewriterConfig.MemOptType.DEFAULT_MEM_OPT.getNumber()) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeEnumSize(4, memoryOptimization_);
-    }
-    if (autoParallel_ != null) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(5, getAutoParallel());
-    }
-    if (!getMemoryOptimizerTargetNodeNameScopeBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(6, memoryOptimizerTargetNodeNameScope_);
-    }
-    if (arithmeticOptimization_ != org.tensorflow.proto.framework.RewriterConfig.Toggle.DEFAULT.getNumber()) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeEnumSize(7, arithmeticOptimization_);
-    }
-    if (dependencyOptimization_ != org.tensorflow.proto.framework.RewriterConfig.Toggle.DEFAULT.getNumber()) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeEnumSize(8, dependencyOptimization_);
-    }
-    if (loopOptimization_ != org.tensorflow.proto.framework.RewriterConfig.Toggle.DEFAULT.getNumber()) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeEnumSize(9, loopOptimization_);
-    }
-    if (functionOptimization_ != org.tensorflow.proto.framework.RewriterConfig.Toggle.DEFAULT.getNumber()) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeEnumSize(10, functionOptimization_);
-    }
-    if (debugStripper_ != org.tensorflow.proto.framework.RewriterConfig.Toggle.DEFAULT.getNumber()) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeEnumSize(11, debugStripper_);
-    }
-    if (metaOptimizerIterations_ != org.tensorflow.proto.framework.RewriterConfig.NumIterationsType.DEFAULT_NUM_ITERS.getNumber()) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeEnumSize(12, metaOptimizerIterations_);
-    }
-    if (shapeOptimization_ != org.tensorflow.proto.framework.RewriterConfig.Toggle.DEFAULT.getNumber()) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeEnumSize(13, shapeOptimization_);
-    }
-    if (remapping_ != org.tensorflow.proto.framework.RewriterConfig.Toggle.DEFAULT.getNumber()) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeEnumSize(14, remapping_);
-    }
-    if (scopedAllocatorOptimization_ != org.tensorflow.proto.framework.RewriterConfig.Toggle.DEFAULT.getNumber()) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeEnumSize(15, scopedAllocatorOptimization_);
-    }
-    if (scopedAllocatorOpts_ != null) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(16, getScopedAllocatorOpts());
-    }
-    if (minGraphNodes_ != 0) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt32Size(17, minGraphNodes_);
-    }
-    if (pinToHostOptimization_ != org.tensorflow.proto.framework.RewriterConfig.Toggle.DEFAULT.getNumber()) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeEnumSize(18, pinToHostOptimization_);
-    }
-    if (disableMetaOptimizer_ != false) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeBoolSize(19, disableMetaOptimizer_);
-    }
-    if (metaOptimizerTimeoutMs_ != 0L) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt64Size(20, metaOptimizerTimeoutMs_);
-    }
-    if (failOnOptimizerErrors_ != false) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeBoolSize(21, failOnOptimizerErrors_);
-    }
-    if (implementationSelector_ != org.tensorflow.proto.framework.RewriterConfig.Toggle.DEFAULT.getNumber()) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeEnumSize(22, implementationSelector_);
-    }
-    if (autoMixedPrecision_ != org.tensorflow.proto.framework.RewriterConfig.Toggle.DEFAULT.getNumber()) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeEnumSize(23, autoMixedPrecision_);
-    }
-    if (commonSubgraphElimination_ != org.tensorflow.proto.framework.RewriterConfig.Toggle.DEFAULT.getNumber()) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeEnumSize(24, commonSubgraphElimination_);
-    }
-    if (autoMixedPrecisionMkl_ != org.tensorflow.proto.framework.RewriterConfig.Toggle.DEFAULT.getNumber()) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeEnumSize(25, autoMixedPrecisionMkl_);
-    }
-    {
-      int dataSize = 0;
-      for (int i = 0; i < optimizers_.size(); i++) {
-        dataSize += computeStringSizeNoTag(optimizers_.getRaw(i));
-      }
-      size += dataSize;
-      size += 2 * getOptimizersList().size();
-    }
-    for (int i = 0; i < customOptimizers_.size(); i++) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(200, customOptimizers_.get(i));
-    }
-    if (interOptimizerVerifierConfig_ != null) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(300, getInterOptimizerVerifierConfig());
-    }
-    if (postOptimizationVerifierConfig_ != null) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(301, getPostOptimizationVerifierConfig());
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.framework.RewriterConfig)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.framework.RewriterConfig other = (org.tensorflow.proto.framework.RewriterConfig) obj;
-
-    if (layoutOptimizer_ != other.layoutOptimizer_) return false;
-    if (constantFolding_ != other.constantFolding_) return false;
-    if (shapeOptimization_ != other.shapeOptimization_) return false;
-    if (remapping_ != other.remapping_) return false;
-    if (commonSubgraphElimination_ != other.commonSubgraphElimination_) return false;
-    if (arithmeticOptimization_ != other.arithmeticOptimization_) return false;
-    if (dependencyOptimization_ != other.dependencyOptimization_) return false;
-    if (loopOptimization_ != other.loopOptimization_) return false;
-    if (functionOptimization_ != other.functionOptimization_) return false;
-    if (debugStripper_ != other.debugStripper_) return false;
-    if (getDisableModelPruning()
-        != other.getDisableModelPruning()) return false;
-    if (scopedAllocatorOptimization_ != other.scopedAllocatorOptimization_) return false;
-    if (pinToHostOptimization_ != other.pinToHostOptimization_) return false;
-    if (implementationSelector_ != other.implementationSelector_) return false;
-    if (autoMixedPrecision_ != other.autoMixedPrecision_) return false;
-    if (autoMixedPrecisionMkl_ != other.autoMixedPrecisionMkl_) return false;
-    if (getDisableMetaOptimizer()
-        != other.getDisableMetaOptimizer()) return false;
-    if (metaOptimizerIterations_ != other.metaOptimizerIterations_) return false;
-    if (getMinGraphNodes()
-        != other.getMinGraphNodes()) return false;
-    if (memoryOptimization_ != other.memoryOptimization_) return false;
-    if (!getMemoryOptimizerTargetNodeNameScope()
-        .equals(other.getMemoryOptimizerTargetNodeNameScope())) return false;
-    if (getMetaOptimizerTimeoutMs()
-        != other.getMetaOptimizerTimeoutMs()) return false;
-    if (hasAutoParallel() != other.hasAutoParallel()) return false;
-    if (hasAutoParallel()) {
-      if (!getAutoParallel()
-          .equals(other.getAutoParallel())) return false;
-    }
-    if (getFailOnOptimizerErrors()
-        != other.getFailOnOptimizerErrors()) return false;
-    if (hasScopedAllocatorOpts() != other.hasScopedAllocatorOpts()) return false;
-    if (hasScopedAllocatorOpts()) {
-      if (!getScopedAllocatorOpts()
-          .equals(other.getScopedAllocatorOpts())) return false;
-    }
-    if (!getOptimizersList()
-        .equals(other.getOptimizersList())) return false;
-    if (!getCustomOptimizersList()
-        .equals(other.getCustomOptimizersList())) return false;
-    if (hasInterOptimizerVerifierConfig() != other.hasInterOptimizerVerifierConfig()) return false;
-    if (hasInterOptimizerVerifierConfig()) {
-      if (!getInterOptimizerVerifierConfig()
-          .equals(other.getInterOptimizerVerifierConfig())) return false;
-    }
-    if (hasPostOptimizationVerifierConfig() != other.hasPostOptimizationVerifierConfig()) return false;
-    if (hasPostOptimizationVerifierConfig()) {
-      if (!getPostOptimizationVerifierConfig()
-          .equals(other.getPostOptimizationVerifierConfig())) return false;
-    }
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    hash = (37 * hash) + LAYOUT_OPTIMIZER_FIELD_NUMBER;
-    hash = (53 * hash) + layoutOptimizer_;
-    hash = (37 * hash) + CONSTANT_FOLDING_FIELD_NUMBER;
-    hash = (53 * hash) + constantFolding_;
-    hash = (37 * hash) + SHAPE_OPTIMIZATION_FIELD_NUMBER;
-    hash = (53 * hash) + shapeOptimization_;
-    hash = (37 * hash) + REMAPPING_FIELD_NUMBER;
-    hash = (53 * hash) + remapping_;
-    hash = (37 * hash) + COMMON_SUBGRAPH_ELIMINATION_FIELD_NUMBER;
-    hash = (53 * hash) + commonSubgraphElimination_;
-    hash = (37 * hash) + ARITHMETIC_OPTIMIZATION_FIELD_NUMBER;
-    hash = (53 * hash) + arithmeticOptimization_;
-    hash = (37 * hash) + DEPENDENCY_OPTIMIZATION_FIELD_NUMBER;
-    hash = (53 * hash) + dependencyOptimization_;
-    hash = (37 * hash) + LOOP_OPTIMIZATION_FIELD_NUMBER;
-    hash = (53 * hash) + loopOptimization_;
-    hash = (37 * hash) + FUNCTION_OPTIMIZATION_FIELD_NUMBER;
-    hash = (53 * hash) + functionOptimization_;
-    hash = (37 * hash) + DEBUG_STRIPPER_FIELD_NUMBER;
-    hash = (53 * hash) + debugStripper_;
-    hash = (37 * hash) + DISABLE_MODEL_PRUNING_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashBoolean(
-        getDisableModelPruning());
-    hash = (37 * hash) + SCOPED_ALLOCATOR_OPTIMIZATION_FIELD_NUMBER;
-    hash = (53 * hash) + scopedAllocatorOptimization_;
-    hash = (37 * hash) + PIN_TO_HOST_OPTIMIZATION_FIELD_NUMBER;
-    hash = (53 * hash) + pinToHostOptimization_;
-    hash = (37 * hash) + IMPLEMENTATION_SELECTOR_FIELD_NUMBER;
-    hash = (53 * hash) + implementationSelector_;
-    hash = (37 * hash) + AUTO_MIXED_PRECISION_FIELD_NUMBER;
-    hash = (53 * hash) + autoMixedPrecision_;
-    hash = (37 * hash) + AUTO_MIXED_PRECISION_MKL_FIELD_NUMBER;
-    hash = (53 * hash) + autoMixedPrecisionMkl_;
-    hash = (37 * hash) + DISABLE_META_OPTIMIZER_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashBoolean(
-        getDisableMetaOptimizer());
-    hash = (37 * hash) + META_OPTIMIZER_ITERATIONS_FIELD_NUMBER;
-    hash = (53 * hash) + metaOptimizerIterations_;
-    hash = (37 * hash) + MIN_GRAPH_NODES_FIELD_NUMBER;
-    hash = (53 * hash) + getMinGraphNodes();
-    hash = (37 * hash) + MEMORY_OPTIMIZATION_FIELD_NUMBER;
-    hash = (53 * hash) + memoryOptimization_;
-    hash = (37 * hash) + MEMORY_OPTIMIZER_TARGET_NODE_NAME_SCOPE_FIELD_NUMBER;
-    hash = (53 * hash) + getMemoryOptimizerTargetNodeNameScope().hashCode();
-    hash = (37 * hash) + META_OPTIMIZER_TIMEOUT_MS_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-        getMetaOptimizerTimeoutMs());
-    if (hasAutoParallel()) {
-      hash = (37 * hash) + AUTO_PARALLEL_FIELD_NUMBER;
-      hash = (53 * hash) + getAutoParallel().hashCode();
-    }
-    hash = (37 * hash) + FAIL_ON_OPTIMIZER_ERRORS_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashBoolean(
-        getFailOnOptimizerErrors());
-    if (hasScopedAllocatorOpts()) {
-      hash = (37 * hash) + SCOPED_ALLOCATOR_OPTS_FIELD_NUMBER;
-      hash = (53 * hash) + getScopedAllocatorOpts().hashCode();
-    }
-    if (getOptimizersCount() > 0) {
-      hash = (37 * hash) + OPTIMIZERS_FIELD_NUMBER;
-      hash = (53 * hash) + getOptimizersList().hashCode();
-    }
-    if (getCustomOptimizersCount() > 0) {
-      hash = (37 * hash) + CUSTOM_OPTIMIZERS_FIELD_NUMBER;
-      hash = (53 * hash) + getCustomOptimizersList().hashCode();
-    }
-    if (hasInterOptimizerVerifierConfig()) {
-      hash = (37 * hash) + INTER_OPTIMIZER_VERIFIER_CONFIG_FIELD_NUMBER;
-      hash = (53 * hash) + getInterOptimizerVerifierConfig().hashCode();
-    }
-    if (hasPostOptimizationVerifierConfig()) {
-      hash = (37 * hash) + POST_OPTIMIZATION_VERIFIER_CONFIG_FIELD_NUMBER;
-      hash = (53 * hash) + getPostOptimizationVerifierConfig().hashCode();
-    }
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.framework.RewriterConfig parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.RewriterConfig parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.RewriterConfig parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.RewriterConfig parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.RewriterConfig parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.RewriterConfig parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.RewriterConfig parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.RewriterConfig parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.RewriterConfig parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.RewriterConfig parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.RewriterConfig parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.RewriterConfig parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.framework.RewriterConfig prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * <pre>
-   * Graph rewriting is experimental and subject to change, not covered by any
-   * API stability guarantees.
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.RewriterConfig}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.RewriterConfig)
-      org.tensorflow.proto.framework.RewriterConfigOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.RewriterConfigProtos.internal_static_tensorflow_RewriterConfig_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.RewriterConfigProtos.internal_static_tensorflow_RewriterConfig_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.RewriterConfig.class, org.tensorflow.proto.framework.RewriterConfig.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.framework.RewriterConfig.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-        getCustomOptimizersFieldBuilder();
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      layoutOptimizer_ = 0;
-
-      constantFolding_ = 0;
-
-      shapeOptimization_ = 0;
-
-      remapping_ = 0;
-
-      commonSubgraphElimination_ = 0;
-
-      arithmeticOptimization_ = 0;
-
-      dependencyOptimization_ = 0;
-
-      loopOptimization_ = 0;
-
-      functionOptimization_ = 0;
-
-      debugStripper_ = 0;
-
-      disableModelPruning_ = false;
-
-      scopedAllocatorOptimization_ = 0;
-
-      pinToHostOptimization_ = 0;
-
-      implementationSelector_ = 0;
-
-      autoMixedPrecision_ = 0;
-
-      autoMixedPrecisionMkl_ = 0;
-
-      disableMetaOptimizer_ = false;
-
-      metaOptimizerIterations_ = 0;
-
-      minGraphNodes_ = 0;
-
-      memoryOptimization_ = 0;
-
-      memoryOptimizerTargetNodeNameScope_ = "";
-
-      metaOptimizerTimeoutMs_ = 0L;
-
-      if (autoParallelBuilder_ == null) {
-        autoParallel_ = null;
-      } else {
-        autoParallel_ = null;
-        autoParallelBuilder_ = null;
-      }
-      failOnOptimizerErrors_ = false;
-
-      if (scopedAllocatorOptsBuilder_ == null) {
-        scopedAllocatorOpts_ = null;
-      } else {
-        scopedAllocatorOpts_ = null;
-        scopedAllocatorOptsBuilder_ = null;
-      }
-      optimizers_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-      bitField0_ = (bitField0_ & ~0x00000001);
-      if (customOptimizersBuilder_ == null) {
-        customOptimizers_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000002);
-      } else {
-        customOptimizersBuilder_.clear();
-      }
-      if (interOptimizerVerifierConfigBuilder_ == null) {
-        interOptimizerVerifierConfig_ = null;
-      } else {
-        interOptimizerVerifierConfig_ = null;
-        interOptimizerVerifierConfigBuilder_ = null;
-      }
-      if (postOptimizationVerifierConfigBuilder_ == null) {
-        postOptimizationVerifierConfig_ = null;
-      } else {
-        postOptimizationVerifierConfig_ = null;
-        postOptimizationVerifierConfigBuilder_ = null;
-      }
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.framework.RewriterConfigProtos.internal_static_tensorflow_RewriterConfig_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.RewriterConfig getDefaultInstanceForType() {
-      return org.tensorflow.proto.framework.RewriterConfig.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.RewriterConfig build() {
-      org.tensorflow.proto.framework.RewriterConfig result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.RewriterConfig buildPartial() {
-      org.tensorflow.proto.framework.RewriterConfig result = new org.tensorflow.proto.framework.RewriterConfig(this);
-      int from_bitField0_ = bitField0_;
-      result.layoutOptimizer_ = layoutOptimizer_;
-      result.constantFolding_ = constantFolding_;
-      result.shapeOptimization_ = shapeOptimization_;
-      result.remapping_ = remapping_;
-      result.commonSubgraphElimination_ = commonSubgraphElimination_;
-      result.arithmeticOptimization_ = arithmeticOptimization_;
-      result.dependencyOptimization_ = dependencyOptimization_;
-      result.loopOptimization_ = loopOptimization_;
-      result.functionOptimization_ = functionOptimization_;
-      result.debugStripper_ = debugStripper_;
-      result.disableModelPruning_ = disableModelPruning_;
-      result.scopedAllocatorOptimization_ = scopedAllocatorOptimization_;
-      result.pinToHostOptimization_ = pinToHostOptimization_;
-      result.implementationSelector_ = implementationSelector_;
-      result.autoMixedPrecision_ = autoMixedPrecision_;
-      result.autoMixedPrecisionMkl_ = autoMixedPrecisionMkl_;
-      result.disableMetaOptimizer_ = disableMetaOptimizer_;
-      result.metaOptimizerIterations_ = metaOptimizerIterations_;
-      result.minGraphNodes_ = minGraphNodes_;
-      result.memoryOptimization_ = memoryOptimization_;
-      result.memoryOptimizerTargetNodeNameScope_ = memoryOptimizerTargetNodeNameScope_;
-      result.metaOptimizerTimeoutMs_ = metaOptimizerTimeoutMs_;
-      if (autoParallelBuilder_ == null) {
-        result.autoParallel_ = autoParallel_;
-      } else {
-        result.autoParallel_ = autoParallelBuilder_.build();
-      }
-      result.failOnOptimizerErrors_ = failOnOptimizerErrors_;
-      if (scopedAllocatorOptsBuilder_ == null) {
-        result.scopedAllocatorOpts_ = scopedAllocatorOpts_;
-      } else {
-        result.scopedAllocatorOpts_ = scopedAllocatorOptsBuilder_.build();
-      }
-      if (((bitField0_ & 0x00000001) != 0)) {
-        optimizers_ = optimizers_.getUnmodifiableView();
-        bitField0_ = (bitField0_ & ~0x00000001);
-      }
-      result.optimizers_ = optimizers_;
-      if (customOptimizersBuilder_ == null) {
-        if (((bitField0_ & 0x00000002) != 0)) {
-          customOptimizers_ = java.util.Collections.unmodifiableList(customOptimizers_);
-          bitField0_ = (bitField0_ & ~0x00000002);
-        }
-        result.customOptimizers_ = customOptimizers_;
-      } else {
-        result.customOptimizers_ = customOptimizersBuilder_.build();
-      }
-      if (interOptimizerVerifierConfigBuilder_ == null) {
-        result.interOptimizerVerifierConfig_ = interOptimizerVerifierConfig_;
-      } else {
-        result.interOptimizerVerifierConfig_ = interOptimizerVerifierConfigBuilder_.build();
-      }
-      if (postOptimizationVerifierConfigBuilder_ == null) {
-        result.postOptimizationVerifierConfig_ = postOptimizationVerifierConfig_;
-      } else {
-        result.postOptimizationVerifierConfig_ = postOptimizationVerifierConfigBuilder_.build();
-      }
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.framework.RewriterConfig) {
-        return mergeFrom((org.tensorflow.proto.framework.RewriterConfig)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.framework.RewriterConfig other) {
-      if (other == org.tensorflow.proto.framework.RewriterConfig.getDefaultInstance()) return this;
-      if (other.layoutOptimizer_ != 0) {
-        setLayoutOptimizerValue(other.getLayoutOptimizerValue());
-      }
-      if (other.constantFolding_ != 0) {
-        setConstantFoldingValue(other.getConstantFoldingValue());
-      }
-      if (other.shapeOptimization_ != 0) {
-        setShapeOptimizationValue(other.getShapeOptimizationValue());
-      }
-      if (other.remapping_ != 0) {
-        setRemappingValue(other.getRemappingValue());
-      }
-      if (other.commonSubgraphElimination_ != 0) {
-        setCommonSubgraphEliminationValue(other.getCommonSubgraphEliminationValue());
-      }
-      if (other.arithmeticOptimization_ != 0) {
-        setArithmeticOptimizationValue(other.getArithmeticOptimizationValue());
-      }
-      if (other.dependencyOptimization_ != 0) {
-        setDependencyOptimizationValue(other.getDependencyOptimizationValue());
-      }
-      if (other.loopOptimization_ != 0) {
-        setLoopOptimizationValue(other.getLoopOptimizationValue());
-      }
-      if (other.functionOptimization_ != 0) {
-        setFunctionOptimizationValue(other.getFunctionOptimizationValue());
-      }
-      if (other.debugStripper_ != 0) {
-        setDebugStripperValue(other.getDebugStripperValue());
-      }
-      if (other.getDisableModelPruning() != false) {
-        setDisableModelPruning(other.getDisableModelPruning());
-      }
-      if (other.scopedAllocatorOptimization_ != 0) {
-        setScopedAllocatorOptimizationValue(other.getScopedAllocatorOptimizationValue());
-      }
-      if (other.pinToHostOptimization_ != 0) {
-        setPinToHostOptimizationValue(other.getPinToHostOptimizationValue());
-      }
-      if (other.implementationSelector_ != 0) {
-        setImplementationSelectorValue(other.getImplementationSelectorValue());
-      }
-      if (other.autoMixedPrecision_ != 0) {
-        setAutoMixedPrecisionValue(other.getAutoMixedPrecisionValue());
-      }
-      if (other.autoMixedPrecisionMkl_ != 0) {
-        setAutoMixedPrecisionMklValue(other.getAutoMixedPrecisionMklValue());
-      }
-      if (other.getDisableMetaOptimizer() != false) {
-        setDisableMetaOptimizer(other.getDisableMetaOptimizer());
-      }
-      if (other.metaOptimizerIterations_ != 0) {
-        setMetaOptimizerIterationsValue(other.getMetaOptimizerIterationsValue());
-      }
-      if (other.getMinGraphNodes() != 0) {
-        setMinGraphNodes(other.getMinGraphNodes());
-      }
-      if (other.memoryOptimization_ != 0) {
-        setMemoryOptimizationValue(other.getMemoryOptimizationValue());
-      }
-      if (!other.getMemoryOptimizerTargetNodeNameScope().isEmpty()) {
-        memoryOptimizerTargetNodeNameScope_ = other.memoryOptimizerTargetNodeNameScope_;
-        onChanged();
-      }
-      if (other.getMetaOptimizerTimeoutMs() != 0L) {
-        setMetaOptimizerTimeoutMs(other.getMetaOptimizerTimeoutMs());
-      }
-      if (other.hasAutoParallel()) {
-        mergeAutoParallel(other.getAutoParallel());
-      }
-      if (other.getFailOnOptimizerErrors() != false) {
-        setFailOnOptimizerErrors(other.getFailOnOptimizerErrors());
-      }
-      if (other.hasScopedAllocatorOpts()) {
-        mergeScopedAllocatorOpts(other.getScopedAllocatorOpts());
-      }
-      if (!other.optimizers_.isEmpty()) {
-        if (optimizers_.isEmpty()) {
-          optimizers_ = other.optimizers_;
-          bitField0_ = (bitField0_ & ~0x00000001);
-        } else {
-          ensureOptimizersIsMutable();
-          optimizers_.addAll(other.optimizers_);
-        }
-        onChanged();
-      }
-      if (customOptimizersBuilder_ == null) {
-        if (!other.customOptimizers_.isEmpty()) {
-          if (customOptimizers_.isEmpty()) {
-            customOptimizers_ = other.customOptimizers_;
-            bitField0_ = (bitField0_ & ~0x00000002);
-          } else {
-            ensureCustomOptimizersIsMutable();
-            customOptimizers_.addAll(other.customOptimizers_);
-          }
-          onChanged();
-        }
-      } else {
-        if (!other.customOptimizers_.isEmpty()) {
-          if (customOptimizersBuilder_.isEmpty()) {
-            customOptimizersBuilder_.dispose();
-            customOptimizersBuilder_ = null;
-            customOptimizers_ = other.customOptimizers_;
-            bitField0_ = (bitField0_ & ~0x00000002);
-            customOptimizersBuilder_ = 
-              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
-                 getCustomOptimizersFieldBuilder() : null;
-          } else {
-            customOptimizersBuilder_.addAllMessages(other.customOptimizers_);
-          }
-        }
-      }
-      if (other.hasInterOptimizerVerifierConfig()) {
-        mergeInterOptimizerVerifierConfig(other.getInterOptimizerVerifierConfig());
-      }
-      if (other.hasPostOptimizationVerifierConfig()) {
-        mergePostOptimizationVerifierConfig(other.getPostOptimizationVerifierConfig());
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.framework.RewriterConfig parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.framework.RewriterConfig) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-    private int bitField0_;
-
-    private int layoutOptimizer_ = 0;
-    /**
-     * <pre>
-     * Optimize tensor layouts (default is ON)
-     * e.g. This will try to use NCHW layout on GPU which is faster.
-     * </pre>
-     *
-     * <code>.tensorflow.RewriterConfig.Toggle layout_optimizer = 1;</code>
-     */
-    public int getLayoutOptimizerValue() {
-      return layoutOptimizer_;
-    }
-    /**
-     * <pre>
-     * Optimize tensor layouts (default is ON)
-     * e.g. This will try to use NCHW layout on GPU which is faster.
-     * </pre>
-     *
-     * <code>.tensorflow.RewriterConfig.Toggle layout_optimizer = 1;</code>
-     */
-    public Builder setLayoutOptimizerValue(int value) {
-      layoutOptimizer_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Optimize tensor layouts (default is ON)
-     * e.g. This will try to use NCHW layout on GPU which is faster.
-     * </pre>
-     *
-     * <code>.tensorflow.RewriterConfig.Toggle layout_optimizer = 1;</code>
-     */
-    public org.tensorflow.proto.framework.RewriterConfig.Toggle getLayoutOptimizer() {
-      @SuppressWarnings("deprecation")
-      org.tensorflow.proto.framework.RewriterConfig.Toggle result = org.tensorflow.proto.framework.RewriterConfig.Toggle.valueOf(layoutOptimizer_);
-      return result == null ? org.tensorflow.proto.framework.RewriterConfig.Toggle.UNRECOGNIZED : result;
-    }
-    /**
-     * <pre>
-     * Optimize tensor layouts (default is ON)
-     * e.g. This will try to use NCHW layout on GPU which is faster.
-     * </pre>
-     *
-     * <code>.tensorflow.RewriterConfig.Toggle layout_optimizer = 1;</code>
-     */
-    public Builder setLayoutOptimizer(org.tensorflow.proto.framework.RewriterConfig.Toggle value) {
-      if (value == null) {
-        throw new NullPointerException();
-      }
-      
-      layoutOptimizer_ = value.getNumber();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Optimize tensor layouts (default is ON)
-     * e.g. This will try to use NCHW layout on GPU which is faster.
-     * </pre>
-     *
-     * <code>.tensorflow.RewriterConfig.Toggle layout_optimizer = 1;</code>
-     */
-    public Builder clearLayoutOptimizer() {
-      
-      layoutOptimizer_ = 0;
-      onChanged();
-      return this;
-    }
-
-    private int constantFolding_ = 0;
-    /**
-     * <pre>
-     * Fold constants (default is ON)
-     * Statically infer the value of tensors when possible, and materialize the
-     * result using constants.
-     * </pre>
-     *
-     * <code>.tensorflow.RewriterConfig.Toggle constant_folding = 3;</code>
-     */
-    public int getConstantFoldingValue() {
-      return constantFolding_;
-    }
-    /**
-     * <pre>
-     * Fold constants (default is ON)
-     * Statically infer the value of tensors when possible, and materialize the
-     * result using constants.
-     * </pre>
-     *
-     * <code>.tensorflow.RewriterConfig.Toggle constant_folding = 3;</code>
-     */
-    public Builder setConstantFoldingValue(int value) {
-      constantFolding_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Fold constants (default is ON)
-     * Statically infer the value of tensors when possible, and materialize the
-     * result using constants.
-     * </pre>
-     *
-     * <code>.tensorflow.RewriterConfig.Toggle constant_folding = 3;</code>
-     */
-    public org.tensorflow.proto.framework.RewriterConfig.Toggle getConstantFolding() {
-      @SuppressWarnings("deprecation")
-      org.tensorflow.proto.framework.RewriterConfig.Toggle result = org.tensorflow.proto.framework.RewriterConfig.Toggle.valueOf(constantFolding_);
-      return result == null ? org.tensorflow.proto.framework.RewriterConfig.Toggle.UNRECOGNIZED : result;
-    }
-    /**
-     * <pre>
-     * Fold constants (default is ON)
-     * Statically infer the value of tensors when possible, and materialize the
-     * result using constants.
-     * </pre>
-     *
-     * <code>.tensorflow.RewriterConfig.Toggle constant_folding = 3;</code>
-     */
-    public Builder setConstantFolding(org.tensorflow.proto.framework.RewriterConfig.Toggle value) {
-      if (value == null) {
-        throw new NullPointerException();
-      }
-      
-      constantFolding_ = value.getNumber();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Fold constants (default is ON)
-     * Statically infer the value of tensors when possible, and materialize the
-     * result using constants.
-     * </pre>
-     *
-     * <code>.tensorflow.RewriterConfig.Toggle constant_folding = 3;</code>
-     */
-    public Builder clearConstantFolding() {
-      
-      constantFolding_ = 0;
-      onChanged();
-      return this;
-    }
-
-    private int shapeOptimization_ = 0;
-    /**
-     * <pre>
-     * Shape optimizations (default is ON)
-     * Simplify computations made on shapes.
-     * </pre>
-     *
-     * <code>.tensorflow.RewriterConfig.Toggle shape_optimization = 13;</code>
-     */
-    public int getShapeOptimizationValue() {
-      return shapeOptimization_;
-    }
-    /**
-     * <pre>
-     * Shape optimizations (default is ON)
-     * Simplify computations made on shapes.
-     * </pre>
-     *
-     * <code>.tensorflow.RewriterConfig.Toggle shape_optimization = 13;</code>
-     */
-    public Builder setShapeOptimizationValue(int value) {
-      shapeOptimization_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Shape optimizations (default is ON)
-     * Simplify computations made on shapes.
-     * </pre>
-     *
-     * <code>.tensorflow.RewriterConfig.Toggle shape_optimization = 13;</code>
-     */
-    public org.tensorflow.proto.framework.RewriterConfig.Toggle getShapeOptimization() {
-      @SuppressWarnings("deprecation")
-      org.tensorflow.proto.framework.RewriterConfig.Toggle result = org.tensorflow.proto.framework.RewriterConfig.Toggle.valueOf(shapeOptimization_);
-      return result == null ? org.tensorflow.proto.framework.RewriterConfig.Toggle.UNRECOGNIZED : result;
-    }
-    /**
-     * <pre>
-     * Shape optimizations (default is ON)
-     * Simplify computations made on shapes.
-     * </pre>
-     *
-     * <code>.tensorflow.RewriterConfig.Toggle shape_optimization = 13;</code>
-     */
-    public Builder setShapeOptimization(org.tensorflow.proto.framework.RewriterConfig.Toggle value) {
-      if (value == null) {
-        throw new NullPointerException();
-      }
-      
-      shapeOptimization_ = value.getNumber();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Shape optimizations (default is ON)
-     * Simplify computations made on shapes.
-     * </pre>
-     *
-     * <code>.tensorflow.RewriterConfig.Toggle shape_optimization = 13;</code>
-     */
-    public Builder clearShapeOptimization() {
-      
-      shapeOptimization_ = 0;
-      onChanged();
-      return this;
-    }
-
-    private int remapping_ = 0;
-    /**
-     * <pre>
-     * Remapping (default is ON)
-     * Remap subgraphs onto more efficient implementations.
-     * </pre>
-     *
-     * <code>.tensorflow.RewriterConfig.Toggle remapping = 14;</code>
-     */
-    public int getRemappingValue() {
-      return remapping_;
-    }
-    /**
-     * <pre>
-     * Remapping (default is ON)
-     * Remap subgraphs onto more efficient implementations.
-     * </pre>
-     *
-     * <code>.tensorflow.RewriterConfig.Toggle remapping = 14;</code>
-     */
-    public Builder setRemappingValue(int value) {
-      remapping_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Remapping (default is ON)
-     * Remap subgraphs onto more efficient implementations.
-     * </pre>
-     *
-     * <code>.tensorflow.RewriterConfig.Toggle remapping = 14;</code>
-     */
-    public org.tensorflow.proto.framework.RewriterConfig.Toggle getRemapping() {
-      @SuppressWarnings("deprecation")
-      org.tensorflow.proto.framework.RewriterConfig.Toggle result = org.tensorflow.proto.framework.RewriterConfig.Toggle.valueOf(remapping_);
-      return result == null ? org.tensorflow.proto.framework.RewriterConfig.Toggle.UNRECOGNIZED : result;
-    }
-    /**
-     * <pre>
-     * Remapping (default is ON)
-     * Remap subgraphs onto more efficient implementations.
-     * </pre>
-     *
-     * <code>.tensorflow.RewriterConfig.Toggle remapping = 14;</code>
-     */
-    public Builder setRemapping(org.tensorflow.proto.framework.RewriterConfig.Toggle value) {
-      if (value == null) {
-        throw new NullPointerException();
-      }
-      
-      remapping_ = value.getNumber();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Remapping (default is ON)
-     * Remap subgraphs onto more efficient implementations.
-     * </pre>
-     *
-     * <code>.tensorflow.RewriterConfig.Toggle remapping = 14;</code>
-     */
-    public Builder clearRemapping() {
-      
-      remapping_ = 0;
-      onChanged();
-      return this;
-    }
-
-    private int commonSubgraphElimination_ = 0;
-    /**
-     * <pre>
-     * Common subgraph elimination (default is ON)
-     * e.g. Simplify arithmetic ops; merge ops with same value (like constants).
-     * </pre>
-     *
-     * <code>.tensorflow.RewriterConfig.Toggle common_subgraph_elimination = 24;</code>
-     */
-    public int getCommonSubgraphEliminationValue() {
-      return commonSubgraphElimination_;
-    }
-    /**
-     * <pre>
-     * Common subgraph elimination (default is ON)
-     * e.g. Simplify arithmetic ops; merge ops with same value (like constants).
-     * </pre>
-     *
-     * <code>.tensorflow.RewriterConfig.Toggle common_subgraph_elimination = 24;</code>
-     */
-    public Builder setCommonSubgraphEliminationValue(int value) {
-      commonSubgraphElimination_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Common subgraph elimination (default is ON)
-     * e.g. Simplify arithmetic ops; merge ops with same value (like constants).
-     * </pre>
-     *
-     * <code>.tensorflow.RewriterConfig.Toggle common_subgraph_elimination = 24;</code>
-     */
-    public org.tensorflow.proto.framework.RewriterConfig.Toggle getCommonSubgraphElimination() {
-      @SuppressWarnings("deprecation")
-      org.tensorflow.proto.framework.RewriterConfig.Toggle result = org.tensorflow.proto.framework.RewriterConfig.Toggle.valueOf(commonSubgraphElimination_);
-      return result == null ? org.tensorflow.proto.framework.RewriterConfig.Toggle.UNRECOGNIZED : result;
-    }
-    /**
-     * <pre>
-     * Common subgraph elimination (default is ON)
-     * e.g. Simplify arithmetic ops; merge ops with same value (like constants).
-     * </pre>
-     *
-     * <code>.tensorflow.RewriterConfig.Toggle common_subgraph_elimination = 24;</code>
-     */
-    public Builder setCommonSubgraphElimination(org.tensorflow.proto.framework.RewriterConfig.Toggle value) {
-      if (value == null) {
-        throw new NullPointerException();
-      }
-      
-      commonSubgraphElimination_ = value.getNumber();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Common subgraph elimination (default is ON)
-     * e.g. Simplify arithmetic ops; merge ops with same value (like constants).
-     * </pre>
-     *
-     * <code>.tensorflow.RewriterConfig.Toggle common_subgraph_elimination = 24;</code>
-     */
-    public Builder clearCommonSubgraphElimination() {
-      
-      commonSubgraphElimination_ = 0;
-      onChanged();
-      return this;
-    }
-
-    private int arithmeticOptimization_ = 0;
-    /**
-     * <pre>
-     * Arithmetic optimizations (default is ON)
-     * e.g. Simplify arithmetic ops; merge ops with same value (like constants).
-     * </pre>
-     *
-     * <code>.tensorflow.RewriterConfig.Toggle arithmetic_optimization = 7;</code>
-     */
-    public int getArithmeticOptimizationValue() {
-      return arithmeticOptimization_;
-    }
-    /**
-     * <pre>
-     * Arithmetic optimizations (default is ON)
-     * e.g. Simplify arithmetic ops; merge ops with same value (like constants).
-     * </pre>
-     *
-     * <code>.tensorflow.RewriterConfig.Toggle arithmetic_optimization = 7;</code>
-     */
-    public Builder setArithmeticOptimizationValue(int value) {
-      arithmeticOptimization_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Arithmetic optimizations (default is ON)
-     * e.g. Simplify arithmetic ops; merge ops with same value (like constants).
-     * </pre>
-     *
-     * <code>.tensorflow.RewriterConfig.Toggle arithmetic_optimization = 7;</code>
-     */
-    public org.tensorflow.proto.framework.RewriterConfig.Toggle getArithmeticOptimization() {
-      @SuppressWarnings("deprecation")
-      org.tensorflow.proto.framework.RewriterConfig.Toggle result = org.tensorflow.proto.framework.RewriterConfig.Toggle.valueOf(arithmeticOptimization_);
-      return result == null ? org.tensorflow.proto.framework.RewriterConfig.Toggle.UNRECOGNIZED : result;
-    }
-    /**
-     * <pre>
-     * Arithmetic optimizations (default is ON)
-     * e.g. Simplify arithmetic ops; merge ops with same value (like constants).
-     * </pre>
-     *
-     * <code>.tensorflow.RewriterConfig.Toggle arithmetic_optimization = 7;</code>
-     */
-    public Builder setArithmeticOptimization(org.tensorflow.proto.framework.RewriterConfig.Toggle value) {
-      if (value == null) {
-        throw new NullPointerException();
-      }
-      
-      arithmeticOptimization_ = value.getNumber();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Arithmetic optimizations (default is ON)
-     * e.g. Simplify arithmetic ops; merge ops with same value (like constants).
-     * </pre>
-     *
-     * <code>.tensorflow.RewriterConfig.Toggle arithmetic_optimization = 7;</code>
-     */
-    public Builder clearArithmeticOptimization() {
-      
-      arithmeticOptimization_ = 0;
-      onChanged();
-      return this;
-    }
-
-    private int dependencyOptimization_ = 0;
-    /**
-     * <pre>
-     * Control dependency optimizations (default is ON).
-     * Remove redundant control dependencies, which may enable other optimization.
-     * </pre>
-     *
-     * <code>.tensorflow.RewriterConfig.Toggle dependency_optimization = 8;</code>
-     */
-    public int getDependencyOptimizationValue() {
-      return dependencyOptimization_;
-    }
-    /**
-     * <pre>
-     * Control dependency optimizations (default is ON).
-     * Remove redundant control dependencies, which may enable other optimization.
-     * </pre>
-     *
-     * <code>.tensorflow.RewriterConfig.Toggle dependency_optimization = 8;</code>
-     */
-    public Builder setDependencyOptimizationValue(int value) {
-      dependencyOptimization_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Control dependency optimizations (default is ON).
-     * Remove redundant control dependencies, which may enable other optimization.
-     * </pre>
-     *
-     * <code>.tensorflow.RewriterConfig.Toggle dependency_optimization = 8;</code>
-     */
-    public org.tensorflow.proto.framework.RewriterConfig.Toggle getDependencyOptimization() {
-      @SuppressWarnings("deprecation")
-      org.tensorflow.proto.framework.RewriterConfig.Toggle result = org.tensorflow.proto.framework.RewriterConfig.Toggle.valueOf(dependencyOptimization_);
-      return result == null ? org.tensorflow.proto.framework.RewriterConfig.Toggle.UNRECOGNIZED : result;
-    }
-    /**
-     * <pre>
-     * Control dependency optimizations (default is ON).
-     * Remove redundant control dependencies, which may enable other optimization.
-     * </pre>
-     *
-     * <code>.tensorflow.RewriterConfig.Toggle dependency_optimization = 8;</code>
-     */
-    public Builder setDependencyOptimization(org.tensorflow.proto.framework.RewriterConfig.Toggle value) {
-      if (value == null) {
-        throw new NullPointerException();
-      }
-      
-      dependencyOptimization_ = value.getNumber();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Control dependency optimizations (default is ON).
-     * Remove redundant control dependencies, which may enable other optimization.
-     * </pre>
-     *
-     * <code>.tensorflow.RewriterConfig.Toggle dependency_optimization = 8;</code>
-     */
-    public Builder clearDependencyOptimization() {
-      
-      dependencyOptimization_ = 0;
-      onChanged();
-      return this;
-    }
-
-    private int loopOptimization_ = 0;
-    /**
-     * <pre>
-     * Loop optimizations (default is ON).
-     * </pre>
-     *
-     * <code>.tensorflow.RewriterConfig.Toggle loop_optimization = 9;</code>
-     */
-    public int getLoopOptimizationValue() {
-      return loopOptimization_;
-    }
-    /**
-     * <pre>
-     * Loop optimizations (default is ON).
-     * </pre>
-     *
-     * <code>.tensorflow.RewriterConfig.Toggle loop_optimization = 9;</code>
-     */
-    public Builder setLoopOptimizationValue(int value) {
-      loopOptimization_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Loop optimizations (default is ON).
-     * </pre>
-     *
-     * <code>.tensorflow.RewriterConfig.Toggle loop_optimization = 9;</code>
-     */
-    public org.tensorflow.proto.framework.RewriterConfig.Toggle getLoopOptimization() {
-      @SuppressWarnings("deprecation")
-      org.tensorflow.proto.framework.RewriterConfig.Toggle result = org.tensorflow.proto.framework.RewriterConfig.Toggle.valueOf(loopOptimization_);
-      return result == null ? org.tensorflow.proto.framework.RewriterConfig.Toggle.UNRECOGNIZED : result;
-    }
-    /**
-     * <pre>
-     * Loop optimizations (default is ON).
-     * </pre>
-     *
-     * <code>.tensorflow.RewriterConfig.Toggle loop_optimization = 9;</code>
-     */
-    public Builder setLoopOptimization(org.tensorflow.proto.framework.RewriterConfig.Toggle value) {
-      if (value == null) {
-        throw new NullPointerException();
-      }
-      
-      loopOptimization_ = value.getNumber();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Loop optimizations (default is ON).
-     * </pre>
-     *
-     * <code>.tensorflow.RewriterConfig.Toggle loop_optimization = 9;</code>
-     */
-    public Builder clearLoopOptimization() {
-      
-      loopOptimization_ = 0;
-      onChanged();
-      return this;
-    }
-
-    private int functionOptimization_ = 0;
-    /**
-     * <pre>
-     * Function optimizations (default is ON).
-     * </pre>
-     *
-     * <code>.tensorflow.RewriterConfig.Toggle function_optimization = 10;</code>
-     */
-    public int getFunctionOptimizationValue() {
-      return functionOptimization_;
-    }
-    /**
-     * <pre>
-     * Function optimizations (default is ON).
-     * </pre>
-     *
-     * <code>.tensorflow.RewriterConfig.Toggle function_optimization = 10;</code>
-     */
-    public Builder setFunctionOptimizationValue(int value) {
-      functionOptimization_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Function optimizations (default is ON).
-     * </pre>
-     *
-     * <code>.tensorflow.RewriterConfig.Toggle function_optimization = 10;</code>
-     */
-    public org.tensorflow.proto.framework.RewriterConfig.Toggle getFunctionOptimization() {
-      @SuppressWarnings("deprecation")
-      org.tensorflow.proto.framework.RewriterConfig.Toggle result = org.tensorflow.proto.framework.RewriterConfig.Toggle.valueOf(functionOptimization_);
-      return result == null ? org.tensorflow.proto.framework.RewriterConfig.Toggle.UNRECOGNIZED : result;
-    }
-    /**
-     * <pre>
-     * Function optimizations (default is ON).
-     * </pre>
-     *
-     * <code>.tensorflow.RewriterConfig.Toggle function_optimization = 10;</code>
-     */
-    public Builder setFunctionOptimization(org.tensorflow.proto.framework.RewriterConfig.Toggle value) {
-      if (value == null) {
-        throw new NullPointerException();
-      }
-      
-      functionOptimization_ = value.getNumber();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Function optimizations (default is ON).
-     * </pre>
-     *
-     * <code>.tensorflow.RewriterConfig.Toggle function_optimization = 10;</code>
-     */
-    public Builder clearFunctionOptimization() {
-      
-      functionOptimization_ = 0;
-      onChanged();
-      return this;
-    }
-
-    private int debugStripper_ = 0;
-    /**
-     * <pre>
-     * Strips debug-related nodes from the graph (off by default).
-     * </pre>
-     *
-     * <code>.tensorflow.RewriterConfig.Toggle debug_stripper = 11;</code>
-     */
-    public int getDebugStripperValue() {
-      return debugStripper_;
-    }
-    /**
-     * <pre>
-     * Strips debug-related nodes from the graph (off by default).
-     * </pre>
-     *
-     * <code>.tensorflow.RewriterConfig.Toggle debug_stripper = 11;</code>
-     */
-    public Builder setDebugStripperValue(int value) {
-      debugStripper_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Strips debug-related nodes from the graph (off by default).
-     * </pre>
-     *
-     * <code>.tensorflow.RewriterConfig.Toggle debug_stripper = 11;</code>
-     */
-    public org.tensorflow.proto.framework.RewriterConfig.Toggle getDebugStripper() {
-      @SuppressWarnings("deprecation")
-      org.tensorflow.proto.framework.RewriterConfig.Toggle result = org.tensorflow.proto.framework.RewriterConfig.Toggle.valueOf(debugStripper_);
-      return result == null ? org.tensorflow.proto.framework.RewriterConfig.Toggle.UNRECOGNIZED : result;
-    }
-    /**
-     * <pre>
-     * Strips debug-related nodes from the graph (off by default).
-     * </pre>
-     *
-     * <code>.tensorflow.RewriterConfig.Toggle debug_stripper = 11;</code>
-     */
-    public Builder setDebugStripper(org.tensorflow.proto.framework.RewriterConfig.Toggle value) {
-      if (value == null) {
-        throw new NullPointerException();
-      }
-      
-      debugStripper_ = value.getNumber();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Strips debug-related nodes from the graph (off by default).
-     * </pre>
-     *
-     * <code>.tensorflow.RewriterConfig.Toggle debug_stripper = 11;</code>
-     */
-    public Builder clearDebugStripper() {
-      
-      debugStripper_ = 0;
-      onChanged();
-      return this;
-    }
-
-    private boolean disableModelPruning_ ;
-    /**
-     * <pre>
-     * If true, don't remove unnecessary ops from the graph
-     * </pre>
-     *
-     * <code>bool disable_model_pruning = 2;</code>
-     */
-    public boolean getDisableModelPruning() {
-      return disableModelPruning_;
-    }
-    /**
-     * <pre>
-     * If true, don't remove unnecessary ops from the graph
-     * </pre>
-     *
-     * <code>bool disable_model_pruning = 2;</code>
-     */
-    public Builder setDisableModelPruning(boolean value) {
-      
-      disableModelPruning_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * If true, don't remove unnecessary ops from the graph
-     * </pre>
-     *
-     * <code>bool disable_model_pruning = 2;</code>
-     */
-    public Builder clearDisableModelPruning() {
-      
-      disableModelPruning_ = false;
-      onChanged();
-      return this;
-    }
-
-    private int scopedAllocatorOptimization_ = 0;
-    /**
-     * <pre>
-     * Try to allocate some independent Op outputs contiguously in order to
-     * merge or eliminate downstream Ops (off by default).
-     * </pre>
-     *
-     * <code>.tensorflow.RewriterConfig.Toggle scoped_allocator_optimization = 15;</code>
-     */
-    public int getScopedAllocatorOptimizationValue() {
-      return scopedAllocatorOptimization_;
-    }
-    /**
-     * <pre>
-     * Try to allocate some independent Op outputs contiguously in order to
-     * merge or eliminate downstream Ops (off by default).
-     * </pre>
-     *
-     * <code>.tensorflow.RewriterConfig.Toggle scoped_allocator_optimization = 15;</code>
-     */
-    public Builder setScopedAllocatorOptimizationValue(int value) {
-      scopedAllocatorOptimization_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Try to allocate some independent Op outputs contiguously in order to
-     * merge or eliminate downstream Ops (off by default).
-     * </pre>
-     *
-     * <code>.tensorflow.RewriterConfig.Toggle scoped_allocator_optimization = 15;</code>
-     */
-    public org.tensorflow.proto.framework.RewriterConfig.Toggle getScopedAllocatorOptimization() {
-      @SuppressWarnings("deprecation")
-      org.tensorflow.proto.framework.RewriterConfig.Toggle result = org.tensorflow.proto.framework.RewriterConfig.Toggle.valueOf(scopedAllocatorOptimization_);
-      return result == null ? org.tensorflow.proto.framework.RewriterConfig.Toggle.UNRECOGNIZED : result;
-    }
-    /**
-     * <pre>
-     * Try to allocate some independent Op outputs contiguously in order to
-     * merge or eliminate downstream Ops (off by default).
-     * </pre>
-     *
-     * <code>.tensorflow.RewriterConfig.Toggle scoped_allocator_optimization = 15;</code>
-     */
-    public Builder setScopedAllocatorOptimization(org.tensorflow.proto.framework.RewriterConfig.Toggle value) {
-      if (value == null) {
-        throw new NullPointerException();
-      }
-      
-      scopedAllocatorOptimization_ = value.getNumber();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Try to allocate some independent Op outputs contiguously in order to
-     * merge or eliminate downstream Ops (off by default).
-     * </pre>
-     *
-     * <code>.tensorflow.RewriterConfig.Toggle scoped_allocator_optimization = 15;</code>
-     */
-    public Builder clearScopedAllocatorOptimization() {
-      
-      scopedAllocatorOptimization_ = 0;
-      onChanged();
-      return this;
-    }
-
-    private int pinToHostOptimization_ = 0;
-    /**
-     * <pre>
-     * Force small ops onto the CPU (default is OFF).
-     * </pre>
-     *
-     * <code>.tensorflow.RewriterConfig.Toggle pin_to_host_optimization = 18;</code>
-     */
-    public int getPinToHostOptimizationValue() {
-      return pinToHostOptimization_;
-    }
-    /**
-     * <pre>
-     * Force small ops onto the CPU (default is OFF).
-     * </pre>
-     *
-     * <code>.tensorflow.RewriterConfig.Toggle pin_to_host_optimization = 18;</code>
-     */
-    public Builder setPinToHostOptimizationValue(int value) {
-      pinToHostOptimization_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Force small ops onto the CPU (default is OFF).
-     * </pre>
-     *
-     * <code>.tensorflow.RewriterConfig.Toggle pin_to_host_optimization = 18;</code>
-     */
-    public org.tensorflow.proto.framework.RewriterConfig.Toggle getPinToHostOptimization() {
-      @SuppressWarnings("deprecation")
-      org.tensorflow.proto.framework.RewriterConfig.Toggle result = org.tensorflow.proto.framework.RewriterConfig.Toggle.valueOf(pinToHostOptimization_);
-      return result == null ? org.tensorflow.proto.framework.RewriterConfig.Toggle.UNRECOGNIZED : result;
-    }
-    /**
-     * <pre>
-     * Force small ops onto the CPU (default is OFF).
-     * </pre>
-     *
-     * <code>.tensorflow.RewriterConfig.Toggle pin_to_host_optimization = 18;</code>
-     */
-    public Builder setPinToHostOptimization(org.tensorflow.proto.framework.RewriterConfig.Toggle value) {
-      if (value == null) {
-        throw new NullPointerException();
-      }
-      
-      pinToHostOptimization_ = value.getNumber();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Force small ops onto the CPU (default is OFF).
-     * </pre>
-     *
-     * <code>.tensorflow.RewriterConfig.Toggle pin_to_host_optimization = 18;</code>
-     */
-    public Builder clearPinToHostOptimization() {
-      
-      pinToHostOptimization_ = 0;
-      onChanged();
-      return this;
-    }
-
-    private int implementationSelector_ = 0;
-    /**
-     * <pre>
-     * Enable the swap of kernel implementations based on the device placement
-     * (default is ON).
-     * </pre>
-     *
-     * <code>.tensorflow.RewriterConfig.Toggle implementation_selector = 22;</code>
-     */
-    public int getImplementationSelectorValue() {
-      return implementationSelector_;
-    }
-    /**
-     * <pre>
-     * Enable the swap of kernel implementations based on the device placement
-     * (default is ON).
-     * </pre>
-     *
-     * <code>.tensorflow.RewriterConfig.Toggle implementation_selector = 22;</code>
-     */
-    public Builder setImplementationSelectorValue(int value) {
-      implementationSelector_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Enable the swap of kernel implementations based on the device placement
-     * (default is ON).
-     * </pre>
-     *
-     * <code>.tensorflow.RewriterConfig.Toggle implementation_selector = 22;</code>
-     */
-    public org.tensorflow.proto.framework.RewriterConfig.Toggle getImplementationSelector() {
-      @SuppressWarnings("deprecation")
-      org.tensorflow.proto.framework.RewriterConfig.Toggle result = org.tensorflow.proto.framework.RewriterConfig.Toggle.valueOf(implementationSelector_);
-      return result == null ? org.tensorflow.proto.framework.RewriterConfig.Toggle.UNRECOGNIZED : result;
-    }
-    /**
-     * <pre>
-     * Enable the swap of kernel implementations based on the device placement
-     * (default is ON).
-     * </pre>
-     *
-     * <code>.tensorflow.RewriterConfig.Toggle implementation_selector = 22;</code>
-     */
-    public Builder setImplementationSelector(org.tensorflow.proto.framework.RewriterConfig.Toggle value) {
-      if (value == null) {
-        throw new NullPointerException();
-      }
-      
-      implementationSelector_ = value.getNumber();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Enable the swap of kernel implementations based on the device placement
-     * (default is ON).
-     * </pre>
-     *
-     * <code>.tensorflow.RewriterConfig.Toggle implementation_selector = 22;</code>
-     */
-    public Builder clearImplementationSelector() {
-      
-      implementationSelector_ = 0;
-      onChanged();
-      return this;
-    }
-
-    private int autoMixedPrecision_ = 0;
-    /**
-     * <pre>
-     * Optimize data types for CUDA (default is OFF).
-     * This will try to use float16 on GPU which is faster.
-     * Note that this can change the numerical stability of the graph and may
-     * require the use of loss scaling to maintain model convergence.
-     * </pre>
-     *
-     * <code>.tensorflow.RewriterConfig.Toggle auto_mixed_precision = 23;</code>
-     */
-    public int getAutoMixedPrecisionValue() {
-      return autoMixedPrecision_;
-    }
-    /**
-     * <pre>
-     * Optimize data types for CUDA (default is OFF).
-     * This will try to use float16 on GPU which is faster.
-     * Note that this can change the numerical stability of the graph and may
-     * require the use of loss scaling to maintain model convergence.
-     * </pre>
-     *
-     * <code>.tensorflow.RewriterConfig.Toggle auto_mixed_precision = 23;</code>
-     */
-    public Builder setAutoMixedPrecisionValue(int value) {
-      autoMixedPrecision_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Optimize data types for CUDA (default is OFF).
-     * This will try to use float16 on GPU which is faster.
-     * Note that this can change the numerical stability of the graph and may
-     * require the use of loss scaling to maintain model convergence.
-     * </pre>
-     *
-     * <code>.tensorflow.RewriterConfig.Toggle auto_mixed_precision = 23;</code>
-     */
-    public org.tensorflow.proto.framework.RewriterConfig.Toggle getAutoMixedPrecision() {
-      @SuppressWarnings("deprecation")
-      org.tensorflow.proto.framework.RewriterConfig.Toggle result = org.tensorflow.proto.framework.RewriterConfig.Toggle.valueOf(autoMixedPrecision_);
-      return result == null ? org.tensorflow.proto.framework.RewriterConfig.Toggle.UNRECOGNIZED : result;
-    }
-    /**
-     * <pre>
-     * Optimize data types for CUDA (default is OFF).
-     * This will try to use float16 on GPU which is faster.
-     * Note that this can change the numerical stability of the graph and may
-     * require the use of loss scaling to maintain model convergence.
-     * </pre>
-     *
-     * <code>.tensorflow.RewriterConfig.Toggle auto_mixed_precision = 23;</code>
-     */
-    public Builder setAutoMixedPrecision(org.tensorflow.proto.framework.RewriterConfig.Toggle value) {
-      if (value == null) {
-        throw new NullPointerException();
-      }
-      
-      autoMixedPrecision_ = value.getNumber();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Optimize data types for CUDA (default is OFF).
-     * This will try to use float16 on GPU which is faster.
-     * Note that this can change the numerical stability of the graph and may
-     * require the use of loss scaling to maintain model convergence.
-     * </pre>
-     *
-     * <code>.tensorflow.RewriterConfig.Toggle auto_mixed_precision = 23;</code>
-     */
-    public Builder clearAutoMixedPrecision() {
-      
-      autoMixedPrecision_ = 0;
-      onChanged();
-      return this;
-    }
-
-    private int autoMixedPrecisionMkl_ = 0;
-    /**
-     * <pre>
-     * Optimize data types for MKL (default is OFF).
-     * This will try to use bfloat16 on CPUs, which is faster.
-     * Note that this can change the numerical stability of the graph.
-     * </pre>
-     *
-     * <code>.tensorflow.RewriterConfig.Toggle auto_mixed_precision_mkl = 25;</code>
-     */
-    public int getAutoMixedPrecisionMklValue() {
-      return autoMixedPrecisionMkl_;
-    }
-    /**
-     * <pre>
-     * Optimize data types for MKL (default is OFF).
-     * This will try to use bfloat16 on CPUs, which is faster.
-     * Note that this can change the numerical stability of the graph.
-     * </pre>
-     *
-     * <code>.tensorflow.RewriterConfig.Toggle auto_mixed_precision_mkl = 25;</code>
-     */
-    public Builder setAutoMixedPrecisionMklValue(int value) {
-      autoMixedPrecisionMkl_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Optimize data types for MKL (default is OFF).
-     * This will try to use bfloat16 on CPUs, which is faster.
-     * Note that this can change the numerical stability of the graph.
-     * </pre>
-     *
-     * <code>.tensorflow.RewriterConfig.Toggle auto_mixed_precision_mkl = 25;</code>
-     */
-    public org.tensorflow.proto.framework.RewriterConfig.Toggle getAutoMixedPrecisionMkl() {
-      @SuppressWarnings("deprecation")
-      org.tensorflow.proto.framework.RewriterConfig.Toggle result = org.tensorflow.proto.framework.RewriterConfig.Toggle.valueOf(autoMixedPrecisionMkl_);
-      return result == null ? org.tensorflow.proto.framework.RewriterConfig.Toggle.UNRECOGNIZED : result;
-    }
-    /**
-     * <pre>
-     * Optimize data types for MKL (default is OFF).
-     * This will try to use bfloat16 on CPUs, which is faster.
-     * Note that this can change the numerical stability of the graph.
-     * </pre>
-     *
-     * <code>.tensorflow.RewriterConfig.Toggle auto_mixed_precision_mkl = 25;</code>
-     */
-    public Builder setAutoMixedPrecisionMkl(org.tensorflow.proto.framework.RewriterConfig.Toggle value) {
-      if (value == null) {
-        throw new NullPointerException();
-      }
-      
-      autoMixedPrecisionMkl_ = value.getNumber();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Optimize data types for MKL (default is OFF).
-     * This will try to use bfloat16 on CPUs, which is faster.
-     * Note that this can change the numerical stability of the graph.
-     * </pre>
-     *
-     * <code>.tensorflow.RewriterConfig.Toggle auto_mixed_precision_mkl = 25;</code>
-     */
-    public Builder clearAutoMixedPrecisionMkl() {
-      
-      autoMixedPrecisionMkl_ = 0;
-      onChanged();
-      return this;
-    }
-
-    private boolean disableMetaOptimizer_ ;
-    /**
-     * <pre>
-     * Disable the entire meta optimizer (off by default).
-     * </pre>
-     *
-     * <code>bool disable_meta_optimizer = 19;</code>
-     */
-    public boolean getDisableMetaOptimizer() {
-      return disableMetaOptimizer_;
-    }
-    /**
-     * <pre>
-     * Disable the entire meta optimizer (off by default).
-     * </pre>
-     *
-     * <code>bool disable_meta_optimizer = 19;</code>
-     */
-    public Builder setDisableMetaOptimizer(boolean value) {
-      
-      disableMetaOptimizer_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Disable the entire meta optimizer (off by default).
-     * </pre>
-     *
-     * <code>bool disable_meta_optimizer = 19;</code>
-     */
-    public Builder clearDisableMetaOptimizer() {
-      
-      disableMetaOptimizer_ = false;
-      onChanged();
-      return this;
-    }
-
-    private int metaOptimizerIterations_ = 0;
-    /**
-     * <pre>
-     * Controls how many times we run the optimizers in meta optimizer (default
-     * is once).
-     * </pre>
-     *
-     * <code>.tensorflow.RewriterConfig.NumIterationsType meta_optimizer_iterations = 12;</code>
-     */
-    public int getMetaOptimizerIterationsValue() {
-      return metaOptimizerIterations_;
-    }
-    /**
-     * <pre>
-     * Controls how many times we run the optimizers in meta optimizer (default
-     * is once).
-     * </pre>
-     *
-     * <code>.tensorflow.RewriterConfig.NumIterationsType meta_optimizer_iterations = 12;</code>
-     */
-    public Builder setMetaOptimizerIterationsValue(int value) {
-      metaOptimizerIterations_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Controls how many times we run the optimizers in meta optimizer (default
-     * is once).
-     * </pre>
-     *
-     * <code>.tensorflow.RewriterConfig.NumIterationsType meta_optimizer_iterations = 12;</code>
-     */
-    public org.tensorflow.proto.framework.RewriterConfig.NumIterationsType getMetaOptimizerIterations() {
-      @SuppressWarnings("deprecation")
-      org.tensorflow.proto.framework.RewriterConfig.NumIterationsType result = org.tensorflow.proto.framework.RewriterConfig.NumIterationsType.valueOf(metaOptimizerIterations_);
-      return result == null ? org.tensorflow.proto.framework.RewriterConfig.NumIterationsType.UNRECOGNIZED : result;
-    }
-    /**
-     * <pre>
-     * Controls how many times we run the optimizers in meta optimizer (default
-     * is once).
-     * </pre>
-     *
-     * <code>.tensorflow.RewriterConfig.NumIterationsType meta_optimizer_iterations = 12;</code>
-     */
-    public Builder setMetaOptimizerIterations(org.tensorflow.proto.framework.RewriterConfig.NumIterationsType value) {
-      if (value == null) {
-        throw new NullPointerException();
-      }
-      
-      metaOptimizerIterations_ = value.getNumber();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Controls how many times we run the optimizers in meta optimizer (default
-     * is once).
-     * </pre>
-     *
-     * <code>.tensorflow.RewriterConfig.NumIterationsType meta_optimizer_iterations = 12;</code>
-     */
-    public Builder clearMetaOptimizerIterations() {
-      
-      metaOptimizerIterations_ = 0;
-      onChanged();
-      return this;
-    }
-
-    private int minGraphNodes_ ;
-    /**
-     * <pre>
-     * The minimum number of nodes in a graph to optimizer. For smaller graphs,
-     * optimization is skipped.
-     * 0 means the system picks an appropriate number.
-     * &lt; 0 means do not skip optimization.
-     * </pre>
-     *
-     * <code>int32 min_graph_nodes = 17;</code>
-     */
-    public int getMinGraphNodes() {
-      return minGraphNodes_;
-    }
-    /**
-     * <pre>
-     * The minimum number of nodes in a graph to optimizer. For smaller graphs,
-     * optimization is skipped.
-     * 0 means the system picks an appropriate number.
-     * &lt; 0 means do not skip optimization.
-     * </pre>
-     *
-     * <code>int32 min_graph_nodes = 17;</code>
-     */
-    public Builder setMinGraphNodes(int value) {
-      
-      minGraphNodes_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * The minimum number of nodes in a graph to optimizer. For smaller graphs,
-     * optimization is skipped.
-     * 0 means the system picks an appropriate number.
-     * &lt; 0 means do not skip optimization.
-     * </pre>
-     *
-     * <code>int32 min_graph_nodes = 17;</code>
-     */
-    public Builder clearMinGraphNodes() {
-      
-      minGraphNodes_ = 0;
-      onChanged();
-      return this;
-    }
-
-    private int memoryOptimization_ = 0;
-    /**
-     * <pre>
-     * Configures memory optimization passes through the meta-optimizer. Has no
-     * effect on manually requested memory optimization passes in the optimizers
-     * field.
-     * </pre>
-     *
-     * <code>.tensorflow.RewriterConfig.MemOptType memory_optimization = 4;</code>
-     */
-    public int getMemoryOptimizationValue() {
-      return memoryOptimization_;
-    }
-    /**
-     * <pre>
-     * Configures memory optimization passes through the meta-optimizer. Has no
-     * effect on manually requested memory optimization passes in the optimizers
-     * field.
-     * </pre>
-     *
-     * <code>.tensorflow.RewriterConfig.MemOptType memory_optimization = 4;</code>
-     */
-    public Builder setMemoryOptimizationValue(int value) {
-      memoryOptimization_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Configures memory optimization passes through the meta-optimizer. Has no
-     * effect on manually requested memory optimization passes in the optimizers
-     * field.
-     * </pre>
-     *
-     * <code>.tensorflow.RewriterConfig.MemOptType memory_optimization = 4;</code>
-     */
-    public org.tensorflow.proto.framework.RewriterConfig.MemOptType getMemoryOptimization() {
-      @SuppressWarnings("deprecation")
-      org.tensorflow.proto.framework.RewriterConfig.MemOptType result = org.tensorflow.proto.framework.RewriterConfig.MemOptType.valueOf(memoryOptimization_);
-      return result == null ? org.tensorflow.proto.framework.RewriterConfig.MemOptType.UNRECOGNIZED : result;
-    }
-    /**
-     * <pre>
-     * Configures memory optimization passes through the meta-optimizer. Has no
-     * effect on manually requested memory optimization passes in the optimizers
-     * field.
-     * </pre>
-     *
-     * <code>.tensorflow.RewriterConfig.MemOptType memory_optimization = 4;</code>
-     */
-    public Builder setMemoryOptimization(org.tensorflow.proto.framework.RewriterConfig.MemOptType value) {
-      if (value == null) {
-        throw new NullPointerException();
-      }
-      
-      memoryOptimization_ = value.getNumber();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Configures memory optimization passes through the meta-optimizer. Has no
-     * effect on manually requested memory optimization passes in the optimizers
-     * field.
-     * </pre>
-     *
-     * <code>.tensorflow.RewriterConfig.MemOptType memory_optimization = 4;</code>
-     */
-    public Builder clearMemoryOptimization() {
-      
-      memoryOptimization_ = 0;
-      onChanged();
-      return this;
-    }
-
-    private java.lang.Object memoryOptimizerTargetNodeNameScope_ = "";
-    /**
-     * <pre>
-     * A node name scope for node names which are valid outputs of recomputations.
-     * Inputs to nodes that match this scope may be recomputed (subject either to
-     * manual annotation of those input nodes or to manual annotation and
-     * heuristics depending on memory_optimization), but the nodes themselves will
-     * not be recomputed. This matches any sub-scopes as well, meaning the scope
-     * can appear not just as a top-level scope. For example, if the value is
-     * "gradients/", the default, it will match node name "gradients/foo",
-     * "foo/gradients/bar", but not "foo_gradients/"
-     * </pre>
-     *
-     * <code>string memory_optimizer_target_node_name_scope = 6;</code>
-     */
-    public java.lang.String getMemoryOptimizerTargetNodeNameScope() {
-      java.lang.Object ref = memoryOptimizerTargetNodeNameScope_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        memoryOptimizerTargetNodeNameScope_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * A node name scope for node names which are valid outputs of recomputations.
-     * Inputs to nodes that match this scope may be recomputed (subject either to
-     * manual annotation of those input nodes or to manual annotation and
-     * heuristics depending on memory_optimization), but the nodes themselves will
-     * not be recomputed. This matches any sub-scopes as well, meaning the scope
-     * can appear not just as a top-level scope. For example, if the value is
-     * "gradients/", the default, it will match node name "gradients/foo",
-     * "foo/gradients/bar", but not "foo_gradients/"
-     * </pre>
-     *
-     * <code>string memory_optimizer_target_node_name_scope = 6;</code>
-     */
-    public com.google.protobuf.ByteString
-        getMemoryOptimizerTargetNodeNameScopeBytes() {
-      java.lang.Object ref = memoryOptimizerTargetNodeNameScope_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        memoryOptimizerTargetNodeNameScope_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * A node name scope for node names which are valid outputs of recomputations.
-     * Inputs to nodes that match this scope may be recomputed (subject either to
-     * manual annotation of those input nodes or to manual annotation and
-     * heuristics depending on memory_optimization), but the nodes themselves will
-     * not be recomputed. This matches any sub-scopes as well, meaning the scope
-     * can appear not just as a top-level scope. For example, if the value is
-     * "gradients/", the default, it will match node name "gradients/foo",
-     * "foo/gradients/bar", but not "foo_gradients/"
-     * </pre>
-     *
-     * <code>string memory_optimizer_target_node_name_scope = 6;</code>
-     */
-    public Builder setMemoryOptimizerTargetNodeNameScope(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      memoryOptimizerTargetNodeNameScope_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * A node name scope for node names which are valid outputs of recomputations.
-     * Inputs to nodes that match this scope may be recomputed (subject either to
-     * manual annotation of those input nodes or to manual annotation and
-     * heuristics depending on memory_optimization), but the nodes themselves will
-     * not be recomputed. This matches any sub-scopes as well, meaning the scope
-     * can appear not just as a top-level scope. For example, if the value is
-     * "gradients/", the default, it will match node name "gradients/foo",
-     * "foo/gradients/bar", but not "foo_gradients/"
-     * </pre>
-     *
-     * <code>string memory_optimizer_target_node_name_scope = 6;</code>
-     */
-    public Builder clearMemoryOptimizerTargetNodeNameScope() {
-      
-      memoryOptimizerTargetNodeNameScope_ = getDefaultInstance().getMemoryOptimizerTargetNodeNameScope();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * A node name scope for node names which are valid outputs of recomputations.
-     * Inputs to nodes that match this scope may be recomputed (subject either to
-     * manual annotation of those input nodes or to manual annotation and
-     * heuristics depending on memory_optimization), but the nodes themselves will
-     * not be recomputed. This matches any sub-scopes as well, meaning the scope
-     * can appear not just as a top-level scope. For example, if the value is
-     * "gradients/", the default, it will match node name "gradients/foo",
-     * "foo/gradients/bar", but not "foo_gradients/"
-     * </pre>
-     *
-     * <code>string memory_optimizer_target_node_name_scope = 6;</code>
-     */
-    public Builder setMemoryOptimizerTargetNodeNameScopeBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      memoryOptimizerTargetNodeNameScope_ = value;
-      onChanged();
-      return this;
-    }
-
-    private long metaOptimizerTimeoutMs_ ;
-    /**
-     * <pre>
-     * Maximum number of milliseconds to spend optimizing a single graph before
-     * timing out. If equal to 0 the system picks a default (currently 5 minutes).
-     * If less than 0 the optimizer will never time out.
-     * </pre>
-     *
-     * <code>int64 meta_optimizer_timeout_ms = 20;</code>
-     */
-    public long getMetaOptimizerTimeoutMs() {
-      return metaOptimizerTimeoutMs_;
-    }
-    /**
-     * <pre>
-     * Maximum number of milliseconds to spend optimizing a single graph before
-     * timing out. If equal to 0 the system picks a default (currently 5 minutes).
-     * If less than 0 the optimizer will never time out.
-     * </pre>
-     *
-     * <code>int64 meta_optimizer_timeout_ms = 20;</code>
-     */
-    public Builder setMetaOptimizerTimeoutMs(long value) {
-      
-      metaOptimizerTimeoutMs_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Maximum number of milliseconds to spend optimizing a single graph before
-     * timing out. If equal to 0 the system picks a default (currently 5 minutes).
-     * If less than 0 the optimizer will never time out.
-     * </pre>
-     *
-     * <code>int64 meta_optimizer_timeout_ms = 20;</code>
-     */
-    public Builder clearMetaOptimizerTimeoutMs() {
-      
-      metaOptimizerTimeoutMs_ = 0L;
-      onChanged();
-      return this;
-    }
-
-    private org.tensorflow.proto.framework.AutoParallelOptions autoParallel_;
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.AutoParallelOptions, org.tensorflow.proto.framework.AutoParallelOptions.Builder, org.tensorflow.proto.framework.AutoParallelOptionsOrBuilder> autoParallelBuilder_;
-    /**
-     * <pre>
-     * Configures AutoParallel optimization passes either through the
-     * meta-optimizer or when manually specified through the optimizers field.
-     * </pre>
-     *
-     * <code>.tensorflow.AutoParallelOptions auto_parallel = 5;</code>
-     */
-    public boolean hasAutoParallel() {
-      return autoParallelBuilder_ != null || autoParallel_ != null;
-    }
-    /**
-     * <pre>
-     * Configures AutoParallel optimization passes either through the
-     * meta-optimizer or when manually specified through the optimizers field.
-     * </pre>
-     *
-     * <code>.tensorflow.AutoParallelOptions auto_parallel = 5;</code>
-     */
-    public org.tensorflow.proto.framework.AutoParallelOptions getAutoParallel() {
-      if (autoParallelBuilder_ == null) {
-        return autoParallel_ == null ? org.tensorflow.proto.framework.AutoParallelOptions.getDefaultInstance() : autoParallel_;
-      } else {
-        return autoParallelBuilder_.getMessage();
-      }
-    }
-    /**
-     * <pre>
-     * Configures AutoParallel optimization passes either through the
-     * meta-optimizer or when manually specified through the optimizers field.
-     * </pre>
-     *
-     * <code>.tensorflow.AutoParallelOptions auto_parallel = 5;</code>
-     */
-    public Builder setAutoParallel(org.tensorflow.proto.framework.AutoParallelOptions value) {
-      if (autoParallelBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        autoParallel_ = value;
-        onChanged();
-      } else {
-        autoParallelBuilder_.setMessage(value);
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Configures AutoParallel optimization passes either through the
-     * meta-optimizer or when manually specified through the optimizers field.
-     * </pre>
-     *
-     * <code>.tensorflow.AutoParallelOptions auto_parallel = 5;</code>
-     */
-    public Builder setAutoParallel(
-        org.tensorflow.proto.framework.AutoParallelOptions.Builder builderForValue) {
-      if (autoParallelBuilder_ == null) {
-        autoParallel_ = builderForValue.build();
-        onChanged();
-      } else {
-        autoParallelBuilder_.setMessage(builderForValue.build());
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Configures AutoParallel optimization passes either through the
-     * meta-optimizer or when manually specified through the optimizers field.
-     * </pre>
-     *
-     * <code>.tensorflow.AutoParallelOptions auto_parallel = 5;</code>
-     */
-    public Builder mergeAutoParallel(org.tensorflow.proto.framework.AutoParallelOptions value) {
-      if (autoParallelBuilder_ == null) {
-        if (autoParallel_ != null) {
-          autoParallel_ =
-            org.tensorflow.proto.framework.AutoParallelOptions.newBuilder(autoParallel_).mergeFrom(value).buildPartial();
-        } else {
-          autoParallel_ = value;
-        }
-        onChanged();
-      } else {
-        autoParallelBuilder_.mergeFrom(value);
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Configures AutoParallel optimization passes either through the
-     * meta-optimizer or when manually specified through the optimizers field.
-     * </pre>
-     *
-     * <code>.tensorflow.AutoParallelOptions auto_parallel = 5;</code>
-     */
-    public Builder clearAutoParallel() {
-      if (autoParallelBuilder_ == null) {
-        autoParallel_ = null;
-        onChanged();
-      } else {
-        autoParallel_ = null;
-        autoParallelBuilder_ = null;
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Configures AutoParallel optimization passes either through the
-     * meta-optimizer or when manually specified through the optimizers field.
-     * </pre>
-     *
-     * <code>.tensorflow.AutoParallelOptions auto_parallel = 5;</code>
-     */
-    public org.tensorflow.proto.framework.AutoParallelOptions.Builder getAutoParallelBuilder() {
-      
-      onChanged();
-      return getAutoParallelFieldBuilder().getBuilder();
-    }
-    /**
-     * <pre>
-     * Configures AutoParallel optimization passes either through the
-     * meta-optimizer or when manually specified through the optimizers field.
-     * </pre>
-     *
-     * <code>.tensorflow.AutoParallelOptions auto_parallel = 5;</code>
-     */
-    public org.tensorflow.proto.framework.AutoParallelOptionsOrBuilder getAutoParallelOrBuilder() {
-      if (autoParallelBuilder_ != null) {
-        return autoParallelBuilder_.getMessageOrBuilder();
-      } else {
-        return autoParallel_ == null ?
-            org.tensorflow.proto.framework.AutoParallelOptions.getDefaultInstance() : autoParallel_;
-      }
-    }
-    /**
-     * <pre>
-     * Configures AutoParallel optimization passes either through the
-     * meta-optimizer or when manually specified through the optimizers field.
-     * </pre>
-     *
-     * <code>.tensorflow.AutoParallelOptions auto_parallel = 5;</code>
-     */
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.AutoParallelOptions, org.tensorflow.proto.framework.AutoParallelOptions.Builder, org.tensorflow.proto.framework.AutoParallelOptionsOrBuilder> 
-        getAutoParallelFieldBuilder() {
-      if (autoParallelBuilder_ == null) {
-        autoParallelBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-            org.tensorflow.proto.framework.AutoParallelOptions, org.tensorflow.proto.framework.AutoParallelOptions.Builder, org.tensorflow.proto.framework.AutoParallelOptionsOrBuilder>(
-                getAutoParallel(),
-                getParentForChildren(),
-                isClean());
-        autoParallel_ = null;
-      }
-      return autoParallelBuilder_;
-    }
-
-    private boolean failOnOptimizerErrors_ ;
-    /**
-     * <pre>
-     * If true, any optimization pass failing will cause the MetaOptimizer to
-     * stop with an error. By default - or when set to false, failing passes are
-     * skipped silently.
-     * </pre>
-     *
-     * <code>bool fail_on_optimizer_errors = 21;</code>
-     */
-    public boolean getFailOnOptimizerErrors() {
-      return failOnOptimizerErrors_;
-    }
-    /**
-     * <pre>
-     * If true, any optimization pass failing will cause the MetaOptimizer to
-     * stop with an error. By default - or when set to false, failing passes are
-     * skipped silently.
-     * </pre>
-     *
-     * <code>bool fail_on_optimizer_errors = 21;</code>
-     */
-    public Builder setFailOnOptimizerErrors(boolean value) {
-      
-      failOnOptimizerErrors_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * If true, any optimization pass failing will cause the MetaOptimizer to
-     * stop with an error. By default - or when set to false, failing passes are
-     * skipped silently.
-     * </pre>
-     *
-     * <code>bool fail_on_optimizer_errors = 21;</code>
-     */
-    public Builder clearFailOnOptimizerErrors() {
-      
-      failOnOptimizerErrors_ = false;
-      onChanged();
-      return this;
-    }
-
-    private org.tensorflow.proto.framework.ScopedAllocatorOptions scopedAllocatorOpts_;
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.ScopedAllocatorOptions, org.tensorflow.proto.framework.ScopedAllocatorOptions.Builder, org.tensorflow.proto.framework.ScopedAllocatorOptionsOrBuilder> scopedAllocatorOptsBuilder_;
-    /**
-     * <code>.tensorflow.ScopedAllocatorOptions scoped_allocator_opts = 16;</code>
-     */
-    public boolean hasScopedAllocatorOpts() {
-      return scopedAllocatorOptsBuilder_ != null || scopedAllocatorOpts_ != null;
-    }
-    /**
-     * <code>.tensorflow.ScopedAllocatorOptions scoped_allocator_opts = 16;</code>
-     */
-    public org.tensorflow.proto.framework.ScopedAllocatorOptions getScopedAllocatorOpts() {
-      if (scopedAllocatorOptsBuilder_ == null) {
-        return scopedAllocatorOpts_ == null ? org.tensorflow.proto.framework.ScopedAllocatorOptions.getDefaultInstance() : scopedAllocatorOpts_;
-      } else {
-        return scopedAllocatorOptsBuilder_.getMessage();
-      }
-    }
-    /**
-     * <code>.tensorflow.ScopedAllocatorOptions scoped_allocator_opts = 16;</code>
-     */
-    public Builder setScopedAllocatorOpts(org.tensorflow.proto.framework.ScopedAllocatorOptions value) {
-      if (scopedAllocatorOptsBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        scopedAllocatorOpts_ = value;
-        onChanged();
-      } else {
-        scopedAllocatorOptsBuilder_.setMessage(value);
-      }
-
-      return this;
-    }
-    /**
-     * <code>.tensorflow.ScopedAllocatorOptions scoped_allocator_opts = 16;</code>
-     */
-    public Builder setScopedAllocatorOpts(
-        org.tensorflow.proto.framework.ScopedAllocatorOptions.Builder builderForValue) {
-      if (scopedAllocatorOptsBuilder_ == null) {
-        scopedAllocatorOpts_ = builderForValue.build();
-        onChanged();
-      } else {
-        scopedAllocatorOptsBuilder_.setMessage(builderForValue.build());
-      }
-
-      return this;
-    }
-    /**
-     * <code>.tensorflow.ScopedAllocatorOptions scoped_allocator_opts = 16;</code>
-     */
-    public Builder mergeScopedAllocatorOpts(org.tensorflow.proto.framework.ScopedAllocatorOptions value) {
-      if (scopedAllocatorOptsBuilder_ == null) {
-        if (scopedAllocatorOpts_ != null) {
-          scopedAllocatorOpts_ =
-            org.tensorflow.proto.framework.ScopedAllocatorOptions.newBuilder(scopedAllocatorOpts_).mergeFrom(value).buildPartial();
-        } else {
-          scopedAllocatorOpts_ = value;
-        }
-        onChanged();
-      } else {
-        scopedAllocatorOptsBuilder_.mergeFrom(value);
-      }
-
-      return this;
-    }
-    /**
-     * <code>.tensorflow.ScopedAllocatorOptions scoped_allocator_opts = 16;</code>
-     */
-    public Builder clearScopedAllocatorOpts() {
-      if (scopedAllocatorOptsBuilder_ == null) {
-        scopedAllocatorOpts_ = null;
-        onChanged();
-      } else {
-        scopedAllocatorOpts_ = null;
-        scopedAllocatorOptsBuilder_ = null;
-      }
-
-      return this;
-    }
-    /**
-     * <code>.tensorflow.ScopedAllocatorOptions scoped_allocator_opts = 16;</code>
-     */
-    public org.tensorflow.proto.framework.ScopedAllocatorOptions.Builder getScopedAllocatorOptsBuilder() {
-      
-      onChanged();
-      return getScopedAllocatorOptsFieldBuilder().getBuilder();
-    }
-    /**
-     * <code>.tensorflow.ScopedAllocatorOptions scoped_allocator_opts = 16;</code>
-     */
-    public org.tensorflow.proto.framework.ScopedAllocatorOptionsOrBuilder getScopedAllocatorOptsOrBuilder() {
-      if (scopedAllocatorOptsBuilder_ != null) {
-        return scopedAllocatorOptsBuilder_.getMessageOrBuilder();
-      } else {
-        return scopedAllocatorOpts_ == null ?
-            org.tensorflow.proto.framework.ScopedAllocatorOptions.getDefaultInstance() : scopedAllocatorOpts_;
-      }
-    }
-    /**
-     * <code>.tensorflow.ScopedAllocatorOptions scoped_allocator_opts = 16;</code>
-     */
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.ScopedAllocatorOptions, org.tensorflow.proto.framework.ScopedAllocatorOptions.Builder, org.tensorflow.proto.framework.ScopedAllocatorOptionsOrBuilder> 
-        getScopedAllocatorOptsFieldBuilder() {
-      if (scopedAllocatorOptsBuilder_ == null) {
-        scopedAllocatorOptsBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-            org.tensorflow.proto.framework.ScopedAllocatorOptions, org.tensorflow.proto.framework.ScopedAllocatorOptions.Builder, org.tensorflow.proto.framework.ScopedAllocatorOptionsOrBuilder>(
-                getScopedAllocatorOpts(),
-                getParentForChildren(),
-                isClean());
-        scopedAllocatorOpts_ = null;
-      }
-      return scopedAllocatorOptsBuilder_;
-    }
-
-    private com.google.protobuf.LazyStringList optimizers_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-    private void ensureOptimizersIsMutable() {
-      if (!((bitField0_ & 0x00000001) != 0)) {
-        optimizers_ = new com.google.protobuf.LazyStringArrayList(optimizers_);
-        bitField0_ |= 0x00000001;
-       }
-    }
-    /**
-     * <pre>
-     * If non-empty, will use this as an alternative way to specify a list of
-     * optimizations to turn on and the order of the optimizations (replacing the
-     * meta-optimizer).
-     * Of the RewriterConfig options, only the AutoParallel configuration options
-     * (the auto_parallel field) apply to manually requested optimization passes
-     * ("autoparallel"). Memory optimization passes ("memory") invoked here are
-     * not configurable (in contrast to memory optimization passes through the
-     * meta-optimizer) and act only on manual op annotations.
-     * Custom optimizers (see custom_optimizers) that are not part of this
-     * schedule will be run after - in the order that they were specified.
-     * </pre>
-     *
-     * <code>repeated string optimizers = 100;</code>
-     */
-    public com.google.protobuf.ProtocolStringList
-        getOptimizersList() {
-      return optimizers_.getUnmodifiableView();
-    }
-    /**
-     * <pre>
-     * If non-empty, will use this as an alternative way to specify a list of
-     * optimizations to turn on and the order of the optimizations (replacing the
-     * meta-optimizer).
-     * Of the RewriterConfig options, only the AutoParallel configuration options
-     * (the auto_parallel field) apply to manually requested optimization passes
-     * ("autoparallel"). Memory optimization passes ("memory") invoked here are
-     * not configurable (in contrast to memory optimization passes through the
-     * meta-optimizer) and act only on manual op annotations.
-     * Custom optimizers (see custom_optimizers) that are not part of this
-     * schedule will be run after - in the order that they were specified.
-     * </pre>
-     *
-     * <code>repeated string optimizers = 100;</code>
-     */
-    public int getOptimizersCount() {
-      return optimizers_.size();
-    }
-    /**
-     * <pre>
-     * If non-empty, will use this as an alternative way to specify a list of
-     * optimizations to turn on and the order of the optimizations (replacing the
-     * meta-optimizer).
-     * Of the RewriterConfig options, only the AutoParallel configuration options
-     * (the auto_parallel field) apply to manually requested optimization passes
-     * ("autoparallel"). Memory optimization passes ("memory") invoked here are
-     * not configurable (in contrast to memory optimization passes through the
-     * meta-optimizer) and act only on manual op annotations.
-     * Custom optimizers (see custom_optimizers) that are not part of this
-     * schedule will be run after - in the order that they were specified.
-     * </pre>
-     *
-     * <code>repeated string optimizers = 100;</code>
-     */
-    public java.lang.String getOptimizers(int index) {
-      return optimizers_.get(index);
-    }
-    /**
-     * <pre>
-     * If non-empty, will use this as an alternative way to specify a list of
-     * optimizations to turn on and the order of the optimizations (replacing the
-     * meta-optimizer).
-     * Of the RewriterConfig options, only the AutoParallel configuration options
-     * (the auto_parallel field) apply to manually requested optimization passes
-     * ("autoparallel"). Memory optimization passes ("memory") invoked here are
-     * not configurable (in contrast to memory optimization passes through the
-     * meta-optimizer) and act only on manual op annotations.
-     * Custom optimizers (see custom_optimizers) that are not part of this
-     * schedule will be run after - in the order that they were specified.
-     * </pre>
-     *
-     * <code>repeated string optimizers = 100;</code>
-     */
-    public com.google.protobuf.ByteString
-        getOptimizersBytes(int index) {
-      return optimizers_.getByteString(index);
-    }
-    /**
-     * <pre>
-     * If non-empty, will use this as an alternative way to specify a list of
-     * optimizations to turn on and the order of the optimizations (replacing the
-     * meta-optimizer).
-     * Of the RewriterConfig options, only the AutoParallel configuration options
-     * (the auto_parallel field) apply to manually requested optimization passes
-     * ("autoparallel"). Memory optimization passes ("memory") invoked here are
-     * not configurable (in contrast to memory optimization passes through the
-     * meta-optimizer) and act only on manual op annotations.
-     * Custom optimizers (see custom_optimizers) that are not part of this
-     * schedule will be run after - in the order that they were specified.
-     * </pre>
-     *
-     * <code>repeated string optimizers = 100;</code>
-     */
-    public Builder setOptimizers(
-        int index, java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  ensureOptimizersIsMutable();
-      optimizers_.set(index, value);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * If non-empty, will use this as an alternative way to specify a list of
-     * optimizations to turn on and the order of the optimizations (replacing the
-     * meta-optimizer).
-     * Of the RewriterConfig options, only the AutoParallel configuration options
-     * (the auto_parallel field) apply to manually requested optimization passes
-     * ("autoparallel"). Memory optimization passes ("memory") invoked here are
-     * not configurable (in contrast to memory optimization passes through the
-     * meta-optimizer) and act only on manual op annotations.
-     * Custom optimizers (see custom_optimizers) that are not part of this
-     * schedule will be run after - in the order that they were specified.
-     * </pre>
-     *
-     * <code>repeated string optimizers = 100;</code>
-     */
-    public Builder addOptimizers(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  ensureOptimizersIsMutable();
-      optimizers_.add(value);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * If non-empty, will use this as an alternative way to specify a list of
-     * optimizations to turn on and the order of the optimizations (replacing the
-     * meta-optimizer).
-     * Of the RewriterConfig options, only the AutoParallel configuration options
-     * (the auto_parallel field) apply to manually requested optimization passes
-     * ("autoparallel"). Memory optimization passes ("memory") invoked here are
-     * not configurable (in contrast to memory optimization passes through the
-     * meta-optimizer) and act only on manual op annotations.
-     * Custom optimizers (see custom_optimizers) that are not part of this
-     * schedule will be run after - in the order that they were specified.
-     * </pre>
-     *
-     * <code>repeated string optimizers = 100;</code>
-     */
-    public Builder addAllOptimizers(
-        java.lang.Iterable<java.lang.String> values) {
-      ensureOptimizersIsMutable();
-      com.google.protobuf.AbstractMessageLite.Builder.addAll(
-          values, optimizers_);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * If non-empty, will use this as an alternative way to specify a list of
-     * optimizations to turn on and the order of the optimizations (replacing the
-     * meta-optimizer).
-     * Of the RewriterConfig options, only the AutoParallel configuration options
-     * (the auto_parallel field) apply to manually requested optimization passes
-     * ("autoparallel"). Memory optimization passes ("memory") invoked here are
-     * not configurable (in contrast to memory optimization passes through the
-     * meta-optimizer) and act only on manual op annotations.
-     * Custom optimizers (see custom_optimizers) that are not part of this
-     * schedule will be run after - in the order that they were specified.
-     * </pre>
-     *
-     * <code>repeated string optimizers = 100;</code>
-     */
-    public Builder clearOptimizers() {
-      optimizers_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-      bitField0_ = (bitField0_ & ~0x00000001);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * If non-empty, will use this as an alternative way to specify a list of
-     * optimizations to turn on and the order of the optimizations (replacing the
-     * meta-optimizer).
-     * Of the RewriterConfig options, only the AutoParallel configuration options
-     * (the auto_parallel field) apply to manually requested optimization passes
-     * ("autoparallel"). Memory optimization passes ("memory") invoked here are
-     * not configurable (in contrast to memory optimization passes through the
-     * meta-optimizer) and act only on manual op annotations.
-     * Custom optimizers (see custom_optimizers) that are not part of this
-     * schedule will be run after - in the order that they were specified.
-     * </pre>
-     *
-     * <code>repeated string optimizers = 100;</code>
-     */
-    public Builder addOptimizersBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      ensureOptimizersIsMutable();
-      optimizers_.add(value);
-      onChanged();
-      return this;
-    }
-
-    private java.util.List<org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizer> customOptimizers_ =
-      java.util.Collections.emptyList();
-    private void ensureCustomOptimizersIsMutable() {
-      if (!((bitField0_ & 0x00000002) != 0)) {
-        customOptimizers_ = new java.util.ArrayList<org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizer>(customOptimizers_);
-        bitField0_ |= 0x00000002;
-       }
-    }
-
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizer, org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizer.Builder, org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizerOrBuilder> customOptimizersBuilder_;
-
-    /**
-     * <pre>
-     * list of CustomGraphOptimizers to apply.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.RewriterConfig.CustomGraphOptimizer custom_optimizers = 200;</code>
-     */
-    public java.util.List<org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizer> getCustomOptimizersList() {
-      if (customOptimizersBuilder_ == null) {
-        return java.util.Collections.unmodifiableList(customOptimizers_);
-      } else {
-        return customOptimizersBuilder_.getMessageList();
-      }
-    }
-    /**
-     * <pre>
-     * list of CustomGraphOptimizers to apply.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.RewriterConfig.CustomGraphOptimizer custom_optimizers = 200;</code>
-     */
-    public int getCustomOptimizersCount() {
-      if (customOptimizersBuilder_ == null) {
-        return customOptimizers_.size();
-      } else {
-        return customOptimizersBuilder_.getCount();
-      }
-    }
-    /**
-     * <pre>
-     * list of CustomGraphOptimizers to apply.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.RewriterConfig.CustomGraphOptimizer custom_optimizers = 200;</code>
-     */
-    public org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizer getCustomOptimizers(int index) {
-      if (customOptimizersBuilder_ == null) {
-        return customOptimizers_.get(index);
-      } else {
-        return customOptimizersBuilder_.getMessage(index);
-      }
-    }
-    /**
-     * <pre>
-     * list of CustomGraphOptimizers to apply.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.RewriterConfig.CustomGraphOptimizer custom_optimizers = 200;</code>
-     */
-    public Builder setCustomOptimizers(
-        int index, org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizer value) {
-      if (customOptimizersBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureCustomOptimizersIsMutable();
-        customOptimizers_.set(index, value);
-        onChanged();
-      } else {
-        customOptimizersBuilder_.setMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * list of CustomGraphOptimizers to apply.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.RewriterConfig.CustomGraphOptimizer custom_optimizers = 200;</code>
-     */
-    public Builder setCustomOptimizers(
-        int index, org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizer.Builder builderForValue) {
-      if (customOptimizersBuilder_ == null) {
-        ensureCustomOptimizersIsMutable();
-        customOptimizers_.set(index, builderForValue.build());
-        onChanged();
-      } else {
-        customOptimizersBuilder_.setMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * list of CustomGraphOptimizers to apply.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.RewriterConfig.CustomGraphOptimizer custom_optimizers = 200;</code>
-     */
-    public Builder addCustomOptimizers(org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizer value) {
-      if (customOptimizersBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureCustomOptimizersIsMutable();
-        customOptimizers_.add(value);
-        onChanged();
-      } else {
-        customOptimizersBuilder_.addMessage(value);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * list of CustomGraphOptimizers to apply.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.RewriterConfig.CustomGraphOptimizer custom_optimizers = 200;</code>
-     */
-    public Builder addCustomOptimizers(
-        int index, org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizer value) {
-      if (customOptimizersBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureCustomOptimizersIsMutable();
-        customOptimizers_.add(index, value);
-        onChanged();
-      } else {
-        customOptimizersBuilder_.addMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * list of CustomGraphOptimizers to apply.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.RewriterConfig.CustomGraphOptimizer custom_optimizers = 200;</code>
-     */
-    public Builder addCustomOptimizers(
-        org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizer.Builder builderForValue) {
-      if (customOptimizersBuilder_ == null) {
-        ensureCustomOptimizersIsMutable();
-        customOptimizers_.add(builderForValue.build());
-        onChanged();
-      } else {
-        customOptimizersBuilder_.addMessage(builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * list of CustomGraphOptimizers to apply.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.RewriterConfig.CustomGraphOptimizer custom_optimizers = 200;</code>
-     */
-    public Builder addCustomOptimizers(
-        int index, org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizer.Builder builderForValue) {
-      if (customOptimizersBuilder_ == null) {
-        ensureCustomOptimizersIsMutable();
-        customOptimizers_.add(index, builderForValue.build());
-        onChanged();
-      } else {
-        customOptimizersBuilder_.addMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * list of CustomGraphOptimizers to apply.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.RewriterConfig.CustomGraphOptimizer custom_optimizers = 200;</code>
-     */
-    public Builder addAllCustomOptimizers(
-        java.lang.Iterable<? extends org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizer> values) {
-      if (customOptimizersBuilder_ == null) {
-        ensureCustomOptimizersIsMutable();
-        com.google.protobuf.AbstractMessageLite.Builder.addAll(
-            values, customOptimizers_);
-        onChanged();
-      } else {
-        customOptimizersBuilder_.addAllMessages(values);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * list of CustomGraphOptimizers to apply.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.RewriterConfig.CustomGraphOptimizer custom_optimizers = 200;</code>
-     */
-    public Builder clearCustomOptimizers() {
-      if (customOptimizersBuilder_ == null) {
-        customOptimizers_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000002);
-        onChanged();
-      } else {
-        customOptimizersBuilder_.clear();
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * list of CustomGraphOptimizers to apply.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.RewriterConfig.CustomGraphOptimizer custom_optimizers = 200;</code>
-     */
-    public Builder removeCustomOptimizers(int index) {
-      if (customOptimizersBuilder_ == null) {
-        ensureCustomOptimizersIsMutable();
-        customOptimizers_.remove(index);
-        onChanged();
-      } else {
-        customOptimizersBuilder_.remove(index);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * list of CustomGraphOptimizers to apply.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.RewriterConfig.CustomGraphOptimizer custom_optimizers = 200;</code>
-     */
-    public org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizer.Builder getCustomOptimizersBuilder(
-        int index) {
-      return getCustomOptimizersFieldBuilder().getBuilder(index);
-    }
-    /**
-     * <pre>
-     * list of CustomGraphOptimizers to apply.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.RewriterConfig.CustomGraphOptimizer custom_optimizers = 200;</code>
-     */
-    public org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizerOrBuilder getCustomOptimizersOrBuilder(
-        int index) {
-      if (customOptimizersBuilder_ == null) {
-        return customOptimizers_.get(index);  } else {
-        return customOptimizersBuilder_.getMessageOrBuilder(index);
-      }
-    }
-    /**
-     * <pre>
-     * list of CustomGraphOptimizers to apply.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.RewriterConfig.CustomGraphOptimizer custom_optimizers = 200;</code>
-     */
-    public java.util.List<? extends org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizerOrBuilder> 
-         getCustomOptimizersOrBuilderList() {
-      if (customOptimizersBuilder_ != null) {
-        return customOptimizersBuilder_.getMessageOrBuilderList();
-      } else {
-        return java.util.Collections.unmodifiableList(customOptimizers_);
-      }
-    }
-    /**
-     * <pre>
-     * list of CustomGraphOptimizers to apply.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.RewriterConfig.CustomGraphOptimizer custom_optimizers = 200;</code>
-     */
-    public org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizer.Builder addCustomOptimizersBuilder() {
-      return getCustomOptimizersFieldBuilder().addBuilder(
-          org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizer.getDefaultInstance());
-    }
-    /**
-     * <pre>
-     * list of CustomGraphOptimizers to apply.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.RewriterConfig.CustomGraphOptimizer custom_optimizers = 200;</code>
-     */
-    public org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizer.Builder addCustomOptimizersBuilder(
-        int index) {
-      return getCustomOptimizersFieldBuilder().addBuilder(
-          index, org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizer.getDefaultInstance());
-    }
-    /**
-     * <pre>
-     * list of CustomGraphOptimizers to apply.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.RewriterConfig.CustomGraphOptimizer custom_optimizers = 200;</code>
-     */
-    public java.util.List<org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizer.Builder> 
-         getCustomOptimizersBuilderList() {
-      return getCustomOptimizersFieldBuilder().getBuilderList();
-    }
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizer, org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizer.Builder, org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizerOrBuilder> 
-        getCustomOptimizersFieldBuilder() {
-      if (customOptimizersBuilder_ == null) {
-        customOptimizersBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
-            org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizer, org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizer.Builder, org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizerOrBuilder>(
-                customOptimizers_,
-                ((bitField0_ & 0x00000002) != 0),
-                getParentForChildren(),
-                isClean());
-        customOptimizers_ = null;
-      }
-      return customOptimizersBuilder_;
-    }
-
-    private org.tensorflow.proto.framework.VerifierConfig interOptimizerVerifierConfig_;
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.VerifierConfig, org.tensorflow.proto.framework.VerifierConfig.Builder, org.tensorflow.proto.framework.VerifierConfigOrBuilder> interOptimizerVerifierConfigBuilder_;
-    /**
-     * <pre>
-     * VerifierConfig specifying the verifiers to be run after every optimizer.
-     * </pre>
-     *
-     * <code>.tensorflow.VerifierConfig inter_optimizer_verifier_config = 300;</code>
-     */
-    public boolean hasInterOptimizerVerifierConfig() {
-      return interOptimizerVerifierConfigBuilder_ != null || interOptimizerVerifierConfig_ != null;
-    }
-    /**
-     * <pre>
-     * VerifierConfig specifying the verifiers to be run after every optimizer.
-     * </pre>
-     *
-     * <code>.tensorflow.VerifierConfig inter_optimizer_verifier_config = 300;</code>
-     */
-    public org.tensorflow.proto.framework.VerifierConfig getInterOptimizerVerifierConfig() {
-      if (interOptimizerVerifierConfigBuilder_ == null) {
-        return interOptimizerVerifierConfig_ == null ? org.tensorflow.proto.framework.VerifierConfig.getDefaultInstance() : interOptimizerVerifierConfig_;
-      } else {
-        return interOptimizerVerifierConfigBuilder_.getMessage();
-      }
-    }
-    /**
-     * <pre>
-     * VerifierConfig specifying the verifiers to be run after every optimizer.
-     * </pre>
-     *
-     * <code>.tensorflow.VerifierConfig inter_optimizer_verifier_config = 300;</code>
-     */
-    public Builder setInterOptimizerVerifierConfig(org.tensorflow.proto.framework.VerifierConfig value) {
-      if (interOptimizerVerifierConfigBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        interOptimizerVerifierConfig_ = value;
-        onChanged();
-      } else {
-        interOptimizerVerifierConfigBuilder_.setMessage(value);
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * VerifierConfig specifying the verifiers to be run after every optimizer.
-     * </pre>
-     *
-     * <code>.tensorflow.VerifierConfig inter_optimizer_verifier_config = 300;</code>
-     */
-    public Builder setInterOptimizerVerifierConfig(
-        org.tensorflow.proto.framework.VerifierConfig.Builder builderForValue) {
-      if (interOptimizerVerifierConfigBuilder_ == null) {
-        interOptimizerVerifierConfig_ = builderForValue.build();
-        onChanged();
-      } else {
-        interOptimizerVerifierConfigBuilder_.setMessage(builderForValue.build());
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * VerifierConfig specifying the verifiers to be run after every optimizer.
-     * </pre>
-     *
-     * <code>.tensorflow.VerifierConfig inter_optimizer_verifier_config = 300;</code>
-     */
-    public Builder mergeInterOptimizerVerifierConfig(org.tensorflow.proto.framework.VerifierConfig value) {
-      if (interOptimizerVerifierConfigBuilder_ == null) {
-        if (interOptimizerVerifierConfig_ != null) {
-          interOptimizerVerifierConfig_ =
-            org.tensorflow.proto.framework.VerifierConfig.newBuilder(interOptimizerVerifierConfig_).mergeFrom(value).buildPartial();
-        } else {
-          interOptimizerVerifierConfig_ = value;
-        }
-        onChanged();
-      } else {
-        interOptimizerVerifierConfigBuilder_.mergeFrom(value);
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * VerifierConfig specifying the verifiers to be run after every optimizer.
-     * </pre>
-     *
-     * <code>.tensorflow.VerifierConfig inter_optimizer_verifier_config = 300;</code>
-     */
-    public Builder clearInterOptimizerVerifierConfig() {
-      if (interOptimizerVerifierConfigBuilder_ == null) {
-        interOptimizerVerifierConfig_ = null;
-        onChanged();
-      } else {
-        interOptimizerVerifierConfig_ = null;
-        interOptimizerVerifierConfigBuilder_ = null;
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * VerifierConfig specifying the verifiers to be run after every optimizer.
-     * </pre>
-     *
-     * <code>.tensorflow.VerifierConfig inter_optimizer_verifier_config = 300;</code>
-     */
-    public org.tensorflow.proto.framework.VerifierConfig.Builder getInterOptimizerVerifierConfigBuilder() {
-      
-      onChanged();
-      return getInterOptimizerVerifierConfigFieldBuilder().getBuilder();
-    }
-    /**
-     * <pre>
-     * VerifierConfig specifying the verifiers to be run after every optimizer.
-     * </pre>
-     *
-     * <code>.tensorflow.VerifierConfig inter_optimizer_verifier_config = 300;</code>
-     */
-    public org.tensorflow.proto.framework.VerifierConfigOrBuilder getInterOptimizerVerifierConfigOrBuilder() {
-      if (interOptimizerVerifierConfigBuilder_ != null) {
-        return interOptimizerVerifierConfigBuilder_.getMessageOrBuilder();
-      } else {
-        return interOptimizerVerifierConfig_ == null ?
-            org.tensorflow.proto.framework.VerifierConfig.getDefaultInstance() : interOptimizerVerifierConfig_;
-      }
-    }
-    /**
-     * <pre>
-     * VerifierConfig specifying the verifiers to be run after every optimizer.
-     * </pre>
-     *
-     * <code>.tensorflow.VerifierConfig inter_optimizer_verifier_config = 300;</code>
-     */
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.VerifierConfig, org.tensorflow.proto.framework.VerifierConfig.Builder, org.tensorflow.proto.framework.VerifierConfigOrBuilder> 
-        getInterOptimizerVerifierConfigFieldBuilder() {
-      if (interOptimizerVerifierConfigBuilder_ == null) {
-        interOptimizerVerifierConfigBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-            org.tensorflow.proto.framework.VerifierConfig, org.tensorflow.proto.framework.VerifierConfig.Builder, org.tensorflow.proto.framework.VerifierConfigOrBuilder>(
-                getInterOptimizerVerifierConfig(),
-                getParentForChildren(),
-                isClean());
-        interOptimizerVerifierConfig_ = null;
-      }
-      return interOptimizerVerifierConfigBuilder_;
-    }
-
-    private org.tensorflow.proto.framework.VerifierConfig postOptimizationVerifierConfig_;
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.VerifierConfig, org.tensorflow.proto.framework.VerifierConfig.Builder, org.tensorflow.proto.framework.VerifierConfigOrBuilder> postOptimizationVerifierConfigBuilder_;
-    /**
-     * <pre>
-     * VerifierConfig specifying the verifiers to be run at the end, after all
-     * optimizers have run.
-     * </pre>
-     *
-     * <code>.tensorflow.VerifierConfig post_optimization_verifier_config = 301;</code>
-     */
-    public boolean hasPostOptimizationVerifierConfig() {
-      return postOptimizationVerifierConfigBuilder_ != null || postOptimizationVerifierConfig_ != null;
-    }
-    /**
-     * <pre>
-     * VerifierConfig specifying the verifiers to be run at the end, after all
-     * optimizers have run.
-     * </pre>
-     *
-     * <code>.tensorflow.VerifierConfig post_optimization_verifier_config = 301;</code>
-     */
-    public org.tensorflow.proto.framework.VerifierConfig getPostOptimizationVerifierConfig() {
-      if (postOptimizationVerifierConfigBuilder_ == null) {
-        return postOptimizationVerifierConfig_ == null ? org.tensorflow.proto.framework.VerifierConfig.getDefaultInstance() : postOptimizationVerifierConfig_;
-      } else {
-        return postOptimizationVerifierConfigBuilder_.getMessage();
-      }
-    }
-    /**
-     * <pre>
-     * VerifierConfig specifying the verifiers to be run at the end, after all
-     * optimizers have run.
-     * </pre>
-     *
-     * <code>.tensorflow.VerifierConfig post_optimization_verifier_config = 301;</code>
-     */
-    public Builder setPostOptimizationVerifierConfig(org.tensorflow.proto.framework.VerifierConfig value) {
-      if (postOptimizationVerifierConfigBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        postOptimizationVerifierConfig_ = value;
-        onChanged();
-      } else {
-        postOptimizationVerifierConfigBuilder_.setMessage(value);
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * VerifierConfig specifying the verifiers to be run at the end, after all
-     * optimizers have run.
-     * </pre>
-     *
-     * <code>.tensorflow.VerifierConfig post_optimization_verifier_config = 301;</code>
-     */
-    public Builder setPostOptimizationVerifierConfig(
-        org.tensorflow.proto.framework.VerifierConfig.Builder builderForValue) {
-      if (postOptimizationVerifierConfigBuilder_ == null) {
-        postOptimizationVerifierConfig_ = builderForValue.build();
-        onChanged();
-      } else {
-        postOptimizationVerifierConfigBuilder_.setMessage(builderForValue.build());
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * VerifierConfig specifying the verifiers to be run at the end, after all
-     * optimizers have run.
-     * </pre>
-     *
-     * <code>.tensorflow.VerifierConfig post_optimization_verifier_config = 301;</code>
-     */
-    public Builder mergePostOptimizationVerifierConfig(org.tensorflow.proto.framework.VerifierConfig value) {
-      if (postOptimizationVerifierConfigBuilder_ == null) {
-        if (postOptimizationVerifierConfig_ != null) {
-          postOptimizationVerifierConfig_ =
-            org.tensorflow.proto.framework.VerifierConfig.newBuilder(postOptimizationVerifierConfig_).mergeFrom(value).buildPartial();
-        } else {
-          postOptimizationVerifierConfig_ = value;
-        }
-        onChanged();
-      } else {
-        postOptimizationVerifierConfigBuilder_.mergeFrom(value);
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * VerifierConfig specifying the verifiers to be run at the end, after all
-     * optimizers have run.
-     * </pre>
-     *
-     * <code>.tensorflow.VerifierConfig post_optimization_verifier_config = 301;</code>
-     */
-    public Builder clearPostOptimizationVerifierConfig() {
-      if (postOptimizationVerifierConfigBuilder_ == null) {
-        postOptimizationVerifierConfig_ = null;
-        onChanged();
-      } else {
-        postOptimizationVerifierConfig_ = null;
-        postOptimizationVerifierConfigBuilder_ = null;
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * VerifierConfig specifying the verifiers to be run at the end, after all
-     * optimizers have run.
-     * </pre>
-     *
-     * <code>.tensorflow.VerifierConfig post_optimization_verifier_config = 301;</code>
-     */
-    public org.tensorflow.proto.framework.VerifierConfig.Builder getPostOptimizationVerifierConfigBuilder() {
-      
-      onChanged();
-      return getPostOptimizationVerifierConfigFieldBuilder().getBuilder();
-    }
-    /**
-     * <pre>
-     * VerifierConfig specifying the verifiers to be run at the end, after all
-     * optimizers have run.
-     * </pre>
-     *
-     * <code>.tensorflow.VerifierConfig post_optimization_verifier_config = 301;</code>
-     */
-    public org.tensorflow.proto.framework.VerifierConfigOrBuilder getPostOptimizationVerifierConfigOrBuilder() {
-      if (postOptimizationVerifierConfigBuilder_ != null) {
-        return postOptimizationVerifierConfigBuilder_.getMessageOrBuilder();
-      } else {
-        return postOptimizationVerifierConfig_ == null ?
-            org.tensorflow.proto.framework.VerifierConfig.getDefaultInstance() : postOptimizationVerifierConfig_;
-      }
-    }
-    /**
-     * <pre>
-     * VerifierConfig specifying the verifiers to be run at the end, after all
-     * optimizers have run.
-     * </pre>
-     *
-     * <code>.tensorflow.VerifierConfig post_optimization_verifier_config = 301;</code>
-     */
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.VerifierConfig, org.tensorflow.proto.framework.VerifierConfig.Builder, org.tensorflow.proto.framework.VerifierConfigOrBuilder> 
-        getPostOptimizationVerifierConfigFieldBuilder() {
-      if (postOptimizationVerifierConfigBuilder_ == null) {
-        postOptimizationVerifierConfigBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-            org.tensorflow.proto.framework.VerifierConfig, org.tensorflow.proto.framework.VerifierConfig.Builder, org.tensorflow.proto.framework.VerifierConfigOrBuilder>(
-                getPostOptimizationVerifierConfig(),
-                getParentForChildren(),
-                isClean());
-        postOptimizationVerifierConfig_ = null;
-      }
-      return postOptimizationVerifierConfigBuilder_;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.RewriterConfig)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.RewriterConfig)
-  private static final org.tensorflow.proto.framework.RewriterConfig DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.RewriterConfig();
-  }
-
-  public static org.tensorflow.proto.framework.RewriterConfig getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<RewriterConfig>
-      PARSER = new com.google.protobuf.AbstractParser<RewriterConfig>() {
-    @java.lang.Override
-    public RewriterConfig parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new RewriterConfig(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<RewriterConfig> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<RewriterConfig> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.framework.RewriterConfig getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/RewriterConfigOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/RewriterConfigOrBuilder.java
deleted file mode 100644
index a8a1c6f1ed0..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/RewriterConfigOrBuilder.java
+++ /dev/null
@@ -1,627 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/rewriter_config.proto
-
-package org.tensorflow.proto.framework;
-
-public interface RewriterConfigOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.RewriterConfig)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <pre>
-   * Optimize tensor layouts (default is ON)
-   * e.g. This will try to use NCHW layout on GPU which is faster.
-   * </pre>
-   *
-   * <code>.tensorflow.RewriterConfig.Toggle layout_optimizer = 1;</code>
-   */
-  int getLayoutOptimizerValue();
-  /**
-   * <pre>
-   * Optimize tensor layouts (default is ON)
-   * e.g. This will try to use NCHW layout on GPU which is faster.
-   * </pre>
-   *
-   * <code>.tensorflow.RewriterConfig.Toggle layout_optimizer = 1;</code>
-   */
-  org.tensorflow.proto.framework.RewriterConfig.Toggle getLayoutOptimizer();
-
-  /**
-   * <pre>
-   * Fold constants (default is ON)
-   * Statically infer the value of tensors when possible, and materialize the
-   * result using constants.
-   * </pre>
-   *
-   * <code>.tensorflow.RewriterConfig.Toggle constant_folding = 3;</code>
-   */
-  int getConstantFoldingValue();
-  /**
-   * <pre>
-   * Fold constants (default is ON)
-   * Statically infer the value of tensors when possible, and materialize the
-   * result using constants.
-   * </pre>
-   *
-   * <code>.tensorflow.RewriterConfig.Toggle constant_folding = 3;</code>
-   */
-  org.tensorflow.proto.framework.RewriterConfig.Toggle getConstantFolding();
-
-  /**
-   * <pre>
-   * Shape optimizations (default is ON)
-   * Simplify computations made on shapes.
-   * </pre>
-   *
-   * <code>.tensorflow.RewriterConfig.Toggle shape_optimization = 13;</code>
-   */
-  int getShapeOptimizationValue();
-  /**
-   * <pre>
-   * Shape optimizations (default is ON)
-   * Simplify computations made on shapes.
-   * </pre>
-   *
-   * <code>.tensorflow.RewriterConfig.Toggle shape_optimization = 13;</code>
-   */
-  org.tensorflow.proto.framework.RewriterConfig.Toggle getShapeOptimization();
-
-  /**
-   * <pre>
-   * Remapping (default is ON)
-   * Remap subgraphs onto more efficient implementations.
-   * </pre>
-   *
-   * <code>.tensorflow.RewriterConfig.Toggle remapping = 14;</code>
-   */
-  int getRemappingValue();
-  /**
-   * <pre>
-   * Remapping (default is ON)
-   * Remap subgraphs onto more efficient implementations.
-   * </pre>
-   *
-   * <code>.tensorflow.RewriterConfig.Toggle remapping = 14;</code>
-   */
-  org.tensorflow.proto.framework.RewriterConfig.Toggle getRemapping();
-
-  /**
-   * <pre>
-   * Common subgraph elimination (default is ON)
-   * e.g. Simplify arithmetic ops; merge ops with same value (like constants).
-   * </pre>
-   *
-   * <code>.tensorflow.RewriterConfig.Toggle common_subgraph_elimination = 24;</code>
-   */
-  int getCommonSubgraphEliminationValue();
-  /**
-   * <pre>
-   * Common subgraph elimination (default is ON)
-   * e.g. Simplify arithmetic ops; merge ops with same value (like constants).
-   * </pre>
-   *
-   * <code>.tensorflow.RewriterConfig.Toggle common_subgraph_elimination = 24;</code>
-   */
-  org.tensorflow.proto.framework.RewriterConfig.Toggle getCommonSubgraphElimination();
-
-  /**
-   * <pre>
-   * Arithmetic optimizations (default is ON)
-   * e.g. Simplify arithmetic ops; merge ops with same value (like constants).
-   * </pre>
-   *
-   * <code>.tensorflow.RewriterConfig.Toggle arithmetic_optimization = 7;</code>
-   */
-  int getArithmeticOptimizationValue();
-  /**
-   * <pre>
-   * Arithmetic optimizations (default is ON)
-   * e.g. Simplify arithmetic ops; merge ops with same value (like constants).
-   * </pre>
-   *
-   * <code>.tensorflow.RewriterConfig.Toggle arithmetic_optimization = 7;</code>
-   */
-  org.tensorflow.proto.framework.RewriterConfig.Toggle getArithmeticOptimization();
-
-  /**
-   * <pre>
-   * Control dependency optimizations (default is ON).
-   * Remove redundant control dependencies, which may enable other optimization.
-   * </pre>
-   *
-   * <code>.tensorflow.RewriterConfig.Toggle dependency_optimization = 8;</code>
-   */
-  int getDependencyOptimizationValue();
-  /**
-   * <pre>
-   * Control dependency optimizations (default is ON).
-   * Remove redundant control dependencies, which may enable other optimization.
-   * </pre>
-   *
-   * <code>.tensorflow.RewriterConfig.Toggle dependency_optimization = 8;</code>
-   */
-  org.tensorflow.proto.framework.RewriterConfig.Toggle getDependencyOptimization();
-
-  /**
-   * <pre>
-   * Loop optimizations (default is ON).
-   * </pre>
-   *
-   * <code>.tensorflow.RewriterConfig.Toggle loop_optimization = 9;</code>
-   */
-  int getLoopOptimizationValue();
-  /**
-   * <pre>
-   * Loop optimizations (default is ON).
-   * </pre>
-   *
-   * <code>.tensorflow.RewriterConfig.Toggle loop_optimization = 9;</code>
-   */
-  org.tensorflow.proto.framework.RewriterConfig.Toggle getLoopOptimization();
-
-  /**
-   * <pre>
-   * Function optimizations (default is ON).
-   * </pre>
-   *
-   * <code>.tensorflow.RewriterConfig.Toggle function_optimization = 10;</code>
-   */
-  int getFunctionOptimizationValue();
-  /**
-   * <pre>
-   * Function optimizations (default is ON).
-   * </pre>
-   *
-   * <code>.tensorflow.RewriterConfig.Toggle function_optimization = 10;</code>
-   */
-  org.tensorflow.proto.framework.RewriterConfig.Toggle getFunctionOptimization();
-
-  /**
-   * <pre>
-   * Strips debug-related nodes from the graph (off by default).
-   * </pre>
-   *
-   * <code>.tensorflow.RewriterConfig.Toggle debug_stripper = 11;</code>
-   */
-  int getDebugStripperValue();
-  /**
-   * <pre>
-   * Strips debug-related nodes from the graph (off by default).
-   * </pre>
-   *
-   * <code>.tensorflow.RewriterConfig.Toggle debug_stripper = 11;</code>
-   */
-  org.tensorflow.proto.framework.RewriterConfig.Toggle getDebugStripper();
-
-  /**
-   * <pre>
-   * If true, don't remove unnecessary ops from the graph
-   * </pre>
-   *
-   * <code>bool disable_model_pruning = 2;</code>
-   */
-  boolean getDisableModelPruning();
-
-  /**
-   * <pre>
-   * Try to allocate some independent Op outputs contiguously in order to
-   * merge or eliminate downstream Ops (off by default).
-   * </pre>
-   *
-   * <code>.tensorflow.RewriterConfig.Toggle scoped_allocator_optimization = 15;</code>
-   */
-  int getScopedAllocatorOptimizationValue();
-  /**
-   * <pre>
-   * Try to allocate some independent Op outputs contiguously in order to
-   * merge or eliminate downstream Ops (off by default).
-   * </pre>
-   *
-   * <code>.tensorflow.RewriterConfig.Toggle scoped_allocator_optimization = 15;</code>
-   */
-  org.tensorflow.proto.framework.RewriterConfig.Toggle getScopedAllocatorOptimization();
-
-  /**
-   * <pre>
-   * Force small ops onto the CPU (default is OFF).
-   * </pre>
-   *
-   * <code>.tensorflow.RewriterConfig.Toggle pin_to_host_optimization = 18;</code>
-   */
-  int getPinToHostOptimizationValue();
-  /**
-   * <pre>
-   * Force small ops onto the CPU (default is OFF).
-   * </pre>
-   *
-   * <code>.tensorflow.RewriterConfig.Toggle pin_to_host_optimization = 18;</code>
-   */
-  org.tensorflow.proto.framework.RewriterConfig.Toggle getPinToHostOptimization();
-
-  /**
-   * <pre>
-   * Enable the swap of kernel implementations based on the device placement
-   * (default is ON).
-   * </pre>
-   *
-   * <code>.tensorflow.RewriterConfig.Toggle implementation_selector = 22;</code>
-   */
-  int getImplementationSelectorValue();
-  /**
-   * <pre>
-   * Enable the swap of kernel implementations based on the device placement
-   * (default is ON).
-   * </pre>
-   *
-   * <code>.tensorflow.RewriterConfig.Toggle implementation_selector = 22;</code>
-   */
-  org.tensorflow.proto.framework.RewriterConfig.Toggle getImplementationSelector();
-
-  /**
-   * <pre>
-   * Optimize data types for CUDA (default is OFF).
-   * This will try to use float16 on GPU which is faster.
-   * Note that this can change the numerical stability of the graph and may
-   * require the use of loss scaling to maintain model convergence.
-   * </pre>
-   *
-   * <code>.tensorflow.RewriterConfig.Toggle auto_mixed_precision = 23;</code>
-   */
-  int getAutoMixedPrecisionValue();
-  /**
-   * <pre>
-   * Optimize data types for CUDA (default is OFF).
-   * This will try to use float16 on GPU which is faster.
-   * Note that this can change the numerical stability of the graph and may
-   * require the use of loss scaling to maintain model convergence.
-   * </pre>
-   *
-   * <code>.tensorflow.RewriterConfig.Toggle auto_mixed_precision = 23;</code>
-   */
-  org.tensorflow.proto.framework.RewriterConfig.Toggle getAutoMixedPrecision();
-
-  /**
-   * <pre>
-   * Optimize data types for MKL (default is OFF).
-   * This will try to use bfloat16 on CPUs, which is faster.
-   * Note that this can change the numerical stability of the graph.
-   * </pre>
-   *
-   * <code>.tensorflow.RewriterConfig.Toggle auto_mixed_precision_mkl = 25;</code>
-   */
-  int getAutoMixedPrecisionMklValue();
-  /**
-   * <pre>
-   * Optimize data types for MKL (default is OFF).
-   * This will try to use bfloat16 on CPUs, which is faster.
-   * Note that this can change the numerical stability of the graph.
-   * </pre>
-   *
-   * <code>.tensorflow.RewriterConfig.Toggle auto_mixed_precision_mkl = 25;</code>
-   */
-  org.tensorflow.proto.framework.RewriterConfig.Toggle getAutoMixedPrecisionMkl();
-
-  /**
-   * <pre>
-   * Disable the entire meta optimizer (off by default).
-   * </pre>
-   *
-   * <code>bool disable_meta_optimizer = 19;</code>
-   */
-  boolean getDisableMetaOptimizer();
-
-  /**
-   * <pre>
-   * Controls how many times we run the optimizers in meta optimizer (default
-   * is once).
-   * </pre>
-   *
-   * <code>.tensorflow.RewriterConfig.NumIterationsType meta_optimizer_iterations = 12;</code>
-   */
-  int getMetaOptimizerIterationsValue();
-  /**
-   * <pre>
-   * Controls how many times we run the optimizers in meta optimizer (default
-   * is once).
-   * </pre>
-   *
-   * <code>.tensorflow.RewriterConfig.NumIterationsType meta_optimizer_iterations = 12;</code>
-   */
-  org.tensorflow.proto.framework.RewriterConfig.NumIterationsType getMetaOptimizerIterations();
-
-  /**
-   * <pre>
-   * The minimum number of nodes in a graph to optimizer. For smaller graphs,
-   * optimization is skipped.
-   * 0 means the system picks an appropriate number.
-   * &lt; 0 means do not skip optimization.
-   * </pre>
-   *
-   * <code>int32 min_graph_nodes = 17;</code>
-   */
-  int getMinGraphNodes();
-
-  /**
-   * <pre>
-   * Configures memory optimization passes through the meta-optimizer. Has no
-   * effect on manually requested memory optimization passes in the optimizers
-   * field.
-   * </pre>
-   *
-   * <code>.tensorflow.RewriterConfig.MemOptType memory_optimization = 4;</code>
-   */
-  int getMemoryOptimizationValue();
-  /**
-   * <pre>
-   * Configures memory optimization passes through the meta-optimizer. Has no
-   * effect on manually requested memory optimization passes in the optimizers
-   * field.
-   * </pre>
-   *
-   * <code>.tensorflow.RewriterConfig.MemOptType memory_optimization = 4;</code>
-   */
-  org.tensorflow.proto.framework.RewriterConfig.MemOptType getMemoryOptimization();
-
-  /**
-   * <pre>
-   * A node name scope for node names which are valid outputs of recomputations.
-   * Inputs to nodes that match this scope may be recomputed (subject either to
-   * manual annotation of those input nodes or to manual annotation and
-   * heuristics depending on memory_optimization), but the nodes themselves will
-   * not be recomputed. This matches any sub-scopes as well, meaning the scope
-   * can appear not just as a top-level scope. For example, if the value is
-   * "gradients/", the default, it will match node name "gradients/foo",
-   * "foo/gradients/bar", but not "foo_gradients/"
-   * </pre>
-   *
-   * <code>string memory_optimizer_target_node_name_scope = 6;</code>
-   */
-  java.lang.String getMemoryOptimizerTargetNodeNameScope();
-  /**
-   * <pre>
-   * A node name scope for node names which are valid outputs of recomputations.
-   * Inputs to nodes that match this scope may be recomputed (subject either to
-   * manual annotation of those input nodes or to manual annotation and
-   * heuristics depending on memory_optimization), but the nodes themselves will
-   * not be recomputed. This matches any sub-scopes as well, meaning the scope
-   * can appear not just as a top-level scope. For example, if the value is
-   * "gradients/", the default, it will match node name "gradients/foo",
-   * "foo/gradients/bar", but not "foo_gradients/"
-   * </pre>
-   *
-   * <code>string memory_optimizer_target_node_name_scope = 6;</code>
-   */
-  com.google.protobuf.ByteString
-      getMemoryOptimizerTargetNodeNameScopeBytes();
-
-  /**
-   * <pre>
-   * Maximum number of milliseconds to spend optimizing a single graph before
-   * timing out. If equal to 0 the system picks a default (currently 5 minutes).
-   * If less than 0 the optimizer will never time out.
-   * </pre>
-   *
-   * <code>int64 meta_optimizer_timeout_ms = 20;</code>
-   */
-  long getMetaOptimizerTimeoutMs();
-
-  /**
-   * <pre>
-   * Configures AutoParallel optimization passes either through the
-   * meta-optimizer or when manually specified through the optimizers field.
-   * </pre>
-   *
-   * <code>.tensorflow.AutoParallelOptions auto_parallel = 5;</code>
-   */
-  boolean hasAutoParallel();
-  /**
-   * <pre>
-   * Configures AutoParallel optimization passes either through the
-   * meta-optimizer or when manually specified through the optimizers field.
-   * </pre>
-   *
-   * <code>.tensorflow.AutoParallelOptions auto_parallel = 5;</code>
-   */
-  org.tensorflow.proto.framework.AutoParallelOptions getAutoParallel();
-  /**
-   * <pre>
-   * Configures AutoParallel optimization passes either through the
-   * meta-optimizer or when manually specified through the optimizers field.
-   * </pre>
-   *
-   * <code>.tensorflow.AutoParallelOptions auto_parallel = 5;</code>
-   */
-  org.tensorflow.proto.framework.AutoParallelOptionsOrBuilder getAutoParallelOrBuilder();
-
-  /**
-   * <pre>
-   * If true, any optimization pass failing will cause the MetaOptimizer to
-   * stop with an error. By default - or when set to false, failing passes are
-   * skipped silently.
-   * </pre>
-   *
-   * <code>bool fail_on_optimizer_errors = 21;</code>
-   */
-  boolean getFailOnOptimizerErrors();
-
-  /**
-   * <code>.tensorflow.ScopedAllocatorOptions scoped_allocator_opts = 16;</code>
-   */
-  boolean hasScopedAllocatorOpts();
-  /**
-   * <code>.tensorflow.ScopedAllocatorOptions scoped_allocator_opts = 16;</code>
-   */
-  org.tensorflow.proto.framework.ScopedAllocatorOptions getScopedAllocatorOpts();
-  /**
-   * <code>.tensorflow.ScopedAllocatorOptions scoped_allocator_opts = 16;</code>
-   */
-  org.tensorflow.proto.framework.ScopedAllocatorOptionsOrBuilder getScopedAllocatorOptsOrBuilder();
-
-  /**
-   * <pre>
-   * If non-empty, will use this as an alternative way to specify a list of
-   * optimizations to turn on and the order of the optimizations (replacing the
-   * meta-optimizer).
-   * Of the RewriterConfig options, only the AutoParallel configuration options
-   * (the auto_parallel field) apply to manually requested optimization passes
-   * ("autoparallel"). Memory optimization passes ("memory") invoked here are
-   * not configurable (in contrast to memory optimization passes through the
-   * meta-optimizer) and act only on manual op annotations.
-   * Custom optimizers (see custom_optimizers) that are not part of this
-   * schedule will be run after - in the order that they were specified.
-   * </pre>
-   *
-   * <code>repeated string optimizers = 100;</code>
-   */
-  java.util.List<java.lang.String>
-      getOptimizersList();
-  /**
-   * <pre>
-   * If non-empty, will use this as an alternative way to specify a list of
-   * optimizations to turn on and the order of the optimizations (replacing the
-   * meta-optimizer).
-   * Of the RewriterConfig options, only the AutoParallel configuration options
-   * (the auto_parallel field) apply to manually requested optimization passes
-   * ("autoparallel"). Memory optimization passes ("memory") invoked here are
-   * not configurable (in contrast to memory optimization passes through the
-   * meta-optimizer) and act only on manual op annotations.
-   * Custom optimizers (see custom_optimizers) that are not part of this
-   * schedule will be run after - in the order that they were specified.
-   * </pre>
-   *
-   * <code>repeated string optimizers = 100;</code>
-   */
-  int getOptimizersCount();
-  /**
-   * <pre>
-   * If non-empty, will use this as an alternative way to specify a list of
-   * optimizations to turn on and the order of the optimizations (replacing the
-   * meta-optimizer).
-   * Of the RewriterConfig options, only the AutoParallel configuration options
-   * (the auto_parallel field) apply to manually requested optimization passes
-   * ("autoparallel"). Memory optimization passes ("memory") invoked here are
-   * not configurable (in contrast to memory optimization passes through the
-   * meta-optimizer) and act only on manual op annotations.
-   * Custom optimizers (see custom_optimizers) that are not part of this
-   * schedule will be run after - in the order that they were specified.
-   * </pre>
-   *
-   * <code>repeated string optimizers = 100;</code>
-   */
-  java.lang.String getOptimizers(int index);
-  /**
-   * <pre>
-   * If non-empty, will use this as an alternative way to specify a list of
-   * optimizations to turn on and the order of the optimizations (replacing the
-   * meta-optimizer).
-   * Of the RewriterConfig options, only the AutoParallel configuration options
-   * (the auto_parallel field) apply to manually requested optimization passes
-   * ("autoparallel"). Memory optimization passes ("memory") invoked here are
-   * not configurable (in contrast to memory optimization passes through the
-   * meta-optimizer) and act only on manual op annotations.
-   * Custom optimizers (see custom_optimizers) that are not part of this
-   * schedule will be run after - in the order that they were specified.
-   * </pre>
-   *
-   * <code>repeated string optimizers = 100;</code>
-   */
-  com.google.protobuf.ByteString
-      getOptimizersBytes(int index);
-
-  /**
-   * <pre>
-   * list of CustomGraphOptimizers to apply.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.RewriterConfig.CustomGraphOptimizer custom_optimizers = 200;</code>
-   */
-  java.util.List<org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizer> 
-      getCustomOptimizersList();
-  /**
-   * <pre>
-   * list of CustomGraphOptimizers to apply.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.RewriterConfig.CustomGraphOptimizer custom_optimizers = 200;</code>
-   */
-  org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizer getCustomOptimizers(int index);
-  /**
-   * <pre>
-   * list of CustomGraphOptimizers to apply.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.RewriterConfig.CustomGraphOptimizer custom_optimizers = 200;</code>
-   */
-  int getCustomOptimizersCount();
-  /**
-   * <pre>
-   * list of CustomGraphOptimizers to apply.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.RewriterConfig.CustomGraphOptimizer custom_optimizers = 200;</code>
-   */
-  java.util.List<? extends org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizerOrBuilder> 
-      getCustomOptimizersOrBuilderList();
-  /**
-   * <pre>
-   * list of CustomGraphOptimizers to apply.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.RewriterConfig.CustomGraphOptimizer custom_optimizers = 200;</code>
-   */
-  org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizerOrBuilder getCustomOptimizersOrBuilder(
-      int index);
-
-  /**
-   * <pre>
-   * VerifierConfig specifying the verifiers to be run after every optimizer.
-   * </pre>
-   *
-   * <code>.tensorflow.VerifierConfig inter_optimizer_verifier_config = 300;</code>
-   */
-  boolean hasInterOptimizerVerifierConfig();
-  /**
-   * <pre>
-   * VerifierConfig specifying the verifiers to be run after every optimizer.
-   * </pre>
-   *
-   * <code>.tensorflow.VerifierConfig inter_optimizer_verifier_config = 300;</code>
-   */
-  org.tensorflow.proto.framework.VerifierConfig getInterOptimizerVerifierConfig();
-  /**
-   * <pre>
-   * VerifierConfig specifying the verifiers to be run after every optimizer.
-   * </pre>
-   *
-   * <code>.tensorflow.VerifierConfig inter_optimizer_verifier_config = 300;</code>
-   */
-  org.tensorflow.proto.framework.VerifierConfigOrBuilder getInterOptimizerVerifierConfigOrBuilder();
-
-  /**
-   * <pre>
-   * VerifierConfig specifying the verifiers to be run at the end, after all
-   * optimizers have run.
-   * </pre>
-   *
-   * <code>.tensorflow.VerifierConfig post_optimization_verifier_config = 301;</code>
-   */
-  boolean hasPostOptimizationVerifierConfig();
-  /**
-   * <pre>
-   * VerifierConfig specifying the verifiers to be run at the end, after all
-   * optimizers have run.
-   * </pre>
-   *
-   * <code>.tensorflow.VerifierConfig post_optimization_verifier_config = 301;</code>
-   */
-  org.tensorflow.proto.framework.VerifierConfig getPostOptimizationVerifierConfig();
-  /**
-   * <pre>
-   * VerifierConfig specifying the verifiers to be run at the end, after all
-   * optimizers have run.
-   * </pre>
-   *
-   * <code>.tensorflow.VerifierConfig post_optimization_verifier_config = 301;</code>
-   */
-  org.tensorflow.proto.framework.VerifierConfigOrBuilder getPostOptimizationVerifierConfigOrBuilder();
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/RewriterConfigProtos.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/RewriterConfigProtos.java
deleted file mode 100644
index 01debd9ea7f..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/RewriterConfigProtos.java
+++ /dev/null
@@ -1,159 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/rewriter_config.proto
-
-package org.tensorflow.proto.framework;
-
-public final class RewriterConfigProtos {
-  private RewriterConfigProtos() {}
-  public static void registerAllExtensions(
-      com.google.protobuf.ExtensionRegistryLite registry) {
-  }
-
-  public static void registerAllExtensions(
-      com.google.protobuf.ExtensionRegistry registry) {
-    registerAllExtensions(
-        (com.google.protobuf.ExtensionRegistryLite) registry);
-  }
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_AutoParallelOptions_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_AutoParallelOptions_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_ScopedAllocatorOptions_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_ScopedAllocatorOptions_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_RewriterConfig_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_RewriterConfig_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_RewriterConfig_CustomGraphOptimizer_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_RewriterConfig_CustomGraphOptimizer_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_RewriterConfig_CustomGraphOptimizer_ParameterMapEntry_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_RewriterConfig_CustomGraphOptimizer_ParameterMapEntry_fieldAccessorTable;
-
-  public static com.google.protobuf.Descriptors.FileDescriptor
-      getDescriptor() {
-    return descriptor;
-  }
-  private static  com.google.protobuf.Descriptors.FileDescriptor
-      descriptor;
-  static {
-    java.lang.String[] descriptorData = {
-      "\n.tensorflow/core/protobuf/rewriter_conf" +
-      "ig.proto\022\ntensorflow\032*tensorflow/core/fr" +
-      "amework/attr_value.proto\032.tensorflow/cor" +
-      "e/protobuf/verifier_config.proto\";\n\023Auto" +
-      "ParallelOptions\022\016\n\006enable\030\001 \001(\010\022\024\n\014num_r" +
-      "eplicas\030\002 \001(\005\"+\n\026ScopedAllocatorOptions\022" +
-      "\021\n\tenable_op\030\001 \003(\t\"\225\021\n\016RewriterConfig\022;\n" +
-      "\020layout_optimizer\030\001 \001(\0162!.tensorflow.Rew" +
-      "riterConfig.Toggle\022;\n\020constant_folding\030\003" +
-      " \001(\0162!.tensorflow.RewriterConfig.Toggle\022" +
-      "=\n\022shape_optimization\030\r \001(\0162!.tensorflow" +
-      ".RewriterConfig.Toggle\0224\n\tremapping\030\016 \001(" +
-      "\0162!.tensorflow.RewriterConfig.Toggle\022F\n\033" +
-      "common_subgraph_elimination\030\030 \001(\0162!.tens" +
-      "orflow.RewriterConfig.Toggle\022B\n\027arithmet" +
-      "ic_optimization\030\007 \001(\0162!.tensorflow.Rewri" +
-      "terConfig.Toggle\022B\n\027dependency_optimizat" +
-      "ion\030\010 \001(\0162!.tensorflow.RewriterConfig.To" +
-      "ggle\022<\n\021loop_optimization\030\t \001(\0162!.tensor" +
-      "flow.RewriterConfig.Toggle\022@\n\025function_o" +
-      "ptimization\030\n \001(\0162!.tensorflow.RewriterC" +
-      "onfig.Toggle\0229\n\016debug_stripper\030\013 \001(\0162!.t" +
-      "ensorflow.RewriterConfig.Toggle\022\035\n\025disab" +
-      "le_model_pruning\030\002 \001(\010\022H\n\035scoped_allocat" +
-      "or_optimization\030\017 \001(\0162!.tensorflow.Rewri" +
-      "terConfig.Toggle\022C\n\030pin_to_host_optimiza" +
-      "tion\030\022 \001(\0162!.tensorflow.RewriterConfig.T" +
-      "oggle\022B\n\027implementation_selector\030\026 \001(\0162!" +
-      ".tensorflow.RewriterConfig.Toggle\022?\n\024aut" +
-      "o_mixed_precision\030\027 \001(\0162!.tensorflow.Rew" +
-      "riterConfig.Toggle\022C\n\030auto_mixed_precisi" +
-      "on_mkl\030\031 \001(\0162!.tensorflow.RewriterConfig" +
-      ".Toggle\022\036\n\026disable_meta_optimizer\030\023 \001(\010\022" +
-      "O\n\031meta_optimizer_iterations\030\014 \001(\0162,.ten" +
-      "sorflow.RewriterConfig.NumIterationsType" +
-      "\022\027\n\017min_graph_nodes\030\021 \001(\005\022B\n\023memory_opti" +
-      "mization\030\004 \001(\0162%.tensorflow.RewriterConf" +
-      "ig.MemOptType\022/\n\'memory_optimizer_target" +
-      "_node_name_scope\030\006 \001(\t\022!\n\031meta_optimizer" +
-      "_timeout_ms\030\024 \001(\003\0226\n\rauto_parallel\030\005 \001(\013" +
-      "2\037.tensorflow.AutoParallelOptions\022 \n\030fai" +
-      "l_on_optimizer_errors\030\025 \001(\010\022A\n\025scoped_al" +
-      "locator_opts\030\020 \001(\0132\".tensorflow.ScopedAl" +
-      "locatorOptions\022\022\n\noptimizers\030d \003(\t\022K\n\021cu" +
-      "stom_optimizers\030\310\001 \003(\0132/.tensorflow.Rewr" +
-      "iterConfig.CustomGraphOptimizer\022D\n\037inter" +
-      "_optimizer_verifier_config\030\254\002 \001(\0132\032.tens" +
-      "orflow.VerifierConfig\022F\n!post_optimizati" +
-      "on_verifier_config\030\255\002 \001(\0132\032.tensorflow.V" +
-      "erifierConfig\032\312\001\n\024CustomGraphOptimizer\022\014" +
-      "\n\004name\030\001 \001(\t\022X\n\rparameter_map\030\002 \003(\0132A.te" +
-      "nsorflow.RewriterConfig.CustomGraphOptim" +
-      "izer.ParameterMapEntry\032J\n\021ParameterMapEn" +
-      "try\022\013\n\003key\030\001 \001(\t\022$\n\005value\030\002 \001(\0132\025.tensor" +
-      "flow.AttrValue:\0028\001\"6\n\006Toggle\022\013\n\007DEFAULT\020" +
-      "\000\022\006\n\002ON\020\001\022\007\n\003OFF\020\002\022\016\n\nAGGRESSIVE\020\003\"<\n\021Nu" +
-      "mIterationsType\022\025\n\021DEFAULT_NUM_ITERS\020\000\022\007" +
-      "\n\003ONE\020\001\022\007\n\003TWO\020\002\"\237\001\n\nMemOptType\022\023\n\017DEFAU" +
-      "LT_MEM_OPT\020\000\022\016\n\nNO_MEM_OPT\020\001\022\n\n\006MANUAL\020\002" +
-      "\022\027\n\023SWAPPING_HEURISTICS\020\004\022\034\n\030RECOMPUTATI" +
-      "ON_HEURISTICS\020\005\022\031\n\025SCHEDULING_HEURISTICS" +
-      "\020\006\022\016\n\nHEURISTICS\020\003B\205\001\n\036org.tensorflow.pr" +
-      "oto.frameworkB\024RewriterConfigProtosP\001ZHg" +
-      "ithub.com/tensorflow/tensorflow/tensorfl" +
-      "ow/go/core/core_protos_go_proto\370\001\001b\006prot" +
-      "o3"
-    };
-    descriptor = com.google.protobuf.Descriptors.FileDescriptor
-      .internalBuildGeneratedFileFrom(descriptorData,
-        new com.google.protobuf.Descriptors.FileDescriptor[] {
-          org.tensorflow.proto.framework.AttrValueProtos.getDescriptor(),
-          org.tensorflow.proto.framework.VerifierConfigProtos.getDescriptor(),
-        });
-    internal_static_tensorflow_AutoParallelOptions_descriptor =
-      getDescriptor().getMessageTypes().get(0);
-    internal_static_tensorflow_AutoParallelOptions_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_AutoParallelOptions_descriptor,
-        new java.lang.String[] { "Enable", "NumReplicas", });
-    internal_static_tensorflow_ScopedAllocatorOptions_descriptor =
-      getDescriptor().getMessageTypes().get(1);
-    internal_static_tensorflow_ScopedAllocatorOptions_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_ScopedAllocatorOptions_descriptor,
-        new java.lang.String[] { "EnableOp", });
-    internal_static_tensorflow_RewriterConfig_descriptor =
-      getDescriptor().getMessageTypes().get(2);
-    internal_static_tensorflow_RewriterConfig_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_RewriterConfig_descriptor,
-        new java.lang.String[] { "LayoutOptimizer", "ConstantFolding", "ShapeOptimization", "Remapping", "CommonSubgraphElimination", "ArithmeticOptimization", "DependencyOptimization", "LoopOptimization", "FunctionOptimization", "DebugStripper", "DisableModelPruning", "ScopedAllocatorOptimization", "PinToHostOptimization", "ImplementationSelector", "AutoMixedPrecision", "AutoMixedPrecisionMkl", "DisableMetaOptimizer", "MetaOptimizerIterations", "MinGraphNodes", "MemoryOptimization", "MemoryOptimizerTargetNodeNameScope", "MetaOptimizerTimeoutMs", "AutoParallel", "FailOnOptimizerErrors", "ScopedAllocatorOpts", "Optimizers", "CustomOptimizers", "InterOptimizerVerifierConfig", "PostOptimizationVerifierConfig", });
-    internal_static_tensorflow_RewriterConfig_CustomGraphOptimizer_descriptor =
-      internal_static_tensorflow_RewriterConfig_descriptor.getNestedTypes().get(0);
-    internal_static_tensorflow_RewriterConfig_CustomGraphOptimizer_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_RewriterConfig_CustomGraphOptimizer_descriptor,
-        new java.lang.String[] { "Name", "ParameterMap", });
-    internal_static_tensorflow_RewriterConfig_CustomGraphOptimizer_ParameterMapEntry_descriptor =
-      internal_static_tensorflow_RewriterConfig_CustomGraphOptimizer_descriptor.getNestedTypes().get(0);
-    internal_static_tensorflow_RewriterConfig_CustomGraphOptimizer_ParameterMapEntry_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_RewriterConfig_CustomGraphOptimizer_ParameterMapEntry_descriptor,
-        new java.lang.String[] { "Key", "Value", });
-    org.tensorflow.proto.framework.AttrValueProtos.getDescriptor();
-    org.tensorflow.proto.framework.VerifierConfigProtos.getDescriptor();
-  }
-
-  // @@protoc_insertion_point(outer_class_scope)
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/RunMetadata.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/RunMetadata.java
deleted file mode 100644
index 47f4aa27082..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/RunMetadata.java
+++ /dev/null
@@ -1,3277 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/config.proto
-
-package org.tensorflow.proto.framework;
-
-/**
- * <pre>
- * Metadata output (i.e., non-Tensor) for a single Run() call.
- * </pre>
- *
- * Protobuf type {@code tensorflow.RunMetadata}
- */
-public  final class RunMetadata extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.RunMetadata)
-    RunMetadataOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use RunMetadata.newBuilder() to construct.
-  private RunMetadata(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private RunMetadata() {
-    partitionGraphs_ = java.util.Collections.emptyList();
-    functionGraphs_ = java.util.Collections.emptyList();
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new RunMetadata();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private RunMetadata(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    int mutable_bitField0_ = 0;
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 10: {
-            org.tensorflow.proto.framework.StepStats.Builder subBuilder = null;
-            if (stepStats_ != null) {
-              subBuilder = stepStats_.toBuilder();
-            }
-            stepStats_ = input.readMessage(org.tensorflow.proto.framework.StepStats.parser(), extensionRegistry);
-            if (subBuilder != null) {
-              subBuilder.mergeFrom(stepStats_);
-              stepStats_ = subBuilder.buildPartial();
-            }
-
-            break;
-          }
-          case 18: {
-            org.tensorflow.proto.framework.CostGraphDef.Builder subBuilder = null;
-            if (costGraph_ != null) {
-              subBuilder = costGraph_.toBuilder();
-            }
-            costGraph_ = input.readMessage(org.tensorflow.proto.framework.CostGraphDef.parser(), extensionRegistry);
-            if (subBuilder != null) {
-              subBuilder.mergeFrom(costGraph_);
-              costGraph_ = subBuilder.buildPartial();
-            }
-
-            break;
-          }
-          case 26: {
-            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
-              partitionGraphs_ = new java.util.ArrayList<org.tensorflow.proto.framework.GraphDef>();
-              mutable_bitField0_ |= 0x00000001;
-            }
-            partitionGraphs_.add(
-                input.readMessage(org.tensorflow.proto.framework.GraphDef.parser(), extensionRegistry));
-            break;
-          }
-          case 34: {
-            if (!((mutable_bitField0_ & 0x00000002) != 0)) {
-              functionGraphs_ = new java.util.ArrayList<org.tensorflow.proto.framework.RunMetadata.FunctionGraphs>();
-              mutable_bitField0_ |= 0x00000002;
-            }
-            functionGraphs_.add(
-                input.readMessage(org.tensorflow.proto.framework.RunMetadata.FunctionGraphs.parser(), extensionRegistry));
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      if (((mutable_bitField0_ & 0x00000001) != 0)) {
-        partitionGraphs_ = java.util.Collections.unmodifiableList(partitionGraphs_);
-      }
-      if (((mutable_bitField0_ & 0x00000002) != 0)) {
-        functionGraphs_ = java.util.Collections.unmodifiableList(functionGraphs_);
-      }
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_RunMetadata_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_RunMetadata_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.framework.RunMetadata.class, org.tensorflow.proto.framework.RunMetadata.Builder.class);
-  }
-
-  public interface FunctionGraphsOrBuilder extends
-      // @@protoc_insertion_point(interface_extends:tensorflow.RunMetadata.FunctionGraphs)
-      com.google.protobuf.MessageOrBuilder {
-
-    /**
-     * <pre>
-     * TODO(nareshmodi): Include some sort of function/cache-key identifier?
-     * </pre>
-     *
-     * <code>repeated .tensorflow.GraphDef partition_graphs = 1;</code>
-     */
-    java.util.List<org.tensorflow.proto.framework.GraphDef> 
-        getPartitionGraphsList();
-    /**
-     * <pre>
-     * TODO(nareshmodi): Include some sort of function/cache-key identifier?
-     * </pre>
-     *
-     * <code>repeated .tensorflow.GraphDef partition_graphs = 1;</code>
-     */
-    org.tensorflow.proto.framework.GraphDef getPartitionGraphs(int index);
-    /**
-     * <pre>
-     * TODO(nareshmodi): Include some sort of function/cache-key identifier?
-     * </pre>
-     *
-     * <code>repeated .tensorflow.GraphDef partition_graphs = 1;</code>
-     */
-    int getPartitionGraphsCount();
-    /**
-     * <pre>
-     * TODO(nareshmodi): Include some sort of function/cache-key identifier?
-     * </pre>
-     *
-     * <code>repeated .tensorflow.GraphDef partition_graphs = 1;</code>
-     */
-    java.util.List<? extends org.tensorflow.proto.framework.GraphDefOrBuilder> 
-        getPartitionGraphsOrBuilderList();
-    /**
-     * <pre>
-     * TODO(nareshmodi): Include some sort of function/cache-key identifier?
-     * </pre>
-     *
-     * <code>repeated .tensorflow.GraphDef partition_graphs = 1;</code>
-     */
-    org.tensorflow.proto.framework.GraphDefOrBuilder getPartitionGraphsOrBuilder(
-        int index);
-
-    /**
-     * <code>.tensorflow.GraphDef pre_optimization_graph = 2;</code>
-     */
-    boolean hasPreOptimizationGraph();
-    /**
-     * <code>.tensorflow.GraphDef pre_optimization_graph = 2;</code>
-     */
-    org.tensorflow.proto.framework.GraphDef getPreOptimizationGraph();
-    /**
-     * <code>.tensorflow.GraphDef pre_optimization_graph = 2;</code>
-     */
-    org.tensorflow.proto.framework.GraphDefOrBuilder getPreOptimizationGraphOrBuilder();
-
-    /**
-     * <code>.tensorflow.GraphDef post_optimization_graph = 3;</code>
-     */
-    boolean hasPostOptimizationGraph();
-    /**
-     * <code>.tensorflow.GraphDef post_optimization_graph = 3;</code>
-     */
-    org.tensorflow.proto.framework.GraphDef getPostOptimizationGraph();
-    /**
-     * <code>.tensorflow.GraphDef post_optimization_graph = 3;</code>
-     */
-    org.tensorflow.proto.framework.GraphDefOrBuilder getPostOptimizationGraphOrBuilder();
-  }
-  /**
-   * Protobuf type {@code tensorflow.RunMetadata.FunctionGraphs}
-   */
-  public  static final class FunctionGraphs extends
-      com.google.protobuf.GeneratedMessageV3 implements
-      // @@protoc_insertion_point(message_implements:tensorflow.RunMetadata.FunctionGraphs)
-      FunctionGraphsOrBuilder {
-  private static final long serialVersionUID = 0L;
-    // Use FunctionGraphs.newBuilder() to construct.
-    private FunctionGraphs(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-      super(builder);
-    }
-    private FunctionGraphs() {
-      partitionGraphs_ = java.util.Collections.emptyList();
-    }
-
-    @java.lang.Override
-    @SuppressWarnings({"unused"})
-    protected java.lang.Object newInstance(
-        UnusedPrivateParameter unused) {
-      return new FunctionGraphs();
-    }
-
-    @java.lang.Override
-    public final com.google.protobuf.UnknownFieldSet
-    getUnknownFields() {
-      return this.unknownFields;
-    }
-    private FunctionGraphs(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      this();
-      if (extensionRegistry == null) {
-        throw new java.lang.NullPointerException();
-      }
-      int mutable_bitField0_ = 0;
-      com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-          com.google.protobuf.UnknownFieldSet.newBuilder();
-      try {
-        boolean done = false;
-        while (!done) {
-          int tag = input.readTag();
-          switch (tag) {
-            case 0:
-              done = true;
-              break;
-            case 10: {
-              if (!((mutable_bitField0_ & 0x00000001) != 0)) {
-                partitionGraphs_ = new java.util.ArrayList<org.tensorflow.proto.framework.GraphDef>();
-                mutable_bitField0_ |= 0x00000001;
-              }
-              partitionGraphs_.add(
-                  input.readMessage(org.tensorflow.proto.framework.GraphDef.parser(), extensionRegistry));
-              break;
-            }
-            case 18: {
-              org.tensorflow.proto.framework.GraphDef.Builder subBuilder = null;
-              if (preOptimizationGraph_ != null) {
-                subBuilder = preOptimizationGraph_.toBuilder();
-              }
-              preOptimizationGraph_ = input.readMessage(org.tensorflow.proto.framework.GraphDef.parser(), extensionRegistry);
-              if (subBuilder != null) {
-                subBuilder.mergeFrom(preOptimizationGraph_);
-                preOptimizationGraph_ = subBuilder.buildPartial();
-              }
-
-              break;
-            }
-            case 26: {
-              org.tensorflow.proto.framework.GraphDef.Builder subBuilder = null;
-              if (postOptimizationGraph_ != null) {
-                subBuilder = postOptimizationGraph_.toBuilder();
-              }
-              postOptimizationGraph_ = input.readMessage(org.tensorflow.proto.framework.GraphDef.parser(), extensionRegistry);
-              if (subBuilder != null) {
-                subBuilder.mergeFrom(postOptimizationGraph_);
-                postOptimizationGraph_ = subBuilder.buildPartial();
-              }
-
-              break;
-            }
-            default: {
-              if (!parseUnknownField(
-                  input, unknownFields, extensionRegistry, tag)) {
-                done = true;
-              }
-              break;
-            }
-          }
-        }
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        throw e.setUnfinishedMessage(this);
-      } catch (java.io.IOException e) {
-        throw new com.google.protobuf.InvalidProtocolBufferException(
-            e).setUnfinishedMessage(this);
-      } finally {
-        if (((mutable_bitField0_ & 0x00000001) != 0)) {
-          partitionGraphs_ = java.util.Collections.unmodifiableList(partitionGraphs_);
-        }
-        this.unknownFields = unknownFields.build();
-        makeExtensionsImmutable();
-      }
-    }
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_RunMetadata_FunctionGraphs_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_RunMetadata_FunctionGraphs_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.RunMetadata.FunctionGraphs.class, org.tensorflow.proto.framework.RunMetadata.FunctionGraphs.Builder.class);
-    }
-
-    public static final int PARTITION_GRAPHS_FIELD_NUMBER = 1;
-    private java.util.List<org.tensorflow.proto.framework.GraphDef> partitionGraphs_;
-    /**
-     * <pre>
-     * TODO(nareshmodi): Include some sort of function/cache-key identifier?
-     * </pre>
-     *
-     * <code>repeated .tensorflow.GraphDef partition_graphs = 1;</code>
-     */
-    public java.util.List<org.tensorflow.proto.framework.GraphDef> getPartitionGraphsList() {
-      return partitionGraphs_;
-    }
-    /**
-     * <pre>
-     * TODO(nareshmodi): Include some sort of function/cache-key identifier?
-     * </pre>
-     *
-     * <code>repeated .tensorflow.GraphDef partition_graphs = 1;</code>
-     */
-    public java.util.List<? extends org.tensorflow.proto.framework.GraphDefOrBuilder> 
-        getPartitionGraphsOrBuilderList() {
-      return partitionGraphs_;
-    }
-    /**
-     * <pre>
-     * TODO(nareshmodi): Include some sort of function/cache-key identifier?
-     * </pre>
-     *
-     * <code>repeated .tensorflow.GraphDef partition_graphs = 1;</code>
-     */
-    public int getPartitionGraphsCount() {
-      return partitionGraphs_.size();
-    }
-    /**
-     * <pre>
-     * TODO(nareshmodi): Include some sort of function/cache-key identifier?
-     * </pre>
-     *
-     * <code>repeated .tensorflow.GraphDef partition_graphs = 1;</code>
-     */
-    public org.tensorflow.proto.framework.GraphDef getPartitionGraphs(int index) {
-      return partitionGraphs_.get(index);
-    }
-    /**
-     * <pre>
-     * TODO(nareshmodi): Include some sort of function/cache-key identifier?
-     * </pre>
-     *
-     * <code>repeated .tensorflow.GraphDef partition_graphs = 1;</code>
-     */
-    public org.tensorflow.proto.framework.GraphDefOrBuilder getPartitionGraphsOrBuilder(
-        int index) {
-      return partitionGraphs_.get(index);
-    }
-
-    public static final int PRE_OPTIMIZATION_GRAPH_FIELD_NUMBER = 2;
-    private org.tensorflow.proto.framework.GraphDef preOptimizationGraph_;
-    /**
-     * <code>.tensorflow.GraphDef pre_optimization_graph = 2;</code>
-     */
-    public boolean hasPreOptimizationGraph() {
-      return preOptimizationGraph_ != null;
-    }
-    /**
-     * <code>.tensorflow.GraphDef pre_optimization_graph = 2;</code>
-     */
-    public org.tensorflow.proto.framework.GraphDef getPreOptimizationGraph() {
-      return preOptimizationGraph_ == null ? org.tensorflow.proto.framework.GraphDef.getDefaultInstance() : preOptimizationGraph_;
-    }
-    /**
-     * <code>.tensorflow.GraphDef pre_optimization_graph = 2;</code>
-     */
-    public org.tensorflow.proto.framework.GraphDefOrBuilder getPreOptimizationGraphOrBuilder() {
-      return getPreOptimizationGraph();
-    }
-
-    public static final int POST_OPTIMIZATION_GRAPH_FIELD_NUMBER = 3;
-    private org.tensorflow.proto.framework.GraphDef postOptimizationGraph_;
-    /**
-     * <code>.tensorflow.GraphDef post_optimization_graph = 3;</code>
-     */
-    public boolean hasPostOptimizationGraph() {
-      return postOptimizationGraph_ != null;
-    }
-    /**
-     * <code>.tensorflow.GraphDef post_optimization_graph = 3;</code>
-     */
-    public org.tensorflow.proto.framework.GraphDef getPostOptimizationGraph() {
-      return postOptimizationGraph_ == null ? org.tensorflow.proto.framework.GraphDef.getDefaultInstance() : postOptimizationGraph_;
-    }
-    /**
-     * <code>.tensorflow.GraphDef post_optimization_graph = 3;</code>
-     */
-    public org.tensorflow.proto.framework.GraphDefOrBuilder getPostOptimizationGraphOrBuilder() {
-      return getPostOptimizationGraph();
-    }
-
-    private byte memoizedIsInitialized = -1;
-    @java.lang.Override
-    public final boolean isInitialized() {
-      byte isInitialized = memoizedIsInitialized;
-      if (isInitialized == 1) return true;
-      if (isInitialized == 0) return false;
-
-      memoizedIsInitialized = 1;
-      return true;
-    }
-
-    @java.lang.Override
-    public void writeTo(com.google.protobuf.CodedOutputStream output)
-                        throws java.io.IOException {
-      for (int i = 0; i < partitionGraphs_.size(); i++) {
-        output.writeMessage(1, partitionGraphs_.get(i));
-      }
-      if (preOptimizationGraph_ != null) {
-        output.writeMessage(2, getPreOptimizationGraph());
-      }
-      if (postOptimizationGraph_ != null) {
-        output.writeMessage(3, getPostOptimizationGraph());
-      }
-      unknownFields.writeTo(output);
-    }
-
-    @java.lang.Override
-    public int getSerializedSize() {
-      int size = memoizedSize;
-      if (size != -1) return size;
-
-      size = 0;
-      for (int i = 0; i < partitionGraphs_.size(); i++) {
-        size += com.google.protobuf.CodedOutputStream
-          .computeMessageSize(1, partitionGraphs_.get(i));
-      }
-      if (preOptimizationGraph_ != null) {
-        size += com.google.protobuf.CodedOutputStream
-          .computeMessageSize(2, getPreOptimizationGraph());
-      }
-      if (postOptimizationGraph_ != null) {
-        size += com.google.protobuf.CodedOutputStream
-          .computeMessageSize(3, getPostOptimizationGraph());
-      }
-      size += unknownFields.getSerializedSize();
-      memoizedSize = size;
-      return size;
-    }
-
-    @java.lang.Override
-    public boolean equals(final java.lang.Object obj) {
-      if (obj == this) {
-       return true;
-      }
-      if (!(obj instanceof org.tensorflow.proto.framework.RunMetadata.FunctionGraphs)) {
-        return super.equals(obj);
-      }
-      org.tensorflow.proto.framework.RunMetadata.FunctionGraphs other = (org.tensorflow.proto.framework.RunMetadata.FunctionGraphs) obj;
-
-      if (!getPartitionGraphsList()
-          .equals(other.getPartitionGraphsList())) return false;
-      if (hasPreOptimizationGraph() != other.hasPreOptimizationGraph()) return false;
-      if (hasPreOptimizationGraph()) {
-        if (!getPreOptimizationGraph()
-            .equals(other.getPreOptimizationGraph())) return false;
-      }
-      if (hasPostOptimizationGraph() != other.hasPostOptimizationGraph()) return false;
-      if (hasPostOptimizationGraph()) {
-        if (!getPostOptimizationGraph()
-            .equals(other.getPostOptimizationGraph())) return false;
-      }
-      if (!unknownFields.equals(other.unknownFields)) return false;
-      return true;
-    }
-
-    @java.lang.Override
-    public int hashCode() {
-      if (memoizedHashCode != 0) {
-        return memoizedHashCode;
-      }
-      int hash = 41;
-      hash = (19 * hash) + getDescriptor().hashCode();
-      if (getPartitionGraphsCount() > 0) {
-        hash = (37 * hash) + PARTITION_GRAPHS_FIELD_NUMBER;
-        hash = (53 * hash) + getPartitionGraphsList().hashCode();
-      }
-      if (hasPreOptimizationGraph()) {
-        hash = (37 * hash) + PRE_OPTIMIZATION_GRAPH_FIELD_NUMBER;
-        hash = (53 * hash) + getPreOptimizationGraph().hashCode();
-      }
-      if (hasPostOptimizationGraph()) {
-        hash = (37 * hash) + POST_OPTIMIZATION_GRAPH_FIELD_NUMBER;
-        hash = (53 * hash) + getPostOptimizationGraph().hashCode();
-      }
-      hash = (29 * hash) + unknownFields.hashCode();
-      memoizedHashCode = hash;
-      return hash;
-    }
-
-    public static org.tensorflow.proto.framework.RunMetadata.FunctionGraphs parseFrom(
-        java.nio.ByteBuffer data)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data);
-    }
-    public static org.tensorflow.proto.framework.RunMetadata.FunctionGraphs parseFrom(
-        java.nio.ByteBuffer data,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.RunMetadata.FunctionGraphs parseFrom(
-        com.google.protobuf.ByteString data)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data);
-    }
-    public static org.tensorflow.proto.framework.RunMetadata.FunctionGraphs parseFrom(
-        com.google.protobuf.ByteString data,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.RunMetadata.FunctionGraphs parseFrom(byte[] data)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data);
-    }
-    public static org.tensorflow.proto.framework.RunMetadata.FunctionGraphs parseFrom(
-        byte[] data,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.RunMetadata.FunctionGraphs parseFrom(java.io.InputStream input)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input);
-    }
-    public static org.tensorflow.proto.framework.RunMetadata.FunctionGraphs parseFrom(
-        java.io.InputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.RunMetadata.FunctionGraphs parseDelimitedFrom(java.io.InputStream input)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseDelimitedWithIOException(PARSER, input);
-    }
-    public static org.tensorflow.proto.framework.RunMetadata.FunctionGraphs parseDelimitedFrom(
-        java.io.InputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.RunMetadata.FunctionGraphs parseFrom(
-        com.google.protobuf.CodedInputStream input)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input);
-    }
-    public static org.tensorflow.proto.framework.RunMetadata.FunctionGraphs parseFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input, extensionRegistry);
-    }
-
-    @java.lang.Override
-    public Builder newBuilderForType() { return newBuilder(); }
-    public static Builder newBuilder() {
-      return DEFAULT_INSTANCE.toBuilder();
-    }
-    public static Builder newBuilder(org.tensorflow.proto.framework.RunMetadata.FunctionGraphs prototype) {
-      return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-    }
-    @java.lang.Override
-    public Builder toBuilder() {
-      return this == DEFAULT_INSTANCE
-          ? new Builder() : new Builder().mergeFrom(this);
-    }
-
-    @java.lang.Override
-    protected Builder newBuilderForType(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      Builder builder = new Builder(parent);
-      return builder;
-    }
-    /**
-     * Protobuf type {@code tensorflow.RunMetadata.FunctionGraphs}
-     */
-    public static final class Builder extends
-        com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-        // @@protoc_insertion_point(builder_implements:tensorflow.RunMetadata.FunctionGraphs)
-        org.tensorflow.proto.framework.RunMetadata.FunctionGraphsOrBuilder {
-      public static final com.google.protobuf.Descriptors.Descriptor
-          getDescriptor() {
-        return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_RunMetadata_FunctionGraphs_descriptor;
-      }
-
-      @java.lang.Override
-      protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-          internalGetFieldAccessorTable() {
-        return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_RunMetadata_FunctionGraphs_fieldAccessorTable
-            .ensureFieldAccessorsInitialized(
-                org.tensorflow.proto.framework.RunMetadata.FunctionGraphs.class, org.tensorflow.proto.framework.RunMetadata.FunctionGraphs.Builder.class);
-      }
-
-      // Construct using org.tensorflow.proto.framework.RunMetadata.FunctionGraphs.newBuilder()
-      private Builder() {
-        maybeForceBuilderInitialization();
-      }
-
-      private Builder(
-          com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-        super(parent);
-        maybeForceBuilderInitialization();
-      }
-      private void maybeForceBuilderInitialization() {
-        if (com.google.protobuf.GeneratedMessageV3
-                .alwaysUseFieldBuilders) {
-          getPartitionGraphsFieldBuilder();
-        }
-      }
-      @java.lang.Override
-      public Builder clear() {
-        super.clear();
-        if (partitionGraphsBuilder_ == null) {
-          partitionGraphs_ = java.util.Collections.emptyList();
-          bitField0_ = (bitField0_ & ~0x00000001);
-        } else {
-          partitionGraphsBuilder_.clear();
-        }
-        if (preOptimizationGraphBuilder_ == null) {
-          preOptimizationGraph_ = null;
-        } else {
-          preOptimizationGraph_ = null;
-          preOptimizationGraphBuilder_ = null;
-        }
-        if (postOptimizationGraphBuilder_ == null) {
-          postOptimizationGraph_ = null;
-        } else {
-          postOptimizationGraph_ = null;
-          postOptimizationGraphBuilder_ = null;
-        }
-        return this;
-      }
-
-      @java.lang.Override
-      public com.google.protobuf.Descriptors.Descriptor
-          getDescriptorForType() {
-        return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_RunMetadata_FunctionGraphs_descriptor;
-      }
-
-      @java.lang.Override
-      public org.tensorflow.proto.framework.RunMetadata.FunctionGraphs getDefaultInstanceForType() {
-        return org.tensorflow.proto.framework.RunMetadata.FunctionGraphs.getDefaultInstance();
-      }
-
-      @java.lang.Override
-      public org.tensorflow.proto.framework.RunMetadata.FunctionGraphs build() {
-        org.tensorflow.proto.framework.RunMetadata.FunctionGraphs result = buildPartial();
-        if (!result.isInitialized()) {
-          throw newUninitializedMessageException(result);
-        }
-        return result;
-      }
-
-      @java.lang.Override
-      public org.tensorflow.proto.framework.RunMetadata.FunctionGraphs buildPartial() {
-        org.tensorflow.proto.framework.RunMetadata.FunctionGraphs result = new org.tensorflow.proto.framework.RunMetadata.FunctionGraphs(this);
-        int from_bitField0_ = bitField0_;
-        if (partitionGraphsBuilder_ == null) {
-          if (((bitField0_ & 0x00000001) != 0)) {
-            partitionGraphs_ = java.util.Collections.unmodifiableList(partitionGraphs_);
-            bitField0_ = (bitField0_ & ~0x00000001);
-          }
-          result.partitionGraphs_ = partitionGraphs_;
-        } else {
-          result.partitionGraphs_ = partitionGraphsBuilder_.build();
-        }
-        if (preOptimizationGraphBuilder_ == null) {
-          result.preOptimizationGraph_ = preOptimizationGraph_;
-        } else {
-          result.preOptimizationGraph_ = preOptimizationGraphBuilder_.build();
-        }
-        if (postOptimizationGraphBuilder_ == null) {
-          result.postOptimizationGraph_ = postOptimizationGraph_;
-        } else {
-          result.postOptimizationGraph_ = postOptimizationGraphBuilder_.build();
-        }
-        onBuilt();
-        return result;
-      }
-
-      @java.lang.Override
-      public Builder clone() {
-        return super.clone();
-      }
-      @java.lang.Override
-      public Builder setField(
-          com.google.protobuf.Descriptors.FieldDescriptor field,
-          java.lang.Object value) {
-        return super.setField(field, value);
-      }
-      @java.lang.Override
-      public Builder clearField(
-          com.google.protobuf.Descriptors.FieldDescriptor field) {
-        return super.clearField(field);
-      }
-      @java.lang.Override
-      public Builder clearOneof(
-          com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-        return super.clearOneof(oneof);
-      }
-      @java.lang.Override
-      public Builder setRepeatedField(
-          com.google.protobuf.Descriptors.FieldDescriptor field,
-          int index, java.lang.Object value) {
-        return super.setRepeatedField(field, index, value);
-      }
-      @java.lang.Override
-      public Builder addRepeatedField(
-          com.google.protobuf.Descriptors.FieldDescriptor field,
-          java.lang.Object value) {
-        return super.addRepeatedField(field, value);
-      }
-      @java.lang.Override
-      public Builder mergeFrom(com.google.protobuf.Message other) {
-        if (other instanceof org.tensorflow.proto.framework.RunMetadata.FunctionGraphs) {
-          return mergeFrom((org.tensorflow.proto.framework.RunMetadata.FunctionGraphs)other);
-        } else {
-          super.mergeFrom(other);
-          return this;
-        }
-      }
-
-      public Builder mergeFrom(org.tensorflow.proto.framework.RunMetadata.FunctionGraphs other) {
-        if (other == org.tensorflow.proto.framework.RunMetadata.FunctionGraphs.getDefaultInstance()) return this;
-        if (partitionGraphsBuilder_ == null) {
-          if (!other.partitionGraphs_.isEmpty()) {
-            if (partitionGraphs_.isEmpty()) {
-              partitionGraphs_ = other.partitionGraphs_;
-              bitField0_ = (bitField0_ & ~0x00000001);
-            } else {
-              ensurePartitionGraphsIsMutable();
-              partitionGraphs_.addAll(other.partitionGraphs_);
-            }
-            onChanged();
-          }
-        } else {
-          if (!other.partitionGraphs_.isEmpty()) {
-            if (partitionGraphsBuilder_.isEmpty()) {
-              partitionGraphsBuilder_.dispose();
-              partitionGraphsBuilder_ = null;
-              partitionGraphs_ = other.partitionGraphs_;
-              bitField0_ = (bitField0_ & ~0x00000001);
-              partitionGraphsBuilder_ = 
-                com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
-                   getPartitionGraphsFieldBuilder() : null;
-            } else {
-              partitionGraphsBuilder_.addAllMessages(other.partitionGraphs_);
-            }
-          }
-        }
-        if (other.hasPreOptimizationGraph()) {
-          mergePreOptimizationGraph(other.getPreOptimizationGraph());
-        }
-        if (other.hasPostOptimizationGraph()) {
-          mergePostOptimizationGraph(other.getPostOptimizationGraph());
-        }
-        this.mergeUnknownFields(other.unknownFields);
-        onChanged();
-        return this;
-      }
-
-      @java.lang.Override
-      public final boolean isInitialized() {
-        return true;
-      }
-
-      @java.lang.Override
-      public Builder mergeFrom(
-          com.google.protobuf.CodedInputStream input,
-          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-          throws java.io.IOException {
-        org.tensorflow.proto.framework.RunMetadata.FunctionGraphs parsedMessage = null;
-        try {
-          parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-        } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-          parsedMessage = (org.tensorflow.proto.framework.RunMetadata.FunctionGraphs) e.getUnfinishedMessage();
-          throw e.unwrapIOException();
-        } finally {
-          if (parsedMessage != null) {
-            mergeFrom(parsedMessage);
-          }
-        }
-        return this;
-      }
-      private int bitField0_;
-
-      private java.util.List<org.tensorflow.proto.framework.GraphDef> partitionGraphs_ =
-        java.util.Collections.emptyList();
-      private void ensurePartitionGraphsIsMutable() {
-        if (!((bitField0_ & 0x00000001) != 0)) {
-          partitionGraphs_ = new java.util.ArrayList<org.tensorflow.proto.framework.GraphDef>(partitionGraphs_);
-          bitField0_ |= 0x00000001;
-         }
-      }
-
-      private com.google.protobuf.RepeatedFieldBuilderV3<
-          org.tensorflow.proto.framework.GraphDef, org.tensorflow.proto.framework.GraphDef.Builder, org.tensorflow.proto.framework.GraphDefOrBuilder> partitionGraphsBuilder_;
-
-      /**
-       * <pre>
-       * TODO(nareshmodi): Include some sort of function/cache-key identifier?
-       * </pre>
-       *
-       * <code>repeated .tensorflow.GraphDef partition_graphs = 1;</code>
-       */
-      public java.util.List<org.tensorflow.proto.framework.GraphDef> getPartitionGraphsList() {
-        if (partitionGraphsBuilder_ == null) {
-          return java.util.Collections.unmodifiableList(partitionGraphs_);
-        } else {
-          return partitionGraphsBuilder_.getMessageList();
-        }
-      }
-      /**
-       * <pre>
-       * TODO(nareshmodi): Include some sort of function/cache-key identifier?
-       * </pre>
-       *
-       * <code>repeated .tensorflow.GraphDef partition_graphs = 1;</code>
-       */
-      public int getPartitionGraphsCount() {
-        if (partitionGraphsBuilder_ == null) {
-          return partitionGraphs_.size();
-        } else {
-          return partitionGraphsBuilder_.getCount();
-        }
-      }
-      /**
-       * <pre>
-       * TODO(nareshmodi): Include some sort of function/cache-key identifier?
-       * </pre>
-       *
-       * <code>repeated .tensorflow.GraphDef partition_graphs = 1;</code>
-       */
-      public org.tensorflow.proto.framework.GraphDef getPartitionGraphs(int index) {
-        if (partitionGraphsBuilder_ == null) {
-          return partitionGraphs_.get(index);
-        } else {
-          return partitionGraphsBuilder_.getMessage(index);
-        }
-      }
-      /**
-       * <pre>
-       * TODO(nareshmodi): Include some sort of function/cache-key identifier?
-       * </pre>
-       *
-       * <code>repeated .tensorflow.GraphDef partition_graphs = 1;</code>
-       */
-      public Builder setPartitionGraphs(
-          int index, org.tensorflow.proto.framework.GraphDef value) {
-        if (partitionGraphsBuilder_ == null) {
-          if (value == null) {
-            throw new NullPointerException();
-          }
-          ensurePartitionGraphsIsMutable();
-          partitionGraphs_.set(index, value);
-          onChanged();
-        } else {
-          partitionGraphsBuilder_.setMessage(index, value);
-        }
-        return this;
-      }
-      /**
-       * <pre>
-       * TODO(nareshmodi): Include some sort of function/cache-key identifier?
-       * </pre>
-       *
-       * <code>repeated .tensorflow.GraphDef partition_graphs = 1;</code>
-       */
-      public Builder setPartitionGraphs(
-          int index, org.tensorflow.proto.framework.GraphDef.Builder builderForValue) {
-        if (partitionGraphsBuilder_ == null) {
-          ensurePartitionGraphsIsMutable();
-          partitionGraphs_.set(index, builderForValue.build());
-          onChanged();
-        } else {
-          partitionGraphsBuilder_.setMessage(index, builderForValue.build());
-        }
-        return this;
-      }
-      /**
-       * <pre>
-       * TODO(nareshmodi): Include some sort of function/cache-key identifier?
-       * </pre>
-       *
-       * <code>repeated .tensorflow.GraphDef partition_graphs = 1;</code>
-       */
-      public Builder addPartitionGraphs(org.tensorflow.proto.framework.GraphDef value) {
-        if (partitionGraphsBuilder_ == null) {
-          if (value == null) {
-            throw new NullPointerException();
-          }
-          ensurePartitionGraphsIsMutable();
-          partitionGraphs_.add(value);
-          onChanged();
-        } else {
-          partitionGraphsBuilder_.addMessage(value);
-        }
-        return this;
-      }
-      /**
-       * <pre>
-       * TODO(nareshmodi): Include some sort of function/cache-key identifier?
-       * </pre>
-       *
-       * <code>repeated .tensorflow.GraphDef partition_graphs = 1;</code>
-       */
-      public Builder addPartitionGraphs(
-          int index, org.tensorflow.proto.framework.GraphDef value) {
-        if (partitionGraphsBuilder_ == null) {
-          if (value == null) {
-            throw new NullPointerException();
-          }
-          ensurePartitionGraphsIsMutable();
-          partitionGraphs_.add(index, value);
-          onChanged();
-        } else {
-          partitionGraphsBuilder_.addMessage(index, value);
-        }
-        return this;
-      }
-      /**
-       * <pre>
-       * TODO(nareshmodi): Include some sort of function/cache-key identifier?
-       * </pre>
-       *
-       * <code>repeated .tensorflow.GraphDef partition_graphs = 1;</code>
-       */
-      public Builder addPartitionGraphs(
-          org.tensorflow.proto.framework.GraphDef.Builder builderForValue) {
-        if (partitionGraphsBuilder_ == null) {
-          ensurePartitionGraphsIsMutable();
-          partitionGraphs_.add(builderForValue.build());
-          onChanged();
-        } else {
-          partitionGraphsBuilder_.addMessage(builderForValue.build());
-        }
-        return this;
-      }
-      /**
-       * <pre>
-       * TODO(nareshmodi): Include some sort of function/cache-key identifier?
-       * </pre>
-       *
-       * <code>repeated .tensorflow.GraphDef partition_graphs = 1;</code>
-       */
-      public Builder addPartitionGraphs(
-          int index, org.tensorflow.proto.framework.GraphDef.Builder builderForValue) {
-        if (partitionGraphsBuilder_ == null) {
-          ensurePartitionGraphsIsMutable();
-          partitionGraphs_.add(index, builderForValue.build());
-          onChanged();
-        } else {
-          partitionGraphsBuilder_.addMessage(index, builderForValue.build());
-        }
-        return this;
-      }
-      /**
-       * <pre>
-       * TODO(nareshmodi): Include some sort of function/cache-key identifier?
-       * </pre>
-       *
-       * <code>repeated .tensorflow.GraphDef partition_graphs = 1;</code>
-       */
-      public Builder addAllPartitionGraphs(
-          java.lang.Iterable<? extends org.tensorflow.proto.framework.GraphDef> values) {
-        if (partitionGraphsBuilder_ == null) {
-          ensurePartitionGraphsIsMutable();
-          com.google.protobuf.AbstractMessageLite.Builder.addAll(
-              values, partitionGraphs_);
-          onChanged();
-        } else {
-          partitionGraphsBuilder_.addAllMessages(values);
-        }
-        return this;
-      }
-      /**
-       * <pre>
-       * TODO(nareshmodi): Include some sort of function/cache-key identifier?
-       * </pre>
-       *
-       * <code>repeated .tensorflow.GraphDef partition_graphs = 1;</code>
-       */
-      public Builder clearPartitionGraphs() {
-        if (partitionGraphsBuilder_ == null) {
-          partitionGraphs_ = java.util.Collections.emptyList();
-          bitField0_ = (bitField0_ & ~0x00000001);
-          onChanged();
-        } else {
-          partitionGraphsBuilder_.clear();
-        }
-        return this;
-      }
-      /**
-       * <pre>
-       * TODO(nareshmodi): Include some sort of function/cache-key identifier?
-       * </pre>
-       *
-       * <code>repeated .tensorflow.GraphDef partition_graphs = 1;</code>
-       */
-      public Builder removePartitionGraphs(int index) {
-        if (partitionGraphsBuilder_ == null) {
-          ensurePartitionGraphsIsMutable();
-          partitionGraphs_.remove(index);
-          onChanged();
-        } else {
-          partitionGraphsBuilder_.remove(index);
-        }
-        return this;
-      }
-      /**
-       * <pre>
-       * TODO(nareshmodi): Include some sort of function/cache-key identifier?
-       * </pre>
-       *
-       * <code>repeated .tensorflow.GraphDef partition_graphs = 1;</code>
-       */
-      public org.tensorflow.proto.framework.GraphDef.Builder getPartitionGraphsBuilder(
-          int index) {
-        return getPartitionGraphsFieldBuilder().getBuilder(index);
-      }
-      /**
-       * <pre>
-       * TODO(nareshmodi): Include some sort of function/cache-key identifier?
-       * </pre>
-       *
-       * <code>repeated .tensorflow.GraphDef partition_graphs = 1;</code>
-       */
-      public org.tensorflow.proto.framework.GraphDefOrBuilder getPartitionGraphsOrBuilder(
-          int index) {
-        if (partitionGraphsBuilder_ == null) {
-          return partitionGraphs_.get(index);  } else {
-          return partitionGraphsBuilder_.getMessageOrBuilder(index);
-        }
-      }
-      /**
-       * <pre>
-       * TODO(nareshmodi): Include some sort of function/cache-key identifier?
-       * </pre>
-       *
-       * <code>repeated .tensorflow.GraphDef partition_graphs = 1;</code>
-       */
-      public java.util.List<? extends org.tensorflow.proto.framework.GraphDefOrBuilder> 
-           getPartitionGraphsOrBuilderList() {
-        if (partitionGraphsBuilder_ != null) {
-          return partitionGraphsBuilder_.getMessageOrBuilderList();
-        } else {
-          return java.util.Collections.unmodifiableList(partitionGraphs_);
-        }
-      }
-      /**
-       * <pre>
-       * TODO(nareshmodi): Include some sort of function/cache-key identifier?
-       * </pre>
-       *
-       * <code>repeated .tensorflow.GraphDef partition_graphs = 1;</code>
-       */
-      public org.tensorflow.proto.framework.GraphDef.Builder addPartitionGraphsBuilder() {
-        return getPartitionGraphsFieldBuilder().addBuilder(
-            org.tensorflow.proto.framework.GraphDef.getDefaultInstance());
-      }
-      /**
-       * <pre>
-       * TODO(nareshmodi): Include some sort of function/cache-key identifier?
-       * </pre>
-       *
-       * <code>repeated .tensorflow.GraphDef partition_graphs = 1;</code>
-       */
-      public org.tensorflow.proto.framework.GraphDef.Builder addPartitionGraphsBuilder(
-          int index) {
-        return getPartitionGraphsFieldBuilder().addBuilder(
-            index, org.tensorflow.proto.framework.GraphDef.getDefaultInstance());
-      }
-      /**
-       * <pre>
-       * TODO(nareshmodi): Include some sort of function/cache-key identifier?
-       * </pre>
-       *
-       * <code>repeated .tensorflow.GraphDef partition_graphs = 1;</code>
-       */
-      public java.util.List<org.tensorflow.proto.framework.GraphDef.Builder> 
-           getPartitionGraphsBuilderList() {
-        return getPartitionGraphsFieldBuilder().getBuilderList();
-      }
-      private com.google.protobuf.RepeatedFieldBuilderV3<
-          org.tensorflow.proto.framework.GraphDef, org.tensorflow.proto.framework.GraphDef.Builder, org.tensorflow.proto.framework.GraphDefOrBuilder> 
-          getPartitionGraphsFieldBuilder() {
-        if (partitionGraphsBuilder_ == null) {
-          partitionGraphsBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
-              org.tensorflow.proto.framework.GraphDef, org.tensorflow.proto.framework.GraphDef.Builder, org.tensorflow.proto.framework.GraphDefOrBuilder>(
-                  partitionGraphs_,
-                  ((bitField0_ & 0x00000001) != 0),
-                  getParentForChildren(),
-                  isClean());
-          partitionGraphs_ = null;
-        }
-        return partitionGraphsBuilder_;
-      }
-
-      private org.tensorflow.proto.framework.GraphDef preOptimizationGraph_;
-      private com.google.protobuf.SingleFieldBuilderV3<
-          org.tensorflow.proto.framework.GraphDef, org.tensorflow.proto.framework.GraphDef.Builder, org.tensorflow.proto.framework.GraphDefOrBuilder> preOptimizationGraphBuilder_;
-      /**
-       * <code>.tensorflow.GraphDef pre_optimization_graph = 2;</code>
-       */
-      public boolean hasPreOptimizationGraph() {
-        return preOptimizationGraphBuilder_ != null || preOptimizationGraph_ != null;
-      }
-      /**
-       * <code>.tensorflow.GraphDef pre_optimization_graph = 2;</code>
-       */
-      public org.tensorflow.proto.framework.GraphDef getPreOptimizationGraph() {
-        if (preOptimizationGraphBuilder_ == null) {
-          return preOptimizationGraph_ == null ? org.tensorflow.proto.framework.GraphDef.getDefaultInstance() : preOptimizationGraph_;
-        } else {
-          return preOptimizationGraphBuilder_.getMessage();
-        }
-      }
-      /**
-       * <code>.tensorflow.GraphDef pre_optimization_graph = 2;</code>
-       */
-      public Builder setPreOptimizationGraph(org.tensorflow.proto.framework.GraphDef value) {
-        if (preOptimizationGraphBuilder_ == null) {
-          if (value == null) {
-            throw new NullPointerException();
-          }
-          preOptimizationGraph_ = value;
-          onChanged();
-        } else {
-          preOptimizationGraphBuilder_.setMessage(value);
-        }
-
-        return this;
-      }
-      /**
-       * <code>.tensorflow.GraphDef pre_optimization_graph = 2;</code>
-       */
-      public Builder setPreOptimizationGraph(
-          org.tensorflow.proto.framework.GraphDef.Builder builderForValue) {
-        if (preOptimizationGraphBuilder_ == null) {
-          preOptimizationGraph_ = builderForValue.build();
-          onChanged();
-        } else {
-          preOptimizationGraphBuilder_.setMessage(builderForValue.build());
-        }
-
-        return this;
-      }
-      /**
-       * <code>.tensorflow.GraphDef pre_optimization_graph = 2;</code>
-       */
-      public Builder mergePreOptimizationGraph(org.tensorflow.proto.framework.GraphDef value) {
-        if (preOptimizationGraphBuilder_ == null) {
-          if (preOptimizationGraph_ != null) {
-            preOptimizationGraph_ =
-              org.tensorflow.proto.framework.GraphDef.newBuilder(preOptimizationGraph_).mergeFrom(value).buildPartial();
-          } else {
-            preOptimizationGraph_ = value;
-          }
-          onChanged();
-        } else {
-          preOptimizationGraphBuilder_.mergeFrom(value);
-        }
-
-        return this;
-      }
-      /**
-       * <code>.tensorflow.GraphDef pre_optimization_graph = 2;</code>
-       */
-      public Builder clearPreOptimizationGraph() {
-        if (preOptimizationGraphBuilder_ == null) {
-          preOptimizationGraph_ = null;
-          onChanged();
-        } else {
-          preOptimizationGraph_ = null;
-          preOptimizationGraphBuilder_ = null;
-        }
-
-        return this;
-      }
-      /**
-       * <code>.tensorflow.GraphDef pre_optimization_graph = 2;</code>
-       */
-      public org.tensorflow.proto.framework.GraphDef.Builder getPreOptimizationGraphBuilder() {
-        
-        onChanged();
-        return getPreOptimizationGraphFieldBuilder().getBuilder();
-      }
-      /**
-       * <code>.tensorflow.GraphDef pre_optimization_graph = 2;</code>
-       */
-      public org.tensorflow.proto.framework.GraphDefOrBuilder getPreOptimizationGraphOrBuilder() {
-        if (preOptimizationGraphBuilder_ != null) {
-          return preOptimizationGraphBuilder_.getMessageOrBuilder();
-        } else {
-          return preOptimizationGraph_ == null ?
-              org.tensorflow.proto.framework.GraphDef.getDefaultInstance() : preOptimizationGraph_;
-        }
-      }
-      /**
-       * <code>.tensorflow.GraphDef pre_optimization_graph = 2;</code>
-       */
-      private com.google.protobuf.SingleFieldBuilderV3<
-          org.tensorflow.proto.framework.GraphDef, org.tensorflow.proto.framework.GraphDef.Builder, org.tensorflow.proto.framework.GraphDefOrBuilder> 
-          getPreOptimizationGraphFieldBuilder() {
-        if (preOptimizationGraphBuilder_ == null) {
-          preOptimizationGraphBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-              org.tensorflow.proto.framework.GraphDef, org.tensorflow.proto.framework.GraphDef.Builder, org.tensorflow.proto.framework.GraphDefOrBuilder>(
-                  getPreOptimizationGraph(),
-                  getParentForChildren(),
-                  isClean());
-          preOptimizationGraph_ = null;
-        }
-        return preOptimizationGraphBuilder_;
-      }
-
-      private org.tensorflow.proto.framework.GraphDef postOptimizationGraph_;
-      private com.google.protobuf.SingleFieldBuilderV3<
-          org.tensorflow.proto.framework.GraphDef, org.tensorflow.proto.framework.GraphDef.Builder, org.tensorflow.proto.framework.GraphDefOrBuilder> postOptimizationGraphBuilder_;
-      /**
-       * <code>.tensorflow.GraphDef post_optimization_graph = 3;</code>
-       */
-      public boolean hasPostOptimizationGraph() {
-        return postOptimizationGraphBuilder_ != null || postOptimizationGraph_ != null;
-      }
-      /**
-       * <code>.tensorflow.GraphDef post_optimization_graph = 3;</code>
-       */
-      public org.tensorflow.proto.framework.GraphDef getPostOptimizationGraph() {
-        if (postOptimizationGraphBuilder_ == null) {
-          return postOptimizationGraph_ == null ? org.tensorflow.proto.framework.GraphDef.getDefaultInstance() : postOptimizationGraph_;
-        } else {
-          return postOptimizationGraphBuilder_.getMessage();
-        }
-      }
-      /**
-       * <code>.tensorflow.GraphDef post_optimization_graph = 3;</code>
-       */
-      public Builder setPostOptimizationGraph(org.tensorflow.proto.framework.GraphDef value) {
-        if (postOptimizationGraphBuilder_ == null) {
-          if (value == null) {
-            throw new NullPointerException();
-          }
-          postOptimizationGraph_ = value;
-          onChanged();
-        } else {
-          postOptimizationGraphBuilder_.setMessage(value);
-        }
-
-        return this;
-      }
-      /**
-       * <code>.tensorflow.GraphDef post_optimization_graph = 3;</code>
-       */
-      public Builder setPostOptimizationGraph(
-          org.tensorflow.proto.framework.GraphDef.Builder builderForValue) {
-        if (postOptimizationGraphBuilder_ == null) {
-          postOptimizationGraph_ = builderForValue.build();
-          onChanged();
-        } else {
-          postOptimizationGraphBuilder_.setMessage(builderForValue.build());
-        }
-
-        return this;
-      }
-      /**
-       * <code>.tensorflow.GraphDef post_optimization_graph = 3;</code>
-       */
-      public Builder mergePostOptimizationGraph(org.tensorflow.proto.framework.GraphDef value) {
-        if (postOptimizationGraphBuilder_ == null) {
-          if (postOptimizationGraph_ != null) {
-            postOptimizationGraph_ =
-              org.tensorflow.proto.framework.GraphDef.newBuilder(postOptimizationGraph_).mergeFrom(value).buildPartial();
-          } else {
-            postOptimizationGraph_ = value;
-          }
-          onChanged();
-        } else {
-          postOptimizationGraphBuilder_.mergeFrom(value);
-        }
-
-        return this;
-      }
-      /**
-       * <code>.tensorflow.GraphDef post_optimization_graph = 3;</code>
-       */
-      public Builder clearPostOptimizationGraph() {
-        if (postOptimizationGraphBuilder_ == null) {
-          postOptimizationGraph_ = null;
-          onChanged();
-        } else {
-          postOptimizationGraph_ = null;
-          postOptimizationGraphBuilder_ = null;
-        }
-
-        return this;
-      }
-      /**
-       * <code>.tensorflow.GraphDef post_optimization_graph = 3;</code>
-       */
-      public org.tensorflow.proto.framework.GraphDef.Builder getPostOptimizationGraphBuilder() {
-        
-        onChanged();
-        return getPostOptimizationGraphFieldBuilder().getBuilder();
-      }
-      /**
-       * <code>.tensorflow.GraphDef post_optimization_graph = 3;</code>
-       */
-      public org.tensorflow.proto.framework.GraphDefOrBuilder getPostOptimizationGraphOrBuilder() {
-        if (postOptimizationGraphBuilder_ != null) {
-          return postOptimizationGraphBuilder_.getMessageOrBuilder();
-        } else {
-          return postOptimizationGraph_ == null ?
-              org.tensorflow.proto.framework.GraphDef.getDefaultInstance() : postOptimizationGraph_;
-        }
-      }
-      /**
-       * <code>.tensorflow.GraphDef post_optimization_graph = 3;</code>
-       */
-      private com.google.protobuf.SingleFieldBuilderV3<
-          org.tensorflow.proto.framework.GraphDef, org.tensorflow.proto.framework.GraphDef.Builder, org.tensorflow.proto.framework.GraphDefOrBuilder> 
-          getPostOptimizationGraphFieldBuilder() {
-        if (postOptimizationGraphBuilder_ == null) {
-          postOptimizationGraphBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-              org.tensorflow.proto.framework.GraphDef, org.tensorflow.proto.framework.GraphDef.Builder, org.tensorflow.proto.framework.GraphDefOrBuilder>(
-                  getPostOptimizationGraph(),
-                  getParentForChildren(),
-                  isClean());
-          postOptimizationGraph_ = null;
-        }
-        return postOptimizationGraphBuilder_;
-      }
-      @java.lang.Override
-      public final Builder setUnknownFields(
-          final com.google.protobuf.UnknownFieldSet unknownFields) {
-        return super.setUnknownFields(unknownFields);
-      }
-
-      @java.lang.Override
-      public final Builder mergeUnknownFields(
-          final com.google.protobuf.UnknownFieldSet unknownFields) {
-        return super.mergeUnknownFields(unknownFields);
-      }
-
-
-      // @@protoc_insertion_point(builder_scope:tensorflow.RunMetadata.FunctionGraphs)
-    }
-
-    // @@protoc_insertion_point(class_scope:tensorflow.RunMetadata.FunctionGraphs)
-    private static final org.tensorflow.proto.framework.RunMetadata.FunctionGraphs DEFAULT_INSTANCE;
-    static {
-      DEFAULT_INSTANCE = new org.tensorflow.proto.framework.RunMetadata.FunctionGraphs();
-    }
-
-    public static org.tensorflow.proto.framework.RunMetadata.FunctionGraphs getDefaultInstance() {
-      return DEFAULT_INSTANCE;
-    }
-
-    private static final com.google.protobuf.Parser<FunctionGraphs>
-        PARSER = new com.google.protobuf.AbstractParser<FunctionGraphs>() {
-      @java.lang.Override
-      public FunctionGraphs parsePartialFrom(
-          com.google.protobuf.CodedInputStream input,
-          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-          throws com.google.protobuf.InvalidProtocolBufferException {
-        return new FunctionGraphs(input, extensionRegistry);
-      }
-    };
-
-    public static com.google.protobuf.Parser<FunctionGraphs> parser() {
-      return PARSER;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Parser<FunctionGraphs> getParserForType() {
-      return PARSER;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.RunMetadata.FunctionGraphs getDefaultInstanceForType() {
-      return DEFAULT_INSTANCE;
-    }
-
-  }
-
-  public static final int STEP_STATS_FIELD_NUMBER = 1;
-  private org.tensorflow.proto.framework.StepStats stepStats_;
-  /**
-   * <pre>
-   * Statistics traced for this step. Populated if tracing is turned on via the
-   * "RunOptions" proto.
-   * EXPERIMENTAL: The format and set of events may change in future versions.
-   * </pre>
-   *
-   * <code>.tensorflow.StepStats step_stats = 1;</code>
-   */
-  public boolean hasStepStats() {
-    return stepStats_ != null;
-  }
-  /**
-   * <pre>
-   * Statistics traced for this step. Populated if tracing is turned on via the
-   * "RunOptions" proto.
-   * EXPERIMENTAL: The format and set of events may change in future versions.
-   * </pre>
-   *
-   * <code>.tensorflow.StepStats step_stats = 1;</code>
-   */
-  public org.tensorflow.proto.framework.StepStats getStepStats() {
-    return stepStats_ == null ? org.tensorflow.proto.framework.StepStats.getDefaultInstance() : stepStats_;
-  }
-  /**
-   * <pre>
-   * Statistics traced for this step. Populated if tracing is turned on via the
-   * "RunOptions" proto.
-   * EXPERIMENTAL: The format and set of events may change in future versions.
-   * </pre>
-   *
-   * <code>.tensorflow.StepStats step_stats = 1;</code>
-   */
-  public org.tensorflow.proto.framework.StepStatsOrBuilder getStepStatsOrBuilder() {
-    return getStepStats();
-  }
-
-  public static final int COST_GRAPH_FIELD_NUMBER = 2;
-  private org.tensorflow.proto.framework.CostGraphDef costGraph_;
-  /**
-   * <pre>
-   * The cost graph for the computation defined by the run call.
-   * </pre>
-   *
-   * <code>.tensorflow.CostGraphDef cost_graph = 2;</code>
-   */
-  public boolean hasCostGraph() {
-    return costGraph_ != null;
-  }
-  /**
-   * <pre>
-   * The cost graph for the computation defined by the run call.
-   * </pre>
-   *
-   * <code>.tensorflow.CostGraphDef cost_graph = 2;</code>
-   */
-  public org.tensorflow.proto.framework.CostGraphDef getCostGraph() {
-    return costGraph_ == null ? org.tensorflow.proto.framework.CostGraphDef.getDefaultInstance() : costGraph_;
-  }
-  /**
-   * <pre>
-   * The cost graph for the computation defined by the run call.
-   * </pre>
-   *
-   * <code>.tensorflow.CostGraphDef cost_graph = 2;</code>
-   */
-  public org.tensorflow.proto.framework.CostGraphDefOrBuilder getCostGraphOrBuilder() {
-    return getCostGraph();
-  }
-
-  public static final int PARTITION_GRAPHS_FIELD_NUMBER = 3;
-  private java.util.List<org.tensorflow.proto.framework.GraphDef> partitionGraphs_;
-  /**
-   * <pre>
-   * Graphs of the partitions executed by executors.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.GraphDef partition_graphs = 3;</code>
-   */
-  public java.util.List<org.tensorflow.proto.framework.GraphDef> getPartitionGraphsList() {
-    return partitionGraphs_;
-  }
-  /**
-   * <pre>
-   * Graphs of the partitions executed by executors.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.GraphDef partition_graphs = 3;</code>
-   */
-  public java.util.List<? extends org.tensorflow.proto.framework.GraphDefOrBuilder> 
-      getPartitionGraphsOrBuilderList() {
-    return partitionGraphs_;
-  }
-  /**
-   * <pre>
-   * Graphs of the partitions executed by executors.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.GraphDef partition_graphs = 3;</code>
-   */
-  public int getPartitionGraphsCount() {
-    return partitionGraphs_.size();
-  }
-  /**
-   * <pre>
-   * Graphs of the partitions executed by executors.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.GraphDef partition_graphs = 3;</code>
-   */
-  public org.tensorflow.proto.framework.GraphDef getPartitionGraphs(int index) {
-    return partitionGraphs_.get(index);
-  }
-  /**
-   * <pre>
-   * Graphs of the partitions executed by executors.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.GraphDef partition_graphs = 3;</code>
-   */
-  public org.tensorflow.proto.framework.GraphDefOrBuilder getPartitionGraphsOrBuilder(
-      int index) {
-    return partitionGraphs_.get(index);
-  }
-
-  public static final int FUNCTION_GRAPHS_FIELD_NUMBER = 4;
-  private java.util.List<org.tensorflow.proto.framework.RunMetadata.FunctionGraphs> functionGraphs_;
-  /**
-   * <pre>
-   * This is only populated for graphs that are run as functions in TensorFlow
-   * V2. There will be an entry below for each function that is traced.
-   * The main use cases of the post_optimization_graph and the partition_graphs
-   * is to give the caller insight into the graphs that were actually run by the
-   * runtime. Additional information (such as those in step_stats) will match
-   * these graphs.
-   * We also include the pre_optimization_graph since it is usually easier to
-   * read, and is helpful in situations where the caller wants to get a high
-   * level idea of what the built graph looks like (since the various graph
-   * optimization passes might change the structure of the graph significantly).
-   * </pre>
-   *
-   * <code>repeated .tensorflow.RunMetadata.FunctionGraphs function_graphs = 4;</code>
-   */
-  public java.util.List<org.tensorflow.proto.framework.RunMetadata.FunctionGraphs> getFunctionGraphsList() {
-    return functionGraphs_;
-  }
-  /**
-   * <pre>
-   * This is only populated for graphs that are run as functions in TensorFlow
-   * V2. There will be an entry below for each function that is traced.
-   * The main use cases of the post_optimization_graph and the partition_graphs
-   * is to give the caller insight into the graphs that were actually run by the
-   * runtime. Additional information (such as those in step_stats) will match
-   * these graphs.
-   * We also include the pre_optimization_graph since it is usually easier to
-   * read, and is helpful in situations where the caller wants to get a high
-   * level idea of what the built graph looks like (since the various graph
-   * optimization passes might change the structure of the graph significantly).
-   * </pre>
-   *
-   * <code>repeated .tensorflow.RunMetadata.FunctionGraphs function_graphs = 4;</code>
-   */
-  public java.util.List<? extends org.tensorflow.proto.framework.RunMetadata.FunctionGraphsOrBuilder> 
-      getFunctionGraphsOrBuilderList() {
-    return functionGraphs_;
-  }
-  /**
-   * <pre>
-   * This is only populated for graphs that are run as functions in TensorFlow
-   * V2. There will be an entry below for each function that is traced.
-   * The main use cases of the post_optimization_graph and the partition_graphs
-   * is to give the caller insight into the graphs that were actually run by the
-   * runtime. Additional information (such as those in step_stats) will match
-   * these graphs.
-   * We also include the pre_optimization_graph since it is usually easier to
-   * read, and is helpful in situations where the caller wants to get a high
-   * level idea of what the built graph looks like (since the various graph
-   * optimization passes might change the structure of the graph significantly).
-   * </pre>
-   *
-   * <code>repeated .tensorflow.RunMetadata.FunctionGraphs function_graphs = 4;</code>
-   */
-  public int getFunctionGraphsCount() {
-    return functionGraphs_.size();
-  }
-  /**
-   * <pre>
-   * This is only populated for graphs that are run as functions in TensorFlow
-   * V2. There will be an entry below for each function that is traced.
-   * The main use cases of the post_optimization_graph and the partition_graphs
-   * is to give the caller insight into the graphs that were actually run by the
-   * runtime. Additional information (such as those in step_stats) will match
-   * these graphs.
-   * We also include the pre_optimization_graph since it is usually easier to
-   * read, and is helpful in situations where the caller wants to get a high
-   * level idea of what the built graph looks like (since the various graph
-   * optimization passes might change the structure of the graph significantly).
-   * </pre>
-   *
-   * <code>repeated .tensorflow.RunMetadata.FunctionGraphs function_graphs = 4;</code>
-   */
-  public org.tensorflow.proto.framework.RunMetadata.FunctionGraphs getFunctionGraphs(int index) {
-    return functionGraphs_.get(index);
-  }
-  /**
-   * <pre>
-   * This is only populated for graphs that are run as functions in TensorFlow
-   * V2. There will be an entry below for each function that is traced.
-   * The main use cases of the post_optimization_graph and the partition_graphs
-   * is to give the caller insight into the graphs that were actually run by the
-   * runtime. Additional information (such as those in step_stats) will match
-   * these graphs.
-   * We also include the pre_optimization_graph since it is usually easier to
-   * read, and is helpful in situations where the caller wants to get a high
-   * level idea of what the built graph looks like (since the various graph
-   * optimization passes might change the structure of the graph significantly).
-   * </pre>
-   *
-   * <code>repeated .tensorflow.RunMetadata.FunctionGraphs function_graphs = 4;</code>
-   */
-  public org.tensorflow.proto.framework.RunMetadata.FunctionGraphsOrBuilder getFunctionGraphsOrBuilder(
-      int index) {
-    return functionGraphs_.get(index);
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    if (stepStats_ != null) {
-      output.writeMessage(1, getStepStats());
-    }
-    if (costGraph_ != null) {
-      output.writeMessage(2, getCostGraph());
-    }
-    for (int i = 0; i < partitionGraphs_.size(); i++) {
-      output.writeMessage(3, partitionGraphs_.get(i));
-    }
-    for (int i = 0; i < functionGraphs_.size(); i++) {
-      output.writeMessage(4, functionGraphs_.get(i));
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (stepStats_ != null) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(1, getStepStats());
-    }
-    if (costGraph_ != null) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(2, getCostGraph());
-    }
-    for (int i = 0; i < partitionGraphs_.size(); i++) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(3, partitionGraphs_.get(i));
-    }
-    for (int i = 0; i < functionGraphs_.size(); i++) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(4, functionGraphs_.get(i));
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.framework.RunMetadata)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.framework.RunMetadata other = (org.tensorflow.proto.framework.RunMetadata) obj;
-
-    if (hasStepStats() != other.hasStepStats()) return false;
-    if (hasStepStats()) {
-      if (!getStepStats()
-          .equals(other.getStepStats())) return false;
-    }
-    if (hasCostGraph() != other.hasCostGraph()) return false;
-    if (hasCostGraph()) {
-      if (!getCostGraph()
-          .equals(other.getCostGraph())) return false;
-    }
-    if (!getPartitionGraphsList()
-        .equals(other.getPartitionGraphsList())) return false;
-    if (!getFunctionGraphsList()
-        .equals(other.getFunctionGraphsList())) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    if (hasStepStats()) {
-      hash = (37 * hash) + STEP_STATS_FIELD_NUMBER;
-      hash = (53 * hash) + getStepStats().hashCode();
-    }
-    if (hasCostGraph()) {
-      hash = (37 * hash) + COST_GRAPH_FIELD_NUMBER;
-      hash = (53 * hash) + getCostGraph().hashCode();
-    }
-    if (getPartitionGraphsCount() > 0) {
-      hash = (37 * hash) + PARTITION_GRAPHS_FIELD_NUMBER;
-      hash = (53 * hash) + getPartitionGraphsList().hashCode();
-    }
-    if (getFunctionGraphsCount() > 0) {
-      hash = (37 * hash) + FUNCTION_GRAPHS_FIELD_NUMBER;
-      hash = (53 * hash) + getFunctionGraphsList().hashCode();
-    }
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.framework.RunMetadata parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.RunMetadata parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.RunMetadata parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.RunMetadata parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.RunMetadata parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.RunMetadata parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.RunMetadata parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.RunMetadata parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.RunMetadata parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.RunMetadata parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.RunMetadata parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.RunMetadata parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.framework.RunMetadata prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * <pre>
-   * Metadata output (i.e., non-Tensor) for a single Run() call.
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.RunMetadata}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.RunMetadata)
-      org.tensorflow.proto.framework.RunMetadataOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_RunMetadata_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_RunMetadata_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.RunMetadata.class, org.tensorflow.proto.framework.RunMetadata.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.framework.RunMetadata.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-        getPartitionGraphsFieldBuilder();
-        getFunctionGraphsFieldBuilder();
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      if (stepStatsBuilder_ == null) {
-        stepStats_ = null;
-      } else {
-        stepStats_ = null;
-        stepStatsBuilder_ = null;
-      }
-      if (costGraphBuilder_ == null) {
-        costGraph_ = null;
-      } else {
-        costGraph_ = null;
-        costGraphBuilder_ = null;
-      }
-      if (partitionGraphsBuilder_ == null) {
-        partitionGraphs_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000001);
-      } else {
-        partitionGraphsBuilder_.clear();
-      }
-      if (functionGraphsBuilder_ == null) {
-        functionGraphs_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000002);
-      } else {
-        functionGraphsBuilder_.clear();
-      }
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_RunMetadata_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.RunMetadata getDefaultInstanceForType() {
-      return org.tensorflow.proto.framework.RunMetadata.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.RunMetadata build() {
-      org.tensorflow.proto.framework.RunMetadata result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.RunMetadata buildPartial() {
-      org.tensorflow.proto.framework.RunMetadata result = new org.tensorflow.proto.framework.RunMetadata(this);
-      int from_bitField0_ = bitField0_;
-      if (stepStatsBuilder_ == null) {
-        result.stepStats_ = stepStats_;
-      } else {
-        result.stepStats_ = stepStatsBuilder_.build();
-      }
-      if (costGraphBuilder_ == null) {
-        result.costGraph_ = costGraph_;
-      } else {
-        result.costGraph_ = costGraphBuilder_.build();
-      }
-      if (partitionGraphsBuilder_ == null) {
-        if (((bitField0_ & 0x00000001) != 0)) {
-          partitionGraphs_ = java.util.Collections.unmodifiableList(partitionGraphs_);
-          bitField0_ = (bitField0_ & ~0x00000001);
-        }
-        result.partitionGraphs_ = partitionGraphs_;
-      } else {
-        result.partitionGraphs_ = partitionGraphsBuilder_.build();
-      }
-      if (functionGraphsBuilder_ == null) {
-        if (((bitField0_ & 0x00000002) != 0)) {
-          functionGraphs_ = java.util.Collections.unmodifiableList(functionGraphs_);
-          bitField0_ = (bitField0_ & ~0x00000002);
-        }
-        result.functionGraphs_ = functionGraphs_;
-      } else {
-        result.functionGraphs_ = functionGraphsBuilder_.build();
-      }
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.framework.RunMetadata) {
-        return mergeFrom((org.tensorflow.proto.framework.RunMetadata)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.framework.RunMetadata other) {
-      if (other == org.tensorflow.proto.framework.RunMetadata.getDefaultInstance()) return this;
-      if (other.hasStepStats()) {
-        mergeStepStats(other.getStepStats());
-      }
-      if (other.hasCostGraph()) {
-        mergeCostGraph(other.getCostGraph());
-      }
-      if (partitionGraphsBuilder_ == null) {
-        if (!other.partitionGraphs_.isEmpty()) {
-          if (partitionGraphs_.isEmpty()) {
-            partitionGraphs_ = other.partitionGraphs_;
-            bitField0_ = (bitField0_ & ~0x00000001);
-          } else {
-            ensurePartitionGraphsIsMutable();
-            partitionGraphs_.addAll(other.partitionGraphs_);
-          }
-          onChanged();
-        }
-      } else {
-        if (!other.partitionGraphs_.isEmpty()) {
-          if (partitionGraphsBuilder_.isEmpty()) {
-            partitionGraphsBuilder_.dispose();
-            partitionGraphsBuilder_ = null;
-            partitionGraphs_ = other.partitionGraphs_;
-            bitField0_ = (bitField0_ & ~0x00000001);
-            partitionGraphsBuilder_ = 
-              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
-                 getPartitionGraphsFieldBuilder() : null;
-          } else {
-            partitionGraphsBuilder_.addAllMessages(other.partitionGraphs_);
-          }
-        }
-      }
-      if (functionGraphsBuilder_ == null) {
-        if (!other.functionGraphs_.isEmpty()) {
-          if (functionGraphs_.isEmpty()) {
-            functionGraphs_ = other.functionGraphs_;
-            bitField0_ = (bitField0_ & ~0x00000002);
-          } else {
-            ensureFunctionGraphsIsMutable();
-            functionGraphs_.addAll(other.functionGraphs_);
-          }
-          onChanged();
-        }
-      } else {
-        if (!other.functionGraphs_.isEmpty()) {
-          if (functionGraphsBuilder_.isEmpty()) {
-            functionGraphsBuilder_.dispose();
-            functionGraphsBuilder_ = null;
-            functionGraphs_ = other.functionGraphs_;
-            bitField0_ = (bitField0_ & ~0x00000002);
-            functionGraphsBuilder_ = 
-              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
-                 getFunctionGraphsFieldBuilder() : null;
-          } else {
-            functionGraphsBuilder_.addAllMessages(other.functionGraphs_);
-          }
-        }
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.framework.RunMetadata parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.framework.RunMetadata) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-    private int bitField0_;
-
-    private org.tensorflow.proto.framework.StepStats stepStats_;
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.StepStats, org.tensorflow.proto.framework.StepStats.Builder, org.tensorflow.proto.framework.StepStatsOrBuilder> stepStatsBuilder_;
-    /**
-     * <pre>
-     * Statistics traced for this step. Populated if tracing is turned on via the
-     * "RunOptions" proto.
-     * EXPERIMENTAL: The format and set of events may change in future versions.
-     * </pre>
-     *
-     * <code>.tensorflow.StepStats step_stats = 1;</code>
-     */
-    public boolean hasStepStats() {
-      return stepStatsBuilder_ != null || stepStats_ != null;
-    }
-    /**
-     * <pre>
-     * Statistics traced for this step. Populated if tracing is turned on via the
-     * "RunOptions" proto.
-     * EXPERIMENTAL: The format and set of events may change in future versions.
-     * </pre>
-     *
-     * <code>.tensorflow.StepStats step_stats = 1;</code>
-     */
-    public org.tensorflow.proto.framework.StepStats getStepStats() {
-      if (stepStatsBuilder_ == null) {
-        return stepStats_ == null ? org.tensorflow.proto.framework.StepStats.getDefaultInstance() : stepStats_;
-      } else {
-        return stepStatsBuilder_.getMessage();
-      }
-    }
-    /**
-     * <pre>
-     * Statistics traced for this step. Populated if tracing is turned on via the
-     * "RunOptions" proto.
-     * EXPERIMENTAL: The format and set of events may change in future versions.
-     * </pre>
-     *
-     * <code>.tensorflow.StepStats step_stats = 1;</code>
-     */
-    public Builder setStepStats(org.tensorflow.proto.framework.StepStats value) {
-      if (stepStatsBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        stepStats_ = value;
-        onChanged();
-      } else {
-        stepStatsBuilder_.setMessage(value);
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Statistics traced for this step. Populated if tracing is turned on via the
-     * "RunOptions" proto.
-     * EXPERIMENTAL: The format and set of events may change in future versions.
-     * </pre>
-     *
-     * <code>.tensorflow.StepStats step_stats = 1;</code>
-     */
-    public Builder setStepStats(
-        org.tensorflow.proto.framework.StepStats.Builder builderForValue) {
-      if (stepStatsBuilder_ == null) {
-        stepStats_ = builderForValue.build();
-        onChanged();
-      } else {
-        stepStatsBuilder_.setMessage(builderForValue.build());
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Statistics traced for this step. Populated if tracing is turned on via the
-     * "RunOptions" proto.
-     * EXPERIMENTAL: The format and set of events may change in future versions.
-     * </pre>
-     *
-     * <code>.tensorflow.StepStats step_stats = 1;</code>
-     */
-    public Builder mergeStepStats(org.tensorflow.proto.framework.StepStats value) {
-      if (stepStatsBuilder_ == null) {
-        if (stepStats_ != null) {
-          stepStats_ =
-            org.tensorflow.proto.framework.StepStats.newBuilder(stepStats_).mergeFrom(value).buildPartial();
-        } else {
-          stepStats_ = value;
-        }
-        onChanged();
-      } else {
-        stepStatsBuilder_.mergeFrom(value);
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Statistics traced for this step. Populated if tracing is turned on via the
-     * "RunOptions" proto.
-     * EXPERIMENTAL: The format and set of events may change in future versions.
-     * </pre>
-     *
-     * <code>.tensorflow.StepStats step_stats = 1;</code>
-     */
-    public Builder clearStepStats() {
-      if (stepStatsBuilder_ == null) {
-        stepStats_ = null;
-        onChanged();
-      } else {
-        stepStats_ = null;
-        stepStatsBuilder_ = null;
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Statistics traced for this step. Populated if tracing is turned on via the
-     * "RunOptions" proto.
-     * EXPERIMENTAL: The format and set of events may change in future versions.
-     * </pre>
-     *
-     * <code>.tensorflow.StepStats step_stats = 1;</code>
-     */
-    public org.tensorflow.proto.framework.StepStats.Builder getStepStatsBuilder() {
-      
-      onChanged();
-      return getStepStatsFieldBuilder().getBuilder();
-    }
-    /**
-     * <pre>
-     * Statistics traced for this step. Populated if tracing is turned on via the
-     * "RunOptions" proto.
-     * EXPERIMENTAL: The format and set of events may change in future versions.
-     * </pre>
-     *
-     * <code>.tensorflow.StepStats step_stats = 1;</code>
-     */
-    public org.tensorflow.proto.framework.StepStatsOrBuilder getStepStatsOrBuilder() {
-      if (stepStatsBuilder_ != null) {
-        return stepStatsBuilder_.getMessageOrBuilder();
-      } else {
-        return stepStats_ == null ?
-            org.tensorflow.proto.framework.StepStats.getDefaultInstance() : stepStats_;
-      }
-    }
-    /**
-     * <pre>
-     * Statistics traced for this step. Populated if tracing is turned on via the
-     * "RunOptions" proto.
-     * EXPERIMENTAL: The format and set of events may change in future versions.
-     * </pre>
-     *
-     * <code>.tensorflow.StepStats step_stats = 1;</code>
-     */
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.StepStats, org.tensorflow.proto.framework.StepStats.Builder, org.tensorflow.proto.framework.StepStatsOrBuilder> 
-        getStepStatsFieldBuilder() {
-      if (stepStatsBuilder_ == null) {
-        stepStatsBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-            org.tensorflow.proto.framework.StepStats, org.tensorflow.proto.framework.StepStats.Builder, org.tensorflow.proto.framework.StepStatsOrBuilder>(
-                getStepStats(),
-                getParentForChildren(),
-                isClean());
-        stepStats_ = null;
-      }
-      return stepStatsBuilder_;
-    }
-
-    private org.tensorflow.proto.framework.CostGraphDef costGraph_;
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.CostGraphDef, org.tensorflow.proto.framework.CostGraphDef.Builder, org.tensorflow.proto.framework.CostGraphDefOrBuilder> costGraphBuilder_;
-    /**
-     * <pre>
-     * The cost graph for the computation defined by the run call.
-     * </pre>
-     *
-     * <code>.tensorflow.CostGraphDef cost_graph = 2;</code>
-     */
-    public boolean hasCostGraph() {
-      return costGraphBuilder_ != null || costGraph_ != null;
-    }
-    /**
-     * <pre>
-     * The cost graph for the computation defined by the run call.
-     * </pre>
-     *
-     * <code>.tensorflow.CostGraphDef cost_graph = 2;</code>
-     */
-    public org.tensorflow.proto.framework.CostGraphDef getCostGraph() {
-      if (costGraphBuilder_ == null) {
-        return costGraph_ == null ? org.tensorflow.proto.framework.CostGraphDef.getDefaultInstance() : costGraph_;
-      } else {
-        return costGraphBuilder_.getMessage();
-      }
-    }
-    /**
-     * <pre>
-     * The cost graph for the computation defined by the run call.
-     * </pre>
-     *
-     * <code>.tensorflow.CostGraphDef cost_graph = 2;</code>
-     */
-    public Builder setCostGraph(org.tensorflow.proto.framework.CostGraphDef value) {
-      if (costGraphBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        costGraph_ = value;
-        onChanged();
-      } else {
-        costGraphBuilder_.setMessage(value);
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * The cost graph for the computation defined by the run call.
-     * </pre>
-     *
-     * <code>.tensorflow.CostGraphDef cost_graph = 2;</code>
-     */
-    public Builder setCostGraph(
-        org.tensorflow.proto.framework.CostGraphDef.Builder builderForValue) {
-      if (costGraphBuilder_ == null) {
-        costGraph_ = builderForValue.build();
-        onChanged();
-      } else {
-        costGraphBuilder_.setMessage(builderForValue.build());
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * The cost graph for the computation defined by the run call.
-     * </pre>
-     *
-     * <code>.tensorflow.CostGraphDef cost_graph = 2;</code>
-     */
-    public Builder mergeCostGraph(org.tensorflow.proto.framework.CostGraphDef value) {
-      if (costGraphBuilder_ == null) {
-        if (costGraph_ != null) {
-          costGraph_ =
-            org.tensorflow.proto.framework.CostGraphDef.newBuilder(costGraph_).mergeFrom(value).buildPartial();
-        } else {
-          costGraph_ = value;
-        }
-        onChanged();
-      } else {
-        costGraphBuilder_.mergeFrom(value);
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * The cost graph for the computation defined by the run call.
-     * </pre>
-     *
-     * <code>.tensorflow.CostGraphDef cost_graph = 2;</code>
-     */
-    public Builder clearCostGraph() {
-      if (costGraphBuilder_ == null) {
-        costGraph_ = null;
-        onChanged();
-      } else {
-        costGraph_ = null;
-        costGraphBuilder_ = null;
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * The cost graph for the computation defined by the run call.
-     * </pre>
-     *
-     * <code>.tensorflow.CostGraphDef cost_graph = 2;</code>
-     */
-    public org.tensorflow.proto.framework.CostGraphDef.Builder getCostGraphBuilder() {
-      
-      onChanged();
-      return getCostGraphFieldBuilder().getBuilder();
-    }
-    /**
-     * <pre>
-     * The cost graph for the computation defined by the run call.
-     * </pre>
-     *
-     * <code>.tensorflow.CostGraphDef cost_graph = 2;</code>
-     */
-    public org.tensorflow.proto.framework.CostGraphDefOrBuilder getCostGraphOrBuilder() {
-      if (costGraphBuilder_ != null) {
-        return costGraphBuilder_.getMessageOrBuilder();
-      } else {
-        return costGraph_ == null ?
-            org.tensorflow.proto.framework.CostGraphDef.getDefaultInstance() : costGraph_;
-      }
-    }
-    /**
-     * <pre>
-     * The cost graph for the computation defined by the run call.
-     * </pre>
-     *
-     * <code>.tensorflow.CostGraphDef cost_graph = 2;</code>
-     */
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.CostGraphDef, org.tensorflow.proto.framework.CostGraphDef.Builder, org.tensorflow.proto.framework.CostGraphDefOrBuilder> 
-        getCostGraphFieldBuilder() {
-      if (costGraphBuilder_ == null) {
-        costGraphBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-            org.tensorflow.proto.framework.CostGraphDef, org.tensorflow.proto.framework.CostGraphDef.Builder, org.tensorflow.proto.framework.CostGraphDefOrBuilder>(
-                getCostGraph(),
-                getParentForChildren(),
-                isClean());
-        costGraph_ = null;
-      }
-      return costGraphBuilder_;
-    }
-
-    private java.util.List<org.tensorflow.proto.framework.GraphDef> partitionGraphs_ =
-      java.util.Collections.emptyList();
-    private void ensurePartitionGraphsIsMutable() {
-      if (!((bitField0_ & 0x00000001) != 0)) {
-        partitionGraphs_ = new java.util.ArrayList<org.tensorflow.proto.framework.GraphDef>(partitionGraphs_);
-        bitField0_ |= 0x00000001;
-       }
-    }
-
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.framework.GraphDef, org.tensorflow.proto.framework.GraphDef.Builder, org.tensorflow.proto.framework.GraphDefOrBuilder> partitionGraphsBuilder_;
-
-    /**
-     * <pre>
-     * Graphs of the partitions executed by executors.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.GraphDef partition_graphs = 3;</code>
-     */
-    public java.util.List<org.tensorflow.proto.framework.GraphDef> getPartitionGraphsList() {
-      if (partitionGraphsBuilder_ == null) {
-        return java.util.Collections.unmodifiableList(partitionGraphs_);
-      } else {
-        return partitionGraphsBuilder_.getMessageList();
-      }
-    }
-    /**
-     * <pre>
-     * Graphs of the partitions executed by executors.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.GraphDef partition_graphs = 3;</code>
-     */
-    public int getPartitionGraphsCount() {
-      if (partitionGraphsBuilder_ == null) {
-        return partitionGraphs_.size();
-      } else {
-        return partitionGraphsBuilder_.getCount();
-      }
-    }
-    /**
-     * <pre>
-     * Graphs of the partitions executed by executors.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.GraphDef partition_graphs = 3;</code>
-     */
-    public org.tensorflow.proto.framework.GraphDef getPartitionGraphs(int index) {
-      if (partitionGraphsBuilder_ == null) {
-        return partitionGraphs_.get(index);
-      } else {
-        return partitionGraphsBuilder_.getMessage(index);
-      }
-    }
-    /**
-     * <pre>
-     * Graphs of the partitions executed by executors.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.GraphDef partition_graphs = 3;</code>
-     */
-    public Builder setPartitionGraphs(
-        int index, org.tensorflow.proto.framework.GraphDef value) {
-      if (partitionGraphsBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensurePartitionGraphsIsMutable();
-        partitionGraphs_.set(index, value);
-        onChanged();
-      } else {
-        partitionGraphsBuilder_.setMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Graphs of the partitions executed by executors.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.GraphDef partition_graphs = 3;</code>
-     */
-    public Builder setPartitionGraphs(
-        int index, org.tensorflow.proto.framework.GraphDef.Builder builderForValue) {
-      if (partitionGraphsBuilder_ == null) {
-        ensurePartitionGraphsIsMutable();
-        partitionGraphs_.set(index, builderForValue.build());
-        onChanged();
-      } else {
-        partitionGraphsBuilder_.setMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Graphs of the partitions executed by executors.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.GraphDef partition_graphs = 3;</code>
-     */
-    public Builder addPartitionGraphs(org.tensorflow.proto.framework.GraphDef value) {
-      if (partitionGraphsBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensurePartitionGraphsIsMutable();
-        partitionGraphs_.add(value);
-        onChanged();
-      } else {
-        partitionGraphsBuilder_.addMessage(value);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Graphs of the partitions executed by executors.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.GraphDef partition_graphs = 3;</code>
-     */
-    public Builder addPartitionGraphs(
-        int index, org.tensorflow.proto.framework.GraphDef value) {
-      if (partitionGraphsBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensurePartitionGraphsIsMutable();
-        partitionGraphs_.add(index, value);
-        onChanged();
-      } else {
-        partitionGraphsBuilder_.addMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Graphs of the partitions executed by executors.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.GraphDef partition_graphs = 3;</code>
-     */
-    public Builder addPartitionGraphs(
-        org.tensorflow.proto.framework.GraphDef.Builder builderForValue) {
-      if (partitionGraphsBuilder_ == null) {
-        ensurePartitionGraphsIsMutable();
-        partitionGraphs_.add(builderForValue.build());
-        onChanged();
-      } else {
-        partitionGraphsBuilder_.addMessage(builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Graphs of the partitions executed by executors.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.GraphDef partition_graphs = 3;</code>
-     */
-    public Builder addPartitionGraphs(
-        int index, org.tensorflow.proto.framework.GraphDef.Builder builderForValue) {
-      if (partitionGraphsBuilder_ == null) {
-        ensurePartitionGraphsIsMutable();
-        partitionGraphs_.add(index, builderForValue.build());
-        onChanged();
-      } else {
-        partitionGraphsBuilder_.addMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Graphs of the partitions executed by executors.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.GraphDef partition_graphs = 3;</code>
-     */
-    public Builder addAllPartitionGraphs(
-        java.lang.Iterable<? extends org.tensorflow.proto.framework.GraphDef> values) {
-      if (partitionGraphsBuilder_ == null) {
-        ensurePartitionGraphsIsMutable();
-        com.google.protobuf.AbstractMessageLite.Builder.addAll(
-            values, partitionGraphs_);
-        onChanged();
-      } else {
-        partitionGraphsBuilder_.addAllMessages(values);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Graphs of the partitions executed by executors.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.GraphDef partition_graphs = 3;</code>
-     */
-    public Builder clearPartitionGraphs() {
-      if (partitionGraphsBuilder_ == null) {
-        partitionGraphs_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000001);
-        onChanged();
-      } else {
-        partitionGraphsBuilder_.clear();
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Graphs of the partitions executed by executors.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.GraphDef partition_graphs = 3;</code>
-     */
-    public Builder removePartitionGraphs(int index) {
-      if (partitionGraphsBuilder_ == null) {
-        ensurePartitionGraphsIsMutable();
-        partitionGraphs_.remove(index);
-        onChanged();
-      } else {
-        partitionGraphsBuilder_.remove(index);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Graphs of the partitions executed by executors.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.GraphDef partition_graphs = 3;</code>
-     */
-    public org.tensorflow.proto.framework.GraphDef.Builder getPartitionGraphsBuilder(
-        int index) {
-      return getPartitionGraphsFieldBuilder().getBuilder(index);
-    }
-    /**
-     * <pre>
-     * Graphs of the partitions executed by executors.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.GraphDef partition_graphs = 3;</code>
-     */
-    public org.tensorflow.proto.framework.GraphDefOrBuilder getPartitionGraphsOrBuilder(
-        int index) {
-      if (partitionGraphsBuilder_ == null) {
-        return partitionGraphs_.get(index);  } else {
-        return partitionGraphsBuilder_.getMessageOrBuilder(index);
-      }
-    }
-    /**
-     * <pre>
-     * Graphs of the partitions executed by executors.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.GraphDef partition_graphs = 3;</code>
-     */
-    public java.util.List<? extends org.tensorflow.proto.framework.GraphDefOrBuilder> 
-         getPartitionGraphsOrBuilderList() {
-      if (partitionGraphsBuilder_ != null) {
-        return partitionGraphsBuilder_.getMessageOrBuilderList();
-      } else {
-        return java.util.Collections.unmodifiableList(partitionGraphs_);
-      }
-    }
-    /**
-     * <pre>
-     * Graphs of the partitions executed by executors.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.GraphDef partition_graphs = 3;</code>
-     */
-    public org.tensorflow.proto.framework.GraphDef.Builder addPartitionGraphsBuilder() {
-      return getPartitionGraphsFieldBuilder().addBuilder(
-          org.tensorflow.proto.framework.GraphDef.getDefaultInstance());
-    }
-    /**
-     * <pre>
-     * Graphs of the partitions executed by executors.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.GraphDef partition_graphs = 3;</code>
-     */
-    public org.tensorflow.proto.framework.GraphDef.Builder addPartitionGraphsBuilder(
-        int index) {
-      return getPartitionGraphsFieldBuilder().addBuilder(
-          index, org.tensorflow.proto.framework.GraphDef.getDefaultInstance());
-    }
-    /**
-     * <pre>
-     * Graphs of the partitions executed by executors.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.GraphDef partition_graphs = 3;</code>
-     */
-    public java.util.List<org.tensorflow.proto.framework.GraphDef.Builder> 
-         getPartitionGraphsBuilderList() {
-      return getPartitionGraphsFieldBuilder().getBuilderList();
-    }
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.framework.GraphDef, org.tensorflow.proto.framework.GraphDef.Builder, org.tensorflow.proto.framework.GraphDefOrBuilder> 
-        getPartitionGraphsFieldBuilder() {
-      if (partitionGraphsBuilder_ == null) {
-        partitionGraphsBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
-            org.tensorflow.proto.framework.GraphDef, org.tensorflow.proto.framework.GraphDef.Builder, org.tensorflow.proto.framework.GraphDefOrBuilder>(
-                partitionGraphs_,
-                ((bitField0_ & 0x00000001) != 0),
-                getParentForChildren(),
-                isClean());
-        partitionGraphs_ = null;
-      }
-      return partitionGraphsBuilder_;
-    }
-
-    private java.util.List<org.tensorflow.proto.framework.RunMetadata.FunctionGraphs> functionGraphs_ =
-      java.util.Collections.emptyList();
-    private void ensureFunctionGraphsIsMutable() {
-      if (!((bitField0_ & 0x00000002) != 0)) {
-        functionGraphs_ = new java.util.ArrayList<org.tensorflow.proto.framework.RunMetadata.FunctionGraphs>(functionGraphs_);
-        bitField0_ |= 0x00000002;
-       }
-    }
-
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.framework.RunMetadata.FunctionGraphs, org.tensorflow.proto.framework.RunMetadata.FunctionGraphs.Builder, org.tensorflow.proto.framework.RunMetadata.FunctionGraphsOrBuilder> functionGraphsBuilder_;
-
-    /**
-     * <pre>
-     * This is only populated for graphs that are run as functions in TensorFlow
-     * V2. There will be an entry below for each function that is traced.
-     * The main use cases of the post_optimization_graph and the partition_graphs
-     * is to give the caller insight into the graphs that were actually run by the
-     * runtime. Additional information (such as those in step_stats) will match
-     * these graphs.
-     * We also include the pre_optimization_graph since it is usually easier to
-     * read, and is helpful in situations where the caller wants to get a high
-     * level idea of what the built graph looks like (since the various graph
-     * optimization passes might change the structure of the graph significantly).
-     * </pre>
-     *
-     * <code>repeated .tensorflow.RunMetadata.FunctionGraphs function_graphs = 4;</code>
-     */
-    public java.util.List<org.tensorflow.proto.framework.RunMetadata.FunctionGraphs> getFunctionGraphsList() {
-      if (functionGraphsBuilder_ == null) {
-        return java.util.Collections.unmodifiableList(functionGraphs_);
-      } else {
-        return functionGraphsBuilder_.getMessageList();
-      }
-    }
-    /**
-     * <pre>
-     * This is only populated for graphs that are run as functions in TensorFlow
-     * V2. There will be an entry below for each function that is traced.
-     * The main use cases of the post_optimization_graph and the partition_graphs
-     * is to give the caller insight into the graphs that were actually run by the
-     * runtime. Additional information (such as those in step_stats) will match
-     * these graphs.
-     * We also include the pre_optimization_graph since it is usually easier to
-     * read, and is helpful in situations where the caller wants to get a high
-     * level idea of what the built graph looks like (since the various graph
-     * optimization passes might change the structure of the graph significantly).
-     * </pre>
-     *
-     * <code>repeated .tensorflow.RunMetadata.FunctionGraphs function_graphs = 4;</code>
-     */
-    public int getFunctionGraphsCount() {
-      if (functionGraphsBuilder_ == null) {
-        return functionGraphs_.size();
-      } else {
-        return functionGraphsBuilder_.getCount();
-      }
-    }
-    /**
-     * <pre>
-     * This is only populated for graphs that are run as functions in TensorFlow
-     * V2. There will be an entry below for each function that is traced.
-     * The main use cases of the post_optimization_graph and the partition_graphs
-     * is to give the caller insight into the graphs that were actually run by the
-     * runtime. Additional information (such as those in step_stats) will match
-     * these graphs.
-     * We also include the pre_optimization_graph since it is usually easier to
-     * read, and is helpful in situations where the caller wants to get a high
-     * level idea of what the built graph looks like (since the various graph
-     * optimization passes might change the structure of the graph significantly).
-     * </pre>
-     *
-     * <code>repeated .tensorflow.RunMetadata.FunctionGraphs function_graphs = 4;</code>
-     */
-    public org.tensorflow.proto.framework.RunMetadata.FunctionGraphs getFunctionGraphs(int index) {
-      if (functionGraphsBuilder_ == null) {
-        return functionGraphs_.get(index);
-      } else {
-        return functionGraphsBuilder_.getMessage(index);
-      }
-    }
-    /**
-     * <pre>
-     * This is only populated for graphs that are run as functions in TensorFlow
-     * V2. There will be an entry below for each function that is traced.
-     * The main use cases of the post_optimization_graph and the partition_graphs
-     * is to give the caller insight into the graphs that were actually run by the
-     * runtime. Additional information (such as those in step_stats) will match
-     * these graphs.
-     * We also include the pre_optimization_graph since it is usually easier to
-     * read, and is helpful in situations where the caller wants to get a high
-     * level idea of what the built graph looks like (since the various graph
-     * optimization passes might change the structure of the graph significantly).
-     * </pre>
-     *
-     * <code>repeated .tensorflow.RunMetadata.FunctionGraphs function_graphs = 4;</code>
-     */
-    public Builder setFunctionGraphs(
-        int index, org.tensorflow.proto.framework.RunMetadata.FunctionGraphs value) {
-      if (functionGraphsBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureFunctionGraphsIsMutable();
-        functionGraphs_.set(index, value);
-        onChanged();
-      } else {
-        functionGraphsBuilder_.setMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * This is only populated for graphs that are run as functions in TensorFlow
-     * V2. There will be an entry below for each function that is traced.
-     * The main use cases of the post_optimization_graph and the partition_graphs
-     * is to give the caller insight into the graphs that were actually run by the
-     * runtime. Additional information (such as those in step_stats) will match
-     * these graphs.
-     * We also include the pre_optimization_graph since it is usually easier to
-     * read, and is helpful in situations where the caller wants to get a high
-     * level idea of what the built graph looks like (since the various graph
-     * optimization passes might change the structure of the graph significantly).
-     * </pre>
-     *
-     * <code>repeated .tensorflow.RunMetadata.FunctionGraphs function_graphs = 4;</code>
-     */
-    public Builder setFunctionGraphs(
-        int index, org.tensorflow.proto.framework.RunMetadata.FunctionGraphs.Builder builderForValue) {
-      if (functionGraphsBuilder_ == null) {
-        ensureFunctionGraphsIsMutable();
-        functionGraphs_.set(index, builderForValue.build());
-        onChanged();
-      } else {
-        functionGraphsBuilder_.setMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * This is only populated for graphs that are run as functions in TensorFlow
-     * V2. There will be an entry below for each function that is traced.
-     * The main use cases of the post_optimization_graph and the partition_graphs
-     * is to give the caller insight into the graphs that were actually run by the
-     * runtime. Additional information (such as those in step_stats) will match
-     * these graphs.
-     * We also include the pre_optimization_graph since it is usually easier to
-     * read, and is helpful in situations where the caller wants to get a high
-     * level idea of what the built graph looks like (since the various graph
-     * optimization passes might change the structure of the graph significantly).
-     * </pre>
-     *
-     * <code>repeated .tensorflow.RunMetadata.FunctionGraphs function_graphs = 4;</code>
-     */
-    public Builder addFunctionGraphs(org.tensorflow.proto.framework.RunMetadata.FunctionGraphs value) {
-      if (functionGraphsBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureFunctionGraphsIsMutable();
-        functionGraphs_.add(value);
-        onChanged();
-      } else {
-        functionGraphsBuilder_.addMessage(value);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * This is only populated for graphs that are run as functions in TensorFlow
-     * V2. There will be an entry below for each function that is traced.
-     * The main use cases of the post_optimization_graph and the partition_graphs
-     * is to give the caller insight into the graphs that were actually run by the
-     * runtime. Additional information (such as those in step_stats) will match
-     * these graphs.
-     * We also include the pre_optimization_graph since it is usually easier to
-     * read, and is helpful in situations where the caller wants to get a high
-     * level idea of what the built graph looks like (since the various graph
-     * optimization passes might change the structure of the graph significantly).
-     * </pre>
-     *
-     * <code>repeated .tensorflow.RunMetadata.FunctionGraphs function_graphs = 4;</code>
-     */
-    public Builder addFunctionGraphs(
-        int index, org.tensorflow.proto.framework.RunMetadata.FunctionGraphs value) {
-      if (functionGraphsBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureFunctionGraphsIsMutable();
-        functionGraphs_.add(index, value);
-        onChanged();
-      } else {
-        functionGraphsBuilder_.addMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * This is only populated for graphs that are run as functions in TensorFlow
-     * V2. There will be an entry below for each function that is traced.
-     * The main use cases of the post_optimization_graph and the partition_graphs
-     * is to give the caller insight into the graphs that were actually run by the
-     * runtime. Additional information (such as those in step_stats) will match
-     * these graphs.
-     * We also include the pre_optimization_graph since it is usually easier to
-     * read, and is helpful in situations where the caller wants to get a high
-     * level idea of what the built graph looks like (since the various graph
-     * optimization passes might change the structure of the graph significantly).
-     * </pre>
-     *
-     * <code>repeated .tensorflow.RunMetadata.FunctionGraphs function_graphs = 4;</code>
-     */
-    public Builder addFunctionGraphs(
-        org.tensorflow.proto.framework.RunMetadata.FunctionGraphs.Builder builderForValue) {
-      if (functionGraphsBuilder_ == null) {
-        ensureFunctionGraphsIsMutable();
-        functionGraphs_.add(builderForValue.build());
-        onChanged();
-      } else {
-        functionGraphsBuilder_.addMessage(builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * This is only populated for graphs that are run as functions in TensorFlow
-     * V2. There will be an entry below for each function that is traced.
-     * The main use cases of the post_optimization_graph and the partition_graphs
-     * is to give the caller insight into the graphs that were actually run by the
-     * runtime. Additional information (such as those in step_stats) will match
-     * these graphs.
-     * We also include the pre_optimization_graph since it is usually easier to
-     * read, and is helpful in situations where the caller wants to get a high
-     * level idea of what the built graph looks like (since the various graph
-     * optimization passes might change the structure of the graph significantly).
-     * </pre>
-     *
-     * <code>repeated .tensorflow.RunMetadata.FunctionGraphs function_graphs = 4;</code>
-     */
-    public Builder addFunctionGraphs(
-        int index, org.tensorflow.proto.framework.RunMetadata.FunctionGraphs.Builder builderForValue) {
-      if (functionGraphsBuilder_ == null) {
-        ensureFunctionGraphsIsMutable();
-        functionGraphs_.add(index, builderForValue.build());
-        onChanged();
-      } else {
-        functionGraphsBuilder_.addMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * This is only populated for graphs that are run as functions in TensorFlow
-     * V2. There will be an entry below for each function that is traced.
-     * The main use cases of the post_optimization_graph and the partition_graphs
-     * is to give the caller insight into the graphs that were actually run by the
-     * runtime. Additional information (such as those in step_stats) will match
-     * these graphs.
-     * We also include the pre_optimization_graph since it is usually easier to
-     * read, and is helpful in situations where the caller wants to get a high
-     * level idea of what the built graph looks like (since the various graph
-     * optimization passes might change the structure of the graph significantly).
-     * </pre>
-     *
-     * <code>repeated .tensorflow.RunMetadata.FunctionGraphs function_graphs = 4;</code>
-     */
-    public Builder addAllFunctionGraphs(
-        java.lang.Iterable<? extends org.tensorflow.proto.framework.RunMetadata.FunctionGraphs> values) {
-      if (functionGraphsBuilder_ == null) {
-        ensureFunctionGraphsIsMutable();
-        com.google.protobuf.AbstractMessageLite.Builder.addAll(
-            values, functionGraphs_);
-        onChanged();
-      } else {
-        functionGraphsBuilder_.addAllMessages(values);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * This is only populated for graphs that are run as functions in TensorFlow
-     * V2. There will be an entry below for each function that is traced.
-     * The main use cases of the post_optimization_graph and the partition_graphs
-     * is to give the caller insight into the graphs that were actually run by the
-     * runtime. Additional information (such as those in step_stats) will match
-     * these graphs.
-     * We also include the pre_optimization_graph since it is usually easier to
-     * read, and is helpful in situations where the caller wants to get a high
-     * level idea of what the built graph looks like (since the various graph
-     * optimization passes might change the structure of the graph significantly).
-     * </pre>
-     *
-     * <code>repeated .tensorflow.RunMetadata.FunctionGraphs function_graphs = 4;</code>
-     */
-    public Builder clearFunctionGraphs() {
-      if (functionGraphsBuilder_ == null) {
-        functionGraphs_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000002);
-        onChanged();
-      } else {
-        functionGraphsBuilder_.clear();
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * This is only populated for graphs that are run as functions in TensorFlow
-     * V2. There will be an entry below for each function that is traced.
-     * The main use cases of the post_optimization_graph and the partition_graphs
-     * is to give the caller insight into the graphs that were actually run by the
-     * runtime. Additional information (such as those in step_stats) will match
-     * these graphs.
-     * We also include the pre_optimization_graph since it is usually easier to
-     * read, and is helpful in situations where the caller wants to get a high
-     * level idea of what the built graph looks like (since the various graph
-     * optimization passes might change the structure of the graph significantly).
-     * </pre>
-     *
-     * <code>repeated .tensorflow.RunMetadata.FunctionGraphs function_graphs = 4;</code>
-     */
-    public Builder removeFunctionGraphs(int index) {
-      if (functionGraphsBuilder_ == null) {
-        ensureFunctionGraphsIsMutable();
-        functionGraphs_.remove(index);
-        onChanged();
-      } else {
-        functionGraphsBuilder_.remove(index);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * This is only populated for graphs that are run as functions in TensorFlow
-     * V2. There will be an entry below for each function that is traced.
-     * The main use cases of the post_optimization_graph and the partition_graphs
-     * is to give the caller insight into the graphs that were actually run by the
-     * runtime. Additional information (such as those in step_stats) will match
-     * these graphs.
-     * We also include the pre_optimization_graph since it is usually easier to
-     * read, and is helpful in situations where the caller wants to get a high
-     * level idea of what the built graph looks like (since the various graph
-     * optimization passes might change the structure of the graph significantly).
-     * </pre>
-     *
-     * <code>repeated .tensorflow.RunMetadata.FunctionGraphs function_graphs = 4;</code>
-     */
-    public org.tensorflow.proto.framework.RunMetadata.FunctionGraphs.Builder getFunctionGraphsBuilder(
-        int index) {
-      return getFunctionGraphsFieldBuilder().getBuilder(index);
-    }
-    /**
-     * <pre>
-     * This is only populated for graphs that are run as functions in TensorFlow
-     * V2. There will be an entry below for each function that is traced.
-     * The main use cases of the post_optimization_graph and the partition_graphs
-     * is to give the caller insight into the graphs that were actually run by the
-     * runtime. Additional information (such as those in step_stats) will match
-     * these graphs.
-     * We also include the pre_optimization_graph since it is usually easier to
-     * read, and is helpful in situations where the caller wants to get a high
-     * level idea of what the built graph looks like (since the various graph
-     * optimization passes might change the structure of the graph significantly).
-     * </pre>
-     *
-     * <code>repeated .tensorflow.RunMetadata.FunctionGraphs function_graphs = 4;</code>
-     */
-    public org.tensorflow.proto.framework.RunMetadata.FunctionGraphsOrBuilder getFunctionGraphsOrBuilder(
-        int index) {
-      if (functionGraphsBuilder_ == null) {
-        return functionGraphs_.get(index);  } else {
-        return functionGraphsBuilder_.getMessageOrBuilder(index);
-      }
-    }
-    /**
-     * <pre>
-     * This is only populated for graphs that are run as functions in TensorFlow
-     * V2. There will be an entry below for each function that is traced.
-     * The main use cases of the post_optimization_graph and the partition_graphs
-     * is to give the caller insight into the graphs that were actually run by the
-     * runtime. Additional information (such as those in step_stats) will match
-     * these graphs.
-     * We also include the pre_optimization_graph since it is usually easier to
-     * read, and is helpful in situations where the caller wants to get a high
-     * level idea of what the built graph looks like (since the various graph
-     * optimization passes might change the structure of the graph significantly).
-     * </pre>
-     *
-     * <code>repeated .tensorflow.RunMetadata.FunctionGraphs function_graphs = 4;</code>
-     */
-    public java.util.List<? extends org.tensorflow.proto.framework.RunMetadata.FunctionGraphsOrBuilder> 
-         getFunctionGraphsOrBuilderList() {
-      if (functionGraphsBuilder_ != null) {
-        return functionGraphsBuilder_.getMessageOrBuilderList();
-      } else {
-        return java.util.Collections.unmodifiableList(functionGraphs_);
-      }
-    }
-    /**
-     * <pre>
-     * This is only populated for graphs that are run as functions in TensorFlow
-     * V2. There will be an entry below for each function that is traced.
-     * The main use cases of the post_optimization_graph and the partition_graphs
-     * is to give the caller insight into the graphs that were actually run by the
-     * runtime. Additional information (such as those in step_stats) will match
-     * these graphs.
-     * We also include the pre_optimization_graph since it is usually easier to
-     * read, and is helpful in situations where the caller wants to get a high
-     * level idea of what the built graph looks like (since the various graph
-     * optimization passes might change the structure of the graph significantly).
-     * </pre>
-     *
-     * <code>repeated .tensorflow.RunMetadata.FunctionGraphs function_graphs = 4;</code>
-     */
-    public org.tensorflow.proto.framework.RunMetadata.FunctionGraphs.Builder addFunctionGraphsBuilder() {
-      return getFunctionGraphsFieldBuilder().addBuilder(
-          org.tensorflow.proto.framework.RunMetadata.FunctionGraphs.getDefaultInstance());
-    }
-    /**
-     * <pre>
-     * This is only populated for graphs that are run as functions in TensorFlow
-     * V2. There will be an entry below for each function that is traced.
-     * The main use cases of the post_optimization_graph and the partition_graphs
-     * is to give the caller insight into the graphs that were actually run by the
-     * runtime. Additional information (such as those in step_stats) will match
-     * these graphs.
-     * We also include the pre_optimization_graph since it is usually easier to
-     * read, and is helpful in situations where the caller wants to get a high
-     * level idea of what the built graph looks like (since the various graph
-     * optimization passes might change the structure of the graph significantly).
-     * </pre>
-     *
-     * <code>repeated .tensorflow.RunMetadata.FunctionGraphs function_graphs = 4;</code>
-     */
-    public org.tensorflow.proto.framework.RunMetadata.FunctionGraphs.Builder addFunctionGraphsBuilder(
-        int index) {
-      return getFunctionGraphsFieldBuilder().addBuilder(
-          index, org.tensorflow.proto.framework.RunMetadata.FunctionGraphs.getDefaultInstance());
-    }
-    /**
-     * <pre>
-     * This is only populated for graphs that are run as functions in TensorFlow
-     * V2. There will be an entry below for each function that is traced.
-     * The main use cases of the post_optimization_graph and the partition_graphs
-     * is to give the caller insight into the graphs that were actually run by the
-     * runtime. Additional information (such as those in step_stats) will match
-     * these graphs.
-     * We also include the pre_optimization_graph since it is usually easier to
-     * read, and is helpful in situations where the caller wants to get a high
-     * level idea of what the built graph looks like (since the various graph
-     * optimization passes might change the structure of the graph significantly).
-     * </pre>
-     *
-     * <code>repeated .tensorflow.RunMetadata.FunctionGraphs function_graphs = 4;</code>
-     */
-    public java.util.List<org.tensorflow.proto.framework.RunMetadata.FunctionGraphs.Builder> 
-         getFunctionGraphsBuilderList() {
-      return getFunctionGraphsFieldBuilder().getBuilderList();
-    }
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.framework.RunMetadata.FunctionGraphs, org.tensorflow.proto.framework.RunMetadata.FunctionGraphs.Builder, org.tensorflow.proto.framework.RunMetadata.FunctionGraphsOrBuilder> 
-        getFunctionGraphsFieldBuilder() {
-      if (functionGraphsBuilder_ == null) {
-        functionGraphsBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
-            org.tensorflow.proto.framework.RunMetadata.FunctionGraphs, org.tensorflow.proto.framework.RunMetadata.FunctionGraphs.Builder, org.tensorflow.proto.framework.RunMetadata.FunctionGraphsOrBuilder>(
-                functionGraphs_,
-                ((bitField0_ & 0x00000002) != 0),
-                getParentForChildren(),
-                isClean());
-        functionGraphs_ = null;
-      }
-      return functionGraphsBuilder_;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.RunMetadata)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.RunMetadata)
-  private static final org.tensorflow.proto.framework.RunMetadata DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.RunMetadata();
-  }
-
-  public static org.tensorflow.proto.framework.RunMetadata getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<RunMetadata>
-      PARSER = new com.google.protobuf.AbstractParser<RunMetadata>() {
-    @java.lang.Override
-    public RunMetadata parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new RunMetadata(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<RunMetadata> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<RunMetadata> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.framework.RunMetadata getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/RunMetadataOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/RunMetadataOrBuilder.java
deleted file mode 100644
index 21e15741316..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/RunMetadataOrBuilder.java
+++ /dev/null
@@ -1,198 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/config.proto
-
-package org.tensorflow.proto.framework;
-
-public interface RunMetadataOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.RunMetadata)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <pre>
-   * Statistics traced for this step. Populated if tracing is turned on via the
-   * "RunOptions" proto.
-   * EXPERIMENTAL: The format and set of events may change in future versions.
-   * </pre>
-   *
-   * <code>.tensorflow.StepStats step_stats = 1;</code>
-   */
-  boolean hasStepStats();
-  /**
-   * <pre>
-   * Statistics traced for this step. Populated if tracing is turned on via the
-   * "RunOptions" proto.
-   * EXPERIMENTAL: The format and set of events may change in future versions.
-   * </pre>
-   *
-   * <code>.tensorflow.StepStats step_stats = 1;</code>
-   */
-  org.tensorflow.proto.framework.StepStats getStepStats();
-  /**
-   * <pre>
-   * Statistics traced for this step. Populated if tracing is turned on via the
-   * "RunOptions" proto.
-   * EXPERIMENTAL: The format and set of events may change in future versions.
-   * </pre>
-   *
-   * <code>.tensorflow.StepStats step_stats = 1;</code>
-   */
-  org.tensorflow.proto.framework.StepStatsOrBuilder getStepStatsOrBuilder();
-
-  /**
-   * <pre>
-   * The cost graph for the computation defined by the run call.
-   * </pre>
-   *
-   * <code>.tensorflow.CostGraphDef cost_graph = 2;</code>
-   */
-  boolean hasCostGraph();
-  /**
-   * <pre>
-   * The cost graph for the computation defined by the run call.
-   * </pre>
-   *
-   * <code>.tensorflow.CostGraphDef cost_graph = 2;</code>
-   */
-  org.tensorflow.proto.framework.CostGraphDef getCostGraph();
-  /**
-   * <pre>
-   * The cost graph for the computation defined by the run call.
-   * </pre>
-   *
-   * <code>.tensorflow.CostGraphDef cost_graph = 2;</code>
-   */
-  org.tensorflow.proto.framework.CostGraphDefOrBuilder getCostGraphOrBuilder();
-
-  /**
-   * <pre>
-   * Graphs of the partitions executed by executors.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.GraphDef partition_graphs = 3;</code>
-   */
-  java.util.List<org.tensorflow.proto.framework.GraphDef> 
-      getPartitionGraphsList();
-  /**
-   * <pre>
-   * Graphs of the partitions executed by executors.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.GraphDef partition_graphs = 3;</code>
-   */
-  org.tensorflow.proto.framework.GraphDef getPartitionGraphs(int index);
-  /**
-   * <pre>
-   * Graphs of the partitions executed by executors.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.GraphDef partition_graphs = 3;</code>
-   */
-  int getPartitionGraphsCount();
-  /**
-   * <pre>
-   * Graphs of the partitions executed by executors.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.GraphDef partition_graphs = 3;</code>
-   */
-  java.util.List<? extends org.tensorflow.proto.framework.GraphDefOrBuilder> 
-      getPartitionGraphsOrBuilderList();
-  /**
-   * <pre>
-   * Graphs of the partitions executed by executors.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.GraphDef partition_graphs = 3;</code>
-   */
-  org.tensorflow.proto.framework.GraphDefOrBuilder getPartitionGraphsOrBuilder(
-      int index);
-
-  /**
-   * <pre>
-   * This is only populated for graphs that are run as functions in TensorFlow
-   * V2. There will be an entry below for each function that is traced.
-   * The main use cases of the post_optimization_graph and the partition_graphs
-   * is to give the caller insight into the graphs that were actually run by the
-   * runtime. Additional information (such as those in step_stats) will match
-   * these graphs.
-   * We also include the pre_optimization_graph since it is usually easier to
-   * read, and is helpful in situations where the caller wants to get a high
-   * level idea of what the built graph looks like (since the various graph
-   * optimization passes might change the structure of the graph significantly).
-   * </pre>
-   *
-   * <code>repeated .tensorflow.RunMetadata.FunctionGraphs function_graphs = 4;</code>
-   */
-  java.util.List<org.tensorflow.proto.framework.RunMetadata.FunctionGraphs> 
-      getFunctionGraphsList();
-  /**
-   * <pre>
-   * This is only populated for graphs that are run as functions in TensorFlow
-   * V2. There will be an entry below for each function that is traced.
-   * The main use cases of the post_optimization_graph and the partition_graphs
-   * is to give the caller insight into the graphs that were actually run by the
-   * runtime. Additional information (such as those in step_stats) will match
-   * these graphs.
-   * We also include the pre_optimization_graph since it is usually easier to
-   * read, and is helpful in situations where the caller wants to get a high
-   * level idea of what the built graph looks like (since the various graph
-   * optimization passes might change the structure of the graph significantly).
-   * </pre>
-   *
-   * <code>repeated .tensorflow.RunMetadata.FunctionGraphs function_graphs = 4;</code>
-   */
-  org.tensorflow.proto.framework.RunMetadata.FunctionGraphs getFunctionGraphs(int index);
-  /**
-   * <pre>
-   * This is only populated for graphs that are run as functions in TensorFlow
-   * V2. There will be an entry below for each function that is traced.
-   * The main use cases of the post_optimization_graph and the partition_graphs
-   * is to give the caller insight into the graphs that were actually run by the
-   * runtime. Additional information (such as those in step_stats) will match
-   * these graphs.
-   * We also include the pre_optimization_graph since it is usually easier to
-   * read, and is helpful in situations where the caller wants to get a high
-   * level idea of what the built graph looks like (since the various graph
-   * optimization passes might change the structure of the graph significantly).
-   * </pre>
-   *
-   * <code>repeated .tensorflow.RunMetadata.FunctionGraphs function_graphs = 4;</code>
-   */
-  int getFunctionGraphsCount();
-  /**
-   * <pre>
-   * This is only populated for graphs that are run as functions in TensorFlow
-   * V2. There will be an entry below for each function that is traced.
-   * The main use cases of the post_optimization_graph and the partition_graphs
-   * is to give the caller insight into the graphs that were actually run by the
-   * runtime. Additional information (such as those in step_stats) will match
-   * these graphs.
-   * We also include the pre_optimization_graph since it is usually easier to
-   * read, and is helpful in situations where the caller wants to get a high
-   * level idea of what the built graph looks like (since the various graph
-   * optimization passes might change the structure of the graph significantly).
-   * </pre>
-   *
-   * <code>repeated .tensorflow.RunMetadata.FunctionGraphs function_graphs = 4;</code>
-   */
-  java.util.List<? extends org.tensorflow.proto.framework.RunMetadata.FunctionGraphsOrBuilder> 
-      getFunctionGraphsOrBuilderList();
-  /**
-   * <pre>
-   * This is only populated for graphs that are run as functions in TensorFlow
-   * V2. There will be an entry below for each function that is traced.
-   * The main use cases of the post_optimization_graph and the partition_graphs
-   * is to give the caller insight into the graphs that were actually run by the
-   * runtime. Additional information (such as those in step_stats) will match
-   * these graphs.
-   * We also include the pre_optimization_graph since it is usually easier to
-   * read, and is helpful in situations where the caller wants to get a high
-   * level idea of what the built graph looks like (since the various graph
-   * optimization passes might change the structure of the graph significantly).
-   * </pre>
-   *
-   * <code>repeated .tensorflow.RunMetadata.FunctionGraphs function_graphs = 4;</code>
-   */
-  org.tensorflow.proto.framework.RunMetadata.FunctionGraphsOrBuilder getFunctionGraphsOrBuilder(
-      int index);
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/RunOptions.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/RunOptions.java
deleted file mode 100644
index 9ccdb06d2b6..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/RunOptions.java
+++ /dev/null
@@ -1,2708 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/config.proto
-
-package org.tensorflow.proto.framework;
-
-/**
- * <pre>
- * Options for a single Run() call.
- * </pre>
- *
- * Protobuf type {@code tensorflow.RunOptions}
- */
-public  final class RunOptions extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.RunOptions)
-    RunOptionsOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use RunOptions.newBuilder() to construct.
-  private RunOptions(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private RunOptions() {
-    traceLevel_ = 0;
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new RunOptions();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private RunOptions(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 8: {
-            int rawValue = input.readEnum();
-
-            traceLevel_ = rawValue;
-            break;
-          }
-          case 16: {
-
-            timeoutInMs_ = input.readInt64();
-            break;
-          }
-          case 24: {
-
-            interOpThreadPool_ = input.readInt32();
-            break;
-          }
-          case 40: {
-
-            outputPartitionGraphs_ = input.readBool();
-            break;
-          }
-          case 50: {
-            org.tensorflow.proto.framework.DebugOptions.Builder subBuilder = null;
-            if (debugOptions_ != null) {
-              subBuilder = debugOptions_.toBuilder();
-            }
-            debugOptions_ = input.readMessage(org.tensorflow.proto.framework.DebugOptions.parser(), extensionRegistry);
-            if (subBuilder != null) {
-              subBuilder.mergeFrom(debugOptions_);
-              debugOptions_ = subBuilder.buildPartial();
-            }
-
-            break;
-          }
-          case 56: {
-
-            reportTensorAllocationsUponOom_ = input.readBool();
-            break;
-          }
-          case 66: {
-            org.tensorflow.proto.framework.RunOptions.Experimental.Builder subBuilder = null;
-            if (experimental_ != null) {
-              subBuilder = experimental_.toBuilder();
-            }
-            experimental_ = input.readMessage(org.tensorflow.proto.framework.RunOptions.Experimental.parser(), extensionRegistry);
-            if (subBuilder != null) {
-              subBuilder.mergeFrom(experimental_);
-              experimental_ = subBuilder.buildPartial();
-            }
-
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_RunOptions_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_RunOptions_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.framework.RunOptions.class, org.tensorflow.proto.framework.RunOptions.Builder.class);
-  }
-
-  /**
-   * <pre>
-   * TODO(pbar) Turn this into a TraceOptions proto which allows
-   * tracing to be controlled in a more orthogonal manner?
-   * </pre>
-   *
-   * Protobuf enum {@code tensorflow.RunOptions.TraceLevel}
-   */
-  public enum TraceLevel
-      implements com.google.protobuf.ProtocolMessageEnum {
-    /**
-     * <code>NO_TRACE = 0;</code>
-     */
-    NO_TRACE(0),
-    /**
-     * <code>SOFTWARE_TRACE = 1;</code>
-     */
-    SOFTWARE_TRACE(1),
-    /**
-     * <code>HARDWARE_TRACE = 2;</code>
-     */
-    HARDWARE_TRACE(2),
-    /**
-     * <code>FULL_TRACE = 3;</code>
-     */
-    FULL_TRACE(3),
-    UNRECOGNIZED(-1),
-    ;
-
-    /**
-     * <code>NO_TRACE = 0;</code>
-     */
-    public static final int NO_TRACE_VALUE = 0;
-    /**
-     * <code>SOFTWARE_TRACE = 1;</code>
-     */
-    public static final int SOFTWARE_TRACE_VALUE = 1;
-    /**
-     * <code>HARDWARE_TRACE = 2;</code>
-     */
-    public static final int HARDWARE_TRACE_VALUE = 2;
-    /**
-     * <code>FULL_TRACE = 3;</code>
-     */
-    public static final int FULL_TRACE_VALUE = 3;
-
-
-    public final int getNumber() {
-      if (this == UNRECOGNIZED) {
-        throw new java.lang.IllegalArgumentException(
-            "Can't get the number of an unknown enum value.");
-      }
-      return value;
-    }
-
-    /**
-     * @deprecated Use {@link #forNumber(int)} instead.
-     */
-    @java.lang.Deprecated
-    public static TraceLevel valueOf(int value) {
-      return forNumber(value);
-    }
-
-    public static TraceLevel forNumber(int value) {
-      switch (value) {
-        case 0: return NO_TRACE;
-        case 1: return SOFTWARE_TRACE;
-        case 2: return HARDWARE_TRACE;
-        case 3: return FULL_TRACE;
-        default: return null;
-      }
-    }
-
-    public static com.google.protobuf.Internal.EnumLiteMap<TraceLevel>
-        internalGetValueMap() {
-      return internalValueMap;
-    }
-    private static final com.google.protobuf.Internal.EnumLiteMap<
-        TraceLevel> internalValueMap =
-          new com.google.protobuf.Internal.EnumLiteMap<TraceLevel>() {
-            public TraceLevel findValueByNumber(int number) {
-              return TraceLevel.forNumber(number);
-            }
-          };
-
-    public final com.google.protobuf.Descriptors.EnumValueDescriptor
-        getValueDescriptor() {
-      return getDescriptor().getValues().get(ordinal());
-    }
-    public final com.google.protobuf.Descriptors.EnumDescriptor
-        getDescriptorForType() {
-      return getDescriptor();
-    }
-    public static final com.google.protobuf.Descriptors.EnumDescriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.RunOptions.getDescriptor().getEnumTypes().get(0);
-    }
-
-    private static final TraceLevel[] VALUES = values();
-
-    public static TraceLevel valueOf(
-        com.google.protobuf.Descriptors.EnumValueDescriptor desc) {
-      if (desc.getType() != getDescriptor()) {
-        throw new java.lang.IllegalArgumentException(
-          "EnumValueDescriptor is not for this type.");
-      }
-      if (desc.getIndex() == -1) {
-        return UNRECOGNIZED;
-      }
-      return VALUES[desc.getIndex()];
-    }
-
-    private final int value;
-
-    private TraceLevel(int value) {
-      this.value = value;
-    }
-
-    // @@protoc_insertion_point(enum_scope:tensorflow.RunOptions.TraceLevel)
-  }
-
-  public interface ExperimentalOrBuilder extends
-      // @@protoc_insertion_point(interface_extends:tensorflow.RunOptions.Experimental)
-      com.google.protobuf.MessageOrBuilder {
-
-    /**
-     * <pre>
-     * If non-zero, declares that this graph is going to use collective
-     * ops and must synchronize step_ids with any other graph with this
-     * same group_key value (in a distributed computation where tasks
-     * run disjoint graphs).
-     * </pre>
-     *
-     * <code>int64 collective_graph_key = 1;</code>
-     */
-    long getCollectiveGraphKey();
-
-    /**
-     * <pre>
-     * If true, then operations (using the inter-op pool) across all
-     * session::run() calls will be centrally scheduled, optimizing for (median
-     * and tail) latency.
-     * Consider using this option for CPU-bound workloads like inference.
-     * </pre>
-     *
-     * <code>bool use_run_handler_pool = 2;</code>
-     */
-    boolean getUseRunHandlerPool();
-
-    /**
-     * <code>.tensorflow.RunOptions.Experimental.RunHandlerPoolOptions run_handler_pool_options = 3;</code>
-     */
-    boolean hasRunHandlerPoolOptions();
-    /**
-     * <code>.tensorflow.RunOptions.Experimental.RunHandlerPoolOptions run_handler_pool_options = 3;</code>
-     */
-    org.tensorflow.proto.framework.RunOptions.Experimental.RunHandlerPoolOptions getRunHandlerPoolOptions();
-    /**
-     * <code>.tensorflow.RunOptions.Experimental.RunHandlerPoolOptions run_handler_pool_options = 3;</code>
-     */
-    org.tensorflow.proto.framework.RunOptions.Experimental.RunHandlerPoolOptionsOrBuilder getRunHandlerPoolOptionsOrBuilder();
-  }
-  /**
-   * <pre>
-   * Everything inside Experimental is subject to change and is not subject
-   * to API stability guarantees in
-   * https://www.tensorflow.org/guide/version_compat.
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.RunOptions.Experimental}
-   */
-  public  static final class Experimental extends
-      com.google.protobuf.GeneratedMessageV3 implements
-      // @@protoc_insertion_point(message_implements:tensorflow.RunOptions.Experimental)
-      ExperimentalOrBuilder {
-  private static final long serialVersionUID = 0L;
-    // Use Experimental.newBuilder() to construct.
-    private Experimental(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-      super(builder);
-    }
-    private Experimental() {
-    }
-
-    @java.lang.Override
-    @SuppressWarnings({"unused"})
-    protected java.lang.Object newInstance(
-        UnusedPrivateParameter unused) {
-      return new Experimental();
-    }
-
-    @java.lang.Override
-    public final com.google.protobuf.UnknownFieldSet
-    getUnknownFields() {
-      return this.unknownFields;
-    }
-    private Experimental(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      this();
-      if (extensionRegistry == null) {
-        throw new java.lang.NullPointerException();
-      }
-      com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-          com.google.protobuf.UnknownFieldSet.newBuilder();
-      try {
-        boolean done = false;
-        while (!done) {
-          int tag = input.readTag();
-          switch (tag) {
-            case 0:
-              done = true;
-              break;
-            case 8: {
-
-              collectiveGraphKey_ = input.readInt64();
-              break;
-            }
-            case 16: {
-
-              useRunHandlerPool_ = input.readBool();
-              break;
-            }
-            case 26: {
-              org.tensorflow.proto.framework.RunOptions.Experimental.RunHandlerPoolOptions.Builder subBuilder = null;
-              if (runHandlerPoolOptions_ != null) {
-                subBuilder = runHandlerPoolOptions_.toBuilder();
-              }
-              runHandlerPoolOptions_ = input.readMessage(org.tensorflow.proto.framework.RunOptions.Experimental.RunHandlerPoolOptions.parser(), extensionRegistry);
-              if (subBuilder != null) {
-                subBuilder.mergeFrom(runHandlerPoolOptions_);
-                runHandlerPoolOptions_ = subBuilder.buildPartial();
-              }
-
-              break;
-            }
-            default: {
-              if (!parseUnknownField(
-                  input, unknownFields, extensionRegistry, tag)) {
-                done = true;
-              }
-              break;
-            }
-          }
-        }
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        throw e.setUnfinishedMessage(this);
-      } catch (java.io.IOException e) {
-        throw new com.google.protobuf.InvalidProtocolBufferException(
-            e).setUnfinishedMessage(this);
-      } finally {
-        this.unknownFields = unknownFields.build();
-        makeExtensionsImmutable();
-      }
-    }
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_RunOptions_Experimental_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_RunOptions_Experimental_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.RunOptions.Experimental.class, org.tensorflow.proto.framework.RunOptions.Experimental.Builder.class);
-    }
-
-    public interface RunHandlerPoolOptionsOrBuilder extends
-        // @@protoc_insertion_point(interface_extends:tensorflow.RunOptions.Experimental.RunHandlerPoolOptions)
-        com.google.protobuf.MessageOrBuilder {
-
-      /**
-       * <pre>
-       * Priority of the request. The run handler thread pool will schedule ops
-       * based on the priority number. The larger number means higher priority.
-       * </pre>
-       *
-       * <code>int64 priority = 1;</code>
-       */
-      long getPriority();
-    }
-    /**
-     * <pre>
-     * Options for run handler thread pool.
-     * </pre>
-     *
-     * Protobuf type {@code tensorflow.RunOptions.Experimental.RunHandlerPoolOptions}
-     */
-    public  static final class RunHandlerPoolOptions extends
-        com.google.protobuf.GeneratedMessageV3 implements
-        // @@protoc_insertion_point(message_implements:tensorflow.RunOptions.Experimental.RunHandlerPoolOptions)
-        RunHandlerPoolOptionsOrBuilder {
-    private static final long serialVersionUID = 0L;
-      // Use RunHandlerPoolOptions.newBuilder() to construct.
-      private RunHandlerPoolOptions(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-        super(builder);
-      }
-      private RunHandlerPoolOptions() {
-      }
-
-      @java.lang.Override
-      @SuppressWarnings({"unused"})
-      protected java.lang.Object newInstance(
-          UnusedPrivateParameter unused) {
-        return new RunHandlerPoolOptions();
-      }
-
-      @java.lang.Override
-      public final com.google.protobuf.UnknownFieldSet
-      getUnknownFields() {
-        return this.unknownFields;
-      }
-      private RunHandlerPoolOptions(
-          com.google.protobuf.CodedInputStream input,
-          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-          throws com.google.protobuf.InvalidProtocolBufferException {
-        this();
-        if (extensionRegistry == null) {
-          throw new java.lang.NullPointerException();
-        }
-        com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-            com.google.protobuf.UnknownFieldSet.newBuilder();
-        try {
-          boolean done = false;
-          while (!done) {
-            int tag = input.readTag();
-            switch (tag) {
-              case 0:
-                done = true;
-                break;
-              case 8: {
-
-                priority_ = input.readInt64();
-                break;
-              }
-              default: {
-                if (!parseUnknownField(
-                    input, unknownFields, extensionRegistry, tag)) {
-                  done = true;
-                }
-                break;
-              }
-            }
-          }
-        } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-          throw e.setUnfinishedMessage(this);
-        } catch (java.io.IOException e) {
-          throw new com.google.protobuf.InvalidProtocolBufferException(
-              e).setUnfinishedMessage(this);
-        } finally {
-          this.unknownFields = unknownFields.build();
-          makeExtensionsImmutable();
-        }
-      }
-      public static final com.google.protobuf.Descriptors.Descriptor
-          getDescriptor() {
-        return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_RunOptions_Experimental_RunHandlerPoolOptions_descriptor;
-      }
-
-      @java.lang.Override
-      protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-          internalGetFieldAccessorTable() {
-        return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_RunOptions_Experimental_RunHandlerPoolOptions_fieldAccessorTable
-            .ensureFieldAccessorsInitialized(
-                org.tensorflow.proto.framework.RunOptions.Experimental.RunHandlerPoolOptions.class, org.tensorflow.proto.framework.RunOptions.Experimental.RunHandlerPoolOptions.Builder.class);
-      }
-
-      public static final int PRIORITY_FIELD_NUMBER = 1;
-      private long priority_;
-      /**
-       * <pre>
-       * Priority of the request. The run handler thread pool will schedule ops
-       * based on the priority number. The larger number means higher priority.
-       * </pre>
-       *
-       * <code>int64 priority = 1;</code>
-       */
-      public long getPriority() {
-        return priority_;
-      }
-
-      private byte memoizedIsInitialized = -1;
-      @java.lang.Override
-      public final boolean isInitialized() {
-        byte isInitialized = memoizedIsInitialized;
-        if (isInitialized == 1) return true;
-        if (isInitialized == 0) return false;
-
-        memoizedIsInitialized = 1;
-        return true;
-      }
-
-      @java.lang.Override
-      public void writeTo(com.google.protobuf.CodedOutputStream output)
-                          throws java.io.IOException {
-        if (priority_ != 0L) {
-          output.writeInt64(1, priority_);
-        }
-        unknownFields.writeTo(output);
-      }
-
-      @java.lang.Override
-      public int getSerializedSize() {
-        int size = memoizedSize;
-        if (size != -1) return size;
-
-        size = 0;
-        if (priority_ != 0L) {
-          size += com.google.protobuf.CodedOutputStream
-            .computeInt64Size(1, priority_);
-        }
-        size += unknownFields.getSerializedSize();
-        memoizedSize = size;
-        return size;
-      }
-
-      @java.lang.Override
-      public boolean equals(final java.lang.Object obj) {
-        if (obj == this) {
-         return true;
-        }
-        if (!(obj instanceof org.tensorflow.proto.framework.RunOptions.Experimental.RunHandlerPoolOptions)) {
-          return super.equals(obj);
-        }
-        org.tensorflow.proto.framework.RunOptions.Experimental.RunHandlerPoolOptions other = (org.tensorflow.proto.framework.RunOptions.Experimental.RunHandlerPoolOptions) obj;
-
-        if (getPriority()
-            != other.getPriority()) return false;
-        if (!unknownFields.equals(other.unknownFields)) return false;
-        return true;
-      }
-
-      @java.lang.Override
-      public int hashCode() {
-        if (memoizedHashCode != 0) {
-          return memoizedHashCode;
-        }
-        int hash = 41;
-        hash = (19 * hash) + getDescriptor().hashCode();
-        hash = (37 * hash) + PRIORITY_FIELD_NUMBER;
-        hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-            getPriority());
-        hash = (29 * hash) + unknownFields.hashCode();
-        memoizedHashCode = hash;
-        return hash;
-      }
-
-      public static org.tensorflow.proto.framework.RunOptions.Experimental.RunHandlerPoolOptions parseFrom(
-          java.nio.ByteBuffer data)
-          throws com.google.protobuf.InvalidProtocolBufferException {
-        return PARSER.parseFrom(data);
-      }
-      public static org.tensorflow.proto.framework.RunOptions.Experimental.RunHandlerPoolOptions parseFrom(
-          java.nio.ByteBuffer data,
-          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-          throws com.google.protobuf.InvalidProtocolBufferException {
-        return PARSER.parseFrom(data, extensionRegistry);
-      }
-      public static org.tensorflow.proto.framework.RunOptions.Experimental.RunHandlerPoolOptions parseFrom(
-          com.google.protobuf.ByteString data)
-          throws com.google.protobuf.InvalidProtocolBufferException {
-        return PARSER.parseFrom(data);
-      }
-      public static org.tensorflow.proto.framework.RunOptions.Experimental.RunHandlerPoolOptions parseFrom(
-          com.google.protobuf.ByteString data,
-          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-          throws com.google.protobuf.InvalidProtocolBufferException {
-        return PARSER.parseFrom(data, extensionRegistry);
-      }
-      public static org.tensorflow.proto.framework.RunOptions.Experimental.RunHandlerPoolOptions parseFrom(byte[] data)
-          throws com.google.protobuf.InvalidProtocolBufferException {
-        return PARSER.parseFrom(data);
-      }
-      public static org.tensorflow.proto.framework.RunOptions.Experimental.RunHandlerPoolOptions parseFrom(
-          byte[] data,
-          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-          throws com.google.protobuf.InvalidProtocolBufferException {
-        return PARSER.parseFrom(data, extensionRegistry);
-      }
-      public static org.tensorflow.proto.framework.RunOptions.Experimental.RunHandlerPoolOptions parseFrom(java.io.InputStream input)
-          throws java.io.IOException {
-        return com.google.protobuf.GeneratedMessageV3
-            .parseWithIOException(PARSER, input);
-      }
-      public static org.tensorflow.proto.framework.RunOptions.Experimental.RunHandlerPoolOptions parseFrom(
-          java.io.InputStream input,
-          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-          throws java.io.IOException {
-        return com.google.protobuf.GeneratedMessageV3
-            .parseWithIOException(PARSER, input, extensionRegistry);
-      }
-      public static org.tensorflow.proto.framework.RunOptions.Experimental.RunHandlerPoolOptions parseDelimitedFrom(java.io.InputStream input)
-          throws java.io.IOException {
-        return com.google.protobuf.GeneratedMessageV3
-            .parseDelimitedWithIOException(PARSER, input);
-      }
-      public static org.tensorflow.proto.framework.RunOptions.Experimental.RunHandlerPoolOptions parseDelimitedFrom(
-          java.io.InputStream input,
-          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-          throws java.io.IOException {
-        return com.google.protobuf.GeneratedMessageV3
-            .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-      }
-      public static org.tensorflow.proto.framework.RunOptions.Experimental.RunHandlerPoolOptions parseFrom(
-          com.google.protobuf.CodedInputStream input)
-          throws java.io.IOException {
-        return com.google.protobuf.GeneratedMessageV3
-            .parseWithIOException(PARSER, input);
-      }
-      public static org.tensorflow.proto.framework.RunOptions.Experimental.RunHandlerPoolOptions parseFrom(
-          com.google.protobuf.CodedInputStream input,
-          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-          throws java.io.IOException {
-        return com.google.protobuf.GeneratedMessageV3
-            .parseWithIOException(PARSER, input, extensionRegistry);
-      }
-
-      @java.lang.Override
-      public Builder newBuilderForType() { return newBuilder(); }
-      public static Builder newBuilder() {
-        return DEFAULT_INSTANCE.toBuilder();
-      }
-      public static Builder newBuilder(org.tensorflow.proto.framework.RunOptions.Experimental.RunHandlerPoolOptions prototype) {
-        return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-      }
-      @java.lang.Override
-      public Builder toBuilder() {
-        return this == DEFAULT_INSTANCE
-            ? new Builder() : new Builder().mergeFrom(this);
-      }
-
-      @java.lang.Override
-      protected Builder newBuilderForType(
-          com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-        Builder builder = new Builder(parent);
-        return builder;
-      }
-      /**
-       * <pre>
-       * Options for run handler thread pool.
-       * </pre>
-       *
-       * Protobuf type {@code tensorflow.RunOptions.Experimental.RunHandlerPoolOptions}
-       */
-      public static final class Builder extends
-          com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-          // @@protoc_insertion_point(builder_implements:tensorflow.RunOptions.Experimental.RunHandlerPoolOptions)
-          org.tensorflow.proto.framework.RunOptions.Experimental.RunHandlerPoolOptionsOrBuilder {
-        public static final com.google.protobuf.Descriptors.Descriptor
-            getDescriptor() {
-          return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_RunOptions_Experimental_RunHandlerPoolOptions_descriptor;
-        }
-
-        @java.lang.Override
-        protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-            internalGetFieldAccessorTable() {
-          return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_RunOptions_Experimental_RunHandlerPoolOptions_fieldAccessorTable
-              .ensureFieldAccessorsInitialized(
-                  org.tensorflow.proto.framework.RunOptions.Experimental.RunHandlerPoolOptions.class, org.tensorflow.proto.framework.RunOptions.Experimental.RunHandlerPoolOptions.Builder.class);
-        }
-
-        // Construct using org.tensorflow.proto.framework.RunOptions.Experimental.RunHandlerPoolOptions.newBuilder()
-        private Builder() {
-          maybeForceBuilderInitialization();
-        }
-
-        private Builder(
-            com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-          super(parent);
-          maybeForceBuilderInitialization();
-        }
-        private void maybeForceBuilderInitialization() {
-          if (com.google.protobuf.GeneratedMessageV3
-                  .alwaysUseFieldBuilders) {
-          }
-        }
-        @java.lang.Override
-        public Builder clear() {
-          super.clear();
-          priority_ = 0L;
-
-          return this;
-        }
-
-        @java.lang.Override
-        public com.google.protobuf.Descriptors.Descriptor
-            getDescriptorForType() {
-          return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_RunOptions_Experimental_RunHandlerPoolOptions_descriptor;
-        }
-
-        @java.lang.Override
-        public org.tensorflow.proto.framework.RunOptions.Experimental.RunHandlerPoolOptions getDefaultInstanceForType() {
-          return org.tensorflow.proto.framework.RunOptions.Experimental.RunHandlerPoolOptions.getDefaultInstance();
-        }
-
-        @java.lang.Override
-        public org.tensorflow.proto.framework.RunOptions.Experimental.RunHandlerPoolOptions build() {
-          org.tensorflow.proto.framework.RunOptions.Experimental.RunHandlerPoolOptions result = buildPartial();
-          if (!result.isInitialized()) {
-            throw newUninitializedMessageException(result);
-          }
-          return result;
-        }
-
-        @java.lang.Override
-        public org.tensorflow.proto.framework.RunOptions.Experimental.RunHandlerPoolOptions buildPartial() {
-          org.tensorflow.proto.framework.RunOptions.Experimental.RunHandlerPoolOptions result = new org.tensorflow.proto.framework.RunOptions.Experimental.RunHandlerPoolOptions(this);
-          result.priority_ = priority_;
-          onBuilt();
-          return result;
-        }
-
-        @java.lang.Override
-        public Builder clone() {
-          return super.clone();
-        }
-        @java.lang.Override
-        public Builder setField(
-            com.google.protobuf.Descriptors.FieldDescriptor field,
-            java.lang.Object value) {
-          return super.setField(field, value);
-        }
-        @java.lang.Override
-        public Builder clearField(
-            com.google.protobuf.Descriptors.FieldDescriptor field) {
-          return super.clearField(field);
-        }
-        @java.lang.Override
-        public Builder clearOneof(
-            com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-          return super.clearOneof(oneof);
-        }
-        @java.lang.Override
-        public Builder setRepeatedField(
-            com.google.protobuf.Descriptors.FieldDescriptor field,
-            int index, java.lang.Object value) {
-          return super.setRepeatedField(field, index, value);
-        }
-        @java.lang.Override
-        public Builder addRepeatedField(
-            com.google.protobuf.Descriptors.FieldDescriptor field,
-            java.lang.Object value) {
-          return super.addRepeatedField(field, value);
-        }
-        @java.lang.Override
-        public Builder mergeFrom(com.google.protobuf.Message other) {
-          if (other instanceof org.tensorflow.proto.framework.RunOptions.Experimental.RunHandlerPoolOptions) {
-            return mergeFrom((org.tensorflow.proto.framework.RunOptions.Experimental.RunHandlerPoolOptions)other);
-          } else {
-            super.mergeFrom(other);
-            return this;
-          }
-        }
-
-        public Builder mergeFrom(org.tensorflow.proto.framework.RunOptions.Experimental.RunHandlerPoolOptions other) {
-          if (other == org.tensorflow.proto.framework.RunOptions.Experimental.RunHandlerPoolOptions.getDefaultInstance()) return this;
-          if (other.getPriority() != 0L) {
-            setPriority(other.getPriority());
-          }
-          this.mergeUnknownFields(other.unknownFields);
-          onChanged();
-          return this;
-        }
-
-        @java.lang.Override
-        public final boolean isInitialized() {
-          return true;
-        }
-
-        @java.lang.Override
-        public Builder mergeFrom(
-            com.google.protobuf.CodedInputStream input,
-            com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-            throws java.io.IOException {
-          org.tensorflow.proto.framework.RunOptions.Experimental.RunHandlerPoolOptions parsedMessage = null;
-          try {
-            parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-          } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-            parsedMessage = (org.tensorflow.proto.framework.RunOptions.Experimental.RunHandlerPoolOptions) e.getUnfinishedMessage();
-            throw e.unwrapIOException();
-          } finally {
-            if (parsedMessage != null) {
-              mergeFrom(parsedMessage);
-            }
-          }
-          return this;
-        }
-
-        private long priority_ ;
-        /**
-         * <pre>
-         * Priority of the request. The run handler thread pool will schedule ops
-         * based on the priority number. The larger number means higher priority.
-         * </pre>
-         *
-         * <code>int64 priority = 1;</code>
-         */
-        public long getPriority() {
-          return priority_;
-        }
-        /**
-         * <pre>
-         * Priority of the request. The run handler thread pool will schedule ops
-         * based on the priority number. The larger number means higher priority.
-         * </pre>
-         *
-         * <code>int64 priority = 1;</code>
-         */
-        public Builder setPriority(long value) {
-          
-          priority_ = value;
-          onChanged();
-          return this;
-        }
-        /**
-         * <pre>
-         * Priority of the request. The run handler thread pool will schedule ops
-         * based on the priority number. The larger number means higher priority.
-         * </pre>
-         *
-         * <code>int64 priority = 1;</code>
-         */
-        public Builder clearPriority() {
-          
-          priority_ = 0L;
-          onChanged();
-          return this;
-        }
-        @java.lang.Override
-        public final Builder setUnknownFields(
-            final com.google.protobuf.UnknownFieldSet unknownFields) {
-          return super.setUnknownFields(unknownFields);
-        }
-
-        @java.lang.Override
-        public final Builder mergeUnknownFields(
-            final com.google.protobuf.UnknownFieldSet unknownFields) {
-          return super.mergeUnknownFields(unknownFields);
-        }
-
-
-        // @@protoc_insertion_point(builder_scope:tensorflow.RunOptions.Experimental.RunHandlerPoolOptions)
-      }
-
-      // @@protoc_insertion_point(class_scope:tensorflow.RunOptions.Experimental.RunHandlerPoolOptions)
-      private static final org.tensorflow.proto.framework.RunOptions.Experimental.RunHandlerPoolOptions DEFAULT_INSTANCE;
-      static {
-        DEFAULT_INSTANCE = new org.tensorflow.proto.framework.RunOptions.Experimental.RunHandlerPoolOptions();
-      }
-
-      public static org.tensorflow.proto.framework.RunOptions.Experimental.RunHandlerPoolOptions getDefaultInstance() {
-        return DEFAULT_INSTANCE;
-      }
-
-      private static final com.google.protobuf.Parser<RunHandlerPoolOptions>
-          PARSER = new com.google.protobuf.AbstractParser<RunHandlerPoolOptions>() {
-        @java.lang.Override
-        public RunHandlerPoolOptions parsePartialFrom(
-            com.google.protobuf.CodedInputStream input,
-            com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-            throws com.google.protobuf.InvalidProtocolBufferException {
-          return new RunHandlerPoolOptions(input, extensionRegistry);
-        }
-      };
-
-      public static com.google.protobuf.Parser<RunHandlerPoolOptions> parser() {
-        return PARSER;
-      }
-
-      @java.lang.Override
-      public com.google.protobuf.Parser<RunHandlerPoolOptions> getParserForType() {
-        return PARSER;
-      }
-
-      @java.lang.Override
-      public org.tensorflow.proto.framework.RunOptions.Experimental.RunHandlerPoolOptions getDefaultInstanceForType() {
-        return DEFAULT_INSTANCE;
-      }
-
-    }
-
-    public static final int COLLECTIVE_GRAPH_KEY_FIELD_NUMBER = 1;
-    private long collectiveGraphKey_;
-    /**
-     * <pre>
-     * If non-zero, declares that this graph is going to use collective
-     * ops and must synchronize step_ids with any other graph with this
-     * same group_key value (in a distributed computation where tasks
-     * run disjoint graphs).
-     * </pre>
-     *
-     * <code>int64 collective_graph_key = 1;</code>
-     */
-    public long getCollectiveGraphKey() {
-      return collectiveGraphKey_;
-    }
-
-    public static final int USE_RUN_HANDLER_POOL_FIELD_NUMBER = 2;
-    private boolean useRunHandlerPool_;
-    /**
-     * <pre>
-     * If true, then operations (using the inter-op pool) across all
-     * session::run() calls will be centrally scheduled, optimizing for (median
-     * and tail) latency.
-     * Consider using this option for CPU-bound workloads like inference.
-     * </pre>
-     *
-     * <code>bool use_run_handler_pool = 2;</code>
-     */
-    public boolean getUseRunHandlerPool() {
-      return useRunHandlerPool_;
-    }
-
-    public static final int RUN_HANDLER_POOL_OPTIONS_FIELD_NUMBER = 3;
-    private org.tensorflow.proto.framework.RunOptions.Experimental.RunHandlerPoolOptions runHandlerPoolOptions_;
-    /**
-     * <code>.tensorflow.RunOptions.Experimental.RunHandlerPoolOptions run_handler_pool_options = 3;</code>
-     */
-    public boolean hasRunHandlerPoolOptions() {
-      return runHandlerPoolOptions_ != null;
-    }
-    /**
-     * <code>.tensorflow.RunOptions.Experimental.RunHandlerPoolOptions run_handler_pool_options = 3;</code>
-     */
-    public org.tensorflow.proto.framework.RunOptions.Experimental.RunHandlerPoolOptions getRunHandlerPoolOptions() {
-      return runHandlerPoolOptions_ == null ? org.tensorflow.proto.framework.RunOptions.Experimental.RunHandlerPoolOptions.getDefaultInstance() : runHandlerPoolOptions_;
-    }
-    /**
-     * <code>.tensorflow.RunOptions.Experimental.RunHandlerPoolOptions run_handler_pool_options = 3;</code>
-     */
-    public org.tensorflow.proto.framework.RunOptions.Experimental.RunHandlerPoolOptionsOrBuilder getRunHandlerPoolOptionsOrBuilder() {
-      return getRunHandlerPoolOptions();
-    }
-
-    private byte memoizedIsInitialized = -1;
-    @java.lang.Override
-    public final boolean isInitialized() {
-      byte isInitialized = memoizedIsInitialized;
-      if (isInitialized == 1) return true;
-      if (isInitialized == 0) return false;
-
-      memoizedIsInitialized = 1;
-      return true;
-    }
-
-    @java.lang.Override
-    public void writeTo(com.google.protobuf.CodedOutputStream output)
-                        throws java.io.IOException {
-      if (collectiveGraphKey_ != 0L) {
-        output.writeInt64(1, collectiveGraphKey_);
-      }
-      if (useRunHandlerPool_ != false) {
-        output.writeBool(2, useRunHandlerPool_);
-      }
-      if (runHandlerPoolOptions_ != null) {
-        output.writeMessage(3, getRunHandlerPoolOptions());
-      }
-      unknownFields.writeTo(output);
-    }
-
-    @java.lang.Override
-    public int getSerializedSize() {
-      int size = memoizedSize;
-      if (size != -1) return size;
-
-      size = 0;
-      if (collectiveGraphKey_ != 0L) {
-        size += com.google.protobuf.CodedOutputStream
-          .computeInt64Size(1, collectiveGraphKey_);
-      }
-      if (useRunHandlerPool_ != false) {
-        size += com.google.protobuf.CodedOutputStream
-          .computeBoolSize(2, useRunHandlerPool_);
-      }
-      if (runHandlerPoolOptions_ != null) {
-        size += com.google.protobuf.CodedOutputStream
-          .computeMessageSize(3, getRunHandlerPoolOptions());
-      }
-      size += unknownFields.getSerializedSize();
-      memoizedSize = size;
-      return size;
-    }
-
-    @java.lang.Override
-    public boolean equals(final java.lang.Object obj) {
-      if (obj == this) {
-       return true;
-      }
-      if (!(obj instanceof org.tensorflow.proto.framework.RunOptions.Experimental)) {
-        return super.equals(obj);
-      }
-      org.tensorflow.proto.framework.RunOptions.Experimental other = (org.tensorflow.proto.framework.RunOptions.Experimental) obj;
-
-      if (getCollectiveGraphKey()
-          != other.getCollectiveGraphKey()) return false;
-      if (getUseRunHandlerPool()
-          != other.getUseRunHandlerPool()) return false;
-      if (hasRunHandlerPoolOptions() != other.hasRunHandlerPoolOptions()) return false;
-      if (hasRunHandlerPoolOptions()) {
-        if (!getRunHandlerPoolOptions()
-            .equals(other.getRunHandlerPoolOptions())) return false;
-      }
-      if (!unknownFields.equals(other.unknownFields)) return false;
-      return true;
-    }
-
-    @java.lang.Override
-    public int hashCode() {
-      if (memoizedHashCode != 0) {
-        return memoizedHashCode;
-      }
-      int hash = 41;
-      hash = (19 * hash) + getDescriptor().hashCode();
-      hash = (37 * hash) + COLLECTIVE_GRAPH_KEY_FIELD_NUMBER;
-      hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-          getCollectiveGraphKey());
-      hash = (37 * hash) + USE_RUN_HANDLER_POOL_FIELD_NUMBER;
-      hash = (53 * hash) + com.google.protobuf.Internal.hashBoolean(
-          getUseRunHandlerPool());
-      if (hasRunHandlerPoolOptions()) {
-        hash = (37 * hash) + RUN_HANDLER_POOL_OPTIONS_FIELD_NUMBER;
-        hash = (53 * hash) + getRunHandlerPoolOptions().hashCode();
-      }
-      hash = (29 * hash) + unknownFields.hashCode();
-      memoizedHashCode = hash;
-      return hash;
-    }
-
-    public static org.tensorflow.proto.framework.RunOptions.Experimental parseFrom(
-        java.nio.ByteBuffer data)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data);
-    }
-    public static org.tensorflow.proto.framework.RunOptions.Experimental parseFrom(
-        java.nio.ByteBuffer data,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.RunOptions.Experimental parseFrom(
-        com.google.protobuf.ByteString data)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data);
-    }
-    public static org.tensorflow.proto.framework.RunOptions.Experimental parseFrom(
-        com.google.protobuf.ByteString data,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.RunOptions.Experimental parseFrom(byte[] data)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data);
-    }
-    public static org.tensorflow.proto.framework.RunOptions.Experimental parseFrom(
-        byte[] data,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.RunOptions.Experimental parseFrom(java.io.InputStream input)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input);
-    }
-    public static org.tensorflow.proto.framework.RunOptions.Experimental parseFrom(
-        java.io.InputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.RunOptions.Experimental parseDelimitedFrom(java.io.InputStream input)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseDelimitedWithIOException(PARSER, input);
-    }
-    public static org.tensorflow.proto.framework.RunOptions.Experimental parseDelimitedFrom(
-        java.io.InputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.RunOptions.Experimental parseFrom(
-        com.google.protobuf.CodedInputStream input)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input);
-    }
-    public static org.tensorflow.proto.framework.RunOptions.Experimental parseFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input, extensionRegistry);
-    }
-
-    @java.lang.Override
-    public Builder newBuilderForType() { return newBuilder(); }
-    public static Builder newBuilder() {
-      return DEFAULT_INSTANCE.toBuilder();
-    }
-    public static Builder newBuilder(org.tensorflow.proto.framework.RunOptions.Experimental prototype) {
-      return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-    }
-    @java.lang.Override
-    public Builder toBuilder() {
-      return this == DEFAULT_INSTANCE
-          ? new Builder() : new Builder().mergeFrom(this);
-    }
-
-    @java.lang.Override
-    protected Builder newBuilderForType(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      Builder builder = new Builder(parent);
-      return builder;
-    }
-    /**
-     * <pre>
-     * Everything inside Experimental is subject to change and is not subject
-     * to API stability guarantees in
-     * https://www.tensorflow.org/guide/version_compat.
-     * </pre>
-     *
-     * Protobuf type {@code tensorflow.RunOptions.Experimental}
-     */
-    public static final class Builder extends
-        com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-        // @@protoc_insertion_point(builder_implements:tensorflow.RunOptions.Experimental)
-        org.tensorflow.proto.framework.RunOptions.ExperimentalOrBuilder {
-      public static final com.google.protobuf.Descriptors.Descriptor
-          getDescriptor() {
-        return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_RunOptions_Experimental_descriptor;
-      }
-
-      @java.lang.Override
-      protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-          internalGetFieldAccessorTable() {
-        return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_RunOptions_Experimental_fieldAccessorTable
-            .ensureFieldAccessorsInitialized(
-                org.tensorflow.proto.framework.RunOptions.Experimental.class, org.tensorflow.proto.framework.RunOptions.Experimental.Builder.class);
-      }
-
-      // Construct using org.tensorflow.proto.framework.RunOptions.Experimental.newBuilder()
-      private Builder() {
-        maybeForceBuilderInitialization();
-      }
-
-      private Builder(
-          com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-        super(parent);
-        maybeForceBuilderInitialization();
-      }
-      private void maybeForceBuilderInitialization() {
-        if (com.google.protobuf.GeneratedMessageV3
-                .alwaysUseFieldBuilders) {
-        }
-      }
-      @java.lang.Override
-      public Builder clear() {
-        super.clear();
-        collectiveGraphKey_ = 0L;
-
-        useRunHandlerPool_ = false;
-
-        if (runHandlerPoolOptionsBuilder_ == null) {
-          runHandlerPoolOptions_ = null;
-        } else {
-          runHandlerPoolOptions_ = null;
-          runHandlerPoolOptionsBuilder_ = null;
-        }
-        return this;
-      }
-
-      @java.lang.Override
-      public com.google.protobuf.Descriptors.Descriptor
-          getDescriptorForType() {
-        return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_RunOptions_Experimental_descriptor;
-      }
-
-      @java.lang.Override
-      public org.tensorflow.proto.framework.RunOptions.Experimental getDefaultInstanceForType() {
-        return org.tensorflow.proto.framework.RunOptions.Experimental.getDefaultInstance();
-      }
-
-      @java.lang.Override
-      public org.tensorflow.proto.framework.RunOptions.Experimental build() {
-        org.tensorflow.proto.framework.RunOptions.Experimental result = buildPartial();
-        if (!result.isInitialized()) {
-          throw newUninitializedMessageException(result);
-        }
-        return result;
-      }
-
-      @java.lang.Override
-      public org.tensorflow.proto.framework.RunOptions.Experimental buildPartial() {
-        org.tensorflow.proto.framework.RunOptions.Experimental result = new org.tensorflow.proto.framework.RunOptions.Experimental(this);
-        result.collectiveGraphKey_ = collectiveGraphKey_;
-        result.useRunHandlerPool_ = useRunHandlerPool_;
-        if (runHandlerPoolOptionsBuilder_ == null) {
-          result.runHandlerPoolOptions_ = runHandlerPoolOptions_;
-        } else {
-          result.runHandlerPoolOptions_ = runHandlerPoolOptionsBuilder_.build();
-        }
-        onBuilt();
-        return result;
-      }
-
-      @java.lang.Override
-      public Builder clone() {
-        return super.clone();
-      }
-      @java.lang.Override
-      public Builder setField(
-          com.google.protobuf.Descriptors.FieldDescriptor field,
-          java.lang.Object value) {
-        return super.setField(field, value);
-      }
-      @java.lang.Override
-      public Builder clearField(
-          com.google.protobuf.Descriptors.FieldDescriptor field) {
-        return super.clearField(field);
-      }
-      @java.lang.Override
-      public Builder clearOneof(
-          com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-        return super.clearOneof(oneof);
-      }
-      @java.lang.Override
-      public Builder setRepeatedField(
-          com.google.protobuf.Descriptors.FieldDescriptor field,
-          int index, java.lang.Object value) {
-        return super.setRepeatedField(field, index, value);
-      }
-      @java.lang.Override
-      public Builder addRepeatedField(
-          com.google.protobuf.Descriptors.FieldDescriptor field,
-          java.lang.Object value) {
-        return super.addRepeatedField(field, value);
-      }
-      @java.lang.Override
-      public Builder mergeFrom(com.google.protobuf.Message other) {
-        if (other instanceof org.tensorflow.proto.framework.RunOptions.Experimental) {
-          return mergeFrom((org.tensorflow.proto.framework.RunOptions.Experimental)other);
-        } else {
-          super.mergeFrom(other);
-          return this;
-        }
-      }
-
-      public Builder mergeFrom(org.tensorflow.proto.framework.RunOptions.Experimental other) {
-        if (other == org.tensorflow.proto.framework.RunOptions.Experimental.getDefaultInstance()) return this;
-        if (other.getCollectiveGraphKey() != 0L) {
-          setCollectiveGraphKey(other.getCollectiveGraphKey());
-        }
-        if (other.getUseRunHandlerPool() != false) {
-          setUseRunHandlerPool(other.getUseRunHandlerPool());
-        }
-        if (other.hasRunHandlerPoolOptions()) {
-          mergeRunHandlerPoolOptions(other.getRunHandlerPoolOptions());
-        }
-        this.mergeUnknownFields(other.unknownFields);
-        onChanged();
-        return this;
-      }
-
-      @java.lang.Override
-      public final boolean isInitialized() {
-        return true;
-      }
-
-      @java.lang.Override
-      public Builder mergeFrom(
-          com.google.protobuf.CodedInputStream input,
-          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-          throws java.io.IOException {
-        org.tensorflow.proto.framework.RunOptions.Experimental parsedMessage = null;
-        try {
-          parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-        } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-          parsedMessage = (org.tensorflow.proto.framework.RunOptions.Experimental) e.getUnfinishedMessage();
-          throw e.unwrapIOException();
-        } finally {
-          if (parsedMessage != null) {
-            mergeFrom(parsedMessage);
-          }
-        }
-        return this;
-      }
-
-      private long collectiveGraphKey_ ;
-      /**
-       * <pre>
-       * If non-zero, declares that this graph is going to use collective
-       * ops and must synchronize step_ids with any other graph with this
-       * same group_key value (in a distributed computation where tasks
-       * run disjoint graphs).
-       * </pre>
-       *
-       * <code>int64 collective_graph_key = 1;</code>
-       */
-      public long getCollectiveGraphKey() {
-        return collectiveGraphKey_;
-      }
-      /**
-       * <pre>
-       * If non-zero, declares that this graph is going to use collective
-       * ops and must synchronize step_ids with any other graph with this
-       * same group_key value (in a distributed computation where tasks
-       * run disjoint graphs).
-       * </pre>
-       *
-       * <code>int64 collective_graph_key = 1;</code>
-       */
-      public Builder setCollectiveGraphKey(long value) {
-        
-        collectiveGraphKey_ = value;
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * If non-zero, declares that this graph is going to use collective
-       * ops and must synchronize step_ids with any other graph with this
-       * same group_key value (in a distributed computation where tasks
-       * run disjoint graphs).
-       * </pre>
-       *
-       * <code>int64 collective_graph_key = 1;</code>
-       */
-      public Builder clearCollectiveGraphKey() {
-        
-        collectiveGraphKey_ = 0L;
-        onChanged();
-        return this;
-      }
-
-      private boolean useRunHandlerPool_ ;
-      /**
-       * <pre>
-       * If true, then operations (using the inter-op pool) across all
-       * session::run() calls will be centrally scheduled, optimizing for (median
-       * and tail) latency.
-       * Consider using this option for CPU-bound workloads like inference.
-       * </pre>
-       *
-       * <code>bool use_run_handler_pool = 2;</code>
-       */
-      public boolean getUseRunHandlerPool() {
-        return useRunHandlerPool_;
-      }
-      /**
-       * <pre>
-       * If true, then operations (using the inter-op pool) across all
-       * session::run() calls will be centrally scheduled, optimizing for (median
-       * and tail) latency.
-       * Consider using this option for CPU-bound workloads like inference.
-       * </pre>
-       *
-       * <code>bool use_run_handler_pool = 2;</code>
-       */
-      public Builder setUseRunHandlerPool(boolean value) {
-        
-        useRunHandlerPool_ = value;
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * If true, then operations (using the inter-op pool) across all
-       * session::run() calls will be centrally scheduled, optimizing for (median
-       * and tail) latency.
-       * Consider using this option for CPU-bound workloads like inference.
-       * </pre>
-       *
-       * <code>bool use_run_handler_pool = 2;</code>
-       */
-      public Builder clearUseRunHandlerPool() {
-        
-        useRunHandlerPool_ = false;
-        onChanged();
-        return this;
-      }
-
-      private org.tensorflow.proto.framework.RunOptions.Experimental.RunHandlerPoolOptions runHandlerPoolOptions_;
-      private com.google.protobuf.SingleFieldBuilderV3<
-          org.tensorflow.proto.framework.RunOptions.Experimental.RunHandlerPoolOptions, org.tensorflow.proto.framework.RunOptions.Experimental.RunHandlerPoolOptions.Builder, org.tensorflow.proto.framework.RunOptions.Experimental.RunHandlerPoolOptionsOrBuilder> runHandlerPoolOptionsBuilder_;
-      /**
-       * <code>.tensorflow.RunOptions.Experimental.RunHandlerPoolOptions run_handler_pool_options = 3;</code>
-       */
-      public boolean hasRunHandlerPoolOptions() {
-        return runHandlerPoolOptionsBuilder_ != null || runHandlerPoolOptions_ != null;
-      }
-      /**
-       * <code>.tensorflow.RunOptions.Experimental.RunHandlerPoolOptions run_handler_pool_options = 3;</code>
-       */
-      public org.tensorflow.proto.framework.RunOptions.Experimental.RunHandlerPoolOptions getRunHandlerPoolOptions() {
-        if (runHandlerPoolOptionsBuilder_ == null) {
-          return runHandlerPoolOptions_ == null ? org.tensorflow.proto.framework.RunOptions.Experimental.RunHandlerPoolOptions.getDefaultInstance() : runHandlerPoolOptions_;
-        } else {
-          return runHandlerPoolOptionsBuilder_.getMessage();
-        }
-      }
-      /**
-       * <code>.tensorflow.RunOptions.Experimental.RunHandlerPoolOptions run_handler_pool_options = 3;</code>
-       */
-      public Builder setRunHandlerPoolOptions(org.tensorflow.proto.framework.RunOptions.Experimental.RunHandlerPoolOptions value) {
-        if (runHandlerPoolOptionsBuilder_ == null) {
-          if (value == null) {
-            throw new NullPointerException();
-          }
-          runHandlerPoolOptions_ = value;
-          onChanged();
-        } else {
-          runHandlerPoolOptionsBuilder_.setMessage(value);
-        }
-
-        return this;
-      }
-      /**
-       * <code>.tensorflow.RunOptions.Experimental.RunHandlerPoolOptions run_handler_pool_options = 3;</code>
-       */
-      public Builder setRunHandlerPoolOptions(
-          org.tensorflow.proto.framework.RunOptions.Experimental.RunHandlerPoolOptions.Builder builderForValue) {
-        if (runHandlerPoolOptionsBuilder_ == null) {
-          runHandlerPoolOptions_ = builderForValue.build();
-          onChanged();
-        } else {
-          runHandlerPoolOptionsBuilder_.setMessage(builderForValue.build());
-        }
-
-        return this;
-      }
-      /**
-       * <code>.tensorflow.RunOptions.Experimental.RunHandlerPoolOptions run_handler_pool_options = 3;</code>
-       */
-      public Builder mergeRunHandlerPoolOptions(org.tensorflow.proto.framework.RunOptions.Experimental.RunHandlerPoolOptions value) {
-        if (runHandlerPoolOptionsBuilder_ == null) {
-          if (runHandlerPoolOptions_ != null) {
-            runHandlerPoolOptions_ =
-              org.tensorflow.proto.framework.RunOptions.Experimental.RunHandlerPoolOptions.newBuilder(runHandlerPoolOptions_).mergeFrom(value).buildPartial();
-          } else {
-            runHandlerPoolOptions_ = value;
-          }
-          onChanged();
-        } else {
-          runHandlerPoolOptionsBuilder_.mergeFrom(value);
-        }
-
-        return this;
-      }
-      /**
-       * <code>.tensorflow.RunOptions.Experimental.RunHandlerPoolOptions run_handler_pool_options = 3;</code>
-       */
-      public Builder clearRunHandlerPoolOptions() {
-        if (runHandlerPoolOptionsBuilder_ == null) {
-          runHandlerPoolOptions_ = null;
-          onChanged();
-        } else {
-          runHandlerPoolOptions_ = null;
-          runHandlerPoolOptionsBuilder_ = null;
-        }
-
-        return this;
-      }
-      /**
-       * <code>.tensorflow.RunOptions.Experimental.RunHandlerPoolOptions run_handler_pool_options = 3;</code>
-       */
-      public org.tensorflow.proto.framework.RunOptions.Experimental.RunHandlerPoolOptions.Builder getRunHandlerPoolOptionsBuilder() {
-        
-        onChanged();
-        return getRunHandlerPoolOptionsFieldBuilder().getBuilder();
-      }
-      /**
-       * <code>.tensorflow.RunOptions.Experimental.RunHandlerPoolOptions run_handler_pool_options = 3;</code>
-       */
-      public org.tensorflow.proto.framework.RunOptions.Experimental.RunHandlerPoolOptionsOrBuilder getRunHandlerPoolOptionsOrBuilder() {
-        if (runHandlerPoolOptionsBuilder_ != null) {
-          return runHandlerPoolOptionsBuilder_.getMessageOrBuilder();
-        } else {
-          return runHandlerPoolOptions_ == null ?
-              org.tensorflow.proto.framework.RunOptions.Experimental.RunHandlerPoolOptions.getDefaultInstance() : runHandlerPoolOptions_;
-        }
-      }
-      /**
-       * <code>.tensorflow.RunOptions.Experimental.RunHandlerPoolOptions run_handler_pool_options = 3;</code>
-       */
-      private com.google.protobuf.SingleFieldBuilderV3<
-          org.tensorflow.proto.framework.RunOptions.Experimental.RunHandlerPoolOptions, org.tensorflow.proto.framework.RunOptions.Experimental.RunHandlerPoolOptions.Builder, org.tensorflow.proto.framework.RunOptions.Experimental.RunHandlerPoolOptionsOrBuilder> 
-          getRunHandlerPoolOptionsFieldBuilder() {
-        if (runHandlerPoolOptionsBuilder_ == null) {
-          runHandlerPoolOptionsBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-              org.tensorflow.proto.framework.RunOptions.Experimental.RunHandlerPoolOptions, org.tensorflow.proto.framework.RunOptions.Experimental.RunHandlerPoolOptions.Builder, org.tensorflow.proto.framework.RunOptions.Experimental.RunHandlerPoolOptionsOrBuilder>(
-                  getRunHandlerPoolOptions(),
-                  getParentForChildren(),
-                  isClean());
-          runHandlerPoolOptions_ = null;
-        }
-        return runHandlerPoolOptionsBuilder_;
-      }
-      @java.lang.Override
-      public final Builder setUnknownFields(
-          final com.google.protobuf.UnknownFieldSet unknownFields) {
-        return super.setUnknownFields(unknownFields);
-      }
-
-      @java.lang.Override
-      public final Builder mergeUnknownFields(
-          final com.google.protobuf.UnknownFieldSet unknownFields) {
-        return super.mergeUnknownFields(unknownFields);
-      }
-
-
-      // @@protoc_insertion_point(builder_scope:tensorflow.RunOptions.Experimental)
-    }
-
-    // @@protoc_insertion_point(class_scope:tensorflow.RunOptions.Experimental)
-    private static final org.tensorflow.proto.framework.RunOptions.Experimental DEFAULT_INSTANCE;
-    static {
-      DEFAULT_INSTANCE = new org.tensorflow.proto.framework.RunOptions.Experimental();
-    }
-
-    public static org.tensorflow.proto.framework.RunOptions.Experimental getDefaultInstance() {
-      return DEFAULT_INSTANCE;
-    }
-
-    private static final com.google.protobuf.Parser<Experimental>
-        PARSER = new com.google.protobuf.AbstractParser<Experimental>() {
-      @java.lang.Override
-      public Experimental parsePartialFrom(
-          com.google.protobuf.CodedInputStream input,
-          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-          throws com.google.protobuf.InvalidProtocolBufferException {
-        return new Experimental(input, extensionRegistry);
-      }
-    };
-
-    public static com.google.protobuf.Parser<Experimental> parser() {
-      return PARSER;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Parser<Experimental> getParserForType() {
-      return PARSER;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.RunOptions.Experimental getDefaultInstanceForType() {
-      return DEFAULT_INSTANCE;
-    }
-
-  }
-
-  public static final int TRACE_LEVEL_FIELD_NUMBER = 1;
-  private int traceLevel_;
-  /**
-   * <code>.tensorflow.RunOptions.TraceLevel trace_level = 1;</code>
-   */
-  public int getTraceLevelValue() {
-    return traceLevel_;
-  }
-  /**
-   * <code>.tensorflow.RunOptions.TraceLevel trace_level = 1;</code>
-   */
-  public org.tensorflow.proto.framework.RunOptions.TraceLevel getTraceLevel() {
-    @SuppressWarnings("deprecation")
-    org.tensorflow.proto.framework.RunOptions.TraceLevel result = org.tensorflow.proto.framework.RunOptions.TraceLevel.valueOf(traceLevel_);
-    return result == null ? org.tensorflow.proto.framework.RunOptions.TraceLevel.UNRECOGNIZED : result;
-  }
-
-  public static final int TIMEOUT_IN_MS_FIELD_NUMBER = 2;
-  private long timeoutInMs_;
-  /**
-   * <pre>
-   * Time to wait for operation to complete in milliseconds.
-   * </pre>
-   *
-   * <code>int64 timeout_in_ms = 2;</code>
-   */
-  public long getTimeoutInMs() {
-    return timeoutInMs_;
-  }
-
-  public static final int INTER_OP_THREAD_POOL_FIELD_NUMBER = 3;
-  private int interOpThreadPool_;
-  /**
-   * <pre>
-   * The thread pool to use, if session_inter_op_thread_pool is configured.
-   * To use the caller thread set this to -1 - this uses the caller thread
-   * to execute Session::Run() and thus avoids a context switch. Using the
-   * caller thread to execute Session::Run() should be done ONLY for simple
-   * graphs, where the overhead of an additional context switch is
-   * comparable with the overhead of Session::Run().
-   * </pre>
-   *
-   * <code>int32 inter_op_thread_pool = 3;</code>
-   */
-  public int getInterOpThreadPool() {
-    return interOpThreadPool_;
-  }
-
-  public static final int OUTPUT_PARTITION_GRAPHS_FIELD_NUMBER = 5;
-  private boolean outputPartitionGraphs_;
-  /**
-   * <pre>
-   * Whether the partition graph(s) executed by the executor(s) should be
-   * outputted via RunMetadata.
-   * </pre>
-   *
-   * <code>bool output_partition_graphs = 5;</code>
-   */
-  public boolean getOutputPartitionGraphs() {
-    return outputPartitionGraphs_;
-  }
-
-  public static final int DEBUG_OPTIONS_FIELD_NUMBER = 6;
-  private org.tensorflow.proto.framework.DebugOptions debugOptions_;
-  /**
-   * <pre>
-   * EXPERIMENTAL.  Options used to initialize DebuggerState, if enabled.
-   * </pre>
-   *
-   * <code>.tensorflow.DebugOptions debug_options = 6;</code>
-   */
-  public boolean hasDebugOptions() {
-    return debugOptions_ != null;
-  }
-  /**
-   * <pre>
-   * EXPERIMENTAL.  Options used to initialize DebuggerState, if enabled.
-   * </pre>
-   *
-   * <code>.tensorflow.DebugOptions debug_options = 6;</code>
-   */
-  public org.tensorflow.proto.framework.DebugOptions getDebugOptions() {
-    return debugOptions_ == null ? org.tensorflow.proto.framework.DebugOptions.getDefaultInstance() : debugOptions_;
-  }
-  /**
-   * <pre>
-   * EXPERIMENTAL.  Options used to initialize DebuggerState, if enabled.
-   * </pre>
-   *
-   * <code>.tensorflow.DebugOptions debug_options = 6;</code>
-   */
-  public org.tensorflow.proto.framework.DebugOptionsOrBuilder getDebugOptionsOrBuilder() {
-    return getDebugOptions();
-  }
-
-  public static final int REPORT_TENSOR_ALLOCATIONS_UPON_OOM_FIELD_NUMBER = 7;
-  private boolean reportTensorAllocationsUponOom_;
-  /**
-   * <pre>
-   * When enabled, causes tensor allocation information to be included in
-   * the error message when the Run() call fails because the allocator ran
-   * out of memory (OOM).
-   * Enabling this option can slow down the Run() call.
-   * </pre>
-   *
-   * <code>bool report_tensor_allocations_upon_oom = 7;</code>
-   */
-  public boolean getReportTensorAllocationsUponOom() {
-    return reportTensorAllocationsUponOom_;
-  }
-
-  public static final int EXPERIMENTAL_FIELD_NUMBER = 8;
-  private org.tensorflow.proto.framework.RunOptions.Experimental experimental_;
-  /**
-   * <code>.tensorflow.RunOptions.Experimental experimental = 8;</code>
-   */
-  public boolean hasExperimental() {
-    return experimental_ != null;
-  }
-  /**
-   * <code>.tensorflow.RunOptions.Experimental experimental = 8;</code>
-   */
-  public org.tensorflow.proto.framework.RunOptions.Experimental getExperimental() {
-    return experimental_ == null ? org.tensorflow.proto.framework.RunOptions.Experimental.getDefaultInstance() : experimental_;
-  }
-  /**
-   * <code>.tensorflow.RunOptions.Experimental experimental = 8;</code>
-   */
-  public org.tensorflow.proto.framework.RunOptions.ExperimentalOrBuilder getExperimentalOrBuilder() {
-    return getExperimental();
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    if (traceLevel_ != org.tensorflow.proto.framework.RunOptions.TraceLevel.NO_TRACE.getNumber()) {
-      output.writeEnum(1, traceLevel_);
-    }
-    if (timeoutInMs_ != 0L) {
-      output.writeInt64(2, timeoutInMs_);
-    }
-    if (interOpThreadPool_ != 0) {
-      output.writeInt32(3, interOpThreadPool_);
-    }
-    if (outputPartitionGraphs_ != false) {
-      output.writeBool(5, outputPartitionGraphs_);
-    }
-    if (debugOptions_ != null) {
-      output.writeMessage(6, getDebugOptions());
-    }
-    if (reportTensorAllocationsUponOom_ != false) {
-      output.writeBool(7, reportTensorAllocationsUponOom_);
-    }
-    if (experimental_ != null) {
-      output.writeMessage(8, getExperimental());
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (traceLevel_ != org.tensorflow.proto.framework.RunOptions.TraceLevel.NO_TRACE.getNumber()) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeEnumSize(1, traceLevel_);
-    }
-    if (timeoutInMs_ != 0L) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt64Size(2, timeoutInMs_);
-    }
-    if (interOpThreadPool_ != 0) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt32Size(3, interOpThreadPool_);
-    }
-    if (outputPartitionGraphs_ != false) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeBoolSize(5, outputPartitionGraphs_);
-    }
-    if (debugOptions_ != null) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(6, getDebugOptions());
-    }
-    if (reportTensorAllocationsUponOom_ != false) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeBoolSize(7, reportTensorAllocationsUponOom_);
-    }
-    if (experimental_ != null) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(8, getExperimental());
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.framework.RunOptions)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.framework.RunOptions other = (org.tensorflow.proto.framework.RunOptions) obj;
-
-    if (traceLevel_ != other.traceLevel_) return false;
-    if (getTimeoutInMs()
-        != other.getTimeoutInMs()) return false;
-    if (getInterOpThreadPool()
-        != other.getInterOpThreadPool()) return false;
-    if (getOutputPartitionGraphs()
-        != other.getOutputPartitionGraphs()) return false;
-    if (hasDebugOptions() != other.hasDebugOptions()) return false;
-    if (hasDebugOptions()) {
-      if (!getDebugOptions()
-          .equals(other.getDebugOptions())) return false;
-    }
-    if (getReportTensorAllocationsUponOom()
-        != other.getReportTensorAllocationsUponOom()) return false;
-    if (hasExperimental() != other.hasExperimental()) return false;
-    if (hasExperimental()) {
-      if (!getExperimental()
-          .equals(other.getExperimental())) return false;
-    }
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    hash = (37 * hash) + TRACE_LEVEL_FIELD_NUMBER;
-    hash = (53 * hash) + traceLevel_;
-    hash = (37 * hash) + TIMEOUT_IN_MS_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-        getTimeoutInMs());
-    hash = (37 * hash) + INTER_OP_THREAD_POOL_FIELD_NUMBER;
-    hash = (53 * hash) + getInterOpThreadPool();
-    hash = (37 * hash) + OUTPUT_PARTITION_GRAPHS_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashBoolean(
-        getOutputPartitionGraphs());
-    if (hasDebugOptions()) {
-      hash = (37 * hash) + DEBUG_OPTIONS_FIELD_NUMBER;
-      hash = (53 * hash) + getDebugOptions().hashCode();
-    }
-    hash = (37 * hash) + REPORT_TENSOR_ALLOCATIONS_UPON_OOM_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashBoolean(
-        getReportTensorAllocationsUponOom());
-    if (hasExperimental()) {
-      hash = (37 * hash) + EXPERIMENTAL_FIELD_NUMBER;
-      hash = (53 * hash) + getExperimental().hashCode();
-    }
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.framework.RunOptions parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.RunOptions parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.RunOptions parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.RunOptions parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.RunOptions parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.RunOptions parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.RunOptions parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.RunOptions parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.RunOptions parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.RunOptions parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.RunOptions parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.RunOptions parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.framework.RunOptions prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * <pre>
-   * Options for a single Run() call.
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.RunOptions}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.RunOptions)
-      org.tensorflow.proto.framework.RunOptionsOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_RunOptions_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_RunOptions_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.RunOptions.class, org.tensorflow.proto.framework.RunOptions.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.framework.RunOptions.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      traceLevel_ = 0;
-
-      timeoutInMs_ = 0L;
-
-      interOpThreadPool_ = 0;
-
-      outputPartitionGraphs_ = false;
-
-      if (debugOptionsBuilder_ == null) {
-        debugOptions_ = null;
-      } else {
-        debugOptions_ = null;
-        debugOptionsBuilder_ = null;
-      }
-      reportTensorAllocationsUponOom_ = false;
-
-      if (experimentalBuilder_ == null) {
-        experimental_ = null;
-      } else {
-        experimental_ = null;
-        experimentalBuilder_ = null;
-      }
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_RunOptions_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.RunOptions getDefaultInstanceForType() {
-      return org.tensorflow.proto.framework.RunOptions.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.RunOptions build() {
-      org.tensorflow.proto.framework.RunOptions result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.RunOptions buildPartial() {
-      org.tensorflow.proto.framework.RunOptions result = new org.tensorflow.proto.framework.RunOptions(this);
-      result.traceLevel_ = traceLevel_;
-      result.timeoutInMs_ = timeoutInMs_;
-      result.interOpThreadPool_ = interOpThreadPool_;
-      result.outputPartitionGraphs_ = outputPartitionGraphs_;
-      if (debugOptionsBuilder_ == null) {
-        result.debugOptions_ = debugOptions_;
-      } else {
-        result.debugOptions_ = debugOptionsBuilder_.build();
-      }
-      result.reportTensorAllocationsUponOom_ = reportTensorAllocationsUponOom_;
-      if (experimentalBuilder_ == null) {
-        result.experimental_ = experimental_;
-      } else {
-        result.experimental_ = experimentalBuilder_.build();
-      }
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.framework.RunOptions) {
-        return mergeFrom((org.tensorflow.proto.framework.RunOptions)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.framework.RunOptions other) {
-      if (other == org.tensorflow.proto.framework.RunOptions.getDefaultInstance()) return this;
-      if (other.traceLevel_ != 0) {
-        setTraceLevelValue(other.getTraceLevelValue());
-      }
-      if (other.getTimeoutInMs() != 0L) {
-        setTimeoutInMs(other.getTimeoutInMs());
-      }
-      if (other.getInterOpThreadPool() != 0) {
-        setInterOpThreadPool(other.getInterOpThreadPool());
-      }
-      if (other.getOutputPartitionGraphs() != false) {
-        setOutputPartitionGraphs(other.getOutputPartitionGraphs());
-      }
-      if (other.hasDebugOptions()) {
-        mergeDebugOptions(other.getDebugOptions());
-      }
-      if (other.getReportTensorAllocationsUponOom() != false) {
-        setReportTensorAllocationsUponOom(other.getReportTensorAllocationsUponOom());
-      }
-      if (other.hasExperimental()) {
-        mergeExperimental(other.getExperimental());
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.framework.RunOptions parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.framework.RunOptions) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-
-    private int traceLevel_ = 0;
-    /**
-     * <code>.tensorflow.RunOptions.TraceLevel trace_level = 1;</code>
-     */
-    public int getTraceLevelValue() {
-      return traceLevel_;
-    }
-    /**
-     * <code>.tensorflow.RunOptions.TraceLevel trace_level = 1;</code>
-     */
-    public Builder setTraceLevelValue(int value) {
-      traceLevel_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>.tensorflow.RunOptions.TraceLevel trace_level = 1;</code>
-     */
-    public org.tensorflow.proto.framework.RunOptions.TraceLevel getTraceLevel() {
-      @SuppressWarnings("deprecation")
-      org.tensorflow.proto.framework.RunOptions.TraceLevel result = org.tensorflow.proto.framework.RunOptions.TraceLevel.valueOf(traceLevel_);
-      return result == null ? org.tensorflow.proto.framework.RunOptions.TraceLevel.UNRECOGNIZED : result;
-    }
-    /**
-     * <code>.tensorflow.RunOptions.TraceLevel trace_level = 1;</code>
-     */
-    public Builder setTraceLevel(org.tensorflow.proto.framework.RunOptions.TraceLevel value) {
-      if (value == null) {
-        throw new NullPointerException();
-      }
-      
-      traceLevel_ = value.getNumber();
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>.tensorflow.RunOptions.TraceLevel trace_level = 1;</code>
-     */
-    public Builder clearTraceLevel() {
-      
-      traceLevel_ = 0;
-      onChanged();
-      return this;
-    }
-
-    private long timeoutInMs_ ;
-    /**
-     * <pre>
-     * Time to wait for operation to complete in milliseconds.
-     * </pre>
-     *
-     * <code>int64 timeout_in_ms = 2;</code>
-     */
-    public long getTimeoutInMs() {
-      return timeoutInMs_;
-    }
-    /**
-     * <pre>
-     * Time to wait for operation to complete in milliseconds.
-     * </pre>
-     *
-     * <code>int64 timeout_in_ms = 2;</code>
-     */
-    public Builder setTimeoutInMs(long value) {
-      
-      timeoutInMs_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Time to wait for operation to complete in milliseconds.
-     * </pre>
-     *
-     * <code>int64 timeout_in_ms = 2;</code>
-     */
-    public Builder clearTimeoutInMs() {
-      
-      timeoutInMs_ = 0L;
-      onChanged();
-      return this;
-    }
-
-    private int interOpThreadPool_ ;
-    /**
-     * <pre>
-     * The thread pool to use, if session_inter_op_thread_pool is configured.
-     * To use the caller thread set this to -1 - this uses the caller thread
-     * to execute Session::Run() and thus avoids a context switch. Using the
-     * caller thread to execute Session::Run() should be done ONLY for simple
-     * graphs, where the overhead of an additional context switch is
-     * comparable with the overhead of Session::Run().
-     * </pre>
-     *
-     * <code>int32 inter_op_thread_pool = 3;</code>
-     */
-    public int getInterOpThreadPool() {
-      return interOpThreadPool_;
-    }
-    /**
-     * <pre>
-     * The thread pool to use, if session_inter_op_thread_pool is configured.
-     * To use the caller thread set this to -1 - this uses the caller thread
-     * to execute Session::Run() and thus avoids a context switch. Using the
-     * caller thread to execute Session::Run() should be done ONLY for simple
-     * graphs, where the overhead of an additional context switch is
-     * comparable with the overhead of Session::Run().
-     * </pre>
-     *
-     * <code>int32 inter_op_thread_pool = 3;</code>
-     */
-    public Builder setInterOpThreadPool(int value) {
-      
-      interOpThreadPool_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * The thread pool to use, if session_inter_op_thread_pool is configured.
-     * To use the caller thread set this to -1 - this uses the caller thread
-     * to execute Session::Run() and thus avoids a context switch. Using the
-     * caller thread to execute Session::Run() should be done ONLY for simple
-     * graphs, where the overhead of an additional context switch is
-     * comparable with the overhead of Session::Run().
-     * </pre>
-     *
-     * <code>int32 inter_op_thread_pool = 3;</code>
-     */
-    public Builder clearInterOpThreadPool() {
-      
-      interOpThreadPool_ = 0;
-      onChanged();
-      return this;
-    }
-
-    private boolean outputPartitionGraphs_ ;
-    /**
-     * <pre>
-     * Whether the partition graph(s) executed by the executor(s) should be
-     * outputted via RunMetadata.
-     * </pre>
-     *
-     * <code>bool output_partition_graphs = 5;</code>
-     */
-    public boolean getOutputPartitionGraphs() {
-      return outputPartitionGraphs_;
-    }
-    /**
-     * <pre>
-     * Whether the partition graph(s) executed by the executor(s) should be
-     * outputted via RunMetadata.
-     * </pre>
-     *
-     * <code>bool output_partition_graphs = 5;</code>
-     */
-    public Builder setOutputPartitionGraphs(boolean value) {
-      
-      outputPartitionGraphs_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Whether the partition graph(s) executed by the executor(s) should be
-     * outputted via RunMetadata.
-     * </pre>
-     *
-     * <code>bool output_partition_graphs = 5;</code>
-     */
-    public Builder clearOutputPartitionGraphs() {
-      
-      outputPartitionGraphs_ = false;
-      onChanged();
-      return this;
-    }
-
-    private org.tensorflow.proto.framework.DebugOptions debugOptions_;
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.DebugOptions, org.tensorflow.proto.framework.DebugOptions.Builder, org.tensorflow.proto.framework.DebugOptionsOrBuilder> debugOptionsBuilder_;
-    /**
-     * <pre>
-     * EXPERIMENTAL.  Options used to initialize DebuggerState, if enabled.
-     * </pre>
-     *
-     * <code>.tensorflow.DebugOptions debug_options = 6;</code>
-     */
-    public boolean hasDebugOptions() {
-      return debugOptionsBuilder_ != null || debugOptions_ != null;
-    }
-    /**
-     * <pre>
-     * EXPERIMENTAL.  Options used to initialize DebuggerState, if enabled.
-     * </pre>
-     *
-     * <code>.tensorflow.DebugOptions debug_options = 6;</code>
-     */
-    public org.tensorflow.proto.framework.DebugOptions getDebugOptions() {
-      if (debugOptionsBuilder_ == null) {
-        return debugOptions_ == null ? org.tensorflow.proto.framework.DebugOptions.getDefaultInstance() : debugOptions_;
-      } else {
-        return debugOptionsBuilder_.getMessage();
-      }
-    }
-    /**
-     * <pre>
-     * EXPERIMENTAL.  Options used to initialize DebuggerState, if enabled.
-     * </pre>
-     *
-     * <code>.tensorflow.DebugOptions debug_options = 6;</code>
-     */
-    public Builder setDebugOptions(org.tensorflow.proto.framework.DebugOptions value) {
-      if (debugOptionsBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        debugOptions_ = value;
-        onChanged();
-      } else {
-        debugOptionsBuilder_.setMessage(value);
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * EXPERIMENTAL.  Options used to initialize DebuggerState, if enabled.
-     * </pre>
-     *
-     * <code>.tensorflow.DebugOptions debug_options = 6;</code>
-     */
-    public Builder setDebugOptions(
-        org.tensorflow.proto.framework.DebugOptions.Builder builderForValue) {
-      if (debugOptionsBuilder_ == null) {
-        debugOptions_ = builderForValue.build();
-        onChanged();
-      } else {
-        debugOptionsBuilder_.setMessage(builderForValue.build());
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * EXPERIMENTAL.  Options used to initialize DebuggerState, if enabled.
-     * </pre>
-     *
-     * <code>.tensorflow.DebugOptions debug_options = 6;</code>
-     */
-    public Builder mergeDebugOptions(org.tensorflow.proto.framework.DebugOptions value) {
-      if (debugOptionsBuilder_ == null) {
-        if (debugOptions_ != null) {
-          debugOptions_ =
-            org.tensorflow.proto.framework.DebugOptions.newBuilder(debugOptions_).mergeFrom(value).buildPartial();
-        } else {
-          debugOptions_ = value;
-        }
-        onChanged();
-      } else {
-        debugOptionsBuilder_.mergeFrom(value);
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * EXPERIMENTAL.  Options used to initialize DebuggerState, if enabled.
-     * </pre>
-     *
-     * <code>.tensorflow.DebugOptions debug_options = 6;</code>
-     */
-    public Builder clearDebugOptions() {
-      if (debugOptionsBuilder_ == null) {
-        debugOptions_ = null;
-        onChanged();
-      } else {
-        debugOptions_ = null;
-        debugOptionsBuilder_ = null;
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * EXPERIMENTAL.  Options used to initialize DebuggerState, if enabled.
-     * </pre>
-     *
-     * <code>.tensorflow.DebugOptions debug_options = 6;</code>
-     */
-    public org.tensorflow.proto.framework.DebugOptions.Builder getDebugOptionsBuilder() {
-      
-      onChanged();
-      return getDebugOptionsFieldBuilder().getBuilder();
-    }
-    /**
-     * <pre>
-     * EXPERIMENTAL.  Options used to initialize DebuggerState, if enabled.
-     * </pre>
-     *
-     * <code>.tensorflow.DebugOptions debug_options = 6;</code>
-     */
-    public org.tensorflow.proto.framework.DebugOptionsOrBuilder getDebugOptionsOrBuilder() {
-      if (debugOptionsBuilder_ != null) {
-        return debugOptionsBuilder_.getMessageOrBuilder();
-      } else {
-        return debugOptions_ == null ?
-            org.tensorflow.proto.framework.DebugOptions.getDefaultInstance() : debugOptions_;
-      }
-    }
-    /**
-     * <pre>
-     * EXPERIMENTAL.  Options used to initialize DebuggerState, if enabled.
-     * </pre>
-     *
-     * <code>.tensorflow.DebugOptions debug_options = 6;</code>
-     */
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.DebugOptions, org.tensorflow.proto.framework.DebugOptions.Builder, org.tensorflow.proto.framework.DebugOptionsOrBuilder> 
-        getDebugOptionsFieldBuilder() {
-      if (debugOptionsBuilder_ == null) {
-        debugOptionsBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-            org.tensorflow.proto.framework.DebugOptions, org.tensorflow.proto.framework.DebugOptions.Builder, org.tensorflow.proto.framework.DebugOptionsOrBuilder>(
-                getDebugOptions(),
-                getParentForChildren(),
-                isClean());
-        debugOptions_ = null;
-      }
-      return debugOptionsBuilder_;
-    }
-
-    private boolean reportTensorAllocationsUponOom_ ;
-    /**
-     * <pre>
-     * When enabled, causes tensor allocation information to be included in
-     * the error message when the Run() call fails because the allocator ran
-     * out of memory (OOM).
-     * Enabling this option can slow down the Run() call.
-     * </pre>
-     *
-     * <code>bool report_tensor_allocations_upon_oom = 7;</code>
-     */
-    public boolean getReportTensorAllocationsUponOom() {
-      return reportTensorAllocationsUponOom_;
-    }
-    /**
-     * <pre>
-     * When enabled, causes tensor allocation information to be included in
-     * the error message when the Run() call fails because the allocator ran
-     * out of memory (OOM).
-     * Enabling this option can slow down the Run() call.
-     * </pre>
-     *
-     * <code>bool report_tensor_allocations_upon_oom = 7;</code>
-     */
-    public Builder setReportTensorAllocationsUponOom(boolean value) {
-      
-      reportTensorAllocationsUponOom_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * When enabled, causes tensor allocation information to be included in
-     * the error message when the Run() call fails because the allocator ran
-     * out of memory (OOM).
-     * Enabling this option can slow down the Run() call.
-     * </pre>
-     *
-     * <code>bool report_tensor_allocations_upon_oom = 7;</code>
-     */
-    public Builder clearReportTensorAllocationsUponOom() {
-      
-      reportTensorAllocationsUponOom_ = false;
-      onChanged();
-      return this;
-    }
-
-    private org.tensorflow.proto.framework.RunOptions.Experimental experimental_;
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.RunOptions.Experimental, org.tensorflow.proto.framework.RunOptions.Experimental.Builder, org.tensorflow.proto.framework.RunOptions.ExperimentalOrBuilder> experimentalBuilder_;
-    /**
-     * <code>.tensorflow.RunOptions.Experimental experimental = 8;</code>
-     */
-    public boolean hasExperimental() {
-      return experimentalBuilder_ != null || experimental_ != null;
-    }
-    /**
-     * <code>.tensorflow.RunOptions.Experimental experimental = 8;</code>
-     */
-    public org.tensorflow.proto.framework.RunOptions.Experimental getExperimental() {
-      if (experimentalBuilder_ == null) {
-        return experimental_ == null ? org.tensorflow.proto.framework.RunOptions.Experimental.getDefaultInstance() : experimental_;
-      } else {
-        return experimentalBuilder_.getMessage();
-      }
-    }
-    /**
-     * <code>.tensorflow.RunOptions.Experimental experimental = 8;</code>
-     */
-    public Builder setExperimental(org.tensorflow.proto.framework.RunOptions.Experimental value) {
-      if (experimentalBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        experimental_ = value;
-        onChanged();
-      } else {
-        experimentalBuilder_.setMessage(value);
-      }
-
-      return this;
-    }
-    /**
-     * <code>.tensorflow.RunOptions.Experimental experimental = 8;</code>
-     */
-    public Builder setExperimental(
-        org.tensorflow.proto.framework.RunOptions.Experimental.Builder builderForValue) {
-      if (experimentalBuilder_ == null) {
-        experimental_ = builderForValue.build();
-        onChanged();
-      } else {
-        experimentalBuilder_.setMessage(builderForValue.build());
-      }
-
-      return this;
-    }
-    /**
-     * <code>.tensorflow.RunOptions.Experimental experimental = 8;</code>
-     */
-    public Builder mergeExperimental(org.tensorflow.proto.framework.RunOptions.Experimental value) {
-      if (experimentalBuilder_ == null) {
-        if (experimental_ != null) {
-          experimental_ =
-            org.tensorflow.proto.framework.RunOptions.Experimental.newBuilder(experimental_).mergeFrom(value).buildPartial();
-        } else {
-          experimental_ = value;
-        }
-        onChanged();
-      } else {
-        experimentalBuilder_.mergeFrom(value);
-      }
-
-      return this;
-    }
-    /**
-     * <code>.tensorflow.RunOptions.Experimental experimental = 8;</code>
-     */
-    public Builder clearExperimental() {
-      if (experimentalBuilder_ == null) {
-        experimental_ = null;
-        onChanged();
-      } else {
-        experimental_ = null;
-        experimentalBuilder_ = null;
-      }
-
-      return this;
-    }
-    /**
-     * <code>.tensorflow.RunOptions.Experimental experimental = 8;</code>
-     */
-    public org.tensorflow.proto.framework.RunOptions.Experimental.Builder getExperimentalBuilder() {
-      
-      onChanged();
-      return getExperimentalFieldBuilder().getBuilder();
-    }
-    /**
-     * <code>.tensorflow.RunOptions.Experimental experimental = 8;</code>
-     */
-    public org.tensorflow.proto.framework.RunOptions.ExperimentalOrBuilder getExperimentalOrBuilder() {
-      if (experimentalBuilder_ != null) {
-        return experimentalBuilder_.getMessageOrBuilder();
-      } else {
-        return experimental_ == null ?
-            org.tensorflow.proto.framework.RunOptions.Experimental.getDefaultInstance() : experimental_;
-      }
-    }
-    /**
-     * <code>.tensorflow.RunOptions.Experimental experimental = 8;</code>
-     */
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.RunOptions.Experimental, org.tensorflow.proto.framework.RunOptions.Experimental.Builder, org.tensorflow.proto.framework.RunOptions.ExperimentalOrBuilder> 
-        getExperimentalFieldBuilder() {
-      if (experimentalBuilder_ == null) {
-        experimentalBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-            org.tensorflow.proto.framework.RunOptions.Experimental, org.tensorflow.proto.framework.RunOptions.Experimental.Builder, org.tensorflow.proto.framework.RunOptions.ExperimentalOrBuilder>(
-                getExperimental(),
-                getParentForChildren(),
-                isClean());
-        experimental_ = null;
-      }
-      return experimentalBuilder_;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.RunOptions)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.RunOptions)
-  private static final org.tensorflow.proto.framework.RunOptions DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.RunOptions();
-  }
-
-  public static org.tensorflow.proto.framework.RunOptions getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<RunOptions>
-      PARSER = new com.google.protobuf.AbstractParser<RunOptions>() {
-    @java.lang.Override
-    public RunOptions parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new RunOptions(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<RunOptions> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<RunOptions> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.framework.RunOptions getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/RunOptionsOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/RunOptionsOrBuilder.java
deleted file mode 100644
index c3f4a9d7205..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/RunOptionsOrBuilder.java
+++ /dev/null
@@ -1,101 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/config.proto
-
-package org.tensorflow.proto.framework;
-
-public interface RunOptionsOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.RunOptions)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <code>.tensorflow.RunOptions.TraceLevel trace_level = 1;</code>
-   */
-  int getTraceLevelValue();
-  /**
-   * <code>.tensorflow.RunOptions.TraceLevel trace_level = 1;</code>
-   */
-  org.tensorflow.proto.framework.RunOptions.TraceLevel getTraceLevel();
-
-  /**
-   * <pre>
-   * Time to wait for operation to complete in milliseconds.
-   * </pre>
-   *
-   * <code>int64 timeout_in_ms = 2;</code>
-   */
-  long getTimeoutInMs();
-
-  /**
-   * <pre>
-   * The thread pool to use, if session_inter_op_thread_pool is configured.
-   * To use the caller thread set this to -1 - this uses the caller thread
-   * to execute Session::Run() and thus avoids a context switch. Using the
-   * caller thread to execute Session::Run() should be done ONLY for simple
-   * graphs, where the overhead of an additional context switch is
-   * comparable with the overhead of Session::Run().
-   * </pre>
-   *
-   * <code>int32 inter_op_thread_pool = 3;</code>
-   */
-  int getInterOpThreadPool();
-
-  /**
-   * <pre>
-   * Whether the partition graph(s) executed by the executor(s) should be
-   * outputted via RunMetadata.
-   * </pre>
-   *
-   * <code>bool output_partition_graphs = 5;</code>
-   */
-  boolean getOutputPartitionGraphs();
-
-  /**
-   * <pre>
-   * EXPERIMENTAL.  Options used to initialize DebuggerState, if enabled.
-   * </pre>
-   *
-   * <code>.tensorflow.DebugOptions debug_options = 6;</code>
-   */
-  boolean hasDebugOptions();
-  /**
-   * <pre>
-   * EXPERIMENTAL.  Options used to initialize DebuggerState, if enabled.
-   * </pre>
-   *
-   * <code>.tensorflow.DebugOptions debug_options = 6;</code>
-   */
-  org.tensorflow.proto.framework.DebugOptions getDebugOptions();
-  /**
-   * <pre>
-   * EXPERIMENTAL.  Options used to initialize DebuggerState, if enabled.
-   * </pre>
-   *
-   * <code>.tensorflow.DebugOptions debug_options = 6;</code>
-   */
-  org.tensorflow.proto.framework.DebugOptionsOrBuilder getDebugOptionsOrBuilder();
-
-  /**
-   * <pre>
-   * When enabled, causes tensor allocation information to be included in
-   * the error message when the Run() call fails because the allocator ran
-   * out of memory (OOM).
-   * Enabling this option can slow down the Run() call.
-   * </pre>
-   *
-   * <code>bool report_tensor_allocations_upon_oom = 7;</code>
-   */
-  boolean getReportTensorAllocationsUponOom();
-
-  /**
-   * <code>.tensorflow.RunOptions.Experimental experimental = 8;</code>
-   */
-  boolean hasExperimental();
-  /**
-   * <code>.tensorflow.RunOptions.Experimental experimental = 8;</code>
-   */
-  org.tensorflow.proto.framework.RunOptions.Experimental getExperimental();
-  /**
-   * <code>.tensorflow.RunOptions.Experimental experimental = 8;</code>
-   */
-  org.tensorflow.proto.framework.RunOptions.ExperimentalOrBuilder getExperimentalOrBuilder();
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SaveSliceInfoDef.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SaveSliceInfoDef.java
deleted file mode 100644
index 3cf8b1b4b12..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SaveSliceInfoDef.java
+++ /dev/null
@@ -1,1175 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/framework/variable.proto
-
-package org.tensorflow.proto.framework;
-
-/**
- * Protobuf type {@code tensorflow.SaveSliceInfoDef}
- */
-public  final class SaveSliceInfoDef extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.SaveSliceInfoDef)
-    SaveSliceInfoDefOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use SaveSliceInfoDef.newBuilder() to construct.
-  private SaveSliceInfoDef(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private SaveSliceInfoDef() {
-    fullName_ = "";
-    fullShape_ = emptyLongList();
-    varOffset_ = emptyLongList();
-    varShape_ = emptyLongList();
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new SaveSliceInfoDef();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private SaveSliceInfoDef(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    int mutable_bitField0_ = 0;
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 10: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            fullName_ = s;
-            break;
-          }
-          case 16: {
-            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
-              fullShape_ = newLongList();
-              mutable_bitField0_ |= 0x00000001;
-            }
-            fullShape_.addLong(input.readInt64());
-            break;
-          }
-          case 18: {
-            int length = input.readRawVarint32();
-            int limit = input.pushLimit(length);
-            if (!((mutable_bitField0_ & 0x00000001) != 0) && input.getBytesUntilLimit() > 0) {
-              fullShape_ = newLongList();
-              mutable_bitField0_ |= 0x00000001;
-            }
-            while (input.getBytesUntilLimit() > 0) {
-              fullShape_.addLong(input.readInt64());
-            }
-            input.popLimit(limit);
-            break;
-          }
-          case 24: {
-            if (!((mutable_bitField0_ & 0x00000002) != 0)) {
-              varOffset_ = newLongList();
-              mutable_bitField0_ |= 0x00000002;
-            }
-            varOffset_.addLong(input.readInt64());
-            break;
-          }
-          case 26: {
-            int length = input.readRawVarint32();
-            int limit = input.pushLimit(length);
-            if (!((mutable_bitField0_ & 0x00000002) != 0) && input.getBytesUntilLimit() > 0) {
-              varOffset_ = newLongList();
-              mutable_bitField0_ |= 0x00000002;
-            }
-            while (input.getBytesUntilLimit() > 0) {
-              varOffset_.addLong(input.readInt64());
-            }
-            input.popLimit(limit);
-            break;
-          }
-          case 32: {
-            if (!((mutable_bitField0_ & 0x00000004) != 0)) {
-              varShape_ = newLongList();
-              mutable_bitField0_ |= 0x00000004;
-            }
-            varShape_.addLong(input.readInt64());
-            break;
-          }
-          case 34: {
-            int length = input.readRawVarint32();
-            int limit = input.pushLimit(length);
-            if (!((mutable_bitField0_ & 0x00000004) != 0) && input.getBytesUntilLimit() > 0) {
-              varShape_ = newLongList();
-              mutable_bitField0_ |= 0x00000004;
-            }
-            while (input.getBytesUntilLimit() > 0) {
-              varShape_.addLong(input.readInt64());
-            }
-            input.popLimit(limit);
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      if (((mutable_bitField0_ & 0x00000001) != 0)) {
-        fullShape_.makeImmutable(); // C
-      }
-      if (((mutable_bitField0_ & 0x00000002) != 0)) {
-        varOffset_.makeImmutable(); // C
-      }
-      if (((mutable_bitField0_ & 0x00000004) != 0)) {
-        varShape_.makeImmutable(); // C
-      }
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.framework.VariableProtos.internal_static_tensorflow_SaveSliceInfoDef_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.framework.VariableProtos.internal_static_tensorflow_SaveSliceInfoDef_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.framework.SaveSliceInfoDef.class, org.tensorflow.proto.framework.SaveSliceInfoDef.Builder.class);
-  }
-
-  public static final int FULL_NAME_FIELD_NUMBER = 1;
-  private volatile java.lang.Object fullName_;
-  /**
-   * <pre>
-   * Name of the full variable of which this is a slice.
-   * </pre>
-   *
-   * <code>string full_name = 1;</code>
-   */
-  public java.lang.String getFullName() {
-    java.lang.Object ref = fullName_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      fullName_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * Name of the full variable of which this is a slice.
-   * </pre>
-   *
-   * <code>string full_name = 1;</code>
-   */
-  public com.google.protobuf.ByteString
-      getFullNameBytes() {
-    java.lang.Object ref = fullName_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      fullName_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int FULL_SHAPE_FIELD_NUMBER = 2;
-  private com.google.protobuf.Internal.LongList fullShape_;
-  /**
-   * <pre>
-   * Shape of the full variable.
-   * </pre>
-   *
-   * <code>repeated int64 full_shape = 2;</code>
-   */
-  public java.util.List<java.lang.Long>
-      getFullShapeList() {
-    return fullShape_;
-  }
-  /**
-   * <pre>
-   * Shape of the full variable.
-   * </pre>
-   *
-   * <code>repeated int64 full_shape = 2;</code>
-   */
-  public int getFullShapeCount() {
-    return fullShape_.size();
-  }
-  /**
-   * <pre>
-   * Shape of the full variable.
-   * </pre>
-   *
-   * <code>repeated int64 full_shape = 2;</code>
-   */
-  public long getFullShape(int index) {
-    return fullShape_.getLong(index);
-  }
-  private int fullShapeMemoizedSerializedSize = -1;
-
-  public static final int VAR_OFFSET_FIELD_NUMBER = 3;
-  private com.google.protobuf.Internal.LongList varOffset_;
-  /**
-   * <pre>
-   * Offset of this variable into the full variable.
-   * </pre>
-   *
-   * <code>repeated int64 var_offset = 3;</code>
-   */
-  public java.util.List<java.lang.Long>
-      getVarOffsetList() {
-    return varOffset_;
-  }
-  /**
-   * <pre>
-   * Offset of this variable into the full variable.
-   * </pre>
-   *
-   * <code>repeated int64 var_offset = 3;</code>
-   */
-  public int getVarOffsetCount() {
-    return varOffset_.size();
-  }
-  /**
-   * <pre>
-   * Offset of this variable into the full variable.
-   * </pre>
-   *
-   * <code>repeated int64 var_offset = 3;</code>
-   */
-  public long getVarOffset(int index) {
-    return varOffset_.getLong(index);
-  }
-  private int varOffsetMemoizedSerializedSize = -1;
-
-  public static final int VAR_SHAPE_FIELD_NUMBER = 4;
-  private com.google.protobuf.Internal.LongList varShape_;
-  /**
-   * <pre>
-   * Shape of this variable.
-   * </pre>
-   *
-   * <code>repeated int64 var_shape = 4;</code>
-   */
-  public java.util.List<java.lang.Long>
-      getVarShapeList() {
-    return varShape_;
-  }
-  /**
-   * <pre>
-   * Shape of this variable.
-   * </pre>
-   *
-   * <code>repeated int64 var_shape = 4;</code>
-   */
-  public int getVarShapeCount() {
-    return varShape_.size();
-  }
-  /**
-   * <pre>
-   * Shape of this variable.
-   * </pre>
-   *
-   * <code>repeated int64 var_shape = 4;</code>
-   */
-  public long getVarShape(int index) {
-    return varShape_.getLong(index);
-  }
-  private int varShapeMemoizedSerializedSize = -1;
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    getSerializedSize();
-    if (!getFullNameBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 1, fullName_);
-    }
-    if (getFullShapeList().size() > 0) {
-      output.writeUInt32NoTag(18);
-      output.writeUInt32NoTag(fullShapeMemoizedSerializedSize);
-    }
-    for (int i = 0; i < fullShape_.size(); i++) {
-      output.writeInt64NoTag(fullShape_.getLong(i));
-    }
-    if (getVarOffsetList().size() > 0) {
-      output.writeUInt32NoTag(26);
-      output.writeUInt32NoTag(varOffsetMemoizedSerializedSize);
-    }
-    for (int i = 0; i < varOffset_.size(); i++) {
-      output.writeInt64NoTag(varOffset_.getLong(i));
-    }
-    if (getVarShapeList().size() > 0) {
-      output.writeUInt32NoTag(34);
-      output.writeUInt32NoTag(varShapeMemoizedSerializedSize);
-    }
-    for (int i = 0; i < varShape_.size(); i++) {
-      output.writeInt64NoTag(varShape_.getLong(i));
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (!getFullNameBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, fullName_);
-    }
-    {
-      int dataSize = 0;
-      for (int i = 0; i < fullShape_.size(); i++) {
-        dataSize += com.google.protobuf.CodedOutputStream
-          .computeInt64SizeNoTag(fullShape_.getLong(i));
-      }
-      size += dataSize;
-      if (!getFullShapeList().isEmpty()) {
-        size += 1;
-        size += com.google.protobuf.CodedOutputStream
-            .computeInt32SizeNoTag(dataSize);
-      }
-      fullShapeMemoizedSerializedSize = dataSize;
-    }
-    {
-      int dataSize = 0;
-      for (int i = 0; i < varOffset_.size(); i++) {
-        dataSize += com.google.protobuf.CodedOutputStream
-          .computeInt64SizeNoTag(varOffset_.getLong(i));
-      }
-      size += dataSize;
-      if (!getVarOffsetList().isEmpty()) {
-        size += 1;
-        size += com.google.protobuf.CodedOutputStream
-            .computeInt32SizeNoTag(dataSize);
-      }
-      varOffsetMemoizedSerializedSize = dataSize;
-    }
-    {
-      int dataSize = 0;
-      for (int i = 0; i < varShape_.size(); i++) {
-        dataSize += com.google.protobuf.CodedOutputStream
-          .computeInt64SizeNoTag(varShape_.getLong(i));
-      }
-      size += dataSize;
-      if (!getVarShapeList().isEmpty()) {
-        size += 1;
-        size += com.google.protobuf.CodedOutputStream
-            .computeInt32SizeNoTag(dataSize);
-      }
-      varShapeMemoizedSerializedSize = dataSize;
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.framework.SaveSliceInfoDef)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.framework.SaveSliceInfoDef other = (org.tensorflow.proto.framework.SaveSliceInfoDef) obj;
-
-    if (!getFullName()
-        .equals(other.getFullName())) return false;
-    if (!getFullShapeList()
-        .equals(other.getFullShapeList())) return false;
-    if (!getVarOffsetList()
-        .equals(other.getVarOffsetList())) return false;
-    if (!getVarShapeList()
-        .equals(other.getVarShapeList())) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    hash = (37 * hash) + FULL_NAME_FIELD_NUMBER;
-    hash = (53 * hash) + getFullName().hashCode();
-    if (getFullShapeCount() > 0) {
-      hash = (37 * hash) + FULL_SHAPE_FIELD_NUMBER;
-      hash = (53 * hash) + getFullShapeList().hashCode();
-    }
-    if (getVarOffsetCount() > 0) {
-      hash = (37 * hash) + VAR_OFFSET_FIELD_NUMBER;
-      hash = (53 * hash) + getVarOffsetList().hashCode();
-    }
-    if (getVarShapeCount() > 0) {
-      hash = (37 * hash) + VAR_SHAPE_FIELD_NUMBER;
-      hash = (53 * hash) + getVarShapeList().hashCode();
-    }
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.framework.SaveSliceInfoDef parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.SaveSliceInfoDef parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.SaveSliceInfoDef parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.SaveSliceInfoDef parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.SaveSliceInfoDef parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.SaveSliceInfoDef parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.SaveSliceInfoDef parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.SaveSliceInfoDef parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.SaveSliceInfoDef parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.SaveSliceInfoDef parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.SaveSliceInfoDef parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.SaveSliceInfoDef parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.framework.SaveSliceInfoDef prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * Protobuf type {@code tensorflow.SaveSliceInfoDef}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.SaveSliceInfoDef)
-      org.tensorflow.proto.framework.SaveSliceInfoDefOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.VariableProtos.internal_static_tensorflow_SaveSliceInfoDef_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.VariableProtos.internal_static_tensorflow_SaveSliceInfoDef_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.SaveSliceInfoDef.class, org.tensorflow.proto.framework.SaveSliceInfoDef.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.framework.SaveSliceInfoDef.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      fullName_ = "";
-
-      fullShape_ = emptyLongList();
-      bitField0_ = (bitField0_ & ~0x00000001);
-      varOffset_ = emptyLongList();
-      bitField0_ = (bitField0_ & ~0x00000002);
-      varShape_ = emptyLongList();
-      bitField0_ = (bitField0_ & ~0x00000004);
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.framework.VariableProtos.internal_static_tensorflow_SaveSliceInfoDef_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.SaveSliceInfoDef getDefaultInstanceForType() {
-      return org.tensorflow.proto.framework.SaveSliceInfoDef.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.SaveSliceInfoDef build() {
-      org.tensorflow.proto.framework.SaveSliceInfoDef result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.SaveSliceInfoDef buildPartial() {
-      org.tensorflow.proto.framework.SaveSliceInfoDef result = new org.tensorflow.proto.framework.SaveSliceInfoDef(this);
-      int from_bitField0_ = bitField0_;
-      result.fullName_ = fullName_;
-      if (((bitField0_ & 0x00000001) != 0)) {
-        fullShape_.makeImmutable();
-        bitField0_ = (bitField0_ & ~0x00000001);
-      }
-      result.fullShape_ = fullShape_;
-      if (((bitField0_ & 0x00000002) != 0)) {
-        varOffset_.makeImmutable();
-        bitField0_ = (bitField0_ & ~0x00000002);
-      }
-      result.varOffset_ = varOffset_;
-      if (((bitField0_ & 0x00000004) != 0)) {
-        varShape_.makeImmutable();
-        bitField0_ = (bitField0_ & ~0x00000004);
-      }
-      result.varShape_ = varShape_;
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.framework.SaveSliceInfoDef) {
-        return mergeFrom((org.tensorflow.proto.framework.SaveSliceInfoDef)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.framework.SaveSliceInfoDef other) {
-      if (other == org.tensorflow.proto.framework.SaveSliceInfoDef.getDefaultInstance()) return this;
-      if (!other.getFullName().isEmpty()) {
-        fullName_ = other.fullName_;
-        onChanged();
-      }
-      if (!other.fullShape_.isEmpty()) {
-        if (fullShape_.isEmpty()) {
-          fullShape_ = other.fullShape_;
-          bitField0_ = (bitField0_ & ~0x00000001);
-        } else {
-          ensureFullShapeIsMutable();
-          fullShape_.addAll(other.fullShape_);
-        }
-        onChanged();
-      }
-      if (!other.varOffset_.isEmpty()) {
-        if (varOffset_.isEmpty()) {
-          varOffset_ = other.varOffset_;
-          bitField0_ = (bitField0_ & ~0x00000002);
-        } else {
-          ensureVarOffsetIsMutable();
-          varOffset_.addAll(other.varOffset_);
-        }
-        onChanged();
-      }
-      if (!other.varShape_.isEmpty()) {
-        if (varShape_.isEmpty()) {
-          varShape_ = other.varShape_;
-          bitField0_ = (bitField0_ & ~0x00000004);
-        } else {
-          ensureVarShapeIsMutable();
-          varShape_.addAll(other.varShape_);
-        }
-        onChanged();
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.framework.SaveSliceInfoDef parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.framework.SaveSliceInfoDef) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-    private int bitField0_;
-
-    private java.lang.Object fullName_ = "";
-    /**
-     * <pre>
-     * Name of the full variable of which this is a slice.
-     * </pre>
-     *
-     * <code>string full_name = 1;</code>
-     */
-    public java.lang.String getFullName() {
-      java.lang.Object ref = fullName_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        fullName_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Name of the full variable of which this is a slice.
-     * </pre>
-     *
-     * <code>string full_name = 1;</code>
-     */
-    public com.google.protobuf.ByteString
-        getFullNameBytes() {
-      java.lang.Object ref = fullName_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        fullName_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Name of the full variable of which this is a slice.
-     * </pre>
-     *
-     * <code>string full_name = 1;</code>
-     */
-    public Builder setFullName(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      fullName_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Name of the full variable of which this is a slice.
-     * </pre>
-     *
-     * <code>string full_name = 1;</code>
-     */
-    public Builder clearFullName() {
-      
-      fullName_ = getDefaultInstance().getFullName();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Name of the full variable of which this is a slice.
-     * </pre>
-     *
-     * <code>string full_name = 1;</code>
-     */
-    public Builder setFullNameBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      fullName_ = value;
-      onChanged();
-      return this;
-    }
-
-    private com.google.protobuf.Internal.LongList fullShape_ = emptyLongList();
-    private void ensureFullShapeIsMutable() {
-      if (!((bitField0_ & 0x00000001) != 0)) {
-        fullShape_ = mutableCopy(fullShape_);
-        bitField0_ |= 0x00000001;
-       }
-    }
-    /**
-     * <pre>
-     * Shape of the full variable.
-     * </pre>
-     *
-     * <code>repeated int64 full_shape = 2;</code>
-     */
-    public java.util.List<java.lang.Long>
-        getFullShapeList() {
-      return ((bitField0_ & 0x00000001) != 0) ?
-               java.util.Collections.unmodifiableList(fullShape_) : fullShape_;
-    }
-    /**
-     * <pre>
-     * Shape of the full variable.
-     * </pre>
-     *
-     * <code>repeated int64 full_shape = 2;</code>
-     */
-    public int getFullShapeCount() {
-      return fullShape_.size();
-    }
-    /**
-     * <pre>
-     * Shape of the full variable.
-     * </pre>
-     *
-     * <code>repeated int64 full_shape = 2;</code>
-     */
-    public long getFullShape(int index) {
-      return fullShape_.getLong(index);
-    }
-    /**
-     * <pre>
-     * Shape of the full variable.
-     * </pre>
-     *
-     * <code>repeated int64 full_shape = 2;</code>
-     */
-    public Builder setFullShape(
-        int index, long value) {
-      ensureFullShapeIsMutable();
-      fullShape_.setLong(index, value);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Shape of the full variable.
-     * </pre>
-     *
-     * <code>repeated int64 full_shape = 2;</code>
-     */
-    public Builder addFullShape(long value) {
-      ensureFullShapeIsMutable();
-      fullShape_.addLong(value);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Shape of the full variable.
-     * </pre>
-     *
-     * <code>repeated int64 full_shape = 2;</code>
-     */
-    public Builder addAllFullShape(
-        java.lang.Iterable<? extends java.lang.Long> values) {
-      ensureFullShapeIsMutable();
-      com.google.protobuf.AbstractMessageLite.Builder.addAll(
-          values, fullShape_);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Shape of the full variable.
-     * </pre>
-     *
-     * <code>repeated int64 full_shape = 2;</code>
-     */
-    public Builder clearFullShape() {
-      fullShape_ = emptyLongList();
-      bitField0_ = (bitField0_ & ~0x00000001);
-      onChanged();
-      return this;
-    }
-
-    private com.google.protobuf.Internal.LongList varOffset_ = emptyLongList();
-    private void ensureVarOffsetIsMutable() {
-      if (!((bitField0_ & 0x00000002) != 0)) {
-        varOffset_ = mutableCopy(varOffset_);
-        bitField0_ |= 0x00000002;
-       }
-    }
-    /**
-     * <pre>
-     * Offset of this variable into the full variable.
-     * </pre>
-     *
-     * <code>repeated int64 var_offset = 3;</code>
-     */
-    public java.util.List<java.lang.Long>
-        getVarOffsetList() {
-      return ((bitField0_ & 0x00000002) != 0) ?
-               java.util.Collections.unmodifiableList(varOffset_) : varOffset_;
-    }
-    /**
-     * <pre>
-     * Offset of this variable into the full variable.
-     * </pre>
-     *
-     * <code>repeated int64 var_offset = 3;</code>
-     */
-    public int getVarOffsetCount() {
-      return varOffset_.size();
-    }
-    /**
-     * <pre>
-     * Offset of this variable into the full variable.
-     * </pre>
-     *
-     * <code>repeated int64 var_offset = 3;</code>
-     */
-    public long getVarOffset(int index) {
-      return varOffset_.getLong(index);
-    }
-    /**
-     * <pre>
-     * Offset of this variable into the full variable.
-     * </pre>
-     *
-     * <code>repeated int64 var_offset = 3;</code>
-     */
-    public Builder setVarOffset(
-        int index, long value) {
-      ensureVarOffsetIsMutable();
-      varOffset_.setLong(index, value);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Offset of this variable into the full variable.
-     * </pre>
-     *
-     * <code>repeated int64 var_offset = 3;</code>
-     */
-    public Builder addVarOffset(long value) {
-      ensureVarOffsetIsMutable();
-      varOffset_.addLong(value);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Offset of this variable into the full variable.
-     * </pre>
-     *
-     * <code>repeated int64 var_offset = 3;</code>
-     */
-    public Builder addAllVarOffset(
-        java.lang.Iterable<? extends java.lang.Long> values) {
-      ensureVarOffsetIsMutable();
-      com.google.protobuf.AbstractMessageLite.Builder.addAll(
-          values, varOffset_);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Offset of this variable into the full variable.
-     * </pre>
-     *
-     * <code>repeated int64 var_offset = 3;</code>
-     */
-    public Builder clearVarOffset() {
-      varOffset_ = emptyLongList();
-      bitField0_ = (bitField0_ & ~0x00000002);
-      onChanged();
-      return this;
-    }
-
-    private com.google.protobuf.Internal.LongList varShape_ = emptyLongList();
-    private void ensureVarShapeIsMutable() {
-      if (!((bitField0_ & 0x00000004) != 0)) {
-        varShape_ = mutableCopy(varShape_);
-        bitField0_ |= 0x00000004;
-       }
-    }
-    /**
-     * <pre>
-     * Shape of this variable.
-     * </pre>
-     *
-     * <code>repeated int64 var_shape = 4;</code>
-     */
-    public java.util.List<java.lang.Long>
-        getVarShapeList() {
-      return ((bitField0_ & 0x00000004) != 0) ?
-               java.util.Collections.unmodifiableList(varShape_) : varShape_;
-    }
-    /**
-     * <pre>
-     * Shape of this variable.
-     * </pre>
-     *
-     * <code>repeated int64 var_shape = 4;</code>
-     */
-    public int getVarShapeCount() {
-      return varShape_.size();
-    }
-    /**
-     * <pre>
-     * Shape of this variable.
-     * </pre>
-     *
-     * <code>repeated int64 var_shape = 4;</code>
-     */
-    public long getVarShape(int index) {
-      return varShape_.getLong(index);
-    }
-    /**
-     * <pre>
-     * Shape of this variable.
-     * </pre>
-     *
-     * <code>repeated int64 var_shape = 4;</code>
-     */
-    public Builder setVarShape(
-        int index, long value) {
-      ensureVarShapeIsMutable();
-      varShape_.setLong(index, value);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Shape of this variable.
-     * </pre>
-     *
-     * <code>repeated int64 var_shape = 4;</code>
-     */
-    public Builder addVarShape(long value) {
-      ensureVarShapeIsMutable();
-      varShape_.addLong(value);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Shape of this variable.
-     * </pre>
-     *
-     * <code>repeated int64 var_shape = 4;</code>
-     */
-    public Builder addAllVarShape(
-        java.lang.Iterable<? extends java.lang.Long> values) {
-      ensureVarShapeIsMutable();
-      com.google.protobuf.AbstractMessageLite.Builder.addAll(
-          values, varShape_);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Shape of this variable.
-     * </pre>
-     *
-     * <code>repeated int64 var_shape = 4;</code>
-     */
-    public Builder clearVarShape() {
-      varShape_ = emptyLongList();
-      bitField0_ = (bitField0_ & ~0x00000004);
-      onChanged();
-      return this;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.SaveSliceInfoDef)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.SaveSliceInfoDef)
-  private static final org.tensorflow.proto.framework.SaveSliceInfoDef DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.SaveSliceInfoDef();
-  }
-
-  public static org.tensorflow.proto.framework.SaveSliceInfoDef getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<SaveSliceInfoDef>
-      PARSER = new com.google.protobuf.AbstractParser<SaveSliceInfoDef>() {
-    @java.lang.Override
-    public SaveSliceInfoDef parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new SaveSliceInfoDef(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<SaveSliceInfoDef> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<SaveSliceInfoDef> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.framework.SaveSliceInfoDef getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SaveSliceInfoDefOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SaveSliceInfoDefOrBuilder.java
deleted file mode 100644
index c06b53fbb61..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SaveSliceInfoDefOrBuilder.java
+++ /dev/null
@@ -1,102 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/framework/variable.proto
-
-package org.tensorflow.proto.framework;
-
-public interface SaveSliceInfoDefOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.SaveSliceInfoDef)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <pre>
-   * Name of the full variable of which this is a slice.
-   * </pre>
-   *
-   * <code>string full_name = 1;</code>
-   */
-  java.lang.String getFullName();
-  /**
-   * <pre>
-   * Name of the full variable of which this is a slice.
-   * </pre>
-   *
-   * <code>string full_name = 1;</code>
-   */
-  com.google.protobuf.ByteString
-      getFullNameBytes();
-
-  /**
-   * <pre>
-   * Shape of the full variable.
-   * </pre>
-   *
-   * <code>repeated int64 full_shape = 2;</code>
-   */
-  java.util.List<java.lang.Long> getFullShapeList();
-  /**
-   * <pre>
-   * Shape of the full variable.
-   * </pre>
-   *
-   * <code>repeated int64 full_shape = 2;</code>
-   */
-  int getFullShapeCount();
-  /**
-   * <pre>
-   * Shape of the full variable.
-   * </pre>
-   *
-   * <code>repeated int64 full_shape = 2;</code>
-   */
-  long getFullShape(int index);
-
-  /**
-   * <pre>
-   * Offset of this variable into the full variable.
-   * </pre>
-   *
-   * <code>repeated int64 var_offset = 3;</code>
-   */
-  java.util.List<java.lang.Long> getVarOffsetList();
-  /**
-   * <pre>
-   * Offset of this variable into the full variable.
-   * </pre>
-   *
-   * <code>repeated int64 var_offset = 3;</code>
-   */
-  int getVarOffsetCount();
-  /**
-   * <pre>
-   * Offset of this variable into the full variable.
-   * </pre>
-   *
-   * <code>repeated int64 var_offset = 3;</code>
-   */
-  long getVarOffset(int index);
-
-  /**
-   * <pre>
-   * Shape of this variable.
-   * </pre>
-   *
-   * <code>repeated int64 var_shape = 4;</code>
-   */
-  java.util.List<java.lang.Long> getVarShapeList();
-  /**
-   * <pre>
-   * Shape of this variable.
-   * </pre>
-   *
-   * <code>repeated int64 var_shape = 4;</code>
-   */
-  int getVarShapeCount();
-  /**
-   * <pre>
-   * Shape of this variable.
-   * </pre>
-   *
-   * <code>repeated int64 var_shape = 4;</code>
-   */
-  long getVarShape(int index);
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SaveableObject.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SaveableObject.java
deleted file mode 100644
index c2b8ae241d9..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SaveableObject.java
+++ /dev/null
@@ -1,549 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/saved_object_graph.proto
-
-package org.tensorflow.proto.framework;
-
-/**
- * Protobuf type {@code tensorflow.SaveableObject}
- */
-public  final class SaveableObject extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.SaveableObject)
-    SaveableObjectOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use SaveableObject.newBuilder() to construct.
-  private SaveableObject(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private SaveableObject() {
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new SaveableObject();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private SaveableObject(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 16: {
-
-            saveFunction_ = input.readInt32();
-            break;
-          }
-          case 24: {
-
-            restoreFunction_ = input.readInt32();
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.framework.SavedObjectGraphProtos.internal_static_tensorflow_SaveableObject_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.framework.SavedObjectGraphProtos.internal_static_tensorflow_SaveableObject_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.framework.SaveableObject.class, org.tensorflow.proto.framework.SaveableObject.Builder.class);
-  }
-
-  public static final int SAVE_FUNCTION_FIELD_NUMBER = 2;
-  private int saveFunction_;
-  /**
-   * <pre>
-   * Node ids of concrete functions for saving and loading from a checkpoint.
-   * </pre>
-   *
-   * <code>int32 save_function = 2;</code>
-   */
-  public int getSaveFunction() {
-    return saveFunction_;
-  }
-
-  public static final int RESTORE_FUNCTION_FIELD_NUMBER = 3;
-  private int restoreFunction_;
-  /**
-   * <code>int32 restore_function = 3;</code>
-   */
-  public int getRestoreFunction() {
-    return restoreFunction_;
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    if (saveFunction_ != 0) {
-      output.writeInt32(2, saveFunction_);
-    }
-    if (restoreFunction_ != 0) {
-      output.writeInt32(3, restoreFunction_);
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (saveFunction_ != 0) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt32Size(2, saveFunction_);
-    }
-    if (restoreFunction_ != 0) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt32Size(3, restoreFunction_);
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.framework.SaveableObject)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.framework.SaveableObject other = (org.tensorflow.proto.framework.SaveableObject) obj;
-
-    if (getSaveFunction()
-        != other.getSaveFunction()) return false;
-    if (getRestoreFunction()
-        != other.getRestoreFunction()) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    hash = (37 * hash) + SAVE_FUNCTION_FIELD_NUMBER;
-    hash = (53 * hash) + getSaveFunction();
-    hash = (37 * hash) + RESTORE_FUNCTION_FIELD_NUMBER;
-    hash = (53 * hash) + getRestoreFunction();
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.framework.SaveableObject parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.SaveableObject parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.SaveableObject parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.SaveableObject parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.SaveableObject parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.SaveableObject parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.SaveableObject parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.SaveableObject parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.SaveableObject parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.SaveableObject parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.SaveableObject parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.SaveableObject parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.framework.SaveableObject prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * Protobuf type {@code tensorflow.SaveableObject}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.SaveableObject)
-      org.tensorflow.proto.framework.SaveableObjectOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.SavedObjectGraphProtos.internal_static_tensorflow_SaveableObject_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.SavedObjectGraphProtos.internal_static_tensorflow_SaveableObject_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.SaveableObject.class, org.tensorflow.proto.framework.SaveableObject.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.framework.SaveableObject.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      saveFunction_ = 0;
-
-      restoreFunction_ = 0;
-
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.framework.SavedObjectGraphProtos.internal_static_tensorflow_SaveableObject_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.SaveableObject getDefaultInstanceForType() {
-      return org.tensorflow.proto.framework.SaveableObject.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.SaveableObject build() {
-      org.tensorflow.proto.framework.SaveableObject result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.SaveableObject buildPartial() {
-      org.tensorflow.proto.framework.SaveableObject result = new org.tensorflow.proto.framework.SaveableObject(this);
-      result.saveFunction_ = saveFunction_;
-      result.restoreFunction_ = restoreFunction_;
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.framework.SaveableObject) {
-        return mergeFrom((org.tensorflow.proto.framework.SaveableObject)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.framework.SaveableObject other) {
-      if (other == org.tensorflow.proto.framework.SaveableObject.getDefaultInstance()) return this;
-      if (other.getSaveFunction() != 0) {
-        setSaveFunction(other.getSaveFunction());
-      }
-      if (other.getRestoreFunction() != 0) {
-        setRestoreFunction(other.getRestoreFunction());
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.framework.SaveableObject parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.framework.SaveableObject) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-
-    private int saveFunction_ ;
-    /**
-     * <pre>
-     * Node ids of concrete functions for saving and loading from a checkpoint.
-     * </pre>
-     *
-     * <code>int32 save_function = 2;</code>
-     */
-    public int getSaveFunction() {
-      return saveFunction_;
-    }
-    /**
-     * <pre>
-     * Node ids of concrete functions for saving and loading from a checkpoint.
-     * </pre>
-     *
-     * <code>int32 save_function = 2;</code>
-     */
-    public Builder setSaveFunction(int value) {
-      
-      saveFunction_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Node ids of concrete functions for saving and loading from a checkpoint.
-     * </pre>
-     *
-     * <code>int32 save_function = 2;</code>
-     */
-    public Builder clearSaveFunction() {
-      
-      saveFunction_ = 0;
-      onChanged();
-      return this;
-    }
-
-    private int restoreFunction_ ;
-    /**
-     * <code>int32 restore_function = 3;</code>
-     */
-    public int getRestoreFunction() {
-      return restoreFunction_;
-    }
-    /**
-     * <code>int32 restore_function = 3;</code>
-     */
-    public Builder setRestoreFunction(int value) {
-      
-      restoreFunction_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>int32 restore_function = 3;</code>
-     */
-    public Builder clearRestoreFunction() {
-      
-      restoreFunction_ = 0;
-      onChanged();
-      return this;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.SaveableObject)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.SaveableObject)
-  private static final org.tensorflow.proto.framework.SaveableObject DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.SaveableObject();
-  }
-
-  public static org.tensorflow.proto.framework.SaveableObject getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<SaveableObject>
-      PARSER = new com.google.protobuf.AbstractParser<SaveableObject>() {
-    @java.lang.Override
-    public SaveableObject parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new SaveableObject(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<SaveableObject> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<SaveableObject> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.framework.SaveableObject getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SaveableObjectOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SaveableObjectOrBuilder.java
deleted file mode 100644
index 896dcfa8b34..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SaveableObjectOrBuilder.java
+++ /dev/null
@@ -1,23 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/saved_object_graph.proto
-
-package org.tensorflow.proto.framework;
-
-public interface SaveableObjectOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.SaveableObject)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <pre>
-   * Node ids of concrete functions for saving and loading from a checkpoint.
-   * </pre>
-   *
-   * <code>int32 save_function = 2;</code>
-   */
-  int getSaveFunction();
-
-  /**
-   * <code>int32 restore_function = 3;</code>
-   */
-  int getRestoreFunction();
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedAsset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedAsset.java
deleted file mode 100644
index 2558719f7c3..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedAsset.java
+++ /dev/null
@@ -1,514 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/saved_object_graph.proto
-
-package org.tensorflow.proto.framework;
-
-/**
- * <pre>
- * A SavedAsset points to an asset in the MetaGraph.
- * When bound to a function this object evaluates to a tensor with the absolute
- * filename. Users should not depend on a particular part of the filename to
- * remain stable (e.g. basename could be changed).
- * </pre>
- *
- * Protobuf type {@code tensorflow.SavedAsset}
- */
-public  final class SavedAsset extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.SavedAsset)
-    SavedAssetOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use SavedAsset.newBuilder() to construct.
-  private SavedAsset(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private SavedAsset() {
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new SavedAsset();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private SavedAsset(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 8: {
-
-            assetFileDefIndex_ = input.readInt32();
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.framework.SavedObjectGraphProtos.internal_static_tensorflow_SavedAsset_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.framework.SavedObjectGraphProtos.internal_static_tensorflow_SavedAsset_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.framework.SavedAsset.class, org.tensorflow.proto.framework.SavedAsset.Builder.class);
-  }
-
-  public static final int ASSET_FILE_DEF_INDEX_FIELD_NUMBER = 1;
-  private int assetFileDefIndex_;
-  /**
-   * <pre>
-   * Index into `MetaGraphDef.asset_file_def[]` that describes the Asset.
-   * Only the field `AssetFileDef.filename` is used. Other fields, such as
-   * `AssetFileDef.tensor_info`, MUST be ignored.
-   * </pre>
-   *
-   * <code>int32 asset_file_def_index = 1;</code>
-   */
-  public int getAssetFileDefIndex() {
-    return assetFileDefIndex_;
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    if (assetFileDefIndex_ != 0) {
-      output.writeInt32(1, assetFileDefIndex_);
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (assetFileDefIndex_ != 0) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt32Size(1, assetFileDefIndex_);
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.framework.SavedAsset)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.framework.SavedAsset other = (org.tensorflow.proto.framework.SavedAsset) obj;
-
-    if (getAssetFileDefIndex()
-        != other.getAssetFileDefIndex()) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    hash = (37 * hash) + ASSET_FILE_DEF_INDEX_FIELD_NUMBER;
-    hash = (53 * hash) + getAssetFileDefIndex();
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.framework.SavedAsset parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.SavedAsset parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.SavedAsset parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.SavedAsset parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.SavedAsset parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.SavedAsset parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.SavedAsset parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.SavedAsset parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.SavedAsset parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.SavedAsset parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.SavedAsset parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.SavedAsset parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.framework.SavedAsset prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * <pre>
-   * A SavedAsset points to an asset in the MetaGraph.
-   * When bound to a function this object evaluates to a tensor with the absolute
-   * filename. Users should not depend on a particular part of the filename to
-   * remain stable (e.g. basename could be changed).
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.SavedAsset}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.SavedAsset)
-      org.tensorflow.proto.framework.SavedAssetOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.SavedObjectGraphProtos.internal_static_tensorflow_SavedAsset_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.SavedObjectGraphProtos.internal_static_tensorflow_SavedAsset_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.SavedAsset.class, org.tensorflow.proto.framework.SavedAsset.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.framework.SavedAsset.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      assetFileDefIndex_ = 0;
-
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.framework.SavedObjectGraphProtos.internal_static_tensorflow_SavedAsset_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.SavedAsset getDefaultInstanceForType() {
-      return org.tensorflow.proto.framework.SavedAsset.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.SavedAsset build() {
-      org.tensorflow.proto.framework.SavedAsset result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.SavedAsset buildPartial() {
-      org.tensorflow.proto.framework.SavedAsset result = new org.tensorflow.proto.framework.SavedAsset(this);
-      result.assetFileDefIndex_ = assetFileDefIndex_;
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.framework.SavedAsset) {
-        return mergeFrom((org.tensorflow.proto.framework.SavedAsset)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.framework.SavedAsset other) {
-      if (other == org.tensorflow.proto.framework.SavedAsset.getDefaultInstance()) return this;
-      if (other.getAssetFileDefIndex() != 0) {
-        setAssetFileDefIndex(other.getAssetFileDefIndex());
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.framework.SavedAsset parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.framework.SavedAsset) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-
-    private int assetFileDefIndex_ ;
-    /**
-     * <pre>
-     * Index into `MetaGraphDef.asset_file_def[]` that describes the Asset.
-     * Only the field `AssetFileDef.filename` is used. Other fields, such as
-     * `AssetFileDef.tensor_info`, MUST be ignored.
-     * </pre>
-     *
-     * <code>int32 asset_file_def_index = 1;</code>
-     */
-    public int getAssetFileDefIndex() {
-      return assetFileDefIndex_;
-    }
-    /**
-     * <pre>
-     * Index into `MetaGraphDef.asset_file_def[]` that describes the Asset.
-     * Only the field `AssetFileDef.filename` is used. Other fields, such as
-     * `AssetFileDef.tensor_info`, MUST be ignored.
-     * </pre>
-     *
-     * <code>int32 asset_file_def_index = 1;</code>
-     */
-    public Builder setAssetFileDefIndex(int value) {
-      
-      assetFileDefIndex_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Index into `MetaGraphDef.asset_file_def[]` that describes the Asset.
-     * Only the field `AssetFileDef.filename` is used. Other fields, such as
-     * `AssetFileDef.tensor_info`, MUST be ignored.
-     * </pre>
-     *
-     * <code>int32 asset_file_def_index = 1;</code>
-     */
-    public Builder clearAssetFileDefIndex() {
-      
-      assetFileDefIndex_ = 0;
-      onChanged();
-      return this;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.SavedAsset)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.SavedAsset)
-  private static final org.tensorflow.proto.framework.SavedAsset DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.SavedAsset();
-  }
-
-  public static org.tensorflow.proto.framework.SavedAsset getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<SavedAsset>
-      PARSER = new com.google.protobuf.AbstractParser<SavedAsset>() {
-    @java.lang.Override
-    public SavedAsset parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new SavedAsset(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<SavedAsset> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<SavedAsset> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.framework.SavedAsset getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedAssetOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedAssetOrBuilder.java
deleted file mode 100644
index a12a5ff6ad0..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedAssetOrBuilder.java
+++ /dev/null
@@ -1,20 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/saved_object_graph.proto
-
-package org.tensorflow.proto.framework;
-
-public interface SavedAssetOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.SavedAsset)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <pre>
-   * Index into `MetaGraphDef.asset_file_def[]` that describes the Asset.
-   * Only the field `AssetFileDef.filename` is used. Other fields, such as
-   * `AssetFileDef.tensor_info`, MUST be ignored.
-   * </pre>
-   *
-   * <code>int32 asset_file_def_index = 1;</code>
-   */
-  int getAssetFileDefIndex();
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedBareConcreteFunction.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedBareConcreteFunction.java
deleted file mode 100644
index e7fadf9f362..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedBareConcreteFunction.java
+++ /dev/null
@@ -1,873 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/saved_object_graph.proto
-
-package org.tensorflow.proto.framework;
-
-/**
- * Protobuf type {@code tensorflow.SavedBareConcreteFunction}
- */
-public  final class SavedBareConcreteFunction extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.SavedBareConcreteFunction)
-    SavedBareConcreteFunctionOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use SavedBareConcreteFunction.newBuilder() to construct.
-  private SavedBareConcreteFunction(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private SavedBareConcreteFunction() {
-    concreteFunctionName_ = "";
-    argumentKeywords_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new SavedBareConcreteFunction();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private SavedBareConcreteFunction(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    int mutable_bitField0_ = 0;
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 10: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            concreteFunctionName_ = s;
-            break;
-          }
-          case 18: {
-            java.lang.String s = input.readStringRequireUtf8();
-            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
-              argumentKeywords_ = new com.google.protobuf.LazyStringArrayList();
-              mutable_bitField0_ |= 0x00000001;
-            }
-            argumentKeywords_.add(s);
-            break;
-          }
-          case 24: {
-
-            allowedPositionalArguments_ = input.readInt64();
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      if (((mutable_bitField0_ & 0x00000001) != 0)) {
-        argumentKeywords_ = argumentKeywords_.getUnmodifiableView();
-      }
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.framework.SavedObjectGraphProtos.internal_static_tensorflow_SavedBareConcreteFunction_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.framework.SavedObjectGraphProtos.internal_static_tensorflow_SavedBareConcreteFunction_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.framework.SavedBareConcreteFunction.class, org.tensorflow.proto.framework.SavedBareConcreteFunction.Builder.class);
-  }
-
-  public static final int CONCRETE_FUNCTION_NAME_FIELD_NUMBER = 1;
-  private volatile java.lang.Object concreteFunctionName_;
-  /**
-   * <pre>
-   * Identifies a SavedConcreteFunction.
-   * </pre>
-   *
-   * <code>string concrete_function_name = 1;</code>
-   */
-  public java.lang.String getConcreteFunctionName() {
-    java.lang.Object ref = concreteFunctionName_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      concreteFunctionName_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * Identifies a SavedConcreteFunction.
-   * </pre>
-   *
-   * <code>string concrete_function_name = 1;</code>
-   */
-  public com.google.protobuf.ByteString
-      getConcreteFunctionNameBytes() {
-    java.lang.Object ref = concreteFunctionName_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      concreteFunctionName_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int ARGUMENT_KEYWORDS_FIELD_NUMBER = 2;
-  private com.google.protobuf.LazyStringList argumentKeywords_;
-  /**
-   * <pre>
-   * A sequence of unique strings, one per Tensor argument.
-   * </pre>
-   *
-   * <code>repeated string argument_keywords = 2;</code>
-   */
-  public com.google.protobuf.ProtocolStringList
-      getArgumentKeywordsList() {
-    return argumentKeywords_;
-  }
-  /**
-   * <pre>
-   * A sequence of unique strings, one per Tensor argument.
-   * </pre>
-   *
-   * <code>repeated string argument_keywords = 2;</code>
-   */
-  public int getArgumentKeywordsCount() {
-    return argumentKeywords_.size();
-  }
-  /**
-   * <pre>
-   * A sequence of unique strings, one per Tensor argument.
-   * </pre>
-   *
-   * <code>repeated string argument_keywords = 2;</code>
-   */
-  public java.lang.String getArgumentKeywords(int index) {
-    return argumentKeywords_.get(index);
-  }
-  /**
-   * <pre>
-   * A sequence of unique strings, one per Tensor argument.
-   * </pre>
-   *
-   * <code>repeated string argument_keywords = 2;</code>
-   */
-  public com.google.protobuf.ByteString
-      getArgumentKeywordsBytes(int index) {
-    return argumentKeywords_.getByteString(index);
-  }
-
-  public static final int ALLOWED_POSITIONAL_ARGUMENTS_FIELD_NUMBER = 3;
-  private long allowedPositionalArguments_;
-  /**
-   * <pre>
-   * The prefix of `argument_keywords` which may be identified by position.
-   * </pre>
-   *
-   * <code>int64 allowed_positional_arguments = 3;</code>
-   */
-  public long getAllowedPositionalArguments() {
-    return allowedPositionalArguments_;
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    if (!getConcreteFunctionNameBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 1, concreteFunctionName_);
-    }
-    for (int i = 0; i < argumentKeywords_.size(); i++) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 2, argumentKeywords_.getRaw(i));
-    }
-    if (allowedPositionalArguments_ != 0L) {
-      output.writeInt64(3, allowedPositionalArguments_);
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (!getConcreteFunctionNameBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, concreteFunctionName_);
-    }
-    {
-      int dataSize = 0;
-      for (int i = 0; i < argumentKeywords_.size(); i++) {
-        dataSize += computeStringSizeNoTag(argumentKeywords_.getRaw(i));
-      }
-      size += dataSize;
-      size += 1 * getArgumentKeywordsList().size();
-    }
-    if (allowedPositionalArguments_ != 0L) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt64Size(3, allowedPositionalArguments_);
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.framework.SavedBareConcreteFunction)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.framework.SavedBareConcreteFunction other = (org.tensorflow.proto.framework.SavedBareConcreteFunction) obj;
-
-    if (!getConcreteFunctionName()
-        .equals(other.getConcreteFunctionName())) return false;
-    if (!getArgumentKeywordsList()
-        .equals(other.getArgumentKeywordsList())) return false;
-    if (getAllowedPositionalArguments()
-        != other.getAllowedPositionalArguments()) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    hash = (37 * hash) + CONCRETE_FUNCTION_NAME_FIELD_NUMBER;
-    hash = (53 * hash) + getConcreteFunctionName().hashCode();
-    if (getArgumentKeywordsCount() > 0) {
-      hash = (37 * hash) + ARGUMENT_KEYWORDS_FIELD_NUMBER;
-      hash = (53 * hash) + getArgumentKeywordsList().hashCode();
-    }
-    hash = (37 * hash) + ALLOWED_POSITIONAL_ARGUMENTS_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-        getAllowedPositionalArguments());
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.framework.SavedBareConcreteFunction parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.SavedBareConcreteFunction parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.SavedBareConcreteFunction parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.SavedBareConcreteFunction parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.SavedBareConcreteFunction parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.SavedBareConcreteFunction parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.SavedBareConcreteFunction parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.SavedBareConcreteFunction parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.SavedBareConcreteFunction parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.SavedBareConcreteFunction parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.SavedBareConcreteFunction parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.SavedBareConcreteFunction parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.framework.SavedBareConcreteFunction prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * Protobuf type {@code tensorflow.SavedBareConcreteFunction}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.SavedBareConcreteFunction)
-      org.tensorflow.proto.framework.SavedBareConcreteFunctionOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.SavedObjectGraphProtos.internal_static_tensorflow_SavedBareConcreteFunction_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.SavedObjectGraphProtos.internal_static_tensorflow_SavedBareConcreteFunction_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.SavedBareConcreteFunction.class, org.tensorflow.proto.framework.SavedBareConcreteFunction.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.framework.SavedBareConcreteFunction.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      concreteFunctionName_ = "";
-
-      argumentKeywords_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-      bitField0_ = (bitField0_ & ~0x00000001);
-      allowedPositionalArguments_ = 0L;
-
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.framework.SavedObjectGraphProtos.internal_static_tensorflow_SavedBareConcreteFunction_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.SavedBareConcreteFunction getDefaultInstanceForType() {
-      return org.tensorflow.proto.framework.SavedBareConcreteFunction.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.SavedBareConcreteFunction build() {
-      org.tensorflow.proto.framework.SavedBareConcreteFunction result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.SavedBareConcreteFunction buildPartial() {
-      org.tensorflow.proto.framework.SavedBareConcreteFunction result = new org.tensorflow.proto.framework.SavedBareConcreteFunction(this);
-      int from_bitField0_ = bitField0_;
-      result.concreteFunctionName_ = concreteFunctionName_;
-      if (((bitField0_ & 0x00000001) != 0)) {
-        argumentKeywords_ = argumentKeywords_.getUnmodifiableView();
-        bitField0_ = (bitField0_ & ~0x00000001);
-      }
-      result.argumentKeywords_ = argumentKeywords_;
-      result.allowedPositionalArguments_ = allowedPositionalArguments_;
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.framework.SavedBareConcreteFunction) {
-        return mergeFrom((org.tensorflow.proto.framework.SavedBareConcreteFunction)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.framework.SavedBareConcreteFunction other) {
-      if (other == org.tensorflow.proto.framework.SavedBareConcreteFunction.getDefaultInstance()) return this;
-      if (!other.getConcreteFunctionName().isEmpty()) {
-        concreteFunctionName_ = other.concreteFunctionName_;
-        onChanged();
-      }
-      if (!other.argumentKeywords_.isEmpty()) {
-        if (argumentKeywords_.isEmpty()) {
-          argumentKeywords_ = other.argumentKeywords_;
-          bitField0_ = (bitField0_ & ~0x00000001);
-        } else {
-          ensureArgumentKeywordsIsMutable();
-          argumentKeywords_.addAll(other.argumentKeywords_);
-        }
-        onChanged();
-      }
-      if (other.getAllowedPositionalArguments() != 0L) {
-        setAllowedPositionalArguments(other.getAllowedPositionalArguments());
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.framework.SavedBareConcreteFunction parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.framework.SavedBareConcreteFunction) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-    private int bitField0_;
-
-    private java.lang.Object concreteFunctionName_ = "";
-    /**
-     * <pre>
-     * Identifies a SavedConcreteFunction.
-     * </pre>
-     *
-     * <code>string concrete_function_name = 1;</code>
-     */
-    public java.lang.String getConcreteFunctionName() {
-      java.lang.Object ref = concreteFunctionName_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        concreteFunctionName_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Identifies a SavedConcreteFunction.
-     * </pre>
-     *
-     * <code>string concrete_function_name = 1;</code>
-     */
-    public com.google.protobuf.ByteString
-        getConcreteFunctionNameBytes() {
-      java.lang.Object ref = concreteFunctionName_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        concreteFunctionName_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Identifies a SavedConcreteFunction.
-     * </pre>
-     *
-     * <code>string concrete_function_name = 1;</code>
-     */
-    public Builder setConcreteFunctionName(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      concreteFunctionName_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Identifies a SavedConcreteFunction.
-     * </pre>
-     *
-     * <code>string concrete_function_name = 1;</code>
-     */
-    public Builder clearConcreteFunctionName() {
-      
-      concreteFunctionName_ = getDefaultInstance().getConcreteFunctionName();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Identifies a SavedConcreteFunction.
-     * </pre>
-     *
-     * <code>string concrete_function_name = 1;</code>
-     */
-    public Builder setConcreteFunctionNameBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      concreteFunctionName_ = value;
-      onChanged();
-      return this;
-    }
-
-    private com.google.protobuf.LazyStringList argumentKeywords_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-    private void ensureArgumentKeywordsIsMutable() {
-      if (!((bitField0_ & 0x00000001) != 0)) {
-        argumentKeywords_ = new com.google.protobuf.LazyStringArrayList(argumentKeywords_);
-        bitField0_ |= 0x00000001;
-       }
-    }
-    /**
-     * <pre>
-     * A sequence of unique strings, one per Tensor argument.
-     * </pre>
-     *
-     * <code>repeated string argument_keywords = 2;</code>
-     */
-    public com.google.protobuf.ProtocolStringList
-        getArgumentKeywordsList() {
-      return argumentKeywords_.getUnmodifiableView();
-    }
-    /**
-     * <pre>
-     * A sequence of unique strings, one per Tensor argument.
-     * </pre>
-     *
-     * <code>repeated string argument_keywords = 2;</code>
-     */
-    public int getArgumentKeywordsCount() {
-      return argumentKeywords_.size();
-    }
-    /**
-     * <pre>
-     * A sequence of unique strings, one per Tensor argument.
-     * </pre>
-     *
-     * <code>repeated string argument_keywords = 2;</code>
-     */
-    public java.lang.String getArgumentKeywords(int index) {
-      return argumentKeywords_.get(index);
-    }
-    /**
-     * <pre>
-     * A sequence of unique strings, one per Tensor argument.
-     * </pre>
-     *
-     * <code>repeated string argument_keywords = 2;</code>
-     */
-    public com.google.protobuf.ByteString
-        getArgumentKeywordsBytes(int index) {
-      return argumentKeywords_.getByteString(index);
-    }
-    /**
-     * <pre>
-     * A sequence of unique strings, one per Tensor argument.
-     * </pre>
-     *
-     * <code>repeated string argument_keywords = 2;</code>
-     */
-    public Builder setArgumentKeywords(
-        int index, java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  ensureArgumentKeywordsIsMutable();
-      argumentKeywords_.set(index, value);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * A sequence of unique strings, one per Tensor argument.
-     * </pre>
-     *
-     * <code>repeated string argument_keywords = 2;</code>
-     */
-    public Builder addArgumentKeywords(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  ensureArgumentKeywordsIsMutable();
-      argumentKeywords_.add(value);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * A sequence of unique strings, one per Tensor argument.
-     * </pre>
-     *
-     * <code>repeated string argument_keywords = 2;</code>
-     */
-    public Builder addAllArgumentKeywords(
-        java.lang.Iterable<java.lang.String> values) {
-      ensureArgumentKeywordsIsMutable();
-      com.google.protobuf.AbstractMessageLite.Builder.addAll(
-          values, argumentKeywords_);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * A sequence of unique strings, one per Tensor argument.
-     * </pre>
-     *
-     * <code>repeated string argument_keywords = 2;</code>
-     */
-    public Builder clearArgumentKeywords() {
-      argumentKeywords_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-      bitField0_ = (bitField0_ & ~0x00000001);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * A sequence of unique strings, one per Tensor argument.
-     * </pre>
-     *
-     * <code>repeated string argument_keywords = 2;</code>
-     */
-    public Builder addArgumentKeywordsBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      ensureArgumentKeywordsIsMutable();
-      argumentKeywords_.add(value);
-      onChanged();
-      return this;
-    }
-
-    private long allowedPositionalArguments_ ;
-    /**
-     * <pre>
-     * The prefix of `argument_keywords` which may be identified by position.
-     * </pre>
-     *
-     * <code>int64 allowed_positional_arguments = 3;</code>
-     */
-    public long getAllowedPositionalArguments() {
-      return allowedPositionalArguments_;
-    }
-    /**
-     * <pre>
-     * The prefix of `argument_keywords` which may be identified by position.
-     * </pre>
-     *
-     * <code>int64 allowed_positional_arguments = 3;</code>
-     */
-    public Builder setAllowedPositionalArguments(long value) {
-      
-      allowedPositionalArguments_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * The prefix of `argument_keywords` which may be identified by position.
-     * </pre>
-     *
-     * <code>int64 allowed_positional_arguments = 3;</code>
-     */
-    public Builder clearAllowedPositionalArguments() {
-      
-      allowedPositionalArguments_ = 0L;
-      onChanged();
-      return this;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.SavedBareConcreteFunction)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.SavedBareConcreteFunction)
-  private static final org.tensorflow.proto.framework.SavedBareConcreteFunction DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.SavedBareConcreteFunction();
-  }
-
-  public static org.tensorflow.proto.framework.SavedBareConcreteFunction getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<SavedBareConcreteFunction>
-      PARSER = new com.google.protobuf.AbstractParser<SavedBareConcreteFunction>() {
-    @java.lang.Override
-    public SavedBareConcreteFunction parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new SavedBareConcreteFunction(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<SavedBareConcreteFunction> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<SavedBareConcreteFunction> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.framework.SavedBareConcreteFunction getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedBareConcreteFunctionOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedBareConcreteFunctionOrBuilder.java
deleted file mode 100644
index cf9a625e8b3..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedBareConcreteFunctionOrBuilder.java
+++ /dev/null
@@ -1,71 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/saved_object_graph.proto
-
-package org.tensorflow.proto.framework;
-
-public interface SavedBareConcreteFunctionOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.SavedBareConcreteFunction)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <pre>
-   * Identifies a SavedConcreteFunction.
-   * </pre>
-   *
-   * <code>string concrete_function_name = 1;</code>
-   */
-  java.lang.String getConcreteFunctionName();
-  /**
-   * <pre>
-   * Identifies a SavedConcreteFunction.
-   * </pre>
-   *
-   * <code>string concrete_function_name = 1;</code>
-   */
-  com.google.protobuf.ByteString
-      getConcreteFunctionNameBytes();
-
-  /**
-   * <pre>
-   * A sequence of unique strings, one per Tensor argument.
-   * </pre>
-   *
-   * <code>repeated string argument_keywords = 2;</code>
-   */
-  java.util.List<java.lang.String>
-      getArgumentKeywordsList();
-  /**
-   * <pre>
-   * A sequence of unique strings, one per Tensor argument.
-   * </pre>
-   *
-   * <code>repeated string argument_keywords = 2;</code>
-   */
-  int getArgumentKeywordsCount();
-  /**
-   * <pre>
-   * A sequence of unique strings, one per Tensor argument.
-   * </pre>
-   *
-   * <code>repeated string argument_keywords = 2;</code>
-   */
-  java.lang.String getArgumentKeywords(int index);
-  /**
-   * <pre>
-   * A sequence of unique strings, one per Tensor argument.
-   * </pre>
-   *
-   * <code>repeated string argument_keywords = 2;</code>
-   */
-  com.google.protobuf.ByteString
-      getArgumentKeywordsBytes(int index);
-
-  /**
-   * <pre>
-   * The prefix of `argument_keywords` which may be identified by position.
-   * </pre>
-   *
-   * <code>int64 allowed_positional_arguments = 3;</code>
-   */
-  long getAllowedPositionalArguments();
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedConcreteFunction.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedConcreteFunction.java
deleted file mode 100644
index 62106345581..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedConcreteFunction.java
+++ /dev/null
@@ -1,1156 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/saved_object_graph.proto
-
-package org.tensorflow.proto.framework;
-
-/**
- * <pre>
- * Stores low-level information about a concrete function. Referenced in either
- * a SavedFunction or a SavedBareConcreteFunction.
- * </pre>
- *
- * Protobuf type {@code tensorflow.SavedConcreteFunction}
- */
-public  final class SavedConcreteFunction extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.SavedConcreteFunction)
-    SavedConcreteFunctionOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use SavedConcreteFunction.newBuilder() to construct.
-  private SavedConcreteFunction(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private SavedConcreteFunction() {
-    boundInputs_ = emptyIntList();
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new SavedConcreteFunction();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private SavedConcreteFunction(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    int mutable_bitField0_ = 0;
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 16: {
-            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
-              boundInputs_ = newIntList();
-              mutable_bitField0_ |= 0x00000001;
-            }
-            boundInputs_.addInt(input.readInt32());
-            break;
-          }
-          case 18: {
-            int length = input.readRawVarint32();
-            int limit = input.pushLimit(length);
-            if (!((mutable_bitField0_ & 0x00000001) != 0) && input.getBytesUntilLimit() > 0) {
-              boundInputs_ = newIntList();
-              mutable_bitField0_ |= 0x00000001;
-            }
-            while (input.getBytesUntilLimit() > 0) {
-              boundInputs_.addInt(input.readInt32());
-            }
-            input.popLimit(limit);
-            break;
-          }
-          case 26: {
-            org.tensorflow.proto.framework.StructuredValue.Builder subBuilder = null;
-            if (canonicalizedInputSignature_ != null) {
-              subBuilder = canonicalizedInputSignature_.toBuilder();
-            }
-            canonicalizedInputSignature_ = input.readMessage(org.tensorflow.proto.framework.StructuredValue.parser(), extensionRegistry);
-            if (subBuilder != null) {
-              subBuilder.mergeFrom(canonicalizedInputSignature_);
-              canonicalizedInputSignature_ = subBuilder.buildPartial();
-            }
-
-            break;
-          }
-          case 34: {
-            org.tensorflow.proto.framework.StructuredValue.Builder subBuilder = null;
-            if (outputSignature_ != null) {
-              subBuilder = outputSignature_.toBuilder();
-            }
-            outputSignature_ = input.readMessage(org.tensorflow.proto.framework.StructuredValue.parser(), extensionRegistry);
-            if (subBuilder != null) {
-              subBuilder.mergeFrom(outputSignature_);
-              outputSignature_ = subBuilder.buildPartial();
-            }
-
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      if (((mutable_bitField0_ & 0x00000001) != 0)) {
-        boundInputs_.makeImmutable(); // C
-      }
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.framework.SavedObjectGraphProtos.internal_static_tensorflow_SavedConcreteFunction_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.framework.SavedObjectGraphProtos.internal_static_tensorflow_SavedConcreteFunction_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.framework.SavedConcreteFunction.class, org.tensorflow.proto.framework.SavedConcreteFunction.Builder.class);
-  }
-
-  public static final int BOUND_INPUTS_FIELD_NUMBER = 2;
-  private com.google.protobuf.Internal.IntList boundInputs_;
-  /**
-   * <pre>
-   * Bound inputs to the function. The SavedObjects identified by the node ids
-   * given here are appended as extra inputs to the caller-supplied inputs.
-   * The only types of SavedObjects valid here are SavedVariable, SavedResource
-   * and SavedAsset.
-   * </pre>
-   *
-   * <code>repeated int32 bound_inputs = 2;</code>
-   */
-  public java.util.List<java.lang.Integer>
-      getBoundInputsList() {
-    return boundInputs_;
-  }
-  /**
-   * <pre>
-   * Bound inputs to the function. The SavedObjects identified by the node ids
-   * given here are appended as extra inputs to the caller-supplied inputs.
-   * The only types of SavedObjects valid here are SavedVariable, SavedResource
-   * and SavedAsset.
-   * </pre>
-   *
-   * <code>repeated int32 bound_inputs = 2;</code>
-   */
-  public int getBoundInputsCount() {
-    return boundInputs_.size();
-  }
-  /**
-   * <pre>
-   * Bound inputs to the function. The SavedObjects identified by the node ids
-   * given here are appended as extra inputs to the caller-supplied inputs.
-   * The only types of SavedObjects valid here are SavedVariable, SavedResource
-   * and SavedAsset.
-   * </pre>
-   *
-   * <code>repeated int32 bound_inputs = 2;</code>
-   */
-  public int getBoundInputs(int index) {
-    return boundInputs_.getInt(index);
-  }
-  private int boundInputsMemoizedSerializedSize = -1;
-
-  public static final int CANONICALIZED_INPUT_SIGNATURE_FIELD_NUMBER = 3;
-  private org.tensorflow.proto.framework.StructuredValue canonicalizedInputSignature_;
-  /**
-   * <pre>
-   * Input in canonicalized form that was received to create this concrete
-   * function.
-   * </pre>
-   *
-   * <code>.tensorflow.StructuredValue canonicalized_input_signature = 3;</code>
-   */
-  public boolean hasCanonicalizedInputSignature() {
-    return canonicalizedInputSignature_ != null;
-  }
-  /**
-   * <pre>
-   * Input in canonicalized form that was received to create this concrete
-   * function.
-   * </pre>
-   *
-   * <code>.tensorflow.StructuredValue canonicalized_input_signature = 3;</code>
-   */
-  public org.tensorflow.proto.framework.StructuredValue getCanonicalizedInputSignature() {
-    return canonicalizedInputSignature_ == null ? org.tensorflow.proto.framework.StructuredValue.getDefaultInstance() : canonicalizedInputSignature_;
-  }
-  /**
-   * <pre>
-   * Input in canonicalized form that was received to create this concrete
-   * function.
-   * </pre>
-   *
-   * <code>.tensorflow.StructuredValue canonicalized_input_signature = 3;</code>
-   */
-  public org.tensorflow.proto.framework.StructuredValueOrBuilder getCanonicalizedInputSignatureOrBuilder() {
-    return getCanonicalizedInputSignature();
-  }
-
-  public static final int OUTPUT_SIGNATURE_FIELD_NUMBER = 4;
-  private org.tensorflow.proto.framework.StructuredValue outputSignature_;
-  /**
-   * <pre>
-   * Output that was the return value of this function after replacing all
-   * Tensors with TensorSpecs. This can be an arbitrary nested function and will
-   * be used to reconstruct the full structure from pure tensors.
-   * </pre>
-   *
-   * <code>.tensorflow.StructuredValue output_signature = 4;</code>
-   */
-  public boolean hasOutputSignature() {
-    return outputSignature_ != null;
-  }
-  /**
-   * <pre>
-   * Output that was the return value of this function after replacing all
-   * Tensors with TensorSpecs. This can be an arbitrary nested function and will
-   * be used to reconstruct the full structure from pure tensors.
-   * </pre>
-   *
-   * <code>.tensorflow.StructuredValue output_signature = 4;</code>
-   */
-  public org.tensorflow.proto.framework.StructuredValue getOutputSignature() {
-    return outputSignature_ == null ? org.tensorflow.proto.framework.StructuredValue.getDefaultInstance() : outputSignature_;
-  }
-  /**
-   * <pre>
-   * Output that was the return value of this function after replacing all
-   * Tensors with TensorSpecs. This can be an arbitrary nested function and will
-   * be used to reconstruct the full structure from pure tensors.
-   * </pre>
-   *
-   * <code>.tensorflow.StructuredValue output_signature = 4;</code>
-   */
-  public org.tensorflow.proto.framework.StructuredValueOrBuilder getOutputSignatureOrBuilder() {
-    return getOutputSignature();
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    getSerializedSize();
-    if (getBoundInputsList().size() > 0) {
-      output.writeUInt32NoTag(18);
-      output.writeUInt32NoTag(boundInputsMemoizedSerializedSize);
-    }
-    for (int i = 0; i < boundInputs_.size(); i++) {
-      output.writeInt32NoTag(boundInputs_.getInt(i));
-    }
-    if (canonicalizedInputSignature_ != null) {
-      output.writeMessage(3, getCanonicalizedInputSignature());
-    }
-    if (outputSignature_ != null) {
-      output.writeMessage(4, getOutputSignature());
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    {
-      int dataSize = 0;
-      for (int i = 0; i < boundInputs_.size(); i++) {
-        dataSize += com.google.protobuf.CodedOutputStream
-          .computeInt32SizeNoTag(boundInputs_.getInt(i));
-      }
-      size += dataSize;
-      if (!getBoundInputsList().isEmpty()) {
-        size += 1;
-        size += com.google.protobuf.CodedOutputStream
-            .computeInt32SizeNoTag(dataSize);
-      }
-      boundInputsMemoizedSerializedSize = dataSize;
-    }
-    if (canonicalizedInputSignature_ != null) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(3, getCanonicalizedInputSignature());
-    }
-    if (outputSignature_ != null) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(4, getOutputSignature());
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.framework.SavedConcreteFunction)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.framework.SavedConcreteFunction other = (org.tensorflow.proto.framework.SavedConcreteFunction) obj;
-
-    if (!getBoundInputsList()
-        .equals(other.getBoundInputsList())) return false;
-    if (hasCanonicalizedInputSignature() != other.hasCanonicalizedInputSignature()) return false;
-    if (hasCanonicalizedInputSignature()) {
-      if (!getCanonicalizedInputSignature()
-          .equals(other.getCanonicalizedInputSignature())) return false;
-    }
-    if (hasOutputSignature() != other.hasOutputSignature()) return false;
-    if (hasOutputSignature()) {
-      if (!getOutputSignature()
-          .equals(other.getOutputSignature())) return false;
-    }
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    if (getBoundInputsCount() > 0) {
-      hash = (37 * hash) + BOUND_INPUTS_FIELD_NUMBER;
-      hash = (53 * hash) + getBoundInputsList().hashCode();
-    }
-    if (hasCanonicalizedInputSignature()) {
-      hash = (37 * hash) + CANONICALIZED_INPUT_SIGNATURE_FIELD_NUMBER;
-      hash = (53 * hash) + getCanonicalizedInputSignature().hashCode();
-    }
-    if (hasOutputSignature()) {
-      hash = (37 * hash) + OUTPUT_SIGNATURE_FIELD_NUMBER;
-      hash = (53 * hash) + getOutputSignature().hashCode();
-    }
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.framework.SavedConcreteFunction parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.SavedConcreteFunction parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.SavedConcreteFunction parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.SavedConcreteFunction parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.SavedConcreteFunction parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.SavedConcreteFunction parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.SavedConcreteFunction parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.SavedConcreteFunction parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.SavedConcreteFunction parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.SavedConcreteFunction parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.SavedConcreteFunction parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.SavedConcreteFunction parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.framework.SavedConcreteFunction prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * <pre>
-   * Stores low-level information about a concrete function. Referenced in either
-   * a SavedFunction or a SavedBareConcreteFunction.
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.SavedConcreteFunction}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.SavedConcreteFunction)
-      org.tensorflow.proto.framework.SavedConcreteFunctionOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.SavedObjectGraphProtos.internal_static_tensorflow_SavedConcreteFunction_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.SavedObjectGraphProtos.internal_static_tensorflow_SavedConcreteFunction_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.SavedConcreteFunction.class, org.tensorflow.proto.framework.SavedConcreteFunction.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.framework.SavedConcreteFunction.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      boundInputs_ = emptyIntList();
-      bitField0_ = (bitField0_ & ~0x00000001);
-      if (canonicalizedInputSignatureBuilder_ == null) {
-        canonicalizedInputSignature_ = null;
-      } else {
-        canonicalizedInputSignature_ = null;
-        canonicalizedInputSignatureBuilder_ = null;
-      }
-      if (outputSignatureBuilder_ == null) {
-        outputSignature_ = null;
-      } else {
-        outputSignature_ = null;
-        outputSignatureBuilder_ = null;
-      }
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.framework.SavedObjectGraphProtos.internal_static_tensorflow_SavedConcreteFunction_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.SavedConcreteFunction getDefaultInstanceForType() {
-      return org.tensorflow.proto.framework.SavedConcreteFunction.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.SavedConcreteFunction build() {
-      org.tensorflow.proto.framework.SavedConcreteFunction result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.SavedConcreteFunction buildPartial() {
-      org.tensorflow.proto.framework.SavedConcreteFunction result = new org.tensorflow.proto.framework.SavedConcreteFunction(this);
-      int from_bitField0_ = bitField0_;
-      if (((bitField0_ & 0x00000001) != 0)) {
-        boundInputs_.makeImmutable();
-        bitField0_ = (bitField0_ & ~0x00000001);
-      }
-      result.boundInputs_ = boundInputs_;
-      if (canonicalizedInputSignatureBuilder_ == null) {
-        result.canonicalizedInputSignature_ = canonicalizedInputSignature_;
-      } else {
-        result.canonicalizedInputSignature_ = canonicalizedInputSignatureBuilder_.build();
-      }
-      if (outputSignatureBuilder_ == null) {
-        result.outputSignature_ = outputSignature_;
-      } else {
-        result.outputSignature_ = outputSignatureBuilder_.build();
-      }
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.framework.SavedConcreteFunction) {
-        return mergeFrom((org.tensorflow.proto.framework.SavedConcreteFunction)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.framework.SavedConcreteFunction other) {
-      if (other == org.tensorflow.proto.framework.SavedConcreteFunction.getDefaultInstance()) return this;
-      if (!other.boundInputs_.isEmpty()) {
-        if (boundInputs_.isEmpty()) {
-          boundInputs_ = other.boundInputs_;
-          bitField0_ = (bitField0_ & ~0x00000001);
-        } else {
-          ensureBoundInputsIsMutable();
-          boundInputs_.addAll(other.boundInputs_);
-        }
-        onChanged();
-      }
-      if (other.hasCanonicalizedInputSignature()) {
-        mergeCanonicalizedInputSignature(other.getCanonicalizedInputSignature());
-      }
-      if (other.hasOutputSignature()) {
-        mergeOutputSignature(other.getOutputSignature());
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.framework.SavedConcreteFunction parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.framework.SavedConcreteFunction) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-    private int bitField0_;
-
-    private com.google.protobuf.Internal.IntList boundInputs_ = emptyIntList();
-    private void ensureBoundInputsIsMutable() {
-      if (!((bitField0_ & 0x00000001) != 0)) {
-        boundInputs_ = mutableCopy(boundInputs_);
-        bitField0_ |= 0x00000001;
-       }
-    }
-    /**
-     * <pre>
-     * Bound inputs to the function. The SavedObjects identified by the node ids
-     * given here are appended as extra inputs to the caller-supplied inputs.
-     * The only types of SavedObjects valid here are SavedVariable, SavedResource
-     * and SavedAsset.
-     * </pre>
-     *
-     * <code>repeated int32 bound_inputs = 2;</code>
-     */
-    public java.util.List<java.lang.Integer>
-        getBoundInputsList() {
-      return ((bitField0_ & 0x00000001) != 0) ?
-               java.util.Collections.unmodifiableList(boundInputs_) : boundInputs_;
-    }
-    /**
-     * <pre>
-     * Bound inputs to the function. The SavedObjects identified by the node ids
-     * given here are appended as extra inputs to the caller-supplied inputs.
-     * The only types of SavedObjects valid here are SavedVariable, SavedResource
-     * and SavedAsset.
-     * </pre>
-     *
-     * <code>repeated int32 bound_inputs = 2;</code>
-     */
-    public int getBoundInputsCount() {
-      return boundInputs_.size();
-    }
-    /**
-     * <pre>
-     * Bound inputs to the function. The SavedObjects identified by the node ids
-     * given here are appended as extra inputs to the caller-supplied inputs.
-     * The only types of SavedObjects valid here are SavedVariable, SavedResource
-     * and SavedAsset.
-     * </pre>
-     *
-     * <code>repeated int32 bound_inputs = 2;</code>
-     */
-    public int getBoundInputs(int index) {
-      return boundInputs_.getInt(index);
-    }
-    /**
-     * <pre>
-     * Bound inputs to the function. The SavedObjects identified by the node ids
-     * given here are appended as extra inputs to the caller-supplied inputs.
-     * The only types of SavedObjects valid here are SavedVariable, SavedResource
-     * and SavedAsset.
-     * </pre>
-     *
-     * <code>repeated int32 bound_inputs = 2;</code>
-     */
-    public Builder setBoundInputs(
-        int index, int value) {
-      ensureBoundInputsIsMutable();
-      boundInputs_.setInt(index, value);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Bound inputs to the function. The SavedObjects identified by the node ids
-     * given here are appended as extra inputs to the caller-supplied inputs.
-     * The only types of SavedObjects valid here are SavedVariable, SavedResource
-     * and SavedAsset.
-     * </pre>
-     *
-     * <code>repeated int32 bound_inputs = 2;</code>
-     */
-    public Builder addBoundInputs(int value) {
-      ensureBoundInputsIsMutable();
-      boundInputs_.addInt(value);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Bound inputs to the function. The SavedObjects identified by the node ids
-     * given here are appended as extra inputs to the caller-supplied inputs.
-     * The only types of SavedObjects valid here are SavedVariable, SavedResource
-     * and SavedAsset.
-     * </pre>
-     *
-     * <code>repeated int32 bound_inputs = 2;</code>
-     */
-    public Builder addAllBoundInputs(
-        java.lang.Iterable<? extends java.lang.Integer> values) {
-      ensureBoundInputsIsMutable();
-      com.google.protobuf.AbstractMessageLite.Builder.addAll(
-          values, boundInputs_);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Bound inputs to the function. The SavedObjects identified by the node ids
-     * given here are appended as extra inputs to the caller-supplied inputs.
-     * The only types of SavedObjects valid here are SavedVariable, SavedResource
-     * and SavedAsset.
-     * </pre>
-     *
-     * <code>repeated int32 bound_inputs = 2;</code>
-     */
-    public Builder clearBoundInputs() {
-      boundInputs_ = emptyIntList();
-      bitField0_ = (bitField0_ & ~0x00000001);
-      onChanged();
-      return this;
-    }
-
-    private org.tensorflow.proto.framework.StructuredValue canonicalizedInputSignature_;
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.StructuredValue, org.tensorflow.proto.framework.StructuredValue.Builder, org.tensorflow.proto.framework.StructuredValueOrBuilder> canonicalizedInputSignatureBuilder_;
-    /**
-     * <pre>
-     * Input in canonicalized form that was received to create this concrete
-     * function.
-     * </pre>
-     *
-     * <code>.tensorflow.StructuredValue canonicalized_input_signature = 3;</code>
-     */
-    public boolean hasCanonicalizedInputSignature() {
-      return canonicalizedInputSignatureBuilder_ != null || canonicalizedInputSignature_ != null;
-    }
-    /**
-     * <pre>
-     * Input in canonicalized form that was received to create this concrete
-     * function.
-     * </pre>
-     *
-     * <code>.tensorflow.StructuredValue canonicalized_input_signature = 3;</code>
-     */
-    public org.tensorflow.proto.framework.StructuredValue getCanonicalizedInputSignature() {
-      if (canonicalizedInputSignatureBuilder_ == null) {
-        return canonicalizedInputSignature_ == null ? org.tensorflow.proto.framework.StructuredValue.getDefaultInstance() : canonicalizedInputSignature_;
-      } else {
-        return canonicalizedInputSignatureBuilder_.getMessage();
-      }
-    }
-    /**
-     * <pre>
-     * Input in canonicalized form that was received to create this concrete
-     * function.
-     * </pre>
-     *
-     * <code>.tensorflow.StructuredValue canonicalized_input_signature = 3;</code>
-     */
-    public Builder setCanonicalizedInputSignature(org.tensorflow.proto.framework.StructuredValue value) {
-      if (canonicalizedInputSignatureBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        canonicalizedInputSignature_ = value;
-        onChanged();
-      } else {
-        canonicalizedInputSignatureBuilder_.setMessage(value);
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Input in canonicalized form that was received to create this concrete
-     * function.
-     * </pre>
-     *
-     * <code>.tensorflow.StructuredValue canonicalized_input_signature = 3;</code>
-     */
-    public Builder setCanonicalizedInputSignature(
-        org.tensorflow.proto.framework.StructuredValue.Builder builderForValue) {
-      if (canonicalizedInputSignatureBuilder_ == null) {
-        canonicalizedInputSignature_ = builderForValue.build();
-        onChanged();
-      } else {
-        canonicalizedInputSignatureBuilder_.setMessage(builderForValue.build());
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Input in canonicalized form that was received to create this concrete
-     * function.
-     * </pre>
-     *
-     * <code>.tensorflow.StructuredValue canonicalized_input_signature = 3;</code>
-     */
-    public Builder mergeCanonicalizedInputSignature(org.tensorflow.proto.framework.StructuredValue value) {
-      if (canonicalizedInputSignatureBuilder_ == null) {
-        if (canonicalizedInputSignature_ != null) {
-          canonicalizedInputSignature_ =
-            org.tensorflow.proto.framework.StructuredValue.newBuilder(canonicalizedInputSignature_).mergeFrom(value).buildPartial();
-        } else {
-          canonicalizedInputSignature_ = value;
-        }
-        onChanged();
-      } else {
-        canonicalizedInputSignatureBuilder_.mergeFrom(value);
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Input in canonicalized form that was received to create this concrete
-     * function.
-     * </pre>
-     *
-     * <code>.tensorflow.StructuredValue canonicalized_input_signature = 3;</code>
-     */
-    public Builder clearCanonicalizedInputSignature() {
-      if (canonicalizedInputSignatureBuilder_ == null) {
-        canonicalizedInputSignature_ = null;
-        onChanged();
-      } else {
-        canonicalizedInputSignature_ = null;
-        canonicalizedInputSignatureBuilder_ = null;
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Input in canonicalized form that was received to create this concrete
-     * function.
-     * </pre>
-     *
-     * <code>.tensorflow.StructuredValue canonicalized_input_signature = 3;</code>
-     */
-    public org.tensorflow.proto.framework.StructuredValue.Builder getCanonicalizedInputSignatureBuilder() {
-      
-      onChanged();
-      return getCanonicalizedInputSignatureFieldBuilder().getBuilder();
-    }
-    /**
-     * <pre>
-     * Input in canonicalized form that was received to create this concrete
-     * function.
-     * </pre>
-     *
-     * <code>.tensorflow.StructuredValue canonicalized_input_signature = 3;</code>
-     */
-    public org.tensorflow.proto.framework.StructuredValueOrBuilder getCanonicalizedInputSignatureOrBuilder() {
-      if (canonicalizedInputSignatureBuilder_ != null) {
-        return canonicalizedInputSignatureBuilder_.getMessageOrBuilder();
-      } else {
-        return canonicalizedInputSignature_ == null ?
-            org.tensorflow.proto.framework.StructuredValue.getDefaultInstance() : canonicalizedInputSignature_;
-      }
-    }
-    /**
-     * <pre>
-     * Input in canonicalized form that was received to create this concrete
-     * function.
-     * </pre>
-     *
-     * <code>.tensorflow.StructuredValue canonicalized_input_signature = 3;</code>
-     */
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.StructuredValue, org.tensorflow.proto.framework.StructuredValue.Builder, org.tensorflow.proto.framework.StructuredValueOrBuilder> 
-        getCanonicalizedInputSignatureFieldBuilder() {
-      if (canonicalizedInputSignatureBuilder_ == null) {
-        canonicalizedInputSignatureBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-            org.tensorflow.proto.framework.StructuredValue, org.tensorflow.proto.framework.StructuredValue.Builder, org.tensorflow.proto.framework.StructuredValueOrBuilder>(
-                getCanonicalizedInputSignature(),
-                getParentForChildren(),
-                isClean());
-        canonicalizedInputSignature_ = null;
-      }
-      return canonicalizedInputSignatureBuilder_;
-    }
-
-    private org.tensorflow.proto.framework.StructuredValue outputSignature_;
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.StructuredValue, org.tensorflow.proto.framework.StructuredValue.Builder, org.tensorflow.proto.framework.StructuredValueOrBuilder> outputSignatureBuilder_;
-    /**
-     * <pre>
-     * Output that was the return value of this function after replacing all
-     * Tensors with TensorSpecs. This can be an arbitrary nested function and will
-     * be used to reconstruct the full structure from pure tensors.
-     * </pre>
-     *
-     * <code>.tensorflow.StructuredValue output_signature = 4;</code>
-     */
-    public boolean hasOutputSignature() {
-      return outputSignatureBuilder_ != null || outputSignature_ != null;
-    }
-    /**
-     * <pre>
-     * Output that was the return value of this function after replacing all
-     * Tensors with TensorSpecs. This can be an arbitrary nested function and will
-     * be used to reconstruct the full structure from pure tensors.
-     * </pre>
-     *
-     * <code>.tensorflow.StructuredValue output_signature = 4;</code>
-     */
-    public org.tensorflow.proto.framework.StructuredValue getOutputSignature() {
-      if (outputSignatureBuilder_ == null) {
-        return outputSignature_ == null ? org.tensorflow.proto.framework.StructuredValue.getDefaultInstance() : outputSignature_;
-      } else {
-        return outputSignatureBuilder_.getMessage();
-      }
-    }
-    /**
-     * <pre>
-     * Output that was the return value of this function after replacing all
-     * Tensors with TensorSpecs. This can be an arbitrary nested function and will
-     * be used to reconstruct the full structure from pure tensors.
-     * </pre>
-     *
-     * <code>.tensorflow.StructuredValue output_signature = 4;</code>
-     */
-    public Builder setOutputSignature(org.tensorflow.proto.framework.StructuredValue value) {
-      if (outputSignatureBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        outputSignature_ = value;
-        onChanged();
-      } else {
-        outputSignatureBuilder_.setMessage(value);
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Output that was the return value of this function after replacing all
-     * Tensors with TensorSpecs. This can be an arbitrary nested function and will
-     * be used to reconstruct the full structure from pure tensors.
-     * </pre>
-     *
-     * <code>.tensorflow.StructuredValue output_signature = 4;</code>
-     */
-    public Builder setOutputSignature(
-        org.tensorflow.proto.framework.StructuredValue.Builder builderForValue) {
-      if (outputSignatureBuilder_ == null) {
-        outputSignature_ = builderForValue.build();
-        onChanged();
-      } else {
-        outputSignatureBuilder_.setMessage(builderForValue.build());
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Output that was the return value of this function after replacing all
-     * Tensors with TensorSpecs. This can be an arbitrary nested function and will
-     * be used to reconstruct the full structure from pure tensors.
-     * </pre>
-     *
-     * <code>.tensorflow.StructuredValue output_signature = 4;</code>
-     */
-    public Builder mergeOutputSignature(org.tensorflow.proto.framework.StructuredValue value) {
-      if (outputSignatureBuilder_ == null) {
-        if (outputSignature_ != null) {
-          outputSignature_ =
-            org.tensorflow.proto.framework.StructuredValue.newBuilder(outputSignature_).mergeFrom(value).buildPartial();
-        } else {
-          outputSignature_ = value;
-        }
-        onChanged();
-      } else {
-        outputSignatureBuilder_.mergeFrom(value);
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Output that was the return value of this function after replacing all
-     * Tensors with TensorSpecs. This can be an arbitrary nested function and will
-     * be used to reconstruct the full structure from pure tensors.
-     * </pre>
-     *
-     * <code>.tensorflow.StructuredValue output_signature = 4;</code>
-     */
-    public Builder clearOutputSignature() {
-      if (outputSignatureBuilder_ == null) {
-        outputSignature_ = null;
-        onChanged();
-      } else {
-        outputSignature_ = null;
-        outputSignatureBuilder_ = null;
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Output that was the return value of this function after replacing all
-     * Tensors with TensorSpecs. This can be an arbitrary nested function and will
-     * be used to reconstruct the full structure from pure tensors.
-     * </pre>
-     *
-     * <code>.tensorflow.StructuredValue output_signature = 4;</code>
-     */
-    public org.tensorflow.proto.framework.StructuredValue.Builder getOutputSignatureBuilder() {
-      
-      onChanged();
-      return getOutputSignatureFieldBuilder().getBuilder();
-    }
-    /**
-     * <pre>
-     * Output that was the return value of this function after replacing all
-     * Tensors with TensorSpecs. This can be an arbitrary nested function and will
-     * be used to reconstruct the full structure from pure tensors.
-     * </pre>
-     *
-     * <code>.tensorflow.StructuredValue output_signature = 4;</code>
-     */
-    public org.tensorflow.proto.framework.StructuredValueOrBuilder getOutputSignatureOrBuilder() {
-      if (outputSignatureBuilder_ != null) {
-        return outputSignatureBuilder_.getMessageOrBuilder();
-      } else {
-        return outputSignature_ == null ?
-            org.tensorflow.proto.framework.StructuredValue.getDefaultInstance() : outputSignature_;
-      }
-    }
-    /**
-     * <pre>
-     * Output that was the return value of this function after replacing all
-     * Tensors with TensorSpecs. This can be an arbitrary nested function and will
-     * be used to reconstruct the full structure from pure tensors.
-     * </pre>
-     *
-     * <code>.tensorflow.StructuredValue output_signature = 4;</code>
-     */
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.StructuredValue, org.tensorflow.proto.framework.StructuredValue.Builder, org.tensorflow.proto.framework.StructuredValueOrBuilder> 
-        getOutputSignatureFieldBuilder() {
-      if (outputSignatureBuilder_ == null) {
-        outputSignatureBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-            org.tensorflow.proto.framework.StructuredValue, org.tensorflow.proto.framework.StructuredValue.Builder, org.tensorflow.proto.framework.StructuredValueOrBuilder>(
-                getOutputSignature(),
-                getParentForChildren(),
-                isClean());
-        outputSignature_ = null;
-      }
-      return outputSignatureBuilder_;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.SavedConcreteFunction)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.SavedConcreteFunction)
-  private static final org.tensorflow.proto.framework.SavedConcreteFunction DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.SavedConcreteFunction();
-  }
-
-  public static org.tensorflow.proto.framework.SavedConcreteFunction getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<SavedConcreteFunction>
-      PARSER = new com.google.protobuf.AbstractParser<SavedConcreteFunction>() {
-    @java.lang.Override
-    public SavedConcreteFunction parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new SavedConcreteFunction(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<SavedConcreteFunction> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<SavedConcreteFunction> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.framework.SavedConcreteFunction getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedConcreteFunctionOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedConcreteFunctionOrBuilder.java
deleted file mode 100644
index 49145b8ca01..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedConcreteFunctionOrBuilder.java
+++ /dev/null
@@ -1,102 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/saved_object_graph.proto
-
-package org.tensorflow.proto.framework;
-
-public interface SavedConcreteFunctionOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.SavedConcreteFunction)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <pre>
-   * Bound inputs to the function. The SavedObjects identified by the node ids
-   * given here are appended as extra inputs to the caller-supplied inputs.
-   * The only types of SavedObjects valid here are SavedVariable, SavedResource
-   * and SavedAsset.
-   * </pre>
-   *
-   * <code>repeated int32 bound_inputs = 2;</code>
-   */
-  java.util.List<java.lang.Integer> getBoundInputsList();
-  /**
-   * <pre>
-   * Bound inputs to the function. The SavedObjects identified by the node ids
-   * given here are appended as extra inputs to the caller-supplied inputs.
-   * The only types of SavedObjects valid here are SavedVariable, SavedResource
-   * and SavedAsset.
-   * </pre>
-   *
-   * <code>repeated int32 bound_inputs = 2;</code>
-   */
-  int getBoundInputsCount();
-  /**
-   * <pre>
-   * Bound inputs to the function. The SavedObjects identified by the node ids
-   * given here are appended as extra inputs to the caller-supplied inputs.
-   * The only types of SavedObjects valid here are SavedVariable, SavedResource
-   * and SavedAsset.
-   * </pre>
-   *
-   * <code>repeated int32 bound_inputs = 2;</code>
-   */
-  int getBoundInputs(int index);
-
-  /**
-   * <pre>
-   * Input in canonicalized form that was received to create this concrete
-   * function.
-   * </pre>
-   *
-   * <code>.tensorflow.StructuredValue canonicalized_input_signature = 3;</code>
-   */
-  boolean hasCanonicalizedInputSignature();
-  /**
-   * <pre>
-   * Input in canonicalized form that was received to create this concrete
-   * function.
-   * </pre>
-   *
-   * <code>.tensorflow.StructuredValue canonicalized_input_signature = 3;</code>
-   */
-  org.tensorflow.proto.framework.StructuredValue getCanonicalizedInputSignature();
-  /**
-   * <pre>
-   * Input in canonicalized form that was received to create this concrete
-   * function.
-   * </pre>
-   *
-   * <code>.tensorflow.StructuredValue canonicalized_input_signature = 3;</code>
-   */
-  org.tensorflow.proto.framework.StructuredValueOrBuilder getCanonicalizedInputSignatureOrBuilder();
-
-  /**
-   * <pre>
-   * Output that was the return value of this function after replacing all
-   * Tensors with TensorSpecs. This can be an arbitrary nested function and will
-   * be used to reconstruct the full structure from pure tensors.
-   * </pre>
-   *
-   * <code>.tensorflow.StructuredValue output_signature = 4;</code>
-   */
-  boolean hasOutputSignature();
-  /**
-   * <pre>
-   * Output that was the return value of this function after replacing all
-   * Tensors with TensorSpecs. This can be an arbitrary nested function and will
-   * be used to reconstruct the full structure from pure tensors.
-   * </pre>
-   *
-   * <code>.tensorflow.StructuredValue output_signature = 4;</code>
-   */
-  org.tensorflow.proto.framework.StructuredValue getOutputSignature();
-  /**
-   * <pre>
-   * Output that was the return value of this function after replacing all
-   * Tensors with TensorSpecs. This can be an arbitrary nested function and will
-   * be used to reconstruct the full structure from pure tensors.
-   * </pre>
-   *
-   * <code>.tensorflow.StructuredValue output_signature = 4;</code>
-   */
-  org.tensorflow.proto.framework.StructuredValueOrBuilder getOutputSignatureOrBuilder();
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedConstant.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedConstant.java
deleted file mode 100644
index 4ba709c9c98..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedConstant.java
+++ /dev/null
@@ -1,574 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/saved_object_graph.proto
-
-package org.tensorflow.proto.framework;
-
-/**
- * Protobuf type {@code tensorflow.SavedConstant}
- */
-public  final class SavedConstant extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.SavedConstant)
-    SavedConstantOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use SavedConstant.newBuilder() to construct.
-  private SavedConstant(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private SavedConstant() {
-    operation_ = "";
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new SavedConstant();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private SavedConstant(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 10: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            operation_ = s;
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.framework.SavedObjectGraphProtos.internal_static_tensorflow_SavedConstant_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.framework.SavedObjectGraphProtos.internal_static_tensorflow_SavedConstant_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.framework.SavedConstant.class, org.tensorflow.proto.framework.SavedConstant.Builder.class);
-  }
-
-  public static final int OPERATION_FIELD_NUMBER = 1;
-  private volatile java.lang.Object operation_;
-  /**
-   * <pre>
-   * An Operation name for a ConstantOp in this SavedObjectGraph's MetaGraph.
-   * </pre>
-   *
-   * <code>string operation = 1;</code>
-   */
-  public java.lang.String getOperation() {
-    java.lang.Object ref = operation_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      operation_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * An Operation name for a ConstantOp in this SavedObjectGraph's MetaGraph.
-   * </pre>
-   *
-   * <code>string operation = 1;</code>
-   */
-  public com.google.protobuf.ByteString
-      getOperationBytes() {
-    java.lang.Object ref = operation_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      operation_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    if (!getOperationBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 1, operation_);
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (!getOperationBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, operation_);
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.framework.SavedConstant)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.framework.SavedConstant other = (org.tensorflow.proto.framework.SavedConstant) obj;
-
-    if (!getOperation()
-        .equals(other.getOperation())) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    hash = (37 * hash) + OPERATION_FIELD_NUMBER;
-    hash = (53 * hash) + getOperation().hashCode();
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.framework.SavedConstant parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.SavedConstant parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.SavedConstant parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.SavedConstant parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.SavedConstant parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.SavedConstant parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.SavedConstant parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.SavedConstant parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.SavedConstant parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.SavedConstant parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.SavedConstant parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.SavedConstant parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.framework.SavedConstant prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * Protobuf type {@code tensorflow.SavedConstant}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.SavedConstant)
-      org.tensorflow.proto.framework.SavedConstantOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.SavedObjectGraphProtos.internal_static_tensorflow_SavedConstant_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.SavedObjectGraphProtos.internal_static_tensorflow_SavedConstant_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.SavedConstant.class, org.tensorflow.proto.framework.SavedConstant.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.framework.SavedConstant.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      operation_ = "";
-
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.framework.SavedObjectGraphProtos.internal_static_tensorflow_SavedConstant_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.SavedConstant getDefaultInstanceForType() {
-      return org.tensorflow.proto.framework.SavedConstant.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.SavedConstant build() {
-      org.tensorflow.proto.framework.SavedConstant result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.SavedConstant buildPartial() {
-      org.tensorflow.proto.framework.SavedConstant result = new org.tensorflow.proto.framework.SavedConstant(this);
-      result.operation_ = operation_;
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.framework.SavedConstant) {
-        return mergeFrom((org.tensorflow.proto.framework.SavedConstant)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.framework.SavedConstant other) {
-      if (other == org.tensorflow.proto.framework.SavedConstant.getDefaultInstance()) return this;
-      if (!other.getOperation().isEmpty()) {
-        operation_ = other.operation_;
-        onChanged();
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.framework.SavedConstant parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.framework.SavedConstant) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-
-    private java.lang.Object operation_ = "";
-    /**
-     * <pre>
-     * An Operation name for a ConstantOp in this SavedObjectGraph's MetaGraph.
-     * </pre>
-     *
-     * <code>string operation = 1;</code>
-     */
-    public java.lang.String getOperation() {
-      java.lang.Object ref = operation_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        operation_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * An Operation name for a ConstantOp in this SavedObjectGraph's MetaGraph.
-     * </pre>
-     *
-     * <code>string operation = 1;</code>
-     */
-    public com.google.protobuf.ByteString
-        getOperationBytes() {
-      java.lang.Object ref = operation_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        operation_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * An Operation name for a ConstantOp in this SavedObjectGraph's MetaGraph.
-     * </pre>
-     *
-     * <code>string operation = 1;</code>
-     */
-    public Builder setOperation(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      operation_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * An Operation name for a ConstantOp in this SavedObjectGraph's MetaGraph.
-     * </pre>
-     *
-     * <code>string operation = 1;</code>
-     */
-    public Builder clearOperation() {
-      
-      operation_ = getDefaultInstance().getOperation();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * An Operation name for a ConstantOp in this SavedObjectGraph's MetaGraph.
-     * </pre>
-     *
-     * <code>string operation = 1;</code>
-     */
-    public Builder setOperationBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      operation_ = value;
-      onChanged();
-      return this;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.SavedConstant)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.SavedConstant)
-  private static final org.tensorflow.proto.framework.SavedConstant DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.SavedConstant();
-  }
-
-  public static org.tensorflow.proto.framework.SavedConstant getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<SavedConstant>
-      PARSER = new com.google.protobuf.AbstractParser<SavedConstant>() {
-    @java.lang.Override
-    public SavedConstant parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new SavedConstant(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<SavedConstant> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<SavedConstant> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.framework.SavedConstant getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedConstantOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedConstantOrBuilder.java
deleted file mode 100644
index ed435bc83ba..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedConstantOrBuilder.java
+++ /dev/null
@@ -1,27 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/saved_object_graph.proto
-
-package org.tensorflow.proto.framework;
-
-public interface SavedConstantOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.SavedConstant)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <pre>
-   * An Operation name for a ConstantOp in this SavedObjectGraph's MetaGraph.
-   * </pre>
-   *
-   * <code>string operation = 1;</code>
-   */
-  java.lang.String getOperation();
-  /**
-   * <pre>
-   * An Operation name for a ConstantOp in this SavedObjectGraph's MetaGraph.
-   * </pre>
-   *
-   * <code>string operation = 1;</code>
-   */
-  com.google.protobuf.ByteString
-      getOperationBytes();
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedFunction.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedFunction.java
deleted file mode 100644
index d873063784e..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedFunction.java
+++ /dev/null
@@ -1,781 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/saved_object_graph.proto
-
-package org.tensorflow.proto.framework;
-
-/**
- * <pre>
- * A function with multiple signatures, possibly with non-Tensor arguments.
- * </pre>
- *
- * Protobuf type {@code tensorflow.SavedFunction}
- */
-public  final class SavedFunction extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.SavedFunction)
-    SavedFunctionOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use SavedFunction.newBuilder() to construct.
-  private SavedFunction(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private SavedFunction() {
-    concreteFunctions_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new SavedFunction();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private SavedFunction(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    int mutable_bitField0_ = 0;
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 10: {
-            java.lang.String s = input.readStringRequireUtf8();
-            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
-              concreteFunctions_ = new com.google.protobuf.LazyStringArrayList();
-              mutable_bitField0_ |= 0x00000001;
-            }
-            concreteFunctions_.add(s);
-            break;
-          }
-          case 18: {
-            org.tensorflow.proto.framework.FunctionSpec.Builder subBuilder = null;
-            if (functionSpec_ != null) {
-              subBuilder = functionSpec_.toBuilder();
-            }
-            functionSpec_ = input.readMessage(org.tensorflow.proto.framework.FunctionSpec.parser(), extensionRegistry);
-            if (subBuilder != null) {
-              subBuilder.mergeFrom(functionSpec_);
-              functionSpec_ = subBuilder.buildPartial();
-            }
-
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      if (((mutable_bitField0_ & 0x00000001) != 0)) {
-        concreteFunctions_ = concreteFunctions_.getUnmodifiableView();
-      }
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.framework.SavedObjectGraphProtos.internal_static_tensorflow_SavedFunction_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.framework.SavedObjectGraphProtos.internal_static_tensorflow_SavedFunction_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.framework.SavedFunction.class, org.tensorflow.proto.framework.SavedFunction.Builder.class);
-  }
-
-  public static final int CONCRETE_FUNCTIONS_FIELD_NUMBER = 1;
-  private com.google.protobuf.LazyStringList concreteFunctions_;
-  /**
-   * <code>repeated string concrete_functions = 1;</code>
-   */
-  public com.google.protobuf.ProtocolStringList
-      getConcreteFunctionsList() {
-    return concreteFunctions_;
-  }
-  /**
-   * <code>repeated string concrete_functions = 1;</code>
-   */
-  public int getConcreteFunctionsCount() {
-    return concreteFunctions_.size();
-  }
-  /**
-   * <code>repeated string concrete_functions = 1;</code>
-   */
-  public java.lang.String getConcreteFunctions(int index) {
-    return concreteFunctions_.get(index);
-  }
-  /**
-   * <code>repeated string concrete_functions = 1;</code>
-   */
-  public com.google.protobuf.ByteString
-      getConcreteFunctionsBytes(int index) {
-    return concreteFunctions_.getByteString(index);
-  }
-
-  public static final int FUNCTION_SPEC_FIELD_NUMBER = 2;
-  private org.tensorflow.proto.framework.FunctionSpec functionSpec_;
-  /**
-   * <code>.tensorflow.FunctionSpec function_spec = 2;</code>
-   */
-  public boolean hasFunctionSpec() {
-    return functionSpec_ != null;
-  }
-  /**
-   * <code>.tensorflow.FunctionSpec function_spec = 2;</code>
-   */
-  public org.tensorflow.proto.framework.FunctionSpec getFunctionSpec() {
-    return functionSpec_ == null ? org.tensorflow.proto.framework.FunctionSpec.getDefaultInstance() : functionSpec_;
-  }
-  /**
-   * <code>.tensorflow.FunctionSpec function_spec = 2;</code>
-   */
-  public org.tensorflow.proto.framework.FunctionSpecOrBuilder getFunctionSpecOrBuilder() {
-    return getFunctionSpec();
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    for (int i = 0; i < concreteFunctions_.size(); i++) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 1, concreteFunctions_.getRaw(i));
-    }
-    if (functionSpec_ != null) {
-      output.writeMessage(2, getFunctionSpec());
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    {
-      int dataSize = 0;
-      for (int i = 0; i < concreteFunctions_.size(); i++) {
-        dataSize += computeStringSizeNoTag(concreteFunctions_.getRaw(i));
-      }
-      size += dataSize;
-      size += 1 * getConcreteFunctionsList().size();
-    }
-    if (functionSpec_ != null) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(2, getFunctionSpec());
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.framework.SavedFunction)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.framework.SavedFunction other = (org.tensorflow.proto.framework.SavedFunction) obj;
-
-    if (!getConcreteFunctionsList()
-        .equals(other.getConcreteFunctionsList())) return false;
-    if (hasFunctionSpec() != other.hasFunctionSpec()) return false;
-    if (hasFunctionSpec()) {
-      if (!getFunctionSpec()
-          .equals(other.getFunctionSpec())) return false;
-    }
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    if (getConcreteFunctionsCount() > 0) {
-      hash = (37 * hash) + CONCRETE_FUNCTIONS_FIELD_NUMBER;
-      hash = (53 * hash) + getConcreteFunctionsList().hashCode();
-    }
-    if (hasFunctionSpec()) {
-      hash = (37 * hash) + FUNCTION_SPEC_FIELD_NUMBER;
-      hash = (53 * hash) + getFunctionSpec().hashCode();
-    }
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.framework.SavedFunction parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.SavedFunction parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.SavedFunction parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.SavedFunction parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.SavedFunction parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.SavedFunction parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.SavedFunction parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.SavedFunction parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.SavedFunction parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.SavedFunction parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.SavedFunction parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.SavedFunction parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.framework.SavedFunction prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * <pre>
-   * A function with multiple signatures, possibly with non-Tensor arguments.
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.SavedFunction}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.SavedFunction)
-      org.tensorflow.proto.framework.SavedFunctionOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.SavedObjectGraphProtos.internal_static_tensorflow_SavedFunction_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.SavedObjectGraphProtos.internal_static_tensorflow_SavedFunction_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.SavedFunction.class, org.tensorflow.proto.framework.SavedFunction.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.framework.SavedFunction.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      concreteFunctions_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-      bitField0_ = (bitField0_ & ~0x00000001);
-      if (functionSpecBuilder_ == null) {
-        functionSpec_ = null;
-      } else {
-        functionSpec_ = null;
-        functionSpecBuilder_ = null;
-      }
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.framework.SavedObjectGraphProtos.internal_static_tensorflow_SavedFunction_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.SavedFunction getDefaultInstanceForType() {
-      return org.tensorflow.proto.framework.SavedFunction.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.SavedFunction build() {
-      org.tensorflow.proto.framework.SavedFunction result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.SavedFunction buildPartial() {
-      org.tensorflow.proto.framework.SavedFunction result = new org.tensorflow.proto.framework.SavedFunction(this);
-      int from_bitField0_ = bitField0_;
-      if (((bitField0_ & 0x00000001) != 0)) {
-        concreteFunctions_ = concreteFunctions_.getUnmodifiableView();
-        bitField0_ = (bitField0_ & ~0x00000001);
-      }
-      result.concreteFunctions_ = concreteFunctions_;
-      if (functionSpecBuilder_ == null) {
-        result.functionSpec_ = functionSpec_;
-      } else {
-        result.functionSpec_ = functionSpecBuilder_.build();
-      }
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.framework.SavedFunction) {
-        return mergeFrom((org.tensorflow.proto.framework.SavedFunction)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.framework.SavedFunction other) {
-      if (other == org.tensorflow.proto.framework.SavedFunction.getDefaultInstance()) return this;
-      if (!other.concreteFunctions_.isEmpty()) {
-        if (concreteFunctions_.isEmpty()) {
-          concreteFunctions_ = other.concreteFunctions_;
-          bitField0_ = (bitField0_ & ~0x00000001);
-        } else {
-          ensureConcreteFunctionsIsMutable();
-          concreteFunctions_.addAll(other.concreteFunctions_);
-        }
-        onChanged();
-      }
-      if (other.hasFunctionSpec()) {
-        mergeFunctionSpec(other.getFunctionSpec());
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.framework.SavedFunction parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.framework.SavedFunction) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-    private int bitField0_;
-
-    private com.google.protobuf.LazyStringList concreteFunctions_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-    private void ensureConcreteFunctionsIsMutable() {
-      if (!((bitField0_ & 0x00000001) != 0)) {
-        concreteFunctions_ = new com.google.protobuf.LazyStringArrayList(concreteFunctions_);
-        bitField0_ |= 0x00000001;
-       }
-    }
-    /**
-     * <code>repeated string concrete_functions = 1;</code>
-     */
-    public com.google.protobuf.ProtocolStringList
-        getConcreteFunctionsList() {
-      return concreteFunctions_.getUnmodifiableView();
-    }
-    /**
-     * <code>repeated string concrete_functions = 1;</code>
-     */
-    public int getConcreteFunctionsCount() {
-      return concreteFunctions_.size();
-    }
-    /**
-     * <code>repeated string concrete_functions = 1;</code>
-     */
-    public java.lang.String getConcreteFunctions(int index) {
-      return concreteFunctions_.get(index);
-    }
-    /**
-     * <code>repeated string concrete_functions = 1;</code>
-     */
-    public com.google.protobuf.ByteString
-        getConcreteFunctionsBytes(int index) {
-      return concreteFunctions_.getByteString(index);
-    }
-    /**
-     * <code>repeated string concrete_functions = 1;</code>
-     */
-    public Builder setConcreteFunctions(
-        int index, java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  ensureConcreteFunctionsIsMutable();
-      concreteFunctions_.set(index, value);
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>repeated string concrete_functions = 1;</code>
-     */
-    public Builder addConcreteFunctions(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  ensureConcreteFunctionsIsMutable();
-      concreteFunctions_.add(value);
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>repeated string concrete_functions = 1;</code>
-     */
-    public Builder addAllConcreteFunctions(
-        java.lang.Iterable<java.lang.String> values) {
-      ensureConcreteFunctionsIsMutable();
-      com.google.protobuf.AbstractMessageLite.Builder.addAll(
-          values, concreteFunctions_);
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>repeated string concrete_functions = 1;</code>
-     */
-    public Builder clearConcreteFunctions() {
-      concreteFunctions_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-      bitField0_ = (bitField0_ & ~0x00000001);
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>repeated string concrete_functions = 1;</code>
-     */
-    public Builder addConcreteFunctionsBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      ensureConcreteFunctionsIsMutable();
-      concreteFunctions_.add(value);
-      onChanged();
-      return this;
-    }
-
-    private org.tensorflow.proto.framework.FunctionSpec functionSpec_;
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.FunctionSpec, org.tensorflow.proto.framework.FunctionSpec.Builder, org.tensorflow.proto.framework.FunctionSpecOrBuilder> functionSpecBuilder_;
-    /**
-     * <code>.tensorflow.FunctionSpec function_spec = 2;</code>
-     */
-    public boolean hasFunctionSpec() {
-      return functionSpecBuilder_ != null || functionSpec_ != null;
-    }
-    /**
-     * <code>.tensorflow.FunctionSpec function_spec = 2;</code>
-     */
-    public org.tensorflow.proto.framework.FunctionSpec getFunctionSpec() {
-      if (functionSpecBuilder_ == null) {
-        return functionSpec_ == null ? org.tensorflow.proto.framework.FunctionSpec.getDefaultInstance() : functionSpec_;
-      } else {
-        return functionSpecBuilder_.getMessage();
-      }
-    }
-    /**
-     * <code>.tensorflow.FunctionSpec function_spec = 2;</code>
-     */
-    public Builder setFunctionSpec(org.tensorflow.proto.framework.FunctionSpec value) {
-      if (functionSpecBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        functionSpec_ = value;
-        onChanged();
-      } else {
-        functionSpecBuilder_.setMessage(value);
-      }
-
-      return this;
-    }
-    /**
-     * <code>.tensorflow.FunctionSpec function_spec = 2;</code>
-     */
-    public Builder setFunctionSpec(
-        org.tensorflow.proto.framework.FunctionSpec.Builder builderForValue) {
-      if (functionSpecBuilder_ == null) {
-        functionSpec_ = builderForValue.build();
-        onChanged();
-      } else {
-        functionSpecBuilder_.setMessage(builderForValue.build());
-      }
-
-      return this;
-    }
-    /**
-     * <code>.tensorflow.FunctionSpec function_spec = 2;</code>
-     */
-    public Builder mergeFunctionSpec(org.tensorflow.proto.framework.FunctionSpec value) {
-      if (functionSpecBuilder_ == null) {
-        if (functionSpec_ != null) {
-          functionSpec_ =
-            org.tensorflow.proto.framework.FunctionSpec.newBuilder(functionSpec_).mergeFrom(value).buildPartial();
-        } else {
-          functionSpec_ = value;
-        }
-        onChanged();
-      } else {
-        functionSpecBuilder_.mergeFrom(value);
-      }
-
-      return this;
-    }
-    /**
-     * <code>.tensorflow.FunctionSpec function_spec = 2;</code>
-     */
-    public Builder clearFunctionSpec() {
-      if (functionSpecBuilder_ == null) {
-        functionSpec_ = null;
-        onChanged();
-      } else {
-        functionSpec_ = null;
-        functionSpecBuilder_ = null;
-      }
-
-      return this;
-    }
-    /**
-     * <code>.tensorflow.FunctionSpec function_spec = 2;</code>
-     */
-    public org.tensorflow.proto.framework.FunctionSpec.Builder getFunctionSpecBuilder() {
-      
-      onChanged();
-      return getFunctionSpecFieldBuilder().getBuilder();
-    }
-    /**
-     * <code>.tensorflow.FunctionSpec function_spec = 2;</code>
-     */
-    public org.tensorflow.proto.framework.FunctionSpecOrBuilder getFunctionSpecOrBuilder() {
-      if (functionSpecBuilder_ != null) {
-        return functionSpecBuilder_.getMessageOrBuilder();
-      } else {
-        return functionSpec_ == null ?
-            org.tensorflow.proto.framework.FunctionSpec.getDefaultInstance() : functionSpec_;
-      }
-    }
-    /**
-     * <code>.tensorflow.FunctionSpec function_spec = 2;</code>
-     */
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.FunctionSpec, org.tensorflow.proto.framework.FunctionSpec.Builder, org.tensorflow.proto.framework.FunctionSpecOrBuilder> 
-        getFunctionSpecFieldBuilder() {
-      if (functionSpecBuilder_ == null) {
-        functionSpecBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-            org.tensorflow.proto.framework.FunctionSpec, org.tensorflow.proto.framework.FunctionSpec.Builder, org.tensorflow.proto.framework.FunctionSpecOrBuilder>(
-                getFunctionSpec(),
-                getParentForChildren(),
-                isClean());
-        functionSpec_ = null;
-      }
-      return functionSpecBuilder_;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.SavedFunction)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.SavedFunction)
-  private static final org.tensorflow.proto.framework.SavedFunction DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.SavedFunction();
-  }
-
-  public static org.tensorflow.proto.framework.SavedFunction getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<SavedFunction>
-      PARSER = new com.google.protobuf.AbstractParser<SavedFunction>() {
-    @java.lang.Override
-    public SavedFunction parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new SavedFunction(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<SavedFunction> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<SavedFunction> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.framework.SavedFunction getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedFunctionOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedFunctionOrBuilder.java
deleted file mode 100644
index 19b27a54fb3..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedFunctionOrBuilder.java
+++ /dev/null
@@ -1,41 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/saved_object_graph.proto
-
-package org.tensorflow.proto.framework;
-
-public interface SavedFunctionOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.SavedFunction)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <code>repeated string concrete_functions = 1;</code>
-   */
-  java.util.List<java.lang.String>
-      getConcreteFunctionsList();
-  /**
-   * <code>repeated string concrete_functions = 1;</code>
-   */
-  int getConcreteFunctionsCount();
-  /**
-   * <code>repeated string concrete_functions = 1;</code>
-   */
-  java.lang.String getConcreteFunctions(int index);
-  /**
-   * <code>repeated string concrete_functions = 1;</code>
-   */
-  com.google.protobuf.ByteString
-      getConcreteFunctionsBytes(int index);
-
-  /**
-   * <code>.tensorflow.FunctionSpec function_spec = 2;</code>
-   */
-  boolean hasFunctionSpec();
-  /**
-   * <code>.tensorflow.FunctionSpec function_spec = 2;</code>
-   */
-  org.tensorflow.proto.framework.FunctionSpec getFunctionSpec();
-  /**
-   * <code>.tensorflow.FunctionSpec function_spec = 2;</code>
-   */
-  org.tensorflow.proto.framework.FunctionSpecOrBuilder getFunctionSpecOrBuilder();
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedModel.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedModel.java
deleted file mode 100644
index ac25416e31a..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedModel.java
+++ /dev/null
@@ -1,949 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/saved_model.proto
-
-package org.tensorflow.proto.framework;
-
-/**
- * <pre>
- * SavedModel is the high level serialization format for TensorFlow Models.
- * See [todo: doc links, similar to session_bundle] for more information.
- * </pre>
- *
- * Protobuf type {@code tensorflow.SavedModel}
- */
-public  final class SavedModel extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.SavedModel)
-    SavedModelOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use SavedModel.newBuilder() to construct.
-  private SavedModel(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private SavedModel() {
-    metaGraphs_ = java.util.Collections.emptyList();
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new SavedModel();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private SavedModel(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    int mutable_bitField0_ = 0;
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 8: {
-
-            savedModelSchemaVersion_ = input.readInt64();
-            break;
-          }
-          case 18: {
-            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
-              metaGraphs_ = new java.util.ArrayList<org.tensorflow.proto.framework.MetaGraphDef>();
-              mutable_bitField0_ |= 0x00000001;
-            }
-            metaGraphs_.add(
-                input.readMessage(org.tensorflow.proto.framework.MetaGraphDef.parser(), extensionRegistry));
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      if (((mutable_bitField0_ & 0x00000001) != 0)) {
-        metaGraphs_ = java.util.Collections.unmodifiableList(metaGraphs_);
-      }
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.framework.SavedModelProtos.internal_static_tensorflow_SavedModel_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.framework.SavedModelProtos.internal_static_tensorflow_SavedModel_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.framework.SavedModel.class, org.tensorflow.proto.framework.SavedModel.Builder.class);
-  }
-
-  public static final int SAVED_MODEL_SCHEMA_VERSION_FIELD_NUMBER = 1;
-  private long savedModelSchemaVersion_;
-  /**
-   * <pre>
-   * The schema version of the SavedModel instance. Used for versioning when
-   * making future changes to the specification/implementation. Initial value
-   * at release will be 1.
-   * </pre>
-   *
-   * <code>int64 saved_model_schema_version = 1;</code>
-   */
-  public long getSavedModelSchemaVersion() {
-    return savedModelSchemaVersion_;
-  }
-
-  public static final int META_GRAPHS_FIELD_NUMBER = 2;
-  private java.util.List<org.tensorflow.proto.framework.MetaGraphDef> metaGraphs_;
-  /**
-   * <pre>
-   * One or more MetaGraphs.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.MetaGraphDef meta_graphs = 2;</code>
-   */
-  public java.util.List<org.tensorflow.proto.framework.MetaGraphDef> getMetaGraphsList() {
-    return metaGraphs_;
-  }
-  /**
-   * <pre>
-   * One or more MetaGraphs.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.MetaGraphDef meta_graphs = 2;</code>
-   */
-  public java.util.List<? extends org.tensorflow.proto.framework.MetaGraphDefOrBuilder> 
-      getMetaGraphsOrBuilderList() {
-    return metaGraphs_;
-  }
-  /**
-   * <pre>
-   * One or more MetaGraphs.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.MetaGraphDef meta_graphs = 2;</code>
-   */
-  public int getMetaGraphsCount() {
-    return metaGraphs_.size();
-  }
-  /**
-   * <pre>
-   * One or more MetaGraphs.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.MetaGraphDef meta_graphs = 2;</code>
-   */
-  public org.tensorflow.proto.framework.MetaGraphDef getMetaGraphs(int index) {
-    return metaGraphs_.get(index);
-  }
-  /**
-   * <pre>
-   * One or more MetaGraphs.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.MetaGraphDef meta_graphs = 2;</code>
-   */
-  public org.tensorflow.proto.framework.MetaGraphDefOrBuilder getMetaGraphsOrBuilder(
-      int index) {
-    return metaGraphs_.get(index);
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    if (savedModelSchemaVersion_ != 0L) {
-      output.writeInt64(1, savedModelSchemaVersion_);
-    }
-    for (int i = 0; i < metaGraphs_.size(); i++) {
-      output.writeMessage(2, metaGraphs_.get(i));
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (savedModelSchemaVersion_ != 0L) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt64Size(1, savedModelSchemaVersion_);
-    }
-    for (int i = 0; i < metaGraphs_.size(); i++) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(2, metaGraphs_.get(i));
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.framework.SavedModel)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.framework.SavedModel other = (org.tensorflow.proto.framework.SavedModel) obj;
-
-    if (getSavedModelSchemaVersion()
-        != other.getSavedModelSchemaVersion()) return false;
-    if (!getMetaGraphsList()
-        .equals(other.getMetaGraphsList())) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    hash = (37 * hash) + SAVED_MODEL_SCHEMA_VERSION_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-        getSavedModelSchemaVersion());
-    if (getMetaGraphsCount() > 0) {
-      hash = (37 * hash) + META_GRAPHS_FIELD_NUMBER;
-      hash = (53 * hash) + getMetaGraphsList().hashCode();
-    }
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.framework.SavedModel parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.SavedModel parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.SavedModel parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.SavedModel parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.SavedModel parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.SavedModel parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.SavedModel parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.SavedModel parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.SavedModel parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.SavedModel parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.SavedModel parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.SavedModel parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.framework.SavedModel prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * <pre>
-   * SavedModel is the high level serialization format for TensorFlow Models.
-   * See [todo: doc links, similar to session_bundle] for more information.
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.SavedModel}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.SavedModel)
-      org.tensorflow.proto.framework.SavedModelOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.SavedModelProtos.internal_static_tensorflow_SavedModel_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.SavedModelProtos.internal_static_tensorflow_SavedModel_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.SavedModel.class, org.tensorflow.proto.framework.SavedModel.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.framework.SavedModel.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-        getMetaGraphsFieldBuilder();
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      savedModelSchemaVersion_ = 0L;
-
-      if (metaGraphsBuilder_ == null) {
-        metaGraphs_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000001);
-      } else {
-        metaGraphsBuilder_.clear();
-      }
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.framework.SavedModelProtos.internal_static_tensorflow_SavedModel_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.SavedModel getDefaultInstanceForType() {
-      return org.tensorflow.proto.framework.SavedModel.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.SavedModel build() {
-      org.tensorflow.proto.framework.SavedModel result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.SavedModel buildPartial() {
-      org.tensorflow.proto.framework.SavedModel result = new org.tensorflow.proto.framework.SavedModel(this);
-      int from_bitField0_ = bitField0_;
-      result.savedModelSchemaVersion_ = savedModelSchemaVersion_;
-      if (metaGraphsBuilder_ == null) {
-        if (((bitField0_ & 0x00000001) != 0)) {
-          metaGraphs_ = java.util.Collections.unmodifiableList(metaGraphs_);
-          bitField0_ = (bitField0_ & ~0x00000001);
-        }
-        result.metaGraphs_ = metaGraphs_;
-      } else {
-        result.metaGraphs_ = metaGraphsBuilder_.build();
-      }
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.framework.SavedModel) {
-        return mergeFrom((org.tensorflow.proto.framework.SavedModel)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.framework.SavedModel other) {
-      if (other == org.tensorflow.proto.framework.SavedModel.getDefaultInstance()) return this;
-      if (other.getSavedModelSchemaVersion() != 0L) {
-        setSavedModelSchemaVersion(other.getSavedModelSchemaVersion());
-      }
-      if (metaGraphsBuilder_ == null) {
-        if (!other.metaGraphs_.isEmpty()) {
-          if (metaGraphs_.isEmpty()) {
-            metaGraphs_ = other.metaGraphs_;
-            bitField0_ = (bitField0_ & ~0x00000001);
-          } else {
-            ensureMetaGraphsIsMutable();
-            metaGraphs_.addAll(other.metaGraphs_);
-          }
-          onChanged();
-        }
-      } else {
-        if (!other.metaGraphs_.isEmpty()) {
-          if (metaGraphsBuilder_.isEmpty()) {
-            metaGraphsBuilder_.dispose();
-            metaGraphsBuilder_ = null;
-            metaGraphs_ = other.metaGraphs_;
-            bitField0_ = (bitField0_ & ~0x00000001);
-            metaGraphsBuilder_ = 
-              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
-                 getMetaGraphsFieldBuilder() : null;
-          } else {
-            metaGraphsBuilder_.addAllMessages(other.metaGraphs_);
-          }
-        }
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.framework.SavedModel parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.framework.SavedModel) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-    private int bitField0_;
-
-    private long savedModelSchemaVersion_ ;
-    /**
-     * <pre>
-     * The schema version of the SavedModel instance. Used for versioning when
-     * making future changes to the specification/implementation. Initial value
-     * at release will be 1.
-     * </pre>
-     *
-     * <code>int64 saved_model_schema_version = 1;</code>
-     */
-    public long getSavedModelSchemaVersion() {
-      return savedModelSchemaVersion_;
-    }
-    /**
-     * <pre>
-     * The schema version of the SavedModel instance. Used for versioning when
-     * making future changes to the specification/implementation. Initial value
-     * at release will be 1.
-     * </pre>
-     *
-     * <code>int64 saved_model_schema_version = 1;</code>
-     */
-    public Builder setSavedModelSchemaVersion(long value) {
-      
-      savedModelSchemaVersion_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * The schema version of the SavedModel instance. Used for versioning when
-     * making future changes to the specification/implementation. Initial value
-     * at release will be 1.
-     * </pre>
-     *
-     * <code>int64 saved_model_schema_version = 1;</code>
-     */
-    public Builder clearSavedModelSchemaVersion() {
-      
-      savedModelSchemaVersion_ = 0L;
-      onChanged();
-      return this;
-    }
-
-    private java.util.List<org.tensorflow.proto.framework.MetaGraphDef> metaGraphs_ =
-      java.util.Collections.emptyList();
-    private void ensureMetaGraphsIsMutable() {
-      if (!((bitField0_ & 0x00000001) != 0)) {
-        metaGraphs_ = new java.util.ArrayList<org.tensorflow.proto.framework.MetaGraphDef>(metaGraphs_);
-        bitField0_ |= 0x00000001;
-       }
-    }
-
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.framework.MetaGraphDef, org.tensorflow.proto.framework.MetaGraphDef.Builder, org.tensorflow.proto.framework.MetaGraphDefOrBuilder> metaGraphsBuilder_;
-
-    /**
-     * <pre>
-     * One or more MetaGraphs.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.MetaGraphDef meta_graphs = 2;</code>
-     */
-    public java.util.List<org.tensorflow.proto.framework.MetaGraphDef> getMetaGraphsList() {
-      if (metaGraphsBuilder_ == null) {
-        return java.util.Collections.unmodifiableList(metaGraphs_);
-      } else {
-        return metaGraphsBuilder_.getMessageList();
-      }
-    }
-    /**
-     * <pre>
-     * One or more MetaGraphs.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.MetaGraphDef meta_graphs = 2;</code>
-     */
-    public int getMetaGraphsCount() {
-      if (metaGraphsBuilder_ == null) {
-        return metaGraphs_.size();
-      } else {
-        return metaGraphsBuilder_.getCount();
-      }
-    }
-    /**
-     * <pre>
-     * One or more MetaGraphs.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.MetaGraphDef meta_graphs = 2;</code>
-     */
-    public org.tensorflow.proto.framework.MetaGraphDef getMetaGraphs(int index) {
-      if (metaGraphsBuilder_ == null) {
-        return metaGraphs_.get(index);
-      } else {
-        return metaGraphsBuilder_.getMessage(index);
-      }
-    }
-    /**
-     * <pre>
-     * One or more MetaGraphs.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.MetaGraphDef meta_graphs = 2;</code>
-     */
-    public Builder setMetaGraphs(
-        int index, org.tensorflow.proto.framework.MetaGraphDef value) {
-      if (metaGraphsBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureMetaGraphsIsMutable();
-        metaGraphs_.set(index, value);
-        onChanged();
-      } else {
-        metaGraphsBuilder_.setMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * One or more MetaGraphs.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.MetaGraphDef meta_graphs = 2;</code>
-     */
-    public Builder setMetaGraphs(
-        int index, org.tensorflow.proto.framework.MetaGraphDef.Builder builderForValue) {
-      if (metaGraphsBuilder_ == null) {
-        ensureMetaGraphsIsMutable();
-        metaGraphs_.set(index, builderForValue.build());
-        onChanged();
-      } else {
-        metaGraphsBuilder_.setMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * One or more MetaGraphs.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.MetaGraphDef meta_graphs = 2;</code>
-     */
-    public Builder addMetaGraphs(org.tensorflow.proto.framework.MetaGraphDef value) {
-      if (metaGraphsBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureMetaGraphsIsMutable();
-        metaGraphs_.add(value);
-        onChanged();
-      } else {
-        metaGraphsBuilder_.addMessage(value);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * One or more MetaGraphs.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.MetaGraphDef meta_graphs = 2;</code>
-     */
-    public Builder addMetaGraphs(
-        int index, org.tensorflow.proto.framework.MetaGraphDef value) {
-      if (metaGraphsBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureMetaGraphsIsMutable();
-        metaGraphs_.add(index, value);
-        onChanged();
-      } else {
-        metaGraphsBuilder_.addMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * One or more MetaGraphs.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.MetaGraphDef meta_graphs = 2;</code>
-     */
-    public Builder addMetaGraphs(
-        org.tensorflow.proto.framework.MetaGraphDef.Builder builderForValue) {
-      if (metaGraphsBuilder_ == null) {
-        ensureMetaGraphsIsMutable();
-        metaGraphs_.add(builderForValue.build());
-        onChanged();
-      } else {
-        metaGraphsBuilder_.addMessage(builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * One or more MetaGraphs.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.MetaGraphDef meta_graphs = 2;</code>
-     */
-    public Builder addMetaGraphs(
-        int index, org.tensorflow.proto.framework.MetaGraphDef.Builder builderForValue) {
-      if (metaGraphsBuilder_ == null) {
-        ensureMetaGraphsIsMutable();
-        metaGraphs_.add(index, builderForValue.build());
-        onChanged();
-      } else {
-        metaGraphsBuilder_.addMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * One or more MetaGraphs.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.MetaGraphDef meta_graphs = 2;</code>
-     */
-    public Builder addAllMetaGraphs(
-        java.lang.Iterable<? extends org.tensorflow.proto.framework.MetaGraphDef> values) {
-      if (metaGraphsBuilder_ == null) {
-        ensureMetaGraphsIsMutable();
-        com.google.protobuf.AbstractMessageLite.Builder.addAll(
-            values, metaGraphs_);
-        onChanged();
-      } else {
-        metaGraphsBuilder_.addAllMessages(values);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * One or more MetaGraphs.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.MetaGraphDef meta_graphs = 2;</code>
-     */
-    public Builder clearMetaGraphs() {
-      if (metaGraphsBuilder_ == null) {
-        metaGraphs_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000001);
-        onChanged();
-      } else {
-        metaGraphsBuilder_.clear();
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * One or more MetaGraphs.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.MetaGraphDef meta_graphs = 2;</code>
-     */
-    public Builder removeMetaGraphs(int index) {
-      if (metaGraphsBuilder_ == null) {
-        ensureMetaGraphsIsMutable();
-        metaGraphs_.remove(index);
-        onChanged();
-      } else {
-        metaGraphsBuilder_.remove(index);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * One or more MetaGraphs.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.MetaGraphDef meta_graphs = 2;</code>
-     */
-    public org.tensorflow.proto.framework.MetaGraphDef.Builder getMetaGraphsBuilder(
-        int index) {
-      return getMetaGraphsFieldBuilder().getBuilder(index);
-    }
-    /**
-     * <pre>
-     * One or more MetaGraphs.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.MetaGraphDef meta_graphs = 2;</code>
-     */
-    public org.tensorflow.proto.framework.MetaGraphDefOrBuilder getMetaGraphsOrBuilder(
-        int index) {
-      if (metaGraphsBuilder_ == null) {
-        return metaGraphs_.get(index);  } else {
-        return metaGraphsBuilder_.getMessageOrBuilder(index);
-      }
-    }
-    /**
-     * <pre>
-     * One or more MetaGraphs.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.MetaGraphDef meta_graphs = 2;</code>
-     */
-    public java.util.List<? extends org.tensorflow.proto.framework.MetaGraphDefOrBuilder> 
-         getMetaGraphsOrBuilderList() {
-      if (metaGraphsBuilder_ != null) {
-        return metaGraphsBuilder_.getMessageOrBuilderList();
-      } else {
-        return java.util.Collections.unmodifiableList(metaGraphs_);
-      }
-    }
-    /**
-     * <pre>
-     * One or more MetaGraphs.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.MetaGraphDef meta_graphs = 2;</code>
-     */
-    public org.tensorflow.proto.framework.MetaGraphDef.Builder addMetaGraphsBuilder() {
-      return getMetaGraphsFieldBuilder().addBuilder(
-          org.tensorflow.proto.framework.MetaGraphDef.getDefaultInstance());
-    }
-    /**
-     * <pre>
-     * One or more MetaGraphs.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.MetaGraphDef meta_graphs = 2;</code>
-     */
-    public org.tensorflow.proto.framework.MetaGraphDef.Builder addMetaGraphsBuilder(
-        int index) {
-      return getMetaGraphsFieldBuilder().addBuilder(
-          index, org.tensorflow.proto.framework.MetaGraphDef.getDefaultInstance());
-    }
-    /**
-     * <pre>
-     * One or more MetaGraphs.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.MetaGraphDef meta_graphs = 2;</code>
-     */
-    public java.util.List<org.tensorflow.proto.framework.MetaGraphDef.Builder> 
-         getMetaGraphsBuilderList() {
-      return getMetaGraphsFieldBuilder().getBuilderList();
-    }
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.framework.MetaGraphDef, org.tensorflow.proto.framework.MetaGraphDef.Builder, org.tensorflow.proto.framework.MetaGraphDefOrBuilder> 
-        getMetaGraphsFieldBuilder() {
-      if (metaGraphsBuilder_ == null) {
-        metaGraphsBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
-            org.tensorflow.proto.framework.MetaGraphDef, org.tensorflow.proto.framework.MetaGraphDef.Builder, org.tensorflow.proto.framework.MetaGraphDefOrBuilder>(
-                metaGraphs_,
-                ((bitField0_ & 0x00000001) != 0),
-                getParentForChildren(),
-                isClean());
-        metaGraphs_ = null;
-      }
-      return metaGraphsBuilder_;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.SavedModel)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.SavedModel)
-  private static final org.tensorflow.proto.framework.SavedModel DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.SavedModel();
-  }
-
-  public static org.tensorflow.proto.framework.SavedModel getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<SavedModel>
-      PARSER = new com.google.protobuf.AbstractParser<SavedModel>() {
-    @java.lang.Override
-    public SavedModel parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new SavedModel(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<SavedModel> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<SavedModel> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.framework.SavedModel getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedModelOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedModelOrBuilder.java
deleted file mode 100644
index 9714aee34dd..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedModelOrBuilder.java
+++ /dev/null
@@ -1,64 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/saved_model.proto
-
-package org.tensorflow.proto.framework;
-
-public interface SavedModelOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.SavedModel)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <pre>
-   * The schema version of the SavedModel instance. Used for versioning when
-   * making future changes to the specification/implementation. Initial value
-   * at release will be 1.
-   * </pre>
-   *
-   * <code>int64 saved_model_schema_version = 1;</code>
-   */
-  long getSavedModelSchemaVersion();
-
-  /**
-   * <pre>
-   * One or more MetaGraphs.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.MetaGraphDef meta_graphs = 2;</code>
-   */
-  java.util.List<org.tensorflow.proto.framework.MetaGraphDef> 
-      getMetaGraphsList();
-  /**
-   * <pre>
-   * One or more MetaGraphs.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.MetaGraphDef meta_graphs = 2;</code>
-   */
-  org.tensorflow.proto.framework.MetaGraphDef getMetaGraphs(int index);
-  /**
-   * <pre>
-   * One or more MetaGraphs.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.MetaGraphDef meta_graphs = 2;</code>
-   */
-  int getMetaGraphsCount();
-  /**
-   * <pre>
-   * One or more MetaGraphs.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.MetaGraphDef meta_graphs = 2;</code>
-   */
-  java.util.List<? extends org.tensorflow.proto.framework.MetaGraphDefOrBuilder> 
-      getMetaGraphsOrBuilderList();
-  /**
-   * <pre>
-   * One or more MetaGraphs.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.MetaGraphDef meta_graphs = 2;</code>
-   */
-  org.tensorflow.proto.framework.MetaGraphDefOrBuilder getMetaGraphsOrBuilder(
-      int index);
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedModelProtos.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedModelProtos.java
deleted file mode 100644
index 05b4edd36f5..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedModelProtos.java
+++ /dev/null
@@ -1,56 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/saved_model.proto
-
-package org.tensorflow.proto.framework;
-
-public final class SavedModelProtos {
-  private SavedModelProtos() {}
-  public static void registerAllExtensions(
-      com.google.protobuf.ExtensionRegistryLite registry) {
-  }
-
-  public static void registerAllExtensions(
-      com.google.protobuf.ExtensionRegistry registry) {
-    registerAllExtensions(
-        (com.google.protobuf.ExtensionRegistryLite) registry);
-  }
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_SavedModel_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_SavedModel_fieldAccessorTable;
-
-  public static com.google.protobuf.Descriptors.FileDescriptor
-      getDescriptor() {
-    return descriptor;
-  }
-  private static  com.google.protobuf.Descriptors.FileDescriptor
-      descriptor;
-  static {
-    java.lang.String[] descriptorData = {
-      "\n*tensorflow/core/protobuf/saved_model.p" +
-      "roto\022\ntensorflow\032)tensorflow/core/protob" +
-      "uf/meta_graph.proto\"_\n\nSavedModel\022\"\n\032sav" +
-      "ed_model_schema_version\030\001 \001(\003\022-\n\013meta_gr" +
-      "aphs\030\002 \003(\0132\030.tensorflow.MetaGraphDefB\201\001\n" +
-      "\036org.tensorflow.proto.frameworkB\020SavedMo" +
-      "delProtosP\001ZHgithub.com/tensorflow/tenso" +
-      "rflow/tensorflow/go/core/core_protos_go_" +
-      "proto\370\001\001b\006proto3"
-    };
-    descriptor = com.google.protobuf.Descriptors.FileDescriptor
-      .internalBuildGeneratedFileFrom(descriptorData,
-        new com.google.protobuf.Descriptors.FileDescriptor[] {
-          org.tensorflow.proto.framework.MetaGraphProtos.getDescriptor(),
-        });
-    internal_static_tensorflow_SavedModel_descriptor =
-      getDescriptor().getMessageTypes().get(0);
-    internal_static_tensorflow_SavedModel_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_SavedModel_descriptor,
-        new java.lang.String[] { "SavedModelSchemaVersion", "MetaGraphs", });
-    org.tensorflow.proto.framework.MetaGraphProtos.getDescriptor();
-  }
-
-  // @@protoc_insertion_point(outer_class_scope)
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedObject.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedObject.java
deleted file mode 100644
index d38d76b85a1..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedObject.java
+++ /dev/null
@@ -1,3174 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/saved_object_graph.proto
-
-package org.tensorflow.proto.framework;
-
-/**
- * Protobuf type {@code tensorflow.SavedObject}
- */
-public  final class SavedObject extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.SavedObject)
-    SavedObjectOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use SavedObject.newBuilder() to construct.
-  private SavedObject(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private SavedObject() {
-    children_ = java.util.Collections.emptyList();
-    slotVariables_ = java.util.Collections.emptyList();
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new SavedObject();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private SavedObject(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    int mutable_bitField0_ = 0;
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 10: {
-            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
-              children_ = new java.util.ArrayList<org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference>();
-              mutable_bitField0_ |= 0x00000001;
-            }
-            children_.add(
-                input.readMessage(org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference.parser(), extensionRegistry));
-            break;
-          }
-          case 26: {
-            if (!((mutable_bitField0_ & 0x00000002) != 0)) {
-              slotVariables_ = new java.util.ArrayList<org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference>();
-              mutable_bitField0_ |= 0x00000002;
-            }
-            slotVariables_.add(
-                input.readMessage(org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference.parser(), extensionRegistry));
-            break;
-          }
-          case 34: {
-            org.tensorflow.proto.framework.SavedUserObject.Builder subBuilder = null;
-            if (kindCase_ == 4) {
-              subBuilder = ((org.tensorflow.proto.framework.SavedUserObject) kind_).toBuilder();
-            }
-            kind_ =
-                input.readMessage(org.tensorflow.proto.framework.SavedUserObject.parser(), extensionRegistry);
-            if (subBuilder != null) {
-              subBuilder.mergeFrom((org.tensorflow.proto.framework.SavedUserObject) kind_);
-              kind_ = subBuilder.buildPartial();
-            }
-            kindCase_ = 4;
-            break;
-          }
-          case 42: {
-            org.tensorflow.proto.framework.SavedAsset.Builder subBuilder = null;
-            if (kindCase_ == 5) {
-              subBuilder = ((org.tensorflow.proto.framework.SavedAsset) kind_).toBuilder();
-            }
-            kind_ =
-                input.readMessage(org.tensorflow.proto.framework.SavedAsset.parser(), extensionRegistry);
-            if (subBuilder != null) {
-              subBuilder.mergeFrom((org.tensorflow.proto.framework.SavedAsset) kind_);
-              kind_ = subBuilder.buildPartial();
-            }
-            kindCase_ = 5;
-            break;
-          }
-          case 50: {
-            org.tensorflow.proto.framework.SavedFunction.Builder subBuilder = null;
-            if (kindCase_ == 6) {
-              subBuilder = ((org.tensorflow.proto.framework.SavedFunction) kind_).toBuilder();
-            }
-            kind_ =
-                input.readMessage(org.tensorflow.proto.framework.SavedFunction.parser(), extensionRegistry);
-            if (subBuilder != null) {
-              subBuilder.mergeFrom((org.tensorflow.proto.framework.SavedFunction) kind_);
-              kind_ = subBuilder.buildPartial();
-            }
-            kindCase_ = 6;
-            break;
-          }
-          case 58: {
-            org.tensorflow.proto.framework.SavedVariable.Builder subBuilder = null;
-            if (kindCase_ == 7) {
-              subBuilder = ((org.tensorflow.proto.framework.SavedVariable) kind_).toBuilder();
-            }
-            kind_ =
-                input.readMessage(org.tensorflow.proto.framework.SavedVariable.parser(), extensionRegistry);
-            if (subBuilder != null) {
-              subBuilder.mergeFrom((org.tensorflow.proto.framework.SavedVariable) kind_);
-              kind_ = subBuilder.buildPartial();
-            }
-            kindCase_ = 7;
-            break;
-          }
-          case 66: {
-            org.tensorflow.proto.framework.SavedBareConcreteFunction.Builder subBuilder = null;
-            if (kindCase_ == 8) {
-              subBuilder = ((org.tensorflow.proto.framework.SavedBareConcreteFunction) kind_).toBuilder();
-            }
-            kind_ =
-                input.readMessage(org.tensorflow.proto.framework.SavedBareConcreteFunction.parser(), extensionRegistry);
-            if (subBuilder != null) {
-              subBuilder.mergeFrom((org.tensorflow.proto.framework.SavedBareConcreteFunction) kind_);
-              kind_ = subBuilder.buildPartial();
-            }
-            kindCase_ = 8;
-            break;
-          }
-          case 74: {
-            org.tensorflow.proto.framework.SavedConstant.Builder subBuilder = null;
-            if (kindCase_ == 9) {
-              subBuilder = ((org.tensorflow.proto.framework.SavedConstant) kind_).toBuilder();
-            }
-            kind_ =
-                input.readMessage(org.tensorflow.proto.framework.SavedConstant.parser(), extensionRegistry);
-            if (subBuilder != null) {
-              subBuilder.mergeFrom((org.tensorflow.proto.framework.SavedConstant) kind_);
-              kind_ = subBuilder.buildPartial();
-            }
-            kindCase_ = 9;
-            break;
-          }
-          case 82: {
-            org.tensorflow.proto.framework.SavedResource.Builder subBuilder = null;
-            if (kindCase_ == 10) {
-              subBuilder = ((org.tensorflow.proto.framework.SavedResource) kind_).toBuilder();
-            }
-            kind_ =
-                input.readMessage(org.tensorflow.proto.framework.SavedResource.parser(), extensionRegistry);
-            if (subBuilder != null) {
-              subBuilder.mergeFrom((org.tensorflow.proto.framework.SavedResource) kind_);
-              kind_ = subBuilder.buildPartial();
-            }
-            kindCase_ = 10;
-            break;
-          }
-          case 90: {
-            if (!((mutable_bitField0_ & 0x00000004) != 0)) {
-              saveableObjects_ = com.google.protobuf.MapField.newMapField(
-                  SaveableObjectsDefaultEntryHolder.defaultEntry);
-              mutable_bitField0_ |= 0x00000004;
-            }
-            com.google.protobuf.MapEntry<java.lang.String, org.tensorflow.proto.framework.SaveableObject>
-            saveableObjects__ = input.readMessage(
-                SaveableObjectsDefaultEntryHolder.defaultEntry.getParserForType(), extensionRegistry);
-            saveableObjects_.getMutableMap().put(
-                saveableObjects__.getKey(), saveableObjects__.getValue());
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      if (((mutable_bitField0_ & 0x00000001) != 0)) {
-        children_ = java.util.Collections.unmodifiableList(children_);
-      }
-      if (((mutable_bitField0_ & 0x00000002) != 0)) {
-        slotVariables_ = java.util.Collections.unmodifiableList(slotVariables_);
-      }
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.framework.SavedObjectGraphProtos.internal_static_tensorflow_SavedObject_descriptor;
-  }
-
-  @SuppressWarnings({"rawtypes"})
-  @java.lang.Override
-  protected com.google.protobuf.MapField internalGetMapField(
-      int number) {
-    switch (number) {
-      case 11:
-        return internalGetSaveableObjects();
-      default:
-        throw new RuntimeException(
-            "Invalid map field number: " + number);
-    }
-  }
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.framework.SavedObjectGraphProtos.internal_static_tensorflow_SavedObject_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.framework.SavedObject.class, org.tensorflow.proto.framework.SavedObject.Builder.class);
-  }
-
-  private int kindCase_ = 0;
-  private java.lang.Object kind_;
-  public enum KindCase
-      implements com.google.protobuf.Internal.EnumLite {
-    USER_OBJECT(4),
-    ASSET(5),
-    FUNCTION(6),
-    VARIABLE(7),
-    BARE_CONCRETE_FUNCTION(8),
-    CONSTANT(9),
-    RESOURCE(10),
-    KIND_NOT_SET(0);
-    private final int value;
-    private KindCase(int value) {
-      this.value = value;
-    }
-    /**
-     * @deprecated Use {@link #forNumber(int)} instead.
-     */
-    @java.lang.Deprecated
-    public static KindCase valueOf(int value) {
-      return forNumber(value);
-    }
-
-    public static KindCase forNumber(int value) {
-      switch (value) {
-        case 4: return USER_OBJECT;
-        case 5: return ASSET;
-        case 6: return FUNCTION;
-        case 7: return VARIABLE;
-        case 8: return BARE_CONCRETE_FUNCTION;
-        case 9: return CONSTANT;
-        case 10: return RESOURCE;
-        case 0: return KIND_NOT_SET;
-        default: return null;
-      }
-    }
-    public int getNumber() {
-      return this.value;
-    }
-  };
-
-  public KindCase
-  getKindCase() {
-    return KindCase.forNumber(
-        kindCase_);
-  }
-
-  public static final int CHILDREN_FIELD_NUMBER = 1;
-  private java.util.List<org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference> children_;
-  /**
-   * <pre>
-   * Objects which this object depends on: named edges in the dependency
-   * graph.
-   * Note: currently only valid if kind == "user_object".
-   * </pre>
-   *
-   * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.ObjectReference children = 1;</code>
-   */
-  public java.util.List<org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference> getChildrenList() {
-    return children_;
-  }
-  /**
-   * <pre>
-   * Objects which this object depends on: named edges in the dependency
-   * graph.
-   * Note: currently only valid if kind == "user_object".
-   * </pre>
-   *
-   * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.ObjectReference children = 1;</code>
-   */
-  public java.util.List<? extends org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReferenceOrBuilder> 
-      getChildrenOrBuilderList() {
-    return children_;
-  }
-  /**
-   * <pre>
-   * Objects which this object depends on: named edges in the dependency
-   * graph.
-   * Note: currently only valid if kind == "user_object".
-   * </pre>
-   *
-   * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.ObjectReference children = 1;</code>
-   */
-  public int getChildrenCount() {
-    return children_.size();
-  }
-  /**
-   * <pre>
-   * Objects which this object depends on: named edges in the dependency
-   * graph.
-   * Note: currently only valid if kind == "user_object".
-   * </pre>
-   *
-   * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.ObjectReference children = 1;</code>
-   */
-  public org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference getChildren(int index) {
-    return children_.get(index);
-  }
-  /**
-   * <pre>
-   * Objects which this object depends on: named edges in the dependency
-   * graph.
-   * Note: currently only valid if kind == "user_object".
-   * </pre>
-   *
-   * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.ObjectReference children = 1;</code>
-   */
-  public org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReferenceOrBuilder getChildrenOrBuilder(
-      int index) {
-    return children_.get(index);
-  }
-
-  public static final int SLOT_VARIABLES_FIELD_NUMBER = 3;
-  private java.util.List<org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference> slotVariables_;
-  /**
-   * <pre>
-   * Slot variables owned by this object. This describes the three-way
-   * (optimizer, variable, slot variable) relationship; none of the three
-   * depend on the others directly.
-   * Note: currently only valid if kind == "user_object".
-   * </pre>
-   *
-   * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SlotVariableReference slot_variables = 3;</code>
-   */
-  public java.util.List<org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference> getSlotVariablesList() {
-    return slotVariables_;
-  }
-  /**
-   * <pre>
-   * Slot variables owned by this object. This describes the three-way
-   * (optimizer, variable, slot variable) relationship; none of the three
-   * depend on the others directly.
-   * Note: currently only valid if kind == "user_object".
-   * </pre>
-   *
-   * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SlotVariableReference slot_variables = 3;</code>
-   */
-  public java.util.List<? extends org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReferenceOrBuilder> 
-      getSlotVariablesOrBuilderList() {
-    return slotVariables_;
-  }
-  /**
-   * <pre>
-   * Slot variables owned by this object. This describes the three-way
-   * (optimizer, variable, slot variable) relationship; none of the three
-   * depend on the others directly.
-   * Note: currently only valid if kind == "user_object".
-   * </pre>
-   *
-   * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SlotVariableReference slot_variables = 3;</code>
-   */
-  public int getSlotVariablesCount() {
-    return slotVariables_.size();
-  }
-  /**
-   * <pre>
-   * Slot variables owned by this object. This describes the three-way
-   * (optimizer, variable, slot variable) relationship; none of the three
-   * depend on the others directly.
-   * Note: currently only valid if kind == "user_object".
-   * </pre>
-   *
-   * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SlotVariableReference slot_variables = 3;</code>
-   */
-  public org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference getSlotVariables(int index) {
-    return slotVariables_.get(index);
-  }
-  /**
-   * <pre>
-   * Slot variables owned by this object. This describes the three-way
-   * (optimizer, variable, slot variable) relationship; none of the three
-   * depend on the others directly.
-   * Note: currently only valid if kind == "user_object".
-   * </pre>
-   *
-   * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SlotVariableReference slot_variables = 3;</code>
-   */
-  public org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReferenceOrBuilder getSlotVariablesOrBuilder(
-      int index) {
-    return slotVariables_.get(index);
-  }
-
-  public static final int USER_OBJECT_FIELD_NUMBER = 4;
-  /**
-   * <code>.tensorflow.SavedUserObject user_object = 4;</code>
-   */
-  public boolean hasUserObject() {
-    return kindCase_ == 4;
-  }
-  /**
-   * <code>.tensorflow.SavedUserObject user_object = 4;</code>
-   */
-  public org.tensorflow.proto.framework.SavedUserObject getUserObject() {
-    if (kindCase_ == 4) {
-       return (org.tensorflow.proto.framework.SavedUserObject) kind_;
-    }
-    return org.tensorflow.proto.framework.SavedUserObject.getDefaultInstance();
-  }
-  /**
-   * <code>.tensorflow.SavedUserObject user_object = 4;</code>
-   */
-  public org.tensorflow.proto.framework.SavedUserObjectOrBuilder getUserObjectOrBuilder() {
-    if (kindCase_ == 4) {
-       return (org.tensorflow.proto.framework.SavedUserObject) kind_;
-    }
-    return org.tensorflow.proto.framework.SavedUserObject.getDefaultInstance();
-  }
-
-  public static final int ASSET_FIELD_NUMBER = 5;
-  /**
-   * <code>.tensorflow.SavedAsset asset = 5;</code>
-   */
-  public boolean hasAsset() {
-    return kindCase_ == 5;
-  }
-  /**
-   * <code>.tensorflow.SavedAsset asset = 5;</code>
-   */
-  public org.tensorflow.proto.framework.SavedAsset getAsset() {
-    if (kindCase_ == 5) {
-       return (org.tensorflow.proto.framework.SavedAsset) kind_;
-    }
-    return org.tensorflow.proto.framework.SavedAsset.getDefaultInstance();
-  }
-  /**
-   * <code>.tensorflow.SavedAsset asset = 5;</code>
-   */
-  public org.tensorflow.proto.framework.SavedAssetOrBuilder getAssetOrBuilder() {
-    if (kindCase_ == 5) {
-       return (org.tensorflow.proto.framework.SavedAsset) kind_;
-    }
-    return org.tensorflow.proto.framework.SavedAsset.getDefaultInstance();
-  }
-
-  public static final int FUNCTION_FIELD_NUMBER = 6;
-  /**
-   * <code>.tensorflow.SavedFunction function = 6;</code>
-   */
-  public boolean hasFunction() {
-    return kindCase_ == 6;
-  }
-  /**
-   * <code>.tensorflow.SavedFunction function = 6;</code>
-   */
-  public org.tensorflow.proto.framework.SavedFunction getFunction() {
-    if (kindCase_ == 6) {
-       return (org.tensorflow.proto.framework.SavedFunction) kind_;
-    }
-    return org.tensorflow.proto.framework.SavedFunction.getDefaultInstance();
-  }
-  /**
-   * <code>.tensorflow.SavedFunction function = 6;</code>
-   */
-  public org.tensorflow.proto.framework.SavedFunctionOrBuilder getFunctionOrBuilder() {
-    if (kindCase_ == 6) {
-       return (org.tensorflow.proto.framework.SavedFunction) kind_;
-    }
-    return org.tensorflow.proto.framework.SavedFunction.getDefaultInstance();
-  }
-
-  public static final int VARIABLE_FIELD_NUMBER = 7;
-  /**
-   * <code>.tensorflow.SavedVariable variable = 7;</code>
-   */
-  public boolean hasVariable() {
-    return kindCase_ == 7;
-  }
-  /**
-   * <code>.tensorflow.SavedVariable variable = 7;</code>
-   */
-  public org.tensorflow.proto.framework.SavedVariable getVariable() {
-    if (kindCase_ == 7) {
-       return (org.tensorflow.proto.framework.SavedVariable) kind_;
-    }
-    return org.tensorflow.proto.framework.SavedVariable.getDefaultInstance();
-  }
-  /**
-   * <code>.tensorflow.SavedVariable variable = 7;</code>
-   */
-  public org.tensorflow.proto.framework.SavedVariableOrBuilder getVariableOrBuilder() {
-    if (kindCase_ == 7) {
-       return (org.tensorflow.proto.framework.SavedVariable) kind_;
-    }
-    return org.tensorflow.proto.framework.SavedVariable.getDefaultInstance();
-  }
-
-  public static final int BARE_CONCRETE_FUNCTION_FIELD_NUMBER = 8;
-  /**
-   * <code>.tensorflow.SavedBareConcreteFunction bare_concrete_function = 8;</code>
-   */
-  public boolean hasBareConcreteFunction() {
-    return kindCase_ == 8;
-  }
-  /**
-   * <code>.tensorflow.SavedBareConcreteFunction bare_concrete_function = 8;</code>
-   */
-  public org.tensorflow.proto.framework.SavedBareConcreteFunction getBareConcreteFunction() {
-    if (kindCase_ == 8) {
-       return (org.tensorflow.proto.framework.SavedBareConcreteFunction) kind_;
-    }
-    return org.tensorflow.proto.framework.SavedBareConcreteFunction.getDefaultInstance();
-  }
-  /**
-   * <code>.tensorflow.SavedBareConcreteFunction bare_concrete_function = 8;</code>
-   */
-  public org.tensorflow.proto.framework.SavedBareConcreteFunctionOrBuilder getBareConcreteFunctionOrBuilder() {
-    if (kindCase_ == 8) {
-       return (org.tensorflow.proto.framework.SavedBareConcreteFunction) kind_;
-    }
-    return org.tensorflow.proto.framework.SavedBareConcreteFunction.getDefaultInstance();
-  }
-
-  public static final int CONSTANT_FIELD_NUMBER = 9;
-  /**
-   * <code>.tensorflow.SavedConstant constant = 9;</code>
-   */
-  public boolean hasConstant() {
-    return kindCase_ == 9;
-  }
-  /**
-   * <code>.tensorflow.SavedConstant constant = 9;</code>
-   */
-  public org.tensorflow.proto.framework.SavedConstant getConstant() {
-    if (kindCase_ == 9) {
-       return (org.tensorflow.proto.framework.SavedConstant) kind_;
-    }
-    return org.tensorflow.proto.framework.SavedConstant.getDefaultInstance();
-  }
-  /**
-   * <code>.tensorflow.SavedConstant constant = 9;</code>
-   */
-  public org.tensorflow.proto.framework.SavedConstantOrBuilder getConstantOrBuilder() {
-    if (kindCase_ == 9) {
-       return (org.tensorflow.proto.framework.SavedConstant) kind_;
-    }
-    return org.tensorflow.proto.framework.SavedConstant.getDefaultInstance();
-  }
-
-  public static final int RESOURCE_FIELD_NUMBER = 10;
-  /**
-   * <code>.tensorflow.SavedResource resource = 10;</code>
-   */
-  public boolean hasResource() {
-    return kindCase_ == 10;
-  }
-  /**
-   * <code>.tensorflow.SavedResource resource = 10;</code>
-   */
-  public org.tensorflow.proto.framework.SavedResource getResource() {
-    if (kindCase_ == 10) {
-       return (org.tensorflow.proto.framework.SavedResource) kind_;
-    }
-    return org.tensorflow.proto.framework.SavedResource.getDefaultInstance();
-  }
-  /**
-   * <code>.tensorflow.SavedResource resource = 10;</code>
-   */
-  public org.tensorflow.proto.framework.SavedResourceOrBuilder getResourceOrBuilder() {
-    if (kindCase_ == 10) {
-       return (org.tensorflow.proto.framework.SavedResource) kind_;
-    }
-    return org.tensorflow.proto.framework.SavedResource.getDefaultInstance();
-  }
-
-  public static final int SAVEABLE_OBJECTS_FIELD_NUMBER = 11;
-  private static final class SaveableObjectsDefaultEntryHolder {
-    static final com.google.protobuf.MapEntry<
-        java.lang.String, org.tensorflow.proto.framework.SaveableObject> defaultEntry =
-            com.google.protobuf.MapEntry
-            .<java.lang.String, org.tensorflow.proto.framework.SaveableObject>newDefaultInstance(
-                org.tensorflow.proto.framework.SavedObjectGraphProtos.internal_static_tensorflow_SavedObject_SaveableObjectsEntry_descriptor, 
-                com.google.protobuf.WireFormat.FieldType.STRING,
-                "",
-                com.google.protobuf.WireFormat.FieldType.MESSAGE,
-                org.tensorflow.proto.framework.SaveableObject.getDefaultInstance());
-  }
-  private com.google.protobuf.MapField<
-      java.lang.String, org.tensorflow.proto.framework.SaveableObject> saveableObjects_;
-  private com.google.protobuf.MapField<java.lang.String, org.tensorflow.proto.framework.SaveableObject>
-  internalGetSaveableObjects() {
-    if (saveableObjects_ == null) {
-      return com.google.protobuf.MapField.emptyMapField(
-          SaveableObjectsDefaultEntryHolder.defaultEntry);
-    }
-    return saveableObjects_;
-  }
-
-  public int getSaveableObjectsCount() {
-    return internalGetSaveableObjects().getMap().size();
-  }
-  /**
-   * <code>map&lt;string, .tensorflow.SaveableObject&gt; saveable_objects = 11;</code>
-   */
-
-  public boolean containsSaveableObjects(
-      java.lang.String key) {
-    if (key == null) { throw new java.lang.NullPointerException(); }
-    return internalGetSaveableObjects().getMap().containsKey(key);
-  }
-  /**
-   * Use {@link #getSaveableObjectsMap()} instead.
-   */
-  @java.lang.Deprecated
-  public java.util.Map<java.lang.String, org.tensorflow.proto.framework.SaveableObject> getSaveableObjects() {
-    return getSaveableObjectsMap();
-  }
-  /**
-   * <code>map&lt;string, .tensorflow.SaveableObject&gt; saveable_objects = 11;</code>
-   */
-
-  public java.util.Map<java.lang.String, org.tensorflow.proto.framework.SaveableObject> getSaveableObjectsMap() {
-    return internalGetSaveableObjects().getMap();
-  }
-  /**
-   * <code>map&lt;string, .tensorflow.SaveableObject&gt; saveable_objects = 11;</code>
-   */
-
-  public org.tensorflow.proto.framework.SaveableObject getSaveableObjectsOrDefault(
-      java.lang.String key,
-      org.tensorflow.proto.framework.SaveableObject defaultValue) {
-    if (key == null) { throw new java.lang.NullPointerException(); }
-    java.util.Map<java.lang.String, org.tensorflow.proto.framework.SaveableObject> map =
-        internalGetSaveableObjects().getMap();
-    return map.containsKey(key) ? map.get(key) : defaultValue;
-  }
-  /**
-   * <code>map&lt;string, .tensorflow.SaveableObject&gt; saveable_objects = 11;</code>
-   */
-
-  public org.tensorflow.proto.framework.SaveableObject getSaveableObjectsOrThrow(
-      java.lang.String key) {
-    if (key == null) { throw new java.lang.NullPointerException(); }
-    java.util.Map<java.lang.String, org.tensorflow.proto.framework.SaveableObject> map =
-        internalGetSaveableObjects().getMap();
-    if (!map.containsKey(key)) {
-      throw new java.lang.IllegalArgumentException();
-    }
-    return map.get(key);
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    for (int i = 0; i < children_.size(); i++) {
-      output.writeMessage(1, children_.get(i));
-    }
-    for (int i = 0; i < slotVariables_.size(); i++) {
-      output.writeMessage(3, slotVariables_.get(i));
-    }
-    if (kindCase_ == 4) {
-      output.writeMessage(4, (org.tensorflow.proto.framework.SavedUserObject) kind_);
-    }
-    if (kindCase_ == 5) {
-      output.writeMessage(5, (org.tensorflow.proto.framework.SavedAsset) kind_);
-    }
-    if (kindCase_ == 6) {
-      output.writeMessage(6, (org.tensorflow.proto.framework.SavedFunction) kind_);
-    }
-    if (kindCase_ == 7) {
-      output.writeMessage(7, (org.tensorflow.proto.framework.SavedVariable) kind_);
-    }
-    if (kindCase_ == 8) {
-      output.writeMessage(8, (org.tensorflow.proto.framework.SavedBareConcreteFunction) kind_);
-    }
-    if (kindCase_ == 9) {
-      output.writeMessage(9, (org.tensorflow.proto.framework.SavedConstant) kind_);
-    }
-    if (kindCase_ == 10) {
-      output.writeMessage(10, (org.tensorflow.proto.framework.SavedResource) kind_);
-    }
-    com.google.protobuf.GeneratedMessageV3
-      .serializeStringMapTo(
-        output,
-        internalGetSaveableObjects(),
-        SaveableObjectsDefaultEntryHolder.defaultEntry,
-        11);
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    for (int i = 0; i < children_.size(); i++) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(1, children_.get(i));
-    }
-    for (int i = 0; i < slotVariables_.size(); i++) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(3, slotVariables_.get(i));
-    }
-    if (kindCase_ == 4) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(4, (org.tensorflow.proto.framework.SavedUserObject) kind_);
-    }
-    if (kindCase_ == 5) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(5, (org.tensorflow.proto.framework.SavedAsset) kind_);
-    }
-    if (kindCase_ == 6) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(6, (org.tensorflow.proto.framework.SavedFunction) kind_);
-    }
-    if (kindCase_ == 7) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(7, (org.tensorflow.proto.framework.SavedVariable) kind_);
-    }
-    if (kindCase_ == 8) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(8, (org.tensorflow.proto.framework.SavedBareConcreteFunction) kind_);
-    }
-    if (kindCase_ == 9) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(9, (org.tensorflow.proto.framework.SavedConstant) kind_);
-    }
-    if (kindCase_ == 10) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(10, (org.tensorflow.proto.framework.SavedResource) kind_);
-    }
-    for (java.util.Map.Entry<java.lang.String, org.tensorflow.proto.framework.SaveableObject> entry
-         : internalGetSaveableObjects().getMap().entrySet()) {
-      com.google.protobuf.MapEntry<java.lang.String, org.tensorflow.proto.framework.SaveableObject>
-      saveableObjects__ = SaveableObjectsDefaultEntryHolder.defaultEntry.newBuilderForType()
-          .setKey(entry.getKey())
-          .setValue(entry.getValue())
-          .build();
-      size += com.google.protobuf.CodedOutputStream
-          .computeMessageSize(11, saveableObjects__);
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.framework.SavedObject)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.framework.SavedObject other = (org.tensorflow.proto.framework.SavedObject) obj;
-
-    if (!getChildrenList()
-        .equals(other.getChildrenList())) return false;
-    if (!getSlotVariablesList()
-        .equals(other.getSlotVariablesList())) return false;
-    if (!internalGetSaveableObjects().equals(
-        other.internalGetSaveableObjects())) return false;
-    if (!getKindCase().equals(other.getKindCase())) return false;
-    switch (kindCase_) {
-      case 4:
-        if (!getUserObject()
-            .equals(other.getUserObject())) return false;
-        break;
-      case 5:
-        if (!getAsset()
-            .equals(other.getAsset())) return false;
-        break;
-      case 6:
-        if (!getFunction()
-            .equals(other.getFunction())) return false;
-        break;
-      case 7:
-        if (!getVariable()
-            .equals(other.getVariable())) return false;
-        break;
-      case 8:
-        if (!getBareConcreteFunction()
-            .equals(other.getBareConcreteFunction())) return false;
-        break;
-      case 9:
-        if (!getConstant()
-            .equals(other.getConstant())) return false;
-        break;
-      case 10:
-        if (!getResource()
-            .equals(other.getResource())) return false;
-        break;
-      case 0:
-      default:
-    }
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    if (getChildrenCount() > 0) {
-      hash = (37 * hash) + CHILDREN_FIELD_NUMBER;
-      hash = (53 * hash) + getChildrenList().hashCode();
-    }
-    if (getSlotVariablesCount() > 0) {
-      hash = (37 * hash) + SLOT_VARIABLES_FIELD_NUMBER;
-      hash = (53 * hash) + getSlotVariablesList().hashCode();
-    }
-    if (!internalGetSaveableObjects().getMap().isEmpty()) {
-      hash = (37 * hash) + SAVEABLE_OBJECTS_FIELD_NUMBER;
-      hash = (53 * hash) + internalGetSaveableObjects().hashCode();
-    }
-    switch (kindCase_) {
-      case 4:
-        hash = (37 * hash) + USER_OBJECT_FIELD_NUMBER;
-        hash = (53 * hash) + getUserObject().hashCode();
-        break;
-      case 5:
-        hash = (37 * hash) + ASSET_FIELD_NUMBER;
-        hash = (53 * hash) + getAsset().hashCode();
-        break;
-      case 6:
-        hash = (37 * hash) + FUNCTION_FIELD_NUMBER;
-        hash = (53 * hash) + getFunction().hashCode();
-        break;
-      case 7:
-        hash = (37 * hash) + VARIABLE_FIELD_NUMBER;
-        hash = (53 * hash) + getVariable().hashCode();
-        break;
-      case 8:
-        hash = (37 * hash) + BARE_CONCRETE_FUNCTION_FIELD_NUMBER;
-        hash = (53 * hash) + getBareConcreteFunction().hashCode();
-        break;
-      case 9:
-        hash = (37 * hash) + CONSTANT_FIELD_NUMBER;
-        hash = (53 * hash) + getConstant().hashCode();
-        break;
-      case 10:
-        hash = (37 * hash) + RESOURCE_FIELD_NUMBER;
-        hash = (53 * hash) + getResource().hashCode();
-        break;
-      case 0:
-      default:
-    }
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.framework.SavedObject parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.SavedObject parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.SavedObject parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.SavedObject parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.SavedObject parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.SavedObject parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.SavedObject parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.SavedObject parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.SavedObject parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.SavedObject parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.SavedObject parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.SavedObject parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.framework.SavedObject prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * Protobuf type {@code tensorflow.SavedObject}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.SavedObject)
-      org.tensorflow.proto.framework.SavedObjectOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.SavedObjectGraphProtos.internal_static_tensorflow_SavedObject_descriptor;
-    }
-
-    @SuppressWarnings({"rawtypes"})
-    protected com.google.protobuf.MapField internalGetMapField(
-        int number) {
-      switch (number) {
-        case 11:
-          return internalGetSaveableObjects();
-        default:
-          throw new RuntimeException(
-              "Invalid map field number: " + number);
-      }
-    }
-    @SuppressWarnings({"rawtypes"})
-    protected com.google.protobuf.MapField internalGetMutableMapField(
-        int number) {
-      switch (number) {
-        case 11:
-          return internalGetMutableSaveableObjects();
-        default:
-          throw new RuntimeException(
-              "Invalid map field number: " + number);
-      }
-    }
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.SavedObjectGraphProtos.internal_static_tensorflow_SavedObject_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.SavedObject.class, org.tensorflow.proto.framework.SavedObject.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.framework.SavedObject.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-        getChildrenFieldBuilder();
-        getSlotVariablesFieldBuilder();
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      if (childrenBuilder_ == null) {
-        children_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000001);
-      } else {
-        childrenBuilder_.clear();
-      }
-      if (slotVariablesBuilder_ == null) {
-        slotVariables_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000002);
-      } else {
-        slotVariablesBuilder_.clear();
-      }
-      internalGetMutableSaveableObjects().clear();
-      kindCase_ = 0;
-      kind_ = null;
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.framework.SavedObjectGraphProtos.internal_static_tensorflow_SavedObject_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.SavedObject getDefaultInstanceForType() {
-      return org.tensorflow.proto.framework.SavedObject.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.SavedObject build() {
-      org.tensorflow.proto.framework.SavedObject result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.SavedObject buildPartial() {
-      org.tensorflow.proto.framework.SavedObject result = new org.tensorflow.proto.framework.SavedObject(this);
-      int from_bitField0_ = bitField0_;
-      if (childrenBuilder_ == null) {
-        if (((bitField0_ & 0x00000001) != 0)) {
-          children_ = java.util.Collections.unmodifiableList(children_);
-          bitField0_ = (bitField0_ & ~0x00000001);
-        }
-        result.children_ = children_;
-      } else {
-        result.children_ = childrenBuilder_.build();
-      }
-      if (slotVariablesBuilder_ == null) {
-        if (((bitField0_ & 0x00000002) != 0)) {
-          slotVariables_ = java.util.Collections.unmodifiableList(slotVariables_);
-          bitField0_ = (bitField0_ & ~0x00000002);
-        }
-        result.slotVariables_ = slotVariables_;
-      } else {
-        result.slotVariables_ = slotVariablesBuilder_.build();
-      }
-      if (kindCase_ == 4) {
-        if (userObjectBuilder_ == null) {
-          result.kind_ = kind_;
-        } else {
-          result.kind_ = userObjectBuilder_.build();
-        }
-      }
-      if (kindCase_ == 5) {
-        if (assetBuilder_ == null) {
-          result.kind_ = kind_;
-        } else {
-          result.kind_ = assetBuilder_.build();
-        }
-      }
-      if (kindCase_ == 6) {
-        if (functionBuilder_ == null) {
-          result.kind_ = kind_;
-        } else {
-          result.kind_ = functionBuilder_.build();
-        }
-      }
-      if (kindCase_ == 7) {
-        if (variableBuilder_ == null) {
-          result.kind_ = kind_;
-        } else {
-          result.kind_ = variableBuilder_.build();
-        }
-      }
-      if (kindCase_ == 8) {
-        if (bareConcreteFunctionBuilder_ == null) {
-          result.kind_ = kind_;
-        } else {
-          result.kind_ = bareConcreteFunctionBuilder_.build();
-        }
-      }
-      if (kindCase_ == 9) {
-        if (constantBuilder_ == null) {
-          result.kind_ = kind_;
-        } else {
-          result.kind_ = constantBuilder_.build();
-        }
-      }
-      if (kindCase_ == 10) {
-        if (resourceBuilder_ == null) {
-          result.kind_ = kind_;
-        } else {
-          result.kind_ = resourceBuilder_.build();
-        }
-      }
-      result.saveableObjects_ = internalGetSaveableObjects();
-      result.saveableObjects_.makeImmutable();
-      result.kindCase_ = kindCase_;
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.framework.SavedObject) {
-        return mergeFrom((org.tensorflow.proto.framework.SavedObject)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.framework.SavedObject other) {
-      if (other == org.tensorflow.proto.framework.SavedObject.getDefaultInstance()) return this;
-      if (childrenBuilder_ == null) {
-        if (!other.children_.isEmpty()) {
-          if (children_.isEmpty()) {
-            children_ = other.children_;
-            bitField0_ = (bitField0_ & ~0x00000001);
-          } else {
-            ensureChildrenIsMutable();
-            children_.addAll(other.children_);
-          }
-          onChanged();
-        }
-      } else {
-        if (!other.children_.isEmpty()) {
-          if (childrenBuilder_.isEmpty()) {
-            childrenBuilder_.dispose();
-            childrenBuilder_ = null;
-            children_ = other.children_;
-            bitField0_ = (bitField0_ & ~0x00000001);
-            childrenBuilder_ = 
-              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
-                 getChildrenFieldBuilder() : null;
-          } else {
-            childrenBuilder_.addAllMessages(other.children_);
-          }
-        }
-      }
-      if (slotVariablesBuilder_ == null) {
-        if (!other.slotVariables_.isEmpty()) {
-          if (slotVariables_.isEmpty()) {
-            slotVariables_ = other.slotVariables_;
-            bitField0_ = (bitField0_ & ~0x00000002);
-          } else {
-            ensureSlotVariablesIsMutable();
-            slotVariables_.addAll(other.slotVariables_);
-          }
-          onChanged();
-        }
-      } else {
-        if (!other.slotVariables_.isEmpty()) {
-          if (slotVariablesBuilder_.isEmpty()) {
-            slotVariablesBuilder_.dispose();
-            slotVariablesBuilder_ = null;
-            slotVariables_ = other.slotVariables_;
-            bitField0_ = (bitField0_ & ~0x00000002);
-            slotVariablesBuilder_ = 
-              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
-                 getSlotVariablesFieldBuilder() : null;
-          } else {
-            slotVariablesBuilder_.addAllMessages(other.slotVariables_);
-          }
-        }
-      }
-      internalGetMutableSaveableObjects().mergeFrom(
-          other.internalGetSaveableObjects());
-      switch (other.getKindCase()) {
-        case USER_OBJECT: {
-          mergeUserObject(other.getUserObject());
-          break;
-        }
-        case ASSET: {
-          mergeAsset(other.getAsset());
-          break;
-        }
-        case FUNCTION: {
-          mergeFunction(other.getFunction());
-          break;
-        }
-        case VARIABLE: {
-          mergeVariable(other.getVariable());
-          break;
-        }
-        case BARE_CONCRETE_FUNCTION: {
-          mergeBareConcreteFunction(other.getBareConcreteFunction());
-          break;
-        }
-        case CONSTANT: {
-          mergeConstant(other.getConstant());
-          break;
-        }
-        case RESOURCE: {
-          mergeResource(other.getResource());
-          break;
-        }
-        case KIND_NOT_SET: {
-          break;
-        }
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.framework.SavedObject parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.framework.SavedObject) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-    private int kindCase_ = 0;
-    private java.lang.Object kind_;
-    public KindCase
-        getKindCase() {
-      return KindCase.forNumber(
-          kindCase_);
-    }
-
-    public Builder clearKind() {
-      kindCase_ = 0;
-      kind_ = null;
-      onChanged();
-      return this;
-    }
-
-    private int bitField0_;
-
-    private java.util.List<org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference> children_ =
-      java.util.Collections.emptyList();
-    private void ensureChildrenIsMutable() {
-      if (!((bitField0_ & 0x00000001) != 0)) {
-        children_ = new java.util.ArrayList<org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference>(children_);
-        bitField0_ |= 0x00000001;
-       }
-    }
-
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference.Builder, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReferenceOrBuilder> childrenBuilder_;
-
-    /**
-     * <pre>
-     * Objects which this object depends on: named edges in the dependency
-     * graph.
-     * Note: currently only valid if kind == "user_object".
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.ObjectReference children = 1;</code>
-     */
-    public java.util.List<org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference> getChildrenList() {
-      if (childrenBuilder_ == null) {
-        return java.util.Collections.unmodifiableList(children_);
-      } else {
-        return childrenBuilder_.getMessageList();
-      }
-    }
-    /**
-     * <pre>
-     * Objects which this object depends on: named edges in the dependency
-     * graph.
-     * Note: currently only valid if kind == "user_object".
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.ObjectReference children = 1;</code>
-     */
-    public int getChildrenCount() {
-      if (childrenBuilder_ == null) {
-        return children_.size();
-      } else {
-        return childrenBuilder_.getCount();
-      }
-    }
-    /**
-     * <pre>
-     * Objects which this object depends on: named edges in the dependency
-     * graph.
-     * Note: currently only valid if kind == "user_object".
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.ObjectReference children = 1;</code>
-     */
-    public org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference getChildren(int index) {
-      if (childrenBuilder_ == null) {
-        return children_.get(index);
-      } else {
-        return childrenBuilder_.getMessage(index);
-      }
-    }
-    /**
-     * <pre>
-     * Objects which this object depends on: named edges in the dependency
-     * graph.
-     * Note: currently only valid if kind == "user_object".
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.ObjectReference children = 1;</code>
-     */
-    public Builder setChildren(
-        int index, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference value) {
-      if (childrenBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureChildrenIsMutable();
-        children_.set(index, value);
-        onChanged();
-      } else {
-        childrenBuilder_.setMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Objects which this object depends on: named edges in the dependency
-     * graph.
-     * Note: currently only valid if kind == "user_object".
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.ObjectReference children = 1;</code>
-     */
-    public Builder setChildren(
-        int index, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference.Builder builderForValue) {
-      if (childrenBuilder_ == null) {
-        ensureChildrenIsMutable();
-        children_.set(index, builderForValue.build());
-        onChanged();
-      } else {
-        childrenBuilder_.setMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Objects which this object depends on: named edges in the dependency
-     * graph.
-     * Note: currently only valid if kind == "user_object".
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.ObjectReference children = 1;</code>
-     */
-    public Builder addChildren(org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference value) {
-      if (childrenBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureChildrenIsMutable();
-        children_.add(value);
-        onChanged();
-      } else {
-        childrenBuilder_.addMessage(value);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Objects which this object depends on: named edges in the dependency
-     * graph.
-     * Note: currently only valid if kind == "user_object".
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.ObjectReference children = 1;</code>
-     */
-    public Builder addChildren(
-        int index, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference value) {
-      if (childrenBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureChildrenIsMutable();
-        children_.add(index, value);
-        onChanged();
-      } else {
-        childrenBuilder_.addMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Objects which this object depends on: named edges in the dependency
-     * graph.
-     * Note: currently only valid if kind == "user_object".
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.ObjectReference children = 1;</code>
-     */
-    public Builder addChildren(
-        org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference.Builder builderForValue) {
-      if (childrenBuilder_ == null) {
-        ensureChildrenIsMutable();
-        children_.add(builderForValue.build());
-        onChanged();
-      } else {
-        childrenBuilder_.addMessage(builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Objects which this object depends on: named edges in the dependency
-     * graph.
-     * Note: currently only valid if kind == "user_object".
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.ObjectReference children = 1;</code>
-     */
-    public Builder addChildren(
-        int index, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference.Builder builderForValue) {
-      if (childrenBuilder_ == null) {
-        ensureChildrenIsMutable();
-        children_.add(index, builderForValue.build());
-        onChanged();
-      } else {
-        childrenBuilder_.addMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Objects which this object depends on: named edges in the dependency
-     * graph.
-     * Note: currently only valid if kind == "user_object".
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.ObjectReference children = 1;</code>
-     */
-    public Builder addAllChildren(
-        java.lang.Iterable<? extends org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference> values) {
-      if (childrenBuilder_ == null) {
-        ensureChildrenIsMutable();
-        com.google.protobuf.AbstractMessageLite.Builder.addAll(
-            values, children_);
-        onChanged();
-      } else {
-        childrenBuilder_.addAllMessages(values);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Objects which this object depends on: named edges in the dependency
-     * graph.
-     * Note: currently only valid if kind == "user_object".
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.ObjectReference children = 1;</code>
-     */
-    public Builder clearChildren() {
-      if (childrenBuilder_ == null) {
-        children_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000001);
-        onChanged();
-      } else {
-        childrenBuilder_.clear();
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Objects which this object depends on: named edges in the dependency
-     * graph.
-     * Note: currently only valid if kind == "user_object".
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.ObjectReference children = 1;</code>
-     */
-    public Builder removeChildren(int index) {
-      if (childrenBuilder_ == null) {
-        ensureChildrenIsMutable();
-        children_.remove(index);
-        onChanged();
-      } else {
-        childrenBuilder_.remove(index);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Objects which this object depends on: named edges in the dependency
-     * graph.
-     * Note: currently only valid if kind == "user_object".
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.ObjectReference children = 1;</code>
-     */
-    public org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference.Builder getChildrenBuilder(
-        int index) {
-      return getChildrenFieldBuilder().getBuilder(index);
-    }
-    /**
-     * <pre>
-     * Objects which this object depends on: named edges in the dependency
-     * graph.
-     * Note: currently only valid if kind == "user_object".
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.ObjectReference children = 1;</code>
-     */
-    public org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReferenceOrBuilder getChildrenOrBuilder(
-        int index) {
-      if (childrenBuilder_ == null) {
-        return children_.get(index);  } else {
-        return childrenBuilder_.getMessageOrBuilder(index);
-      }
-    }
-    /**
-     * <pre>
-     * Objects which this object depends on: named edges in the dependency
-     * graph.
-     * Note: currently only valid if kind == "user_object".
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.ObjectReference children = 1;</code>
-     */
-    public java.util.List<? extends org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReferenceOrBuilder> 
-         getChildrenOrBuilderList() {
-      if (childrenBuilder_ != null) {
-        return childrenBuilder_.getMessageOrBuilderList();
-      } else {
-        return java.util.Collections.unmodifiableList(children_);
-      }
-    }
-    /**
-     * <pre>
-     * Objects which this object depends on: named edges in the dependency
-     * graph.
-     * Note: currently only valid if kind == "user_object".
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.ObjectReference children = 1;</code>
-     */
-    public org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference.Builder addChildrenBuilder() {
-      return getChildrenFieldBuilder().addBuilder(
-          org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference.getDefaultInstance());
-    }
-    /**
-     * <pre>
-     * Objects which this object depends on: named edges in the dependency
-     * graph.
-     * Note: currently only valid if kind == "user_object".
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.ObjectReference children = 1;</code>
-     */
-    public org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference.Builder addChildrenBuilder(
-        int index) {
-      return getChildrenFieldBuilder().addBuilder(
-          index, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference.getDefaultInstance());
-    }
-    /**
-     * <pre>
-     * Objects which this object depends on: named edges in the dependency
-     * graph.
-     * Note: currently only valid if kind == "user_object".
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.ObjectReference children = 1;</code>
-     */
-    public java.util.List<org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference.Builder> 
-         getChildrenBuilderList() {
-      return getChildrenFieldBuilder().getBuilderList();
-    }
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference.Builder, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReferenceOrBuilder> 
-        getChildrenFieldBuilder() {
-      if (childrenBuilder_ == null) {
-        childrenBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
-            org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference.Builder, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReferenceOrBuilder>(
-                children_,
-                ((bitField0_ & 0x00000001) != 0),
-                getParentForChildren(),
-                isClean());
-        children_ = null;
-      }
-      return childrenBuilder_;
-    }
-
-    private java.util.List<org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference> slotVariables_ =
-      java.util.Collections.emptyList();
-    private void ensureSlotVariablesIsMutable() {
-      if (!((bitField0_ & 0x00000002) != 0)) {
-        slotVariables_ = new java.util.ArrayList<org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference>(slotVariables_);
-        bitField0_ |= 0x00000002;
-       }
-    }
-
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference.Builder, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReferenceOrBuilder> slotVariablesBuilder_;
-
-    /**
-     * <pre>
-     * Slot variables owned by this object. This describes the three-way
-     * (optimizer, variable, slot variable) relationship; none of the three
-     * depend on the others directly.
-     * Note: currently only valid if kind == "user_object".
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SlotVariableReference slot_variables = 3;</code>
-     */
-    public java.util.List<org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference> getSlotVariablesList() {
-      if (slotVariablesBuilder_ == null) {
-        return java.util.Collections.unmodifiableList(slotVariables_);
-      } else {
-        return slotVariablesBuilder_.getMessageList();
-      }
-    }
-    /**
-     * <pre>
-     * Slot variables owned by this object. This describes the three-way
-     * (optimizer, variable, slot variable) relationship; none of the three
-     * depend on the others directly.
-     * Note: currently only valid if kind == "user_object".
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SlotVariableReference slot_variables = 3;</code>
-     */
-    public int getSlotVariablesCount() {
-      if (slotVariablesBuilder_ == null) {
-        return slotVariables_.size();
-      } else {
-        return slotVariablesBuilder_.getCount();
-      }
-    }
-    /**
-     * <pre>
-     * Slot variables owned by this object. This describes the three-way
-     * (optimizer, variable, slot variable) relationship; none of the three
-     * depend on the others directly.
-     * Note: currently only valid if kind == "user_object".
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SlotVariableReference slot_variables = 3;</code>
-     */
-    public org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference getSlotVariables(int index) {
-      if (slotVariablesBuilder_ == null) {
-        return slotVariables_.get(index);
-      } else {
-        return slotVariablesBuilder_.getMessage(index);
-      }
-    }
-    /**
-     * <pre>
-     * Slot variables owned by this object. This describes the three-way
-     * (optimizer, variable, slot variable) relationship; none of the three
-     * depend on the others directly.
-     * Note: currently only valid if kind == "user_object".
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SlotVariableReference slot_variables = 3;</code>
-     */
-    public Builder setSlotVariables(
-        int index, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference value) {
-      if (slotVariablesBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureSlotVariablesIsMutable();
-        slotVariables_.set(index, value);
-        onChanged();
-      } else {
-        slotVariablesBuilder_.setMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Slot variables owned by this object. This describes the three-way
-     * (optimizer, variable, slot variable) relationship; none of the three
-     * depend on the others directly.
-     * Note: currently only valid if kind == "user_object".
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SlotVariableReference slot_variables = 3;</code>
-     */
-    public Builder setSlotVariables(
-        int index, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference.Builder builderForValue) {
-      if (slotVariablesBuilder_ == null) {
-        ensureSlotVariablesIsMutable();
-        slotVariables_.set(index, builderForValue.build());
-        onChanged();
-      } else {
-        slotVariablesBuilder_.setMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Slot variables owned by this object. This describes the three-way
-     * (optimizer, variable, slot variable) relationship; none of the three
-     * depend on the others directly.
-     * Note: currently only valid if kind == "user_object".
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SlotVariableReference slot_variables = 3;</code>
-     */
-    public Builder addSlotVariables(org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference value) {
-      if (slotVariablesBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureSlotVariablesIsMutable();
-        slotVariables_.add(value);
-        onChanged();
-      } else {
-        slotVariablesBuilder_.addMessage(value);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Slot variables owned by this object. This describes the three-way
-     * (optimizer, variable, slot variable) relationship; none of the three
-     * depend on the others directly.
-     * Note: currently only valid if kind == "user_object".
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SlotVariableReference slot_variables = 3;</code>
-     */
-    public Builder addSlotVariables(
-        int index, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference value) {
-      if (slotVariablesBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureSlotVariablesIsMutable();
-        slotVariables_.add(index, value);
-        onChanged();
-      } else {
-        slotVariablesBuilder_.addMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Slot variables owned by this object. This describes the three-way
-     * (optimizer, variable, slot variable) relationship; none of the three
-     * depend on the others directly.
-     * Note: currently only valid if kind == "user_object".
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SlotVariableReference slot_variables = 3;</code>
-     */
-    public Builder addSlotVariables(
-        org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference.Builder builderForValue) {
-      if (slotVariablesBuilder_ == null) {
-        ensureSlotVariablesIsMutable();
-        slotVariables_.add(builderForValue.build());
-        onChanged();
-      } else {
-        slotVariablesBuilder_.addMessage(builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Slot variables owned by this object. This describes the three-way
-     * (optimizer, variable, slot variable) relationship; none of the three
-     * depend on the others directly.
-     * Note: currently only valid if kind == "user_object".
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SlotVariableReference slot_variables = 3;</code>
-     */
-    public Builder addSlotVariables(
-        int index, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference.Builder builderForValue) {
-      if (slotVariablesBuilder_ == null) {
-        ensureSlotVariablesIsMutable();
-        slotVariables_.add(index, builderForValue.build());
-        onChanged();
-      } else {
-        slotVariablesBuilder_.addMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Slot variables owned by this object. This describes the three-way
-     * (optimizer, variable, slot variable) relationship; none of the three
-     * depend on the others directly.
-     * Note: currently only valid if kind == "user_object".
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SlotVariableReference slot_variables = 3;</code>
-     */
-    public Builder addAllSlotVariables(
-        java.lang.Iterable<? extends org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference> values) {
-      if (slotVariablesBuilder_ == null) {
-        ensureSlotVariablesIsMutable();
-        com.google.protobuf.AbstractMessageLite.Builder.addAll(
-            values, slotVariables_);
-        onChanged();
-      } else {
-        slotVariablesBuilder_.addAllMessages(values);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Slot variables owned by this object. This describes the three-way
-     * (optimizer, variable, slot variable) relationship; none of the three
-     * depend on the others directly.
-     * Note: currently only valid if kind == "user_object".
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SlotVariableReference slot_variables = 3;</code>
-     */
-    public Builder clearSlotVariables() {
-      if (slotVariablesBuilder_ == null) {
-        slotVariables_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000002);
-        onChanged();
-      } else {
-        slotVariablesBuilder_.clear();
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Slot variables owned by this object. This describes the three-way
-     * (optimizer, variable, slot variable) relationship; none of the three
-     * depend on the others directly.
-     * Note: currently only valid if kind == "user_object".
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SlotVariableReference slot_variables = 3;</code>
-     */
-    public Builder removeSlotVariables(int index) {
-      if (slotVariablesBuilder_ == null) {
-        ensureSlotVariablesIsMutable();
-        slotVariables_.remove(index);
-        onChanged();
-      } else {
-        slotVariablesBuilder_.remove(index);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Slot variables owned by this object. This describes the three-way
-     * (optimizer, variable, slot variable) relationship; none of the three
-     * depend on the others directly.
-     * Note: currently only valid if kind == "user_object".
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SlotVariableReference slot_variables = 3;</code>
-     */
-    public org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference.Builder getSlotVariablesBuilder(
-        int index) {
-      return getSlotVariablesFieldBuilder().getBuilder(index);
-    }
-    /**
-     * <pre>
-     * Slot variables owned by this object. This describes the three-way
-     * (optimizer, variable, slot variable) relationship; none of the three
-     * depend on the others directly.
-     * Note: currently only valid if kind == "user_object".
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SlotVariableReference slot_variables = 3;</code>
-     */
-    public org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReferenceOrBuilder getSlotVariablesOrBuilder(
-        int index) {
-      if (slotVariablesBuilder_ == null) {
-        return slotVariables_.get(index);  } else {
-        return slotVariablesBuilder_.getMessageOrBuilder(index);
-      }
-    }
-    /**
-     * <pre>
-     * Slot variables owned by this object. This describes the three-way
-     * (optimizer, variable, slot variable) relationship; none of the three
-     * depend on the others directly.
-     * Note: currently only valid if kind == "user_object".
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SlotVariableReference slot_variables = 3;</code>
-     */
-    public java.util.List<? extends org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReferenceOrBuilder> 
-         getSlotVariablesOrBuilderList() {
-      if (slotVariablesBuilder_ != null) {
-        return slotVariablesBuilder_.getMessageOrBuilderList();
-      } else {
-        return java.util.Collections.unmodifiableList(slotVariables_);
-      }
-    }
-    /**
-     * <pre>
-     * Slot variables owned by this object. This describes the three-way
-     * (optimizer, variable, slot variable) relationship; none of the three
-     * depend on the others directly.
-     * Note: currently only valid if kind == "user_object".
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SlotVariableReference slot_variables = 3;</code>
-     */
-    public org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference.Builder addSlotVariablesBuilder() {
-      return getSlotVariablesFieldBuilder().addBuilder(
-          org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference.getDefaultInstance());
-    }
-    /**
-     * <pre>
-     * Slot variables owned by this object. This describes the three-way
-     * (optimizer, variable, slot variable) relationship; none of the three
-     * depend on the others directly.
-     * Note: currently only valid if kind == "user_object".
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SlotVariableReference slot_variables = 3;</code>
-     */
-    public org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference.Builder addSlotVariablesBuilder(
-        int index) {
-      return getSlotVariablesFieldBuilder().addBuilder(
-          index, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference.getDefaultInstance());
-    }
-    /**
-     * <pre>
-     * Slot variables owned by this object. This describes the three-way
-     * (optimizer, variable, slot variable) relationship; none of the three
-     * depend on the others directly.
-     * Note: currently only valid if kind == "user_object".
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SlotVariableReference slot_variables = 3;</code>
-     */
-    public java.util.List<org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference.Builder> 
-         getSlotVariablesBuilderList() {
-      return getSlotVariablesFieldBuilder().getBuilderList();
-    }
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference.Builder, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReferenceOrBuilder> 
-        getSlotVariablesFieldBuilder() {
-      if (slotVariablesBuilder_ == null) {
-        slotVariablesBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
-            org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference.Builder, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReferenceOrBuilder>(
-                slotVariables_,
-                ((bitField0_ & 0x00000002) != 0),
-                getParentForChildren(),
-                isClean());
-        slotVariables_ = null;
-      }
-      return slotVariablesBuilder_;
-    }
-
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.SavedUserObject, org.tensorflow.proto.framework.SavedUserObject.Builder, org.tensorflow.proto.framework.SavedUserObjectOrBuilder> userObjectBuilder_;
-    /**
-     * <code>.tensorflow.SavedUserObject user_object = 4;</code>
-     */
-    public boolean hasUserObject() {
-      return kindCase_ == 4;
-    }
-    /**
-     * <code>.tensorflow.SavedUserObject user_object = 4;</code>
-     */
-    public org.tensorflow.proto.framework.SavedUserObject getUserObject() {
-      if (userObjectBuilder_ == null) {
-        if (kindCase_ == 4) {
-          return (org.tensorflow.proto.framework.SavedUserObject) kind_;
-        }
-        return org.tensorflow.proto.framework.SavedUserObject.getDefaultInstance();
-      } else {
-        if (kindCase_ == 4) {
-          return userObjectBuilder_.getMessage();
-        }
-        return org.tensorflow.proto.framework.SavedUserObject.getDefaultInstance();
-      }
-    }
-    /**
-     * <code>.tensorflow.SavedUserObject user_object = 4;</code>
-     */
-    public Builder setUserObject(org.tensorflow.proto.framework.SavedUserObject value) {
-      if (userObjectBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        kind_ = value;
-        onChanged();
-      } else {
-        userObjectBuilder_.setMessage(value);
-      }
-      kindCase_ = 4;
-      return this;
-    }
-    /**
-     * <code>.tensorflow.SavedUserObject user_object = 4;</code>
-     */
-    public Builder setUserObject(
-        org.tensorflow.proto.framework.SavedUserObject.Builder builderForValue) {
-      if (userObjectBuilder_ == null) {
-        kind_ = builderForValue.build();
-        onChanged();
-      } else {
-        userObjectBuilder_.setMessage(builderForValue.build());
-      }
-      kindCase_ = 4;
-      return this;
-    }
-    /**
-     * <code>.tensorflow.SavedUserObject user_object = 4;</code>
-     */
-    public Builder mergeUserObject(org.tensorflow.proto.framework.SavedUserObject value) {
-      if (userObjectBuilder_ == null) {
-        if (kindCase_ == 4 &&
-            kind_ != org.tensorflow.proto.framework.SavedUserObject.getDefaultInstance()) {
-          kind_ = org.tensorflow.proto.framework.SavedUserObject.newBuilder((org.tensorflow.proto.framework.SavedUserObject) kind_)
-              .mergeFrom(value).buildPartial();
-        } else {
-          kind_ = value;
-        }
-        onChanged();
-      } else {
-        if (kindCase_ == 4) {
-          userObjectBuilder_.mergeFrom(value);
-        }
-        userObjectBuilder_.setMessage(value);
-      }
-      kindCase_ = 4;
-      return this;
-    }
-    /**
-     * <code>.tensorflow.SavedUserObject user_object = 4;</code>
-     */
-    public Builder clearUserObject() {
-      if (userObjectBuilder_ == null) {
-        if (kindCase_ == 4) {
-          kindCase_ = 0;
-          kind_ = null;
-          onChanged();
-        }
-      } else {
-        if (kindCase_ == 4) {
-          kindCase_ = 0;
-          kind_ = null;
-        }
-        userObjectBuilder_.clear();
-      }
-      return this;
-    }
-    /**
-     * <code>.tensorflow.SavedUserObject user_object = 4;</code>
-     */
-    public org.tensorflow.proto.framework.SavedUserObject.Builder getUserObjectBuilder() {
-      return getUserObjectFieldBuilder().getBuilder();
-    }
-    /**
-     * <code>.tensorflow.SavedUserObject user_object = 4;</code>
-     */
-    public org.tensorflow.proto.framework.SavedUserObjectOrBuilder getUserObjectOrBuilder() {
-      if ((kindCase_ == 4) && (userObjectBuilder_ != null)) {
-        return userObjectBuilder_.getMessageOrBuilder();
-      } else {
-        if (kindCase_ == 4) {
-          return (org.tensorflow.proto.framework.SavedUserObject) kind_;
-        }
-        return org.tensorflow.proto.framework.SavedUserObject.getDefaultInstance();
-      }
-    }
-    /**
-     * <code>.tensorflow.SavedUserObject user_object = 4;</code>
-     */
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.SavedUserObject, org.tensorflow.proto.framework.SavedUserObject.Builder, org.tensorflow.proto.framework.SavedUserObjectOrBuilder> 
-        getUserObjectFieldBuilder() {
-      if (userObjectBuilder_ == null) {
-        if (!(kindCase_ == 4)) {
-          kind_ = org.tensorflow.proto.framework.SavedUserObject.getDefaultInstance();
-        }
-        userObjectBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-            org.tensorflow.proto.framework.SavedUserObject, org.tensorflow.proto.framework.SavedUserObject.Builder, org.tensorflow.proto.framework.SavedUserObjectOrBuilder>(
-                (org.tensorflow.proto.framework.SavedUserObject) kind_,
-                getParentForChildren(),
-                isClean());
-        kind_ = null;
-      }
-      kindCase_ = 4;
-      onChanged();;
-      return userObjectBuilder_;
-    }
-
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.SavedAsset, org.tensorflow.proto.framework.SavedAsset.Builder, org.tensorflow.proto.framework.SavedAssetOrBuilder> assetBuilder_;
-    /**
-     * <code>.tensorflow.SavedAsset asset = 5;</code>
-     */
-    public boolean hasAsset() {
-      return kindCase_ == 5;
-    }
-    /**
-     * <code>.tensorflow.SavedAsset asset = 5;</code>
-     */
-    public org.tensorflow.proto.framework.SavedAsset getAsset() {
-      if (assetBuilder_ == null) {
-        if (kindCase_ == 5) {
-          return (org.tensorflow.proto.framework.SavedAsset) kind_;
-        }
-        return org.tensorflow.proto.framework.SavedAsset.getDefaultInstance();
-      } else {
-        if (kindCase_ == 5) {
-          return assetBuilder_.getMessage();
-        }
-        return org.tensorflow.proto.framework.SavedAsset.getDefaultInstance();
-      }
-    }
-    /**
-     * <code>.tensorflow.SavedAsset asset = 5;</code>
-     */
-    public Builder setAsset(org.tensorflow.proto.framework.SavedAsset value) {
-      if (assetBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        kind_ = value;
-        onChanged();
-      } else {
-        assetBuilder_.setMessage(value);
-      }
-      kindCase_ = 5;
-      return this;
-    }
-    /**
-     * <code>.tensorflow.SavedAsset asset = 5;</code>
-     */
-    public Builder setAsset(
-        org.tensorflow.proto.framework.SavedAsset.Builder builderForValue) {
-      if (assetBuilder_ == null) {
-        kind_ = builderForValue.build();
-        onChanged();
-      } else {
-        assetBuilder_.setMessage(builderForValue.build());
-      }
-      kindCase_ = 5;
-      return this;
-    }
-    /**
-     * <code>.tensorflow.SavedAsset asset = 5;</code>
-     */
-    public Builder mergeAsset(org.tensorflow.proto.framework.SavedAsset value) {
-      if (assetBuilder_ == null) {
-        if (kindCase_ == 5 &&
-            kind_ != org.tensorflow.proto.framework.SavedAsset.getDefaultInstance()) {
-          kind_ = org.tensorflow.proto.framework.SavedAsset.newBuilder((org.tensorflow.proto.framework.SavedAsset) kind_)
-              .mergeFrom(value).buildPartial();
-        } else {
-          kind_ = value;
-        }
-        onChanged();
-      } else {
-        if (kindCase_ == 5) {
-          assetBuilder_.mergeFrom(value);
-        }
-        assetBuilder_.setMessage(value);
-      }
-      kindCase_ = 5;
-      return this;
-    }
-    /**
-     * <code>.tensorflow.SavedAsset asset = 5;</code>
-     */
-    public Builder clearAsset() {
-      if (assetBuilder_ == null) {
-        if (kindCase_ == 5) {
-          kindCase_ = 0;
-          kind_ = null;
-          onChanged();
-        }
-      } else {
-        if (kindCase_ == 5) {
-          kindCase_ = 0;
-          kind_ = null;
-        }
-        assetBuilder_.clear();
-      }
-      return this;
-    }
-    /**
-     * <code>.tensorflow.SavedAsset asset = 5;</code>
-     */
-    public org.tensorflow.proto.framework.SavedAsset.Builder getAssetBuilder() {
-      return getAssetFieldBuilder().getBuilder();
-    }
-    /**
-     * <code>.tensorflow.SavedAsset asset = 5;</code>
-     */
-    public org.tensorflow.proto.framework.SavedAssetOrBuilder getAssetOrBuilder() {
-      if ((kindCase_ == 5) && (assetBuilder_ != null)) {
-        return assetBuilder_.getMessageOrBuilder();
-      } else {
-        if (kindCase_ == 5) {
-          return (org.tensorflow.proto.framework.SavedAsset) kind_;
-        }
-        return org.tensorflow.proto.framework.SavedAsset.getDefaultInstance();
-      }
-    }
-    /**
-     * <code>.tensorflow.SavedAsset asset = 5;</code>
-     */
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.SavedAsset, org.tensorflow.proto.framework.SavedAsset.Builder, org.tensorflow.proto.framework.SavedAssetOrBuilder> 
-        getAssetFieldBuilder() {
-      if (assetBuilder_ == null) {
-        if (!(kindCase_ == 5)) {
-          kind_ = org.tensorflow.proto.framework.SavedAsset.getDefaultInstance();
-        }
-        assetBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-            org.tensorflow.proto.framework.SavedAsset, org.tensorflow.proto.framework.SavedAsset.Builder, org.tensorflow.proto.framework.SavedAssetOrBuilder>(
-                (org.tensorflow.proto.framework.SavedAsset) kind_,
-                getParentForChildren(),
-                isClean());
-        kind_ = null;
-      }
-      kindCase_ = 5;
-      onChanged();;
-      return assetBuilder_;
-    }
-
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.SavedFunction, org.tensorflow.proto.framework.SavedFunction.Builder, org.tensorflow.proto.framework.SavedFunctionOrBuilder> functionBuilder_;
-    /**
-     * <code>.tensorflow.SavedFunction function = 6;</code>
-     */
-    public boolean hasFunction() {
-      return kindCase_ == 6;
-    }
-    /**
-     * <code>.tensorflow.SavedFunction function = 6;</code>
-     */
-    public org.tensorflow.proto.framework.SavedFunction getFunction() {
-      if (functionBuilder_ == null) {
-        if (kindCase_ == 6) {
-          return (org.tensorflow.proto.framework.SavedFunction) kind_;
-        }
-        return org.tensorflow.proto.framework.SavedFunction.getDefaultInstance();
-      } else {
-        if (kindCase_ == 6) {
-          return functionBuilder_.getMessage();
-        }
-        return org.tensorflow.proto.framework.SavedFunction.getDefaultInstance();
-      }
-    }
-    /**
-     * <code>.tensorflow.SavedFunction function = 6;</code>
-     */
-    public Builder setFunction(org.tensorflow.proto.framework.SavedFunction value) {
-      if (functionBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        kind_ = value;
-        onChanged();
-      } else {
-        functionBuilder_.setMessage(value);
-      }
-      kindCase_ = 6;
-      return this;
-    }
-    /**
-     * <code>.tensorflow.SavedFunction function = 6;</code>
-     */
-    public Builder setFunction(
-        org.tensorflow.proto.framework.SavedFunction.Builder builderForValue) {
-      if (functionBuilder_ == null) {
-        kind_ = builderForValue.build();
-        onChanged();
-      } else {
-        functionBuilder_.setMessage(builderForValue.build());
-      }
-      kindCase_ = 6;
-      return this;
-    }
-    /**
-     * <code>.tensorflow.SavedFunction function = 6;</code>
-     */
-    public Builder mergeFunction(org.tensorflow.proto.framework.SavedFunction value) {
-      if (functionBuilder_ == null) {
-        if (kindCase_ == 6 &&
-            kind_ != org.tensorflow.proto.framework.SavedFunction.getDefaultInstance()) {
-          kind_ = org.tensorflow.proto.framework.SavedFunction.newBuilder((org.tensorflow.proto.framework.SavedFunction) kind_)
-              .mergeFrom(value).buildPartial();
-        } else {
-          kind_ = value;
-        }
-        onChanged();
-      } else {
-        if (kindCase_ == 6) {
-          functionBuilder_.mergeFrom(value);
-        }
-        functionBuilder_.setMessage(value);
-      }
-      kindCase_ = 6;
-      return this;
-    }
-    /**
-     * <code>.tensorflow.SavedFunction function = 6;</code>
-     */
-    public Builder clearFunction() {
-      if (functionBuilder_ == null) {
-        if (kindCase_ == 6) {
-          kindCase_ = 0;
-          kind_ = null;
-          onChanged();
-        }
-      } else {
-        if (kindCase_ == 6) {
-          kindCase_ = 0;
-          kind_ = null;
-        }
-        functionBuilder_.clear();
-      }
-      return this;
-    }
-    /**
-     * <code>.tensorflow.SavedFunction function = 6;</code>
-     */
-    public org.tensorflow.proto.framework.SavedFunction.Builder getFunctionBuilder() {
-      return getFunctionFieldBuilder().getBuilder();
-    }
-    /**
-     * <code>.tensorflow.SavedFunction function = 6;</code>
-     */
-    public org.tensorflow.proto.framework.SavedFunctionOrBuilder getFunctionOrBuilder() {
-      if ((kindCase_ == 6) && (functionBuilder_ != null)) {
-        return functionBuilder_.getMessageOrBuilder();
-      } else {
-        if (kindCase_ == 6) {
-          return (org.tensorflow.proto.framework.SavedFunction) kind_;
-        }
-        return org.tensorflow.proto.framework.SavedFunction.getDefaultInstance();
-      }
-    }
-    /**
-     * <code>.tensorflow.SavedFunction function = 6;</code>
-     */
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.SavedFunction, org.tensorflow.proto.framework.SavedFunction.Builder, org.tensorflow.proto.framework.SavedFunctionOrBuilder> 
-        getFunctionFieldBuilder() {
-      if (functionBuilder_ == null) {
-        if (!(kindCase_ == 6)) {
-          kind_ = org.tensorflow.proto.framework.SavedFunction.getDefaultInstance();
-        }
-        functionBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-            org.tensorflow.proto.framework.SavedFunction, org.tensorflow.proto.framework.SavedFunction.Builder, org.tensorflow.proto.framework.SavedFunctionOrBuilder>(
-                (org.tensorflow.proto.framework.SavedFunction) kind_,
-                getParentForChildren(),
-                isClean());
-        kind_ = null;
-      }
-      kindCase_ = 6;
-      onChanged();;
-      return functionBuilder_;
-    }
-
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.SavedVariable, org.tensorflow.proto.framework.SavedVariable.Builder, org.tensorflow.proto.framework.SavedVariableOrBuilder> variableBuilder_;
-    /**
-     * <code>.tensorflow.SavedVariable variable = 7;</code>
-     */
-    public boolean hasVariable() {
-      return kindCase_ == 7;
-    }
-    /**
-     * <code>.tensorflow.SavedVariable variable = 7;</code>
-     */
-    public org.tensorflow.proto.framework.SavedVariable getVariable() {
-      if (variableBuilder_ == null) {
-        if (kindCase_ == 7) {
-          return (org.tensorflow.proto.framework.SavedVariable) kind_;
-        }
-        return org.tensorflow.proto.framework.SavedVariable.getDefaultInstance();
-      } else {
-        if (kindCase_ == 7) {
-          return variableBuilder_.getMessage();
-        }
-        return org.tensorflow.proto.framework.SavedVariable.getDefaultInstance();
-      }
-    }
-    /**
-     * <code>.tensorflow.SavedVariable variable = 7;</code>
-     */
-    public Builder setVariable(org.tensorflow.proto.framework.SavedVariable value) {
-      if (variableBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        kind_ = value;
-        onChanged();
-      } else {
-        variableBuilder_.setMessage(value);
-      }
-      kindCase_ = 7;
-      return this;
-    }
-    /**
-     * <code>.tensorflow.SavedVariable variable = 7;</code>
-     */
-    public Builder setVariable(
-        org.tensorflow.proto.framework.SavedVariable.Builder builderForValue) {
-      if (variableBuilder_ == null) {
-        kind_ = builderForValue.build();
-        onChanged();
-      } else {
-        variableBuilder_.setMessage(builderForValue.build());
-      }
-      kindCase_ = 7;
-      return this;
-    }
-    /**
-     * <code>.tensorflow.SavedVariable variable = 7;</code>
-     */
-    public Builder mergeVariable(org.tensorflow.proto.framework.SavedVariable value) {
-      if (variableBuilder_ == null) {
-        if (kindCase_ == 7 &&
-            kind_ != org.tensorflow.proto.framework.SavedVariable.getDefaultInstance()) {
-          kind_ = org.tensorflow.proto.framework.SavedVariable.newBuilder((org.tensorflow.proto.framework.SavedVariable) kind_)
-              .mergeFrom(value).buildPartial();
-        } else {
-          kind_ = value;
-        }
-        onChanged();
-      } else {
-        if (kindCase_ == 7) {
-          variableBuilder_.mergeFrom(value);
-        }
-        variableBuilder_.setMessage(value);
-      }
-      kindCase_ = 7;
-      return this;
-    }
-    /**
-     * <code>.tensorflow.SavedVariable variable = 7;</code>
-     */
-    public Builder clearVariable() {
-      if (variableBuilder_ == null) {
-        if (kindCase_ == 7) {
-          kindCase_ = 0;
-          kind_ = null;
-          onChanged();
-        }
-      } else {
-        if (kindCase_ == 7) {
-          kindCase_ = 0;
-          kind_ = null;
-        }
-        variableBuilder_.clear();
-      }
-      return this;
-    }
-    /**
-     * <code>.tensorflow.SavedVariable variable = 7;</code>
-     */
-    public org.tensorflow.proto.framework.SavedVariable.Builder getVariableBuilder() {
-      return getVariableFieldBuilder().getBuilder();
-    }
-    /**
-     * <code>.tensorflow.SavedVariable variable = 7;</code>
-     */
-    public org.tensorflow.proto.framework.SavedVariableOrBuilder getVariableOrBuilder() {
-      if ((kindCase_ == 7) && (variableBuilder_ != null)) {
-        return variableBuilder_.getMessageOrBuilder();
-      } else {
-        if (kindCase_ == 7) {
-          return (org.tensorflow.proto.framework.SavedVariable) kind_;
-        }
-        return org.tensorflow.proto.framework.SavedVariable.getDefaultInstance();
-      }
-    }
-    /**
-     * <code>.tensorflow.SavedVariable variable = 7;</code>
-     */
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.SavedVariable, org.tensorflow.proto.framework.SavedVariable.Builder, org.tensorflow.proto.framework.SavedVariableOrBuilder> 
-        getVariableFieldBuilder() {
-      if (variableBuilder_ == null) {
-        if (!(kindCase_ == 7)) {
-          kind_ = org.tensorflow.proto.framework.SavedVariable.getDefaultInstance();
-        }
-        variableBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-            org.tensorflow.proto.framework.SavedVariable, org.tensorflow.proto.framework.SavedVariable.Builder, org.tensorflow.proto.framework.SavedVariableOrBuilder>(
-                (org.tensorflow.proto.framework.SavedVariable) kind_,
-                getParentForChildren(),
-                isClean());
-        kind_ = null;
-      }
-      kindCase_ = 7;
-      onChanged();;
-      return variableBuilder_;
-    }
-
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.SavedBareConcreteFunction, org.tensorflow.proto.framework.SavedBareConcreteFunction.Builder, org.tensorflow.proto.framework.SavedBareConcreteFunctionOrBuilder> bareConcreteFunctionBuilder_;
-    /**
-     * <code>.tensorflow.SavedBareConcreteFunction bare_concrete_function = 8;</code>
-     */
-    public boolean hasBareConcreteFunction() {
-      return kindCase_ == 8;
-    }
-    /**
-     * <code>.tensorflow.SavedBareConcreteFunction bare_concrete_function = 8;</code>
-     */
-    public org.tensorflow.proto.framework.SavedBareConcreteFunction getBareConcreteFunction() {
-      if (bareConcreteFunctionBuilder_ == null) {
-        if (kindCase_ == 8) {
-          return (org.tensorflow.proto.framework.SavedBareConcreteFunction) kind_;
-        }
-        return org.tensorflow.proto.framework.SavedBareConcreteFunction.getDefaultInstance();
-      } else {
-        if (kindCase_ == 8) {
-          return bareConcreteFunctionBuilder_.getMessage();
-        }
-        return org.tensorflow.proto.framework.SavedBareConcreteFunction.getDefaultInstance();
-      }
-    }
-    /**
-     * <code>.tensorflow.SavedBareConcreteFunction bare_concrete_function = 8;</code>
-     */
-    public Builder setBareConcreteFunction(org.tensorflow.proto.framework.SavedBareConcreteFunction value) {
-      if (bareConcreteFunctionBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        kind_ = value;
-        onChanged();
-      } else {
-        bareConcreteFunctionBuilder_.setMessage(value);
-      }
-      kindCase_ = 8;
-      return this;
-    }
-    /**
-     * <code>.tensorflow.SavedBareConcreteFunction bare_concrete_function = 8;</code>
-     */
-    public Builder setBareConcreteFunction(
-        org.tensorflow.proto.framework.SavedBareConcreteFunction.Builder builderForValue) {
-      if (bareConcreteFunctionBuilder_ == null) {
-        kind_ = builderForValue.build();
-        onChanged();
-      } else {
-        bareConcreteFunctionBuilder_.setMessage(builderForValue.build());
-      }
-      kindCase_ = 8;
-      return this;
-    }
-    /**
-     * <code>.tensorflow.SavedBareConcreteFunction bare_concrete_function = 8;</code>
-     */
-    public Builder mergeBareConcreteFunction(org.tensorflow.proto.framework.SavedBareConcreteFunction value) {
-      if (bareConcreteFunctionBuilder_ == null) {
-        if (kindCase_ == 8 &&
-            kind_ != org.tensorflow.proto.framework.SavedBareConcreteFunction.getDefaultInstance()) {
-          kind_ = org.tensorflow.proto.framework.SavedBareConcreteFunction.newBuilder((org.tensorflow.proto.framework.SavedBareConcreteFunction) kind_)
-              .mergeFrom(value).buildPartial();
-        } else {
-          kind_ = value;
-        }
-        onChanged();
-      } else {
-        if (kindCase_ == 8) {
-          bareConcreteFunctionBuilder_.mergeFrom(value);
-        }
-        bareConcreteFunctionBuilder_.setMessage(value);
-      }
-      kindCase_ = 8;
-      return this;
-    }
-    /**
-     * <code>.tensorflow.SavedBareConcreteFunction bare_concrete_function = 8;</code>
-     */
-    public Builder clearBareConcreteFunction() {
-      if (bareConcreteFunctionBuilder_ == null) {
-        if (kindCase_ == 8) {
-          kindCase_ = 0;
-          kind_ = null;
-          onChanged();
-        }
-      } else {
-        if (kindCase_ == 8) {
-          kindCase_ = 0;
-          kind_ = null;
-        }
-        bareConcreteFunctionBuilder_.clear();
-      }
-      return this;
-    }
-    /**
-     * <code>.tensorflow.SavedBareConcreteFunction bare_concrete_function = 8;</code>
-     */
-    public org.tensorflow.proto.framework.SavedBareConcreteFunction.Builder getBareConcreteFunctionBuilder() {
-      return getBareConcreteFunctionFieldBuilder().getBuilder();
-    }
-    /**
-     * <code>.tensorflow.SavedBareConcreteFunction bare_concrete_function = 8;</code>
-     */
-    public org.tensorflow.proto.framework.SavedBareConcreteFunctionOrBuilder getBareConcreteFunctionOrBuilder() {
-      if ((kindCase_ == 8) && (bareConcreteFunctionBuilder_ != null)) {
-        return bareConcreteFunctionBuilder_.getMessageOrBuilder();
-      } else {
-        if (kindCase_ == 8) {
-          return (org.tensorflow.proto.framework.SavedBareConcreteFunction) kind_;
-        }
-        return org.tensorflow.proto.framework.SavedBareConcreteFunction.getDefaultInstance();
-      }
-    }
-    /**
-     * <code>.tensorflow.SavedBareConcreteFunction bare_concrete_function = 8;</code>
-     */
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.SavedBareConcreteFunction, org.tensorflow.proto.framework.SavedBareConcreteFunction.Builder, org.tensorflow.proto.framework.SavedBareConcreteFunctionOrBuilder> 
-        getBareConcreteFunctionFieldBuilder() {
-      if (bareConcreteFunctionBuilder_ == null) {
-        if (!(kindCase_ == 8)) {
-          kind_ = org.tensorflow.proto.framework.SavedBareConcreteFunction.getDefaultInstance();
-        }
-        bareConcreteFunctionBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-            org.tensorflow.proto.framework.SavedBareConcreteFunction, org.tensorflow.proto.framework.SavedBareConcreteFunction.Builder, org.tensorflow.proto.framework.SavedBareConcreteFunctionOrBuilder>(
-                (org.tensorflow.proto.framework.SavedBareConcreteFunction) kind_,
-                getParentForChildren(),
-                isClean());
-        kind_ = null;
-      }
-      kindCase_ = 8;
-      onChanged();;
-      return bareConcreteFunctionBuilder_;
-    }
-
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.SavedConstant, org.tensorflow.proto.framework.SavedConstant.Builder, org.tensorflow.proto.framework.SavedConstantOrBuilder> constantBuilder_;
-    /**
-     * <code>.tensorflow.SavedConstant constant = 9;</code>
-     */
-    public boolean hasConstant() {
-      return kindCase_ == 9;
-    }
-    /**
-     * <code>.tensorflow.SavedConstant constant = 9;</code>
-     */
-    public org.tensorflow.proto.framework.SavedConstant getConstant() {
-      if (constantBuilder_ == null) {
-        if (kindCase_ == 9) {
-          return (org.tensorflow.proto.framework.SavedConstant) kind_;
-        }
-        return org.tensorflow.proto.framework.SavedConstant.getDefaultInstance();
-      } else {
-        if (kindCase_ == 9) {
-          return constantBuilder_.getMessage();
-        }
-        return org.tensorflow.proto.framework.SavedConstant.getDefaultInstance();
-      }
-    }
-    /**
-     * <code>.tensorflow.SavedConstant constant = 9;</code>
-     */
-    public Builder setConstant(org.tensorflow.proto.framework.SavedConstant value) {
-      if (constantBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        kind_ = value;
-        onChanged();
-      } else {
-        constantBuilder_.setMessage(value);
-      }
-      kindCase_ = 9;
-      return this;
-    }
-    /**
-     * <code>.tensorflow.SavedConstant constant = 9;</code>
-     */
-    public Builder setConstant(
-        org.tensorflow.proto.framework.SavedConstant.Builder builderForValue) {
-      if (constantBuilder_ == null) {
-        kind_ = builderForValue.build();
-        onChanged();
-      } else {
-        constantBuilder_.setMessage(builderForValue.build());
-      }
-      kindCase_ = 9;
-      return this;
-    }
-    /**
-     * <code>.tensorflow.SavedConstant constant = 9;</code>
-     */
-    public Builder mergeConstant(org.tensorflow.proto.framework.SavedConstant value) {
-      if (constantBuilder_ == null) {
-        if (kindCase_ == 9 &&
-            kind_ != org.tensorflow.proto.framework.SavedConstant.getDefaultInstance()) {
-          kind_ = org.tensorflow.proto.framework.SavedConstant.newBuilder((org.tensorflow.proto.framework.SavedConstant) kind_)
-              .mergeFrom(value).buildPartial();
-        } else {
-          kind_ = value;
-        }
-        onChanged();
-      } else {
-        if (kindCase_ == 9) {
-          constantBuilder_.mergeFrom(value);
-        }
-        constantBuilder_.setMessage(value);
-      }
-      kindCase_ = 9;
-      return this;
-    }
-    /**
-     * <code>.tensorflow.SavedConstant constant = 9;</code>
-     */
-    public Builder clearConstant() {
-      if (constantBuilder_ == null) {
-        if (kindCase_ == 9) {
-          kindCase_ = 0;
-          kind_ = null;
-          onChanged();
-        }
-      } else {
-        if (kindCase_ == 9) {
-          kindCase_ = 0;
-          kind_ = null;
-        }
-        constantBuilder_.clear();
-      }
-      return this;
-    }
-    /**
-     * <code>.tensorflow.SavedConstant constant = 9;</code>
-     */
-    public org.tensorflow.proto.framework.SavedConstant.Builder getConstantBuilder() {
-      return getConstantFieldBuilder().getBuilder();
-    }
-    /**
-     * <code>.tensorflow.SavedConstant constant = 9;</code>
-     */
-    public org.tensorflow.proto.framework.SavedConstantOrBuilder getConstantOrBuilder() {
-      if ((kindCase_ == 9) && (constantBuilder_ != null)) {
-        return constantBuilder_.getMessageOrBuilder();
-      } else {
-        if (kindCase_ == 9) {
-          return (org.tensorflow.proto.framework.SavedConstant) kind_;
-        }
-        return org.tensorflow.proto.framework.SavedConstant.getDefaultInstance();
-      }
-    }
-    /**
-     * <code>.tensorflow.SavedConstant constant = 9;</code>
-     */
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.SavedConstant, org.tensorflow.proto.framework.SavedConstant.Builder, org.tensorflow.proto.framework.SavedConstantOrBuilder> 
-        getConstantFieldBuilder() {
-      if (constantBuilder_ == null) {
-        if (!(kindCase_ == 9)) {
-          kind_ = org.tensorflow.proto.framework.SavedConstant.getDefaultInstance();
-        }
-        constantBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-            org.tensorflow.proto.framework.SavedConstant, org.tensorflow.proto.framework.SavedConstant.Builder, org.tensorflow.proto.framework.SavedConstantOrBuilder>(
-                (org.tensorflow.proto.framework.SavedConstant) kind_,
-                getParentForChildren(),
-                isClean());
-        kind_ = null;
-      }
-      kindCase_ = 9;
-      onChanged();;
-      return constantBuilder_;
-    }
-
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.SavedResource, org.tensorflow.proto.framework.SavedResource.Builder, org.tensorflow.proto.framework.SavedResourceOrBuilder> resourceBuilder_;
-    /**
-     * <code>.tensorflow.SavedResource resource = 10;</code>
-     */
-    public boolean hasResource() {
-      return kindCase_ == 10;
-    }
-    /**
-     * <code>.tensorflow.SavedResource resource = 10;</code>
-     */
-    public org.tensorflow.proto.framework.SavedResource getResource() {
-      if (resourceBuilder_ == null) {
-        if (kindCase_ == 10) {
-          return (org.tensorflow.proto.framework.SavedResource) kind_;
-        }
-        return org.tensorflow.proto.framework.SavedResource.getDefaultInstance();
-      } else {
-        if (kindCase_ == 10) {
-          return resourceBuilder_.getMessage();
-        }
-        return org.tensorflow.proto.framework.SavedResource.getDefaultInstance();
-      }
-    }
-    /**
-     * <code>.tensorflow.SavedResource resource = 10;</code>
-     */
-    public Builder setResource(org.tensorflow.proto.framework.SavedResource value) {
-      if (resourceBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        kind_ = value;
-        onChanged();
-      } else {
-        resourceBuilder_.setMessage(value);
-      }
-      kindCase_ = 10;
-      return this;
-    }
-    /**
-     * <code>.tensorflow.SavedResource resource = 10;</code>
-     */
-    public Builder setResource(
-        org.tensorflow.proto.framework.SavedResource.Builder builderForValue) {
-      if (resourceBuilder_ == null) {
-        kind_ = builderForValue.build();
-        onChanged();
-      } else {
-        resourceBuilder_.setMessage(builderForValue.build());
-      }
-      kindCase_ = 10;
-      return this;
-    }
-    /**
-     * <code>.tensorflow.SavedResource resource = 10;</code>
-     */
-    public Builder mergeResource(org.tensorflow.proto.framework.SavedResource value) {
-      if (resourceBuilder_ == null) {
-        if (kindCase_ == 10 &&
-            kind_ != org.tensorflow.proto.framework.SavedResource.getDefaultInstance()) {
-          kind_ = org.tensorflow.proto.framework.SavedResource.newBuilder((org.tensorflow.proto.framework.SavedResource) kind_)
-              .mergeFrom(value).buildPartial();
-        } else {
-          kind_ = value;
-        }
-        onChanged();
-      } else {
-        if (kindCase_ == 10) {
-          resourceBuilder_.mergeFrom(value);
-        }
-        resourceBuilder_.setMessage(value);
-      }
-      kindCase_ = 10;
-      return this;
-    }
-    /**
-     * <code>.tensorflow.SavedResource resource = 10;</code>
-     */
-    public Builder clearResource() {
-      if (resourceBuilder_ == null) {
-        if (kindCase_ == 10) {
-          kindCase_ = 0;
-          kind_ = null;
-          onChanged();
-        }
-      } else {
-        if (kindCase_ == 10) {
-          kindCase_ = 0;
-          kind_ = null;
-        }
-        resourceBuilder_.clear();
-      }
-      return this;
-    }
-    /**
-     * <code>.tensorflow.SavedResource resource = 10;</code>
-     */
-    public org.tensorflow.proto.framework.SavedResource.Builder getResourceBuilder() {
-      return getResourceFieldBuilder().getBuilder();
-    }
-    /**
-     * <code>.tensorflow.SavedResource resource = 10;</code>
-     */
-    public org.tensorflow.proto.framework.SavedResourceOrBuilder getResourceOrBuilder() {
-      if ((kindCase_ == 10) && (resourceBuilder_ != null)) {
-        return resourceBuilder_.getMessageOrBuilder();
-      } else {
-        if (kindCase_ == 10) {
-          return (org.tensorflow.proto.framework.SavedResource) kind_;
-        }
-        return org.tensorflow.proto.framework.SavedResource.getDefaultInstance();
-      }
-    }
-    /**
-     * <code>.tensorflow.SavedResource resource = 10;</code>
-     */
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.SavedResource, org.tensorflow.proto.framework.SavedResource.Builder, org.tensorflow.proto.framework.SavedResourceOrBuilder> 
-        getResourceFieldBuilder() {
-      if (resourceBuilder_ == null) {
-        if (!(kindCase_ == 10)) {
-          kind_ = org.tensorflow.proto.framework.SavedResource.getDefaultInstance();
-        }
-        resourceBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-            org.tensorflow.proto.framework.SavedResource, org.tensorflow.proto.framework.SavedResource.Builder, org.tensorflow.proto.framework.SavedResourceOrBuilder>(
-                (org.tensorflow.proto.framework.SavedResource) kind_,
-                getParentForChildren(),
-                isClean());
-        kind_ = null;
-      }
-      kindCase_ = 10;
-      onChanged();;
-      return resourceBuilder_;
-    }
-
-    private com.google.protobuf.MapField<
-        java.lang.String, org.tensorflow.proto.framework.SaveableObject> saveableObjects_;
-    private com.google.protobuf.MapField<java.lang.String, org.tensorflow.proto.framework.SaveableObject>
-    internalGetSaveableObjects() {
-      if (saveableObjects_ == null) {
-        return com.google.protobuf.MapField.emptyMapField(
-            SaveableObjectsDefaultEntryHolder.defaultEntry);
-      }
-      return saveableObjects_;
-    }
-    private com.google.protobuf.MapField<java.lang.String, org.tensorflow.proto.framework.SaveableObject>
-    internalGetMutableSaveableObjects() {
-      onChanged();;
-      if (saveableObjects_ == null) {
-        saveableObjects_ = com.google.protobuf.MapField.newMapField(
-            SaveableObjectsDefaultEntryHolder.defaultEntry);
-      }
-      if (!saveableObjects_.isMutable()) {
-        saveableObjects_ = saveableObjects_.copy();
-      }
-      return saveableObjects_;
-    }
-
-    public int getSaveableObjectsCount() {
-      return internalGetSaveableObjects().getMap().size();
-    }
-    /**
-     * <code>map&lt;string, .tensorflow.SaveableObject&gt; saveable_objects = 11;</code>
-     */
-
-    public boolean containsSaveableObjects(
-        java.lang.String key) {
-      if (key == null) { throw new java.lang.NullPointerException(); }
-      return internalGetSaveableObjects().getMap().containsKey(key);
-    }
-    /**
-     * Use {@link #getSaveableObjectsMap()} instead.
-     */
-    @java.lang.Deprecated
-    public java.util.Map<java.lang.String, org.tensorflow.proto.framework.SaveableObject> getSaveableObjects() {
-      return getSaveableObjectsMap();
-    }
-    /**
-     * <code>map&lt;string, .tensorflow.SaveableObject&gt; saveable_objects = 11;</code>
-     */
-
-    public java.util.Map<java.lang.String, org.tensorflow.proto.framework.SaveableObject> getSaveableObjectsMap() {
-      return internalGetSaveableObjects().getMap();
-    }
-    /**
-     * <code>map&lt;string, .tensorflow.SaveableObject&gt; saveable_objects = 11;</code>
-     */
-
-    public org.tensorflow.proto.framework.SaveableObject getSaveableObjectsOrDefault(
-        java.lang.String key,
-        org.tensorflow.proto.framework.SaveableObject defaultValue) {
-      if (key == null) { throw new java.lang.NullPointerException(); }
-      java.util.Map<java.lang.String, org.tensorflow.proto.framework.SaveableObject> map =
-          internalGetSaveableObjects().getMap();
-      return map.containsKey(key) ? map.get(key) : defaultValue;
-    }
-    /**
-     * <code>map&lt;string, .tensorflow.SaveableObject&gt; saveable_objects = 11;</code>
-     */
-
-    public org.tensorflow.proto.framework.SaveableObject getSaveableObjectsOrThrow(
-        java.lang.String key) {
-      if (key == null) { throw new java.lang.NullPointerException(); }
-      java.util.Map<java.lang.String, org.tensorflow.proto.framework.SaveableObject> map =
-          internalGetSaveableObjects().getMap();
-      if (!map.containsKey(key)) {
-        throw new java.lang.IllegalArgumentException();
-      }
-      return map.get(key);
-    }
-
-    public Builder clearSaveableObjects() {
-      internalGetMutableSaveableObjects().getMutableMap()
-          .clear();
-      return this;
-    }
-    /**
-     * <code>map&lt;string, .tensorflow.SaveableObject&gt; saveable_objects = 11;</code>
-     */
-
-    public Builder removeSaveableObjects(
-        java.lang.String key) {
-      if (key == null) { throw new java.lang.NullPointerException(); }
-      internalGetMutableSaveableObjects().getMutableMap()
-          .remove(key);
-      return this;
-    }
-    /**
-     * Use alternate mutation accessors instead.
-     */
-    @java.lang.Deprecated
-    public java.util.Map<java.lang.String, org.tensorflow.proto.framework.SaveableObject>
-    getMutableSaveableObjects() {
-      return internalGetMutableSaveableObjects().getMutableMap();
-    }
-    /**
-     * <code>map&lt;string, .tensorflow.SaveableObject&gt; saveable_objects = 11;</code>
-     */
-    public Builder putSaveableObjects(
-        java.lang.String key,
-        org.tensorflow.proto.framework.SaveableObject value) {
-      if (key == null) { throw new java.lang.NullPointerException(); }
-      if (value == null) { throw new java.lang.NullPointerException(); }
-      internalGetMutableSaveableObjects().getMutableMap()
-          .put(key, value);
-      return this;
-    }
-    /**
-     * <code>map&lt;string, .tensorflow.SaveableObject&gt; saveable_objects = 11;</code>
-     */
-
-    public Builder putAllSaveableObjects(
-        java.util.Map<java.lang.String, org.tensorflow.proto.framework.SaveableObject> values) {
-      internalGetMutableSaveableObjects().getMutableMap()
-          .putAll(values);
-      return this;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.SavedObject)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.SavedObject)
-  private static final org.tensorflow.proto.framework.SavedObject DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.SavedObject();
-  }
-
-  public static org.tensorflow.proto.framework.SavedObject getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<SavedObject>
-      PARSER = new com.google.protobuf.AbstractParser<SavedObject>() {
-    @java.lang.Override
-    public SavedObject parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new SavedObject(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<SavedObject> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<SavedObject> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.framework.SavedObject getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedObjectGraph.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedObjectGraph.java
deleted file mode 100644
index 40a1e497db4..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedObjectGraph.java
+++ /dev/null
@@ -1,1231 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/saved_object_graph.proto
-
-package org.tensorflow.proto.framework;
-
-/**
- * Protobuf type {@code tensorflow.SavedObjectGraph}
- */
-public  final class SavedObjectGraph extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.SavedObjectGraph)
-    SavedObjectGraphOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use SavedObjectGraph.newBuilder() to construct.
-  private SavedObjectGraph(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private SavedObjectGraph() {
-    nodes_ = java.util.Collections.emptyList();
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new SavedObjectGraph();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private SavedObjectGraph(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    int mutable_bitField0_ = 0;
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 10: {
-            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
-              nodes_ = new java.util.ArrayList<org.tensorflow.proto.framework.SavedObject>();
-              mutable_bitField0_ |= 0x00000001;
-            }
-            nodes_.add(
-                input.readMessage(org.tensorflow.proto.framework.SavedObject.parser(), extensionRegistry));
-            break;
-          }
-          case 18: {
-            if (!((mutable_bitField0_ & 0x00000002) != 0)) {
-              concreteFunctions_ = com.google.protobuf.MapField.newMapField(
-                  ConcreteFunctionsDefaultEntryHolder.defaultEntry);
-              mutable_bitField0_ |= 0x00000002;
-            }
-            com.google.protobuf.MapEntry<java.lang.String, org.tensorflow.proto.framework.SavedConcreteFunction>
-            concreteFunctions__ = input.readMessage(
-                ConcreteFunctionsDefaultEntryHolder.defaultEntry.getParserForType(), extensionRegistry);
-            concreteFunctions_.getMutableMap().put(
-                concreteFunctions__.getKey(), concreteFunctions__.getValue());
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      if (((mutable_bitField0_ & 0x00000001) != 0)) {
-        nodes_ = java.util.Collections.unmodifiableList(nodes_);
-      }
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.framework.SavedObjectGraphProtos.internal_static_tensorflow_SavedObjectGraph_descriptor;
-  }
-
-  @SuppressWarnings({"rawtypes"})
-  @java.lang.Override
-  protected com.google.protobuf.MapField internalGetMapField(
-      int number) {
-    switch (number) {
-      case 2:
-        return internalGetConcreteFunctions();
-      default:
-        throw new RuntimeException(
-            "Invalid map field number: " + number);
-    }
-  }
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.framework.SavedObjectGraphProtos.internal_static_tensorflow_SavedObjectGraph_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.framework.SavedObjectGraph.class, org.tensorflow.proto.framework.SavedObjectGraph.Builder.class);
-  }
-
-  public static final int NODES_FIELD_NUMBER = 1;
-  private java.util.List<org.tensorflow.proto.framework.SavedObject> nodes_;
-  /**
-   * <pre>
-   * Flattened list of objects in the object graph.
-   * The position of the object in this list indicates its id.
-   * Nodes[0] is considered the root node.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.SavedObject nodes = 1;</code>
-   */
-  public java.util.List<org.tensorflow.proto.framework.SavedObject> getNodesList() {
-    return nodes_;
-  }
-  /**
-   * <pre>
-   * Flattened list of objects in the object graph.
-   * The position of the object in this list indicates its id.
-   * Nodes[0] is considered the root node.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.SavedObject nodes = 1;</code>
-   */
-  public java.util.List<? extends org.tensorflow.proto.framework.SavedObjectOrBuilder> 
-      getNodesOrBuilderList() {
-    return nodes_;
-  }
-  /**
-   * <pre>
-   * Flattened list of objects in the object graph.
-   * The position of the object in this list indicates its id.
-   * Nodes[0] is considered the root node.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.SavedObject nodes = 1;</code>
-   */
-  public int getNodesCount() {
-    return nodes_.size();
-  }
-  /**
-   * <pre>
-   * Flattened list of objects in the object graph.
-   * The position of the object in this list indicates its id.
-   * Nodes[0] is considered the root node.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.SavedObject nodes = 1;</code>
-   */
-  public org.tensorflow.proto.framework.SavedObject getNodes(int index) {
-    return nodes_.get(index);
-  }
-  /**
-   * <pre>
-   * Flattened list of objects in the object graph.
-   * The position of the object in this list indicates its id.
-   * Nodes[0] is considered the root node.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.SavedObject nodes = 1;</code>
-   */
-  public org.tensorflow.proto.framework.SavedObjectOrBuilder getNodesOrBuilder(
-      int index) {
-    return nodes_.get(index);
-  }
-
-  public static final int CONCRETE_FUNCTIONS_FIELD_NUMBER = 2;
-  private static final class ConcreteFunctionsDefaultEntryHolder {
-    static final com.google.protobuf.MapEntry<
-        java.lang.String, org.tensorflow.proto.framework.SavedConcreteFunction> defaultEntry =
-            com.google.protobuf.MapEntry
-            .<java.lang.String, org.tensorflow.proto.framework.SavedConcreteFunction>newDefaultInstance(
-                org.tensorflow.proto.framework.SavedObjectGraphProtos.internal_static_tensorflow_SavedObjectGraph_ConcreteFunctionsEntry_descriptor, 
-                com.google.protobuf.WireFormat.FieldType.STRING,
-                "",
-                com.google.protobuf.WireFormat.FieldType.MESSAGE,
-                org.tensorflow.proto.framework.SavedConcreteFunction.getDefaultInstance());
-  }
-  private com.google.protobuf.MapField<
-      java.lang.String, org.tensorflow.proto.framework.SavedConcreteFunction> concreteFunctions_;
-  private com.google.protobuf.MapField<java.lang.String, org.tensorflow.proto.framework.SavedConcreteFunction>
-  internalGetConcreteFunctions() {
-    if (concreteFunctions_ == null) {
-      return com.google.protobuf.MapField.emptyMapField(
-          ConcreteFunctionsDefaultEntryHolder.defaultEntry);
-    }
-    return concreteFunctions_;
-  }
-
-  public int getConcreteFunctionsCount() {
-    return internalGetConcreteFunctions().getMap().size();
-  }
-  /**
-   * <pre>
-   * Information about captures and output structures in concrete functions.
-   * Referenced from SavedBareConcreteFunction and SavedFunction.
-   * </pre>
-   *
-   * <code>map&lt;string, .tensorflow.SavedConcreteFunction&gt; concrete_functions = 2;</code>
-   */
-
-  public boolean containsConcreteFunctions(
-      java.lang.String key) {
-    if (key == null) { throw new java.lang.NullPointerException(); }
-    return internalGetConcreteFunctions().getMap().containsKey(key);
-  }
-  /**
-   * Use {@link #getConcreteFunctionsMap()} instead.
-   */
-  @java.lang.Deprecated
-  public java.util.Map<java.lang.String, org.tensorflow.proto.framework.SavedConcreteFunction> getConcreteFunctions() {
-    return getConcreteFunctionsMap();
-  }
-  /**
-   * <pre>
-   * Information about captures and output structures in concrete functions.
-   * Referenced from SavedBareConcreteFunction and SavedFunction.
-   * </pre>
-   *
-   * <code>map&lt;string, .tensorflow.SavedConcreteFunction&gt; concrete_functions = 2;</code>
-   */
-
-  public java.util.Map<java.lang.String, org.tensorflow.proto.framework.SavedConcreteFunction> getConcreteFunctionsMap() {
-    return internalGetConcreteFunctions().getMap();
-  }
-  /**
-   * <pre>
-   * Information about captures and output structures in concrete functions.
-   * Referenced from SavedBareConcreteFunction and SavedFunction.
-   * </pre>
-   *
-   * <code>map&lt;string, .tensorflow.SavedConcreteFunction&gt; concrete_functions = 2;</code>
-   */
-
-  public org.tensorflow.proto.framework.SavedConcreteFunction getConcreteFunctionsOrDefault(
-      java.lang.String key,
-      org.tensorflow.proto.framework.SavedConcreteFunction defaultValue) {
-    if (key == null) { throw new java.lang.NullPointerException(); }
-    java.util.Map<java.lang.String, org.tensorflow.proto.framework.SavedConcreteFunction> map =
-        internalGetConcreteFunctions().getMap();
-    return map.containsKey(key) ? map.get(key) : defaultValue;
-  }
-  /**
-   * <pre>
-   * Information about captures and output structures in concrete functions.
-   * Referenced from SavedBareConcreteFunction and SavedFunction.
-   * </pre>
-   *
-   * <code>map&lt;string, .tensorflow.SavedConcreteFunction&gt; concrete_functions = 2;</code>
-   */
-
-  public org.tensorflow.proto.framework.SavedConcreteFunction getConcreteFunctionsOrThrow(
-      java.lang.String key) {
-    if (key == null) { throw new java.lang.NullPointerException(); }
-    java.util.Map<java.lang.String, org.tensorflow.proto.framework.SavedConcreteFunction> map =
-        internalGetConcreteFunctions().getMap();
-    if (!map.containsKey(key)) {
-      throw new java.lang.IllegalArgumentException();
-    }
-    return map.get(key);
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    for (int i = 0; i < nodes_.size(); i++) {
-      output.writeMessage(1, nodes_.get(i));
-    }
-    com.google.protobuf.GeneratedMessageV3
-      .serializeStringMapTo(
-        output,
-        internalGetConcreteFunctions(),
-        ConcreteFunctionsDefaultEntryHolder.defaultEntry,
-        2);
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    for (int i = 0; i < nodes_.size(); i++) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(1, nodes_.get(i));
-    }
-    for (java.util.Map.Entry<java.lang.String, org.tensorflow.proto.framework.SavedConcreteFunction> entry
-         : internalGetConcreteFunctions().getMap().entrySet()) {
-      com.google.protobuf.MapEntry<java.lang.String, org.tensorflow.proto.framework.SavedConcreteFunction>
-      concreteFunctions__ = ConcreteFunctionsDefaultEntryHolder.defaultEntry.newBuilderForType()
-          .setKey(entry.getKey())
-          .setValue(entry.getValue())
-          .build();
-      size += com.google.protobuf.CodedOutputStream
-          .computeMessageSize(2, concreteFunctions__);
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.framework.SavedObjectGraph)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.framework.SavedObjectGraph other = (org.tensorflow.proto.framework.SavedObjectGraph) obj;
-
-    if (!getNodesList()
-        .equals(other.getNodesList())) return false;
-    if (!internalGetConcreteFunctions().equals(
-        other.internalGetConcreteFunctions())) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    if (getNodesCount() > 0) {
-      hash = (37 * hash) + NODES_FIELD_NUMBER;
-      hash = (53 * hash) + getNodesList().hashCode();
-    }
-    if (!internalGetConcreteFunctions().getMap().isEmpty()) {
-      hash = (37 * hash) + CONCRETE_FUNCTIONS_FIELD_NUMBER;
-      hash = (53 * hash) + internalGetConcreteFunctions().hashCode();
-    }
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.framework.SavedObjectGraph parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.SavedObjectGraph parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.SavedObjectGraph parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.SavedObjectGraph parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.SavedObjectGraph parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.SavedObjectGraph parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.SavedObjectGraph parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.SavedObjectGraph parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.SavedObjectGraph parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.SavedObjectGraph parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.SavedObjectGraph parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.SavedObjectGraph parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.framework.SavedObjectGraph prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * Protobuf type {@code tensorflow.SavedObjectGraph}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.SavedObjectGraph)
-      org.tensorflow.proto.framework.SavedObjectGraphOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.SavedObjectGraphProtos.internal_static_tensorflow_SavedObjectGraph_descriptor;
-    }
-
-    @SuppressWarnings({"rawtypes"})
-    protected com.google.protobuf.MapField internalGetMapField(
-        int number) {
-      switch (number) {
-        case 2:
-          return internalGetConcreteFunctions();
-        default:
-          throw new RuntimeException(
-              "Invalid map field number: " + number);
-      }
-    }
-    @SuppressWarnings({"rawtypes"})
-    protected com.google.protobuf.MapField internalGetMutableMapField(
-        int number) {
-      switch (number) {
-        case 2:
-          return internalGetMutableConcreteFunctions();
-        default:
-          throw new RuntimeException(
-              "Invalid map field number: " + number);
-      }
-    }
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.SavedObjectGraphProtos.internal_static_tensorflow_SavedObjectGraph_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.SavedObjectGraph.class, org.tensorflow.proto.framework.SavedObjectGraph.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.framework.SavedObjectGraph.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-        getNodesFieldBuilder();
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      if (nodesBuilder_ == null) {
-        nodes_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000001);
-      } else {
-        nodesBuilder_.clear();
-      }
-      internalGetMutableConcreteFunctions().clear();
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.framework.SavedObjectGraphProtos.internal_static_tensorflow_SavedObjectGraph_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.SavedObjectGraph getDefaultInstanceForType() {
-      return org.tensorflow.proto.framework.SavedObjectGraph.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.SavedObjectGraph build() {
-      org.tensorflow.proto.framework.SavedObjectGraph result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.SavedObjectGraph buildPartial() {
-      org.tensorflow.proto.framework.SavedObjectGraph result = new org.tensorflow.proto.framework.SavedObjectGraph(this);
-      int from_bitField0_ = bitField0_;
-      if (nodesBuilder_ == null) {
-        if (((bitField0_ & 0x00000001) != 0)) {
-          nodes_ = java.util.Collections.unmodifiableList(nodes_);
-          bitField0_ = (bitField0_ & ~0x00000001);
-        }
-        result.nodes_ = nodes_;
-      } else {
-        result.nodes_ = nodesBuilder_.build();
-      }
-      result.concreteFunctions_ = internalGetConcreteFunctions();
-      result.concreteFunctions_.makeImmutable();
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.framework.SavedObjectGraph) {
-        return mergeFrom((org.tensorflow.proto.framework.SavedObjectGraph)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.framework.SavedObjectGraph other) {
-      if (other == org.tensorflow.proto.framework.SavedObjectGraph.getDefaultInstance()) return this;
-      if (nodesBuilder_ == null) {
-        if (!other.nodes_.isEmpty()) {
-          if (nodes_.isEmpty()) {
-            nodes_ = other.nodes_;
-            bitField0_ = (bitField0_ & ~0x00000001);
-          } else {
-            ensureNodesIsMutable();
-            nodes_.addAll(other.nodes_);
-          }
-          onChanged();
-        }
-      } else {
-        if (!other.nodes_.isEmpty()) {
-          if (nodesBuilder_.isEmpty()) {
-            nodesBuilder_.dispose();
-            nodesBuilder_ = null;
-            nodes_ = other.nodes_;
-            bitField0_ = (bitField0_ & ~0x00000001);
-            nodesBuilder_ = 
-              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
-                 getNodesFieldBuilder() : null;
-          } else {
-            nodesBuilder_.addAllMessages(other.nodes_);
-          }
-        }
-      }
-      internalGetMutableConcreteFunctions().mergeFrom(
-          other.internalGetConcreteFunctions());
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.framework.SavedObjectGraph parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.framework.SavedObjectGraph) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-    private int bitField0_;
-
-    private java.util.List<org.tensorflow.proto.framework.SavedObject> nodes_ =
-      java.util.Collections.emptyList();
-    private void ensureNodesIsMutable() {
-      if (!((bitField0_ & 0x00000001) != 0)) {
-        nodes_ = new java.util.ArrayList<org.tensorflow.proto.framework.SavedObject>(nodes_);
-        bitField0_ |= 0x00000001;
-       }
-    }
-
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.framework.SavedObject, org.tensorflow.proto.framework.SavedObject.Builder, org.tensorflow.proto.framework.SavedObjectOrBuilder> nodesBuilder_;
-
-    /**
-     * <pre>
-     * Flattened list of objects in the object graph.
-     * The position of the object in this list indicates its id.
-     * Nodes[0] is considered the root node.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.SavedObject nodes = 1;</code>
-     */
-    public java.util.List<org.tensorflow.proto.framework.SavedObject> getNodesList() {
-      if (nodesBuilder_ == null) {
-        return java.util.Collections.unmodifiableList(nodes_);
-      } else {
-        return nodesBuilder_.getMessageList();
-      }
-    }
-    /**
-     * <pre>
-     * Flattened list of objects in the object graph.
-     * The position of the object in this list indicates its id.
-     * Nodes[0] is considered the root node.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.SavedObject nodes = 1;</code>
-     */
-    public int getNodesCount() {
-      if (nodesBuilder_ == null) {
-        return nodes_.size();
-      } else {
-        return nodesBuilder_.getCount();
-      }
-    }
-    /**
-     * <pre>
-     * Flattened list of objects in the object graph.
-     * The position of the object in this list indicates its id.
-     * Nodes[0] is considered the root node.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.SavedObject nodes = 1;</code>
-     */
-    public org.tensorflow.proto.framework.SavedObject getNodes(int index) {
-      if (nodesBuilder_ == null) {
-        return nodes_.get(index);
-      } else {
-        return nodesBuilder_.getMessage(index);
-      }
-    }
-    /**
-     * <pre>
-     * Flattened list of objects in the object graph.
-     * The position of the object in this list indicates its id.
-     * Nodes[0] is considered the root node.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.SavedObject nodes = 1;</code>
-     */
-    public Builder setNodes(
-        int index, org.tensorflow.proto.framework.SavedObject value) {
-      if (nodesBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureNodesIsMutable();
-        nodes_.set(index, value);
-        onChanged();
-      } else {
-        nodesBuilder_.setMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Flattened list of objects in the object graph.
-     * The position of the object in this list indicates its id.
-     * Nodes[0] is considered the root node.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.SavedObject nodes = 1;</code>
-     */
-    public Builder setNodes(
-        int index, org.tensorflow.proto.framework.SavedObject.Builder builderForValue) {
-      if (nodesBuilder_ == null) {
-        ensureNodesIsMutable();
-        nodes_.set(index, builderForValue.build());
-        onChanged();
-      } else {
-        nodesBuilder_.setMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Flattened list of objects in the object graph.
-     * The position of the object in this list indicates its id.
-     * Nodes[0] is considered the root node.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.SavedObject nodes = 1;</code>
-     */
-    public Builder addNodes(org.tensorflow.proto.framework.SavedObject value) {
-      if (nodesBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureNodesIsMutable();
-        nodes_.add(value);
-        onChanged();
-      } else {
-        nodesBuilder_.addMessage(value);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Flattened list of objects in the object graph.
-     * The position of the object in this list indicates its id.
-     * Nodes[0] is considered the root node.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.SavedObject nodes = 1;</code>
-     */
-    public Builder addNodes(
-        int index, org.tensorflow.proto.framework.SavedObject value) {
-      if (nodesBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureNodesIsMutable();
-        nodes_.add(index, value);
-        onChanged();
-      } else {
-        nodesBuilder_.addMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Flattened list of objects in the object graph.
-     * The position of the object in this list indicates its id.
-     * Nodes[0] is considered the root node.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.SavedObject nodes = 1;</code>
-     */
-    public Builder addNodes(
-        org.tensorflow.proto.framework.SavedObject.Builder builderForValue) {
-      if (nodesBuilder_ == null) {
-        ensureNodesIsMutable();
-        nodes_.add(builderForValue.build());
-        onChanged();
-      } else {
-        nodesBuilder_.addMessage(builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Flattened list of objects in the object graph.
-     * The position of the object in this list indicates its id.
-     * Nodes[0] is considered the root node.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.SavedObject nodes = 1;</code>
-     */
-    public Builder addNodes(
-        int index, org.tensorflow.proto.framework.SavedObject.Builder builderForValue) {
-      if (nodesBuilder_ == null) {
-        ensureNodesIsMutable();
-        nodes_.add(index, builderForValue.build());
-        onChanged();
-      } else {
-        nodesBuilder_.addMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Flattened list of objects in the object graph.
-     * The position of the object in this list indicates its id.
-     * Nodes[0] is considered the root node.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.SavedObject nodes = 1;</code>
-     */
-    public Builder addAllNodes(
-        java.lang.Iterable<? extends org.tensorflow.proto.framework.SavedObject> values) {
-      if (nodesBuilder_ == null) {
-        ensureNodesIsMutable();
-        com.google.protobuf.AbstractMessageLite.Builder.addAll(
-            values, nodes_);
-        onChanged();
-      } else {
-        nodesBuilder_.addAllMessages(values);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Flattened list of objects in the object graph.
-     * The position of the object in this list indicates its id.
-     * Nodes[0] is considered the root node.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.SavedObject nodes = 1;</code>
-     */
-    public Builder clearNodes() {
-      if (nodesBuilder_ == null) {
-        nodes_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000001);
-        onChanged();
-      } else {
-        nodesBuilder_.clear();
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Flattened list of objects in the object graph.
-     * The position of the object in this list indicates its id.
-     * Nodes[0] is considered the root node.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.SavedObject nodes = 1;</code>
-     */
-    public Builder removeNodes(int index) {
-      if (nodesBuilder_ == null) {
-        ensureNodesIsMutable();
-        nodes_.remove(index);
-        onChanged();
-      } else {
-        nodesBuilder_.remove(index);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Flattened list of objects in the object graph.
-     * The position of the object in this list indicates its id.
-     * Nodes[0] is considered the root node.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.SavedObject nodes = 1;</code>
-     */
-    public org.tensorflow.proto.framework.SavedObject.Builder getNodesBuilder(
-        int index) {
-      return getNodesFieldBuilder().getBuilder(index);
-    }
-    /**
-     * <pre>
-     * Flattened list of objects in the object graph.
-     * The position of the object in this list indicates its id.
-     * Nodes[0] is considered the root node.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.SavedObject nodes = 1;</code>
-     */
-    public org.tensorflow.proto.framework.SavedObjectOrBuilder getNodesOrBuilder(
-        int index) {
-      if (nodesBuilder_ == null) {
-        return nodes_.get(index);  } else {
-        return nodesBuilder_.getMessageOrBuilder(index);
-      }
-    }
-    /**
-     * <pre>
-     * Flattened list of objects in the object graph.
-     * The position of the object in this list indicates its id.
-     * Nodes[0] is considered the root node.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.SavedObject nodes = 1;</code>
-     */
-    public java.util.List<? extends org.tensorflow.proto.framework.SavedObjectOrBuilder> 
-         getNodesOrBuilderList() {
-      if (nodesBuilder_ != null) {
-        return nodesBuilder_.getMessageOrBuilderList();
-      } else {
-        return java.util.Collections.unmodifiableList(nodes_);
-      }
-    }
-    /**
-     * <pre>
-     * Flattened list of objects in the object graph.
-     * The position of the object in this list indicates its id.
-     * Nodes[0] is considered the root node.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.SavedObject nodes = 1;</code>
-     */
-    public org.tensorflow.proto.framework.SavedObject.Builder addNodesBuilder() {
-      return getNodesFieldBuilder().addBuilder(
-          org.tensorflow.proto.framework.SavedObject.getDefaultInstance());
-    }
-    /**
-     * <pre>
-     * Flattened list of objects in the object graph.
-     * The position of the object in this list indicates its id.
-     * Nodes[0] is considered the root node.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.SavedObject nodes = 1;</code>
-     */
-    public org.tensorflow.proto.framework.SavedObject.Builder addNodesBuilder(
-        int index) {
-      return getNodesFieldBuilder().addBuilder(
-          index, org.tensorflow.proto.framework.SavedObject.getDefaultInstance());
-    }
-    /**
-     * <pre>
-     * Flattened list of objects in the object graph.
-     * The position of the object in this list indicates its id.
-     * Nodes[0] is considered the root node.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.SavedObject nodes = 1;</code>
-     */
-    public java.util.List<org.tensorflow.proto.framework.SavedObject.Builder> 
-         getNodesBuilderList() {
-      return getNodesFieldBuilder().getBuilderList();
-    }
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.framework.SavedObject, org.tensorflow.proto.framework.SavedObject.Builder, org.tensorflow.proto.framework.SavedObjectOrBuilder> 
-        getNodesFieldBuilder() {
-      if (nodesBuilder_ == null) {
-        nodesBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
-            org.tensorflow.proto.framework.SavedObject, org.tensorflow.proto.framework.SavedObject.Builder, org.tensorflow.proto.framework.SavedObjectOrBuilder>(
-                nodes_,
-                ((bitField0_ & 0x00000001) != 0),
-                getParentForChildren(),
-                isClean());
-        nodes_ = null;
-      }
-      return nodesBuilder_;
-    }
-
-    private com.google.protobuf.MapField<
-        java.lang.String, org.tensorflow.proto.framework.SavedConcreteFunction> concreteFunctions_;
-    private com.google.protobuf.MapField<java.lang.String, org.tensorflow.proto.framework.SavedConcreteFunction>
-    internalGetConcreteFunctions() {
-      if (concreteFunctions_ == null) {
-        return com.google.protobuf.MapField.emptyMapField(
-            ConcreteFunctionsDefaultEntryHolder.defaultEntry);
-      }
-      return concreteFunctions_;
-    }
-    private com.google.protobuf.MapField<java.lang.String, org.tensorflow.proto.framework.SavedConcreteFunction>
-    internalGetMutableConcreteFunctions() {
-      onChanged();;
-      if (concreteFunctions_ == null) {
-        concreteFunctions_ = com.google.protobuf.MapField.newMapField(
-            ConcreteFunctionsDefaultEntryHolder.defaultEntry);
-      }
-      if (!concreteFunctions_.isMutable()) {
-        concreteFunctions_ = concreteFunctions_.copy();
-      }
-      return concreteFunctions_;
-    }
-
-    public int getConcreteFunctionsCount() {
-      return internalGetConcreteFunctions().getMap().size();
-    }
-    /**
-     * <pre>
-     * Information about captures and output structures in concrete functions.
-     * Referenced from SavedBareConcreteFunction and SavedFunction.
-     * </pre>
-     *
-     * <code>map&lt;string, .tensorflow.SavedConcreteFunction&gt; concrete_functions = 2;</code>
-     */
-
-    public boolean containsConcreteFunctions(
-        java.lang.String key) {
-      if (key == null) { throw new java.lang.NullPointerException(); }
-      return internalGetConcreteFunctions().getMap().containsKey(key);
-    }
-    /**
-     * Use {@link #getConcreteFunctionsMap()} instead.
-     */
-    @java.lang.Deprecated
-    public java.util.Map<java.lang.String, org.tensorflow.proto.framework.SavedConcreteFunction> getConcreteFunctions() {
-      return getConcreteFunctionsMap();
-    }
-    /**
-     * <pre>
-     * Information about captures and output structures in concrete functions.
-     * Referenced from SavedBareConcreteFunction and SavedFunction.
-     * </pre>
-     *
-     * <code>map&lt;string, .tensorflow.SavedConcreteFunction&gt; concrete_functions = 2;</code>
-     */
-
-    public java.util.Map<java.lang.String, org.tensorflow.proto.framework.SavedConcreteFunction> getConcreteFunctionsMap() {
-      return internalGetConcreteFunctions().getMap();
-    }
-    /**
-     * <pre>
-     * Information about captures and output structures in concrete functions.
-     * Referenced from SavedBareConcreteFunction and SavedFunction.
-     * </pre>
-     *
-     * <code>map&lt;string, .tensorflow.SavedConcreteFunction&gt; concrete_functions = 2;</code>
-     */
-
-    public org.tensorflow.proto.framework.SavedConcreteFunction getConcreteFunctionsOrDefault(
-        java.lang.String key,
-        org.tensorflow.proto.framework.SavedConcreteFunction defaultValue) {
-      if (key == null) { throw new java.lang.NullPointerException(); }
-      java.util.Map<java.lang.String, org.tensorflow.proto.framework.SavedConcreteFunction> map =
-          internalGetConcreteFunctions().getMap();
-      return map.containsKey(key) ? map.get(key) : defaultValue;
-    }
-    /**
-     * <pre>
-     * Information about captures and output structures in concrete functions.
-     * Referenced from SavedBareConcreteFunction and SavedFunction.
-     * </pre>
-     *
-     * <code>map&lt;string, .tensorflow.SavedConcreteFunction&gt; concrete_functions = 2;</code>
-     */
-
-    public org.tensorflow.proto.framework.SavedConcreteFunction getConcreteFunctionsOrThrow(
-        java.lang.String key) {
-      if (key == null) { throw new java.lang.NullPointerException(); }
-      java.util.Map<java.lang.String, org.tensorflow.proto.framework.SavedConcreteFunction> map =
-          internalGetConcreteFunctions().getMap();
-      if (!map.containsKey(key)) {
-        throw new java.lang.IllegalArgumentException();
-      }
-      return map.get(key);
-    }
-
-    public Builder clearConcreteFunctions() {
-      internalGetMutableConcreteFunctions().getMutableMap()
-          .clear();
-      return this;
-    }
-    /**
-     * <pre>
-     * Information about captures and output structures in concrete functions.
-     * Referenced from SavedBareConcreteFunction and SavedFunction.
-     * </pre>
-     *
-     * <code>map&lt;string, .tensorflow.SavedConcreteFunction&gt; concrete_functions = 2;</code>
-     */
-
-    public Builder removeConcreteFunctions(
-        java.lang.String key) {
-      if (key == null) { throw new java.lang.NullPointerException(); }
-      internalGetMutableConcreteFunctions().getMutableMap()
-          .remove(key);
-      return this;
-    }
-    /**
-     * Use alternate mutation accessors instead.
-     */
-    @java.lang.Deprecated
-    public java.util.Map<java.lang.String, org.tensorflow.proto.framework.SavedConcreteFunction>
-    getMutableConcreteFunctions() {
-      return internalGetMutableConcreteFunctions().getMutableMap();
-    }
-    /**
-     * <pre>
-     * Information about captures and output structures in concrete functions.
-     * Referenced from SavedBareConcreteFunction and SavedFunction.
-     * </pre>
-     *
-     * <code>map&lt;string, .tensorflow.SavedConcreteFunction&gt; concrete_functions = 2;</code>
-     */
-    public Builder putConcreteFunctions(
-        java.lang.String key,
-        org.tensorflow.proto.framework.SavedConcreteFunction value) {
-      if (key == null) { throw new java.lang.NullPointerException(); }
-      if (value == null) { throw new java.lang.NullPointerException(); }
-      internalGetMutableConcreteFunctions().getMutableMap()
-          .put(key, value);
-      return this;
-    }
-    /**
-     * <pre>
-     * Information about captures and output structures in concrete functions.
-     * Referenced from SavedBareConcreteFunction and SavedFunction.
-     * </pre>
-     *
-     * <code>map&lt;string, .tensorflow.SavedConcreteFunction&gt; concrete_functions = 2;</code>
-     */
-
-    public Builder putAllConcreteFunctions(
-        java.util.Map<java.lang.String, org.tensorflow.proto.framework.SavedConcreteFunction> values) {
-      internalGetMutableConcreteFunctions().getMutableMap()
-          .putAll(values);
-      return this;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.SavedObjectGraph)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.SavedObjectGraph)
-  private static final org.tensorflow.proto.framework.SavedObjectGraph DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.SavedObjectGraph();
-  }
-
-  public static org.tensorflow.proto.framework.SavedObjectGraph getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<SavedObjectGraph>
-      PARSER = new com.google.protobuf.AbstractParser<SavedObjectGraph>() {
-    @java.lang.Override
-    public SavedObjectGraph parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new SavedObjectGraph(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<SavedObjectGraph> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<SavedObjectGraph> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.framework.SavedObjectGraph getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedObjectGraphOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedObjectGraphOrBuilder.java
deleted file mode 100644
index 3d80fbc1306..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedObjectGraphOrBuilder.java
+++ /dev/null
@@ -1,122 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/saved_object_graph.proto
-
-package org.tensorflow.proto.framework;
-
-public interface SavedObjectGraphOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.SavedObjectGraph)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <pre>
-   * Flattened list of objects in the object graph.
-   * The position of the object in this list indicates its id.
-   * Nodes[0] is considered the root node.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.SavedObject nodes = 1;</code>
-   */
-  java.util.List<org.tensorflow.proto.framework.SavedObject> 
-      getNodesList();
-  /**
-   * <pre>
-   * Flattened list of objects in the object graph.
-   * The position of the object in this list indicates its id.
-   * Nodes[0] is considered the root node.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.SavedObject nodes = 1;</code>
-   */
-  org.tensorflow.proto.framework.SavedObject getNodes(int index);
-  /**
-   * <pre>
-   * Flattened list of objects in the object graph.
-   * The position of the object in this list indicates its id.
-   * Nodes[0] is considered the root node.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.SavedObject nodes = 1;</code>
-   */
-  int getNodesCount();
-  /**
-   * <pre>
-   * Flattened list of objects in the object graph.
-   * The position of the object in this list indicates its id.
-   * Nodes[0] is considered the root node.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.SavedObject nodes = 1;</code>
-   */
-  java.util.List<? extends org.tensorflow.proto.framework.SavedObjectOrBuilder> 
-      getNodesOrBuilderList();
-  /**
-   * <pre>
-   * Flattened list of objects in the object graph.
-   * The position of the object in this list indicates its id.
-   * Nodes[0] is considered the root node.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.SavedObject nodes = 1;</code>
-   */
-  org.tensorflow.proto.framework.SavedObjectOrBuilder getNodesOrBuilder(
-      int index);
-
-  /**
-   * <pre>
-   * Information about captures and output structures in concrete functions.
-   * Referenced from SavedBareConcreteFunction and SavedFunction.
-   * </pre>
-   *
-   * <code>map&lt;string, .tensorflow.SavedConcreteFunction&gt; concrete_functions = 2;</code>
-   */
-  int getConcreteFunctionsCount();
-  /**
-   * <pre>
-   * Information about captures and output structures in concrete functions.
-   * Referenced from SavedBareConcreteFunction and SavedFunction.
-   * </pre>
-   *
-   * <code>map&lt;string, .tensorflow.SavedConcreteFunction&gt; concrete_functions = 2;</code>
-   */
-  boolean containsConcreteFunctions(
-      java.lang.String key);
-  /**
-   * Use {@link #getConcreteFunctionsMap()} instead.
-   */
-  @java.lang.Deprecated
-  java.util.Map<java.lang.String, org.tensorflow.proto.framework.SavedConcreteFunction>
-  getConcreteFunctions();
-  /**
-   * <pre>
-   * Information about captures and output structures in concrete functions.
-   * Referenced from SavedBareConcreteFunction and SavedFunction.
-   * </pre>
-   *
-   * <code>map&lt;string, .tensorflow.SavedConcreteFunction&gt; concrete_functions = 2;</code>
-   */
-  java.util.Map<java.lang.String, org.tensorflow.proto.framework.SavedConcreteFunction>
-  getConcreteFunctionsMap();
-  /**
-   * <pre>
-   * Information about captures and output structures in concrete functions.
-   * Referenced from SavedBareConcreteFunction and SavedFunction.
-   * </pre>
-   *
-   * <code>map&lt;string, .tensorflow.SavedConcreteFunction&gt; concrete_functions = 2;</code>
-   */
-
-  org.tensorflow.proto.framework.SavedConcreteFunction getConcreteFunctionsOrDefault(
-      java.lang.String key,
-      org.tensorflow.proto.framework.SavedConcreteFunction defaultValue);
-  /**
-   * <pre>
-   * Information about captures and output structures in concrete functions.
-   * Referenced from SavedBareConcreteFunction and SavedFunction.
-   * </pre>
-   *
-   * <code>map&lt;string, .tensorflow.SavedConcreteFunction&gt; concrete_functions = 2;</code>
-   */
-
-  org.tensorflow.proto.framework.SavedConcreteFunction getConcreteFunctionsOrThrow(
-      java.lang.String key);
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedObjectGraphProtos.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedObjectGraphProtos.java
deleted file mode 100644
index a7baea2d222..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedObjectGraphProtos.java
+++ /dev/null
@@ -1,263 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/saved_object_graph.proto
-
-package org.tensorflow.proto.framework;
-
-public final class SavedObjectGraphProtos {
-  private SavedObjectGraphProtos() {}
-  public static void registerAllExtensions(
-      com.google.protobuf.ExtensionRegistryLite registry) {
-  }
-
-  public static void registerAllExtensions(
-      com.google.protobuf.ExtensionRegistry registry) {
-    registerAllExtensions(
-        (com.google.protobuf.ExtensionRegistryLite) registry);
-  }
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_SavedObjectGraph_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_SavedObjectGraph_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_SavedObjectGraph_ConcreteFunctionsEntry_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_SavedObjectGraph_ConcreteFunctionsEntry_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_SavedObject_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_SavedObject_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_SavedObject_SaveableObjectsEntry_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_SavedObject_SaveableObjectsEntry_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_SavedUserObject_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_SavedUserObject_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_SavedAsset_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_SavedAsset_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_SavedFunction_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_SavedFunction_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_SavedConcreteFunction_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_SavedConcreteFunction_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_SavedBareConcreteFunction_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_SavedBareConcreteFunction_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_SavedConstant_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_SavedConstant_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_SavedVariable_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_SavedVariable_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_FunctionSpec_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_FunctionSpec_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_SavedResource_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_SavedResource_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_SaveableObject_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_SaveableObject_fieldAccessorTable;
-
-  public static com.google.protobuf.Descriptors.FileDescriptor
-      getDescriptor() {
-    return descriptor;
-  }
-  private static  com.google.protobuf.Descriptors.FileDescriptor
-      descriptor;
-  static {
-    java.lang.String[] descriptorData = {
-      "\n1tensorflow/core/protobuf/saved_object_" +
-      "graph.proto\022\ntensorflow\032,tensorflow/core" +
-      "/framework/tensor_shape.proto\032%tensorflo" +
-      "w/core/framework/types.proto\032(tensorflow" +
-      "/core/framework/variable.proto\032(tensorfl" +
-      "ow/core/framework/versions.proto\032%tensor" +
-      "flow/core/protobuf/struct.proto\0325tensorf" +
-      "low/core/protobuf/trackable_object_graph" +
-      ".proto\"\350\001\n\020SavedObjectGraph\022&\n\005nodes\030\001 \003" +
-      "(\0132\027.tensorflow.SavedObject\022O\n\022concrete_" +
-      "functions\030\002 \003(\01323.tensorflow.SavedObject" +
-      "Graph.ConcreteFunctionsEntry\032[\n\026Concrete" +
-      "FunctionsEntry\022\013\n\003key\030\001 \001(\t\0220\n\005value\030\002 \001" +
-      "(\0132!.tensorflow.SavedConcreteFunction:\0028" +
-      "\001\"\331\005\n\013SavedObject\022R\n\010children\030\001 \003(\0132@.te" +
-      "nsorflow.TrackableObjectGraph.TrackableO" +
-      "bject.ObjectReference\022^\n\016slot_variables\030" +
-      "\003 \003(\0132F.tensorflow.TrackableObjectGraph." +
-      "TrackableObject.SlotVariableReference\0222\n" +
-      "\013user_object\030\004 \001(\0132\033.tensorflow.SavedUse" +
-      "rObjectH\000\022\'\n\005asset\030\005 \001(\0132\026.tensorflow.Sa" +
-      "vedAssetH\000\022-\n\010function\030\006 \001(\0132\031.tensorflo" +
-      "w.SavedFunctionH\000\022-\n\010variable\030\007 \001(\0132\031.te" +
-      "nsorflow.SavedVariableH\000\022G\n\026bare_concret" +
-      "e_function\030\010 \001(\0132%.tensorflow.SavedBareC" +
-      "oncreteFunctionH\000\022-\n\010constant\030\t \001(\0132\031.te" +
-      "nsorflow.SavedConstantH\000\022-\n\010resource\030\n \001" +
-      "(\0132\031.tensorflow.SavedResourceH\000\022F\n\020savea" +
-      "ble_objects\030\013 \003(\0132,.tensorflow.SavedObje" +
-      "ct.SaveableObjectsEntry\032R\n\024SaveableObjec" +
-      "tsEntry\022\013\n\003key\030\001 \001(\t\022)\n\005value\030\002 \001(\0132\032.te" +
-      "nsorflow.SaveableObject:\0028\001B\006\n\004kindJ\004\010\002\020" +
-      "\003R\nattributes\"`\n\017SavedUserObject\022\022\n\niden" +
-      "tifier\030\001 \001(\t\022\'\n\007version\030\002 \001(\0132\026.tensorfl" +
-      "ow.VersionDef\022\020\n\010metadata\030\003 \001(\t\"*\n\nSaved" +
-      "Asset\022\034\n\024asset_file_def_index\030\001 \001(\005\"\\\n\rS" +
-      "avedFunction\022\032\n\022concrete_functions\030\001 \003(\t" +
-      "\022/\n\rfunction_spec\030\002 \001(\0132\030.tensorflow.Fun" +
-      "ctionSpec\"\250\001\n\025SavedConcreteFunction\022\024\n\014b" +
-      "ound_inputs\030\002 \003(\005\022B\n\035canonicalized_input" +
-      "_signature\030\003 \001(\0132\033.tensorflow.Structured" +
-      "Value\0225\n\020output_signature\030\004 \001(\0132\033.tensor" +
-      "flow.StructuredValue\"|\n\031SavedBareConcret" +
-      "eFunction\022\036\n\026concrete_function_name\030\001 \001(" +
-      "\t\022\031\n\021argument_keywords\030\002 \003(\t\022$\n\034allowed_" +
-      "positional_arguments\030\003 \001(\003\"\"\n\rSavedConst" +
-      "ant\022\021\n\toperation\030\001 \001(\t\"\366\001\n\rSavedVariable" +
-      "\022#\n\005dtype\030\001 \001(\0162\024.tensorflow.DataType\022+\n" +
-      "\005shape\030\002 \001(\0132\034.tensorflow.TensorShapePro" +
-      "to\022\021\n\ttrainable\030\003 \001(\010\022<\n\017synchronization" +
-      "\030\004 \001(\0162#.tensorflow.VariableSynchronizat" +
-      "ion\0224\n\013aggregation\030\005 \001(\0162\037.tensorflow.Va" +
-      "riableAggregation\022\014\n\004name\030\006 \001(\t\"\225\001\n\014Func" +
-      "tionSpec\0220\n\013fullargspec\030\001 \001(\0132\033.tensorfl" +
-      "ow.StructuredValue\022\021\n\tis_method\030\002 \001(\010\0224\n" +
-      "\017input_signature\030\005 \001(\0132\033.tensorflow.Stru" +
-      "cturedValueJ\004\010\003\020\004J\004\010\004\020\005\"\037\n\rSavedResource" +
-      "\022\016\n\006device\030\001 \001(\t\"A\n\016SaveableObject\022\025\n\rsa" +
-      "ve_function\030\002 \001(\005\022\030\n\020restore_function\030\003 " +
-      "\001(\005B\207\001\n\036org.tensorflow.proto.frameworkB\026" +
-      "SavedObjectGraphProtosP\001ZHgithub.com/ten" +
-      "sorflow/tensorflow/tensorflow/go/core/co" +
-      "re_protos_go_proto\370\001\001b\006proto3"
-    };
-    descriptor = com.google.protobuf.Descriptors.FileDescriptor
-      .internalBuildGeneratedFileFrom(descriptorData,
-        new com.google.protobuf.Descriptors.FileDescriptor[] {
-          org.tensorflow.proto.framework.TensorShapeProtos.getDescriptor(),
-          org.tensorflow.proto.framework.TypesProtos.getDescriptor(),
-          org.tensorflow.proto.framework.VariableProtos.getDescriptor(),
-          org.tensorflow.proto.framework.VersionsProtos.getDescriptor(),
-          org.tensorflow.proto.framework.StructProtos.getDescriptor(),
-          org.tensorflow.proto.framework.TrackableObjectGraphProtos.getDescriptor(),
-        });
-    internal_static_tensorflow_SavedObjectGraph_descriptor =
-      getDescriptor().getMessageTypes().get(0);
-    internal_static_tensorflow_SavedObjectGraph_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_SavedObjectGraph_descriptor,
-        new java.lang.String[] { "Nodes", "ConcreteFunctions", });
-    internal_static_tensorflow_SavedObjectGraph_ConcreteFunctionsEntry_descriptor =
-      internal_static_tensorflow_SavedObjectGraph_descriptor.getNestedTypes().get(0);
-    internal_static_tensorflow_SavedObjectGraph_ConcreteFunctionsEntry_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_SavedObjectGraph_ConcreteFunctionsEntry_descriptor,
-        new java.lang.String[] { "Key", "Value", });
-    internal_static_tensorflow_SavedObject_descriptor =
-      getDescriptor().getMessageTypes().get(1);
-    internal_static_tensorflow_SavedObject_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_SavedObject_descriptor,
-        new java.lang.String[] { "Children", "SlotVariables", "UserObject", "Asset", "Function", "Variable", "BareConcreteFunction", "Constant", "Resource", "SaveableObjects", "Kind", });
-    internal_static_tensorflow_SavedObject_SaveableObjectsEntry_descriptor =
-      internal_static_tensorflow_SavedObject_descriptor.getNestedTypes().get(0);
-    internal_static_tensorflow_SavedObject_SaveableObjectsEntry_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_SavedObject_SaveableObjectsEntry_descriptor,
-        new java.lang.String[] { "Key", "Value", });
-    internal_static_tensorflow_SavedUserObject_descriptor =
-      getDescriptor().getMessageTypes().get(2);
-    internal_static_tensorflow_SavedUserObject_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_SavedUserObject_descriptor,
-        new java.lang.String[] { "Identifier", "Version", "Metadata", });
-    internal_static_tensorflow_SavedAsset_descriptor =
-      getDescriptor().getMessageTypes().get(3);
-    internal_static_tensorflow_SavedAsset_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_SavedAsset_descriptor,
-        new java.lang.String[] { "AssetFileDefIndex", });
-    internal_static_tensorflow_SavedFunction_descriptor =
-      getDescriptor().getMessageTypes().get(4);
-    internal_static_tensorflow_SavedFunction_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_SavedFunction_descriptor,
-        new java.lang.String[] { "ConcreteFunctions", "FunctionSpec", });
-    internal_static_tensorflow_SavedConcreteFunction_descriptor =
-      getDescriptor().getMessageTypes().get(5);
-    internal_static_tensorflow_SavedConcreteFunction_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_SavedConcreteFunction_descriptor,
-        new java.lang.String[] { "BoundInputs", "CanonicalizedInputSignature", "OutputSignature", });
-    internal_static_tensorflow_SavedBareConcreteFunction_descriptor =
-      getDescriptor().getMessageTypes().get(6);
-    internal_static_tensorflow_SavedBareConcreteFunction_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_SavedBareConcreteFunction_descriptor,
-        new java.lang.String[] { "ConcreteFunctionName", "ArgumentKeywords", "AllowedPositionalArguments", });
-    internal_static_tensorflow_SavedConstant_descriptor =
-      getDescriptor().getMessageTypes().get(7);
-    internal_static_tensorflow_SavedConstant_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_SavedConstant_descriptor,
-        new java.lang.String[] { "Operation", });
-    internal_static_tensorflow_SavedVariable_descriptor =
-      getDescriptor().getMessageTypes().get(8);
-    internal_static_tensorflow_SavedVariable_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_SavedVariable_descriptor,
-        new java.lang.String[] { "Dtype", "Shape", "Trainable", "Synchronization", "Aggregation", "Name", });
-    internal_static_tensorflow_FunctionSpec_descriptor =
-      getDescriptor().getMessageTypes().get(9);
-    internal_static_tensorflow_FunctionSpec_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_FunctionSpec_descriptor,
-        new java.lang.String[] { "Fullargspec", "IsMethod", "InputSignature", });
-    internal_static_tensorflow_SavedResource_descriptor =
-      getDescriptor().getMessageTypes().get(10);
-    internal_static_tensorflow_SavedResource_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_SavedResource_descriptor,
-        new java.lang.String[] { "Device", });
-    internal_static_tensorflow_SaveableObject_descriptor =
-      getDescriptor().getMessageTypes().get(11);
-    internal_static_tensorflow_SaveableObject_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_SaveableObject_descriptor,
-        new java.lang.String[] { "SaveFunction", "RestoreFunction", });
-    org.tensorflow.proto.framework.TensorShapeProtos.getDescriptor();
-    org.tensorflow.proto.framework.TypesProtos.getDescriptor();
-    org.tensorflow.proto.framework.VariableProtos.getDescriptor();
-    org.tensorflow.proto.framework.VersionsProtos.getDescriptor();
-    org.tensorflow.proto.framework.StructProtos.getDescriptor();
-    org.tensorflow.proto.framework.TrackableObjectGraphProtos.getDescriptor();
-  }
-
-  // @@protoc_insertion_point(outer_class_scope)
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedObjectOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedObjectOrBuilder.java
deleted file mode 100644
index 269ee74c9f2..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedObjectOrBuilder.java
+++ /dev/null
@@ -1,249 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/saved_object_graph.proto
-
-package org.tensorflow.proto.framework;
-
-public interface SavedObjectOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.SavedObject)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <pre>
-   * Objects which this object depends on: named edges in the dependency
-   * graph.
-   * Note: currently only valid if kind == "user_object".
-   * </pre>
-   *
-   * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.ObjectReference children = 1;</code>
-   */
-  java.util.List<org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference> 
-      getChildrenList();
-  /**
-   * <pre>
-   * Objects which this object depends on: named edges in the dependency
-   * graph.
-   * Note: currently only valid if kind == "user_object".
-   * </pre>
-   *
-   * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.ObjectReference children = 1;</code>
-   */
-  org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference getChildren(int index);
-  /**
-   * <pre>
-   * Objects which this object depends on: named edges in the dependency
-   * graph.
-   * Note: currently only valid if kind == "user_object".
-   * </pre>
-   *
-   * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.ObjectReference children = 1;</code>
-   */
-  int getChildrenCount();
-  /**
-   * <pre>
-   * Objects which this object depends on: named edges in the dependency
-   * graph.
-   * Note: currently only valid if kind == "user_object".
-   * </pre>
-   *
-   * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.ObjectReference children = 1;</code>
-   */
-  java.util.List<? extends org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReferenceOrBuilder> 
-      getChildrenOrBuilderList();
-  /**
-   * <pre>
-   * Objects which this object depends on: named edges in the dependency
-   * graph.
-   * Note: currently only valid if kind == "user_object".
-   * </pre>
-   *
-   * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.ObjectReference children = 1;</code>
-   */
-  org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReferenceOrBuilder getChildrenOrBuilder(
-      int index);
-
-  /**
-   * <pre>
-   * Slot variables owned by this object. This describes the three-way
-   * (optimizer, variable, slot variable) relationship; none of the three
-   * depend on the others directly.
-   * Note: currently only valid if kind == "user_object".
-   * </pre>
-   *
-   * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SlotVariableReference slot_variables = 3;</code>
-   */
-  java.util.List<org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference> 
-      getSlotVariablesList();
-  /**
-   * <pre>
-   * Slot variables owned by this object. This describes the three-way
-   * (optimizer, variable, slot variable) relationship; none of the three
-   * depend on the others directly.
-   * Note: currently only valid if kind == "user_object".
-   * </pre>
-   *
-   * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SlotVariableReference slot_variables = 3;</code>
-   */
-  org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference getSlotVariables(int index);
-  /**
-   * <pre>
-   * Slot variables owned by this object. This describes the three-way
-   * (optimizer, variable, slot variable) relationship; none of the three
-   * depend on the others directly.
-   * Note: currently only valid if kind == "user_object".
-   * </pre>
-   *
-   * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SlotVariableReference slot_variables = 3;</code>
-   */
-  int getSlotVariablesCount();
-  /**
-   * <pre>
-   * Slot variables owned by this object. This describes the three-way
-   * (optimizer, variable, slot variable) relationship; none of the three
-   * depend on the others directly.
-   * Note: currently only valid if kind == "user_object".
-   * </pre>
-   *
-   * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SlotVariableReference slot_variables = 3;</code>
-   */
-  java.util.List<? extends org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReferenceOrBuilder> 
-      getSlotVariablesOrBuilderList();
-  /**
-   * <pre>
-   * Slot variables owned by this object. This describes the three-way
-   * (optimizer, variable, slot variable) relationship; none of the three
-   * depend on the others directly.
-   * Note: currently only valid if kind == "user_object".
-   * </pre>
-   *
-   * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SlotVariableReference slot_variables = 3;</code>
-   */
-  org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReferenceOrBuilder getSlotVariablesOrBuilder(
-      int index);
-
-  /**
-   * <code>.tensorflow.SavedUserObject user_object = 4;</code>
-   */
-  boolean hasUserObject();
-  /**
-   * <code>.tensorflow.SavedUserObject user_object = 4;</code>
-   */
-  org.tensorflow.proto.framework.SavedUserObject getUserObject();
-  /**
-   * <code>.tensorflow.SavedUserObject user_object = 4;</code>
-   */
-  org.tensorflow.proto.framework.SavedUserObjectOrBuilder getUserObjectOrBuilder();
-
-  /**
-   * <code>.tensorflow.SavedAsset asset = 5;</code>
-   */
-  boolean hasAsset();
-  /**
-   * <code>.tensorflow.SavedAsset asset = 5;</code>
-   */
-  org.tensorflow.proto.framework.SavedAsset getAsset();
-  /**
-   * <code>.tensorflow.SavedAsset asset = 5;</code>
-   */
-  org.tensorflow.proto.framework.SavedAssetOrBuilder getAssetOrBuilder();
-
-  /**
-   * <code>.tensorflow.SavedFunction function = 6;</code>
-   */
-  boolean hasFunction();
-  /**
-   * <code>.tensorflow.SavedFunction function = 6;</code>
-   */
-  org.tensorflow.proto.framework.SavedFunction getFunction();
-  /**
-   * <code>.tensorflow.SavedFunction function = 6;</code>
-   */
-  org.tensorflow.proto.framework.SavedFunctionOrBuilder getFunctionOrBuilder();
-
-  /**
-   * <code>.tensorflow.SavedVariable variable = 7;</code>
-   */
-  boolean hasVariable();
-  /**
-   * <code>.tensorflow.SavedVariable variable = 7;</code>
-   */
-  org.tensorflow.proto.framework.SavedVariable getVariable();
-  /**
-   * <code>.tensorflow.SavedVariable variable = 7;</code>
-   */
-  org.tensorflow.proto.framework.SavedVariableOrBuilder getVariableOrBuilder();
-
-  /**
-   * <code>.tensorflow.SavedBareConcreteFunction bare_concrete_function = 8;</code>
-   */
-  boolean hasBareConcreteFunction();
-  /**
-   * <code>.tensorflow.SavedBareConcreteFunction bare_concrete_function = 8;</code>
-   */
-  org.tensorflow.proto.framework.SavedBareConcreteFunction getBareConcreteFunction();
-  /**
-   * <code>.tensorflow.SavedBareConcreteFunction bare_concrete_function = 8;</code>
-   */
-  org.tensorflow.proto.framework.SavedBareConcreteFunctionOrBuilder getBareConcreteFunctionOrBuilder();
-
-  /**
-   * <code>.tensorflow.SavedConstant constant = 9;</code>
-   */
-  boolean hasConstant();
-  /**
-   * <code>.tensorflow.SavedConstant constant = 9;</code>
-   */
-  org.tensorflow.proto.framework.SavedConstant getConstant();
-  /**
-   * <code>.tensorflow.SavedConstant constant = 9;</code>
-   */
-  org.tensorflow.proto.framework.SavedConstantOrBuilder getConstantOrBuilder();
-
-  /**
-   * <code>.tensorflow.SavedResource resource = 10;</code>
-   */
-  boolean hasResource();
-  /**
-   * <code>.tensorflow.SavedResource resource = 10;</code>
-   */
-  org.tensorflow.proto.framework.SavedResource getResource();
-  /**
-   * <code>.tensorflow.SavedResource resource = 10;</code>
-   */
-  org.tensorflow.proto.framework.SavedResourceOrBuilder getResourceOrBuilder();
-
-  /**
-   * <code>map&lt;string, .tensorflow.SaveableObject&gt; saveable_objects = 11;</code>
-   */
-  int getSaveableObjectsCount();
-  /**
-   * <code>map&lt;string, .tensorflow.SaveableObject&gt; saveable_objects = 11;</code>
-   */
-  boolean containsSaveableObjects(
-      java.lang.String key);
-  /**
-   * Use {@link #getSaveableObjectsMap()} instead.
-   */
-  @java.lang.Deprecated
-  java.util.Map<java.lang.String, org.tensorflow.proto.framework.SaveableObject>
-  getSaveableObjects();
-  /**
-   * <code>map&lt;string, .tensorflow.SaveableObject&gt; saveable_objects = 11;</code>
-   */
-  java.util.Map<java.lang.String, org.tensorflow.proto.framework.SaveableObject>
-  getSaveableObjectsMap();
-  /**
-   * <code>map&lt;string, .tensorflow.SaveableObject&gt; saveable_objects = 11;</code>
-   */
-
-  org.tensorflow.proto.framework.SaveableObject getSaveableObjectsOrDefault(
-      java.lang.String key,
-      org.tensorflow.proto.framework.SaveableObject defaultValue);
-  /**
-   * <code>map&lt;string, .tensorflow.SaveableObject&gt; saveable_objects = 11;</code>
-   */
-
-  org.tensorflow.proto.framework.SaveableObject getSaveableObjectsOrThrow(
-      java.lang.String key);
-
-  public org.tensorflow.proto.framework.SavedObject.KindCase getKindCase();
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedResource.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedResource.java
deleted file mode 100644
index b45f7ff73a9..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedResource.java
+++ /dev/null
@@ -1,600 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/saved_object_graph.proto
-
-package org.tensorflow.proto.framework;
-
-/**
- * <pre>
- * A SavedResource represents a TF object that holds state during its lifetime.
- * An object of this type can have a reference to a:
- * create_resource() and an initialize() function.
- * </pre>
- *
- * Protobuf type {@code tensorflow.SavedResource}
- */
-public  final class SavedResource extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.SavedResource)
-    SavedResourceOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use SavedResource.newBuilder() to construct.
-  private SavedResource(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private SavedResource() {
-    device_ = "";
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new SavedResource();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private SavedResource(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 10: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            device_ = s;
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.framework.SavedObjectGraphProtos.internal_static_tensorflow_SavedResource_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.framework.SavedObjectGraphProtos.internal_static_tensorflow_SavedResource_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.framework.SavedResource.class, org.tensorflow.proto.framework.SavedResource.Builder.class);
-  }
-
-  public static final int DEVICE_FIELD_NUMBER = 1;
-  private volatile java.lang.Object device_;
-  /**
-   * <pre>
-   * A device specification indicating a required placement for the resource
-   * creation function, e.g. "CPU". An empty string allows the user to select a
-   * device.
-   * </pre>
-   *
-   * <code>string device = 1;</code>
-   */
-  public java.lang.String getDevice() {
-    java.lang.Object ref = device_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      device_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * A device specification indicating a required placement for the resource
-   * creation function, e.g. "CPU". An empty string allows the user to select a
-   * device.
-   * </pre>
-   *
-   * <code>string device = 1;</code>
-   */
-  public com.google.protobuf.ByteString
-      getDeviceBytes() {
-    java.lang.Object ref = device_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      device_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    if (!getDeviceBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 1, device_);
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (!getDeviceBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, device_);
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.framework.SavedResource)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.framework.SavedResource other = (org.tensorflow.proto.framework.SavedResource) obj;
-
-    if (!getDevice()
-        .equals(other.getDevice())) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    hash = (37 * hash) + DEVICE_FIELD_NUMBER;
-    hash = (53 * hash) + getDevice().hashCode();
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.framework.SavedResource parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.SavedResource parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.SavedResource parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.SavedResource parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.SavedResource parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.SavedResource parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.SavedResource parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.SavedResource parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.SavedResource parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.SavedResource parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.SavedResource parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.SavedResource parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.framework.SavedResource prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * <pre>
-   * A SavedResource represents a TF object that holds state during its lifetime.
-   * An object of this type can have a reference to a:
-   * create_resource() and an initialize() function.
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.SavedResource}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.SavedResource)
-      org.tensorflow.proto.framework.SavedResourceOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.SavedObjectGraphProtos.internal_static_tensorflow_SavedResource_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.SavedObjectGraphProtos.internal_static_tensorflow_SavedResource_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.SavedResource.class, org.tensorflow.proto.framework.SavedResource.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.framework.SavedResource.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      device_ = "";
-
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.framework.SavedObjectGraphProtos.internal_static_tensorflow_SavedResource_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.SavedResource getDefaultInstanceForType() {
-      return org.tensorflow.proto.framework.SavedResource.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.SavedResource build() {
-      org.tensorflow.proto.framework.SavedResource result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.SavedResource buildPartial() {
-      org.tensorflow.proto.framework.SavedResource result = new org.tensorflow.proto.framework.SavedResource(this);
-      result.device_ = device_;
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.framework.SavedResource) {
-        return mergeFrom((org.tensorflow.proto.framework.SavedResource)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.framework.SavedResource other) {
-      if (other == org.tensorflow.proto.framework.SavedResource.getDefaultInstance()) return this;
-      if (!other.getDevice().isEmpty()) {
-        device_ = other.device_;
-        onChanged();
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.framework.SavedResource parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.framework.SavedResource) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-
-    private java.lang.Object device_ = "";
-    /**
-     * <pre>
-     * A device specification indicating a required placement for the resource
-     * creation function, e.g. "CPU". An empty string allows the user to select a
-     * device.
-     * </pre>
-     *
-     * <code>string device = 1;</code>
-     */
-    public java.lang.String getDevice() {
-      java.lang.Object ref = device_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        device_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * A device specification indicating a required placement for the resource
-     * creation function, e.g. "CPU". An empty string allows the user to select a
-     * device.
-     * </pre>
-     *
-     * <code>string device = 1;</code>
-     */
-    public com.google.protobuf.ByteString
-        getDeviceBytes() {
-      java.lang.Object ref = device_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        device_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * A device specification indicating a required placement for the resource
-     * creation function, e.g. "CPU". An empty string allows the user to select a
-     * device.
-     * </pre>
-     *
-     * <code>string device = 1;</code>
-     */
-    public Builder setDevice(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      device_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * A device specification indicating a required placement for the resource
-     * creation function, e.g. "CPU". An empty string allows the user to select a
-     * device.
-     * </pre>
-     *
-     * <code>string device = 1;</code>
-     */
-    public Builder clearDevice() {
-      
-      device_ = getDefaultInstance().getDevice();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * A device specification indicating a required placement for the resource
-     * creation function, e.g. "CPU". An empty string allows the user to select a
-     * device.
-     * </pre>
-     *
-     * <code>string device = 1;</code>
-     */
-    public Builder setDeviceBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      device_ = value;
-      onChanged();
-      return this;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.SavedResource)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.SavedResource)
-  private static final org.tensorflow.proto.framework.SavedResource DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.SavedResource();
-  }
-
-  public static org.tensorflow.proto.framework.SavedResource getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<SavedResource>
-      PARSER = new com.google.protobuf.AbstractParser<SavedResource>() {
-    @java.lang.Override
-    public SavedResource parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new SavedResource(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<SavedResource> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<SavedResource> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.framework.SavedResource getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedResourceOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedResourceOrBuilder.java
deleted file mode 100644
index 5d78c2eea9b..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedResourceOrBuilder.java
+++ /dev/null
@@ -1,31 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/saved_object_graph.proto
-
-package org.tensorflow.proto.framework;
-
-public interface SavedResourceOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.SavedResource)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <pre>
-   * A device specification indicating a required placement for the resource
-   * creation function, e.g. "CPU". An empty string allows the user to select a
-   * device.
-   * </pre>
-   *
-   * <code>string device = 1;</code>
-   */
-  java.lang.String getDevice();
-  /**
-   * <pre>
-   * A device specification indicating a required placement for the resource
-   * creation function, e.g. "CPU". An empty string allows the user to select a
-   * device.
-   * </pre>
-   *
-   * <code>string device = 1;</code>
-   */
-  com.google.protobuf.ByteString
-      getDeviceBytes();
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedUserObject.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedUserObject.java
deleted file mode 100644
index f66874096f5..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedUserObject.java
+++ /dev/null
@@ -1,974 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/saved_object_graph.proto
-
-package org.tensorflow.proto.framework;
-
-/**
- * <pre>
- * A SavedUserObject is an object (in the object-oriented language of the
- * TensorFlow program) of some user- or framework-defined class other than
- * those handled specifically by the other kinds of SavedObjects.
- * This object cannot be evaluated as a tensor, and therefore cannot be bound
- * to an input of a function.
- * </pre>
- *
- * Protobuf type {@code tensorflow.SavedUserObject}
- */
-public  final class SavedUserObject extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.SavedUserObject)
-    SavedUserObjectOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use SavedUserObject.newBuilder() to construct.
-  private SavedUserObject(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private SavedUserObject() {
-    identifier_ = "";
-    metadata_ = "";
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new SavedUserObject();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private SavedUserObject(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 10: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            identifier_ = s;
-            break;
-          }
-          case 18: {
-            org.tensorflow.proto.framework.VersionDef.Builder subBuilder = null;
-            if (version_ != null) {
-              subBuilder = version_.toBuilder();
-            }
-            version_ = input.readMessage(org.tensorflow.proto.framework.VersionDef.parser(), extensionRegistry);
-            if (subBuilder != null) {
-              subBuilder.mergeFrom(version_);
-              version_ = subBuilder.buildPartial();
-            }
-
-            break;
-          }
-          case 26: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            metadata_ = s;
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.framework.SavedObjectGraphProtos.internal_static_tensorflow_SavedUserObject_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.framework.SavedObjectGraphProtos.internal_static_tensorflow_SavedUserObject_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.framework.SavedUserObject.class, org.tensorflow.proto.framework.SavedUserObject.Builder.class);
-  }
-
-  public static final int IDENTIFIER_FIELD_NUMBER = 1;
-  private volatile java.lang.Object identifier_;
-  /**
-   * <pre>
-   * Corresponds to a registration of the type to use in the loading program.
-   * </pre>
-   *
-   * <code>string identifier = 1;</code>
-   */
-  public java.lang.String getIdentifier() {
-    java.lang.Object ref = identifier_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      identifier_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * Corresponds to a registration of the type to use in the loading program.
-   * </pre>
-   *
-   * <code>string identifier = 1;</code>
-   */
-  public com.google.protobuf.ByteString
-      getIdentifierBytes() {
-    java.lang.Object ref = identifier_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      identifier_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int VERSION_FIELD_NUMBER = 2;
-  private org.tensorflow.proto.framework.VersionDef version_;
-  /**
-   * <pre>
-   * Version information from the producer of this SavedUserObject.
-   * </pre>
-   *
-   * <code>.tensorflow.VersionDef version = 2;</code>
-   */
-  public boolean hasVersion() {
-    return version_ != null;
-  }
-  /**
-   * <pre>
-   * Version information from the producer of this SavedUserObject.
-   * </pre>
-   *
-   * <code>.tensorflow.VersionDef version = 2;</code>
-   */
-  public org.tensorflow.proto.framework.VersionDef getVersion() {
-    return version_ == null ? org.tensorflow.proto.framework.VersionDef.getDefaultInstance() : version_;
-  }
-  /**
-   * <pre>
-   * Version information from the producer of this SavedUserObject.
-   * </pre>
-   *
-   * <code>.tensorflow.VersionDef version = 2;</code>
-   */
-  public org.tensorflow.proto.framework.VersionDefOrBuilder getVersionOrBuilder() {
-    return getVersion();
-  }
-
-  public static final int METADATA_FIELD_NUMBER = 3;
-  private volatile java.lang.Object metadata_;
-  /**
-   * <pre>
-   * Initialization-related metadata.
-   * </pre>
-   *
-   * <code>string metadata = 3;</code>
-   */
-  public java.lang.String getMetadata() {
-    java.lang.Object ref = metadata_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      metadata_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * Initialization-related metadata.
-   * </pre>
-   *
-   * <code>string metadata = 3;</code>
-   */
-  public com.google.protobuf.ByteString
-      getMetadataBytes() {
-    java.lang.Object ref = metadata_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      metadata_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    if (!getIdentifierBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 1, identifier_);
-    }
-    if (version_ != null) {
-      output.writeMessage(2, getVersion());
-    }
-    if (!getMetadataBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 3, metadata_);
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (!getIdentifierBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, identifier_);
-    }
-    if (version_ != null) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(2, getVersion());
-    }
-    if (!getMetadataBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(3, metadata_);
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.framework.SavedUserObject)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.framework.SavedUserObject other = (org.tensorflow.proto.framework.SavedUserObject) obj;
-
-    if (!getIdentifier()
-        .equals(other.getIdentifier())) return false;
-    if (hasVersion() != other.hasVersion()) return false;
-    if (hasVersion()) {
-      if (!getVersion()
-          .equals(other.getVersion())) return false;
-    }
-    if (!getMetadata()
-        .equals(other.getMetadata())) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    hash = (37 * hash) + IDENTIFIER_FIELD_NUMBER;
-    hash = (53 * hash) + getIdentifier().hashCode();
-    if (hasVersion()) {
-      hash = (37 * hash) + VERSION_FIELD_NUMBER;
-      hash = (53 * hash) + getVersion().hashCode();
-    }
-    hash = (37 * hash) + METADATA_FIELD_NUMBER;
-    hash = (53 * hash) + getMetadata().hashCode();
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.framework.SavedUserObject parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.SavedUserObject parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.SavedUserObject parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.SavedUserObject parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.SavedUserObject parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.SavedUserObject parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.SavedUserObject parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.SavedUserObject parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.SavedUserObject parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.SavedUserObject parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.SavedUserObject parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.SavedUserObject parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.framework.SavedUserObject prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * <pre>
-   * A SavedUserObject is an object (in the object-oriented language of the
-   * TensorFlow program) of some user- or framework-defined class other than
-   * those handled specifically by the other kinds of SavedObjects.
-   * This object cannot be evaluated as a tensor, and therefore cannot be bound
-   * to an input of a function.
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.SavedUserObject}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.SavedUserObject)
-      org.tensorflow.proto.framework.SavedUserObjectOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.SavedObjectGraphProtos.internal_static_tensorflow_SavedUserObject_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.SavedObjectGraphProtos.internal_static_tensorflow_SavedUserObject_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.SavedUserObject.class, org.tensorflow.proto.framework.SavedUserObject.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.framework.SavedUserObject.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      identifier_ = "";
-
-      if (versionBuilder_ == null) {
-        version_ = null;
-      } else {
-        version_ = null;
-        versionBuilder_ = null;
-      }
-      metadata_ = "";
-
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.framework.SavedObjectGraphProtos.internal_static_tensorflow_SavedUserObject_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.SavedUserObject getDefaultInstanceForType() {
-      return org.tensorflow.proto.framework.SavedUserObject.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.SavedUserObject build() {
-      org.tensorflow.proto.framework.SavedUserObject result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.SavedUserObject buildPartial() {
-      org.tensorflow.proto.framework.SavedUserObject result = new org.tensorflow.proto.framework.SavedUserObject(this);
-      result.identifier_ = identifier_;
-      if (versionBuilder_ == null) {
-        result.version_ = version_;
-      } else {
-        result.version_ = versionBuilder_.build();
-      }
-      result.metadata_ = metadata_;
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.framework.SavedUserObject) {
-        return mergeFrom((org.tensorflow.proto.framework.SavedUserObject)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.framework.SavedUserObject other) {
-      if (other == org.tensorflow.proto.framework.SavedUserObject.getDefaultInstance()) return this;
-      if (!other.getIdentifier().isEmpty()) {
-        identifier_ = other.identifier_;
-        onChanged();
-      }
-      if (other.hasVersion()) {
-        mergeVersion(other.getVersion());
-      }
-      if (!other.getMetadata().isEmpty()) {
-        metadata_ = other.metadata_;
-        onChanged();
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.framework.SavedUserObject parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.framework.SavedUserObject) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-
-    private java.lang.Object identifier_ = "";
-    /**
-     * <pre>
-     * Corresponds to a registration of the type to use in the loading program.
-     * </pre>
-     *
-     * <code>string identifier = 1;</code>
-     */
-    public java.lang.String getIdentifier() {
-      java.lang.Object ref = identifier_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        identifier_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Corresponds to a registration of the type to use in the loading program.
-     * </pre>
-     *
-     * <code>string identifier = 1;</code>
-     */
-    public com.google.protobuf.ByteString
-        getIdentifierBytes() {
-      java.lang.Object ref = identifier_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        identifier_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Corresponds to a registration of the type to use in the loading program.
-     * </pre>
-     *
-     * <code>string identifier = 1;</code>
-     */
-    public Builder setIdentifier(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      identifier_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Corresponds to a registration of the type to use in the loading program.
-     * </pre>
-     *
-     * <code>string identifier = 1;</code>
-     */
-    public Builder clearIdentifier() {
-      
-      identifier_ = getDefaultInstance().getIdentifier();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Corresponds to a registration of the type to use in the loading program.
-     * </pre>
-     *
-     * <code>string identifier = 1;</code>
-     */
-    public Builder setIdentifierBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      identifier_ = value;
-      onChanged();
-      return this;
-    }
-
-    private org.tensorflow.proto.framework.VersionDef version_;
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.VersionDef, org.tensorflow.proto.framework.VersionDef.Builder, org.tensorflow.proto.framework.VersionDefOrBuilder> versionBuilder_;
-    /**
-     * <pre>
-     * Version information from the producer of this SavedUserObject.
-     * </pre>
-     *
-     * <code>.tensorflow.VersionDef version = 2;</code>
-     */
-    public boolean hasVersion() {
-      return versionBuilder_ != null || version_ != null;
-    }
-    /**
-     * <pre>
-     * Version information from the producer of this SavedUserObject.
-     * </pre>
-     *
-     * <code>.tensorflow.VersionDef version = 2;</code>
-     */
-    public org.tensorflow.proto.framework.VersionDef getVersion() {
-      if (versionBuilder_ == null) {
-        return version_ == null ? org.tensorflow.proto.framework.VersionDef.getDefaultInstance() : version_;
-      } else {
-        return versionBuilder_.getMessage();
-      }
-    }
-    /**
-     * <pre>
-     * Version information from the producer of this SavedUserObject.
-     * </pre>
-     *
-     * <code>.tensorflow.VersionDef version = 2;</code>
-     */
-    public Builder setVersion(org.tensorflow.proto.framework.VersionDef value) {
-      if (versionBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        version_ = value;
-        onChanged();
-      } else {
-        versionBuilder_.setMessage(value);
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Version information from the producer of this SavedUserObject.
-     * </pre>
-     *
-     * <code>.tensorflow.VersionDef version = 2;</code>
-     */
-    public Builder setVersion(
-        org.tensorflow.proto.framework.VersionDef.Builder builderForValue) {
-      if (versionBuilder_ == null) {
-        version_ = builderForValue.build();
-        onChanged();
-      } else {
-        versionBuilder_.setMessage(builderForValue.build());
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Version information from the producer of this SavedUserObject.
-     * </pre>
-     *
-     * <code>.tensorflow.VersionDef version = 2;</code>
-     */
-    public Builder mergeVersion(org.tensorflow.proto.framework.VersionDef value) {
-      if (versionBuilder_ == null) {
-        if (version_ != null) {
-          version_ =
-            org.tensorflow.proto.framework.VersionDef.newBuilder(version_).mergeFrom(value).buildPartial();
-        } else {
-          version_ = value;
-        }
-        onChanged();
-      } else {
-        versionBuilder_.mergeFrom(value);
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Version information from the producer of this SavedUserObject.
-     * </pre>
-     *
-     * <code>.tensorflow.VersionDef version = 2;</code>
-     */
-    public Builder clearVersion() {
-      if (versionBuilder_ == null) {
-        version_ = null;
-        onChanged();
-      } else {
-        version_ = null;
-        versionBuilder_ = null;
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Version information from the producer of this SavedUserObject.
-     * </pre>
-     *
-     * <code>.tensorflow.VersionDef version = 2;</code>
-     */
-    public org.tensorflow.proto.framework.VersionDef.Builder getVersionBuilder() {
-      
-      onChanged();
-      return getVersionFieldBuilder().getBuilder();
-    }
-    /**
-     * <pre>
-     * Version information from the producer of this SavedUserObject.
-     * </pre>
-     *
-     * <code>.tensorflow.VersionDef version = 2;</code>
-     */
-    public org.tensorflow.proto.framework.VersionDefOrBuilder getVersionOrBuilder() {
-      if (versionBuilder_ != null) {
-        return versionBuilder_.getMessageOrBuilder();
-      } else {
-        return version_ == null ?
-            org.tensorflow.proto.framework.VersionDef.getDefaultInstance() : version_;
-      }
-    }
-    /**
-     * <pre>
-     * Version information from the producer of this SavedUserObject.
-     * </pre>
-     *
-     * <code>.tensorflow.VersionDef version = 2;</code>
-     */
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.VersionDef, org.tensorflow.proto.framework.VersionDef.Builder, org.tensorflow.proto.framework.VersionDefOrBuilder> 
-        getVersionFieldBuilder() {
-      if (versionBuilder_ == null) {
-        versionBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-            org.tensorflow.proto.framework.VersionDef, org.tensorflow.proto.framework.VersionDef.Builder, org.tensorflow.proto.framework.VersionDefOrBuilder>(
-                getVersion(),
-                getParentForChildren(),
-                isClean());
-        version_ = null;
-      }
-      return versionBuilder_;
-    }
-
-    private java.lang.Object metadata_ = "";
-    /**
-     * <pre>
-     * Initialization-related metadata.
-     * </pre>
-     *
-     * <code>string metadata = 3;</code>
-     */
-    public java.lang.String getMetadata() {
-      java.lang.Object ref = metadata_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        metadata_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Initialization-related metadata.
-     * </pre>
-     *
-     * <code>string metadata = 3;</code>
-     */
-    public com.google.protobuf.ByteString
-        getMetadataBytes() {
-      java.lang.Object ref = metadata_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        metadata_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Initialization-related metadata.
-     * </pre>
-     *
-     * <code>string metadata = 3;</code>
-     */
-    public Builder setMetadata(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      metadata_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Initialization-related metadata.
-     * </pre>
-     *
-     * <code>string metadata = 3;</code>
-     */
-    public Builder clearMetadata() {
-      
-      metadata_ = getDefaultInstance().getMetadata();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Initialization-related metadata.
-     * </pre>
-     *
-     * <code>string metadata = 3;</code>
-     */
-    public Builder setMetadataBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      metadata_ = value;
-      onChanged();
-      return this;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.SavedUserObject)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.SavedUserObject)
-  private static final org.tensorflow.proto.framework.SavedUserObject DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.SavedUserObject();
-  }
-
-  public static org.tensorflow.proto.framework.SavedUserObject getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<SavedUserObject>
-      PARSER = new com.google.protobuf.AbstractParser<SavedUserObject>() {
-    @java.lang.Override
-    public SavedUserObject parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new SavedUserObject(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<SavedUserObject> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<SavedUserObject> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.framework.SavedUserObject getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedUserObjectOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedUserObjectOrBuilder.java
deleted file mode 100644
index bd7cbadcbd3..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedUserObjectOrBuilder.java
+++ /dev/null
@@ -1,70 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/saved_object_graph.proto
-
-package org.tensorflow.proto.framework;
-
-public interface SavedUserObjectOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.SavedUserObject)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <pre>
-   * Corresponds to a registration of the type to use in the loading program.
-   * </pre>
-   *
-   * <code>string identifier = 1;</code>
-   */
-  java.lang.String getIdentifier();
-  /**
-   * <pre>
-   * Corresponds to a registration of the type to use in the loading program.
-   * </pre>
-   *
-   * <code>string identifier = 1;</code>
-   */
-  com.google.protobuf.ByteString
-      getIdentifierBytes();
-
-  /**
-   * <pre>
-   * Version information from the producer of this SavedUserObject.
-   * </pre>
-   *
-   * <code>.tensorflow.VersionDef version = 2;</code>
-   */
-  boolean hasVersion();
-  /**
-   * <pre>
-   * Version information from the producer of this SavedUserObject.
-   * </pre>
-   *
-   * <code>.tensorflow.VersionDef version = 2;</code>
-   */
-  org.tensorflow.proto.framework.VersionDef getVersion();
-  /**
-   * <pre>
-   * Version information from the producer of this SavedUserObject.
-   * </pre>
-   *
-   * <code>.tensorflow.VersionDef version = 2;</code>
-   */
-  org.tensorflow.proto.framework.VersionDefOrBuilder getVersionOrBuilder();
-
-  /**
-   * <pre>
-   * Initialization-related metadata.
-   * </pre>
-   *
-   * <code>string metadata = 3;</code>
-   */
-  java.lang.String getMetadata();
-  /**
-   * <pre>
-   * Initialization-related metadata.
-   * </pre>
-   *
-   * <code>string metadata = 3;</code>
-   */
-  com.google.protobuf.ByteString
-      getMetadataBytes();
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedVariable.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedVariable.java
deleted file mode 100644
index 17ce45af6d7..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedVariable.java
+++ /dev/null
@@ -1,1050 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/saved_object_graph.proto
-
-package org.tensorflow.proto.framework;
-
-/**
- * <pre>
- * Represents a Variable that is initialized by loading the contents from the
- * checkpoint.
- * </pre>
- *
- * Protobuf type {@code tensorflow.SavedVariable}
- */
-public  final class SavedVariable extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.SavedVariable)
-    SavedVariableOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use SavedVariable.newBuilder() to construct.
-  private SavedVariable(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private SavedVariable() {
-    dtype_ = 0;
-    synchronization_ = 0;
-    aggregation_ = 0;
-    name_ = "";
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new SavedVariable();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private SavedVariable(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 8: {
-            int rawValue = input.readEnum();
-
-            dtype_ = rawValue;
-            break;
-          }
-          case 18: {
-            org.tensorflow.proto.framework.TensorShapeProto.Builder subBuilder = null;
-            if (shape_ != null) {
-              subBuilder = shape_.toBuilder();
-            }
-            shape_ = input.readMessage(org.tensorflow.proto.framework.TensorShapeProto.parser(), extensionRegistry);
-            if (subBuilder != null) {
-              subBuilder.mergeFrom(shape_);
-              shape_ = subBuilder.buildPartial();
-            }
-
-            break;
-          }
-          case 24: {
-
-            trainable_ = input.readBool();
-            break;
-          }
-          case 32: {
-            int rawValue = input.readEnum();
-
-            synchronization_ = rawValue;
-            break;
-          }
-          case 40: {
-            int rawValue = input.readEnum();
-
-            aggregation_ = rawValue;
-            break;
-          }
-          case 50: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            name_ = s;
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.framework.SavedObjectGraphProtos.internal_static_tensorflow_SavedVariable_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.framework.SavedObjectGraphProtos.internal_static_tensorflow_SavedVariable_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.framework.SavedVariable.class, org.tensorflow.proto.framework.SavedVariable.Builder.class);
-  }
-
-  public static final int DTYPE_FIELD_NUMBER = 1;
-  private int dtype_;
-  /**
-   * <code>.tensorflow.DataType dtype = 1;</code>
-   */
-  public int getDtypeValue() {
-    return dtype_;
-  }
-  /**
-   * <code>.tensorflow.DataType dtype = 1;</code>
-   */
-  public org.tensorflow.proto.framework.DataType getDtype() {
-    @SuppressWarnings("deprecation")
-    org.tensorflow.proto.framework.DataType result = org.tensorflow.proto.framework.DataType.valueOf(dtype_);
-    return result == null ? org.tensorflow.proto.framework.DataType.UNRECOGNIZED : result;
-  }
-
-  public static final int SHAPE_FIELD_NUMBER = 2;
-  private org.tensorflow.proto.framework.TensorShapeProto shape_;
-  /**
-   * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-   */
-  public boolean hasShape() {
-    return shape_ != null;
-  }
-  /**
-   * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-   */
-  public org.tensorflow.proto.framework.TensorShapeProto getShape() {
-    return shape_ == null ? org.tensorflow.proto.framework.TensorShapeProto.getDefaultInstance() : shape_;
-  }
-  /**
-   * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-   */
-  public org.tensorflow.proto.framework.TensorShapeProtoOrBuilder getShapeOrBuilder() {
-    return getShape();
-  }
-
-  public static final int TRAINABLE_FIELD_NUMBER = 3;
-  private boolean trainable_;
-  /**
-   * <code>bool trainable = 3;</code>
-   */
-  public boolean getTrainable() {
-    return trainable_;
-  }
-
-  public static final int SYNCHRONIZATION_FIELD_NUMBER = 4;
-  private int synchronization_;
-  /**
-   * <code>.tensorflow.VariableSynchronization synchronization = 4;</code>
-   */
-  public int getSynchronizationValue() {
-    return synchronization_;
-  }
-  /**
-   * <code>.tensorflow.VariableSynchronization synchronization = 4;</code>
-   */
-  public org.tensorflow.proto.framework.VariableSynchronization getSynchronization() {
-    @SuppressWarnings("deprecation")
-    org.tensorflow.proto.framework.VariableSynchronization result = org.tensorflow.proto.framework.VariableSynchronization.valueOf(synchronization_);
-    return result == null ? org.tensorflow.proto.framework.VariableSynchronization.UNRECOGNIZED : result;
-  }
-
-  public static final int AGGREGATION_FIELD_NUMBER = 5;
-  private int aggregation_;
-  /**
-   * <code>.tensorflow.VariableAggregation aggregation = 5;</code>
-   */
-  public int getAggregationValue() {
-    return aggregation_;
-  }
-  /**
-   * <code>.tensorflow.VariableAggregation aggregation = 5;</code>
-   */
-  public org.tensorflow.proto.framework.VariableAggregation getAggregation() {
-    @SuppressWarnings("deprecation")
-    org.tensorflow.proto.framework.VariableAggregation result = org.tensorflow.proto.framework.VariableAggregation.valueOf(aggregation_);
-    return result == null ? org.tensorflow.proto.framework.VariableAggregation.UNRECOGNIZED : result;
-  }
-
-  public static final int NAME_FIELD_NUMBER = 6;
-  private volatile java.lang.Object name_;
-  /**
-   * <code>string name = 6;</code>
-   */
-  public java.lang.String getName() {
-    java.lang.Object ref = name_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      name_ = s;
-      return s;
-    }
-  }
-  /**
-   * <code>string name = 6;</code>
-   */
-  public com.google.protobuf.ByteString
-      getNameBytes() {
-    java.lang.Object ref = name_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      name_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    if (dtype_ != org.tensorflow.proto.framework.DataType.DT_INVALID.getNumber()) {
-      output.writeEnum(1, dtype_);
-    }
-    if (shape_ != null) {
-      output.writeMessage(2, getShape());
-    }
-    if (trainable_ != false) {
-      output.writeBool(3, trainable_);
-    }
-    if (synchronization_ != org.tensorflow.proto.framework.VariableSynchronization.VARIABLE_SYNCHRONIZATION_AUTO.getNumber()) {
-      output.writeEnum(4, synchronization_);
-    }
-    if (aggregation_ != org.tensorflow.proto.framework.VariableAggregation.VARIABLE_AGGREGATION_NONE.getNumber()) {
-      output.writeEnum(5, aggregation_);
-    }
-    if (!getNameBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 6, name_);
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (dtype_ != org.tensorflow.proto.framework.DataType.DT_INVALID.getNumber()) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeEnumSize(1, dtype_);
-    }
-    if (shape_ != null) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(2, getShape());
-    }
-    if (trainable_ != false) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeBoolSize(3, trainable_);
-    }
-    if (synchronization_ != org.tensorflow.proto.framework.VariableSynchronization.VARIABLE_SYNCHRONIZATION_AUTO.getNumber()) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeEnumSize(4, synchronization_);
-    }
-    if (aggregation_ != org.tensorflow.proto.framework.VariableAggregation.VARIABLE_AGGREGATION_NONE.getNumber()) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeEnumSize(5, aggregation_);
-    }
-    if (!getNameBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(6, name_);
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.framework.SavedVariable)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.framework.SavedVariable other = (org.tensorflow.proto.framework.SavedVariable) obj;
-
-    if (dtype_ != other.dtype_) return false;
-    if (hasShape() != other.hasShape()) return false;
-    if (hasShape()) {
-      if (!getShape()
-          .equals(other.getShape())) return false;
-    }
-    if (getTrainable()
-        != other.getTrainable()) return false;
-    if (synchronization_ != other.synchronization_) return false;
-    if (aggregation_ != other.aggregation_) return false;
-    if (!getName()
-        .equals(other.getName())) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    hash = (37 * hash) + DTYPE_FIELD_NUMBER;
-    hash = (53 * hash) + dtype_;
-    if (hasShape()) {
-      hash = (37 * hash) + SHAPE_FIELD_NUMBER;
-      hash = (53 * hash) + getShape().hashCode();
-    }
-    hash = (37 * hash) + TRAINABLE_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashBoolean(
-        getTrainable());
-    hash = (37 * hash) + SYNCHRONIZATION_FIELD_NUMBER;
-    hash = (53 * hash) + synchronization_;
-    hash = (37 * hash) + AGGREGATION_FIELD_NUMBER;
-    hash = (53 * hash) + aggregation_;
-    hash = (37 * hash) + NAME_FIELD_NUMBER;
-    hash = (53 * hash) + getName().hashCode();
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.framework.SavedVariable parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.SavedVariable parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.SavedVariable parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.SavedVariable parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.SavedVariable parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.SavedVariable parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.SavedVariable parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.SavedVariable parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.SavedVariable parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.SavedVariable parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.SavedVariable parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.SavedVariable parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.framework.SavedVariable prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * <pre>
-   * Represents a Variable that is initialized by loading the contents from the
-   * checkpoint.
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.SavedVariable}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.SavedVariable)
-      org.tensorflow.proto.framework.SavedVariableOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.SavedObjectGraphProtos.internal_static_tensorflow_SavedVariable_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.SavedObjectGraphProtos.internal_static_tensorflow_SavedVariable_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.SavedVariable.class, org.tensorflow.proto.framework.SavedVariable.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.framework.SavedVariable.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      dtype_ = 0;
-
-      if (shapeBuilder_ == null) {
-        shape_ = null;
-      } else {
-        shape_ = null;
-        shapeBuilder_ = null;
-      }
-      trainable_ = false;
-
-      synchronization_ = 0;
-
-      aggregation_ = 0;
-
-      name_ = "";
-
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.framework.SavedObjectGraphProtos.internal_static_tensorflow_SavedVariable_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.SavedVariable getDefaultInstanceForType() {
-      return org.tensorflow.proto.framework.SavedVariable.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.SavedVariable build() {
-      org.tensorflow.proto.framework.SavedVariable result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.SavedVariable buildPartial() {
-      org.tensorflow.proto.framework.SavedVariable result = new org.tensorflow.proto.framework.SavedVariable(this);
-      result.dtype_ = dtype_;
-      if (shapeBuilder_ == null) {
-        result.shape_ = shape_;
-      } else {
-        result.shape_ = shapeBuilder_.build();
-      }
-      result.trainable_ = trainable_;
-      result.synchronization_ = synchronization_;
-      result.aggregation_ = aggregation_;
-      result.name_ = name_;
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.framework.SavedVariable) {
-        return mergeFrom((org.tensorflow.proto.framework.SavedVariable)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.framework.SavedVariable other) {
-      if (other == org.tensorflow.proto.framework.SavedVariable.getDefaultInstance()) return this;
-      if (other.dtype_ != 0) {
-        setDtypeValue(other.getDtypeValue());
-      }
-      if (other.hasShape()) {
-        mergeShape(other.getShape());
-      }
-      if (other.getTrainable() != false) {
-        setTrainable(other.getTrainable());
-      }
-      if (other.synchronization_ != 0) {
-        setSynchronizationValue(other.getSynchronizationValue());
-      }
-      if (other.aggregation_ != 0) {
-        setAggregationValue(other.getAggregationValue());
-      }
-      if (!other.getName().isEmpty()) {
-        name_ = other.name_;
-        onChanged();
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.framework.SavedVariable parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.framework.SavedVariable) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-
-    private int dtype_ = 0;
-    /**
-     * <code>.tensorflow.DataType dtype = 1;</code>
-     */
-    public int getDtypeValue() {
-      return dtype_;
-    }
-    /**
-     * <code>.tensorflow.DataType dtype = 1;</code>
-     */
-    public Builder setDtypeValue(int value) {
-      dtype_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>.tensorflow.DataType dtype = 1;</code>
-     */
-    public org.tensorflow.proto.framework.DataType getDtype() {
-      @SuppressWarnings("deprecation")
-      org.tensorflow.proto.framework.DataType result = org.tensorflow.proto.framework.DataType.valueOf(dtype_);
-      return result == null ? org.tensorflow.proto.framework.DataType.UNRECOGNIZED : result;
-    }
-    /**
-     * <code>.tensorflow.DataType dtype = 1;</code>
-     */
-    public Builder setDtype(org.tensorflow.proto.framework.DataType value) {
-      if (value == null) {
-        throw new NullPointerException();
-      }
-      
-      dtype_ = value.getNumber();
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>.tensorflow.DataType dtype = 1;</code>
-     */
-    public Builder clearDtype() {
-      
-      dtype_ = 0;
-      onChanged();
-      return this;
-    }
-
-    private org.tensorflow.proto.framework.TensorShapeProto shape_;
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.TensorShapeProto, org.tensorflow.proto.framework.TensorShapeProto.Builder, org.tensorflow.proto.framework.TensorShapeProtoOrBuilder> shapeBuilder_;
-    /**
-     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-     */
-    public boolean hasShape() {
-      return shapeBuilder_ != null || shape_ != null;
-    }
-    /**
-     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-     */
-    public org.tensorflow.proto.framework.TensorShapeProto getShape() {
-      if (shapeBuilder_ == null) {
-        return shape_ == null ? org.tensorflow.proto.framework.TensorShapeProto.getDefaultInstance() : shape_;
-      } else {
-        return shapeBuilder_.getMessage();
-      }
-    }
-    /**
-     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-     */
-    public Builder setShape(org.tensorflow.proto.framework.TensorShapeProto value) {
-      if (shapeBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        shape_ = value;
-        onChanged();
-      } else {
-        shapeBuilder_.setMessage(value);
-      }
-
-      return this;
-    }
-    /**
-     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-     */
-    public Builder setShape(
-        org.tensorflow.proto.framework.TensorShapeProto.Builder builderForValue) {
-      if (shapeBuilder_ == null) {
-        shape_ = builderForValue.build();
-        onChanged();
-      } else {
-        shapeBuilder_.setMessage(builderForValue.build());
-      }
-
-      return this;
-    }
-    /**
-     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-     */
-    public Builder mergeShape(org.tensorflow.proto.framework.TensorShapeProto value) {
-      if (shapeBuilder_ == null) {
-        if (shape_ != null) {
-          shape_ =
-            org.tensorflow.proto.framework.TensorShapeProto.newBuilder(shape_).mergeFrom(value).buildPartial();
-        } else {
-          shape_ = value;
-        }
-        onChanged();
-      } else {
-        shapeBuilder_.mergeFrom(value);
-      }
-
-      return this;
-    }
-    /**
-     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-     */
-    public Builder clearShape() {
-      if (shapeBuilder_ == null) {
-        shape_ = null;
-        onChanged();
-      } else {
-        shape_ = null;
-        shapeBuilder_ = null;
-      }
-
-      return this;
-    }
-    /**
-     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-     */
-    public org.tensorflow.proto.framework.TensorShapeProto.Builder getShapeBuilder() {
-      
-      onChanged();
-      return getShapeFieldBuilder().getBuilder();
-    }
-    /**
-     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-     */
-    public org.tensorflow.proto.framework.TensorShapeProtoOrBuilder getShapeOrBuilder() {
-      if (shapeBuilder_ != null) {
-        return shapeBuilder_.getMessageOrBuilder();
-      } else {
-        return shape_ == null ?
-            org.tensorflow.proto.framework.TensorShapeProto.getDefaultInstance() : shape_;
-      }
-    }
-    /**
-     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-     */
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.TensorShapeProto, org.tensorflow.proto.framework.TensorShapeProto.Builder, org.tensorflow.proto.framework.TensorShapeProtoOrBuilder> 
-        getShapeFieldBuilder() {
-      if (shapeBuilder_ == null) {
-        shapeBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-            org.tensorflow.proto.framework.TensorShapeProto, org.tensorflow.proto.framework.TensorShapeProto.Builder, org.tensorflow.proto.framework.TensorShapeProtoOrBuilder>(
-                getShape(),
-                getParentForChildren(),
-                isClean());
-        shape_ = null;
-      }
-      return shapeBuilder_;
-    }
-
-    private boolean trainable_ ;
-    /**
-     * <code>bool trainable = 3;</code>
-     */
-    public boolean getTrainable() {
-      return trainable_;
-    }
-    /**
-     * <code>bool trainable = 3;</code>
-     */
-    public Builder setTrainable(boolean value) {
-      
-      trainable_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>bool trainable = 3;</code>
-     */
-    public Builder clearTrainable() {
-      
-      trainable_ = false;
-      onChanged();
-      return this;
-    }
-
-    private int synchronization_ = 0;
-    /**
-     * <code>.tensorflow.VariableSynchronization synchronization = 4;</code>
-     */
-    public int getSynchronizationValue() {
-      return synchronization_;
-    }
-    /**
-     * <code>.tensorflow.VariableSynchronization synchronization = 4;</code>
-     */
-    public Builder setSynchronizationValue(int value) {
-      synchronization_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>.tensorflow.VariableSynchronization synchronization = 4;</code>
-     */
-    public org.tensorflow.proto.framework.VariableSynchronization getSynchronization() {
-      @SuppressWarnings("deprecation")
-      org.tensorflow.proto.framework.VariableSynchronization result = org.tensorflow.proto.framework.VariableSynchronization.valueOf(synchronization_);
-      return result == null ? org.tensorflow.proto.framework.VariableSynchronization.UNRECOGNIZED : result;
-    }
-    /**
-     * <code>.tensorflow.VariableSynchronization synchronization = 4;</code>
-     */
-    public Builder setSynchronization(org.tensorflow.proto.framework.VariableSynchronization value) {
-      if (value == null) {
-        throw new NullPointerException();
-      }
-      
-      synchronization_ = value.getNumber();
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>.tensorflow.VariableSynchronization synchronization = 4;</code>
-     */
-    public Builder clearSynchronization() {
-      
-      synchronization_ = 0;
-      onChanged();
-      return this;
-    }
-
-    private int aggregation_ = 0;
-    /**
-     * <code>.tensorflow.VariableAggregation aggregation = 5;</code>
-     */
-    public int getAggregationValue() {
-      return aggregation_;
-    }
-    /**
-     * <code>.tensorflow.VariableAggregation aggregation = 5;</code>
-     */
-    public Builder setAggregationValue(int value) {
-      aggregation_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>.tensorflow.VariableAggregation aggregation = 5;</code>
-     */
-    public org.tensorflow.proto.framework.VariableAggregation getAggregation() {
-      @SuppressWarnings("deprecation")
-      org.tensorflow.proto.framework.VariableAggregation result = org.tensorflow.proto.framework.VariableAggregation.valueOf(aggregation_);
-      return result == null ? org.tensorflow.proto.framework.VariableAggregation.UNRECOGNIZED : result;
-    }
-    /**
-     * <code>.tensorflow.VariableAggregation aggregation = 5;</code>
-     */
-    public Builder setAggregation(org.tensorflow.proto.framework.VariableAggregation value) {
-      if (value == null) {
-        throw new NullPointerException();
-      }
-      
-      aggregation_ = value.getNumber();
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>.tensorflow.VariableAggregation aggregation = 5;</code>
-     */
-    public Builder clearAggregation() {
-      
-      aggregation_ = 0;
-      onChanged();
-      return this;
-    }
-
-    private java.lang.Object name_ = "";
-    /**
-     * <code>string name = 6;</code>
-     */
-    public java.lang.String getName() {
-      java.lang.Object ref = name_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        name_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <code>string name = 6;</code>
-     */
-    public com.google.protobuf.ByteString
-        getNameBytes() {
-      java.lang.Object ref = name_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        name_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <code>string name = 6;</code>
-     */
-    public Builder setName(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      name_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>string name = 6;</code>
-     */
-    public Builder clearName() {
-      
-      name_ = getDefaultInstance().getName();
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>string name = 6;</code>
-     */
-    public Builder setNameBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      name_ = value;
-      onChanged();
-      return this;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.SavedVariable)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.SavedVariable)
-  private static final org.tensorflow.proto.framework.SavedVariable DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.SavedVariable();
-  }
-
-  public static org.tensorflow.proto.framework.SavedVariable getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<SavedVariable>
-      PARSER = new com.google.protobuf.AbstractParser<SavedVariable>() {
-    @java.lang.Override
-    public SavedVariable parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new SavedVariable(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<SavedVariable> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<SavedVariable> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.framework.SavedVariable getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedVariableOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedVariableOrBuilder.java
deleted file mode 100644
index 27da7b6abfa..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedVariableOrBuilder.java
+++ /dev/null
@@ -1,64 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/saved_object_graph.proto
-
-package org.tensorflow.proto.framework;
-
-public interface SavedVariableOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.SavedVariable)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <code>.tensorflow.DataType dtype = 1;</code>
-   */
-  int getDtypeValue();
-  /**
-   * <code>.tensorflow.DataType dtype = 1;</code>
-   */
-  org.tensorflow.proto.framework.DataType getDtype();
-
-  /**
-   * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-   */
-  boolean hasShape();
-  /**
-   * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-   */
-  org.tensorflow.proto.framework.TensorShapeProto getShape();
-  /**
-   * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-   */
-  org.tensorflow.proto.framework.TensorShapeProtoOrBuilder getShapeOrBuilder();
-
-  /**
-   * <code>bool trainable = 3;</code>
-   */
-  boolean getTrainable();
-
-  /**
-   * <code>.tensorflow.VariableSynchronization synchronization = 4;</code>
-   */
-  int getSynchronizationValue();
-  /**
-   * <code>.tensorflow.VariableSynchronization synchronization = 4;</code>
-   */
-  org.tensorflow.proto.framework.VariableSynchronization getSynchronization();
-
-  /**
-   * <code>.tensorflow.VariableAggregation aggregation = 5;</code>
-   */
-  int getAggregationValue();
-  /**
-   * <code>.tensorflow.VariableAggregation aggregation = 5;</code>
-   */
-  org.tensorflow.proto.framework.VariableAggregation getAggregation();
-
-  /**
-   * <code>string name = 6;</code>
-   */
-  java.lang.String getName();
-  /**
-   * <code>string name = 6;</code>
-   */
-  com.google.protobuf.ByteString
-      getNameBytes();
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ScopedAllocatorOptions.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ScopedAllocatorOptions.java
deleted file mode 100644
index 37bf6f9c4d1..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ScopedAllocatorOptions.java
+++ /dev/null
@@ -1,644 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/rewriter_config.proto
-
-package org.tensorflow.proto.framework;
-
-/**
- * Protobuf type {@code tensorflow.ScopedAllocatorOptions}
- */
-public  final class ScopedAllocatorOptions extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.ScopedAllocatorOptions)
-    ScopedAllocatorOptionsOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use ScopedAllocatorOptions.newBuilder() to construct.
-  private ScopedAllocatorOptions(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private ScopedAllocatorOptions() {
-    enableOp_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new ScopedAllocatorOptions();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private ScopedAllocatorOptions(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    int mutable_bitField0_ = 0;
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 10: {
-            java.lang.String s = input.readStringRequireUtf8();
-            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
-              enableOp_ = new com.google.protobuf.LazyStringArrayList();
-              mutable_bitField0_ |= 0x00000001;
-            }
-            enableOp_.add(s);
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      if (((mutable_bitField0_ & 0x00000001) != 0)) {
-        enableOp_ = enableOp_.getUnmodifiableView();
-      }
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.framework.RewriterConfigProtos.internal_static_tensorflow_ScopedAllocatorOptions_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.framework.RewriterConfigProtos.internal_static_tensorflow_ScopedAllocatorOptions_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.framework.ScopedAllocatorOptions.class, org.tensorflow.proto.framework.ScopedAllocatorOptions.Builder.class);
-  }
-
-  public static final int ENABLE_OP_FIELD_NUMBER = 1;
-  private com.google.protobuf.LazyStringList enableOp_;
-  /**
-   * <pre>
-   * If present, only perform optimization for these ops.
-   * </pre>
-   *
-   * <code>repeated string enable_op = 1;</code>
-   */
-  public com.google.protobuf.ProtocolStringList
-      getEnableOpList() {
-    return enableOp_;
-  }
-  /**
-   * <pre>
-   * If present, only perform optimization for these ops.
-   * </pre>
-   *
-   * <code>repeated string enable_op = 1;</code>
-   */
-  public int getEnableOpCount() {
-    return enableOp_.size();
-  }
-  /**
-   * <pre>
-   * If present, only perform optimization for these ops.
-   * </pre>
-   *
-   * <code>repeated string enable_op = 1;</code>
-   */
-  public java.lang.String getEnableOp(int index) {
-    return enableOp_.get(index);
-  }
-  /**
-   * <pre>
-   * If present, only perform optimization for these ops.
-   * </pre>
-   *
-   * <code>repeated string enable_op = 1;</code>
-   */
-  public com.google.protobuf.ByteString
-      getEnableOpBytes(int index) {
-    return enableOp_.getByteString(index);
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    for (int i = 0; i < enableOp_.size(); i++) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 1, enableOp_.getRaw(i));
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    {
-      int dataSize = 0;
-      for (int i = 0; i < enableOp_.size(); i++) {
-        dataSize += computeStringSizeNoTag(enableOp_.getRaw(i));
-      }
-      size += dataSize;
-      size += 1 * getEnableOpList().size();
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.framework.ScopedAllocatorOptions)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.framework.ScopedAllocatorOptions other = (org.tensorflow.proto.framework.ScopedAllocatorOptions) obj;
-
-    if (!getEnableOpList()
-        .equals(other.getEnableOpList())) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    if (getEnableOpCount() > 0) {
-      hash = (37 * hash) + ENABLE_OP_FIELD_NUMBER;
-      hash = (53 * hash) + getEnableOpList().hashCode();
-    }
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.framework.ScopedAllocatorOptions parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.ScopedAllocatorOptions parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.ScopedAllocatorOptions parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.ScopedAllocatorOptions parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.ScopedAllocatorOptions parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.ScopedAllocatorOptions parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.ScopedAllocatorOptions parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.ScopedAllocatorOptions parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.ScopedAllocatorOptions parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.ScopedAllocatorOptions parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.ScopedAllocatorOptions parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.ScopedAllocatorOptions parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.framework.ScopedAllocatorOptions prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * Protobuf type {@code tensorflow.ScopedAllocatorOptions}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.ScopedAllocatorOptions)
-      org.tensorflow.proto.framework.ScopedAllocatorOptionsOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.RewriterConfigProtos.internal_static_tensorflow_ScopedAllocatorOptions_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.RewriterConfigProtos.internal_static_tensorflow_ScopedAllocatorOptions_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.ScopedAllocatorOptions.class, org.tensorflow.proto.framework.ScopedAllocatorOptions.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.framework.ScopedAllocatorOptions.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      enableOp_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-      bitField0_ = (bitField0_ & ~0x00000001);
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.framework.RewriterConfigProtos.internal_static_tensorflow_ScopedAllocatorOptions_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.ScopedAllocatorOptions getDefaultInstanceForType() {
-      return org.tensorflow.proto.framework.ScopedAllocatorOptions.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.ScopedAllocatorOptions build() {
-      org.tensorflow.proto.framework.ScopedAllocatorOptions result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.ScopedAllocatorOptions buildPartial() {
-      org.tensorflow.proto.framework.ScopedAllocatorOptions result = new org.tensorflow.proto.framework.ScopedAllocatorOptions(this);
-      int from_bitField0_ = bitField0_;
-      if (((bitField0_ & 0x00000001) != 0)) {
-        enableOp_ = enableOp_.getUnmodifiableView();
-        bitField0_ = (bitField0_ & ~0x00000001);
-      }
-      result.enableOp_ = enableOp_;
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.framework.ScopedAllocatorOptions) {
-        return mergeFrom((org.tensorflow.proto.framework.ScopedAllocatorOptions)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.framework.ScopedAllocatorOptions other) {
-      if (other == org.tensorflow.proto.framework.ScopedAllocatorOptions.getDefaultInstance()) return this;
-      if (!other.enableOp_.isEmpty()) {
-        if (enableOp_.isEmpty()) {
-          enableOp_ = other.enableOp_;
-          bitField0_ = (bitField0_ & ~0x00000001);
-        } else {
-          ensureEnableOpIsMutable();
-          enableOp_.addAll(other.enableOp_);
-        }
-        onChanged();
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.framework.ScopedAllocatorOptions parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.framework.ScopedAllocatorOptions) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-    private int bitField0_;
-
-    private com.google.protobuf.LazyStringList enableOp_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-    private void ensureEnableOpIsMutable() {
-      if (!((bitField0_ & 0x00000001) != 0)) {
-        enableOp_ = new com.google.protobuf.LazyStringArrayList(enableOp_);
-        bitField0_ |= 0x00000001;
-       }
-    }
-    /**
-     * <pre>
-     * If present, only perform optimization for these ops.
-     * </pre>
-     *
-     * <code>repeated string enable_op = 1;</code>
-     */
-    public com.google.protobuf.ProtocolStringList
-        getEnableOpList() {
-      return enableOp_.getUnmodifiableView();
-    }
-    /**
-     * <pre>
-     * If present, only perform optimization for these ops.
-     * </pre>
-     *
-     * <code>repeated string enable_op = 1;</code>
-     */
-    public int getEnableOpCount() {
-      return enableOp_.size();
-    }
-    /**
-     * <pre>
-     * If present, only perform optimization for these ops.
-     * </pre>
-     *
-     * <code>repeated string enable_op = 1;</code>
-     */
-    public java.lang.String getEnableOp(int index) {
-      return enableOp_.get(index);
-    }
-    /**
-     * <pre>
-     * If present, only perform optimization for these ops.
-     * </pre>
-     *
-     * <code>repeated string enable_op = 1;</code>
-     */
-    public com.google.protobuf.ByteString
-        getEnableOpBytes(int index) {
-      return enableOp_.getByteString(index);
-    }
-    /**
-     * <pre>
-     * If present, only perform optimization for these ops.
-     * </pre>
-     *
-     * <code>repeated string enable_op = 1;</code>
-     */
-    public Builder setEnableOp(
-        int index, java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  ensureEnableOpIsMutable();
-      enableOp_.set(index, value);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * If present, only perform optimization for these ops.
-     * </pre>
-     *
-     * <code>repeated string enable_op = 1;</code>
-     */
-    public Builder addEnableOp(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  ensureEnableOpIsMutable();
-      enableOp_.add(value);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * If present, only perform optimization for these ops.
-     * </pre>
-     *
-     * <code>repeated string enable_op = 1;</code>
-     */
-    public Builder addAllEnableOp(
-        java.lang.Iterable<java.lang.String> values) {
-      ensureEnableOpIsMutable();
-      com.google.protobuf.AbstractMessageLite.Builder.addAll(
-          values, enableOp_);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * If present, only perform optimization for these ops.
-     * </pre>
-     *
-     * <code>repeated string enable_op = 1;</code>
-     */
-    public Builder clearEnableOp() {
-      enableOp_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-      bitField0_ = (bitField0_ & ~0x00000001);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * If present, only perform optimization for these ops.
-     * </pre>
-     *
-     * <code>repeated string enable_op = 1;</code>
-     */
-    public Builder addEnableOpBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      ensureEnableOpIsMutable();
-      enableOp_.add(value);
-      onChanged();
-      return this;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.ScopedAllocatorOptions)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.ScopedAllocatorOptions)
-  private static final org.tensorflow.proto.framework.ScopedAllocatorOptions DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.ScopedAllocatorOptions();
-  }
-
-  public static org.tensorflow.proto.framework.ScopedAllocatorOptions getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<ScopedAllocatorOptions>
-      PARSER = new com.google.protobuf.AbstractParser<ScopedAllocatorOptions>() {
-    @java.lang.Override
-    public ScopedAllocatorOptions parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new ScopedAllocatorOptions(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<ScopedAllocatorOptions> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<ScopedAllocatorOptions> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.framework.ScopedAllocatorOptions getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ScopedAllocatorOptionsOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ScopedAllocatorOptionsOrBuilder.java
deleted file mode 100644
index df536ed18a9..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ScopedAllocatorOptionsOrBuilder.java
+++ /dev/null
@@ -1,44 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/rewriter_config.proto
-
-package org.tensorflow.proto.framework;
-
-public interface ScopedAllocatorOptionsOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.ScopedAllocatorOptions)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <pre>
-   * If present, only perform optimization for these ops.
-   * </pre>
-   *
-   * <code>repeated string enable_op = 1;</code>
-   */
-  java.util.List<java.lang.String>
-      getEnableOpList();
-  /**
-   * <pre>
-   * If present, only perform optimization for these ops.
-   * </pre>
-   *
-   * <code>repeated string enable_op = 1;</code>
-   */
-  int getEnableOpCount();
-  /**
-   * <pre>
-   * If present, only perform optimization for these ops.
-   * </pre>
-   *
-   * <code>repeated string enable_op = 1;</code>
-   */
-  java.lang.String getEnableOp(int index);
-  /**
-   * <pre>
-   * If present, only perform optimization for these ops.
-   * </pre>
-   *
-   * <code>repeated string enable_op = 1;</code>
-   */
-  com.google.protobuf.ByteString
-      getEnableOpBytes(int index);
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SessionMetadata.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SessionMetadata.java
deleted file mode 100644
index a825437b468..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SessionMetadata.java
+++ /dev/null
@@ -1,636 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/config.proto
-
-package org.tensorflow.proto.framework;
-
-/**
- * <pre>
- * Metadata about the session.
- * This can be used by the runtime and the Ops for debugging, monitoring, etc.
- * The (name, version) tuple is expected to be a unique identifier for
- * sessions within the same process.
- * NOTE: This is currently used and propagated only by the direct session.
- * </pre>
- *
- * Protobuf type {@code tensorflow.SessionMetadata}
- */
-public  final class SessionMetadata extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.SessionMetadata)
-    SessionMetadataOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use SessionMetadata.newBuilder() to construct.
-  private SessionMetadata(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private SessionMetadata() {
-    name_ = "";
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new SessionMetadata();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private SessionMetadata(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 10: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            name_ = s;
-            break;
-          }
-          case 16: {
-
-            version_ = input.readInt64();
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_SessionMetadata_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_SessionMetadata_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.framework.SessionMetadata.class, org.tensorflow.proto.framework.SessionMetadata.Builder.class);
-  }
-
-  public static final int NAME_FIELD_NUMBER = 1;
-  private volatile java.lang.Object name_;
-  /**
-   * <code>string name = 1;</code>
-   */
-  public java.lang.String getName() {
-    java.lang.Object ref = name_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      name_ = s;
-      return s;
-    }
-  }
-  /**
-   * <code>string name = 1;</code>
-   */
-  public com.google.protobuf.ByteString
-      getNameBytes() {
-    java.lang.Object ref = name_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      name_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int VERSION_FIELD_NUMBER = 2;
-  private long version_;
-  /**
-   * <pre>
-   * The version is optional. If set, needs to be &gt;= 0.
-   * </pre>
-   *
-   * <code>int64 version = 2;</code>
-   */
-  public long getVersion() {
-    return version_;
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    if (!getNameBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 1, name_);
-    }
-    if (version_ != 0L) {
-      output.writeInt64(2, version_);
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (!getNameBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, name_);
-    }
-    if (version_ != 0L) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt64Size(2, version_);
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.framework.SessionMetadata)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.framework.SessionMetadata other = (org.tensorflow.proto.framework.SessionMetadata) obj;
-
-    if (!getName()
-        .equals(other.getName())) return false;
-    if (getVersion()
-        != other.getVersion()) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    hash = (37 * hash) + NAME_FIELD_NUMBER;
-    hash = (53 * hash) + getName().hashCode();
-    hash = (37 * hash) + VERSION_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-        getVersion());
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.framework.SessionMetadata parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.SessionMetadata parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.SessionMetadata parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.SessionMetadata parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.SessionMetadata parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.SessionMetadata parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.SessionMetadata parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.SessionMetadata parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.SessionMetadata parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.SessionMetadata parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.SessionMetadata parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.SessionMetadata parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.framework.SessionMetadata prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * <pre>
-   * Metadata about the session.
-   * This can be used by the runtime and the Ops for debugging, monitoring, etc.
-   * The (name, version) tuple is expected to be a unique identifier for
-   * sessions within the same process.
-   * NOTE: This is currently used and propagated only by the direct session.
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.SessionMetadata}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.SessionMetadata)
-      org.tensorflow.proto.framework.SessionMetadataOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_SessionMetadata_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_SessionMetadata_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.SessionMetadata.class, org.tensorflow.proto.framework.SessionMetadata.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.framework.SessionMetadata.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      name_ = "";
-
-      version_ = 0L;
-
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_SessionMetadata_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.SessionMetadata getDefaultInstanceForType() {
-      return org.tensorflow.proto.framework.SessionMetadata.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.SessionMetadata build() {
-      org.tensorflow.proto.framework.SessionMetadata result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.SessionMetadata buildPartial() {
-      org.tensorflow.proto.framework.SessionMetadata result = new org.tensorflow.proto.framework.SessionMetadata(this);
-      result.name_ = name_;
-      result.version_ = version_;
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.framework.SessionMetadata) {
-        return mergeFrom((org.tensorflow.proto.framework.SessionMetadata)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.framework.SessionMetadata other) {
-      if (other == org.tensorflow.proto.framework.SessionMetadata.getDefaultInstance()) return this;
-      if (!other.getName().isEmpty()) {
-        name_ = other.name_;
-        onChanged();
-      }
-      if (other.getVersion() != 0L) {
-        setVersion(other.getVersion());
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.framework.SessionMetadata parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.framework.SessionMetadata) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-
-    private java.lang.Object name_ = "";
-    /**
-     * <code>string name = 1;</code>
-     */
-    public java.lang.String getName() {
-      java.lang.Object ref = name_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        name_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <code>string name = 1;</code>
-     */
-    public com.google.protobuf.ByteString
-        getNameBytes() {
-      java.lang.Object ref = name_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        name_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <code>string name = 1;</code>
-     */
-    public Builder setName(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      name_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>string name = 1;</code>
-     */
-    public Builder clearName() {
-      
-      name_ = getDefaultInstance().getName();
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>string name = 1;</code>
-     */
-    public Builder setNameBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      name_ = value;
-      onChanged();
-      return this;
-    }
-
-    private long version_ ;
-    /**
-     * <pre>
-     * The version is optional. If set, needs to be &gt;= 0.
-     * </pre>
-     *
-     * <code>int64 version = 2;</code>
-     */
-    public long getVersion() {
-      return version_;
-    }
-    /**
-     * <pre>
-     * The version is optional. If set, needs to be &gt;= 0.
-     * </pre>
-     *
-     * <code>int64 version = 2;</code>
-     */
-    public Builder setVersion(long value) {
-      
-      version_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * The version is optional. If set, needs to be &gt;= 0.
-     * </pre>
-     *
-     * <code>int64 version = 2;</code>
-     */
-    public Builder clearVersion() {
-      
-      version_ = 0L;
-      onChanged();
-      return this;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.SessionMetadata)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.SessionMetadata)
-  private static final org.tensorflow.proto.framework.SessionMetadata DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.SessionMetadata();
-  }
-
-  public static org.tensorflow.proto.framework.SessionMetadata getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<SessionMetadata>
-      PARSER = new com.google.protobuf.AbstractParser<SessionMetadata>() {
-    @java.lang.Override
-    public SessionMetadata parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new SessionMetadata(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<SessionMetadata> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<SessionMetadata> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.framework.SessionMetadata getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SessionMetadataOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SessionMetadataOrBuilder.java
deleted file mode 100644
index eae7295e772..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SessionMetadataOrBuilder.java
+++ /dev/null
@@ -1,28 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/config.proto
-
-package org.tensorflow.proto.framework;
-
-public interface SessionMetadataOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.SessionMetadata)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <code>string name = 1;</code>
-   */
-  java.lang.String getName();
-  /**
-   * <code>string name = 1;</code>
-   */
-  com.google.protobuf.ByteString
-      getNameBytes();
-
-  /**
-   * <pre>
-   * The version is optional. If set, needs to be &gt;= 0.
-   * </pre>
-   *
-   * <code>int64 version = 2;</code>
-   */
-  long getVersion();
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SignatureDef.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SignatureDef.java
deleted file mode 100644
index fe588917c86..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SignatureDef.java
+++ /dev/null
@@ -1,1339 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/meta_graph.proto
-
-package org.tensorflow.proto.framework;
-
-/**
- * <pre>
- * SignatureDef defines the signature of a computation supported by a TensorFlow
- * graph.
- * For example, a model with two loss computations, sharing a single input,
- * might have the following signature_def map.
- * Note that across the two SignatureDefs "loss_A" and "loss_B", the input key,
- * output key, and method_name are identical, and will be used by system(s) that
- * implement or rely upon this particular loss method. The output tensor names
- * differ, demonstrating how different outputs can exist for the same method.
- * signature_def {
- *   key: "loss_A"
- *   value {
- *     inputs {
- *       key: "input"
- *       value {
- *         name: "input:0"
- *         dtype: DT_STRING
- *         tensor_shape: ...
- *       }
- *     }
- *     outputs {
- *       key: "loss_output"
- *       value {
- *         name: "loss_output_A:0"
- *         dtype: DT_FLOAT
- *         tensor_shape: ...
- *       }
- *     }
- *   }
- *   ...
- *   method_name: "some/package/compute_loss"
- * }
- * signature_def {
- *   key: "loss_B"
- *   value {
- *     inputs {
- *       key: "input"
- *       value {
- *         name: "input:0"
- *         dtype: DT_STRING
- *         tensor_shape: ...
- *       }
- *     }
- *     outputs {
- *       key: "loss_output"
- *       value {
- *         name: "loss_output_B:0"
- *         dtype: DT_FLOAT
- *         tensor_shape: ...
- *       }
- *     }
- *   }
- *   ...
- *   method_name: "some/package/compute_loss"
- * }
- * </pre>
- *
- * Protobuf type {@code tensorflow.SignatureDef}
- */
-public  final class SignatureDef extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.SignatureDef)
-    SignatureDefOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use SignatureDef.newBuilder() to construct.
-  private SignatureDef(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private SignatureDef() {
-    methodName_ = "";
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new SignatureDef();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private SignatureDef(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    int mutable_bitField0_ = 0;
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 10: {
-            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
-              inputs_ = com.google.protobuf.MapField.newMapField(
-                  InputsDefaultEntryHolder.defaultEntry);
-              mutable_bitField0_ |= 0x00000001;
-            }
-            com.google.protobuf.MapEntry<java.lang.String, org.tensorflow.proto.framework.TensorInfo>
-            inputs__ = input.readMessage(
-                InputsDefaultEntryHolder.defaultEntry.getParserForType(), extensionRegistry);
-            inputs_.getMutableMap().put(
-                inputs__.getKey(), inputs__.getValue());
-            break;
-          }
-          case 18: {
-            if (!((mutable_bitField0_ & 0x00000002) != 0)) {
-              outputs_ = com.google.protobuf.MapField.newMapField(
-                  OutputsDefaultEntryHolder.defaultEntry);
-              mutable_bitField0_ |= 0x00000002;
-            }
-            com.google.protobuf.MapEntry<java.lang.String, org.tensorflow.proto.framework.TensorInfo>
-            outputs__ = input.readMessage(
-                OutputsDefaultEntryHolder.defaultEntry.getParserForType(), extensionRegistry);
-            outputs_.getMutableMap().put(
-                outputs__.getKey(), outputs__.getValue());
-            break;
-          }
-          case 26: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            methodName_ = s;
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.framework.MetaGraphProtos.internal_static_tensorflow_SignatureDef_descriptor;
-  }
-
-  @SuppressWarnings({"rawtypes"})
-  @java.lang.Override
-  protected com.google.protobuf.MapField internalGetMapField(
-      int number) {
-    switch (number) {
-      case 1:
-        return internalGetInputs();
-      case 2:
-        return internalGetOutputs();
-      default:
-        throw new RuntimeException(
-            "Invalid map field number: " + number);
-    }
-  }
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.framework.MetaGraphProtos.internal_static_tensorflow_SignatureDef_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.framework.SignatureDef.class, org.tensorflow.proto.framework.SignatureDef.Builder.class);
-  }
-
-  public static final int INPUTS_FIELD_NUMBER = 1;
-  private static final class InputsDefaultEntryHolder {
-    static final com.google.protobuf.MapEntry<
-        java.lang.String, org.tensorflow.proto.framework.TensorInfo> defaultEntry =
-            com.google.protobuf.MapEntry
-            .<java.lang.String, org.tensorflow.proto.framework.TensorInfo>newDefaultInstance(
-                org.tensorflow.proto.framework.MetaGraphProtos.internal_static_tensorflow_SignatureDef_InputsEntry_descriptor, 
-                com.google.protobuf.WireFormat.FieldType.STRING,
-                "",
-                com.google.protobuf.WireFormat.FieldType.MESSAGE,
-                org.tensorflow.proto.framework.TensorInfo.getDefaultInstance());
-  }
-  private com.google.protobuf.MapField<
-      java.lang.String, org.tensorflow.proto.framework.TensorInfo> inputs_;
-  private com.google.protobuf.MapField<java.lang.String, org.tensorflow.proto.framework.TensorInfo>
-  internalGetInputs() {
-    if (inputs_ == null) {
-      return com.google.protobuf.MapField.emptyMapField(
-          InputsDefaultEntryHolder.defaultEntry);
-    }
-    return inputs_;
-  }
-
-  public int getInputsCount() {
-    return internalGetInputs().getMap().size();
-  }
-  /**
-   * <pre>
-   * Named input parameters.
-   * </pre>
-   *
-   * <code>map&lt;string, .tensorflow.TensorInfo&gt; inputs = 1;</code>
-   */
-
-  public boolean containsInputs(
-      java.lang.String key) {
-    if (key == null) { throw new java.lang.NullPointerException(); }
-    return internalGetInputs().getMap().containsKey(key);
-  }
-  /**
-   * Use {@link #getInputsMap()} instead.
-   */
-  @java.lang.Deprecated
-  public java.util.Map<java.lang.String, org.tensorflow.proto.framework.TensorInfo> getInputs() {
-    return getInputsMap();
-  }
-  /**
-   * <pre>
-   * Named input parameters.
-   * </pre>
-   *
-   * <code>map&lt;string, .tensorflow.TensorInfo&gt; inputs = 1;</code>
-   */
-
-  public java.util.Map<java.lang.String, org.tensorflow.proto.framework.TensorInfo> getInputsMap() {
-    return internalGetInputs().getMap();
-  }
-  /**
-   * <pre>
-   * Named input parameters.
-   * </pre>
-   *
-   * <code>map&lt;string, .tensorflow.TensorInfo&gt; inputs = 1;</code>
-   */
-
-  public org.tensorflow.proto.framework.TensorInfo getInputsOrDefault(
-      java.lang.String key,
-      org.tensorflow.proto.framework.TensorInfo defaultValue) {
-    if (key == null) { throw new java.lang.NullPointerException(); }
-    java.util.Map<java.lang.String, org.tensorflow.proto.framework.TensorInfo> map =
-        internalGetInputs().getMap();
-    return map.containsKey(key) ? map.get(key) : defaultValue;
-  }
-  /**
-   * <pre>
-   * Named input parameters.
-   * </pre>
-   *
-   * <code>map&lt;string, .tensorflow.TensorInfo&gt; inputs = 1;</code>
-   */
-
-  public org.tensorflow.proto.framework.TensorInfo getInputsOrThrow(
-      java.lang.String key) {
-    if (key == null) { throw new java.lang.NullPointerException(); }
-    java.util.Map<java.lang.String, org.tensorflow.proto.framework.TensorInfo> map =
-        internalGetInputs().getMap();
-    if (!map.containsKey(key)) {
-      throw new java.lang.IllegalArgumentException();
-    }
-    return map.get(key);
-  }
-
-  public static final int OUTPUTS_FIELD_NUMBER = 2;
-  private static final class OutputsDefaultEntryHolder {
-    static final com.google.protobuf.MapEntry<
-        java.lang.String, org.tensorflow.proto.framework.TensorInfo> defaultEntry =
-            com.google.protobuf.MapEntry
-            .<java.lang.String, org.tensorflow.proto.framework.TensorInfo>newDefaultInstance(
-                org.tensorflow.proto.framework.MetaGraphProtos.internal_static_tensorflow_SignatureDef_OutputsEntry_descriptor, 
-                com.google.protobuf.WireFormat.FieldType.STRING,
-                "",
-                com.google.protobuf.WireFormat.FieldType.MESSAGE,
-                org.tensorflow.proto.framework.TensorInfo.getDefaultInstance());
-  }
-  private com.google.protobuf.MapField<
-      java.lang.String, org.tensorflow.proto.framework.TensorInfo> outputs_;
-  private com.google.protobuf.MapField<java.lang.String, org.tensorflow.proto.framework.TensorInfo>
-  internalGetOutputs() {
-    if (outputs_ == null) {
-      return com.google.protobuf.MapField.emptyMapField(
-          OutputsDefaultEntryHolder.defaultEntry);
-    }
-    return outputs_;
-  }
-
-  public int getOutputsCount() {
-    return internalGetOutputs().getMap().size();
-  }
-  /**
-   * <pre>
-   * Named output parameters.
-   * </pre>
-   *
-   * <code>map&lt;string, .tensorflow.TensorInfo&gt; outputs = 2;</code>
-   */
-
-  public boolean containsOutputs(
-      java.lang.String key) {
-    if (key == null) { throw new java.lang.NullPointerException(); }
-    return internalGetOutputs().getMap().containsKey(key);
-  }
-  /**
-   * Use {@link #getOutputsMap()} instead.
-   */
-  @java.lang.Deprecated
-  public java.util.Map<java.lang.String, org.tensorflow.proto.framework.TensorInfo> getOutputs() {
-    return getOutputsMap();
-  }
-  /**
-   * <pre>
-   * Named output parameters.
-   * </pre>
-   *
-   * <code>map&lt;string, .tensorflow.TensorInfo&gt; outputs = 2;</code>
-   */
-
-  public java.util.Map<java.lang.String, org.tensorflow.proto.framework.TensorInfo> getOutputsMap() {
-    return internalGetOutputs().getMap();
-  }
-  /**
-   * <pre>
-   * Named output parameters.
-   * </pre>
-   *
-   * <code>map&lt;string, .tensorflow.TensorInfo&gt; outputs = 2;</code>
-   */
-
-  public org.tensorflow.proto.framework.TensorInfo getOutputsOrDefault(
-      java.lang.String key,
-      org.tensorflow.proto.framework.TensorInfo defaultValue) {
-    if (key == null) { throw new java.lang.NullPointerException(); }
-    java.util.Map<java.lang.String, org.tensorflow.proto.framework.TensorInfo> map =
-        internalGetOutputs().getMap();
-    return map.containsKey(key) ? map.get(key) : defaultValue;
-  }
-  /**
-   * <pre>
-   * Named output parameters.
-   * </pre>
-   *
-   * <code>map&lt;string, .tensorflow.TensorInfo&gt; outputs = 2;</code>
-   */
-
-  public org.tensorflow.proto.framework.TensorInfo getOutputsOrThrow(
-      java.lang.String key) {
-    if (key == null) { throw new java.lang.NullPointerException(); }
-    java.util.Map<java.lang.String, org.tensorflow.proto.framework.TensorInfo> map =
-        internalGetOutputs().getMap();
-    if (!map.containsKey(key)) {
-      throw new java.lang.IllegalArgumentException();
-    }
-    return map.get(key);
-  }
-
-  public static final int METHOD_NAME_FIELD_NUMBER = 3;
-  private volatile java.lang.Object methodName_;
-  /**
-   * <pre>
-   * Extensible method_name information enabling third-party users to mark a
-   * SignatureDef as supporting a particular method. This enables producers and
-   * consumers of SignatureDefs, e.g. a model definition library and a serving
-   * library to have a clear hand-off regarding the semantics of a computation.
-   * Note that multiple SignatureDefs in a single MetaGraphDef may have the same
-   * method_name. This is commonly used to support multi-headed computation,
-   * where a single graph computation may return multiple results.
-   * </pre>
-   *
-   * <code>string method_name = 3;</code>
-   */
-  public java.lang.String getMethodName() {
-    java.lang.Object ref = methodName_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      methodName_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * Extensible method_name information enabling third-party users to mark a
-   * SignatureDef as supporting a particular method. This enables producers and
-   * consumers of SignatureDefs, e.g. a model definition library and a serving
-   * library to have a clear hand-off regarding the semantics of a computation.
-   * Note that multiple SignatureDefs in a single MetaGraphDef may have the same
-   * method_name. This is commonly used to support multi-headed computation,
-   * where a single graph computation may return multiple results.
-   * </pre>
-   *
-   * <code>string method_name = 3;</code>
-   */
-  public com.google.protobuf.ByteString
-      getMethodNameBytes() {
-    java.lang.Object ref = methodName_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      methodName_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    com.google.protobuf.GeneratedMessageV3
-      .serializeStringMapTo(
-        output,
-        internalGetInputs(),
-        InputsDefaultEntryHolder.defaultEntry,
-        1);
-    com.google.protobuf.GeneratedMessageV3
-      .serializeStringMapTo(
-        output,
-        internalGetOutputs(),
-        OutputsDefaultEntryHolder.defaultEntry,
-        2);
-    if (!getMethodNameBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 3, methodName_);
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    for (java.util.Map.Entry<java.lang.String, org.tensorflow.proto.framework.TensorInfo> entry
-         : internalGetInputs().getMap().entrySet()) {
-      com.google.protobuf.MapEntry<java.lang.String, org.tensorflow.proto.framework.TensorInfo>
-      inputs__ = InputsDefaultEntryHolder.defaultEntry.newBuilderForType()
-          .setKey(entry.getKey())
-          .setValue(entry.getValue())
-          .build();
-      size += com.google.protobuf.CodedOutputStream
-          .computeMessageSize(1, inputs__);
-    }
-    for (java.util.Map.Entry<java.lang.String, org.tensorflow.proto.framework.TensorInfo> entry
-         : internalGetOutputs().getMap().entrySet()) {
-      com.google.protobuf.MapEntry<java.lang.String, org.tensorflow.proto.framework.TensorInfo>
-      outputs__ = OutputsDefaultEntryHolder.defaultEntry.newBuilderForType()
-          .setKey(entry.getKey())
-          .setValue(entry.getValue())
-          .build();
-      size += com.google.protobuf.CodedOutputStream
-          .computeMessageSize(2, outputs__);
-    }
-    if (!getMethodNameBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(3, methodName_);
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.framework.SignatureDef)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.framework.SignatureDef other = (org.tensorflow.proto.framework.SignatureDef) obj;
-
-    if (!internalGetInputs().equals(
-        other.internalGetInputs())) return false;
-    if (!internalGetOutputs().equals(
-        other.internalGetOutputs())) return false;
-    if (!getMethodName()
-        .equals(other.getMethodName())) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    if (!internalGetInputs().getMap().isEmpty()) {
-      hash = (37 * hash) + INPUTS_FIELD_NUMBER;
-      hash = (53 * hash) + internalGetInputs().hashCode();
-    }
-    if (!internalGetOutputs().getMap().isEmpty()) {
-      hash = (37 * hash) + OUTPUTS_FIELD_NUMBER;
-      hash = (53 * hash) + internalGetOutputs().hashCode();
-    }
-    hash = (37 * hash) + METHOD_NAME_FIELD_NUMBER;
-    hash = (53 * hash) + getMethodName().hashCode();
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.framework.SignatureDef parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.SignatureDef parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.SignatureDef parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.SignatureDef parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.SignatureDef parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.SignatureDef parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.SignatureDef parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.SignatureDef parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.SignatureDef parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.SignatureDef parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.SignatureDef parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.SignatureDef parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.framework.SignatureDef prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * <pre>
-   * SignatureDef defines the signature of a computation supported by a TensorFlow
-   * graph.
-   * For example, a model with two loss computations, sharing a single input,
-   * might have the following signature_def map.
-   * Note that across the two SignatureDefs "loss_A" and "loss_B", the input key,
-   * output key, and method_name are identical, and will be used by system(s) that
-   * implement or rely upon this particular loss method. The output tensor names
-   * differ, demonstrating how different outputs can exist for the same method.
-   * signature_def {
-   *   key: "loss_A"
-   *   value {
-   *     inputs {
-   *       key: "input"
-   *       value {
-   *         name: "input:0"
-   *         dtype: DT_STRING
-   *         tensor_shape: ...
-   *       }
-   *     }
-   *     outputs {
-   *       key: "loss_output"
-   *       value {
-   *         name: "loss_output_A:0"
-   *         dtype: DT_FLOAT
-   *         tensor_shape: ...
-   *       }
-   *     }
-   *   }
-   *   ...
-   *   method_name: "some/package/compute_loss"
-   * }
-   * signature_def {
-   *   key: "loss_B"
-   *   value {
-   *     inputs {
-   *       key: "input"
-   *       value {
-   *         name: "input:0"
-   *         dtype: DT_STRING
-   *         tensor_shape: ...
-   *       }
-   *     }
-   *     outputs {
-   *       key: "loss_output"
-   *       value {
-   *         name: "loss_output_B:0"
-   *         dtype: DT_FLOAT
-   *         tensor_shape: ...
-   *       }
-   *     }
-   *   }
-   *   ...
-   *   method_name: "some/package/compute_loss"
-   * }
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.SignatureDef}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.SignatureDef)
-      org.tensorflow.proto.framework.SignatureDefOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.MetaGraphProtos.internal_static_tensorflow_SignatureDef_descriptor;
-    }
-
-    @SuppressWarnings({"rawtypes"})
-    protected com.google.protobuf.MapField internalGetMapField(
-        int number) {
-      switch (number) {
-        case 1:
-          return internalGetInputs();
-        case 2:
-          return internalGetOutputs();
-        default:
-          throw new RuntimeException(
-              "Invalid map field number: " + number);
-      }
-    }
-    @SuppressWarnings({"rawtypes"})
-    protected com.google.protobuf.MapField internalGetMutableMapField(
-        int number) {
-      switch (number) {
-        case 1:
-          return internalGetMutableInputs();
-        case 2:
-          return internalGetMutableOutputs();
-        default:
-          throw new RuntimeException(
-              "Invalid map field number: " + number);
-      }
-    }
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.MetaGraphProtos.internal_static_tensorflow_SignatureDef_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.SignatureDef.class, org.tensorflow.proto.framework.SignatureDef.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.framework.SignatureDef.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      internalGetMutableInputs().clear();
-      internalGetMutableOutputs().clear();
-      methodName_ = "";
-
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.framework.MetaGraphProtos.internal_static_tensorflow_SignatureDef_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.SignatureDef getDefaultInstanceForType() {
-      return org.tensorflow.proto.framework.SignatureDef.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.SignatureDef build() {
-      org.tensorflow.proto.framework.SignatureDef result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.SignatureDef buildPartial() {
-      org.tensorflow.proto.framework.SignatureDef result = new org.tensorflow.proto.framework.SignatureDef(this);
-      int from_bitField0_ = bitField0_;
-      result.inputs_ = internalGetInputs();
-      result.inputs_.makeImmutable();
-      result.outputs_ = internalGetOutputs();
-      result.outputs_.makeImmutable();
-      result.methodName_ = methodName_;
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.framework.SignatureDef) {
-        return mergeFrom((org.tensorflow.proto.framework.SignatureDef)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.framework.SignatureDef other) {
-      if (other == org.tensorflow.proto.framework.SignatureDef.getDefaultInstance()) return this;
-      internalGetMutableInputs().mergeFrom(
-          other.internalGetInputs());
-      internalGetMutableOutputs().mergeFrom(
-          other.internalGetOutputs());
-      if (!other.getMethodName().isEmpty()) {
-        methodName_ = other.methodName_;
-        onChanged();
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.framework.SignatureDef parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.framework.SignatureDef) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-    private int bitField0_;
-
-    private com.google.protobuf.MapField<
-        java.lang.String, org.tensorflow.proto.framework.TensorInfo> inputs_;
-    private com.google.protobuf.MapField<java.lang.String, org.tensorflow.proto.framework.TensorInfo>
-    internalGetInputs() {
-      if (inputs_ == null) {
-        return com.google.protobuf.MapField.emptyMapField(
-            InputsDefaultEntryHolder.defaultEntry);
-      }
-      return inputs_;
-    }
-    private com.google.protobuf.MapField<java.lang.String, org.tensorflow.proto.framework.TensorInfo>
-    internalGetMutableInputs() {
-      onChanged();;
-      if (inputs_ == null) {
-        inputs_ = com.google.protobuf.MapField.newMapField(
-            InputsDefaultEntryHolder.defaultEntry);
-      }
-      if (!inputs_.isMutable()) {
-        inputs_ = inputs_.copy();
-      }
-      return inputs_;
-    }
-
-    public int getInputsCount() {
-      return internalGetInputs().getMap().size();
-    }
-    /**
-     * <pre>
-     * Named input parameters.
-     * </pre>
-     *
-     * <code>map&lt;string, .tensorflow.TensorInfo&gt; inputs = 1;</code>
-     */
-
-    public boolean containsInputs(
-        java.lang.String key) {
-      if (key == null) { throw new java.lang.NullPointerException(); }
-      return internalGetInputs().getMap().containsKey(key);
-    }
-    /**
-     * Use {@link #getInputsMap()} instead.
-     */
-    @java.lang.Deprecated
-    public java.util.Map<java.lang.String, org.tensorflow.proto.framework.TensorInfo> getInputs() {
-      return getInputsMap();
-    }
-    /**
-     * <pre>
-     * Named input parameters.
-     * </pre>
-     *
-     * <code>map&lt;string, .tensorflow.TensorInfo&gt; inputs = 1;</code>
-     */
-
-    public java.util.Map<java.lang.String, org.tensorflow.proto.framework.TensorInfo> getInputsMap() {
-      return internalGetInputs().getMap();
-    }
-    /**
-     * <pre>
-     * Named input parameters.
-     * </pre>
-     *
-     * <code>map&lt;string, .tensorflow.TensorInfo&gt; inputs = 1;</code>
-     */
-
-    public org.tensorflow.proto.framework.TensorInfo getInputsOrDefault(
-        java.lang.String key,
-        org.tensorflow.proto.framework.TensorInfo defaultValue) {
-      if (key == null) { throw new java.lang.NullPointerException(); }
-      java.util.Map<java.lang.String, org.tensorflow.proto.framework.TensorInfo> map =
-          internalGetInputs().getMap();
-      return map.containsKey(key) ? map.get(key) : defaultValue;
-    }
-    /**
-     * <pre>
-     * Named input parameters.
-     * </pre>
-     *
-     * <code>map&lt;string, .tensorflow.TensorInfo&gt; inputs = 1;</code>
-     */
-
-    public org.tensorflow.proto.framework.TensorInfo getInputsOrThrow(
-        java.lang.String key) {
-      if (key == null) { throw new java.lang.NullPointerException(); }
-      java.util.Map<java.lang.String, org.tensorflow.proto.framework.TensorInfo> map =
-          internalGetInputs().getMap();
-      if (!map.containsKey(key)) {
-        throw new java.lang.IllegalArgumentException();
-      }
-      return map.get(key);
-    }
-
-    public Builder clearInputs() {
-      internalGetMutableInputs().getMutableMap()
-          .clear();
-      return this;
-    }
-    /**
-     * <pre>
-     * Named input parameters.
-     * </pre>
-     *
-     * <code>map&lt;string, .tensorflow.TensorInfo&gt; inputs = 1;</code>
-     */
-
-    public Builder removeInputs(
-        java.lang.String key) {
-      if (key == null) { throw new java.lang.NullPointerException(); }
-      internalGetMutableInputs().getMutableMap()
-          .remove(key);
-      return this;
-    }
-    /**
-     * Use alternate mutation accessors instead.
-     */
-    @java.lang.Deprecated
-    public java.util.Map<java.lang.String, org.tensorflow.proto.framework.TensorInfo>
-    getMutableInputs() {
-      return internalGetMutableInputs().getMutableMap();
-    }
-    /**
-     * <pre>
-     * Named input parameters.
-     * </pre>
-     *
-     * <code>map&lt;string, .tensorflow.TensorInfo&gt; inputs = 1;</code>
-     */
-    public Builder putInputs(
-        java.lang.String key,
-        org.tensorflow.proto.framework.TensorInfo value) {
-      if (key == null) { throw new java.lang.NullPointerException(); }
-      if (value == null) { throw new java.lang.NullPointerException(); }
-      internalGetMutableInputs().getMutableMap()
-          .put(key, value);
-      return this;
-    }
-    /**
-     * <pre>
-     * Named input parameters.
-     * </pre>
-     *
-     * <code>map&lt;string, .tensorflow.TensorInfo&gt; inputs = 1;</code>
-     */
-
-    public Builder putAllInputs(
-        java.util.Map<java.lang.String, org.tensorflow.proto.framework.TensorInfo> values) {
-      internalGetMutableInputs().getMutableMap()
-          .putAll(values);
-      return this;
-    }
-
-    private com.google.protobuf.MapField<
-        java.lang.String, org.tensorflow.proto.framework.TensorInfo> outputs_;
-    private com.google.protobuf.MapField<java.lang.String, org.tensorflow.proto.framework.TensorInfo>
-    internalGetOutputs() {
-      if (outputs_ == null) {
-        return com.google.protobuf.MapField.emptyMapField(
-            OutputsDefaultEntryHolder.defaultEntry);
-      }
-      return outputs_;
-    }
-    private com.google.protobuf.MapField<java.lang.String, org.tensorflow.proto.framework.TensorInfo>
-    internalGetMutableOutputs() {
-      onChanged();;
-      if (outputs_ == null) {
-        outputs_ = com.google.protobuf.MapField.newMapField(
-            OutputsDefaultEntryHolder.defaultEntry);
-      }
-      if (!outputs_.isMutable()) {
-        outputs_ = outputs_.copy();
-      }
-      return outputs_;
-    }
-
-    public int getOutputsCount() {
-      return internalGetOutputs().getMap().size();
-    }
-    /**
-     * <pre>
-     * Named output parameters.
-     * </pre>
-     *
-     * <code>map&lt;string, .tensorflow.TensorInfo&gt; outputs = 2;</code>
-     */
-
-    public boolean containsOutputs(
-        java.lang.String key) {
-      if (key == null) { throw new java.lang.NullPointerException(); }
-      return internalGetOutputs().getMap().containsKey(key);
-    }
-    /**
-     * Use {@link #getOutputsMap()} instead.
-     */
-    @java.lang.Deprecated
-    public java.util.Map<java.lang.String, org.tensorflow.proto.framework.TensorInfo> getOutputs() {
-      return getOutputsMap();
-    }
-    /**
-     * <pre>
-     * Named output parameters.
-     * </pre>
-     *
-     * <code>map&lt;string, .tensorflow.TensorInfo&gt; outputs = 2;</code>
-     */
-
-    public java.util.Map<java.lang.String, org.tensorflow.proto.framework.TensorInfo> getOutputsMap() {
-      return internalGetOutputs().getMap();
-    }
-    /**
-     * <pre>
-     * Named output parameters.
-     * </pre>
-     *
-     * <code>map&lt;string, .tensorflow.TensorInfo&gt; outputs = 2;</code>
-     */
-
-    public org.tensorflow.proto.framework.TensorInfo getOutputsOrDefault(
-        java.lang.String key,
-        org.tensorflow.proto.framework.TensorInfo defaultValue) {
-      if (key == null) { throw new java.lang.NullPointerException(); }
-      java.util.Map<java.lang.String, org.tensorflow.proto.framework.TensorInfo> map =
-          internalGetOutputs().getMap();
-      return map.containsKey(key) ? map.get(key) : defaultValue;
-    }
-    /**
-     * <pre>
-     * Named output parameters.
-     * </pre>
-     *
-     * <code>map&lt;string, .tensorflow.TensorInfo&gt; outputs = 2;</code>
-     */
-
-    public org.tensorflow.proto.framework.TensorInfo getOutputsOrThrow(
-        java.lang.String key) {
-      if (key == null) { throw new java.lang.NullPointerException(); }
-      java.util.Map<java.lang.String, org.tensorflow.proto.framework.TensorInfo> map =
-          internalGetOutputs().getMap();
-      if (!map.containsKey(key)) {
-        throw new java.lang.IllegalArgumentException();
-      }
-      return map.get(key);
-    }
-
-    public Builder clearOutputs() {
-      internalGetMutableOutputs().getMutableMap()
-          .clear();
-      return this;
-    }
-    /**
-     * <pre>
-     * Named output parameters.
-     * </pre>
-     *
-     * <code>map&lt;string, .tensorflow.TensorInfo&gt; outputs = 2;</code>
-     */
-
-    public Builder removeOutputs(
-        java.lang.String key) {
-      if (key == null) { throw new java.lang.NullPointerException(); }
-      internalGetMutableOutputs().getMutableMap()
-          .remove(key);
-      return this;
-    }
-    /**
-     * Use alternate mutation accessors instead.
-     */
-    @java.lang.Deprecated
-    public java.util.Map<java.lang.String, org.tensorflow.proto.framework.TensorInfo>
-    getMutableOutputs() {
-      return internalGetMutableOutputs().getMutableMap();
-    }
-    /**
-     * <pre>
-     * Named output parameters.
-     * </pre>
-     *
-     * <code>map&lt;string, .tensorflow.TensorInfo&gt; outputs = 2;</code>
-     */
-    public Builder putOutputs(
-        java.lang.String key,
-        org.tensorflow.proto.framework.TensorInfo value) {
-      if (key == null) { throw new java.lang.NullPointerException(); }
-      if (value == null) { throw new java.lang.NullPointerException(); }
-      internalGetMutableOutputs().getMutableMap()
-          .put(key, value);
-      return this;
-    }
-    /**
-     * <pre>
-     * Named output parameters.
-     * </pre>
-     *
-     * <code>map&lt;string, .tensorflow.TensorInfo&gt; outputs = 2;</code>
-     */
-
-    public Builder putAllOutputs(
-        java.util.Map<java.lang.String, org.tensorflow.proto.framework.TensorInfo> values) {
-      internalGetMutableOutputs().getMutableMap()
-          .putAll(values);
-      return this;
-    }
-
-    private java.lang.Object methodName_ = "";
-    /**
-     * <pre>
-     * Extensible method_name information enabling third-party users to mark a
-     * SignatureDef as supporting a particular method. This enables producers and
-     * consumers of SignatureDefs, e.g. a model definition library and a serving
-     * library to have a clear hand-off regarding the semantics of a computation.
-     * Note that multiple SignatureDefs in a single MetaGraphDef may have the same
-     * method_name. This is commonly used to support multi-headed computation,
-     * where a single graph computation may return multiple results.
-     * </pre>
-     *
-     * <code>string method_name = 3;</code>
-     */
-    public java.lang.String getMethodName() {
-      java.lang.Object ref = methodName_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        methodName_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Extensible method_name information enabling third-party users to mark a
-     * SignatureDef as supporting a particular method. This enables producers and
-     * consumers of SignatureDefs, e.g. a model definition library and a serving
-     * library to have a clear hand-off regarding the semantics of a computation.
-     * Note that multiple SignatureDefs in a single MetaGraphDef may have the same
-     * method_name. This is commonly used to support multi-headed computation,
-     * where a single graph computation may return multiple results.
-     * </pre>
-     *
-     * <code>string method_name = 3;</code>
-     */
-    public com.google.protobuf.ByteString
-        getMethodNameBytes() {
-      java.lang.Object ref = methodName_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        methodName_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Extensible method_name information enabling third-party users to mark a
-     * SignatureDef as supporting a particular method. This enables producers and
-     * consumers of SignatureDefs, e.g. a model definition library and a serving
-     * library to have a clear hand-off regarding the semantics of a computation.
-     * Note that multiple SignatureDefs in a single MetaGraphDef may have the same
-     * method_name. This is commonly used to support multi-headed computation,
-     * where a single graph computation may return multiple results.
-     * </pre>
-     *
-     * <code>string method_name = 3;</code>
-     */
-    public Builder setMethodName(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      methodName_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Extensible method_name information enabling third-party users to mark a
-     * SignatureDef as supporting a particular method. This enables producers and
-     * consumers of SignatureDefs, e.g. a model definition library and a serving
-     * library to have a clear hand-off regarding the semantics of a computation.
-     * Note that multiple SignatureDefs in a single MetaGraphDef may have the same
-     * method_name. This is commonly used to support multi-headed computation,
-     * where a single graph computation may return multiple results.
-     * </pre>
-     *
-     * <code>string method_name = 3;</code>
-     */
-    public Builder clearMethodName() {
-      
-      methodName_ = getDefaultInstance().getMethodName();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Extensible method_name information enabling third-party users to mark a
-     * SignatureDef as supporting a particular method. This enables producers and
-     * consumers of SignatureDefs, e.g. a model definition library and a serving
-     * library to have a clear hand-off regarding the semantics of a computation.
-     * Note that multiple SignatureDefs in a single MetaGraphDef may have the same
-     * method_name. This is commonly used to support multi-headed computation,
-     * where a single graph computation may return multiple results.
-     * </pre>
-     *
-     * <code>string method_name = 3;</code>
-     */
-    public Builder setMethodNameBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      methodName_ = value;
-      onChanged();
-      return this;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.SignatureDef)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.SignatureDef)
-  private static final org.tensorflow.proto.framework.SignatureDef DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.SignatureDef();
-  }
-
-  public static org.tensorflow.proto.framework.SignatureDef getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<SignatureDef>
-      PARSER = new com.google.protobuf.AbstractParser<SignatureDef>() {
-    @java.lang.Override
-    public SignatureDef parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new SignatureDef(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<SignatureDef> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<SignatureDef> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.framework.SignatureDef getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SignatureDefOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SignatureDefOrBuilder.java
deleted file mode 100644
index e9234adf5f7..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SignatureDefOrBuilder.java
+++ /dev/null
@@ -1,147 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/meta_graph.proto
-
-package org.tensorflow.proto.framework;
-
-public interface SignatureDefOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.SignatureDef)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <pre>
-   * Named input parameters.
-   * </pre>
-   *
-   * <code>map&lt;string, .tensorflow.TensorInfo&gt; inputs = 1;</code>
-   */
-  int getInputsCount();
-  /**
-   * <pre>
-   * Named input parameters.
-   * </pre>
-   *
-   * <code>map&lt;string, .tensorflow.TensorInfo&gt; inputs = 1;</code>
-   */
-  boolean containsInputs(
-      java.lang.String key);
-  /**
-   * Use {@link #getInputsMap()} instead.
-   */
-  @java.lang.Deprecated
-  java.util.Map<java.lang.String, org.tensorflow.proto.framework.TensorInfo>
-  getInputs();
-  /**
-   * <pre>
-   * Named input parameters.
-   * </pre>
-   *
-   * <code>map&lt;string, .tensorflow.TensorInfo&gt; inputs = 1;</code>
-   */
-  java.util.Map<java.lang.String, org.tensorflow.proto.framework.TensorInfo>
-  getInputsMap();
-  /**
-   * <pre>
-   * Named input parameters.
-   * </pre>
-   *
-   * <code>map&lt;string, .tensorflow.TensorInfo&gt; inputs = 1;</code>
-   */
-
-  org.tensorflow.proto.framework.TensorInfo getInputsOrDefault(
-      java.lang.String key,
-      org.tensorflow.proto.framework.TensorInfo defaultValue);
-  /**
-   * <pre>
-   * Named input parameters.
-   * </pre>
-   *
-   * <code>map&lt;string, .tensorflow.TensorInfo&gt; inputs = 1;</code>
-   */
-
-  org.tensorflow.proto.framework.TensorInfo getInputsOrThrow(
-      java.lang.String key);
-
-  /**
-   * <pre>
-   * Named output parameters.
-   * </pre>
-   *
-   * <code>map&lt;string, .tensorflow.TensorInfo&gt; outputs = 2;</code>
-   */
-  int getOutputsCount();
-  /**
-   * <pre>
-   * Named output parameters.
-   * </pre>
-   *
-   * <code>map&lt;string, .tensorflow.TensorInfo&gt; outputs = 2;</code>
-   */
-  boolean containsOutputs(
-      java.lang.String key);
-  /**
-   * Use {@link #getOutputsMap()} instead.
-   */
-  @java.lang.Deprecated
-  java.util.Map<java.lang.String, org.tensorflow.proto.framework.TensorInfo>
-  getOutputs();
-  /**
-   * <pre>
-   * Named output parameters.
-   * </pre>
-   *
-   * <code>map&lt;string, .tensorflow.TensorInfo&gt; outputs = 2;</code>
-   */
-  java.util.Map<java.lang.String, org.tensorflow.proto.framework.TensorInfo>
-  getOutputsMap();
-  /**
-   * <pre>
-   * Named output parameters.
-   * </pre>
-   *
-   * <code>map&lt;string, .tensorflow.TensorInfo&gt; outputs = 2;</code>
-   */
-
-  org.tensorflow.proto.framework.TensorInfo getOutputsOrDefault(
-      java.lang.String key,
-      org.tensorflow.proto.framework.TensorInfo defaultValue);
-  /**
-   * <pre>
-   * Named output parameters.
-   * </pre>
-   *
-   * <code>map&lt;string, .tensorflow.TensorInfo&gt; outputs = 2;</code>
-   */
-
-  org.tensorflow.proto.framework.TensorInfo getOutputsOrThrow(
-      java.lang.String key);
-
-  /**
-   * <pre>
-   * Extensible method_name information enabling third-party users to mark a
-   * SignatureDef as supporting a particular method. This enables producers and
-   * consumers of SignatureDefs, e.g. a model definition library and a serving
-   * library to have a clear hand-off regarding the semantics of a computation.
-   * Note that multiple SignatureDefs in a single MetaGraphDef may have the same
-   * method_name. This is commonly used to support multi-headed computation,
-   * where a single graph computation may return multiple results.
-   * </pre>
-   *
-   * <code>string method_name = 3;</code>
-   */
-  java.lang.String getMethodName();
-  /**
-   * <pre>
-   * Extensible method_name information enabling third-party users to mark a
-   * SignatureDef as supporting a particular method. This enables producers and
-   * consumers of SignatureDefs, e.g. a model definition library and a serving
-   * library to have a clear hand-off regarding the semantics of a computation.
-   * Note that multiple SignatureDefs in a single MetaGraphDef may have the same
-   * method_name. This is commonly used to support multi-headed computation,
-   * where a single graph computation may return multiple results.
-   * </pre>
-   *
-   * <code>string method_name = 3;</code>
-   */
-  com.google.protobuf.ByteString
-      getMethodNameBytes();
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/StepStats.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/StepStats.java
deleted file mode 100644
index 01a979698bd..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/StepStats.java
+++ /dev/null
@@ -1,765 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/framework/step_stats.proto
-
-package org.tensorflow.proto.framework;
-
-/**
- * Protobuf type {@code tensorflow.StepStats}
- */
-public  final class StepStats extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.StepStats)
-    StepStatsOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use StepStats.newBuilder() to construct.
-  private StepStats(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private StepStats() {
-    devStats_ = java.util.Collections.emptyList();
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new StepStats();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private StepStats(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    int mutable_bitField0_ = 0;
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 10: {
-            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
-              devStats_ = new java.util.ArrayList<org.tensorflow.proto.framework.DeviceStepStats>();
-              mutable_bitField0_ |= 0x00000001;
-            }
-            devStats_.add(
-                input.readMessage(org.tensorflow.proto.framework.DeviceStepStats.parser(), extensionRegistry));
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      if (((mutable_bitField0_ & 0x00000001) != 0)) {
-        devStats_ = java.util.Collections.unmodifiableList(devStats_);
-      }
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.framework.StepStatsProtos.internal_static_tensorflow_StepStats_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.framework.StepStatsProtos.internal_static_tensorflow_StepStats_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.framework.StepStats.class, org.tensorflow.proto.framework.StepStats.Builder.class);
-  }
-
-  public static final int DEV_STATS_FIELD_NUMBER = 1;
-  private java.util.List<org.tensorflow.proto.framework.DeviceStepStats> devStats_;
-  /**
-   * <code>repeated .tensorflow.DeviceStepStats dev_stats = 1;</code>
-   */
-  public java.util.List<org.tensorflow.proto.framework.DeviceStepStats> getDevStatsList() {
-    return devStats_;
-  }
-  /**
-   * <code>repeated .tensorflow.DeviceStepStats dev_stats = 1;</code>
-   */
-  public java.util.List<? extends org.tensorflow.proto.framework.DeviceStepStatsOrBuilder> 
-      getDevStatsOrBuilderList() {
-    return devStats_;
-  }
-  /**
-   * <code>repeated .tensorflow.DeviceStepStats dev_stats = 1;</code>
-   */
-  public int getDevStatsCount() {
-    return devStats_.size();
-  }
-  /**
-   * <code>repeated .tensorflow.DeviceStepStats dev_stats = 1;</code>
-   */
-  public org.tensorflow.proto.framework.DeviceStepStats getDevStats(int index) {
-    return devStats_.get(index);
-  }
-  /**
-   * <code>repeated .tensorflow.DeviceStepStats dev_stats = 1;</code>
-   */
-  public org.tensorflow.proto.framework.DeviceStepStatsOrBuilder getDevStatsOrBuilder(
-      int index) {
-    return devStats_.get(index);
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    for (int i = 0; i < devStats_.size(); i++) {
-      output.writeMessage(1, devStats_.get(i));
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    for (int i = 0; i < devStats_.size(); i++) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(1, devStats_.get(i));
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.framework.StepStats)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.framework.StepStats other = (org.tensorflow.proto.framework.StepStats) obj;
-
-    if (!getDevStatsList()
-        .equals(other.getDevStatsList())) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    if (getDevStatsCount() > 0) {
-      hash = (37 * hash) + DEV_STATS_FIELD_NUMBER;
-      hash = (53 * hash) + getDevStatsList().hashCode();
-    }
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.framework.StepStats parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.StepStats parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.StepStats parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.StepStats parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.StepStats parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.StepStats parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.StepStats parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.StepStats parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.StepStats parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.StepStats parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.StepStats parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.StepStats parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.framework.StepStats prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * Protobuf type {@code tensorflow.StepStats}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.StepStats)
-      org.tensorflow.proto.framework.StepStatsOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.StepStatsProtos.internal_static_tensorflow_StepStats_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.StepStatsProtos.internal_static_tensorflow_StepStats_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.StepStats.class, org.tensorflow.proto.framework.StepStats.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.framework.StepStats.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-        getDevStatsFieldBuilder();
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      if (devStatsBuilder_ == null) {
-        devStats_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000001);
-      } else {
-        devStatsBuilder_.clear();
-      }
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.framework.StepStatsProtos.internal_static_tensorflow_StepStats_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.StepStats getDefaultInstanceForType() {
-      return org.tensorflow.proto.framework.StepStats.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.StepStats build() {
-      org.tensorflow.proto.framework.StepStats result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.StepStats buildPartial() {
-      org.tensorflow.proto.framework.StepStats result = new org.tensorflow.proto.framework.StepStats(this);
-      int from_bitField0_ = bitField0_;
-      if (devStatsBuilder_ == null) {
-        if (((bitField0_ & 0x00000001) != 0)) {
-          devStats_ = java.util.Collections.unmodifiableList(devStats_);
-          bitField0_ = (bitField0_ & ~0x00000001);
-        }
-        result.devStats_ = devStats_;
-      } else {
-        result.devStats_ = devStatsBuilder_.build();
-      }
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.framework.StepStats) {
-        return mergeFrom((org.tensorflow.proto.framework.StepStats)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.framework.StepStats other) {
-      if (other == org.tensorflow.proto.framework.StepStats.getDefaultInstance()) return this;
-      if (devStatsBuilder_ == null) {
-        if (!other.devStats_.isEmpty()) {
-          if (devStats_.isEmpty()) {
-            devStats_ = other.devStats_;
-            bitField0_ = (bitField0_ & ~0x00000001);
-          } else {
-            ensureDevStatsIsMutable();
-            devStats_.addAll(other.devStats_);
-          }
-          onChanged();
-        }
-      } else {
-        if (!other.devStats_.isEmpty()) {
-          if (devStatsBuilder_.isEmpty()) {
-            devStatsBuilder_.dispose();
-            devStatsBuilder_ = null;
-            devStats_ = other.devStats_;
-            bitField0_ = (bitField0_ & ~0x00000001);
-            devStatsBuilder_ = 
-              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
-                 getDevStatsFieldBuilder() : null;
-          } else {
-            devStatsBuilder_.addAllMessages(other.devStats_);
-          }
-        }
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.framework.StepStats parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.framework.StepStats) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-    private int bitField0_;
-
-    private java.util.List<org.tensorflow.proto.framework.DeviceStepStats> devStats_ =
-      java.util.Collections.emptyList();
-    private void ensureDevStatsIsMutable() {
-      if (!((bitField0_ & 0x00000001) != 0)) {
-        devStats_ = new java.util.ArrayList<org.tensorflow.proto.framework.DeviceStepStats>(devStats_);
-        bitField0_ |= 0x00000001;
-       }
-    }
-
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.framework.DeviceStepStats, org.tensorflow.proto.framework.DeviceStepStats.Builder, org.tensorflow.proto.framework.DeviceStepStatsOrBuilder> devStatsBuilder_;
-
-    /**
-     * <code>repeated .tensorflow.DeviceStepStats dev_stats = 1;</code>
-     */
-    public java.util.List<org.tensorflow.proto.framework.DeviceStepStats> getDevStatsList() {
-      if (devStatsBuilder_ == null) {
-        return java.util.Collections.unmodifiableList(devStats_);
-      } else {
-        return devStatsBuilder_.getMessageList();
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.DeviceStepStats dev_stats = 1;</code>
-     */
-    public int getDevStatsCount() {
-      if (devStatsBuilder_ == null) {
-        return devStats_.size();
-      } else {
-        return devStatsBuilder_.getCount();
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.DeviceStepStats dev_stats = 1;</code>
-     */
-    public org.tensorflow.proto.framework.DeviceStepStats getDevStats(int index) {
-      if (devStatsBuilder_ == null) {
-        return devStats_.get(index);
-      } else {
-        return devStatsBuilder_.getMessage(index);
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.DeviceStepStats dev_stats = 1;</code>
-     */
-    public Builder setDevStats(
-        int index, org.tensorflow.proto.framework.DeviceStepStats value) {
-      if (devStatsBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureDevStatsIsMutable();
-        devStats_.set(index, value);
-        onChanged();
-      } else {
-        devStatsBuilder_.setMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.DeviceStepStats dev_stats = 1;</code>
-     */
-    public Builder setDevStats(
-        int index, org.tensorflow.proto.framework.DeviceStepStats.Builder builderForValue) {
-      if (devStatsBuilder_ == null) {
-        ensureDevStatsIsMutable();
-        devStats_.set(index, builderForValue.build());
-        onChanged();
-      } else {
-        devStatsBuilder_.setMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.DeviceStepStats dev_stats = 1;</code>
-     */
-    public Builder addDevStats(org.tensorflow.proto.framework.DeviceStepStats value) {
-      if (devStatsBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureDevStatsIsMutable();
-        devStats_.add(value);
-        onChanged();
-      } else {
-        devStatsBuilder_.addMessage(value);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.DeviceStepStats dev_stats = 1;</code>
-     */
-    public Builder addDevStats(
-        int index, org.tensorflow.proto.framework.DeviceStepStats value) {
-      if (devStatsBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureDevStatsIsMutable();
-        devStats_.add(index, value);
-        onChanged();
-      } else {
-        devStatsBuilder_.addMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.DeviceStepStats dev_stats = 1;</code>
-     */
-    public Builder addDevStats(
-        org.tensorflow.proto.framework.DeviceStepStats.Builder builderForValue) {
-      if (devStatsBuilder_ == null) {
-        ensureDevStatsIsMutable();
-        devStats_.add(builderForValue.build());
-        onChanged();
-      } else {
-        devStatsBuilder_.addMessage(builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.DeviceStepStats dev_stats = 1;</code>
-     */
-    public Builder addDevStats(
-        int index, org.tensorflow.proto.framework.DeviceStepStats.Builder builderForValue) {
-      if (devStatsBuilder_ == null) {
-        ensureDevStatsIsMutable();
-        devStats_.add(index, builderForValue.build());
-        onChanged();
-      } else {
-        devStatsBuilder_.addMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.DeviceStepStats dev_stats = 1;</code>
-     */
-    public Builder addAllDevStats(
-        java.lang.Iterable<? extends org.tensorflow.proto.framework.DeviceStepStats> values) {
-      if (devStatsBuilder_ == null) {
-        ensureDevStatsIsMutable();
-        com.google.protobuf.AbstractMessageLite.Builder.addAll(
-            values, devStats_);
-        onChanged();
-      } else {
-        devStatsBuilder_.addAllMessages(values);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.DeviceStepStats dev_stats = 1;</code>
-     */
-    public Builder clearDevStats() {
-      if (devStatsBuilder_ == null) {
-        devStats_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000001);
-        onChanged();
-      } else {
-        devStatsBuilder_.clear();
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.DeviceStepStats dev_stats = 1;</code>
-     */
-    public Builder removeDevStats(int index) {
-      if (devStatsBuilder_ == null) {
-        ensureDevStatsIsMutable();
-        devStats_.remove(index);
-        onChanged();
-      } else {
-        devStatsBuilder_.remove(index);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.DeviceStepStats dev_stats = 1;</code>
-     */
-    public org.tensorflow.proto.framework.DeviceStepStats.Builder getDevStatsBuilder(
-        int index) {
-      return getDevStatsFieldBuilder().getBuilder(index);
-    }
-    /**
-     * <code>repeated .tensorflow.DeviceStepStats dev_stats = 1;</code>
-     */
-    public org.tensorflow.proto.framework.DeviceStepStatsOrBuilder getDevStatsOrBuilder(
-        int index) {
-      if (devStatsBuilder_ == null) {
-        return devStats_.get(index);  } else {
-        return devStatsBuilder_.getMessageOrBuilder(index);
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.DeviceStepStats dev_stats = 1;</code>
-     */
-    public java.util.List<? extends org.tensorflow.proto.framework.DeviceStepStatsOrBuilder> 
-         getDevStatsOrBuilderList() {
-      if (devStatsBuilder_ != null) {
-        return devStatsBuilder_.getMessageOrBuilderList();
-      } else {
-        return java.util.Collections.unmodifiableList(devStats_);
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.DeviceStepStats dev_stats = 1;</code>
-     */
-    public org.tensorflow.proto.framework.DeviceStepStats.Builder addDevStatsBuilder() {
-      return getDevStatsFieldBuilder().addBuilder(
-          org.tensorflow.proto.framework.DeviceStepStats.getDefaultInstance());
-    }
-    /**
-     * <code>repeated .tensorflow.DeviceStepStats dev_stats = 1;</code>
-     */
-    public org.tensorflow.proto.framework.DeviceStepStats.Builder addDevStatsBuilder(
-        int index) {
-      return getDevStatsFieldBuilder().addBuilder(
-          index, org.tensorflow.proto.framework.DeviceStepStats.getDefaultInstance());
-    }
-    /**
-     * <code>repeated .tensorflow.DeviceStepStats dev_stats = 1;</code>
-     */
-    public java.util.List<org.tensorflow.proto.framework.DeviceStepStats.Builder> 
-         getDevStatsBuilderList() {
-      return getDevStatsFieldBuilder().getBuilderList();
-    }
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.framework.DeviceStepStats, org.tensorflow.proto.framework.DeviceStepStats.Builder, org.tensorflow.proto.framework.DeviceStepStatsOrBuilder> 
-        getDevStatsFieldBuilder() {
-      if (devStatsBuilder_ == null) {
-        devStatsBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
-            org.tensorflow.proto.framework.DeviceStepStats, org.tensorflow.proto.framework.DeviceStepStats.Builder, org.tensorflow.proto.framework.DeviceStepStatsOrBuilder>(
-                devStats_,
-                ((bitField0_ & 0x00000001) != 0),
-                getParentForChildren(),
-                isClean());
-        devStats_ = null;
-      }
-      return devStatsBuilder_;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.StepStats)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.StepStats)
-  private static final org.tensorflow.proto.framework.StepStats DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.StepStats();
-  }
-
-  public static org.tensorflow.proto.framework.StepStats getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<StepStats>
-      PARSER = new com.google.protobuf.AbstractParser<StepStats>() {
-    @java.lang.Override
-    public StepStats parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new StepStats(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<StepStats> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<StepStats> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.framework.StepStats getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/StepStatsOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/StepStatsOrBuilder.java
deleted file mode 100644
index bb7cdf41046..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/StepStatsOrBuilder.java
+++ /dev/null
@@ -1,33 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/framework/step_stats.proto
-
-package org.tensorflow.proto.framework;
-
-public interface StepStatsOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.StepStats)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <code>repeated .tensorflow.DeviceStepStats dev_stats = 1;</code>
-   */
-  java.util.List<org.tensorflow.proto.framework.DeviceStepStats> 
-      getDevStatsList();
-  /**
-   * <code>repeated .tensorflow.DeviceStepStats dev_stats = 1;</code>
-   */
-  org.tensorflow.proto.framework.DeviceStepStats getDevStats(int index);
-  /**
-   * <code>repeated .tensorflow.DeviceStepStats dev_stats = 1;</code>
-   */
-  int getDevStatsCount();
-  /**
-   * <code>repeated .tensorflow.DeviceStepStats dev_stats = 1;</code>
-   */
-  java.util.List<? extends org.tensorflow.proto.framework.DeviceStepStatsOrBuilder> 
-      getDevStatsOrBuilderList();
-  /**
-   * <code>repeated .tensorflow.DeviceStepStats dev_stats = 1;</code>
-   */
-  org.tensorflow.proto.framework.DeviceStepStatsOrBuilder getDevStatsOrBuilder(
-      int index);
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/StepStatsProtos.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/StepStatsProtos.java
deleted file mode 100644
index 15ef4a6b554..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/StepStatsProtos.java
+++ /dev/null
@@ -1,169 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/framework/step_stats.proto
-
-package org.tensorflow.proto.framework;
-
-public final class StepStatsProtos {
-  private StepStatsProtos() {}
-  public static void registerAllExtensions(
-      com.google.protobuf.ExtensionRegistryLite registry) {
-  }
-
-  public static void registerAllExtensions(
-      com.google.protobuf.ExtensionRegistry registry) {
-    registerAllExtensions(
-        (com.google.protobuf.ExtensionRegistryLite) registry);
-  }
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_AllocationRecord_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_AllocationRecord_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_AllocatorMemoryUsed_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_AllocatorMemoryUsed_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_NodeOutput_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_NodeOutput_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_MemoryStats_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_MemoryStats_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_NodeExecStats_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_NodeExecStats_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_DeviceStepStats_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_DeviceStepStats_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_DeviceStepStats_ThreadNamesEntry_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_DeviceStepStats_ThreadNamesEntry_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_StepStats_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_StepStats_fieldAccessorTable;
-
-  public static com.google.protobuf.Descriptors.FileDescriptor
-      getDescriptor() {
-    return descriptor;
-  }
-  private static  com.google.protobuf.Descriptors.FileDescriptor
-      descriptor;
-  static {
-    java.lang.String[] descriptorData = {
-      "\n*tensorflow/core/framework/step_stats.p" +
-      "roto\022\ntensorflow\0326tensorflow/core/framew" +
-      "ork/allocation_description.proto\0322tensor" +
-      "flow/core/framework/tensor_description.p" +
-      "roto\"=\n\020AllocationRecord\022\024\n\014alloc_micros" +
-      "\030\001 \001(\003\022\023\n\013alloc_bytes\030\002 \001(\003\"\304\001\n\023Allocato" +
-      "rMemoryUsed\022\026\n\016allocator_name\030\001 \001(\t\022\023\n\013t" +
-      "otal_bytes\030\002 \001(\003\022\022\n\npeak_bytes\030\003 \001(\003\022\022\n\n" +
-      "live_bytes\030\004 \001(\003\0228\n\022allocation_records\030\006" +
-      " \003(\0132\034.tensorflow.AllocationRecord\022\036\n\026al" +
-      "locator_bytes_in_use\030\005 \001(\003\"U\n\nNodeOutput" +
-      "\022\014\n\004slot\030\001 \001(\005\0229\n\022tensor_description\030\003 \001" +
-      "(\0132\035.tensorflow.TensorDescription\"\354\001\n\013Me" +
-      "moryStats\022\030\n\020temp_memory_size\030\001 \001(\003\022\036\n\026p" +
-      "ersistent_memory_size\030\003 \001(\003\022#\n\033persisten" +
-      "t_tensor_alloc_ids\030\005 \003(\003\022#\n\027device_temp_" +
-      "memory_size\030\002 \001(\003B\002\030\001\022)\n\035device_persiste" +
-      "nt_memory_size\030\004 \001(\003B\002\030\001\022.\n\"device_persi" +
-      "stent_tensor_alloc_ids\030\006 \003(\003B\002\030\001\"\236\004\n\rNod" +
-      "eExecStats\022\021\n\tnode_name\030\001 \001(\t\022\030\n\020all_sta" +
-      "rt_micros\030\002 \001(\003\022\033\n\023op_start_rel_micros\030\003" +
-      " \001(\003\022\031\n\021op_end_rel_micros\030\004 \001(\003\022\032\n\022all_e" +
-      "nd_rel_micros\030\005 \001(\003\022/\n\006memory\030\006 \003(\0132\037.te" +
-      "nsorflow.AllocatorMemoryUsed\022&\n\006output\030\007" +
-      " \003(\0132\026.tensorflow.NodeOutput\022\026\n\016timeline" +
-      "_label\030\010 \001(\t\022\030\n\020scheduled_micros\030\t \001(\003\022\021" +
-      "\n\tthread_id\030\n \001(\r\022<\n\021referenced_tensor\030\013" +
-      " \003(\0132!.tensorflow.AllocationDescription\022" +
-      "-\n\014memory_stats\030\014 \001(\0132\027.tensorflow.Memor" +
-      "yStats\022\027\n\017all_start_nanos\030\r \001(\003\022\032\n\022op_st" +
-      "art_rel_nanos\030\016 \001(\003\022\030\n\020op_end_rel_nanos\030" +
-      "\017 \001(\003\022\031\n\021all_end_rel_nanos\030\020 \001(\003\022\027\n\017sche" +
-      "duled_nanos\030\021 \001(\003\"\310\001\n\017DeviceStepStats\022\016\n" +
-      "\006device\030\001 \001(\t\022-\n\nnode_stats\030\002 \003(\0132\031.tens" +
-      "orflow.NodeExecStats\022B\n\014thread_names\030\003 \003" +
-      "(\0132,.tensorflow.DeviceStepStats.ThreadNa" +
-      "mesEntry\0322\n\020ThreadNamesEntry\022\013\n\003key\030\001 \001(" +
-      "\r\022\r\n\005value\030\002 \001(\t:\0028\001\";\n\tStepStats\022.\n\tdev" +
-      "_stats\030\001 \003(\0132\033.tensorflow.DeviceStepStat" +
-      "sB\211\001\n\036org.tensorflow.proto.frameworkB\017St" +
-      "epStatsProtosP\001ZQgithub.com/tensorflow/t" +
-      "ensorflow/tensorflow/go/core/framework/s" +
-      "tep_stats_go_proto\370\001\001b\006proto3"
-    };
-    descriptor = com.google.protobuf.Descriptors.FileDescriptor
-      .internalBuildGeneratedFileFrom(descriptorData,
-        new com.google.protobuf.Descriptors.FileDescriptor[] {
-          org.tensorflow.proto.framework.AllocationDescriptionProtos.getDescriptor(),
-          org.tensorflow.proto.framework.TensorDescriptionProtos.getDescriptor(),
-        });
-    internal_static_tensorflow_AllocationRecord_descriptor =
-      getDescriptor().getMessageTypes().get(0);
-    internal_static_tensorflow_AllocationRecord_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_AllocationRecord_descriptor,
-        new java.lang.String[] { "AllocMicros", "AllocBytes", });
-    internal_static_tensorflow_AllocatorMemoryUsed_descriptor =
-      getDescriptor().getMessageTypes().get(1);
-    internal_static_tensorflow_AllocatorMemoryUsed_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_AllocatorMemoryUsed_descriptor,
-        new java.lang.String[] { "AllocatorName", "TotalBytes", "PeakBytes", "LiveBytes", "AllocationRecords", "AllocatorBytesInUse", });
-    internal_static_tensorflow_NodeOutput_descriptor =
-      getDescriptor().getMessageTypes().get(2);
-    internal_static_tensorflow_NodeOutput_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_NodeOutput_descriptor,
-        new java.lang.String[] { "Slot", "TensorDescription", });
-    internal_static_tensorflow_MemoryStats_descriptor =
-      getDescriptor().getMessageTypes().get(3);
-    internal_static_tensorflow_MemoryStats_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_MemoryStats_descriptor,
-        new java.lang.String[] { "TempMemorySize", "PersistentMemorySize", "PersistentTensorAllocIds", "DeviceTempMemorySize", "DevicePersistentMemorySize", "DevicePersistentTensorAllocIds", });
-    internal_static_tensorflow_NodeExecStats_descriptor =
-      getDescriptor().getMessageTypes().get(4);
-    internal_static_tensorflow_NodeExecStats_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_NodeExecStats_descriptor,
-        new java.lang.String[] { "NodeName", "AllStartMicros", "OpStartRelMicros", "OpEndRelMicros", "AllEndRelMicros", "Memory", "Output", "TimelineLabel", "ScheduledMicros", "ThreadId", "ReferencedTensor", "MemoryStats", "AllStartNanos", "OpStartRelNanos", "OpEndRelNanos", "AllEndRelNanos", "ScheduledNanos", });
-    internal_static_tensorflow_DeviceStepStats_descriptor =
-      getDescriptor().getMessageTypes().get(5);
-    internal_static_tensorflow_DeviceStepStats_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_DeviceStepStats_descriptor,
-        new java.lang.String[] { "Device", "NodeStats", "ThreadNames", });
-    internal_static_tensorflow_DeviceStepStats_ThreadNamesEntry_descriptor =
-      internal_static_tensorflow_DeviceStepStats_descriptor.getNestedTypes().get(0);
-    internal_static_tensorflow_DeviceStepStats_ThreadNamesEntry_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_DeviceStepStats_ThreadNamesEntry_descriptor,
-        new java.lang.String[] { "Key", "Value", });
-    internal_static_tensorflow_StepStats_descriptor =
-      getDescriptor().getMessageTypes().get(6);
-    internal_static_tensorflow_StepStats_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_StepStats_descriptor,
-        new java.lang.String[] { "DevStats", });
-    org.tensorflow.proto.framework.AllocationDescriptionProtos.getDescriptor();
-    org.tensorflow.proto.framework.TensorDescriptionProtos.getDescriptor();
-  }
-
-  // @@protoc_insertion_point(outer_class_scope)
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/StructProtos.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/StructProtos.java
deleted file mode 100644
index 6b25fbe0fa5..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/StructProtos.java
+++ /dev/null
@@ -1,215 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/struct.proto
-
-package org.tensorflow.proto.framework;
-
-public final class StructProtos {
-  private StructProtos() {}
-  public static void registerAllExtensions(
-      com.google.protobuf.ExtensionRegistryLite registry) {
-  }
-
-  public static void registerAllExtensions(
-      com.google.protobuf.ExtensionRegistry registry) {
-    registerAllExtensions(
-        (com.google.protobuf.ExtensionRegistryLite) registry);
-  }
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_StructuredValue_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_StructuredValue_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_NoneValue_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_NoneValue_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_ListValue_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_ListValue_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_TupleValue_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_TupleValue_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_DictValue_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_DictValue_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_DictValue_FieldsEntry_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_DictValue_FieldsEntry_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_PairValue_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_PairValue_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_NamedTupleValue_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_NamedTupleValue_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_TensorSpecProto_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_TensorSpecProto_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_BoundedTensorSpecProto_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_BoundedTensorSpecProto_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_TypeSpecProto_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_TypeSpecProto_fieldAccessorTable;
-
-  public static com.google.protobuf.Descriptors.FileDescriptor
-      getDescriptor() {
-    return descriptor;
-  }
-  private static  com.google.protobuf.Descriptors.FileDescriptor
-      descriptor;
-  static {
-    java.lang.String[] descriptorData = {
-      "\n%tensorflow/core/protobuf/struct.proto\022" +
-      "\ntensorflow\032&tensorflow/core/framework/t" +
-      "ensor.proto\032,tensorflow/core/framework/t" +
-      "ensor_shape.proto\032%tensorflow/core/frame" +
-      "work/types.proto\"\220\005\n\017StructuredValue\022+\n\n" +
-      "none_value\030\001 \001(\0132\025.tensorflow.NoneValueH" +
-      "\000\022\027\n\rfloat64_value\030\013 \001(\001H\000\022\025\n\013int64_valu" +
-      "e\030\014 \001(\022H\000\022\026\n\014string_value\030\r \001(\tH\000\022\024\n\nboo" +
-      "l_value\030\016 \001(\010H\000\022:\n\022tensor_shape_value\030\037 " +
-      "\001(\0132\034.tensorflow.TensorShapeProtoH\000\0222\n\022t" +
-      "ensor_dtype_value\030  \001(\0162\024.tensorflow.Dat" +
-      "aTypeH\000\0228\n\021tensor_spec_value\030! \001(\0132\033.ten" +
-      "sorflow.TensorSpecProtoH\000\0224\n\017type_spec_v" +
-      "alue\030\" \001(\0132\031.tensorflow.TypeSpecProtoH\000\022" +
-      "G\n\031bounded_tensor_spec_value\030# \001(\0132\".ten" +
-      "sorflow.BoundedTensorSpecProtoH\000\022+\n\nlist" +
-      "_value\0303 \001(\0132\025.tensorflow.ListValueH\000\022-\n" +
-      "\013tuple_value\0304 \001(\0132\026.tensorflow.TupleVal" +
-      "ueH\000\022+\n\ndict_value\0305 \001(\0132\025.tensorflow.Di" +
-      "ctValueH\000\0228\n\021named_tuple_value\0306 \001(\0132\033.t" +
-      "ensorflow.NamedTupleValueH\000B\006\n\004kind\"\013\n\tN" +
-      "oneValue\"8\n\tListValue\022+\n\006values\030\001 \003(\0132\033." +
-      "tensorflow.StructuredValue\"9\n\nTupleValue" +
-      "\022+\n\006values\030\001 \003(\0132\033.tensorflow.Structured" +
-      "Value\"\212\001\n\tDictValue\0221\n\006fields\030\001 \003(\0132!.te" +
-      "nsorflow.DictValue.FieldsEntry\032J\n\013Fields" +
-      "Entry\022\013\n\003key\030\001 \001(\t\022*\n\005value\030\002 \001(\0132\033.tens" +
-      "orflow.StructuredValue:\0028\001\"D\n\tPairValue\022" +
-      "\013\n\003key\030\001 \001(\t\022*\n\005value\030\002 \001(\0132\033.tensorflow" +
-      ".StructuredValue\"F\n\017NamedTupleValue\022\014\n\004n" +
-      "ame\030\001 \001(\t\022%\n\006values\030\002 \003(\0132\025.tensorflow.P" +
-      "airValue\"q\n\017TensorSpecProto\022\014\n\004name\030\001 \001(" +
-      "\t\022+\n\005shape\030\002 \001(\0132\034.tensorflow.TensorShap" +
-      "eProto\022#\n\005dtype\030\003 \001(\0162\024.tensorflow.DataT" +
-      "ype\"\314\001\n\026BoundedTensorSpecProto\022\014\n\004name\030\001" +
-      " \001(\t\022+\n\005shape\030\002 \001(\0132\034.tensorflow.TensorS" +
-      "hapeProto\022#\n\005dtype\030\003 \001(\0162\024.tensorflow.Da" +
-      "taType\022(\n\007minimum\030\004 \001(\0132\027.tensorflow.Ten" +
-      "sorProto\022(\n\007maximum\030\005 \001(\0132\027.tensorflow.T" +
-      "ensorProto\"\242\003\n\rTypeSpecProto\022@\n\017type_spe" +
-      "c_class\030\001 \001(\0162\'.tensorflow.TypeSpecProto" +
-      ".TypeSpecClass\022/\n\ntype_state\030\002 \001(\0132\033.ten" +
-      "sorflow.StructuredValue\022\034\n\024type_spec_cla" +
-      "ss_name\030\003 \001(\t\"\377\001\n\rTypeSpecClass\022\013\n\007UNKNO" +
-      "WN\020\000\022\026\n\022SPARSE_TENSOR_SPEC\020\001\022\027\n\023INDEXED_" +
-      "SLICES_SPEC\020\002\022\026\n\022RAGGED_TENSOR_SPEC\020\003\022\025\n" +
-      "\021TENSOR_ARRAY_SPEC\020\004\022\025\n\021DATA_DATASET_SPE" +
-      "C\020\005\022\026\n\022DATA_ITERATOR_SPEC\020\006\022\021\n\rOPTIONAL_" +
-      "SPEC\020\007\022\024\n\020PER_REPLICA_SPEC\020\010\022\021\n\rVARIABLE" +
-      "_SPEC\020\t\022\026\n\022ROW_PARTITION_SPEC\020\nBz\n\036org.t" +
-      "ensorflow.proto.frameworkB\014StructProtosP" +
-      "\001ZHgithub.com/tensorflow/tensorflow/tens" +
-      "orflow/go/core/core_protos_go_protob\006pro" +
-      "to3"
-    };
-    descriptor = com.google.protobuf.Descriptors.FileDescriptor
-      .internalBuildGeneratedFileFrom(descriptorData,
-        new com.google.protobuf.Descriptors.FileDescriptor[] {
-          org.tensorflow.proto.framework.TensorProtos.getDescriptor(),
-          org.tensorflow.proto.framework.TensorShapeProtos.getDescriptor(),
-          org.tensorflow.proto.framework.TypesProtos.getDescriptor(),
-        });
-    internal_static_tensorflow_StructuredValue_descriptor =
-      getDescriptor().getMessageTypes().get(0);
-    internal_static_tensorflow_StructuredValue_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_StructuredValue_descriptor,
-        new java.lang.String[] { "NoneValue", "Float64Value", "Int64Value", "StringValue", "BoolValue", "TensorShapeValue", "TensorDtypeValue", "TensorSpecValue", "TypeSpecValue", "BoundedTensorSpecValue", "ListValue", "TupleValue", "DictValue", "NamedTupleValue", "Kind", });
-    internal_static_tensorflow_NoneValue_descriptor =
-      getDescriptor().getMessageTypes().get(1);
-    internal_static_tensorflow_NoneValue_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_NoneValue_descriptor,
-        new java.lang.String[] { });
-    internal_static_tensorflow_ListValue_descriptor =
-      getDescriptor().getMessageTypes().get(2);
-    internal_static_tensorflow_ListValue_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_ListValue_descriptor,
-        new java.lang.String[] { "Values", });
-    internal_static_tensorflow_TupleValue_descriptor =
-      getDescriptor().getMessageTypes().get(3);
-    internal_static_tensorflow_TupleValue_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_TupleValue_descriptor,
-        new java.lang.String[] { "Values", });
-    internal_static_tensorflow_DictValue_descriptor =
-      getDescriptor().getMessageTypes().get(4);
-    internal_static_tensorflow_DictValue_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_DictValue_descriptor,
-        new java.lang.String[] { "Fields", });
-    internal_static_tensorflow_DictValue_FieldsEntry_descriptor =
-      internal_static_tensorflow_DictValue_descriptor.getNestedTypes().get(0);
-    internal_static_tensorflow_DictValue_FieldsEntry_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_DictValue_FieldsEntry_descriptor,
-        new java.lang.String[] { "Key", "Value", });
-    internal_static_tensorflow_PairValue_descriptor =
-      getDescriptor().getMessageTypes().get(5);
-    internal_static_tensorflow_PairValue_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_PairValue_descriptor,
-        new java.lang.String[] { "Key", "Value", });
-    internal_static_tensorflow_NamedTupleValue_descriptor =
-      getDescriptor().getMessageTypes().get(6);
-    internal_static_tensorflow_NamedTupleValue_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_NamedTupleValue_descriptor,
-        new java.lang.String[] { "Name", "Values", });
-    internal_static_tensorflow_TensorSpecProto_descriptor =
-      getDescriptor().getMessageTypes().get(7);
-    internal_static_tensorflow_TensorSpecProto_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_TensorSpecProto_descriptor,
-        new java.lang.String[] { "Name", "Shape", "Dtype", });
-    internal_static_tensorflow_BoundedTensorSpecProto_descriptor =
-      getDescriptor().getMessageTypes().get(8);
-    internal_static_tensorflow_BoundedTensorSpecProto_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_BoundedTensorSpecProto_descriptor,
-        new java.lang.String[] { "Name", "Shape", "Dtype", "Minimum", "Maximum", });
-    internal_static_tensorflow_TypeSpecProto_descriptor =
-      getDescriptor().getMessageTypes().get(9);
-    internal_static_tensorflow_TypeSpecProto_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_TypeSpecProto_descriptor,
-        new java.lang.String[] { "TypeSpecClass", "TypeState", "TypeSpecClassName", });
-    org.tensorflow.proto.framework.TensorProtos.getDescriptor();
-    org.tensorflow.proto.framework.TensorShapeProtos.getDescriptor();
-    org.tensorflow.proto.framework.TypesProtos.getDescriptor();
-  }
-
-  // @@protoc_insertion_point(outer_class_scope)
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/StructuredValue.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/StructuredValue.java
deleted file mode 100644
index 327e919c2ab..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/StructuredValue.java
+++ /dev/null
@@ -1,3423 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/struct.proto
-
-package org.tensorflow.proto.framework;
-
-/**
- * <pre>
- * `StructuredValue` represents a dynamically typed value representing various
- * data structures that are inspired by Python data structures typically used in
- * TensorFlow functions as inputs and outputs.
- * For example when saving a Layer there may be a `training` argument. If the
- * user passes a boolean True/False, that switches between two concrete
- * TensorFlow functions. In order to switch between them in the same way after
- * loading the SavedModel, we need to represent "True" and "False".
- * A more advanced example might be a function which takes a list of
- * dictionaries mapping from strings to Tensors. In order to map from
- * user-specified arguments `[{"a": tf.constant(1.)}, {"q": tf.constant(3.)}]`
- * after load to the right saved TensorFlow function, we need to represent the
- * nested structure and the strings, recording that we have a trace for anything
- * matching `[{"a": tf.TensorSpec(None, tf.float32)}, {"q": tf.TensorSpec([],
- * tf.float64)}]` as an example.
- * Likewise functions may return nested structures of Tensors, for example
- * returning a dictionary mapping from strings to Tensors. In order for the
- * loaded function to return the same structure we need to serialize it.
- * This is an ergonomic aid for working with loaded SavedModels, not a promise
- * to serialize all possible function signatures. For example we do not expect
- * to pickle generic Python objects, and ideally we'd stay language-agnostic.
- * </pre>
- *
- * Protobuf type {@code tensorflow.StructuredValue}
- */
-public  final class StructuredValue extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.StructuredValue)
-    StructuredValueOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use StructuredValue.newBuilder() to construct.
-  private StructuredValue(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private StructuredValue() {
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new StructuredValue();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private StructuredValue(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 10: {
-            org.tensorflow.proto.framework.NoneValue.Builder subBuilder = null;
-            if (kindCase_ == 1) {
-              subBuilder = ((org.tensorflow.proto.framework.NoneValue) kind_).toBuilder();
-            }
-            kind_ =
-                input.readMessage(org.tensorflow.proto.framework.NoneValue.parser(), extensionRegistry);
-            if (subBuilder != null) {
-              subBuilder.mergeFrom((org.tensorflow.proto.framework.NoneValue) kind_);
-              kind_ = subBuilder.buildPartial();
-            }
-            kindCase_ = 1;
-            break;
-          }
-          case 89: {
-            kindCase_ = 11;
-            kind_ = input.readDouble();
-            break;
-          }
-          case 96: {
-            kindCase_ = 12;
-            kind_ = input.readSInt64();
-            break;
-          }
-          case 106: {
-            java.lang.String s = input.readStringRequireUtf8();
-            kindCase_ = 13;
-            kind_ = s;
-            break;
-          }
-          case 112: {
-            kindCase_ = 14;
-            kind_ = input.readBool();
-            break;
-          }
-          case 250: {
-            org.tensorflow.proto.framework.TensorShapeProto.Builder subBuilder = null;
-            if (kindCase_ == 31) {
-              subBuilder = ((org.tensorflow.proto.framework.TensorShapeProto) kind_).toBuilder();
-            }
-            kind_ =
-                input.readMessage(org.tensorflow.proto.framework.TensorShapeProto.parser(), extensionRegistry);
-            if (subBuilder != null) {
-              subBuilder.mergeFrom((org.tensorflow.proto.framework.TensorShapeProto) kind_);
-              kind_ = subBuilder.buildPartial();
-            }
-            kindCase_ = 31;
-            break;
-          }
-          case 256: {
-            int rawValue = input.readEnum();
-            kindCase_ = 32;
-            kind_ = rawValue;
-            break;
-          }
-          case 266: {
-            org.tensorflow.proto.framework.TensorSpecProto.Builder subBuilder = null;
-            if (kindCase_ == 33) {
-              subBuilder = ((org.tensorflow.proto.framework.TensorSpecProto) kind_).toBuilder();
-            }
-            kind_ =
-                input.readMessage(org.tensorflow.proto.framework.TensorSpecProto.parser(), extensionRegistry);
-            if (subBuilder != null) {
-              subBuilder.mergeFrom((org.tensorflow.proto.framework.TensorSpecProto) kind_);
-              kind_ = subBuilder.buildPartial();
-            }
-            kindCase_ = 33;
-            break;
-          }
-          case 274: {
-            org.tensorflow.proto.framework.TypeSpecProto.Builder subBuilder = null;
-            if (kindCase_ == 34) {
-              subBuilder = ((org.tensorflow.proto.framework.TypeSpecProto) kind_).toBuilder();
-            }
-            kind_ =
-                input.readMessage(org.tensorflow.proto.framework.TypeSpecProto.parser(), extensionRegistry);
-            if (subBuilder != null) {
-              subBuilder.mergeFrom((org.tensorflow.proto.framework.TypeSpecProto) kind_);
-              kind_ = subBuilder.buildPartial();
-            }
-            kindCase_ = 34;
-            break;
-          }
-          case 282: {
-            org.tensorflow.proto.framework.BoundedTensorSpecProto.Builder subBuilder = null;
-            if (kindCase_ == 35) {
-              subBuilder = ((org.tensorflow.proto.framework.BoundedTensorSpecProto) kind_).toBuilder();
-            }
-            kind_ =
-                input.readMessage(org.tensorflow.proto.framework.BoundedTensorSpecProto.parser(), extensionRegistry);
-            if (subBuilder != null) {
-              subBuilder.mergeFrom((org.tensorflow.proto.framework.BoundedTensorSpecProto) kind_);
-              kind_ = subBuilder.buildPartial();
-            }
-            kindCase_ = 35;
-            break;
-          }
-          case 410: {
-            org.tensorflow.proto.framework.ListValue.Builder subBuilder = null;
-            if (kindCase_ == 51) {
-              subBuilder = ((org.tensorflow.proto.framework.ListValue) kind_).toBuilder();
-            }
-            kind_ =
-                input.readMessage(org.tensorflow.proto.framework.ListValue.parser(), extensionRegistry);
-            if (subBuilder != null) {
-              subBuilder.mergeFrom((org.tensorflow.proto.framework.ListValue) kind_);
-              kind_ = subBuilder.buildPartial();
-            }
-            kindCase_ = 51;
-            break;
-          }
-          case 418: {
-            org.tensorflow.proto.framework.TupleValue.Builder subBuilder = null;
-            if (kindCase_ == 52) {
-              subBuilder = ((org.tensorflow.proto.framework.TupleValue) kind_).toBuilder();
-            }
-            kind_ =
-                input.readMessage(org.tensorflow.proto.framework.TupleValue.parser(), extensionRegistry);
-            if (subBuilder != null) {
-              subBuilder.mergeFrom((org.tensorflow.proto.framework.TupleValue) kind_);
-              kind_ = subBuilder.buildPartial();
-            }
-            kindCase_ = 52;
-            break;
-          }
-          case 426: {
-            org.tensorflow.proto.framework.DictValue.Builder subBuilder = null;
-            if (kindCase_ == 53) {
-              subBuilder = ((org.tensorflow.proto.framework.DictValue) kind_).toBuilder();
-            }
-            kind_ =
-                input.readMessage(org.tensorflow.proto.framework.DictValue.parser(), extensionRegistry);
-            if (subBuilder != null) {
-              subBuilder.mergeFrom((org.tensorflow.proto.framework.DictValue) kind_);
-              kind_ = subBuilder.buildPartial();
-            }
-            kindCase_ = 53;
-            break;
-          }
-          case 434: {
-            org.tensorflow.proto.framework.NamedTupleValue.Builder subBuilder = null;
-            if (kindCase_ == 54) {
-              subBuilder = ((org.tensorflow.proto.framework.NamedTupleValue) kind_).toBuilder();
-            }
-            kind_ =
-                input.readMessage(org.tensorflow.proto.framework.NamedTupleValue.parser(), extensionRegistry);
-            if (subBuilder != null) {
-              subBuilder.mergeFrom((org.tensorflow.proto.framework.NamedTupleValue) kind_);
-              kind_ = subBuilder.buildPartial();
-            }
-            kindCase_ = 54;
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.framework.StructProtos.internal_static_tensorflow_StructuredValue_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.framework.StructProtos.internal_static_tensorflow_StructuredValue_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.framework.StructuredValue.class, org.tensorflow.proto.framework.StructuredValue.Builder.class);
-  }
-
-  private int kindCase_ = 0;
-  private java.lang.Object kind_;
-  public enum KindCase
-      implements com.google.protobuf.Internal.EnumLite {
-    NONE_VALUE(1),
-    FLOAT64_VALUE(11),
-    INT64_VALUE(12),
-    STRING_VALUE(13),
-    BOOL_VALUE(14),
-    TENSOR_SHAPE_VALUE(31),
-    TENSOR_DTYPE_VALUE(32),
-    TENSOR_SPEC_VALUE(33),
-    TYPE_SPEC_VALUE(34),
-    BOUNDED_TENSOR_SPEC_VALUE(35),
-    LIST_VALUE(51),
-    TUPLE_VALUE(52),
-    DICT_VALUE(53),
-    NAMED_TUPLE_VALUE(54),
-    KIND_NOT_SET(0);
-    private final int value;
-    private KindCase(int value) {
-      this.value = value;
-    }
-    /**
-     * @deprecated Use {@link #forNumber(int)} instead.
-     */
-    @java.lang.Deprecated
-    public static KindCase valueOf(int value) {
-      return forNumber(value);
-    }
-
-    public static KindCase forNumber(int value) {
-      switch (value) {
-        case 1: return NONE_VALUE;
-        case 11: return FLOAT64_VALUE;
-        case 12: return INT64_VALUE;
-        case 13: return STRING_VALUE;
-        case 14: return BOOL_VALUE;
-        case 31: return TENSOR_SHAPE_VALUE;
-        case 32: return TENSOR_DTYPE_VALUE;
-        case 33: return TENSOR_SPEC_VALUE;
-        case 34: return TYPE_SPEC_VALUE;
-        case 35: return BOUNDED_TENSOR_SPEC_VALUE;
-        case 51: return LIST_VALUE;
-        case 52: return TUPLE_VALUE;
-        case 53: return DICT_VALUE;
-        case 54: return NAMED_TUPLE_VALUE;
-        case 0: return KIND_NOT_SET;
-        default: return null;
-      }
-    }
-    public int getNumber() {
-      return this.value;
-    }
-  };
-
-  public KindCase
-  getKindCase() {
-    return KindCase.forNumber(
-        kindCase_);
-  }
-
-  public static final int NONE_VALUE_FIELD_NUMBER = 1;
-  /**
-   * <pre>
-   * Represents None.
-   * </pre>
-   *
-   * <code>.tensorflow.NoneValue none_value = 1;</code>
-   */
-  public boolean hasNoneValue() {
-    return kindCase_ == 1;
-  }
-  /**
-   * <pre>
-   * Represents None.
-   * </pre>
-   *
-   * <code>.tensorflow.NoneValue none_value = 1;</code>
-   */
-  public org.tensorflow.proto.framework.NoneValue getNoneValue() {
-    if (kindCase_ == 1) {
-       return (org.tensorflow.proto.framework.NoneValue) kind_;
-    }
-    return org.tensorflow.proto.framework.NoneValue.getDefaultInstance();
-  }
-  /**
-   * <pre>
-   * Represents None.
-   * </pre>
-   *
-   * <code>.tensorflow.NoneValue none_value = 1;</code>
-   */
-  public org.tensorflow.proto.framework.NoneValueOrBuilder getNoneValueOrBuilder() {
-    if (kindCase_ == 1) {
-       return (org.tensorflow.proto.framework.NoneValue) kind_;
-    }
-    return org.tensorflow.proto.framework.NoneValue.getDefaultInstance();
-  }
-
-  public static final int FLOAT64_VALUE_FIELD_NUMBER = 11;
-  /**
-   * <pre>
-   * Represents a double-precision floating-point value (a Python `float`).
-   * </pre>
-   *
-   * <code>double float64_value = 11;</code>
-   */
-  public double getFloat64Value() {
-    if (kindCase_ == 11) {
-      return (java.lang.Double) kind_;
-    }
-    return 0D;
-  }
-
-  public static final int INT64_VALUE_FIELD_NUMBER = 12;
-  /**
-   * <pre>
-   * Represents a signed integer value, limited to 64 bits.
-   * Larger values from Python's arbitrary-precision integers are unsupported.
-   * </pre>
-   *
-   * <code>sint64 int64_value = 12;</code>
-   */
-  public long getInt64Value() {
-    if (kindCase_ == 12) {
-      return (java.lang.Long) kind_;
-    }
-    return 0L;
-  }
-
-  public static final int STRING_VALUE_FIELD_NUMBER = 13;
-  /**
-   * <pre>
-   * Represents a string of Unicode characters stored in a Python `str`.
-   * In Python 3, this is exactly what type `str` is.
-   * In Python 2, this is the UTF-8 encoding of the characters.
-   * For strings with ASCII characters only (as often used in TensorFlow code)
-   * there is effectively no difference between the language versions.
-   * The obsolescent `unicode` type of Python 2 is not supported here.
-   * </pre>
-   *
-   * <code>string string_value = 13;</code>
-   */
-  public java.lang.String getStringValue() {
-    java.lang.Object ref = "";
-    if (kindCase_ == 13) {
-      ref = kind_;
-    }
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      if (kindCase_ == 13) {
-        kind_ = s;
-      }
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * Represents a string of Unicode characters stored in a Python `str`.
-   * In Python 3, this is exactly what type `str` is.
-   * In Python 2, this is the UTF-8 encoding of the characters.
-   * For strings with ASCII characters only (as often used in TensorFlow code)
-   * there is effectively no difference between the language versions.
-   * The obsolescent `unicode` type of Python 2 is not supported here.
-   * </pre>
-   *
-   * <code>string string_value = 13;</code>
-   */
-  public com.google.protobuf.ByteString
-      getStringValueBytes() {
-    java.lang.Object ref = "";
-    if (kindCase_ == 13) {
-      ref = kind_;
-    }
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      if (kindCase_ == 13) {
-        kind_ = b;
-      }
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int BOOL_VALUE_FIELD_NUMBER = 14;
-  /**
-   * <pre>
-   * Represents a boolean value.
-   * </pre>
-   *
-   * <code>bool bool_value = 14;</code>
-   */
-  public boolean getBoolValue() {
-    if (kindCase_ == 14) {
-      return (java.lang.Boolean) kind_;
-    }
-    return false;
-  }
-
-  public static final int TENSOR_SHAPE_VALUE_FIELD_NUMBER = 31;
-  /**
-   * <pre>
-   * Represents a TensorShape.
-   * </pre>
-   *
-   * <code>.tensorflow.TensorShapeProto tensor_shape_value = 31;</code>
-   */
-  public boolean hasTensorShapeValue() {
-    return kindCase_ == 31;
-  }
-  /**
-   * <pre>
-   * Represents a TensorShape.
-   * </pre>
-   *
-   * <code>.tensorflow.TensorShapeProto tensor_shape_value = 31;</code>
-   */
-  public org.tensorflow.proto.framework.TensorShapeProto getTensorShapeValue() {
-    if (kindCase_ == 31) {
-       return (org.tensorflow.proto.framework.TensorShapeProto) kind_;
-    }
-    return org.tensorflow.proto.framework.TensorShapeProto.getDefaultInstance();
-  }
-  /**
-   * <pre>
-   * Represents a TensorShape.
-   * </pre>
-   *
-   * <code>.tensorflow.TensorShapeProto tensor_shape_value = 31;</code>
-   */
-  public org.tensorflow.proto.framework.TensorShapeProtoOrBuilder getTensorShapeValueOrBuilder() {
-    if (kindCase_ == 31) {
-       return (org.tensorflow.proto.framework.TensorShapeProto) kind_;
-    }
-    return org.tensorflow.proto.framework.TensorShapeProto.getDefaultInstance();
-  }
-
-  public static final int TENSOR_DTYPE_VALUE_FIELD_NUMBER = 32;
-  /**
-   * <pre>
-   * Represents an enum value for dtype.
-   * </pre>
-   *
-   * <code>.tensorflow.DataType tensor_dtype_value = 32;</code>
-   */
-  public int getTensorDtypeValueValue() {
-    if (kindCase_ == 32) {
-      return (java.lang.Integer) kind_;
-    }
-    return 0;
-  }
-  /**
-   * <pre>
-   * Represents an enum value for dtype.
-   * </pre>
-   *
-   * <code>.tensorflow.DataType tensor_dtype_value = 32;</code>
-   */
-  public org.tensorflow.proto.framework.DataType getTensorDtypeValue() {
-    if (kindCase_ == 32) {
-      @SuppressWarnings("deprecation")
-      org.tensorflow.proto.framework.DataType result = org.tensorflow.proto.framework.DataType.valueOf(
-          (java.lang.Integer) kind_);
-      return result == null ? org.tensorflow.proto.framework.DataType.UNRECOGNIZED : result;
-    }
-    return org.tensorflow.proto.framework.DataType.DT_INVALID;
-  }
-
-  public static final int TENSOR_SPEC_VALUE_FIELD_NUMBER = 33;
-  /**
-   * <pre>
-   * Represents a value for tf.TensorSpec.
-   * </pre>
-   *
-   * <code>.tensorflow.TensorSpecProto tensor_spec_value = 33;</code>
-   */
-  public boolean hasTensorSpecValue() {
-    return kindCase_ == 33;
-  }
-  /**
-   * <pre>
-   * Represents a value for tf.TensorSpec.
-   * </pre>
-   *
-   * <code>.tensorflow.TensorSpecProto tensor_spec_value = 33;</code>
-   */
-  public org.tensorflow.proto.framework.TensorSpecProto getTensorSpecValue() {
-    if (kindCase_ == 33) {
-       return (org.tensorflow.proto.framework.TensorSpecProto) kind_;
-    }
-    return org.tensorflow.proto.framework.TensorSpecProto.getDefaultInstance();
-  }
-  /**
-   * <pre>
-   * Represents a value for tf.TensorSpec.
-   * </pre>
-   *
-   * <code>.tensorflow.TensorSpecProto tensor_spec_value = 33;</code>
-   */
-  public org.tensorflow.proto.framework.TensorSpecProtoOrBuilder getTensorSpecValueOrBuilder() {
-    if (kindCase_ == 33) {
-       return (org.tensorflow.proto.framework.TensorSpecProto) kind_;
-    }
-    return org.tensorflow.proto.framework.TensorSpecProto.getDefaultInstance();
-  }
-
-  public static final int TYPE_SPEC_VALUE_FIELD_NUMBER = 34;
-  /**
-   * <pre>
-   * Represents a value for tf.TypeSpec.
-   * </pre>
-   *
-   * <code>.tensorflow.TypeSpecProto type_spec_value = 34;</code>
-   */
-  public boolean hasTypeSpecValue() {
-    return kindCase_ == 34;
-  }
-  /**
-   * <pre>
-   * Represents a value for tf.TypeSpec.
-   * </pre>
-   *
-   * <code>.tensorflow.TypeSpecProto type_spec_value = 34;</code>
-   */
-  public org.tensorflow.proto.framework.TypeSpecProto getTypeSpecValue() {
-    if (kindCase_ == 34) {
-       return (org.tensorflow.proto.framework.TypeSpecProto) kind_;
-    }
-    return org.tensorflow.proto.framework.TypeSpecProto.getDefaultInstance();
-  }
-  /**
-   * <pre>
-   * Represents a value for tf.TypeSpec.
-   * </pre>
-   *
-   * <code>.tensorflow.TypeSpecProto type_spec_value = 34;</code>
-   */
-  public org.tensorflow.proto.framework.TypeSpecProtoOrBuilder getTypeSpecValueOrBuilder() {
-    if (kindCase_ == 34) {
-       return (org.tensorflow.proto.framework.TypeSpecProto) kind_;
-    }
-    return org.tensorflow.proto.framework.TypeSpecProto.getDefaultInstance();
-  }
-
-  public static final int BOUNDED_TENSOR_SPEC_VALUE_FIELD_NUMBER = 35;
-  /**
-   * <pre>
-   * Represents a value for tf.BoundedTensorSpec.
-   * </pre>
-   *
-   * <code>.tensorflow.BoundedTensorSpecProto bounded_tensor_spec_value = 35;</code>
-   */
-  public boolean hasBoundedTensorSpecValue() {
-    return kindCase_ == 35;
-  }
-  /**
-   * <pre>
-   * Represents a value for tf.BoundedTensorSpec.
-   * </pre>
-   *
-   * <code>.tensorflow.BoundedTensorSpecProto bounded_tensor_spec_value = 35;</code>
-   */
-  public org.tensorflow.proto.framework.BoundedTensorSpecProto getBoundedTensorSpecValue() {
-    if (kindCase_ == 35) {
-       return (org.tensorflow.proto.framework.BoundedTensorSpecProto) kind_;
-    }
-    return org.tensorflow.proto.framework.BoundedTensorSpecProto.getDefaultInstance();
-  }
-  /**
-   * <pre>
-   * Represents a value for tf.BoundedTensorSpec.
-   * </pre>
-   *
-   * <code>.tensorflow.BoundedTensorSpecProto bounded_tensor_spec_value = 35;</code>
-   */
-  public org.tensorflow.proto.framework.BoundedTensorSpecProtoOrBuilder getBoundedTensorSpecValueOrBuilder() {
-    if (kindCase_ == 35) {
-       return (org.tensorflow.proto.framework.BoundedTensorSpecProto) kind_;
-    }
-    return org.tensorflow.proto.framework.BoundedTensorSpecProto.getDefaultInstance();
-  }
-
-  public static final int LIST_VALUE_FIELD_NUMBER = 51;
-  /**
-   * <pre>
-   * Represents a list of `Value`.
-   * </pre>
-   *
-   * <code>.tensorflow.ListValue list_value = 51;</code>
-   */
-  public boolean hasListValue() {
-    return kindCase_ == 51;
-  }
-  /**
-   * <pre>
-   * Represents a list of `Value`.
-   * </pre>
-   *
-   * <code>.tensorflow.ListValue list_value = 51;</code>
-   */
-  public org.tensorflow.proto.framework.ListValue getListValue() {
-    if (kindCase_ == 51) {
-       return (org.tensorflow.proto.framework.ListValue) kind_;
-    }
-    return org.tensorflow.proto.framework.ListValue.getDefaultInstance();
-  }
-  /**
-   * <pre>
-   * Represents a list of `Value`.
-   * </pre>
-   *
-   * <code>.tensorflow.ListValue list_value = 51;</code>
-   */
-  public org.tensorflow.proto.framework.ListValueOrBuilder getListValueOrBuilder() {
-    if (kindCase_ == 51) {
-       return (org.tensorflow.proto.framework.ListValue) kind_;
-    }
-    return org.tensorflow.proto.framework.ListValue.getDefaultInstance();
-  }
-
-  public static final int TUPLE_VALUE_FIELD_NUMBER = 52;
-  /**
-   * <pre>
-   * Represents a tuple of `Value`.
-   * </pre>
-   *
-   * <code>.tensorflow.TupleValue tuple_value = 52;</code>
-   */
-  public boolean hasTupleValue() {
-    return kindCase_ == 52;
-  }
-  /**
-   * <pre>
-   * Represents a tuple of `Value`.
-   * </pre>
-   *
-   * <code>.tensorflow.TupleValue tuple_value = 52;</code>
-   */
-  public org.tensorflow.proto.framework.TupleValue getTupleValue() {
-    if (kindCase_ == 52) {
-       return (org.tensorflow.proto.framework.TupleValue) kind_;
-    }
-    return org.tensorflow.proto.framework.TupleValue.getDefaultInstance();
-  }
-  /**
-   * <pre>
-   * Represents a tuple of `Value`.
-   * </pre>
-   *
-   * <code>.tensorflow.TupleValue tuple_value = 52;</code>
-   */
-  public org.tensorflow.proto.framework.TupleValueOrBuilder getTupleValueOrBuilder() {
-    if (kindCase_ == 52) {
-       return (org.tensorflow.proto.framework.TupleValue) kind_;
-    }
-    return org.tensorflow.proto.framework.TupleValue.getDefaultInstance();
-  }
-
-  public static final int DICT_VALUE_FIELD_NUMBER = 53;
-  /**
-   * <pre>
-   * Represents a dict `Value`.
-   * </pre>
-   *
-   * <code>.tensorflow.DictValue dict_value = 53;</code>
-   */
-  public boolean hasDictValue() {
-    return kindCase_ == 53;
-  }
-  /**
-   * <pre>
-   * Represents a dict `Value`.
-   * </pre>
-   *
-   * <code>.tensorflow.DictValue dict_value = 53;</code>
-   */
-  public org.tensorflow.proto.framework.DictValue getDictValue() {
-    if (kindCase_ == 53) {
-       return (org.tensorflow.proto.framework.DictValue) kind_;
-    }
-    return org.tensorflow.proto.framework.DictValue.getDefaultInstance();
-  }
-  /**
-   * <pre>
-   * Represents a dict `Value`.
-   * </pre>
-   *
-   * <code>.tensorflow.DictValue dict_value = 53;</code>
-   */
-  public org.tensorflow.proto.framework.DictValueOrBuilder getDictValueOrBuilder() {
-    if (kindCase_ == 53) {
-       return (org.tensorflow.proto.framework.DictValue) kind_;
-    }
-    return org.tensorflow.proto.framework.DictValue.getDefaultInstance();
-  }
-
-  public static final int NAMED_TUPLE_VALUE_FIELD_NUMBER = 54;
-  /**
-   * <pre>
-   * Represents Python's namedtuple.
-   * </pre>
-   *
-   * <code>.tensorflow.NamedTupleValue named_tuple_value = 54;</code>
-   */
-  public boolean hasNamedTupleValue() {
-    return kindCase_ == 54;
-  }
-  /**
-   * <pre>
-   * Represents Python's namedtuple.
-   * </pre>
-   *
-   * <code>.tensorflow.NamedTupleValue named_tuple_value = 54;</code>
-   */
-  public org.tensorflow.proto.framework.NamedTupleValue getNamedTupleValue() {
-    if (kindCase_ == 54) {
-       return (org.tensorflow.proto.framework.NamedTupleValue) kind_;
-    }
-    return org.tensorflow.proto.framework.NamedTupleValue.getDefaultInstance();
-  }
-  /**
-   * <pre>
-   * Represents Python's namedtuple.
-   * </pre>
-   *
-   * <code>.tensorflow.NamedTupleValue named_tuple_value = 54;</code>
-   */
-  public org.tensorflow.proto.framework.NamedTupleValueOrBuilder getNamedTupleValueOrBuilder() {
-    if (kindCase_ == 54) {
-       return (org.tensorflow.proto.framework.NamedTupleValue) kind_;
-    }
-    return org.tensorflow.proto.framework.NamedTupleValue.getDefaultInstance();
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    if (kindCase_ == 1) {
-      output.writeMessage(1, (org.tensorflow.proto.framework.NoneValue) kind_);
-    }
-    if (kindCase_ == 11) {
-      output.writeDouble(
-          11, (double)((java.lang.Double) kind_));
-    }
-    if (kindCase_ == 12) {
-      output.writeSInt64(
-          12, (long)((java.lang.Long) kind_));
-    }
-    if (kindCase_ == 13) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 13, kind_);
-    }
-    if (kindCase_ == 14) {
-      output.writeBool(
-          14, (boolean)((java.lang.Boolean) kind_));
-    }
-    if (kindCase_ == 31) {
-      output.writeMessage(31, (org.tensorflow.proto.framework.TensorShapeProto) kind_);
-    }
-    if (kindCase_ == 32) {
-      output.writeEnum(32, ((java.lang.Integer) kind_));
-    }
-    if (kindCase_ == 33) {
-      output.writeMessage(33, (org.tensorflow.proto.framework.TensorSpecProto) kind_);
-    }
-    if (kindCase_ == 34) {
-      output.writeMessage(34, (org.tensorflow.proto.framework.TypeSpecProto) kind_);
-    }
-    if (kindCase_ == 35) {
-      output.writeMessage(35, (org.tensorflow.proto.framework.BoundedTensorSpecProto) kind_);
-    }
-    if (kindCase_ == 51) {
-      output.writeMessage(51, (org.tensorflow.proto.framework.ListValue) kind_);
-    }
-    if (kindCase_ == 52) {
-      output.writeMessage(52, (org.tensorflow.proto.framework.TupleValue) kind_);
-    }
-    if (kindCase_ == 53) {
-      output.writeMessage(53, (org.tensorflow.proto.framework.DictValue) kind_);
-    }
-    if (kindCase_ == 54) {
-      output.writeMessage(54, (org.tensorflow.proto.framework.NamedTupleValue) kind_);
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (kindCase_ == 1) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(1, (org.tensorflow.proto.framework.NoneValue) kind_);
-    }
-    if (kindCase_ == 11) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeDoubleSize(
-            11, (double)((java.lang.Double) kind_));
-    }
-    if (kindCase_ == 12) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeSInt64Size(
-            12, (long)((java.lang.Long) kind_));
-    }
-    if (kindCase_ == 13) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(13, kind_);
-    }
-    if (kindCase_ == 14) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeBoolSize(
-            14, (boolean)((java.lang.Boolean) kind_));
-    }
-    if (kindCase_ == 31) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(31, (org.tensorflow.proto.framework.TensorShapeProto) kind_);
-    }
-    if (kindCase_ == 32) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeEnumSize(32, ((java.lang.Integer) kind_));
-    }
-    if (kindCase_ == 33) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(33, (org.tensorflow.proto.framework.TensorSpecProto) kind_);
-    }
-    if (kindCase_ == 34) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(34, (org.tensorflow.proto.framework.TypeSpecProto) kind_);
-    }
-    if (kindCase_ == 35) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(35, (org.tensorflow.proto.framework.BoundedTensorSpecProto) kind_);
-    }
-    if (kindCase_ == 51) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(51, (org.tensorflow.proto.framework.ListValue) kind_);
-    }
-    if (kindCase_ == 52) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(52, (org.tensorflow.proto.framework.TupleValue) kind_);
-    }
-    if (kindCase_ == 53) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(53, (org.tensorflow.proto.framework.DictValue) kind_);
-    }
-    if (kindCase_ == 54) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(54, (org.tensorflow.proto.framework.NamedTupleValue) kind_);
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.framework.StructuredValue)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.framework.StructuredValue other = (org.tensorflow.proto.framework.StructuredValue) obj;
-
-    if (!getKindCase().equals(other.getKindCase())) return false;
-    switch (kindCase_) {
-      case 1:
-        if (!getNoneValue()
-            .equals(other.getNoneValue())) return false;
-        break;
-      case 11:
-        if (java.lang.Double.doubleToLongBits(getFloat64Value())
-            != java.lang.Double.doubleToLongBits(
-                other.getFloat64Value())) return false;
-        break;
-      case 12:
-        if (getInt64Value()
-            != other.getInt64Value()) return false;
-        break;
-      case 13:
-        if (!getStringValue()
-            .equals(other.getStringValue())) return false;
-        break;
-      case 14:
-        if (getBoolValue()
-            != other.getBoolValue()) return false;
-        break;
-      case 31:
-        if (!getTensorShapeValue()
-            .equals(other.getTensorShapeValue())) return false;
-        break;
-      case 32:
-        if (getTensorDtypeValueValue()
-            != other.getTensorDtypeValueValue()) return false;
-        break;
-      case 33:
-        if (!getTensorSpecValue()
-            .equals(other.getTensorSpecValue())) return false;
-        break;
-      case 34:
-        if (!getTypeSpecValue()
-            .equals(other.getTypeSpecValue())) return false;
-        break;
-      case 35:
-        if (!getBoundedTensorSpecValue()
-            .equals(other.getBoundedTensorSpecValue())) return false;
-        break;
-      case 51:
-        if (!getListValue()
-            .equals(other.getListValue())) return false;
-        break;
-      case 52:
-        if (!getTupleValue()
-            .equals(other.getTupleValue())) return false;
-        break;
-      case 53:
-        if (!getDictValue()
-            .equals(other.getDictValue())) return false;
-        break;
-      case 54:
-        if (!getNamedTupleValue()
-            .equals(other.getNamedTupleValue())) return false;
-        break;
-      case 0:
-      default:
-    }
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    switch (kindCase_) {
-      case 1:
-        hash = (37 * hash) + NONE_VALUE_FIELD_NUMBER;
-        hash = (53 * hash) + getNoneValue().hashCode();
-        break;
-      case 11:
-        hash = (37 * hash) + FLOAT64_VALUE_FIELD_NUMBER;
-        hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-            java.lang.Double.doubleToLongBits(getFloat64Value()));
-        break;
-      case 12:
-        hash = (37 * hash) + INT64_VALUE_FIELD_NUMBER;
-        hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-            getInt64Value());
-        break;
-      case 13:
-        hash = (37 * hash) + STRING_VALUE_FIELD_NUMBER;
-        hash = (53 * hash) + getStringValue().hashCode();
-        break;
-      case 14:
-        hash = (37 * hash) + BOOL_VALUE_FIELD_NUMBER;
-        hash = (53 * hash) + com.google.protobuf.Internal.hashBoolean(
-            getBoolValue());
-        break;
-      case 31:
-        hash = (37 * hash) + TENSOR_SHAPE_VALUE_FIELD_NUMBER;
-        hash = (53 * hash) + getTensorShapeValue().hashCode();
-        break;
-      case 32:
-        hash = (37 * hash) + TENSOR_DTYPE_VALUE_FIELD_NUMBER;
-        hash = (53 * hash) + getTensorDtypeValueValue();
-        break;
-      case 33:
-        hash = (37 * hash) + TENSOR_SPEC_VALUE_FIELD_NUMBER;
-        hash = (53 * hash) + getTensorSpecValue().hashCode();
-        break;
-      case 34:
-        hash = (37 * hash) + TYPE_SPEC_VALUE_FIELD_NUMBER;
-        hash = (53 * hash) + getTypeSpecValue().hashCode();
-        break;
-      case 35:
-        hash = (37 * hash) + BOUNDED_TENSOR_SPEC_VALUE_FIELD_NUMBER;
-        hash = (53 * hash) + getBoundedTensorSpecValue().hashCode();
-        break;
-      case 51:
-        hash = (37 * hash) + LIST_VALUE_FIELD_NUMBER;
-        hash = (53 * hash) + getListValue().hashCode();
-        break;
-      case 52:
-        hash = (37 * hash) + TUPLE_VALUE_FIELD_NUMBER;
-        hash = (53 * hash) + getTupleValue().hashCode();
-        break;
-      case 53:
-        hash = (37 * hash) + DICT_VALUE_FIELD_NUMBER;
-        hash = (53 * hash) + getDictValue().hashCode();
-        break;
-      case 54:
-        hash = (37 * hash) + NAMED_TUPLE_VALUE_FIELD_NUMBER;
-        hash = (53 * hash) + getNamedTupleValue().hashCode();
-        break;
-      case 0:
-      default:
-    }
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.framework.StructuredValue parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.StructuredValue parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.StructuredValue parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.StructuredValue parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.StructuredValue parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.StructuredValue parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.StructuredValue parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.StructuredValue parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.StructuredValue parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.StructuredValue parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.StructuredValue parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.StructuredValue parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.framework.StructuredValue prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * <pre>
-   * `StructuredValue` represents a dynamically typed value representing various
-   * data structures that are inspired by Python data structures typically used in
-   * TensorFlow functions as inputs and outputs.
-   * For example when saving a Layer there may be a `training` argument. If the
-   * user passes a boolean True/False, that switches between two concrete
-   * TensorFlow functions. In order to switch between them in the same way after
-   * loading the SavedModel, we need to represent "True" and "False".
-   * A more advanced example might be a function which takes a list of
-   * dictionaries mapping from strings to Tensors. In order to map from
-   * user-specified arguments `[{"a": tf.constant(1.)}, {"q": tf.constant(3.)}]`
-   * after load to the right saved TensorFlow function, we need to represent the
-   * nested structure and the strings, recording that we have a trace for anything
-   * matching `[{"a": tf.TensorSpec(None, tf.float32)}, {"q": tf.TensorSpec([],
-   * tf.float64)}]` as an example.
-   * Likewise functions may return nested structures of Tensors, for example
-   * returning a dictionary mapping from strings to Tensors. In order for the
-   * loaded function to return the same structure we need to serialize it.
-   * This is an ergonomic aid for working with loaded SavedModels, not a promise
-   * to serialize all possible function signatures. For example we do not expect
-   * to pickle generic Python objects, and ideally we'd stay language-agnostic.
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.StructuredValue}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.StructuredValue)
-      org.tensorflow.proto.framework.StructuredValueOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.StructProtos.internal_static_tensorflow_StructuredValue_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.StructProtos.internal_static_tensorflow_StructuredValue_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.StructuredValue.class, org.tensorflow.proto.framework.StructuredValue.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.framework.StructuredValue.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      kindCase_ = 0;
-      kind_ = null;
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.framework.StructProtos.internal_static_tensorflow_StructuredValue_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.StructuredValue getDefaultInstanceForType() {
-      return org.tensorflow.proto.framework.StructuredValue.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.StructuredValue build() {
-      org.tensorflow.proto.framework.StructuredValue result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.StructuredValue buildPartial() {
-      org.tensorflow.proto.framework.StructuredValue result = new org.tensorflow.proto.framework.StructuredValue(this);
-      if (kindCase_ == 1) {
-        if (noneValueBuilder_ == null) {
-          result.kind_ = kind_;
-        } else {
-          result.kind_ = noneValueBuilder_.build();
-        }
-      }
-      if (kindCase_ == 11) {
-        result.kind_ = kind_;
-      }
-      if (kindCase_ == 12) {
-        result.kind_ = kind_;
-      }
-      if (kindCase_ == 13) {
-        result.kind_ = kind_;
-      }
-      if (kindCase_ == 14) {
-        result.kind_ = kind_;
-      }
-      if (kindCase_ == 31) {
-        if (tensorShapeValueBuilder_ == null) {
-          result.kind_ = kind_;
-        } else {
-          result.kind_ = tensorShapeValueBuilder_.build();
-        }
-      }
-      if (kindCase_ == 32) {
-        result.kind_ = kind_;
-      }
-      if (kindCase_ == 33) {
-        if (tensorSpecValueBuilder_ == null) {
-          result.kind_ = kind_;
-        } else {
-          result.kind_ = tensorSpecValueBuilder_.build();
-        }
-      }
-      if (kindCase_ == 34) {
-        if (typeSpecValueBuilder_ == null) {
-          result.kind_ = kind_;
-        } else {
-          result.kind_ = typeSpecValueBuilder_.build();
-        }
-      }
-      if (kindCase_ == 35) {
-        if (boundedTensorSpecValueBuilder_ == null) {
-          result.kind_ = kind_;
-        } else {
-          result.kind_ = boundedTensorSpecValueBuilder_.build();
-        }
-      }
-      if (kindCase_ == 51) {
-        if (listValueBuilder_ == null) {
-          result.kind_ = kind_;
-        } else {
-          result.kind_ = listValueBuilder_.build();
-        }
-      }
-      if (kindCase_ == 52) {
-        if (tupleValueBuilder_ == null) {
-          result.kind_ = kind_;
-        } else {
-          result.kind_ = tupleValueBuilder_.build();
-        }
-      }
-      if (kindCase_ == 53) {
-        if (dictValueBuilder_ == null) {
-          result.kind_ = kind_;
-        } else {
-          result.kind_ = dictValueBuilder_.build();
-        }
-      }
-      if (kindCase_ == 54) {
-        if (namedTupleValueBuilder_ == null) {
-          result.kind_ = kind_;
-        } else {
-          result.kind_ = namedTupleValueBuilder_.build();
-        }
-      }
-      result.kindCase_ = kindCase_;
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.framework.StructuredValue) {
-        return mergeFrom((org.tensorflow.proto.framework.StructuredValue)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.framework.StructuredValue other) {
-      if (other == org.tensorflow.proto.framework.StructuredValue.getDefaultInstance()) return this;
-      switch (other.getKindCase()) {
-        case NONE_VALUE: {
-          mergeNoneValue(other.getNoneValue());
-          break;
-        }
-        case FLOAT64_VALUE: {
-          setFloat64Value(other.getFloat64Value());
-          break;
-        }
-        case INT64_VALUE: {
-          setInt64Value(other.getInt64Value());
-          break;
-        }
-        case STRING_VALUE: {
-          kindCase_ = 13;
-          kind_ = other.kind_;
-          onChanged();
-          break;
-        }
-        case BOOL_VALUE: {
-          setBoolValue(other.getBoolValue());
-          break;
-        }
-        case TENSOR_SHAPE_VALUE: {
-          mergeTensorShapeValue(other.getTensorShapeValue());
-          break;
-        }
-        case TENSOR_DTYPE_VALUE: {
-          setTensorDtypeValueValue(other.getTensorDtypeValueValue());
-          break;
-        }
-        case TENSOR_SPEC_VALUE: {
-          mergeTensorSpecValue(other.getTensorSpecValue());
-          break;
-        }
-        case TYPE_SPEC_VALUE: {
-          mergeTypeSpecValue(other.getTypeSpecValue());
-          break;
-        }
-        case BOUNDED_TENSOR_SPEC_VALUE: {
-          mergeBoundedTensorSpecValue(other.getBoundedTensorSpecValue());
-          break;
-        }
-        case LIST_VALUE: {
-          mergeListValue(other.getListValue());
-          break;
-        }
-        case TUPLE_VALUE: {
-          mergeTupleValue(other.getTupleValue());
-          break;
-        }
-        case DICT_VALUE: {
-          mergeDictValue(other.getDictValue());
-          break;
-        }
-        case NAMED_TUPLE_VALUE: {
-          mergeNamedTupleValue(other.getNamedTupleValue());
-          break;
-        }
-        case KIND_NOT_SET: {
-          break;
-        }
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.framework.StructuredValue parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.framework.StructuredValue) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-    private int kindCase_ = 0;
-    private java.lang.Object kind_;
-    public KindCase
-        getKindCase() {
-      return KindCase.forNumber(
-          kindCase_);
-    }
-
-    public Builder clearKind() {
-      kindCase_ = 0;
-      kind_ = null;
-      onChanged();
-      return this;
-    }
-
-
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.NoneValue, org.tensorflow.proto.framework.NoneValue.Builder, org.tensorflow.proto.framework.NoneValueOrBuilder> noneValueBuilder_;
-    /**
-     * <pre>
-     * Represents None.
-     * </pre>
-     *
-     * <code>.tensorflow.NoneValue none_value = 1;</code>
-     */
-    public boolean hasNoneValue() {
-      return kindCase_ == 1;
-    }
-    /**
-     * <pre>
-     * Represents None.
-     * </pre>
-     *
-     * <code>.tensorflow.NoneValue none_value = 1;</code>
-     */
-    public org.tensorflow.proto.framework.NoneValue getNoneValue() {
-      if (noneValueBuilder_ == null) {
-        if (kindCase_ == 1) {
-          return (org.tensorflow.proto.framework.NoneValue) kind_;
-        }
-        return org.tensorflow.proto.framework.NoneValue.getDefaultInstance();
-      } else {
-        if (kindCase_ == 1) {
-          return noneValueBuilder_.getMessage();
-        }
-        return org.tensorflow.proto.framework.NoneValue.getDefaultInstance();
-      }
-    }
-    /**
-     * <pre>
-     * Represents None.
-     * </pre>
-     *
-     * <code>.tensorflow.NoneValue none_value = 1;</code>
-     */
-    public Builder setNoneValue(org.tensorflow.proto.framework.NoneValue value) {
-      if (noneValueBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        kind_ = value;
-        onChanged();
-      } else {
-        noneValueBuilder_.setMessage(value);
-      }
-      kindCase_ = 1;
-      return this;
-    }
-    /**
-     * <pre>
-     * Represents None.
-     * </pre>
-     *
-     * <code>.tensorflow.NoneValue none_value = 1;</code>
-     */
-    public Builder setNoneValue(
-        org.tensorflow.proto.framework.NoneValue.Builder builderForValue) {
-      if (noneValueBuilder_ == null) {
-        kind_ = builderForValue.build();
-        onChanged();
-      } else {
-        noneValueBuilder_.setMessage(builderForValue.build());
-      }
-      kindCase_ = 1;
-      return this;
-    }
-    /**
-     * <pre>
-     * Represents None.
-     * </pre>
-     *
-     * <code>.tensorflow.NoneValue none_value = 1;</code>
-     */
-    public Builder mergeNoneValue(org.tensorflow.proto.framework.NoneValue value) {
-      if (noneValueBuilder_ == null) {
-        if (kindCase_ == 1 &&
-            kind_ != org.tensorflow.proto.framework.NoneValue.getDefaultInstance()) {
-          kind_ = org.tensorflow.proto.framework.NoneValue.newBuilder((org.tensorflow.proto.framework.NoneValue) kind_)
-              .mergeFrom(value).buildPartial();
-        } else {
-          kind_ = value;
-        }
-        onChanged();
-      } else {
-        if (kindCase_ == 1) {
-          noneValueBuilder_.mergeFrom(value);
-        }
-        noneValueBuilder_.setMessage(value);
-      }
-      kindCase_ = 1;
-      return this;
-    }
-    /**
-     * <pre>
-     * Represents None.
-     * </pre>
-     *
-     * <code>.tensorflow.NoneValue none_value = 1;</code>
-     */
-    public Builder clearNoneValue() {
-      if (noneValueBuilder_ == null) {
-        if (kindCase_ == 1) {
-          kindCase_ = 0;
-          kind_ = null;
-          onChanged();
-        }
-      } else {
-        if (kindCase_ == 1) {
-          kindCase_ = 0;
-          kind_ = null;
-        }
-        noneValueBuilder_.clear();
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Represents None.
-     * </pre>
-     *
-     * <code>.tensorflow.NoneValue none_value = 1;</code>
-     */
-    public org.tensorflow.proto.framework.NoneValue.Builder getNoneValueBuilder() {
-      return getNoneValueFieldBuilder().getBuilder();
-    }
-    /**
-     * <pre>
-     * Represents None.
-     * </pre>
-     *
-     * <code>.tensorflow.NoneValue none_value = 1;</code>
-     */
-    public org.tensorflow.proto.framework.NoneValueOrBuilder getNoneValueOrBuilder() {
-      if ((kindCase_ == 1) && (noneValueBuilder_ != null)) {
-        return noneValueBuilder_.getMessageOrBuilder();
-      } else {
-        if (kindCase_ == 1) {
-          return (org.tensorflow.proto.framework.NoneValue) kind_;
-        }
-        return org.tensorflow.proto.framework.NoneValue.getDefaultInstance();
-      }
-    }
-    /**
-     * <pre>
-     * Represents None.
-     * </pre>
-     *
-     * <code>.tensorflow.NoneValue none_value = 1;</code>
-     */
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.NoneValue, org.tensorflow.proto.framework.NoneValue.Builder, org.tensorflow.proto.framework.NoneValueOrBuilder> 
-        getNoneValueFieldBuilder() {
-      if (noneValueBuilder_ == null) {
-        if (!(kindCase_ == 1)) {
-          kind_ = org.tensorflow.proto.framework.NoneValue.getDefaultInstance();
-        }
-        noneValueBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-            org.tensorflow.proto.framework.NoneValue, org.tensorflow.proto.framework.NoneValue.Builder, org.tensorflow.proto.framework.NoneValueOrBuilder>(
-                (org.tensorflow.proto.framework.NoneValue) kind_,
-                getParentForChildren(),
-                isClean());
-        kind_ = null;
-      }
-      kindCase_ = 1;
-      onChanged();;
-      return noneValueBuilder_;
-    }
-
-    /**
-     * <pre>
-     * Represents a double-precision floating-point value (a Python `float`).
-     * </pre>
-     *
-     * <code>double float64_value = 11;</code>
-     */
-    public double getFloat64Value() {
-      if (kindCase_ == 11) {
-        return (java.lang.Double) kind_;
-      }
-      return 0D;
-    }
-    /**
-     * <pre>
-     * Represents a double-precision floating-point value (a Python `float`).
-     * </pre>
-     *
-     * <code>double float64_value = 11;</code>
-     */
-    public Builder setFloat64Value(double value) {
-      kindCase_ = 11;
-      kind_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Represents a double-precision floating-point value (a Python `float`).
-     * </pre>
-     *
-     * <code>double float64_value = 11;</code>
-     */
-    public Builder clearFloat64Value() {
-      if (kindCase_ == 11) {
-        kindCase_ = 0;
-        kind_ = null;
-        onChanged();
-      }
-      return this;
-    }
-
-    /**
-     * <pre>
-     * Represents a signed integer value, limited to 64 bits.
-     * Larger values from Python's arbitrary-precision integers are unsupported.
-     * </pre>
-     *
-     * <code>sint64 int64_value = 12;</code>
-     */
-    public long getInt64Value() {
-      if (kindCase_ == 12) {
-        return (java.lang.Long) kind_;
-      }
-      return 0L;
-    }
-    /**
-     * <pre>
-     * Represents a signed integer value, limited to 64 bits.
-     * Larger values from Python's arbitrary-precision integers are unsupported.
-     * </pre>
-     *
-     * <code>sint64 int64_value = 12;</code>
-     */
-    public Builder setInt64Value(long value) {
-      kindCase_ = 12;
-      kind_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Represents a signed integer value, limited to 64 bits.
-     * Larger values from Python's arbitrary-precision integers are unsupported.
-     * </pre>
-     *
-     * <code>sint64 int64_value = 12;</code>
-     */
-    public Builder clearInt64Value() {
-      if (kindCase_ == 12) {
-        kindCase_ = 0;
-        kind_ = null;
-        onChanged();
-      }
-      return this;
-    }
-
-    /**
-     * <pre>
-     * Represents a string of Unicode characters stored in a Python `str`.
-     * In Python 3, this is exactly what type `str` is.
-     * In Python 2, this is the UTF-8 encoding of the characters.
-     * For strings with ASCII characters only (as often used in TensorFlow code)
-     * there is effectively no difference between the language versions.
-     * The obsolescent `unicode` type of Python 2 is not supported here.
-     * </pre>
-     *
-     * <code>string string_value = 13;</code>
-     */
-    public java.lang.String getStringValue() {
-      java.lang.Object ref = "";
-      if (kindCase_ == 13) {
-        ref = kind_;
-      }
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        if (kindCase_ == 13) {
-          kind_ = s;
-        }
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Represents a string of Unicode characters stored in a Python `str`.
-     * In Python 3, this is exactly what type `str` is.
-     * In Python 2, this is the UTF-8 encoding of the characters.
-     * For strings with ASCII characters only (as often used in TensorFlow code)
-     * there is effectively no difference between the language versions.
-     * The obsolescent `unicode` type of Python 2 is not supported here.
-     * </pre>
-     *
-     * <code>string string_value = 13;</code>
-     */
-    public com.google.protobuf.ByteString
-        getStringValueBytes() {
-      java.lang.Object ref = "";
-      if (kindCase_ == 13) {
-        ref = kind_;
-      }
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        if (kindCase_ == 13) {
-          kind_ = b;
-        }
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Represents a string of Unicode characters stored in a Python `str`.
-     * In Python 3, this is exactly what type `str` is.
-     * In Python 2, this is the UTF-8 encoding of the characters.
-     * For strings with ASCII characters only (as often used in TensorFlow code)
-     * there is effectively no difference between the language versions.
-     * The obsolescent `unicode` type of Python 2 is not supported here.
-     * </pre>
-     *
-     * <code>string string_value = 13;</code>
-     */
-    public Builder setStringValue(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  kindCase_ = 13;
-      kind_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Represents a string of Unicode characters stored in a Python `str`.
-     * In Python 3, this is exactly what type `str` is.
-     * In Python 2, this is the UTF-8 encoding of the characters.
-     * For strings with ASCII characters only (as often used in TensorFlow code)
-     * there is effectively no difference between the language versions.
-     * The obsolescent `unicode` type of Python 2 is not supported here.
-     * </pre>
-     *
-     * <code>string string_value = 13;</code>
-     */
-    public Builder clearStringValue() {
-      if (kindCase_ == 13) {
-        kindCase_ = 0;
-        kind_ = null;
-        onChanged();
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Represents a string of Unicode characters stored in a Python `str`.
-     * In Python 3, this is exactly what type `str` is.
-     * In Python 2, this is the UTF-8 encoding of the characters.
-     * For strings with ASCII characters only (as often used in TensorFlow code)
-     * there is effectively no difference between the language versions.
-     * The obsolescent `unicode` type of Python 2 is not supported here.
-     * </pre>
-     *
-     * <code>string string_value = 13;</code>
-     */
-    public Builder setStringValueBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      kindCase_ = 13;
-      kind_ = value;
-      onChanged();
-      return this;
-    }
-
-    /**
-     * <pre>
-     * Represents a boolean value.
-     * </pre>
-     *
-     * <code>bool bool_value = 14;</code>
-     */
-    public boolean getBoolValue() {
-      if (kindCase_ == 14) {
-        return (java.lang.Boolean) kind_;
-      }
-      return false;
-    }
-    /**
-     * <pre>
-     * Represents a boolean value.
-     * </pre>
-     *
-     * <code>bool bool_value = 14;</code>
-     */
-    public Builder setBoolValue(boolean value) {
-      kindCase_ = 14;
-      kind_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Represents a boolean value.
-     * </pre>
-     *
-     * <code>bool bool_value = 14;</code>
-     */
-    public Builder clearBoolValue() {
-      if (kindCase_ == 14) {
-        kindCase_ = 0;
-        kind_ = null;
-        onChanged();
-      }
-      return this;
-    }
-
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.TensorShapeProto, org.tensorflow.proto.framework.TensorShapeProto.Builder, org.tensorflow.proto.framework.TensorShapeProtoOrBuilder> tensorShapeValueBuilder_;
-    /**
-     * <pre>
-     * Represents a TensorShape.
-     * </pre>
-     *
-     * <code>.tensorflow.TensorShapeProto tensor_shape_value = 31;</code>
-     */
-    public boolean hasTensorShapeValue() {
-      return kindCase_ == 31;
-    }
-    /**
-     * <pre>
-     * Represents a TensorShape.
-     * </pre>
-     *
-     * <code>.tensorflow.TensorShapeProto tensor_shape_value = 31;</code>
-     */
-    public org.tensorflow.proto.framework.TensorShapeProto getTensorShapeValue() {
-      if (tensorShapeValueBuilder_ == null) {
-        if (kindCase_ == 31) {
-          return (org.tensorflow.proto.framework.TensorShapeProto) kind_;
-        }
-        return org.tensorflow.proto.framework.TensorShapeProto.getDefaultInstance();
-      } else {
-        if (kindCase_ == 31) {
-          return tensorShapeValueBuilder_.getMessage();
-        }
-        return org.tensorflow.proto.framework.TensorShapeProto.getDefaultInstance();
-      }
-    }
-    /**
-     * <pre>
-     * Represents a TensorShape.
-     * </pre>
-     *
-     * <code>.tensorflow.TensorShapeProto tensor_shape_value = 31;</code>
-     */
-    public Builder setTensorShapeValue(org.tensorflow.proto.framework.TensorShapeProto value) {
-      if (tensorShapeValueBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        kind_ = value;
-        onChanged();
-      } else {
-        tensorShapeValueBuilder_.setMessage(value);
-      }
-      kindCase_ = 31;
-      return this;
-    }
-    /**
-     * <pre>
-     * Represents a TensorShape.
-     * </pre>
-     *
-     * <code>.tensorflow.TensorShapeProto tensor_shape_value = 31;</code>
-     */
-    public Builder setTensorShapeValue(
-        org.tensorflow.proto.framework.TensorShapeProto.Builder builderForValue) {
-      if (tensorShapeValueBuilder_ == null) {
-        kind_ = builderForValue.build();
-        onChanged();
-      } else {
-        tensorShapeValueBuilder_.setMessage(builderForValue.build());
-      }
-      kindCase_ = 31;
-      return this;
-    }
-    /**
-     * <pre>
-     * Represents a TensorShape.
-     * </pre>
-     *
-     * <code>.tensorflow.TensorShapeProto tensor_shape_value = 31;</code>
-     */
-    public Builder mergeTensorShapeValue(org.tensorflow.proto.framework.TensorShapeProto value) {
-      if (tensorShapeValueBuilder_ == null) {
-        if (kindCase_ == 31 &&
-            kind_ != org.tensorflow.proto.framework.TensorShapeProto.getDefaultInstance()) {
-          kind_ = org.tensorflow.proto.framework.TensorShapeProto.newBuilder((org.tensorflow.proto.framework.TensorShapeProto) kind_)
-              .mergeFrom(value).buildPartial();
-        } else {
-          kind_ = value;
-        }
-        onChanged();
-      } else {
-        if (kindCase_ == 31) {
-          tensorShapeValueBuilder_.mergeFrom(value);
-        }
-        tensorShapeValueBuilder_.setMessage(value);
-      }
-      kindCase_ = 31;
-      return this;
-    }
-    /**
-     * <pre>
-     * Represents a TensorShape.
-     * </pre>
-     *
-     * <code>.tensorflow.TensorShapeProto tensor_shape_value = 31;</code>
-     */
-    public Builder clearTensorShapeValue() {
-      if (tensorShapeValueBuilder_ == null) {
-        if (kindCase_ == 31) {
-          kindCase_ = 0;
-          kind_ = null;
-          onChanged();
-        }
-      } else {
-        if (kindCase_ == 31) {
-          kindCase_ = 0;
-          kind_ = null;
-        }
-        tensorShapeValueBuilder_.clear();
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Represents a TensorShape.
-     * </pre>
-     *
-     * <code>.tensorflow.TensorShapeProto tensor_shape_value = 31;</code>
-     */
-    public org.tensorflow.proto.framework.TensorShapeProto.Builder getTensorShapeValueBuilder() {
-      return getTensorShapeValueFieldBuilder().getBuilder();
-    }
-    /**
-     * <pre>
-     * Represents a TensorShape.
-     * </pre>
-     *
-     * <code>.tensorflow.TensorShapeProto tensor_shape_value = 31;</code>
-     */
-    public org.tensorflow.proto.framework.TensorShapeProtoOrBuilder getTensorShapeValueOrBuilder() {
-      if ((kindCase_ == 31) && (tensorShapeValueBuilder_ != null)) {
-        return tensorShapeValueBuilder_.getMessageOrBuilder();
-      } else {
-        if (kindCase_ == 31) {
-          return (org.tensorflow.proto.framework.TensorShapeProto) kind_;
-        }
-        return org.tensorflow.proto.framework.TensorShapeProto.getDefaultInstance();
-      }
-    }
-    /**
-     * <pre>
-     * Represents a TensorShape.
-     * </pre>
-     *
-     * <code>.tensorflow.TensorShapeProto tensor_shape_value = 31;</code>
-     */
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.TensorShapeProto, org.tensorflow.proto.framework.TensorShapeProto.Builder, org.tensorflow.proto.framework.TensorShapeProtoOrBuilder> 
-        getTensorShapeValueFieldBuilder() {
-      if (tensorShapeValueBuilder_ == null) {
-        if (!(kindCase_ == 31)) {
-          kind_ = org.tensorflow.proto.framework.TensorShapeProto.getDefaultInstance();
-        }
-        tensorShapeValueBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-            org.tensorflow.proto.framework.TensorShapeProto, org.tensorflow.proto.framework.TensorShapeProto.Builder, org.tensorflow.proto.framework.TensorShapeProtoOrBuilder>(
-                (org.tensorflow.proto.framework.TensorShapeProto) kind_,
-                getParentForChildren(),
-                isClean());
-        kind_ = null;
-      }
-      kindCase_ = 31;
-      onChanged();;
-      return tensorShapeValueBuilder_;
-    }
-
-    /**
-     * <pre>
-     * Represents an enum value for dtype.
-     * </pre>
-     *
-     * <code>.tensorflow.DataType tensor_dtype_value = 32;</code>
-     */
-    public int getTensorDtypeValueValue() {
-      if (kindCase_ == 32) {
-        return ((java.lang.Integer) kind_).intValue();
-      }
-      return 0;
-    }
-    /**
-     * <pre>
-     * Represents an enum value for dtype.
-     * </pre>
-     *
-     * <code>.tensorflow.DataType tensor_dtype_value = 32;</code>
-     */
-    public Builder setTensorDtypeValueValue(int value) {
-      kindCase_ = 32;
-      kind_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Represents an enum value for dtype.
-     * </pre>
-     *
-     * <code>.tensorflow.DataType tensor_dtype_value = 32;</code>
-     */
-    public org.tensorflow.proto.framework.DataType getTensorDtypeValue() {
-      if (kindCase_ == 32) {
-        @SuppressWarnings("deprecation")
-        org.tensorflow.proto.framework.DataType result = org.tensorflow.proto.framework.DataType.valueOf(
-            (java.lang.Integer) kind_);
-        return result == null ? org.tensorflow.proto.framework.DataType.UNRECOGNIZED : result;
-      }
-      return org.tensorflow.proto.framework.DataType.DT_INVALID;
-    }
-    /**
-     * <pre>
-     * Represents an enum value for dtype.
-     * </pre>
-     *
-     * <code>.tensorflow.DataType tensor_dtype_value = 32;</code>
-     */
-    public Builder setTensorDtypeValue(org.tensorflow.proto.framework.DataType value) {
-      if (value == null) {
-        throw new NullPointerException();
-      }
-      kindCase_ = 32;
-      kind_ = value.getNumber();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Represents an enum value for dtype.
-     * </pre>
-     *
-     * <code>.tensorflow.DataType tensor_dtype_value = 32;</code>
-     */
-    public Builder clearTensorDtypeValue() {
-      if (kindCase_ == 32) {
-        kindCase_ = 0;
-        kind_ = null;
-        onChanged();
-      }
-      return this;
-    }
-
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.TensorSpecProto, org.tensorflow.proto.framework.TensorSpecProto.Builder, org.tensorflow.proto.framework.TensorSpecProtoOrBuilder> tensorSpecValueBuilder_;
-    /**
-     * <pre>
-     * Represents a value for tf.TensorSpec.
-     * </pre>
-     *
-     * <code>.tensorflow.TensorSpecProto tensor_spec_value = 33;</code>
-     */
-    public boolean hasTensorSpecValue() {
-      return kindCase_ == 33;
-    }
-    /**
-     * <pre>
-     * Represents a value for tf.TensorSpec.
-     * </pre>
-     *
-     * <code>.tensorflow.TensorSpecProto tensor_spec_value = 33;</code>
-     */
-    public org.tensorflow.proto.framework.TensorSpecProto getTensorSpecValue() {
-      if (tensorSpecValueBuilder_ == null) {
-        if (kindCase_ == 33) {
-          return (org.tensorflow.proto.framework.TensorSpecProto) kind_;
-        }
-        return org.tensorflow.proto.framework.TensorSpecProto.getDefaultInstance();
-      } else {
-        if (kindCase_ == 33) {
-          return tensorSpecValueBuilder_.getMessage();
-        }
-        return org.tensorflow.proto.framework.TensorSpecProto.getDefaultInstance();
-      }
-    }
-    /**
-     * <pre>
-     * Represents a value for tf.TensorSpec.
-     * </pre>
-     *
-     * <code>.tensorflow.TensorSpecProto tensor_spec_value = 33;</code>
-     */
-    public Builder setTensorSpecValue(org.tensorflow.proto.framework.TensorSpecProto value) {
-      if (tensorSpecValueBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        kind_ = value;
-        onChanged();
-      } else {
-        tensorSpecValueBuilder_.setMessage(value);
-      }
-      kindCase_ = 33;
-      return this;
-    }
-    /**
-     * <pre>
-     * Represents a value for tf.TensorSpec.
-     * </pre>
-     *
-     * <code>.tensorflow.TensorSpecProto tensor_spec_value = 33;</code>
-     */
-    public Builder setTensorSpecValue(
-        org.tensorflow.proto.framework.TensorSpecProto.Builder builderForValue) {
-      if (tensorSpecValueBuilder_ == null) {
-        kind_ = builderForValue.build();
-        onChanged();
-      } else {
-        tensorSpecValueBuilder_.setMessage(builderForValue.build());
-      }
-      kindCase_ = 33;
-      return this;
-    }
-    /**
-     * <pre>
-     * Represents a value for tf.TensorSpec.
-     * </pre>
-     *
-     * <code>.tensorflow.TensorSpecProto tensor_spec_value = 33;</code>
-     */
-    public Builder mergeTensorSpecValue(org.tensorflow.proto.framework.TensorSpecProto value) {
-      if (tensorSpecValueBuilder_ == null) {
-        if (kindCase_ == 33 &&
-            kind_ != org.tensorflow.proto.framework.TensorSpecProto.getDefaultInstance()) {
-          kind_ = org.tensorflow.proto.framework.TensorSpecProto.newBuilder((org.tensorflow.proto.framework.TensorSpecProto) kind_)
-              .mergeFrom(value).buildPartial();
-        } else {
-          kind_ = value;
-        }
-        onChanged();
-      } else {
-        if (kindCase_ == 33) {
-          tensorSpecValueBuilder_.mergeFrom(value);
-        }
-        tensorSpecValueBuilder_.setMessage(value);
-      }
-      kindCase_ = 33;
-      return this;
-    }
-    /**
-     * <pre>
-     * Represents a value for tf.TensorSpec.
-     * </pre>
-     *
-     * <code>.tensorflow.TensorSpecProto tensor_spec_value = 33;</code>
-     */
-    public Builder clearTensorSpecValue() {
-      if (tensorSpecValueBuilder_ == null) {
-        if (kindCase_ == 33) {
-          kindCase_ = 0;
-          kind_ = null;
-          onChanged();
-        }
-      } else {
-        if (kindCase_ == 33) {
-          kindCase_ = 0;
-          kind_ = null;
-        }
-        tensorSpecValueBuilder_.clear();
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Represents a value for tf.TensorSpec.
-     * </pre>
-     *
-     * <code>.tensorflow.TensorSpecProto tensor_spec_value = 33;</code>
-     */
-    public org.tensorflow.proto.framework.TensorSpecProto.Builder getTensorSpecValueBuilder() {
-      return getTensorSpecValueFieldBuilder().getBuilder();
-    }
-    /**
-     * <pre>
-     * Represents a value for tf.TensorSpec.
-     * </pre>
-     *
-     * <code>.tensorflow.TensorSpecProto tensor_spec_value = 33;</code>
-     */
-    public org.tensorflow.proto.framework.TensorSpecProtoOrBuilder getTensorSpecValueOrBuilder() {
-      if ((kindCase_ == 33) && (tensorSpecValueBuilder_ != null)) {
-        return tensorSpecValueBuilder_.getMessageOrBuilder();
-      } else {
-        if (kindCase_ == 33) {
-          return (org.tensorflow.proto.framework.TensorSpecProto) kind_;
-        }
-        return org.tensorflow.proto.framework.TensorSpecProto.getDefaultInstance();
-      }
-    }
-    /**
-     * <pre>
-     * Represents a value for tf.TensorSpec.
-     * </pre>
-     *
-     * <code>.tensorflow.TensorSpecProto tensor_spec_value = 33;</code>
-     */
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.TensorSpecProto, org.tensorflow.proto.framework.TensorSpecProto.Builder, org.tensorflow.proto.framework.TensorSpecProtoOrBuilder> 
-        getTensorSpecValueFieldBuilder() {
-      if (tensorSpecValueBuilder_ == null) {
-        if (!(kindCase_ == 33)) {
-          kind_ = org.tensorflow.proto.framework.TensorSpecProto.getDefaultInstance();
-        }
-        tensorSpecValueBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-            org.tensorflow.proto.framework.TensorSpecProto, org.tensorflow.proto.framework.TensorSpecProto.Builder, org.tensorflow.proto.framework.TensorSpecProtoOrBuilder>(
-                (org.tensorflow.proto.framework.TensorSpecProto) kind_,
-                getParentForChildren(),
-                isClean());
-        kind_ = null;
-      }
-      kindCase_ = 33;
-      onChanged();;
-      return tensorSpecValueBuilder_;
-    }
-
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.TypeSpecProto, org.tensorflow.proto.framework.TypeSpecProto.Builder, org.tensorflow.proto.framework.TypeSpecProtoOrBuilder> typeSpecValueBuilder_;
-    /**
-     * <pre>
-     * Represents a value for tf.TypeSpec.
-     * </pre>
-     *
-     * <code>.tensorflow.TypeSpecProto type_spec_value = 34;</code>
-     */
-    public boolean hasTypeSpecValue() {
-      return kindCase_ == 34;
-    }
-    /**
-     * <pre>
-     * Represents a value for tf.TypeSpec.
-     * </pre>
-     *
-     * <code>.tensorflow.TypeSpecProto type_spec_value = 34;</code>
-     */
-    public org.tensorflow.proto.framework.TypeSpecProto getTypeSpecValue() {
-      if (typeSpecValueBuilder_ == null) {
-        if (kindCase_ == 34) {
-          return (org.tensorflow.proto.framework.TypeSpecProto) kind_;
-        }
-        return org.tensorflow.proto.framework.TypeSpecProto.getDefaultInstance();
-      } else {
-        if (kindCase_ == 34) {
-          return typeSpecValueBuilder_.getMessage();
-        }
-        return org.tensorflow.proto.framework.TypeSpecProto.getDefaultInstance();
-      }
-    }
-    /**
-     * <pre>
-     * Represents a value for tf.TypeSpec.
-     * </pre>
-     *
-     * <code>.tensorflow.TypeSpecProto type_spec_value = 34;</code>
-     */
-    public Builder setTypeSpecValue(org.tensorflow.proto.framework.TypeSpecProto value) {
-      if (typeSpecValueBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        kind_ = value;
-        onChanged();
-      } else {
-        typeSpecValueBuilder_.setMessage(value);
-      }
-      kindCase_ = 34;
-      return this;
-    }
-    /**
-     * <pre>
-     * Represents a value for tf.TypeSpec.
-     * </pre>
-     *
-     * <code>.tensorflow.TypeSpecProto type_spec_value = 34;</code>
-     */
-    public Builder setTypeSpecValue(
-        org.tensorflow.proto.framework.TypeSpecProto.Builder builderForValue) {
-      if (typeSpecValueBuilder_ == null) {
-        kind_ = builderForValue.build();
-        onChanged();
-      } else {
-        typeSpecValueBuilder_.setMessage(builderForValue.build());
-      }
-      kindCase_ = 34;
-      return this;
-    }
-    /**
-     * <pre>
-     * Represents a value for tf.TypeSpec.
-     * </pre>
-     *
-     * <code>.tensorflow.TypeSpecProto type_spec_value = 34;</code>
-     */
-    public Builder mergeTypeSpecValue(org.tensorflow.proto.framework.TypeSpecProto value) {
-      if (typeSpecValueBuilder_ == null) {
-        if (kindCase_ == 34 &&
-            kind_ != org.tensorflow.proto.framework.TypeSpecProto.getDefaultInstance()) {
-          kind_ = org.tensorflow.proto.framework.TypeSpecProto.newBuilder((org.tensorflow.proto.framework.TypeSpecProto) kind_)
-              .mergeFrom(value).buildPartial();
-        } else {
-          kind_ = value;
-        }
-        onChanged();
-      } else {
-        if (kindCase_ == 34) {
-          typeSpecValueBuilder_.mergeFrom(value);
-        }
-        typeSpecValueBuilder_.setMessage(value);
-      }
-      kindCase_ = 34;
-      return this;
-    }
-    /**
-     * <pre>
-     * Represents a value for tf.TypeSpec.
-     * </pre>
-     *
-     * <code>.tensorflow.TypeSpecProto type_spec_value = 34;</code>
-     */
-    public Builder clearTypeSpecValue() {
-      if (typeSpecValueBuilder_ == null) {
-        if (kindCase_ == 34) {
-          kindCase_ = 0;
-          kind_ = null;
-          onChanged();
-        }
-      } else {
-        if (kindCase_ == 34) {
-          kindCase_ = 0;
-          kind_ = null;
-        }
-        typeSpecValueBuilder_.clear();
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Represents a value for tf.TypeSpec.
-     * </pre>
-     *
-     * <code>.tensorflow.TypeSpecProto type_spec_value = 34;</code>
-     */
-    public org.tensorflow.proto.framework.TypeSpecProto.Builder getTypeSpecValueBuilder() {
-      return getTypeSpecValueFieldBuilder().getBuilder();
-    }
-    /**
-     * <pre>
-     * Represents a value for tf.TypeSpec.
-     * </pre>
-     *
-     * <code>.tensorflow.TypeSpecProto type_spec_value = 34;</code>
-     */
-    public org.tensorflow.proto.framework.TypeSpecProtoOrBuilder getTypeSpecValueOrBuilder() {
-      if ((kindCase_ == 34) && (typeSpecValueBuilder_ != null)) {
-        return typeSpecValueBuilder_.getMessageOrBuilder();
-      } else {
-        if (kindCase_ == 34) {
-          return (org.tensorflow.proto.framework.TypeSpecProto) kind_;
-        }
-        return org.tensorflow.proto.framework.TypeSpecProto.getDefaultInstance();
-      }
-    }
-    /**
-     * <pre>
-     * Represents a value for tf.TypeSpec.
-     * </pre>
-     *
-     * <code>.tensorflow.TypeSpecProto type_spec_value = 34;</code>
-     */
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.TypeSpecProto, org.tensorflow.proto.framework.TypeSpecProto.Builder, org.tensorflow.proto.framework.TypeSpecProtoOrBuilder> 
-        getTypeSpecValueFieldBuilder() {
-      if (typeSpecValueBuilder_ == null) {
-        if (!(kindCase_ == 34)) {
-          kind_ = org.tensorflow.proto.framework.TypeSpecProto.getDefaultInstance();
-        }
-        typeSpecValueBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-            org.tensorflow.proto.framework.TypeSpecProto, org.tensorflow.proto.framework.TypeSpecProto.Builder, org.tensorflow.proto.framework.TypeSpecProtoOrBuilder>(
-                (org.tensorflow.proto.framework.TypeSpecProto) kind_,
-                getParentForChildren(),
-                isClean());
-        kind_ = null;
-      }
-      kindCase_ = 34;
-      onChanged();;
-      return typeSpecValueBuilder_;
-    }
-
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.BoundedTensorSpecProto, org.tensorflow.proto.framework.BoundedTensorSpecProto.Builder, org.tensorflow.proto.framework.BoundedTensorSpecProtoOrBuilder> boundedTensorSpecValueBuilder_;
-    /**
-     * <pre>
-     * Represents a value for tf.BoundedTensorSpec.
-     * </pre>
-     *
-     * <code>.tensorflow.BoundedTensorSpecProto bounded_tensor_spec_value = 35;</code>
-     */
-    public boolean hasBoundedTensorSpecValue() {
-      return kindCase_ == 35;
-    }
-    /**
-     * <pre>
-     * Represents a value for tf.BoundedTensorSpec.
-     * </pre>
-     *
-     * <code>.tensorflow.BoundedTensorSpecProto bounded_tensor_spec_value = 35;</code>
-     */
-    public org.tensorflow.proto.framework.BoundedTensorSpecProto getBoundedTensorSpecValue() {
-      if (boundedTensorSpecValueBuilder_ == null) {
-        if (kindCase_ == 35) {
-          return (org.tensorflow.proto.framework.BoundedTensorSpecProto) kind_;
-        }
-        return org.tensorflow.proto.framework.BoundedTensorSpecProto.getDefaultInstance();
-      } else {
-        if (kindCase_ == 35) {
-          return boundedTensorSpecValueBuilder_.getMessage();
-        }
-        return org.tensorflow.proto.framework.BoundedTensorSpecProto.getDefaultInstance();
-      }
-    }
-    /**
-     * <pre>
-     * Represents a value for tf.BoundedTensorSpec.
-     * </pre>
-     *
-     * <code>.tensorflow.BoundedTensorSpecProto bounded_tensor_spec_value = 35;</code>
-     */
-    public Builder setBoundedTensorSpecValue(org.tensorflow.proto.framework.BoundedTensorSpecProto value) {
-      if (boundedTensorSpecValueBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        kind_ = value;
-        onChanged();
-      } else {
-        boundedTensorSpecValueBuilder_.setMessage(value);
-      }
-      kindCase_ = 35;
-      return this;
-    }
-    /**
-     * <pre>
-     * Represents a value for tf.BoundedTensorSpec.
-     * </pre>
-     *
-     * <code>.tensorflow.BoundedTensorSpecProto bounded_tensor_spec_value = 35;</code>
-     */
-    public Builder setBoundedTensorSpecValue(
-        org.tensorflow.proto.framework.BoundedTensorSpecProto.Builder builderForValue) {
-      if (boundedTensorSpecValueBuilder_ == null) {
-        kind_ = builderForValue.build();
-        onChanged();
-      } else {
-        boundedTensorSpecValueBuilder_.setMessage(builderForValue.build());
-      }
-      kindCase_ = 35;
-      return this;
-    }
-    /**
-     * <pre>
-     * Represents a value for tf.BoundedTensorSpec.
-     * </pre>
-     *
-     * <code>.tensorflow.BoundedTensorSpecProto bounded_tensor_spec_value = 35;</code>
-     */
-    public Builder mergeBoundedTensorSpecValue(org.tensorflow.proto.framework.BoundedTensorSpecProto value) {
-      if (boundedTensorSpecValueBuilder_ == null) {
-        if (kindCase_ == 35 &&
-            kind_ != org.tensorflow.proto.framework.BoundedTensorSpecProto.getDefaultInstance()) {
-          kind_ = org.tensorflow.proto.framework.BoundedTensorSpecProto.newBuilder((org.tensorflow.proto.framework.BoundedTensorSpecProto) kind_)
-              .mergeFrom(value).buildPartial();
-        } else {
-          kind_ = value;
-        }
-        onChanged();
-      } else {
-        if (kindCase_ == 35) {
-          boundedTensorSpecValueBuilder_.mergeFrom(value);
-        }
-        boundedTensorSpecValueBuilder_.setMessage(value);
-      }
-      kindCase_ = 35;
-      return this;
-    }
-    /**
-     * <pre>
-     * Represents a value for tf.BoundedTensorSpec.
-     * </pre>
-     *
-     * <code>.tensorflow.BoundedTensorSpecProto bounded_tensor_spec_value = 35;</code>
-     */
-    public Builder clearBoundedTensorSpecValue() {
-      if (boundedTensorSpecValueBuilder_ == null) {
-        if (kindCase_ == 35) {
-          kindCase_ = 0;
-          kind_ = null;
-          onChanged();
-        }
-      } else {
-        if (kindCase_ == 35) {
-          kindCase_ = 0;
-          kind_ = null;
-        }
-        boundedTensorSpecValueBuilder_.clear();
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Represents a value for tf.BoundedTensorSpec.
-     * </pre>
-     *
-     * <code>.tensorflow.BoundedTensorSpecProto bounded_tensor_spec_value = 35;</code>
-     */
-    public org.tensorflow.proto.framework.BoundedTensorSpecProto.Builder getBoundedTensorSpecValueBuilder() {
-      return getBoundedTensorSpecValueFieldBuilder().getBuilder();
-    }
-    /**
-     * <pre>
-     * Represents a value for tf.BoundedTensorSpec.
-     * </pre>
-     *
-     * <code>.tensorflow.BoundedTensorSpecProto bounded_tensor_spec_value = 35;</code>
-     */
-    public org.tensorflow.proto.framework.BoundedTensorSpecProtoOrBuilder getBoundedTensorSpecValueOrBuilder() {
-      if ((kindCase_ == 35) && (boundedTensorSpecValueBuilder_ != null)) {
-        return boundedTensorSpecValueBuilder_.getMessageOrBuilder();
-      } else {
-        if (kindCase_ == 35) {
-          return (org.tensorflow.proto.framework.BoundedTensorSpecProto) kind_;
-        }
-        return org.tensorflow.proto.framework.BoundedTensorSpecProto.getDefaultInstance();
-      }
-    }
-    /**
-     * <pre>
-     * Represents a value for tf.BoundedTensorSpec.
-     * </pre>
-     *
-     * <code>.tensorflow.BoundedTensorSpecProto bounded_tensor_spec_value = 35;</code>
-     */
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.BoundedTensorSpecProto, org.tensorflow.proto.framework.BoundedTensorSpecProto.Builder, org.tensorflow.proto.framework.BoundedTensorSpecProtoOrBuilder> 
-        getBoundedTensorSpecValueFieldBuilder() {
-      if (boundedTensorSpecValueBuilder_ == null) {
-        if (!(kindCase_ == 35)) {
-          kind_ = org.tensorflow.proto.framework.BoundedTensorSpecProto.getDefaultInstance();
-        }
-        boundedTensorSpecValueBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-            org.tensorflow.proto.framework.BoundedTensorSpecProto, org.tensorflow.proto.framework.BoundedTensorSpecProto.Builder, org.tensorflow.proto.framework.BoundedTensorSpecProtoOrBuilder>(
-                (org.tensorflow.proto.framework.BoundedTensorSpecProto) kind_,
-                getParentForChildren(),
-                isClean());
-        kind_ = null;
-      }
-      kindCase_ = 35;
-      onChanged();;
-      return boundedTensorSpecValueBuilder_;
-    }
-
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.ListValue, org.tensorflow.proto.framework.ListValue.Builder, org.tensorflow.proto.framework.ListValueOrBuilder> listValueBuilder_;
-    /**
-     * <pre>
-     * Represents a list of `Value`.
-     * </pre>
-     *
-     * <code>.tensorflow.ListValue list_value = 51;</code>
-     */
-    public boolean hasListValue() {
-      return kindCase_ == 51;
-    }
-    /**
-     * <pre>
-     * Represents a list of `Value`.
-     * </pre>
-     *
-     * <code>.tensorflow.ListValue list_value = 51;</code>
-     */
-    public org.tensorflow.proto.framework.ListValue getListValue() {
-      if (listValueBuilder_ == null) {
-        if (kindCase_ == 51) {
-          return (org.tensorflow.proto.framework.ListValue) kind_;
-        }
-        return org.tensorflow.proto.framework.ListValue.getDefaultInstance();
-      } else {
-        if (kindCase_ == 51) {
-          return listValueBuilder_.getMessage();
-        }
-        return org.tensorflow.proto.framework.ListValue.getDefaultInstance();
-      }
-    }
-    /**
-     * <pre>
-     * Represents a list of `Value`.
-     * </pre>
-     *
-     * <code>.tensorflow.ListValue list_value = 51;</code>
-     */
-    public Builder setListValue(org.tensorflow.proto.framework.ListValue value) {
-      if (listValueBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        kind_ = value;
-        onChanged();
-      } else {
-        listValueBuilder_.setMessage(value);
-      }
-      kindCase_ = 51;
-      return this;
-    }
-    /**
-     * <pre>
-     * Represents a list of `Value`.
-     * </pre>
-     *
-     * <code>.tensorflow.ListValue list_value = 51;</code>
-     */
-    public Builder setListValue(
-        org.tensorflow.proto.framework.ListValue.Builder builderForValue) {
-      if (listValueBuilder_ == null) {
-        kind_ = builderForValue.build();
-        onChanged();
-      } else {
-        listValueBuilder_.setMessage(builderForValue.build());
-      }
-      kindCase_ = 51;
-      return this;
-    }
-    /**
-     * <pre>
-     * Represents a list of `Value`.
-     * </pre>
-     *
-     * <code>.tensorflow.ListValue list_value = 51;</code>
-     */
-    public Builder mergeListValue(org.tensorflow.proto.framework.ListValue value) {
-      if (listValueBuilder_ == null) {
-        if (kindCase_ == 51 &&
-            kind_ != org.tensorflow.proto.framework.ListValue.getDefaultInstance()) {
-          kind_ = org.tensorflow.proto.framework.ListValue.newBuilder((org.tensorflow.proto.framework.ListValue) kind_)
-              .mergeFrom(value).buildPartial();
-        } else {
-          kind_ = value;
-        }
-        onChanged();
-      } else {
-        if (kindCase_ == 51) {
-          listValueBuilder_.mergeFrom(value);
-        }
-        listValueBuilder_.setMessage(value);
-      }
-      kindCase_ = 51;
-      return this;
-    }
-    /**
-     * <pre>
-     * Represents a list of `Value`.
-     * </pre>
-     *
-     * <code>.tensorflow.ListValue list_value = 51;</code>
-     */
-    public Builder clearListValue() {
-      if (listValueBuilder_ == null) {
-        if (kindCase_ == 51) {
-          kindCase_ = 0;
-          kind_ = null;
-          onChanged();
-        }
-      } else {
-        if (kindCase_ == 51) {
-          kindCase_ = 0;
-          kind_ = null;
-        }
-        listValueBuilder_.clear();
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Represents a list of `Value`.
-     * </pre>
-     *
-     * <code>.tensorflow.ListValue list_value = 51;</code>
-     */
-    public org.tensorflow.proto.framework.ListValue.Builder getListValueBuilder() {
-      return getListValueFieldBuilder().getBuilder();
-    }
-    /**
-     * <pre>
-     * Represents a list of `Value`.
-     * </pre>
-     *
-     * <code>.tensorflow.ListValue list_value = 51;</code>
-     */
-    public org.tensorflow.proto.framework.ListValueOrBuilder getListValueOrBuilder() {
-      if ((kindCase_ == 51) && (listValueBuilder_ != null)) {
-        return listValueBuilder_.getMessageOrBuilder();
-      } else {
-        if (kindCase_ == 51) {
-          return (org.tensorflow.proto.framework.ListValue) kind_;
-        }
-        return org.tensorflow.proto.framework.ListValue.getDefaultInstance();
-      }
-    }
-    /**
-     * <pre>
-     * Represents a list of `Value`.
-     * </pre>
-     *
-     * <code>.tensorflow.ListValue list_value = 51;</code>
-     */
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.ListValue, org.tensorflow.proto.framework.ListValue.Builder, org.tensorflow.proto.framework.ListValueOrBuilder> 
-        getListValueFieldBuilder() {
-      if (listValueBuilder_ == null) {
-        if (!(kindCase_ == 51)) {
-          kind_ = org.tensorflow.proto.framework.ListValue.getDefaultInstance();
-        }
-        listValueBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-            org.tensorflow.proto.framework.ListValue, org.tensorflow.proto.framework.ListValue.Builder, org.tensorflow.proto.framework.ListValueOrBuilder>(
-                (org.tensorflow.proto.framework.ListValue) kind_,
-                getParentForChildren(),
-                isClean());
-        kind_ = null;
-      }
-      kindCase_ = 51;
-      onChanged();;
-      return listValueBuilder_;
-    }
-
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.TupleValue, org.tensorflow.proto.framework.TupleValue.Builder, org.tensorflow.proto.framework.TupleValueOrBuilder> tupleValueBuilder_;
-    /**
-     * <pre>
-     * Represents a tuple of `Value`.
-     * </pre>
-     *
-     * <code>.tensorflow.TupleValue tuple_value = 52;</code>
-     */
-    public boolean hasTupleValue() {
-      return kindCase_ == 52;
-    }
-    /**
-     * <pre>
-     * Represents a tuple of `Value`.
-     * </pre>
-     *
-     * <code>.tensorflow.TupleValue tuple_value = 52;</code>
-     */
-    public org.tensorflow.proto.framework.TupleValue getTupleValue() {
-      if (tupleValueBuilder_ == null) {
-        if (kindCase_ == 52) {
-          return (org.tensorflow.proto.framework.TupleValue) kind_;
-        }
-        return org.tensorflow.proto.framework.TupleValue.getDefaultInstance();
-      } else {
-        if (kindCase_ == 52) {
-          return tupleValueBuilder_.getMessage();
-        }
-        return org.tensorflow.proto.framework.TupleValue.getDefaultInstance();
-      }
-    }
-    /**
-     * <pre>
-     * Represents a tuple of `Value`.
-     * </pre>
-     *
-     * <code>.tensorflow.TupleValue tuple_value = 52;</code>
-     */
-    public Builder setTupleValue(org.tensorflow.proto.framework.TupleValue value) {
-      if (tupleValueBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        kind_ = value;
-        onChanged();
-      } else {
-        tupleValueBuilder_.setMessage(value);
-      }
-      kindCase_ = 52;
-      return this;
-    }
-    /**
-     * <pre>
-     * Represents a tuple of `Value`.
-     * </pre>
-     *
-     * <code>.tensorflow.TupleValue tuple_value = 52;</code>
-     */
-    public Builder setTupleValue(
-        org.tensorflow.proto.framework.TupleValue.Builder builderForValue) {
-      if (tupleValueBuilder_ == null) {
-        kind_ = builderForValue.build();
-        onChanged();
-      } else {
-        tupleValueBuilder_.setMessage(builderForValue.build());
-      }
-      kindCase_ = 52;
-      return this;
-    }
-    /**
-     * <pre>
-     * Represents a tuple of `Value`.
-     * </pre>
-     *
-     * <code>.tensorflow.TupleValue tuple_value = 52;</code>
-     */
-    public Builder mergeTupleValue(org.tensorflow.proto.framework.TupleValue value) {
-      if (tupleValueBuilder_ == null) {
-        if (kindCase_ == 52 &&
-            kind_ != org.tensorflow.proto.framework.TupleValue.getDefaultInstance()) {
-          kind_ = org.tensorflow.proto.framework.TupleValue.newBuilder((org.tensorflow.proto.framework.TupleValue) kind_)
-              .mergeFrom(value).buildPartial();
-        } else {
-          kind_ = value;
-        }
-        onChanged();
-      } else {
-        if (kindCase_ == 52) {
-          tupleValueBuilder_.mergeFrom(value);
-        }
-        tupleValueBuilder_.setMessage(value);
-      }
-      kindCase_ = 52;
-      return this;
-    }
-    /**
-     * <pre>
-     * Represents a tuple of `Value`.
-     * </pre>
-     *
-     * <code>.tensorflow.TupleValue tuple_value = 52;</code>
-     */
-    public Builder clearTupleValue() {
-      if (tupleValueBuilder_ == null) {
-        if (kindCase_ == 52) {
-          kindCase_ = 0;
-          kind_ = null;
-          onChanged();
-        }
-      } else {
-        if (kindCase_ == 52) {
-          kindCase_ = 0;
-          kind_ = null;
-        }
-        tupleValueBuilder_.clear();
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Represents a tuple of `Value`.
-     * </pre>
-     *
-     * <code>.tensorflow.TupleValue tuple_value = 52;</code>
-     */
-    public org.tensorflow.proto.framework.TupleValue.Builder getTupleValueBuilder() {
-      return getTupleValueFieldBuilder().getBuilder();
-    }
-    /**
-     * <pre>
-     * Represents a tuple of `Value`.
-     * </pre>
-     *
-     * <code>.tensorflow.TupleValue tuple_value = 52;</code>
-     */
-    public org.tensorflow.proto.framework.TupleValueOrBuilder getTupleValueOrBuilder() {
-      if ((kindCase_ == 52) && (tupleValueBuilder_ != null)) {
-        return tupleValueBuilder_.getMessageOrBuilder();
-      } else {
-        if (kindCase_ == 52) {
-          return (org.tensorflow.proto.framework.TupleValue) kind_;
-        }
-        return org.tensorflow.proto.framework.TupleValue.getDefaultInstance();
-      }
-    }
-    /**
-     * <pre>
-     * Represents a tuple of `Value`.
-     * </pre>
-     *
-     * <code>.tensorflow.TupleValue tuple_value = 52;</code>
-     */
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.TupleValue, org.tensorflow.proto.framework.TupleValue.Builder, org.tensorflow.proto.framework.TupleValueOrBuilder> 
-        getTupleValueFieldBuilder() {
-      if (tupleValueBuilder_ == null) {
-        if (!(kindCase_ == 52)) {
-          kind_ = org.tensorflow.proto.framework.TupleValue.getDefaultInstance();
-        }
-        tupleValueBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-            org.tensorflow.proto.framework.TupleValue, org.tensorflow.proto.framework.TupleValue.Builder, org.tensorflow.proto.framework.TupleValueOrBuilder>(
-                (org.tensorflow.proto.framework.TupleValue) kind_,
-                getParentForChildren(),
-                isClean());
-        kind_ = null;
-      }
-      kindCase_ = 52;
-      onChanged();;
-      return tupleValueBuilder_;
-    }
-
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.DictValue, org.tensorflow.proto.framework.DictValue.Builder, org.tensorflow.proto.framework.DictValueOrBuilder> dictValueBuilder_;
-    /**
-     * <pre>
-     * Represents a dict `Value`.
-     * </pre>
-     *
-     * <code>.tensorflow.DictValue dict_value = 53;</code>
-     */
-    public boolean hasDictValue() {
-      return kindCase_ == 53;
-    }
-    /**
-     * <pre>
-     * Represents a dict `Value`.
-     * </pre>
-     *
-     * <code>.tensorflow.DictValue dict_value = 53;</code>
-     */
-    public org.tensorflow.proto.framework.DictValue getDictValue() {
-      if (dictValueBuilder_ == null) {
-        if (kindCase_ == 53) {
-          return (org.tensorflow.proto.framework.DictValue) kind_;
-        }
-        return org.tensorflow.proto.framework.DictValue.getDefaultInstance();
-      } else {
-        if (kindCase_ == 53) {
-          return dictValueBuilder_.getMessage();
-        }
-        return org.tensorflow.proto.framework.DictValue.getDefaultInstance();
-      }
-    }
-    /**
-     * <pre>
-     * Represents a dict `Value`.
-     * </pre>
-     *
-     * <code>.tensorflow.DictValue dict_value = 53;</code>
-     */
-    public Builder setDictValue(org.tensorflow.proto.framework.DictValue value) {
-      if (dictValueBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        kind_ = value;
-        onChanged();
-      } else {
-        dictValueBuilder_.setMessage(value);
-      }
-      kindCase_ = 53;
-      return this;
-    }
-    /**
-     * <pre>
-     * Represents a dict `Value`.
-     * </pre>
-     *
-     * <code>.tensorflow.DictValue dict_value = 53;</code>
-     */
-    public Builder setDictValue(
-        org.tensorflow.proto.framework.DictValue.Builder builderForValue) {
-      if (dictValueBuilder_ == null) {
-        kind_ = builderForValue.build();
-        onChanged();
-      } else {
-        dictValueBuilder_.setMessage(builderForValue.build());
-      }
-      kindCase_ = 53;
-      return this;
-    }
-    /**
-     * <pre>
-     * Represents a dict `Value`.
-     * </pre>
-     *
-     * <code>.tensorflow.DictValue dict_value = 53;</code>
-     */
-    public Builder mergeDictValue(org.tensorflow.proto.framework.DictValue value) {
-      if (dictValueBuilder_ == null) {
-        if (kindCase_ == 53 &&
-            kind_ != org.tensorflow.proto.framework.DictValue.getDefaultInstance()) {
-          kind_ = org.tensorflow.proto.framework.DictValue.newBuilder((org.tensorflow.proto.framework.DictValue) kind_)
-              .mergeFrom(value).buildPartial();
-        } else {
-          kind_ = value;
-        }
-        onChanged();
-      } else {
-        if (kindCase_ == 53) {
-          dictValueBuilder_.mergeFrom(value);
-        }
-        dictValueBuilder_.setMessage(value);
-      }
-      kindCase_ = 53;
-      return this;
-    }
-    /**
-     * <pre>
-     * Represents a dict `Value`.
-     * </pre>
-     *
-     * <code>.tensorflow.DictValue dict_value = 53;</code>
-     */
-    public Builder clearDictValue() {
-      if (dictValueBuilder_ == null) {
-        if (kindCase_ == 53) {
-          kindCase_ = 0;
-          kind_ = null;
-          onChanged();
-        }
-      } else {
-        if (kindCase_ == 53) {
-          kindCase_ = 0;
-          kind_ = null;
-        }
-        dictValueBuilder_.clear();
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Represents a dict `Value`.
-     * </pre>
-     *
-     * <code>.tensorflow.DictValue dict_value = 53;</code>
-     */
-    public org.tensorflow.proto.framework.DictValue.Builder getDictValueBuilder() {
-      return getDictValueFieldBuilder().getBuilder();
-    }
-    /**
-     * <pre>
-     * Represents a dict `Value`.
-     * </pre>
-     *
-     * <code>.tensorflow.DictValue dict_value = 53;</code>
-     */
-    public org.tensorflow.proto.framework.DictValueOrBuilder getDictValueOrBuilder() {
-      if ((kindCase_ == 53) && (dictValueBuilder_ != null)) {
-        return dictValueBuilder_.getMessageOrBuilder();
-      } else {
-        if (kindCase_ == 53) {
-          return (org.tensorflow.proto.framework.DictValue) kind_;
-        }
-        return org.tensorflow.proto.framework.DictValue.getDefaultInstance();
-      }
-    }
-    /**
-     * <pre>
-     * Represents a dict `Value`.
-     * </pre>
-     *
-     * <code>.tensorflow.DictValue dict_value = 53;</code>
-     */
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.DictValue, org.tensorflow.proto.framework.DictValue.Builder, org.tensorflow.proto.framework.DictValueOrBuilder> 
-        getDictValueFieldBuilder() {
-      if (dictValueBuilder_ == null) {
-        if (!(kindCase_ == 53)) {
-          kind_ = org.tensorflow.proto.framework.DictValue.getDefaultInstance();
-        }
-        dictValueBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-            org.tensorflow.proto.framework.DictValue, org.tensorflow.proto.framework.DictValue.Builder, org.tensorflow.proto.framework.DictValueOrBuilder>(
-                (org.tensorflow.proto.framework.DictValue) kind_,
-                getParentForChildren(),
-                isClean());
-        kind_ = null;
-      }
-      kindCase_ = 53;
-      onChanged();;
-      return dictValueBuilder_;
-    }
-
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.NamedTupleValue, org.tensorflow.proto.framework.NamedTupleValue.Builder, org.tensorflow.proto.framework.NamedTupleValueOrBuilder> namedTupleValueBuilder_;
-    /**
-     * <pre>
-     * Represents Python's namedtuple.
-     * </pre>
-     *
-     * <code>.tensorflow.NamedTupleValue named_tuple_value = 54;</code>
-     */
-    public boolean hasNamedTupleValue() {
-      return kindCase_ == 54;
-    }
-    /**
-     * <pre>
-     * Represents Python's namedtuple.
-     * </pre>
-     *
-     * <code>.tensorflow.NamedTupleValue named_tuple_value = 54;</code>
-     */
-    public org.tensorflow.proto.framework.NamedTupleValue getNamedTupleValue() {
-      if (namedTupleValueBuilder_ == null) {
-        if (kindCase_ == 54) {
-          return (org.tensorflow.proto.framework.NamedTupleValue) kind_;
-        }
-        return org.tensorflow.proto.framework.NamedTupleValue.getDefaultInstance();
-      } else {
-        if (kindCase_ == 54) {
-          return namedTupleValueBuilder_.getMessage();
-        }
-        return org.tensorflow.proto.framework.NamedTupleValue.getDefaultInstance();
-      }
-    }
-    /**
-     * <pre>
-     * Represents Python's namedtuple.
-     * </pre>
-     *
-     * <code>.tensorflow.NamedTupleValue named_tuple_value = 54;</code>
-     */
-    public Builder setNamedTupleValue(org.tensorflow.proto.framework.NamedTupleValue value) {
-      if (namedTupleValueBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        kind_ = value;
-        onChanged();
-      } else {
-        namedTupleValueBuilder_.setMessage(value);
-      }
-      kindCase_ = 54;
-      return this;
-    }
-    /**
-     * <pre>
-     * Represents Python's namedtuple.
-     * </pre>
-     *
-     * <code>.tensorflow.NamedTupleValue named_tuple_value = 54;</code>
-     */
-    public Builder setNamedTupleValue(
-        org.tensorflow.proto.framework.NamedTupleValue.Builder builderForValue) {
-      if (namedTupleValueBuilder_ == null) {
-        kind_ = builderForValue.build();
-        onChanged();
-      } else {
-        namedTupleValueBuilder_.setMessage(builderForValue.build());
-      }
-      kindCase_ = 54;
-      return this;
-    }
-    /**
-     * <pre>
-     * Represents Python's namedtuple.
-     * </pre>
-     *
-     * <code>.tensorflow.NamedTupleValue named_tuple_value = 54;</code>
-     */
-    public Builder mergeNamedTupleValue(org.tensorflow.proto.framework.NamedTupleValue value) {
-      if (namedTupleValueBuilder_ == null) {
-        if (kindCase_ == 54 &&
-            kind_ != org.tensorflow.proto.framework.NamedTupleValue.getDefaultInstance()) {
-          kind_ = org.tensorflow.proto.framework.NamedTupleValue.newBuilder((org.tensorflow.proto.framework.NamedTupleValue) kind_)
-              .mergeFrom(value).buildPartial();
-        } else {
-          kind_ = value;
-        }
-        onChanged();
-      } else {
-        if (kindCase_ == 54) {
-          namedTupleValueBuilder_.mergeFrom(value);
-        }
-        namedTupleValueBuilder_.setMessage(value);
-      }
-      kindCase_ = 54;
-      return this;
-    }
-    /**
-     * <pre>
-     * Represents Python's namedtuple.
-     * </pre>
-     *
-     * <code>.tensorflow.NamedTupleValue named_tuple_value = 54;</code>
-     */
-    public Builder clearNamedTupleValue() {
-      if (namedTupleValueBuilder_ == null) {
-        if (kindCase_ == 54) {
-          kindCase_ = 0;
-          kind_ = null;
-          onChanged();
-        }
-      } else {
-        if (kindCase_ == 54) {
-          kindCase_ = 0;
-          kind_ = null;
-        }
-        namedTupleValueBuilder_.clear();
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Represents Python's namedtuple.
-     * </pre>
-     *
-     * <code>.tensorflow.NamedTupleValue named_tuple_value = 54;</code>
-     */
-    public org.tensorflow.proto.framework.NamedTupleValue.Builder getNamedTupleValueBuilder() {
-      return getNamedTupleValueFieldBuilder().getBuilder();
-    }
-    /**
-     * <pre>
-     * Represents Python's namedtuple.
-     * </pre>
-     *
-     * <code>.tensorflow.NamedTupleValue named_tuple_value = 54;</code>
-     */
-    public org.tensorflow.proto.framework.NamedTupleValueOrBuilder getNamedTupleValueOrBuilder() {
-      if ((kindCase_ == 54) && (namedTupleValueBuilder_ != null)) {
-        return namedTupleValueBuilder_.getMessageOrBuilder();
-      } else {
-        if (kindCase_ == 54) {
-          return (org.tensorflow.proto.framework.NamedTupleValue) kind_;
-        }
-        return org.tensorflow.proto.framework.NamedTupleValue.getDefaultInstance();
-      }
-    }
-    /**
-     * <pre>
-     * Represents Python's namedtuple.
-     * </pre>
-     *
-     * <code>.tensorflow.NamedTupleValue named_tuple_value = 54;</code>
-     */
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.NamedTupleValue, org.tensorflow.proto.framework.NamedTupleValue.Builder, org.tensorflow.proto.framework.NamedTupleValueOrBuilder> 
-        getNamedTupleValueFieldBuilder() {
-      if (namedTupleValueBuilder_ == null) {
-        if (!(kindCase_ == 54)) {
-          kind_ = org.tensorflow.proto.framework.NamedTupleValue.getDefaultInstance();
-        }
-        namedTupleValueBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-            org.tensorflow.proto.framework.NamedTupleValue, org.tensorflow.proto.framework.NamedTupleValue.Builder, org.tensorflow.proto.framework.NamedTupleValueOrBuilder>(
-                (org.tensorflow.proto.framework.NamedTupleValue) kind_,
-                getParentForChildren(),
-                isClean());
-        kind_ = null;
-      }
-      kindCase_ = 54;
-      onChanged();;
-      return namedTupleValueBuilder_;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.StructuredValue)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.StructuredValue)
-  private static final org.tensorflow.proto.framework.StructuredValue DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.StructuredValue();
-  }
-
-  public static org.tensorflow.proto.framework.StructuredValue getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<StructuredValue>
-      PARSER = new com.google.protobuf.AbstractParser<StructuredValue>() {
-    @java.lang.Override
-    public StructuredValue parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new StructuredValue(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<StructuredValue> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<StructuredValue> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.framework.StructuredValue getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/StructuredValueOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/StructuredValueOrBuilder.java
deleted file mode 100644
index 3ffb498eb22..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/StructuredValueOrBuilder.java
+++ /dev/null
@@ -1,309 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/struct.proto
-
-package org.tensorflow.proto.framework;
-
-public interface StructuredValueOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.StructuredValue)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <pre>
-   * Represents None.
-   * </pre>
-   *
-   * <code>.tensorflow.NoneValue none_value = 1;</code>
-   */
-  boolean hasNoneValue();
-  /**
-   * <pre>
-   * Represents None.
-   * </pre>
-   *
-   * <code>.tensorflow.NoneValue none_value = 1;</code>
-   */
-  org.tensorflow.proto.framework.NoneValue getNoneValue();
-  /**
-   * <pre>
-   * Represents None.
-   * </pre>
-   *
-   * <code>.tensorflow.NoneValue none_value = 1;</code>
-   */
-  org.tensorflow.proto.framework.NoneValueOrBuilder getNoneValueOrBuilder();
-
-  /**
-   * <pre>
-   * Represents a double-precision floating-point value (a Python `float`).
-   * </pre>
-   *
-   * <code>double float64_value = 11;</code>
-   */
-  double getFloat64Value();
-
-  /**
-   * <pre>
-   * Represents a signed integer value, limited to 64 bits.
-   * Larger values from Python's arbitrary-precision integers are unsupported.
-   * </pre>
-   *
-   * <code>sint64 int64_value = 12;</code>
-   */
-  long getInt64Value();
-
-  /**
-   * <pre>
-   * Represents a string of Unicode characters stored in a Python `str`.
-   * In Python 3, this is exactly what type `str` is.
-   * In Python 2, this is the UTF-8 encoding of the characters.
-   * For strings with ASCII characters only (as often used in TensorFlow code)
-   * there is effectively no difference between the language versions.
-   * The obsolescent `unicode` type of Python 2 is not supported here.
-   * </pre>
-   *
-   * <code>string string_value = 13;</code>
-   */
-  java.lang.String getStringValue();
-  /**
-   * <pre>
-   * Represents a string of Unicode characters stored in a Python `str`.
-   * In Python 3, this is exactly what type `str` is.
-   * In Python 2, this is the UTF-8 encoding of the characters.
-   * For strings with ASCII characters only (as often used in TensorFlow code)
-   * there is effectively no difference between the language versions.
-   * The obsolescent `unicode` type of Python 2 is not supported here.
-   * </pre>
-   *
-   * <code>string string_value = 13;</code>
-   */
-  com.google.protobuf.ByteString
-      getStringValueBytes();
-
-  /**
-   * <pre>
-   * Represents a boolean value.
-   * </pre>
-   *
-   * <code>bool bool_value = 14;</code>
-   */
-  boolean getBoolValue();
-
-  /**
-   * <pre>
-   * Represents a TensorShape.
-   * </pre>
-   *
-   * <code>.tensorflow.TensorShapeProto tensor_shape_value = 31;</code>
-   */
-  boolean hasTensorShapeValue();
-  /**
-   * <pre>
-   * Represents a TensorShape.
-   * </pre>
-   *
-   * <code>.tensorflow.TensorShapeProto tensor_shape_value = 31;</code>
-   */
-  org.tensorflow.proto.framework.TensorShapeProto getTensorShapeValue();
-  /**
-   * <pre>
-   * Represents a TensorShape.
-   * </pre>
-   *
-   * <code>.tensorflow.TensorShapeProto tensor_shape_value = 31;</code>
-   */
-  org.tensorflow.proto.framework.TensorShapeProtoOrBuilder getTensorShapeValueOrBuilder();
-
-  /**
-   * <pre>
-   * Represents an enum value for dtype.
-   * </pre>
-   *
-   * <code>.tensorflow.DataType tensor_dtype_value = 32;</code>
-   */
-  int getTensorDtypeValueValue();
-  /**
-   * <pre>
-   * Represents an enum value for dtype.
-   * </pre>
-   *
-   * <code>.tensorflow.DataType tensor_dtype_value = 32;</code>
-   */
-  org.tensorflow.proto.framework.DataType getTensorDtypeValue();
-
-  /**
-   * <pre>
-   * Represents a value for tf.TensorSpec.
-   * </pre>
-   *
-   * <code>.tensorflow.TensorSpecProto tensor_spec_value = 33;</code>
-   */
-  boolean hasTensorSpecValue();
-  /**
-   * <pre>
-   * Represents a value for tf.TensorSpec.
-   * </pre>
-   *
-   * <code>.tensorflow.TensorSpecProto tensor_spec_value = 33;</code>
-   */
-  org.tensorflow.proto.framework.TensorSpecProto getTensorSpecValue();
-  /**
-   * <pre>
-   * Represents a value for tf.TensorSpec.
-   * </pre>
-   *
-   * <code>.tensorflow.TensorSpecProto tensor_spec_value = 33;</code>
-   */
-  org.tensorflow.proto.framework.TensorSpecProtoOrBuilder getTensorSpecValueOrBuilder();
-
-  /**
-   * <pre>
-   * Represents a value for tf.TypeSpec.
-   * </pre>
-   *
-   * <code>.tensorflow.TypeSpecProto type_spec_value = 34;</code>
-   */
-  boolean hasTypeSpecValue();
-  /**
-   * <pre>
-   * Represents a value for tf.TypeSpec.
-   * </pre>
-   *
-   * <code>.tensorflow.TypeSpecProto type_spec_value = 34;</code>
-   */
-  org.tensorflow.proto.framework.TypeSpecProto getTypeSpecValue();
-  /**
-   * <pre>
-   * Represents a value for tf.TypeSpec.
-   * </pre>
-   *
-   * <code>.tensorflow.TypeSpecProto type_spec_value = 34;</code>
-   */
-  org.tensorflow.proto.framework.TypeSpecProtoOrBuilder getTypeSpecValueOrBuilder();
-
-  /**
-   * <pre>
-   * Represents a value for tf.BoundedTensorSpec.
-   * </pre>
-   *
-   * <code>.tensorflow.BoundedTensorSpecProto bounded_tensor_spec_value = 35;</code>
-   */
-  boolean hasBoundedTensorSpecValue();
-  /**
-   * <pre>
-   * Represents a value for tf.BoundedTensorSpec.
-   * </pre>
-   *
-   * <code>.tensorflow.BoundedTensorSpecProto bounded_tensor_spec_value = 35;</code>
-   */
-  org.tensorflow.proto.framework.BoundedTensorSpecProto getBoundedTensorSpecValue();
-  /**
-   * <pre>
-   * Represents a value for tf.BoundedTensorSpec.
-   * </pre>
-   *
-   * <code>.tensorflow.BoundedTensorSpecProto bounded_tensor_spec_value = 35;</code>
-   */
-  org.tensorflow.proto.framework.BoundedTensorSpecProtoOrBuilder getBoundedTensorSpecValueOrBuilder();
-
-  /**
-   * <pre>
-   * Represents a list of `Value`.
-   * </pre>
-   *
-   * <code>.tensorflow.ListValue list_value = 51;</code>
-   */
-  boolean hasListValue();
-  /**
-   * <pre>
-   * Represents a list of `Value`.
-   * </pre>
-   *
-   * <code>.tensorflow.ListValue list_value = 51;</code>
-   */
-  org.tensorflow.proto.framework.ListValue getListValue();
-  /**
-   * <pre>
-   * Represents a list of `Value`.
-   * </pre>
-   *
-   * <code>.tensorflow.ListValue list_value = 51;</code>
-   */
-  org.tensorflow.proto.framework.ListValueOrBuilder getListValueOrBuilder();
-
-  /**
-   * <pre>
-   * Represents a tuple of `Value`.
-   * </pre>
-   *
-   * <code>.tensorflow.TupleValue tuple_value = 52;</code>
-   */
-  boolean hasTupleValue();
-  /**
-   * <pre>
-   * Represents a tuple of `Value`.
-   * </pre>
-   *
-   * <code>.tensorflow.TupleValue tuple_value = 52;</code>
-   */
-  org.tensorflow.proto.framework.TupleValue getTupleValue();
-  /**
-   * <pre>
-   * Represents a tuple of `Value`.
-   * </pre>
-   *
-   * <code>.tensorflow.TupleValue tuple_value = 52;</code>
-   */
-  org.tensorflow.proto.framework.TupleValueOrBuilder getTupleValueOrBuilder();
-
-  /**
-   * <pre>
-   * Represents a dict `Value`.
-   * </pre>
-   *
-   * <code>.tensorflow.DictValue dict_value = 53;</code>
-   */
-  boolean hasDictValue();
-  /**
-   * <pre>
-   * Represents a dict `Value`.
-   * </pre>
-   *
-   * <code>.tensorflow.DictValue dict_value = 53;</code>
-   */
-  org.tensorflow.proto.framework.DictValue getDictValue();
-  /**
-   * <pre>
-   * Represents a dict `Value`.
-   * </pre>
-   *
-   * <code>.tensorflow.DictValue dict_value = 53;</code>
-   */
-  org.tensorflow.proto.framework.DictValueOrBuilder getDictValueOrBuilder();
-
-  /**
-   * <pre>
-   * Represents Python's namedtuple.
-   * </pre>
-   *
-   * <code>.tensorflow.NamedTupleValue named_tuple_value = 54;</code>
-   */
-  boolean hasNamedTupleValue();
-  /**
-   * <pre>
-   * Represents Python's namedtuple.
-   * </pre>
-   *
-   * <code>.tensorflow.NamedTupleValue named_tuple_value = 54;</code>
-   */
-  org.tensorflow.proto.framework.NamedTupleValue getNamedTupleValue();
-  /**
-   * <pre>
-   * Represents Python's namedtuple.
-   * </pre>
-   *
-   * <code>.tensorflow.NamedTupleValue named_tuple_value = 54;</code>
-   */
-  org.tensorflow.proto.framework.NamedTupleValueOrBuilder getNamedTupleValueOrBuilder();
-
-  public org.tensorflow.proto.framework.StructuredValue.KindCase getKindCase();
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/Summary.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/Summary.java
deleted file mode 100644
index 19c5c6664a3..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/Summary.java
+++ /dev/null
@@ -1,4725 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/framework/summary.proto
-
-package org.tensorflow.proto.framework;
-
-/**
- * <pre>
- * A Summary is a set of named values to be displayed by the
- * visualizer.
- * Summaries are produced regularly during training, as controlled by
- * the "summary_interval_secs" attribute of the training operation.
- * Summaries are also produced at the end of an evaluation.
- * </pre>
- *
- * Protobuf type {@code tensorflow.Summary}
- */
-public  final class Summary extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.Summary)
-    SummaryOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use Summary.newBuilder() to construct.
-  private Summary(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private Summary() {
-    value_ = java.util.Collections.emptyList();
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new Summary();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private Summary(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    int mutable_bitField0_ = 0;
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 10: {
-            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
-              value_ = new java.util.ArrayList<org.tensorflow.proto.framework.Summary.Value>();
-              mutable_bitField0_ |= 0x00000001;
-            }
-            value_.add(
-                input.readMessage(org.tensorflow.proto.framework.Summary.Value.parser(), extensionRegistry));
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      if (((mutable_bitField0_ & 0x00000001) != 0)) {
-        value_ = java.util.Collections.unmodifiableList(value_);
-      }
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.framework.SummaryProtos.internal_static_tensorflow_Summary_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.framework.SummaryProtos.internal_static_tensorflow_Summary_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.framework.Summary.class, org.tensorflow.proto.framework.Summary.Builder.class);
-  }
-
-  public interface ImageOrBuilder extends
-      // @@protoc_insertion_point(interface_extends:tensorflow.Summary.Image)
-      com.google.protobuf.MessageOrBuilder {
-
-    /**
-     * <pre>
-     * Dimensions of the image.
-     * </pre>
-     *
-     * <code>int32 height = 1;</code>
-     */
-    int getHeight();
-
-    /**
-     * <code>int32 width = 2;</code>
-     */
-    int getWidth();
-
-    /**
-     * <pre>
-     * Valid colorspace values are
-     *   1 - grayscale
-     *   2 - grayscale + alpha
-     *   3 - RGB
-     *   4 - RGBA
-     *   5 - DIGITAL_YUV
-     *   6 - BGRA
-     * </pre>
-     *
-     * <code>int32 colorspace = 3;</code>
-     */
-    int getColorspace();
-
-    /**
-     * <pre>
-     * Image data in encoded format.  All image formats supported by
-     * image_codec::CoderUtil can be stored here.
-     * </pre>
-     *
-     * <code>bytes encoded_image_string = 4;</code>
-     */
-    com.google.protobuf.ByteString getEncodedImageString();
-  }
-  /**
-   * Protobuf type {@code tensorflow.Summary.Image}
-   */
-  public  static final class Image extends
-      com.google.protobuf.GeneratedMessageV3 implements
-      // @@protoc_insertion_point(message_implements:tensorflow.Summary.Image)
-      ImageOrBuilder {
-  private static final long serialVersionUID = 0L;
-    // Use Image.newBuilder() to construct.
-    private Image(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-      super(builder);
-    }
-    private Image() {
-      encodedImageString_ = com.google.protobuf.ByteString.EMPTY;
-    }
-
-    @java.lang.Override
-    @SuppressWarnings({"unused"})
-    protected java.lang.Object newInstance(
-        UnusedPrivateParameter unused) {
-      return new Image();
-    }
-
-    @java.lang.Override
-    public final com.google.protobuf.UnknownFieldSet
-    getUnknownFields() {
-      return this.unknownFields;
-    }
-    private Image(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      this();
-      if (extensionRegistry == null) {
-        throw new java.lang.NullPointerException();
-      }
-      com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-          com.google.protobuf.UnknownFieldSet.newBuilder();
-      try {
-        boolean done = false;
-        while (!done) {
-          int tag = input.readTag();
-          switch (tag) {
-            case 0:
-              done = true;
-              break;
-            case 8: {
-
-              height_ = input.readInt32();
-              break;
-            }
-            case 16: {
-
-              width_ = input.readInt32();
-              break;
-            }
-            case 24: {
-
-              colorspace_ = input.readInt32();
-              break;
-            }
-            case 34: {
-
-              encodedImageString_ = input.readBytes();
-              break;
-            }
-            default: {
-              if (!parseUnknownField(
-                  input, unknownFields, extensionRegistry, tag)) {
-                done = true;
-              }
-              break;
-            }
-          }
-        }
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        throw e.setUnfinishedMessage(this);
-      } catch (java.io.IOException e) {
-        throw new com.google.protobuf.InvalidProtocolBufferException(
-            e).setUnfinishedMessage(this);
-      } finally {
-        this.unknownFields = unknownFields.build();
-        makeExtensionsImmutable();
-      }
-    }
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.SummaryProtos.internal_static_tensorflow_Summary_Image_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.SummaryProtos.internal_static_tensorflow_Summary_Image_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.Summary.Image.class, org.tensorflow.proto.framework.Summary.Image.Builder.class);
-    }
-
-    public static final int HEIGHT_FIELD_NUMBER = 1;
-    private int height_;
-    /**
-     * <pre>
-     * Dimensions of the image.
-     * </pre>
-     *
-     * <code>int32 height = 1;</code>
-     */
-    public int getHeight() {
-      return height_;
-    }
-
-    public static final int WIDTH_FIELD_NUMBER = 2;
-    private int width_;
-    /**
-     * <code>int32 width = 2;</code>
-     */
-    public int getWidth() {
-      return width_;
-    }
-
-    public static final int COLORSPACE_FIELD_NUMBER = 3;
-    private int colorspace_;
-    /**
-     * <pre>
-     * Valid colorspace values are
-     *   1 - grayscale
-     *   2 - grayscale + alpha
-     *   3 - RGB
-     *   4 - RGBA
-     *   5 - DIGITAL_YUV
-     *   6 - BGRA
-     * </pre>
-     *
-     * <code>int32 colorspace = 3;</code>
-     */
-    public int getColorspace() {
-      return colorspace_;
-    }
-
-    public static final int ENCODED_IMAGE_STRING_FIELD_NUMBER = 4;
-    private com.google.protobuf.ByteString encodedImageString_;
-    /**
-     * <pre>
-     * Image data in encoded format.  All image formats supported by
-     * image_codec::CoderUtil can be stored here.
-     * </pre>
-     *
-     * <code>bytes encoded_image_string = 4;</code>
-     */
-    public com.google.protobuf.ByteString getEncodedImageString() {
-      return encodedImageString_;
-    }
-
-    private byte memoizedIsInitialized = -1;
-    @java.lang.Override
-    public final boolean isInitialized() {
-      byte isInitialized = memoizedIsInitialized;
-      if (isInitialized == 1) return true;
-      if (isInitialized == 0) return false;
-
-      memoizedIsInitialized = 1;
-      return true;
-    }
-
-    @java.lang.Override
-    public void writeTo(com.google.protobuf.CodedOutputStream output)
-                        throws java.io.IOException {
-      if (height_ != 0) {
-        output.writeInt32(1, height_);
-      }
-      if (width_ != 0) {
-        output.writeInt32(2, width_);
-      }
-      if (colorspace_ != 0) {
-        output.writeInt32(3, colorspace_);
-      }
-      if (!encodedImageString_.isEmpty()) {
-        output.writeBytes(4, encodedImageString_);
-      }
-      unknownFields.writeTo(output);
-    }
-
-    @java.lang.Override
-    public int getSerializedSize() {
-      int size = memoizedSize;
-      if (size != -1) return size;
-
-      size = 0;
-      if (height_ != 0) {
-        size += com.google.protobuf.CodedOutputStream
-          .computeInt32Size(1, height_);
-      }
-      if (width_ != 0) {
-        size += com.google.protobuf.CodedOutputStream
-          .computeInt32Size(2, width_);
-      }
-      if (colorspace_ != 0) {
-        size += com.google.protobuf.CodedOutputStream
-          .computeInt32Size(3, colorspace_);
-      }
-      if (!encodedImageString_.isEmpty()) {
-        size += com.google.protobuf.CodedOutputStream
-          .computeBytesSize(4, encodedImageString_);
-      }
-      size += unknownFields.getSerializedSize();
-      memoizedSize = size;
-      return size;
-    }
-
-    @java.lang.Override
-    public boolean equals(final java.lang.Object obj) {
-      if (obj == this) {
-       return true;
-      }
-      if (!(obj instanceof org.tensorflow.proto.framework.Summary.Image)) {
-        return super.equals(obj);
-      }
-      org.tensorflow.proto.framework.Summary.Image other = (org.tensorflow.proto.framework.Summary.Image) obj;
-
-      if (getHeight()
-          != other.getHeight()) return false;
-      if (getWidth()
-          != other.getWidth()) return false;
-      if (getColorspace()
-          != other.getColorspace()) return false;
-      if (!getEncodedImageString()
-          .equals(other.getEncodedImageString())) return false;
-      if (!unknownFields.equals(other.unknownFields)) return false;
-      return true;
-    }
-
-    @java.lang.Override
-    public int hashCode() {
-      if (memoizedHashCode != 0) {
-        return memoizedHashCode;
-      }
-      int hash = 41;
-      hash = (19 * hash) + getDescriptor().hashCode();
-      hash = (37 * hash) + HEIGHT_FIELD_NUMBER;
-      hash = (53 * hash) + getHeight();
-      hash = (37 * hash) + WIDTH_FIELD_NUMBER;
-      hash = (53 * hash) + getWidth();
-      hash = (37 * hash) + COLORSPACE_FIELD_NUMBER;
-      hash = (53 * hash) + getColorspace();
-      hash = (37 * hash) + ENCODED_IMAGE_STRING_FIELD_NUMBER;
-      hash = (53 * hash) + getEncodedImageString().hashCode();
-      hash = (29 * hash) + unknownFields.hashCode();
-      memoizedHashCode = hash;
-      return hash;
-    }
-
-    public static org.tensorflow.proto.framework.Summary.Image parseFrom(
-        java.nio.ByteBuffer data)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data);
-    }
-    public static org.tensorflow.proto.framework.Summary.Image parseFrom(
-        java.nio.ByteBuffer data,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.Summary.Image parseFrom(
-        com.google.protobuf.ByteString data)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data);
-    }
-    public static org.tensorflow.proto.framework.Summary.Image parseFrom(
-        com.google.protobuf.ByteString data,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.Summary.Image parseFrom(byte[] data)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data);
-    }
-    public static org.tensorflow.proto.framework.Summary.Image parseFrom(
-        byte[] data,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.Summary.Image parseFrom(java.io.InputStream input)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input);
-    }
-    public static org.tensorflow.proto.framework.Summary.Image parseFrom(
-        java.io.InputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.Summary.Image parseDelimitedFrom(java.io.InputStream input)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseDelimitedWithIOException(PARSER, input);
-    }
-    public static org.tensorflow.proto.framework.Summary.Image parseDelimitedFrom(
-        java.io.InputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.Summary.Image parseFrom(
-        com.google.protobuf.CodedInputStream input)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input);
-    }
-    public static org.tensorflow.proto.framework.Summary.Image parseFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input, extensionRegistry);
-    }
-
-    @java.lang.Override
-    public Builder newBuilderForType() { return newBuilder(); }
-    public static Builder newBuilder() {
-      return DEFAULT_INSTANCE.toBuilder();
-    }
-    public static Builder newBuilder(org.tensorflow.proto.framework.Summary.Image prototype) {
-      return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-    }
-    @java.lang.Override
-    public Builder toBuilder() {
-      return this == DEFAULT_INSTANCE
-          ? new Builder() : new Builder().mergeFrom(this);
-    }
-
-    @java.lang.Override
-    protected Builder newBuilderForType(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      Builder builder = new Builder(parent);
-      return builder;
-    }
-    /**
-     * Protobuf type {@code tensorflow.Summary.Image}
-     */
-    public static final class Builder extends
-        com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-        // @@protoc_insertion_point(builder_implements:tensorflow.Summary.Image)
-        org.tensorflow.proto.framework.Summary.ImageOrBuilder {
-      public static final com.google.protobuf.Descriptors.Descriptor
-          getDescriptor() {
-        return org.tensorflow.proto.framework.SummaryProtos.internal_static_tensorflow_Summary_Image_descriptor;
-      }
-
-      @java.lang.Override
-      protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-          internalGetFieldAccessorTable() {
-        return org.tensorflow.proto.framework.SummaryProtos.internal_static_tensorflow_Summary_Image_fieldAccessorTable
-            .ensureFieldAccessorsInitialized(
-                org.tensorflow.proto.framework.Summary.Image.class, org.tensorflow.proto.framework.Summary.Image.Builder.class);
-      }
-
-      // Construct using org.tensorflow.proto.framework.Summary.Image.newBuilder()
-      private Builder() {
-        maybeForceBuilderInitialization();
-      }
-
-      private Builder(
-          com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-        super(parent);
-        maybeForceBuilderInitialization();
-      }
-      private void maybeForceBuilderInitialization() {
-        if (com.google.protobuf.GeneratedMessageV3
-                .alwaysUseFieldBuilders) {
-        }
-      }
-      @java.lang.Override
-      public Builder clear() {
-        super.clear();
-        height_ = 0;
-
-        width_ = 0;
-
-        colorspace_ = 0;
-
-        encodedImageString_ = com.google.protobuf.ByteString.EMPTY;
-
-        return this;
-      }
-
-      @java.lang.Override
-      public com.google.protobuf.Descriptors.Descriptor
-          getDescriptorForType() {
-        return org.tensorflow.proto.framework.SummaryProtos.internal_static_tensorflow_Summary_Image_descriptor;
-      }
-
-      @java.lang.Override
-      public org.tensorflow.proto.framework.Summary.Image getDefaultInstanceForType() {
-        return org.tensorflow.proto.framework.Summary.Image.getDefaultInstance();
-      }
-
-      @java.lang.Override
-      public org.tensorflow.proto.framework.Summary.Image build() {
-        org.tensorflow.proto.framework.Summary.Image result = buildPartial();
-        if (!result.isInitialized()) {
-          throw newUninitializedMessageException(result);
-        }
-        return result;
-      }
-
-      @java.lang.Override
-      public org.tensorflow.proto.framework.Summary.Image buildPartial() {
-        org.tensorflow.proto.framework.Summary.Image result = new org.tensorflow.proto.framework.Summary.Image(this);
-        result.height_ = height_;
-        result.width_ = width_;
-        result.colorspace_ = colorspace_;
-        result.encodedImageString_ = encodedImageString_;
-        onBuilt();
-        return result;
-      }
-
-      @java.lang.Override
-      public Builder clone() {
-        return super.clone();
-      }
-      @java.lang.Override
-      public Builder setField(
-          com.google.protobuf.Descriptors.FieldDescriptor field,
-          java.lang.Object value) {
-        return super.setField(field, value);
-      }
-      @java.lang.Override
-      public Builder clearField(
-          com.google.protobuf.Descriptors.FieldDescriptor field) {
-        return super.clearField(field);
-      }
-      @java.lang.Override
-      public Builder clearOneof(
-          com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-        return super.clearOneof(oneof);
-      }
-      @java.lang.Override
-      public Builder setRepeatedField(
-          com.google.protobuf.Descriptors.FieldDescriptor field,
-          int index, java.lang.Object value) {
-        return super.setRepeatedField(field, index, value);
-      }
-      @java.lang.Override
-      public Builder addRepeatedField(
-          com.google.protobuf.Descriptors.FieldDescriptor field,
-          java.lang.Object value) {
-        return super.addRepeatedField(field, value);
-      }
-      @java.lang.Override
-      public Builder mergeFrom(com.google.protobuf.Message other) {
-        if (other instanceof org.tensorflow.proto.framework.Summary.Image) {
-          return mergeFrom((org.tensorflow.proto.framework.Summary.Image)other);
-        } else {
-          super.mergeFrom(other);
-          return this;
-        }
-      }
-
-      public Builder mergeFrom(org.tensorflow.proto.framework.Summary.Image other) {
-        if (other == org.tensorflow.proto.framework.Summary.Image.getDefaultInstance()) return this;
-        if (other.getHeight() != 0) {
-          setHeight(other.getHeight());
-        }
-        if (other.getWidth() != 0) {
-          setWidth(other.getWidth());
-        }
-        if (other.getColorspace() != 0) {
-          setColorspace(other.getColorspace());
-        }
-        if (other.getEncodedImageString() != com.google.protobuf.ByteString.EMPTY) {
-          setEncodedImageString(other.getEncodedImageString());
-        }
-        this.mergeUnknownFields(other.unknownFields);
-        onChanged();
-        return this;
-      }
-
-      @java.lang.Override
-      public final boolean isInitialized() {
-        return true;
-      }
-
-      @java.lang.Override
-      public Builder mergeFrom(
-          com.google.protobuf.CodedInputStream input,
-          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-          throws java.io.IOException {
-        org.tensorflow.proto.framework.Summary.Image parsedMessage = null;
-        try {
-          parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-        } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-          parsedMessage = (org.tensorflow.proto.framework.Summary.Image) e.getUnfinishedMessage();
-          throw e.unwrapIOException();
-        } finally {
-          if (parsedMessage != null) {
-            mergeFrom(parsedMessage);
-          }
-        }
-        return this;
-      }
-
-      private int height_ ;
-      /**
-       * <pre>
-       * Dimensions of the image.
-       * </pre>
-       *
-       * <code>int32 height = 1;</code>
-       */
-      public int getHeight() {
-        return height_;
-      }
-      /**
-       * <pre>
-       * Dimensions of the image.
-       * </pre>
-       *
-       * <code>int32 height = 1;</code>
-       */
-      public Builder setHeight(int value) {
-        
-        height_ = value;
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * Dimensions of the image.
-       * </pre>
-       *
-       * <code>int32 height = 1;</code>
-       */
-      public Builder clearHeight() {
-        
-        height_ = 0;
-        onChanged();
-        return this;
-      }
-
-      private int width_ ;
-      /**
-       * <code>int32 width = 2;</code>
-       */
-      public int getWidth() {
-        return width_;
-      }
-      /**
-       * <code>int32 width = 2;</code>
-       */
-      public Builder setWidth(int value) {
-        
-        width_ = value;
-        onChanged();
-        return this;
-      }
-      /**
-       * <code>int32 width = 2;</code>
-       */
-      public Builder clearWidth() {
-        
-        width_ = 0;
-        onChanged();
-        return this;
-      }
-
-      private int colorspace_ ;
-      /**
-       * <pre>
-       * Valid colorspace values are
-       *   1 - grayscale
-       *   2 - grayscale + alpha
-       *   3 - RGB
-       *   4 - RGBA
-       *   5 - DIGITAL_YUV
-       *   6 - BGRA
-       * </pre>
-       *
-       * <code>int32 colorspace = 3;</code>
-       */
-      public int getColorspace() {
-        return colorspace_;
-      }
-      /**
-       * <pre>
-       * Valid colorspace values are
-       *   1 - grayscale
-       *   2 - grayscale + alpha
-       *   3 - RGB
-       *   4 - RGBA
-       *   5 - DIGITAL_YUV
-       *   6 - BGRA
-       * </pre>
-       *
-       * <code>int32 colorspace = 3;</code>
-       */
-      public Builder setColorspace(int value) {
-        
-        colorspace_ = value;
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * Valid colorspace values are
-       *   1 - grayscale
-       *   2 - grayscale + alpha
-       *   3 - RGB
-       *   4 - RGBA
-       *   5 - DIGITAL_YUV
-       *   6 - BGRA
-       * </pre>
-       *
-       * <code>int32 colorspace = 3;</code>
-       */
-      public Builder clearColorspace() {
-        
-        colorspace_ = 0;
-        onChanged();
-        return this;
-      }
-
-      private com.google.protobuf.ByteString encodedImageString_ = com.google.protobuf.ByteString.EMPTY;
-      /**
-       * <pre>
-       * Image data in encoded format.  All image formats supported by
-       * image_codec::CoderUtil can be stored here.
-       * </pre>
-       *
-       * <code>bytes encoded_image_string = 4;</code>
-       */
-      public com.google.protobuf.ByteString getEncodedImageString() {
-        return encodedImageString_;
-      }
-      /**
-       * <pre>
-       * Image data in encoded format.  All image formats supported by
-       * image_codec::CoderUtil can be stored here.
-       * </pre>
-       *
-       * <code>bytes encoded_image_string = 4;</code>
-       */
-      public Builder setEncodedImageString(com.google.protobuf.ByteString value) {
-        if (value == null) {
-    throw new NullPointerException();
-  }
-  
-        encodedImageString_ = value;
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * Image data in encoded format.  All image formats supported by
-       * image_codec::CoderUtil can be stored here.
-       * </pre>
-       *
-       * <code>bytes encoded_image_string = 4;</code>
-       */
-      public Builder clearEncodedImageString() {
-        
-        encodedImageString_ = getDefaultInstance().getEncodedImageString();
-        onChanged();
-        return this;
-      }
-      @java.lang.Override
-      public final Builder setUnknownFields(
-          final com.google.protobuf.UnknownFieldSet unknownFields) {
-        return super.setUnknownFields(unknownFields);
-      }
-
-      @java.lang.Override
-      public final Builder mergeUnknownFields(
-          final com.google.protobuf.UnknownFieldSet unknownFields) {
-        return super.mergeUnknownFields(unknownFields);
-      }
-
-
-      // @@protoc_insertion_point(builder_scope:tensorflow.Summary.Image)
-    }
-
-    // @@protoc_insertion_point(class_scope:tensorflow.Summary.Image)
-    private static final org.tensorflow.proto.framework.Summary.Image DEFAULT_INSTANCE;
-    static {
-      DEFAULT_INSTANCE = new org.tensorflow.proto.framework.Summary.Image();
-    }
-
-    public static org.tensorflow.proto.framework.Summary.Image getDefaultInstance() {
-      return DEFAULT_INSTANCE;
-    }
-
-    private static final com.google.protobuf.Parser<Image>
-        PARSER = new com.google.protobuf.AbstractParser<Image>() {
-      @java.lang.Override
-      public Image parsePartialFrom(
-          com.google.protobuf.CodedInputStream input,
-          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-          throws com.google.protobuf.InvalidProtocolBufferException {
-        return new Image(input, extensionRegistry);
-      }
-    };
-
-    public static com.google.protobuf.Parser<Image> parser() {
-      return PARSER;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Parser<Image> getParserForType() {
-      return PARSER;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.Summary.Image getDefaultInstanceForType() {
-      return DEFAULT_INSTANCE;
-    }
-
-  }
-
-  public interface AudioOrBuilder extends
-      // @@protoc_insertion_point(interface_extends:tensorflow.Summary.Audio)
-      com.google.protobuf.MessageOrBuilder {
-
-    /**
-     * <pre>
-     * Sample rate of the audio in Hz.
-     * </pre>
-     *
-     * <code>float sample_rate = 1;</code>
-     */
-    float getSampleRate();
-
-    /**
-     * <pre>
-     * Number of channels of audio.
-     * </pre>
-     *
-     * <code>int64 num_channels = 2;</code>
-     */
-    long getNumChannels();
-
-    /**
-     * <pre>
-     * Length of the audio in frames (samples per channel).
-     * </pre>
-     *
-     * <code>int64 length_frames = 3;</code>
-     */
-    long getLengthFrames();
-
-    /**
-     * <pre>
-     * Encoded audio data and its associated RFC 2045 content type (e.g.
-     * "audio/wav").
-     * </pre>
-     *
-     * <code>bytes encoded_audio_string = 4;</code>
-     */
-    com.google.protobuf.ByteString getEncodedAudioString();
-
-    /**
-     * <code>string content_type = 5;</code>
-     */
-    java.lang.String getContentType();
-    /**
-     * <code>string content_type = 5;</code>
-     */
-    com.google.protobuf.ByteString
-        getContentTypeBytes();
-  }
-  /**
-   * Protobuf type {@code tensorflow.Summary.Audio}
-   */
-  public  static final class Audio extends
-      com.google.protobuf.GeneratedMessageV3 implements
-      // @@protoc_insertion_point(message_implements:tensorflow.Summary.Audio)
-      AudioOrBuilder {
-  private static final long serialVersionUID = 0L;
-    // Use Audio.newBuilder() to construct.
-    private Audio(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-      super(builder);
-    }
-    private Audio() {
-      encodedAudioString_ = com.google.protobuf.ByteString.EMPTY;
-      contentType_ = "";
-    }
-
-    @java.lang.Override
-    @SuppressWarnings({"unused"})
-    protected java.lang.Object newInstance(
-        UnusedPrivateParameter unused) {
-      return new Audio();
-    }
-
-    @java.lang.Override
-    public final com.google.protobuf.UnknownFieldSet
-    getUnknownFields() {
-      return this.unknownFields;
-    }
-    private Audio(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      this();
-      if (extensionRegistry == null) {
-        throw new java.lang.NullPointerException();
-      }
-      com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-          com.google.protobuf.UnknownFieldSet.newBuilder();
-      try {
-        boolean done = false;
-        while (!done) {
-          int tag = input.readTag();
-          switch (tag) {
-            case 0:
-              done = true;
-              break;
-            case 13: {
-
-              sampleRate_ = input.readFloat();
-              break;
-            }
-            case 16: {
-
-              numChannels_ = input.readInt64();
-              break;
-            }
-            case 24: {
-
-              lengthFrames_ = input.readInt64();
-              break;
-            }
-            case 34: {
-
-              encodedAudioString_ = input.readBytes();
-              break;
-            }
-            case 42: {
-              java.lang.String s = input.readStringRequireUtf8();
-
-              contentType_ = s;
-              break;
-            }
-            default: {
-              if (!parseUnknownField(
-                  input, unknownFields, extensionRegistry, tag)) {
-                done = true;
-              }
-              break;
-            }
-          }
-        }
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        throw e.setUnfinishedMessage(this);
-      } catch (java.io.IOException e) {
-        throw new com.google.protobuf.InvalidProtocolBufferException(
-            e).setUnfinishedMessage(this);
-      } finally {
-        this.unknownFields = unknownFields.build();
-        makeExtensionsImmutable();
-      }
-    }
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.SummaryProtos.internal_static_tensorflow_Summary_Audio_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.SummaryProtos.internal_static_tensorflow_Summary_Audio_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.Summary.Audio.class, org.tensorflow.proto.framework.Summary.Audio.Builder.class);
-    }
-
-    public static final int SAMPLE_RATE_FIELD_NUMBER = 1;
-    private float sampleRate_;
-    /**
-     * <pre>
-     * Sample rate of the audio in Hz.
-     * </pre>
-     *
-     * <code>float sample_rate = 1;</code>
-     */
-    public float getSampleRate() {
-      return sampleRate_;
-    }
-
-    public static final int NUM_CHANNELS_FIELD_NUMBER = 2;
-    private long numChannels_;
-    /**
-     * <pre>
-     * Number of channels of audio.
-     * </pre>
-     *
-     * <code>int64 num_channels = 2;</code>
-     */
-    public long getNumChannels() {
-      return numChannels_;
-    }
-
-    public static final int LENGTH_FRAMES_FIELD_NUMBER = 3;
-    private long lengthFrames_;
-    /**
-     * <pre>
-     * Length of the audio in frames (samples per channel).
-     * </pre>
-     *
-     * <code>int64 length_frames = 3;</code>
-     */
-    public long getLengthFrames() {
-      return lengthFrames_;
-    }
-
-    public static final int ENCODED_AUDIO_STRING_FIELD_NUMBER = 4;
-    private com.google.protobuf.ByteString encodedAudioString_;
-    /**
-     * <pre>
-     * Encoded audio data and its associated RFC 2045 content type (e.g.
-     * "audio/wav").
-     * </pre>
-     *
-     * <code>bytes encoded_audio_string = 4;</code>
-     */
-    public com.google.protobuf.ByteString getEncodedAudioString() {
-      return encodedAudioString_;
-    }
-
-    public static final int CONTENT_TYPE_FIELD_NUMBER = 5;
-    private volatile java.lang.Object contentType_;
-    /**
-     * <code>string content_type = 5;</code>
-     */
-    public java.lang.String getContentType() {
-      java.lang.Object ref = contentType_;
-      if (ref instanceof java.lang.String) {
-        return (java.lang.String) ref;
-      } else {
-        com.google.protobuf.ByteString bs = 
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        contentType_ = s;
-        return s;
-      }
-    }
-    /**
-     * <code>string content_type = 5;</code>
-     */
-    public com.google.protobuf.ByteString
-        getContentTypeBytes() {
-      java.lang.Object ref = contentType_;
-      if (ref instanceof java.lang.String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        contentType_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-
-    private byte memoizedIsInitialized = -1;
-    @java.lang.Override
-    public final boolean isInitialized() {
-      byte isInitialized = memoizedIsInitialized;
-      if (isInitialized == 1) return true;
-      if (isInitialized == 0) return false;
-
-      memoizedIsInitialized = 1;
-      return true;
-    }
-
-    @java.lang.Override
-    public void writeTo(com.google.protobuf.CodedOutputStream output)
-                        throws java.io.IOException {
-      if (sampleRate_ != 0F) {
-        output.writeFloat(1, sampleRate_);
-      }
-      if (numChannels_ != 0L) {
-        output.writeInt64(2, numChannels_);
-      }
-      if (lengthFrames_ != 0L) {
-        output.writeInt64(3, lengthFrames_);
-      }
-      if (!encodedAudioString_.isEmpty()) {
-        output.writeBytes(4, encodedAudioString_);
-      }
-      if (!getContentTypeBytes().isEmpty()) {
-        com.google.protobuf.GeneratedMessageV3.writeString(output, 5, contentType_);
-      }
-      unknownFields.writeTo(output);
-    }
-
-    @java.lang.Override
-    public int getSerializedSize() {
-      int size = memoizedSize;
-      if (size != -1) return size;
-
-      size = 0;
-      if (sampleRate_ != 0F) {
-        size += com.google.protobuf.CodedOutputStream
-          .computeFloatSize(1, sampleRate_);
-      }
-      if (numChannels_ != 0L) {
-        size += com.google.protobuf.CodedOutputStream
-          .computeInt64Size(2, numChannels_);
-      }
-      if (lengthFrames_ != 0L) {
-        size += com.google.protobuf.CodedOutputStream
-          .computeInt64Size(3, lengthFrames_);
-      }
-      if (!encodedAudioString_.isEmpty()) {
-        size += com.google.protobuf.CodedOutputStream
-          .computeBytesSize(4, encodedAudioString_);
-      }
-      if (!getContentTypeBytes().isEmpty()) {
-        size += com.google.protobuf.GeneratedMessageV3.computeStringSize(5, contentType_);
-      }
-      size += unknownFields.getSerializedSize();
-      memoizedSize = size;
-      return size;
-    }
-
-    @java.lang.Override
-    public boolean equals(final java.lang.Object obj) {
-      if (obj == this) {
-       return true;
-      }
-      if (!(obj instanceof org.tensorflow.proto.framework.Summary.Audio)) {
-        return super.equals(obj);
-      }
-      org.tensorflow.proto.framework.Summary.Audio other = (org.tensorflow.proto.framework.Summary.Audio) obj;
-
-      if (java.lang.Float.floatToIntBits(getSampleRate())
-          != java.lang.Float.floatToIntBits(
-              other.getSampleRate())) return false;
-      if (getNumChannels()
-          != other.getNumChannels()) return false;
-      if (getLengthFrames()
-          != other.getLengthFrames()) return false;
-      if (!getEncodedAudioString()
-          .equals(other.getEncodedAudioString())) return false;
-      if (!getContentType()
-          .equals(other.getContentType())) return false;
-      if (!unknownFields.equals(other.unknownFields)) return false;
-      return true;
-    }
-
-    @java.lang.Override
-    public int hashCode() {
-      if (memoizedHashCode != 0) {
-        return memoizedHashCode;
-      }
-      int hash = 41;
-      hash = (19 * hash) + getDescriptor().hashCode();
-      hash = (37 * hash) + SAMPLE_RATE_FIELD_NUMBER;
-      hash = (53 * hash) + java.lang.Float.floatToIntBits(
-          getSampleRate());
-      hash = (37 * hash) + NUM_CHANNELS_FIELD_NUMBER;
-      hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-          getNumChannels());
-      hash = (37 * hash) + LENGTH_FRAMES_FIELD_NUMBER;
-      hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-          getLengthFrames());
-      hash = (37 * hash) + ENCODED_AUDIO_STRING_FIELD_NUMBER;
-      hash = (53 * hash) + getEncodedAudioString().hashCode();
-      hash = (37 * hash) + CONTENT_TYPE_FIELD_NUMBER;
-      hash = (53 * hash) + getContentType().hashCode();
-      hash = (29 * hash) + unknownFields.hashCode();
-      memoizedHashCode = hash;
-      return hash;
-    }
-
-    public static org.tensorflow.proto.framework.Summary.Audio parseFrom(
-        java.nio.ByteBuffer data)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data);
-    }
-    public static org.tensorflow.proto.framework.Summary.Audio parseFrom(
-        java.nio.ByteBuffer data,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.Summary.Audio parseFrom(
-        com.google.protobuf.ByteString data)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data);
-    }
-    public static org.tensorflow.proto.framework.Summary.Audio parseFrom(
-        com.google.protobuf.ByteString data,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.Summary.Audio parseFrom(byte[] data)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data);
-    }
-    public static org.tensorflow.proto.framework.Summary.Audio parseFrom(
-        byte[] data,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.Summary.Audio parseFrom(java.io.InputStream input)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input);
-    }
-    public static org.tensorflow.proto.framework.Summary.Audio parseFrom(
-        java.io.InputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.Summary.Audio parseDelimitedFrom(java.io.InputStream input)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseDelimitedWithIOException(PARSER, input);
-    }
-    public static org.tensorflow.proto.framework.Summary.Audio parseDelimitedFrom(
-        java.io.InputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.Summary.Audio parseFrom(
-        com.google.protobuf.CodedInputStream input)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input);
-    }
-    public static org.tensorflow.proto.framework.Summary.Audio parseFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input, extensionRegistry);
-    }
-
-    @java.lang.Override
-    public Builder newBuilderForType() { return newBuilder(); }
-    public static Builder newBuilder() {
-      return DEFAULT_INSTANCE.toBuilder();
-    }
-    public static Builder newBuilder(org.tensorflow.proto.framework.Summary.Audio prototype) {
-      return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-    }
-    @java.lang.Override
-    public Builder toBuilder() {
-      return this == DEFAULT_INSTANCE
-          ? new Builder() : new Builder().mergeFrom(this);
-    }
-
-    @java.lang.Override
-    protected Builder newBuilderForType(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      Builder builder = new Builder(parent);
-      return builder;
-    }
-    /**
-     * Protobuf type {@code tensorflow.Summary.Audio}
-     */
-    public static final class Builder extends
-        com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-        // @@protoc_insertion_point(builder_implements:tensorflow.Summary.Audio)
-        org.tensorflow.proto.framework.Summary.AudioOrBuilder {
-      public static final com.google.protobuf.Descriptors.Descriptor
-          getDescriptor() {
-        return org.tensorflow.proto.framework.SummaryProtos.internal_static_tensorflow_Summary_Audio_descriptor;
-      }
-
-      @java.lang.Override
-      protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-          internalGetFieldAccessorTable() {
-        return org.tensorflow.proto.framework.SummaryProtos.internal_static_tensorflow_Summary_Audio_fieldAccessorTable
-            .ensureFieldAccessorsInitialized(
-                org.tensorflow.proto.framework.Summary.Audio.class, org.tensorflow.proto.framework.Summary.Audio.Builder.class);
-      }
-
-      // Construct using org.tensorflow.proto.framework.Summary.Audio.newBuilder()
-      private Builder() {
-        maybeForceBuilderInitialization();
-      }
-
-      private Builder(
-          com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-        super(parent);
-        maybeForceBuilderInitialization();
-      }
-      private void maybeForceBuilderInitialization() {
-        if (com.google.protobuf.GeneratedMessageV3
-                .alwaysUseFieldBuilders) {
-        }
-      }
-      @java.lang.Override
-      public Builder clear() {
-        super.clear();
-        sampleRate_ = 0F;
-
-        numChannels_ = 0L;
-
-        lengthFrames_ = 0L;
-
-        encodedAudioString_ = com.google.protobuf.ByteString.EMPTY;
-
-        contentType_ = "";
-
-        return this;
-      }
-
-      @java.lang.Override
-      public com.google.protobuf.Descriptors.Descriptor
-          getDescriptorForType() {
-        return org.tensorflow.proto.framework.SummaryProtos.internal_static_tensorflow_Summary_Audio_descriptor;
-      }
-
-      @java.lang.Override
-      public org.tensorflow.proto.framework.Summary.Audio getDefaultInstanceForType() {
-        return org.tensorflow.proto.framework.Summary.Audio.getDefaultInstance();
-      }
-
-      @java.lang.Override
-      public org.tensorflow.proto.framework.Summary.Audio build() {
-        org.tensorflow.proto.framework.Summary.Audio result = buildPartial();
-        if (!result.isInitialized()) {
-          throw newUninitializedMessageException(result);
-        }
-        return result;
-      }
-
-      @java.lang.Override
-      public org.tensorflow.proto.framework.Summary.Audio buildPartial() {
-        org.tensorflow.proto.framework.Summary.Audio result = new org.tensorflow.proto.framework.Summary.Audio(this);
-        result.sampleRate_ = sampleRate_;
-        result.numChannels_ = numChannels_;
-        result.lengthFrames_ = lengthFrames_;
-        result.encodedAudioString_ = encodedAudioString_;
-        result.contentType_ = contentType_;
-        onBuilt();
-        return result;
-      }
-
-      @java.lang.Override
-      public Builder clone() {
-        return super.clone();
-      }
-      @java.lang.Override
-      public Builder setField(
-          com.google.protobuf.Descriptors.FieldDescriptor field,
-          java.lang.Object value) {
-        return super.setField(field, value);
-      }
-      @java.lang.Override
-      public Builder clearField(
-          com.google.protobuf.Descriptors.FieldDescriptor field) {
-        return super.clearField(field);
-      }
-      @java.lang.Override
-      public Builder clearOneof(
-          com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-        return super.clearOneof(oneof);
-      }
-      @java.lang.Override
-      public Builder setRepeatedField(
-          com.google.protobuf.Descriptors.FieldDescriptor field,
-          int index, java.lang.Object value) {
-        return super.setRepeatedField(field, index, value);
-      }
-      @java.lang.Override
-      public Builder addRepeatedField(
-          com.google.protobuf.Descriptors.FieldDescriptor field,
-          java.lang.Object value) {
-        return super.addRepeatedField(field, value);
-      }
-      @java.lang.Override
-      public Builder mergeFrom(com.google.protobuf.Message other) {
-        if (other instanceof org.tensorflow.proto.framework.Summary.Audio) {
-          return mergeFrom((org.tensorflow.proto.framework.Summary.Audio)other);
-        } else {
-          super.mergeFrom(other);
-          return this;
-        }
-      }
-
-      public Builder mergeFrom(org.tensorflow.proto.framework.Summary.Audio other) {
-        if (other == org.tensorflow.proto.framework.Summary.Audio.getDefaultInstance()) return this;
-        if (other.getSampleRate() != 0F) {
-          setSampleRate(other.getSampleRate());
-        }
-        if (other.getNumChannels() != 0L) {
-          setNumChannels(other.getNumChannels());
-        }
-        if (other.getLengthFrames() != 0L) {
-          setLengthFrames(other.getLengthFrames());
-        }
-        if (other.getEncodedAudioString() != com.google.protobuf.ByteString.EMPTY) {
-          setEncodedAudioString(other.getEncodedAudioString());
-        }
-        if (!other.getContentType().isEmpty()) {
-          contentType_ = other.contentType_;
-          onChanged();
-        }
-        this.mergeUnknownFields(other.unknownFields);
-        onChanged();
-        return this;
-      }
-
-      @java.lang.Override
-      public final boolean isInitialized() {
-        return true;
-      }
-
-      @java.lang.Override
-      public Builder mergeFrom(
-          com.google.protobuf.CodedInputStream input,
-          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-          throws java.io.IOException {
-        org.tensorflow.proto.framework.Summary.Audio parsedMessage = null;
-        try {
-          parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-        } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-          parsedMessage = (org.tensorflow.proto.framework.Summary.Audio) e.getUnfinishedMessage();
-          throw e.unwrapIOException();
-        } finally {
-          if (parsedMessage != null) {
-            mergeFrom(parsedMessage);
-          }
-        }
-        return this;
-      }
-
-      private float sampleRate_ ;
-      /**
-       * <pre>
-       * Sample rate of the audio in Hz.
-       * </pre>
-       *
-       * <code>float sample_rate = 1;</code>
-       */
-      public float getSampleRate() {
-        return sampleRate_;
-      }
-      /**
-       * <pre>
-       * Sample rate of the audio in Hz.
-       * </pre>
-       *
-       * <code>float sample_rate = 1;</code>
-       */
-      public Builder setSampleRate(float value) {
-        
-        sampleRate_ = value;
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * Sample rate of the audio in Hz.
-       * </pre>
-       *
-       * <code>float sample_rate = 1;</code>
-       */
-      public Builder clearSampleRate() {
-        
-        sampleRate_ = 0F;
-        onChanged();
-        return this;
-      }
-
-      private long numChannels_ ;
-      /**
-       * <pre>
-       * Number of channels of audio.
-       * </pre>
-       *
-       * <code>int64 num_channels = 2;</code>
-       */
-      public long getNumChannels() {
-        return numChannels_;
-      }
-      /**
-       * <pre>
-       * Number of channels of audio.
-       * </pre>
-       *
-       * <code>int64 num_channels = 2;</code>
-       */
-      public Builder setNumChannels(long value) {
-        
-        numChannels_ = value;
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * Number of channels of audio.
-       * </pre>
-       *
-       * <code>int64 num_channels = 2;</code>
-       */
-      public Builder clearNumChannels() {
-        
-        numChannels_ = 0L;
-        onChanged();
-        return this;
-      }
-
-      private long lengthFrames_ ;
-      /**
-       * <pre>
-       * Length of the audio in frames (samples per channel).
-       * </pre>
-       *
-       * <code>int64 length_frames = 3;</code>
-       */
-      public long getLengthFrames() {
-        return lengthFrames_;
-      }
-      /**
-       * <pre>
-       * Length of the audio in frames (samples per channel).
-       * </pre>
-       *
-       * <code>int64 length_frames = 3;</code>
-       */
-      public Builder setLengthFrames(long value) {
-        
-        lengthFrames_ = value;
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * Length of the audio in frames (samples per channel).
-       * </pre>
-       *
-       * <code>int64 length_frames = 3;</code>
-       */
-      public Builder clearLengthFrames() {
-        
-        lengthFrames_ = 0L;
-        onChanged();
-        return this;
-      }
-
-      private com.google.protobuf.ByteString encodedAudioString_ = com.google.protobuf.ByteString.EMPTY;
-      /**
-       * <pre>
-       * Encoded audio data and its associated RFC 2045 content type (e.g.
-       * "audio/wav").
-       * </pre>
-       *
-       * <code>bytes encoded_audio_string = 4;</code>
-       */
-      public com.google.protobuf.ByteString getEncodedAudioString() {
-        return encodedAudioString_;
-      }
-      /**
-       * <pre>
-       * Encoded audio data and its associated RFC 2045 content type (e.g.
-       * "audio/wav").
-       * </pre>
-       *
-       * <code>bytes encoded_audio_string = 4;</code>
-       */
-      public Builder setEncodedAudioString(com.google.protobuf.ByteString value) {
-        if (value == null) {
-    throw new NullPointerException();
-  }
-  
-        encodedAudioString_ = value;
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * Encoded audio data and its associated RFC 2045 content type (e.g.
-       * "audio/wav").
-       * </pre>
-       *
-       * <code>bytes encoded_audio_string = 4;</code>
-       */
-      public Builder clearEncodedAudioString() {
-        
-        encodedAudioString_ = getDefaultInstance().getEncodedAudioString();
-        onChanged();
-        return this;
-      }
-
-      private java.lang.Object contentType_ = "";
-      /**
-       * <code>string content_type = 5;</code>
-       */
-      public java.lang.String getContentType() {
-        java.lang.Object ref = contentType_;
-        if (!(ref instanceof java.lang.String)) {
-          com.google.protobuf.ByteString bs =
-              (com.google.protobuf.ByteString) ref;
-          java.lang.String s = bs.toStringUtf8();
-          contentType_ = s;
-          return s;
-        } else {
-          return (java.lang.String) ref;
-        }
-      }
-      /**
-       * <code>string content_type = 5;</code>
-       */
-      public com.google.protobuf.ByteString
-          getContentTypeBytes() {
-        java.lang.Object ref = contentType_;
-        if (ref instanceof String) {
-          com.google.protobuf.ByteString b = 
-              com.google.protobuf.ByteString.copyFromUtf8(
-                  (java.lang.String) ref);
-          contentType_ = b;
-          return b;
-        } else {
-          return (com.google.protobuf.ByteString) ref;
-        }
-      }
-      /**
-       * <code>string content_type = 5;</code>
-       */
-      public Builder setContentType(
-          java.lang.String value) {
-        if (value == null) {
-    throw new NullPointerException();
-  }
-  
-        contentType_ = value;
-        onChanged();
-        return this;
-      }
-      /**
-       * <code>string content_type = 5;</code>
-       */
-      public Builder clearContentType() {
-        
-        contentType_ = getDefaultInstance().getContentType();
-        onChanged();
-        return this;
-      }
-      /**
-       * <code>string content_type = 5;</code>
-       */
-      public Builder setContentTypeBytes(
-          com.google.protobuf.ByteString value) {
-        if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-        
-        contentType_ = value;
-        onChanged();
-        return this;
-      }
-      @java.lang.Override
-      public final Builder setUnknownFields(
-          final com.google.protobuf.UnknownFieldSet unknownFields) {
-        return super.setUnknownFields(unknownFields);
-      }
-
-      @java.lang.Override
-      public final Builder mergeUnknownFields(
-          final com.google.protobuf.UnknownFieldSet unknownFields) {
-        return super.mergeUnknownFields(unknownFields);
-      }
-
-
-      // @@protoc_insertion_point(builder_scope:tensorflow.Summary.Audio)
-    }
-
-    // @@protoc_insertion_point(class_scope:tensorflow.Summary.Audio)
-    private static final org.tensorflow.proto.framework.Summary.Audio DEFAULT_INSTANCE;
-    static {
-      DEFAULT_INSTANCE = new org.tensorflow.proto.framework.Summary.Audio();
-    }
-
-    public static org.tensorflow.proto.framework.Summary.Audio getDefaultInstance() {
-      return DEFAULT_INSTANCE;
-    }
-
-    private static final com.google.protobuf.Parser<Audio>
-        PARSER = new com.google.protobuf.AbstractParser<Audio>() {
-      @java.lang.Override
-      public Audio parsePartialFrom(
-          com.google.protobuf.CodedInputStream input,
-          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-          throws com.google.protobuf.InvalidProtocolBufferException {
-        return new Audio(input, extensionRegistry);
-      }
-    };
-
-    public static com.google.protobuf.Parser<Audio> parser() {
-      return PARSER;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Parser<Audio> getParserForType() {
-      return PARSER;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.Summary.Audio getDefaultInstanceForType() {
-      return DEFAULT_INSTANCE;
-    }
-
-  }
-
-  public interface ValueOrBuilder extends
-      // @@protoc_insertion_point(interface_extends:tensorflow.Summary.Value)
-      com.google.protobuf.MessageOrBuilder {
-
-    /**
-     * <pre>
-     * This field is deprecated and will not be set.
-     * </pre>
-     *
-     * <code>string node_name = 7;</code>
-     */
-    java.lang.String getNodeName();
-    /**
-     * <pre>
-     * This field is deprecated and will not be set.
-     * </pre>
-     *
-     * <code>string node_name = 7;</code>
-     */
-    com.google.protobuf.ByteString
-        getNodeNameBytes();
-
-    /**
-     * <pre>
-     * Tag name for the data. Used by TensorBoard plugins to organize data. Tags
-     * are often organized by scope (which contains slashes to convey
-     * hierarchy). For example: foo/bar/0
-     * </pre>
-     *
-     * <code>string tag = 1;</code>
-     */
-    java.lang.String getTag();
-    /**
-     * <pre>
-     * Tag name for the data. Used by TensorBoard plugins to organize data. Tags
-     * are often organized by scope (which contains slashes to convey
-     * hierarchy). For example: foo/bar/0
-     * </pre>
-     *
-     * <code>string tag = 1;</code>
-     */
-    com.google.protobuf.ByteString
-        getTagBytes();
-
-    /**
-     * <pre>
-     * Contains metadata on the summary value such as which plugins may use it.
-     * Take note that many summary values may lack a metadata field. This is
-     * because the FileWriter only keeps a metadata object on the first summary
-     * value with a certain tag for each tag. TensorBoard then remembers which
-     * tags are associated with which plugins. This saves space.
-     * </pre>
-     *
-     * <code>.tensorflow.SummaryMetadata metadata = 9;</code>
-     */
-    boolean hasMetadata();
-    /**
-     * <pre>
-     * Contains metadata on the summary value such as which plugins may use it.
-     * Take note that many summary values may lack a metadata field. This is
-     * because the FileWriter only keeps a metadata object on the first summary
-     * value with a certain tag for each tag. TensorBoard then remembers which
-     * tags are associated with which plugins. This saves space.
-     * </pre>
-     *
-     * <code>.tensorflow.SummaryMetadata metadata = 9;</code>
-     */
-    org.tensorflow.proto.framework.SummaryMetadata getMetadata();
-    /**
-     * <pre>
-     * Contains metadata on the summary value such as which plugins may use it.
-     * Take note that many summary values may lack a metadata field. This is
-     * because the FileWriter only keeps a metadata object on the first summary
-     * value with a certain tag for each tag. TensorBoard then remembers which
-     * tags are associated with which plugins. This saves space.
-     * </pre>
-     *
-     * <code>.tensorflow.SummaryMetadata metadata = 9;</code>
-     */
-    org.tensorflow.proto.framework.SummaryMetadataOrBuilder getMetadataOrBuilder();
-
-    /**
-     * <code>float simple_value = 2;</code>
-     */
-    float getSimpleValue();
-
-    /**
-     * <code>bytes obsolete_old_style_histogram = 3;</code>
-     */
-    com.google.protobuf.ByteString getObsoleteOldStyleHistogram();
-
-    /**
-     * <code>.tensorflow.Summary.Image image = 4;</code>
-     */
-    boolean hasImage();
-    /**
-     * <code>.tensorflow.Summary.Image image = 4;</code>
-     */
-    org.tensorflow.proto.framework.Summary.Image getImage();
-    /**
-     * <code>.tensorflow.Summary.Image image = 4;</code>
-     */
-    org.tensorflow.proto.framework.Summary.ImageOrBuilder getImageOrBuilder();
-
-    /**
-     * <code>.tensorflow.HistogramProto histo = 5;</code>
-     */
-    boolean hasHisto();
-    /**
-     * <code>.tensorflow.HistogramProto histo = 5;</code>
-     */
-    org.tensorflow.proto.framework.HistogramProto getHisto();
-    /**
-     * <code>.tensorflow.HistogramProto histo = 5;</code>
-     */
-    org.tensorflow.proto.framework.HistogramProtoOrBuilder getHistoOrBuilder();
-
-    /**
-     * <code>.tensorflow.Summary.Audio audio = 6;</code>
-     */
-    boolean hasAudio();
-    /**
-     * <code>.tensorflow.Summary.Audio audio = 6;</code>
-     */
-    org.tensorflow.proto.framework.Summary.Audio getAudio();
-    /**
-     * <code>.tensorflow.Summary.Audio audio = 6;</code>
-     */
-    org.tensorflow.proto.framework.Summary.AudioOrBuilder getAudioOrBuilder();
-
-    /**
-     * <code>.tensorflow.TensorProto tensor = 8;</code>
-     */
-    boolean hasTensor();
-    /**
-     * <code>.tensorflow.TensorProto tensor = 8;</code>
-     */
-    org.tensorflow.proto.framework.TensorProto getTensor();
-    /**
-     * <code>.tensorflow.TensorProto tensor = 8;</code>
-     */
-    org.tensorflow.proto.framework.TensorProtoOrBuilder getTensorOrBuilder();
-
-    public org.tensorflow.proto.framework.Summary.Value.ValueCase getValueCase();
-  }
-  /**
-   * Protobuf type {@code tensorflow.Summary.Value}
-   */
-  public  static final class Value extends
-      com.google.protobuf.GeneratedMessageV3 implements
-      // @@protoc_insertion_point(message_implements:tensorflow.Summary.Value)
-      ValueOrBuilder {
-  private static final long serialVersionUID = 0L;
-    // Use Value.newBuilder() to construct.
-    private Value(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-      super(builder);
-    }
-    private Value() {
-      nodeName_ = "";
-      tag_ = "";
-    }
-
-    @java.lang.Override
-    @SuppressWarnings({"unused"})
-    protected java.lang.Object newInstance(
-        UnusedPrivateParameter unused) {
-      return new Value();
-    }
-
-    @java.lang.Override
-    public final com.google.protobuf.UnknownFieldSet
-    getUnknownFields() {
-      return this.unknownFields;
-    }
-    private Value(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      this();
-      if (extensionRegistry == null) {
-        throw new java.lang.NullPointerException();
-      }
-      com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-          com.google.protobuf.UnknownFieldSet.newBuilder();
-      try {
-        boolean done = false;
-        while (!done) {
-          int tag = input.readTag();
-          switch (tag) {
-            case 0:
-              done = true;
-              break;
-            case 10: {
-              java.lang.String s = input.readStringRequireUtf8();
-
-              tag_ = s;
-              break;
-            }
-            case 21: {
-              valueCase_ = 2;
-              value_ = input.readFloat();
-              break;
-            }
-            case 26: {
-              valueCase_ = 3;
-              value_ = input.readBytes();
-              break;
-            }
-            case 34: {
-              org.tensorflow.proto.framework.Summary.Image.Builder subBuilder = null;
-              if (valueCase_ == 4) {
-                subBuilder = ((org.tensorflow.proto.framework.Summary.Image) value_).toBuilder();
-              }
-              value_ =
-                  input.readMessage(org.tensorflow.proto.framework.Summary.Image.parser(), extensionRegistry);
-              if (subBuilder != null) {
-                subBuilder.mergeFrom((org.tensorflow.proto.framework.Summary.Image) value_);
-                value_ = subBuilder.buildPartial();
-              }
-              valueCase_ = 4;
-              break;
-            }
-            case 42: {
-              org.tensorflow.proto.framework.HistogramProto.Builder subBuilder = null;
-              if (valueCase_ == 5) {
-                subBuilder = ((org.tensorflow.proto.framework.HistogramProto) value_).toBuilder();
-              }
-              value_ =
-                  input.readMessage(org.tensorflow.proto.framework.HistogramProto.parser(), extensionRegistry);
-              if (subBuilder != null) {
-                subBuilder.mergeFrom((org.tensorflow.proto.framework.HistogramProto) value_);
-                value_ = subBuilder.buildPartial();
-              }
-              valueCase_ = 5;
-              break;
-            }
-            case 50: {
-              org.tensorflow.proto.framework.Summary.Audio.Builder subBuilder = null;
-              if (valueCase_ == 6) {
-                subBuilder = ((org.tensorflow.proto.framework.Summary.Audio) value_).toBuilder();
-              }
-              value_ =
-                  input.readMessage(org.tensorflow.proto.framework.Summary.Audio.parser(), extensionRegistry);
-              if (subBuilder != null) {
-                subBuilder.mergeFrom((org.tensorflow.proto.framework.Summary.Audio) value_);
-                value_ = subBuilder.buildPartial();
-              }
-              valueCase_ = 6;
-              break;
-            }
-            case 58: {
-              java.lang.String s = input.readStringRequireUtf8();
-
-              nodeName_ = s;
-              break;
-            }
-            case 66: {
-              org.tensorflow.proto.framework.TensorProto.Builder subBuilder = null;
-              if (valueCase_ == 8) {
-                subBuilder = ((org.tensorflow.proto.framework.TensorProto) value_).toBuilder();
-              }
-              value_ =
-                  input.readMessage(org.tensorflow.proto.framework.TensorProto.parser(), extensionRegistry);
-              if (subBuilder != null) {
-                subBuilder.mergeFrom((org.tensorflow.proto.framework.TensorProto) value_);
-                value_ = subBuilder.buildPartial();
-              }
-              valueCase_ = 8;
-              break;
-            }
-            case 74: {
-              org.tensorflow.proto.framework.SummaryMetadata.Builder subBuilder = null;
-              if (metadata_ != null) {
-                subBuilder = metadata_.toBuilder();
-              }
-              metadata_ = input.readMessage(org.tensorflow.proto.framework.SummaryMetadata.parser(), extensionRegistry);
-              if (subBuilder != null) {
-                subBuilder.mergeFrom(metadata_);
-                metadata_ = subBuilder.buildPartial();
-              }
-
-              break;
-            }
-            default: {
-              if (!parseUnknownField(
-                  input, unknownFields, extensionRegistry, tag)) {
-                done = true;
-              }
-              break;
-            }
-          }
-        }
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        throw e.setUnfinishedMessage(this);
-      } catch (java.io.IOException e) {
-        throw new com.google.protobuf.InvalidProtocolBufferException(
-            e).setUnfinishedMessage(this);
-      } finally {
-        this.unknownFields = unknownFields.build();
-        makeExtensionsImmutable();
-      }
-    }
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.SummaryProtos.internal_static_tensorflow_Summary_Value_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.SummaryProtos.internal_static_tensorflow_Summary_Value_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.Summary.Value.class, org.tensorflow.proto.framework.Summary.Value.Builder.class);
-    }
-
-    private int valueCase_ = 0;
-    private java.lang.Object value_;
-    public enum ValueCase
-        implements com.google.protobuf.Internal.EnumLite {
-      SIMPLE_VALUE(2),
-      OBSOLETE_OLD_STYLE_HISTOGRAM(3),
-      IMAGE(4),
-      HISTO(5),
-      AUDIO(6),
-      TENSOR(8),
-      VALUE_NOT_SET(0);
-      private final int value;
-      private ValueCase(int value) {
-        this.value = value;
-      }
-      /**
-       * @deprecated Use {@link #forNumber(int)} instead.
-       */
-      @java.lang.Deprecated
-      public static ValueCase valueOf(int value) {
-        return forNumber(value);
-      }
-
-      public static ValueCase forNumber(int value) {
-        switch (value) {
-          case 2: return SIMPLE_VALUE;
-          case 3: return OBSOLETE_OLD_STYLE_HISTOGRAM;
-          case 4: return IMAGE;
-          case 5: return HISTO;
-          case 6: return AUDIO;
-          case 8: return TENSOR;
-          case 0: return VALUE_NOT_SET;
-          default: return null;
-        }
-      }
-      public int getNumber() {
-        return this.value;
-      }
-    };
-
-    public ValueCase
-    getValueCase() {
-      return ValueCase.forNumber(
-          valueCase_);
-    }
-
-    public static final int NODE_NAME_FIELD_NUMBER = 7;
-    private volatile java.lang.Object nodeName_;
-    /**
-     * <pre>
-     * This field is deprecated and will not be set.
-     * </pre>
-     *
-     * <code>string node_name = 7;</code>
-     */
-    public java.lang.String getNodeName() {
-      java.lang.Object ref = nodeName_;
-      if (ref instanceof java.lang.String) {
-        return (java.lang.String) ref;
-      } else {
-        com.google.protobuf.ByteString bs = 
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        nodeName_ = s;
-        return s;
-      }
-    }
-    /**
-     * <pre>
-     * This field is deprecated and will not be set.
-     * </pre>
-     *
-     * <code>string node_name = 7;</code>
-     */
-    public com.google.protobuf.ByteString
-        getNodeNameBytes() {
-      java.lang.Object ref = nodeName_;
-      if (ref instanceof java.lang.String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        nodeName_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-
-    public static final int TAG_FIELD_NUMBER = 1;
-    private volatile java.lang.Object tag_;
-    /**
-     * <pre>
-     * Tag name for the data. Used by TensorBoard plugins to organize data. Tags
-     * are often organized by scope (which contains slashes to convey
-     * hierarchy). For example: foo/bar/0
-     * </pre>
-     *
-     * <code>string tag = 1;</code>
-     */
-    public java.lang.String getTag() {
-      java.lang.Object ref = tag_;
-      if (ref instanceof java.lang.String) {
-        return (java.lang.String) ref;
-      } else {
-        com.google.protobuf.ByteString bs = 
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        tag_ = s;
-        return s;
-      }
-    }
-    /**
-     * <pre>
-     * Tag name for the data. Used by TensorBoard plugins to organize data. Tags
-     * are often organized by scope (which contains slashes to convey
-     * hierarchy). For example: foo/bar/0
-     * </pre>
-     *
-     * <code>string tag = 1;</code>
-     */
-    public com.google.protobuf.ByteString
-        getTagBytes() {
-      java.lang.Object ref = tag_;
-      if (ref instanceof java.lang.String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        tag_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-
-    public static final int METADATA_FIELD_NUMBER = 9;
-    private org.tensorflow.proto.framework.SummaryMetadata metadata_;
-    /**
-     * <pre>
-     * Contains metadata on the summary value such as which plugins may use it.
-     * Take note that many summary values may lack a metadata field. This is
-     * because the FileWriter only keeps a metadata object on the first summary
-     * value with a certain tag for each tag. TensorBoard then remembers which
-     * tags are associated with which plugins. This saves space.
-     * </pre>
-     *
-     * <code>.tensorflow.SummaryMetadata metadata = 9;</code>
-     */
-    public boolean hasMetadata() {
-      return metadata_ != null;
-    }
-    /**
-     * <pre>
-     * Contains metadata on the summary value such as which plugins may use it.
-     * Take note that many summary values may lack a metadata field. This is
-     * because the FileWriter only keeps a metadata object on the first summary
-     * value with a certain tag for each tag. TensorBoard then remembers which
-     * tags are associated with which plugins. This saves space.
-     * </pre>
-     *
-     * <code>.tensorflow.SummaryMetadata metadata = 9;</code>
-     */
-    public org.tensorflow.proto.framework.SummaryMetadata getMetadata() {
-      return metadata_ == null ? org.tensorflow.proto.framework.SummaryMetadata.getDefaultInstance() : metadata_;
-    }
-    /**
-     * <pre>
-     * Contains metadata on the summary value such as which plugins may use it.
-     * Take note that many summary values may lack a metadata field. This is
-     * because the FileWriter only keeps a metadata object on the first summary
-     * value with a certain tag for each tag. TensorBoard then remembers which
-     * tags are associated with which plugins. This saves space.
-     * </pre>
-     *
-     * <code>.tensorflow.SummaryMetadata metadata = 9;</code>
-     */
-    public org.tensorflow.proto.framework.SummaryMetadataOrBuilder getMetadataOrBuilder() {
-      return getMetadata();
-    }
-
-    public static final int SIMPLE_VALUE_FIELD_NUMBER = 2;
-    /**
-     * <code>float simple_value = 2;</code>
-     */
-    public float getSimpleValue() {
-      if (valueCase_ == 2) {
-        return (java.lang.Float) value_;
-      }
-      return 0F;
-    }
-
-    public static final int OBSOLETE_OLD_STYLE_HISTOGRAM_FIELD_NUMBER = 3;
-    /**
-     * <code>bytes obsolete_old_style_histogram = 3;</code>
-     */
-    public com.google.protobuf.ByteString getObsoleteOldStyleHistogram() {
-      if (valueCase_ == 3) {
-        return (com.google.protobuf.ByteString) value_;
-      }
-      return com.google.protobuf.ByteString.EMPTY;
-    }
-
-    public static final int IMAGE_FIELD_NUMBER = 4;
-    /**
-     * <code>.tensorflow.Summary.Image image = 4;</code>
-     */
-    public boolean hasImage() {
-      return valueCase_ == 4;
-    }
-    /**
-     * <code>.tensorflow.Summary.Image image = 4;</code>
-     */
-    public org.tensorflow.proto.framework.Summary.Image getImage() {
-      if (valueCase_ == 4) {
-         return (org.tensorflow.proto.framework.Summary.Image) value_;
-      }
-      return org.tensorflow.proto.framework.Summary.Image.getDefaultInstance();
-    }
-    /**
-     * <code>.tensorflow.Summary.Image image = 4;</code>
-     */
-    public org.tensorflow.proto.framework.Summary.ImageOrBuilder getImageOrBuilder() {
-      if (valueCase_ == 4) {
-         return (org.tensorflow.proto.framework.Summary.Image) value_;
-      }
-      return org.tensorflow.proto.framework.Summary.Image.getDefaultInstance();
-    }
-
-    public static final int HISTO_FIELD_NUMBER = 5;
-    /**
-     * <code>.tensorflow.HistogramProto histo = 5;</code>
-     */
-    public boolean hasHisto() {
-      return valueCase_ == 5;
-    }
-    /**
-     * <code>.tensorflow.HistogramProto histo = 5;</code>
-     */
-    public org.tensorflow.proto.framework.HistogramProto getHisto() {
-      if (valueCase_ == 5) {
-         return (org.tensorflow.proto.framework.HistogramProto) value_;
-      }
-      return org.tensorflow.proto.framework.HistogramProto.getDefaultInstance();
-    }
-    /**
-     * <code>.tensorflow.HistogramProto histo = 5;</code>
-     */
-    public org.tensorflow.proto.framework.HistogramProtoOrBuilder getHistoOrBuilder() {
-      if (valueCase_ == 5) {
-         return (org.tensorflow.proto.framework.HistogramProto) value_;
-      }
-      return org.tensorflow.proto.framework.HistogramProto.getDefaultInstance();
-    }
-
-    public static final int AUDIO_FIELD_NUMBER = 6;
-    /**
-     * <code>.tensorflow.Summary.Audio audio = 6;</code>
-     */
-    public boolean hasAudio() {
-      return valueCase_ == 6;
-    }
-    /**
-     * <code>.tensorflow.Summary.Audio audio = 6;</code>
-     */
-    public org.tensorflow.proto.framework.Summary.Audio getAudio() {
-      if (valueCase_ == 6) {
-         return (org.tensorflow.proto.framework.Summary.Audio) value_;
-      }
-      return org.tensorflow.proto.framework.Summary.Audio.getDefaultInstance();
-    }
-    /**
-     * <code>.tensorflow.Summary.Audio audio = 6;</code>
-     */
-    public org.tensorflow.proto.framework.Summary.AudioOrBuilder getAudioOrBuilder() {
-      if (valueCase_ == 6) {
-         return (org.tensorflow.proto.framework.Summary.Audio) value_;
-      }
-      return org.tensorflow.proto.framework.Summary.Audio.getDefaultInstance();
-    }
-
-    public static final int TENSOR_FIELD_NUMBER = 8;
-    /**
-     * <code>.tensorflow.TensorProto tensor = 8;</code>
-     */
-    public boolean hasTensor() {
-      return valueCase_ == 8;
-    }
-    /**
-     * <code>.tensorflow.TensorProto tensor = 8;</code>
-     */
-    public org.tensorflow.proto.framework.TensorProto getTensor() {
-      if (valueCase_ == 8) {
-         return (org.tensorflow.proto.framework.TensorProto) value_;
-      }
-      return org.tensorflow.proto.framework.TensorProto.getDefaultInstance();
-    }
-    /**
-     * <code>.tensorflow.TensorProto tensor = 8;</code>
-     */
-    public org.tensorflow.proto.framework.TensorProtoOrBuilder getTensorOrBuilder() {
-      if (valueCase_ == 8) {
-         return (org.tensorflow.proto.framework.TensorProto) value_;
-      }
-      return org.tensorflow.proto.framework.TensorProto.getDefaultInstance();
-    }
-
-    private byte memoizedIsInitialized = -1;
-    @java.lang.Override
-    public final boolean isInitialized() {
-      byte isInitialized = memoizedIsInitialized;
-      if (isInitialized == 1) return true;
-      if (isInitialized == 0) return false;
-
-      memoizedIsInitialized = 1;
-      return true;
-    }
-
-    @java.lang.Override
-    public void writeTo(com.google.protobuf.CodedOutputStream output)
-                        throws java.io.IOException {
-      if (!getTagBytes().isEmpty()) {
-        com.google.protobuf.GeneratedMessageV3.writeString(output, 1, tag_);
-      }
-      if (valueCase_ == 2) {
-        output.writeFloat(
-            2, (float)((java.lang.Float) value_));
-      }
-      if (valueCase_ == 3) {
-        output.writeBytes(
-            3, (com.google.protobuf.ByteString) value_);
-      }
-      if (valueCase_ == 4) {
-        output.writeMessage(4, (org.tensorflow.proto.framework.Summary.Image) value_);
-      }
-      if (valueCase_ == 5) {
-        output.writeMessage(5, (org.tensorflow.proto.framework.HistogramProto) value_);
-      }
-      if (valueCase_ == 6) {
-        output.writeMessage(6, (org.tensorflow.proto.framework.Summary.Audio) value_);
-      }
-      if (!getNodeNameBytes().isEmpty()) {
-        com.google.protobuf.GeneratedMessageV3.writeString(output, 7, nodeName_);
-      }
-      if (valueCase_ == 8) {
-        output.writeMessage(8, (org.tensorflow.proto.framework.TensorProto) value_);
-      }
-      if (metadata_ != null) {
-        output.writeMessage(9, getMetadata());
-      }
-      unknownFields.writeTo(output);
-    }
-
-    @java.lang.Override
-    public int getSerializedSize() {
-      int size = memoizedSize;
-      if (size != -1) return size;
-
-      size = 0;
-      if (!getTagBytes().isEmpty()) {
-        size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, tag_);
-      }
-      if (valueCase_ == 2) {
-        size += com.google.protobuf.CodedOutputStream
-          .computeFloatSize(
-              2, (float)((java.lang.Float) value_));
-      }
-      if (valueCase_ == 3) {
-        size += com.google.protobuf.CodedOutputStream
-          .computeBytesSize(
-              3, (com.google.protobuf.ByteString) value_);
-      }
-      if (valueCase_ == 4) {
-        size += com.google.protobuf.CodedOutputStream
-          .computeMessageSize(4, (org.tensorflow.proto.framework.Summary.Image) value_);
-      }
-      if (valueCase_ == 5) {
-        size += com.google.protobuf.CodedOutputStream
-          .computeMessageSize(5, (org.tensorflow.proto.framework.HistogramProto) value_);
-      }
-      if (valueCase_ == 6) {
-        size += com.google.protobuf.CodedOutputStream
-          .computeMessageSize(6, (org.tensorflow.proto.framework.Summary.Audio) value_);
-      }
-      if (!getNodeNameBytes().isEmpty()) {
-        size += com.google.protobuf.GeneratedMessageV3.computeStringSize(7, nodeName_);
-      }
-      if (valueCase_ == 8) {
-        size += com.google.protobuf.CodedOutputStream
-          .computeMessageSize(8, (org.tensorflow.proto.framework.TensorProto) value_);
-      }
-      if (metadata_ != null) {
-        size += com.google.protobuf.CodedOutputStream
-          .computeMessageSize(9, getMetadata());
-      }
-      size += unknownFields.getSerializedSize();
-      memoizedSize = size;
-      return size;
-    }
-
-    @java.lang.Override
-    public boolean equals(final java.lang.Object obj) {
-      if (obj == this) {
-       return true;
-      }
-      if (!(obj instanceof org.tensorflow.proto.framework.Summary.Value)) {
-        return super.equals(obj);
-      }
-      org.tensorflow.proto.framework.Summary.Value other = (org.tensorflow.proto.framework.Summary.Value) obj;
-
-      if (!getNodeName()
-          .equals(other.getNodeName())) return false;
-      if (!getTag()
-          .equals(other.getTag())) return false;
-      if (hasMetadata() != other.hasMetadata()) return false;
-      if (hasMetadata()) {
-        if (!getMetadata()
-            .equals(other.getMetadata())) return false;
-      }
-      if (!getValueCase().equals(other.getValueCase())) return false;
-      switch (valueCase_) {
-        case 2:
-          if (java.lang.Float.floatToIntBits(getSimpleValue())
-              != java.lang.Float.floatToIntBits(
-                  other.getSimpleValue())) return false;
-          break;
-        case 3:
-          if (!getObsoleteOldStyleHistogram()
-              .equals(other.getObsoleteOldStyleHistogram())) return false;
-          break;
-        case 4:
-          if (!getImage()
-              .equals(other.getImage())) return false;
-          break;
-        case 5:
-          if (!getHisto()
-              .equals(other.getHisto())) return false;
-          break;
-        case 6:
-          if (!getAudio()
-              .equals(other.getAudio())) return false;
-          break;
-        case 8:
-          if (!getTensor()
-              .equals(other.getTensor())) return false;
-          break;
-        case 0:
-        default:
-      }
-      if (!unknownFields.equals(other.unknownFields)) return false;
-      return true;
-    }
-
-    @java.lang.Override
-    public int hashCode() {
-      if (memoizedHashCode != 0) {
-        return memoizedHashCode;
-      }
-      int hash = 41;
-      hash = (19 * hash) + getDescriptor().hashCode();
-      hash = (37 * hash) + NODE_NAME_FIELD_NUMBER;
-      hash = (53 * hash) + getNodeName().hashCode();
-      hash = (37 * hash) + TAG_FIELD_NUMBER;
-      hash = (53 * hash) + getTag().hashCode();
-      if (hasMetadata()) {
-        hash = (37 * hash) + METADATA_FIELD_NUMBER;
-        hash = (53 * hash) + getMetadata().hashCode();
-      }
-      switch (valueCase_) {
-        case 2:
-          hash = (37 * hash) + SIMPLE_VALUE_FIELD_NUMBER;
-          hash = (53 * hash) + java.lang.Float.floatToIntBits(
-              getSimpleValue());
-          break;
-        case 3:
-          hash = (37 * hash) + OBSOLETE_OLD_STYLE_HISTOGRAM_FIELD_NUMBER;
-          hash = (53 * hash) + getObsoleteOldStyleHistogram().hashCode();
-          break;
-        case 4:
-          hash = (37 * hash) + IMAGE_FIELD_NUMBER;
-          hash = (53 * hash) + getImage().hashCode();
-          break;
-        case 5:
-          hash = (37 * hash) + HISTO_FIELD_NUMBER;
-          hash = (53 * hash) + getHisto().hashCode();
-          break;
-        case 6:
-          hash = (37 * hash) + AUDIO_FIELD_NUMBER;
-          hash = (53 * hash) + getAudio().hashCode();
-          break;
-        case 8:
-          hash = (37 * hash) + TENSOR_FIELD_NUMBER;
-          hash = (53 * hash) + getTensor().hashCode();
-          break;
-        case 0:
-        default:
-      }
-      hash = (29 * hash) + unknownFields.hashCode();
-      memoizedHashCode = hash;
-      return hash;
-    }
-
-    public static org.tensorflow.proto.framework.Summary.Value parseFrom(
-        java.nio.ByteBuffer data)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data);
-    }
-    public static org.tensorflow.proto.framework.Summary.Value parseFrom(
-        java.nio.ByteBuffer data,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.Summary.Value parseFrom(
-        com.google.protobuf.ByteString data)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data);
-    }
-    public static org.tensorflow.proto.framework.Summary.Value parseFrom(
-        com.google.protobuf.ByteString data,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.Summary.Value parseFrom(byte[] data)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data);
-    }
-    public static org.tensorflow.proto.framework.Summary.Value parseFrom(
-        byte[] data,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.Summary.Value parseFrom(java.io.InputStream input)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input);
-    }
-    public static org.tensorflow.proto.framework.Summary.Value parseFrom(
-        java.io.InputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.Summary.Value parseDelimitedFrom(java.io.InputStream input)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseDelimitedWithIOException(PARSER, input);
-    }
-    public static org.tensorflow.proto.framework.Summary.Value parseDelimitedFrom(
-        java.io.InputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.Summary.Value parseFrom(
-        com.google.protobuf.CodedInputStream input)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input);
-    }
-    public static org.tensorflow.proto.framework.Summary.Value parseFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input, extensionRegistry);
-    }
-
-    @java.lang.Override
-    public Builder newBuilderForType() { return newBuilder(); }
-    public static Builder newBuilder() {
-      return DEFAULT_INSTANCE.toBuilder();
-    }
-    public static Builder newBuilder(org.tensorflow.proto.framework.Summary.Value prototype) {
-      return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-    }
-    @java.lang.Override
-    public Builder toBuilder() {
-      return this == DEFAULT_INSTANCE
-          ? new Builder() : new Builder().mergeFrom(this);
-    }
-
-    @java.lang.Override
-    protected Builder newBuilderForType(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      Builder builder = new Builder(parent);
-      return builder;
-    }
-    /**
-     * Protobuf type {@code tensorflow.Summary.Value}
-     */
-    public static final class Builder extends
-        com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-        // @@protoc_insertion_point(builder_implements:tensorflow.Summary.Value)
-        org.tensorflow.proto.framework.Summary.ValueOrBuilder {
-      public static final com.google.protobuf.Descriptors.Descriptor
-          getDescriptor() {
-        return org.tensorflow.proto.framework.SummaryProtos.internal_static_tensorflow_Summary_Value_descriptor;
-      }
-
-      @java.lang.Override
-      protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-          internalGetFieldAccessorTable() {
-        return org.tensorflow.proto.framework.SummaryProtos.internal_static_tensorflow_Summary_Value_fieldAccessorTable
-            .ensureFieldAccessorsInitialized(
-                org.tensorflow.proto.framework.Summary.Value.class, org.tensorflow.proto.framework.Summary.Value.Builder.class);
-      }
-
-      // Construct using org.tensorflow.proto.framework.Summary.Value.newBuilder()
-      private Builder() {
-        maybeForceBuilderInitialization();
-      }
-
-      private Builder(
-          com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-        super(parent);
-        maybeForceBuilderInitialization();
-      }
-      private void maybeForceBuilderInitialization() {
-        if (com.google.protobuf.GeneratedMessageV3
-                .alwaysUseFieldBuilders) {
-        }
-      }
-      @java.lang.Override
-      public Builder clear() {
-        super.clear();
-        nodeName_ = "";
-
-        tag_ = "";
-
-        if (metadataBuilder_ == null) {
-          metadata_ = null;
-        } else {
-          metadata_ = null;
-          metadataBuilder_ = null;
-        }
-        valueCase_ = 0;
-        value_ = null;
-        return this;
-      }
-
-      @java.lang.Override
-      public com.google.protobuf.Descriptors.Descriptor
-          getDescriptorForType() {
-        return org.tensorflow.proto.framework.SummaryProtos.internal_static_tensorflow_Summary_Value_descriptor;
-      }
-
-      @java.lang.Override
-      public org.tensorflow.proto.framework.Summary.Value getDefaultInstanceForType() {
-        return org.tensorflow.proto.framework.Summary.Value.getDefaultInstance();
-      }
-
-      @java.lang.Override
-      public org.tensorflow.proto.framework.Summary.Value build() {
-        org.tensorflow.proto.framework.Summary.Value result = buildPartial();
-        if (!result.isInitialized()) {
-          throw newUninitializedMessageException(result);
-        }
-        return result;
-      }
-
-      @java.lang.Override
-      public org.tensorflow.proto.framework.Summary.Value buildPartial() {
-        org.tensorflow.proto.framework.Summary.Value result = new org.tensorflow.proto.framework.Summary.Value(this);
-        result.nodeName_ = nodeName_;
-        result.tag_ = tag_;
-        if (metadataBuilder_ == null) {
-          result.metadata_ = metadata_;
-        } else {
-          result.metadata_ = metadataBuilder_.build();
-        }
-        if (valueCase_ == 2) {
-          result.value_ = value_;
-        }
-        if (valueCase_ == 3) {
-          result.value_ = value_;
-        }
-        if (valueCase_ == 4) {
-          if (imageBuilder_ == null) {
-            result.value_ = value_;
-          } else {
-            result.value_ = imageBuilder_.build();
-          }
-        }
-        if (valueCase_ == 5) {
-          if (histoBuilder_ == null) {
-            result.value_ = value_;
-          } else {
-            result.value_ = histoBuilder_.build();
-          }
-        }
-        if (valueCase_ == 6) {
-          if (audioBuilder_ == null) {
-            result.value_ = value_;
-          } else {
-            result.value_ = audioBuilder_.build();
-          }
-        }
-        if (valueCase_ == 8) {
-          if (tensorBuilder_ == null) {
-            result.value_ = value_;
-          } else {
-            result.value_ = tensorBuilder_.build();
-          }
-        }
-        result.valueCase_ = valueCase_;
-        onBuilt();
-        return result;
-      }
-
-      @java.lang.Override
-      public Builder clone() {
-        return super.clone();
-      }
-      @java.lang.Override
-      public Builder setField(
-          com.google.protobuf.Descriptors.FieldDescriptor field,
-          java.lang.Object value) {
-        return super.setField(field, value);
-      }
-      @java.lang.Override
-      public Builder clearField(
-          com.google.protobuf.Descriptors.FieldDescriptor field) {
-        return super.clearField(field);
-      }
-      @java.lang.Override
-      public Builder clearOneof(
-          com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-        return super.clearOneof(oneof);
-      }
-      @java.lang.Override
-      public Builder setRepeatedField(
-          com.google.protobuf.Descriptors.FieldDescriptor field,
-          int index, java.lang.Object value) {
-        return super.setRepeatedField(field, index, value);
-      }
-      @java.lang.Override
-      public Builder addRepeatedField(
-          com.google.protobuf.Descriptors.FieldDescriptor field,
-          java.lang.Object value) {
-        return super.addRepeatedField(field, value);
-      }
-      @java.lang.Override
-      public Builder mergeFrom(com.google.protobuf.Message other) {
-        if (other instanceof org.tensorflow.proto.framework.Summary.Value) {
-          return mergeFrom((org.tensorflow.proto.framework.Summary.Value)other);
-        } else {
-          super.mergeFrom(other);
-          return this;
-        }
-      }
-
-      public Builder mergeFrom(org.tensorflow.proto.framework.Summary.Value other) {
-        if (other == org.tensorflow.proto.framework.Summary.Value.getDefaultInstance()) return this;
-        if (!other.getNodeName().isEmpty()) {
-          nodeName_ = other.nodeName_;
-          onChanged();
-        }
-        if (!other.getTag().isEmpty()) {
-          tag_ = other.tag_;
-          onChanged();
-        }
-        if (other.hasMetadata()) {
-          mergeMetadata(other.getMetadata());
-        }
-        switch (other.getValueCase()) {
-          case SIMPLE_VALUE: {
-            setSimpleValue(other.getSimpleValue());
-            break;
-          }
-          case OBSOLETE_OLD_STYLE_HISTOGRAM: {
-            setObsoleteOldStyleHistogram(other.getObsoleteOldStyleHistogram());
-            break;
-          }
-          case IMAGE: {
-            mergeImage(other.getImage());
-            break;
-          }
-          case HISTO: {
-            mergeHisto(other.getHisto());
-            break;
-          }
-          case AUDIO: {
-            mergeAudio(other.getAudio());
-            break;
-          }
-          case TENSOR: {
-            mergeTensor(other.getTensor());
-            break;
-          }
-          case VALUE_NOT_SET: {
-            break;
-          }
-        }
-        this.mergeUnknownFields(other.unknownFields);
-        onChanged();
-        return this;
-      }
-
-      @java.lang.Override
-      public final boolean isInitialized() {
-        return true;
-      }
-
-      @java.lang.Override
-      public Builder mergeFrom(
-          com.google.protobuf.CodedInputStream input,
-          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-          throws java.io.IOException {
-        org.tensorflow.proto.framework.Summary.Value parsedMessage = null;
-        try {
-          parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-        } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-          parsedMessage = (org.tensorflow.proto.framework.Summary.Value) e.getUnfinishedMessage();
-          throw e.unwrapIOException();
-        } finally {
-          if (parsedMessage != null) {
-            mergeFrom(parsedMessage);
-          }
-        }
-        return this;
-      }
-      private int valueCase_ = 0;
-      private java.lang.Object value_;
-      public ValueCase
-          getValueCase() {
-        return ValueCase.forNumber(
-            valueCase_);
-      }
-
-      public Builder clearValue() {
-        valueCase_ = 0;
-        value_ = null;
-        onChanged();
-        return this;
-      }
-
-
-      private java.lang.Object nodeName_ = "";
-      /**
-       * <pre>
-       * This field is deprecated and will not be set.
-       * </pre>
-       *
-       * <code>string node_name = 7;</code>
-       */
-      public java.lang.String getNodeName() {
-        java.lang.Object ref = nodeName_;
-        if (!(ref instanceof java.lang.String)) {
-          com.google.protobuf.ByteString bs =
-              (com.google.protobuf.ByteString) ref;
-          java.lang.String s = bs.toStringUtf8();
-          nodeName_ = s;
-          return s;
-        } else {
-          return (java.lang.String) ref;
-        }
-      }
-      /**
-       * <pre>
-       * This field is deprecated and will not be set.
-       * </pre>
-       *
-       * <code>string node_name = 7;</code>
-       */
-      public com.google.protobuf.ByteString
-          getNodeNameBytes() {
-        java.lang.Object ref = nodeName_;
-        if (ref instanceof String) {
-          com.google.protobuf.ByteString b = 
-              com.google.protobuf.ByteString.copyFromUtf8(
-                  (java.lang.String) ref);
-          nodeName_ = b;
-          return b;
-        } else {
-          return (com.google.protobuf.ByteString) ref;
-        }
-      }
-      /**
-       * <pre>
-       * This field is deprecated and will not be set.
-       * </pre>
-       *
-       * <code>string node_name = 7;</code>
-       */
-      public Builder setNodeName(
-          java.lang.String value) {
-        if (value == null) {
-    throw new NullPointerException();
-  }
-  
-        nodeName_ = value;
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * This field is deprecated and will not be set.
-       * </pre>
-       *
-       * <code>string node_name = 7;</code>
-       */
-      public Builder clearNodeName() {
-        
-        nodeName_ = getDefaultInstance().getNodeName();
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * This field is deprecated and will not be set.
-       * </pre>
-       *
-       * <code>string node_name = 7;</code>
-       */
-      public Builder setNodeNameBytes(
-          com.google.protobuf.ByteString value) {
-        if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-        
-        nodeName_ = value;
-        onChanged();
-        return this;
-      }
-
-      private java.lang.Object tag_ = "";
-      /**
-       * <pre>
-       * Tag name for the data. Used by TensorBoard plugins to organize data. Tags
-       * are often organized by scope (which contains slashes to convey
-       * hierarchy). For example: foo/bar/0
-       * </pre>
-       *
-       * <code>string tag = 1;</code>
-       */
-      public java.lang.String getTag() {
-        java.lang.Object ref = tag_;
-        if (!(ref instanceof java.lang.String)) {
-          com.google.protobuf.ByteString bs =
-              (com.google.protobuf.ByteString) ref;
-          java.lang.String s = bs.toStringUtf8();
-          tag_ = s;
-          return s;
-        } else {
-          return (java.lang.String) ref;
-        }
-      }
-      /**
-       * <pre>
-       * Tag name for the data. Used by TensorBoard plugins to organize data. Tags
-       * are often organized by scope (which contains slashes to convey
-       * hierarchy). For example: foo/bar/0
-       * </pre>
-       *
-       * <code>string tag = 1;</code>
-       */
-      public com.google.protobuf.ByteString
-          getTagBytes() {
-        java.lang.Object ref = tag_;
-        if (ref instanceof String) {
-          com.google.protobuf.ByteString b = 
-              com.google.protobuf.ByteString.copyFromUtf8(
-                  (java.lang.String) ref);
-          tag_ = b;
-          return b;
-        } else {
-          return (com.google.protobuf.ByteString) ref;
-        }
-      }
-      /**
-       * <pre>
-       * Tag name for the data. Used by TensorBoard plugins to organize data. Tags
-       * are often organized by scope (which contains slashes to convey
-       * hierarchy). For example: foo/bar/0
-       * </pre>
-       *
-       * <code>string tag = 1;</code>
-       */
-      public Builder setTag(
-          java.lang.String value) {
-        if (value == null) {
-    throw new NullPointerException();
-  }
-  
-        tag_ = value;
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * Tag name for the data. Used by TensorBoard plugins to organize data. Tags
-       * are often organized by scope (which contains slashes to convey
-       * hierarchy). For example: foo/bar/0
-       * </pre>
-       *
-       * <code>string tag = 1;</code>
-       */
-      public Builder clearTag() {
-        
-        tag_ = getDefaultInstance().getTag();
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * Tag name for the data. Used by TensorBoard plugins to organize data. Tags
-       * are often organized by scope (which contains slashes to convey
-       * hierarchy). For example: foo/bar/0
-       * </pre>
-       *
-       * <code>string tag = 1;</code>
-       */
-      public Builder setTagBytes(
-          com.google.protobuf.ByteString value) {
-        if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-        
-        tag_ = value;
-        onChanged();
-        return this;
-      }
-
-      private org.tensorflow.proto.framework.SummaryMetadata metadata_;
-      private com.google.protobuf.SingleFieldBuilderV3<
-          org.tensorflow.proto.framework.SummaryMetadata, org.tensorflow.proto.framework.SummaryMetadata.Builder, org.tensorflow.proto.framework.SummaryMetadataOrBuilder> metadataBuilder_;
-      /**
-       * <pre>
-       * Contains metadata on the summary value such as which plugins may use it.
-       * Take note that many summary values may lack a metadata field. This is
-       * because the FileWriter only keeps a metadata object on the first summary
-       * value with a certain tag for each tag. TensorBoard then remembers which
-       * tags are associated with which plugins. This saves space.
-       * </pre>
-       *
-       * <code>.tensorflow.SummaryMetadata metadata = 9;</code>
-       */
-      public boolean hasMetadata() {
-        return metadataBuilder_ != null || metadata_ != null;
-      }
-      /**
-       * <pre>
-       * Contains metadata on the summary value such as which plugins may use it.
-       * Take note that many summary values may lack a metadata field. This is
-       * because the FileWriter only keeps a metadata object on the first summary
-       * value with a certain tag for each tag. TensorBoard then remembers which
-       * tags are associated with which plugins. This saves space.
-       * </pre>
-       *
-       * <code>.tensorflow.SummaryMetadata metadata = 9;</code>
-       */
-      public org.tensorflow.proto.framework.SummaryMetadata getMetadata() {
-        if (metadataBuilder_ == null) {
-          return metadata_ == null ? org.tensorflow.proto.framework.SummaryMetadata.getDefaultInstance() : metadata_;
-        } else {
-          return metadataBuilder_.getMessage();
-        }
-      }
-      /**
-       * <pre>
-       * Contains metadata on the summary value such as which plugins may use it.
-       * Take note that many summary values may lack a metadata field. This is
-       * because the FileWriter only keeps a metadata object on the first summary
-       * value with a certain tag for each tag. TensorBoard then remembers which
-       * tags are associated with which plugins. This saves space.
-       * </pre>
-       *
-       * <code>.tensorflow.SummaryMetadata metadata = 9;</code>
-       */
-      public Builder setMetadata(org.tensorflow.proto.framework.SummaryMetadata value) {
-        if (metadataBuilder_ == null) {
-          if (value == null) {
-            throw new NullPointerException();
-          }
-          metadata_ = value;
-          onChanged();
-        } else {
-          metadataBuilder_.setMessage(value);
-        }
-
-        return this;
-      }
-      /**
-       * <pre>
-       * Contains metadata on the summary value such as which plugins may use it.
-       * Take note that many summary values may lack a metadata field. This is
-       * because the FileWriter only keeps a metadata object on the first summary
-       * value with a certain tag for each tag. TensorBoard then remembers which
-       * tags are associated with which plugins. This saves space.
-       * </pre>
-       *
-       * <code>.tensorflow.SummaryMetadata metadata = 9;</code>
-       */
-      public Builder setMetadata(
-          org.tensorflow.proto.framework.SummaryMetadata.Builder builderForValue) {
-        if (metadataBuilder_ == null) {
-          metadata_ = builderForValue.build();
-          onChanged();
-        } else {
-          metadataBuilder_.setMessage(builderForValue.build());
-        }
-
-        return this;
-      }
-      /**
-       * <pre>
-       * Contains metadata on the summary value such as which plugins may use it.
-       * Take note that many summary values may lack a metadata field. This is
-       * because the FileWriter only keeps a metadata object on the first summary
-       * value with a certain tag for each tag. TensorBoard then remembers which
-       * tags are associated with which plugins. This saves space.
-       * </pre>
-       *
-       * <code>.tensorflow.SummaryMetadata metadata = 9;</code>
-       */
-      public Builder mergeMetadata(org.tensorflow.proto.framework.SummaryMetadata value) {
-        if (metadataBuilder_ == null) {
-          if (metadata_ != null) {
-            metadata_ =
-              org.tensorflow.proto.framework.SummaryMetadata.newBuilder(metadata_).mergeFrom(value).buildPartial();
-          } else {
-            metadata_ = value;
-          }
-          onChanged();
-        } else {
-          metadataBuilder_.mergeFrom(value);
-        }
-
-        return this;
-      }
-      /**
-       * <pre>
-       * Contains metadata on the summary value such as which plugins may use it.
-       * Take note that many summary values may lack a metadata field. This is
-       * because the FileWriter only keeps a metadata object on the first summary
-       * value with a certain tag for each tag. TensorBoard then remembers which
-       * tags are associated with which plugins. This saves space.
-       * </pre>
-       *
-       * <code>.tensorflow.SummaryMetadata metadata = 9;</code>
-       */
-      public Builder clearMetadata() {
-        if (metadataBuilder_ == null) {
-          metadata_ = null;
-          onChanged();
-        } else {
-          metadata_ = null;
-          metadataBuilder_ = null;
-        }
-
-        return this;
-      }
-      /**
-       * <pre>
-       * Contains metadata on the summary value such as which plugins may use it.
-       * Take note that many summary values may lack a metadata field. This is
-       * because the FileWriter only keeps a metadata object on the first summary
-       * value with a certain tag for each tag. TensorBoard then remembers which
-       * tags are associated with which plugins. This saves space.
-       * </pre>
-       *
-       * <code>.tensorflow.SummaryMetadata metadata = 9;</code>
-       */
-      public org.tensorflow.proto.framework.SummaryMetadata.Builder getMetadataBuilder() {
-        
-        onChanged();
-        return getMetadataFieldBuilder().getBuilder();
-      }
-      /**
-       * <pre>
-       * Contains metadata on the summary value such as which plugins may use it.
-       * Take note that many summary values may lack a metadata field. This is
-       * because the FileWriter only keeps a metadata object on the first summary
-       * value with a certain tag for each tag. TensorBoard then remembers which
-       * tags are associated with which plugins. This saves space.
-       * </pre>
-       *
-       * <code>.tensorflow.SummaryMetadata metadata = 9;</code>
-       */
-      public org.tensorflow.proto.framework.SummaryMetadataOrBuilder getMetadataOrBuilder() {
-        if (metadataBuilder_ != null) {
-          return metadataBuilder_.getMessageOrBuilder();
-        } else {
-          return metadata_ == null ?
-              org.tensorflow.proto.framework.SummaryMetadata.getDefaultInstance() : metadata_;
-        }
-      }
-      /**
-       * <pre>
-       * Contains metadata on the summary value such as which plugins may use it.
-       * Take note that many summary values may lack a metadata field. This is
-       * because the FileWriter only keeps a metadata object on the first summary
-       * value with a certain tag for each tag. TensorBoard then remembers which
-       * tags are associated with which plugins. This saves space.
-       * </pre>
-       *
-       * <code>.tensorflow.SummaryMetadata metadata = 9;</code>
-       */
-      private com.google.protobuf.SingleFieldBuilderV3<
-          org.tensorflow.proto.framework.SummaryMetadata, org.tensorflow.proto.framework.SummaryMetadata.Builder, org.tensorflow.proto.framework.SummaryMetadataOrBuilder> 
-          getMetadataFieldBuilder() {
-        if (metadataBuilder_ == null) {
-          metadataBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-              org.tensorflow.proto.framework.SummaryMetadata, org.tensorflow.proto.framework.SummaryMetadata.Builder, org.tensorflow.proto.framework.SummaryMetadataOrBuilder>(
-                  getMetadata(),
-                  getParentForChildren(),
-                  isClean());
-          metadata_ = null;
-        }
-        return metadataBuilder_;
-      }
-
-      /**
-       * <code>float simple_value = 2;</code>
-       */
-      public float getSimpleValue() {
-        if (valueCase_ == 2) {
-          return (java.lang.Float) value_;
-        }
-        return 0F;
-      }
-      /**
-       * <code>float simple_value = 2;</code>
-       */
-      public Builder setSimpleValue(float value) {
-        valueCase_ = 2;
-        value_ = value;
-        onChanged();
-        return this;
-      }
-      /**
-       * <code>float simple_value = 2;</code>
-       */
-      public Builder clearSimpleValue() {
-        if (valueCase_ == 2) {
-          valueCase_ = 0;
-          value_ = null;
-          onChanged();
-        }
-        return this;
-      }
-
-      /**
-       * <code>bytes obsolete_old_style_histogram = 3;</code>
-       */
-      public com.google.protobuf.ByteString getObsoleteOldStyleHistogram() {
-        if (valueCase_ == 3) {
-          return (com.google.protobuf.ByteString) value_;
-        }
-        return com.google.protobuf.ByteString.EMPTY;
-      }
-      /**
-       * <code>bytes obsolete_old_style_histogram = 3;</code>
-       */
-      public Builder setObsoleteOldStyleHistogram(com.google.protobuf.ByteString value) {
-        if (value == null) {
-    throw new NullPointerException();
-  }
-  valueCase_ = 3;
-        value_ = value;
-        onChanged();
-        return this;
-      }
-      /**
-       * <code>bytes obsolete_old_style_histogram = 3;</code>
-       */
-      public Builder clearObsoleteOldStyleHistogram() {
-        if (valueCase_ == 3) {
-          valueCase_ = 0;
-          value_ = null;
-          onChanged();
-        }
-        return this;
-      }
-
-      private com.google.protobuf.SingleFieldBuilderV3<
-          org.tensorflow.proto.framework.Summary.Image, org.tensorflow.proto.framework.Summary.Image.Builder, org.tensorflow.proto.framework.Summary.ImageOrBuilder> imageBuilder_;
-      /**
-       * <code>.tensorflow.Summary.Image image = 4;</code>
-       */
-      public boolean hasImage() {
-        return valueCase_ == 4;
-      }
-      /**
-       * <code>.tensorflow.Summary.Image image = 4;</code>
-       */
-      public org.tensorflow.proto.framework.Summary.Image getImage() {
-        if (imageBuilder_ == null) {
-          if (valueCase_ == 4) {
-            return (org.tensorflow.proto.framework.Summary.Image) value_;
-          }
-          return org.tensorflow.proto.framework.Summary.Image.getDefaultInstance();
-        } else {
-          if (valueCase_ == 4) {
-            return imageBuilder_.getMessage();
-          }
-          return org.tensorflow.proto.framework.Summary.Image.getDefaultInstance();
-        }
-      }
-      /**
-       * <code>.tensorflow.Summary.Image image = 4;</code>
-       */
-      public Builder setImage(org.tensorflow.proto.framework.Summary.Image value) {
-        if (imageBuilder_ == null) {
-          if (value == null) {
-            throw new NullPointerException();
-          }
-          value_ = value;
-          onChanged();
-        } else {
-          imageBuilder_.setMessage(value);
-        }
-        valueCase_ = 4;
-        return this;
-      }
-      /**
-       * <code>.tensorflow.Summary.Image image = 4;</code>
-       */
-      public Builder setImage(
-          org.tensorflow.proto.framework.Summary.Image.Builder builderForValue) {
-        if (imageBuilder_ == null) {
-          value_ = builderForValue.build();
-          onChanged();
-        } else {
-          imageBuilder_.setMessage(builderForValue.build());
-        }
-        valueCase_ = 4;
-        return this;
-      }
-      /**
-       * <code>.tensorflow.Summary.Image image = 4;</code>
-       */
-      public Builder mergeImage(org.tensorflow.proto.framework.Summary.Image value) {
-        if (imageBuilder_ == null) {
-          if (valueCase_ == 4 &&
-              value_ != org.tensorflow.proto.framework.Summary.Image.getDefaultInstance()) {
-            value_ = org.tensorflow.proto.framework.Summary.Image.newBuilder((org.tensorflow.proto.framework.Summary.Image) value_)
-                .mergeFrom(value).buildPartial();
-          } else {
-            value_ = value;
-          }
-          onChanged();
-        } else {
-          if (valueCase_ == 4) {
-            imageBuilder_.mergeFrom(value);
-          }
-          imageBuilder_.setMessage(value);
-        }
-        valueCase_ = 4;
-        return this;
-      }
-      /**
-       * <code>.tensorflow.Summary.Image image = 4;</code>
-       */
-      public Builder clearImage() {
-        if (imageBuilder_ == null) {
-          if (valueCase_ == 4) {
-            valueCase_ = 0;
-            value_ = null;
-            onChanged();
-          }
-        } else {
-          if (valueCase_ == 4) {
-            valueCase_ = 0;
-            value_ = null;
-          }
-          imageBuilder_.clear();
-        }
-        return this;
-      }
-      /**
-       * <code>.tensorflow.Summary.Image image = 4;</code>
-       */
-      public org.tensorflow.proto.framework.Summary.Image.Builder getImageBuilder() {
-        return getImageFieldBuilder().getBuilder();
-      }
-      /**
-       * <code>.tensorflow.Summary.Image image = 4;</code>
-       */
-      public org.tensorflow.proto.framework.Summary.ImageOrBuilder getImageOrBuilder() {
-        if ((valueCase_ == 4) && (imageBuilder_ != null)) {
-          return imageBuilder_.getMessageOrBuilder();
-        } else {
-          if (valueCase_ == 4) {
-            return (org.tensorflow.proto.framework.Summary.Image) value_;
-          }
-          return org.tensorflow.proto.framework.Summary.Image.getDefaultInstance();
-        }
-      }
-      /**
-       * <code>.tensorflow.Summary.Image image = 4;</code>
-       */
-      private com.google.protobuf.SingleFieldBuilderV3<
-          org.tensorflow.proto.framework.Summary.Image, org.tensorflow.proto.framework.Summary.Image.Builder, org.tensorflow.proto.framework.Summary.ImageOrBuilder> 
-          getImageFieldBuilder() {
-        if (imageBuilder_ == null) {
-          if (!(valueCase_ == 4)) {
-            value_ = org.tensorflow.proto.framework.Summary.Image.getDefaultInstance();
-          }
-          imageBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-              org.tensorflow.proto.framework.Summary.Image, org.tensorflow.proto.framework.Summary.Image.Builder, org.tensorflow.proto.framework.Summary.ImageOrBuilder>(
-                  (org.tensorflow.proto.framework.Summary.Image) value_,
-                  getParentForChildren(),
-                  isClean());
-          value_ = null;
-        }
-        valueCase_ = 4;
-        onChanged();;
-        return imageBuilder_;
-      }
-
-      private com.google.protobuf.SingleFieldBuilderV3<
-          org.tensorflow.proto.framework.HistogramProto, org.tensorflow.proto.framework.HistogramProto.Builder, org.tensorflow.proto.framework.HistogramProtoOrBuilder> histoBuilder_;
-      /**
-       * <code>.tensorflow.HistogramProto histo = 5;</code>
-       */
-      public boolean hasHisto() {
-        return valueCase_ == 5;
-      }
-      /**
-       * <code>.tensorflow.HistogramProto histo = 5;</code>
-       */
-      public org.tensorflow.proto.framework.HistogramProto getHisto() {
-        if (histoBuilder_ == null) {
-          if (valueCase_ == 5) {
-            return (org.tensorflow.proto.framework.HistogramProto) value_;
-          }
-          return org.tensorflow.proto.framework.HistogramProto.getDefaultInstance();
-        } else {
-          if (valueCase_ == 5) {
-            return histoBuilder_.getMessage();
-          }
-          return org.tensorflow.proto.framework.HistogramProto.getDefaultInstance();
-        }
-      }
-      /**
-       * <code>.tensorflow.HistogramProto histo = 5;</code>
-       */
-      public Builder setHisto(org.tensorflow.proto.framework.HistogramProto value) {
-        if (histoBuilder_ == null) {
-          if (value == null) {
-            throw new NullPointerException();
-          }
-          value_ = value;
-          onChanged();
-        } else {
-          histoBuilder_.setMessage(value);
-        }
-        valueCase_ = 5;
-        return this;
-      }
-      /**
-       * <code>.tensorflow.HistogramProto histo = 5;</code>
-       */
-      public Builder setHisto(
-          org.tensorflow.proto.framework.HistogramProto.Builder builderForValue) {
-        if (histoBuilder_ == null) {
-          value_ = builderForValue.build();
-          onChanged();
-        } else {
-          histoBuilder_.setMessage(builderForValue.build());
-        }
-        valueCase_ = 5;
-        return this;
-      }
-      /**
-       * <code>.tensorflow.HistogramProto histo = 5;</code>
-       */
-      public Builder mergeHisto(org.tensorflow.proto.framework.HistogramProto value) {
-        if (histoBuilder_ == null) {
-          if (valueCase_ == 5 &&
-              value_ != org.tensorflow.proto.framework.HistogramProto.getDefaultInstance()) {
-            value_ = org.tensorflow.proto.framework.HistogramProto.newBuilder((org.tensorflow.proto.framework.HistogramProto) value_)
-                .mergeFrom(value).buildPartial();
-          } else {
-            value_ = value;
-          }
-          onChanged();
-        } else {
-          if (valueCase_ == 5) {
-            histoBuilder_.mergeFrom(value);
-          }
-          histoBuilder_.setMessage(value);
-        }
-        valueCase_ = 5;
-        return this;
-      }
-      /**
-       * <code>.tensorflow.HistogramProto histo = 5;</code>
-       */
-      public Builder clearHisto() {
-        if (histoBuilder_ == null) {
-          if (valueCase_ == 5) {
-            valueCase_ = 0;
-            value_ = null;
-            onChanged();
-          }
-        } else {
-          if (valueCase_ == 5) {
-            valueCase_ = 0;
-            value_ = null;
-          }
-          histoBuilder_.clear();
-        }
-        return this;
-      }
-      /**
-       * <code>.tensorflow.HistogramProto histo = 5;</code>
-       */
-      public org.tensorflow.proto.framework.HistogramProto.Builder getHistoBuilder() {
-        return getHistoFieldBuilder().getBuilder();
-      }
-      /**
-       * <code>.tensorflow.HistogramProto histo = 5;</code>
-       */
-      public org.tensorflow.proto.framework.HistogramProtoOrBuilder getHistoOrBuilder() {
-        if ((valueCase_ == 5) && (histoBuilder_ != null)) {
-          return histoBuilder_.getMessageOrBuilder();
-        } else {
-          if (valueCase_ == 5) {
-            return (org.tensorflow.proto.framework.HistogramProto) value_;
-          }
-          return org.tensorflow.proto.framework.HistogramProto.getDefaultInstance();
-        }
-      }
-      /**
-       * <code>.tensorflow.HistogramProto histo = 5;</code>
-       */
-      private com.google.protobuf.SingleFieldBuilderV3<
-          org.tensorflow.proto.framework.HistogramProto, org.tensorflow.proto.framework.HistogramProto.Builder, org.tensorflow.proto.framework.HistogramProtoOrBuilder> 
-          getHistoFieldBuilder() {
-        if (histoBuilder_ == null) {
-          if (!(valueCase_ == 5)) {
-            value_ = org.tensorflow.proto.framework.HistogramProto.getDefaultInstance();
-          }
-          histoBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-              org.tensorflow.proto.framework.HistogramProto, org.tensorflow.proto.framework.HistogramProto.Builder, org.tensorflow.proto.framework.HistogramProtoOrBuilder>(
-                  (org.tensorflow.proto.framework.HistogramProto) value_,
-                  getParentForChildren(),
-                  isClean());
-          value_ = null;
-        }
-        valueCase_ = 5;
-        onChanged();;
-        return histoBuilder_;
-      }
-
-      private com.google.protobuf.SingleFieldBuilderV3<
-          org.tensorflow.proto.framework.Summary.Audio, org.tensorflow.proto.framework.Summary.Audio.Builder, org.tensorflow.proto.framework.Summary.AudioOrBuilder> audioBuilder_;
-      /**
-       * <code>.tensorflow.Summary.Audio audio = 6;</code>
-       */
-      public boolean hasAudio() {
-        return valueCase_ == 6;
-      }
-      /**
-       * <code>.tensorflow.Summary.Audio audio = 6;</code>
-       */
-      public org.tensorflow.proto.framework.Summary.Audio getAudio() {
-        if (audioBuilder_ == null) {
-          if (valueCase_ == 6) {
-            return (org.tensorflow.proto.framework.Summary.Audio) value_;
-          }
-          return org.tensorflow.proto.framework.Summary.Audio.getDefaultInstance();
-        } else {
-          if (valueCase_ == 6) {
-            return audioBuilder_.getMessage();
-          }
-          return org.tensorflow.proto.framework.Summary.Audio.getDefaultInstance();
-        }
-      }
-      /**
-       * <code>.tensorflow.Summary.Audio audio = 6;</code>
-       */
-      public Builder setAudio(org.tensorflow.proto.framework.Summary.Audio value) {
-        if (audioBuilder_ == null) {
-          if (value == null) {
-            throw new NullPointerException();
-          }
-          value_ = value;
-          onChanged();
-        } else {
-          audioBuilder_.setMessage(value);
-        }
-        valueCase_ = 6;
-        return this;
-      }
-      /**
-       * <code>.tensorflow.Summary.Audio audio = 6;</code>
-       */
-      public Builder setAudio(
-          org.tensorflow.proto.framework.Summary.Audio.Builder builderForValue) {
-        if (audioBuilder_ == null) {
-          value_ = builderForValue.build();
-          onChanged();
-        } else {
-          audioBuilder_.setMessage(builderForValue.build());
-        }
-        valueCase_ = 6;
-        return this;
-      }
-      /**
-       * <code>.tensorflow.Summary.Audio audio = 6;</code>
-       */
-      public Builder mergeAudio(org.tensorflow.proto.framework.Summary.Audio value) {
-        if (audioBuilder_ == null) {
-          if (valueCase_ == 6 &&
-              value_ != org.tensorflow.proto.framework.Summary.Audio.getDefaultInstance()) {
-            value_ = org.tensorflow.proto.framework.Summary.Audio.newBuilder((org.tensorflow.proto.framework.Summary.Audio) value_)
-                .mergeFrom(value).buildPartial();
-          } else {
-            value_ = value;
-          }
-          onChanged();
-        } else {
-          if (valueCase_ == 6) {
-            audioBuilder_.mergeFrom(value);
-          }
-          audioBuilder_.setMessage(value);
-        }
-        valueCase_ = 6;
-        return this;
-      }
-      /**
-       * <code>.tensorflow.Summary.Audio audio = 6;</code>
-       */
-      public Builder clearAudio() {
-        if (audioBuilder_ == null) {
-          if (valueCase_ == 6) {
-            valueCase_ = 0;
-            value_ = null;
-            onChanged();
-          }
-        } else {
-          if (valueCase_ == 6) {
-            valueCase_ = 0;
-            value_ = null;
-          }
-          audioBuilder_.clear();
-        }
-        return this;
-      }
-      /**
-       * <code>.tensorflow.Summary.Audio audio = 6;</code>
-       */
-      public org.tensorflow.proto.framework.Summary.Audio.Builder getAudioBuilder() {
-        return getAudioFieldBuilder().getBuilder();
-      }
-      /**
-       * <code>.tensorflow.Summary.Audio audio = 6;</code>
-       */
-      public org.tensorflow.proto.framework.Summary.AudioOrBuilder getAudioOrBuilder() {
-        if ((valueCase_ == 6) && (audioBuilder_ != null)) {
-          return audioBuilder_.getMessageOrBuilder();
-        } else {
-          if (valueCase_ == 6) {
-            return (org.tensorflow.proto.framework.Summary.Audio) value_;
-          }
-          return org.tensorflow.proto.framework.Summary.Audio.getDefaultInstance();
-        }
-      }
-      /**
-       * <code>.tensorflow.Summary.Audio audio = 6;</code>
-       */
-      private com.google.protobuf.SingleFieldBuilderV3<
-          org.tensorflow.proto.framework.Summary.Audio, org.tensorflow.proto.framework.Summary.Audio.Builder, org.tensorflow.proto.framework.Summary.AudioOrBuilder> 
-          getAudioFieldBuilder() {
-        if (audioBuilder_ == null) {
-          if (!(valueCase_ == 6)) {
-            value_ = org.tensorflow.proto.framework.Summary.Audio.getDefaultInstance();
-          }
-          audioBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-              org.tensorflow.proto.framework.Summary.Audio, org.tensorflow.proto.framework.Summary.Audio.Builder, org.tensorflow.proto.framework.Summary.AudioOrBuilder>(
-                  (org.tensorflow.proto.framework.Summary.Audio) value_,
-                  getParentForChildren(),
-                  isClean());
-          value_ = null;
-        }
-        valueCase_ = 6;
-        onChanged();;
-        return audioBuilder_;
-      }
-
-      private com.google.protobuf.SingleFieldBuilderV3<
-          org.tensorflow.proto.framework.TensorProto, org.tensorflow.proto.framework.TensorProto.Builder, org.tensorflow.proto.framework.TensorProtoOrBuilder> tensorBuilder_;
-      /**
-       * <code>.tensorflow.TensorProto tensor = 8;</code>
-       */
-      public boolean hasTensor() {
-        return valueCase_ == 8;
-      }
-      /**
-       * <code>.tensorflow.TensorProto tensor = 8;</code>
-       */
-      public org.tensorflow.proto.framework.TensorProto getTensor() {
-        if (tensorBuilder_ == null) {
-          if (valueCase_ == 8) {
-            return (org.tensorflow.proto.framework.TensorProto) value_;
-          }
-          return org.tensorflow.proto.framework.TensorProto.getDefaultInstance();
-        } else {
-          if (valueCase_ == 8) {
-            return tensorBuilder_.getMessage();
-          }
-          return org.tensorflow.proto.framework.TensorProto.getDefaultInstance();
-        }
-      }
-      /**
-       * <code>.tensorflow.TensorProto tensor = 8;</code>
-       */
-      public Builder setTensor(org.tensorflow.proto.framework.TensorProto value) {
-        if (tensorBuilder_ == null) {
-          if (value == null) {
-            throw new NullPointerException();
-          }
-          value_ = value;
-          onChanged();
-        } else {
-          tensorBuilder_.setMessage(value);
-        }
-        valueCase_ = 8;
-        return this;
-      }
-      /**
-       * <code>.tensorflow.TensorProto tensor = 8;</code>
-       */
-      public Builder setTensor(
-          org.tensorflow.proto.framework.TensorProto.Builder builderForValue) {
-        if (tensorBuilder_ == null) {
-          value_ = builderForValue.build();
-          onChanged();
-        } else {
-          tensorBuilder_.setMessage(builderForValue.build());
-        }
-        valueCase_ = 8;
-        return this;
-      }
-      /**
-       * <code>.tensorflow.TensorProto tensor = 8;</code>
-       */
-      public Builder mergeTensor(org.tensorflow.proto.framework.TensorProto value) {
-        if (tensorBuilder_ == null) {
-          if (valueCase_ == 8 &&
-              value_ != org.tensorflow.proto.framework.TensorProto.getDefaultInstance()) {
-            value_ = org.tensorflow.proto.framework.TensorProto.newBuilder((org.tensorflow.proto.framework.TensorProto) value_)
-                .mergeFrom(value).buildPartial();
-          } else {
-            value_ = value;
-          }
-          onChanged();
-        } else {
-          if (valueCase_ == 8) {
-            tensorBuilder_.mergeFrom(value);
-          }
-          tensorBuilder_.setMessage(value);
-        }
-        valueCase_ = 8;
-        return this;
-      }
-      /**
-       * <code>.tensorflow.TensorProto tensor = 8;</code>
-       */
-      public Builder clearTensor() {
-        if (tensorBuilder_ == null) {
-          if (valueCase_ == 8) {
-            valueCase_ = 0;
-            value_ = null;
-            onChanged();
-          }
-        } else {
-          if (valueCase_ == 8) {
-            valueCase_ = 0;
-            value_ = null;
-          }
-          tensorBuilder_.clear();
-        }
-        return this;
-      }
-      /**
-       * <code>.tensorflow.TensorProto tensor = 8;</code>
-       */
-      public org.tensorflow.proto.framework.TensorProto.Builder getTensorBuilder() {
-        return getTensorFieldBuilder().getBuilder();
-      }
-      /**
-       * <code>.tensorflow.TensorProto tensor = 8;</code>
-       */
-      public org.tensorflow.proto.framework.TensorProtoOrBuilder getTensorOrBuilder() {
-        if ((valueCase_ == 8) && (tensorBuilder_ != null)) {
-          return tensorBuilder_.getMessageOrBuilder();
-        } else {
-          if (valueCase_ == 8) {
-            return (org.tensorflow.proto.framework.TensorProto) value_;
-          }
-          return org.tensorflow.proto.framework.TensorProto.getDefaultInstance();
-        }
-      }
-      /**
-       * <code>.tensorflow.TensorProto tensor = 8;</code>
-       */
-      private com.google.protobuf.SingleFieldBuilderV3<
-          org.tensorflow.proto.framework.TensorProto, org.tensorflow.proto.framework.TensorProto.Builder, org.tensorflow.proto.framework.TensorProtoOrBuilder> 
-          getTensorFieldBuilder() {
-        if (tensorBuilder_ == null) {
-          if (!(valueCase_ == 8)) {
-            value_ = org.tensorflow.proto.framework.TensorProto.getDefaultInstance();
-          }
-          tensorBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-              org.tensorflow.proto.framework.TensorProto, org.tensorflow.proto.framework.TensorProto.Builder, org.tensorflow.proto.framework.TensorProtoOrBuilder>(
-                  (org.tensorflow.proto.framework.TensorProto) value_,
-                  getParentForChildren(),
-                  isClean());
-          value_ = null;
-        }
-        valueCase_ = 8;
-        onChanged();;
-        return tensorBuilder_;
-      }
-      @java.lang.Override
-      public final Builder setUnknownFields(
-          final com.google.protobuf.UnknownFieldSet unknownFields) {
-        return super.setUnknownFields(unknownFields);
-      }
-
-      @java.lang.Override
-      public final Builder mergeUnknownFields(
-          final com.google.protobuf.UnknownFieldSet unknownFields) {
-        return super.mergeUnknownFields(unknownFields);
-      }
-
-
-      // @@protoc_insertion_point(builder_scope:tensorflow.Summary.Value)
-    }
-
-    // @@protoc_insertion_point(class_scope:tensorflow.Summary.Value)
-    private static final org.tensorflow.proto.framework.Summary.Value DEFAULT_INSTANCE;
-    static {
-      DEFAULT_INSTANCE = new org.tensorflow.proto.framework.Summary.Value();
-    }
-
-    public static org.tensorflow.proto.framework.Summary.Value getDefaultInstance() {
-      return DEFAULT_INSTANCE;
-    }
-
-    private static final com.google.protobuf.Parser<Value>
-        PARSER = new com.google.protobuf.AbstractParser<Value>() {
-      @java.lang.Override
-      public Value parsePartialFrom(
-          com.google.protobuf.CodedInputStream input,
-          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-          throws com.google.protobuf.InvalidProtocolBufferException {
-        return new Value(input, extensionRegistry);
-      }
-    };
-
-    public static com.google.protobuf.Parser<Value> parser() {
-      return PARSER;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Parser<Value> getParserForType() {
-      return PARSER;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.Summary.Value getDefaultInstanceForType() {
-      return DEFAULT_INSTANCE;
-    }
-
-  }
-
-  public static final int VALUE_FIELD_NUMBER = 1;
-  private java.util.List<org.tensorflow.proto.framework.Summary.Value> value_;
-  /**
-   * <pre>
-   * Set of values for the summary.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.Summary.Value value = 1;</code>
-   */
-  public java.util.List<org.tensorflow.proto.framework.Summary.Value> getValueList() {
-    return value_;
-  }
-  /**
-   * <pre>
-   * Set of values for the summary.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.Summary.Value value = 1;</code>
-   */
-  public java.util.List<? extends org.tensorflow.proto.framework.Summary.ValueOrBuilder> 
-      getValueOrBuilderList() {
-    return value_;
-  }
-  /**
-   * <pre>
-   * Set of values for the summary.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.Summary.Value value = 1;</code>
-   */
-  public int getValueCount() {
-    return value_.size();
-  }
-  /**
-   * <pre>
-   * Set of values for the summary.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.Summary.Value value = 1;</code>
-   */
-  public org.tensorflow.proto.framework.Summary.Value getValue(int index) {
-    return value_.get(index);
-  }
-  /**
-   * <pre>
-   * Set of values for the summary.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.Summary.Value value = 1;</code>
-   */
-  public org.tensorflow.proto.framework.Summary.ValueOrBuilder getValueOrBuilder(
-      int index) {
-    return value_.get(index);
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    for (int i = 0; i < value_.size(); i++) {
-      output.writeMessage(1, value_.get(i));
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    for (int i = 0; i < value_.size(); i++) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(1, value_.get(i));
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.framework.Summary)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.framework.Summary other = (org.tensorflow.proto.framework.Summary) obj;
-
-    if (!getValueList()
-        .equals(other.getValueList())) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    if (getValueCount() > 0) {
-      hash = (37 * hash) + VALUE_FIELD_NUMBER;
-      hash = (53 * hash) + getValueList().hashCode();
-    }
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.framework.Summary parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.Summary parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.Summary parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.Summary parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.Summary parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.Summary parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.Summary parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.Summary parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.Summary parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.Summary parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.Summary parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.Summary parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.framework.Summary prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * <pre>
-   * A Summary is a set of named values to be displayed by the
-   * visualizer.
-   * Summaries are produced regularly during training, as controlled by
-   * the "summary_interval_secs" attribute of the training operation.
-   * Summaries are also produced at the end of an evaluation.
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.Summary}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.Summary)
-      org.tensorflow.proto.framework.SummaryOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.SummaryProtos.internal_static_tensorflow_Summary_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.SummaryProtos.internal_static_tensorflow_Summary_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.Summary.class, org.tensorflow.proto.framework.Summary.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.framework.Summary.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-        getValueFieldBuilder();
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      if (valueBuilder_ == null) {
-        value_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000001);
-      } else {
-        valueBuilder_.clear();
-      }
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.framework.SummaryProtos.internal_static_tensorflow_Summary_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.Summary getDefaultInstanceForType() {
-      return org.tensorflow.proto.framework.Summary.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.Summary build() {
-      org.tensorflow.proto.framework.Summary result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.Summary buildPartial() {
-      org.tensorflow.proto.framework.Summary result = new org.tensorflow.proto.framework.Summary(this);
-      int from_bitField0_ = bitField0_;
-      if (valueBuilder_ == null) {
-        if (((bitField0_ & 0x00000001) != 0)) {
-          value_ = java.util.Collections.unmodifiableList(value_);
-          bitField0_ = (bitField0_ & ~0x00000001);
-        }
-        result.value_ = value_;
-      } else {
-        result.value_ = valueBuilder_.build();
-      }
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.framework.Summary) {
-        return mergeFrom((org.tensorflow.proto.framework.Summary)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.framework.Summary other) {
-      if (other == org.tensorflow.proto.framework.Summary.getDefaultInstance()) return this;
-      if (valueBuilder_ == null) {
-        if (!other.value_.isEmpty()) {
-          if (value_.isEmpty()) {
-            value_ = other.value_;
-            bitField0_ = (bitField0_ & ~0x00000001);
-          } else {
-            ensureValueIsMutable();
-            value_.addAll(other.value_);
-          }
-          onChanged();
-        }
-      } else {
-        if (!other.value_.isEmpty()) {
-          if (valueBuilder_.isEmpty()) {
-            valueBuilder_.dispose();
-            valueBuilder_ = null;
-            value_ = other.value_;
-            bitField0_ = (bitField0_ & ~0x00000001);
-            valueBuilder_ = 
-              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
-                 getValueFieldBuilder() : null;
-          } else {
-            valueBuilder_.addAllMessages(other.value_);
-          }
-        }
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.framework.Summary parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.framework.Summary) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-    private int bitField0_;
-
-    private java.util.List<org.tensorflow.proto.framework.Summary.Value> value_ =
-      java.util.Collections.emptyList();
-    private void ensureValueIsMutable() {
-      if (!((bitField0_ & 0x00000001) != 0)) {
-        value_ = new java.util.ArrayList<org.tensorflow.proto.framework.Summary.Value>(value_);
-        bitField0_ |= 0x00000001;
-       }
-    }
-
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.framework.Summary.Value, org.tensorflow.proto.framework.Summary.Value.Builder, org.tensorflow.proto.framework.Summary.ValueOrBuilder> valueBuilder_;
-
-    /**
-     * <pre>
-     * Set of values for the summary.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.Summary.Value value = 1;</code>
-     */
-    public java.util.List<org.tensorflow.proto.framework.Summary.Value> getValueList() {
-      if (valueBuilder_ == null) {
-        return java.util.Collections.unmodifiableList(value_);
-      } else {
-        return valueBuilder_.getMessageList();
-      }
-    }
-    /**
-     * <pre>
-     * Set of values for the summary.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.Summary.Value value = 1;</code>
-     */
-    public int getValueCount() {
-      if (valueBuilder_ == null) {
-        return value_.size();
-      } else {
-        return valueBuilder_.getCount();
-      }
-    }
-    /**
-     * <pre>
-     * Set of values for the summary.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.Summary.Value value = 1;</code>
-     */
-    public org.tensorflow.proto.framework.Summary.Value getValue(int index) {
-      if (valueBuilder_ == null) {
-        return value_.get(index);
-      } else {
-        return valueBuilder_.getMessage(index);
-      }
-    }
-    /**
-     * <pre>
-     * Set of values for the summary.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.Summary.Value value = 1;</code>
-     */
-    public Builder setValue(
-        int index, org.tensorflow.proto.framework.Summary.Value value) {
-      if (valueBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureValueIsMutable();
-        value_.set(index, value);
-        onChanged();
-      } else {
-        valueBuilder_.setMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Set of values for the summary.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.Summary.Value value = 1;</code>
-     */
-    public Builder setValue(
-        int index, org.tensorflow.proto.framework.Summary.Value.Builder builderForValue) {
-      if (valueBuilder_ == null) {
-        ensureValueIsMutable();
-        value_.set(index, builderForValue.build());
-        onChanged();
-      } else {
-        valueBuilder_.setMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Set of values for the summary.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.Summary.Value value = 1;</code>
-     */
-    public Builder addValue(org.tensorflow.proto.framework.Summary.Value value) {
-      if (valueBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureValueIsMutable();
-        value_.add(value);
-        onChanged();
-      } else {
-        valueBuilder_.addMessage(value);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Set of values for the summary.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.Summary.Value value = 1;</code>
-     */
-    public Builder addValue(
-        int index, org.tensorflow.proto.framework.Summary.Value value) {
-      if (valueBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureValueIsMutable();
-        value_.add(index, value);
-        onChanged();
-      } else {
-        valueBuilder_.addMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Set of values for the summary.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.Summary.Value value = 1;</code>
-     */
-    public Builder addValue(
-        org.tensorflow.proto.framework.Summary.Value.Builder builderForValue) {
-      if (valueBuilder_ == null) {
-        ensureValueIsMutable();
-        value_.add(builderForValue.build());
-        onChanged();
-      } else {
-        valueBuilder_.addMessage(builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Set of values for the summary.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.Summary.Value value = 1;</code>
-     */
-    public Builder addValue(
-        int index, org.tensorflow.proto.framework.Summary.Value.Builder builderForValue) {
-      if (valueBuilder_ == null) {
-        ensureValueIsMutable();
-        value_.add(index, builderForValue.build());
-        onChanged();
-      } else {
-        valueBuilder_.addMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Set of values for the summary.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.Summary.Value value = 1;</code>
-     */
-    public Builder addAllValue(
-        java.lang.Iterable<? extends org.tensorflow.proto.framework.Summary.Value> values) {
-      if (valueBuilder_ == null) {
-        ensureValueIsMutable();
-        com.google.protobuf.AbstractMessageLite.Builder.addAll(
-            values, value_);
-        onChanged();
-      } else {
-        valueBuilder_.addAllMessages(values);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Set of values for the summary.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.Summary.Value value = 1;</code>
-     */
-    public Builder clearValue() {
-      if (valueBuilder_ == null) {
-        value_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000001);
-        onChanged();
-      } else {
-        valueBuilder_.clear();
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Set of values for the summary.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.Summary.Value value = 1;</code>
-     */
-    public Builder removeValue(int index) {
-      if (valueBuilder_ == null) {
-        ensureValueIsMutable();
-        value_.remove(index);
-        onChanged();
-      } else {
-        valueBuilder_.remove(index);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Set of values for the summary.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.Summary.Value value = 1;</code>
-     */
-    public org.tensorflow.proto.framework.Summary.Value.Builder getValueBuilder(
-        int index) {
-      return getValueFieldBuilder().getBuilder(index);
-    }
-    /**
-     * <pre>
-     * Set of values for the summary.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.Summary.Value value = 1;</code>
-     */
-    public org.tensorflow.proto.framework.Summary.ValueOrBuilder getValueOrBuilder(
-        int index) {
-      if (valueBuilder_ == null) {
-        return value_.get(index);  } else {
-        return valueBuilder_.getMessageOrBuilder(index);
-      }
-    }
-    /**
-     * <pre>
-     * Set of values for the summary.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.Summary.Value value = 1;</code>
-     */
-    public java.util.List<? extends org.tensorflow.proto.framework.Summary.ValueOrBuilder> 
-         getValueOrBuilderList() {
-      if (valueBuilder_ != null) {
-        return valueBuilder_.getMessageOrBuilderList();
-      } else {
-        return java.util.Collections.unmodifiableList(value_);
-      }
-    }
-    /**
-     * <pre>
-     * Set of values for the summary.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.Summary.Value value = 1;</code>
-     */
-    public org.tensorflow.proto.framework.Summary.Value.Builder addValueBuilder() {
-      return getValueFieldBuilder().addBuilder(
-          org.tensorflow.proto.framework.Summary.Value.getDefaultInstance());
-    }
-    /**
-     * <pre>
-     * Set of values for the summary.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.Summary.Value value = 1;</code>
-     */
-    public org.tensorflow.proto.framework.Summary.Value.Builder addValueBuilder(
-        int index) {
-      return getValueFieldBuilder().addBuilder(
-          index, org.tensorflow.proto.framework.Summary.Value.getDefaultInstance());
-    }
-    /**
-     * <pre>
-     * Set of values for the summary.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.Summary.Value value = 1;</code>
-     */
-    public java.util.List<org.tensorflow.proto.framework.Summary.Value.Builder> 
-         getValueBuilderList() {
-      return getValueFieldBuilder().getBuilderList();
-    }
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.framework.Summary.Value, org.tensorflow.proto.framework.Summary.Value.Builder, org.tensorflow.proto.framework.Summary.ValueOrBuilder> 
-        getValueFieldBuilder() {
-      if (valueBuilder_ == null) {
-        valueBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
-            org.tensorflow.proto.framework.Summary.Value, org.tensorflow.proto.framework.Summary.Value.Builder, org.tensorflow.proto.framework.Summary.ValueOrBuilder>(
-                value_,
-                ((bitField0_ & 0x00000001) != 0),
-                getParentForChildren(),
-                isClean());
-        value_ = null;
-      }
-      return valueBuilder_;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.Summary)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.Summary)
-  private static final org.tensorflow.proto.framework.Summary DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.Summary();
-  }
-
-  public static org.tensorflow.proto.framework.Summary getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<Summary>
-      PARSER = new com.google.protobuf.AbstractParser<Summary>() {
-    @java.lang.Override
-    public Summary parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new Summary(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<Summary> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<Summary> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.framework.Summary getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SummaryDescription.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SummaryDescription.java
deleted file mode 100644
index 1543fd841d1..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SummaryDescription.java
+++ /dev/null
@@ -1,589 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/framework/summary.proto
-
-package org.tensorflow.proto.framework;
-
-/**
- * <pre>
- * Metadata associated with a series of Summary data
- * </pre>
- *
- * Protobuf type {@code tensorflow.SummaryDescription}
- */
-public  final class SummaryDescription extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.SummaryDescription)
-    SummaryDescriptionOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use SummaryDescription.newBuilder() to construct.
-  private SummaryDescription(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private SummaryDescription() {
-    typeHint_ = "";
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new SummaryDescription();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private SummaryDescription(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 10: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            typeHint_ = s;
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.framework.SummaryProtos.internal_static_tensorflow_SummaryDescription_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.framework.SummaryProtos.internal_static_tensorflow_SummaryDescription_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.framework.SummaryDescription.class, org.tensorflow.proto.framework.SummaryDescription.Builder.class);
-  }
-
-  public static final int TYPE_HINT_FIELD_NUMBER = 1;
-  private volatile java.lang.Object typeHint_;
-  /**
-   * <pre>
-   * Hint on how plugins should process the data in this series.
-   * Supported values include "scalar", "histogram", "image", "audio"
-   * </pre>
-   *
-   * <code>string type_hint = 1;</code>
-   */
-  public java.lang.String getTypeHint() {
-    java.lang.Object ref = typeHint_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      typeHint_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * Hint on how plugins should process the data in this series.
-   * Supported values include "scalar", "histogram", "image", "audio"
-   * </pre>
-   *
-   * <code>string type_hint = 1;</code>
-   */
-  public com.google.protobuf.ByteString
-      getTypeHintBytes() {
-    java.lang.Object ref = typeHint_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      typeHint_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    if (!getTypeHintBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 1, typeHint_);
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (!getTypeHintBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, typeHint_);
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.framework.SummaryDescription)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.framework.SummaryDescription other = (org.tensorflow.proto.framework.SummaryDescription) obj;
-
-    if (!getTypeHint()
-        .equals(other.getTypeHint())) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    hash = (37 * hash) + TYPE_HINT_FIELD_NUMBER;
-    hash = (53 * hash) + getTypeHint().hashCode();
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.framework.SummaryDescription parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.SummaryDescription parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.SummaryDescription parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.SummaryDescription parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.SummaryDescription parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.SummaryDescription parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.SummaryDescription parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.SummaryDescription parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.SummaryDescription parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.SummaryDescription parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.SummaryDescription parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.SummaryDescription parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.framework.SummaryDescription prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * <pre>
-   * Metadata associated with a series of Summary data
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.SummaryDescription}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.SummaryDescription)
-      org.tensorflow.proto.framework.SummaryDescriptionOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.SummaryProtos.internal_static_tensorflow_SummaryDescription_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.SummaryProtos.internal_static_tensorflow_SummaryDescription_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.SummaryDescription.class, org.tensorflow.proto.framework.SummaryDescription.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.framework.SummaryDescription.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      typeHint_ = "";
-
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.framework.SummaryProtos.internal_static_tensorflow_SummaryDescription_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.SummaryDescription getDefaultInstanceForType() {
-      return org.tensorflow.proto.framework.SummaryDescription.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.SummaryDescription build() {
-      org.tensorflow.proto.framework.SummaryDescription result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.SummaryDescription buildPartial() {
-      org.tensorflow.proto.framework.SummaryDescription result = new org.tensorflow.proto.framework.SummaryDescription(this);
-      result.typeHint_ = typeHint_;
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.framework.SummaryDescription) {
-        return mergeFrom((org.tensorflow.proto.framework.SummaryDescription)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.framework.SummaryDescription other) {
-      if (other == org.tensorflow.proto.framework.SummaryDescription.getDefaultInstance()) return this;
-      if (!other.getTypeHint().isEmpty()) {
-        typeHint_ = other.typeHint_;
-        onChanged();
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.framework.SummaryDescription parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.framework.SummaryDescription) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-
-    private java.lang.Object typeHint_ = "";
-    /**
-     * <pre>
-     * Hint on how plugins should process the data in this series.
-     * Supported values include "scalar", "histogram", "image", "audio"
-     * </pre>
-     *
-     * <code>string type_hint = 1;</code>
-     */
-    public java.lang.String getTypeHint() {
-      java.lang.Object ref = typeHint_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        typeHint_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Hint on how plugins should process the data in this series.
-     * Supported values include "scalar", "histogram", "image", "audio"
-     * </pre>
-     *
-     * <code>string type_hint = 1;</code>
-     */
-    public com.google.protobuf.ByteString
-        getTypeHintBytes() {
-      java.lang.Object ref = typeHint_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        typeHint_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Hint on how plugins should process the data in this series.
-     * Supported values include "scalar", "histogram", "image", "audio"
-     * </pre>
-     *
-     * <code>string type_hint = 1;</code>
-     */
-    public Builder setTypeHint(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      typeHint_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Hint on how plugins should process the data in this series.
-     * Supported values include "scalar", "histogram", "image", "audio"
-     * </pre>
-     *
-     * <code>string type_hint = 1;</code>
-     */
-    public Builder clearTypeHint() {
-      
-      typeHint_ = getDefaultInstance().getTypeHint();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Hint on how plugins should process the data in this series.
-     * Supported values include "scalar", "histogram", "image", "audio"
-     * </pre>
-     *
-     * <code>string type_hint = 1;</code>
-     */
-    public Builder setTypeHintBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      typeHint_ = value;
-      onChanged();
-      return this;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.SummaryDescription)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.SummaryDescription)
-  private static final org.tensorflow.proto.framework.SummaryDescription DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.SummaryDescription();
-  }
-
-  public static org.tensorflow.proto.framework.SummaryDescription getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<SummaryDescription>
-      PARSER = new com.google.protobuf.AbstractParser<SummaryDescription>() {
-    @java.lang.Override
-    public SummaryDescription parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new SummaryDescription(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<SummaryDescription> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<SummaryDescription> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.framework.SummaryDescription getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SummaryDescriptionOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SummaryDescriptionOrBuilder.java
deleted file mode 100644
index a6b581e65e4..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SummaryDescriptionOrBuilder.java
+++ /dev/null
@@ -1,29 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/framework/summary.proto
-
-package org.tensorflow.proto.framework;
-
-public interface SummaryDescriptionOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.SummaryDescription)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <pre>
-   * Hint on how plugins should process the data in this series.
-   * Supported values include "scalar", "histogram", "image", "audio"
-   * </pre>
-   *
-   * <code>string type_hint = 1;</code>
-   */
-  java.lang.String getTypeHint();
-  /**
-   * <pre>
-   * Hint on how plugins should process the data in this series.
-   * Supported values include "scalar", "histogram", "image", "audio"
-   * </pre>
-   *
-   * <code>string type_hint = 1;</code>
-   */
-  com.google.protobuf.ByteString
-      getTypeHintBytes();
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SummaryMetadata.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SummaryMetadata.java
deleted file mode 100644
index 969ed4d6d71..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SummaryMetadata.java
+++ /dev/null
@@ -1,1784 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/framework/summary.proto
-
-package org.tensorflow.proto.framework;
-
-/**
- * <pre>
- * A SummaryMetadata encapsulates information on which plugins are able to make
- * use of a certain summary value.
- * </pre>
- *
- * Protobuf type {@code tensorflow.SummaryMetadata}
- */
-public  final class SummaryMetadata extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.SummaryMetadata)
-    SummaryMetadataOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use SummaryMetadata.newBuilder() to construct.
-  private SummaryMetadata(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private SummaryMetadata() {
-    displayName_ = "";
-    summaryDescription_ = "";
-    dataClass_ = 0;
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new SummaryMetadata();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private SummaryMetadata(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 10: {
-            org.tensorflow.proto.framework.SummaryMetadata.PluginData.Builder subBuilder = null;
-            if (pluginData_ != null) {
-              subBuilder = pluginData_.toBuilder();
-            }
-            pluginData_ = input.readMessage(org.tensorflow.proto.framework.SummaryMetadata.PluginData.parser(), extensionRegistry);
-            if (subBuilder != null) {
-              subBuilder.mergeFrom(pluginData_);
-              pluginData_ = subBuilder.buildPartial();
-            }
-
-            break;
-          }
-          case 18: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            displayName_ = s;
-            break;
-          }
-          case 26: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            summaryDescription_ = s;
-            break;
-          }
-          case 32: {
-            int rawValue = input.readEnum();
-
-            dataClass_ = rawValue;
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.framework.SummaryProtos.internal_static_tensorflow_SummaryMetadata_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.framework.SummaryProtos.internal_static_tensorflow_SummaryMetadata_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.framework.SummaryMetadata.class, org.tensorflow.proto.framework.SummaryMetadata.Builder.class);
-  }
-
-  public interface PluginDataOrBuilder extends
-      // @@protoc_insertion_point(interface_extends:tensorflow.SummaryMetadata.PluginData)
-      com.google.protobuf.MessageOrBuilder {
-
-    /**
-     * <pre>
-     * The name of the plugin this data pertains to.
-     * </pre>
-     *
-     * <code>string plugin_name = 1;</code>
-     */
-    java.lang.String getPluginName();
-    /**
-     * <pre>
-     * The name of the plugin this data pertains to.
-     * </pre>
-     *
-     * <code>string plugin_name = 1;</code>
-     */
-    com.google.protobuf.ByteString
-        getPluginNameBytes();
-
-    /**
-     * <pre>
-     * The content to store for the plugin. The best practice is for this to be
-     * a binary serialized protocol buffer.
-     * </pre>
-     *
-     * <code>bytes content = 2;</code>
-     */
-    com.google.protobuf.ByteString getContent();
-  }
-  /**
-   * Protobuf type {@code tensorflow.SummaryMetadata.PluginData}
-   */
-  public  static final class PluginData extends
-      com.google.protobuf.GeneratedMessageV3 implements
-      // @@protoc_insertion_point(message_implements:tensorflow.SummaryMetadata.PluginData)
-      PluginDataOrBuilder {
-  private static final long serialVersionUID = 0L;
-    // Use PluginData.newBuilder() to construct.
-    private PluginData(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-      super(builder);
-    }
-    private PluginData() {
-      pluginName_ = "";
-      content_ = com.google.protobuf.ByteString.EMPTY;
-    }
-
-    @java.lang.Override
-    @SuppressWarnings({"unused"})
-    protected java.lang.Object newInstance(
-        UnusedPrivateParameter unused) {
-      return new PluginData();
-    }
-
-    @java.lang.Override
-    public final com.google.protobuf.UnknownFieldSet
-    getUnknownFields() {
-      return this.unknownFields;
-    }
-    private PluginData(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      this();
-      if (extensionRegistry == null) {
-        throw new java.lang.NullPointerException();
-      }
-      com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-          com.google.protobuf.UnknownFieldSet.newBuilder();
-      try {
-        boolean done = false;
-        while (!done) {
-          int tag = input.readTag();
-          switch (tag) {
-            case 0:
-              done = true;
-              break;
-            case 10: {
-              java.lang.String s = input.readStringRequireUtf8();
-
-              pluginName_ = s;
-              break;
-            }
-            case 18: {
-
-              content_ = input.readBytes();
-              break;
-            }
-            default: {
-              if (!parseUnknownField(
-                  input, unknownFields, extensionRegistry, tag)) {
-                done = true;
-              }
-              break;
-            }
-          }
-        }
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        throw e.setUnfinishedMessage(this);
-      } catch (java.io.IOException e) {
-        throw new com.google.protobuf.InvalidProtocolBufferException(
-            e).setUnfinishedMessage(this);
-      } finally {
-        this.unknownFields = unknownFields.build();
-        makeExtensionsImmutable();
-      }
-    }
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.SummaryProtos.internal_static_tensorflow_SummaryMetadata_PluginData_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.SummaryProtos.internal_static_tensorflow_SummaryMetadata_PluginData_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.SummaryMetadata.PluginData.class, org.tensorflow.proto.framework.SummaryMetadata.PluginData.Builder.class);
-    }
-
-    public static final int PLUGIN_NAME_FIELD_NUMBER = 1;
-    private volatile java.lang.Object pluginName_;
-    /**
-     * <pre>
-     * The name of the plugin this data pertains to.
-     * </pre>
-     *
-     * <code>string plugin_name = 1;</code>
-     */
-    public java.lang.String getPluginName() {
-      java.lang.Object ref = pluginName_;
-      if (ref instanceof java.lang.String) {
-        return (java.lang.String) ref;
-      } else {
-        com.google.protobuf.ByteString bs = 
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        pluginName_ = s;
-        return s;
-      }
-    }
-    /**
-     * <pre>
-     * The name of the plugin this data pertains to.
-     * </pre>
-     *
-     * <code>string plugin_name = 1;</code>
-     */
-    public com.google.protobuf.ByteString
-        getPluginNameBytes() {
-      java.lang.Object ref = pluginName_;
-      if (ref instanceof java.lang.String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        pluginName_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-
-    public static final int CONTENT_FIELD_NUMBER = 2;
-    private com.google.protobuf.ByteString content_;
-    /**
-     * <pre>
-     * The content to store for the plugin. The best practice is for this to be
-     * a binary serialized protocol buffer.
-     * </pre>
-     *
-     * <code>bytes content = 2;</code>
-     */
-    public com.google.protobuf.ByteString getContent() {
-      return content_;
-    }
-
-    private byte memoizedIsInitialized = -1;
-    @java.lang.Override
-    public final boolean isInitialized() {
-      byte isInitialized = memoizedIsInitialized;
-      if (isInitialized == 1) return true;
-      if (isInitialized == 0) return false;
-
-      memoizedIsInitialized = 1;
-      return true;
-    }
-
-    @java.lang.Override
-    public void writeTo(com.google.protobuf.CodedOutputStream output)
-                        throws java.io.IOException {
-      if (!getPluginNameBytes().isEmpty()) {
-        com.google.protobuf.GeneratedMessageV3.writeString(output, 1, pluginName_);
-      }
-      if (!content_.isEmpty()) {
-        output.writeBytes(2, content_);
-      }
-      unknownFields.writeTo(output);
-    }
-
-    @java.lang.Override
-    public int getSerializedSize() {
-      int size = memoizedSize;
-      if (size != -1) return size;
-
-      size = 0;
-      if (!getPluginNameBytes().isEmpty()) {
-        size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, pluginName_);
-      }
-      if (!content_.isEmpty()) {
-        size += com.google.protobuf.CodedOutputStream
-          .computeBytesSize(2, content_);
-      }
-      size += unknownFields.getSerializedSize();
-      memoizedSize = size;
-      return size;
-    }
-
-    @java.lang.Override
-    public boolean equals(final java.lang.Object obj) {
-      if (obj == this) {
-       return true;
-      }
-      if (!(obj instanceof org.tensorflow.proto.framework.SummaryMetadata.PluginData)) {
-        return super.equals(obj);
-      }
-      org.tensorflow.proto.framework.SummaryMetadata.PluginData other = (org.tensorflow.proto.framework.SummaryMetadata.PluginData) obj;
-
-      if (!getPluginName()
-          .equals(other.getPluginName())) return false;
-      if (!getContent()
-          .equals(other.getContent())) return false;
-      if (!unknownFields.equals(other.unknownFields)) return false;
-      return true;
-    }
-
-    @java.lang.Override
-    public int hashCode() {
-      if (memoizedHashCode != 0) {
-        return memoizedHashCode;
-      }
-      int hash = 41;
-      hash = (19 * hash) + getDescriptor().hashCode();
-      hash = (37 * hash) + PLUGIN_NAME_FIELD_NUMBER;
-      hash = (53 * hash) + getPluginName().hashCode();
-      hash = (37 * hash) + CONTENT_FIELD_NUMBER;
-      hash = (53 * hash) + getContent().hashCode();
-      hash = (29 * hash) + unknownFields.hashCode();
-      memoizedHashCode = hash;
-      return hash;
-    }
-
-    public static org.tensorflow.proto.framework.SummaryMetadata.PluginData parseFrom(
-        java.nio.ByteBuffer data)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data);
-    }
-    public static org.tensorflow.proto.framework.SummaryMetadata.PluginData parseFrom(
-        java.nio.ByteBuffer data,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.SummaryMetadata.PluginData parseFrom(
-        com.google.protobuf.ByteString data)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data);
-    }
-    public static org.tensorflow.proto.framework.SummaryMetadata.PluginData parseFrom(
-        com.google.protobuf.ByteString data,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.SummaryMetadata.PluginData parseFrom(byte[] data)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data);
-    }
-    public static org.tensorflow.proto.framework.SummaryMetadata.PluginData parseFrom(
-        byte[] data,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.SummaryMetadata.PluginData parseFrom(java.io.InputStream input)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input);
-    }
-    public static org.tensorflow.proto.framework.SummaryMetadata.PluginData parseFrom(
-        java.io.InputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.SummaryMetadata.PluginData parseDelimitedFrom(java.io.InputStream input)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseDelimitedWithIOException(PARSER, input);
-    }
-    public static org.tensorflow.proto.framework.SummaryMetadata.PluginData parseDelimitedFrom(
-        java.io.InputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.SummaryMetadata.PluginData parseFrom(
-        com.google.protobuf.CodedInputStream input)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input);
-    }
-    public static org.tensorflow.proto.framework.SummaryMetadata.PluginData parseFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input, extensionRegistry);
-    }
-
-    @java.lang.Override
-    public Builder newBuilderForType() { return newBuilder(); }
-    public static Builder newBuilder() {
-      return DEFAULT_INSTANCE.toBuilder();
-    }
-    public static Builder newBuilder(org.tensorflow.proto.framework.SummaryMetadata.PluginData prototype) {
-      return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-    }
-    @java.lang.Override
-    public Builder toBuilder() {
-      return this == DEFAULT_INSTANCE
-          ? new Builder() : new Builder().mergeFrom(this);
-    }
-
-    @java.lang.Override
-    protected Builder newBuilderForType(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      Builder builder = new Builder(parent);
-      return builder;
-    }
-    /**
-     * Protobuf type {@code tensorflow.SummaryMetadata.PluginData}
-     */
-    public static final class Builder extends
-        com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-        // @@protoc_insertion_point(builder_implements:tensorflow.SummaryMetadata.PluginData)
-        org.tensorflow.proto.framework.SummaryMetadata.PluginDataOrBuilder {
-      public static final com.google.protobuf.Descriptors.Descriptor
-          getDescriptor() {
-        return org.tensorflow.proto.framework.SummaryProtos.internal_static_tensorflow_SummaryMetadata_PluginData_descriptor;
-      }
-
-      @java.lang.Override
-      protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-          internalGetFieldAccessorTable() {
-        return org.tensorflow.proto.framework.SummaryProtos.internal_static_tensorflow_SummaryMetadata_PluginData_fieldAccessorTable
-            .ensureFieldAccessorsInitialized(
-                org.tensorflow.proto.framework.SummaryMetadata.PluginData.class, org.tensorflow.proto.framework.SummaryMetadata.PluginData.Builder.class);
-      }
-
-      // Construct using org.tensorflow.proto.framework.SummaryMetadata.PluginData.newBuilder()
-      private Builder() {
-        maybeForceBuilderInitialization();
-      }
-
-      private Builder(
-          com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-        super(parent);
-        maybeForceBuilderInitialization();
-      }
-      private void maybeForceBuilderInitialization() {
-        if (com.google.protobuf.GeneratedMessageV3
-                .alwaysUseFieldBuilders) {
-        }
-      }
-      @java.lang.Override
-      public Builder clear() {
-        super.clear();
-        pluginName_ = "";
-
-        content_ = com.google.protobuf.ByteString.EMPTY;
-
-        return this;
-      }
-
-      @java.lang.Override
-      public com.google.protobuf.Descriptors.Descriptor
-          getDescriptorForType() {
-        return org.tensorflow.proto.framework.SummaryProtos.internal_static_tensorflow_SummaryMetadata_PluginData_descriptor;
-      }
-
-      @java.lang.Override
-      public org.tensorflow.proto.framework.SummaryMetadata.PluginData getDefaultInstanceForType() {
-        return org.tensorflow.proto.framework.SummaryMetadata.PluginData.getDefaultInstance();
-      }
-
-      @java.lang.Override
-      public org.tensorflow.proto.framework.SummaryMetadata.PluginData build() {
-        org.tensorflow.proto.framework.SummaryMetadata.PluginData result = buildPartial();
-        if (!result.isInitialized()) {
-          throw newUninitializedMessageException(result);
-        }
-        return result;
-      }
-
-      @java.lang.Override
-      public org.tensorflow.proto.framework.SummaryMetadata.PluginData buildPartial() {
-        org.tensorflow.proto.framework.SummaryMetadata.PluginData result = new org.tensorflow.proto.framework.SummaryMetadata.PluginData(this);
-        result.pluginName_ = pluginName_;
-        result.content_ = content_;
-        onBuilt();
-        return result;
-      }
-
-      @java.lang.Override
-      public Builder clone() {
-        return super.clone();
-      }
-      @java.lang.Override
-      public Builder setField(
-          com.google.protobuf.Descriptors.FieldDescriptor field,
-          java.lang.Object value) {
-        return super.setField(field, value);
-      }
-      @java.lang.Override
-      public Builder clearField(
-          com.google.protobuf.Descriptors.FieldDescriptor field) {
-        return super.clearField(field);
-      }
-      @java.lang.Override
-      public Builder clearOneof(
-          com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-        return super.clearOneof(oneof);
-      }
-      @java.lang.Override
-      public Builder setRepeatedField(
-          com.google.protobuf.Descriptors.FieldDescriptor field,
-          int index, java.lang.Object value) {
-        return super.setRepeatedField(field, index, value);
-      }
-      @java.lang.Override
-      public Builder addRepeatedField(
-          com.google.protobuf.Descriptors.FieldDescriptor field,
-          java.lang.Object value) {
-        return super.addRepeatedField(field, value);
-      }
-      @java.lang.Override
-      public Builder mergeFrom(com.google.protobuf.Message other) {
-        if (other instanceof org.tensorflow.proto.framework.SummaryMetadata.PluginData) {
-          return mergeFrom((org.tensorflow.proto.framework.SummaryMetadata.PluginData)other);
-        } else {
-          super.mergeFrom(other);
-          return this;
-        }
-      }
-
-      public Builder mergeFrom(org.tensorflow.proto.framework.SummaryMetadata.PluginData other) {
-        if (other == org.tensorflow.proto.framework.SummaryMetadata.PluginData.getDefaultInstance()) return this;
-        if (!other.getPluginName().isEmpty()) {
-          pluginName_ = other.pluginName_;
-          onChanged();
-        }
-        if (other.getContent() != com.google.protobuf.ByteString.EMPTY) {
-          setContent(other.getContent());
-        }
-        this.mergeUnknownFields(other.unknownFields);
-        onChanged();
-        return this;
-      }
-
-      @java.lang.Override
-      public final boolean isInitialized() {
-        return true;
-      }
-
-      @java.lang.Override
-      public Builder mergeFrom(
-          com.google.protobuf.CodedInputStream input,
-          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-          throws java.io.IOException {
-        org.tensorflow.proto.framework.SummaryMetadata.PluginData parsedMessage = null;
-        try {
-          parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-        } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-          parsedMessage = (org.tensorflow.proto.framework.SummaryMetadata.PluginData) e.getUnfinishedMessage();
-          throw e.unwrapIOException();
-        } finally {
-          if (parsedMessage != null) {
-            mergeFrom(parsedMessage);
-          }
-        }
-        return this;
-      }
-
-      private java.lang.Object pluginName_ = "";
-      /**
-       * <pre>
-       * The name of the plugin this data pertains to.
-       * </pre>
-       *
-       * <code>string plugin_name = 1;</code>
-       */
-      public java.lang.String getPluginName() {
-        java.lang.Object ref = pluginName_;
-        if (!(ref instanceof java.lang.String)) {
-          com.google.protobuf.ByteString bs =
-              (com.google.protobuf.ByteString) ref;
-          java.lang.String s = bs.toStringUtf8();
-          pluginName_ = s;
-          return s;
-        } else {
-          return (java.lang.String) ref;
-        }
-      }
-      /**
-       * <pre>
-       * The name of the plugin this data pertains to.
-       * </pre>
-       *
-       * <code>string plugin_name = 1;</code>
-       */
-      public com.google.protobuf.ByteString
-          getPluginNameBytes() {
-        java.lang.Object ref = pluginName_;
-        if (ref instanceof String) {
-          com.google.protobuf.ByteString b = 
-              com.google.protobuf.ByteString.copyFromUtf8(
-                  (java.lang.String) ref);
-          pluginName_ = b;
-          return b;
-        } else {
-          return (com.google.protobuf.ByteString) ref;
-        }
-      }
-      /**
-       * <pre>
-       * The name of the plugin this data pertains to.
-       * </pre>
-       *
-       * <code>string plugin_name = 1;</code>
-       */
-      public Builder setPluginName(
-          java.lang.String value) {
-        if (value == null) {
-    throw new NullPointerException();
-  }
-  
-        pluginName_ = value;
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * The name of the plugin this data pertains to.
-       * </pre>
-       *
-       * <code>string plugin_name = 1;</code>
-       */
-      public Builder clearPluginName() {
-        
-        pluginName_ = getDefaultInstance().getPluginName();
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * The name of the plugin this data pertains to.
-       * </pre>
-       *
-       * <code>string plugin_name = 1;</code>
-       */
-      public Builder setPluginNameBytes(
-          com.google.protobuf.ByteString value) {
-        if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-        
-        pluginName_ = value;
-        onChanged();
-        return this;
-      }
-
-      private com.google.protobuf.ByteString content_ = com.google.protobuf.ByteString.EMPTY;
-      /**
-       * <pre>
-       * The content to store for the plugin. The best practice is for this to be
-       * a binary serialized protocol buffer.
-       * </pre>
-       *
-       * <code>bytes content = 2;</code>
-       */
-      public com.google.protobuf.ByteString getContent() {
-        return content_;
-      }
-      /**
-       * <pre>
-       * The content to store for the plugin. The best practice is for this to be
-       * a binary serialized protocol buffer.
-       * </pre>
-       *
-       * <code>bytes content = 2;</code>
-       */
-      public Builder setContent(com.google.protobuf.ByteString value) {
-        if (value == null) {
-    throw new NullPointerException();
-  }
-  
-        content_ = value;
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * The content to store for the plugin. The best practice is for this to be
-       * a binary serialized protocol buffer.
-       * </pre>
-       *
-       * <code>bytes content = 2;</code>
-       */
-      public Builder clearContent() {
-        
-        content_ = getDefaultInstance().getContent();
-        onChanged();
-        return this;
-      }
-      @java.lang.Override
-      public final Builder setUnknownFields(
-          final com.google.protobuf.UnknownFieldSet unknownFields) {
-        return super.setUnknownFields(unknownFields);
-      }
-
-      @java.lang.Override
-      public final Builder mergeUnknownFields(
-          final com.google.protobuf.UnknownFieldSet unknownFields) {
-        return super.mergeUnknownFields(unknownFields);
-      }
-
-
-      // @@protoc_insertion_point(builder_scope:tensorflow.SummaryMetadata.PluginData)
-    }
-
-    // @@protoc_insertion_point(class_scope:tensorflow.SummaryMetadata.PluginData)
-    private static final org.tensorflow.proto.framework.SummaryMetadata.PluginData DEFAULT_INSTANCE;
-    static {
-      DEFAULT_INSTANCE = new org.tensorflow.proto.framework.SummaryMetadata.PluginData();
-    }
-
-    public static org.tensorflow.proto.framework.SummaryMetadata.PluginData getDefaultInstance() {
-      return DEFAULT_INSTANCE;
-    }
-
-    private static final com.google.protobuf.Parser<PluginData>
-        PARSER = new com.google.protobuf.AbstractParser<PluginData>() {
-      @java.lang.Override
-      public PluginData parsePartialFrom(
-          com.google.protobuf.CodedInputStream input,
-          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-          throws com.google.protobuf.InvalidProtocolBufferException {
-        return new PluginData(input, extensionRegistry);
-      }
-    };
-
-    public static com.google.protobuf.Parser<PluginData> parser() {
-      return PARSER;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Parser<PluginData> getParserForType() {
-      return PARSER;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.SummaryMetadata.PluginData getDefaultInstanceForType() {
-      return DEFAULT_INSTANCE;
-    }
-
-  }
-
-  public static final int PLUGIN_DATA_FIELD_NUMBER = 1;
-  private org.tensorflow.proto.framework.SummaryMetadata.PluginData pluginData_;
-  /**
-   * <pre>
-   * Data that associates a summary with a certain plugin.
-   * </pre>
-   *
-   * <code>.tensorflow.SummaryMetadata.PluginData plugin_data = 1;</code>
-   */
-  public boolean hasPluginData() {
-    return pluginData_ != null;
-  }
-  /**
-   * <pre>
-   * Data that associates a summary with a certain plugin.
-   * </pre>
-   *
-   * <code>.tensorflow.SummaryMetadata.PluginData plugin_data = 1;</code>
-   */
-  public org.tensorflow.proto.framework.SummaryMetadata.PluginData getPluginData() {
-    return pluginData_ == null ? org.tensorflow.proto.framework.SummaryMetadata.PluginData.getDefaultInstance() : pluginData_;
-  }
-  /**
-   * <pre>
-   * Data that associates a summary with a certain plugin.
-   * </pre>
-   *
-   * <code>.tensorflow.SummaryMetadata.PluginData plugin_data = 1;</code>
-   */
-  public org.tensorflow.proto.framework.SummaryMetadata.PluginDataOrBuilder getPluginDataOrBuilder() {
-    return getPluginData();
-  }
-
-  public static final int DISPLAY_NAME_FIELD_NUMBER = 2;
-  private volatile java.lang.Object displayName_;
-  /**
-   * <pre>
-   * Display name for viewing in TensorBoard.
-   * </pre>
-   *
-   * <code>string display_name = 2;</code>
-   */
-  public java.lang.String getDisplayName() {
-    java.lang.Object ref = displayName_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      displayName_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * Display name for viewing in TensorBoard.
-   * </pre>
-   *
-   * <code>string display_name = 2;</code>
-   */
-  public com.google.protobuf.ByteString
-      getDisplayNameBytes() {
-    java.lang.Object ref = displayName_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      displayName_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int SUMMARY_DESCRIPTION_FIELD_NUMBER = 3;
-  private volatile java.lang.Object summaryDescription_;
-  /**
-   * <pre>
-   * Longform readable description of the summary sequence. Markdown supported.
-   * </pre>
-   *
-   * <code>string summary_description = 3;</code>
-   */
-  public java.lang.String getSummaryDescription() {
-    java.lang.Object ref = summaryDescription_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      summaryDescription_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * Longform readable description of the summary sequence. Markdown supported.
-   * </pre>
-   *
-   * <code>string summary_description = 3;</code>
-   */
-  public com.google.protobuf.ByteString
-      getSummaryDescriptionBytes() {
-    java.lang.Object ref = summaryDescription_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      summaryDescription_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int DATA_CLASS_FIELD_NUMBER = 4;
-  private int dataClass_;
-  /**
-   * <pre>
-   * Class of data stored in this time series. Required for compatibility with
-   * TensorBoard's generic data facilities (`DataProvider`, et al.). This value
-   * imposes constraints on the dtype and shape of the corresponding tensor
-   * values. See `DataClass` docs for details.
-   * </pre>
-   *
-   * <code>.tensorflow.DataClass data_class = 4;</code>
-   */
-  public int getDataClassValue() {
-    return dataClass_;
-  }
-  /**
-   * <pre>
-   * Class of data stored in this time series. Required for compatibility with
-   * TensorBoard's generic data facilities (`DataProvider`, et al.). This value
-   * imposes constraints on the dtype and shape of the corresponding tensor
-   * values. See `DataClass` docs for details.
-   * </pre>
-   *
-   * <code>.tensorflow.DataClass data_class = 4;</code>
-   */
-  public org.tensorflow.proto.framework.DataClass getDataClass() {
-    @SuppressWarnings("deprecation")
-    org.tensorflow.proto.framework.DataClass result = org.tensorflow.proto.framework.DataClass.valueOf(dataClass_);
-    return result == null ? org.tensorflow.proto.framework.DataClass.UNRECOGNIZED : result;
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    if (pluginData_ != null) {
-      output.writeMessage(1, getPluginData());
-    }
-    if (!getDisplayNameBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 2, displayName_);
-    }
-    if (!getSummaryDescriptionBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 3, summaryDescription_);
-    }
-    if (dataClass_ != org.tensorflow.proto.framework.DataClass.DATA_CLASS_UNKNOWN.getNumber()) {
-      output.writeEnum(4, dataClass_);
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (pluginData_ != null) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(1, getPluginData());
-    }
-    if (!getDisplayNameBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(2, displayName_);
-    }
-    if (!getSummaryDescriptionBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(3, summaryDescription_);
-    }
-    if (dataClass_ != org.tensorflow.proto.framework.DataClass.DATA_CLASS_UNKNOWN.getNumber()) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeEnumSize(4, dataClass_);
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.framework.SummaryMetadata)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.framework.SummaryMetadata other = (org.tensorflow.proto.framework.SummaryMetadata) obj;
-
-    if (hasPluginData() != other.hasPluginData()) return false;
-    if (hasPluginData()) {
-      if (!getPluginData()
-          .equals(other.getPluginData())) return false;
-    }
-    if (!getDisplayName()
-        .equals(other.getDisplayName())) return false;
-    if (!getSummaryDescription()
-        .equals(other.getSummaryDescription())) return false;
-    if (dataClass_ != other.dataClass_) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    if (hasPluginData()) {
-      hash = (37 * hash) + PLUGIN_DATA_FIELD_NUMBER;
-      hash = (53 * hash) + getPluginData().hashCode();
-    }
-    hash = (37 * hash) + DISPLAY_NAME_FIELD_NUMBER;
-    hash = (53 * hash) + getDisplayName().hashCode();
-    hash = (37 * hash) + SUMMARY_DESCRIPTION_FIELD_NUMBER;
-    hash = (53 * hash) + getSummaryDescription().hashCode();
-    hash = (37 * hash) + DATA_CLASS_FIELD_NUMBER;
-    hash = (53 * hash) + dataClass_;
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.framework.SummaryMetadata parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.SummaryMetadata parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.SummaryMetadata parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.SummaryMetadata parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.SummaryMetadata parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.SummaryMetadata parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.SummaryMetadata parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.SummaryMetadata parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.SummaryMetadata parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.SummaryMetadata parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.SummaryMetadata parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.SummaryMetadata parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.framework.SummaryMetadata prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * <pre>
-   * A SummaryMetadata encapsulates information on which plugins are able to make
-   * use of a certain summary value.
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.SummaryMetadata}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.SummaryMetadata)
-      org.tensorflow.proto.framework.SummaryMetadataOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.SummaryProtos.internal_static_tensorflow_SummaryMetadata_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.SummaryProtos.internal_static_tensorflow_SummaryMetadata_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.SummaryMetadata.class, org.tensorflow.proto.framework.SummaryMetadata.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.framework.SummaryMetadata.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      if (pluginDataBuilder_ == null) {
-        pluginData_ = null;
-      } else {
-        pluginData_ = null;
-        pluginDataBuilder_ = null;
-      }
-      displayName_ = "";
-
-      summaryDescription_ = "";
-
-      dataClass_ = 0;
-
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.framework.SummaryProtos.internal_static_tensorflow_SummaryMetadata_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.SummaryMetadata getDefaultInstanceForType() {
-      return org.tensorflow.proto.framework.SummaryMetadata.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.SummaryMetadata build() {
-      org.tensorflow.proto.framework.SummaryMetadata result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.SummaryMetadata buildPartial() {
-      org.tensorflow.proto.framework.SummaryMetadata result = new org.tensorflow.proto.framework.SummaryMetadata(this);
-      if (pluginDataBuilder_ == null) {
-        result.pluginData_ = pluginData_;
-      } else {
-        result.pluginData_ = pluginDataBuilder_.build();
-      }
-      result.displayName_ = displayName_;
-      result.summaryDescription_ = summaryDescription_;
-      result.dataClass_ = dataClass_;
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.framework.SummaryMetadata) {
-        return mergeFrom((org.tensorflow.proto.framework.SummaryMetadata)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.framework.SummaryMetadata other) {
-      if (other == org.tensorflow.proto.framework.SummaryMetadata.getDefaultInstance()) return this;
-      if (other.hasPluginData()) {
-        mergePluginData(other.getPluginData());
-      }
-      if (!other.getDisplayName().isEmpty()) {
-        displayName_ = other.displayName_;
-        onChanged();
-      }
-      if (!other.getSummaryDescription().isEmpty()) {
-        summaryDescription_ = other.summaryDescription_;
-        onChanged();
-      }
-      if (other.dataClass_ != 0) {
-        setDataClassValue(other.getDataClassValue());
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.framework.SummaryMetadata parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.framework.SummaryMetadata) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-
-    private org.tensorflow.proto.framework.SummaryMetadata.PluginData pluginData_;
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.SummaryMetadata.PluginData, org.tensorflow.proto.framework.SummaryMetadata.PluginData.Builder, org.tensorflow.proto.framework.SummaryMetadata.PluginDataOrBuilder> pluginDataBuilder_;
-    /**
-     * <pre>
-     * Data that associates a summary with a certain plugin.
-     * </pre>
-     *
-     * <code>.tensorflow.SummaryMetadata.PluginData plugin_data = 1;</code>
-     */
-    public boolean hasPluginData() {
-      return pluginDataBuilder_ != null || pluginData_ != null;
-    }
-    /**
-     * <pre>
-     * Data that associates a summary with a certain plugin.
-     * </pre>
-     *
-     * <code>.tensorflow.SummaryMetadata.PluginData plugin_data = 1;</code>
-     */
-    public org.tensorflow.proto.framework.SummaryMetadata.PluginData getPluginData() {
-      if (pluginDataBuilder_ == null) {
-        return pluginData_ == null ? org.tensorflow.proto.framework.SummaryMetadata.PluginData.getDefaultInstance() : pluginData_;
-      } else {
-        return pluginDataBuilder_.getMessage();
-      }
-    }
-    /**
-     * <pre>
-     * Data that associates a summary with a certain plugin.
-     * </pre>
-     *
-     * <code>.tensorflow.SummaryMetadata.PluginData plugin_data = 1;</code>
-     */
-    public Builder setPluginData(org.tensorflow.proto.framework.SummaryMetadata.PluginData value) {
-      if (pluginDataBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        pluginData_ = value;
-        onChanged();
-      } else {
-        pluginDataBuilder_.setMessage(value);
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Data that associates a summary with a certain plugin.
-     * </pre>
-     *
-     * <code>.tensorflow.SummaryMetadata.PluginData plugin_data = 1;</code>
-     */
-    public Builder setPluginData(
-        org.tensorflow.proto.framework.SummaryMetadata.PluginData.Builder builderForValue) {
-      if (pluginDataBuilder_ == null) {
-        pluginData_ = builderForValue.build();
-        onChanged();
-      } else {
-        pluginDataBuilder_.setMessage(builderForValue.build());
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Data that associates a summary with a certain plugin.
-     * </pre>
-     *
-     * <code>.tensorflow.SummaryMetadata.PluginData plugin_data = 1;</code>
-     */
-    public Builder mergePluginData(org.tensorflow.proto.framework.SummaryMetadata.PluginData value) {
-      if (pluginDataBuilder_ == null) {
-        if (pluginData_ != null) {
-          pluginData_ =
-            org.tensorflow.proto.framework.SummaryMetadata.PluginData.newBuilder(pluginData_).mergeFrom(value).buildPartial();
-        } else {
-          pluginData_ = value;
-        }
-        onChanged();
-      } else {
-        pluginDataBuilder_.mergeFrom(value);
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Data that associates a summary with a certain plugin.
-     * </pre>
-     *
-     * <code>.tensorflow.SummaryMetadata.PluginData plugin_data = 1;</code>
-     */
-    public Builder clearPluginData() {
-      if (pluginDataBuilder_ == null) {
-        pluginData_ = null;
-        onChanged();
-      } else {
-        pluginData_ = null;
-        pluginDataBuilder_ = null;
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Data that associates a summary with a certain plugin.
-     * </pre>
-     *
-     * <code>.tensorflow.SummaryMetadata.PluginData plugin_data = 1;</code>
-     */
-    public org.tensorflow.proto.framework.SummaryMetadata.PluginData.Builder getPluginDataBuilder() {
-      
-      onChanged();
-      return getPluginDataFieldBuilder().getBuilder();
-    }
-    /**
-     * <pre>
-     * Data that associates a summary with a certain plugin.
-     * </pre>
-     *
-     * <code>.tensorflow.SummaryMetadata.PluginData plugin_data = 1;</code>
-     */
-    public org.tensorflow.proto.framework.SummaryMetadata.PluginDataOrBuilder getPluginDataOrBuilder() {
-      if (pluginDataBuilder_ != null) {
-        return pluginDataBuilder_.getMessageOrBuilder();
-      } else {
-        return pluginData_ == null ?
-            org.tensorflow.proto.framework.SummaryMetadata.PluginData.getDefaultInstance() : pluginData_;
-      }
-    }
-    /**
-     * <pre>
-     * Data that associates a summary with a certain plugin.
-     * </pre>
-     *
-     * <code>.tensorflow.SummaryMetadata.PluginData plugin_data = 1;</code>
-     */
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.SummaryMetadata.PluginData, org.tensorflow.proto.framework.SummaryMetadata.PluginData.Builder, org.tensorflow.proto.framework.SummaryMetadata.PluginDataOrBuilder> 
-        getPluginDataFieldBuilder() {
-      if (pluginDataBuilder_ == null) {
-        pluginDataBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-            org.tensorflow.proto.framework.SummaryMetadata.PluginData, org.tensorflow.proto.framework.SummaryMetadata.PluginData.Builder, org.tensorflow.proto.framework.SummaryMetadata.PluginDataOrBuilder>(
-                getPluginData(),
-                getParentForChildren(),
-                isClean());
-        pluginData_ = null;
-      }
-      return pluginDataBuilder_;
-    }
-
-    private java.lang.Object displayName_ = "";
-    /**
-     * <pre>
-     * Display name for viewing in TensorBoard.
-     * </pre>
-     *
-     * <code>string display_name = 2;</code>
-     */
-    public java.lang.String getDisplayName() {
-      java.lang.Object ref = displayName_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        displayName_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Display name for viewing in TensorBoard.
-     * </pre>
-     *
-     * <code>string display_name = 2;</code>
-     */
-    public com.google.protobuf.ByteString
-        getDisplayNameBytes() {
-      java.lang.Object ref = displayName_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        displayName_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Display name for viewing in TensorBoard.
-     * </pre>
-     *
-     * <code>string display_name = 2;</code>
-     */
-    public Builder setDisplayName(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      displayName_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Display name for viewing in TensorBoard.
-     * </pre>
-     *
-     * <code>string display_name = 2;</code>
-     */
-    public Builder clearDisplayName() {
-      
-      displayName_ = getDefaultInstance().getDisplayName();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Display name for viewing in TensorBoard.
-     * </pre>
-     *
-     * <code>string display_name = 2;</code>
-     */
-    public Builder setDisplayNameBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      displayName_ = value;
-      onChanged();
-      return this;
-    }
-
-    private java.lang.Object summaryDescription_ = "";
-    /**
-     * <pre>
-     * Longform readable description of the summary sequence. Markdown supported.
-     * </pre>
-     *
-     * <code>string summary_description = 3;</code>
-     */
-    public java.lang.String getSummaryDescription() {
-      java.lang.Object ref = summaryDescription_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        summaryDescription_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Longform readable description of the summary sequence. Markdown supported.
-     * </pre>
-     *
-     * <code>string summary_description = 3;</code>
-     */
-    public com.google.protobuf.ByteString
-        getSummaryDescriptionBytes() {
-      java.lang.Object ref = summaryDescription_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        summaryDescription_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Longform readable description of the summary sequence. Markdown supported.
-     * </pre>
-     *
-     * <code>string summary_description = 3;</code>
-     */
-    public Builder setSummaryDescription(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      summaryDescription_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Longform readable description of the summary sequence. Markdown supported.
-     * </pre>
-     *
-     * <code>string summary_description = 3;</code>
-     */
-    public Builder clearSummaryDescription() {
-      
-      summaryDescription_ = getDefaultInstance().getSummaryDescription();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Longform readable description of the summary sequence. Markdown supported.
-     * </pre>
-     *
-     * <code>string summary_description = 3;</code>
-     */
-    public Builder setSummaryDescriptionBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      summaryDescription_ = value;
-      onChanged();
-      return this;
-    }
-
-    private int dataClass_ = 0;
-    /**
-     * <pre>
-     * Class of data stored in this time series. Required for compatibility with
-     * TensorBoard's generic data facilities (`DataProvider`, et al.). This value
-     * imposes constraints on the dtype and shape of the corresponding tensor
-     * values. See `DataClass` docs for details.
-     * </pre>
-     *
-     * <code>.tensorflow.DataClass data_class = 4;</code>
-     */
-    public int getDataClassValue() {
-      return dataClass_;
-    }
-    /**
-     * <pre>
-     * Class of data stored in this time series. Required for compatibility with
-     * TensorBoard's generic data facilities (`DataProvider`, et al.). This value
-     * imposes constraints on the dtype and shape of the corresponding tensor
-     * values. See `DataClass` docs for details.
-     * </pre>
-     *
-     * <code>.tensorflow.DataClass data_class = 4;</code>
-     */
-    public Builder setDataClassValue(int value) {
-      dataClass_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Class of data stored in this time series. Required for compatibility with
-     * TensorBoard's generic data facilities (`DataProvider`, et al.). This value
-     * imposes constraints on the dtype and shape of the corresponding tensor
-     * values. See `DataClass` docs for details.
-     * </pre>
-     *
-     * <code>.tensorflow.DataClass data_class = 4;</code>
-     */
-    public org.tensorflow.proto.framework.DataClass getDataClass() {
-      @SuppressWarnings("deprecation")
-      org.tensorflow.proto.framework.DataClass result = org.tensorflow.proto.framework.DataClass.valueOf(dataClass_);
-      return result == null ? org.tensorflow.proto.framework.DataClass.UNRECOGNIZED : result;
-    }
-    /**
-     * <pre>
-     * Class of data stored in this time series. Required for compatibility with
-     * TensorBoard's generic data facilities (`DataProvider`, et al.). This value
-     * imposes constraints on the dtype and shape of the corresponding tensor
-     * values. See `DataClass` docs for details.
-     * </pre>
-     *
-     * <code>.tensorflow.DataClass data_class = 4;</code>
-     */
-    public Builder setDataClass(org.tensorflow.proto.framework.DataClass value) {
-      if (value == null) {
-        throw new NullPointerException();
-      }
-      
-      dataClass_ = value.getNumber();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Class of data stored in this time series. Required for compatibility with
-     * TensorBoard's generic data facilities (`DataProvider`, et al.). This value
-     * imposes constraints on the dtype and shape of the corresponding tensor
-     * values. See `DataClass` docs for details.
-     * </pre>
-     *
-     * <code>.tensorflow.DataClass data_class = 4;</code>
-     */
-    public Builder clearDataClass() {
-      
-      dataClass_ = 0;
-      onChanged();
-      return this;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.SummaryMetadata)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.SummaryMetadata)
-  private static final org.tensorflow.proto.framework.SummaryMetadata DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.SummaryMetadata();
-  }
-
-  public static org.tensorflow.proto.framework.SummaryMetadata getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<SummaryMetadata>
-      PARSER = new com.google.protobuf.AbstractParser<SummaryMetadata>() {
-    @java.lang.Override
-    public SummaryMetadata parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new SummaryMetadata(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<SummaryMetadata> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<SummaryMetadata> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.framework.SummaryMetadata getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SummaryMetadataOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SummaryMetadataOrBuilder.java
deleted file mode 100644
index 9fd35f7d1de..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SummaryMetadataOrBuilder.java
+++ /dev/null
@@ -1,93 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/framework/summary.proto
-
-package org.tensorflow.proto.framework;
-
-public interface SummaryMetadataOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.SummaryMetadata)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <pre>
-   * Data that associates a summary with a certain plugin.
-   * </pre>
-   *
-   * <code>.tensorflow.SummaryMetadata.PluginData plugin_data = 1;</code>
-   */
-  boolean hasPluginData();
-  /**
-   * <pre>
-   * Data that associates a summary with a certain plugin.
-   * </pre>
-   *
-   * <code>.tensorflow.SummaryMetadata.PluginData plugin_data = 1;</code>
-   */
-  org.tensorflow.proto.framework.SummaryMetadata.PluginData getPluginData();
-  /**
-   * <pre>
-   * Data that associates a summary with a certain plugin.
-   * </pre>
-   *
-   * <code>.tensorflow.SummaryMetadata.PluginData plugin_data = 1;</code>
-   */
-  org.tensorflow.proto.framework.SummaryMetadata.PluginDataOrBuilder getPluginDataOrBuilder();
-
-  /**
-   * <pre>
-   * Display name for viewing in TensorBoard.
-   * </pre>
-   *
-   * <code>string display_name = 2;</code>
-   */
-  java.lang.String getDisplayName();
-  /**
-   * <pre>
-   * Display name for viewing in TensorBoard.
-   * </pre>
-   *
-   * <code>string display_name = 2;</code>
-   */
-  com.google.protobuf.ByteString
-      getDisplayNameBytes();
-
-  /**
-   * <pre>
-   * Longform readable description of the summary sequence. Markdown supported.
-   * </pre>
-   *
-   * <code>string summary_description = 3;</code>
-   */
-  java.lang.String getSummaryDescription();
-  /**
-   * <pre>
-   * Longform readable description of the summary sequence. Markdown supported.
-   * </pre>
-   *
-   * <code>string summary_description = 3;</code>
-   */
-  com.google.protobuf.ByteString
-      getSummaryDescriptionBytes();
-
-  /**
-   * <pre>
-   * Class of data stored in this time series. Required for compatibility with
-   * TensorBoard's generic data facilities (`DataProvider`, et al.). This value
-   * imposes constraints on the dtype and shape of the corresponding tensor
-   * values. See `DataClass` docs for details.
-   * </pre>
-   *
-   * <code>.tensorflow.DataClass data_class = 4;</code>
-   */
-  int getDataClassValue();
-  /**
-   * <pre>
-   * Class of data stored in this time series. Required for compatibility with
-   * TensorBoard's generic data facilities (`DataProvider`, et al.). This value
-   * imposes constraints on the dtype and shape of the corresponding tensor
-   * values. See `DataClass` docs for details.
-   * </pre>
-   *
-   * <code>.tensorflow.DataClass data_class = 4;</code>
-   */
-  org.tensorflow.proto.framework.DataClass getDataClass();
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SummaryOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SummaryOrBuilder.java
deleted file mode 100644
index 09de7983ac5..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SummaryOrBuilder.java
+++ /dev/null
@@ -1,53 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/framework/summary.proto
-
-package org.tensorflow.proto.framework;
-
-public interface SummaryOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.Summary)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <pre>
-   * Set of values for the summary.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.Summary.Value value = 1;</code>
-   */
-  java.util.List<org.tensorflow.proto.framework.Summary.Value> 
-      getValueList();
-  /**
-   * <pre>
-   * Set of values for the summary.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.Summary.Value value = 1;</code>
-   */
-  org.tensorflow.proto.framework.Summary.Value getValue(int index);
-  /**
-   * <pre>
-   * Set of values for the summary.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.Summary.Value value = 1;</code>
-   */
-  int getValueCount();
-  /**
-   * <pre>
-   * Set of values for the summary.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.Summary.Value value = 1;</code>
-   */
-  java.util.List<? extends org.tensorflow.proto.framework.Summary.ValueOrBuilder> 
-      getValueOrBuilderList();
-  /**
-   * <pre>
-   * Set of values for the summary.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.Summary.Value value = 1;</code>
-   */
-  org.tensorflow.proto.framework.Summary.ValueOrBuilder getValueOrBuilder(
-      int index);
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SummaryProtos.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SummaryProtos.java
deleted file mode 100644
index 8e0373a5f18..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SummaryProtos.java
+++ /dev/null
@@ -1,159 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/framework/summary.proto
-
-package org.tensorflow.proto.framework;
-
-public final class SummaryProtos {
-  private SummaryProtos() {}
-  public static void registerAllExtensions(
-      com.google.protobuf.ExtensionRegistryLite registry) {
-  }
-
-  public static void registerAllExtensions(
-      com.google.protobuf.ExtensionRegistry registry) {
-    registerAllExtensions(
-        (com.google.protobuf.ExtensionRegistryLite) registry);
-  }
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_SummaryDescription_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_SummaryDescription_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_HistogramProto_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_HistogramProto_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_SummaryMetadata_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_SummaryMetadata_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_SummaryMetadata_PluginData_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_SummaryMetadata_PluginData_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_Summary_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_Summary_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_Summary_Image_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_Summary_Image_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_Summary_Audio_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_Summary_Audio_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_Summary_Value_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_Summary_Value_fieldAccessorTable;
-
-  public static com.google.protobuf.Descriptors.FileDescriptor
-      getDescriptor() {
-    return descriptor;
-  }
-  private static  com.google.protobuf.Descriptors.FileDescriptor
-      descriptor;
-  static {
-    java.lang.String[] descriptorData = {
-      "\n\'tensorflow/core/framework/summary.prot" +
-      "o\022\ntensorflow\032&tensorflow/core/framework" +
-      "/tensor.proto\"\'\n\022SummaryDescription\022\021\n\tt" +
-      "ype_hint\030\001 \001(\t\"\207\001\n\016HistogramProto\022\013\n\003min" +
-      "\030\001 \001(\001\022\013\n\003max\030\002 \001(\001\022\013\n\003num\030\003 \001(\001\022\013\n\003sum\030" +
-      "\004 \001(\001\022\023\n\013sum_squares\030\005 \001(\001\022\030\n\014bucket_lim" +
-      "it\030\006 \003(\001B\002\020\001\022\022\n\006bucket\030\007 \003(\001B\002\020\001\"\340\001\n\017Sum" +
-      "maryMetadata\022;\n\013plugin_data\030\001 \001(\0132&.tens" +
-      "orflow.SummaryMetadata.PluginData\022\024\n\014dis" +
-      "play_name\030\002 \001(\t\022\033\n\023summary_description\030\003" +
-      " \001(\t\022)\n\ndata_class\030\004 \001(\0162\025.tensorflow.Da" +
-      "taClass\0322\n\nPluginData\022\023\n\013plugin_name\030\001 \001" +
-      "(\t\022\017\n\007content\030\002 \001(\014\"\336\004\n\007Summary\022(\n\005value" +
-      "\030\001 \003(\0132\031.tensorflow.Summary.Value\032X\n\005Ima" +
-      "ge\022\016\n\006height\030\001 \001(\005\022\r\n\005width\030\002 \001(\005\022\022\n\ncol" +
-      "orspace\030\003 \001(\005\022\034\n\024encoded_image_string\030\004 " +
-      "\001(\014\032}\n\005Audio\022\023\n\013sample_rate\030\001 \001(\002\022\024\n\014num" +
-      "_channels\030\002 \001(\003\022\025\n\rlength_frames\030\003 \001(\003\022\034" +
-      "\n\024encoded_audio_string\030\004 \001(\014\022\024\n\014content_" +
-      "type\030\005 \001(\t\032\317\002\n\005Value\022\021\n\tnode_name\030\007 \001(\t\022" +
-      "\013\n\003tag\030\001 \001(\t\022-\n\010metadata\030\t \001(\0132\033.tensorf" +
-      "low.SummaryMetadata\022\026\n\014simple_value\030\002 \001(" +
-      "\002H\000\022&\n\034obsolete_old_style_histogram\030\003 \001(" +
-      "\014H\000\022*\n\005image\030\004 \001(\0132\031.tensorflow.Summary." +
-      "ImageH\000\022+\n\005histo\030\005 \001(\0132\032.tensorflow.Hist" +
-      "ogramProtoH\000\022*\n\005audio\030\006 \001(\0132\031.tensorflow" +
-      ".Summary.AudioH\000\022)\n\006tensor\030\010 \001(\0132\027.tenso" +
-      "rflow.TensorProtoH\000B\007\n\005value*o\n\tDataClas" +
-      "s\022\026\n\022DATA_CLASS_UNKNOWN\020\000\022\025\n\021DATA_CLASS_" +
-      "SCALAR\020\001\022\025\n\021DATA_CLASS_TENSOR\020\002\022\034\n\030DATA_" +
-      "CLASS_BLOB_SEQUENCE\020\003B\204\001\n\036org.tensorflow" +
-      ".proto.frameworkB\rSummaryProtosP\001ZNgithu" +
-      "b.com/tensorflow/tensorflow/tensorflow/g" +
-      "o/core/framework/summary_go_proto\370\001\001b\006pr" +
-      "oto3"
-    };
-    descriptor = com.google.protobuf.Descriptors.FileDescriptor
-      .internalBuildGeneratedFileFrom(descriptorData,
-        new com.google.protobuf.Descriptors.FileDescriptor[] {
-          org.tensorflow.proto.framework.TensorProtos.getDescriptor(),
-        });
-    internal_static_tensorflow_SummaryDescription_descriptor =
-      getDescriptor().getMessageTypes().get(0);
-    internal_static_tensorflow_SummaryDescription_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_SummaryDescription_descriptor,
-        new java.lang.String[] { "TypeHint", });
-    internal_static_tensorflow_HistogramProto_descriptor =
-      getDescriptor().getMessageTypes().get(1);
-    internal_static_tensorflow_HistogramProto_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_HistogramProto_descriptor,
-        new java.lang.String[] { "Min", "Max", "Num", "Sum", "SumSquares", "BucketLimit", "Bucket", });
-    internal_static_tensorflow_SummaryMetadata_descriptor =
-      getDescriptor().getMessageTypes().get(2);
-    internal_static_tensorflow_SummaryMetadata_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_SummaryMetadata_descriptor,
-        new java.lang.String[] { "PluginData", "DisplayName", "SummaryDescription", "DataClass", });
-    internal_static_tensorflow_SummaryMetadata_PluginData_descriptor =
-      internal_static_tensorflow_SummaryMetadata_descriptor.getNestedTypes().get(0);
-    internal_static_tensorflow_SummaryMetadata_PluginData_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_SummaryMetadata_PluginData_descriptor,
-        new java.lang.String[] { "PluginName", "Content", });
-    internal_static_tensorflow_Summary_descriptor =
-      getDescriptor().getMessageTypes().get(3);
-    internal_static_tensorflow_Summary_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_Summary_descriptor,
-        new java.lang.String[] { "Value", });
-    internal_static_tensorflow_Summary_Image_descriptor =
-      internal_static_tensorflow_Summary_descriptor.getNestedTypes().get(0);
-    internal_static_tensorflow_Summary_Image_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_Summary_Image_descriptor,
-        new java.lang.String[] { "Height", "Width", "Colorspace", "EncodedImageString", });
-    internal_static_tensorflow_Summary_Audio_descriptor =
-      internal_static_tensorflow_Summary_descriptor.getNestedTypes().get(1);
-    internal_static_tensorflow_Summary_Audio_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_Summary_Audio_descriptor,
-        new java.lang.String[] { "SampleRate", "NumChannels", "LengthFrames", "EncodedAudioString", "ContentType", });
-    internal_static_tensorflow_Summary_Value_descriptor =
-      internal_static_tensorflow_Summary_descriptor.getNestedTypes().get(2);
-    internal_static_tensorflow_Summary_Value_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_Summary_Value_descriptor,
-        new java.lang.String[] { "NodeName", "Tag", "Metadata", "SimpleValue", "ObsoleteOldStyleHistogram", "Image", "Histo", "Audio", "Tensor", "Value", });
-    org.tensorflow.proto.framework.TensorProtos.getDescriptor();
-  }
-
-  // @@protoc_insertion_point(outer_class_scope)
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorConnection.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorConnection.java
deleted file mode 100644
index 5be946ae717..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorConnection.java
+++ /dev/null
@@ -1,751 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/config.proto
-
-package org.tensorflow.proto.framework;
-
-/**
- * <pre>
- * Defines a connection between two tensors in a `GraphDef`.
- * </pre>
- *
- * Protobuf type {@code tensorflow.TensorConnection}
- */
-public  final class TensorConnection extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.TensorConnection)
-    TensorConnectionOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use TensorConnection.newBuilder() to construct.
-  private TensorConnection(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private TensorConnection() {
-    fromTensor_ = "";
-    toTensor_ = "";
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new TensorConnection();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private TensorConnection(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 10: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            fromTensor_ = s;
-            break;
-          }
-          case 18: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            toTensor_ = s;
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_TensorConnection_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_TensorConnection_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.framework.TensorConnection.class, org.tensorflow.proto.framework.TensorConnection.Builder.class);
-  }
-
-  public static final int FROM_TENSOR_FIELD_NUMBER = 1;
-  private volatile java.lang.Object fromTensor_;
-  /**
-   * <pre>
-   * A tensor name. The value of this tensor will be substituted for
-   * the tensor named in `to_tensor`.
-   * </pre>
-   *
-   * <code>string from_tensor = 1;</code>
-   */
-  public java.lang.String getFromTensor() {
-    java.lang.Object ref = fromTensor_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      fromTensor_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * A tensor name. The value of this tensor will be substituted for
-   * the tensor named in `to_tensor`.
-   * </pre>
-   *
-   * <code>string from_tensor = 1;</code>
-   */
-  public com.google.protobuf.ByteString
-      getFromTensorBytes() {
-    java.lang.Object ref = fromTensor_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      fromTensor_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int TO_TENSOR_FIELD_NUMBER = 2;
-  private volatile java.lang.Object toTensor_;
-  /**
-   * <pre>
-   * A tensor name. The value of this tensor will be bound to the
-   * value of the tensor named in `from_tensor`.
-   * </pre>
-   *
-   * <code>string to_tensor = 2;</code>
-   */
-  public java.lang.String getToTensor() {
-    java.lang.Object ref = toTensor_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      toTensor_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * A tensor name. The value of this tensor will be bound to the
-   * value of the tensor named in `from_tensor`.
-   * </pre>
-   *
-   * <code>string to_tensor = 2;</code>
-   */
-  public com.google.protobuf.ByteString
-      getToTensorBytes() {
-    java.lang.Object ref = toTensor_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      toTensor_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    if (!getFromTensorBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 1, fromTensor_);
-    }
-    if (!getToTensorBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 2, toTensor_);
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (!getFromTensorBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, fromTensor_);
-    }
-    if (!getToTensorBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(2, toTensor_);
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.framework.TensorConnection)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.framework.TensorConnection other = (org.tensorflow.proto.framework.TensorConnection) obj;
-
-    if (!getFromTensor()
-        .equals(other.getFromTensor())) return false;
-    if (!getToTensor()
-        .equals(other.getToTensor())) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    hash = (37 * hash) + FROM_TENSOR_FIELD_NUMBER;
-    hash = (53 * hash) + getFromTensor().hashCode();
-    hash = (37 * hash) + TO_TENSOR_FIELD_NUMBER;
-    hash = (53 * hash) + getToTensor().hashCode();
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.framework.TensorConnection parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.TensorConnection parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.TensorConnection parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.TensorConnection parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.TensorConnection parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.TensorConnection parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.TensorConnection parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.TensorConnection parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.TensorConnection parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.TensorConnection parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.TensorConnection parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.TensorConnection parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.framework.TensorConnection prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * <pre>
-   * Defines a connection between two tensors in a `GraphDef`.
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.TensorConnection}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.TensorConnection)
-      org.tensorflow.proto.framework.TensorConnectionOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_TensorConnection_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_TensorConnection_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.TensorConnection.class, org.tensorflow.proto.framework.TensorConnection.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.framework.TensorConnection.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      fromTensor_ = "";
-
-      toTensor_ = "";
-
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_TensorConnection_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.TensorConnection getDefaultInstanceForType() {
-      return org.tensorflow.proto.framework.TensorConnection.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.TensorConnection build() {
-      org.tensorflow.proto.framework.TensorConnection result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.TensorConnection buildPartial() {
-      org.tensorflow.proto.framework.TensorConnection result = new org.tensorflow.proto.framework.TensorConnection(this);
-      result.fromTensor_ = fromTensor_;
-      result.toTensor_ = toTensor_;
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.framework.TensorConnection) {
-        return mergeFrom((org.tensorflow.proto.framework.TensorConnection)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.framework.TensorConnection other) {
-      if (other == org.tensorflow.proto.framework.TensorConnection.getDefaultInstance()) return this;
-      if (!other.getFromTensor().isEmpty()) {
-        fromTensor_ = other.fromTensor_;
-        onChanged();
-      }
-      if (!other.getToTensor().isEmpty()) {
-        toTensor_ = other.toTensor_;
-        onChanged();
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.framework.TensorConnection parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.framework.TensorConnection) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-
-    private java.lang.Object fromTensor_ = "";
-    /**
-     * <pre>
-     * A tensor name. The value of this tensor will be substituted for
-     * the tensor named in `to_tensor`.
-     * </pre>
-     *
-     * <code>string from_tensor = 1;</code>
-     */
-    public java.lang.String getFromTensor() {
-      java.lang.Object ref = fromTensor_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        fromTensor_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * A tensor name. The value of this tensor will be substituted for
-     * the tensor named in `to_tensor`.
-     * </pre>
-     *
-     * <code>string from_tensor = 1;</code>
-     */
-    public com.google.protobuf.ByteString
-        getFromTensorBytes() {
-      java.lang.Object ref = fromTensor_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        fromTensor_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * A tensor name. The value of this tensor will be substituted for
-     * the tensor named in `to_tensor`.
-     * </pre>
-     *
-     * <code>string from_tensor = 1;</code>
-     */
-    public Builder setFromTensor(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      fromTensor_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * A tensor name. The value of this tensor will be substituted for
-     * the tensor named in `to_tensor`.
-     * </pre>
-     *
-     * <code>string from_tensor = 1;</code>
-     */
-    public Builder clearFromTensor() {
-      
-      fromTensor_ = getDefaultInstance().getFromTensor();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * A tensor name. The value of this tensor will be substituted for
-     * the tensor named in `to_tensor`.
-     * </pre>
-     *
-     * <code>string from_tensor = 1;</code>
-     */
-    public Builder setFromTensorBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      fromTensor_ = value;
-      onChanged();
-      return this;
-    }
-
-    private java.lang.Object toTensor_ = "";
-    /**
-     * <pre>
-     * A tensor name. The value of this tensor will be bound to the
-     * value of the tensor named in `from_tensor`.
-     * </pre>
-     *
-     * <code>string to_tensor = 2;</code>
-     */
-    public java.lang.String getToTensor() {
-      java.lang.Object ref = toTensor_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        toTensor_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * A tensor name. The value of this tensor will be bound to the
-     * value of the tensor named in `from_tensor`.
-     * </pre>
-     *
-     * <code>string to_tensor = 2;</code>
-     */
-    public com.google.protobuf.ByteString
-        getToTensorBytes() {
-      java.lang.Object ref = toTensor_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        toTensor_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * A tensor name. The value of this tensor will be bound to the
-     * value of the tensor named in `from_tensor`.
-     * </pre>
-     *
-     * <code>string to_tensor = 2;</code>
-     */
-    public Builder setToTensor(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      toTensor_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * A tensor name. The value of this tensor will be bound to the
-     * value of the tensor named in `from_tensor`.
-     * </pre>
-     *
-     * <code>string to_tensor = 2;</code>
-     */
-    public Builder clearToTensor() {
-      
-      toTensor_ = getDefaultInstance().getToTensor();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * A tensor name. The value of this tensor will be bound to the
-     * value of the tensor named in `from_tensor`.
-     * </pre>
-     *
-     * <code>string to_tensor = 2;</code>
-     */
-    public Builder setToTensorBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      toTensor_ = value;
-      onChanged();
-      return this;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.TensorConnection)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.TensorConnection)
-  private static final org.tensorflow.proto.framework.TensorConnection DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.TensorConnection();
-  }
-
-  public static org.tensorflow.proto.framework.TensorConnection getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<TensorConnection>
-      PARSER = new com.google.protobuf.AbstractParser<TensorConnection>() {
-    @java.lang.Override
-    public TensorConnection parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new TensorConnection(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<TensorConnection> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<TensorConnection> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.framework.TensorConnection getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorConnectionOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorConnectionOrBuilder.java
deleted file mode 100644
index 377a25aa7ad..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorConnectionOrBuilder.java
+++ /dev/null
@@ -1,49 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/config.proto
-
-package org.tensorflow.proto.framework;
-
-public interface TensorConnectionOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.TensorConnection)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <pre>
-   * A tensor name. The value of this tensor will be substituted for
-   * the tensor named in `to_tensor`.
-   * </pre>
-   *
-   * <code>string from_tensor = 1;</code>
-   */
-  java.lang.String getFromTensor();
-  /**
-   * <pre>
-   * A tensor name. The value of this tensor will be substituted for
-   * the tensor named in `to_tensor`.
-   * </pre>
-   *
-   * <code>string from_tensor = 1;</code>
-   */
-  com.google.protobuf.ByteString
-      getFromTensorBytes();
-
-  /**
-   * <pre>
-   * A tensor name. The value of this tensor will be bound to the
-   * value of the tensor named in `from_tensor`.
-   * </pre>
-   *
-   * <code>string to_tensor = 2;</code>
-   */
-  java.lang.String getToTensor();
-  /**
-   * <pre>
-   * A tensor name. The value of this tensor will be bound to the
-   * value of the tensor named in `from_tensor`.
-   * </pre>
-   *
-   * <code>string to_tensor = 2;</code>
-   */
-  com.google.protobuf.ByteString
-      getToTensorBytes();
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorDescription.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorDescription.java
deleted file mode 100644
index 7a79838fb16..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorDescription.java
+++ /dev/null
@@ -1,990 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/framework/tensor_description.proto
-
-package org.tensorflow.proto.framework;
-
-/**
- * Protobuf type {@code tensorflow.TensorDescription}
- */
-public  final class TensorDescription extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.TensorDescription)
-    TensorDescriptionOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use TensorDescription.newBuilder() to construct.
-  private TensorDescription(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private TensorDescription() {
-    dtype_ = 0;
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new TensorDescription();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private TensorDescription(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 8: {
-            int rawValue = input.readEnum();
-
-            dtype_ = rawValue;
-            break;
-          }
-          case 18: {
-            org.tensorflow.proto.framework.TensorShapeProto.Builder subBuilder = null;
-            if (shape_ != null) {
-              subBuilder = shape_.toBuilder();
-            }
-            shape_ = input.readMessage(org.tensorflow.proto.framework.TensorShapeProto.parser(), extensionRegistry);
-            if (subBuilder != null) {
-              subBuilder.mergeFrom(shape_);
-              shape_ = subBuilder.buildPartial();
-            }
-
-            break;
-          }
-          case 34: {
-            org.tensorflow.proto.framework.AllocationDescription.Builder subBuilder = null;
-            if (allocationDescription_ != null) {
-              subBuilder = allocationDescription_.toBuilder();
-            }
-            allocationDescription_ = input.readMessage(org.tensorflow.proto.framework.AllocationDescription.parser(), extensionRegistry);
-            if (subBuilder != null) {
-              subBuilder.mergeFrom(allocationDescription_);
-              allocationDescription_ = subBuilder.buildPartial();
-            }
-
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.framework.TensorDescriptionProtos.internal_static_tensorflow_TensorDescription_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.framework.TensorDescriptionProtos.internal_static_tensorflow_TensorDescription_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.framework.TensorDescription.class, org.tensorflow.proto.framework.TensorDescription.Builder.class);
-  }
-
-  public static final int DTYPE_FIELD_NUMBER = 1;
-  private int dtype_;
-  /**
-   * <pre>
-   * Data type of tensor elements
-   * </pre>
-   *
-   * <code>.tensorflow.DataType dtype = 1;</code>
-   */
-  public int getDtypeValue() {
-    return dtype_;
-  }
-  /**
-   * <pre>
-   * Data type of tensor elements
-   * </pre>
-   *
-   * <code>.tensorflow.DataType dtype = 1;</code>
-   */
-  public org.tensorflow.proto.framework.DataType getDtype() {
-    @SuppressWarnings("deprecation")
-    org.tensorflow.proto.framework.DataType result = org.tensorflow.proto.framework.DataType.valueOf(dtype_);
-    return result == null ? org.tensorflow.proto.framework.DataType.UNRECOGNIZED : result;
-  }
-
-  public static final int SHAPE_FIELD_NUMBER = 2;
-  private org.tensorflow.proto.framework.TensorShapeProto shape_;
-  /**
-   * <pre>
-   * Shape of the tensor.
-   * </pre>
-   *
-   * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-   */
-  public boolean hasShape() {
-    return shape_ != null;
-  }
-  /**
-   * <pre>
-   * Shape of the tensor.
-   * </pre>
-   *
-   * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-   */
-  public org.tensorflow.proto.framework.TensorShapeProto getShape() {
-    return shape_ == null ? org.tensorflow.proto.framework.TensorShapeProto.getDefaultInstance() : shape_;
-  }
-  /**
-   * <pre>
-   * Shape of the tensor.
-   * </pre>
-   *
-   * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-   */
-  public org.tensorflow.proto.framework.TensorShapeProtoOrBuilder getShapeOrBuilder() {
-    return getShape();
-  }
-
-  public static final int ALLOCATION_DESCRIPTION_FIELD_NUMBER = 4;
-  private org.tensorflow.proto.framework.AllocationDescription allocationDescription_;
-  /**
-   * <pre>
-   * Information about the size and allocator used for the data
-   * </pre>
-   *
-   * <code>.tensorflow.AllocationDescription allocation_description = 4;</code>
-   */
-  public boolean hasAllocationDescription() {
-    return allocationDescription_ != null;
-  }
-  /**
-   * <pre>
-   * Information about the size and allocator used for the data
-   * </pre>
-   *
-   * <code>.tensorflow.AllocationDescription allocation_description = 4;</code>
-   */
-  public org.tensorflow.proto.framework.AllocationDescription getAllocationDescription() {
-    return allocationDescription_ == null ? org.tensorflow.proto.framework.AllocationDescription.getDefaultInstance() : allocationDescription_;
-  }
-  /**
-   * <pre>
-   * Information about the size and allocator used for the data
-   * </pre>
-   *
-   * <code>.tensorflow.AllocationDescription allocation_description = 4;</code>
-   */
-  public org.tensorflow.proto.framework.AllocationDescriptionOrBuilder getAllocationDescriptionOrBuilder() {
-    return getAllocationDescription();
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    if (dtype_ != org.tensorflow.proto.framework.DataType.DT_INVALID.getNumber()) {
-      output.writeEnum(1, dtype_);
-    }
-    if (shape_ != null) {
-      output.writeMessage(2, getShape());
-    }
-    if (allocationDescription_ != null) {
-      output.writeMessage(4, getAllocationDescription());
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (dtype_ != org.tensorflow.proto.framework.DataType.DT_INVALID.getNumber()) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeEnumSize(1, dtype_);
-    }
-    if (shape_ != null) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(2, getShape());
-    }
-    if (allocationDescription_ != null) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(4, getAllocationDescription());
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.framework.TensorDescription)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.framework.TensorDescription other = (org.tensorflow.proto.framework.TensorDescription) obj;
-
-    if (dtype_ != other.dtype_) return false;
-    if (hasShape() != other.hasShape()) return false;
-    if (hasShape()) {
-      if (!getShape()
-          .equals(other.getShape())) return false;
-    }
-    if (hasAllocationDescription() != other.hasAllocationDescription()) return false;
-    if (hasAllocationDescription()) {
-      if (!getAllocationDescription()
-          .equals(other.getAllocationDescription())) return false;
-    }
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    hash = (37 * hash) + DTYPE_FIELD_NUMBER;
-    hash = (53 * hash) + dtype_;
-    if (hasShape()) {
-      hash = (37 * hash) + SHAPE_FIELD_NUMBER;
-      hash = (53 * hash) + getShape().hashCode();
-    }
-    if (hasAllocationDescription()) {
-      hash = (37 * hash) + ALLOCATION_DESCRIPTION_FIELD_NUMBER;
-      hash = (53 * hash) + getAllocationDescription().hashCode();
-    }
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.framework.TensorDescription parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.TensorDescription parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.TensorDescription parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.TensorDescription parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.TensorDescription parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.TensorDescription parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.TensorDescription parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.TensorDescription parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.TensorDescription parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.TensorDescription parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.TensorDescription parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.TensorDescription parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.framework.TensorDescription prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * Protobuf type {@code tensorflow.TensorDescription}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.TensorDescription)
-      org.tensorflow.proto.framework.TensorDescriptionOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.TensorDescriptionProtos.internal_static_tensorflow_TensorDescription_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.TensorDescriptionProtos.internal_static_tensorflow_TensorDescription_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.TensorDescription.class, org.tensorflow.proto.framework.TensorDescription.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.framework.TensorDescription.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      dtype_ = 0;
-
-      if (shapeBuilder_ == null) {
-        shape_ = null;
-      } else {
-        shape_ = null;
-        shapeBuilder_ = null;
-      }
-      if (allocationDescriptionBuilder_ == null) {
-        allocationDescription_ = null;
-      } else {
-        allocationDescription_ = null;
-        allocationDescriptionBuilder_ = null;
-      }
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.framework.TensorDescriptionProtos.internal_static_tensorflow_TensorDescription_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.TensorDescription getDefaultInstanceForType() {
-      return org.tensorflow.proto.framework.TensorDescription.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.TensorDescription build() {
-      org.tensorflow.proto.framework.TensorDescription result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.TensorDescription buildPartial() {
-      org.tensorflow.proto.framework.TensorDescription result = new org.tensorflow.proto.framework.TensorDescription(this);
-      result.dtype_ = dtype_;
-      if (shapeBuilder_ == null) {
-        result.shape_ = shape_;
-      } else {
-        result.shape_ = shapeBuilder_.build();
-      }
-      if (allocationDescriptionBuilder_ == null) {
-        result.allocationDescription_ = allocationDescription_;
-      } else {
-        result.allocationDescription_ = allocationDescriptionBuilder_.build();
-      }
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.framework.TensorDescription) {
-        return mergeFrom((org.tensorflow.proto.framework.TensorDescription)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.framework.TensorDescription other) {
-      if (other == org.tensorflow.proto.framework.TensorDescription.getDefaultInstance()) return this;
-      if (other.dtype_ != 0) {
-        setDtypeValue(other.getDtypeValue());
-      }
-      if (other.hasShape()) {
-        mergeShape(other.getShape());
-      }
-      if (other.hasAllocationDescription()) {
-        mergeAllocationDescription(other.getAllocationDescription());
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.framework.TensorDescription parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.framework.TensorDescription) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-
-    private int dtype_ = 0;
-    /**
-     * <pre>
-     * Data type of tensor elements
-     * </pre>
-     *
-     * <code>.tensorflow.DataType dtype = 1;</code>
-     */
-    public int getDtypeValue() {
-      return dtype_;
-    }
-    /**
-     * <pre>
-     * Data type of tensor elements
-     * </pre>
-     *
-     * <code>.tensorflow.DataType dtype = 1;</code>
-     */
-    public Builder setDtypeValue(int value) {
-      dtype_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Data type of tensor elements
-     * </pre>
-     *
-     * <code>.tensorflow.DataType dtype = 1;</code>
-     */
-    public org.tensorflow.proto.framework.DataType getDtype() {
-      @SuppressWarnings("deprecation")
-      org.tensorflow.proto.framework.DataType result = org.tensorflow.proto.framework.DataType.valueOf(dtype_);
-      return result == null ? org.tensorflow.proto.framework.DataType.UNRECOGNIZED : result;
-    }
-    /**
-     * <pre>
-     * Data type of tensor elements
-     * </pre>
-     *
-     * <code>.tensorflow.DataType dtype = 1;</code>
-     */
-    public Builder setDtype(org.tensorflow.proto.framework.DataType value) {
-      if (value == null) {
-        throw new NullPointerException();
-      }
-      
-      dtype_ = value.getNumber();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Data type of tensor elements
-     * </pre>
-     *
-     * <code>.tensorflow.DataType dtype = 1;</code>
-     */
-    public Builder clearDtype() {
-      
-      dtype_ = 0;
-      onChanged();
-      return this;
-    }
-
-    private org.tensorflow.proto.framework.TensorShapeProto shape_;
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.TensorShapeProto, org.tensorflow.proto.framework.TensorShapeProto.Builder, org.tensorflow.proto.framework.TensorShapeProtoOrBuilder> shapeBuilder_;
-    /**
-     * <pre>
-     * Shape of the tensor.
-     * </pre>
-     *
-     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-     */
-    public boolean hasShape() {
-      return shapeBuilder_ != null || shape_ != null;
-    }
-    /**
-     * <pre>
-     * Shape of the tensor.
-     * </pre>
-     *
-     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-     */
-    public org.tensorflow.proto.framework.TensorShapeProto getShape() {
-      if (shapeBuilder_ == null) {
-        return shape_ == null ? org.tensorflow.proto.framework.TensorShapeProto.getDefaultInstance() : shape_;
-      } else {
-        return shapeBuilder_.getMessage();
-      }
-    }
-    /**
-     * <pre>
-     * Shape of the tensor.
-     * </pre>
-     *
-     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-     */
-    public Builder setShape(org.tensorflow.proto.framework.TensorShapeProto value) {
-      if (shapeBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        shape_ = value;
-        onChanged();
-      } else {
-        shapeBuilder_.setMessage(value);
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Shape of the tensor.
-     * </pre>
-     *
-     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-     */
-    public Builder setShape(
-        org.tensorflow.proto.framework.TensorShapeProto.Builder builderForValue) {
-      if (shapeBuilder_ == null) {
-        shape_ = builderForValue.build();
-        onChanged();
-      } else {
-        shapeBuilder_.setMessage(builderForValue.build());
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Shape of the tensor.
-     * </pre>
-     *
-     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-     */
-    public Builder mergeShape(org.tensorflow.proto.framework.TensorShapeProto value) {
-      if (shapeBuilder_ == null) {
-        if (shape_ != null) {
-          shape_ =
-            org.tensorflow.proto.framework.TensorShapeProto.newBuilder(shape_).mergeFrom(value).buildPartial();
-        } else {
-          shape_ = value;
-        }
-        onChanged();
-      } else {
-        shapeBuilder_.mergeFrom(value);
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Shape of the tensor.
-     * </pre>
-     *
-     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-     */
-    public Builder clearShape() {
-      if (shapeBuilder_ == null) {
-        shape_ = null;
-        onChanged();
-      } else {
-        shape_ = null;
-        shapeBuilder_ = null;
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Shape of the tensor.
-     * </pre>
-     *
-     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-     */
-    public org.tensorflow.proto.framework.TensorShapeProto.Builder getShapeBuilder() {
-      
-      onChanged();
-      return getShapeFieldBuilder().getBuilder();
-    }
-    /**
-     * <pre>
-     * Shape of the tensor.
-     * </pre>
-     *
-     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-     */
-    public org.tensorflow.proto.framework.TensorShapeProtoOrBuilder getShapeOrBuilder() {
-      if (shapeBuilder_ != null) {
-        return shapeBuilder_.getMessageOrBuilder();
-      } else {
-        return shape_ == null ?
-            org.tensorflow.proto.framework.TensorShapeProto.getDefaultInstance() : shape_;
-      }
-    }
-    /**
-     * <pre>
-     * Shape of the tensor.
-     * </pre>
-     *
-     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-     */
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.TensorShapeProto, org.tensorflow.proto.framework.TensorShapeProto.Builder, org.tensorflow.proto.framework.TensorShapeProtoOrBuilder> 
-        getShapeFieldBuilder() {
-      if (shapeBuilder_ == null) {
-        shapeBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-            org.tensorflow.proto.framework.TensorShapeProto, org.tensorflow.proto.framework.TensorShapeProto.Builder, org.tensorflow.proto.framework.TensorShapeProtoOrBuilder>(
-                getShape(),
-                getParentForChildren(),
-                isClean());
-        shape_ = null;
-      }
-      return shapeBuilder_;
-    }
-
-    private org.tensorflow.proto.framework.AllocationDescription allocationDescription_;
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.AllocationDescription, org.tensorflow.proto.framework.AllocationDescription.Builder, org.tensorflow.proto.framework.AllocationDescriptionOrBuilder> allocationDescriptionBuilder_;
-    /**
-     * <pre>
-     * Information about the size and allocator used for the data
-     * </pre>
-     *
-     * <code>.tensorflow.AllocationDescription allocation_description = 4;</code>
-     */
-    public boolean hasAllocationDescription() {
-      return allocationDescriptionBuilder_ != null || allocationDescription_ != null;
-    }
-    /**
-     * <pre>
-     * Information about the size and allocator used for the data
-     * </pre>
-     *
-     * <code>.tensorflow.AllocationDescription allocation_description = 4;</code>
-     */
-    public org.tensorflow.proto.framework.AllocationDescription getAllocationDescription() {
-      if (allocationDescriptionBuilder_ == null) {
-        return allocationDescription_ == null ? org.tensorflow.proto.framework.AllocationDescription.getDefaultInstance() : allocationDescription_;
-      } else {
-        return allocationDescriptionBuilder_.getMessage();
-      }
-    }
-    /**
-     * <pre>
-     * Information about the size and allocator used for the data
-     * </pre>
-     *
-     * <code>.tensorflow.AllocationDescription allocation_description = 4;</code>
-     */
-    public Builder setAllocationDescription(org.tensorflow.proto.framework.AllocationDescription value) {
-      if (allocationDescriptionBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        allocationDescription_ = value;
-        onChanged();
-      } else {
-        allocationDescriptionBuilder_.setMessage(value);
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Information about the size and allocator used for the data
-     * </pre>
-     *
-     * <code>.tensorflow.AllocationDescription allocation_description = 4;</code>
-     */
-    public Builder setAllocationDescription(
-        org.tensorflow.proto.framework.AllocationDescription.Builder builderForValue) {
-      if (allocationDescriptionBuilder_ == null) {
-        allocationDescription_ = builderForValue.build();
-        onChanged();
-      } else {
-        allocationDescriptionBuilder_.setMessage(builderForValue.build());
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Information about the size and allocator used for the data
-     * </pre>
-     *
-     * <code>.tensorflow.AllocationDescription allocation_description = 4;</code>
-     */
-    public Builder mergeAllocationDescription(org.tensorflow.proto.framework.AllocationDescription value) {
-      if (allocationDescriptionBuilder_ == null) {
-        if (allocationDescription_ != null) {
-          allocationDescription_ =
-            org.tensorflow.proto.framework.AllocationDescription.newBuilder(allocationDescription_).mergeFrom(value).buildPartial();
-        } else {
-          allocationDescription_ = value;
-        }
-        onChanged();
-      } else {
-        allocationDescriptionBuilder_.mergeFrom(value);
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Information about the size and allocator used for the data
-     * </pre>
-     *
-     * <code>.tensorflow.AllocationDescription allocation_description = 4;</code>
-     */
-    public Builder clearAllocationDescription() {
-      if (allocationDescriptionBuilder_ == null) {
-        allocationDescription_ = null;
-        onChanged();
-      } else {
-        allocationDescription_ = null;
-        allocationDescriptionBuilder_ = null;
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Information about the size and allocator used for the data
-     * </pre>
-     *
-     * <code>.tensorflow.AllocationDescription allocation_description = 4;</code>
-     */
-    public org.tensorflow.proto.framework.AllocationDescription.Builder getAllocationDescriptionBuilder() {
-      
-      onChanged();
-      return getAllocationDescriptionFieldBuilder().getBuilder();
-    }
-    /**
-     * <pre>
-     * Information about the size and allocator used for the data
-     * </pre>
-     *
-     * <code>.tensorflow.AllocationDescription allocation_description = 4;</code>
-     */
-    public org.tensorflow.proto.framework.AllocationDescriptionOrBuilder getAllocationDescriptionOrBuilder() {
-      if (allocationDescriptionBuilder_ != null) {
-        return allocationDescriptionBuilder_.getMessageOrBuilder();
-      } else {
-        return allocationDescription_ == null ?
-            org.tensorflow.proto.framework.AllocationDescription.getDefaultInstance() : allocationDescription_;
-      }
-    }
-    /**
-     * <pre>
-     * Information about the size and allocator used for the data
-     * </pre>
-     *
-     * <code>.tensorflow.AllocationDescription allocation_description = 4;</code>
-     */
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.AllocationDescription, org.tensorflow.proto.framework.AllocationDescription.Builder, org.tensorflow.proto.framework.AllocationDescriptionOrBuilder> 
-        getAllocationDescriptionFieldBuilder() {
-      if (allocationDescriptionBuilder_ == null) {
-        allocationDescriptionBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-            org.tensorflow.proto.framework.AllocationDescription, org.tensorflow.proto.framework.AllocationDescription.Builder, org.tensorflow.proto.framework.AllocationDescriptionOrBuilder>(
-                getAllocationDescription(),
-                getParentForChildren(),
-                isClean());
-        allocationDescription_ = null;
-      }
-      return allocationDescriptionBuilder_;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.TensorDescription)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.TensorDescription)
-  private static final org.tensorflow.proto.framework.TensorDescription DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.TensorDescription();
-  }
-
-  public static org.tensorflow.proto.framework.TensorDescription getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<TensorDescription>
-      PARSER = new com.google.protobuf.AbstractParser<TensorDescription>() {
-    @java.lang.Override
-    public TensorDescription parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new TensorDescription(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<TensorDescription> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<TensorDescription> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.framework.TensorDescription getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorDescriptionOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorDescriptionOrBuilder.java
deleted file mode 100644
index e10f0e90d6b..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorDescriptionOrBuilder.java
+++ /dev/null
@@ -1,76 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/framework/tensor_description.proto
-
-package org.tensorflow.proto.framework;
-
-public interface TensorDescriptionOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.TensorDescription)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <pre>
-   * Data type of tensor elements
-   * </pre>
-   *
-   * <code>.tensorflow.DataType dtype = 1;</code>
-   */
-  int getDtypeValue();
-  /**
-   * <pre>
-   * Data type of tensor elements
-   * </pre>
-   *
-   * <code>.tensorflow.DataType dtype = 1;</code>
-   */
-  org.tensorflow.proto.framework.DataType getDtype();
-
-  /**
-   * <pre>
-   * Shape of the tensor.
-   * </pre>
-   *
-   * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-   */
-  boolean hasShape();
-  /**
-   * <pre>
-   * Shape of the tensor.
-   * </pre>
-   *
-   * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-   */
-  org.tensorflow.proto.framework.TensorShapeProto getShape();
-  /**
-   * <pre>
-   * Shape of the tensor.
-   * </pre>
-   *
-   * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-   */
-  org.tensorflow.proto.framework.TensorShapeProtoOrBuilder getShapeOrBuilder();
-
-  /**
-   * <pre>
-   * Information about the size and allocator used for the data
-   * </pre>
-   *
-   * <code>.tensorflow.AllocationDescription allocation_description = 4;</code>
-   */
-  boolean hasAllocationDescription();
-  /**
-   * <pre>
-   * Information about the size and allocator used for the data
-   * </pre>
-   *
-   * <code>.tensorflow.AllocationDescription allocation_description = 4;</code>
-   */
-  org.tensorflow.proto.framework.AllocationDescription getAllocationDescription();
-  /**
-   * <pre>
-   * Information about the size and allocator used for the data
-   * </pre>
-   *
-   * <code>.tensorflow.AllocationDescription allocation_description = 4;</code>
-   */
-  org.tensorflow.proto.framework.AllocationDescriptionOrBuilder getAllocationDescriptionOrBuilder();
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorDescriptionProtos.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorDescriptionProtos.java
deleted file mode 100644
index 840b3c4068d..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorDescriptionProtos.java
+++ /dev/null
@@ -1,65 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/framework/tensor_description.proto
-
-package org.tensorflow.proto.framework;
-
-public final class TensorDescriptionProtos {
-  private TensorDescriptionProtos() {}
-  public static void registerAllExtensions(
-      com.google.protobuf.ExtensionRegistryLite registry) {
-  }
-
-  public static void registerAllExtensions(
-      com.google.protobuf.ExtensionRegistry registry) {
-    registerAllExtensions(
-        (com.google.protobuf.ExtensionRegistryLite) registry);
-  }
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_TensorDescription_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_TensorDescription_fieldAccessorTable;
-
-  public static com.google.protobuf.Descriptors.FileDescriptor
-      getDescriptor() {
-    return descriptor;
-  }
-  private static  com.google.protobuf.Descriptors.FileDescriptor
-      descriptor;
-  static {
-    java.lang.String[] descriptorData = {
-      "\n2tensorflow/core/framework/tensor_descr" +
-      "iption.proto\022\ntensorflow\0326tensorflow/cor" +
-      "e/framework/allocation_description.proto" +
-      "\032,tensorflow/core/framework/tensor_shape" +
-      ".proto\032%tensorflow/core/framework/types." +
-      "proto\"\250\001\n\021TensorDescription\022#\n\005dtype\030\001 \001" +
-      "(\0162\024.tensorflow.DataType\022+\n\005shape\030\002 \001(\0132" +
-      "\034.tensorflow.TensorShapeProto\022A\n\026allocat" +
-      "ion_description\030\004 \001(\0132!.tensorflow.Alloc" +
-      "ationDescriptionB\231\001\n\036org.tensorflow.prot" +
-      "o.frameworkB\027TensorDescriptionProtosP\001ZY" +
-      "github.com/tensorflow/tensorflow/tensorf" +
-      "low/go/core/framework/tensor_description" +
-      "_go_proto\370\001\001b\006proto3"
-    };
-    descriptor = com.google.protobuf.Descriptors.FileDescriptor
-      .internalBuildGeneratedFileFrom(descriptorData,
-        new com.google.protobuf.Descriptors.FileDescriptor[] {
-          org.tensorflow.proto.framework.AllocationDescriptionProtos.getDescriptor(),
-          org.tensorflow.proto.framework.TensorShapeProtos.getDescriptor(),
-          org.tensorflow.proto.framework.TypesProtos.getDescriptor(),
-        });
-    internal_static_tensorflow_TensorDescription_descriptor =
-      getDescriptor().getMessageTypes().get(0);
-    internal_static_tensorflow_TensorDescription_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_TensorDescription_descriptor,
-        new java.lang.String[] { "Dtype", "Shape", "AllocationDescription", });
-    org.tensorflow.proto.framework.AllocationDescriptionProtos.getDescriptor();
-    org.tensorflow.proto.framework.TensorShapeProtos.getDescriptor();
-    org.tensorflow.proto.framework.TypesProtos.getDescriptor();
-  }
-
-  // @@protoc_insertion_point(outer_class_scope)
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorInfo.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorInfo.java
deleted file mode 100644
index 4e873da94be..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorInfo.java
+++ /dev/null
@@ -1,3680 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/meta_graph.proto
-
-package org.tensorflow.proto.framework;
-
-/**
- * <pre>
- * Information about a Tensor necessary for feeding or retrieval.
- * </pre>
- *
- * Protobuf type {@code tensorflow.TensorInfo}
- */
-public  final class TensorInfo extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.TensorInfo)
-    TensorInfoOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use TensorInfo.newBuilder() to construct.
-  private TensorInfo(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private TensorInfo() {
-    dtype_ = 0;
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new TensorInfo();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private TensorInfo(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 10: {
-            java.lang.String s = input.readStringRequireUtf8();
-            encodingCase_ = 1;
-            encoding_ = s;
-            break;
-          }
-          case 16: {
-            int rawValue = input.readEnum();
-
-            dtype_ = rawValue;
-            break;
-          }
-          case 26: {
-            org.tensorflow.proto.framework.TensorShapeProto.Builder subBuilder = null;
-            if (tensorShape_ != null) {
-              subBuilder = tensorShape_.toBuilder();
-            }
-            tensorShape_ = input.readMessage(org.tensorflow.proto.framework.TensorShapeProto.parser(), extensionRegistry);
-            if (subBuilder != null) {
-              subBuilder.mergeFrom(tensorShape_);
-              tensorShape_ = subBuilder.buildPartial();
-            }
-
-            break;
-          }
-          case 34: {
-            org.tensorflow.proto.framework.TensorInfo.CooSparse.Builder subBuilder = null;
-            if (encodingCase_ == 4) {
-              subBuilder = ((org.tensorflow.proto.framework.TensorInfo.CooSparse) encoding_).toBuilder();
-            }
-            encoding_ =
-                input.readMessage(org.tensorflow.proto.framework.TensorInfo.CooSparse.parser(), extensionRegistry);
-            if (subBuilder != null) {
-              subBuilder.mergeFrom((org.tensorflow.proto.framework.TensorInfo.CooSparse) encoding_);
-              encoding_ = subBuilder.buildPartial();
-            }
-            encodingCase_ = 4;
-            break;
-          }
-          case 42: {
-            org.tensorflow.proto.framework.TensorInfo.CompositeTensor.Builder subBuilder = null;
-            if (encodingCase_ == 5) {
-              subBuilder = ((org.tensorflow.proto.framework.TensorInfo.CompositeTensor) encoding_).toBuilder();
-            }
-            encoding_ =
-                input.readMessage(org.tensorflow.proto.framework.TensorInfo.CompositeTensor.parser(), extensionRegistry);
-            if (subBuilder != null) {
-              subBuilder.mergeFrom((org.tensorflow.proto.framework.TensorInfo.CompositeTensor) encoding_);
-              encoding_ = subBuilder.buildPartial();
-            }
-            encodingCase_ = 5;
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.framework.MetaGraphProtos.internal_static_tensorflow_TensorInfo_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.framework.MetaGraphProtos.internal_static_tensorflow_TensorInfo_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.framework.TensorInfo.class, org.tensorflow.proto.framework.TensorInfo.Builder.class);
-  }
-
-  public interface CooSparseOrBuilder extends
-      // @@protoc_insertion_point(interface_extends:tensorflow.TensorInfo.CooSparse)
-      com.google.protobuf.MessageOrBuilder {
-
-    /**
-     * <pre>
-     * The shape of the values Tensor is [?].  Its dtype must be the dtype of
-     * the SparseTensor as a whole, given in the enclosing TensorInfo.
-     * </pre>
-     *
-     * <code>string values_tensor_name = 1;</code>
-     */
-    java.lang.String getValuesTensorName();
-    /**
-     * <pre>
-     * The shape of the values Tensor is [?].  Its dtype must be the dtype of
-     * the SparseTensor as a whole, given in the enclosing TensorInfo.
-     * </pre>
-     *
-     * <code>string values_tensor_name = 1;</code>
-     */
-    com.google.protobuf.ByteString
-        getValuesTensorNameBytes();
-
-    /**
-     * <pre>
-     * The indices Tensor must have dtype int64 and shape [?, ?].
-     * </pre>
-     *
-     * <code>string indices_tensor_name = 2;</code>
-     */
-    java.lang.String getIndicesTensorName();
-    /**
-     * <pre>
-     * The indices Tensor must have dtype int64 and shape [?, ?].
-     * </pre>
-     *
-     * <code>string indices_tensor_name = 2;</code>
-     */
-    com.google.protobuf.ByteString
-        getIndicesTensorNameBytes();
-
-    /**
-     * <pre>
-     * The dynamic logical shape represented by the SparseTensor is recorded in
-     * the Tensor referenced here.  It must have dtype int64 and shape [?].
-     * </pre>
-     *
-     * <code>string dense_shape_tensor_name = 3;</code>
-     */
-    java.lang.String getDenseShapeTensorName();
-    /**
-     * <pre>
-     * The dynamic logical shape represented by the SparseTensor is recorded in
-     * the Tensor referenced here.  It must have dtype int64 and shape [?].
-     * </pre>
-     *
-     * <code>string dense_shape_tensor_name = 3;</code>
-     */
-    com.google.protobuf.ByteString
-        getDenseShapeTensorNameBytes();
-  }
-  /**
-   * <pre>
-   * For sparse tensors, The COO encoding stores a triple of values, indices,
-   * and shape.
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.TensorInfo.CooSparse}
-   */
-  public  static final class CooSparse extends
-      com.google.protobuf.GeneratedMessageV3 implements
-      // @@protoc_insertion_point(message_implements:tensorflow.TensorInfo.CooSparse)
-      CooSparseOrBuilder {
-  private static final long serialVersionUID = 0L;
-    // Use CooSparse.newBuilder() to construct.
-    private CooSparse(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-      super(builder);
-    }
-    private CooSparse() {
-      valuesTensorName_ = "";
-      indicesTensorName_ = "";
-      denseShapeTensorName_ = "";
-    }
-
-    @java.lang.Override
-    @SuppressWarnings({"unused"})
-    protected java.lang.Object newInstance(
-        UnusedPrivateParameter unused) {
-      return new CooSparse();
-    }
-
-    @java.lang.Override
-    public final com.google.protobuf.UnknownFieldSet
-    getUnknownFields() {
-      return this.unknownFields;
-    }
-    private CooSparse(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      this();
-      if (extensionRegistry == null) {
-        throw new java.lang.NullPointerException();
-      }
-      com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-          com.google.protobuf.UnknownFieldSet.newBuilder();
-      try {
-        boolean done = false;
-        while (!done) {
-          int tag = input.readTag();
-          switch (tag) {
-            case 0:
-              done = true;
-              break;
-            case 10: {
-              java.lang.String s = input.readStringRequireUtf8();
-
-              valuesTensorName_ = s;
-              break;
-            }
-            case 18: {
-              java.lang.String s = input.readStringRequireUtf8();
-
-              indicesTensorName_ = s;
-              break;
-            }
-            case 26: {
-              java.lang.String s = input.readStringRequireUtf8();
-
-              denseShapeTensorName_ = s;
-              break;
-            }
-            default: {
-              if (!parseUnknownField(
-                  input, unknownFields, extensionRegistry, tag)) {
-                done = true;
-              }
-              break;
-            }
-          }
-        }
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        throw e.setUnfinishedMessage(this);
-      } catch (java.io.IOException e) {
-        throw new com.google.protobuf.InvalidProtocolBufferException(
-            e).setUnfinishedMessage(this);
-      } finally {
-        this.unknownFields = unknownFields.build();
-        makeExtensionsImmutable();
-      }
-    }
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.MetaGraphProtos.internal_static_tensorflow_TensorInfo_CooSparse_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.MetaGraphProtos.internal_static_tensorflow_TensorInfo_CooSparse_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.TensorInfo.CooSparse.class, org.tensorflow.proto.framework.TensorInfo.CooSparse.Builder.class);
-    }
-
-    public static final int VALUES_TENSOR_NAME_FIELD_NUMBER = 1;
-    private volatile java.lang.Object valuesTensorName_;
-    /**
-     * <pre>
-     * The shape of the values Tensor is [?].  Its dtype must be the dtype of
-     * the SparseTensor as a whole, given in the enclosing TensorInfo.
-     * </pre>
-     *
-     * <code>string values_tensor_name = 1;</code>
-     */
-    public java.lang.String getValuesTensorName() {
-      java.lang.Object ref = valuesTensorName_;
-      if (ref instanceof java.lang.String) {
-        return (java.lang.String) ref;
-      } else {
-        com.google.protobuf.ByteString bs = 
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        valuesTensorName_ = s;
-        return s;
-      }
-    }
-    /**
-     * <pre>
-     * The shape of the values Tensor is [?].  Its dtype must be the dtype of
-     * the SparseTensor as a whole, given in the enclosing TensorInfo.
-     * </pre>
-     *
-     * <code>string values_tensor_name = 1;</code>
-     */
-    public com.google.protobuf.ByteString
-        getValuesTensorNameBytes() {
-      java.lang.Object ref = valuesTensorName_;
-      if (ref instanceof java.lang.String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        valuesTensorName_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-
-    public static final int INDICES_TENSOR_NAME_FIELD_NUMBER = 2;
-    private volatile java.lang.Object indicesTensorName_;
-    /**
-     * <pre>
-     * The indices Tensor must have dtype int64 and shape [?, ?].
-     * </pre>
-     *
-     * <code>string indices_tensor_name = 2;</code>
-     */
-    public java.lang.String getIndicesTensorName() {
-      java.lang.Object ref = indicesTensorName_;
-      if (ref instanceof java.lang.String) {
-        return (java.lang.String) ref;
-      } else {
-        com.google.protobuf.ByteString bs = 
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        indicesTensorName_ = s;
-        return s;
-      }
-    }
-    /**
-     * <pre>
-     * The indices Tensor must have dtype int64 and shape [?, ?].
-     * </pre>
-     *
-     * <code>string indices_tensor_name = 2;</code>
-     */
-    public com.google.protobuf.ByteString
-        getIndicesTensorNameBytes() {
-      java.lang.Object ref = indicesTensorName_;
-      if (ref instanceof java.lang.String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        indicesTensorName_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-
-    public static final int DENSE_SHAPE_TENSOR_NAME_FIELD_NUMBER = 3;
-    private volatile java.lang.Object denseShapeTensorName_;
-    /**
-     * <pre>
-     * The dynamic logical shape represented by the SparseTensor is recorded in
-     * the Tensor referenced here.  It must have dtype int64 and shape [?].
-     * </pre>
-     *
-     * <code>string dense_shape_tensor_name = 3;</code>
-     */
-    public java.lang.String getDenseShapeTensorName() {
-      java.lang.Object ref = denseShapeTensorName_;
-      if (ref instanceof java.lang.String) {
-        return (java.lang.String) ref;
-      } else {
-        com.google.protobuf.ByteString bs = 
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        denseShapeTensorName_ = s;
-        return s;
-      }
-    }
-    /**
-     * <pre>
-     * The dynamic logical shape represented by the SparseTensor is recorded in
-     * the Tensor referenced here.  It must have dtype int64 and shape [?].
-     * </pre>
-     *
-     * <code>string dense_shape_tensor_name = 3;</code>
-     */
-    public com.google.protobuf.ByteString
-        getDenseShapeTensorNameBytes() {
-      java.lang.Object ref = denseShapeTensorName_;
-      if (ref instanceof java.lang.String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        denseShapeTensorName_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-
-    private byte memoizedIsInitialized = -1;
-    @java.lang.Override
-    public final boolean isInitialized() {
-      byte isInitialized = memoizedIsInitialized;
-      if (isInitialized == 1) return true;
-      if (isInitialized == 0) return false;
-
-      memoizedIsInitialized = 1;
-      return true;
-    }
-
-    @java.lang.Override
-    public void writeTo(com.google.protobuf.CodedOutputStream output)
-                        throws java.io.IOException {
-      if (!getValuesTensorNameBytes().isEmpty()) {
-        com.google.protobuf.GeneratedMessageV3.writeString(output, 1, valuesTensorName_);
-      }
-      if (!getIndicesTensorNameBytes().isEmpty()) {
-        com.google.protobuf.GeneratedMessageV3.writeString(output, 2, indicesTensorName_);
-      }
-      if (!getDenseShapeTensorNameBytes().isEmpty()) {
-        com.google.protobuf.GeneratedMessageV3.writeString(output, 3, denseShapeTensorName_);
-      }
-      unknownFields.writeTo(output);
-    }
-
-    @java.lang.Override
-    public int getSerializedSize() {
-      int size = memoizedSize;
-      if (size != -1) return size;
-
-      size = 0;
-      if (!getValuesTensorNameBytes().isEmpty()) {
-        size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, valuesTensorName_);
-      }
-      if (!getIndicesTensorNameBytes().isEmpty()) {
-        size += com.google.protobuf.GeneratedMessageV3.computeStringSize(2, indicesTensorName_);
-      }
-      if (!getDenseShapeTensorNameBytes().isEmpty()) {
-        size += com.google.protobuf.GeneratedMessageV3.computeStringSize(3, denseShapeTensorName_);
-      }
-      size += unknownFields.getSerializedSize();
-      memoizedSize = size;
-      return size;
-    }
-
-    @java.lang.Override
-    public boolean equals(final java.lang.Object obj) {
-      if (obj == this) {
-       return true;
-      }
-      if (!(obj instanceof org.tensorflow.proto.framework.TensorInfo.CooSparse)) {
-        return super.equals(obj);
-      }
-      org.tensorflow.proto.framework.TensorInfo.CooSparse other = (org.tensorflow.proto.framework.TensorInfo.CooSparse) obj;
-
-      if (!getValuesTensorName()
-          .equals(other.getValuesTensorName())) return false;
-      if (!getIndicesTensorName()
-          .equals(other.getIndicesTensorName())) return false;
-      if (!getDenseShapeTensorName()
-          .equals(other.getDenseShapeTensorName())) return false;
-      if (!unknownFields.equals(other.unknownFields)) return false;
-      return true;
-    }
-
-    @java.lang.Override
-    public int hashCode() {
-      if (memoizedHashCode != 0) {
-        return memoizedHashCode;
-      }
-      int hash = 41;
-      hash = (19 * hash) + getDescriptor().hashCode();
-      hash = (37 * hash) + VALUES_TENSOR_NAME_FIELD_NUMBER;
-      hash = (53 * hash) + getValuesTensorName().hashCode();
-      hash = (37 * hash) + INDICES_TENSOR_NAME_FIELD_NUMBER;
-      hash = (53 * hash) + getIndicesTensorName().hashCode();
-      hash = (37 * hash) + DENSE_SHAPE_TENSOR_NAME_FIELD_NUMBER;
-      hash = (53 * hash) + getDenseShapeTensorName().hashCode();
-      hash = (29 * hash) + unknownFields.hashCode();
-      memoizedHashCode = hash;
-      return hash;
-    }
-
-    public static org.tensorflow.proto.framework.TensorInfo.CooSparse parseFrom(
-        java.nio.ByteBuffer data)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data);
-    }
-    public static org.tensorflow.proto.framework.TensorInfo.CooSparse parseFrom(
-        java.nio.ByteBuffer data,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.TensorInfo.CooSparse parseFrom(
-        com.google.protobuf.ByteString data)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data);
-    }
-    public static org.tensorflow.proto.framework.TensorInfo.CooSparse parseFrom(
-        com.google.protobuf.ByteString data,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.TensorInfo.CooSparse parseFrom(byte[] data)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data);
-    }
-    public static org.tensorflow.proto.framework.TensorInfo.CooSparse parseFrom(
-        byte[] data,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.TensorInfo.CooSparse parseFrom(java.io.InputStream input)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input);
-    }
-    public static org.tensorflow.proto.framework.TensorInfo.CooSparse parseFrom(
-        java.io.InputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.TensorInfo.CooSparse parseDelimitedFrom(java.io.InputStream input)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseDelimitedWithIOException(PARSER, input);
-    }
-    public static org.tensorflow.proto.framework.TensorInfo.CooSparse parseDelimitedFrom(
-        java.io.InputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.TensorInfo.CooSparse parseFrom(
-        com.google.protobuf.CodedInputStream input)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input);
-    }
-    public static org.tensorflow.proto.framework.TensorInfo.CooSparse parseFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input, extensionRegistry);
-    }
-
-    @java.lang.Override
-    public Builder newBuilderForType() { return newBuilder(); }
-    public static Builder newBuilder() {
-      return DEFAULT_INSTANCE.toBuilder();
-    }
-    public static Builder newBuilder(org.tensorflow.proto.framework.TensorInfo.CooSparse prototype) {
-      return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-    }
-    @java.lang.Override
-    public Builder toBuilder() {
-      return this == DEFAULT_INSTANCE
-          ? new Builder() : new Builder().mergeFrom(this);
-    }
-
-    @java.lang.Override
-    protected Builder newBuilderForType(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      Builder builder = new Builder(parent);
-      return builder;
-    }
-    /**
-     * <pre>
-     * For sparse tensors, The COO encoding stores a triple of values, indices,
-     * and shape.
-     * </pre>
-     *
-     * Protobuf type {@code tensorflow.TensorInfo.CooSparse}
-     */
-    public static final class Builder extends
-        com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-        // @@protoc_insertion_point(builder_implements:tensorflow.TensorInfo.CooSparse)
-        org.tensorflow.proto.framework.TensorInfo.CooSparseOrBuilder {
-      public static final com.google.protobuf.Descriptors.Descriptor
-          getDescriptor() {
-        return org.tensorflow.proto.framework.MetaGraphProtos.internal_static_tensorflow_TensorInfo_CooSparse_descriptor;
-      }
-
-      @java.lang.Override
-      protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-          internalGetFieldAccessorTable() {
-        return org.tensorflow.proto.framework.MetaGraphProtos.internal_static_tensorflow_TensorInfo_CooSparse_fieldAccessorTable
-            .ensureFieldAccessorsInitialized(
-                org.tensorflow.proto.framework.TensorInfo.CooSparse.class, org.tensorflow.proto.framework.TensorInfo.CooSparse.Builder.class);
-      }
-
-      // Construct using org.tensorflow.proto.framework.TensorInfo.CooSparse.newBuilder()
-      private Builder() {
-        maybeForceBuilderInitialization();
-      }
-
-      private Builder(
-          com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-        super(parent);
-        maybeForceBuilderInitialization();
-      }
-      private void maybeForceBuilderInitialization() {
-        if (com.google.protobuf.GeneratedMessageV3
-                .alwaysUseFieldBuilders) {
-        }
-      }
-      @java.lang.Override
-      public Builder clear() {
-        super.clear();
-        valuesTensorName_ = "";
-
-        indicesTensorName_ = "";
-
-        denseShapeTensorName_ = "";
-
-        return this;
-      }
-
-      @java.lang.Override
-      public com.google.protobuf.Descriptors.Descriptor
-          getDescriptorForType() {
-        return org.tensorflow.proto.framework.MetaGraphProtos.internal_static_tensorflow_TensorInfo_CooSparse_descriptor;
-      }
-
-      @java.lang.Override
-      public org.tensorflow.proto.framework.TensorInfo.CooSparse getDefaultInstanceForType() {
-        return org.tensorflow.proto.framework.TensorInfo.CooSparse.getDefaultInstance();
-      }
-
-      @java.lang.Override
-      public org.tensorflow.proto.framework.TensorInfo.CooSparse build() {
-        org.tensorflow.proto.framework.TensorInfo.CooSparse result = buildPartial();
-        if (!result.isInitialized()) {
-          throw newUninitializedMessageException(result);
-        }
-        return result;
-      }
-
-      @java.lang.Override
-      public org.tensorflow.proto.framework.TensorInfo.CooSparse buildPartial() {
-        org.tensorflow.proto.framework.TensorInfo.CooSparse result = new org.tensorflow.proto.framework.TensorInfo.CooSparse(this);
-        result.valuesTensorName_ = valuesTensorName_;
-        result.indicesTensorName_ = indicesTensorName_;
-        result.denseShapeTensorName_ = denseShapeTensorName_;
-        onBuilt();
-        return result;
-      }
-
-      @java.lang.Override
-      public Builder clone() {
-        return super.clone();
-      }
-      @java.lang.Override
-      public Builder setField(
-          com.google.protobuf.Descriptors.FieldDescriptor field,
-          java.lang.Object value) {
-        return super.setField(field, value);
-      }
-      @java.lang.Override
-      public Builder clearField(
-          com.google.protobuf.Descriptors.FieldDescriptor field) {
-        return super.clearField(field);
-      }
-      @java.lang.Override
-      public Builder clearOneof(
-          com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-        return super.clearOneof(oneof);
-      }
-      @java.lang.Override
-      public Builder setRepeatedField(
-          com.google.protobuf.Descriptors.FieldDescriptor field,
-          int index, java.lang.Object value) {
-        return super.setRepeatedField(field, index, value);
-      }
-      @java.lang.Override
-      public Builder addRepeatedField(
-          com.google.protobuf.Descriptors.FieldDescriptor field,
-          java.lang.Object value) {
-        return super.addRepeatedField(field, value);
-      }
-      @java.lang.Override
-      public Builder mergeFrom(com.google.protobuf.Message other) {
-        if (other instanceof org.tensorflow.proto.framework.TensorInfo.CooSparse) {
-          return mergeFrom((org.tensorflow.proto.framework.TensorInfo.CooSparse)other);
-        } else {
-          super.mergeFrom(other);
-          return this;
-        }
-      }
-
-      public Builder mergeFrom(org.tensorflow.proto.framework.TensorInfo.CooSparse other) {
-        if (other == org.tensorflow.proto.framework.TensorInfo.CooSparse.getDefaultInstance()) return this;
-        if (!other.getValuesTensorName().isEmpty()) {
-          valuesTensorName_ = other.valuesTensorName_;
-          onChanged();
-        }
-        if (!other.getIndicesTensorName().isEmpty()) {
-          indicesTensorName_ = other.indicesTensorName_;
-          onChanged();
-        }
-        if (!other.getDenseShapeTensorName().isEmpty()) {
-          denseShapeTensorName_ = other.denseShapeTensorName_;
-          onChanged();
-        }
-        this.mergeUnknownFields(other.unknownFields);
-        onChanged();
-        return this;
-      }
-
-      @java.lang.Override
-      public final boolean isInitialized() {
-        return true;
-      }
-
-      @java.lang.Override
-      public Builder mergeFrom(
-          com.google.protobuf.CodedInputStream input,
-          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-          throws java.io.IOException {
-        org.tensorflow.proto.framework.TensorInfo.CooSparse parsedMessage = null;
-        try {
-          parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-        } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-          parsedMessage = (org.tensorflow.proto.framework.TensorInfo.CooSparse) e.getUnfinishedMessage();
-          throw e.unwrapIOException();
-        } finally {
-          if (parsedMessage != null) {
-            mergeFrom(parsedMessage);
-          }
-        }
-        return this;
-      }
-
-      private java.lang.Object valuesTensorName_ = "";
-      /**
-       * <pre>
-       * The shape of the values Tensor is [?].  Its dtype must be the dtype of
-       * the SparseTensor as a whole, given in the enclosing TensorInfo.
-       * </pre>
-       *
-       * <code>string values_tensor_name = 1;</code>
-       */
-      public java.lang.String getValuesTensorName() {
-        java.lang.Object ref = valuesTensorName_;
-        if (!(ref instanceof java.lang.String)) {
-          com.google.protobuf.ByteString bs =
-              (com.google.protobuf.ByteString) ref;
-          java.lang.String s = bs.toStringUtf8();
-          valuesTensorName_ = s;
-          return s;
-        } else {
-          return (java.lang.String) ref;
-        }
-      }
-      /**
-       * <pre>
-       * The shape of the values Tensor is [?].  Its dtype must be the dtype of
-       * the SparseTensor as a whole, given in the enclosing TensorInfo.
-       * </pre>
-       *
-       * <code>string values_tensor_name = 1;</code>
-       */
-      public com.google.protobuf.ByteString
-          getValuesTensorNameBytes() {
-        java.lang.Object ref = valuesTensorName_;
-        if (ref instanceof String) {
-          com.google.protobuf.ByteString b = 
-              com.google.protobuf.ByteString.copyFromUtf8(
-                  (java.lang.String) ref);
-          valuesTensorName_ = b;
-          return b;
-        } else {
-          return (com.google.protobuf.ByteString) ref;
-        }
-      }
-      /**
-       * <pre>
-       * The shape of the values Tensor is [?].  Its dtype must be the dtype of
-       * the SparseTensor as a whole, given in the enclosing TensorInfo.
-       * </pre>
-       *
-       * <code>string values_tensor_name = 1;</code>
-       */
-      public Builder setValuesTensorName(
-          java.lang.String value) {
-        if (value == null) {
-    throw new NullPointerException();
-  }
-  
-        valuesTensorName_ = value;
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * The shape of the values Tensor is [?].  Its dtype must be the dtype of
-       * the SparseTensor as a whole, given in the enclosing TensorInfo.
-       * </pre>
-       *
-       * <code>string values_tensor_name = 1;</code>
-       */
-      public Builder clearValuesTensorName() {
-        
-        valuesTensorName_ = getDefaultInstance().getValuesTensorName();
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * The shape of the values Tensor is [?].  Its dtype must be the dtype of
-       * the SparseTensor as a whole, given in the enclosing TensorInfo.
-       * </pre>
-       *
-       * <code>string values_tensor_name = 1;</code>
-       */
-      public Builder setValuesTensorNameBytes(
-          com.google.protobuf.ByteString value) {
-        if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-        
-        valuesTensorName_ = value;
-        onChanged();
-        return this;
-      }
-
-      private java.lang.Object indicesTensorName_ = "";
-      /**
-       * <pre>
-       * The indices Tensor must have dtype int64 and shape [?, ?].
-       * </pre>
-       *
-       * <code>string indices_tensor_name = 2;</code>
-       */
-      public java.lang.String getIndicesTensorName() {
-        java.lang.Object ref = indicesTensorName_;
-        if (!(ref instanceof java.lang.String)) {
-          com.google.protobuf.ByteString bs =
-              (com.google.protobuf.ByteString) ref;
-          java.lang.String s = bs.toStringUtf8();
-          indicesTensorName_ = s;
-          return s;
-        } else {
-          return (java.lang.String) ref;
-        }
-      }
-      /**
-       * <pre>
-       * The indices Tensor must have dtype int64 and shape [?, ?].
-       * </pre>
-       *
-       * <code>string indices_tensor_name = 2;</code>
-       */
-      public com.google.protobuf.ByteString
-          getIndicesTensorNameBytes() {
-        java.lang.Object ref = indicesTensorName_;
-        if (ref instanceof String) {
-          com.google.protobuf.ByteString b = 
-              com.google.protobuf.ByteString.copyFromUtf8(
-                  (java.lang.String) ref);
-          indicesTensorName_ = b;
-          return b;
-        } else {
-          return (com.google.protobuf.ByteString) ref;
-        }
-      }
-      /**
-       * <pre>
-       * The indices Tensor must have dtype int64 and shape [?, ?].
-       * </pre>
-       *
-       * <code>string indices_tensor_name = 2;</code>
-       */
-      public Builder setIndicesTensorName(
-          java.lang.String value) {
-        if (value == null) {
-    throw new NullPointerException();
-  }
-  
-        indicesTensorName_ = value;
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * The indices Tensor must have dtype int64 and shape [?, ?].
-       * </pre>
-       *
-       * <code>string indices_tensor_name = 2;</code>
-       */
-      public Builder clearIndicesTensorName() {
-        
-        indicesTensorName_ = getDefaultInstance().getIndicesTensorName();
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * The indices Tensor must have dtype int64 and shape [?, ?].
-       * </pre>
-       *
-       * <code>string indices_tensor_name = 2;</code>
-       */
-      public Builder setIndicesTensorNameBytes(
-          com.google.protobuf.ByteString value) {
-        if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-        
-        indicesTensorName_ = value;
-        onChanged();
-        return this;
-      }
-
-      private java.lang.Object denseShapeTensorName_ = "";
-      /**
-       * <pre>
-       * The dynamic logical shape represented by the SparseTensor is recorded in
-       * the Tensor referenced here.  It must have dtype int64 and shape [?].
-       * </pre>
-       *
-       * <code>string dense_shape_tensor_name = 3;</code>
-       */
-      public java.lang.String getDenseShapeTensorName() {
-        java.lang.Object ref = denseShapeTensorName_;
-        if (!(ref instanceof java.lang.String)) {
-          com.google.protobuf.ByteString bs =
-              (com.google.protobuf.ByteString) ref;
-          java.lang.String s = bs.toStringUtf8();
-          denseShapeTensorName_ = s;
-          return s;
-        } else {
-          return (java.lang.String) ref;
-        }
-      }
-      /**
-       * <pre>
-       * The dynamic logical shape represented by the SparseTensor is recorded in
-       * the Tensor referenced here.  It must have dtype int64 and shape [?].
-       * </pre>
-       *
-       * <code>string dense_shape_tensor_name = 3;</code>
-       */
-      public com.google.protobuf.ByteString
-          getDenseShapeTensorNameBytes() {
-        java.lang.Object ref = denseShapeTensorName_;
-        if (ref instanceof String) {
-          com.google.protobuf.ByteString b = 
-              com.google.protobuf.ByteString.copyFromUtf8(
-                  (java.lang.String) ref);
-          denseShapeTensorName_ = b;
-          return b;
-        } else {
-          return (com.google.protobuf.ByteString) ref;
-        }
-      }
-      /**
-       * <pre>
-       * The dynamic logical shape represented by the SparseTensor is recorded in
-       * the Tensor referenced here.  It must have dtype int64 and shape [?].
-       * </pre>
-       *
-       * <code>string dense_shape_tensor_name = 3;</code>
-       */
-      public Builder setDenseShapeTensorName(
-          java.lang.String value) {
-        if (value == null) {
-    throw new NullPointerException();
-  }
-  
-        denseShapeTensorName_ = value;
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * The dynamic logical shape represented by the SparseTensor is recorded in
-       * the Tensor referenced here.  It must have dtype int64 and shape [?].
-       * </pre>
-       *
-       * <code>string dense_shape_tensor_name = 3;</code>
-       */
-      public Builder clearDenseShapeTensorName() {
-        
-        denseShapeTensorName_ = getDefaultInstance().getDenseShapeTensorName();
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * The dynamic logical shape represented by the SparseTensor is recorded in
-       * the Tensor referenced here.  It must have dtype int64 and shape [?].
-       * </pre>
-       *
-       * <code>string dense_shape_tensor_name = 3;</code>
-       */
-      public Builder setDenseShapeTensorNameBytes(
-          com.google.protobuf.ByteString value) {
-        if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-        
-        denseShapeTensorName_ = value;
-        onChanged();
-        return this;
-      }
-      @java.lang.Override
-      public final Builder setUnknownFields(
-          final com.google.protobuf.UnknownFieldSet unknownFields) {
-        return super.setUnknownFields(unknownFields);
-      }
-
-      @java.lang.Override
-      public final Builder mergeUnknownFields(
-          final com.google.protobuf.UnknownFieldSet unknownFields) {
-        return super.mergeUnknownFields(unknownFields);
-      }
-
-
-      // @@protoc_insertion_point(builder_scope:tensorflow.TensorInfo.CooSparse)
-    }
-
-    // @@protoc_insertion_point(class_scope:tensorflow.TensorInfo.CooSparse)
-    private static final org.tensorflow.proto.framework.TensorInfo.CooSparse DEFAULT_INSTANCE;
-    static {
-      DEFAULT_INSTANCE = new org.tensorflow.proto.framework.TensorInfo.CooSparse();
-    }
-
-    public static org.tensorflow.proto.framework.TensorInfo.CooSparse getDefaultInstance() {
-      return DEFAULT_INSTANCE;
-    }
-
-    private static final com.google.protobuf.Parser<CooSparse>
-        PARSER = new com.google.protobuf.AbstractParser<CooSparse>() {
-      @java.lang.Override
-      public CooSparse parsePartialFrom(
-          com.google.protobuf.CodedInputStream input,
-          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-          throws com.google.protobuf.InvalidProtocolBufferException {
-        return new CooSparse(input, extensionRegistry);
-      }
-    };
-
-    public static com.google.protobuf.Parser<CooSparse> parser() {
-      return PARSER;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Parser<CooSparse> getParserForType() {
-      return PARSER;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.TensorInfo.CooSparse getDefaultInstanceForType() {
-      return DEFAULT_INSTANCE;
-    }
-
-  }
-
-  public interface CompositeTensorOrBuilder extends
-      // @@protoc_insertion_point(interface_extends:tensorflow.TensorInfo.CompositeTensor)
-      com.google.protobuf.MessageOrBuilder {
-
-    /**
-     * <pre>
-     * The serialized TypeSpec for the composite tensor.
-     * </pre>
-     *
-     * <code>.tensorflow.TypeSpecProto type_spec = 1;</code>
-     */
-    boolean hasTypeSpec();
-    /**
-     * <pre>
-     * The serialized TypeSpec for the composite tensor.
-     * </pre>
-     *
-     * <code>.tensorflow.TypeSpecProto type_spec = 1;</code>
-     */
-    org.tensorflow.proto.framework.TypeSpecProto getTypeSpec();
-    /**
-     * <pre>
-     * The serialized TypeSpec for the composite tensor.
-     * </pre>
-     *
-     * <code>.tensorflow.TypeSpecProto type_spec = 1;</code>
-     */
-    org.tensorflow.proto.framework.TypeSpecProtoOrBuilder getTypeSpecOrBuilder();
-
-    /**
-     * <pre>
-     * A TensorInfo for each flattened component tensor.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorInfo components = 2;</code>
-     */
-    java.util.List<org.tensorflow.proto.framework.TensorInfo> 
-        getComponentsList();
-    /**
-     * <pre>
-     * A TensorInfo for each flattened component tensor.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorInfo components = 2;</code>
-     */
-    org.tensorflow.proto.framework.TensorInfo getComponents(int index);
-    /**
-     * <pre>
-     * A TensorInfo for each flattened component tensor.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorInfo components = 2;</code>
-     */
-    int getComponentsCount();
-    /**
-     * <pre>
-     * A TensorInfo for each flattened component tensor.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorInfo components = 2;</code>
-     */
-    java.util.List<? extends org.tensorflow.proto.framework.TensorInfoOrBuilder> 
-        getComponentsOrBuilderList();
-    /**
-     * <pre>
-     * A TensorInfo for each flattened component tensor.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorInfo components = 2;</code>
-     */
-    org.tensorflow.proto.framework.TensorInfoOrBuilder getComponentsOrBuilder(
-        int index);
-  }
-  /**
-   * <pre>
-   * Generic encoding for composite tensors.
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.TensorInfo.CompositeTensor}
-   */
-  public  static final class CompositeTensor extends
-      com.google.protobuf.GeneratedMessageV3 implements
-      // @@protoc_insertion_point(message_implements:tensorflow.TensorInfo.CompositeTensor)
-      CompositeTensorOrBuilder {
-  private static final long serialVersionUID = 0L;
-    // Use CompositeTensor.newBuilder() to construct.
-    private CompositeTensor(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-      super(builder);
-    }
-    private CompositeTensor() {
-      components_ = java.util.Collections.emptyList();
-    }
-
-    @java.lang.Override
-    @SuppressWarnings({"unused"})
-    protected java.lang.Object newInstance(
-        UnusedPrivateParameter unused) {
-      return new CompositeTensor();
-    }
-
-    @java.lang.Override
-    public final com.google.protobuf.UnknownFieldSet
-    getUnknownFields() {
-      return this.unknownFields;
-    }
-    private CompositeTensor(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      this();
-      if (extensionRegistry == null) {
-        throw new java.lang.NullPointerException();
-      }
-      int mutable_bitField0_ = 0;
-      com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-          com.google.protobuf.UnknownFieldSet.newBuilder();
-      try {
-        boolean done = false;
-        while (!done) {
-          int tag = input.readTag();
-          switch (tag) {
-            case 0:
-              done = true;
-              break;
-            case 10: {
-              org.tensorflow.proto.framework.TypeSpecProto.Builder subBuilder = null;
-              if (typeSpec_ != null) {
-                subBuilder = typeSpec_.toBuilder();
-              }
-              typeSpec_ = input.readMessage(org.tensorflow.proto.framework.TypeSpecProto.parser(), extensionRegistry);
-              if (subBuilder != null) {
-                subBuilder.mergeFrom(typeSpec_);
-                typeSpec_ = subBuilder.buildPartial();
-              }
-
-              break;
-            }
-            case 18: {
-              if (!((mutable_bitField0_ & 0x00000001) != 0)) {
-                components_ = new java.util.ArrayList<org.tensorflow.proto.framework.TensorInfo>();
-                mutable_bitField0_ |= 0x00000001;
-              }
-              components_.add(
-                  input.readMessage(org.tensorflow.proto.framework.TensorInfo.parser(), extensionRegistry));
-              break;
-            }
-            default: {
-              if (!parseUnknownField(
-                  input, unknownFields, extensionRegistry, tag)) {
-                done = true;
-              }
-              break;
-            }
-          }
-        }
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        throw e.setUnfinishedMessage(this);
-      } catch (java.io.IOException e) {
-        throw new com.google.protobuf.InvalidProtocolBufferException(
-            e).setUnfinishedMessage(this);
-      } finally {
-        if (((mutable_bitField0_ & 0x00000001) != 0)) {
-          components_ = java.util.Collections.unmodifiableList(components_);
-        }
-        this.unknownFields = unknownFields.build();
-        makeExtensionsImmutable();
-      }
-    }
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.MetaGraphProtos.internal_static_tensorflow_TensorInfo_CompositeTensor_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.MetaGraphProtos.internal_static_tensorflow_TensorInfo_CompositeTensor_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.TensorInfo.CompositeTensor.class, org.tensorflow.proto.framework.TensorInfo.CompositeTensor.Builder.class);
-    }
-
-    public static final int TYPE_SPEC_FIELD_NUMBER = 1;
-    private org.tensorflow.proto.framework.TypeSpecProto typeSpec_;
-    /**
-     * <pre>
-     * The serialized TypeSpec for the composite tensor.
-     * </pre>
-     *
-     * <code>.tensorflow.TypeSpecProto type_spec = 1;</code>
-     */
-    public boolean hasTypeSpec() {
-      return typeSpec_ != null;
-    }
-    /**
-     * <pre>
-     * The serialized TypeSpec for the composite tensor.
-     * </pre>
-     *
-     * <code>.tensorflow.TypeSpecProto type_spec = 1;</code>
-     */
-    public org.tensorflow.proto.framework.TypeSpecProto getTypeSpec() {
-      return typeSpec_ == null ? org.tensorflow.proto.framework.TypeSpecProto.getDefaultInstance() : typeSpec_;
-    }
-    /**
-     * <pre>
-     * The serialized TypeSpec for the composite tensor.
-     * </pre>
-     *
-     * <code>.tensorflow.TypeSpecProto type_spec = 1;</code>
-     */
-    public org.tensorflow.proto.framework.TypeSpecProtoOrBuilder getTypeSpecOrBuilder() {
-      return getTypeSpec();
-    }
-
-    public static final int COMPONENTS_FIELD_NUMBER = 2;
-    private java.util.List<org.tensorflow.proto.framework.TensorInfo> components_;
-    /**
-     * <pre>
-     * A TensorInfo for each flattened component tensor.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorInfo components = 2;</code>
-     */
-    public java.util.List<org.tensorflow.proto.framework.TensorInfo> getComponentsList() {
-      return components_;
-    }
-    /**
-     * <pre>
-     * A TensorInfo for each flattened component tensor.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorInfo components = 2;</code>
-     */
-    public java.util.List<? extends org.tensorflow.proto.framework.TensorInfoOrBuilder> 
-        getComponentsOrBuilderList() {
-      return components_;
-    }
-    /**
-     * <pre>
-     * A TensorInfo for each flattened component tensor.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorInfo components = 2;</code>
-     */
-    public int getComponentsCount() {
-      return components_.size();
-    }
-    /**
-     * <pre>
-     * A TensorInfo for each flattened component tensor.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorInfo components = 2;</code>
-     */
-    public org.tensorflow.proto.framework.TensorInfo getComponents(int index) {
-      return components_.get(index);
-    }
-    /**
-     * <pre>
-     * A TensorInfo for each flattened component tensor.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorInfo components = 2;</code>
-     */
-    public org.tensorflow.proto.framework.TensorInfoOrBuilder getComponentsOrBuilder(
-        int index) {
-      return components_.get(index);
-    }
-
-    private byte memoizedIsInitialized = -1;
-    @java.lang.Override
-    public final boolean isInitialized() {
-      byte isInitialized = memoizedIsInitialized;
-      if (isInitialized == 1) return true;
-      if (isInitialized == 0) return false;
-
-      memoizedIsInitialized = 1;
-      return true;
-    }
-
-    @java.lang.Override
-    public void writeTo(com.google.protobuf.CodedOutputStream output)
-                        throws java.io.IOException {
-      if (typeSpec_ != null) {
-        output.writeMessage(1, getTypeSpec());
-      }
-      for (int i = 0; i < components_.size(); i++) {
-        output.writeMessage(2, components_.get(i));
-      }
-      unknownFields.writeTo(output);
-    }
-
-    @java.lang.Override
-    public int getSerializedSize() {
-      int size = memoizedSize;
-      if (size != -1) return size;
-
-      size = 0;
-      if (typeSpec_ != null) {
-        size += com.google.protobuf.CodedOutputStream
-          .computeMessageSize(1, getTypeSpec());
-      }
-      for (int i = 0; i < components_.size(); i++) {
-        size += com.google.protobuf.CodedOutputStream
-          .computeMessageSize(2, components_.get(i));
-      }
-      size += unknownFields.getSerializedSize();
-      memoizedSize = size;
-      return size;
-    }
-
-    @java.lang.Override
-    public boolean equals(final java.lang.Object obj) {
-      if (obj == this) {
-       return true;
-      }
-      if (!(obj instanceof org.tensorflow.proto.framework.TensorInfo.CompositeTensor)) {
-        return super.equals(obj);
-      }
-      org.tensorflow.proto.framework.TensorInfo.CompositeTensor other = (org.tensorflow.proto.framework.TensorInfo.CompositeTensor) obj;
-
-      if (hasTypeSpec() != other.hasTypeSpec()) return false;
-      if (hasTypeSpec()) {
-        if (!getTypeSpec()
-            .equals(other.getTypeSpec())) return false;
-      }
-      if (!getComponentsList()
-          .equals(other.getComponentsList())) return false;
-      if (!unknownFields.equals(other.unknownFields)) return false;
-      return true;
-    }
-
-    @java.lang.Override
-    public int hashCode() {
-      if (memoizedHashCode != 0) {
-        return memoizedHashCode;
-      }
-      int hash = 41;
-      hash = (19 * hash) + getDescriptor().hashCode();
-      if (hasTypeSpec()) {
-        hash = (37 * hash) + TYPE_SPEC_FIELD_NUMBER;
-        hash = (53 * hash) + getTypeSpec().hashCode();
-      }
-      if (getComponentsCount() > 0) {
-        hash = (37 * hash) + COMPONENTS_FIELD_NUMBER;
-        hash = (53 * hash) + getComponentsList().hashCode();
-      }
-      hash = (29 * hash) + unknownFields.hashCode();
-      memoizedHashCode = hash;
-      return hash;
-    }
-
-    public static org.tensorflow.proto.framework.TensorInfo.CompositeTensor parseFrom(
-        java.nio.ByteBuffer data)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data);
-    }
-    public static org.tensorflow.proto.framework.TensorInfo.CompositeTensor parseFrom(
-        java.nio.ByteBuffer data,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.TensorInfo.CompositeTensor parseFrom(
-        com.google.protobuf.ByteString data)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data);
-    }
-    public static org.tensorflow.proto.framework.TensorInfo.CompositeTensor parseFrom(
-        com.google.protobuf.ByteString data,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.TensorInfo.CompositeTensor parseFrom(byte[] data)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data);
-    }
-    public static org.tensorflow.proto.framework.TensorInfo.CompositeTensor parseFrom(
-        byte[] data,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.TensorInfo.CompositeTensor parseFrom(java.io.InputStream input)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input);
-    }
-    public static org.tensorflow.proto.framework.TensorInfo.CompositeTensor parseFrom(
-        java.io.InputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.TensorInfo.CompositeTensor parseDelimitedFrom(java.io.InputStream input)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseDelimitedWithIOException(PARSER, input);
-    }
-    public static org.tensorflow.proto.framework.TensorInfo.CompositeTensor parseDelimitedFrom(
-        java.io.InputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.TensorInfo.CompositeTensor parseFrom(
-        com.google.protobuf.CodedInputStream input)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input);
-    }
-    public static org.tensorflow.proto.framework.TensorInfo.CompositeTensor parseFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input, extensionRegistry);
-    }
-
-    @java.lang.Override
-    public Builder newBuilderForType() { return newBuilder(); }
-    public static Builder newBuilder() {
-      return DEFAULT_INSTANCE.toBuilder();
-    }
-    public static Builder newBuilder(org.tensorflow.proto.framework.TensorInfo.CompositeTensor prototype) {
-      return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-    }
-    @java.lang.Override
-    public Builder toBuilder() {
-      return this == DEFAULT_INSTANCE
-          ? new Builder() : new Builder().mergeFrom(this);
-    }
-
-    @java.lang.Override
-    protected Builder newBuilderForType(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      Builder builder = new Builder(parent);
-      return builder;
-    }
-    /**
-     * <pre>
-     * Generic encoding for composite tensors.
-     * </pre>
-     *
-     * Protobuf type {@code tensorflow.TensorInfo.CompositeTensor}
-     */
-    public static final class Builder extends
-        com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-        // @@protoc_insertion_point(builder_implements:tensorflow.TensorInfo.CompositeTensor)
-        org.tensorflow.proto.framework.TensorInfo.CompositeTensorOrBuilder {
-      public static final com.google.protobuf.Descriptors.Descriptor
-          getDescriptor() {
-        return org.tensorflow.proto.framework.MetaGraphProtos.internal_static_tensorflow_TensorInfo_CompositeTensor_descriptor;
-      }
-
-      @java.lang.Override
-      protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-          internalGetFieldAccessorTable() {
-        return org.tensorflow.proto.framework.MetaGraphProtos.internal_static_tensorflow_TensorInfo_CompositeTensor_fieldAccessorTable
-            .ensureFieldAccessorsInitialized(
-                org.tensorflow.proto.framework.TensorInfo.CompositeTensor.class, org.tensorflow.proto.framework.TensorInfo.CompositeTensor.Builder.class);
-      }
-
-      // Construct using org.tensorflow.proto.framework.TensorInfo.CompositeTensor.newBuilder()
-      private Builder() {
-        maybeForceBuilderInitialization();
-      }
-
-      private Builder(
-          com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-        super(parent);
-        maybeForceBuilderInitialization();
-      }
-      private void maybeForceBuilderInitialization() {
-        if (com.google.protobuf.GeneratedMessageV3
-                .alwaysUseFieldBuilders) {
-          getComponentsFieldBuilder();
-        }
-      }
-      @java.lang.Override
-      public Builder clear() {
-        super.clear();
-        if (typeSpecBuilder_ == null) {
-          typeSpec_ = null;
-        } else {
-          typeSpec_ = null;
-          typeSpecBuilder_ = null;
-        }
-        if (componentsBuilder_ == null) {
-          components_ = java.util.Collections.emptyList();
-          bitField0_ = (bitField0_ & ~0x00000001);
-        } else {
-          componentsBuilder_.clear();
-        }
-        return this;
-      }
-
-      @java.lang.Override
-      public com.google.protobuf.Descriptors.Descriptor
-          getDescriptorForType() {
-        return org.tensorflow.proto.framework.MetaGraphProtos.internal_static_tensorflow_TensorInfo_CompositeTensor_descriptor;
-      }
-
-      @java.lang.Override
-      public org.tensorflow.proto.framework.TensorInfo.CompositeTensor getDefaultInstanceForType() {
-        return org.tensorflow.proto.framework.TensorInfo.CompositeTensor.getDefaultInstance();
-      }
-
-      @java.lang.Override
-      public org.tensorflow.proto.framework.TensorInfo.CompositeTensor build() {
-        org.tensorflow.proto.framework.TensorInfo.CompositeTensor result = buildPartial();
-        if (!result.isInitialized()) {
-          throw newUninitializedMessageException(result);
-        }
-        return result;
-      }
-
-      @java.lang.Override
-      public org.tensorflow.proto.framework.TensorInfo.CompositeTensor buildPartial() {
-        org.tensorflow.proto.framework.TensorInfo.CompositeTensor result = new org.tensorflow.proto.framework.TensorInfo.CompositeTensor(this);
-        int from_bitField0_ = bitField0_;
-        if (typeSpecBuilder_ == null) {
-          result.typeSpec_ = typeSpec_;
-        } else {
-          result.typeSpec_ = typeSpecBuilder_.build();
-        }
-        if (componentsBuilder_ == null) {
-          if (((bitField0_ & 0x00000001) != 0)) {
-            components_ = java.util.Collections.unmodifiableList(components_);
-            bitField0_ = (bitField0_ & ~0x00000001);
-          }
-          result.components_ = components_;
-        } else {
-          result.components_ = componentsBuilder_.build();
-        }
-        onBuilt();
-        return result;
-      }
-
-      @java.lang.Override
-      public Builder clone() {
-        return super.clone();
-      }
-      @java.lang.Override
-      public Builder setField(
-          com.google.protobuf.Descriptors.FieldDescriptor field,
-          java.lang.Object value) {
-        return super.setField(field, value);
-      }
-      @java.lang.Override
-      public Builder clearField(
-          com.google.protobuf.Descriptors.FieldDescriptor field) {
-        return super.clearField(field);
-      }
-      @java.lang.Override
-      public Builder clearOneof(
-          com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-        return super.clearOneof(oneof);
-      }
-      @java.lang.Override
-      public Builder setRepeatedField(
-          com.google.protobuf.Descriptors.FieldDescriptor field,
-          int index, java.lang.Object value) {
-        return super.setRepeatedField(field, index, value);
-      }
-      @java.lang.Override
-      public Builder addRepeatedField(
-          com.google.protobuf.Descriptors.FieldDescriptor field,
-          java.lang.Object value) {
-        return super.addRepeatedField(field, value);
-      }
-      @java.lang.Override
-      public Builder mergeFrom(com.google.protobuf.Message other) {
-        if (other instanceof org.tensorflow.proto.framework.TensorInfo.CompositeTensor) {
-          return mergeFrom((org.tensorflow.proto.framework.TensorInfo.CompositeTensor)other);
-        } else {
-          super.mergeFrom(other);
-          return this;
-        }
-      }
-
-      public Builder mergeFrom(org.tensorflow.proto.framework.TensorInfo.CompositeTensor other) {
-        if (other == org.tensorflow.proto.framework.TensorInfo.CompositeTensor.getDefaultInstance()) return this;
-        if (other.hasTypeSpec()) {
-          mergeTypeSpec(other.getTypeSpec());
-        }
-        if (componentsBuilder_ == null) {
-          if (!other.components_.isEmpty()) {
-            if (components_.isEmpty()) {
-              components_ = other.components_;
-              bitField0_ = (bitField0_ & ~0x00000001);
-            } else {
-              ensureComponentsIsMutable();
-              components_.addAll(other.components_);
-            }
-            onChanged();
-          }
-        } else {
-          if (!other.components_.isEmpty()) {
-            if (componentsBuilder_.isEmpty()) {
-              componentsBuilder_.dispose();
-              componentsBuilder_ = null;
-              components_ = other.components_;
-              bitField0_ = (bitField0_ & ~0x00000001);
-              componentsBuilder_ = 
-                com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
-                   getComponentsFieldBuilder() : null;
-            } else {
-              componentsBuilder_.addAllMessages(other.components_);
-            }
-          }
-        }
-        this.mergeUnknownFields(other.unknownFields);
-        onChanged();
-        return this;
-      }
-
-      @java.lang.Override
-      public final boolean isInitialized() {
-        return true;
-      }
-
-      @java.lang.Override
-      public Builder mergeFrom(
-          com.google.protobuf.CodedInputStream input,
-          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-          throws java.io.IOException {
-        org.tensorflow.proto.framework.TensorInfo.CompositeTensor parsedMessage = null;
-        try {
-          parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-        } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-          parsedMessage = (org.tensorflow.proto.framework.TensorInfo.CompositeTensor) e.getUnfinishedMessage();
-          throw e.unwrapIOException();
-        } finally {
-          if (parsedMessage != null) {
-            mergeFrom(parsedMessage);
-          }
-        }
-        return this;
-      }
-      private int bitField0_;
-
-      private org.tensorflow.proto.framework.TypeSpecProto typeSpec_;
-      private com.google.protobuf.SingleFieldBuilderV3<
-          org.tensorflow.proto.framework.TypeSpecProto, org.tensorflow.proto.framework.TypeSpecProto.Builder, org.tensorflow.proto.framework.TypeSpecProtoOrBuilder> typeSpecBuilder_;
-      /**
-       * <pre>
-       * The serialized TypeSpec for the composite tensor.
-       * </pre>
-       *
-       * <code>.tensorflow.TypeSpecProto type_spec = 1;</code>
-       */
-      public boolean hasTypeSpec() {
-        return typeSpecBuilder_ != null || typeSpec_ != null;
-      }
-      /**
-       * <pre>
-       * The serialized TypeSpec for the composite tensor.
-       * </pre>
-       *
-       * <code>.tensorflow.TypeSpecProto type_spec = 1;</code>
-       */
-      public org.tensorflow.proto.framework.TypeSpecProto getTypeSpec() {
-        if (typeSpecBuilder_ == null) {
-          return typeSpec_ == null ? org.tensorflow.proto.framework.TypeSpecProto.getDefaultInstance() : typeSpec_;
-        } else {
-          return typeSpecBuilder_.getMessage();
-        }
-      }
-      /**
-       * <pre>
-       * The serialized TypeSpec for the composite tensor.
-       * </pre>
-       *
-       * <code>.tensorflow.TypeSpecProto type_spec = 1;</code>
-       */
-      public Builder setTypeSpec(org.tensorflow.proto.framework.TypeSpecProto value) {
-        if (typeSpecBuilder_ == null) {
-          if (value == null) {
-            throw new NullPointerException();
-          }
-          typeSpec_ = value;
-          onChanged();
-        } else {
-          typeSpecBuilder_.setMessage(value);
-        }
-
-        return this;
-      }
-      /**
-       * <pre>
-       * The serialized TypeSpec for the composite tensor.
-       * </pre>
-       *
-       * <code>.tensorflow.TypeSpecProto type_spec = 1;</code>
-       */
-      public Builder setTypeSpec(
-          org.tensorflow.proto.framework.TypeSpecProto.Builder builderForValue) {
-        if (typeSpecBuilder_ == null) {
-          typeSpec_ = builderForValue.build();
-          onChanged();
-        } else {
-          typeSpecBuilder_.setMessage(builderForValue.build());
-        }
-
-        return this;
-      }
-      /**
-       * <pre>
-       * The serialized TypeSpec for the composite tensor.
-       * </pre>
-       *
-       * <code>.tensorflow.TypeSpecProto type_spec = 1;</code>
-       */
-      public Builder mergeTypeSpec(org.tensorflow.proto.framework.TypeSpecProto value) {
-        if (typeSpecBuilder_ == null) {
-          if (typeSpec_ != null) {
-            typeSpec_ =
-              org.tensorflow.proto.framework.TypeSpecProto.newBuilder(typeSpec_).mergeFrom(value).buildPartial();
-          } else {
-            typeSpec_ = value;
-          }
-          onChanged();
-        } else {
-          typeSpecBuilder_.mergeFrom(value);
-        }
-
-        return this;
-      }
-      /**
-       * <pre>
-       * The serialized TypeSpec for the composite tensor.
-       * </pre>
-       *
-       * <code>.tensorflow.TypeSpecProto type_spec = 1;</code>
-       */
-      public Builder clearTypeSpec() {
-        if (typeSpecBuilder_ == null) {
-          typeSpec_ = null;
-          onChanged();
-        } else {
-          typeSpec_ = null;
-          typeSpecBuilder_ = null;
-        }
-
-        return this;
-      }
-      /**
-       * <pre>
-       * The serialized TypeSpec for the composite tensor.
-       * </pre>
-       *
-       * <code>.tensorflow.TypeSpecProto type_spec = 1;</code>
-       */
-      public org.tensorflow.proto.framework.TypeSpecProto.Builder getTypeSpecBuilder() {
-        
-        onChanged();
-        return getTypeSpecFieldBuilder().getBuilder();
-      }
-      /**
-       * <pre>
-       * The serialized TypeSpec for the composite tensor.
-       * </pre>
-       *
-       * <code>.tensorflow.TypeSpecProto type_spec = 1;</code>
-       */
-      public org.tensorflow.proto.framework.TypeSpecProtoOrBuilder getTypeSpecOrBuilder() {
-        if (typeSpecBuilder_ != null) {
-          return typeSpecBuilder_.getMessageOrBuilder();
-        } else {
-          return typeSpec_ == null ?
-              org.tensorflow.proto.framework.TypeSpecProto.getDefaultInstance() : typeSpec_;
-        }
-      }
-      /**
-       * <pre>
-       * The serialized TypeSpec for the composite tensor.
-       * </pre>
-       *
-       * <code>.tensorflow.TypeSpecProto type_spec = 1;</code>
-       */
-      private com.google.protobuf.SingleFieldBuilderV3<
-          org.tensorflow.proto.framework.TypeSpecProto, org.tensorflow.proto.framework.TypeSpecProto.Builder, org.tensorflow.proto.framework.TypeSpecProtoOrBuilder> 
-          getTypeSpecFieldBuilder() {
-        if (typeSpecBuilder_ == null) {
-          typeSpecBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-              org.tensorflow.proto.framework.TypeSpecProto, org.tensorflow.proto.framework.TypeSpecProto.Builder, org.tensorflow.proto.framework.TypeSpecProtoOrBuilder>(
-                  getTypeSpec(),
-                  getParentForChildren(),
-                  isClean());
-          typeSpec_ = null;
-        }
-        return typeSpecBuilder_;
-      }
-
-      private java.util.List<org.tensorflow.proto.framework.TensorInfo> components_ =
-        java.util.Collections.emptyList();
-      private void ensureComponentsIsMutable() {
-        if (!((bitField0_ & 0x00000001) != 0)) {
-          components_ = new java.util.ArrayList<org.tensorflow.proto.framework.TensorInfo>(components_);
-          bitField0_ |= 0x00000001;
-         }
-      }
-
-      private com.google.protobuf.RepeatedFieldBuilderV3<
-          org.tensorflow.proto.framework.TensorInfo, org.tensorflow.proto.framework.TensorInfo.Builder, org.tensorflow.proto.framework.TensorInfoOrBuilder> componentsBuilder_;
-
-      /**
-       * <pre>
-       * A TensorInfo for each flattened component tensor.
-       * </pre>
-       *
-       * <code>repeated .tensorflow.TensorInfo components = 2;</code>
-       */
-      public java.util.List<org.tensorflow.proto.framework.TensorInfo> getComponentsList() {
-        if (componentsBuilder_ == null) {
-          return java.util.Collections.unmodifiableList(components_);
-        } else {
-          return componentsBuilder_.getMessageList();
-        }
-      }
-      /**
-       * <pre>
-       * A TensorInfo for each flattened component tensor.
-       * </pre>
-       *
-       * <code>repeated .tensorflow.TensorInfo components = 2;</code>
-       */
-      public int getComponentsCount() {
-        if (componentsBuilder_ == null) {
-          return components_.size();
-        } else {
-          return componentsBuilder_.getCount();
-        }
-      }
-      /**
-       * <pre>
-       * A TensorInfo for each flattened component tensor.
-       * </pre>
-       *
-       * <code>repeated .tensorflow.TensorInfo components = 2;</code>
-       */
-      public org.tensorflow.proto.framework.TensorInfo getComponents(int index) {
-        if (componentsBuilder_ == null) {
-          return components_.get(index);
-        } else {
-          return componentsBuilder_.getMessage(index);
-        }
-      }
-      /**
-       * <pre>
-       * A TensorInfo for each flattened component tensor.
-       * </pre>
-       *
-       * <code>repeated .tensorflow.TensorInfo components = 2;</code>
-       */
-      public Builder setComponents(
-          int index, org.tensorflow.proto.framework.TensorInfo value) {
-        if (componentsBuilder_ == null) {
-          if (value == null) {
-            throw new NullPointerException();
-          }
-          ensureComponentsIsMutable();
-          components_.set(index, value);
-          onChanged();
-        } else {
-          componentsBuilder_.setMessage(index, value);
-        }
-        return this;
-      }
-      /**
-       * <pre>
-       * A TensorInfo for each flattened component tensor.
-       * </pre>
-       *
-       * <code>repeated .tensorflow.TensorInfo components = 2;</code>
-       */
-      public Builder setComponents(
-          int index, org.tensorflow.proto.framework.TensorInfo.Builder builderForValue) {
-        if (componentsBuilder_ == null) {
-          ensureComponentsIsMutable();
-          components_.set(index, builderForValue.build());
-          onChanged();
-        } else {
-          componentsBuilder_.setMessage(index, builderForValue.build());
-        }
-        return this;
-      }
-      /**
-       * <pre>
-       * A TensorInfo for each flattened component tensor.
-       * </pre>
-       *
-       * <code>repeated .tensorflow.TensorInfo components = 2;</code>
-       */
-      public Builder addComponents(org.tensorflow.proto.framework.TensorInfo value) {
-        if (componentsBuilder_ == null) {
-          if (value == null) {
-            throw new NullPointerException();
-          }
-          ensureComponentsIsMutable();
-          components_.add(value);
-          onChanged();
-        } else {
-          componentsBuilder_.addMessage(value);
-        }
-        return this;
-      }
-      /**
-       * <pre>
-       * A TensorInfo for each flattened component tensor.
-       * </pre>
-       *
-       * <code>repeated .tensorflow.TensorInfo components = 2;</code>
-       */
-      public Builder addComponents(
-          int index, org.tensorflow.proto.framework.TensorInfo value) {
-        if (componentsBuilder_ == null) {
-          if (value == null) {
-            throw new NullPointerException();
-          }
-          ensureComponentsIsMutable();
-          components_.add(index, value);
-          onChanged();
-        } else {
-          componentsBuilder_.addMessage(index, value);
-        }
-        return this;
-      }
-      /**
-       * <pre>
-       * A TensorInfo for each flattened component tensor.
-       * </pre>
-       *
-       * <code>repeated .tensorflow.TensorInfo components = 2;</code>
-       */
-      public Builder addComponents(
-          org.tensorflow.proto.framework.TensorInfo.Builder builderForValue) {
-        if (componentsBuilder_ == null) {
-          ensureComponentsIsMutable();
-          components_.add(builderForValue.build());
-          onChanged();
-        } else {
-          componentsBuilder_.addMessage(builderForValue.build());
-        }
-        return this;
-      }
-      /**
-       * <pre>
-       * A TensorInfo for each flattened component tensor.
-       * </pre>
-       *
-       * <code>repeated .tensorflow.TensorInfo components = 2;</code>
-       */
-      public Builder addComponents(
-          int index, org.tensorflow.proto.framework.TensorInfo.Builder builderForValue) {
-        if (componentsBuilder_ == null) {
-          ensureComponentsIsMutable();
-          components_.add(index, builderForValue.build());
-          onChanged();
-        } else {
-          componentsBuilder_.addMessage(index, builderForValue.build());
-        }
-        return this;
-      }
-      /**
-       * <pre>
-       * A TensorInfo for each flattened component tensor.
-       * </pre>
-       *
-       * <code>repeated .tensorflow.TensorInfo components = 2;</code>
-       */
-      public Builder addAllComponents(
-          java.lang.Iterable<? extends org.tensorflow.proto.framework.TensorInfo> values) {
-        if (componentsBuilder_ == null) {
-          ensureComponentsIsMutable();
-          com.google.protobuf.AbstractMessageLite.Builder.addAll(
-              values, components_);
-          onChanged();
-        } else {
-          componentsBuilder_.addAllMessages(values);
-        }
-        return this;
-      }
-      /**
-       * <pre>
-       * A TensorInfo for each flattened component tensor.
-       * </pre>
-       *
-       * <code>repeated .tensorflow.TensorInfo components = 2;</code>
-       */
-      public Builder clearComponents() {
-        if (componentsBuilder_ == null) {
-          components_ = java.util.Collections.emptyList();
-          bitField0_ = (bitField0_ & ~0x00000001);
-          onChanged();
-        } else {
-          componentsBuilder_.clear();
-        }
-        return this;
-      }
-      /**
-       * <pre>
-       * A TensorInfo for each flattened component tensor.
-       * </pre>
-       *
-       * <code>repeated .tensorflow.TensorInfo components = 2;</code>
-       */
-      public Builder removeComponents(int index) {
-        if (componentsBuilder_ == null) {
-          ensureComponentsIsMutable();
-          components_.remove(index);
-          onChanged();
-        } else {
-          componentsBuilder_.remove(index);
-        }
-        return this;
-      }
-      /**
-       * <pre>
-       * A TensorInfo for each flattened component tensor.
-       * </pre>
-       *
-       * <code>repeated .tensorflow.TensorInfo components = 2;</code>
-       */
-      public org.tensorflow.proto.framework.TensorInfo.Builder getComponentsBuilder(
-          int index) {
-        return getComponentsFieldBuilder().getBuilder(index);
-      }
-      /**
-       * <pre>
-       * A TensorInfo for each flattened component tensor.
-       * </pre>
-       *
-       * <code>repeated .tensorflow.TensorInfo components = 2;</code>
-       */
-      public org.tensorflow.proto.framework.TensorInfoOrBuilder getComponentsOrBuilder(
-          int index) {
-        if (componentsBuilder_ == null) {
-          return components_.get(index);  } else {
-          return componentsBuilder_.getMessageOrBuilder(index);
-        }
-      }
-      /**
-       * <pre>
-       * A TensorInfo for each flattened component tensor.
-       * </pre>
-       *
-       * <code>repeated .tensorflow.TensorInfo components = 2;</code>
-       */
-      public java.util.List<? extends org.tensorflow.proto.framework.TensorInfoOrBuilder> 
-           getComponentsOrBuilderList() {
-        if (componentsBuilder_ != null) {
-          return componentsBuilder_.getMessageOrBuilderList();
-        } else {
-          return java.util.Collections.unmodifiableList(components_);
-        }
-      }
-      /**
-       * <pre>
-       * A TensorInfo for each flattened component tensor.
-       * </pre>
-       *
-       * <code>repeated .tensorflow.TensorInfo components = 2;</code>
-       */
-      public org.tensorflow.proto.framework.TensorInfo.Builder addComponentsBuilder() {
-        return getComponentsFieldBuilder().addBuilder(
-            org.tensorflow.proto.framework.TensorInfo.getDefaultInstance());
-      }
-      /**
-       * <pre>
-       * A TensorInfo for each flattened component tensor.
-       * </pre>
-       *
-       * <code>repeated .tensorflow.TensorInfo components = 2;</code>
-       */
-      public org.tensorflow.proto.framework.TensorInfo.Builder addComponentsBuilder(
-          int index) {
-        return getComponentsFieldBuilder().addBuilder(
-            index, org.tensorflow.proto.framework.TensorInfo.getDefaultInstance());
-      }
-      /**
-       * <pre>
-       * A TensorInfo for each flattened component tensor.
-       * </pre>
-       *
-       * <code>repeated .tensorflow.TensorInfo components = 2;</code>
-       */
-      public java.util.List<org.tensorflow.proto.framework.TensorInfo.Builder> 
-           getComponentsBuilderList() {
-        return getComponentsFieldBuilder().getBuilderList();
-      }
-      private com.google.protobuf.RepeatedFieldBuilderV3<
-          org.tensorflow.proto.framework.TensorInfo, org.tensorflow.proto.framework.TensorInfo.Builder, org.tensorflow.proto.framework.TensorInfoOrBuilder> 
-          getComponentsFieldBuilder() {
-        if (componentsBuilder_ == null) {
-          componentsBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
-              org.tensorflow.proto.framework.TensorInfo, org.tensorflow.proto.framework.TensorInfo.Builder, org.tensorflow.proto.framework.TensorInfoOrBuilder>(
-                  components_,
-                  ((bitField0_ & 0x00000001) != 0),
-                  getParentForChildren(),
-                  isClean());
-          components_ = null;
-        }
-        return componentsBuilder_;
-      }
-      @java.lang.Override
-      public final Builder setUnknownFields(
-          final com.google.protobuf.UnknownFieldSet unknownFields) {
-        return super.setUnknownFields(unknownFields);
-      }
-
-      @java.lang.Override
-      public final Builder mergeUnknownFields(
-          final com.google.protobuf.UnknownFieldSet unknownFields) {
-        return super.mergeUnknownFields(unknownFields);
-      }
-
-
-      // @@protoc_insertion_point(builder_scope:tensorflow.TensorInfo.CompositeTensor)
-    }
-
-    // @@protoc_insertion_point(class_scope:tensorflow.TensorInfo.CompositeTensor)
-    private static final org.tensorflow.proto.framework.TensorInfo.CompositeTensor DEFAULT_INSTANCE;
-    static {
-      DEFAULT_INSTANCE = new org.tensorflow.proto.framework.TensorInfo.CompositeTensor();
-    }
-
-    public static org.tensorflow.proto.framework.TensorInfo.CompositeTensor getDefaultInstance() {
-      return DEFAULT_INSTANCE;
-    }
-
-    private static final com.google.protobuf.Parser<CompositeTensor>
-        PARSER = new com.google.protobuf.AbstractParser<CompositeTensor>() {
-      @java.lang.Override
-      public CompositeTensor parsePartialFrom(
-          com.google.protobuf.CodedInputStream input,
-          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-          throws com.google.protobuf.InvalidProtocolBufferException {
-        return new CompositeTensor(input, extensionRegistry);
-      }
-    };
-
-    public static com.google.protobuf.Parser<CompositeTensor> parser() {
-      return PARSER;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Parser<CompositeTensor> getParserForType() {
-      return PARSER;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.TensorInfo.CompositeTensor getDefaultInstanceForType() {
-      return DEFAULT_INSTANCE;
-    }
-
-  }
-
-  private int encodingCase_ = 0;
-  private java.lang.Object encoding_;
-  public enum EncodingCase
-      implements com.google.protobuf.Internal.EnumLite {
-    NAME(1),
-    COO_SPARSE(4),
-    COMPOSITE_TENSOR(5),
-    ENCODING_NOT_SET(0);
-    private final int value;
-    private EncodingCase(int value) {
-      this.value = value;
-    }
-    /**
-     * @deprecated Use {@link #forNumber(int)} instead.
-     */
-    @java.lang.Deprecated
-    public static EncodingCase valueOf(int value) {
-      return forNumber(value);
-    }
-
-    public static EncodingCase forNumber(int value) {
-      switch (value) {
-        case 1: return NAME;
-        case 4: return COO_SPARSE;
-        case 5: return COMPOSITE_TENSOR;
-        case 0: return ENCODING_NOT_SET;
-        default: return null;
-      }
-    }
-    public int getNumber() {
-      return this.value;
-    }
-  };
-
-  public EncodingCase
-  getEncodingCase() {
-    return EncodingCase.forNumber(
-        encodingCase_);
-  }
-
-  public static final int NAME_FIELD_NUMBER = 1;
-  /**
-   * <pre>
-   * For dense `Tensor`s, the name of the tensor in the graph.
-   * </pre>
-   *
-   * <code>string name = 1;</code>
-   */
-  public java.lang.String getName() {
-    java.lang.Object ref = "";
-    if (encodingCase_ == 1) {
-      ref = encoding_;
-    }
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      if (encodingCase_ == 1) {
-        encoding_ = s;
-      }
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * For dense `Tensor`s, the name of the tensor in the graph.
-   * </pre>
-   *
-   * <code>string name = 1;</code>
-   */
-  public com.google.protobuf.ByteString
-      getNameBytes() {
-    java.lang.Object ref = "";
-    if (encodingCase_ == 1) {
-      ref = encoding_;
-    }
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      if (encodingCase_ == 1) {
-        encoding_ = b;
-      }
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int COO_SPARSE_FIELD_NUMBER = 4;
-  /**
-   * <pre>
-   * There are many possible encodings of sparse matrices
-   * (https://en.wikipedia.org/wiki/Sparse_matrix).  Currently, TensorFlow
-   * uses only the COO encoding.  This is supported and documented in the
-   * SparseTensor Python class.
-   * </pre>
-   *
-   * <code>.tensorflow.TensorInfo.CooSparse coo_sparse = 4;</code>
-   */
-  public boolean hasCooSparse() {
-    return encodingCase_ == 4;
-  }
-  /**
-   * <pre>
-   * There are many possible encodings of sparse matrices
-   * (https://en.wikipedia.org/wiki/Sparse_matrix).  Currently, TensorFlow
-   * uses only the COO encoding.  This is supported and documented in the
-   * SparseTensor Python class.
-   * </pre>
-   *
-   * <code>.tensorflow.TensorInfo.CooSparse coo_sparse = 4;</code>
-   */
-  public org.tensorflow.proto.framework.TensorInfo.CooSparse getCooSparse() {
-    if (encodingCase_ == 4) {
-       return (org.tensorflow.proto.framework.TensorInfo.CooSparse) encoding_;
-    }
-    return org.tensorflow.proto.framework.TensorInfo.CooSparse.getDefaultInstance();
-  }
-  /**
-   * <pre>
-   * There are many possible encodings of sparse matrices
-   * (https://en.wikipedia.org/wiki/Sparse_matrix).  Currently, TensorFlow
-   * uses only the COO encoding.  This is supported and documented in the
-   * SparseTensor Python class.
-   * </pre>
-   *
-   * <code>.tensorflow.TensorInfo.CooSparse coo_sparse = 4;</code>
-   */
-  public org.tensorflow.proto.framework.TensorInfo.CooSparseOrBuilder getCooSparseOrBuilder() {
-    if (encodingCase_ == 4) {
-       return (org.tensorflow.proto.framework.TensorInfo.CooSparse) encoding_;
-    }
-    return org.tensorflow.proto.framework.TensorInfo.CooSparse.getDefaultInstance();
-  }
-
-  public static final int COMPOSITE_TENSOR_FIELD_NUMBER = 5;
-  /**
-   * <pre>
-   * Generic encoding for CompositeTensors.
-   * </pre>
-   *
-   * <code>.tensorflow.TensorInfo.CompositeTensor composite_tensor = 5;</code>
-   */
-  public boolean hasCompositeTensor() {
-    return encodingCase_ == 5;
-  }
-  /**
-   * <pre>
-   * Generic encoding for CompositeTensors.
-   * </pre>
-   *
-   * <code>.tensorflow.TensorInfo.CompositeTensor composite_tensor = 5;</code>
-   */
-  public org.tensorflow.proto.framework.TensorInfo.CompositeTensor getCompositeTensor() {
-    if (encodingCase_ == 5) {
-       return (org.tensorflow.proto.framework.TensorInfo.CompositeTensor) encoding_;
-    }
-    return org.tensorflow.proto.framework.TensorInfo.CompositeTensor.getDefaultInstance();
-  }
-  /**
-   * <pre>
-   * Generic encoding for CompositeTensors.
-   * </pre>
-   *
-   * <code>.tensorflow.TensorInfo.CompositeTensor composite_tensor = 5;</code>
-   */
-  public org.tensorflow.proto.framework.TensorInfo.CompositeTensorOrBuilder getCompositeTensorOrBuilder() {
-    if (encodingCase_ == 5) {
-       return (org.tensorflow.proto.framework.TensorInfo.CompositeTensor) encoding_;
-    }
-    return org.tensorflow.proto.framework.TensorInfo.CompositeTensor.getDefaultInstance();
-  }
-
-  public static final int DTYPE_FIELD_NUMBER = 2;
-  private int dtype_;
-  /**
-   * <code>.tensorflow.DataType dtype = 2;</code>
-   */
-  public int getDtypeValue() {
-    return dtype_;
-  }
-  /**
-   * <code>.tensorflow.DataType dtype = 2;</code>
-   */
-  public org.tensorflow.proto.framework.DataType getDtype() {
-    @SuppressWarnings("deprecation")
-    org.tensorflow.proto.framework.DataType result = org.tensorflow.proto.framework.DataType.valueOf(dtype_);
-    return result == null ? org.tensorflow.proto.framework.DataType.UNRECOGNIZED : result;
-  }
-
-  public static final int TENSOR_SHAPE_FIELD_NUMBER = 3;
-  private org.tensorflow.proto.framework.TensorShapeProto tensorShape_;
-  /**
-   * <pre>
-   * The static shape should be recorded here, to the extent that it can
-   * be known in advance.  In the case of a SparseTensor, this field describes
-   * the logical shape of the represented tensor (aka dense_shape).
-   * </pre>
-   *
-   * <code>.tensorflow.TensorShapeProto tensor_shape = 3;</code>
-   */
-  public boolean hasTensorShape() {
-    return tensorShape_ != null;
-  }
-  /**
-   * <pre>
-   * The static shape should be recorded here, to the extent that it can
-   * be known in advance.  In the case of a SparseTensor, this field describes
-   * the logical shape of the represented tensor (aka dense_shape).
-   * </pre>
-   *
-   * <code>.tensorflow.TensorShapeProto tensor_shape = 3;</code>
-   */
-  public org.tensorflow.proto.framework.TensorShapeProto getTensorShape() {
-    return tensorShape_ == null ? org.tensorflow.proto.framework.TensorShapeProto.getDefaultInstance() : tensorShape_;
-  }
-  /**
-   * <pre>
-   * The static shape should be recorded here, to the extent that it can
-   * be known in advance.  In the case of a SparseTensor, this field describes
-   * the logical shape of the represented tensor (aka dense_shape).
-   * </pre>
-   *
-   * <code>.tensorflow.TensorShapeProto tensor_shape = 3;</code>
-   */
-  public org.tensorflow.proto.framework.TensorShapeProtoOrBuilder getTensorShapeOrBuilder() {
-    return getTensorShape();
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    if (encodingCase_ == 1) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 1, encoding_);
-    }
-    if (dtype_ != org.tensorflow.proto.framework.DataType.DT_INVALID.getNumber()) {
-      output.writeEnum(2, dtype_);
-    }
-    if (tensorShape_ != null) {
-      output.writeMessage(3, getTensorShape());
-    }
-    if (encodingCase_ == 4) {
-      output.writeMessage(4, (org.tensorflow.proto.framework.TensorInfo.CooSparse) encoding_);
-    }
-    if (encodingCase_ == 5) {
-      output.writeMessage(5, (org.tensorflow.proto.framework.TensorInfo.CompositeTensor) encoding_);
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (encodingCase_ == 1) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, encoding_);
-    }
-    if (dtype_ != org.tensorflow.proto.framework.DataType.DT_INVALID.getNumber()) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeEnumSize(2, dtype_);
-    }
-    if (tensorShape_ != null) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(3, getTensorShape());
-    }
-    if (encodingCase_ == 4) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(4, (org.tensorflow.proto.framework.TensorInfo.CooSparse) encoding_);
-    }
-    if (encodingCase_ == 5) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(5, (org.tensorflow.proto.framework.TensorInfo.CompositeTensor) encoding_);
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.framework.TensorInfo)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.framework.TensorInfo other = (org.tensorflow.proto.framework.TensorInfo) obj;
-
-    if (dtype_ != other.dtype_) return false;
-    if (hasTensorShape() != other.hasTensorShape()) return false;
-    if (hasTensorShape()) {
-      if (!getTensorShape()
-          .equals(other.getTensorShape())) return false;
-    }
-    if (!getEncodingCase().equals(other.getEncodingCase())) return false;
-    switch (encodingCase_) {
-      case 1:
-        if (!getName()
-            .equals(other.getName())) return false;
-        break;
-      case 4:
-        if (!getCooSparse()
-            .equals(other.getCooSparse())) return false;
-        break;
-      case 5:
-        if (!getCompositeTensor()
-            .equals(other.getCompositeTensor())) return false;
-        break;
-      case 0:
-      default:
-    }
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    hash = (37 * hash) + DTYPE_FIELD_NUMBER;
-    hash = (53 * hash) + dtype_;
-    if (hasTensorShape()) {
-      hash = (37 * hash) + TENSOR_SHAPE_FIELD_NUMBER;
-      hash = (53 * hash) + getTensorShape().hashCode();
-    }
-    switch (encodingCase_) {
-      case 1:
-        hash = (37 * hash) + NAME_FIELD_NUMBER;
-        hash = (53 * hash) + getName().hashCode();
-        break;
-      case 4:
-        hash = (37 * hash) + COO_SPARSE_FIELD_NUMBER;
-        hash = (53 * hash) + getCooSparse().hashCode();
-        break;
-      case 5:
-        hash = (37 * hash) + COMPOSITE_TENSOR_FIELD_NUMBER;
-        hash = (53 * hash) + getCompositeTensor().hashCode();
-        break;
-      case 0:
-      default:
-    }
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.framework.TensorInfo parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.TensorInfo parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.TensorInfo parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.TensorInfo parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.TensorInfo parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.TensorInfo parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.TensorInfo parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.TensorInfo parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.TensorInfo parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.TensorInfo parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.TensorInfo parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.TensorInfo parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.framework.TensorInfo prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * <pre>
-   * Information about a Tensor necessary for feeding or retrieval.
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.TensorInfo}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.TensorInfo)
-      org.tensorflow.proto.framework.TensorInfoOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.MetaGraphProtos.internal_static_tensorflow_TensorInfo_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.MetaGraphProtos.internal_static_tensorflow_TensorInfo_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.TensorInfo.class, org.tensorflow.proto.framework.TensorInfo.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.framework.TensorInfo.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      dtype_ = 0;
-
-      if (tensorShapeBuilder_ == null) {
-        tensorShape_ = null;
-      } else {
-        tensorShape_ = null;
-        tensorShapeBuilder_ = null;
-      }
-      encodingCase_ = 0;
-      encoding_ = null;
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.framework.MetaGraphProtos.internal_static_tensorflow_TensorInfo_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.TensorInfo getDefaultInstanceForType() {
-      return org.tensorflow.proto.framework.TensorInfo.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.TensorInfo build() {
-      org.tensorflow.proto.framework.TensorInfo result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.TensorInfo buildPartial() {
-      org.tensorflow.proto.framework.TensorInfo result = new org.tensorflow.proto.framework.TensorInfo(this);
-      if (encodingCase_ == 1) {
-        result.encoding_ = encoding_;
-      }
-      if (encodingCase_ == 4) {
-        if (cooSparseBuilder_ == null) {
-          result.encoding_ = encoding_;
-        } else {
-          result.encoding_ = cooSparseBuilder_.build();
-        }
-      }
-      if (encodingCase_ == 5) {
-        if (compositeTensorBuilder_ == null) {
-          result.encoding_ = encoding_;
-        } else {
-          result.encoding_ = compositeTensorBuilder_.build();
-        }
-      }
-      result.dtype_ = dtype_;
-      if (tensorShapeBuilder_ == null) {
-        result.tensorShape_ = tensorShape_;
-      } else {
-        result.tensorShape_ = tensorShapeBuilder_.build();
-      }
-      result.encodingCase_ = encodingCase_;
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.framework.TensorInfo) {
-        return mergeFrom((org.tensorflow.proto.framework.TensorInfo)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.framework.TensorInfo other) {
-      if (other == org.tensorflow.proto.framework.TensorInfo.getDefaultInstance()) return this;
-      if (other.dtype_ != 0) {
-        setDtypeValue(other.getDtypeValue());
-      }
-      if (other.hasTensorShape()) {
-        mergeTensorShape(other.getTensorShape());
-      }
-      switch (other.getEncodingCase()) {
-        case NAME: {
-          encodingCase_ = 1;
-          encoding_ = other.encoding_;
-          onChanged();
-          break;
-        }
-        case COO_SPARSE: {
-          mergeCooSparse(other.getCooSparse());
-          break;
-        }
-        case COMPOSITE_TENSOR: {
-          mergeCompositeTensor(other.getCompositeTensor());
-          break;
-        }
-        case ENCODING_NOT_SET: {
-          break;
-        }
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.framework.TensorInfo parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.framework.TensorInfo) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-    private int encodingCase_ = 0;
-    private java.lang.Object encoding_;
-    public EncodingCase
-        getEncodingCase() {
-      return EncodingCase.forNumber(
-          encodingCase_);
-    }
-
-    public Builder clearEncoding() {
-      encodingCase_ = 0;
-      encoding_ = null;
-      onChanged();
-      return this;
-    }
-
-
-    /**
-     * <pre>
-     * For dense `Tensor`s, the name of the tensor in the graph.
-     * </pre>
-     *
-     * <code>string name = 1;</code>
-     */
-    public java.lang.String getName() {
-      java.lang.Object ref = "";
-      if (encodingCase_ == 1) {
-        ref = encoding_;
-      }
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        if (encodingCase_ == 1) {
-          encoding_ = s;
-        }
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * For dense `Tensor`s, the name of the tensor in the graph.
-     * </pre>
-     *
-     * <code>string name = 1;</code>
-     */
-    public com.google.protobuf.ByteString
-        getNameBytes() {
-      java.lang.Object ref = "";
-      if (encodingCase_ == 1) {
-        ref = encoding_;
-      }
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        if (encodingCase_ == 1) {
-          encoding_ = b;
-        }
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * For dense `Tensor`s, the name of the tensor in the graph.
-     * </pre>
-     *
-     * <code>string name = 1;</code>
-     */
-    public Builder setName(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  encodingCase_ = 1;
-      encoding_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * For dense `Tensor`s, the name of the tensor in the graph.
-     * </pre>
-     *
-     * <code>string name = 1;</code>
-     */
-    public Builder clearName() {
-      if (encodingCase_ == 1) {
-        encodingCase_ = 0;
-        encoding_ = null;
-        onChanged();
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * For dense `Tensor`s, the name of the tensor in the graph.
-     * </pre>
-     *
-     * <code>string name = 1;</code>
-     */
-    public Builder setNameBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      encodingCase_ = 1;
-      encoding_ = value;
-      onChanged();
-      return this;
-    }
-
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.TensorInfo.CooSparse, org.tensorflow.proto.framework.TensorInfo.CooSparse.Builder, org.tensorflow.proto.framework.TensorInfo.CooSparseOrBuilder> cooSparseBuilder_;
-    /**
-     * <pre>
-     * There are many possible encodings of sparse matrices
-     * (https://en.wikipedia.org/wiki/Sparse_matrix).  Currently, TensorFlow
-     * uses only the COO encoding.  This is supported and documented in the
-     * SparseTensor Python class.
-     * </pre>
-     *
-     * <code>.tensorflow.TensorInfo.CooSparse coo_sparse = 4;</code>
-     */
-    public boolean hasCooSparse() {
-      return encodingCase_ == 4;
-    }
-    /**
-     * <pre>
-     * There are many possible encodings of sparse matrices
-     * (https://en.wikipedia.org/wiki/Sparse_matrix).  Currently, TensorFlow
-     * uses only the COO encoding.  This is supported and documented in the
-     * SparseTensor Python class.
-     * </pre>
-     *
-     * <code>.tensorflow.TensorInfo.CooSparse coo_sparse = 4;</code>
-     */
-    public org.tensorflow.proto.framework.TensorInfo.CooSparse getCooSparse() {
-      if (cooSparseBuilder_ == null) {
-        if (encodingCase_ == 4) {
-          return (org.tensorflow.proto.framework.TensorInfo.CooSparse) encoding_;
-        }
-        return org.tensorflow.proto.framework.TensorInfo.CooSparse.getDefaultInstance();
-      } else {
-        if (encodingCase_ == 4) {
-          return cooSparseBuilder_.getMessage();
-        }
-        return org.tensorflow.proto.framework.TensorInfo.CooSparse.getDefaultInstance();
-      }
-    }
-    /**
-     * <pre>
-     * There are many possible encodings of sparse matrices
-     * (https://en.wikipedia.org/wiki/Sparse_matrix).  Currently, TensorFlow
-     * uses only the COO encoding.  This is supported and documented in the
-     * SparseTensor Python class.
-     * </pre>
-     *
-     * <code>.tensorflow.TensorInfo.CooSparse coo_sparse = 4;</code>
-     */
-    public Builder setCooSparse(org.tensorflow.proto.framework.TensorInfo.CooSparse value) {
-      if (cooSparseBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        encoding_ = value;
-        onChanged();
-      } else {
-        cooSparseBuilder_.setMessage(value);
-      }
-      encodingCase_ = 4;
-      return this;
-    }
-    /**
-     * <pre>
-     * There are many possible encodings of sparse matrices
-     * (https://en.wikipedia.org/wiki/Sparse_matrix).  Currently, TensorFlow
-     * uses only the COO encoding.  This is supported and documented in the
-     * SparseTensor Python class.
-     * </pre>
-     *
-     * <code>.tensorflow.TensorInfo.CooSparse coo_sparse = 4;</code>
-     */
-    public Builder setCooSparse(
-        org.tensorflow.proto.framework.TensorInfo.CooSparse.Builder builderForValue) {
-      if (cooSparseBuilder_ == null) {
-        encoding_ = builderForValue.build();
-        onChanged();
-      } else {
-        cooSparseBuilder_.setMessage(builderForValue.build());
-      }
-      encodingCase_ = 4;
-      return this;
-    }
-    /**
-     * <pre>
-     * There are many possible encodings of sparse matrices
-     * (https://en.wikipedia.org/wiki/Sparse_matrix).  Currently, TensorFlow
-     * uses only the COO encoding.  This is supported and documented in the
-     * SparseTensor Python class.
-     * </pre>
-     *
-     * <code>.tensorflow.TensorInfo.CooSparse coo_sparse = 4;</code>
-     */
-    public Builder mergeCooSparse(org.tensorflow.proto.framework.TensorInfo.CooSparse value) {
-      if (cooSparseBuilder_ == null) {
-        if (encodingCase_ == 4 &&
-            encoding_ != org.tensorflow.proto.framework.TensorInfo.CooSparse.getDefaultInstance()) {
-          encoding_ = org.tensorflow.proto.framework.TensorInfo.CooSparse.newBuilder((org.tensorflow.proto.framework.TensorInfo.CooSparse) encoding_)
-              .mergeFrom(value).buildPartial();
-        } else {
-          encoding_ = value;
-        }
-        onChanged();
-      } else {
-        if (encodingCase_ == 4) {
-          cooSparseBuilder_.mergeFrom(value);
-        }
-        cooSparseBuilder_.setMessage(value);
-      }
-      encodingCase_ = 4;
-      return this;
-    }
-    /**
-     * <pre>
-     * There are many possible encodings of sparse matrices
-     * (https://en.wikipedia.org/wiki/Sparse_matrix).  Currently, TensorFlow
-     * uses only the COO encoding.  This is supported and documented in the
-     * SparseTensor Python class.
-     * </pre>
-     *
-     * <code>.tensorflow.TensorInfo.CooSparse coo_sparse = 4;</code>
-     */
-    public Builder clearCooSparse() {
-      if (cooSparseBuilder_ == null) {
-        if (encodingCase_ == 4) {
-          encodingCase_ = 0;
-          encoding_ = null;
-          onChanged();
-        }
-      } else {
-        if (encodingCase_ == 4) {
-          encodingCase_ = 0;
-          encoding_ = null;
-        }
-        cooSparseBuilder_.clear();
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * There are many possible encodings of sparse matrices
-     * (https://en.wikipedia.org/wiki/Sparse_matrix).  Currently, TensorFlow
-     * uses only the COO encoding.  This is supported and documented in the
-     * SparseTensor Python class.
-     * </pre>
-     *
-     * <code>.tensorflow.TensorInfo.CooSparse coo_sparse = 4;</code>
-     */
-    public org.tensorflow.proto.framework.TensorInfo.CooSparse.Builder getCooSparseBuilder() {
-      return getCooSparseFieldBuilder().getBuilder();
-    }
-    /**
-     * <pre>
-     * There are many possible encodings of sparse matrices
-     * (https://en.wikipedia.org/wiki/Sparse_matrix).  Currently, TensorFlow
-     * uses only the COO encoding.  This is supported and documented in the
-     * SparseTensor Python class.
-     * </pre>
-     *
-     * <code>.tensorflow.TensorInfo.CooSparse coo_sparse = 4;</code>
-     */
-    public org.tensorflow.proto.framework.TensorInfo.CooSparseOrBuilder getCooSparseOrBuilder() {
-      if ((encodingCase_ == 4) && (cooSparseBuilder_ != null)) {
-        return cooSparseBuilder_.getMessageOrBuilder();
-      } else {
-        if (encodingCase_ == 4) {
-          return (org.tensorflow.proto.framework.TensorInfo.CooSparse) encoding_;
-        }
-        return org.tensorflow.proto.framework.TensorInfo.CooSparse.getDefaultInstance();
-      }
-    }
-    /**
-     * <pre>
-     * There are many possible encodings of sparse matrices
-     * (https://en.wikipedia.org/wiki/Sparse_matrix).  Currently, TensorFlow
-     * uses only the COO encoding.  This is supported and documented in the
-     * SparseTensor Python class.
-     * </pre>
-     *
-     * <code>.tensorflow.TensorInfo.CooSparse coo_sparse = 4;</code>
-     */
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.TensorInfo.CooSparse, org.tensorflow.proto.framework.TensorInfo.CooSparse.Builder, org.tensorflow.proto.framework.TensorInfo.CooSparseOrBuilder> 
-        getCooSparseFieldBuilder() {
-      if (cooSparseBuilder_ == null) {
-        if (!(encodingCase_ == 4)) {
-          encoding_ = org.tensorflow.proto.framework.TensorInfo.CooSparse.getDefaultInstance();
-        }
-        cooSparseBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-            org.tensorflow.proto.framework.TensorInfo.CooSparse, org.tensorflow.proto.framework.TensorInfo.CooSparse.Builder, org.tensorflow.proto.framework.TensorInfo.CooSparseOrBuilder>(
-                (org.tensorflow.proto.framework.TensorInfo.CooSparse) encoding_,
-                getParentForChildren(),
-                isClean());
-        encoding_ = null;
-      }
-      encodingCase_ = 4;
-      onChanged();;
-      return cooSparseBuilder_;
-    }
-
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.TensorInfo.CompositeTensor, org.tensorflow.proto.framework.TensorInfo.CompositeTensor.Builder, org.tensorflow.proto.framework.TensorInfo.CompositeTensorOrBuilder> compositeTensorBuilder_;
-    /**
-     * <pre>
-     * Generic encoding for CompositeTensors.
-     * </pre>
-     *
-     * <code>.tensorflow.TensorInfo.CompositeTensor composite_tensor = 5;</code>
-     */
-    public boolean hasCompositeTensor() {
-      return encodingCase_ == 5;
-    }
-    /**
-     * <pre>
-     * Generic encoding for CompositeTensors.
-     * </pre>
-     *
-     * <code>.tensorflow.TensorInfo.CompositeTensor composite_tensor = 5;</code>
-     */
-    public org.tensorflow.proto.framework.TensorInfo.CompositeTensor getCompositeTensor() {
-      if (compositeTensorBuilder_ == null) {
-        if (encodingCase_ == 5) {
-          return (org.tensorflow.proto.framework.TensorInfo.CompositeTensor) encoding_;
-        }
-        return org.tensorflow.proto.framework.TensorInfo.CompositeTensor.getDefaultInstance();
-      } else {
-        if (encodingCase_ == 5) {
-          return compositeTensorBuilder_.getMessage();
-        }
-        return org.tensorflow.proto.framework.TensorInfo.CompositeTensor.getDefaultInstance();
-      }
-    }
-    /**
-     * <pre>
-     * Generic encoding for CompositeTensors.
-     * </pre>
-     *
-     * <code>.tensorflow.TensorInfo.CompositeTensor composite_tensor = 5;</code>
-     */
-    public Builder setCompositeTensor(org.tensorflow.proto.framework.TensorInfo.CompositeTensor value) {
-      if (compositeTensorBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        encoding_ = value;
-        onChanged();
-      } else {
-        compositeTensorBuilder_.setMessage(value);
-      }
-      encodingCase_ = 5;
-      return this;
-    }
-    /**
-     * <pre>
-     * Generic encoding for CompositeTensors.
-     * </pre>
-     *
-     * <code>.tensorflow.TensorInfo.CompositeTensor composite_tensor = 5;</code>
-     */
-    public Builder setCompositeTensor(
-        org.tensorflow.proto.framework.TensorInfo.CompositeTensor.Builder builderForValue) {
-      if (compositeTensorBuilder_ == null) {
-        encoding_ = builderForValue.build();
-        onChanged();
-      } else {
-        compositeTensorBuilder_.setMessage(builderForValue.build());
-      }
-      encodingCase_ = 5;
-      return this;
-    }
-    /**
-     * <pre>
-     * Generic encoding for CompositeTensors.
-     * </pre>
-     *
-     * <code>.tensorflow.TensorInfo.CompositeTensor composite_tensor = 5;</code>
-     */
-    public Builder mergeCompositeTensor(org.tensorflow.proto.framework.TensorInfo.CompositeTensor value) {
-      if (compositeTensorBuilder_ == null) {
-        if (encodingCase_ == 5 &&
-            encoding_ != org.tensorflow.proto.framework.TensorInfo.CompositeTensor.getDefaultInstance()) {
-          encoding_ = org.tensorflow.proto.framework.TensorInfo.CompositeTensor.newBuilder((org.tensorflow.proto.framework.TensorInfo.CompositeTensor) encoding_)
-              .mergeFrom(value).buildPartial();
-        } else {
-          encoding_ = value;
-        }
-        onChanged();
-      } else {
-        if (encodingCase_ == 5) {
-          compositeTensorBuilder_.mergeFrom(value);
-        }
-        compositeTensorBuilder_.setMessage(value);
-      }
-      encodingCase_ = 5;
-      return this;
-    }
-    /**
-     * <pre>
-     * Generic encoding for CompositeTensors.
-     * </pre>
-     *
-     * <code>.tensorflow.TensorInfo.CompositeTensor composite_tensor = 5;</code>
-     */
-    public Builder clearCompositeTensor() {
-      if (compositeTensorBuilder_ == null) {
-        if (encodingCase_ == 5) {
-          encodingCase_ = 0;
-          encoding_ = null;
-          onChanged();
-        }
-      } else {
-        if (encodingCase_ == 5) {
-          encodingCase_ = 0;
-          encoding_ = null;
-        }
-        compositeTensorBuilder_.clear();
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Generic encoding for CompositeTensors.
-     * </pre>
-     *
-     * <code>.tensorflow.TensorInfo.CompositeTensor composite_tensor = 5;</code>
-     */
-    public org.tensorflow.proto.framework.TensorInfo.CompositeTensor.Builder getCompositeTensorBuilder() {
-      return getCompositeTensorFieldBuilder().getBuilder();
-    }
-    /**
-     * <pre>
-     * Generic encoding for CompositeTensors.
-     * </pre>
-     *
-     * <code>.tensorflow.TensorInfo.CompositeTensor composite_tensor = 5;</code>
-     */
-    public org.tensorflow.proto.framework.TensorInfo.CompositeTensorOrBuilder getCompositeTensorOrBuilder() {
-      if ((encodingCase_ == 5) && (compositeTensorBuilder_ != null)) {
-        return compositeTensorBuilder_.getMessageOrBuilder();
-      } else {
-        if (encodingCase_ == 5) {
-          return (org.tensorflow.proto.framework.TensorInfo.CompositeTensor) encoding_;
-        }
-        return org.tensorflow.proto.framework.TensorInfo.CompositeTensor.getDefaultInstance();
-      }
-    }
-    /**
-     * <pre>
-     * Generic encoding for CompositeTensors.
-     * </pre>
-     *
-     * <code>.tensorflow.TensorInfo.CompositeTensor composite_tensor = 5;</code>
-     */
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.TensorInfo.CompositeTensor, org.tensorflow.proto.framework.TensorInfo.CompositeTensor.Builder, org.tensorflow.proto.framework.TensorInfo.CompositeTensorOrBuilder> 
-        getCompositeTensorFieldBuilder() {
-      if (compositeTensorBuilder_ == null) {
-        if (!(encodingCase_ == 5)) {
-          encoding_ = org.tensorflow.proto.framework.TensorInfo.CompositeTensor.getDefaultInstance();
-        }
-        compositeTensorBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-            org.tensorflow.proto.framework.TensorInfo.CompositeTensor, org.tensorflow.proto.framework.TensorInfo.CompositeTensor.Builder, org.tensorflow.proto.framework.TensorInfo.CompositeTensorOrBuilder>(
-                (org.tensorflow.proto.framework.TensorInfo.CompositeTensor) encoding_,
-                getParentForChildren(),
-                isClean());
-        encoding_ = null;
-      }
-      encodingCase_ = 5;
-      onChanged();;
-      return compositeTensorBuilder_;
-    }
-
-    private int dtype_ = 0;
-    /**
-     * <code>.tensorflow.DataType dtype = 2;</code>
-     */
-    public int getDtypeValue() {
-      return dtype_;
-    }
-    /**
-     * <code>.tensorflow.DataType dtype = 2;</code>
-     */
-    public Builder setDtypeValue(int value) {
-      dtype_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>.tensorflow.DataType dtype = 2;</code>
-     */
-    public org.tensorflow.proto.framework.DataType getDtype() {
-      @SuppressWarnings("deprecation")
-      org.tensorflow.proto.framework.DataType result = org.tensorflow.proto.framework.DataType.valueOf(dtype_);
-      return result == null ? org.tensorflow.proto.framework.DataType.UNRECOGNIZED : result;
-    }
-    /**
-     * <code>.tensorflow.DataType dtype = 2;</code>
-     */
-    public Builder setDtype(org.tensorflow.proto.framework.DataType value) {
-      if (value == null) {
-        throw new NullPointerException();
-      }
-      
-      dtype_ = value.getNumber();
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>.tensorflow.DataType dtype = 2;</code>
-     */
-    public Builder clearDtype() {
-      
-      dtype_ = 0;
-      onChanged();
-      return this;
-    }
-
-    private org.tensorflow.proto.framework.TensorShapeProto tensorShape_;
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.TensorShapeProto, org.tensorflow.proto.framework.TensorShapeProto.Builder, org.tensorflow.proto.framework.TensorShapeProtoOrBuilder> tensorShapeBuilder_;
-    /**
-     * <pre>
-     * The static shape should be recorded here, to the extent that it can
-     * be known in advance.  In the case of a SparseTensor, this field describes
-     * the logical shape of the represented tensor (aka dense_shape).
-     * </pre>
-     *
-     * <code>.tensorflow.TensorShapeProto tensor_shape = 3;</code>
-     */
-    public boolean hasTensorShape() {
-      return tensorShapeBuilder_ != null || tensorShape_ != null;
-    }
-    /**
-     * <pre>
-     * The static shape should be recorded here, to the extent that it can
-     * be known in advance.  In the case of a SparseTensor, this field describes
-     * the logical shape of the represented tensor (aka dense_shape).
-     * </pre>
-     *
-     * <code>.tensorflow.TensorShapeProto tensor_shape = 3;</code>
-     */
-    public org.tensorflow.proto.framework.TensorShapeProto getTensorShape() {
-      if (tensorShapeBuilder_ == null) {
-        return tensorShape_ == null ? org.tensorflow.proto.framework.TensorShapeProto.getDefaultInstance() : tensorShape_;
-      } else {
-        return tensorShapeBuilder_.getMessage();
-      }
-    }
-    /**
-     * <pre>
-     * The static shape should be recorded here, to the extent that it can
-     * be known in advance.  In the case of a SparseTensor, this field describes
-     * the logical shape of the represented tensor (aka dense_shape).
-     * </pre>
-     *
-     * <code>.tensorflow.TensorShapeProto tensor_shape = 3;</code>
-     */
-    public Builder setTensorShape(org.tensorflow.proto.framework.TensorShapeProto value) {
-      if (tensorShapeBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        tensorShape_ = value;
-        onChanged();
-      } else {
-        tensorShapeBuilder_.setMessage(value);
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * The static shape should be recorded here, to the extent that it can
-     * be known in advance.  In the case of a SparseTensor, this field describes
-     * the logical shape of the represented tensor (aka dense_shape).
-     * </pre>
-     *
-     * <code>.tensorflow.TensorShapeProto tensor_shape = 3;</code>
-     */
-    public Builder setTensorShape(
-        org.tensorflow.proto.framework.TensorShapeProto.Builder builderForValue) {
-      if (tensorShapeBuilder_ == null) {
-        tensorShape_ = builderForValue.build();
-        onChanged();
-      } else {
-        tensorShapeBuilder_.setMessage(builderForValue.build());
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * The static shape should be recorded here, to the extent that it can
-     * be known in advance.  In the case of a SparseTensor, this field describes
-     * the logical shape of the represented tensor (aka dense_shape).
-     * </pre>
-     *
-     * <code>.tensorflow.TensorShapeProto tensor_shape = 3;</code>
-     */
-    public Builder mergeTensorShape(org.tensorflow.proto.framework.TensorShapeProto value) {
-      if (tensorShapeBuilder_ == null) {
-        if (tensorShape_ != null) {
-          tensorShape_ =
-            org.tensorflow.proto.framework.TensorShapeProto.newBuilder(tensorShape_).mergeFrom(value).buildPartial();
-        } else {
-          tensorShape_ = value;
-        }
-        onChanged();
-      } else {
-        tensorShapeBuilder_.mergeFrom(value);
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * The static shape should be recorded here, to the extent that it can
-     * be known in advance.  In the case of a SparseTensor, this field describes
-     * the logical shape of the represented tensor (aka dense_shape).
-     * </pre>
-     *
-     * <code>.tensorflow.TensorShapeProto tensor_shape = 3;</code>
-     */
-    public Builder clearTensorShape() {
-      if (tensorShapeBuilder_ == null) {
-        tensorShape_ = null;
-        onChanged();
-      } else {
-        tensorShape_ = null;
-        tensorShapeBuilder_ = null;
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * The static shape should be recorded here, to the extent that it can
-     * be known in advance.  In the case of a SparseTensor, this field describes
-     * the logical shape of the represented tensor (aka dense_shape).
-     * </pre>
-     *
-     * <code>.tensorflow.TensorShapeProto tensor_shape = 3;</code>
-     */
-    public org.tensorflow.proto.framework.TensorShapeProto.Builder getTensorShapeBuilder() {
-      
-      onChanged();
-      return getTensorShapeFieldBuilder().getBuilder();
-    }
-    /**
-     * <pre>
-     * The static shape should be recorded here, to the extent that it can
-     * be known in advance.  In the case of a SparseTensor, this field describes
-     * the logical shape of the represented tensor (aka dense_shape).
-     * </pre>
-     *
-     * <code>.tensorflow.TensorShapeProto tensor_shape = 3;</code>
-     */
-    public org.tensorflow.proto.framework.TensorShapeProtoOrBuilder getTensorShapeOrBuilder() {
-      if (tensorShapeBuilder_ != null) {
-        return tensorShapeBuilder_.getMessageOrBuilder();
-      } else {
-        return tensorShape_ == null ?
-            org.tensorflow.proto.framework.TensorShapeProto.getDefaultInstance() : tensorShape_;
-      }
-    }
-    /**
-     * <pre>
-     * The static shape should be recorded here, to the extent that it can
-     * be known in advance.  In the case of a SparseTensor, this field describes
-     * the logical shape of the represented tensor (aka dense_shape).
-     * </pre>
-     *
-     * <code>.tensorflow.TensorShapeProto tensor_shape = 3;</code>
-     */
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.TensorShapeProto, org.tensorflow.proto.framework.TensorShapeProto.Builder, org.tensorflow.proto.framework.TensorShapeProtoOrBuilder> 
-        getTensorShapeFieldBuilder() {
-      if (tensorShapeBuilder_ == null) {
-        tensorShapeBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-            org.tensorflow.proto.framework.TensorShapeProto, org.tensorflow.proto.framework.TensorShapeProto.Builder, org.tensorflow.proto.framework.TensorShapeProtoOrBuilder>(
-                getTensorShape(),
-                getParentForChildren(),
-                isClean());
-        tensorShape_ = null;
-      }
-      return tensorShapeBuilder_;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.TensorInfo)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.TensorInfo)
-  private static final org.tensorflow.proto.framework.TensorInfo DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.TensorInfo();
-  }
-
-  public static org.tensorflow.proto.framework.TensorInfo getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<TensorInfo>
-      PARSER = new com.google.protobuf.AbstractParser<TensorInfo>() {
-    @java.lang.Override
-    public TensorInfo parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new TensorInfo(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<TensorInfo> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<TensorInfo> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.framework.TensorInfo getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorInfoOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorInfoOrBuilder.java
deleted file mode 100644
index d30570a1969..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorInfoOrBuilder.java
+++ /dev/null
@@ -1,128 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/meta_graph.proto
-
-package org.tensorflow.proto.framework;
-
-public interface TensorInfoOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.TensorInfo)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <pre>
-   * For dense `Tensor`s, the name of the tensor in the graph.
-   * </pre>
-   *
-   * <code>string name = 1;</code>
-   */
-  java.lang.String getName();
-  /**
-   * <pre>
-   * For dense `Tensor`s, the name of the tensor in the graph.
-   * </pre>
-   *
-   * <code>string name = 1;</code>
-   */
-  com.google.protobuf.ByteString
-      getNameBytes();
-
-  /**
-   * <pre>
-   * There are many possible encodings of sparse matrices
-   * (https://en.wikipedia.org/wiki/Sparse_matrix).  Currently, TensorFlow
-   * uses only the COO encoding.  This is supported and documented in the
-   * SparseTensor Python class.
-   * </pre>
-   *
-   * <code>.tensorflow.TensorInfo.CooSparse coo_sparse = 4;</code>
-   */
-  boolean hasCooSparse();
-  /**
-   * <pre>
-   * There are many possible encodings of sparse matrices
-   * (https://en.wikipedia.org/wiki/Sparse_matrix).  Currently, TensorFlow
-   * uses only the COO encoding.  This is supported and documented in the
-   * SparseTensor Python class.
-   * </pre>
-   *
-   * <code>.tensorflow.TensorInfo.CooSparse coo_sparse = 4;</code>
-   */
-  org.tensorflow.proto.framework.TensorInfo.CooSparse getCooSparse();
-  /**
-   * <pre>
-   * There are many possible encodings of sparse matrices
-   * (https://en.wikipedia.org/wiki/Sparse_matrix).  Currently, TensorFlow
-   * uses only the COO encoding.  This is supported and documented in the
-   * SparseTensor Python class.
-   * </pre>
-   *
-   * <code>.tensorflow.TensorInfo.CooSparse coo_sparse = 4;</code>
-   */
-  org.tensorflow.proto.framework.TensorInfo.CooSparseOrBuilder getCooSparseOrBuilder();
-
-  /**
-   * <pre>
-   * Generic encoding for CompositeTensors.
-   * </pre>
-   *
-   * <code>.tensorflow.TensorInfo.CompositeTensor composite_tensor = 5;</code>
-   */
-  boolean hasCompositeTensor();
-  /**
-   * <pre>
-   * Generic encoding for CompositeTensors.
-   * </pre>
-   *
-   * <code>.tensorflow.TensorInfo.CompositeTensor composite_tensor = 5;</code>
-   */
-  org.tensorflow.proto.framework.TensorInfo.CompositeTensor getCompositeTensor();
-  /**
-   * <pre>
-   * Generic encoding for CompositeTensors.
-   * </pre>
-   *
-   * <code>.tensorflow.TensorInfo.CompositeTensor composite_tensor = 5;</code>
-   */
-  org.tensorflow.proto.framework.TensorInfo.CompositeTensorOrBuilder getCompositeTensorOrBuilder();
-
-  /**
-   * <code>.tensorflow.DataType dtype = 2;</code>
-   */
-  int getDtypeValue();
-  /**
-   * <code>.tensorflow.DataType dtype = 2;</code>
-   */
-  org.tensorflow.proto.framework.DataType getDtype();
-
-  /**
-   * <pre>
-   * The static shape should be recorded here, to the extent that it can
-   * be known in advance.  In the case of a SparseTensor, this field describes
-   * the logical shape of the represented tensor (aka dense_shape).
-   * </pre>
-   *
-   * <code>.tensorflow.TensorShapeProto tensor_shape = 3;</code>
-   */
-  boolean hasTensorShape();
-  /**
-   * <pre>
-   * The static shape should be recorded here, to the extent that it can
-   * be known in advance.  In the case of a SparseTensor, this field describes
-   * the logical shape of the represented tensor (aka dense_shape).
-   * </pre>
-   *
-   * <code>.tensorflow.TensorShapeProto tensor_shape = 3;</code>
-   */
-  org.tensorflow.proto.framework.TensorShapeProto getTensorShape();
-  /**
-   * <pre>
-   * The static shape should be recorded here, to the extent that it can
-   * be known in advance.  In the case of a SparseTensor, this field describes
-   * the logical shape of the represented tensor (aka dense_shape).
-   * </pre>
-   *
-   * <code>.tensorflow.TensorShapeProto tensor_shape = 3;</code>
-   */
-  org.tensorflow.proto.framework.TensorShapeProtoOrBuilder getTensorShapeOrBuilder();
-
-  public org.tensorflow.proto.framework.TensorInfo.EncodingCase getEncodingCase();
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorProto.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorProto.java
deleted file mode 100644
index 756d17fef2a..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorProto.java
+++ /dev/null
@@ -1,3980 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/framework/tensor.proto
-
-package org.tensorflow.proto.framework;
-
-/**
- * <pre>
- * Protocol buffer representing a tensor.
- * </pre>
- *
- * Protobuf type {@code tensorflow.TensorProto}
- */
-public  final class TensorProto extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.TensorProto)
-    TensorProtoOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use TensorProto.newBuilder() to construct.
-  private TensorProto(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private TensorProto() {
-    dtype_ = 0;
-    tensorContent_ = com.google.protobuf.ByteString.EMPTY;
-    halfVal_ = emptyIntList();
-    floatVal_ = emptyFloatList();
-    doubleVal_ = emptyDoubleList();
-    intVal_ = emptyIntList();
-    stringVal_ = java.util.Collections.emptyList();
-    scomplexVal_ = emptyFloatList();
-    int64Val_ = emptyLongList();
-    boolVal_ = emptyBooleanList();
-    dcomplexVal_ = emptyDoubleList();
-    resourceHandleVal_ = java.util.Collections.emptyList();
-    variantVal_ = java.util.Collections.emptyList();
-    uint32Val_ = emptyIntList();
-    uint64Val_ = emptyLongList();
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new TensorProto();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private TensorProto(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    int mutable_bitField0_ = 0;
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 8: {
-            int rawValue = input.readEnum();
-
-            dtype_ = rawValue;
-            break;
-          }
-          case 18: {
-            org.tensorflow.proto.framework.TensorShapeProto.Builder subBuilder = null;
-            if (tensorShape_ != null) {
-              subBuilder = tensorShape_.toBuilder();
-            }
-            tensorShape_ = input.readMessage(org.tensorflow.proto.framework.TensorShapeProto.parser(), extensionRegistry);
-            if (subBuilder != null) {
-              subBuilder.mergeFrom(tensorShape_);
-              tensorShape_ = subBuilder.buildPartial();
-            }
-
-            break;
-          }
-          case 24: {
-
-            versionNumber_ = input.readInt32();
-            break;
-          }
-          case 34: {
-
-            tensorContent_ = input.readBytes();
-            break;
-          }
-          case 45: {
-            if (!((mutable_bitField0_ & 0x00000002) != 0)) {
-              floatVal_ = newFloatList();
-              mutable_bitField0_ |= 0x00000002;
-            }
-            floatVal_.addFloat(input.readFloat());
-            break;
-          }
-          case 42: {
-            int length = input.readRawVarint32();
-            int limit = input.pushLimit(length);
-            if (!((mutable_bitField0_ & 0x00000002) != 0) && input.getBytesUntilLimit() > 0) {
-              floatVal_ = newFloatList();
-              mutable_bitField0_ |= 0x00000002;
-            }
-            while (input.getBytesUntilLimit() > 0) {
-              floatVal_.addFloat(input.readFloat());
-            }
-            input.popLimit(limit);
-            break;
-          }
-          case 49: {
-            if (!((mutable_bitField0_ & 0x00000004) != 0)) {
-              doubleVal_ = newDoubleList();
-              mutable_bitField0_ |= 0x00000004;
-            }
-            doubleVal_.addDouble(input.readDouble());
-            break;
-          }
-          case 50: {
-            int length = input.readRawVarint32();
-            int limit = input.pushLimit(length);
-            if (!((mutable_bitField0_ & 0x00000004) != 0) && input.getBytesUntilLimit() > 0) {
-              doubleVal_ = newDoubleList();
-              mutable_bitField0_ |= 0x00000004;
-            }
-            while (input.getBytesUntilLimit() > 0) {
-              doubleVal_.addDouble(input.readDouble());
-            }
-            input.popLimit(limit);
-            break;
-          }
-          case 56: {
-            if (!((mutable_bitField0_ & 0x00000008) != 0)) {
-              intVal_ = newIntList();
-              mutable_bitField0_ |= 0x00000008;
-            }
-            intVal_.addInt(input.readInt32());
-            break;
-          }
-          case 58: {
-            int length = input.readRawVarint32();
-            int limit = input.pushLimit(length);
-            if (!((mutable_bitField0_ & 0x00000008) != 0) && input.getBytesUntilLimit() > 0) {
-              intVal_ = newIntList();
-              mutable_bitField0_ |= 0x00000008;
-            }
-            while (input.getBytesUntilLimit() > 0) {
-              intVal_.addInt(input.readInt32());
-            }
-            input.popLimit(limit);
-            break;
-          }
-          case 66: {
-            if (!((mutable_bitField0_ & 0x00000010) != 0)) {
-              stringVal_ = new java.util.ArrayList<com.google.protobuf.ByteString>();
-              mutable_bitField0_ |= 0x00000010;
-            }
-            stringVal_.add(input.readBytes());
-            break;
-          }
-          case 77: {
-            if (!((mutable_bitField0_ & 0x00000020) != 0)) {
-              scomplexVal_ = newFloatList();
-              mutable_bitField0_ |= 0x00000020;
-            }
-            scomplexVal_.addFloat(input.readFloat());
-            break;
-          }
-          case 74: {
-            int length = input.readRawVarint32();
-            int limit = input.pushLimit(length);
-            if (!((mutable_bitField0_ & 0x00000020) != 0) && input.getBytesUntilLimit() > 0) {
-              scomplexVal_ = newFloatList();
-              mutable_bitField0_ |= 0x00000020;
-            }
-            while (input.getBytesUntilLimit() > 0) {
-              scomplexVal_.addFloat(input.readFloat());
-            }
-            input.popLimit(limit);
-            break;
-          }
-          case 80: {
-            if (!((mutable_bitField0_ & 0x00000040) != 0)) {
-              int64Val_ = newLongList();
-              mutable_bitField0_ |= 0x00000040;
-            }
-            int64Val_.addLong(input.readInt64());
-            break;
-          }
-          case 82: {
-            int length = input.readRawVarint32();
-            int limit = input.pushLimit(length);
-            if (!((mutable_bitField0_ & 0x00000040) != 0) && input.getBytesUntilLimit() > 0) {
-              int64Val_ = newLongList();
-              mutable_bitField0_ |= 0x00000040;
-            }
-            while (input.getBytesUntilLimit() > 0) {
-              int64Val_.addLong(input.readInt64());
-            }
-            input.popLimit(limit);
-            break;
-          }
-          case 88: {
-            if (!((mutable_bitField0_ & 0x00000080) != 0)) {
-              boolVal_ = newBooleanList();
-              mutable_bitField0_ |= 0x00000080;
-            }
-            boolVal_.addBoolean(input.readBool());
-            break;
-          }
-          case 90: {
-            int length = input.readRawVarint32();
-            int limit = input.pushLimit(length);
-            if (!((mutable_bitField0_ & 0x00000080) != 0) && input.getBytesUntilLimit() > 0) {
-              boolVal_ = newBooleanList();
-              mutable_bitField0_ |= 0x00000080;
-            }
-            while (input.getBytesUntilLimit() > 0) {
-              boolVal_.addBoolean(input.readBool());
-            }
-            input.popLimit(limit);
-            break;
-          }
-          case 97: {
-            if (!((mutable_bitField0_ & 0x00000100) != 0)) {
-              dcomplexVal_ = newDoubleList();
-              mutable_bitField0_ |= 0x00000100;
-            }
-            dcomplexVal_.addDouble(input.readDouble());
-            break;
-          }
-          case 98: {
-            int length = input.readRawVarint32();
-            int limit = input.pushLimit(length);
-            if (!((mutable_bitField0_ & 0x00000100) != 0) && input.getBytesUntilLimit() > 0) {
-              dcomplexVal_ = newDoubleList();
-              mutable_bitField0_ |= 0x00000100;
-            }
-            while (input.getBytesUntilLimit() > 0) {
-              dcomplexVal_.addDouble(input.readDouble());
-            }
-            input.popLimit(limit);
-            break;
-          }
-          case 104: {
-            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
-              halfVal_ = newIntList();
-              mutable_bitField0_ |= 0x00000001;
-            }
-            halfVal_.addInt(input.readInt32());
-            break;
-          }
-          case 106: {
-            int length = input.readRawVarint32();
-            int limit = input.pushLimit(length);
-            if (!((mutable_bitField0_ & 0x00000001) != 0) && input.getBytesUntilLimit() > 0) {
-              halfVal_ = newIntList();
-              mutable_bitField0_ |= 0x00000001;
-            }
-            while (input.getBytesUntilLimit() > 0) {
-              halfVal_.addInt(input.readInt32());
-            }
-            input.popLimit(limit);
-            break;
-          }
-          case 114: {
-            if (!((mutable_bitField0_ & 0x00000200) != 0)) {
-              resourceHandleVal_ = new java.util.ArrayList<org.tensorflow.proto.framework.ResourceHandleProto>();
-              mutable_bitField0_ |= 0x00000200;
-            }
-            resourceHandleVal_.add(
-                input.readMessage(org.tensorflow.proto.framework.ResourceHandleProto.parser(), extensionRegistry));
-            break;
-          }
-          case 122: {
-            if (!((mutable_bitField0_ & 0x00000400) != 0)) {
-              variantVal_ = new java.util.ArrayList<org.tensorflow.proto.framework.VariantTensorDataProto>();
-              mutable_bitField0_ |= 0x00000400;
-            }
-            variantVal_.add(
-                input.readMessage(org.tensorflow.proto.framework.VariantTensorDataProto.parser(), extensionRegistry));
-            break;
-          }
-          case 128: {
-            if (!((mutable_bitField0_ & 0x00000800) != 0)) {
-              uint32Val_ = newIntList();
-              mutable_bitField0_ |= 0x00000800;
-            }
-            uint32Val_.addInt(input.readUInt32());
-            break;
-          }
-          case 130: {
-            int length = input.readRawVarint32();
-            int limit = input.pushLimit(length);
-            if (!((mutable_bitField0_ & 0x00000800) != 0) && input.getBytesUntilLimit() > 0) {
-              uint32Val_ = newIntList();
-              mutable_bitField0_ |= 0x00000800;
-            }
-            while (input.getBytesUntilLimit() > 0) {
-              uint32Val_.addInt(input.readUInt32());
-            }
-            input.popLimit(limit);
-            break;
-          }
-          case 136: {
-            if (!((mutable_bitField0_ & 0x00001000) != 0)) {
-              uint64Val_ = newLongList();
-              mutable_bitField0_ |= 0x00001000;
-            }
-            uint64Val_.addLong(input.readUInt64());
-            break;
-          }
-          case 138: {
-            int length = input.readRawVarint32();
-            int limit = input.pushLimit(length);
-            if (!((mutable_bitField0_ & 0x00001000) != 0) && input.getBytesUntilLimit() > 0) {
-              uint64Val_ = newLongList();
-              mutable_bitField0_ |= 0x00001000;
-            }
-            while (input.getBytesUntilLimit() > 0) {
-              uint64Val_.addLong(input.readUInt64());
-            }
-            input.popLimit(limit);
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      if (((mutable_bitField0_ & 0x00000002) != 0)) {
-        floatVal_.makeImmutable(); // C
-      }
-      if (((mutable_bitField0_ & 0x00000004) != 0)) {
-        doubleVal_.makeImmutable(); // C
-      }
-      if (((mutable_bitField0_ & 0x00000008) != 0)) {
-        intVal_.makeImmutable(); // C
-      }
-      if (((mutable_bitField0_ & 0x00000010) != 0)) {
-        stringVal_ = java.util.Collections.unmodifiableList(stringVal_); // C
-      }
-      if (((mutable_bitField0_ & 0x00000020) != 0)) {
-        scomplexVal_.makeImmutable(); // C
-      }
-      if (((mutable_bitField0_ & 0x00000040) != 0)) {
-        int64Val_.makeImmutable(); // C
-      }
-      if (((mutable_bitField0_ & 0x00000080) != 0)) {
-        boolVal_.makeImmutable(); // C
-      }
-      if (((mutable_bitField0_ & 0x00000100) != 0)) {
-        dcomplexVal_.makeImmutable(); // C
-      }
-      if (((mutable_bitField0_ & 0x00000001) != 0)) {
-        halfVal_.makeImmutable(); // C
-      }
-      if (((mutable_bitField0_ & 0x00000200) != 0)) {
-        resourceHandleVal_ = java.util.Collections.unmodifiableList(resourceHandleVal_);
-      }
-      if (((mutable_bitField0_ & 0x00000400) != 0)) {
-        variantVal_ = java.util.Collections.unmodifiableList(variantVal_);
-      }
-      if (((mutable_bitField0_ & 0x00000800) != 0)) {
-        uint32Val_.makeImmutable(); // C
-      }
-      if (((mutable_bitField0_ & 0x00001000) != 0)) {
-        uint64Val_.makeImmutable(); // C
-      }
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.framework.TensorProtos.internal_static_tensorflow_TensorProto_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.framework.TensorProtos.internal_static_tensorflow_TensorProto_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.framework.TensorProto.class, org.tensorflow.proto.framework.TensorProto.Builder.class);
-  }
-
-  public static final int DTYPE_FIELD_NUMBER = 1;
-  private int dtype_;
-  /**
-   * <code>.tensorflow.DataType dtype = 1;</code>
-   */
-  public int getDtypeValue() {
-    return dtype_;
-  }
-  /**
-   * <code>.tensorflow.DataType dtype = 1;</code>
-   */
-  public org.tensorflow.proto.framework.DataType getDtype() {
-    @SuppressWarnings("deprecation")
-    org.tensorflow.proto.framework.DataType result = org.tensorflow.proto.framework.DataType.valueOf(dtype_);
-    return result == null ? org.tensorflow.proto.framework.DataType.UNRECOGNIZED : result;
-  }
-
-  public static final int TENSOR_SHAPE_FIELD_NUMBER = 2;
-  private org.tensorflow.proto.framework.TensorShapeProto tensorShape_;
-  /**
-   * <pre>
-   * Shape of the tensor.  TODO(touts): sort out the 0-rank issues.
-   * </pre>
-   *
-   * <code>.tensorflow.TensorShapeProto tensor_shape = 2;</code>
-   */
-  public boolean hasTensorShape() {
-    return tensorShape_ != null;
-  }
-  /**
-   * <pre>
-   * Shape of the tensor.  TODO(touts): sort out the 0-rank issues.
-   * </pre>
-   *
-   * <code>.tensorflow.TensorShapeProto tensor_shape = 2;</code>
-   */
-  public org.tensorflow.proto.framework.TensorShapeProto getTensorShape() {
-    return tensorShape_ == null ? org.tensorflow.proto.framework.TensorShapeProto.getDefaultInstance() : tensorShape_;
-  }
-  /**
-   * <pre>
-   * Shape of the tensor.  TODO(touts): sort out the 0-rank issues.
-   * </pre>
-   *
-   * <code>.tensorflow.TensorShapeProto tensor_shape = 2;</code>
-   */
-  public org.tensorflow.proto.framework.TensorShapeProtoOrBuilder getTensorShapeOrBuilder() {
-    return getTensorShape();
-  }
-
-  public static final int VERSION_NUMBER_FIELD_NUMBER = 3;
-  private int versionNumber_;
-  /**
-   * <pre>
-   * Version number.
-   * In version 0, if the "repeated xxx" representations contain only one
-   * element, that element is repeated to fill the shape.  This makes it easy
-   * to represent a constant Tensor with a single value.
-   * </pre>
-   *
-   * <code>int32 version_number = 3;</code>
-   */
-  public int getVersionNumber() {
-    return versionNumber_;
-  }
-
-  public static final int TENSOR_CONTENT_FIELD_NUMBER = 4;
-  private com.google.protobuf.ByteString tensorContent_;
-  /**
-   * <pre>
-   * Serialized raw tensor content from either Tensor::AsProtoTensorContent or
-   * memcpy in tensorflow::grpc::EncodeTensorToByteBuffer. This representation
-   * can be used for all tensor types. The purpose of this representation is to
-   * reduce serialization overhead during RPC call by avoiding serialization of
-   * many repeated small items.
-   * </pre>
-   *
-   * <code>bytes tensor_content = 4;</code>
-   */
-  public com.google.protobuf.ByteString getTensorContent() {
-    return tensorContent_;
-  }
-
-  public static final int HALF_VAL_FIELD_NUMBER = 13;
-  private com.google.protobuf.Internal.IntList halfVal_;
-  /**
-   * <pre>
-   * DT_HALF, DT_BFLOAT16. Note that since protobuf has no int16 type, we'll
-   * have some pointless zero padding for each value here.
-   * </pre>
-   *
-   * <code>repeated int32 half_val = 13 [packed = true];</code>
-   */
-  public java.util.List<java.lang.Integer>
-      getHalfValList() {
-    return halfVal_;
-  }
-  /**
-   * <pre>
-   * DT_HALF, DT_BFLOAT16. Note that since protobuf has no int16 type, we'll
-   * have some pointless zero padding for each value here.
-   * </pre>
-   *
-   * <code>repeated int32 half_val = 13 [packed = true];</code>
-   */
-  public int getHalfValCount() {
-    return halfVal_.size();
-  }
-  /**
-   * <pre>
-   * DT_HALF, DT_BFLOAT16. Note that since protobuf has no int16 type, we'll
-   * have some pointless zero padding for each value here.
-   * </pre>
-   *
-   * <code>repeated int32 half_val = 13 [packed = true];</code>
-   */
-  public int getHalfVal(int index) {
-    return halfVal_.getInt(index);
-  }
-  private int halfValMemoizedSerializedSize = -1;
-
-  public static final int FLOAT_VAL_FIELD_NUMBER = 5;
-  private com.google.protobuf.Internal.FloatList floatVal_;
-  /**
-   * <pre>
-   * DT_FLOAT.
-   * </pre>
-   *
-   * <code>repeated float float_val = 5 [packed = true];</code>
-   */
-  public java.util.List<java.lang.Float>
-      getFloatValList() {
-    return floatVal_;
-  }
-  /**
-   * <pre>
-   * DT_FLOAT.
-   * </pre>
-   *
-   * <code>repeated float float_val = 5 [packed = true];</code>
-   */
-  public int getFloatValCount() {
-    return floatVal_.size();
-  }
-  /**
-   * <pre>
-   * DT_FLOAT.
-   * </pre>
-   *
-   * <code>repeated float float_val = 5 [packed = true];</code>
-   */
-  public float getFloatVal(int index) {
-    return floatVal_.getFloat(index);
-  }
-  private int floatValMemoizedSerializedSize = -1;
-
-  public static final int DOUBLE_VAL_FIELD_NUMBER = 6;
-  private com.google.protobuf.Internal.DoubleList doubleVal_;
-  /**
-   * <pre>
-   * DT_DOUBLE.
-   * </pre>
-   *
-   * <code>repeated double double_val = 6 [packed = true];</code>
-   */
-  public java.util.List<java.lang.Double>
-      getDoubleValList() {
-    return doubleVal_;
-  }
-  /**
-   * <pre>
-   * DT_DOUBLE.
-   * </pre>
-   *
-   * <code>repeated double double_val = 6 [packed = true];</code>
-   */
-  public int getDoubleValCount() {
-    return doubleVal_.size();
-  }
-  /**
-   * <pre>
-   * DT_DOUBLE.
-   * </pre>
-   *
-   * <code>repeated double double_val = 6 [packed = true];</code>
-   */
-  public double getDoubleVal(int index) {
-    return doubleVal_.getDouble(index);
-  }
-  private int doubleValMemoizedSerializedSize = -1;
-
-  public static final int INT_VAL_FIELD_NUMBER = 7;
-  private com.google.protobuf.Internal.IntList intVal_;
-  /**
-   * <pre>
-   * DT_INT32, DT_INT16, DT_INT8, DT_UINT8.
-   * </pre>
-   *
-   * <code>repeated int32 int_val = 7 [packed = true];</code>
-   */
-  public java.util.List<java.lang.Integer>
-      getIntValList() {
-    return intVal_;
-  }
-  /**
-   * <pre>
-   * DT_INT32, DT_INT16, DT_INT8, DT_UINT8.
-   * </pre>
-   *
-   * <code>repeated int32 int_val = 7 [packed = true];</code>
-   */
-  public int getIntValCount() {
-    return intVal_.size();
-  }
-  /**
-   * <pre>
-   * DT_INT32, DT_INT16, DT_INT8, DT_UINT8.
-   * </pre>
-   *
-   * <code>repeated int32 int_val = 7 [packed = true];</code>
-   */
-  public int getIntVal(int index) {
-    return intVal_.getInt(index);
-  }
-  private int intValMemoizedSerializedSize = -1;
-
-  public static final int STRING_VAL_FIELD_NUMBER = 8;
-  private java.util.List<com.google.protobuf.ByteString> stringVal_;
-  /**
-   * <pre>
-   * DT_STRING
-   * </pre>
-   *
-   * <code>repeated bytes string_val = 8;</code>
-   */
-  public java.util.List<com.google.protobuf.ByteString>
-      getStringValList() {
-    return stringVal_;
-  }
-  /**
-   * <pre>
-   * DT_STRING
-   * </pre>
-   *
-   * <code>repeated bytes string_val = 8;</code>
-   */
-  public int getStringValCount() {
-    return stringVal_.size();
-  }
-  /**
-   * <pre>
-   * DT_STRING
-   * </pre>
-   *
-   * <code>repeated bytes string_val = 8;</code>
-   */
-  public com.google.protobuf.ByteString getStringVal(int index) {
-    return stringVal_.get(index);
-  }
-
-  public static final int SCOMPLEX_VAL_FIELD_NUMBER = 9;
-  private com.google.protobuf.Internal.FloatList scomplexVal_;
-  /**
-   * <pre>
-   * DT_COMPLEX64. scomplex_val(2*i) and scomplex_val(2*i+1) are real
-   * and imaginary parts of i-th single precision complex.
-   * </pre>
-   *
-   * <code>repeated float scomplex_val = 9 [packed = true];</code>
-   */
-  public java.util.List<java.lang.Float>
-      getScomplexValList() {
-    return scomplexVal_;
-  }
-  /**
-   * <pre>
-   * DT_COMPLEX64. scomplex_val(2*i) and scomplex_val(2*i+1) are real
-   * and imaginary parts of i-th single precision complex.
-   * </pre>
-   *
-   * <code>repeated float scomplex_val = 9 [packed = true];</code>
-   */
-  public int getScomplexValCount() {
-    return scomplexVal_.size();
-  }
-  /**
-   * <pre>
-   * DT_COMPLEX64. scomplex_val(2*i) and scomplex_val(2*i+1) are real
-   * and imaginary parts of i-th single precision complex.
-   * </pre>
-   *
-   * <code>repeated float scomplex_val = 9 [packed = true];</code>
-   */
-  public float getScomplexVal(int index) {
-    return scomplexVal_.getFloat(index);
-  }
-  private int scomplexValMemoizedSerializedSize = -1;
-
-  public static final int INT64_VAL_FIELD_NUMBER = 10;
-  private com.google.protobuf.Internal.LongList int64Val_;
-  /**
-   * <pre>
-   * DT_INT64
-   * </pre>
-   *
-   * <code>repeated int64 int64_val = 10 [packed = true];</code>
-   */
-  public java.util.List<java.lang.Long>
-      getInt64ValList() {
-    return int64Val_;
-  }
-  /**
-   * <pre>
-   * DT_INT64
-   * </pre>
-   *
-   * <code>repeated int64 int64_val = 10 [packed = true];</code>
-   */
-  public int getInt64ValCount() {
-    return int64Val_.size();
-  }
-  /**
-   * <pre>
-   * DT_INT64
-   * </pre>
-   *
-   * <code>repeated int64 int64_val = 10 [packed = true];</code>
-   */
-  public long getInt64Val(int index) {
-    return int64Val_.getLong(index);
-  }
-  private int int64ValMemoizedSerializedSize = -1;
-
-  public static final int BOOL_VAL_FIELD_NUMBER = 11;
-  private com.google.protobuf.Internal.BooleanList boolVal_;
-  /**
-   * <pre>
-   * DT_BOOL
-   * </pre>
-   *
-   * <code>repeated bool bool_val = 11 [packed = true];</code>
-   */
-  public java.util.List<java.lang.Boolean>
-      getBoolValList() {
-    return boolVal_;
-  }
-  /**
-   * <pre>
-   * DT_BOOL
-   * </pre>
-   *
-   * <code>repeated bool bool_val = 11 [packed = true];</code>
-   */
-  public int getBoolValCount() {
-    return boolVal_.size();
-  }
-  /**
-   * <pre>
-   * DT_BOOL
-   * </pre>
-   *
-   * <code>repeated bool bool_val = 11 [packed = true];</code>
-   */
-  public boolean getBoolVal(int index) {
-    return boolVal_.getBoolean(index);
-  }
-  private int boolValMemoizedSerializedSize = -1;
-
-  public static final int DCOMPLEX_VAL_FIELD_NUMBER = 12;
-  private com.google.protobuf.Internal.DoubleList dcomplexVal_;
-  /**
-   * <pre>
-   * DT_COMPLEX128. dcomplex_val(2*i) and dcomplex_val(2*i+1) are real
-   * and imaginary parts of i-th double precision complex.
-   * </pre>
-   *
-   * <code>repeated double dcomplex_val = 12 [packed = true];</code>
-   */
-  public java.util.List<java.lang.Double>
-      getDcomplexValList() {
-    return dcomplexVal_;
-  }
-  /**
-   * <pre>
-   * DT_COMPLEX128. dcomplex_val(2*i) and dcomplex_val(2*i+1) are real
-   * and imaginary parts of i-th double precision complex.
-   * </pre>
-   *
-   * <code>repeated double dcomplex_val = 12 [packed = true];</code>
-   */
-  public int getDcomplexValCount() {
-    return dcomplexVal_.size();
-  }
-  /**
-   * <pre>
-   * DT_COMPLEX128. dcomplex_val(2*i) and dcomplex_val(2*i+1) are real
-   * and imaginary parts of i-th double precision complex.
-   * </pre>
-   *
-   * <code>repeated double dcomplex_val = 12 [packed = true];</code>
-   */
-  public double getDcomplexVal(int index) {
-    return dcomplexVal_.getDouble(index);
-  }
-  private int dcomplexValMemoizedSerializedSize = -1;
-
-  public static final int RESOURCE_HANDLE_VAL_FIELD_NUMBER = 14;
-  private java.util.List<org.tensorflow.proto.framework.ResourceHandleProto> resourceHandleVal_;
-  /**
-   * <pre>
-   * DT_RESOURCE
-   * </pre>
-   *
-   * <code>repeated .tensorflow.ResourceHandleProto resource_handle_val = 14;</code>
-   */
-  public java.util.List<org.tensorflow.proto.framework.ResourceHandleProto> getResourceHandleValList() {
-    return resourceHandleVal_;
-  }
-  /**
-   * <pre>
-   * DT_RESOURCE
-   * </pre>
-   *
-   * <code>repeated .tensorflow.ResourceHandleProto resource_handle_val = 14;</code>
-   */
-  public java.util.List<? extends org.tensorflow.proto.framework.ResourceHandleProtoOrBuilder> 
-      getResourceHandleValOrBuilderList() {
-    return resourceHandleVal_;
-  }
-  /**
-   * <pre>
-   * DT_RESOURCE
-   * </pre>
-   *
-   * <code>repeated .tensorflow.ResourceHandleProto resource_handle_val = 14;</code>
-   */
-  public int getResourceHandleValCount() {
-    return resourceHandleVal_.size();
-  }
-  /**
-   * <pre>
-   * DT_RESOURCE
-   * </pre>
-   *
-   * <code>repeated .tensorflow.ResourceHandleProto resource_handle_val = 14;</code>
-   */
-  public org.tensorflow.proto.framework.ResourceHandleProto getResourceHandleVal(int index) {
-    return resourceHandleVal_.get(index);
-  }
-  /**
-   * <pre>
-   * DT_RESOURCE
-   * </pre>
-   *
-   * <code>repeated .tensorflow.ResourceHandleProto resource_handle_val = 14;</code>
-   */
-  public org.tensorflow.proto.framework.ResourceHandleProtoOrBuilder getResourceHandleValOrBuilder(
-      int index) {
-    return resourceHandleVal_.get(index);
-  }
-
-  public static final int VARIANT_VAL_FIELD_NUMBER = 15;
-  private java.util.List<org.tensorflow.proto.framework.VariantTensorDataProto> variantVal_;
-  /**
-   * <pre>
-   * DT_VARIANT
-   * </pre>
-   *
-   * <code>repeated .tensorflow.VariantTensorDataProto variant_val = 15;</code>
-   */
-  public java.util.List<org.tensorflow.proto.framework.VariantTensorDataProto> getVariantValList() {
-    return variantVal_;
-  }
-  /**
-   * <pre>
-   * DT_VARIANT
-   * </pre>
-   *
-   * <code>repeated .tensorflow.VariantTensorDataProto variant_val = 15;</code>
-   */
-  public java.util.List<? extends org.tensorflow.proto.framework.VariantTensorDataProtoOrBuilder> 
-      getVariantValOrBuilderList() {
-    return variantVal_;
-  }
-  /**
-   * <pre>
-   * DT_VARIANT
-   * </pre>
-   *
-   * <code>repeated .tensorflow.VariantTensorDataProto variant_val = 15;</code>
-   */
-  public int getVariantValCount() {
-    return variantVal_.size();
-  }
-  /**
-   * <pre>
-   * DT_VARIANT
-   * </pre>
-   *
-   * <code>repeated .tensorflow.VariantTensorDataProto variant_val = 15;</code>
-   */
-  public org.tensorflow.proto.framework.VariantTensorDataProto getVariantVal(int index) {
-    return variantVal_.get(index);
-  }
-  /**
-   * <pre>
-   * DT_VARIANT
-   * </pre>
-   *
-   * <code>repeated .tensorflow.VariantTensorDataProto variant_val = 15;</code>
-   */
-  public org.tensorflow.proto.framework.VariantTensorDataProtoOrBuilder getVariantValOrBuilder(
-      int index) {
-    return variantVal_.get(index);
-  }
-
-  public static final int UINT32_VAL_FIELD_NUMBER = 16;
-  private com.google.protobuf.Internal.IntList uint32Val_;
-  /**
-   * <pre>
-   * DT_UINT32
-   * </pre>
-   *
-   * <code>repeated uint32 uint32_val = 16 [packed = true];</code>
-   */
-  public java.util.List<java.lang.Integer>
-      getUint32ValList() {
-    return uint32Val_;
-  }
-  /**
-   * <pre>
-   * DT_UINT32
-   * </pre>
-   *
-   * <code>repeated uint32 uint32_val = 16 [packed = true];</code>
-   */
-  public int getUint32ValCount() {
-    return uint32Val_.size();
-  }
-  /**
-   * <pre>
-   * DT_UINT32
-   * </pre>
-   *
-   * <code>repeated uint32 uint32_val = 16 [packed = true];</code>
-   */
-  public int getUint32Val(int index) {
-    return uint32Val_.getInt(index);
-  }
-  private int uint32ValMemoizedSerializedSize = -1;
-
-  public static final int UINT64_VAL_FIELD_NUMBER = 17;
-  private com.google.protobuf.Internal.LongList uint64Val_;
-  /**
-   * <pre>
-   * DT_UINT64
-   * </pre>
-   *
-   * <code>repeated uint64 uint64_val = 17 [packed = true];</code>
-   */
-  public java.util.List<java.lang.Long>
-      getUint64ValList() {
-    return uint64Val_;
-  }
-  /**
-   * <pre>
-   * DT_UINT64
-   * </pre>
-   *
-   * <code>repeated uint64 uint64_val = 17 [packed = true];</code>
-   */
-  public int getUint64ValCount() {
-    return uint64Val_.size();
-  }
-  /**
-   * <pre>
-   * DT_UINT64
-   * </pre>
-   *
-   * <code>repeated uint64 uint64_val = 17 [packed = true];</code>
-   */
-  public long getUint64Val(int index) {
-    return uint64Val_.getLong(index);
-  }
-  private int uint64ValMemoizedSerializedSize = -1;
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    getSerializedSize();
-    if (dtype_ != org.tensorflow.proto.framework.DataType.DT_INVALID.getNumber()) {
-      output.writeEnum(1, dtype_);
-    }
-    if (tensorShape_ != null) {
-      output.writeMessage(2, getTensorShape());
-    }
-    if (versionNumber_ != 0) {
-      output.writeInt32(3, versionNumber_);
-    }
-    if (!tensorContent_.isEmpty()) {
-      output.writeBytes(4, tensorContent_);
-    }
-    if (getFloatValList().size() > 0) {
-      output.writeUInt32NoTag(42);
-      output.writeUInt32NoTag(floatValMemoizedSerializedSize);
-    }
-    for (int i = 0; i < floatVal_.size(); i++) {
-      output.writeFloatNoTag(floatVal_.getFloat(i));
-    }
-    if (getDoubleValList().size() > 0) {
-      output.writeUInt32NoTag(50);
-      output.writeUInt32NoTag(doubleValMemoizedSerializedSize);
-    }
-    for (int i = 0; i < doubleVal_.size(); i++) {
-      output.writeDoubleNoTag(doubleVal_.getDouble(i));
-    }
-    if (getIntValList().size() > 0) {
-      output.writeUInt32NoTag(58);
-      output.writeUInt32NoTag(intValMemoizedSerializedSize);
-    }
-    for (int i = 0; i < intVal_.size(); i++) {
-      output.writeInt32NoTag(intVal_.getInt(i));
-    }
-    for (int i = 0; i < stringVal_.size(); i++) {
-      output.writeBytes(8, stringVal_.get(i));
-    }
-    if (getScomplexValList().size() > 0) {
-      output.writeUInt32NoTag(74);
-      output.writeUInt32NoTag(scomplexValMemoizedSerializedSize);
-    }
-    for (int i = 0; i < scomplexVal_.size(); i++) {
-      output.writeFloatNoTag(scomplexVal_.getFloat(i));
-    }
-    if (getInt64ValList().size() > 0) {
-      output.writeUInt32NoTag(82);
-      output.writeUInt32NoTag(int64ValMemoizedSerializedSize);
-    }
-    for (int i = 0; i < int64Val_.size(); i++) {
-      output.writeInt64NoTag(int64Val_.getLong(i));
-    }
-    if (getBoolValList().size() > 0) {
-      output.writeUInt32NoTag(90);
-      output.writeUInt32NoTag(boolValMemoizedSerializedSize);
-    }
-    for (int i = 0; i < boolVal_.size(); i++) {
-      output.writeBoolNoTag(boolVal_.getBoolean(i));
-    }
-    if (getDcomplexValList().size() > 0) {
-      output.writeUInt32NoTag(98);
-      output.writeUInt32NoTag(dcomplexValMemoizedSerializedSize);
-    }
-    for (int i = 0; i < dcomplexVal_.size(); i++) {
-      output.writeDoubleNoTag(dcomplexVal_.getDouble(i));
-    }
-    if (getHalfValList().size() > 0) {
-      output.writeUInt32NoTag(106);
-      output.writeUInt32NoTag(halfValMemoizedSerializedSize);
-    }
-    for (int i = 0; i < halfVal_.size(); i++) {
-      output.writeInt32NoTag(halfVal_.getInt(i));
-    }
-    for (int i = 0; i < resourceHandleVal_.size(); i++) {
-      output.writeMessage(14, resourceHandleVal_.get(i));
-    }
-    for (int i = 0; i < variantVal_.size(); i++) {
-      output.writeMessage(15, variantVal_.get(i));
-    }
-    if (getUint32ValList().size() > 0) {
-      output.writeUInt32NoTag(130);
-      output.writeUInt32NoTag(uint32ValMemoizedSerializedSize);
-    }
-    for (int i = 0; i < uint32Val_.size(); i++) {
-      output.writeUInt32NoTag(uint32Val_.getInt(i));
-    }
-    if (getUint64ValList().size() > 0) {
-      output.writeUInt32NoTag(138);
-      output.writeUInt32NoTag(uint64ValMemoizedSerializedSize);
-    }
-    for (int i = 0; i < uint64Val_.size(); i++) {
-      output.writeUInt64NoTag(uint64Val_.getLong(i));
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (dtype_ != org.tensorflow.proto.framework.DataType.DT_INVALID.getNumber()) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeEnumSize(1, dtype_);
-    }
-    if (tensorShape_ != null) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(2, getTensorShape());
-    }
-    if (versionNumber_ != 0) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt32Size(3, versionNumber_);
-    }
-    if (!tensorContent_.isEmpty()) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeBytesSize(4, tensorContent_);
-    }
-    {
-      int dataSize = 0;
-      dataSize = 4 * getFloatValList().size();
-      size += dataSize;
-      if (!getFloatValList().isEmpty()) {
-        size += 1;
-        size += com.google.protobuf.CodedOutputStream
-            .computeInt32SizeNoTag(dataSize);
-      }
-      floatValMemoizedSerializedSize = dataSize;
-    }
-    {
-      int dataSize = 0;
-      dataSize = 8 * getDoubleValList().size();
-      size += dataSize;
-      if (!getDoubleValList().isEmpty()) {
-        size += 1;
-        size += com.google.protobuf.CodedOutputStream
-            .computeInt32SizeNoTag(dataSize);
-      }
-      doubleValMemoizedSerializedSize = dataSize;
-    }
-    {
-      int dataSize = 0;
-      for (int i = 0; i < intVal_.size(); i++) {
-        dataSize += com.google.protobuf.CodedOutputStream
-          .computeInt32SizeNoTag(intVal_.getInt(i));
-      }
-      size += dataSize;
-      if (!getIntValList().isEmpty()) {
-        size += 1;
-        size += com.google.protobuf.CodedOutputStream
-            .computeInt32SizeNoTag(dataSize);
-      }
-      intValMemoizedSerializedSize = dataSize;
-    }
-    {
-      int dataSize = 0;
-      for (int i = 0; i < stringVal_.size(); i++) {
-        dataSize += com.google.protobuf.CodedOutputStream
-          .computeBytesSizeNoTag(stringVal_.get(i));
-      }
-      size += dataSize;
-      size += 1 * getStringValList().size();
-    }
-    {
-      int dataSize = 0;
-      dataSize = 4 * getScomplexValList().size();
-      size += dataSize;
-      if (!getScomplexValList().isEmpty()) {
-        size += 1;
-        size += com.google.protobuf.CodedOutputStream
-            .computeInt32SizeNoTag(dataSize);
-      }
-      scomplexValMemoizedSerializedSize = dataSize;
-    }
-    {
-      int dataSize = 0;
-      for (int i = 0; i < int64Val_.size(); i++) {
-        dataSize += com.google.protobuf.CodedOutputStream
-          .computeInt64SizeNoTag(int64Val_.getLong(i));
-      }
-      size += dataSize;
-      if (!getInt64ValList().isEmpty()) {
-        size += 1;
-        size += com.google.protobuf.CodedOutputStream
-            .computeInt32SizeNoTag(dataSize);
-      }
-      int64ValMemoizedSerializedSize = dataSize;
-    }
-    {
-      int dataSize = 0;
-      dataSize = 1 * getBoolValList().size();
-      size += dataSize;
-      if (!getBoolValList().isEmpty()) {
-        size += 1;
-        size += com.google.protobuf.CodedOutputStream
-            .computeInt32SizeNoTag(dataSize);
-      }
-      boolValMemoizedSerializedSize = dataSize;
-    }
-    {
-      int dataSize = 0;
-      dataSize = 8 * getDcomplexValList().size();
-      size += dataSize;
-      if (!getDcomplexValList().isEmpty()) {
-        size += 1;
-        size += com.google.protobuf.CodedOutputStream
-            .computeInt32SizeNoTag(dataSize);
-      }
-      dcomplexValMemoizedSerializedSize = dataSize;
-    }
-    {
-      int dataSize = 0;
-      for (int i = 0; i < halfVal_.size(); i++) {
-        dataSize += com.google.protobuf.CodedOutputStream
-          .computeInt32SizeNoTag(halfVal_.getInt(i));
-      }
-      size += dataSize;
-      if (!getHalfValList().isEmpty()) {
-        size += 1;
-        size += com.google.protobuf.CodedOutputStream
-            .computeInt32SizeNoTag(dataSize);
-      }
-      halfValMemoizedSerializedSize = dataSize;
-    }
-    for (int i = 0; i < resourceHandleVal_.size(); i++) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(14, resourceHandleVal_.get(i));
-    }
-    for (int i = 0; i < variantVal_.size(); i++) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(15, variantVal_.get(i));
-    }
-    {
-      int dataSize = 0;
-      for (int i = 0; i < uint32Val_.size(); i++) {
-        dataSize += com.google.protobuf.CodedOutputStream
-          .computeUInt32SizeNoTag(uint32Val_.getInt(i));
-      }
-      size += dataSize;
-      if (!getUint32ValList().isEmpty()) {
-        size += 2;
-        size += com.google.protobuf.CodedOutputStream
-            .computeInt32SizeNoTag(dataSize);
-      }
-      uint32ValMemoizedSerializedSize = dataSize;
-    }
-    {
-      int dataSize = 0;
-      for (int i = 0; i < uint64Val_.size(); i++) {
-        dataSize += com.google.protobuf.CodedOutputStream
-          .computeUInt64SizeNoTag(uint64Val_.getLong(i));
-      }
-      size += dataSize;
-      if (!getUint64ValList().isEmpty()) {
-        size += 2;
-        size += com.google.protobuf.CodedOutputStream
-            .computeInt32SizeNoTag(dataSize);
-      }
-      uint64ValMemoizedSerializedSize = dataSize;
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.framework.TensorProto)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.framework.TensorProto other = (org.tensorflow.proto.framework.TensorProto) obj;
-
-    if (dtype_ != other.dtype_) return false;
-    if (hasTensorShape() != other.hasTensorShape()) return false;
-    if (hasTensorShape()) {
-      if (!getTensorShape()
-          .equals(other.getTensorShape())) return false;
-    }
-    if (getVersionNumber()
-        != other.getVersionNumber()) return false;
-    if (!getTensorContent()
-        .equals(other.getTensorContent())) return false;
-    if (!getHalfValList()
-        .equals(other.getHalfValList())) return false;
-    if (!getFloatValList()
-        .equals(other.getFloatValList())) return false;
-    if (!getDoubleValList()
-        .equals(other.getDoubleValList())) return false;
-    if (!getIntValList()
-        .equals(other.getIntValList())) return false;
-    if (!getStringValList()
-        .equals(other.getStringValList())) return false;
-    if (!getScomplexValList()
-        .equals(other.getScomplexValList())) return false;
-    if (!getInt64ValList()
-        .equals(other.getInt64ValList())) return false;
-    if (!getBoolValList()
-        .equals(other.getBoolValList())) return false;
-    if (!getDcomplexValList()
-        .equals(other.getDcomplexValList())) return false;
-    if (!getResourceHandleValList()
-        .equals(other.getResourceHandleValList())) return false;
-    if (!getVariantValList()
-        .equals(other.getVariantValList())) return false;
-    if (!getUint32ValList()
-        .equals(other.getUint32ValList())) return false;
-    if (!getUint64ValList()
-        .equals(other.getUint64ValList())) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    hash = (37 * hash) + DTYPE_FIELD_NUMBER;
-    hash = (53 * hash) + dtype_;
-    if (hasTensorShape()) {
-      hash = (37 * hash) + TENSOR_SHAPE_FIELD_NUMBER;
-      hash = (53 * hash) + getTensorShape().hashCode();
-    }
-    hash = (37 * hash) + VERSION_NUMBER_FIELD_NUMBER;
-    hash = (53 * hash) + getVersionNumber();
-    hash = (37 * hash) + TENSOR_CONTENT_FIELD_NUMBER;
-    hash = (53 * hash) + getTensorContent().hashCode();
-    if (getHalfValCount() > 0) {
-      hash = (37 * hash) + HALF_VAL_FIELD_NUMBER;
-      hash = (53 * hash) + getHalfValList().hashCode();
-    }
-    if (getFloatValCount() > 0) {
-      hash = (37 * hash) + FLOAT_VAL_FIELD_NUMBER;
-      hash = (53 * hash) + getFloatValList().hashCode();
-    }
-    if (getDoubleValCount() > 0) {
-      hash = (37 * hash) + DOUBLE_VAL_FIELD_NUMBER;
-      hash = (53 * hash) + getDoubleValList().hashCode();
-    }
-    if (getIntValCount() > 0) {
-      hash = (37 * hash) + INT_VAL_FIELD_NUMBER;
-      hash = (53 * hash) + getIntValList().hashCode();
-    }
-    if (getStringValCount() > 0) {
-      hash = (37 * hash) + STRING_VAL_FIELD_NUMBER;
-      hash = (53 * hash) + getStringValList().hashCode();
-    }
-    if (getScomplexValCount() > 0) {
-      hash = (37 * hash) + SCOMPLEX_VAL_FIELD_NUMBER;
-      hash = (53 * hash) + getScomplexValList().hashCode();
-    }
-    if (getInt64ValCount() > 0) {
-      hash = (37 * hash) + INT64_VAL_FIELD_NUMBER;
-      hash = (53 * hash) + getInt64ValList().hashCode();
-    }
-    if (getBoolValCount() > 0) {
-      hash = (37 * hash) + BOOL_VAL_FIELD_NUMBER;
-      hash = (53 * hash) + getBoolValList().hashCode();
-    }
-    if (getDcomplexValCount() > 0) {
-      hash = (37 * hash) + DCOMPLEX_VAL_FIELD_NUMBER;
-      hash = (53 * hash) + getDcomplexValList().hashCode();
-    }
-    if (getResourceHandleValCount() > 0) {
-      hash = (37 * hash) + RESOURCE_HANDLE_VAL_FIELD_NUMBER;
-      hash = (53 * hash) + getResourceHandleValList().hashCode();
-    }
-    if (getVariantValCount() > 0) {
-      hash = (37 * hash) + VARIANT_VAL_FIELD_NUMBER;
-      hash = (53 * hash) + getVariantValList().hashCode();
-    }
-    if (getUint32ValCount() > 0) {
-      hash = (37 * hash) + UINT32_VAL_FIELD_NUMBER;
-      hash = (53 * hash) + getUint32ValList().hashCode();
-    }
-    if (getUint64ValCount() > 0) {
-      hash = (37 * hash) + UINT64_VAL_FIELD_NUMBER;
-      hash = (53 * hash) + getUint64ValList().hashCode();
-    }
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.framework.TensorProto parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.TensorProto parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.TensorProto parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.TensorProto parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.TensorProto parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.TensorProto parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.TensorProto parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.TensorProto parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.TensorProto parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.TensorProto parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.TensorProto parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.TensorProto parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.framework.TensorProto prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * <pre>
-   * Protocol buffer representing a tensor.
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.TensorProto}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.TensorProto)
-      org.tensorflow.proto.framework.TensorProtoOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.TensorProtos.internal_static_tensorflow_TensorProto_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.TensorProtos.internal_static_tensorflow_TensorProto_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.TensorProto.class, org.tensorflow.proto.framework.TensorProto.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.framework.TensorProto.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-        getResourceHandleValFieldBuilder();
-        getVariantValFieldBuilder();
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      dtype_ = 0;
-
-      if (tensorShapeBuilder_ == null) {
-        tensorShape_ = null;
-      } else {
-        tensorShape_ = null;
-        tensorShapeBuilder_ = null;
-      }
-      versionNumber_ = 0;
-
-      tensorContent_ = com.google.protobuf.ByteString.EMPTY;
-
-      halfVal_ = emptyIntList();
-      bitField0_ = (bitField0_ & ~0x00000001);
-      floatVal_ = emptyFloatList();
-      bitField0_ = (bitField0_ & ~0x00000002);
-      doubleVal_ = emptyDoubleList();
-      bitField0_ = (bitField0_ & ~0x00000004);
-      intVal_ = emptyIntList();
-      bitField0_ = (bitField0_ & ~0x00000008);
-      stringVal_ = java.util.Collections.emptyList();
-      bitField0_ = (bitField0_ & ~0x00000010);
-      scomplexVal_ = emptyFloatList();
-      bitField0_ = (bitField0_ & ~0x00000020);
-      int64Val_ = emptyLongList();
-      bitField0_ = (bitField0_ & ~0x00000040);
-      boolVal_ = emptyBooleanList();
-      bitField0_ = (bitField0_ & ~0x00000080);
-      dcomplexVal_ = emptyDoubleList();
-      bitField0_ = (bitField0_ & ~0x00000100);
-      if (resourceHandleValBuilder_ == null) {
-        resourceHandleVal_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000200);
-      } else {
-        resourceHandleValBuilder_.clear();
-      }
-      if (variantValBuilder_ == null) {
-        variantVal_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000400);
-      } else {
-        variantValBuilder_.clear();
-      }
-      uint32Val_ = emptyIntList();
-      bitField0_ = (bitField0_ & ~0x00000800);
-      uint64Val_ = emptyLongList();
-      bitField0_ = (bitField0_ & ~0x00001000);
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.framework.TensorProtos.internal_static_tensorflow_TensorProto_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.TensorProto getDefaultInstanceForType() {
-      return org.tensorflow.proto.framework.TensorProto.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.TensorProto build() {
-      org.tensorflow.proto.framework.TensorProto result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.TensorProto buildPartial() {
-      org.tensorflow.proto.framework.TensorProto result = new org.tensorflow.proto.framework.TensorProto(this);
-      int from_bitField0_ = bitField0_;
-      result.dtype_ = dtype_;
-      if (tensorShapeBuilder_ == null) {
-        result.tensorShape_ = tensorShape_;
-      } else {
-        result.tensorShape_ = tensorShapeBuilder_.build();
-      }
-      result.versionNumber_ = versionNumber_;
-      result.tensorContent_ = tensorContent_;
-      if (((bitField0_ & 0x00000001) != 0)) {
-        halfVal_.makeImmutable();
-        bitField0_ = (bitField0_ & ~0x00000001);
-      }
-      result.halfVal_ = halfVal_;
-      if (((bitField0_ & 0x00000002) != 0)) {
-        floatVal_.makeImmutable();
-        bitField0_ = (bitField0_ & ~0x00000002);
-      }
-      result.floatVal_ = floatVal_;
-      if (((bitField0_ & 0x00000004) != 0)) {
-        doubleVal_.makeImmutable();
-        bitField0_ = (bitField0_ & ~0x00000004);
-      }
-      result.doubleVal_ = doubleVal_;
-      if (((bitField0_ & 0x00000008) != 0)) {
-        intVal_.makeImmutable();
-        bitField0_ = (bitField0_ & ~0x00000008);
-      }
-      result.intVal_ = intVal_;
-      if (((bitField0_ & 0x00000010) != 0)) {
-        stringVal_ = java.util.Collections.unmodifiableList(stringVal_);
-        bitField0_ = (bitField0_ & ~0x00000010);
-      }
-      result.stringVal_ = stringVal_;
-      if (((bitField0_ & 0x00000020) != 0)) {
-        scomplexVal_.makeImmutable();
-        bitField0_ = (bitField0_ & ~0x00000020);
-      }
-      result.scomplexVal_ = scomplexVal_;
-      if (((bitField0_ & 0x00000040) != 0)) {
-        int64Val_.makeImmutable();
-        bitField0_ = (bitField0_ & ~0x00000040);
-      }
-      result.int64Val_ = int64Val_;
-      if (((bitField0_ & 0x00000080) != 0)) {
-        boolVal_.makeImmutable();
-        bitField0_ = (bitField0_ & ~0x00000080);
-      }
-      result.boolVal_ = boolVal_;
-      if (((bitField0_ & 0x00000100) != 0)) {
-        dcomplexVal_.makeImmutable();
-        bitField0_ = (bitField0_ & ~0x00000100);
-      }
-      result.dcomplexVal_ = dcomplexVal_;
-      if (resourceHandleValBuilder_ == null) {
-        if (((bitField0_ & 0x00000200) != 0)) {
-          resourceHandleVal_ = java.util.Collections.unmodifiableList(resourceHandleVal_);
-          bitField0_ = (bitField0_ & ~0x00000200);
-        }
-        result.resourceHandleVal_ = resourceHandleVal_;
-      } else {
-        result.resourceHandleVal_ = resourceHandleValBuilder_.build();
-      }
-      if (variantValBuilder_ == null) {
-        if (((bitField0_ & 0x00000400) != 0)) {
-          variantVal_ = java.util.Collections.unmodifiableList(variantVal_);
-          bitField0_ = (bitField0_ & ~0x00000400);
-        }
-        result.variantVal_ = variantVal_;
-      } else {
-        result.variantVal_ = variantValBuilder_.build();
-      }
-      if (((bitField0_ & 0x00000800) != 0)) {
-        uint32Val_.makeImmutable();
-        bitField0_ = (bitField0_ & ~0x00000800);
-      }
-      result.uint32Val_ = uint32Val_;
-      if (((bitField0_ & 0x00001000) != 0)) {
-        uint64Val_.makeImmutable();
-        bitField0_ = (bitField0_ & ~0x00001000);
-      }
-      result.uint64Val_ = uint64Val_;
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.framework.TensorProto) {
-        return mergeFrom((org.tensorflow.proto.framework.TensorProto)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.framework.TensorProto other) {
-      if (other == org.tensorflow.proto.framework.TensorProto.getDefaultInstance()) return this;
-      if (other.dtype_ != 0) {
-        setDtypeValue(other.getDtypeValue());
-      }
-      if (other.hasTensorShape()) {
-        mergeTensorShape(other.getTensorShape());
-      }
-      if (other.getVersionNumber() != 0) {
-        setVersionNumber(other.getVersionNumber());
-      }
-      if (other.getTensorContent() != com.google.protobuf.ByteString.EMPTY) {
-        setTensorContent(other.getTensorContent());
-      }
-      if (!other.halfVal_.isEmpty()) {
-        if (halfVal_.isEmpty()) {
-          halfVal_ = other.halfVal_;
-          bitField0_ = (bitField0_ & ~0x00000001);
-        } else {
-          ensureHalfValIsMutable();
-          halfVal_.addAll(other.halfVal_);
-        }
-        onChanged();
-      }
-      if (!other.floatVal_.isEmpty()) {
-        if (floatVal_.isEmpty()) {
-          floatVal_ = other.floatVal_;
-          bitField0_ = (bitField0_ & ~0x00000002);
-        } else {
-          ensureFloatValIsMutable();
-          floatVal_.addAll(other.floatVal_);
-        }
-        onChanged();
-      }
-      if (!other.doubleVal_.isEmpty()) {
-        if (doubleVal_.isEmpty()) {
-          doubleVal_ = other.doubleVal_;
-          bitField0_ = (bitField0_ & ~0x00000004);
-        } else {
-          ensureDoubleValIsMutable();
-          doubleVal_.addAll(other.doubleVal_);
-        }
-        onChanged();
-      }
-      if (!other.intVal_.isEmpty()) {
-        if (intVal_.isEmpty()) {
-          intVal_ = other.intVal_;
-          bitField0_ = (bitField0_ & ~0x00000008);
-        } else {
-          ensureIntValIsMutable();
-          intVal_.addAll(other.intVal_);
-        }
-        onChanged();
-      }
-      if (!other.stringVal_.isEmpty()) {
-        if (stringVal_.isEmpty()) {
-          stringVal_ = other.stringVal_;
-          bitField0_ = (bitField0_ & ~0x00000010);
-        } else {
-          ensureStringValIsMutable();
-          stringVal_.addAll(other.stringVal_);
-        }
-        onChanged();
-      }
-      if (!other.scomplexVal_.isEmpty()) {
-        if (scomplexVal_.isEmpty()) {
-          scomplexVal_ = other.scomplexVal_;
-          bitField0_ = (bitField0_ & ~0x00000020);
-        } else {
-          ensureScomplexValIsMutable();
-          scomplexVal_.addAll(other.scomplexVal_);
-        }
-        onChanged();
-      }
-      if (!other.int64Val_.isEmpty()) {
-        if (int64Val_.isEmpty()) {
-          int64Val_ = other.int64Val_;
-          bitField0_ = (bitField0_ & ~0x00000040);
-        } else {
-          ensureInt64ValIsMutable();
-          int64Val_.addAll(other.int64Val_);
-        }
-        onChanged();
-      }
-      if (!other.boolVal_.isEmpty()) {
-        if (boolVal_.isEmpty()) {
-          boolVal_ = other.boolVal_;
-          bitField0_ = (bitField0_ & ~0x00000080);
-        } else {
-          ensureBoolValIsMutable();
-          boolVal_.addAll(other.boolVal_);
-        }
-        onChanged();
-      }
-      if (!other.dcomplexVal_.isEmpty()) {
-        if (dcomplexVal_.isEmpty()) {
-          dcomplexVal_ = other.dcomplexVal_;
-          bitField0_ = (bitField0_ & ~0x00000100);
-        } else {
-          ensureDcomplexValIsMutable();
-          dcomplexVal_.addAll(other.dcomplexVal_);
-        }
-        onChanged();
-      }
-      if (resourceHandleValBuilder_ == null) {
-        if (!other.resourceHandleVal_.isEmpty()) {
-          if (resourceHandleVal_.isEmpty()) {
-            resourceHandleVal_ = other.resourceHandleVal_;
-            bitField0_ = (bitField0_ & ~0x00000200);
-          } else {
-            ensureResourceHandleValIsMutable();
-            resourceHandleVal_.addAll(other.resourceHandleVal_);
-          }
-          onChanged();
-        }
-      } else {
-        if (!other.resourceHandleVal_.isEmpty()) {
-          if (resourceHandleValBuilder_.isEmpty()) {
-            resourceHandleValBuilder_.dispose();
-            resourceHandleValBuilder_ = null;
-            resourceHandleVal_ = other.resourceHandleVal_;
-            bitField0_ = (bitField0_ & ~0x00000200);
-            resourceHandleValBuilder_ = 
-              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
-                 getResourceHandleValFieldBuilder() : null;
-          } else {
-            resourceHandleValBuilder_.addAllMessages(other.resourceHandleVal_);
-          }
-        }
-      }
-      if (variantValBuilder_ == null) {
-        if (!other.variantVal_.isEmpty()) {
-          if (variantVal_.isEmpty()) {
-            variantVal_ = other.variantVal_;
-            bitField0_ = (bitField0_ & ~0x00000400);
-          } else {
-            ensureVariantValIsMutable();
-            variantVal_.addAll(other.variantVal_);
-          }
-          onChanged();
-        }
-      } else {
-        if (!other.variantVal_.isEmpty()) {
-          if (variantValBuilder_.isEmpty()) {
-            variantValBuilder_.dispose();
-            variantValBuilder_ = null;
-            variantVal_ = other.variantVal_;
-            bitField0_ = (bitField0_ & ~0x00000400);
-            variantValBuilder_ = 
-              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
-                 getVariantValFieldBuilder() : null;
-          } else {
-            variantValBuilder_.addAllMessages(other.variantVal_);
-          }
-        }
-      }
-      if (!other.uint32Val_.isEmpty()) {
-        if (uint32Val_.isEmpty()) {
-          uint32Val_ = other.uint32Val_;
-          bitField0_ = (bitField0_ & ~0x00000800);
-        } else {
-          ensureUint32ValIsMutable();
-          uint32Val_.addAll(other.uint32Val_);
-        }
-        onChanged();
-      }
-      if (!other.uint64Val_.isEmpty()) {
-        if (uint64Val_.isEmpty()) {
-          uint64Val_ = other.uint64Val_;
-          bitField0_ = (bitField0_ & ~0x00001000);
-        } else {
-          ensureUint64ValIsMutable();
-          uint64Val_.addAll(other.uint64Val_);
-        }
-        onChanged();
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.framework.TensorProto parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.framework.TensorProto) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-    private int bitField0_;
-
-    private int dtype_ = 0;
-    /**
-     * <code>.tensorflow.DataType dtype = 1;</code>
-     */
-    public int getDtypeValue() {
-      return dtype_;
-    }
-    /**
-     * <code>.tensorflow.DataType dtype = 1;</code>
-     */
-    public Builder setDtypeValue(int value) {
-      dtype_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>.tensorflow.DataType dtype = 1;</code>
-     */
-    public org.tensorflow.proto.framework.DataType getDtype() {
-      @SuppressWarnings("deprecation")
-      org.tensorflow.proto.framework.DataType result = org.tensorflow.proto.framework.DataType.valueOf(dtype_);
-      return result == null ? org.tensorflow.proto.framework.DataType.UNRECOGNIZED : result;
-    }
-    /**
-     * <code>.tensorflow.DataType dtype = 1;</code>
-     */
-    public Builder setDtype(org.tensorflow.proto.framework.DataType value) {
-      if (value == null) {
-        throw new NullPointerException();
-      }
-      
-      dtype_ = value.getNumber();
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>.tensorflow.DataType dtype = 1;</code>
-     */
-    public Builder clearDtype() {
-      
-      dtype_ = 0;
-      onChanged();
-      return this;
-    }
-
-    private org.tensorflow.proto.framework.TensorShapeProto tensorShape_;
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.TensorShapeProto, org.tensorflow.proto.framework.TensorShapeProto.Builder, org.tensorflow.proto.framework.TensorShapeProtoOrBuilder> tensorShapeBuilder_;
-    /**
-     * <pre>
-     * Shape of the tensor.  TODO(touts): sort out the 0-rank issues.
-     * </pre>
-     *
-     * <code>.tensorflow.TensorShapeProto tensor_shape = 2;</code>
-     */
-    public boolean hasTensorShape() {
-      return tensorShapeBuilder_ != null || tensorShape_ != null;
-    }
-    /**
-     * <pre>
-     * Shape of the tensor.  TODO(touts): sort out the 0-rank issues.
-     * </pre>
-     *
-     * <code>.tensorflow.TensorShapeProto tensor_shape = 2;</code>
-     */
-    public org.tensorflow.proto.framework.TensorShapeProto getTensorShape() {
-      if (tensorShapeBuilder_ == null) {
-        return tensorShape_ == null ? org.tensorflow.proto.framework.TensorShapeProto.getDefaultInstance() : tensorShape_;
-      } else {
-        return tensorShapeBuilder_.getMessage();
-      }
-    }
-    /**
-     * <pre>
-     * Shape of the tensor.  TODO(touts): sort out the 0-rank issues.
-     * </pre>
-     *
-     * <code>.tensorflow.TensorShapeProto tensor_shape = 2;</code>
-     */
-    public Builder setTensorShape(org.tensorflow.proto.framework.TensorShapeProto value) {
-      if (tensorShapeBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        tensorShape_ = value;
-        onChanged();
-      } else {
-        tensorShapeBuilder_.setMessage(value);
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Shape of the tensor.  TODO(touts): sort out the 0-rank issues.
-     * </pre>
-     *
-     * <code>.tensorflow.TensorShapeProto tensor_shape = 2;</code>
-     */
-    public Builder setTensorShape(
-        org.tensorflow.proto.framework.TensorShapeProto.Builder builderForValue) {
-      if (tensorShapeBuilder_ == null) {
-        tensorShape_ = builderForValue.build();
-        onChanged();
-      } else {
-        tensorShapeBuilder_.setMessage(builderForValue.build());
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Shape of the tensor.  TODO(touts): sort out the 0-rank issues.
-     * </pre>
-     *
-     * <code>.tensorflow.TensorShapeProto tensor_shape = 2;</code>
-     */
-    public Builder mergeTensorShape(org.tensorflow.proto.framework.TensorShapeProto value) {
-      if (tensorShapeBuilder_ == null) {
-        if (tensorShape_ != null) {
-          tensorShape_ =
-            org.tensorflow.proto.framework.TensorShapeProto.newBuilder(tensorShape_).mergeFrom(value).buildPartial();
-        } else {
-          tensorShape_ = value;
-        }
-        onChanged();
-      } else {
-        tensorShapeBuilder_.mergeFrom(value);
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Shape of the tensor.  TODO(touts): sort out the 0-rank issues.
-     * </pre>
-     *
-     * <code>.tensorflow.TensorShapeProto tensor_shape = 2;</code>
-     */
-    public Builder clearTensorShape() {
-      if (tensorShapeBuilder_ == null) {
-        tensorShape_ = null;
-        onChanged();
-      } else {
-        tensorShape_ = null;
-        tensorShapeBuilder_ = null;
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Shape of the tensor.  TODO(touts): sort out the 0-rank issues.
-     * </pre>
-     *
-     * <code>.tensorflow.TensorShapeProto tensor_shape = 2;</code>
-     */
-    public org.tensorflow.proto.framework.TensorShapeProto.Builder getTensorShapeBuilder() {
-      
-      onChanged();
-      return getTensorShapeFieldBuilder().getBuilder();
-    }
-    /**
-     * <pre>
-     * Shape of the tensor.  TODO(touts): sort out the 0-rank issues.
-     * </pre>
-     *
-     * <code>.tensorflow.TensorShapeProto tensor_shape = 2;</code>
-     */
-    public org.tensorflow.proto.framework.TensorShapeProtoOrBuilder getTensorShapeOrBuilder() {
-      if (tensorShapeBuilder_ != null) {
-        return tensorShapeBuilder_.getMessageOrBuilder();
-      } else {
-        return tensorShape_ == null ?
-            org.tensorflow.proto.framework.TensorShapeProto.getDefaultInstance() : tensorShape_;
-      }
-    }
-    /**
-     * <pre>
-     * Shape of the tensor.  TODO(touts): sort out the 0-rank issues.
-     * </pre>
-     *
-     * <code>.tensorflow.TensorShapeProto tensor_shape = 2;</code>
-     */
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.TensorShapeProto, org.tensorflow.proto.framework.TensorShapeProto.Builder, org.tensorflow.proto.framework.TensorShapeProtoOrBuilder> 
-        getTensorShapeFieldBuilder() {
-      if (tensorShapeBuilder_ == null) {
-        tensorShapeBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-            org.tensorflow.proto.framework.TensorShapeProto, org.tensorflow.proto.framework.TensorShapeProto.Builder, org.tensorflow.proto.framework.TensorShapeProtoOrBuilder>(
-                getTensorShape(),
-                getParentForChildren(),
-                isClean());
-        tensorShape_ = null;
-      }
-      return tensorShapeBuilder_;
-    }
-
-    private int versionNumber_ ;
-    /**
-     * <pre>
-     * Version number.
-     * In version 0, if the "repeated xxx" representations contain only one
-     * element, that element is repeated to fill the shape.  This makes it easy
-     * to represent a constant Tensor with a single value.
-     * </pre>
-     *
-     * <code>int32 version_number = 3;</code>
-     */
-    public int getVersionNumber() {
-      return versionNumber_;
-    }
-    /**
-     * <pre>
-     * Version number.
-     * In version 0, if the "repeated xxx" representations contain only one
-     * element, that element is repeated to fill the shape.  This makes it easy
-     * to represent a constant Tensor with a single value.
-     * </pre>
-     *
-     * <code>int32 version_number = 3;</code>
-     */
-    public Builder setVersionNumber(int value) {
-      
-      versionNumber_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Version number.
-     * In version 0, if the "repeated xxx" representations contain only one
-     * element, that element is repeated to fill the shape.  This makes it easy
-     * to represent a constant Tensor with a single value.
-     * </pre>
-     *
-     * <code>int32 version_number = 3;</code>
-     */
-    public Builder clearVersionNumber() {
-      
-      versionNumber_ = 0;
-      onChanged();
-      return this;
-    }
-
-    private com.google.protobuf.ByteString tensorContent_ = com.google.protobuf.ByteString.EMPTY;
-    /**
-     * <pre>
-     * Serialized raw tensor content from either Tensor::AsProtoTensorContent or
-     * memcpy in tensorflow::grpc::EncodeTensorToByteBuffer. This representation
-     * can be used for all tensor types. The purpose of this representation is to
-     * reduce serialization overhead during RPC call by avoiding serialization of
-     * many repeated small items.
-     * </pre>
-     *
-     * <code>bytes tensor_content = 4;</code>
-     */
-    public com.google.protobuf.ByteString getTensorContent() {
-      return tensorContent_;
-    }
-    /**
-     * <pre>
-     * Serialized raw tensor content from either Tensor::AsProtoTensorContent or
-     * memcpy in tensorflow::grpc::EncodeTensorToByteBuffer. This representation
-     * can be used for all tensor types. The purpose of this representation is to
-     * reduce serialization overhead during RPC call by avoiding serialization of
-     * many repeated small items.
-     * </pre>
-     *
-     * <code>bytes tensor_content = 4;</code>
-     */
-    public Builder setTensorContent(com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      tensorContent_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Serialized raw tensor content from either Tensor::AsProtoTensorContent or
-     * memcpy in tensorflow::grpc::EncodeTensorToByteBuffer. This representation
-     * can be used for all tensor types. The purpose of this representation is to
-     * reduce serialization overhead during RPC call by avoiding serialization of
-     * many repeated small items.
-     * </pre>
-     *
-     * <code>bytes tensor_content = 4;</code>
-     */
-    public Builder clearTensorContent() {
-      
-      tensorContent_ = getDefaultInstance().getTensorContent();
-      onChanged();
-      return this;
-    }
-
-    private com.google.protobuf.Internal.IntList halfVal_ = emptyIntList();
-    private void ensureHalfValIsMutable() {
-      if (!((bitField0_ & 0x00000001) != 0)) {
-        halfVal_ = mutableCopy(halfVal_);
-        bitField0_ |= 0x00000001;
-       }
-    }
-    /**
-     * <pre>
-     * DT_HALF, DT_BFLOAT16. Note that since protobuf has no int16 type, we'll
-     * have some pointless zero padding for each value here.
-     * </pre>
-     *
-     * <code>repeated int32 half_val = 13 [packed = true];</code>
-     */
-    public java.util.List<java.lang.Integer>
-        getHalfValList() {
-      return ((bitField0_ & 0x00000001) != 0) ?
-               java.util.Collections.unmodifiableList(halfVal_) : halfVal_;
-    }
-    /**
-     * <pre>
-     * DT_HALF, DT_BFLOAT16. Note that since protobuf has no int16 type, we'll
-     * have some pointless zero padding for each value here.
-     * </pre>
-     *
-     * <code>repeated int32 half_val = 13 [packed = true];</code>
-     */
-    public int getHalfValCount() {
-      return halfVal_.size();
-    }
-    /**
-     * <pre>
-     * DT_HALF, DT_BFLOAT16. Note that since protobuf has no int16 type, we'll
-     * have some pointless zero padding for each value here.
-     * </pre>
-     *
-     * <code>repeated int32 half_val = 13 [packed = true];</code>
-     */
-    public int getHalfVal(int index) {
-      return halfVal_.getInt(index);
-    }
-    /**
-     * <pre>
-     * DT_HALF, DT_BFLOAT16. Note that since protobuf has no int16 type, we'll
-     * have some pointless zero padding for each value here.
-     * </pre>
-     *
-     * <code>repeated int32 half_val = 13 [packed = true];</code>
-     */
-    public Builder setHalfVal(
-        int index, int value) {
-      ensureHalfValIsMutable();
-      halfVal_.setInt(index, value);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * DT_HALF, DT_BFLOAT16. Note that since protobuf has no int16 type, we'll
-     * have some pointless zero padding for each value here.
-     * </pre>
-     *
-     * <code>repeated int32 half_val = 13 [packed = true];</code>
-     */
-    public Builder addHalfVal(int value) {
-      ensureHalfValIsMutable();
-      halfVal_.addInt(value);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * DT_HALF, DT_BFLOAT16. Note that since protobuf has no int16 type, we'll
-     * have some pointless zero padding for each value here.
-     * </pre>
-     *
-     * <code>repeated int32 half_val = 13 [packed = true];</code>
-     */
-    public Builder addAllHalfVal(
-        java.lang.Iterable<? extends java.lang.Integer> values) {
-      ensureHalfValIsMutable();
-      com.google.protobuf.AbstractMessageLite.Builder.addAll(
-          values, halfVal_);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * DT_HALF, DT_BFLOAT16. Note that since protobuf has no int16 type, we'll
-     * have some pointless zero padding for each value here.
-     * </pre>
-     *
-     * <code>repeated int32 half_val = 13 [packed = true];</code>
-     */
-    public Builder clearHalfVal() {
-      halfVal_ = emptyIntList();
-      bitField0_ = (bitField0_ & ~0x00000001);
-      onChanged();
-      return this;
-    }
-
-    private com.google.protobuf.Internal.FloatList floatVal_ = emptyFloatList();
-    private void ensureFloatValIsMutable() {
-      if (!((bitField0_ & 0x00000002) != 0)) {
-        floatVal_ = mutableCopy(floatVal_);
-        bitField0_ |= 0x00000002;
-       }
-    }
-    /**
-     * <pre>
-     * DT_FLOAT.
-     * </pre>
-     *
-     * <code>repeated float float_val = 5 [packed = true];</code>
-     */
-    public java.util.List<java.lang.Float>
-        getFloatValList() {
-      return ((bitField0_ & 0x00000002) != 0) ?
-               java.util.Collections.unmodifiableList(floatVal_) : floatVal_;
-    }
-    /**
-     * <pre>
-     * DT_FLOAT.
-     * </pre>
-     *
-     * <code>repeated float float_val = 5 [packed = true];</code>
-     */
-    public int getFloatValCount() {
-      return floatVal_.size();
-    }
-    /**
-     * <pre>
-     * DT_FLOAT.
-     * </pre>
-     *
-     * <code>repeated float float_val = 5 [packed = true];</code>
-     */
-    public float getFloatVal(int index) {
-      return floatVal_.getFloat(index);
-    }
-    /**
-     * <pre>
-     * DT_FLOAT.
-     * </pre>
-     *
-     * <code>repeated float float_val = 5 [packed = true];</code>
-     */
-    public Builder setFloatVal(
-        int index, float value) {
-      ensureFloatValIsMutable();
-      floatVal_.setFloat(index, value);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * DT_FLOAT.
-     * </pre>
-     *
-     * <code>repeated float float_val = 5 [packed = true];</code>
-     */
-    public Builder addFloatVal(float value) {
-      ensureFloatValIsMutable();
-      floatVal_.addFloat(value);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * DT_FLOAT.
-     * </pre>
-     *
-     * <code>repeated float float_val = 5 [packed = true];</code>
-     */
-    public Builder addAllFloatVal(
-        java.lang.Iterable<? extends java.lang.Float> values) {
-      ensureFloatValIsMutable();
-      com.google.protobuf.AbstractMessageLite.Builder.addAll(
-          values, floatVal_);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * DT_FLOAT.
-     * </pre>
-     *
-     * <code>repeated float float_val = 5 [packed = true];</code>
-     */
-    public Builder clearFloatVal() {
-      floatVal_ = emptyFloatList();
-      bitField0_ = (bitField0_ & ~0x00000002);
-      onChanged();
-      return this;
-    }
-
-    private com.google.protobuf.Internal.DoubleList doubleVal_ = emptyDoubleList();
-    private void ensureDoubleValIsMutable() {
-      if (!((bitField0_ & 0x00000004) != 0)) {
-        doubleVal_ = mutableCopy(doubleVal_);
-        bitField0_ |= 0x00000004;
-       }
-    }
-    /**
-     * <pre>
-     * DT_DOUBLE.
-     * </pre>
-     *
-     * <code>repeated double double_val = 6 [packed = true];</code>
-     */
-    public java.util.List<java.lang.Double>
-        getDoubleValList() {
-      return ((bitField0_ & 0x00000004) != 0) ?
-               java.util.Collections.unmodifiableList(doubleVal_) : doubleVal_;
-    }
-    /**
-     * <pre>
-     * DT_DOUBLE.
-     * </pre>
-     *
-     * <code>repeated double double_val = 6 [packed = true];</code>
-     */
-    public int getDoubleValCount() {
-      return doubleVal_.size();
-    }
-    /**
-     * <pre>
-     * DT_DOUBLE.
-     * </pre>
-     *
-     * <code>repeated double double_val = 6 [packed = true];</code>
-     */
-    public double getDoubleVal(int index) {
-      return doubleVal_.getDouble(index);
-    }
-    /**
-     * <pre>
-     * DT_DOUBLE.
-     * </pre>
-     *
-     * <code>repeated double double_val = 6 [packed = true];</code>
-     */
-    public Builder setDoubleVal(
-        int index, double value) {
-      ensureDoubleValIsMutable();
-      doubleVal_.setDouble(index, value);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * DT_DOUBLE.
-     * </pre>
-     *
-     * <code>repeated double double_val = 6 [packed = true];</code>
-     */
-    public Builder addDoubleVal(double value) {
-      ensureDoubleValIsMutable();
-      doubleVal_.addDouble(value);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * DT_DOUBLE.
-     * </pre>
-     *
-     * <code>repeated double double_val = 6 [packed = true];</code>
-     */
-    public Builder addAllDoubleVal(
-        java.lang.Iterable<? extends java.lang.Double> values) {
-      ensureDoubleValIsMutable();
-      com.google.protobuf.AbstractMessageLite.Builder.addAll(
-          values, doubleVal_);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * DT_DOUBLE.
-     * </pre>
-     *
-     * <code>repeated double double_val = 6 [packed = true];</code>
-     */
-    public Builder clearDoubleVal() {
-      doubleVal_ = emptyDoubleList();
-      bitField0_ = (bitField0_ & ~0x00000004);
-      onChanged();
-      return this;
-    }
-
-    private com.google.protobuf.Internal.IntList intVal_ = emptyIntList();
-    private void ensureIntValIsMutable() {
-      if (!((bitField0_ & 0x00000008) != 0)) {
-        intVal_ = mutableCopy(intVal_);
-        bitField0_ |= 0x00000008;
-       }
-    }
-    /**
-     * <pre>
-     * DT_INT32, DT_INT16, DT_INT8, DT_UINT8.
-     * </pre>
-     *
-     * <code>repeated int32 int_val = 7 [packed = true];</code>
-     */
-    public java.util.List<java.lang.Integer>
-        getIntValList() {
-      return ((bitField0_ & 0x00000008) != 0) ?
-               java.util.Collections.unmodifiableList(intVal_) : intVal_;
-    }
-    /**
-     * <pre>
-     * DT_INT32, DT_INT16, DT_INT8, DT_UINT8.
-     * </pre>
-     *
-     * <code>repeated int32 int_val = 7 [packed = true];</code>
-     */
-    public int getIntValCount() {
-      return intVal_.size();
-    }
-    /**
-     * <pre>
-     * DT_INT32, DT_INT16, DT_INT8, DT_UINT8.
-     * </pre>
-     *
-     * <code>repeated int32 int_val = 7 [packed = true];</code>
-     */
-    public int getIntVal(int index) {
-      return intVal_.getInt(index);
-    }
-    /**
-     * <pre>
-     * DT_INT32, DT_INT16, DT_INT8, DT_UINT8.
-     * </pre>
-     *
-     * <code>repeated int32 int_val = 7 [packed = true];</code>
-     */
-    public Builder setIntVal(
-        int index, int value) {
-      ensureIntValIsMutable();
-      intVal_.setInt(index, value);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * DT_INT32, DT_INT16, DT_INT8, DT_UINT8.
-     * </pre>
-     *
-     * <code>repeated int32 int_val = 7 [packed = true];</code>
-     */
-    public Builder addIntVal(int value) {
-      ensureIntValIsMutable();
-      intVal_.addInt(value);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * DT_INT32, DT_INT16, DT_INT8, DT_UINT8.
-     * </pre>
-     *
-     * <code>repeated int32 int_val = 7 [packed = true];</code>
-     */
-    public Builder addAllIntVal(
-        java.lang.Iterable<? extends java.lang.Integer> values) {
-      ensureIntValIsMutable();
-      com.google.protobuf.AbstractMessageLite.Builder.addAll(
-          values, intVal_);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * DT_INT32, DT_INT16, DT_INT8, DT_UINT8.
-     * </pre>
-     *
-     * <code>repeated int32 int_val = 7 [packed = true];</code>
-     */
-    public Builder clearIntVal() {
-      intVal_ = emptyIntList();
-      bitField0_ = (bitField0_ & ~0x00000008);
-      onChanged();
-      return this;
-    }
-
-    private java.util.List<com.google.protobuf.ByteString> stringVal_ = java.util.Collections.emptyList();
-    private void ensureStringValIsMutable() {
-      if (!((bitField0_ & 0x00000010) != 0)) {
-        stringVal_ = new java.util.ArrayList<com.google.protobuf.ByteString>(stringVal_);
-        bitField0_ |= 0x00000010;
-       }
-    }
-    /**
-     * <pre>
-     * DT_STRING
-     * </pre>
-     *
-     * <code>repeated bytes string_val = 8;</code>
-     */
-    public java.util.List<com.google.protobuf.ByteString>
-        getStringValList() {
-      return ((bitField0_ & 0x00000010) != 0) ?
-               java.util.Collections.unmodifiableList(stringVal_) : stringVal_;
-    }
-    /**
-     * <pre>
-     * DT_STRING
-     * </pre>
-     *
-     * <code>repeated bytes string_val = 8;</code>
-     */
-    public int getStringValCount() {
-      return stringVal_.size();
-    }
-    /**
-     * <pre>
-     * DT_STRING
-     * </pre>
-     *
-     * <code>repeated bytes string_val = 8;</code>
-     */
-    public com.google.protobuf.ByteString getStringVal(int index) {
-      return stringVal_.get(index);
-    }
-    /**
-     * <pre>
-     * DT_STRING
-     * </pre>
-     *
-     * <code>repeated bytes string_val = 8;</code>
-     */
-    public Builder setStringVal(
-        int index, com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  ensureStringValIsMutable();
-      stringVal_.set(index, value);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * DT_STRING
-     * </pre>
-     *
-     * <code>repeated bytes string_val = 8;</code>
-     */
-    public Builder addStringVal(com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  ensureStringValIsMutable();
-      stringVal_.add(value);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * DT_STRING
-     * </pre>
-     *
-     * <code>repeated bytes string_val = 8;</code>
-     */
-    public Builder addAllStringVal(
-        java.lang.Iterable<? extends com.google.protobuf.ByteString> values) {
-      ensureStringValIsMutable();
-      com.google.protobuf.AbstractMessageLite.Builder.addAll(
-          values, stringVal_);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * DT_STRING
-     * </pre>
-     *
-     * <code>repeated bytes string_val = 8;</code>
-     */
-    public Builder clearStringVal() {
-      stringVal_ = java.util.Collections.emptyList();
-      bitField0_ = (bitField0_ & ~0x00000010);
-      onChanged();
-      return this;
-    }
-
-    private com.google.protobuf.Internal.FloatList scomplexVal_ = emptyFloatList();
-    private void ensureScomplexValIsMutable() {
-      if (!((bitField0_ & 0x00000020) != 0)) {
-        scomplexVal_ = mutableCopy(scomplexVal_);
-        bitField0_ |= 0x00000020;
-       }
-    }
-    /**
-     * <pre>
-     * DT_COMPLEX64. scomplex_val(2*i) and scomplex_val(2*i+1) are real
-     * and imaginary parts of i-th single precision complex.
-     * </pre>
-     *
-     * <code>repeated float scomplex_val = 9 [packed = true];</code>
-     */
-    public java.util.List<java.lang.Float>
-        getScomplexValList() {
-      return ((bitField0_ & 0x00000020) != 0) ?
-               java.util.Collections.unmodifiableList(scomplexVal_) : scomplexVal_;
-    }
-    /**
-     * <pre>
-     * DT_COMPLEX64. scomplex_val(2*i) and scomplex_val(2*i+1) are real
-     * and imaginary parts of i-th single precision complex.
-     * </pre>
-     *
-     * <code>repeated float scomplex_val = 9 [packed = true];</code>
-     */
-    public int getScomplexValCount() {
-      return scomplexVal_.size();
-    }
-    /**
-     * <pre>
-     * DT_COMPLEX64. scomplex_val(2*i) and scomplex_val(2*i+1) are real
-     * and imaginary parts of i-th single precision complex.
-     * </pre>
-     *
-     * <code>repeated float scomplex_val = 9 [packed = true];</code>
-     */
-    public float getScomplexVal(int index) {
-      return scomplexVal_.getFloat(index);
-    }
-    /**
-     * <pre>
-     * DT_COMPLEX64. scomplex_val(2*i) and scomplex_val(2*i+1) are real
-     * and imaginary parts of i-th single precision complex.
-     * </pre>
-     *
-     * <code>repeated float scomplex_val = 9 [packed = true];</code>
-     */
-    public Builder setScomplexVal(
-        int index, float value) {
-      ensureScomplexValIsMutable();
-      scomplexVal_.setFloat(index, value);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * DT_COMPLEX64. scomplex_val(2*i) and scomplex_val(2*i+1) are real
-     * and imaginary parts of i-th single precision complex.
-     * </pre>
-     *
-     * <code>repeated float scomplex_val = 9 [packed = true];</code>
-     */
-    public Builder addScomplexVal(float value) {
-      ensureScomplexValIsMutable();
-      scomplexVal_.addFloat(value);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * DT_COMPLEX64. scomplex_val(2*i) and scomplex_val(2*i+1) are real
-     * and imaginary parts of i-th single precision complex.
-     * </pre>
-     *
-     * <code>repeated float scomplex_val = 9 [packed = true];</code>
-     */
-    public Builder addAllScomplexVal(
-        java.lang.Iterable<? extends java.lang.Float> values) {
-      ensureScomplexValIsMutable();
-      com.google.protobuf.AbstractMessageLite.Builder.addAll(
-          values, scomplexVal_);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * DT_COMPLEX64. scomplex_val(2*i) and scomplex_val(2*i+1) are real
-     * and imaginary parts of i-th single precision complex.
-     * </pre>
-     *
-     * <code>repeated float scomplex_val = 9 [packed = true];</code>
-     */
-    public Builder clearScomplexVal() {
-      scomplexVal_ = emptyFloatList();
-      bitField0_ = (bitField0_ & ~0x00000020);
-      onChanged();
-      return this;
-    }
-
-    private com.google.protobuf.Internal.LongList int64Val_ = emptyLongList();
-    private void ensureInt64ValIsMutable() {
-      if (!((bitField0_ & 0x00000040) != 0)) {
-        int64Val_ = mutableCopy(int64Val_);
-        bitField0_ |= 0x00000040;
-       }
-    }
-    /**
-     * <pre>
-     * DT_INT64
-     * </pre>
-     *
-     * <code>repeated int64 int64_val = 10 [packed = true];</code>
-     */
-    public java.util.List<java.lang.Long>
-        getInt64ValList() {
-      return ((bitField0_ & 0x00000040) != 0) ?
-               java.util.Collections.unmodifiableList(int64Val_) : int64Val_;
-    }
-    /**
-     * <pre>
-     * DT_INT64
-     * </pre>
-     *
-     * <code>repeated int64 int64_val = 10 [packed = true];</code>
-     */
-    public int getInt64ValCount() {
-      return int64Val_.size();
-    }
-    /**
-     * <pre>
-     * DT_INT64
-     * </pre>
-     *
-     * <code>repeated int64 int64_val = 10 [packed = true];</code>
-     */
-    public long getInt64Val(int index) {
-      return int64Val_.getLong(index);
-    }
-    /**
-     * <pre>
-     * DT_INT64
-     * </pre>
-     *
-     * <code>repeated int64 int64_val = 10 [packed = true];</code>
-     */
-    public Builder setInt64Val(
-        int index, long value) {
-      ensureInt64ValIsMutable();
-      int64Val_.setLong(index, value);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * DT_INT64
-     * </pre>
-     *
-     * <code>repeated int64 int64_val = 10 [packed = true];</code>
-     */
-    public Builder addInt64Val(long value) {
-      ensureInt64ValIsMutable();
-      int64Val_.addLong(value);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * DT_INT64
-     * </pre>
-     *
-     * <code>repeated int64 int64_val = 10 [packed = true];</code>
-     */
-    public Builder addAllInt64Val(
-        java.lang.Iterable<? extends java.lang.Long> values) {
-      ensureInt64ValIsMutable();
-      com.google.protobuf.AbstractMessageLite.Builder.addAll(
-          values, int64Val_);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * DT_INT64
-     * </pre>
-     *
-     * <code>repeated int64 int64_val = 10 [packed = true];</code>
-     */
-    public Builder clearInt64Val() {
-      int64Val_ = emptyLongList();
-      bitField0_ = (bitField0_ & ~0x00000040);
-      onChanged();
-      return this;
-    }
-
-    private com.google.protobuf.Internal.BooleanList boolVal_ = emptyBooleanList();
-    private void ensureBoolValIsMutable() {
-      if (!((bitField0_ & 0x00000080) != 0)) {
-        boolVal_ = mutableCopy(boolVal_);
-        bitField0_ |= 0x00000080;
-       }
-    }
-    /**
-     * <pre>
-     * DT_BOOL
-     * </pre>
-     *
-     * <code>repeated bool bool_val = 11 [packed = true];</code>
-     */
-    public java.util.List<java.lang.Boolean>
-        getBoolValList() {
-      return ((bitField0_ & 0x00000080) != 0) ?
-               java.util.Collections.unmodifiableList(boolVal_) : boolVal_;
-    }
-    /**
-     * <pre>
-     * DT_BOOL
-     * </pre>
-     *
-     * <code>repeated bool bool_val = 11 [packed = true];</code>
-     */
-    public int getBoolValCount() {
-      return boolVal_.size();
-    }
-    /**
-     * <pre>
-     * DT_BOOL
-     * </pre>
-     *
-     * <code>repeated bool bool_val = 11 [packed = true];</code>
-     */
-    public boolean getBoolVal(int index) {
-      return boolVal_.getBoolean(index);
-    }
-    /**
-     * <pre>
-     * DT_BOOL
-     * </pre>
-     *
-     * <code>repeated bool bool_val = 11 [packed = true];</code>
-     */
-    public Builder setBoolVal(
-        int index, boolean value) {
-      ensureBoolValIsMutable();
-      boolVal_.setBoolean(index, value);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * DT_BOOL
-     * </pre>
-     *
-     * <code>repeated bool bool_val = 11 [packed = true];</code>
-     */
-    public Builder addBoolVal(boolean value) {
-      ensureBoolValIsMutable();
-      boolVal_.addBoolean(value);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * DT_BOOL
-     * </pre>
-     *
-     * <code>repeated bool bool_val = 11 [packed = true];</code>
-     */
-    public Builder addAllBoolVal(
-        java.lang.Iterable<? extends java.lang.Boolean> values) {
-      ensureBoolValIsMutable();
-      com.google.protobuf.AbstractMessageLite.Builder.addAll(
-          values, boolVal_);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * DT_BOOL
-     * </pre>
-     *
-     * <code>repeated bool bool_val = 11 [packed = true];</code>
-     */
-    public Builder clearBoolVal() {
-      boolVal_ = emptyBooleanList();
-      bitField0_ = (bitField0_ & ~0x00000080);
-      onChanged();
-      return this;
-    }
-
-    private com.google.protobuf.Internal.DoubleList dcomplexVal_ = emptyDoubleList();
-    private void ensureDcomplexValIsMutable() {
-      if (!((bitField0_ & 0x00000100) != 0)) {
-        dcomplexVal_ = mutableCopy(dcomplexVal_);
-        bitField0_ |= 0x00000100;
-       }
-    }
-    /**
-     * <pre>
-     * DT_COMPLEX128. dcomplex_val(2*i) and dcomplex_val(2*i+1) are real
-     * and imaginary parts of i-th double precision complex.
-     * </pre>
-     *
-     * <code>repeated double dcomplex_val = 12 [packed = true];</code>
-     */
-    public java.util.List<java.lang.Double>
-        getDcomplexValList() {
-      return ((bitField0_ & 0x00000100) != 0) ?
-               java.util.Collections.unmodifiableList(dcomplexVal_) : dcomplexVal_;
-    }
-    /**
-     * <pre>
-     * DT_COMPLEX128. dcomplex_val(2*i) and dcomplex_val(2*i+1) are real
-     * and imaginary parts of i-th double precision complex.
-     * </pre>
-     *
-     * <code>repeated double dcomplex_val = 12 [packed = true];</code>
-     */
-    public int getDcomplexValCount() {
-      return dcomplexVal_.size();
-    }
-    /**
-     * <pre>
-     * DT_COMPLEX128. dcomplex_val(2*i) and dcomplex_val(2*i+1) are real
-     * and imaginary parts of i-th double precision complex.
-     * </pre>
-     *
-     * <code>repeated double dcomplex_val = 12 [packed = true];</code>
-     */
-    public double getDcomplexVal(int index) {
-      return dcomplexVal_.getDouble(index);
-    }
-    /**
-     * <pre>
-     * DT_COMPLEX128. dcomplex_val(2*i) and dcomplex_val(2*i+1) are real
-     * and imaginary parts of i-th double precision complex.
-     * </pre>
-     *
-     * <code>repeated double dcomplex_val = 12 [packed = true];</code>
-     */
-    public Builder setDcomplexVal(
-        int index, double value) {
-      ensureDcomplexValIsMutable();
-      dcomplexVal_.setDouble(index, value);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * DT_COMPLEX128. dcomplex_val(2*i) and dcomplex_val(2*i+1) are real
-     * and imaginary parts of i-th double precision complex.
-     * </pre>
-     *
-     * <code>repeated double dcomplex_val = 12 [packed = true];</code>
-     */
-    public Builder addDcomplexVal(double value) {
-      ensureDcomplexValIsMutable();
-      dcomplexVal_.addDouble(value);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * DT_COMPLEX128. dcomplex_val(2*i) and dcomplex_val(2*i+1) are real
-     * and imaginary parts of i-th double precision complex.
-     * </pre>
-     *
-     * <code>repeated double dcomplex_val = 12 [packed = true];</code>
-     */
-    public Builder addAllDcomplexVal(
-        java.lang.Iterable<? extends java.lang.Double> values) {
-      ensureDcomplexValIsMutable();
-      com.google.protobuf.AbstractMessageLite.Builder.addAll(
-          values, dcomplexVal_);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * DT_COMPLEX128. dcomplex_val(2*i) and dcomplex_val(2*i+1) are real
-     * and imaginary parts of i-th double precision complex.
-     * </pre>
-     *
-     * <code>repeated double dcomplex_val = 12 [packed = true];</code>
-     */
-    public Builder clearDcomplexVal() {
-      dcomplexVal_ = emptyDoubleList();
-      bitField0_ = (bitField0_ & ~0x00000100);
-      onChanged();
-      return this;
-    }
-
-    private java.util.List<org.tensorflow.proto.framework.ResourceHandleProto> resourceHandleVal_ =
-      java.util.Collections.emptyList();
-    private void ensureResourceHandleValIsMutable() {
-      if (!((bitField0_ & 0x00000200) != 0)) {
-        resourceHandleVal_ = new java.util.ArrayList<org.tensorflow.proto.framework.ResourceHandleProto>(resourceHandleVal_);
-        bitField0_ |= 0x00000200;
-       }
-    }
-
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.framework.ResourceHandleProto, org.tensorflow.proto.framework.ResourceHandleProto.Builder, org.tensorflow.proto.framework.ResourceHandleProtoOrBuilder> resourceHandleValBuilder_;
-
-    /**
-     * <pre>
-     * DT_RESOURCE
-     * </pre>
-     *
-     * <code>repeated .tensorflow.ResourceHandleProto resource_handle_val = 14;</code>
-     */
-    public java.util.List<org.tensorflow.proto.framework.ResourceHandleProto> getResourceHandleValList() {
-      if (resourceHandleValBuilder_ == null) {
-        return java.util.Collections.unmodifiableList(resourceHandleVal_);
-      } else {
-        return resourceHandleValBuilder_.getMessageList();
-      }
-    }
-    /**
-     * <pre>
-     * DT_RESOURCE
-     * </pre>
-     *
-     * <code>repeated .tensorflow.ResourceHandleProto resource_handle_val = 14;</code>
-     */
-    public int getResourceHandleValCount() {
-      if (resourceHandleValBuilder_ == null) {
-        return resourceHandleVal_.size();
-      } else {
-        return resourceHandleValBuilder_.getCount();
-      }
-    }
-    /**
-     * <pre>
-     * DT_RESOURCE
-     * </pre>
-     *
-     * <code>repeated .tensorflow.ResourceHandleProto resource_handle_val = 14;</code>
-     */
-    public org.tensorflow.proto.framework.ResourceHandleProto getResourceHandleVal(int index) {
-      if (resourceHandleValBuilder_ == null) {
-        return resourceHandleVal_.get(index);
-      } else {
-        return resourceHandleValBuilder_.getMessage(index);
-      }
-    }
-    /**
-     * <pre>
-     * DT_RESOURCE
-     * </pre>
-     *
-     * <code>repeated .tensorflow.ResourceHandleProto resource_handle_val = 14;</code>
-     */
-    public Builder setResourceHandleVal(
-        int index, org.tensorflow.proto.framework.ResourceHandleProto value) {
-      if (resourceHandleValBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureResourceHandleValIsMutable();
-        resourceHandleVal_.set(index, value);
-        onChanged();
-      } else {
-        resourceHandleValBuilder_.setMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * DT_RESOURCE
-     * </pre>
-     *
-     * <code>repeated .tensorflow.ResourceHandleProto resource_handle_val = 14;</code>
-     */
-    public Builder setResourceHandleVal(
-        int index, org.tensorflow.proto.framework.ResourceHandleProto.Builder builderForValue) {
-      if (resourceHandleValBuilder_ == null) {
-        ensureResourceHandleValIsMutable();
-        resourceHandleVal_.set(index, builderForValue.build());
-        onChanged();
-      } else {
-        resourceHandleValBuilder_.setMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * DT_RESOURCE
-     * </pre>
-     *
-     * <code>repeated .tensorflow.ResourceHandleProto resource_handle_val = 14;</code>
-     */
-    public Builder addResourceHandleVal(org.tensorflow.proto.framework.ResourceHandleProto value) {
-      if (resourceHandleValBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureResourceHandleValIsMutable();
-        resourceHandleVal_.add(value);
-        onChanged();
-      } else {
-        resourceHandleValBuilder_.addMessage(value);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * DT_RESOURCE
-     * </pre>
-     *
-     * <code>repeated .tensorflow.ResourceHandleProto resource_handle_val = 14;</code>
-     */
-    public Builder addResourceHandleVal(
-        int index, org.tensorflow.proto.framework.ResourceHandleProto value) {
-      if (resourceHandleValBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureResourceHandleValIsMutable();
-        resourceHandleVal_.add(index, value);
-        onChanged();
-      } else {
-        resourceHandleValBuilder_.addMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * DT_RESOURCE
-     * </pre>
-     *
-     * <code>repeated .tensorflow.ResourceHandleProto resource_handle_val = 14;</code>
-     */
-    public Builder addResourceHandleVal(
-        org.tensorflow.proto.framework.ResourceHandleProto.Builder builderForValue) {
-      if (resourceHandleValBuilder_ == null) {
-        ensureResourceHandleValIsMutable();
-        resourceHandleVal_.add(builderForValue.build());
-        onChanged();
-      } else {
-        resourceHandleValBuilder_.addMessage(builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * DT_RESOURCE
-     * </pre>
-     *
-     * <code>repeated .tensorflow.ResourceHandleProto resource_handle_val = 14;</code>
-     */
-    public Builder addResourceHandleVal(
-        int index, org.tensorflow.proto.framework.ResourceHandleProto.Builder builderForValue) {
-      if (resourceHandleValBuilder_ == null) {
-        ensureResourceHandleValIsMutable();
-        resourceHandleVal_.add(index, builderForValue.build());
-        onChanged();
-      } else {
-        resourceHandleValBuilder_.addMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * DT_RESOURCE
-     * </pre>
-     *
-     * <code>repeated .tensorflow.ResourceHandleProto resource_handle_val = 14;</code>
-     */
-    public Builder addAllResourceHandleVal(
-        java.lang.Iterable<? extends org.tensorflow.proto.framework.ResourceHandleProto> values) {
-      if (resourceHandleValBuilder_ == null) {
-        ensureResourceHandleValIsMutable();
-        com.google.protobuf.AbstractMessageLite.Builder.addAll(
-            values, resourceHandleVal_);
-        onChanged();
-      } else {
-        resourceHandleValBuilder_.addAllMessages(values);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * DT_RESOURCE
-     * </pre>
-     *
-     * <code>repeated .tensorflow.ResourceHandleProto resource_handle_val = 14;</code>
-     */
-    public Builder clearResourceHandleVal() {
-      if (resourceHandleValBuilder_ == null) {
-        resourceHandleVal_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000200);
-        onChanged();
-      } else {
-        resourceHandleValBuilder_.clear();
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * DT_RESOURCE
-     * </pre>
-     *
-     * <code>repeated .tensorflow.ResourceHandleProto resource_handle_val = 14;</code>
-     */
-    public Builder removeResourceHandleVal(int index) {
-      if (resourceHandleValBuilder_ == null) {
-        ensureResourceHandleValIsMutable();
-        resourceHandleVal_.remove(index);
-        onChanged();
-      } else {
-        resourceHandleValBuilder_.remove(index);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * DT_RESOURCE
-     * </pre>
-     *
-     * <code>repeated .tensorflow.ResourceHandleProto resource_handle_val = 14;</code>
-     */
-    public org.tensorflow.proto.framework.ResourceHandleProto.Builder getResourceHandleValBuilder(
-        int index) {
-      return getResourceHandleValFieldBuilder().getBuilder(index);
-    }
-    /**
-     * <pre>
-     * DT_RESOURCE
-     * </pre>
-     *
-     * <code>repeated .tensorflow.ResourceHandleProto resource_handle_val = 14;</code>
-     */
-    public org.tensorflow.proto.framework.ResourceHandleProtoOrBuilder getResourceHandleValOrBuilder(
-        int index) {
-      if (resourceHandleValBuilder_ == null) {
-        return resourceHandleVal_.get(index);  } else {
-        return resourceHandleValBuilder_.getMessageOrBuilder(index);
-      }
-    }
-    /**
-     * <pre>
-     * DT_RESOURCE
-     * </pre>
-     *
-     * <code>repeated .tensorflow.ResourceHandleProto resource_handle_val = 14;</code>
-     */
-    public java.util.List<? extends org.tensorflow.proto.framework.ResourceHandleProtoOrBuilder> 
-         getResourceHandleValOrBuilderList() {
-      if (resourceHandleValBuilder_ != null) {
-        return resourceHandleValBuilder_.getMessageOrBuilderList();
-      } else {
-        return java.util.Collections.unmodifiableList(resourceHandleVal_);
-      }
-    }
-    /**
-     * <pre>
-     * DT_RESOURCE
-     * </pre>
-     *
-     * <code>repeated .tensorflow.ResourceHandleProto resource_handle_val = 14;</code>
-     */
-    public org.tensorflow.proto.framework.ResourceHandleProto.Builder addResourceHandleValBuilder() {
-      return getResourceHandleValFieldBuilder().addBuilder(
-          org.tensorflow.proto.framework.ResourceHandleProto.getDefaultInstance());
-    }
-    /**
-     * <pre>
-     * DT_RESOURCE
-     * </pre>
-     *
-     * <code>repeated .tensorflow.ResourceHandleProto resource_handle_val = 14;</code>
-     */
-    public org.tensorflow.proto.framework.ResourceHandleProto.Builder addResourceHandleValBuilder(
-        int index) {
-      return getResourceHandleValFieldBuilder().addBuilder(
-          index, org.tensorflow.proto.framework.ResourceHandleProto.getDefaultInstance());
-    }
-    /**
-     * <pre>
-     * DT_RESOURCE
-     * </pre>
-     *
-     * <code>repeated .tensorflow.ResourceHandleProto resource_handle_val = 14;</code>
-     */
-    public java.util.List<org.tensorflow.proto.framework.ResourceHandleProto.Builder> 
-         getResourceHandleValBuilderList() {
-      return getResourceHandleValFieldBuilder().getBuilderList();
-    }
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.framework.ResourceHandleProto, org.tensorflow.proto.framework.ResourceHandleProto.Builder, org.tensorflow.proto.framework.ResourceHandleProtoOrBuilder> 
-        getResourceHandleValFieldBuilder() {
-      if (resourceHandleValBuilder_ == null) {
-        resourceHandleValBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
-            org.tensorflow.proto.framework.ResourceHandleProto, org.tensorflow.proto.framework.ResourceHandleProto.Builder, org.tensorflow.proto.framework.ResourceHandleProtoOrBuilder>(
-                resourceHandleVal_,
-                ((bitField0_ & 0x00000200) != 0),
-                getParentForChildren(),
-                isClean());
-        resourceHandleVal_ = null;
-      }
-      return resourceHandleValBuilder_;
-    }
-
-    private java.util.List<org.tensorflow.proto.framework.VariantTensorDataProto> variantVal_ =
-      java.util.Collections.emptyList();
-    private void ensureVariantValIsMutable() {
-      if (!((bitField0_ & 0x00000400) != 0)) {
-        variantVal_ = new java.util.ArrayList<org.tensorflow.proto.framework.VariantTensorDataProto>(variantVal_);
-        bitField0_ |= 0x00000400;
-       }
-    }
-
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.framework.VariantTensorDataProto, org.tensorflow.proto.framework.VariantTensorDataProto.Builder, org.tensorflow.proto.framework.VariantTensorDataProtoOrBuilder> variantValBuilder_;
-
-    /**
-     * <pre>
-     * DT_VARIANT
-     * </pre>
-     *
-     * <code>repeated .tensorflow.VariantTensorDataProto variant_val = 15;</code>
-     */
-    public java.util.List<org.tensorflow.proto.framework.VariantTensorDataProto> getVariantValList() {
-      if (variantValBuilder_ == null) {
-        return java.util.Collections.unmodifiableList(variantVal_);
-      } else {
-        return variantValBuilder_.getMessageList();
-      }
-    }
-    /**
-     * <pre>
-     * DT_VARIANT
-     * </pre>
-     *
-     * <code>repeated .tensorflow.VariantTensorDataProto variant_val = 15;</code>
-     */
-    public int getVariantValCount() {
-      if (variantValBuilder_ == null) {
-        return variantVal_.size();
-      } else {
-        return variantValBuilder_.getCount();
-      }
-    }
-    /**
-     * <pre>
-     * DT_VARIANT
-     * </pre>
-     *
-     * <code>repeated .tensorflow.VariantTensorDataProto variant_val = 15;</code>
-     */
-    public org.tensorflow.proto.framework.VariantTensorDataProto getVariantVal(int index) {
-      if (variantValBuilder_ == null) {
-        return variantVal_.get(index);
-      } else {
-        return variantValBuilder_.getMessage(index);
-      }
-    }
-    /**
-     * <pre>
-     * DT_VARIANT
-     * </pre>
-     *
-     * <code>repeated .tensorflow.VariantTensorDataProto variant_val = 15;</code>
-     */
-    public Builder setVariantVal(
-        int index, org.tensorflow.proto.framework.VariantTensorDataProto value) {
-      if (variantValBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureVariantValIsMutable();
-        variantVal_.set(index, value);
-        onChanged();
-      } else {
-        variantValBuilder_.setMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * DT_VARIANT
-     * </pre>
-     *
-     * <code>repeated .tensorflow.VariantTensorDataProto variant_val = 15;</code>
-     */
-    public Builder setVariantVal(
-        int index, org.tensorflow.proto.framework.VariantTensorDataProto.Builder builderForValue) {
-      if (variantValBuilder_ == null) {
-        ensureVariantValIsMutable();
-        variantVal_.set(index, builderForValue.build());
-        onChanged();
-      } else {
-        variantValBuilder_.setMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * DT_VARIANT
-     * </pre>
-     *
-     * <code>repeated .tensorflow.VariantTensorDataProto variant_val = 15;</code>
-     */
-    public Builder addVariantVal(org.tensorflow.proto.framework.VariantTensorDataProto value) {
-      if (variantValBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureVariantValIsMutable();
-        variantVal_.add(value);
-        onChanged();
-      } else {
-        variantValBuilder_.addMessage(value);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * DT_VARIANT
-     * </pre>
-     *
-     * <code>repeated .tensorflow.VariantTensorDataProto variant_val = 15;</code>
-     */
-    public Builder addVariantVal(
-        int index, org.tensorflow.proto.framework.VariantTensorDataProto value) {
-      if (variantValBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureVariantValIsMutable();
-        variantVal_.add(index, value);
-        onChanged();
-      } else {
-        variantValBuilder_.addMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * DT_VARIANT
-     * </pre>
-     *
-     * <code>repeated .tensorflow.VariantTensorDataProto variant_val = 15;</code>
-     */
-    public Builder addVariantVal(
-        org.tensorflow.proto.framework.VariantTensorDataProto.Builder builderForValue) {
-      if (variantValBuilder_ == null) {
-        ensureVariantValIsMutable();
-        variantVal_.add(builderForValue.build());
-        onChanged();
-      } else {
-        variantValBuilder_.addMessage(builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * DT_VARIANT
-     * </pre>
-     *
-     * <code>repeated .tensorflow.VariantTensorDataProto variant_val = 15;</code>
-     */
-    public Builder addVariantVal(
-        int index, org.tensorflow.proto.framework.VariantTensorDataProto.Builder builderForValue) {
-      if (variantValBuilder_ == null) {
-        ensureVariantValIsMutable();
-        variantVal_.add(index, builderForValue.build());
-        onChanged();
-      } else {
-        variantValBuilder_.addMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * DT_VARIANT
-     * </pre>
-     *
-     * <code>repeated .tensorflow.VariantTensorDataProto variant_val = 15;</code>
-     */
-    public Builder addAllVariantVal(
-        java.lang.Iterable<? extends org.tensorflow.proto.framework.VariantTensorDataProto> values) {
-      if (variantValBuilder_ == null) {
-        ensureVariantValIsMutable();
-        com.google.protobuf.AbstractMessageLite.Builder.addAll(
-            values, variantVal_);
-        onChanged();
-      } else {
-        variantValBuilder_.addAllMessages(values);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * DT_VARIANT
-     * </pre>
-     *
-     * <code>repeated .tensorflow.VariantTensorDataProto variant_val = 15;</code>
-     */
-    public Builder clearVariantVal() {
-      if (variantValBuilder_ == null) {
-        variantVal_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000400);
-        onChanged();
-      } else {
-        variantValBuilder_.clear();
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * DT_VARIANT
-     * </pre>
-     *
-     * <code>repeated .tensorflow.VariantTensorDataProto variant_val = 15;</code>
-     */
-    public Builder removeVariantVal(int index) {
-      if (variantValBuilder_ == null) {
-        ensureVariantValIsMutable();
-        variantVal_.remove(index);
-        onChanged();
-      } else {
-        variantValBuilder_.remove(index);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * DT_VARIANT
-     * </pre>
-     *
-     * <code>repeated .tensorflow.VariantTensorDataProto variant_val = 15;</code>
-     */
-    public org.tensorflow.proto.framework.VariantTensorDataProto.Builder getVariantValBuilder(
-        int index) {
-      return getVariantValFieldBuilder().getBuilder(index);
-    }
-    /**
-     * <pre>
-     * DT_VARIANT
-     * </pre>
-     *
-     * <code>repeated .tensorflow.VariantTensorDataProto variant_val = 15;</code>
-     */
-    public org.tensorflow.proto.framework.VariantTensorDataProtoOrBuilder getVariantValOrBuilder(
-        int index) {
-      if (variantValBuilder_ == null) {
-        return variantVal_.get(index);  } else {
-        return variantValBuilder_.getMessageOrBuilder(index);
-      }
-    }
-    /**
-     * <pre>
-     * DT_VARIANT
-     * </pre>
-     *
-     * <code>repeated .tensorflow.VariantTensorDataProto variant_val = 15;</code>
-     */
-    public java.util.List<? extends org.tensorflow.proto.framework.VariantTensorDataProtoOrBuilder> 
-         getVariantValOrBuilderList() {
-      if (variantValBuilder_ != null) {
-        return variantValBuilder_.getMessageOrBuilderList();
-      } else {
-        return java.util.Collections.unmodifiableList(variantVal_);
-      }
-    }
-    /**
-     * <pre>
-     * DT_VARIANT
-     * </pre>
-     *
-     * <code>repeated .tensorflow.VariantTensorDataProto variant_val = 15;</code>
-     */
-    public org.tensorflow.proto.framework.VariantTensorDataProto.Builder addVariantValBuilder() {
-      return getVariantValFieldBuilder().addBuilder(
-          org.tensorflow.proto.framework.VariantTensorDataProto.getDefaultInstance());
-    }
-    /**
-     * <pre>
-     * DT_VARIANT
-     * </pre>
-     *
-     * <code>repeated .tensorflow.VariantTensorDataProto variant_val = 15;</code>
-     */
-    public org.tensorflow.proto.framework.VariantTensorDataProto.Builder addVariantValBuilder(
-        int index) {
-      return getVariantValFieldBuilder().addBuilder(
-          index, org.tensorflow.proto.framework.VariantTensorDataProto.getDefaultInstance());
-    }
-    /**
-     * <pre>
-     * DT_VARIANT
-     * </pre>
-     *
-     * <code>repeated .tensorflow.VariantTensorDataProto variant_val = 15;</code>
-     */
-    public java.util.List<org.tensorflow.proto.framework.VariantTensorDataProto.Builder> 
-         getVariantValBuilderList() {
-      return getVariantValFieldBuilder().getBuilderList();
-    }
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.framework.VariantTensorDataProto, org.tensorflow.proto.framework.VariantTensorDataProto.Builder, org.tensorflow.proto.framework.VariantTensorDataProtoOrBuilder> 
-        getVariantValFieldBuilder() {
-      if (variantValBuilder_ == null) {
-        variantValBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
-            org.tensorflow.proto.framework.VariantTensorDataProto, org.tensorflow.proto.framework.VariantTensorDataProto.Builder, org.tensorflow.proto.framework.VariantTensorDataProtoOrBuilder>(
-                variantVal_,
-                ((bitField0_ & 0x00000400) != 0),
-                getParentForChildren(),
-                isClean());
-        variantVal_ = null;
-      }
-      return variantValBuilder_;
-    }
-
-    private com.google.protobuf.Internal.IntList uint32Val_ = emptyIntList();
-    private void ensureUint32ValIsMutable() {
-      if (!((bitField0_ & 0x00000800) != 0)) {
-        uint32Val_ = mutableCopy(uint32Val_);
-        bitField0_ |= 0x00000800;
-       }
-    }
-    /**
-     * <pre>
-     * DT_UINT32
-     * </pre>
-     *
-     * <code>repeated uint32 uint32_val = 16 [packed = true];</code>
-     */
-    public java.util.List<java.lang.Integer>
-        getUint32ValList() {
-      return ((bitField0_ & 0x00000800) != 0) ?
-               java.util.Collections.unmodifiableList(uint32Val_) : uint32Val_;
-    }
-    /**
-     * <pre>
-     * DT_UINT32
-     * </pre>
-     *
-     * <code>repeated uint32 uint32_val = 16 [packed = true];</code>
-     */
-    public int getUint32ValCount() {
-      return uint32Val_.size();
-    }
-    /**
-     * <pre>
-     * DT_UINT32
-     * </pre>
-     *
-     * <code>repeated uint32 uint32_val = 16 [packed = true];</code>
-     */
-    public int getUint32Val(int index) {
-      return uint32Val_.getInt(index);
-    }
-    /**
-     * <pre>
-     * DT_UINT32
-     * </pre>
-     *
-     * <code>repeated uint32 uint32_val = 16 [packed = true];</code>
-     */
-    public Builder setUint32Val(
-        int index, int value) {
-      ensureUint32ValIsMutable();
-      uint32Val_.setInt(index, value);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * DT_UINT32
-     * </pre>
-     *
-     * <code>repeated uint32 uint32_val = 16 [packed = true];</code>
-     */
-    public Builder addUint32Val(int value) {
-      ensureUint32ValIsMutable();
-      uint32Val_.addInt(value);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * DT_UINT32
-     * </pre>
-     *
-     * <code>repeated uint32 uint32_val = 16 [packed = true];</code>
-     */
-    public Builder addAllUint32Val(
-        java.lang.Iterable<? extends java.lang.Integer> values) {
-      ensureUint32ValIsMutable();
-      com.google.protobuf.AbstractMessageLite.Builder.addAll(
-          values, uint32Val_);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * DT_UINT32
-     * </pre>
-     *
-     * <code>repeated uint32 uint32_val = 16 [packed = true];</code>
-     */
-    public Builder clearUint32Val() {
-      uint32Val_ = emptyIntList();
-      bitField0_ = (bitField0_ & ~0x00000800);
-      onChanged();
-      return this;
-    }
-
-    private com.google.protobuf.Internal.LongList uint64Val_ = emptyLongList();
-    private void ensureUint64ValIsMutable() {
-      if (!((bitField0_ & 0x00001000) != 0)) {
-        uint64Val_ = mutableCopy(uint64Val_);
-        bitField0_ |= 0x00001000;
-       }
-    }
-    /**
-     * <pre>
-     * DT_UINT64
-     * </pre>
-     *
-     * <code>repeated uint64 uint64_val = 17 [packed = true];</code>
-     */
-    public java.util.List<java.lang.Long>
-        getUint64ValList() {
-      return ((bitField0_ & 0x00001000) != 0) ?
-               java.util.Collections.unmodifiableList(uint64Val_) : uint64Val_;
-    }
-    /**
-     * <pre>
-     * DT_UINT64
-     * </pre>
-     *
-     * <code>repeated uint64 uint64_val = 17 [packed = true];</code>
-     */
-    public int getUint64ValCount() {
-      return uint64Val_.size();
-    }
-    /**
-     * <pre>
-     * DT_UINT64
-     * </pre>
-     *
-     * <code>repeated uint64 uint64_val = 17 [packed = true];</code>
-     */
-    public long getUint64Val(int index) {
-      return uint64Val_.getLong(index);
-    }
-    /**
-     * <pre>
-     * DT_UINT64
-     * </pre>
-     *
-     * <code>repeated uint64 uint64_val = 17 [packed = true];</code>
-     */
-    public Builder setUint64Val(
-        int index, long value) {
-      ensureUint64ValIsMutable();
-      uint64Val_.setLong(index, value);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * DT_UINT64
-     * </pre>
-     *
-     * <code>repeated uint64 uint64_val = 17 [packed = true];</code>
-     */
-    public Builder addUint64Val(long value) {
-      ensureUint64ValIsMutable();
-      uint64Val_.addLong(value);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * DT_UINT64
-     * </pre>
-     *
-     * <code>repeated uint64 uint64_val = 17 [packed = true];</code>
-     */
-    public Builder addAllUint64Val(
-        java.lang.Iterable<? extends java.lang.Long> values) {
-      ensureUint64ValIsMutable();
-      com.google.protobuf.AbstractMessageLite.Builder.addAll(
-          values, uint64Val_);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * DT_UINT64
-     * </pre>
-     *
-     * <code>repeated uint64 uint64_val = 17 [packed = true];</code>
-     */
-    public Builder clearUint64Val() {
-      uint64Val_ = emptyLongList();
-      bitField0_ = (bitField0_ & ~0x00001000);
-      onChanged();
-      return this;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.TensorProto)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.TensorProto)
-  private static final org.tensorflow.proto.framework.TensorProto DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.TensorProto();
-  }
-
-  public static org.tensorflow.proto.framework.TensorProto getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<TensorProto>
-      PARSER = new com.google.protobuf.AbstractParser<TensorProto>() {
-    @java.lang.Override
-    public TensorProto parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new TensorProto(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<TensorProto> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<TensorProto> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.framework.TensorProto getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorProtoOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorProtoOrBuilder.java
deleted file mode 100644
index b5369066023..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorProtoOrBuilder.java
+++ /dev/null
@@ -1,440 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/framework/tensor.proto
-
-package org.tensorflow.proto.framework;
-
-public interface TensorProtoOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.TensorProto)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <code>.tensorflow.DataType dtype = 1;</code>
-   */
-  int getDtypeValue();
-  /**
-   * <code>.tensorflow.DataType dtype = 1;</code>
-   */
-  org.tensorflow.proto.framework.DataType getDtype();
-
-  /**
-   * <pre>
-   * Shape of the tensor.  TODO(touts): sort out the 0-rank issues.
-   * </pre>
-   *
-   * <code>.tensorflow.TensorShapeProto tensor_shape = 2;</code>
-   */
-  boolean hasTensorShape();
-  /**
-   * <pre>
-   * Shape of the tensor.  TODO(touts): sort out the 0-rank issues.
-   * </pre>
-   *
-   * <code>.tensorflow.TensorShapeProto tensor_shape = 2;</code>
-   */
-  org.tensorflow.proto.framework.TensorShapeProto getTensorShape();
-  /**
-   * <pre>
-   * Shape of the tensor.  TODO(touts): sort out the 0-rank issues.
-   * </pre>
-   *
-   * <code>.tensorflow.TensorShapeProto tensor_shape = 2;</code>
-   */
-  org.tensorflow.proto.framework.TensorShapeProtoOrBuilder getTensorShapeOrBuilder();
-
-  /**
-   * <pre>
-   * Version number.
-   * In version 0, if the "repeated xxx" representations contain only one
-   * element, that element is repeated to fill the shape.  This makes it easy
-   * to represent a constant Tensor with a single value.
-   * </pre>
-   *
-   * <code>int32 version_number = 3;</code>
-   */
-  int getVersionNumber();
-
-  /**
-   * <pre>
-   * Serialized raw tensor content from either Tensor::AsProtoTensorContent or
-   * memcpy in tensorflow::grpc::EncodeTensorToByteBuffer. This representation
-   * can be used for all tensor types. The purpose of this representation is to
-   * reduce serialization overhead during RPC call by avoiding serialization of
-   * many repeated small items.
-   * </pre>
-   *
-   * <code>bytes tensor_content = 4;</code>
-   */
-  com.google.protobuf.ByteString getTensorContent();
-
-  /**
-   * <pre>
-   * DT_HALF, DT_BFLOAT16. Note that since protobuf has no int16 type, we'll
-   * have some pointless zero padding for each value here.
-   * </pre>
-   *
-   * <code>repeated int32 half_val = 13 [packed = true];</code>
-   */
-  java.util.List<java.lang.Integer> getHalfValList();
-  /**
-   * <pre>
-   * DT_HALF, DT_BFLOAT16. Note that since protobuf has no int16 type, we'll
-   * have some pointless zero padding for each value here.
-   * </pre>
-   *
-   * <code>repeated int32 half_val = 13 [packed = true];</code>
-   */
-  int getHalfValCount();
-  /**
-   * <pre>
-   * DT_HALF, DT_BFLOAT16. Note that since protobuf has no int16 type, we'll
-   * have some pointless zero padding for each value here.
-   * </pre>
-   *
-   * <code>repeated int32 half_val = 13 [packed = true];</code>
-   */
-  int getHalfVal(int index);
-
-  /**
-   * <pre>
-   * DT_FLOAT.
-   * </pre>
-   *
-   * <code>repeated float float_val = 5 [packed = true];</code>
-   */
-  java.util.List<java.lang.Float> getFloatValList();
-  /**
-   * <pre>
-   * DT_FLOAT.
-   * </pre>
-   *
-   * <code>repeated float float_val = 5 [packed = true];</code>
-   */
-  int getFloatValCount();
-  /**
-   * <pre>
-   * DT_FLOAT.
-   * </pre>
-   *
-   * <code>repeated float float_val = 5 [packed = true];</code>
-   */
-  float getFloatVal(int index);
-
-  /**
-   * <pre>
-   * DT_DOUBLE.
-   * </pre>
-   *
-   * <code>repeated double double_val = 6 [packed = true];</code>
-   */
-  java.util.List<java.lang.Double> getDoubleValList();
-  /**
-   * <pre>
-   * DT_DOUBLE.
-   * </pre>
-   *
-   * <code>repeated double double_val = 6 [packed = true];</code>
-   */
-  int getDoubleValCount();
-  /**
-   * <pre>
-   * DT_DOUBLE.
-   * </pre>
-   *
-   * <code>repeated double double_val = 6 [packed = true];</code>
-   */
-  double getDoubleVal(int index);
-
-  /**
-   * <pre>
-   * DT_INT32, DT_INT16, DT_INT8, DT_UINT8.
-   * </pre>
-   *
-   * <code>repeated int32 int_val = 7 [packed = true];</code>
-   */
-  java.util.List<java.lang.Integer> getIntValList();
-  /**
-   * <pre>
-   * DT_INT32, DT_INT16, DT_INT8, DT_UINT8.
-   * </pre>
-   *
-   * <code>repeated int32 int_val = 7 [packed = true];</code>
-   */
-  int getIntValCount();
-  /**
-   * <pre>
-   * DT_INT32, DT_INT16, DT_INT8, DT_UINT8.
-   * </pre>
-   *
-   * <code>repeated int32 int_val = 7 [packed = true];</code>
-   */
-  int getIntVal(int index);
-
-  /**
-   * <pre>
-   * DT_STRING
-   * </pre>
-   *
-   * <code>repeated bytes string_val = 8;</code>
-   */
-  java.util.List<com.google.protobuf.ByteString> getStringValList();
-  /**
-   * <pre>
-   * DT_STRING
-   * </pre>
-   *
-   * <code>repeated bytes string_val = 8;</code>
-   */
-  int getStringValCount();
-  /**
-   * <pre>
-   * DT_STRING
-   * </pre>
-   *
-   * <code>repeated bytes string_val = 8;</code>
-   */
-  com.google.protobuf.ByteString getStringVal(int index);
-
-  /**
-   * <pre>
-   * DT_COMPLEX64. scomplex_val(2*i) and scomplex_val(2*i+1) are real
-   * and imaginary parts of i-th single precision complex.
-   * </pre>
-   *
-   * <code>repeated float scomplex_val = 9 [packed = true];</code>
-   */
-  java.util.List<java.lang.Float> getScomplexValList();
-  /**
-   * <pre>
-   * DT_COMPLEX64. scomplex_val(2*i) and scomplex_val(2*i+1) are real
-   * and imaginary parts of i-th single precision complex.
-   * </pre>
-   *
-   * <code>repeated float scomplex_val = 9 [packed = true];</code>
-   */
-  int getScomplexValCount();
-  /**
-   * <pre>
-   * DT_COMPLEX64. scomplex_val(2*i) and scomplex_val(2*i+1) are real
-   * and imaginary parts of i-th single precision complex.
-   * </pre>
-   *
-   * <code>repeated float scomplex_val = 9 [packed = true];</code>
-   */
-  float getScomplexVal(int index);
-
-  /**
-   * <pre>
-   * DT_INT64
-   * </pre>
-   *
-   * <code>repeated int64 int64_val = 10 [packed = true];</code>
-   */
-  java.util.List<java.lang.Long> getInt64ValList();
-  /**
-   * <pre>
-   * DT_INT64
-   * </pre>
-   *
-   * <code>repeated int64 int64_val = 10 [packed = true];</code>
-   */
-  int getInt64ValCount();
-  /**
-   * <pre>
-   * DT_INT64
-   * </pre>
-   *
-   * <code>repeated int64 int64_val = 10 [packed = true];</code>
-   */
-  long getInt64Val(int index);
-
-  /**
-   * <pre>
-   * DT_BOOL
-   * </pre>
-   *
-   * <code>repeated bool bool_val = 11 [packed = true];</code>
-   */
-  java.util.List<java.lang.Boolean> getBoolValList();
-  /**
-   * <pre>
-   * DT_BOOL
-   * </pre>
-   *
-   * <code>repeated bool bool_val = 11 [packed = true];</code>
-   */
-  int getBoolValCount();
-  /**
-   * <pre>
-   * DT_BOOL
-   * </pre>
-   *
-   * <code>repeated bool bool_val = 11 [packed = true];</code>
-   */
-  boolean getBoolVal(int index);
-
-  /**
-   * <pre>
-   * DT_COMPLEX128. dcomplex_val(2*i) and dcomplex_val(2*i+1) are real
-   * and imaginary parts of i-th double precision complex.
-   * </pre>
-   *
-   * <code>repeated double dcomplex_val = 12 [packed = true];</code>
-   */
-  java.util.List<java.lang.Double> getDcomplexValList();
-  /**
-   * <pre>
-   * DT_COMPLEX128. dcomplex_val(2*i) and dcomplex_val(2*i+1) are real
-   * and imaginary parts of i-th double precision complex.
-   * </pre>
-   *
-   * <code>repeated double dcomplex_val = 12 [packed = true];</code>
-   */
-  int getDcomplexValCount();
-  /**
-   * <pre>
-   * DT_COMPLEX128. dcomplex_val(2*i) and dcomplex_val(2*i+1) are real
-   * and imaginary parts of i-th double precision complex.
-   * </pre>
-   *
-   * <code>repeated double dcomplex_val = 12 [packed = true];</code>
-   */
-  double getDcomplexVal(int index);
-
-  /**
-   * <pre>
-   * DT_RESOURCE
-   * </pre>
-   *
-   * <code>repeated .tensorflow.ResourceHandleProto resource_handle_val = 14;</code>
-   */
-  java.util.List<org.tensorflow.proto.framework.ResourceHandleProto> 
-      getResourceHandleValList();
-  /**
-   * <pre>
-   * DT_RESOURCE
-   * </pre>
-   *
-   * <code>repeated .tensorflow.ResourceHandleProto resource_handle_val = 14;</code>
-   */
-  org.tensorflow.proto.framework.ResourceHandleProto getResourceHandleVal(int index);
-  /**
-   * <pre>
-   * DT_RESOURCE
-   * </pre>
-   *
-   * <code>repeated .tensorflow.ResourceHandleProto resource_handle_val = 14;</code>
-   */
-  int getResourceHandleValCount();
-  /**
-   * <pre>
-   * DT_RESOURCE
-   * </pre>
-   *
-   * <code>repeated .tensorflow.ResourceHandleProto resource_handle_val = 14;</code>
-   */
-  java.util.List<? extends org.tensorflow.proto.framework.ResourceHandleProtoOrBuilder> 
-      getResourceHandleValOrBuilderList();
-  /**
-   * <pre>
-   * DT_RESOURCE
-   * </pre>
-   *
-   * <code>repeated .tensorflow.ResourceHandleProto resource_handle_val = 14;</code>
-   */
-  org.tensorflow.proto.framework.ResourceHandleProtoOrBuilder getResourceHandleValOrBuilder(
-      int index);
-
-  /**
-   * <pre>
-   * DT_VARIANT
-   * </pre>
-   *
-   * <code>repeated .tensorflow.VariantTensorDataProto variant_val = 15;</code>
-   */
-  java.util.List<org.tensorflow.proto.framework.VariantTensorDataProto> 
-      getVariantValList();
-  /**
-   * <pre>
-   * DT_VARIANT
-   * </pre>
-   *
-   * <code>repeated .tensorflow.VariantTensorDataProto variant_val = 15;</code>
-   */
-  org.tensorflow.proto.framework.VariantTensorDataProto getVariantVal(int index);
-  /**
-   * <pre>
-   * DT_VARIANT
-   * </pre>
-   *
-   * <code>repeated .tensorflow.VariantTensorDataProto variant_val = 15;</code>
-   */
-  int getVariantValCount();
-  /**
-   * <pre>
-   * DT_VARIANT
-   * </pre>
-   *
-   * <code>repeated .tensorflow.VariantTensorDataProto variant_val = 15;</code>
-   */
-  java.util.List<? extends org.tensorflow.proto.framework.VariantTensorDataProtoOrBuilder> 
-      getVariantValOrBuilderList();
-  /**
-   * <pre>
-   * DT_VARIANT
-   * </pre>
-   *
-   * <code>repeated .tensorflow.VariantTensorDataProto variant_val = 15;</code>
-   */
-  org.tensorflow.proto.framework.VariantTensorDataProtoOrBuilder getVariantValOrBuilder(
-      int index);
-
-  /**
-   * <pre>
-   * DT_UINT32
-   * </pre>
-   *
-   * <code>repeated uint32 uint32_val = 16 [packed = true];</code>
-   */
-  java.util.List<java.lang.Integer> getUint32ValList();
-  /**
-   * <pre>
-   * DT_UINT32
-   * </pre>
-   *
-   * <code>repeated uint32 uint32_val = 16 [packed = true];</code>
-   */
-  int getUint32ValCount();
-  /**
-   * <pre>
-   * DT_UINT32
-   * </pre>
-   *
-   * <code>repeated uint32 uint32_val = 16 [packed = true];</code>
-   */
-  int getUint32Val(int index);
-
-  /**
-   * <pre>
-   * DT_UINT64
-   * </pre>
-   *
-   * <code>repeated uint64 uint64_val = 17 [packed = true];</code>
-   */
-  java.util.List<java.lang.Long> getUint64ValList();
-  /**
-   * <pre>
-   * DT_UINT64
-   * </pre>
-   *
-   * <code>repeated uint64 uint64_val = 17 [packed = true];</code>
-   */
-  int getUint64ValCount();
-  /**
-   * <pre>
-   * DT_UINT64
-   * </pre>
-   *
-   * <code>repeated uint64 uint64_val = 17 [packed = true];</code>
-   */
-  long getUint64Val(int index);
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorProtos.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorProtos.java
deleted file mode 100644
index 02a9fdadd6e..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorProtos.java
+++ /dev/null
@@ -1,86 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/framework/tensor.proto
-
-package org.tensorflow.proto.framework;
-
-public final class TensorProtos {
-  private TensorProtos() {}
-  public static void registerAllExtensions(
-      com.google.protobuf.ExtensionRegistryLite registry) {
-  }
-
-  public static void registerAllExtensions(
-      com.google.protobuf.ExtensionRegistry registry) {
-    registerAllExtensions(
-        (com.google.protobuf.ExtensionRegistryLite) registry);
-  }
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_TensorProto_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_TensorProto_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_VariantTensorDataProto_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_VariantTensorDataProto_fieldAccessorTable;
-
-  public static com.google.protobuf.Descriptors.FileDescriptor
-      getDescriptor() {
-    return descriptor;
-  }
-  private static  com.google.protobuf.Descriptors.FileDescriptor
-      descriptor;
-  static {
-    java.lang.String[] descriptorData = {
-      "\n&tensorflow/core/framework/tensor.proto" +
-      "\022\ntensorflow\032/tensorflow/core/framework/" +
-      "resource_handle.proto\032,tensorflow/core/f" +
-      "ramework/tensor_shape.proto\032%tensorflow/" +
-      "core/framework/types.proto\"\214\004\n\013TensorPro" +
-      "to\022#\n\005dtype\030\001 \001(\0162\024.tensorflow.DataType\022" +
-      "2\n\014tensor_shape\030\002 \001(\0132\034.tensorflow.Tenso" +
-      "rShapeProto\022\026\n\016version_number\030\003 \001(\005\022\026\n\016t" +
-      "ensor_content\030\004 \001(\014\022\024\n\010half_val\030\r \003(\005B\002\020" +
-      "\001\022\025\n\tfloat_val\030\005 \003(\002B\002\020\001\022\026\n\ndouble_val\030\006" +
-      " \003(\001B\002\020\001\022\023\n\007int_val\030\007 \003(\005B\002\020\001\022\022\n\nstring_" +
-      "val\030\010 \003(\014\022\030\n\014scomplex_val\030\t \003(\002B\002\020\001\022\025\n\ti" +
-      "nt64_val\030\n \003(\003B\002\020\001\022\024\n\010bool_val\030\013 \003(\010B\002\020\001" +
-      "\022\030\n\014dcomplex_val\030\014 \003(\001B\002\020\001\022<\n\023resource_h" +
-      "andle_val\030\016 \003(\0132\037.tensorflow.ResourceHan" +
-      "dleProto\0227\n\013variant_val\030\017 \003(\0132\".tensorfl" +
-      "ow.VariantTensorDataProto\022\026\n\nuint32_val\030" +
-      "\020 \003(\rB\002\020\001\022\026\n\nuint64_val\030\021 \003(\004B\002\020\001\"g\n\026Var" +
-      "iantTensorDataProto\022\021\n\ttype_name\030\001 \001(\t\022\020" +
-      "\n\010metadata\030\002 \001(\014\022(\n\007tensors\030\003 \003(\0132\027.tens" +
-      "orflow.TensorProtoB\202\001\n\036org.tensorflow.pr" +
-      "oto.frameworkB\014TensorProtosP\001ZMgithub.co" +
-      "m/tensorflow/tensorflow/tensorflow/go/co" +
-      "re/framework/tensor_go_proto\370\001\001b\006proto3"
-    };
-    descriptor = com.google.protobuf.Descriptors.FileDescriptor
-      .internalBuildGeneratedFileFrom(descriptorData,
-        new com.google.protobuf.Descriptors.FileDescriptor[] {
-          org.tensorflow.proto.framework.ResourceHandle.getDescriptor(),
-          org.tensorflow.proto.framework.TensorShapeProtos.getDescriptor(),
-          org.tensorflow.proto.framework.TypesProtos.getDescriptor(),
-        });
-    internal_static_tensorflow_TensorProto_descriptor =
-      getDescriptor().getMessageTypes().get(0);
-    internal_static_tensorflow_TensorProto_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_TensorProto_descriptor,
-        new java.lang.String[] { "Dtype", "TensorShape", "VersionNumber", "TensorContent", "HalfVal", "FloatVal", "DoubleVal", "IntVal", "StringVal", "ScomplexVal", "Int64Val", "BoolVal", "DcomplexVal", "ResourceHandleVal", "VariantVal", "Uint32Val", "Uint64Val", });
-    internal_static_tensorflow_VariantTensorDataProto_descriptor =
-      getDescriptor().getMessageTypes().get(1);
-    internal_static_tensorflow_VariantTensorDataProto_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_VariantTensorDataProto_descriptor,
-        new java.lang.String[] { "TypeName", "Metadata", "Tensors", });
-    org.tensorflow.proto.framework.ResourceHandle.getDescriptor();
-    org.tensorflow.proto.framework.TensorShapeProtos.getDescriptor();
-    org.tensorflow.proto.framework.TypesProtos.getDescriptor();
-  }
-
-  // @@protoc_insertion_point(outer_class_scope)
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorShapeProto.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorShapeProto.java
deleted file mode 100644
index 1d0d543c656..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorShapeProto.java
+++ /dev/null
@@ -1,1852 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/framework/tensor_shape.proto
-
-package org.tensorflow.proto.framework;
-
-/**
- * <pre>
- * Dimensions of a tensor.
- * </pre>
- *
- * Protobuf type {@code tensorflow.TensorShapeProto}
- */
-public  final class TensorShapeProto extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.TensorShapeProto)
-    TensorShapeProtoOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use TensorShapeProto.newBuilder() to construct.
-  private TensorShapeProto(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private TensorShapeProto() {
-    dim_ = java.util.Collections.emptyList();
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new TensorShapeProto();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private TensorShapeProto(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    int mutable_bitField0_ = 0;
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 18: {
-            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
-              dim_ = new java.util.ArrayList<org.tensorflow.proto.framework.TensorShapeProto.Dim>();
-              mutable_bitField0_ |= 0x00000001;
-            }
-            dim_.add(
-                input.readMessage(org.tensorflow.proto.framework.TensorShapeProto.Dim.parser(), extensionRegistry));
-            break;
-          }
-          case 24: {
-
-            unknownRank_ = input.readBool();
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      if (((mutable_bitField0_ & 0x00000001) != 0)) {
-        dim_ = java.util.Collections.unmodifiableList(dim_);
-      }
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.framework.TensorShapeProtos.internal_static_tensorflow_TensorShapeProto_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.framework.TensorShapeProtos.internal_static_tensorflow_TensorShapeProto_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.framework.TensorShapeProto.class, org.tensorflow.proto.framework.TensorShapeProto.Builder.class);
-  }
-
-  public interface DimOrBuilder extends
-      // @@protoc_insertion_point(interface_extends:tensorflow.TensorShapeProto.Dim)
-      com.google.protobuf.MessageOrBuilder {
-
-    /**
-     * <pre>
-     * Size of the tensor in that dimension.
-     * This value must be &gt;= -1, but values of -1 are reserved for "unknown"
-     * shapes (values of -1 mean "unknown" dimension).  Certain wrappers
-     * that work with TensorShapeProto may fail at runtime when deserializing
-     * a TensorShapeProto containing a dim value of -1.
-     * </pre>
-     *
-     * <code>int64 size = 1;</code>
-     */
-    long getSize();
-
-    /**
-     * <pre>
-     * Optional name of the tensor dimension.
-     * </pre>
-     *
-     * <code>string name = 2;</code>
-     */
-    java.lang.String getName();
-    /**
-     * <pre>
-     * Optional name of the tensor dimension.
-     * </pre>
-     *
-     * <code>string name = 2;</code>
-     */
-    com.google.protobuf.ByteString
-        getNameBytes();
-  }
-  /**
-   * <pre>
-   * One dimension of the tensor.
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.TensorShapeProto.Dim}
-   */
-  public  static final class Dim extends
-      com.google.protobuf.GeneratedMessageV3 implements
-      // @@protoc_insertion_point(message_implements:tensorflow.TensorShapeProto.Dim)
-      DimOrBuilder {
-  private static final long serialVersionUID = 0L;
-    // Use Dim.newBuilder() to construct.
-    private Dim(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-      super(builder);
-    }
-    private Dim() {
-      name_ = "";
-    }
-
-    @java.lang.Override
-    @SuppressWarnings({"unused"})
-    protected java.lang.Object newInstance(
-        UnusedPrivateParameter unused) {
-      return new Dim();
-    }
-
-    @java.lang.Override
-    public final com.google.protobuf.UnknownFieldSet
-    getUnknownFields() {
-      return this.unknownFields;
-    }
-    private Dim(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      this();
-      if (extensionRegistry == null) {
-        throw new java.lang.NullPointerException();
-      }
-      com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-          com.google.protobuf.UnknownFieldSet.newBuilder();
-      try {
-        boolean done = false;
-        while (!done) {
-          int tag = input.readTag();
-          switch (tag) {
-            case 0:
-              done = true;
-              break;
-            case 8: {
-
-              size_ = input.readInt64();
-              break;
-            }
-            case 18: {
-              java.lang.String s = input.readStringRequireUtf8();
-
-              name_ = s;
-              break;
-            }
-            default: {
-              if (!parseUnknownField(
-                  input, unknownFields, extensionRegistry, tag)) {
-                done = true;
-              }
-              break;
-            }
-          }
-        }
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        throw e.setUnfinishedMessage(this);
-      } catch (java.io.IOException e) {
-        throw new com.google.protobuf.InvalidProtocolBufferException(
-            e).setUnfinishedMessage(this);
-      } finally {
-        this.unknownFields = unknownFields.build();
-        makeExtensionsImmutable();
-      }
-    }
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.TensorShapeProtos.internal_static_tensorflow_TensorShapeProto_Dim_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.TensorShapeProtos.internal_static_tensorflow_TensorShapeProto_Dim_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.TensorShapeProto.Dim.class, org.tensorflow.proto.framework.TensorShapeProto.Dim.Builder.class);
-    }
-
-    public static final int SIZE_FIELD_NUMBER = 1;
-    private long size_;
-    /**
-     * <pre>
-     * Size of the tensor in that dimension.
-     * This value must be &gt;= -1, but values of -1 are reserved for "unknown"
-     * shapes (values of -1 mean "unknown" dimension).  Certain wrappers
-     * that work with TensorShapeProto may fail at runtime when deserializing
-     * a TensorShapeProto containing a dim value of -1.
-     * </pre>
-     *
-     * <code>int64 size = 1;</code>
-     */
-    public long getSize() {
-      return size_;
-    }
-
-    public static final int NAME_FIELD_NUMBER = 2;
-    private volatile java.lang.Object name_;
-    /**
-     * <pre>
-     * Optional name of the tensor dimension.
-     * </pre>
-     *
-     * <code>string name = 2;</code>
-     */
-    public java.lang.String getName() {
-      java.lang.Object ref = name_;
-      if (ref instanceof java.lang.String) {
-        return (java.lang.String) ref;
-      } else {
-        com.google.protobuf.ByteString bs = 
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        name_ = s;
-        return s;
-      }
-    }
-    /**
-     * <pre>
-     * Optional name of the tensor dimension.
-     * </pre>
-     *
-     * <code>string name = 2;</code>
-     */
-    public com.google.protobuf.ByteString
-        getNameBytes() {
-      java.lang.Object ref = name_;
-      if (ref instanceof java.lang.String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        name_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-
-    private byte memoizedIsInitialized = -1;
-    @java.lang.Override
-    public final boolean isInitialized() {
-      byte isInitialized = memoizedIsInitialized;
-      if (isInitialized == 1) return true;
-      if (isInitialized == 0) return false;
-
-      memoizedIsInitialized = 1;
-      return true;
-    }
-
-    @java.lang.Override
-    public void writeTo(com.google.protobuf.CodedOutputStream output)
-                        throws java.io.IOException {
-      if (size_ != 0L) {
-        output.writeInt64(1, size_);
-      }
-      if (!getNameBytes().isEmpty()) {
-        com.google.protobuf.GeneratedMessageV3.writeString(output, 2, name_);
-      }
-      unknownFields.writeTo(output);
-    }
-
-    @java.lang.Override
-    public int getSerializedSize() {
-      int size = memoizedSize;
-      if (size != -1) return size;
-
-      size = 0;
-      if (size_ != 0L) {
-        size += com.google.protobuf.CodedOutputStream
-          .computeInt64Size(1, size_);
-      }
-      if (!getNameBytes().isEmpty()) {
-        size += com.google.protobuf.GeneratedMessageV3.computeStringSize(2, name_);
-      }
-      size += unknownFields.getSerializedSize();
-      memoizedSize = size;
-      return size;
-    }
-
-    @java.lang.Override
-    public boolean equals(final java.lang.Object obj) {
-      if (obj == this) {
-       return true;
-      }
-      if (!(obj instanceof org.tensorflow.proto.framework.TensorShapeProto.Dim)) {
-        return super.equals(obj);
-      }
-      org.tensorflow.proto.framework.TensorShapeProto.Dim other = (org.tensorflow.proto.framework.TensorShapeProto.Dim) obj;
-
-      if (getSize()
-          != other.getSize()) return false;
-      if (!getName()
-          .equals(other.getName())) return false;
-      if (!unknownFields.equals(other.unknownFields)) return false;
-      return true;
-    }
-
-    @java.lang.Override
-    public int hashCode() {
-      if (memoizedHashCode != 0) {
-        return memoizedHashCode;
-      }
-      int hash = 41;
-      hash = (19 * hash) + getDescriptor().hashCode();
-      hash = (37 * hash) + SIZE_FIELD_NUMBER;
-      hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-          getSize());
-      hash = (37 * hash) + NAME_FIELD_NUMBER;
-      hash = (53 * hash) + getName().hashCode();
-      hash = (29 * hash) + unknownFields.hashCode();
-      memoizedHashCode = hash;
-      return hash;
-    }
-
-    public static org.tensorflow.proto.framework.TensorShapeProto.Dim parseFrom(
-        java.nio.ByteBuffer data)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data);
-    }
-    public static org.tensorflow.proto.framework.TensorShapeProto.Dim parseFrom(
-        java.nio.ByteBuffer data,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.TensorShapeProto.Dim parseFrom(
-        com.google.protobuf.ByteString data)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data);
-    }
-    public static org.tensorflow.proto.framework.TensorShapeProto.Dim parseFrom(
-        com.google.protobuf.ByteString data,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.TensorShapeProto.Dim parseFrom(byte[] data)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data);
-    }
-    public static org.tensorflow.proto.framework.TensorShapeProto.Dim parseFrom(
-        byte[] data,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.TensorShapeProto.Dim parseFrom(java.io.InputStream input)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input);
-    }
-    public static org.tensorflow.proto.framework.TensorShapeProto.Dim parseFrom(
-        java.io.InputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.TensorShapeProto.Dim parseDelimitedFrom(java.io.InputStream input)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseDelimitedWithIOException(PARSER, input);
-    }
-    public static org.tensorflow.proto.framework.TensorShapeProto.Dim parseDelimitedFrom(
-        java.io.InputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.TensorShapeProto.Dim parseFrom(
-        com.google.protobuf.CodedInputStream input)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input);
-    }
-    public static org.tensorflow.proto.framework.TensorShapeProto.Dim parseFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input, extensionRegistry);
-    }
-
-    @java.lang.Override
-    public Builder newBuilderForType() { return newBuilder(); }
-    public static Builder newBuilder() {
-      return DEFAULT_INSTANCE.toBuilder();
-    }
-    public static Builder newBuilder(org.tensorflow.proto.framework.TensorShapeProto.Dim prototype) {
-      return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-    }
-    @java.lang.Override
-    public Builder toBuilder() {
-      return this == DEFAULT_INSTANCE
-          ? new Builder() : new Builder().mergeFrom(this);
-    }
-
-    @java.lang.Override
-    protected Builder newBuilderForType(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      Builder builder = new Builder(parent);
-      return builder;
-    }
-    /**
-     * <pre>
-     * One dimension of the tensor.
-     * </pre>
-     *
-     * Protobuf type {@code tensorflow.TensorShapeProto.Dim}
-     */
-    public static final class Builder extends
-        com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-        // @@protoc_insertion_point(builder_implements:tensorflow.TensorShapeProto.Dim)
-        org.tensorflow.proto.framework.TensorShapeProto.DimOrBuilder {
-      public static final com.google.protobuf.Descriptors.Descriptor
-          getDescriptor() {
-        return org.tensorflow.proto.framework.TensorShapeProtos.internal_static_tensorflow_TensorShapeProto_Dim_descriptor;
-      }
-
-      @java.lang.Override
-      protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-          internalGetFieldAccessorTable() {
-        return org.tensorflow.proto.framework.TensorShapeProtos.internal_static_tensorflow_TensorShapeProto_Dim_fieldAccessorTable
-            .ensureFieldAccessorsInitialized(
-                org.tensorflow.proto.framework.TensorShapeProto.Dim.class, org.tensorflow.proto.framework.TensorShapeProto.Dim.Builder.class);
-      }
-
-      // Construct using org.tensorflow.proto.framework.TensorShapeProto.Dim.newBuilder()
-      private Builder() {
-        maybeForceBuilderInitialization();
-      }
-
-      private Builder(
-          com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-        super(parent);
-        maybeForceBuilderInitialization();
-      }
-      private void maybeForceBuilderInitialization() {
-        if (com.google.protobuf.GeneratedMessageV3
-                .alwaysUseFieldBuilders) {
-        }
-      }
-      @java.lang.Override
-      public Builder clear() {
-        super.clear();
-        size_ = 0L;
-
-        name_ = "";
-
-        return this;
-      }
-
-      @java.lang.Override
-      public com.google.protobuf.Descriptors.Descriptor
-          getDescriptorForType() {
-        return org.tensorflow.proto.framework.TensorShapeProtos.internal_static_tensorflow_TensorShapeProto_Dim_descriptor;
-      }
-
-      @java.lang.Override
-      public org.tensorflow.proto.framework.TensorShapeProto.Dim getDefaultInstanceForType() {
-        return org.tensorflow.proto.framework.TensorShapeProto.Dim.getDefaultInstance();
-      }
-
-      @java.lang.Override
-      public org.tensorflow.proto.framework.TensorShapeProto.Dim build() {
-        org.tensorflow.proto.framework.TensorShapeProto.Dim result = buildPartial();
-        if (!result.isInitialized()) {
-          throw newUninitializedMessageException(result);
-        }
-        return result;
-      }
-
-      @java.lang.Override
-      public org.tensorflow.proto.framework.TensorShapeProto.Dim buildPartial() {
-        org.tensorflow.proto.framework.TensorShapeProto.Dim result = new org.tensorflow.proto.framework.TensorShapeProto.Dim(this);
-        result.size_ = size_;
-        result.name_ = name_;
-        onBuilt();
-        return result;
-      }
-
-      @java.lang.Override
-      public Builder clone() {
-        return super.clone();
-      }
-      @java.lang.Override
-      public Builder setField(
-          com.google.protobuf.Descriptors.FieldDescriptor field,
-          java.lang.Object value) {
-        return super.setField(field, value);
-      }
-      @java.lang.Override
-      public Builder clearField(
-          com.google.protobuf.Descriptors.FieldDescriptor field) {
-        return super.clearField(field);
-      }
-      @java.lang.Override
-      public Builder clearOneof(
-          com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-        return super.clearOneof(oneof);
-      }
-      @java.lang.Override
-      public Builder setRepeatedField(
-          com.google.protobuf.Descriptors.FieldDescriptor field,
-          int index, java.lang.Object value) {
-        return super.setRepeatedField(field, index, value);
-      }
-      @java.lang.Override
-      public Builder addRepeatedField(
-          com.google.protobuf.Descriptors.FieldDescriptor field,
-          java.lang.Object value) {
-        return super.addRepeatedField(field, value);
-      }
-      @java.lang.Override
-      public Builder mergeFrom(com.google.protobuf.Message other) {
-        if (other instanceof org.tensorflow.proto.framework.TensorShapeProto.Dim) {
-          return mergeFrom((org.tensorflow.proto.framework.TensorShapeProto.Dim)other);
-        } else {
-          super.mergeFrom(other);
-          return this;
-        }
-      }
-
-      public Builder mergeFrom(org.tensorflow.proto.framework.TensorShapeProto.Dim other) {
-        if (other == org.tensorflow.proto.framework.TensorShapeProto.Dim.getDefaultInstance()) return this;
-        if (other.getSize() != 0L) {
-          setSize(other.getSize());
-        }
-        if (!other.getName().isEmpty()) {
-          name_ = other.name_;
-          onChanged();
-        }
-        this.mergeUnknownFields(other.unknownFields);
-        onChanged();
-        return this;
-      }
-
-      @java.lang.Override
-      public final boolean isInitialized() {
-        return true;
-      }
-
-      @java.lang.Override
-      public Builder mergeFrom(
-          com.google.protobuf.CodedInputStream input,
-          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-          throws java.io.IOException {
-        org.tensorflow.proto.framework.TensorShapeProto.Dim parsedMessage = null;
-        try {
-          parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-        } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-          parsedMessage = (org.tensorflow.proto.framework.TensorShapeProto.Dim) e.getUnfinishedMessage();
-          throw e.unwrapIOException();
-        } finally {
-          if (parsedMessage != null) {
-            mergeFrom(parsedMessage);
-          }
-        }
-        return this;
-      }
-
-      private long size_ ;
-      /**
-       * <pre>
-       * Size of the tensor in that dimension.
-       * This value must be &gt;= -1, but values of -1 are reserved for "unknown"
-       * shapes (values of -1 mean "unknown" dimension).  Certain wrappers
-       * that work with TensorShapeProto may fail at runtime when deserializing
-       * a TensorShapeProto containing a dim value of -1.
-       * </pre>
-       *
-       * <code>int64 size = 1;</code>
-       */
-      public long getSize() {
-        return size_;
-      }
-      /**
-       * <pre>
-       * Size of the tensor in that dimension.
-       * This value must be &gt;= -1, but values of -1 are reserved for "unknown"
-       * shapes (values of -1 mean "unknown" dimension).  Certain wrappers
-       * that work with TensorShapeProto may fail at runtime when deserializing
-       * a TensorShapeProto containing a dim value of -1.
-       * </pre>
-       *
-       * <code>int64 size = 1;</code>
-       */
-      public Builder setSize(long value) {
-        
-        size_ = value;
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * Size of the tensor in that dimension.
-       * This value must be &gt;= -1, but values of -1 are reserved for "unknown"
-       * shapes (values of -1 mean "unknown" dimension).  Certain wrappers
-       * that work with TensorShapeProto may fail at runtime when deserializing
-       * a TensorShapeProto containing a dim value of -1.
-       * </pre>
-       *
-       * <code>int64 size = 1;</code>
-       */
-      public Builder clearSize() {
-        
-        size_ = 0L;
-        onChanged();
-        return this;
-      }
-
-      private java.lang.Object name_ = "";
-      /**
-       * <pre>
-       * Optional name of the tensor dimension.
-       * </pre>
-       *
-       * <code>string name = 2;</code>
-       */
-      public java.lang.String getName() {
-        java.lang.Object ref = name_;
-        if (!(ref instanceof java.lang.String)) {
-          com.google.protobuf.ByteString bs =
-              (com.google.protobuf.ByteString) ref;
-          java.lang.String s = bs.toStringUtf8();
-          name_ = s;
-          return s;
-        } else {
-          return (java.lang.String) ref;
-        }
-      }
-      /**
-       * <pre>
-       * Optional name of the tensor dimension.
-       * </pre>
-       *
-       * <code>string name = 2;</code>
-       */
-      public com.google.protobuf.ByteString
-          getNameBytes() {
-        java.lang.Object ref = name_;
-        if (ref instanceof String) {
-          com.google.protobuf.ByteString b = 
-              com.google.protobuf.ByteString.copyFromUtf8(
-                  (java.lang.String) ref);
-          name_ = b;
-          return b;
-        } else {
-          return (com.google.protobuf.ByteString) ref;
-        }
-      }
-      /**
-       * <pre>
-       * Optional name of the tensor dimension.
-       * </pre>
-       *
-       * <code>string name = 2;</code>
-       */
-      public Builder setName(
-          java.lang.String value) {
-        if (value == null) {
-    throw new NullPointerException();
-  }
-  
-        name_ = value;
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * Optional name of the tensor dimension.
-       * </pre>
-       *
-       * <code>string name = 2;</code>
-       */
-      public Builder clearName() {
-        
-        name_ = getDefaultInstance().getName();
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * Optional name of the tensor dimension.
-       * </pre>
-       *
-       * <code>string name = 2;</code>
-       */
-      public Builder setNameBytes(
-          com.google.protobuf.ByteString value) {
-        if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-        
-        name_ = value;
-        onChanged();
-        return this;
-      }
-      @java.lang.Override
-      public final Builder setUnknownFields(
-          final com.google.protobuf.UnknownFieldSet unknownFields) {
-        return super.setUnknownFields(unknownFields);
-      }
-
-      @java.lang.Override
-      public final Builder mergeUnknownFields(
-          final com.google.protobuf.UnknownFieldSet unknownFields) {
-        return super.mergeUnknownFields(unknownFields);
-      }
-
-
-      // @@protoc_insertion_point(builder_scope:tensorflow.TensorShapeProto.Dim)
-    }
-
-    // @@protoc_insertion_point(class_scope:tensorflow.TensorShapeProto.Dim)
-    private static final org.tensorflow.proto.framework.TensorShapeProto.Dim DEFAULT_INSTANCE;
-    static {
-      DEFAULT_INSTANCE = new org.tensorflow.proto.framework.TensorShapeProto.Dim();
-    }
-
-    public static org.tensorflow.proto.framework.TensorShapeProto.Dim getDefaultInstance() {
-      return DEFAULT_INSTANCE;
-    }
-
-    private static final com.google.protobuf.Parser<Dim>
-        PARSER = new com.google.protobuf.AbstractParser<Dim>() {
-      @java.lang.Override
-      public Dim parsePartialFrom(
-          com.google.protobuf.CodedInputStream input,
-          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-          throws com.google.protobuf.InvalidProtocolBufferException {
-        return new Dim(input, extensionRegistry);
-      }
-    };
-
-    public static com.google.protobuf.Parser<Dim> parser() {
-      return PARSER;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Parser<Dim> getParserForType() {
-      return PARSER;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.TensorShapeProto.Dim getDefaultInstanceForType() {
-      return DEFAULT_INSTANCE;
-    }
-
-  }
-
-  public static final int DIM_FIELD_NUMBER = 2;
-  private java.util.List<org.tensorflow.proto.framework.TensorShapeProto.Dim> dim_;
-  /**
-   * <pre>
-   * Dimensions of the tensor, such as {"input", 30}, {"output", 40}
-   * for a 30 x 40 2D tensor.  If an entry has size -1, this
-   * corresponds to a dimension of unknown size. The names are
-   * optional.
-   * The order of entries in "dim" matters: It indicates the layout of the
-   * values in the tensor in-memory representation.
-   * The first entry in "dim" is the outermost dimension used to layout the
-   * values, the last entry is the innermost dimension.  This matches the
-   * in-memory layout of RowMajor Eigen tensors.
-   * If "dim.size()" &gt; 0, "unknown_rank" must be false.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.TensorShapeProto.Dim dim = 2;</code>
-   */
-  public java.util.List<org.tensorflow.proto.framework.TensorShapeProto.Dim> getDimList() {
-    return dim_;
-  }
-  /**
-   * <pre>
-   * Dimensions of the tensor, such as {"input", 30}, {"output", 40}
-   * for a 30 x 40 2D tensor.  If an entry has size -1, this
-   * corresponds to a dimension of unknown size. The names are
-   * optional.
-   * The order of entries in "dim" matters: It indicates the layout of the
-   * values in the tensor in-memory representation.
-   * The first entry in "dim" is the outermost dimension used to layout the
-   * values, the last entry is the innermost dimension.  This matches the
-   * in-memory layout of RowMajor Eigen tensors.
-   * If "dim.size()" &gt; 0, "unknown_rank" must be false.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.TensorShapeProto.Dim dim = 2;</code>
-   */
-  public java.util.List<? extends org.tensorflow.proto.framework.TensorShapeProto.DimOrBuilder> 
-      getDimOrBuilderList() {
-    return dim_;
-  }
-  /**
-   * <pre>
-   * Dimensions of the tensor, such as {"input", 30}, {"output", 40}
-   * for a 30 x 40 2D tensor.  If an entry has size -1, this
-   * corresponds to a dimension of unknown size. The names are
-   * optional.
-   * The order of entries in "dim" matters: It indicates the layout of the
-   * values in the tensor in-memory representation.
-   * The first entry in "dim" is the outermost dimension used to layout the
-   * values, the last entry is the innermost dimension.  This matches the
-   * in-memory layout of RowMajor Eigen tensors.
-   * If "dim.size()" &gt; 0, "unknown_rank" must be false.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.TensorShapeProto.Dim dim = 2;</code>
-   */
-  public int getDimCount() {
-    return dim_.size();
-  }
-  /**
-   * <pre>
-   * Dimensions of the tensor, such as {"input", 30}, {"output", 40}
-   * for a 30 x 40 2D tensor.  If an entry has size -1, this
-   * corresponds to a dimension of unknown size. The names are
-   * optional.
-   * The order of entries in "dim" matters: It indicates the layout of the
-   * values in the tensor in-memory representation.
-   * The first entry in "dim" is the outermost dimension used to layout the
-   * values, the last entry is the innermost dimension.  This matches the
-   * in-memory layout of RowMajor Eigen tensors.
-   * If "dim.size()" &gt; 0, "unknown_rank" must be false.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.TensorShapeProto.Dim dim = 2;</code>
-   */
-  public org.tensorflow.proto.framework.TensorShapeProto.Dim getDim(int index) {
-    return dim_.get(index);
-  }
-  /**
-   * <pre>
-   * Dimensions of the tensor, such as {"input", 30}, {"output", 40}
-   * for a 30 x 40 2D tensor.  If an entry has size -1, this
-   * corresponds to a dimension of unknown size. The names are
-   * optional.
-   * The order of entries in "dim" matters: It indicates the layout of the
-   * values in the tensor in-memory representation.
-   * The first entry in "dim" is the outermost dimension used to layout the
-   * values, the last entry is the innermost dimension.  This matches the
-   * in-memory layout of RowMajor Eigen tensors.
-   * If "dim.size()" &gt; 0, "unknown_rank" must be false.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.TensorShapeProto.Dim dim = 2;</code>
-   */
-  public org.tensorflow.proto.framework.TensorShapeProto.DimOrBuilder getDimOrBuilder(
-      int index) {
-    return dim_.get(index);
-  }
-
-  public static final int UNKNOWN_RANK_FIELD_NUMBER = 3;
-  private boolean unknownRank_;
-  /**
-   * <pre>
-   * If true, the number of dimensions in the shape is unknown.
-   * If true, "dim.size()" must be 0.
-   * </pre>
-   *
-   * <code>bool unknown_rank = 3;</code>
-   */
-  public boolean getUnknownRank() {
-    return unknownRank_;
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    for (int i = 0; i < dim_.size(); i++) {
-      output.writeMessage(2, dim_.get(i));
-    }
-    if (unknownRank_ != false) {
-      output.writeBool(3, unknownRank_);
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    for (int i = 0; i < dim_.size(); i++) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(2, dim_.get(i));
-    }
-    if (unknownRank_ != false) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeBoolSize(3, unknownRank_);
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.framework.TensorShapeProto)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.framework.TensorShapeProto other = (org.tensorflow.proto.framework.TensorShapeProto) obj;
-
-    if (!getDimList()
-        .equals(other.getDimList())) return false;
-    if (getUnknownRank()
-        != other.getUnknownRank()) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    if (getDimCount() > 0) {
-      hash = (37 * hash) + DIM_FIELD_NUMBER;
-      hash = (53 * hash) + getDimList().hashCode();
-    }
-    hash = (37 * hash) + UNKNOWN_RANK_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashBoolean(
-        getUnknownRank());
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.framework.TensorShapeProto parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.TensorShapeProto parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.TensorShapeProto parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.TensorShapeProto parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.TensorShapeProto parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.TensorShapeProto parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.TensorShapeProto parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.TensorShapeProto parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.TensorShapeProto parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.TensorShapeProto parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.TensorShapeProto parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.TensorShapeProto parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.framework.TensorShapeProto prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * <pre>
-   * Dimensions of a tensor.
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.TensorShapeProto}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.TensorShapeProto)
-      org.tensorflow.proto.framework.TensorShapeProtoOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.TensorShapeProtos.internal_static_tensorflow_TensorShapeProto_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.TensorShapeProtos.internal_static_tensorflow_TensorShapeProto_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.TensorShapeProto.class, org.tensorflow.proto.framework.TensorShapeProto.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.framework.TensorShapeProto.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-        getDimFieldBuilder();
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      if (dimBuilder_ == null) {
-        dim_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000001);
-      } else {
-        dimBuilder_.clear();
-      }
-      unknownRank_ = false;
-
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.framework.TensorShapeProtos.internal_static_tensorflow_TensorShapeProto_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.TensorShapeProto getDefaultInstanceForType() {
-      return org.tensorflow.proto.framework.TensorShapeProto.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.TensorShapeProto build() {
-      org.tensorflow.proto.framework.TensorShapeProto result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.TensorShapeProto buildPartial() {
-      org.tensorflow.proto.framework.TensorShapeProto result = new org.tensorflow.proto.framework.TensorShapeProto(this);
-      int from_bitField0_ = bitField0_;
-      if (dimBuilder_ == null) {
-        if (((bitField0_ & 0x00000001) != 0)) {
-          dim_ = java.util.Collections.unmodifiableList(dim_);
-          bitField0_ = (bitField0_ & ~0x00000001);
-        }
-        result.dim_ = dim_;
-      } else {
-        result.dim_ = dimBuilder_.build();
-      }
-      result.unknownRank_ = unknownRank_;
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.framework.TensorShapeProto) {
-        return mergeFrom((org.tensorflow.proto.framework.TensorShapeProto)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.framework.TensorShapeProto other) {
-      if (other == org.tensorflow.proto.framework.TensorShapeProto.getDefaultInstance()) return this;
-      if (dimBuilder_ == null) {
-        if (!other.dim_.isEmpty()) {
-          if (dim_.isEmpty()) {
-            dim_ = other.dim_;
-            bitField0_ = (bitField0_ & ~0x00000001);
-          } else {
-            ensureDimIsMutable();
-            dim_.addAll(other.dim_);
-          }
-          onChanged();
-        }
-      } else {
-        if (!other.dim_.isEmpty()) {
-          if (dimBuilder_.isEmpty()) {
-            dimBuilder_.dispose();
-            dimBuilder_ = null;
-            dim_ = other.dim_;
-            bitField0_ = (bitField0_ & ~0x00000001);
-            dimBuilder_ = 
-              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
-                 getDimFieldBuilder() : null;
-          } else {
-            dimBuilder_.addAllMessages(other.dim_);
-          }
-        }
-      }
-      if (other.getUnknownRank() != false) {
-        setUnknownRank(other.getUnknownRank());
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.framework.TensorShapeProto parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.framework.TensorShapeProto) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-    private int bitField0_;
-
-    private java.util.List<org.tensorflow.proto.framework.TensorShapeProto.Dim> dim_ =
-      java.util.Collections.emptyList();
-    private void ensureDimIsMutable() {
-      if (!((bitField0_ & 0x00000001) != 0)) {
-        dim_ = new java.util.ArrayList<org.tensorflow.proto.framework.TensorShapeProto.Dim>(dim_);
-        bitField0_ |= 0x00000001;
-       }
-    }
-
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.framework.TensorShapeProto.Dim, org.tensorflow.proto.framework.TensorShapeProto.Dim.Builder, org.tensorflow.proto.framework.TensorShapeProto.DimOrBuilder> dimBuilder_;
-
-    /**
-     * <pre>
-     * Dimensions of the tensor, such as {"input", 30}, {"output", 40}
-     * for a 30 x 40 2D tensor.  If an entry has size -1, this
-     * corresponds to a dimension of unknown size. The names are
-     * optional.
-     * The order of entries in "dim" matters: It indicates the layout of the
-     * values in the tensor in-memory representation.
-     * The first entry in "dim" is the outermost dimension used to layout the
-     * values, the last entry is the innermost dimension.  This matches the
-     * in-memory layout of RowMajor Eigen tensors.
-     * If "dim.size()" &gt; 0, "unknown_rank" must be false.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorShapeProto.Dim dim = 2;</code>
-     */
-    public java.util.List<org.tensorflow.proto.framework.TensorShapeProto.Dim> getDimList() {
-      if (dimBuilder_ == null) {
-        return java.util.Collections.unmodifiableList(dim_);
-      } else {
-        return dimBuilder_.getMessageList();
-      }
-    }
-    /**
-     * <pre>
-     * Dimensions of the tensor, such as {"input", 30}, {"output", 40}
-     * for a 30 x 40 2D tensor.  If an entry has size -1, this
-     * corresponds to a dimension of unknown size. The names are
-     * optional.
-     * The order of entries in "dim" matters: It indicates the layout of the
-     * values in the tensor in-memory representation.
-     * The first entry in "dim" is the outermost dimension used to layout the
-     * values, the last entry is the innermost dimension.  This matches the
-     * in-memory layout of RowMajor Eigen tensors.
-     * If "dim.size()" &gt; 0, "unknown_rank" must be false.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorShapeProto.Dim dim = 2;</code>
-     */
-    public int getDimCount() {
-      if (dimBuilder_ == null) {
-        return dim_.size();
-      } else {
-        return dimBuilder_.getCount();
-      }
-    }
-    /**
-     * <pre>
-     * Dimensions of the tensor, such as {"input", 30}, {"output", 40}
-     * for a 30 x 40 2D tensor.  If an entry has size -1, this
-     * corresponds to a dimension of unknown size. The names are
-     * optional.
-     * The order of entries in "dim" matters: It indicates the layout of the
-     * values in the tensor in-memory representation.
-     * The first entry in "dim" is the outermost dimension used to layout the
-     * values, the last entry is the innermost dimension.  This matches the
-     * in-memory layout of RowMajor Eigen tensors.
-     * If "dim.size()" &gt; 0, "unknown_rank" must be false.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorShapeProto.Dim dim = 2;</code>
-     */
-    public org.tensorflow.proto.framework.TensorShapeProto.Dim getDim(int index) {
-      if (dimBuilder_ == null) {
-        return dim_.get(index);
-      } else {
-        return dimBuilder_.getMessage(index);
-      }
-    }
-    /**
-     * <pre>
-     * Dimensions of the tensor, such as {"input", 30}, {"output", 40}
-     * for a 30 x 40 2D tensor.  If an entry has size -1, this
-     * corresponds to a dimension of unknown size. The names are
-     * optional.
-     * The order of entries in "dim" matters: It indicates the layout of the
-     * values in the tensor in-memory representation.
-     * The first entry in "dim" is the outermost dimension used to layout the
-     * values, the last entry is the innermost dimension.  This matches the
-     * in-memory layout of RowMajor Eigen tensors.
-     * If "dim.size()" &gt; 0, "unknown_rank" must be false.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorShapeProto.Dim dim = 2;</code>
-     */
-    public Builder setDim(
-        int index, org.tensorflow.proto.framework.TensorShapeProto.Dim value) {
-      if (dimBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureDimIsMutable();
-        dim_.set(index, value);
-        onChanged();
-      } else {
-        dimBuilder_.setMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Dimensions of the tensor, such as {"input", 30}, {"output", 40}
-     * for a 30 x 40 2D tensor.  If an entry has size -1, this
-     * corresponds to a dimension of unknown size. The names are
-     * optional.
-     * The order of entries in "dim" matters: It indicates the layout of the
-     * values in the tensor in-memory representation.
-     * The first entry in "dim" is the outermost dimension used to layout the
-     * values, the last entry is the innermost dimension.  This matches the
-     * in-memory layout of RowMajor Eigen tensors.
-     * If "dim.size()" &gt; 0, "unknown_rank" must be false.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorShapeProto.Dim dim = 2;</code>
-     */
-    public Builder setDim(
-        int index, org.tensorflow.proto.framework.TensorShapeProto.Dim.Builder builderForValue) {
-      if (dimBuilder_ == null) {
-        ensureDimIsMutable();
-        dim_.set(index, builderForValue.build());
-        onChanged();
-      } else {
-        dimBuilder_.setMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Dimensions of the tensor, such as {"input", 30}, {"output", 40}
-     * for a 30 x 40 2D tensor.  If an entry has size -1, this
-     * corresponds to a dimension of unknown size. The names are
-     * optional.
-     * The order of entries in "dim" matters: It indicates the layout of the
-     * values in the tensor in-memory representation.
-     * The first entry in "dim" is the outermost dimension used to layout the
-     * values, the last entry is the innermost dimension.  This matches the
-     * in-memory layout of RowMajor Eigen tensors.
-     * If "dim.size()" &gt; 0, "unknown_rank" must be false.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorShapeProto.Dim dim = 2;</code>
-     */
-    public Builder addDim(org.tensorflow.proto.framework.TensorShapeProto.Dim value) {
-      if (dimBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureDimIsMutable();
-        dim_.add(value);
-        onChanged();
-      } else {
-        dimBuilder_.addMessage(value);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Dimensions of the tensor, such as {"input", 30}, {"output", 40}
-     * for a 30 x 40 2D tensor.  If an entry has size -1, this
-     * corresponds to a dimension of unknown size. The names are
-     * optional.
-     * The order of entries in "dim" matters: It indicates the layout of the
-     * values in the tensor in-memory representation.
-     * The first entry in "dim" is the outermost dimension used to layout the
-     * values, the last entry is the innermost dimension.  This matches the
-     * in-memory layout of RowMajor Eigen tensors.
-     * If "dim.size()" &gt; 0, "unknown_rank" must be false.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorShapeProto.Dim dim = 2;</code>
-     */
-    public Builder addDim(
-        int index, org.tensorflow.proto.framework.TensorShapeProto.Dim value) {
-      if (dimBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureDimIsMutable();
-        dim_.add(index, value);
-        onChanged();
-      } else {
-        dimBuilder_.addMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Dimensions of the tensor, such as {"input", 30}, {"output", 40}
-     * for a 30 x 40 2D tensor.  If an entry has size -1, this
-     * corresponds to a dimension of unknown size. The names are
-     * optional.
-     * The order of entries in "dim" matters: It indicates the layout of the
-     * values in the tensor in-memory representation.
-     * The first entry in "dim" is the outermost dimension used to layout the
-     * values, the last entry is the innermost dimension.  This matches the
-     * in-memory layout of RowMajor Eigen tensors.
-     * If "dim.size()" &gt; 0, "unknown_rank" must be false.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorShapeProto.Dim dim = 2;</code>
-     */
-    public Builder addDim(
-        org.tensorflow.proto.framework.TensorShapeProto.Dim.Builder builderForValue) {
-      if (dimBuilder_ == null) {
-        ensureDimIsMutable();
-        dim_.add(builderForValue.build());
-        onChanged();
-      } else {
-        dimBuilder_.addMessage(builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Dimensions of the tensor, such as {"input", 30}, {"output", 40}
-     * for a 30 x 40 2D tensor.  If an entry has size -1, this
-     * corresponds to a dimension of unknown size. The names are
-     * optional.
-     * The order of entries in "dim" matters: It indicates the layout of the
-     * values in the tensor in-memory representation.
-     * The first entry in "dim" is the outermost dimension used to layout the
-     * values, the last entry is the innermost dimension.  This matches the
-     * in-memory layout of RowMajor Eigen tensors.
-     * If "dim.size()" &gt; 0, "unknown_rank" must be false.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorShapeProto.Dim dim = 2;</code>
-     */
-    public Builder addDim(
-        int index, org.tensorflow.proto.framework.TensorShapeProto.Dim.Builder builderForValue) {
-      if (dimBuilder_ == null) {
-        ensureDimIsMutable();
-        dim_.add(index, builderForValue.build());
-        onChanged();
-      } else {
-        dimBuilder_.addMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Dimensions of the tensor, such as {"input", 30}, {"output", 40}
-     * for a 30 x 40 2D tensor.  If an entry has size -1, this
-     * corresponds to a dimension of unknown size. The names are
-     * optional.
-     * The order of entries in "dim" matters: It indicates the layout of the
-     * values in the tensor in-memory representation.
-     * The first entry in "dim" is the outermost dimension used to layout the
-     * values, the last entry is the innermost dimension.  This matches the
-     * in-memory layout of RowMajor Eigen tensors.
-     * If "dim.size()" &gt; 0, "unknown_rank" must be false.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorShapeProto.Dim dim = 2;</code>
-     */
-    public Builder addAllDim(
-        java.lang.Iterable<? extends org.tensorflow.proto.framework.TensorShapeProto.Dim> values) {
-      if (dimBuilder_ == null) {
-        ensureDimIsMutable();
-        com.google.protobuf.AbstractMessageLite.Builder.addAll(
-            values, dim_);
-        onChanged();
-      } else {
-        dimBuilder_.addAllMessages(values);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Dimensions of the tensor, such as {"input", 30}, {"output", 40}
-     * for a 30 x 40 2D tensor.  If an entry has size -1, this
-     * corresponds to a dimension of unknown size. The names are
-     * optional.
-     * The order of entries in "dim" matters: It indicates the layout of the
-     * values in the tensor in-memory representation.
-     * The first entry in "dim" is the outermost dimension used to layout the
-     * values, the last entry is the innermost dimension.  This matches the
-     * in-memory layout of RowMajor Eigen tensors.
-     * If "dim.size()" &gt; 0, "unknown_rank" must be false.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorShapeProto.Dim dim = 2;</code>
-     */
-    public Builder clearDim() {
-      if (dimBuilder_ == null) {
-        dim_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000001);
-        onChanged();
-      } else {
-        dimBuilder_.clear();
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Dimensions of the tensor, such as {"input", 30}, {"output", 40}
-     * for a 30 x 40 2D tensor.  If an entry has size -1, this
-     * corresponds to a dimension of unknown size. The names are
-     * optional.
-     * The order of entries in "dim" matters: It indicates the layout of the
-     * values in the tensor in-memory representation.
-     * The first entry in "dim" is the outermost dimension used to layout the
-     * values, the last entry is the innermost dimension.  This matches the
-     * in-memory layout of RowMajor Eigen tensors.
-     * If "dim.size()" &gt; 0, "unknown_rank" must be false.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorShapeProto.Dim dim = 2;</code>
-     */
-    public Builder removeDim(int index) {
-      if (dimBuilder_ == null) {
-        ensureDimIsMutable();
-        dim_.remove(index);
-        onChanged();
-      } else {
-        dimBuilder_.remove(index);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Dimensions of the tensor, such as {"input", 30}, {"output", 40}
-     * for a 30 x 40 2D tensor.  If an entry has size -1, this
-     * corresponds to a dimension of unknown size. The names are
-     * optional.
-     * The order of entries in "dim" matters: It indicates the layout of the
-     * values in the tensor in-memory representation.
-     * The first entry in "dim" is the outermost dimension used to layout the
-     * values, the last entry is the innermost dimension.  This matches the
-     * in-memory layout of RowMajor Eigen tensors.
-     * If "dim.size()" &gt; 0, "unknown_rank" must be false.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorShapeProto.Dim dim = 2;</code>
-     */
-    public org.tensorflow.proto.framework.TensorShapeProto.Dim.Builder getDimBuilder(
-        int index) {
-      return getDimFieldBuilder().getBuilder(index);
-    }
-    /**
-     * <pre>
-     * Dimensions of the tensor, such as {"input", 30}, {"output", 40}
-     * for a 30 x 40 2D tensor.  If an entry has size -1, this
-     * corresponds to a dimension of unknown size. The names are
-     * optional.
-     * The order of entries in "dim" matters: It indicates the layout of the
-     * values in the tensor in-memory representation.
-     * The first entry in "dim" is the outermost dimension used to layout the
-     * values, the last entry is the innermost dimension.  This matches the
-     * in-memory layout of RowMajor Eigen tensors.
-     * If "dim.size()" &gt; 0, "unknown_rank" must be false.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorShapeProto.Dim dim = 2;</code>
-     */
-    public org.tensorflow.proto.framework.TensorShapeProto.DimOrBuilder getDimOrBuilder(
-        int index) {
-      if (dimBuilder_ == null) {
-        return dim_.get(index);  } else {
-        return dimBuilder_.getMessageOrBuilder(index);
-      }
-    }
-    /**
-     * <pre>
-     * Dimensions of the tensor, such as {"input", 30}, {"output", 40}
-     * for a 30 x 40 2D tensor.  If an entry has size -1, this
-     * corresponds to a dimension of unknown size. The names are
-     * optional.
-     * The order of entries in "dim" matters: It indicates the layout of the
-     * values in the tensor in-memory representation.
-     * The first entry in "dim" is the outermost dimension used to layout the
-     * values, the last entry is the innermost dimension.  This matches the
-     * in-memory layout of RowMajor Eigen tensors.
-     * If "dim.size()" &gt; 0, "unknown_rank" must be false.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorShapeProto.Dim dim = 2;</code>
-     */
-    public java.util.List<? extends org.tensorflow.proto.framework.TensorShapeProto.DimOrBuilder> 
-         getDimOrBuilderList() {
-      if (dimBuilder_ != null) {
-        return dimBuilder_.getMessageOrBuilderList();
-      } else {
-        return java.util.Collections.unmodifiableList(dim_);
-      }
-    }
-    /**
-     * <pre>
-     * Dimensions of the tensor, such as {"input", 30}, {"output", 40}
-     * for a 30 x 40 2D tensor.  If an entry has size -1, this
-     * corresponds to a dimension of unknown size. The names are
-     * optional.
-     * The order of entries in "dim" matters: It indicates the layout of the
-     * values in the tensor in-memory representation.
-     * The first entry in "dim" is the outermost dimension used to layout the
-     * values, the last entry is the innermost dimension.  This matches the
-     * in-memory layout of RowMajor Eigen tensors.
-     * If "dim.size()" &gt; 0, "unknown_rank" must be false.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorShapeProto.Dim dim = 2;</code>
-     */
-    public org.tensorflow.proto.framework.TensorShapeProto.Dim.Builder addDimBuilder() {
-      return getDimFieldBuilder().addBuilder(
-          org.tensorflow.proto.framework.TensorShapeProto.Dim.getDefaultInstance());
-    }
-    /**
-     * <pre>
-     * Dimensions of the tensor, such as {"input", 30}, {"output", 40}
-     * for a 30 x 40 2D tensor.  If an entry has size -1, this
-     * corresponds to a dimension of unknown size. The names are
-     * optional.
-     * The order of entries in "dim" matters: It indicates the layout of the
-     * values in the tensor in-memory representation.
-     * The first entry in "dim" is the outermost dimension used to layout the
-     * values, the last entry is the innermost dimension.  This matches the
-     * in-memory layout of RowMajor Eigen tensors.
-     * If "dim.size()" &gt; 0, "unknown_rank" must be false.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorShapeProto.Dim dim = 2;</code>
-     */
-    public org.tensorflow.proto.framework.TensorShapeProto.Dim.Builder addDimBuilder(
-        int index) {
-      return getDimFieldBuilder().addBuilder(
-          index, org.tensorflow.proto.framework.TensorShapeProto.Dim.getDefaultInstance());
-    }
-    /**
-     * <pre>
-     * Dimensions of the tensor, such as {"input", 30}, {"output", 40}
-     * for a 30 x 40 2D tensor.  If an entry has size -1, this
-     * corresponds to a dimension of unknown size. The names are
-     * optional.
-     * The order of entries in "dim" matters: It indicates the layout of the
-     * values in the tensor in-memory representation.
-     * The first entry in "dim" is the outermost dimension used to layout the
-     * values, the last entry is the innermost dimension.  This matches the
-     * in-memory layout of RowMajor Eigen tensors.
-     * If "dim.size()" &gt; 0, "unknown_rank" must be false.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorShapeProto.Dim dim = 2;</code>
-     */
-    public java.util.List<org.tensorflow.proto.framework.TensorShapeProto.Dim.Builder> 
-         getDimBuilderList() {
-      return getDimFieldBuilder().getBuilderList();
-    }
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.framework.TensorShapeProto.Dim, org.tensorflow.proto.framework.TensorShapeProto.Dim.Builder, org.tensorflow.proto.framework.TensorShapeProto.DimOrBuilder> 
-        getDimFieldBuilder() {
-      if (dimBuilder_ == null) {
-        dimBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
-            org.tensorflow.proto.framework.TensorShapeProto.Dim, org.tensorflow.proto.framework.TensorShapeProto.Dim.Builder, org.tensorflow.proto.framework.TensorShapeProto.DimOrBuilder>(
-                dim_,
-                ((bitField0_ & 0x00000001) != 0),
-                getParentForChildren(),
-                isClean());
-        dim_ = null;
-      }
-      return dimBuilder_;
-    }
-
-    private boolean unknownRank_ ;
-    /**
-     * <pre>
-     * If true, the number of dimensions in the shape is unknown.
-     * If true, "dim.size()" must be 0.
-     * </pre>
-     *
-     * <code>bool unknown_rank = 3;</code>
-     */
-    public boolean getUnknownRank() {
-      return unknownRank_;
-    }
-    /**
-     * <pre>
-     * If true, the number of dimensions in the shape is unknown.
-     * If true, "dim.size()" must be 0.
-     * </pre>
-     *
-     * <code>bool unknown_rank = 3;</code>
-     */
-    public Builder setUnknownRank(boolean value) {
-      
-      unknownRank_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * If true, the number of dimensions in the shape is unknown.
-     * If true, "dim.size()" must be 0.
-     * </pre>
-     *
-     * <code>bool unknown_rank = 3;</code>
-     */
-    public Builder clearUnknownRank() {
-      
-      unknownRank_ = false;
-      onChanged();
-      return this;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.TensorShapeProto)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.TensorShapeProto)
-  private static final org.tensorflow.proto.framework.TensorShapeProto DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.TensorShapeProto();
-  }
-
-  public static org.tensorflow.proto.framework.TensorShapeProto getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<TensorShapeProto>
-      PARSER = new com.google.protobuf.AbstractParser<TensorShapeProto>() {
-    @java.lang.Override
-    public TensorShapeProto parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new TensorShapeProto(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<TensorShapeProto> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<TensorShapeProto> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.framework.TensorShapeProto getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorShapeProtoOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorShapeProtoOrBuilder.java
deleted file mode 100644
index 7835368954b..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorShapeProtoOrBuilder.java
+++ /dev/null
@@ -1,108 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/framework/tensor_shape.proto
-
-package org.tensorflow.proto.framework;
-
-public interface TensorShapeProtoOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.TensorShapeProto)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <pre>
-   * Dimensions of the tensor, such as {"input", 30}, {"output", 40}
-   * for a 30 x 40 2D tensor.  If an entry has size -1, this
-   * corresponds to a dimension of unknown size. The names are
-   * optional.
-   * The order of entries in "dim" matters: It indicates the layout of the
-   * values in the tensor in-memory representation.
-   * The first entry in "dim" is the outermost dimension used to layout the
-   * values, the last entry is the innermost dimension.  This matches the
-   * in-memory layout of RowMajor Eigen tensors.
-   * If "dim.size()" &gt; 0, "unknown_rank" must be false.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.TensorShapeProto.Dim dim = 2;</code>
-   */
-  java.util.List<org.tensorflow.proto.framework.TensorShapeProto.Dim> 
-      getDimList();
-  /**
-   * <pre>
-   * Dimensions of the tensor, such as {"input", 30}, {"output", 40}
-   * for a 30 x 40 2D tensor.  If an entry has size -1, this
-   * corresponds to a dimension of unknown size. The names are
-   * optional.
-   * The order of entries in "dim" matters: It indicates the layout of the
-   * values in the tensor in-memory representation.
-   * The first entry in "dim" is the outermost dimension used to layout the
-   * values, the last entry is the innermost dimension.  This matches the
-   * in-memory layout of RowMajor Eigen tensors.
-   * If "dim.size()" &gt; 0, "unknown_rank" must be false.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.TensorShapeProto.Dim dim = 2;</code>
-   */
-  org.tensorflow.proto.framework.TensorShapeProto.Dim getDim(int index);
-  /**
-   * <pre>
-   * Dimensions of the tensor, such as {"input", 30}, {"output", 40}
-   * for a 30 x 40 2D tensor.  If an entry has size -1, this
-   * corresponds to a dimension of unknown size. The names are
-   * optional.
-   * The order of entries in "dim" matters: It indicates the layout of the
-   * values in the tensor in-memory representation.
-   * The first entry in "dim" is the outermost dimension used to layout the
-   * values, the last entry is the innermost dimension.  This matches the
-   * in-memory layout of RowMajor Eigen tensors.
-   * If "dim.size()" &gt; 0, "unknown_rank" must be false.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.TensorShapeProto.Dim dim = 2;</code>
-   */
-  int getDimCount();
-  /**
-   * <pre>
-   * Dimensions of the tensor, such as {"input", 30}, {"output", 40}
-   * for a 30 x 40 2D tensor.  If an entry has size -1, this
-   * corresponds to a dimension of unknown size. The names are
-   * optional.
-   * The order of entries in "dim" matters: It indicates the layout of the
-   * values in the tensor in-memory representation.
-   * The first entry in "dim" is the outermost dimension used to layout the
-   * values, the last entry is the innermost dimension.  This matches the
-   * in-memory layout of RowMajor Eigen tensors.
-   * If "dim.size()" &gt; 0, "unknown_rank" must be false.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.TensorShapeProto.Dim dim = 2;</code>
-   */
-  java.util.List<? extends org.tensorflow.proto.framework.TensorShapeProto.DimOrBuilder> 
-      getDimOrBuilderList();
-  /**
-   * <pre>
-   * Dimensions of the tensor, such as {"input", 30}, {"output", 40}
-   * for a 30 x 40 2D tensor.  If an entry has size -1, this
-   * corresponds to a dimension of unknown size. The names are
-   * optional.
-   * The order of entries in "dim" matters: It indicates the layout of the
-   * values in the tensor in-memory representation.
-   * The first entry in "dim" is the outermost dimension used to layout the
-   * values, the last entry is the innermost dimension.  This matches the
-   * in-memory layout of RowMajor Eigen tensors.
-   * If "dim.size()" &gt; 0, "unknown_rank" must be false.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.TensorShapeProto.Dim dim = 2;</code>
-   */
-  org.tensorflow.proto.framework.TensorShapeProto.DimOrBuilder getDimOrBuilder(
-      int index);
-
-  /**
-   * <pre>
-   * If true, the number of dimensions in the shape is unknown.
-   * If true, "dim.size()" must be 0.
-   * </pre>
-   *
-   * <code>bool unknown_rank = 3;</code>
-   */
-  boolean getUnknownRank();
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorShapeProtos.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorShapeProtos.java
deleted file mode 100644
index e8214c97ec1..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorShapeProtos.java
+++ /dev/null
@@ -1,65 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/framework/tensor_shape.proto
-
-package org.tensorflow.proto.framework;
-
-public final class TensorShapeProtos {
-  private TensorShapeProtos() {}
-  public static void registerAllExtensions(
-      com.google.protobuf.ExtensionRegistryLite registry) {
-  }
-
-  public static void registerAllExtensions(
-      com.google.protobuf.ExtensionRegistry registry) {
-    registerAllExtensions(
-        (com.google.protobuf.ExtensionRegistryLite) registry);
-  }
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_TensorShapeProto_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_TensorShapeProto_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_TensorShapeProto_Dim_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_TensorShapeProto_Dim_fieldAccessorTable;
-
-  public static com.google.protobuf.Descriptors.FileDescriptor
-      getDescriptor() {
-    return descriptor;
-  }
-  private static  com.google.protobuf.Descriptors.FileDescriptor
-      descriptor;
-  static {
-    java.lang.String[] descriptorData = {
-      "\n,tensorflow/core/framework/tensor_shape" +
-      ".proto\022\ntensorflow\"z\n\020TensorShapeProto\022-" +
-      "\n\003dim\030\002 \003(\0132 .tensorflow.TensorShapeProt" +
-      "o.Dim\022\024\n\014unknown_rank\030\003 \001(\010\032!\n\003Dim\022\014\n\004si" +
-      "ze\030\001 \001(\003\022\014\n\004name\030\002 \001(\tB\215\001\n\036org.tensorflo" +
-      "w.proto.frameworkB\021TensorShapeProtosP\001ZS" +
-      "github.com/tensorflow/tensorflow/tensorf" +
-      "low/go/core/framework/tensor_shape_go_pr" +
-      "oto\370\001\001b\006proto3"
-    };
-    descriptor = com.google.protobuf.Descriptors.FileDescriptor
-      .internalBuildGeneratedFileFrom(descriptorData,
-        new com.google.protobuf.Descriptors.FileDescriptor[] {
-        });
-    internal_static_tensorflow_TensorShapeProto_descriptor =
-      getDescriptor().getMessageTypes().get(0);
-    internal_static_tensorflow_TensorShapeProto_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_TensorShapeProto_descriptor,
-        new java.lang.String[] { "Dim", "UnknownRank", });
-    internal_static_tensorflow_TensorShapeProto_Dim_descriptor =
-      internal_static_tensorflow_TensorShapeProto_descriptor.getNestedTypes().get(0);
-    internal_static_tensorflow_TensorShapeProto_Dim_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_TensorShapeProto_Dim_descriptor,
-        new java.lang.String[] { "Size", "Name", });
-  }
-
-  // @@protoc_insertion_point(outer_class_scope)
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorSliceProto.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorSliceProto.java
deleted file mode 100644
index f1f7b1cf89d..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorSliceProto.java
+++ /dev/null
@@ -1,1589 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/framework/tensor_slice.proto
-
-package org.tensorflow.proto.framework;
-
-/**
- * <pre>
- * Can only be interpreted if you know the corresponding TensorShape.
- * </pre>
- *
- * Protobuf type {@code tensorflow.TensorSliceProto}
- */
-public  final class TensorSliceProto extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.TensorSliceProto)
-    TensorSliceProtoOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use TensorSliceProto.newBuilder() to construct.
-  private TensorSliceProto(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private TensorSliceProto() {
-    extent_ = java.util.Collections.emptyList();
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new TensorSliceProto();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private TensorSliceProto(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    int mutable_bitField0_ = 0;
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 10: {
-            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
-              extent_ = new java.util.ArrayList<org.tensorflow.proto.framework.TensorSliceProto.Extent>();
-              mutable_bitField0_ |= 0x00000001;
-            }
-            extent_.add(
-                input.readMessage(org.tensorflow.proto.framework.TensorSliceProto.Extent.parser(), extensionRegistry));
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      if (((mutable_bitField0_ & 0x00000001) != 0)) {
-        extent_ = java.util.Collections.unmodifiableList(extent_);
-      }
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.framework.TensorSliceProtos.internal_static_tensorflow_TensorSliceProto_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.framework.TensorSliceProtos.internal_static_tensorflow_TensorSliceProto_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.framework.TensorSliceProto.class, org.tensorflow.proto.framework.TensorSliceProto.Builder.class);
-  }
-
-  public interface ExtentOrBuilder extends
-      // @@protoc_insertion_point(interface_extends:tensorflow.TensorSliceProto.Extent)
-      com.google.protobuf.MessageOrBuilder {
-
-    /**
-     * <pre>
-     * Start index of the slice, starting at 0.
-     * </pre>
-     *
-     * <code>int64 start = 1;</code>
-     */
-    long getStart();
-
-    /**
-     * <code>int64 length = 2;</code>
-     */
-    long getLength();
-
-    public org.tensorflow.proto.framework.TensorSliceProto.Extent.HasLengthCase getHasLengthCase();
-  }
-  /**
-   * <pre>
-   * Extent of the slice in one dimension.
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.TensorSliceProto.Extent}
-   */
-  public  static final class Extent extends
-      com.google.protobuf.GeneratedMessageV3 implements
-      // @@protoc_insertion_point(message_implements:tensorflow.TensorSliceProto.Extent)
-      ExtentOrBuilder {
-  private static final long serialVersionUID = 0L;
-    // Use Extent.newBuilder() to construct.
-    private Extent(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-      super(builder);
-    }
-    private Extent() {
-    }
-
-    @java.lang.Override
-    @SuppressWarnings({"unused"})
-    protected java.lang.Object newInstance(
-        UnusedPrivateParameter unused) {
-      return new Extent();
-    }
-
-    @java.lang.Override
-    public final com.google.protobuf.UnknownFieldSet
-    getUnknownFields() {
-      return this.unknownFields;
-    }
-    private Extent(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      this();
-      if (extensionRegistry == null) {
-        throw new java.lang.NullPointerException();
-      }
-      com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-          com.google.protobuf.UnknownFieldSet.newBuilder();
-      try {
-        boolean done = false;
-        while (!done) {
-          int tag = input.readTag();
-          switch (tag) {
-            case 0:
-              done = true;
-              break;
-            case 8: {
-
-              start_ = input.readInt64();
-              break;
-            }
-            case 16: {
-              hasLengthCase_ = 2;
-              hasLength_ = input.readInt64();
-              break;
-            }
-            default: {
-              if (!parseUnknownField(
-                  input, unknownFields, extensionRegistry, tag)) {
-                done = true;
-              }
-              break;
-            }
-          }
-        }
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        throw e.setUnfinishedMessage(this);
-      } catch (java.io.IOException e) {
-        throw new com.google.protobuf.InvalidProtocolBufferException(
-            e).setUnfinishedMessage(this);
-      } finally {
-        this.unknownFields = unknownFields.build();
-        makeExtensionsImmutable();
-      }
-    }
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.TensorSliceProtos.internal_static_tensorflow_TensorSliceProto_Extent_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.TensorSliceProtos.internal_static_tensorflow_TensorSliceProto_Extent_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.TensorSliceProto.Extent.class, org.tensorflow.proto.framework.TensorSliceProto.Extent.Builder.class);
-    }
-
-    private int hasLengthCase_ = 0;
-    private java.lang.Object hasLength_;
-    public enum HasLengthCase
-        implements com.google.protobuf.Internal.EnumLite {
-      LENGTH(2),
-      HASLENGTH_NOT_SET(0);
-      private final int value;
-      private HasLengthCase(int value) {
-        this.value = value;
-      }
-      /**
-       * @deprecated Use {@link #forNumber(int)} instead.
-       */
-      @java.lang.Deprecated
-      public static HasLengthCase valueOf(int value) {
-        return forNumber(value);
-      }
-
-      public static HasLengthCase forNumber(int value) {
-        switch (value) {
-          case 2: return LENGTH;
-          case 0: return HASLENGTH_NOT_SET;
-          default: return null;
-        }
-      }
-      public int getNumber() {
-        return this.value;
-      }
-    };
-
-    public HasLengthCase
-    getHasLengthCase() {
-      return HasLengthCase.forNumber(
-          hasLengthCase_);
-    }
-
-    public static final int START_FIELD_NUMBER = 1;
-    private long start_;
-    /**
-     * <pre>
-     * Start index of the slice, starting at 0.
-     * </pre>
-     *
-     * <code>int64 start = 1;</code>
-     */
-    public long getStart() {
-      return start_;
-    }
-
-    public static final int LENGTH_FIELD_NUMBER = 2;
-    /**
-     * <code>int64 length = 2;</code>
-     */
-    public long getLength() {
-      if (hasLengthCase_ == 2) {
-        return (java.lang.Long) hasLength_;
-      }
-      return 0L;
-    }
-
-    private byte memoizedIsInitialized = -1;
-    @java.lang.Override
-    public final boolean isInitialized() {
-      byte isInitialized = memoizedIsInitialized;
-      if (isInitialized == 1) return true;
-      if (isInitialized == 0) return false;
-
-      memoizedIsInitialized = 1;
-      return true;
-    }
-
-    @java.lang.Override
-    public void writeTo(com.google.protobuf.CodedOutputStream output)
-                        throws java.io.IOException {
-      if (start_ != 0L) {
-        output.writeInt64(1, start_);
-      }
-      if (hasLengthCase_ == 2) {
-        output.writeInt64(
-            2, (long)((java.lang.Long) hasLength_));
-      }
-      unknownFields.writeTo(output);
-    }
-
-    @java.lang.Override
-    public int getSerializedSize() {
-      int size = memoizedSize;
-      if (size != -1) return size;
-
-      size = 0;
-      if (start_ != 0L) {
-        size += com.google.protobuf.CodedOutputStream
-          .computeInt64Size(1, start_);
-      }
-      if (hasLengthCase_ == 2) {
-        size += com.google.protobuf.CodedOutputStream
-          .computeInt64Size(
-              2, (long)((java.lang.Long) hasLength_));
-      }
-      size += unknownFields.getSerializedSize();
-      memoizedSize = size;
-      return size;
-    }
-
-    @java.lang.Override
-    public boolean equals(final java.lang.Object obj) {
-      if (obj == this) {
-       return true;
-      }
-      if (!(obj instanceof org.tensorflow.proto.framework.TensorSliceProto.Extent)) {
-        return super.equals(obj);
-      }
-      org.tensorflow.proto.framework.TensorSliceProto.Extent other = (org.tensorflow.proto.framework.TensorSliceProto.Extent) obj;
-
-      if (getStart()
-          != other.getStart()) return false;
-      if (!getHasLengthCase().equals(other.getHasLengthCase())) return false;
-      switch (hasLengthCase_) {
-        case 2:
-          if (getLength()
-              != other.getLength()) return false;
-          break;
-        case 0:
-        default:
-      }
-      if (!unknownFields.equals(other.unknownFields)) return false;
-      return true;
-    }
-
-    @java.lang.Override
-    public int hashCode() {
-      if (memoizedHashCode != 0) {
-        return memoizedHashCode;
-      }
-      int hash = 41;
-      hash = (19 * hash) + getDescriptor().hashCode();
-      hash = (37 * hash) + START_FIELD_NUMBER;
-      hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-          getStart());
-      switch (hasLengthCase_) {
-        case 2:
-          hash = (37 * hash) + LENGTH_FIELD_NUMBER;
-          hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-              getLength());
-          break;
-        case 0:
-        default:
-      }
-      hash = (29 * hash) + unknownFields.hashCode();
-      memoizedHashCode = hash;
-      return hash;
-    }
-
-    public static org.tensorflow.proto.framework.TensorSliceProto.Extent parseFrom(
-        java.nio.ByteBuffer data)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data);
-    }
-    public static org.tensorflow.proto.framework.TensorSliceProto.Extent parseFrom(
-        java.nio.ByteBuffer data,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.TensorSliceProto.Extent parseFrom(
-        com.google.protobuf.ByteString data)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data);
-    }
-    public static org.tensorflow.proto.framework.TensorSliceProto.Extent parseFrom(
-        com.google.protobuf.ByteString data,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.TensorSliceProto.Extent parseFrom(byte[] data)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data);
-    }
-    public static org.tensorflow.proto.framework.TensorSliceProto.Extent parseFrom(
-        byte[] data,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.TensorSliceProto.Extent parseFrom(java.io.InputStream input)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input);
-    }
-    public static org.tensorflow.proto.framework.TensorSliceProto.Extent parseFrom(
-        java.io.InputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.TensorSliceProto.Extent parseDelimitedFrom(java.io.InputStream input)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseDelimitedWithIOException(PARSER, input);
-    }
-    public static org.tensorflow.proto.framework.TensorSliceProto.Extent parseDelimitedFrom(
-        java.io.InputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.TensorSliceProto.Extent parseFrom(
-        com.google.protobuf.CodedInputStream input)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input);
-    }
-    public static org.tensorflow.proto.framework.TensorSliceProto.Extent parseFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input, extensionRegistry);
-    }
-
-    @java.lang.Override
-    public Builder newBuilderForType() { return newBuilder(); }
-    public static Builder newBuilder() {
-      return DEFAULT_INSTANCE.toBuilder();
-    }
-    public static Builder newBuilder(org.tensorflow.proto.framework.TensorSliceProto.Extent prototype) {
-      return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-    }
-    @java.lang.Override
-    public Builder toBuilder() {
-      return this == DEFAULT_INSTANCE
-          ? new Builder() : new Builder().mergeFrom(this);
-    }
-
-    @java.lang.Override
-    protected Builder newBuilderForType(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      Builder builder = new Builder(parent);
-      return builder;
-    }
-    /**
-     * <pre>
-     * Extent of the slice in one dimension.
-     * </pre>
-     *
-     * Protobuf type {@code tensorflow.TensorSliceProto.Extent}
-     */
-    public static final class Builder extends
-        com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-        // @@protoc_insertion_point(builder_implements:tensorflow.TensorSliceProto.Extent)
-        org.tensorflow.proto.framework.TensorSliceProto.ExtentOrBuilder {
-      public static final com.google.protobuf.Descriptors.Descriptor
-          getDescriptor() {
-        return org.tensorflow.proto.framework.TensorSliceProtos.internal_static_tensorflow_TensorSliceProto_Extent_descriptor;
-      }
-
-      @java.lang.Override
-      protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-          internalGetFieldAccessorTable() {
-        return org.tensorflow.proto.framework.TensorSliceProtos.internal_static_tensorflow_TensorSliceProto_Extent_fieldAccessorTable
-            .ensureFieldAccessorsInitialized(
-                org.tensorflow.proto.framework.TensorSliceProto.Extent.class, org.tensorflow.proto.framework.TensorSliceProto.Extent.Builder.class);
-      }
-
-      // Construct using org.tensorflow.proto.framework.TensorSliceProto.Extent.newBuilder()
-      private Builder() {
-        maybeForceBuilderInitialization();
-      }
-
-      private Builder(
-          com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-        super(parent);
-        maybeForceBuilderInitialization();
-      }
-      private void maybeForceBuilderInitialization() {
-        if (com.google.protobuf.GeneratedMessageV3
-                .alwaysUseFieldBuilders) {
-        }
-      }
-      @java.lang.Override
-      public Builder clear() {
-        super.clear();
-        start_ = 0L;
-
-        hasLengthCase_ = 0;
-        hasLength_ = null;
-        return this;
-      }
-
-      @java.lang.Override
-      public com.google.protobuf.Descriptors.Descriptor
-          getDescriptorForType() {
-        return org.tensorflow.proto.framework.TensorSliceProtos.internal_static_tensorflow_TensorSliceProto_Extent_descriptor;
-      }
-
-      @java.lang.Override
-      public org.tensorflow.proto.framework.TensorSliceProto.Extent getDefaultInstanceForType() {
-        return org.tensorflow.proto.framework.TensorSliceProto.Extent.getDefaultInstance();
-      }
-
-      @java.lang.Override
-      public org.tensorflow.proto.framework.TensorSliceProto.Extent build() {
-        org.tensorflow.proto.framework.TensorSliceProto.Extent result = buildPartial();
-        if (!result.isInitialized()) {
-          throw newUninitializedMessageException(result);
-        }
-        return result;
-      }
-
-      @java.lang.Override
-      public org.tensorflow.proto.framework.TensorSliceProto.Extent buildPartial() {
-        org.tensorflow.proto.framework.TensorSliceProto.Extent result = new org.tensorflow.proto.framework.TensorSliceProto.Extent(this);
-        result.start_ = start_;
-        if (hasLengthCase_ == 2) {
-          result.hasLength_ = hasLength_;
-        }
-        result.hasLengthCase_ = hasLengthCase_;
-        onBuilt();
-        return result;
-      }
-
-      @java.lang.Override
-      public Builder clone() {
-        return super.clone();
-      }
-      @java.lang.Override
-      public Builder setField(
-          com.google.protobuf.Descriptors.FieldDescriptor field,
-          java.lang.Object value) {
-        return super.setField(field, value);
-      }
-      @java.lang.Override
-      public Builder clearField(
-          com.google.protobuf.Descriptors.FieldDescriptor field) {
-        return super.clearField(field);
-      }
-      @java.lang.Override
-      public Builder clearOneof(
-          com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-        return super.clearOneof(oneof);
-      }
-      @java.lang.Override
-      public Builder setRepeatedField(
-          com.google.protobuf.Descriptors.FieldDescriptor field,
-          int index, java.lang.Object value) {
-        return super.setRepeatedField(field, index, value);
-      }
-      @java.lang.Override
-      public Builder addRepeatedField(
-          com.google.protobuf.Descriptors.FieldDescriptor field,
-          java.lang.Object value) {
-        return super.addRepeatedField(field, value);
-      }
-      @java.lang.Override
-      public Builder mergeFrom(com.google.protobuf.Message other) {
-        if (other instanceof org.tensorflow.proto.framework.TensorSliceProto.Extent) {
-          return mergeFrom((org.tensorflow.proto.framework.TensorSliceProto.Extent)other);
-        } else {
-          super.mergeFrom(other);
-          return this;
-        }
-      }
-
-      public Builder mergeFrom(org.tensorflow.proto.framework.TensorSliceProto.Extent other) {
-        if (other == org.tensorflow.proto.framework.TensorSliceProto.Extent.getDefaultInstance()) return this;
-        if (other.getStart() != 0L) {
-          setStart(other.getStart());
-        }
-        switch (other.getHasLengthCase()) {
-          case LENGTH: {
-            setLength(other.getLength());
-            break;
-          }
-          case HASLENGTH_NOT_SET: {
-            break;
-          }
-        }
-        this.mergeUnknownFields(other.unknownFields);
-        onChanged();
-        return this;
-      }
-
-      @java.lang.Override
-      public final boolean isInitialized() {
-        return true;
-      }
-
-      @java.lang.Override
-      public Builder mergeFrom(
-          com.google.protobuf.CodedInputStream input,
-          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-          throws java.io.IOException {
-        org.tensorflow.proto.framework.TensorSliceProto.Extent parsedMessage = null;
-        try {
-          parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-        } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-          parsedMessage = (org.tensorflow.proto.framework.TensorSliceProto.Extent) e.getUnfinishedMessage();
-          throw e.unwrapIOException();
-        } finally {
-          if (parsedMessage != null) {
-            mergeFrom(parsedMessage);
-          }
-        }
-        return this;
-      }
-      private int hasLengthCase_ = 0;
-      private java.lang.Object hasLength_;
-      public HasLengthCase
-          getHasLengthCase() {
-        return HasLengthCase.forNumber(
-            hasLengthCase_);
-      }
-
-      public Builder clearHasLength() {
-        hasLengthCase_ = 0;
-        hasLength_ = null;
-        onChanged();
-        return this;
-      }
-
-
-      private long start_ ;
-      /**
-       * <pre>
-       * Start index of the slice, starting at 0.
-       * </pre>
-       *
-       * <code>int64 start = 1;</code>
-       */
-      public long getStart() {
-        return start_;
-      }
-      /**
-       * <pre>
-       * Start index of the slice, starting at 0.
-       * </pre>
-       *
-       * <code>int64 start = 1;</code>
-       */
-      public Builder setStart(long value) {
-        
-        start_ = value;
-        onChanged();
-        return this;
-      }
-      /**
-       * <pre>
-       * Start index of the slice, starting at 0.
-       * </pre>
-       *
-       * <code>int64 start = 1;</code>
-       */
-      public Builder clearStart() {
-        
-        start_ = 0L;
-        onChanged();
-        return this;
-      }
-
-      /**
-       * <code>int64 length = 2;</code>
-       */
-      public long getLength() {
-        if (hasLengthCase_ == 2) {
-          return (java.lang.Long) hasLength_;
-        }
-        return 0L;
-      }
-      /**
-       * <code>int64 length = 2;</code>
-       */
-      public Builder setLength(long value) {
-        hasLengthCase_ = 2;
-        hasLength_ = value;
-        onChanged();
-        return this;
-      }
-      /**
-       * <code>int64 length = 2;</code>
-       */
-      public Builder clearLength() {
-        if (hasLengthCase_ == 2) {
-          hasLengthCase_ = 0;
-          hasLength_ = null;
-          onChanged();
-        }
-        return this;
-      }
-      @java.lang.Override
-      public final Builder setUnknownFields(
-          final com.google.protobuf.UnknownFieldSet unknownFields) {
-        return super.setUnknownFields(unknownFields);
-      }
-
-      @java.lang.Override
-      public final Builder mergeUnknownFields(
-          final com.google.protobuf.UnknownFieldSet unknownFields) {
-        return super.mergeUnknownFields(unknownFields);
-      }
-
-
-      // @@protoc_insertion_point(builder_scope:tensorflow.TensorSliceProto.Extent)
-    }
-
-    // @@protoc_insertion_point(class_scope:tensorflow.TensorSliceProto.Extent)
-    private static final org.tensorflow.proto.framework.TensorSliceProto.Extent DEFAULT_INSTANCE;
-    static {
-      DEFAULT_INSTANCE = new org.tensorflow.proto.framework.TensorSliceProto.Extent();
-    }
-
-    public static org.tensorflow.proto.framework.TensorSliceProto.Extent getDefaultInstance() {
-      return DEFAULT_INSTANCE;
-    }
-
-    private static final com.google.protobuf.Parser<Extent>
-        PARSER = new com.google.protobuf.AbstractParser<Extent>() {
-      @java.lang.Override
-      public Extent parsePartialFrom(
-          com.google.protobuf.CodedInputStream input,
-          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-          throws com.google.protobuf.InvalidProtocolBufferException {
-        return new Extent(input, extensionRegistry);
-      }
-    };
-
-    public static com.google.protobuf.Parser<Extent> parser() {
-      return PARSER;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Parser<Extent> getParserForType() {
-      return PARSER;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.TensorSliceProto.Extent getDefaultInstanceForType() {
-      return DEFAULT_INSTANCE;
-    }
-
-  }
-
-  public static final int EXTENT_FIELD_NUMBER = 1;
-  private java.util.List<org.tensorflow.proto.framework.TensorSliceProto.Extent> extent_;
-  /**
-   * <pre>
-   * Extent of the slice in all tensor dimensions.
-   * Must have one entry for each of the dimension of the tensor that this
-   * slice belongs to.  The order of sizes is the same as the order of
-   * dimensions in the TensorShape.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.TensorSliceProto.Extent extent = 1;</code>
-   */
-  public java.util.List<org.tensorflow.proto.framework.TensorSliceProto.Extent> getExtentList() {
-    return extent_;
-  }
-  /**
-   * <pre>
-   * Extent of the slice in all tensor dimensions.
-   * Must have one entry for each of the dimension of the tensor that this
-   * slice belongs to.  The order of sizes is the same as the order of
-   * dimensions in the TensorShape.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.TensorSliceProto.Extent extent = 1;</code>
-   */
-  public java.util.List<? extends org.tensorflow.proto.framework.TensorSliceProto.ExtentOrBuilder> 
-      getExtentOrBuilderList() {
-    return extent_;
-  }
-  /**
-   * <pre>
-   * Extent of the slice in all tensor dimensions.
-   * Must have one entry for each of the dimension of the tensor that this
-   * slice belongs to.  The order of sizes is the same as the order of
-   * dimensions in the TensorShape.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.TensorSliceProto.Extent extent = 1;</code>
-   */
-  public int getExtentCount() {
-    return extent_.size();
-  }
-  /**
-   * <pre>
-   * Extent of the slice in all tensor dimensions.
-   * Must have one entry for each of the dimension of the tensor that this
-   * slice belongs to.  The order of sizes is the same as the order of
-   * dimensions in the TensorShape.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.TensorSliceProto.Extent extent = 1;</code>
-   */
-  public org.tensorflow.proto.framework.TensorSliceProto.Extent getExtent(int index) {
-    return extent_.get(index);
-  }
-  /**
-   * <pre>
-   * Extent of the slice in all tensor dimensions.
-   * Must have one entry for each of the dimension of the tensor that this
-   * slice belongs to.  The order of sizes is the same as the order of
-   * dimensions in the TensorShape.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.TensorSliceProto.Extent extent = 1;</code>
-   */
-  public org.tensorflow.proto.framework.TensorSliceProto.ExtentOrBuilder getExtentOrBuilder(
-      int index) {
-    return extent_.get(index);
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    for (int i = 0; i < extent_.size(); i++) {
-      output.writeMessage(1, extent_.get(i));
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    for (int i = 0; i < extent_.size(); i++) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(1, extent_.get(i));
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.framework.TensorSliceProto)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.framework.TensorSliceProto other = (org.tensorflow.proto.framework.TensorSliceProto) obj;
-
-    if (!getExtentList()
-        .equals(other.getExtentList())) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    if (getExtentCount() > 0) {
-      hash = (37 * hash) + EXTENT_FIELD_NUMBER;
-      hash = (53 * hash) + getExtentList().hashCode();
-    }
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.framework.TensorSliceProto parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.TensorSliceProto parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.TensorSliceProto parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.TensorSliceProto parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.TensorSliceProto parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.TensorSliceProto parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.TensorSliceProto parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.TensorSliceProto parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.TensorSliceProto parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.TensorSliceProto parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.TensorSliceProto parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.TensorSliceProto parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.framework.TensorSliceProto prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * <pre>
-   * Can only be interpreted if you know the corresponding TensorShape.
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.TensorSliceProto}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.TensorSliceProto)
-      org.tensorflow.proto.framework.TensorSliceProtoOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.TensorSliceProtos.internal_static_tensorflow_TensorSliceProto_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.TensorSliceProtos.internal_static_tensorflow_TensorSliceProto_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.TensorSliceProto.class, org.tensorflow.proto.framework.TensorSliceProto.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.framework.TensorSliceProto.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-        getExtentFieldBuilder();
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      if (extentBuilder_ == null) {
-        extent_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000001);
-      } else {
-        extentBuilder_.clear();
-      }
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.framework.TensorSliceProtos.internal_static_tensorflow_TensorSliceProto_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.TensorSliceProto getDefaultInstanceForType() {
-      return org.tensorflow.proto.framework.TensorSliceProto.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.TensorSliceProto build() {
-      org.tensorflow.proto.framework.TensorSliceProto result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.TensorSliceProto buildPartial() {
-      org.tensorflow.proto.framework.TensorSliceProto result = new org.tensorflow.proto.framework.TensorSliceProto(this);
-      int from_bitField0_ = bitField0_;
-      if (extentBuilder_ == null) {
-        if (((bitField0_ & 0x00000001) != 0)) {
-          extent_ = java.util.Collections.unmodifiableList(extent_);
-          bitField0_ = (bitField0_ & ~0x00000001);
-        }
-        result.extent_ = extent_;
-      } else {
-        result.extent_ = extentBuilder_.build();
-      }
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.framework.TensorSliceProto) {
-        return mergeFrom((org.tensorflow.proto.framework.TensorSliceProto)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.framework.TensorSliceProto other) {
-      if (other == org.tensorflow.proto.framework.TensorSliceProto.getDefaultInstance()) return this;
-      if (extentBuilder_ == null) {
-        if (!other.extent_.isEmpty()) {
-          if (extent_.isEmpty()) {
-            extent_ = other.extent_;
-            bitField0_ = (bitField0_ & ~0x00000001);
-          } else {
-            ensureExtentIsMutable();
-            extent_.addAll(other.extent_);
-          }
-          onChanged();
-        }
-      } else {
-        if (!other.extent_.isEmpty()) {
-          if (extentBuilder_.isEmpty()) {
-            extentBuilder_.dispose();
-            extentBuilder_ = null;
-            extent_ = other.extent_;
-            bitField0_ = (bitField0_ & ~0x00000001);
-            extentBuilder_ = 
-              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
-                 getExtentFieldBuilder() : null;
-          } else {
-            extentBuilder_.addAllMessages(other.extent_);
-          }
-        }
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.framework.TensorSliceProto parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.framework.TensorSliceProto) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-    private int bitField0_;
-
-    private java.util.List<org.tensorflow.proto.framework.TensorSliceProto.Extent> extent_ =
-      java.util.Collections.emptyList();
-    private void ensureExtentIsMutable() {
-      if (!((bitField0_ & 0x00000001) != 0)) {
-        extent_ = new java.util.ArrayList<org.tensorflow.proto.framework.TensorSliceProto.Extent>(extent_);
-        bitField0_ |= 0x00000001;
-       }
-    }
-
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.framework.TensorSliceProto.Extent, org.tensorflow.proto.framework.TensorSliceProto.Extent.Builder, org.tensorflow.proto.framework.TensorSliceProto.ExtentOrBuilder> extentBuilder_;
-
-    /**
-     * <pre>
-     * Extent of the slice in all tensor dimensions.
-     * Must have one entry for each of the dimension of the tensor that this
-     * slice belongs to.  The order of sizes is the same as the order of
-     * dimensions in the TensorShape.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorSliceProto.Extent extent = 1;</code>
-     */
-    public java.util.List<org.tensorflow.proto.framework.TensorSliceProto.Extent> getExtentList() {
-      if (extentBuilder_ == null) {
-        return java.util.Collections.unmodifiableList(extent_);
-      } else {
-        return extentBuilder_.getMessageList();
-      }
-    }
-    /**
-     * <pre>
-     * Extent of the slice in all tensor dimensions.
-     * Must have one entry for each of the dimension of the tensor that this
-     * slice belongs to.  The order of sizes is the same as the order of
-     * dimensions in the TensorShape.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorSliceProto.Extent extent = 1;</code>
-     */
-    public int getExtentCount() {
-      if (extentBuilder_ == null) {
-        return extent_.size();
-      } else {
-        return extentBuilder_.getCount();
-      }
-    }
-    /**
-     * <pre>
-     * Extent of the slice in all tensor dimensions.
-     * Must have one entry for each of the dimension of the tensor that this
-     * slice belongs to.  The order of sizes is the same as the order of
-     * dimensions in the TensorShape.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorSliceProto.Extent extent = 1;</code>
-     */
-    public org.tensorflow.proto.framework.TensorSliceProto.Extent getExtent(int index) {
-      if (extentBuilder_ == null) {
-        return extent_.get(index);
-      } else {
-        return extentBuilder_.getMessage(index);
-      }
-    }
-    /**
-     * <pre>
-     * Extent of the slice in all tensor dimensions.
-     * Must have one entry for each of the dimension of the tensor that this
-     * slice belongs to.  The order of sizes is the same as the order of
-     * dimensions in the TensorShape.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorSliceProto.Extent extent = 1;</code>
-     */
-    public Builder setExtent(
-        int index, org.tensorflow.proto.framework.TensorSliceProto.Extent value) {
-      if (extentBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureExtentIsMutable();
-        extent_.set(index, value);
-        onChanged();
-      } else {
-        extentBuilder_.setMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Extent of the slice in all tensor dimensions.
-     * Must have one entry for each of the dimension of the tensor that this
-     * slice belongs to.  The order of sizes is the same as the order of
-     * dimensions in the TensorShape.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorSliceProto.Extent extent = 1;</code>
-     */
-    public Builder setExtent(
-        int index, org.tensorflow.proto.framework.TensorSliceProto.Extent.Builder builderForValue) {
-      if (extentBuilder_ == null) {
-        ensureExtentIsMutable();
-        extent_.set(index, builderForValue.build());
-        onChanged();
-      } else {
-        extentBuilder_.setMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Extent of the slice in all tensor dimensions.
-     * Must have one entry for each of the dimension of the tensor that this
-     * slice belongs to.  The order of sizes is the same as the order of
-     * dimensions in the TensorShape.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorSliceProto.Extent extent = 1;</code>
-     */
-    public Builder addExtent(org.tensorflow.proto.framework.TensorSliceProto.Extent value) {
-      if (extentBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureExtentIsMutable();
-        extent_.add(value);
-        onChanged();
-      } else {
-        extentBuilder_.addMessage(value);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Extent of the slice in all tensor dimensions.
-     * Must have one entry for each of the dimension of the tensor that this
-     * slice belongs to.  The order of sizes is the same as the order of
-     * dimensions in the TensorShape.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorSliceProto.Extent extent = 1;</code>
-     */
-    public Builder addExtent(
-        int index, org.tensorflow.proto.framework.TensorSliceProto.Extent value) {
-      if (extentBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureExtentIsMutable();
-        extent_.add(index, value);
-        onChanged();
-      } else {
-        extentBuilder_.addMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Extent of the slice in all tensor dimensions.
-     * Must have one entry for each of the dimension of the tensor that this
-     * slice belongs to.  The order of sizes is the same as the order of
-     * dimensions in the TensorShape.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorSliceProto.Extent extent = 1;</code>
-     */
-    public Builder addExtent(
-        org.tensorflow.proto.framework.TensorSliceProto.Extent.Builder builderForValue) {
-      if (extentBuilder_ == null) {
-        ensureExtentIsMutable();
-        extent_.add(builderForValue.build());
-        onChanged();
-      } else {
-        extentBuilder_.addMessage(builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Extent of the slice in all tensor dimensions.
-     * Must have one entry for each of the dimension of the tensor that this
-     * slice belongs to.  The order of sizes is the same as the order of
-     * dimensions in the TensorShape.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorSliceProto.Extent extent = 1;</code>
-     */
-    public Builder addExtent(
-        int index, org.tensorflow.proto.framework.TensorSliceProto.Extent.Builder builderForValue) {
-      if (extentBuilder_ == null) {
-        ensureExtentIsMutable();
-        extent_.add(index, builderForValue.build());
-        onChanged();
-      } else {
-        extentBuilder_.addMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Extent of the slice in all tensor dimensions.
-     * Must have one entry for each of the dimension of the tensor that this
-     * slice belongs to.  The order of sizes is the same as the order of
-     * dimensions in the TensorShape.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorSliceProto.Extent extent = 1;</code>
-     */
-    public Builder addAllExtent(
-        java.lang.Iterable<? extends org.tensorflow.proto.framework.TensorSliceProto.Extent> values) {
-      if (extentBuilder_ == null) {
-        ensureExtentIsMutable();
-        com.google.protobuf.AbstractMessageLite.Builder.addAll(
-            values, extent_);
-        onChanged();
-      } else {
-        extentBuilder_.addAllMessages(values);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Extent of the slice in all tensor dimensions.
-     * Must have one entry for each of the dimension of the tensor that this
-     * slice belongs to.  The order of sizes is the same as the order of
-     * dimensions in the TensorShape.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorSliceProto.Extent extent = 1;</code>
-     */
-    public Builder clearExtent() {
-      if (extentBuilder_ == null) {
-        extent_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000001);
-        onChanged();
-      } else {
-        extentBuilder_.clear();
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Extent of the slice in all tensor dimensions.
-     * Must have one entry for each of the dimension of the tensor that this
-     * slice belongs to.  The order of sizes is the same as the order of
-     * dimensions in the TensorShape.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorSliceProto.Extent extent = 1;</code>
-     */
-    public Builder removeExtent(int index) {
-      if (extentBuilder_ == null) {
-        ensureExtentIsMutable();
-        extent_.remove(index);
-        onChanged();
-      } else {
-        extentBuilder_.remove(index);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Extent of the slice in all tensor dimensions.
-     * Must have one entry for each of the dimension of the tensor that this
-     * slice belongs to.  The order of sizes is the same as the order of
-     * dimensions in the TensorShape.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorSliceProto.Extent extent = 1;</code>
-     */
-    public org.tensorflow.proto.framework.TensorSliceProto.Extent.Builder getExtentBuilder(
-        int index) {
-      return getExtentFieldBuilder().getBuilder(index);
-    }
-    /**
-     * <pre>
-     * Extent of the slice in all tensor dimensions.
-     * Must have one entry for each of the dimension of the tensor that this
-     * slice belongs to.  The order of sizes is the same as the order of
-     * dimensions in the TensorShape.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorSliceProto.Extent extent = 1;</code>
-     */
-    public org.tensorflow.proto.framework.TensorSliceProto.ExtentOrBuilder getExtentOrBuilder(
-        int index) {
-      if (extentBuilder_ == null) {
-        return extent_.get(index);  } else {
-        return extentBuilder_.getMessageOrBuilder(index);
-      }
-    }
-    /**
-     * <pre>
-     * Extent of the slice in all tensor dimensions.
-     * Must have one entry for each of the dimension of the tensor that this
-     * slice belongs to.  The order of sizes is the same as the order of
-     * dimensions in the TensorShape.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorSliceProto.Extent extent = 1;</code>
-     */
-    public java.util.List<? extends org.tensorflow.proto.framework.TensorSliceProto.ExtentOrBuilder> 
-         getExtentOrBuilderList() {
-      if (extentBuilder_ != null) {
-        return extentBuilder_.getMessageOrBuilderList();
-      } else {
-        return java.util.Collections.unmodifiableList(extent_);
-      }
-    }
-    /**
-     * <pre>
-     * Extent of the slice in all tensor dimensions.
-     * Must have one entry for each of the dimension of the tensor that this
-     * slice belongs to.  The order of sizes is the same as the order of
-     * dimensions in the TensorShape.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorSliceProto.Extent extent = 1;</code>
-     */
-    public org.tensorflow.proto.framework.TensorSliceProto.Extent.Builder addExtentBuilder() {
-      return getExtentFieldBuilder().addBuilder(
-          org.tensorflow.proto.framework.TensorSliceProto.Extent.getDefaultInstance());
-    }
-    /**
-     * <pre>
-     * Extent of the slice in all tensor dimensions.
-     * Must have one entry for each of the dimension of the tensor that this
-     * slice belongs to.  The order of sizes is the same as the order of
-     * dimensions in the TensorShape.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorSliceProto.Extent extent = 1;</code>
-     */
-    public org.tensorflow.proto.framework.TensorSliceProto.Extent.Builder addExtentBuilder(
-        int index) {
-      return getExtentFieldBuilder().addBuilder(
-          index, org.tensorflow.proto.framework.TensorSliceProto.Extent.getDefaultInstance());
-    }
-    /**
-     * <pre>
-     * Extent of the slice in all tensor dimensions.
-     * Must have one entry for each of the dimension of the tensor that this
-     * slice belongs to.  The order of sizes is the same as the order of
-     * dimensions in the TensorShape.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorSliceProto.Extent extent = 1;</code>
-     */
-    public java.util.List<org.tensorflow.proto.framework.TensorSliceProto.Extent.Builder> 
-         getExtentBuilderList() {
-      return getExtentFieldBuilder().getBuilderList();
-    }
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.framework.TensorSliceProto.Extent, org.tensorflow.proto.framework.TensorSliceProto.Extent.Builder, org.tensorflow.proto.framework.TensorSliceProto.ExtentOrBuilder> 
-        getExtentFieldBuilder() {
-      if (extentBuilder_ == null) {
-        extentBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
-            org.tensorflow.proto.framework.TensorSliceProto.Extent, org.tensorflow.proto.framework.TensorSliceProto.Extent.Builder, org.tensorflow.proto.framework.TensorSliceProto.ExtentOrBuilder>(
-                extent_,
-                ((bitField0_ & 0x00000001) != 0),
-                getParentForChildren(),
-                isClean());
-        extent_ = null;
-      }
-      return extentBuilder_;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.TensorSliceProto)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.TensorSliceProto)
-  private static final org.tensorflow.proto.framework.TensorSliceProto DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.TensorSliceProto();
-  }
-
-  public static org.tensorflow.proto.framework.TensorSliceProto getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<TensorSliceProto>
-      PARSER = new com.google.protobuf.AbstractParser<TensorSliceProto>() {
-    @java.lang.Override
-    public TensorSliceProto parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new TensorSliceProto(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<TensorSliceProto> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<TensorSliceProto> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.framework.TensorSliceProto getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorSliceProtoOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorSliceProtoOrBuilder.java
deleted file mode 100644
index 187cbc4962f..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorSliceProtoOrBuilder.java
+++ /dev/null
@@ -1,68 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/framework/tensor_slice.proto
-
-package org.tensorflow.proto.framework;
-
-public interface TensorSliceProtoOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.TensorSliceProto)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <pre>
-   * Extent of the slice in all tensor dimensions.
-   * Must have one entry for each of the dimension of the tensor that this
-   * slice belongs to.  The order of sizes is the same as the order of
-   * dimensions in the TensorShape.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.TensorSliceProto.Extent extent = 1;</code>
-   */
-  java.util.List<org.tensorflow.proto.framework.TensorSliceProto.Extent> 
-      getExtentList();
-  /**
-   * <pre>
-   * Extent of the slice in all tensor dimensions.
-   * Must have one entry for each of the dimension of the tensor that this
-   * slice belongs to.  The order of sizes is the same as the order of
-   * dimensions in the TensorShape.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.TensorSliceProto.Extent extent = 1;</code>
-   */
-  org.tensorflow.proto.framework.TensorSliceProto.Extent getExtent(int index);
-  /**
-   * <pre>
-   * Extent of the slice in all tensor dimensions.
-   * Must have one entry for each of the dimension of the tensor that this
-   * slice belongs to.  The order of sizes is the same as the order of
-   * dimensions in the TensorShape.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.TensorSliceProto.Extent extent = 1;</code>
-   */
-  int getExtentCount();
-  /**
-   * <pre>
-   * Extent of the slice in all tensor dimensions.
-   * Must have one entry for each of the dimension of the tensor that this
-   * slice belongs to.  The order of sizes is the same as the order of
-   * dimensions in the TensorShape.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.TensorSliceProto.Extent extent = 1;</code>
-   */
-  java.util.List<? extends org.tensorflow.proto.framework.TensorSliceProto.ExtentOrBuilder> 
-      getExtentOrBuilderList();
-  /**
-   * <pre>
-   * Extent of the slice in all tensor dimensions.
-   * Must have one entry for each of the dimension of the tensor that this
-   * slice belongs to.  The order of sizes is the same as the order of
-   * dimensions in the TensorShape.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.TensorSliceProto.Extent extent = 1;</code>
-   */
-  org.tensorflow.proto.framework.TensorSliceProto.ExtentOrBuilder getExtentOrBuilder(
-      int index);
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorSliceProtos.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorSliceProtos.java
deleted file mode 100644
index 8a3b4497177..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorSliceProtos.java
+++ /dev/null
@@ -1,65 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/framework/tensor_slice.proto
-
-package org.tensorflow.proto.framework;
-
-public final class TensorSliceProtos {
-  private TensorSliceProtos() {}
-  public static void registerAllExtensions(
-      com.google.protobuf.ExtensionRegistryLite registry) {
-  }
-
-  public static void registerAllExtensions(
-      com.google.protobuf.ExtensionRegistry registry) {
-    registerAllExtensions(
-        (com.google.protobuf.ExtensionRegistryLite) registry);
-  }
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_TensorSliceProto_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_TensorSliceProto_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_TensorSliceProto_Extent_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_TensorSliceProto_Extent_fieldAccessorTable;
-
-  public static com.google.protobuf.Descriptors.FileDescriptor
-      getDescriptor() {
-    return descriptor;
-  }
-  private static  com.google.protobuf.Descriptors.FileDescriptor
-      descriptor;
-  static {
-    java.lang.String[] descriptorData = {
-      "\n,tensorflow/core/framework/tensor_slice" +
-      ".proto\022\ntensorflow\"\200\001\n\020TensorSliceProto\022" +
-      "3\n\006extent\030\001 \003(\0132#.tensorflow.TensorSlice" +
-      "Proto.Extent\0327\n\006Extent\022\r\n\005start\030\001 \001(\003\022\020\n" +
-      "\006length\030\002 \001(\003H\000B\014\n\nhas_lengthB\215\001\n\036org.te" +
-      "nsorflow.proto.frameworkB\021TensorSlicePro" +
-      "tosP\001ZSgithub.com/tensorflow/tensorflow/" +
-      "tensorflow/go/core/framework/tensor_slic" +
-      "e_go_proto\370\001\001b\006proto3"
-    };
-    descriptor = com.google.protobuf.Descriptors.FileDescriptor
-      .internalBuildGeneratedFileFrom(descriptorData,
-        new com.google.protobuf.Descriptors.FileDescriptor[] {
-        });
-    internal_static_tensorflow_TensorSliceProto_descriptor =
-      getDescriptor().getMessageTypes().get(0);
-    internal_static_tensorflow_TensorSliceProto_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_TensorSliceProto_descriptor,
-        new java.lang.String[] { "Extent", });
-    internal_static_tensorflow_TensorSliceProto_Extent_descriptor =
-      internal_static_tensorflow_TensorSliceProto_descriptor.getNestedTypes().get(0);
-    internal_static_tensorflow_TensorSliceProto_Extent_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_TensorSliceProto_Extent_descriptor,
-        new java.lang.String[] { "Start", "Length", "HasLength", });
-  }
-
-  // @@protoc_insertion_point(outer_class_scope)
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorSpecProto.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorSpecProto.java
deleted file mode 100644
index 1dadc728091..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorSpecProto.java
+++ /dev/null
@@ -1,820 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/struct.proto
-
-package org.tensorflow.proto.framework;
-
-/**
- * <pre>
- * A protobuf to represent tf.TensorSpec.
- * </pre>
- *
- * Protobuf type {@code tensorflow.TensorSpecProto}
- */
-public  final class TensorSpecProto extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.TensorSpecProto)
-    TensorSpecProtoOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use TensorSpecProto.newBuilder() to construct.
-  private TensorSpecProto(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private TensorSpecProto() {
-    name_ = "";
-    dtype_ = 0;
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new TensorSpecProto();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private TensorSpecProto(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 10: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            name_ = s;
-            break;
-          }
-          case 18: {
-            org.tensorflow.proto.framework.TensorShapeProto.Builder subBuilder = null;
-            if (shape_ != null) {
-              subBuilder = shape_.toBuilder();
-            }
-            shape_ = input.readMessage(org.tensorflow.proto.framework.TensorShapeProto.parser(), extensionRegistry);
-            if (subBuilder != null) {
-              subBuilder.mergeFrom(shape_);
-              shape_ = subBuilder.buildPartial();
-            }
-
-            break;
-          }
-          case 24: {
-            int rawValue = input.readEnum();
-
-            dtype_ = rawValue;
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.framework.StructProtos.internal_static_tensorflow_TensorSpecProto_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.framework.StructProtos.internal_static_tensorflow_TensorSpecProto_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.framework.TensorSpecProto.class, org.tensorflow.proto.framework.TensorSpecProto.Builder.class);
-  }
-
-  public static final int NAME_FIELD_NUMBER = 1;
-  private volatile java.lang.Object name_;
-  /**
-   * <code>string name = 1;</code>
-   */
-  public java.lang.String getName() {
-    java.lang.Object ref = name_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      name_ = s;
-      return s;
-    }
-  }
-  /**
-   * <code>string name = 1;</code>
-   */
-  public com.google.protobuf.ByteString
-      getNameBytes() {
-    java.lang.Object ref = name_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      name_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int SHAPE_FIELD_NUMBER = 2;
-  private org.tensorflow.proto.framework.TensorShapeProto shape_;
-  /**
-   * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-   */
-  public boolean hasShape() {
-    return shape_ != null;
-  }
-  /**
-   * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-   */
-  public org.tensorflow.proto.framework.TensorShapeProto getShape() {
-    return shape_ == null ? org.tensorflow.proto.framework.TensorShapeProto.getDefaultInstance() : shape_;
-  }
-  /**
-   * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-   */
-  public org.tensorflow.proto.framework.TensorShapeProtoOrBuilder getShapeOrBuilder() {
-    return getShape();
-  }
-
-  public static final int DTYPE_FIELD_NUMBER = 3;
-  private int dtype_;
-  /**
-   * <code>.tensorflow.DataType dtype = 3;</code>
-   */
-  public int getDtypeValue() {
-    return dtype_;
-  }
-  /**
-   * <code>.tensorflow.DataType dtype = 3;</code>
-   */
-  public org.tensorflow.proto.framework.DataType getDtype() {
-    @SuppressWarnings("deprecation")
-    org.tensorflow.proto.framework.DataType result = org.tensorflow.proto.framework.DataType.valueOf(dtype_);
-    return result == null ? org.tensorflow.proto.framework.DataType.UNRECOGNIZED : result;
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    if (!getNameBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 1, name_);
-    }
-    if (shape_ != null) {
-      output.writeMessage(2, getShape());
-    }
-    if (dtype_ != org.tensorflow.proto.framework.DataType.DT_INVALID.getNumber()) {
-      output.writeEnum(3, dtype_);
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (!getNameBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, name_);
-    }
-    if (shape_ != null) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(2, getShape());
-    }
-    if (dtype_ != org.tensorflow.proto.framework.DataType.DT_INVALID.getNumber()) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeEnumSize(3, dtype_);
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.framework.TensorSpecProto)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.framework.TensorSpecProto other = (org.tensorflow.proto.framework.TensorSpecProto) obj;
-
-    if (!getName()
-        .equals(other.getName())) return false;
-    if (hasShape() != other.hasShape()) return false;
-    if (hasShape()) {
-      if (!getShape()
-          .equals(other.getShape())) return false;
-    }
-    if (dtype_ != other.dtype_) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    hash = (37 * hash) + NAME_FIELD_NUMBER;
-    hash = (53 * hash) + getName().hashCode();
-    if (hasShape()) {
-      hash = (37 * hash) + SHAPE_FIELD_NUMBER;
-      hash = (53 * hash) + getShape().hashCode();
-    }
-    hash = (37 * hash) + DTYPE_FIELD_NUMBER;
-    hash = (53 * hash) + dtype_;
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.framework.TensorSpecProto parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.TensorSpecProto parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.TensorSpecProto parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.TensorSpecProto parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.TensorSpecProto parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.TensorSpecProto parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.TensorSpecProto parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.TensorSpecProto parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.TensorSpecProto parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.TensorSpecProto parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.TensorSpecProto parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.TensorSpecProto parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.framework.TensorSpecProto prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * <pre>
-   * A protobuf to represent tf.TensorSpec.
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.TensorSpecProto}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.TensorSpecProto)
-      org.tensorflow.proto.framework.TensorSpecProtoOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.StructProtos.internal_static_tensorflow_TensorSpecProto_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.StructProtos.internal_static_tensorflow_TensorSpecProto_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.TensorSpecProto.class, org.tensorflow.proto.framework.TensorSpecProto.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.framework.TensorSpecProto.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      name_ = "";
-
-      if (shapeBuilder_ == null) {
-        shape_ = null;
-      } else {
-        shape_ = null;
-        shapeBuilder_ = null;
-      }
-      dtype_ = 0;
-
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.framework.StructProtos.internal_static_tensorflow_TensorSpecProto_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.TensorSpecProto getDefaultInstanceForType() {
-      return org.tensorflow.proto.framework.TensorSpecProto.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.TensorSpecProto build() {
-      org.tensorflow.proto.framework.TensorSpecProto result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.TensorSpecProto buildPartial() {
-      org.tensorflow.proto.framework.TensorSpecProto result = new org.tensorflow.proto.framework.TensorSpecProto(this);
-      result.name_ = name_;
-      if (shapeBuilder_ == null) {
-        result.shape_ = shape_;
-      } else {
-        result.shape_ = shapeBuilder_.build();
-      }
-      result.dtype_ = dtype_;
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.framework.TensorSpecProto) {
-        return mergeFrom((org.tensorflow.proto.framework.TensorSpecProto)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.framework.TensorSpecProto other) {
-      if (other == org.tensorflow.proto.framework.TensorSpecProto.getDefaultInstance()) return this;
-      if (!other.getName().isEmpty()) {
-        name_ = other.name_;
-        onChanged();
-      }
-      if (other.hasShape()) {
-        mergeShape(other.getShape());
-      }
-      if (other.dtype_ != 0) {
-        setDtypeValue(other.getDtypeValue());
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.framework.TensorSpecProto parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.framework.TensorSpecProto) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-
-    private java.lang.Object name_ = "";
-    /**
-     * <code>string name = 1;</code>
-     */
-    public java.lang.String getName() {
-      java.lang.Object ref = name_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        name_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <code>string name = 1;</code>
-     */
-    public com.google.protobuf.ByteString
-        getNameBytes() {
-      java.lang.Object ref = name_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        name_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <code>string name = 1;</code>
-     */
-    public Builder setName(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      name_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>string name = 1;</code>
-     */
-    public Builder clearName() {
-      
-      name_ = getDefaultInstance().getName();
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>string name = 1;</code>
-     */
-    public Builder setNameBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      name_ = value;
-      onChanged();
-      return this;
-    }
-
-    private org.tensorflow.proto.framework.TensorShapeProto shape_;
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.TensorShapeProto, org.tensorflow.proto.framework.TensorShapeProto.Builder, org.tensorflow.proto.framework.TensorShapeProtoOrBuilder> shapeBuilder_;
-    /**
-     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-     */
-    public boolean hasShape() {
-      return shapeBuilder_ != null || shape_ != null;
-    }
-    /**
-     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-     */
-    public org.tensorflow.proto.framework.TensorShapeProto getShape() {
-      if (shapeBuilder_ == null) {
-        return shape_ == null ? org.tensorflow.proto.framework.TensorShapeProto.getDefaultInstance() : shape_;
-      } else {
-        return shapeBuilder_.getMessage();
-      }
-    }
-    /**
-     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-     */
-    public Builder setShape(org.tensorflow.proto.framework.TensorShapeProto value) {
-      if (shapeBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        shape_ = value;
-        onChanged();
-      } else {
-        shapeBuilder_.setMessage(value);
-      }
-
-      return this;
-    }
-    /**
-     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-     */
-    public Builder setShape(
-        org.tensorflow.proto.framework.TensorShapeProto.Builder builderForValue) {
-      if (shapeBuilder_ == null) {
-        shape_ = builderForValue.build();
-        onChanged();
-      } else {
-        shapeBuilder_.setMessage(builderForValue.build());
-      }
-
-      return this;
-    }
-    /**
-     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-     */
-    public Builder mergeShape(org.tensorflow.proto.framework.TensorShapeProto value) {
-      if (shapeBuilder_ == null) {
-        if (shape_ != null) {
-          shape_ =
-            org.tensorflow.proto.framework.TensorShapeProto.newBuilder(shape_).mergeFrom(value).buildPartial();
-        } else {
-          shape_ = value;
-        }
-        onChanged();
-      } else {
-        shapeBuilder_.mergeFrom(value);
-      }
-
-      return this;
-    }
-    /**
-     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-     */
-    public Builder clearShape() {
-      if (shapeBuilder_ == null) {
-        shape_ = null;
-        onChanged();
-      } else {
-        shape_ = null;
-        shapeBuilder_ = null;
-      }
-
-      return this;
-    }
-    /**
-     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-     */
-    public org.tensorflow.proto.framework.TensorShapeProto.Builder getShapeBuilder() {
-      
-      onChanged();
-      return getShapeFieldBuilder().getBuilder();
-    }
-    /**
-     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-     */
-    public org.tensorflow.proto.framework.TensorShapeProtoOrBuilder getShapeOrBuilder() {
-      if (shapeBuilder_ != null) {
-        return shapeBuilder_.getMessageOrBuilder();
-      } else {
-        return shape_ == null ?
-            org.tensorflow.proto.framework.TensorShapeProto.getDefaultInstance() : shape_;
-      }
-    }
-    /**
-     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-     */
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.TensorShapeProto, org.tensorflow.proto.framework.TensorShapeProto.Builder, org.tensorflow.proto.framework.TensorShapeProtoOrBuilder> 
-        getShapeFieldBuilder() {
-      if (shapeBuilder_ == null) {
-        shapeBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-            org.tensorflow.proto.framework.TensorShapeProto, org.tensorflow.proto.framework.TensorShapeProto.Builder, org.tensorflow.proto.framework.TensorShapeProtoOrBuilder>(
-                getShape(),
-                getParentForChildren(),
-                isClean());
-        shape_ = null;
-      }
-      return shapeBuilder_;
-    }
-
-    private int dtype_ = 0;
-    /**
-     * <code>.tensorflow.DataType dtype = 3;</code>
-     */
-    public int getDtypeValue() {
-      return dtype_;
-    }
-    /**
-     * <code>.tensorflow.DataType dtype = 3;</code>
-     */
-    public Builder setDtypeValue(int value) {
-      dtype_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>.tensorflow.DataType dtype = 3;</code>
-     */
-    public org.tensorflow.proto.framework.DataType getDtype() {
-      @SuppressWarnings("deprecation")
-      org.tensorflow.proto.framework.DataType result = org.tensorflow.proto.framework.DataType.valueOf(dtype_);
-      return result == null ? org.tensorflow.proto.framework.DataType.UNRECOGNIZED : result;
-    }
-    /**
-     * <code>.tensorflow.DataType dtype = 3;</code>
-     */
-    public Builder setDtype(org.tensorflow.proto.framework.DataType value) {
-      if (value == null) {
-        throw new NullPointerException();
-      }
-      
-      dtype_ = value.getNumber();
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>.tensorflow.DataType dtype = 3;</code>
-     */
-    public Builder clearDtype() {
-      
-      dtype_ = 0;
-      onChanged();
-      return this;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.TensorSpecProto)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.TensorSpecProto)
-  private static final org.tensorflow.proto.framework.TensorSpecProto DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.TensorSpecProto();
-  }
-
-  public static org.tensorflow.proto.framework.TensorSpecProto getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<TensorSpecProto>
-      PARSER = new com.google.protobuf.AbstractParser<TensorSpecProto>() {
-    @java.lang.Override
-    public TensorSpecProto parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new TensorSpecProto(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<TensorSpecProto> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<TensorSpecProto> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.framework.TensorSpecProto getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorSpecProtoOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorSpecProtoOrBuilder.java
deleted file mode 100644
index f9f8ec12d9d..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorSpecProtoOrBuilder.java
+++ /dev/null
@@ -1,41 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/struct.proto
-
-package org.tensorflow.proto.framework;
-
-public interface TensorSpecProtoOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.TensorSpecProto)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <code>string name = 1;</code>
-   */
-  java.lang.String getName();
-  /**
-   * <code>string name = 1;</code>
-   */
-  com.google.protobuf.ByteString
-      getNameBytes();
-
-  /**
-   * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-   */
-  boolean hasShape();
-  /**
-   * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-   */
-  org.tensorflow.proto.framework.TensorShapeProto getShape();
-  /**
-   * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-   */
-  org.tensorflow.proto.framework.TensorShapeProtoOrBuilder getShapeOrBuilder();
-
-  /**
-   * <code>.tensorflow.DataType dtype = 3;</code>
-   */
-  int getDtypeValue();
-  /**
-   * <code>.tensorflow.DataType dtype = 3;</code>
-   */
-  org.tensorflow.proto.framework.DataType getDtype();
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ThreadPoolOptionProto.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ThreadPoolOptionProto.java
deleted file mode 100644
index 9894b24788e..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ThreadPoolOptionProto.java
+++ /dev/null
@@ -1,739 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/config.proto
-
-package org.tensorflow.proto.framework;
-
-/**
- * Protobuf type {@code tensorflow.ThreadPoolOptionProto}
- */
-public  final class ThreadPoolOptionProto extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.ThreadPoolOptionProto)
-    ThreadPoolOptionProtoOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use ThreadPoolOptionProto.newBuilder() to construct.
-  private ThreadPoolOptionProto(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private ThreadPoolOptionProto() {
-    globalName_ = "";
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new ThreadPoolOptionProto();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private ThreadPoolOptionProto(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 8: {
-
-            numThreads_ = input.readInt32();
-            break;
-          }
-          case 18: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            globalName_ = s;
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_ThreadPoolOptionProto_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_ThreadPoolOptionProto_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.framework.ThreadPoolOptionProto.class, org.tensorflow.proto.framework.ThreadPoolOptionProto.Builder.class);
-  }
-
-  public static final int NUM_THREADS_FIELD_NUMBER = 1;
-  private int numThreads_;
-  /**
-   * <pre>
-   * The number of threads in the pool.
-   * 0 means the system picks a value based on where this option proto is used
-   * (see the declaration of the specific field for more info).
-   * </pre>
-   *
-   * <code>int32 num_threads = 1;</code>
-   */
-  public int getNumThreads() {
-    return numThreads_;
-  }
-
-  public static final int GLOBAL_NAME_FIELD_NUMBER = 2;
-  private volatile java.lang.Object globalName_;
-  /**
-   * <pre>
-   * The global name of the threadpool.
-   * If empty, then the threadpool is made and used according to the scope it's
-   * in - e.g., for a session threadpool, it is used by that session only.
-   * If non-empty, then:
-   * - a global threadpool associated with this name is looked
-   *   up or created. This allows, for example, sharing one threadpool across
-   *   many sessions (e.g., like the default behavior, if
-   *   inter_op_parallelism_threads is not configured), but still partitioning
-   *   into a large and small pool.
-   * - if the threadpool for this global_name already exists, then it is an
-   *   error if the existing pool was created using a different num_threads
-   *   value as is specified on this call.
-   * - threadpools created this way are never garbage collected.
-   * </pre>
-   *
-   * <code>string global_name = 2;</code>
-   */
-  public java.lang.String getGlobalName() {
-    java.lang.Object ref = globalName_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      globalName_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * The global name of the threadpool.
-   * If empty, then the threadpool is made and used according to the scope it's
-   * in - e.g., for a session threadpool, it is used by that session only.
-   * If non-empty, then:
-   * - a global threadpool associated with this name is looked
-   *   up or created. This allows, for example, sharing one threadpool across
-   *   many sessions (e.g., like the default behavior, if
-   *   inter_op_parallelism_threads is not configured), but still partitioning
-   *   into a large and small pool.
-   * - if the threadpool for this global_name already exists, then it is an
-   *   error if the existing pool was created using a different num_threads
-   *   value as is specified on this call.
-   * - threadpools created this way are never garbage collected.
-   * </pre>
-   *
-   * <code>string global_name = 2;</code>
-   */
-  public com.google.protobuf.ByteString
-      getGlobalNameBytes() {
-    java.lang.Object ref = globalName_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      globalName_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    if (numThreads_ != 0) {
-      output.writeInt32(1, numThreads_);
-    }
-    if (!getGlobalNameBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 2, globalName_);
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (numThreads_ != 0) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt32Size(1, numThreads_);
-    }
-    if (!getGlobalNameBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(2, globalName_);
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.framework.ThreadPoolOptionProto)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.framework.ThreadPoolOptionProto other = (org.tensorflow.proto.framework.ThreadPoolOptionProto) obj;
-
-    if (getNumThreads()
-        != other.getNumThreads()) return false;
-    if (!getGlobalName()
-        .equals(other.getGlobalName())) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    hash = (37 * hash) + NUM_THREADS_FIELD_NUMBER;
-    hash = (53 * hash) + getNumThreads();
-    hash = (37 * hash) + GLOBAL_NAME_FIELD_NUMBER;
-    hash = (53 * hash) + getGlobalName().hashCode();
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.framework.ThreadPoolOptionProto parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.ThreadPoolOptionProto parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.ThreadPoolOptionProto parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.ThreadPoolOptionProto parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.ThreadPoolOptionProto parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.ThreadPoolOptionProto parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.ThreadPoolOptionProto parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.ThreadPoolOptionProto parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.ThreadPoolOptionProto parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.ThreadPoolOptionProto parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.ThreadPoolOptionProto parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.ThreadPoolOptionProto parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.framework.ThreadPoolOptionProto prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * Protobuf type {@code tensorflow.ThreadPoolOptionProto}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.ThreadPoolOptionProto)
-      org.tensorflow.proto.framework.ThreadPoolOptionProtoOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_ThreadPoolOptionProto_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_ThreadPoolOptionProto_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.ThreadPoolOptionProto.class, org.tensorflow.proto.framework.ThreadPoolOptionProto.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.framework.ThreadPoolOptionProto.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      numThreads_ = 0;
-
-      globalName_ = "";
-
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_ThreadPoolOptionProto_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.ThreadPoolOptionProto getDefaultInstanceForType() {
-      return org.tensorflow.proto.framework.ThreadPoolOptionProto.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.ThreadPoolOptionProto build() {
-      org.tensorflow.proto.framework.ThreadPoolOptionProto result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.ThreadPoolOptionProto buildPartial() {
-      org.tensorflow.proto.framework.ThreadPoolOptionProto result = new org.tensorflow.proto.framework.ThreadPoolOptionProto(this);
-      result.numThreads_ = numThreads_;
-      result.globalName_ = globalName_;
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.framework.ThreadPoolOptionProto) {
-        return mergeFrom((org.tensorflow.proto.framework.ThreadPoolOptionProto)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.framework.ThreadPoolOptionProto other) {
-      if (other == org.tensorflow.proto.framework.ThreadPoolOptionProto.getDefaultInstance()) return this;
-      if (other.getNumThreads() != 0) {
-        setNumThreads(other.getNumThreads());
-      }
-      if (!other.getGlobalName().isEmpty()) {
-        globalName_ = other.globalName_;
-        onChanged();
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.framework.ThreadPoolOptionProto parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.framework.ThreadPoolOptionProto) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-
-    private int numThreads_ ;
-    /**
-     * <pre>
-     * The number of threads in the pool.
-     * 0 means the system picks a value based on where this option proto is used
-     * (see the declaration of the specific field for more info).
-     * </pre>
-     *
-     * <code>int32 num_threads = 1;</code>
-     */
-    public int getNumThreads() {
-      return numThreads_;
-    }
-    /**
-     * <pre>
-     * The number of threads in the pool.
-     * 0 means the system picks a value based on where this option proto is used
-     * (see the declaration of the specific field for more info).
-     * </pre>
-     *
-     * <code>int32 num_threads = 1;</code>
-     */
-    public Builder setNumThreads(int value) {
-      
-      numThreads_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * The number of threads in the pool.
-     * 0 means the system picks a value based on where this option proto is used
-     * (see the declaration of the specific field for more info).
-     * </pre>
-     *
-     * <code>int32 num_threads = 1;</code>
-     */
-    public Builder clearNumThreads() {
-      
-      numThreads_ = 0;
-      onChanged();
-      return this;
-    }
-
-    private java.lang.Object globalName_ = "";
-    /**
-     * <pre>
-     * The global name of the threadpool.
-     * If empty, then the threadpool is made and used according to the scope it's
-     * in - e.g., for a session threadpool, it is used by that session only.
-     * If non-empty, then:
-     * - a global threadpool associated with this name is looked
-     *   up or created. This allows, for example, sharing one threadpool across
-     *   many sessions (e.g., like the default behavior, if
-     *   inter_op_parallelism_threads is not configured), but still partitioning
-     *   into a large and small pool.
-     * - if the threadpool for this global_name already exists, then it is an
-     *   error if the existing pool was created using a different num_threads
-     *   value as is specified on this call.
-     * - threadpools created this way are never garbage collected.
-     * </pre>
-     *
-     * <code>string global_name = 2;</code>
-     */
-    public java.lang.String getGlobalName() {
-      java.lang.Object ref = globalName_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        globalName_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * The global name of the threadpool.
-     * If empty, then the threadpool is made and used according to the scope it's
-     * in - e.g., for a session threadpool, it is used by that session only.
-     * If non-empty, then:
-     * - a global threadpool associated with this name is looked
-     *   up or created. This allows, for example, sharing one threadpool across
-     *   many sessions (e.g., like the default behavior, if
-     *   inter_op_parallelism_threads is not configured), but still partitioning
-     *   into a large and small pool.
-     * - if the threadpool for this global_name already exists, then it is an
-     *   error if the existing pool was created using a different num_threads
-     *   value as is specified on this call.
-     * - threadpools created this way are never garbage collected.
-     * </pre>
-     *
-     * <code>string global_name = 2;</code>
-     */
-    public com.google.protobuf.ByteString
-        getGlobalNameBytes() {
-      java.lang.Object ref = globalName_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        globalName_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * The global name of the threadpool.
-     * If empty, then the threadpool is made and used according to the scope it's
-     * in - e.g., for a session threadpool, it is used by that session only.
-     * If non-empty, then:
-     * - a global threadpool associated with this name is looked
-     *   up or created. This allows, for example, sharing one threadpool across
-     *   many sessions (e.g., like the default behavior, if
-     *   inter_op_parallelism_threads is not configured), but still partitioning
-     *   into a large and small pool.
-     * - if the threadpool for this global_name already exists, then it is an
-     *   error if the existing pool was created using a different num_threads
-     *   value as is specified on this call.
-     * - threadpools created this way are never garbage collected.
-     * </pre>
-     *
-     * <code>string global_name = 2;</code>
-     */
-    public Builder setGlobalName(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      globalName_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * The global name of the threadpool.
-     * If empty, then the threadpool is made and used according to the scope it's
-     * in - e.g., for a session threadpool, it is used by that session only.
-     * If non-empty, then:
-     * - a global threadpool associated with this name is looked
-     *   up or created. This allows, for example, sharing one threadpool across
-     *   many sessions (e.g., like the default behavior, if
-     *   inter_op_parallelism_threads is not configured), but still partitioning
-     *   into a large and small pool.
-     * - if the threadpool for this global_name already exists, then it is an
-     *   error if the existing pool was created using a different num_threads
-     *   value as is specified on this call.
-     * - threadpools created this way are never garbage collected.
-     * </pre>
-     *
-     * <code>string global_name = 2;</code>
-     */
-    public Builder clearGlobalName() {
-      
-      globalName_ = getDefaultInstance().getGlobalName();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * The global name of the threadpool.
-     * If empty, then the threadpool is made and used according to the scope it's
-     * in - e.g., for a session threadpool, it is used by that session only.
-     * If non-empty, then:
-     * - a global threadpool associated with this name is looked
-     *   up or created. This allows, for example, sharing one threadpool across
-     *   many sessions (e.g., like the default behavior, if
-     *   inter_op_parallelism_threads is not configured), but still partitioning
-     *   into a large and small pool.
-     * - if the threadpool for this global_name already exists, then it is an
-     *   error if the existing pool was created using a different num_threads
-     *   value as is specified on this call.
-     * - threadpools created this way are never garbage collected.
-     * </pre>
-     *
-     * <code>string global_name = 2;</code>
-     */
-    public Builder setGlobalNameBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      globalName_ = value;
-      onChanged();
-      return this;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.ThreadPoolOptionProto)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.ThreadPoolOptionProto)
-  private static final org.tensorflow.proto.framework.ThreadPoolOptionProto DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.ThreadPoolOptionProto();
-  }
-
-  public static org.tensorflow.proto.framework.ThreadPoolOptionProto getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<ThreadPoolOptionProto>
-      PARSER = new com.google.protobuf.AbstractParser<ThreadPoolOptionProto>() {
-    @java.lang.Override
-    public ThreadPoolOptionProto parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new ThreadPoolOptionProto(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<ThreadPoolOptionProto> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<ThreadPoolOptionProto> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.framework.ThreadPoolOptionProto getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ThreadPoolOptionProtoOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ThreadPoolOptionProtoOrBuilder.java
deleted file mode 100644
index d1d3d837519..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ThreadPoolOptionProtoOrBuilder.java
+++ /dev/null
@@ -1,62 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/config.proto
-
-package org.tensorflow.proto.framework;
-
-public interface ThreadPoolOptionProtoOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.ThreadPoolOptionProto)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <pre>
-   * The number of threads in the pool.
-   * 0 means the system picks a value based on where this option proto is used
-   * (see the declaration of the specific field for more info).
-   * </pre>
-   *
-   * <code>int32 num_threads = 1;</code>
-   */
-  int getNumThreads();
-
-  /**
-   * <pre>
-   * The global name of the threadpool.
-   * If empty, then the threadpool is made and used according to the scope it's
-   * in - e.g., for a session threadpool, it is used by that session only.
-   * If non-empty, then:
-   * - a global threadpool associated with this name is looked
-   *   up or created. This allows, for example, sharing one threadpool across
-   *   many sessions (e.g., like the default behavior, if
-   *   inter_op_parallelism_threads is not configured), but still partitioning
-   *   into a large and small pool.
-   * - if the threadpool for this global_name already exists, then it is an
-   *   error if the existing pool was created using a different num_threads
-   *   value as is specified on this call.
-   * - threadpools created this way are never garbage collected.
-   * </pre>
-   *
-   * <code>string global_name = 2;</code>
-   */
-  java.lang.String getGlobalName();
-  /**
-   * <pre>
-   * The global name of the threadpool.
-   * If empty, then the threadpool is made and used according to the scope it's
-   * in - e.g., for a session threadpool, it is used by that session only.
-   * If non-empty, then:
-   * - a global threadpool associated with this name is looked
-   *   up or created. This allows, for example, sharing one threadpool across
-   *   many sessions (e.g., like the default behavior, if
-   *   inter_op_parallelism_threads is not configured), but still partitioning
-   *   into a large and small pool.
-   * - if the threadpool for this global_name already exists, then it is an
-   *   error if the existing pool was created using a different num_threads
-   *   value as is specified on this call.
-   * - threadpools created this way are never garbage collected.
-   * </pre>
-   *
-   * <code>string global_name = 2;</code>
-   */
-  com.google.protobuf.ByteString
-      getGlobalNameBytes();
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/Trace.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/Trace.java
deleted file mode 100644
index 09d97221dde..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/Trace.java
+++ /dev/null
@@ -1,1193 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/trace_events.proto
-
-package org.tensorflow.proto.framework;
-
-/**
- * <pre>
- * A 'Trace' contains metadata for the individual traces of a system.
- * </pre>
- *
- * Protobuf type {@code tensorflow.profiler.Trace}
- */
-public  final class Trace extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.profiler.Trace)
-    TraceOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use Trace.newBuilder() to construct.
-  private Trace(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private Trace() {
-    traceEvents_ = java.util.Collections.emptyList();
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new Trace();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private Trace(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    int mutable_bitField0_ = 0;
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 10: {
-            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
-              devices_ = com.google.protobuf.MapField.newMapField(
-                  DevicesDefaultEntryHolder.defaultEntry);
-              mutable_bitField0_ |= 0x00000001;
-            }
-            com.google.protobuf.MapEntry<java.lang.Integer, org.tensorflow.proto.framework.Device>
-            devices__ = input.readMessage(
-                DevicesDefaultEntryHolder.defaultEntry.getParserForType(), extensionRegistry);
-            devices_.getMutableMap().put(
-                devices__.getKey(), devices__.getValue());
-            break;
-          }
-          case 34: {
-            if (!((mutable_bitField0_ & 0x00000002) != 0)) {
-              traceEvents_ = new java.util.ArrayList<org.tensorflow.proto.framework.TraceEvent>();
-              mutable_bitField0_ |= 0x00000002;
-            }
-            traceEvents_.add(
-                input.readMessage(org.tensorflow.proto.framework.TraceEvent.parser(), extensionRegistry));
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      if (((mutable_bitField0_ & 0x00000002) != 0)) {
-        traceEvents_ = java.util.Collections.unmodifiableList(traceEvents_);
-      }
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.framework.TraceEventsProtos.internal_static_tensorflow_profiler_Trace_descriptor;
-  }
-
-  @SuppressWarnings({"rawtypes"})
-  @java.lang.Override
-  protected com.google.protobuf.MapField internalGetMapField(
-      int number) {
-    switch (number) {
-      case 1:
-        return internalGetDevices();
-      default:
-        throw new RuntimeException(
-            "Invalid map field number: " + number);
-    }
-  }
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.framework.TraceEventsProtos.internal_static_tensorflow_profiler_Trace_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.framework.Trace.class, org.tensorflow.proto.framework.Trace.Builder.class);
-  }
-
-  public static final int DEVICES_FIELD_NUMBER = 1;
-  private static final class DevicesDefaultEntryHolder {
-    static final com.google.protobuf.MapEntry<
-        java.lang.Integer, org.tensorflow.proto.framework.Device> defaultEntry =
-            com.google.protobuf.MapEntry
-            .<java.lang.Integer, org.tensorflow.proto.framework.Device>newDefaultInstance(
-                org.tensorflow.proto.framework.TraceEventsProtos.internal_static_tensorflow_profiler_Trace_DevicesEntry_descriptor, 
-                com.google.protobuf.WireFormat.FieldType.UINT32,
-                0,
-                com.google.protobuf.WireFormat.FieldType.MESSAGE,
-                org.tensorflow.proto.framework.Device.getDefaultInstance());
-  }
-  private com.google.protobuf.MapField<
-      java.lang.Integer, org.tensorflow.proto.framework.Device> devices_;
-  private com.google.protobuf.MapField<java.lang.Integer, org.tensorflow.proto.framework.Device>
-  internalGetDevices() {
-    if (devices_ == null) {
-      return com.google.protobuf.MapField.emptyMapField(
-          DevicesDefaultEntryHolder.defaultEntry);
-    }
-    return devices_;
-  }
-
-  public int getDevicesCount() {
-    return internalGetDevices().getMap().size();
-  }
-  /**
-   * <pre>
-   * The devices that this trace has information about. Maps from device_id to
-   * more data about the specific device.
-   * </pre>
-   *
-   * <code>map&lt;uint32, .tensorflow.profiler.Device&gt; devices = 1;</code>
-   */
-
-  public boolean containsDevices(
-      int key) {
-    
-    return internalGetDevices().getMap().containsKey(key);
-  }
-  /**
-   * Use {@link #getDevicesMap()} instead.
-   */
-  @java.lang.Deprecated
-  public java.util.Map<java.lang.Integer, org.tensorflow.proto.framework.Device> getDevices() {
-    return getDevicesMap();
-  }
-  /**
-   * <pre>
-   * The devices that this trace has information about. Maps from device_id to
-   * more data about the specific device.
-   * </pre>
-   *
-   * <code>map&lt;uint32, .tensorflow.profiler.Device&gt; devices = 1;</code>
-   */
-
-  public java.util.Map<java.lang.Integer, org.tensorflow.proto.framework.Device> getDevicesMap() {
-    return internalGetDevices().getMap();
-  }
-  /**
-   * <pre>
-   * The devices that this trace has information about. Maps from device_id to
-   * more data about the specific device.
-   * </pre>
-   *
-   * <code>map&lt;uint32, .tensorflow.profiler.Device&gt; devices = 1;</code>
-   */
-
-  public org.tensorflow.proto.framework.Device getDevicesOrDefault(
-      int key,
-      org.tensorflow.proto.framework.Device defaultValue) {
-    
-    java.util.Map<java.lang.Integer, org.tensorflow.proto.framework.Device> map =
-        internalGetDevices().getMap();
-    return map.containsKey(key) ? map.get(key) : defaultValue;
-  }
-  /**
-   * <pre>
-   * The devices that this trace has information about. Maps from device_id to
-   * more data about the specific device.
-   * </pre>
-   *
-   * <code>map&lt;uint32, .tensorflow.profiler.Device&gt; devices = 1;</code>
-   */
-
-  public org.tensorflow.proto.framework.Device getDevicesOrThrow(
-      int key) {
-    
-    java.util.Map<java.lang.Integer, org.tensorflow.proto.framework.Device> map =
-        internalGetDevices().getMap();
-    if (!map.containsKey(key)) {
-      throw new java.lang.IllegalArgumentException();
-    }
-    return map.get(key);
-  }
-
-  public static final int TRACE_EVENTS_FIELD_NUMBER = 4;
-  private java.util.List<org.tensorflow.proto.framework.TraceEvent> traceEvents_;
-  /**
-   * <pre>
-   * All trace events capturing in the profiling period.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.profiler.TraceEvent trace_events = 4;</code>
-   */
-  public java.util.List<org.tensorflow.proto.framework.TraceEvent> getTraceEventsList() {
-    return traceEvents_;
-  }
-  /**
-   * <pre>
-   * All trace events capturing in the profiling period.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.profiler.TraceEvent trace_events = 4;</code>
-   */
-  public java.util.List<? extends org.tensorflow.proto.framework.TraceEventOrBuilder> 
-      getTraceEventsOrBuilderList() {
-    return traceEvents_;
-  }
-  /**
-   * <pre>
-   * All trace events capturing in the profiling period.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.profiler.TraceEvent trace_events = 4;</code>
-   */
-  public int getTraceEventsCount() {
-    return traceEvents_.size();
-  }
-  /**
-   * <pre>
-   * All trace events capturing in the profiling period.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.profiler.TraceEvent trace_events = 4;</code>
-   */
-  public org.tensorflow.proto.framework.TraceEvent getTraceEvents(int index) {
-    return traceEvents_.get(index);
-  }
-  /**
-   * <pre>
-   * All trace events capturing in the profiling period.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.profiler.TraceEvent trace_events = 4;</code>
-   */
-  public org.tensorflow.proto.framework.TraceEventOrBuilder getTraceEventsOrBuilder(
-      int index) {
-    return traceEvents_.get(index);
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    com.google.protobuf.GeneratedMessageV3
-      .serializeIntegerMapTo(
-        output,
-        internalGetDevices(),
-        DevicesDefaultEntryHolder.defaultEntry,
-        1);
-    for (int i = 0; i < traceEvents_.size(); i++) {
-      output.writeMessage(4, traceEvents_.get(i));
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    for (java.util.Map.Entry<java.lang.Integer, org.tensorflow.proto.framework.Device> entry
-         : internalGetDevices().getMap().entrySet()) {
-      com.google.protobuf.MapEntry<java.lang.Integer, org.tensorflow.proto.framework.Device>
-      devices__ = DevicesDefaultEntryHolder.defaultEntry.newBuilderForType()
-          .setKey(entry.getKey())
-          .setValue(entry.getValue())
-          .build();
-      size += com.google.protobuf.CodedOutputStream
-          .computeMessageSize(1, devices__);
-    }
-    for (int i = 0; i < traceEvents_.size(); i++) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(4, traceEvents_.get(i));
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.framework.Trace)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.framework.Trace other = (org.tensorflow.proto.framework.Trace) obj;
-
-    if (!internalGetDevices().equals(
-        other.internalGetDevices())) return false;
-    if (!getTraceEventsList()
-        .equals(other.getTraceEventsList())) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    if (!internalGetDevices().getMap().isEmpty()) {
-      hash = (37 * hash) + DEVICES_FIELD_NUMBER;
-      hash = (53 * hash) + internalGetDevices().hashCode();
-    }
-    if (getTraceEventsCount() > 0) {
-      hash = (37 * hash) + TRACE_EVENTS_FIELD_NUMBER;
-      hash = (53 * hash) + getTraceEventsList().hashCode();
-    }
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.framework.Trace parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.Trace parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.Trace parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.Trace parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.Trace parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.Trace parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.Trace parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.Trace parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.Trace parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.Trace parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.Trace parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.Trace parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.framework.Trace prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * <pre>
-   * A 'Trace' contains metadata for the individual traces of a system.
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.profiler.Trace}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.profiler.Trace)
-      org.tensorflow.proto.framework.TraceOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.TraceEventsProtos.internal_static_tensorflow_profiler_Trace_descriptor;
-    }
-
-    @SuppressWarnings({"rawtypes"})
-    protected com.google.protobuf.MapField internalGetMapField(
-        int number) {
-      switch (number) {
-        case 1:
-          return internalGetDevices();
-        default:
-          throw new RuntimeException(
-              "Invalid map field number: " + number);
-      }
-    }
-    @SuppressWarnings({"rawtypes"})
-    protected com.google.protobuf.MapField internalGetMutableMapField(
-        int number) {
-      switch (number) {
-        case 1:
-          return internalGetMutableDevices();
-        default:
-          throw new RuntimeException(
-              "Invalid map field number: " + number);
-      }
-    }
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.TraceEventsProtos.internal_static_tensorflow_profiler_Trace_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.Trace.class, org.tensorflow.proto.framework.Trace.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.framework.Trace.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-        getTraceEventsFieldBuilder();
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      internalGetMutableDevices().clear();
-      if (traceEventsBuilder_ == null) {
-        traceEvents_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000002);
-      } else {
-        traceEventsBuilder_.clear();
-      }
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.framework.TraceEventsProtos.internal_static_tensorflow_profiler_Trace_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.Trace getDefaultInstanceForType() {
-      return org.tensorflow.proto.framework.Trace.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.Trace build() {
-      org.tensorflow.proto.framework.Trace result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.Trace buildPartial() {
-      org.tensorflow.proto.framework.Trace result = new org.tensorflow.proto.framework.Trace(this);
-      int from_bitField0_ = bitField0_;
-      result.devices_ = internalGetDevices();
-      result.devices_.makeImmutable();
-      if (traceEventsBuilder_ == null) {
-        if (((bitField0_ & 0x00000002) != 0)) {
-          traceEvents_ = java.util.Collections.unmodifiableList(traceEvents_);
-          bitField0_ = (bitField0_ & ~0x00000002);
-        }
-        result.traceEvents_ = traceEvents_;
-      } else {
-        result.traceEvents_ = traceEventsBuilder_.build();
-      }
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.framework.Trace) {
-        return mergeFrom((org.tensorflow.proto.framework.Trace)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.framework.Trace other) {
-      if (other == org.tensorflow.proto.framework.Trace.getDefaultInstance()) return this;
-      internalGetMutableDevices().mergeFrom(
-          other.internalGetDevices());
-      if (traceEventsBuilder_ == null) {
-        if (!other.traceEvents_.isEmpty()) {
-          if (traceEvents_.isEmpty()) {
-            traceEvents_ = other.traceEvents_;
-            bitField0_ = (bitField0_ & ~0x00000002);
-          } else {
-            ensureTraceEventsIsMutable();
-            traceEvents_.addAll(other.traceEvents_);
-          }
-          onChanged();
-        }
-      } else {
-        if (!other.traceEvents_.isEmpty()) {
-          if (traceEventsBuilder_.isEmpty()) {
-            traceEventsBuilder_.dispose();
-            traceEventsBuilder_ = null;
-            traceEvents_ = other.traceEvents_;
-            bitField0_ = (bitField0_ & ~0x00000002);
-            traceEventsBuilder_ = 
-              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
-                 getTraceEventsFieldBuilder() : null;
-          } else {
-            traceEventsBuilder_.addAllMessages(other.traceEvents_);
-          }
-        }
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.framework.Trace parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.framework.Trace) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-    private int bitField0_;
-
-    private com.google.protobuf.MapField<
-        java.lang.Integer, org.tensorflow.proto.framework.Device> devices_;
-    private com.google.protobuf.MapField<java.lang.Integer, org.tensorflow.proto.framework.Device>
-    internalGetDevices() {
-      if (devices_ == null) {
-        return com.google.protobuf.MapField.emptyMapField(
-            DevicesDefaultEntryHolder.defaultEntry);
-      }
-      return devices_;
-    }
-    private com.google.protobuf.MapField<java.lang.Integer, org.tensorflow.proto.framework.Device>
-    internalGetMutableDevices() {
-      onChanged();;
-      if (devices_ == null) {
-        devices_ = com.google.protobuf.MapField.newMapField(
-            DevicesDefaultEntryHolder.defaultEntry);
-      }
-      if (!devices_.isMutable()) {
-        devices_ = devices_.copy();
-      }
-      return devices_;
-    }
-
-    public int getDevicesCount() {
-      return internalGetDevices().getMap().size();
-    }
-    /**
-     * <pre>
-     * The devices that this trace has information about. Maps from device_id to
-     * more data about the specific device.
-     * </pre>
-     *
-     * <code>map&lt;uint32, .tensorflow.profiler.Device&gt; devices = 1;</code>
-     */
-
-    public boolean containsDevices(
-        int key) {
-      
-      return internalGetDevices().getMap().containsKey(key);
-    }
-    /**
-     * Use {@link #getDevicesMap()} instead.
-     */
-    @java.lang.Deprecated
-    public java.util.Map<java.lang.Integer, org.tensorflow.proto.framework.Device> getDevices() {
-      return getDevicesMap();
-    }
-    /**
-     * <pre>
-     * The devices that this trace has information about. Maps from device_id to
-     * more data about the specific device.
-     * </pre>
-     *
-     * <code>map&lt;uint32, .tensorflow.profiler.Device&gt; devices = 1;</code>
-     */
-
-    public java.util.Map<java.lang.Integer, org.tensorflow.proto.framework.Device> getDevicesMap() {
-      return internalGetDevices().getMap();
-    }
-    /**
-     * <pre>
-     * The devices that this trace has information about. Maps from device_id to
-     * more data about the specific device.
-     * </pre>
-     *
-     * <code>map&lt;uint32, .tensorflow.profiler.Device&gt; devices = 1;</code>
-     */
-
-    public org.tensorflow.proto.framework.Device getDevicesOrDefault(
-        int key,
-        org.tensorflow.proto.framework.Device defaultValue) {
-      
-      java.util.Map<java.lang.Integer, org.tensorflow.proto.framework.Device> map =
-          internalGetDevices().getMap();
-      return map.containsKey(key) ? map.get(key) : defaultValue;
-    }
-    /**
-     * <pre>
-     * The devices that this trace has information about. Maps from device_id to
-     * more data about the specific device.
-     * </pre>
-     *
-     * <code>map&lt;uint32, .tensorflow.profiler.Device&gt; devices = 1;</code>
-     */
-
-    public org.tensorflow.proto.framework.Device getDevicesOrThrow(
-        int key) {
-      
-      java.util.Map<java.lang.Integer, org.tensorflow.proto.framework.Device> map =
-          internalGetDevices().getMap();
-      if (!map.containsKey(key)) {
-        throw new java.lang.IllegalArgumentException();
-      }
-      return map.get(key);
-    }
-
-    public Builder clearDevices() {
-      internalGetMutableDevices().getMutableMap()
-          .clear();
-      return this;
-    }
-    /**
-     * <pre>
-     * The devices that this trace has information about. Maps from device_id to
-     * more data about the specific device.
-     * </pre>
-     *
-     * <code>map&lt;uint32, .tensorflow.profiler.Device&gt; devices = 1;</code>
-     */
-
-    public Builder removeDevices(
-        int key) {
-      
-      internalGetMutableDevices().getMutableMap()
-          .remove(key);
-      return this;
-    }
-    /**
-     * Use alternate mutation accessors instead.
-     */
-    @java.lang.Deprecated
-    public java.util.Map<java.lang.Integer, org.tensorflow.proto.framework.Device>
-    getMutableDevices() {
-      return internalGetMutableDevices().getMutableMap();
-    }
-    /**
-     * <pre>
-     * The devices that this trace has information about. Maps from device_id to
-     * more data about the specific device.
-     * </pre>
-     *
-     * <code>map&lt;uint32, .tensorflow.profiler.Device&gt; devices = 1;</code>
-     */
-    public Builder putDevices(
-        int key,
-        org.tensorflow.proto.framework.Device value) {
-      
-      if (value == null) { throw new java.lang.NullPointerException(); }
-      internalGetMutableDevices().getMutableMap()
-          .put(key, value);
-      return this;
-    }
-    /**
-     * <pre>
-     * The devices that this trace has information about. Maps from device_id to
-     * more data about the specific device.
-     * </pre>
-     *
-     * <code>map&lt;uint32, .tensorflow.profiler.Device&gt; devices = 1;</code>
-     */
-
-    public Builder putAllDevices(
-        java.util.Map<java.lang.Integer, org.tensorflow.proto.framework.Device> values) {
-      internalGetMutableDevices().getMutableMap()
-          .putAll(values);
-      return this;
-    }
-
-    private java.util.List<org.tensorflow.proto.framework.TraceEvent> traceEvents_ =
-      java.util.Collections.emptyList();
-    private void ensureTraceEventsIsMutable() {
-      if (!((bitField0_ & 0x00000002) != 0)) {
-        traceEvents_ = new java.util.ArrayList<org.tensorflow.proto.framework.TraceEvent>(traceEvents_);
-        bitField0_ |= 0x00000002;
-       }
-    }
-
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.framework.TraceEvent, org.tensorflow.proto.framework.TraceEvent.Builder, org.tensorflow.proto.framework.TraceEventOrBuilder> traceEventsBuilder_;
-
-    /**
-     * <pre>
-     * All trace events capturing in the profiling period.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.profiler.TraceEvent trace_events = 4;</code>
-     */
-    public java.util.List<org.tensorflow.proto.framework.TraceEvent> getTraceEventsList() {
-      if (traceEventsBuilder_ == null) {
-        return java.util.Collections.unmodifiableList(traceEvents_);
-      } else {
-        return traceEventsBuilder_.getMessageList();
-      }
-    }
-    /**
-     * <pre>
-     * All trace events capturing in the profiling period.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.profiler.TraceEvent trace_events = 4;</code>
-     */
-    public int getTraceEventsCount() {
-      if (traceEventsBuilder_ == null) {
-        return traceEvents_.size();
-      } else {
-        return traceEventsBuilder_.getCount();
-      }
-    }
-    /**
-     * <pre>
-     * All trace events capturing in the profiling period.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.profiler.TraceEvent trace_events = 4;</code>
-     */
-    public org.tensorflow.proto.framework.TraceEvent getTraceEvents(int index) {
-      if (traceEventsBuilder_ == null) {
-        return traceEvents_.get(index);
-      } else {
-        return traceEventsBuilder_.getMessage(index);
-      }
-    }
-    /**
-     * <pre>
-     * All trace events capturing in the profiling period.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.profiler.TraceEvent trace_events = 4;</code>
-     */
-    public Builder setTraceEvents(
-        int index, org.tensorflow.proto.framework.TraceEvent value) {
-      if (traceEventsBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureTraceEventsIsMutable();
-        traceEvents_.set(index, value);
-        onChanged();
-      } else {
-        traceEventsBuilder_.setMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * All trace events capturing in the profiling period.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.profiler.TraceEvent trace_events = 4;</code>
-     */
-    public Builder setTraceEvents(
-        int index, org.tensorflow.proto.framework.TraceEvent.Builder builderForValue) {
-      if (traceEventsBuilder_ == null) {
-        ensureTraceEventsIsMutable();
-        traceEvents_.set(index, builderForValue.build());
-        onChanged();
-      } else {
-        traceEventsBuilder_.setMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * All trace events capturing in the profiling period.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.profiler.TraceEvent trace_events = 4;</code>
-     */
-    public Builder addTraceEvents(org.tensorflow.proto.framework.TraceEvent value) {
-      if (traceEventsBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureTraceEventsIsMutable();
-        traceEvents_.add(value);
-        onChanged();
-      } else {
-        traceEventsBuilder_.addMessage(value);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * All trace events capturing in the profiling period.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.profiler.TraceEvent trace_events = 4;</code>
-     */
-    public Builder addTraceEvents(
-        int index, org.tensorflow.proto.framework.TraceEvent value) {
-      if (traceEventsBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureTraceEventsIsMutable();
-        traceEvents_.add(index, value);
-        onChanged();
-      } else {
-        traceEventsBuilder_.addMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * All trace events capturing in the profiling period.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.profiler.TraceEvent trace_events = 4;</code>
-     */
-    public Builder addTraceEvents(
-        org.tensorflow.proto.framework.TraceEvent.Builder builderForValue) {
-      if (traceEventsBuilder_ == null) {
-        ensureTraceEventsIsMutable();
-        traceEvents_.add(builderForValue.build());
-        onChanged();
-      } else {
-        traceEventsBuilder_.addMessage(builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * All trace events capturing in the profiling period.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.profiler.TraceEvent trace_events = 4;</code>
-     */
-    public Builder addTraceEvents(
-        int index, org.tensorflow.proto.framework.TraceEvent.Builder builderForValue) {
-      if (traceEventsBuilder_ == null) {
-        ensureTraceEventsIsMutable();
-        traceEvents_.add(index, builderForValue.build());
-        onChanged();
-      } else {
-        traceEventsBuilder_.addMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * All trace events capturing in the profiling period.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.profiler.TraceEvent trace_events = 4;</code>
-     */
-    public Builder addAllTraceEvents(
-        java.lang.Iterable<? extends org.tensorflow.proto.framework.TraceEvent> values) {
-      if (traceEventsBuilder_ == null) {
-        ensureTraceEventsIsMutable();
-        com.google.protobuf.AbstractMessageLite.Builder.addAll(
-            values, traceEvents_);
-        onChanged();
-      } else {
-        traceEventsBuilder_.addAllMessages(values);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * All trace events capturing in the profiling period.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.profiler.TraceEvent trace_events = 4;</code>
-     */
-    public Builder clearTraceEvents() {
-      if (traceEventsBuilder_ == null) {
-        traceEvents_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000002);
-        onChanged();
-      } else {
-        traceEventsBuilder_.clear();
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * All trace events capturing in the profiling period.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.profiler.TraceEvent trace_events = 4;</code>
-     */
-    public Builder removeTraceEvents(int index) {
-      if (traceEventsBuilder_ == null) {
-        ensureTraceEventsIsMutable();
-        traceEvents_.remove(index);
-        onChanged();
-      } else {
-        traceEventsBuilder_.remove(index);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * All trace events capturing in the profiling period.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.profiler.TraceEvent trace_events = 4;</code>
-     */
-    public org.tensorflow.proto.framework.TraceEvent.Builder getTraceEventsBuilder(
-        int index) {
-      return getTraceEventsFieldBuilder().getBuilder(index);
-    }
-    /**
-     * <pre>
-     * All trace events capturing in the profiling period.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.profiler.TraceEvent trace_events = 4;</code>
-     */
-    public org.tensorflow.proto.framework.TraceEventOrBuilder getTraceEventsOrBuilder(
-        int index) {
-      if (traceEventsBuilder_ == null) {
-        return traceEvents_.get(index);  } else {
-        return traceEventsBuilder_.getMessageOrBuilder(index);
-      }
-    }
-    /**
-     * <pre>
-     * All trace events capturing in the profiling period.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.profiler.TraceEvent trace_events = 4;</code>
-     */
-    public java.util.List<? extends org.tensorflow.proto.framework.TraceEventOrBuilder> 
-         getTraceEventsOrBuilderList() {
-      if (traceEventsBuilder_ != null) {
-        return traceEventsBuilder_.getMessageOrBuilderList();
-      } else {
-        return java.util.Collections.unmodifiableList(traceEvents_);
-      }
-    }
-    /**
-     * <pre>
-     * All trace events capturing in the profiling period.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.profiler.TraceEvent trace_events = 4;</code>
-     */
-    public org.tensorflow.proto.framework.TraceEvent.Builder addTraceEventsBuilder() {
-      return getTraceEventsFieldBuilder().addBuilder(
-          org.tensorflow.proto.framework.TraceEvent.getDefaultInstance());
-    }
-    /**
-     * <pre>
-     * All trace events capturing in the profiling period.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.profiler.TraceEvent trace_events = 4;</code>
-     */
-    public org.tensorflow.proto.framework.TraceEvent.Builder addTraceEventsBuilder(
-        int index) {
-      return getTraceEventsFieldBuilder().addBuilder(
-          index, org.tensorflow.proto.framework.TraceEvent.getDefaultInstance());
-    }
-    /**
-     * <pre>
-     * All trace events capturing in the profiling period.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.profiler.TraceEvent trace_events = 4;</code>
-     */
-    public java.util.List<org.tensorflow.proto.framework.TraceEvent.Builder> 
-         getTraceEventsBuilderList() {
-      return getTraceEventsFieldBuilder().getBuilderList();
-    }
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.framework.TraceEvent, org.tensorflow.proto.framework.TraceEvent.Builder, org.tensorflow.proto.framework.TraceEventOrBuilder> 
-        getTraceEventsFieldBuilder() {
-      if (traceEventsBuilder_ == null) {
-        traceEventsBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
-            org.tensorflow.proto.framework.TraceEvent, org.tensorflow.proto.framework.TraceEvent.Builder, org.tensorflow.proto.framework.TraceEventOrBuilder>(
-                traceEvents_,
-                ((bitField0_ & 0x00000002) != 0),
-                getParentForChildren(),
-                isClean());
-        traceEvents_ = null;
-      }
-      return traceEventsBuilder_;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.profiler.Trace)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.profiler.Trace)
-  private static final org.tensorflow.proto.framework.Trace DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.Trace();
-  }
-
-  public static org.tensorflow.proto.framework.Trace getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<Trace>
-      PARSER = new com.google.protobuf.AbstractParser<Trace>() {
-    @java.lang.Override
-    public Trace parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new Trace(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<Trace> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<Trace> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.framework.Trace getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TraceEvent.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TraceEvent.java
deleted file mode 100644
index 8f34acfde34..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TraceEvent.java
+++ /dev/null
@@ -1,1208 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/trace_events.proto
-
-package org.tensorflow.proto.framework;
-
-/**
- * Protobuf type {@code tensorflow.profiler.TraceEvent}
- */
-public  final class TraceEvent extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.profiler.TraceEvent)
-    TraceEventOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use TraceEvent.newBuilder() to construct.
-  private TraceEvent(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private TraceEvent() {
-    name_ = "";
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new TraceEvent();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private TraceEvent(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    int mutable_bitField0_ = 0;
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 8: {
-
-            deviceId_ = input.readUInt32();
-            break;
-          }
-          case 16: {
-
-            resourceId_ = input.readUInt32();
-            break;
-          }
-          case 26: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            name_ = s;
-            break;
-          }
-          case 72: {
-
-            timestampPs_ = input.readUInt64();
-            break;
-          }
-          case 80: {
-
-            durationPs_ = input.readUInt64();
-            break;
-          }
-          case 90: {
-            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
-              args_ = com.google.protobuf.MapField.newMapField(
-                  ArgsDefaultEntryHolder.defaultEntry);
-              mutable_bitField0_ |= 0x00000001;
-            }
-            com.google.protobuf.MapEntry<java.lang.String, java.lang.String>
-            args__ = input.readMessage(
-                ArgsDefaultEntryHolder.defaultEntry.getParserForType(), extensionRegistry);
-            args_.getMutableMap().put(
-                args__.getKey(), args__.getValue());
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.framework.TraceEventsProtos.internal_static_tensorflow_profiler_TraceEvent_descriptor;
-  }
-
-  @SuppressWarnings({"rawtypes"})
-  @java.lang.Override
-  protected com.google.protobuf.MapField internalGetMapField(
-      int number) {
-    switch (number) {
-      case 11:
-        return internalGetArgs();
-      default:
-        throw new RuntimeException(
-            "Invalid map field number: " + number);
-    }
-  }
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.framework.TraceEventsProtos.internal_static_tensorflow_profiler_TraceEvent_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.framework.TraceEvent.class, org.tensorflow.proto.framework.TraceEvent.Builder.class);
-  }
-
-  public static final int DEVICE_ID_FIELD_NUMBER = 1;
-  private int deviceId_;
-  /**
-   * <pre>
-   * The id of the device that this event occurred on. The full dataset should
-   * have this device present in the Trace object.
-   * </pre>
-   *
-   * <code>uint32 device_id = 1;</code>
-   */
-  public int getDeviceId() {
-    return deviceId_;
-  }
-
-  public static final int RESOURCE_ID_FIELD_NUMBER = 2;
-  private int resourceId_;
-  /**
-   * <pre>
-   * The id of the resource that this event occurred on. The full dataset should
-   * have this resource present in the Device object of the Trace object. A
-   * resource_id is unique on a specific device, but not necessarily within the
-   * trace.
-   * </pre>
-   *
-   * <code>uint32 resource_id = 2;</code>
-   */
-  public int getResourceId() {
-    return resourceId_;
-  }
-
-  public static final int NAME_FIELD_NUMBER = 3;
-  private volatile java.lang.Object name_;
-  /**
-   * <pre>
-   * The name of this trace event.
-   * </pre>
-   *
-   * <code>string name = 3;</code>
-   */
-  public java.lang.String getName() {
-    java.lang.Object ref = name_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      name_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * The name of this trace event.
-   * </pre>
-   *
-   * <code>string name = 3;</code>
-   */
-  public com.google.protobuf.ByteString
-      getNameBytes() {
-    java.lang.Object ref = name_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      name_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int TIMESTAMP_PS_FIELD_NUMBER = 9;
-  private long timestampPs_;
-  /**
-   * <pre>
-   * The timestamp that this event occurred at (in picos since tracing started).
-   * </pre>
-   *
-   * <code>uint64 timestamp_ps = 9;</code>
-   */
-  public long getTimestampPs() {
-    return timestampPs_;
-  }
-
-  public static final int DURATION_PS_FIELD_NUMBER = 10;
-  private long durationPs_;
-  /**
-   * <pre>
-   * The duration of the event in picoseconds if applicable.
-   * Events without duration are called instant events.
-   * </pre>
-   *
-   * <code>uint64 duration_ps = 10;</code>
-   */
-  public long getDurationPs() {
-    return durationPs_;
-  }
-
-  public static final int ARGS_FIELD_NUMBER = 11;
-  private static final class ArgsDefaultEntryHolder {
-    static final com.google.protobuf.MapEntry<
-        java.lang.String, java.lang.String> defaultEntry =
-            com.google.protobuf.MapEntry
-            .<java.lang.String, java.lang.String>newDefaultInstance(
-                org.tensorflow.proto.framework.TraceEventsProtos.internal_static_tensorflow_profiler_TraceEvent_ArgsEntry_descriptor, 
-                com.google.protobuf.WireFormat.FieldType.STRING,
-                "",
-                com.google.protobuf.WireFormat.FieldType.STRING,
-                "");
-  }
-  private com.google.protobuf.MapField<
-      java.lang.String, java.lang.String> args_;
-  private com.google.protobuf.MapField<java.lang.String, java.lang.String>
-  internalGetArgs() {
-    if (args_ == null) {
-      return com.google.protobuf.MapField.emptyMapField(
-          ArgsDefaultEntryHolder.defaultEntry);
-    }
-    return args_;
-  }
-
-  public int getArgsCount() {
-    return internalGetArgs().getMap().size();
-  }
-  /**
-   * <pre>
-   * Extra arguments that will be displayed in trace view.
-   * </pre>
-   *
-   * <code>map&lt;string, string&gt; args = 11;</code>
-   */
-
-  public boolean containsArgs(
-      java.lang.String key) {
-    if (key == null) { throw new java.lang.NullPointerException(); }
-    return internalGetArgs().getMap().containsKey(key);
-  }
-  /**
-   * Use {@link #getArgsMap()} instead.
-   */
-  @java.lang.Deprecated
-  public java.util.Map<java.lang.String, java.lang.String> getArgs() {
-    return getArgsMap();
-  }
-  /**
-   * <pre>
-   * Extra arguments that will be displayed in trace view.
-   * </pre>
-   *
-   * <code>map&lt;string, string&gt; args = 11;</code>
-   */
-
-  public java.util.Map<java.lang.String, java.lang.String> getArgsMap() {
-    return internalGetArgs().getMap();
-  }
-  /**
-   * <pre>
-   * Extra arguments that will be displayed in trace view.
-   * </pre>
-   *
-   * <code>map&lt;string, string&gt; args = 11;</code>
-   */
-
-  public java.lang.String getArgsOrDefault(
-      java.lang.String key,
-      java.lang.String defaultValue) {
-    if (key == null) { throw new java.lang.NullPointerException(); }
-    java.util.Map<java.lang.String, java.lang.String> map =
-        internalGetArgs().getMap();
-    return map.containsKey(key) ? map.get(key) : defaultValue;
-  }
-  /**
-   * <pre>
-   * Extra arguments that will be displayed in trace view.
-   * </pre>
-   *
-   * <code>map&lt;string, string&gt; args = 11;</code>
-   */
-
-  public java.lang.String getArgsOrThrow(
-      java.lang.String key) {
-    if (key == null) { throw new java.lang.NullPointerException(); }
-    java.util.Map<java.lang.String, java.lang.String> map =
-        internalGetArgs().getMap();
-    if (!map.containsKey(key)) {
-      throw new java.lang.IllegalArgumentException();
-    }
-    return map.get(key);
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    if (deviceId_ != 0) {
-      output.writeUInt32(1, deviceId_);
-    }
-    if (resourceId_ != 0) {
-      output.writeUInt32(2, resourceId_);
-    }
-    if (!getNameBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 3, name_);
-    }
-    if (timestampPs_ != 0L) {
-      output.writeUInt64(9, timestampPs_);
-    }
-    if (durationPs_ != 0L) {
-      output.writeUInt64(10, durationPs_);
-    }
-    com.google.protobuf.GeneratedMessageV3
-      .serializeStringMapTo(
-        output,
-        internalGetArgs(),
-        ArgsDefaultEntryHolder.defaultEntry,
-        11);
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (deviceId_ != 0) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeUInt32Size(1, deviceId_);
-    }
-    if (resourceId_ != 0) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeUInt32Size(2, resourceId_);
-    }
-    if (!getNameBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(3, name_);
-    }
-    if (timestampPs_ != 0L) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeUInt64Size(9, timestampPs_);
-    }
-    if (durationPs_ != 0L) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeUInt64Size(10, durationPs_);
-    }
-    for (java.util.Map.Entry<java.lang.String, java.lang.String> entry
-         : internalGetArgs().getMap().entrySet()) {
-      com.google.protobuf.MapEntry<java.lang.String, java.lang.String>
-      args__ = ArgsDefaultEntryHolder.defaultEntry.newBuilderForType()
-          .setKey(entry.getKey())
-          .setValue(entry.getValue())
-          .build();
-      size += com.google.protobuf.CodedOutputStream
-          .computeMessageSize(11, args__);
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.framework.TraceEvent)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.framework.TraceEvent other = (org.tensorflow.proto.framework.TraceEvent) obj;
-
-    if (getDeviceId()
-        != other.getDeviceId()) return false;
-    if (getResourceId()
-        != other.getResourceId()) return false;
-    if (!getName()
-        .equals(other.getName())) return false;
-    if (getTimestampPs()
-        != other.getTimestampPs()) return false;
-    if (getDurationPs()
-        != other.getDurationPs()) return false;
-    if (!internalGetArgs().equals(
-        other.internalGetArgs())) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    hash = (37 * hash) + DEVICE_ID_FIELD_NUMBER;
-    hash = (53 * hash) + getDeviceId();
-    hash = (37 * hash) + RESOURCE_ID_FIELD_NUMBER;
-    hash = (53 * hash) + getResourceId();
-    hash = (37 * hash) + NAME_FIELD_NUMBER;
-    hash = (53 * hash) + getName().hashCode();
-    hash = (37 * hash) + TIMESTAMP_PS_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-        getTimestampPs());
-    hash = (37 * hash) + DURATION_PS_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-        getDurationPs());
-    if (!internalGetArgs().getMap().isEmpty()) {
-      hash = (37 * hash) + ARGS_FIELD_NUMBER;
-      hash = (53 * hash) + internalGetArgs().hashCode();
-    }
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.framework.TraceEvent parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.TraceEvent parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.TraceEvent parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.TraceEvent parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.TraceEvent parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.TraceEvent parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.TraceEvent parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.TraceEvent parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.TraceEvent parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.TraceEvent parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.TraceEvent parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.TraceEvent parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.framework.TraceEvent prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * Protobuf type {@code tensorflow.profiler.TraceEvent}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.profiler.TraceEvent)
-      org.tensorflow.proto.framework.TraceEventOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.TraceEventsProtos.internal_static_tensorflow_profiler_TraceEvent_descriptor;
-    }
-
-    @SuppressWarnings({"rawtypes"})
-    protected com.google.protobuf.MapField internalGetMapField(
-        int number) {
-      switch (number) {
-        case 11:
-          return internalGetArgs();
-        default:
-          throw new RuntimeException(
-              "Invalid map field number: " + number);
-      }
-    }
-    @SuppressWarnings({"rawtypes"})
-    protected com.google.protobuf.MapField internalGetMutableMapField(
-        int number) {
-      switch (number) {
-        case 11:
-          return internalGetMutableArgs();
-        default:
-          throw new RuntimeException(
-              "Invalid map field number: " + number);
-      }
-    }
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.TraceEventsProtos.internal_static_tensorflow_profiler_TraceEvent_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.TraceEvent.class, org.tensorflow.proto.framework.TraceEvent.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.framework.TraceEvent.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      deviceId_ = 0;
-
-      resourceId_ = 0;
-
-      name_ = "";
-
-      timestampPs_ = 0L;
-
-      durationPs_ = 0L;
-
-      internalGetMutableArgs().clear();
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.framework.TraceEventsProtos.internal_static_tensorflow_profiler_TraceEvent_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.TraceEvent getDefaultInstanceForType() {
-      return org.tensorflow.proto.framework.TraceEvent.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.TraceEvent build() {
-      org.tensorflow.proto.framework.TraceEvent result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.TraceEvent buildPartial() {
-      org.tensorflow.proto.framework.TraceEvent result = new org.tensorflow.proto.framework.TraceEvent(this);
-      int from_bitField0_ = bitField0_;
-      result.deviceId_ = deviceId_;
-      result.resourceId_ = resourceId_;
-      result.name_ = name_;
-      result.timestampPs_ = timestampPs_;
-      result.durationPs_ = durationPs_;
-      result.args_ = internalGetArgs();
-      result.args_.makeImmutable();
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.framework.TraceEvent) {
-        return mergeFrom((org.tensorflow.proto.framework.TraceEvent)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.framework.TraceEvent other) {
-      if (other == org.tensorflow.proto.framework.TraceEvent.getDefaultInstance()) return this;
-      if (other.getDeviceId() != 0) {
-        setDeviceId(other.getDeviceId());
-      }
-      if (other.getResourceId() != 0) {
-        setResourceId(other.getResourceId());
-      }
-      if (!other.getName().isEmpty()) {
-        name_ = other.name_;
-        onChanged();
-      }
-      if (other.getTimestampPs() != 0L) {
-        setTimestampPs(other.getTimestampPs());
-      }
-      if (other.getDurationPs() != 0L) {
-        setDurationPs(other.getDurationPs());
-      }
-      internalGetMutableArgs().mergeFrom(
-          other.internalGetArgs());
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.framework.TraceEvent parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.framework.TraceEvent) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-    private int bitField0_;
-
-    private int deviceId_ ;
-    /**
-     * <pre>
-     * The id of the device that this event occurred on. The full dataset should
-     * have this device present in the Trace object.
-     * </pre>
-     *
-     * <code>uint32 device_id = 1;</code>
-     */
-    public int getDeviceId() {
-      return deviceId_;
-    }
-    /**
-     * <pre>
-     * The id of the device that this event occurred on. The full dataset should
-     * have this device present in the Trace object.
-     * </pre>
-     *
-     * <code>uint32 device_id = 1;</code>
-     */
-    public Builder setDeviceId(int value) {
-      
-      deviceId_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * The id of the device that this event occurred on. The full dataset should
-     * have this device present in the Trace object.
-     * </pre>
-     *
-     * <code>uint32 device_id = 1;</code>
-     */
-    public Builder clearDeviceId() {
-      
-      deviceId_ = 0;
-      onChanged();
-      return this;
-    }
-
-    private int resourceId_ ;
-    /**
-     * <pre>
-     * The id of the resource that this event occurred on. The full dataset should
-     * have this resource present in the Device object of the Trace object. A
-     * resource_id is unique on a specific device, but not necessarily within the
-     * trace.
-     * </pre>
-     *
-     * <code>uint32 resource_id = 2;</code>
-     */
-    public int getResourceId() {
-      return resourceId_;
-    }
-    /**
-     * <pre>
-     * The id of the resource that this event occurred on. The full dataset should
-     * have this resource present in the Device object of the Trace object. A
-     * resource_id is unique on a specific device, but not necessarily within the
-     * trace.
-     * </pre>
-     *
-     * <code>uint32 resource_id = 2;</code>
-     */
-    public Builder setResourceId(int value) {
-      
-      resourceId_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * The id of the resource that this event occurred on. The full dataset should
-     * have this resource present in the Device object of the Trace object. A
-     * resource_id is unique on a specific device, but not necessarily within the
-     * trace.
-     * </pre>
-     *
-     * <code>uint32 resource_id = 2;</code>
-     */
-    public Builder clearResourceId() {
-      
-      resourceId_ = 0;
-      onChanged();
-      return this;
-    }
-
-    private java.lang.Object name_ = "";
-    /**
-     * <pre>
-     * The name of this trace event.
-     * </pre>
-     *
-     * <code>string name = 3;</code>
-     */
-    public java.lang.String getName() {
-      java.lang.Object ref = name_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        name_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * The name of this trace event.
-     * </pre>
-     *
-     * <code>string name = 3;</code>
-     */
-    public com.google.protobuf.ByteString
-        getNameBytes() {
-      java.lang.Object ref = name_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        name_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * The name of this trace event.
-     * </pre>
-     *
-     * <code>string name = 3;</code>
-     */
-    public Builder setName(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      name_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * The name of this trace event.
-     * </pre>
-     *
-     * <code>string name = 3;</code>
-     */
-    public Builder clearName() {
-      
-      name_ = getDefaultInstance().getName();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * The name of this trace event.
-     * </pre>
-     *
-     * <code>string name = 3;</code>
-     */
-    public Builder setNameBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      name_ = value;
-      onChanged();
-      return this;
-    }
-
-    private long timestampPs_ ;
-    /**
-     * <pre>
-     * The timestamp that this event occurred at (in picos since tracing started).
-     * </pre>
-     *
-     * <code>uint64 timestamp_ps = 9;</code>
-     */
-    public long getTimestampPs() {
-      return timestampPs_;
-    }
-    /**
-     * <pre>
-     * The timestamp that this event occurred at (in picos since tracing started).
-     * </pre>
-     *
-     * <code>uint64 timestamp_ps = 9;</code>
-     */
-    public Builder setTimestampPs(long value) {
-      
-      timestampPs_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * The timestamp that this event occurred at (in picos since tracing started).
-     * </pre>
-     *
-     * <code>uint64 timestamp_ps = 9;</code>
-     */
-    public Builder clearTimestampPs() {
-      
-      timestampPs_ = 0L;
-      onChanged();
-      return this;
-    }
-
-    private long durationPs_ ;
-    /**
-     * <pre>
-     * The duration of the event in picoseconds if applicable.
-     * Events without duration are called instant events.
-     * </pre>
-     *
-     * <code>uint64 duration_ps = 10;</code>
-     */
-    public long getDurationPs() {
-      return durationPs_;
-    }
-    /**
-     * <pre>
-     * The duration of the event in picoseconds if applicable.
-     * Events without duration are called instant events.
-     * </pre>
-     *
-     * <code>uint64 duration_ps = 10;</code>
-     */
-    public Builder setDurationPs(long value) {
-      
-      durationPs_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * The duration of the event in picoseconds if applicable.
-     * Events without duration are called instant events.
-     * </pre>
-     *
-     * <code>uint64 duration_ps = 10;</code>
-     */
-    public Builder clearDurationPs() {
-      
-      durationPs_ = 0L;
-      onChanged();
-      return this;
-    }
-
-    private com.google.protobuf.MapField<
-        java.lang.String, java.lang.String> args_;
-    private com.google.protobuf.MapField<java.lang.String, java.lang.String>
-    internalGetArgs() {
-      if (args_ == null) {
-        return com.google.protobuf.MapField.emptyMapField(
-            ArgsDefaultEntryHolder.defaultEntry);
-      }
-      return args_;
-    }
-    private com.google.protobuf.MapField<java.lang.String, java.lang.String>
-    internalGetMutableArgs() {
-      onChanged();;
-      if (args_ == null) {
-        args_ = com.google.protobuf.MapField.newMapField(
-            ArgsDefaultEntryHolder.defaultEntry);
-      }
-      if (!args_.isMutable()) {
-        args_ = args_.copy();
-      }
-      return args_;
-    }
-
-    public int getArgsCount() {
-      return internalGetArgs().getMap().size();
-    }
-    /**
-     * <pre>
-     * Extra arguments that will be displayed in trace view.
-     * </pre>
-     *
-     * <code>map&lt;string, string&gt; args = 11;</code>
-     */
-
-    public boolean containsArgs(
-        java.lang.String key) {
-      if (key == null) { throw new java.lang.NullPointerException(); }
-      return internalGetArgs().getMap().containsKey(key);
-    }
-    /**
-     * Use {@link #getArgsMap()} instead.
-     */
-    @java.lang.Deprecated
-    public java.util.Map<java.lang.String, java.lang.String> getArgs() {
-      return getArgsMap();
-    }
-    /**
-     * <pre>
-     * Extra arguments that will be displayed in trace view.
-     * </pre>
-     *
-     * <code>map&lt;string, string&gt; args = 11;</code>
-     */
-
-    public java.util.Map<java.lang.String, java.lang.String> getArgsMap() {
-      return internalGetArgs().getMap();
-    }
-    /**
-     * <pre>
-     * Extra arguments that will be displayed in trace view.
-     * </pre>
-     *
-     * <code>map&lt;string, string&gt; args = 11;</code>
-     */
-
-    public java.lang.String getArgsOrDefault(
-        java.lang.String key,
-        java.lang.String defaultValue) {
-      if (key == null) { throw new java.lang.NullPointerException(); }
-      java.util.Map<java.lang.String, java.lang.String> map =
-          internalGetArgs().getMap();
-      return map.containsKey(key) ? map.get(key) : defaultValue;
-    }
-    /**
-     * <pre>
-     * Extra arguments that will be displayed in trace view.
-     * </pre>
-     *
-     * <code>map&lt;string, string&gt; args = 11;</code>
-     */
-
-    public java.lang.String getArgsOrThrow(
-        java.lang.String key) {
-      if (key == null) { throw new java.lang.NullPointerException(); }
-      java.util.Map<java.lang.String, java.lang.String> map =
-          internalGetArgs().getMap();
-      if (!map.containsKey(key)) {
-        throw new java.lang.IllegalArgumentException();
-      }
-      return map.get(key);
-    }
-
-    public Builder clearArgs() {
-      internalGetMutableArgs().getMutableMap()
-          .clear();
-      return this;
-    }
-    /**
-     * <pre>
-     * Extra arguments that will be displayed in trace view.
-     * </pre>
-     *
-     * <code>map&lt;string, string&gt; args = 11;</code>
-     */
-
-    public Builder removeArgs(
-        java.lang.String key) {
-      if (key == null) { throw new java.lang.NullPointerException(); }
-      internalGetMutableArgs().getMutableMap()
-          .remove(key);
-      return this;
-    }
-    /**
-     * Use alternate mutation accessors instead.
-     */
-    @java.lang.Deprecated
-    public java.util.Map<java.lang.String, java.lang.String>
-    getMutableArgs() {
-      return internalGetMutableArgs().getMutableMap();
-    }
-    /**
-     * <pre>
-     * Extra arguments that will be displayed in trace view.
-     * </pre>
-     *
-     * <code>map&lt;string, string&gt; args = 11;</code>
-     */
-    public Builder putArgs(
-        java.lang.String key,
-        java.lang.String value) {
-      if (key == null) { throw new java.lang.NullPointerException(); }
-      if (value == null) { throw new java.lang.NullPointerException(); }
-      internalGetMutableArgs().getMutableMap()
-          .put(key, value);
-      return this;
-    }
-    /**
-     * <pre>
-     * Extra arguments that will be displayed in trace view.
-     * </pre>
-     *
-     * <code>map&lt;string, string&gt; args = 11;</code>
-     */
-
-    public Builder putAllArgs(
-        java.util.Map<java.lang.String, java.lang.String> values) {
-      internalGetMutableArgs().getMutableMap()
-          .putAll(values);
-      return this;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.profiler.TraceEvent)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.profiler.TraceEvent)
-  private static final org.tensorflow.proto.framework.TraceEvent DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.TraceEvent();
-  }
-
-  public static org.tensorflow.proto.framework.TraceEvent getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<TraceEvent>
-      PARSER = new com.google.protobuf.AbstractParser<TraceEvent>() {
-    @java.lang.Override
-    public TraceEvent parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new TraceEvent(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<TraceEvent> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<TraceEvent> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.framework.TraceEvent getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TraceEventOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TraceEventOrBuilder.java
deleted file mode 100644
index 2bebd63d083..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TraceEventOrBuilder.java
+++ /dev/null
@@ -1,122 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/trace_events.proto
-
-package org.tensorflow.proto.framework;
-
-public interface TraceEventOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.profiler.TraceEvent)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <pre>
-   * The id of the device that this event occurred on. The full dataset should
-   * have this device present in the Trace object.
-   * </pre>
-   *
-   * <code>uint32 device_id = 1;</code>
-   */
-  int getDeviceId();
-
-  /**
-   * <pre>
-   * The id of the resource that this event occurred on. The full dataset should
-   * have this resource present in the Device object of the Trace object. A
-   * resource_id is unique on a specific device, but not necessarily within the
-   * trace.
-   * </pre>
-   *
-   * <code>uint32 resource_id = 2;</code>
-   */
-  int getResourceId();
-
-  /**
-   * <pre>
-   * The name of this trace event.
-   * </pre>
-   *
-   * <code>string name = 3;</code>
-   */
-  java.lang.String getName();
-  /**
-   * <pre>
-   * The name of this trace event.
-   * </pre>
-   *
-   * <code>string name = 3;</code>
-   */
-  com.google.protobuf.ByteString
-      getNameBytes();
-
-  /**
-   * <pre>
-   * The timestamp that this event occurred at (in picos since tracing started).
-   * </pre>
-   *
-   * <code>uint64 timestamp_ps = 9;</code>
-   */
-  long getTimestampPs();
-
-  /**
-   * <pre>
-   * The duration of the event in picoseconds if applicable.
-   * Events without duration are called instant events.
-   * </pre>
-   *
-   * <code>uint64 duration_ps = 10;</code>
-   */
-  long getDurationPs();
-
-  /**
-   * <pre>
-   * Extra arguments that will be displayed in trace view.
-   * </pre>
-   *
-   * <code>map&lt;string, string&gt; args = 11;</code>
-   */
-  int getArgsCount();
-  /**
-   * <pre>
-   * Extra arguments that will be displayed in trace view.
-   * </pre>
-   *
-   * <code>map&lt;string, string&gt; args = 11;</code>
-   */
-  boolean containsArgs(
-      java.lang.String key);
-  /**
-   * Use {@link #getArgsMap()} instead.
-   */
-  @java.lang.Deprecated
-  java.util.Map<java.lang.String, java.lang.String>
-  getArgs();
-  /**
-   * <pre>
-   * Extra arguments that will be displayed in trace view.
-   * </pre>
-   *
-   * <code>map&lt;string, string&gt; args = 11;</code>
-   */
-  java.util.Map<java.lang.String, java.lang.String>
-  getArgsMap();
-  /**
-   * <pre>
-   * Extra arguments that will be displayed in trace view.
-   * </pre>
-   *
-   * <code>map&lt;string, string&gt; args = 11;</code>
-   */
-
-  java.lang.String getArgsOrDefault(
-      java.lang.String key,
-      java.lang.String defaultValue);
-  /**
-   * <pre>
-   * Extra arguments that will be displayed in trace view.
-   * </pre>
-   *
-   * <code>map&lt;string, string&gt; args = 11;</code>
-   */
-
-  java.lang.String getArgsOrThrow(
-      java.lang.String key);
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TraceEventsProtos.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TraceEventsProtos.java
deleted file mode 100644
index 4e2e8c8d80a..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TraceEventsProtos.java
+++ /dev/null
@@ -1,133 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/trace_events.proto
-
-package org.tensorflow.proto.framework;
-
-public final class TraceEventsProtos {
-  private TraceEventsProtos() {}
-  public static void registerAllExtensions(
-      com.google.protobuf.ExtensionRegistryLite registry) {
-  }
-
-  public static void registerAllExtensions(
-      com.google.protobuf.ExtensionRegistry registry) {
-    registerAllExtensions(
-        (com.google.protobuf.ExtensionRegistryLite) registry);
-  }
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_profiler_Trace_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_profiler_Trace_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_profiler_Trace_DevicesEntry_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_profiler_Trace_DevicesEntry_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_profiler_Device_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_profiler_Device_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_profiler_Device_ResourcesEntry_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_profiler_Device_ResourcesEntry_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_profiler_Resource_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_profiler_Resource_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_profiler_TraceEvent_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_profiler_TraceEvent_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_profiler_TraceEvent_ArgsEntry_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_profiler_TraceEvent_ArgsEntry_fieldAccessorTable;
-
-  public static com.google.protobuf.Descriptors.FileDescriptor
-      getDescriptor() {
-    return descriptor;
-  }
-  private static  com.google.protobuf.Descriptors.FileDescriptor
-      descriptor;
-  static {
-    java.lang.String[] descriptorData = {
-      "\n+tensorflow/core/protobuf/trace_events." +
-      "proto\022\023tensorflow.profiler\"\305\001\n\005Trace\0228\n\007" +
-      "devices\030\001 \003(\0132\'.tensorflow.profiler.Trac" +
-      "e.DevicesEntry\0225\n\014trace_events\030\004 \003(\0132\037.t" +
-      "ensorflow.profiler.TraceEvent\032K\n\014Devices" +
-      "Entry\022\013\n\003key\030\001 \001(\r\022*\n\005value\030\002 \001(\0132\033.tens" +
-      "orflow.profiler.Device:\0028\001\"\271\001\n\006Device\022\014\n" +
-      "\004name\030\001 \001(\t\022\021\n\tdevice_id\030\002 \001(\r\022=\n\tresour" +
-      "ces\030\003 \003(\0132*.tensorflow.profiler.Device.R" +
-      "esourcesEntry\032O\n\016ResourcesEntry\022\013\n\003key\030\001" +
-      " \001(\r\022,\n\005value\030\002 \001(\0132\035.tensorflow.profile" +
-      "r.Resource:\0028\001\"-\n\010Resource\022\014\n\004name\030\001 \001(\t" +
-      "\022\023\n\013resource_id\030\002 \001(\r\"\323\001\n\nTraceEvent\022\021\n\t" +
-      "device_id\030\001 \001(\r\022\023\n\013resource_id\030\002 \001(\r\022\014\n\004" +
-      "name\030\003 \001(\t\022\024\n\014timestamp_ps\030\t \001(\004\022\023\n\013dura" +
-      "tion_ps\030\n \001(\004\0227\n\004args\030\013 \003(\0132).tensorflow" +
-      ".profiler.TraceEvent.ArgsEntry\032+\n\tArgsEn" +
-      "try\022\013\n\003key\030\001 \001(\t\022\r\n\005value\030\002 \001(\t:\0028\001B\202\001\n\036" +
-      "org.tensorflow.proto.frameworkB\021TraceEve" +
-      "ntsProtosP\001ZHgithub.com/tensorflow/tenso" +
-      "rflow/tensorflow/go/core/core_protos_go_" +
-      "proto\370\001\001b\006proto3"
-    };
-    descriptor = com.google.protobuf.Descriptors.FileDescriptor
-      .internalBuildGeneratedFileFrom(descriptorData,
-        new com.google.protobuf.Descriptors.FileDescriptor[] {
-        });
-    internal_static_tensorflow_profiler_Trace_descriptor =
-      getDescriptor().getMessageTypes().get(0);
-    internal_static_tensorflow_profiler_Trace_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_profiler_Trace_descriptor,
-        new java.lang.String[] { "Devices", "TraceEvents", });
-    internal_static_tensorflow_profiler_Trace_DevicesEntry_descriptor =
-      internal_static_tensorflow_profiler_Trace_descriptor.getNestedTypes().get(0);
-    internal_static_tensorflow_profiler_Trace_DevicesEntry_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_profiler_Trace_DevicesEntry_descriptor,
-        new java.lang.String[] { "Key", "Value", });
-    internal_static_tensorflow_profiler_Device_descriptor =
-      getDescriptor().getMessageTypes().get(1);
-    internal_static_tensorflow_profiler_Device_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_profiler_Device_descriptor,
-        new java.lang.String[] { "Name", "DeviceId", "Resources", });
-    internal_static_tensorflow_profiler_Device_ResourcesEntry_descriptor =
-      internal_static_tensorflow_profiler_Device_descriptor.getNestedTypes().get(0);
-    internal_static_tensorflow_profiler_Device_ResourcesEntry_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_profiler_Device_ResourcesEntry_descriptor,
-        new java.lang.String[] { "Key", "Value", });
-    internal_static_tensorflow_profiler_Resource_descriptor =
-      getDescriptor().getMessageTypes().get(2);
-    internal_static_tensorflow_profiler_Resource_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_profiler_Resource_descriptor,
-        new java.lang.String[] { "Name", "ResourceId", });
-    internal_static_tensorflow_profiler_TraceEvent_descriptor =
-      getDescriptor().getMessageTypes().get(3);
-    internal_static_tensorflow_profiler_TraceEvent_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_profiler_TraceEvent_descriptor,
-        new java.lang.String[] { "DeviceId", "ResourceId", "Name", "TimestampPs", "DurationPs", "Args", });
-    internal_static_tensorflow_profiler_TraceEvent_ArgsEntry_descriptor =
-      internal_static_tensorflow_profiler_TraceEvent_descriptor.getNestedTypes().get(0);
-    internal_static_tensorflow_profiler_TraceEvent_ArgsEntry_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_profiler_TraceEvent_ArgsEntry_descriptor,
-        new java.lang.String[] { "Key", "Value", });
-  }
-
-  // @@protoc_insertion_point(outer_class_scope)
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TraceOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TraceOrBuilder.java
deleted file mode 100644
index 78fa25b0aa9..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TraceOrBuilder.java
+++ /dev/null
@@ -1,112 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/trace_events.proto
-
-package org.tensorflow.proto.framework;
-
-public interface TraceOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.profiler.Trace)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <pre>
-   * The devices that this trace has information about. Maps from device_id to
-   * more data about the specific device.
-   * </pre>
-   *
-   * <code>map&lt;uint32, .tensorflow.profiler.Device&gt; devices = 1;</code>
-   */
-  int getDevicesCount();
-  /**
-   * <pre>
-   * The devices that this trace has information about. Maps from device_id to
-   * more data about the specific device.
-   * </pre>
-   *
-   * <code>map&lt;uint32, .tensorflow.profiler.Device&gt; devices = 1;</code>
-   */
-  boolean containsDevices(
-      int key);
-  /**
-   * Use {@link #getDevicesMap()} instead.
-   */
-  @java.lang.Deprecated
-  java.util.Map<java.lang.Integer, org.tensorflow.proto.framework.Device>
-  getDevices();
-  /**
-   * <pre>
-   * The devices that this trace has information about. Maps from device_id to
-   * more data about the specific device.
-   * </pre>
-   *
-   * <code>map&lt;uint32, .tensorflow.profiler.Device&gt; devices = 1;</code>
-   */
-  java.util.Map<java.lang.Integer, org.tensorflow.proto.framework.Device>
-  getDevicesMap();
-  /**
-   * <pre>
-   * The devices that this trace has information about. Maps from device_id to
-   * more data about the specific device.
-   * </pre>
-   *
-   * <code>map&lt;uint32, .tensorflow.profiler.Device&gt; devices = 1;</code>
-   */
-
-  org.tensorflow.proto.framework.Device getDevicesOrDefault(
-      int key,
-      org.tensorflow.proto.framework.Device defaultValue);
-  /**
-   * <pre>
-   * The devices that this trace has information about. Maps from device_id to
-   * more data about the specific device.
-   * </pre>
-   *
-   * <code>map&lt;uint32, .tensorflow.profiler.Device&gt; devices = 1;</code>
-   */
-
-  org.tensorflow.proto.framework.Device getDevicesOrThrow(
-      int key);
-
-  /**
-   * <pre>
-   * All trace events capturing in the profiling period.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.profiler.TraceEvent trace_events = 4;</code>
-   */
-  java.util.List<org.tensorflow.proto.framework.TraceEvent> 
-      getTraceEventsList();
-  /**
-   * <pre>
-   * All trace events capturing in the profiling period.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.profiler.TraceEvent trace_events = 4;</code>
-   */
-  org.tensorflow.proto.framework.TraceEvent getTraceEvents(int index);
-  /**
-   * <pre>
-   * All trace events capturing in the profiling period.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.profiler.TraceEvent trace_events = 4;</code>
-   */
-  int getTraceEventsCount();
-  /**
-   * <pre>
-   * All trace events capturing in the profiling period.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.profiler.TraceEvent trace_events = 4;</code>
-   */
-  java.util.List<? extends org.tensorflow.proto.framework.TraceEventOrBuilder> 
-      getTraceEventsOrBuilderList();
-  /**
-   * <pre>
-   * All trace events capturing in the profiling period.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.profiler.TraceEvent trace_events = 4;</code>
-   */
-  org.tensorflow.proto.framework.TraceEventOrBuilder getTraceEventsOrBuilder(
-      int index);
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TrackableObjectGraph.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TrackableObjectGraph.java
deleted file mode 100644
index 5d516f926dc..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TrackableObjectGraph.java
+++ /dev/null
@@ -1,5141 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/trackable_object_graph.proto
-
-package org.tensorflow.proto.framework;
-
-/**
- * Protobuf type {@code tensorflow.TrackableObjectGraph}
- */
-public  final class TrackableObjectGraph extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.TrackableObjectGraph)
-    TrackableObjectGraphOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use TrackableObjectGraph.newBuilder() to construct.
-  private TrackableObjectGraph(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private TrackableObjectGraph() {
-    nodes_ = java.util.Collections.emptyList();
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new TrackableObjectGraph();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private TrackableObjectGraph(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    int mutable_bitField0_ = 0;
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 10: {
-            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
-              nodes_ = new java.util.ArrayList<org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject>();
-              mutable_bitField0_ |= 0x00000001;
-            }
-            nodes_.add(
-                input.readMessage(org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.parser(), extensionRegistry));
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      if (((mutable_bitField0_ & 0x00000001) != 0)) {
-        nodes_ = java.util.Collections.unmodifiableList(nodes_);
-      }
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.framework.TrackableObjectGraphProtos.internal_static_tensorflow_TrackableObjectGraph_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.framework.TrackableObjectGraphProtos.internal_static_tensorflow_TrackableObjectGraph_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.framework.TrackableObjectGraph.class, org.tensorflow.proto.framework.TrackableObjectGraph.Builder.class);
-  }
-
-  public interface TrackableObjectOrBuilder extends
-      // @@protoc_insertion_point(interface_extends:tensorflow.TrackableObjectGraph.TrackableObject)
-      com.google.protobuf.MessageOrBuilder {
-
-    /**
-     * <pre>
-     * Objects which this object depends on.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.ObjectReference children = 1;</code>
-     */
-    java.util.List<org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference> 
-        getChildrenList();
-    /**
-     * <pre>
-     * Objects which this object depends on.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.ObjectReference children = 1;</code>
-     */
-    org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference getChildren(int index);
-    /**
-     * <pre>
-     * Objects which this object depends on.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.ObjectReference children = 1;</code>
-     */
-    int getChildrenCount();
-    /**
-     * <pre>
-     * Objects which this object depends on.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.ObjectReference children = 1;</code>
-     */
-    java.util.List<? extends org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReferenceOrBuilder> 
-        getChildrenOrBuilderList();
-    /**
-     * <pre>
-     * Objects which this object depends on.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.ObjectReference children = 1;</code>
-     */
-    org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReferenceOrBuilder getChildrenOrBuilder(
-        int index);
-
-    /**
-     * <pre>
-     * Serialized data specific to this object.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SerializedTensor attributes = 2;</code>
-     */
-    java.util.List<org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensor> 
-        getAttributesList();
-    /**
-     * <pre>
-     * Serialized data specific to this object.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SerializedTensor attributes = 2;</code>
-     */
-    org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensor getAttributes(int index);
-    /**
-     * <pre>
-     * Serialized data specific to this object.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SerializedTensor attributes = 2;</code>
-     */
-    int getAttributesCount();
-    /**
-     * <pre>
-     * Serialized data specific to this object.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SerializedTensor attributes = 2;</code>
-     */
-    java.util.List<? extends org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensorOrBuilder> 
-        getAttributesOrBuilderList();
-    /**
-     * <pre>
-     * Serialized data specific to this object.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SerializedTensor attributes = 2;</code>
-     */
-    org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensorOrBuilder getAttributesOrBuilder(
-        int index);
-
-    /**
-     * <pre>
-     * Slot variables owned by this object.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SlotVariableReference slot_variables = 3;</code>
-     */
-    java.util.List<org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference> 
-        getSlotVariablesList();
-    /**
-     * <pre>
-     * Slot variables owned by this object.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SlotVariableReference slot_variables = 3;</code>
-     */
-    org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference getSlotVariables(int index);
-    /**
-     * <pre>
-     * Slot variables owned by this object.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SlotVariableReference slot_variables = 3;</code>
-     */
-    int getSlotVariablesCount();
-    /**
-     * <pre>
-     * Slot variables owned by this object.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SlotVariableReference slot_variables = 3;</code>
-     */
-    java.util.List<? extends org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReferenceOrBuilder> 
-        getSlotVariablesOrBuilderList();
-    /**
-     * <pre>
-     * Slot variables owned by this object.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SlotVariableReference slot_variables = 3;</code>
-     */
-    org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReferenceOrBuilder getSlotVariablesOrBuilder(
-        int index);
-  }
-  /**
-   * Protobuf type {@code tensorflow.TrackableObjectGraph.TrackableObject}
-   */
-  public  static final class TrackableObject extends
-      com.google.protobuf.GeneratedMessageV3 implements
-      // @@protoc_insertion_point(message_implements:tensorflow.TrackableObjectGraph.TrackableObject)
-      TrackableObjectOrBuilder {
-  private static final long serialVersionUID = 0L;
-    // Use TrackableObject.newBuilder() to construct.
-    private TrackableObject(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-      super(builder);
-    }
-    private TrackableObject() {
-      children_ = java.util.Collections.emptyList();
-      attributes_ = java.util.Collections.emptyList();
-      slotVariables_ = java.util.Collections.emptyList();
-    }
-
-    @java.lang.Override
-    @SuppressWarnings({"unused"})
-    protected java.lang.Object newInstance(
-        UnusedPrivateParameter unused) {
-      return new TrackableObject();
-    }
-
-    @java.lang.Override
-    public final com.google.protobuf.UnknownFieldSet
-    getUnknownFields() {
-      return this.unknownFields;
-    }
-    private TrackableObject(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      this();
-      if (extensionRegistry == null) {
-        throw new java.lang.NullPointerException();
-      }
-      int mutable_bitField0_ = 0;
-      com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-          com.google.protobuf.UnknownFieldSet.newBuilder();
-      try {
-        boolean done = false;
-        while (!done) {
-          int tag = input.readTag();
-          switch (tag) {
-            case 0:
-              done = true;
-              break;
-            case 10: {
-              if (!((mutable_bitField0_ & 0x00000001) != 0)) {
-                children_ = new java.util.ArrayList<org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference>();
-                mutable_bitField0_ |= 0x00000001;
-              }
-              children_.add(
-                  input.readMessage(org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference.parser(), extensionRegistry));
-              break;
-            }
-            case 18: {
-              if (!((mutable_bitField0_ & 0x00000002) != 0)) {
-                attributes_ = new java.util.ArrayList<org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensor>();
-                mutable_bitField0_ |= 0x00000002;
-              }
-              attributes_.add(
-                  input.readMessage(org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensor.parser(), extensionRegistry));
-              break;
-            }
-            case 26: {
-              if (!((mutable_bitField0_ & 0x00000004) != 0)) {
-                slotVariables_ = new java.util.ArrayList<org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference>();
-                mutable_bitField0_ |= 0x00000004;
-              }
-              slotVariables_.add(
-                  input.readMessage(org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference.parser(), extensionRegistry));
-              break;
-            }
-            default: {
-              if (!parseUnknownField(
-                  input, unknownFields, extensionRegistry, tag)) {
-                done = true;
-              }
-              break;
-            }
-          }
-        }
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        throw e.setUnfinishedMessage(this);
-      } catch (java.io.IOException e) {
-        throw new com.google.protobuf.InvalidProtocolBufferException(
-            e).setUnfinishedMessage(this);
-      } finally {
-        if (((mutable_bitField0_ & 0x00000001) != 0)) {
-          children_ = java.util.Collections.unmodifiableList(children_);
-        }
-        if (((mutable_bitField0_ & 0x00000002) != 0)) {
-          attributes_ = java.util.Collections.unmodifiableList(attributes_);
-        }
-        if (((mutable_bitField0_ & 0x00000004) != 0)) {
-          slotVariables_ = java.util.Collections.unmodifiableList(slotVariables_);
-        }
-        this.unknownFields = unknownFields.build();
-        makeExtensionsImmutable();
-      }
-    }
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.TrackableObjectGraphProtos.internal_static_tensorflow_TrackableObjectGraph_TrackableObject_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.TrackableObjectGraphProtos.internal_static_tensorflow_TrackableObjectGraph_TrackableObject_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.class, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.Builder.class);
-    }
-
-    public interface ObjectReferenceOrBuilder extends
-        // @@protoc_insertion_point(interface_extends:tensorflow.TrackableObjectGraph.TrackableObject.ObjectReference)
-        com.google.protobuf.MessageOrBuilder {
-
-      /**
-       * <pre>
-       * An index into `TrackableObjectGraph.nodes`, indicating the object
-       * being referenced.
-       * </pre>
-       *
-       * <code>int32 node_id = 1;</code>
-       */
-      int getNodeId();
-
-      /**
-       * <pre>
-       * A user-provided name for the edge.
-       * </pre>
-       *
-       * <code>string local_name = 2;</code>
-       */
-      java.lang.String getLocalName();
-      /**
-       * <pre>
-       * A user-provided name for the edge.
-       * </pre>
-       *
-       * <code>string local_name = 2;</code>
-       */
-      com.google.protobuf.ByteString
-          getLocalNameBytes();
-    }
-    /**
-     * Protobuf type {@code tensorflow.TrackableObjectGraph.TrackableObject.ObjectReference}
-     */
-    public  static final class ObjectReference extends
-        com.google.protobuf.GeneratedMessageV3 implements
-        // @@protoc_insertion_point(message_implements:tensorflow.TrackableObjectGraph.TrackableObject.ObjectReference)
-        ObjectReferenceOrBuilder {
-    private static final long serialVersionUID = 0L;
-      // Use ObjectReference.newBuilder() to construct.
-      private ObjectReference(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-        super(builder);
-      }
-      private ObjectReference() {
-        localName_ = "";
-      }
-
-      @java.lang.Override
-      @SuppressWarnings({"unused"})
-      protected java.lang.Object newInstance(
-          UnusedPrivateParameter unused) {
-        return new ObjectReference();
-      }
-
-      @java.lang.Override
-      public final com.google.protobuf.UnknownFieldSet
-      getUnknownFields() {
-        return this.unknownFields;
-      }
-      private ObjectReference(
-          com.google.protobuf.CodedInputStream input,
-          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-          throws com.google.protobuf.InvalidProtocolBufferException {
-        this();
-        if (extensionRegistry == null) {
-          throw new java.lang.NullPointerException();
-        }
-        com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-            com.google.protobuf.UnknownFieldSet.newBuilder();
-        try {
-          boolean done = false;
-          while (!done) {
-            int tag = input.readTag();
-            switch (tag) {
-              case 0:
-                done = true;
-                break;
-              case 8: {
-
-                nodeId_ = input.readInt32();
-                break;
-              }
-              case 18: {
-                java.lang.String s = input.readStringRequireUtf8();
-
-                localName_ = s;
-                break;
-              }
-              default: {
-                if (!parseUnknownField(
-                    input, unknownFields, extensionRegistry, tag)) {
-                  done = true;
-                }
-                break;
-              }
-            }
-          }
-        } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-          throw e.setUnfinishedMessage(this);
-        } catch (java.io.IOException e) {
-          throw new com.google.protobuf.InvalidProtocolBufferException(
-              e).setUnfinishedMessage(this);
-        } finally {
-          this.unknownFields = unknownFields.build();
-          makeExtensionsImmutable();
-        }
-      }
-      public static final com.google.protobuf.Descriptors.Descriptor
-          getDescriptor() {
-        return org.tensorflow.proto.framework.TrackableObjectGraphProtos.internal_static_tensorflow_TrackableObjectGraph_TrackableObject_ObjectReference_descriptor;
-      }
-
-      @java.lang.Override
-      protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-          internalGetFieldAccessorTable() {
-        return org.tensorflow.proto.framework.TrackableObjectGraphProtos.internal_static_tensorflow_TrackableObjectGraph_TrackableObject_ObjectReference_fieldAccessorTable
-            .ensureFieldAccessorsInitialized(
-                org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference.class, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference.Builder.class);
-      }
-
-      public static final int NODE_ID_FIELD_NUMBER = 1;
-      private int nodeId_;
-      /**
-       * <pre>
-       * An index into `TrackableObjectGraph.nodes`, indicating the object
-       * being referenced.
-       * </pre>
-       *
-       * <code>int32 node_id = 1;</code>
-       */
-      public int getNodeId() {
-        return nodeId_;
-      }
-
-      public static final int LOCAL_NAME_FIELD_NUMBER = 2;
-      private volatile java.lang.Object localName_;
-      /**
-       * <pre>
-       * A user-provided name for the edge.
-       * </pre>
-       *
-       * <code>string local_name = 2;</code>
-       */
-      public java.lang.String getLocalName() {
-        java.lang.Object ref = localName_;
-        if (ref instanceof java.lang.String) {
-          return (java.lang.String) ref;
-        } else {
-          com.google.protobuf.ByteString bs = 
-              (com.google.protobuf.ByteString) ref;
-          java.lang.String s = bs.toStringUtf8();
-          localName_ = s;
-          return s;
-        }
-      }
-      /**
-       * <pre>
-       * A user-provided name for the edge.
-       * </pre>
-       *
-       * <code>string local_name = 2;</code>
-       */
-      public com.google.protobuf.ByteString
-          getLocalNameBytes() {
-        java.lang.Object ref = localName_;
-        if (ref instanceof java.lang.String) {
-          com.google.protobuf.ByteString b = 
-              com.google.protobuf.ByteString.copyFromUtf8(
-                  (java.lang.String) ref);
-          localName_ = b;
-          return b;
-        } else {
-          return (com.google.protobuf.ByteString) ref;
-        }
-      }
-
-      private byte memoizedIsInitialized = -1;
-      @java.lang.Override
-      public final boolean isInitialized() {
-        byte isInitialized = memoizedIsInitialized;
-        if (isInitialized == 1) return true;
-        if (isInitialized == 0) return false;
-
-        memoizedIsInitialized = 1;
-        return true;
-      }
-
-      @java.lang.Override
-      public void writeTo(com.google.protobuf.CodedOutputStream output)
-                          throws java.io.IOException {
-        if (nodeId_ != 0) {
-          output.writeInt32(1, nodeId_);
-        }
-        if (!getLocalNameBytes().isEmpty()) {
-          com.google.protobuf.GeneratedMessageV3.writeString(output, 2, localName_);
-        }
-        unknownFields.writeTo(output);
-      }
-
-      @java.lang.Override
-      public int getSerializedSize() {
-        int size = memoizedSize;
-        if (size != -1) return size;
-
-        size = 0;
-        if (nodeId_ != 0) {
-          size += com.google.protobuf.CodedOutputStream
-            .computeInt32Size(1, nodeId_);
-        }
-        if (!getLocalNameBytes().isEmpty()) {
-          size += com.google.protobuf.GeneratedMessageV3.computeStringSize(2, localName_);
-        }
-        size += unknownFields.getSerializedSize();
-        memoizedSize = size;
-        return size;
-      }
-
-      @java.lang.Override
-      public boolean equals(final java.lang.Object obj) {
-        if (obj == this) {
-         return true;
-        }
-        if (!(obj instanceof org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference)) {
-          return super.equals(obj);
-        }
-        org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference other = (org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference) obj;
-
-        if (getNodeId()
-            != other.getNodeId()) return false;
-        if (!getLocalName()
-            .equals(other.getLocalName())) return false;
-        if (!unknownFields.equals(other.unknownFields)) return false;
-        return true;
-      }
-
-      @java.lang.Override
-      public int hashCode() {
-        if (memoizedHashCode != 0) {
-          return memoizedHashCode;
-        }
-        int hash = 41;
-        hash = (19 * hash) + getDescriptor().hashCode();
-        hash = (37 * hash) + NODE_ID_FIELD_NUMBER;
-        hash = (53 * hash) + getNodeId();
-        hash = (37 * hash) + LOCAL_NAME_FIELD_NUMBER;
-        hash = (53 * hash) + getLocalName().hashCode();
-        hash = (29 * hash) + unknownFields.hashCode();
-        memoizedHashCode = hash;
-        return hash;
-      }
-
-      public static org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference parseFrom(
-          java.nio.ByteBuffer data)
-          throws com.google.protobuf.InvalidProtocolBufferException {
-        return PARSER.parseFrom(data);
-      }
-      public static org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference parseFrom(
-          java.nio.ByteBuffer data,
-          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-          throws com.google.protobuf.InvalidProtocolBufferException {
-        return PARSER.parseFrom(data, extensionRegistry);
-      }
-      public static org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference parseFrom(
-          com.google.protobuf.ByteString data)
-          throws com.google.protobuf.InvalidProtocolBufferException {
-        return PARSER.parseFrom(data);
-      }
-      public static org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference parseFrom(
-          com.google.protobuf.ByteString data,
-          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-          throws com.google.protobuf.InvalidProtocolBufferException {
-        return PARSER.parseFrom(data, extensionRegistry);
-      }
-      public static org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference parseFrom(byte[] data)
-          throws com.google.protobuf.InvalidProtocolBufferException {
-        return PARSER.parseFrom(data);
-      }
-      public static org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference parseFrom(
-          byte[] data,
-          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-          throws com.google.protobuf.InvalidProtocolBufferException {
-        return PARSER.parseFrom(data, extensionRegistry);
-      }
-      public static org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference parseFrom(java.io.InputStream input)
-          throws java.io.IOException {
-        return com.google.protobuf.GeneratedMessageV3
-            .parseWithIOException(PARSER, input);
-      }
-      public static org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference parseFrom(
-          java.io.InputStream input,
-          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-          throws java.io.IOException {
-        return com.google.protobuf.GeneratedMessageV3
-            .parseWithIOException(PARSER, input, extensionRegistry);
-      }
-      public static org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference parseDelimitedFrom(java.io.InputStream input)
-          throws java.io.IOException {
-        return com.google.protobuf.GeneratedMessageV3
-            .parseDelimitedWithIOException(PARSER, input);
-      }
-      public static org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference parseDelimitedFrom(
-          java.io.InputStream input,
-          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-          throws java.io.IOException {
-        return com.google.protobuf.GeneratedMessageV3
-            .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-      }
-      public static org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference parseFrom(
-          com.google.protobuf.CodedInputStream input)
-          throws java.io.IOException {
-        return com.google.protobuf.GeneratedMessageV3
-            .parseWithIOException(PARSER, input);
-      }
-      public static org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference parseFrom(
-          com.google.protobuf.CodedInputStream input,
-          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-          throws java.io.IOException {
-        return com.google.protobuf.GeneratedMessageV3
-            .parseWithIOException(PARSER, input, extensionRegistry);
-      }
-
-      @java.lang.Override
-      public Builder newBuilderForType() { return newBuilder(); }
-      public static Builder newBuilder() {
-        return DEFAULT_INSTANCE.toBuilder();
-      }
-      public static Builder newBuilder(org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference prototype) {
-        return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-      }
-      @java.lang.Override
-      public Builder toBuilder() {
-        return this == DEFAULT_INSTANCE
-            ? new Builder() : new Builder().mergeFrom(this);
-      }
-
-      @java.lang.Override
-      protected Builder newBuilderForType(
-          com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-        Builder builder = new Builder(parent);
-        return builder;
-      }
-      /**
-       * Protobuf type {@code tensorflow.TrackableObjectGraph.TrackableObject.ObjectReference}
-       */
-      public static final class Builder extends
-          com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-          // @@protoc_insertion_point(builder_implements:tensorflow.TrackableObjectGraph.TrackableObject.ObjectReference)
-          org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReferenceOrBuilder {
-        public static final com.google.protobuf.Descriptors.Descriptor
-            getDescriptor() {
-          return org.tensorflow.proto.framework.TrackableObjectGraphProtos.internal_static_tensorflow_TrackableObjectGraph_TrackableObject_ObjectReference_descriptor;
-        }
-
-        @java.lang.Override
-        protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-            internalGetFieldAccessorTable() {
-          return org.tensorflow.proto.framework.TrackableObjectGraphProtos.internal_static_tensorflow_TrackableObjectGraph_TrackableObject_ObjectReference_fieldAccessorTable
-              .ensureFieldAccessorsInitialized(
-                  org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference.class, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference.Builder.class);
-        }
-
-        // Construct using org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference.newBuilder()
-        private Builder() {
-          maybeForceBuilderInitialization();
-        }
-
-        private Builder(
-            com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-          super(parent);
-          maybeForceBuilderInitialization();
-        }
-        private void maybeForceBuilderInitialization() {
-          if (com.google.protobuf.GeneratedMessageV3
-                  .alwaysUseFieldBuilders) {
-          }
-        }
-        @java.lang.Override
-        public Builder clear() {
-          super.clear();
-          nodeId_ = 0;
-
-          localName_ = "";
-
-          return this;
-        }
-
-        @java.lang.Override
-        public com.google.protobuf.Descriptors.Descriptor
-            getDescriptorForType() {
-          return org.tensorflow.proto.framework.TrackableObjectGraphProtos.internal_static_tensorflow_TrackableObjectGraph_TrackableObject_ObjectReference_descriptor;
-        }
-
-        @java.lang.Override
-        public org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference getDefaultInstanceForType() {
-          return org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference.getDefaultInstance();
-        }
-
-        @java.lang.Override
-        public org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference build() {
-          org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference result = buildPartial();
-          if (!result.isInitialized()) {
-            throw newUninitializedMessageException(result);
-          }
-          return result;
-        }
-
-        @java.lang.Override
-        public org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference buildPartial() {
-          org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference result = new org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference(this);
-          result.nodeId_ = nodeId_;
-          result.localName_ = localName_;
-          onBuilt();
-          return result;
-        }
-
-        @java.lang.Override
-        public Builder clone() {
-          return super.clone();
-        }
-        @java.lang.Override
-        public Builder setField(
-            com.google.protobuf.Descriptors.FieldDescriptor field,
-            java.lang.Object value) {
-          return super.setField(field, value);
-        }
-        @java.lang.Override
-        public Builder clearField(
-            com.google.protobuf.Descriptors.FieldDescriptor field) {
-          return super.clearField(field);
-        }
-        @java.lang.Override
-        public Builder clearOneof(
-            com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-          return super.clearOneof(oneof);
-        }
-        @java.lang.Override
-        public Builder setRepeatedField(
-            com.google.protobuf.Descriptors.FieldDescriptor field,
-            int index, java.lang.Object value) {
-          return super.setRepeatedField(field, index, value);
-        }
-        @java.lang.Override
-        public Builder addRepeatedField(
-            com.google.protobuf.Descriptors.FieldDescriptor field,
-            java.lang.Object value) {
-          return super.addRepeatedField(field, value);
-        }
-        @java.lang.Override
-        public Builder mergeFrom(com.google.protobuf.Message other) {
-          if (other instanceof org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference) {
-            return mergeFrom((org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference)other);
-          } else {
-            super.mergeFrom(other);
-            return this;
-          }
-        }
-
-        public Builder mergeFrom(org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference other) {
-          if (other == org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference.getDefaultInstance()) return this;
-          if (other.getNodeId() != 0) {
-            setNodeId(other.getNodeId());
-          }
-          if (!other.getLocalName().isEmpty()) {
-            localName_ = other.localName_;
-            onChanged();
-          }
-          this.mergeUnknownFields(other.unknownFields);
-          onChanged();
-          return this;
-        }
-
-        @java.lang.Override
-        public final boolean isInitialized() {
-          return true;
-        }
-
-        @java.lang.Override
-        public Builder mergeFrom(
-            com.google.protobuf.CodedInputStream input,
-            com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-            throws java.io.IOException {
-          org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference parsedMessage = null;
-          try {
-            parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-          } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-            parsedMessage = (org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference) e.getUnfinishedMessage();
-            throw e.unwrapIOException();
-          } finally {
-            if (parsedMessage != null) {
-              mergeFrom(parsedMessage);
-            }
-          }
-          return this;
-        }
-
-        private int nodeId_ ;
-        /**
-         * <pre>
-         * An index into `TrackableObjectGraph.nodes`, indicating the object
-         * being referenced.
-         * </pre>
-         *
-         * <code>int32 node_id = 1;</code>
-         */
-        public int getNodeId() {
-          return nodeId_;
-        }
-        /**
-         * <pre>
-         * An index into `TrackableObjectGraph.nodes`, indicating the object
-         * being referenced.
-         * </pre>
-         *
-         * <code>int32 node_id = 1;</code>
-         */
-        public Builder setNodeId(int value) {
-          
-          nodeId_ = value;
-          onChanged();
-          return this;
-        }
-        /**
-         * <pre>
-         * An index into `TrackableObjectGraph.nodes`, indicating the object
-         * being referenced.
-         * </pre>
-         *
-         * <code>int32 node_id = 1;</code>
-         */
-        public Builder clearNodeId() {
-          
-          nodeId_ = 0;
-          onChanged();
-          return this;
-        }
-
-        private java.lang.Object localName_ = "";
-        /**
-         * <pre>
-         * A user-provided name for the edge.
-         * </pre>
-         *
-         * <code>string local_name = 2;</code>
-         */
-        public java.lang.String getLocalName() {
-          java.lang.Object ref = localName_;
-          if (!(ref instanceof java.lang.String)) {
-            com.google.protobuf.ByteString bs =
-                (com.google.protobuf.ByteString) ref;
-            java.lang.String s = bs.toStringUtf8();
-            localName_ = s;
-            return s;
-          } else {
-            return (java.lang.String) ref;
-          }
-        }
-        /**
-         * <pre>
-         * A user-provided name for the edge.
-         * </pre>
-         *
-         * <code>string local_name = 2;</code>
-         */
-        public com.google.protobuf.ByteString
-            getLocalNameBytes() {
-          java.lang.Object ref = localName_;
-          if (ref instanceof String) {
-            com.google.protobuf.ByteString b = 
-                com.google.protobuf.ByteString.copyFromUtf8(
-                    (java.lang.String) ref);
-            localName_ = b;
-            return b;
-          } else {
-            return (com.google.protobuf.ByteString) ref;
-          }
-        }
-        /**
-         * <pre>
-         * A user-provided name for the edge.
-         * </pre>
-         *
-         * <code>string local_name = 2;</code>
-         */
-        public Builder setLocalName(
-            java.lang.String value) {
-          if (value == null) {
-    throw new NullPointerException();
-  }
-  
-          localName_ = value;
-          onChanged();
-          return this;
-        }
-        /**
-         * <pre>
-         * A user-provided name for the edge.
-         * </pre>
-         *
-         * <code>string local_name = 2;</code>
-         */
-        public Builder clearLocalName() {
-          
-          localName_ = getDefaultInstance().getLocalName();
-          onChanged();
-          return this;
-        }
-        /**
-         * <pre>
-         * A user-provided name for the edge.
-         * </pre>
-         *
-         * <code>string local_name = 2;</code>
-         */
-        public Builder setLocalNameBytes(
-            com.google.protobuf.ByteString value) {
-          if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-          
-          localName_ = value;
-          onChanged();
-          return this;
-        }
-        @java.lang.Override
-        public final Builder setUnknownFields(
-            final com.google.protobuf.UnknownFieldSet unknownFields) {
-          return super.setUnknownFields(unknownFields);
-        }
-
-        @java.lang.Override
-        public final Builder mergeUnknownFields(
-            final com.google.protobuf.UnknownFieldSet unknownFields) {
-          return super.mergeUnknownFields(unknownFields);
-        }
-
-
-        // @@protoc_insertion_point(builder_scope:tensorflow.TrackableObjectGraph.TrackableObject.ObjectReference)
-      }
-
-      // @@protoc_insertion_point(class_scope:tensorflow.TrackableObjectGraph.TrackableObject.ObjectReference)
-      private static final org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference DEFAULT_INSTANCE;
-      static {
-        DEFAULT_INSTANCE = new org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference();
-      }
-
-      public static org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference getDefaultInstance() {
-        return DEFAULT_INSTANCE;
-      }
-
-      private static final com.google.protobuf.Parser<ObjectReference>
-          PARSER = new com.google.protobuf.AbstractParser<ObjectReference>() {
-        @java.lang.Override
-        public ObjectReference parsePartialFrom(
-            com.google.protobuf.CodedInputStream input,
-            com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-            throws com.google.protobuf.InvalidProtocolBufferException {
-          return new ObjectReference(input, extensionRegistry);
-        }
-      };
-
-      public static com.google.protobuf.Parser<ObjectReference> parser() {
-        return PARSER;
-      }
-
-      @java.lang.Override
-      public com.google.protobuf.Parser<ObjectReference> getParserForType() {
-        return PARSER;
-      }
-
-      @java.lang.Override
-      public org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference getDefaultInstanceForType() {
-        return DEFAULT_INSTANCE;
-      }
-
-    }
-
-    public interface SerializedTensorOrBuilder extends
-        // @@protoc_insertion_point(interface_extends:tensorflow.TrackableObjectGraph.TrackableObject.SerializedTensor)
-        com.google.protobuf.MessageOrBuilder {
-
-      /**
-       * <pre>
-       * A name for the Tensor. Simple variables have only one
-       * `SerializedTensor` named "VARIABLE_VALUE" by convention. This value may
-       * be restored on object creation as an optimization.
-       * </pre>
-       *
-       * <code>string name = 1;</code>
-       */
-      java.lang.String getName();
-      /**
-       * <pre>
-       * A name for the Tensor. Simple variables have only one
-       * `SerializedTensor` named "VARIABLE_VALUE" by convention. This value may
-       * be restored on object creation as an optimization.
-       * </pre>
-       *
-       * <code>string name = 1;</code>
-       */
-      com.google.protobuf.ByteString
-          getNameBytes();
-
-      /**
-       * <pre>
-       * The full name of the variable/tensor, if applicable. Used to allow
-       * name-based loading of checkpoints which were saved using an
-       * object-based API. Should match the checkpoint key which would have been
-       * assigned by tf.train.Saver.
-       * </pre>
-       *
-       * <code>string full_name = 2;</code>
-       */
-      java.lang.String getFullName();
-      /**
-       * <pre>
-       * The full name of the variable/tensor, if applicable. Used to allow
-       * name-based loading of checkpoints which were saved using an
-       * object-based API. Should match the checkpoint key which would have been
-       * assigned by tf.train.Saver.
-       * </pre>
-       *
-       * <code>string full_name = 2;</code>
-       */
-      com.google.protobuf.ByteString
-          getFullNameBytes();
-
-      /**
-       * <pre>
-       * The generated name of the Tensor in the checkpoint.
-       * </pre>
-       *
-       * <code>string checkpoint_key = 3;</code>
-       */
-      java.lang.String getCheckpointKey();
-      /**
-       * <pre>
-       * The generated name of the Tensor in the checkpoint.
-       * </pre>
-       *
-       * <code>string checkpoint_key = 3;</code>
-       */
-      com.google.protobuf.ByteString
-          getCheckpointKeyBytes();
-
-      /**
-       * <pre>
-       * Whether checkpoints should be considered as matching even without this
-       * value restored. Used for non-critical values which don't affect the
-       * TensorFlow graph, such as layer configurations.
-       * </pre>
-       *
-       * <code>bool optional_restore = 4;</code>
-       */
-      boolean getOptionalRestore();
-    }
-    /**
-     * Protobuf type {@code tensorflow.TrackableObjectGraph.TrackableObject.SerializedTensor}
-     */
-    public  static final class SerializedTensor extends
-        com.google.protobuf.GeneratedMessageV3 implements
-        // @@protoc_insertion_point(message_implements:tensorflow.TrackableObjectGraph.TrackableObject.SerializedTensor)
-        SerializedTensorOrBuilder {
-    private static final long serialVersionUID = 0L;
-      // Use SerializedTensor.newBuilder() to construct.
-      private SerializedTensor(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-        super(builder);
-      }
-      private SerializedTensor() {
-        name_ = "";
-        fullName_ = "";
-        checkpointKey_ = "";
-      }
-
-      @java.lang.Override
-      @SuppressWarnings({"unused"})
-      protected java.lang.Object newInstance(
-          UnusedPrivateParameter unused) {
-        return new SerializedTensor();
-      }
-
-      @java.lang.Override
-      public final com.google.protobuf.UnknownFieldSet
-      getUnknownFields() {
-        return this.unknownFields;
-      }
-      private SerializedTensor(
-          com.google.protobuf.CodedInputStream input,
-          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-          throws com.google.protobuf.InvalidProtocolBufferException {
-        this();
-        if (extensionRegistry == null) {
-          throw new java.lang.NullPointerException();
-        }
-        com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-            com.google.protobuf.UnknownFieldSet.newBuilder();
-        try {
-          boolean done = false;
-          while (!done) {
-            int tag = input.readTag();
-            switch (tag) {
-              case 0:
-                done = true;
-                break;
-              case 10: {
-                java.lang.String s = input.readStringRequireUtf8();
-
-                name_ = s;
-                break;
-              }
-              case 18: {
-                java.lang.String s = input.readStringRequireUtf8();
-
-                fullName_ = s;
-                break;
-              }
-              case 26: {
-                java.lang.String s = input.readStringRequireUtf8();
-
-                checkpointKey_ = s;
-                break;
-              }
-              case 32: {
-
-                optionalRestore_ = input.readBool();
-                break;
-              }
-              default: {
-                if (!parseUnknownField(
-                    input, unknownFields, extensionRegistry, tag)) {
-                  done = true;
-                }
-                break;
-              }
-            }
-          }
-        } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-          throw e.setUnfinishedMessage(this);
-        } catch (java.io.IOException e) {
-          throw new com.google.protobuf.InvalidProtocolBufferException(
-              e).setUnfinishedMessage(this);
-        } finally {
-          this.unknownFields = unknownFields.build();
-          makeExtensionsImmutable();
-        }
-      }
-      public static final com.google.protobuf.Descriptors.Descriptor
-          getDescriptor() {
-        return org.tensorflow.proto.framework.TrackableObjectGraphProtos.internal_static_tensorflow_TrackableObjectGraph_TrackableObject_SerializedTensor_descriptor;
-      }
-
-      @java.lang.Override
-      protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-          internalGetFieldAccessorTable() {
-        return org.tensorflow.proto.framework.TrackableObjectGraphProtos.internal_static_tensorflow_TrackableObjectGraph_TrackableObject_SerializedTensor_fieldAccessorTable
-            .ensureFieldAccessorsInitialized(
-                org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensor.class, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensor.Builder.class);
-      }
-
-      public static final int NAME_FIELD_NUMBER = 1;
-      private volatile java.lang.Object name_;
-      /**
-       * <pre>
-       * A name for the Tensor. Simple variables have only one
-       * `SerializedTensor` named "VARIABLE_VALUE" by convention. This value may
-       * be restored on object creation as an optimization.
-       * </pre>
-       *
-       * <code>string name = 1;</code>
-       */
-      public java.lang.String getName() {
-        java.lang.Object ref = name_;
-        if (ref instanceof java.lang.String) {
-          return (java.lang.String) ref;
-        } else {
-          com.google.protobuf.ByteString bs = 
-              (com.google.protobuf.ByteString) ref;
-          java.lang.String s = bs.toStringUtf8();
-          name_ = s;
-          return s;
-        }
-      }
-      /**
-       * <pre>
-       * A name for the Tensor. Simple variables have only one
-       * `SerializedTensor` named "VARIABLE_VALUE" by convention. This value may
-       * be restored on object creation as an optimization.
-       * </pre>
-       *
-       * <code>string name = 1;</code>
-       */
-      public com.google.protobuf.ByteString
-          getNameBytes() {
-        java.lang.Object ref = name_;
-        if (ref instanceof java.lang.String) {
-          com.google.protobuf.ByteString b = 
-              com.google.protobuf.ByteString.copyFromUtf8(
-                  (java.lang.String) ref);
-          name_ = b;
-          return b;
-        } else {
-          return (com.google.protobuf.ByteString) ref;
-        }
-      }
-
-      public static final int FULL_NAME_FIELD_NUMBER = 2;
-      private volatile java.lang.Object fullName_;
-      /**
-       * <pre>
-       * The full name of the variable/tensor, if applicable. Used to allow
-       * name-based loading of checkpoints which were saved using an
-       * object-based API. Should match the checkpoint key which would have been
-       * assigned by tf.train.Saver.
-       * </pre>
-       *
-       * <code>string full_name = 2;</code>
-       */
-      public java.lang.String getFullName() {
-        java.lang.Object ref = fullName_;
-        if (ref instanceof java.lang.String) {
-          return (java.lang.String) ref;
-        } else {
-          com.google.protobuf.ByteString bs = 
-              (com.google.protobuf.ByteString) ref;
-          java.lang.String s = bs.toStringUtf8();
-          fullName_ = s;
-          return s;
-        }
-      }
-      /**
-       * <pre>
-       * The full name of the variable/tensor, if applicable. Used to allow
-       * name-based loading of checkpoints which were saved using an
-       * object-based API. Should match the checkpoint key which would have been
-       * assigned by tf.train.Saver.
-       * </pre>
-       *
-       * <code>string full_name = 2;</code>
-       */
-      public com.google.protobuf.ByteString
-          getFullNameBytes() {
-        java.lang.Object ref = fullName_;
-        if (ref instanceof java.lang.String) {
-          com.google.protobuf.ByteString b = 
-              com.google.protobuf.ByteString.copyFromUtf8(
-                  (java.lang.String) ref);
-          fullName_ = b;
-          return b;
-        } else {
-          return (com.google.protobuf.ByteString) ref;
-        }
-      }
-
-      public static final int CHECKPOINT_KEY_FIELD_NUMBER = 3;
-      private volatile java.lang.Object checkpointKey_;
-      /**
-       * <pre>
-       * The generated name of the Tensor in the checkpoint.
-       * </pre>
-       *
-       * <code>string checkpoint_key = 3;</code>
-       */
-      public java.lang.String getCheckpointKey() {
-        java.lang.Object ref = checkpointKey_;
-        if (ref instanceof java.lang.String) {
-          return (java.lang.String) ref;
-        } else {
-          com.google.protobuf.ByteString bs = 
-              (com.google.protobuf.ByteString) ref;
-          java.lang.String s = bs.toStringUtf8();
-          checkpointKey_ = s;
-          return s;
-        }
-      }
-      /**
-       * <pre>
-       * The generated name of the Tensor in the checkpoint.
-       * </pre>
-       *
-       * <code>string checkpoint_key = 3;</code>
-       */
-      public com.google.protobuf.ByteString
-          getCheckpointKeyBytes() {
-        java.lang.Object ref = checkpointKey_;
-        if (ref instanceof java.lang.String) {
-          com.google.protobuf.ByteString b = 
-              com.google.protobuf.ByteString.copyFromUtf8(
-                  (java.lang.String) ref);
-          checkpointKey_ = b;
-          return b;
-        } else {
-          return (com.google.protobuf.ByteString) ref;
-        }
-      }
-
-      public static final int OPTIONAL_RESTORE_FIELD_NUMBER = 4;
-      private boolean optionalRestore_;
-      /**
-       * <pre>
-       * Whether checkpoints should be considered as matching even without this
-       * value restored. Used for non-critical values which don't affect the
-       * TensorFlow graph, such as layer configurations.
-       * </pre>
-       *
-       * <code>bool optional_restore = 4;</code>
-       */
-      public boolean getOptionalRestore() {
-        return optionalRestore_;
-      }
-
-      private byte memoizedIsInitialized = -1;
-      @java.lang.Override
-      public final boolean isInitialized() {
-        byte isInitialized = memoizedIsInitialized;
-        if (isInitialized == 1) return true;
-        if (isInitialized == 0) return false;
-
-        memoizedIsInitialized = 1;
-        return true;
-      }
-
-      @java.lang.Override
-      public void writeTo(com.google.protobuf.CodedOutputStream output)
-                          throws java.io.IOException {
-        if (!getNameBytes().isEmpty()) {
-          com.google.protobuf.GeneratedMessageV3.writeString(output, 1, name_);
-        }
-        if (!getFullNameBytes().isEmpty()) {
-          com.google.protobuf.GeneratedMessageV3.writeString(output, 2, fullName_);
-        }
-        if (!getCheckpointKeyBytes().isEmpty()) {
-          com.google.protobuf.GeneratedMessageV3.writeString(output, 3, checkpointKey_);
-        }
-        if (optionalRestore_ != false) {
-          output.writeBool(4, optionalRestore_);
-        }
-        unknownFields.writeTo(output);
-      }
-
-      @java.lang.Override
-      public int getSerializedSize() {
-        int size = memoizedSize;
-        if (size != -1) return size;
-
-        size = 0;
-        if (!getNameBytes().isEmpty()) {
-          size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, name_);
-        }
-        if (!getFullNameBytes().isEmpty()) {
-          size += com.google.protobuf.GeneratedMessageV3.computeStringSize(2, fullName_);
-        }
-        if (!getCheckpointKeyBytes().isEmpty()) {
-          size += com.google.protobuf.GeneratedMessageV3.computeStringSize(3, checkpointKey_);
-        }
-        if (optionalRestore_ != false) {
-          size += com.google.protobuf.CodedOutputStream
-            .computeBoolSize(4, optionalRestore_);
-        }
-        size += unknownFields.getSerializedSize();
-        memoizedSize = size;
-        return size;
-      }
-
-      @java.lang.Override
-      public boolean equals(final java.lang.Object obj) {
-        if (obj == this) {
-         return true;
-        }
-        if (!(obj instanceof org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensor)) {
-          return super.equals(obj);
-        }
-        org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensor other = (org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensor) obj;
-
-        if (!getName()
-            .equals(other.getName())) return false;
-        if (!getFullName()
-            .equals(other.getFullName())) return false;
-        if (!getCheckpointKey()
-            .equals(other.getCheckpointKey())) return false;
-        if (getOptionalRestore()
-            != other.getOptionalRestore()) return false;
-        if (!unknownFields.equals(other.unknownFields)) return false;
-        return true;
-      }
-
-      @java.lang.Override
-      public int hashCode() {
-        if (memoizedHashCode != 0) {
-          return memoizedHashCode;
-        }
-        int hash = 41;
-        hash = (19 * hash) + getDescriptor().hashCode();
-        hash = (37 * hash) + NAME_FIELD_NUMBER;
-        hash = (53 * hash) + getName().hashCode();
-        hash = (37 * hash) + FULL_NAME_FIELD_NUMBER;
-        hash = (53 * hash) + getFullName().hashCode();
-        hash = (37 * hash) + CHECKPOINT_KEY_FIELD_NUMBER;
-        hash = (53 * hash) + getCheckpointKey().hashCode();
-        hash = (37 * hash) + OPTIONAL_RESTORE_FIELD_NUMBER;
-        hash = (53 * hash) + com.google.protobuf.Internal.hashBoolean(
-            getOptionalRestore());
-        hash = (29 * hash) + unknownFields.hashCode();
-        memoizedHashCode = hash;
-        return hash;
-      }
-
-      public static org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensor parseFrom(
-          java.nio.ByteBuffer data)
-          throws com.google.protobuf.InvalidProtocolBufferException {
-        return PARSER.parseFrom(data);
-      }
-      public static org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensor parseFrom(
-          java.nio.ByteBuffer data,
-          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-          throws com.google.protobuf.InvalidProtocolBufferException {
-        return PARSER.parseFrom(data, extensionRegistry);
-      }
-      public static org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensor parseFrom(
-          com.google.protobuf.ByteString data)
-          throws com.google.protobuf.InvalidProtocolBufferException {
-        return PARSER.parseFrom(data);
-      }
-      public static org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensor parseFrom(
-          com.google.protobuf.ByteString data,
-          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-          throws com.google.protobuf.InvalidProtocolBufferException {
-        return PARSER.parseFrom(data, extensionRegistry);
-      }
-      public static org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensor parseFrom(byte[] data)
-          throws com.google.protobuf.InvalidProtocolBufferException {
-        return PARSER.parseFrom(data);
-      }
-      public static org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensor parseFrom(
-          byte[] data,
-          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-          throws com.google.protobuf.InvalidProtocolBufferException {
-        return PARSER.parseFrom(data, extensionRegistry);
-      }
-      public static org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensor parseFrom(java.io.InputStream input)
-          throws java.io.IOException {
-        return com.google.protobuf.GeneratedMessageV3
-            .parseWithIOException(PARSER, input);
-      }
-      public static org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensor parseFrom(
-          java.io.InputStream input,
-          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-          throws java.io.IOException {
-        return com.google.protobuf.GeneratedMessageV3
-            .parseWithIOException(PARSER, input, extensionRegistry);
-      }
-      public static org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensor parseDelimitedFrom(java.io.InputStream input)
-          throws java.io.IOException {
-        return com.google.protobuf.GeneratedMessageV3
-            .parseDelimitedWithIOException(PARSER, input);
-      }
-      public static org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensor parseDelimitedFrom(
-          java.io.InputStream input,
-          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-          throws java.io.IOException {
-        return com.google.protobuf.GeneratedMessageV3
-            .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-      }
-      public static org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensor parseFrom(
-          com.google.protobuf.CodedInputStream input)
-          throws java.io.IOException {
-        return com.google.protobuf.GeneratedMessageV3
-            .parseWithIOException(PARSER, input);
-      }
-      public static org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensor parseFrom(
-          com.google.protobuf.CodedInputStream input,
-          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-          throws java.io.IOException {
-        return com.google.protobuf.GeneratedMessageV3
-            .parseWithIOException(PARSER, input, extensionRegistry);
-      }
-
-      @java.lang.Override
-      public Builder newBuilderForType() { return newBuilder(); }
-      public static Builder newBuilder() {
-        return DEFAULT_INSTANCE.toBuilder();
-      }
-      public static Builder newBuilder(org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensor prototype) {
-        return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-      }
-      @java.lang.Override
-      public Builder toBuilder() {
-        return this == DEFAULT_INSTANCE
-            ? new Builder() : new Builder().mergeFrom(this);
-      }
-
-      @java.lang.Override
-      protected Builder newBuilderForType(
-          com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-        Builder builder = new Builder(parent);
-        return builder;
-      }
-      /**
-       * Protobuf type {@code tensorflow.TrackableObjectGraph.TrackableObject.SerializedTensor}
-       */
-      public static final class Builder extends
-          com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-          // @@protoc_insertion_point(builder_implements:tensorflow.TrackableObjectGraph.TrackableObject.SerializedTensor)
-          org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensorOrBuilder {
-        public static final com.google.protobuf.Descriptors.Descriptor
-            getDescriptor() {
-          return org.tensorflow.proto.framework.TrackableObjectGraphProtos.internal_static_tensorflow_TrackableObjectGraph_TrackableObject_SerializedTensor_descriptor;
-        }
-
-        @java.lang.Override
-        protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-            internalGetFieldAccessorTable() {
-          return org.tensorflow.proto.framework.TrackableObjectGraphProtos.internal_static_tensorflow_TrackableObjectGraph_TrackableObject_SerializedTensor_fieldAccessorTable
-              .ensureFieldAccessorsInitialized(
-                  org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensor.class, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensor.Builder.class);
-        }
-
-        // Construct using org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensor.newBuilder()
-        private Builder() {
-          maybeForceBuilderInitialization();
-        }
-
-        private Builder(
-            com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-          super(parent);
-          maybeForceBuilderInitialization();
-        }
-        private void maybeForceBuilderInitialization() {
-          if (com.google.protobuf.GeneratedMessageV3
-                  .alwaysUseFieldBuilders) {
-          }
-        }
-        @java.lang.Override
-        public Builder clear() {
-          super.clear();
-          name_ = "";
-
-          fullName_ = "";
-
-          checkpointKey_ = "";
-
-          optionalRestore_ = false;
-
-          return this;
-        }
-
-        @java.lang.Override
-        public com.google.protobuf.Descriptors.Descriptor
-            getDescriptorForType() {
-          return org.tensorflow.proto.framework.TrackableObjectGraphProtos.internal_static_tensorflow_TrackableObjectGraph_TrackableObject_SerializedTensor_descriptor;
-        }
-
-        @java.lang.Override
-        public org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensor getDefaultInstanceForType() {
-          return org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensor.getDefaultInstance();
-        }
-
-        @java.lang.Override
-        public org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensor build() {
-          org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensor result = buildPartial();
-          if (!result.isInitialized()) {
-            throw newUninitializedMessageException(result);
-          }
-          return result;
-        }
-
-        @java.lang.Override
-        public org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensor buildPartial() {
-          org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensor result = new org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensor(this);
-          result.name_ = name_;
-          result.fullName_ = fullName_;
-          result.checkpointKey_ = checkpointKey_;
-          result.optionalRestore_ = optionalRestore_;
-          onBuilt();
-          return result;
-        }
-
-        @java.lang.Override
-        public Builder clone() {
-          return super.clone();
-        }
-        @java.lang.Override
-        public Builder setField(
-            com.google.protobuf.Descriptors.FieldDescriptor field,
-            java.lang.Object value) {
-          return super.setField(field, value);
-        }
-        @java.lang.Override
-        public Builder clearField(
-            com.google.protobuf.Descriptors.FieldDescriptor field) {
-          return super.clearField(field);
-        }
-        @java.lang.Override
-        public Builder clearOneof(
-            com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-          return super.clearOneof(oneof);
-        }
-        @java.lang.Override
-        public Builder setRepeatedField(
-            com.google.protobuf.Descriptors.FieldDescriptor field,
-            int index, java.lang.Object value) {
-          return super.setRepeatedField(field, index, value);
-        }
-        @java.lang.Override
-        public Builder addRepeatedField(
-            com.google.protobuf.Descriptors.FieldDescriptor field,
-            java.lang.Object value) {
-          return super.addRepeatedField(field, value);
-        }
-        @java.lang.Override
-        public Builder mergeFrom(com.google.protobuf.Message other) {
-          if (other instanceof org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensor) {
-            return mergeFrom((org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensor)other);
-          } else {
-            super.mergeFrom(other);
-            return this;
-          }
-        }
-
-        public Builder mergeFrom(org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensor other) {
-          if (other == org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensor.getDefaultInstance()) return this;
-          if (!other.getName().isEmpty()) {
-            name_ = other.name_;
-            onChanged();
-          }
-          if (!other.getFullName().isEmpty()) {
-            fullName_ = other.fullName_;
-            onChanged();
-          }
-          if (!other.getCheckpointKey().isEmpty()) {
-            checkpointKey_ = other.checkpointKey_;
-            onChanged();
-          }
-          if (other.getOptionalRestore() != false) {
-            setOptionalRestore(other.getOptionalRestore());
-          }
-          this.mergeUnknownFields(other.unknownFields);
-          onChanged();
-          return this;
-        }
-
-        @java.lang.Override
-        public final boolean isInitialized() {
-          return true;
-        }
-
-        @java.lang.Override
-        public Builder mergeFrom(
-            com.google.protobuf.CodedInputStream input,
-            com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-            throws java.io.IOException {
-          org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensor parsedMessage = null;
-          try {
-            parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-          } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-            parsedMessage = (org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensor) e.getUnfinishedMessage();
-            throw e.unwrapIOException();
-          } finally {
-            if (parsedMessage != null) {
-              mergeFrom(parsedMessage);
-            }
-          }
-          return this;
-        }
-
-        private java.lang.Object name_ = "";
-        /**
-         * <pre>
-         * A name for the Tensor. Simple variables have only one
-         * `SerializedTensor` named "VARIABLE_VALUE" by convention. This value may
-         * be restored on object creation as an optimization.
-         * </pre>
-         *
-         * <code>string name = 1;</code>
-         */
-        public java.lang.String getName() {
-          java.lang.Object ref = name_;
-          if (!(ref instanceof java.lang.String)) {
-            com.google.protobuf.ByteString bs =
-                (com.google.protobuf.ByteString) ref;
-            java.lang.String s = bs.toStringUtf8();
-            name_ = s;
-            return s;
-          } else {
-            return (java.lang.String) ref;
-          }
-        }
-        /**
-         * <pre>
-         * A name for the Tensor. Simple variables have only one
-         * `SerializedTensor` named "VARIABLE_VALUE" by convention. This value may
-         * be restored on object creation as an optimization.
-         * </pre>
-         *
-         * <code>string name = 1;</code>
-         */
-        public com.google.protobuf.ByteString
-            getNameBytes() {
-          java.lang.Object ref = name_;
-          if (ref instanceof String) {
-            com.google.protobuf.ByteString b = 
-                com.google.protobuf.ByteString.copyFromUtf8(
-                    (java.lang.String) ref);
-            name_ = b;
-            return b;
-          } else {
-            return (com.google.protobuf.ByteString) ref;
-          }
-        }
-        /**
-         * <pre>
-         * A name for the Tensor. Simple variables have only one
-         * `SerializedTensor` named "VARIABLE_VALUE" by convention. This value may
-         * be restored on object creation as an optimization.
-         * </pre>
-         *
-         * <code>string name = 1;</code>
-         */
-        public Builder setName(
-            java.lang.String value) {
-          if (value == null) {
-    throw new NullPointerException();
-  }
-  
-          name_ = value;
-          onChanged();
-          return this;
-        }
-        /**
-         * <pre>
-         * A name for the Tensor. Simple variables have only one
-         * `SerializedTensor` named "VARIABLE_VALUE" by convention. This value may
-         * be restored on object creation as an optimization.
-         * </pre>
-         *
-         * <code>string name = 1;</code>
-         */
-        public Builder clearName() {
-          
-          name_ = getDefaultInstance().getName();
-          onChanged();
-          return this;
-        }
-        /**
-         * <pre>
-         * A name for the Tensor. Simple variables have only one
-         * `SerializedTensor` named "VARIABLE_VALUE" by convention. This value may
-         * be restored on object creation as an optimization.
-         * </pre>
-         *
-         * <code>string name = 1;</code>
-         */
-        public Builder setNameBytes(
-            com.google.protobuf.ByteString value) {
-          if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-          
-          name_ = value;
-          onChanged();
-          return this;
-        }
-
-        private java.lang.Object fullName_ = "";
-        /**
-         * <pre>
-         * The full name of the variable/tensor, if applicable. Used to allow
-         * name-based loading of checkpoints which were saved using an
-         * object-based API. Should match the checkpoint key which would have been
-         * assigned by tf.train.Saver.
-         * </pre>
-         *
-         * <code>string full_name = 2;</code>
-         */
-        public java.lang.String getFullName() {
-          java.lang.Object ref = fullName_;
-          if (!(ref instanceof java.lang.String)) {
-            com.google.protobuf.ByteString bs =
-                (com.google.protobuf.ByteString) ref;
-            java.lang.String s = bs.toStringUtf8();
-            fullName_ = s;
-            return s;
-          } else {
-            return (java.lang.String) ref;
-          }
-        }
-        /**
-         * <pre>
-         * The full name of the variable/tensor, if applicable. Used to allow
-         * name-based loading of checkpoints which were saved using an
-         * object-based API. Should match the checkpoint key which would have been
-         * assigned by tf.train.Saver.
-         * </pre>
-         *
-         * <code>string full_name = 2;</code>
-         */
-        public com.google.protobuf.ByteString
-            getFullNameBytes() {
-          java.lang.Object ref = fullName_;
-          if (ref instanceof String) {
-            com.google.protobuf.ByteString b = 
-                com.google.protobuf.ByteString.copyFromUtf8(
-                    (java.lang.String) ref);
-            fullName_ = b;
-            return b;
-          } else {
-            return (com.google.protobuf.ByteString) ref;
-          }
-        }
-        /**
-         * <pre>
-         * The full name of the variable/tensor, if applicable. Used to allow
-         * name-based loading of checkpoints which were saved using an
-         * object-based API. Should match the checkpoint key which would have been
-         * assigned by tf.train.Saver.
-         * </pre>
-         *
-         * <code>string full_name = 2;</code>
-         */
-        public Builder setFullName(
-            java.lang.String value) {
-          if (value == null) {
-    throw new NullPointerException();
-  }
-  
-          fullName_ = value;
-          onChanged();
-          return this;
-        }
-        /**
-         * <pre>
-         * The full name of the variable/tensor, if applicable. Used to allow
-         * name-based loading of checkpoints which were saved using an
-         * object-based API. Should match the checkpoint key which would have been
-         * assigned by tf.train.Saver.
-         * </pre>
-         *
-         * <code>string full_name = 2;</code>
-         */
-        public Builder clearFullName() {
-          
-          fullName_ = getDefaultInstance().getFullName();
-          onChanged();
-          return this;
-        }
-        /**
-         * <pre>
-         * The full name of the variable/tensor, if applicable. Used to allow
-         * name-based loading of checkpoints which were saved using an
-         * object-based API. Should match the checkpoint key which would have been
-         * assigned by tf.train.Saver.
-         * </pre>
-         *
-         * <code>string full_name = 2;</code>
-         */
-        public Builder setFullNameBytes(
-            com.google.protobuf.ByteString value) {
-          if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-          
-          fullName_ = value;
-          onChanged();
-          return this;
-        }
-
-        private java.lang.Object checkpointKey_ = "";
-        /**
-         * <pre>
-         * The generated name of the Tensor in the checkpoint.
-         * </pre>
-         *
-         * <code>string checkpoint_key = 3;</code>
-         */
-        public java.lang.String getCheckpointKey() {
-          java.lang.Object ref = checkpointKey_;
-          if (!(ref instanceof java.lang.String)) {
-            com.google.protobuf.ByteString bs =
-                (com.google.protobuf.ByteString) ref;
-            java.lang.String s = bs.toStringUtf8();
-            checkpointKey_ = s;
-            return s;
-          } else {
-            return (java.lang.String) ref;
-          }
-        }
-        /**
-         * <pre>
-         * The generated name of the Tensor in the checkpoint.
-         * </pre>
-         *
-         * <code>string checkpoint_key = 3;</code>
-         */
-        public com.google.protobuf.ByteString
-            getCheckpointKeyBytes() {
-          java.lang.Object ref = checkpointKey_;
-          if (ref instanceof String) {
-            com.google.protobuf.ByteString b = 
-                com.google.protobuf.ByteString.copyFromUtf8(
-                    (java.lang.String) ref);
-            checkpointKey_ = b;
-            return b;
-          } else {
-            return (com.google.protobuf.ByteString) ref;
-          }
-        }
-        /**
-         * <pre>
-         * The generated name of the Tensor in the checkpoint.
-         * </pre>
-         *
-         * <code>string checkpoint_key = 3;</code>
-         */
-        public Builder setCheckpointKey(
-            java.lang.String value) {
-          if (value == null) {
-    throw new NullPointerException();
-  }
-  
-          checkpointKey_ = value;
-          onChanged();
-          return this;
-        }
-        /**
-         * <pre>
-         * The generated name of the Tensor in the checkpoint.
-         * </pre>
-         *
-         * <code>string checkpoint_key = 3;</code>
-         */
-        public Builder clearCheckpointKey() {
-          
-          checkpointKey_ = getDefaultInstance().getCheckpointKey();
-          onChanged();
-          return this;
-        }
-        /**
-         * <pre>
-         * The generated name of the Tensor in the checkpoint.
-         * </pre>
-         *
-         * <code>string checkpoint_key = 3;</code>
-         */
-        public Builder setCheckpointKeyBytes(
-            com.google.protobuf.ByteString value) {
-          if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-          
-          checkpointKey_ = value;
-          onChanged();
-          return this;
-        }
-
-        private boolean optionalRestore_ ;
-        /**
-         * <pre>
-         * Whether checkpoints should be considered as matching even without this
-         * value restored. Used for non-critical values which don't affect the
-         * TensorFlow graph, such as layer configurations.
-         * </pre>
-         *
-         * <code>bool optional_restore = 4;</code>
-         */
-        public boolean getOptionalRestore() {
-          return optionalRestore_;
-        }
-        /**
-         * <pre>
-         * Whether checkpoints should be considered as matching even without this
-         * value restored. Used for non-critical values which don't affect the
-         * TensorFlow graph, such as layer configurations.
-         * </pre>
-         *
-         * <code>bool optional_restore = 4;</code>
-         */
-        public Builder setOptionalRestore(boolean value) {
-          
-          optionalRestore_ = value;
-          onChanged();
-          return this;
-        }
-        /**
-         * <pre>
-         * Whether checkpoints should be considered as matching even without this
-         * value restored. Used for non-critical values which don't affect the
-         * TensorFlow graph, such as layer configurations.
-         * </pre>
-         *
-         * <code>bool optional_restore = 4;</code>
-         */
-        public Builder clearOptionalRestore() {
-          
-          optionalRestore_ = false;
-          onChanged();
-          return this;
-        }
-        @java.lang.Override
-        public final Builder setUnknownFields(
-            final com.google.protobuf.UnknownFieldSet unknownFields) {
-          return super.setUnknownFields(unknownFields);
-        }
-
-        @java.lang.Override
-        public final Builder mergeUnknownFields(
-            final com.google.protobuf.UnknownFieldSet unknownFields) {
-          return super.mergeUnknownFields(unknownFields);
-        }
-
-
-        // @@protoc_insertion_point(builder_scope:tensorflow.TrackableObjectGraph.TrackableObject.SerializedTensor)
-      }
-
-      // @@protoc_insertion_point(class_scope:tensorflow.TrackableObjectGraph.TrackableObject.SerializedTensor)
-      private static final org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensor DEFAULT_INSTANCE;
-      static {
-        DEFAULT_INSTANCE = new org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensor();
-      }
-
-      public static org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensor getDefaultInstance() {
-        return DEFAULT_INSTANCE;
-      }
-
-      private static final com.google.protobuf.Parser<SerializedTensor>
-          PARSER = new com.google.protobuf.AbstractParser<SerializedTensor>() {
-        @java.lang.Override
-        public SerializedTensor parsePartialFrom(
-            com.google.protobuf.CodedInputStream input,
-            com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-            throws com.google.protobuf.InvalidProtocolBufferException {
-          return new SerializedTensor(input, extensionRegistry);
-        }
-      };
-
-      public static com.google.protobuf.Parser<SerializedTensor> parser() {
-        return PARSER;
-      }
-
-      @java.lang.Override
-      public com.google.protobuf.Parser<SerializedTensor> getParserForType() {
-        return PARSER;
-      }
-
-      @java.lang.Override
-      public org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensor getDefaultInstanceForType() {
-        return DEFAULT_INSTANCE;
-      }
-
-    }
-
-    public interface SlotVariableReferenceOrBuilder extends
-        // @@protoc_insertion_point(interface_extends:tensorflow.TrackableObjectGraph.TrackableObject.SlotVariableReference)
-        com.google.protobuf.MessageOrBuilder {
-
-      /**
-       * <pre>
-       * An index into `TrackableObjectGraph.nodes`, indicating the
-       * variable object this slot was created for.
-       * </pre>
-       *
-       * <code>int32 original_variable_node_id = 1;</code>
-       */
-      int getOriginalVariableNodeId();
-
-      /**
-       * <pre>
-       * The name of the slot (e.g. "m"/"v").
-       * </pre>
-       *
-       * <code>string slot_name = 2;</code>
-       */
-      java.lang.String getSlotName();
-      /**
-       * <pre>
-       * The name of the slot (e.g. "m"/"v").
-       * </pre>
-       *
-       * <code>string slot_name = 2;</code>
-       */
-      com.google.protobuf.ByteString
-          getSlotNameBytes();
-
-      /**
-       * <pre>
-       * An index into `TrackableObjectGraph.nodes`, indicating the
-       * `Object` with the value of the slot variable.
-       * </pre>
-       *
-       * <code>int32 slot_variable_node_id = 3;</code>
-       */
-      int getSlotVariableNodeId();
-    }
-    /**
-     * Protobuf type {@code tensorflow.TrackableObjectGraph.TrackableObject.SlotVariableReference}
-     */
-    public  static final class SlotVariableReference extends
-        com.google.protobuf.GeneratedMessageV3 implements
-        // @@protoc_insertion_point(message_implements:tensorflow.TrackableObjectGraph.TrackableObject.SlotVariableReference)
-        SlotVariableReferenceOrBuilder {
-    private static final long serialVersionUID = 0L;
-      // Use SlotVariableReference.newBuilder() to construct.
-      private SlotVariableReference(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-        super(builder);
-      }
-      private SlotVariableReference() {
-        slotName_ = "";
-      }
-
-      @java.lang.Override
-      @SuppressWarnings({"unused"})
-      protected java.lang.Object newInstance(
-          UnusedPrivateParameter unused) {
-        return new SlotVariableReference();
-      }
-
-      @java.lang.Override
-      public final com.google.protobuf.UnknownFieldSet
-      getUnknownFields() {
-        return this.unknownFields;
-      }
-      private SlotVariableReference(
-          com.google.protobuf.CodedInputStream input,
-          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-          throws com.google.protobuf.InvalidProtocolBufferException {
-        this();
-        if (extensionRegistry == null) {
-          throw new java.lang.NullPointerException();
-        }
-        com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-            com.google.protobuf.UnknownFieldSet.newBuilder();
-        try {
-          boolean done = false;
-          while (!done) {
-            int tag = input.readTag();
-            switch (tag) {
-              case 0:
-                done = true;
-                break;
-              case 8: {
-
-                originalVariableNodeId_ = input.readInt32();
-                break;
-              }
-              case 18: {
-                java.lang.String s = input.readStringRequireUtf8();
-
-                slotName_ = s;
-                break;
-              }
-              case 24: {
-
-                slotVariableNodeId_ = input.readInt32();
-                break;
-              }
-              default: {
-                if (!parseUnknownField(
-                    input, unknownFields, extensionRegistry, tag)) {
-                  done = true;
-                }
-                break;
-              }
-            }
-          }
-        } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-          throw e.setUnfinishedMessage(this);
-        } catch (java.io.IOException e) {
-          throw new com.google.protobuf.InvalidProtocolBufferException(
-              e).setUnfinishedMessage(this);
-        } finally {
-          this.unknownFields = unknownFields.build();
-          makeExtensionsImmutable();
-        }
-      }
-      public static final com.google.protobuf.Descriptors.Descriptor
-          getDescriptor() {
-        return org.tensorflow.proto.framework.TrackableObjectGraphProtos.internal_static_tensorflow_TrackableObjectGraph_TrackableObject_SlotVariableReference_descriptor;
-      }
-
-      @java.lang.Override
-      protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-          internalGetFieldAccessorTable() {
-        return org.tensorflow.proto.framework.TrackableObjectGraphProtos.internal_static_tensorflow_TrackableObjectGraph_TrackableObject_SlotVariableReference_fieldAccessorTable
-            .ensureFieldAccessorsInitialized(
-                org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference.class, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference.Builder.class);
-      }
-
-      public static final int ORIGINAL_VARIABLE_NODE_ID_FIELD_NUMBER = 1;
-      private int originalVariableNodeId_;
-      /**
-       * <pre>
-       * An index into `TrackableObjectGraph.nodes`, indicating the
-       * variable object this slot was created for.
-       * </pre>
-       *
-       * <code>int32 original_variable_node_id = 1;</code>
-       */
-      public int getOriginalVariableNodeId() {
-        return originalVariableNodeId_;
-      }
-
-      public static final int SLOT_NAME_FIELD_NUMBER = 2;
-      private volatile java.lang.Object slotName_;
-      /**
-       * <pre>
-       * The name of the slot (e.g. "m"/"v").
-       * </pre>
-       *
-       * <code>string slot_name = 2;</code>
-       */
-      public java.lang.String getSlotName() {
-        java.lang.Object ref = slotName_;
-        if (ref instanceof java.lang.String) {
-          return (java.lang.String) ref;
-        } else {
-          com.google.protobuf.ByteString bs = 
-              (com.google.protobuf.ByteString) ref;
-          java.lang.String s = bs.toStringUtf8();
-          slotName_ = s;
-          return s;
-        }
-      }
-      /**
-       * <pre>
-       * The name of the slot (e.g. "m"/"v").
-       * </pre>
-       *
-       * <code>string slot_name = 2;</code>
-       */
-      public com.google.protobuf.ByteString
-          getSlotNameBytes() {
-        java.lang.Object ref = slotName_;
-        if (ref instanceof java.lang.String) {
-          com.google.protobuf.ByteString b = 
-              com.google.protobuf.ByteString.copyFromUtf8(
-                  (java.lang.String) ref);
-          slotName_ = b;
-          return b;
-        } else {
-          return (com.google.protobuf.ByteString) ref;
-        }
-      }
-
-      public static final int SLOT_VARIABLE_NODE_ID_FIELD_NUMBER = 3;
-      private int slotVariableNodeId_;
-      /**
-       * <pre>
-       * An index into `TrackableObjectGraph.nodes`, indicating the
-       * `Object` with the value of the slot variable.
-       * </pre>
-       *
-       * <code>int32 slot_variable_node_id = 3;</code>
-       */
-      public int getSlotVariableNodeId() {
-        return slotVariableNodeId_;
-      }
-
-      private byte memoizedIsInitialized = -1;
-      @java.lang.Override
-      public final boolean isInitialized() {
-        byte isInitialized = memoizedIsInitialized;
-        if (isInitialized == 1) return true;
-        if (isInitialized == 0) return false;
-
-        memoizedIsInitialized = 1;
-        return true;
-      }
-
-      @java.lang.Override
-      public void writeTo(com.google.protobuf.CodedOutputStream output)
-                          throws java.io.IOException {
-        if (originalVariableNodeId_ != 0) {
-          output.writeInt32(1, originalVariableNodeId_);
-        }
-        if (!getSlotNameBytes().isEmpty()) {
-          com.google.protobuf.GeneratedMessageV3.writeString(output, 2, slotName_);
-        }
-        if (slotVariableNodeId_ != 0) {
-          output.writeInt32(3, slotVariableNodeId_);
-        }
-        unknownFields.writeTo(output);
-      }
-
-      @java.lang.Override
-      public int getSerializedSize() {
-        int size = memoizedSize;
-        if (size != -1) return size;
-
-        size = 0;
-        if (originalVariableNodeId_ != 0) {
-          size += com.google.protobuf.CodedOutputStream
-            .computeInt32Size(1, originalVariableNodeId_);
-        }
-        if (!getSlotNameBytes().isEmpty()) {
-          size += com.google.protobuf.GeneratedMessageV3.computeStringSize(2, slotName_);
-        }
-        if (slotVariableNodeId_ != 0) {
-          size += com.google.protobuf.CodedOutputStream
-            .computeInt32Size(3, slotVariableNodeId_);
-        }
-        size += unknownFields.getSerializedSize();
-        memoizedSize = size;
-        return size;
-      }
-
-      @java.lang.Override
-      public boolean equals(final java.lang.Object obj) {
-        if (obj == this) {
-         return true;
-        }
-        if (!(obj instanceof org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference)) {
-          return super.equals(obj);
-        }
-        org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference other = (org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference) obj;
-
-        if (getOriginalVariableNodeId()
-            != other.getOriginalVariableNodeId()) return false;
-        if (!getSlotName()
-            .equals(other.getSlotName())) return false;
-        if (getSlotVariableNodeId()
-            != other.getSlotVariableNodeId()) return false;
-        if (!unknownFields.equals(other.unknownFields)) return false;
-        return true;
-      }
-
-      @java.lang.Override
-      public int hashCode() {
-        if (memoizedHashCode != 0) {
-          return memoizedHashCode;
-        }
-        int hash = 41;
-        hash = (19 * hash) + getDescriptor().hashCode();
-        hash = (37 * hash) + ORIGINAL_VARIABLE_NODE_ID_FIELD_NUMBER;
-        hash = (53 * hash) + getOriginalVariableNodeId();
-        hash = (37 * hash) + SLOT_NAME_FIELD_NUMBER;
-        hash = (53 * hash) + getSlotName().hashCode();
-        hash = (37 * hash) + SLOT_VARIABLE_NODE_ID_FIELD_NUMBER;
-        hash = (53 * hash) + getSlotVariableNodeId();
-        hash = (29 * hash) + unknownFields.hashCode();
-        memoizedHashCode = hash;
-        return hash;
-      }
-
-      public static org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference parseFrom(
-          java.nio.ByteBuffer data)
-          throws com.google.protobuf.InvalidProtocolBufferException {
-        return PARSER.parseFrom(data);
-      }
-      public static org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference parseFrom(
-          java.nio.ByteBuffer data,
-          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-          throws com.google.protobuf.InvalidProtocolBufferException {
-        return PARSER.parseFrom(data, extensionRegistry);
-      }
-      public static org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference parseFrom(
-          com.google.protobuf.ByteString data)
-          throws com.google.protobuf.InvalidProtocolBufferException {
-        return PARSER.parseFrom(data);
-      }
-      public static org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference parseFrom(
-          com.google.protobuf.ByteString data,
-          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-          throws com.google.protobuf.InvalidProtocolBufferException {
-        return PARSER.parseFrom(data, extensionRegistry);
-      }
-      public static org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference parseFrom(byte[] data)
-          throws com.google.protobuf.InvalidProtocolBufferException {
-        return PARSER.parseFrom(data);
-      }
-      public static org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference parseFrom(
-          byte[] data,
-          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-          throws com.google.protobuf.InvalidProtocolBufferException {
-        return PARSER.parseFrom(data, extensionRegistry);
-      }
-      public static org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference parseFrom(java.io.InputStream input)
-          throws java.io.IOException {
-        return com.google.protobuf.GeneratedMessageV3
-            .parseWithIOException(PARSER, input);
-      }
-      public static org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference parseFrom(
-          java.io.InputStream input,
-          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-          throws java.io.IOException {
-        return com.google.protobuf.GeneratedMessageV3
-            .parseWithIOException(PARSER, input, extensionRegistry);
-      }
-      public static org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference parseDelimitedFrom(java.io.InputStream input)
-          throws java.io.IOException {
-        return com.google.protobuf.GeneratedMessageV3
-            .parseDelimitedWithIOException(PARSER, input);
-      }
-      public static org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference parseDelimitedFrom(
-          java.io.InputStream input,
-          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-          throws java.io.IOException {
-        return com.google.protobuf.GeneratedMessageV3
-            .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-      }
-      public static org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference parseFrom(
-          com.google.protobuf.CodedInputStream input)
-          throws java.io.IOException {
-        return com.google.protobuf.GeneratedMessageV3
-            .parseWithIOException(PARSER, input);
-      }
-      public static org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference parseFrom(
-          com.google.protobuf.CodedInputStream input,
-          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-          throws java.io.IOException {
-        return com.google.protobuf.GeneratedMessageV3
-            .parseWithIOException(PARSER, input, extensionRegistry);
-      }
-
-      @java.lang.Override
-      public Builder newBuilderForType() { return newBuilder(); }
-      public static Builder newBuilder() {
-        return DEFAULT_INSTANCE.toBuilder();
-      }
-      public static Builder newBuilder(org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference prototype) {
-        return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-      }
-      @java.lang.Override
-      public Builder toBuilder() {
-        return this == DEFAULT_INSTANCE
-            ? new Builder() : new Builder().mergeFrom(this);
-      }
-
-      @java.lang.Override
-      protected Builder newBuilderForType(
-          com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-        Builder builder = new Builder(parent);
-        return builder;
-      }
-      /**
-       * Protobuf type {@code tensorflow.TrackableObjectGraph.TrackableObject.SlotVariableReference}
-       */
-      public static final class Builder extends
-          com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-          // @@protoc_insertion_point(builder_implements:tensorflow.TrackableObjectGraph.TrackableObject.SlotVariableReference)
-          org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReferenceOrBuilder {
-        public static final com.google.protobuf.Descriptors.Descriptor
-            getDescriptor() {
-          return org.tensorflow.proto.framework.TrackableObjectGraphProtos.internal_static_tensorflow_TrackableObjectGraph_TrackableObject_SlotVariableReference_descriptor;
-        }
-
-        @java.lang.Override
-        protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-            internalGetFieldAccessorTable() {
-          return org.tensorflow.proto.framework.TrackableObjectGraphProtos.internal_static_tensorflow_TrackableObjectGraph_TrackableObject_SlotVariableReference_fieldAccessorTable
-              .ensureFieldAccessorsInitialized(
-                  org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference.class, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference.Builder.class);
-        }
-
-        // Construct using org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference.newBuilder()
-        private Builder() {
-          maybeForceBuilderInitialization();
-        }
-
-        private Builder(
-            com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-          super(parent);
-          maybeForceBuilderInitialization();
-        }
-        private void maybeForceBuilderInitialization() {
-          if (com.google.protobuf.GeneratedMessageV3
-                  .alwaysUseFieldBuilders) {
-          }
-        }
-        @java.lang.Override
-        public Builder clear() {
-          super.clear();
-          originalVariableNodeId_ = 0;
-
-          slotName_ = "";
-
-          slotVariableNodeId_ = 0;
-
-          return this;
-        }
-
-        @java.lang.Override
-        public com.google.protobuf.Descriptors.Descriptor
-            getDescriptorForType() {
-          return org.tensorflow.proto.framework.TrackableObjectGraphProtos.internal_static_tensorflow_TrackableObjectGraph_TrackableObject_SlotVariableReference_descriptor;
-        }
-
-        @java.lang.Override
-        public org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference getDefaultInstanceForType() {
-          return org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference.getDefaultInstance();
-        }
-
-        @java.lang.Override
-        public org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference build() {
-          org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference result = buildPartial();
-          if (!result.isInitialized()) {
-            throw newUninitializedMessageException(result);
-          }
-          return result;
-        }
-
-        @java.lang.Override
-        public org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference buildPartial() {
-          org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference result = new org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference(this);
-          result.originalVariableNodeId_ = originalVariableNodeId_;
-          result.slotName_ = slotName_;
-          result.slotVariableNodeId_ = slotVariableNodeId_;
-          onBuilt();
-          return result;
-        }
-
-        @java.lang.Override
-        public Builder clone() {
-          return super.clone();
-        }
-        @java.lang.Override
-        public Builder setField(
-            com.google.protobuf.Descriptors.FieldDescriptor field,
-            java.lang.Object value) {
-          return super.setField(field, value);
-        }
-        @java.lang.Override
-        public Builder clearField(
-            com.google.protobuf.Descriptors.FieldDescriptor field) {
-          return super.clearField(field);
-        }
-        @java.lang.Override
-        public Builder clearOneof(
-            com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-          return super.clearOneof(oneof);
-        }
-        @java.lang.Override
-        public Builder setRepeatedField(
-            com.google.protobuf.Descriptors.FieldDescriptor field,
-            int index, java.lang.Object value) {
-          return super.setRepeatedField(field, index, value);
-        }
-        @java.lang.Override
-        public Builder addRepeatedField(
-            com.google.protobuf.Descriptors.FieldDescriptor field,
-            java.lang.Object value) {
-          return super.addRepeatedField(field, value);
-        }
-        @java.lang.Override
-        public Builder mergeFrom(com.google.protobuf.Message other) {
-          if (other instanceof org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference) {
-            return mergeFrom((org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference)other);
-          } else {
-            super.mergeFrom(other);
-            return this;
-          }
-        }
-
-        public Builder mergeFrom(org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference other) {
-          if (other == org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference.getDefaultInstance()) return this;
-          if (other.getOriginalVariableNodeId() != 0) {
-            setOriginalVariableNodeId(other.getOriginalVariableNodeId());
-          }
-          if (!other.getSlotName().isEmpty()) {
-            slotName_ = other.slotName_;
-            onChanged();
-          }
-          if (other.getSlotVariableNodeId() != 0) {
-            setSlotVariableNodeId(other.getSlotVariableNodeId());
-          }
-          this.mergeUnknownFields(other.unknownFields);
-          onChanged();
-          return this;
-        }
-
-        @java.lang.Override
-        public final boolean isInitialized() {
-          return true;
-        }
-
-        @java.lang.Override
-        public Builder mergeFrom(
-            com.google.protobuf.CodedInputStream input,
-            com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-            throws java.io.IOException {
-          org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference parsedMessage = null;
-          try {
-            parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-          } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-            parsedMessage = (org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference) e.getUnfinishedMessage();
-            throw e.unwrapIOException();
-          } finally {
-            if (parsedMessage != null) {
-              mergeFrom(parsedMessage);
-            }
-          }
-          return this;
-        }
-
-        private int originalVariableNodeId_ ;
-        /**
-         * <pre>
-         * An index into `TrackableObjectGraph.nodes`, indicating the
-         * variable object this slot was created for.
-         * </pre>
-         *
-         * <code>int32 original_variable_node_id = 1;</code>
-         */
-        public int getOriginalVariableNodeId() {
-          return originalVariableNodeId_;
-        }
-        /**
-         * <pre>
-         * An index into `TrackableObjectGraph.nodes`, indicating the
-         * variable object this slot was created for.
-         * </pre>
-         *
-         * <code>int32 original_variable_node_id = 1;</code>
-         */
-        public Builder setOriginalVariableNodeId(int value) {
-          
-          originalVariableNodeId_ = value;
-          onChanged();
-          return this;
-        }
-        /**
-         * <pre>
-         * An index into `TrackableObjectGraph.nodes`, indicating the
-         * variable object this slot was created for.
-         * </pre>
-         *
-         * <code>int32 original_variable_node_id = 1;</code>
-         */
-        public Builder clearOriginalVariableNodeId() {
-          
-          originalVariableNodeId_ = 0;
-          onChanged();
-          return this;
-        }
-
-        private java.lang.Object slotName_ = "";
-        /**
-         * <pre>
-         * The name of the slot (e.g. "m"/"v").
-         * </pre>
-         *
-         * <code>string slot_name = 2;</code>
-         */
-        public java.lang.String getSlotName() {
-          java.lang.Object ref = slotName_;
-          if (!(ref instanceof java.lang.String)) {
-            com.google.protobuf.ByteString bs =
-                (com.google.protobuf.ByteString) ref;
-            java.lang.String s = bs.toStringUtf8();
-            slotName_ = s;
-            return s;
-          } else {
-            return (java.lang.String) ref;
-          }
-        }
-        /**
-         * <pre>
-         * The name of the slot (e.g. "m"/"v").
-         * </pre>
-         *
-         * <code>string slot_name = 2;</code>
-         */
-        public com.google.protobuf.ByteString
-            getSlotNameBytes() {
-          java.lang.Object ref = slotName_;
-          if (ref instanceof String) {
-            com.google.protobuf.ByteString b = 
-                com.google.protobuf.ByteString.copyFromUtf8(
-                    (java.lang.String) ref);
-            slotName_ = b;
-            return b;
-          } else {
-            return (com.google.protobuf.ByteString) ref;
-          }
-        }
-        /**
-         * <pre>
-         * The name of the slot (e.g. "m"/"v").
-         * </pre>
-         *
-         * <code>string slot_name = 2;</code>
-         */
-        public Builder setSlotName(
-            java.lang.String value) {
-          if (value == null) {
-    throw new NullPointerException();
-  }
-  
-          slotName_ = value;
-          onChanged();
-          return this;
-        }
-        /**
-         * <pre>
-         * The name of the slot (e.g. "m"/"v").
-         * </pre>
-         *
-         * <code>string slot_name = 2;</code>
-         */
-        public Builder clearSlotName() {
-          
-          slotName_ = getDefaultInstance().getSlotName();
-          onChanged();
-          return this;
-        }
-        /**
-         * <pre>
-         * The name of the slot (e.g. "m"/"v").
-         * </pre>
-         *
-         * <code>string slot_name = 2;</code>
-         */
-        public Builder setSlotNameBytes(
-            com.google.protobuf.ByteString value) {
-          if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-          
-          slotName_ = value;
-          onChanged();
-          return this;
-        }
-
-        private int slotVariableNodeId_ ;
-        /**
-         * <pre>
-         * An index into `TrackableObjectGraph.nodes`, indicating the
-         * `Object` with the value of the slot variable.
-         * </pre>
-         *
-         * <code>int32 slot_variable_node_id = 3;</code>
-         */
-        public int getSlotVariableNodeId() {
-          return slotVariableNodeId_;
-        }
-        /**
-         * <pre>
-         * An index into `TrackableObjectGraph.nodes`, indicating the
-         * `Object` with the value of the slot variable.
-         * </pre>
-         *
-         * <code>int32 slot_variable_node_id = 3;</code>
-         */
-        public Builder setSlotVariableNodeId(int value) {
-          
-          slotVariableNodeId_ = value;
-          onChanged();
-          return this;
-        }
-        /**
-         * <pre>
-         * An index into `TrackableObjectGraph.nodes`, indicating the
-         * `Object` with the value of the slot variable.
-         * </pre>
-         *
-         * <code>int32 slot_variable_node_id = 3;</code>
-         */
-        public Builder clearSlotVariableNodeId() {
-          
-          slotVariableNodeId_ = 0;
-          onChanged();
-          return this;
-        }
-        @java.lang.Override
-        public final Builder setUnknownFields(
-            final com.google.protobuf.UnknownFieldSet unknownFields) {
-          return super.setUnknownFields(unknownFields);
-        }
-
-        @java.lang.Override
-        public final Builder mergeUnknownFields(
-            final com.google.protobuf.UnknownFieldSet unknownFields) {
-          return super.mergeUnknownFields(unknownFields);
-        }
-
-
-        // @@protoc_insertion_point(builder_scope:tensorflow.TrackableObjectGraph.TrackableObject.SlotVariableReference)
-      }
-
-      // @@protoc_insertion_point(class_scope:tensorflow.TrackableObjectGraph.TrackableObject.SlotVariableReference)
-      private static final org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference DEFAULT_INSTANCE;
-      static {
-        DEFAULT_INSTANCE = new org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference();
-      }
-
-      public static org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference getDefaultInstance() {
-        return DEFAULT_INSTANCE;
-      }
-
-      private static final com.google.protobuf.Parser<SlotVariableReference>
-          PARSER = new com.google.protobuf.AbstractParser<SlotVariableReference>() {
-        @java.lang.Override
-        public SlotVariableReference parsePartialFrom(
-            com.google.protobuf.CodedInputStream input,
-            com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-            throws com.google.protobuf.InvalidProtocolBufferException {
-          return new SlotVariableReference(input, extensionRegistry);
-        }
-      };
-
-      public static com.google.protobuf.Parser<SlotVariableReference> parser() {
-        return PARSER;
-      }
-
-      @java.lang.Override
-      public com.google.protobuf.Parser<SlotVariableReference> getParserForType() {
-        return PARSER;
-      }
-
-      @java.lang.Override
-      public org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference getDefaultInstanceForType() {
-        return DEFAULT_INSTANCE;
-      }
-
-    }
-
-    public static final int CHILDREN_FIELD_NUMBER = 1;
-    private java.util.List<org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference> children_;
-    /**
-     * <pre>
-     * Objects which this object depends on.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.ObjectReference children = 1;</code>
-     */
-    public java.util.List<org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference> getChildrenList() {
-      return children_;
-    }
-    /**
-     * <pre>
-     * Objects which this object depends on.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.ObjectReference children = 1;</code>
-     */
-    public java.util.List<? extends org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReferenceOrBuilder> 
-        getChildrenOrBuilderList() {
-      return children_;
-    }
-    /**
-     * <pre>
-     * Objects which this object depends on.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.ObjectReference children = 1;</code>
-     */
-    public int getChildrenCount() {
-      return children_.size();
-    }
-    /**
-     * <pre>
-     * Objects which this object depends on.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.ObjectReference children = 1;</code>
-     */
-    public org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference getChildren(int index) {
-      return children_.get(index);
-    }
-    /**
-     * <pre>
-     * Objects which this object depends on.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.ObjectReference children = 1;</code>
-     */
-    public org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReferenceOrBuilder getChildrenOrBuilder(
-        int index) {
-      return children_.get(index);
-    }
-
-    public static final int ATTRIBUTES_FIELD_NUMBER = 2;
-    private java.util.List<org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensor> attributes_;
-    /**
-     * <pre>
-     * Serialized data specific to this object.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SerializedTensor attributes = 2;</code>
-     */
-    public java.util.List<org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensor> getAttributesList() {
-      return attributes_;
-    }
-    /**
-     * <pre>
-     * Serialized data specific to this object.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SerializedTensor attributes = 2;</code>
-     */
-    public java.util.List<? extends org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensorOrBuilder> 
-        getAttributesOrBuilderList() {
-      return attributes_;
-    }
-    /**
-     * <pre>
-     * Serialized data specific to this object.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SerializedTensor attributes = 2;</code>
-     */
-    public int getAttributesCount() {
-      return attributes_.size();
-    }
-    /**
-     * <pre>
-     * Serialized data specific to this object.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SerializedTensor attributes = 2;</code>
-     */
-    public org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensor getAttributes(int index) {
-      return attributes_.get(index);
-    }
-    /**
-     * <pre>
-     * Serialized data specific to this object.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SerializedTensor attributes = 2;</code>
-     */
-    public org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensorOrBuilder getAttributesOrBuilder(
-        int index) {
-      return attributes_.get(index);
-    }
-
-    public static final int SLOT_VARIABLES_FIELD_NUMBER = 3;
-    private java.util.List<org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference> slotVariables_;
-    /**
-     * <pre>
-     * Slot variables owned by this object.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SlotVariableReference slot_variables = 3;</code>
-     */
-    public java.util.List<org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference> getSlotVariablesList() {
-      return slotVariables_;
-    }
-    /**
-     * <pre>
-     * Slot variables owned by this object.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SlotVariableReference slot_variables = 3;</code>
-     */
-    public java.util.List<? extends org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReferenceOrBuilder> 
-        getSlotVariablesOrBuilderList() {
-      return slotVariables_;
-    }
-    /**
-     * <pre>
-     * Slot variables owned by this object.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SlotVariableReference slot_variables = 3;</code>
-     */
-    public int getSlotVariablesCount() {
-      return slotVariables_.size();
-    }
-    /**
-     * <pre>
-     * Slot variables owned by this object.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SlotVariableReference slot_variables = 3;</code>
-     */
-    public org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference getSlotVariables(int index) {
-      return slotVariables_.get(index);
-    }
-    /**
-     * <pre>
-     * Slot variables owned by this object.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SlotVariableReference slot_variables = 3;</code>
-     */
-    public org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReferenceOrBuilder getSlotVariablesOrBuilder(
-        int index) {
-      return slotVariables_.get(index);
-    }
-
-    private byte memoizedIsInitialized = -1;
-    @java.lang.Override
-    public final boolean isInitialized() {
-      byte isInitialized = memoizedIsInitialized;
-      if (isInitialized == 1) return true;
-      if (isInitialized == 0) return false;
-
-      memoizedIsInitialized = 1;
-      return true;
-    }
-
-    @java.lang.Override
-    public void writeTo(com.google.protobuf.CodedOutputStream output)
-                        throws java.io.IOException {
-      for (int i = 0; i < children_.size(); i++) {
-        output.writeMessage(1, children_.get(i));
-      }
-      for (int i = 0; i < attributes_.size(); i++) {
-        output.writeMessage(2, attributes_.get(i));
-      }
-      for (int i = 0; i < slotVariables_.size(); i++) {
-        output.writeMessage(3, slotVariables_.get(i));
-      }
-      unknownFields.writeTo(output);
-    }
-
-    @java.lang.Override
-    public int getSerializedSize() {
-      int size = memoizedSize;
-      if (size != -1) return size;
-
-      size = 0;
-      for (int i = 0; i < children_.size(); i++) {
-        size += com.google.protobuf.CodedOutputStream
-          .computeMessageSize(1, children_.get(i));
-      }
-      for (int i = 0; i < attributes_.size(); i++) {
-        size += com.google.protobuf.CodedOutputStream
-          .computeMessageSize(2, attributes_.get(i));
-      }
-      for (int i = 0; i < slotVariables_.size(); i++) {
-        size += com.google.protobuf.CodedOutputStream
-          .computeMessageSize(3, slotVariables_.get(i));
-      }
-      size += unknownFields.getSerializedSize();
-      memoizedSize = size;
-      return size;
-    }
-
-    @java.lang.Override
-    public boolean equals(final java.lang.Object obj) {
-      if (obj == this) {
-       return true;
-      }
-      if (!(obj instanceof org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject)) {
-        return super.equals(obj);
-      }
-      org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject other = (org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject) obj;
-
-      if (!getChildrenList()
-          .equals(other.getChildrenList())) return false;
-      if (!getAttributesList()
-          .equals(other.getAttributesList())) return false;
-      if (!getSlotVariablesList()
-          .equals(other.getSlotVariablesList())) return false;
-      if (!unknownFields.equals(other.unknownFields)) return false;
-      return true;
-    }
-
-    @java.lang.Override
-    public int hashCode() {
-      if (memoizedHashCode != 0) {
-        return memoizedHashCode;
-      }
-      int hash = 41;
-      hash = (19 * hash) + getDescriptor().hashCode();
-      if (getChildrenCount() > 0) {
-        hash = (37 * hash) + CHILDREN_FIELD_NUMBER;
-        hash = (53 * hash) + getChildrenList().hashCode();
-      }
-      if (getAttributesCount() > 0) {
-        hash = (37 * hash) + ATTRIBUTES_FIELD_NUMBER;
-        hash = (53 * hash) + getAttributesList().hashCode();
-      }
-      if (getSlotVariablesCount() > 0) {
-        hash = (37 * hash) + SLOT_VARIABLES_FIELD_NUMBER;
-        hash = (53 * hash) + getSlotVariablesList().hashCode();
-      }
-      hash = (29 * hash) + unknownFields.hashCode();
-      memoizedHashCode = hash;
-      return hash;
-    }
-
-    public static org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject parseFrom(
-        java.nio.ByteBuffer data)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data);
-    }
-    public static org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject parseFrom(
-        java.nio.ByteBuffer data,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject parseFrom(
-        com.google.protobuf.ByteString data)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data);
-    }
-    public static org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject parseFrom(
-        com.google.protobuf.ByteString data,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject parseFrom(byte[] data)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data);
-    }
-    public static org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject parseFrom(
-        byte[] data,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return PARSER.parseFrom(data, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject parseFrom(java.io.InputStream input)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input);
-    }
-    public static org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject parseFrom(
-        java.io.InputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject parseDelimitedFrom(java.io.InputStream input)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseDelimitedWithIOException(PARSER, input);
-    }
-    public static org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject parseDelimitedFrom(
-        java.io.InputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-    }
-    public static org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject parseFrom(
-        com.google.protobuf.CodedInputStream input)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input);
-    }
-    public static org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject parseFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      return com.google.protobuf.GeneratedMessageV3
-          .parseWithIOException(PARSER, input, extensionRegistry);
-    }
-
-    @java.lang.Override
-    public Builder newBuilderForType() { return newBuilder(); }
-    public static Builder newBuilder() {
-      return DEFAULT_INSTANCE.toBuilder();
-    }
-    public static Builder newBuilder(org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject prototype) {
-      return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-    }
-    @java.lang.Override
-    public Builder toBuilder() {
-      return this == DEFAULT_INSTANCE
-          ? new Builder() : new Builder().mergeFrom(this);
-    }
-
-    @java.lang.Override
-    protected Builder newBuilderForType(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      Builder builder = new Builder(parent);
-      return builder;
-    }
-    /**
-     * Protobuf type {@code tensorflow.TrackableObjectGraph.TrackableObject}
-     */
-    public static final class Builder extends
-        com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-        // @@protoc_insertion_point(builder_implements:tensorflow.TrackableObjectGraph.TrackableObject)
-        org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObjectOrBuilder {
-      public static final com.google.protobuf.Descriptors.Descriptor
-          getDescriptor() {
-        return org.tensorflow.proto.framework.TrackableObjectGraphProtos.internal_static_tensorflow_TrackableObjectGraph_TrackableObject_descriptor;
-      }
-
-      @java.lang.Override
-      protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-          internalGetFieldAccessorTable() {
-        return org.tensorflow.proto.framework.TrackableObjectGraphProtos.internal_static_tensorflow_TrackableObjectGraph_TrackableObject_fieldAccessorTable
-            .ensureFieldAccessorsInitialized(
-                org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.class, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.Builder.class);
-      }
-
-      // Construct using org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.newBuilder()
-      private Builder() {
-        maybeForceBuilderInitialization();
-      }
-
-      private Builder(
-          com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-        super(parent);
-        maybeForceBuilderInitialization();
-      }
-      private void maybeForceBuilderInitialization() {
-        if (com.google.protobuf.GeneratedMessageV3
-                .alwaysUseFieldBuilders) {
-          getChildrenFieldBuilder();
-          getAttributesFieldBuilder();
-          getSlotVariablesFieldBuilder();
-        }
-      }
-      @java.lang.Override
-      public Builder clear() {
-        super.clear();
-        if (childrenBuilder_ == null) {
-          children_ = java.util.Collections.emptyList();
-          bitField0_ = (bitField0_ & ~0x00000001);
-        } else {
-          childrenBuilder_.clear();
-        }
-        if (attributesBuilder_ == null) {
-          attributes_ = java.util.Collections.emptyList();
-          bitField0_ = (bitField0_ & ~0x00000002);
-        } else {
-          attributesBuilder_.clear();
-        }
-        if (slotVariablesBuilder_ == null) {
-          slotVariables_ = java.util.Collections.emptyList();
-          bitField0_ = (bitField0_ & ~0x00000004);
-        } else {
-          slotVariablesBuilder_.clear();
-        }
-        return this;
-      }
-
-      @java.lang.Override
-      public com.google.protobuf.Descriptors.Descriptor
-          getDescriptorForType() {
-        return org.tensorflow.proto.framework.TrackableObjectGraphProtos.internal_static_tensorflow_TrackableObjectGraph_TrackableObject_descriptor;
-      }
-
-      @java.lang.Override
-      public org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject getDefaultInstanceForType() {
-        return org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.getDefaultInstance();
-      }
-
-      @java.lang.Override
-      public org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject build() {
-        org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject result = buildPartial();
-        if (!result.isInitialized()) {
-          throw newUninitializedMessageException(result);
-        }
-        return result;
-      }
-
-      @java.lang.Override
-      public org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject buildPartial() {
-        org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject result = new org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject(this);
-        int from_bitField0_ = bitField0_;
-        if (childrenBuilder_ == null) {
-          if (((bitField0_ & 0x00000001) != 0)) {
-            children_ = java.util.Collections.unmodifiableList(children_);
-            bitField0_ = (bitField0_ & ~0x00000001);
-          }
-          result.children_ = children_;
-        } else {
-          result.children_ = childrenBuilder_.build();
-        }
-        if (attributesBuilder_ == null) {
-          if (((bitField0_ & 0x00000002) != 0)) {
-            attributes_ = java.util.Collections.unmodifiableList(attributes_);
-            bitField0_ = (bitField0_ & ~0x00000002);
-          }
-          result.attributes_ = attributes_;
-        } else {
-          result.attributes_ = attributesBuilder_.build();
-        }
-        if (slotVariablesBuilder_ == null) {
-          if (((bitField0_ & 0x00000004) != 0)) {
-            slotVariables_ = java.util.Collections.unmodifiableList(slotVariables_);
-            bitField0_ = (bitField0_ & ~0x00000004);
-          }
-          result.slotVariables_ = slotVariables_;
-        } else {
-          result.slotVariables_ = slotVariablesBuilder_.build();
-        }
-        onBuilt();
-        return result;
-      }
-
-      @java.lang.Override
-      public Builder clone() {
-        return super.clone();
-      }
-      @java.lang.Override
-      public Builder setField(
-          com.google.protobuf.Descriptors.FieldDescriptor field,
-          java.lang.Object value) {
-        return super.setField(field, value);
-      }
-      @java.lang.Override
-      public Builder clearField(
-          com.google.protobuf.Descriptors.FieldDescriptor field) {
-        return super.clearField(field);
-      }
-      @java.lang.Override
-      public Builder clearOneof(
-          com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-        return super.clearOneof(oneof);
-      }
-      @java.lang.Override
-      public Builder setRepeatedField(
-          com.google.protobuf.Descriptors.FieldDescriptor field,
-          int index, java.lang.Object value) {
-        return super.setRepeatedField(field, index, value);
-      }
-      @java.lang.Override
-      public Builder addRepeatedField(
-          com.google.protobuf.Descriptors.FieldDescriptor field,
-          java.lang.Object value) {
-        return super.addRepeatedField(field, value);
-      }
-      @java.lang.Override
-      public Builder mergeFrom(com.google.protobuf.Message other) {
-        if (other instanceof org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject) {
-          return mergeFrom((org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject)other);
-        } else {
-          super.mergeFrom(other);
-          return this;
-        }
-      }
-
-      public Builder mergeFrom(org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject other) {
-        if (other == org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.getDefaultInstance()) return this;
-        if (childrenBuilder_ == null) {
-          if (!other.children_.isEmpty()) {
-            if (children_.isEmpty()) {
-              children_ = other.children_;
-              bitField0_ = (bitField0_ & ~0x00000001);
-            } else {
-              ensureChildrenIsMutable();
-              children_.addAll(other.children_);
-            }
-            onChanged();
-          }
-        } else {
-          if (!other.children_.isEmpty()) {
-            if (childrenBuilder_.isEmpty()) {
-              childrenBuilder_.dispose();
-              childrenBuilder_ = null;
-              children_ = other.children_;
-              bitField0_ = (bitField0_ & ~0x00000001);
-              childrenBuilder_ = 
-                com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
-                   getChildrenFieldBuilder() : null;
-            } else {
-              childrenBuilder_.addAllMessages(other.children_);
-            }
-          }
-        }
-        if (attributesBuilder_ == null) {
-          if (!other.attributes_.isEmpty()) {
-            if (attributes_.isEmpty()) {
-              attributes_ = other.attributes_;
-              bitField0_ = (bitField0_ & ~0x00000002);
-            } else {
-              ensureAttributesIsMutable();
-              attributes_.addAll(other.attributes_);
-            }
-            onChanged();
-          }
-        } else {
-          if (!other.attributes_.isEmpty()) {
-            if (attributesBuilder_.isEmpty()) {
-              attributesBuilder_.dispose();
-              attributesBuilder_ = null;
-              attributes_ = other.attributes_;
-              bitField0_ = (bitField0_ & ~0x00000002);
-              attributesBuilder_ = 
-                com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
-                   getAttributesFieldBuilder() : null;
-            } else {
-              attributesBuilder_.addAllMessages(other.attributes_);
-            }
-          }
-        }
-        if (slotVariablesBuilder_ == null) {
-          if (!other.slotVariables_.isEmpty()) {
-            if (slotVariables_.isEmpty()) {
-              slotVariables_ = other.slotVariables_;
-              bitField0_ = (bitField0_ & ~0x00000004);
-            } else {
-              ensureSlotVariablesIsMutable();
-              slotVariables_.addAll(other.slotVariables_);
-            }
-            onChanged();
-          }
-        } else {
-          if (!other.slotVariables_.isEmpty()) {
-            if (slotVariablesBuilder_.isEmpty()) {
-              slotVariablesBuilder_.dispose();
-              slotVariablesBuilder_ = null;
-              slotVariables_ = other.slotVariables_;
-              bitField0_ = (bitField0_ & ~0x00000004);
-              slotVariablesBuilder_ = 
-                com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
-                   getSlotVariablesFieldBuilder() : null;
-            } else {
-              slotVariablesBuilder_.addAllMessages(other.slotVariables_);
-            }
-          }
-        }
-        this.mergeUnknownFields(other.unknownFields);
-        onChanged();
-        return this;
-      }
-
-      @java.lang.Override
-      public final boolean isInitialized() {
-        return true;
-      }
-
-      @java.lang.Override
-      public Builder mergeFrom(
-          com.google.protobuf.CodedInputStream input,
-          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-          throws java.io.IOException {
-        org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject parsedMessage = null;
-        try {
-          parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-        } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-          parsedMessage = (org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject) e.getUnfinishedMessage();
-          throw e.unwrapIOException();
-        } finally {
-          if (parsedMessage != null) {
-            mergeFrom(parsedMessage);
-          }
-        }
-        return this;
-      }
-      private int bitField0_;
-
-      private java.util.List<org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference> children_ =
-        java.util.Collections.emptyList();
-      private void ensureChildrenIsMutable() {
-        if (!((bitField0_ & 0x00000001) != 0)) {
-          children_ = new java.util.ArrayList<org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference>(children_);
-          bitField0_ |= 0x00000001;
-         }
-      }
-
-      private com.google.protobuf.RepeatedFieldBuilderV3<
-          org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference.Builder, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReferenceOrBuilder> childrenBuilder_;
-
-      /**
-       * <pre>
-       * Objects which this object depends on.
-       * </pre>
-       *
-       * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.ObjectReference children = 1;</code>
-       */
-      public java.util.List<org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference> getChildrenList() {
-        if (childrenBuilder_ == null) {
-          return java.util.Collections.unmodifiableList(children_);
-        } else {
-          return childrenBuilder_.getMessageList();
-        }
-      }
-      /**
-       * <pre>
-       * Objects which this object depends on.
-       * </pre>
-       *
-       * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.ObjectReference children = 1;</code>
-       */
-      public int getChildrenCount() {
-        if (childrenBuilder_ == null) {
-          return children_.size();
-        } else {
-          return childrenBuilder_.getCount();
-        }
-      }
-      /**
-       * <pre>
-       * Objects which this object depends on.
-       * </pre>
-       *
-       * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.ObjectReference children = 1;</code>
-       */
-      public org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference getChildren(int index) {
-        if (childrenBuilder_ == null) {
-          return children_.get(index);
-        } else {
-          return childrenBuilder_.getMessage(index);
-        }
-      }
-      /**
-       * <pre>
-       * Objects which this object depends on.
-       * </pre>
-       *
-       * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.ObjectReference children = 1;</code>
-       */
-      public Builder setChildren(
-          int index, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference value) {
-        if (childrenBuilder_ == null) {
-          if (value == null) {
-            throw new NullPointerException();
-          }
-          ensureChildrenIsMutable();
-          children_.set(index, value);
-          onChanged();
-        } else {
-          childrenBuilder_.setMessage(index, value);
-        }
-        return this;
-      }
-      /**
-       * <pre>
-       * Objects which this object depends on.
-       * </pre>
-       *
-       * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.ObjectReference children = 1;</code>
-       */
-      public Builder setChildren(
-          int index, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference.Builder builderForValue) {
-        if (childrenBuilder_ == null) {
-          ensureChildrenIsMutable();
-          children_.set(index, builderForValue.build());
-          onChanged();
-        } else {
-          childrenBuilder_.setMessage(index, builderForValue.build());
-        }
-        return this;
-      }
-      /**
-       * <pre>
-       * Objects which this object depends on.
-       * </pre>
-       *
-       * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.ObjectReference children = 1;</code>
-       */
-      public Builder addChildren(org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference value) {
-        if (childrenBuilder_ == null) {
-          if (value == null) {
-            throw new NullPointerException();
-          }
-          ensureChildrenIsMutable();
-          children_.add(value);
-          onChanged();
-        } else {
-          childrenBuilder_.addMessage(value);
-        }
-        return this;
-      }
-      /**
-       * <pre>
-       * Objects which this object depends on.
-       * </pre>
-       *
-       * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.ObjectReference children = 1;</code>
-       */
-      public Builder addChildren(
-          int index, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference value) {
-        if (childrenBuilder_ == null) {
-          if (value == null) {
-            throw new NullPointerException();
-          }
-          ensureChildrenIsMutable();
-          children_.add(index, value);
-          onChanged();
-        } else {
-          childrenBuilder_.addMessage(index, value);
-        }
-        return this;
-      }
-      /**
-       * <pre>
-       * Objects which this object depends on.
-       * </pre>
-       *
-       * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.ObjectReference children = 1;</code>
-       */
-      public Builder addChildren(
-          org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference.Builder builderForValue) {
-        if (childrenBuilder_ == null) {
-          ensureChildrenIsMutable();
-          children_.add(builderForValue.build());
-          onChanged();
-        } else {
-          childrenBuilder_.addMessage(builderForValue.build());
-        }
-        return this;
-      }
-      /**
-       * <pre>
-       * Objects which this object depends on.
-       * </pre>
-       *
-       * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.ObjectReference children = 1;</code>
-       */
-      public Builder addChildren(
-          int index, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference.Builder builderForValue) {
-        if (childrenBuilder_ == null) {
-          ensureChildrenIsMutable();
-          children_.add(index, builderForValue.build());
-          onChanged();
-        } else {
-          childrenBuilder_.addMessage(index, builderForValue.build());
-        }
-        return this;
-      }
-      /**
-       * <pre>
-       * Objects which this object depends on.
-       * </pre>
-       *
-       * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.ObjectReference children = 1;</code>
-       */
-      public Builder addAllChildren(
-          java.lang.Iterable<? extends org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference> values) {
-        if (childrenBuilder_ == null) {
-          ensureChildrenIsMutable();
-          com.google.protobuf.AbstractMessageLite.Builder.addAll(
-              values, children_);
-          onChanged();
-        } else {
-          childrenBuilder_.addAllMessages(values);
-        }
-        return this;
-      }
-      /**
-       * <pre>
-       * Objects which this object depends on.
-       * </pre>
-       *
-       * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.ObjectReference children = 1;</code>
-       */
-      public Builder clearChildren() {
-        if (childrenBuilder_ == null) {
-          children_ = java.util.Collections.emptyList();
-          bitField0_ = (bitField0_ & ~0x00000001);
-          onChanged();
-        } else {
-          childrenBuilder_.clear();
-        }
-        return this;
-      }
-      /**
-       * <pre>
-       * Objects which this object depends on.
-       * </pre>
-       *
-       * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.ObjectReference children = 1;</code>
-       */
-      public Builder removeChildren(int index) {
-        if (childrenBuilder_ == null) {
-          ensureChildrenIsMutable();
-          children_.remove(index);
-          onChanged();
-        } else {
-          childrenBuilder_.remove(index);
-        }
-        return this;
-      }
-      /**
-       * <pre>
-       * Objects which this object depends on.
-       * </pre>
-       *
-       * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.ObjectReference children = 1;</code>
-       */
-      public org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference.Builder getChildrenBuilder(
-          int index) {
-        return getChildrenFieldBuilder().getBuilder(index);
-      }
-      /**
-       * <pre>
-       * Objects which this object depends on.
-       * </pre>
-       *
-       * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.ObjectReference children = 1;</code>
-       */
-      public org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReferenceOrBuilder getChildrenOrBuilder(
-          int index) {
-        if (childrenBuilder_ == null) {
-          return children_.get(index);  } else {
-          return childrenBuilder_.getMessageOrBuilder(index);
-        }
-      }
-      /**
-       * <pre>
-       * Objects which this object depends on.
-       * </pre>
-       *
-       * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.ObjectReference children = 1;</code>
-       */
-      public java.util.List<? extends org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReferenceOrBuilder> 
-           getChildrenOrBuilderList() {
-        if (childrenBuilder_ != null) {
-          return childrenBuilder_.getMessageOrBuilderList();
-        } else {
-          return java.util.Collections.unmodifiableList(children_);
-        }
-      }
-      /**
-       * <pre>
-       * Objects which this object depends on.
-       * </pre>
-       *
-       * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.ObjectReference children = 1;</code>
-       */
-      public org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference.Builder addChildrenBuilder() {
-        return getChildrenFieldBuilder().addBuilder(
-            org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference.getDefaultInstance());
-      }
-      /**
-       * <pre>
-       * Objects which this object depends on.
-       * </pre>
-       *
-       * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.ObjectReference children = 1;</code>
-       */
-      public org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference.Builder addChildrenBuilder(
-          int index) {
-        return getChildrenFieldBuilder().addBuilder(
-            index, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference.getDefaultInstance());
-      }
-      /**
-       * <pre>
-       * Objects which this object depends on.
-       * </pre>
-       *
-       * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.ObjectReference children = 1;</code>
-       */
-      public java.util.List<org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference.Builder> 
-           getChildrenBuilderList() {
-        return getChildrenFieldBuilder().getBuilderList();
-      }
-      private com.google.protobuf.RepeatedFieldBuilderV3<
-          org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference.Builder, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReferenceOrBuilder> 
-          getChildrenFieldBuilder() {
-        if (childrenBuilder_ == null) {
-          childrenBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
-              org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference.Builder, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReferenceOrBuilder>(
-                  children_,
-                  ((bitField0_ & 0x00000001) != 0),
-                  getParentForChildren(),
-                  isClean());
-          children_ = null;
-        }
-        return childrenBuilder_;
-      }
-
-      private java.util.List<org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensor> attributes_ =
-        java.util.Collections.emptyList();
-      private void ensureAttributesIsMutable() {
-        if (!((bitField0_ & 0x00000002) != 0)) {
-          attributes_ = new java.util.ArrayList<org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensor>(attributes_);
-          bitField0_ |= 0x00000002;
-         }
-      }
-
-      private com.google.protobuf.RepeatedFieldBuilderV3<
-          org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensor, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensor.Builder, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensorOrBuilder> attributesBuilder_;
-
-      /**
-       * <pre>
-       * Serialized data specific to this object.
-       * </pre>
-       *
-       * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SerializedTensor attributes = 2;</code>
-       */
-      public java.util.List<org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensor> getAttributesList() {
-        if (attributesBuilder_ == null) {
-          return java.util.Collections.unmodifiableList(attributes_);
-        } else {
-          return attributesBuilder_.getMessageList();
-        }
-      }
-      /**
-       * <pre>
-       * Serialized data specific to this object.
-       * </pre>
-       *
-       * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SerializedTensor attributes = 2;</code>
-       */
-      public int getAttributesCount() {
-        if (attributesBuilder_ == null) {
-          return attributes_.size();
-        } else {
-          return attributesBuilder_.getCount();
-        }
-      }
-      /**
-       * <pre>
-       * Serialized data specific to this object.
-       * </pre>
-       *
-       * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SerializedTensor attributes = 2;</code>
-       */
-      public org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensor getAttributes(int index) {
-        if (attributesBuilder_ == null) {
-          return attributes_.get(index);
-        } else {
-          return attributesBuilder_.getMessage(index);
-        }
-      }
-      /**
-       * <pre>
-       * Serialized data specific to this object.
-       * </pre>
-       *
-       * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SerializedTensor attributes = 2;</code>
-       */
-      public Builder setAttributes(
-          int index, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensor value) {
-        if (attributesBuilder_ == null) {
-          if (value == null) {
-            throw new NullPointerException();
-          }
-          ensureAttributesIsMutable();
-          attributes_.set(index, value);
-          onChanged();
-        } else {
-          attributesBuilder_.setMessage(index, value);
-        }
-        return this;
-      }
-      /**
-       * <pre>
-       * Serialized data specific to this object.
-       * </pre>
-       *
-       * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SerializedTensor attributes = 2;</code>
-       */
-      public Builder setAttributes(
-          int index, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensor.Builder builderForValue) {
-        if (attributesBuilder_ == null) {
-          ensureAttributesIsMutable();
-          attributes_.set(index, builderForValue.build());
-          onChanged();
-        } else {
-          attributesBuilder_.setMessage(index, builderForValue.build());
-        }
-        return this;
-      }
-      /**
-       * <pre>
-       * Serialized data specific to this object.
-       * </pre>
-       *
-       * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SerializedTensor attributes = 2;</code>
-       */
-      public Builder addAttributes(org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensor value) {
-        if (attributesBuilder_ == null) {
-          if (value == null) {
-            throw new NullPointerException();
-          }
-          ensureAttributesIsMutable();
-          attributes_.add(value);
-          onChanged();
-        } else {
-          attributesBuilder_.addMessage(value);
-        }
-        return this;
-      }
-      /**
-       * <pre>
-       * Serialized data specific to this object.
-       * </pre>
-       *
-       * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SerializedTensor attributes = 2;</code>
-       */
-      public Builder addAttributes(
-          int index, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensor value) {
-        if (attributesBuilder_ == null) {
-          if (value == null) {
-            throw new NullPointerException();
-          }
-          ensureAttributesIsMutable();
-          attributes_.add(index, value);
-          onChanged();
-        } else {
-          attributesBuilder_.addMessage(index, value);
-        }
-        return this;
-      }
-      /**
-       * <pre>
-       * Serialized data specific to this object.
-       * </pre>
-       *
-       * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SerializedTensor attributes = 2;</code>
-       */
-      public Builder addAttributes(
-          org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensor.Builder builderForValue) {
-        if (attributesBuilder_ == null) {
-          ensureAttributesIsMutable();
-          attributes_.add(builderForValue.build());
-          onChanged();
-        } else {
-          attributesBuilder_.addMessage(builderForValue.build());
-        }
-        return this;
-      }
-      /**
-       * <pre>
-       * Serialized data specific to this object.
-       * </pre>
-       *
-       * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SerializedTensor attributes = 2;</code>
-       */
-      public Builder addAttributes(
-          int index, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensor.Builder builderForValue) {
-        if (attributesBuilder_ == null) {
-          ensureAttributesIsMutable();
-          attributes_.add(index, builderForValue.build());
-          onChanged();
-        } else {
-          attributesBuilder_.addMessage(index, builderForValue.build());
-        }
-        return this;
-      }
-      /**
-       * <pre>
-       * Serialized data specific to this object.
-       * </pre>
-       *
-       * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SerializedTensor attributes = 2;</code>
-       */
-      public Builder addAllAttributes(
-          java.lang.Iterable<? extends org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensor> values) {
-        if (attributesBuilder_ == null) {
-          ensureAttributesIsMutable();
-          com.google.protobuf.AbstractMessageLite.Builder.addAll(
-              values, attributes_);
-          onChanged();
-        } else {
-          attributesBuilder_.addAllMessages(values);
-        }
-        return this;
-      }
-      /**
-       * <pre>
-       * Serialized data specific to this object.
-       * </pre>
-       *
-       * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SerializedTensor attributes = 2;</code>
-       */
-      public Builder clearAttributes() {
-        if (attributesBuilder_ == null) {
-          attributes_ = java.util.Collections.emptyList();
-          bitField0_ = (bitField0_ & ~0x00000002);
-          onChanged();
-        } else {
-          attributesBuilder_.clear();
-        }
-        return this;
-      }
-      /**
-       * <pre>
-       * Serialized data specific to this object.
-       * </pre>
-       *
-       * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SerializedTensor attributes = 2;</code>
-       */
-      public Builder removeAttributes(int index) {
-        if (attributesBuilder_ == null) {
-          ensureAttributesIsMutable();
-          attributes_.remove(index);
-          onChanged();
-        } else {
-          attributesBuilder_.remove(index);
-        }
-        return this;
-      }
-      /**
-       * <pre>
-       * Serialized data specific to this object.
-       * </pre>
-       *
-       * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SerializedTensor attributes = 2;</code>
-       */
-      public org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensor.Builder getAttributesBuilder(
-          int index) {
-        return getAttributesFieldBuilder().getBuilder(index);
-      }
-      /**
-       * <pre>
-       * Serialized data specific to this object.
-       * </pre>
-       *
-       * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SerializedTensor attributes = 2;</code>
-       */
-      public org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensorOrBuilder getAttributesOrBuilder(
-          int index) {
-        if (attributesBuilder_ == null) {
-          return attributes_.get(index);  } else {
-          return attributesBuilder_.getMessageOrBuilder(index);
-        }
-      }
-      /**
-       * <pre>
-       * Serialized data specific to this object.
-       * </pre>
-       *
-       * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SerializedTensor attributes = 2;</code>
-       */
-      public java.util.List<? extends org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensorOrBuilder> 
-           getAttributesOrBuilderList() {
-        if (attributesBuilder_ != null) {
-          return attributesBuilder_.getMessageOrBuilderList();
-        } else {
-          return java.util.Collections.unmodifiableList(attributes_);
-        }
-      }
-      /**
-       * <pre>
-       * Serialized data specific to this object.
-       * </pre>
-       *
-       * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SerializedTensor attributes = 2;</code>
-       */
-      public org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensor.Builder addAttributesBuilder() {
-        return getAttributesFieldBuilder().addBuilder(
-            org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensor.getDefaultInstance());
-      }
-      /**
-       * <pre>
-       * Serialized data specific to this object.
-       * </pre>
-       *
-       * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SerializedTensor attributes = 2;</code>
-       */
-      public org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensor.Builder addAttributesBuilder(
-          int index) {
-        return getAttributesFieldBuilder().addBuilder(
-            index, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensor.getDefaultInstance());
-      }
-      /**
-       * <pre>
-       * Serialized data specific to this object.
-       * </pre>
-       *
-       * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SerializedTensor attributes = 2;</code>
-       */
-      public java.util.List<org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensor.Builder> 
-           getAttributesBuilderList() {
-        return getAttributesFieldBuilder().getBuilderList();
-      }
-      private com.google.protobuf.RepeatedFieldBuilderV3<
-          org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensor, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensor.Builder, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensorOrBuilder> 
-          getAttributesFieldBuilder() {
-        if (attributesBuilder_ == null) {
-          attributesBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
-              org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensor, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensor.Builder, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensorOrBuilder>(
-                  attributes_,
-                  ((bitField0_ & 0x00000002) != 0),
-                  getParentForChildren(),
-                  isClean());
-          attributes_ = null;
-        }
-        return attributesBuilder_;
-      }
-
-      private java.util.List<org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference> slotVariables_ =
-        java.util.Collections.emptyList();
-      private void ensureSlotVariablesIsMutable() {
-        if (!((bitField0_ & 0x00000004) != 0)) {
-          slotVariables_ = new java.util.ArrayList<org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference>(slotVariables_);
-          bitField0_ |= 0x00000004;
-         }
-      }
-
-      private com.google.protobuf.RepeatedFieldBuilderV3<
-          org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference.Builder, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReferenceOrBuilder> slotVariablesBuilder_;
-
-      /**
-       * <pre>
-       * Slot variables owned by this object.
-       * </pre>
-       *
-       * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SlotVariableReference slot_variables = 3;</code>
-       */
-      public java.util.List<org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference> getSlotVariablesList() {
-        if (slotVariablesBuilder_ == null) {
-          return java.util.Collections.unmodifiableList(slotVariables_);
-        } else {
-          return slotVariablesBuilder_.getMessageList();
-        }
-      }
-      /**
-       * <pre>
-       * Slot variables owned by this object.
-       * </pre>
-       *
-       * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SlotVariableReference slot_variables = 3;</code>
-       */
-      public int getSlotVariablesCount() {
-        if (slotVariablesBuilder_ == null) {
-          return slotVariables_.size();
-        } else {
-          return slotVariablesBuilder_.getCount();
-        }
-      }
-      /**
-       * <pre>
-       * Slot variables owned by this object.
-       * </pre>
-       *
-       * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SlotVariableReference slot_variables = 3;</code>
-       */
-      public org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference getSlotVariables(int index) {
-        if (slotVariablesBuilder_ == null) {
-          return slotVariables_.get(index);
-        } else {
-          return slotVariablesBuilder_.getMessage(index);
-        }
-      }
-      /**
-       * <pre>
-       * Slot variables owned by this object.
-       * </pre>
-       *
-       * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SlotVariableReference slot_variables = 3;</code>
-       */
-      public Builder setSlotVariables(
-          int index, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference value) {
-        if (slotVariablesBuilder_ == null) {
-          if (value == null) {
-            throw new NullPointerException();
-          }
-          ensureSlotVariablesIsMutable();
-          slotVariables_.set(index, value);
-          onChanged();
-        } else {
-          slotVariablesBuilder_.setMessage(index, value);
-        }
-        return this;
-      }
-      /**
-       * <pre>
-       * Slot variables owned by this object.
-       * </pre>
-       *
-       * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SlotVariableReference slot_variables = 3;</code>
-       */
-      public Builder setSlotVariables(
-          int index, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference.Builder builderForValue) {
-        if (slotVariablesBuilder_ == null) {
-          ensureSlotVariablesIsMutable();
-          slotVariables_.set(index, builderForValue.build());
-          onChanged();
-        } else {
-          slotVariablesBuilder_.setMessage(index, builderForValue.build());
-        }
-        return this;
-      }
-      /**
-       * <pre>
-       * Slot variables owned by this object.
-       * </pre>
-       *
-       * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SlotVariableReference slot_variables = 3;</code>
-       */
-      public Builder addSlotVariables(org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference value) {
-        if (slotVariablesBuilder_ == null) {
-          if (value == null) {
-            throw new NullPointerException();
-          }
-          ensureSlotVariablesIsMutable();
-          slotVariables_.add(value);
-          onChanged();
-        } else {
-          slotVariablesBuilder_.addMessage(value);
-        }
-        return this;
-      }
-      /**
-       * <pre>
-       * Slot variables owned by this object.
-       * </pre>
-       *
-       * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SlotVariableReference slot_variables = 3;</code>
-       */
-      public Builder addSlotVariables(
-          int index, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference value) {
-        if (slotVariablesBuilder_ == null) {
-          if (value == null) {
-            throw new NullPointerException();
-          }
-          ensureSlotVariablesIsMutable();
-          slotVariables_.add(index, value);
-          onChanged();
-        } else {
-          slotVariablesBuilder_.addMessage(index, value);
-        }
-        return this;
-      }
-      /**
-       * <pre>
-       * Slot variables owned by this object.
-       * </pre>
-       *
-       * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SlotVariableReference slot_variables = 3;</code>
-       */
-      public Builder addSlotVariables(
-          org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference.Builder builderForValue) {
-        if (slotVariablesBuilder_ == null) {
-          ensureSlotVariablesIsMutable();
-          slotVariables_.add(builderForValue.build());
-          onChanged();
-        } else {
-          slotVariablesBuilder_.addMessage(builderForValue.build());
-        }
-        return this;
-      }
-      /**
-       * <pre>
-       * Slot variables owned by this object.
-       * </pre>
-       *
-       * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SlotVariableReference slot_variables = 3;</code>
-       */
-      public Builder addSlotVariables(
-          int index, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference.Builder builderForValue) {
-        if (slotVariablesBuilder_ == null) {
-          ensureSlotVariablesIsMutable();
-          slotVariables_.add(index, builderForValue.build());
-          onChanged();
-        } else {
-          slotVariablesBuilder_.addMessage(index, builderForValue.build());
-        }
-        return this;
-      }
-      /**
-       * <pre>
-       * Slot variables owned by this object.
-       * </pre>
-       *
-       * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SlotVariableReference slot_variables = 3;</code>
-       */
-      public Builder addAllSlotVariables(
-          java.lang.Iterable<? extends org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference> values) {
-        if (slotVariablesBuilder_ == null) {
-          ensureSlotVariablesIsMutable();
-          com.google.protobuf.AbstractMessageLite.Builder.addAll(
-              values, slotVariables_);
-          onChanged();
-        } else {
-          slotVariablesBuilder_.addAllMessages(values);
-        }
-        return this;
-      }
-      /**
-       * <pre>
-       * Slot variables owned by this object.
-       * </pre>
-       *
-       * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SlotVariableReference slot_variables = 3;</code>
-       */
-      public Builder clearSlotVariables() {
-        if (slotVariablesBuilder_ == null) {
-          slotVariables_ = java.util.Collections.emptyList();
-          bitField0_ = (bitField0_ & ~0x00000004);
-          onChanged();
-        } else {
-          slotVariablesBuilder_.clear();
-        }
-        return this;
-      }
-      /**
-       * <pre>
-       * Slot variables owned by this object.
-       * </pre>
-       *
-       * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SlotVariableReference slot_variables = 3;</code>
-       */
-      public Builder removeSlotVariables(int index) {
-        if (slotVariablesBuilder_ == null) {
-          ensureSlotVariablesIsMutable();
-          slotVariables_.remove(index);
-          onChanged();
-        } else {
-          slotVariablesBuilder_.remove(index);
-        }
-        return this;
-      }
-      /**
-       * <pre>
-       * Slot variables owned by this object.
-       * </pre>
-       *
-       * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SlotVariableReference slot_variables = 3;</code>
-       */
-      public org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference.Builder getSlotVariablesBuilder(
-          int index) {
-        return getSlotVariablesFieldBuilder().getBuilder(index);
-      }
-      /**
-       * <pre>
-       * Slot variables owned by this object.
-       * </pre>
-       *
-       * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SlotVariableReference slot_variables = 3;</code>
-       */
-      public org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReferenceOrBuilder getSlotVariablesOrBuilder(
-          int index) {
-        if (slotVariablesBuilder_ == null) {
-          return slotVariables_.get(index);  } else {
-          return slotVariablesBuilder_.getMessageOrBuilder(index);
-        }
-      }
-      /**
-       * <pre>
-       * Slot variables owned by this object.
-       * </pre>
-       *
-       * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SlotVariableReference slot_variables = 3;</code>
-       */
-      public java.util.List<? extends org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReferenceOrBuilder> 
-           getSlotVariablesOrBuilderList() {
-        if (slotVariablesBuilder_ != null) {
-          return slotVariablesBuilder_.getMessageOrBuilderList();
-        } else {
-          return java.util.Collections.unmodifiableList(slotVariables_);
-        }
-      }
-      /**
-       * <pre>
-       * Slot variables owned by this object.
-       * </pre>
-       *
-       * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SlotVariableReference slot_variables = 3;</code>
-       */
-      public org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference.Builder addSlotVariablesBuilder() {
-        return getSlotVariablesFieldBuilder().addBuilder(
-            org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference.getDefaultInstance());
-      }
-      /**
-       * <pre>
-       * Slot variables owned by this object.
-       * </pre>
-       *
-       * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SlotVariableReference slot_variables = 3;</code>
-       */
-      public org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference.Builder addSlotVariablesBuilder(
-          int index) {
-        return getSlotVariablesFieldBuilder().addBuilder(
-            index, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference.getDefaultInstance());
-      }
-      /**
-       * <pre>
-       * Slot variables owned by this object.
-       * </pre>
-       *
-       * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SlotVariableReference slot_variables = 3;</code>
-       */
-      public java.util.List<org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference.Builder> 
-           getSlotVariablesBuilderList() {
-        return getSlotVariablesFieldBuilder().getBuilderList();
-      }
-      private com.google.protobuf.RepeatedFieldBuilderV3<
-          org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference.Builder, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReferenceOrBuilder> 
-          getSlotVariablesFieldBuilder() {
-        if (slotVariablesBuilder_ == null) {
-          slotVariablesBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
-              org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference.Builder, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReferenceOrBuilder>(
-                  slotVariables_,
-                  ((bitField0_ & 0x00000004) != 0),
-                  getParentForChildren(),
-                  isClean());
-          slotVariables_ = null;
-        }
-        return slotVariablesBuilder_;
-      }
-      @java.lang.Override
-      public final Builder setUnknownFields(
-          final com.google.protobuf.UnknownFieldSet unknownFields) {
-        return super.setUnknownFields(unknownFields);
-      }
-
-      @java.lang.Override
-      public final Builder mergeUnknownFields(
-          final com.google.protobuf.UnknownFieldSet unknownFields) {
-        return super.mergeUnknownFields(unknownFields);
-      }
-
-
-      // @@protoc_insertion_point(builder_scope:tensorflow.TrackableObjectGraph.TrackableObject)
-    }
-
-    // @@protoc_insertion_point(class_scope:tensorflow.TrackableObjectGraph.TrackableObject)
-    private static final org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject DEFAULT_INSTANCE;
-    static {
-      DEFAULT_INSTANCE = new org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject();
-    }
-
-    public static org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject getDefaultInstance() {
-      return DEFAULT_INSTANCE;
-    }
-
-    private static final com.google.protobuf.Parser<TrackableObject>
-        PARSER = new com.google.protobuf.AbstractParser<TrackableObject>() {
-      @java.lang.Override
-      public TrackableObject parsePartialFrom(
-          com.google.protobuf.CodedInputStream input,
-          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-          throws com.google.protobuf.InvalidProtocolBufferException {
-        return new TrackableObject(input, extensionRegistry);
-      }
-    };
-
-    public static com.google.protobuf.Parser<TrackableObject> parser() {
-      return PARSER;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Parser<TrackableObject> getParserForType() {
-      return PARSER;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject getDefaultInstanceForType() {
-      return DEFAULT_INSTANCE;
-    }
-
-  }
-
-  public static final int NODES_FIELD_NUMBER = 1;
-  private java.util.List<org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject> nodes_;
-  /**
-   * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject nodes = 1;</code>
-   */
-  public java.util.List<org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject> getNodesList() {
-    return nodes_;
-  }
-  /**
-   * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject nodes = 1;</code>
-   */
-  public java.util.List<? extends org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObjectOrBuilder> 
-      getNodesOrBuilderList() {
-    return nodes_;
-  }
-  /**
-   * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject nodes = 1;</code>
-   */
-  public int getNodesCount() {
-    return nodes_.size();
-  }
-  /**
-   * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject nodes = 1;</code>
-   */
-  public org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject getNodes(int index) {
-    return nodes_.get(index);
-  }
-  /**
-   * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject nodes = 1;</code>
-   */
-  public org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObjectOrBuilder getNodesOrBuilder(
-      int index) {
-    return nodes_.get(index);
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    for (int i = 0; i < nodes_.size(); i++) {
-      output.writeMessage(1, nodes_.get(i));
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    for (int i = 0; i < nodes_.size(); i++) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(1, nodes_.get(i));
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.framework.TrackableObjectGraph)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.framework.TrackableObjectGraph other = (org.tensorflow.proto.framework.TrackableObjectGraph) obj;
-
-    if (!getNodesList()
-        .equals(other.getNodesList())) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    if (getNodesCount() > 0) {
-      hash = (37 * hash) + NODES_FIELD_NUMBER;
-      hash = (53 * hash) + getNodesList().hashCode();
-    }
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.framework.TrackableObjectGraph parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.TrackableObjectGraph parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.TrackableObjectGraph parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.TrackableObjectGraph parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.TrackableObjectGraph parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.TrackableObjectGraph parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.TrackableObjectGraph parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.TrackableObjectGraph parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.TrackableObjectGraph parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.TrackableObjectGraph parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.TrackableObjectGraph parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.TrackableObjectGraph parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.framework.TrackableObjectGraph prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * Protobuf type {@code tensorflow.TrackableObjectGraph}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.TrackableObjectGraph)
-      org.tensorflow.proto.framework.TrackableObjectGraphOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.TrackableObjectGraphProtos.internal_static_tensorflow_TrackableObjectGraph_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.TrackableObjectGraphProtos.internal_static_tensorflow_TrackableObjectGraph_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.TrackableObjectGraph.class, org.tensorflow.proto.framework.TrackableObjectGraph.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.framework.TrackableObjectGraph.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-        getNodesFieldBuilder();
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      if (nodesBuilder_ == null) {
-        nodes_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000001);
-      } else {
-        nodesBuilder_.clear();
-      }
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.framework.TrackableObjectGraphProtos.internal_static_tensorflow_TrackableObjectGraph_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.TrackableObjectGraph getDefaultInstanceForType() {
-      return org.tensorflow.proto.framework.TrackableObjectGraph.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.TrackableObjectGraph build() {
-      org.tensorflow.proto.framework.TrackableObjectGraph result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.TrackableObjectGraph buildPartial() {
-      org.tensorflow.proto.framework.TrackableObjectGraph result = new org.tensorflow.proto.framework.TrackableObjectGraph(this);
-      int from_bitField0_ = bitField0_;
-      if (nodesBuilder_ == null) {
-        if (((bitField0_ & 0x00000001) != 0)) {
-          nodes_ = java.util.Collections.unmodifiableList(nodes_);
-          bitField0_ = (bitField0_ & ~0x00000001);
-        }
-        result.nodes_ = nodes_;
-      } else {
-        result.nodes_ = nodesBuilder_.build();
-      }
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.framework.TrackableObjectGraph) {
-        return mergeFrom((org.tensorflow.proto.framework.TrackableObjectGraph)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.framework.TrackableObjectGraph other) {
-      if (other == org.tensorflow.proto.framework.TrackableObjectGraph.getDefaultInstance()) return this;
-      if (nodesBuilder_ == null) {
-        if (!other.nodes_.isEmpty()) {
-          if (nodes_.isEmpty()) {
-            nodes_ = other.nodes_;
-            bitField0_ = (bitField0_ & ~0x00000001);
-          } else {
-            ensureNodesIsMutable();
-            nodes_.addAll(other.nodes_);
-          }
-          onChanged();
-        }
-      } else {
-        if (!other.nodes_.isEmpty()) {
-          if (nodesBuilder_.isEmpty()) {
-            nodesBuilder_.dispose();
-            nodesBuilder_ = null;
-            nodes_ = other.nodes_;
-            bitField0_ = (bitField0_ & ~0x00000001);
-            nodesBuilder_ = 
-              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
-                 getNodesFieldBuilder() : null;
-          } else {
-            nodesBuilder_.addAllMessages(other.nodes_);
-          }
-        }
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.framework.TrackableObjectGraph parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.framework.TrackableObjectGraph) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-    private int bitField0_;
-
-    private java.util.List<org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject> nodes_ =
-      java.util.Collections.emptyList();
-    private void ensureNodesIsMutable() {
-      if (!((bitField0_ & 0x00000001) != 0)) {
-        nodes_ = new java.util.ArrayList<org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject>(nodes_);
-        bitField0_ |= 0x00000001;
-       }
-    }
-
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.Builder, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObjectOrBuilder> nodesBuilder_;
-
-    /**
-     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject nodes = 1;</code>
-     */
-    public java.util.List<org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject> getNodesList() {
-      if (nodesBuilder_ == null) {
-        return java.util.Collections.unmodifiableList(nodes_);
-      } else {
-        return nodesBuilder_.getMessageList();
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject nodes = 1;</code>
-     */
-    public int getNodesCount() {
-      if (nodesBuilder_ == null) {
-        return nodes_.size();
-      } else {
-        return nodesBuilder_.getCount();
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject nodes = 1;</code>
-     */
-    public org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject getNodes(int index) {
-      if (nodesBuilder_ == null) {
-        return nodes_.get(index);
-      } else {
-        return nodesBuilder_.getMessage(index);
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject nodes = 1;</code>
-     */
-    public Builder setNodes(
-        int index, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject value) {
-      if (nodesBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureNodesIsMutable();
-        nodes_.set(index, value);
-        onChanged();
-      } else {
-        nodesBuilder_.setMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject nodes = 1;</code>
-     */
-    public Builder setNodes(
-        int index, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.Builder builderForValue) {
-      if (nodesBuilder_ == null) {
-        ensureNodesIsMutable();
-        nodes_.set(index, builderForValue.build());
-        onChanged();
-      } else {
-        nodesBuilder_.setMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject nodes = 1;</code>
-     */
-    public Builder addNodes(org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject value) {
-      if (nodesBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureNodesIsMutable();
-        nodes_.add(value);
-        onChanged();
-      } else {
-        nodesBuilder_.addMessage(value);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject nodes = 1;</code>
-     */
-    public Builder addNodes(
-        int index, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject value) {
-      if (nodesBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureNodesIsMutable();
-        nodes_.add(index, value);
-        onChanged();
-      } else {
-        nodesBuilder_.addMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject nodes = 1;</code>
-     */
-    public Builder addNodes(
-        org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.Builder builderForValue) {
-      if (nodesBuilder_ == null) {
-        ensureNodesIsMutable();
-        nodes_.add(builderForValue.build());
-        onChanged();
-      } else {
-        nodesBuilder_.addMessage(builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject nodes = 1;</code>
-     */
-    public Builder addNodes(
-        int index, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.Builder builderForValue) {
-      if (nodesBuilder_ == null) {
-        ensureNodesIsMutable();
-        nodes_.add(index, builderForValue.build());
-        onChanged();
-      } else {
-        nodesBuilder_.addMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject nodes = 1;</code>
-     */
-    public Builder addAllNodes(
-        java.lang.Iterable<? extends org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject> values) {
-      if (nodesBuilder_ == null) {
-        ensureNodesIsMutable();
-        com.google.protobuf.AbstractMessageLite.Builder.addAll(
-            values, nodes_);
-        onChanged();
-      } else {
-        nodesBuilder_.addAllMessages(values);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject nodes = 1;</code>
-     */
-    public Builder clearNodes() {
-      if (nodesBuilder_ == null) {
-        nodes_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000001);
-        onChanged();
-      } else {
-        nodesBuilder_.clear();
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject nodes = 1;</code>
-     */
-    public Builder removeNodes(int index) {
-      if (nodesBuilder_ == null) {
-        ensureNodesIsMutable();
-        nodes_.remove(index);
-        onChanged();
-      } else {
-        nodesBuilder_.remove(index);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject nodes = 1;</code>
-     */
-    public org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.Builder getNodesBuilder(
-        int index) {
-      return getNodesFieldBuilder().getBuilder(index);
-    }
-    /**
-     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject nodes = 1;</code>
-     */
-    public org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObjectOrBuilder getNodesOrBuilder(
-        int index) {
-      if (nodesBuilder_ == null) {
-        return nodes_.get(index);  } else {
-        return nodesBuilder_.getMessageOrBuilder(index);
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject nodes = 1;</code>
-     */
-    public java.util.List<? extends org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObjectOrBuilder> 
-         getNodesOrBuilderList() {
-      if (nodesBuilder_ != null) {
-        return nodesBuilder_.getMessageOrBuilderList();
-      } else {
-        return java.util.Collections.unmodifiableList(nodes_);
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject nodes = 1;</code>
-     */
-    public org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.Builder addNodesBuilder() {
-      return getNodesFieldBuilder().addBuilder(
-          org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.getDefaultInstance());
-    }
-    /**
-     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject nodes = 1;</code>
-     */
-    public org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.Builder addNodesBuilder(
-        int index) {
-      return getNodesFieldBuilder().addBuilder(
-          index, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.getDefaultInstance());
-    }
-    /**
-     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject nodes = 1;</code>
-     */
-    public java.util.List<org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.Builder> 
-         getNodesBuilderList() {
-      return getNodesFieldBuilder().getBuilderList();
-    }
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.Builder, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObjectOrBuilder> 
-        getNodesFieldBuilder() {
-      if (nodesBuilder_ == null) {
-        nodesBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
-            org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.Builder, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObjectOrBuilder>(
-                nodes_,
-                ((bitField0_ & 0x00000001) != 0),
-                getParentForChildren(),
-                isClean());
-        nodes_ = null;
-      }
-      return nodesBuilder_;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.TrackableObjectGraph)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.TrackableObjectGraph)
-  private static final org.tensorflow.proto.framework.TrackableObjectGraph DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.TrackableObjectGraph();
-  }
-
-  public static org.tensorflow.proto.framework.TrackableObjectGraph getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<TrackableObjectGraph>
-      PARSER = new com.google.protobuf.AbstractParser<TrackableObjectGraph>() {
-    @java.lang.Override
-    public TrackableObjectGraph parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new TrackableObjectGraph(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<TrackableObjectGraph> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<TrackableObjectGraph> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.framework.TrackableObjectGraph getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TrackableObjectGraphOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TrackableObjectGraphOrBuilder.java
deleted file mode 100644
index b099d054d9a..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TrackableObjectGraphOrBuilder.java
+++ /dev/null
@@ -1,33 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/trackable_object_graph.proto
-
-package org.tensorflow.proto.framework;
-
-public interface TrackableObjectGraphOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.TrackableObjectGraph)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject nodes = 1;</code>
-   */
-  java.util.List<org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject> 
-      getNodesList();
-  /**
-   * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject nodes = 1;</code>
-   */
-  org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject getNodes(int index);
-  /**
-   * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject nodes = 1;</code>
-   */
-  int getNodesCount();
-  /**
-   * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject nodes = 1;</code>
-   */
-  java.util.List<? extends org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObjectOrBuilder> 
-      getNodesOrBuilderList();
-  /**
-   * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject nodes = 1;</code>
-   */
-  org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObjectOrBuilder getNodesOrBuilder(
-      int index);
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TrackableObjectGraphProtos.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TrackableObjectGraphProtos.java
deleted file mode 100644
index 22a3eef371a..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TrackableObjectGraphProtos.java
+++ /dev/null
@@ -1,111 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/trackable_object_graph.proto
-
-package org.tensorflow.proto.framework;
-
-public final class TrackableObjectGraphProtos {
-  private TrackableObjectGraphProtos() {}
-  public static void registerAllExtensions(
-      com.google.protobuf.ExtensionRegistryLite registry) {
-  }
-
-  public static void registerAllExtensions(
-      com.google.protobuf.ExtensionRegistry registry) {
-    registerAllExtensions(
-        (com.google.protobuf.ExtensionRegistryLite) registry);
-  }
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_TrackableObjectGraph_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_TrackableObjectGraph_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_TrackableObjectGraph_TrackableObject_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_TrackableObjectGraph_TrackableObject_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_TrackableObjectGraph_TrackableObject_ObjectReference_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_TrackableObjectGraph_TrackableObject_ObjectReference_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_TrackableObjectGraph_TrackableObject_SerializedTensor_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_TrackableObjectGraph_TrackableObject_SerializedTensor_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_TrackableObjectGraph_TrackableObject_SlotVariableReference_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_TrackableObjectGraph_TrackableObject_SlotVariableReference_fieldAccessorTable;
-
-  public static com.google.protobuf.Descriptors.FileDescriptor
-      getDescriptor() {
-    return descriptor;
-  }
-  private static  com.google.protobuf.Descriptors.FileDescriptor
-      descriptor;
-  static {
-    java.lang.String[] descriptorData = {
-      "\n5tensorflow/core/protobuf/trackable_obj" +
-      "ect_graph.proto\022\ntensorflow\"\203\005\n\024Trackabl" +
-      "eObjectGraph\022?\n\005nodes\030\001 \003(\01320.tensorflow" +
-      ".TrackableObjectGraph.TrackableObject\032\251\004" +
-      "\n\017TrackableObject\022R\n\010children\030\001 \003(\0132@.te" +
-      "nsorflow.TrackableObjectGraph.TrackableO" +
-      "bject.ObjectReference\022U\n\nattributes\030\002 \003(" +
-      "\0132A.tensorflow.TrackableObjectGraph.Trac" +
-      "kableObject.SerializedTensor\022^\n\016slot_var" +
-      "iables\030\003 \003(\0132F.tensorflow.TrackableObjec" +
-      "tGraph.TrackableObject.SlotVariableRefer" +
-      "ence\0326\n\017ObjectReference\022\017\n\007node_id\030\001 \001(\005" +
-      "\022\022\n\nlocal_name\030\002 \001(\t\032e\n\020SerializedTensor" +
-      "\022\014\n\004name\030\001 \001(\t\022\021\n\tfull_name\030\002 \001(\t\022\026\n\016che" +
-      "ckpoint_key\030\003 \001(\t\022\030\n\020optional_restore\030\004 " +
-      "\001(\010\032l\n\025SlotVariableReference\022!\n\031original" +
-      "_variable_node_id\030\001 \001(\005\022\021\n\tslot_name\030\002 \001" +
-      "(\t\022\035\n\025slot_variable_node_id\030\003 \001(\005B\213\001\n\036or" +
-      "g.tensorflow.proto.frameworkB\032TrackableO" +
-      "bjectGraphProtosP\001ZHgithub.com/tensorflo" +
-      "w/tensorflow/tensorflow/go/core/core_pro" +
-      "tos_go_proto\370\001\001b\006proto3"
-    };
-    descriptor = com.google.protobuf.Descriptors.FileDescriptor
-      .internalBuildGeneratedFileFrom(descriptorData,
-        new com.google.protobuf.Descriptors.FileDescriptor[] {
-        });
-    internal_static_tensorflow_TrackableObjectGraph_descriptor =
-      getDescriptor().getMessageTypes().get(0);
-    internal_static_tensorflow_TrackableObjectGraph_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_TrackableObjectGraph_descriptor,
-        new java.lang.String[] { "Nodes", });
-    internal_static_tensorflow_TrackableObjectGraph_TrackableObject_descriptor =
-      internal_static_tensorflow_TrackableObjectGraph_descriptor.getNestedTypes().get(0);
-    internal_static_tensorflow_TrackableObjectGraph_TrackableObject_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_TrackableObjectGraph_TrackableObject_descriptor,
-        new java.lang.String[] { "Children", "Attributes", "SlotVariables", });
-    internal_static_tensorflow_TrackableObjectGraph_TrackableObject_ObjectReference_descriptor =
-      internal_static_tensorflow_TrackableObjectGraph_TrackableObject_descriptor.getNestedTypes().get(0);
-    internal_static_tensorflow_TrackableObjectGraph_TrackableObject_ObjectReference_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_TrackableObjectGraph_TrackableObject_ObjectReference_descriptor,
-        new java.lang.String[] { "NodeId", "LocalName", });
-    internal_static_tensorflow_TrackableObjectGraph_TrackableObject_SerializedTensor_descriptor =
-      internal_static_tensorflow_TrackableObjectGraph_TrackableObject_descriptor.getNestedTypes().get(1);
-    internal_static_tensorflow_TrackableObjectGraph_TrackableObject_SerializedTensor_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_TrackableObjectGraph_TrackableObject_SerializedTensor_descriptor,
-        new java.lang.String[] { "Name", "FullName", "CheckpointKey", "OptionalRestore", });
-    internal_static_tensorflow_TrackableObjectGraph_TrackableObject_SlotVariableReference_descriptor =
-      internal_static_tensorflow_TrackableObjectGraph_TrackableObject_descriptor.getNestedTypes().get(2);
-    internal_static_tensorflow_TrackableObjectGraph_TrackableObject_SlotVariableReference_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_TrackableObjectGraph_TrackableObject_SlotVariableReference_descriptor,
-        new java.lang.String[] { "OriginalVariableNodeId", "SlotName", "SlotVariableNodeId", });
-  }
-
-  // @@protoc_insertion_point(outer_class_scope)
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TupleValue.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TupleValue.java
deleted file mode 100644
index 68f4c9d8d0b..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TupleValue.java
+++ /dev/null
@@ -1,773 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/struct.proto
-
-package org.tensorflow.proto.framework;
-
-/**
- * <pre>
- * Represents a Python tuple.
- * </pre>
- *
- * Protobuf type {@code tensorflow.TupleValue}
- */
-public  final class TupleValue extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.TupleValue)
-    TupleValueOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use TupleValue.newBuilder() to construct.
-  private TupleValue(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private TupleValue() {
-    values_ = java.util.Collections.emptyList();
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new TupleValue();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private TupleValue(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    int mutable_bitField0_ = 0;
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 10: {
-            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
-              values_ = new java.util.ArrayList<org.tensorflow.proto.framework.StructuredValue>();
-              mutable_bitField0_ |= 0x00000001;
-            }
-            values_.add(
-                input.readMessage(org.tensorflow.proto.framework.StructuredValue.parser(), extensionRegistry));
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      if (((mutable_bitField0_ & 0x00000001) != 0)) {
-        values_ = java.util.Collections.unmodifiableList(values_);
-      }
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.framework.StructProtos.internal_static_tensorflow_TupleValue_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.framework.StructProtos.internal_static_tensorflow_TupleValue_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.framework.TupleValue.class, org.tensorflow.proto.framework.TupleValue.Builder.class);
-  }
-
-  public static final int VALUES_FIELD_NUMBER = 1;
-  private java.util.List<org.tensorflow.proto.framework.StructuredValue> values_;
-  /**
-   * <code>repeated .tensorflow.StructuredValue values = 1;</code>
-   */
-  public java.util.List<org.tensorflow.proto.framework.StructuredValue> getValuesList() {
-    return values_;
-  }
-  /**
-   * <code>repeated .tensorflow.StructuredValue values = 1;</code>
-   */
-  public java.util.List<? extends org.tensorflow.proto.framework.StructuredValueOrBuilder> 
-      getValuesOrBuilderList() {
-    return values_;
-  }
-  /**
-   * <code>repeated .tensorflow.StructuredValue values = 1;</code>
-   */
-  public int getValuesCount() {
-    return values_.size();
-  }
-  /**
-   * <code>repeated .tensorflow.StructuredValue values = 1;</code>
-   */
-  public org.tensorflow.proto.framework.StructuredValue getValues(int index) {
-    return values_.get(index);
-  }
-  /**
-   * <code>repeated .tensorflow.StructuredValue values = 1;</code>
-   */
-  public org.tensorflow.proto.framework.StructuredValueOrBuilder getValuesOrBuilder(
-      int index) {
-    return values_.get(index);
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    for (int i = 0; i < values_.size(); i++) {
-      output.writeMessage(1, values_.get(i));
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    for (int i = 0; i < values_.size(); i++) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(1, values_.get(i));
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.framework.TupleValue)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.framework.TupleValue other = (org.tensorflow.proto.framework.TupleValue) obj;
-
-    if (!getValuesList()
-        .equals(other.getValuesList())) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    if (getValuesCount() > 0) {
-      hash = (37 * hash) + VALUES_FIELD_NUMBER;
-      hash = (53 * hash) + getValuesList().hashCode();
-    }
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.framework.TupleValue parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.TupleValue parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.TupleValue parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.TupleValue parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.TupleValue parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.TupleValue parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.TupleValue parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.TupleValue parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.TupleValue parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.TupleValue parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.TupleValue parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.TupleValue parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.framework.TupleValue prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * <pre>
-   * Represents a Python tuple.
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.TupleValue}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.TupleValue)
-      org.tensorflow.proto.framework.TupleValueOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.StructProtos.internal_static_tensorflow_TupleValue_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.StructProtos.internal_static_tensorflow_TupleValue_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.TupleValue.class, org.tensorflow.proto.framework.TupleValue.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.framework.TupleValue.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-        getValuesFieldBuilder();
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      if (valuesBuilder_ == null) {
-        values_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000001);
-      } else {
-        valuesBuilder_.clear();
-      }
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.framework.StructProtos.internal_static_tensorflow_TupleValue_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.TupleValue getDefaultInstanceForType() {
-      return org.tensorflow.proto.framework.TupleValue.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.TupleValue build() {
-      org.tensorflow.proto.framework.TupleValue result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.TupleValue buildPartial() {
-      org.tensorflow.proto.framework.TupleValue result = new org.tensorflow.proto.framework.TupleValue(this);
-      int from_bitField0_ = bitField0_;
-      if (valuesBuilder_ == null) {
-        if (((bitField0_ & 0x00000001) != 0)) {
-          values_ = java.util.Collections.unmodifiableList(values_);
-          bitField0_ = (bitField0_ & ~0x00000001);
-        }
-        result.values_ = values_;
-      } else {
-        result.values_ = valuesBuilder_.build();
-      }
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.framework.TupleValue) {
-        return mergeFrom((org.tensorflow.proto.framework.TupleValue)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.framework.TupleValue other) {
-      if (other == org.tensorflow.proto.framework.TupleValue.getDefaultInstance()) return this;
-      if (valuesBuilder_ == null) {
-        if (!other.values_.isEmpty()) {
-          if (values_.isEmpty()) {
-            values_ = other.values_;
-            bitField0_ = (bitField0_ & ~0x00000001);
-          } else {
-            ensureValuesIsMutable();
-            values_.addAll(other.values_);
-          }
-          onChanged();
-        }
-      } else {
-        if (!other.values_.isEmpty()) {
-          if (valuesBuilder_.isEmpty()) {
-            valuesBuilder_.dispose();
-            valuesBuilder_ = null;
-            values_ = other.values_;
-            bitField0_ = (bitField0_ & ~0x00000001);
-            valuesBuilder_ = 
-              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
-                 getValuesFieldBuilder() : null;
-          } else {
-            valuesBuilder_.addAllMessages(other.values_);
-          }
-        }
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.framework.TupleValue parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.framework.TupleValue) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-    private int bitField0_;
-
-    private java.util.List<org.tensorflow.proto.framework.StructuredValue> values_ =
-      java.util.Collections.emptyList();
-    private void ensureValuesIsMutable() {
-      if (!((bitField0_ & 0x00000001) != 0)) {
-        values_ = new java.util.ArrayList<org.tensorflow.proto.framework.StructuredValue>(values_);
-        bitField0_ |= 0x00000001;
-       }
-    }
-
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.framework.StructuredValue, org.tensorflow.proto.framework.StructuredValue.Builder, org.tensorflow.proto.framework.StructuredValueOrBuilder> valuesBuilder_;
-
-    /**
-     * <code>repeated .tensorflow.StructuredValue values = 1;</code>
-     */
-    public java.util.List<org.tensorflow.proto.framework.StructuredValue> getValuesList() {
-      if (valuesBuilder_ == null) {
-        return java.util.Collections.unmodifiableList(values_);
-      } else {
-        return valuesBuilder_.getMessageList();
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.StructuredValue values = 1;</code>
-     */
-    public int getValuesCount() {
-      if (valuesBuilder_ == null) {
-        return values_.size();
-      } else {
-        return valuesBuilder_.getCount();
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.StructuredValue values = 1;</code>
-     */
-    public org.tensorflow.proto.framework.StructuredValue getValues(int index) {
-      if (valuesBuilder_ == null) {
-        return values_.get(index);
-      } else {
-        return valuesBuilder_.getMessage(index);
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.StructuredValue values = 1;</code>
-     */
-    public Builder setValues(
-        int index, org.tensorflow.proto.framework.StructuredValue value) {
-      if (valuesBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureValuesIsMutable();
-        values_.set(index, value);
-        onChanged();
-      } else {
-        valuesBuilder_.setMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.StructuredValue values = 1;</code>
-     */
-    public Builder setValues(
-        int index, org.tensorflow.proto.framework.StructuredValue.Builder builderForValue) {
-      if (valuesBuilder_ == null) {
-        ensureValuesIsMutable();
-        values_.set(index, builderForValue.build());
-        onChanged();
-      } else {
-        valuesBuilder_.setMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.StructuredValue values = 1;</code>
-     */
-    public Builder addValues(org.tensorflow.proto.framework.StructuredValue value) {
-      if (valuesBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureValuesIsMutable();
-        values_.add(value);
-        onChanged();
-      } else {
-        valuesBuilder_.addMessage(value);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.StructuredValue values = 1;</code>
-     */
-    public Builder addValues(
-        int index, org.tensorflow.proto.framework.StructuredValue value) {
-      if (valuesBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureValuesIsMutable();
-        values_.add(index, value);
-        onChanged();
-      } else {
-        valuesBuilder_.addMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.StructuredValue values = 1;</code>
-     */
-    public Builder addValues(
-        org.tensorflow.proto.framework.StructuredValue.Builder builderForValue) {
-      if (valuesBuilder_ == null) {
-        ensureValuesIsMutable();
-        values_.add(builderForValue.build());
-        onChanged();
-      } else {
-        valuesBuilder_.addMessage(builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.StructuredValue values = 1;</code>
-     */
-    public Builder addValues(
-        int index, org.tensorflow.proto.framework.StructuredValue.Builder builderForValue) {
-      if (valuesBuilder_ == null) {
-        ensureValuesIsMutable();
-        values_.add(index, builderForValue.build());
-        onChanged();
-      } else {
-        valuesBuilder_.addMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.StructuredValue values = 1;</code>
-     */
-    public Builder addAllValues(
-        java.lang.Iterable<? extends org.tensorflow.proto.framework.StructuredValue> values) {
-      if (valuesBuilder_ == null) {
-        ensureValuesIsMutable();
-        com.google.protobuf.AbstractMessageLite.Builder.addAll(
-            values, values_);
-        onChanged();
-      } else {
-        valuesBuilder_.addAllMessages(values);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.StructuredValue values = 1;</code>
-     */
-    public Builder clearValues() {
-      if (valuesBuilder_ == null) {
-        values_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000001);
-        onChanged();
-      } else {
-        valuesBuilder_.clear();
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.StructuredValue values = 1;</code>
-     */
-    public Builder removeValues(int index) {
-      if (valuesBuilder_ == null) {
-        ensureValuesIsMutable();
-        values_.remove(index);
-        onChanged();
-      } else {
-        valuesBuilder_.remove(index);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.StructuredValue values = 1;</code>
-     */
-    public org.tensorflow.proto.framework.StructuredValue.Builder getValuesBuilder(
-        int index) {
-      return getValuesFieldBuilder().getBuilder(index);
-    }
-    /**
-     * <code>repeated .tensorflow.StructuredValue values = 1;</code>
-     */
-    public org.tensorflow.proto.framework.StructuredValueOrBuilder getValuesOrBuilder(
-        int index) {
-      if (valuesBuilder_ == null) {
-        return values_.get(index);  } else {
-        return valuesBuilder_.getMessageOrBuilder(index);
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.StructuredValue values = 1;</code>
-     */
-    public java.util.List<? extends org.tensorflow.proto.framework.StructuredValueOrBuilder> 
-         getValuesOrBuilderList() {
-      if (valuesBuilder_ != null) {
-        return valuesBuilder_.getMessageOrBuilderList();
-      } else {
-        return java.util.Collections.unmodifiableList(values_);
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.StructuredValue values = 1;</code>
-     */
-    public org.tensorflow.proto.framework.StructuredValue.Builder addValuesBuilder() {
-      return getValuesFieldBuilder().addBuilder(
-          org.tensorflow.proto.framework.StructuredValue.getDefaultInstance());
-    }
-    /**
-     * <code>repeated .tensorflow.StructuredValue values = 1;</code>
-     */
-    public org.tensorflow.proto.framework.StructuredValue.Builder addValuesBuilder(
-        int index) {
-      return getValuesFieldBuilder().addBuilder(
-          index, org.tensorflow.proto.framework.StructuredValue.getDefaultInstance());
-    }
-    /**
-     * <code>repeated .tensorflow.StructuredValue values = 1;</code>
-     */
-    public java.util.List<org.tensorflow.proto.framework.StructuredValue.Builder> 
-         getValuesBuilderList() {
-      return getValuesFieldBuilder().getBuilderList();
-    }
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.framework.StructuredValue, org.tensorflow.proto.framework.StructuredValue.Builder, org.tensorflow.proto.framework.StructuredValueOrBuilder> 
-        getValuesFieldBuilder() {
-      if (valuesBuilder_ == null) {
-        valuesBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
-            org.tensorflow.proto.framework.StructuredValue, org.tensorflow.proto.framework.StructuredValue.Builder, org.tensorflow.proto.framework.StructuredValueOrBuilder>(
-                values_,
-                ((bitField0_ & 0x00000001) != 0),
-                getParentForChildren(),
-                isClean());
-        values_ = null;
-      }
-      return valuesBuilder_;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.TupleValue)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.TupleValue)
-  private static final org.tensorflow.proto.framework.TupleValue DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.TupleValue();
-  }
-
-  public static org.tensorflow.proto.framework.TupleValue getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<TupleValue>
-      PARSER = new com.google.protobuf.AbstractParser<TupleValue>() {
-    @java.lang.Override
-    public TupleValue parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new TupleValue(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<TupleValue> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<TupleValue> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.framework.TupleValue getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TupleValueOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TupleValueOrBuilder.java
deleted file mode 100644
index aea74c1d73b..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TupleValueOrBuilder.java
+++ /dev/null
@@ -1,33 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/struct.proto
-
-package org.tensorflow.proto.framework;
-
-public interface TupleValueOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.TupleValue)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <code>repeated .tensorflow.StructuredValue values = 1;</code>
-   */
-  java.util.List<org.tensorflow.proto.framework.StructuredValue> 
-      getValuesList();
-  /**
-   * <code>repeated .tensorflow.StructuredValue values = 1;</code>
-   */
-  org.tensorflow.proto.framework.StructuredValue getValues(int index);
-  /**
-   * <code>repeated .tensorflow.StructuredValue values = 1;</code>
-   */
-  int getValuesCount();
-  /**
-   * <code>repeated .tensorflow.StructuredValue values = 1;</code>
-   */
-  java.util.List<? extends org.tensorflow.proto.framework.StructuredValueOrBuilder> 
-      getValuesOrBuilderList();
-  /**
-   * <code>repeated .tensorflow.StructuredValue values = 1;</code>
-   */
-  org.tensorflow.proto.framework.StructuredValueOrBuilder getValuesOrBuilder(
-      int index);
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TypeSpecProto.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TypeSpecProto.java
deleted file mode 100644
index 10c64eeb41c..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TypeSpecProto.java
+++ /dev/null
@@ -1,1176 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/struct.proto
-
-package org.tensorflow.proto.framework;
-
-/**
- * <pre>
- * Represents a tf.TypeSpec
- * </pre>
- *
- * Protobuf type {@code tensorflow.TypeSpecProto}
- */
-public  final class TypeSpecProto extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.TypeSpecProto)
-    TypeSpecProtoOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use TypeSpecProto.newBuilder() to construct.
-  private TypeSpecProto(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private TypeSpecProto() {
-    typeSpecClass_ = 0;
-    typeSpecClassName_ = "";
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new TypeSpecProto();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private TypeSpecProto(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 8: {
-            int rawValue = input.readEnum();
-
-            typeSpecClass_ = rawValue;
-            break;
-          }
-          case 18: {
-            org.tensorflow.proto.framework.StructuredValue.Builder subBuilder = null;
-            if (typeState_ != null) {
-              subBuilder = typeState_.toBuilder();
-            }
-            typeState_ = input.readMessage(org.tensorflow.proto.framework.StructuredValue.parser(), extensionRegistry);
-            if (subBuilder != null) {
-              subBuilder.mergeFrom(typeState_);
-              typeState_ = subBuilder.buildPartial();
-            }
-
-            break;
-          }
-          case 26: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            typeSpecClassName_ = s;
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.framework.StructProtos.internal_static_tensorflow_TypeSpecProto_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.framework.StructProtos.internal_static_tensorflow_TypeSpecProto_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.framework.TypeSpecProto.class, org.tensorflow.proto.framework.TypeSpecProto.Builder.class);
-  }
-
-  /**
-   * Protobuf enum {@code tensorflow.TypeSpecProto.TypeSpecClass}
-   */
-  public enum TypeSpecClass
-      implements com.google.protobuf.ProtocolMessageEnum {
-    /**
-     * <code>UNKNOWN = 0;</code>
-     */
-    UNKNOWN(0),
-    /**
-     * <pre>
-     * tf.SparseTensorSpec
-     * </pre>
-     *
-     * <code>SPARSE_TENSOR_SPEC = 1;</code>
-     */
-    SPARSE_TENSOR_SPEC(1),
-    /**
-     * <pre>
-     * tf.IndexedSlicesSpec
-     * </pre>
-     *
-     * <code>INDEXED_SLICES_SPEC = 2;</code>
-     */
-    INDEXED_SLICES_SPEC(2),
-    /**
-     * <pre>
-     * tf.RaggedTensorSpec
-     * </pre>
-     *
-     * <code>RAGGED_TENSOR_SPEC = 3;</code>
-     */
-    RAGGED_TENSOR_SPEC(3),
-    /**
-     * <pre>
-     * tf.TensorArraySpec
-     * </pre>
-     *
-     * <code>TENSOR_ARRAY_SPEC = 4;</code>
-     */
-    TENSOR_ARRAY_SPEC(4),
-    /**
-     * <pre>
-     * tf.data.DatasetSpec
-     * </pre>
-     *
-     * <code>DATA_DATASET_SPEC = 5;</code>
-     */
-    DATA_DATASET_SPEC(5),
-    /**
-     * <pre>
-     * IteratorSpec from data/ops/iterator_ops.py
-     * </pre>
-     *
-     * <code>DATA_ITERATOR_SPEC = 6;</code>
-     */
-    DATA_ITERATOR_SPEC(6),
-    /**
-     * <pre>
-     * tf.OptionalSpec
-     * </pre>
-     *
-     * <code>OPTIONAL_SPEC = 7;</code>
-     */
-    OPTIONAL_SPEC(7),
-    /**
-     * <pre>
-     * PerReplicaSpec from distribute/values.py
-     * </pre>
-     *
-     * <code>PER_REPLICA_SPEC = 8;</code>
-     */
-    PER_REPLICA_SPEC(8),
-    /**
-     * <pre>
-     * tf.VariableSpec
-     * </pre>
-     *
-     * <code>VARIABLE_SPEC = 9;</code>
-     */
-    VARIABLE_SPEC(9),
-    /**
-     * <pre>
-     * RowPartitionSpec from ragged/row_partition.py
-     * </pre>
-     *
-     * <code>ROW_PARTITION_SPEC = 10;</code>
-     */
-    ROW_PARTITION_SPEC(10),
-    UNRECOGNIZED(-1),
-    ;
-
-    /**
-     * <code>UNKNOWN = 0;</code>
-     */
-    public static final int UNKNOWN_VALUE = 0;
-    /**
-     * <pre>
-     * tf.SparseTensorSpec
-     * </pre>
-     *
-     * <code>SPARSE_TENSOR_SPEC = 1;</code>
-     */
-    public static final int SPARSE_TENSOR_SPEC_VALUE = 1;
-    /**
-     * <pre>
-     * tf.IndexedSlicesSpec
-     * </pre>
-     *
-     * <code>INDEXED_SLICES_SPEC = 2;</code>
-     */
-    public static final int INDEXED_SLICES_SPEC_VALUE = 2;
-    /**
-     * <pre>
-     * tf.RaggedTensorSpec
-     * </pre>
-     *
-     * <code>RAGGED_TENSOR_SPEC = 3;</code>
-     */
-    public static final int RAGGED_TENSOR_SPEC_VALUE = 3;
-    /**
-     * <pre>
-     * tf.TensorArraySpec
-     * </pre>
-     *
-     * <code>TENSOR_ARRAY_SPEC = 4;</code>
-     */
-    public static final int TENSOR_ARRAY_SPEC_VALUE = 4;
-    /**
-     * <pre>
-     * tf.data.DatasetSpec
-     * </pre>
-     *
-     * <code>DATA_DATASET_SPEC = 5;</code>
-     */
-    public static final int DATA_DATASET_SPEC_VALUE = 5;
-    /**
-     * <pre>
-     * IteratorSpec from data/ops/iterator_ops.py
-     * </pre>
-     *
-     * <code>DATA_ITERATOR_SPEC = 6;</code>
-     */
-    public static final int DATA_ITERATOR_SPEC_VALUE = 6;
-    /**
-     * <pre>
-     * tf.OptionalSpec
-     * </pre>
-     *
-     * <code>OPTIONAL_SPEC = 7;</code>
-     */
-    public static final int OPTIONAL_SPEC_VALUE = 7;
-    /**
-     * <pre>
-     * PerReplicaSpec from distribute/values.py
-     * </pre>
-     *
-     * <code>PER_REPLICA_SPEC = 8;</code>
-     */
-    public static final int PER_REPLICA_SPEC_VALUE = 8;
-    /**
-     * <pre>
-     * tf.VariableSpec
-     * </pre>
-     *
-     * <code>VARIABLE_SPEC = 9;</code>
-     */
-    public static final int VARIABLE_SPEC_VALUE = 9;
-    /**
-     * <pre>
-     * RowPartitionSpec from ragged/row_partition.py
-     * </pre>
-     *
-     * <code>ROW_PARTITION_SPEC = 10;</code>
-     */
-    public static final int ROW_PARTITION_SPEC_VALUE = 10;
-
-
-    public final int getNumber() {
-      if (this == UNRECOGNIZED) {
-        throw new java.lang.IllegalArgumentException(
-            "Can't get the number of an unknown enum value.");
-      }
-      return value;
-    }
-
-    /**
-     * @deprecated Use {@link #forNumber(int)} instead.
-     */
-    @java.lang.Deprecated
-    public static TypeSpecClass valueOf(int value) {
-      return forNumber(value);
-    }
-
-    public static TypeSpecClass forNumber(int value) {
-      switch (value) {
-        case 0: return UNKNOWN;
-        case 1: return SPARSE_TENSOR_SPEC;
-        case 2: return INDEXED_SLICES_SPEC;
-        case 3: return RAGGED_TENSOR_SPEC;
-        case 4: return TENSOR_ARRAY_SPEC;
-        case 5: return DATA_DATASET_SPEC;
-        case 6: return DATA_ITERATOR_SPEC;
-        case 7: return OPTIONAL_SPEC;
-        case 8: return PER_REPLICA_SPEC;
-        case 9: return VARIABLE_SPEC;
-        case 10: return ROW_PARTITION_SPEC;
-        default: return null;
-      }
-    }
-
-    public static com.google.protobuf.Internal.EnumLiteMap<TypeSpecClass>
-        internalGetValueMap() {
-      return internalValueMap;
-    }
-    private static final com.google.protobuf.Internal.EnumLiteMap<
-        TypeSpecClass> internalValueMap =
-          new com.google.protobuf.Internal.EnumLiteMap<TypeSpecClass>() {
-            public TypeSpecClass findValueByNumber(int number) {
-              return TypeSpecClass.forNumber(number);
-            }
-          };
-
-    public final com.google.protobuf.Descriptors.EnumValueDescriptor
-        getValueDescriptor() {
-      return getDescriptor().getValues().get(ordinal());
-    }
-    public final com.google.protobuf.Descriptors.EnumDescriptor
-        getDescriptorForType() {
-      return getDescriptor();
-    }
-    public static final com.google.protobuf.Descriptors.EnumDescriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.TypeSpecProto.getDescriptor().getEnumTypes().get(0);
-    }
-
-    private static final TypeSpecClass[] VALUES = values();
-
-    public static TypeSpecClass valueOf(
-        com.google.protobuf.Descriptors.EnumValueDescriptor desc) {
-      if (desc.getType() != getDescriptor()) {
-        throw new java.lang.IllegalArgumentException(
-          "EnumValueDescriptor is not for this type.");
-      }
-      if (desc.getIndex() == -1) {
-        return UNRECOGNIZED;
-      }
-      return VALUES[desc.getIndex()];
-    }
-
-    private final int value;
-
-    private TypeSpecClass(int value) {
-      this.value = value;
-    }
-
-    // @@protoc_insertion_point(enum_scope:tensorflow.TypeSpecProto.TypeSpecClass)
-  }
-
-  public static final int TYPE_SPEC_CLASS_FIELD_NUMBER = 1;
-  private int typeSpecClass_;
-  /**
-   * <code>.tensorflow.TypeSpecProto.TypeSpecClass type_spec_class = 1;</code>
-   */
-  public int getTypeSpecClassValue() {
-    return typeSpecClass_;
-  }
-  /**
-   * <code>.tensorflow.TypeSpecProto.TypeSpecClass type_spec_class = 1;</code>
-   */
-  public org.tensorflow.proto.framework.TypeSpecProto.TypeSpecClass getTypeSpecClass() {
-    @SuppressWarnings("deprecation")
-    org.tensorflow.proto.framework.TypeSpecProto.TypeSpecClass result = org.tensorflow.proto.framework.TypeSpecProto.TypeSpecClass.valueOf(typeSpecClass_);
-    return result == null ? org.tensorflow.proto.framework.TypeSpecProto.TypeSpecClass.UNRECOGNIZED : result;
-  }
-
-  public static final int TYPE_STATE_FIELD_NUMBER = 2;
-  private org.tensorflow.proto.framework.StructuredValue typeState_;
-  /**
-   * <pre>
-   * The value returned by TypeSpec._serialize().
-   * </pre>
-   *
-   * <code>.tensorflow.StructuredValue type_state = 2;</code>
-   */
-  public boolean hasTypeState() {
-    return typeState_ != null;
-  }
-  /**
-   * <pre>
-   * The value returned by TypeSpec._serialize().
-   * </pre>
-   *
-   * <code>.tensorflow.StructuredValue type_state = 2;</code>
-   */
-  public org.tensorflow.proto.framework.StructuredValue getTypeState() {
-    return typeState_ == null ? org.tensorflow.proto.framework.StructuredValue.getDefaultInstance() : typeState_;
-  }
-  /**
-   * <pre>
-   * The value returned by TypeSpec._serialize().
-   * </pre>
-   *
-   * <code>.tensorflow.StructuredValue type_state = 2;</code>
-   */
-  public org.tensorflow.proto.framework.StructuredValueOrBuilder getTypeStateOrBuilder() {
-    return getTypeState();
-  }
-
-  public static final int TYPE_SPEC_CLASS_NAME_FIELD_NUMBER = 3;
-  private volatile java.lang.Object typeSpecClassName_;
-  /**
-   * <pre>
-   * This is currently redundant with the type_spec_class enum, and is only
-   * used for error reporting.  In particular, if you use an older binary to
-   * load a newer model, and the model uses a TypeSpecClass that the older
-   * binary doesn't support, then this lets us display a useful error message.
-   * </pre>
-   *
-   * <code>string type_spec_class_name = 3;</code>
-   */
-  public java.lang.String getTypeSpecClassName() {
-    java.lang.Object ref = typeSpecClassName_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      typeSpecClassName_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * This is currently redundant with the type_spec_class enum, and is only
-   * used for error reporting.  In particular, if you use an older binary to
-   * load a newer model, and the model uses a TypeSpecClass that the older
-   * binary doesn't support, then this lets us display a useful error message.
-   * </pre>
-   *
-   * <code>string type_spec_class_name = 3;</code>
-   */
-  public com.google.protobuf.ByteString
-      getTypeSpecClassNameBytes() {
-    java.lang.Object ref = typeSpecClassName_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      typeSpecClassName_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    if (typeSpecClass_ != org.tensorflow.proto.framework.TypeSpecProto.TypeSpecClass.UNKNOWN.getNumber()) {
-      output.writeEnum(1, typeSpecClass_);
-    }
-    if (typeState_ != null) {
-      output.writeMessage(2, getTypeState());
-    }
-    if (!getTypeSpecClassNameBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 3, typeSpecClassName_);
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (typeSpecClass_ != org.tensorflow.proto.framework.TypeSpecProto.TypeSpecClass.UNKNOWN.getNumber()) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeEnumSize(1, typeSpecClass_);
-    }
-    if (typeState_ != null) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(2, getTypeState());
-    }
-    if (!getTypeSpecClassNameBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(3, typeSpecClassName_);
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.framework.TypeSpecProto)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.framework.TypeSpecProto other = (org.tensorflow.proto.framework.TypeSpecProto) obj;
-
-    if (typeSpecClass_ != other.typeSpecClass_) return false;
-    if (hasTypeState() != other.hasTypeState()) return false;
-    if (hasTypeState()) {
-      if (!getTypeState()
-          .equals(other.getTypeState())) return false;
-    }
-    if (!getTypeSpecClassName()
-        .equals(other.getTypeSpecClassName())) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    hash = (37 * hash) + TYPE_SPEC_CLASS_FIELD_NUMBER;
-    hash = (53 * hash) + typeSpecClass_;
-    if (hasTypeState()) {
-      hash = (37 * hash) + TYPE_STATE_FIELD_NUMBER;
-      hash = (53 * hash) + getTypeState().hashCode();
-    }
-    hash = (37 * hash) + TYPE_SPEC_CLASS_NAME_FIELD_NUMBER;
-    hash = (53 * hash) + getTypeSpecClassName().hashCode();
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.framework.TypeSpecProto parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.TypeSpecProto parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.TypeSpecProto parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.TypeSpecProto parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.TypeSpecProto parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.TypeSpecProto parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.TypeSpecProto parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.TypeSpecProto parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.TypeSpecProto parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.TypeSpecProto parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.TypeSpecProto parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.TypeSpecProto parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.framework.TypeSpecProto prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * <pre>
-   * Represents a tf.TypeSpec
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.TypeSpecProto}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.TypeSpecProto)
-      org.tensorflow.proto.framework.TypeSpecProtoOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.StructProtos.internal_static_tensorflow_TypeSpecProto_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.StructProtos.internal_static_tensorflow_TypeSpecProto_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.TypeSpecProto.class, org.tensorflow.proto.framework.TypeSpecProto.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.framework.TypeSpecProto.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      typeSpecClass_ = 0;
-
-      if (typeStateBuilder_ == null) {
-        typeState_ = null;
-      } else {
-        typeState_ = null;
-        typeStateBuilder_ = null;
-      }
-      typeSpecClassName_ = "";
-
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.framework.StructProtos.internal_static_tensorflow_TypeSpecProto_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.TypeSpecProto getDefaultInstanceForType() {
-      return org.tensorflow.proto.framework.TypeSpecProto.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.TypeSpecProto build() {
-      org.tensorflow.proto.framework.TypeSpecProto result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.TypeSpecProto buildPartial() {
-      org.tensorflow.proto.framework.TypeSpecProto result = new org.tensorflow.proto.framework.TypeSpecProto(this);
-      result.typeSpecClass_ = typeSpecClass_;
-      if (typeStateBuilder_ == null) {
-        result.typeState_ = typeState_;
-      } else {
-        result.typeState_ = typeStateBuilder_.build();
-      }
-      result.typeSpecClassName_ = typeSpecClassName_;
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.framework.TypeSpecProto) {
-        return mergeFrom((org.tensorflow.proto.framework.TypeSpecProto)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.framework.TypeSpecProto other) {
-      if (other == org.tensorflow.proto.framework.TypeSpecProto.getDefaultInstance()) return this;
-      if (other.typeSpecClass_ != 0) {
-        setTypeSpecClassValue(other.getTypeSpecClassValue());
-      }
-      if (other.hasTypeState()) {
-        mergeTypeState(other.getTypeState());
-      }
-      if (!other.getTypeSpecClassName().isEmpty()) {
-        typeSpecClassName_ = other.typeSpecClassName_;
-        onChanged();
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.framework.TypeSpecProto parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.framework.TypeSpecProto) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-
-    private int typeSpecClass_ = 0;
-    /**
-     * <code>.tensorflow.TypeSpecProto.TypeSpecClass type_spec_class = 1;</code>
-     */
-    public int getTypeSpecClassValue() {
-      return typeSpecClass_;
-    }
-    /**
-     * <code>.tensorflow.TypeSpecProto.TypeSpecClass type_spec_class = 1;</code>
-     */
-    public Builder setTypeSpecClassValue(int value) {
-      typeSpecClass_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>.tensorflow.TypeSpecProto.TypeSpecClass type_spec_class = 1;</code>
-     */
-    public org.tensorflow.proto.framework.TypeSpecProto.TypeSpecClass getTypeSpecClass() {
-      @SuppressWarnings("deprecation")
-      org.tensorflow.proto.framework.TypeSpecProto.TypeSpecClass result = org.tensorflow.proto.framework.TypeSpecProto.TypeSpecClass.valueOf(typeSpecClass_);
-      return result == null ? org.tensorflow.proto.framework.TypeSpecProto.TypeSpecClass.UNRECOGNIZED : result;
-    }
-    /**
-     * <code>.tensorflow.TypeSpecProto.TypeSpecClass type_spec_class = 1;</code>
-     */
-    public Builder setTypeSpecClass(org.tensorflow.proto.framework.TypeSpecProto.TypeSpecClass value) {
-      if (value == null) {
-        throw new NullPointerException();
-      }
-      
-      typeSpecClass_ = value.getNumber();
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>.tensorflow.TypeSpecProto.TypeSpecClass type_spec_class = 1;</code>
-     */
-    public Builder clearTypeSpecClass() {
-      
-      typeSpecClass_ = 0;
-      onChanged();
-      return this;
-    }
-
-    private org.tensorflow.proto.framework.StructuredValue typeState_;
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.StructuredValue, org.tensorflow.proto.framework.StructuredValue.Builder, org.tensorflow.proto.framework.StructuredValueOrBuilder> typeStateBuilder_;
-    /**
-     * <pre>
-     * The value returned by TypeSpec._serialize().
-     * </pre>
-     *
-     * <code>.tensorflow.StructuredValue type_state = 2;</code>
-     */
-    public boolean hasTypeState() {
-      return typeStateBuilder_ != null || typeState_ != null;
-    }
-    /**
-     * <pre>
-     * The value returned by TypeSpec._serialize().
-     * </pre>
-     *
-     * <code>.tensorflow.StructuredValue type_state = 2;</code>
-     */
-    public org.tensorflow.proto.framework.StructuredValue getTypeState() {
-      if (typeStateBuilder_ == null) {
-        return typeState_ == null ? org.tensorflow.proto.framework.StructuredValue.getDefaultInstance() : typeState_;
-      } else {
-        return typeStateBuilder_.getMessage();
-      }
-    }
-    /**
-     * <pre>
-     * The value returned by TypeSpec._serialize().
-     * </pre>
-     *
-     * <code>.tensorflow.StructuredValue type_state = 2;</code>
-     */
-    public Builder setTypeState(org.tensorflow.proto.framework.StructuredValue value) {
-      if (typeStateBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        typeState_ = value;
-        onChanged();
-      } else {
-        typeStateBuilder_.setMessage(value);
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * The value returned by TypeSpec._serialize().
-     * </pre>
-     *
-     * <code>.tensorflow.StructuredValue type_state = 2;</code>
-     */
-    public Builder setTypeState(
-        org.tensorflow.proto.framework.StructuredValue.Builder builderForValue) {
-      if (typeStateBuilder_ == null) {
-        typeState_ = builderForValue.build();
-        onChanged();
-      } else {
-        typeStateBuilder_.setMessage(builderForValue.build());
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * The value returned by TypeSpec._serialize().
-     * </pre>
-     *
-     * <code>.tensorflow.StructuredValue type_state = 2;</code>
-     */
-    public Builder mergeTypeState(org.tensorflow.proto.framework.StructuredValue value) {
-      if (typeStateBuilder_ == null) {
-        if (typeState_ != null) {
-          typeState_ =
-            org.tensorflow.proto.framework.StructuredValue.newBuilder(typeState_).mergeFrom(value).buildPartial();
-        } else {
-          typeState_ = value;
-        }
-        onChanged();
-      } else {
-        typeStateBuilder_.mergeFrom(value);
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * The value returned by TypeSpec._serialize().
-     * </pre>
-     *
-     * <code>.tensorflow.StructuredValue type_state = 2;</code>
-     */
-    public Builder clearTypeState() {
-      if (typeStateBuilder_ == null) {
-        typeState_ = null;
-        onChanged();
-      } else {
-        typeState_ = null;
-        typeStateBuilder_ = null;
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * The value returned by TypeSpec._serialize().
-     * </pre>
-     *
-     * <code>.tensorflow.StructuredValue type_state = 2;</code>
-     */
-    public org.tensorflow.proto.framework.StructuredValue.Builder getTypeStateBuilder() {
-      
-      onChanged();
-      return getTypeStateFieldBuilder().getBuilder();
-    }
-    /**
-     * <pre>
-     * The value returned by TypeSpec._serialize().
-     * </pre>
-     *
-     * <code>.tensorflow.StructuredValue type_state = 2;</code>
-     */
-    public org.tensorflow.proto.framework.StructuredValueOrBuilder getTypeStateOrBuilder() {
-      if (typeStateBuilder_ != null) {
-        return typeStateBuilder_.getMessageOrBuilder();
-      } else {
-        return typeState_ == null ?
-            org.tensorflow.proto.framework.StructuredValue.getDefaultInstance() : typeState_;
-      }
-    }
-    /**
-     * <pre>
-     * The value returned by TypeSpec._serialize().
-     * </pre>
-     *
-     * <code>.tensorflow.StructuredValue type_state = 2;</code>
-     */
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.StructuredValue, org.tensorflow.proto.framework.StructuredValue.Builder, org.tensorflow.proto.framework.StructuredValueOrBuilder> 
-        getTypeStateFieldBuilder() {
-      if (typeStateBuilder_ == null) {
-        typeStateBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-            org.tensorflow.proto.framework.StructuredValue, org.tensorflow.proto.framework.StructuredValue.Builder, org.tensorflow.proto.framework.StructuredValueOrBuilder>(
-                getTypeState(),
-                getParentForChildren(),
-                isClean());
-        typeState_ = null;
-      }
-      return typeStateBuilder_;
-    }
-
-    private java.lang.Object typeSpecClassName_ = "";
-    /**
-     * <pre>
-     * This is currently redundant with the type_spec_class enum, and is only
-     * used for error reporting.  In particular, if you use an older binary to
-     * load a newer model, and the model uses a TypeSpecClass that the older
-     * binary doesn't support, then this lets us display a useful error message.
-     * </pre>
-     *
-     * <code>string type_spec_class_name = 3;</code>
-     */
-    public java.lang.String getTypeSpecClassName() {
-      java.lang.Object ref = typeSpecClassName_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        typeSpecClassName_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * This is currently redundant with the type_spec_class enum, and is only
-     * used for error reporting.  In particular, if you use an older binary to
-     * load a newer model, and the model uses a TypeSpecClass that the older
-     * binary doesn't support, then this lets us display a useful error message.
-     * </pre>
-     *
-     * <code>string type_spec_class_name = 3;</code>
-     */
-    public com.google.protobuf.ByteString
-        getTypeSpecClassNameBytes() {
-      java.lang.Object ref = typeSpecClassName_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        typeSpecClassName_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * This is currently redundant with the type_spec_class enum, and is only
-     * used for error reporting.  In particular, if you use an older binary to
-     * load a newer model, and the model uses a TypeSpecClass that the older
-     * binary doesn't support, then this lets us display a useful error message.
-     * </pre>
-     *
-     * <code>string type_spec_class_name = 3;</code>
-     */
-    public Builder setTypeSpecClassName(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      typeSpecClassName_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * This is currently redundant with the type_spec_class enum, and is only
-     * used for error reporting.  In particular, if you use an older binary to
-     * load a newer model, and the model uses a TypeSpecClass that the older
-     * binary doesn't support, then this lets us display a useful error message.
-     * </pre>
-     *
-     * <code>string type_spec_class_name = 3;</code>
-     */
-    public Builder clearTypeSpecClassName() {
-      
-      typeSpecClassName_ = getDefaultInstance().getTypeSpecClassName();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * This is currently redundant with the type_spec_class enum, and is only
-     * used for error reporting.  In particular, if you use an older binary to
-     * load a newer model, and the model uses a TypeSpecClass that the older
-     * binary doesn't support, then this lets us display a useful error message.
-     * </pre>
-     *
-     * <code>string type_spec_class_name = 3;</code>
-     */
-    public Builder setTypeSpecClassNameBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      typeSpecClassName_ = value;
-      onChanged();
-      return this;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.TypeSpecProto)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.TypeSpecProto)
-  private static final org.tensorflow.proto.framework.TypeSpecProto DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.TypeSpecProto();
-  }
-
-  public static org.tensorflow.proto.framework.TypeSpecProto getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<TypeSpecProto>
-      PARSER = new com.google.protobuf.AbstractParser<TypeSpecProto>() {
-    @java.lang.Override
-    public TypeSpecProto parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new TypeSpecProto(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<TypeSpecProto> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<TypeSpecProto> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.framework.TypeSpecProto getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TypeSpecProtoOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TypeSpecProtoOrBuilder.java
deleted file mode 100644
index db9ce3b7ab8..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TypeSpecProtoOrBuilder.java
+++ /dev/null
@@ -1,67 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/struct.proto
-
-package org.tensorflow.proto.framework;
-
-public interface TypeSpecProtoOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.TypeSpecProto)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <code>.tensorflow.TypeSpecProto.TypeSpecClass type_spec_class = 1;</code>
-   */
-  int getTypeSpecClassValue();
-  /**
-   * <code>.tensorflow.TypeSpecProto.TypeSpecClass type_spec_class = 1;</code>
-   */
-  org.tensorflow.proto.framework.TypeSpecProto.TypeSpecClass getTypeSpecClass();
-
-  /**
-   * <pre>
-   * The value returned by TypeSpec._serialize().
-   * </pre>
-   *
-   * <code>.tensorflow.StructuredValue type_state = 2;</code>
-   */
-  boolean hasTypeState();
-  /**
-   * <pre>
-   * The value returned by TypeSpec._serialize().
-   * </pre>
-   *
-   * <code>.tensorflow.StructuredValue type_state = 2;</code>
-   */
-  org.tensorflow.proto.framework.StructuredValue getTypeState();
-  /**
-   * <pre>
-   * The value returned by TypeSpec._serialize().
-   * </pre>
-   *
-   * <code>.tensorflow.StructuredValue type_state = 2;</code>
-   */
-  org.tensorflow.proto.framework.StructuredValueOrBuilder getTypeStateOrBuilder();
-
-  /**
-   * <pre>
-   * This is currently redundant with the type_spec_class enum, and is only
-   * used for error reporting.  In particular, if you use an older binary to
-   * load a newer model, and the model uses a TypeSpecClass that the older
-   * binary doesn't support, then this lets us display a useful error message.
-   * </pre>
-   *
-   * <code>string type_spec_class_name = 3;</code>
-   */
-  java.lang.String getTypeSpecClassName();
-  /**
-   * <pre>
-   * This is currently redundant with the type_spec_class enum, and is only
-   * used for error reporting.  In particular, if you use an older binary to
-   * load a newer model, and the model uses a TypeSpecClass that the older
-   * binary doesn't support, then this lets us display a useful error message.
-   * </pre>
-   *
-   * <code>string type_spec_class_name = 3;</code>
-   */
-  com.google.protobuf.ByteString
-      getTypeSpecClassNameBytes();
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TypesProtos.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TypesProtos.java
deleted file mode 100644
index 9c343efcb7c..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TypesProtos.java
+++ /dev/null
@@ -1,60 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/framework/types.proto
-
-package org.tensorflow.proto.framework;
-
-public final class TypesProtos {
-  private TypesProtos() {}
-  public static void registerAllExtensions(
-      com.google.protobuf.ExtensionRegistryLite registry) {
-  }
-
-  public static void registerAllExtensions(
-      com.google.protobuf.ExtensionRegistry registry) {
-    registerAllExtensions(
-        (com.google.protobuf.ExtensionRegistryLite) registry);
-  }
-
-  public static com.google.protobuf.Descriptors.FileDescriptor
-      getDescriptor() {
-    return descriptor;
-  }
-  private static  com.google.protobuf.Descriptors.FileDescriptor
-      descriptor;
-  static {
-    java.lang.String[] descriptorData = {
-      "\n%tensorflow/core/framework/types.proto\022" +
-      "\ntensorflow*\252\006\n\010DataType\022\016\n\nDT_INVALID\020\000" +
-      "\022\014\n\010DT_FLOAT\020\001\022\r\n\tDT_DOUBLE\020\002\022\014\n\010DT_INT3" +
-      "2\020\003\022\014\n\010DT_UINT8\020\004\022\014\n\010DT_INT16\020\005\022\013\n\007DT_IN" +
-      "T8\020\006\022\r\n\tDT_STRING\020\007\022\020\n\014DT_COMPLEX64\020\010\022\014\n" +
-      "\010DT_INT64\020\t\022\013\n\007DT_BOOL\020\n\022\014\n\010DT_QINT8\020\013\022\r" +
-      "\n\tDT_QUINT8\020\014\022\r\n\tDT_QINT32\020\r\022\017\n\013DT_BFLOA" +
-      "T16\020\016\022\r\n\tDT_QINT16\020\017\022\016\n\nDT_QUINT16\020\020\022\r\n\t" +
-      "DT_UINT16\020\021\022\021\n\rDT_COMPLEX128\020\022\022\013\n\007DT_HAL" +
-      "F\020\023\022\017\n\013DT_RESOURCE\020\024\022\016\n\nDT_VARIANT\020\025\022\r\n\t" +
-      "DT_UINT32\020\026\022\r\n\tDT_UINT64\020\027\022\020\n\014DT_FLOAT_R" +
-      "EF\020e\022\021\n\rDT_DOUBLE_REF\020f\022\020\n\014DT_INT32_REF\020" +
-      "g\022\020\n\014DT_UINT8_REF\020h\022\020\n\014DT_INT16_REF\020i\022\017\n" +
-      "\013DT_INT8_REF\020j\022\021\n\rDT_STRING_REF\020k\022\024\n\020DT_" +
-      "COMPLEX64_REF\020l\022\020\n\014DT_INT64_REF\020m\022\017\n\013DT_" +
-      "BOOL_REF\020n\022\020\n\014DT_QINT8_REF\020o\022\021\n\rDT_QUINT" +
-      "8_REF\020p\022\021\n\rDT_QINT32_REF\020q\022\023\n\017DT_BFLOAT1" +
-      "6_REF\020r\022\021\n\rDT_QINT16_REF\020s\022\022\n\016DT_QUINT16" +
-      "_REF\020t\022\021\n\rDT_UINT16_REF\020u\022\025\n\021DT_COMPLEX1" +
-      "28_REF\020v\022\017\n\013DT_HALF_REF\020w\022\023\n\017DT_RESOURCE" +
-      "_REF\020x\022\022\n\016DT_VARIANT_REF\020y\022\021\n\rDT_UINT32_" +
-      "REF\020z\022\021\n\rDT_UINT64_REF\020{B\200\001\n\036org.tensorf" +
-      "low.proto.frameworkB\013TypesProtosP\001ZLgith" +
-      "ub.com/tensorflow/tensorflow/tensorflow/" +
-      "go/core/framework/types_go_proto\370\001\001b\006pro" +
-      "to3"
-    };
-    descriptor = com.google.protobuf.Descriptors.FileDescriptor
-      .internalBuildGeneratedFileFrom(descriptorData,
-        new com.google.protobuf.Descriptors.FileDescriptor[] {
-        });
-  }
-
-  // @@protoc_insertion_point(outer_class_scope)
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ValuesDef.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ValuesDef.java
deleted file mode 100644
index d6819b5bbfa..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ValuesDef.java
+++ /dev/null
@@ -1,969 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/control_flow.proto
-
-package org.tensorflow.proto.framework;
-
-/**
- * <pre>
- * Protocol buffer representing the values in ControlFlowContext.
- * </pre>
- *
- * Protobuf type {@code tensorflow.ValuesDef}
- */
-public  final class ValuesDef extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.ValuesDef)
-    ValuesDefOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use ValuesDef.newBuilder() to construct.
-  private ValuesDef(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private ValuesDef() {
-    values_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new ValuesDef();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private ValuesDef(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    int mutable_bitField0_ = 0;
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 10: {
-            java.lang.String s = input.readStringRequireUtf8();
-            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
-              values_ = new com.google.protobuf.LazyStringArrayList();
-              mutable_bitField0_ |= 0x00000001;
-            }
-            values_.add(s);
-            break;
-          }
-          case 18: {
-            if (!((mutable_bitField0_ & 0x00000002) != 0)) {
-              externalValues_ = com.google.protobuf.MapField.newMapField(
-                  ExternalValuesDefaultEntryHolder.defaultEntry);
-              mutable_bitField0_ |= 0x00000002;
-            }
-            com.google.protobuf.MapEntry<java.lang.String, java.lang.String>
-            externalValues__ = input.readMessage(
-                ExternalValuesDefaultEntryHolder.defaultEntry.getParserForType(), extensionRegistry);
-            externalValues_.getMutableMap().put(
-                externalValues__.getKey(), externalValues__.getValue());
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      if (((mutable_bitField0_ & 0x00000001) != 0)) {
-        values_ = values_.getUnmodifiableView();
-      }
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.framework.ControlFlowProtos.internal_static_tensorflow_ValuesDef_descriptor;
-  }
-
-  @SuppressWarnings({"rawtypes"})
-  @java.lang.Override
-  protected com.google.protobuf.MapField internalGetMapField(
-      int number) {
-    switch (number) {
-      case 2:
-        return internalGetExternalValues();
-      default:
-        throw new RuntimeException(
-            "Invalid map field number: " + number);
-    }
-  }
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.framework.ControlFlowProtos.internal_static_tensorflow_ValuesDef_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.framework.ValuesDef.class, org.tensorflow.proto.framework.ValuesDef.Builder.class);
-  }
-
-  public static final int VALUES_FIELD_NUMBER = 1;
-  private com.google.protobuf.LazyStringList values_;
-  /**
-   * <pre>
-   * Value names that have been seen in this context.
-   * </pre>
-   *
-   * <code>repeated string values = 1;</code>
-   */
-  public com.google.protobuf.ProtocolStringList
-      getValuesList() {
-    return values_;
-  }
-  /**
-   * <pre>
-   * Value names that have been seen in this context.
-   * </pre>
-   *
-   * <code>repeated string values = 1;</code>
-   */
-  public int getValuesCount() {
-    return values_.size();
-  }
-  /**
-   * <pre>
-   * Value names that have been seen in this context.
-   * </pre>
-   *
-   * <code>repeated string values = 1;</code>
-   */
-  public java.lang.String getValues(int index) {
-    return values_.get(index);
-  }
-  /**
-   * <pre>
-   * Value names that have been seen in this context.
-   * </pre>
-   *
-   * <code>repeated string values = 1;</code>
-   */
-  public com.google.protobuf.ByteString
-      getValuesBytes(int index) {
-    return values_.getByteString(index);
-  }
-
-  public static final int EXTERNAL_VALUES_FIELD_NUMBER = 2;
-  private static final class ExternalValuesDefaultEntryHolder {
-    static final com.google.protobuf.MapEntry<
-        java.lang.String, java.lang.String> defaultEntry =
-            com.google.protobuf.MapEntry
-            .<java.lang.String, java.lang.String>newDefaultInstance(
-                org.tensorflow.proto.framework.ControlFlowProtos.internal_static_tensorflow_ValuesDef_ExternalValuesEntry_descriptor, 
-                com.google.protobuf.WireFormat.FieldType.STRING,
-                "",
-                com.google.protobuf.WireFormat.FieldType.STRING,
-                "");
-  }
-  private com.google.protobuf.MapField<
-      java.lang.String, java.lang.String> externalValues_;
-  private com.google.protobuf.MapField<java.lang.String, java.lang.String>
-  internalGetExternalValues() {
-    if (externalValues_ == null) {
-      return com.google.protobuf.MapField.emptyMapField(
-          ExternalValuesDefaultEntryHolder.defaultEntry);
-    }
-    return externalValues_;
-  }
-
-  public int getExternalValuesCount() {
-    return internalGetExternalValues().getMap().size();
-  }
-  /**
-   * <pre>
-   * Value names referenced by but external to this context.
-   * </pre>
-   *
-   * <code>map&lt;string, string&gt; external_values = 2;</code>
-   */
-
-  public boolean containsExternalValues(
-      java.lang.String key) {
-    if (key == null) { throw new java.lang.NullPointerException(); }
-    return internalGetExternalValues().getMap().containsKey(key);
-  }
-  /**
-   * Use {@link #getExternalValuesMap()} instead.
-   */
-  @java.lang.Deprecated
-  public java.util.Map<java.lang.String, java.lang.String> getExternalValues() {
-    return getExternalValuesMap();
-  }
-  /**
-   * <pre>
-   * Value names referenced by but external to this context.
-   * </pre>
-   *
-   * <code>map&lt;string, string&gt; external_values = 2;</code>
-   */
-
-  public java.util.Map<java.lang.String, java.lang.String> getExternalValuesMap() {
-    return internalGetExternalValues().getMap();
-  }
-  /**
-   * <pre>
-   * Value names referenced by but external to this context.
-   * </pre>
-   *
-   * <code>map&lt;string, string&gt; external_values = 2;</code>
-   */
-
-  public java.lang.String getExternalValuesOrDefault(
-      java.lang.String key,
-      java.lang.String defaultValue) {
-    if (key == null) { throw new java.lang.NullPointerException(); }
-    java.util.Map<java.lang.String, java.lang.String> map =
-        internalGetExternalValues().getMap();
-    return map.containsKey(key) ? map.get(key) : defaultValue;
-  }
-  /**
-   * <pre>
-   * Value names referenced by but external to this context.
-   * </pre>
-   *
-   * <code>map&lt;string, string&gt; external_values = 2;</code>
-   */
-
-  public java.lang.String getExternalValuesOrThrow(
-      java.lang.String key) {
-    if (key == null) { throw new java.lang.NullPointerException(); }
-    java.util.Map<java.lang.String, java.lang.String> map =
-        internalGetExternalValues().getMap();
-    if (!map.containsKey(key)) {
-      throw new java.lang.IllegalArgumentException();
-    }
-    return map.get(key);
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    for (int i = 0; i < values_.size(); i++) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 1, values_.getRaw(i));
-    }
-    com.google.protobuf.GeneratedMessageV3
-      .serializeStringMapTo(
-        output,
-        internalGetExternalValues(),
-        ExternalValuesDefaultEntryHolder.defaultEntry,
-        2);
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    {
-      int dataSize = 0;
-      for (int i = 0; i < values_.size(); i++) {
-        dataSize += computeStringSizeNoTag(values_.getRaw(i));
-      }
-      size += dataSize;
-      size += 1 * getValuesList().size();
-    }
-    for (java.util.Map.Entry<java.lang.String, java.lang.String> entry
-         : internalGetExternalValues().getMap().entrySet()) {
-      com.google.protobuf.MapEntry<java.lang.String, java.lang.String>
-      externalValues__ = ExternalValuesDefaultEntryHolder.defaultEntry.newBuilderForType()
-          .setKey(entry.getKey())
-          .setValue(entry.getValue())
-          .build();
-      size += com.google.protobuf.CodedOutputStream
-          .computeMessageSize(2, externalValues__);
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.framework.ValuesDef)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.framework.ValuesDef other = (org.tensorflow.proto.framework.ValuesDef) obj;
-
-    if (!getValuesList()
-        .equals(other.getValuesList())) return false;
-    if (!internalGetExternalValues().equals(
-        other.internalGetExternalValues())) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    if (getValuesCount() > 0) {
-      hash = (37 * hash) + VALUES_FIELD_NUMBER;
-      hash = (53 * hash) + getValuesList().hashCode();
-    }
-    if (!internalGetExternalValues().getMap().isEmpty()) {
-      hash = (37 * hash) + EXTERNAL_VALUES_FIELD_NUMBER;
-      hash = (53 * hash) + internalGetExternalValues().hashCode();
-    }
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.framework.ValuesDef parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.ValuesDef parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.ValuesDef parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.ValuesDef parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.ValuesDef parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.ValuesDef parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.ValuesDef parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.ValuesDef parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.ValuesDef parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.ValuesDef parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.ValuesDef parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.ValuesDef parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.framework.ValuesDef prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * <pre>
-   * Protocol buffer representing the values in ControlFlowContext.
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.ValuesDef}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.ValuesDef)
-      org.tensorflow.proto.framework.ValuesDefOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.ControlFlowProtos.internal_static_tensorflow_ValuesDef_descriptor;
-    }
-
-    @SuppressWarnings({"rawtypes"})
-    protected com.google.protobuf.MapField internalGetMapField(
-        int number) {
-      switch (number) {
-        case 2:
-          return internalGetExternalValues();
-        default:
-          throw new RuntimeException(
-              "Invalid map field number: " + number);
-      }
-    }
-    @SuppressWarnings({"rawtypes"})
-    protected com.google.protobuf.MapField internalGetMutableMapField(
-        int number) {
-      switch (number) {
-        case 2:
-          return internalGetMutableExternalValues();
-        default:
-          throw new RuntimeException(
-              "Invalid map field number: " + number);
-      }
-    }
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.ControlFlowProtos.internal_static_tensorflow_ValuesDef_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.ValuesDef.class, org.tensorflow.proto.framework.ValuesDef.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.framework.ValuesDef.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      values_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-      bitField0_ = (bitField0_ & ~0x00000001);
-      internalGetMutableExternalValues().clear();
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.framework.ControlFlowProtos.internal_static_tensorflow_ValuesDef_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.ValuesDef getDefaultInstanceForType() {
-      return org.tensorflow.proto.framework.ValuesDef.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.ValuesDef build() {
-      org.tensorflow.proto.framework.ValuesDef result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.ValuesDef buildPartial() {
-      org.tensorflow.proto.framework.ValuesDef result = new org.tensorflow.proto.framework.ValuesDef(this);
-      int from_bitField0_ = bitField0_;
-      if (((bitField0_ & 0x00000001) != 0)) {
-        values_ = values_.getUnmodifiableView();
-        bitField0_ = (bitField0_ & ~0x00000001);
-      }
-      result.values_ = values_;
-      result.externalValues_ = internalGetExternalValues();
-      result.externalValues_.makeImmutable();
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.framework.ValuesDef) {
-        return mergeFrom((org.tensorflow.proto.framework.ValuesDef)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.framework.ValuesDef other) {
-      if (other == org.tensorflow.proto.framework.ValuesDef.getDefaultInstance()) return this;
-      if (!other.values_.isEmpty()) {
-        if (values_.isEmpty()) {
-          values_ = other.values_;
-          bitField0_ = (bitField0_ & ~0x00000001);
-        } else {
-          ensureValuesIsMutable();
-          values_.addAll(other.values_);
-        }
-        onChanged();
-      }
-      internalGetMutableExternalValues().mergeFrom(
-          other.internalGetExternalValues());
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.framework.ValuesDef parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.framework.ValuesDef) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-    private int bitField0_;
-
-    private com.google.protobuf.LazyStringList values_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-    private void ensureValuesIsMutable() {
-      if (!((bitField0_ & 0x00000001) != 0)) {
-        values_ = new com.google.protobuf.LazyStringArrayList(values_);
-        bitField0_ |= 0x00000001;
-       }
-    }
-    /**
-     * <pre>
-     * Value names that have been seen in this context.
-     * </pre>
-     *
-     * <code>repeated string values = 1;</code>
-     */
-    public com.google.protobuf.ProtocolStringList
-        getValuesList() {
-      return values_.getUnmodifiableView();
-    }
-    /**
-     * <pre>
-     * Value names that have been seen in this context.
-     * </pre>
-     *
-     * <code>repeated string values = 1;</code>
-     */
-    public int getValuesCount() {
-      return values_.size();
-    }
-    /**
-     * <pre>
-     * Value names that have been seen in this context.
-     * </pre>
-     *
-     * <code>repeated string values = 1;</code>
-     */
-    public java.lang.String getValues(int index) {
-      return values_.get(index);
-    }
-    /**
-     * <pre>
-     * Value names that have been seen in this context.
-     * </pre>
-     *
-     * <code>repeated string values = 1;</code>
-     */
-    public com.google.protobuf.ByteString
-        getValuesBytes(int index) {
-      return values_.getByteString(index);
-    }
-    /**
-     * <pre>
-     * Value names that have been seen in this context.
-     * </pre>
-     *
-     * <code>repeated string values = 1;</code>
-     */
-    public Builder setValues(
-        int index, java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  ensureValuesIsMutable();
-      values_.set(index, value);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Value names that have been seen in this context.
-     * </pre>
-     *
-     * <code>repeated string values = 1;</code>
-     */
-    public Builder addValues(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  ensureValuesIsMutable();
-      values_.add(value);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Value names that have been seen in this context.
-     * </pre>
-     *
-     * <code>repeated string values = 1;</code>
-     */
-    public Builder addAllValues(
-        java.lang.Iterable<java.lang.String> values) {
-      ensureValuesIsMutable();
-      com.google.protobuf.AbstractMessageLite.Builder.addAll(
-          values, values_);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Value names that have been seen in this context.
-     * </pre>
-     *
-     * <code>repeated string values = 1;</code>
-     */
-    public Builder clearValues() {
-      values_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-      bitField0_ = (bitField0_ & ~0x00000001);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Value names that have been seen in this context.
-     * </pre>
-     *
-     * <code>repeated string values = 1;</code>
-     */
-    public Builder addValuesBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      ensureValuesIsMutable();
-      values_.add(value);
-      onChanged();
-      return this;
-    }
-
-    private com.google.protobuf.MapField<
-        java.lang.String, java.lang.String> externalValues_;
-    private com.google.protobuf.MapField<java.lang.String, java.lang.String>
-    internalGetExternalValues() {
-      if (externalValues_ == null) {
-        return com.google.protobuf.MapField.emptyMapField(
-            ExternalValuesDefaultEntryHolder.defaultEntry);
-      }
-      return externalValues_;
-    }
-    private com.google.protobuf.MapField<java.lang.String, java.lang.String>
-    internalGetMutableExternalValues() {
-      onChanged();;
-      if (externalValues_ == null) {
-        externalValues_ = com.google.protobuf.MapField.newMapField(
-            ExternalValuesDefaultEntryHolder.defaultEntry);
-      }
-      if (!externalValues_.isMutable()) {
-        externalValues_ = externalValues_.copy();
-      }
-      return externalValues_;
-    }
-
-    public int getExternalValuesCount() {
-      return internalGetExternalValues().getMap().size();
-    }
-    /**
-     * <pre>
-     * Value names referenced by but external to this context.
-     * </pre>
-     *
-     * <code>map&lt;string, string&gt; external_values = 2;</code>
-     */
-
-    public boolean containsExternalValues(
-        java.lang.String key) {
-      if (key == null) { throw new java.lang.NullPointerException(); }
-      return internalGetExternalValues().getMap().containsKey(key);
-    }
-    /**
-     * Use {@link #getExternalValuesMap()} instead.
-     */
-    @java.lang.Deprecated
-    public java.util.Map<java.lang.String, java.lang.String> getExternalValues() {
-      return getExternalValuesMap();
-    }
-    /**
-     * <pre>
-     * Value names referenced by but external to this context.
-     * </pre>
-     *
-     * <code>map&lt;string, string&gt; external_values = 2;</code>
-     */
-
-    public java.util.Map<java.lang.String, java.lang.String> getExternalValuesMap() {
-      return internalGetExternalValues().getMap();
-    }
-    /**
-     * <pre>
-     * Value names referenced by but external to this context.
-     * </pre>
-     *
-     * <code>map&lt;string, string&gt; external_values = 2;</code>
-     */
-
-    public java.lang.String getExternalValuesOrDefault(
-        java.lang.String key,
-        java.lang.String defaultValue) {
-      if (key == null) { throw new java.lang.NullPointerException(); }
-      java.util.Map<java.lang.String, java.lang.String> map =
-          internalGetExternalValues().getMap();
-      return map.containsKey(key) ? map.get(key) : defaultValue;
-    }
-    /**
-     * <pre>
-     * Value names referenced by but external to this context.
-     * </pre>
-     *
-     * <code>map&lt;string, string&gt; external_values = 2;</code>
-     */
-
-    public java.lang.String getExternalValuesOrThrow(
-        java.lang.String key) {
-      if (key == null) { throw new java.lang.NullPointerException(); }
-      java.util.Map<java.lang.String, java.lang.String> map =
-          internalGetExternalValues().getMap();
-      if (!map.containsKey(key)) {
-        throw new java.lang.IllegalArgumentException();
-      }
-      return map.get(key);
-    }
-
-    public Builder clearExternalValues() {
-      internalGetMutableExternalValues().getMutableMap()
-          .clear();
-      return this;
-    }
-    /**
-     * <pre>
-     * Value names referenced by but external to this context.
-     * </pre>
-     *
-     * <code>map&lt;string, string&gt; external_values = 2;</code>
-     */
-
-    public Builder removeExternalValues(
-        java.lang.String key) {
-      if (key == null) { throw new java.lang.NullPointerException(); }
-      internalGetMutableExternalValues().getMutableMap()
-          .remove(key);
-      return this;
-    }
-    /**
-     * Use alternate mutation accessors instead.
-     */
-    @java.lang.Deprecated
-    public java.util.Map<java.lang.String, java.lang.String>
-    getMutableExternalValues() {
-      return internalGetMutableExternalValues().getMutableMap();
-    }
-    /**
-     * <pre>
-     * Value names referenced by but external to this context.
-     * </pre>
-     *
-     * <code>map&lt;string, string&gt; external_values = 2;</code>
-     */
-    public Builder putExternalValues(
-        java.lang.String key,
-        java.lang.String value) {
-      if (key == null) { throw new java.lang.NullPointerException(); }
-      if (value == null) { throw new java.lang.NullPointerException(); }
-      internalGetMutableExternalValues().getMutableMap()
-          .put(key, value);
-      return this;
-    }
-    /**
-     * <pre>
-     * Value names referenced by but external to this context.
-     * </pre>
-     *
-     * <code>map&lt;string, string&gt; external_values = 2;</code>
-     */
-
-    public Builder putAllExternalValues(
-        java.util.Map<java.lang.String, java.lang.String> values) {
-      internalGetMutableExternalValues().getMutableMap()
-          .putAll(values);
-      return this;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.ValuesDef)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.ValuesDef)
-  private static final org.tensorflow.proto.framework.ValuesDef DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.ValuesDef();
-  }
-
-  public static org.tensorflow.proto.framework.ValuesDef getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<ValuesDef>
-      PARSER = new com.google.protobuf.AbstractParser<ValuesDef>() {
-    @java.lang.Override
-    public ValuesDef parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new ValuesDef(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<ValuesDef> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<ValuesDef> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.framework.ValuesDef getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ValuesDefOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ValuesDefOrBuilder.java
deleted file mode 100644
index d9135689d22..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ValuesDefOrBuilder.java
+++ /dev/null
@@ -1,98 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/control_flow.proto
-
-package org.tensorflow.proto.framework;
-
-public interface ValuesDefOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.ValuesDef)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <pre>
-   * Value names that have been seen in this context.
-   * </pre>
-   *
-   * <code>repeated string values = 1;</code>
-   */
-  java.util.List<java.lang.String>
-      getValuesList();
-  /**
-   * <pre>
-   * Value names that have been seen in this context.
-   * </pre>
-   *
-   * <code>repeated string values = 1;</code>
-   */
-  int getValuesCount();
-  /**
-   * <pre>
-   * Value names that have been seen in this context.
-   * </pre>
-   *
-   * <code>repeated string values = 1;</code>
-   */
-  java.lang.String getValues(int index);
-  /**
-   * <pre>
-   * Value names that have been seen in this context.
-   * </pre>
-   *
-   * <code>repeated string values = 1;</code>
-   */
-  com.google.protobuf.ByteString
-      getValuesBytes(int index);
-
-  /**
-   * <pre>
-   * Value names referenced by but external to this context.
-   * </pre>
-   *
-   * <code>map&lt;string, string&gt; external_values = 2;</code>
-   */
-  int getExternalValuesCount();
-  /**
-   * <pre>
-   * Value names referenced by but external to this context.
-   * </pre>
-   *
-   * <code>map&lt;string, string&gt; external_values = 2;</code>
-   */
-  boolean containsExternalValues(
-      java.lang.String key);
-  /**
-   * Use {@link #getExternalValuesMap()} instead.
-   */
-  @java.lang.Deprecated
-  java.util.Map<java.lang.String, java.lang.String>
-  getExternalValues();
-  /**
-   * <pre>
-   * Value names referenced by but external to this context.
-   * </pre>
-   *
-   * <code>map&lt;string, string&gt; external_values = 2;</code>
-   */
-  java.util.Map<java.lang.String, java.lang.String>
-  getExternalValuesMap();
-  /**
-   * <pre>
-   * Value names referenced by but external to this context.
-   * </pre>
-   *
-   * <code>map&lt;string, string&gt; external_values = 2;</code>
-   */
-
-  java.lang.String getExternalValuesOrDefault(
-      java.lang.String key,
-      java.lang.String defaultValue);
-  /**
-   * <pre>
-   * Value names referenced by but external to this context.
-   * </pre>
-   *
-   * <code>map&lt;string, string&gt; external_values = 2;</code>
-   */
-
-  java.lang.String getExternalValuesOrThrow(
-      java.lang.String key);
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/VariableAggregation.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/VariableAggregation.java
deleted file mode 100644
index 584cd9b19bd..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/VariableAggregation.java
+++ /dev/null
@@ -1,165 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/framework/variable.proto
-
-package org.tensorflow.proto.framework;
-
-/**
- * <pre>
- * Indicates how a distributed variable will be aggregated.
- * </pre>
- *
- * Protobuf enum {@code tensorflow.VariableAggregation}
- */
-public enum VariableAggregation
-    implements com.google.protobuf.ProtocolMessageEnum {
-  /**
-   * <pre>
-   * `NONE`: This is the default, giving an error if you use a
-   * variable-update operation with multiple replicas.
-   * </pre>
-   *
-   * <code>VARIABLE_AGGREGATION_NONE = 0;</code>
-   */
-  VARIABLE_AGGREGATION_NONE(0),
-  /**
-   * <pre>
-   * `SUM`: Add the updates across replicas.
-   * </pre>
-   *
-   * <code>VARIABLE_AGGREGATION_SUM = 1;</code>
-   */
-  VARIABLE_AGGREGATION_SUM(1),
-  /**
-   * <pre>
-   * `MEAN`: Take the arithmetic mean ("average") of the updates across
-   * replicas.
-   * </pre>
-   *
-   * <code>VARIABLE_AGGREGATION_MEAN = 2;</code>
-   */
-  VARIABLE_AGGREGATION_MEAN(2),
-  /**
-   * <pre>
-   * `ONLY_FIRST_REPLICA`: This is for when every replica is performing the same
-   * update, but we only want to perform the update once. Used, e.g., for the
-   * global step counter.
-   * </pre>
-   *
-   * <code>VARIABLE_AGGREGATION_ONLY_FIRST_REPLICA = 3;</code>
-   */
-  VARIABLE_AGGREGATION_ONLY_FIRST_REPLICA(3),
-  UNRECOGNIZED(-1),
-  ;
-
-  /**
-   * <pre>
-   * `NONE`: This is the default, giving an error if you use a
-   * variable-update operation with multiple replicas.
-   * </pre>
-   *
-   * <code>VARIABLE_AGGREGATION_NONE = 0;</code>
-   */
-  public static final int VARIABLE_AGGREGATION_NONE_VALUE = 0;
-  /**
-   * <pre>
-   * `SUM`: Add the updates across replicas.
-   * </pre>
-   *
-   * <code>VARIABLE_AGGREGATION_SUM = 1;</code>
-   */
-  public static final int VARIABLE_AGGREGATION_SUM_VALUE = 1;
-  /**
-   * <pre>
-   * `MEAN`: Take the arithmetic mean ("average") of the updates across
-   * replicas.
-   * </pre>
-   *
-   * <code>VARIABLE_AGGREGATION_MEAN = 2;</code>
-   */
-  public static final int VARIABLE_AGGREGATION_MEAN_VALUE = 2;
-  /**
-   * <pre>
-   * `ONLY_FIRST_REPLICA`: This is for when every replica is performing the same
-   * update, but we only want to perform the update once. Used, e.g., for the
-   * global step counter.
-   * </pre>
-   *
-   * <code>VARIABLE_AGGREGATION_ONLY_FIRST_REPLICA = 3;</code>
-   */
-  public static final int VARIABLE_AGGREGATION_ONLY_FIRST_REPLICA_VALUE = 3;
-
-
-  public final int getNumber() {
-    if (this == UNRECOGNIZED) {
-      throw new java.lang.IllegalArgumentException(
-          "Can't get the number of an unknown enum value.");
-    }
-    return value;
-  }
-
-  /**
-   * @deprecated Use {@link #forNumber(int)} instead.
-   */
-  @java.lang.Deprecated
-  public static VariableAggregation valueOf(int value) {
-    return forNumber(value);
-  }
-
-  public static VariableAggregation forNumber(int value) {
-    switch (value) {
-      case 0: return VARIABLE_AGGREGATION_NONE;
-      case 1: return VARIABLE_AGGREGATION_SUM;
-      case 2: return VARIABLE_AGGREGATION_MEAN;
-      case 3: return VARIABLE_AGGREGATION_ONLY_FIRST_REPLICA;
-      default: return null;
-    }
-  }
-
-  public static com.google.protobuf.Internal.EnumLiteMap<VariableAggregation>
-      internalGetValueMap() {
-    return internalValueMap;
-  }
-  private static final com.google.protobuf.Internal.EnumLiteMap<
-      VariableAggregation> internalValueMap =
-        new com.google.protobuf.Internal.EnumLiteMap<VariableAggregation>() {
-          public VariableAggregation findValueByNumber(int number) {
-            return VariableAggregation.forNumber(number);
-          }
-        };
-
-  public final com.google.protobuf.Descriptors.EnumValueDescriptor
-      getValueDescriptor() {
-    return getDescriptor().getValues().get(ordinal());
-  }
-  public final com.google.protobuf.Descriptors.EnumDescriptor
-      getDescriptorForType() {
-    return getDescriptor();
-  }
-  public static final com.google.protobuf.Descriptors.EnumDescriptor
-      getDescriptor() {
-    return org.tensorflow.proto.framework.VariableProtos.getDescriptor().getEnumTypes().get(1);
-  }
-
-  private static final VariableAggregation[] VALUES = values();
-
-  public static VariableAggregation valueOf(
-      com.google.protobuf.Descriptors.EnumValueDescriptor desc) {
-    if (desc.getType() != getDescriptor()) {
-      throw new java.lang.IllegalArgumentException(
-        "EnumValueDescriptor is not for this type.");
-    }
-    if (desc.getIndex() == -1) {
-      return UNRECOGNIZED;
-    }
-    return VALUES[desc.getIndex()];
-  }
-
-  private final int value;
-
-  private VariableAggregation(int value) {
-    this.value = value;
-  }
-
-  // @@protoc_insertion_point(enum_scope:tensorflow.VariableAggregation)
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/VariableDef.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/VariableDef.java
deleted file mode 100644
index e0ecee5dc06..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/VariableDef.java
+++ /dev/null
@@ -1,1650 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/framework/variable.proto
-
-package org.tensorflow.proto.framework;
-
-/**
- * <pre>
- * Protocol buffer representing a Variable.
- * </pre>
- *
- * Protobuf type {@code tensorflow.VariableDef}
- */
-public  final class VariableDef extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.VariableDef)
-    VariableDefOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use VariableDef.newBuilder() to construct.
-  private VariableDef(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private VariableDef() {
-    variableName_ = "";
-    initialValueName_ = "";
-    initializerName_ = "";
-    snapshotName_ = "";
-    synchronization_ = 0;
-    aggregation_ = 0;
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new VariableDef();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private VariableDef(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 10: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            variableName_ = s;
-            break;
-          }
-          case 18: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            initializerName_ = s;
-            break;
-          }
-          case 26: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            snapshotName_ = s;
-            break;
-          }
-          case 34: {
-            org.tensorflow.proto.framework.SaveSliceInfoDef.Builder subBuilder = null;
-            if (saveSliceInfoDef_ != null) {
-              subBuilder = saveSliceInfoDef_.toBuilder();
-            }
-            saveSliceInfoDef_ = input.readMessage(org.tensorflow.proto.framework.SaveSliceInfoDef.parser(), extensionRegistry);
-            if (subBuilder != null) {
-              subBuilder.mergeFrom(saveSliceInfoDef_);
-              saveSliceInfoDef_ = subBuilder.buildPartial();
-            }
-
-            break;
-          }
-          case 40: {
-
-            isResource_ = input.readBool();
-            break;
-          }
-          case 50: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            initialValueName_ = s;
-            break;
-          }
-          case 56: {
-
-            trainable_ = input.readBool();
-            break;
-          }
-          case 64: {
-            int rawValue = input.readEnum();
-
-            synchronization_ = rawValue;
-            break;
-          }
-          case 72: {
-            int rawValue = input.readEnum();
-
-            aggregation_ = rawValue;
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.framework.VariableProtos.internal_static_tensorflow_VariableDef_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.framework.VariableProtos.internal_static_tensorflow_VariableDef_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.framework.VariableDef.class, org.tensorflow.proto.framework.VariableDef.Builder.class);
-  }
-
-  public static final int VARIABLE_NAME_FIELD_NUMBER = 1;
-  private volatile java.lang.Object variableName_;
-  /**
-   * <pre>
-   * Name of the variable tensor.
-   * </pre>
-   *
-   * <code>string variable_name = 1;</code>
-   */
-  public java.lang.String getVariableName() {
-    java.lang.Object ref = variableName_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      variableName_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * Name of the variable tensor.
-   * </pre>
-   *
-   * <code>string variable_name = 1;</code>
-   */
-  public com.google.protobuf.ByteString
-      getVariableNameBytes() {
-    java.lang.Object ref = variableName_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      variableName_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int INITIAL_VALUE_NAME_FIELD_NUMBER = 6;
-  private volatile java.lang.Object initialValueName_;
-  /**
-   * <pre>
-   * Name of the tensor holding the variable's initial value.
-   * </pre>
-   *
-   * <code>string initial_value_name = 6;</code>
-   */
-  public java.lang.String getInitialValueName() {
-    java.lang.Object ref = initialValueName_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      initialValueName_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * Name of the tensor holding the variable's initial value.
-   * </pre>
-   *
-   * <code>string initial_value_name = 6;</code>
-   */
-  public com.google.protobuf.ByteString
-      getInitialValueNameBytes() {
-    java.lang.Object ref = initialValueName_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      initialValueName_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int INITIALIZER_NAME_FIELD_NUMBER = 2;
-  private volatile java.lang.Object initializerName_;
-  /**
-   * <pre>
-   * Name of the initializer op.
-   * </pre>
-   *
-   * <code>string initializer_name = 2;</code>
-   */
-  public java.lang.String getInitializerName() {
-    java.lang.Object ref = initializerName_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      initializerName_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * Name of the initializer op.
-   * </pre>
-   *
-   * <code>string initializer_name = 2;</code>
-   */
-  public com.google.protobuf.ByteString
-      getInitializerNameBytes() {
-    java.lang.Object ref = initializerName_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      initializerName_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int SNAPSHOT_NAME_FIELD_NUMBER = 3;
-  private volatile java.lang.Object snapshotName_;
-  /**
-   * <pre>
-   * Name of the snapshot tensor.
-   * </pre>
-   *
-   * <code>string snapshot_name = 3;</code>
-   */
-  public java.lang.String getSnapshotName() {
-    java.lang.Object ref = snapshotName_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      snapshotName_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * Name of the snapshot tensor.
-   * </pre>
-   *
-   * <code>string snapshot_name = 3;</code>
-   */
-  public com.google.protobuf.ByteString
-      getSnapshotNameBytes() {
-    java.lang.Object ref = snapshotName_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      snapshotName_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int SAVE_SLICE_INFO_DEF_FIELD_NUMBER = 4;
-  private org.tensorflow.proto.framework.SaveSliceInfoDef saveSliceInfoDef_;
-  /**
-   * <pre>
-   * Support for saving variables as slices of a larger variable.
-   * </pre>
-   *
-   * <code>.tensorflow.SaveSliceInfoDef save_slice_info_def = 4;</code>
-   */
-  public boolean hasSaveSliceInfoDef() {
-    return saveSliceInfoDef_ != null;
-  }
-  /**
-   * <pre>
-   * Support for saving variables as slices of a larger variable.
-   * </pre>
-   *
-   * <code>.tensorflow.SaveSliceInfoDef save_slice_info_def = 4;</code>
-   */
-  public org.tensorflow.proto.framework.SaveSliceInfoDef getSaveSliceInfoDef() {
-    return saveSliceInfoDef_ == null ? org.tensorflow.proto.framework.SaveSliceInfoDef.getDefaultInstance() : saveSliceInfoDef_;
-  }
-  /**
-   * <pre>
-   * Support for saving variables as slices of a larger variable.
-   * </pre>
-   *
-   * <code>.tensorflow.SaveSliceInfoDef save_slice_info_def = 4;</code>
-   */
-  public org.tensorflow.proto.framework.SaveSliceInfoDefOrBuilder getSaveSliceInfoDefOrBuilder() {
-    return getSaveSliceInfoDef();
-  }
-
-  public static final int IS_RESOURCE_FIELD_NUMBER = 5;
-  private boolean isResource_;
-  /**
-   * <pre>
-   * Whether to represent this as a ResourceVariable.
-   * </pre>
-   *
-   * <code>bool is_resource = 5;</code>
-   */
-  public boolean getIsResource() {
-    return isResource_;
-  }
-
-  public static final int TRAINABLE_FIELD_NUMBER = 7;
-  private boolean trainable_;
-  /**
-   * <pre>
-   * Whether this variable should be trained.
-   * </pre>
-   *
-   * <code>bool trainable = 7;</code>
-   */
-  public boolean getTrainable() {
-    return trainable_;
-  }
-
-  public static final int SYNCHRONIZATION_FIELD_NUMBER = 8;
-  private int synchronization_;
-  /**
-   * <pre>
-   * Indicates when a distributed variable will be synced.
-   * </pre>
-   *
-   * <code>.tensorflow.VariableSynchronization synchronization = 8;</code>
-   */
-  public int getSynchronizationValue() {
-    return synchronization_;
-  }
-  /**
-   * <pre>
-   * Indicates when a distributed variable will be synced.
-   * </pre>
-   *
-   * <code>.tensorflow.VariableSynchronization synchronization = 8;</code>
-   */
-  public org.tensorflow.proto.framework.VariableSynchronization getSynchronization() {
-    @SuppressWarnings("deprecation")
-    org.tensorflow.proto.framework.VariableSynchronization result = org.tensorflow.proto.framework.VariableSynchronization.valueOf(synchronization_);
-    return result == null ? org.tensorflow.proto.framework.VariableSynchronization.UNRECOGNIZED : result;
-  }
-
-  public static final int AGGREGATION_FIELD_NUMBER = 9;
-  private int aggregation_;
-  /**
-   * <pre>
-   * Indicates how a distributed variable will be aggregated.
-   * </pre>
-   *
-   * <code>.tensorflow.VariableAggregation aggregation = 9;</code>
-   */
-  public int getAggregationValue() {
-    return aggregation_;
-  }
-  /**
-   * <pre>
-   * Indicates how a distributed variable will be aggregated.
-   * </pre>
-   *
-   * <code>.tensorflow.VariableAggregation aggregation = 9;</code>
-   */
-  public org.tensorflow.proto.framework.VariableAggregation getAggregation() {
-    @SuppressWarnings("deprecation")
-    org.tensorflow.proto.framework.VariableAggregation result = org.tensorflow.proto.framework.VariableAggregation.valueOf(aggregation_);
-    return result == null ? org.tensorflow.proto.framework.VariableAggregation.UNRECOGNIZED : result;
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    if (!getVariableNameBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 1, variableName_);
-    }
-    if (!getInitializerNameBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 2, initializerName_);
-    }
-    if (!getSnapshotNameBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 3, snapshotName_);
-    }
-    if (saveSliceInfoDef_ != null) {
-      output.writeMessage(4, getSaveSliceInfoDef());
-    }
-    if (isResource_ != false) {
-      output.writeBool(5, isResource_);
-    }
-    if (!getInitialValueNameBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 6, initialValueName_);
-    }
-    if (trainable_ != false) {
-      output.writeBool(7, trainable_);
-    }
-    if (synchronization_ != org.tensorflow.proto.framework.VariableSynchronization.VARIABLE_SYNCHRONIZATION_AUTO.getNumber()) {
-      output.writeEnum(8, synchronization_);
-    }
-    if (aggregation_ != org.tensorflow.proto.framework.VariableAggregation.VARIABLE_AGGREGATION_NONE.getNumber()) {
-      output.writeEnum(9, aggregation_);
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (!getVariableNameBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, variableName_);
-    }
-    if (!getInitializerNameBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(2, initializerName_);
-    }
-    if (!getSnapshotNameBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(3, snapshotName_);
-    }
-    if (saveSliceInfoDef_ != null) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(4, getSaveSliceInfoDef());
-    }
-    if (isResource_ != false) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeBoolSize(5, isResource_);
-    }
-    if (!getInitialValueNameBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(6, initialValueName_);
-    }
-    if (trainable_ != false) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeBoolSize(7, trainable_);
-    }
-    if (synchronization_ != org.tensorflow.proto.framework.VariableSynchronization.VARIABLE_SYNCHRONIZATION_AUTO.getNumber()) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeEnumSize(8, synchronization_);
-    }
-    if (aggregation_ != org.tensorflow.proto.framework.VariableAggregation.VARIABLE_AGGREGATION_NONE.getNumber()) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeEnumSize(9, aggregation_);
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.framework.VariableDef)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.framework.VariableDef other = (org.tensorflow.proto.framework.VariableDef) obj;
-
-    if (!getVariableName()
-        .equals(other.getVariableName())) return false;
-    if (!getInitialValueName()
-        .equals(other.getInitialValueName())) return false;
-    if (!getInitializerName()
-        .equals(other.getInitializerName())) return false;
-    if (!getSnapshotName()
-        .equals(other.getSnapshotName())) return false;
-    if (hasSaveSliceInfoDef() != other.hasSaveSliceInfoDef()) return false;
-    if (hasSaveSliceInfoDef()) {
-      if (!getSaveSliceInfoDef()
-          .equals(other.getSaveSliceInfoDef())) return false;
-    }
-    if (getIsResource()
-        != other.getIsResource()) return false;
-    if (getTrainable()
-        != other.getTrainable()) return false;
-    if (synchronization_ != other.synchronization_) return false;
-    if (aggregation_ != other.aggregation_) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    hash = (37 * hash) + VARIABLE_NAME_FIELD_NUMBER;
-    hash = (53 * hash) + getVariableName().hashCode();
-    hash = (37 * hash) + INITIAL_VALUE_NAME_FIELD_NUMBER;
-    hash = (53 * hash) + getInitialValueName().hashCode();
-    hash = (37 * hash) + INITIALIZER_NAME_FIELD_NUMBER;
-    hash = (53 * hash) + getInitializerName().hashCode();
-    hash = (37 * hash) + SNAPSHOT_NAME_FIELD_NUMBER;
-    hash = (53 * hash) + getSnapshotName().hashCode();
-    if (hasSaveSliceInfoDef()) {
-      hash = (37 * hash) + SAVE_SLICE_INFO_DEF_FIELD_NUMBER;
-      hash = (53 * hash) + getSaveSliceInfoDef().hashCode();
-    }
-    hash = (37 * hash) + IS_RESOURCE_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashBoolean(
-        getIsResource());
-    hash = (37 * hash) + TRAINABLE_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashBoolean(
-        getTrainable());
-    hash = (37 * hash) + SYNCHRONIZATION_FIELD_NUMBER;
-    hash = (53 * hash) + synchronization_;
-    hash = (37 * hash) + AGGREGATION_FIELD_NUMBER;
-    hash = (53 * hash) + aggregation_;
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.framework.VariableDef parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.VariableDef parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.VariableDef parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.VariableDef parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.VariableDef parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.VariableDef parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.VariableDef parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.VariableDef parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.VariableDef parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.VariableDef parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.VariableDef parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.VariableDef parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.framework.VariableDef prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * <pre>
-   * Protocol buffer representing a Variable.
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.VariableDef}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.VariableDef)
-      org.tensorflow.proto.framework.VariableDefOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.VariableProtos.internal_static_tensorflow_VariableDef_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.VariableProtos.internal_static_tensorflow_VariableDef_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.VariableDef.class, org.tensorflow.proto.framework.VariableDef.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.framework.VariableDef.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      variableName_ = "";
-
-      initialValueName_ = "";
-
-      initializerName_ = "";
-
-      snapshotName_ = "";
-
-      if (saveSliceInfoDefBuilder_ == null) {
-        saveSliceInfoDef_ = null;
-      } else {
-        saveSliceInfoDef_ = null;
-        saveSliceInfoDefBuilder_ = null;
-      }
-      isResource_ = false;
-
-      trainable_ = false;
-
-      synchronization_ = 0;
-
-      aggregation_ = 0;
-
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.framework.VariableProtos.internal_static_tensorflow_VariableDef_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.VariableDef getDefaultInstanceForType() {
-      return org.tensorflow.proto.framework.VariableDef.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.VariableDef build() {
-      org.tensorflow.proto.framework.VariableDef result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.VariableDef buildPartial() {
-      org.tensorflow.proto.framework.VariableDef result = new org.tensorflow.proto.framework.VariableDef(this);
-      result.variableName_ = variableName_;
-      result.initialValueName_ = initialValueName_;
-      result.initializerName_ = initializerName_;
-      result.snapshotName_ = snapshotName_;
-      if (saveSliceInfoDefBuilder_ == null) {
-        result.saveSliceInfoDef_ = saveSliceInfoDef_;
-      } else {
-        result.saveSliceInfoDef_ = saveSliceInfoDefBuilder_.build();
-      }
-      result.isResource_ = isResource_;
-      result.trainable_ = trainable_;
-      result.synchronization_ = synchronization_;
-      result.aggregation_ = aggregation_;
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.framework.VariableDef) {
-        return mergeFrom((org.tensorflow.proto.framework.VariableDef)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.framework.VariableDef other) {
-      if (other == org.tensorflow.proto.framework.VariableDef.getDefaultInstance()) return this;
-      if (!other.getVariableName().isEmpty()) {
-        variableName_ = other.variableName_;
-        onChanged();
-      }
-      if (!other.getInitialValueName().isEmpty()) {
-        initialValueName_ = other.initialValueName_;
-        onChanged();
-      }
-      if (!other.getInitializerName().isEmpty()) {
-        initializerName_ = other.initializerName_;
-        onChanged();
-      }
-      if (!other.getSnapshotName().isEmpty()) {
-        snapshotName_ = other.snapshotName_;
-        onChanged();
-      }
-      if (other.hasSaveSliceInfoDef()) {
-        mergeSaveSliceInfoDef(other.getSaveSliceInfoDef());
-      }
-      if (other.getIsResource() != false) {
-        setIsResource(other.getIsResource());
-      }
-      if (other.getTrainable() != false) {
-        setTrainable(other.getTrainable());
-      }
-      if (other.synchronization_ != 0) {
-        setSynchronizationValue(other.getSynchronizationValue());
-      }
-      if (other.aggregation_ != 0) {
-        setAggregationValue(other.getAggregationValue());
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.framework.VariableDef parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.framework.VariableDef) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-
-    private java.lang.Object variableName_ = "";
-    /**
-     * <pre>
-     * Name of the variable tensor.
-     * </pre>
-     *
-     * <code>string variable_name = 1;</code>
-     */
-    public java.lang.String getVariableName() {
-      java.lang.Object ref = variableName_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        variableName_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Name of the variable tensor.
-     * </pre>
-     *
-     * <code>string variable_name = 1;</code>
-     */
-    public com.google.protobuf.ByteString
-        getVariableNameBytes() {
-      java.lang.Object ref = variableName_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        variableName_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Name of the variable tensor.
-     * </pre>
-     *
-     * <code>string variable_name = 1;</code>
-     */
-    public Builder setVariableName(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      variableName_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Name of the variable tensor.
-     * </pre>
-     *
-     * <code>string variable_name = 1;</code>
-     */
-    public Builder clearVariableName() {
-      
-      variableName_ = getDefaultInstance().getVariableName();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Name of the variable tensor.
-     * </pre>
-     *
-     * <code>string variable_name = 1;</code>
-     */
-    public Builder setVariableNameBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      variableName_ = value;
-      onChanged();
-      return this;
-    }
-
-    private java.lang.Object initialValueName_ = "";
-    /**
-     * <pre>
-     * Name of the tensor holding the variable's initial value.
-     * </pre>
-     *
-     * <code>string initial_value_name = 6;</code>
-     */
-    public java.lang.String getInitialValueName() {
-      java.lang.Object ref = initialValueName_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        initialValueName_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Name of the tensor holding the variable's initial value.
-     * </pre>
-     *
-     * <code>string initial_value_name = 6;</code>
-     */
-    public com.google.protobuf.ByteString
-        getInitialValueNameBytes() {
-      java.lang.Object ref = initialValueName_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        initialValueName_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Name of the tensor holding the variable's initial value.
-     * </pre>
-     *
-     * <code>string initial_value_name = 6;</code>
-     */
-    public Builder setInitialValueName(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      initialValueName_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Name of the tensor holding the variable's initial value.
-     * </pre>
-     *
-     * <code>string initial_value_name = 6;</code>
-     */
-    public Builder clearInitialValueName() {
-      
-      initialValueName_ = getDefaultInstance().getInitialValueName();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Name of the tensor holding the variable's initial value.
-     * </pre>
-     *
-     * <code>string initial_value_name = 6;</code>
-     */
-    public Builder setInitialValueNameBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      initialValueName_ = value;
-      onChanged();
-      return this;
-    }
-
-    private java.lang.Object initializerName_ = "";
-    /**
-     * <pre>
-     * Name of the initializer op.
-     * </pre>
-     *
-     * <code>string initializer_name = 2;</code>
-     */
-    public java.lang.String getInitializerName() {
-      java.lang.Object ref = initializerName_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        initializerName_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Name of the initializer op.
-     * </pre>
-     *
-     * <code>string initializer_name = 2;</code>
-     */
-    public com.google.protobuf.ByteString
-        getInitializerNameBytes() {
-      java.lang.Object ref = initializerName_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        initializerName_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Name of the initializer op.
-     * </pre>
-     *
-     * <code>string initializer_name = 2;</code>
-     */
-    public Builder setInitializerName(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      initializerName_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Name of the initializer op.
-     * </pre>
-     *
-     * <code>string initializer_name = 2;</code>
-     */
-    public Builder clearInitializerName() {
-      
-      initializerName_ = getDefaultInstance().getInitializerName();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Name of the initializer op.
-     * </pre>
-     *
-     * <code>string initializer_name = 2;</code>
-     */
-    public Builder setInitializerNameBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      initializerName_ = value;
-      onChanged();
-      return this;
-    }
-
-    private java.lang.Object snapshotName_ = "";
-    /**
-     * <pre>
-     * Name of the snapshot tensor.
-     * </pre>
-     *
-     * <code>string snapshot_name = 3;</code>
-     */
-    public java.lang.String getSnapshotName() {
-      java.lang.Object ref = snapshotName_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        snapshotName_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Name of the snapshot tensor.
-     * </pre>
-     *
-     * <code>string snapshot_name = 3;</code>
-     */
-    public com.google.protobuf.ByteString
-        getSnapshotNameBytes() {
-      java.lang.Object ref = snapshotName_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        snapshotName_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Name of the snapshot tensor.
-     * </pre>
-     *
-     * <code>string snapshot_name = 3;</code>
-     */
-    public Builder setSnapshotName(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      snapshotName_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Name of the snapshot tensor.
-     * </pre>
-     *
-     * <code>string snapshot_name = 3;</code>
-     */
-    public Builder clearSnapshotName() {
-      
-      snapshotName_ = getDefaultInstance().getSnapshotName();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Name of the snapshot tensor.
-     * </pre>
-     *
-     * <code>string snapshot_name = 3;</code>
-     */
-    public Builder setSnapshotNameBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      snapshotName_ = value;
-      onChanged();
-      return this;
-    }
-
-    private org.tensorflow.proto.framework.SaveSliceInfoDef saveSliceInfoDef_;
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.SaveSliceInfoDef, org.tensorflow.proto.framework.SaveSliceInfoDef.Builder, org.tensorflow.proto.framework.SaveSliceInfoDefOrBuilder> saveSliceInfoDefBuilder_;
-    /**
-     * <pre>
-     * Support for saving variables as slices of a larger variable.
-     * </pre>
-     *
-     * <code>.tensorflow.SaveSliceInfoDef save_slice_info_def = 4;</code>
-     */
-    public boolean hasSaveSliceInfoDef() {
-      return saveSliceInfoDefBuilder_ != null || saveSliceInfoDef_ != null;
-    }
-    /**
-     * <pre>
-     * Support for saving variables as slices of a larger variable.
-     * </pre>
-     *
-     * <code>.tensorflow.SaveSliceInfoDef save_slice_info_def = 4;</code>
-     */
-    public org.tensorflow.proto.framework.SaveSliceInfoDef getSaveSliceInfoDef() {
-      if (saveSliceInfoDefBuilder_ == null) {
-        return saveSliceInfoDef_ == null ? org.tensorflow.proto.framework.SaveSliceInfoDef.getDefaultInstance() : saveSliceInfoDef_;
-      } else {
-        return saveSliceInfoDefBuilder_.getMessage();
-      }
-    }
-    /**
-     * <pre>
-     * Support for saving variables as slices of a larger variable.
-     * </pre>
-     *
-     * <code>.tensorflow.SaveSliceInfoDef save_slice_info_def = 4;</code>
-     */
-    public Builder setSaveSliceInfoDef(org.tensorflow.proto.framework.SaveSliceInfoDef value) {
-      if (saveSliceInfoDefBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        saveSliceInfoDef_ = value;
-        onChanged();
-      } else {
-        saveSliceInfoDefBuilder_.setMessage(value);
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Support for saving variables as slices of a larger variable.
-     * </pre>
-     *
-     * <code>.tensorflow.SaveSliceInfoDef save_slice_info_def = 4;</code>
-     */
-    public Builder setSaveSliceInfoDef(
-        org.tensorflow.proto.framework.SaveSliceInfoDef.Builder builderForValue) {
-      if (saveSliceInfoDefBuilder_ == null) {
-        saveSliceInfoDef_ = builderForValue.build();
-        onChanged();
-      } else {
-        saveSliceInfoDefBuilder_.setMessage(builderForValue.build());
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Support for saving variables as slices of a larger variable.
-     * </pre>
-     *
-     * <code>.tensorflow.SaveSliceInfoDef save_slice_info_def = 4;</code>
-     */
-    public Builder mergeSaveSliceInfoDef(org.tensorflow.proto.framework.SaveSliceInfoDef value) {
-      if (saveSliceInfoDefBuilder_ == null) {
-        if (saveSliceInfoDef_ != null) {
-          saveSliceInfoDef_ =
-            org.tensorflow.proto.framework.SaveSliceInfoDef.newBuilder(saveSliceInfoDef_).mergeFrom(value).buildPartial();
-        } else {
-          saveSliceInfoDef_ = value;
-        }
-        onChanged();
-      } else {
-        saveSliceInfoDefBuilder_.mergeFrom(value);
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Support for saving variables as slices of a larger variable.
-     * </pre>
-     *
-     * <code>.tensorflow.SaveSliceInfoDef save_slice_info_def = 4;</code>
-     */
-    public Builder clearSaveSliceInfoDef() {
-      if (saveSliceInfoDefBuilder_ == null) {
-        saveSliceInfoDef_ = null;
-        onChanged();
-      } else {
-        saveSliceInfoDef_ = null;
-        saveSliceInfoDefBuilder_ = null;
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Support for saving variables as slices of a larger variable.
-     * </pre>
-     *
-     * <code>.tensorflow.SaveSliceInfoDef save_slice_info_def = 4;</code>
-     */
-    public org.tensorflow.proto.framework.SaveSliceInfoDef.Builder getSaveSliceInfoDefBuilder() {
-      
-      onChanged();
-      return getSaveSliceInfoDefFieldBuilder().getBuilder();
-    }
-    /**
-     * <pre>
-     * Support for saving variables as slices of a larger variable.
-     * </pre>
-     *
-     * <code>.tensorflow.SaveSliceInfoDef save_slice_info_def = 4;</code>
-     */
-    public org.tensorflow.proto.framework.SaveSliceInfoDefOrBuilder getSaveSliceInfoDefOrBuilder() {
-      if (saveSliceInfoDefBuilder_ != null) {
-        return saveSliceInfoDefBuilder_.getMessageOrBuilder();
-      } else {
-        return saveSliceInfoDef_ == null ?
-            org.tensorflow.proto.framework.SaveSliceInfoDef.getDefaultInstance() : saveSliceInfoDef_;
-      }
-    }
-    /**
-     * <pre>
-     * Support for saving variables as slices of a larger variable.
-     * </pre>
-     *
-     * <code>.tensorflow.SaveSliceInfoDef save_slice_info_def = 4;</code>
-     */
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.SaveSliceInfoDef, org.tensorflow.proto.framework.SaveSliceInfoDef.Builder, org.tensorflow.proto.framework.SaveSliceInfoDefOrBuilder> 
-        getSaveSliceInfoDefFieldBuilder() {
-      if (saveSliceInfoDefBuilder_ == null) {
-        saveSliceInfoDefBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-            org.tensorflow.proto.framework.SaveSliceInfoDef, org.tensorflow.proto.framework.SaveSliceInfoDef.Builder, org.tensorflow.proto.framework.SaveSliceInfoDefOrBuilder>(
-                getSaveSliceInfoDef(),
-                getParentForChildren(),
-                isClean());
-        saveSliceInfoDef_ = null;
-      }
-      return saveSliceInfoDefBuilder_;
-    }
-
-    private boolean isResource_ ;
-    /**
-     * <pre>
-     * Whether to represent this as a ResourceVariable.
-     * </pre>
-     *
-     * <code>bool is_resource = 5;</code>
-     */
-    public boolean getIsResource() {
-      return isResource_;
-    }
-    /**
-     * <pre>
-     * Whether to represent this as a ResourceVariable.
-     * </pre>
-     *
-     * <code>bool is_resource = 5;</code>
-     */
-    public Builder setIsResource(boolean value) {
-      
-      isResource_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Whether to represent this as a ResourceVariable.
-     * </pre>
-     *
-     * <code>bool is_resource = 5;</code>
-     */
-    public Builder clearIsResource() {
-      
-      isResource_ = false;
-      onChanged();
-      return this;
-    }
-
-    private boolean trainable_ ;
-    /**
-     * <pre>
-     * Whether this variable should be trained.
-     * </pre>
-     *
-     * <code>bool trainable = 7;</code>
-     */
-    public boolean getTrainable() {
-      return trainable_;
-    }
-    /**
-     * <pre>
-     * Whether this variable should be trained.
-     * </pre>
-     *
-     * <code>bool trainable = 7;</code>
-     */
-    public Builder setTrainable(boolean value) {
-      
-      trainable_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Whether this variable should be trained.
-     * </pre>
-     *
-     * <code>bool trainable = 7;</code>
-     */
-    public Builder clearTrainable() {
-      
-      trainable_ = false;
-      onChanged();
-      return this;
-    }
-
-    private int synchronization_ = 0;
-    /**
-     * <pre>
-     * Indicates when a distributed variable will be synced.
-     * </pre>
-     *
-     * <code>.tensorflow.VariableSynchronization synchronization = 8;</code>
-     */
-    public int getSynchronizationValue() {
-      return synchronization_;
-    }
-    /**
-     * <pre>
-     * Indicates when a distributed variable will be synced.
-     * </pre>
-     *
-     * <code>.tensorflow.VariableSynchronization synchronization = 8;</code>
-     */
-    public Builder setSynchronizationValue(int value) {
-      synchronization_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Indicates when a distributed variable will be synced.
-     * </pre>
-     *
-     * <code>.tensorflow.VariableSynchronization synchronization = 8;</code>
-     */
-    public org.tensorflow.proto.framework.VariableSynchronization getSynchronization() {
-      @SuppressWarnings("deprecation")
-      org.tensorflow.proto.framework.VariableSynchronization result = org.tensorflow.proto.framework.VariableSynchronization.valueOf(synchronization_);
-      return result == null ? org.tensorflow.proto.framework.VariableSynchronization.UNRECOGNIZED : result;
-    }
-    /**
-     * <pre>
-     * Indicates when a distributed variable will be synced.
-     * </pre>
-     *
-     * <code>.tensorflow.VariableSynchronization synchronization = 8;</code>
-     */
-    public Builder setSynchronization(org.tensorflow.proto.framework.VariableSynchronization value) {
-      if (value == null) {
-        throw new NullPointerException();
-      }
-      
-      synchronization_ = value.getNumber();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Indicates when a distributed variable will be synced.
-     * </pre>
-     *
-     * <code>.tensorflow.VariableSynchronization synchronization = 8;</code>
-     */
-    public Builder clearSynchronization() {
-      
-      synchronization_ = 0;
-      onChanged();
-      return this;
-    }
-
-    private int aggregation_ = 0;
-    /**
-     * <pre>
-     * Indicates how a distributed variable will be aggregated.
-     * </pre>
-     *
-     * <code>.tensorflow.VariableAggregation aggregation = 9;</code>
-     */
-    public int getAggregationValue() {
-      return aggregation_;
-    }
-    /**
-     * <pre>
-     * Indicates how a distributed variable will be aggregated.
-     * </pre>
-     *
-     * <code>.tensorflow.VariableAggregation aggregation = 9;</code>
-     */
-    public Builder setAggregationValue(int value) {
-      aggregation_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Indicates how a distributed variable will be aggregated.
-     * </pre>
-     *
-     * <code>.tensorflow.VariableAggregation aggregation = 9;</code>
-     */
-    public org.tensorflow.proto.framework.VariableAggregation getAggregation() {
-      @SuppressWarnings("deprecation")
-      org.tensorflow.proto.framework.VariableAggregation result = org.tensorflow.proto.framework.VariableAggregation.valueOf(aggregation_);
-      return result == null ? org.tensorflow.proto.framework.VariableAggregation.UNRECOGNIZED : result;
-    }
-    /**
-     * <pre>
-     * Indicates how a distributed variable will be aggregated.
-     * </pre>
-     *
-     * <code>.tensorflow.VariableAggregation aggregation = 9;</code>
-     */
-    public Builder setAggregation(org.tensorflow.proto.framework.VariableAggregation value) {
-      if (value == null) {
-        throw new NullPointerException();
-      }
-      
-      aggregation_ = value.getNumber();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Indicates how a distributed variable will be aggregated.
-     * </pre>
-     *
-     * <code>.tensorflow.VariableAggregation aggregation = 9;</code>
-     */
-    public Builder clearAggregation() {
-      
-      aggregation_ = 0;
-      onChanged();
-      return this;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.VariableDef)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.VariableDef)
-  private static final org.tensorflow.proto.framework.VariableDef DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.VariableDef();
-  }
-
-  public static org.tensorflow.proto.framework.VariableDef getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<VariableDef>
-      PARSER = new com.google.protobuf.AbstractParser<VariableDef>() {
-    @java.lang.Override
-    public VariableDef parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new VariableDef(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<VariableDef> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<VariableDef> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.framework.VariableDef getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/VariableDefOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/VariableDefOrBuilder.java
deleted file mode 100644
index 31cf5f0207a..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/VariableDefOrBuilder.java
+++ /dev/null
@@ -1,158 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/framework/variable.proto
-
-package org.tensorflow.proto.framework;
-
-public interface VariableDefOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.VariableDef)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <pre>
-   * Name of the variable tensor.
-   * </pre>
-   *
-   * <code>string variable_name = 1;</code>
-   */
-  java.lang.String getVariableName();
-  /**
-   * <pre>
-   * Name of the variable tensor.
-   * </pre>
-   *
-   * <code>string variable_name = 1;</code>
-   */
-  com.google.protobuf.ByteString
-      getVariableNameBytes();
-
-  /**
-   * <pre>
-   * Name of the tensor holding the variable's initial value.
-   * </pre>
-   *
-   * <code>string initial_value_name = 6;</code>
-   */
-  java.lang.String getInitialValueName();
-  /**
-   * <pre>
-   * Name of the tensor holding the variable's initial value.
-   * </pre>
-   *
-   * <code>string initial_value_name = 6;</code>
-   */
-  com.google.protobuf.ByteString
-      getInitialValueNameBytes();
-
-  /**
-   * <pre>
-   * Name of the initializer op.
-   * </pre>
-   *
-   * <code>string initializer_name = 2;</code>
-   */
-  java.lang.String getInitializerName();
-  /**
-   * <pre>
-   * Name of the initializer op.
-   * </pre>
-   *
-   * <code>string initializer_name = 2;</code>
-   */
-  com.google.protobuf.ByteString
-      getInitializerNameBytes();
-
-  /**
-   * <pre>
-   * Name of the snapshot tensor.
-   * </pre>
-   *
-   * <code>string snapshot_name = 3;</code>
-   */
-  java.lang.String getSnapshotName();
-  /**
-   * <pre>
-   * Name of the snapshot tensor.
-   * </pre>
-   *
-   * <code>string snapshot_name = 3;</code>
-   */
-  com.google.protobuf.ByteString
-      getSnapshotNameBytes();
-
-  /**
-   * <pre>
-   * Support for saving variables as slices of a larger variable.
-   * </pre>
-   *
-   * <code>.tensorflow.SaveSliceInfoDef save_slice_info_def = 4;</code>
-   */
-  boolean hasSaveSliceInfoDef();
-  /**
-   * <pre>
-   * Support for saving variables as slices of a larger variable.
-   * </pre>
-   *
-   * <code>.tensorflow.SaveSliceInfoDef save_slice_info_def = 4;</code>
-   */
-  org.tensorflow.proto.framework.SaveSliceInfoDef getSaveSliceInfoDef();
-  /**
-   * <pre>
-   * Support for saving variables as slices of a larger variable.
-   * </pre>
-   *
-   * <code>.tensorflow.SaveSliceInfoDef save_slice_info_def = 4;</code>
-   */
-  org.tensorflow.proto.framework.SaveSliceInfoDefOrBuilder getSaveSliceInfoDefOrBuilder();
-
-  /**
-   * <pre>
-   * Whether to represent this as a ResourceVariable.
-   * </pre>
-   *
-   * <code>bool is_resource = 5;</code>
-   */
-  boolean getIsResource();
-
-  /**
-   * <pre>
-   * Whether this variable should be trained.
-   * </pre>
-   *
-   * <code>bool trainable = 7;</code>
-   */
-  boolean getTrainable();
-
-  /**
-   * <pre>
-   * Indicates when a distributed variable will be synced.
-   * </pre>
-   *
-   * <code>.tensorflow.VariableSynchronization synchronization = 8;</code>
-   */
-  int getSynchronizationValue();
-  /**
-   * <pre>
-   * Indicates when a distributed variable will be synced.
-   * </pre>
-   *
-   * <code>.tensorflow.VariableSynchronization synchronization = 8;</code>
-   */
-  org.tensorflow.proto.framework.VariableSynchronization getSynchronization();
-
-  /**
-   * <pre>
-   * Indicates how a distributed variable will be aggregated.
-   * </pre>
-   *
-   * <code>.tensorflow.VariableAggregation aggregation = 9;</code>
-   */
-  int getAggregationValue();
-  /**
-   * <pre>
-   * Indicates how a distributed variable will be aggregated.
-   * </pre>
-   *
-   * <code>.tensorflow.VariableAggregation aggregation = 9;</code>
-   */
-  org.tensorflow.proto.framework.VariableAggregation getAggregation();
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/VariableProtos.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/VariableProtos.java
deleted file mode 100644
index 1d8c7d85e0a..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/VariableProtos.java
+++ /dev/null
@@ -1,81 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/framework/variable.proto
-
-package org.tensorflow.proto.framework;
-
-public final class VariableProtos {
-  private VariableProtos() {}
-  public static void registerAllExtensions(
-      com.google.protobuf.ExtensionRegistryLite registry) {
-  }
-
-  public static void registerAllExtensions(
-      com.google.protobuf.ExtensionRegistry registry) {
-    registerAllExtensions(
-        (com.google.protobuf.ExtensionRegistryLite) registry);
-  }
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_VariableDef_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_VariableDef_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_SaveSliceInfoDef_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_SaveSliceInfoDef_fieldAccessorTable;
-
-  public static com.google.protobuf.Descriptors.FileDescriptor
-      getDescriptor() {
-    return descriptor;
-  }
-  private static  com.google.protobuf.Descriptors.FileDescriptor
-      descriptor;
-  static {
-    java.lang.String[] descriptorData = {
-      "\n(tensorflow/core/framework/variable.pro" +
-      "to\022\ntensorflow\"\310\002\n\013VariableDef\022\025\n\rvariab" +
-      "le_name\030\001 \001(\t\022\032\n\022initial_value_name\030\006 \001(" +
-      "\t\022\030\n\020initializer_name\030\002 \001(\t\022\025\n\rsnapshot_" +
-      "name\030\003 \001(\t\0229\n\023save_slice_info_def\030\004 \001(\0132" +
-      "\034.tensorflow.SaveSliceInfoDef\022\023\n\013is_reso" +
-      "urce\030\005 \001(\010\022\021\n\ttrainable\030\007 \001(\010\022<\n\017synchro" +
-      "nization\030\010 \001(\0162#.tensorflow.VariableSync" +
-      "hronization\0224\n\013aggregation\030\t \001(\0162\037.tenso" +
-      "rflow.VariableAggregation\"`\n\020SaveSliceIn" +
-      "foDef\022\021\n\tfull_name\030\001 \001(\t\022\022\n\nfull_shape\030\002" +
-      " \003(\003\022\022\n\nvar_offset\030\003 \003(\003\022\021\n\tvar_shape\030\004 " +
-      "\003(\003*\254\001\n\027VariableSynchronization\022!\n\035VARIA" +
-      "BLE_SYNCHRONIZATION_AUTO\020\000\022!\n\035VARIABLE_S" +
-      "YNCHRONIZATION_NONE\020\001\022%\n!VARIABLE_SYNCHR" +
-      "ONIZATION_ON_WRITE\020\002\022$\n VARIABLE_SYNCHRO" +
-      "NIZATION_ON_READ\020\003*\236\001\n\023VariableAggregati" +
-      "on\022\035\n\031VARIABLE_AGGREGATION_NONE\020\000\022\034\n\030VAR" +
-      "IABLE_AGGREGATION_SUM\020\001\022\035\n\031VARIABLE_AGGR" +
-      "EGATION_MEAN\020\002\022+\n\'VARIABLE_AGGREGATION_O" +
-      "NLY_FIRST_REPLICA\020\003B\206\001\n\036org.tensorflow.p" +
-      "roto.frameworkB\016VariableProtosP\001ZOgithub" +
-      ".com/tensorflow/tensorflow/tensorflow/go" +
-      "/core/framework/variable_go_proto\370\001\001b\006pr" +
-      "oto3"
-    };
-    descriptor = com.google.protobuf.Descriptors.FileDescriptor
-      .internalBuildGeneratedFileFrom(descriptorData,
-        new com.google.protobuf.Descriptors.FileDescriptor[] {
-        });
-    internal_static_tensorflow_VariableDef_descriptor =
-      getDescriptor().getMessageTypes().get(0);
-    internal_static_tensorflow_VariableDef_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_VariableDef_descriptor,
-        new java.lang.String[] { "VariableName", "InitialValueName", "InitializerName", "SnapshotName", "SaveSliceInfoDef", "IsResource", "Trainable", "Synchronization", "Aggregation", });
-    internal_static_tensorflow_SaveSliceInfoDef_descriptor =
-      getDescriptor().getMessageTypes().get(1);
-    internal_static_tensorflow_SaveSliceInfoDef_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_SaveSliceInfoDef_descriptor,
-        new java.lang.String[] { "FullName", "FullShape", "VarOffset", "VarShape", });
-  }
-
-  // @@protoc_insertion_point(outer_class_scope)
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/VariableSynchronization.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/VariableSynchronization.java
deleted file mode 100644
index b71b26cdcc1..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/VariableSynchronization.java
+++ /dev/null
@@ -1,169 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/framework/variable.proto
-
-package org.tensorflow.proto.framework;
-
-/**
- * <pre>
- * Indicates when a distributed variable will be synced.
- * </pre>
- *
- * Protobuf enum {@code tensorflow.VariableSynchronization}
- */
-public enum VariableSynchronization
-    implements com.google.protobuf.ProtocolMessageEnum {
-  /**
-   * <pre>
-   * `AUTO`: Indicates that the synchronization will be determined by the
-   * current `DistributionStrategy` (eg. With `MirroredStrategy` this would be
-   * `ON_WRITE`).
-   * </pre>
-   *
-   * <code>VARIABLE_SYNCHRONIZATION_AUTO = 0;</code>
-   */
-  VARIABLE_SYNCHRONIZATION_AUTO(0),
-  /**
-   * <pre>
-   * `NONE`: Indicates that there will only be one copy of the variable, so
-   * there is no need to sync.
-   * </pre>
-   *
-   * <code>VARIABLE_SYNCHRONIZATION_NONE = 1;</code>
-   */
-  VARIABLE_SYNCHRONIZATION_NONE(1),
-  /**
-   * <pre>
-   * `ON_WRITE`: Indicates that the variable will be updated across devices
-   * every time it is written.
-   * </pre>
-   *
-   * <code>VARIABLE_SYNCHRONIZATION_ON_WRITE = 2;</code>
-   */
-  VARIABLE_SYNCHRONIZATION_ON_WRITE(2),
-  /**
-   * <pre>
-   * `ON_READ`: Indicates that the variable will be aggregated across devices
-   * when it is read (eg. when checkpointing or when evaluating an op that uses
-   * the variable).
-   * </pre>
-   *
-   * <code>VARIABLE_SYNCHRONIZATION_ON_READ = 3;</code>
-   */
-  VARIABLE_SYNCHRONIZATION_ON_READ(3),
-  UNRECOGNIZED(-1),
-  ;
-
-  /**
-   * <pre>
-   * `AUTO`: Indicates that the synchronization will be determined by the
-   * current `DistributionStrategy` (eg. With `MirroredStrategy` this would be
-   * `ON_WRITE`).
-   * </pre>
-   *
-   * <code>VARIABLE_SYNCHRONIZATION_AUTO = 0;</code>
-   */
-  public static final int VARIABLE_SYNCHRONIZATION_AUTO_VALUE = 0;
-  /**
-   * <pre>
-   * `NONE`: Indicates that there will only be one copy of the variable, so
-   * there is no need to sync.
-   * </pre>
-   *
-   * <code>VARIABLE_SYNCHRONIZATION_NONE = 1;</code>
-   */
-  public static final int VARIABLE_SYNCHRONIZATION_NONE_VALUE = 1;
-  /**
-   * <pre>
-   * `ON_WRITE`: Indicates that the variable will be updated across devices
-   * every time it is written.
-   * </pre>
-   *
-   * <code>VARIABLE_SYNCHRONIZATION_ON_WRITE = 2;</code>
-   */
-  public static final int VARIABLE_SYNCHRONIZATION_ON_WRITE_VALUE = 2;
-  /**
-   * <pre>
-   * `ON_READ`: Indicates that the variable will be aggregated across devices
-   * when it is read (eg. when checkpointing or when evaluating an op that uses
-   * the variable).
-   * </pre>
-   *
-   * <code>VARIABLE_SYNCHRONIZATION_ON_READ = 3;</code>
-   */
-  public static final int VARIABLE_SYNCHRONIZATION_ON_READ_VALUE = 3;
-
-
-  public final int getNumber() {
-    if (this == UNRECOGNIZED) {
-      throw new java.lang.IllegalArgumentException(
-          "Can't get the number of an unknown enum value.");
-    }
-    return value;
-  }
-
-  /**
-   * @deprecated Use {@link #forNumber(int)} instead.
-   */
-  @java.lang.Deprecated
-  public static VariableSynchronization valueOf(int value) {
-    return forNumber(value);
-  }
-
-  public static VariableSynchronization forNumber(int value) {
-    switch (value) {
-      case 0: return VARIABLE_SYNCHRONIZATION_AUTO;
-      case 1: return VARIABLE_SYNCHRONIZATION_NONE;
-      case 2: return VARIABLE_SYNCHRONIZATION_ON_WRITE;
-      case 3: return VARIABLE_SYNCHRONIZATION_ON_READ;
-      default: return null;
-    }
-  }
-
-  public static com.google.protobuf.Internal.EnumLiteMap<VariableSynchronization>
-      internalGetValueMap() {
-    return internalValueMap;
-  }
-  private static final com.google.protobuf.Internal.EnumLiteMap<
-      VariableSynchronization> internalValueMap =
-        new com.google.protobuf.Internal.EnumLiteMap<VariableSynchronization>() {
-          public VariableSynchronization findValueByNumber(int number) {
-            return VariableSynchronization.forNumber(number);
-          }
-        };
-
-  public final com.google.protobuf.Descriptors.EnumValueDescriptor
-      getValueDescriptor() {
-    return getDescriptor().getValues().get(ordinal());
-  }
-  public final com.google.protobuf.Descriptors.EnumDescriptor
-      getDescriptorForType() {
-    return getDescriptor();
-  }
-  public static final com.google.protobuf.Descriptors.EnumDescriptor
-      getDescriptor() {
-    return org.tensorflow.proto.framework.VariableProtos.getDescriptor().getEnumTypes().get(0);
-  }
-
-  private static final VariableSynchronization[] VALUES = values();
-
-  public static VariableSynchronization valueOf(
-      com.google.protobuf.Descriptors.EnumValueDescriptor desc) {
-    if (desc.getType() != getDescriptor()) {
-      throw new java.lang.IllegalArgumentException(
-        "EnumValueDescriptor is not for this type.");
-    }
-    if (desc.getIndex() == -1) {
-      return UNRECOGNIZED;
-    }
-    return VALUES[desc.getIndex()];
-  }
-
-  private final int value;
-
-  private VariableSynchronization(int value) {
-    this.value = value;
-  }
-
-  // @@protoc_insertion_point(enum_scope:tensorflow.VariableSynchronization)
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/VariantTensorDataProto.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/VariantTensorDataProto.java
deleted file mode 100644
index 47555bc3d7a..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/VariantTensorDataProto.java
+++ /dev/null
@@ -1,1097 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/framework/tensor.proto
-
-package org.tensorflow.proto.framework;
-
-/**
- * <pre>
- * Protocol buffer representing the serialization format of DT_VARIANT tensors.
- * </pre>
- *
- * Protobuf type {@code tensorflow.VariantTensorDataProto}
- */
-public  final class VariantTensorDataProto extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.VariantTensorDataProto)
-    VariantTensorDataProtoOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use VariantTensorDataProto.newBuilder() to construct.
-  private VariantTensorDataProto(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private VariantTensorDataProto() {
-    typeName_ = "";
-    metadata_ = com.google.protobuf.ByteString.EMPTY;
-    tensors_ = java.util.Collections.emptyList();
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new VariantTensorDataProto();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private VariantTensorDataProto(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    int mutable_bitField0_ = 0;
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 10: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            typeName_ = s;
-            break;
-          }
-          case 18: {
-
-            metadata_ = input.readBytes();
-            break;
-          }
-          case 26: {
-            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
-              tensors_ = new java.util.ArrayList<org.tensorflow.proto.framework.TensorProto>();
-              mutable_bitField0_ |= 0x00000001;
-            }
-            tensors_.add(
-                input.readMessage(org.tensorflow.proto.framework.TensorProto.parser(), extensionRegistry));
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      if (((mutable_bitField0_ & 0x00000001) != 0)) {
-        tensors_ = java.util.Collections.unmodifiableList(tensors_);
-      }
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.framework.TensorProtos.internal_static_tensorflow_VariantTensorDataProto_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.framework.TensorProtos.internal_static_tensorflow_VariantTensorDataProto_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.framework.VariantTensorDataProto.class, org.tensorflow.proto.framework.VariantTensorDataProto.Builder.class);
-  }
-
-  public static final int TYPE_NAME_FIELD_NUMBER = 1;
-  private volatile java.lang.Object typeName_;
-  /**
-   * <pre>
-   * Name of the type of objects being serialized.
-   * </pre>
-   *
-   * <code>string type_name = 1;</code>
-   */
-  public java.lang.String getTypeName() {
-    java.lang.Object ref = typeName_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      typeName_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * Name of the type of objects being serialized.
-   * </pre>
-   *
-   * <code>string type_name = 1;</code>
-   */
-  public com.google.protobuf.ByteString
-      getTypeNameBytes() {
-    java.lang.Object ref = typeName_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      typeName_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int METADATA_FIELD_NUMBER = 2;
-  private com.google.protobuf.ByteString metadata_;
-  /**
-   * <pre>
-   * Portions of the object that are not Tensors.
-   * </pre>
-   *
-   * <code>bytes metadata = 2;</code>
-   */
-  public com.google.protobuf.ByteString getMetadata() {
-    return metadata_;
-  }
-
-  public static final int TENSORS_FIELD_NUMBER = 3;
-  private java.util.List<org.tensorflow.proto.framework.TensorProto> tensors_;
-  /**
-   * <pre>
-   * Tensors contained within objects being serialized.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.TensorProto tensors = 3;</code>
-   */
-  public java.util.List<org.tensorflow.proto.framework.TensorProto> getTensorsList() {
-    return tensors_;
-  }
-  /**
-   * <pre>
-   * Tensors contained within objects being serialized.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.TensorProto tensors = 3;</code>
-   */
-  public java.util.List<? extends org.tensorflow.proto.framework.TensorProtoOrBuilder> 
-      getTensorsOrBuilderList() {
-    return tensors_;
-  }
-  /**
-   * <pre>
-   * Tensors contained within objects being serialized.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.TensorProto tensors = 3;</code>
-   */
-  public int getTensorsCount() {
-    return tensors_.size();
-  }
-  /**
-   * <pre>
-   * Tensors contained within objects being serialized.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.TensorProto tensors = 3;</code>
-   */
-  public org.tensorflow.proto.framework.TensorProto getTensors(int index) {
-    return tensors_.get(index);
-  }
-  /**
-   * <pre>
-   * Tensors contained within objects being serialized.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.TensorProto tensors = 3;</code>
-   */
-  public org.tensorflow.proto.framework.TensorProtoOrBuilder getTensorsOrBuilder(
-      int index) {
-    return tensors_.get(index);
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    if (!getTypeNameBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 1, typeName_);
-    }
-    if (!metadata_.isEmpty()) {
-      output.writeBytes(2, metadata_);
-    }
-    for (int i = 0; i < tensors_.size(); i++) {
-      output.writeMessage(3, tensors_.get(i));
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (!getTypeNameBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, typeName_);
-    }
-    if (!metadata_.isEmpty()) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeBytesSize(2, metadata_);
-    }
-    for (int i = 0; i < tensors_.size(); i++) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(3, tensors_.get(i));
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.framework.VariantTensorDataProto)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.framework.VariantTensorDataProto other = (org.tensorflow.proto.framework.VariantTensorDataProto) obj;
-
-    if (!getTypeName()
-        .equals(other.getTypeName())) return false;
-    if (!getMetadata()
-        .equals(other.getMetadata())) return false;
-    if (!getTensorsList()
-        .equals(other.getTensorsList())) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    hash = (37 * hash) + TYPE_NAME_FIELD_NUMBER;
-    hash = (53 * hash) + getTypeName().hashCode();
-    hash = (37 * hash) + METADATA_FIELD_NUMBER;
-    hash = (53 * hash) + getMetadata().hashCode();
-    if (getTensorsCount() > 0) {
-      hash = (37 * hash) + TENSORS_FIELD_NUMBER;
-      hash = (53 * hash) + getTensorsList().hashCode();
-    }
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.framework.VariantTensorDataProto parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.VariantTensorDataProto parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.VariantTensorDataProto parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.VariantTensorDataProto parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.VariantTensorDataProto parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.VariantTensorDataProto parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.VariantTensorDataProto parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.VariantTensorDataProto parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.VariantTensorDataProto parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.VariantTensorDataProto parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.VariantTensorDataProto parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.VariantTensorDataProto parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.framework.VariantTensorDataProto prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * <pre>
-   * Protocol buffer representing the serialization format of DT_VARIANT tensors.
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.VariantTensorDataProto}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.VariantTensorDataProto)
-      org.tensorflow.proto.framework.VariantTensorDataProtoOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.TensorProtos.internal_static_tensorflow_VariantTensorDataProto_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.TensorProtos.internal_static_tensorflow_VariantTensorDataProto_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.VariantTensorDataProto.class, org.tensorflow.proto.framework.VariantTensorDataProto.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.framework.VariantTensorDataProto.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-        getTensorsFieldBuilder();
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      typeName_ = "";
-
-      metadata_ = com.google.protobuf.ByteString.EMPTY;
-
-      if (tensorsBuilder_ == null) {
-        tensors_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000001);
-      } else {
-        tensorsBuilder_.clear();
-      }
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.framework.TensorProtos.internal_static_tensorflow_VariantTensorDataProto_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.VariantTensorDataProto getDefaultInstanceForType() {
-      return org.tensorflow.proto.framework.VariantTensorDataProto.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.VariantTensorDataProto build() {
-      org.tensorflow.proto.framework.VariantTensorDataProto result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.VariantTensorDataProto buildPartial() {
-      org.tensorflow.proto.framework.VariantTensorDataProto result = new org.tensorflow.proto.framework.VariantTensorDataProto(this);
-      int from_bitField0_ = bitField0_;
-      result.typeName_ = typeName_;
-      result.metadata_ = metadata_;
-      if (tensorsBuilder_ == null) {
-        if (((bitField0_ & 0x00000001) != 0)) {
-          tensors_ = java.util.Collections.unmodifiableList(tensors_);
-          bitField0_ = (bitField0_ & ~0x00000001);
-        }
-        result.tensors_ = tensors_;
-      } else {
-        result.tensors_ = tensorsBuilder_.build();
-      }
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.framework.VariantTensorDataProto) {
-        return mergeFrom((org.tensorflow.proto.framework.VariantTensorDataProto)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.framework.VariantTensorDataProto other) {
-      if (other == org.tensorflow.proto.framework.VariantTensorDataProto.getDefaultInstance()) return this;
-      if (!other.getTypeName().isEmpty()) {
-        typeName_ = other.typeName_;
-        onChanged();
-      }
-      if (other.getMetadata() != com.google.protobuf.ByteString.EMPTY) {
-        setMetadata(other.getMetadata());
-      }
-      if (tensorsBuilder_ == null) {
-        if (!other.tensors_.isEmpty()) {
-          if (tensors_.isEmpty()) {
-            tensors_ = other.tensors_;
-            bitField0_ = (bitField0_ & ~0x00000001);
-          } else {
-            ensureTensorsIsMutable();
-            tensors_.addAll(other.tensors_);
-          }
-          onChanged();
-        }
-      } else {
-        if (!other.tensors_.isEmpty()) {
-          if (tensorsBuilder_.isEmpty()) {
-            tensorsBuilder_.dispose();
-            tensorsBuilder_ = null;
-            tensors_ = other.tensors_;
-            bitField0_ = (bitField0_ & ~0x00000001);
-            tensorsBuilder_ = 
-              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
-                 getTensorsFieldBuilder() : null;
-          } else {
-            tensorsBuilder_.addAllMessages(other.tensors_);
-          }
-        }
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.framework.VariantTensorDataProto parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.framework.VariantTensorDataProto) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-    private int bitField0_;
-
-    private java.lang.Object typeName_ = "";
-    /**
-     * <pre>
-     * Name of the type of objects being serialized.
-     * </pre>
-     *
-     * <code>string type_name = 1;</code>
-     */
-    public java.lang.String getTypeName() {
-      java.lang.Object ref = typeName_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        typeName_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Name of the type of objects being serialized.
-     * </pre>
-     *
-     * <code>string type_name = 1;</code>
-     */
-    public com.google.protobuf.ByteString
-        getTypeNameBytes() {
-      java.lang.Object ref = typeName_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        typeName_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Name of the type of objects being serialized.
-     * </pre>
-     *
-     * <code>string type_name = 1;</code>
-     */
-    public Builder setTypeName(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      typeName_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Name of the type of objects being serialized.
-     * </pre>
-     *
-     * <code>string type_name = 1;</code>
-     */
-    public Builder clearTypeName() {
-      
-      typeName_ = getDefaultInstance().getTypeName();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Name of the type of objects being serialized.
-     * </pre>
-     *
-     * <code>string type_name = 1;</code>
-     */
-    public Builder setTypeNameBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      typeName_ = value;
-      onChanged();
-      return this;
-    }
-
-    private com.google.protobuf.ByteString metadata_ = com.google.protobuf.ByteString.EMPTY;
-    /**
-     * <pre>
-     * Portions of the object that are not Tensors.
-     * </pre>
-     *
-     * <code>bytes metadata = 2;</code>
-     */
-    public com.google.protobuf.ByteString getMetadata() {
-      return metadata_;
-    }
-    /**
-     * <pre>
-     * Portions of the object that are not Tensors.
-     * </pre>
-     *
-     * <code>bytes metadata = 2;</code>
-     */
-    public Builder setMetadata(com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      metadata_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Portions of the object that are not Tensors.
-     * </pre>
-     *
-     * <code>bytes metadata = 2;</code>
-     */
-    public Builder clearMetadata() {
-      
-      metadata_ = getDefaultInstance().getMetadata();
-      onChanged();
-      return this;
-    }
-
-    private java.util.List<org.tensorflow.proto.framework.TensorProto> tensors_ =
-      java.util.Collections.emptyList();
-    private void ensureTensorsIsMutable() {
-      if (!((bitField0_ & 0x00000001) != 0)) {
-        tensors_ = new java.util.ArrayList<org.tensorflow.proto.framework.TensorProto>(tensors_);
-        bitField0_ |= 0x00000001;
-       }
-    }
-
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.framework.TensorProto, org.tensorflow.proto.framework.TensorProto.Builder, org.tensorflow.proto.framework.TensorProtoOrBuilder> tensorsBuilder_;
-
-    /**
-     * <pre>
-     * Tensors contained within objects being serialized.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorProto tensors = 3;</code>
-     */
-    public java.util.List<org.tensorflow.proto.framework.TensorProto> getTensorsList() {
-      if (tensorsBuilder_ == null) {
-        return java.util.Collections.unmodifiableList(tensors_);
-      } else {
-        return tensorsBuilder_.getMessageList();
-      }
-    }
-    /**
-     * <pre>
-     * Tensors contained within objects being serialized.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorProto tensors = 3;</code>
-     */
-    public int getTensorsCount() {
-      if (tensorsBuilder_ == null) {
-        return tensors_.size();
-      } else {
-        return tensorsBuilder_.getCount();
-      }
-    }
-    /**
-     * <pre>
-     * Tensors contained within objects being serialized.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorProto tensors = 3;</code>
-     */
-    public org.tensorflow.proto.framework.TensorProto getTensors(int index) {
-      if (tensorsBuilder_ == null) {
-        return tensors_.get(index);
-      } else {
-        return tensorsBuilder_.getMessage(index);
-      }
-    }
-    /**
-     * <pre>
-     * Tensors contained within objects being serialized.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorProto tensors = 3;</code>
-     */
-    public Builder setTensors(
-        int index, org.tensorflow.proto.framework.TensorProto value) {
-      if (tensorsBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureTensorsIsMutable();
-        tensors_.set(index, value);
-        onChanged();
-      } else {
-        tensorsBuilder_.setMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Tensors contained within objects being serialized.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorProto tensors = 3;</code>
-     */
-    public Builder setTensors(
-        int index, org.tensorflow.proto.framework.TensorProto.Builder builderForValue) {
-      if (tensorsBuilder_ == null) {
-        ensureTensorsIsMutable();
-        tensors_.set(index, builderForValue.build());
-        onChanged();
-      } else {
-        tensorsBuilder_.setMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Tensors contained within objects being serialized.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorProto tensors = 3;</code>
-     */
-    public Builder addTensors(org.tensorflow.proto.framework.TensorProto value) {
-      if (tensorsBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureTensorsIsMutable();
-        tensors_.add(value);
-        onChanged();
-      } else {
-        tensorsBuilder_.addMessage(value);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Tensors contained within objects being serialized.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorProto tensors = 3;</code>
-     */
-    public Builder addTensors(
-        int index, org.tensorflow.proto.framework.TensorProto value) {
-      if (tensorsBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureTensorsIsMutable();
-        tensors_.add(index, value);
-        onChanged();
-      } else {
-        tensorsBuilder_.addMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Tensors contained within objects being serialized.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorProto tensors = 3;</code>
-     */
-    public Builder addTensors(
-        org.tensorflow.proto.framework.TensorProto.Builder builderForValue) {
-      if (tensorsBuilder_ == null) {
-        ensureTensorsIsMutable();
-        tensors_.add(builderForValue.build());
-        onChanged();
-      } else {
-        tensorsBuilder_.addMessage(builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Tensors contained within objects being serialized.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorProto tensors = 3;</code>
-     */
-    public Builder addTensors(
-        int index, org.tensorflow.proto.framework.TensorProto.Builder builderForValue) {
-      if (tensorsBuilder_ == null) {
-        ensureTensorsIsMutable();
-        tensors_.add(index, builderForValue.build());
-        onChanged();
-      } else {
-        tensorsBuilder_.addMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Tensors contained within objects being serialized.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorProto tensors = 3;</code>
-     */
-    public Builder addAllTensors(
-        java.lang.Iterable<? extends org.tensorflow.proto.framework.TensorProto> values) {
-      if (tensorsBuilder_ == null) {
-        ensureTensorsIsMutable();
-        com.google.protobuf.AbstractMessageLite.Builder.addAll(
-            values, tensors_);
-        onChanged();
-      } else {
-        tensorsBuilder_.addAllMessages(values);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Tensors contained within objects being serialized.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorProto tensors = 3;</code>
-     */
-    public Builder clearTensors() {
-      if (tensorsBuilder_ == null) {
-        tensors_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000001);
-        onChanged();
-      } else {
-        tensorsBuilder_.clear();
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Tensors contained within objects being serialized.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorProto tensors = 3;</code>
-     */
-    public Builder removeTensors(int index) {
-      if (tensorsBuilder_ == null) {
-        ensureTensorsIsMutable();
-        tensors_.remove(index);
-        onChanged();
-      } else {
-        tensorsBuilder_.remove(index);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Tensors contained within objects being serialized.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorProto tensors = 3;</code>
-     */
-    public org.tensorflow.proto.framework.TensorProto.Builder getTensorsBuilder(
-        int index) {
-      return getTensorsFieldBuilder().getBuilder(index);
-    }
-    /**
-     * <pre>
-     * Tensors contained within objects being serialized.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorProto tensors = 3;</code>
-     */
-    public org.tensorflow.proto.framework.TensorProtoOrBuilder getTensorsOrBuilder(
-        int index) {
-      if (tensorsBuilder_ == null) {
-        return tensors_.get(index);  } else {
-        return tensorsBuilder_.getMessageOrBuilder(index);
-      }
-    }
-    /**
-     * <pre>
-     * Tensors contained within objects being serialized.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorProto tensors = 3;</code>
-     */
-    public java.util.List<? extends org.tensorflow.proto.framework.TensorProtoOrBuilder> 
-         getTensorsOrBuilderList() {
-      if (tensorsBuilder_ != null) {
-        return tensorsBuilder_.getMessageOrBuilderList();
-      } else {
-        return java.util.Collections.unmodifiableList(tensors_);
-      }
-    }
-    /**
-     * <pre>
-     * Tensors contained within objects being serialized.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorProto tensors = 3;</code>
-     */
-    public org.tensorflow.proto.framework.TensorProto.Builder addTensorsBuilder() {
-      return getTensorsFieldBuilder().addBuilder(
-          org.tensorflow.proto.framework.TensorProto.getDefaultInstance());
-    }
-    /**
-     * <pre>
-     * Tensors contained within objects being serialized.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorProto tensors = 3;</code>
-     */
-    public org.tensorflow.proto.framework.TensorProto.Builder addTensorsBuilder(
-        int index) {
-      return getTensorsFieldBuilder().addBuilder(
-          index, org.tensorflow.proto.framework.TensorProto.getDefaultInstance());
-    }
-    /**
-     * <pre>
-     * Tensors contained within objects being serialized.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorProto tensors = 3;</code>
-     */
-    public java.util.List<org.tensorflow.proto.framework.TensorProto.Builder> 
-         getTensorsBuilderList() {
-      return getTensorsFieldBuilder().getBuilderList();
-    }
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.framework.TensorProto, org.tensorflow.proto.framework.TensorProto.Builder, org.tensorflow.proto.framework.TensorProtoOrBuilder> 
-        getTensorsFieldBuilder() {
-      if (tensorsBuilder_ == null) {
-        tensorsBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
-            org.tensorflow.proto.framework.TensorProto, org.tensorflow.proto.framework.TensorProto.Builder, org.tensorflow.proto.framework.TensorProtoOrBuilder>(
-                tensors_,
-                ((bitField0_ & 0x00000001) != 0),
-                getParentForChildren(),
-                isClean());
-        tensors_ = null;
-      }
-      return tensorsBuilder_;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.VariantTensorDataProto)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.VariantTensorDataProto)
-  private static final org.tensorflow.proto.framework.VariantTensorDataProto DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.VariantTensorDataProto();
-  }
-
-  public static org.tensorflow.proto.framework.VariantTensorDataProto getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<VariantTensorDataProto>
-      PARSER = new com.google.protobuf.AbstractParser<VariantTensorDataProto>() {
-    @java.lang.Override
-    public VariantTensorDataProto parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new VariantTensorDataProto(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<VariantTensorDataProto> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<VariantTensorDataProto> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.framework.VariantTensorDataProto getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/VariantTensorDataProtoOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/VariantTensorDataProtoOrBuilder.java
deleted file mode 100644
index 916d583c328..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/VariantTensorDataProtoOrBuilder.java
+++ /dev/null
@@ -1,80 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/framework/tensor.proto
-
-package org.tensorflow.proto.framework;
-
-public interface VariantTensorDataProtoOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.VariantTensorDataProto)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <pre>
-   * Name of the type of objects being serialized.
-   * </pre>
-   *
-   * <code>string type_name = 1;</code>
-   */
-  java.lang.String getTypeName();
-  /**
-   * <pre>
-   * Name of the type of objects being serialized.
-   * </pre>
-   *
-   * <code>string type_name = 1;</code>
-   */
-  com.google.protobuf.ByteString
-      getTypeNameBytes();
-
-  /**
-   * <pre>
-   * Portions of the object that are not Tensors.
-   * </pre>
-   *
-   * <code>bytes metadata = 2;</code>
-   */
-  com.google.protobuf.ByteString getMetadata();
-
-  /**
-   * <pre>
-   * Tensors contained within objects being serialized.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.TensorProto tensors = 3;</code>
-   */
-  java.util.List<org.tensorflow.proto.framework.TensorProto> 
-      getTensorsList();
-  /**
-   * <pre>
-   * Tensors contained within objects being serialized.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.TensorProto tensors = 3;</code>
-   */
-  org.tensorflow.proto.framework.TensorProto getTensors(int index);
-  /**
-   * <pre>
-   * Tensors contained within objects being serialized.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.TensorProto tensors = 3;</code>
-   */
-  int getTensorsCount();
-  /**
-   * <pre>
-   * Tensors contained within objects being serialized.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.TensorProto tensors = 3;</code>
-   */
-  java.util.List<? extends org.tensorflow.proto.framework.TensorProtoOrBuilder> 
-      getTensorsOrBuilderList();
-  /**
-   * <pre>
-   * Tensors contained within objects being serialized.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.TensorProto tensors = 3;</code>
-   */
-  org.tensorflow.proto.framework.TensorProtoOrBuilder getTensorsOrBuilder(
-      int index);
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/VerifierConfig.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/VerifierConfig.java
deleted file mode 100644
index c6099d9d2df..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/VerifierConfig.java
+++ /dev/null
@@ -1,725 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/verifier_config.proto
-
-package org.tensorflow.proto.framework;
-
-/**
- * <pre>
- * The config for graph verifiers.
- * </pre>
- *
- * Protobuf type {@code tensorflow.VerifierConfig}
- */
-public  final class VerifierConfig extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.VerifierConfig)
-    VerifierConfigOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use VerifierConfig.newBuilder() to construct.
-  private VerifierConfig(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private VerifierConfig() {
-    structureVerifier_ = 0;
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new VerifierConfig();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private VerifierConfig(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 8: {
-
-            verificationTimeoutInMs_ = input.readInt64();
-            break;
-          }
-          case 16: {
-            int rawValue = input.readEnum();
-
-            structureVerifier_ = rawValue;
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.framework.VerifierConfigProtos.internal_static_tensorflow_VerifierConfig_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.framework.VerifierConfigProtos.internal_static_tensorflow_VerifierConfig_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.framework.VerifierConfig.class, org.tensorflow.proto.framework.VerifierConfig.Builder.class);
-  }
-
-  /**
-   * Protobuf enum {@code tensorflow.VerifierConfig.Toggle}
-   */
-  public enum Toggle
-      implements com.google.protobuf.ProtocolMessageEnum {
-    /**
-     * <code>DEFAULT = 0;</code>
-     */
-    DEFAULT(0),
-    /**
-     * <code>ON = 1;</code>
-     */
-    ON(1),
-    /**
-     * <code>OFF = 2;</code>
-     */
-    OFF(2),
-    UNRECOGNIZED(-1),
-    ;
-
-    /**
-     * <code>DEFAULT = 0;</code>
-     */
-    public static final int DEFAULT_VALUE = 0;
-    /**
-     * <code>ON = 1;</code>
-     */
-    public static final int ON_VALUE = 1;
-    /**
-     * <code>OFF = 2;</code>
-     */
-    public static final int OFF_VALUE = 2;
-
-
-    public final int getNumber() {
-      if (this == UNRECOGNIZED) {
-        throw new java.lang.IllegalArgumentException(
-            "Can't get the number of an unknown enum value.");
-      }
-      return value;
-    }
-
-    /**
-     * @deprecated Use {@link #forNumber(int)} instead.
-     */
-    @java.lang.Deprecated
-    public static Toggle valueOf(int value) {
-      return forNumber(value);
-    }
-
-    public static Toggle forNumber(int value) {
-      switch (value) {
-        case 0: return DEFAULT;
-        case 1: return ON;
-        case 2: return OFF;
-        default: return null;
-      }
-    }
-
-    public static com.google.protobuf.Internal.EnumLiteMap<Toggle>
-        internalGetValueMap() {
-      return internalValueMap;
-    }
-    private static final com.google.protobuf.Internal.EnumLiteMap<
-        Toggle> internalValueMap =
-          new com.google.protobuf.Internal.EnumLiteMap<Toggle>() {
-            public Toggle findValueByNumber(int number) {
-              return Toggle.forNumber(number);
-            }
-          };
-
-    public final com.google.protobuf.Descriptors.EnumValueDescriptor
-        getValueDescriptor() {
-      return getDescriptor().getValues().get(ordinal());
-    }
-    public final com.google.protobuf.Descriptors.EnumDescriptor
-        getDescriptorForType() {
-      return getDescriptor();
-    }
-    public static final com.google.protobuf.Descriptors.EnumDescriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.VerifierConfig.getDescriptor().getEnumTypes().get(0);
-    }
-
-    private static final Toggle[] VALUES = values();
-
-    public static Toggle valueOf(
-        com.google.protobuf.Descriptors.EnumValueDescriptor desc) {
-      if (desc.getType() != getDescriptor()) {
-        throw new java.lang.IllegalArgumentException(
-          "EnumValueDescriptor is not for this type.");
-      }
-      if (desc.getIndex() == -1) {
-        return UNRECOGNIZED;
-      }
-      return VALUES[desc.getIndex()];
-    }
-
-    private final int value;
-
-    private Toggle(int value) {
-      this.value = value;
-    }
-
-    // @@protoc_insertion_point(enum_scope:tensorflow.VerifierConfig.Toggle)
-  }
-
-  public static final int VERIFICATION_TIMEOUT_IN_MS_FIELD_NUMBER = 1;
-  private long verificationTimeoutInMs_;
-  /**
-   * <pre>
-   * Deadline for completion of all verification i.e. all the Toggle ON
-   * verifiers must complete execution within this time.
-   * </pre>
-   *
-   * <code>int64 verification_timeout_in_ms = 1;</code>
-   */
-  public long getVerificationTimeoutInMs() {
-    return verificationTimeoutInMs_;
-  }
-
-  public static final int STRUCTURE_VERIFIER_FIELD_NUMBER = 2;
-  private int structureVerifier_;
-  /**
-   * <pre>
-   * Perform structural validation on a tensorflow graph. Default is OFF.
-   * </pre>
-   *
-   * <code>.tensorflow.VerifierConfig.Toggle structure_verifier = 2;</code>
-   */
-  public int getStructureVerifierValue() {
-    return structureVerifier_;
-  }
-  /**
-   * <pre>
-   * Perform structural validation on a tensorflow graph. Default is OFF.
-   * </pre>
-   *
-   * <code>.tensorflow.VerifierConfig.Toggle structure_verifier = 2;</code>
-   */
-  public org.tensorflow.proto.framework.VerifierConfig.Toggle getStructureVerifier() {
-    @SuppressWarnings("deprecation")
-    org.tensorflow.proto.framework.VerifierConfig.Toggle result = org.tensorflow.proto.framework.VerifierConfig.Toggle.valueOf(structureVerifier_);
-    return result == null ? org.tensorflow.proto.framework.VerifierConfig.Toggle.UNRECOGNIZED : result;
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    if (verificationTimeoutInMs_ != 0L) {
-      output.writeInt64(1, verificationTimeoutInMs_);
-    }
-    if (structureVerifier_ != org.tensorflow.proto.framework.VerifierConfig.Toggle.DEFAULT.getNumber()) {
-      output.writeEnum(2, structureVerifier_);
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (verificationTimeoutInMs_ != 0L) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt64Size(1, verificationTimeoutInMs_);
-    }
-    if (structureVerifier_ != org.tensorflow.proto.framework.VerifierConfig.Toggle.DEFAULT.getNumber()) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeEnumSize(2, structureVerifier_);
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.framework.VerifierConfig)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.framework.VerifierConfig other = (org.tensorflow.proto.framework.VerifierConfig) obj;
-
-    if (getVerificationTimeoutInMs()
-        != other.getVerificationTimeoutInMs()) return false;
-    if (structureVerifier_ != other.structureVerifier_) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    hash = (37 * hash) + VERIFICATION_TIMEOUT_IN_MS_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-        getVerificationTimeoutInMs());
-    hash = (37 * hash) + STRUCTURE_VERIFIER_FIELD_NUMBER;
-    hash = (53 * hash) + structureVerifier_;
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.framework.VerifierConfig parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.VerifierConfig parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.VerifierConfig parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.VerifierConfig parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.VerifierConfig parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.VerifierConfig parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.VerifierConfig parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.VerifierConfig parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.VerifierConfig parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.VerifierConfig parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.VerifierConfig parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.VerifierConfig parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.framework.VerifierConfig prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * <pre>
-   * The config for graph verifiers.
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.VerifierConfig}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.VerifierConfig)
-      org.tensorflow.proto.framework.VerifierConfigOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.VerifierConfigProtos.internal_static_tensorflow_VerifierConfig_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.VerifierConfigProtos.internal_static_tensorflow_VerifierConfig_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.VerifierConfig.class, org.tensorflow.proto.framework.VerifierConfig.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.framework.VerifierConfig.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      verificationTimeoutInMs_ = 0L;
-
-      structureVerifier_ = 0;
-
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.framework.VerifierConfigProtos.internal_static_tensorflow_VerifierConfig_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.VerifierConfig getDefaultInstanceForType() {
-      return org.tensorflow.proto.framework.VerifierConfig.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.VerifierConfig build() {
-      org.tensorflow.proto.framework.VerifierConfig result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.VerifierConfig buildPartial() {
-      org.tensorflow.proto.framework.VerifierConfig result = new org.tensorflow.proto.framework.VerifierConfig(this);
-      result.verificationTimeoutInMs_ = verificationTimeoutInMs_;
-      result.structureVerifier_ = structureVerifier_;
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.framework.VerifierConfig) {
-        return mergeFrom((org.tensorflow.proto.framework.VerifierConfig)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.framework.VerifierConfig other) {
-      if (other == org.tensorflow.proto.framework.VerifierConfig.getDefaultInstance()) return this;
-      if (other.getVerificationTimeoutInMs() != 0L) {
-        setVerificationTimeoutInMs(other.getVerificationTimeoutInMs());
-      }
-      if (other.structureVerifier_ != 0) {
-        setStructureVerifierValue(other.getStructureVerifierValue());
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.framework.VerifierConfig parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.framework.VerifierConfig) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-
-    private long verificationTimeoutInMs_ ;
-    /**
-     * <pre>
-     * Deadline for completion of all verification i.e. all the Toggle ON
-     * verifiers must complete execution within this time.
-     * </pre>
-     *
-     * <code>int64 verification_timeout_in_ms = 1;</code>
-     */
-    public long getVerificationTimeoutInMs() {
-      return verificationTimeoutInMs_;
-    }
-    /**
-     * <pre>
-     * Deadline for completion of all verification i.e. all the Toggle ON
-     * verifiers must complete execution within this time.
-     * </pre>
-     *
-     * <code>int64 verification_timeout_in_ms = 1;</code>
-     */
-    public Builder setVerificationTimeoutInMs(long value) {
-      
-      verificationTimeoutInMs_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Deadline for completion of all verification i.e. all the Toggle ON
-     * verifiers must complete execution within this time.
-     * </pre>
-     *
-     * <code>int64 verification_timeout_in_ms = 1;</code>
-     */
-    public Builder clearVerificationTimeoutInMs() {
-      
-      verificationTimeoutInMs_ = 0L;
-      onChanged();
-      return this;
-    }
-
-    private int structureVerifier_ = 0;
-    /**
-     * <pre>
-     * Perform structural validation on a tensorflow graph. Default is OFF.
-     * </pre>
-     *
-     * <code>.tensorflow.VerifierConfig.Toggle structure_verifier = 2;</code>
-     */
-    public int getStructureVerifierValue() {
-      return structureVerifier_;
-    }
-    /**
-     * <pre>
-     * Perform structural validation on a tensorflow graph. Default is OFF.
-     * </pre>
-     *
-     * <code>.tensorflow.VerifierConfig.Toggle structure_verifier = 2;</code>
-     */
-    public Builder setStructureVerifierValue(int value) {
-      structureVerifier_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Perform structural validation on a tensorflow graph. Default is OFF.
-     * </pre>
-     *
-     * <code>.tensorflow.VerifierConfig.Toggle structure_verifier = 2;</code>
-     */
-    public org.tensorflow.proto.framework.VerifierConfig.Toggle getStructureVerifier() {
-      @SuppressWarnings("deprecation")
-      org.tensorflow.proto.framework.VerifierConfig.Toggle result = org.tensorflow.proto.framework.VerifierConfig.Toggle.valueOf(structureVerifier_);
-      return result == null ? org.tensorflow.proto.framework.VerifierConfig.Toggle.UNRECOGNIZED : result;
-    }
-    /**
-     * <pre>
-     * Perform structural validation on a tensorflow graph. Default is OFF.
-     * </pre>
-     *
-     * <code>.tensorflow.VerifierConfig.Toggle structure_verifier = 2;</code>
-     */
-    public Builder setStructureVerifier(org.tensorflow.proto.framework.VerifierConfig.Toggle value) {
-      if (value == null) {
-        throw new NullPointerException();
-      }
-      
-      structureVerifier_ = value.getNumber();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Perform structural validation on a tensorflow graph. Default is OFF.
-     * </pre>
-     *
-     * <code>.tensorflow.VerifierConfig.Toggle structure_verifier = 2;</code>
-     */
-    public Builder clearStructureVerifier() {
-      
-      structureVerifier_ = 0;
-      onChanged();
-      return this;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.VerifierConfig)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.VerifierConfig)
-  private static final org.tensorflow.proto.framework.VerifierConfig DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.VerifierConfig();
-  }
-
-  public static org.tensorflow.proto.framework.VerifierConfig getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<VerifierConfig>
-      PARSER = new com.google.protobuf.AbstractParser<VerifierConfig>() {
-    @java.lang.Override
-    public VerifierConfig parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new VerifierConfig(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<VerifierConfig> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<VerifierConfig> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.framework.VerifierConfig getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/VerifierConfigOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/VerifierConfigOrBuilder.java
deleted file mode 100644
index 230a728e05f..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/VerifierConfigOrBuilder.java
+++ /dev/null
@@ -1,36 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/verifier_config.proto
-
-package org.tensorflow.proto.framework;
-
-public interface VerifierConfigOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.VerifierConfig)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <pre>
-   * Deadline for completion of all verification i.e. all the Toggle ON
-   * verifiers must complete execution within this time.
-   * </pre>
-   *
-   * <code>int64 verification_timeout_in_ms = 1;</code>
-   */
-  long getVerificationTimeoutInMs();
-
-  /**
-   * <pre>
-   * Perform structural validation on a tensorflow graph. Default is OFF.
-   * </pre>
-   *
-   * <code>.tensorflow.VerifierConfig.Toggle structure_verifier = 2;</code>
-   */
-  int getStructureVerifierValue();
-  /**
-   * <pre>
-   * Perform structural validation on a tensorflow graph. Default is OFF.
-   * </pre>
-   *
-   * <code>.tensorflow.VerifierConfig.Toggle structure_verifier = 2;</code>
-   */
-  org.tensorflow.proto.framework.VerifierConfig.Toggle getStructureVerifier();
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/VerifierConfigProtos.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/VerifierConfigProtos.java
deleted file mode 100644
index 45b39000486..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/VerifierConfigProtos.java
+++ /dev/null
@@ -1,55 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/verifier_config.proto
-
-package org.tensorflow.proto.framework;
-
-public final class VerifierConfigProtos {
-  private VerifierConfigProtos() {}
-  public static void registerAllExtensions(
-      com.google.protobuf.ExtensionRegistryLite registry) {
-  }
-
-  public static void registerAllExtensions(
-      com.google.protobuf.ExtensionRegistry registry) {
-    registerAllExtensions(
-        (com.google.protobuf.ExtensionRegistryLite) registry);
-  }
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_VerifierConfig_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_VerifierConfig_fieldAccessorTable;
-
-  public static com.google.protobuf.Descriptors.FileDescriptor
-      getDescriptor() {
-    return descriptor;
-  }
-  private static  com.google.protobuf.Descriptors.FileDescriptor
-      descriptor;
-  static {
-    java.lang.String[] descriptorData = {
-      "\n.tensorflow/core/protobuf/verifier_conf" +
-      "ig.proto\022\ntensorflow\"\233\001\n\016VerifierConfig\022" +
-      "\"\n\032verification_timeout_in_ms\030\001 \001(\003\022=\n\022s" +
-      "tructure_verifier\030\002 \001(\0162!.tensorflow.Ver" +
-      "ifierConfig.Toggle\"&\n\006Toggle\022\013\n\007DEFAULT\020" +
-      "\000\022\006\n\002ON\020\001\022\007\n\003OFF\020\002B\205\001\n\036org.tensorflow.pr" +
-      "oto.frameworkB\024VerifierConfigProtosP\001ZHg" +
-      "ithub.com/tensorflow/tensorflow/tensorfl" +
-      "ow/go/core/core_protos_go_proto\370\001\001b\006prot" +
-      "o3"
-    };
-    descriptor = com.google.protobuf.Descriptors.FileDescriptor
-      .internalBuildGeneratedFileFrom(descriptorData,
-        new com.google.protobuf.Descriptors.FileDescriptor[] {
-        });
-    internal_static_tensorflow_VerifierConfig_descriptor =
-      getDescriptor().getMessageTypes().get(0);
-    internal_static_tensorflow_VerifierConfig_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_VerifierConfig_descriptor,
-        new java.lang.String[] { "VerificationTimeoutInMs", "StructureVerifier", });
-  }
-
-  // @@protoc_insertion_point(outer_class_scope)
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/VersionDef.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/VersionDef.java
deleted file mode 100644
index 6e83b2aa792..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/VersionDef.java
+++ /dev/null
@@ -1,792 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/framework/versions.proto
-
-package org.tensorflow.proto.framework;
-
-/**
- * <pre>
- * Version information for a piece of serialized data
- * There are different types of versions for each type of data
- * (GraphDef, etc.), but they all have the same common shape
- * described here.
- * Each consumer has "consumer" and "min_producer" versions (specified
- * elsewhere).  A consumer is allowed to consume this data if
- *   producer &gt;= min_producer
- *   consumer &gt;= min_consumer
- *   consumer not in bad_consumers
- * </pre>
- *
- * Protobuf type {@code tensorflow.VersionDef}
- */
-public  final class VersionDef extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.VersionDef)
-    VersionDefOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use VersionDef.newBuilder() to construct.
-  private VersionDef(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private VersionDef() {
-    badConsumers_ = emptyIntList();
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new VersionDef();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private VersionDef(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    int mutable_bitField0_ = 0;
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 8: {
-
-            producer_ = input.readInt32();
-            break;
-          }
-          case 16: {
-
-            minConsumer_ = input.readInt32();
-            break;
-          }
-          case 24: {
-            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
-              badConsumers_ = newIntList();
-              mutable_bitField0_ |= 0x00000001;
-            }
-            badConsumers_.addInt(input.readInt32());
-            break;
-          }
-          case 26: {
-            int length = input.readRawVarint32();
-            int limit = input.pushLimit(length);
-            if (!((mutable_bitField0_ & 0x00000001) != 0) && input.getBytesUntilLimit() > 0) {
-              badConsumers_ = newIntList();
-              mutable_bitField0_ |= 0x00000001;
-            }
-            while (input.getBytesUntilLimit() > 0) {
-              badConsumers_.addInt(input.readInt32());
-            }
-            input.popLimit(limit);
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      if (((mutable_bitField0_ & 0x00000001) != 0)) {
-        badConsumers_.makeImmutable(); // C
-      }
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.framework.VersionsProtos.internal_static_tensorflow_VersionDef_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.framework.VersionsProtos.internal_static_tensorflow_VersionDef_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.framework.VersionDef.class, org.tensorflow.proto.framework.VersionDef.Builder.class);
-  }
-
-  public static final int PRODUCER_FIELD_NUMBER = 1;
-  private int producer_;
-  /**
-   * <pre>
-   * The version of the code that produced this data.
-   * </pre>
-   *
-   * <code>int32 producer = 1;</code>
-   */
-  public int getProducer() {
-    return producer_;
-  }
-
-  public static final int MIN_CONSUMER_FIELD_NUMBER = 2;
-  private int minConsumer_;
-  /**
-   * <pre>
-   * Any consumer below this version is not allowed to consume this data.
-   * </pre>
-   *
-   * <code>int32 min_consumer = 2;</code>
-   */
-  public int getMinConsumer() {
-    return minConsumer_;
-  }
-
-  public static final int BAD_CONSUMERS_FIELD_NUMBER = 3;
-  private com.google.protobuf.Internal.IntList badConsumers_;
-  /**
-   * <pre>
-   * Specific consumer versions which are disallowed (e.g. due to bugs).
-   * </pre>
-   *
-   * <code>repeated int32 bad_consumers = 3;</code>
-   */
-  public java.util.List<java.lang.Integer>
-      getBadConsumersList() {
-    return badConsumers_;
-  }
-  /**
-   * <pre>
-   * Specific consumer versions which are disallowed (e.g. due to bugs).
-   * </pre>
-   *
-   * <code>repeated int32 bad_consumers = 3;</code>
-   */
-  public int getBadConsumersCount() {
-    return badConsumers_.size();
-  }
-  /**
-   * <pre>
-   * Specific consumer versions which are disallowed (e.g. due to bugs).
-   * </pre>
-   *
-   * <code>repeated int32 bad_consumers = 3;</code>
-   */
-  public int getBadConsumers(int index) {
-    return badConsumers_.getInt(index);
-  }
-  private int badConsumersMemoizedSerializedSize = -1;
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    getSerializedSize();
-    if (producer_ != 0) {
-      output.writeInt32(1, producer_);
-    }
-    if (minConsumer_ != 0) {
-      output.writeInt32(2, minConsumer_);
-    }
-    if (getBadConsumersList().size() > 0) {
-      output.writeUInt32NoTag(26);
-      output.writeUInt32NoTag(badConsumersMemoizedSerializedSize);
-    }
-    for (int i = 0; i < badConsumers_.size(); i++) {
-      output.writeInt32NoTag(badConsumers_.getInt(i));
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (producer_ != 0) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt32Size(1, producer_);
-    }
-    if (minConsumer_ != 0) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt32Size(2, minConsumer_);
-    }
-    {
-      int dataSize = 0;
-      for (int i = 0; i < badConsumers_.size(); i++) {
-        dataSize += com.google.protobuf.CodedOutputStream
-          .computeInt32SizeNoTag(badConsumers_.getInt(i));
-      }
-      size += dataSize;
-      if (!getBadConsumersList().isEmpty()) {
-        size += 1;
-        size += com.google.protobuf.CodedOutputStream
-            .computeInt32SizeNoTag(dataSize);
-      }
-      badConsumersMemoizedSerializedSize = dataSize;
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.framework.VersionDef)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.framework.VersionDef other = (org.tensorflow.proto.framework.VersionDef) obj;
-
-    if (getProducer()
-        != other.getProducer()) return false;
-    if (getMinConsumer()
-        != other.getMinConsumer()) return false;
-    if (!getBadConsumersList()
-        .equals(other.getBadConsumersList())) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    hash = (37 * hash) + PRODUCER_FIELD_NUMBER;
-    hash = (53 * hash) + getProducer();
-    hash = (37 * hash) + MIN_CONSUMER_FIELD_NUMBER;
-    hash = (53 * hash) + getMinConsumer();
-    if (getBadConsumersCount() > 0) {
-      hash = (37 * hash) + BAD_CONSUMERS_FIELD_NUMBER;
-      hash = (53 * hash) + getBadConsumersList().hashCode();
-    }
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.framework.VersionDef parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.VersionDef parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.VersionDef parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.VersionDef parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.VersionDef parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.VersionDef parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.VersionDef parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.VersionDef parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.VersionDef parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.VersionDef parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.VersionDef parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.VersionDef parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.framework.VersionDef prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * <pre>
-   * Version information for a piece of serialized data
-   * There are different types of versions for each type of data
-   * (GraphDef, etc.), but they all have the same common shape
-   * described here.
-   * Each consumer has "consumer" and "min_producer" versions (specified
-   * elsewhere).  A consumer is allowed to consume this data if
-   *   producer &gt;= min_producer
-   *   consumer &gt;= min_consumer
-   *   consumer not in bad_consumers
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.VersionDef}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.VersionDef)
-      org.tensorflow.proto.framework.VersionDefOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.VersionsProtos.internal_static_tensorflow_VersionDef_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.VersionsProtos.internal_static_tensorflow_VersionDef_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.VersionDef.class, org.tensorflow.proto.framework.VersionDef.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.framework.VersionDef.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      producer_ = 0;
-
-      minConsumer_ = 0;
-
-      badConsumers_ = emptyIntList();
-      bitField0_ = (bitField0_ & ~0x00000001);
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.framework.VersionsProtos.internal_static_tensorflow_VersionDef_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.VersionDef getDefaultInstanceForType() {
-      return org.tensorflow.proto.framework.VersionDef.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.VersionDef build() {
-      org.tensorflow.proto.framework.VersionDef result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.VersionDef buildPartial() {
-      org.tensorflow.proto.framework.VersionDef result = new org.tensorflow.proto.framework.VersionDef(this);
-      int from_bitField0_ = bitField0_;
-      result.producer_ = producer_;
-      result.minConsumer_ = minConsumer_;
-      if (((bitField0_ & 0x00000001) != 0)) {
-        badConsumers_.makeImmutable();
-        bitField0_ = (bitField0_ & ~0x00000001);
-      }
-      result.badConsumers_ = badConsumers_;
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.framework.VersionDef) {
-        return mergeFrom((org.tensorflow.proto.framework.VersionDef)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.framework.VersionDef other) {
-      if (other == org.tensorflow.proto.framework.VersionDef.getDefaultInstance()) return this;
-      if (other.getProducer() != 0) {
-        setProducer(other.getProducer());
-      }
-      if (other.getMinConsumer() != 0) {
-        setMinConsumer(other.getMinConsumer());
-      }
-      if (!other.badConsumers_.isEmpty()) {
-        if (badConsumers_.isEmpty()) {
-          badConsumers_ = other.badConsumers_;
-          bitField0_ = (bitField0_ & ~0x00000001);
-        } else {
-          ensureBadConsumersIsMutable();
-          badConsumers_.addAll(other.badConsumers_);
-        }
-        onChanged();
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.framework.VersionDef parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.framework.VersionDef) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-    private int bitField0_;
-
-    private int producer_ ;
-    /**
-     * <pre>
-     * The version of the code that produced this data.
-     * </pre>
-     *
-     * <code>int32 producer = 1;</code>
-     */
-    public int getProducer() {
-      return producer_;
-    }
-    /**
-     * <pre>
-     * The version of the code that produced this data.
-     * </pre>
-     *
-     * <code>int32 producer = 1;</code>
-     */
-    public Builder setProducer(int value) {
-      
-      producer_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * The version of the code that produced this data.
-     * </pre>
-     *
-     * <code>int32 producer = 1;</code>
-     */
-    public Builder clearProducer() {
-      
-      producer_ = 0;
-      onChanged();
-      return this;
-    }
-
-    private int minConsumer_ ;
-    /**
-     * <pre>
-     * Any consumer below this version is not allowed to consume this data.
-     * </pre>
-     *
-     * <code>int32 min_consumer = 2;</code>
-     */
-    public int getMinConsumer() {
-      return minConsumer_;
-    }
-    /**
-     * <pre>
-     * Any consumer below this version is not allowed to consume this data.
-     * </pre>
-     *
-     * <code>int32 min_consumer = 2;</code>
-     */
-    public Builder setMinConsumer(int value) {
-      
-      minConsumer_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Any consumer below this version is not allowed to consume this data.
-     * </pre>
-     *
-     * <code>int32 min_consumer = 2;</code>
-     */
-    public Builder clearMinConsumer() {
-      
-      minConsumer_ = 0;
-      onChanged();
-      return this;
-    }
-
-    private com.google.protobuf.Internal.IntList badConsumers_ = emptyIntList();
-    private void ensureBadConsumersIsMutable() {
-      if (!((bitField0_ & 0x00000001) != 0)) {
-        badConsumers_ = mutableCopy(badConsumers_);
-        bitField0_ |= 0x00000001;
-       }
-    }
-    /**
-     * <pre>
-     * Specific consumer versions which are disallowed (e.g. due to bugs).
-     * </pre>
-     *
-     * <code>repeated int32 bad_consumers = 3;</code>
-     */
-    public java.util.List<java.lang.Integer>
-        getBadConsumersList() {
-      return ((bitField0_ & 0x00000001) != 0) ?
-               java.util.Collections.unmodifiableList(badConsumers_) : badConsumers_;
-    }
-    /**
-     * <pre>
-     * Specific consumer versions which are disallowed (e.g. due to bugs).
-     * </pre>
-     *
-     * <code>repeated int32 bad_consumers = 3;</code>
-     */
-    public int getBadConsumersCount() {
-      return badConsumers_.size();
-    }
-    /**
-     * <pre>
-     * Specific consumer versions which are disallowed (e.g. due to bugs).
-     * </pre>
-     *
-     * <code>repeated int32 bad_consumers = 3;</code>
-     */
-    public int getBadConsumers(int index) {
-      return badConsumers_.getInt(index);
-    }
-    /**
-     * <pre>
-     * Specific consumer versions which are disallowed (e.g. due to bugs).
-     * </pre>
-     *
-     * <code>repeated int32 bad_consumers = 3;</code>
-     */
-    public Builder setBadConsumers(
-        int index, int value) {
-      ensureBadConsumersIsMutable();
-      badConsumers_.setInt(index, value);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Specific consumer versions which are disallowed (e.g. due to bugs).
-     * </pre>
-     *
-     * <code>repeated int32 bad_consumers = 3;</code>
-     */
-    public Builder addBadConsumers(int value) {
-      ensureBadConsumersIsMutable();
-      badConsumers_.addInt(value);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Specific consumer versions which are disallowed (e.g. due to bugs).
-     * </pre>
-     *
-     * <code>repeated int32 bad_consumers = 3;</code>
-     */
-    public Builder addAllBadConsumers(
-        java.lang.Iterable<? extends java.lang.Integer> values) {
-      ensureBadConsumersIsMutable();
-      com.google.protobuf.AbstractMessageLite.Builder.addAll(
-          values, badConsumers_);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Specific consumer versions which are disallowed (e.g. due to bugs).
-     * </pre>
-     *
-     * <code>repeated int32 bad_consumers = 3;</code>
-     */
-    public Builder clearBadConsumers() {
-      badConsumers_ = emptyIntList();
-      bitField0_ = (bitField0_ & ~0x00000001);
-      onChanged();
-      return this;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.VersionDef)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.VersionDef)
-  private static final org.tensorflow.proto.framework.VersionDef DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.VersionDef();
-  }
-
-  public static org.tensorflow.proto.framework.VersionDef getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<VersionDef>
-      PARSER = new com.google.protobuf.AbstractParser<VersionDef>() {
-    @java.lang.Override
-    public VersionDef parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new VersionDef(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<VersionDef> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<VersionDef> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.framework.VersionDef getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/VersionDefOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/VersionDefOrBuilder.java
deleted file mode 100644
index 519769551aa..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/VersionDefOrBuilder.java
+++ /dev/null
@@ -1,52 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/framework/versions.proto
-
-package org.tensorflow.proto.framework;
-
-public interface VersionDefOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.VersionDef)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <pre>
-   * The version of the code that produced this data.
-   * </pre>
-   *
-   * <code>int32 producer = 1;</code>
-   */
-  int getProducer();
-
-  /**
-   * <pre>
-   * Any consumer below this version is not allowed to consume this data.
-   * </pre>
-   *
-   * <code>int32 min_consumer = 2;</code>
-   */
-  int getMinConsumer();
-
-  /**
-   * <pre>
-   * Specific consumer versions which are disallowed (e.g. due to bugs).
-   * </pre>
-   *
-   * <code>repeated int32 bad_consumers = 3;</code>
-   */
-  java.util.List<java.lang.Integer> getBadConsumersList();
-  /**
-   * <pre>
-   * Specific consumer versions which are disallowed (e.g. due to bugs).
-   * </pre>
-   *
-   * <code>repeated int32 bad_consumers = 3;</code>
-   */
-  int getBadConsumersCount();
-  /**
-   * <pre>
-   * Specific consumer versions which are disallowed (e.g. due to bugs).
-   * </pre>
-   *
-   * <code>repeated int32 bad_consumers = 3;</code>
-   */
-  int getBadConsumers(int index);
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/VersionsProtos.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/VersionsProtos.java
deleted file mode 100644
index 0597ff6e213..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/VersionsProtos.java
+++ /dev/null
@@ -1,52 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/framework/versions.proto
-
-package org.tensorflow.proto.framework;
-
-public final class VersionsProtos {
-  private VersionsProtos() {}
-  public static void registerAllExtensions(
-      com.google.protobuf.ExtensionRegistryLite registry) {
-  }
-
-  public static void registerAllExtensions(
-      com.google.protobuf.ExtensionRegistry registry) {
-    registerAllExtensions(
-        (com.google.protobuf.ExtensionRegistryLite) registry);
-  }
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_VersionDef_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_VersionDef_fieldAccessorTable;
-
-  public static com.google.protobuf.Descriptors.FileDescriptor
-      getDescriptor() {
-    return descriptor;
-  }
-  private static  com.google.protobuf.Descriptors.FileDescriptor
-      descriptor;
-  static {
-    java.lang.String[] descriptorData = {
-      "\n(tensorflow/core/framework/versions.pro" +
-      "to\022\ntensorflow\"K\n\nVersionDef\022\020\n\010producer" +
-      "\030\001 \001(\005\022\024\n\014min_consumer\030\002 \001(\005\022\025\n\rbad_cons" +
-      "umers\030\003 \003(\005B\206\001\n\036org.tensorflow.proto.fra" +
-      "meworkB\016VersionsProtosP\001ZOgithub.com/ten" +
-      "sorflow/tensorflow/tensorflow/go/core/fr" +
-      "amework/versions_go_proto\370\001\001b\006proto3"
-    };
-    descriptor = com.google.protobuf.Descriptors.FileDescriptor
-      .internalBuildGeneratedFileFrom(descriptorData,
-        new com.google.protobuf.Descriptors.FileDescriptor[] {
-        });
-    internal_static_tensorflow_VersionDef_descriptor =
-      getDescriptor().getMessageTypes().get(0);
-    internal_static_tensorflow_VersionDef_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_VersionDef_descriptor,
-        new java.lang.String[] { "Producer", "MinConsumer", "BadConsumers", });
-  }
-
-  // @@protoc_insertion_point(outer_class_scope)
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/WhileContextDef.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/WhileContextDef.java
deleted file mode 100644
index 9bd67ab7bcd..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/WhileContextDef.java
+++ /dev/null
@@ -1,2534 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/control_flow.proto
-
-package org.tensorflow.proto.framework;
-
-/**
- * <pre>
- * Protocol buffer representing a WhileContext object.
- * </pre>
- *
- * Protobuf type {@code tensorflow.WhileContextDef}
- */
-public  final class WhileContextDef extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.WhileContextDef)
-    WhileContextDefOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use WhileContextDef.newBuilder() to construct.
-  private WhileContextDef(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private WhileContextDef() {
-    contextName_ = "";
-    pivotName_ = "";
-    pivotForPredName_ = "";
-    pivotForBodyName_ = "";
-    loopExitNames_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-    loopEnterNames_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-    maximumIterationsName_ = "";
-    nestedContexts_ = java.util.Collections.emptyList();
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new WhileContextDef();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private WhileContextDef(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    int mutable_bitField0_ = 0;
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 10: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            contextName_ = s;
-            break;
-          }
-          case 16: {
-
-            parallelIterations_ = input.readInt32();
-            break;
-          }
-          case 24: {
-
-            backProp_ = input.readBool();
-            break;
-          }
-          case 32: {
-
-            swapMemory_ = input.readBool();
-            break;
-          }
-          case 42: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            pivotName_ = s;
-            break;
-          }
-          case 50: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            pivotForPredName_ = s;
-            break;
-          }
-          case 58: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            pivotForBodyName_ = s;
-            break;
-          }
-          case 66: {
-            java.lang.String s = input.readStringRequireUtf8();
-            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
-              loopExitNames_ = new com.google.protobuf.LazyStringArrayList();
-              mutable_bitField0_ |= 0x00000001;
-            }
-            loopExitNames_.add(s);
-            break;
-          }
-          case 74: {
-            org.tensorflow.proto.framework.ValuesDef.Builder subBuilder = null;
-            if (valuesDef_ != null) {
-              subBuilder = valuesDef_.toBuilder();
-            }
-            valuesDef_ = input.readMessage(org.tensorflow.proto.framework.ValuesDef.parser(), extensionRegistry);
-            if (subBuilder != null) {
-              subBuilder.mergeFrom(valuesDef_);
-              valuesDef_ = subBuilder.buildPartial();
-            }
-
-            break;
-          }
-          case 82: {
-            java.lang.String s = input.readStringRequireUtf8();
-            if (!((mutable_bitField0_ & 0x00000002) != 0)) {
-              loopEnterNames_ = new com.google.protobuf.LazyStringArrayList();
-              mutable_bitField0_ |= 0x00000002;
-            }
-            loopEnterNames_.add(s);
-            break;
-          }
-          case 90: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            maximumIterationsName_ = s;
-            break;
-          }
-          case 98: {
-            if (!((mutable_bitField0_ & 0x00000004) != 0)) {
-              nestedContexts_ = new java.util.ArrayList<org.tensorflow.proto.framework.ControlFlowContextDef>();
-              mutable_bitField0_ |= 0x00000004;
-            }
-            nestedContexts_.add(
-                input.readMessage(org.tensorflow.proto.framework.ControlFlowContextDef.parser(), extensionRegistry));
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      if (((mutable_bitField0_ & 0x00000001) != 0)) {
-        loopExitNames_ = loopExitNames_.getUnmodifiableView();
-      }
-      if (((mutable_bitField0_ & 0x00000002) != 0)) {
-        loopEnterNames_ = loopEnterNames_.getUnmodifiableView();
-      }
-      if (((mutable_bitField0_ & 0x00000004) != 0)) {
-        nestedContexts_ = java.util.Collections.unmodifiableList(nestedContexts_);
-      }
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.framework.ControlFlowProtos.internal_static_tensorflow_WhileContextDef_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.framework.ControlFlowProtos.internal_static_tensorflow_WhileContextDef_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.framework.WhileContextDef.class, org.tensorflow.proto.framework.WhileContextDef.Builder.class);
-  }
-
-  public static final int CONTEXT_NAME_FIELD_NUMBER = 1;
-  private volatile java.lang.Object contextName_;
-  /**
-   * <pre>
-   * Name of the context.
-   * </pre>
-   *
-   * <code>string context_name = 1;</code>
-   */
-  public java.lang.String getContextName() {
-    java.lang.Object ref = contextName_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      contextName_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * Name of the context.
-   * </pre>
-   *
-   * <code>string context_name = 1;</code>
-   */
-  public com.google.protobuf.ByteString
-      getContextNameBytes() {
-    java.lang.Object ref = contextName_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      contextName_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int PARALLEL_ITERATIONS_FIELD_NUMBER = 2;
-  private int parallelIterations_;
-  /**
-   * <pre>
-   * The number of iterations allowed to run in parallel.
-   * </pre>
-   *
-   * <code>int32 parallel_iterations = 2;</code>
-   */
-  public int getParallelIterations() {
-    return parallelIterations_;
-  }
-
-  public static final int BACK_PROP_FIELD_NUMBER = 3;
-  private boolean backProp_;
-  /**
-   * <pre>
-   * Whether backprop is enabled for this while loop.
-   * </pre>
-   *
-   * <code>bool back_prop = 3;</code>
-   */
-  public boolean getBackProp() {
-    return backProp_;
-  }
-
-  public static final int SWAP_MEMORY_FIELD_NUMBER = 4;
-  private boolean swapMemory_;
-  /**
-   * <pre>
-   * Whether GPU-CPU memory swap is enabled for this loop.
-   * </pre>
-   *
-   * <code>bool swap_memory = 4;</code>
-   */
-  public boolean getSwapMemory() {
-    return swapMemory_;
-  }
-
-  public static final int PIVOT_NAME_FIELD_NUMBER = 5;
-  private volatile java.lang.Object pivotName_;
-  /**
-   * <pre>
-   * Name of the pivot tensor.
-   * </pre>
-   *
-   * <code>string pivot_name = 5;</code>
-   */
-  public java.lang.String getPivotName() {
-    java.lang.Object ref = pivotName_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      pivotName_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * Name of the pivot tensor.
-   * </pre>
-   *
-   * <code>string pivot_name = 5;</code>
-   */
-  public com.google.protobuf.ByteString
-      getPivotNameBytes() {
-    java.lang.Object ref = pivotName_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      pivotName_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int PIVOT_FOR_PRED_NAME_FIELD_NUMBER = 6;
-  private volatile java.lang.Object pivotForPredName_;
-  /**
-   * <pre>
-   * Name of the pivot_for_pred tensor.
-   * </pre>
-   *
-   * <code>string pivot_for_pred_name = 6;</code>
-   */
-  public java.lang.String getPivotForPredName() {
-    java.lang.Object ref = pivotForPredName_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      pivotForPredName_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * Name of the pivot_for_pred tensor.
-   * </pre>
-   *
-   * <code>string pivot_for_pred_name = 6;</code>
-   */
-  public com.google.protobuf.ByteString
-      getPivotForPredNameBytes() {
-    java.lang.Object ref = pivotForPredName_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      pivotForPredName_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int PIVOT_FOR_BODY_NAME_FIELD_NUMBER = 7;
-  private volatile java.lang.Object pivotForBodyName_;
-  /**
-   * <pre>
-   * Name of the pivot_for_body tensor.
-   * </pre>
-   *
-   * <code>string pivot_for_body_name = 7;</code>
-   */
-  public java.lang.String getPivotForBodyName() {
-    java.lang.Object ref = pivotForBodyName_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      pivotForBodyName_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * Name of the pivot_for_body tensor.
-   * </pre>
-   *
-   * <code>string pivot_for_body_name = 7;</code>
-   */
-  public com.google.protobuf.ByteString
-      getPivotForBodyNameBytes() {
-    java.lang.Object ref = pivotForBodyName_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      pivotForBodyName_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int LOOP_EXIT_NAMES_FIELD_NUMBER = 8;
-  private com.google.protobuf.LazyStringList loopExitNames_;
-  /**
-   * <pre>
-   * List of names for exit tensors.
-   * </pre>
-   *
-   * <code>repeated string loop_exit_names = 8;</code>
-   */
-  public com.google.protobuf.ProtocolStringList
-      getLoopExitNamesList() {
-    return loopExitNames_;
-  }
-  /**
-   * <pre>
-   * List of names for exit tensors.
-   * </pre>
-   *
-   * <code>repeated string loop_exit_names = 8;</code>
-   */
-  public int getLoopExitNamesCount() {
-    return loopExitNames_.size();
-  }
-  /**
-   * <pre>
-   * List of names for exit tensors.
-   * </pre>
-   *
-   * <code>repeated string loop_exit_names = 8;</code>
-   */
-  public java.lang.String getLoopExitNames(int index) {
-    return loopExitNames_.get(index);
-  }
-  /**
-   * <pre>
-   * List of names for exit tensors.
-   * </pre>
-   *
-   * <code>repeated string loop_exit_names = 8;</code>
-   */
-  public com.google.protobuf.ByteString
-      getLoopExitNamesBytes(int index) {
-    return loopExitNames_.getByteString(index);
-  }
-
-  public static final int LOOP_ENTER_NAMES_FIELD_NUMBER = 10;
-  private com.google.protobuf.LazyStringList loopEnterNames_;
-  /**
-   * <pre>
-   * List of names for enter tensors.
-   * </pre>
-   *
-   * <code>repeated string loop_enter_names = 10;</code>
-   */
-  public com.google.protobuf.ProtocolStringList
-      getLoopEnterNamesList() {
-    return loopEnterNames_;
-  }
-  /**
-   * <pre>
-   * List of names for enter tensors.
-   * </pre>
-   *
-   * <code>repeated string loop_enter_names = 10;</code>
-   */
-  public int getLoopEnterNamesCount() {
-    return loopEnterNames_.size();
-  }
-  /**
-   * <pre>
-   * List of names for enter tensors.
-   * </pre>
-   *
-   * <code>repeated string loop_enter_names = 10;</code>
-   */
-  public java.lang.String getLoopEnterNames(int index) {
-    return loopEnterNames_.get(index);
-  }
-  /**
-   * <pre>
-   * List of names for enter tensors.
-   * </pre>
-   *
-   * <code>repeated string loop_enter_names = 10;</code>
-   */
-  public com.google.protobuf.ByteString
-      getLoopEnterNamesBytes(int index) {
-    return loopEnterNames_.getByteString(index);
-  }
-
-  public static final int VALUES_DEF_FIELD_NUMBER = 9;
-  private org.tensorflow.proto.framework.ValuesDef valuesDef_;
-  /**
-   * <pre>
-   * Values and external values in control flow context.
-   * </pre>
-   *
-   * <code>.tensorflow.ValuesDef values_def = 9;</code>
-   */
-  public boolean hasValuesDef() {
-    return valuesDef_ != null;
-  }
-  /**
-   * <pre>
-   * Values and external values in control flow context.
-   * </pre>
-   *
-   * <code>.tensorflow.ValuesDef values_def = 9;</code>
-   */
-  public org.tensorflow.proto.framework.ValuesDef getValuesDef() {
-    return valuesDef_ == null ? org.tensorflow.proto.framework.ValuesDef.getDefaultInstance() : valuesDef_;
-  }
-  /**
-   * <pre>
-   * Values and external values in control flow context.
-   * </pre>
-   *
-   * <code>.tensorflow.ValuesDef values_def = 9;</code>
-   */
-  public org.tensorflow.proto.framework.ValuesDefOrBuilder getValuesDefOrBuilder() {
-    return getValuesDef();
-  }
-
-  public static final int MAXIMUM_ITERATIONS_NAME_FIELD_NUMBER = 11;
-  private volatile java.lang.Object maximumIterationsName_;
-  /**
-   * <pre>
-   * Optional name of the maximum_iterations tensor.
-   * </pre>
-   *
-   * <code>string maximum_iterations_name = 11;</code>
-   */
-  public java.lang.String getMaximumIterationsName() {
-    java.lang.Object ref = maximumIterationsName_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      maximumIterationsName_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * Optional name of the maximum_iterations tensor.
-   * </pre>
-   *
-   * <code>string maximum_iterations_name = 11;</code>
-   */
-  public com.google.protobuf.ByteString
-      getMaximumIterationsNameBytes() {
-    java.lang.Object ref = maximumIterationsName_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      maximumIterationsName_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int NESTED_CONTEXTS_FIELD_NUMBER = 12;
-  private java.util.List<org.tensorflow.proto.framework.ControlFlowContextDef> nestedContexts_;
-  /**
-   * <pre>
-   * Contexts contained inside this context (e.g. nested whiles).
-   * </pre>
-   *
-   * <code>repeated .tensorflow.ControlFlowContextDef nested_contexts = 12;</code>
-   */
-  public java.util.List<org.tensorflow.proto.framework.ControlFlowContextDef> getNestedContextsList() {
-    return nestedContexts_;
-  }
-  /**
-   * <pre>
-   * Contexts contained inside this context (e.g. nested whiles).
-   * </pre>
-   *
-   * <code>repeated .tensorflow.ControlFlowContextDef nested_contexts = 12;</code>
-   */
-  public java.util.List<? extends org.tensorflow.proto.framework.ControlFlowContextDefOrBuilder> 
-      getNestedContextsOrBuilderList() {
-    return nestedContexts_;
-  }
-  /**
-   * <pre>
-   * Contexts contained inside this context (e.g. nested whiles).
-   * </pre>
-   *
-   * <code>repeated .tensorflow.ControlFlowContextDef nested_contexts = 12;</code>
-   */
-  public int getNestedContextsCount() {
-    return nestedContexts_.size();
-  }
-  /**
-   * <pre>
-   * Contexts contained inside this context (e.g. nested whiles).
-   * </pre>
-   *
-   * <code>repeated .tensorflow.ControlFlowContextDef nested_contexts = 12;</code>
-   */
-  public org.tensorflow.proto.framework.ControlFlowContextDef getNestedContexts(int index) {
-    return nestedContexts_.get(index);
-  }
-  /**
-   * <pre>
-   * Contexts contained inside this context (e.g. nested whiles).
-   * </pre>
-   *
-   * <code>repeated .tensorflow.ControlFlowContextDef nested_contexts = 12;</code>
-   */
-  public org.tensorflow.proto.framework.ControlFlowContextDefOrBuilder getNestedContextsOrBuilder(
-      int index) {
-    return nestedContexts_.get(index);
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    if (!getContextNameBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 1, contextName_);
-    }
-    if (parallelIterations_ != 0) {
-      output.writeInt32(2, parallelIterations_);
-    }
-    if (backProp_ != false) {
-      output.writeBool(3, backProp_);
-    }
-    if (swapMemory_ != false) {
-      output.writeBool(4, swapMemory_);
-    }
-    if (!getPivotNameBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 5, pivotName_);
-    }
-    if (!getPivotForPredNameBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 6, pivotForPredName_);
-    }
-    if (!getPivotForBodyNameBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 7, pivotForBodyName_);
-    }
-    for (int i = 0; i < loopExitNames_.size(); i++) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 8, loopExitNames_.getRaw(i));
-    }
-    if (valuesDef_ != null) {
-      output.writeMessage(9, getValuesDef());
-    }
-    for (int i = 0; i < loopEnterNames_.size(); i++) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 10, loopEnterNames_.getRaw(i));
-    }
-    if (!getMaximumIterationsNameBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 11, maximumIterationsName_);
-    }
-    for (int i = 0; i < nestedContexts_.size(); i++) {
-      output.writeMessage(12, nestedContexts_.get(i));
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (!getContextNameBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, contextName_);
-    }
-    if (parallelIterations_ != 0) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt32Size(2, parallelIterations_);
-    }
-    if (backProp_ != false) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeBoolSize(3, backProp_);
-    }
-    if (swapMemory_ != false) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeBoolSize(4, swapMemory_);
-    }
-    if (!getPivotNameBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(5, pivotName_);
-    }
-    if (!getPivotForPredNameBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(6, pivotForPredName_);
-    }
-    if (!getPivotForBodyNameBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(7, pivotForBodyName_);
-    }
-    {
-      int dataSize = 0;
-      for (int i = 0; i < loopExitNames_.size(); i++) {
-        dataSize += computeStringSizeNoTag(loopExitNames_.getRaw(i));
-      }
-      size += dataSize;
-      size += 1 * getLoopExitNamesList().size();
-    }
-    if (valuesDef_ != null) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(9, getValuesDef());
-    }
-    {
-      int dataSize = 0;
-      for (int i = 0; i < loopEnterNames_.size(); i++) {
-        dataSize += computeStringSizeNoTag(loopEnterNames_.getRaw(i));
-      }
-      size += dataSize;
-      size += 1 * getLoopEnterNamesList().size();
-    }
-    if (!getMaximumIterationsNameBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(11, maximumIterationsName_);
-    }
-    for (int i = 0; i < nestedContexts_.size(); i++) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(12, nestedContexts_.get(i));
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.framework.WhileContextDef)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.framework.WhileContextDef other = (org.tensorflow.proto.framework.WhileContextDef) obj;
-
-    if (!getContextName()
-        .equals(other.getContextName())) return false;
-    if (getParallelIterations()
-        != other.getParallelIterations()) return false;
-    if (getBackProp()
-        != other.getBackProp()) return false;
-    if (getSwapMemory()
-        != other.getSwapMemory()) return false;
-    if (!getPivotName()
-        .equals(other.getPivotName())) return false;
-    if (!getPivotForPredName()
-        .equals(other.getPivotForPredName())) return false;
-    if (!getPivotForBodyName()
-        .equals(other.getPivotForBodyName())) return false;
-    if (!getLoopExitNamesList()
-        .equals(other.getLoopExitNamesList())) return false;
-    if (!getLoopEnterNamesList()
-        .equals(other.getLoopEnterNamesList())) return false;
-    if (hasValuesDef() != other.hasValuesDef()) return false;
-    if (hasValuesDef()) {
-      if (!getValuesDef()
-          .equals(other.getValuesDef())) return false;
-    }
-    if (!getMaximumIterationsName()
-        .equals(other.getMaximumIterationsName())) return false;
-    if (!getNestedContextsList()
-        .equals(other.getNestedContextsList())) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    hash = (37 * hash) + CONTEXT_NAME_FIELD_NUMBER;
-    hash = (53 * hash) + getContextName().hashCode();
-    hash = (37 * hash) + PARALLEL_ITERATIONS_FIELD_NUMBER;
-    hash = (53 * hash) + getParallelIterations();
-    hash = (37 * hash) + BACK_PROP_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashBoolean(
-        getBackProp());
-    hash = (37 * hash) + SWAP_MEMORY_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashBoolean(
-        getSwapMemory());
-    hash = (37 * hash) + PIVOT_NAME_FIELD_NUMBER;
-    hash = (53 * hash) + getPivotName().hashCode();
-    hash = (37 * hash) + PIVOT_FOR_PRED_NAME_FIELD_NUMBER;
-    hash = (53 * hash) + getPivotForPredName().hashCode();
-    hash = (37 * hash) + PIVOT_FOR_BODY_NAME_FIELD_NUMBER;
-    hash = (53 * hash) + getPivotForBodyName().hashCode();
-    if (getLoopExitNamesCount() > 0) {
-      hash = (37 * hash) + LOOP_EXIT_NAMES_FIELD_NUMBER;
-      hash = (53 * hash) + getLoopExitNamesList().hashCode();
-    }
-    if (getLoopEnterNamesCount() > 0) {
-      hash = (37 * hash) + LOOP_ENTER_NAMES_FIELD_NUMBER;
-      hash = (53 * hash) + getLoopEnterNamesList().hashCode();
-    }
-    if (hasValuesDef()) {
-      hash = (37 * hash) + VALUES_DEF_FIELD_NUMBER;
-      hash = (53 * hash) + getValuesDef().hashCode();
-    }
-    hash = (37 * hash) + MAXIMUM_ITERATIONS_NAME_FIELD_NUMBER;
-    hash = (53 * hash) + getMaximumIterationsName().hashCode();
-    if (getNestedContextsCount() > 0) {
-      hash = (37 * hash) + NESTED_CONTEXTS_FIELD_NUMBER;
-      hash = (53 * hash) + getNestedContextsList().hashCode();
-    }
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.framework.WhileContextDef parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.WhileContextDef parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.WhileContextDef parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.WhileContextDef parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.WhileContextDef parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.framework.WhileContextDef parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.WhileContextDef parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.WhileContextDef parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.WhileContextDef parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.WhileContextDef parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.framework.WhileContextDef parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.framework.WhileContextDef parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.framework.WhileContextDef prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * <pre>
-   * Protocol buffer representing a WhileContext object.
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.WhileContextDef}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.WhileContextDef)
-      org.tensorflow.proto.framework.WhileContextDefOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.framework.ControlFlowProtos.internal_static_tensorflow_WhileContextDef_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.framework.ControlFlowProtos.internal_static_tensorflow_WhileContextDef_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.framework.WhileContextDef.class, org.tensorflow.proto.framework.WhileContextDef.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.framework.WhileContextDef.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-        getNestedContextsFieldBuilder();
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      contextName_ = "";
-
-      parallelIterations_ = 0;
-
-      backProp_ = false;
-
-      swapMemory_ = false;
-
-      pivotName_ = "";
-
-      pivotForPredName_ = "";
-
-      pivotForBodyName_ = "";
-
-      loopExitNames_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-      bitField0_ = (bitField0_ & ~0x00000001);
-      loopEnterNames_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-      bitField0_ = (bitField0_ & ~0x00000002);
-      if (valuesDefBuilder_ == null) {
-        valuesDef_ = null;
-      } else {
-        valuesDef_ = null;
-        valuesDefBuilder_ = null;
-      }
-      maximumIterationsName_ = "";
-
-      if (nestedContextsBuilder_ == null) {
-        nestedContexts_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000004);
-      } else {
-        nestedContextsBuilder_.clear();
-      }
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.framework.ControlFlowProtos.internal_static_tensorflow_WhileContextDef_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.WhileContextDef getDefaultInstanceForType() {
-      return org.tensorflow.proto.framework.WhileContextDef.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.WhileContextDef build() {
-      org.tensorflow.proto.framework.WhileContextDef result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.framework.WhileContextDef buildPartial() {
-      org.tensorflow.proto.framework.WhileContextDef result = new org.tensorflow.proto.framework.WhileContextDef(this);
-      int from_bitField0_ = bitField0_;
-      result.contextName_ = contextName_;
-      result.parallelIterations_ = parallelIterations_;
-      result.backProp_ = backProp_;
-      result.swapMemory_ = swapMemory_;
-      result.pivotName_ = pivotName_;
-      result.pivotForPredName_ = pivotForPredName_;
-      result.pivotForBodyName_ = pivotForBodyName_;
-      if (((bitField0_ & 0x00000001) != 0)) {
-        loopExitNames_ = loopExitNames_.getUnmodifiableView();
-        bitField0_ = (bitField0_ & ~0x00000001);
-      }
-      result.loopExitNames_ = loopExitNames_;
-      if (((bitField0_ & 0x00000002) != 0)) {
-        loopEnterNames_ = loopEnterNames_.getUnmodifiableView();
-        bitField0_ = (bitField0_ & ~0x00000002);
-      }
-      result.loopEnterNames_ = loopEnterNames_;
-      if (valuesDefBuilder_ == null) {
-        result.valuesDef_ = valuesDef_;
-      } else {
-        result.valuesDef_ = valuesDefBuilder_.build();
-      }
-      result.maximumIterationsName_ = maximumIterationsName_;
-      if (nestedContextsBuilder_ == null) {
-        if (((bitField0_ & 0x00000004) != 0)) {
-          nestedContexts_ = java.util.Collections.unmodifiableList(nestedContexts_);
-          bitField0_ = (bitField0_ & ~0x00000004);
-        }
-        result.nestedContexts_ = nestedContexts_;
-      } else {
-        result.nestedContexts_ = nestedContextsBuilder_.build();
-      }
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.framework.WhileContextDef) {
-        return mergeFrom((org.tensorflow.proto.framework.WhileContextDef)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.framework.WhileContextDef other) {
-      if (other == org.tensorflow.proto.framework.WhileContextDef.getDefaultInstance()) return this;
-      if (!other.getContextName().isEmpty()) {
-        contextName_ = other.contextName_;
-        onChanged();
-      }
-      if (other.getParallelIterations() != 0) {
-        setParallelIterations(other.getParallelIterations());
-      }
-      if (other.getBackProp() != false) {
-        setBackProp(other.getBackProp());
-      }
-      if (other.getSwapMemory() != false) {
-        setSwapMemory(other.getSwapMemory());
-      }
-      if (!other.getPivotName().isEmpty()) {
-        pivotName_ = other.pivotName_;
-        onChanged();
-      }
-      if (!other.getPivotForPredName().isEmpty()) {
-        pivotForPredName_ = other.pivotForPredName_;
-        onChanged();
-      }
-      if (!other.getPivotForBodyName().isEmpty()) {
-        pivotForBodyName_ = other.pivotForBodyName_;
-        onChanged();
-      }
-      if (!other.loopExitNames_.isEmpty()) {
-        if (loopExitNames_.isEmpty()) {
-          loopExitNames_ = other.loopExitNames_;
-          bitField0_ = (bitField0_ & ~0x00000001);
-        } else {
-          ensureLoopExitNamesIsMutable();
-          loopExitNames_.addAll(other.loopExitNames_);
-        }
-        onChanged();
-      }
-      if (!other.loopEnterNames_.isEmpty()) {
-        if (loopEnterNames_.isEmpty()) {
-          loopEnterNames_ = other.loopEnterNames_;
-          bitField0_ = (bitField0_ & ~0x00000002);
-        } else {
-          ensureLoopEnterNamesIsMutable();
-          loopEnterNames_.addAll(other.loopEnterNames_);
-        }
-        onChanged();
-      }
-      if (other.hasValuesDef()) {
-        mergeValuesDef(other.getValuesDef());
-      }
-      if (!other.getMaximumIterationsName().isEmpty()) {
-        maximumIterationsName_ = other.maximumIterationsName_;
-        onChanged();
-      }
-      if (nestedContextsBuilder_ == null) {
-        if (!other.nestedContexts_.isEmpty()) {
-          if (nestedContexts_.isEmpty()) {
-            nestedContexts_ = other.nestedContexts_;
-            bitField0_ = (bitField0_ & ~0x00000004);
-          } else {
-            ensureNestedContextsIsMutable();
-            nestedContexts_.addAll(other.nestedContexts_);
-          }
-          onChanged();
-        }
-      } else {
-        if (!other.nestedContexts_.isEmpty()) {
-          if (nestedContextsBuilder_.isEmpty()) {
-            nestedContextsBuilder_.dispose();
-            nestedContextsBuilder_ = null;
-            nestedContexts_ = other.nestedContexts_;
-            bitField0_ = (bitField0_ & ~0x00000004);
-            nestedContextsBuilder_ = 
-              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
-                 getNestedContextsFieldBuilder() : null;
-          } else {
-            nestedContextsBuilder_.addAllMessages(other.nestedContexts_);
-          }
-        }
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.framework.WhileContextDef parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.framework.WhileContextDef) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-    private int bitField0_;
-
-    private java.lang.Object contextName_ = "";
-    /**
-     * <pre>
-     * Name of the context.
-     * </pre>
-     *
-     * <code>string context_name = 1;</code>
-     */
-    public java.lang.String getContextName() {
-      java.lang.Object ref = contextName_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        contextName_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Name of the context.
-     * </pre>
-     *
-     * <code>string context_name = 1;</code>
-     */
-    public com.google.protobuf.ByteString
-        getContextNameBytes() {
-      java.lang.Object ref = contextName_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        contextName_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Name of the context.
-     * </pre>
-     *
-     * <code>string context_name = 1;</code>
-     */
-    public Builder setContextName(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      contextName_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Name of the context.
-     * </pre>
-     *
-     * <code>string context_name = 1;</code>
-     */
-    public Builder clearContextName() {
-      
-      contextName_ = getDefaultInstance().getContextName();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Name of the context.
-     * </pre>
-     *
-     * <code>string context_name = 1;</code>
-     */
-    public Builder setContextNameBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      contextName_ = value;
-      onChanged();
-      return this;
-    }
-
-    private int parallelIterations_ ;
-    /**
-     * <pre>
-     * The number of iterations allowed to run in parallel.
-     * </pre>
-     *
-     * <code>int32 parallel_iterations = 2;</code>
-     */
-    public int getParallelIterations() {
-      return parallelIterations_;
-    }
-    /**
-     * <pre>
-     * The number of iterations allowed to run in parallel.
-     * </pre>
-     *
-     * <code>int32 parallel_iterations = 2;</code>
-     */
-    public Builder setParallelIterations(int value) {
-      
-      parallelIterations_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * The number of iterations allowed to run in parallel.
-     * </pre>
-     *
-     * <code>int32 parallel_iterations = 2;</code>
-     */
-    public Builder clearParallelIterations() {
-      
-      parallelIterations_ = 0;
-      onChanged();
-      return this;
-    }
-
-    private boolean backProp_ ;
-    /**
-     * <pre>
-     * Whether backprop is enabled for this while loop.
-     * </pre>
-     *
-     * <code>bool back_prop = 3;</code>
-     */
-    public boolean getBackProp() {
-      return backProp_;
-    }
-    /**
-     * <pre>
-     * Whether backprop is enabled for this while loop.
-     * </pre>
-     *
-     * <code>bool back_prop = 3;</code>
-     */
-    public Builder setBackProp(boolean value) {
-      
-      backProp_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Whether backprop is enabled for this while loop.
-     * </pre>
-     *
-     * <code>bool back_prop = 3;</code>
-     */
-    public Builder clearBackProp() {
-      
-      backProp_ = false;
-      onChanged();
-      return this;
-    }
-
-    private boolean swapMemory_ ;
-    /**
-     * <pre>
-     * Whether GPU-CPU memory swap is enabled for this loop.
-     * </pre>
-     *
-     * <code>bool swap_memory = 4;</code>
-     */
-    public boolean getSwapMemory() {
-      return swapMemory_;
-    }
-    /**
-     * <pre>
-     * Whether GPU-CPU memory swap is enabled for this loop.
-     * </pre>
-     *
-     * <code>bool swap_memory = 4;</code>
-     */
-    public Builder setSwapMemory(boolean value) {
-      
-      swapMemory_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Whether GPU-CPU memory swap is enabled for this loop.
-     * </pre>
-     *
-     * <code>bool swap_memory = 4;</code>
-     */
-    public Builder clearSwapMemory() {
-      
-      swapMemory_ = false;
-      onChanged();
-      return this;
-    }
-
-    private java.lang.Object pivotName_ = "";
-    /**
-     * <pre>
-     * Name of the pivot tensor.
-     * </pre>
-     *
-     * <code>string pivot_name = 5;</code>
-     */
-    public java.lang.String getPivotName() {
-      java.lang.Object ref = pivotName_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        pivotName_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Name of the pivot tensor.
-     * </pre>
-     *
-     * <code>string pivot_name = 5;</code>
-     */
-    public com.google.protobuf.ByteString
-        getPivotNameBytes() {
-      java.lang.Object ref = pivotName_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        pivotName_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Name of the pivot tensor.
-     * </pre>
-     *
-     * <code>string pivot_name = 5;</code>
-     */
-    public Builder setPivotName(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      pivotName_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Name of the pivot tensor.
-     * </pre>
-     *
-     * <code>string pivot_name = 5;</code>
-     */
-    public Builder clearPivotName() {
-      
-      pivotName_ = getDefaultInstance().getPivotName();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Name of the pivot tensor.
-     * </pre>
-     *
-     * <code>string pivot_name = 5;</code>
-     */
-    public Builder setPivotNameBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      pivotName_ = value;
-      onChanged();
-      return this;
-    }
-
-    private java.lang.Object pivotForPredName_ = "";
-    /**
-     * <pre>
-     * Name of the pivot_for_pred tensor.
-     * </pre>
-     *
-     * <code>string pivot_for_pred_name = 6;</code>
-     */
-    public java.lang.String getPivotForPredName() {
-      java.lang.Object ref = pivotForPredName_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        pivotForPredName_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Name of the pivot_for_pred tensor.
-     * </pre>
-     *
-     * <code>string pivot_for_pred_name = 6;</code>
-     */
-    public com.google.protobuf.ByteString
-        getPivotForPredNameBytes() {
-      java.lang.Object ref = pivotForPredName_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        pivotForPredName_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Name of the pivot_for_pred tensor.
-     * </pre>
-     *
-     * <code>string pivot_for_pred_name = 6;</code>
-     */
-    public Builder setPivotForPredName(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      pivotForPredName_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Name of the pivot_for_pred tensor.
-     * </pre>
-     *
-     * <code>string pivot_for_pred_name = 6;</code>
-     */
-    public Builder clearPivotForPredName() {
-      
-      pivotForPredName_ = getDefaultInstance().getPivotForPredName();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Name of the pivot_for_pred tensor.
-     * </pre>
-     *
-     * <code>string pivot_for_pred_name = 6;</code>
-     */
-    public Builder setPivotForPredNameBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      pivotForPredName_ = value;
-      onChanged();
-      return this;
-    }
-
-    private java.lang.Object pivotForBodyName_ = "";
-    /**
-     * <pre>
-     * Name of the pivot_for_body tensor.
-     * </pre>
-     *
-     * <code>string pivot_for_body_name = 7;</code>
-     */
-    public java.lang.String getPivotForBodyName() {
-      java.lang.Object ref = pivotForBodyName_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        pivotForBodyName_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Name of the pivot_for_body tensor.
-     * </pre>
-     *
-     * <code>string pivot_for_body_name = 7;</code>
-     */
-    public com.google.protobuf.ByteString
-        getPivotForBodyNameBytes() {
-      java.lang.Object ref = pivotForBodyName_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        pivotForBodyName_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Name of the pivot_for_body tensor.
-     * </pre>
-     *
-     * <code>string pivot_for_body_name = 7;</code>
-     */
-    public Builder setPivotForBodyName(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      pivotForBodyName_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Name of the pivot_for_body tensor.
-     * </pre>
-     *
-     * <code>string pivot_for_body_name = 7;</code>
-     */
-    public Builder clearPivotForBodyName() {
-      
-      pivotForBodyName_ = getDefaultInstance().getPivotForBodyName();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Name of the pivot_for_body tensor.
-     * </pre>
-     *
-     * <code>string pivot_for_body_name = 7;</code>
-     */
-    public Builder setPivotForBodyNameBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      pivotForBodyName_ = value;
-      onChanged();
-      return this;
-    }
-
-    private com.google.protobuf.LazyStringList loopExitNames_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-    private void ensureLoopExitNamesIsMutable() {
-      if (!((bitField0_ & 0x00000001) != 0)) {
-        loopExitNames_ = new com.google.protobuf.LazyStringArrayList(loopExitNames_);
-        bitField0_ |= 0x00000001;
-       }
-    }
-    /**
-     * <pre>
-     * List of names for exit tensors.
-     * </pre>
-     *
-     * <code>repeated string loop_exit_names = 8;</code>
-     */
-    public com.google.protobuf.ProtocolStringList
-        getLoopExitNamesList() {
-      return loopExitNames_.getUnmodifiableView();
-    }
-    /**
-     * <pre>
-     * List of names for exit tensors.
-     * </pre>
-     *
-     * <code>repeated string loop_exit_names = 8;</code>
-     */
-    public int getLoopExitNamesCount() {
-      return loopExitNames_.size();
-    }
-    /**
-     * <pre>
-     * List of names for exit tensors.
-     * </pre>
-     *
-     * <code>repeated string loop_exit_names = 8;</code>
-     */
-    public java.lang.String getLoopExitNames(int index) {
-      return loopExitNames_.get(index);
-    }
-    /**
-     * <pre>
-     * List of names for exit tensors.
-     * </pre>
-     *
-     * <code>repeated string loop_exit_names = 8;</code>
-     */
-    public com.google.protobuf.ByteString
-        getLoopExitNamesBytes(int index) {
-      return loopExitNames_.getByteString(index);
-    }
-    /**
-     * <pre>
-     * List of names for exit tensors.
-     * </pre>
-     *
-     * <code>repeated string loop_exit_names = 8;</code>
-     */
-    public Builder setLoopExitNames(
-        int index, java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  ensureLoopExitNamesIsMutable();
-      loopExitNames_.set(index, value);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * List of names for exit tensors.
-     * </pre>
-     *
-     * <code>repeated string loop_exit_names = 8;</code>
-     */
-    public Builder addLoopExitNames(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  ensureLoopExitNamesIsMutable();
-      loopExitNames_.add(value);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * List of names for exit tensors.
-     * </pre>
-     *
-     * <code>repeated string loop_exit_names = 8;</code>
-     */
-    public Builder addAllLoopExitNames(
-        java.lang.Iterable<java.lang.String> values) {
-      ensureLoopExitNamesIsMutable();
-      com.google.protobuf.AbstractMessageLite.Builder.addAll(
-          values, loopExitNames_);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * List of names for exit tensors.
-     * </pre>
-     *
-     * <code>repeated string loop_exit_names = 8;</code>
-     */
-    public Builder clearLoopExitNames() {
-      loopExitNames_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-      bitField0_ = (bitField0_ & ~0x00000001);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * List of names for exit tensors.
-     * </pre>
-     *
-     * <code>repeated string loop_exit_names = 8;</code>
-     */
-    public Builder addLoopExitNamesBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      ensureLoopExitNamesIsMutable();
-      loopExitNames_.add(value);
-      onChanged();
-      return this;
-    }
-
-    private com.google.protobuf.LazyStringList loopEnterNames_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-    private void ensureLoopEnterNamesIsMutable() {
-      if (!((bitField0_ & 0x00000002) != 0)) {
-        loopEnterNames_ = new com.google.protobuf.LazyStringArrayList(loopEnterNames_);
-        bitField0_ |= 0x00000002;
-       }
-    }
-    /**
-     * <pre>
-     * List of names for enter tensors.
-     * </pre>
-     *
-     * <code>repeated string loop_enter_names = 10;</code>
-     */
-    public com.google.protobuf.ProtocolStringList
-        getLoopEnterNamesList() {
-      return loopEnterNames_.getUnmodifiableView();
-    }
-    /**
-     * <pre>
-     * List of names for enter tensors.
-     * </pre>
-     *
-     * <code>repeated string loop_enter_names = 10;</code>
-     */
-    public int getLoopEnterNamesCount() {
-      return loopEnterNames_.size();
-    }
-    /**
-     * <pre>
-     * List of names for enter tensors.
-     * </pre>
-     *
-     * <code>repeated string loop_enter_names = 10;</code>
-     */
-    public java.lang.String getLoopEnterNames(int index) {
-      return loopEnterNames_.get(index);
-    }
-    /**
-     * <pre>
-     * List of names for enter tensors.
-     * </pre>
-     *
-     * <code>repeated string loop_enter_names = 10;</code>
-     */
-    public com.google.protobuf.ByteString
-        getLoopEnterNamesBytes(int index) {
-      return loopEnterNames_.getByteString(index);
-    }
-    /**
-     * <pre>
-     * List of names for enter tensors.
-     * </pre>
-     *
-     * <code>repeated string loop_enter_names = 10;</code>
-     */
-    public Builder setLoopEnterNames(
-        int index, java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  ensureLoopEnterNamesIsMutable();
-      loopEnterNames_.set(index, value);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * List of names for enter tensors.
-     * </pre>
-     *
-     * <code>repeated string loop_enter_names = 10;</code>
-     */
-    public Builder addLoopEnterNames(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  ensureLoopEnterNamesIsMutable();
-      loopEnterNames_.add(value);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * List of names for enter tensors.
-     * </pre>
-     *
-     * <code>repeated string loop_enter_names = 10;</code>
-     */
-    public Builder addAllLoopEnterNames(
-        java.lang.Iterable<java.lang.String> values) {
-      ensureLoopEnterNamesIsMutable();
-      com.google.protobuf.AbstractMessageLite.Builder.addAll(
-          values, loopEnterNames_);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * List of names for enter tensors.
-     * </pre>
-     *
-     * <code>repeated string loop_enter_names = 10;</code>
-     */
-    public Builder clearLoopEnterNames() {
-      loopEnterNames_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-      bitField0_ = (bitField0_ & ~0x00000002);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * List of names for enter tensors.
-     * </pre>
-     *
-     * <code>repeated string loop_enter_names = 10;</code>
-     */
-    public Builder addLoopEnterNamesBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      ensureLoopEnterNamesIsMutable();
-      loopEnterNames_.add(value);
-      onChanged();
-      return this;
-    }
-
-    private org.tensorflow.proto.framework.ValuesDef valuesDef_;
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.ValuesDef, org.tensorflow.proto.framework.ValuesDef.Builder, org.tensorflow.proto.framework.ValuesDefOrBuilder> valuesDefBuilder_;
-    /**
-     * <pre>
-     * Values and external values in control flow context.
-     * </pre>
-     *
-     * <code>.tensorflow.ValuesDef values_def = 9;</code>
-     */
-    public boolean hasValuesDef() {
-      return valuesDefBuilder_ != null || valuesDef_ != null;
-    }
-    /**
-     * <pre>
-     * Values and external values in control flow context.
-     * </pre>
-     *
-     * <code>.tensorflow.ValuesDef values_def = 9;</code>
-     */
-    public org.tensorflow.proto.framework.ValuesDef getValuesDef() {
-      if (valuesDefBuilder_ == null) {
-        return valuesDef_ == null ? org.tensorflow.proto.framework.ValuesDef.getDefaultInstance() : valuesDef_;
-      } else {
-        return valuesDefBuilder_.getMessage();
-      }
-    }
-    /**
-     * <pre>
-     * Values and external values in control flow context.
-     * </pre>
-     *
-     * <code>.tensorflow.ValuesDef values_def = 9;</code>
-     */
-    public Builder setValuesDef(org.tensorflow.proto.framework.ValuesDef value) {
-      if (valuesDefBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        valuesDef_ = value;
-        onChanged();
-      } else {
-        valuesDefBuilder_.setMessage(value);
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Values and external values in control flow context.
-     * </pre>
-     *
-     * <code>.tensorflow.ValuesDef values_def = 9;</code>
-     */
-    public Builder setValuesDef(
-        org.tensorflow.proto.framework.ValuesDef.Builder builderForValue) {
-      if (valuesDefBuilder_ == null) {
-        valuesDef_ = builderForValue.build();
-        onChanged();
-      } else {
-        valuesDefBuilder_.setMessage(builderForValue.build());
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Values and external values in control flow context.
-     * </pre>
-     *
-     * <code>.tensorflow.ValuesDef values_def = 9;</code>
-     */
-    public Builder mergeValuesDef(org.tensorflow.proto.framework.ValuesDef value) {
-      if (valuesDefBuilder_ == null) {
-        if (valuesDef_ != null) {
-          valuesDef_ =
-            org.tensorflow.proto.framework.ValuesDef.newBuilder(valuesDef_).mergeFrom(value).buildPartial();
-        } else {
-          valuesDef_ = value;
-        }
-        onChanged();
-      } else {
-        valuesDefBuilder_.mergeFrom(value);
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Values and external values in control flow context.
-     * </pre>
-     *
-     * <code>.tensorflow.ValuesDef values_def = 9;</code>
-     */
-    public Builder clearValuesDef() {
-      if (valuesDefBuilder_ == null) {
-        valuesDef_ = null;
-        onChanged();
-      } else {
-        valuesDef_ = null;
-        valuesDefBuilder_ = null;
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Values and external values in control flow context.
-     * </pre>
-     *
-     * <code>.tensorflow.ValuesDef values_def = 9;</code>
-     */
-    public org.tensorflow.proto.framework.ValuesDef.Builder getValuesDefBuilder() {
-      
-      onChanged();
-      return getValuesDefFieldBuilder().getBuilder();
-    }
-    /**
-     * <pre>
-     * Values and external values in control flow context.
-     * </pre>
-     *
-     * <code>.tensorflow.ValuesDef values_def = 9;</code>
-     */
-    public org.tensorflow.proto.framework.ValuesDefOrBuilder getValuesDefOrBuilder() {
-      if (valuesDefBuilder_ != null) {
-        return valuesDefBuilder_.getMessageOrBuilder();
-      } else {
-        return valuesDef_ == null ?
-            org.tensorflow.proto.framework.ValuesDef.getDefaultInstance() : valuesDef_;
-      }
-    }
-    /**
-     * <pre>
-     * Values and external values in control flow context.
-     * </pre>
-     *
-     * <code>.tensorflow.ValuesDef values_def = 9;</code>
-     */
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.ValuesDef, org.tensorflow.proto.framework.ValuesDef.Builder, org.tensorflow.proto.framework.ValuesDefOrBuilder> 
-        getValuesDefFieldBuilder() {
-      if (valuesDefBuilder_ == null) {
-        valuesDefBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-            org.tensorflow.proto.framework.ValuesDef, org.tensorflow.proto.framework.ValuesDef.Builder, org.tensorflow.proto.framework.ValuesDefOrBuilder>(
-                getValuesDef(),
-                getParentForChildren(),
-                isClean());
-        valuesDef_ = null;
-      }
-      return valuesDefBuilder_;
-    }
-
-    private java.lang.Object maximumIterationsName_ = "";
-    /**
-     * <pre>
-     * Optional name of the maximum_iterations tensor.
-     * </pre>
-     *
-     * <code>string maximum_iterations_name = 11;</code>
-     */
-    public java.lang.String getMaximumIterationsName() {
-      java.lang.Object ref = maximumIterationsName_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        maximumIterationsName_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Optional name of the maximum_iterations tensor.
-     * </pre>
-     *
-     * <code>string maximum_iterations_name = 11;</code>
-     */
-    public com.google.protobuf.ByteString
-        getMaximumIterationsNameBytes() {
-      java.lang.Object ref = maximumIterationsName_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        maximumIterationsName_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Optional name of the maximum_iterations tensor.
-     * </pre>
-     *
-     * <code>string maximum_iterations_name = 11;</code>
-     */
-    public Builder setMaximumIterationsName(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      maximumIterationsName_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Optional name of the maximum_iterations tensor.
-     * </pre>
-     *
-     * <code>string maximum_iterations_name = 11;</code>
-     */
-    public Builder clearMaximumIterationsName() {
-      
-      maximumIterationsName_ = getDefaultInstance().getMaximumIterationsName();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Optional name of the maximum_iterations tensor.
-     * </pre>
-     *
-     * <code>string maximum_iterations_name = 11;</code>
-     */
-    public Builder setMaximumIterationsNameBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      maximumIterationsName_ = value;
-      onChanged();
-      return this;
-    }
-
-    private java.util.List<org.tensorflow.proto.framework.ControlFlowContextDef> nestedContexts_ =
-      java.util.Collections.emptyList();
-    private void ensureNestedContextsIsMutable() {
-      if (!((bitField0_ & 0x00000004) != 0)) {
-        nestedContexts_ = new java.util.ArrayList<org.tensorflow.proto.framework.ControlFlowContextDef>(nestedContexts_);
-        bitField0_ |= 0x00000004;
-       }
-    }
-
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.framework.ControlFlowContextDef, org.tensorflow.proto.framework.ControlFlowContextDef.Builder, org.tensorflow.proto.framework.ControlFlowContextDefOrBuilder> nestedContextsBuilder_;
-
-    /**
-     * <pre>
-     * Contexts contained inside this context (e.g. nested whiles).
-     * </pre>
-     *
-     * <code>repeated .tensorflow.ControlFlowContextDef nested_contexts = 12;</code>
-     */
-    public java.util.List<org.tensorflow.proto.framework.ControlFlowContextDef> getNestedContextsList() {
-      if (nestedContextsBuilder_ == null) {
-        return java.util.Collections.unmodifiableList(nestedContexts_);
-      } else {
-        return nestedContextsBuilder_.getMessageList();
-      }
-    }
-    /**
-     * <pre>
-     * Contexts contained inside this context (e.g. nested whiles).
-     * </pre>
-     *
-     * <code>repeated .tensorflow.ControlFlowContextDef nested_contexts = 12;</code>
-     */
-    public int getNestedContextsCount() {
-      if (nestedContextsBuilder_ == null) {
-        return nestedContexts_.size();
-      } else {
-        return nestedContextsBuilder_.getCount();
-      }
-    }
-    /**
-     * <pre>
-     * Contexts contained inside this context (e.g. nested whiles).
-     * </pre>
-     *
-     * <code>repeated .tensorflow.ControlFlowContextDef nested_contexts = 12;</code>
-     */
-    public org.tensorflow.proto.framework.ControlFlowContextDef getNestedContexts(int index) {
-      if (nestedContextsBuilder_ == null) {
-        return nestedContexts_.get(index);
-      } else {
-        return nestedContextsBuilder_.getMessage(index);
-      }
-    }
-    /**
-     * <pre>
-     * Contexts contained inside this context (e.g. nested whiles).
-     * </pre>
-     *
-     * <code>repeated .tensorflow.ControlFlowContextDef nested_contexts = 12;</code>
-     */
-    public Builder setNestedContexts(
-        int index, org.tensorflow.proto.framework.ControlFlowContextDef value) {
-      if (nestedContextsBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureNestedContextsIsMutable();
-        nestedContexts_.set(index, value);
-        onChanged();
-      } else {
-        nestedContextsBuilder_.setMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Contexts contained inside this context (e.g. nested whiles).
-     * </pre>
-     *
-     * <code>repeated .tensorflow.ControlFlowContextDef nested_contexts = 12;</code>
-     */
-    public Builder setNestedContexts(
-        int index, org.tensorflow.proto.framework.ControlFlowContextDef.Builder builderForValue) {
-      if (nestedContextsBuilder_ == null) {
-        ensureNestedContextsIsMutable();
-        nestedContexts_.set(index, builderForValue.build());
-        onChanged();
-      } else {
-        nestedContextsBuilder_.setMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Contexts contained inside this context (e.g. nested whiles).
-     * </pre>
-     *
-     * <code>repeated .tensorflow.ControlFlowContextDef nested_contexts = 12;</code>
-     */
-    public Builder addNestedContexts(org.tensorflow.proto.framework.ControlFlowContextDef value) {
-      if (nestedContextsBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureNestedContextsIsMutable();
-        nestedContexts_.add(value);
-        onChanged();
-      } else {
-        nestedContextsBuilder_.addMessage(value);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Contexts contained inside this context (e.g. nested whiles).
-     * </pre>
-     *
-     * <code>repeated .tensorflow.ControlFlowContextDef nested_contexts = 12;</code>
-     */
-    public Builder addNestedContexts(
-        int index, org.tensorflow.proto.framework.ControlFlowContextDef value) {
-      if (nestedContextsBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureNestedContextsIsMutable();
-        nestedContexts_.add(index, value);
-        onChanged();
-      } else {
-        nestedContextsBuilder_.addMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Contexts contained inside this context (e.g. nested whiles).
-     * </pre>
-     *
-     * <code>repeated .tensorflow.ControlFlowContextDef nested_contexts = 12;</code>
-     */
-    public Builder addNestedContexts(
-        org.tensorflow.proto.framework.ControlFlowContextDef.Builder builderForValue) {
-      if (nestedContextsBuilder_ == null) {
-        ensureNestedContextsIsMutable();
-        nestedContexts_.add(builderForValue.build());
-        onChanged();
-      } else {
-        nestedContextsBuilder_.addMessage(builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Contexts contained inside this context (e.g. nested whiles).
-     * </pre>
-     *
-     * <code>repeated .tensorflow.ControlFlowContextDef nested_contexts = 12;</code>
-     */
-    public Builder addNestedContexts(
-        int index, org.tensorflow.proto.framework.ControlFlowContextDef.Builder builderForValue) {
-      if (nestedContextsBuilder_ == null) {
-        ensureNestedContextsIsMutable();
-        nestedContexts_.add(index, builderForValue.build());
-        onChanged();
-      } else {
-        nestedContextsBuilder_.addMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Contexts contained inside this context (e.g. nested whiles).
-     * </pre>
-     *
-     * <code>repeated .tensorflow.ControlFlowContextDef nested_contexts = 12;</code>
-     */
-    public Builder addAllNestedContexts(
-        java.lang.Iterable<? extends org.tensorflow.proto.framework.ControlFlowContextDef> values) {
-      if (nestedContextsBuilder_ == null) {
-        ensureNestedContextsIsMutable();
-        com.google.protobuf.AbstractMessageLite.Builder.addAll(
-            values, nestedContexts_);
-        onChanged();
-      } else {
-        nestedContextsBuilder_.addAllMessages(values);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Contexts contained inside this context (e.g. nested whiles).
-     * </pre>
-     *
-     * <code>repeated .tensorflow.ControlFlowContextDef nested_contexts = 12;</code>
-     */
-    public Builder clearNestedContexts() {
-      if (nestedContextsBuilder_ == null) {
-        nestedContexts_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000004);
-        onChanged();
-      } else {
-        nestedContextsBuilder_.clear();
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Contexts contained inside this context (e.g. nested whiles).
-     * </pre>
-     *
-     * <code>repeated .tensorflow.ControlFlowContextDef nested_contexts = 12;</code>
-     */
-    public Builder removeNestedContexts(int index) {
-      if (nestedContextsBuilder_ == null) {
-        ensureNestedContextsIsMutable();
-        nestedContexts_.remove(index);
-        onChanged();
-      } else {
-        nestedContextsBuilder_.remove(index);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Contexts contained inside this context (e.g. nested whiles).
-     * </pre>
-     *
-     * <code>repeated .tensorflow.ControlFlowContextDef nested_contexts = 12;</code>
-     */
-    public org.tensorflow.proto.framework.ControlFlowContextDef.Builder getNestedContextsBuilder(
-        int index) {
-      return getNestedContextsFieldBuilder().getBuilder(index);
-    }
-    /**
-     * <pre>
-     * Contexts contained inside this context (e.g. nested whiles).
-     * </pre>
-     *
-     * <code>repeated .tensorflow.ControlFlowContextDef nested_contexts = 12;</code>
-     */
-    public org.tensorflow.proto.framework.ControlFlowContextDefOrBuilder getNestedContextsOrBuilder(
-        int index) {
-      if (nestedContextsBuilder_ == null) {
-        return nestedContexts_.get(index);  } else {
-        return nestedContextsBuilder_.getMessageOrBuilder(index);
-      }
-    }
-    /**
-     * <pre>
-     * Contexts contained inside this context (e.g. nested whiles).
-     * </pre>
-     *
-     * <code>repeated .tensorflow.ControlFlowContextDef nested_contexts = 12;</code>
-     */
-    public java.util.List<? extends org.tensorflow.proto.framework.ControlFlowContextDefOrBuilder> 
-         getNestedContextsOrBuilderList() {
-      if (nestedContextsBuilder_ != null) {
-        return nestedContextsBuilder_.getMessageOrBuilderList();
-      } else {
-        return java.util.Collections.unmodifiableList(nestedContexts_);
-      }
-    }
-    /**
-     * <pre>
-     * Contexts contained inside this context (e.g. nested whiles).
-     * </pre>
-     *
-     * <code>repeated .tensorflow.ControlFlowContextDef nested_contexts = 12;</code>
-     */
-    public org.tensorflow.proto.framework.ControlFlowContextDef.Builder addNestedContextsBuilder() {
-      return getNestedContextsFieldBuilder().addBuilder(
-          org.tensorflow.proto.framework.ControlFlowContextDef.getDefaultInstance());
-    }
-    /**
-     * <pre>
-     * Contexts contained inside this context (e.g. nested whiles).
-     * </pre>
-     *
-     * <code>repeated .tensorflow.ControlFlowContextDef nested_contexts = 12;</code>
-     */
-    public org.tensorflow.proto.framework.ControlFlowContextDef.Builder addNestedContextsBuilder(
-        int index) {
-      return getNestedContextsFieldBuilder().addBuilder(
-          index, org.tensorflow.proto.framework.ControlFlowContextDef.getDefaultInstance());
-    }
-    /**
-     * <pre>
-     * Contexts contained inside this context (e.g. nested whiles).
-     * </pre>
-     *
-     * <code>repeated .tensorflow.ControlFlowContextDef nested_contexts = 12;</code>
-     */
-    public java.util.List<org.tensorflow.proto.framework.ControlFlowContextDef.Builder> 
-         getNestedContextsBuilderList() {
-      return getNestedContextsFieldBuilder().getBuilderList();
-    }
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.framework.ControlFlowContextDef, org.tensorflow.proto.framework.ControlFlowContextDef.Builder, org.tensorflow.proto.framework.ControlFlowContextDefOrBuilder> 
-        getNestedContextsFieldBuilder() {
-      if (nestedContextsBuilder_ == null) {
-        nestedContextsBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
-            org.tensorflow.proto.framework.ControlFlowContextDef, org.tensorflow.proto.framework.ControlFlowContextDef.Builder, org.tensorflow.proto.framework.ControlFlowContextDefOrBuilder>(
-                nestedContexts_,
-                ((bitField0_ & 0x00000004) != 0),
-                getParentForChildren(),
-                isClean());
-        nestedContexts_ = null;
-      }
-      return nestedContextsBuilder_;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.WhileContextDef)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.WhileContextDef)
-  private static final org.tensorflow.proto.framework.WhileContextDef DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.WhileContextDef();
-  }
-
-  public static org.tensorflow.proto.framework.WhileContextDef getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<WhileContextDef>
-      PARSER = new com.google.protobuf.AbstractParser<WhileContextDef>() {
-    @java.lang.Override
-    public WhileContextDef parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new WhileContextDef(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<WhileContextDef> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<WhileContextDef> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.framework.WhileContextDef getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/WhileContextDefOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/WhileContextDefOrBuilder.java
deleted file mode 100644
index 6fd3e2639e1..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/WhileContextDefOrBuilder.java
+++ /dev/null
@@ -1,265 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/control_flow.proto
-
-package org.tensorflow.proto.framework;
-
-public interface WhileContextDefOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.WhileContextDef)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <pre>
-   * Name of the context.
-   * </pre>
-   *
-   * <code>string context_name = 1;</code>
-   */
-  java.lang.String getContextName();
-  /**
-   * <pre>
-   * Name of the context.
-   * </pre>
-   *
-   * <code>string context_name = 1;</code>
-   */
-  com.google.protobuf.ByteString
-      getContextNameBytes();
-
-  /**
-   * <pre>
-   * The number of iterations allowed to run in parallel.
-   * </pre>
-   *
-   * <code>int32 parallel_iterations = 2;</code>
-   */
-  int getParallelIterations();
-
-  /**
-   * <pre>
-   * Whether backprop is enabled for this while loop.
-   * </pre>
-   *
-   * <code>bool back_prop = 3;</code>
-   */
-  boolean getBackProp();
-
-  /**
-   * <pre>
-   * Whether GPU-CPU memory swap is enabled for this loop.
-   * </pre>
-   *
-   * <code>bool swap_memory = 4;</code>
-   */
-  boolean getSwapMemory();
-
-  /**
-   * <pre>
-   * Name of the pivot tensor.
-   * </pre>
-   *
-   * <code>string pivot_name = 5;</code>
-   */
-  java.lang.String getPivotName();
-  /**
-   * <pre>
-   * Name of the pivot tensor.
-   * </pre>
-   *
-   * <code>string pivot_name = 5;</code>
-   */
-  com.google.protobuf.ByteString
-      getPivotNameBytes();
-
-  /**
-   * <pre>
-   * Name of the pivot_for_pred tensor.
-   * </pre>
-   *
-   * <code>string pivot_for_pred_name = 6;</code>
-   */
-  java.lang.String getPivotForPredName();
-  /**
-   * <pre>
-   * Name of the pivot_for_pred tensor.
-   * </pre>
-   *
-   * <code>string pivot_for_pred_name = 6;</code>
-   */
-  com.google.protobuf.ByteString
-      getPivotForPredNameBytes();
-
-  /**
-   * <pre>
-   * Name of the pivot_for_body tensor.
-   * </pre>
-   *
-   * <code>string pivot_for_body_name = 7;</code>
-   */
-  java.lang.String getPivotForBodyName();
-  /**
-   * <pre>
-   * Name of the pivot_for_body tensor.
-   * </pre>
-   *
-   * <code>string pivot_for_body_name = 7;</code>
-   */
-  com.google.protobuf.ByteString
-      getPivotForBodyNameBytes();
-
-  /**
-   * <pre>
-   * List of names for exit tensors.
-   * </pre>
-   *
-   * <code>repeated string loop_exit_names = 8;</code>
-   */
-  java.util.List<java.lang.String>
-      getLoopExitNamesList();
-  /**
-   * <pre>
-   * List of names for exit tensors.
-   * </pre>
-   *
-   * <code>repeated string loop_exit_names = 8;</code>
-   */
-  int getLoopExitNamesCount();
-  /**
-   * <pre>
-   * List of names for exit tensors.
-   * </pre>
-   *
-   * <code>repeated string loop_exit_names = 8;</code>
-   */
-  java.lang.String getLoopExitNames(int index);
-  /**
-   * <pre>
-   * List of names for exit tensors.
-   * </pre>
-   *
-   * <code>repeated string loop_exit_names = 8;</code>
-   */
-  com.google.protobuf.ByteString
-      getLoopExitNamesBytes(int index);
-
-  /**
-   * <pre>
-   * List of names for enter tensors.
-   * </pre>
-   *
-   * <code>repeated string loop_enter_names = 10;</code>
-   */
-  java.util.List<java.lang.String>
-      getLoopEnterNamesList();
-  /**
-   * <pre>
-   * List of names for enter tensors.
-   * </pre>
-   *
-   * <code>repeated string loop_enter_names = 10;</code>
-   */
-  int getLoopEnterNamesCount();
-  /**
-   * <pre>
-   * List of names for enter tensors.
-   * </pre>
-   *
-   * <code>repeated string loop_enter_names = 10;</code>
-   */
-  java.lang.String getLoopEnterNames(int index);
-  /**
-   * <pre>
-   * List of names for enter tensors.
-   * </pre>
-   *
-   * <code>repeated string loop_enter_names = 10;</code>
-   */
-  com.google.protobuf.ByteString
-      getLoopEnterNamesBytes(int index);
-
-  /**
-   * <pre>
-   * Values and external values in control flow context.
-   * </pre>
-   *
-   * <code>.tensorflow.ValuesDef values_def = 9;</code>
-   */
-  boolean hasValuesDef();
-  /**
-   * <pre>
-   * Values and external values in control flow context.
-   * </pre>
-   *
-   * <code>.tensorflow.ValuesDef values_def = 9;</code>
-   */
-  org.tensorflow.proto.framework.ValuesDef getValuesDef();
-  /**
-   * <pre>
-   * Values and external values in control flow context.
-   * </pre>
-   *
-   * <code>.tensorflow.ValuesDef values_def = 9;</code>
-   */
-  org.tensorflow.proto.framework.ValuesDefOrBuilder getValuesDefOrBuilder();
-
-  /**
-   * <pre>
-   * Optional name of the maximum_iterations tensor.
-   * </pre>
-   *
-   * <code>string maximum_iterations_name = 11;</code>
-   */
-  java.lang.String getMaximumIterationsName();
-  /**
-   * <pre>
-   * Optional name of the maximum_iterations tensor.
-   * </pre>
-   *
-   * <code>string maximum_iterations_name = 11;</code>
-   */
-  com.google.protobuf.ByteString
-      getMaximumIterationsNameBytes();
-
-  /**
-   * <pre>
-   * Contexts contained inside this context (e.g. nested whiles).
-   * </pre>
-   *
-   * <code>repeated .tensorflow.ControlFlowContextDef nested_contexts = 12;</code>
-   */
-  java.util.List<org.tensorflow.proto.framework.ControlFlowContextDef> 
-      getNestedContextsList();
-  /**
-   * <pre>
-   * Contexts contained inside this context (e.g. nested whiles).
-   * </pre>
-   *
-   * <code>repeated .tensorflow.ControlFlowContextDef nested_contexts = 12;</code>
-   */
-  org.tensorflow.proto.framework.ControlFlowContextDef getNestedContexts(int index);
-  /**
-   * <pre>
-   * Contexts contained inside this context (e.g. nested whiles).
-   * </pre>
-   *
-   * <code>repeated .tensorflow.ControlFlowContextDef nested_contexts = 12;</code>
-   */
-  int getNestedContextsCount();
-  /**
-   * <pre>
-   * Contexts contained inside this context (e.g. nested whiles).
-   * </pre>
-   *
-   * <code>repeated .tensorflow.ControlFlowContextDef nested_contexts = 12;</code>
-   */
-  java.util.List<? extends org.tensorflow.proto.framework.ControlFlowContextDefOrBuilder> 
-      getNestedContextsOrBuilderList();
-  /**
-   * <pre>
-   * Contexts contained inside this context (e.g. nested whiles).
-   * </pre>
-   *
-   * <code>repeated .tensorflow.ControlFlowContextDef nested_contexts = 12;</code>
-   */
-  org.tensorflow.proto.framework.ControlFlowContextDefOrBuilder getNestedContextsOrBuilder(
-      int index);
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/ProfileOptions.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/ProfileOptions.java
deleted file mode 100644
index 1a071369549..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/ProfileOptions.java
+++ /dev/null
@@ -1,1176 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/profiler/profiler_options.proto
-
-package org.tensorflow.proto.profiler;
-
-/**
- * Protobuf type {@code tensorflow.ProfileOptions}
- */
-public  final class ProfileOptions extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.ProfileOptions)
-    ProfileOptionsOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use ProfileOptions.newBuilder() to construct.
-  private ProfileOptions(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private ProfileOptions() {
-    deviceType_ = 0;
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new ProfileOptions();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private ProfileOptions(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 8: {
-
-            includeDatasetOps_ = input.readBool();
-            break;
-          }
-          case 16: {
-
-            hostTracerLevel_ = input.readUInt32();
-            break;
-          }
-          case 24: {
-
-            deviceTracerLevel_ = input.readUInt32();
-            break;
-          }
-          case 32: {
-
-            pythonTracerLevel_ = input.readUInt32();
-            break;
-          }
-          case 40: {
-
-            version_ = input.readUInt32();
-            break;
-          }
-          case 48: {
-            int rawValue = input.readEnum();
-
-            deviceType_ = rawValue;
-            break;
-          }
-          case 56: {
-
-            enableHloProto_ = input.readBool();
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.profiler.ProfilerOptionsProtos.internal_static_tensorflow_ProfileOptions_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.profiler.ProfilerOptionsProtos.internal_static_tensorflow_ProfileOptions_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.profiler.ProfileOptions.class, org.tensorflow.proto.profiler.ProfileOptions.Builder.class);
-  }
-
-  /**
-   * Protobuf enum {@code tensorflow.ProfileOptions.DeviceType}
-   */
-  public enum DeviceType
-      implements com.google.protobuf.ProtocolMessageEnum {
-    /**
-     * <code>UNSPECIFIED = 0;</code>
-     */
-    UNSPECIFIED(0),
-    /**
-     * <code>CPU = 1;</code>
-     */
-    CPU(1),
-    /**
-     * <code>GPU = 2;</code>
-     */
-    GPU(2),
-    /**
-     * <code>TPU = 3;</code>
-     */
-    TPU(3),
-    UNRECOGNIZED(-1),
-    ;
-
-    /**
-     * <code>UNSPECIFIED = 0;</code>
-     */
-    public static final int UNSPECIFIED_VALUE = 0;
-    /**
-     * <code>CPU = 1;</code>
-     */
-    public static final int CPU_VALUE = 1;
-    /**
-     * <code>GPU = 2;</code>
-     */
-    public static final int GPU_VALUE = 2;
-    /**
-     * <code>TPU = 3;</code>
-     */
-    public static final int TPU_VALUE = 3;
-
-
-    public final int getNumber() {
-      if (this == UNRECOGNIZED) {
-        throw new java.lang.IllegalArgumentException(
-            "Can't get the number of an unknown enum value.");
-      }
-      return value;
-    }
-
-    /**
-     * @deprecated Use {@link #forNumber(int)} instead.
-     */
-    @java.lang.Deprecated
-    public static DeviceType valueOf(int value) {
-      return forNumber(value);
-    }
-
-    public static DeviceType forNumber(int value) {
-      switch (value) {
-        case 0: return UNSPECIFIED;
-        case 1: return CPU;
-        case 2: return GPU;
-        case 3: return TPU;
-        default: return null;
-      }
-    }
-
-    public static com.google.protobuf.Internal.EnumLiteMap<DeviceType>
-        internalGetValueMap() {
-      return internalValueMap;
-    }
-    private static final com.google.protobuf.Internal.EnumLiteMap<
-        DeviceType> internalValueMap =
-          new com.google.protobuf.Internal.EnumLiteMap<DeviceType>() {
-            public DeviceType findValueByNumber(int number) {
-              return DeviceType.forNumber(number);
-            }
-          };
-
-    public final com.google.protobuf.Descriptors.EnumValueDescriptor
-        getValueDescriptor() {
-      return getDescriptor().getValues().get(ordinal());
-    }
-    public final com.google.protobuf.Descriptors.EnumDescriptor
-        getDescriptorForType() {
-      return getDescriptor();
-    }
-    public static final com.google.protobuf.Descriptors.EnumDescriptor
-        getDescriptor() {
-      return org.tensorflow.proto.profiler.ProfileOptions.getDescriptor().getEnumTypes().get(0);
-    }
-
-    private static final DeviceType[] VALUES = values();
-
-    public static DeviceType valueOf(
-        com.google.protobuf.Descriptors.EnumValueDescriptor desc) {
-      if (desc.getType() != getDescriptor()) {
-        throw new java.lang.IllegalArgumentException(
-          "EnumValueDescriptor is not for this type.");
-      }
-      if (desc.getIndex() == -1) {
-        return UNRECOGNIZED;
-      }
-      return VALUES[desc.getIndex()];
-    }
-
-    private final int value;
-
-    private DeviceType(int value) {
-      this.value = value;
-    }
-
-    // @@protoc_insertion_point(enum_scope:tensorflow.ProfileOptions.DeviceType)
-  }
-
-  public static final int VERSION_FIELD_NUMBER = 5;
-  private int version_;
-  /**
-   * <pre>
-   * Some default value of option are not proto3 default value. Use this version
-   * to determine if we should use default option value instead of proto3
-   * default value.
-   * </pre>
-   *
-   * <code>uint32 version = 5;</code>
-   */
-  public int getVersion() {
-    return version_;
-  }
-
-  public static final int DEVICE_TYPE_FIELD_NUMBER = 6;
-  private int deviceType_;
-  /**
-   * <pre>
-   * Device type to profile/trace: (version &gt;= 1)
-   * DeviceType::UNSPECIFIED: All registered device profiler will be enabled.
-   * DeviceType::CPU: only CPU will be profiled.
-   * DeviceType::GPU: only CPU/GPU will be profiled.
-   * DeviceType::TPU: only CPU/TPU will be profiled.
-   * </pre>
-   *
-   * <code>.tensorflow.ProfileOptions.DeviceType device_type = 6;</code>
-   */
-  public int getDeviceTypeValue() {
-    return deviceType_;
-  }
-  /**
-   * <pre>
-   * Device type to profile/trace: (version &gt;= 1)
-   * DeviceType::UNSPECIFIED: All registered device profiler will be enabled.
-   * DeviceType::CPU: only CPU will be profiled.
-   * DeviceType::GPU: only CPU/GPU will be profiled.
-   * DeviceType::TPU: only CPU/TPU will be profiled.
-   * </pre>
-   *
-   * <code>.tensorflow.ProfileOptions.DeviceType device_type = 6;</code>
-   */
-  public org.tensorflow.proto.profiler.ProfileOptions.DeviceType getDeviceType() {
-    @SuppressWarnings("deprecation")
-    org.tensorflow.proto.profiler.ProfileOptions.DeviceType result = org.tensorflow.proto.profiler.ProfileOptions.DeviceType.valueOf(deviceType_);
-    return result == null ? org.tensorflow.proto.profiler.ProfileOptions.DeviceType.UNRECOGNIZED : result;
-  }
-
-  public static final int INCLUDE_DATASET_OPS_FIELD_NUMBER = 1;
-  private boolean includeDatasetOps_;
-  /**
-   * <pre>
-   * We don't collect the dataset ops by default for better trace-viewer
-   * scalability. The caller can mannually set this field to include the ops.
-   * </pre>
-   *
-   * <code>bool include_dataset_ops = 1;</code>
-   */
-  public boolean getIncludeDatasetOps() {
-    return includeDatasetOps_;
-  }
-
-  public static final int HOST_TRACER_LEVEL_FIELD_NUMBER = 2;
-  private int hostTracerLevel_;
-  /**
-   * <pre>
-   * Levels of host tracing: (version &gt;= 1)
-   * - Level 0 is used to disable host traces.
-   * - Level 1 enables tracing of only user instrumented (or default) TraceMe.
-   * - Level 2 enables tracing of all level 1 TraceMe(s) and instrumented high
-   *           level program execution details (expensive TF ops, XLA ops, etc).
-   *           This is the default.
-   * - Level 3 enables tracing of all level 2 TraceMe(s) and more verbose
-   *           (low-level) program execution details (cheap TF ops, etc).
-   * </pre>
-   *
-   * <code>uint32 host_tracer_level = 2;</code>
-   */
-  public int getHostTracerLevel() {
-    return hostTracerLevel_;
-  }
-
-  public static final int DEVICE_TRACER_LEVEL_FIELD_NUMBER = 3;
-  private int deviceTracerLevel_;
-  /**
-   * <pre>
-   * Levels of device tracing: (version &gt;= 1)
-   * - Level 0 is used to disable device traces.
-   * - Level 1 is used to enable device traces.
-   * - More levels might be defined for specific device for controlling the
-   *   verbosity of the trace.
-   * </pre>
-   *
-   * <code>uint32 device_tracer_level = 3;</code>
-   */
-  public int getDeviceTracerLevel() {
-    return deviceTracerLevel_;
-  }
-
-  public static final int PYTHON_TRACER_LEVEL_FIELD_NUMBER = 4;
-  private int pythonTracerLevel_;
-  /**
-   * <pre>
-   * Whether enable python function calls tracing. Runtime overhead ensues if
-   * enabled. Default off. (version &gt;= 1)
-   * </pre>
-   *
-   * <code>uint32 python_tracer_level = 4;</code>
-   */
-  public int getPythonTracerLevel() {
-    return pythonTracerLevel_;
-  }
-
-  public static final int ENABLE_HLO_PROTO_FIELD_NUMBER = 7;
-  private boolean enableHloProto_;
-  /**
-   * <pre>
-   * Whether serialize hlo_proto when XLA is used. (version &gt;= 1)
-   * </pre>
-   *
-   * <code>bool enable_hlo_proto = 7;</code>
-   */
-  public boolean getEnableHloProto() {
-    return enableHloProto_;
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    if (includeDatasetOps_ != false) {
-      output.writeBool(1, includeDatasetOps_);
-    }
-    if (hostTracerLevel_ != 0) {
-      output.writeUInt32(2, hostTracerLevel_);
-    }
-    if (deviceTracerLevel_ != 0) {
-      output.writeUInt32(3, deviceTracerLevel_);
-    }
-    if (pythonTracerLevel_ != 0) {
-      output.writeUInt32(4, pythonTracerLevel_);
-    }
-    if (version_ != 0) {
-      output.writeUInt32(5, version_);
-    }
-    if (deviceType_ != org.tensorflow.proto.profiler.ProfileOptions.DeviceType.UNSPECIFIED.getNumber()) {
-      output.writeEnum(6, deviceType_);
-    }
-    if (enableHloProto_ != false) {
-      output.writeBool(7, enableHloProto_);
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (includeDatasetOps_ != false) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeBoolSize(1, includeDatasetOps_);
-    }
-    if (hostTracerLevel_ != 0) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeUInt32Size(2, hostTracerLevel_);
-    }
-    if (deviceTracerLevel_ != 0) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeUInt32Size(3, deviceTracerLevel_);
-    }
-    if (pythonTracerLevel_ != 0) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeUInt32Size(4, pythonTracerLevel_);
-    }
-    if (version_ != 0) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeUInt32Size(5, version_);
-    }
-    if (deviceType_ != org.tensorflow.proto.profiler.ProfileOptions.DeviceType.UNSPECIFIED.getNumber()) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeEnumSize(6, deviceType_);
-    }
-    if (enableHloProto_ != false) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeBoolSize(7, enableHloProto_);
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.profiler.ProfileOptions)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.profiler.ProfileOptions other = (org.tensorflow.proto.profiler.ProfileOptions) obj;
-
-    if (getVersion()
-        != other.getVersion()) return false;
-    if (deviceType_ != other.deviceType_) return false;
-    if (getIncludeDatasetOps()
-        != other.getIncludeDatasetOps()) return false;
-    if (getHostTracerLevel()
-        != other.getHostTracerLevel()) return false;
-    if (getDeviceTracerLevel()
-        != other.getDeviceTracerLevel()) return false;
-    if (getPythonTracerLevel()
-        != other.getPythonTracerLevel()) return false;
-    if (getEnableHloProto()
-        != other.getEnableHloProto()) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    hash = (37 * hash) + VERSION_FIELD_NUMBER;
-    hash = (53 * hash) + getVersion();
-    hash = (37 * hash) + DEVICE_TYPE_FIELD_NUMBER;
-    hash = (53 * hash) + deviceType_;
-    hash = (37 * hash) + INCLUDE_DATASET_OPS_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashBoolean(
-        getIncludeDatasetOps());
-    hash = (37 * hash) + HOST_TRACER_LEVEL_FIELD_NUMBER;
-    hash = (53 * hash) + getHostTracerLevel();
-    hash = (37 * hash) + DEVICE_TRACER_LEVEL_FIELD_NUMBER;
-    hash = (53 * hash) + getDeviceTracerLevel();
-    hash = (37 * hash) + PYTHON_TRACER_LEVEL_FIELD_NUMBER;
-    hash = (53 * hash) + getPythonTracerLevel();
-    hash = (37 * hash) + ENABLE_HLO_PROTO_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashBoolean(
-        getEnableHloProto());
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.profiler.ProfileOptions parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.profiler.ProfileOptions parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.profiler.ProfileOptions parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.profiler.ProfileOptions parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.profiler.ProfileOptions parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.profiler.ProfileOptions parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.profiler.ProfileOptions parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.profiler.ProfileOptions parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.profiler.ProfileOptions parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.profiler.ProfileOptions parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.profiler.ProfileOptions parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.profiler.ProfileOptions parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.profiler.ProfileOptions prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * Protobuf type {@code tensorflow.ProfileOptions}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.ProfileOptions)
-      org.tensorflow.proto.profiler.ProfileOptionsOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.profiler.ProfilerOptionsProtos.internal_static_tensorflow_ProfileOptions_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.profiler.ProfilerOptionsProtos.internal_static_tensorflow_ProfileOptions_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.profiler.ProfileOptions.class, org.tensorflow.proto.profiler.ProfileOptions.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.profiler.ProfileOptions.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      version_ = 0;
-
-      deviceType_ = 0;
-
-      includeDatasetOps_ = false;
-
-      hostTracerLevel_ = 0;
-
-      deviceTracerLevel_ = 0;
-
-      pythonTracerLevel_ = 0;
-
-      enableHloProto_ = false;
-
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.profiler.ProfilerOptionsProtos.internal_static_tensorflow_ProfileOptions_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.profiler.ProfileOptions getDefaultInstanceForType() {
-      return org.tensorflow.proto.profiler.ProfileOptions.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.profiler.ProfileOptions build() {
-      org.tensorflow.proto.profiler.ProfileOptions result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.profiler.ProfileOptions buildPartial() {
-      org.tensorflow.proto.profiler.ProfileOptions result = new org.tensorflow.proto.profiler.ProfileOptions(this);
-      result.version_ = version_;
-      result.deviceType_ = deviceType_;
-      result.includeDatasetOps_ = includeDatasetOps_;
-      result.hostTracerLevel_ = hostTracerLevel_;
-      result.deviceTracerLevel_ = deviceTracerLevel_;
-      result.pythonTracerLevel_ = pythonTracerLevel_;
-      result.enableHloProto_ = enableHloProto_;
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.profiler.ProfileOptions) {
-        return mergeFrom((org.tensorflow.proto.profiler.ProfileOptions)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.profiler.ProfileOptions other) {
-      if (other == org.tensorflow.proto.profiler.ProfileOptions.getDefaultInstance()) return this;
-      if (other.getVersion() != 0) {
-        setVersion(other.getVersion());
-      }
-      if (other.deviceType_ != 0) {
-        setDeviceTypeValue(other.getDeviceTypeValue());
-      }
-      if (other.getIncludeDatasetOps() != false) {
-        setIncludeDatasetOps(other.getIncludeDatasetOps());
-      }
-      if (other.getHostTracerLevel() != 0) {
-        setHostTracerLevel(other.getHostTracerLevel());
-      }
-      if (other.getDeviceTracerLevel() != 0) {
-        setDeviceTracerLevel(other.getDeviceTracerLevel());
-      }
-      if (other.getPythonTracerLevel() != 0) {
-        setPythonTracerLevel(other.getPythonTracerLevel());
-      }
-      if (other.getEnableHloProto() != false) {
-        setEnableHloProto(other.getEnableHloProto());
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.profiler.ProfileOptions parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.profiler.ProfileOptions) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-
-    private int version_ ;
-    /**
-     * <pre>
-     * Some default value of option are not proto3 default value. Use this version
-     * to determine if we should use default option value instead of proto3
-     * default value.
-     * </pre>
-     *
-     * <code>uint32 version = 5;</code>
-     */
-    public int getVersion() {
-      return version_;
-    }
-    /**
-     * <pre>
-     * Some default value of option are not proto3 default value. Use this version
-     * to determine if we should use default option value instead of proto3
-     * default value.
-     * </pre>
-     *
-     * <code>uint32 version = 5;</code>
-     */
-    public Builder setVersion(int value) {
-      
-      version_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Some default value of option are not proto3 default value. Use this version
-     * to determine if we should use default option value instead of proto3
-     * default value.
-     * </pre>
-     *
-     * <code>uint32 version = 5;</code>
-     */
-    public Builder clearVersion() {
-      
-      version_ = 0;
-      onChanged();
-      return this;
-    }
-
-    private int deviceType_ = 0;
-    /**
-     * <pre>
-     * Device type to profile/trace: (version &gt;= 1)
-     * DeviceType::UNSPECIFIED: All registered device profiler will be enabled.
-     * DeviceType::CPU: only CPU will be profiled.
-     * DeviceType::GPU: only CPU/GPU will be profiled.
-     * DeviceType::TPU: only CPU/TPU will be profiled.
-     * </pre>
-     *
-     * <code>.tensorflow.ProfileOptions.DeviceType device_type = 6;</code>
-     */
-    public int getDeviceTypeValue() {
-      return deviceType_;
-    }
-    /**
-     * <pre>
-     * Device type to profile/trace: (version &gt;= 1)
-     * DeviceType::UNSPECIFIED: All registered device profiler will be enabled.
-     * DeviceType::CPU: only CPU will be profiled.
-     * DeviceType::GPU: only CPU/GPU will be profiled.
-     * DeviceType::TPU: only CPU/TPU will be profiled.
-     * </pre>
-     *
-     * <code>.tensorflow.ProfileOptions.DeviceType device_type = 6;</code>
-     */
-    public Builder setDeviceTypeValue(int value) {
-      deviceType_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Device type to profile/trace: (version &gt;= 1)
-     * DeviceType::UNSPECIFIED: All registered device profiler will be enabled.
-     * DeviceType::CPU: only CPU will be profiled.
-     * DeviceType::GPU: only CPU/GPU will be profiled.
-     * DeviceType::TPU: only CPU/TPU will be profiled.
-     * </pre>
-     *
-     * <code>.tensorflow.ProfileOptions.DeviceType device_type = 6;</code>
-     */
-    public org.tensorflow.proto.profiler.ProfileOptions.DeviceType getDeviceType() {
-      @SuppressWarnings("deprecation")
-      org.tensorflow.proto.profiler.ProfileOptions.DeviceType result = org.tensorflow.proto.profiler.ProfileOptions.DeviceType.valueOf(deviceType_);
-      return result == null ? org.tensorflow.proto.profiler.ProfileOptions.DeviceType.UNRECOGNIZED : result;
-    }
-    /**
-     * <pre>
-     * Device type to profile/trace: (version &gt;= 1)
-     * DeviceType::UNSPECIFIED: All registered device profiler will be enabled.
-     * DeviceType::CPU: only CPU will be profiled.
-     * DeviceType::GPU: only CPU/GPU will be profiled.
-     * DeviceType::TPU: only CPU/TPU will be profiled.
-     * </pre>
-     *
-     * <code>.tensorflow.ProfileOptions.DeviceType device_type = 6;</code>
-     */
-    public Builder setDeviceType(org.tensorflow.proto.profiler.ProfileOptions.DeviceType value) {
-      if (value == null) {
-        throw new NullPointerException();
-      }
-      
-      deviceType_ = value.getNumber();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Device type to profile/trace: (version &gt;= 1)
-     * DeviceType::UNSPECIFIED: All registered device profiler will be enabled.
-     * DeviceType::CPU: only CPU will be profiled.
-     * DeviceType::GPU: only CPU/GPU will be profiled.
-     * DeviceType::TPU: only CPU/TPU will be profiled.
-     * </pre>
-     *
-     * <code>.tensorflow.ProfileOptions.DeviceType device_type = 6;</code>
-     */
-    public Builder clearDeviceType() {
-      
-      deviceType_ = 0;
-      onChanged();
-      return this;
-    }
-
-    private boolean includeDatasetOps_ ;
-    /**
-     * <pre>
-     * We don't collect the dataset ops by default for better trace-viewer
-     * scalability. The caller can mannually set this field to include the ops.
-     * </pre>
-     *
-     * <code>bool include_dataset_ops = 1;</code>
-     */
-    public boolean getIncludeDatasetOps() {
-      return includeDatasetOps_;
-    }
-    /**
-     * <pre>
-     * We don't collect the dataset ops by default for better trace-viewer
-     * scalability. The caller can mannually set this field to include the ops.
-     * </pre>
-     *
-     * <code>bool include_dataset_ops = 1;</code>
-     */
-    public Builder setIncludeDatasetOps(boolean value) {
-      
-      includeDatasetOps_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * We don't collect the dataset ops by default for better trace-viewer
-     * scalability. The caller can mannually set this field to include the ops.
-     * </pre>
-     *
-     * <code>bool include_dataset_ops = 1;</code>
-     */
-    public Builder clearIncludeDatasetOps() {
-      
-      includeDatasetOps_ = false;
-      onChanged();
-      return this;
-    }
-
-    private int hostTracerLevel_ ;
-    /**
-     * <pre>
-     * Levels of host tracing: (version &gt;= 1)
-     * - Level 0 is used to disable host traces.
-     * - Level 1 enables tracing of only user instrumented (or default) TraceMe.
-     * - Level 2 enables tracing of all level 1 TraceMe(s) and instrumented high
-     *           level program execution details (expensive TF ops, XLA ops, etc).
-     *           This is the default.
-     * - Level 3 enables tracing of all level 2 TraceMe(s) and more verbose
-     *           (low-level) program execution details (cheap TF ops, etc).
-     * </pre>
-     *
-     * <code>uint32 host_tracer_level = 2;</code>
-     */
-    public int getHostTracerLevel() {
-      return hostTracerLevel_;
-    }
-    /**
-     * <pre>
-     * Levels of host tracing: (version &gt;= 1)
-     * - Level 0 is used to disable host traces.
-     * - Level 1 enables tracing of only user instrumented (or default) TraceMe.
-     * - Level 2 enables tracing of all level 1 TraceMe(s) and instrumented high
-     *           level program execution details (expensive TF ops, XLA ops, etc).
-     *           This is the default.
-     * - Level 3 enables tracing of all level 2 TraceMe(s) and more verbose
-     *           (low-level) program execution details (cheap TF ops, etc).
-     * </pre>
-     *
-     * <code>uint32 host_tracer_level = 2;</code>
-     */
-    public Builder setHostTracerLevel(int value) {
-      
-      hostTracerLevel_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Levels of host tracing: (version &gt;= 1)
-     * - Level 0 is used to disable host traces.
-     * - Level 1 enables tracing of only user instrumented (or default) TraceMe.
-     * - Level 2 enables tracing of all level 1 TraceMe(s) and instrumented high
-     *           level program execution details (expensive TF ops, XLA ops, etc).
-     *           This is the default.
-     * - Level 3 enables tracing of all level 2 TraceMe(s) and more verbose
-     *           (low-level) program execution details (cheap TF ops, etc).
-     * </pre>
-     *
-     * <code>uint32 host_tracer_level = 2;</code>
-     */
-    public Builder clearHostTracerLevel() {
-      
-      hostTracerLevel_ = 0;
-      onChanged();
-      return this;
-    }
-
-    private int deviceTracerLevel_ ;
-    /**
-     * <pre>
-     * Levels of device tracing: (version &gt;= 1)
-     * - Level 0 is used to disable device traces.
-     * - Level 1 is used to enable device traces.
-     * - More levels might be defined for specific device for controlling the
-     *   verbosity of the trace.
-     * </pre>
-     *
-     * <code>uint32 device_tracer_level = 3;</code>
-     */
-    public int getDeviceTracerLevel() {
-      return deviceTracerLevel_;
-    }
-    /**
-     * <pre>
-     * Levels of device tracing: (version &gt;= 1)
-     * - Level 0 is used to disable device traces.
-     * - Level 1 is used to enable device traces.
-     * - More levels might be defined for specific device for controlling the
-     *   verbosity of the trace.
-     * </pre>
-     *
-     * <code>uint32 device_tracer_level = 3;</code>
-     */
-    public Builder setDeviceTracerLevel(int value) {
-      
-      deviceTracerLevel_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Levels of device tracing: (version &gt;= 1)
-     * - Level 0 is used to disable device traces.
-     * - Level 1 is used to enable device traces.
-     * - More levels might be defined for specific device for controlling the
-     *   verbosity of the trace.
-     * </pre>
-     *
-     * <code>uint32 device_tracer_level = 3;</code>
-     */
-    public Builder clearDeviceTracerLevel() {
-      
-      deviceTracerLevel_ = 0;
-      onChanged();
-      return this;
-    }
-
-    private int pythonTracerLevel_ ;
-    /**
-     * <pre>
-     * Whether enable python function calls tracing. Runtime overhead ensues if
-     * enabled. Default off. (version &gt;= 1)
-     * </pre>
-     *
-     * <code>uint32 python_tracer_level = 4;</code>
-     */
-    public int getPythonTracerLevel() {
-      return pythonTracerLevel_;
-    }
-    /**
-     * <pre>
-     * Whether enable python function calls tracing. Runtime overhead ensues if
-     * enabled. Default off. (version &gt;= 1)
-     * </pre>
-     *
-     * <code>uint32 python_tracer_level = 4;</code>
-     */
-    public Builder setPythonTracerLevel(int value) {
-      
-      pythonTracerLevel_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Whether enable python function calls tracing. Runtime overhead ensues if
-     * enabled. Default off. (version &gt;= 1)
-     * </pre>
-     *
-     * <code>uint32 python_tracer_level = 4;</code>
-     */
-    public Builder clearPythonTracerLevel() {
-      
-      pythonTracerLevel_ = 0;
-      onChanged();
-      return this;
-    }
-
-    private boolean enableHloProto_ ;
-    /**
-     * <pre>
-     * Whether serialize hlo_proto when XLA is used. (version &gt;= 1)
-     * </pre>
-     *
-     * <code>bool enable_hlo_proto = 7;</code>
-     */
-    public boolean getEnableHloProto() {
-      return enableHloProto_;
-    }
-    /**
-     * <pre>
-     * Whether serialize hlo_proto when XLA is used. (version &gt;= 1)
-     * </pre>
-     *
-     * <code>bool enable_hlo_proto = 7;</code>
-     */
-    public Builder setEnableHloProto(boolean value) {
-      
-      enableHloProto_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Whether serialize hlo_proto when XLA is used. (version &gt;= 1)
-     * </pre>
-     *
-     * <code>bool enable_hlo_proto = 7;</code>
-     */
-    public Builder clearEnableHloProto() {
-      
-      enableHloProto_ = false;
-      onChanged();
-      return this;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.ProfileOptions)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.ProfileOptions)
-  private static final org.tensorflow.proto.profiler.ProfileOptions DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.profiler.ProfileOptions();
-  }
-
-  public static org.tensorflow.proto.profiler.ProfileOptions getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<ProfileOptions>
-      PARSER = new com.google.protobuf.AbstractParser<ProfileOptions>() {
-    @java.lang.Override
-    public ProfileOptions parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new ProfileOptions(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<ProfileOptions> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<ProfileOptions> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.profiler.ProfileOptions getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/ProfileOptionsOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/ProfileOptionsOrBuilder.java
deleted file mode 100644
index e508bb52c79..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/ProfileOptionsOrBuilder.java
+++ /dev/null
@@ -1,103 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/profiler/profiler_options.proto
-
-package org.tensorflow.proto.profiler;
-
-public interface ProfileOptionsOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.ProfileOptions)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <pre>
-   * Some default value of option are not proto3 default value. Use this version
-   * to determine if we should use default option value instead of proto3
-   * default value.
-   * </pre>
-   *
-   * <code>uint32 version = 5;</code>
-   */
-  int getVersion();
-
-  /**
-   * <pre>
-   * Device type to profile/trace: (version &gt;= 1)
-   * DeviceType::UNSPECIFIED: All registered device profiler will be enabled.
-   * DeviceType::CPU: only CPU will be profiled.
-   * DeviceType::GPU: only CPU/GPU will be profiled.
-   * DeviceType::TPU: only CPU/TPU will be profiled.
-   * </pre>
-   *
-   * <code>.tensorflow.ProfileOptions.DeviceType device_type = 6;</code>
-   */
-  int getDeviceTypeValue();
-  /**
-   * <pre>
-   * Device type to profile/trace: (version &gt;= 1)
-   * DeviceType::UNSPECIFIED: All registered device profiler will be enabled.
-   * DeviceType::CPU: only CPU will be profiled.
-   * DeviceType::GPU: only CPU/GPU will be profiled.
-   * DeviceType::TPU: only CPU/TPU will be profiled.
-   * </pre>
-   *
-   * <code>.tensorflow.ProfileOptions.DeviceType device_type = 6;</code>
-   */
-  org.tensorflow.proto.profiler.ProfileOptions.DeviceType getDeviceType();
-
-  /**
-   * <pre>
-   * We don't collect the dataset ops by default for better trace-viewer
-   * scalability. The caller can mannually set this field to include the ops.
-   * </pre>
-   *
-   * <code>bool include_dataset_ops = 1;</code>
-   */
-  boolean getIncludeDatasetOps();
-
-  /**
-   * <pre>
-   * Levels of host tracing: (version &gt;= 1)
-   * - Level 0 is used to disable host traces.
-   * - Level 1 enables tracing of only user instrumented (or default) TraceMe.
-   * - Level 2 enables tracing of all level 1 TraceMe(s) and instrumented high
-   *           level program execution details (expensive TF ops, XLA ops, etc).
-   *           This is the default.
-   * - Level 3 enables tracing of all level 2 TraceMe(s) and more verbose
-   *           (low-level) program execution details (cheap TF ops, etc).
-   * </pre>
-   *
-   * <code>uint32 host_tracer_level = 2;</code>
-   */
-  int getHostTracerLevel();
-
-  /**
-   * <pre>
-   * Levels of device tracing: (version &gt;= 1)
-   * - Level 0 is used to disable device traces.
-   * - Level 1 is used to enable device traces.
-   * - More levels might be defined for specific device for controlling the
-   *   verbosity of the trace.
-   * </pre>
-   *
-   * <code>uint32 device_tracer_level = 3;</code>
-   */
-  int getDeviceTracerLevel();
-
-  /**
-   * <pre>
-   * Whether enable python function calls tracing. Runtime overhead ensues if
-   * enabled. Default off. (version &gt;= 1)
-   * </pre>
-   *
-   * <code>uint32 python_tracer_level = 4;</code>
-   */
-  int getPythonTracerLevel();
-
-  /**
-   * <pre>
-   * Whether serialize hlo_proto when XLA is used. (version &gt;= 1)
-   * </pre>
-   *
-   * <code>bool enable_hlo_proto = 7;</code>
-   */
-  boolean getEnableHloProto();
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/ProfilerOptionsProtos.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/ProfilerOptionsProtos.java
deleted file mode 100644
index 669d1ee04b5..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/ProfilerOptionsProtos.java
+++ /dev/null
@@ -1,56 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/profiler/profiler_options.proto
-
-package org.tensorflow.proto.profiler;
-
-public final class ProfilerOptionsProtos {
-  private ProfilerOptionsProtos() {}
-  public static void registerAllExtensions(
-      com.google.protobuf.ExtensionRegistryLite registry) {
-  }
-
-  public static void registerAllExtensions(
-      com.google.protobuf.ExtensionRegistry registry) {
-    registerAllExtensions(
-        (com.google.protobuf.ExtensionRegistryLite) registry);
-  }
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_ProfileOptions_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_ProfileOptions_fieldAccessorTable;
-
-  public static com.google.protobuf.Descriptors.FileDescriptor
-      getDescriptor() {
-    return descriptor;
-  }
-  private static  com.google.protobuf.Descriptors.FileDescriptor
-      descriptor;
-  static {
-    java.lang.String[] descriptorData = {
-      "\n/tensorflow/core/profiler/profiler_opti" +
-      "ons.proto\022\ntensorflow\"\243\002\n\016ProfileOptions" +
-      "\022\017\n\007version\030\005 \001(\r\022:\n\013device_type\030\006 \001(\0162%" +
-      ".tensorflow.ProfileOptions.DeviceType\022\033\n" +
-      "\023include_dataset_ops\030\001 \001(\010\022\031\n\021host_trace" +
-      "r_level\030\002 \001(\r\022\033\n\023device_tracer_level\030\003 \001" +
-      "(\r\022\033\n\023python_tracer_level\030\004 \001(\r\022\030\n\020enabl" +
-      "e_hlo_proto\030\007 \001(\010\"8\n\nDeviceType\022\017\n\013UNSPE" +
-      "CIFIED\020\000\022\007\n\003CPU\020\001\022\007\n\003GPU\020\002\022\007\n\003TPU\020\003B8\n\035o" +
-      "rg.tensorflow.proto.profilerB\025ProfilerOp" +
-      "tionsProtosP\001b\006proto3"
-    };
-    descriptor = com.google.protobuf.Descriptors.FileDescriptor
-      .internalBuildGeneratedFileFrom(descriptorData,
-        new com.google.protobuf.Descriptors.FileDescriptor[] {
-        });
-    internal_static_tensorflow_ProfileOptions_descriptor =
-      getDescriptor().getMessageTypes().get(0);
-    internal_static_tensorflow_ProfileOptions_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_ProfileOptions_descriptor,
-        new java.lang.String[] { "Version", "DeviceType", "IncludeDatasetOps", "HostTracerLevel", "DeviceTracerLevel", "PythonTracerLevel", "EnableHloProto", });
-  }
-
-  // @@protoc_insertion_point(outer_class_scope)
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/XEvent.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/XEvent.java
deleted file mode 100644
index 2921881fee2..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/XEvent.java
+++ /dev/null
@@ -1,1263 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/profiler/protobuf/xplane.proto
-
-package org.tensorflow.proto.profiler;
-
-/**
- * <pre>
- * An XEvent is a trace event, optionally annotated with XStats.
- * Next ID: 6
- * </pre>
- *
- * Protobuf type {@code tensorflow.profiler.XEvent}
- */
-public  final class XEvent extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.profiler.XEvent)
-    XEventOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use XEvent.newBuilder() to construct.
-  private XEvent(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private XEvent() {
-    stats_ = java.util.Collections.emptyList();
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new XEvent();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private XEvent(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    int mutable_bitField0_ = 0;
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 8: {
-
-            metadataId_ = input.readInt64();
-            break;
-          }
-          case 16: {
-            dataCase_ = 2;
-            data_ = input.readInt64();
-            break;
-          }
-          case 24: {
-
-            durationPs_ = input.readInt64();
-            break;
-          }
-          case 34: {
-            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
-              stats_ = new java.util.ArrayList<org.tensorflow.proto.profiler.XStat>();
-              mutable_bitField0_ |= 0x00000001;
-            }
-            stats_.add(
-                input.readMessage(org.tensorflow.proto.profiler.XStat.parser(), extensionRegistry));
-            break;
-          }
-          case 40: {
-            dataCase_ = 5;
-            data_ = input.readInt64();
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      if (((mutable_bitField0_ & 0x00000001) != 0)) {
-        stats_ = java.util.Collections.unmodifiableList(stats_);
-      }
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.profiler.XPlaneProtos.internal_static_tensorflow_profiler_XEvent_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.profiler.XPlaneProtos.internal_static_tensorflow_profiler_XEvent_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.profiler.XEvent.class, org.tensorflow.proto.profiler.XEvent.Builder.class);
-  }
-
-  private int dataCase_ = 0;
-  private java.lang.Object data_;
-  public enum DataCase
-      implements com.google.protobuf.Internal.EnumLite {
-    OFFSET_PS(2),
-    NUM_OCCURRENCES(5),
-    DATA_NOT_SET(0);
-    private final int value;
-    private DataCase(int value) {
-      this.value = value;
-    }
-    /**
-     * @deprecated Use {@link #forNumber(int)} instead.
-     */
-    @java.lang.Deprecated
-    public static DataCase valueOf(int value) {
-      return forNumber(value);
-    }
-
-    public static DataCase forNumber(int value) {
-      switch (value) {
-        case 2: return OFFSET_PS;
-        case 5: return NUM_OCCURRENCES;
-        case 0: return DATA_NOT_SET;
-        default: return null;
-      }
-    }
-    public int getNumber() {
-      return this.value;
-    }
-  };
-
-  public DataCase
-  getDataCase() {
-    return DataCase.forNumber(
-        dataCase_);
-  }
-
-  public static final int METADATA_ID_FIELD_NUMBER = 1;
-  private long metadataId_;
-  /**
-   * <pre>
-   * XEventMetadata.id of corresponding metadata.
-   * </pre>
-   *
-   * <code>int64 metadata_id = 1;</code>
-   */
-  public long getMetadataId() {
-    return metadataId_;
-  }
-
-  public static final int OFFSET_PS_FIELD_NUMBER = 2;
-  /**
-   * <pre>
-   * Start time of the event in picoseconds, as offset from
-   * XLine.timestamp_ns().
-   * </pre>
-   *
-   * <code>int64 offset_ps = 2;</code>
-   */
-  public long getOffsetPs() {
-    if (dataCase_ == 2) {
-      return (java.lang.Long) data_;
-    }
-    return 0L;
-  }
-
-  public static final int NUM_OCCURRENCES_FIELD_NUMBER = 5;
-  /**
-   * <pre>
-   * Number of occurrences of the event, if aggregated.
-   * </pre>
-   *
-   * <code>int64 num_occurrences = 5;</code>
-   */
-  public long getNumOccurrences() {
-    if (dataCase_ == 5) {
-      return (java.lang.Long) data_;
-    }
-    return 0L;
-  }
-
-  public static final int DURATION_PS_FIELD_NUMBER = 3;
-  private long durationPs_;
-  /**
-   * <pre>
-   * Duration of the event in picoseconds. Can be zero for an instant event.
-   * </pre>
-   *
-   * <code>int64 duration_ps = 3;</code>
-   */
-  public long getDurationPs() {
-    return durationPs_;
-  }
-
-  public static final int STATS_FIELD_NUMBER = 4;
-  private java.util.List<org.tensorflow.proto.profiler.XStat> stats_;
-  /**
-   * <pre>
-   * XStats associated with the event.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.profiler.XStat stats = 4;</code>
-   */
-  public java.util.List<org.tensorflow.proto.profiler.XStat> getStatsList() {
-    return stats_;
-  }
-  /**
-   * <pre>
-   * XStats associated with the event.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.profiler.XStat stats = 4;</code>
-   */
-  public java.util.List<? extends org.tensorflow.proto.profiler.XStatOrBuilder> 
-      getStatsOrBuilderList() {
-    return stats_;
-  }
-  /**
-   * <pre>
-   * XStats associated with the event.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.profiler.XStat stats = 4;</code>
-   */
-  public int getStatsCount() {
-    return stats_.size();
-  }
-  /**
-   * <pre>
-   * XStats associated with the event.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.profiler.XStat stats = 4;</code>
-   */
-  public org.tensorflow.proto.profiler.XStat getStats(int index) {
-    return stats_.get(index);
-  }
-  /**
-   * <pre>
-   * XStats associated with the event.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.profiler.XStat stats = 4;</code>
-   */
-  public org.tensorflow.proto.profiler.XStatOrBuilder getStatsOrBuilder(
-      int index) {
-    return stats_.get(index);
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    if (metadataId_ != 0L) {
-      output.writeInt64(1, metadataId_);
-    }
-    if (dataCase_ == 2) {
-      output.writeInt64(
-          2, (long)((java.lang.Long) data_));
-    }
-    if (durationPs_ != 0L) {
-      output.writeInt64(3, durationPs_);
-    }
-    for (int i = 0; i < stats_.size(); i++) {
-      output.writeMessage(4, stats_.get(i));
-    }
-    if (dataCase_ == 5) {
-      output.writeInt64(
-          5, (long)((java.lang.Long) data_));
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (metadataId_ != 0L) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt64Size(1, metadataId_);
-    }
-    if (dataCase_ == 2) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt64Size(
-            2, (long)((java.lang.Long) data_));
-    }
-    if (durationPs_ != 0L) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt64Size(3, durationPs_);
-    }
-    for (int i = 0; i < stats_.size(); i++) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(4, stats_.get(i));
-    }
-    if (dataCase_ == 5) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt64Size(
-            5, (long)((java.lang.Long) data_));
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.profiler.XEvent)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.profiler.XEvent other = (org.tensorflow.proto.profiler.XEvent) obj;
-
-    if (getMetadataId()
-        != other.getMetadataId()) return false;
-    if (getDurationPs()
-        != other.getDurationPs()) return false;
-    if (!getStatsList()
-        .equals(other.getStatsList())) return false;
-    if (!getDataCase().equals(other.getDataCase())) return false;
-    switch (dataCase_) {
-      case 2:
-        if (getOffsetPs()
-            != other.getOffsetPs()) return false;
-        break;
-      case 5:
-        if (getNumOccurrences()
-            != other.getNumOccurrences()) return false;
-        break;
-      case 0:
-      default:
-    }
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    hash = (37 * hash) + METADATA_ID_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-        getMetadataId());
-    hash = (37 * hash) + DURATION_PS_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-        getDurationPs());
-    if (getStatsCount() > 0) {
-      hash = (37 * hash) + STATS_FIELD_NUMBER;
-      hash = (53 * hash) + getStatsList().hashCode();
-    }
-    switch (dataCase_) {
-      case 2:
-        hash = (37 * hash) + OFFSET_PS_FIELD_NUMBER;
-        hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-            getOffsetPs());
-        break;
-      case 5:
-        hash = (37 * hash) + NUM_OCCURRENCES_FIELD_NUMBER;
-        hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-            getNumOccurrences());
-        break;
-      case 0:
-      default:
-    }
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.profiler.XEvent parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.profiler.XEvent parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.profiler.XEvent parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.profiler.XEvent parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.profiler.XEvent parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.profiler.XEvent parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.profiler.XEvent parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.profiler.XEvent parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.profiler.XEvent parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.profiler.XEvent parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.profiler.XEvent parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.profiler.XEvent parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.profiler.XEvent prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * <pre>
-   * An XEvent is a trace event, optionally annotated with XStats.
-   * Next ID: 6
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.profiler.XEvent}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.profiler.XEvent)
-      org.tensorflow.proto.profiler.XEventOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.profiler.XPlaneProtos.internal_static_tensorflow_profiler_XEvent_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.profiler.XPlaneProtos.internal_static_tensorflow_profiler_XEvent_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.profiler.XEvent.class, org.tensorflow.proto.profiler.XEvent.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.profiler.XEvent.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-        getStatsFieldBuilder();
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      metadataId_ = 0L;
-
-      durationPs_ = 0L;
-
-      if (statsBuilder_ == null) {
-        stats_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000001);
-      } else {
-        statsBuilder_.clear();
-      }
-      dataCase_ = 0;
-      data_ = null;
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.profiler.XPlaneProtos.internal_static_tensorflow_profiler_XEvent_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.profiler.XEvent getDefaultInstanceForType() {
-      return org.tensorflow.proto.profiler.XEvent.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.profiler.XEvent build() {
-      org.tensorflow.proto.profiler.XEvent result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.profiler.XEvent buildPartial() {
-      org.tensorflow.proto.profiler.XEvent result = new org.tensorflow.proto.profiler.XEvent(this);
-      int from_bitField0_ = bitField0_;
-      result.metadataId_ = metadataId_;
-      if (dataCase_ == 2) {
-        result.data_ = data_;
-      }
-      if (dataCase_ == 5) {
-        result.data_ = data_;
-      }
-      result.durationPs_ = durationPs_;
-      if (statsBuilder_ == null) {
-        if (((bitField0_ & 0x00000001) != 0)) {
-          stats_ = java.util.Collections.unmodifiableList(stats_);
-          bitField0_ = (bitField0_ & ~0x00000001);
-        }
-        result.stats_ = stats_;
-      } else {
-        result.stats_ = statsBuilder_.build();
-      }
-      result.dataCase_ = dataCase_;
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.profiler.XEvent) {
-        return mergeFrom((org.tensorflow.proto.profiler.XEvent)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.profiler.XEvent other) {
-      if (other == org.tensorflow.proto.profiler.XEvent.getDefaultInstance()) return this;
-      if (other.getMetadataId() != 0L) {
-        setMetadataId(other.getMetadataId());
-      }
-      if (other.getDurationPs() != 0L) {
-        setDurationPs(other.getDurationPs());
-      }
-      if (statsBuilder_ == null) {
-        if (!other.stats_.isEmpty()) {
-          if (stats_.isEmpty()) {
-            stats_ = other.stats_;
-            bitField0_ = (bitField0_ & ~0x00000001);
-          } else {
-            ensureStatsIsMutable();
-            stats_.addAll(other.stats_);
-          }
-          onChanged();
-        }
-      } else {
-        if (!other.stats_.isEmpty()) {
-          if (statsBuilder_.isEmpty()) {
-            statsBuilder_.dispose();
-            statsBuilder_ = null;
-            stats_ = other.stats_;
-            bitField0_ = (bitField0_ & ~0x00000001);
-            statsBuilder_ = 
-              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
-                 getStatsFieldBuilder() : null;
-          } else {
-            statsBuilder_.addAllMessages(other.stats_);
-          }
-        }
-      }
-      switch (other.getDataCase()) {
-        case OFFSET_PS: {
-          setOffsetPs(other.getOffsetPs());
-          break;
-        }
-        case NUM_OCCURRENCES: {
-          setNumOccurrences(other.getNumOccurrences());
-          break;
-        }
-        case DATA_NOT_SET: {
-          break;
-        }
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.profiler.XEvent parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.profiler.XEvent) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-    private int dataCase_ = 0;
-    private java.lang.Object data_;
-    public DataCase
-        getDataCase() {
-      return DataCase.forNumber(
-          dataCase_);
-    }
-
-    public Builder clearData() {
-      dataCase_ = 0;
-      data_ = null;
-      onChanged();
-      return this;
-    }
-
-    private int bitField0_;
-
-    private long metadataId_ ;
-    /**
-     * <pre>
-     * XEventMetadata.id of corresponding metadata.
-     * </pre>
-     *
-     * <code>int64 metadata_id = 1;</code>
-     */
-    public long getMetadataId() {
-      return metadataId_;
-    }
-    /**
-     * <pre>
-     * XEventMetadata.id of corresponding metadata.
-     * </pre>
-     *
-     * <code>int64 metadata_id = 1;</code>
-     */
-    public Builder setMetadataId(long value) {
-      
-      metadataId_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * XEventMetadata.id of corresponding metadata.
-     * </pre>
-     *
-     * <code>int64 metadata_id = 1;</code>
-     */
-    public Builder clearMetadataId() {
-      
-      metadataId_ = 0L;
-      onChanged();
-      return this;
-    }
-
-    /**
-     * <pre>
-     * Start time of the event in picoseconds, as offset from
-     * XLine.timestamp_ns().
-     * </pre>
-     *
-     * <code>int64 offset_ps = 2;</code>
-     */
-    public long getOffsetPs() {
-      if (dataCase_ == 2) {
-        return (java.lang.Long) data_;
-      }
-      return 0L;
-    }
-    /**
-     * <pre>
-     * Start time of the event in picoseconds, as offset from
-     * XLine.timestamp_ns().
-     * </pre>
-     *
-     * <code>int64 offset_ps = 2;</code>
-     */
-    public Builder setOffsetPs(long value) {
-      dataCase_ = 2;
-      data_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Start time of the event in picoseconds, as offset from
-     * XLine.timestamp_ns().
-     * </pre>
-     *
-     * <code>int64 offset_ps = 2;</code>
-     */
-    public Builder clearOffsetPs() {
-      if (dataCase_ == 2) {
-        dataCase_ = 0;
-        data_ = null;
-        onChanged();
-      }
-      return this;
-    }
-
-    /**
-     * <pre>
-     * Number of occurrences of the event, if aggregated.
-     * </pre>
-     *
-     * <code>int64 num_occurrences = 5;</code>
-     */
-    public long getNumOccurrences() {
-      if (dataCase_ == 5) {
-        return (java.lang.Long) data_;
-      }
-      return 0L;
-    }
-    /**
-     * <pre>
-     * Number of occurrences of the event, if aggregated.
-     * </pre>
-     *
-     * <code>int64 num_occurrences = 5;</code>
-     */
-    public Builder setNumOccurrences(long value) {
-      dataCase_ = 5;
-      data_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Number of occurrences of the event, if aggregated.
-     * </pre>
-     *
-     * <code>int64 num_occurrences = 5;</code>
-     */
-    public Builder clearNumOccurrences() {
-      if (dataCase_ == 5) {
-        dataCase_ = 0;
-        data_ = null;
-        onChanged();
-      }
-      return this;
-    }
-
-    private long durationPs_ ;
-    /**
-     * <pre>
-     * Duration of the event in picoseconds. Can be zero for an instant event.
-     * </pre>
-     *
-     * <code>int64 duration_ps = 3;</code>
-     */
-    public long getDurationPs() {
-      return durationPs_;
-    }
-    /**
-     * <pre>
-     * Duration of the event in picoseconds. Can be zero for an instant event.
-     * </pre>
-     *
-     * <code>int64 duration_ps = 3;</code>
-     */
-    public Builder setDurationPs(long value) {
-      
-      durationPs_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Duration of the event in picoseconds. Can be zero for an instant event.
-     * </pre>
-     *
-     * <code>int64 duration_ps = 3;</code>
-     */
-    public Builder clearDurationPs() {
-      
-      durationPs_ = 0L;
-      onChanged();
-      return this;
-    }
-
-    private java.util.List<org.tensorflow.proto.profiler.XStat> stats_ =
-      java.util.Collections.emptyList();
-    private void ensureStatsIsMutable() {
-      if (!((bitField0_ & 0x00000001) != 0)) {
-        stats_ = new java.util.ArrayList<org.tensorflow.proto.profiler.XStat>(stats_);
-        bitField0_ |= 0x00000001;
-       }
-    }
-
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.profiler.XStat, org.tensorflow.proto.profiler.XStat.Builder, org.tensorflow.proto.profiler.XStatOrBuilder> statsBuilder_;
-
-    /**
-     * <pre>
-     * XStats associated with the event.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.profiler.XStat stats = 4;</code>
-     */
-    public java.util.List<org.tensorflow.proto.profiler.XStat> getStatsList() {
-      if (statsBuilder_ == null) {
-        return java.util.Collections.unmodifiableList(stats_);
-      } else {
-        return statsBuilder_.getMessageList();
-      }
-    }
-    /**
-     * <pre>
-     * XStats associated with the event.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.profiler.XStat stats = 4;</code>
-     */
-    public int getStatsCount() {
-      if (statsBuilder_ == null) {
-        return stats_.size();
-      } else {
-        return statsBuilder_.getCount();
-      }
-    }
-    /**
-     * <pre>
-     * XStats associated with the event.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.profiler.XStat stats = 4;</code>
-     */
-    public org.tensorflow.proto.profiler.XStat getStats(int index) {
-      if (statsBuilder_ == null) {
-        return stats_.get(index);
-      } else {
-        return statsBuilder_.getMessage(index);
-      }
-    }
-    /**
-     * <pre>
-     * XStats associated with the event.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.profiler.XStat stats = 4;</code>
-     */
-    public Builder setStats(
-        int index, org.tensorflow.proto.profiler.XStat value) {
-      if (statsBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureStatsIsMutable();
-        stats_.set(index, value);
-        onChanged();
-      } else {
-        statsBuilder_.setMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * XStats associated with the event.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.profiler.XStat stats = 4;</code>
-     */
-    public Builder setStats(
-        int index, org.tensorflow.proto.profiler.XStat.Builder builderForValue) {
-      if (statsBuilder_ == null) {
-        ensureStatsIsMutable();
-        stats_.set(index, builderForValue.build());
-        onChanged();
-      } else {
-        statsBuilder_.setMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * XStats associated with the event.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.profiler.XStat stats = 4;</code>
-     */
-    public Builder addStats(org.tensorflow.proto.profiler.XStat value) {
-      if (statsBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureStatsIsMutable();
-        stats_.add(value);
-        onChanged();
-      } else {
-        statsBuilder_.addMessage(value);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * XStats associated with the event.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.profiler.XStat stats = 4;</code>
-     */
-    public Builder addStats(
-        int index, org.tensorflow.proto.profiler.XStat value) {
-      if (statsBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureStatsIsMutable();
-        stats_.add(index, value);
-        onChanged();
-      } else {
-        statsBuilder_.addMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * XStats associated with the event.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.profiler.XStat stats = 4;</code>
-     */
-    public Builder addStats(
-        org.tensorflow.proto.profiler.XStat.Builder builderForValue) {
-      if (statsBuilder_ == null) {
-        ensureStatsIsMutable();
-        stats_.add(builderForValue.build());
-        onChanged();
-      } else {
-        statsBuilder_.addMessage(builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * XStats associated with the event.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.profiler.XStat stats = 4;</code>
-     */
-    public Builder addStats(
-        int index, org.tensorflow.proto.profiler.XStat.Builder builderForValue) {
-      if (statsBuilder_ == null) {
-        ensureStatsIsMutable();
-        stats_.add(index, builderForValue.build());
-        onChanged();
-      } else {
-        statsBuilder_.addMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * XStats associated with the event.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.profiler.XStat stats = 4;</code>
-     */
-    public Builder addAllStats(
-        java.lang.Iterable<? extends org.tensorflow.proto.profiler.XStat> values) {
-      if (statsBuilder_ == null) {
-        ensureStatsIsMutable();
-        com.google.protobuf.AbstractMessageLite.Builder.addAll(
-            values, stats_);
-        onChanged();
-      } else {
-        statsBuilder_.addAllMessages(values);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * XStats associated with the event.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.profiler.XStat stats = 4;</code>
-     */
-    public Builder clearStats() {
-      if (statsBuilder_ == null) {
-        stats_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000001);
-        onChanged();
-      } else {
-        statsBuilder_.clear();
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * XStats associated with the event.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.profiler.XStat stats = 4;</code>
-     */
-    public Builder removeStats(int index) {
-      if (statsBuilder_ == null) {
-        ensureStatsIsMutable();
-        stats_.remove(index);
-        onChanged();
-      } else {
-        statsBuilder_.remove(index);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * XStats associated with the event.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.profiler.XStat stats = 4;</code>
-     */
-    public org.tensorflow.proto.profiler.XStat.Builder getStatsBuilder(
-        int index) {
-      return getStatsFieldBuilder().getBuilder(index);
-    }
-    /**
-     * <pre>
-     * XStats associated with the event.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.profiler.XStat stats = 4;</code>
-     */
-    public org.tensorflow.proto.profiler.XStatOrBuilder getStatsOrBuilder(
-        int index) {
-      if (statsBuilder_ == null) {
-        return stats_.get(index);  } else {
-        return statsBuilder_.getMessageOrBuilder(index);
-      }
-    }
-    /**
-     * <pre>
-     * XStats associated with the event.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.profiler.XStat stats = 4;</code>
-     */
-    public java.util.List<? extends org.tensorflow.proto.profiler.XStatOrBuilder> 
-         getStatsOrBuilderList() {
-      if (statsBuilder_ != null) {
-        return statsBuilder_.getMessageOrBuilderList();
-      } else {
-        return java.util.Collections.unmodifiableList(stats_);
-      }
-    }
-    /**
-     * <pre>
-     * XStats associated with the event.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.profiler.XStat stats = 4;</code>
-     */
-    public org.tensorflow.proto.profiler.XStat.Builder addStatsBuilder() {
-      return getStatsFieldBuilder().addBuilder(
-          org.tensorflow.proto.profiler.XStat.getDefaultInstance());
-    }
-    /**
-     * <pre>
-     * XStats associated with the event.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.profiler.XStat stats = 4;</code>
-     */
-    public org.tensorflow.proto.profiler.XStat.Builder addStatsBuilder(
-        int index) {
-      return getStatsFieldBuilder().addBuilder(
-          index, org.tensorflow.proto.profiler.XStat.getDefaultInstance());
-    }
-    /**
-     * <pre>
-     * XStats associated with the event.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.profiler.XStat stats = 4;</code>
-     */
-    public java.util.List<org.tensorflow.proto.profiler.XStat.Builder> 
-         getStatsBuilderList() {
-      return getStatsFieldBuilder().getBuilderList();
-    }
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.profiler.XStat, org.tensorflow.proto.profiler.XStat.Builder, org.tensorflow.proto.profiler.XStatOrBuilder> 
-        getStatsFieldBuilder() {
-      if (statsBuilder_ == null) {
-        statsBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
-            org.tensorflow.proto.profiler.XStat, org.tensorflow.proto.profiler.XStat.Builder, org.tensorflow.proto.profiler.XStatOrBuilder>(
-                stats_,
-                ((bitField0_ & 0x00000001) != 0),
-                getParentForChildren(),
-                isClean());
-        stats_ = null;
-      }
-      return statsBuilder_;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.profiler.XEvent)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.profiler.XEvent)
-  private static final org.tensorflow.proto.profiler.XEvent DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.profiler.XEvent();
-  }
-
-  public static org.tensorflow.proto.profiler.XEvent getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<XEvent>
-      PARSER = new com.google.protobuf.AbstractParser<XEvent>() {
-    @java.lang.Override
-    public XEvent parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new XEvent(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<XEvent> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<XEvent> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.profiler.XEvent getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/XEventMetadata.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/XEventMetadata.java
deleted file mode 100644
index c7551d7f4f5..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/XEventMetadata.java
+++ /dev/null
@@ -1,890 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/profiler/protobuf/xplane.proto
-
-package org.tensorflow.proto.profiler;
-
-/**
- * <pre>
- * Metadata for an XEvent, shared by all instances of the same event.
- * Next ID: 5
- * </pre>
- *
- * Protobuf type {@code tensorflow.profiler.XEventMetadata}
- */
-public  final class XEventMetadata extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.profiler.XEventMetadata)
-    XEventMetadataOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use XEventMetadata.newBuilder() to construct.
-  private XEventMetadata(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private XEventMetadata() {
-    name_ = "";
-    displayName_ = "";
-    metadata_ = com.google.protobuf.ByteString.EMPTY;
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new XEventMetadata();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private XEventMetadata(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 8: {
-
-            id_ = input.readInt64();
-            break;
-          }
-          case 18: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            name_ = s;
-            break;
-          }
-          case 26: {
-
-            metadata_ = input.readBytes();
-            break;
-          }
-          case 34: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            displayName_ = s;
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.profiler.XPlaneProtos.internal_static_tensorflow_profiler_XEventMetadata_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.profiler.XPlaneProtos.internal_static_tensorflow_profiler_XEventMetadata_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.profiler.XEventMetadata.class, org.tensorflow.proto.profiler.XEventMetadata.Builder.class);
-  }
-
-  public static final int ID_FIELD_NUMBER = 1;
-  private long id_;
-  /**
-   * <pre>
-   * XPlane.event_metadata map key.
-   * </pre>
-   *
-   * <code>int64 id = 1;</code>
-   */
-  public long getId() {
-    return id_;
-  }
-
-  public static final int NAME_FIELD_NUMBER = 2;
-  private volatile java.lang.Object name_;
-  /**
-   * <pre>
-   * Name of the event.
-   * </pre>
-   *
-   * <code>string name = 2;</code>
-   */
-  public java.lang.String getName() {
-    java.lang.Object ref = name_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      name_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * Name of the event.
-   * </pre>
-   *
-   * <code>string name = 2;</code>
-   */
-  public com.google.protobuf.ByteString
-      getNameBytes() {
-    java.lang.Object ref = name_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      name_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int DISPLAY_NAME_FIELD_NUMBER = 4;
-  private volatile java.lang.Object displayName_;
-  /**
-   * <pre>
-   * Name of the event shown in trace viewer.
-   * </pre>
-   *
-   * <code>string display_name = 4;</code>
-   */
-  public java.lang.String getDisplayName() {
-    java.lang.Object ref = displayName_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      displayName_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * Name of the event shown in trace viewer.
-   * </pre>
-   *
-   * <code>string display_name = 4;</code>
-   */
-  public com.google.protobuf.ByteString
-      getDisplayNameBytes() {
-    java.lang.Object ref = displayName_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      displayName_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int METADATA_FIELD_NUMBER = 3;
-  private com.google.protobuf.ByteString metadata_;
-  /**
-   * <pre>
-   * Additional metadata in serialized format.
-   * </pre>
-   *
-   * <code>bytes metadata = 3;</code>
-   */
-  public com.google.protobuf.ByteString getMetadata() {
-    return metadata_;
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    if (id_ != 0L) {
-      output.writeInt64(1, id_);
-    }
-    if (!getNameBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 2, name_);
-    }
-    if (!metadata_.isEmpty()) {
-      output.writeBytes(3, metadata_);
-    }
-    if (!getDisplayNameBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 4, displayName_);
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (id_ != 0L) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt64Size(1, id_);
-    }
-    if (!getNameBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(2, name_);
-    }
-    if (!metadata_.isEmpty()) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeBytesSize(3, metadata_);
-    }
-    if (!getDisplayNameBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(4, displayName_);
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.profiler.XEventMetadata)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.profiler.XEventMetadata other = (org.tensorflow.proto.profiler.XEventMetadata) obj;
-
-    if (getId()
-        != other.getId()) return false;
-    if (!getName()
-        .equals(other.getName())) return false;
-    if (!getDisplayName()
-        .equals(other.getDisplayName())) return false;
-    if (!getMetadata()
-        .equals(other.getMetadata())) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    hash = (37 * hash) + ID_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-        getId());
-    hash = (37 * hash) + NAME_FIELD_NUMBER;
-    hash = (53 * hash) + getName().hashCode();
-    hash = (37 * hash) + DISPLAY_NAME_FIELD_NUMBER;
-    hash = (53 * hash) + getDisplayName().hashCode();
-    hash = (37 * hash) + METADATA_FIELD_NUMBER;
-    hash = (53 * hash) + getMetadata().hashCode();
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.profiler.XEventMetadata parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.profiler.XEventMetadata parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.profiler.XEventMetadata parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.profiler.XEventMetadata parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.profiler.XEventMetadata parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.profiler.XEventMetadata parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.profiler.XEventMetadata parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.profiler.XEventMetadata parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.profiler.XEventMetadata parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.profiler.XEventMetadata parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.profiler.XEventMetadata parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.profiler.XEventMetadata parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.profiler.XEventMetadata prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * <pre>
-   * Metadata for an XEvent, shared by all instances of the same event.
-   * Next ID: 5
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.profiler.XEventMetadata}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.profiler.XEventMetadata)
-      org.tensorflow.proto.profiler.XEventMetadataOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.profiler.XPlaneProtos.internal_static_tensorflow_profiler_XEventMetadata_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.profiler.XPlaneProtos.internal_static_tensorflow_profiler_XEventMetadata_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.profiler.XEventMetadata.class, org.tensorflow.proto.profiler.XEventMetadata.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.profiler.XEventMetadata.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      id_ = 0L;
-
-      name_ = "";
-
-      displayName_ = "";
-
-      metadata_ = com.google.protobuf.ByteString.EMPTY;
-
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.profiler.XPlaneProtos.internal_static_tensorflow_profiler_XEventMetadata_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.profiler.XEventMetadata getDefaultInstanceForType() {
-      return org.tensorflow.proto.profiler.XEventMetadata.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.profiler.XEventMetadata build() {
-      org.tensorflow.proto.profiler.XEventMetadata result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.profiler.XEventMetadata buildPartial() {
-      org.tensorflow.proto.profiler.XEventMetadata result = new org.tensorflow.proto.profiler.XEventMetadata(this);
-      result.id_ = id_;
-      result.name_ = name_;
-      result.displayName_ = displayName_;
-      result.metadata_ = metadata_;
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.profiler.XEventMetadata) {
-        return mergeFrom((org.tensorflow.proto.profiler.XEventMetadata)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.profiler.XEventMetadata other) {
-      if (other == org.tensorflow.proto.profiler.XEventMetadata.getDefaultInstance()) return this;
-      if (other.getId() != 0L) {
-        setId(other.getId());
-      }
-      if (!other.getName().isEmpty()) {
-        name_ = other.name_;
-        onChanged();
-      }
-      if (!other.getDisplayName().isEmpty()) {
-        displayName_ = other.displayName_;
-        onChanged();
-      }
-      if (other.getMetadata() != com.google.protobuf.ByteString.EMPTY) {
-        setMetadata(other.getMetadata());
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.profiler.XEventMetadata parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.profiler.XEventMetadata) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-
-    private long id_ ;
-    /**
-     * <pre>
-     * XPlane.event_metadata map key.
-     * </pre>
-     *
-     * <code>int64 id = 1;</code>
-     */
-    public long getId() {
-      return id_;
-    }
-    /**
-     * <pre>
-     * XPlane.event_metadata map key.
-     * </pre>
-     *
-     * <code>int64 id = 1;</code>
-     */
-    public Builder setId(long value) {
-      
-      id_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * XPlane.event_metadata map key.
-     * </pre>
-     *
-     * <code>int64 id = 1;</code>
-     */
-    public Builder clearId() {
-      
-      id_ = 0L;
-      onChanged();
-      return this;
-    }
-
-    private java.lang.Object name_ = "";
-    /**
-     * <pre>
-     * Name of the event.
-     * </pre>
-     *
-     * <code>string name = 2;</code>
-     */
-    public java.lang.String getName() {
-      java.lang.Object ref = name_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        name_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Name of the event.
-     * </pre>
-     *
-     * <code>string name = 2;</code>
-     */
-    public com.google.protobuf.ByteString
-        getNameBytes() {
-      java.lang.Object ref = name_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        name_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Name of the event.
-     * </pre>
-     *
-     * <code>string name = 2;</code>
-     */
-    public Builder setName(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      name_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Name of the event.
-     * </pre>
-     *
-     * <code>string name = 2;</code>
-     */
-    public Builder clearName() {
-      
-      name_ = getDefaultInstance().getName();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Name of the event.
-     * </pre>
-     *
-     * <code>string name = 2;</code>
-     */
-    public Builder setNameBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      name_ = value;
-      onChanged();
-      return this;
-    }
-
-    private java.lang.Object displayName_ = "";
-    /**
-     * <pre>
-     * Name of the event shown in trace viewer.
-     * </pre>
-     *
-     * <code>string display_name = 4;</code>
-     */
-    public java.lang.String getDisplayName() {
-      java.lang.Object ref = displayName_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        displayName_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Name of the event shown in trace viewer.
-     * </pre>
-     *
-     * <code>string display_name = 4;</code>
-     */
-    public com.google.protobuf.ByteString
-        getDisplayNameBytes() {
-      java.lang.Object ref = displayName_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        displayName_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Name of the event shown in trace viewer.
-     * </pre>
-     *
-     * <code>string display_name = 4;</code>
-     */
-    public Builder setDisplayName(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      displayName_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Name of the event shown in trace viewer.
-     * </pre>
-     *
-     * <code>string display_name = 4;</code>
-     */
-    public Builder clearDisplayName() {
-      
-      displayName_ = getDefaultInstance().getDisplayName();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Name of the event shown in trace viewer.
-     * </pre>
-     *
-     * <code>string display_name = 4;</code>
-     */
-    public Builder setDisplayNameBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      displayName_ = value;
-      onChanged();
-      return this;
-    }
-
-    private com.google.protobuf.ByteString metadata_ = com.google.protobuf.ByteString.EMPTY;
-    /**
-     * <pre>
-     * Additional metadata in serialized format.
-     * </pre>
-     *
-     * <code>bytes metadata = 3;</code>
-     */
-    public com.google.protobuf.ByteString getMetadata() {
-      return metadata_;
-    }
-    /**
-     * <pre>
-     * Additional metadata in serialized format.
-     * </pre>
-     *
-     * <code>bytes metadata = 3;</code>
-     */
-    public Builder setMetadata(com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      metadata_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Additional metadata in serialized format.
-     * </pre>
-     *
-     * <code>bytes metadata = 3;</code>
-     */
-    public Builder clearMetadata() {
-      
-      metadata_ = getDefaultInstance().getMetadata();
-      onChanged();
-      return this;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.profiler.XEventMetadata)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.profiler.XEventMetadata)
-  private static final org.tensorflow.proto.profiler.XEventMetadata DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.profiler.XEventMetadata();
-  }
-
-  public static org.tensorflow.proto.profiler.XEventMetadata getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<XEventMetadata>
-      PARSER = new com.google.protobuf.AbstractParser<XEventMetadata>() {
-    @java.lang.Override
-    public XEventMetadata parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new XEventMetadata(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<XEventMetadata> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<XEventMetadata> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.profiler.XEventMetadata getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/XEventMetadataOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/XEventMetadataOrBuilder.java
deleted file mode 100644
index b150f6e1850..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/XEventMetadataOrBuilder.java
+++ /dev/null
@@ -1,63 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/profiler/protobuf/xplane.proto
-
-package org.tensorflow.proto.profiler;
-
-public interface XEventMetadataOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.profiler.XEventMetadata)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <pre>
-   * XPlane.event_metadata map key.
-   * </pre>
-   *
-   * <code>int64 id = 1;</code>
-   */
-  long getId();
-
-  /**
-   * <pre>
-   * Name of the event.
-   * </pre>
-   *
-   * <code>string name = 2;</code>
-   */
-  java.lang.String getName();
-  /**
-   * <pre>
-   * Name of the event.
-   * </pre>
-   *
-   * <code>string name = 2;</code>
-   */
-  com.google.protobuf.ByteString
-      getNameBytes();
-
-  /**
-   * <pre>
-   * Name of the event shown in trace viewer.
-   * </pre>
-   *
-   * <code>string display_name = 4;</code>
-   */
-  java.lang.String getDisplayName();
-  /**
-   * <pre>
-   * Name of the event shown in trace viewer.
-   * </pre>
-   *
-   * <code>string display_name = 4;</code>
-   */
-  com.google.protobuf.ByteString
-      getDisplayNameBytes();
-
-  /**
-   * <pre>
-   * Additional metadata in serialized format.
-   * </pre>
-   *
-   * <code>bytes metadata = 3;</code>
-   */
-  com.google.protobuf.ByteString getMetadata();
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/XEventOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/XEventOrBuilder.java
deleted file mode 100644
index 3d9e4c2eec5..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/XEventOrBuilder.java
+++ /dev/null
@@ -1,92 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/profiler/protobuf/xplane.proto
-
-package org.tensorflow.proto.profiler;
-
-public interface XEventOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.profiler.XEvent)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <pre>
-   * XEventMetadata.id of corresponding metadata.
-   * </pre>
-   *
-   * <code>int64 metadata_id = 1;</code>
-   */
-  long getMetadataId();
-
-  /**
-   * <pre>
-   * Start time of the event in picoseconds, as offset from
-   * XLine.timestamp_ns().
-   * </pre>
-   *
-   * <code>int64 offset_ps = 2;</code>
-   */
-  long getOffsetPs();
-
-  /**
-   * <pre>
-   * Number of occurrences of the event, if aggregated.
-   * </pre>
-   *
-   * <code>int64 num_occurrences = 5;</code>
-   */
-  long getNumOccurrences();
-
-  /**
-   * <pre>
-   * Duration of the event in picoseconds. Can be zero for an instant event.
-   * </pre>
-   *
-   * <code>int64 duration_ps = 3;</code>
-   */
-  long getDurationPs();
-
-  /**
-   * <pre>
-   * XStats associated with the event.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.profiler.XStat stats = 4;</code>
-   */
-  java.util.List<org.tensorflow.proto.profiler.XStat> 
-      getStatsList();
-  /**
-   * <pre>
-   * XStats associated with the event.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.profiler.XStat stats = 4;</code>
-   */
-  org.tensorflow.proto.profiler.XStat getStats(int index);
-  /**
-   * <pre>
-   * XStats associated with the event.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.profiler.XStat stats = 4;</code>
-   */
-  int getStatsCount();
-  /**
-   * <pre>
-   * XStats associated with the event.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.profiler.XStat stats = 4;</code>
-   */
-  java.util.List<? extends org.tensorflow.proto.profiler.XStatOrBuilder> 
-      getStatsOrBuilderList();
-  /**
-   * <pre>
-   * XStats associated with the event.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.profiler.XStat stats = 4;</code>
-   */
-  org.tensorflow.proto.profiler.XStatOrBuilder getStatsOrBuilder(
-      int index);
-
-  public org.tensorflow.proto.profiler.XEvent.DataCase getDataCase();
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/XLine.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/XLine.java
deleted file mode 100644
index e9342bbb66f..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/XLine.java
+++ /dev/null
@@ -1,1508 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/profiler/protobuf/xplane.proto
-
-package org.tensorflow.proto.profiler;
-
-/**
- * <pre>
- * An XLine is a timeline of trace events (XEvents).
- * Next ID: 12
- * </pre>
- *
- * Protobuf type {@code tensorflow.profiler.XLine}
- */
-public  final class XLine extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.profiler.XLine)
-    XLineOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use XLine.newBuilder() to construct.
-  private XLine(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private XLine() {
-    name_ = "";
-    displayName_ = "";
-    events_ = java.util.Collections.emptyList();
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new XLine();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private XLine(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    int mutable_bitField0_ = 0;
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 8: {
-
-            id_ = input.readInt64();
-            break;
-          }
-          case 18: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            name_ = s;
-            break;
-          }
-          case 24: {
-
-            timestampNs_ = input.readInt64();
-            break;
-          }
-          case 34: {
-            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
-              events_ = new java.util.ArrayList<org.tensorflow.proto.profiler.XEvent>();
-              mutable_bitField0_ |= 0x00000001;
-            }
-            events_.add(
-                input.readMessage(org.tensorflow.proto.profiler.XEvent.parser(), extensionRegistry));
-            break;
-          }
-          case 72: {
-
-            durationPs_ = input.readInt64();
-            break;
-          }
-          case 80: {
-
-            displayId_ = input.readInt64();
-            break;
-          }
-          case 90: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            displayName_ = s;
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      if (((mutable_bitField0_ & 0x00000001) != 0)) {
-        events_ = java.util.Collections.unmodifiableList(events_);
-      }
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.profiler.XPlaneProtos.internal_static_tensorflow_profiler_XLine_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.profiler.XPlaneProtos.internal_static_tensorflow_profiler_XLine_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.profiler.XLine.class, org.tensorflow.proto.profiler.XLine.Builder.class);
-  }
-
-  public static final int ID_FIELD_NUMBER = 1;
-  private long id_;
-  /**
-   * <pre>
-   * Id of this line, can be repeated within an XPlane. All XLines with the
-   * same id are effectively the same timeline.
-   * </pre>
-   *
-   * <code>int64 id = 1;</code>
-   */
-  public long getId() {
-    return id_;
-  }
-
-  public static final int DISPLAY_ID_FIELD_NUMBER = 10;
-  private long displayId_;
-  /**
-   * <pre>
-   * Display id of this line. Multiple lines with the same display_id are
-   * grouped together in the same trace viewer row.
-   * </pre>
-   *
-   * <code>int64 display_id = 10;</code>
-   */
-  public long getDisplayId() {
-    return displayId_;
-  }
-
-  public static final int NAME_FIELD_NUMBER = 2;
-  private volatile java.lang.Object name_;
-  /**
-   * <pre>
-   * Name of this XLine.
-   * </pre>
-   *
-   * <code>string name = 2;</code>
-   */
-  public java.lang.String getName() {
-    java.lang.Object ref = name_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      name_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * Name of this XLine.
-   * </pre>
-   *
-   * <code>string name = 2;</code>
-   */
-  public com.google.protobuf.ByteString
-      getNameBytes() {
-    java.lang.Object ref = name_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      name_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int DISPLAY_NAME_FIELD_NUMBER = 11;
-  private volatile java.lang.Object displayName_;
-  /**
-   * <pre>
-   * Name of this XLine to display in trace viewer.
-   * </pre>
-   *
-   * <code>string display_name = 11;</code>
-   */
-  public java.lang.String getDisplayName() {
-    java.lang.Object ref = displayName_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      displayName_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * Name of this XLine to display in trace viewer.
-   * </pre>
-   *
-   * <code>string display_name = 11;</code>
-   */
-  public com.google.protobuf.ByteString
-      getDisplayNameBytes() {
-    java.lang.Object ref = displayName_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      displayName_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int TIMESTAMP_NS_FIELD_NUMBER = 3;
-  private long timestampNs_;
-  /**
-   * <pre>
-   * Start time of this line in nanoseconds since the UNIX epoch.
-   * XEvent.offset_ps is relative to this timestamp.
-   * </pre>
-   *
-   * <code>int64 timestamp_ns = 3;</code>
-   */
-  public long getTimestampNs() {
-    return timestampNs_;
-  }
-
-  public static final int DURATION_PS_FIELD_NUMBER = 9;
-  private long durationPs_;
-  /**
-   * <pre>
-   * Profiling duration for this line in picoseconds.
-   * </pre>
-   *
-   * <code>int64 duration_ps = 9;</code>
-   */
-  public long getDurationPs() {
-    return durationPs_;
-  }
-
-  public static final int EVENTS_FIELD_NUMBER = 4;
-  private java.util.List<org.tensorflow.proto.profiler.XEvent> events_;
-  /**
-   * <pre>
-   * XEvents within the same XLine should not overlap in time, but they can be
-   * nested.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.profiler.XEvent events = 4;</code>
-   */
-  public java.util.List<org.tensorflow.proto.profiler.XEvent> getEventsList() {
-    return events_;
-  }
-  /**
-   * <pre>
-   * XEvents within the same XLine should not overlap in time, but they can be
-   * nested.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.profiler.XEvent events = 4;</code>
-   */
-  public java.util.List<? extends org.tensorflow.proto.profiler.XEventOrBuilder> 
-      getEventsOrBuilderList() {
-    return events_;
-  }
-  /**
-   * <pre>
-   * XEvents within the same XLine should not overlap in time, but they can be
-   * nested.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.profiler.XEvent events = 4;</code>
-   */
-  public int getEventsCount() {
-    return events_.size();
-  }
-  /**
-   * <pre>
-   * XEvents within the same XLine should not overlap in time, but they can be
-   * nested.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.profiler.XEvent events = 4;</code>
-   */
-  public org.tensorflow.proto.profiler.XEvent getEvents(int index) {
-    return events_.get(index);
-  }
-  /**
-   * <pre>
-   * XEvents within the same XLine should not overlap in time, but they can be
-   * nested.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.profiler.XEvent events = 4;</code>
-   */
-  public org.tensorflow.proto.profiler.XEventOrBuilder getEventsOrBuilder(
-      int index) {
-    return events_.get(index);
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    if (id_ != 0L) {
-      output.writeInt64(1, id_);
-    }
-    if (!getNameBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 2, name_);
-    }
-    if (timestampNs_ != 0L) {
-      output.writeInt64(3, timestampNs_);
-    }
-    for (int i = 0; i < events_.size(); i++) {
-      output.writeMessage(4, events_.get(i));
-    }
-    if (durationPs_ != 0L) {
-      output.writeInt64(9, durationPs_);
-    }
-    if (displayId_ != 0L) {
-      output.writeInt64(10, displayId_);
-    }
-    if (!getDisplayNameBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 11, displayName_);
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (id_ != 0L) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt64Size(1, id_);
-    }
-    if (!getNameBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(2, name_);
-    }
-    if (timestampNs_ != 0L) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt64Size(3, timestampNs_);
-    }
-    for (int i = 0; i < events_.size(); i++) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(4, events_.get(i));
-    }
-    if (durationPs_ != 0L) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt64Size(9, durationPs_);
-    }
-    if (displayId_ != 0L) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt64Size(10, displayId_);
-    }
-    if (!getDisplayNameBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(11, displayName_);
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.profiler.XLine)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.profiler.XLine other = (org.tensorflow.proto.profiler.XLine) obj;
-
-    if (getId()
-        != other.getId()) return false;
-    if (getDisplayId()
-        != other.getDisplayId()) return false;
-    if (!getName()
-        .equals(other.getName())) return false;
-    if (!getDisplayName()
-        .equals(other.getDisplayName())) return false;
-    if (getTimestampNs()
-        != other.getTimestampNs()) return false;
-    if (getDurationPs()
-        != other.getDurationPs()) return false;
-    if (!getEventsList()
-        .equals(other.getEventsList())) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    hash = (37 * hash) + ID_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-        getId());
-    hash = (37 * hash) + DISPLAY_ID_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-        getDisplayId());
-    hash = (37 * hash) + NAME_FIELD_NUMBER;
-    hash = (53 * hash) + getName().hashCode();
-    hash = (37 * hash) + DISPLAY_NAME_FIELD_NUMBER;
-    hash = (53 * hash) + getDisplayName().hashCode();
-    hash = (37 * hash) + TIMESTAMP_NS_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-        getTimestampNs());
-    hash = (37 * hash) + DURATION_PS_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-        getDurationPs());
-    if (getEventsCount() > 0) {
-      hash = (37 * hash) + EVENTS_FIELD_NUMBER;
-      hash = (53 * hash) + getEventsList().hashCode();
-    }
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.profiler.XLine parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.profiler.XLine parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.profiler.XLine parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.profiler.XLine parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.profiler.XLine parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.profiler.XLine parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.profiler.XLine parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.profiler.XLine parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.profiler.XLine parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.profiler.XLine parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.profiler.XLine parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.profiler.XLine parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.profiler.XLine prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * <pre>
-   * An XLine is a timeline of trace events (XEvents).
-   * Next ID: 12
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.profiler.XLine}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.profiler.XLine)
-      org.tensorflow.proto.profiler.XLineOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.profiler.XPlaneProtos.internal_static_tensorflow_profiler_XLine_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.profiler.XPlaneProtos.internal_static_tensorflow_profiler_XLine_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.profiler.XLine.class, org.tensorflow.proto.profiler.XLine.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.profiler.XLine.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-        getEventsFieldBuilder();
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      id_ = 0L;
-
-      displayId_ = 0L;
-
-      name_ = "";
-
-      displayName_ = "";
-
-      timestampNs_ = 0L;
-
-      durationPs_ = 0L;
-
-      if (eventsBuilder_ == null) {
-        events_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000001);
-      } else {
-        eventsBuilder_.clear();
-      }
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.profiler.XPlaneProtos.internal_static_tensorflow_profiler_XLine_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.profiler.XLine getDefaultInstanceForType() {
-      return org.tensorflow.proto.profiler.XLine.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.profiler.XLine build() {
-      org.tensorflow.proto.profiler.XLine result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.profiler.XLine buildPartial() {
-      org.tensorflow.proto.profiler.XLine result = new org.tensorflow.proto.profiler.XLine(this);
-      int from_bitField0_ = bitField0_;
-      result.id_ = id_;
-      result.displayId_ = displayId_;
-      result.name_ = name_;
-      result.displayName_ = displayName_;
-      result.timestampNs_ = timestampNs_;
-      result.durationPs_ = durationPs_;
-      if (eventsBuilder_ == null) {
-        if (((bitField0_ & 0x00000001) != 0)) {
-          events_ = java.util.Collections.unmodifiableList(events_);
-          bitField0_ = (bitField0_ & ~0x00000001);
-        }
-        result.events_ = events_;
-      } else {
-        result.events_ = eventsBuilder_.build();
-      }
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.profiler.XLine) {
-        return mergeFrom((org.tensorflow.proto.profiler.XLine)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.profiler.XLine other) {
-      if (other == org.tensorflow.proto.profiler.XLine.getDefaultInstance()) return this;
-      if (other.getId() != 0L) {
-        setId(other.getId());
-      }
-      if (other.getDisplayId() != 0L) {
-        setDisplayId(other.getDisplayId());
-      }
-      if (!other.getName().isEmpty()) {
-        name_ = other.name_;
-        onChanged();
-      }
-      if (!other.getDisplayName().isEmpty()) {
-        displayName_ = other.displayName_;
-        onChanged();
-      }
-      if (other.getTimestampNs() != 0L) {
-        setTimestampNs(other.getTimestampNs());
-      }
-      if (other.getDurationPs() != 0L) {
-        setDurationPs(other.getDurationPs());
-      }
-      if (eventsBuilder_ == null) {
-        if (!other.events_.isEmpty()) {
-          if (events_.isEmpty()) {
-            events_ = other.events_;
-            bitField0_ = (bitField0_ & ~0x00000001);
-          } else {
-            ensureEventsIsMutable();
-            events_.addAll(other.events_);
-          }
-          onChanged();
-        }
-      } else {
-        if (!other.events_.isEmpty()) {
-          if (eventsBuilder_.isEmpty()) {
-            eventsBuilder_.dispose();
-            eventsBuilder_ = null;
-            events_ = other.events_;
-            bitField0_ = (bitField0_ & ~0x00000001);
-            eventsBuilder_ = 
-              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
-                 getEventsFieldBuilder() : null;
-          } else {
-            eventsBuilder_.addAllMessages(other.events_);
-          }
-        }
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.profiler.XLine parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.profiler.XLine) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-    private int bitField0_;
-
-    private long id_ ;
-    /**
-     * <pre>
-     * Id of this line, can be repeated within an XPlane. All XLines with the
-     * same id are effectively the same timeline.
-     * </pre>
-     *
-     * <code>int64 id = 1;</code>
-     */
-    public long getId() {
-      return id_;
-    }
-    /**
-     * <pre>
-     * Id of this line, can be repeated within an XPlane. All XLines with the
-     * same id are effectively the same timeline.
-     * </pre>
-     *
-     * <code>int64 id = 1;</code>
-     */
-    public Builder setId(long value) {
-      
-      id_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Id of this line, can be repeated within an XPlane. All XLines with the
-     * same id are effectively the same timeline.
-     * </pre>
-     *
-     * <code>int64 id = 1;</code>
-     */
-    public Builder clearId() {
-      
-      id_ = 0L;
-      onChanged();
-      return this;
-    }
-
-    private long displayId_ ;
-    /**
-     * <pre>
-     * Display id of this line. Multiple lines with the same display_id are
-     * grouped together in the same trace viewer row.
-     * </pre>
-     *
-     * <code>int64 display_id = 10;</code>
-     */
-    public long getDisplayId() {
-      return displayId_;
-    }
-    /**
-     * <pre>
-     * Display id of this line. Multiple lines with the same display_id are
-     * grouped together in the same trace viewer row.
-     * </pre>
-     *
-     * <code>int64 display_id = 10;</code>
-     */
-    public Builder setDisplayId(long value) {
-      
-      displayId_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Display id of this line. Multiple lines with the same display_id are
-     * grouped together in the same trace viewer row.
-     * </pre>
-     *
-     * <code>int64 display_id = 10;</code>
-     */
-    public Builder clearDisplayId() {
-      
-      displayId_ = 0L;
-      onChanged();
-      return this;
-    }
-
-    private java.lang.Object name_ = "";
-    /**
-     * <pre>
-     * Name of this XLine.
-     * </pre>
-     *
-     * <code>string name = 2;</code>
-     */
-    public java.lang.String getName() {
-      java.lang.Object ref = name_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        name_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Name of this XLine.
-     * </pre>
-     *
-     * <code>string name = 2;</code>
-     */
-    public com.google.protobuf.ByteString
-        getNameBytes() {
-      java.lang.Object ref = name_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        name_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Name of this XLine.
-     * </pre>
-     *
-     * <code>string name = 2;</code>
-     */
-    public Builder setName(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      name_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Name of this XLine.
-     * </pre>
-     *
-     * <code>string name = 2;</code>
-     */
-    public Builder clearName() {
-      
-      name_ = getDefaultInstance().getName();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Name of this XLine.
-     * </pre>
-     *
-     * <code>string name = 2;</code>
-     */
-    public Builder setNameBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      name_ = value;
-      onChanged();
-      return this;
-    }
-
-    private java.lang.Object displayName_ = "";
-    /**
-     * <pre>
-     * Name of this XLine to display in trace viewer.
-     * </pre>
-     *
-     * <code>string display_name = 11;</code>
-     */
-    public java.lang.String getDisplayName() {
-      java.lang.Object ref = displayName_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        displayName_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Name of this XLine to display in trace viewer.
-     * </pre>
-     *
-     * <code>string display_name = 11;</code>
-     */
-    public com.google.protobuf.ByteString
-        getDisplayNameBytes() {
-      java.lang.Object ref = displayName_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        displayName_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Name of this XLine to display in trace viewer.
-     * </pre>
-     *
-     * <code>string display_name = 11;</code>
-     */
-    public Builder setDisplayName(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      displayName_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Name of this XLine to display in trace viewer.
-     * </pre>
-     *
-     * <code>string display_name = 11;</code>
-     */
-    public Builder clearDisplayName() {
-      
-      displayName_ = getDefaultInstance().getDisplayName();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Name of this XLine to display in trace viewer.
-     * </pre>
-     *
-     * <code>string display_name = 11;</code>
-     */
-    public Builder setDisplayNameBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      displayName_ = value;
-      onChanged();
-      return this;
-    }
-
-    private long timestampNs_ ;
-    /**
-     * <pre>
-     * Start time of this line in nanoseconds since the UNIX epoch.
-     * XEvent.offset_ps is relative to this timestamp.
-     * </pre>
-     *
-     * <code>int64 timestamp_ns = 3;</code>
-     */
-    public long getTimestampNs() {
-      return timestampNs_;
-    }
-    /**
-     * <pre>
-     * Start time of this line in nanoseconds since the UNIX epoch.
-     * XEvent.offset_ps is relative to this timestamp.
-     * </pre>
-     *
-     * <code>int64 timestamp_ns = 3;</code>
-     */
-    public Builder setTimestampNs(long value) {
-      
-      timestampNs_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Start time of this line in nanoseconds since the UNIX epoch.
-     * XEvent.offset_ps is relative to this timestamp.
-     * </pre>
-     *
-     * <code>int64 timestamp_ns = 3;</code>
-     */
-    public Builder clearTimestampNs() {
-      
-      timestampNs_ = 0L;
-      onChanged();
-      return this;
-    }
-
-    private long durationPs_ ;
-    /**
-     * <pre>
-     * Profiling duration for this line in picoseconds.
-     * </pre>
-     *
-     * <code>int64 duration_ps = 9;</code>
-     */
-    public long getDurationPs() {
-      return durationPs_;
-    }
-    /**
-     * <pre>
-     * Profiling duration for this line in picoseconds.
-     * </pre>
-     *
-     * <code>int64 duration_ps = 9;</code>
-     */
-    public Builder setDurationPs(long value) {
-      
-      durationPs_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Profiling duration for this line in picoseconds.
-     * </pre>
-     *
-     * <code>int64 duration_ps = 9;</code>
-     */
-    public Builder clearDurationPs() {
-      
-      durationPs_ = 0L;
-      onChanged();
-      return this;
-    }
-
-    private java.util.List<org.tensorflow.proto.profiler.XEvent> events_ =
-      java.util.Collections.emptyList();
-    private void ensureEventsIsMutable() {
-      if (!((bitField0_ & 0x00000001) != 0)) {
-        events_ = new java.util.ArrayList<org.tensorflow.proto.profiler.XEvent>(events_);
-        bitField0_ |= 0x00000001;
-       }
-    }
-
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.profiler.XEvent, org.tensorflow.proto.profiler.XEvent.Builder, org.tensorflow.proto.profiler.XEventOrBuilder> eventsBuilder_;
-
-    /**
-     * <pre>
-     * XEvents within the same XLine should not overlap in time, but they can be
-     * nested.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.profiler.XEvent events = 4;</code>
-     */
-    public java.util.List<org.tensorflow.proto.profiler.XEvent> getEventsList() {
-      if (eventsBuilder_ == null) {
-        return java.util.Collections.unmodifiableList(events_);
-      } else {
-        return eventsBuilder_.getMessageList();
-      }
-    }
-    /**
-     * <pre>
-     * XEvents within the same XLine should not overlap in time, but they can be
-     * nested.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.profiler.XEvent events = 4;</code>
-     */
-    public int getEventsCount() {
-      if (eventsBuilder_ == null) {
-        return events_.size();
-      } else {
-        return eventsBuilder_.getCount();
-      }
-    }
-    /**
-     * <pre>
-     * XEvents within the same XLine should not overlap in time, but they can be
-     * nested.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.profiler.XEvent events = 4;</code>
-     */
-    public org.tensorflow.proto.profiler.XEvent getEvents(int index) {
-      if (eventsBuilder_ == null) {
-        return events_.get(index);
-      } else {
-        return eventsBuilder_.getMessage(index);
-      }
-    }
-    /**
-     * <pre>
-     * XEvents within the same XLine should not overlap in time, but they can be
-     * nested.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.profiler.XEvent events = 4;</code>
-     */
-    public Builder setEvents(
-        int index, org.tensorflow.proto.profiler.XEvent value) {
-      if (eventsBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureEventsIsMutable();
-        events_.set(index, value);
-        onChanged();
-      } else {
-        eventsBuilder_.setMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * XEvents within the same XLine should not overlap in time, but they can be
-     * nested.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.profiler.XEvent events = 4;</code>
-     */
-    public Builder setEvents(
-        int index, org.tensorflow.proto.profiler.XEvent.Builder builderForValue) {
-      if (eventsBuilder_ == null) {
-        ensureEventsIsMutable();
-        events_.set(index, builderForValue.build());
-        onChanged();
-      } else {
-        eventsBuilder_.setMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * XEvents within the same XLine should not overlap in time, but they can be
-     * nested.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.profiler.XEvent events = 4;</code>
-     */
-    public Builder addEvents(org.tensorflow.proto.profiler.XEvent value) {
-      if (eventsBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureEventsIsMutable();
-        events_.add(value);
-        onChanged();
-      } else {
-        eventsBuilder_.addMessage(value);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * XEvents within the same XLine should not overlap in time, but they can be
-     * nested.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.profiler.XEvent events = 4;</code>
-     */
-    public Builder addEvents(
-        int index, org.tensorflow.proto.profiler.XEvent value) {
-      if (eventsBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureEventsIsMutable();
-        events_.add(index, value);
-        onChanged();
-      } else {
-        eventsBuilder_.addMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * XEvents within the same XLine should not overlap in time, but they can be
-     * nested.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.profiler.XEvent events = 4;</code>
-     */
-    public Builder addEvents(
-        org.tensorflow.proto.profiler.XEvent.Builder builderForValue) {
-      if (eventsBuilder_ == null) {
-        ensureEventsIsMutable();
-        events_.add(builderForValue.build());
-        onChanged();
-      } else {
-        eventsBuilder_.addMessage(builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * XEvents within the same XLine should not overlap in time, but they can be
-     * nested.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.profiler.XEvent events = 4;</code>
-     */
-    public Builder addEvents(
-        int index, org.tensorflow.proto.profiler.XEvent.Builder builderForValue) {
-      if (eventsBuilder_ == null) {
-        ensureEventsIsMutable();
-        events_.add(index, builderForValue.build());
-        onChanged();
-      } else {
-        eventsBuilder_.addMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * XEvents within the same XLine should not overlap in time, but they can be
-     * nested.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.profiler.XEvent events = 4;</code>
-     */
-    public Builder addAllEvents(
-        java.lang.Iterable<? extends org.tensorflow.proto.profiler.XEvent> values) {
-      if (eventsBuilder_ == null) {
-        ensureEventsIsMutable();
-        com.google.protobuf.AbstractMessageLite.Builder.addAll(
-            values, events_);
-        onChanged();
-      } else {
-        eventsBuilder_.addAllMessages(values);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * XEvents within the same XLine should not overlap in time, but they can be
-     * nested.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.profiler.XEvent events = 4;</code>
-     */
-    public Builder clearEvents() {
-      if (eventsBuilder_ == null) {
-        events_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000001);
-        onChanged();
-      } else {
-        eventsBuilder_.clear();
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * XEvents within the same XLine should not overlap in time, but they can be
-     * nested.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.profiler.XEvent events = 4;</code>
-     */
-    public Builder removeEvents(int index) {
-      if (eventsBuilder_ == null) {
-        ensureEventsIsMutable();
-        events_.remove(index);
-        onChanged();
-      } else {
-        eventsBuilder_.remove(index);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * XEvents within the same XLine should not overlap in time, but they can be
-     * nested.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.profiler.XEvent events = 4;</code>
-     */
-    public org.tensorflow.proto.profiler.XEvent.Builder getEventsBuilder(
-        int index) {
-      return getEventsFieldBuilder().getBuilder(index);
-    }
-    /**
-     * <pre>
-     * XEvents within the same XLine should not overlap in time, but they can be
-     * nested.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.profiler.XEvent events = 4;</code>
-     */
-    public org.tensorflow.proto.profiler.XEventOrBuilder getEventsOrBuilder(
-        int index) {
-      if (eventsBuilder_ == null) {
-        return events_.get(index);  } else {
-        return eventsBuilder_.getMessageOrBuilder(index);
-      }
-    }
-    /**
-     * <pre>
-     * XEvents within the same XLine should not overlap in time, but they can be
-     * nested.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.profiler.XEvent events = 4;</code>
-     */
-    public java.util.List<? extends org.tensorflow.proto.profiler.XEventOrBuilder> 
-         getEventsOrBuilderList() {
-      if (eventsBuilder_ != null) {
-        return eventsBuilder_.getMessageOrBuilderList();
-      } else {
-        return java.util.Collections.unmodifiableList(events_);
-      }
-    }
-    /**
-     * <pre>
-     * XEvents within the same XLine should not overlap in time, but they can be
-     * nested.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.profiler.XEvent events = 4;</code>
-     */
-    public org.tensorflow.proto.profiler.XEvent.Builder addEventsBuilder() {
-      return getEventsFieldBuilder().addBuilder(
-          org.tensorflow.proto.profiler.XEvent.getDefaultInstance());
-    }
-    /**
-     * <pre>
-     * XEvents within the same XLine should not overlap in time, but they can be
-     * nested.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.profiler.XEvent events = 4;</code>
-     */
-    public org.tensorflow.proto.profiler.XEvent.Builder addEventsBuilder(
-        int index) {
-      return getEventsFieldBuilder().addBuilder(
-          index, org.tensorflow.proto.profiler.XEvent.getDefaultInstance());
-    }
-    /**
-     * <pre>
-     * XEvents within the same XLine should not overlap in time, but they can be
-     * nested.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.profiler.XEvent events = 4;</code>
-     */
-    public java.util.List<org.tensorflow.proto.profiler.XEvent.Builder> 
-         getEventsBuilderList() {
-      return getEventsFieldBuilder().getBuilderList();
-    }
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.profiler.XEvent, org.tensorflow.proto.profiler.XEvent.Builder, org.tensorflow.proto.profiler.XEventOrBuilder> 
-        getEventsFieldBuilder() {
-      if (eventsBuilder_ == null) {
-        eventsBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
-            org.tensorflow.proto.profiler.XEvent, org.tensorflow.proto.profiler.XEvent.Builder, org.tensorflow.proto.profiler.XEventOrBuilder>(
-                events_,
-                ((bitField0_ & 0x00000001) != 0),
-                getParentForChildren(),
-                isClean());
-        events_ = null;
-      }
-      return eventsBuilder_;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.profiler.XLine)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.profiler.XLine)
-  private static final org.tensorflow.proto.profiler.XLine DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.profiler.XLine();
-  }
-
-  public static org.tensorflow.proto.profiler.XLine getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<XLine>
-      PARSER = new com.google.protobuf.AbstractParser<XLine>() {
-    @java.lang.Override
-    public XLine parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new XLine(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<XLine> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<XLine> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.profiler.XLine getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/XLineOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/XLineOrBuilder.java
deleted file mode 100644
index 1cf5477dca7..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/XLineOrBuilder.java
+++ /dev/null
@@ -1,133 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/profiler/protobuf/xplane.proto
-
-package org.tensorflow.proto.profiler;
-
-public interface XLineOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.profiler.XLine)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <pre>
-   * Id of this line, can be repeated within an XPlane. All XLines with the
-   * same id are effectively the same timeline.
-   * </pre>
-   *
-   * <code>int64 id = 1;</code>
-   */
-  long getId();
-
-  /**
-   * <pre>
-   * Display id of this line. Multiple lines with the same display_id are
-   * grouped together in the same trace viewer row.
-   * </pre>
-   *
-   * <code>int64 display_id = 10;</code>
-   */
-  long getDisplayId();
-
-  /**
-   * <pre>
-   * Name of this XLine.
-   * </pre>
-   *
-   * <code>string name = 2;</code>
-   */
-  java.lang.String getName();
-  /**
-   * <pre>
-   * Name of this XLine.
-   * </pre>
-   *
-   * <code>string name = 2;</code>
-   */
-  com.google.protobuf.ByteString
-      getNameBytes();
-
-  /**
-   * <pre>
-   * Name of this XLine to display in trace viewer.
-   * </pre>
-   *
-   * <code>string display_name = 11;</code>
-   */
-  java.lang.String getDisplayName();
-  /**
-   * <pre>
-   * Name of this XLine to display in trace viewer.
-   * </pre>
-   *
-   * <code>string display_name = 11;</code>
-   */
-  com.google.protobuf.ByteString
-      getDisplayNameBytes();
-
-  /**
-   * <pre>
-   * Start time of this line in nanoseconds since the UNIX epoch.
-   * XEvent.offset_ps is relative to this timestamp.
-   * </pre>
-   *
-   * <code>int64 timestamp_ns = 3;</code>
-   */
-  long getTimestampNs();
-
-  /**
-   * <pre>
-   * Profiling duration for this line in picoseconds.
-   * </pre>
-   *
-   * <code>int64 duration_ps = 9;</code>
-   */
-  long getDurationPs();
-
-  /**
-   * <pre>
-   * XEvents within the same XLine should not overlap in time, but they can be
-   * nested.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.profiler.XEvent events = 4;</code>
-   */
-  java.util.List<org.tensorflow.proto.profiler.XEvent> 
-      getEventsList();
-  /**
-   * <pre>
-   * XEvents within the same XLine should not overlap in time, but they can be
-   * nested.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.profiler.XEvent events = 4;</code>
-   */
-  org.tensorflow.proto.profiler.XEvent getEvents(int index);
-  /**
-   * <pre>
-   * XEvents within the same XLine should not overlap in time, but they can be
-   * nested.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.profiler.XEvent events = 4;</code>
-   */
-  int getEventsCount();
-  /**
-   * <pre>
-   * XEvents within the same XLine should not overlap in time, but they can be
-   * nested.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.profiler.XEvent events = 4;</code>
-   */
-  java.util.List<? extends org.tensorflow.proto.profiler.XEventOrBuilder> 
-      getEventsOrBuilderList();
-  /**
-   * <pre>
-   * XEvents within the same XLine should not overlap in time, but they can be
-   * nested.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.profiler.XEvent events = 4;</code>
-   */
-  org.tensorflow.proto.profiler.XEventOrBuilder getEventsOrBuilder(
-      int index);
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/XPlane.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/XPlane.java
deleted file mode 100644
index 7668a7f1d74..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/XPlane.java
+++ /dev/null
@@ -1,2168 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/profiler/protobuf/xplane.proto
-
-package org.tensorflow.proto.profiler;
-
-/**
- * <pre>
- * An XPlane is a container of parallel timelines (XLines), generated by a
- * profiling source or by post-processing one or more XPlanes.
- * Next ID: 7
- * </pre>
- *
- * Protobuf type {@code tensorflow.profiler.XPlane}
- */
-public  final class XPlane extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.profiler.XPlane)
-    XPlaneOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use XPlane.newBuilder() to construct.
-  private XPlane(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private XPlane() {
-    name_ = "";
-    lines_ = java.util.Collections.emptyList();
-    stats_ = java.util.Collections.emptyList();
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new XPlane();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private XPlane(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    int mutable_bitField0_ = 0;
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 8: {
-
-            id_ = input.readInt64();
-            break;
-          }
-          case 18: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            name_ = s;
-            break;
-          }
-          case 26: {
-            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
-              lines_ = new java.util.ArrayList<org.tensorflow.proto.profiler.XLine>();
-              mutable_bitField0_ |= 0x00000001;
-            }
-            lines_.add(
-                input.readMessage(org.tensorflow.proto.profiler.XLine.parser(), extensionRegistry));
-            break;
-          }
-          case 34: {
-            if (!((mutable_bitField0_ & 0x00000002) != 0)) {
-              eventMetadata_ = com.google.protobuf.MapField.newMapField(
-                  EventMetadataDefaultEntryHolder.defaultEntry);
-              mutable_bitField0_ |= 0x00000002;
-            }
-            com.google.protobuf.MapEntry<java.lang.Long, org.tensorflow.proto.profiler.XEventMetadata>
-            eventMetadata__ = input.readMessage(
-                EventMetadataDefaultEntryHolder.defaultEntry.getParserForType(), extensionRegistry);
-            eventMetadata_.getMutableMap().put(
-                eventMetadata__.getKey(), eventMetadata__.getValue());
-            break;
-          }
-          case 42: {
-            if (!((mutable_bitField0_ & 0x00000004) != 0)) {
-              statMetadata_ = com.google.protobuf.MapField.newMapField(
-                  StatMetadataDefaultEntryHolder.defaultEntry);
-              mutable_bitField0_ |= 0x00000004;
-            }
-            com.google.protobuf.MapEntry<java.lang.Long, org.tensorflow.proto.profiler.XStatMetadata>
-            statMetadata__ = input.readMessage(
-                StatMetadataDefaultEntryHolder.defaultEntry.getParserForType(), extensionRegistry);
-            statMetadata_.getMutableMap().put(
-                statMetadata__.getKey(), statMetadata__.getValue());
-            break;
-          }
-          case 50: {
-            if (!((mutable_bitField0_ & 0x00000008) != 0)) {
-              stats_ = new java.util.ArrayList<org.tensorflow.proto.profiler.XStat>();
-              mutable_bitField0_ |= 0x00000008;
-            }
-            stats_.add(
-                input.readMessage(org.tensorflow.proto.profiler.XStat.parser(), extensionRegistry));
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      if (((mutable_bitField0_ & 0x00000001) != 0)) {
-        lines_ = java.util.Collections.unmodifiableList(lines_);
-      }
-      if (((mutable_bitField0_ & 0x00000008) != 0)) {
-        stats_ = java.util.Collections.unmodifiableList(stats_);
-      }
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.profiler.XPlaneProtos.internal_static_tensorflow_profiler_XPlane_descriptor;
-  }
-
-  @SuppressWarnings({"rawtypes"})
-  @java.lang.Override
-  protected com.google.protobuf.MapField internalGetMapField(
-      int number) {
-    switch (number) {
-      case 4:
-        return internalGetEventMetadata();
-      case 5:
-        return internalGetStatMetadata();
-      default:
-        throw new RuntimeException(
-            "Invalid map field number: " + number);
-    }
-  }
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.profiler.XPlaneProtos.internal_static_tensorflow_profiler_XPlane_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.profiler.XPlane.class, org.tensorflow.proto.profiler.XPlane.Builder.class);
-  }
-
-  public static final int ID_FIELD_NUMBER = 1;
-  private long id_;
-  /**
-   * <code>int64 id = 1;</code>
-   */
-  public long getId() {
-    return id_;
-  }
-
-  public static final int NAME_FIELD_NUMBER = 2;
-  private volatile java.lang.Object name_;
-  /**
-   * <pre>
-   * Name of this line.
-   * </pre>
-   *
-   * <code>string name = 2;</code>
-   */
-  public java.lang.String getName() {
-    java.lang.Object ref = name_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      name_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * Name of this line.
-   * </pre>
-   *
-   * <code>string name = 2;</code>
-   */
-  public com.google.protobuf.ByteString
-      getNameBytes() {
-    java.lang.Object ref = name_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      name_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int LINES_FIELD_NUMBER = 3;
-  private java.util.List<org.tensorflow.proto.profiler.XLine> lines_;
-  /**
-   * <pre>
-   * Parallel timelines grouped in this plane. XLines with the same id
-   * are effectively the same timeline.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.profiler.XLine lines = 3;</code>
-   */
-  public java.util.List<org.tensorflow.proto.profiler.XLine> getLinesList() {
-    return lines_;
-  }
-  /**
-   * <pre>
-   * Parallel timelines grouped in this plane. XLines with the same id
-   * are effectively the same timeline.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.profiler.XLine lines = 3;</code>
-   */
-  public java.util.List<? extends org.tensorflow.proto.profiler.XLineOrBuilder> 
-      getLinesOrBuilderList() {
-    return lines_;
-  }
-  /**
-   * <pre>
-   * Parallel timelines grouped in this plane. XLines with the same id
-   * are effectively the same timeline.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.profiler.XLine lines = 3;</code>
-   */
-  public int getLinesCount() {
-    return lines_.size();
-  }
-  /**
-   * <pre>
-   * Parallel timelines grouped in this plane. XLines with the same id
-   * are effectively the same timeline.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.profiler.XLine lines = 3;</code>
-   */
-  public org.tensorflow.proto.profiler.XLine getLines(int index) {
-    return lines_.get(index);
-  }
-  /**
-   * <pre>
-   * Parallel timelines grouped in this plane. XLines with the same id
-   * are effectively the same timeline.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.profiler.XLine lines = 3;</code>
-   */
-  public org.tensorflow.proto.profiler.XLineOrBuilder getLinesOrBuilder(
-      int index) {
-    return lines_.get(index);
-  }
-
-  public static final int EVENT_METADATA_FIELD_NUMBER = 4;
-  private static final class EventMetadataDefaultEntryHolder {
-    static final com.google.protobuf.MapEntry<
-        java.lang.Long, org.tensorflow.proto.profiler.XEventMetadata> defaultEntry =
-            com.google.protobuf.MapEntry
-            .<java.lang.Long, org.tensorflow.proto.profiler.XEventMetadata>newDefaultInstance(
-                org.tensorflow.proto.profiler.XPlaneProtos.internal_static_tensorflow_profiler_XPlane_EventMetadataEntry_descriptor, 
-                com.google.protobuf.WireFormat.FieldType.INT64,
-                0L,
-                com.google.protobuf.WireFormat.FieldType.MESSAGE,
-                org.tensorflow.proto.profiler.XEventMetadata.getDefaultInstance());
-  }
-  private com.google.protobuf.MapField<
-      java.lang.Long, org.tensorflow.proto.profiler.XEventMetadata> eventMetadata_;
-  private com.google.protobuf.MapField<java.lang.Long, org.tensorflow.proto.profiler.XEventMetadata>
-  internalGetEventMetadata() {
-    if (eventMetadata_ == null) {
-      return com.google.protobuf.MapField.emptyMapField(
-          EventMetadataDefaultEntryHolder.defaultEntry);
-    }
-    return eventMetadata_;
-  }
-
-  public int getEventMetadataCount() {
-    return internalGetEventMetadata().getMap().size();
-  }
-  /**
-   * <pre>
-   * XEventMetadata map, each entry uses the XEventMetadata.id as key. This map
-   * should be used for events that share the same ID over the whole XPlane.
-   * </pre>
-   *
-   * <code>map&lt;int64, .tensorflow.profiler.XEventMetadata&gt; event_metadata = 4;</code>
-   */
-
-  public boolean containsEventMetadata(
-      long key) {
-    
-    return internalGetEventMetadata().getMap().containsKey(key);
-  }
-  /**
-   * Use {@link #getEventMetadataMap()} instead.
-   */
-  @java.lang.Deprecated
-  public java.util.Map<java.lang.Long, org.tensorflow.proto.profiler.XEventMetadata> getEventMetadata() {
-    return getEventMetadataMap();
-  }
-  /**
-   * <pre>
-   * XEventMetadata map, each entry uses the XEventMetadata.id as key. This map
-   * should be used for events that share the same ID over the whole XPlane.
-   * </pre>
-   *
-   * <code>map&lt;int64, .tensorflow.profiler.XEventMetadata&gt; event_metadata = 4;</code>
-   */
-
-  public java.util.Map<java.lang.Long, org.tensorflow.proto.profiler.XEventMetadata> getEventMetadataMap() {
-    return internalGetEventMetadata().getMap();
-  }
-  /**
-   * <pre>
-   * XEventMetadata map, each entry uses the XEventMetadata.id as key. This map
-   * should be used for events that share the same ID over the whole XPlane.
-   * </pre>
-   *
-   * <code>map&lt;int64, .tensorflow.profiler.XEventMetadata&gt; event_metadata = 4;</code>
-   */
-
-  public org.tensorflow.proto.profiler.XEventMetadata getEventMetadataOrDefault(
-      long key,
-      org.tensorflow.proto.profiler.XEventMetadata defaultValue) {
-    
-    java.util.Map<java.lang.Long, org.tensorflow.proto.profiler.XEventMetadata> map =
-        internalGetEventMetadata().getMap();
-    return map.containsKey(key) ? map.get(key) : defaultValue;
-  }
-  /**
-   * <pre>
-   * XEventMetadata map, each entry uses the XEventMetadata.id as key. This map
-   * should be used for events that share the same ID over the whole XPlane.
-   * </pre>
-   *
-   * <code>map&lt;int64, .tensorflow.profiler.XEventMetadata&gt; event_metadata = 4;</code>
-   */
-
-  public org.tensorflow.proto.profiler.XEventMetadata getEventMetadataOrThrow(
-      long key) {
-    
-    java.util.Map<java.lang.Long, org.tensorflow.proto.profiler.XEventMetadata> map =
-        internalGetEventMetadata().getMap();
-    if (!map.containsKey(key)) {
-      throw new java.lang.IllegalArgumentException();
-    }
-    return map.get(key);
-  }
-
-  public static final int STAT_METADATA_FIELD_NUMBER = 5;
-  private static final class StatMetadataDefaultEntryHolder {
-    static final com.google.protobuf.MapEntry<
-        java.lang.Long, org.tensorflow.proto.profiler.XStatMetadata> defaultEntry =
-            com.google.protobuf.MapEntry
-            .<java.lang.Long, org.tensorflow.proto.profiler.XStatMetadata>newDefaultInstance(
-                org.tensorflow.proto.profiler.XPlaneProtos.internal_static_tensorflow_profiler_XPlane_StatMetadataEntry_descriptor, 
-                com.google.protobuf.WireFormat.FieldType.INT64,
-                0L,
-                com.google.protobuf.WireFormat.FieldType.MESSAGE,
-                org.tensorflow.proto.profiler.XStatMetadata.getDefaultInstance());
-  }
-  private com.google.protobuf.MapField<
-      java.lang.Long, org.tensorflow.proto.profiler.XStatMetadata> statMetadata_;
-  private com.google.protobuf.MapField<java.lang.Long, org.tensorflow.proto.profiler.XStatMetadata>
-  internalGetStatMetadata() {
-    if (statMetadata_ == null) {
-      return com.google.protobuf.MapField.emptyMapField(
-          StatMetadataDefaultEntryHolder.defaultEntry);
-    }
-    return statMetadata_;
-  }
-
-  public int getStatMetadataCount() {
-    return internalGetStatMetadata().getMap().size();
-  }
-  /**
-   * <pre>
-   * XStatMetadata map, each entry uses the XStatMetadata.id as key. This map
-   * should be used for stats that share the same ID over the whole XPlane.
-   * </pre>
-   *
-   * <code>map&lt;int64, .tensorflow.profiler.XStatMetadata&gt; stat_metadata = 5;</code>
-   */
-
-  public boolean containsStatMetadata(
-      long key) {
-    
-    return internalGetStatMetadata().getMap().containsKey(key);
-  }
-  /**
-   * Use {@link #getStatMetadataMap()} instead.
-   */
-  @java.lang.Deprecated
-  public java.util.Map<java.lang.Long, org.tensorflow.proto.profiler.XStatMetadata> getStatMetadata() {
-    return getStatMetadataMap();
-  }
-  /**
-   * <pre>
-   * XStatMetadata map, each entry uses the XStatMetadata.id as key. This map
-   * should be used for stats that share the same ID over the whole XPlane.
-   * </pre>
-   *
-   * <code>map&lt;int64, .tensorflow.profiler.XStatMetadata&gt; stat_metadata = 5;</code>
-   */
-
-  public java.util.Map<java.lang.Long, org.tensorflow.proto.profiler.XStatMetadata> getStatMetadataMap() {
-    return internalGetStatMetadata().getMap();
-  }
-  /**
-   * <pre>
-   * XStatMetadata map, each entry uses the XStatMetadata.id as key. This map
-   * should be used for stats that share the same ID over the whole XPlane.
-   * </pre>
-   *
-   * <code>map&lt;int64, .tensorflow.profiler.XStatMetadata&gt; stat_metadata = 5;</code>
-   */
-
-  public org.tensorflow.proto.profiler.XStatMetadata getStatMetadataOrDefault(
-      long key,
-      org.tensorflow.proto.profiler.XStatMetadata defaultValue) {
-    
-    java.util.Map<java.lang.Long, org.tensorflow.proto.profiler.XStatMetadata> map =
-        internalGetStatMetadata().getMap();
-    return map.containsKey(key) ? map.get(key) : defaultValue;
-  }
-  /**
-   * <pre>
-   * XStatMetadata map, each entry uses the XStatMetadata.id as key. This map
-   * should be used for stats that share the same ID over the whole XPlane.
-   * </pre>
-   *
-   * <code>map&lt;int64, .tensorflow.profiler.XStatMetadata&gt; stat_metadata = 5;</code>
-   */
-
-  public org.tensorflow.proto.profiler.XStatMetadata getStatMetadataOrThrow(
-      long key) {
-    
-    java.util.Map<java.lang.Long, org.tensorflow.proto.profiler.XStatMetadata> map =
-        internalGetStatMetadata().getMap();
-    if (!map.containsKey(key)) {
-      throw new java.lang.IllegalArgumentException();
-    }
-    return map.get(key);
-  }
-
-  public static final int STATS_FIELD_NUMBER = 6;
-  private java.util.List<org.tensorflow.proto.profiler.XStat> stats_;
-  /**
-   * <pre>
-   * XStats associated with this plane, e.g. device capabilities.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.profiler.XStat stats = 6;</code>
-   */
-  public java.util.List<org.tensorflow.proto.profiler.XStat> getStatsList() {
-    return stats_;
-  }
-  /**
-   * <pre>
-   * XStats associated with this plane, e.g. device capabilities.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.profiler.XStat stats = 6;</code>
-   */
-  public java.util.List<? extends org.tensorflow.proto.profiler.XStatOrBuilder> 
-      getStatsOrBuilderList() {
-    return stats_;
-  }
-  /**
-   * <pre>
-   * XStats associated with this plane, e.g. device capabilities.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.profiler.XStat stats = 6;</code>
-   */
-  public int getStatsCount() {
-    return stats_.size();
-  }
-  /**
-   * <pre>
-   * XStats associated with this plane, e.g. device capabilities.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.profiler.XStat stats = 6;</code>
-   */
-  public org.tensorflow.proto.profiler.XStat getStats(int index) {
-    return stats_.get(index);
-  }
-  /**
-   * <pre>
-   * XStats associated with this plane, e.g. device capabilities.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.profiler.XStat stats = 6;</code>
-   */
-  public org.tensorflow.proto.profiler.XStatOrBuilder getStatsOrBuilder(
-      int index) {
-    return stats_.get(index);
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    if (id_ != 0L) {
-      output.writeInt64(1, id_);
-    }
-    if (!getNameBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 2, name_);
-    }
-    for (int i = 0; i < lines_.size(); i++) {
-      output.writeMessage(3, lines_.get(i));
-    }
-    com.google.protobuf.GeneratedMessageV3
-      .serializeLongMapTo(
-        output,
-        internalGetEventMetadata(),
-        EventMetadataDefaultEntryHolder.defaultEntry,
-        4);
-    com.google.protobuf.GeneratedMessageV3
-      .serializeLongMapTo(
-        output,
-        internalGetStatMetadata(),
-        StatMetadataDefaultEntryHolder.defaultEntry,
-        5);
-    for (int i = 0; i < stats_.size(); i++) {
-      output.writeMessage(6, stats_.get(i));
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (id_ != 0L) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt64Size(1, id_);
-    }
-    if (!getNameBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(2, name_);
-    }
-    for (int i = 0; i < lines_.size(); i++) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(3, lines_.get(i));
-    }
-    for (java.util.Map.Entry<java.lang.Long, org.tensorflow.proto.profiler.XEventMetadata> entry
-         : internalGetEventMetadata().getMap().entrySet()) {
-      com.google.protobuf.MapEntry<java.lang.Long, org.tensorflow.proto.profiler.XEventMetadata>
-      eventMetadata__ = EventMetadataDefaultEntryHolder.defaultEntry.newBuilderForType()
-          .setKey(entry.getKey())
-          .setValue(entry.getValue())
-          .build();
-      size += com.google.protobuf.CodedOutputStream
-          .computeMessageSize(4, eventMetadata__);
-    }
-    for (java.util.Map.Entry<java.lang.Long, org.tensorflow.proto.profiler.XStatMetadata> entry
-         : internalGetStatMetadata().getMap().entrySet()) {
-      com.google.protobuf.MapEntry<java.lang.Long, org.tensorflow.proto.profiler.XStatMetadata>
-      statMetadata__ = StatMetadataDefaultEntryHolder.defaultEntry.newBuilderForType()
-          .setKey(entry.getKey())
-          .setValue(entry.getValue())
-          .build();
-      size += com.google.protobuf.CodedOutputStream
-          .computeMessageSize(5, statMetadata__);
-    }
-    for (int i = 0; i < stats_.size(); i++) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(6, stats_.get(i));
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.profiler.XPlane)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.profiler.XPlane other = (org.tensorflow.proto.profiler.XPlane) obj;
-
-    if (getId()
-        != other.getId()) return false;
-    if (!getName()
-        .equals(other.getName())) return false;
-    if (!getLinesList()
-        .equals(other.getLinesList())) return false;
-    if (!internalGetEventMetadata().equals(
-        other.internalGetEventMetadata())) return false;
-    if (!internalGetStatMetadata().equals(
-        other.internalGetStatMetadata())) return false;
-    if (!getStatsList()
-        .equals(other.getStatsList())) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    hash = (37 * hash) + ID_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-        getId());
-    hash = (37 * hash) + NAME_FIELD_NUMBER;
-    hash = (53 * hash) + getName().hashCode();
-    if (getLinesCount() > 0) {
-      hash = (37 * hash) + LINES_FIELD_NUMBER;
-      hash = (53 * hash) + getLinesList().hashCode();
-    }
-    if (!internalGetEventMetadata().getMap().isEmpty()) {
-      hash = (37 * hash) + EVENT_METADATA_FIELD_NUMBER;
-      hash = (53 * hash) + internalGetEventMetadata().hashCode();
-    }
-    if (!internalGetStatMetadata().getMap().isEmpty()) {
-      hash = (37 * hash) + STAT_METADATA_FIELD_NUMBER;
-      hash = (53 * hash) + internalGetStatMetadata().hashCode();
-    }
-    if (getStatsCount() > 0) {
-      hash = (37 * hash) + STATS_FIELD_NUMBER;
-      hash = (53 * hash) + getStatsList().hashCode();
-    }
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.profiler.XPlane parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.profiler.XPlane parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.profiler.XPlane parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.profiler.XPlane parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.profiler.XPlane parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.profiler.XPlane parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.profiler.XPlane parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.profiler.XPlane parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.profiler.XPlane parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.profiler.XPlane parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.profiler.XPlane parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.profiler.XPlane parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.profiler.XPlane prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * <pre>
-   * An XPlane is a container of parallel timelines (XLines), generated by a
-   * profiling source or by post-processing one or more XPlanes.
-   * Next ID: 7
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.profiler.XPlane}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.profiler.XPlane)
-      org.tensorflow.proto.profiler.XPlaneOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.profiler.XPlaneProtos.internal_static_tensorflow_profiler_XPlane_descriptor;
-    }
-
-    @SuppressWarnings({"rawtypes"})
-    protected com.google.protobuf.MapField internalGetMapField(
-        int number) {
-      switch (number) {
-        case 4:
-          return internalGetEventMetadata();
-        case 5:
-          return internalGetStatMetadata();
-        default:
-          throw new RuntimeException(
-              "Invalid map field number: " + number);
-      }
-    }
-    @SuppressWarnings({"rawtypes"})
-    protected com.google.protobuf.MapField internalGetMutableMapField(
-        int number) {
-      switch (number) {
-        case 4:
-          return internalGetMutableEventMetadata();
-        case 5:
-          return internalGetMutableStatMetadata();
-        default:
-          throw new RuntimeException(
-              "Invalid map field number: " + number);
-      }
-    }
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.profiler.XPlaneProtos.internal_static_tensorflow_profiler_XPlane_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.profiler.XPlane.class, org.tensorflow.proto.profiler.XPlane.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.profiler.XPlane.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-        getLinesFieldBuilder();
-        getStatsFieldBuilder();
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      id_ = 0L;
-
-      name_ = "";
-
-      if (linesBuilder_ == null) {
-        lines_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000001);
-      } else {
-        linesBuilder_.clear();
-      }
-      internalGetMutableEventMetadata().clear();
-      internalGetMutableStatMetadata().clear();
-      if (statsBuilder_ == null) {
-        stats_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000008);
-      } else {
-        statsBuilder_.clear();
-      }
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.profiler.XPlaneProtos.internal_static_tensorflow_profiler_XPlane_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.profiler.XPlane getDefaultInstanceForType() {
-      return org.tensorflow.proto.profiler.XPlane.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.profiler.XPlane build() {
-      org.tensorflow.proto.profiler.XPlane result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.profiler.XPlane buildPartial() {
-      org.tensorflow.proto.profiler.XPlane result = new org.tensorflow.proto.profiler.XPlane(this);
-      int from_bitField0_ = bitField0_;
-      result.id_ = id_;
-      result.name_ = name_;
-      if (linesBuilder_ == null) {
-        if (((bitField0_ & 0x00000001) != 0)) {
-          lines_ = java.util.Collections.unmodifiableList(lines_);
-          bitField0_ = (bitField0_ & ~0x00000001);
-        }
-        result.lines_ = lines_;
-      } else {
-        result.lines_ = linesBuilder_.build();
-      }
-      result.eventMetadata_ = internalGetEventMetadata();
-      result.eventMetadata_.makeImmutable();
-      result.statMetadata_ = internalGetStatMetadata();
-      result.statMetadata_.makeImmutable();
-      if (statsBuilder_ == null) {
-        if (((bitField0_ & 0x00000008) != 0)) {
-          stats_ = java.util.Collections.unmodifiableList(stats_);
-          bitField0_ = (bitField0_ & ~0x00000008);
-        }
-        result.stats_ = stats_;
-      } else {
-        result.stats_ = statsBuilder_.build();
-      }
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.profiler.XPlane) {
-        return mergeFrom((org.tensorflow.proto.profiler.XPlane)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.profiler.XPlane other) {
-      if (other == org.tensorflow.proto.profiler.XPlane.getDefaultInstance()) return this;
-      if (other.getId() != 0L) {
-        setId(other.getId());
-      }
-      if (!other.getName().isEmpty()) {
-        name_ = other.name_;
-        onChanged();
-      }
-      if (linesBuilder_ == null) {
-        if (!other.lines_.isEmpty()) {
-          if (lines_.isEmpty()) {
-            lines_ = other.lines_;
-            bitField0_ = (bitField0_ & ~0x00000001);
-          } else {
-            ensureLinesIsMutable();
-            lines_.addAll(other.lines_);
-          }
-          onChanged();
-        }
-      } else {
-        if (!other.lines_.isEmpty()) {
-          if (linesBuilder_.isEmpty()) {
-            linesBuilder_.dispose();
-            linesBuilder_ = null;
-            lines_ = other.lines_;
-            bitField0_ = (bitField0_ & ~0x00000001);
-            linesBuilder_ = 
-              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
-                 getLinesFieldBuilder() : null;
-          } else {
-            linesBuilder_.addAllMessages(other.lines_);
-          }
-        }
-      }
-      internalGetMutableEventMetadata().mergeFrom(
-          other.internalGetEventMetadata());
-      internalGetMutableStatMetadata().mergeFrom(
-          other.internalGetStatMetadata());
-      if (statsBuilder_ == null) {
-        if (!other.stats_.isEmpty()) {
-          if (stats_.isEmpty()) {
-            stats_ = other.stats_;
-            bitField0_ = (bitField0_ & ~0x00000008);
-          } else {
-            ensureStatsIsMutable();
-            stats_.addAll(other.stats_);
-          }
-          onChanged();
-        }
-      } else {
-        if (!other.stats_.isEmpty()) {
-          if (statsBuilder_.isEmpty()) {
-            statsBuilder_.dispose();
-            statsBuilder_ = null;
-            stats_ = other.stats_;
-            bitField0_ = (bitField0_ & ~0x00000008);
-            statsBuilder_ = 
-              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
-                 getStatsFieldBuilder() : null;
-          } else {
-            statsBuilder_.addAllMessages(other.stats_);
-          }
-        }
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.profiler.XPlane parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.profiler.XPlane) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-    private int bitField0_;
-
-    private long id_ ;
-    /**
-     * <code>int64 id = 1;</code>
-     */
-    public long getId() {
-      return id_;
-    }
-    /**
-     * <code>int64 id = 1;</code>
-     */
-    public Builder setId(long value) {
-      
-      id_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>int64 id = 1;</code>
-     */
-    public Builder clearId() {
-      
-      id_ = 0L;
-      onChanged();
-      return this;
-    }
-
-    private java.lang.Object name_ = "";
-    /**
-     * <pre>
-     * Name of this line.
-     * </pre>
-     *
-     * <code>string name = 2;</code>
-     */
-    public java.lang.String getName() {
-      java.lang.Object ref = name_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        name_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Name of this line.
-     * </pre>
-     *
-     * <code>string name = 2;</code>
-     */
-    public com.google.protobuf.ByteString
-        getNameBytes() {
-      java.lang.Object ref = name_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        name_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Name of this line.
-     * </pre>
-     *
-     * <code>string name = 2;</code>
-     */
-    public Builder setName(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      name_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Name of this line.
-     * </pre>
-     *
-     * <code>string name = 2;</code>
-     */
-    public Builder clearName() {
-      
-      name_ = getDefaultInstance().getName();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Name of this line.
-     * </pre>
-     *
-     * <code>string name = 2;</code>
-     */
-    public Builder setNameBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      name_ = value;
-      onChanged();
-      return this;
-    }
-
-    private java.util.List<org.tensorflow.proto.profiler.XLine> lines_ =
-      java.util.Collections.emptyList();
-    private void ensureLinesIsMutable() {
-      if (!((bitField0_ & 0x00000001) != 0)) {
-        lines_ = new java.util.ArrayList<org.tensorflow.proto.profiler.XLine>(lines_);
-        bitField0_ |= 0x00000001;
-       }
-    }
-
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.profiler.XLine, org.tensorflow.proto.profiler.XLine.Builder, org.tensorflow.proto.profiler.XLineOrBuilder> linesBuilder_;
-
-    /**
-     * <pre>
-     * Parallel timelines grouped in this plane. XLines with the same id
-     * are effectively the same timeline.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.profiler.XLine lines = 3;</code>
-     */
-    public java.util.List<org.tensorflow.proto.profiler.XLine> getLinesList() {
-      if (linesBuilder_ == null) {
-        return java.util.Collections.unmodifiableList(lines_);
-      } else {
-        return linesBuilder_.getMessageList();
-      }
-    }
-    /**
-     * <pre>
-     * Parallel timelines grouped in this plane. XLines with the same id
-     * are effectively the same timeline.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.profiler.XLine lines = 3;</code>
-     */
-    public int getLinesCount() {
-      if (linesBuilder_ == null) {
-        return lines_.size();
-      } else {
-        return linesBuilder_.getCount();
-      }
-    }
-    /**
-     * <pre>
-     * Parallel timelines grouped in this plane. XLines with the same id
-     * are effectively the same timeline.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.profiler.XLine lines = 3;</code>
-     */
-    public org.tensorflow.proto.profiler.XLine getLines(int index) {
-      if (linesBuilder_ == null) {
-        return lines_.get(index);
-      } else {
-        return linesBuilder_.getMessage(index);
-      }
-    }
-    /**
-     * <pre>
-     * Parallel timelines grouped in this plane. XLines with the same id
-     * are effectively the same timeline.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.profiler.XLine lines = 3;</code>
-     */
-    public Builder setLines(
-        int index, org.tensorflow.proto.profiler.XLine value) {
-      if (linesBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureLinesIsMutable();
-        lines_.set(index, value);
-        onChanged();
-      } else {
-        linesBuilder_.setMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Parallel timelines grouped in this plane. XLines with the same id
-     * are effectively the same timeline.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.profiler.XLine lines = 3;</code>
-     */
-    public Builder setLines(
-        int index, org.tensorflow.proto.profiler.XLine.Builder builderForValue) {
-      if (linesBuilder_ == null) {
-        ensureLinesIsMutable();
-        lines_.set(index, builderForValue.build());
-        onChanged();
-      } else {
-        linesBuilder_.setMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Parallel timelines grouped in this plane. XLines with the same id
-     * are effectively the same timeline.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.profiler.XLine lines = 3;</code>
-     */
-    public Builder addLines(org.tensorflow.proto.profiler.XLine value) {
-      if (linesBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureLinesIsMutable();
-        lines_.add(value);
-        onChanged();
-      } else {
-        linesBuilder_.addMessage(value);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Parallel timelines grouped in this plane. XLines with the same id
-     * are effectively the same timeline.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.profiler.XLine lines = 3;</code>
-     */
-    public Builder addLines(
-        int index, org.tensorflow.proto.profiler.XLine value) {
-      if (linesBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureLinesIsMutable();
-        lines_.add(index, value);
-        onChanged();
-      } else {
-        linesBuilder_.addMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Parallel timelines grouped in this plane. XLines with the same id
-     * are effectively the same timeline.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.profiler.XLine lines = 3;</code>
-     */
-    public Builder addLines(
-        org.tensorflow.proto.profiler.XLine.Builder builderForValue) {
-      if (linesBuilder_ == null) {
-        ensureLinesIsMutable();
-        lines_.add(builderForValue.build());
-        onChanged();
-      } else {
-        linesBuilder_.addMessage(builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Parallel timelines grouped in this plane. XLines with the same id
-     * are effectively the same timeline.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.profiler.XLine lines = 3;</code>
-     */
-    public Builder addLines(
-        int index, org.tensorflow.proto.profiler.XLine.Builder builderForValue) {
-      if (linesBuilder_ == null) {
-        ensureLinesIsMutable();
-        lines_.add(index, builderForValue.build());
-        onChanged();
-      } else {
-        linesBuilder_.addMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Parallel timelines grouped in this plane. XLines with the same id
-     * are effectively the same timeline.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.profiler.XLine lines = 3;</code>
-     */
-    public Builder addAllLines(
-        java.lang.Iterable<? extends org.tensorflow.proto.profiler.XLine> values) {
-      if (linesBuilder_ == null) {
-        ensureLinesIsMutable();
-        com.google.protobuf.AbstractMessageLite.Builder.addAll(
-            values, lines_);
-        onChanged();
-      } else {
-        linesBuilder_.addAllMessages(values);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Parallel timelines grouped in this plane. XLines with the same id
-     * are effectively the same timeline.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.profiler.XLine lines = 3;</code>
-     */
-    public Builder clearLines() {
-      if (linesBuilder_ == null) {
-        lines_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000001);
-        onChanged();
-      } else {
-        linesBuilder_.clear();
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Parallel timelines grouped in this plane. XLines with the same id
-     * are effectively the same timeline.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.profiler.XLine lines = 3;</code>
-     */
-    public Builder removeLines(int index) {
-      if (linesBuilder_ == null) {
-        ensureLinesIsMutable();
-        lines_.remove(index);
-        onChanged();
-      } else {
-        linesBuilder_.remove(index);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Parallel timelines grouped in this plane. XLines with the same id
-     * are effectively the same timeline.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.profiler.XLine lines = 3;</code>
-     */
-    public org.tensorflow.proto.profiler.XLine.Builder getLinesBuilder(
-        int index) {
-      return getLinesFieldBuilder().getBuilder(index);
-    }
-    /**
-     * <pre>
-     * Parallel timelines grouped in this plane. XLines with the same id
-     * are effectively the same timeline.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.profiler.XLine lines = 3;</code>
-     */
-    public org.tensorflow.proto.profiler.XLineOrBuilder getLinesOrBuilder(
-        int index) {
-      if (linesBuilder_ == null) {
-        return lines_.get(index);  } else {
-        return linesBuilder_.getMessageOrBuilder(index);
-      }
-    }
-    /**
-     * <pre>
-     * Parallel timelines grouped in this plane. XLines with the same id
-     * are effectively the same timeline.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.profiler.XLine lines = 3;</code>
-     */
-    public java.util.List<? extends org.tensorflow.proto.profiler.XLineOrBuilder> 
-         getLinesOrBuilderList() {
-      if (linesBuilder_ != null) {
-        return linesBuilder_.getMessageOrBuilderList();
-      } else {
-        return java.util.Collections.unmodifiableList(lines_);
-      }
-    }
-    /**
-     * <pre>
-     * Parallel timelines grouped in this plane. XLines with the same id
-     * are effectively the same timeline.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.profiler.XLine lines = 3;</code>
-     */
-    public org.tensorflow.proto.profiler.XLine.Builder addLinesBuilder() {
-      return getLinesFieldBuilder().addBuilder(
-          org.tensorflow.proto.profiler.XLine.getDefaultInstance());
-    }
-    /**
-     * <pre>
-     * Parallel timelines grouped in this plane. XLines with the same id
-     * are effectively the same timeline.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.profiler.XLine lines = 3;</code>
-     */
-    public org.tensorflow.proto.profiler.XLine.Builder addLinesBuilder(
-        int index) {
-      return getLinesFieldBuilder().addBuilder(
-          index, org.tensorflow.proto.profiler.XLine.getDefaultInstance());
-    }
-    /**
-     * <pre>
-     * Parallel timelines grouped in this plane. XLines with the same id
-     * are effectively the same timeline.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.profiler.XLine lines = 3;</code>
-     */
-    public java.util.List<org.tensorflow.proto.profiler.XLine.Builder> 
-         getLinesBuilderList() {
-      return getLinesFieldBuilder().getBuilderList();
-    }
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.profiler.XLine, org.tensorflow.proto.profiler.XLine.Builder, org.tensorflow.proto.profiler.XLineOrBuilder> 
-        getLinesFieldBuilder() {
-      if (linesBuilder_ == null) {
-        linesBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
-            org.tensorflow.proto.profiler.XLine, org.tensorflow.proto.profiler.XLine.Builder, org.tensorflow.proto.profiler.XLineOrBuilder>(
-                lines_,
-                ((bitField0_ & 0x00000001) != 0),
-                getParentForChildren(),
-                isClean());
-        lines_ = null;
-      }
-      return linesBuilder_;
-    }
-
-    private com.google.protobuf.MapField<
-        java.lang.Long, org.tensorflow.proto.profiler.XEventMetadata> eventMetadata_;
-    private com.google.protobuf.MapField<java.lang.Long, org.tensorflow.proto.profiler.XEventMetadata>
-    internalGetEventMetadata() {
-      if (eventMetadata_ == null) {
-        return com.google.protobuf.MapField.emptyMapField(
-            EventMetadataDefaultEntryHolder.defaultEntry);
-      }
-      return eventMetadata_;
-    }
-    private com.google.protobuf.MapField<java.lang.Long, org.tensorflow.proto.profiler.XEventMetadata>
-    internalGetMutableEventMetadata() {
-      onChanged();;
-      if (eventMetadata_ == null) {
-        eventMetadata_ = com.google.protobuf.MapField.newMapField(
-            EventMetadataDefaultEntryHolder.defaultEntry);
-      }
-      if (!eventMetadata_.isMutable()) {
-        eventMetadata_ = eventMetadata_.copy();
-      }
-      return eventMetadata_;
-    }
-
-    public int getEventMetadataCount() {
-      return internalGetEventMetadata().getMap().size();
-    }
-    /**
-     * <pre>
-     * XEventMetadata map, each entry uses the XEventMetadata.id as key. This map
-     * should be used for events that share the same ID over the whole XPlane.
-     * </pre>
-     *
-     * <code>map&lt;int64, .tensorflow.profiler.XEventMetadata&gt; event_metadata = 4;</code>
-     */
-
-    public boolean containsEventMetadata(
-        long key) {
-      
-      return internalGetEventMetadata().getMap().containsKey(key);
-    }
-    /**
-     * Use {@link #getEventMetadataMap()} instead.
-     */
-    @java.lang.Deprecated
-    public java.util.Map<java.lang.Long, org.tensorflow.proto.profiler.XEventMetadata> getEventMetadata() {
-      return getEventMetadataMap();
-    }
-    /**
-     * <pre>
-     * XEventMetadata map, each entry uses the XEventMetadata.id as key. This map
-     * should be used for events that share the same ID over the whole XPlane.
-     * </pre>
-     *
-     * <code>map&lt;int64, .tensorflow.profiler.XEventMetadata&gt; event_metadata = 4;</code>
-     */
-
-    public java.util.Map<java.lang.Long, org.tensorflow.proto.profiler.XEventMetadata> getEventMetadataMap() {
-      return internalGetEventMetadata().getMap();
-    }
-    /**
-     * <pre>
-     * XEventMetadata map, each entry uses the XEventMetadata.id as key. This map
-     * should be used for events that share the same ID over the whole XPlane.
-     * </pre>
-     *
-     * <code>map&lt;int64, .tensorflow.profiler.XEventMetadata&gt; event_metadata = 4;</code>
-     */
-
-    public org.tensorflow.proto.profiler.XEventMetadata getEventMetadataOrDefault(
-        long key,
-        org.tensorflow.proto.profiler.XEventMetadata defaultValue) {
-      
-      java.util.Map<java.lang.Long, org.tensorflow.proto.profiler.XEventMetadata> map =
-          internalGetEventMetadata().getMap();
-      return map.containsKey(key) ? map.get(key) : defaultValue;
-    }
-    /**
-     * <pre>
-     * XEventMetadata map, each entry uses the XEventMetadata.id as key. This map
-     * should be used for events that share the same ID over the whole XPlane.
-     * </pre>
-     *
-     * <code>map&lt;int64, .tensorflow.profiler.XEventMetadata&gt; event_metadata = 4;</code>
-     */
-
-    public org.tensorflow.proto.profiler.XEventMetadata getEventMetadataOrThrow(
-        long key) {
-      
-      java.util.Map<java.lang.Long, org.tensorflow.proto.profiler.XEventMetadata> map =
-          internalGetEventMetadata().getMap();
-      if (!map.containsKey(key)) {
-        throw new java.lang.IllegalArgumentException();
-      }
-      return map.get(key);
-    }
-
-    public Builder clearEventMetadata() {
-      internalGetMutableEventMetadata().getMutableMap()
-          .clear();
-      return this;
-    }
-    /**
-     * <pre>
-     * XEventMetadata map, each entry uses the XEventMetadata.id as key. This map
-     * should be used for events that share the same ID over the whole XPlane.
-     * </pre>
-     *
-     * <code>map&lt;int64, .tensorflow.profiler.XEventMetadata&gt; event_metadata = 4;</code>
-     */
-
-    public Builder removeEventMetadata(
-        long key) {
-      
-      internalGetMutableEventMetadata().getMutableMap()
-          .remove(key);
-      return this;
-    }
-    /**
-     * Use alternate mutation accessors instead.
-     */
-    @java.lang.Deprecated
-    public java.util.Map<java.lang.Long, org.tensorflow.proto.profiler.XEventMetadata>
-    getMutableEventMetadata() {
-      return internalGetMutableEventMetadata().getMutableMap();
-    }
-    /**
-     * <pre>
-     * XEventMetadata map, each entry uses the XEventMetadata.id as key. This map
-     * should be used for events that share the same ID over the whole XPlane.
-     * </pre>
-     *
-     * <code>map&lt;int64, .tensorflow.profiler.XEventMetadata&gt; event_metadata = 4;</code>
-     */
-    public Builder putEventMetadata(
-        long key,
-        org.tensorflow.proto.profiler.XEventMetadata value) {
-      
-      if (value == null) { throw new java.lang.NullPointerException(); }
-      internalGetMutableEventMetadata().getMutableMap()
-          .put(key, value);
-      return this;
-    }
-    /**
-     * <pre>
-     * XEventMetadata map, each entry uses the XEventMetadata.id as key. This map
-     * should be used for events that share the same ID over the whole XPlane.
-     * </pre>
-     *
-     * <code>map&lt;int64, .tensorflow.profiler.XEventMetadata&gt; event_metadata = 4;</code>
-     */
-
-    public Builder putAllEventMetadata(
-        java.util.Map<java.lang.Long, org.tensorflow.proto.profiler.XEventMetadata> values) {
-      internalGetMutableEventMetadata().getMutableMap()
-          .putAll(values);
-      return this;
-    }
-
-    private com.google.protobuf.MapField<
-        java.lang.Long, org.tensorflow.proto.profiler.XStatMetadata> statMetadata_;
-    private com.google.protobuf.MapField<java.lang.Long, org.tensorflow.proto.profiler.XStatMetadata>
-    internalGetStatMetadata() {
-      if (statMetadata_ == null) {
-        return com.google.protobuf.MapField.emptyMapField(
-            StatMetadataDefaultEntryHolder.defaultEntry);
-      }
-      return statMetadata_;
-    }
-    private com.google.protobuf.MapField<java.lang.Long, org.tensorflow.proto.profiler.XStatMetadata>
-    internalGetMutableStatMetadata() {
-      onChanged();;
-      if (statMetadata_ == null) {
-        statMetadata_ = com.google.protobuf.MapField.newMapField(
-            StatMetadataDefaultEntryHolder.defaultEntry);
-      }
-      if (!statMetadata_.isMutable()) {
-        statMetadata_ = statMetadata_.copy();
-      }
-      return statMetadata_;
-    }
-
-    public int getStatMetadataCount() {
-      return internalGetStatMetadata().getMap().size();
-    }
-    /**
-     * <pre>
-     * XStatMetadata map, each entry uses the XStatMetadata.id as key. This map
-     * should be used for stats that share the same ID over the whole XPlane.
-     * </pre>
-     *
-     * <code>map&lt;int64, .tensorflow.profiler.XStatMetadata&gt; stat_metadata = 5;</code>
-     */
-
-    public boolean containsStatMetadata(
-        long key) {
-      
-      return internalGetStatMetadata().getMap().containsKey(key);
-    }
-    /**
-     * Use {@link #getStatMetadataMap()} instead.
-     */
-    @java.lang.Deprecated
-    public java.util.Map<java.lang.Long, org.tensorflow.proto.profiler.XStatMetadata> getStatMetadata() {
-      return getStatMetadataMap();
-    }
-    /**
-     * <pre>
-     * XStatMetadata map, each entry uses the XStatMetadata.id as key. This map
-     * should be used for stats that share the same ID over the whole XPlane.
-     * </pre>
-     *
-     * <code>map&lt;int64, .tensorflow.profiler.XStatMetadata&gt; stat_metadata = 5;</code>
-     */
-
-    public java.util.Map<java.lang.Long, org.tensorflow.proto.profiler.XStatMetadata> getStatMetadataMap() {
-      return internalGetStatMetadata().getMap();
-    }
-    /**
-     * <pre>
-     * XStatMetadata map, each entry uses the XStatMetadata.id as key. This map
-     * should be used for stats that share the same ID over the whole XPlane.
-     * </pre>
-     *
-     * <code>map&lt;int64, .tensorflow.profiler.XStatMetadata&gt; stat_metadata = 5;</code>
-     */
-
-    public org.tensorflow.proto.profiler.XStatMetadata getStatMetadataOrDefault(
-        long key,
-        org.tensorflow.proto.profiler.XStatMetadata defaultValue) {
-      
-      java.util.Map<java.lang.Long, org.tensorflow.proto.profiler.XStatMetadata> map =
-          internalGetStatMetadata().getMap();
-      return map.containsKey(key) ? map.get(key) : defaultValue;
-    }
-    /**
-     * <pre>
-     * XStatMetadata map, each entry uses the XStatMetadata.id as key. This map
-     * should be used for stats that share the same ID over the whole XPlane.
-     * </pre>
-     *
-     * <code>map&lt;int64, .tensorflow.profiler.XStatMetadata&gt; stat_metadata = 5;</code>
-     */
-
-    public org.tensorflow.proto.profiler.XStatMetadata getStatMetadataOrThrow(
-        long key) {
-      
-      java.util.Map<java.lang.Long, org.tensorflow.proto.profiler.XStatMetadata> map =
-          internalGetStatMetadata().getMap();
-      if (!map.containsKey(key)) {
-        throw new java.lang.IllegalArgumentException();
-      }
-      return map.get(key);
-    }
-
-    public Builder clearStatMetadata() {
-      internalGetMutableStatMetadata().getMutableMap()
-          .clear();
-      return this;
-    }
-    /**
-     * <pre>
-     * XStatMetadata map, each entry uses the XStatMetadata.id as key. This map
-     * should be used for stats that share the same ID over the whole XPlane.
-     * </pre>
-     *
-     * <code>map&lt;int64, .tensorflow.profiler.XStatMetadata&gt; stat_metadata = 5;</code>
-     */
-
-    public Builder removeStatMetadata(
-        long key) {
-      
-      internalGetMutableStatMetadata().getMutableMap()
-          .remove(key);
-      return this;
-    }
-    /**
-     * Use alternate mutation accessors instead.
-     */
-    @java.lang.Deprecated
-    public java.util.Map<java.lang.Long, org.tensorflow.proto.profiler.XStatMetadata>
-    getMutableStatMetadata() {
-      return internalGetMutableStatMetadata().getMutableMap();
-    }
-    /**
-     * <pre>
-     * XStatMetadata map, each entry uses the XStatMetadata.id as key. This map
-     * should be used for stats that share the same ID over the whole XPlane.
-     * </pre>
-     *
-     * <code>map&lt;int64, .tensorflow.profiler.XStatMetadata&gt; stat_metadata = 5;</code>
-     */
-    public Builder putStatMetadata(
-        long key,
-        org.tensorflow.proto.profiler.XStatMetadata value) {
-      
-      if (value == null) { throw new java.lang.NullPointerException(); }
-      internalGetMutableStatMetadata().getMutableMap()
-          .put(key, value);
-      return this;
-    }
-    /**
-     * <pre>
-     * XStatMetadata map, each entry uses the XStatMetadata.id as key. This map
-     * should be used for stats that share the same ID over the whole XPlane.
-     * </pre>
-     *
-     * <code>map&lt;int64, .tensorflow.profiler.XStatMetadata&gt; stat_metadata = 5;</code>
-     */
-
-    public Builder putAllStatMetadata(
-        java.util.Map<java.lang.Long, org.tensorflow.proto.profiler.XStatMetadata> values) {
-      internalGetMutableStatMetadata().getMutableMap()
-          .putAll(values);
-      return this;
-    }
-
-    private java.util.List<org.tensorflow.proto.profiler.XStat> stats_ =
-      java.util.Collections.emptyList();
-    private void ensureStatsIsMutable() {
-      if (!((bitField0_ & 0x00000008) != 0)) {
-        stats_ = new java.util.ArrayList<org.tensorflow.proto.profiler.XStat>(stats_);
-        bitField0_ |= 0x00000008;
-       }
-    }
-
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.profiler.XStat, org.tensorflow.proto.profiler.XStat.Builder, org.tensorflow.proto.profiler.XStatOrBuilder> statsBuilder_;
-
-    /**
-     * <pre>
-     * XStats associated with this plane, e.g. device capabilities.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.profiler.XStat stats = 6;</code>
-     */
-    public java.util.List<org.tensorflow.proto.profiler.XStat> getStatsList() {
-      if (statsBuilder_ == null) {
-        return java.util.Collections.unmodifiableList(stats_);
-      } else {
-        return statsBuilder_.getMessageList();
-      }
-    }
-    /**
-     * <pre>
-     * XStats associated with this plane, e.g. device capabilities.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.profiler.XStat stats = 6;</code>
-     */
-    public int getStatsCount() {
-      if (statsBuilder_ == null) {
-        return stats_.size();
-      } else {
-        return statsBuilder_.getCount();
-      }
-    }
-    /**
-     * <pre>
-     * XStats associated with this plane, e.g. device capabilities.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.profiler.XStat stats = 6;</code>
-     */
-    public org.tensorflow.proto.profiler.XStat getStats(int index) {
-      if (statsBuilder_ == null) {
-        return stats_.get(index);
-      } else {
-        return statsBuilder_.getMessage(index);
-      }
-    }
-    /**
-     * <pre>
-     * XStats associated with this plane, e.g. device capabilities.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.profiler.XStat stats = 6;</code>
-     */
-    public Builder setStats(
-        int index, org.tensorflow.proto.profiler.XStat value) {
-      if (statsBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureStatsIsMutable();
-        stats_.set(index, value);
-        onChanged();
-      } else {
-        statsBuilder_.setMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * XStats associated with this plane, e.g. device capabilities.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.profiler.XStat stats = 6;</code>
-     */
-    public Builder setStats(
-        int index, org.tensorflow.proto.profiler.XStat.Builder builderForValue) {
-      if (statsBuilder_ == null) {
-        ensureStatsIsMutable();
-        stats_.set(index, builderForValue.build());
-        onChanged();
-      } else {
-        statsBuilder_.setMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * XStats associated with this plane, e.g. device capabilities.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.profiler.XStat stats = 6;</code>
-     */
-    public Builder addStats(org.tensorflow.proto.profiler.XStat value) {
-      if (statsBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureStatsIsMutable();
-        stats_.add(value);
-        onChanged();
-      } else {
-        statsBuilder_.addMessage(value);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * XStats associated with this plane, e.g. device capabilities.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.profiler.XStat stats = 6;</code>
-     */
-    public Builder addStats(
-        int index, org.tensorflow.proto.profiler.XStat value) {
-      if (statsBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureStatsIsMutable();
-        stats_.add(index, value);
-        onChanged();
-      } else {
-        statsBuilder_.addMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * XStats associated with this plane, e.g. device capabilities.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.profiler.XStat stats = 6;</code>
-     */
-    public Builder addStats(
-        org.tensorflow.proto.profiler.XStat.Builder builderForValue) {
-      if (statsBuilder_ == null) {
-        ensureStatsIsMutable();
-        stats_.add(builderForValue.build());
-        onChanged();
-      } else {
-        statsBuilder_.addMessage(builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * XStats associated with this plane, e.g. device capabilities.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.profiler.XStat stats = 6;</code>
-     */
-    public Builder addStats(
-        int index, org.tensorflow.proto.profiler.XStat.Builder builderForValue) {
-      if (statsBuilder_ == null) {
-        ensureStatsIsMutable();
-        stats_.add(index, builderForValue.build());
-        onChanged();
-      } else {
-        statsBuilder_.addMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * XStats associated with this plane, e.g. device capabilities.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.profiler.XStat stats = 6;</code>
-     */
-    public Builder addAllStats(
-        java.lang.Iterable<? extends org.tensorflow.proto.profiler.XStat> values) {
-      if (statsBuilder_ == null) {
-        ensureStatsIsMutable();
-        com.google.protobuf.AbstractMessageLite.Builder.addAll(
-            values, stats_);
-        onChanged();
-      } else {
-        statsBuilder_.addAllMessages(values);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * XStats associated with this plane, e.g. device capabilities.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.profiler.XStat stats = 6;</code>
-     */
-    public Builder clearStats() {
-      if (statsBuilder_ == null) {
-        stats_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000008);
-        onChanged();
-      } else {
-        statsBuilder_.clear();
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * XStats associated with this plane, e.g. device capabilities.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.profiler.XStat stats = 6;</code>
-     */
-    public Builder removeStats(int index) {
-      if (statsBuilder_ == null) {
-        ensureStatsIsMutable();
-        stats_.remove(index);
-        onChanged();
-      } else {
-        statsBuilder_.remove(index);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * XStats associated with this plane, e.g. device capabilities.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.profiler.XStat stats = 6;</code>
-     */
-    public org.tensorflow.proto.profiler.XStat.Builder getStatsBuilder(
-        int index) {
-      return getStatsFieldBuilder().getBuilder(index);
-    }
-    /**
-     * <pre>
-     * XStats associated with this plane, e.g. device capabilities.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.profiler.XStat stats = 6;</code>
-     */
-    public org.tensorflow.proto.profiler.XStatOrBuilder getStatsOrBuilder(
-        int index) {
-      if (statsBuilder_ == null) {
-        return stats_.get(index);  } else {
-        return statsBuilder_.getMessageOrBuilder(index);
-      }
-    }
-    /**
-     * <pre>
-     * XStats associated with this plane, e.g. device capabilities.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.profiler.XStat stats = 6;</code>
-     */
-    public java.util.List<? extends org.tensorflow.proto.profiler.XStatOrBuilder> 
-         getStatsOrBuilderList() {
-      if (statsBuilder_ != null) {
-        return statsBuilder_.getMessageOrBuilderList();
-      } else {
-        return java.util.Collections.unmodifiableList(stats_);
-      }
-    }
-    /**
-     * <pre>
-     * XStats associated with this plane, e.g. device capabilities.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.profiler.XStat stats = 6;</code>
-     */
-    public org.tensorflow.proto.profiler.XStat.Builder addStatsBuilder() {
-      return getStatsFieldBuilder().addBuilder(
-          org.tensorflow.proto.profiler.XStat.getDefaultInstance());
-    }
-    /**
-     * <pre>
-     * XStats associated with this plane, e.g. device capabilities.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.profiler.XStat stats = 6;</code>
-     */
-    public org.tensorflow.proto.profiler.XStat.Builder addStatsBuilder(
-        int index) {
-      return getStatsFieldBuilder().addBuilder(
-          index, org.tensorflow.proto.profiler.XStat.getDefaultInstance());
-    }
-    /**
-     * <pre>
-     * XStats associated with this plane, e.g. device capabilities.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.profiler.XStat stats = 6;</code>
-     */
-    public java.util.List<org.tensorflow.proto.profiler.XStat.Builder> 
-         getStatsBuilderList() {
-      return getStatsFieldBuilder().getBuilderList();
-    }
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.profiler.XStat, org.tensorflow.proto.profiler.XStat.Builder, org.tensorflow.proto.profiler.XStatOrBuilder> 
-        getStatsFieldBuilder() {
-      if (statsBuilder_ == null) {
-        statsBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
-            org.tensorflow.proto.profiler.XStat, org.tensorflow.proto.profiler.XStat.Builder, org.tensorflow.proto.profiler.XStatOrBuilder>(
-                stats_,
-                ((bitField0_ & 0x00000008) != 0),
-                getParentForChildren(),
-                isClean());
-        stats_ = null;
-      }
-      return statsBuilder_;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.profiler.XPlane)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.profiler.XPlane)
-  private static final org.tensorflow.proto.profiler.XPlane DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.profiler.XPlane();
-  }
-
-  public static org.tensorflow.proto.profiler.XPlane getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<XPlane>
-      PARSER = new com.google.protobuf.AbstractParser<XPlane>() {
-    @java.lang.Override
-    public XPlane parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new XPlane(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<XPlane> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<XPlane> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.profiler.XPlane getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/XPlaneOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/XPlaneOrBuilder.java
deleted file mode 100644
index a6b97a4c6dd..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/XPlaneOrBuilder.java
+++ /dev/null
@@ -1,243 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/profiler/protobuf/xplane.proto
-
-package org.tensorflow.proto.profiler;
-
-public interface XPlaneOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.profiler.XPlane)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <code>int64 id = 1;</code>
-   */
-  long getId();
-
-  /**
-   * <pre>
-   * Name of this line.
-   * </pre>
-   *
-   * <code>string name = 2;</code>
-   */
-  java.lang.String getName();
-  /**
-   * <pre>
-   * Name of this line.
-   * </pre>
-   *
-   * <code>string name = 2;</code>
-   */
-  com.google.protobuf.ByteString
-      getNameBytes();
-
-  /**
-   * <pre>
-   * Parallel timelines grouped in this plane. XLines with the same id
-   * are effectively the same timeline.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.profiler.XLine lines = 3;</code>
-   */
-  java.util.List<org.tensorflow.proto.profiler.XLine> 
-      getLinesList();
-  /**
-   * <pre>
-   * Parallel timelines grouped in this plane. XLines with the same id
-   * are effectively the same timeline.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.profiler.XLine lines = 3;</code>
-   */
-  org.tensorflow.proto.profiler.XLine getLines(int index);
-  /**
-   * <pre>
-   * Parallel timelines grouped in this plane. XLines with the same id
-   * are effectively the same timeline.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.profiler.XLine lines = 3;</code>
-   */
-  int getLinesCount();
-  /**
-   * <pre>
-   * Parallel timelines grouped in this plane. XLines with the same id
-   * are effectively the same timeline.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.profiler.XLine lines = 3;</code>
-   */
-  java.util.List<? extends org.tensorflow.proto.profiler.XLineOrBuilder> 
-      getLinesOrBuilderList();
-  /**
-   * <pre>
-   * Parallel timelines grouped in this plane. XLines with the same id
-   * are effectively the same timeline.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.profiler.XLine lines = 3;</code>
-   */
-  org.tensorflow.proto.profiler.XLineOrBuilder getLinesOrBuilder(
-      int index);
-
-  /**
-   * <pre>
-   * XEventMetadata map, each entry uses the XEventMetadata.id as key. This map
-   * should be used for events that share the same ID over the whole XPlane.
-   * </pre>
-   *
-   * <code>map&lt;int64, .tensorflow.profiler.XEventMetadata&gt; event_metadata = 4;</code>
-   */
-  int getEventMetadataCount();
-  /**
-   * <pre>
-   * XEventMetadata map, each entry uses the XEventMetadata.id as key. This map
-   * should be used for events that share the same ID over the whole XPlane.
-   * </pre>
-   *
-   * <code>map&lt;int64, .tensorflow.profiler.XEventMetadata&gt; event_metadata = 4;</code>
-   */
-  boolean containsEventMetadata(
-      long key);
-  /**
-   * Use {@link #getEventMetadataMap()} instead.
-   */
-  @java.lang.Deprecated
-  java.util.Map<java.lang.Long, org.tensorflow.proto.profiler.XEventMetadata>
-  getEventMetadata();
-  /**
-   * <pre>
-   * XEventMetadata map, each entry uses the XEventMetadata.id as key. This map
-   * should be used for events that share the same ID over the whole XPlane.
-   * </pre>
-   *
-   * <code>map&lt;int64, .tensorflow.profiler.XEventMetadata&gt; event_metadata = 4;</code>
-   */
-  java.util.Map<java.lang.Long, org.tensorflow.proto.profiler.XEventMetadata>
-  getEventMetadataMap();
-  /**
-   * <pre>
-   * XEventMetadata map, each entry uses the XEventMetadata.id as key. This map
-   * should be used for events that share the same ID over the whole XPlane.
-   * </pre>
-   *
-   * <code>map&lt;int64, .tensorflow.profiler.XEventMetadata&gt; event_metadata = 4;</code>
-   */
-
-  org.tensorflow.proto.profiler.XEventMetadata getEventMetadataOrDefault(
-      long key,
-      org.tensorflow.proto.profiler.XEventMetadata defaultValue);
-  /**
-   * <pre>
-   * XEventMetadata map, each entry uses the XEventMetadata.id as key. This map
-   * should be used for events that share the same ID over the whole XPlane.
-   * </pre>
-   *
-   * <code>map&lt;int64, .tensorflow.profiler.XEventMetadata&gt; event_metadata = 4;</code>
-   */
-
-  org.tensorflow.proto.profiler.XEventMetadata getEventMetadataOrThrow(
-      long key);
-
-  /**
-   * <pre>
-   * XStatMetadata map, each entry uses the XStatMetadata.id as key. This map
-   * should be used for stats that share the same ID over the whole XPlane.
-   * </pre>
-   *
-   * <code>map&lt;int64, .tensorflow.profiler.XStatMetadata&gt; stat_metadata = 5;</code>
-   */
-  int getStatMetadataCount();
-  /**
-   * <pre>
-   * XStatMetadata map, each entry uses the XStatMetadata.id as key. This map
-   * should be used for stats that share the same ID over the whole XPlane.
-   * </pre>
-   *
-   * <code>map&lt;int64, .tensorflow.profiler.XStatMetadata&gt; stat_metadata = 5;</code>
-   */
-  boolean containsStatMetadata(
-      long key);
-  /**
-   * Use {@link #getStatMetadataMap()} instead.
-   */
-  @java.lang.Deprecated
-  java.util.Map<java.lang.Long, org.tensorflow.proto.profiler.XStatMetadata>
-  getStatMetadata();
-  /**
-   * <pre>
-   * XStatMetadata map, each entry uses the XStatMetadata.id as key. This map
-   * should be used for stats that share the same ID over the whole XPlane.
-   * </pre>
-   *
-   * <code>map&lt;int64, .tensorflow.profiler.XStatMetadata&gt; stat_metadata = 5;</code>
-   */
-  java.util.Map<java.lang.Long, org.tensorflow.proto.profiler.XStatMetadata>
-  getStatMetadataMap();
-  /**
-   * <pre>
-   * XStatMetadata map, each entry uses the XStatMetadata.id as key. This map
-   * should be used for stats that share the same ID over the whole XPlane.
-   * </pre>
-   *
-   * <code>map&lt;int64, .tensorflow.profiler.XStatMetadata&gt; stat_metadata = 5;</code>
-   */
-
-  org.tensorflow.proto.profiler.XStatMetadata getStatMetadataOrDefault(
-      long key,
-      org.tensorflow.proto.profiler.XStatMetadata defaultValue);
-  /**
-   * <pre>
-   * XStatMetadata map, each entry uses the XStatMetadata.id as key. This map
-   * should be used for stats that share the same ID over the whole XPlane.
-   * </pre>
-   *
-   * <code>map&lt;int64, .tensorflow.profiler.XStatMetadata&gt; stat_metadata = 5;</code>
-   */
-
-  org.tensorflow.proto.profiler.XStatMetadata getStatMetadataOrThrow(
-      long key);
-
-  /**
-   * <pre>
-   * XStats associated with this plane, e.g. device capabilities.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.profiler.XStat stats = 6;</code>
-   */
-  java.util.List<org.tensorflow.proto.profiler.XStat> 
-      getStatsList();
-  /**
-   * <pre>
-   * XStats associated with this plane, e.g. device capabilities.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.profiler.XStat stats = 6;</code>
-   */
-  org.tensorflow.proto.profiler.XStat getStats(int index);
-  /**
-   * <pre>
-   * XStats associated with this plane, e.g. device capabilities.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.profiler.XStat stats = 6;</code>
-   */
-  int getStatsCount();
-  /**
-   * <pre>
-   * XStats associated with this plane, e.g. device capabilities.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.profiler.XStat stats = 6;</code>
-   */
-  java.util.List<? extends org.tensorflow.proto.profiler.XStatOrBuilder> 
-      getStatsOrBuilderList();
-  /**
-   * <pre>
-   * XStats associated with this plane, e.g. device capabilities.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.profiler.XStat stats = 6;</code>
-   */
-  org.tensorflow.proto.profiler.XStatOrBuilder getStatsOrBuilder(
-      int index);
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/XPlaneProtos.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/XPlaneProtos.java
deleted file mode 100644
index 4fdb2cce99d..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/XPlaneProtos.java
+++ /dev/null
@@ -1,167 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/profiler/protobuf/xplane.proto
-
-package org.tensorflow.proto.profiler;
-
-public final class XPlaneProtos {
-  private XPlaneProtos() {}
-  public static void registerAllExtensions(
-      com.google.protobuf.ExtensionRegistryLite registry) {
-  }
-
-  public static void registerAllExtensions(
-      com.google.protobuf.ExtensionRegistry registry) {
-    registerAllExtensions(
-        (com.google.protobuf.ExtensionRegistryLite) registry);
-  }
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_profiler_XSpace_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_profiler_XSpace_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_profiler_XPlane_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_profiler_XPlane_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_profiler_XPlane_EventMetadataEntry_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_profiler_XPlane_EventMetadataEntry_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_profiler_XPlane_StatMetadataEntry_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_profiler_XPlane_StatMetadataEntry_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_profiler_XLine_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_profiler_XLine_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_profiler_XEvent_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_profiler_XEvent_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_profiler_XStat_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_profiler_XStat_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_profiler_XEventMetadata_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_profiler_XEventMetadata_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_profiler_XStatMetadata_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_profiler_XStatMetadata_fieldAccessorTable;
-
-  public static com.google.protobuf.Descriptors.FileDescriptor
-      getDescriptor() {
-    return descriptor;
-  }
-  private static  com.google.protobuf.Descriptors.FileDescriptor
-      descriptor;
-  static {
-    java.lang.String[] descriptorData = {
-      "\n.tensorflow/core/profiler/protobuf/xpla" +
-      "ne.proto\022\023tensorflow.profiler\"W\n\006XSpace\022" +
-      "+\n\006planes\030\001 \003(\0132\033.tensorflow.profiler.XP" +
-      "lane\022\016\n\006errors\030\002 \003(\t\022\020\n\010warnings\030\003 \003(\t\"\272" +
-      "\003\n\006XPlane\022\n\n\002id\030\001 \001(\003\022\014\n\004name\030\002 \001(\t\022)\n\005l" +
-      "ines\030\003 \003(\0132\032.tensorflow.profiler.XLine\022F" +
-      "\n\016event_metadata\030\004 \003(\0132..tensorflow.prof" +
-      "iler.XPlane.EventMetadataEntry\022D\n\rstat_m" +
-      "etadata\030\005 \003(\0132-.tensorflow.profiler.XPla" +
-      "ne.StatMetadataEntry\022)\n\005stats\030\006 \003(\0132\032.te" +
-      "nsorflow.profiler.XStat\032Y\n\022EventMetadata" +
-      "Entry\022\013\n\003key\030\001 \001(\003\0222\n\005value\030\002 \001(\0132#.tens" +
-      "orflow.profiler.XEventMetadata:\0028\001\032W\n\021St" +
-      "atMetadataEntry\022\013\n\003key\030\001 \001(\003\0221\n\005value\030\002 " +
-      "\001(\0132\".tensorflow.profiler.XStatMetadata:" +
-      "\0028\001\"\273\001\n\005XLine\022\n\n\002id\030\001 \001(\003\022\022\n\ndisplay_id\030" +
-      "\n \001(\003\022\014\n\004name\030\002 \001(\t\022\024\n\014display_name\030\013 \001(" +
-      "\t\022\024\n\014timestamp_ns\030\003 \001(\003\022\023\n\013duration_ps\030\t" +
-      " \001(\003\022+\n\006events\030\004 \003(\0132\033.tensorflow.profil" +
-      "er.XEventJ\004\010\005\020\006J\004\010\006\020\007J\004\010\007\020\010J\004\010\010\020\t\"\225\001\n\006XE" +
-      "vent\022\023\n\013metadata_id\030\001 \001(\003\022\023\n\toffset_ps\030\002" +
-      " \001(\003H\000\022\031\n\017num_occurrences\030\005 \001(\003H\000\022\023\n\013dur" +
-      "ation_ps\030\003 \001(\003\022)\n\005stats\030\004 \003(\0132\032.tensorfl" +
-      "ow.profiler.XStatB\006\n\004data\"\255\001\n\005XStat\022\023\n\013m" +
-      "etadata_id\030\001 \001(\003\022\026\n\014double_value\030\002 \001(\001H\000" +
-      "\022\026\n\014uint64_value\030\003 \001(\004H\000\022\025\n\013int64_value\030" +
-      "\004 \001(\003H\000\022\023\n\tstr_value\030\005 \001(\tH\000\022\025\n\013bytes_va" +
-      "lue\030\006 \001(\014H\000\022\023\n\tref_value\030\007 \001(\004H\000B\007\n\005valu" +
-      "e\"R\n\016XEventMetadata\022\n\n\002id\030\001 \001(\003\022\014\n\004name\030" +
-      "\002 \001(\t\022\024\n\014display_name\030\004 \001(\t\022\020\n\010metadata\030" +
-      "\003 \001(\014\">\n\rXStatMetadata\022\n\n\002id\030\001 \001(\003\022\014\n\004na" +
-      "me\030\002 \001(\t\022\023\n\013description\030\003 \001(\tB2\n\035org.ten" +
-      "sorflow.proto.profilerB\014XPlaneProtosP\001\370\001" +
-      "\001b\006proto3"
-    };
-    descriptor = com.google.protobuf.Descriptors.FileDescriptor
-      .internalBuildGeneratedFileFrom(descriptorData,
-        new com.google.protobuf.Descriptors.FileDescriptor[] {
-        });
-    internal_static_tensorflow_profiler_XSpace_descriptor =
-      getDescriptor().getMessageTypes().get(0);
-    internal_static_tensorflow_profiler_XSpace_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_profiler_XSpace_descriptor,
-        new java.lang.String[] { "Planes", "Errors", "Warnings", });
-    internal_static_tensorflow_profiler_XPlane_descriptor =
-      getDescriptor().getMessageTypes().get(1);
-    internal_static_tensorflow_profiler_XPlane_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_profiler_XPlane_descriptor,
-        new java.lang.String[] { "Id", "Name", "Lines", "EventMetadata", "StatMetadata", "Stats", });
-    internal_static_tensorflow_profiler_XPlane_EventMetadataEntry_descriptor =
-      internal_static_tensorflow_profiler_XPlane_descriptor.getNestedTypes().get(0);
-    internal_static_tensorflow_profiler_XPlane_EventMetadataEntry_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_profiler_XPlane_EventMetadataEntry_descriptor,
-        new java.lang.String[] { "Key", "Value", });
-    internal_static_tensorflow_profiler_XPlane_StatMetadataEntry_descriptor =
-      internal_static_tensorflow_profiler_XPlane_descriptor.getNestedTypes().get(1);
-    internal_static_tensorflow_profiler_XPlane_StatMetadataEntry_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_profiler_XPlane_StatMetadataEntry_descriptor,
-        new java.lang.String[] { "Key", "Value", });
-    internal_static_tensorflow_profiler_XLine_descriptor =
-      getDescriptor().getMessageTypes().get(2);
-    internal_static_tensorflow_profiler_XLine_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_profiler_XLine_descriptor,
-        new java.lang.String[] { "Id", "DisplayId", "Name", "DisplayName", "TimestampNs", "DurationPs", "Events", });
-    internal_static_tensorflow_profiler_XEvent_descriptor =
-      getDescriptor().getMessageTypes().get(3);
-    internal_static_tensorflow_profiler_XEvent_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_profiler_XEvent_descriptor,
-        new java.lang.String[] { "MetadataId", "OffsetPs", "NumOccurrences", "DurationPs", "Stats", "Data", });
-    internal_static_tensorflow_profiler_XStat_descriptor =
-      getDescriptor().getMessageTypes().get(4);
-    internal_static_tensorflow_profiler_XStat_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_profiler_XStat_descriptor,
-        new java.lang.String[] { "MetadataId", "DoubleValue", "Uint64Value", "Int64Value", "StrValue", "BytesValue", "RefValue", "Value", });
-    internal_static_tensorflow_profiler_XEventMetadata_descriptor =
-      getDescriptor().getMessageTypes().get(5);
-    internal_static_tensorflow_profiler_XEventMetadata_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_profiler_XEventMetadata_descriptor,
-        new java.lang.String[] { "Id", "Name", "DisplayName", "Metadata", });
-    internal_static_tensorflow_profiler_XStatMetadata_descriptor =
-      getDescriptor().getMessageTypes().get(6);
-    internal_static_tensorflow_profiler_XStatMetadata_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_profiler_XStatMetadata_descriptor,
-        new java.lang.String[] { "Id", "Name", "Description", });
-  }
-
-  // @@protoc_insertion_point(outer_class_scope)
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/XSpace.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/XSpace.java
deleted file mode 100644
index 956b5ab793b..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/XSpace.java
+++ /dev/null
@@ -1,1219 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/profiler/protobuf/xplane.proto
-
-package org.tensorflow.proto.profiler;
-
-/**
- * <pre>
- * A container of parallel XPlanes, generated by one or more profiling sources.
- * Next ID: 4
- * </pre>
- *
- * Protobuf type {@code tensorflow.profiler.XSpace}
- */
-public  final class XSpace extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.profiler.XSpace)
-    XSpaceOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use XSpace.newBuilder() to construct.
-  private XSpace(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private XSpace() {
-    planes_ = java.util.Collections.emptyList();
-    errors_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-    warnings_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new XSpace();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private XSpace(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    int mutable_bitField0_ = 0;
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 10: {
-            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
-              planes_ = new java.util.ArrayList<org.tensorflow.proto.profiler.XPlane>();
-              mutable_bitField0_ |= 0x00000001;
-            }
-            planes_.add(
-                input.readMessage(org.tensorflow.proto.profiler.XPlane.parser(), extensionRegistry));
-            break;
-          }
-          case 18: {
-            java.lang.String s = input.readStringRequireUtf8();
-            if (!((mutable_bitField0_ & 0x00000002) != 0)) {
-              errors_ = new com.google.protobuf.LazyStringArrayList();
-              mutable_bitField0_ |= 0x00000002;
-            }
-            errors_.add(s);
-            break;
-          }
-          case 26: {
-            java.lang.String s = input.readStringRequireUtf8();
-            if (!((mutable_bitField0_ & 0x00000004) != 0)) {
-              warnings_ = new com.google.protobuf.LazyStringArrayList();
-              mutable_bitField0_ |= 0x00000004;
-            }
-            warnings_.add(s);
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      if (((mutable_bitField0_ & 0x00000001) != 0)) {
-        planes_ = java.util.Collections.unmodifiableList(planes_);
-      }
-      if (((mutable_bitField0_ & 0x00000002) != 0)) {
-        errors_ = errors_.getUnmodifiableView();
-      }
-      if (((mutable_bitField0_ & 0x00000004) != 0)) {
-        warnings_ = warnings_.getUnmodifiableView();
-      }
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.profiler.XPlaneProtos.internal_static_tensorflow_profiler_XSpace_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.profiler.XPlaneProtos.internal_static_tensorflow_profiler_XSpace_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.profiler.XSpace.class, org.tensorflow.proto.profiler.XSpace.Builder.class);
-  }
-
-  public static final int PLANES_FIELD_NUMBER = 1;
-  private java.util.List<org.tensorflow.proto.profiler.XPlane> planes_;
-  /**
-   * <code>repeated .tensorflow.profiler.XPlane planes = 1;</code>
-   */
-  public java.util.List<org.tensorflow.proto.profiler.XPlane> getPlanesList() {
-    return planes_;
-  }
-  /**
-   * <code>repeated .tensorflow.profiler.XPlane planes = 1;</code>
-   */
-  public java.util.List<? extends org.tensorflow.proto.profiler.XPlaneOrBuilder> 
-      getPlanesOrBuilderList() {
-    return planes_;
-  }
-  /**
-   * <code>repeated .tensorflow.profiler.XPlane planes = 1;</code>
-   */
-  public int getPlanesCount() {
-    return planes_.size();
-  }
-  /**
-   * <code>repeated .tensorflow.profiler.XPlane planes = 1;</code>
-   */
-  public org.tensorflow.proto.profiler.XPlane getPlanes(int index) {
-    return planes_.get(index);
-  }
-  /**
-   * <code>repeated .tensorflow.profiler.XPlane planes = 1;</code>
-   */
-  public org.tensorflow.proto.profiler.XPlaneOrBuilder getPlanesOrBuilder(
-      int index) {
-    return planes_.get(index);
-  }
-
-  public static final int ERRORS_FIELD_NUMBER = 2;
-  private com.google.protobuf.LazyStringList errors_;
-  /**
-   * <pre>
-   * Errors (if any) in the generation of planes.
-   * </pre>
-   *
-   * <code>repeated string errors = 2;</code>
-   */
-  public com.google.protobuf.ProtocolStringList
-      getErrorsList() {
-    return errors_;
-  }
-  /**
-   * <pre>
-   * Errors (if any) in the generation of planes.
-   * </pre>
-   *
-   * <code>repeated string errors = 2;</code>
-   */
-  public int getErrorsCount() {
-    return errors_.size();
-  }
-  /**
-   * <pre>
-   * Errors (if any) in the generation of planes.
-   * </pre>
-   *
-   * <code>repeated string errors = 2;</code>
-   */
-  public java.lang.String getErrors(int index) {
-    return errors_.get(index);
-  }
-  /**
-   * <pre>
-   * Errors (if any) in the generation of planes.
-   * </pre>
-   *
-   * <code>repeated string errors = 2;</code>
-   */
-  public com.google.protobuf.ByteString
-      getErrorsBytes(int index) {
-    return errors_.getByteString(index);
-  }
-
-  public static final int WARNINGS_FIELD_NUMBER = 3;
-  private com.google.protobuf.LazyStringList warnings_;
-  /**
-   * <pre>
-   * Warnings (if any) in the generation of planes;
-   * </pre>
-   *
-   * <code>repeated string warnings = 3;</code>
-   */
-  public com.google.protobuf.ProtocolStringList
-      getWarningsList() {
-    return warnings_;
-  }
-  /**
-   * <pre>
-   * Warnings (if any) in the generation of planes;
-   * </pre>
-   *
-   * <code>repeated string warnings = 3;</code>
-   */
-  public int getWarningsCount() {
-    return warnings_.size();
-  }
-  /**
-   * <pre>
-   * Warnings (if any) in the generation of planes;
-   * </pre>
-   *
-   * <code>repeated string warnings = 3;</code>
-   */
-  public java.lang.String getWarnings(int index) {
-    return warnings_.get(index);
-  }
-  /**
-   * <pre>
-   * Warnings (if any) in the generation of planes;
-   * </pre>
-   *
-   * <code>repeated string warnings = 3;</code>
-   */
-  public com.google.protobuf.ByteString
-      getWarningsBytes(int index) {
-    return warnings_.getByteString(index);
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    for (int i = 0; i < planes_.size(); i++) {
-      output.writeMessage(1, planes_.get(i));
-    }
-    for (int i = 0; i < errors_.size(); i++) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 2, errors_.getRaw(i));
-    }
-    for (int i = 0; i < warnings_.size(); i++) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 3, warnings_.getRaw(i));
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    for (int i = 0; i < planes_.size(); i++) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(1, planes_.get(i));
-    }
-    {
-      int dataSize = 0;
-      for (int i = 0; i < errors_.size(); i++) {
-        dataSize += computeStringSizeNoTag(errors_.getRaw(i));
-      }
-      size += dataSize;
-      size += 1 * getErrorsList().size();
-    }
-    {
-      int dataSize = 0;
-      for (int i = 0; i < warnings_.size(); i++) {
-        dataSize += computeStringSizeNoTag(warnings_.getRaw(i));
-      }
-      size += dataSize;
-      size += 1 * getWarningsList().size();
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.profiler.XSpace)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.profiler.XSpace other = (org.tensorflow.proto.profiler.XSpace) obj;
-
-    if (!getPlanesList()
-        .equals(other.getPlanesList())) return false;
-    if (!getErrorsList()
-        .equals(other.getErrorsList())) return false;
-    if (!getWarningsList()
-        .equals(other.getWarningsList())) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    if (getPlanesCount() > 0) {
-      hash = (37 * hash) + PLANES_FIELD_NUMBER;
-      hash = (53 * hash) + getPlanesList().hashCode();
-    }
-    if (getErrorsCount() > 0) {
-      hash = (37 * hash) + ERRORS_FIELD_NUMBER;
-      hash = (53 * hash) + getErrorsList().hashCode();
-    }
-    if (getWarningsCount() > 0) {
-      hash = (37 * hash) + WARNINGS_FIELD_NUMBER;
-      hash = (53 * hash) + getWarningsList().hashCode();
-    }
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.profiler.XSpace parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.profiler.XSpace parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.profiler.XSpace parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.profiler.XSpace parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.profiler.XSpace parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.profiler.XSpace parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.profiler.XSpace parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.profiler.XSpace parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.profiler.XSpace parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.profiler.XSpace parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.profiler.XSpace parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.profiler.XSpace parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.profiler.XSpace prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * <pre>
-   * A container of parallel XPlanes, generated by one or more profiling sources.
-   * Next ID: 4
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.profiler.XSpace}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.profiler.XSpace)
-      org.tensorflow.proto.profiler.XSpaceOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.profiler.XPlaneProtos.internal_static_tensorflow_profiler_XSpace_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.profiler.XPlaneProtos.internal_static_tensorflow_profiler_XSpace_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.profiler.XSpace.class, org.tensorflow.proto.profiler.XSpace.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.profiler.XSpace.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-        getPlanesFieldBuilder();
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      if (planesBuilder_ == null) {
-        planes_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000001);
-      } else {
-        planesBuilder_.clear();
-      }
-      errors_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-      bitField0_ = (bitField0_ & ~0x00000002);
-      warnings_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-      bitField0_ = (bitField0_ & ~0x00000004);
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.profiler.XPlaneProtos.internal_static_tensorflow_profiler_XSpace_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.profiler.XSpace getDefaultInstanceForType() {
-      return org.tensorflow.proto.profiler.XSpace.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.profiler.XSpace build() {
-      org.tensorflow.proto.profiler.XSpace result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.profiler.XSpace buildPartial() {
-      org.tensorflow.proto.profiler.XSpace result = new org.tensorflow.proto.profiler.XSpace(this);
-      int from_bitField0_ = bitField0_;
-      if (planesBuilder_ == null) {
-        if (((bitField0_ & 0x00000001) != 0)) {
-          planes_ = java.util.Collections.unmodifiableList(planes_);
-          bitField0_ = (bitField0_ & ~0x00000001);
-        }
-        result.planes_ = planes_;
-      } else {
-        result.planes_ = planesBuilder_.build();
-      }
-      if (((bitField0_ & 0x00000002) != 0)) {
-        errors_ = errors_.getUnmodifiableView();
-        bitField0_ = (bitField0_ & ~0x00000002);
-      }
-      result.errors_ = errors_;
-      if (((bitField0_ & 0x00000004) != 0)) {
-        warnings_ = warnings_.getUnmodifiableView();
-        bitField0_ = (bitField0_ & ~0x00000004);
-      }
-      result.warnings_ = warnings_;
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.profiler.XSpace) {
-        return mergeFrom((org.tensorflow.proto.profiler.XSpace)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.profiler.XSpace other) {
-      if (other == org.tensorflow.proto.profiler.XSpace.getDefaultInstance()) return this;
-      if (planesBuilder_ == null) {
-        if (!other.planes_.isEmpty()) {
-          if (planes_.isEmpty()) {
-            planes_ = other.planes_;
-            bitField0_ = (bitField0_ & ~0x00000001);
-          } else {
-            ensurePlanesIsMutable();
-            planes_.addAll(other.planes_);
-          }
-          onChanged();
-        }
-      } else {
-        if (!other.planes_.isEmpty()) {
-          if (planesBuilder_.isEmpty()) {
-            planesBuilder_.dispose();
-            planesBuilder_ = null;
-            planes_ = other.planes_;
-            bitField0_ = (bitField0_ & ~0x00000001);
-            planesBuilder_ = 
-              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
-                 getPlanesFieldBuilder() : null;
-          } else {
-            planesBuilder_.addAllMessages(other.planes_);
-          }
-        }
-      }
-      if (!other.errors_.isEmpty()) {
-        if (errors_.isEmpty()) {
-          errors_ = other.errors_;
-          bitField0_ = (bitField0_ & ~0x00000002);
-        } else {
-          ensureErrorsIsMutable();
-          errors_.addAll(other.errors_);
-        }
-        onChanged();
-      }
-      if (!other.warnings_.isEmpty()) {
-        if (warnings_.isEmpty()) {
-          warnings_ = other.warnings_;
-          bitField0_ = (bitField0_ & ~0x00000004);
-        } else {
-          ensureWarningsIsMutable();
-          warnings_.addAll(other.warnings_);
-        }
-        onChanged();
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.profiler.XSpace parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.profiler.XSpace) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-    private int bitField0_;
-
-    private java.util.List<org.tensorflow.proto.profiler.XPlane> planes_ =
-      java.util.Collections.emptyList();
-    private void ensurePlanesIsMutable() {
-      if (!((bitField0_ & 0x00000001) != 0)) {
-        planes_ = new java.util.ArrayList<org.tensorflow.proto.profiler.XPlane>(planes_);
-        bitField0_ |= 0x00000001;
-       }
-    }
-
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.profiler.XPlane, org.tensorflow.proto.profiler.XPlane.Builder, org.tensorflow.proto.profiler.XPlaneOrBuilder> planesBuilder_;
-
-    /**
-     * <code>repeated .tensorflow.profiler.XPlane planes = 1;</code>
-     */
-    public java.util.List<org.tensorflow.proto.profiler.XPlane> getPlanesList() {
-      if (planesBuilder_ == null) {
-        return java.util.Collections.unmodifiableList(planes_);
-      } else {
-        return planesBuilder_.getMessageList();
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.profiler.XPlane planes = 1;</code>
-     */
-    public int getPlanesCount() {
-      if (planesBuilder_ == null) {
-        return planes_.size();
-      } else {
-        return planesBuilder_.getCount();
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.profiler.XPlane planes = 1;</code>
-     */
-    public org.tensorflow.proto.profiler.XPlane getPlanes(int index) {
-      if (planesBuilder_ == null) {
-        return planes_.get(index);
-      } else {
-        return planesBuilder_.getMessage(index);
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.profiler.XPlane planes = 1;</code>
-     */
-    public Builder setPlanes(
-        int index, org.tensorflow.proto.profiler.XPlane value) {
-      if (planesBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensurePlanesIsMutable();
-        planes_.set(index, value);
-        onChanged();
-      } else {
-        planesBuilder_.setMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.profiler.XPlane planes = 1;</code>
-     */
-    public Builder setPlanes(
-        int index, org.tensorflow.proto.profiler.XPlane.Builder builderForValue) {
-      if (planesBuilder_ == null) {
-        ensurePlanesIsMutable();
-        planes_.set(index, builderForValue.build());
-        onChanged();
-      } else {
-        planesBuilder_.setMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.profiler.XPlane planes = 1;</code>
-     */
-    public Builder addPlanes(org.tensorflow.proto.profiler.XPlane value) {
-      if (planesBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensurePlanesIsMutable();
-        planes_.add(value);
-        onChanged();
-      } else {
-        planesBuilder_.addMessage(value);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.profiler.XPlane planes = 1;</code>
-     */
-    public Builder addPlanes(
-        int index, org.tensorflow.proto.profiler.XPlane value) {
-      if (planesBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensurePlanesIsMutable();
-        planes_.add(index, value);
-        onChanged();
-      } else {
-        planesBuilder_.addMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.profiler.XPlane planes = 1;</code>
-     */
-    public Builder addPlanes(
-        org.tensorflow.proto.profiler.XPlane.Builder builderForValue) {
-      if (planesBuilder_ == null) {
-        ensurePlanesIsMutable();
-        planes_.add(builderForValue.build());
-        onChanged();
-      } else {
-        planesBuilder_.addMessage(builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.profiler.XPlane planes = 1;</code>
-     */
-    public Builder addPlanes(
-        int index, org.tensorflow.proto.profiler.XPlane.Builder builderForValue) {
-      if (planesBuilder_ == null) {
-        ensurePlanesIsMutable();
-        planes_.add(index, builderForValue.build());
-        onChanged();
-      } else {
-        planesBuilder_.addMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.profiler.XPlane planes = 1;</code>
-     */
-    public Builder addAllPlanes(
-        java.lang.Iterable<? extends org.tensorflow.proto.profiler.XPlane> values) {
-      if (planesBuilder_ == null) {
-        ensurePlanesIsMutable();
-        com.google.protobuf.AbstractMessageLite.Builder.addAll(
-            values, planes_);
-        onChanged();
-      } else {
-        planesBuilder_.addAllMessages(values);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.profiler.XPlane planes = 1;</code>
-     */
-    public Builder clearPlanes() {
-      if (planesBuilder_ == null) {
-        planes_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000001);
-        onChanged();
-      } else {
-        planesBuilder_.clear();
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.profiler.XPlane planes = 1;</code>
-     */
-    public Builder removePlanes(int index) {
-      if (planesBuilder_ == null) {
-        ensurePlanesIsMutable();
-        planes_.remove(index);
-        onChanged();
-      } else {
-        planesBuilder_.remove(index);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.profiler.XPlane planes = 1;</code>
-     */
-    public org.tensorflow.proto.profiler.XPlane.Builder getPlanesBuilder(
-        int index) {
-      return getPlanesFieldBuilder().getBuilder(index);
-    }
-    /**
-     * <code>repeated .tensorflow.profiler.XPlane planes = 1;</code>
-     */
-    public org.tensorflow.proto.profiler.XPlaneOrBuilder getPlanesOrBuilder(
-        int index) {
-      if (planesBuilder_ == null) {
-        return planes_.get(index);  } else {
-        return planesBuilder_.getMessageOrBuilder(index);
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.profiler.XPlane planes = 1;</code>
-     */
-    public java.util.List<? extends org.tensorflow.proto.profiler.XPlaneOrBuilder> 
-         getPlanesOrBuilderList() {
-      if (planesBuilder_ != null) {
-        return planesBuilder_.getMessageOrBuilderList();
-      } else {
-        return java.util.Collections.unmodifiableList(planes_);
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.profiler.XPlane planes = 1;</code>
-     */
-    public org.tensorflow.proto.profiler.XPlane.Builder addPlanesBuilder() {
-      return getPlanesFieldBuilder().addBuilder(
-          org.tensorflow.proto.profiler.XPlane.getDefaultInstance());
-    }
-    /**
-     * <code>repeated .tensorflow.profiler.XPlane planes = 1;</code>
-     */
-    public org.tensorflow.proto.profiler.XPlane.Builder addPlanesBuilder(
-        int index) {
-      return getPlanesFieldBuilder().addBuilder(
-          index, org.tensorflow.proto.profiler.XPlane.getDefaultInstance());
-    }
-    /**
-     * <code>repeated .tensorflow.profiler.XPlane planes = 1;</code>
-     */
-    public java.util.List<org.tensorflow.proto.profiler.XPlane.Builder> 
-         getPlanesBuilderList() {
-      return getPlanesFieldBuilder().getBuilderList();
-    }
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.profiler.XPlane, org.tensorflow.proto.profiler.XPlane.Builder, org.tensorflow.proto.profiler.XPlaneOrBuilder> 
-        getPlanesFieldBuilder() {
-      if (planesBuilder_ == null) {
-        planesBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
-            org.tensorflow.proto.profiler.XPlane, org.tensorflow.proto.profiler.XPlane.Builder, org.tensorflow.proto.profiler.XPlaneOrBuilder>(
-                planes_,
-                ((bitField0_ & 0x00000001) != 0),
-                getParentForChildren(),
-                isClean());
-        planes_ = null;
-      }
-      return planesBuilder_;
-    }
-
-    private com.google.protobuf.LazyStringList errors_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-    private void ensureErrorsIsMutable() {
-      if (!((bitField0_ & 0x00000002) != 0)) {
-        errors_ = new com.google.protobuf.LazyStringArrayList(errors_);
-        bitField0_ |= 0x00000002;
-       }
-    }
-    /**
-     * <pre>
-     * Errors (if any) in the generation of planes.
-     * </pre>
-     *
-     * <code>repeated string errors = 2;</code>
-     */
-    public com.google.protobuf.ProtocolStringList
-        getErrorsList() {
-      return errors_.getUnmodifiableView();
-    }
-    /**
-     * <pre>
-     * Errors (if any) in the generation of planes.
-     * </pre>
-     *
-     * <code>repeated string errors = 2;</code>
-     */
-    public int getErrorsCount() {
-      return errors_.size();
-    }
-    /**
-     * <pre>
-     * Errors (if any) in the generation of planes.
-     * </pre>
-     *
-     * <code>repeated string errors = 2;</code>
-     */
-    public java.lang.String getErrors(int index) {
-      return errors_.get(index);
-    }
-    /**
-     * <pre>
-     * Errors (if any) in the generation of planes.
-     * </pre>
-     *
-     * <code>repeated string errors = 2;</code>
-     */
-    public com.google.protobuf.ByteString
-        getErrorsBytes(int index) {
-      return errors_.getByteString(index);
-    }
-    /**
-     * <pre>
-     * Errors (if any) in the generation of planes.
-     * </pre>
-     *
-     * <code>repeated string errors = 2;</code>
-     */
-    public Builder setErrors(
-        int index, java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  ensureErrorsIsMutable();
-      errors_.set(index, value);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Errors (if any) in the generation of planes.
-     * </pre>
-     *
-     * <code>repeated string errors = 2;</code>
-     */
-    public Builder addErrors(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  ensureErrorsIsMutable();
-      errors_.add(value);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Errors (if any) in the generation of planes.
-     * </pre>
-     *
-     * <code>repeated string errors = 2;</code>
-     */
-    public Builder addAllErrors(
-        java.lang.Iterable<java.lang.String> values) {
-      ensureErrorsIsMutable();
-      com.google.protobuf.AbstractMessageLite.Builder.addAll(
-          values, errors_);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Errors (if any) in the generation of planes.
-     * </pre>
-     *
-     * <code>repeated string errors = 2;</code>
-     */
-    public Builder clearErrors() {
-      errors_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-      bitField0_ = (bitField0_ & ~0x00000002);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Errors (if any) in the generation of planes.
-     * </pre>
-     *
-     * <code>repeated string errors = 2;</code>
-     */
-    public Builder addErrorsBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      ensureErrorsIsMutable();
-      errors_.add(value);
-      onChanged();
-      return this;
-    }
-
-    private com.google.protobuf.LazyStringList warnings_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-    private void ensureWarningsIsMutable() {
-      if (!((bitField0_ & 0x00000004) != 0)) {
-        warnings_ = new com.google.protobuf.LazyStringArrayList(warnings_);
-        bitField0_ |= 0x00000004;
-       }
-    }
-    /**
-     * <pre>
-     * Warnings (if any) in the generation of planes;
-     * </pre>
-     *
-     * <code>repeated string warnings = 3;</code>
-     */
-    public com.google.protobuf.ProtocolStringList
-        getWarningsList() {
-      return warnings_.getUnmodifiableView();
-    }
-    /**
-     * <pre>
-     * Warnings (if any) in the generation of planes;
-     * </pre>
-     *
-     * <code>repeated string warnings = 3;</code>
-     */
-    public int getWarningsCount() {
-      return warnings_.size();
-    }
-    /**
-     * <pre>
-     * Warnings (if any) in the generation of planes;
-     * </pre>
-     *
-     * <code>repeated string warnings = 3;</code>
-     */
-    public java.lang.String getWarnings(int index) {
-      return warnings_.get(index);
-    }
-    /**
-     * <pre>
-     * Warnings (if any) in the generation of planes;
-     * </pre>
-     *
-     * <code>repeated string warnings = 3;</code>
-     */
-    public com.google.protobuf.ByteString
-        getWarningsBytes(int index) {
-      return warnings_.getByteString(index);
-    }
-    /**
-     * <pre>
-     * Warnings (if any) in the generation of planes;
-     * </pre>
-     *
-     * <code>repeated string warnings = 3;</code>
-     */
-    public Builder setWarnings(
-        int index, java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  ensureWarningsIsMutable();
-      warnings_.set(index, value);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Warnings (if any) in the generation of planes;
-     * </pre>
-     *
-     * <code>repeated string warnings = 3;</code>
-     */
-    public Builder addWarnings(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  ensureWarningsIsMutable();
-      warnings_.add(value);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Warnings (if any) in the generation of planes;
-     * </pre>
-     *
-     * <code>repeated string warnings = 3;</code>
-     */
-    public Builder addAllWarnings(
-        java.lang.Iterable<java.lang.String> values) {
-      ensureWarningsIsMutable();
-      com.google.protobuf.AbstractMessageLite.Builder.addAll(
-          values, warnings_);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Warnings (if any) in the generation of planes;
-     * </pre>
-     *
-     * <code>repeated string warnings = 3;</code>
-     */
-    public Builder clearWarnings() {
-      warnings_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-      bitField0_ = (bitField0_ & ~0x00000004);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Warnings (if any) in the generation of planes;
-     * </pre>
-     *
-     * <code>repeated string warnings = 3;</code>
-     */
-    public Builder addWarningsBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      ensureWarningsIsMutable();
-      warnings_.add(value);
-      onChanged();
-      return this;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.profiler.XSpace)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.profiler.XSpace)
-  private static final org.tensorflow.proto.profiler.XSpace DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.profiler.XSpace();
-  }
-
-  public static org.tensorflow.proto.profiler.XSpace getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<XSpace>
-      PARSER = new com.google.protobuf.AbstractParser<XSpace>() {
-    @java.lang.Override
-    public XSpace parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new XSpace(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<XSpace> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<XSpace> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.profiler.XSpace getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/XSpaceOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/XSpaceOrBuilder.java
deleted file mode 100644
index dd081ad954f..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/XSpaceOrBuilder.java
+++ /dev/null
@@ -1,103 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/profiler/protobuf/xplane.proto
-
-package org.tensorflow.proto.profiler;
-
-public interface XSpaceOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.profiler.XSpace)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <code>repeated .tensorflow.profiler.XPlane planes = 1;</code>
-   */
-  java.util.List<org.tensorflow.proto.profiler.XPlane> 
-      getPlanesList();
-  /**
-   * <code>repeated .tensorflow.profiler.XPlane planes = 1;</code>
-   */
-  org.tensorflow.proto.profiler.XPlane getPlanes(int index);
-  /**
-   * <code>repeated .tensorflow.profiler.XPlane planes = 1;</code>
-   */
-  int getPlanesCount();
-  /**
-   * <code>repeated .tensorflow.profiler.XPlane planes = 1;</code>
-   */
-  java.util.List<? extends org.tensorflow.proto.profiler.XPlaneOrBuilder> 
-      getPlanesOrBuilderList();
-  /**
-   * <code>repeated .tensorflow.profiler.XPlane planes = 1;</code>
-   */
-  org.tensorflow.proto.profiler.XPlaneOrBuilder getPlanesOrBuilder(
-      int index);
-
-  /**
-   * <pre>
-   * Errors (if any) in the generation of planes.
-   * </pre>
-   *
-   * <code>repeated string errors = 2;</code>
-   */
-  java.util.List<java.lang.String>
-      getErrorsList();
-  /**
-   * <pre>
-   * Errors (if any) in the generation of planes.
-   * </pre>
-   *
-   * <code>repeated string errors = 2;</code>
-   */
-  int getErrorsCount();
-  /**
-   * <pre>
-   * Errors (if any) in the generation of planes.
-   * </pre>
-   *
-   * <code>repeated string errors = 2;</code>
-   */
-  java.lang.String getErrors(int index);
-  /**
-   * <pre>
-   * Errors (if any) in the generation of planes.
-   * </pre>
-   *
-   * <code>repeated string errors = 2;</code>
-   */
-  com.google.protobuf.ByteString
-      getErrorsBytes(int index);
-
-  /**
-   * <pre>
-   * Warnings (if any) in the generation of planes;
-   * </pre>
-   *
-   * <code>repeated string warnings = 3;</code>
-   */
-  java.util.List<java.lang.String>
-      getWarningsList();
-  /**
-   * <pre>
-   * Warnings (if any) in the generation of planes;
-   * </pre>
-   *
-   * <code>repeated string warnings = 3;</code>
-   */
-  int getWarningsCount();
-  /**
-   * <pre>
-   * Warnings (if any) in the generation of planes;
-   * </pre>
-   *
-   * <code>repeated string warnings = 3;</code>
-   */
-  java.lang.String getWarnings(int index);
-  /**
-   * <pre>
-   * Warnings (if any) in the generation of planes;
-   * </pre>
-   *
-   * <code>repeated string warnings = 3;</code>
-   */
-  com.google.protobuf.ByteString
-      getWarningsBytes(int index);
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/XStat.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/XStat.java
deleted file mode 100644
index 3ab8c18a43a..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/XStat.java
+++ /dev/null
@@ -1,1111 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/profiler/protobuf/xplane.proto
-
-package org.tensorflow.proto.profiler;
-
-/**
- * <pre>
- * An XStat is a named value associated with an XEvent, e.g., a performance
- * counter value, a metric computed by a formula applied over nested XEvents
- * and XStats.
- * Next ID: 8
- * </pre>
- *
- * Protobuf type {@code tensorflow.profiler.XStat}
- */
-public  final class XStat extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.profiler.XStat)
-    XStatOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use XStat.newBuilder() to construct.
-  private XStat(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private XStat() {
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new XStat();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private XStat(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 8: {
-
-            metadataId_ = input.readInt64();
-            break;
-          }
-          case 17: {
-            valueCase_ = 2;
-            value_ = input.readDouble();
-            break;
-          }
-          case 24: {
-            valueCase_ = 3;
-            value_ = input.readUInt64();
-            break;
-          }
-          case 32: {
-            valueCase_ = 4;
-            value_ = input.readInt64();
-            break;
-          }
-          case 42: {
-            java.lang.String s = input.readStringRequireUtf8();
-            valueCase_ = 5;
-            value_ = s;
-            break;
-          }
-          case 50: {
-            valueCase_ = 6;
-            value_ = input.readBytes();
-            break;
-          }
-          case 56: {
-            valueCase_ = 7;
-            value_ = input.readUInt64();
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.profiler.XPlaneProtos.internal_static_tensorflow_profiler_XStat_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.profiler.XPlaneProtos.internal_static_tensorflow_profiler_XStat_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.profiler.XStat.class, org.tensorflow.proto.profiler.XStat.Builder.class);
-  }
-
-  private int valueCase_ = 0;
-  private java.lang.Object value_;
-  public enum ValueCase
-      implements com.google.protobuf.Internal.EnumLite {
-    DOUBLE_VALUE(2),
-    UINT64_VALUE(3),
-    INT64_VALUE(4),
-    STR_VALUE(5),
-    BYTES_VALUE(6),
-    REF_VALUE(7),
-    VALUE_NOT_SET(0);
-    private final int value;
-    private ValueCase(int value) {
-      this.value = value;
-    }
-    /**
-     * @deprecated Use {@link #forNumber(int)} instead.
-     */
-    @java.lang.Deprecated
-    public static ValueCase valueOf(int value) {
-      return forNumber(value);
-    }
-
-    public static ValueCase forNumber(int value) {
-      switch (value) {
-        case 2: return DOUBLE_VALUE;
-        case 3: return UINT64_VALUE;
-        case 4: return INT64_VALUE;
-        case 5: return STR_VALUE;
-        case 6: return BYTES_VALUE;
-        case 7: return REF_VALUE;
-        case 0: return VALUE_NOT_SET;
-        default: return null;
-      }
-    }
-    public int getNumber() {
-      return this.value;
-    }
-  };
-
-  public ValueCase
-  getValueCase() {
-    return ValueCase.forNumber(
-        valueCase_);
-  }
-
-  public static final int METADATA_ID_FIELD_NUMBER = 1;
-  private long metadataId_;
-  /**
-   * <pre>
-   * XStatMetadata.id of corresponding metadata.
-   * </pre>
-   *
-   * <code>int64 metadata_id = 1;</code>
-   */
-  public long getMetadataId() {
-    return metadataId_;
-  }
-
-  public static final int DOUBLE_VALUE_FIELD_NUMBER = 2;
-  /**
-   * <code>double double_value = 2;</code>
-   */
-  public double getDoubleValue() {
-    if (valueCase_ == 2) {
-      return (java.lang.Double) value_;
-    }
-    return 0D;
-  }
-
-  public static final int UINT64_VALUE_FIELD_NUMBER = 3;
-  /**
-   * <code>uint64 uint64_value = 3;</code>
-   */
-  public long getUint64Value() {
-    if (valueCase_ == 3) {
-      return (java.lang.Long) value_;
-    }
-    return 0L;
-  }
-
-  public static final int INT64_VALUE_FIELD_NUMBER = 4;
-  /**
-   * <code>int64 int64_value = 4;</code>
-   */
-  public long getInt64Value() {
-    if (valueCase_ == 4) {
-      return (java.lang.Long) value_;
-    }
-    return 0L;
-  }
-
-  public static final int STR_VALUE_FIELD_NUMBER = 5;
-  /**
-   * <code>string str_value = 5;</code>
-   */
-  public java.lang.String getStrValue() {
-    java.lang.Object ref = "";
-    if (valueCase_ == 5) {
-      ref = value_;
-    }
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      if (valueCase_ == 5) {
-        value_ = s;
-      }
-      return s;
-    }
-  }
-  /**
-   * <code>string str_value = 5;</code>
-   */
-  public com.google.protobuf.ByteString
-      getStrValueBytes() {
-    java.lang.Object ref = "";
-    if (valueCase_ == 5) {
-      ref = value_;
-    }
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      if (valueCase_ == 5) {
-        value_ = b;
-      }
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int BYTES_VALUE_FIELD_NUMBER = 6;
-  /**
-   * <code>bytes bytes_value = 6;</code>
-   */
-  public com.google.protobuf.ByteString getBytesValue() {
-    if (valueCase_ == 6) {
-      return (com.google.protobuf.ByteString) value_;
-    }
-    return com.google.protobuf.ByteString.EMPTY;
-  }
-
-  public static final int REF_VALUE_FIELD_NUMBER = 7;
-  /**
-   * <pre>
-   * A string value that stored in XStatMetadata::name.
-   * </pre>
-   *
-   * <code>uint64 ref_value = 7;</code>
-   */
-  public long getRefValue() {
-    if (valueCase_ == 7) {
-      return (java.lang.Long) value_;
-    }
-    return 0L;
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    if (metadataId_ != 0L) {
-      output.writeInt64(1, metadataId_);
-    }
-    if (valueCase_ == 2) {
-      output.writeDouble(
-          2, (double)((java.lang.Double) value_));
-    }
-    if (valueCase_ == 3) {
-      output.writeUInt64(
-          3, (long)((java.lang.Long) value_));
-    }
-    if (valueCase_ == 4) {
-      output.writeInt64(
-          4, (long)((java.lang.Long) value_));
-    }
-    if (valueCase_ == 5) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 5, value_);
-    }
-    if (valueCase_ == 6) {
-      output.writeBytes(
-          6, (com.google.protobuf.ByteString) value_);
-    }
-    if (valueCase_ == 7) {
-      output.writeUInt64(
-          7, (long)((java.lang.Long) value_));
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (metadataId_ != 0L) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt64Size(1, metadataId_);
-    }
-    if (valueCase_ == 2) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeDoubleSize(
-            2, (double)((java.lang.Double) value_));
-    }
-    if (valueCase_ == 3) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeUInt64Size(
-            3, (long)((java.lang.Long) value_));
-    }
-    if (valueCase_ == 4) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt64Size(
-            4, (long)((java.lang.Long) value_));
-    }
-    if (valueCase_ == 5) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(5, value_);
-    }
-    if (valueCase_ == 6) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeBytesSize(
-            6, (com.google.protobuf.ByteString) value_);
-    }
-    if (valueCase_ == 7) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeUInt64Size(
-            7, (long)((java.lang.Long) value_));
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.profiler.XStat)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.profiler.XStat other = (org.tensorflow.proto.profiler.XStat) obj;
-
-    if (getMetadataId()
-        != other.getMetadataId()) return false;
-    if (!getValueCase().equals(other.getValueCase())) return false;
-    switch (valueCase_) {
-      case 2:
-        if (java.lang.Double.doubleToLongBits(getDoubleValue())
-            != java.lang.Double.doubleToLongBits(
-                other.getDoubleValue())) return false;
-        break;
-      case 3:
-        if (getUint64Value()
-            != other.getUint64Value()) return false;
-        break;
-      case 4:
-        if (getInt64Value()
-            != other.getInt64Value()) return false;
-        break;
-      case 5:
-        if (!getStrValue()
-            .equals(other.getStrValue())) return false;
-        break;
-      case 6:
-        if (!getBytesValue()
-            .equals(other.getBytesValue())) return false;
-        break;
-      case 7:
-        if (getRefValue()
-            != other.getRefValue()) return false;
-        break;
-      case 0:
-      default:
-    }
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    hash = (37 * hash) + METADATA_ID_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-        getMetadataId());
-    switch (valueCase_) {
-      case 2:
-        hash = (37 * hash) + DOUBLE_VALUE_FIELD_NUMBER;
-        hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-            java.lang.Double.doubleToLongBits(getDoubleValue()));
-        break;
-      case 3:
-        hash = (37 * hash) + UINT64_VALUE_FIELD_NUMBER;
-        hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-            getUint64Value());
-        break;
-      case 4:
-        hash = (37 * hash) + INT64_VALUE_FIELD_NUMBER;
-        hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-            getInt64Value());
-        break;
-      case 5:
-        hash = (37 * hash) + STR_VALUE_FIELD_NUMBER;
-        hash = (53 * hash) + getStrValue().hashCode();
-        break;
-      case 6:
-        hash = (37 * hash) + BYTES_VALUE_FIELD_NUMBER;
-        hash = (53 * hash) + getBytesValue().hashCode();
-        break;
-      case 7:
-        hash = (37 * hash) + REF_VALUE_FIELD_NUMBER;
-        hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-            getRefValue());
-        break;
-      case 0:
-      default:
-    }
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.profiler.XStat parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.profiler.XStat parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.profiler.XStat parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.profiler.XStat parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.profiler.XStat parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.profiler.XStat parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.profiler.XStat parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.profiler.XStat parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.profiler.XStat parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.profiler.XStat parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.profiler.XStat parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.profiler.XStat parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.profiler.XStat prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * <pre>
-   * An XStat is a named value associated with an XEvent, e.g., a performance
-   * counter value, a metric computed by a formula applied over nested XEvents
-   * and XStats.
-   * Next ID: 8
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.profiler.XStat}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.profiler.XStat)
-      org.tensorflow.proto.profiler.XStatOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.profiler.XPlaneProtos.internal_static_tensorflow_profiler_XStat_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.profiler.XPlaneProtos.internal_static_tensorflow_profiler_XStat_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.profiler.XStat.class, org.tensorflow.proto.profiler.XStat.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.profiler.XStat.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      metadataId_ = 0L;
-
-      valueCase_ = 0;
-      value_ = null;
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.profiler.XPlaneProtos.internal_static_tensorflow_profiler_XStat_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.profiler.XStat getDefaultInstanceForType() {
-      return org.tensorflow.proto.profiler.XStat.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.profiler.XStat build() {
-      org.tensorflow.proto.profiler.XStat result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.profiler.XStat buildPartial() {
-      org.tensorflow.proto.profiler.XStat result = new org.tensorflow.proto.profiler.XStat(this);
-      result.metadataId_ = metadataId_;
-      if (valueCase_ == 2) {
-        result.value_ = value_;
-      }
-      if (valueCase_ == 3) {
-        result.value_ = value_;
-      }
-      if (valueCase_ == 4) {
-        result.value_ = value_;
-      }
-      if (valueCase_ == 5) {
-        result.value_ = value_;
-      }
-      if (valueCase_ == 6) {
-        result.value_ = value_;
-      }
-      if (valueCase_ == 7) {
-        result.value_ = value_;
-      }
-      result.valueCase_ = valueCase_;
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.profiler.XStat) {
-        return mergeFrom((org.tensorflow.proto.profiler.XStat)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.profiler.XStat other) {
-      if (other == org.tensorflow.proto.profiler.XStat.getDefaultInstance()) return this;
-      if (other.getMetadataId() != 0L) {
-        setMetadataId(other.getMetadataId());
-      }
-      switch (other.getValueCase()) {
-        case DOUBLE_VALUE: {
-          setDoubleValue(other.getDoubleValue());
-          break;
-        }
-        case UINT64_VALUE: {
-          setUint64Value(other.getUint64Value());
-          break;
-        }
-        case INT64_VALUE: {
-          setInt64Value(other.getInt64Value());
-          break;
-        }
-        case STR_VALUE: {
-          valueCase_ = 5;
-          value_ = other.value_;
-          onChanged();
-          break;
-        }
-        case BYTES_VALUE: {
-          setBytesValue(other.getBytesValue());
-          break;
-        }
-        case REF_VALUE: {
-          setRefValue(other.getRefValue());
-          break;
-        }
-        case VALUE_NOT_SET: {
-          break;
-        }
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.profiler.XStat parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.profiler.XStat) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-    private int valueCase_ = 0;
-    private java.lang.Object value_;
-    public ValueCase
-        getValueCase() {
-      return ValueCase.forNumber(
-          valueCase_);
-    }
-
-    public Builder clearValue() {
-      valueCase_ = 0;
-      value_ = null;
-      onChanged();
-      return this;
-    }
-
-
-    private long metadataId_ ;
-    /**
-     * <pre>
-     * XStatMetadata.id of corresponding metadata.
-     * </pre>
-     *
-     * <code>int64 metadata_id = 1;</code>
-     */
-    public long getMetadataId() {
-      return metadataId_;
-    }
-    /**
-     * <pre>
-     * XStatMetadata.id of corresponding metadata.
-     * </pre>
-     *
-     * <code>int64 metadata_id = 1;</code>
-     */
-    public Builder setMetadataId(long value) {
-      
-      metadataId_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * XStatMetadata.id of corresponding metadata.
-     * </pre>
-     *
-     * <code>int64 metadata_id = 1;</code>
-     */
-    public Builder clearMetadataId() {
-      
-      metadataId_ = 0L;
-      onChanged();
-      return this;
-    }
-
-    /**
-     * <code>double double_value = 2;</code>
-     */
-    public double getDoubleValue() {
-      if (valueCase_ == 2) {
-        return (java.lang.Double) value_;
-      }
-      return 0D;
-    }
-    /**
-     * <code>double double_value = 2;</code>
-     */
-    public Builder setDoubleValue(double value) {
-      valueCase_ = 2;
-      value_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>double double_value = 2;</code>
-     */
-    public Builder clearDoubleValue() {
-      if (valueCase_ == 2) {
-        valueCase_ = 0;
-        value_ = null;
-        onChanged();
-      }
-      return this;
-    }
-
-    /**
-     * <code>uint64 uint64_value = 3;</code>
-     */
-    public long getUint64Value() {
-      if (valueCase_ == 3) {
-        return (java.lang.Long) value_;
-      }
-      return 0L;
-    }
-    /**
-     * <code>uint64 uint64_value = 3;</code>
-     */
-    public Builder setUint64Value(long value) {
-      valueCase_ = 3;
-      value_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>uint64 uint64_value = 3;</code>
-     */
-    public Builder clearUint64Value() {
-      if (valueCase_ == 3) {
-        valueCase_ = 0;
-        value_ = null;
-        onChanged();
-      }
-      return this;
-    }
-
-    /**
-     * <code>int64 int64_value = 4;</code>
-     */
-    public long getInt64Value() {
-      if (valueCase_ == 4) {
-        return (java.lang.Long) value_;
-      }
-      return 0L;
-    }
-    /**
-     * <code>int64 int64_value = 4;</code>
-     */
-    public Builder setInt64Value(long value) {
-      valueCase_ = 4;
-      value_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>int64 int64_value = 4;</code>
-     */
-    public Builder clearInt64Value() {
-      if (valueCase_ == 4) {
-        valueCase_ = 0;
-        value_ = null;
-        onChanged();
-      }
-      return this;
-    }
-
-    /**
-     * <code>string str_value = 5;</code>
-     */
-    public java.lang.String getStrValue() {
-      java.lang.Object ref = "";
-      if (valueCase_ == 5) {
-        ref = value_;
-      }
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        if (valueCase_ == 5) {
-          value_ = s;
-        }
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <code>string str_value = 5;</code>
-     */
-    public com.google.protobuf.ByteString
-        getStrValueBytes() {
-      java.lang.Object ref = "";
-      if (valueCase_ == 5) {
-        ref = value_;
-      }
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        if (valueCase_ == 5) {
-          value_ = b;
-        }
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <code>string str_value = 5;</code>
-     */
-    public Builder setStrValue(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  valueCase_ = 5;
-      value_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>string str_value = 5;</code>
-     */
-    public Builder clearStrValue() {
-      if (valueCase_ == 5) {
-        valueCase_ = 0;
-        value_ = null;
-        onChanged();
-      }
-      return this;
-    }
-    /**
-     * <code>string str_value = 5;</code>
-     */
-    public Builder setStrValueBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      valueCase_ = 5;
-      value_ = value;
-      onChanged();
-      return this;
-    }
-
-    /**
-     * <code>bytes bytes_value = 6;</code>
-     */
-    public com.google.protobuf.ByteString getBytesValue() {
-      if (valueCase_ == 6) {
-        return (com.google.protobuf.ByteString) value_;
-      }
-      return com.google.protobuf.ByteString.EMPTY;
-    }
-    /**
-     * <code>bytes bytes_value = 6;</code>
-     */
-    public Builder setBytesValue(com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  valueCase_ = 6;
-      value_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>bytes bytes_value = 6;</code>
-     */
-    public Builder clearBytesValue() {
-      if (valueCase_ == 6) {
-        valueCase_ = 0;
-        value_ = null;
-        onChanged();
-      }
-      return this;
-    }
-
-    /**
-     * <pre>
-     * A string value that stored in XStatMetadata::name.
-     * </pre>
-     *
-     * <code>uint64 ref_value = 7;</code>
-     */
-    public long getRefValue() {
-      if (valueCase_ == 7) {
-        return (java.lang.Long) value_;
-      }
-      return 0L;
-    }
-    /**
-     * <pre>
-     * A string value that stored in XStatMetadata::name.
-     * </pre>
-     *
-     * <code>uint64 ref_value = 7;</code>
-     */
-    public Builder setRefValue(long value) {
-      valueCase_ = 7;
-      value_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * A string value that stored in XStatMetadata::name.
-     * </pre>
-     *
-     * <code>uint64 ref_value = 7;</code>
-     */
-    public Builder clearRefValue() {
-      if (valueCase_ == 7) {
-        valueCase_ = 0;
-        value_ = null;
-        onChanged();
-      }
-      return this;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.profiler.XStat)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.profiler.XStat)
-  private static final org.tensorflow.proto.profiler.XStat DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.profiler.XStat();
-  }
-
-  public static org.tensorflow.proto.profiler.XStat getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<XStat>
-      PARSER = new com.google.protobuf.AbstractParser<XStat>() {
-    @java.lang.Override
-    public XStat parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new XStat(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<XStat> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<XStat> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.profiler.XStat getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/XStatMetadata.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/XStatMetadata.java
deleted file mode 100644
index 8967f6e6c1e..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/XStatMetadata.java
+++ /dev/null
@@ -1,813 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/profiler/protobuf/xplane.proto
-
-package org.tensorflow.proto.profiler;
-
-/**
- * <pre>
- * Metadata for an XStat, shared by all instances of the same stat.
- * Next ID: 4
- * </pre>
- *
- * Protobuf type {@code tensorflow.profiler.XStatMetadata}
- */
-public  final class XStatMetadata extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.profiler.XStatMetadata)
-    XStatMetadataOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use XStatMetadata.newBuilder() to construct.
-  private XStatMetadata(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private XStatMetadata() {
-    name_ = "";
-    description_ = "";
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new XStatMetadata();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private XStatMetadata(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 8: {
-
-            id_ = input.readInt64();
-            break;
-          }
-          case 18: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            name_ = s;
-            break;
-          }
-          case 26: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            description_ = s;
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.profiler.XPlaneProtos.internal_static_tensorflow_profiler_XStatMetadata_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.profiler.XPlaneProtos.internal_static_tensorflow_profiler_XStatMetadata_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.profiler.XStatMetadata.class, org.tensorflow.proto.profiler.XStatMetadata.Builder.class);
-  }
-
-  public static final int ID_FIELD_NUMBER = 1;
-  private long id_;
-  /**
-   * <pre>
-   * XPlane.stat_metadata map key.
-   * </pre>
-   *
-   * <code>int64 id = 1;</code>
-   */
-  public long getId() {
-    return id_;
-  }
-
-  public static final int NAME_FIELD_NUMBER = 2;
-  private volatile java.lang.Object name_;
-  /**
-   * <pre>
-   * Name of the stat (should be short).
-   * </pre>
-   *
-   * <code>string name = 2;</code>
-   */
-  public java.lang.String getName() {
-    java.lang.Object ref = name_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      name_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * Name of the stat (should be short).
-   * </pre>
-   *
-   * <code>string name = 2;</code>
-   */
-  public com.google.protobuf.ByteString
-      getNameBytes() {
-    java.lang.Object ref = name_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      name_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int DESCRIPTION_FIELD_NUMBER = 3;
-  private volatile java.lang.Object description_;
-  /**
-   * <pre>
-   * Description of the stat (might be long).
-   * </pre>
-   *
-   * <code>string description = 3;</code>
-   */
-  public java.lang.String getDescription() {
-    java.lang.Object ref = description_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      description_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * Description of the stat (might be long).
-   * </pre>
-   *
-   * <code>string description = 3;</code>
-   */
-  public com.google.protobuf.ByteString
-      getDescriptionBytes() {
-    java.lang.Object ref = description_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      description_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    if (id_ != 0L) {
-      output.writeInt64(1, id_);
-    }
-    if (!getNameBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 2, name_);
-    }
-    if (!getDescriptionBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 3, description_);
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (id_ != 0L) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt64Size(1, id_);
-    }
-    if (!getNameBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(2, name_);
-    }
-    if (!getDescriptionBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(3, description_);
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.profiler.XStatMetadata)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.profiler.XStatMetadata other = (org.tensorflow.proto.profiler.XStatMetadata) obj;
-
-    if (getId()
-        != other.getId()) return false;
-    if (!getName()
-        .equals(other.getName())) return false;
-    if (!getDescription()
-        .equals(other.getDescription())) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    hash = (37 * hash) + ID_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-        getId());
-    hash = (37 * hash) + NAME_FIELD_NUMBER;
-    hash = (53 * hash) + getName().hashCode();
-    hash = (37 * hash) + DESCRIPTION_FIELD_NUMBER;
-    hash = (53 * hash) + getDescription().hashCode();
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.profiler.XStatMetadata parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.profiler.XStatMetadata parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.profiler.XStatMetadata parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.profiler.XStatMetadata parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.profiler.XStatMetadata parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.profiler.XStatMetadata parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.profiler.XStatMetadata parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.profiler.XStatMetadata parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.profiler.XStatMetadata parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.profiler.XStatMetadata parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.profiler.XStatMetadata parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.profiler.XStatMetadata parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.profiler.XStatMetadata prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * <pre>
-   * Metadata for an XStat, shared by all instances of the same stat.
-   * Next ID: 4
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.profiler.XStatMetadata}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.profiler.XStatMetadata)
-      org.tensorflow.proto.profiler.XStatMetadataOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.profiler.XPlaneProtos.internal_static_tensorflow_profiler_XStatMetadata_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.profiler.XPlaneProtos.internal_static_tensorflow_profiler_XStatMetadata_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.profiler.XStatMetadata.class, org.tensorflow.proto.profiler.XStatMetadata.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.profiler.XStatMetadata.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      id_ = 0L;
-
-      name_ = "";
-
-      description_ = "";
-
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.profiler.XPlaneProtos.internal_static_tensorflow_profiler_XStatMetadata_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.profiler.XStatMetadata getDefaultInstanceForType() {
-      return org.tensorflow.proto.profiler.XStatMetadata.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.profiler.XStatMetadata build() {
-      org.tensorflow.proto.profiler.XStatMetadata result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.profiler.XStatMetadata buildPartial() {
-      org.tensorflow.proto.profiler.XStatMetadata result = new org.tensorflow.proto.profiler.XStatMetadata(this);
-      result.id_ = id_;
-      result.name_ = name_;
-      result.description_ = description_;
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.profiler.XStatMetadata) {
-        return mergeFrom((org.tensorflow.proto.profiler.XStatMetadata)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.profiler.XStatMetadata other) {
-      if (other == org.tensorflow.proto.profiler.XStatMetadata.getDefaultInstance()) return this;
-      if (other.getId() != 0L) {
-        setId(other.getId());
-      }
-      if (!other.getName().isEmpty()) {
-        name_ = other.name_;
-        onChanged();
-      }
-      if (!other.getDescription().isEmpty()) {
-        description_ = other.description_;
-        onChanged();
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.profiler.XStatMetadata parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.profiler.XStatMetadata) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-
-    private long id_ ;
-    /**
-     * <pre>
-     * XPlane.stat_metadata map key.
-     * </pre>
-     *
-     * <code>int64 id = 1;</code>
-     */
-    public long getId() {
-      return id_;
-    }
-    /**
-     * <pre>
-     * XPlane.stat_metadata map key.
-     * </pre>
-     *
-     * <code>int64 id = 1;</code>
-     */
-    public Builder setId(long value) {
-      
-      id_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * XPlane.stat_metadata map key.
-     * </pre>
-     *
-     * <code>int64 id = 1;</code>
-     */
-    public Builder clearId() {
-      
-      id_ = 0L;
-      onChanged();
-      return this;
-    }
-
-    private java.lang.Object name_ = "";
-    /**
-     * <pre>
-     * Name of the stat (should be short).
-     * </pre>
-     *
-     * <code>string name = 2;</code>
-     */
-    public java.lang.String getName() {
-      java.lang.Object ref = name_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        name_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Name of the stat (should be short).
-     * </pre>
-     *
-     * <code>string name = 2;</code>
-     */
-    public com.google.protobuf.ByteString
-        getNameBytes() {
-      java.lang.Object ref = name_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        name_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Name of the stat (should be short).
-     * </pre>
-     *
-     * <code>string name = 2;</code>
-     */
-    public Builder setName(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      name_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Name of the stat (should be short).
-     * </pre>
-     *
-     * <code>string name = 2;</code>
-     */
-    public Builder clearName() {
-      
-      name_ = getDefaultInstance().getName();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Name of the stat (should be short).
-     * </pre>
-     *
-     * <code>string name = 2;</code>
-     */
-    public Builder setNameBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      name_ = value;
-      onChanged();
-      return this;
-    }
-
-    private java.lang.Object description_ = "";
-    /**
-     * <pre>
-     * Description of the stat (might be long).
-     * </pre>
-     *
-     * <code>string description = 3;</code>
-     */
-    public java.lang.String getDescription() {
-      java.lang.Object ref = description_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        description_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Description of the stat (might be long).
-     * </pre>
-     *
-     * <code>string description = 3;</code>
-     */
-    public com.google.protobuf.ByteString
-        getDescriptionBytes() {
-      java.lang.Object ref = description_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        description_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Description of the stat (might be long).
-     * </pre>
-     *
-     * <code>string description = 3;</code>
-     */
-    public Builder setDescription(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      description_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Description of the stat (might be long).
-     * </pre>
-     *
-     * <code>string description = 3;</code>
-     */
-    public Builder clearDescription() {
-      
-      description_ = getDefaultInstance().getDescription();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Description of the stat (might be long).
-     * </pre>
-     *
-     * <code>string description = 3;</code>
-     */
-    public Builder setDescriptionBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      description_ = value;
-      onChanged();
-      return this;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.profiler.XStatMetadata)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.profiler.XStatMetadata)
-  private static final org.tensorflow.proto.profiler.XStatMetadata DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.profiler.XStatMetadata();
-  }
-
-  public static org.tensorflow.proto.profiler.XStatMetadata getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<XStatMetadata>
-      PARSER = new com.google.protobuf.AbstractParser<XStatMetadata>() {
-    @java.lang.Override
-    public XStatMetadata parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new XStatMetadata(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<XStatMetadata> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<XStatMetadata> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.profiler.XStatMetadata getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/XStatMetadataOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/XStatMetadataOrBuilder.java
deleted file mode 100644
index ebd901428ad..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/XStatMetadataOrBuilder.java
+++ /dev/null
@@ -1,54 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/profiler/protobuf/xplane.proto
-
-package org.tensorflow.proto.profiler;
-
-public interface XStatMetadataOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.profiler.XStatMetadata)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <pre>
-   * XPlane.stat_metadata map key.
-   * </pre>
-   *
-   * <code>int64 id = 1;</code>
-   */
-  long getId();
-
-  /**
-   * <pre>
-   * Name of the stat (should be short).
-   * </pre>
-   *
-   * <code>string name = 2;</code>
-   */
-  java.lang.String getName();
-  /**
-   * <pre>
-   * Name of the stat (should be short).
-   * </pre>
-   *
-   * <code>string name = 2;</code>
-   */
-  com.google.protobuf.ByteString
-      getNameBytes();
-
-  /**
-   * <pre>
-   * Description of the stat (might be long).
-   * </pre>
-   *
-   * <code>string description = 3;</code>
-   */
-  java.lang.String getDescription();
-  /**
-   * <pre>
-   * Description of the stat (might be long).
-   * </pre>
-   *
-   * <code>string description = 3;</code>
-   */
-  com.google.protobuf.ByteString
-      getDescriptionBytes();
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/XStatOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/XStatOrBuilder.java
deleted file mode 100644
index ddbd0ef1584..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/XStatOrBuilder.java
+++ /dev/null
@@ -1,59 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/profiler/protobuf/xplane.proto
-
-package org.tensorflow.proto.profiler;
-
-public interface XStatOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.profiler.XStat)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <pre>
-   * XStatMetadata.id of corresponding metadata.
-   * </pre>
-   *
-   * <code>int64 metadata_id = 1;</code>
-   */
-  long getMetadataId();
-
-  /**
-   * <code>double double_value = 2;</code>
-   */
-  double getDoubleValue();
-
-  /**
-   * <code>uint64 uint64_value = 3;</code>
-   */
-  long getUint64Value();
-
-  /**
-   * <code>int64 int64_value = 4;</code>
-   */
-  long getInt64Value();
-
-  /**
-   * <code>string str_value = 5;</code>
-   */
-  java.lang.String getStrValue();
-  /**
-   * <code>string str_value = 5;</code>
-   */
-  com.google.protobuf.ByteString
-      getStrValueBytes();
-
-  /**
-   * <code>bytes bytes_value = 6;</code>
-   */
-  com.google.protobuf.ByteString getBytesValue();
-
-  /**
-   * <pre>
-   * A string value that stored in XStatMetadata::name.
-   * </pre>
-   *
-   * <code>uint64 ref_value = 7;</code>
-   */
-  long getRefValue();
-
-  public org.tensorflow.proto.profiler.XStat.ValueCase getValueCase();
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/BfcMemoryMapProtos.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/BfcMemoryMapProtos.java
deleted file mode 100644
index b5870cc34d5..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/BfcMemoryMapProtos.java
+++ /dev/null
@@ -1,112 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/bfc_memory_map.proto
-
-package org.tensorflow.proto.util;
-
-public final class BfcMemoryMapProtos {
-  private BfcMemoryMapProtos() {}
-  public static void registerAllExtensions(
-      com.google.protobuf.ExtensionRegistryLite registry) {
-  }
-
-  public static void registerAllExtensions(
-      com.google.protobuf.ExtensionRegistry registry) {
-    registerAllExtensions(
-        (com.google.protobuf.ExtensionRegistryLite) registry);
-  }
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_MemAllocatorStats_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_MemAllocatorStats_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_MemChunk_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_MemChunk_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_BinSummary_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_BinSummary_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_SnapShot_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_SnapShot_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_MemoryDump_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_MemoryDump_fieldAccessorTable;
-
-  public static com.google.protobuf.Descriptors.FileDescriptor
-      getDescriptor() {
-    return descriptor;
-  }
-  private static  com.google.protobuf.Descriptors.FileDescriptor
-      descriptor;
-  static {
-    java.lang.String[] descriptorData = {
-      "\n-tensorflow/core/protobuf/bfc_memory_ma" +
-      "p.proto\022\ntensorflow\"\222\001\n\021MemAllocatorStat" +
-      "s\022\022\n\nnum_allocs\030\001 \001(\003\022\024\n\014bytes_in_use\030\002 " +
-      "\001(\003\022\031\n\021peak_bytes_in_use\030\003 \001(\003\022\032\n\022larges" +
-      "t_alloc_size\030\004 \001(\003\022\034\n\024fragmentation_metr" +
-      "ic\030\005 \001(\002\"\256\001\n\010MemChunk\022\017\n\007address\030\001 \001(\004\022\014" +
-      "\n\004size\030\002 \001(\003\022\026\n\016requested_size\030\003 \001(\003\022\013\n\003" +
-      "bin\030\004 \001(\005\022\017\n\007op_name\030\005 \001(\t\022\026\n\016freed_at_c" +
-      "ount\030\006 \001(\004\022\024\n\014action_count\030\007 \001(\004\022\016\n\006in_u" +
-      "se\030\010 \001(\010\022\017\n\007step_id\030\t \001(\004\"\213\001\n\nBinSummary" +
-      "\022\013\n\003bin\030\001 \001(\005\022\032\n\022total_bytes_in_use\030\002 \001(" +
-      "\003\022\032\n\022total_bytes_in_bin\030\003 \001(\003\022\033\n\023total_c" +
-      "hunks_in_use\030\004 \001(\003\022\033\n\023total_chunks_in_bi" +
-      "n\030\005 \001(\003\".\n\010SnapShot\022\024\n\014action_count\030\001 \001(" +
-      "\004\022\014\n\004size\030\002 \001(\003\"\315\001\n\nMemoryDump\022\026\n\016alloca" +
-      "tor_name\030\001 \001(\t\022+\n\013bin_summary\030\002 \003(\0132\026.te" +
-      "nsorflow.BinSummary\022#\n\005chunk\030\003 \003(\0132\024.ten" +
-      "sorflow.MemChunk\022\'\n\tsnap_shot\030\004 \003(\0132\024.te" +
-      "nsorflow.SnapShot\022,\n\005stats\030\005 \001(\0132\035.tenso" +
-      "rflow.MemAllocatorStatsB{\n\031org.tensorflo" +
-      "w.proto.utilB\022BfcMemoryMapProtosP\001ZHgith" +
-      "ub.com/tensorflow/tensorflow/tensorflow/" +
-      "go/core/core_protos_go_protob\006proto3"
-    };
-    descriptor = com.google.protobuf.Descriptors.FileDescriptor
-      .internalBuildGeneratedFileFrom(descriptorData,
-        new com.google.protobuf.Descriptors.FileDescriptor[] {
-        });
-    internal_static_tensorflow_MemAllocatorStats_descriptor =
-      getDescriptor().getMessageTypes().get(0);
-    internal_static_tensorflow_MemAllocatorStats_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_MemAllocatorStats_descriptor,
-        new java.lang.String[] { "NumAllocs", "BytesInUse", "PeakBytesInUse", "LargestAllocSize", "FragmentationMetric", });
-    internal_static_tensorflow_MemChunk_descriptor =
-      getDescriptor().getMessageTypes().get(1);
-    internal_static_tensorflow_MemChunk_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_MemChunk_descriptor,
-        new java.lang.String[] { "Address", "Size", "RequestedSize", "Bin", "OpName", "FreedAtCount", "ActionCount", "InUse", "StepId", });
-    internal_static_tensorflow_BinSummary_descriptor =
-      getDescriptor().getMessageTypes().get(2);
-    internal_static_tensorflow_BinSummary_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_BinSummary_descriptor,
-        new java.lang.String[] { "Bin", "TotalBytesInUse", "TotalBytesInBin", "TotalChunksInUse", "TotalChunksInBin", });
-    internal_static_tensorflow_SnapShot_descriptor =
-      getDescriptor().getMessageTypes().get(3);
-    internal_static_tensorflow_SnapShot_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_SnapShot_descriptor,
-        new java.lang.String[] { "ActionCount", "Size", });
-    internal_static_tensorflow_MemoryDump_descriptor =
-      getDescriptor().getMessageTypes().get(4);
-    internal_static_tensorflow_MemoryDump_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_MemoryDump_descriptor,
-        new java.lang.String[] { "AllocatorName", "BinSummary", "Chunk", "SnapShot", "Stats", });
-  }
-
-  // @@protoc_insertion_point(outer_class_scope)
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/BinSummary.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/BinSummary.java
deleted file mode 100644
index 66f221b62f0..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/BinSummary.java
+++ /dev/null
@@ -1,708 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/bfc_memory_map.proto
-
-package org.tensorflow.proto.util;
-
-/**
- * Protobuf type {@code tensorflow.BinSummary}
- */
-public  final class BinSummary extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.BinSummary)
-    BinSummaryOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use BinSummary.newBuilder() to construct.
-  private BinSummary(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private BinSummary() {
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new BinSummary();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private BinSummary(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 8: {
-
-            bin_ = input.readInt32();
-            break;
-          }
-          case 16: {
-
-            totalBytesInUse_ = input.readInt64();
-            break;
-          }
-          case 24: {
-
-            totalBytesInBin_ = input.readInt64();
-            break;
-          }
-          case 32: {
-
-            totalChunksInUse_ = input.readInt64();
-            break;
-          }
-          case 40: {
-
-            totalChunksInBin_ = input.readInt64();
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.util.BfcMemoryMapProtos.internal_static_tensorflow_BinSummary_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.util.BfcMemoryMapProtos.internal_static_tensorflow_BinSummary_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.util.BinSummary.class, org.tensorflow.proto.util.BinSummary.Builder.class);
-  }
-
-  public static final int BIN_FIELD_NUMBER = 1;
-  private int bin_;
-  /**
-   * <code>int32 bin = 1;</code>
-   */
-  public int getBin() {
-    return bin_;
-  }
-
-  public static final int TOTAL_BYTES_IN_USE_FIELD_NUMBER = 2;
-  private long totalBytesInUse_;
-  /**
-   * <code>int64 total_bytes_in_use = 2;</code>
-   */
-  public long getTotalBytesInUse() {
-    return totalBytesInUse_;
-  }
-
-  public static final int TOTAL_BYTES_IN_BIN_FIELD_NUMBER = 3;
-  private long totalBytesInBin_;
-  /**
-   * <code>int64 total_bytes_in_bin = 3;</code>
-   */
-  public long getTotalBytesInBin() {
-    return totalBytesInBin_;
-  }
-
-  public static final int TOTAL_CHUNKS_IN_USE_FIELD_NUMBER = 4;
-  private long totalChunksInUse_;
-  /**
-   * <code>int64 total_chunks_in_use = 4;</code>
-   */
-  public long getTotalChunksInUse() {
-    return totalChunksInUse_;
-  }
-
-  public static final int TOTAL_CHUNKS_IN_BIN_FIELD_NUMBER = 5;
-  private long totalChunksInBin_;
-  /**
-   * <code>int64 total_chunks_in_bin = 5;</code>
-   */
-  public long getTotalChunksInBin() {
-    return totalChunksInBin_;
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    if (bin_ != 0) {
-      output.writeInt32(1, bin_);
-    }
-    if (totalBytesInUse_ != 0L) {
-      output.writeInt64(2, totalBytesInUse_);
-    }
-    if (totalBytesInBin_ != 0L) {
-      output.writeInt64(3, totalBytesInBin_);
-    }
-    if (totalChunksInUse_ != 0L) {
-      output.writeInt64(4, totalChunksInUse_);
-    }
-    if (totalChunksInBin_ != 0L) {
-      output.writeInt64(5, totalChunksInBin_);
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (bin_ != 0) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt32Size(1, bin_);
-    }
-    if (totalBytesInUse_ != 0L) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt64Size(2, totalBytesInUse_);
-    }
-    if (totalBytesInBin_ != 0L) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt64Size(3, totalBytesInBin_);
-    }
-    if (totalChunksInUse_ != 0L) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt64Size(4, totalChunksInUse_);
-    }
-    if (totalChunksInBin_ != 0L) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt64Size(5, totalChunksInBin_);
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.util.BinSummary)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.util.BinSummary other = (org.tensorflow.proto.util.BinSummary) obj;
-
-    if (getBin()
-        != other.getBin()) return false;
-    if (getTotalBytesInUse()
-        != other.getTotalBytesInUse()) return false;
-    if (getTotalBytesInBin()
-        != other.getTotalBytesInBin()) return false;
-    if (getTotalChunksInUse()
-        != other.getTotalChunksInUse()) return false;
-    if (getTotalChunksInBin()
-        != other.getTotalChunksInBin()) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    hash = (37 * hash) + BIN_FIELD_NUMBER;
-    hash = (53 * hash) + getBin();
-    hash = (37 * hash) + TOTAL_BYTES_IN_USE_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-        getTotalBytesInUse());
-    hash = (37 * hash) + TOTAL_BYTES_IN_BIN_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-        getTotalBytesInBin());
-    hash = (37 * hash) + TOTAL_CHUNKS_IN_USE_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-        getTotalChunksInUse());
-    hash = (37 * hash) + TOTAL_CHUNKS_IN_BIN_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-        getTotalChunksInBin());
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.util.BinSummary parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.BinSummary parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.BinSummary parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.BinSummary parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.BinSummary parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.BinSummary parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.BinSummary parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.BinSummary parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.BinSummary parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.BinSummary parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.BinSummary parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.BinSummary parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.util.BinSummary prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * Protobuf type {@code tensorflow.BinSummary}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.BinSummary)
-      org.tensorflow.proto.util.BinSummaryOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.util.BfcMemoryMapProtos.internal_static_tensorflow_BinSummary_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.util.BfcMemoryMapProtos.internal_static_tensorflow_BinSummary_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.util.BinSummary.class, org.tensorflow.proto.util.BinSummary.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.util.BinSummary.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      bin_ = 0;
-
-      totalBytesInUse_ = 0L;
-
-      totalBytesInBin_ = 0L;
-
-      totalChunksInUse_ = 0L;
-
-      totalChunksInBin_ = 0L;
-
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.util.BfcMemoryMapProtos.internal_static_tensorflow_BinSummary_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.BinSummary getDefaultInstanceForType() {
-      return org.tensorflow.proto.util.BinSummary.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.BinSummary build() {
-      org.tensorflow.proto.util.BinSummary result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.BinSummary buildPartial() {
-      org.tensorflow.proto.util.BinSummary result = new org.tensorflow.proto.util.BinSummary(this);
-      result.bin_ = bin_;
-      result.totalBytesInUse_ = totalBytesInUse_;
-      result.totalBytesInBin_ = totalBytesInBin_;
-      result.totalChunksInUse_ = totalChunksInUse_;
-      result.totalChunksInBin_ = totalChunksInBin_;
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.util.BinSummary) {
-        return mergeFrom((org.tensorflow.proto.util.BinSummary)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.util.BinSummary other) {
-      if (other == org.tensorflow.proto.util.BinSummary.getDefaultInstance()) return this;
-      if (other.getBin() != 0) {
-        setBin(other.getBin());
-      }
-      if (other.getTotalBytesInUse() != 0L) {
-        setTotalBytesInUse(other.getTotalBytesInUse());
-      }
-      if (other.getTotalBytesInBin() != 0L) {
-        setTotalBytesInBin(other.getTotalBytesInBin());
-      }
-      if (other.getTotalChunksInUse() != 0L) {
-        setTotalChunksInUse(other.getTotalChunksInUse());
-      }
-      if (other.getTotalChunksInBin() != 0L) {
-        setTotalChunksInBin(other.getTotalChunksInBin());
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.util.BinSummary parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.util.BinSummary) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-
-    private int bin_ ;
-    /**
-     * <code>int32 bin = 1;</code>
-     */
-    public int getBin() {
-      return bin_;
-    }
-    /**
-     * <code>int32 bin = 1;</code>
-     */
-    public Builder setBin(int value) {
-      
-      bin_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>int32 bin = 1;</code>
-     */
-    public Builder clearBin() {
-      
-      bin_ = 0;
-      onChanged();
-      return this;
-    }
-
-    private long totalBytesInUse_ ;
-    /**
-     * <code>int64 total_bytes_in_use = 2;</code>
-     */
-    public long getTotalBytesInUse() {
-      return totalBytesInUse_;
-    }
-    /**
-     * <code>int64 total_bytes_in_use = 2;</code>
-     */
-    public Builder setTotalBytesInUse(long value) {
-      
-      totalBytesInUse_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>int64 total_bytes_in_use = 2;</code>
-     */
-    public Builder clearTotalBytesInUse() {
-      
-      totalBytesInUse_ = 0L;
-      onChanged();
-      return this;
-    }
-
-    private long totalBytesInBin_ ;
-    /**
-     * <code>int64 total_bytes_in_bin = 3;</code>
-     */
-    public long getTotalBytesInBin() {
-      return totalBytesInBin_;
-    }
-    /**
-     * <code>int64 total_bytes_in_bin = 3;</code>
-     */
-    public Builder setTotalBytesInBin(long value) {
-      
-      totalBytesInBin_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>int64 total_bytes_in_bin = 3;</code>
-     */
-    public Builder clearTotalBytesInBin() {
-      
-      totalBytesInBin_ = 0L;
-      onChanged();
-      return this;
-    }
-
-    private long totalChunksInUse_ ;
-    /**
-     * <code>int64 total_chunks_in_use = 4;</code>
-     */
-    public long getTotalChunksInUse() {
-      return totalChunksInUse_;
-    }
-    /**
-     * <code>int64 total_chunks_in_use = 4;</code>
-     */
-    public Builder setTotalChunksInUse(long value) {
-      
-      totalChunksInUse_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>int64 total_chunks_in_use = 4;</code>
-     */
-    public Builder clearTotalChunksInUse() {
-      
-      totalChunksInUse_ = 0L;
-      onChanged();
-      return this;
-    }
-
-    private long totalChunksInBin_ ;
-    /**
-     * <code>int64 total_chunks_in_bin = 5;</code>
-     */
-    public long getTotalChunksInBin() {
-      return totalChunksInBin_;
-    }
-    /**
-     * <code>int64 total_chunks_in_bin = 5;</code>
-     */
-    public Builder setTotalChunksInBin(long value) {
-      
-      totalChunksInBin_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>int64 total_chunks_in_bin = 5;</code>
-     */
-    public Builder clearTotalChunksInBin() {
-      
-      totalChunksInBin_ = 0L;
-      onChanged();
-      return this;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.BinSummary)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.BinSummary)
-  private static final org.tensorflow.proto.util.BinSummary DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.util.BinSummary();
-  }
-
-  public static org.tensorflow.proto.util.BinSummary getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<BinSummary>
-      PARSER = new com.google.protobuf.AbstractParser<BinSummary>() {
-    @java.lang.Override
-    public BinSummary parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new BinSummary(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<BinSummary> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<BinSummary> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.util.BinSummary getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/BinSummaryOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/BinSummaryOrBuilder.java
deleted file mode 100644
index 077107aa85f..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/BinSummaryOrBuilder.java
+++ /dev/null
@@ -1,34 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/bfc_memory_map.proto
-
-package org.tensorflow.proto.util;
-
-public interface BinSummaryOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.BinSummary)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <code>int32 bin = 1;</code>
-   */
-  int getBin();
-
-  /**
-   * <code>int64 total_bytes_in_use = 2;</code>
-   */
-  long getTotalBytesInUse();
-
-  /**
-   * <code>int64 total_bytes_in_bin = 3;</code>
-   */
-  long getTotalBytesInBin();
-
-  /**
-   * <code>int64 total_chunks_in_use = 4;</code>
-   */
-  long getTotalChunksInUse();
-
-  /**
-   * <code>int64 total_chunks_in_bin = 5;</code>
-   */
-  long getTotalChunksInBin();
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/BundleEntryProto.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/BundleEntryProto.java
deleted file mode 100644
index a74911cdde6..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/BundleEntryProto.java
+++ /dev/null
@@ -1,1540 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/tensor_bundle.proto
-
-package org.tensorflow.proto.util;
-
-/**
- * <pre>
- * Describes the metadata related to a checkpointed tensor.
- * </pre>
- *
- * Protobuf type {@code tensorflow.BundleEntryProto}
- */
-public  final class BundleEntryProto extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.BundleEntryProto)
-    BundleEntryProtoOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use BundleEntryProto.newBuilder() to construct.
-  private BundleEntryProto(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private BundleEntryProto() {
-    dtype_ = 0;
-    slices_ = java.util.Collections.emptyList();
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new BundleEntryProto();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private BundleEntryProto(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    int mutable_bitField0_ = 0;
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 8: {
-            int rawValue = input.readEnum();
-
-            dtype_ = rawValue;
-            break;
-          }
-          case 18: {
-            org.tensorflow.proto.framework.TensorShapeProto.Builder subBuilder = null;
-            if (shape_ != null) {
-              subBuilder = shape_.toBuilder();
-            }
-            shape_ = input.readMessage(org.tensorflow.proto.framework.TensorShapeProto.parser(), extensionRegistry);
-            if (subBuilder != null) {
-              subBuilder.mergeFrom(shape_);
-              shape_ = subBuilder.buildPartial();
-            }
-
-            break;
-          }
-          case 24: {
-
-            shardId_ = input.readInt32();
-            break;
-          }
-          case 32: {
-
-            offset_ = input.readInt64();
-            break;
-          }
-          case 40: {
-
-            size_ = input.readInt64();
-            break;
-          }
-          case 53: {
-
-            crc32C_ = input.readFixed32();
-            break;
-          }
-          case 58: {
-            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
-              slices_ = new java.util.ArrayList<org.tensorflow.proto.framework.TensorSliceProto>();
-              mutable_bitField0_ |= 0x00000001;
-            }
-            slices_.add(
-                input.readMessage(org.tensorflow.proto.framework.TensorSliceProto.parser(), extensionRegistry));
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      if (((mutable_bitField0_ & 0x00000001) != 0)) {
-        slices_ = java.util.Collections.unmodifiableList(slices_);
-      }
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.util.TensorBundleProtos.internal_static_tensorflow_BundleEntryProto_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.util.TensorBundleProtos.internal_static_tensorflow_BundleEntryProto_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.util.BundleEntryProto.class, org.tensorflow.proto.util.BundleEntryProto.Builder.class);
-  }
-
-  public static final int DTYPE_FIELD_NUMBER = 1;
-  private int dtype_;
-  /**
-   * <pre>
-   * The tensor dtype and shape.
-   * </pre>
-   *
-   * <code>.tensorflow.DataType dtype = 1;</code>
-   */
-  public int getDtypeValue() {
-    return dtype_;
-  }
-  /**
-   * <pre>
-   * The tensor dtype and shape.
-   * </pre>
-   *
-   * <code>.tensorflow.DataType dtype = 1;</code>
-   */
-  public org.tensorflow.proto.framework.DataType getDtype() {
-    @SuppressWarnings("deprecation")
-    org.tensorflow.proto.framework.DataType result = org.tensorflow.proto.framework.DataType.valueOf(dtype_);
-    return result == null ? org.tensorflow.proto.framework.DataType.UNRECOGNIZED : result;
-  }
-
-  public static final int SHAPE_FIELD_NUMBER = 2;
-  private org.tensorflow.proto.framework.TensorShapeProto shape_;
-  /**
-   * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-   */
-  public boolean hasShape() {
-    return shape_ != null;
-  }
-  /**
-   * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-   */
-  public org.tensorflow.proto.framework.TensorShapeProto getShape() {
-    return shape_ == null ? org.tensorflow.proto.framework.TensorShapeProto.getDefaultInstance() : shape_;
-  }
-  /**
-   * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-   */
-  public org.tensorflow.proto.framework.TensorShapeProtoOrBuilder getShapeOrBuilder() {
-    return getShape();
-  }
-
-  public static final int SHARD_ID_FIELD_NUMBER = 3;
-  private int shardId_;
-  /**
-   * <pre>
-   * The binary content of the tensor lies in:
-   *   File "shard_id": bytes [offset, offset + size).
-   * </pre>
-   *
-   * <code>int32 shard_id = 3;</code>
-   */
-  public int getShardId() {
-    return shardId_;
-  }
-
-  public static final int OFFSET_FIELD_NUMBER = 4;
-  private long offset_;
-  /**
-   * <code>int64 offset = 4;</code>
-   */
-  public long getOffset() {
-    return offset_;
-  }
-
-  public static final int SIZE_FIELD_NUMBER = 5;
-  private long size_;
-  /**
-   * <code>int64 size = 5;</code>
-   */
-  public long getSize() {
-    return size_;
-  }
-
-  public static final int CRC32C_FIELD_NUMBER = 6;
-  private int crc32C_;
-  /**
-   * <pre>
-   * The CRC32C checksum of the tensor bytes.
-   * </pre>
-   *
-   * <code>fixed32 crc32c = 6;</code>
-   */
-  public int getCrc32C() {
-    return crc32C_;
-  }
-
-  public static final int SLICES_FIELD_NUMBER = 7;
-  private java.util.List<org.tensorflow.proto.framework.TensorSliceProto> slices_;
-  /**
-   * <pre>
-   * Iff present, this entry represents a partitioned tensor.  The previous
-   * fields are interpreted as follows:
-   *   "dtype", "shape": describe the full tensor.
-   *   "shard_id", "offset", "size", "crc32c": all IGNORED.
-   *      These information for each slice can be looked up in their own
-   *      BundleEntryProto, keyed by each "slice_name".
-   * </pre>
-   *
-   * <code>repeated .tensorflow.TensorSliceProto slices = 7;</code>
-   */
-  public java.util.List<org.tensorflow.proto.framework.TensorSliceProto> getSlicesList() {
-    return slices_;
-  }
-  /**
-   * <pre>
-   * Iff present, this entry represents a partitioned tensor.  The previous
-   * fields are interpreted as follows:
-   *   "dtype", "shape": describe the full tensor.
-   *   "shard_id", "offset", "size", "crc32c": all IGNORED.
-   *      These information for each slice can be looked up in their own
-   *      BundleEntryProto, keyed by each "slice_name".
-   * </pre>
-   *
-   * <code>repeated .tensorflow.TensorSliceProto slices = 7;</code>
-   */
-  public java.util.List<? extends org.tensorflow.proto.framework.TensorSliceProtoOrBuilder> 
-      getSlicesOrBuilderList() {
-    return slices_;
-  }
-  /**
-   * <pre>
-   * Iff present, this entry represents a partitioned tensor.  The previous
-   * fields are interpreted as follows:
-   *   "dtype", "shape": describe the full tensor.
-   *   "shard_id", "offset", "size", "crc32c": all IGNORED.
-   *      These information for each slice can be looked up in their own
-   *      BundleEntryProto, keyed by each "slice_name".
-   * </pre>
-   *
-   * <code>repeated .tensorflow.TensorSliceProto slices = 7;</code>
-   */
-  public int getSlicesCount() {
-    return slices_.size();
-  }
-  /**
-   * <pre>
-   * Iff present, this entry represents a partitioned tensor.  The previous
-   * fields are interpreted as follows:
-   *   "dtype", "shape": describe the full tensor.
-   *   "shard_id", "offset", "size", "crc32c": all IGNORED.
-   *      These information for each slice can be looked up in their own
-   *      BundleEntryProto, keyed by each "slice_name".
-   * </pre>
-   *
-   * <code>repeated .tensorflow.TensorSliceProto slices = 7;</code>
-   */
-  public org.tensorflow.proto.framework.TensorSliceProto getSlices(int index) {
-    return slices_.get(index);
-  }
-  /**
-   * <pre>
-   * Iff present, this entry represents a partitioned tensor.  The previous
-   * fields are interpreted as follows:
-   *   "dtype", "shape": describe the full tensor.
-   *   "shard_id", "offset", "size", "crc32c": all IGNORED.
-   *      These information for each slice can be looked up in their own
-   *      BundleEntryProto, keyed by each "slice_name".
-   * </pre>
-   *
-   * <code>repeated .tensorflow.TensorSliceProto slices = 7;</code>
-   */
-  public org.tensorflow.proto.framework.TensorSliceProtoOrBuilder getSlicesOrBuilder(
-      int index) {
-    return slices_.get(index);
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    if (dtype_ != org.tensorflow.proto.framework.DataType.DT_INVALID.getNumber()) {
-      output.writeEnum(1, dtype_);
-    }
-    if (shape_ != null) {
-      output.writeMessage(2, getShape());
-    }
-    if (shardId_ != 0) {
-      output.writeInt32(3, shardId_);
-    }
-    if (offset_ != 0L) {
-      output.writeInt64(4, offset_);
-    }
-    if (size_ != 0L) {
-      output.writeInt64(5, size_);
-    }
-    if (crc32C_ != 0) {
-      output.writeFixed32(6, crc32C_);
-    }
-    for (int i = 0; i < slices_.size(); i++) {
-      output.writeMessage(7, slices_.get(i));
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (dtype_ != org.tensorflow.proto.framework.DataType.DT_INVALID.getNumber()) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeEnumSize(1, dtype_);
-    }
-    if (shape_ != null) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(2, getShape());
-    }
-    if (shardId_ != 0) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt32Size(3, shardId_);
-    }
-    if (offset_ != 0L) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt64Size(4, offset_);
-    }
-    if (size_ != 0L) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt64Size(5, size_);
-    }
-    if (crc32C_ != 0) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeFixed32Size(6, crc32C_);
-    }
-    for (int i = 0; i < slices_.size(); i++) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(7, slices_.get(i));
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.util.BundleEntryProto)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.util.BundleEntryProto other = (org.tensorflow.proto.util.BundleEntryProto) obj;
-
-    if (dtype_ != other.dtype_) return false;
-    if (hasShape() != other.hasShape()) return false;
-    if (hasShape()) {
-      if (!getShape()
-          .equals(other.getShape())) return false;
-    }
-    if (getShardId()
-        != other.getShardId()) return false;
-    if (getOffset()
-        != other.getOffset()) return false;
-    if (getSize()
-        != other.getSize()) return false;
-    if (getCrc32C()
-        != other.getCrc32C()) return false;
-    if (!getSlicesList()
-        .equals(other.getSlicesList())) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    hash = (37 * hash) + DTYPE_FIELD_NUMBER;
-    hash = (53 * hash) + dtype_;
-    if (hasShape()) {
-      hash = (37 * hash) + SHAPE_FIELD_NUMBER;
-      hash = (53 * hash) + getShape().hashCode();
-    }
-    hash = (37 * hash) + SHARD_ID_FIELD_NUMBER;
-    hash = (53 * hash) + getShardId();
-    hash = (37 * hash) + OFFSET_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-        getOffset());
-    hash = (37 * hash) + SIZE_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-        getSize());
-    hash = (37 * hash) + CRC32C_FIELD_NUMBER;
-    hash = (53 * hash) + getCrc32C();
-    if (getSlicesCount() > 0) {
-      hash = (37 * hash) + SLICES_FIELD_NUMBER;
-      hash = (53 * hash) + getSlicesList().hashCode();
-    }
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.util.BundleEntryProto parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.BundleEntryProto parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.BundleEntryProto parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.BundleEntryProto parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.BundleEntryProto parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.BundleEntryProto parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.BundleEntryProto parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.BundleEntryProto parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.BundleEntryProto parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.BundleEntryProto parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.BundleEntryProto parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.BundleEntryProto parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.util.BundleEntryProto prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * <pre>
-   * Describes the metadata related to a checkpointed tensor.
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.BundleEntryProto}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.BundleEntryProto)
-      org.tensorflow.proto.util.BundleEntryProtoOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.util.TensorBundleProtos.internal_static_tensorflow_BundleEntryProto_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.util.TensorBundleProtos.internal_static_tensorflow_BundleEntryProto_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.util.BundleEntryProto.class, org.tensorflow.proto.util.BundleEntryProto.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.util.BundleEntryProto.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-        getSlicesFieldBuilder();
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      dtype_ = 0;
-
-      if (shapeBuilder_ == null) {
-        shape_ = null;
-      } else {
-        shape_ = null;
-        shapeBuilder_ = null;
-      }
-      shardId_ = 0;
-
-      offset_ = 0L;
-
-      size_ = 0L;
-
-      crc32C_ = 0;
-
-      if (slicesBuilder_ == null) {
-        slices_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000001);
-      } else {
-        slicesBuilder_.clear();
-      }
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.util.TensorBundleProtos.internal_static_tensorflow_BundleEntryProto_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.BundleEntryProto getDefaultInstanceForType() {
-      return org.tensorflow.proto.util.BundleEntryProto.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.BundleEntryProto build() {
-      org.tensorflow.proto.util.BundleEntryProto result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.BundleEntryProto buildPartial() {
-      org.tensorflow.proto.util.BundleEntryProto result = new org.tensorflow.proto.util.BundleEntryProto(this);
-      int from_bitField0_ = bitField0_;
-      result.dtype_ = dtype_;
-      if (shapeBuilder_ == null) {
-        result.shape_ = shape_;
-      } else {
-        result.shape_ = shapeBuilder_.build();
-      }
-      result.shardId_ = shardId_;
-      result.offset_ = offset_;
-      result.size_ = size_;
-      result.crc32C_ = crc32C_;
-      if (slicesBuilder_ == null) {
-        if (((bitField0_ & 0x00000001) != 0)) {
-          slices_ = java.util.Collections.unmodifiableList(slices_);
-          bitField0_ = (bitField0_ & ~0x00000001);
-        }
-        result.slices_ = slices_;
-      } else {
-        result.slices_ = slicesBuilder_.build();
-      }
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.util.BundleEntryProto) {
-        return mergeFrom((org.tensorflow.proto.util.BundleEntryProto)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.util.BundleEntryProto other) {
-      if (other == org.tensorflow.proto.util.BundleEntryProto.getDefaultInstance()) return this;
-      if (other.dtype_ != 0) {
-        setDtypeValue(other.getDtypeValue());
-      }
-      if (other.hasShape()) {
-        mergeShape(other.getShape());
-      }
-      if (other.getShardId() != 0) {
-        setShardId(other.getShardId());
-      }
-      if (other.getOffset() != 0L) {
-        setOffset(other.getOffset());
-      }
-      if (other.getSize() != 0L) {
-        setSize(other.getSize());
-      }
-      if (other.getCrc32C() != 0) {
-        setCrc32C(other.getCrc32C());
-      }
-      if (slicesBuilder_ == null) {
-        if (!other.slices_.isEmpty()) {
-          if (slices_.isEmpty()) {
-            slices_ = other.slices_;
-            bitField0_ = (bitField0_ & ~0x00000001);
-          } else {
-            ensureSlicesIsMutable();
-            slices_.addAll(other.slices_);
-          }
-          onChanged();
-        }
-      } else {
-        if (!other.slices_.isEmpty()) {
-          if (slicesBuilder_.isEmpty()) {
-            slicesBuilder_.dispose();
-            slicesBuilder_ = null;
-            slices_ = other.slices_;
-            bitField0_ = (bitField0_ & ~0x00000001);
-            slicesBuilder_ = 
-              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
-                 getSlicesFieldBuilder() : null;
-          } else {
-            slicesBuilder_.addAllMessages(other.slices_);
-          }
-        }
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.util.BundleEntryProto parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.util.BundleEntryProto) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-    private int bitField0_;
-
-    private int dtype_ = 0;
-    /**
-     * <pre>
-     * The tensor dtype and shape.
-     * </pre>
-     *
-     * <code>.tensorflow.DataType dtype = 1;</code>
-     */
-    public int getDtypeValue() {
-      return dtype_;
-    }
-    /**
-     * <pre>
-     * The tensor dtype and shape.
-     * </pre>
-     *
-     * <code>.tensorflow.DataType dtype = 1;</code>
-     */
-    public Builder setDtypeValue(int value) {
-      dtype_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * The tensor dtype and shape.
-     * </pre>
-     *
-     * <code>.tensorflow.DataType dtype = 1;</code>
-     */
-    public org.tensorflow.proto.framework.DataType getDtype() {
-      @SuppressWarnings("deprecation")
-      org.tensorflow.proto.framework.DataType result = org.tensorflow.proto.framework.DataType.valueOf(dtype_);
-      return result == null ? org.tensorflow.proto.framework.DataType.UNRECOGNIZED : result;
-    }
-    /**
-     * <pre>
-     * The tensor dtype and shape.
-     * </pre>
-     *
-     * <code>.tensorflow.DataType dtype = 1;</code>
-     */
-    public Builder setDtype(org.tensorflow.proto.framework.DataType value) {
-      if (value == null) {
-        throw new NullPointerException();
-      }
-      
-      dtype_ = value.getNumber();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * The tensor dtype and shape.
-     * </pre>
-     *
-     * <code>.tensorflow.DataType dtype = 1;</code>
-     */
-    public Builder clearDtype() {
-      
-      dtype_ = 0;
-      onChanged();
-      return this;
-    }
-
-    private org.tensorflow.proto.framework.TensorShapeProto shape_;
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.TensorShapeProto, org.tensorflow.proto.framework.TensorShapeProto.Builder, org.tensorflow.proto.framework.TensorShapeProtoOrBuilder> shapeBuilder_;
-    /**
-     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-     */
-    public boolean hasShape() {
-      return shapeBuilder_ != null || shape_ != null;
-    }
-    /**
-     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-     */
-    public org.tensorflow.proto.framework.TensorShapeProto getShape() {
-      if (shapeBuilder_ == null) {
-        return shape_ == null ? org.tensorflow.proto.framework.TensorShapeProto.getDefaultInstance() : shape_;
-      } else {
-        return shapeBuilder_.getMessage();
-      }
-    }
-    /**
-     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-     */
-    public Builder setShape(org.tensorflow.proto.framework.TensorShapeProto value) {
-      if (shapeBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        shape_ = value;
-        onChanged();
-      } else {
-        shapeBuilder_.setMessage(value);
-      }
-
-      return this;
-    }
-    /**
-     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-     */
-    public Builder setShape(
-        org.tensorflow.proto.framework.TensorShapeProto.Builder builderForValue) {
-      if (shapeBuilder_ == null) {
-        shape_ = builderForValue.build();
-        onChanged();
-      } else {
-        shapeBuilder_.setMessage(builderForValue.build());
-      }
-
-      return this;
-    }
-    /**
-     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-     */
-    public Builder mergeShape(org.tensorflow.proto.framework.TensorShapeProto value) {
-      if (shapeBuilder_ == null) {
-        if (shape_ != null) {
-          shape_ =
-            org.tensorflow.proto.framework.TensorShapeProto.newBuilder(shape_).mergeFrom(value).buildPartial();
-        } else {
-          shape_ = value;
-        }
-        onChanged();
-      } else {
-        shapeBuilder_.mergeFrom(value);
-      }
-
-      return this;
-    }
-    /**
-     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-     */
-    public Builder clearShape() {
-      if (shapeBuilder_ == null) {
-        shape_ = null;
-        onChanged();
-      } else {
-        shape_ = null;
-        shapeBuilder_ = null;
-      }
-
-      return this;
-    }
-    /**
-     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-     */
-    public org.tensorflow.proto.framework.TensorShapeProto.Builder getShapeBuilder() {
-      
-      onChanged();
-      return getShapeFieldBuilder().getBuilder();
-    }
-    /**
-     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-     */
-    public org.tensorflow.proto.framework.TensorShapeProtoOrBuilder getShapeOrBuilder() {
-      if (shapeBuilder_ != null) {
-        return shapeBuilder_.getMessageOrBuilder();
-      } else {
-        return shape_ == null ?
-            org.tensorflow.proto.framework.TensorShapeProto.getDefaultInstance() : shape_;
-      }
-    }
-    /**
-     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-     */
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.TensorShapeProto, org.tensorflow.proto.framework.TensorShapeProto.Builder, org.tensorflow.proto.framework.TensorShapeProtoOrBuilder> 
-        getShapeFieldBuilder() {
-      if (shapeBuilder_ == null) {
-        shapeBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-            org.tensorflow.proto.framework.TensorShapeProto, org.tensorflow.proto.framework.TensorShapeProto.Builder, org.tensorflow.proto.framework.TensorShapeProtoOrBuilder>(
-                getShape(),
-                getParentForChildren(),
-                isClean());
-        shape_ = null;
-      }
-      return shapeBuilder_;
-    }
-
-    private int shardId_ ;
-    /**
-     * <pre>
-     * The binary content of the tensor lies in:
-     *   File "shard_id": bytes [offset, offset + size).
-     * </pre>
-     *
-     * <code>int32 shard_id = 3;</code>
-     */
-    public int getShardId() {
-      return shardId_;
-    }
-    /**
-     * <pre>
-     * The binary content of the tensor lies in:
-     *   File "shard_id": bytes [offset, offset + size).
-     * </pre>
-     *
-     * <code>int32 shard_id = 3;</code>
-     */
-    public Builder setShardId(int value) {
-      
-      shardId_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * The binary content of the tensor lies in:
-     *   File "shard_id": bytes [offset, offset + size).
-     * </pre>
-     *
-     * <code>int32 shard_id = 3;</code>
-     */
-    public Builder clearShardId() {
-      
-      shardId_ = 0;
-      onChanged();
-      return this;
-    }
-
-    private long offset_ ;
-    /**
-     * <code>int64 offset = 4;</code>
-     */
-    public long getOffset() {
-      return offset_;
-    }
-    /**
-     * <code>int64 offset = 4;</code>
-     */
-    public Builder setOffset(long value) {
-      
-      offset_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>int64 offset = 4;</code>
-     */
-    public Builder clearOffset() {
-      
-      offset_ = 0L;
-      onChanged();
-      return this;
-    }
-
-    private long size_ ;
-    /**
-     * <code>int64 size = 5;</code>
-     */
-    public long getSize() {
-      return size_;
-    }
-    /**
-     * <code>int64 size = 5;</code>
-     */
-    public Builder setSize(long value) {
-      
-      size_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>int64 size = 5;</code>
-     */
-    public Builder clearSize() {
-      
-      size_ = 0L;
-      onChanged();
-      return this;
-    }
-
-    private int crc32C_ ;
-    /**
-     * <pre>
-     * The CRC32C checksum of the tensor bytes.
-     * </pre>
-     *
-     * <code>fixed32 crc32c = 6;</code>
-     */
-    public int getCrc32C() {
-      return crc32C_;
-    }
-    /**
-     * <pre>
-     * The CRC32C checksum of the tensor bytes.
-     * </pre>
-     *
-     * <code>fixed32 crc32c = 6;</code>
-     */
-    public Builder setCrc32C(int value) {
-      
-      crc32C_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * The CRC32C checksum of the tensor bytes.
-     * </pre>
-     *
-     * <code>fixed32 crc32c = 6;</code>
-     */
-    public Builder clearCrc32C() {
-      
-      crc32C_ = 0;
-      onChanged();
-      return this;
-    }
-
-    private java.util.List<org.tensorflow.proto.framework.TensorSliceProto> slices_ =
-      java.util.Collections.emptyList();
-    private void ensureSlicesIsMutable() {
-      if (!((bitField0_ & 0x00000001) != 0)) {
-        slices_ = new java.util.ArrayList<org.tensorflow.proto.framework.TensorSliceProto>(slices_);
-        bitField0_ |= 0x00000001;
-       }
-    }
-
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.framework.TensorSliceProto, org.tensorflow.proto.framework.TensorSliceProto.Builder, org.tensorflow.proto.framework.TensorSliceProtoOrBuilder> slicesBuilder_;
-
-    /**
-     * <pre>
-     * Iff present, this entry represents a partitioned tensor.  The previous
-     * fields are interpreted as follows:
-     *   "dtype", "shape": describe the full tensor.
-     *   "shard_id", "offset", "size", "crc32c": all IGNORED.
-     *      These information for each slice can be looked up in their own
-     *      BundleEntryProto, keyed by each "slice_name".
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorSliceProto slices = 7;</code>
-     */
-    public java.util.List<org.tensorflow.proto.framework.TensorSliceProto> getSlicesList() {
-      if (slicesBuilder_ == null) {
-        return java.util.Collections.unmodifiableList(slices_);
-      } else {
-        return slicesBuilder_.getMessageList();
-      }
-    }
-    /**
-     * <pre>
-     * Iff present, this entry represents a partitioned tensor.  The previous
-     * fields are interpreted as follows:
-     *   "dtype", "shape": describe the full tensor.
-     *   "shard_id", "offset", "size", "crc32c": all IGNORED.
-     *      These information for each slice can be looked up in their own
-     *      BundleEntryProto, keyed by each "slice_name".
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorSliceProto slices = 7;</code>
-     */
-    public int getSlicesCount() {
-      if (slicesBuilder_ == null) {
-        return slices_.size();
-      } else {
-        return slicesBuilder_.getCount();
-      }
-    }
-    /**
-     * <pre>
-     * Iff present, this entry represents a partitioned tensor.  The previous
-     * fields are interpreted as follows:
-     *   "dtype", "shape": describe the full tensor.
-     *   "shard_id", "offset", "size", "crc32c": all IGNORED.
-     *      These information for each slice can be looked up in their own
-     *      BundleEntryProto, keyed by each "slice_name".
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorSliceProto slices = 7;</code>
-     */
-    public org.tensorflow.proto.framework.TensorSliceProto getSlices(int index) {
-      if (slicesBuilder_ == null) {
-        return slices_.get(index);
-      } else {
-        return slicesBuilder_.getMessage(index);
-      }
-    }
-    /**
-     * <pre>
-     * Iff present, this entry represents a partitioned tensor.  The previous
-     * fields are interpreted as follows:
-     *   "dtype", "shape": describe the full tensor.
-     *   "shard_id", "offset", "size", "crc32c": all IGNORED.
-     *      These information for each slice can be looked up in their own
-     *      BundleEntryProto, keyed by each "slice_name".
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorSliceProto slices = 7;</code>
-     */
-    public Builder setSlices(
-        int index, org.tensorflow.proto.framework.TensorSliceProto value) {
-      if (slicesBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureSlicesIsMutable();
-        slices_.set(index, value);
-        onChanged();
-      } else {
-        slicesBuilder_.setMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Iff present, this entry represents a partitioned tensor.  The previous
-     * fields are interpreted as follows:
-     *   "dtype", "shape": describe the full tensor.
-     *   "shard_id", "offset", "size", "crc32c": all IGNORED.
-     *      These information for each slice can be looked up in their own
-     *      BundleEntryProto, keyed by each "slice_name".
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorSliceProto slices = 7;</code>
-     */
-    public Builder setSlices(
-        int index, org.tensorflow.proto.framework.TensorSliceProto.Builder builderForValue) {
-      if (slicesBuilder_ == null) {
-        ensureSlicesIsMutable();
-        slices_.set(index, builderForValue.build());
-        onChanged();
-      } else {
-        slicesBuilder_.setMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Iff present, this entry represents a partitioned tensor.  The previous
-     * fields are interpreted as follows:
-     *   "dtype", "shape": describe the full tensor.
-     *   "shard_id", "offset", "size", "crc32c": all IGNORED.
-     *      These information for each slice can be looked up in their own
-     *      BundleEntryProto, keyed by each "slice_name".
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorSliceProto slices = 7;</code>
-     */
-    public Builder addSlices(org.tensorflow.proto.framework.TensorSliceProto value) {
-      if (slicesBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureSlicesIsMutable();
-        slices_.add(value);
-        onChanged();
-      } else {
-        slicesBuilder_.addMessage(value);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Iff present, this entry represents a partitioned tensor.  The previous
-     * fields are interpreted as follows:
-     *   "dtype", "shape": describe the full tensor.
-     *   "shard_id", "offset", "size", "crc32c": all IGNORED.
-     *      These information for each slice can be looked up in their own
-     *      BundleEntryProto, keyed by each "slice_name".
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorSliceProto slices = 7;</code>
-     */
-    public Builder addSlices(
-        int index, org.tensorflow.proto.framework.TensorSliceProto value) {
-      if (slicesBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureSlicesIsMutable();
-        slices_.add(index, value);
-        onChanged();
-      } else {
-        slicesBuilder_.addMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Iff present, this entry represents a partitioned tensor.  The previous
-     * fields are interpreted as follows:
-     *   "dtype", "shape": describe the full tensor.
-     *   "shard_id", "offset", "size", "crc32c": all IGNORED.
-     *      These information for each slice can be looked up in their own
-     *      BundleEntryProto, keyed by each "slice_name".
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorSliceProto slices = 7;</code>
-     */
-    public Builder addSlices(
-        org.tensorflow.proto.framework.TensorSliceProto.Builder builderForValue) {
-      if (slicesBuilder_ == null) {
-        ensureSlicesIsMutable();
-        slices_.add(builderForValue.build());
-        onChanged();
-      } else {
-        slicesBuilder_.addMessage(builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Iff present, this entry represents a partitioned tensor.  The previous
-     * fields are interpreted as follows:
-     *   "dtype", "shape": describe the full tensor.
-     *   "shard_id", "offset", "size", "crc32c": all IGNORED.
-     *      These information for each slice can be looked up in their own
-     *      BundleEntryProto, keyed by each "slice_name".
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorSliceProto slices = 7;</code>
-     */
-    public Builder addSlices(
-        int index, org.tensorflow.proto.framework.TensorSliceProto.Builder builderForValue) {
-      if (slicesBuilder_ == null) {
-        ensureSlicesIsMutable();
-        slices_.add(index, builderForValue.build());
-        onChanged();
-      } else {
-        slicesBuilder_.addMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Iff present, this entry represents a partitioned tensor.  The previous
-     * fields are interpreted as follows:
-     *   "dtype", "shape": describe the full tensor.
-     *   "shard_id", "offset", "size", "crc32c": all IGNORED.
-     *      These information for each slice can be looked up in their own
-     *      BundleEntryProto, keyed by each "slice_name".
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorSliceProto slices = 7;</code>
-     */
-    public Builder addAllSlices(
-        java.lang.Iterable<? extends org.tensorflow.proto.framework.TensorSliceProto> values) {
-      if (slicesBuilder_ == null) {
-        ensureSlicesIsMutable();
-        com.google.protobuf.AbstractMessageLite.Builder.addAll(
-            values, slices_);
-        onChanged();
-      } else {
-        slicesBuilder_.addAllMessages(values);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Iff present, this entry represents a partitioned tensor.  The previous
-     * fields are interpreted as follows:
-     *   "dtype", "shape": describe the full tensor.
-     *   "shard_id", "offset", "size", "crc32c": all IGNORED.
-     *      These information for each slice can be looked up in their own
-     *      BundleEntryProto, keyed by each "slice_name".
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorSliceProto slices = 7;</code>
-     */
-    public Builder clearSlices() {
-      if (slicesBuilder_ == null) {
-        slices_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000001);
-        onChanged();
-      } else {
-        slicesBuilder_.clear();
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Iff present, this entry represents a partitioned tensor.  The previous
-     * fields are interpreted as follows:
-     *   "dtype", "shape": describe the full tensor.
-     *   "shard_id", "offset", "size", "crc32c": all IGNORED.
-     *      These information for each slice can be looked up in their own
-     *      BundleEntryProto, keyed by each "slice_name".
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorSliceProto slices = 7;</code>
-     */
-    public Builder removeSlices(int index) {
-      if (slicesBuilder_ == null) {
-        ensureSlicesIsMutable();
-        slices_.remove(index);
-        onChanged();
-      } else {
-        slicesBuilder_.remove(index);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Iff present, this entry represents a partitioned tensor.  The previous
-     * fields are interpreted as follows:
-     *   "dtype", "shape": describe the full tensor.
-     *   "shard_id", "offset", "size", "crc32c": all IGNORED.
-     *      These information for each slice can be looked up in their own
-     *      BundleEntryProto, keyed by each "slice_name".
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorSliceProto slices = 7;</code>
-     */
-    public org.tensorflow.proto.framework.TensorSliceProto.Builder getSlicesBuilder(
-        int index) {
-      return getSlicesFieldBuilder().getBuilder(index);
-    }
-    /**
-     * <pre>
-     * Iff present, this entry represents a partitioned tensor.  The previous
-     * fields are interpreted as follows:
-     *   "dtype", "shape": describe the full tensor.
-     *   "shard_id", "offset", "size", "crc32c": all IGNORED.
-     *      These information for each slice can be looked up in their own
-     *      BundleEntryProto, keyed by each "slice_name".
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorSliceProto slices = 7;</code>
-     */
-    public org.tensorflow.proto.framework.TensorSliceProtoOrBuilder getSlicesOrBuilder(
-        int index) {
-      if (slicesBuilder_ == null) {
-        return slices_.get(index);  } else {
-        return slicesBuilder_.getMessageOrBuilder(index);
-      }
-    }
-    /**
-     * <pre>
-     * Iff present, this entry represents a partitioned tensor.  The previous
-     * fields are interpreted as follows:
-     *   "dtype", "shape": describe the full tensor.
-     *   "shard_id", "offset", "size", "crc32c": all IGNORED.
-     *      These information for each slice can be looked up in their own
-     *      BundleEntryProto, keyed by each "slice_name".
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorSliceProto slices = 7;</code>
-     */
-    public java.util.List<? extends org.tensorflow.proto.framework.TensorSliceProtoOrBuilder> 
-         getSlicesOrBuilderList() {
-      if (slicesBuilder_ != null) {
-        return slicesBuilder_.getMessageOrBuilderList();
-      } else {
-        return java.util.Collections.unmodifiableList(slices_);
-      }
-    }
-    /**
-     * <pre>
-     * Iff present, this entry represents a partitioned tensor.  The previous
-     * fields are interpreted as follows:
-     *   "dtype", "shape": describe the full tensor.
-     *   "shard_id", "offset", "size", "crc32c": all IGNORED.
-     *      These information for each slice can be looked up in their own
-     *      BundleEntryProto, keyed by each "slice_name".
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorSliceProto slices = 7;</code>
-     */
-    public org.tensorflow.proto.framework.TensorSliceProto.Builder addSlicesBuilder() {
-      return getSlicesFieldBuilder().addBuilder(
-          org.tensorflow.proto.framework.TensorSliceProto.getDefaultInstance());
-    }
-    /**
-     * <pre>
-     * Iff present, this entry represents a partitioned tensor.  The previous
-     * fields are interpreted as follows:
-     *   "dtype", "shape": describe the full tensor.
-     *   "shard_id", "offset", "size", "crc32c": all IGNORED.
-     *      These information for each slice can be looked up in their own
-     *      BundleEntryProto, keyed by each "slice_name".
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorSliceProto slices = 7;</code>
-     */
-    public org.tensorflow.proto.framework.TensorSliceProto.Builder addSlicesBuilder(
-        int index) {
-      return getSlicesFieldBuilder().addBuilder(
-          index, org.tensorflow.proto.framework.TensorSliceProto.getDefaultInstance());
-    }
-    /**
-     * <pre>
-     * Iff present, this entry represents a partitioned tensor.  The previous
-     * fields are interpreted as follows:
-     *   "dtype", "shape": describe the full tensor.
-     *   "shard_id", "offset", "size", "crc32c": all IGNORED.
-     *      These information for each slice can be looked up in their own
-     *      BundleEntryProto, keyed by each "slice_name".
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorSliceProto slices = 7;</code>
-     */
-    public java.util.List<org.tensorflow.proto.framework.TensorSliceProto.Builder> 
-         getSlicesBuilderList() {
-      return getSlicesFieldBuilder().getBuilderList();
-    }
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.framework.TensorSliceProto, org.tensorflow.proto.framework.TensorSliceProto.Builder, org.tensorflow.proto.framework.TensorSliceProtoOrBuilder> 
-        getSlicesFieldBuilder() {
-      if (slicesBuilder_ == null) {
-        slicesBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
-            org.tensorflow.proto.framework.TensorSliceProto, org.tensorflow.proto.framework.TensorSliceProto.Builder, org.tensorflow.proto.framework.TensorSliceProtoOrBuilder>(
-                slices_,
-                ((bitField0_ & 0x00000001) != 0),
-                getParentForChildren(),
-                isClean());
-        slices_ = null;
-      }
-      return slicesBuilder_;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.BundleEntryProto)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.BundleEntryProto)
-  private static final org.tensorflow.proto.util.BundleEntryProto DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.util.BundleEntryProto();
-  }
-
-  public static org.tensorflow.proto.util.BundleEntryProto getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<BundleEntryProto>
-      PARSER = new com.google.protobuf.AbstractParser<BundleEntryProto>() {
-    @java.lang.Override
-    public BundleEntryProto parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new BundleEntryProto(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<BundleEntryProto> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<BundleEntryProto> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.util.BundleEntryProto getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/BundleEntryProtoOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/BundleEntryProtoOrBuilder.java
deleted file mode 100644
index a95ca5d54e6..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/BundleEntryProtoOrBuilder.java
+++ /dev/null
@@ -1,137 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/tensor_bundle.proto
-
-package org.tensorflow.proto.util;
-
-public interface BundleEntryProtoOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.BundleEntryProto)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <pre>
-   * The tensor dtype and shape.
-   * </pre>
-   *
-   * <code>.tensorflow.DataType dtype = 1;</code>
-   */
-  int getDtypeValue();
-  /**
-   * <pre>
-   * The tensor dtype and shape.
-   * </pre>
-   *
-   * <code>.tensorflow.DataType dtype = 1;</code>
-   */
-  org.tensorflow.proto.framework.DataType getDtype();
-
-  /**
-   * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-   */
-  boolean hasShape();
-  /**
-   * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-   */
-  org.tensorflow.proto.framework.TensorShapeProto getShape();
-  /**
-   * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-   */
-  org.tensorflow.proto.framework.TensorShapeProtoOrBuilder getShapeOrBuilder();
-
-  /**
-   * <pre>
-   * The binary content of the tensor lies in:
-   *   File "shard_id": bytes [offset, offset + size).
-   * </pre>
-   *
-   * <code>int32 shard_id = 3;</code>
-   */
-  int getShardId();
-
-  /**
-   * <code>int64 offset = 4;</code>
-   */
-  long getOffset();
-
-  /**
-   * <code>int64 size = 5;</code>
-   */
-  long getSize();
-
-  /**
-   * <pre>
-   * The CRC32C checksum of the tensor bytes.
-   * </pre>
-   *
-   * <code>fixed32 crc32c = 6;</code>
-   */
-  int getCrc32C();
-
-  /**
-   * <pre>
-   * Iff present, this entry represents a partitioned tensor.  The previous
-   * fields are interpreted as follows:
-   *   "dtype", "shape": describe the full tensor.
-   *   "shard_id", "offset", "size", "crc32c": all IGNORED.
-   *      These information for each slice can be looked up in their own
-   *      BundleEntryProto, keyed by each "slice_name".
-   * </pre>
-   *
-   * <code>repeated .tensorflow.TensorSliceProto slices = 7;</code>
-   */
-  java.util.List<org.tensorflow.proto.framework.TensorSliceProto> 
-      getSlicesList();
-  /**
-   * <pre>
-   * Iff present, this entry represents a partitioned tensor.  The previous
-   * fields are interpreted as follows:
-   *   "dtype", "shape": describe the full tensor.
-   *   "shard_id", "offset", "size", "crc32c": all IGNORED.
-   *      These information for each slice can be looked up in their own
-   *      BundleEntryProto, keyed by each "slice_name".
-   * </pre>
-   *
-   * <code>repeated .tensorflow.TensorSliceProto slices = 7;</code>
-   */
-  org.tensorflow.proto.framework.TensorSliceProto getSlices(int index);
-  /**
-   * <pre>
-   * Iff present, this entry represents a partitioned tensor.  The previous
-   * fields are interpreted as follows:
-   *   "dtype", "shape": describe the full tensor.
-   *   "shard_id", "offset", "size", "crc32c": all IGNORED.
-   *      These information for each slice can be looked up in their own
-   *      BundleEntryProto, keyed by each "slice_name".
-   * </pre>
-   *
-   * <code>repeated .tensorflow.TensorSliceProto slices = 7;</code>
-   */
-  int getSlicesCount();
-  /**
-   * <pre>
-   * Iff present, this entry represents a partitioned tensor.  The previous
-   * fields are interpreted as follows:
-   *   "dtype", "shape": describe the full tensor.
-   *   "shard_id", "offset", "size", "crc32c": all IGNORED.
-   *      These information for each slice can be looked up in their own
-   *      BundleEntryProto, keyed by each "slice_name".
-   * </pre>
-   *
-   * <code>repeated .tensorflow.TensorSliceProto slices = 7;</code>
-   */
-  java.util.List<? extends org.tensorflow.proto.framework.TensorSliceProtoOrBuilder> 
-      getSlicesOrBuilderList();
-  /**
-   * <pre>
-   * Iff present, this entry represents a partitioned tensor.  The previous
-   * fields are interpreted as follows:
-   *   "dtype", "shape": describe the full tensor.
-   *   "shard_id", "offset", "size", "crc32c": all IGNORED.
-   *      These information for each slice can be looked up in their own
-   *      BundleEntryProto, keyed by each "slice_name".
-   * </pre>
-   *
-   * <code>repeated .tensorflow.TensorSliceProto slices = 7;</code>
-   */
-  org.tensorflow.proto.framework.TensorSliceProtoOrBuilder getSlicesOrBuilder(
-      int index);
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/BundleHeaderProto.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/BundleHeaderProto.java
deleted file mode 100644
index 1a30188c8a5..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/BundleHeaderProto.java
+++ /dev/null
@@ -1,929 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/tensor_bundle.proto
-
-package org.tensorflow.proto.util;
-
-/**
- * <pre>
- * Special header that is associated with a bundle.
- * TODO(zongheng,zhifengc): maybe in the future, we can add information about
- * which binary produced this checkpoint, timestamp, etc. Sometime, these can be
- * valuable debugging information. And if needed, these can be used as defensive
- * information ensuring reader (binary version) of the checkpoint and the writer
- * (binary version) must match within certain range, etc.
- * </pre>
- *
- * Protobuf type {@code tensorflow.BundleHeaderProto}
- */
-public  final class BundleHeaderProto extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.BundleHeaderProto)
-    BundleHeaderProtoOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use BundleHeaderProto.newBuilder() to construct.
-  private BundleHeaderProto(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private BundleHeaderProto() {
-    endianness_ = 0;
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new BundleHeaderProto();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private BundleHeaderProto(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 8: {
-
-            numShards_ = input.readInt32();
-            break;
-          }
-          case 16: {
-            int rawValue = input.readEnum();
-
-            endianness_ = rawValue;
-            break;
-          }
-          case 26: {
-            org.tensorflow.proto.framework.VersionDef.Builder subBuilder = null;
-            if (version_ != null) {
-              subBuilder = version_.toBuilder();
-            }
-            version_ = input.readMessage(org.tensorflow.proto.framework.VersionDef.parser(), extensionRegistry);
-            if (subBuilder != null) {
-              subBuilder.mergeFrom(version_);
-              version_ = subBuilder.buildPartial();
-            }
-
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.util.TensorBundleProtos.internal_static_tensorflow_BundleHeaderProto_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.util.TensorBundleProtos.internal_static_tensorflow_BundleHeaderProto_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.util.BundleHeaderProto.class, org.tensorflow.proto.util.BundleHeaderProto.Builder.class);
-  }
-
-  /**
-   * <pre>
-   * An enum indicating the endianness of the platform that produced this
-   * bundle.  A bundle can only be read by a platform with matching endianness.
-   * Defaults to LITTLE, as most modern platforms are little-endian.
-   * Affects the binary tensor data bytes only, not the metadata in protobufs.
-   * </pre>
-   *
-   * Protobuf enum {@code tensorflow.BundleHeaderProto.Endianness}
-   */
-  public enum Endianness
-      implements com.google.protobuf.ProtocolMessageEnum {
-    /**
-     * <code>LITTLE = 0;</code>
-     */
-    LITTLE(0),
-    /**
-     * <code>BIG = 1;</code>
-     */
-    BIG(1),
-    UNRECOGNIZED(-1),
-    ;
-
-    /**
-     * <code>LITTLE = 0;</code>
-     */
-    public static final int LITTLE_VALUE = 0;
-    /**
-     * <code>BIG = 1;</code>
-     */
-    public static final int BIG_VALUE = 1;
-
-
-    public final int getNumber() {
-      if (this == UNRECOGNIZED) {
-        throw new java.lang.IllegalArgumentException(
-            "Can't get the number of an unknown enum value.");
-      }
-      return value;
-    }
-
-    /**
-     * @deprecated Use {@link #forNumber(int)} instead.
-     */
-    @java.lang.Deprecated
-    public static Endianness valueOf(int value) {
-      return forNumber(value);
-    }
-
-    public static Endianness forNumber(int value) {
-      switch (value) {
-        case 0: return LITTLE;
-        case 1: return BIG;
-        default: return null;
-      }
-    }
-
-    public static com.google.protobuf.Internal.EnumLiteMap<Endianness>
-        internalGetValueMap() {
-      return internalValueMap;
-    }
-    private static final com.google.protobuf.Internal.EnumLiteMap<
-        Endianness> internalValueMap =
-          new com.google.protobuf.Internal.EnumLiteMap<Endianness>() {
-            public Endianness findValueByNumber(int number) {
-              return Endianness.forNumber(number);
-            }
-          };
-
-    public final com.google.protobuf.Descriptors.EnumValueDescriptor
-        getValueDescriptor() {
-      return getDescriptor().getValues().get(ordinal());
-    }
-    public final com.google.protobuf.Descriptors.EnumDescriptor
-        getDescriptorForType() {
-      return getDescriptor();
-    }
-    public static final com.google.protobuf.Descriptors.EnumDescriptor
-        getDescriptor() {
-      return org.tensorflow.proto.util.BundleHeaderProto.getDescriptor().getEnumTypes().get(0);
-    }
-
-    private static final Endianness[] VALUES = values();
-
-    public static Endianness valueOf(
-        com.google.protobuf.Descriptors.EnumValueDescriptor desc) {
-      if (desc.getType() != getDescriptor()) {
-        throw new java.lang.IllegalArgumentException(
-          "EnumValueDescriptor is not for this type.");
-      }
-      if (desc.getIndex() == -1) {
-        return UNRECOGNIZED;
-      }
-      return VALUES[desc.getIndex()];
-    }
-
-    private final int value;
-
-    private Endianness(int value) {
-      this.value = value;
-    }
-
-    // @@protoc_insertion_point(enum_scope:tensorflow.BundleHeaderProto.Endianness)
-  }
-
-  public static final int NUM_SHARDS_FIELD_NUMBER = 1;
-  private int numShards_;
-  /**
-   * <pre>
-   * Number of data files in the bundle.
-   * </pre>
-   *
-   * <code>int32 num_shards = 1;</code>
-   */
-  public int getNumShards() {
-    return numShards_;
-  }
-
-  public static final int ENDIANNESS_FIELD_NUMBER = 2;
-  private int endianness_;
-  /**
-   * <code>.tensorflow.BundleHeaderProto.Endianness endianness = 2;</code>
-   */
-  public int getEndiannessValue() {
-    return endianness_;
-  }
-  /**
-   * <code>.tensorflow.BundleHeaderProto.Endianness endianness = 2;</code>
-   */
-  public org.tensorflow.proto.util.BundleHeaderProto.Endianness getEndianness() {
-    @SuppressWarnings("deprecation")
-    org.tensorflow.proto.util.BundleHeaderProto.Endianness result = org.tensorflow.proto.util.BundleHeaderProto.Endianness.valueOf(endianness_);
-    return result == null ? org.tensorflow.proto.util.BundleHeaderProto.Endianness.UNRECOGNIZED : result;
-  }
-
-  public static final int VERSION_FIELD_NUMBER = 3;
-  private org.tensorflow.proto.framework.VersionDef version_;
-  /**
-   * <pre>
-   * Versioning of the tensor bundle format.
-   * </pre>
-   *
-   * <code>.tensorflow.VersionDef version = 3;</code>
-   */
-  public boolean hasVersion() {
-    return version_ != null;
-  }
-  /**
-   * <pre>
-   * Versioning of the tensor bundle format.
-   * </pre>
-   *
-   * <code>.tensorflow.VersionDef version = 3;</code>
-   */
-  public org.tensorflow.proto.framework.VersionDef getVersion() {
-    return version_ == null ? org.tensorflow.proto.framework.VersionDef.getDefaultInstance() : version_;
-  }
-  /**
-   * <pre>
-   * Versioning of the tensor bundle format.
-   * </pre>
-   *
-   * <code>.tensorflow.VersionDef version = 3;</code>
-   */
-  public org.tensorflow.proto.framework.VersionDefOrBuilder getVersionOrBuilder() {
-    return getVersion();
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    if (numShards_ != 0) {
-      output.writeInt32(1, numShards_);
-    }
-    if (endianness_ != org.tensorflow.proto.util.BundleHeaderProto.Endianness.LITTLE.getNumber()) {
-      output.writeEnum(2, endianness_);
-    }
-    if (version_ != null) {
-      output.writeMessage(3, getVersion());
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (numShards_ != 0) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt32Size(1, numShards_);
-    }
-    if (endianness_ != org.tensorflow.proto.util.BundleHeaderProto.Endianness.LITTLE.getNumber()) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeEnumSize(2, endianness_);
-    }
-    if (version_ != null) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(3, getVersion());
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.util.BundleHeaderProto)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.util.BundleHeaderProto other = (org.tensorflow.proto.util.BundleHeaderProto) obj;
-
-    if (getNumShards()
-        != other.getNumShards()) return false;
-    if (endianness_ != other.endianness_) return false;
-    if (hasVersion() != other.hasVersion()) return false;
-    if (hasVersion()) {
-      if (!getVersion()
-          .equals(other.getVersion())) return false;
-    }
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    hash = (37 * hash) + NUM_SHARDS_FIELD_NUMBER;
-    hash = (53 * hash) + getNumShards();
-    hash = (37 * hash) + ENDIANNESS_FIELD_NUMBER;
-    hash = (53 * hash) + endianness_;
-    if (hasVersion()) {
-      hash = (37 * hash) + VERSION_FIELD_NUMBER;
-      hash = (53 * hash) + getVersion().hashCode();
-    }
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.util.BundleHeaderProto parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.BundleHeaderProto parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.BundleHeaderProto parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.BundleHeaderProto parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.BundleHeaderProto parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.BundleHeaderProto parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.BundleHeaderProto parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.BundleHeaderProto parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.BundleHeaderProto parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.BundleHeaderProto parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.BundleHeaderProto parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.BundleHeaderProto parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.util.BundleHeaderProto prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * <pre>
-   * Special header that is associated with a bundle.
-   * TODO(zongheng,zhifengc): maybe in the future, we can add information about
-   * which binary produced this checkpoint, timestamp, etc. Sometime, these can be
-   * valuable debugging information. And if needed, these can be used as defensive
-   * information ensuring reader (binary version) of the checkpoint and the writer
-   * (binary version) must match within certain range, etc.
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.BundleHeaderProto}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.BundleHeaderProto)
-      org.tensorflow.proto.util.BundleHeaderProtoOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.util.TensorBundleProtos.internal_static_tensorflow_BundleHeaderProto_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.util.TensorBundleProtos.internal_static_tensorflow_BundleHeaderProto_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.util.BundleHeaderProto.class, org.tensorflow.proto.util.BundleHeaderProto.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.util.BundleHeaderProto.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      numShards_ = 0;
-
-      endianness_ = 0;
-
-      if (versionBuilder_ == null) {
-        version_ = null;
-      } else {
-        version_ = null;
-        versionBuilder_ = null;
-      }
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.util.TensorBundleProtos.internal_static_tensorflow_BundleHeaderProto_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.BundleHeaderProto getDefaultInstanceForType() {
-      return org.tensorflow.proto.util.BundleHeaderProto.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.BundleHeaderProto build() {
-      org.tensorflow.proto.util.BundleHeaderProto result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.BundleHeaderProto buildPartial() {
-      org.tensorflow.proto.util.BundleHeaderProto result = new org.tensorflow.proto.util.BundleHeaderProto(this);
-      result.numShards_ = numShards_;
-      result.endianness_ = endianness_;
-      if (versionBuilder_ == null) {
-        result.version_ = version_;
-      } else {
-        result.version_ = versionBuilder_.build();
-      }
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.util.BundleHeaderProto) {
-        return mergeFrom((org.tensorflow.proto.util.BundleHeaderProto)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.util.BundleHeaderProto other) {
-      if (other == org.tensorflow.proto.util.BundleHeaderProto.getDefaultInstance()) return this;
-      if (other.getNumShards() != 0) {
-        setNumShards(other.getNumShards());
-      }
-      if (other.endianness_ != 0) {
-        setEndiannessValue(other.getEndiannessValue());
-      }
-      if (other.hasVersion()) {
-        mergeVersion(other.getVersion());
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.util.BundleHeaderProto parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.util.BundleHeaderProto) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-
-    private int numShards_ ;
-    /**
-     * <pre>
-     * Number of data files in the bundle.
-     * </pre>
-     *
-     * <code>int32 num_shards = 1;</code>
-     */
-    public int getNumShards() {
-      return numShards_;
-    }
-    /**
-     * <pre>
-     * Number of data files in the bundle.
-     * </pre>
-     *
-     * <code>int32 num_shards = 1;</code>
-     */
-    public Builder setNumShards(int value) {
-      
-      numShards_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Number of data files in the bundle.
-     * </pre>
-     *
-     * <code>int32 num_shards = 1;</code>
-     */
-    public Builder clearNumShards() {
-      
-      numShards_ = 0;
-      onChanged();
-      return this;
-    }
-
-    private int endianness_ = 0;
-    /**
-     * <code>.tensorflow.BundleHeaderProto.Endianness endianness = 2;</code>
-     */
-    public int getEndiannessValue() {
-      return endianness_;
-    }
-    /**
-     * <code>.tensorflow.BundleHeaderProto.Endianness endianness = 2;</code>
-     */
-    public Builder setEndiannessValue(int value) {
-      endianness_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>.tensorflow.BundleHeaderProto.Endianness endianness = 2;</code>
-     */
-    public org.tensorflow.proto.util.BundleHeaderProto.Endianness getEndianness() {
-      @SuppressWarnings("deprecation")
-      org.tensorflow.proto.util.BundleHeaderProto.Endianness result = org.tensorflow.proto.util.BundleHeaderProto.Endianness.valueOf(endianness_);
-      return result == null ? org.tensorflow.proto.util.BundleHeaderProto.Endianness.UNRECOGNIZED : result;
-    }
-    /**
-     * <code>.tensorflow.BundleHeaderProto.Endianness endianness = 2;</code>
-     */
-    public Builder setEndianness(org.tensorflow.proto.util.BundleHeaderProto.Endianness value) {
-      if (value == null) {
-        throw new NullPointerException();
-      }
-      
-      endianness_ = value.getNumber();
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>.tensorflow.BundleHeaderProto.Endianness endianness = 2;</code>
-     */
-    public Builder clearEndianness() {
-      
-      endianness_ = 0;
-      onChanged();
-      return this;
-    }
-
-    private org.tensorflow.proto.framework.VersionDef version_;
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.VersionDef, org.tensorflow.proto.framework.VersionDef.Builder, org.tensorflow.proto.framework.VersionDefOrBuilder> versionBuilder_;
-    /**
-     * <pre>
-     * Versioning of the tensor bundle format.
-     * </pre>
-     *
-     * <code>.tensorflow.VersionDef version = 3;</code>
-     */
-    public boolean hasVersion() {
-      return versionBuilder_ != null || version_ != null;
-    }
-    /**
-     * <pre>
-     * Versioning of the tensor bundle format.
-     * </pre>
-     *
-     * <code>.tensorflow.VersionDef version = 3;</code>
-     */
-    public org.tensorflow.proto.framework.VersionDef getVersion() {
-      if (versionBuilder_ == null) {
-        return version_ == null ? org.tensorflow.proto.framework.VersionDef.getDefaultInstance() : version_;
-      } else {
-        return versionBuilder_.getMessage();
-      }
-    }
-    /**
-     * <pre>
-     * Versioning of the tensor bundle format.
-     * </pre>
-     *
-     * <code>.tensorflow.VersionDef version = 3;</code>
-     */
-    public Builder setVersion(org.tensorflow.proto.framework.VersionDef value) {
-      if (versionBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        version_ = value;
-        onChanged();
-      } else {
-        versionBuilder_.setMessage(value);
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Versioning of the tensor bundle format.
-     * </pre>
-     *
-     * <code>.tensorflow.VersionDef version = 3;</code>
-     */
-    public Builder setVersion(
-        org.tensorflow.proto.framework.VersionDef.Builder builderForValue) {
-      if (versionBuilder_ == null) {
-        version_ = builderForValue.build();
-        onChanged();
-      } else {
-        versionBuilder_.setMessage(builderForValue.build());
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Versioning of the tensor bundle format.
-     * </pre>
-     *
-     * <code>.tensorflow.VersionDef version = 3;</code>
-     */
-    public Builder mergeVersion(org.tensorflow.proto.framework.VersionDef value) {
-      if (versionBuilder_ == null) {
-        if (version_ != null) {
-          version_ =
-            org.tensorflow.proto.framework.VersionDef.newBuilder(version_).mergeFrom(value).buildPartial();
-        } else {
-          version_ = value;
-        }
-        onChanged();
-      } else {
-        versionBuilder_.mergeFrom(value);
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Versioning of the tensor bundle format.
-     * </pre>
-     *
-     * <code>.tensorflow.VersionDef version = 3;</code>
-     */
-    public Builder clearVersion() {
-      if (versionBuilder_ == null) {
-        version_ = null;
-        onChanged();
-      } else {
-        version_ = null;
-        versionBuilder_ = null;
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Versioning of the tensor bundle format.
-     * </pre>
-     *
-     * <code>.tensorflow.VersionDef version = 3;</code>
-     */
-    public org.tensorflow.proto.framework.VersionDef.Builder getVersionBuilder() {
-      
-      onChanged();
-      return getVersionFieldBuilder().getBuilder();
-    }
-    /**
-     * <pre>
-     * Versioning of the tensor bundle format.
-     * </pre>
-     *
-     * <code>.tensorflow.VersionDef version = 3;</code>
-     */
-    public org.tensorflow.proto.framework.VersionDefOrBuilder getVersionOrBuilder() {
-      if (versionBuilder_ != null) {
-        return versionBuilder_.getMessageOrBuilder();
-      } else {
-        return version_ == null ?
-            org.tensorflow.proto.framework.VersionDef.getDefaultInstance() : version_;
-      }
-    }
-    /**
-     * <pre>
-     * Versioning of the tensor bundle format.
-     * </pre>
-     *
-     * <code>.tensorflow.VersionDef version = 3;</code>
-     */
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.VersionDef, org.tensorflow.proto.framework.VersionDef.Builder, org.tensorflow.proto.framework.VersionDefOrBuilder> 
-        getVersionFieldBuilder() {
-      if (versionBuilder_ == null) {
-        versionBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-            org.tensorflow.proto.framework.VersionDef, org.tensorflow.proto.framework.VersionDef.Builder, org.tensorflow.proto.framework.VersionDefOrBuilder>(
-                getVersion(),
-                getParentForChildren(),
-                isClean());
-        version_ = null;
-      }
-      return versionBuilder_;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.BundleHeaderProto)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.BundleHeaderProto)
-  private static final org.tensorflow.proto.util.BundleHeaderProto DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.util.BundleHeaderProto();
-  }
-
-  public static org.tensorflow.proto.util.BundleHeaderProto getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<BundleHeaderProto>
-      PARSER = new com.google.protobuf.AbstractParser<BundleHeaderProto>() {
-    @java.lang.Override
-    public BundleHeaderProto parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new BundleHeaderProto(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<BundleHeaderProto> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<BundleHeaderProto> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.util.BundleHeaderProto getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/BundleHeaderProtoOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/BundleHeaderProtoOrBuilder.java
deleted file mode 100644
index f80c5fcef1c..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/BundleHeaderProtoOrBuilder.java
+++ /dev/null
@@ -1,52 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/tensor_bundle.proto
-
-package org.tensorflow.proto.util;
-
-public interface BundleHeaderProtoOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.BundleHeaderProto)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <pre>
-   * Number of data files in the bundle.
-   * </pre>
-   *
-   * <code>int32 num_shards = 1;</code>
-   */
-  int getNumShards();
-
-  /**
-   * <code>.tensorflow.BundleHeaderProto.Endianness endianness = 2;</code>
-   */
-  int getEndiannessValue();
-  /**
-   * <code>.tensorflow.BundleHeaderProto.Endianness endianness = 2;</code>
-   */
-  org.tensorflow.proto.util.BundleHeaderProto.Endianness getEndianness();
-
-  /**
-   * <pre>
-   * Versioning of the tensor bundle format.
-   * </pre>
-   *
-   * <code>.tensorflow.VersionDef version = 3;</code>
-   */
-  boolean hasVersion();
-  /**
-   * <pre>
-   * Versioning of the tensor bundle format.
-   * </pre>
-   *
-   * <code>.tensorflow.VersionDef version = 3;</code>
-   */
-  org.tensorflow.proto.framework.VersionDef getVersion();
-  /**
-   * <pre>
-   * Versioning of the tensor bundle format.
-   * </pre>
-   *
-   * <code>.tensorflow.VersionDef version = 3;</code>
-   */
-  org.tensorflow.proto.framework.VersionDefOrBuilder getVersionOrBuilder();
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/CodeLocation.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/CodeLocation.java
deleted file mode 100644
index 28d856a52f3..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/CodeLocation.java
+++ /dev/null
@@ -1,837 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/debug_event.proto
-
-package org.tensorflow.proto.util;
-
-/**
- * <pre>
- * Code location information: A stack trace with host-name information.
- * Instead of encoding the detailed stack trace, this proto refers to IDs of
- * stack frames stored as `StackFrameWithId` protos.
- * </pre>
- *
- * Protobuf type {@code tensorflow.CodeLocation}
- */
-public  final class CodeLocation extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.CodeLocation)
-    CodeLocationOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use CodeLocation.newBuilder() to construct.
-  private CodeLocation(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private CodeLocation() {
-    hostName_ = "";
-    stackFrameIds_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new CodeLocation();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private CodeLocation(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    int mutable_bitField0_ = 0;
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 10: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            hostName_ = s;
-            break;
-          }
-          case 18: {
-            java.lang.String s = input.readStringRequireUtf8();
-            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
-              stackFrameIds_ = new com.google.protobuf.LazyStringArrayList();
-              mutable_bitField0_ |= 0x00000001;
-            }
-            stackFrameIds_.add(s);
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      if (((mutable_bitField0_ & 0x00000001) != 0)) {
-        stackFrameIds_ = stackFrameIds_.getUnmodifiableView();
-      }
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.util.DebugEventProtos.internal_static_tensorflow_CodeLocation_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.util.DebugEventProtos.internal_static_tensorflow_CodeLocation_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.util.CodeLocation.class, org.tensorflow.proto.util.CodeLocation.Builder.class);
-  }
-
-  public static final int HOST_NAME_FIELD_NUMBER = 1;
-  private volatile java.lang.Object hostName_;
-  /**
-   * <pre>
-   * Host name on which the source files are located.
-   * </pre>
-   *
-   * <code>string host_name = 1;</code>
-   */
-  public java.lang.String getHostName() {
-    java.lang.Object ref = hostName_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      hostName_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * Host name on which the source files are located.
-   * </pre>
-   *
-   * <code>string host_name = 1;</code>
-   */
-  public com.google.protobuf.ByteString
-      getHostNameBytes() {
-    java.lang.Object ref = hostName_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      hostName_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int STACK_FRAME_IDS_FIELD_NUMBER = 2;
-  private com.google.protobuf.LazyStringList stackFrameIds_;
-  /**
-   * <pre>
-   * ID to a stack frame, each of which is pointed to
-   * by a unique ID. The ordering of the frames is consistent with Python's
-   * `traceback.extract_tb()`.
-   * </pre>
-   *
-   * <code>repeated string stack_frame_ids = 2;</code>
-   */
-  public com.google.protobuf.ProtocolStringList
-      getStackFrameIdsList() {
-    return stackFrameIds_;
-  }
-  /**
-   * <pre>
-   * ID to a stack frame, each of which is pointed to
-   * by a unique ID. The ordering of the frames is consistent with Python's
-   * `traceback.extract_tb()`.
-   * </pre>
-   *
-   * <code>repeated string stack_frame_ids = 2;</code>
-   */
-  public int getStackFrameIdsCount() {
-    return stackFrameIds_.size();
-  }
-  /**
-   * <pre>
-   * ID to a stack frame, each of which is pointed to
-   * by a unique ID. The ordering of the frames is consistent with Python's
-   * `traceback.extract_tb()`.
-   * </pre>
-   *
-   * <code>repeated string stack_frame_ids = 2;</code>
-   */
-  public java.lang.String getStackFrameIds(int index) {
-    return stackFrameIds_.get(index);
-  }
-  /**
-   * <pre>
-   * ID to a stack frame, each of which is pointed to
-   * by a unique ID. The ordering of the frames is consistent with Python's
-   * `traceback.extract_tb()`.
-   * </pre>
-   *
-   * <code>repeated string stack_frame_ids = 2;</code>
-   */
-  public com.google.protobuf.ByteString
-      getStackFrameIdsBytes(int index) {
-    return stackFrameIds_.getByteString(index);
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    if (!getHostNameBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 1, hostName_);
-    }
-    for (int i = 0; i < stackFrameIds_.size(); i++) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 2, stackFrameIds_.getRaw(i));
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (!getHostNameBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, hostName_);
-    }
-    {
-      int dataSize = 0;
-      for (int i = 0; i < stackFrameIds_.size(); i++) {
-        dataSize += computeStringSizeNoTag(stackFrameIds_.getRaw(i));
-      }
-      size += dataSize;
-      size += 1 * getStackFrameIdsList().size();
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.util.CodeLocation)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.util.CodeLocation other = (org.tensorflow.proto.util.CodeLocation) obj;
-
-    if (!getHostName()
-        .equals(other.getHostName())) return false;
-    if (!getStackFrameIdsList()
-        .equals(other.getStackFrameIdsList())) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    hash = (37 * hash) + HOST_NAME_FIELD_NUMBER;
-    hash = (53 * hash) + getHostName().hashCode();
-    if (getStackFrameIdsCount() > 0) {
-      hash = (37 * hash) + STACK_FRAME_IDS_FIELD_NUMBER;
-      hash = (53 * hash) + getStackFrameIdsList().hashCode();
-    }
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.util.CodeLocation parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.CodeLocation parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.CodeLocation parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.CodeLocation parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.CodeLocation parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.CodeLocation parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.CodeLocation parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.CodeLocation parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.CodeLocation parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.CodeLocation parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.CodeLocation parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.CodeLocation parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.util.CodeLocation prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * <pre>
-   * Code location information: A stack trace with host-name information.
-   * Instead of encoding the detailed stack trace, this proto refers to IDs of
-   * stack frames stored as `StackFrameWithId` protos.
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.CodeLocation}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.CodeLocation)
-      org.tensorflow.proto.util.CodeLocationOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.util.DebugEventProtos.internal_static_tensorflow_CodeLocation_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.util.DebugEventProtos.internal_static_tensorflow_CodeLocation_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.util.CodeLocation.class, org.tensorflow.proto.util.CodeLocation.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.util.CodeLocation.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      hostName_ = "";
-
-      stackFrameIds_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-      bitField0_ = (bitField0_ & ~0x00000001);
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.util.DebugEventProtos.internal_static_tensorflow_CodeLocation_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.CodeLocation getDefaultInstanceForType() {
-      return org.tensorflow.proto.util.CodeLocation.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.CodeLocation build() {
-      org.tensorflow.proto.util.CodeLocation result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.CodeLocation buildPartial() {
-      org.tensorflow.proto.util.CodeLocation result = new org.tensorflow.proto.util.CodeLocation(this);
-      int from_bitField0_ = bitField0_;
-      result.hostName_ = hostName_;
-      if (((bitField0_ & 0x00000001) != 0)) {
-        stackFrameIds_ = stackFrameIds_.getUnmodifiableView();
-        bitField0_ = (bitField0_ & ~0x00000001);
-      }
-      result.stackFrameIds_ = stackFrameIds_;
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.util.CodeLocation) {
-        return mergeFrom((org.tensorflow.proto.util.CodeLocation)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.util.CodeLocation other) {
-      if (other == org.tensorflow.proto.util.CodeLocation.getDefaultInstance()) return this;
-      if (!other.getHostName().isEmpty()) {
-        hostName_ = other.hostName_;
-        onChanged();
-      }
-      if (!other.stackFrameIds_.isEmpty()) {
-        if (stackFrameIds_.isEmpty()) {
-          stackFrameIds_ = other.stackFrameIds_;
-          bitField0_ = (bitField0_ & ~0x00000001);
-        } else {
-          ensureStackFrameIdsIsMutable();
-          stackFrameIds_.addAll(other.stackFrameIds_);
-        }
-        onChanged();
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.util.CodeLocation parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.util.CodeLocation) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-    private int bitField0_;
-
-    private java.lang.Object hostName_ = "";
-    /**
-     * <pre>
-     * Host name on which the source files are located.
-     * </pre>
-     *
-     * <code>string host_name = 1;</code>
-     */
-    public java.lang.String getHostName() {
-      java.lang.Object ref = hostName_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        hostName_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Host name on which the source files are located.
-     * </pre>
-     *
-     * <code>string host_name = 1;</code>
-     */
-    public com.google.protobuf.ByteString
-        getHostNameBytes() {
-      java.lang.Object ref = hostName_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        hostName_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Host name on which the source files are located.
-     * </pre>
-     *
-     * <code>string host_name = 1;</code>
-     */
-    public Builder setHostName(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      hostName_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Host name on which the source files are located.
-     * </pre>
-     *
-     * <code>string host_name = 1;</code>
-     */
-    public Builder clearHostName() {
-      
-      hostName_ = getDefaultInstance().getHostName();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Host name on which the source files are located.
-     * </pre>
-     *
-     * <code>string host_name = 1;</code>
-     */
-    public Builder setHostNameBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      hostName_ = value;
-      onChanged();
-      return this;
-    }
-
-    private com.google.protobuf.LazyStringList stackFrameIds_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-    private void ensureStackFrameIdsIsMutable() {
-      if (!((bitField0_ & 0x00000001) != 0)) {
-        stackFrameIds_ = new com.google.protobuf.LazyStringArrayList(stackFrameIds_);
-        bitField0_ |= 0x00000001;
-       }
-    }
-    /**
-     * <pre>
-     * ID to a stack frame, each of which is pointed to
-     * by a unique ID. The ordering of the frames is consistent with Python's
-     * `traceback.extract_tb()`.
-     * </pre>
-     *
-     * <code>repeated string stack_frame_ids = 2;</code>
-     */
-    public com.google.protobuf.ProtocolStringList
-        getStackFrameIdsList() {
-      return stackFrameIds_.getUnmodifiableView();
-    }
-    /**
-     * <pre>
-     * ID to a stack frame, each of which is pointed to
-     * by a unique ID. The ordering of the frames is consistent with Python's
-     * `traceback.extract_tb()`.
-     * </pre>
-     *
-     * <code>repeated string stack_frame_ids = 2;</code>
-     */
-    public int getStackFrameIdsCount() {
-      return stackFrameIds_.size();
-    }
-    /**
-     * <pre>
-     * ID to a stack frame, each of which is pointed to
-     * by a unique ID. The ordering of the frames is consistent with Python's
-     * `traceback.extract_tb()`.
-     * </pre>
-     *
-     * <code>repeated string stack_frame_ids = 2;</code>
-     */
-    public java.lang.String getStackFrameIds(int index) {
-      return stackFrameIds_.get(index);
-    }
-    /**
-     * <pre>
-     * ID to a stack frame, each of which is pointed to
-     * by a unique ID. The ordering of the frames is consistent with Python's
-     * `traceback.extract_tb()`.
-     * </pre>
-     *
-     * <code>repeated string stack_frame_ids = 2;</code>
-     */
-    public com.google.protobuf.ByteString
-        getStackFrameIdsBytes(int index) {
-      return stackFrameIds_.getByteString(index);
-    }
-    /**
-     * <pre>
-     * ID to a stack frame, each of which is pointed to
-     * by a unique ID. The ordering of the frames is consistent with Python's
-     * `traceback.extract_tb()`.
-     * </pre>
-     *
-     * <code>repeated string stack_frame_ids = 2;</code>
-     */
-    public Builder setStackFrameIds(
-        int index, java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  ensureStackFrameIdsIsMutable();
-      stackFrameIds_.set(index, value);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * ID to a stack frame, each of which is pointed to
-     * by a unique ID. The ordering of the frames is consistent with Python's
-     * `traceback.extract_tb()`.
-     * </pre>
-     *
-     * <code>repeated string stack_frame_ids = 2;</code>
-     */
-    public Builder addStackFrameIds(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  ensureStackFrameIdsIsMutable();
-      stackFrameIds_.add(value);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * ID to a stack frame, each of which is pointed to
-     * by a unique ID. The ordering of the frames is consistent with Python's
-     * `traceback.extract_tb()`.
-     * </pre>
-     *
-     * <code>repeated string stack_frame_ids = 2;</code>
-     */
-    public Builder addAllStackFrameIds(
-        java.lang.Iterable<java.lang.String> values) {
-      ensureStackFrameIdsIsMutable();
-      com.google.protobuf.AbstractMessageLite.Builder.addAll(
-          values, stackFrameIds_);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * ID to a stack frame, each of which is pointed to
-     * by a unique ID. The ordering of the frames is consistent with Python's
-     * `traceback.extract_tb()`.
-     * </pre>
-     *
-     * <code>repeated string stack_frame_ids = 2;</code>
-     */
-    public Builder clearStackFrameIds() {
-      stackFrameIds_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-      bitField0_ = (bitField0_ & ~0x00000001);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * ID to a stack frame, each of which is pointed to
-     * by a unique ID. The ordering of the frames is consistent with Python's
-     * `traceback.extract_tb()`.
-     * </pre>
-     *
-     * <code>repeated string stack_frame_ids = 2;</code>
-     */
-    public Builder addStackFrameIdsBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      ensureStackFrameIdsIsMutable();
-      stackFrameIds_.add(value);
-      onChanged();
-      return this;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.CodeLocation)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.CodeLocation)
-  private static final org.tensorflow.proto.util.CodeLocation DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.util.CodeLocation();
-  }
-
-  public static org.tensorflow.proto.util.CodeLocation getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<CodeLocation>
-      PARSER = new com.google.protobuf.AbstractParser<CodeLocation>() {
-    @java.lang.Override
-    public CodeLocation parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new CodeLocation(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<CodeLocation> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<CodeLocation> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.util.CodeLocation getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/CodeLocationOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/CodeLocationOrBuilder.java
deleted file mode 100644
index b75306258be..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/CodeLocationOrBuilder.java
+++ /dev/null
@@ -1,70 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/debug_event.proto
-
-package org.tensorflow.proto.util;
-
-public interface CodeLocationOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.CodeLocation)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <pre>
-   * Host name on which the source files are located.
-   * </pre>
-   *
-   * <code>string host_name = 1;</code>
-   */
-  java.lang.String getHostName();
-  /**
-   * <pre>
-   * Host name on which the source files are located.
-   * </pre>
-   *
-   * <code>string host_name = 1;</code>
-   */
-  com.google.protobuf.ByteString
-      getHostNameBytes();
-
-  /**
-   * <pre>
-   * ID to a stack frame, each of which is pointed to
-   * by a unique ID. The ordering of the frames is consistent with Python's
-   * `traceback.extract_tb()`.
-   * </pre>
-   *
-   * <code>repeated string stack_frame_ids = 2;</code>
-   */
-  java.util.List<java.lang.String>
-      getStackFrameIdsList();
-  /**
-   * <pre>
-   * ID to a stack frame, each of which is pointed to
-   * by a unique ID. The ordering of the frames is consistent with Python's
-   * `traceback.extract_tb()`.
-   * </pre>
-   *
-   * <code>repeated string stack_frame_ids = 2;</code>
-   */
-  int getStackFrameIdsCount();
-  /**
-   * <pre>
-   * ID to a stack frame, each of which is pointed to
-   * by a unique ID. The ordering of the frames is consistent with Python's
-   * `traceback.extract_tb()`.
-   * </pre>
-   *
-   * <code>repeated string stack_frame_ids = 2;</code>
-   */
-  java.lang.String getStackFrameIds(int index);
-  /**
-   * <pre>
-   * ID to a stack frame, each of which is pointed to
-   * by a unique ID. The ordering of the frames is consistent with Python's
-   * `traceback.extract_tb()`.
-   * </pre>
-   *
-   * <code>repeated string stack_frame_ids = 2;</code>
-   */
-  com.google.protobuf.ByteString
-      getStackFrameIdsBytes(int index);
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/DebugEvent.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/DebugEvent.java
deleted file mode 100644
index a1b0b986a3d..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/DebugEvent.java
+++ /dev/null
@@ -1,2883 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/debug_event.proto
-
-package org.tensorflow.proto.util;
-
-/**
- * <pre>
- * An Event related to the debugging of a TensorFlow program.
- * </pre>
- *
- * Protobuf type {@code tensorflow.DebugEvent}
- */
-public  final class DebugEvent extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.DebugEvent)
-    DebugEventOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use DebugEvent.newBuilder() to construct.
-  private DebugEvent(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private DebugEvent() {
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new DebugEvent();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private DebugEvent(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 9: {
-
-            wallTime_ = input.readDouble();
-            break;
-          }
-          case 16: {
-
-            step_ = input.readInt64();
-            break;
-          }
-          case 26: {
-            org.tensorflow.proto.util.DebugMetadata.Builder subBuilder = null;
-            if (whatCase_ == 3) {
-              subBuilder = ((org.tensorflow.proto.util.DebugMetadata) what_).toBuilder();
-            }
-            what_ =
-                input.readMessage(org.tensorflow.proto.util.DebugMetadata.parser(), extensionRegistry);
-            if (subBuilder != null) {
-              subBuilder.mergeFrom((org.tensorflow.proto.util.DebugMetadata) what_);
-              what_ = subBuilder.buildPartial();
-            }
-            whatCase_ = 3;
-            break;
-          }
-          case 34: {
-            org.tensorflow.proto.util.SourceFile.Builder subBuilder = null;
-            if (whatCase_ == 4) {
-              subBuilder = ((org.tensorflow.proto.util.SourceFile) what_).toBuilder();
-            }
-            what_ =
-                input.readMessage(org.tensorflow.proto.util.SourceFile.parser(), extensionRegistry);
-            if (subBuilder != null) {
-              subBuilder.mergeFrom((org.tensorflow.proto.util.SourceFile) what_);
-              what_ = subBuilder.buildPartial();
-            }
-            whatCase_ = 4;
-            break;
-          }
-          case 50: {
-            org.tensorflow.proto.util.StackFrameWithId.Builder subBuilder = null;
-            if (whatCase_ == 6) {
-              subBuilder = ((org.tensorflow.proto.util.StackFrameWithId) what_).toBuilder();
-            }
-            what_ =
-                input.readMessage(org.tensorflow.proto.util.StackFrameWithId.parser(), extensionRegistry);
-            if (subBuilder != null) {
-              subBuilder.mergeFrom((org.tensorflow.proto.util.StackFrameWithId) what_);
-              what_ = subBuilder.buildPartial();
-            }
-            whatCase_ = 6;
-            break;
-          }
-          case 58: {
-            org.tensorflow.proto.util.GraphOpCreation.Builder subBuilder = null;
-            if (whatCase_ == 7) {
-              subBuilder = ((org.tensorflow.proto.util.GraphOpCreation) what_).toBuilder();
-            }
-            what_ =
-                input.readMessage(org.tensorflow.proto.util.GraphOpCreation.parser(), extensionRegistry);
-            if (subBuilder != null) {
-              subBuilder.mergeFrom((org.tensorflow.proto.util.GraphOpCreation) what_);
-              what_ = subBuilder.buildPartial();
-            }
-            whatCase_ = 7;
-            break;
-          }
-          case 66: {
-            org.tensorflow.proto.util.DebuggedGraph.Builder subBuilder = null;
-            if (whatCase_ == 8) {
-              subBuilder = ((org.tensorflow.proto.util.DebuggedGraph) what_).toBuilder();
-            }
-            what_ =
-                input.readMessage(org.tensorflow.proto.util.DebuggedGraph.parser(), extensionRegistry);
-            if (subBuilder != null) {
-              subBuilder.mergeFrom((org.tensorflow.proto.util.DebuggedGraph) what_);
-              what_ = subBuilder.buildPartial();
-            }
-            whatCase_ = 8;
-            break;
-          }
-          case 74: {
-            org.tensorflow.proto.util.Execution.Builder subBuilder = null;
-            if (whatCase_ == 9) {
-              subBuilder = ((org.tensorflow.proto.util.Execution) what_).toBuilder();
-            }
-            what_ =
-                input.readMessage(org.tensorflow.proto.util.Execution.parser(), extensionRegistry);
-            if (subBuilder != null) {
-              subBuilder.mergeFrom((org.tensorflow.proto.util.Execution) what_);
-              what_ = subBuilder.buildPartial();
-            }
-            whatCase_ = 9;
-            break;
-          }
-          case 82: {
-            org.tensorflow.proto.util.GraphExecutionTrace.Builder subBuilder = null;
-            if (whatCase_ == 10) {
-              subBuilder = ((org.tensorflow.proto.util.GraphExecutionTrace) what_).toBuilder();
-            }
-            what_ =
-                input.readMessage(org.tensorflow.proto.util.GraphExecutionTrace.parser(), extensionRegistry);
-            if (subBuilder != null) {
-              subBuilder.mergeFrom((org.tensorflow.proto.util.GraphExecutionTrace) what_);
-              what_ = subBuilder.buildPartial();
-            }
-            whatCase_ = 10;
-            break;
-          }
-          case 90: {
-            java.lang.String s = input.readStringRequireUtf8();
-            whatCase_ = 11;
-            what_ = s;
-            break;
-          }
-          case 98: {
-            org.tensorflow.proto.util.DebuggedDevice.Builder subBuilder = null;
-            if (whatCase_ == 12) {
-              subBuilder = ((org.tensorflow.proto.util.DebuggedDevice) what_).toBuilder();
-            }
-            what_ =
-                input.readMessage(org.tensorflow.proto.util.DebuggedDevice.parser(), extensionRegistry);
-            if (subBuilder != null) {
-              subBuilder.mergeFrom((org.tensorflow.proto.util.DebuggedDevice) what_);
-              what_ = subBuilder.buildPartial();
-            }
-            whatCase_ = 12;
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.util.DebugEventProtos.internal_static_tensorflow_DebugEvent_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.util.DebugEventProtos.internal_static_tensorflow_DebugEvent_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.util.DebugEvent.class, org.tensorflow.proto.util.DebugEvent.Builder.class);
-  }
-
-  private int whatCase_ = 0;
-  private java.lang.Object what_;
-  public enum WhatCase
-      implements com.google.protobuf.Internal.EnumLite {
-    DEBUG_METADATA(3),
-    SOURCE_FILE(4),
-    STACK_FRAME_WITH_ID(6),
-    GRAPH_OP_CREATION(7),
-    DEBUGGED_GRAPH(8),
-    EXECUTION(9),
-    GRAPH_EXECUTION_TRACE(10),
-    GRAPH_ID(11),
-    DEBUGGED_DEVICE(12),
-    WHAT_NOT_SET(0);
-    private final int value;
-    private WhatCase(int value) {
-      this.value = value;
-    }
-    /**
-     * @deprecated Use {@link #forNumber(int)} instead.
-     */
-    @java.lang.Deprecated
-    public static WhatCase valueOf(int value) {
-      return forNumber(value);
-    }
-
-    public static WhatCase forNumber(int value) {
-      switch (value) {
-        case 3: return DEBUG_METADATA;
-        case 4: return SOURCE_FILE;
-        case 6: return STACK_FRAME_WITH_ID;
-        case 7: return GRAPH_OP_CREATION;
-        case 8: return DEBUGGED_GRAPH;
-        case 9: return EXECUTION;
-        case 10: return GRAPH_EXECUTION_TRACE;
-        case 11: return GRAPH_ID;
-        case 12: return DEBUGGED_DEVICE;
-        case 0: return WHAT_NOT_SET;
-        default: return null;
-      }
-    }
-    public int getNumber() {
-      return this.value;
-    }
-  };
-
-  public WhatCase
-  getWhatCase() {
-    return WhatCase.forNumber(
-        whatCase_);
-  }
-
-  public static final int WALL_TIME_FIELD_NUMBER = 1;
-  private double wallTime_;
-  /**
-   * <pre>
-   * Timestamp in seconds (with microsecond precision).
-   * </pre>
-   *
-   * <code>double wall_time = 1;</code>
-   */
-  public double getWallTime() {
-    return wallTime_;
-  }
-
-  public static final int STEP_FIELD_NUMBER = 2;
-  private long step_;
-  /**
-   * <pre>
-   * Step of training (if available).
-   * </pre>
-   *
-   * <code>int64 step = 2;</code>
-   */
-  public long getStep() {
-    return step_;
-  }
-
-  public static final int DEBUG_METADATA_FIELD_NUMBER = 3;
-  /**
-   * <pre>
-   * Metadata related to this debugging data.
-   * </pre>
-   *
-   * <code>.tensorflow.DebugMetadata debug_metadata = 3;</code>
-   */
-  public boolean hasDebugMetadata() {
-    return whatCase_ == 3;
-  }
-  /**
-   * <pre>
-   * Metadata related to this debugging data.
-   * </pre>
-   *
-   * <code>.tensorflow.DebugMetadata debug_metadata = 3;</code>
-   */
-  public org.tensorflow.proto.util.DebugMetadata getDebugMetadata() {
-    if (whatCase_ == 3) {
-       return (org.tensorflow.proto.util.DebugMetadata) what_;
-    }
-    return org.tensorflow.proto.util.DebugMetadata.getDefaultInstance();
-  }
-  /**
-   * <pre>
-   * Metadata related to this debugging data.
-   * </pre>
-   *
-   * <code>.tensorflow.DebugMetadata debug_metadata = 3;</code>
-   */
-  public org.tensorflow.proto.util.DebugMetadataOrBuilder getDebugMetadataOrBuilder() {
-    if (whatCase_ == 3) {
-       return (org.tensorflow.proto.util.DebugMetadata) what_;
-    }
-    return org.tensorflow.proto.util.DebugMetadata.getDefaultInstance();
-  }
-
-  public static final int SOURCE_FILE_FIELD_NUMBER = 4;
-  /**
-   * <pre>
-   * The content of a source file.
-   * </pre>
-   *
-   * <code>.tensorflow.SourceFile source_file = 4;</code>
-   */
-  public boolean hasSourceFile() {
-    return whatCase_ == 4;
-  }
-  /**
-   * <pre>
-   * The content of a source file.
-   * </pre>
-   *
-   * <code>.tensorflow.SourceFile source_file = 4;</code>
-   */
-  public org.tensorflow.proto.util.SourceFile getSourceFile() {
-    if (whatCase_ == 4) {
-       return (org.tensorflow.proto.util.SourceFile) what_;
-    }
-    return org.tensorflow.proto.util.SourceFile.getDefaultInstance();
-  }
-  /**
-   * <pre>
-   * The content of a source file.
-   * </pre>
-   *
-   * <code>.tensorflow.SourceFile source_file = 4;</code>
-   */
-  public org.tensorflow.proto.util.SourceFileOrBuilder getSourceFileOrBuilder() {
-    if (whatCase_ == 4) {
-       return (org.tensorflow.proto.util.SourceFile) what_;
-    }
-    return org.tensorflow.proto.util.SourceFile.getDefaultInstance();
-  }
-
-  public static final int STACK_FRAME_WITH_ID_FIELD_NUMBER = 6;
-  /**
-   * <pre>
-   * A stack frame (filename, line number and column number, function name and
-   * code string) with ID.
-   * </pre>
-   *
-   * <code>.tensorflow.StackFrameWithId stack_frame_with_id = 6;</code>
-   */
-  public boolean hasStackFrameWithId() {
-    return whatCase_ == 6;
-  }
-  /**
-   * <pre>
-   * A stack frame (filename, line number and column number, function name and
-   * code string) with ID.
-   * </pre>
-   *
-   * <code>.tensorflow.StackFrameWithId stack_frame_with_id = 6;</code>
-   */
-  public org.tensorflow.proto.util.StackFrameWithId getStackFrameWithId() {
-    if (whatCase_ == 6) {
-       return (org.tensorflow.proto.util.StackFrameWithId) what_;
-    }
-    return org.tensorflow.proto.util.StackFrameWithId.getDefaultInstance();
-  }
-  /**
-   * <pre>
-   * A stack frame (filename, line number and column number, function name and
-   * code string) with ID.
-   * </pre>
-   *
-   * <code>.tensorflow.StackFrameWithId stack_frame_with_id = 6;</code>
-   */
-  public org.tensorflow.proto.util.StackFrameWithIdOrBuilder getStackFrameWithIdOrBuilder() {
-    if (whatCase_ == 6) {
-       return (org.tensorflow.proto.util.StackFrameWithId) what_;
-    }
-    return org.tensorflow.proto.util.StackFrameWithId.getDefaultInstance();
-  }
-
-  public static final int GRAPH_OP_CREATION_FIELD_NUMBER = 7;
-  /**
-   * <pre>
-   * The creation of an op within a graph (e.g., a FuncGraph compiled from
-   * a Python function).
-   * </pre>
-   *
-   * <code>.tensorflow.GraphOpCreation graph_op_creation = 7;</code>
-   */
-  public boolean hasGraphOpCreation() {
-    return whatCase_ == 7;
-  }
-  /**
-   * <pre>
-   * The creation of an op within a graph (e.g., a FuncGraph compiled from
-   * a Python function).
-   * </pre>
-   *
-   * <code>.tensorflow.GraphOpCreation graph_op_creation = 7;</code>
-   */
-  public org.tensorflow.proto.util.GraphOpCreation getGraphOpCreation() {
-    if (whatCase_ == 7) {
-       return (org.tensorflow.proto.util.GraphOpCreation) what_;
-    }
-    return org.tensorflow.proto.util.GraphOpCreation.getDefaultInstance();
-  }
-  /**
-   * <pre>
-   * The creation of an op within a graph (e.g., a FuncGraph compiled from
-   * a Python function).
-   * </pre>
-   *
-   * <code>.tensorflow.GraphOpCreation graph_op_creation = 7;</code>
-   */
-  public org.tensorflow.proto.util.GraphOpCreationOrBuilder getGraphOpCreationOrBuilder() {
-    if (whatCase_ == 7) {
-       return (org.tensorflow.proto.util.GraphOpCreation) what_;
-    }
-    return org.tensorflow.proto.util.GraphOpCreation.getDefaultInstance();
-  }
-
-  public static final int DEBUGGED_GRAPH_FIELD_NUMBER = 8;
-  /**
-   * <pre>
-   * Information about a debugged graph.
-   * </pre>
-   *
-   * <code>.tensorflow.DebuggedGraph debugged_graph = 8;</code>
-   */
-  public boolean hasDebuggedGraph() {
-    return whatCase_ == 8;
-  }
-  /**
-   * <pre>
-   * Information about a debugged graph.
-   * </pre>
-   *
-   * <code>.tensorflow.DebuggedGraph debugged_graph = 8;</code>
-   */
-  public org.tensorflow.proto.util.DebuggedGraph getDebuggedGraph() {
-    if (whatCase_ == 8) {
-       return (org.tensorflow.proto.util.DebuggedGraph) what_;
-    }
-    return org.tensorflow.proto.util.DebuggedGraph.getDefaultInstance();
-  }
-  /**
-   * <pre>
-   * Information about a debugged graph.
-   * </pre>
-   *
-   * <code>.tensorflow.DebuggedGraph debugged_graph = 8;</code>
-   */
-  public org.tensorflow.proto.util.DebuggedGraphOrBuilder getDebuggedGraphOrBuilder() {
-    if (whatCase_ == 8) {
-       return (org.tensorflow.proto.util.DebuggedGraph) what_;
-    }
-    return org.tensorflow.proto.util.DebuggedGraph.getDefaultInstance();
-  }
-
-  public static final int EXECUTION_FIELD_NUMBER = 9;
-  /**
-   * <pre>
-   * Execution of an op or a Graph (e.g., a tf.function).
-   * </pre>
-   *
-   * <code>.tensorflow.Execution execution = 9;</code>
-   */
-  public boolean hasExecution() {
-    return whatCase_ == 9;
-  }
-  /**
-   * <pre>
-   * Execution of an op or a Graph (e.g., a tf.function).
-   * </pre>
-   *
-   * <code>.tensorflow.Execution execution = 9;</code>
-   */
-  public org.tensorflow.proto.util.Execution getExecution() {
-    if (whatCase_ == 9) {
-       return (org.tensorflow.proto.util.Execution) what_;
-    }
-    return org.tensorflow.proto.util.Execution.getDefaultInstance();
-  }
-  /**
-   * <pre>
-   * Execution of an op or a Graph (e.g., a tf.function).
-   * </pre>
-   *
-   * <code>.tensorflow.Execution execution = 9;</code>
-   */
-  public org.tensorflow.proto.util.ExecutionOrBuilder getExecutionOrBuilder() {
-    if (whatCase_ == 9) {
-       return (org.tensorflow.proto.util.Execution) what_;
-    }
-    return org.tensorflow.proto.util.Execution.getDefaultInstance();
-  }
-
-  public static final int GRAPH_EXECUTION_TRACE_FIELD_NUMBER = 10;
-  /**
-   * <pre>
-   * A graph execution trace: Contains information about the intermediate
-   * tensors computed during the graph execution.
-   * </pre>
-   *
-   * <code>.tensorflow.GraphExecutionTrace graph_execution_trace = 10;</code>
-   */
-  public boolean hasGraphExecutionTrace() {
-    return whatCase_ == 10;
-  }
-  /**
-   * <pre>
-   * A graph execution trace: Contains information about the intermediate
-   * tensors computed during the graph execution.
-   * </pre>
-   *
-   * <code>.tensorflow.GraphExecutionTrace graph_execution_trace = 10;</code>
-   */
-  public org.tensorflow.proto.util.GraphExecutionTrace getGraphExecutionTrace() {
-    if (whatCase_ == 10) {
-       return (org.tensorflow.proto.util.GraphExecutionTrace) what_;
-    }
-    return org.tensorflow.proto.util.GraphExecutionTrace.getDefaultInstance();
-  }
-  /**
-   * <pre>
-   * A graph execution trace: Contains information about the intermediate
-   * tensors computed during the graph execution.
-   * </pre>
-   *
-   * <code>.tensorflow.GraphExecutionTrace graph_execution_trace = 10;</code>
-   */
-  public org.tensorflow.proto.util.GraphExecutionTraceOrBuilder getGraphExecutionTraceOrBuilder() {
-    if (whatCase_ == 10) {
-       return (org.tensorflow.proto.util.GraphExecutionTrace) what_;
-    }
-    return org.tensorflow.proto.util.GraphExecutionTrace.getDefaultInstance();
-  }
-
-  public static final int GRAPH_ID_FIELD_NUMBER = 11;
-  /**
-   * <pre>
-   * The ID of the graph (i.e., FuncGraph) executed here: applicable only
-   * to the execution of a FuncGraph.
-   * </pre>
-   *
-   * <code>string graph_id = 11;</code>
-   */
-  public java.lang.String getGraphId() {
-    java.lang.Object ref = "";
-    if (whatCase_ == 11) {
-      ref = what_;
-    }
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      if (whatCase_ == 11) {
-        what_ = s;
-      }
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * The ID of the graph (i.e., FuncGraph) executed here: applicable only
-   * to the execution of a FuncGraph.
-   * </pre>
-   *
-   * <code>string graph_id = 11;</code>
-   */
-  public com.google.protobuf.ByteString
-      getGraphIdBytes() {
-    java.lang.Object ref = "";
-    if (whatCase_ == 11) {
-      ref = what_;
-    }
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      if (whatCase_ == 11) {
-        what_ = b;
-      }
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int DEBUGGED_DEVICE_FIELD_NUMBER = 12;
-  /**
-   * <pre>
-   * A device on which debugger-instrumented ops and/or tensors reside.
-   * </pre>
-   *
-   * <code>.tensorflow.DebuggedDevice debugged_device = 12;</code>
-   */
-  public boolean hasDebuggedDevice() {
-    return whatCase_ == 12;
-  }
-  /**
-   * <pre>
-   * A device on which debugger-instrumented ops and/or tensors reside.
-   * </pre>
-   *
-   * <code>.tensorflow.DebuggedDevice debugged_device = 12;</code>
-   */
-  public org.tensorflow.proto.util.DebuggedDevice getDebuggedDevice() {
-    if (whatCase_ == 12) {
-       return (org.tensorflow.proto.util.DebuggedDevice) what_;
-    }
-    return org.tensorflow.proto.util.DebuggedDevice.getDefaultInstance();
-  }
-  /**
-   * <pre>
-   * A device on which debugger-instrumented ops and/or tensors reside.
-   * </pre>
-   *
-   * <code>.tensorflow.DebuggedDevice debugged_device = 12;</code>
-   */
-  public org.tensorflow.proto.util.DebuggedDeviceOrBuilder getDebuggedDeviceOrBuilder() {
-    if (whatCase_ == 12) {
-       return (org.tensorflow.proto.util.DebuggedDevice) what_;
-    }
-    return org.tensorflow.proto.util.DebuggedDevice.getDefaultInstance();
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    if (wallTime_ != 0D) {
-      output.writeDouble(1, wallTime_);
-    }
-    if (step_ != 0L) {
-      output.writeInt64(2, step_);
-    }
-    if (whatCase_ == 3) {
-      output.writeMessage(3, (org.tensorflow.proto.util.DebugMetadata) what_);
-    }
-    if (whatCase_ == 4) {
-      output.writeMessage(4, (org.tensorflow.proto.util.SourceFile) what_);
-    }
-    if (whatCase_ == 6) {
-      output.writeMessage(6, (org.tensorflow.proto.util.StackFrameWithId) what_);
-    }
-    if (whatCase_ == 7) {
-      output.writeMessage(7, (org.tensorflow.proto.util.GraphOpCreation) what_);
-    }
-    if (whatCase_ == 8) {
-      output.writeMessage(8, (org.tensorflow.proto.util.DebuggedGraph) what_);
-    }
-    if (whatCase_ == 9) {
-      output.writeMessage(9, (org.tensorflow.proto.util.Execution) what_);
-    }
-    if (whatCase_ == 10) {
-      output.writeMessage(10, (org.tensorflow.proto.util.GraphExecutionTrace) what_);
-    }
-    if (whatCase_ == 11) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 11, what_);
-    }
-    if (whatCase_ == 12) {
-      output.writeMessage(12, (org.tensorflow.proto.util.DebuggedDevice) what_);
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (wallTime_ != 0D) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeDoubleSize(1, wallTime_);
-    }
-    if (step_ != 0L) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt64Size(2, step_);
-    }
-    if (whatCase_ == 3) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(3, (org.tensorflow.proto.util.DebugMetadata) what_);
-    }
-    if (whatCase_ == 4) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(4, (org.tensorflow.proto.util.SourceFile) what_);
-    }
-    if (whatCase_ == 6) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(6, (org.tensorflow.proto.util.StackFrameWithId) what_);
-    }
-    if (whatCase_ == 7) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(7, (org.tensorflow.proto.util.GraphOpCreation) what_);
-    }
-    if (whatCase_ == 8) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(8, (org.tensorflow.proto.util.DebuggedGraph) what_);
-    }
-    if (whatCase_ == 9) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(9, (org.tensorflow.proto.util.Execution) what_);
-    }
-    if (whatCase_ == 10) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(10, (org.tensorflow.proto.util.GraphExecutionTrace) what_);
-    }
-    if (whatCase_ == 11) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(11, what_);
-    }
-    if (whatCase_ == 12) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(12, (org.tensorflow.proto.util.DebuggedDevice) what_);
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.util.DebugEvent)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.util.DebugEvent other = (org.tensorflow.proto.util.DebugEvent) obj;
-
-    if (java.lang.Double.doubleToLongBits(getWallTime())
-        != java.lang.Double.doubleToLongBits(
-            other.getWallTime())) return false;
-    if (getStep()
-        != other.getStep()) return false;
-    if (!getWhatCase().equals(other.getWhatCase())) return false;
-    switch (whatCase_) {
-      case 3:
-        if (!getDebugMetadata()
-            .equals(other.getDebugMetadata())) return false;
-        break;
-      case 4:
-        if (!getSourceFile()
-            .equals(other.getSourceFile())) return false;
-        break;
-      case 6:
-        if (!getStackFrameWithId()
-            .equals(other.getStackFrameWithId())) return false;
-        break;
-      case 7:
-        if (!getGraphOpCreation()
-            .equals(other.getGraphOpCreation())) return false;
-        break;
-      case 8:
-        if (!getDebuggedGraph()
-            .equals(other.getDebuggedGraph())) return false;
-        break;
-      case 9:
-        if (!getExecution()
-            .equals(other.getExecution())) return false;
-        break;
-      case 10:
-        if (!getGraphExecutionTrace()
-            .equals(other.getGraphExecutionTrace())) return false;
-        break;
-      case 11:
-        if (!getGraphId()
-            .equals(other.getGraphId())) return false;
-        break;
-      case 12:
-        if (!getDebuggedDevice()
-            .equals(other.getDebuggedDevice())) return false;
-        break;
-      case 0:
-      default:
-    }
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    hash = (37 * hash) + WALL_TIME_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-        java.lang.Double.doubleToLongBits(getWallTime()));
-    hash = (37 * hash) + STEP_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-        getStep());
-    switch (whatCase_) {
-      case 3:
-        hash = (37 * hash) + DEBUG_METADATA_FIELD_NUMBER;
-        hash = (53 * hash) + getDebugMetadata().hashCode();
-        break;
-      case 4:
-        hash = (37 * hash) + SOURCE_FILE_FIELD_NUMBER;
-        hash = (53 * hash) + getSourceFile().hashCode();
-        break;
-      case 6:
-        hash = (37 * hash) + STACK_FRAME_WITH_ID_FIELD_NUMBER;
-        hash = (53 * hash) + getStackFrameWithId().hashCode();
-        break;
-      case 7:
-        hash = (37 * hash) + GRAPH_OP_CREATION_FIELD_NUMBER;
-        hash = (53 * hash) + getGraphOpCreation().hashCode();
-        break;
-      case 8:
-        hash = (37 * hash) + DEBUGGED_GRAPH_FIELD_NUMBER;
-        hash = (53 * hash) + getDebuggedGraph().hashCode();
-        break;
-      case 9:
-        hash = (37 * hash) + EXECUTION_FIELD_NUMBER;
-        hash = (53 * hash) + getExecution().hashCode();
-        break;
-      case 10:
-        hash = (37 * hash) + GRAPH_EXECUTION_TRACE_FIELD_NUMBER;
-        hash = (53 * hash) + getGraphExecutionTrace().hashCode();
-        break;
-      case 11:
-        hash = (37 * hash) + GRAPH_ID_FIELD_NUMBER;
-        hash = (53 * hash) + getGraphId().hashCode();
-        break;
-      case 12:
-        hash = (37 * hash) + DEBUGGED_DEVICE_FIELD_NUMBER;
-        hash = (53 * hash) + getDebuggedDevice().hashCode();
-        break;
-      case 0:
-      default:
-    }
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.util.DebugEvent parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.DebugEvent parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.DebugEvent parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.DebugEvent parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.DebugEvent parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.DebugEvent parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.DebugEvent parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.DebugEvent parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.DebugEvent parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.DebugEvent parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.DebugEvent parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.DebugEvent parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.util.DebugEvent prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * <pre>
-   * An Event related to the debugging of a TensorFlow program.
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.DebugEvent}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.DebugEvent)
-      org.tensorflow.proto.util.DebugEventOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.util.DebugEventProtos.internal_static_tensorflow_DebugEvent_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.util.DebugEventProtos.internal_static_tensorflow_DebugEvent_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.util.DebugEvent.class, org.tensorflow.proto.util.DebugEvent.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.util.DebugEvent.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      wallTime_ = 0D;
-
-      step_ = 0L;
-
-      whatCase_ = 0;
-      what_ = null;
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.util.DebugEventProtos.internal_static_tensorflow_DebugEvent_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.DebugEvent getDefaultInstanceForType() {
-      return org.tensorflow.proto.util.DebugEvent.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.DebugEvent build() {
-      org.tensorflow.proto.util.DebugEvent result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.DebugEvent buildPartial() {
-      org.tensorflow.proto.util.DebugEvent result = new org.tensorflow.proto.util.DebugEvent(this);
-      result.wallTime_ = wallTime_;
-      result.step_ = step_;
-      if (whatCase_ == 3) {
-        if (debugMetadataBuilder_ == null) {
-          result.what_ = what_;
-        } else {
-          result.what_ = debugMetadataBuilder_.build();
-        }
-      }
-      if (whatCase_ == 4) {
-        if (sourceFileBuilder_ == null) {
-          result.what_ = what_;
-        } else {
-          result.what_ = sourceFileBuilder_.build();
-        }
-      }
-      if (whatCase_ == 6) {
-        if (stackFrameWithIdBuilder_ == null) {
-          result.what_ = what_;
-        } else {
-          result.what_ = stackFrameWithIdBuilder_.build();
-        }
-      }
-      if (whatCase_ == 7) {
-        if (graphOpCreationBuilder_ == null) {
-          result.what_ = what_;
-        } else {
-          result.what_ = graphOpCreationBuilder_.build();
-        }
-      }
-      if (whatCase_ == 8) {
-        if (debuggedGraphBuilder_ == null) {
-          result.what_ = what_;
-        } else {
-          result.what_ = debuggedGraphBuilder_.build();
-        }
-      }
-      if (whatCase_ == 9) {
-        if (executionBuilder_ == null) {
-          result.what_ = what_;
-        } else {
-          result.what_ = executionBuilder_.build();
-        }
-      }
-      if (whatCase_ == 10) {
-        if (graphExecutionTraceBuilder_ == null) {
-          result.what_ = what_;
-        } else {
-          result.what_ = graphExecutionTraceBuilder_.build();
-        }
-      }
-      if (whatCase_ == 11) {
-        result.what_ = what_;
-      }
-      if (whatCase_ == 12) {
-        if (debuggedDeviceBuilder_ == null) {
-          result.what_ = what_;
-        } else {
-          result.what_ = debuggedDeviceBuilder_.build();
-        }
-      }
-      result.whatCase_ = whatCase_;
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.util.DebugEvent) {
-        return mergeFrom((org.tensorflow.proto.util.DebugEvent)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.util.DebugEvent other) {
-      if (other == org.tensorflow.proto.util.DebugEvent.getDefaultInstance()) return this;
-      if (other.getWallTime() != 0D) {
-        setWallTime(other.getWallTime());
-      }
-      if (other.getStep() != 0L) {
-        setStep(other.getStep());
-      }
-      switch (other.getWhatCase()) {
-        case DEBUG_METADATA: {
-          mergeDebugMetadata(other.getDebugMetadata());
-          break;
-        }
-        case SOURCE_FILE: {
-          mergeSourceFile(other.getSourceFile());
-          break;
-        }
-        case STACK_FRAME_WITH_ID: {
-          mergeStackFrameWithId(other.getStackFrameWithId());
-          break;
-        }
-        case GRAPH_OP_CREATION: {
-          mergeGraphOpCreation(other.getGraphOpCreation());
-          break;
-        }
-        case DEBUGGED_GRAPH: {
-          mergeDebuggedGraph(other.getDebuggedGraph());
-          break;
-        }
-        case EXECUTION: {
-          mergeExecution(other.getExecution());
-          break;
-        }
-        case GRAPH_EXECUTION_TRACE: {
-          mergeGraphExecutionTrace(other.getGraphExecutionTrace());
-          break;
-        }
-        case GRAPH_ID: {
-          whatCase_ = 11;
-          what_ = other.what_;
-          onChanged();
-          break;
-        }
-        case DEBUGGED_DEVICE: {
-          mergeDebuggedDevice(other.getDebuggedDevice());
-          break;
-        }
-        case WHAT_NOT_SET: {
-          break;
-        }
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.util.DebugEvent parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.util.DebugEvent) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-    private int whatCase_ = 0;
-    private java.lang.Object what_;
-    public WhatCase
-        getWhatCase() {
-      return WhatCase.forNumber(
-          whatCase_);
-    }
-
-    public Builder clearWhat() {
-      whatCase_ = 0;
-      what_ = null;
-      onChanged();
-      return this;
-    }
-
-
-    private double wallTime_ ;
-    /**
-     * <pre>
-     * Timestamp in seconds (with microsecond precision).
-     * </pre>
-     *
-     * <code>double wall_time = 1;</code>
-     */
-    public double getWallTime() {
-      return wallTime_;
-    }
-    /**
-     * <pre>
-     * Timestamp in seconds (with microsecond precision).
-     * </pre>
-     *
-     * <code>double wall_time = 1;</code>
-     */
-    public Builder setWallTime(double value) {
-      
-      wallTime_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Timestamp in seconds (with microsecond precision).
-     * </pre>
-     *
-     * <code>double wall_time = 1;</code>
-     */
-    public Builder clearWallTime() {
-      
-      wallTime_ = 0D;
-      onChanged();
-      return this;
-    }
-
-    private long step_ ;
-    /**
-     * <pre>
-     * Step of training (if available).
-     * </pre>
-     *
-     * <code>int64 step = 2;</code>
-     */
-    public long getStep() {
-      return step_;
-    }
-    /**
-     * <pre>
-     * Step of training (if available).
-     * </pre>
-     *
-     * <code>int64 step = 2;</code>
-     */
-    public Builder setStep(long value) {
-      
-      step_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Step of training (if available).
-     * </pre>
-     *
-     * <code>int64 step = 2;</code>
-     */
-    public Builder clearStep() {
-      
-      step_ = 0L;
-      onChanged();
-      return this;
-    }
-
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.util.DebugMetadata, org.tensorflow.proto.util.DebugMetadata.Builder, org.tensorflow.proto.util.DebugMetadataOrBuilder> debugMetadataBuilder_;
-    /**
-     * <pre>
-     * Metadata related to this debugging data.
-     * </pre>
-     *
-     * <code>.tensorflow.DebugMetadata debug_metadata = 3;</code>
-     */
-    public boolean hasDebugMetadata() {
-      return whatCase_ == 3;
-    }
-    /**
-     * <pre>
-     * Metadata related to this debugging data.
-     * </pre>
-     *
-     * <code>.tensorflow.DebugMetadata debug_metadata = 3;</code>
-     */
-    public org.tensorflow.proto.util.DebugMetadata getDebugMetadata() {
-      if (debugMetadataBuilder_ == null) {
-        if (whatCase_ == 3) {
-          return (org.tensorflow.proto.util.DebugMetadata) what_;
-        }
-        return org.tensorflow.proto.util.DebugMetadata.getDefaultInstance();
-      } else {
-        if (whatCase_ == 3) {
-          return debugMetadataBuilder_.getMessage();
-        }
-        return org.tensorflow.proto.util.DebugMetadata.getDefaultInstance();
-      }
-    }
-    /**
-     * <pre>
-     * Metadata related to this debugging data.
-     * </pre>
-     *
-     * <code>.tensorflow.DebugMetadata debug_metadata = 3;</code>
-     */
-    public Builder setDebugMetadata(org.tensorflow.proto.util.DebugMetadata value) {
-      if (debugMetadataBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        what_ = value;
-        onChanged();
-      } else {
-        debugMetadataBuilder_.setMessage(value);
-      }
-      whatCase_ = 3;
-      return this;
-    }
-    /**
-     * <pre>
-     * Metadata related to this debugging data.
-     * </pre>
-     *
-     * <code>.tensorflow.DebugMetadata debug_metadata = 3;</code>
-     */
-    public Builder setDebugMetadata(
-        org.tensorflow.proto.util.DebugMetadata.Builder builderForValue) {
-      if (debugMetadataBuilder_ == null) {
-        what_ = builderForValue.build();
-        onChanged();
-      } else {
-        debugMetadataBuilder_.setMessage(builderForValue.build());
-      }
-      whatCase_ = 3;
-      return this;
-    }
-    /**
-     * <pre>
-     * Metadata related to this debugging data.
-     * </pre>
-     *
-     * <code>.tensorflow.DebugMetadata debug_metadata = 3;</code>
-     */
-    public Builder mergeDebugMetadata(org.tensorflow.proto.util.DebugMetadata value) {
-      if (debugMetadataBuilder_ == null) {
-        if (whatCase_ == 3 &&
-            what_ != org.tensorflow.proto.util.DebugMetadata.getDefaultInstance()) {
-          what_ = org.tensorflow.proto.util.DebugMetadata.newBuilder((org.tensorflow.proto.util.DebugMetadata) what_)
-              .mergeFrom(value).buildPartial();
-        } else {
-          what_ = value;
-        }
-        onChanged();
-      } else {
-        if (whatCase_ == 3) {
-          debugMetadataBuilder_.mergeFrom(value);
-        }
-        debugMetadataBuilder_.setMessage(value);
-      }
-      whatCase_ = 3;
-      return this;
-    }
-    /**
-     * <pre>
-     * Metadata related to this debugging data.
-     * </pre>
-     *
-     * <code>.tensorflow.DebugMetadata debug_metadata = 3;</code>
-     */
-    public Builder clearDebugMetadata() {
-      if (debugMetadataBuilder_ == null) {
-        if (whatCase_ == 3) {
-          whatCase_ = 0;
-          what_ = null;
-          onChanged();
-        }
-      } else {
-        if (whatCase_ == 3) {
-          whatCase_ = 0;
-          what_ = null;
-        }
-        debugMetadataBuilder_.clear();
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Metadata related to this debugging data.
-     * </pre>
-     *
-     * <code>.tensorflow.DebugMetadata debug_metadata = 3;</code>
-     */
-    public org.tensorflow.proto.util.DebugMetadata.Builder getDebugMetadataBuilder() {
-      return getDebugMetadataFieldBuilder().getBuilder();
-    }
-    /**
-     * <pre>
-     * Metadata related to this debugging data.
-     * </pre>
-     *
-     * <code>.tensorflow.DebugMetadata debug_metadata = 3;</code>
-     */
-    public org.tensorflow.proto.util.DebugMetadataOrBuilder getDebugMetadataOrBuilder() {
-      if ((whatCase_ == 3) && (debugMetadataBuilder_ != null)) {
-        return debugMetadataBuilder_.getMessageOrBuilder();
-      } else {
-        if (whatCase_ == 3) {
-          return (org.tensorflow.proto.util.DebugMetadata) what_;
-        }
-        return org.tensorflow.proto.util.DebugMetadata.getDefaultInstance();
-      }
-    }
-    /**
-     * <pre>
-     * Metadata related to this debugging data.
-     * </pre>
-     *
-     * <code>.tensorflow.DebugMetadata debug_metadata = 3;</code>
-     */
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.util.DebugMetadata, org.tensorflow.proto.util.DebugMetadata.Builder, org.tensorflow.proto.util.DebugMetadataOrBuilder> 
-        getDebugMetadataFieldBuilder() {
-      if (debugMetadataBuilder_ == null) {
-        if (!(whatCase_ == 3)) {
-          what_ = org.tensorflow.proto.util.DebugMetadata.getDefaultInstance();
-        }
-        debugMetadataBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-            org.tensorflow.proto.util.DebugMetadata, org.tensorflow.proto.util.DebugMetadata.Builder, org.tensorflow.proto.util.DebugMetadataOrBuilder>(
-                (org.tensorflow.proto.util.DebugMetadata) what_,
-                getParentForChildren(),
-                isClean());
-        what_ = null;
-      }
-      whatCase_ = 3;
-      onChanged();;
-      return debugMetadataBuilder_;
-    }
-
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.util.SourceFile, org.tensorflow.proto.util.SourceFile.Builder, org.tensorflow.proto.util.SourceFileOrBuilder> sourceFileBuilder_;
-    /**
-     * <pre>
-     * The content of a source file.
-     * </pre>
-     *
-     * <code>.tensorflow.SourceFile source_file = 4;</code>
-     */
-    public boolean hasSourceFile() {
-      return whatCase_ == 4;
-    }
-    /**
-     * <pre>
-     * The content of a source file.
-     * </pre>
-     *
-     * <code>.tensorflow.SourceFile source_file = 4;</code>
-     */
-    public org.tensorflow.proto.util.SourceFile getSourceFile() {
-      if (sourceFileBuilder_ == null) {
-        if (whatCase_ == 4) {
-          return (org.tensorflow.proto.util.SourceFile) what_;
-        }
-        return org.tensorflow.proto.util.SourceFile.getDefaultInstance();
-      } else {
-        if (whatCase_ == 4) {
-          return sourceFileBuilder_.getMessage();
-        }
-        return org.tensorflow.proto.util.SourceFile.getDefaultInstance();
-      }
-    }
-    /**
-     * <pre>
-     * The content of a source file.
-     * </pre>
-     *
-     * <code>.tensorflow.SourceFile source_file = 4;</code>
-     */
-    public Builder setSourceFile(org.tensorflow.proto.util.SourceFile value) {
-      if (sourceFileBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        what_ = value;
-        onChanged();
-      } else {
-        sourceFileBuilder_.setMessage(value);
-      }
-      whatCase_ = 4;
-      return this;
-    }
-    /**
-     * <pre>
-     * The content of a source file.
-     * </pre>
-     *
-     * <code>.tensorflow.SourceFile source_file = 4;</code>
-     */
-    public Builder setSourceFile(
-        org.tensorflow.proto.util.SourceFile.Builder builderForValue) {
-      if (sourceFileBuilder_ == null) {
-        what_ = builderForValue.build();
-        onChanged();
-      } else {
-        sourceFileBuilder_.setMessage(builderForValue.build());
-      }
-      whatCase_ = 4;
-      return this;
-    }
-    /**
-     * <pre>
-     * The content of a source file.
-     * </pre>
-     *
-     * <code>.tensorflow.SourceFile source_file = 4;</code>
-     */
-    public Builder mergeSourceFile(org.tensorflow.proto.util.SourceFile value) {
-      if (sourceFileBuilder_ == null) {
-        if (whatCase_ == 4 &&
-            what_ != org.tensorflow.proto.util.SourceFile.getDefaultInstance()) {
-          what_ = org.tensorflow.proto.util.SourceFile.newBuilder((org.tensorflow.proto.util.SourceFile) what_)
-              .mergeFrom(value).buildPartial();
-        } else {
-          what_ = value;
-        }
-        onChanged();
-      } else {
-        if (whatCase_ == 4) {
-          sourceFileBuilder_.mergeFrom(value);
-        }
-        sourceFileBuilder_.setMessage(value);
-      }
-      whatCase_ = 4;
-      return this;
-    }
-    /**
-     * <pre>
-     * The content of a source file.
-     * </pre>
-     *
-     * <code>.tensorflow.SourceFile source_file = 4;</code>
-     */
-    public Builder clearSourceFile() {
-      if (sourceFileBuilder_ == null) {
-        if (whatCase_ == 4) {
-          whatCase_ = 0;
-          what_ = null;
-          onChanged();
-        }
-      } else {
-        if (whatCase_ == 4) {
-          whatCase_ = 0;
-          what_ = null;
-        }
-        sourceFileBuilder_.clear();
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * The content of a source file.
-     * </pre>
-     *
-     * <code>.tensorflow.SourceFile source_file = 4;</code>
-     */
-    public org.tensorflow.proto.util.SourceFile.Builder getSourceFileBuilder() {
-      return getSourceFileFieldBuilder().getBuilder();
-    }
-    /**
-     * <pre>
-     * The content of a source file.
-     * </pre>
-     *
-     * <code>.tensorflow.SourceFile source_file = 4;</code>
-     */
-    public org.tensorflow.proto.util.SourceFileOrBuilder getSourceFileOrBuilder() {
-      if ((whatCase_ == 4) && (sourceFileBuilder_ != null)) {
-        return sourceFileBuilder_.getMessageOrBuilder();
-      } else {
-        if (whatCase_ == 4) {
-          return (org.tensorflow.proto.util.SourceFile) what_;
-        }
-        return org.tensorflow.proto.util.SourceFile.getDefaultInstance();
-      }
-    }
-    /**
-     * <pre>
-     * The content of a source file.
-     * </pre>
-     *
-     * <code>.tensorflow.SourceFile source_file = 4;</code>
-     */
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.util.SourceFile, org.tensorflow.proto.util.SourceFile.Builder, org.tensorflow.proto.util.SourceFileOrBuilder> 
-        getSourceFileFieldBuilder() {
-      if (sourceFileBuilder_ == null) {
-        if (!(whatCase_ == 4)) {
-          what_ = org.tensorflow.proto.util.SourceFile.getDefaultInstance();
-        }
-        sourceFileBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-            org.tensorflow.proto.util.SourceFile, org.tensorflow.proto.util.SourceFile.Builder, org.tensorflow.proto.util.SourceFileOrBuilder>(
-                (org.tensorflow.proto.util.SourceFile) what_,
-                getParentForChildren(),
-                isClean());
-        what_ = null;
-      }
-      whatCase_ = 4;
-      onChanged();;
-      return sourceFileBuilder_;
-    }
-
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.util.StackFrameWithId, org.tensorflow.proto.util.StackFrameWithId.Builder, org.tensorflow.proto.util.StackFrameWithIdOrBuilder> stackFrameWithIdBuilder_;
-    /**
-     * <pre>
-     * A stack frame (filename, line number and column number, function name and
-     * code string) with ID.
-     * </pre>
-     *
-     * <code>.tensorflow.StackFrameWithId stack_frame_with_id = 6;</code>
-     */
-    public boolean hasStackFrameWithId() {
-      return whatCase_ == 6;
-    }
-    /**
-     * <pre>
-     * A stack frame (filename, line number and column number, function name and
-     * code string) with ID.
-     * </pre>
-     *
-     * <code>.tensorflow.StackFrameWithId stack_frame_with_id = 6;</code>
-     */
-    public org.tensorflow.proto.util.StackFrameWithId getStackFrameWithId() {
-      if (stackFrameWithIdBuilder_ == null) {
-        if (whatCase_ == 6) {
-          return (org.tensorflow.proto.util.StackFrameWithId) what_;
-        }
-        return org.tensorflow.proto.util.StackFrameWithId.getDefaultInstance();
-      } else {
-        if (whatCase_ == 6) {
-          return stackFrameWithIdBuilder_.getMessage();
-        }
-        return org.tensorflow.proto.util.StackFrameWithId.getDefaultInstance();
-      }
-    }
-    /**
-     * <pre>
-     * A stack frame (filename, line number and column number, function name and
-     * code string) with ID.
-     * </pre>
-     *
-     * <code>.tensorflow.StackFrameWithId stack_frame_with_id = 6;</code>
-     */
-    public Builder setStackFrameWithId(org.tensorflow.proto.util.StackFrameWithId value) {
-      if (stackFrameWithIdBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        what_ = value;
-        onChanged();
-      } else {
-        stackFrameWithIdBuilder_.setMessage(value);
-      }
-      whatCase_ = 6;
-      return this;
-    }
-    /**
-     * <pre>
-     * A stack frame (filename, line number and column number, function name and
-     * code string) with ID.
-     * </pre>
-     *
-     * <code>.tensorflow.StackFrameWithId stack_frame_with_id = 6;</code>
-     */
-    public Builder setStackFrameWithId(
-        org.tensorflow.proto.util.StackFrameWithId.Builder builderForValue) {
-      if (stackFrameWithIdBuilder_ == null) {
-        what_ = builderForValue.build();
-        onChanged();
-      } else {
-        stackFrameWithIdBuilder_.setMessage(builderForValue.build());
-      }
-      whatCase_ = 6;
-      return this;
-    }
-    /**
-     * <pre>
-     * A stack frame (filename, line number and column number, function name and
-     * code string) with ID.
-     * </pre>
-     *
-     * <code>.tensorflow.StackFrameWithId stack_frame_with_id = 6;</code>
-     */
-    public Builder mergeStackFrameWithId(org.tensorflow.proto.util.StackFrameWithId value) {
-      if (stackFrameWithIdBuilder_ == null) {
-        if (whatCase_ == 6 &&
-            what_ != org.tensorflow.proto.util.StackFrameWithId.getDefaultInstance()) {
-          what_ = org.tensorflow.proto.util.StackFrameWithId.newBuilder((org.tensorflow.proto.util.StackFrameWithId) what_)
-              .mergeFrom(value).buildPartial();
-        } else {
-          what_ = value;
-        }
-        onChanged();
-      } else {
-        if (whatCase_ == 6) {
-          stackFrameWithIdBuilder_.mergeFrom(value);
-        }
-        stackFrameWithIdBuilder_.setMessage(value);
-      }
-      whatCase_ = 6;
-      return this;
-    }
-    /**
-     * <pre>
-     * A stack frame (filename, line number and column number, function name and
-     * code string) with ID.
-     * </pre>
-     *
-     * <code>.tensorflow.StackFrameWithId stack_frame_with_id = 6;</code>
-     */
-    public Builder clearStackFrameWithId() {
-      if (stackFrameWithIdBuilder_ == null) {
-        if (whatCase_ == 6) {
-          whatCase_ = 0;
-          what_ = null;
-          onChanged();
-        }
-      } else {
-        if (whatCase_ == 6) {
-          whatCase_ = 0;
-          what_ = null;
-        }
-        stackFrameWithIdBuilder_.clear();
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * A stack frame (filename, line number and column number, function name and
-     * code string) with ID.
-     * </pre>
-     *
-     * <code>.tensorflow.StackFrameWithId stack_frame_with_id = 6;</code>
-     */
-    public org.tensorflow.proto.util.StackFrameWithId.Builder getStackFrameWithIdBuilder() {
-      return getStackFrameWithIdFieldBuilder().getBuilder();
-    }
-    /**
-     * <pre>
-     * A stack frame (filename, line number and column number, function name and
-     * code string) with ID.
-     * </pre>
-     *
-     * <code>.tensorflow.StackFrameWithId stack_frame_with_id = 6;</code>
-     */
-    public org.tensorflow.proto.util.StackFrameWithIdOrBuilder getStackFrameWithIdOrBuilder() {
-      if ((whatCase_ == 6) && (stackFrameWithIdBuilder_ != null)) {
-        return stackFrameWithIdBuilder_.getMessageOrBuilder();
-      } else {
-        if (whatCase_ == 6) {
-          return (org.tensorflow.proto.util.StackFrameWithId) what_;
-        }
-        return org.tensorflow.proto.util.StackFrameWithId.getDefaultInstance();
-      }
-    }
-    /**
-     * <pre>
-     * A stack frame (filename, line number and column number, function name and
-     * code string) with ID.
-     * </pre>
-     *
-     * <code>.tensorflow.StackFrameWithId stack_frame_with_id = 6;</code>
-     */
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.util.StackFrameWithId, org.tensorflow.proto.util.StackFrameWithId.Builder, org.tensorflow.proto.util.StackFrameWithIdOrBuilder> 
-        getStackFrameWithIdFieldBuilder() {
-      if (stackFrameWithIdBuilder_ == null) {
-        if (!(whatCase_ == 6)) {
-          what_ = org.tensorflow.proto.util.StackFrameWithId.getDefaultInstance();
-        }
-        stackFrameWithIdBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-            org.tensorflow.proto.util.StackFrameWithId, org.tensorflow.proto.util.StackFrameWithId.Builder, org.tensorflow.proto.util.StackFrameWithIdOrBuilder>(
-                (org.tensorflow.proto.util.StackFrameWithId) what_,
-                getParentForChildren(),
-                isClean());
-        what_ = null;
-      }
-      whatCase_ = 6;
-      onChanged();;
-      return stackFrameWithIdBuilder_;
-    }
-
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.util.GraphOpCreation, org.tensorflow.proto.util.GraphOpCreation.Builder, org.tensorflow.proto.util.GraphOpCreationOrBuilder> graphOpCreationBuilder_;
-    /**
-     * <pre>
-     * The creation of an op within a graph (e.g., a FuncGraph compiled from
-     * a Python function).
-     * </pre>
-     *
-     * <code>.tensorflow.GraphOpCreation graph_op_creation = 7;</code>
-     */
-    public boolean hasGraphOpCreation() {
-      return whatCase_ == 7;
-    }
-    /**
-     * <pre>
-     * The creation of an op within a graph (e.g., a FuncGraph compiled from
-     * a Python function).
-     * </pre>
-     *
-     * <code>.tensorflow.GraphOpCreation graph_op_creation = 7;</code>
-     */
-    public org.tensorflow.proto.util.GraphOpCreation getGraphOpCreation() {
-      if (graphOpCreationBuilder_ == null) {
-        if (whatCase_ == 7) {
-          return (org.tensorflow.proto.util.GraphOpCreation) what_;
-        }
-        return org.tensorflow.proto.util.GraphOpCreation.getDefaultInstance();
-      } else {
-        if (whatCase_ == 7) {
-          return graphOpCreationBuilder_.getMessage();
-        }
-        return org.tensorflow.proto.util.GraphOpCreation.getDefaultInstance();
-      }
-    }
-    /**
-     * <pre>
-     * The creation of an op within a graph (e.g., a FuncGraph compiled from
-     * a Python function).
-     * </pre>
-     *
-     * <code>.tensorflow.GraphOpCreation graph_op_creation = 7;</code>
-     */
-    public Builder setGraphOpCreation(org.tensorflow.proto.util.GraphOpCreation value) {
-      if (graphOpCreationBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        what_ = value;
-        onChanged();
-      } else {
-        graphOpCreationBuilder_.setMessage(value);
-      }
-      whatCase_ = 7;
-      return this;
-    }
-    /**
-     * <pre>
-     * The creation of an op within a graph (e.g., a FuncGraph compiled from
-     * a Python function).
-     * </pre>
-     *
-     * <code>.tensorflow.GraphOpCreation graph_op_creation = 7;</code>
-     */
-    public Builder setGraphOpCreation(
-        org.tensorflow.proto.util.GraphOpCreation.Builder builderForValue) {
-      if (graphOpCreationBuilder_ == null) {
-        what_ = builderForValue.build();
-        onChanged();
-      } else {
-        graphOpCreationBuilder_.setMessage(builderForValue.build());
-      }
-      whatCase_ = 7;
-      return this;
-    }
-    /**
-     * <pre>
-     * The creation of an op within a graph (e.g., a FuncGraph compiled from
-     * a Python function).
-     * </pre>
-     *
-     * <code>.tensorflow.GraphOpCreation graph_op_creation = 7;</code>
-     */
-    public Builder mergeGraphOpCreation(org.tensorflow.proto.util.GraphOpCreation value) {
-      if (graphOpCreationBuilder_ == null) {
-        if (whatCase_ == 7 &&
-            what_ != org.tensorflow.proto.util.GraphOpCreation.getDefaultInstance()) {
-          what_ = org.tensorflow.proto.util.GraphOpCreation.newBuilder((org.tensorflow.proto.util.GraphOpCreation) what_)
-              .mergeFrom(value).buildPartial();
-        } else {
-          what_ = value;
-        }
-        onChanged();
-      } else {
-        if (whatCase_ == 7) {
-          graphOpCreationBuilder_.mergeFrom(value);
-        }
-        graphOpCreationBuilder_.setMessage(value);
-      }
-      whatCase_ = 7;
-      return this;
-    }
-    /**
-     * <pre>
-     * The creation of an op within a graph (e.g., a FuncGraph compiled from
-     * a Python function).
-     * </pre>
-     *
-     * <code>.tensorflow.GraphOpCreation graph_op_creation = 7;</code>
-     */
-    public Builder clearGraphOpCreation() {
-      if (graphOpCreationBuilder_ == null) {
-        if (whatCase_ == 7) {
-          whatCase_ = 0;
-          what_ = null;
-          onChanged();
-        }
-      } else {
-        if (whatCase_ == 7) {
-          whatCase_ = 0;
-          what_ = null;
-        }
-        graphOpCreationBuilder_.clear();
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * The creation of an op within a graph (e.g., a FuncGraph compiled from
-     * a Python function).
-     * </pre>
-     *
-     * <code>.tensorflow.GraphOpCreation graph_op_creation = 7;</code>
-     */
-    public org.tensorflow.proto.util.GraphOpCreation.Builder getGraphOpCreationBuilder() {
-      return getGraphOpCreationFieldBuilder().getBuilder();
-    }
-    /**
-     * <pre>
-     * The creation of an op within a graph (e.g., a FuncGraph compiled from
-     * a Python function).
-     * </pre>
-     *
-     * <code>.tensorflow.GraphOpCreation graph_op_creation = 7;</code>
-     */
-    public org.tensorflow.proto.util.GraphOpCreationOrBuilder getGraphOpCreationOrBuilder() {
-      if ((whatCase_ == 7) && (graphOpCreationBuilder_ != null)) {
-        return graphOpCreationBuilder_.getMessageOrBuilder();
-      } else {
-        if (whatCase_ == 7) {
-          return (org.tensorflow.proto.util.GraphOpCreation) what_;
-        }
-        return org.tensorflow.proto.util.GraphOpCreation.getDefaultInstance();
-      }
-    }
-    /**
-     * <pre>
-     * The creation of an op within a graph (e.g., a FuncGraph compiled from
-     * a Python function).
-     * </pre>
-     *
-     * <code>.tensorflow.GraphOpCreation graph_op_creation = 7;</code>
-     */
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.util.GraphOpCreation, org.tensorflow.proto.util.GraphOpCreation.Builder, org.tensorflow.proto.util.GraphOpCreationOrBuilder> 
-        getGraphOpCreationFieldBuilder() {
-      if (graphOpCreationBuilder_ == null) {
-        if (!(whatCase_ == 7)) {
-          what_ = org.tensorflow.proto.util.GraphOpCreation.getDefaultInstance();
-        }
-        graphOpCreationBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-            org.tensorflow.proto.util.GraphOpCreation, org.tensorflow.proto.util.GraphOpCreation.Builder, org.tensorflow.proto.util.GraphOpCreationOrBuilder>(
-                (org.tensorflow.proto.util.GraphOpCreation) what_,
-                getParentForChildren(),
-                isClean());
-        what_ = null;
-      }
-      whatCase_ = 7;
-      onChanged();;
-      return graphOpCreationBuilder_;
-    }
-
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.util.DebuggedGraph, org.tensorflow.proto.util.DebuggedGraph.Builder, org.tensorflow.proto.util.DebuggedGraphOrBuilder> debuggedGraphBuilder_;
-    /**
-     * <pre>
-     * Information about a debugged graph.
-     * </pre>
-     *
-     * <code>.tensorflow.DebuggedGraph debugged_graph = 8;</code>
-     */
-    public boolean hasDebuggedGraph() {
-      return whatCase_ == 8;
-    }
-    /**
-     * <pre>
-     * Information about a debugged graph.
-     * </pre>
-     *
-     * <code>.tensorflow.DebuggedGraph debugged_graph = 8;</code>
-     */
-    public org.tensorflow.proto.util.DebuggedGraph getDebuggedGraph() {
-      if (debuggedGraphBuilder_ == null) {
-        if (whatCase_ == 8) {
-          return (org.tensorflow.proto.util.DebuggedGraph) what_;
-        }
-        return org.tensorflow.proto.util.DebuggedGraph.getDefaultInstance();
-      } else {
-        if (whatCase_ == 8) {
-          return debuggedGraphBuilder_.getMessage();
-        }
-        return org.tensorflow.proto.util.DebuggedGraph.getDefaultInstance();
-      }
-    }
-    /**
-     * <pre>
-     * Information about a debugged graph.
-     * </pre>
-     *
-     * <code>.tensorflow.DebuggedGraph debugged_graph = 8;</code>
-     */
-    public Builder setDebuggedGraph(org.tensorflow.proto.util.DebuggedGraph value) {
-      if (debuggedGraphBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        what_ = value;
-        onChanged();
-      } else {
-        debuggedGraphBuilder_.setMessage(value);
-      }
-      whatCase_ = 8;
-      return this;
-    }
-    /**
-     * <pre>
-     * Information about a debugged graph.
-     * </pre>
-     *
-     * <code>.tensorflow.DebuggedGraph debugged_graph = 8;</code>
-     */
-    public Builder setDebuggedGraph(
-        org.tensorflow.proto.util.DebuggedGraph.Builder builderForValue) {
-      if (debuggedGraphBuilder_ == null) {
-        what_ = builderForValue.build();
-        onChanged();
-      } else {
-        debuggedGraphBuilder_.setMessage(builderForValue.build());
-      }
-      whatCase_ = 8;
-      return this;
-    }
-    /**
-     * <pre>
-     * Information about a debugged graph.
-     * </pre>
-     *
-     * <code>.tensorflow.DebuggedGraph debugged_graph = 8;</code>
-     */
-    public Builder mergeDebuggedGraph(org.tensorflow.proto.util.DebuggedGraph value) {
-      if (debuggedGraphBuilder_ == null) {
-        if (whatCase_ == 8 &&
-            what_ != org.tensorflow.proto.util.DebuggedGraph.getDefaultInstance()) {
-          what_ = org.tensorflow.proto.util.DebuggedGraph.newBuilder((org.tensorflow.proto.util.DebuggedGraph) what_)
-              .mergeFrom(value).buildPartial();
-        } else {
-          what_ = value;
-        }
-        onChanged();
-      } else {
-        if (whatCase_ == 8) {
-          debuggedGraphBuilder_.mergeFrom(value);
-        }
-        debuggedGraphBuilder_.setMessage(value);
-      }
-      whatCase_ = 8;
-      return this;
-    }
-    /**
-     * <pre>
-     * Information about a debugged graph.
-     * </pre>
-     *
-     * <code>.tensorflow.DebuggedGraph debugged_graph = 8;</code>
-     */
-    public Builder clearDebuggedGraph() {
-      if (debuggedGraphBuilder_ == null) {
-        if (whatCase_ == 8) {
-          whatCase_ = 0;
-          what_ = null;
-          onChanged();
-        }
-      } else {
-        if (whatCase_ == 8) {
-          whatCase_ = 0;
-          what_ = null;
-        }
-        debuggedGraphBuilder_.clear();
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Information about a debugged graph.
-     * </pre>
-     *
-     * <code>.tensorflow.DebuggedGraph debugged_graph = 8;</code>
-     */
-    public org.tensorflow.proto.util.DebuggedGraph.Builder getDebuggedGraphBuilder() {
-      return getDebuggedGraphFieldBuilder().getBuilder();
-    }
-    /**
-     * <pre>
-     * Information about a debugged graph.
-     * </pre>
-     *
-     * <code>.tensorflow.DebuggedGraph debugged_graph = 8;</code>
-     */
-    public org.tensorflow.proto.util.DebuggedGraphOrBuilder getDebuggedGraphOrBuilder() {
-      if ((whatCase_ == 8) && (debuggedGraphBuilder_ != null)) {
-        return debuggedGraphBuilder_.getMessageOrBuilder();
-      } else {
-        if (whatCase_ == 8) {
-          return (org.tensorflow.proto.util.DebuggedGraph) what_;
-        }
-        return org.tensorflow.proto.util.DebuggedGraph.getDefaultInstance();
-      }
-    }
-    /**
-     * <pre>
-     * Information about a debugged graph.
-     * </pre>
-     *
-     * <code>.tensorflow.DebuggedGraph debugged_graph = 8;</code>
-     */
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.util.DebuggedGraph, org.tensorflow.proto.util.DebuggedGraph.Builder, org.tensorflow.proto.util.DebuggedGraphOrBuilder> 
-        getDebuggedGraphFieldBuilder() {
-      if (debuggedGraphBuilder_ == null) {
-        if (!(whatCase_ == 8)) {
-          what_ = org.tensorflow.proto.util.DebuggedGraph.getDefaultInstance();
-        }
-        debuggedGraphBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-            org.tensorflow.proto.util.DebuggedGraph, org.tensorflow.proto.util.DebuggedGraph.Builder, org.tensorflow.proto.util.DebuggedGraphOrBuilder>(
-                (org.tensorflow.proto.util.DebuggedGraph) what_,
-                getParentForChildren(),
-                isClean());
-        what_ = null;
-      }
-      whatCase_ = 8;
-      onChanged();;
-      return debuggedGraphBuilder_;
-    }
-
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.util.Execution, org.tensorflow.proto.util.Execution.Builder, org.tensorflow.proto.util.ExecutionOrBuilder> executionBuilder_;
-    /**
-     * <pre>
-     * Execution of an op or a Graph (e.g., a tf.function).
-     * </pre>
-     *
-     * <code>.tensorflow.Execution execution = 9;</code>
-     */
-    public boolean hasExecution() {
-      return whatCase_ == 9;
-    }
-    /**
-     * <pre>
-     * Execution of an op or a Graph (e.g., a tf.function).
-     * </pre>
-     *
-     * <code>.tensorflow.Execution execution = 9;</code>
-     */
-    public org.tensorflow.proto.util.Execution getExecution() {
-      if (executionBuilder_ == null) {
-        if (whatCase_ == 9) {
-          return (org.tensorflow.proto.util.Execution) what_;
-        }
-        return org.tensorflow.proto.util.Execution.getDefaultInstance();
-      } else {
-        if (whatCase_ == 9) {
-          return executionBuilder_.getMessage();
-        }
-        return org.tensorflow.proto.util.Execution.getDefaultInstance();
-      }
-    }
-    /**
-     * <pre>
-     * Execution of an op or a Graph (e.g., a tf.function).
-     * </pre>
-     *
-     * <code>.tensorflow.Execution execution = 9;</code>
-     */
-    public Builder setExecution(org.tensorflow.proto.util.Execution value) {
-      if (executionBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        what_ = value;
-        onChanged();
-      } else {
-        executionBuilder_.setMessage(value);
-      }
-      whatCase_ = 9;
-      return this;
-    }
-    /**
-     * <pre>
-     * Execution of an op or a Graph (e.g., a tf.function).
-     * </pre>
-     *
-     * <code>.tensorflow.Execution execution = 9;</code>
-     */
-    public Builder setExecution(
-        org.tensorflow.proto.util.Execution.Builder builderForValue) {
-      if (executionBuilder_ == null) {
-        what_ = builderForValue.build();
-        onChanged();
-      } else {
-        executionBuilder_.setMessage(builderForValue.build());
-      }
-      whatCase_ = 9;
-      return this;
-    }
-    /**
-     * <pre>
-     * Execution of an op or a Graph (e.g., a tf.function).
-     * </pre>
-     *
-     * <code>.tensorflow.Execution execution = 9;</code>
-     */
-    public Builder mergeExecution(org.tensorflow.proto.util.Execution value) {
-      if (executionBuilder_ == null) {
-        if (whatCase_ == 9 &&
-            what_ != org.tensorflow.proto.util.Execution.getDefaultInstance()) {
-          what_ = org.tensorflow.proto.util.Execution.newBuilder((org.tensorflow.proto.util.Execution) what_)
-              .mergeFrom(value).buildPartial();
-        } else {
-          what_ = value;
-        }
-        onChanged();
-      } else {
-        if (whatCase_ == 9) {
-          executionBuilder_.mergeFrom(value);
-        }
-        executionBuilder_.setMessage(value);
-      }
-      whatCase_ = 9;
-      return this;
-    }
-    /**
-     * <pre>
-     * Execution of an op or a Graph (e.g., a tf.function).
-     * </pre>
-     *
-     * <code>.tensorflow.Execution execution = 9;</code>
-     */
-    public Builder clearExecution() {
-      if (executionBuilder_ == null) {
-        if (whatCase_ == 9) {
-          whatCase_ = 0;
-          what_ = null;
-          onChanged();
-        }
-      } else {
-        if (whatCase_ == 9) {
-          whatCase_ = 0;
-          what_ = null;
-        }
-        executionBuilder_.clear();
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Execution of an op or a Graph (e.g., a tf.function).
-     * </pre>
-     *
-     * <code>.tensorflow.Execution execution = 9;</code>
-     */
-    public org.tensorflow.proto.util.Execution.Builder getExecutionBuilder() {
-      return getExecutionFieldBuilder().getBuilder();
-    }
-    /**
-     * <pre>
-     * Execution of an op or a Graph (e.g., a tf.function).
-     * </pre>
-     *
-     * <code>.tensorflow.Execution execution = 9;</code>
-     */
-    public org.tensorflow.proto.util.ExecutionOrBuilder getExecutionOrBuilder() {
-      if ((whatCase_ == 9) && (executionBuilder_ != null)) {
-        return executionBuilder_.getMessageOrBuilder();
-      } else {
-        if (whatCase_ == 9) {
-          return (org.tensorflow.proto.util.Execution) what_;
-        }
-        return org.tensorflow.proto.util.Execution.getDefaultInstance();
-      }
-    }
-    /**
-     * <pre>
-     * Execution of an op or a Graph (e.g., a tf.function).
-     * </pre>
-     *
-     * <code>.tensorflow.Execution execution = 9;</code>
-     */
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.util.Execution, org.tensorflow.proto.util.Execution.Builder, org.tensorflow.proto.util.ExecutionOrBuilder> 
-        getExecutionFieldBuilder() {
-      if (executionBuilder_ == null) {
-        if (!(whatCase_ == 9)) {
-          what_ = org.tensorflow.proto.util.Execution.getDefaultInstance();
-        }
-        executionBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-            org.tensorflow.proto.util.Execution, org.tensorflow.proto.util.Execution.Builder, org.tensorflow.proto.util.ExecutionOrBuilder>(
-                (org.tensorflow.proto.util.Execution) what_,
-                getParentForChildren(),
-                isClean());
-        what_ = null;
-      }
-      whatCase_ = 9;
-      onChanged();;
-      return executionBuilder_;
-    }
-
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.util.GraphExecutionTrace, org.tensorflow.proto.util.GraphExecutionTrace.Builder, org.tensorflow.proto.util.GraphExecutionTraceOrBuilder> graphExecutionTraceBuilder_;
-    /**
-     * <pre>
-     * A graph execution trace: Contains information about the intermediate
-     * tensors computed during the graph execution.
-     * </pre>
-     *
-     * <code>.tensorflow.GraphExecutionTrace graph_execution_trace = 10;</code>
-     */
-    public boolean hasGraphExecutionTrace() {
-      return whatCase_ == 10;
-    }
-    /**
-     * <pre>
-     * A graph execution trace: Contains information about the intermediate
-     * tensors computed during the graph execution.
-     * </pre>
-     *
-     * <code>.tensorflow.GraphExecutionTrace graph_execution_trace = 10;</code>
-     */
-    public org.tensorflow.proto.util.GraphExecutionTrace getGraphExecutionTrace() {
-      if (graphExecutionTraceBuilder_ == null) {
-        if (whatCase_ == 10) {
-          return (org.tensorflow.proto.util.GraphExecutionTrace) what_;
-        }
-        return org.tensorflow.proto.util.GraphExecutionTrace.getDefaultInstance();
-      } else {
-        if (whatCase_ == 10) {
-          return graphExecutionTraceBuilder_.getMessage();
-        }
-        return org.tensorflow.proto.util.GraphExecutionTrace.getDefaultInstance();
-      }
-    }
-    /**
-     * <pre>
-     * A graph execution trace: Contains information about the intermediate
-     * tensors computed during the graph execution.
-     * </pre>
-     *
-     * <code>.tensorflow.GraphExecutionTrace graph_execution_trace = 10;</code>
-     */
-    public Builder setGraphExecutionTrace(org.tensorflow.proto.util.GraphExecutionTrace value) {
-      if (graphExecutionTraceBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        what_ = value;
-        onChanged();
-      } else {
-        graphExecutionTraceBuilder_.setMessage(value);
-      }
-      whatCase_ = 10;
-      return this;
-    }
-    /**
-     * <pre>
-     * A graph execution trace: Contains information about the intermediate
-     * tensors computed during the graph execution.
-     * </pre>
-     *
-     * <code>.tensorflow.GraphExecutionTrace graph_execution_trace = 10;</code>
-     */
-    public Builder setGraphExecutionTrace(
-        org.tensorflow.proto.util.GraphExecutionTrace.Builder builderForValue) {
-      if (graphExecutionTraceBuilder_ == null) {
-        what_ = builderForValue.build();
-        onChanged();
-      } else {
-        graphExecutionTraceBuilder_.setMessage(builderForValue.build());
-      }
-      whatCase_ = 10;
-      return this;
-    }
-    /**
-     * <pre>
-     * A graph execution trace: Contains information about the intermediate
-     * tensors computed during the graph execution.
-     * </pre>
-     *
-     * <code>.tensorflow.GraphExecutionTrace graph_execution_trace = 10;</code>
-     */
-    public Builder mergeGraphExecutionTrace(org.tensorflow.proto.util.GraphExecutionTrace value) {
-      if (graphExecutionTraceBuilder_ == null) {
-        if (whatCase_ == 10 &&
-            what_ != org.tensorflow.proto.util.GraphExecutionTrace.getDefaultInstance()) {
-          what_ = org.tensorflow.proto.util.GraphExecutionTrace.newBuilder((org.tensorflow.proto.util.GraphExecutionTrace) what_)
-              .mergeFrom(value).buildPartial();
-        } else {
-          what_ = value;
-        }
-        onChanged();
-      } else {
-        if (whatCase_ == 10) {
-          graphExecutionTraceBuilder_.mergeFrom(value);
-        }
-        graphExecutionTraceBuilder_.setMessage(value);
-      }
-      whatCase_ = 10;
-      return this;
-    }
-    /**
-     * <pre>
-     * A graph execution trace: Contains information about the intermediate
-     * tensors computed during the graph execution.
-     * </pre>
-     *
-     * <code>.tensorflow.GraphExecutionTrace graph_execution_trace = 10;</code>
-     */
-    public Builder clearGraphExecutionTrace() {
-      if (graphExecutionTraceBuilder_ == null) {
-        if (whatCase_ == 10) {
-          whatCase_ = 0;
-          what_ = null;
-          onChanged();
-        }
-      } else {
-        if (whatCase_ == 10) {
-          whatCase_ = 0;
-          what_ = null;
-        }
-        graphExecutionTraceBuilder_.clear();
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * A graph execution trace: Contains information about the intermediate
-     * tensors computed during the graph execution.
-     * </pre>
-     *
-     * <code>.tensorflow.GraphExecutionTrace graph_execution_trace = 10;</code>
-     */
-    public org.tensorflow.proto.util.GraphExecutionTrace.Builder getGraphExecutionTraceBuilder() {
-      return getGraphExecutionTraceFieldBuilder().getBuilder();
-    }
-    /**
-     * <pre>
-     * A graph execution trace: Contains information about the intermediate
-     * tensors computed during the graph execution.
-     * </pre>
-     *
-     * <code>.tensorflow.GraphExecutionTrace graph_execution_trace = 10;</code>
-     */
-    public org.tensorflow.proto.util.GraphExecutionTraceOrBuilder getGraphExecutionTraceOrBuilder() {
-      if ((whatCase_ == 10) && (graphExecutionTraceBuilder_ != null)) {
-        return graphExecutionTraceBuilder_.getMessageOrBuilder();
-      } else {
-        if (whatCase_ == 10) {
-          return (org.tensorflow.proto.util.GraphExecutionTrace) what_;
-        }
-        return org.tensorflow.proto.util.GraphExecutionTrace.getDefaultInstance();
-      }
-    }
-    /**
-     * <pre>
-     * A graph execution trace: Contains information about the intermediate
-     * tensors computed during the graph execution.
-     * </pre>
-     *
-     * <code>.tensorflow.GraphExecutionTrace graph_execution_trace = 10;</code>
-     */
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.util.GraphExecutionTrace, org.tensorflow.proto.util.GraphExecutionTrace.Builder, org.tensorflow.proto.util.GraphExecutionTraceOrBuilder> 
-        getGraphExecutionTraceFieldBuilder() {
-      if (graphExecutionTraceBuilder_ == null) {
-        if (!(whatCase_ == 10)) {
-          what_ = org.tensorflow.proto.util.GraphExecutionTrace.getDefaultInstance();
-        }
-        graphExecutionTraceBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-            org.tensorflow.proto.util.GraphExecutionTrace, org.tensorflow.proto.util.GraphExecutionTrace.Builder, org.tensorflow.proto.util.GraphExecutionTraceOrBuilder>(
-                (org.tensorflow.proto.util.GraphExecutionTrace) what_,
-                getParentForChildren(),
-                isClean());
-        what_ = null;
-      }
-      whatCase_ = 10;
-      onChanged();;
-      return graphExecutionTraceBuilder_;
-    }
-
-    /**
-     * <pre>
-     * The ID of the graph (i.e., FuncGraph) executed here: applicable only
-     * to the execution of a FuncGraph.
-     * </pre>
-     *
-     * <code>string graph_id = 11;</code>
-     */
-    public java.lang.String getGraphId() {
-      java.lang.Object ref = "";
-      if (whatCase_ == 11) {
-        ref = what_;
-      }
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        if (whatCase_ == 11) {
-          what_ = s;
-        }
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * The ID of the graph (i.e., FuncGraph) executed here: applicable only
-     * to the execution of a FuncGraph.
-     * </pre>
-     *
-     * <code>string graph_id = 11;</code>
-     */
-    public com.google.protobuf.ByteString
-        getGraphIdBytes() {
-      java.lang.Object ref = "";
-      if (whatCase_ == 11) {
-        ref = what_;
-      }
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        if (whatCase_ == 11) {
-          what_ = b;
-        }
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * The ID of the graph (i.e., FuncGraph) executed here: applicable only
-     * to the execution of a FuncGraph.
-     * </pre>
-     *
-     * <code>string graph_id = 11;</code>
-     */
-    public Builder setGraphId(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  whatCase_ = 11;
-      what_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * The ID of the graph (i.e., FuncGraph) executed here: applicable only
-     * to the execution of a FuncGraph.
-     * </pre>
-     *
-     * <code>string graph_id = 11;</code>
-     */
-    public Builder clearGraphId() {
-      if (whatCase_ == 11) {
-        whatCase_ = 0;
-        what_ = null;
-        onChanged();
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * The ID of the graph (i.e., FuncGraph) executed here: applicable only
-     * to the execution of a FuncGraph.
-     * </pre>
-     *
-     * <code>string graph_id = 11;</code>
-     */
-    public Builder setGraphIdBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      whatCase_ = 11;
-      what_ = value;
-      onChanged();
-      return this;
-    }
-
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.util.DebuggedDevice, org.tensorflow.proto.util.DebuggedDevice.Builder, org.tensorflow.proto.util.DebuggedDeviceOrBuilder> debuggedDeviceBuilder_;
-    /**
-     * <pre>
-     * A device on which debugger-instrumented ops and/or tensors reside.
-     * </pre>
-     *
-     * <code>.tensorflow.DebuggedDevice debugged_device = 12;</code>
-     */
-    public boolean hasDebuggedDevice() {
-      return whatCase_ == 12;
-    }
-    /**
-     * <pre>
-     * A device on which debugger-instrumented ops and/or tensors reside.
-     * </pre>
-     *
-     * <code>.tensorflow.DebuggedDevice debugged_device = 12;</code>
-     */
-    public org.tensorflow.proto.util.DebuggedDevice getDebuggedDevice() {
-      if (debuggedDeviceBuilder_ == null) {
-        if (whatCase_ == 12) {
-          return (org.tensorflow.proto.util.DebuggedDevice) what_;
-        }
-        return org.tensorflow.proto.util.DebuggedDevice.getDefaultInstance();
-      } else {
-        if (whatCase_ == 12) {
-          return debuggedDeviceBuilder_.getMessage();
-        }
-        return org.tensorflow.proto.util.DebuggedDevice.getDefaultInstance();
-      }
-    }
-    /**
-     * <pre>
-     * A device on which debugger-instrumented ops and/or tensors reside.
-     * </pre>
-     *
-     * <code>.tensorflow.DebuggedDevice debugged_device = 12;</code>
-     */
-    public Builder setDebuggedDevice(org.tensorflow.proto.util.DebuggedDevice value) {
-      if (debuggedDeviceBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        what_ = value;
-        onChanged();
-      } else {
-        debuggedDeviceBuilder_.setMessage(value);
-      }
-      whatCase_ = 12;
-      return this;
-    }
-    /**
-     * <pre>
-     * A device on which debugger-instrumented ops and/or tensors reside.
-     * </pre>
-     *
-     * <code>.tensorflow.DebuggedDevice debugged_device = 12;</code>
-     */
-    public Builder setDebuggedDevice(
-        org.tensorflow.proto.util.DebuggedDevice.Builder builderForValue) {
-      if (debuggedDeviceBuilder_ == null) {
-        what_ = builderForValue.build();
-        onChanged();
-      } else {
-        debuggedDeviceBuilder_.setMessage(builderForValue.build());
-      }
-      whatCase_ = 12;
-      return this;
-    }
-    /**
-     * <pre>
-     * A device on which debugger-instrumented ops and/or tensors reside.
-     * </pre>
-     *
-     * <code>.tensorflow.DebuggedDevice debugged_device = 12;</code>
-     */
-    public Builder mergeDebuggedDevice(org.tensorflow.proto.util.DebuggedDevice value) {
-      if (debuggedDeviceBuilder_ == null) {
-        if (whatCase_ == 12 &&
-            what_ != org.tensorflow.proto.util.DebuggedDevice.getDefaultInstance()) {
-          what_ = org.tensorflow.proto.util.DebuggedDevice.newBuilder((org.tensorflow.proto.util.DebuggedDevice) what_)
-              .mergeFrom(value).buildPartial();
-        } else {
-          what_ = value;
-        }
-        onChanged();
-      } else {
-        if (whatCase_ == 12) {
-          debuggedDeviceBuilder_.mergeFrom(value);
-        }
-        debuggedDeviceBuilder_.setMessage(value);
-      }
-      whatCase_ = 12;
-      return this;
-    }
-    /**
-     * <pre>
-     * A device on which debugger-instrumented ops and/or tensors reside.
-     * </pre>
-     *
-     * <code>.tensorflow.DebuggedDevice debugged_device = 12;</code>
-     */
-    public Builder clearDebuggedDevice() {
-      if (debuggedDeviceBuilder_ == null) {
-        if (whatCase_ == 12) {
-          whatCase_ = 0;
-          what_ = null;
-          onChanged();
-        }
-      } else {
-        if (whatCase_ == 12) {
-          whatCase_ = 0;
-          what_ = null;
-        }
-        debuggedDeviceBuilder_.clear();
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * A device on which debugger-instrumented ops and/or tensors reside.
-     * </pre>
-     *
-     * <code>.tensorflow.DebuggedDevice debugged_device = 12;</code>
-     */
-    public org.tensorflow.proto.util.DebuggedDevice.Builder getDebuggedDeviceBuilder() {
-      return getDebuggedDeviceFieldBuilder().getBuilder();
-    }
-    /**
-     * <pre>
-     * A device on which debugger-instrumented ops and/or tensors reside.
-     * </pre>
-     *
-     * <code>.tensorflow.DebuggedDevice debugged_device = 12;</code>
-     */
-    public org.tensorflow.proto.util.DebuggedDeviceOrBuilder getDebuggedDeviceOrBuilder() {
-      if ((whatCase_ == 12) && (debuggedDeviceBuilder_ != null)) {
-        return debuggedDeviceBuilder_.getMessageOrBuilder();
-      } else {
-        if (whatCase_ == 12) {
-          return (org.tensorflow.proto.util.DebuggedDevice) what_;
-        }
-        return org.tensorflow.proto.util.DebuggedDevice.getDefaultInstance();
-      }
-    }
-    /**
-     * <pre>
-     * A device on which debugger-instrumented ops and/or tensors reside.
-     * </pre>
-     *
-     * <code>.tensorflow.DebuggedDevice debugged_device = 12;</code>
-     */
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.util.DebuggedDevice, org.tensorflow.proto.util.DebuggedDevice.Builder, org.tensorflow.proto.util.DebuggedDeviceOrBuilder> 
-        getDebuggedDeviceFieldBuilder() {
-      if (debuggedDeviceBuilder_ == null) {
-        if (!(whatCase_ == 12)) {
-          what_ = org.tensorflow.proto.util.DebuggedDevice.getDefaultInstance();
-        }
-        debuggedDeviceBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-            org.tensorflow.proto.util.DebuggedDevice, org.tensorflow.proto.util.DebuggedDevice.Builder, org.tensorflow.proto.util.DebuggedDeviceOrBuilder>(
-                (org.tensorflow.proto.util.DebuggedDevice) what_,
-                getParentForChildren(),
-                isClean());
-        what_ = null;
-      }
-      whatCase_ = 12;
-      onChanged();;
-      return debuggedDeviceBuilder_;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.DebugEvent)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.DebugEvent)
-  private static final org.tensorflow.proto.util.DebugEvent DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.util.DebugEvent();
-  }
-
-  public static org.tensorflow.proto.util.DebugEvent getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<DebugEvent>
-      PARSER = new com.google.protobuf.AbstractParser<DebugEvent>() {
-    @java.lang.Override
-    public DebugEvent parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new DebugEvent(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<DebugEvent> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<DebugEvent> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.util.DebugEvent getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/DebugEventOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/DebugEventOrBuilder.java
deleted file mode 100644
index dd322c5d995..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/DebugEventOrBuilder.java
+++ /dev/null
@@ -1,258 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/debug_event.proto
-
-package org.tensorflow.proto.util;
-
-public interface DebugEventOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.DebugEvent)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <pre>
-   * Timestamp in seconds (with microsecond precision).
-   * </pre>
-   *
-   * <code>double wall_time = 1;</code>
-   */
-  double getWallTime();
-
-  /**
-   * <pre>
-   * Step of training (if available).
-   * </pre>
-   *
-   * <code>int64 step = 2;</code>
-   */
-  long getStep();
-
-  /**
-   * <pre>
-   * Metadata related to this debugging data.
-   * </pre>
-   *
-   * <code>.tensorflow.DebugMetadata debug_metadata = 3;</code>
-   */
-  boolean hasDebugMetadata();
-  /**
-   * <pre>
-   * Metadata related to this debugging data.
-   * </pre>
-   *
-   * <code>.tensorflow.DebugMetadata debug_metadata = 3;</code>
-   */
-  org.tensorflow.proto.util.DebugMetadata getDebugMetadata();
-  /**
-   * <pre>
-   * Metadata related to this debugging data.
-   * </pre>
-   *
-   * <code>.tensorflow.DebugMetadata debug_metadata = 3;</code>
-   */
-  org.tensorflow.proto.util.DebugMetadataOrBuilder getDebugMetadataOrBuilder();
-
-  /**
-   * <pre>
-   * The content of a source file.
-   * </pre>
-   *
-   * <code>.tensorflow.SourceFile source_file = 4;</code>
-   */
-  boolean hasSourceFile();
-  /**
-   * <pre>
-   * The content of a source file.
-   * </pre>
-   *
-   * <code>.tensorflow.SourceFile source_file = 4;</code>
-   */
-  org.tensorflow.proto.util.SourceFile getSourceFile();
-  /**
-   * <pre>
-   * The content of a source file.
-   * </pre>
-   *
-   * <code>.tensorflow.SourceFile source_file = 4;</code>
-   */
-  org.tensorflow.proto.util.SourceFileOrBuilder getSourceFileOrBuilder();
-
-  /**
-   * <pre>
-   * A stack frame (filename, line number and column number, function name and
-   * code string) with ID.
-   * </pre>
-   *
-   * <code>.tensorflow.StackFrameWithId stack_frame_with_id = 6;</code>
-   */
-  boolean hasStackFrameWithId();
-  /**
-   * <pre>
-   * A stack frame (filename, line number and column number, function name and
-   * code string) with ID.
-   * </pre>
-   *
-   * <code>.tensorflow.StackFrameWithId stack_frame_with_id = 6;</code>
-   */
-  org.tensorflow.proto.util.StackFrameWithId getStackFrameWithId();
-  /**
-   * <pre>
-   * A stack frame (filename, line number and column number, function name and
-   * code string) with ID.
-   * </pre>
-   *
-   * <code>.tensorflow.StackFrameWithId stack_frame_with_id = 6;</code>
-   */
-  org.tensorflow.proto.util.StackFrameWithIdOrBuilder getStackFrameWithIdOrBuilder();
-
-  /**
-   * <pre>
-   * The creation of an op within a graph (e.g., a FuncGraph compiled from
-   * a Python function).
-   * </pre>
-   *
-   * <code>.tensorflow.GraphOpCreation graph_op_creation = 7;</code>
-   */
-  boolean hasGraphOpCreation();
-  /**
-   * <pre>
-   * The creation of an op within a graph (e.g., a FuncGraph compiled from
-   * a Python function).
-   * </pre>
-   *
-   * <code>.tensorflow.GraphOpCreation graph_op_creation = 7;</code>
-   */
-  org.tensorflow.proto.util.GraphOpCreation getGraphOpCreation();
-  /**
-   * <pre>
-   * The creation of an op within a graph (e.g., a FuncGraph compiled from
-   * a Python function).
-   * </pre>
-   *
-   * <code>.tensorflow.GraphOpCreation graph_op_creation = 7;</code>
-   */
-  org.tensorflow.proto.util.GraphOpCreationOrBuilder getGraphOpCreationOrBuilder();
-
-  /**
-   * <pre>
-   * Information about a debugged graph.
-   * </pre>
-   *
-   * <code>.tensorflow.DebuggedGraph debugged_graph = 8;</code>
-   */
-  boolean hasDebuggedGraph();
-  /**
-   * <pre>
-   * Information about a debugged graph.
-   * </pre>
-   *
-   * <code>.tensorflow.DebuggedGraph debugged_graph = 8;</code>
-   */
-  org.tensorflow.proto.util.DebuggedGraph getDebuggedGraph();
-  /**
-   * <pre>
-   * Information about a debugged graph.
-   * </pre>
-   *
-   * <code>.tensorflow.DebuggedGraph debugged_graph = 8;</code>
-   */
-  org.tensorflow.proto.util.DebuggedGraphOrBuilder getDebuggedGraphOrBuilder();
-
-  /**
-   * <pre>
-   * Execution of an op or a Graph (e.g., a tf.function).
-   * </pre>
-   *
-   * <code>.tensorflow.Execution execution = 9;</code>
-   */
-  boolean hasExecution();
-  /**
-   * <pre>
-   * Execution of an op or a Graph (e.g., a tf.function).
-   * </pre>
-   *
-   * <code>.tensorflow.Execution execution = 9;</code>
-   */
-  org.tensorflow.proto.util.Execution getExecution();
-  /**
-   * <pre>
-   * Execution of an op or a Graph (e.g., a tf.function).
-   * </pre>
-   *
-   * <code>.tensorflow.Execution execution = 9;</code>
-   */
-  org.tensorflow.proto.util.ExecutionOrBuilder getExecutionOrBuilder();
-
-  /**
-   * <pre>
-   * A graph execution trace: Contains information about the intermediate
-   * tensors computed during the graph execution.
-   * </pre>
-   *
-   * <code>.tensorflow.GraphExecutionTrace graph_execution_trace = 10;</code>
-   */
-  boolean hasGraphExecutionTrace();
-  /**
-   * <pre>
-   * A graph execution trace: Contains information about the intermediate
-   * tensors computed during the graph execution.
-   * </pre>
-   *
-   * <code>.tensorflow.GraphExecutionTrace graph_execution_trace = 10;</code>
-   */
-  org.tensorflow.proto.util.GraphExecutionTrace getGraphExecutionTrace();
-  /**
-   * <pre>
-   * A graph execution trace: Contains information about the intermediate
-   * tensors computed during the graph execution.
-   * </pre>
-   *
-   * <code>.tensorflow.GraphExecutionTrace graph_execution_trace = 10;</code>
-   */
-  org.tensorflow.proto.util.GraphExecutionTraceOrBuilder getGraphExecutionTraceOrBuilder();
-
-  /**
-   * <pre>
-   * The ID of the graph (i.e., FuncGraph) executed here: applicable only
-   * to the execution of a FuncGraph.
-   * </pre>
-   *
-   * <code>string graph_id = 11;</code>
-   */
-  java.lang.String getGraphId();
-  /**
-   * <pre>
-   * The ID of the graph (i.e., FuncGraph) executed here: applicable only
-   * to the execution of a FuncGraph.
-   * </pre>
-   *
-   * <code>string graph_id = 11;</code>
-   */
-  com.google.protobuf.ByteString
-      getGraphIdBytes();
-
-  /**
-   * <pre>
-   * A device on which debugger-instrumented ops and/or tensors reside.
-   * </pre>
-   *
-   * <code>.tensorflow.DebuggedDevice debugged_device = 12;</code>
-   */
-  boolean hasDebuggedDevice();
-  /**
-   * <pre>
-   * A device on which debugger-instrumented ops and/or tensors reside.
-   * </pre>
-   *
-   * <code>.tensorflow.DebuggedDevice debugged_device = 12;</code>
-   */
-  org.tensorflow.proto.util.DebuggedDevice getDebuggedDevice();
-  /**
-   * <pre>
-   * A device on which debugger-instrumented ops and/or tensors reside.
-   * </pre>
-   *
-   * <code>.tensorflow.DebuggedDevice debugged_device = 12;</code>
-   */
-  org.tensorflow.proto.util.DebuggedDeviceOrBuilder getDebuggedDeviceOrBuilder();
-
-  public org.tensorflow.proto.util.DebugEvent.WhatCase getWhatCase();
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/DebugEventProtos.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/DebugEventProtos.java
deleted file mode 100644
index 411b002619a..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/DebugEventProtos.java
+++ /dev/null
@@ -1,205 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/debug_event.proto
-
-package org.tensorflow.proto.util;
-
-public final class DebugEventProtos {
-  private DebugEventProtos() {}
-  public static void registerAllExtensions(
-      com.google.protobuf.ExtensionRegistryLite registry) {
-  }
-
-  public static void registerAllExtensions(
-      com.google.protobuf.ExtensionRegistry registry) {
-    registerAllExtensions(
-        (com.google.protobuf.ExtensionRegistryLite) registry);
-  }
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_DebugEvent_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_DebugEvent_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_DebugMetadata_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_DebugMetadata_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_SourceFile_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_SourceFile_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_StackFrameWithId_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_StackFrameWithId_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_CodeLocation_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_CodeLocation_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_GraphOpCreation_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_GraphOpCreation_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_DebuggedGraph_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_DebuggedGraph_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_DebuggedDevice_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_DebuggedDevice_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_Execution_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_Execution_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_GraphExecutionTrace_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_GraphExecutionTrace_fieldAccessorTable;
-
-  public static com.google.protobuf.Descriptors.FileDescriptor
-      getDescriptor() {
-    return descriptor;
-  }
-  private static  com.google.protobuf.Descriptors.FileDescriptor
-      descriptor;
-  static {
-    java.lang.String[] descriptorData = {
-      "\n*tensorflow/core/protobuf/debug_event.p" +
-      "roto\022\ntensorflow\032&tensorflow/core/framew" +
-      "ork/tensor.proto\032/tensorflow/core/protob" +
-      "uf/graph_debug_info.proto\"\376\003\n\nDebugEvent" +
-      "\022\021\n\twall_time\030\001 \001(\001\022\014\n\004step\030\002 \001(\003\0223\n\016deb" +
-      "ug_metadata\030\003 \001(\0132\031.tensorflow.DebugMeta" +
-      "dataH\000\022-\n\013source_file\030\004 \001(\0132\026.tensorflow" +
-      ".SourceFileH\000\022;\n\023stack_frame_with_id\030\006 \001" +
-      "(\0132\034.tensorflow.StackFrameWithIdH\000\0228\n\021gr" +
-      "aph_op_creation\030\007 \001(\0132\033.tensorflow.Graph" +
-      "OpCreationH\000\0223\n\016debugged_graph\030\010 \001(\0132\031.t" +
-      "ensorflow.DebuggedGraphH\000\022*\n\texecution\030\t" +
-      " \001(\0132\025.tensorflow.ExecutionH\000\022@\n\025graph_e" +
-      "xecution_trace\030\n \001(\0132\037.tensorflow.GraphE" +
-      "xecutionTraceH\000\022\022\n\010graph_id\030\013 \001(\tH\000\0225\n\017d" +
-      "ebugged_device\030\014 \001(\0132\032.tensorflow.Debugg" +
-      "edDeviceH\000B\006\n\004what\"W\n\rDebugMetadata\022\032\n\022t" +
-      "ensorflow_version\030\001 \001(\t\022\024\n\014file_version\030" +
-      "\002 \001(\t\022\024\n\014tfdbg_run_id\030\003 \001(\t\"A\n\nSourceFil" +
-      "e\022\021\n\tfile_path\030\001 \001(\t\022\021\n\thost_name\030\002 \001(\t\022" +
-      "\r\n\005lines\030\003 \003(\t\"]\n\020StackFrameWithId\022\n\n\002id" +
-      "\030\001 \001(\t\022=\n\rfile_line_col\030\002 \001(\0132&.tensorfl" +
-      "ow.GraphDebugInfo.FileLineCol\":\n\014CodeLoc" +
-      "ation\022\021\n\thost_name\030\001 \001(\t\022\027\n\017stack_frame_" +
-      "ids\030\002 \003(\t\"\344\001\n\017GraphOpCreation\022\017\n\007op_type" +
-      "\030\001 \001(\t\022\017\n\007op_name\030\002 \001(\t\022\022\n\ngraph_name\030\003 " +
-      "\001(\t\022\020\n\010graph_id\030\004 \001(\t\022\023\n\013device_name\030\005 \001" +
-      "(\t\022\023\n\013input_names\030\006 \003(\t\022\023\n\013num_outputs\030\007" +
-      " \001(\005\022/\n\rcode_location\030\010 \001(\0132\030.tensorflow" +
-      ".CodeLocation\022\031\n\021output_tensor_ids\030\t \003(\005" +
-      "\"\245\001\n\rDebuggedGraph\022\020\n\010graph_id\030\001 \001(\t\022\022\n\n" +
-      "graph_name\030\002 \001(\t\022\030\n\020instrumented_ops\030\003 \003" +
-      "(\t\022\032\n\022original_graph_def\030\004 \001(\014\022\036\n\026instru" +
-      "mented_graph_def\030\005 \001(\014\022\030\n\020outer_context_" +
-      "id\030\006 \001(\t\"8\n\016DebuggedDevice\022\023\n\013device_nam" +
-      "e\030\001 \001(\t\022\021\n\tdevice_id\030\002 \001(\005\"\263\002\n\tExecution" +
-      "\022\017\n\007op_type\030\001 \001(\t\022\023\n\013num_outputs\030\002 \001(\005\022\020" +
-      "\n\010graph_id\030\003 \001(\t\022\030\n\020input_tensor_ids\030\004 \003" +
-      "(\003\022\031\n\021output_tensor_ids\030\005 \003(\003\0226\n\021tensor_" +
-      "debug_mode\030\006 \001(\0162\033.tensorflow.TensorDebu" +
-      "gMode\022.\n\rtensor_protos\030\007 \003(\0132\027.tensorflo" +
-      "w.TensorProto\022/\n\rcode_location\030\010 \001(\0132\030.t" +
-      "ensorflow.CodeLocation\022 \n\030output_tensor_" +
-      "device_ids\030\t \003(\005\"\321\001\n\023GraphExecutionTrace" +
-      "\022\030\n\020tfdbg_context_id\030\001 \001(\t\022\017\n\007op_name\030\002 " +
-      "\001(\t\022\023\n\013output_slot\030\003 \001(\005\0226\n\021tensor_debug" +
-      "_mode\030\004 \001(\0162\033.tensorflow.TensorDebugMode" +
-      "\022-\n\014tensor_proto\030\005 \001(\0132\027.tensorflow.Tens" +
-      "orProto\022\023\n\013device_name\030\006 \001(\t*\266\001\n\017TensorD" +
-      "ebugMode\022\017\n\013UNSPECIFIED\020\000\022\r\n\tNO_TENSOR\020\001" +
-      "\022\017\n\013CURT_HEALTH\020\002\022\022\n\016CONCISE_HEALTH\020\003\022\017\n" +
-      "\013FULL_HEALTH\020\004\022\t\n\005SHAPE\020\005\022\021\n\rFULL_NUMERI" +
-      "CS\020\006\022\017\n\013FULL_TENSOR\020\007\022\036\n\032REDUCE_INF_NAN_" +
-      "THREE_SLOTS\020\010B|\n\031org.tensorflow.proto.ut" +
-      "ilB\020DebugEventProtosP\001ZHgithub.com/tenso" +
-      "rflow/tensorflow/tensorflow/go/core/core" +
-      "_protos_go_proto\370\001\001b\006proto3"
-    };
-    descriptor = com.google.protobuf.Descriptors.FileDescriptor
-      .internalBuildGeneratedFileFrom(descriptorData,
-        new com.google.protobuf.Descriptors.FileDescriptor[] {
-          org.tensorflow.proto.framework.TensorProtos.getDescriptor(),
-          org.tensorflow.proto.framework.GraphDebugInfoProtos.getDescriptor(),
-        });
-    internal_static_tensorflow_DebugEvent_descriptor =
-      getDescriptor().getMessageTypes().get(0);
-    internal_static_tensorflow_DebugEvent_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_DebugEvent_descriptor,
-        new java.lang.String[] { "WallTime", "Step", "DebugMetadata", "SourceFile", "StackFrameWithId", "GraphOpCreation", "DebuggedGraph", "Execution", "GraphExecutionTrace", "GraphId", "DebuggedDevice", "What", });
-    internal_static_tensorflow_DebugMetadata_descriptor =
-      getDescriptor().getMessageTypes().get(1);
-    internal_static_tensorflow_DebugMetadata_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_DebugMetadata_descriptor,
-        new java.lang.String[] { "TensorflowVersion", "FileVersion", "TfdbgRunId", });
-    internal_static_tensorflow_SourceFile_descriptor =
-      getDescriptor().getMessageTypes().get(2);
-    internal_static_tensorflow_SourceFile_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_SourceFile_descriptor,
-        new java.lang.String[] { "FilePath", "HostName", "Lines", });
-    internal_static_tensorflow_StackFrameWithId_descriptor =
-      getDescriptor().getMessageTypes().get(3);
-    internal_static_tensorflow_StackFrameWithId_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_StackFrameWithId_descriptor,
-        new java.lang.String[] { "Id", "FileLineCol", });
-    internal_static_tensorflow_CodeLocation_descriptor =
-      getDescriptor().getMessageTypes().get(4);
-    internal_static_tensorflow_CodeLocation_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_CodeLocation_descriptor,
-        new java.lang.String[] { "HostName", "StackFrameIds", });
-    internal_static_tensorflow_GraphOpCreation_descriptor =
-      getDescriptor().getMessageTypes().get(5);
-    internal_static_tensorflow_GraphOpCreation_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_GraphOpCreation_descriptor,
-        new java.lang.String[] { "OpType", "OpName", "GraphName", "GraphId", "DeviceName", "InputNames", "NumOutputs", "CodeLocation", "OutputTensorIds", });
-    internal_static_tensorflow_DebuggedGraph_descriptor =
-      getDescriptor().getMessageTypes().get(6);
-    internal_static_tensorflow_DebuggedGraph_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_DebuggedGraph_descriptor,
-        new java.lang.String[] { "GraphId", "GraphName", "InstrumentedOps", "OriginalGraphDef", "InstrumentedGraphDef", "OuterContextId", });
-    internal_static_tensorflow_DebuggedDevice_descriptor =
-      getDescriptor().getMessageTypes().get(7);
-    internal_static_tensorflow_DebuggedDevice_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_DebuggedDevice_descriptor,
-        new java.lang.String[] { "DeviceName", "DeviceId", });
-    internal_static_tensorflow_Execution_descriptor =
-      getDescriptor().getMessageTypes().get(8);
-    internal_static_tensorflow_Execution_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_Execution_descriptor,
-        new java.lang.String[] { "OpType", "NumOutputs", "GraphId", "InputTensorIds", "OutputTensorIds", "TensorDebugMode", "TensorProtos", "CodeLocation", "OutputTensorDeviceIds", });
-    internal_static_tensorflow_GraphExecutionTrace_descriptor =
-      getDescriptor().getMessageTypes().get(9);
-    internal_static_tensorflow_GraphExecutionTrace_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_GraphExecutionTrace_descriptor,
-        new java.lang.String[] { "TfdbgContextId", "OpName", "OutputSlot", "TensorDebugMode", "TensorProto", "DeviceName", });
-    org.tensorflow.proto.framework.TensorProtos.getDescriptor();
-    org.tensorflow.proto.framework.GraphDebugInfoProtos.getDescriptor();
-  }
-
-  // @@protoc_insertion_point(outer_class_scope)
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/DebugMetadata.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/DebugMetadata.java
deleted file mode 100644
index e0679182c6e..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/DebugMetadata.java
+++ /dev/null
@@ -1,920 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/debug_event.proto
-
-package org.tensorflow.proto.util;
-
-/**
- * <pre>
- * Metadata about the debugger and the debugged TensorFlow program.
- * </pre>
- *
- * Protobuf type {@code tensorflow.DebugMetadata}
- */
-public  final class DebugMetadata extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.DebugMetadata)
-    DebugMetadataOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use DebugMetadata.newBuilder() to construct.
-  private DebugMetadata(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private DebugMetadata() {
-    tensorflowVersion_ = "";
-    fileVersion_ = "";
-    tfdbgRunId_ = "";
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new DebugMetadata();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private DebugMetadata(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 10: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            tensorflowVersion_ = s;
-            break;
-          }
-          case 18: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            fileVersion_ = s;
-            break;
-          }
-          case 26: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            tfdbgRunId_ = s;
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.util.DebugEventProtos.internal_static_tensorflow_DebugMetadata_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.util.DebugEventProtos.internal_static_tensorflow_DebugMetadata_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.util.DebugMetadata.class, org.tensorflow.proto.util.DebugMetadata.Builder.class);
-  }
-
-  public static final int TENSORFLOW_VERSION_FIELD_NUMBER = 1;
-  private volatile java.lang.Object tensorflowVersion_;
-  /**
-   * <pre>
-   * Version of TensorFlow.
-   * </pre>
-   *
-   * <code>string tensorflow_version = 1;</code>
-   */
-  public java.lang.String getTensorflowVersion() {
-    java.lang.Object ref = tensorflowVersion_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      tensorflowVersion_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * Version of TensorFlow.
-   * </pre>
-   *
-   * <code>string tensorflow_version = 1;</code>
-   */
-  public com.google.protobuf.ByteString
-      getTensorflowVersionBytes() {
-    java.lang.Object ref = tensorflowVersion_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      tensorflowVersion_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int FILE_VERSION_FIELD_NUMBER = 2;
-  private volatile java.lang.Object fileVersion_;
-  /**
-   * <pre>
-   * Version of the DebugEvent file format.
-   * Has a format of "debug.Event:&lt;number&gt;", e.g., "debug.Event:1".
-   * </pre>
-   *
-   * <code>string file_version = 2;</code>
-   */
-  public java.lang.String getFileVersion() {
-    java.lang.Object ref = fileVersion_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      fileVersion_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * Version of the DebugEvent file format.
-   * Has a format of "debug.Event:&lt;number&gt;", e.g., "debug.Event:1".
-   * </pre>
-   *
-   * <code>string file_version = 2;</code>
-   */
-  public com.google.protobuf.ByteString
-      getFileVersionBytes() {
-    java.lang.Object ref = fileVersion_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      fileVersion_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int TFDBG_RUN_ID_FIELD_NUMBER = 3;
-  private volatile java.lang.Object tfdbgRunId_;
-  /**
-   * <pre>
-   * A unique ID for the current run of tfdbg.
-   * A run of tfdbg is defined as a TensorFlow job instrumented by tfdbg.
-   * Multiple hosts in a distributed TensorFlow job instrumented by tfdbg
-   * have the same ID.
-   * </pre>
-   *
-   * <code>string tfdbg_run_id = 3;</code>
-   */
-  public java.lang.String getTfdbgRunId() {
-    java.lang.Object ref = tfdbgRunId_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      tfdbgRunId_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * A unique ID for the current run of tfdbg.
-   * A run of tfdbg is defined as a TensorFlow job instrumented by tfdbg.
-   * Multiple hosts in a distributed TensorFlow job instrumented by tfdbg
-   * have the same ID.
-   * </pre>
-   *
-   * <code>string tfdbg_run_id = 3;</code>
-   */
-  public com.google.protobuf.ByteString
-      getTfdbgRunIdBytes() {
-    java.lang.Object ref = tfdbgRunId_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      tfdbgRunId_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    if (!getTensorflowVersionBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 1, tensorflowVersion_);
-    }
-    if (!getFileVersionBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 2, fileVersion_);
-    }
-    if (!getTfdbgRunIdBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 3, tfdbgRunId_);
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (!getTensorflowVersionBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, tensorflowVersion_);
-    }
-    if (!getFileVersionBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(2, fileVersion_);
-    }
-    if (!getTfdbgRunIdBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(3, tfdbgRunId_);
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.util.DebugMetadata)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.util.DebugMetadata other = (org.tensorflow.proto.util.DebugMetadata) obj;
-
-    if (!getTensorflowVersion()
-        .equals(other.getTensorflowVersion())) return false;
-    if (!getFileVersion()
-        .equals(other.getFileVersion())) return false;
-    if (!getTfdbgRunId()
-        .equals(other.getTfdbgRunId())) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    hash = (37 * hash) + TENSORFLOW_VERSION_FIELD_NUMBER;
-    hash = (53 * hash) + getTensorflowVersion().hashCode();
-    hash = (37 * hash) + FILE_VERSION_FIELD_NUMBER;
-    hash = (53 * hash) + getFileVersion().hashCode();
-    hash = (37 * hash) + TFDBG_RUN_ID_FIELD_NUMBER;
-    hash = (53 * hash) + getTfdbgRunId().hashCode();
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.util.DebugMetadata parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.DebugMetadata parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.DebugMetadata parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.DebugMetadata parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.DebugMetadata parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.DebugMetadata parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.DebugMetadata parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.DebugMetadata parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.DebugMetadata parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.DebugMetadata parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.DebugMetadata parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.DebugMetadata parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.util.DebugMetadata prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * <pre>
-   * Metadata about the debugger and the debugged TensorFlow program.
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.DebugMetadata}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.DebugMetadata)
-      org.tensorflow.proto.util.DebugMetadataOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.util.DebugEventProtos.internal_static_tensorflow_DebugMetadata_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.util.DebugEventProtos.internal_static_tensorflow_DebugMetadata_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.util.DebugMetadata.class, org.tensorflow.proto.util.DebugMetadata.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.util.DebugMetadata.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      tensorflowVersion_ = "";
-
-      fileVersion_ = "";
-
-      tfdbgRunId_ = "";
-
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.util.DebugEventProtos.internal_static_tensorflow_DebugMetadata_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.DebugMetadata getDefaultInstanceForType() {
-      return org.tensorflow.proto.util.DebugMetadata.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.DebugMetadata build() {
-      org.tensorflow.proto.util.DebugMetadata result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.DebugMetadata buildPartial() {
-      org.tensorflow.proto.util.DebugMetadata result = new org.tensorflow.proto.util.DebugMetadata(this);
-      result.tensorflowVersion_ = tensorflowVersion_;
-      result.fileVersion_ = fileVersion_;
-      result.tfdbgRunId_ = tfdbgRunId_;
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.util.DebugMetadata) {
-        return mergeFrom((org.tensorflow.proto.util.DebugMetadata)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.util.DebugMetadata other) {
-      if (other == org.tensorflow.proto.util.DebugMetadata.getDefaultInstance()) return this;
-      if (!other.getTensorflowVersion().isEmpty()) {
-        tensorflowVersion_ = other.tensorflowVersion_;
-        onChanged();
-      }
-      if (!other.getFileVersion().isEmpty()) {
-        fileVersion_ = other.fileVersion_;
-        onChanged();
-      }
-      if (!other.getTfdbgRunId().isEmpty()) {
-        tfdbgRunId_ = other.tfdbgRunId_;
-        onChanged();
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.util.DebugMetadata parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.util.DebugMetadata) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-
-    private java.lang.Object tensorflowVersion_ = "";
-    /**
-     * <pre>
-     * Version of TensorFlow.
-     * </pre>
-     *
-     * <code>string tensorflow_version = 1;</code>
-     */
-    public java.lang.String getTensorflowVersion() {
-      java.lang.Object ref = tensorflowVersion_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        tensorflowVersion_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Version of TensorFlow.
-     * </pre>
-     *
-     * <code>string tensorflow_version = 1;</code>
-     */
-    public com.google.protobuf.ByteString
-        getTensorflowVersionBytes() {
-      java.lang.Object ref = tensorflowVersion_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        tensorflowVersion_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Version of TensorFlow.
-     * </pre>
-     *
-     * <code>string tensorflow_version = 1;</code>
-     */
-    public Builder setTensorflowVersion(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      tensorflowVersion_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Version of TensorFlow.
-     * </pre>
-     *
-     * <code>string tensorflow_version = 1;</code>
-     */
-    public Builder clearTensorflowVersion() {
-      
-      tensorflowVersion_ = getDefaultInstance().getTensorflowVersion();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Version of TensorFlow.
-     * </pre>
-     *
-     * <code>string tensorflow_version = 1;</code>
-     */
-    public Builder setTensorflowVersionBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      tensorflowVersion_ = value;
-      onChanged();
-      return this;
-    }
-
-    private java.lang.Object fileVersion_ = "";
-    /**
-     * <pre>
-     * Version of the DebugEvent file format.
-     * Has a format of "debug.Event:&lt;number&gt;", e.g., "debug.Event:1".
-     * </pre>
-     *
-     * <code>string file_version = 2;</code>
-     */
-    public java.lang.String getFileVersion() {
-      java.lang.Object ref = fileVersion_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        fileVersion_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Version of the DebugEvent file format.
-     * Has a format of "debug.Event:&lt;number&gt;", e.g., "debug.Event:1".
-     * </pre>
-     *
-     * <code>string file_version = 2;</code>
-     */
-    public com.google.protobuf.ByteString
-        getFileVersionBytes() {
-      java.lang.Object ref = fileVersion_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        fileVersion_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Version of the DebugEvent file format.
-     * Has a format of "debug.Event:&lt;number&gt;", e.g., "debug.Event:1".
-     * </pre>
-     *
-     * <code>string file_version = 2;</code>
-     */
-    public Builder setFileVersion(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      fileVersion_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Version of the DebugEvent file format.
-     * Has a format of "debug.Event:&lt;number&gt;", e.g., "debug.Event:1".
-     * </pre>
-     *
-     * <code>string file_version = 2;</code>
-     */
-    public Builder clearFileVersion() {
-      
-      fileVersion_ = getDefaultInstance().getFileVersion();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Version of the DebugEvent file format.
-     * Has a format of "debug.Event:&lt;number&gt;", e.g., "debug.Event:1".
-     * </pre>
-     *
-     * <code>string file_version = 2;</code>
-     */
-    public Builder setFileVersionBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      fileVersion_ = value;
-      onChanged();
-      return this;
-    }
-
-    private java.lang.Object tfdbgRunId_ = "";
-    /**
-     * <pre>
-     * A unique ID for the current run of tfdbg.
-     * A run of tfdbg is defined as a TensorFlow job instrumented by tfdbg.
-     * Multiple hosts in a distributed TensorFlow job instrumented by tfdbg
-     * have the same ID.
-     * </pre>
-     *
-     * <code>string tfdbg_run_id = 3;</code>
-     */
-    public java.lang.String getTfdbgRunId() {
-      java.lang.Object ref = tfdbgRunId_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        tfdbgRunId_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * A unique ID for the current run of tfdbg.
-     * A run of tfdbg is defined as a TensorFlow job instrumented by tfdbg.
-     * Multiple hosts in a distributed TensorFlow job instrumented by tfdbg
-     * have the same ID.
-     * </pre>
-     *
-     * <code>string tfdbg_run_id = 3;</code>
-     */
-    public com.google.protobuf.ByteString
-        getTfdbgRunIdBytes() {
-      java.lang.Object ref = tfdbgRunId_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        tfdbgRunId_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * A unique ID for the current run of tfdbg.
-     * A run of tfdbg is defined as a TensorFlow job instrumented by tfdbg.
-     * Multiple hosts in a distributed TensorFlow job instrumented by tfdbg
-     * have the same ID.
-     * </pre>
-     *
-     * <code>string tfdbg_run_id = 3;</code>
-     */
-    public Builder setTfdbgRunId(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      tfdbgRunId_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * A unique ID for the current run of tfdbg.
-     * A run of tfdbg is defined as a TensorFlow job instrumented by tfdbg.
-     * Multiple hosts in a distributed TensorFlow job instrumented by tfdbg
-     * have the same ID.
-     * </pre>
-     *
-     * <code>string tfdbg_run_id = 3;</code>
-     */
-    public Builder clearTfdbgRunId() {
-      
-      tfdbgRunId_ = getDefaultInstance().getTfdbgRunId();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * A unique ID for the current run of tfdbg.
-     * A run of tfdbg is defined as a TensorFlow job instrumented by tfdbg.
-     * Multiple hosts in a distributed TensorFlow job instrumented by tfdbg
-     * have the same ID.
-     * </pre>
-     *
-     * <code>string tfdbg_run_id = 3;</code>
-     */
-    public Builder setTfdbgRunIdBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      tfdbgRunId_ = value;
-      onChanged();
-      return this;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.DebugMetadata)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.DebugMetadata)
-  private static final org.tensorflow.proto.util.DebugMetadata DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.util.DebugMetadata();
-  }
-
-  public static org.tensorflow.proto.util.DebugMetadata getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<DebugMetadata>
-      PARSER = new com.google.protobuf.AbstractParser<DebugMetadata>() {
-    @java.lang.Override
-    public DebugMetadata parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new DebugMetadata(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<DebugMetadata> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<DebugMetadata> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.util.DebugMetadata getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/DebugMetadataOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/DebugMetadataOrBuilder.java
deleted file mode 100644
index 91015599f7d..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/DebugMetadataOrBuilder.java
+++ /dev/null
@@ -1,71 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/debug_event.proto
-
-package org.tensorflow.proto.util;
-
-public interface DebugMetadataOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.DebugMetadata)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <pre>
-   * Version of TensorFlow.
-   * </pre>
-   *
-   * <code>string tensorflow_version = 1;</code>
-   */
-  java.lang.String getTensorflowVersion();
-  /**
-   * <pre>
-   * Version of TensorFlow.
-   * </pre>
-   *
-   * <code>string tensorflow_version = 1;</code>
-   */
-  com.google.protobuf.ByteString
-      getTensorflowVersionBytes();
-
-  /**
-   * <pre>
-   * Version of the DebugEvent file format.
-   * Has a format of "debug.Event:&lt;number&gt;", e.g., "debug.Event:1".
-   * </pre>
-   *
-   * <code>string file_version = 2;</code>
-   */
-  java.lang.String getFileVersion();
-  /**
-   * <pre>
-   * Version of the DebugEvent file format.
-   * Has a format of "debug.Event:&lt;number&gt;", e.g., "debug.Event:1".
-   * </pre>
-   *
-   * <code>string file_version = 2;</code>
-   */
-  com.google.protobuf.ByteString
-      getFileVersionBytes();
-
-  /**
-   * <pre>
-   * A unique ID for the current run of tfdbg.
-   * A run of tfdbg is defined as a TensorFlow job instrumented by tfdbg.
-   * Multiple hosts in a distributed TensorFlow job instrumented by tfdbg
-   * have the same ID.
-   * </pre>
-   *
-   * <code>string tfdbg_run_id = 3;</code>
-   */
-  java.lang.String getTfdbgRunId();
-  /**
-   * <pre>
-   * A unique ID for the current run of tfdbg.
-   * A run of tfdbg is defined as a TensorFlow job instrumented by tfdbg.
-   * Multiple hosts in a distributed TensorFlow job instrumented by tfdbg
-   * have the same ID.
-   * </pre>
-   *
-   * <code>string tfdbg_run_id = 3;</code>
-   */
-  com.google.protobuf.ByteString
-      getTfdbgRunIdBytes();
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/DebuggedDevice.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/DebuggedDevice.java
deleted file mode 100644
index 18f528481cd..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/DebuggedDevice.java
+++ /dev/null
@@ -1,667 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/debug_event.proto
-
-package org.tensorflow.proto.util;
-
-/**
- * <pre>
- * A device on which ops and/or tensors are instrumented by the debugger.
- * </pre>
- *
- * Protobuf type {@code tensorflow.DebuggedDevice}
- */
-public  final class DebuggedDevice extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.DebuggedDevice)
-    DebuggedDeviceOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use DebuggedDevice.newBuilder() to construct.
-  private DebuggedDevice(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private DebuggedDevice() {
-    deviceName_ = "";
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new DebuggedDevice();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private DebuggedDevice(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 10: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            deviceName_ = s;
-            break;
-          }
-          case 16: {
-
-            deviceId_ = input.readInt32();
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.util.DebugEventProtos.internal_static_tensorflow_DebuggedDevice_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.util.DebugEventProtos.internal_static_tensorflow_DebuggedDevice_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.util.DebuggedDevice.class, org.tensorflow.proto.util.DebuggedDevice.Builder.class);
-  }
-
-  public static final int DEVICE_NAME_FIELD_NUMBER = 1;
-  private volatile java.lang.Object deviceName_;
-  /**
-   * <pre>
-   * Name of the device.
-   * </pre>
-   *
-   * <code>string device_name = 1;</code>
-   */
-  public java.lang.String getDeviceName() {
-    java.lang.Object ref = deviceName_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      deviceName_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * Name of the device.
-   * </pre>
-   *
-   * <code>string device_name = 1;</code>
-   */
-  public com.google.protobuf.ByteString
-      getDeviceNameBytes() {
-    java.lang.Object ref = deviceName_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      deviceName_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int DEVICE_ID_FIELD_NUMBER = 2;
-  private int deviceId_;
-  /**
-   * <pre>
-   * A debugger-generated ID for the device. Guaranteed to be unique within
-   * the scope of the debugged TensorFlow program, including single-host and
-   * multi-host settings.
-   * TODO(cais): Test the uniqueness guarantee in multi-host settings.
-   * </pre>
-   *
-   * <code>int32 device_id = 2;</code>
-   */
-  public int getDeviceId() {
-    return deviceId_;
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    if (!getDeviceNameBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 1, deviceName_);
-    }
-    if (deviceId_ != 0) {
-      output.writeInt32(2, deviceId_);
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (!getDeviceNameBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, deviceName_);
-    }
-    if (deviceId_ != 0) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt32Size(2, deviceId_);
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.util.DebuggedDevice)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.util.DebuggedDevice other = (org.tensorflow.proto.util.DebuggedDevice) obj;
-
-    if (!getDeviceName()
-        .equals(other.getDeviceName())) return false;
-    if (getDeviceId()
-        != other.getDeviceId()) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    hash = (37 * hash) + DEVICE_NAME_FIELD_NUMBER;
-    hash = (53 * hash) + getDeviceName().hashCode();
-    hash = (37 * hash) + DEVICE_ID_FIELD_NUMBER;
-    hash = (53 * hash) + getDeviceId();
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.util.DebuggedDevice parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.DebuggedDevice parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.DebuggedDevice parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.DebuggedDevice parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.DebuggedDevice parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.DebuggedDevice parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.DebuggedDevice parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.DebuggedDevice parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.DebuggedDevice parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.DebuggedDevice parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.DebuggedDevice parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.DebuggedDevice parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.util.DebuggedDevice prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * <pre>
-   * A device on which ops and/or tensors are instrumented by the debugger.
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.DebuggedDevice}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.DebuggedDevice)
-      org.tensorflow.proto.util.DebuggedDeviceOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.util.DebugEventProtos.internal_static_tensorflow_DebuggedDevice_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.util.DebugEventProtos.internal_static_tensorflow_DebuggedDevice_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.util.DebuggedDevice.class, org.tensorflow.proto.util.DebuggedDevice.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.util.DebuggedDevice.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      deviceName_ = "";
-
-      deviceId_ = 0;
-
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.util.DebugEventProtos.internal_static_tensorflow_DebuggedDevice_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.DebuggedDevice getDefaultInstanceForType() {
-      return org.tensorflow.proto.util.DebuggedDevice.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.DebuggedDevice build() {
-      org.tensorflow.proto.util.DebuggedDevice result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.DebuggedDevice buildPartial() {
-      org.tensorflow.proto.util.DebuggedDevice result = new org.tensorflow.proto.util.DebuggedDevice(this);
-      result.deviceName_ = deviceName_;
-      result.deviceId_ = deviceId_;
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.util.DebuggedDevice) {
-        return mergeFrom((org.tensorflow.proto.util.DebuggedDevice)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.util.DebuggedDevice other) {
-      if (other == org.tensorflow.proto.util.DebuggedDevice.getDefaultInstance()) return this;
-      if (!other.getDeviceName().isEmpty()) {
-        deviceName_ = other.deviceName_;
-        onChanged();
-      }
-      if (other.getDeviceId() != 0) {
-        setDeviceId(other.getDeviceId());
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.util.DebuggedDevice parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.util.DebuggedDevice) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-
-    private java.lang.Object deviceName_ = "";
-    /**
-     * <pre>
-     * Name of the device.
-     * </pre>
-     *
-     * <code>string device_name = 1;</code>
-     */
-    public java.lang.String getDeviceName() {
-      java.lang.Object ref = deviceName_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        deviceName_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Name of the device.
-     * </pre>
-     *
-     * <code>string device_name = 1;</code>
-     */
-    public com.google.protobuf.ByteString
-        getDeviceNameBytes() {
-      java.lang.Object ref = deviceName_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        deviceName_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Name of the device.
-     * </pre>
-     *
-     * <code>string device_name = 1;</code>
-     */
-    public Builder setDeviceName(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      deviceName_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Name of the device.
-     * </pre>
-     *
-     * <code>string device_name = 1;</code>
-     */
-    public Builder clearDeviceName() {
-      
-      deviceName_ = getDefaultInstance().getDeviceName();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Name of the device.
-     * </pre>
-     *
-     * <code>string device_name = 1;</code>
-     */
-    public Builder setDeviceNameBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      deviceName_ = value;
-      onChanged();
-      return this;
-    }
-
-    private int deviceId_ ;
-    /**
-     * <pre>
-     * A debugger-generated ID for the device. Guaranteed to be unique within
-     * the scope of the debugged TensorFlow program, including single-host and
-     * multi-host settings.
-     * TODO(cais): Test the uniqueness guarantee in multi-host settings.
-     * </pre>
-     *
-     * <code>int32 device_id = 2;</code>
-     */
-    public int getDeviceId() {
-      return deviceId_;
-    }
-    /**
-     * <pre>
-     * A debugger-generated ID for the device. Guaranteed to be unique within
-     * the scope of the debugged TensorFlow program, including single-host and
-     * multi-host settings.
-     * TODO(cais): Test the uniqueness guarantee in multi-host settings.
-     * </pre>
-     *
-     * <code>int32 device_id = 2;</code>
-     */
-    public Builder setDeviceId(int value) {
-      
-      deviceId_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * A debugger-generated ID for the device. Guaranteed to be unique within
-     * the scope of the debugged TensorFlow program, including single-host and
-     * multi-host settings.
-     * TODO(cais): Test the uniqueness guarantee in multi-host settings.
-     * </pre>
-     *
-     * <code>int32 device_id = 2;</code>
-     */
-    public Builder clearDeviceId() {
-      
-      deviceId_ = 0;
-      onChanged();
-      return this;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.DebuggedDevice)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.DebuggedDevice)
-  private static final org.tensorflow.proto.util.DebuggedDevice DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.util.DebuggedDevice();
-  }
-
-  public static org.tensorflow.proto.util.DebuggedDevice getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<DebuggedDevice>
-      PARSER = new com.google.protobuf.AbstractParser<DebuggedDevice>() {
-    @java.lang.Override
-    public DebuggedDevice parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new DebuggedDevice(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<DebuggedDevice> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<DebuggedDevice> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.util.DebuggedDevice getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/DebuggedDeviceOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/DebuggedDeviceOrBuilder.java
deleted file mode 100644
index 3915849b3ae..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/DebuggedDeviceOrBuilder.java
+++ /dev/null
@@ -1,39 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/debug_event.proto
-
-package org.tensorflow.proto.util;
-
-public interface DebuggedDeviceOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.DebuggedDevice)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <pre>
-   * Name of the device.
-   * </pre>
-   *
-   * <code>string device_name = 1;</code>
-   */
-  java.lang.String getDeviceName();
-  /**
-   * <pre>
-   * Name of the device.
-   * </pre>
-   *
-   * <code>string device_name = 1;</code>
-   */
-  com.google.protobuf.ByteString
-      getDeviceNameBytes();
-
-  /**
-   * <pre>
-   * A debugger-generated ID for the device. Guaranteed to be unique within
-   * the scope of the debugged TensorFlow program, including single-host and
-   * multi-host settings.
-   * TODO(cais): Test the uniqueness guarantee in multi-host settings.
-   * </pre>
-   *
-   * <code>int32 device_id = 2;</code>
-   */
-  int getDeviceId();
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/DebuggedGraph.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/DebuggedGraph.java
deleted file mode 100644
index 6fa3e5adcff..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/DebuggedGraph.java
+++ /dev/null
@@ -1,1295 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/debug_event.proto
-
-package org.tensorflow.proto.util;
-
-/**
- * <pre>
- * A debugger-instrumented graph.
- * </pre>
- *
- * Protobuf type {@code tensorflow.DebuggedGraph}
- */
-public  final class DebuggedGraph extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.DebuggedGraph)
-    DebuggedGraphOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use DebuggedGraph.newBuilder() to construct.
-  private DebuggedGraph(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private DebuggedGraph() {
-    graphId_ = "";
-    graphName_ = "";
-    instrumentedOps_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-    originalGraphDef_ = com.google.protobuf.ByteString.EMPTY;
-    instrumentedGraphDef_ = com.google.protobuf.ByteString.EMPTY;
-    outerContextId_ = "";
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new DebuggedGraph();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private DebuggedGraph(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    int mutable_bitField0_ = 0;
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 10: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            graphId_ = s;
-            break;
-          }
-          case 18: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            graphName_ = s;
-            break;
-          }
-          case 26: {
-            java.lang.String s = input.readStringRequireUtf8();
-            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
-              instrumentedOps_ = new com.google.protobuf.LazyStringArrayList();
-              mutable_bitField0_ |= 0x00000001;
-            }
-            instrumentedOps_.add(s);
-            break;
-          }
-          case 34: {
-
-            originalGraphDef_ = input.readBytes();
-            break;
-          }
-          case 42: {
-
-            instrumentedGraphDef_ = input.readBytes();
-            break;
-          }
-          case 50: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            outerContextId_ = s;
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      if (((mutable_bitField0_ & 0x00000001) != 0)) {
-        instrumentedOps_ = instrumentedOps_.getUnmodifiableView();
-      }
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.util.DebugEventProtos.internal_static_tensorflow_DebuggedGraph_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.util.DebugEventProtos.internal_static_tensorflow_DebuggedGraph_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.util.DebuggedGraph.class, org.tensorflow.proto.util.DebuggedGraph.Builder.class);
-  }
-
-  public static final int GRAPH_ID_FIELD_NUMBER = 1;
-  private volatile java.lang.Object graphId_;
-  /**
-   * <pre>
-   * An ID for the graph.
-   * This can be used up to look up graph names. Generated by the debugger.
-   * </pre>
-   *
-   * <code>string graph_id = 1;</code>
-   */
-  public java.lang.String getGraphId() {
-    java.lang.Object ref = graphId_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      graphId_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * An ID for the graph.
-   * This can be used up to look up graph names. Generated by the debugger.
-   * </pre>
-   *
-   * <code>string graph_id = 1;</code>
-   */
-  public com.google.protobuf.ByteString
-      getGraphIdBytes() {
-    java.lang.Object ref = graphId_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      graphId_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int GRAPH_NAME_FIELD_NUMBER = 2;
-  private volatile java.lang.Object graphName_;
-  /**
-   * <pre>
-   * Name of the graph (if available).
-   * </pre>
-   *
-   * <code>string graph_name = 2;</code>
-   */
-  public java.lang.String getGraphName() {
-    java.lang.Object ref = graphName_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      graphName_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * Name of the graph (if available).
-   * </pre>
-   *
-   * <code>string graph_name = 2;</code>
-   */
-  public com.google.protobuf.ByteString
-      getGraphNameBytes() {
-    java.lang.Object ref = graphName_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      graphName_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int INSTRUMENTED_OPS_FIELD_NUMBER = 3;
-  private com.google.protobuf.LazyStringList instrumentedOps_;
-  /**
-   * <pre>
-   * Names of the instrumented ops. This can be used to look up op name
-   * based on the numeric-summary tensors (2nd column).
-   * </pre>
-   *
-   * <code>repeated string instrumented_ops = 3;</code>
-   */
-  public com.google.protobuf.ProtocolStringList
-      getInstrumentedOpsList() {
-    return instrumentedOps_;
-  }
-  /**
-   * <pre>
-   * Names of the instrumented ops. This can be used to look up op name
-   * based on the numeric-summary tensors (2nd column).
-   * </pre>
-   *
-   * <code>repeated string instrumented_ops = 3;</code>
-   */
-  public int getInstrumentedOpsCount() {
-    return instrumentedOps_.size();
-  }
-  /**
-   * <pre>
-   * Names of the instrumented ops. This can be used to look up op name
-   * based on the numeric-summary tensors (2nd column).
-   * </pre>
-   *
-   * <code>repeated string instrumented_ops = 3;</code>
-   */
-  public java.lang.String getInstrumentedOps(int index) {
-    return instrumentedOps_.get(index);
-  }
-  /**
-   * <pre>
-   * Names of the instrumented ops. This can be used to look up op name
-   * based on the numeric-summary tensors (2nd column).
-   * </pre>
-   *
-   * <code>repeated string instrumented_ops = 3;</code>
-   */
-  public com.google.protobuf.ByteString
-      getInstrumentedOpsBytes(int index) {
-    return instrumentedOps_.getByteString(index);
-  }
-
-  public static final int ORIGINAL_GRAPH_DEF_FIELD_NUMBER = 4;
-  private com.google.protobuf.ByteString originalGraphDef_;
-  /**
-   * <pre>
-   * Original (uninstrumented) GraphDef (if available).
-   * </pre>
-   *
-   * <code>bytes original_graph_def = 4;</code>
-   */
-  public com.google.protobuf.ByteString getOriginalGraphDef() {
-    return originalGraphDef_;
-  }
-
-  public static final int INSTRUMENTED_GRAPH_DEF_FIELD_NUMBER = 5;
-  private com.google.protobuf.ByteString instrumentedGraphDef_;
-  /**
-   * <pre>
-   * An encoded version of a GraphDef.
-   * This graph may include the debugger-inserted ops.
-   * </pre>
-   *
-   * <code>bytes instrumented_graph_def = 5;</code>
-   */
-  public com.google.protobuf.ByteString getInstrumentedGraphDef() {
-    return instrumentedGraphDef_;
-  }
-
-  public static final int OUTER_CONTEXT_ID_FIELD_NUMBER = 6;
-  private volatile java.lang.Object outerContextId_;
-  /**
-   * <pre>
-   * IDs of the immediate enclosing context (graph), if any.
-   * </pre>
-   *
-   * <code>string outer_context_id = 6;</code>
-   */
-  public java.lang.String getOuterContextId() {
-    java.lang.Object ref = outerContextId_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      outerContextId_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * IDs of the immediate enclosing context (graph), if any.
-   * </pre>
-   *
-   * <code>string outer_context_id = 6;</code>
-   */
-  public com.google.protobuf.ByteString
-      getOuterContextIdBytes() {
-    java.lang.Object ref = outerContextId_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      outerContextId_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    if (!getGraphIdBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 1, graphId_);
-    }
-    if (!getGraphNameBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 2, graphName_);
-    }
-    for (int i = 0; i < instrumentedOps_.size(); i++) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 3, instrumentedOps_.getRaw(i));
-    }
-    if (!originalGraphDef_.isEmpty()) {
-      output.writeBytes(4, originalGraphDef_);
-    }
-    if (!instrumentedGraphDef_.isEmpty()) {
-      output.writeBytes(5, instrumentedGraphDef_);
-    }
-    if (!getOuterContextIdBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 6, outerContextId_);
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (!getGraphIdBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, graphId_);
-    }
-    if (!getGraphNameBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(2, graphName_);
-    }
-    {
-      int dataSize = 0;
-      for (int i = 0; i < instrumentedOps_.size(); i++) {
-        dataSize += computeStringSizeNoTag(instrumentedOps_.getRaw(i));
-      }
-      size += dataSize;
-      size += 1 * getInstrumentedOpsList().size();
-    }
-    if (!originalGraphDef_.isEmpty()) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeBytesSize(4, originalGraphDef_);
-    }
-    if (!instrumentedGraphDef_.isEmpty()) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeBytesSize(5, instrumentedGraphDef_);
-    }
-    if (!getOuterContextIdBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(6, outerContextId_);
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.util.DebuggedGraph)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.util.DebuggedGraph other = (org.tensorflow.proto.util.DebuggedGraph) obj;
-
-    if (!getGraphId()
-        .equals(other.getGraphId())) return false;
-    if (!getGraphName()
-        .equals(other.getGraphName())) return false;
-    if (!getInstrumentedOpsList()
-        .equals(other.getInstrumentedOpsList())) return false;
-    if (!getOriginalGraphDef()
-        .equals(other.getOriginalGraphDef())) return false;
-    if (!getInstrumentedGraphDef()
-        .equals(other.getInstrumentedGraphDef())) return false;
-    if (!getOuterContextId()
-        .equals(other.getOuterContextId())) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    hash = (37 * hash) + GRAPH_ID_FIELD_NUMBER;
-    hash = (53 * hash) + getGraphId().hashCode();
-    hash = (37 * hash) + GRAPH_NAME_FIELD_NUMBER;
-    hash = (53 * hash) + getGraphName().hashCode();
-    if (getInstrumentedOpsCount() > 0) {
-      hash = (37 * hash) + INSTRUMENTED_OPS_FIELD_NUMBER;
-      hash = (53 * hash) + getInstrumentedOpsList().hashCode();
-    }
-    hash = (37 * hash) + ORIGINAL_GRAPH_DEF_FIELD_NUMBER;
-    hash = (53 * hash) + getOriginalGraphDef().hashCode();
-    hash = (37 * hash) + INSTRUMENTED_GRAPH_DEF_FIELD_NUMBER;
-    hash = (53 * hash) + getInstrumentedGraphDef().hashCode();
-    hash = (37 * hash) + OUTER_CONTEXT_ID_FIELD_NUMBER;
-    hash = (53 * hash) + getOuterContextId().hashCode();
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.util.DebuggedGraph parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.DebuggedGraph parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.DebuggedGraph parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.DebuggedGraph parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.DebuggedGraph parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.DebuggedGraph parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.DebuggedGraph parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.DebuggedGraph parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.DebuggedGraph parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.DebuggedGraph parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.DebuggedGraph parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.DebuggedGraph parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.util.DebuggedGraph prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * <pre>
-   * A debugger-instrumented graph.
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.DebuggedGraph}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.DebuggedGraph)
-      org.tensorflow.proto.util.DebuggedGraphOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.util.DebugEventProtos.internal_static_tensorflow_DebuggedGraph_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.util.DebugEventProtos.internal_static_tensorflow_DebuggedGraph_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.util.DebuggedGraph.class, org.tensorflow.proto.util.DebuggedGraph.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.util.DebuggedGraph.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      graphId_ = "";
-
-      graphName_ = "";
-
-      instrumentedOps_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-      bitField0_ = (bitField0_ & ~0x00000001);
-      originalGraphDef_ = com.google.protobuf.ByteString.EMPTY;
-
-      instrumentedGraphDef_ = com.google.protobuf.ByteString.EMPTY;
-
-      outerContextId_ = "";
-
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.util.DebugEventProtos.internal_static_tensorflow_DebuggedGraph_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.DebuggedGraph getDefaultInstanceForType() {
-      return org.tensorflow.proto.util.DebuggedGraph.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.DebuggedGraph build() {
-      org.tensorflow.proto.util.DebuggedGraph result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.DebuggedGraph buildPartial() {
-      org.tensorflow.proto.util.DebuggedGraph result = new org.tensorflow.proto.util.DebuggedGraph(this);
-      int from_bitField0_ = bitField0_;
-      result.graphId_ = graphId_;
-      result.graphName_ = graphName_;
-      if (((bitField0_ & 0x00000001) != 0)) {
-        instrumentedOps_ = instrumentedOps_.getUnmodifiableView();
-        bitField0_ = (bitField0_ & ~0x00000001);
-      }
-      result.instrumentedOps_ = instrumentedOps_;
-      result.originalGraphDef_ = originalGraphDef_;
-      result.instrumentedGraphDef_ = instrumentedGraphDef_;
-      result.outerContextId_ = outerContextId_;
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.util.DebuggedGraph) {
-        return mergeFrom((org.tensorflow.proto.util.DebuggedGraph)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.util.DebuggedGraph other) {
-      if (other == org.tensorflow.proto.util.DebuggedGraph.getDefaultInstance()) return this;
-      if (!other.getGraphId().isEmpty()) {
-        graphId_ = other.graphId_;
-        onChanged();
-      }
-      if (!other.getGraphName().isEmpty()) {
-        graphName_ = other.graphName_;
-        onChanged();
-      }
-      if (!other.instrumentedOps_.isEmpty()) {
-        if (instrumentedOps_.isEmpty()) {
-          instrumentedOps_ = other.instrumentedOps_;
-          bitField0_ = (bitField0_ & ~0x00000001);
-        } else {
-          ensureInstrumentedOpsIsMutable();
-          instrumentedOps_.addAll(other.instrumentedOps_);
-        }
-        onChanged();
-      }
-      if (other.getOriginalGraphDef() != com.google.protobuf.ByteString.EMPTY) {
-        setOriginalGraphDef(other.getOriginalGraphDef());
-      }
-      if (other.getInstrumentedGraphDef() != com.google.protobuf.ByteString.EMPTY) {
-        setInstrumentedGraphDef(other.getInstrumentedGraphDef());
-      }
-      if (!other.getOuterContextId().isEmpty()) {
-        outerContextId_ = other.outerContextId_;
-        onChanged();
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.util.DebuggedGraph parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.util.DebuggedGraph) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-    private int bitField0_;
-
-    private java.lang.Object graphId_ = "";
-    /**
-     * <pre>
-     * An ID for the graph.
-     * This can be used up to look up graph names. Generated by the debugger.
-     * </pre>
-     *
-     * <code>string graph_id = 1;</code>
-     */
-    public java.lang.String getGraphId() {
-      java.lang.Object ref = graphId_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        graphId_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * An ID for the graph.
-     * This can be used up to look up graph names. Generated by the debugger.
-     * </pre>
-     *
-     * <code>string graph_id = 1;</code>
-     */
-    public com.google.protobuf.ByteString
-        getGraphIdBytes() {
-      java.lang.Object ref = graphId_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        graphId_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * An ID for the graph.
-     * This can be used up to look up graph names. Generated by the debugger.
-     * </pre>
-     *
-     * <code>string graph_id = 1;</code>
-     */
-    public Builder setGraphId(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      graphId_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * An ID for the graph.
-     * This can be used up to look up graph names. Generated by the debugger.
-     * </pre>
-     *
-     * <code>string graph_id = 1;</code>
-     */
-    public Builder clearGraphId() {
-      
-      graphId_ = getDefaultInstance().getGraphId();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * An ID for the graph.
-     * This can be used up to look up graph names. Generated by the debugger.
-     * </pre>
-     *
-     * <code>string graph_id = 1;</code>
-     */
-    public Builder setGraphIdBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      graphId_ = value;
-      onChanged();
-      return this;
-    }
-
-    private java.lang.Object graphName_ = "";
-    /**
-     * <pre>
-     * Name of the graph (if available).
-     * </pre>
-     *
-     * <code>string graph_name = 2;</code>
-     */
-    public java.lang.String getGraphName() {
-      java.lang.Object ref = graphName_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        graphName_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Name of the graph (if available).
-     * </pre>
-     *
-     * <code>string graph_name = 2;</code>
-     */
-    public com.google.protobuf.ByteString
-        getGraphNameBytes() {
-      java.lang.Object ref = graphName_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        graphName_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Name of the graph (if available).
-     * </pre>
-     *
-     * <code>string graph_name = 2;</code>
-     */
-    public Builder setGraphName(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      graphName_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Name of the graph (if available).
-     * </pre>
-     *
-     * <code>string graph_name = 2;</code>
-     */
-    public Builder clearGraphName() {
-      
-      graphName_ = getDefaultInstance().getGraphName();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Name of the graph (if available).
-     * </pre>
-     *
-     * <code>string graph_name = 2;</code>
-     */
-    public Builder setGraphNameBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      graphName_ = value;
-      onChanged();
-      return this;
-    }
-
-    private com.google.protobuf.LazyStringList instrumentedOps_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-    private void ensureInstrumentedOpsIsMutable() {
-      if (!((bitField0_ & 0x00000001) != 0)) {
-        instrumentedOps_ = new com.google.protobuf.LazyStringArrayList(instrumentedOps_);
-        bitField0_ |= 0x00000001;
-       }
-    }
-    /**
-     * <pre>
-     * Names of the instrumented ops. This can be used to look up op name
-     * based on the numeric-summary tensors (2nd column).
-     * </pre>
-     *
-     * <code>repeated string instrumented_ops = 3;</code>
-     */
-    public com.google.protobuf.ProtocolStringList
-        getInstrumentedOpsList() {
-      return instrumentedOps_.getUnmodifiableView();
-    }
-    /**
-     * <pre>
-     * Names of the instrumented ops. This can be used to look up op name
-     * based on the numeric-summary tensors (2nd column).
-     * </pre>
-     *
-     * <code>repeated string instrumented_ops = 3;</code>
-     */
-    public int getInstrumentedOpsCount() {
-      return instrumentedOps_.size();
-    }
-    /**
-     * <pre>
-     * Names of the instrumented ops. This can be used to look up op name
-     * based on the numeric-summary tensors (2nd column).
-     * </pre>
-     *
-     * <code>repeated string instrumented_ops = 3;</code>
-     */
-    public java.lang.String getInstrumentedOps(int index) {
-      return instrumentedOps_.get(index);
-    }
-    /**
-     * <pre>
-     * Names of the instrumented ops. This can be used to look up op name
-     * based on the numeric-summary tensors (2nd column).
-     * </pre>
-     *
-     * <code>repeated string instrumented_ops = 3;</code>
-     */
-    public com.google.protobuf.ByteString
-        getInstrumentedOpsBytes(int index) {
-      return instrumentedOps_.getByteString(index);
-    }
-    /**
-     * <pre>
-     * Names of the instrumented ops. This can be used to look up op name
-     * based on the numeric-summary tensors (2nd column).
-     * </pre>
-     *
-     * <code>repeated string instrumented_ops = 3;</code>
-     */
-    public Builder setInstrumentedOps(
-        int index, java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  ensureInstrumentedOpsIsMutable();
-      instrumentedOps_.set(index, value);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Names of the instrumented ops. This can be used to look up op name
-     * based on the numeric-summary tensors (2nd column).
-     * </pre>
-     *
-     * <code>repeated string instrumented_ops = 3;</code>
-     */
-    public Builder addInstrumentedOps(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  ensureInstrumentedOpsIsMutable();
-      instrumentedOps_.add(value);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Names of the instrumented ops. This can be used to look up op name
-     * based on the numeric-summary tensors (2nd column).
-     * </pre>
-     *
-     * <code>repeated string instrumented_ops = 3;</code>
-     */
-    public Builder addAllInstrumentedOps(
-        java.lang.Iterable<java.lang.String> values) {
-      ensureInstrumentedOpsIsMutable();
-      com.google.protobuf.AbstractMessageLite.Builder.addAll(
-          values, instrumentedOps_);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Names of the instrumented ops. This can be used to look up op name
-     * based on the numeric-summary tensors (2nd column).
-     * </pre>
-     *
-     * <code>repeated string instrumented_ops = 3;</code>
-     */
-    public Builder clearInstrumentedOps() {
-      instrumentedOps_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-      bitField0_ = (bitField0_ & ~0x00000001);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Names of the instrumented ops. This can be used to look up op name
-     * based on the numeric-summary tensors (2nd column).
-     * </pre>
-     *
-     * <code>repeated string instrumented_ops = 3;</code>
-     */
-    public Builder addInstrumentedOpsBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      ensureInstrumentedOpsIsMutable();
-      instrumentedOps_.add(value);
-      onChanged();
-      return this;
-    }
-
-    private com.google.protobuf.ByteString originalGraphDef_ = com.google.protobuf.ByteString.EMPTY;
-    /**
-     * <pre>
-     * Original (uninstrumented) GraphDef (if available).
-     * </pre>
-     *
-     * <code>bytes original_graph_def = 4;</code>
-     */
-    public com.google.protobuf.ByteString getOriginalGraphDef() {
-      return originalGraphDef_;
-    }
-    /**
-     * <pre>
-     * Original (uninstrumented) GraphDef (if available).
-     * </pre>
-     *
-     * <code>bytes original_graph_def = 4;</code>
-     */
-    public Builder setOriginalGraphDef(com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      originalGraphDef_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Original (uninstrumented) GraphDef (if available).
-     * </pre>
-     *
-     * <code>bytes original_graph_def = 4;</code>
-     */
-    public Builder clearOriginalGraphDef() {
-      
-      originalGraphDef_ = getDefaultInstance().getOriginalGraphDef();
-      onChanged();
-      return this;
-    }
-
-    private com.google.protobuf.ByteString instrumentedGraphDef_ = com.google.protobuf.ByteString.EMPTY;
-    /**
-     * <pre>
-     * An encoded version of a GraphDef.
-     * This graph may include the debugger-inserted ops.
-     * </pre>
-     *
-     * <code>bytes instrumented_graph_def = 5;</code>
-     */
-    public com.google.protobuf.ByteString getInstrumentedGraphDef() {
-      return instrumentedGraphDef_;
-    }
-    /**
-     * <pre>
-     * An encoded version of a GraphDef.
-     * This graph may include the debugger-inserted ops.
-     * </pre>
-     *
-     * <code>bytes instrumented_graph_def = 5;</code>
-     */
-    public Builder setInstrumentedGraphDef(com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      instrumentedGraphDef_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * An encoded version of a GraphDef.
-     * This graph may include the debugger-inserted ops.
-     * </pre>
-     *
-     * <code>bytes instrumented_graph_def = 5;</code>
-     */
-    public Builder clearInstrumentedGraphDef() {
-      
-      instrumentedGraphDef_ = getDefaultInstance().getInstrumentedGraphDef();
-      onChanged();
-      return this;
-    }
-
-    private java.lang.Object outerContextId_ = "";
-    /**
-     * <pre>
-     * IDs of the immediate enclosing context (graph), if any.
-     * </pre>
-     *
-     * <code>string outer_context_id = 6;</code>
-     */
-    public java.lang.String getOuterContextId() {
-      java.lang.Object ref = outerContextId_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        outerContextId_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * IDs of the immediate enclosing context (graph), if any.
-     * </pre>
-     *
-     * <code>string outer_context_id = 6;</code>
-     */
-    public com.google.protobuf.ByteString
-        getOuterContextIdBytes() {
-      java.lang.Object ref = outerContextId_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        outerContextId_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * IDs of the immediate enclosing context (graph), if any.
-     * </pre>
-     *
-     * <code>string outer_context_id = 6;</code>
-     */
-    public Builder setOuterContextId(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      outerContextId_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * IDs of the immediate enclosing context (graph), if any.
-     * </pre>
-     *
-     * <code>string outer_context_id = 6;</code>
-     */
-    public Builder clearOuterContextId() {
-      
-      outerContextId_ = getDefaultInstance().getOuterContextId();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * IDs of the immediate enclosing context (graph), if any.
-     * </pre>
-     *
-     * <code>string outer_context_id = 6;</code>
-     */
-    public Builder setOuterContextIdBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      outerContextId_ = value;
-      onChanged();
-      return this;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.DebuggedGraph)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.DebuggedGraph)
-  private static final org.tensorflow.proto.util.DebuggedGraph DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.util.DebuggedGraph();
-  }
-
-  public static org.tensorflow.proto.util.DebuggedGraph getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<DebuggedGraph>
-      PARSER = new com.google.protobuf.AbstractParser<DebuggedGraph>() {
-    @java.lang.Override
-    public DebuggedGraph parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new DebuggedGraph(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<DebuggedGraph> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<DebuggedGraph> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.util.DebuggedGraph getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/DebuggedGraphOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/DebuggedGraphOrBuilder.java
deleted file mode 100644
index c5f78a334e6..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/DebuggedGraphOrBuilder.java
+++ /dev/null
@@ -1,123 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/debug_event.proto
-
-package org.tensorflow.proto.util;
-
-public interface DebuggedGraphOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.DebuggedGraph)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <pre>
-   * An ID for the graph.
-   * This can be used up to look up graph names. Generated by the debugger.
-   * </pre>
-   *
-   * <code>string graph_id = 1;</code>
-   */
-  java.lang.String getGraphId();
-  /**
-   * <pre>
-   * An ID for the graph.
-   * This can be used up to look up graph names. Generated by the debugger.
-   * </pre>
-   *
-   * <code>string graph_id = 1;</code>
-   */
-  com.google.protobuf.ByteString
-      getGraphIdBytes();
-
-  /**
-   * <pre>
-   * Name of the graph (if available).
-   * </pre>
-   *
-   * <code>string graph_name = 2;</code>
-   */
-  java.lang.String getGraphName();
-  /**
-   * <pre>
-   * Name of the graph (if available).
-   * </pre>
-   *
-   * <code>string graph_name = 2;</code>
-   */
-  com.google.protobuf.ByteString
-      getGraphNameBytes();
-
-  /**
-   * <pre>
-   * Names of the instrumented ops. This can be used to look up op name
-   * based on the numeric-summary tensors (2nd column).
-   * </pre>
-   *
-   * <code>repeated string instrumented_ops = 3;</code>
-   */
-  java.util.List<java.lang.String>
-      getInstrumentedOpsList();
-  /**
-   * <pre>
-   * Names of the instrumented ops. This can be used to look up op name
-   * based on the numeric-summary tensors (2nd column).
-   * </pre>
-   *
-   * <code>repeated string instrumented_ops = 3;</code>
-   */
-  int getInstrumentedOpsCount();
-  /**
-   * <pre>
-   * Names of the instrumented ops. This can be used to look up op name
-   * based on the numeric-summary tensors (2nd column).
-   * </pre>
-   *
-   * <code>repeated string instrumented_ops = 3;</code>
-   */
-  java.lang.String getInstrumentedOps(int index);
-  /**
-   * <pre>
-   * Names of the instrumented ops. This can be used to look up op name
-   * based on the numeric-summary tensors (2nd column).
-   * </pre>
-   *
-   * <code>repeated string instrumented_ops = 3;</code>
-   */
-  com.google.protobuf.ByteString
-      getInstrumentedOpsBytes(int index);
-
-  /**
-   * <pre>
-   * Original (uninstrumented) GraphDef (if available).
-   * </pre>
-   *
-   * <code>bytes original_graph_def = 4;</code>
-   */
-  com.google.protobuf.ByteString getOriginalGraphDef();
-
-  /**
-   * <pre>
-   * An encoded version of a GraphDef.
-   * This graph may include the debugger-inserted ops.
-   * </pre>
-   *
-   * <code>bytes instrumented_graph_def = 5;</code>
-   */
-  com.google.protobuf.ByteString getInstrumentedGraphDef();
-
-  /**
-   * <pre>
-   * IDs of the immediate enclosing context (graph), if any.
-   * </pre>
-   *
-   * <code>string outer_context_id = 6;</code>
-   */
-  java.lang.String getOuterContextId();
-  /**
-   * <pre>
-   * IDs of the immediate enclosing context (graph), if any.
-   * </pre>
-   *
-   * <code>string outer_context_id = 6;</code>
-   */
-  com.google.protobuf.ByteString
-      getOuterContextIdBytes();
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/Event.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/Event.java
deleted file mode 100644
index 9b30419c091..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/Event.java
+++ /dev/null
@@ -1,2049 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/util/event.proto
-
-package org.tensorflow.proto.util;
-
-/**
- * <pre>
- * Protocol buffer representing an event that happened during
- * the execution of a Brain model.
- * </pre>
- *
- * Protobuf type {@code tensorflow.Event}
- */
-public  final class Event extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.Event)
-    EventOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use Event.newBuilder() to construct.
-  private Event(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private Event() {
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new Event();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private Event(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 9: {
-
-            wallTime_ = input.readDouble();
-            break;
-          }
-          case 16: {
-
-            step_ = input.readInt64();
-            break;
-          }
-          case 26: {
-            java.lang.String s = input.readStringRequireUtf8();
-            whatCase_ = 3;
-            what_ = s;
-            break;
-          }
-          case 34: {
-            whatCase_ = 4;
-            what_ = input.readBytes();
-            break;
-          }
-          case 42: {
-            org.tensorflow.proto.framework.Summary.Builder subBuilder = null;
-            if (whatCase_ == 5) {
-              subBuilder = ((org.tensorflow.proto.framework.Summary) what_).toBuilder();
-            }
-            what_ =
-                input.readMessage(org.tensorflow.proto.framework.Summary.parser(), extensionRegistry);
-            if (subBuilder != null) {
-              subBuilder.mergeFrom((org.tensorflow.proto.framework.Summary) what_);
-              what_ = subBuilder.buildPartial();
-            }
-            whatCase_ = 5;
-            break;
-          }
-          case 50: {
-            org.tensorflow.proto.util.LogMessage.Builder subBuilder = null;
-            if (whatCase_ == 6) {
-              subBuilder = ((org.tensorflow.proto.util.LogMessage) what_).toBuilder();
-            }
-            what_ =
-                input.readMessage(org.tensorflow.proto.util.LogMessage.parser(), extensionRegistry);
-            if (subBuilder != null) {
-              subBuilder.mergeFrom((org.tensorflow.proto.util.LogMessage) what_);
-              what_ = subBuilder.buildPartial();
-            }
-            whatCase_ = 6;
-            break;
-          }
-          case 58: {
-            org.tensorflow.proto.util.SessionLog.Builder subBuilder = null;
-            if (whatCase_ == 7) {
-              subBuilder = ((org.tensorflow.proto.util.SessionLog) what_).toBuilder();
-            }
-            what_ =
-                input.readMessage(org.tensorflow.proto.util.SessionLog.parser(), extensionRegistry);
-            if (subBuilder != null) {
-              subBuilder.mergeFrom((org.tensorflow.proto.util.SessionLog) what_);
-              what_ = subBuilder.buildPartial();
-            }
-            whatCase_ = 7;
-            break;
-          }
-          case 66: {
-            org.tensorflow.proto.util.TaggedRunMetadata.Builder subBuilder = null;
-            if (whatCase_ == 8) {
-              subBuilder = ((org.tensorflow.proto.util.TaggedRunMetadata) what_).toBuilder();
-            }
-            what_ =
-                input.readMessage(org.tensorflow.proto.util.TaggedRunMetadata.parser(), extensionRegistry);
-            if (subBuilder != null) {
-              subBuilder.mergeFrom((org.tensorflow.proto.util.TaggedRunMetadata) what_);
-              what_ = subBuilder.buildPartial();
-            }
-            whatCase_ = 8;
-            break;
-          }
-          case 74: {
-            whatCase_ = 9;
-            what_ = input.readBytes();
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.util.EventProtos.internal_static_tensorflow_Event_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.util.EventProtos.internal_static_tensorflow_Event_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.util.Event.class, org.tensorflow.proto.util.Event.Builder.class);
-  }
-
-  private int whatCase_ = 0;
-  private java.lang.Object what_;
-  public enum WhatCase
-      implements com.google.protobuf.Internal.EnumLite {
-    FILE_VERSION(3),
-    GRAPH_DEF(4),
-    SUMMARY(5),
-    LOG_MESSAGE(6),
-    SESSION_LOG(7),
-    TAGGED_RUN_METADATA(8),
-    META_GRAPH_DEF(9),
-    WHAT_NOT_SET(0);
-    private final int value;
-    private WhatCase(int value) {
-      this.value = value;
-    }
-    /**
-     * @deprecated Use {@link #forNumber(int)} instead.
-     */
-    @java.lang.Deprecated
-    public static WhatCase valueOf(int value) {
-      return forNumber(value);
-    }
-
-    public static WhatCase forNumber(int value) {
-      switch (value) {
-        case 3: return FILE_VERSION;
-        case 4: return GRAPH_DEF;
-        case 5: return SUMMARY;
-        case 6: return LOG_MESSAGE;
-        case 7: return SESSION_LOG;
-        case 8: return TAGGED_RUN_METADATA;
-        case 9: return META_GRAPH_DEF;
-        case 0: return WHAT_NOT_SET;
-        default: return null;
-      }
-    }
-    public int getNumber() {
-      return this.value;
-    }
-  };
-
-  public WhatCase
-  getWhatCase() {
-    return WhatCase.forNumber(
-        whatCase_);
-  }
-
-  public static final int WALL_TIME_FIELD_NUMBER = 1;
-  private double wallTime_;
-  /**
-   * <pre>
-   * Timestamp of the event.
-   * </pre>
-   *
-   * <code>double wall_time = 1;</code>
-   */
-  public double getWallTime() {
-    return wallTime_;
-  }
-
-  public static final int STEP_FIELD_NUMBER = 2;
-  private long step_;
-  /**
-   * <pre>
-   * Global step of the event.
-   * </pre>
-   *
-   * <code>int64 step = 2;</code>
-   */
-  public long getStep() {
-    return step_;
-  }
-
-  public static final int FILE_VERSION_FIELD_NUMBER = 3;
-  /**
-   * <pre>
-   * An event file was started, with the specified version.
-   * This is use to identify the contents of the record IO files
-   * easily.  Current version is "brain.Event:2".  All versions
-   * start with "brain.Event:".
-   * </pre>
-   *
-   * <code>string file_version = 3;</code>
-   */
-  public java.lang.String getFileVersion() {
-    java.lang.Object ref = "";
-    if (whatCase_ == 3) {
-      ref = what_;
-    }
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      if (whatCase_ == 3) {
-        what_ = s;
-      }
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * An event file was started, with the specified version.
-   * This is use to identify the contents of the record IO files
-   * easily.  Current version is "brain.Event:2".  All versions
-   * start with "brain.Event:".
-   * </pre>
-   *
-   * <code>string file_version = 3;</code>
-   */
-  public com.google.protobuf.ByteString
-      getFileVersionBytes() {
-    java.lang.Object ref = "";
-    if (whatCase_ == 3) {
-      ref = what_;
-    }
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      if (whatCase_ == 3) {
-        what_ = b;
-      }
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int GRAPH_DEF_FIELD_NUMBER = 4;
-  /**
-   * <pre>
-   * An encoded version of a GraphDef.
-   * </pre>
-   *
-   * <code>bytes graph_def = 4;</code>
-   */
-  public com.google.protobuf.ByteString getGraphDef() {
-    if (whatCase_ == 4) {
-      return (com.google.protobuf.ByteString) what_;
-    }
-    return com.google.protobuf.ByteString.EMPTY;
-  }
-
-  public static final int SUMMARY_FIELD_NUMBER = 5;
-  /**
-   * <pre>
-   * A summary was generated.
-   * </pre>
-   *
-   * <code>.tensorflow.Summary summary = 5;</code>
-   */
-  public boolean hasSummary() {
-    return whatCase_ == 5;
-  }
-  /**
-   * <pre>
-   * A summary was generated.
-   * </pre>
-   *
-   * <code>.tensorflow.Summary summary = 5;</code>
-   */
-  public org.tensorflow.proto.framework.Summary getSummary() {
-    if (whatCase_ == 5) {
-       return (org.tensorflow.proto.framework.Summary) what_;
-    }
-    return org.tensorflow.proto.framework.Summary.getDefaultInstance();
-  }
-  /**
-   * <pre>
-   * A summary was generated.
-   * </pre>
-   *
-   * <code>.tensorflow.Summary summary = 5;</code>
-   */
-  public org.tensorflow.proto.framework.SummaryOrBuilder getSummaryOrBuilder() {
-    if (whatCase_ == 5) {
-       return (org.tensorflow.proto.framework.Summary) what_;
-    }
-    return org.tensorflow.proto.framework.Summary.getDefaultInstance();
-  }
-
-  public static final int LOG_MESSAGE_FIELD_NUMBER = 6;
-  /**
-   * <pre>
-   * The user output a log message. Not all messages are logged, only ones
-   * generated via the Python tensorboard_logging module.
-   * </pre>
-   *
-   * <code>.tensorflow.LogMessage log_message = 6;</code>
-   */
-  public boolean hasLogMessage() {
-    return whatCase_ == 6;
-  }
-  /**
-   * <pre>
-   * The user output a log message. Not all messages are logged, only ones
-   * generated via the Python tensorboard_logging module.
-   * </pre>
-   *
-   * <code>.tensorflow.LogMessage log_message = 6;</code>
-   */
-  public org.tensorflow.proto.util.LogMessage getLogMessage() {
-    if (whatCase_ == 6) {
-       return (org.tensorflow.proto.util.LogMessage) what_;
-    }
-    return org.tensorflow.proto.util.LogMessage.getDefaultInstance();
-  }
-  /**
-   * <pre>
-   * The user output a log message. Not all messages are logged, only ones
-   * generated via the Python tensorboard_logging module.
-   * </pre>
-   *
-   * <code>.tensorflow.LogMessage log_message = 6;</code>
-   */
-  public org.tensorflow.proto.util.LogMessageOrBuilder getLogMessageOrBuilder() {
-    if (whatCase_ == 6) {
-       return (org.tensorflow.proto.util.LogMessage) what_;
-    }
-    return org.tensorflow.proto.util.LogMessage.getDefaultInstance();
-  }
-
-  public static final int SESSION_LOG_FIELD_NUMBER = 7;
-  /**
-   * <pre>
-   * The state of the session which can be used for restarting after crashes.
-   * </pre>
-   *
-   * <code>.tensorflow.SessionLog session_log = 7;</code>
-   */
-  public boolean hasSessionLog() {
-    return whatCase_ == 7;
-  }
-  /**
-   * <pre>
-   * The state of the session which can be used for restarting after crashes.
-   * </pre>
-   *
-   * <code>.tensorflow.SessionLog session_log = 7;</code>
-   */
-  public org.tensorflow.proto.util.SessionLog getSessionLog() {
-    if (whatCase_ == 7) {
-       return (org.tensorflow.proto.util.SessionLog) what_;
-    }
-    return org.tensorflow.proto.util.SessionLog.getDefaultInstance();
-  }
-  /**
-   * <pre>
-   * The state of the session which can be used for restarting after crashes.
-   * </pre>
-   *
-   * <code>.tensorflow.SessionLog session_log = 7;</code>
-   */
-  public org.tensorflow.proto.util.SessionLogOrBuilder getSessionLogOrBuilder() {
-    if (whatCase_ == 7) {
-       return (org.tensorflow.proto.util.SessionLog) what_;
-    }
-    return org.tensorflow.proto.util.SessionLog.getDefaultInstance();
-  }
-
-  public static final int TAGGED_RUN_METADATA_FIELD_NUMBER = 8;
-  /**
-   * <pre>
-   * The metadata returned by running a session.run() call.
-   * </pre>
-   *
-   * <code>.tensorflow.TaggedRunMetadata tagged_run_metadata = 8;</code>
-   */
-  public boolean hasTaggedRunMetadata() {
-    return whatCase_ == 8;
-  }
-  /**
-   * <pre>
-   * The metadata returned by running a session.run() call.
-   * </pre>
-   *
-   * <code>.tensorflow.TaggedRunMetadata tagged_run_metadata = 8;</code>
-   */
-  public org.tensorflow.proto.util.TaggedRunMetadata getTaggedRunMetadata() {
-    if (whatCase_ == 8) {
-       return (org.tensorflow.proto.util.TaggedRunMetadata) what_;
-    }
-    return org.tensorflow.proto.util.TaggedRunMetadata.getDefaultInstance();
-  }
-  /**
-   * <pre>
-   * The metadata returned by running a session.run() call.
-   * </pre>
-   *
-   * <code>.tensorflow.TaggedRunMetadata tagged_run_metadata = 8;</code>
-   */
-  public org.tensorflow.proto.util.TaggedRunMetadataOrBuilder getTaggedRunMetadataOrBuilder() {
-    if (whatCase_ == 8) {
-       return (org.tensorflow.proto.util.TaggedRunMetadata) what_;
-    }
-    return org.tensorflow.proto.util.TaggedRunMetadata.getDefaultInstance();
-  }
-
-  public static final int META_GRAPH_DEF_FIELD_NUMBER = 9;
-  /**
-   * <pre>
-   * An encoded version of a MetaGraphDef.
-   * </pre>
-   *
-   * <code>bytes meta_graph_def = 9;</code>
-   */
-  public com.google.protobuf.ByteString getMetaGraphDef() {
-    if (whatCase_ == 9) {
-      return (com.google.protobuf.ByteString) what_;
-    }
-    return com.google.protobuf.ByteString.EMPTY;
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    if (wallTime_ != 0D) {
-      output.writeDouble(1, wallTime_);
-    }
-    if (step_ != 0L) {
-      output.writeInt64(2, step_);
-    }
-    if (whatCase_ == 3) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 3, what_);
-    }
-    if (whatCase_ == 4) {
-      output.writeBytes(
-          4, (com.google.protobuf.ByteString) what_);
-    }
-    if (whatCase_ == 5) {
-      output.writeMessage(5, (org.tensorflow.proto.framework.Summary) what_);
-    }
-    if (whatCase_ == 6) {
-      output.writeMessage(6, (org.tensorflow.proto.util.LogMessage) what_);
-    }
-    if (whatCase_ == 7) {
-      output.writeMessage(7, (org.tensorflow.proto.util.SessionLog) what_);
-    }
-    if (whatCase_ == 8) {
-      output.writeMessage(8, (org.tensorflow.proto.util.TaggedRunMetadata) what_);
-    }
-    if (whatCase_ == 9) {
-      output.writeBytes(
-          9, (com.google.protobuf.ByteString) what_);
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (wallTime_ != 0D) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeDoubleSize(1, wallTime_);
-    }
-    if (step_ != 0L) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt64Size(2, step_);
-    }
-    if (whatCase_ == 3) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(3, what_);
-    }
-    if (whatCase_ == 4) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeBytesSize(
-            4, (com.google.protobuf.ByteString) what_);
-    }
-    if (whatCase_ == 5) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(5, (org.tensorflow.proto.framework.Summary) what_);
-    }
-    if (whatCase_ == 6) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(6, (org.tensorflow.proto.util.LogMessage) what_);
-    }
-    if (whatCase_ == 7) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(7, (org.tensorflow.proto.util.SessionLog) what_);
-    }
-    if (whatCase_ == 8) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(8, (org.tensorflow.proto.util.TaggedRunMetadata) what_);
-    }
-    if (whatCase_ == 9) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeBytesSize(
-            9, (com.google.protobuf.ByteString) what_);
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.util.Event)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.util.Event other = (org.tensorflow.proto.util.Event) obj;
-
-    if (java.lang.Double.doubleToLongBits(getWallTime())
-        != java.lang.Double.doubleToLongBits(
-            other.getWallTime())) return false;
-    if (getStep()
-        != other.getStep()) return false;
-    if (!getWhatCase().equals(other.getWhatCase())) return false;
-    switch (whatCase_) {
-      case 3:
-        if (!getFileVersion()
-            .equals(other.getFileVersion())) return false;
-        break;
-      case 4:
-        if (!getGraphDef()
-            .equals(other.getGraphDef())) return false;
-        break;
-      case 5:
-        if (!getSummary()
-            .equals(other.getSummary())) return false;
-        break;
-      case 6:
-        if (!getLogMessage()
-            .equals(other.getLogMessage())) return false;
-        break;
-      case 7:
-        if (!getSessionLog()
-            .equals(other.getSessionLog())) return false;
-        break;
-      case 8:
-        if (!getTaggedRunMetadata()
-            .equals(other.getTaggedRunMetadata())) return false;
-        break;
-      case 9:
-        if (!getMetaGraphDef()
-            .equals(other.getMetaGraphDef())) return false;
-        break;
-      case 0:
-      default:
-    }
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    hash = (37 * hash) + WALL_TIME_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-        java.lang.Double.doubleToLongBits(getWallTime()));
-    hash = (37 * hash) + STEP_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-        getStep());
-    switch (whatCase_) {
-      case 3:
-        hash = (37 * hash) + FILE_VERSION_FIELD_NUMBER;
-        hash = (53 * hash) + getFileVersion().hashCode();
-        break;
-      case 4:
-        hash = (37 * hash) + GRAPH_DEF_FIELD_NUMBER;
-        hash = (53 * hash) + getGraphDef().hashCode();
-        break;
-      case 5:
-        hash = (37 * hash) + SUMMARY_FIELD_NUMBER;
-        hash = (53 * hash) + getSummary().hashCode();
-        break;
-      case 6:
-        hash = (37 * hash) + LOG_MESSAGE_FIELD_NUMBER;
-        hash = (53 * hash) + getLogMessage().hashCode();
-        break;
-      case 7:
-        hash = (37 * hash) + SESSION_LOG_FIELD_NUMBER;
-        hash = (53 * hash) + getSessionLog().hashCode();
-        break;
-      case 8:
-        hash = (37 * hash) + TAGGED_RUN_METADATA_FIELD_NUMBER;
-        hash = (53 * hash) + getTaggedRunMetadata().hashCode();
-        break;
-      case 9:
-        hash = (37 * hash) + META_GRAPH_DEF_FIELD_NUMBER;
-        hash = (53 * hash) + getMetaGraphDef().hashCode();
-        break;
-      case 0:
-      default:
-    }
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.util.Event parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.Event parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.Event parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.Event parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.Event parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.Event parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.Event parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.Event parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.Event parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.Event parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.Event parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.Event parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.util.Event prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * <pre>
-   * Protocol buffer representing an event that happened during
-   * the execution of a Brain model.
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.Event}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.Event)
-      org.tensorflow.proto.util.EventOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.util.EventProtos.internal_static_tensorflow_Event_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.util.EventProtos.internal_static_tensorflow_Event_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.util.Event.class, org.tensorflow.proto.util.Event.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.util.Event.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      wallTime_ = 0D;
-
-      step_ = 0L;
-
-      whatCase_ = 0;
-      what_ = null;
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.util.EventProtos.internal_static_tensorflow_Event_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.Event getDefaultInstanceForType() {
-      return org.tensorflow.proto.util.Event.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.Event build() {
-      org.tensorflow.proto.util.Event result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.Event buildPartial() {
-      org.tensorflow.proto.util.Event result = new org.tensorflow.proto.util.Event(this);
-      result.wallTime_ = wallTime_;
-      result.step_ = step_;
-      if (whatCase_ == 3) {
-        result.what_ = what_;
-      }
-      if (whatCase_ == 4) {
-        result.what_ = what_;
-      }
-      if (whatCase_ == 5) {
-        if (summaryBuilder_ == null) {
-          result.what_ = what_;
-        } else {
-          result.what_ = summaryBuilder_.build();
-        }
-      }
-      if (whatCase_ == 6) {
-        if (logMessageBuilder_ == null) {
-          result.what_ = what_;
-        } else {
-          result.what_ = logMessageBuilder_.build();
-        }
-      }
-      if (whatCase_ == 7) {
-        if (sessionLogBuilder_ == null) {
-          result.what_ = what_;
-        } else {
-          result.what_ = sessionLogBuilder_.build();
-        }
-      }
-      if (whatCase_ == 8) {
-        if (taggedRunMetadataBuilder_ == null) {
-          result.what_ = what_;
-        } else {
-          result.what_ = taggedRunMetadataBuilder_.build();
-        }
-      }
-      if (whatCase_ == 9) {
-        result.what_ = what_;
-      }
-      result.whatCase_ = whatCase_;
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.util.Event) {
-        return mergeFrom((org.tensorflow.proto.util.Event)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.util.Event other) {
-      if (other == org.tensorflow.proto.util.Event.getDefaultInstance()) return this;
-      if (other.getWallTime() != 0D) {
-        setWallTime(other.getWallTime());
-      }
-      if (other.getStep() != 0L) {
-        setStep(other.getStep());
-      }
-      switch (other.getWhatCase()) {
-        case FILE_VERSION: {
-          whatCase_ = 3;
-          what_ = other.what_;
-          onChanged();
-          break;
-        }
-        case GRAPH_DEF: {
-          setGraphDef(other.getGraphDef());
-          break;
-        }
-        case SUMMARY: {
-          mergeSummary(other.getSummary());
-          break;
-        }
-        case LOG_MESSAGE: {
-          mergeLogMessage(other.getLogMessage());
-          break;
-        }
-        case SESSION_LOG: {
-          mergeSessionLog(other.getSessionLog());
-          break;
-        }
-        case TAGGED_RUN_METADATA: {
-          mergeTaggedRunMetadata(other.getTaggedRunMetadata());
-          break;
-        }
-        case META_GRAPH_DEF: {
-          setMetaGraphDef(other.getMetaGraphDef());
-          break;
-        }
-        case WHAT_NOT_SET: {
-          break;
-        }
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.util.Event parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.util.Event) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-    private int whatCase_ = 0;
-    private java.lang.Object what_;
-    public WhatCase
-        getWhatCase() {
-      return WhatCase.forNumber(
-          whatCase_);
-    }
-
-    public Builder clearWhat() {
-      whatCase_ = 0;
-      what_ = null;
-      onChanged();
-      return this;
-    }
-
-
-    private double wallTime_ ;
-    /**
-     * <pre>
-     * Timestamp of the event.
-     * </pre>
-     *
-     * <code>double wall_time = 1;</code>
-     */
-    public double getWallTime() {
-      return wallTime_;
-    }
-    /**
-     * <pre>
-     * Timestamp of the event.
-     * </pre>
-     *
-     * <code>double wall_time = 1;</code>
-     */
-    public Builder setWallTime(double value) {
-      
-      wallTime_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Timestamp of the event.
-     * </pre>
-     *
-     * <code>double wall_time = 1;</code>
-     */
-    public Builder clearWallTime() {
-      
-      wallTime_ = 0D;
-      onChanged();
-      return this;
-    }
-
-    private long step_ ;
-    /**
-     * <pre>
-     * Global step of the event.
-     * </pre>
-     *
-     * <code>int64 step = 2;</code>
-     */
-    public long getStep() {
-      return step_;
-    }
-    /**
-     * <pre>
-     * Global step of the event.
-     * </pre>
-     *
-     * <code>int64 step = 2;</code>
-     */
-    public Builder setStep(long value) {
-      
-      step_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Global step of the event.
-     * </pre>
-     *
-     * <code>int64 step = 2;</code>
-     */
-    public Builder clearStep() {
-      
-      step_ = 0L;
-      onChanged();
-      return this;
-    }
-
-    /**
-     * <pre>
-     * An event file was started, with the specified version.
-     * This is use to identify the contents of the record IO files
-     * easily.  Current version is "brain.Event:2".  All versions
-     * start with "brain.Event:".
-     * </pre>
-     *
-     * <code>string file_version = 3;</code>
-     */
-    public java.lang.String getFileVersion() {
-      java.lang.Object ref = "";
-      if (whatCase_ == 3) {
-        ref = what_;
-      }
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        if (whatCase_ == 3) {
-          what_ = s;
-        }
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * An event file was started, with the specified version.
-     * This is use to identify the contents of the record IO files
-     * easily.  Current version is "brain.Event:2".  All versions
-     * start with "brain.Event:".
-     * </pre>
-     *
-     * <code>string file_version = 3;</code>
-     */
-    public com.google.protobuf.ByteString
-        getFileVersionBytes() {
-      java.lang.Object ref = "";
-      if (whatCase_ == 3) {
-        ref = what_;
-      }
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        if (whatCase_ == 3) {
-          what_ = b;
-        }
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * An event file was started, with the specified version.
-     * This is use to identify the contents of the record IO files
-     * easily.  Current version is "brain.Event:2".  All versions
-     * start with "brain.Event:".
-     * </pre>
-     *
-     * <code>string file_version = 3;</code>
-     */
-    public Builder setFileVersion(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  whatCase_ = 3;
-      what_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * An event file was started, with the specified version.
-     * This is use to identify the contents of the record IO files
-     * easily.  Current version is "brain.Event:2".  All versions
-     * start with "brain.Event:".
-     * </pre>
-     *
-     * <code>string file_version = 3;</code>
-     */
-    public Builder clearFileVersion() {
-      if (whatCase_ == 3) {
-        whatCase_ = 0;
-        what_ = null;
-        onChanged();
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * An event file was started, with the specified version.
-     * This is use to identify the contents of the record IO files
-     * easily.  Current version is "brain.Event:2".  All versions
-     * start with "brain.Event:".
-     * </pre>
-     *
-     * <code>string file_version = 3;</code>
-     */
-    public Builder setFileVersionBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      whatCase_ = 3;
-      what_ = value;
-      onChanged();
-      return this;
-    }
-
-    /**
-     * <pre>
-     * An encoded version of a GraphDef.
-     * </pre>
-     *
-     * <code>bytes graph_def = 4;</code>
-     */
-    public com.google.protobuf.ByteString getGraphDef() {
-      if (whatCase_ == 4) {
-        return (com.google.protobuf.ByteString) what_;
-      }
-      return com.google.protobuf.ByteString.EMPTY;
-    }
-    /**
-     * <pre>
-     * An encoded version of a GraphDef.
-     * </pre>
-     *
-     * <code>bytes graph_def = 4;</code>
-     */
-    public Builder setGraphDef(com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  whatCase_ = 4;
-      what_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * An encoded version of a GraphDef.
-     * </pre>
-     *
-     * <code>bytes graph_def = 4;</code>
-     */
-    public Builder clearGraphDef() {
-      if (whatCase_ == 4) {
-        whatCase_ = 0;
-        what_ = null;
-        onChanged();
-      }
-      return this;
-    }
-
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.Summary, org.tensorflow.proto.framework.Summary.Builder, org.tensorflow.proto.framework.SummaryOrBuilder> summaryBuilder_;
-    /**
-     * <pre>
-     * A summary was generated.
-     * </pre>
-     *
-     * <code>.tensorflow.Summary summary = 5;</code>
-     */
-    public boolean hasSummary() {
-      return whatCase_ == 5;
-    }
-    /**
-     * <pre>
-     * A summary was generated.
-     * </pre>
-     *
-     * <code>.tensorflow.Summary summary = 5;</code>
-     */
-    public org.tensorflow.proto.framework.Summary getSummary() {
-      if (summaryBuilder_ == null) {
-        if (whatCase_ == 5) {
-          return (org.tensorflow.proto.framework.Summary) what_;
-        }
-        return org.tensorflow.proto.framework.Summary.getDefaultInstance();
-      } else {
-        if (whatCase_ == 5) {
-          return summaryBuilder_.getMessage();
-        }
-        return org.tensorflow.proto.framework.Summary.getDefaultInstance();
-      }
-    }
-    /**
-     * <pre>
-     * A summary was generated.
-     * </pre>
-     *
-     * <code>.tensorflow.Summary summary = 5;</code>
-     */
-    public Builder setSummary(org.tensorflow.proto.framework.Summary value) {
-      if (summaryBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        what_ = value;
-        onChanged();
-      } else {
-        summaryBuilder_.setMessage(value);
-      }
-      whatCase_ = 5;
-      return this;
-    }
-    /**
-     * <pre>
-     * A summary was generated.
-     * </pre>
-     *
-     * <code>.tensorflow.Summary summary = 5;</code>
-     */
-    public Builder setSummary(
-        org.tensorflow.proto.framework.Summary.Builder builderForValue) {
-      if (summaryBuilder_ == null) {
-        what_ = builderForValue.build();
-        onChanged();
-      } else {
-        summaryBuilder_.setMessage(builderForValue.build());
-      }
-      whatCase_ = 5;
-      return this;
-    }
-    /**
-     * <pre>
-     * A summary was generated.
-     * </pre>
-     *
-     * <code>.tensorflow.Summary summary = 5;</code>
-     */
-    public Builder mergeSummary(org.tensorflow.proto.framework.Summary value) {
-      if (summaryBuilder_ == null) {
-        if (whatCase_ == 5 &&
-            what_ != org.tensorflow.proto.framework.Summary.getDefaultInstance()) {
-          what_ = org.tensorflow.proto.framework.Summary.newBuilder((org.tensorflow.proto.framework.Summary) what_)
-              .mergeFrom(value).buildPartial();
-        } else {
-          what_ = value;
-        }
-        onChanged();
-      } else {
-        if (whatCase_ == 5) {
-          summaryBuilder_.mergeFrom(value);
-        }
-        summaryBuilder_.setMessage(value);
-      }
-      whatCase_ = 5;
-      return this;
-    }
-    /**
-     * <pre>
-     * A summary was generated.
-     * </pre>
-     *
-     * <code>.tensorflow.Summary summary = 5;</code>
-     */
-    public Builder clearSummary() {
-      if (summaryBuilder_ == null) {
-        if (whatCase_ == 5) {
-          whatCase_ = 0;
-          what_ = null;
-          onChanged();
-        }
-      } else {
-        if (whatCase_ == 5) {
-          whatCase_ = 0;
-          what_ = null;
-        }
-        summaryBuilder_.clear();
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * A summary was generated.
-     * </pre>
-     *
-     * <code>.tensorflow.Summary summary = 5;</code>
-     */
-    public org.tensorflow.proto.framework.Summary.Builder getSummaryBuilder() {
-      return getSummaryFieldBuilder().getBuilder();
-    }
-    /**
-     * <pre>
-     * A summary was generated.
-     * </pre>
-     *
-     * <code>.tensorflow.Summary summary = 5;</code>
-     */
-    public org.tensorflow.proto.framework.SummaryOrBuilder getSummaryOrBuilder() {
-      if ((whatCase_ == 5) && (summaryBuilder_ != null)) {
-        return summaryBuilder_.getMessageOrBuilder();
-      } else {
-        if (whatCase_ == 5) {
-          return (org.tensorflow.proto.framework.Summary) what_;
-        }
-        return org.tensorflow.proto.framework.Summary.getDefaultInstance();
-      }
-    }
-    /**
-     * <pre>
-     * A summary was generated.
-     * </pre>
-     *
-     * <code>.tensorflow.Summary summary = 5;</code>
-     */
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.Summary, org.tensorflow.proto.framework.Summary.Builder, org.tensorflow.proto.framework.SummaryOrBuilder> 
-        getSummaryFieldBuilder() {
-      if (summaryBuilder_ == null) {
-        if (!(whatCase_ == 5)) {
-          what_ = org.tensorflow.proto.framework.Summary.getDefaultInstance();
-        }
-        summaryBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-            org.tensorflow.proto.framework.Summary, org.tensorflow.proto.framework.Summary.Builder, org.tensorflow.proto.framework.SummaryOrBuilder>(
-                (org.tensorflow.proto.framework.Summary) what_,
-                getParentForChildren(),
-                isClean());
-        what_ = null;
-      }
-      whatCase_ = 5;
-      onChanged();;
-      return summaryBuilder_;
-    }
-
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.util.LogMessage, org.tensorflow.proto.util.LogMessage.Builder, org.tensorflow.proto.util.LogMessageOrBuilder> logMessageBuilder_;
-    /**
-     * <pre>
-     * The user output a log message. Not all messages are logged, only ones
-     * generated via the Python tensorboard_logging module.
-     * </pre>
-     *
-     * <code>.tensorflow.LogMessage log_message = 6;</code>
-     */
-    public boolean hasLogMessage() {
-      return whatCase_ == 6;
-    }
-    /**
-     * <pre>
-     * The user output a log message. Not all messages are logged, only ones
-     * generated via the Python tensorboard_logging module.
-     * </pre>
-     *
-     * <code>.tensorflow.LogMessage log_message = 6;</code>
-     */
-    public org.tensorflow.proto.util.LogMessage getLogMessage() {
-      if (logMessageBuilder_ == null) {
-        if (whatCase_ == 6) {
-          return (org.tensorflow.proto.util.LogMessage) what_;
-        }
-        return org.tensorflow.proto.util.LogMessage.getDefaultInstance();
-      } else {
-        if (whatCase_ == 6) {
-          return logMessageBuilder_.getMessage();
-        }
-        return org.tensorflow.proto.util.LogMessage.getDefaultInstance();
-      }
-    }
-    /**
-     * <pre>
-     * The user output a log message. Not all messages are logged, only ones
-     * generated via the Python tensorboard_logging module.
-     * </pre>
-     *
-     * <code>.tensorflow.LogMessage log_message = 6;</code>
-     */
-    public Builder setLogMessage(org.tensorflow.proto.util.LogMessage value) {
-      if (logMessageBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        what_ = value;
-        onChanged();
-      } else {
-        logMessageBuilder_.setMessage(value);
-      }
-      whatCase_ = 6;
-      return this;
-    }
-    /**
-     * <pre>
-     * The user output a log message. Not all messages are logged, only ones
-     * generated via the Python tensorboard_logging module.
-     * </pre>
-     *
-     * <code>.tensorflow.LogMessage log_message = 6;</code>
-     */
-    public Builder setLogMessage(
-        org.tensorflow.proto.util.LogMessage.Builder builderForValue) {
-      if (logMessageBuilder_ == null) {
-        what_ = builderForValue.build();
-        onChanged();
-      } else {
-        logMessageBuilder_.setMessage(builderForValue.build());
-      }
-      whatCase_ = 6;
-      return this;
-    }
-    /**
-     * <pre>
-     * The user output a log message. Not all messages are logged, only ones
-     * generated via the Python tensorboard_logging module.
-     * </pre>
-     *
-     * <code>.tensorflow.LogMessage log_message = 6;</code>
-     */
-    public Builder mergeLogMessage(org.tensorflow.proto.util.LogMessage value) {
-      if (logMessageBuilder_ == null) {
-        if (whatCase_ == 6 &&
-            what_ != org.tensorflow.proto.util.LogMessage.getDefaultInstance()) {
-          what_ = org.tensorflow.proto.util.LogMessage.newBuilder((org.tensorflow.proto.util.LogMessage) what_)
-              .mergeFrom(value).buildPartial();
-        } else {
-          what_ = value;
-        }
-        onChanged();
-      } else {
-        if (whatCase_ == 6) {
-          logMessageBuilder_.mergeFrom(value);
-        }
-        logMessageBuilder_.setMessage(value);
-      }
-      whatCase_ = 6;
-      return this;
-    }
-    /**
-     * <pre>
-     * The user output a log message. Not all messages are logged, only ones
-     * generated via the Python tensorboard_logging module.
-     * </pre>
-     *
-     * <code>.tensorflow.LogMessage log_message = 6;</code>
-     */
-    public Builder clearLogMessage() {
-      if (logMessageBuilder_ == null) {
-        if (whatCase_ == 6) {
-          whatCase_ = 0;
-          what_ = null;
-          onChanged();
-        }
-      } else {
-        if (whatCase_ == 6) {
-          whatCase_ = 0;
-          what_ = null;
-        }
-        logMessageBuilder_.clear();
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * The user output a log message. Not all messages are logged, only ones
-     * generated via the Python tensorboard_logging module.
-     * </pre>
-     *
-     * <code>.tensorflow.LogMessage log_message = 6;</code>
-     */
-    public org.tensorflow.proto.util.LogMessage.Builder getLogMessageBuilder() {
-      return getLogMessageFieldBuilder().getBuilder();
-    }
-    /**
-     * <pre>
-     * The user output a log message. Not all messages are logged, only ones
-     * generated via the Python tensorboard_logging module.
-     * </pre>
-     *
-     * <code>.tensorflow.LogMessage log_message = 6;</code>
-     */
-    public org.tensorflow.proto.util.LogMessageOrBuilder getLogMessageOrBuilder() {
-      if ((whatCase_ == 6) && (logMessageBuilder_ != null)) {
-        return logMessageBuilder_.getMessageOrBuilder();
-      } else {
-        if (whatCase_ == 6) {
-          return (org.tensorflow.proto.util.LogMessage) what_;
-        }
-        return org.tensorflow.proto.util.LogMessage.getDefaultInstance();
-      }
-    }
-    /**
-     * <pre>
-     * The user output a log message. Not all messages are logged, only ones
-     * generated via the Python tensorboard_logging module.
-     * </pre>
-     *
-     * <code>.tensorflow.LogMessage log_message = 6;</code>
-     */
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.util.LogMessage, org.tensorflow.proto.util.LogMessage.Builder, org.tensorflow.proto.util.LogMessageOrBuilder> 
-        getLogMessageFieldBuilder() {
-      if (logMessageBuilder_ == null) {
-        if (!(whatCase_ == 6)) {
-          what_ = org.tensorflow.proto.util.LogMessage.getDefaultInstance();
-        }
-        logMessageBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-            org.tensorflow.proto.util.LogMessage, org.tensorflow.proto.util.LogMessage.Builder, org.tensorflow.proto.util.LogMessageOrBuilder>(
-                (org.tensorflow.proto.util.LogMessage) what_,
-                getParentForChildren(),
-                isClean());
-        what_ = null;
-      }
-      whatCase_ = 6;
-      onChanged();;
-      return logMessageBuilder_;
-    }
-
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.util.SessionLog, org.tensorflow.proto.util.SessionLog.Builder, org.tensorflow.proto.util.SessionLogOrBuilder> sessionLogBuilder_;
-    /**
-     * <pre>
-     * The state of the session which can be used for restarting after crashes.
-     * </pre>
-     *
-     * <code>.tensorflow.SessionLog session_log = 7;</code>
-     */
-    public boolean hasSessionLog() {
-      return whatCase_ == 7;
-    }
-    /**
-     * <pre>
-     * The state of the session which can be used for restarting after crashes.
-     * </pre>
-     *
-     * <code>.tensorflow.SessionLog session_log = 7;</code>
-     */
-    public org.tensorflow.proto.util.SessionLog getSessionLog() {
-      if (sessionLogBuilder_ == null) {
-        if (whatCase_ == 7) {
-          return (org.tensorflow.proto.util.SessionLog) what_;
-        }
-        return org.tensorflow.proto.util.SessionLog.getDefaultInstance();
-      } else {
-        if (whatCase_ == 7) {
-          return sessionLogBuilder_.getMessage();
-        }
-        return org.tensorflow.proto.util.SessionLog.getDefaultInstance();
-      }
-    }
-    /**
-     * <pre>
-     * The state of the session which can be used for restarting after crashes.
-     * </pre>
-     *
-     * <code>.tensorflow.SessionLog session_log = 7;</code>
-     */
-    public Builder setSessionLog(org.tensorflow.proto.util.SessionLog value) {
-      if (sessionLogBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        what_ = value;
-        onChanged();
-      } else {
-        sessionLogBuilder_.setMessage(value);
-      }
-      whatCase_ = 7;
-      return this;
-    }
-    /**
-     * <pre>
-     * The state of the session which can be used for restarting after crashes.
-     * </pre>
-     *
-     * <code>.tensorflow.SessionLog session_log = 7;</code>
-     */
-    public Builder setSessionLog(
-        org.tensorflow.proto.util.SessionLog.Builder builderForValue) {
-      if (sessionLogBuilder_ == null) {
-        what_ = builderForValue.build();
-        onChanged();
-      } else {
-        sessionLogBuilder_.setMessage(builderForValue.build());
-      }
-      whatCase_ = 7;
-      return this;
-    }
-    /**
-     * <pre>
-     * The state of the session which can be used for restarting after crashes.
-     * </pre>
-     *
-     * <code>.tensorflow.SessionLog session_log = 7;</code>
-     */
-    public Builder mergeSessionLog(org.tensorflow.proto.util.SessionLog value) {
-      if (sessionLogBuilder_ == null) {
-        if (whatCase_ == 7 &&
-            what_ != org.tensorflow.proto.util.SessionLog.getDefaultInstance()) {
-          what_ = org.tensorflow.proto.util.SessionLog.newBuilder((org.tensorflow.proto.util.SessionLog) what_)
-              .mergeFrom(value).buildPartial();
-        } else {
-          what_ = value;
-        }
-        onChanged();
-      } else {
-        if (whatCase_ == 7) {
-          sessionLogBuilder_.mergeFrom(value);
-        }
-        sessionLogBuilder_.setMessage(value);
-      }
-      whatCase_ = 7;
-      return this;
-    }
-    /**
-     * <pre>
-     * The state of the session which can be used for restarting after crashes.
-     * </pre>
-     *
-     * <code>.tensorflow.SessionLog session_log = 7;</code>
-     */
-    public Builder clearSessionLog() {
-      if (sessionLogBuilder_ == null) {
-        if (whatCase_ == 7) {
-          whatCase_ = 0;
-          what_ = null;
-          onChanged();
-        }
-      } else {
-        if (whatCase_ == 7) {
-          whatCase_ = 0;
-          what_ = null;
-        }
-        sessionLogBuilder_.clear();
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * The state of the session which can be used for restarting after crashes.
-     * </pre>
-     *
-     * <code>.tensorflow.SessionLog session_log = 7;</code>
-     */
-    public org.tensorflow.proto.util.SessionLog.Builder getSessionLogBuilder() {
-      return getSessionLogFieldBuilder().getBuilder();
-    }
-    /**
-     * <pre>
-     * The state of the session which can be used for restarting after crashes.
-     * </pre>
-     *
-     * <code>.tensorflow.SessionLog session_log = 7;</code>
-     */
-    public org.tensorflow.proto.util.SessionLogOrBuilder getSessionLogOrBuilder() {
-      if ((whatCase_ == 7) && (sessionLogBuilder_ != null)) {
-        return sessionLogBuilder_.getMessageOrBuilder();
-      } else {
-        if (whatCase_ == 7) {
-          return (org.tensorflow.proto.util.SessionLog) what_;
-        }
-        return org.tensorflow.proto.util.SessionLog.getDefaultInstance();
-      }
-    }
-    /**
-     * <pre>
-     * The state of the session which can be used for restarting after crashes.
-     * </pre>
-     *
-     * <code>.tensorflow.SessionLog session_log = 7;</code>
-     */
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.util.SessionLog, org.tensorflow.proto.util.SessionLog.Builder, org.tensorflow.proto.util.SessionLogOrBuilder> 
-        getSessionLogFieldBuilder() {
-      if (sessionLogBuilder_ == null) {
-        if (!(whatCase_ == 7)) {
-          what_ = org.tensorflow.proto.util.SessionLog.getDefaultInstance();
-        }
-        sessionLogBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-            org.tensorflow.proto.util.SessionLog, org.tensorflow.proto.util.SessionLog.Builder, org.tensorflow.proto.util.SessionLogOrBuilder>(
-                (org.tensorflow.proto.util.SessionLog) what_,
-                getParentForChildren(),
-                isClean());
-        what_ = null;
-      }
-      whatCase_ = 7;
-      onChanged();;
-      return sessionLogBuilder_;
-    }
-
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.util.TaggedRunMetadata, org.tensorflow.proto.util.TaggedRunMetadata.Builder, org.tensorflow.proto.util.TaggedRunMetadataOrBuilder> taggedRunMetadataBuilder_;
-    /**
-     * <pre>
-     * The metadata returned by running a session.run() call.
-     * </pre>
-     *
-     * <code>.tensorflow.TaggedRunMetadata tagged_run_metadata = 8;</code>
-     */
-    public boolean hasTaggedRunMetadata() {
-      return whatCase_ == 8;
-    }
-    /**
-     * <pre>
-     * The metadata returned by running a session.run() call.
-     * </pre>
-     *
-     * <code>.tensorflow.TaggedRunMetadata tagged_run_metadata = 8;</code>
-     */
-    public org.tensorflow.proto.util.TaggedRunMetadata getTaggedRunMetadata() {
-      if (taggedRunMetadataBuilder_ == null) {
-        if (whatCase_ == 8) {
-          return (org.tensorflow.proto.util.TaggedRunMetadata) what_;
-        }
-        return org.tensorflow.proto.util.TaggedRunMetadata.getDefaultInstance();
-      } else {
-        if (whatCase_ == 8) {
-          return taggedRunMetadataBuilder_.getMessage();
-        }
-        return org.tensorflow.proto.util.TaggedRunMetadata.getDefaultInstance();
-      }
-    }
-    /**
-     * <pre>
-     * The metadata returned by running a session.run() call.
-     * </pre>
-     *
-     * <code>.tensorflow.TaggedRunMetadata tagged_run_metadata = 8;</code>
-     */
-    public Builder setTaggedRunMetadata(org.tensorflow.proto.util.TaggedRunMetadata value) {
-      if (taggedRunMetadataBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        what_ = value;
-        onChanged();
-      } else {
-        taggedRunMetadataBuilder_.setMessage(value);
-      }
-      whatCase_ = 8;
-      return this;
-    }
-    /**
-     * <pre>
-     * The metadata returned by running a session.run() call.
-     * </pre>
-     *
-     * <code>.tensorflow.TaggedRunMetadata tagged_run_metadata = 8;</code>
-     */
-    public Builder setTaggedRunMetadata(
-        org.tensorflow.proto.util.TaggedRunMetadata.Builder builderForValue) {
-      if (taggedRunMetadataBuilder_ == null) {
-        what_ = builderForValue.build();
-        onChanged();
-      } else {
-        taggedRunMetadataBuilder_.setMessage(builderForValue.build());
-      }
-      whatCase_ = 8;
-      return this;
-    }
-    /**
-     * <pre>
-     * The metadata returned by running a session.run() call.
-     * </pre>
-     *
-     * <code>.tensorflow.TaggedRunMetadata tagged_run_metadata = 8;</code>
-     */
-    public Builder mergeTaggedRunMetadata(org.tensorflow.proto.util.TaggedRunMetadata value) {
-      if (taggedRunMetadataBuilder_ == null) {
-        if (whatCase_ == 8 &&
-            what_ != org.tensorflow.proto.util.TaggedRunMetadata.getDefaultInstance()) {
-          what_ = org.tensorflow.proto.util.TaggedRunMetadata.newBuilder((org.tensorflow.proto.util.TaggedRunMetadata) what_)
-              .mergeFrom(value).buildPartial();
-        } else {
-          what_ = value;
-        }
-        onChanged();
-      } else {
-        if (whatCase_ == 8) {
-          taggedRunMetadataBuilder_.mergeFrom(value);
-        }
-        taggedRunMetadataBuilder_.setMessage(value);
-      }
-      whatCase_ = 8;
-      return this;
-    }
-    /**
-     * <pre>
-     * The metadata returned by running a session.run() call.
-     * </pre>
-     *
-     * <code>.tensorflow.TaggedRunMetadata tagged_run_metadata = 8;</code>
-     */
-    public Builder clearTaggedRunMetadata() {
-      if (taggedRunMetadataBuilder_ == null) {
-        if (whatCase_ == 8) {
-          whatCase_ = 0;
-          what_ = null;
-          onChanged();
-        }
-      } else {
-        if (whatCase_ == 8) {
-          whatCase_ = 0;
-          what_ = null;
-        }
-        taggedRunMetadataBuilder_.clear();
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * The metadata returned by running a session.run() call.
-     * </pre>
-     *
-     * <code>.tensorflow.TaggedRunMetadata tagged_run_metadata = 8;</code>
-     */
-    public org.tensorflow.proto.util.TaggedRunMetadata.Builder getTaggedRunMetadataBuilder() {
-      return getTaggedRunMetadataFieldBuilder().getBuilder();
-    }
-    /**
-     * <pre>
-     * The metadata returned by running a session.run() call.
-     * </pre>
-     *
-     * <code>.tensorflow.TaggedRunMetadata tagged_run_metadata = 8;</code>
-     */
-    public org.tensorflow.proto.util.TaggedRunMetadataOrBuilder getTaggedRunMetadataOrBuilder() {
-      if ((whatCase_ == 8) && (taggedRunMetadataBuilder_ != null)) {
-        return taggedRunMetadataBuilder_.getMessageOrBuilder();
-      } else {
-        if (whatCase_ == 8) {
-          return (org.tensorflow.proto.util.TaggedRunMetadata) what_;
-        }
-        return org.tensorflow.proto.util.TaggedRunMetadata.getDefaultInstance();
-      }
-    }
-    /**
-     * <pre>
-     * The metadata returned by running a session.run() call.
-     * </pre>
-     *
-     * <code>.tensorflow.TaggedRunMetadata tagged_run_metadata = 8;</code>
-     */
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.util.TaggedRunMetadata, org.tensorflow.proto.util.TaggedRunMetadata.Builder, org.tensorflow.proto.util.TaggedRunMetadataOrBuilder> 
-        getTaggedRunMetadataFieldBuilder() {
-      if (taggedRunMetadataBuilder_ == null) {
-        if (!(whatCase_ == 8)) {
-          what_ = org.tensorflow.proto.util.TaggedRunMetadata.getDefaultInstance();
-        }
-        taggedRunMetadataBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-            org.tensorflow.proto.util.TaggedRunMetadata, org.tensorflow.proto.util.TaggedRunMetadata.Builder, org.tensorflow.proto.util.TaggedRunMetadataOrBuilder>(
-                (org.tensorflow.proto.util.TaggedRunMetadata) what_,
-                getParentForChildren(),
-                isClean());
-        what_ = null;
-      }
-      whatCase_ = 8;
-      onChanged();;
-      return taggedRunMetadataBuilder_;
-    }
-
-    /**
-     * <pre>
-     * An encoded version of a MetaGraphDef.
-     * </pre>
-     *
-     * <code>bytes meta_graph_def = 9;</code>
-     */
-    public com.google.protobuf.ByteString getMetaGraphDef() {
-      if (whatCase_ == 9) {
-        return (com.google.protobuf.ByteString) what_;
-      }
-      return com.google.protobuf.ByteString.EMPTY;
-    }
-    /**
-     * <pre>
-     * An encoded version of a MetaGraphDef.
-     * </pre>
-     *
-     * <code>bytes meta_graph_def = 9;</code>
-     */
-    public Builder setMetaGraphDef(com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  whatCase_ = 9;
-      what_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * An encoded version of a MetaGraphDef.
-     * </pre>
-     *
-     * <code>bytes meta_graph_def = 9;</code>
-     */
-    public Builder clearMetaGraphDef() {
-      if (whatCase_ == 9) {
-        whatCase_ = 0;
-        what_ = null;
-        onChanged();
-      }
-      return this;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.Event)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.Event)
-  private static final org.tensorflow.proto.util.Event DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.util.Event();
-  }
-
-  public static org.tensorflow.proto.util.Event getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<Event>
-      PARSER = new com.google.protobuf.AbstractParser<Event>() {
-    @java.lang.Override
-    public Event parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new Event(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<Event> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<Event> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.util.Event getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/EventOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/EventOrBuilder.java
deleted file mode 100644
index dbe86dad114..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/EventOrBuilder.java
+++ /dev/null
@@ -1,174 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/util/event.proto
-
-package org.tensorflow.proto.util;
-
-public interface EventOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.Event)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <pre>
-   * Timestamp of the event.
-   * </pre>
-   *
-   * <code>double wall_time = 1;</code>
-   */
-  double getWallTime();
-
-  /**
-   * <pre>
-   * Global step of the event.
-   * </pre>
-   *
-   * <code>int64 step = 2;</code>
-   */
-  long getStep();
-
-  /**
-   * <pre>
-   * An event file was started, with the specified version.
-   * This is use to identify the contents of the record IO files
-   * easily.  Current version is "brain.Event:2".  All versions
-   * start with "brain.Event:".
-   * </pre>
-   *
-   * <code>string file_version = 3;</code>
-   */
-  java.lang.String getFileVersion();
-  /**
-   * <pre>
-   * An event file was started, with the specified version.
-   * This is use to identify the contents of the record IO files
-   * easily.  Current version is "brain.Event:2".  All versions
-   * start with "brain.Event:".
-   * </pre>
-   *
-   * <code>string file_version = 3;</code>
-   */
-  com.google.protobuf.ByteString
-      getFileVersionBytes();
-
-  /**
-   * <pre>
-   * An encoded version of a GraphDef.
-   * </pre>
-   *
-   * <code>bytes graph_def = 4;</code>
-   */
-  com.google.protobuf.ByteString getGraphDef();
-
-  /**
-   * <pre>
-   * A summary was generated.
-   * </pre>
-   *
-   * <code>.tensorflow.Summary summary = 5;</code>
-   */
-  boolean hasSummary();
-  /**
-   * <pre>
-   * A summary was generated.
-   * </pre>
-   *
-   * <code>.tensorflow.Summary summary = 5;</code>
-   */
-  org.tensorflow.proto.framework.Summary getSummary();
-  /**
-   * <pre>
-   * A summary was generated.
-   * </pre>
-   *
-   * <code>.tensorflow.Summary summary = 5;</code>
-   */
-  org.tensorflow.proto.framework.SummaryOrBuilder getSummaryOrBuilder();
-
-  /**
-   * <pre>
-   * The user output a log message. Not all messages are logged, only ones
-   * generated via the Python tensorboard_logging module.
-   * </pre>
-   *
-   * <code>.tensorflow.LogMessage log_message = 6;</code>
-   */
-  boolean hasLogMessage();
-  /**
-   * <pre>
-   * The user output a log message. Not all messages are logged, only ones
-   * generated via the Python tensorboard_logging module.
-   * </pre>
-   *
-   * <code>.tensorflow.LogMessage log_message = 6;</code>
-   */
-  org.tensorflow.proto.util.LogMessage getLogMessage();
-  /**
-   * <pre>
-   * The user output a log message. Not all messages are logged, only ones
-   * generated via the Python tensorboard_logging module.
-   * </pre>
-   *
-   * <code>.tensorflow.LogMessage log_message = 6;</code>
-   */
-  org.tensorflow.proto.util.LogMessageOrBuilder getLogMessageOrBuilder();
-
-  /**
-   * <pre>
-   * The state of the session which can be used for restarting after crashes.
-   * </pre>
-   *
-   * <code>.tensorflow.SessionLog session_log = 7;</code>
-   */
-  boolean hasSessionLog();
-  /**
-   * <pre>
-   * The state of the session which can be used for restarting after crashes.
-   * </pre>
-   *
-   * <code>.tensorflow.SessionLog session_log = 7;</code>
-   */
-  org.tensorflow.proto.util.SessionLog getSessionLog();
-  /**
-   * <pre>
-   * The state of the session which can be used for restarting after crashes.
-   * </pre>
-   *
-   * <code>.tensorflow.SessionLog session_log = 7;</code>
-   */
-  org.tensorflow.proto.util.SessionLogOrBuilder getSessionLogOrBuilder();
-
-  /**
-   * <pre>
-   * The metadata returned by running a session.run() call.
-   * </pre>
-   *
-   * <code>.tensorflow.TaggedRunMetadata tagged_run_metadata = 8;</code>
-   */
-  boolean hasTaggedRunMetadata();
-  /**
-   * <pre>
-   * The metadata returned by running a session.run() call.
-   * </pre>
-   *
-   * <code>.tensorflow.TaggedRunMetadata tagged_run_metadata = 8;</code>
-   */
-  org.tensorflow.proto.util.TaggedRunMetadata getTaggedRunMetadata();
-  /**
-   * <pre>
-   * The metadata returned by running a session.run() call.
-   * </pre>
-   *
-   * <code>.tensorflow.TaggedRunMetadata tagged_run_metadata = 8;</code>
-   */
-  org.tensorflow.proto.util.TaggedRunMetadataOrBuilder getTaggedRunMetadataOrBuilder();
-
-  /**
-   * <pre>
-   * An encoded version of a MetaGraphDef.
-   * </pre>
-   *
-   * <code>bytes meta_graph_def = 9;</code>
-   */
-  com.google.protobuf.ByteString getMetaGraphDef();
-
-  public org.tensorflow.proto.util.Event.WhatCase getWhatCase();
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/EventProtos.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/EventProtos.java
deleted file mode 100644
index 121a7223e1d..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/EventProtos.java
+++ /dev/null
@@ -1,161 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/util/event.proto
-
-package org.tensorflow.proto.util;
-
-public final class EventProtos {
-  private EventProtos() {}
-  public static void registerAllExtensions(
-      com.google.protobuf.ExtensionRegistryLite registry) {
-  }
-
-  public static void registerAllExtensions(
-      com.google.protobuf.ExtensionRegistry registry) {
-    registerAllExtensions(
-        (com.google.protobuf.ExtensionRegistryLite) registry);
-  }
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_Event_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_Event_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_LogMessage_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_LogMessage_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_SessionLog_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_SessionLog_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_TaggedRunMetadata_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_TaggedRunMetadata_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_WatchdogConfig_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_WatchdogConfig_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_RequestedExitCode_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_RequestedExitCode_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_WorkerHeartbeatRequest_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_WorkerHeartbeatRequest_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_WorkerHeartbeatResponse_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_WorkerHeartbeatResponse_fieldAccessorTable;
-
-  public static com.google.protobuf.Descriptors.FileDescriptor
-      getDescriptor() {
-    return descriptor;
-  }
-  private static  com.google.protobuf.Descriptors.FileDescriptor
-      descriptor;
-  static {
-    java.lang.String[] descriptorData = {
-      "\n tensorflow/core/util/event.proto\022\ntens" +
-      "orflow\032\'tensorflow/core/framework/summar" +
-      "y.proto\"\273\002\n\005Event\022\021\n\twall_time\030\001 \001(\001\022\014\n\004" +
-      "step\030\002 \001(\003\022\026\n\014file_version\030\003 \001(\tH\000\022\023\n\tgr" +
-      "aph_def\030\004 \001(\014H\000\022&\n\007summary\030\005 \001(\0132\023.tenso" +
-      "rflow.SummaryH\000\022-\n\013log_message\030\006 \001(\0132\026.t" +
-      "ensorflow.LogMessageH\000\022-\n\013session_log\030\007 " +
-      "\001(\0132\026.tensorflow.SessionLogH\000\022<\n\023tagged_" +
-      "run_metadata\030\010 \001(\0132\035.tensorflow.TaggedRu" +
-      "nMetadataH\000\022\030\n\016meta_graph_def\030\t \001(\014H\000B\006\n" +
-      "\004what\"\231\001\n\nLogMessage\022+\n\005level\030\001 \001(\0162\034.te" +
-      "nsorflow.LogMessage.Level\022\017\n\007message\030\002 \001" +
-      "(\t\"M\n\005Level\022\013\n\007UNKNOWN\020\000\022\r\n\tDEBUGGING\020\n\022" +
-      "\010\n\004INFO\020\024\022\010\n\004WARN\020\036\022\t\n\005ERROR\020(\022\t\n\005FATAL\020" +
-      "2\"\266\001\n\nSessionLog\0224\n\006status\030\001 \001(\0162$.tenso" +
-      "rflow.SessionLog.SessionStatus\022\027\n\017checkp" +
-      "oint_path\030\002 \001(\t\022\013\n\003msg\030\003 \001(\t\"L\n\rSessionS" +
-      "tatus\022\026\n\022STATUS_UNSPECIFIED\020\000\022\t\n\005START\020\001" +
-      "\022\010\n\004STOP\020\002\022\016\n\nCHECKPOINT\020\003\"6\n\021TaggedRunM" +
-      "etadata\022\013\n\003tag\030\001 \001(\t\022\024\n\014run_metadata\030\002 \001" +
-      "(\014\"$\n\016WatchdogConfig\022\022\n\ntimeout_ms\030\001 \001(\003" +
-      "\"&\n\021RequestedExitCode\022\021\n\texit_code\030\001 \001(\005" +
-      "\"\266\001\n\026WorkerHeartbeatRequest\0225\n\rshutdown_" +
-      "mode\030\001 \001(\0162\036.tensorflow.WorkerShutdownMo" +
-      "de\0223\n\017watchdog_config\030\002 \001(\0132\032.tensorflow" +
-      ".WatchdogConfig\0220\n\texit_code\030\003 \001(\0132\035.ten" +
-      "sorflow.RequestedExitCode\"\203\001\n\027WorkerHear" +
-      "tbeatResponse\022/\n\rhealth_status\030\001 \001(\0162\030.t" +
-      "ensorflow.WorkerHealth\022%\n\nworker_log\030\002 \003" +
-      "(\0132\021.tensorflow.Event\022\020\n\010hostname\030\003 \001(\t*" +
-      "[\n\014WorkerHealth\022\006\n\002OK\020\000\022\034\n\030RECEIVED_SHUT" +
-      "DOWN_SIGNAL\020\001\022\022\n\016INTERNAL_ERROR\020\002\022\021\n\rSHU" +
-      "TTING_DOWN\020\003*k\n\022WorkerShutdownMode\022\013\n\007DE" +
-      "FAULT\020\000\022\022\n\016NOT_CONFIGURED\020\001\022\030\n\024WAIT_FOR_" +
-      "COORDINATOR\020\002\022\032\n\026SHUTDOWN_AFTER_TIMEOUT\020" +
-      "\003B-\n\031org.tensorflow.proto.utilB\013EventPro" +
-      "tosP\001\370\001\001b\006proto3"
-    };
-    descriptor = com.google.protobuf.Descriptors.FileDescriptor
-      .internalBuildGeneratedFileFrom(descriptorData,
-        new com.google.protobuf.Descriptors.FileDescriptor[] {
-          org.tensorflow.proto.framework.SummaryProtos.getDescriptor(),
-        });
-    internal_static_tensorflow_Event_descriptor =
-      getDescriptor().getMessageTypes().get(0);
-    internal_static_tensorflow_Event_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_Event_descriptor,
-        new java.lang.String[] { "WallTime", "Step", "FileVersion", "GraphDef", "Summary", "LogMessage", "SessionLog", "TaggedRunMetadata", "MetaGraphDef", "What", });
-    internal_static_tensorflow_LogMessage_descriptor =
-      getDescriptor().getMessageTypes().get(1);
-    internal_static_tensorflow_LogMessage_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_LogMessage_descriptor,
-        new java.lang.String[] { "Level", "Message", });
-    internal_static_tensorflow_SessionLog_descriptor =
-      getDescriptor().getMessageTypes().get(2);
-    internal_static_tensorflow_SessionLog_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_SessionLog_descriptor,
-        new java.lang.String[] { "Status", "CheckpointPath", "Msg", });
-    internal_static_tensorflow_TaggedRunMetadata_descriptor =
-      getDescriptor().getMessageTypes().get(3);
-    internal_static_tensorflow_TaggedRunMetadata_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_TaggedRunMetadata_descriptor,
-        new java.lang.String[] { "Tag", "RunMetadata", });
-    internal_static_tensorflow_WatchdogConfig_descriptor =
-      getDescriptor().getMessageTypes().get(4);
-    internal_static_tensorflow_WatchdogConfig_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_WatchdogConfig_descriptor,
-        new java.lang.String[] { "TimeoutMs", });
-    internal_static_tensorflow_RequestedExitCode_descriptor =
-      getDescriptor().getMessageTypes().get(5);
-    internal_static_tensorflow_RequestedExitCode_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_RequestedExitCode_descriptor,
-        new java.lang.String[] { "ExitCode", });
-    internal_static_tensorflow_WorkerHeartbeatRequest_descriptor =
-      getDescriptor().getMessageTypes().get(6);
-    internal_static_tensorflow_WorkerHeartbeatRequest_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_WorkerHeartbeatRequest_descriptor,
-        new java.lang.String[] { "ShutdownMode", "WatchdogConfig", "ExitCode", });
-    internal_static_tensorflow_WorkerHeartbeatResponse_descriptor =
-      getDescriptor().getMessageTypes().get(7);
-    internal_static_tensorflow_WorkerHeartbeatResponse_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_WorkerHeartbeatResponse_descriptor,
-        new java.lang.String[] { "HealthStatus", "WorkerLog", "Hostname", });
-    org.tensorflow.proto.framework.SummaryProtos.getDescriptor();
-  }
-
-  // @@protoc_insertion_point(outer_class_scope)
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/Execution.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/Execution.java
deleted file mode 100644
index 42d3a1d835a..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/Execution.java
+++ /dev/null
@@ -1,2259 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/debug_event.proto
-
-package org.tensorflow.proto.util;
-
-/**
- * <pre>
- * Data relating to the eager execution of an op or a Graph.
- * For a op that generates N output tensors (N &gt;= 0), only one
- * Execution proto will be used to describe the execution event.
- * </pre>
- *
- * Protobuf type {@code tensorflow.Execution}
- */
-public  final class Execution extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.Execution)
-    ExecutionOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use Execution.newBuilder() to construct.
-  private Execution(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private Execution() {
-    opType_ = "";
-    graphId_ = "";
-    inputTensorIds_ = emptyLongList();
-    outputTensorIds_ = emptyLongList();
-    tensorDebugMode_ = 0;
-    tensorProtos_ = java.util.Collections.emptyList();
-    outputTensorDeviceIds_ = emptyIntList();
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new Execution();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private Execution(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    int mutable_bitField0_ = 0;
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 10: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            opType_ = s;
-            break;
-          }
-          case 16: {
-
-            numOutputs_ = input.readInt32();
-            break;
-          }
-          case 26: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            graphId_ = s;
-            break;
-          }
-          case 32: {
-            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
-              inputTensorIds_ = newLongList();
-              mutable_bitField0_ |= 0x00000001;
-            }
-            inputTensorIds_.addLong(input.readInt64());
-            break;
-          }
-          case 34: {
-            int length = input.readRawVarint32();
-            int limit = input.pushLimit(length);
-            if (!((mutable_bitField0_ & 0x00000001) != 0) && input.getBytesUntilLimit() > 0) {
-              inputTensorIds_ = newLongList();
-              mutable_bitField0_ |= 0x00000001;
-            }
-            while (input.getBytesUntilLimit() > 0) {
-              inputTensorIds_.addLong(input.readInt64());
-            }
-            input.popLimit(limit);
-            break;
-          }
-          case 40: {
-            if (!((mutable_bitField0_ & 0x00000002) != 0)) {
-              outputTensorIds_ = newLongList();
-              mutable_bitField0_ |= 0x00000002;
-            }
-            outputTensorIds_.addLong(input.readInt64());
-            break;
-          }
-          case 42: {
-            int length = input.readRawVarint32();
-            int limit = input.pushLimit(length);
-            if (!((mutable_bitField0_ & 0x00000002) != 0) && input.getBytesUntilLimit() > 0) {
-              outputTensorIds_ = newLongList();
-              mutable_bitField0_ |= 0x00000002;
-            }
-            while (input.getBytesUntilLimit() > 0) {
-              outputTensorIds_.addLong(input.readInt64());
-            }
-            input.popLimit(limit);
-            break;
-          }
-          case 48: {
-            int rawValue = input.readEnum();
-
-            tensorDebugMode_ = rawValue;
-            break;
-          }
-          case 58: {
-            if (!((mutable_bitField0_ & 0x00000004) != 0)) {
-              tensorProtos_ = new java.util.ArrayList<org.tensorflow.proto.framework.TensorProto>();
-              mutable_bitField0_ |= 0x00000004;
-            }
-            tensorProtos_.add(
-                input.readMessage(org.tensorflow.proto.framework.TensorProto.parser(), extensionRegistry));
-            break;
-          }
-          case 66: {
-            org.tensorflow.proto.util.CodeLocation.Builder subBuilder = null;
-            if (codeLocation_ != null) {
-              subBuilder = codeLocation_.toBuilder();
-            }
-            codeLocation_ = input.readMessage(org.tensorflow.proto.util.CodeLocation.parser(), extensionRegistry);
-            if (subBuilder != null) {
-              subBuilder.mergeFrom(codeLocation_);
-              codeLocation_ = subBuilder.buildPartial();
-            }
-
-            break;
-          }
-          case 72: {
-            if (!((mutable_bitField0_ & 0x00000008) != 0)) {
-              outputTensorDeviceIds_ = newIntList();
-              mutable_bitField0_ |= 0x00000008;
-            }
-            outputTensorDeviceIds_.addInt(input.readInt32());
-            break;
-          }
-          case 74: {
-            int length = input.readRawVarint32();
-            int limit = input.pushLimit(length);
-            if (!((mutable_bitField0_ & 0x00000008) != 0) && input.getBytesUntilLimit() > 0) {
-              outputTensorDeviceIds_ = newIntList();
-              mutable_bitField0_ |= 0x00000008;
-            }
-            while (input.getBytesUntilLimit() > 0) {
-              outputTensorDeviceIds_.addInt(input.readInt32());
-            }
-            input.popLimit(limit);
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      if (((mutable_bitField0_ & 0x00000001) != 0)) {
-        inputTensorIds_.makeImmutable(); // C
-      }
-      if (((mutable_bitField0_ & 0x00000002) != 0)) {
-        outputTensorIds_.makeImmutable(); // C
-      }
-      if (((mutable_bitField0_ & 0x00000004) != 0)) {
-        tensorProtos_ = java.util.Collections.unmodifiableList(tensorProtos_);
-      }
-      if (((mutable_bitField0_ & 0x00000008) != 0)) {
-        outputTensorDeviceIds_.makeImmutable(); // C
-      }
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.util.DebugEventProtos.internal_static_tensorflow_Execution_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.util.DebugEventProtos.internal_static_tensorflow_Execution_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.util.Execution.class, org.tensorflow.proto.util.Execution.Builder.class);
-  }
-
-  public static final int OP_TYPE_FIELD_NUMBER = 1;
-  private volatile java.lang.Object opType_;
-  /**
-   * <pre>
-   * Op type (e.g., "MatMul").
-   * In the case of a Graph, this is the name of the Graph.
-   * </pre>
-   *
-   * <code>string op_type = 1;</code>
-   */
-  public java.lang.String getOpType() {
-    java.lang.Object ref = opType_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      opType_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * Op type (e.g., "MatMul").
-   * In the case of a Graph, this is the name of the Graph.
-   * </pre>
-   *
-   * <code>string op_type = 1;</code>
-   */
-  public com.google.protobuf.ByteString
-      getOpTypeBytes() {
-    java.lang.Object ref = opType_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      opType_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int NUM_OUTPUTS_FIELD_NUMBER = 2;
-  private int numOutputs_;
-  /**
-   * <pre>
-   * Number of output tensors.
-   * </pre>
-   *
-   * <code>int32 num_outputs = 2;</code>
-   */
-  public int getNumOutputs() {
-    return numOutputs_;
-  }
-
-  public static final int GRAPH_ID_FIELD_NUMBER = 3;
-  private volatile java.lang.Object graphId_;
-  /**
-   * <pre>
-   * The graph that's executed: applicable only to the eager
-   * execution of a FuncGraph.
-   * </pre>
-   *
-   * <code>string graph_id = 3;</code>
-   */
-  public java.lang.String getGraphId() {
-    java.lang.Object ref = graphId_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      graphId_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * The graph that's executed: applicable only to the eager
-   * execution of a FuncGraph.
-   * </pre>
-   *
-   * <code>string graph_id = 3;</code>
-   */
-  public com.google.protobuf.ByteString
-      getGraphIdBytes() {
-    java.lang.Object ref = graphId_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      graphId_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int INPUT_TENSOR_IDS_FIELD_NUMBER = 4;
-  private com.google.protobuf.Internal.LongList inputTensorIds_;
-  /**
-   * <pre>
-   * IDs of the input tensors (if available).
-   * </pre>
-   *
-   * <code>repeated int64 input_tensor_ids = 4;</code>
-   */
-  public java.util.List<java.lang.Long>
-      getInputTensorIdsList() {
-    return inputTensorIds_;
-  }
-  /**
-   * <pre>
-   * IDs of the input tensors (if available).
-   * </pre>
-   *
-   * <code>repeated int64 input_tensor_ids = 4;</code>
-   */
-  public int getInputTensorIdsCount() {
-    return inputTensorIds_.size();
-  }
-  /**
-   * <pre>
-   * IDs of the input tensors (if available).
-   * </pre>
-   *
-   * <code>repeated int64 input_tensor_ids = 4;</code>
-   */
-  public long getInputTensorIds(int index) {
-    return inputTensorIds_.getLong(index);
-  }
-  private int inputTensorIdsMemoizedSerializedSize = -1;
-
-  public static final int OUTPUT_TENSOR_IDS_FIELD_NUMBER = 5;
-  private com.google.protobuf.Internal.LongList outputTensorIds_;
-  /**
-   * <pre>
-   * IDs of the output tensors (if availbable).
-   * If specified, must have the same length as tensor_protos.
-   * </pre>
-   *
-   * <code>repeated int64 output_tensor_ids = 5;</code>
-   */
-  public java.util.List<java.lang.Long>
-      getOutputTensorIdsList() {
-    return outputTensorIds_;
-  }
-  /**
-   * <pre>
-   * IDs of the output tensors (if availbable).
-   * If specified, must have the same length as tensor_protos.
-   * </pre>
-   *
-   * <code>repeated int64 output_tensor_ids = 5;</code>
-   */
-  public int getOutputTensorIdsCount() {
-    return outputTensorIds_.size();
-  }
-  /**
-   * <pre>
-   * IDs of the output tensors (if availbable).
-   * If specified, must have the same length as tensor_protos.
-   * </pre>
-   *
-   * <code>repeated int64 output_tensor_ids = 5;</code>
-   */
-  public long getOutputTensorIds(int index) {
-    return outputTensorIds_.getLong(index);
-  }
-  private int outputTensorIdsMemoizedSerializedSize = -1;
-
-  public static final int TENSOR_DEBUG_MODE_FIELD_NUMBER = 6;
-  private int tensorDebugMode_;
-  /**
-   * <pre>
-   * Type of the tensor value encapsulated in this proto.
-   * </pre>
-   *
-   * <code>.tensorflow.TensorDebugMode tensor_debug_mode = 6;</code>
-   */
-  public int getTensorDebugModeValue() {
-    return tensorDebugMode_;
-  }
-  /**
-   * <pre>
-   * Type of the tensor value encapsulated in this proto.
-   * </pre>
-   *
-   * <code>.tensorflow.TensorDebugMode tensor_debug_mode = 6;</code>
-   */
-  public org.tensorflow.proto.util.TensorDebugMode getTensorDebugMode() {
-    @SuppressWarnings("deprecation")
-    org.tensorflow.proto.util.TensorDebugMode result = org.tensorflow.proto.util.TensorDebugMode.valueOf(tensorDebugMode_);
-    return result == null ? org.tensorflow.proto.util.TensorDebugMode.UNRECOGNIZED : result;
-  }
-
-  public static final int TENSOR_PROTOS_FIELD_NUMBER = 7;
-  private java.util.List<org.tensorflow.proto.framework.TensorProto> tensorProtos_;
-  /**
-   * <pre>
-   * Output Tensor values in the type described by `tensor_value_type`.
-   * The length of this should match `num_outputs`.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.TensorProto tensor_protos = 7;</code>
-   */
-  public java.util.List<org.tensorflow.proto.framework.TensorProto> getTensorProtosList() {
-    return tensorProtos_;
-  }
-  /**
-   * <pre>
-   * Output Tensor values in the type described by `tensor_value_type`.
-   * The length of this should match `num_outputs`.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.TensorProto tensor_protos = 7;</code>
-   */
-  public java.util.List<? extends org.tensorflow.proto.framework.TensorProtoOrBuilder> 
-      getTensorProtosOrBuilderList() {
-    return tensorProtos_;
-  }
-  /**
-   * <pre>
-   * Output Tensor values in the type described by `tensor_value_type`.
-   * The length of this should match `num_outputs`.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.TensorProto tensor_protos = 7;</code>
-   */
-  public int getTensorProtosCount() {
-    return tensorProtos_.size();
-  }
-  /**
-   * <pre>
-   * Output Tensor values in the type described by `tensor_value_type`.
-   * The length of this should match `num_outputs`.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.TensorProto tensor_protos = 7;</code>
-   */
-  public org.tensorflow.proto.framework.TensorProto getTensorProtos(int index) {
-    return tensorProtos_.get(index);
-  }
-  /**
-   * <pre>
-   * Output Tensor values in the type described by `tensor_value_type`.
-   * The length of this should match `num_outputs`.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.TensorProto tensor_protos = 7;</code>
-   */
-  public org.tensorflow.proto.framework.TensorProtoOrBuilder getTensorProtosOrBuilder(
-      int index) {
-    return tensorProtos_.get(index);
-  }
-
-  public static final int CODE_LOCATION_FIELD_NUMBER = 8;
-  private org.tensorflow.proto.util.CodeLocation codeLocation_;
-  /**
-   * <pre>
-   * Stack trace of the eager execution.
-   * </pre>
-   *
-   * <code>.tensorflow.CodeLocation code_location = 8;</code>
-   */
-  public boolean hasCodeLocation() {
-    return codeLocation_ != null;
-  }
-  /**
-   * <pre>
-   * Stack trace of the eager execution.
-   * </pre>
-   *
-   * <code>.tensorflow.CodeLocation code_location = 8;</code>
-   */
-  public org.tensorflow.proto.util.CodeLocation getCodeLocation() {
-    return codeLocation_ == null ? org.tensorflow.proto.util.CodeLocation.getDefaultInstance() : codeLocation_;
-  }
-  /**
-   * <pre>
-   * Stack trace of the eager execution.
-   * </pre>
-   *
-   * <code>.tensorflow.CodeLocation code_location = 8;</code>
-   */
-  public org.tensorflow.proto.util.CodeLocationOrBuilder getCodeLocationOrBuilder() {
-    return getCodeLocation();
-  }
-
-  public static final int OUTPUT_TENSOR_DEVICE_IDS_FIELD_NUMBER = 9;
-  private com.google.protobuf.Internal.IntList outputTensorDeviceIds_;
-  /**
-   * <pre>
-   * Debugged-generated IDs of the devices on which the output tensors reside.
-   * To look up details about the device (e.g., name), cross-reference this
-   * field with the DebuggedDevice messages.
-   * </pre>
-   *
-   * <code>repeated int32 output_tensor_device_ids = 9;</code>
-   */
-  public java.util.List<java.lang.Integer>
-      getOutputTensorDeviceIdsList() {
-    return outputTensorDeviceIds_;
-  }
-  /**
-   * <pre>
-   * Debugged-generated IDs of the devices on which the output tensors reside.
-   * To look up details about the device (e.g., name), cross-reference this
-   * field with the DebuggedDevice messages.
-   * </pre>
-   *
-   * <code>repeated int32 output_tensor_device_ids = 9;</code>
-   */
-  public int getOutputTensorDeviceIdsCount() {
-    return outputTensorDeviceIds_.size();
-  }
-  /**
-   * <pre>
-   * Debugged-generated IDs of the devices on which the output tensors reside.
-   * To look up details about the device (e.g., name), cross-reference this
-   * field with the DebuggedDevice messages.
-   * </pre>
-   *
-   * <code>repeated int32 output_tensor_device_ids = 9;</code>
-   */
-  public int getOutputTensorDeviceIds(int index) {
-    return outputTensorDeviceIds_.getInt(index);
-  }
-  private int outputTensorDeviceIdsMemoizedSerializedSize = -1;
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    getSerializedSize();
-    if (!getOpTypeBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 1, opType_);
-    }
-    if (numOutputs_ != 0) {
-      output.writeInt32(2, numOutputs_);
-    }
-    if (!getGraphIdBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 3, graphId_);
-    }
-    if (getInputTensorIdsList().size() > 0) {
-      output.writeUInt32NoTag(34);
-      output.writeUInt32NoTag(inputTensorIdsMemoizedSerializedSize);
-    }
-    for (int i = 0; i < inputTensorIds_.size(); i++) {
-      output.writeInt64NoTag(inputTensorIds_.getLong(i));
-    }
-    if (getOutputTensorIdsList().size() > 0) {
-      output.writeUInt32NoTag(42);
-      output.writeUInt32NoTag(outputTensorIdsMemoizedSerializedSize);
-    }
-    for (int i = 0; i < outputTensorIds_.size(); i++) {
-      output.writeInt64NoTag(outputTensorIds_.getLong(i));
-    }
-    if (tensorDebugMode_ != org.tensorflow.proto.util.TensorDebugMode.UNSPECIFIED.getNumber()) {
-      output.writeEnum(6, tensorDebugMode_);
-    }
-    for (int i = 0; i < tensorProtos_.size(); i++) {
-      output.writeMessage(7, tensorProtos_.get(i));
-    }
-    if (codeLocation_ != null) {
-      output.writeMessage(8, getCodeLocation());
-    }
-    if (getOutputTensorDeviceIdsList().size() > 0) {
-      output.writeUInt32NoTag(74);
-      output.writeUInt32NoTag(outputTensorDeviceIdsMemoizedSerializedSize);
-    }
-    for (int i = 0; i < outputTensorDeviceIds_.size(); i++) {
-      output.writeInt32NoTag(outputTensorDeviceIds_.getInt(i));
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (!getOpTypeBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, opType_);
-    }
-    if (numOutputs_ != 0) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt32Size(2, numOutputs_);
-    }
-    if (!getGraphIdBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(3, graphId_);
-    }
-    {
-      int dataSize = 0;
-      for (int i = 0; i < inputTensorIds_.size(); i++) {
-        dataSize += com.google.protobuf.CodedOutputStream
-          .computeInt64SizeNoTag(inputTensorIds_.getLong(i));
-      }
-      size += dataSize;
-      if (!getInputTensorIdsList().isEmpty()) {
-        size += 1;
-        size += com.google.protobuf.CodedOutputStream
-            .computeInt32SizeNoTag(dataSize);
-      }
-      inputTensorIdsMemoizedSerializedSize = dataSize;
-    }
-    {
-      int dataSize = 0;
-      for (int i = 0; i < outputTensorIds_.size(); i++) {
-        dataSize += com.google.protobuf.CodedOutputStream
-          .computeInt64SizeNoTag(outputTensorIds_.getLong(i));
-      }
-      size += dataSize;
-      if (!getOutputTensorIdsList().isEmpty()) {
-        size += 1;
-        size += com.google.protobuf.CodedOutputStream
-            .computeInt32SizeNoTag(dataSize);
-      }
-      outputTensorIdsMemoizedSerializedSize = dataSize;
-    }
-    if (tensorDebugMode_ != org.tensorflow.proto.util.TensorDebugMode.UNSPECIFIED.getNumber()) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeEnumSize(6, tensorDebugMode_);
-    }
-    for (int i = 0; i < tensorProtos_.size(); i++) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(7, tensorProtos_.get(i));
-    }
-    if (codeLocation_ != null) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(8, getCodeLocation());
-    }
-    {
-      int dataSize = 0;
-      for (int i = 0; i < outputTensorDeviceIds_.size(); i++) {
-        dataSize += com.google.protobuf.CodedOutputStream
-          .computeInt32SizeNoTag(outputTensorDeviceIds_.getInt(i));
-      }
-      size += dataSize;
-      if (!getOutputTensorDeviceIdsList().isEmpty()) {
-        size += 1;
-        size += com.google.protobuf.CodedOutputStream
-            .computeInt32SizeNoTag(dataSize);
-      }
-      outputTensorDeviceIdsMemoizedSerializedSize = dataSize;
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.util.Execution)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.util.Execution other = (org.tensorflow.proto.util.Execution) obj;
-
-    if (!getOpType()
-        .equals(other.getOpType())) return false;
-    if (getNumOutputs()
-        != other.getNumOutputs()) return false;
-    if (!getGraphId()
-        .equals(other.getGraphId())) return false;
-    if (!getInputTensorIdsList()
-        .equals(other.getInputTensorIdsList())) return false;
-    if (!getOutputTensorIdsList()
-        .equals(other.getOutputTensorIdsList())) return false;
-    if (tensorDebugMode_ != other.tensorDebugMode_) return false;
-    if (!getTensorProtosList()
-        .equals(other.getTensorProtosList())) return false;
-    if (hasCodeLocation() != other.hasCodeLocation()) return false;
-    if (hasCodeLocation()) {
-      if (!getCodeLocation()
-          .equals(other.getCodeLocation())) return false;
-    }
-    if (!getOutputTensorDeviceIdsList()
-        .equals(other.getOutputTensorDeviceIdsList())) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    hash = (37 * hash) + OP_TYPE_FIELD_NUMBER;
-    hash = (53 * hash) + getOpType().hashCode();
-    hash = (37 * hash) + NUM_OUTPUTS_FIELD_NUMBER;
-    hash = (53 * hash) + getNumOutputs();
-    hash = (37 * hash) + GRAPH_ID_FIELD_NUMBER;
-    hash = (53 * hash) + getGraphId().hashCode();
-    if (getInputTensorIdsCount() > 0) {
-      hash = (37 * hash) + INPUT_TENSOR_IDS_FIELD_NUMBER;
-      hash = (53 * hash) + getInputTensorIdsList().hashCode();
-    }
-    if (getOutputTensorIdsCount() > 0) {
-      hash = (37 * hash) + OUTPUT_TENSOR_IDS_FIELD_NUMBER;
-      hash = (53 * hash) + getOutputTensorIdsList().hashCode();
-    }
-    hash = (37 * hash) + TENSOR_DEBUG_MODE_FIELD_NUMBER;
-    hash = (53 * hash) + tensorDebugMode_;
-    if (getTensorProtosCount() > 0) {
-      hash = (37 * hash) + TENSOR_PROTOS_FIELD_NUMBER;
-      hash = (53 * hash) + getTensorProtosList().hashCode();
-    }
-    if (hasCodeLocation()) {
-      hash = (37 * hash) + CODE_LOCATION_FIELD_NUMBER;
-      hash = (53 * hash) + getCodeLocation().hashCode();
-    }
-    if (getOutputTensorDeviceIdsCount() > 0) {
-      hash = (37 * hash) + OUTPUT_TENSOR_DEVICE_IDS_FIELD_NUMBER;
-      hash = (53 * hash) + getOutputTensorDeviceIdsList().hashCode();
-    }
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.util.Execution parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.Execution parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.Execution parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.Execution parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.Execution parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.Execution parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.Execution parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.Execution parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.Execution parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.Execution parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.Execution parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.Execution parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.util.Execution prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * <pre>
-   * Data relating to the eager execution of an op or a Graph.
-   * For a op that generates N output tensors (N &gt;= 0), only one
-   * Execution proto will be used to describe the execution event.
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.Execution}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.Execution)
-      org.tensorflow.proto.util.ExecutionOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.util.DebugEventProtos.internal_static_tensorflow_Execution_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.util.DebugEventProtos.internal_static_tensorflow_Execution_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.util.Execution.class, org.tensorflow.proto.util.Execution.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.util.Execution.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-        getTensorProtosFieldBuilder();
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      opType_ = "";
-
-      numOutputs_ = 0;
-
-      graphId_ = "";
-
-      inputTensorIds_ = emptyLongList();
-      bitField0_ = (bitField0_ & ~0x00000001);
-      outputTensorIds_ = emptyLongList();
-      bitField0_ = (bitField0_ & ~0x00000002);
-      tensorDebugMode_ = 0;
-
-      if (tensorProtosBuilder_ == null) {
-        tensorProtos_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000004);
-      } else {
-        tensorProtosBuilder_.clear();
-      }
-      if (codeLocationBuilder_ == null) {
-        codeLocation_ = null;
-      } else {
-        codeLocation_ = null;
-        codeLocationBuilder_ = null;
-      }
-      outputTensorDeviceIds_ = emptyIntList();
-      bitField0_ = (bitField0_ & ~0x00000008);
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.util.DebugEventProtos.internal_static_tensorflow_Execution_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.Execution getDefaultInstanceForType() {
-      return org.tensorflow.proto.util.Execution.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.Execution build() {
-      org.tensorflow.proto.util.Execution result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.Execution buildPartial() {
-      org.tensorflow.proto.util.Execution result = new org.tensorflow.proto.util.Execution(this);
-      int from_bitField0_ = bitField0_;
-      result.opType_ = opType_;
-      result.numOutputs_ = numOutputs_;
-      result.graphId_ = graphId_;
-      if (((bitField0_ & 0x00000001) != 0)) {
-        inputTensorIds_.makeImmutable();
-        bitField0_ = (bitField0_ & ~0x00000001);
-      }
-      result.inputTensorIds_ = inputTensorIds_;
-      if (((bitField0_ & 0x00000002) != 0)) {
-        outputTensorIds_.makeImmutable();
-        bitField0_ = (bitField0_ & ~0x00000002);
-      }
-      result.outputTensorIds_ = outputTensorIds_;
-      result.tensorDebugMode_ = tensorDebugMode_;
-      if (tensorProtosBuilder_ == null) {
-        if (((bitField0_ & 0x00000004) != 0)) {
-          tensorProtos_ = java.util.Collections.unmodifiableList(tensorProtos_);
-          bitField0_ = (bitField0_ & ~0x00000004);
-        }
-        result.tensorProtos_ = tensorProtos_;
-      } else {
-        result.tensorProtos_ = tensorProtosBuilder_.build();
-      }
-      if (codeLocationBuilder_ == null) {
-        result.codeLocation_ = codeLocation_;
-      } else {
-        result.codeLocation_ = codeLocationBuilder_.build();
-      }
-      if (((bitField0_ & 0x00000008) != 0)) {
-        outputTensorDeviceIds_.makeImmutable();
-        bitField0_ = (bitField0_ & ~0x00000008);
-      }
-      result.outputTensorDeviceIds_ = outputTensorDeviceIds_;
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.util.Execution) {
-        return mergeFrom((org.tensorflow.proto.util.Execution)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.util.Execution other) {
-      if (other == org.tensorflow.proto.util.Execution.getDefaultInstance()) return this;
-      if (!other.getOpType().isEmpty()) {
-        opType_ = other.opType_;
-        onChanged();
-      }
-      if (other.getNumOutputs() != 0) {
-        setNumOutputs(other.getNumOutputs());
-      }
-      if (!other.getGraphId().isEmpty()) {
-        graphId_ = other.graphId_;
-        onChanged();
-      }
-      if (!other.inputTensorIds_.isEmpty()) {
-        if (inputTensorIds_.isEmpty()) {
-          inputTensorIds_ = other.inputTensorIds_;
-          bitField0_ = (bitField0_ & ~0x00000001);
-        } else {
-          ensureInputTensorIdsIsMutable();
-          inputTensorIds_.addAll(other.inputTensorIds_);
-        }
-        onChanged();
-      }
-      if (!other.outputTensorIds_.isEmpty()) {
-        if (outputTensorIds_.isEmpty()) {
-          outputTensorIds_ = other.outputTensorIds_;
-          bitField0_ = (bitField0_ & ~0x00000002);
-        } else {
-          ensureOutputTensorIdsIsMutable();
-          outputTensorIds_.addAll(other.outputTensorIds_);
-        }
-        onChanged();
-      }
-      if (other.tensorDebugMode_ != 0) {
-        setTensorDebugModeValue(other.getTensorDebugModeValue());
-      }
-      if (tensorProtosBuilder_ == null) {
-        if (!other.tensorProtos_.isEmpty()) {
-          if (tensorProtos_.isEmpty()) {
-            tensorProtos_ = other.tensorProtos_;
-            bitField0_ = (bitField0_ & ~0x00000004);
-          } else {
-            ensureTensorProtosIsMutable();
-            tensorProtos_.addAll(other.tensorProtos_);
-          }
-          onChanged();
-        }
-      } else {
-        if (!other.tensorProtos_.isEmpty()) {
-          if (tensorProtosBuilder_.isEmpty()) {
-            tensorProtosBuilder_.dispose();
-            tensorProtosBuilder_ = null;
-            tensorProtos_ = other.tensorProtos_;
-            bitField0_ = (bitField0_ & ~0x00000004);
-            tensorProtosBuilder_ = 
-              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
-                 getTensorProtosFieldBuilder() : null;
-          } else {
-            tensorProtosBuilder_.addAllMessages(other.tensorProtos_);
-          }
-        }
-      }
-      if (other.hasCodeLocation()) {
-        mergeCodeLocation(other.getCodeLocation());
-      }
-      if (!other.outputTensorDeviceIds_.isEmpty()) {
-        if (outputTensorDeviceIds_.isEmpty()) {
-          outputTensorDeviceIds_ = other.outputTensorDeviceIds_;
-          bitField0_ = (bitField0_ & ~0x00000008);
-        } else {
-          ensureOutputTensorDeviceIdsIsMutable();
-          outputTensorDeviceIds_.addAll(other.outputTensorDeviceIds_);
-        }
-        onChanged();
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.util.Execution parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.util.Execution) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-    private int bitField0_;
-
-    private java.lang.Object opType_ = "";
-    /**
-     * <pre>
-     * Op type (e.g., "MatMul").
-     * In the case of a Graph, this is the name of the Graph.
-     * </pre>
-     *
-     * <code>string op_type = 1;</code>
-     */
-    public java.lang.String getOpType() {
-      java.lang.Object ref = opType_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        opType_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Op type (e.g., "MatMul").
-     * In the case of a Graph, this is the name of the Graph.
-     * </pre>
-     *
-     * <code>string op_type = 1;</code>
-     */
-    public com.google.protobuf.ByteString
-        getOpTypeBytes() {
-      java.lang.Object ref = opType_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        opType_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Op type (e.g., "MatMul").
-     * In the case of a Graph, this is the name of the Graph.
-     * </pre>
-     *
-     * <code>string op_type = 1;</code>
-     */
-    public Builder setOpType(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      opType_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Op type (e.g., "MatMul").
-     * In the case of a Graph, this is the name of the Graph.
-     * </pre>
-     *
-     * <code>string op_type = 1;</code>
-     */
-    public Builder clearOpType() {
-      
-      opType_ = getDefaultInstance().getOpType();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Op type (e.g., "MatMul").
-     * In the case of a Graph, this is the name of the Graph.
-     * </pre>
-     *
-     * <code>string op_type = 1;</code>
-     */
-    public Builder setOpTypeBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      opType_ = value;
-      onChanged();
-      return this;
-    }
-
-    private int numOutputs_ ;
-    /**
-     * <pre>
-     * Number of output tensors.
-     * </pre>
-     *
-     * <code>int32 num_outputs = 2;</code>
-     */
-    public int getNumOutputs() {
-      return numOutputs_;
-    }
-    /**
-     * <pre>
-     * Number of output tensors.
-     * </pre>
-     *
-     * <code>int32 num_outputs = 2;</code>
-     */
-    public Builder setNumOutputs(int value) {
-      
-      numOutputs_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Number of output tensors.
-     * </pre>
-     *
-     * <code>int32 num_outputs = 2;</code>
-     */
-    public Builder clearNumOutputs() {
-      
-      numOutputs_ = 0;
-      onChanged();
-      return this;
-    }
-
-    private java.lang.Object graphId_ = "";
-    /**
-     * <pre>
-     * The graph that's executed: applicable only to the eager
-     * execution of a FuncGraph.
-     * </pre>
-     *
-     * <code>string graph_id = 3;</code>
-     */
-    public java.lang.String getGraphId() {
-      java.lang.Object ref = graphId_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        graphId_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * The graph that's executed: applicable only to the eager
-     * execution of a FuncGraph.
-     * </pre>
-     *
-     * <code>string graph_id = 3;</code>
-     */
-    public com.google.protobuf.ByteString
-        getGraphIdBytes() {
-      java.lang.Object ref = graphId_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        graphId_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * The graph that's executed: applicable only to the eager
-     * execution of a FuncGraph.
-     * </pre>
-     *
-     * <code>string graph_id = 3;</code>
-     */
-    public Builder setGraphId(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      graphId_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * The graph that's executed: applicable only to the eager
-     * execution of a FuncGraph.
-     * </pre>
-     *
-     * <code>string graph_id = 3;</code>
-     */
-    public Builder clearGraphId() {
-      
-      graphId_ = getDefaultInstance().getGraphId();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * The graph that's executed: applicable only to the eager
-     * execution of a FuncGraph.
-     * </pre>
-     *
-     * <code>string graph_id = 3;</code>
-     */
-    public Builder setGraphIdBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      graphId_ = value;
-      onChanged();
-      return this;
-    }
-
-    private com.google.protobuf.Internal.LongList inputTensorIds_ = emptyLongList();
-    private void ensureInputTensorIdsIsMutable() {
-      if (!((bitField0_ & 0x00000001) != 0)) {
-        inputTensorIds_ = mutableCopy(inputTensorIds_);
-        bitField0_ |= 0x00000001;
-       }
-    }
-    /**
-     * <pre>
-     * IDs of the input tensors (if available).
-     * </pre>
-     *
-     * <code>repeated int64 input_tensor_ids = 4;</code>
-     */
-    public java.util.List<java.lang.Long>
-        getInputTensorIdsList() {
-      return ((bitField0_ & 0x00000001) != 0) ?
-               java.util.Collections.unmodifiableList(inputTensorIds_) : inputTensorIds_;
-    }
-    /**
-     * <pre>
-     * IDs of the input tensors (if available).
-     * </pre>
-     *
-     * <code>repeated int64 input_tensor_ids = 4;</code>
-     */
-    public int getInputTensorIdsCount() {
-      return inputTensorIds_.size();
-    }
-    /**
-     * <pre>
-     * IDs of the input tensors (if available).
-     * </pre>
-     *
-     * <code>repeated int64 input_tensor_ids = 4;</code>
-     */
-    public long getInputTensorIds(int index) {
-      return inputTensorIds_.getLong(index);
-    }
-    /**
-     * <pre>
-     * IDs of the input tensors (if available).
-     * </pre>
-     *
-     * <code>repeated int64 input_tensor_ids = 4;</code>
-     */
-    public Builder setInputTensorIds(
-        int index, long value) {
-      ensureInputTensorIdsIsMutable();
-      inputTensorIds_.setLong(index, value);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * IDs of the input tensors (if available).
-     * </pre>
-     *
-     * <code>repeated int64 input_tensor_ids = 4;</code>
-     */
-    public Builder addInputTensorIds(long value) {
-      ensureInputTensorIdsIsMutable();
-      inputTensorIds_.addLong(value);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * IDs of the input tensors (if available).
-     * </pre>
-     *
-     * <code>repeated int64 input_tensor_ids = 4;</code>
-     */
-    public Builder addAllInputTensorIds(
-        java.lang.Iterable<? extends java.lang.Long> values) {
-      ensureInputTensorIdsIsMutable();
-      com.google.protobuf.AbstractMessageLite.Builder.addAll(
-          values, inputTensorIds_);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * IDs of the input tensors (if available).
-     * </pre>
-     *
-     * <code>repeated int64 input_tensor_ids = 4;</code>
-     */
-    public Builder clearInputTensorIds() {
-      inputTensorIds_ = emptyLongList();
-      bitField0_ = (bitField0_ & ~0x00000001);
-      onChanged();
-      return this;
-    }
-
-    private com.google.protobuf.Internal.LongList outputTensorIds_ = emptyLongList();
-    private void ensureOutputTensorIdsIsMutable() {
-      if (!((bitField0_ & 0x00000002) != 0)) {
-        outputTensorIds_ = mutableCopy(outputTensorIds_);
-        bitField0_ |= 0x00000002;
-       }
-    }
-    /**
-     * <pre>
-     * IDs of the output tensors (if availbable).
-     * If specified, must have the same length as tensor_protos.
-     * </pre>
-     *
-     * <code>repeated int64 output_tensor_ids = 5;</code>
-     */
-    public java.util.List<java.lang.Long>
-        getOutputTensorIdsList() {
-      return ((bitField0_ & 0x00000002) != 0) ?
-               java.util.Collections.unmodifiableList(outputTensorIds_) : outputTensorIds_;
-    }
-    /**
-     * <pre>
-     * IDs of the output tensors (if availbable).
-     * If specified, must have the same length as tensor_protos.
-     * </pre>
-     *
-     * <code>repeated int64 output_tensor_ids = 5;</code>
-     */
-    public int getOutputTensorIdsCount() {
-      return outputTensorIds_.size();
-    }
-    /**
-     * <pre>
-     * IDs of the output tensors (if availbable).
-     * If specified, must have the same length as tensor_protos.
-     * </pre>
-     *
-     * <code>repeated int64 output_tensor_ids = 5;</code>
-     */
-    public long getOutputTensorIds(int index) {
-      return outputTensorIds_.getLong(index);
-    }
-    /**
-     * <pre>
-     * IDs of the output tensors (if availbable).
-     * If specified, must have the same length as tensor_protos.
-     * </pre>
-     *
-     * <code>repeated int64 output_tensor_ids = 5;</code>
-     */
-    public Builder setOutputTensorIds(
-        int index, long value) {
-      ensureOutputTensorIdsIsMutable();
-      outputTensorIds_.setLong(index, value);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * IDs of the output tensors (if availbable).
-     * If specified, must have the same length as tensor_protos.
-     * </pre>
-     *
-     * <code>repeated int64 output_tensor_ids = 5;</code>
-     */
-    public Builder addOutputTensorIds(long value) {
-      ensureOutputTensorIdsIsMutable();
-      outputTensorIds_.addLong(value);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * IDs of the output tensors (if availbable).
-     * If specified, must have the same length as tensor_protos.
-     * </pre>
-     *
-     * <code>repeated int64 output_tensor_ids = 5;</code>
-     */
-    public Builder addAllOutputTensorIds(
-        java.lang.Iterable<? extends java.lang.Long> values) {
-      ensureOutputTensorIdsIsMutable();
-      com.google.protobuf.AbstractMessageLite.Builder.addAll(
-          values, outputTensorIds_);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * IDs of the output tensors (if availbable).
-     * If specified, must have the same length as tensor_protos.
-     * </pre>
-     *
-     * <code>repeated int64 output_tensor_ids = 5;</code>
-     */
-    public Builder clearOutputTensorIds() {
-      outputTensorIds_ = emptyLongList();
-      bitField0_ = (bitField0_ & ~0x00000002);
-      onChanged();
-      return this;
-    }
-
-    private int tensorDebugMode_ = 0;
-    /**
-     * <pre>
-     * Type of the tensor value encapsulated in this proto.
-     * </pre>
-     *
-     * <code>.tensorflow.TensorDebugMode tensor_debug_mode = 6;</code>
-     */
-    public int getTensorDebugModeValue() {
-      return tensorDebugMode_;
-    }
-    /**
-     * <pre>
-     * Type of the tensor value encapsulated in this proto.
-     * </pre>
-     *
-     * <code>.tensorflow.TensorDebugMode tensor_debug_mode = 6;</code>
-     */
-    public Builder setTensorDebugModeValue(int value) {
-      tensorDebugMode_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Type of the tensor value encapsulated in this proto.
-     * </pre>
-     *
-     * <code>.tensorflow.TensorDebugMode tensor_debug_mode = 6;</code>
-     */
-    public org.tensorflow.proto.util.TensorDebugMode getTensorDebugMode() {
-      @SuppressWarnings("deprecation")
-      org.tensorflow.proto.util.TensorDebugMode result = org.tensorflow.proto.util.TensorDebugMode.valueOf(tensorDebugMode_);
-      return result == null ? org.tensorflow.proto.util.TensorDebugMode.UNRECOGNIZED : result;
-    }
-    /**
-     * <pre>
-     * Type of the tensor value encapsulated in this proto.
-     * </pre>
-     *
-     * <code>.tensorflow.TensorDebugMode tensor_debug_mode = 6;</code>
-     */
-    public Builder setTensorDebugMode(org.tensorflow.proto.util.TensorDebugMode value) {
-      if (value == null) {
-        throw new NullPointerException();
-      }
-      
-      tensorDebugMode_ = value.getNumber();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Type of the tensor value encapsulated in this proto.
-     * </pre>
-     *
-     * <code>.tensorflow.TensorDebugMode tensor_debug_mode = 6;</code>
-     */
-    public Builder clearTensorDebugMode() {
-      
-      tensorDebugMode_ = 0;
-      onChanged();
-      return this;
-    }
-
-    private java.util.List<org.tensorflow.proto.framework.TensorProto> tensorProtos_ =
-      java.util.Collections.emptyList();
-    private void ensureTensorProtosIsMutable() {
-      if (!((bitField0_ & 0x00000004) != 0)) {
-        tensorProtos_ = new java.util.ArrayList<org.tensorflow.proto.framework.TensorProto>(tensorProtos_);
-        bitField0_ |= 0x00000004;
-       }
-    }
-
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.framework.TensorProto, org.tensorflow.proto.framework.TensorProto.Builder, org.tensorflow.proto.framework.TensorProtoOrBuilder> tensorProtosBuilder_;
-
-    /**
-     * <pre>
-     * Output Tensor values in the type described by `tensor_value_type`.
-     * The length of this should match `num_outputs`.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorProto tensor_protos = 7;</code>
-     */
-    public java.util.List<org.tensorflow.proto.framework.TensorProto> getTensorProtosList() {
-      if (tensorProtosBuilder_ == null) {
-        return java.util.Collections.unmodifiableList(tensorProtos_);
-      } else {
-        return tensorProtosBuilder_.getMessageList();
-      }
-    }
-    /**
-     * <pre>
-     * Output Tensor values in the type described by `tensor_value_type`.
-     * The length of this should match `num_outputs`.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorProto tensor_protos = 7;</code>
-     */
-    public int getTensorProtosCount() {
-      if (tensorProtosBuilder_ == null) {
-        return tensorProtos_.size();
-      } else {
-        return tensorProtosBuilder_.getCount();
-      }
-    }
-    /**
-     * <pre>
-     * Output Tensor values in the type described by `tensor_value_type`.
-     * The length of this should match `num_outputs`.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorProto tensor_protos = 7;</code>
-     */
-    public org.tensorflow.proto.framework.TensorProto getTensorProtos(int index) {
-      if (tensorProtosBuilder_ == null) {
-        return tensorProtos_.get(index);
-      } else {
-        return tensorProtosBuilder_.getMessage(index);
-      }
-    }
-    /**
-     * <pre>
-     * Output Tensor values in the type described by `tensor_value_type`.
-     * The length of this should match `num_outputs`.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorProto tensor_protos = 7;</code>
-     */
-    public Builder setTensorProtos(
-        int index, org.tensorflow.proto.framework.TensorProto value) {
-      if (tensorProtosBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureTensorProtosIsMutable();
-        tensorProtos_.set(index, value);
-        onChanged();
-      } else {
-        tensorProtosBuilder_.setMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Output Tensor values in the type described by `tensor_value_type`.
-     * The length of this should match `num_outputs`.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorProto tensor_protos = 7;</code>
-     */
-    public Builder setTensorProtos(
-        int index, org.tensorflow.proto.framework.TensorProto.Builder builderForValue) {
-      if (tensorProtosBuilder_ == null) {
-        ensureTensorProtosIsMutable();
-        tensorProtos_.set(index, builderForValue.build());
-        onChanged();
-      } else {
-        tensorProtosBuilder_.setMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Output Tensor values in the type described by `tensor_value_type`.
-     * The length of this should match `num_outputs`.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorProto tensor_protos = 7;</code>
-     */
-    public Builder addTensorProtos(org.tensorflow.proto.framework.TensorProto value) {
-      if (tensorProtosBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureTensorProtosIsMutable();
-        tensorProtos_.add(value);
-        onChanged();
-      } else {
-        tensorProtosBuilder_.addMessage(value);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Output Tensor values in the type described by `tensor_value_type`.
-     * The length of this should match `num_outputs`.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorProto tensor_protos = 7;</code>
-     */
-    public Builder addTensorProtos(
-        int index, org.tensorflow.proto.framework.TensorProto value) {
-      if (tensorProtosBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureTensorProtosIsMutable();
-        tensorProtos_.add(index, value);
-        onChanged();
-      } else {
-        tensorProtosBuilder_.addMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Output Tensor values in the type described by `tensor_value_type`.
-     * The length of this should match `num_outputs`.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorProto tensor_protos = 7;</code>
-     */
-    public Builder addTensorProtos(
-        org.tensorflow.proto.framework.TensorProto.Builder builderForValue) {
-      if (tensorProtosBuilder_ == null) {
-        ensureTensorProtosIsMutable();
-        tensorProtos_.add(builderForValue.build());
-        onChanged();
-      } else {
-        tensorProtosBuilder_.addMessage(builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Output Tensor values in the type described by `tensor_value_type`.
-     * The length of this should match `num_outputs`.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorProto tensor_protos = 7;</code>
-     */
-    public Builder addTensorProtos(
-        int index, org.tensorflow.proto.framework.TensorProto.Builder builderForValue) {
-      if (tensorProtosBuilder_ == null) {
-        ensureTensorProtosIsMutable();
-        tensorProtos_.add(index, builderForValue.build());
-        onChanged();
-      } else {
-        tensorProtosBuilder_.addMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Output Tensor values in the type described by `tensor_value_type`.
-     * The length of this should match `num_outputs`.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorProto tensor_protos = 7;</code>
-     */
-    public Builder addAllTensorProtos(
-        java.lang.Iterable<? extends org.tensorflow.proto.framework.TensorProto> values) {
-      if (tensorProtosBuilder_ == null) {
-        ensureTensorProtosIsMutable();
-        com.google.protobuf.AbstractMessageLite.Builder.addAll(
-            values, tensorProtos_);
-        onChanged();
-      } else {
-        tensorProtosBuilder_.addAllMessages(values);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Output Tensor values in the type described by `tensor_value_type`.
-     * The length of this should match `num_outputs`.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorProto tensor_protos = 7;</code>
-     */
-    public Builder clearTensorProtos() {
-      if (tensorProtosBuilder_ == null) {
-        tensorProtos_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000004);
-        onChanged();
-      } else {
-        tensorProtosBuilder_.clear();
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Output Tensor values in the type described by `tensor_value_type`.
-     * The length of this should match `num_outputs`.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorProto tensor_protos = 7;</code>
-     */
-    public Builder removeTensorProtos(int index) {
-      if (tensorProtosBuilder_ == null) {
-        ensureTensorProtosIsMutable();
-        tensorProtos_.remove(index);
-        onChanged();
-      } else {
-        tensorProtosBuilder_.remove(index);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Output Tensor values in the type described by `tensor_value_type`.
-     * The length of this should match `num_outputs`.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorProto tensor_protos = 7;</code>
-     */
-    public org.tensorflow.proto.framework.TensorProto.Builder getTensorProtosBuilder(
-        int index) {
-      return getTensorProtosFieldBuilder().getBuilder(index);
-    }
-    /**
-     * <pre>
-     * Output Tensor values in the type described by `tensor_value_type`.
-     * The length of this should match `num_outputs`.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorProto tensor_protos = 7;</code>
-     */
-    public org.tensorflow.proto.framework.TensorProtoOrBuilder getTensorProtosOrBuilder(
-        int index) {
-      if (tensorProtosBuilder_ == null) {
-        return tensorProtos_.get(index);  } else {
-        return tensorProtosBuilder_.getMessageOrBuilder(index);
-      }
-    }
-    /**
-     * <pre>
-     * Output Tensor values in the type described by `tensor_value_type`.
-     * The length of this should match `num_outputs`.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorProto tensor_protos = 7;</code>
-     */
-    public java.util.List<? extends org.tensorflow.proto.framework.TensorProtoOrBuilder> 
-         getTensorProtosOrBuilderList() {
-      if (tensorProtosBuilder_ != null) {
-        return tensorProtosBuilder_.getMessageOrBuilderList();
-      } else {
-        return java.util.Collections.unmodifiableList(tensorProtos_);
-      }
-    }
-    /**
-     * <pre>
-     * Output Tensor values in the type described by `tensor_value_type`.
-     * The length of this should match `num_outputs`.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorProto tensor_protos = 7;</code>
-     */
-    public org.tensorflow.proto.framework.TensorProto.Builder addTensorProtosBuilder() {
-      return getTensorProtosFieldBuilder().addBuilder(
-          org.tensorflow.proto.framework.TensorProto.getDefaultInstance());
-    }
-    /**
-     * <pre>
-     * Output Tensor values in the type described by `tensor_value_type`.
-     * The length of this should match `num_outputs`.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorProto tensor_protos = 7;</code>
-     */
-    public org.tensorflow.proto.framework.TensorProto.Builder addTensorProtosBuilder(
-        int index) {
-      return getTensorProtosFieldBuilder().addBuilder(
-          index, org.tensorflow.proto.framework.TensorProto.getDefaultInstance());
-    }
-    /**
-     * <pre>
-     * Output Tensor values in the type described by `tensor_value_type`.
-     * The length of this should match `num_outputs`.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorProto tensor_protos = 7;</code>
-     */
-    public java.util.List<org.tensorflow.proto.framework.TensorProto.Builder> 
-         getTensorProtosBuilderList() {
-      return getTensorProtosFieldBuilder().getBuilderList();
-    }
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.framework.TensorProto, org.tensorflow.proto.framework.TensorProto.Builder, org.tensorflow.proto.framework.TensorProtoOrBuilder> 
-        getTensorProtosFieldBuilder() {
-      if (tensorProtosBuilder_ == null) {
-        tensorProtosBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
-            org.tensorflow.proto.framework.TensorProto, org.tensorflow.proto.framework.TensorProto.Builder, org.tensorflow.proto.framework.TensorProtoOrBuilder>(
-                tensorProtos_,
-                ((bitField0_ & 0x00000004) != 0),
-                getParentForChildren(),
-                isClean());
-        tensorProtos_ = null;
-      }
-      return tensorProtosBuilder_;
-    }
-
-    private org.tensorflow.proto.util.CodeLocation codeLocation_;
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.util.CodeLocation, org.tensorflow.proto.util.CodeLocation.Builder, org.tensorflow.proto.util.CodeLocationOrBuilder> codeLocationBuilder_;
-    /**
-     * <pre>
-     * Stack trace of the eager execution.
-     * </pre>
-     *
-     * <code>.tensorflow.CodeLocation code_location = 8;</code>
-     */
-    public boolean hasCodeLocation() {
-      return codeLocationBuilder_ != null || codeLocation_ != null;
-    }
-    /**
-     * <pre>
-     * Stack trace of the eager execution.
-     * </pre>
-     *
-     * <code>.tensorflow.CodeLocation code_location = 8;</code>
-     */
-    public org.tensorflow.proto.util.CodeLocation getCodeLocation() {
-      if (codeLocationBuilder_ == null) {
-        return codeLocation_ == null ? org.tensorflow.proto.util.CodeLocation.getDefaultInstance() : codeLocation_;
-      } else {
-        return codeLocationBuilder_.getMessage();
-      }
-    }
-    /**
-     * <pre>
-     * Stack trace of the eager execution.
-     * </pre>
-     *
-     * <code>.tensorflow.CodeLocation code_location = 8;</code>
-     */
-    public Builder setCodeLocation(org.tensorflow.proto.util.CodeLocation value) {
-      if (codeLocationBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        codeLocation_ = value;
-        onChanged();
-      } else {
-        codeLocationBuilder_.setMessage(value);
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Stack trace of the eager execution.
-     * </pre>
-     *
-     * <code>.tensorflow.CodeLocation code_location = 8;</code>
-     */
-    public Builder setCodeLocation(
-        org.tensorflow.proto.util.CodeLocation.Builder builderForValue) {
-      if (codeLocationBuilder_ == null) {
-        codeLocation_ = builderForValue.build();
-        onChanged();
-      } else {
-        codeLocationBuilder_.setMessage(builderForValue.build());
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Stack trace of the eager execution.
-     * </pre>
-     *
-     * <code>.tensorflow.CodeLocation code_location = 8;</code>
-     */
-    public Builder mergeCodeLocation(org.tensorflow.proto.util.CodeLocation value) {
-      if (codeLocationBuilder_ == null) {
-        if (codeLocation_ != null) {
-          codeLocation_ =
-            org.tensorflow.proto.util.CodeLocation.newBuilder(codeLocation_).mergeFrom(value).buildPartial();
-        } else {
-          codeLocation_ = value;
-        }
-        onChanged();
-      } else {
-        codeLocationBuilder_.mergeFrom(value);
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Stack trace of the eager execution.
-     * </pre>
-     *
-     * <code>.tensorflow.CodeLocation code_location = 8;</code>
-     */
-    public Builder clearCodeLocation() {
-      if (codeLocationBuilder_ == null) {
-        codeLocation_ = null;
-        onChanged();
-      } else {
-        codeLocation_ = null;
-        codeLocationBuilder_ = null;
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Stack trace of the eager execution.
-     * </pre>
-     *
-     * <code>.tensorflow.CodeLocation code_location = 8;</code>
-     */
-    public org.tensorflow.proto.util.CodeLocation.Builder getCodeLocationBuilder() {
-      
-      onChanged();
-      return getCodeLocationFieldBuilder().getBuilder();
-    }
-    /**
-     * <pre>
-     * Stack trace of the eager execution.
-     * </pre>
-     *
-     * <code>.tensorflow.CodeLocation code_location = 8;</code>
-     */
-    public org.tensorflow.proto.util.CodeLocationOrBuilder getCodeLocationOrBuilder() {
-      if (codeLocationBuilder_ != null) {
-        return codeLocationBuilder_.getMessageOrBuilder();
-      } else {
-        return codeLocation_ == null ?
-            org.tensorflow.proto.util.CodeLocation.getDefaultInstance() : codeLocation_;
-      }
-    }
-    /**
-     * <pre>
-     * Stack trace of the eager execution.
-     * </pre>
-     *
-     * <code>.tensorflow.CodeLocation code_location = 8;</code>
-     */
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.util.CodeLocation, org.tensorflow.proto.util.CodeLocation.Builder, org.tensorflow.proto.util.CodeLocationOrBuilder> 
-        getCodeLocationFieldBuilder() {
-      if (codeLocationBuilder_ == null) {
-        codeLocationBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-            org.tensorflow.proto.util.CodeLocation, org.tensorflow.proto.util.CodeLocation.Builder, org.tensorflow.proto.util.CodeLocationOrBuilder>(
-                getCodeLocation(),
-                getParentForChildren(),
-                isClean());
-        codeLocation_ = null;
-      }
-      return codeLocationBuilder_;
-    }
-
-    private com.google.protobuf.Internal.IntList outputTensorDeviceIds_ = emptyIntList();
-    private void ensureOutputTensorDeviceIdsIsMutable() {
-      if (!((bitField0_ & 0x00000008) != 0)) {
-        outputTensorDeviceIds_ = mutableCopy(outputTensorDeviceIds_);
-        bitField0_ |= 0x00000008;
-       }
-    }
-    /**
-     * <pre>
-     * Debugged-generated IDs of the devices on which the output tensors reside.
-     * To look up details about the device (e.g., name), cross-reference this
-     * field with the DebuggedDevice messages.
-     * </pre>
-     *
-     * <code>repeated int32 output_tensor_device_ids = 9;</code>
-     */
-    public java.util.List<java.lang.Integer>
-        getOutputTensorDeviceIdsList() {
-      return ((bitField0_ & 0x00000008) != 0) ?
-               java.util.Collections.unmodifiableList(outputTensorDeviceIds_) : outputTensorDeviceIds_;
-    }
-    /**
-     * <pre>
-     * Debugged-generated IDs of the devices on which the output tensors reside.
-     * To look up details about the device (e.g., name), cross-reference this
-     * field with the DebuggedDevice messages.
-     * </pre>
-     *
-     * <code>repeated int32 output_tensor_device_ids = 9;</code>
-     */
-    public int getOutputTensorDeviceIdsCount() {
-      return outputTensorDeviceIds_.size();
-    }
-    /**
-     * <pre>
-     * Debugged-generated IDs of the devices on which the output tensors reside.
-     * To look up details about the device (e.g., name), cross-reference this
-     * field with the DebuggedDevice messages.
-     * </pre>
-     *
-     * <code>repeated int32 output_tensor_device_ids = 9;</code>
-     */
-    public int getOutputTensorDeviceIds(int index) {
-      return outputTensorDeviceIds_.getInt(index);
-    }
-    /**
-     * <pre>
-     * Debugged-generated IDs of the devices on which the output tensors reside.
-     * To look up details about the device (e.g., name), cross-reference this
-     * field with the DebuggedDevice messages.
-     * </pre>
-     *
-     * <code>repeated int32 output_tensor_device_ids = 9;</code>
-     */
-    public Builder setOutputTensorDeviceIds(
-        int index, int value) {
-      ensureOutputTensorDeviceIdsIsMutable();
-      outputTensorDeviceIds_.setInt(index, value);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Debugged-generated IDs of the devices on which the output tensors reside.
-     * To look up details about the device (e.g., name), cross-reference this
-     * field with the DebuggedDevice messages.
-     * </pre>
-     *
-     * <code>repeated int32 output_tensor_device_ids = 9;</code>
-     */
-    public Builder addOutputTensorDeviceIds(int value) {
-      ensureOutputTensorDeviceIdsIsMutable();
-      outputTensorDeviceIds_.addInt(value);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Debugged-generated IDs of the devices on which the output tensors reside.
-     * To look up details about the device (e.g., name), cross-reference this
-     * field with the DebuggedDevice messages.
-     * </pre>
-     *
-     * <code>repeated int32 output_tensor_device_ids = 9;</code>
-     */
-    public Builder addAllOutputTensorDeviceIds(
-        java.lang.Iterable<? extends java.lang.Integer> values) {
-      ensureOutputTensorDeviceIdsIsMutable();
-      com.google.protobuf.AbstractMessageLite.Builder.addAll(
-          values, outputTensorDeviceIds_);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Debugged-generated IDs of the devices on which the output tensors reside.
-     * To look up details about the device (e.g., name), cross-reference this
-     * field with the DebuggedDevice messages.
-     * </pre>
-     *
-     * <code>repeated int32 output_tensor_device_ids = 9;</code>
-     */
-    public Builder clearOutputTensorDeviceIds() {
-      outputTensorDeviceIds_ = emptyIntList();
-      bitField0_ = (bitField0_ & ~0x00000008);
-      onChanged();
-      return this;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.Execution)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.Execution)
-  private static final org.tensorflow.proto.util.Execution DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.util.Execution();
-  }
-
-  public static org.tensorflow.proto.util.Execution getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<Execution>
-      PARSER = new com.google.protobuf.AbstractParser<Execution>() {
-    @java.lang.Override
-    public Execution parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new Execution(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<Execution> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<Execution> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.util.Execution getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/ExecutionOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/ExecutionOrBuilder.java
deleted file mode 100644
index c08db8762e5..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/ExecutionOrBuilder.java
+++ /dev/null
@@ -1,233 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/debug_event.proto
-
-package org.tensorflow.proto.util;
-
-public interface ExecutionOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.Execution)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <pre>
-   * Op type (e.g., "MatMul").
-   * In the case of a Graph, this is the name of the Graph.
-   * </pre>
-   *
-   * <code>string op_type = 1;</code>
-   */
-  java.lang.String getOpType();
-  /**
-   * <pre>
-   * Op type (e.g., "MatMul").
-   * In the case of a Graph, this is the name of the Graph.
-   * </pre>
-   *
-   * <code>string op_type = 1;</code>
-   */
-  com.google.protobuf.ByteString
-      getOpTypeBytes();
-
-  /**
-   * <pre>
-   * Number of output tensors.
-   * </pre>
-   *
-   * <code>int32 num_outputs = 2;</code>
-   */
-  int getNumOutputs();
-
-  /**
-   * <pre>
-   * The graph that's executed: applicable only to the eager
-   * execution of a FuncGraph.
-   * </pre>
-   *
-   * <code>string graph_id = 3;</code>
-   */
-  java.lang.String getGraphId();
-  /**
-   * <pre>
-   * The graph that's executed: applicable only to the eager
-   * execution of a FuncGraph.
-   * </pre>
-   *
-   * <code>string graph_id = 3;</code>
-   */
-  com.google.protobuf.ByteString
-      getGraphIdBytes();
-
-  /**
-   * <pre>
-   * IDs of the input tensors (if available).
-   * </pre>
-   *
-   * <code>repeated int64 input_tensor_ids = 4;</code>
-   */
-  java.util.List<java.lang.Long> getInputTensorIdsList();
-  /**
-   * <pre>
-   * IDs of the input tensors (if available).
-   * </pre>
-   *
-   * <code>repeated int64 input_tensor_ids = 4;</code>
-   */
-  int getInputTensorIdsCount();
-  /**
-   * <pre>
-   * IDs of the input tensors (if available).
-   * </pre>
-   *
-   * <code>repeated int64 input_tensor_ids = 4;</code>
-   */
-  long getInputTensorIds(int index);
-
-  /**
-   * <pre>
-   * IDs of the output tensors (if availbable).
-   * If specified, must have the same length as tensor_protos.
-   * </pre>
-   *
-   * <code>repeated int64 output_tensor_ids = 5;</code>
-   */
-  java.util.List<java.lang.Long> getOutputTensorIdsList();
-  /**
-   * <pre>
-   * IDs of the output tensors (if availbable).
-   * If specified, must have the same length as tensor_protos.
-   * </pre>
-   *
-   * <code>repeated int64 output_tensor_ids = 5;</code>
-   */
-  int getOutputTensorIdsCount();
-  /**
-   * <pre>
-   * IDs of the output tensors (if availbable).
-   * If specified, must have the same length as tensor_protos.
-   * </pre>
-   *
-   * <code>repeated int64 output_tensor_ids = 5;</code>
-   */
-  long getOutputTensorIds(int index);
-
-  /**
-   * <pre>
-   * Type of the tensor value encapsulated in this proto.
-   * </pre>
-   *
-   * <code>.tensorflow.TensorDebugMode tensor_debug_mode = 6;</code>
-   */
-  int getTensorDebugModeValue();
-  /**
-   * <pre>
-   * Type of the tensor value encapsulated in this proto.
-   * </pre>
-   *
-   * <code>.tensorflow.TensorDebugMode tensor_debug_mode = 6;</code>
-   */
-  org.tensorflow.proto.util.TensorDebugMode getTensorDebugMode();
-
-  /**
-   * <pre>
-   * Output Tensor values in the type described by `tensor_value_type`.
-   * The length of this should match `num_outputs`.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.TensorProto tensor_protos = 7;</code>
-   */
-  java.util.List<org.tensorflow.proto.framework.TensorProto> 
-      getTensorProtosList();
-  /**
-   * <pre>
-   * Output Tensor values in the type described by `tensor_value_type`.
-   * The length of this should match `num_outputs`.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.TensorProto tensor_protos = 7;</code>
-   */
-  org.tensorflow.proto.framework.TensorProto getTensorProtos(int index);
-  /**
-   * <pre>
-   * Output Tensor values in the type described by `tensor_value_type`.
-   * The length of this should match `num_outputs`.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.TensorProto tensor_protos = 7;</code>
-   */
-  int getTensorProtosCount();
-  /**
-   * <pre>
-   * Output Tensor values in the type described by `tensor_value_type`.
-   * The length of this should match `num_outputs`.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.TensorProto tensor_protos = 7;</code>
-   */
-  java.util.List<? extends org.tensorflow.proto.framework.TensorProtoOrBuilder> 
-      getTensorProtosOrBuilderList();
-  /**
-   * <pre>
-   * Output Tensor values in the type described by `tensor_value_type`.
-   * The length of this should match `num_outputs`.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.TensorProto tensor_protos = 7;</code>
-   */
-  org.tensorflow.proto.framework.TensorProtoOrBuilder getTensorProtosOrBuilder(
-      int index);
-
-  /**
-   * <pre>
-   * Stack trace of the eager execution.
-   * </pre>
-   *
-   * <code>.tensorflow.CodeLocation code_location = 8;</code>
-   */
-  boolean hasCodeLocation();
-  /**
-   * <pre>
-   * Stack trace of the eager execution.
-   * </pre>
-   *
-   * <code>.tensorflow.CodeLocation code_location = 8;</code>
-   */
-  org.tensorflow.proto.util.CodeLocation getCodeLocation();
-  /**
-   * <pre>
-   * Stack trace of the eager execution.
-   * </pre>
-   *
-   * <code>.tensorflow.CodeLocation code_location = 8;</code>
-   */
-  org.tensorflow.proto.util.CodeLocationOrBuilder getCodeLocationOrBuilder();
-
-  /**
-   * <pre>
-   * Debugged-generated IDs of the devices on which the output tensors reside.
-   * To look up details about the device (e.g., name), cross-reference this
-   * field with the DebuggedDevice messages.
-   * </pre>
-   *
-   * <code>repeated int32 output_tensor_device_ids = 9;</code>
-   */
-  java.util.List<java.lang.Integer> getOutputTensorDeviceIdsList();
-  /**
-   * <pre>
-   * Debugged-generated IDs of the devices on which the output tensors reside.
-   * To look up details about the device (e.g., name), cross-reference this
-   * field with the DebuggedDevice messages.
-   * </pre>
-   *
-   * <code>repeated int32 output_tensor_device_ids = 9;</code>
-   */
-  int getOutputTensorDeviceIdsCount();
-  /**
-   * <pre>
-   * Debugged-generated IDs of the devices on which the output tensors reside.
-   * To look up details about the device (e.g., name), cross-reference this
-   * field with the DebuggedDevice messages.
-   * </pre>
-   *
-   * <code>repeated int32 output_tensor_device_ids = 9;</code>
-   */
-  int getOutputTensorDeviceIds(int index);
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/GraphExecutionTrace.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/GraphExecutionTrace.java
deleted file mode 100644
index 3ea9e244ff2..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/GraphExecutionTrace.java
+++ /dev/null
@@ -1,1359 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/debug_event.proto
-
-package org.tensorflow.proto.util;
-
-/**
- * <pre>
- * Data relating to an execution of a Graph (e.g., an eager execution of a
- * FuncGraph).
- * The values of the intermediate tensors computed in the graph are recorded
- * in this proto. A graph execution may correspond to one or more pieces of
- * `GraphExecutionTrace`, depending on whether the instrumented tensor values
- * are summarized in an aggregated or separate fashion.
- * </pre>
- *
- * Protobuf type {@code tensorflow.GraphExecutionTrace}
- */
-public  final class GraphExecutionTrace extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.GraphExecutionTrace)
-    GraphExecutionTraceOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use GraphExecutionTrace.newBuilder() to construct.
-  private GraphExecutionTrace(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private GraphExecutionTrace() {
-    tfdbgContextId_ = "";
-    opName_ = "";
-    tensorDebugMode_ = 0;
-    deviceName_ = "";
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new GraphExecutionTrace();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private GraphExecutionTrace(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 10: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            tfdbgContextId_ = s;
-            break;
-          }
-          case 18: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            opName_ = s;
-            break;
-          }
-          case 24: {
-
-            outputSlot_ = input.readInt32();
-            break;
-          }
-          case 32: {
-            int rawValue = input.readEnum();
-
-            tensorDebugMode_ = rawValue;
-            break;
-          }
-          case 42: {
-            org.tensorflow.proto.framework.TensorProto.Builder subBuilder = null;
-            if (tensorProto_ != null) {
-              subBuilder = tensorProto_.toBuilder();
-            }
-            tensorProto_ = input.readMessage(org.tensorflow.proto.framework.TensorProto.parser(), extensionRegistry);
-            if (subBuilder != null) {
-              subBuilder.mergeFrom(tensorProto_);
-              tensorProto_ = subBuilder.buildPartial();
-            }
-
-            break;
-          }
-          case 50: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            deviceName_ = s;
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.util.DebugEventProtos.internal_static_tensorflow_GraphExecutionTrace_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.util.DebugEventProtos.internal_static_tensorflow_GraphExecutionTrace_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.util.GraphExecutionTrace.class, org.tensorflow.proto.util.GraphExecutionTrace.Builder.class);
-  }
-
-  public static final int TFDBG_CONTEXT_ID_FIELD_NUMBER = 1;
-  private volatile java.lang.Object tfdbgContextId_;
-  /**
-   * <pre>
-   * Unique ID of the context that the executed op(s) belong to (e.g., a
-   * compiled concrete tf.function).
-   * </pre>
-   *
-   * <code>string tfdbg_context_id = 1;</code>
-   */
-  public java.lang.String getTfdbgContextId() {
-    java.lang.Object ref = tfdbgContextId_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      tfdbgContextId_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * Unique ID of the context that the executed op(s) belong to (e.g., a
-   * compiled concrete tf.function).
-   * </pre>
-   *
-   * <code>string tfdbg_context_id = 1;</code>
-   */
-  public com.google.protobuf.ByteString
-      getTfdbgContextIdBytes() {
-    java.lang.Object ref = tfdbgContextId_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      tfdbgContextId_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int OP_NAME_FIELD_NUMBER = 2;
-  private volatile java.lang.Object opName_;
-  /**
-   * <pre>
-   * Name of the op (applicable only in the case of the `FULL_TENSOR` trace
-   * level).
-   * </pre>
-   *
-   * <code>string op_name = 2;</code>
-   */
-  public java.lang.String getOpName() {
-    java.lang.Object ref = opName_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      opName_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * Name of the op (applicable only in the case of the `FULL_TENSOR` trace
-   * level).
-   * </pre>
-   *
-   * <code>string op_name = 2;</code>
-   */
-  public com.google.protobuf.ByteString
-      getOpNameBytes() {
-    java.lang.Object ref = opName_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      opName_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int OUTPUT_SLOT_FIELD_NUMBER = 3;
-  private int outputSlot_;
-  /**
-   * <pre>
-   * Output slot of the tensor (applicable only in the case of the `FULL_TENSOR`
-   * trace level).
-   * </pre>
-   *
-   * <code>int32 output_slot = 3;</code>
-   */
-  public int getOutputSlot() {
-    return outputSlot_;
-  }
-
-  public static final int TENSOR_DEBUG_MODE_FIELD_NUMBER = 4;
-  private int tensorDebugMode_;
-  /**
-   * <pre>
-   * Type of the tensor value encapsulated in this proto.
-   * </pre>
-   *
-   * <code>.tensorflow.TensorDebugMode tensor_debug_mode = 4;</code>
-   */
-  public int getTensorDebugModeValue() {
-    return tensorDebugMode_;
-  }
-  /**
-   * <pre>
-   * Type of the tensor value encapsulated in this proto.
-   * </pre>
-   *
-   * <code>.tensorflow.TensorDebugMode tensor_debug_mode = 4;</code>
-   */
-  public org.tensorflow.proto.util.TensorDebugMode getTensorDebugMode() {
-    @SuppressWarnings("deprecation")
-    org.tensorflow.proto.util.TensorDebugMode result = org.tensorflow.proto.util.TensorDebugMode.valueOf(tensorDebugMode_);
-    return result == null ? org.tensorflow.proto.util.TensorDebugMode.UNRECOGNIZED : result;
-  }
-
-  public static final int TENSOR_PROTO_FIELD_NUMBER = 5;
-  private org.tensorflow.proto.framework.TensorProto tensorProto_;
-  /**
-   * <pre>
-   * Tensor value in the type described by `tensor_value_type`.
-   * This tensor may summarize the value of a single intermediate op of the
-   * graph, or those of multiple intermediate tensors.
-   * </pre>
-   *
-   * <code>.tensorflow.TensorProto tensor_proto = 5;</code>
-   */
-  public boolean hasTensorProto() {
-    return tensorProto_ != null;
-  }
-  /**
-   * <pre>
-   * Tensor value in the type described by `tensor_value_type`.
-   * This tensor may summarize the value of a single intermediate op of the
-   * graph, or those of multiple intermediate tensors.
-   * </pre>
-   *
-   * <code>.tensorflow.TensorProto tensor_proto = 5;</code>
-   */
-  public org.tensorflow.proto.framework.TensorProto getTensorProto() {
-    return tensorProto_ == null ? org.tensorflow.proto.framework.TensorProto.getDefaultInstance() : tensorProto_;
-  }
-  /**
-   * <pre>
-   * Tensor value in the type described by `tensor_value_type`.
-   * This tensor may summarize the value of a single intermediate op of the
-   * graph, or those of multiple intermediate tensors.
-   * </pre>
-   *
-   * <code>.tensorflow.TensorProto tensor_proto = 5;</code>
-   */
-  public org.tensorflow.proto.framework.TensorProtoOrBuilder getTensorProtoOrBuilder() {
-    return getTensorProto();
-  }
-
-  public static final int DEVICE_NAME_FIELD_NUMBER = 6;
-  private volatile java.lang.Object deviceName_;
-  /**
-   * <pre>
-   * Name of the device that the op belongs to.
-   * </pre>
-   *
-   * <code>string device_name = 6;</code>
-   */
-  public java.lang.String getDeviceName() {
-    java.lang.Object ref = deviceName_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      deviceName_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * Name of the device that the op belongs to.
-   * </pre>
-   *
-   * <code>string device_name = 6;</code>
-   */
-  public com.google.protobuf.ByteString
-      getDeviceNameBytes() {
-    java.lang.Object ref = deviceName_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      deviceName_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    if (!getTfdbgContextIdBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 1, tfdbgContextId_);
-    }
-    if (!getOpNameBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 2, opName_);
-    }
-    if (outputSlot_ != 0) {
-      output.writeInt32(3, outputSlot_);
-    }
-    if (tensorDebugMode_ != org.tensorflow.proto.util.TensorDebugMode.UNSPECIFIED.getNumber()) {
-      output.writeEnum(4, tensorDebugMode_);
-    }
-    if (tensorProto_ != null) {
-      output.writeMessage(5, getTensorProto());
-    }
-    if (!getDeviceNameBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 6, deviceName_);
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (!getTfdbgContextIdBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, tfdbgContextId_);
-    }
-    if (!getOpNameBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(2, opName_);
-    }
-    if (outputSlot_ != 0) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt32Size(3, outputSlot_);
-    }
-    if (tensorDebugMode_ != org.tensorflow.proto.util.TensorDebugMode.UNSPECIFIED.getNumber()) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeEnumSize(4, tensorDebugMode_);
-    }
-    if (tensorProto_ != null) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(5, getTensorProto());
-    }
-    if (!getDeviceNameBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(6, deviceName_);
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.util.GraphExecutionTrace)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.util.GraphExecutionTrace other = (org.tensorflow.proto.util.GraphExecutionTrace) obj;
-
-    if (!getTfdbgContextId()
-        .equals(other.getTfdbgContextId())) return false;
-    if (!getOpName()
-        .equals(other.getOpName())) return false;
-    if (getOutputSlot()
-        != other.getOutputSlot()) return false;
-    if (tensorDebugMode_ != other.tensorDebugMode_) return false;
-    if (hasTensorProto() != other.hasTensorProto()) return false;
-    if (hasTensorProto()) {
-      if (!getTensorProto()
-          .equals(other.getTensorProto())) return false;
-    }
-    if (!getDeviceName()
-        .equals(other.getDeviceName())) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    hash = (37 * hash) + TFDBG_CONTEXT_ID_FIELD_NUMBER;
-    hash = (53 * hash) + getTfdbgContextId().hashCode();
-    hash = (37 * hash) + OP_NAME_FIELD_NUMBER;
-    hash = (53 * hash) + getOpName().hashCode();
-    hash = (37 * hash) + OUTPUT_SLOT_FIELD_NUMBER;
-    hash = (53 * hash) + getOutputSlot();
-    hash = (37 * hash) + TENSOR_DEBUG_MODE_FIELD_NUMBER;
-    hash = (53 * hash) + tensorDebugMode_;
-    if (hasTensorProto()) {
-      hash = (37 * hash) + TENSOR_PROTO_FIELD_NUMBER;
-      hash = (53 * hash) + getTensorProto().hashCode();
-    }
-    hash = (37 * hash) + DEVICE_NAME_FIELD_NUMBER;
-    hash = (53 * hash) + getDeviceName().hashCode();
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.util.GraphExecutionTrace parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.GraphExecutionTrace parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.GraphExecutionTrace parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.GraphExecutionTrace parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.GraphExecutionTrace parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.GraphExecutionTrace parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.GraphExecutionTrace parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.GraphExecutionTrace parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.GraphExecutionTrace parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.GraphExecutionTrace parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.GraphExecutionTrace parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.GraphExecutionTrace parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.util.GraphExecutionTrace prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * <pre>
-   * Data relating to an execution of a Graph (e.g., an eager execution of a
-   * FuncGraph).
-   * The values of the intermediate tensors computed in the graph are recorded
-   * in this proto. A graph execution may correspond to one or more pieces of
-   * `GraphExecutionTrace`, depending on whether the instrumented tensor values
-   * are summarized in an aggregated or separate fashion.
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.GraphExecutionTrace}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.GraphExecutionTrace)
-      org.tensorflow.proto.util.GraphExecutionTraceOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.util.DebugEventProtos.internal_static_tensorflow_GraphExecutionTrace_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.util.DebugEventProtos.internal_static_tensorflow_GraphExecutionTrace_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.util.GraphExecutionTrace.class, org.tensorflow.proto.util.GraphExecutionTrace.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.util.GraphExecutionTrace.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      tfdbgContextId_ = "";
-
-      opName_ = "";
-
-      outputSlot_ = 0;
-
-      tensorDebugMode_ = 0;
-
-      if (tensorProtoBuilder_ == null) {
-        tensorProto_ = null;
-      } else {
-        tensorProto_ = null;
-        tensorProtoBuilder_ = null;
-      }
-      deviceName_ = "";
-
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.util.DebugEventProtos.internal_static_tensorflow_GraphExecutionTrace_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.GraphExecutionTrace getDefaultInstanceForType() {
-      return org.tensorflow.proto.util.GraphExecutionTrace.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.GraphExecutionTrace build() {
-      org.tensorflow.proto.util.GraphExecutionTrace result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.GraphExecutionTrace buildPartial() {
-      org.tensorflow.proto.util.GraphExecutionTrace result = new org.tensorflow.proto.util.GraphExecutionTrace(this);
-      result.tfdbgContextId_ = tfdbgContextId_;
-      result.opName_ = opName_;
-      result.outputSlot_ = outputSlot_;
-      result.tensorDebugMode_ = tensorDebugMode_;
-      if (tensorProtoBuilder_ == null) {
-        result.tensorProto_ = tensorProto_;
-      } else {
-        result.tensorProto_ = tensorProtoBuilder_.build();
-      }
-      result.deviceName_ = deviceName_;
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.util.GraphExecutionTrace) {
-        return mergeFrom((org.tensorflow.proto.util.GraphExecutionTrace)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.util.GraphExecutionTrace other) {
-      if (other == org.tensorflow.proto.util.GraphExecutionTrace.getDefaultInstance()) return this;
-      if (!other.getTfdbgContextId().isEmpty()) {
-        tfdbgContextId_ = other.tfdbgContextId_;
-        onChanged();
-      }
-      if (!other.getOpName().isEmpty()) {
-        opName_ = other.opName_;
-        onChanged();
-      }
-      if (other.getOutputSlot() != 0) {
-        setOutputSlot(other.getOutputSlot());
-      }
-      if (other.tensorDebugMode_ != 0) {
-        setTensorDebugModeValue(other.getTensorDebugModeValue());
-      }
-      if (other.hasTensorProto()) {
-        mergeTensorProto(other.getTensorProto());
-      }
-      if (!other.getDeviceName().isEmpty()) {
-        deviceName_ = other.deviceName_;
-        onChanged();
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.util.GraphExecutionTrace parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.util.GraphExecutionTrace) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-
-    private java.lang.Object tfdbgContextId_ = "";
-    /**
-     * <pre>
-     * Unique ID of the context that the executed op(s) belong to (e.g., a
-     * compiled concrete tf.function).
-     * </pre>
-     *
-     * <code>string tfdbg_context_id = 1;</code>
-     */
-    public java.lang.String getTfdbgContextId() {
-      java.lang.Object ref = tfdbgContextId_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        tfdbgContextId_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Unique ID of the context that the executed op(s) belong to (e.g., a
-     * compiled concrete tf.function).
-     * </pre>
-     *
-     * <code>string tfdbg_context_id = 1;</code>
-     */
-    public com.google.protobuf.ByteString
-        getTfdbgContextIdBytes() {
-      java.lang.Object ref = tfdbgContextId_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        tfdbgContextId_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Unique ID of the context that the executed op(s) belong to (e.g., a
-     * compiled concrete tf.function).
-     * </pre>
-     *
-     * <code>string tfdbg_context_id = 1;</code>
-     */
-    public Builder setTfdbgContextId(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      tfdbgContextId_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Unique ID of the context that the executed op(s) belong to (e.g., a
-     * compiled concrete tf.function).
-     * </pre>
-     *
-     * <code>string tfdbg_context_id = 1;</code>
-     */
-    public Builder clearTfdbgContextId() {
-      
-      tfdbgContextId_ = getDefaultInstance().getTfdbgContextId();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Unique ID of the context that the executed op(s) belong to (e.g., a
-     * compiled concrete tf.function).
-     * </pre>
-     *
-     * <code>string tfdbg_context_id = 1;</code>
-     */
-    public Builder setTfdbgContextIdBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      tfdbgContextId_ = value;
-      onChanged();
-      return this;
-    }
-
-    private java.lang.Object opName_ = "";
-    /**
-     * <pre>
-     * Name of the op (applicable only in the case of the `FULL_TENSOR` trace
-     * level).
-     * </pre>
-     *
-     * <code>string op_name = 2;</code>
-     */
-    public java.lang.String getOpName() {
-      java.lang.Object ref = opName_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        opName_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Name of the op (applicable only in the case of the `FULL_TENSOR` trace
-     * level).
-     * </pre>
-     *
-     * <code>string op_name = 2;</code>
-     */
-    public com.google.protobuf.ByteString
-        getOpNameBytes() {
-      java.lang.Object ref = opName_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        opName_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Name of the op (applicable only in the case of the `FULL_TENSOR` trace
-     * level).
-     * </pre>
-     *
-     * <code>string op_name = 2;</code>
-     */
-    public Builder setOpName(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      opName_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Name of the op (applicable only in the case of the `FULL_TENSOR` trace
-     * level).
-     * </pre>
-     *
-     * <code>string op_name = 2;</code>
-     */
-    public Builder clearOpName() {
-      
-      opName_ = getDefaultInstance().getOpName();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Name of the op (applicable only in the case of the `FULL_TENSOR` trace
-     * level).
-     * </pre>
-     *
-     * <code>string op_name = 2;</code>
-     */
-    public Builder setOpNameBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      opName_ = value;
-      onChanged();
-      return this;
-    }
-
-    private int outputSlot_ ;
-    /**
-     * <pre>
-     * Output slot of the tensor (applicable only in the case of the `FULL_TENSOR`
-     * trace level).
-     * </pre>
-     *
-     * <code>int32 output_slot = 3;</code>
-     */
-    public int getOutputSlot() {
-      return outputSlot_;
-    }
-    /**
-     * <pre>
-     * Output slot of the tensor (applicable only in the case of the `FULL_TENSOR`
-     * trace level).
-     * </pre>
-     *
-     * <code>int32 output_slot = 3;</code>
-     */
-    public Builder setOutputSlot(int value) {
-      
-      outputSlot_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Output slot of the tensor (applicable only in the case of the `FULL_TENSOR`
-     * trace level).
-     * </pre>
-     *
-     * <code>int32 output_slot = 3;</code>
-     */
-    public Builder clearOutputSlot() {
-      
-      outputSlot_ = 0;
-      onChanged();
-      return this;
-    }
-
-    private int tensorDebugMode_ = 0;
-    /**
-     * <pre>
-     * Type of the tensor value encapsulated in this proto.
-     * </pre>
-     *
-     * <code>.tensorflow.TensorDebugMode tensor_debug_mode = 4;</code>
-     */
-    public int getTensorDebugModeValue() {
-      return tensorDebugMode_;
-    }
-    /**
-     * <pre>
-     * Type of the tensor value encapsulated in this proto.
-     * </pre>
-     *
-     * <code>.tensorflow.TensorDebugMode tensor_debug_mode = 4;</code>
-     */
-    public Builder setTensorDebugModeValue(int value) {
-      tensorDebugMode_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Type of the tensor value encapsulated in this proto.
-     * </pre>
-     *
-     * <code>.tensorflow.TensorDebugMode tensor_debug_mode = 4;</code>
-     */
-    public org.tensorflow.proto.util.TensorDebugMode getTensorDebugMode() {
-      @SuppressWarnings("deprecation")
-      org.tensorflow.proto.util.TensorDebugMode result = org.tensorflow.proto.util.TensorDebugMode.valueOf(tensorDebugMode_);
-      return result == null ? org.tensorflow.proto.util.TensorDebugMode.UNRECOGNIZED : result;
-    }
-    /**
-     * <pre>
-     * Type of the tensor value encapsulated in this proto.
-     * </pre>
-     *
-     * <code>.tensorflow.TensorDebugMode tensor_debug_mode = 4;</code>
-     */
-    public Builder setTensorDebugMode(org.tensorflow.proto.util.TensorDebugMode value) {
-      if (value == null) {
-        throw new NullPointerException();
-      }
-      
-      tensorDebugMode_ = value.getNumber();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Type of the tensor value encapsulated in this proto.
-     * </pre>
-     *
-     * <code>.tensorflow.TensorDebugMode tensor_debug_mode = 4;</code>
-     */
-    public Builder clearTensorDebugMode() {
-      
-      tensorDebugMode_ = 0;
-      onChanged();
-      return this;
-    }
-
-    private org.tensorflow.proto.framework.TensorProto tensorProto_;
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.TensorProto, org.tensorflow.proto.framework.TensorProto.Builder, org.tensorflow.proto.framework.TensorProtoOrBuilder> tensorProtoBuilder_;
-    /**
-     * <pre>
-     * Tensor value in the type described by `tensor_value_type`.
-     * This tensor may summarize the value of a single intermediate op of the
-     * graph, or those of multiple intermediate tensors.
-     * </pre>
-     *
-     * <code>.tensorflow.TensorProto tensor_proto = 5;</code>
-     */
-    public boolean hasTensorProto() {
-      return tensorProtoBuilder_ != null || tensorProto_ != null;
-    }
-    /**
-     * <pre>
-     * Tensor value in the type described by `tensor_value_type`.
-     * This tensor may summarize the value of a single intermediate op of the
-     * graph, or those of multiple intermediate tensors.
-     * </pre>
-     *
-     * <code>.tensorflow.TensorProto tensor_proto = 5;</code>
-     */
-    public org.tensorflow.proto.framework.TensorProto getTensorProto() {
-      if (tensorProtoBuilder_ == null) {
-        return tensorProto_ == null ? org.tensorflow.proto.framework.TensorProto.getDefaultInstance() : tensorProto_;
-      } else {
-        return tensorProtoBuilder_.getMessage();
-      }
-    }
-    /**
-     * <pre>
-     * Tensor value in the type described by `tensor_value_type`.
-     * This tensor may summarize the value of a single intermediate op of the
-     * graph, or those of multiple intermediate tensors.
-     * </pre>
-     *
-     * <code>.tensorflow.TensorProto tensor_proto = 5;</code>
-     */
-    public Builder setTensorProto(org.tensorflow.proto.framework.TensorProto value) {
-      if (tensorProtoBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        tensorProto_ = value;
-        onChanged();
-      } else {
-        tensorProtoBuilder_.setMessage(value);
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Tensor value in the type described by `tensor_value_type`.
-     * This tensor may summarize the value of a single intermediate op of the
-     * graph, or those of multiple intermediate tensors.
-     * </pre>
-     *
-     * <code>.tensorflow.TensorProto tensor_proto = 5;</code>
-     */
-    public Builder setTensorProto(
-        org.tensorflow.proto.framework.TensorProto.Builder builderForValue) {
-      if (tensorProtoBuilder_ == null) {
-        tensorProto_ = builderForValue.build();
-        onChanged();
-      } else {
-        tensorProtoBuilder_.setMessage(builderForValue.build());
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Tensor value in the type described by `tensor_value_type`.
-     * This tensor may summarize the value of a single intermediate op of the
-     * graph, or those of multiple intermediate tensors.
-     * </pre>
-     *
-     * <code>.tensorflow.TensorProto tensor_proto = 5;</code>
-     */
-    public Builder mergeTensorProto(org.tensorflow.proto.framework.TensorProto value) {
-      if (tensorProtoBuilder_ == null) {
-        if (tensorProto_ != null) {
-          tensorProto_ =
-            org.tensorflow.proto.framework.TensorProto.newBuilder(tensorProto_).mergeFrom(value).buildPartial();
-        } else {
-          tensorProto_ = value;
-        }
-        onChanged();
-      } else {
-        tensorProtoBuilder_.mergeFrom(value);
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Tensor value in the type described by `tensor_value_type`.
-     * This tensor may summarize the value of a single intermediate op of the
-     * graph, or those of multiple intermediate tensors.
-     * </pre>
-     *
-     * <code>.tensorflow.TensorProto tensor_proto = 5;</code>
-     */
-    public Builder clearTensorProto() {
-      if (tensorProtoBuilder_ == null) {
-        tensorProto_ = null;
-        onChanged();
-      } else {
-        tensorProto_ = null;
-        tensorProtoBuilder_ = null;
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Tensor value in the type described by `tensor_value_type`.
-     * This tensor may summarize the value of a single intermediate op of the
-     * graph, or those of multiple intermediate tensors.
-     * </pre>
-     *
-     * <code>.tensorflow.TensorProto tensor_proto = 5;</code>
-     */
-    public org.tensorflow.proto.framework.TensorProto.Builder getTensorProtoBuilder() {
-      
-      onChanged();
-      return getTensorProtoFieldBuilder().getBuilder();
-    }
-    /**
-     * <pre>
-     * Tensor value in the type described by `tensor_value_type`.
-     * This tensor may summarize the value of a single intermediate op of the
-     * graph, or those of multiple intermediate tensors.
-     * </pre>
-     *
-     * <code>.tensorflow.TensorProto tensor_proto = 5;</code>
-     */
-    public org.tensorflow.proto.framework.TensorProtoOrBuilder getTensorProtoOrBuilder() {
-      if (tensorProtoBuilder_ != null) {
-        return tensorProtoBuilder_.getMessageOrBuilder();
-      } else {
-        return tensorProto_ == null ?
-            org.tensorflow.proto.framework.TensorProto.getDefaultInstance() : tensorProto_;
-      }
-    }
-    /**
-     * <pre>
-     * Tensor value in the type described by `tensor_value_type`.
-     * This tensor may summarize the value of a single intermediate op of the
-     * graph, or those of multiple intermediate tensors.
-     * </pre>
-     *
-     * <code>.tensorflow.TensorProto tensor_proto = 5;</code>
-     */
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.TensorProto, org.tensorflow.proto.framework.TensorProto.Builder, org.tensorflow.proto.framework.TensorProtoOrBuilder> 
-        getTensorProtoFieldBuilder() {
-      if (tensorProtoBuilder_ == null) {
-        tensorProtoBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-            org.tensorflow.proto.framework.TensorProto, org.tensorflow.proto.framework.TensorProto.Builder, org.tensorflow.proto.framework.TensorProtoOrBuilder>(
-                getTensorProto(),
-                getParentForChildren(),
-                isClean());
-        tensorProto_ = null;
-      }
-      return tensorProtoBuilder_;
-    }
-
-    private java.lang.Object deviceName_ = "";
-    /**
-     * <pre>
-     * Name of the device that the op belongs to.
-     * </pre>
-     *
-     * <code>string device_name = 6;</code>
-     */
-    public java.lang.String getDeviceName() {
-      java.lang.Object ref = deviceName_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        deviceName_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Name of the device that the op belongs to.
-     * </pre>
-     *
-     * <code>string device_name = 6;</code>
-     */
-    public com.google.protobuf.ByteString
-        getDeviceNameBytes() {
-      java.lang.Object ref = deviceName_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        deviceName_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Name of the device that the op belongs to.
-     * </pre>
-     *
-     * <code>string device_name = 6;</code>
-     */
-    public Builder setDeviceName(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      deviceName_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Name of the device that the op belongs to.
-     * </pre>
-     *
-     * <code>string device_name = 6;</code>
-     */
-    public Builder clearDeviceName() {
-      
-      deviceName_ = getDefaultInstance().getDeviceName();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Name of the device that the op belongs to.
-     * </pre>
-     *
-     * <code>string device_name = 6;</code>
-     */
-    public Builder setDeviceNameBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      deviceName_ = value;
-      onChanged();
-      return this;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.GraphExecutionTrace)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.GraphExecutionTrace)
-  private static final org.tensorflow.proto.util.GraphExecutionTrace DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.util.GraphExecutionTrace();
-  }
-
-  public static org.tensorflow.proto.util.GraphExecutionTrace getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<GraphExecutionTrace>
-      PARSER = new com.google.protobuf.AbstractParser<GraphExecutionTrace>() {
-    @java.lang.Override
-    public GraphExecutionTrace parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new GraphExecutionTrace(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<GraphExecutionTrace> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<GraphExecutionTrace> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.util.GraphExecutionTrace getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/GraphExecutionTraceOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/GraphExecutionTraceOrBuilder.java
deleted file mode 100644
index d83c52096e3..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/GraphExecutionTraceOrBuilder.java
+++ /dev/null
@@ -1,125 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/debug_event.proto
-
-package org.tensorflow.proto.util;
-
-public interface GraphExecutionTraceOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.GraphExecutionTrace)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <pre>
-   * Unique ID of the context that the executed op(s) belong to (e.g., a
-   * compiled concrete tf.function).
-   * </pre>
-   *
-   * <code>string tfdbg_context_id = 1;</code>
-   */
-  java.lang.String getTfdbgContextId();
-  /**
-   * <pre>
-   * Unique ID of the context that the executed op(s) belong to (e.g., a
-   * compiled concrete tf.function).
-   * </pre>
-   *
-   * <code>string tfdbg_context_id = 1;</code>
-   */
-  com.google.protobuf.ByteString
-      getTfdbgContextIdBytes();
-
-  /**
-   * <pre>
-   * Name of the op (applicable only in the case of the `FULL_TENSOR` trace
-   * level).
-   * </pre>
-   *
-   * <code>string op_name = 2;</code>
-   */
-  java.lang.String getOpName();
-  /**
-   * <pre>
-   * Name of the op (applicable only in the case of the `FULL_TENSOR` trace
-   * level).
-   * </pre>
-   *
-   * <code>string op_name = 2;</code>
-   */
-  com.google.protobuf.ByteString
-      getOpNameBytes();
-
-  /**
-   * <pre>
-   * Output slot of the tensor (applicable only in the case of the `FULL_TENSOR`
-   * trace level).
-   * </pre>
-   *
-   * <code>int32 output_slot = 3;</code>
-   */
-  int getOutputSlot();
-
-  /**
-   * <pre>
-   * Type of the tensor value encapsulated in this proto.
-   * </pre>
-   *
-   * <code>.tensorflow.TensorDebugMode tensor_debug_mode = 4;</code>
-   */
-  int getTensorDebugModeValue();
-  /**
-   * <pre>
-   * Type of the tensor value encapsulated in this proto.
-   * </pre>
-   *
-   * <code>.tensorflow.TensorDebugMode tensor_debug_mode = 4;</code>
-   */
-  org.tensorflow.proto.util.TensorDebugMode getTensorDebugMode();
-
-  /**
-   * <pre>
-   * Tensor value in the type described by `tensor_value_type`.
-   * This tensor may summarize the value of a single intermediate op of the
-   * graph, or those of multiple intermediate tensors.
-   * </pre>
-   *
-   * <code>.tensorflow.TensorProto tensor_proto = 5;</code>
-   */
-  boolean hasTensorProto();
-  /**
-   * <pre>
-   * Tensor value in the type described by `tensor_value_type`.
-   * This tensor may summarize the value of a single intermediate op of the
-   * graph, or those of multiple intermediate tensors.
-   * </pre>
-   *
-   * <code>.tensorflow.TensorProto tensor_proto = 5;</code>
-   */
-  org.tensorflow.proto.framework.TensorProto getTensorProto();
-  /**
-   * <pre>
-   * Tensor value in the type described by `tensor_value_type`.
-   * This tensor may summarize the value of a single intermediate op of the
-   * graph, or those of multiple intermediate tensors.
-   * </pre>
-   *
-   * <code>.tensorflow.TensorProto tensor_proto = 5;</code>
-   */
-  org.tensorflow.proto.framework.TensorProtoOrBuilder getTensorProtoOrBuilder();
-
-  /**
-   * <pre>
-   * Name of the device that the op belongs to.
-   * </pre>
-   *
-   * <code>string device_name = 6;</code>
-   */
-  java.lang.String getDeviceName();
-  /**
-   * <pre>
-   * Name of the device that the op belongs to.
-   * </pre>
-   *
-   * <code>string device_name = 6;</code>
-   */
-  com.google.protobuf.ByteString
-      getDeviceNameBytes();
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/GraphOpCreation.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/GraphOpCreation.java
deleted file mode 100644
index da8a4523ec0..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/GraphOpCreation.java
+++ /dev/null
@@ -1,1936 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/debug_event.proto
-
-package org.tensorflow.proto.util;
-
-/**
- * <pre>
- * The creation of an op in a TensorFlow Graph (e.g., FuncGraph in TF2).
- * </pre>
- *
- * Protobuf type {@code tensorflow.GraphOpCreation}
- */
-public  final class GraphOpCreation extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.GraphOpCreation)
-    GraphOpCreationOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use GraphOpCreation.newBuilder() to construct.
-  private GraphOpCreation(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private GraphOpCreation() {
-    opType_ = "";
-    opName_ = "";
-    graphName_ = "";
-    graphId_ = "";
-    deviceName_ = "";
-    inputNames_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-    outputTensorIds_ = emptyIntList();
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new GraphOpCreation();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private GraphOpCreation(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    int mutable_bitField0_ = 0;
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 10: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            opType_ = s;
-            break;
-          }
-          case 18: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            opName_ = s;
-            break;
-          }
-          case 26: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            graphName_ = s;
-            break;
-          }
-          case 34: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            graphId_ = s;
-            break;
-          }
-          case 42: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            deviceName_ = s;
-            break;
-          }
-          case 50: {
-            java.lang.String s = input.readStringRequireUtf8();
-            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
-              inputNames_ = new com.google.protobuf.LazyStringArrayList();
-              mutable_bitField0_ |= 0x00000001;
-            }
-            inputNames_.add(s);
-            break;
-          }
-          case 56: {
-
-            numOutputs_ = input.readInt32();
-            break;
-          }
-          case 66: {
-            org.tensorflow.proto.util.CodeLocation.Builder subBuilder = null;
-            if (codeLocation_ != null) {
-              subBuilder = codeLocation_.toBuilder();
-            }
-            codeLocation_ = input.readMessage(org.tensorflow.proto.util.CodeLocation.parser(), extensionRegistry);
-            if (subBuilder != null) {
-              subBuilder.mergeFrom(codeLocation_);
-              codeLocation_ = subBuilder.buildPartial();
-            }
-
-            break;
-          }
-          case 72: {
-            if (!((mutable_bitField0_ & 0x00000002) != 0)) {
-              outputTensorIds_ = newIntList();
-              mutable_bitField0_ |= 0x00000002;
-            }
-            outputTensorIds_.addInt(input.readInt32());
-            break;
-          }
-          case 74: {
-            int length = input.readRawVarint32();
-            int limit = input.pushLimit(length);
-            if (!((mutable_bitField0_ & 0x00000002) != 0) && input.getBytesUntilLimit() > 0) {
-              outputTensorIds_ = newIntList();
-              mutable_bitField0_ |= 0x00000002;
-            }
-            while (input.getBytesUntilLimit() > 0) {
-              outputTensorIds_.addInt(input.readInt32());
-            }
-            input.popLimit(limit);
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      if (((mutable_bitField0_ & 0x00000001) != 0)) {
-        inputNames_ = inputNames_.getUnmodifiableView();
-      }
-      if (((mutable_bitField0_ & 0x00000002) != 0)) {
-        outputTensorIds_.makeImmutable(); // C
-      }
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.util.DebugEventProtos.internal_static_tensorflow_GraphOpCreation_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.util.DebugEventProtos.internal_static_tensorflow_GraphOpCreation_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.util.GraphOpCreation.class, org.tensorflow.proto.util.GraphOpCreation.Builder.class);
-  }
-
-  public static final int OP_TYPE_FIELD_NUMBER = 1;
-  private volatile java.lang.Object opType_;
-  /**
-   * <pre>
-   * Type of the op (e.g., "MatMul").
-   * </pre>
-   *
-   * <code>string op_type = 1;</code>
-   */
-  public java.lang.String getOpType() {
-    java.lang.Object ref = opType_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      opType_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * Type of the op (e.g., "MatMul").
-   * </pre>
-   *
-   * <code>string op_type = 1;</code>
-   */
-  public com.google.protobuf.ByteString
-      getOpTypeBytes() {
-    java.lang.Object ref = opType_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      opType_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int OP_NAME_FIELD_NUMBER = 2;
-  private volatile java.lang.Object opName_;
-  /**
-   * <pre>
-   * Name of the op (e.g., "Dense/MatMul_1").
-   * </pre>
-   *
-   * <code>string op_name = 2;</code>
-   */
-  public java.lang.String getOpName() {
-    java.lang.Object ref = opName_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      opName_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * Name of the op (e.g., "Dense/MatMul_1").
-   * </pre>
-   *
-   * <code>string op_name = 2;</code>
-   */
-  public com.google.protobuf.ByteString
-      getOpNameBytes() {
-    java.lang.Object ref = opName_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      opName_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int GRAPH_NAME_FIELD_NUMBER = 3;
-  private volatile java.lang.Object graphName_;
-  /**
-   * <pre>
-   * Name of the graph that the op is a part of (if available).
-   * </pre>
-   *
-   * <code>string graph_name = 3;</code>
-   */
-  public java.lang.String getGraphName() {
-    java.lang.Object ref = graphName_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      graphName_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * Name of the graph that the op is a part of (if available).
-   * </pre>
-   *
-   * <code>string graph_name = 3;</code>
-   */
-  public com.google.protobuf.ByteString
-      getGraphNameBytes() {
-    java.lang.Object ref = graphName_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      graphName_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int GRAPH_ID_FIELD_NUMBER = 4;
-  private volatile java.lang.Object graphId_;
-  /**
-   * <pre>
-   * Unique ID of the graph (generated by debugger).
-   * This is the ID of the immediately-enclosing graph.
-   * </pre>
-   *
-   * <code>string graph_id = 4;</code>
-   */
-  public java.lang.String getGraphId() {
-    java.lang.Object ref = graphId_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      graphId_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * Unique ID of the graph (generated by debugger).
-   * This is the ID of the immediately-enclosing graph.
-   * </pre>
-   *
-   * <code>string graph_id = 4;</code>
-   */
-  public com.google.protobuf.ByteString
-      getGraphIdBytes() {
-    java.lang.Object ref = graphId_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      graphId_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int DEVICE_NAME_FIELD_NUMBER = 5;
-  private volatile java.lang.Object deviceName_;
-  /**
-   * <pre>
-   * Name of the device that the op is assigned to (if available).
-   * </pre>
-   *
-   * <code>string device_name = 5;</code>
-   */
-  public java.lang.String getDeviceName() {
-    java.lang.Object ref = deviceName_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      deviceName_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * Name of the device that the op is assigned to (if available).
-   * </pre>
-   *
-   * <code>string device_name = 5;</code>
-   */
-  public com.google.protobuf.ByteString
-      getDeviceNameBytes() {
-    java.lang.Object ref = deviceName_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      deviceName_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int INPUT_NAMES_FIELD_NUMBER = 6;
-  private com.google.protobuf.LazyStringList inputNames_;
-  /**
-   * <pre>
-   * Names of the input tensors to the op.
-   * </pre>
-   *
-   * <code>repeated string input_names = 6;</code>
-   */
-  public com.google.protobuf.ProtocolStringList
-      getInputNamesList() {
-    return inputNames_;
-  }
-  /**
-   * <pre>
-   * Names of the input tensors to the op.
-   * </pre>
-   *
-   * <code>repeated string input_names = 6;</code>
-   */
-  public int getInputNamesCount() {
-    return inputNames_.size();
-  }
-  /**
-   * <pre>
-   * Names of the input tensors to the op.
-   * </pre>
-   *
-   * <code>repeated string input_names = 6;</code>
-   */
-  public java.lang.String getInputNames(int index) {
-    return inputNames_.get(index);
-  }
-  /**
-   * <pre>
-   * Names of the input tensors to the op.
-   * </pre>
-   *
-   * <code>repeated string input_names = 6;</code>
-   */
-  public com.google.protobuf.ByteString
-      getInputNamesBytes(int index) {
-    return inputNames_.getByteString(index);
-  }
-
-  public static final int NUM_OUTPUTS_FIELD_NUMBER = 7;
-  private int numOutputs_;
-  /**
-   * <pre>
-   * Number of output tensors emitted by the op.
-   * </pre>
-   *
-   * <code>int32 num_outputs = 7;</code>
-   */
-  public int getNumOutputs() {
-    return numOutputs_;
-  }
-
-  public static final int CODE_LOCATION_FIELD_NUMBER = 8;
-  private org.tensorflow.proto.util.CodeLocation codeLocation_;
-  /**
-   * <pre>
-   * The unique ID for code location (stack trace) of the op's creation.
-   * </pre>
-   *
-   * <code>.tensorflow.CodeLocation code_location = 8;</code>
-   */
-  public boolean hasCodeLocation() {
-    return codeLocation_ != null;
-  }
-  /**
-   * <pre>
-   * The unique ID for code location (stack trace) of the op's creation.
-   * </pre>
-   *
-   * <code>.tensorflow.CodeLocation code_location = 8;</code>
-   */
-  public org.tensorflow.proto.util.CodeLocation getCodeLocation() {
-    return codeLocation_ == null ? org.tensorflow.proto.util.CodeLocation.getDefaultInstance() : codeLocation_;
-  }
-  /**
-   * <pre>
-   * The unique ID for code location (stack trace) of the op's creation.
-   * </pre>
-   *
-   * <code>.tensorflow.CodeLocation code_location = 8;</code>
-   */
-  public org.tensorflow.proto.util.CodeLocationOrBuilder getCodeLocationOrBuilder() {
-    return getCodeLocation();
-  }
-
-  public static final int OUTPUT_TENSOR_IDS_FIELD_NUMBER = 9;
-  private com.google.protobuf.Internal.IntList outputTensorIds_;
-  /**
-   * <pre>
-   * Unique IDs for the output tensors of this op.
-   * </pre>
-   *
-   * <code>repeated int32 output_tensor_ids = 9;</code>
-   */
-  public java.util.List<java.lang.Integer>
-      getOutputTensorIdsList() {
-    return outputTensorIds_;
-  }
-  /**
-   * <pre>
-   * Unique IDs for the output tensors of this op.
-   * </pre>
-   *
-   * <code>repeated int32 output_tensor_ids = 9;</code>
-   */
-  public int getOutputTensorIdsCount() {
-    return outputTensorIds_.size();
-  }
-  /**
-   * <pre>
-   * Unique IDs for the output tensors of this op.
-   * </pre>
-   *
-   * <code>repeated int32 output_tensor_ids = 9;</code>
-   */
-  public int getOutputTensorIds(int index) {
-    return outputTensorIds_.getInt(index);
-  }
-  private int outputTensorIdsMemoizedSerializedSize = -1;
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    getSerializedSize();
-    if (!getOpTypeBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 1, opType_);
-    }
-    if (!getOpNameBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 2, opName_);
-    }
-    if (!getGraphNameBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 3, graphName_);
-    }
-    if (!getGraphIdBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 4, graphId_);
-    }
-    if (!getDeviceNameBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 5, deviceName_);
-    }
-    for (int i = 0; i < inputNames_.size(); i++) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 6, inputNames_.getRaw(i));
-    }
-    if (numOutputs_ != 0) {
-      output.writeInt32(7, numOutputs_);
-    }
-    if (codeLocation_ != null) {
-      output.writeMessage(8, getCodeLocation());
-    }
-    if (getOutputTensorIdsList().size() > 0) {
-      output.writeUInt32NoTag(74);
-      output.writeUInt32NoTag(outputTensorIdsMemoizedSerializedSize);
-    }
-    for (int i = 0; i < outputTensorIds_.size(); i++) {
-      output.writeInt32NoTag(outputTensorIds_.getInt(i));
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (!getOpTypeBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, opType_);
-    }
-    if (!getOpNameBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(2, opName_);
-    }
-    if (!getGraphNameBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(3, graphName_);
-    }
-    if (!getGraphIdBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(4, graphId_);
-    }
-    if (!getDeviceNameBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(5, deviceName_);
-    }
-    {
-      int dataSize = 0;
-      for (int i = 0; i < inputNames_.size(); i++) {
-        dataSize += computeStringSizeNoTag(inputNames_.getRaw(i));
-      }
-      size += dataSize;
-      size += 1 * getInputNamesList().size();
-    }
-    if (numOutputs_ != 0) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt32Size(7, numOutputs_);
-    }
-    if (codeLocation_ != null) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(8, getCodeLocation());
-    }
-    {
-      int dataSize = 0;
-      for (int i = 0; i < outputTensorIds_.size(); i++) {
-        dataSize += com.google.protobuf.CodedOutputStream
-          .computeInt32SizeNoTag(outputTensorIds_.getInt(i));
-      }
-      size += dataSize;
-      if (!getOutputTensorIdsList().isEmpty()) {
-        size += 1;
-        size += com.google.protobuf.CodedOutputStream
-            .computeInt32SizeNoTag(dataSize);
-      }
-      outputTensorIdsMemoizedSerializedSize = dataSize;
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.util.GraphOpCreation)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.util.GraphOpCreation other = (org.tensorflow.proto.util.GraphOpCreation) obj;
-
-    if (!getOpType()
-        .equals(other.getOpType())) return false;
-    if (!getOpName()
-        .equals(other.getOpName())) return false;
-    if (!getGraphName()
-        .equals(other.getGraphName())) return false;
-    if (!getGraphId()
-        .equals(other.getGraphId())) return false;
-    if (!getDeviceName()
-        .equals(other.getDeviceName())) return false;
-    if (!getInputNamesList()
-        .equals(other.getInputNamesList())) return false;
-    if (getNumOutputs()
-        != other.getNumOutputs()) return false;
-    if (hasCodeLocation() != other.hasCodeLocation()) return false;
-    if (hasCodeLocation()) {
-      if (!getCodeLocation()
-          .equals(other.getCodeLocation())) return false;
-    }
-    if (!getOutputTensorIdsList()
-        .equals(other.getOutputTensorIdsList())) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    hash = (37 * hash) + OP_TYPE_FIELD_NUMBER;
-    hash = (53 * hash) + getOpType().hashCode();
-    hash = (37 * hash) + OP_NAME_FIELD_NUMBER;
-    hash = (53 * hash) + getOpName().hashCode();
-    hash = (37 * hash) + GRAPH_NAME_FIELD_NUMBER;
-    hash = (53 * hash) + getGraphName().hashCode();
-    hash = (37 * hash) + GRAPH_ID_FIELD_NUMBER;
-    hash = (53 * hash) + getGraphId().hashCode();
-    hash = (37 * hash) + DEVICE_NAME_FIELD_NUMBER;
-    hash = (53 * hash) + getDeviceName().hashCode();
-    if (getInputNamesCount() > 0) {
-      hash = (37 * hash) + INPUT_NAMES_FIELD_NUMBER;
-      hash = (53 * hash) + getInputNamesList().hashCode();
-    }
-    hash = (37 * hash) + NUM_OUTPUTS_FIELD_NUMBER;
-    hash = (53 * hash) + getNumOutputs();
-    if (hasCodeLocation()) {
-      hash = (37 * hash) + CODE_LOCATION_FIELD_NUMBER;
-      hash = (53 * hash) + getCodeLocation().hashCode();
-    }
-    if (getOutputTensorIdsCount() > 0) {
-      hash = (37 * hash) + OUTPUT_TENSOR_IDS_FIELD_NUMBER;
-      hash = (53 * hash) + getOutputTensorIdsList().hashCode();
-    }
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.util.GraphOpCreation parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.GraphOpCreation parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.GraphOpCreation parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.GraphOpCreation parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.GraphOpCreation parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.GraphOpCreation parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.GraphOpCreation parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.GraphOpCreation parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.GraphOpCreation parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.GraphOpCreation parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.GraphOpCreation parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.GraphOpCreation parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.util.GraphOpCreation prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * <pre>
-   * The creation of an op in a TensorFlow Graph (e.g., FuncGraph in TF2).
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.GraphOpCreation}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.GraphOpCreation)
-      org.tensorflow.proto.util.GraphOpCreationOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.util.DebugEventProtos.internal_static_tensorflow_GraphOpCreation_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.util.DebugEventProtos.internal_static_tensorflow_GraphOpCreation_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.util.GraphOpCreation.class, org.tensorflow.proto.util.GraphOpCreation.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.util.GraphOpCreation.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      opType_ = "";
-
-      opName_ = "";
-
-      graphName_ = "";
-
-      graphId_ = "";
-
-      deviceName_ = "";
-
-      inputNames_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-      bitField0_ = (bitField0_ & ~0x00000001);
-      numOutputs_ = 0;
-
-      if (codeLocationBuilder_ == null) {
-        codeLocation_ = null;
-      } else {
-        codeLocation_ = null;
-        codeLocationBuilder_ = null;
-      }
-      outputTensorIds_ = emptyIntList();
-      bitField0_ = (bitField0_ & ~0x00000002);
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.util.DebugEventProtos.internal_static_tensorflow_GraphOpCreation_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.GraphOpCreation getDefaultInstanceForType() {
-      return org.tensorflow.proto.util.GraphOpCreation.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.GraphOpCreation build() {
-      org.tensorflow.proto.util.GraphOpCreation result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.GraphOpCreation buildPartial() {
-      org.tensorflow.proto.util.GraphOpCreation result = new org.tensorflow.proto.util.GraphOpCreation(this);
-      int from_bitField0_ = bitField0_;
-      result.opType_ = opType_;
-      result.opName_ = opName_;
-      result.graphName_ = graphName_;
-      result.graphId_ = graphId_;
-      result.deviceName_ = deviceName_;
-      if (((bitField0_ & 0x00000001) != 0)) {
-        inputNames_ = inputNames_.getUnmodifiableView();
-        bitField0_ = (bitField0_ & ~0x00000001);
-      }
-      result.inputNames_ = inputNames_;
-      result.numOutputs_ = numOutputs_;
-      if (codeLocationBuilder_ == null) {
-        result.codeLocation_ = codeLocation_;
-      } else {
-        result.codeLocation_ = codeLocationBuilder_.build();
-      }
-      if (((bitField0_ & 0x00000002) != 0)) {
-        outputTensorIds_.makeImmutable();
-        bitField0_ = (bitField0_ & ~0x00000002);
-      }
-      result.outputTensorIds_ = outputTensorIds_;
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.util.GraphOpCreation) {
-        return mergeFrom((org.tensorflow.proto.util.GraphOpCreation)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.util.GraphOpCreation other) {
-      if (other == org.tensorflow.proto.util.GraphOpCreation.getDefaultInstance()) return this;
-      if (!other.getOpType().isEmpty()) {
-        opType_ = other.opType_;
-        onChanged();
-      }
-      if (!other.getOpName().isEmpty()) {
-        opName_ = other.opName_;
-        onChanged();
-      }
-      if (!other.getGraphName().isEmpty()) {
-        graphName_ = other.graphName_;
-        onChanged();
-      }
-      if (!other.getGraphId().isEmpty()) {
-        graphId_ = other.graphId_;
-        onChanged();
-      }
-      if (!other.getDeviceName().isEmpty()) {
-        deviceName_ = other.deviceName_;
-        onChanged();
-      }
-      if (!other.inputNames_.isEmpty()) {
-        if (inputNames_.isEmpty()) {
-          inputNames_ = other.inputNames_;
-          bitField0_ = (bitField0_ & ~0x00000001);
-        } else {
-          ensureInputNamesIsMutable();
-          inputNames_.addAll(other.inputNames_);
-        }
-        onChanged();
-      }
-      if (other.getNumOutputs() != 0) {
-        setNumOutputs(other.getNumOutputs());
-      }
-      if (other.hasCodeLocation()) {
-        mergeCodeLocation(other.getCodeLocation());
-      }
-      if (!other.outputTensorIds_.isEmpty()) {
-        if (outputTensorIds_.isEmpty()) {
-          outputTensorIds_ = other.outputTensorIds_;
-          bitField0_ = (bitField0_ & ~0x00000002);
-        } else {
-          ensureOutputTensorIdsIsMutable();
-          outputTensorIds_.addAll(other.outputTensorIds_);
-        }
-        onChanged();
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.util.GraphOpCreation parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.util.GraphOpCreation) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-    private int bitField0_;
-
-    private java.lang.Object opType_ = "";
-    /**
-     * <pre>
-     * Type of the op (e.g., "MatMul").
-     * </pre>
-     *
-     * <code>string op_type = 1;</code>
-     */
-    public java.lang.String getOpType() {
-      java.lang.Object ref = opType_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        opType_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Type of the op (e.g., "MatMul").
-     * </pre>
-     *
-     * <code>string op_type = 1;</code>
-     */
-    public com.google.protobuf.ByteString
-        getOpTypeBytes() {
-      java.lang.Object ref = opType_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        opType_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Type of the op (e.g., "MatMul").
-     * </pre>
-     *
-     * <code>string op_type = 1;</code>
-     */
-    public Builder setOpType(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      opType_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Type of the op (e.g., "MatMul").
-     * </pre>
-     *
-     * <code>string op_type = 1;</code>
-     */
-    public Builder clearOpType() {
-      
-      opType_ = getDefaultInstance().getOpType();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Type of the op (e.g., "MatMul").
-     * </pre>
-     *
-     * <code>string op_type = 1;</code>
-     */
-    public Builder setOpTypeBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      opType_ = value;
-      onChanged();
-      return this;
-    }
-
-    private java.lang.Object opName_ = "";
-    /**
-     * <pre>
-     * Name of the op (e.g., "Dense/MatMul_1").
-     * </pre>
-     *
-     * <code>string op_name = 2;</code>
-     */
-    public java.lang.String getOpName() {
-      java.lang.Object ref = opName_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        opName_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Name of the op (e.g., "Dense/MatMul_1").
-     * </pre>
-     *
-     * <code>string op_name = 2;</code>
-     */
-    public com.google.protobuf.ByteString
-        getOpNameBytes() {
-      java.lang.Object ref = opName_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        opName_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Name of the op (e.g., "Dense/MatMul_1").
-     * </pre>
-     *
-     * <code>string op_name = 2;</code>
-     */
-    public Builder setOpName(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      opName_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Name of the op (e.g., "Dense/MatMul_1").
-     * </pre>
-     *
-     * <code>string op_name = 2;</code>
-     */
-    public Builder clearOpName() {
-      
-      opName_ = getDefaultInstance().getOpName();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Name of the op (e.g., "Dense/MatMul_1").
-     * </pre>
-     *
-     * <code>string op_name = 2;</code>
-     */
-    public Builder setOpNameBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      opName_ = value;
-      onChanged();
-      return this;
-    }
-
-    private java.lang.Object graphName_ = "";
-    /**
-     * <pre>
-     * Name of the graph that the op is a part of (if available).
-     * </pre>
-     *
-     * <code>string graph_name = 3;</code>
-     */
-    public java.lang.String getGraphName() {
-      java.lang.Object ref = graphName_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        graphName_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Name of the graph that the op is a part of (if available).
-     * </pre>
-     *
-     * <code>string graph_name = 3;</code>
-     */
-    public com.google.protobuf.ByteString
-        getGraphNameBytes() {
-      java.lang.Object ref = graphName_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        graphName_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Name of the graph that the op is a part of (if available).
-     * </pre>
-     *
-     * <code>string graph_name = 3;</code>
-     */
-    public Builder setGraphName(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      graphName_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Name of the graph that the op is a part of (if available).
-     * </pre>
-     *
-     * <code>string graph_name = 3;</code>
-     */
-    public Builder clearGraphName() {
-      
-      graphName_ = getDefaultInstance().getGraphName();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Name of the graph that the op is a part of (if available).
-     * </pre>
-     *
-     * <code>string graph_name = 3;</code>
-     */
-    public Builder setGraphNameBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      graphName_ = value;
-      onChanged();
-      return this;
-    }
-
-    private java.lang.Object graphId_ = "";
-    /**
-     * <pre>
-     * Unique ID of the graph (generated by debugger).
-     * This is the ID of the immediately-enclosing graph.
-     * </pre>
-     *
-     * <code>string graph_id = 4;</code>
-     */
-    public java.lang.String getGraphId() {
-      java.lang.Object ref = graphId_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        graphId_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Unique ID of the graph (generated by debugger).
-     * This is the ID of the immediately-enclosing graph.
-     * </pre>
-     *
-     * <code>string graph_id = 4;</code>
-     */
-    public com.google.protobuf.ByteString
-        getGraphIdBytes() {
-      java.lang.Object ref = graphId_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        graphId_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Unique ID of the graph (generated by debugger).
-     * This is the ID of the immediately-enclosing graph.
-     * </pre>
-     *
-     * <code>string graph_id = 4;</code>
-     */
-    public Builder setGraphId(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      graphId_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Unique ID of the graph (generated by debugger).
-     * This is the ID of the immediately-enclosing graph.
-     * </pre>
-     *
-     * <code>string graph_id = 4;</code>
-     */
-    public Builder clearGraphId() {
-      
-      graphId_ = getDefaultInstance().getGraphId();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Unique ID of the graph (generated by debugger).
-     * This is the ID of the immediately-enclosing graph.
-     * </pre>
-     *
-     * <code>string graph_id = 4;</code>
-     */
-    public Builder setGraphIdBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      graphId_ = value;
-      onChanged();
-      return this;
-    }
-
-    private java.lang.Object deviceName_ = "";
-    /**
-     * <pre>
-     * Name of the device that the op is assigned to (if available).
-     * </pre>
-     *
-     * <code>string device_name = 5;</code>
-     */
-    public java.lang.String getDeviceName() {
-      java.lang.Object ref = deviceName_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        deviceName_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Name of the device that the op is assigned to (if available).
-     * </pre>
-     *
-     * <code>string device_name = 5;</code>
-     */
-    public com.google.protobuf.ByteString
-        getDeviceNameBytes() {
-      java.lang.Object ref = deviceName_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        deviceName_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Name of the device that the op is assigned to (if available).
-     * </pre>
-     *
-     * <code>string device_name = 5;</code>
-     */
-    public Builder setDeviceName(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      deviceName_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Name of the device that the op is assigned to (if available).
-     * </pre>
-     *
-     * <code>string device_name = 5;</code>
-     */
-    public Builder clearDeviceName() {
-      
-      deviceName_ = getDefaultInstance().getDeviceName();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Name of the device that the op is assigned to (if available).
-     * </pre>
-     *
-     * <code>string device_name = 5;</code>
-     */
-    public Builder setDeviceNameBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      deviceName_ = value;
-      onChanged();
-      return this;
-    }
-
-    private com.google.protobuf.LazyStringList inputNames_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-    private void ensureInputNamesIsMutable() {
-      if (!((bitField0_ & 0x00000001) != 0)) {
-        inputNames_ = new com.google.protobuf.LazyStringArrayList(inputNames_);
-        bitField0_ |= 0x00000001;
-       }
-    }
-    /**
-     * <pre>
-     * Names of the input tensors to the op.
-     * </pre>
-     *
-     * <code>repeated string input_names = 6;</code>
-     */
-    public com.google.protobuf.ProtocolStringList
-        getInputNamesList() {
-      return inputNames_.getUnmodifiableView();
-    }
-    /**
-     * <pre>
-     * Names of the input tensors to the op.
-     * </pre>
-     *
-     * <code>repeated string input_names = 6;</code>
-     */
-    public int getInputNamesCount() {
-      return inputNames_.size();
-    }
-    /**
-     * <pre>
-     * Names of the input tensors to the op.
-     * </pre>
-     *
-     * <code>repeated string input_names = 6;</code>
-     */
-    public java.lang.String getInputNames(int index) {
-      return inputNames_.get(index);
-    }
-    /**
-     * <pre>
-     * Names of the input tensors to the op.
-     * </pre>
-     *
-     * <code>repeated string input_names = 6;</code>
-     */
-    public com.google.protobuf.ByteString
-        getInputNamesBytes(int index) {
-      return inputNames_.getByteString(index);
-    }
-    /**
-     * <pre>
-     * Names of the input tensors to the op.
-     * </pre>
-     *
-     * <code>repeated string input_names = 6;</code>
-     */
-    public Builder setInputNames(
-        int index, java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  ensureInputNamesIsMutable();
-      inputNames_.set(index, value);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Names of the input tensors to the op.
-     * </pre>
-     *
-     * <code>repeated string input_names = 6;</code>
-     */
-    public Builder addInputNames(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  ensureInputNamesIsMutable();
-      inputNames_.add(value);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Names of the input tensors to the op.
-     * </pre>
-     *
-     * <code>repeated string input_names = 6;</code>
-     */
-    public Builder addAllInputNames(
-        java.lang.Iterable<java.lang.String> values) {
-      ensureInputNamesIsMutable();
-      com.google.protobuf.AbstractMessageLite.Builder.addAll(
-          values, inputNames_);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Names of the input tensors to the op.
-     * </pre>
-     *
-     * <code>repeated string input_names = 6;</code>
-     */
-    public Builder clearInputNames() {
-      inputNames_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-      bitField0_ = (bitField0_ & ~0x00000001);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Names of the input tensors to the op.
-     * </pre>
-     *
-     * <code>repeated string input_names = 6;</code>
-     */
-    public Builder addInputNamesBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      ensureInputNamesIsMutable();
-      inputNames_.add(value);
-      onChanged();
-      return this;
-    }
-
-    private int numOutputs_ ;
-    /**
-     * <pre>
-     * Number of output tensors emitted by the op.
-     * </pre>
-     *
-     * <code>int32 num_outputs = 7;</code>
-     */
-    public int getNumOutputs() {
-      return numOutputs_;
-    }
-    /**
-     * <pre>
-     * Number of output tensors emitted by the op.
-     * </pre>
-     *
-     * <code>int32 num_outputs = 7;</code>
-     */
-    public Builder setNumOutputs(int value) {
-      
-      numOutputs_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Number of output tensors emitted by the op.
-     * </pre>
-     *
-     * <code>int32 num_outputs = 7;</code>
-     */
-    public Builder clearNumOutputs() {
-      
-      numOutputs_ = 0;
-      onChanged();
-      return this;
-    }
-
-    private org.tensorflow.proto.util.CodeLocation codeLocation_;
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.util.CodeLocation, org.tensorflow.proto.util.CodeLocation.Builder, org.tensorflow.proto.util.CodeLocationOrBuilder> codeLocationBuilder_;
-    /**
-     * <pre>
-     * The unique ID for code location (stack trace) of the op's creation.
-     * </pre>
-     *
-     * <code>.tensorflow.CodeLocation code_location = 8;</code>
-     */
-    public boolean hasCodeLocation() {
-      return codeLocationBuilder_ != null || codeLocation_ != null;
-    }
-    /**
-     * <pre>
-     * The unique ID for code location (stack trace) of the op's creation.
-     * </pre>
-     *
-     * <code>.tensorflow.CodeLocation code_location = 8;</code>
-     */
-    public org.tensorflow.proto.util.CodeLocation getCodeLocation() {
-      if (codeLocationBuilder_ == null) {
-        return codeLocation_ == null ? org.tensorflow.proto.util.CodeLocation.getDefaultInstance() : codeLocation_;
-      } else {
-        return codeLocationBuilder_.getMessage();
-      }
-    }
-    /**
-     * <pre>
-     * The unique ID for code location (stack trace) of the op's creation.
-     * </pre>
-     *
-     * <code>.tensorflow.CodeLocation code_location = 8;</code>
-     */
-    public Builder setCodeLocation(org.tensorflow.proto.util.CodeLocation value) {
-      if (codeLocationBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        codeLocation_ = value;
-        onChanged();
-      } else {
-        codeLocationBuilder_.setMessage(value);
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * The unique ID for code location (stack trace) of the op's creation.
-     * </pre>
-     *
-     * <code>.tensorflow.CodeLocation code_location = 8;</code>
-     */
-    public Builder setCodeLocation(
-        org.tensorflow.proto.util.CodeLocation.Builder builderForValue) {
-      if (codeLocationBuilder_ == null) {
-        codeLocation_ = builderForValue.build();
-        onChanged();
-      } else {
-        codeLocationBuilder_.setMessage(builderForValue.build());
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * The unique ID for code location (stack trace) of the op's creation.
-     * </pre>
-     *
-     * <code>.tensorflow.CodeLocation code_location = 8;</code>
-     */
-    public Builder mergeCodeLocation(org.tensorflow.proto.util.CodeLocation value) {
-      if (codeLocationBuilder_ == null) {
-        if (codeLocation_ != null) {
-          codeLocation_ =
-            org.tensorflow.proto.util.CodeLocation.newBuilder(codeLocation_).mergeFrom(value).buildPartial();
-        } else {
-          codeLocation_ = value;
-        }
-        onChanged();
-      } else {
-        codeLocationBuilder_.mergeFrom(value);
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * The unique ID for code location (stack trace) of the op's creation.
-     * </pre>
-     *
-     * <code>.tensorflow.CodeLocation code_location = 8;</code>
-     */
-    public Builder clearCodeLocation() {
-      if (codeLocationBuilder_ == null) {
-        codeLocation_ = null;
-        onChanged();
-      } else {
-        codeLocation_ = null;
-        codeLocationBuilder_ = null;
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * The unique ID for code location (stack trace) of the op's creation.
-     * </pre>
-     *
-     * <code>.tensorflow.CodeLocation code_location = 8;</code>
-     */
-    public org.tensorflow.proto.util.CodeLocation.Builder getCodeLocationBuilder() {
-      
-      onChanged();
-      return getCodeLocationFieldBuilder().getBuilder();
-    }
-    /**
-     * <pre>
-     * The unique ID for code location (stack trace) of the op's creation.
-     * </pre>
-     *
-     * <code>.tensorflow.CodeLocation code_location = 8;</code>
-     */
-    public org.tensorflow.proto.util.CodeLocationOrBuilder getCodeLocationOrBuilder() {
-      if (codeLocationBuilder_ != null) {
-        return codeLocationBuilder_.getMessageOrBuilder();
-      } else {
-        return codeLocation_ == null ?
-            org.tensorflow.proto.util.CodeLocation.getDefaultInstance() : codeLocation_;
-      }
-    }
-    /**
-     * <pre>
-     * The unique ID for code location (stack trace) of the op's creation.
-     * </pre>
-     *
-     * <code>.tensorflow.CodeLocation code_location = 8;</code>
-     */
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.util.CodeLocation, org.tensorflow.proto.util.CodeLocation.Builder, org.tensorflow.proto.util.CodeLocationOrBuilder> 
-        getCodeLocationFieldBuilder() {
-      if (codeLocationBuilder_ == null) {
-        codeLocationBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-            org.tensorflow.proto.util.CodeLocation, org.tensorflow.proto.util.CodeLocation.Builder, org.tensorflow.proto.util.CodeLocationOrBuilder>(
-                getCodeLocation(),
-                getParentForChildren(),
-                isClean());
-        codeLocation_ = null;
-      }
-      return codeLocationBuilder_;
-    }
-
-    private com.google.protobuf.Internal.IntList outputTensorIds_ = emptyIntList();
-    private void ensureOutputTensorIdsIsMutable() {
-      if (!((bitField0_ & 0x00000002) != 0)) {
-        outputTensorIds_ = mutableCopy(outputTensorIds_);
-        bitField0_ |= 0x00000002;
-       }
-    }
-    /**
-     * <pre>
-     * Unique IDs for the output tensors of this op.
-     * </pre>
-     *
-     * <code>repeated int32 output_tensor_ids = 9;</code>
-     */
-    public java.util.List<java.lang.Integer>
-        getOutputTensorIdsList() {
-      return ((bitField0_ & 0x00000002) != 0) ?
-               java.util.Collections.unmodifiableList(outputTensorIds_) : outputTensorIds_;
-    }
-    /**
-     * <pre>
-     * Unique IDs for the output tensors of this op.
-     * </pre>
-     *
-     * <code>repeated int32 output_tensor_ids = 9;</code>
-     */
-    public int getOutputTensorIdsCount() {
-      return outputTensorIds_.size();
-    }
-    /**
-     * <pre>
-     * Unique IDs for the output tensors of this op.
-     * </pre>
-     *
-     * <code>repeated int32 output_tensor_ids = 9;</code>
-     */
-    public int getOutputTensorIds(int index) {
-      return outputTensorIds_.getInt(index);
-    }
-    /**
-     * <pre>
-     * Unique IDs for the output tensors of this op.
-     * </pre>
-     *
-     * <code>repeated int32 output_tensor_ids = 9;</code>
-     */
-    public Builder setOutputTensorIds(
-        int index, int value) {
-      ensureOutputTensorIdsIsMutable();
-      outputTensorIds_.setInt(index, value);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Unique IDs for the output tensors of this op.
-     * </pre>
-     *
-     * <code>repeated int32 output_tensor_ids = 9;</code>
-     */
-    public Builder addOutputTensorIds(int value) {
-      ensureOutputTensorIdsIsMutable();
-      outputTensorIds_.addInt(value);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Unique IDs for the output tensors of this op.
-     * </pre>
-     *
-     * <code>repeated int32 output_tensor_ids = 9;</code>
-     */
-    public Builder addAllOutputTensorIds(
-        java.lang.Iterable<? extends java.lang.Integer> values) {
-      ensureOutputTensorIdsIsMutable();
-      com.google.protobuf.AbstractMessageLite.Builder.addAll(
-          values, outputTensorIds_);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Unique IDs for the output tensors of this op.
-     * </pre>
-     *
-     * <code>repeated int32 output_tensor_ids = 9;</code>
-     */
-    public Builder clearOutputTensorIds() {
-      outputTensorIds_ = emptyIntList();
-      bitField0_ = (bitField0_ & ~0x00000002);
-      onChanged();
-      return this;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.GraphOpCreation)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.GraphOpCreation)
-  private static final org.tensorflow.proto.util.GraphOpCreation DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.util.GraphOpCreation();
-  }
-
-  public static org.tensorflow.proto.util.GraphOpCreation getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<GraphOpCreation>
-      PARSER = new com.google.protobuf.AbstractParser<GraphOpCreation>() {
-    @java.lang.Override
-    public GraphOpCreation parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new GraphOpCreation(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<GraphOpCreation> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<GraphOpCreation> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.util.GraphOpCreation getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/GraphOpCreationOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/GraphOpCreationOrBuilder.java
deleted file mode 100644
index 60abaf9cdfe..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/GraphOpCreationOrBuilder.java
+++ /dev/null
@@ -1,195 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/debug_event.proto
-
-package org.tensorflow.proto.util;
-
-public interface GraphOpCreationOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.GraphOpCreation)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <pre>
-   * Type of the op (e.g., "MatMul").
-   * </pre>
-   *
-   * <code>string op_type = 1;</code>
-   */
-  java.lang.String getOpType();
-  /**
-   * <pre>
-   * Type of the op (e.g., "MatMul").
-   * </pre>
-   *
-   * <code>string op_type = 1;</code>
-   */
-  com.google.protobuf.ByteString
-      getOpTypeBytes();
-
-  /**
-   * <pre>
-   * Name of the op (e.g., "Dense/MatMul_1").
-   * </pre>
-   *
-   * <code>string op_name = 2;</code>
-   */
-  java.lang.String getOpName();
-  /**
-   * <pre>
-   * Name of the op (e.g., "Dense/MatMul_1").
-   * </pre>
-   *
-   * <code>string op_name = 2;</code>
-   */
-  com.google.protobuf.ByteString
-      getOpNameBytes();
-
-  /**
-   * <pre>
-   * Name of the graph that the op is a part of (if available).
-   * </pre>
-   *
-   * <code>string graph_name = 3;</code>
-   */
-  java.lang.String getGraphName();
-  /**
-   * <pre>
-   * Name of the graph that the op is a part of (if available).
-   * </pre>
-   *
-   * <code>string graph_name = 3;</code>
-   */
-  com.google.protobuf.ByteString
-      getGraphNameBytes();
-
-  /**
-   * <pre>
-   * Unique ID of the graph (generated by debugger).
-   * This is the ID of the immediately-enclosing graph.
-   * </pre>
-   *
-   * <code>string graph_id = 4;</code>
-   */
-  java.lang.String getGraphId();
-  /**
-   * <pre>
-   * Unique ID of the graph (generated by debugger).
-   * This is the ID of the immediately-enclosing graph.
-   * </pre>
-   *
-   * <code>string graph_id = 4;</code>
-   */
-  com.google.protobuf.ByteString
-      getGraphIdBytes();
-
-  /**
-   * <pre>
-   * Name of the device that the op is assigned to (if available).
-   * </pre>
-   *
-   * <code>string device_name = 5;</code>
-   */
-  java.lang.String getDeviceName();
-  /**
-   * <pre>
-   * Name of the device that the op is assigned to (if available).
-   * </pre>
-   *
-   * <code>string device_name = 5;</code>
-   */
-  com.google.protobuf.ByteString
-      getDeviceNameBytes();
-
-  /**
-   * <pre>
-   * Names of the input tensors to the op.
-   * </pre>
-   *
-   * <code>repeated string input_names = 6;</code>
-   */
-  java.util.List<java.lang.String>
-      getInputNamesList();
-  /**
-   * <pre>
-   * Names of the input tensors to the op.
-   * </pre>
-   *
-   * <code>repeated string input_names = 6;</code>
-   */
-  int getInputNamesCount();
-  /**
-   * <pre>
-   * Names of the input tensors to the op.
-   * </pre>
-   *
-   * <code>repeated string input_names = 6;</code>
-   */
-  java.lang.String getInputNames(int index);
-  /**
-   * <pre>
-   * Names of the input tensors to the op.
-   * </pre>
-   *
-   * <code>repeated string input_names = 6;</code>
-   */
-  com.google.protobuf.ByteString
-      getInputNamesBytes(int index);
-
-  /**
-   * <pre>
-   * Number of output tensors emitted by the op.
-   * </pre>
-   *
-   * <code>int32 num_outputs = 7;</code>
-   */
-  int getNumOutputs();
-
-  /**
-   * <pre>
-   * The unique ID for code location (stack trace) of the op's creation.
-   * </pre>
-   *
-   * <code>.tensorflow.CodeLocation code_location = 8;</code>
-   */
-  boolean hasCodeLocation();
-  /**
-   * <pre>
-   * The unique ID for code location (stack trace) of the op's creation.
-   * </pre>
-   *
-   * <code>.tensorflow.CodeLocation code_location = 8;</code>
-   */
-  org.tensorflow.proto.util.CodeLocation getCodeLocation();
-  /**
-   * <pre>
-   * The unique ID for code location (stack trace) of the op's creation.
-   * </pre>
-   *
-   * <code>.tensorflow.CodeLocation code_location = 8;</code>
-   */
-  org.tensorflow.proto.util.CodeLocationOrBuilder getCodeLocationOrBuilder();
-
-  /**
-   * <pre>
-   * Unique IDs for the output tensors of this op.
-   * </pre>
-   *
-   * <code>repeated int32 output_tensor_ids = 9;</code>
-   */
-  java.util.List<java.lang.Integer> getOutputTensorIdsList();
-  /**
-   * <pre>
-   * Unique IDs for the output tensors of this op.
-   * </pre>
-   *
-   * <code>repeated int32 output_tensor_ids = 9;</code>
-   */
-  int getOutputTensorIdsCount();
-  /**
-   * <pre>
-   * Unique IDs for the output tensors of this op.
-   * </pre>
-   *
-   * <code>repeated int32 output_tensor_ids = 9;</code>
-   */
-  int getOutputTensorIds(int index);
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/LogMessage.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/LogMessage.java
deleted file mode 100644
index 3a7dd55b4c8..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/LogMessage.java
+++ /dev/null
@@ -1,787 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/util/event.proto
-
-package org.tensorflow.proto.util;
-
-/**
- * <pre>
- * Protocol buffer used for logging messages to the events file.
- * </pre>
- *
- * Protobuf type {@code tensorflow.LogMessage}
- */
-public  final class LogMessage extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.LogMessage)
-    LogMessageOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use LogMessage.newBuilder() to construct.
-  private LogMessage(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private LogMessage() {
-    level_ = 0;
-    message_ = "";
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new LogMessage();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private LogMessage(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 8: {
-            int rawValue = input.readEnum();
-
-            level_ = rawValue;
-            break;
-          }
-          case 18: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            message_ = s;
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.util.EventProtos.internal_static_tensorflow_LogMessage_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.util.EventProtos.internal_static_tensorflow_LogMessage_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.util.LogMessage.class, org.tensorflow.proto.util.LogMessage.Builder.class);
-  }
-
-  /**
-   * Protobuf enum {@code tensorflow.LogMessage.Level}
-   */
-  public enum Level
-      implements com.google.protobuf.ProtocolMessageEnum {
-    /**
-     * <code>UNKNOWN = 0;</code>
-     */
-    UNKNOWN(0),
-    /**
-     * <pre>
-     * Note: The logging level 10 cannot be named DEBUG. Some software
-     * projects compile their C/C++ code with -DDEBUG in debug builds. So the
-     * C++ code generated from this file should not have an identifier named
-     * DEBUG.
-     * </pre>
-     *
-     * <code>DEBUGGING = 10;</code>
-     */
-    DEBUGGING(10),
-    /**
-     * <code>INFO = 20;</code>
-     */
-    INFO(20),
-    /**
-     * <code>WARN = 30;</code>
-     */
-    WARN(30),
-    /**
-     * <code>ERROR = 40;</code>
-     */
-    ERROR(40),
-    /**
-     * <code>FATAL = 50;</code>
-     */
-    FATAL(50),
-    UNRECOGNIZED(-1),
-    ;
-
-    /**
-     * <code>UNKNOWN = 0;</code>
-     */
-    public static final int UNKNOWN_VALUE = 0;
-    /**
-     * <pre>
-     * Note: The logging level 10 cannot be named DEBUG. Some software
-     * projects compile their C/C++ code with -DDEBUG in debug builds. So the
-     * C++ code generated from this file should not have an identifier named
-     * DEBUG.
-     * </pre>
-     *
-     * <code>DEBUGGING = 10;</code>
-     */
-    public static final int DEBUGGING_VALUE = 10;
-    /**
-     * <code>INFO = 20;</code>
-     */
-    public static final int INFO_VALUE = 20;
-    /**
-     * <code>WARN = 30;</code>
-     */
-    public static final int WARN_VALUE = 30;
-    /**
-     * <code>ERROR = 40;</code>
-     */
-    public static final int ERROR_VALUE = 40;
-    /**
-     * <code>FATAL = 50;</code>
-     */
-    public static final int FATAL_VALUE = 50;
-
-
-    public final int getNumber() {
-      if (this == UNRECOGNIZED) {
-        throw new java.lang.IllegalArgumentException(
-            "Can't get the number of an unknown enum value.");
-      }
-      return value;
-    }
-
-    /**
-     * @deprecated Use {@link #forNumber(int)} instead.
-     */
-    @java.lang.Deprecated
-    public static Level valueOf(int value) {
-      return forNumber(value);
-    }
-
-    public static Level forNumber(int value) {
-      switch (value) {
-        case 0: return UNKNOWN;
-        case 10: return DEBUGGING;
-        case 20: return INFO;
-        case 30: return WARN;
-        case 40: return ERROR;
-        case 50: return FATAL;
-        default: return null;
-      }
-    }
-
-    public static com.google.protobuf.Internal.EnumLiteMap<Level>
-        internalGetValueMap() {
-      return internalValueMap;
-    }
-    private static final com.google.protobuf.Internal.EnumLiteMap<
-        Level> internalValueMap =
-          new com.google.protobuf.Internal.EnumLiteMap<Level>() {
-            public Level findValueByNumber(int number) {
-              return Level.forNumber(number);
-            }
-          };
-
-    public final com.google.protobuf.Descriptors.EnumValueDescriptor
-        getValueDescriptor() {
-      return getDescriptor().getValues().get(ordinal());
-    }
-    public final com.google.protobuf.Descriptors.EnumDescriptor
-        getDescriptorForType() {
-      return getDescriptor();
-    }
-    public static final com.google.protobuf.Descriptors.EnumDescriptor
-        getDescriptor() {
-      return org.tensorflow.proto.util.LogMessage.getDescriptor().getEnumTypes().get(0);
-    }
-
-    private static final Level[] VALUES = values();
-
-    public static Level valueOf(
-        com.google.protobuf.Descriptors.EnumValueDescriptor desc) {
-      if (desc.getType() != getDescriptor()) {
-        throw new java.lang.IllegalArgumentException(
-          "EnumValueDescriptor is not for this type.");
-      }
-      if (desc.getIndex() == -1) {
-        return UNRECOGNIZED;
-      }
-      return VALUES[desc.getIndex()];
-    }
-
-    private final int value;
-
-    private Level(int value) {
-      this.value = value;
-    }
-
-    // @@protoc_insertion_point(enum_scope:tensorflow.LogMessage.Level)
-  }
-
-  public static final int LEVEL_FIELD_NUMBER = 1;
-  private int level_;
-  /**
-   * <code>.tensorflow.LogMessage.Level level = 1;</code>
-   */
-  public int getLevelValue() {
-    return level_;
-  }
-  /**
-   * <code>.tensorflow.LogMessage.Level level = 1;</code>
-   */
-  public org.tensorflow.proto.util.LogMessage.Level getLevel() {
-    @SuppressWarnings("deprecation")
-    org.tensorflow.proto.util.LogMessage.Level result = org.tensorflow.proto.util.LogMessage.Level.valueOf(level_);
-    return result == null ? org.tensorflow.proto.util.LogMessage.Level.UNRECOGNIZED : result;
-  }
-
-  public static final int MESSAGE_FIELD_NUMBER = 2;
-  private volatile java.lang.Object message_;
-  /**
-   * <code>string message = 2;</code>
-   */
-  public java.lang.String getMessage() {
-    java.lang.Object ref = message_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      message_ = s;
-      return s;
-    }
-  }
-  /**
-   * <code>string message = 2;</code>
-   */
-  public com.google.protobuf.ByteString
-      getMessageBytes() {
-    java.lang.Object ref = message_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      message_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    if (level_ != org.tensorflow.proto.util.LogMessage.Level.UNKNOWN.getNumber()) {
-      output.writeEnum(1, level_);
-    }
-    if (!getMessageBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 2, message_);
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (level_ != org.tensorflow.proto.util.LogMessage.Level.UNKNOWN.getNumber()) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeEnumSize(1, level_);
-    }
-    if (!getMessageBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(2, message_);
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.util.LogMessage)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.util.LogMessage other = (org.tensorflow.proto.util.LogMessage) obj;
-
-    if (level_ != other.level_) return false;
-    if (!getMessage()
-        .equals(other.getMessage())) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    hash = (37 * hash) + LEVEL_FIELD_NUMBER;
-    hash = (53 * hash) + level_;
-    hash = (37 * hash) + MESSAGE_FIELD_NUMBER;
-    hash = (53 * hash) + getMessage().hashCode();
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.util.LogMessage parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.LogMessage parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.LogMessage parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.LogMessage parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.LogMessage parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.LogMessage parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.LogMessage parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.LogMessage parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.LogMessage parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.LogMessage parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.LogMessage parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.LogMessage parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.util.LogMessage prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * <pre>
-   * Protocol buffer used for logging messages to the events file.
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.LogMessage}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.LogMessage)
-      org.tensorflow.proto.util.LogMessageOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.util.EventProtos.internal_static_tensorflow_LogMessage_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.util.EventProtos.internal_static_tensorflow_LogMessage_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.util.LogMessage.class, org.tensorflow.proto.util.LogMessage.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.util.LogMessage.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      level_ = 0;
-
-      message_ = "";
-
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.util.EventProtos.internal_static_tensorflow_LogMessage_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.LogMessage getDefaultInstanceForType() {
-      return org.tensorflow.proto.util.LogMessage.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.LogMessage build() {
-      org.tensorflow.proto.util.LogMessage result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.LogMessage buildPartial() {
-      org.tensorflow.proto.util.LogMessage result = new org.tensorflow.proto.util.LogMessage(this);
-      result.level_ = level_;
-      result.message_ = message_;
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.util.LogMessage) {
-        return mergeFrom((org.tensorflow.proto.util.LogMessage)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.util.LogMessage other) {
-      if (other == org.tensorflow.proto.util.LogMessage.getDefaultInstance()) return this;
-      if (other.level_ != 0) {
-        setLevelValue(other.getLevelValue());
-      }
-      if (!other.getMessage().isEmpty()) {
-        message_ = other.message_;
-        onChanged();
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.util.LogMessage parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.util.LogMessage) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-
-    private int level_ = 0;
-    /**
-     * <code>.tensorflow.LogMessage.Level level = 1;</code>
-     */
-    public int getLevelValue() {
-      return level_;
-    }
-    /**
-     * <code>.tensorflow.LogMessage.Level level = 1;</code>
-     */
-    public Builder setLevelValue(int value) {
-      level_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>.tensorflow.LogMessage.Level level = 1;</code>
-     */
-    public org.tensorflow.proto.util.LogMessage.Level getLevel() {
-      @SuppressWarnings("deprecation")
-      org.tensorflow.proto.util.LogMessage.Level result = org.tensorflow.proto.util.LogMessage.Level.valueOf(level_);
-      return result == null ? org.tensorflow.proto.util.LogMessage.Level.UNRECOGNIZED : result;
-    }
-    /**
-     * <code>.tensorflow.LogMessage.Level level = 1;</code>
-     */
-    public Builder setLevel(org.tensorflow.proto.util.LogMessage.Level value) {
-      if (value == null) {
-        throw new NullPointerException();
-      }
-      
-      level_ = value.getNumber();
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>.tensorflow.LogMessage.Level level = 1;</code>
-     */
-    public Builder clearLevel() {
-      
-      level_ = 0;
-      onChanged();
-      return this;
-    }
-
-    private java.lang.Object message_ = "";
-    /**
-     * <code>string message = 2;</code>
-     */
-    public java.lang.String getMessage() {
-      java.lang.Object ref = message_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        message_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <code>string message = 2;</code>
-     */
-    public com.google.protobuf.ByteString
-        getMessageBytes() {
-      java.lang.Object ref = message_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        message_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <code>string message = 2;</code>
-     */
-    public Builder setMessage(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      message_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>string message = 2;</code>
-     */
-    public Builder clearMessage() {
-      
-      message_ = getDefaultInstance().getMessage();
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>string message = 2;</code>
-     */
-    public Builder setMessageBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      message_ = value;
-      onChanged();
-      return this;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.LogMessage)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.LogMessage)
-  private static final org.tensorflow.proto.util.LogMessage DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.util.LogMessage();
-  }
-
-  public static org.tensorflow.proto.util.LogMessage getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<LogMessage>
-      PARSER = new com.google.protobuf.AbstractParser<LogMessage>() {
-    @java.lang.Override
-    public LogMessage parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new LogMessage(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<LogMessage> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<LogMessage> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.util.LogMessage getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/LogMessageOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/LogMessageOrBuilder.java
deleted file mode 100644
index a19c21eebdf..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/LogMessageOrBuilder.java
+++ /dev/null
@@ -1,28 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/util/event.proto
-
-package org.tensorflow.proto.util;
-
-public interface LogMessageOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.LogMessage)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <code>.tensorflow.LogMessage.Level level = 1;</code>
-   */
-  int getLevelValue();
-  /**
-   * <code>.tensorflow.LogMessage.Level level = 1;</code>
-   */
-  org.tensorflow.proto.util.LogMessage.Level getLevel();
-
-  /**
-   * <code>string message = 2;</code>
-   */
-  java.lang.String getMessage();
-  /**
-   * <code>string message = 2;</code>
-   */
-  com.google.protobuf.ByteString
-      getMessageBytes();
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/MemAllocatorStats.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/MemAllocatorStats.java
deleted file mode 100644
index 78a84b41b9c..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/MemAllocatorStats.java
+++ /dev/null
@@ -1,718 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/bfc_memory_map.proto
-
-package org.tensorflow.proto.util;
-
-/**
- * <pre>
- * Some of the data from AllocatorStats
- * </pre>
- *
- * Protobuf type {@code tensorflow.MemAllocatorStats}
- */
-public  final class MemAllocatorStats extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.MemAllocatorStats)
-    MemAllocatorStatsOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use MemAllocatorStats.newBuilder() to construct.
-  private MemAllocatorStats(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private MemAllocatorStats() {
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new MemAllocatorStats();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private MemAllocatorStats(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 8: {
-
-            numAllocs_ = input.readInt64();
-            break;
-          }
-          case 16: {
-
-            bytesInUse_ = input.readInt64();
-            break;
-          }
-          case 24: {
-
-            peakBytesInUse_ = input.readInt64();
-            break;
-          }
-          case 32: {
-
-            largestAllocSize_ = input.readInt64();
-            break;
-          }
-          case 45: {
-
-            fragmentationMetric_ = input.readFloat();
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.util.BfcMemoryMapProtos.internal_static_tensorflow_MemAllocatorStats_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.util.BfcMemoryMapProtos.internal_static_tensorflow_MemAllocatorStats_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.util.MemAllocatorStats.class, org.tensorflow.proto.util.MemAllocatorStats.Builder.class);
-  }
-
-  public static final int NUM_ALLOCS_FIELD_NUMBER = 1;
-  private long numAllocs_;
-  /**
-   * <code>int64 num_allocs = 1;</code>
-   */
-  public long getNumAllocs() {
-    return numAllocs_;
-  }
-
-  public static final int BYTES_IN_USE_FIELD_NUMBER = 2;
-  private long bytesInUse_;
-  /**
-   * <code>int64 bytes_in_use = 2;</code>
-   */
-  public long getBytesInUse() {
-    return bytesInUse_;
-  }
-
-  public static final int PEAK_BYTES_IN_USE_FIELD_NUMBER = 3;
-  private long peakBytesInUse_;
-  /**
-   * <code>int64 peak_bytes_in_use = 3;</code>
-   */
-  public long getPeakBytesInUse() {
-    return peakBytesInUse_;
-  }
-
-  public static final int LARGEST_ALLOC_SIZE_FIELD_NUMBER = 4;
-  private long largestAllocSize_;
-  /**
-   * <code>int64 largest_alloc_size = 4;</code>
-   */
-  public long getLargestAllocSize() {
-    return largestAllocSize_;
-  }
-
-  public static final int FRAGMENTATION_METRIC_FIELD_NUMBER = 5;
-  private float fragmentationMetric_;
-  /**
-   * <code>float fragmentation_metric = 5;</code>
-   */
-  public float getFragmentationMetric() {
-    return fragmentationMetric_;
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    if (numAllocs_ != 0L) {
-      output.writeInt64(1, numAllocs_);
-    }
-    if (bytesInUse_ != 0L) {
-      output.writeInt64(2, bytesInUse_);
-    }
-    if (peakBytesInUse_ != 0L) {
-      output.writeInt64(3, peakBytesInUse_);
-    }
-    if (largestAllocSize_ != 0L) {
-      output.writeInt64(4, largestAllocSize_);
-    }
-    if (fragmentationMetric_ != 0F) {
-      output.writeFloat(5, fragmentationMetric_);
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (numAllocs_ != 0L) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt64Size(1, numAllocs_);
-    }
-    if (bytesInUse_ != 0L) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt64Size(2, bytesInUse_);
-    }
-    if (peakBytesInUse_ != 0L) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt64Size(3, peakBytesInUse_);
-    }
-    if (largestAllocSize_ != 0L) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt64Size(4, largestAllocSize_);
-    }
-    if (fragmentationMetric_ != 0F) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeFloatSize(5, fragmentationMetric_);
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.util.MemAllocatorStats)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.util.MemAllocatorStats other = (org.tensorflow.proto.util.MemAllocatorStats) obj;
-
-    if (getNumAllocs()
-        != other.getNumAllocs()) return false;
-    if (getBytesInUse()
-        != other.getBytesInUse()) return false;
-    if (getPeakBytesInUse()
-        != other.getPeakBytesInUse()) return false;
-    if (getLargestAllocSize()
-        != other.getLargestAllocSize()) return false;
-    if (java.lang.Float.floatToIntBits(getFragmentationMetric())
-        != java.lang.Float.floatToIntBits(
-            other.getFragmentationMetric())) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    hash = (37 * hash) + NUM_ALLOCS_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-        getNumAllocs());
-    hash = (37 * hash) + BYTES_IN_USE_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-        getBytesInUse());
-    hash = (37 * hash) + PEAK_BYTES_IN_USE_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-        getPeakBytesInUse());
-    hash = (37 * hash) + LARGEST_ALLOC_SIZE_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-        getLargestAllocSize());
-    hash = (37 * hash) + FRAGMENTATION_METRIC_FIELD_NUMBER;
-    hash = (53 * hash) + java.lang.Float.floatToIntBits(
-        getFragmentationMetric());
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.util.MemAllocatorStats parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.MemAllocatorStats parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.MemAllocatorStats parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.MemAllocatorStats parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.MemAllocatorStats parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.MemAllocatorStats parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.MemAllocatorStats parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.MemAllocatorStats parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.MemAllocatorStats parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.MemAllocatorStats parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.MemAllocatorStats parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.MemAllocatorStats parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.util.MemAllocatorStats prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * <pre>
-   * Some of the data from AllocatorStats
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.MemAllocatorStats}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.MemAllocatorStats)
-      org.tensorflow.proto.util.MemAllocatorStatsOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.util.BfcMemoryMapProtos.internal_static_tensorflow_MemAllocatorStats_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.util.BfcMemoryMapProtos.internal_static_tensorflow_MemAllocatorStats_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.util.MemAllocatorStats.class, org.tensorflow.proto.util.MemAllocatorStats.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.util.MemAllocatorStats.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      numAllocs_ = 0L;
-
-      bytesInUse_ = 0L;
-
-      peakBytesInUse_ = 0L;
-
-      largestAllocSize_ = 0L;
-
-      fragmentationMetric_ = 0F;
-
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.util.BfcMemoryMapProtos.internal_static_tensorflow_MemAllocatorStats_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.MemAllocatorStats getDefaultInstanceForType() {
-      return org.tensorflow.proto.util.MemAllocatorStats.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.MemAllocatorStats build() {
-      org.tensorflow.proto.util.MemAllocatorStats result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.MemAllocatorStats buildPartial() {
-      org.tensorflow.proto.util.MemAllocatorStats result = new org.tensorflow.proto.util.MemAllocatorStats(this);
-      result.numAllocs_ = numAllocs_;
-      result.bytesInUse_ = bytesInUse_;
-      result.peakBytesInUse_ = peakBytesInUse_;
-      result.largestAllocSize_ = largestAllocSize_;
-      result.fragmentationMetric_ = fragmentationMetric_;
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.util.MemAllocatorStats) {
-        return mergeFrom((org.tensorflow.proto.util.MemAllocatorStats)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.util.MemAllocatorStats other) {
-      if (other == org.tensorflow.proto.util.MemAllocatorStats.getDefaultInstance()) return this;
-      if (other.getNumAllocs() != 0L) {
-        setNumAllocs(other.getNumAllocs());
-      }
-      if (other.getBytesInUse() != 0L) {
-        setBytesInUse(other.getBytesInUse());
-      }
-      if (other.getPeakBytesInUse() != 0L) {
-        setPeakBytesInUse(other.getPeakBytesInUse());
-      }
-      if (other.getLargestAllocSize() != 0L) {
-        setLargestAllocSize(other.getLargestAllocSize());
-      }
-      if (other.getFragmentationMetric() != 0F) {
-        setFragmentationMetric(other.getFragmentationMetric());
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.util.MemAllocatorStats parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.util.MemAllocatorStats) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-
-    private long numAllocs_ ;
-    /**
-     * <code>int64 num_allocs = 1;</code>
-     */
-    public long getNumAllocs() {
-      return numAllocs_;
-    }
-    /**
-     * <code>int64 num_allocs = 1;</code>
-     */
-    public Builder setNumAllocs(long value) {
-      
-      numAllocs_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>int64 num_allocs = 1;</code>
-     */
-    public Builder clearNumAllocs() {
-      
-      numAllocs_ = 0L;
-      onChanged();
-      return this;
-    }
-
-    private long bytesInUse_ ;
-    /**
-     * <code>int64 bytes_in_use = 2;</code>
-     */
-    public long getBytesInUse() {
-      return bytesInUse_;
-    }
-    /**
-     * <code>int64 bytes_in_use = 2;</code>
-     */
-    public Builder setBytesInUse(long value) {
-      
-      bytesInUse_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>int64 bytes_in_use = 2;</code>
-     */
-    public Builder clearBytesInUse() {
-      
-      bytesInUse_ = 0L;
-      onChanged();
-      return this;
-    }
-
-    private long peakBytesInUse_ ;
-    /**
-     * <code>int64 peak_bytes_in_use = 3;</code>
-     */
-    public long getPeakBytesInUse() {
-      return peakBytesInUse_;
-    }
-    /**
-     * <code>int64 peak_bytes_in_use = 3;</code>
-     */
-    public Builder setPeakBytesInUse(long value) {
-      
-      peakBytesInUse_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>int64 peak_bytes_in_use = 3;</code>
-     */
-    public Builder clearPeakBytesInUse() {
-      
-      peakBytesInUse_ = 0L;
-      onChanged();
-      return this;
-    }
-
-    private long largestAllocSize_ ;
-    /**
-     * <code>int64 largest_alloc_size = 4;</code>
-     */
-    public long getLargestAllocSize() {
-      return largestAllocSize_;
-    }
-    /**
-     * <code>int64 largest_alloc_size = 4;</code>
-     */
-    public Builder setLargestAllocSize(long value) {
-      
-      largestAllocSize_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>int64 largest_alloc_size = 4;</code>
-     */
-    public Builder clearLargestAllocSize() {
-      
-      largestAllocSize_ = 0L;
-      onChanged();
-      return this;
-    }
-
-    private float fragmentationMetric_ ;
-    /**
-     * <code>float fragmentation_metric = 5;</code>
-     */
-    public float getFragmentationMetric() {
-      return fragmentationMetric_;
-    }
-    /**
-     * <code>float fragmentation_metric = 5;</code>
-     */
-    public Builder setFragmentationMetric(float value) {
-      
-      fragmentationMetric_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>float fragmentation_metric = 5;</code>
-     */
-    public Builder clearFragmentationMetric() {
-      
-      fragmentationMetric_ = 0F;
-      onChanged();
-      return this;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.MemAllocatorStats)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.MemAllocatorStats)
-  private static final org.tensorflow.proto.util.MemAllocatorStats DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.util.MemAllocatorStats();
-  }
-
-  public static org.tensorflow.proto.util.MemAllocatorStats getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<MemAllocatorStats>
-      PARSER = new com.google.protobuf.AbstractParser<MemAllocatorStats>() {
-    @java.lang.Override
-    public MemAllocatorStats parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new MemAllocatorStats(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<MemAllocatorStats> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<MemAllocatorStats> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.util.MemAllocatorStats getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/MemAllocatorStatsOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/MemAllocatorStatsOrBuilder.java
deleted file mode 100644
index 119204effcb..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/MemAllocatorStatsOrBuilder.java
+++ /dev/null
@@ -1,34 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/bfc_memory_map.proto
-
-package org.tensorflow.proto.util;
-
-public interface MemAllocatorStatsOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.MemAllocatorStats)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <code>int64 num_allocs = 1;</code>
-   */
-  long getNumAllocs();
-
-  /**
-   * <code>int64 bytes_in_use = 2;</code>
-   */
-  long getBytesInUse();
-
-  /**
-   * <code>int64 peak_bytes_in_use = 3;</code>
-   */
-  long getPeakBytesInUse();
-
-  /**
-   * <code>int64 largest_alloc_size = 4;</code>
-   */
-  long getLargestAllocSize();
-
-  /**
-   * <code>float fragmentation_metric = 5;</code>
-   */
-  float getFragmentationMetric();
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/MemChunk.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/MemChunk.java
deleted file mode 100644
index 3212e0b9c85..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/MemChunk.java
+++ /dev/null
@@ -1,1009 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/bfc_memory_map.proto
-
-package org.tensorflow.proto.util;
-
-/**
- * Protobuf type {@code tensorflow.MemChunk}
- */
-public  final class MemChunk extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.MemChunk)
-    MemChunkOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use MemChunk.newBuilder() to construct.
-  private MemChunk(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private MemChunk() {
-    opName_ = "";
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new MemChunk();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private MemChunk(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 8: {
-
-            address_ = input.readUInt64();
-            break;
-          }
-          case 16: {
-
-            size_ = input.readInt64();
-            break;
-          }
-          case 24: {
-
-            requestedSize_ = input.readInt64();
-            break;
-          }
-          case 32: {
-
-            bin_ = input.readInt32();
-            break;
-          }
-          case 42: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            opName_ = s;
-            break;
-          }
-          case 48: {
-
-            freedAtCount_ = input.readUInt64();
-            break;
-          }
-          case 56: {
-
-            actionCount_ = input.readUInt64();
-            break;
-          }
-          case 64: {
-
-            inUse_ = input.readBool();
-            break;
-          }
-          case 72: {
-
-            stepId_ = input.readUInt64();
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.util.BfcMemoryMapProtos.internal_static_tensorflow_MemChunk_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.util.BfcMemoryMapProtos.internal_static_tensorflow_MemChunk_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.util.MemChunk.class, org.tensorflow.proto.util.MemChunk.Builder.class);
-  }
-
-  public static final int ADDRESS_FIELD_NUMBER = 1;
-  private long address_;
-  /**
-   * <code>uint64 address = 1;</code>
-   */
-  public long getAddress() {
-    return address_;
-  }
-
-  public static final int SIZE_FIELD_NUMBER = 2;
-  private long size_;
-  /**
-   * <code>int64 size = 2;</code>
-   */
-  public long getSize() {
-    return size_;
-  }
-
-  public static final int REQUESTED_SIZE_FIELD_NUMBER = 3;
-  private long requestedSize_;
-  /**
-   * <code>int64 requested_size = 3;</code>
-   */
-  public long getRequestedSize() {
-    return requestedSize_;
-  }
-
-  public static final int BIN_FIELD_NUMBER = 4;
-  private int bin_;
-  /**
-   * <code>int32 bin = 4;</code>
-   */
-  public int getBin() {
-    return bin_;
-  }
-
-  public static final int OP_NAME_FIELD_NUMBER = 5;
-  private volatile java.lang.Object opName_;
-  /**
-   * <code>string op_name = 5;</code>
-   */
-  public java.lang.String getOpName() {
-    java.lang.Object ref = opName_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      opName_ = s;
-      return s;
-    }
-  }
-  /**
-   * <code>string op_name = 5;</code>
-   */
-  public com.google.protobuf.ByteString
-      getOpNameBytes() {
-    java.lang.Object ref = opName_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      opName_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int FREED_AT_COUNT_FIELD_NUMBER = 6;
-  private long freedAtCount_;
-  /**
-   * <code>uint64 freed_at_count = 6;</code>
-   */
-  public long getFreedAtCount() {
-    return freedAtCount_;
-  }
-
-  public static final int ACTION_COUNT_FIELD_NUMBER = 7;
-  private long actionCount_;
-  /**
-   * <code>uint64 action_count = 7;</code>
-   */
-  public long getActionCount() {
-    return actionCount_;
-  }
-
-  public static final int IN_USE_FIELD_NUMBER = 8;
-  private boolean inUse_;
-  /**
-   * <code>bool in_use = 8;</code>
-   */
-  public boolean getInUse() {
-    return inUse_;
-  }
-
-  public static final int STEP_ID_FIELD_NUMBER = 9;
-  private long stepId_;
-  /**
-   * <code>uint64 step_id = 9;</code>
-   */
-  public long getStepId() {
-    return stepId_;
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    if (address_ != 0L) {
-      output.writeUInt64(1, address_);
-    }
-    if (size_ != 0L) {
-      output.writeInt64(2, size_);
-    }
-    if (requestedSize_ != 0L) {
-      output.writeInt64(3, requestedSize_);
-    }
-    if (bin_ != 0) {
-      output.writeInt32(4, bin_);
-    }
-    if (!getOpNameBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 5, opName_);
-    }
-    if (freedAtCount_ != 0L) {
-      output.writeUInt64(6, freedAtCount_);
-    }
-    if (actionCount_ != 0L) {
-      output.writeUInt64(7, actionCount_);
-    }
-    if (inUse_ != false) {
-      output.writeBool(8, inUse_);
-    }
-    if (stepId_ != 0L) {
-      output.writeUInt64(9, stepId_);
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (address_ != 0L) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeUInt64Size(1, address_);
-    }
-    if (size_ != 0L) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt64Size(2, size_);
-    }
-    if (requestedSize_ != 0L) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt64Size(3, requestedSize_);
-    }
-    if (bin_ != 0) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt32Size(4, bin_);
-    }
-    if (!getOpNameBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(5, opName_);
-    }
-    if (freedAtCount_ != 0L) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeUInt64Size(6, freedAtCount_);
-    }
-    if (actionCount_ != 0L) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeUInt64Size(7, actionCount_);
-    }
-    if (inUse_ != false) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeBoolSize(8, inUse_);
-    }
-    if (stepId_ != 0L) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeUInt64Size(9, stepId_);
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.util.MemChunk)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.util.MemChunk other = (org.tensorflow.proto.util.MemChunk) obj;
-
-    if (getAddress()
-        != other.getAddress()) return false;
-    if (getSize()
-        != other.getSize()) return false;
-    if (getRequestedSize()
-        != other.getRequestedSize()) return false;
-    if (getBin()
-        != other.getBin()) return false;
-    if (!getOpName()
-        .equals(other.getOpName())) return false;
-    if (getFreedAtCount()
-        != other.getFreedAtCount()) return false;
-    if (getActionCount()
-        != other.getActionCount()) return false;
-    if (getInUse()
-        != other.getInUse()) return false;
-    if (getStepId()
-        != other.getStepId()) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    hash = (37 * hash) + ADDRESS_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-        getAddress());
-    hash = (37 * hash) + SIZE_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-        getSize());
-    hash = (37 * hash) + REQUESTED_SIZE_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-        getRequestedSize());
-    hash = (37 * hash) + BIN_FIELD_NUMBER;
-    hash = (53 * hash) + getBin();
-    hash = (37 * hash) + OP_NAME_FIELD_NUMBER;
-    hash = (53 * hash) + getOpName().hashCode();
-    hash = (37 * hash) + FREED_AT_COUNT_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-        getFreedAtCount());
-    hash = (37 * hash) + ACTION_COUNT_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-        getActionCount());
-    hash = (37 * hash) + IN_USE_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashBoolean(
-        getInUse());
-    hash = (37 * hash) + STEP_ID_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-        getStepId());
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.util.MemChunk parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.MemChunk parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.MemChunk parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.MemChunk parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.MemChunk parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.MemChunk parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.MemChunk parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.MemChunk parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.MemChunk parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.MemChunk parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.MemChunk parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.MemChunk parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.util.MemChunk prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * Protobuf type {@code tensorflow.MemChunk}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.MemChunk)
-      org.tensorflow.proto.util.MemChunkOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.util.BfcMemoryMapProtos.internal_static_tensorflow_MemChunk_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.util.BfcMemoryMapProtos.internal_static_tensorflow_MemChunk_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.util.MemChunk.class, org.tensorflow.proto.util.MemChunk.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.util.MemChunk.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      address_ = 0L;
-
-      size_ = 0L;
-
-      requestedSize_ = 0L;
-
-      bin_ = 0;
-
-      opName_ = "";
-
-      freedAtCount_ = 0L;
-
-      actionCount_ = 0L;
-
-      inUse_ = false;
-
-      stepId_ = 0L;
-
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.util.BfcMemoryMapProtos.internal_static_tensorflow_MemChunk_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.MemChunk getDefaultInstanceForType() {
-      return org.tensorflow.proto.util.MemChunk.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.MemChunk build() {
-      org.tensorflow.proto.util.MemChunk result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.MemChunk buildPartial() {
-      org.tensorflow.proto.util.MemChunk result = new org.tensorflow.proto.util.MemChunk(this);
-      result.address_ = address_;
-      result.size_ = size_;
-      result.requestedSize_ = requestedSize_;
-      result.bin_ = bin_;
-      result.opName_ = opName_;
-      result.freedAtCount_ = freedAtCount_;
-      result.actionCount_ = actionCount_;
-      result.inUse_ = inUse_;
-      result.stepId_ = stepId_;
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.util.MemChunk) {
-        return mergeFrom((org.tensorflow.proto.util.MemChunk)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.util.MemChunk other) {
-      if (other == org.tensorflow.proto.util.MemChunk.getDefaultInstance()) return this;
-      if (other.getAddress() != 0L) {
-        setAddress(other.getAddress());
-      }
-      if (other.getSize() != 0L) {
-        setSize(other.getSize());
-      }
-      if (other.getRequestedSize() != 0L) {
-        setRequestedSize(other.getRequestedSize());
-      }
-      if (other.getBin() != 0) {
-        setBin(other.getBin());
-      }
-      if (!other.getOpName().isEmpty()) {
-        opName_ = other.opName_;
-        onChanged();
-      }
-      if (other.getFreedAtCount() != 0L) {
-        setFreedAtCount(other.getFreedAtCount());
-      }
-      if (other.getActionCount() != 0L) {
-        setActionCount(other.getActionCount());
-      }
-      if (other.getInUse() != false) {
-        setInUse(other.getInUse());
-      }
-      if (other.getStepId() != 0L) {
-        setStepId(other.getStepId());
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.util.MemChunk parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.util.MemChunk) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-
-    private long address_ ;
-    /**
-     * <code>uint64 address = 1;</code>
-     */
-    public long getAddress() {
-      return address_;
-    }
-    /**
-     * <code>uint64 address = 1;</code>
-     */
-    public Builder setAddress(long value) {
-      
-      address_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>uint64 address = 1;</code>
-     */
-    public Builder clearAddress() {
-      
-      address_ = 0L;
-      onChanged();
-      return this;
-    }
-
-    private long size_ ;
-    /**
-     * <code>int64 size = 2;</code>
-     */
-    public long getSize() {
-      return size_;
-    }
-    /**
-     * <code>int64 size = 2;</code>
-     */
-    public Builder setSize(long value) {
-      
-      size_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>int64 size = 2;</code>
-     */
-    public Builder clearSize() {
-      
-      size_ = 0L;
-      onChanged();
-      return this;
-    }
-
-    private long requestedSize_ ;
-    /**
-     * <code>int64 requested_size = 3;</code>
-     */
-    public long getRequestedSize() {
-      return requestedSize_;
-    }
-    /**
-     * <code>int64 requested_size = 3;</code>
-     */
-    public Builder setRequestedSize(long value) {
-      
-      requestedSize_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>int64 requested_size = 3;</code>
-     */
-    public Builder clearRequestedSize() {
-      
-      requestedSize_ = 0L;
-      onChanged();
-      return this;
-    }
-
-    private int bin_ ;
-    /**
-     * <code>int32 bin = 4;</code>
-     */
-    public int getBin() {
-      return bin_;
-    }
-    /**
-     * <code>int32 bin = 4;</code>
-     */
-    public Builder setBin(int value) {
-      
-      bin_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>int32 bin = 4;</code>
-     */
-    public Builder clearBin() {
-      
-      bin_ = 0;
-      onChanged();
-      return this;
-    }
-
-    private java.lang.Object opName_ = "";
-    /**
-     * <code>string op_name = 5;</code>
-     */
-    public java.lang.String getOpName() {
-      java.lang.Object ref = opName_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        opName_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <code>string op_name = 5;</code>
-     */
-    public com.google.protobuf.ByteString
-        getOpNameBytes() {
-      java.lang.Object ref = opName_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        opName_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <code>string op_name = 5;</code>
-     */
-    public Builder setOpName(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      opName_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>string op_name = 5;</code>
-     */
-    public Builder clearOpName() {
-      
-      opName_ = getDefaultInstance().getOpName();
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>string op_name = 5;</code>
-     */
-    public Builder setOpNameBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      opName_ = value;
-      onChanged();
-      return this;
-    }
-
-    private long freedAtCount_ ;
-    /**
-     * <code>uint64 freed_at_count = 6;</code>
-     */
-    public long getFreedAtCount() {
-      return freedAtCount_;
-    }
-    /**
-     * <code>uint64 freed_at_count = 6;</code>
-     */
-    public Builder setFreedAtCount(long value) {
-      
-      freedAtCount_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>uint64 freed_at_count = 6;</code>
-     */
-    public Builder clearFreedAtCount() {
-      
-      freedAtCount_ = 0L;
-      onChanged();
-      return this;
-    }
-
-    private long actionCount_ ;
-    /**
-     * <code>uint64 action_count = 7;</code>
-     */
-    public long getActionCount() {
-      return actionCount_;
-    }
-    /**
-     * <code>uint64 action_count = 7;</code>
-     */
-    public Builder setActionCount(long value) {
-      
-      actionCount_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>uint64 action_count = 7;</code>
-     */
-    public Builder clearActionCount() {
-      
-      actionCount_ = 0L;
-      onChanged();
-      return this;
-    }
-
-    private boolean inUse_ ;
-    /**
-     * <code>bool in_use = 8;</code>
-     */
-    public boolean getInUse() {
-      return inUse_;
-    }
-    /**
-     * <code>bool in_use = 8;</code>
-     */
-    public Builder setInUse(boolean value) {
-      
-      inUse_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>bool in_use = 8;</code>
-     */
-    public Builder clearInUse() {
-      
-      inUse_ = false;
-      onChanged();
-      return this;
-    }
-
-    private long stepId_ ;
-    /**
-     * <code>uint64 step_id = 9;</code>
-     */
-    public long getStepId() {
-      return stepId_;
-    }
-    /**
-     * <code>uint64 step_id = 9;</code>
-     */
-    public Builder setStepId(long value) {
-      
-      stepId_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>uint64 step_id = 9;</code>
-     */
-    public Builder clearStepId() {
-      
-      stepId_ = 0L;
-      onChanged();
-      return this;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.MemChunk)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.MemChunk)
-  private static final org.tensorflow.proto.util.MemChunk DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.util.MemChunk();
-  }
-
-  public static org.tensorflow.proto.util.MemChunk getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<MemChunk>
-      PARSER = new com.google.protobuf.AbstractParser<MemChunk>() {
-    @java.lang.Override
-    public MemChunk parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new MemChunk(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<MemChunk> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<MemChunk> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.util.MemChunk getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/MemChunkOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/MemChunkOrBuilder.java
deleted file mode 100644
index 6dbce3e91ab..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/MemChunkOrBuilder.java
+++ /dev/null
@@ -1,59 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/bfc_memory_map.proto
-
-package org.tensorflow.proto.util;
-
-public interface MemChunkOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.MemChunk)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <code>uint64 address = 1;</code>
-   */
-  long getAddress();
-
-  /**
-   * <code>int64 size = 2;</code>
-   */
-  long getSize();
-
-  /**
-   * <code>int64 requested_size = 3;</code>
-   */
-  long getRequestedSize();
-
-  /**
-   * <code>int32 bin = 4;</code>
-   */
-  int getBin();
-
-  /**
-   * <code>string op_name = 5;</code>
-   */
-  java.lang.String getOpName();
-  /**
-   * <code>string op_name = 5;</code>
-   */
-  com.google.protobuf.ByteString
-      getOpNameBytes();
-
-  /**
-   * <code>uint64 freed_at_count = 6;</code>
-   */
-  long getFreedAtCount();
-
-  /**
-   * <code>uint64 action_count = 7;</code>
-   */
-  long getActionCount();
-
-  /**
-   * <code>bool in_use = 8;</code>
-   */
-  boolean getInUse();
-
-  /**
-   * <code>uint64 step_id = 9;</code>
-   */
-  long getStepId();
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/MemmappedFileSystemDirectory.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/MemmappedFileSystemDirectory.java
deleted file mode 100644
index 0f7f977819f..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/MemmappedFileSystemDirectory.java
+++ /dev/null
@@ -1,773 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/util/memmapped_file_system.proto
-
-package org.tensorflow.proto.util;
-
-/**
- * <pre>
- * A directory of regions in a memmapped file.
- * </pre>
- *
- * Protobuf type {@code tensorflow.MemmappedFileSystemDirectory}
- */
-public  final class MemmappedFileSystemDirectory extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.MemmappedFileSystemDirectory)
-    MemmappedFileSystemDirectoryOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use MemmappedFileSystemDirectory.newBuilder() to construct.
-  private MemmappedFileSystemDirectory(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private MemmappedFileSystemDirectory() {
-    element_ = java.util.Collections.emptyList();
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new MemmappedFileSystemDirectory();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private MemmappedFileSystemDirectory(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    int mutable_bitField0_ = 0;
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 10: {
-            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
-              element_ = new java.util.ArrayList<org.tensorflow.proto.util.MemmappedFileSystemDirectoryElement>();
-              mutable_bitField0_ |= 0x00000001;
-            }
-            element_.add(
-                input.readMessage(org.tensorflow.proto.util.MemmappedFileSystemDirectoryElement.parser(), extensionRegistry));
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      if (((mutable_bitField0_ & 0x00000001) != 0)) {
-        element_ = java.util.Collections.unmodifiableList(element_);
-      }
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.util.MemmappedFileSystemProtos.internal_static_tensorflow_MemmappedFileSystemDirectory_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.util.MemmappedFileSystemProtos.internal_static_tensorflow_MemmappedFileSystemDirectory_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.util.MemmappedFileSystemDirectory.class, org.tensorflow.proto.util.MemmappedFileSystemDirectory.Builder.class);
-  }
-
-  public static final int ELEMENT_FIELD_NUMBER = 1;
-  private java.util.List<org.tensorflow.proto.util.MemmappedFileSystemDirectoryElement> element_;
-  /**
-   * <code>repeated .tensorflow.MemmappedFileSystemDirectoryElement element = 1;</code>
-   */
-  public java.util.List<org.tensorflow.proto.util.MemmappedFileSystemDirectoryElement> getElementList() {
-    return element_;
-  }
-  /**
-   * <code>repeated .tensorflow.MemmappedFileSystemDirectoryElement element = 1;</code>
-   */
-  public java.util.List<? extends org.tensorflow.proto.util.MemmappedFileSystemDirectoryElementOrBuilder> 
-      getElementOrBuilderList() {
-    return element_;
-  }
-  /**
-   * <code>repeated .tensorflow.MemmappedFileSystemDirectoryElement element = 1;</code>
-   */
-  public int getElementCount() {
-    return element_.size();
-  }
-  /**
-   * <code>repeated .tensorflow.MemmappedFileSystemDirectoryElement element = 1;</code>
-   */
-  public org.tensorflow.proto.util.MemmappedFileSystemDirectoryElement getElement(int index) {
-    return element_.get(index);
-  }
-  /**
-   * <code>repeated .tensorflow.MemmappedFileSystemDirectoryElement element = 1;</code>
-   */
-  public org.tensorflow.proto.util.MemmappedFileSystemDirectoryElementOrBuilder getElementOrBuilder(
-      int index) {
-    return element_.get(index);
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    for (int i = 0; i < element_.size(); i++) {
-      output.writeMessage(1, element_.get(i));
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    for (int i = 0; i < element_.size(); i++) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(1, element_.get(i));
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.util.MemmappedFileSystemDirectory)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.util.MemmappedFileSystemDirectory other = (org.tensorflow.proto.util.MemmappedFileSystemDirectory) obj;
-
-    if (!getElementList()
-        .equals(other.getElementList())) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    if (getElementCount() > 0) {
-      hash = (37 * hash) + ELEMENT_FIELD_NUMBER;
-      hash = (53 * hash) + getElementList().hashCode();
-    }
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.util.MemmappedFileSystemDirectory parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.MemmappedFileSystemDirectory parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.MemmappedFileSystemDirectory parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.MemmappedFileSystemDirectory parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.MemmappedFileSystemDirectory parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.MemmappedFileSystemDirectory parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.MemmappedFileSystemDirectory parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.MemmappedFileSystemDirectory parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.MemmappedFileSystemDirectory parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.MemmappedFileSystemDirectory parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.MemmappedFileSystemDirectory parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.MemmappedFileSystemDirectory parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.util.MemmappedFileSystemDirectory prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * <pre>
-   * A directory of regions in a memmapped file.
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.MemmappedFileSystemDirectory}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.MemmappedFileSystemDirectory)
-      org.tensorflow.proto.util.MemmappedFileSystemDirectoryOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.util.MemmappedFileSystemProtos.internal_static_tensorflow_MemmappedFileSystemDirectory_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.util.MemmappedFileSystemProtos.internal_static_tensorflow_MemmappedFileSystemDirectory_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.util.MemmappedFileSystemDirectory.class, org.tensorflow.proto.util.MemmappedFileSystemDirectory.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.util.MemmappedFileSystemDirectory.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-        getElementFieldBuilder();
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      if (elementBuilder_ == null) {
-        element_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000001);
-      } else {
-        elementBuilder_.clear();
-      }
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.util.MemmappedFileSystemProtos.internal_static_tensorflow_MemmappedFileSystemDirectory_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.MemmappedFileSystemDirectory getDefaultInstanceForType() {
-      return org.tensorflow.proto.util.MemmappedFileSystemDirectory.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.MemmappedFileSystemDirectory build() {
-      org.tensorflow.proto.util.MemmappedFileSystemDirectory result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.MemmappedFileSystemDirectory buildPartial() {
-      org.tensorflow.proto.util.MemmappedFileSystemDirectory result = new org.tensorflow.proto.util.MemmappedFileSystemDirectory(this);
-      int from_bitField0_ = bitField0_;
-      if (elementBuilder_ == null) {
-        if (((bitField0_ & 0x00000001) != 0)) {
-          element_ = java.util.Collections.unmodifiableList(element_);
-          bitField0_ = (bitField0_ & ~0x00000001);
-        }
-        result.element_ = element_;
-      } else {
-        result.element_ = elementBuilder_.build();
-      }
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.util.MemmappedFileSystemDirectory) {
-        return mergeFrom((org.tensorflow.proto.util.MemmappedFileSystemDirectory)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.util.MemmappedFileSystemDirectory other) {
-      if (other == org.tensorflow.proto.util.MemmappedFileSystemDirectory.getDefaultInstance()) return this;
-      if (elementBuilder_ == null) {
-        if (!other.element_.isEmpty()) {
-          if (element_.isEmpty()) {
-            element_ = other.element_;
-            bitField0_ = (bitField0_ & ~0x00000001);
-          } else {
-            ensureElementIsMutable();
-            element_.addAll(other.element_);
-          }
-          onChanged();
-        }
-      } else {
-        if (!other.element_.isEmpty()) {
-          if (elementBuilder_.isEmpty()) {
-            elementBuilder_.dispose();
-            elementBuilder_ = null;
-            element_ = other.element_;
-            bitField0_ = (bitField0_ & ~0x00000001);
-            elementBuilder_ = 
-              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
-                 getElementFieldBuilder() : null;
-          } else {
-            elementBuilder_.addAllMessages(other.element_);
-          }
-        }
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.util.MemmappedFileSystemDirectory parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.util.MemmappedFileSystemDirectory) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-    private int bitField0_;
-
-    private java.util.List<org.tensorflow.proto.util.MemmappedFileSystemDirectoryElement> element_ =
-      java.util.Collections.emptyList();
-    private void ensureElementIsMutable() {
-      if (!((bitField0_ & 0x00000001) != 0)) {
-        element_ = new java.util.ArrayList<org.tensorflow.proto.util.MemmappedFileSystemDirectoryElement>(element_);
-        bitField0_ |= 0x00000001;
-       }
-    }
-
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.util.MemmappedFileSystemDirectoryElement, org.tensorflow.proto.util.MemmappedFileSystemDirectoryElement.Builder, org.tensorflow.proto.util.MemmappedFileSystemDirectoryElementOrBuilder> elementBuilder_;
-
-    /**
-     * <code>repeated .tensorflow.MemmappedFileSystemDirectoryElement element = 1;</code>
-     */
-    public java.util.List<org.tensorflow.proto.util.MemmappedFileSystemDirectoryElement> getElementList() {
-      if (elementBuilder_ == null) {
-        return java.util.Collections.unmodifiableList(element_);
-      } else {
-        return elementBuilder_.getMessageList();
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.MemmappedFileSystemDirectoryElement element = 1;</code>
-     */
-    public int getElementCount() {
-      if (elementBuilder_ == null) {
-        return element_.size();
-      } else {
-        return elementBuilder_.getCount();
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.MemmappedFileSystemDirectoryElement element = 1;</code>
-     */
-    public org.tensorflow.proto.util.MemmappedFileSystemDirectoryElement getElement(int index) {
-      if (elementBuilder_ == null) {
-        return element_.get(index);
-      } else {
-        return elementBuilder_.getMessage(index);
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.MemmappedFileSystemDirectoryElement element = 1;</code>
-     */
-    public Builder setElement(
-        int index, org.tensorflow.proto.util.MemmappedFileSystemDirectoryElement value) {
-      if (elementBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureElementIsMutable();
-        element_.set(index, value);
-        onChanged();
-      } else {
-        elementBuilder_.setMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.MemmappedFileSystemDirectoryElement element = 1;</code>
-     */
-    public Builder setElement(
-        int index, org.tensorflow.proto.util.MemmappedFileSystemDirectoryElement.Builder builderForValue) {
-      if (elementBuilder_ == null) {
-        ensureElementIsMutable();
-        element_.set(index, builderForValue.build());
-        onChanged();
-      } else {
-        elementBuilder_.setMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.MemmappedFileSystemDirectoryElement element = 1;</code>
-     */
-    public Builder addElement(org.tensorflow.proto.util.MemmappedFileSystemDirectoryElement value) {
-      if (elementBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureElementIsMutable();
-        element_.add(value);
-        onChanged();
-      } else {
-        elementBuilder_.addMessage(value);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.MemmappedFileSystemDirectoryElement element = 1;</code>
-     */
-    public Builder addElement(
-        int index, org.tensorflow.proto.util.MemmappedFileSystemDirectoryElement value) {
-      if (elementBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureElementIsMutable();
-        element_.add(index, value);
-        onChanged();
-      } else {
-        elementBuilder_.addMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.MemmappedFileSystemDirectoryElement element = 1;</code>
-     */
-    public Builder addElement(
-        org.tensorflow.proto.util.MemmappedFileSystemDirectoryElement.Builder builderForValue) {
-      if (elementBuilder_ == null) {
-        ensureElementIsMutable();
-        element_.add(builderForValue.build());
-        onChanged();
-      } else {
-        elementBuilder_.addMessage(builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.MemmappedFileSystemDirectoryElement element = 1;</code>
-     */
-    public Builder addElement(
-        int index, org.tensorflow.proto.util.MemmappedFileSystemDirectoryElement.Builder builderForValue) {
-      if (elementBuilder_ == null) {
-        ensureElementIsMutable();
-        element_.add(index, builderForValue.build());
-        onChanged();
-      } else {
-        elementBuilder_.addMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.MemmappedFileSystemDirectoryElement element = 1;</code>
-     */
-    public Builder addAllElement(
-        java.lang.Iterable<? extends org.tensorflow.proto.util.MemmappedFileSystemDirectoryElement> values) {
-      if (elementBuilder_ == null) {
-        ensureElementIsMutable();
-        com.google.protobuf.AbstractMessageLite.Builder.addAll(
-            values, element_);
-        onChanged();
-      } else {
-        elementBuilder_.addAllMessages(values);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.MemmappedFileSystemDirectoryElement element = 1;</code>
-     */
-    public Builder clearElement() {
-      if (elementBuilder_ == null) {
-        element_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000001);
-        onChanged();
-      } else {
-        elementBuilder_.clear();
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.MemmappedFileSystemDirectoryElement element = 1;</code>
-     */
-    public Builder removeElement(int index) {
-      if (elementBuilder_ == null) {
-        ensureElementIsMutable();
-        element_.remove(index);
-        onChanged();
-      } else {
-        elementBuilder_.remove(index);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.MemmappedFileSystemDirectoryElement element = 1;</code>
-     */
-    public org.tensorflow.proto.util.MemmappedFileSystemDirectoryElement.Builder getElementBuilder(
-        int index) {
-      return getElementFieldBuilder().getBuilder(index);
-    }
-    /**
-     * <code>repeated .tensorflow.MemmappedFileSystemDirectoryElement element = 1;</code>
-     */
-    public org.tensorflow.proto.util.MemmappedFileSystemDirectoryElementOrBuilder getElementOrBuilder(
-        int index) {
-      if (elementBuilder_ == null) {
-        return element_.get(index);  } else {
-        return elementBuilder_.getMessageOrBuilder(index);
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.MemmappedFileSystemDirectoryElement element = 1;</code>
-     */
-    public java.util.List<? extends org.tensorflow.proto.util.MemmappedFileSystemDirectoryElementOrBuilder> 
-         getElementOrBuilderList() {
-      if (elementBuilder_ != null) {
-        return elementBuilder_.getMessageOrBuilderList();
-      } else {
-        return java.util.Collections.unmodifiableList(element_);
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.MemmappedFileSystemDirectoryElement element = 1;</code>
-     */
-    public org.tensorflow.proto.util.MemmappedFileSystemDirectoryElement.Builder addElementBuilder() {
-      return getElementFieldBuilder().addBuilder(
-          org.tensorflow.proto.util.MemmappedFileSystemDirectoryElement.getDefaultInstance());
-    }
-    /**
-     * <code>repeated .tensorflow.MemmappedFileSystemDirectoryElement element = 1;</code>
-     */
-    public org.tensorflow.proto.util.MemmappedFileSystemDirectoryElement.Builder addElementBuilder(
-        int index) {
-      return getElementFieldBuilder().addBuilder(
-          index, org.tensorflow.proto.util.MemmappedFileSystemDirectoryElement.getDefaultInstance());
-    }
-    /**
-     * <code>repeated .tensorflow.MemmappedFileSystemDirectoryElement element = 1;</code>
-     */
-    public java.util.List<org.tensorflow.proto.util.MemmappedFileSystemDirectoryElement.Builder> 
-         getElementBuilderList() {
-      return getElementFieldBuilder().getBuilderList();
-    }
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.util.MemmappedFileSystemDirectoryElement, org.tensorflow.proto.util.MemmappedFileSystemDirectoryElement.Builder, org.tensorflow.proto.util.MemmappedFileSystemDirectoryElementOrBuilder> 
-        getElementFieldBuilder() {
-      if (elementBuilder_ == null) {
-        elementBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
-            org.tensorflow.proto.util.MemmappedFileSystemDirectoryElement, org.tensorflow.proto.util.MemmappedFileSystemDirectoryElement.Builder, org.tensorflow.proto.util.MemmappedFileSystemDirectoryElementOrBuilder>(
-                element_,
-                ((bitField0_ & 0x00000001) != 0),
-                getParentForChildren(),
-                isClean());
-        element_ = null;
-      }
-      return elementBuilder_;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.MemmappedFileSystemDirectory)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.MemmappedFileSystemDirectory)
-  private static final org.tensorflow.proto.util.MemmappedFileSystemDirectory DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.util.MemmappedFileSystemDirectory();
-  }
-
-  public static org.tensorflow.proto.util.MemmappedFileSystemDirectory getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<MemmappedFileSystemDirectory>
-      PARSER = new com.google.protobuf.AbstractParser<MemmappedFileSystemDirectory>() {
-    @java.lang.Override
-    public MemmappedFileSystemDirectory parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new MemmappedFileSystemDirectory(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<MemmappedFileSystemDirectory> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<MemmappedFileSystemDirectory> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.util.MemmappedFileSystemDirectory getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/MemmappedFileSystemDirectoryElement.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/MemmappedFileSystemDirectoryElement.java
deleted file mode 100644
index d2dafba3ef2..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/MemmappedFileSystemDirectoryElement.java
+++ /dev/null
@@ -1,670 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/util/memmapped_file_system.proto
-
-package org.tensorflow.proto.util;
-
-/**
- * <pre>
- * A message that describes one region of memmapped file.
- * </pre>
- *
- * Protobuf type {@code tensorflow.MemmappedFileSystemDirectoryElement}
- */
-public  final class MemmappedFileSystemDirectoryElement extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.MemmappedFileSystemDirectoryElement)
-    MemmappedFileSystemDirectoryElementOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use MemmappedFileSystemDirectoryElement.newBuilder() to construct.
-  private MemmappedFileSystemDirectoryElement(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private MemmappedFileSystemDirectoryElement() {
-    name_ = "";
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new MemmappedFileSystemDirectoryElement();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private MemmappedFileSystemDirectoryElement(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 8: {
-
-            offset_ = input.readUInt64();
-            break;
-          }
-          case 18: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            name_ = s;
-            break;
-          }
-          case 24: {
-
-            length_ = input.readUInt64();
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.util.MemmappedFileSystemProtos.internal_static_tensorflow_MemmappedFileSystemDirectoryElement_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.util.MemmappedFileSystemProtos.internal_static_tensorflow_MemmappedFileSystemDirectoryElement_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.util.MemmappedFileSystemDirectoryElement.class, org.tensorflow.proto.util.MemmappedFileSystemDirectoryElement.Builder.class);
-  }
-
-  public static final int OFFSET_FIELD_NUMBER = 1;
-  private long offset_;
-  /**
-   * <code>uint64 offset = 1;</code>
-   */
-  public long getOffset() {
-    return offset_;
-  }
-
-  public static final int NAME_FIELD_NUMBER = 2;
-  private volatile java.lang.Object name_;
-  /**
-   * <code>string name = 2;</code>
-   */
-  public java.lang.String getName() {
-    java.lang.Object ref = name_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      name_ = s;
-      return s;
-    }
-  }
-  /**
-   * <code>string name = 2;</code>
-   */
-  public com.google.protobuf.ByteString
-      getNameBytes() {
-    java.lang.Object ref = name_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      name_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int LENGTH_FIELD_NUMBER = 3;
-  private long length_;
-  /**
-   * <code>uint64 length = 3;</code>
-   */
-  public long getLength() {
-    return length_;
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    if (offset_ != 0L) {
-      output.writeUInt64(1, offset_);
-    }
-    if (!getNameBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 2, name_);
-    }
-    if (length_ != 0L) {
-      output.writeUInt64(3, length_);
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (offset_ != 0L) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeUInt64Size(1, offset_);
-    }
-    if (!getNameBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(2, name_);
-    }
-    if (length_ != 0L) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeUInt64Size(3, length_);
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.util.MemmappedFileSystemDirectoryElement)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.util.MemmappedFileSystemDirectoryElement other = (org.tensorflow.proto.util.MemmappedFileSystemDirectoryElement) obj;
-
-    if (getOffset()
-        != other.getOffset()) return false;
-    if (!getName()
-        .equals(other.getName())) return false;
-    if (getLength()
-        != other.getLength()) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    hash = (37 * hash) + OFFSET_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-        getOffset());
-    hash = (37 * hash) + NAME_FIELD_NUMBER;
-    hash = (53 * hash) + getName().hashCode();
-    hash = (37 * hash) + LENGTH_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-        getLength());
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.util.MemmappedFileSystemDirectoryElement parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.MemmappedFileSystemDirectoryElement parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.MemmappedFileSystemDirectoryElement parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.MemmappedFileSystemDirectoryElement parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.MemmappedFileSystemDirectoryElement parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.MemmappedFileSystemDirectoryElement parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.MemmappedFileSystemDirectoryElement parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.MemmappedFileSystemDirectoryElement parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.MemmappedFileSystemDirectoryElement parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.MemmappedFileSystemDirectoryElement parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.MemmappedFileSystemDirectoryElement parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.MemmappedFileSystemDirectoryElement parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.util.MemmappedFileSystemDirectoryElement prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * <pre>
-   * A message that describes one region of memmapped file.
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.MemmappedFileSystemDirectoryElement}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.MemmappedFileSystemDirectoryElement)
-      org.tensorflow.proto.util.MemmappedFileSystemDirectoryElementOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.util.MemmappedFileSystemProtos.internal_static_tensorflow_MemmappedFileSystemDirectoryElement_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.util.MemmappedFileSystemProtos.internal_static_tensorflow_MemmappedFileSystemDirectoryElement_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.util.MemmappedFileSystemDirectoryElement.class, org.tensorflow.proto.util.MemmappedFileSystemDirectoryElement.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.util.MemmappedFileSystemDirectoryElement.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      offset_ = 0L;
-
-      name_ = "";
-
-      length_ = 0L;
-
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.util.MemmappedFileSystemProtos.internal_static_tensorflow_MemmappedFileSystemDirectoryElement_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.MemmappedFileSystemDirectoryElement getDefaultInstanceForType() {
-      return org.tensorflow.proto.util.MemmappedFileSystemDirectoryElement.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.MemmappedFileSystemDirectoryElement build() {
-      org.tensorflow.proto.util.MemmappedFileSystemDirectoryElement result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.MemmappedFileSystemDirectoryElement buildPartial() {
-      org.tensorflow.proto.util.MemmappedFileSystemDirectoryElement result = new org.tensorflow.proto.util.MemmappedFileSystemDirectoryElement(this);
-      result.offset_ = offset_;
-      result.name_ = name_;
-      result.length_ = length_;
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.util.MemmappedFileSystemDirectoryElement) {
-        return mergeFrom((org.tensorflow.proto.util.MemmappedFileSystemDirectoryElement)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.util.MemmappedFileSystemDirectoryElement other) {
-      if (other == org.tensorflow.proto.util.MemmappedFileSystemDirectoryElement.getDefaultInstance()) return this;
-      if (other.getOffset() != 0L) {
-        setOffset(other.getOffset());
-      }
-      if (!other.getName().isEmpty()) {
-        name_ = other.name_;
-        onChanged();
-      }
-      if (other.getLength() != 0L) {
-        setLength(other.getLength());
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.util.MemmappedFileSystemDirectoryElement parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.util.MemmappedFileSystemDirectoryElement) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-
-    private long offset_ ;
-    /**
-     * <code>uint64 offset = 1;</code>
-     */
-    public long getOffset() {
-      return offset_;
-    }
-    /**
-     * <code>uint64 offset = 1;</code>
-     */
-    public Builder setOffset(long value) {
-      
-      offset_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>uint64 offset = 1;</code>
-     */
-    public Builder clearOffset() {
-      
-      offset_ = 0L;
-      onChanged();
-      return this;
-    }
-
-    private java.lang.Object name_ = "";
-    /**
-     * <code>string name = 2;</code>
-     */
-    public java.lang.String getName() {
-      java.lang.Object ref = name_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        name_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <code>string name = 2;</code>
-     */
-    public com.google.protobuf.ByteString
-        getNameBytes() {
-      java.lang.Object ref = name_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        name_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <code>string name = 2;</code>
-     */
-    public Builder setName(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      name_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>string name = 2;</code>
-     */
-    public Builder clearName() {
-      
-      name_ = getDefaultInstance().getName();
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>string name = 2;</code>
-     */
-    public Builder setNameBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      name_ = value;
-      onChanged();
-      return this;
-    }
-
-    private long length_ ;
-    /**
-     * <code>uint64 length = 3;</code>
-     */
-    public long getLength() {
-      return length_;
-    }
-    /**
-     * <code>uint64 length = 3;</code>
-     */
-    public Builder setLength(long value) {
-      
-      length_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>uint64 length = 3;</code>
-     */
-    public Builder clearLength() {
-      
-      length_ = 0L;
-      onChanged();
-      return this;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.MemmappedFileSystemDirectoryElement)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.MemmappedFileSystemDirectoryElement)
-  private static final org.tensorflow.proto.util.MemmappedFileSystemDirectoryElement DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.util.MemmappedFileSystemDirectoryElement();
-  }
-
-  public static org.tensorflow.proto.util.MemmappedFileSystemDirectoryElement getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<MemmappedFileSystemDirectoryElement>
-      PARSER = new com.google.protobuf.AbstractParser<MemmappedFileSystemDirectoryElement>() {
-    @java.lang.Override
-    public MemmappedFileSystemDirectoryElement parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new MemmappedFileSystemDirectoryElement(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<MemmappedFileSystemDirectoryElement> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<MemmappedFileSystemDirectoryElement> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.util.MemmappedFileSystemDirectoryElement getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/MemmappedFileSystemDirectoryElementOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/MemmappedFileSystemDirectoryElementOrBuilder.java
deleted file mode 100644
index 93fd60f2faa..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/MemmappedFileSystemDirectoryElementOrBuilder.java
+++ /dev/null
@@ -1,29 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/util/memmapped_file_system.proto
-
-package org.tensorflow.proto.util;
-
-public interface MemmappedFileSystemDirectoryElementOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.MemmappedFileSystemDirectoryElement)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <code>uint64 offset = 1;</code>
-   */
-  long getOffset();
-
-  /**
-   * <code>string name = 2;</code>
-   */
-  java.lang.String getName();
-  /**
-   * <code>string name = 2;</code>
-   */
-  com.google.protobuf.ByteString
-      getNameBytes();
-
-  /**
-   * <code>uint64 length = 3;</code>
-   */
-  long getLength();
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/MemmappedFileSystemDirectoryOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/MemmappedFileSystemDirectoryOrBuilder.java
deleted file mode 100644
index c19c30645f5..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/MemmappedFileSystemDirectoryOrBuilder.java
+++ /dev/null
@@ -1,33 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/util/memmapped_file_system.proto
-
-package org.tensorflow.proto.util;
-
-public interface MemmappedFileSystemDirectoryOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.MemmappedFileSystemDirectory)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <code>repeated .tensorflow.MemmappedFileSystemDirectoryElement element = 1;</code>
-   */
-  java.util.List<org.tensorflow.proto.util.MemmappedFileSystemDirectoryElement> 
-      getElementList();
-  /**
-   * <code>repeated .tensorflow.MemmappedFileSystemDirectoryElement element = 1;</code>
-   */
-  org.tensorflow.proto.util.MemmappedFileSystemDirectoryElement getElement(int index);
-  /**
-   * <code>repeated .tensorflow.MemmappedFileSystemDirectoryElement element = 1;</code>
-   */
-  int getElementCount();
-  /**
-   * <code>repeated .tensorflow.MemmappedFileSystemDirectoryElement element = 1;</code>
-   */
-  java.util.List<? extends org.tensorflow.proto.util.MemmappedFileSystemDirectoryElementOrBuilder> 
-      getElementOrBuilderList();
-  /**
-   * <code>repeated .tensorflow.MemmappedFileSystemDirectoryElement element = 1;</code>
-   */
-  org.tensorflow.proto.util.MemmappedFileSystemDirectoryElementOrBuilder getElementOrBuilder(
-      int index);
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/MemmappedFileSystemProtos.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/MemmappedFileSystemProtos.java
deleted file mode 100644
index 2e2809cce45..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/MemmappedFileSystemProtos.java
+++ /dev/null
@@ -1,64 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/util/memmapped_file_system.proto
-
-package org.tensorflow.proto.util;
-
-public final class MemmappedFileSystemProtos {
-  private MemmappedFileSystemProtos() {}
-  public static void registerAllExtensions(
-      com.google.protobuf.ExtensionRegistryLite registry) {
-  }
-
-  public static void registerAllExtensions(
-      com.google.protobuf.ExtensionRegistry registry) {
-    registerAllExtensions(
-        (com.google.protobuf.ExtensionRegistryLite) registry);
-  }
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_MemmappedFileSystemDirectoryElement_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_MemmappedFileSystemDirectoryElement_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_MemmappedFileSystemDirectory_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_MemmappedFileSystemDirectory_fieldAccessorTable;
-
-  public static com.google.protobuf.Descriptors.FileDescriptor
-      getDescriptor() {
-    return descriptor;
-  }
-  private static  com.google.protobuf.Descriptors.FileDescriptor
-      descriptor;
-  static {
-    java.lang.String[] descriptorData = {
-      "\n0tensorflow/core/util/memmapped_file_sy" +
-      "stem.proto\022\ntensorflow\"S\n#MemmappedFileS" +
-      "ystemDirectoryElement\022\016\n\006offset\030\001 \001(\004\022\014\n" +
-      "\004name\030\002 \001(\t\022\016\n\006length\030\003 \001(\004\"`\n\034Memmapped" +
-      "FileSystemDirectory\022@\n\007element\030\001 \003(\0132/.t" +
-      "ensorflow.MemmappedFileSystemDirectoryEl" +
-      "ementB;\n\031org.tensorflow.proto.utilB\031Memm" +
-      "appedFileSystemProtosP\001\370\001\001b\006proto3"
-    };
-    descriptor = com.google.protobuf.Descriptors.FileDescriptor
-      .internalBuildGeneratedFileFrom(descriptorData,
-        new com.google.protobuf.Descriptors.FileDescriptor[] {
-        });
-    internal_static_tensorflow_MemmappedFileSystemDirectoryElement_descriptor =
-      getDescriptor().getMessageTypes().get(0);
-    internal_static_tensorflow_MemmappedFileSystemDirectoryElement_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_MemmappedFileSystemDirectoryElement_descriptor,
-        new java.lang.String[] { "Offset", "Name", "Length", });
-    internal_static_tensorflow_MemmappedFileSystemDirectory_descriptor =
-      getDescriptor().getMessageTypes().get(1);
-    internal_static_tensorflow_MemmappedFileSystemDirectory_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_MemmappedFileSystemDirectory_descriptor,
-        new java.lang.String[] { "Element", });
-  }
-
-  // @@protoc_insertion_point(outer_class_scope)
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/MemoryDump.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/MemoryDump.java
deleted file mode 100644
index e7408bfc997..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/MemoryDump.java
+++ /dev/null
@@ -1,1759 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/bfc_memory_map.proto
-
-package org.tensorflow.proto.util;
-
-/**
- * Protobuf type {@code tensorflow.MemoryDump}
- */
-public  final class MemoryDump extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.MemoryDump)
-    MemoryDumpOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use MemoryDump.newBuilder() to construct.
-  private MemoryDump(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private MemoryDump() {
-    allocatorName_ = "";
-    binSummary_ = java.util.Collections.emptyList();
-    chunk_ = java.util.Collections.emptyList();
-    snapShot_ = java.util.Collections.emptyList();
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new MemoryDump();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private MemoryDump(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    int mutable_bitField0_ = 0;
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 10: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            allocatorName_ = s;
-            break;
-          }
-          case 18: {
-            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
-              binSummary_ = new java.util.ArrayList<org.tensorflow.proto.util.BinSummary>();
-              mutable_bitField0_ |= 0x00000001;
-            }
-            binSummary_.add(
-                input.readMessage(org.tensorflow.proto.util.BinSummary.parser(), extensionRegistry));
-            break;
-          }
-          case 26: {
-            if (!((mutable_bitField0_ & 0x00000002) != 0)) {
-              chunk_ = new java.util.ArrayList<org.tensorflow.proto.util.MemChunk>();
-              mutable_bitField0_ |= 0x00000002;
-            }
-            chunk_.add(
-                input.readMessage(org.tensorflow.proto.util.MemChunk.parser(), extensionRegistry));
-            break;
-          }
-          case 34: {
-            if (!((mutable_bitField0_ & 0x00000004) != 0)) {
-              snapShot_ = new java.util.ArrayList<org.tensorflow.proto.util.SnapShot>();
-              mutable_bitField0_ |= 0x00000004;
-            }
-            snapShot_.add(
-                input.readMessage(org.tensorflow.proto.util.SnapShot.parser(), extensionRegistry));
-            break;
-          }
-          case 42: {
-            org.tensorflow.proto.util.MemAllocatorStats.Builder subBuilder = null;
-            if (stats_ != null) {
-              subBuilder = stats_.toBuilder();
-            }
-            stats_ = input.readMessage(org.tensorflow.proto.util.MemAllocatorStats.parser(), extensionRegistry);
-            if (subBuilder != null) {
-              subBuilder.mergeFrom(stats_);
-              stats_ = subBuilder.buildPartial();
-            }
-
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      if (((mutable_bitField0_ & 0x00000001) != 0)) {
-        binSummary_ = java.util.Collections.unmodifiableList(binSummary_);
-      }
-      if (((mutable_bitField0_ & 0x00000002) != 0)) {
-        chunk_ = java.util.Collections.unmodifiableList(chunk_);
-      }
-      if (((mutable_bitField0_ & 0x00000004) != 0)) {
-        snapShot_ = java.util.Collections.unmodifiableList(snapShot_);
-      }
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.util.BfcMemoryMapProtos.internal_static_tensorflow_MemoryDump_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.util.BfcMemoryMapProtos.internal_static_tensorflow_MemoryDump_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.util.MemoryDump.class, org.tensorflow.proto.util.MemoryDump.Builder.class);
-  }
-
-  public static final int ALLOCATOR_NAME_FIELD_NUMBER = 1;
-  private volatile java.lang.Object allocatorName_;
-  /**
-   * <code>string allocator_name = 1;</code>
-   */
-  public java.lang.String getAllocatorName() {
-    java.lang.Object ref = allocatorName_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      allocatorName_ = s;
-      return s;
-    }
-  }
-  /**
-   * <code>string allocator_name = 1;</code>
-   */
-  public com.google.protobuf.ByteString
-      getAllocatorNameBytes() {
-    java.lang.Object ref = allocatorName_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      allocatorName_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int BIN_SUMMARY_FIELD_NUMBER = 2;
-  private java.util.List<org.tensorflow.proto.util.BinSummary> binSummary_;
-  /**
-   * <code>repeated .tensorflow.BinSummary bin_summary = 2;</code>
-   */
-  public java.util.List<org.tensorflow.proto.util.BinSummary> getBinSummaryList() {
-    return binSummary_;
-  }
-  /**
-   * <code>repeated .tensorflow.BinSummary bin_summary = 2;</code>
-   */
-  public java.util.List<? extends org.tensorflow.proto.util.BinSummaryOrBuilder> 
-      getBinSummaryOrBuilderList() {
-    return binSummary_;
-  }
-  /**
-   * <code>repeated .tensorflow.BinSummary bin_summary = 2;</code>
-   */
-  public int getBinSummaryCount() {
-    return binSummary_.size();
-  }
-  /**
-   * <code>repeated .tensorflow.BinSummary bin_summary = 2;</code>
-   */
-  public org.tensorflow.proto.util.BinSummary getBinSummary(int index) {
-    return binSummary_.get(index);
-  }
-  /**
-   * <code>repeated .tensorflow.BinSummary bin_summary = 2;</code>
-   */
-  public org.tensorflow.proto.util.BinSummaryOrBuilder getBinSummaryOrBuilder(
-      int index) {
-    return binSummary_.get(index);
-  }
-
-  public static final int CHUNK_FIELD_NUMBER = 3;
-  private java.util.List<org.tensorflow.proto.util.MemChunk> chunk_;
-  /**
-   * <code>repeated .tensorflow.MemChunk chunk = 3;</code>
-   */
-  public java.util.List<org.tensorflow.proto.util.MemChunk> getChunkList() {
-    return chunk_;
-  }
-  /**
-   * <code>repeated .tensorflow.MemChunk chunk = 3;</code>
-   */
-  public java.util.List<? extends org.tensorflow.proto.util.MemChunkOrBuilder> 
-      getChunkOrBuilderList() {
-    return chunk_;
-  }
-  /**
-   * <code>repeated .tensorflow.MemChunk chunk = 3;</code>
-   */
-  public int getChunkCount() {
-    return chunk_.size();
-  }
-  /**
-   * <code>repeated .tensorflow.MemChunk chunk = 3;</code>
-   */
-  public org.tensorflow.proto.util.MemChunk getChunk(int index) {
-    return chunk_.get(index);
-  }
-  /**
-   * <code>repeated .tensorflow.MemChunk chunk = 3;</code>
-   */
-  public org.tensorflow.proto.util.MemChunkOrBuilder getChunkOrBuilder(
-      int index) {
-    return chunk_.get(index);
-  }
-
-  public static final int SNAP_SHOT_FIELD_NUMBER = 4;
-  private java.util.List<org.tensorflow.proto.util.SnapShot> snapShot_;
-  /**
-   * <code>repeated .tensorflow.SnapShot snap_shot = 4;</code>
-   */
-  public java.util.List<org.tensorflow.proto.util.SnapShot> getSnapShotList() {
-    return snapShot_;
-  }
-  /**
-   * <code>repeated .tensorflow.SnapShot snap_shot = 4;</code>
-   */
-  public java.util.List<? extends org.tensorflow.proto.util.SnapShotOrBuilder> 
-      getSnapShotOrBuilderList() {
-    return snapShot_;
-  }
-  /**
-   * <code>repeated .tensorflow.SnapShot snap_shot = 4;</code>
-   */
-  public int getSnapShotCount() {
-    return snapShot_.size();
-  }
-  /**
-   * <code>repeated .tensorflow.SnapShot snap_shot = 4;</code>
-   */
-  public org.tensorflow.proto.util.SnapShot getSnapShot(int index) {
-    return snapShot_.get(index);
-  }
-  /**
-   * <code>repeated .tensorflow.SnapShot snap_shot = 4;</code>
-   */
-  public org.tensorflow.proto.util.SnapShotOrBuilder getSnapShotOrBuilder(
-      int index) {
-    return snapShot_.get(index);
-  }
-
-  public static final int STATS_FIELD_NUMBER = 5;
-  private org.tensorflow.proto.util.MemAllocatorStats stats_;
-  /**
-   * <code>.tensorflow.MemAllocatorStats stats = 5;</code>
-   */
-  public boolean hasStats() {
-    return stats_ != null;
-  }
-  /**
-   * <code>.tensorflow.MemAllocatorStats stats = 5;</code>
-   */
-  public org.tensorflow.proto.util.MemAllocatorStats getStats() {
-    return stats_ == null ? org.tensorflow.proto.util.MemAllocatorStats.getDefaultInstance() : stats_;
-  }
-  /**
-   * <code>.tensorflow.MemAllocatorStats stats = 5;</code>
-   */
-  public org.tensorflow.proto.util.MemAllocatorStatsOrBuilder getStatsOrBuilder() {
-    return getStats();
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    if (!getAllocatorNameBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 1, allocatorName_);
-    }
-    for (int i = 0; i < binSummary_.size(); i++) {
-      output.writeMessage(2, binSummary_.get(i));
-    }
-    for (int i = 0; i < chunk_.size(); i++) {
-      output.writeMessage(3, chunk_.get(i));
-    }
-    for (int i = 0; i < snapShot_.size(); i++) {
-      output.writeMessage(4, snapShot_.get(i));
-    }
-    if (stats_ != null) {
-      output.writeMessage(5, getStats());
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (!getAllocatorNameBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, allocatorName_);
-    }
-    for (int i = 0; i < binSummary_.size(); i++) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(2, binSummary_.get(i));
-    }
-    for (int i = 0; i < chunk_.size(); i++) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(3, chunk_.get(i));
-    }
-    for (int i = 0; i < snapShot_.size(); i++) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(4, snapShot_.get(i));
-    }
-    if (stats_ != null) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(5, getStats());
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.util.MemoryDump)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.util.MemoryDump other = (org.tensorflow.proto.util.MemoryDump) obj;
-
-    if (!getAllocatorName()
-        .equals(other.getAllocatorName())) return false;
-    if (!getBinSummaryList()
-        .equals(other.getBinSummaryList())) return false;
-    if (!getChunkList()
-        .equals(other.getChunkList())) return false;
-    if (!getSnapShotList()
-        .equals(other.getSnapShotList())) return false;
-    if (hasStats() != other.hasStats()) return false;
-    if (hasStats()) {
-      if (!getStats()
-          .equals(other.getStats())) return false;
-    }
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    hash = (37 * hash) + ALLOCATOR_NAME_FIELD_NUMBER;
-    hash = (53 * hash) + getAllocatorName().hashCode();
-    if (getBinSummaryCount() > 0) {
-      hash = (37 * hash) + BIN_SUMMARY_FIELD_NUMBER;
-      hash = (53 * hash) + getBinSummaryList().hashCode();
-    }
-    if (getChunkCount() > 0) {
-      hash = (37 * hash) + CHUNK_FIELD_NUMBER;
-      hash = (53 * hash) + getChunkList().hashCode();
-    }
-    if (getSnapShotCount() > 0) {
-      hash = (37 * hash) + SNAP_SHOT_FIELD_NUMBER;
-      hash = (53 * hash) + getSnapShotList().hashCode();
-    }
-    if (hasStats()) {
-      hash = (37 * hash) + STATS_FIELD_NUMBER;
-      hash = (53 * hash) + getStats().hashCode();
-    }
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.util.MemoryDump parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.MemoryDump parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.MemoryDump parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.MemoryDump parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.MemoryDump parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.MemoryDump parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.MemoryDump parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.MemoryDump parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.MemoryDump parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.MemoryDump parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.MemoryDump parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.MemoryDump parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.util.MemoryDump prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * Protobuf type {@code tensorflow.MemoryDump}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.MemoryDump)
-      org.tensorflow.proto.util.MemoryDumpOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.util.BfcMemoryMapProtos.internal_static_tensorflow_MemoryDump_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.util.BfcMemoryMapProtos.internal_static_tensorflow_MemoryDump_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.util.MemoryDump.class, org.tensorflow.proto.util.MemoryDump.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.util.MemoryDump.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-        getBinSummaryFieldBuilder();
-        getChunkFieldBuilder();
-        getSnapShotFieldBuilder();
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      allocatorName_ = "";
-
-      if (binSummaryBuilder_ == null) {
-        binSummary_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000001);
-      } else {
-        binSummaryBuilder_.clear();
-      }
-      if (chunkBuilder_ == null) {
-        chunk_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000002);
-      } else {
-        chunkBuilder_.clear();
-      }
-      if (snapShotBuilder_ == null) {
-        snapShot_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000004);
-      } else {
-        snapShotBuilder_.clear();
-      }
-      if (statsBuilder_ == null) {
-        stats_ = null;
-      } else {
-        stats_ = null;
-        statsBuilder_ = null;
-      }
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.util.BfcMemoryMapProtos.internal_static_tensorflow_MemoryDump_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.MemoryDump getDefaultInstanceForType() {
-      return org.tensorflow.proto.util.MemoryDump.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.MemoryDump build() {
-      org.tensorflow.proto.util.MemoryDump result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.MemoryDump buildPartial() {
-      org.tensorflow.proto.util.MemoryDump result = new org.tensorflow.proto.util.MemoryDump(this);
-      int from_bitField0_ = bitField0_;
-      result.allocatorName_ = allocatorName_;
-      if (binSummaryBuilder_ == null) {
-        if (((bitField0_ & 0x00000001) != 0)) {
-          binSummary_ = java.util.Collections.unmodifiableList(binSummary_);
-          bitField0_ = (bitField0_ & ~0x00000001);
-        }
-        result.binSummary_ = binSummary_;
-      } else {
-        result.binSummary_ = binSummaryBuilder_.build();
-      }
-      if (chunkBuilder_ == null) {
-        if (((bitField0_ & 0x00000002) != 0)) {
-          chunk_ = java.util.Collections.unmodifiableList(chunk_);
-          bitField0_ = (bitField0_ & ~0x00000002);
-        }
-        result.chunk_ = chunk_;
-      } else {
-        result.chunk_ = chunkBuilder_.build();
-      }
-      if (snapShotBuilder_ == null) {
-        if (((bitField0_ & 0x00000004) != 0)) {
-          snapShot_ = java.util.Collections.unmodifiableList(snapShot_);
-          bitField0_ = (bitField0_ & ~0x00000004);
-        }
-        result.snapShot_ = snapShot_;
-      } else {
-        result.snapShot_ = snapShotBuilder_.build();
-      }
-      if (statsBuilder_ == null) {
-        result.stats_ = stats_;
-      } else {
-        result.stats_ = statsBuilder_.build();
-      }
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.util.MemoryDump) {
-        return mergeFrom((org.tensorflow.proto.util.MemoryDump)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.util.MemoryDump other) {
-      if (other == org.tensorflow.proto.util.MemoryDump.getDefaultInstance()) return this;
-      if (!other.getAllocatorName().isEmpty()) {
-        allocatorName_ = other.allocatorName_;
-        onChanged();
-      }
-      if (binSummaryBuilder_ == null) {
-        if (!other.binSummary_.isEmpty()) {
-          if (binSummary_.isEmpty()) {
-            binSummary_ = other.binSummary_;
-            bitField0_ = (bitField0_ & ~0x00000001);
-          } else {
-            ensureBinSummaryIsMutable();
-            binSummary_.addAll(other.binSummary_);
-          }
-          onChanged();
-        }
-      } else {
-        if (!other.binSummary_.isEmpty()) {
-          if (binSummaryBuilder_.isEmpty()) {
-            binSummaryBuilder_.dispose();
-            binSummaryBuilder_ = null;
-            binSummary_ = other.binSummary_;
-            bitField0_ = (bitField0_ & ~0x00000001);
-            binSummaryBuilder_ = 
-              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
-                 getBinSummaryFieldBuilder() : null;
-          } else {
-            binSummaryBuilder_.addAllMessages(other.binSummary_);
-          }
-        }
-      }
-      if (chunkBuilder_ == null) {
-        if (!other.chunk_.isEmpty()) {
-          if (chunk_.isEmpty()) {
-            chunk_ = other.chunk_;
-            bitField0_ = (bitField0_ & ~0x00000002);
-          } else {
-            ensureChunkIsMutable();
-            chunk_.addAll(other.chunk_);
-          }
-          onChanged();
-        }
-      } else {
-        if (!other.chunk_.isEmpty()) {
-          if (chunkBuilder_.isEmpty()) {
-            chunkBuilder_.dispose();
-            chunkBuilder_ = null;
-            chunk_ = other.chunk_;
-            bitField0_ = (bitField0_ & ~0x00000002);
-            chunkBuilder_ = 
-              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
-                 getChunkFieldBuilder() : null;
-          } else {
-            chunkBuilder_.addAllMessages(other.chunk_);
-          }
-        }
-      }
-      if (snapShotBuilder_ == null) {
-        if (!other.snapShot_.isEmpty()) {
-          if (snapShot_.isEmpty()) {
-            snapShot_ = other.snapShot_;
-            bitField0_ = (bitField0_ & ~0x00000004);
-          } else {
-            ensureSnapShotIsMutable();
-            snapShot_.addAll(other.snapShot_);
-          }
-          onChanged();
-        }
-      } else {
-        if (!other.snapShot_.isEmpty()) {
-          if (snapShotBuilder_.isEmpty()) {
-            snapShotBuilder_.dispose();
-            snapShotBuilder_ = null;
-            snapShot_ = other.snapShot_;
-            bitField0_ = (bitField0_ & ~0x00000004);
-            snapShotBuilder_ = 
-              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
-                 getSnapShotFieldBuilder() : null;
-          } else {
-            snapShotBuilder_.addAllMessages(other.snapShot_);
-          }
-        }
-      }
-      if (other.hasStats()) {
-        mergeStats(other.getStats());
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.util.MemoryDump parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.util.MemoryDump) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-    private int bitField0_;
-
-    private java.lang.Object allocatorName_ = "";
-    /**
-     * <code>string allocator_name = 1;</code>
-     */
-    public java.lang.String getAllocatorName() {
-      java.lang.Object ref = allocatorName_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        allocatorName_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <code>string allocator_name = 1;</code>
-     */
-    public com.google.protobuf.ByteString
-        getAllocatorNameBytes() {
-      java.lang.Object ref = allocatorName_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        allocatorName_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <code>string allocator_name = 1;</code>
-     */
-    public Builder setAllocatorName(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      allocatorName_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>string allocator_name = 1;</code>
-     */
-    public Builder clearAllocatorName() {
-      
-      allocatorName_ = getDefaultInstance().getAllocatorName();
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>string allocator_name = 1;</code>
-     */
-    public Builder setAllocatorNameBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      allocatorName_ = value;
-      onChanged();
-      return this;
-    }
-
-    private java.util.List<org.tensorflow.proto.util.BinSummary> binSummary_ =
-      java.util.Collections.emptyList();
-    private void ensureBinSummaryIsMutable() {
-      if (!((bitField0_ & 0x00000001) != 0)) {
-        binSummary_ = new java.util.ArrayList<org.tensorflow.proto.util.BinSummary>(binSummary_);
-        bitField0_ |= 0x00000001;
-       }
-    }
-
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.util.BinSummary, org.tensorflow.proto.util.BinSummary.Builder, org.tensorflow.proto.util.BinSummaryOrBuilder> binSummaryBuilder_;
-
-    /**
-     * <code>repeated .tensorflow.BinSummary bin_summary = 2;</code>
-     */
-    public java.util.List<org.tensorflow.proto.util.BinSummary> getBinSummaryList() {
-      if (binSummaryBuilder_ == null) {
-        return java.util.Collections.unmodifiableList(binSummary_);
-      } else {
-        return binSummaryBuilder_.getMessageList();
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.BinSummary bin_summary = 2;</code>
-     */
-    public int getBinSummaryCount() {
-      if (binSummaryBuilder_ == null) {
-        return binSummary_.size();
-      } else {
-        return binSummaryBuilder_.getCount();
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.BinSummary bin_summary = 2;</code>
-     */
-    public org.tensorflow.proto.util.BinSummary getBinSummary(int index) {
-      if (binSummaryBuilder_ == null) {
-        return binSummary_.get(index);
-      } else {
-        return binSummaryBuilder_.getMessage(index);
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.BinSummary bin_summary = 2;</code>
-     */
-    public Builder setBinSummary(
-        int index, org.tensorflow.proto.util.BinSummary value) {
-      if (binSummaryBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureBinSummaryIsMutable();
-        binSummary_.set(index, value);
-        onChanged();
-      } else {
-        binSummaryBuilder_.setMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.BinSummary bin_summary = 2;</code>
-     */
-    public Builder setBinSummary(
-        int index, org.tensorflow.proto.util.BinSummary.Builder builderForValue) {
-      if (binSummaryBuilder_ == null) {
-        ensureBinSummaryIsMutable();
-        binSummary_.set(index, builderForValue.build());
-        onChanged();
-      } else {
-        binSummaryBuilder_.setMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.BinSummary bin_summary = 2;</code>
-     */
-    public Builder addBinSummary(org.tensorflow.proto.util.BinSummary value) {
-      if (binSummaryBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureBinSummaryIsMutable();
-        binSummary_.add(value);
-        onChanged();
-      } else {
-        binSummaryBuilder_.addMessage(value);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.BinSummary bin_summary = 2;</code>
-     */
-    public Builder addBinSummary(
-        int index, org.tensorflow.proto.util.BinSummary value) {
-      if (binSummaryBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureBinSummaryIsMutable();
-        binSummary_.add(index, value);
-        onChanged();
-      } else {
-        binSummaryBuilder_.addMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.BinSummary bin_summary = 2;</code>
-     */
-    public Builder addBinSummary(
-        org.tensorflow.proto.util.BinSummary.Builder builderForValue) {
-      if (binSummaryBuilder_ == null) {
-        ensureBinSummaryIsMutable();
-        binSummary_.add(builderForValue.build());
-        onChanged();
-      } else {
-        binSummaryBuilder_.addMessage(builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.BinSummary bin_summary = 2;</code>
-     */
-    public Builder addBinSummary(
-        int index, org.tensorflow.proto.util.BinSummary.Builder builderForValue) {
-      if (binSummaryBuilder_ == null) {
-        ensureBinSummaryIsMutable();
-        binSummary_.add(index, builderForValue.build());
-        onChanged();
-      } else {
-        binSummaryBuilder_.addMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.BinSummary bin_summary = 2;</code>
-     */
-    public Builder addAllBinSummary(
-        java.lang.Iterable<? extends org.tensorflow.proto.util.BinSummary> values) {
-      if (binSummaryBuilder_ == null) {
-        ensureBinSummaryIsMutable();
-        com.google.protobuf.AbstractMessageLite.Builder.addAll(
-            values, binSummary_);
-        onChanged();
-      } else {
-        binSummaryBuilder_.addAllMessages(values);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.BinSummary bin_summary = 2;</code>
-     */
-    public Builder clearBinSummary() {
-      if (binSummaryBuilder_ == null) {
-        binSummary_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000001);
-        onChanged();
-      } else {
-        binSummaryBuilder_.clear();
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.BinSummary bin_summary = 2;</code>
-     */
-    public Builder removeBinSummary(int index) {
-      if (binSummaryBuilder_ == null) {
-        ensureBinSummaryIsMutable();
-        binSummary_.remove(index);
-        onChanged();
-      } else {
-        binSummaryBuilder_.remove(index);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.BinSummary bin_summary = 2;</code>
-     */
-    public org.tensorflow.proto.util.BinSummary.Builder getBinSummaryBuilder(
-        int index) {
-      return getBinSummaryFieldBuilder().getBuilder(index);
-    }
-    /**
-     * <code>repeated .tensorflow.BinSummary bin_summary = 2;</code>
-     */
-    public org.tensorflow.proto.util.BinSummaryOrBuilder getBinSummaryOrBuilder(
-        int index) {
-      if (binSummaryBuilder_ == null) {
-        return binSummary_.get(index);  } else {
-        return binSummaryBuilder_.getMessageOrBuilder(index);
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.BinSummary bin_summary = 2;</code>
-     */
-    public java.util.List<? extends org.tensorflow.proto.util.BinSummaryOrBuilder> 
-         getBinSummaryOrBuilderList() {
-      if (binSummaryBuilder_ != null) {
-        return binSummaryBuilder_.getMessageOrBuilderList();
-      } else {
-        return java.util.Collections.unmodifiableList(binSummary_);
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.BinSummary bin_summary = 2;</code>
-     */
-    public org.tensorflow.proto.util.BinSummary.Builder addBinSummaryBuilder() {
-      return getBinSummaryFieldBuilder().addBuilder(
-          org.tensorflow.proto.util.BinSummary.getDefaultInstance());
-    }
-    /**
-     * <code>repeated .tensorflow.BinSummary bin_summary = 2;</code>
-     */
-    public org.tensorflow.proto.util.BinSummary.Builder addBinSummaryBuilder(
-        int index) {
-      return getBinSummaryFieldBuilder().addBuilder(
-          index, org.tensorflow.proto.util.BinSummary.getDefaultInstance());
-    }
-    /**
-     * <code>repeated .tensorflow.BinSummary bin_summary = 2;</code>
-     */
-    public java.util.List<org.tensorflow.proto.util.BinSummary.Builder> 
-         getBinSummaryBuilderList() {
-      return getBinSummaryFieldBuilder().getBuilderList();
-    }
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.util.BinSummary, org.tensorflow.proto.util.BinSummary.Builder, org.tensorflow.proto.util.BinSummaryOrBuilder> 
-        getBinSummaryFieldBuilder() {
-      if (binSummaryBuilder_ == null) {
-        binSummaryBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
-            org.tensorflow.proto.util.BinSummary, org.tensorflow.proto.util.BinSummary.Builder, org.tensorflow.proto.util.BinSummaryOrBuilder>(
-                binSummary_,
-                ((bitField0_ & 0x00000001) != 0),
-                getParentForChildren(),
-                isClean());
-        binSummary_ = null;
-      }
-      return binSummaryBuilder_;
-    }
-
-    private java.util.List<org.tensorflow.proto.util.MemChunk> chunk_ =
-      java.util.Collections.emptyList();
-    private void ensureChunkIsMutable() {
-      if (!((bitField0_ & 0x00000002) != 0)) {
-        chunk_ = new java.util.ArrayList<org.tensorflow.proto.util.MemChunk>(chunk_);
-        bitField0_ |= 0x00000002;
-       }
-    }
-
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.util.MemChunk, org.tensorflow.proto.util.MemChunk.Builder, org.tensorflow.proto.util.MemChunkOrBuilder> chunkBuilder_;
-
-    /**
-     * <code>repeated .tensorflow.MemChunk chunk = 3;</code>
-     */
-    public java.util.List<org.tensorflow.proto.util.MemChunk> getChunkList() {
-      if (chunkBuilder_ == null) {
-        return java.util.Collections.unmodifiableList(chunk_);
-      } else {
-        return chunkBuilder_.getMessageList();
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.MemChunk chunk = 3;</code>
-     */
-    public int getChunkCount() {
-      if (chunkBuilder_ == null) {
-        return chunk_.size();
-      } else {
-        return chunkBuilder_.getCount();
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.MemChunk chunk = 3;</code>
-     */
-    public org.tensorflow.proto.util.MemChunk getChunk(int index) {
-      if (chunkBuilder_ == null) {
-        return chunk_.get(index);
-      } else {
-        return chunkBuilder_.getMessage(index);
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.MemChunk chunk = 3;</code>
-     */
-    public Builder setChunk(
-        int index, org.tensorflow.proto.util.MemChunk value) {
-      if (chunkBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureChunkIsMutable();
-        chunk_.set(index, value);
-        onChanged();
-      } else {
-        chunkBuilder_.setMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.MemChunk chunk = 3;</code>
-     */
-    public Builder setChunk(
-        int index, org.tensorflow.proto.util.MemChunk.Builder builderForValue) {
-      if (chunkBuilder_ == null) {
-        ensureChunkIsMutable();
-        chunk_.set(index, builderForValue.build());
-        onChanged();
-      } else {
-        chunkBuilder_.setMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.MemChunk chunk = 3;</code>
-     */
-    public Builder addChunk(org.tensorflow.proto.util.MemChunk value) {
-      if (chunkBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureChunkIsMutable();
-        chunk_.add(value);
-        onChanged();
-      } else {
-        chunkBuilder_.addMessage(value);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.MemChunk chunk = 3;</code>
-     */
-    public Builder addChunk(
-        int index, org.tensorflow.proto.util.MemChunk value) {
-      if (chunkBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureChunkIsMutable();
-        chunk_.add(index, value);
-        onChanged();
-      } else {
-        chunkBuilder_.addMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.MemChunk chunk = 3;</code>
-     */
-    public Builder addChunk(
-        org.tensorflow.proto.util.MemChunk.Builder builderForValue) {
-      if (chunkBuilder_ == null) {
-        ensureChunkIsMutable();
-        chunk_.add(builderForValue.build());
-        onChanged();
-      } else {
-        chunkBuilder_.addMessage(builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.MemChunk chunk = 3;</code>
-     */
-    public Builder addChunk(
-        int index, org.tensorflow.proto.util.MemChunk.Builder builderForValue) {
-      if (chunkBuilder_ == null) {
-        ensureChunkIsMutable();
-        chunk_.add(index, builderForValue.build());
-        onChanged();
-      } else {
-        chunkBuilder_.addMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.MemChunk chunk = 3;</code>
-     */
-    public Builder addAllChunk(
-        java.lang.Iterable<? extends org.tensorflow.proto.util.MemChunk> values) {
-      if (chunkBuilder_ == null) {
-        ensureChunkIsMutable();
-        com.google.protobuf.AbstractMessageLite.Builder.addAll(
-            values, chunk_);
-        onChanged();
-      } else {
-        chunkBuilder_.addAllMessages(values);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.MemChunk chunk = 3;</code>
-     */
-    public Builder clearChunk() {
-      if (chunkBuilder_ == null) {
-        chunk_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000002);
-        onChanged();
-      } else {
-        chunkBuilder_.clear();
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.MemChunk chunk = 3;</code>
-     */
-    public Builder removeChunk(int index) {
-      if (chunkBuilder_ == null) {
-        ensureChunkIsMutable();
-        chunk_.remove(index);
-        onChanged();
-      } else {
-        chunkBuilder_.remove(index);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.MemChunk chunk = 3;</code>
-     */
-    public org.tensorflow.proto.util.MemChunk.Builder getChunkBuilder(
-        int index) {
-      return getChunkFieldBuilder().getBuilder(index);
-    }
-    /**
-     * <code>repeated .tensorflow.MemChunk chunk = 3;</code>
-     */
-    public org.tensorflow.proto.util.MemChunkOrBuilder getChunkOrBuilder(
-        int index) {
-      if (chunkBuilder_ == null) {
-        return chunk_.get(index);  } else {
-        return chunkBuilder_.getMessageOrBuilder(index);
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.MemChunk chunk = 3;</code>
-     */
-    public java.util.List<? extends org.tensorflow.proto.util.MemChunkOrBuilder> 
-         getChunkOrBuilderList() {
-      if (chunkBuilder_ != null) {
-        return chunkBuilder_.getMessageOrBuilderList();
-      } else {
-        return java.util.Collections.unmodifiableList(chunk_);
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.MemChunk chunk = 3;</code>
-     */
-    public org.tensorflow.proto.util.MemChunk.Builder addChunkBuilder() {
-      return getChunkFieldBuilder().addBuilder(
-          org.tensorflow.proto.util.MemChunk.getDefaultInstance());
-    }
-    /**
-     * <code>repeated .tensorflow.MemChunk chunk = 3;</code>
-     */
-    public org.tensorflow.proto.util.MemChunk.Builder addChunkBuilder(
-        int index) {
-      return getChunkFieldBuilder().addBuilder(
-          index, org.tensorflow.proto.util.MemChunk.getDefaultInstance());
-    }
-    /**
-     * <code>repeated .tensorflow.MemChunk chunk = 3;</code>
-     */
-    public java.util.List<org.tensorflow.proto.util.MemChunk.Builder> 
-         getChunkBuilderList() {
-      return getChunkFieldBuilder().getBuilderList();
-    }
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.util.MemChunk, org.tensorflow.proto.util.MemChunk.Builder, org.tensorflow.proto.util.MemChunkOrBuilder> 
-        getChunkFieldBuilder() {
-      if (chunkBuilder_ == null) {
-        chunkBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
-            org.tensorflow.proto.util.MemChunk, org.tensorflow.proto.util.MemChunk.Builder, org.tensorflow.proto.util.MemChunkOrBuilder>(
-                chunk_,
-                ((bitField0_ & 0x00000002) != 0),
-                getParentForChildren(),
-                isClean());
-        chunk_ = null;
-      }
-      return chunkBuilder_;
-    }
-
-    private java.util.List<org.tensorflow.proto.util.SnapShot> snapShot_ =
-      java.util.Collections.emptyList();
-    private void ensureSnapShotIsMutable() {
-      if (!((bitField0_ & 0x00000004) != 0)) {
-        snapShot_ = new java.util.ArrayList<org.tensorflow.proto.util.SnapShot>(snapShot_);
-        bitField0_ |= 0x00000004;
-       }
-    }
-
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.util.SnapShot, org.tensorflow.proto.util.SnapShot.Builder, org.tensorflow.proto.util.SnapShotOrBuilder> snapShotBuilder_;
-
-    /**
-     * <code>repeated .tensorflow.SnapShot snap_shot = 4;</code>
-     */
-    public java.util.List<org.tensorflow.proto.util.SnapShot> getSnapShotList() {
-      if (snapShotBuilder_ == null) {
-        return java.util.Collections.unmodifiableList(snapShot_);
-      } else {
-        return snapShotBuilder_.getMessageList();
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.SnapShot snap_shot = 4;</code>
-     */
-    public int getSnapShotCount() {
-      if (snapShotBuilder_ == null) {
-        return snapShot_.size();
-      } else {
-        return snapShotBuilder_.getCount();
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.SnapShot snap_shot = 4;</code>
-     */
-    public org.tensorflow.proto.util.SnapShot getSnapShot(int index) {
-      if (snapShotBuilder_ == null) {
-        return snapShot_.get(index);
-      } else {
-        return snapShotBuilder_.getMessage(index);
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.SnapShot snap_shot = 4;</code>
-     */
-    public Builder setSnapShot(
-        int index, org.tensorflow.proto.util.SnapShot value) {
-      if (snapShotBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureSnapShotIsMutable();
-        snapShot_.set(index, value);
-        onChanged();
-      } else {
-        snapShotBuilder_.setMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.SnapShot snap_shot = 4;</code>
-     */
-    public Builder setSnapShot(
-        int index, org.tensorflow.proto.util.SnapShot.Builder builderForValue) {
-      if (snapShotBuilder_ == null) {
-        ensureSnapShotIsMutable();
-        snapShot_.set(index, builderForValue.build());
-        onChanged();
-      } else {
-        snapShotBuilder_.setMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.SnapShot snap_shot = 4;</code>
-     */
-    public Builder addSnapShot(org.tensorflow.proto.util.SnapShot value) {
-      if (snapShotBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureSnapShotIsMutable();
-        snapShot_.add(value);
-        onChanged();
-      } else {
-        snapShotBuilder_.addMessage(value);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.SnapShot snap_shot = 4;</code>
-     */
-    public Builder addSnapShot(
-        int index, org.tensorflow.proto.util.SnapShot value) {
-      if (snapShotBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureSnapShotIsMutable();
-        snapShot_.add(index, value);
-        onChanged();
-      } else {
-        snapShotBuilder_.addMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.SnapShot snap_shot = 4;</code>
-     */
-    public Builder addSnapShot(
-        org.tensorflow.proto.util.SnapShot.Builder builderForValue) {
-      if (snapShotBuilder_ == null) {
-        ensureSnapShotIsMutable();
-        snapShot_.add(builderForValue.build());
-        onChanged();
-      } else {
-        snapShotBuilder_.addMessage(builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.SnapShot snap_shot = 4;</code>
-     */
-    public Builder addSnapShot(
-        int index, org.tensorflow.proto.util.SnapShot.Builder builderForValue) {
-      if (snapShotBuilder_ == null) {
-        ensureSnapShotIsMutable();
-        snapShot_.add(index, builderForValue.build());
-        onChanged();
-      } else {
-        snapShotBuilder_.addMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.SnapShot snap_shot = 4;</code>
-     */
-    public Builder addAllSnapShot(
-        java.lang.Iterable<? extends org.tensorflow.proto.util.SnapShot> values) {
-      if (snapShotBuilder_ == null) {
-        ensureSnapShotIsMutable();
-        com.google.protobuf.AbstractMessageLite.Builder.addAll(
-            values, snapShot_);
-        onChanged();
-      } else {
-        snapShotBuilder_.addAllMessages(values);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.SnapShot snap_shot = 4;</code>
-     */
-    public Builder clearSnapShot() {
-      if (snapShotBuilder_ == null) {
-        snapShot_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000004);
-        onChanged();
-      } else {
-        snapShotBuilder_.clear();
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.SnapShot snap_shot = 4;</code>
-     */
-    public Builder removeSnapShot(int index) {
-      if (snapShotBuilder_ == null) {
-        ensureSnapShotIsMutable();
-        snapShot_.remove(index);
-        onChanged();
-      } else {
-        snapShotBuilder_.remove(index);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.SnapShot snap_shot = 4;</code>
-     */
-    public org.tensorflow.proto.util.SnapShot.Builder getSnapShotBuilder(
-        int index) {
-      return getSnapShotFieldBuilder().getBuilder(index);
-    }
-    /**
-     * <code>repeated .tensorflow.SnapShot snap_shot = 4;</code>
-     */
-    public org.tensorflow.proto.util.SnapShotOrBuilder getSnapShotOrBuilder(
-        int index) {
-      if (snapShotBuilder_ == null) {
-        return snapShot_.get(index);  } else {
-        return snapShotBuilder_.getMessageOrBuilder(index);
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.SnapShot snap_shot = 4;</code>
-     */
-    public java.util.List<? extends org.tensorflow.proto.util.SnapShotOrBuilder> 
-         getSnapShotOrBuilderList() {
-      if (snapShotBuilder_ != null) {
-        return snapShotBuilder_.getMessageOrBuilderList();
-      } else {
-        return java.util.Collections.unmodifiableList(snapShot_);
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.SnapShot snap_shot = 4;</code>
-     */
-    public org.tensorflow.proto.util.SnapShot.Builder addSnapShotBuilder() {
-      return getSnapShotFieldBuilder().addBuilder(
-          org.tensorflow.proto.util.SnapShot.getDefaultInstance());
-    }
-    /**
-     * <code>repeated .tensorflow.SnapShot snap_shot = 4;</code>
-     */
-    public org.tensorflow.proto.util.SnapShot.Builder addSnapShotBuilder(
-        int index) {
-      return getSnapShotFieldBuilder().addBuilder(
-          index, org.tensorflow.proto.util.SnapShot.getDefaultInstance());
-    }
-    /**
-     * <code>repeated .tensorflow.SnapShot snap_shot = 4;</code>
-     */
-    public java.util.List<org.tensorflow.proto.util.SnapShot.Builder> 
-         getSnapShotBuilderList() {
-      return getSnapShotFieldBuilder().getBuilderList();
-    }
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.util.SnapShot, org.tensorflow.proto.util.SnapShot.Builder, org.tensorflow.proto.util.SnapShotOrBuilder> 
-        getSnapShotFieldBuilder() {
-      if (snapShotBuilder_ == null) {
-        snapShotBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
-            org.tensorflow.proto.util.SnapShot, org.tensorflow.proto.util.SnapShot.Builder, org.tensorflow.proto.util.SnapShotOrBuilder>(
-                snapShot_,
-                ((bitField0_ & 0x00000004) != 0),
-                getParentForChildren(),
-                isClean());
-        snapShot_ = null;
-      }
-      return snapShotBuilder_;
-    }
-
-    private org.tensorflow.proto.util.MemAllocatorStats stats_;
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.util.MemAllocatorStats, org.tensorflow.proto.util.MemAllocatorStats.Builder, org.tensorflow.proto.util.MemAllocatorStatsOrBuilder> statsBuilder_;
-    /**
-     * <code>.tensorflow.MemAllocatorStats stats = 5;</code>
-     */
-    public boolean hasStats() {
-      return statsBuilder_ != null || stats_ != null;
-    }
-    /**
-     * <code>.tensorflow.MemAllocatorStats stats = 5;</code>
-     */
-    public org.tensorflow.proto.util.MemAllocatorStats getStats() {
-      if (statsBuilder_ == null) {
-        return stats_ == null ? org.tensorflow.proto.util.MemAllocatorStats.getDefaultInstance() : stats_;
-      } else {
-        return statsBuilder_.getMessage();
-      }
-    }
-    /**
-     * <code>.tensorflow.MemAllocatorStats stats = 5;</code>
-     */
-    public Builder setStats(org.tensorflow.proto.util.MemAllocatorStats value) {
-      if (statsBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        stats_ = value;
-        onChanged();
-      } else {
-        statsBuilder_.setMessage(value);
-      }
-
-      return this;
-    }
-    /**
-     * <code>.tensorflow.MemAllocatorStats stats = 5;</code>
-     */
-    public Builder setStats(
-        org.tensorflow.proto.util.MemAllocatorStats.Builder builderForValue) {
-      if (statsBuilder_ == null) {
-        stats_ = builderForValue.build();
-        onChanged();
-      } else {
-        statsBuilder_.setMessage(builderForValue.build());
-      }
-
-      return this;
-    }
-    /**
-     * <code>.tensorflow.MemAllocatorStats stats = 5;</code>
-     */
-    public Builder mergeStats(org.tensorflow.proto.util.MemAllocatorStats value) {
-      if (statsBuilder_ == null) {
-        if (stats_ != null) {
-          stats_ =
-            org.tensorflow.proto.util.MemAllocatorStats.newBuilder(stats_).mergeFrom(value).buildPartial();
-        } else {
-          stats_ = value;
-        }
-        onChanged();
-      } else {
-        statsBuilder_.mergeFrom(value);
-      }
-
-      return this;
-    }
-    /**
-     * <code>.tensorflow.MemAllocatorStats stats = 5;</code>
-     */
-    public Builder clearStats() {
-      if (statsBuilder_ == null) {
-        stats_ = null;
-        onChanged();
-      } else {
-        stats_ = null;
-        statsBuilder_ = null;
-      }
-
-      return this;
-    }
-    /**
-     * <code>.tensorflow.MemAllocatorStats stats = 5;</code>
-     */
-    public org.tensorflow.proto.util.MemAllocatorStats.Builder getStatsBuilder() {
-      
-      onChanged();
-      return getStatsFieldBuilder().getBuilder();
-    }
-    /**
-     * <code>.tensorflow.MemAllocatorStats stats = 5;</code>
-     */
-    public org.tensorflow.proto.util.MemAllocatorStatsOrBuilder getStatsOrBuilder() {
-      if (statsBuilder_ != null) {
-        return statsBuilder_.getMessageOrBuilder();
-      } else {
-        return stats_ == null ?
-            org.tensorflow.proto.util.MemAllocatorStats.getDefaultInstance() : stats_;
-      }
-    }
-    /**
-     * <code>.tensorflow.MemAllocatorStats stats = 5;</code>
-     */
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.util.MemAllocatorStats, org.tensorflow.proto.util.MemAllocatorStats.Builder, org.tensorflow.proto.util.MemAllocatorStatsOrBuilder> 
-        getStatsFieldBuilder() {
-      if (statsBuilder_ == null) {
-        statsBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-            org.tensorflow.proto.util.MemAllocatorStats, org.tensorflow.proto.util.MemAllocatorStats.Builder, org.tensorflow.proto.util.MemAllocatorStatsOrBuilder>(
-                getStats(),
-                getParentForChildren(),
-                isClean());
-        stats_ = null;
-      }
-      return statsBuilder_;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.MemoryDump)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.MemoryDump)
-  private static final org.tensorflow.proto.util.MemoryDump DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.util.MemoryDump();
-  }
-
-  public static org.tensorflow.proto.util.MemoryDump getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<MemoryDump>
-      PARSER = new com.google.protobuf.AbstractParser<MemoryDump>() {
-    @java.lang.Override
-    public MemoryDump parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new MemoryDump(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<MemoryDump> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<MemoryDump> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.util.MemoryDump getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/MemoryDumpOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/MemoryDumpOrBuilder.java
deleted file mode 100644
index 5d15667734f..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/MemoryDumpOrBuilder.java
+++ /dev/null
@@ -1,104 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/bfc_memory_map.proto
-
-package org.tensorflow.proto.util;
-
-public interface MemoryDumpOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.MemoryDump)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <code>string allocator_name = 1;</code>
-   */
-  java.lang.String getAllocatorName();
-  /**
-   * <code>string allocator_name = 1;</code>
-   */
-  com.google.protobuf.ByteString
-      getAllocatorNameBytes();
-
-  /**
-   * <code>repeated .tensorflow.BinSummary bin_summary = 2;</code>
-   */
-  java.util.List<org.tensorflow.proto.util.BinSummary> 
-      getBinSummaryList();
-  /**
-   * <code>repeated .tensorflow.BinSummary bin_summary = 2;</code>
-   */
-  org.tensorflow.proto.util.BinSummary getBinSummary(int index);
-  /**
-   * <code>repeated .tensorflow.BinSummary bin_summary = 2;</code>
-   */
-  int getBinSummaryCount();
-  /**
-   * <code>repeated .tensorflow.BinSummary bin_summary = 2;</code>
-   */
-  java.util.List<? extends org.tensorflow.proto.util.BinSummaryOrBuilder> 
-      getBinSummaryOrBuilderList();
-  /**
-   * <code>repeated .tensorflow.BinSummary bin_summary = 2;</code>
-   */
-  org.tensorflow.proto.util.BinSummaryOrBuilder getBinSummaryOrBuilder(
-      int index);
-
-  /**
-   * <code>repeated .tensorflow.MemChunk chunk = 3;</code>
-   */
-  java.util.List<org.tensorflow.proto.util.MemChunk> 
-      getChunkList();
-  /**
-   * <code>repeated .tensorflow.MemChunk chunk = 3;</code>
-   */
-  org.tensorflow.proto.util.MemChunk getChunk(int index);
-  /**
-   * <code>repeated .tensorflow.MemChunk chunk = 3;</code>
-   */
-  int getChunkCount();
-  /**
-   * <code>repeated .tensorflow.MemChunk chunk = 3;</code>
-   */
-  java.util.List<? extends org.tensorflow.proto.util.MemChunkOrBuilder> 
-      getChunkOrBuilderList();
-  /**
-   * <code>repeated .tensorflow.MemChunk chunk = 3;</code>
-   */
-  org.tensorflow.proto.util.MemChunkOrBuilder getChunkOrBuilder(
-      int index);
-
-  /**
-   * <code>repeated .tensorflow.SnapShot snap_shot = 4;</code>
-   */
-  java.util.List<org.tensorflow.proto.util.SnapShot> 
-      getSnapShotList();
-  /**
-   * <code>repeated .tensorflow.SnapShot snap_shot = 4;</code>
-   */
-  org.tensorflow.proto.util.SnapShot getSnapShot(int index);
-  /**
-   * <code>repeated .tensorflow.SnapShot snap_shot = 4;</code>
-   */
-  int getSnapShotCount();
-  /**
-   * <code>repeated .tensorflow.SnapShot snap_shot = 4;</code>
-   */
-  java.util.List<? extends org.tensorflow.proto.util.SnapShotOrBuilder> 
-      getSnapShotOrBuilderList();
-  /**
-   * <code>repeated .tensorflow.SnapShot snap_shot = 4;</code>
-   */
-  org.tensorflow.proto.util.SnapShotOrBuilder getSnapShotOrBuilder(
-      int index);
-
-  /**
-   * <code>.tensorflow.MemAllocatorStats stats = 5;</code>
-   */
-  boolean hasStats();
-  /**
-   * <code>.tensorflow.MemAllocatorStats stats = 5;</code>
-   */
-  org.tensorflow.proto.util.MemAllocatorStats getStats();
-  /**
-   * <code>.tensorflow.MemAllocatorStats stats = 5;</code>
-   */
-  org.tensorflow.proto.util.MemAllocatorStatsOrBuilder getStatsOrBuilder();
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/RequestedExitCode.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/RequestedExitCode.java
deleted file mode 100644
index 093cc8f7b5e..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/RequestedExitCode.java
+++ /dev/null
@@ -1,476 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/util/event.proto
-
-package org.tensorflow.proto.util;
-
-/**
- * Protobuf type {@code tensorflow.RequestedExitCode}
- */
-public  final class RequestedExitCode extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.RequestedExitCode)
-    RequestedExitCodeOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use RequestedExitCode.newBuilder() to construct.
-  private RequestedExitCode(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private RequestedExitCode() {
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new RequestedExitCode();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private RequestedExitCode(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 8: {
-
-            exitCode_ = input.readInt32();
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.util.EventProtos.internal_static_tensorflow_RequestedExitCode_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.util.EventProtos.internal_static_tensorflow_RequestedExitCode_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.util.RequestedExitCode.class, org.tensorflow.proto.util.RequestedExitCode.Builder.class);
-  }
-
-  public static final int EXIT_CODE_FIELD_NUMBER = 1;
-  private int exitCode_;
-  /**
-   * <code>int32 exit_code = 1;</code>
-   */
-  public int getExitCode() {
-    return exitCode_;
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    if (exitCode_ != 0) {
-      output.writeInt32(1, exitCode_);
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (exitCode_ != 0) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt32Size(1, exitCode_);
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.util.RequestedExitCode)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.util.RequestedExitCode other = (org.tensorflow.proto.util.RequestedExitCode) obj;
-
-    if (getExitCode()
-        != other.getExitCode()) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    hash = (37 * hash) + EXIT_CODE_FIELD_NUMBER;
-    hash = (53 * hash) + getExitCode();
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.util.RequestedExitCode parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.RequestedExitCode parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.RequestedExitCode parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.RequestedExitCode parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.RequestedExitCode parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.RequestedExitCode parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.RequestedExitCode parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.RequestedExitCode parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.RequestedExitCode parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.RequestedExitCode parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.RequestedExitCode parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.RequestedExitCode parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.util.RequestedExitCode prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * Protobuf type {@code tensorflow.RequestedExitCode}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.RequestedExitCode)
-      org.tensorflow.proto.util.RequestedExitCodeOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.util.EventProtos.internal_static_tensorflow_RequestedExitCode_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.util.EventProtos.internal_static_tensorflow_RequestedExitCode_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.util.RequestedExitCode.class, org.tensorflow.proto.util.RequestedExitCode.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.util.RequestedExitCode.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      exitCode_ = 0;
-
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.util.EventProtos.internal_static_tensorflow_RequestedExitCode_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.RequestedExitCode getDefaultInstanceForType() {
-      return org.tensorflow.proto.util.RequestedExitCode.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.RequestedExitCode build() {
-      org.tensorflow.proto.util.RequestedExitCode result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.RequestedExitCode buildPartial() {
-      org.tensorflow.proto.util.RequestedExitCode result = new org.tensorflow.proto.util.RequestedExitCode(this);
-      result.exitCode_ = exitCode_;
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.util.RequestedExitCode) {
-        return mergeFrom((org.tensorflow.proto.util.RequestedExitCode)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.util.RequestedExitCode other) {
-      if (other == org.tensorflow.proto.util.RequestedExitCode.getDefaultInstance()) return this;
-      if (other.getExitCode() != 0) {
-        setExitCode(other.getExitCode());
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.util.RequestedExitCode parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.util.RequestedExitCode) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-
-    private int exitCode_ ;
-    /**
-     * <code>int32 exit_code = 1;</code>
-     */
-    public int getExitCode() {
-      return exitCode_;
-    }
-    /**
-     * <code>int32 exit_code = 1;</code>
-     */
-    public Builder setExitCode(int value) {
-      
-      exitCode_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>int32 exit_code = 1;</code>
-     */
-    public Builder clearExitCode() {
-      
-      exitCode_ = 0;
-      onChanged();
-      return this;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.RequestedExitCode)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.RequestedExitCode)
-  private static final org.tensorflow.proto.util.RequestedExitCode DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.util.RequestedExitCode();
-  }
-
-  public static org.tensorflow.proto.util.RequestedExitCode getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<RequestedExitCode>
-      PARSER = new com.google.protobuf.AbstractParser<RequestedExitCode>() {
-    @java.lang.Override
-    public RequestedExitCode parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new RequestedExitCode(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<RequestedExitCode> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<RequestedExitCode> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.util.RequestedExitCode getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/RequestedExitCodeOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/RequestedExitCodeOrBuilder.java
deleted file mode 100644
index 260e7609dc7..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/RequestedExitCodeOrBuilder.java
+++ /dev/null
@@ -1,14 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/util/event.proto
-
-package org.tensorflow.proto.util;
-
-public interface RequestedExitCodeOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.RequestedExitCode)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <code>int32 exit_code = 1;</code>
-   */
-  int getExitCode();
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SavedSlice.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SavedSlice.java
deleted file mode 100644
index b1473544ce7..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SavedSlice.java
+++ /dev/null
@@ -1,1073 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/util/saved_tensor_slice.proto
-
-package org.tensorflow.proto.util;
-
-/**
- * <pre>
- * Saved tensor slice: it stores the name of the tensors, the slice, and the
- * raw data.
- * </pre>
- *
- * Protobuf type {@code tensorflow.SavedSlice}
- */
-public  final class SavedSlice extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.SavedSlice)
-    SavedSliceOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use SavedSlice.newBuilder() to construct.
-  private SavedSlice(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private SavedSlice() {
-    name_ = "";
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new SavedSlice();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private SavedSlice(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 10: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            name_ = s;
-            break;
-          }
-          case 18: {
-            org.tensorflow.proto.framework.TensorSliceProto.Builder subBuilder = null;
-            if (slice_ != null) {
-              subBuilder = slice_.toBuilder();
-            }
-            slice_ = input.readMessage(org.tensorflow.proto.framework.TensorSliceProto.parser(), extensionRegistry);
-            if (subBuilder != null) {
-              subBuilder.mergeFrom(slice_);
-              slice_ = subBuilder.buildPartial();
-            }
-
-            break;
-          }
-          case 26: {
-            org.tensorflow.proto.framework.TensorProto.Builder subBuilder = null;
-            if (data_ != null) {
-              subBuilder = data_.toBuilder();
-            }
-            data_ = input.readMessage(org.tensorflow.proto.framework.TensorProto.parser(), extensionRegistry);
-            if (subBuilder != null) {
-              subBuilder.mergeFrom(data_);
-              data_ = subBuilder.buildPartial();
-            }
-
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.util.SavedTensorSliceProtos.internal_static_tensorflow_SavedSlice_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.util.SavedTensorSliceProtos.internal_static_tensorflow_SavedSlice_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.util.SavedSlice.class, org.tensorflow.proto.util.SavedSlice.Builder.class);
-  }
-
-  public static final int NAME_FIELD_NUMBER = 1;
-  private volatile java.lang.Object name_;
-  /**
-   * <pre>
-   * Name of the tensor that this slice belongs to. This must be identical to
-   * the name used to encode the key for this record.
-   * </pre>
-   *
-   * <code>string name = 1;</code>
-   */
-  public java.lang.String getName() {
-    java.lang.Object ref = name_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      name_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * Name of the tensor that this slice belongs to. This must be identical to
-   * the name used to encode the key for this record.
-   * </pre>
-   *
-   * <code>string name = 1;</code>
-   */
-  public com.google.protobuf.ByteString
-      getNameBytes() {
-    java.lang.Object ref = name_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      name_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int SLICE_FIELD_NUMBER = 2;
-  private org.tensorflow.proto.framework.TensorSliceProto slice_;
-  /**
-   * <pre>
-   * Extent of the slice.  Must have one entry for each of the dimension of the
-   * tensor that this slice belongs to.
-   * </pre>
-   *
-   * <code>.tensorflow.TensorSliceProto slice = 2;</code>
-   */
-  public boolean hasSlice() {
-    return slice_ != null;
-  }
-  /**
-   * <pre>
-   * Extent of the slice.  Must have one entry for each of the dimension of the
-   * tensor that this slice belongs to.
-   * </pre>
-   *
-   * <code>.tensorflow.TensorSliceProto slice = 2;</code>
-   */
-  public org.tensorflow.proto.framework.TensorSliceProto getSlice() {
-    return slice_ == null ? org.tensorflow.proto.framework.TensorSliceProto.getDefaultInstance() : slice_;
-  }
-  /**
-   * <pre>
-   * Extent of the slice.  Must have one entry for each of the dimension of the
-   * tensor that this slice belongs to.
-   * </pre>
-   *
-   * <code>.tensorflow.TensorSliceProto slice = 2;</code>
-   */
-  public org.tensorflow.proto.framework.TensorSliceProtoOrBuilder getSliceOrBuilder() {
-    return getSlice();
-  }
-
-  public static final int DATA_FIELD_NUMBER = 3;
-  private org.tensorflow.proto.framework.TensorProto data_;
-  /**
-   * <pre>
-   * The raw data of the slice is stored as a TensorProto. Only raw data are
-   * stored (we don't fill in fields such as dtype or tensor_shape).
-   * </pre>
-   *
-   * <code>.tensorflow.TensorProto data = 3;</code>
-   */
-  public boolean hasData() {
-    return data_ != null;
-  }
-  /**
-   * <pre>
-   * The raw data of the slice is stored as a TensorProto. Only raw data are
-   * stored (we don't fill in fields such as dtype or tensor_shape).
-   * </pre>
-   *
-   * <code>.tensorflow.TensorProto data = 3;</code>
-   */
-  public org.tensorflow.proto.framework.TensorProto getData() {
-    return data_ == null ? org.tensorflow.proto.framework.TensorProto.getDefaultInstance() : data_;
-  }
-  /**
-   * <pre>
-   * The raw data of the slice is stored as a TensorProto. Only raw data are
-   * stored (we don't fill in fields such as dtype or tensor_shape).
-   * </pre>
-   *
-   * <code>.tensorflow.TensorProto data = 3;</code>
-   */
-  public org.tensorflow.proto.framework.TensorProtoOrBuilder getDataOrBuilder() {
-    return getData();
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    if (!getNameBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 1, name_);
-    }
-    if (slice_ != null) {
-      output.writeMessage(2, getSlice());
-    }
-    if (data_ != null) {
-      output.writeMessage(3, getData());
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (!getNameBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, name_);
-    }
-    if (slice_ != null) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(2, getSlice());
-    }
-    if (data_ != null) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(3, getData());
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.util.SavedSlice)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.util.SavedSlice other = (org.tensorflow.proto.util.SavedSlice) obj;
-
-    if (!getName()
-        .equals(other.getName())) return false;
-    if (hasSlice() != other.hasSlice()) return false;
-    if (hasSlice()) {
-      if (!getSlice()
-          .equals(other.getSlice())) return false;
-    }
-    if (hasData() != other.hasData()) return false;
-    if (hasData()) {
-      if (!getData()
-          .equals(other.getData())) return false;
-    }
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    hash = (37 * hash) + NAME_FIELD_NUMBER;
-    hash = (53 * hash) + getName().hashCode();
-    if (hasSlice()) {
-      hash = (37 * hash) + SLICE_FIELD_NUMBER;
-      hash = (53 * hash) + getSlice().hashCode();
-    }
-    if (hasData()) {
-      hash = (37 * hash) + DATA_FIELD_NUMBER;
-      hash = (53 * hash) + getData().hashCode();
-    }
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.util.SavedSlice parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.SavedSlice parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.SavedSlice parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.SavedSlice parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.SavedSlice parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.SavedSlice parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.SavedSlice parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.SavedSlice parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.SavedSlice parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.SavedSlice parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.SavedSlice parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.SavedSlice parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.util.SavedSlice prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * <pre>
-   * Saved tensor slice: it stores the name of the tensors, the slice, and the
-   * raw data.
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.SavedSlice}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.SavedSlice)
-      org.tensorflow.proto.util.SavedSliceOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.util.SavedTensorSliceProtos.internal_static_tensorflow_SavedSlice_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.util.SavedTensorSliceProtos.internal_static_tensorflow_SavedSlice_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.util.SavedSlice.class, org.tensorflow.proto.util.SavedSlice.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.util.SavedSlice.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      name_ = "";
-
-      if (sliceBuilder_ == null) {
-        slice_ = null;
-      } else {
-        slice_ = null;
-        sliceBuilder_ = null;
-      }
-      if (dataBuilder_ == null) {
-        data_ = null;
-      } else {
-        data_ = null;
-        dataBuilder_ = null;
-      }
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.util.SavedTensorSliceProtos.internal_static_tensorflow_SavedSlice_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.SavedSlice getDefaultInstanceForType() {
-      return org.tensorflow.proto.util.SavedSlice.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.SavedSlice build() {
-      org.tensorflow.proto.util.SavedSlice result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.SavedSlice buildPartial() {
-      org.tensorflow.proto.util.SavedSlice result = new org.tensorflow.proto.util.SavedSlice(this);
-      result.name_ = name_;
-      if (sliceBuilder_ == null) {
-        result.slice_ = slice_;
-      } else {
-        result.slice_ = sliceBuilder_.build();
-      }
-      if (dataBuilder_ == null) {
-        result.data_ = data_;
-      } else {
-        result.data_ = dataBuilder_.build();
-      }
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.util.SavedSlice) {
-        return mergeFrom((org.tensorflow.proto.util.SavedSlice)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.util.SavedSlice other) {
-      if (other == org.tensorflow.proto.util.SavedSlice.getDefaultInstance()) return this;
-      if (!other.getName().isEmpty()) {
-        name_ = other.name_;
-        onChanged();
-      }
-      if (other.hasSlice()) {
-        mergeSlice(other.getSlice());
-      }
-      if (other.hasData()) {
-        mergeData(other.getData());
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.util.SavedSlice parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.util.SavedSlice) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-
-    private java.lang.Object name_ = "";
-    /**
-     * <pre>
-     * Name of the tensor that this slice belongs to. This must be identical to
-     * the name used to encode the key for this record.
-     * </pre>
-     *
-     * <code>string name = 1;</code>
-     */
-    public java.lang.String getName() {
-      java.lang.Object ref = name_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        name_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Name of the tensor that this slice belongs to. This must be identical to
-     * the name used to encode the key for this record.
-     * </pre>
-     *
-     * <code>string name = 1;</code>
-     */
-    public com.google.protobuf.ByteString
-        getNameBytes() {
-      java.lang.Object ref = name_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        name_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Name of the tensor that this slice belongs to. This must be identical to
-     * the name used to encode the key for this record.
-     * </pre>
-     *
-     * <code>string name = 1;</code>
-     */
-    public Builder setName(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      name_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Name of the tensor that this slice belongs to. This must be identical to
-     * the name used to encode the key for this record.
-     * </pre>
-     *
-     * <code>string name = 1;</code>
-     */
-    public Builder clearName() {
-      
-      name_ = getDefaultInstance().getName();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Name of the tensor that this slice belongs to. This must be identical to
-     * the name used to encode the key for this record.
-     * </pre>
-     *
-     * <code>string name = 1;</code>
-     */
-    public Builder setNameBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      name_ = value;
-      onChanged();
-      return this;
-    }
-
-    private org.tensorflow.proto.framework.TensorSliceProto slice_;
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.TensorSliceProto, org.tensorflow.proto.framework.TensorSliceProto.Builder, org.tensorflow.proto.framework.TensorSliceProtoOrBuilder> sliceBuilder_;
-    /**
-     * <pre>
-     * Extent of the slice.  Must have one entry for each of the dimension of the
-     * tensor that this slice belongs to.
-     * </pre>
-     *
-     * <code>.tensorflow.TensorSliceProto slice = 2;</code>
-     */
-    public boolean hasSlice() {
-      return sliceBuilder_ != null || slice_ != null;
-    }
-    /**
-     * <pre>
-     * Extent of the slice.  Must have one entry for each of the dimension of the
-     * tensor that this slice belongs to.
-     * </pre>
-     *
-     * <code>.tensorflow.TensorSliceProto slice = 2;</code>
-     */
-    public org.tensorflow.proto.framework.TensorSliceProto getSlice() {
-      if (sliceBuilder_ == null) {
-        return slice_ == null ? org.tensorflow.proto.framework.TensorSliceProto.getDefaultInstance() : slice_;
-      } else {
-        return sliceBuilder_.getMessage();
-      }
-    }
-    /**
-     * <pre>
-     * Extent of the slice.  Must have one entry for each of the dimension of the
-     * tensor that this slice belongs to.
-     * </pre>
-     *
-     * <code>.tensorflow.TensorSliceProto slice = 2;</code>
-     */
-    public Builder setSlice(org.tensorflow.proto.framework.TensorSliceProto value) {
-      if (sliceBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        slice_ = value;
-        onChanged();
-      } else {
-        sliceBuilder_.setMessage(value);
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Extent of the slice.  Must have one entry for each of the dimension of the
-     * tensor that this slice belongs to.
-     * </pre>
-     *
-     * <code>.tensorflow.TensorSliceProto slice = 2;</code>
-     */
-    public Builder setSlice(
-        org.tensorflow.proto.framework.TensorSliceProto.Builder builderForValue) {
-      if (sliceBuilder_ == null) {
-        slice_ = builderForValue.build();
-        onChanged();
-      } else {
-        sliceBuilder_.setMessage(builderForValue.build());
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Extent of the slice.  Must have one entry for each of the dimension of the
-     * tensor that this slice belongs to.
-     * </pre>
-     *
-     * <code>.tensorflow.TensorSliceProto slice = 2;</code>
-     */
-    public Builder mergeSlice(org.tensorflow.proto.framework.TensorSliceProto value) {
-      if (sliceBuilder_ == null) {
-        if (slice_ != null) {
-          slice_ =
-            org.tensorflow.proto.framework.TensorSliceProto.newBuilder(slice_).mergeFrom(value).buildPartial();
-        } else {
-          slice_ = value;
-        }
-        onChanged();
-      } else {
-        sliceBuilder_.mergeFrom(value);
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Extent of the slice.  Must have one entry for each of the dimension of the
-     * tensor that this slice belongs to.
-     * </pre>
-     *
-     * <code>.tensorflow.TensorSliceProto slice = 2;</code>
-     */
-    public Builder clearSlice() {
-      if (sliceBuilder_ == null) {
-        slice_ = null;
-        onChanged();
-      } else {
-        slice_ = null;
-        sliceBuilder_ = null;
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Extent of the slice.  Must have one entry for each of the dimension of the
-     * tensor that this slice belongs to.
-     * </pre>
-     *
-     * <code>.tensorflow.TensorSliceProto slice = 2;</code>
-     */
-    public org.tensorflow.proto.framework.TensorSliceProto.Builder getSliceBuilder() {
-      
-      onChanged();
-      return getSliceFieldBuilder().getBuilder();
-    }
-    /**
-     * <pre>
-     * Extent of the slice.  Must have one entry for each of the dimension of the
-     * tensor that this slice belongs to.
-     * </pre>
-     *
-     * <code>.tensorflow.TensorSliceProto slice = 2;</code>
-     */
-    public org.tensorflow.proto.framework.TensorSliceProtoOrBuilder getSliceOrBuilder() {
-      if (sliceBuilder_ != null) {
-        return sliceBuilder_.getMessageOrBuilder();
-      } else {
-        return slice_ == null ?
-            org.tensorflow.proto.framework.TensorSliceProto.getDefaultInstance() : slice_;
-      }
-    }
-    /**
-     * <pre>
-     * Extent of the slice.  Must have one entry for each of the dimension of the
-     * tensor that this slice belongs to.
-     * </pre>
-     *
-     * <code>.tensorflow.TensorSliceProto slice = 2;</code>
-     */
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.TensorSliceProto, org.tensorflow.proto.framework.TensorSliceProto.Builder, org.tensorflow.proto.framework.TensorSliceProtoOrBuilder> 
-        getSliceFieldBuilder() {
-      if (sliceBuilder_ == null) {
-        sliceBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-            org.tensorflow.proto.framework.TensorSliceProto, org.tensorflow.proto.framework.TensorSliceProto.Builder, org.tensorflow.proto.framework.TensorSliceProtoOrBuilder>(
-                getSlice(),
-                getParentForChildren(),
-                isClean());
-        slice_ = null;
-      }
-      return sliceBuilder_;
-    }
-
-    private org.tensorflow.proto.framework.TensorProto data_;
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.TensorProto, org.tensorflow.proto.framework.TensorProto.Builder, org.tensorflow.proto.framework.TensorProtoOrBuilder> dataBuilder_;
-    /**
-     * <pre>
-     * The raw data of the slice is stored as a TensorProto. Only raw data are
-     * stored (we don't fill in fields such as dtype or tensor_shape).
-     * </pre>
-     *
-     * <code>.tensorflow.TensorProto data = 3;</code>
-     */
-    public boolean hasData() {
-      return dataBuilder_ != null || data_ != null;
-    }
-    /**
-     * <pre>
-     * The raw data of the slice is stored as a TensorProto. Only raw data are
-     * stored (we don't fill in fields such as dtype or tensor_shape).
-     * </pre>
-     *
-     * <code>.tensorflow.TensorProto data = 3;</code>
-     */
-    public org.tensorflow.proto.framework.TensorProto getData() {
-      if (dataBuilder_ == null) {
-        return data_ == null ? org.tensorflow.proto.framework.TensorProto.getDefaultInstance() : data_;
-      } else {
-        return dataBuilder_.getMessage();
-      }
-    }
-    /**
-     * <pre>
-     * The raw data of the slice is stored as a TensorProto. Only raw data are
-     * stored (we don't fill in fields such as dtype or tensor_shape).
-     * </pre>
-     *
-     * <code>.tensorflow.TensorProto data = 3;</code>
-     */
-    public Builder setData(org.tensorflow.proto.framework.TensorProto value) {
-      if (dataBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        data_ = value;
-        onChanged();
-      } else {
-        dataBuilder_.setMessage(value);
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * The raw data of the slice is stored as a TensorProto. Only raw data are
-     * stored (we don't fill in fields such as dtype or tensor_shape).
-     * </pre>
-     *
-     * <code>.tensorflow.TensorProto data = 3;</code>
-     */
-    public Builder setData(
-        org.tensorflow.proto.framework.TensorProto.Builder builderForValue) {
-      if (dataBuilder_ == null) {
-        data_ = builderForValue.build();
-        onChanged();
-      } else {
-        dataBuilder_.setMessage(builderForValue.build());
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * The raw data of the slice is stored as a TensorProto. Only raw data are
-     * stored (we don't fill in fields such as dtype or tensor_shape).
-     * </pre>
-     *
-     * <code>.tensorflow.TensorProto data = 3;</code>
-     */
-    public Builder mergeData(org.tensorflow.proto.framework.TensorProto value) {
-      if (dataBuilder_ == null) {
-        if (data_ != null) {
-          data_ =
-            org.tensorflow.proto.framework.TensorProto.newBuilder(data_).mergeFrom(value).buildPartial();
-        } else {
-          data_ = value;
-        }
-        onChanged();
-      } else {
-        dataBuilder_.mergeFrom(value);
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * The raw data of the slice is stored as a TensorProto. Only raw data are
-     * stored (we don't fill in fields such as dtype or tensor_shape).
-     * </pre>
-     *
-     * <code>.tensorflow.TensorProto data = 3;</code>
-     */
-    public Builder clearData() {
-      if (dataBuilder_ == null) {
-        data_ = null;
-        onChanged();
-      } else {
-        data_ = null;
-        dataBuilder_ = null;
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * The raw data of the slice is stored as a TensorProto. Only raw data are
-     * stored (we don't fill in fields such as dtype or tensor_shape).
-     * </pre>
-     *
-     * <code>.tensorflow.TensorProto data = 3;</code>
-     */
-    public org.tensorflow.proto.framework.TensorProto.Builder getDataBuilder() {
-      
-      onChanged();
-      return getDataFieldBuilder().getBuilder();
-    }
-    /**
-     * <pre>
-     * The raw data of the slice is stored as a TensorProto. Only raw data are
-     * stored (we don't fill in fields such as dtype or tensor_shape).
-     * </pre>
-     *
-     * <code>.tensorflow.TensorProto data = 3;</code>
-     */
-    public org.tensorflow.proto.framework.TensorProtoOrBuilder getDataOrBuilder() {
-      if (dataBuilder_ != null) {
-        return dataBuilder_.getMessageOrBuilder();
-      } else {
-        return data_ == null ?
-            org.tensorflow.proto.framework.TensorProto.getDefaultInstance() : data_;
-      }
-    }
-    /**
-     * <pre>
-     * The raw data of the slice is stored as a TensorProto. Only raw data are
-     * stored (we don't fill in fields such as dtype or tensor_shape).
-     * </pre>
-     *
-     * <code>.tensorflow.TensorProto data = 3;</code>
-     */
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.TensorProto, org.tensorflow.proto.framework.TensorProto.Builder, org.tensorflow.proto.framework.TensorProtoOrBuilder> 
-        getDataFieldBuilder() {
-      if (dataBuilder_ == null) {
-        dataBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-            org.tensorflow.proto.framework.TensorProto, org.tensorflow.proto.framework.TensorProto.Builder, org.tensorflow.proto.framework.TensorProtoOrBuilder>(
-                getData(),
-                getParentForChildren(),
-                isClean());
-        data_ = null;
-      }
-      return dataBuilder_;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.SavedSlice)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.SavedSlice)
-  private static final org.tensorflow.proto.util.SavedSlice DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.util.SavedSlice();
-  }
-
-  public static org.tensorflow.proto.util.SavedSlice getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<SavedSlice>
-      PARSER = new com.google.protobuf.AbstractParser<SavedSlice>() {
-    @java.lang.Override
-    public SavedSlice parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new SavedSlice(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<SavedSlice> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<SavedSlice> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.util.SavedSlice getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SavedSliceMeta.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SavedSliceMeta.java
deleted file mode 100644
index 884792faec0..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SavedSliceMeta.java
+++ /dev/null
@@ -1,1364 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/util/saved_tensor_slice.proto
-
-package org.tensorflow.proto.util;
-
-/**
- * <pre>
- * Metadata describing the set of slices of the same tensor saved in a
- * checkpoint file.
- * </pre>
- *
- * Protobuf type {@code tensorflow.SavedSliceMeta}
- */
-public  final class SavedSliceMeta extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.SavedSliceMeta)
-    SavedSliceMetaOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use SavedSliceMeta.newBuilder() to construct.
-  private SavedSliceMeta(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private SavedSliceMeta() {
-    name_ = "";
-    type_ = 0;
-    slice_ = java.util.Collections.emptyList();
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new SavedSliceMeta();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private SavedSliceMeta(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    int mutable_bitField0_ = 0;
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 10: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            name_ = s;
-            break;
-          }
-          case 18: {
-            org.tensorflow.proto.framework.TensorShapeProto.Builder subBuilder = null;
-            if (shape_ != null) {
-              subBuilder = shape_.toBuilder();
-            }
-            shape_ = input.readMessage(org.tensorflow.proto.framework.TensorShapeProto.parser(), extensionRegistry);
-            if (subBuilder != null) {
-              subBuilder.mergeFrom(shape_);
-              shape_ = subBuilder.buildPartial();
-            }
-
-            break;
-          }
-          case 24: {
-            int rawValue = input.readEnum();
-
-            type_ = rawValue;
-            break;
-          }
-          case 34: {
-            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
-              slice_ = new java.util.ArrayList<org.tensorflow.proto.framework.TensorSliceProto>();
-              mutable_bitField0_ |= 0x00000001;
-            }
-            slice_.add(
-                input.readMessage(org.tensorflow.proto.framework.TensorSliceProto.parser(), extensionRegistry));
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      if (((mutable_bitField0_ & 0x00000001) != 0)) {
-        slice_ = java.util.Collections.unmodifiableList(slice_);
-      }
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.util.SavedTensorSliceProtos.internal_static_tensorflow_SavedSliceMeta_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.util.SavedTensorSliceProtos.internal_static_tensorflow_SavedSliceMeta_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.util.SavedSliceMeta.class, org.tensorflow.proto.util.SavedSliceMeta.Builder.class);
-  }
-
-  public static final int NAME_FIELD_NUMBER = 1;
-  private volatile java.lang.Object name_;
-  /**
-   * <pre>
-   * Name of the tensor.
-   * </pre>
-   *
-   * <code>string name = 1;</code>
-   */
-  public java.lang.String getName() {
-    java.lang.Object ref = name_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      name_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * Name of the tensor.
-   * </pre>
-   *
-   * <code>string name = 1;</code>
-   */
-  public com.google.protobuf.ByteString
-      getNameBytes() {
-    java.lang.Object ref = name_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      name_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int SHAPE_FIELD_NUMBER = 2;
-  private org.tensorflow.proto.framework.TensorShapeProto shape_;
-  /**
-   * <pre>
-   * Shape of the tensor
-   * </pre>
-   *
-   * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-   */
-  public boolean hasShape() {
-    return shape_ != null;
-  }
-  /**
-   * <pre>
-   * Shape of the tensor
-   * </pre>
-   *
-   * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-   */
-  public org.tensorflow.proto.framework.TensorShapeProto getShape() {
-    return shape_ == null ? org.tensorflow.proto.framework.TensorShapeProto.getDefaultInstance() : shape_;
-  }
-  /**
-   * <pre>
-   * Shape of the tensor
-   * </pre>
-   *
-   * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-   */
-  public org.tensorflow.proto.framework.TensorShapeProtoOrBuilder getShapeOrBuilder() {
-    return getShape();
-  }
-
-  public static final int TYPE_FIELD_NUMBER = 3;
-  private int type_;
-  /**
-   * <pre>
-   * Type of the tensor
-   * </pre>
-   *
-   * <code>.tensorflow.DataType type = 3;</code>
-   */
-  public int getTypeValue() {
-    return type_;
-  }
-  /**
-   * <pre>
-   * Type of the tensor
-   * </pre>
-   *
-   * <code>.tensorflow.DataType type = 3;</code>
-   */
-  public org.tensorflow.proto.framework.DataType getType() {
-    @SuppressWarnings("deprecation")
-    org.tensorflow.proto.framework.DataType result = org.tensorflow.proto.framework.DataType.valueOf(type_);
-    return result == null ? org.tensorflow.proto.framework.DataType.UNRECOGNIZED : result;
-  }
-
-  public static final int SLICE_FIELD_NUMBER = 4;
-  private java.util.List<org.tensorflow.proto.framework.TensorSliceProto> slice_;
-  /**
-   * <pre>
-   * Explicit list of slices saved in the checkpoint file.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.TensorSliceProto slice = 4;</code>
-   */
-  public java.util.List<org.tensorflow.proto.framework.TensorSliceProto> getSliceList() {
-    return slice_;
-  }
-  /**
-   * <pre>
-   * Explicit list of slices saved in the checkpoint file.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.TensorSliceProto slice = 4;</code>
-   */
-  public java.util.List<? extends org.tensorflow.proto.framework.TensorSliceProtoOrBuilder> 
-      getSliceOrBuilderList() {
-    return slice_;
-  }
-  /**
-   * <pre>
-   * Explicit list of slices saved in the checkpoint file.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.TensorSliceProto slice = 4;</code>
-   */
-  public int getSliceCount() {
-    return slice_.size();
-  }
-  /**
-   * <pre>
-   * Explicit list of slices saved in the checkpoint file.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.TensorSliceProto slice = 4;</code>
-   */
-  public org.tensorflow.proto.framework.TensorSliceProto getSlice(int index) {
-    return slice_.get(index);
-  }
-  /**
-   * <pre>
-   * Explicit list of slices saved in the checkpoint file.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.TensorSliceProto slice = 4;</code>
-   */
-  public org.tensorflow.proto.framework.TensorSliceProtoOrBuilder getSliceOrBuilder(
-      int index) {
-    return slice_.get(index);
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    if (!getNameBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 1, name_);
-    }
-    if (shape_ != null) {
-      output.writeMessage(2, getShape());
-    }
-    if (type_ != org.tensorflow.proto.framework.DataType.DT_INVALID.getNumber()) {
-      output.writeEnum(3, type_);
-    }
-    for (int i = 0; i < slice_.size(); i++) {
-      output.writeMessage(4, slice_.get(i));
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (!getNameBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, name_);
-    }
-    if (shape_ != null) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(2, getShape());
-    }
-    if (type_ != org.tensorflow.proto.framework.DataType.DT_INVALID.getNumber()) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeEnumSize(3, type_);
-    }
-    for (int i = 0; i < slice_.size(); i++) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(4, slice_.get(i));
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.util.SavedSliceMeta)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.util.SavedSliceMeta other = (org.tensorflow.proto.util.SavedSliceMeta) obj;
-
-    if (!getName()
-        .equals(other.getName())) return false;
-    if (hasShape() != other.hasShape()) return false;
-    if (hasShape()) {
-      if (!getShape()
-          .equals(other.getShape())) return false;
-    }
-    if (type_ != other.type_) return false;
-    if (!getSliceList()
-        .equals(other.getSliceList())) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    hash = (37 * hash) + NAME_FIELD_NUMBER;
-    hash = (53 * hash) + getName().hashCode();
-    if (hasShape()) {
-      hash = (37 * hash) + SHAPE_FIELD_NUMBER;
-      hash = (53 * hash) + getShape().hashCode();
-    }
-    hash = (37 * hash) + TYPE_FIELD_NUMBER;
-    hash = (53 * hash) + type_;
-    if (getSliceCount() > 0) {
-      hash = (37 * hash) + SLICE_FIELD_NUMBER;
-      hash = (53 * hash) + getSliceList().hashCode();
-    }
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.util.SavedSliceMeta parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.SavedSliceMeta parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.SavedSliceMeta parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.SavedSliceMeta parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.SavedSliceMeta parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.SavedSliceMeta parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.SavedSliceMeta parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.SavedSliceMeta parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.SavedSliceMeta parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.SavedSliceMeta parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.SavedSliceMeta parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.SavedSliceMeta parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.util.SavedSliceMeta prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * <pre>
-   * Metadata describing the set of slices of the same tensor saved in a
-   * checkpoint file.
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.SavedSliceMeta}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.SavedSliceMeta)
-      org.tensorflow.proto.util.SavedSliceMetaOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.util.SavedTensorSliceProtos.internal_static_tensorflow_SavedSliceMeta_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.util.SavedTensorSliceProtos.internal_static_tensorflow_SavedSliceMeta_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.util.SavedSliceMeta.class, org.tensorflow.proto.util.SavedSliceMeta.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.util.SavedSliceMeta.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-        getSliceFieldBuilder();
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      name_ = "";
-
-      if (shapeBuilder_ == null) {
-        shape_ = null;
-      } else {
-        shape_ = null;
-        shapeBuilder_ = null;
-      }
-      type_ = 0;
-
-      if (sliceBuilder_ == null) {
-        slice_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000001);
-      } else {
-        sliceBuilder_.clear();
-      }
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.util.SavedTensorSliceProtos.internal_static_tensorflow_SavedSliceMeta_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.SavedSliceMeta getDefaultInstanceForType() {
-      return org.tensorflow.proto.util.SavedSliceMeta.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.SavedSliceMeta build() {
-      org.tensorflow.proto.util.SavedSliceMeta result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.SavedSliceMeta buildPartial() {
-      org.tensorflow.proto.util.SavedSliceMeta result = new org.tensorflow.proto.util.SavedSliceMeta(this);
-      int from_bitField0_ = bitField0_;
-      result.name_ = name_;
-      if (shapeBuilder_ == null) {
-        result.shape_ = shape_;
-      } else {
-        result.shape_ = shapeBuilder_.build();
-      }
-      result.type_ = type_;
-      if (sliceBuilder_ == null) {
-        if (((bitField0_ & 0x00000001) != 0)) {
-          slice_ = java.util.Collections.unmodifiableList(slice_);
-          bitField0_ = (bitField0_ & ~0x00000001);
-        }
-        result.slice_ = slice_;
-      } else {
-        result.slice_ = sliceBuilder_.build();
-      }
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.util.SavedSliceMeta) {
-        return mergeFrom((org.tensorflow.proto.util.SavedSliceMeta)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.util.SavedSliceMeta other) {
-      if (other == org.tensorflow.proto.util.SavedSliceMeta.getDefaultInstance()) return this;
-      if (!other.getName().isEmpty()) {
-        name_ = other.name_;
-        onChanged();
-      }
-      if (other.hasShape()) {
-        mergeShape(other.getShape());
-      }
-      if (other.type_ != 0) {
-        setTypeValue(other.getTypeValue());
-      }
-      if (sliceBuilder_ == null) {
-        if (!other.slice_.isEmpty()) {
-          if (slice_.isEmpty()) {
-            slice_ = other.slice_;
-            bitField0_ = (bitField0_ & ~0x00000001);
-          } else {
-            ensureSliceIsMutable();
-            slice_.addAll(other.slice_);
-          }
-          onChanged();
-        }
-      } else {
-        if (!other.slice_.isEmpty()) {
-          if (sliceBuilder_.isEmpty()) {
-            sliceBuilder_.dispose();
-            sliceBuilder_ = null;
-            slice_ = other.slice_;
-            bitField0_ = (bitField0_ & ~0x00000001);
-            sliceBuilder_ = 
-              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
-                 getSliceFieldBuilder() : null;
-          } else {
-            sliceBuilder_.addAllMessages(other.slice_);
-          }
-        }
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.util.SavedSliceMeta parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.util.SavedSliceMeta) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-    private int bitField0_;
-
-    private java.lang.Object name_ = "";
-    /**
-     * <pre>
-     * Name of the tensor.
-     * </pre>
-     *
-     * <code>string name = 1;</code>
-     */
-    public java.lang.String getName() {
-      java.lang.Object ref = name_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        name_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Name of the tensor.
-     * </pre>
-     *
-     * <code>string name = 1;</code>
-     */
-    public com.google.protobuf.ByteString
-        getNameBytes() {
-      java.lang.Object ref = name_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        name_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Name of the tensor.
-     * </pre>
-     *
-     * <code>string name = 1;</code>
-     */
-    public Builder setName(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      name_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Name of the tensor.
-     * </pre>
-     *
-     * <code>string name = 1;</code>
-     */
-    public Builder clearName() {
-      
-      name_ = getDefaultInstance().getName();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Name of the tensor.
-     * </pre>
-     *
-     * <code>string name = 1;</code>
-     */
-    public Builder setNameBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      name_ = value;
-      onChanged();
-      return this;
-    }
-
-    private org.tensorflow.proto.framework.TensorShapeProto shape_;
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.TensorShapeProto, org.tensorflow.proto.framework.TensorShapeProto.Builder, org.tensorflow.proto.framework.TensorShapeProtoOrBuilder> shapeBuilder_;
-    /**
-     * <pre>
-     * Shape of the tensor
-     * </pre>
-     *
-     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-     */
-    public boolean hasShape() {
-      return shapeBuilder_ != null || shape_ != null;
-    }
-    /**
-     * <pre>
-     * Shape of the tensor
-     * </pre>
-     *
-     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-     */
-    public org.tensorflow.proto.framework.TensorShapeProto getShape() {
-      if (shapeBuilder_ == null) {
-        return shape_ == null ? org.tensorflow.proto.framework.TensorShapeProto.getDefaultInstance() : shape_;
-      } else {
-        return shapeBuilder_.getMessage();
-      }
-    }
-    /**
-     * <pre>
-     * Shape of the tensor
-     * </pre>
-     *
-     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-     */
-    public Builder setShape(org.tensorflow.proto.framework.TensorShapeProto value) {
-      if (shapeBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        shape_ = value;
-        onChanged();
-      } else {
-        shapeBuilder_.setMessage(value);
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Shape of the tensor
-     * </pre>
-     *
-     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-     */
-    public Builder setShape(
-        org.tensorflow.proto.framework.TensorShapeProto.Builder builderForValue) {
-      if (shapeBuilder_ == null) {
-        shape_ = builderForValue.build();
-        onChanged();
-      } else {
-        shapeBuilder_.setMessage(builderForValue.build());
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Shape of the tensor
-     * </pre>
-     *
-     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-     */
-    public Builder mergeShape(org.tensorflow.proto.framework.TensorShapeProto value) {
-      if (shapeBuilder_ == null) {
-        if (shape_ != null) {
-          shape_ =
-            org.tensorflow.proto.framework.TensorShapeProto.newBuilder(shape_).mergeFrom(value).buildPartial();
-        } else {
-          shape_ = value;
-        }
-        onChanged();
-      } else {
-        shapeBuilder_.mergeFrom(value);
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Shape of the tensor
-     * </pre>
-     *
-     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-     */
-    public Builder clearShape() {
-      if (shapeBuilder_ == null) {
-        shape_ = null;
-        onChanged();
-      } else {
-        shape_ = null;
-        shapeBuilder_ = null;
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Shape of the tensor
-     * </pre>
-     *
-     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-     */
-    public org.tensorflow.proto.framework.TensorShapeProto.Builder getShapeBuilder() {
-      
-      onChanged();
-      return getShapeFieldBuilder().getBuilder();
-    }
-    /**
-     * <pre>
-     * Shape of the tensor
-     * </pre>
-     *
-     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-     */
-    public org.tensorflow.proto.framework.TensorShapeProtoOrBuilder getShapeOrBuilder() {
-      if (shapeBuilder_ != null) {
-        return shapeBuilder_.getMessageOrBuilder();
-      } else {
-        return shape_ == null ?
-            org.tensorflow.proto.framework.TensorShapeProto.getDefaultInstance() : shape_;
-      }
-    }
-    /**
-     * <pre>
-     * Shape of the tensor
-     * </pre>
-     *
-     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-     */
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.TensorShapeProto, org.tensorflow.proto.framework.TensorShapeProto.Builder, org.tensorflow.proto.framework.TensorShapeProtoOrBuilder> 
-        getShapeFieldBuilder() {
-      if (shapeBuilder_ == null) {
-        shapeBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-            org.tensorflow.proto.framework.TensorShapeProto, org.tensorflow.proto.framework.TensorShapeProto.Builder, org.tensorflow.proto.framework.TensorShapeProtoOrBuilder>(
-                getShape(),
-                getParentForChildren(),
-                isClean());
-        shape_ = null;
-      }
-      return shapeBuilder_;
-    }
-
-    private int type_ = 0;
-    /**
-     * <pre>
-     * Type of the tensor
-     * </pre>
-     *
-     * <code>.tensorflow.DataType type = 3;</code>
-     */
-    public int getTypeValue() {
-      return type_;
-    }
-    /**
-     * <pre>
-     * Type of the tensor
-     * </pre>
-     *
-     * <code>.tensorflow.DataType type = 3;</code>
-     */
-    public Builder setTypeValue(int value) {
-      type_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Type of the tensor
-     * </pre>
-     *
-     * <code>.tensorflow.DataType type = 3;</code>
-     */
-    public org.tensorflow.proto.framework.DataType getType() {
-      @SuppressWarnings("deprecation")
-      org.tensorflow.proto.framework.DataType result = org.tensorflow.proto.framework.DataType.valueOf(type_);
-      return result == null ? org.tensorflow.proto.framework.DataType.UNRECOGNIZED : result;
-    }
-    /**
-     * <pre>
-     * Type of the tensor
-     * </pre>
-     *
-     * <code>.tensorflow.DataType type = 3;</code>
-     */
-    public Builder setType(org.tensorflow.proto.framework.DataType value) {
-      if (value == null) {
-        throw new NullPointerException();
-      }
-      
-      type_ = value.getNumber();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Type of the tensor
-     * </pre>
-     *
-     * <code>.tensorflow.DataType type = 3;</code>
-     */
-    public Builder clearType() {
-      
-      type_ = 0;
-      onChanged();
-      return this;
-    }
-
-    private java.util.List<org.tensorflow.proto.framework.TensorSliceProto> slice_ =
-      java.util.Collections.emptyList();
-    private void ensureSliceIsMutable() {
-      if (!((bitField0_ & 0x00000001) != 0)) {
-        slice_ = new java.util.ArrayList<org.tensorflow.proto.framework.TensorSliceProto>(slice_);
-        bitField0_ |= 0x00000001;
-       }
-    }
-
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.framework.TensorSliceProto, org.tensorflow.proto.framework.TensorSliceProto.Builder, org.tensorflow.proto.framework.TensorSliceProtoOrBuilder> sliceBuilder_;
-
-    /**
-     * <pre>
-     * Explicit list of slices saved in the checkpoint file.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorSliceProto slice = 4;</code>
-     */
-    public java.util.List<org.tensorflow.proto.framework.TensorSliceProto> getSliceList() {
-      if (sliceBuilder_ == null) {
-        return java.util.Collections.unmodifiableList(slice_);
-      } else {
-        return sliceBuilder_.getMessageList();
-      }
-    }
-    /**
-     * <pre>
-     * Explicit list of slices saved in the checkpoint file.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorSliceProto slice = 4;</code>
-     */
-    public int getSliceCount() {
-      if (sliceBuilder_ == null) {
-        return slice_.size();
-      } else {
-        return sliceBuilder_.getCount();
-      }
-    }
-    /**
-     * <pre>
-     * Explicit list of slices saved in the checkpoint file.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorSliceProto slice = 4;</code>
-     */
-    public org.tensorflow.proto.framework.TensorSliceProto getSlice(int index) {
-      if (sliceBuilder_ == null) {
-        return slice_.get(index);
-      } else {
-        return sliceBuilder_.getMessage(index);
-      }
-    }
-    /**
-     * <pre>
-     * Explicit list of slices saved in the checkpoint file.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorSliceProto slice = 4;</code>
-     */
-    public Builder setSlice(
-        int index, org.tensorflow.proto.framework.TensorSliceProto value) {
-      if (sliceBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureSliceIsMutable();
-        slice_.set(index, value);
-        onChanged();
-      } else {
-        sliceBuilder_.setMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Explicit list of slices saved in the checkpoint file.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorSliceProto slice = 4;</code>
-     */
-    public Builder setSlice(
-        int index, org.tensorflow.proto.framework.TensorSliceProto.Builder builderForValue) {
-      if (sliceBuilder_ == null) {
-        ensureSliceIsMutable();
-        slice_.set(index, builderForValue.build());
-        onChanged();
-      } else {
-        sliceBuilder_.setMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Explicit list of slices saved in the checkpoint file.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorSliceProto slice = 4;</code>
-     */
-    public Builder addSlice(org.tensorflow.proto.framework.TensorSliceProto value) {
-      if (sliceBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureSliceIsMutable();
-        slice_.add(value);
-        onChanged();
-      } else {
-        sliceBuilder_.addMessage(value);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Explicit list of slices saved in the checkpoint file.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorSliceProto slice = 4;</code>
-     */
-    public Builder addSlice(
-        int index, org.tensorflow.proto.framework.TensorSliceProto value) {
-      if (sliceBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureSliceIsMutable();
-        slice_.add(index, value);
-        onChanged();
-      } else {
-        sliceBuilder_.addMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Explicit list of slices saved in the checkpoint file.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorSliceProto slice = 4;</code>
-     */
-    public Builder addSlice(
-        org.tensorflow.proto.framework.TensorSliceProto.Builder builderForValue) {
-      if (sliceBuilder_ == null) {
-        ensureSliceIsMutable();
-        slice_.add(builderForValue.build());
-        onChanged();
-      } else {
-        sliceBuilder_.addMessage(builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Explicit list of slices saved in the checkpoint file.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorSliceProto slice = 4;</code>
-     */
-    public Builder addSlice(
-        int index, org.tensorflow.proto.framework.TensorSliceProto.Builder builderForValue) {
-      if (sliceBuilder_ == null) {
-        ensureSliceIsMutable();
-        slice_.add(index, builderForValue.build());
-        onChanged();
-      } else {
-        sliceBuilder_.addMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Explicit list of slices saved in the checkpoint file.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorSliceProto slice = 4;</code>
-     */
-    public Builder addAllSlice(
-        java.lang.Iterable<? extends org.tensorflow.proto.framework.TensorSliceProto> values) {
-      if (sliceBuilder_ == null) {
-        ensureSliceIsMutable();
-        com.google.protobuf.AbstractMessageLite.Builder.addAll(
-            values, slice_);
-        onChanged();
-      } else {
-        sliceBuilder_.addAllMessages(values);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Explicit list of slices saved in the checkpoint file.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorSliceProto slice = 4;</code>
-     */
-    public Builder clearSlice() {
-      if (sliceBuilder_ == null) {
-        slice_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000001);
-        onChanged();
-      } else {
-        sliceBuilder_.clear();
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Explicit list of slices saved in the checkpoint file.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorSliceProto slice = 4;</code>
-     */
-    public Builder removeSlice(int index) {
-      if (sliceBuilder_ == null) {
-        ensureSliceIsMutable();
-        slice_.remove(index);
-        onChanged();
-      } else {
-        sliceBuilder_.remove(index);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Explicit list of slices saved in the checkpoint file.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorSliceProto slice = 4;</code>
-     */
-    public org.tensorflow.proto.framework.TensorSliceProto.Builder getSliceBuilder(
-        int index) {
-      return getSliceFieldBuilder().getBuilder(index);
-    }
-    /**
-     * <pre>
-     * Explicit list of slices saved in the checkpoint file.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorSliceProto slice = 4;</code>
-     */
-    public org.tensorflow.proto.framework.TensorSliceProtoOrBuilder getSliceOrBuilder(
-        int index) {
-      if (sliceBuilder_ == null) {
-        return slice_.get(index);  } else {
-        return sliceBuilder_.getMessageOrBuilder(index);
-      }
-    }
-    /**
-     * <pre>
-     * Explicit list of slices saved in the checkpoint file.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorSliceProto slice = 4;</code>
-     */
-    public java.util.List<? extends org.tensorflow.proto.framework.TensorSliceProtoOrBuilder> 
-         getSliceOrBuilderList() {
-      if (sliceBuilder_ != null) {
-        return sliceBuilder_.getMessageOrBuilderList();
-      } else {
-        return java.util.Collections.unmodifiableList(slice_);
-      }
-    }
-    /**
-     * <pre>
-     * Explicit list of slices saved in the checkpoint file.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorSliceProto slice = 4;</code>
-     */
-    public org.tensorflow.proto.framework.TensorSliceProto.Builder addSliceBuilder() {
-      return getSliceFieldBuilder().addBuilder(
-          org.tensorflow.proto.framework.TensorSliceProto.getDefaultInstance());
-    }
-    /**
-     * <pre>
-     * Explicit list of slices saved in the checkpoint file.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorSliceProto slice = 4;</code>
-     */
-    public org.tensorflow.proto.framework.TensorSliceProto.Builder addSliceBuilder(
-        int index) {
-      return getSliceFieldBuilder().addBuilder(
-          index, org.tensorflow.proto.framework.TensorSliceProto.getDefaultInstance());
-    }
-    /**
-     * <pre>
-     * Explicit list of slices saved in the checkpoint file.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.TensorSliceProto slice = 4;</code>
-     */
-    public java.util.List<org.tensorflow.proto.framework.TensorSliceProto.Builder> 
-         getSliceBuilderList() {
-      return getSliceFieldBuilder().getBuilderList();
-    }
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.framework.TensorSliceProto, org.tensorflow.proto.framework.TensorSliceProto.Builder, org.tensorflow.proto.framework.TensorSliceProtoOrBuilder> 
-        getSliceFieldBuilder() {
-      if (sliceBuilder_ == null) {
-        sliceBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
-            org.tensorflow.proto.framework.TensorSliceProto, org.tensorflow.proto.framework.TensorSliceProto.Builder, org.tensorflow.proto.framework.TensorSliceProtoOrBuilder>(
-                slice_,
-                ((bitField0_ & 0x00000001) != 0),
-                getParentForChildren(),
-                isClean());
-        slice_ = null;
-      }
-      return sliceBuilder_;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.SavedSliceMeta)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.SavedSliceMeta)
-  private static final org.tensorflow.proto.util.SavedSliceMeta DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.util.SavedSliceMeta();
-  }
-
-  public static org.tensorflow.proto.util.SavedSliceMeta getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<SavedSliceMeta>
-      PARSER = new com.google.protobuf.AbstractParser<SavedSliceMeta>() {
-    @java.lang.Override
-    public SavedSliceMeta parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new SavedSliceMeta(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<SavedSliceMeta> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<SavedSliceMeta> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.util.SavedSliceMeta getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SavedSliceMetaOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SavedSliceMetaOrBuilder.java
deleted file mode 100644
index 76aff4e7eb5..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SavedSliceMetaOrBuilder.java
+++ /dev/null
@@ -1,113 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/util/saved_tensor_slice.proto
-
-package org.tensorflow.proto.util;
-
-public interface SavedSliceMetaOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.SavedSliceMeta)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <pre>
-   * Name of the tensor.
-   * </pre>
-   *
-   * <code>string name = 1;</code>
-   */
-  java.lang.String getName();
-  /**
-   * <pre>
-   * Name of the tensor.
-   * </pre>
-   *
-   * <code>string name = 1;</code>
-   */
-  com.google.protobuf.ByteString
-      getNameBytes();
-
-  /**
-   * <pre>
-   * Shape of the tensor
-   * </pre>
-   *
-   * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-   */
-  boolean hasShape();
-  /**
-   * <pre>
-   * Shape of the tensor
-   * </pre>
-   *
-   * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-   */
-  org.tensorflow.proto.framework.TensorShapeProto getShape();
-  /**
-   * <pre>
-   * Shape of the tensor
-   * </pre>
-   *
-   * <code>.tensorflow.TensorShapeProto shape = 2;</code>
-   */
-  org.tensorflow.proto.framework.TensorShapeProtoOrBuilder getShapeOrBuilder();
-
-  /**
-   * <pre>
-   * Type of the tensor
-   * </pre>
-   *
-   * <code>.tensorflow.DataType type = 3;</code>
-   */
-  int getTypeValue();
-  /**
-   * <pre>
-   * Type of the tensor
-   * </pre>
-   *
-   * <code>.tensorflow.DataType type = 3;</code>
-   */
-  org.tensorflow.proto.framework.DataType getType();
-
-  /**
-   * <pre>
-   * Explicit list of slices saved in the checkpoint file.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.TensorSliceProto slice = 4;</code>
-   */
-  java.util.List<org.tensorflow.proto.framework.TensorSliceProto> 
-      getSliceList();
-  /**
-   * <pre>
-   * Explicit list of slices saved in the checkpoint file.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.TensorSliceProto slice = 4;</code>
-   */
-  org.tensorflow.proto.framework.TensorSliceProto getSlice(int index);
-  /**
-   * <pre>
-   * Explicit list of slices saved in the checkpoint file.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.TensorSliceProto slice = 4;</code>
-   */
-  int getSliceCount();
-  /**
-   * <pre>
-   * Explicit list of slices saved in the checkpoint file.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.TensorSliceProto slice = 4;</code>
-   */
-  java.util.List<? extends org.tensorflow.proto.framework.TensorSliceProtoOrBuilder> 
-      getSliceOrBuilderList();
-  /**
-   * <pre>
-   * Explicit list of slices saved in the checkpoint file.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.TensorSliceProto slice = 4;</code>
-   */
-  org.tensorflow.proto.framework.TensorSliceProtoOrBuilder getSliceOrBuilder(
-      int index);
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SavedSliceOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SavedSliceOrBuilder.java
deleted file mode 100644
index bf76430d534..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SavedSliceOrBuilder.java
+++ /dev/null
@@ -1,85 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/util/saved_tensor_slice.proto
-
-package org.tensorflow.proto.util;
-
-public interface SavedSliceOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.SavedSlice)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <pre>
-   * Name of the tensor that this slice belongs to. This must be identical to
-   * the name used to encode the key for this record.
-   * </pre>
-   *
-   * <code>string name = 1;</code>
-   */
-  java.lang.String getName();
-  /**
-   * <pre>
-   * Name of the tensor that this slice belongs to. This must be identical to
-   * the name used to encode the key for this record.
-   * </pre>
-   *
-   * <code>string name = 1;</code>
-   */
-  com.google.protobuf.ByteString
-      getNameBytes();
-
-  /**
-   * <pre>
-   * Extent of the slice.  Must have one entry for each of the dimension of the
-   * tensor that this slice belongs to.
-   * </pre>
-   *
-   * <code>.tensorflow.TensorSliceProto slice = 2;</code>
-   */
-  boolean hasSlice();
-  /**
-   * <pre>
-   * Extent of the slice.  Must have one entry for each of the dimension of the
-   * tensor that this slice belongs to.
-   * </pre>
-   *
-   * <code>.tensorflow.TensorSliceProto slice = 2;</code>
-   */
-  org.tensorflow.proto.framework.TensorSliceProto getSlice();
-  /**
-   * <pre>
-   * Extent of the slice.  Must have one entry for each of the dimension of the
-   * tensor that this slice belongs to.
-   * </pre>
-   *
-   * <code>.tensorflow.TensorSliceProto slice = 2;</code>
-   */
-  org.tensorflow.proto.framework.TensorSliceProtoOrBuilder getSliceOrBuilder();
-
-  /**
-   * <pre>
-   * The raw data of the slice is stored as a TensorProto. Only raw data are
-   * stored (we don't fill in fields such as dtype or tensor_shape).
-   * </pre>
-   *
-   * <code>.tensorflow.TensorProto data = 3;</code>
-   */
-  boolean hasData();
-  /**
-   * <pre>
-   * The raw data of the slice is stored as a TensorProto. Only raw data are
-   * stored (we don't fill in fields such as dtype or tensor_shape).
-   * </pre>
-   *
-   * <code>.tensorflow.TensorProto data = 3;</code>
-   */
-  org.tensorflow.proto.framework.TensorProto getData();
-  /**
-   * <pre>
-   * The raw data of the slice is stored as a TensorProto. Only raw data are
-   * stored (we don't fill in fields such as dtype or tensor_shape).
-   * </pre>
-   *
-   * <code>.tensorflow.TensorProto data = 3;</code>
-   */
-  org.tensorflow.proto.framework.TensorProtoOrBuilder getDataOrBuilder();
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SavedTensorSliceMeta.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SavedTensorSliceMeta.java
deleted file mode 100644
index 6217e7ed403..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SavedTensorSliceMeta.java
+++ /dev/null
@@ -1,1108 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/util/saved_tensor_slice.proto
-
-package org.tensorflow.proto.util;
-
-/**
- * <pre>
- * Metadata describing the set of tensor slices saved in a checkpoint file.
- * It is always stored at the beginning of each checkpoint file.
- * </pre>
- *
- * Protobuf type {@code tensorflow.SavedTensorSliceMeta}
- */
-public  final class SavedTensorSliceMeta extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.SavedTensorSliceMeta)
-    SavedTensorSliceMetaOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use SavedTensorSliceMeta.newBuilder() to construct.
-  private SavedTensorSliceMeta(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private SavedTensorSliceMeta() {
-    tensor_ = java.util.Collections.emptyList();
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new SavedTensorSliceMeta();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private SavedTensorSliceMeta(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    int mutable_bitField0_ = 0;
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 10: {
-            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
-              tensor_ = new java.util.ArrayList<org.tensorflow.proto.util.SavedSliceMeta>();
-              mutable_bitField0_ |= 0x00000001;
-            }
-            tensor_.add(
-                input.readMessage(org.tensorflow.proto.util.SavedSliceMeta.parser(), extensionRegistry));
-            break;
-          }
-          case 18: {
-            org.tensorflow.proto.framework.VersionDef.Builder subBuilder = null;
-            if (versions_ != null) {
-              subBuilder = versions_.toBuilder();
-            }
-            versions_ = input.readMessage(org.tensorflow.proto.framework.VersionDef.parser(), extensionRegistry);
-            if (subBuilder != null) {
-              subBuilder.mergeFrom(versions_);
-              versions_ = subBuilder.buildPartial();
-            }
-
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      if (((mutable_bitField0_ & 0x00000001) != 0)) {
-        tensor_ = java.util.Collections.unmodifiableList(tensor_);
-      }
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.util.SavedTensorSliceProtos.internal_static_tensorflow_SavedTensorSliceMeta_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.util.SavedTensorSliceProtos.internal_static_tensorflow_SavedTensorSliceMeta_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.util.SavedTensorSliceMeta.class, org.tensorflow.proto.util.SavedTensorSliceMeta.Builder.class);
-  }
-
-  public static final int TENSOR_FIELD_NUMBER = 1;
-  private java.util.List<org.tensorflow.proto.util.SavedSliceMeta> tensor_;
-  /**
-   * <pre>
-   * Each SavedSliceMeta describes the slices for one tensor.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.SavedSliceMeta tensor = 1;</code>
-   */
-  public java.util.List<org.tensorflow.proto.util.SavedSliceMeta> getTensorList() {
-    return tensor_;
-  }
-  /**
-   * <pre>
-   * Each SavedSliceMeta describes the slices for one tensor.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.SavedSliceMeta tensor = 1;</code>
-   */
-  public java.util.List<? extends org.tensorflow.proto.util.SavedSliceMetaOrBuilder> 
-      getTensorOrBuilderList() {
-    return tensor_;
-  }
-  /**
-   * <pre>
-   * Each SavedSliceMeta describes the slices for one tensor.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.SavedSliceMeta tensor = 1;</code>
-   */
-  public int getTensorCount() {
-    return tensor_.size();
-  }
-  /**
-   * <pre>
-   * Each SavedSliceMeta describes the slices for one tensor.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.SavedSliceMeta tensor = 1;</code>
-   */
-  public org.tensorflow.proto.util.SavedSliceMeta getTensor(int index) {
-    return tensor_.get(index);
-  }
-  /**
-   * <pre>
-   * Each SavedSliceMeta describes the slices for one tensor.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.SavedSliceMeta tensor = 1;</code>
-   */
-  public org.tensorflow.proto.util.SavedSliceMetaOrBuilder getTensorOrBuilder(
-      int index) {
-    return tensor_.get(index);
-  }
-
-  public static final int VERSIONS_FIELD_NUMBER = 2;
-  private org.tensorflow.proto.framework.VersionDef versions_;
-  /**
-   * <pre>
-   * Compatibility version of this checkpoint.  See core/public/version.h
-   * for version history.
-   * </pre>
-   *
-   * <code>.tensorflow.VersionDef versions = 2;</code>
-   */
-  public boolean hasVersions() {
-    return versions_ != null;
-  }
-  /**
-   * <pre>
-   * Compatibility version of this checkpoint.  See core/public/version.h
-   * for version history.
-   * </pre>
-   *
-   * <code>.tensorflow.VersionDef versions = 2;</code>
-   */
-  public org.tensorflow.proto.framework.VersionDef getVersions() {
-    return versions_ == null ? org.tensorflow.proto.framework.VersionDef.getDefaultInstance() : versions_;
-  }
-  /**
-   * <pre>
-   * Compatibility version of this checkpoint.  See core/public/version.h
-   * for version history.
-   * </pre>
-   *
-   * <code>.tensorflow.VersionDef versions = 2;</code>
-   */
-  public org.tensorflow.proto.framework.VersionDefOrBuilder getVersionsOrBuilder() {
-    return getVersions();
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    for (int i = 0; i < tensor_.size(); i++) {
-      output.writeMessage(1, tensor_.get(i));
-    }
-    if (versions_ != null) {
-      output.writeMessage(2, getVersions());
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    for (int i = 0; i < tensor_.size(); i++) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(1, tensor_.get(i));
-    }
-    if (versions_ != null) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(2, getVersions());
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.util.SavedTensorSliceMeta)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.util.SavedTensorSliceMeta other = (org.tensorflow.proto.util.SavedTensorSliceMeta) obj;
-
-    if (!getTensorList()
-        .equals(other.getTensorList())) return false;
-    if (hasVersions() != other.hasVersions()) return false;
-    if (hasVersions()) {
-      if (!getVersions()
-          .equals(other.getVersions())) return false;
-    }
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    if (getTensorCount() > 0) {
-      hash = (37 * hash) + TENSOR_FIELD_NUMBER;
-      hash = (53 * hash) + getTensorList().hashCode();
-    }
-    if (hasVersions()) {
-      hash = (37 * hash) + VERSIONS_FIELD_NUMBER;
-      hash = (53 * hash) + getVersions().hashCode();
-    }
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.util.SavedTensorSliceMeta parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.SavedTensorSliceMeta parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.SavedTensorSliceMeta parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.SavedTensorSliceMeta parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.SavedTensorSliceMeta parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.SavedTensorSliceMeta parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.SavedTensorSliceMeta parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.SavedTensorSliceMeta parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.SavedTensorSliceMeta parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.SavedTensorSliceMeta parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.SavedTensorSliceMeta parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.SavedTensorSliceMeta parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.util.SavedTensorSliceMeta prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * <pre>
-   * Metadata describing the set of tensor slices saved in a checkpoint file.
-   * It is always stored at the beginning of each checkpoint file.
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.SavedTensorSliceMeta}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.SavedTensorSliceMeta)
-      org.tensorflow.proto.util.SavedTensorSliceMetaOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.util.SavedTensorSliceProtos.internal_static_tensorflow_SavedTensorSliceMeta_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.util.SavedTensorSliceProtos.internal_static_tensorflow_SavedTensorSliceMeta_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.util.SavedTensorSliceMeta.class, org.tensorflow.proto.util.SavedTensorSliceMeta.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.util.SavedTensorSliceMeta.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-        getTensorFieldBuilder();
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      if (tensorBuilder_ == null) {
-        tensor_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000001);
-      } else {
-        tensorBuilder_.clear();
-      }
-      if (versionsBuilder_ == null) {
-        versions_ = null;
-      } else {
-        versions_ = null;
-        versionsBuilder_ = null;
-      }
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.util.SavedTensorSliceProtos.internal_static_tensorflow_SavedTensorSliceMeta_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.SavedTensorSliceMeta getDefaultInstanceForType() {
-      return org.tensorflow.proto.util.SavedTensorSliceMeta.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.SavedTensorSliceMeta build() {
-      org.tensorflow.proto.util.SavedTensorSliceMeta result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.SavedTensorSliceMeta buildPartial() {
-      org.tensorflow.proto.util.SavedTensorSliceMeta result = new org.tensorflow.proto.util.SavedTensorSliceMeta(this);
-      int from_bitField0_ = bitField0_;
-      if (tensorBuilder_ == null) {
-        if (((bitField0_ & 0x00000001) != 0)) {
-          tensor_ = java.util.Collections.unmodifiableList(tensor_);
-          bitField0_ = (bitField0_ & ~0x00000001);
-        }
-        result.tensor_ = tensor_;
-      } else {
-        result.tensor_ = tensorBuilder_.build();
-      }
-      if (versionsBuilder_ == null) {
-        result.versions_ = versions_;
-      } else {
-        result.versions_ = versionsBuilder_.build();
-      }
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.util.SavedTensorSliceMeta) {
-        return mergeFrom((org.tensorflow.proto.util.SavedTensorSliceMeta)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.util.SavedTensorSliceMeta other) {
-      if (other == org.tensorflow.proto.util.SavedTensorSliceMeta.getDefaultInstance()) return this;
-      if (tensorBuilder_ == null) {
-        if (!other.tensor_.isEmpty()) {
-          if (tensor_.isEmpty()) {
-            tensor_ = other.tensor_;
-            bitField0_ = (bitField0_ & ~0x00000001);
-          } else {
-            ensureTensorIsMutable();
-            tensor_.addAll(other.tensor_);
-          }
-          onChanged();
-        }
-      } else {
-        if (!other.tensor_.isEmpty()) {
-          if (tensorBuilder_.isEmpty()) {
-            tensorBuilder_.dispose();
-            tensorBuilder_ = null;
-            tensor_ = other.tensor_;
-            bitField0_ = (bitField0_ & ~0x00000001);
-            tensorBuilder_ = 
-              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
-                 getTensorFieldBuilder() : null;
-          } else {
-            tensorBuilder_.addAllMessages(other.tensor_);
-          }
-        }
-      }
-      if (other.hasVersions()) {
-        mergeVersions(other.getVersions());
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.util.SavedTensorSliceMeta parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.util.SavedTensorSliceMeta) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-    private int bitField0_;
-
-    private java.util.List<org.tensorflow.proto.util.SavedSliceMeta> tensor_ =
-      java.util.Collections.emptyList();
-    private void ensureTensorIsMutable() {
-      if (!((bitField0_ & 0x00000001) != 0)) {
-        tensor_ = new java.util.ArrayList<org.tensorflow.proto.util.SavedSliceMeta>(tensor_);
-        bitField0_ |= 0x00000001;
-       }
-    }
-
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.util.SavedSliceMeta, org.tensorflow.proto.util.SavedSliceMeta.Builder, org.tensorflow.proto.util.SavedSliceMetaOrBuilder> tensorBuilder_;
-
-    /**
-     * <pre>
-     * Each SavedSliceMeta describes the slices for one tensor.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.SavedSliceMeta tensor = 1;</code>
-     */
-    public java.util.List<org.tensorflow.proto.util.SavedSliceMeta> getTensorList() {
-      if (tensorBuilder_ == null) {
-        return java.util.Collections.unmodifiableList(tensor_);
-      } else {
-        return tensorBuilder_.getMessageList();
-      }
-    }
-    /**
-     * <pre>
-     * Each SavedSliceMeta describes the slices for one tensor.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.SavedSliceMeta tensor = 1;</code>
-     */
-    public int getTensorCount() {
-      if (tensorBuilder_ == null) {
-        return tensor_.size();
-      } else {
-        return tensorBuilder_.getCount();
-      }
-    }
-    /**
-     * <pre>
-     * Each SavedSliceMeta describes the slices for one tensor.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.SavedSliceMeta tensor = 1;</code>
-     */
-    public org.tensorflow.proto.util.SavedSliceMeta getTensor(int index) {
-      if (tensorBuilder_ == null) {
-        return tensor_.get(index);
-      } else {
-        return tensorBuilder_.getMessage(index);
-      }
-    }
-    /**
-     * <pre>
-     * Each SavedSliceMeta describes the slices for one tensor.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.SavedSliceMeta tensor = 1;</code>
-     */
-    public Builder setTensor(
-        int index, org.tensorflow.proto.util.SavedSliceMeta value) {
-      if (tensorBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureTensorIsMutable();
-        tensor_.set(index, value);
-        onChanged();
-      } else {
-        tensorBuilder_.setMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Each SavedSliceMeta describes the slices for one tensor.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.SavedSliceMeta tensor = 1;</code>
-     */
-    public Builder setTensor(
-        int index, org.tensorflow.proto.util.SavedSliceMeta.Builder builderForValue) {
-      if (tensorBuilder_ == null) {
-        ensureTensorIsMutable();
-        tensor_.set(index, builderForValue.build());
-        onChanged();
-      } else {
-        tensorBuilder_.setMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Each SavedSliceMeta describes the slices for one tensor.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.SavedSliceMeta tensor = 1;</code>
-     */
-    public Builder addTensor(org.tensorflow.proto.util.SavedSliceMeta value) {
-      if (tensorBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureTensorIsMutable();
-        tensor_.add(value);
-        onChanged();
-      } else {
-        tensorBuilder_.addMessage(value);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Each SavedSliceMeta describes the slices for one tensor.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.SavedSliceMeta tensor = 1;</code>
-     */
-    public Builder addTensor(
-        int index, org.tensorflow.proto.util.SavedSliceMeta value) {
-      if (tensorBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureTensorIsMutable();
-        tensor_.add(index, value);
-        onChanged();
-      } else {
-        tensorBuilder_.addMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Each SavedSliceMeta describes the slices for one tensor.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.SavedSliceMeta tensor = 1;</code>
-     */
-    public Builder addTensor(
-        org.tensorflow.proto.util.SavedSliceMeta.Builder builderForValue) {
-      if (tensorBuilder_ == null) {
-        ensureTensorIsMutable();
-        tensor_.add(builderForValue.build());
-        onChanged();
-      } else {
-        tensorBuilder_.addMessage(builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Each SavedSliceMeta describes the slices for one tensor.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.SavedSliceMeta tensor = 1;</code>
-     */
-    public Builder addTensor(
-        int index, org.tensorflow.proto.util.SavedSliceMeta.Builder builderForValue) {
-      if (tensorBuilder_ == null) {
-        ensureTensorIsMutable();
-        tensor_.add(index, builderForValue.build());
-        onChanged();
-      } else {
-        tensorBuilder_.addMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Each SavedSliceMeta describes the slices for one tensor.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.SavedSliceMeta tensor = 1;</code>
-     */
-    public Builder addAllTensor(
-        java.lang.Iterable<? extends org.tensorflow.proto.util.SavedSliceMeta> values) {
-      if (tensorBuilder_ == null) {
-        ensureTensorIsMutable();
-        com.google.protobuf.AbstractMessageLite.Builder.addAll(
-            values, tensor_);
-        onChanged();
-      } else {
-        tensorBuilder_.addAllMessages(values);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Each SavedSliceMeta describes the slices for one tensor.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.SavedSliceMeta tensor = 1;</code>
-     */
-    public Builder clearTensor() {
-      if (tensorBuilder_ == null) {
-        tensor_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000001);
-        onChanged();
-      } else {
-        tensorBuilder_.clear();
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Each SavedSliceMeta describes the slices for one tensor.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.SavedSliceMeta tensor = 1;</code>
-     */
-    public Builder removeTensor(int index) {
-      if (tensorBuilder_ == null) {
-        ensureTensorIsMutable();
-        tensor_.remove(index);
-        onChanged();
-      } else {
-        tensorBuilder_.remove(index);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Each SavedSliceMeta describes the slices for one tensor.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.SavedSliceMeta tensor = 1;</code>
-     */
-    public org.tensorflow.proto.util.SavedSliceMeta.Builder getTensorBuilder(
-        int index) {
-      return getTensorFieldBuilder().getBuilder(index);
-    }
-    /**
-     * <pre>
-     * Each SavedSliceMeta describes the slices for one tensor.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.SavedSliceMeta tensor = 1;</code>
-     */
-    public org.tensorflow.proto.util.SavedSliceMetaOrBuilder getTensorOrBuilder(
-        int index) {
-      if (tensorBuilder_ == null) {
-        return tensor_.get(index);  } else {
-        return tensorBuilder_.getMessageOrBuilder(index);
-      }
-    }
-    /**
-     * <pre>
-     * Each SavedSliceMeta describes the slices for one tensor.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.SavedSliceMeta tensor = 1;</code>
-     */
-    public java.util.List<? extends org.tensorflow.proto.util.SavedSliceMetaOrBuilder> 
-         getTensorOrBuilderList() {
-      if (tensorBuilder_ != null) {
-        return tensorBuilder_.getMessageOrBuilderList();
-      } else {
-        return java.util.Collections.unmodifiableList(tensor_);
-      }
-    }
-    /**
-     * <pre>
-     * Each SavedSliceMeta describes the slices for one tensor.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.SavedSliceMeta tensor = 1;</code>
-     */
-    public org.tensorflow.proto.util.SavedSliceMeta.Builder addTensorBuilder() {
-      return getTensorFieldBuilder().addBuilder(
-          org.tensorflow.proto.util.SavedSliceMeta.getDefaultInstance());
-    }
-    /**
-     * <pre>
-     * Each SavedSliceMeta describes the slices for one tensor.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.SavedSliceMeta tensor = 1;</code>
-     */
-    public org.tensorflow.proto.util.SavedSliceMeta.Builder addTensorBuilder(
-        int index) {
-      return getTensorFieldBuilder().addBuilder(
-          index, org.tensorflow.proto.util.SavedSliceMeta.getDefaultInstance());
-    }
-    /**
-     * <pre>
-     * Each SavedSliceMeta describes the slices for one tensor.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.SavedSliceMeta tensor = 1;</code>
-     */
-    public java.util.List<org.tensorflow.proto.util.SavedSliceMeta.Builder> 
-         getTensorBuilderList() {
-      return getTensorFieldBuilder().getBuilderList();
-    }
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.util.SavedSliceMeta, org.tensorflow.proto.util.SavedSliceMeta.Builder, org.tensorflow.proto.util.SavedSliceMetaOrBuilder> 
-        getTensorFieldBuilder() {
-      if (tensorBuilder_ == null) {
-        tensorBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
-            org.tensorflow.proto.util.SavedSliceMeta, org.tensorflow.proto.util.SavedSliceMeta.Builder, org.tensorflow.proto.util.SavedSliceMetaOrBuilder>(
-                tensor_,
-                ((bitField0_ & 0x00000001) != 0),
-                getParentForChildren(),
-                isClean());
-        tensor_ = null;
-      }
-      return tensorBuilder_;
-    }
-
-    private org.tensorflow.proto.framework.VersionDef versions_;
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.VersionDef, org.tensorflow.proto.framework.VersionDef.Builder, org.tensorflow.proto.framework.VersionDefOrBuilder> versionsBuilder_;
-    /**
-     * <pre>
-     * Compatibility version of this checkpoint.  See core/public/version.h
-     * for version history.
-     * </pre>
-     *
-     * <code>.tensorflow.VersionDef versions = 2;</code>
-     */
-    public boolean hasVersions() {
-      return versionsBuilder_ != null || versions_ != null;
-    }
-    /**
-     * <pre>
-     * Compatibility version of this checkpoint.  See core/public/version.h
-     * for version history.
-     * </pre>
-     *
-     * <code>.tensorflow.VersionDef versions = 2;</code>
-     */
-    public org.tensorflow.proto.framework.VersionDef getVersions() {
-      if (versionsBuilder_ == null) {
-        return versions_ == null ? org.tensorflow.proto.framework.VersionDef.getDefaultInstance() : versions_;
-      } else {
-        return versionsBuilder_.getMessage();
-      }
-    }
-    /**
-     * <pre>
-     * Compatibility version of this checkpoint.  See core/public/version.h
-     * for version history.
-     * </pre>
-     *
-     * <code>.tensorflow.VersionDef versions = 2;</code>
-     */
-    public Builder setVersions(org.tensorflow.proto.framework.VersionDef value) {
-      if (versionsBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        versions_ = value;
-        onChanged();
-      } else {
-        versionsBuilder_.setMessage(value);
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Compatibility version of this checkpoint.  See core/public/version.h
-     * for version history.
-     * </pre>
-     *
-     * <code>.tensorflow.VersionDef versions = 2;</code>
-     */
-    public Builder setVersions(
-        org.tensorflow.proto.framework.VersionDef.Builder builderForValue) {
-      if (versionsBuilder_ == null) {
-        versions_ = builderForValue.build();
-        onChanged();
-      } else {
-        versionsBuilder_.setMessage(builderForValue.build());
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Compatibility version of this checkpoint.  See core/public/version.h
-     * for version history.
-     * </pre>
-     *
-     * <code>.tensorflow.VersionDef versions = 2;</code>
-     */
-    public Builder mergeVersions(org.tensorflow.proto.framework.VersionDef value) {
-      if (versionsBuilder_ == null) {
-        if (versions_ != null) {
-          versions_ =
-            org.tensorflow.proto.framework.VersionDef.newBuilder(versions_).mergeFrom(value).buildPartial();
-        } else {
-          versions_ = value;
-        }
-        onChanged();
-      } else {
-        versionsBuilder_.mergeFrom(value);
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Compatibility version of this checkpoint.  See core/public/version.h
-     * for version history.
-     * </pre>
-     *
-     * <code>.tensorflow.VersionDef versions = 2;</code>
-     */
-    public Builder clearVersions() {
-      if (versionsBuilder_ == null) {
-        versions_ = null;
-        onChanged();
-      } else {
-        versions_ = null;
-        versionsBuilder_ = null;
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Compatibility version of this checkpoint.  See core/public/version.h
-     * for version history.
-     * </pre>
-     *
-     * <code>.tensorflow.VersionDef versions = 2;</code>
-     */
-    public org.tensorflow.proto.framework.VersionDef.Builder getVersionsBuilder() {
-      
-      onChanged();
-      return getVersionsFieldBuilder().getBuilder();
-    }
-    /**
-     * <pre>
-     * Compatibility version of this checkpoint.  See core/public/version.h
-     * for version history.
-     * </pre>
-     *
-     * <code>.tensorflow.VersionDef versions = 2;</code>
-     */
-    public org.tensorflow.proto.framework.VersionDefOrBuilder getVersionsOrBuilder() {
-      if (versionsBuilder_ != null) {
-        return versionsBuilder_.getMessageOrBuilder();
-      } else {
-        return versions_ == null ?
-            org.tensorflow.proto.framework.VersionDef.getDefaultInstance() : versions_;
-      }
-    }
-    /**
-     * <pre>
-     * Compatibility version of this checkpoint.  See core/public/version.h
-     * for version history.
-     * </pre>
-     *
-     * <code>.tensorflow.VersionDef versions = 2;</code>
-     */
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.VersionDef, org.tensorflow.proto.framework.VersionDef.Builder, org.tensorflow.proto.framework.VersionDefOrBuilder> 
-        getVersionsFieldBuilder() {
-      if (versionsBuilder_ == null) {
-        versionsBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-            org.tensorflow.proto.framework.VersionDef, org.tensorflow.proto.framework.VersionDef.Builder, org.tensorflow.proto.framework.VersionDefOrBuilder>(
-                getVersions(),
-                getParentForChildren(),
-                isClean());
-        versions_ = null;
-      }
-      return versionsBuilder_;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.SavedTensorSliceMeta)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.SavedTensorSliceMeta)
-  private static final org.tensorflow.proto.util.SavedTensorSliceMeta DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.util.SavedTensorSliceMeta();
-  }
-
-  public static org.tensorflow.proto.util.SavedTensorSliceMeta getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<SavedTensorSliceMeta>
-      PARSER = new com.google.protobuf.AbstractParser<SavedTensorSliceMeta>() {
-    @java.lang.Override
-    public SavedTensorSliceMeta parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new SavedTensorSliceMeta(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<SavedTensorSliceMeta> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<SavedTensorSliceMeta> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.util.SavedTensorSliceMeta getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SavedTensorSliceMetaOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SavedTensorSliceMetaOrBuilder.java
deleted file mode 100644
index 464fc6423f6..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SavedTensorSliceMetaOrBuilder.java
+++ /dev/null
@@ -1,81 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/util/saved_tensor_slice.proto
-
-package org.tensorflow.proto.util;
-
-public interface SavedTensorSliceMetaOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.SavedTensorSliceMeta)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <pre>
-   * Each SavedSliceMeta describes the slices for one tensor.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.SavedSliceMeta tensor = 1;</code>
-   */
-  java.util.List<org.tensorflow.proto.util.SavedSliceMeta> 
-      getTensorList();
-  /**
-   * <pre>
-   * Each SavedSliceMeta describes the slices for one tensor.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.SavedSliceMeta tensor = 1;</code>
-   */
-  org.tensorflow.proto.util.SavedSliceMeta getTensor(int index);
-  /**
-   * <pre>
-   * Each SavedSliceMeta describes the slices for one tensor.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.SavedSliceMeta tensor = 1;</code>
-   */
-  int getTensorCount();
-  /**
-   * <pre>
-   * Each SavedSliceMeta describes the slices for one tensor.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.SavedSliceMeta tensor = 1;</code>
-   */
-  java.util.List<? extends org.tensorflow.proto.util.SavedSliceMetaOrBuilder> 
-      getTensorOrBuilderList();
-  /**
-   * <pre>
-   * Each SavedSliceMeta describes the slices for one tensor.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.SavedSliceMeta tensor = 1;</code>
-   */
-  org.tensorflow.proto.util.SavedSliceMetaOrBuilder getTensorOrBuilder(
-      int index);
-
-  /**
-   * <pre>
-   * Compatibility version of this checkpoint.  See core/public/version.h
-   * for version history.
-   * </pre>
-   *
-   * <code>.tensorflow.VersionDef versions = 2;</code>
-   */
-  boolean hasVersions();
-  /**
-   * <pre>
-   * Compatibility version of this checkpoint.  See core/public/version.h
-   * for version history.
-   * </pre>
-   *
-   * <code>.tensorflow.VersionDef versions = 2;</code>
-   */
-  org.tensorflow.proto.framework.VersionDef getVersions();
-  /**
-   * <pre>
-   * Compatibility version of this checkpoint.  See core/public/version.h
-   * for version history.
-   * </pre>
-   *
-   * <code>.tensorflow.VersionDef versions = 2;</code>
-   */
-  org.tensorflow.proto.framework.VersionDefOrBuilder getVersionsOrBuilder();
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SavedTensorSliceProtos.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SavedTensorSliceProtos.java
deleted file mode 100644
index 94de1b07571..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SavedTensorSliceProtos.java
+++ /dev/null
@@ -1,109 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/util/saved_tensor_slice.proto
-
-package org.tensorflow.proto.util;
-
-public final class SavedTensorSliceProtos {
-  private SavedTensorSliceProtos() {}
-  public static void registerAllExtensions(
-      com.google.protobuf.ExtensionRegistryLite registry) {
-  }
-
-  public static void registerAllExtensions(
-      com.google.protobuf.ExtensionRegistry registry) {
-    registerAllExtensions(
-        (com.google.protobuf.ExtensionRegistryLite) registry);
-  }
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_SavedSliceMeta_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_SavedSliceMeta_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_SavedTensorSliceMeta_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_SavedTensorSliceMeta_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_SavedSlice_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_SavedSlice_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_SavedTensorSlices_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_SavedTensorSlices_fieldAccessorTable;
-
-  public static com.google.protobuf.Descriptors.FileDescriptor
-      getDescriptor() {
-    return descriptor;
-  }
-  private static  com.google.protobuf.Descriptors.FileDescriptor
-      descriptor;
-  static {
-    java.lang.String[] descriptorData = {
-      "\n-tensorflow/core/util/saved_tensor_slic" +
-      "e.proto\022\ntensorflow\032,tensorflow/core/fra" +
-      "mework/tensor_shape.proto\032,tensorflow/co" +
-      "re/framework/tensor_slice.proto\032&tensorf" +
-      "low/core/framework/tensor.proto\032%tensorf" +
-      "low/core/framework/types.proto\032(tensorfl" +
-      "ow/core/framework/versions.proto\"\234\001\n\016Sav" +
-      "edSliceMeta\022\014\n\004name\030\001 \001(\t\022+\n\005shape\030\002 \001(\013" +
-      "2\034.tensorflow.TensorShapeProto\022\"\n\004type\030\003" +
-      " \001(\0162\024.tensorflow.DataType\022+\n\005slice\030\004 \003(" +
-      "\0132\034.tensorflow.TensorSliceProto\"l\n\024Saved" +
-      "TensorSliceMeta\022*\n\006tensor\030\001 \003(\0132\032.tensor" +
-      "flow.SavedSliceMeta\022(\n\010versions\030\002 \001(\0132\026." +
-      "tensorflow.VersionDef\"n\n\nSavedSlice\022\014\n\004n" +
-      "ame\030\001 \001(\t\022+\n\005slice\030\002 \001(\0132\034.tensorflow.Te" +
-      "nsorSliceProto\022%\n\004data\030\003 \001(\0132\027.tensorflo" +
-      "w.TensorProto\"i\n\021SavedTensorSlices\022.\n\004me" +
-      "ta\030\001 \001(\0132 .tensorflow.SavedTensorSliceMe" +
-      "ta\022$\n\004data\030\002 \001(\0132\026.tensorflow.SavedSlice" +
-      "B8\n\031org.tensorflow.proto.utilB\026SavedTens" +
-      "orSliceProtosP\001\370\001\001b\006proto3"
-    };
-    descriptor = com.google.protobuf.Descriptors.FileDescriptor
-      .internalBuildGeneratedFileFrom(descriptorData,
-        new com.google.protobuf.Descriptors.FileDescriptor[] {
-          org.tensorflow.proto.framework.TensorShapeProtos.getDescriptor(),
-          org.tensorflow.proto.framework.TensorSliceProtos.getDescriptor(),
-          org.tensorflow.proto.framework.TensorProtos.getDescriptor(),
-          org.tensorflow.proto.framework.TypesProtos.getDescriptor(),
-          org.tensorflow.proto.framework.VersionsProtos.getDescriptor(),
-        });
-    internal_static_tensorflow_SavedSliceMeta_descriptor =
-      getDescriptor().getMessageTypes().get(0);
-    internal_static_tensorflow_SavedSliceMeta_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_SavedSliceMeta_descriptor,
-        new java.lang.String[] { "Name", "Shape", "Type", "Slice", });
-    internal_static_tensorflow_SavedTensorSliceMeta_descriptor =
-      getDescriptor().getMessageTypes().get(1);
-    internal_static_tensorflow_SavedTensorSliceMeta_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_SavedTensorSliceMeta_descriptor,
-        new java.lang.String[] { "Tensor", "Versions", });
-    internal_static_tensorflow_SavedSlice_descriptor =
-      getDescriptor().getMessageTypes().get(2);
-    internal_static_tensorflow_SavedSlice_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_SavedSlice_descriptor,
-        new java.lang.String[] { "Name", "Slice", "Data", });
-    internal_static_tensorflow_SavedTensorSlices_descriptor =
-      getDescriptor().getMessageTypes().get(3);
-    internal_static_tensorflow_SavedTensorSlices_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_SavedTensorSlices_descriptor,
-        new java.lang.String[] { "Meta", "Data", });
-    org.tensorflow.proto.framework.TensorShapeProtos.getDescriptor();
-    org.tensorflow.proto.framework.TensorSliceProtos.getDescriptor();
-    org.tensorflow.proto.framework.TensorProtos.getDescriptor();
-    org.tensorflow.proto.framework.TypesProtos.getDescriptor();
-    org.tensorflow.proto.framework.VersionsProtos.getDescriptor();
-  }
-
-  // @@protoc_insertion_point(outer_class_scope)
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SavedTensorSlices.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SavedTensorSlices.java
deleted file mode 100644
index e058ca978ee..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SavedTensorSlices.java
+++ /dev/null
@@ -1,899 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/util/saved_tensor_slice.proto
-
-package org.tensorflow.proto.util;
-
-/**
- * <pre>
- * Each record in a v3 checkpoint file is a serialized SavedTensorSlices
- * message.
- * </pre>
- *
- * Protobuf type {@code tensorflow.SavedTensorSlices}
- */
-public  final class SavedTensorSlices extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.SavedTensorSlices)
-    SavedTensorSlicesOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use SavedTensorSlices.newBuilder() to construct.
-  private SavedTensorSlices(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private SavedTensorSlices() {
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new SavedTensorSlices();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private SavedTensorSlices(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 10: {
-            org.tensorflow.proto.util.SavedTensorSliceMeta.Builder subBuilder = null;
-            if (meta_ != null) {
-              subBuilder = meta_.toBuilder();
-            }
-            meta_ = input.readMessage(org.tensorflow.proto.util.SavedTensorSliceMeta.parser(), extensionRegistry);
-            if (subBuilder != null) {
-              subBuilder.mergeFrom(meta_);
-              meta_ = subBuilder.buildPartial();
-            }
-
-            break;
-          }
-          case 18: {
-            org.tensorflow.proto.util.SavedSlice.Builder subBuilder = null;
-            if (data_ != null) {
-              subBuilder = data_.toBuilder();
-            }
-            data_ = input.readMessage(org.tensorflow.proto.util.SavedSlice.parser(), extensionRegistry);
-            if (subBuilder != null) {
-              subBuilder.mergeFrom(data_);
-              data_ = subBuilder.buildPartial();
-            }
-
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.util.SavedTensorSliceProtos.internal_static_tensorflow_SavedTensorSlices_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.util.SavedTensorSliceProtos.internal_static_tensorflow_SavedTensorSlices_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.util.SavedTensorSlices.class, org.tensorflow.proto.util.SavedTensorSlices.Builder.class);
-  }
-
-  public static final int META_FIELD_NUMBER = 1;
-  private org.tensorflow.proto.util.SavedTensorSliceMeta meta_;
-  /**
-   * <pre>
-   * This is only present at the first item of each checkpoint file and serves
-   * as a table of contents, listing all the tensor slices saved in this file.
-   * </pre>
-   *
-   * <code>.tensorflow.SavedTensorSliceMeta meta = 1;</code>
-   */
-  public boolean hasMeta() {
-    return meta_ != null;
-  }
-  /**
-   * <pre>
-   * This is only present at the first item of each checkpoint file and serves
-   * as a table of contents, listing all the tensor slices saved in this file.
-   * </pre>
-   *
-   * <code>.tensorflow.SavedTensorSliceMeta meta = 1;</code>
-   */
-  public org.tensorflow.proto.util.SavedTensorSliceMeta getMeta() {
-    return meta_ == null ? org.tensorflow.proto.util.SavedTensorSliceMeta.getDefaultInstance() : meta_;
-  }
-  /**
-   * <pre>
-   * This is only present at the first item of each checkpoint file and serves
-   * as a table of contents, listing all the tensor slices saved in this file.
-   * </pre>
-   *
-   * <code>.tensorflow.SavedTensorSliceMeta meta = 1;</code>
-   */
-  public org.tensorflow.proto.util.SavedTensorSliceMetaOrBuilder getMetaOrBuilder() {
-    return getMeta();
-  }
-
-  public static final int DATA_FIELD_NUMBER = 2;
-  private org.tensorflow.proto.util.SavedSlice data_;
-  /**
-   * <pre>
-   * This exists in all but the first item of each checkpoint file.
-   * </pre>
-   *
-   * <code>.tensorflow.SavedSlice data = 2;</code>
-   */
-  public boolean hasData() {
-    return data_ != null;
-  }
-  /**
-   * <pre>
-   * This exists in all but the first item of each checkpoint file.
-   * </pre>
-   *
-   * <code>.tensorflow.SavedSlice data = 2;</code>
-   */
-  public org.tensorflow.proto.util.SavedSlice getData() {
-    return data_ == null ? org.tensorflow.proto.util.SavedSlice.getDefaultInstance() : data_;
-  }
-  /**
-   * <pre>
-   * This exists in all but the first item of each checkpoint file.
-   * </pre>
-   *
-   * <code>.tensorflow.SavedSlice data = 2;</code>
-   */
-  public org.tensorflow.proto.util.SavedSliceOrBuilder getDataOrBuilder() {
-    return getData();
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    if (meta_ != null) {
-      output.writeMessage(1, getMeta());
-    }
-    if (data_ != null) {
-      output.writeMessage(2, getData());
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (meta_ != null) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(1, getMeta());
-    }
-    if (data_ != null) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(2, getData());
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.util.SavedTensorSlices)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.util.SavedTensorSlices other = (org.tensorflow.proto.util.SavedTensorSlices) obj;
-
-    if (hasMeta() != other.hasMeta()) return false;
-    if (hasMeta()) {
-      if (!getMeta()
-          .equals(other.getMeta())) return false;
-    }
-    if (hasData() != other.hasData()) return false;
-    if (hasData()) {
-      if (!getData()
-          .equals(other.getData())) return false;
-    }
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    if (hasMeta()) {
-      hash = (37 * hash) + META_FIELD_NUMBER;
-      hash = (53 * hash) + getMeta().hashCode();
-    }
-    if (hasData()) {
-      hash = (37 * hash) + DATA_FIELD_NUMBER;
-      hash = (53 * hash) + getData().hashCode();
-    }
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.util.SavedTensorSlices parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.SavedTensorSlices parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.SavedTensorSlices parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.SavedTensorSlices parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.SavedTensorSlices parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.SavedTensorSlices parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.SavedTensorSlices parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.SavedTensorSlices parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.SavedTensorSlices parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.SavedTensorSlices parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.SavedTensorSlices parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.SavedTensorSlices parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.util.SavedTensorSlices prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * <pre>
-   * Each record in a v3 checkpoint file is a serialized SavedTensorSlices
-   * message.
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.SavedTensorSlices}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.SavedTensorSlices)
-      org.tensorflow.proto.util.SavedTensorSlicesOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.util.SavedTensorSliceProtos.internal_static_tensorflow_SavedTensorSlices_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.util.SavedTensorSliceProtos.internal_static_tensorflow_SavedTensorSlices_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.util.SavedTensorSlices.class, org.tensorflow.proto.util.SavedTensorSlices.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.util.SavedTensorSlices.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      if (metaBuilder_ == null) {
-        meta_ = null;
-      } else {
-        meta_ = null;
-        metaBuilder_ = null;
-      }
-      if (dataBuilder_ == null) {
-        data_ = null;
-      } else {
-        data_ = null;
-        dataBuilder_ = null;
-      }
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.util.SavedTensorSliceProtos.internal_static_tensorflow_SavedTensorSlices_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.SavedTensorSlices getDefaultInstanceForType() {
-      return org.tensorflow.proto.util.SavedTensorSlices.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.SavedTensorSlices build() {
-      org.tensorflow.proto.util.SavedTensorSlices result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.SavedTensorSlices buildPartial() {
-      org.tensorflow.proto.util.SavedTensorSlices result = new org.tensorflow.proto.util.SavedTensorSlices(this);
-      if (metaBuilder_ == null) {
-        result.meta_ = meta_;
-      } else {
-        result.meta_ = metaBuilder_.build();
-      }
-      if (dataBuilder_ == null) {
-        result.data_ = data_;
-      } else {
-        result.data_ = dataBuilder_.build();
-      }
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.util.SavedTensorSlices) {
-        return mergeFrom((org.tensorflow.proto.util.SavedTensorSlices)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.util.SavedTensorSlices other) {
-      if (other == org.tensorflow.proto.util.SavedTensorSlices.getDefaultInstance()) return this;
-      if (other.hasMeta()) {
-        mergeMeta(other.getMeta());
-      }
-      if (other.hasData()) {
-        mergeData(other.getData());
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.util.SavedTensorSlices parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.util.SavedTensorSlices) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-
-    private org.tensorflow.proto.util.SavedTensorSliceMeta meta_;
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.util.SavedTensorSliceMeta, org.tensorflow.proto.util.SavedTensorSliceMeta.Builder, org.tensorflow.proto.util.SavedTensorSliceMetaOrBuilder> metaBuilder_;
-    /**
-     * <pre>
-     * This is only present at the first item of each checkpoint file and serves
-     * as a table of contents, listing all the tensor slices saved in this file.
-     * </pre>
-     *
-     * <code>.tensorflow.SavedTensorSliceMeta meta = 1;</code>
-     */
-    public boolean hasMeta() {
-      return metaBuilder_ != null || meta_ != null;
-    }
-    /**
-     * <pre>
-     * This is only present at the first item of each checkpoint file and serves
-     * as a table of contents, listing all the tensor slices saved in this file.
-     * </pre>
-     *
-     * <code>.tensorflow.SavedTensorSliceMeta meta = 1;</code>
-     */
-    public org.tensorflow.proto.util.SavedTensorSliceMeta getMeta() {
-      if (metaBuilder_ == null) {
-        return meta_ == null ? org.tensorflow.proto.util.SavedTensorSliceMeta.getDefaultInstance() : meta_;
-      } else {
-        return metaBuilder_.getMessage();
-      }
-    }
-    /**
-     * <pre>
-     * This is only present at the first item of each checkpoint file and serves
-     * as a table of contents, listing all the tensor slices saved in this file.
-     * </pre>
-     *
-     * <code>.tensorflow.SavedTensorSliceMeta meta = 1;</code>
-     */
-    public Builder setMeta(org.tensorflow.proto.util.SavedTensorSliceMeta value) {
-      if (metaBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        meta_ = value;
-        onChanged();
-      } else {
-        metaBuilder_.setMessage(value);
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * This is only present at the first item of each checkpoint file and serves
-     * as a table of contents, listing all the tensor slices saved in this file.
-     * </pre>
-     *
-     * <code>.tensorflow.SavedTensorSliceMeta meta = 1;</code>
-     */
-    public Builder setMeta(
-        org.tensorflow.proto.util.SavedTensorSliceMeta.Builder builderForValue) {
-      if (metaBuilder_ == null) {
-        meta_ = builderForValue.build();
-        onChanged();
-      } else {
-        metaBuilder_.setMessage(builderForValue.build());
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * This is only present at the first item of each checkpoint file and serves
-     * as a table of contents, listing all the tensor slices saved in this file.
-     * </pre>
-     *
-     * <code>.tensorflow.SavedTensorSliceMeta meta = 1;</code>
-     */
-    public Builder mergeMeta(org.tensorflow.proto.util.SavedTensorSliceMeta value) {
-      if (metaBuilder_ == null) {
-        if (meta_ != null) {
-          meta_ =
-            org.tensorflow.proto.util.SavedTensorSliceMeta.newBuilder(meta_).mergeFrom(value).buildPartial();
-        } else {
-          meta_ = value;
-        }
-        onChanged();
-      } else {
-        metaBuilder_.mergeFrom(value);
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * This is only present at the first item of each checkpoint file and serves
-     * as a table of contents, listing all the tensor slices saved in this file.
-     * </pre>
-     *
-     * <code>.tensorflow.SavedTensorSliceMeta meta = 1;</code>
-     */
-    public Builder clearMeta() {
-      if (metaBuilder_ == null) {
-        meta_ = null;
-        onChanged();
-      } else {
-        meta_ = null;
-        metaBuilder_ = null;
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * This is only present at the first item of each checkpoint file and serves
-     * as a table of contents, listing all the tensor slices saved in this file.
-     * </pre>
-     *
-     * <code>.tensorflow.SavedTensorSliceMeta meta = 1;</code>
-     */
-    public org.tensorflow.proto.util.SavedTensorSliceMeta.Builder getMetaBuilder() {
-      
-      onChanged();
-      return getMetaFieldBuilder().getBuilder();
-    }
-    /**
-     * <pre>
-     * This is only present at the first item of each checkpoint file and serves
-     * as a table of contents, listing all the tensor slices saved in this file.
-     * </pre>
-     *
-     * <code>.tensorflow.SavedTensorSliceMeta meta = 1;</code>
-     */
-    public org.tensorflow.proto.util.SavedTensorSliceMetaOrBuilder getMetaOrBuilder() {
-      if (metaBuilder_ != null) {
-        return metaBuilder_.getMessageOrBuilder();
-      } else {
-        return meta_ == null ?
-            org.tensorflow.proto.util.SavedTensorSliceMeta.getDefaultInstance() : meta_;
-      }
-    }
-    /**
-     * <pre>
-     * This is only present at the first item of each checkpoint file and serves
-     * as a table of contents, listing all the tensor slices saved in this file.
-     * </pre>
-     *
-     * <code>.tensorflow.SavedTensorSliceMeta meta = 1;</code>
-     */
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.util.SavedTensorSliceMeta, org.tensorflow.proto.util.SavedTensorSliceMeta.Builder, org.tensorflow.proto.util.SavedTensorSliceMetaOrBuilder> 
-        getMetaFieldBuilder() {
-      if (metaBuilder_ == null) {
-        metaBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-            org.tensorflow.proto.util.SavedTensorSliceMeta, org.tensorflow.proto.util.SavedTensorSliceMeta.Builder, org.tensorflow.proto.util.SavedTensorSliceMetaOrBuilder>(
-                getMeta(),
-                getParentForChildren(),
-                isClean());
-        meta_ = null;
-      }
-      return metaBuilder_;
-    }
-
-    private org.tensorflow.proto.util.SavedSlice data_;
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.util.SavedSlice, org.tensorflow.proto.util.SavedSlice.Builder, org.tensorflow.proto.util.SavedSliceOrBuilder> dataBuilder_;
-    /**
-     * <pre>
-     * This exists in all but the first item of each checkpoint file.
-     * </pre>
-     *
-     * <code>.tensorflow.SavedSlice data = 2;</code>
-     */
-    public boolean hasData() {
-      return dataBuilder_ != null || data_ != null;
-    }
-    /**
-     * <pre>
-     * This exists in all but the first item of each checkpoint file.
-     * </pre>
-     *
-     * <code>.tensorflow.SavedSlice data = 2;</code>
-     */
-    public org.tensorflow.proto.util.SavedSlice getData() {
-      if (dataBuilder_ == null) {
-        return data_ == null ? org.tensorflow.proto.util.SavedSlice.getDefaultInstance() : data_;
-      } else {
-        return dataBuilder_.getMessage();
-      }
-    }
-    /**
-     * <pre>
-     * This exists in all but the first item of each checkpoint file.
-     * </pre>
-     *
-     * <code>.tensorflow.SavedSlice data = 2;</code>
-     */
-    public Builder setData(org.tensorflow.proto.util.SavedSlice value) {
-      if (dataBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        data_ = value;
-        onChanged();
-      } else {
-        dataBuilder_.setMessage(value);
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * This exists in all but the first item of each checkpoint file.
-     * </pre>
-     *
-     * <code>.tensorflow.SavedSlice data = 2;</code>
-     */
-    public Builder setData(
-        org.tensorflow.proto.util.SavedSlice.Builder builderForValue) {
-      if (dataBuilder_ == null) {
-        data_ = builderForValue.build();
-        onChanged();
-      } else {
-        dataBuilder_.setMessage(builderForValue.build());
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * This exists in all but the first item of each checkpoint file.
-     * </pre>
-     *
-     * <code>.tensorflow.SavedSlice data = 2;</code>
-     */
-    public Builder mergeData(org.tensorflow.proto.util.SavedSlice value) {
-      if (dataBuilder_ == null) {
-        if (data_ != null) {
-          data_ =
-            org.tensorflow.proto.util.SavedSlice.newBuilder(data_).mergeFrom(value).buildPartial();
-        } else {
-          data_ = value;
-        }
-        onChanged();
-      } else {
-        dataBuilder_.mergeFrom(value);
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * This exists in all but the first item of each checkpoint file.
-     * </pre>
-     *
-     * <code>.tensorflow.SavedSlice data = 2;</code>
-     */
-    public Builder clearData() {
-      if (dataBuilder_ == null) {
-        data_ = null;
-        onChanged();
-      } else {
-        data_ = null;
-        dataBuilder_ = null;
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * This exists in all but the first item of each checkpoint file.
-     * </pre>
-     *
-     * <code>.tensorflow.SavedSlice data = 2;</code>
-     */
-    public org.tensorflow.proto.util.SavedSlice.Builder getDataBuilder() {
-      
-      onChanged();
-      return getDataFieldBuilder().getBuilder();
-    }
-    /**
-     * <pre>
-     * This exists in all but the first item of each checkpoint file.
-     * </pre>
-     *
-     * <code>.tensorflow.SavedSlice data = 2;</code>
-     */
-    public org.tensorflow.proto.util.SavedSliceOrBuilder getDataOrBuilder() {
-      if (dataBuilder_ != null) {
-        return dataBuilder_.getMessageOrBuilder();
-      } else {
-        return data_ == null ?
-            org.tensorflow.proto.util.SavedSlice.getDefaultInstance() : data_;
-      }
-    }
-    /**
-     * <pre>
-     * This exists in all but the first item of each checkpoint file.
-     * </pre>
-     *
-     * <code>.tensorflow.SavedSlice data = 2;</code>
-     */
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.util.SavedSlice, org.tensorflow.proto.util.SavedSlice.Builder, org.tensorflow.proto.util.SavedSliceOrBuilder> 
-        getDataFieldBuilder() {
-      if (dataBuilder_ == null) {
-        dataBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-            org.tensorflow.proto.util.SavedSlice, org.tensorflow.proto.util.SavedSlice.Builder, org.tensorflow.proto.util.SavedSliceOrBuilder>(
-                getData(),
-                getParentForChildren(),
-                isClean());
-        data_ = null;
-      }
-      return dataBuilder_;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.SavedTensorSlices)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.SavedTensorSlices)
-  private static final org.tensorflow.proto.util.SavedTensorSlices DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.util.SavedTensorSlices();
-  }
-
-  public static org.tensorflow.proto.util.SavedTensorSlices getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<SavedTensorSlices>
-      PARSER = new com.google.protobuf.AbstractParser<SavedTensorSlices>() {
-    @java.lang.Override
-    public SavedTensorSlices parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new SavedTensorSlices(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<SavedTensorSlices> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<SavedTensorSlices> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.util.SavedTensorSlices getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SavedTensorSlicesOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SavedTensorSlicesOrBuilder.java
deleted file mode 100644
index f8b12b1c7dd..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SavedTensorSlicesOrBuilder.java
+++ /dev/null
@@ -1,62 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/util/saved_tensor_slice.proto
-
-package org.tensorflow.proto.util;
-
-public interface SavedTensorSlicesOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.SavedTensorSlices)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <pre>
-   * This is only present at the first item of each checkpoint file and serves
-   * as a table of contents, listing all the tensor slices saved in this file.
-   * </pre>
-   *
-   * <code>.tensorflow.SavedTensorSliceMeta meta = 1;</code>
-   */
-  boolean hasMeta();
-  /**
-   * <pre>
-   * This is only present at the first item of each checkpoint file and serves
-   * as a table of contents, listing all the tensor slices saved in this file.
-   * </pre>
-   *
-   * <code>.tensorflow.SavedTensorSliceMeta meta = 1;</code>
-   */
-  org.tensorflow.proto.util.SavedTensorSliceMeta getMeta();
-  /**
-   * <pre>
-   * This is only present at the first item of each checkpoint file and serves
-   * as a table of contents, listing all the tensor slices saved in this file.
-   * </pre>
-   *
-   * <code>.tensorflow.SavedTensorSliceMeta meta = 1;</code>
-   */
-  org.tensorflow.proto.util.SavedTensorSliceMetaOrBuilder getMetaOrBuilder();
-
-  /**
-   * <pre>
-   * This exists in all but the first item of each checkpoint file.
-   * </pre>
-   *
-   * <code>.tensorflow.SavedSlice data = 2;</code>
-   */
-  boolean hasData();
-  /**
-   * <pre>
-   * This exists in all but the first item of each checkpoint file.
-   * </pre>
-   *
-   * <code>.tensorflow.SavedSlice data = 2;</code>
-   */
-  org.tensorflow.proto.util.SavedSlice getData();
-  /**
-   * <pre>
-   * This exists in all but the first item of each checkpoint file.
-   * </pre>
-   *
-   * <code>.tensorflow.SavedSlice data = 2;</code>
-   */
-  org.tensorflow.proto.util.SavedSliceOrBuilder getDataOrBuilder();
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SaverDef.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SaverDef.java
deleted file mode 100644
index daf4a6c91e4..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SaverDef.java
+++ /dev/null
@@ -1,1356 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/saver.proto
-
-package org.tensorflow.proto.util;
-
-/**
- * <pre>
- * Protocol buffer representing the configuration of a Saver.
- * </pre>
- *
- * Protobuf type {@code tensorflow.SaverDef}
- */
-public  final class SaverDef extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.SaverDef)
-    SaverDefOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use SaverDef.newBuilder() to construct.
-  private SaverDef(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private SaverDef() {
-    filenameTensorName_ = "";
-    saveTensorName_ = "";
-    restoreOpName_ = "";
-    version_ = 0;
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new SaverDef();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private SaverDef(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 10: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            filenameTensorName_ = s;
-            break;
-          }
-          case 18: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            saveTensorName_ = s;
-            break;
-          }
-          case 26: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            restoreOpName_ = s;
-            break;
-          }
-          case 32: {
-
-            maxToKeep_ = input.readInt32();
-            break;
-          }
-          case 40: {
-
-            sharded_ = input.readBool();
-            break;
-          }
-          case 53: {
-
-            keepCheckpointEveryNHours_ = input.readFloat();
-            break;
-          }
-          case 56: {
-            int rawValue = input.readEnum();
-
-            version_ = rawValue;
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.util.SaverProtos.internal_static_tensorflow_SaverDef_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.util.SaverProtos.internal_static_tensorflow_SaverDef_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.util.SaverDef.class, org.tensorflow.proto.util.SaverDef.Builder.class);
-  }
-
-  /**
-   * <pre>
-   * A version number that identifies a different on-disk checkpoint format.
-   * Usually, each subclass of BaseSaverBuilder works with a particular
-   * version/format.  However, it is possible that the same builder may be
-   * upgraded to support a newer checkpoint format in the future.
-   * </pre>
-   *
-   * Protobuf enum {@code tensorflow.SaverDef.CheckpointFormatVersion}
-   */
-  public enum CheckpointFormatVersion
-      implements com.google.protobuf.ProtocolMessageEnum {
-    /**
-     * <pre>
-     * Internal legacy format.
-     * </pre>
-     *
-     * <code>LEGACY = 0;</code>
-     */
-    LEGACY(0),
-    /**
-     * <pre>
-     * Deprecated format: tf.Saver() which works with tensorflow::table::Table.
-     * </pre>
-     *
-     * <code>V1 = 1;</code>
-     */
-    V1(1),
-    /**
-     * <pre>
-     * Current format: more efficient.
-     * </pre>
-     *
-     * <code>V2 = 2;</code>
-     */
-    V2(2),
-    UNRECOGNIZED(-1),
-    ;
-
-    /**
-     * <pre>
-     * Internal legacy format.
-     * </pre>
-     *
-     * <code>LEGACY = 0;</code>
-     */
-    public static final int LEGACY_VALUE = 0;
-    /**
-     * <pre>
-     * Deprecated format: tf.Saver() which works with tensorflow::table::Table.
-     * </pre>
-     *
-     * <code>V1 = 1;</code>
-     */
-    public static final int V1_VALUE = 1;
-    /**
-     * <pre>
-     * Current format: more efficient.
-     * </pre>
-     *
-     * <code>V2 = 2;</code>
-     */
-    public static final int V2_VALUE = 2;
-
-
-    public final int getNumber() {
-      if (this == UNRECOGNIZED) {
-        throw new java.lang.IllegalArgumentException(
-            "Can't get the number of an unknown enum value.");
-      }
-      return value;
-    }
-
-    /**
-     * @deprecated Use {@link #forNumber(int)} instead.
-     */
-    @java.lang.Deprecated
-    public static CheckpointFormatVersion valueOf(int value) {
-      return forNumber(value);
-    }
-
-    public static CheckpointFormatVersion forNumber(int value) {
-      switch (value) {
-        case 0: return LEGACY;
-        case 1: return V1;
-        case 2: return V2;
-        default: return null;
-      }
-    }
-
-    public static com.google.protobuf.Internal.EnumLiteMap<CheckpointFormatVersion>
-        internalGetValueMap() {
-      return internalValueMap;
-    }
-    private static final com.google.protobuf.Internal.EnumLiteMap<
-        CheckpointFormatVersion> internalValueMap =
-          new com.google.protobuf.Internal.EnumLiteMap<CheckpointFormatVersion>() {
-            public CheckpointFormatVersion findValueByNumber(int number) {
-              return CheckpointFormatVersion.forNumber(number);
-            }
-          };
-
-    public final com.google.protobuf.Descriptors.EnumValueDescriptor
-        getValueDescriptor() {
-      return getDescriptor().getValues().get(ordinal());
-    }
-    public final com.google.protobuf.Descriptors.EnumDescriptor
-        getDescriptorForType() {
-      return getDescriptor();
-    }
-    public static final com.google.protobuf.Descriptors.EnumDescriptor
-        getDescriptor() {
-      return org.tensorflow.proto.util.SaverDef.getDescriptor().getEnumTypes().get(0);
-    }
-
-    private static final CheckpointFormatVersion[] VALUES = values();
-
-    public static CheckpointFormatVersion valueOf(
-        com.google.protobuf.Descriptors.EnumValueDescriptor desc) {
-      if (desc.getType() != getDescriptor()) {
-        throw new java.lang.IllegalArgumentException(
-          "EnumValueDescriptor is not for this type.");
-      }
-      if (desc.getIndex() == -1) {
-        return UNRECOGNIZED;
-      }
-      return VALUES[desc.getIndex()];
-    }
-
-    private final int value;
-
-    private CheckpointFormatVersion(int value) {
-      this.value = value;
-    }
-
-    // @@protoc_insertion_point(enum_scope:tensorflow.SaverDef.CheckpointFormatVersion)
-  }
-
-  public static final int FILENAME_TENSOR_NAME_FIELD_NUMBER = 1;
-  private volatile java.lang.Object filenameTensorName_;
-  /**
-   * <pre>
-   * The name of the tensor in which to specify the filename when saving or
-   * restoring a model checkpoint.
-   * </pre>
-   *
-   * <code>string filename_tensor_name = 1;</code>
-   */
-  public java.lang.String getFilenameTensorName() {
-    java.lang.Object ref = filenameTensorName_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      filenameTensorName_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * The name of the tensor in which to specify the filename when saving or
-   * restoring a model checkpoint.
-   * </pre>
-   *
-   * <code>string filename_tensor_name = 1;</code>
-   */
-  public com.google.protobuf.ByteString
-      getFilenameTensorNameBytes() {
-    java.lang.Object ref = filenameTensorName_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      filenameTensorName_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int SAVE_TENSOR_NAME_FIELD_NUMBER = 2;
-  private volatile java.lang.Object saveTensorName_;
-  /**
-   * <pre>
-   * The operation to run when saving a model checkpoint.
-   * </pre>
-   *
-   * <code>string save_tensor_name = 2;</code>
-   */
-  public java.lang.String getSaveTensorName() {
-    java.lang.Object ref = saveTensorName_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      saveTensorName_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * The operation to run when saving a model checkpoint.
-   * </pre>
-   *
-   * <code>string save_tensor_name = 2;</code>
-   */
-  public com.google.protobuf.ByteString
-      getSaveTensorNameBytes() {
-    java.lang.Object ref = saveTensorName_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      saveTensorName_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int RESTORE_OP_NAME_FIELD_NUMBER = 3;
-  private volatile java.lang.Object restoreOpName_;
-  /**
-   * <pre>
-   * The operation to run when restoring a model checkpoint.
-   * </pre>
-   *
-   * <code>string restore_op_name = 3;</code>
-   */
-  public java.lang.String getRestoreOpName() {
-    java.lang.Object ref = restoreOpName_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      restoreOpName_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * The operation to run when restoring a model checkpoint.
-   * </pre>
-   *
-   * <code>string restore_op_name = 3;</code>
-   */
-  public com.google.protobuf.ByteString
-      getRestoreOpNameBytes() {
-    java.lang.Object ref = restoreOpName_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      restoreOpName_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int MAX_TO_KEEP_FIELD_NUMBER = 4;
-  private int maxToKeep_;
-  /**
-   * <pre>
-   * Maximum number of checkpoints to keep.  If 0, no checkpoints are deleted.
-   * </pre>
-   *
-   * <code>int32 max_to_keep = 4;</code>
-   */
-  public int getMaxToKeep() {
-    return maxToKeep_;
-  }
-
-  public static final int SHARDED_FIELD_NUMBER = 5;
-  private boolean sharded_;
-  /**
-   * <pre>
-   * Shard the save files, one per device that has Variable nodes.
-   * </pre>
-   *
-   * <code>bool sharded = 5;</code>
-   */
-  public boolean getSharded() {
-    return sharded_;
-  }
-
-  public static final int KEEP_CHECKPOINT_EVERY_N_HOURS_FIELD_NUMBER = 6;
-  private float keepCheckpointEveryNHours_;
-  /**
-   * <pre>
-   * How often to keep an additional checkpoint. If not specified, only the last
-   * "max_to_keep" checkpoints are kept; if specified, in addition to keeping
-   * the last "max_to_keep" checkpoints, an additional checkpoint will be kept
-   * for every n hours of training.
-   * </pre>
-   *
-   * <code>float keep_checkpoint_every_n_hours = 6;</code>
-   */
-  public float getKeepCheckpointEveryNHours() {
-    return keepCheckpointEveryNHours_;
-  }
-
-  public static final int VERSION_FIELD_NUMBER = 7;
-  private int version_;
-  /**
-   * <code>.tensorflow.SaverDef.CheckpointFormatVersion version = 7;</code>
-   */
-  public int getVersionValue() {
-    return version_;
-  }
-  /**
-   * <code>.tensorflow.SaverDef.CheckpointFormatVersion version = 7;</code>
-   */
-  public org.tensorflow.proto.util.SaverDef.CheckpointFormatVersion getVersion() {
-    @SuppressWarnings("deprecation")
-    org.tensorflow.proto.util.SaverDef.CheckpointFormatVersion result = org.tensorflow.proto.util.SaverDef.CheckpointFormatVersion.valueOf(version_);
-    return result == null ? org.tensorflow.proto.util.SaverDef.CheckpointFormatVersion.UNRECOGNIZED : result;
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    if (!getFilenameTensorNameBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 1, filenameTensorName_);
-    }
-    if (!getSaveTensorNameBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 2, saveTensorName_);
-    }
-    if (!getRestoreOpNameBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 3, restoreOpName_);
-    }
-    if (maxToKeep_ != 0) {
-      output.writeInt32(4, maxToKeep_);
-    }
-    if (sharded_ != false) {
-      output.writeBool(5, sharded_);
-    }
-    if (keepCheckpointEveryNHours_ != 0F) {
-      output.writeFloat(6, keepCheckpointEveryNHours_);
-    }
-    if (version_ != org.tensorflow.proto.util.SaverDef.CheckpointFormatVersion.LEGACY.getNumber()) {
-      output.writeEnum(7, version_);
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (!getFilenameTensorNameBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, filenameTensorName_);
-    }
-    if (!getSaveTensorNameBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(2, saveTensorName_);
-    }
-    if (!getRestoreOpNameBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(3, restoreOpName_);
-    }
-    if (maxToKeep_ != 0) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt32Size(4, maxToKeep_);
-    }
-    if (sharded_ != false) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeBoolSize(5, sharded_);
-    }
-    if (keepCheckpointEveryNHours_ != 0F) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeFloatSize(6, keepCheckpointEveryNHours_);
-    }
-    if (version_ != org.tensorflow.proto.util.SaverDef.CheckpointFormatVersion.LEGACY.getNumber()) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeEnumSize(7, version_);
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.util.SaverDef)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.util.SaverDef other = (org.tensorflow.proto.util.SaverDef) obj;
-
-    if (!getFilenameTensorName()
-        .equals(other.getFilenameTensorName())) return false;
-    if (!getSaveTensorName()
-        .equals(other.getSaveTensorName())) return false;
-    if (!getRestoreOpName()
-        .equals(other.getRestoreOpName())) return false;
-    if (getMaxToKeep()
-        != other.getMaxToKeep()) return false;
-    if (getSharded()
-        != other.getSharded()) return false;
-    if (java.lang.Float.floatToIntBits(getKeepCheckpointEveryNHours())
-        != java.lang.Float.floatToIntBits(
-            other.getKeepCheckpointEveryNHours())) return false;
-    if (version_ != other.version_) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    hash = (37 * hash) + FILENAME_TENSOR_NAME_FIELD_NUMBER;
-    hash = (53 * hash) + getFilenameTensorName().hashCode();
-    hash = (37 * hash) + SAVE_TENSOR_NAME_FIELD_NUMBER;
-    hash = (53 * hash) + getSaveTensorName().hashCode();
-    hash = (37 * hash) + RESTORE_OP_NAME_FIELD_NUMBER;
-    hash = (53 * hash) + getRestoreOpName().hashCode();
-    hash = (37 * hash) + MAX_TO_KEEP_FIELD_NUMBER;
-    hash = (53 * hash) + getMaxToKeep();
-    hash = (37 * hash) + SHARDED_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashBoolean(
-        getSharded());
-    hash = (37 * hash) + KEEP_CHECKPOINT_EVERY_N_HOURS_FIELD_NUMBER;
-    hash = (53 * hash) + java.lang.Float.floatToIntBits(
-        getKeepCheckpointEveryNHours());
-    hash = (37 * hash) + VERSION_FIELD_NUMBER;
-    hash = (53 * hash) + version_;
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.util.SaverDef parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.SaverDef parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.SaverDef parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.SaverDef parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.SaverDef parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.SaverDef parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.SaverDef parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.SaverDef parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.SaverDef parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.SaverDef parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.SaverDef parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.SaverDef parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.util.SaverDef prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * <pre>
-   * Protocol buffer representing the configuration of a Saver.
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.SaverDef}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.SaverDef)
-      org.tensorflow.proto.util.SaverDefOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.util.SaverProtos.internal_static_tensorflow_SaverDef_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.util.SaverProtos.internal_static_tensorflow_SaverDef_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.util.SaverDef.class, org.tensorflow.proto.util.SaverDef.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.util.SaverDef.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      filenameTensorName_ = "";
-
-      saveTensorName_ = "";
-
-      restoreOpName_ = "";
-
-      maxToKeep_ = 0;
-
-      sharded_ = false;
-
-      keepCheckpointEveryNHours_ = 0F;
-
-      version_ = 0;
-
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.util.SaverProtos.internal_static_tensorflow_SaverDef_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.SaverDef getDefaultInstanceForType() {
-      return org.tensorflow.proto.util.SaverDef.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.SaverDef build() {
-      org.tensorflow.proto.util.SaverDef result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.SaverDef buildPartial() {
-      org.tensorflow.proto.util.SaverDef result = new org.tensorflow.proto.util.SaverDef(this);
-      result.filenameTensorName_ = filenameTensorName_;
-      result.saveTensorName_ = saveTensorName_;
-      result.restoreOpName_ = restoreOpName_;
-      result.maxToKeep_ = maxToKeep_;
-      result.sharded_ = sharded_;
-      result.keepCheckpointEveryNHours_ = keepCheckpointEveryNHours_;
-      result.version_ = version_;
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.util.SaverDef) {
-        return mergeFrom((org.tensorflow.proto.util.SaverDef)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.util.SaverDef other) {
-      if (other == org.tensorflow.proto.util.SaverDef.getDefaultInstance()) return this;
-      if (!other.getFilenameTensorName().isEmpty()) {
-        filenameTensorName_ = other.filenameTensorName_;
-        onChanged();
-      }
-      if (!other.getSaveTensorName().isEmpty()) {
-        saveTensorName_ = other.saveTensorName_;
-        onChanged();
-      }
-      if (!other.getRestoreOpName().isEmpty()) {
-        restoreOpName_ = other.restoreOpName_;
-        onChanged();
-      }
-      if (other.getMaxToKeep() != 0) {
-        setMaxToKeep(other.getMaxToKeep());
-      }
-      if (other.getSharded() != false) {
-        setSharded(other.getSharded());
-      }
-      if (other.getKeepCheckpointEveryNHours() != 0F) {
-        setKeepCheckpointEveryNHours(other.getKeepCheckpointEveryNHours());
-      }
-      if (other.version_ != 0) {
-        setVersionValue(other.getVersionValue());
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.util.SaverDef parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.util.SaverDef) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-
-    private java.lang.Object filenameTensorName_ = "";
-    /**
-     * <pre>
-     * The name of the tensor in which to specify the filename when saving or
-     * restoring a model checkpoint.
-     * </pre>
-     *
-     * <code>string filename_tensor_name = 1;</code>
-     */
-    public java.lang.String getFilenameTensorName() {
-      java.lang.Object ref = filenameTensorName_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        filenameTensorName_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * The name of the tensor in which to specify the filename when saving or
-     * restoring a model checkpoint.
-     * </pre>
-     *
-     * <code>string filename_tensor_name = 1;</code>
-     */
-    public com.google.protobuf.ByteString
-        getFilenameTensorNameBytes() {
-      java.lang.Object ref = filenameTensorName_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        filenameTensorName_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * The name of the tensor in which to specify the filename when saving or
-     * restoring a model checkpoint.
-     * </pre>
-     *
-     * <code>string filename_tensor_name = 1;</code>
-     */
-    public Builder setFilenameTensorName(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      filenameTensorName_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * The name of the tensor in which to specify the filename when saving or
-     * restoring a model checkpoint.
-     * </pre>
-     *
-     * <code>string filename_tensor_name = 1;</code>
-     */
-    public Builder clearFilenameTensorName() {
-      
-      filenameTensorName_ = getDefaultInstance().getFilenameTensorName();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * The name of the tensor in which to specify the filename when saving or
-     * restoring a model checkpoint.
-     * </pre>
-     *
-     * <code>string filename_tensor_name = 1;</code>
-     */
-    public Builder setFilenameTensorNameBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      filenameTensorName_ = value;
-      onChanged();
-      return this;
-    }
-
-    private java.lang.Object saveTensorName_ = "";
-    /**
-     * <pre>
-     * The operation to run when saving a model checkpoint.
-     * </pre>
-     *
-     * <code>string save_tensor_name = 2;</code>
-     */
-    public java.lang.String getSaveTensorName() {
-      java.lang.Object ref = saveTensorName_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        saveTensorName_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * The operation to run when saving a model checkpoint.
-     * </pre>
-     *
-     * <code>string save_tensor_name = 2;</code>
-     */
-    public com.google.protobuf.ByteString
-        getSaveTensorNameBytes() {
-      java.lang.Object ref = saveTensorName_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        saveTensorName_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * The operation to run when saving a model checkpoint.
-     * </pre>
-     *
-     * <code>string save_tensor_name = 2;</code>
-     */
-    public Builder setSaveTensorName(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      saveTensorName_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * The operation to run when saving a model checkpoint.
-     * </pre>
-     *
-     * <code>string save_tensor_name = 2;</code>
-     */
-    public Builder clearSaveTensorName() {
-      
-      saveTensorName_ = getDefaultInstance().getSaveTensorName();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * The operation to run when saving a model checkpoint.
-     * </pre>
-     *
-     * <code>string save_tensor_name = 2;</code>
-     */
-    public Builder setSaveTensorNameBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      saveTensorName_ = value;
-      onChanged();
-      return this;
-    }
-
-    private java.lang.Object restoreOpName_ = "";
-    /**
-     * <pre>
-     * The operation to run when restoring a model checkpoint.
-     * </pre>
-     *
-     * <code>string restore_op_name = 3;</code>
-     */
-    public java.lang.String getRestoreOpName() {
-      java.lang.Object ref = restoreOpName_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        restoreOpName_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * The operation to run when restoring a model checkpoint.
-     * </pre>
-     *
-     * <code>string restore_op_name = 3;</code>
-     */
-    public com.google.protobuf.ByteString
-        getRestoreOpNameBytes() {
-      java.lang.Object ref = restoreOpName_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        restoreOpName_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * The operation to run when restoring a model checkpoint.
-     * </pre>
-     *
-     * <code>string restore_op_name = 3;</code>
-     */
-    public Builder setRestoreOpName(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      restoreOpName_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * The operation to run when restoring a model checkpoint.
-     * </pre>
-     *
-     * <code>string restore_op_name = 3;</code>
-     */
-    public Builder clearRestoreOpName() {
-      
-      restoreOpName_ = getDefaultInstance().getRestoreOpName();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * The operation to run when restoring a model checkpoint.
-     * </pre>
-     *
-     * <code>string restore_op_name = 3;</code>
-     */
-    public Builder setRestoreOpNameBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      restoreOpName_ = value;
-      onChanged();
-      return this;
-    }
-
-    private int maxToKeep_ ;
-    /**
-     * <pre>
-     * Maximum number of checkpoints to keep.  If 0, no checkpoints are deleted.
-     * </pre>
-     *
-     * <code>int32 max_to_keep = 4;</code>
-     */
-    public int getMaxToKeep() {
-      return maxToKeep_;
-    }
-    /**
-     * <pre>
-     * Maximum number of checkpoints to keep.  If 0, no checkpoints are deleted.
-     * </pre>
-     *
-     * <code>int32 max_to_keep = 4;</code>
-     */
-    public Builder setMaxToKeep(int value) {
-      
-      maxToKeep_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Maximum number of checkpoints to keep.  If 0, no checkpoints are deleted.
-     * </pre>
-     *
-     * <code>int32 max_to_keep = 4;</code>
-     */
-    public Builder clearMaxToKeep() {
-      
-      maxToKeep_ = 0;
-      onChanged();
-      return this;
-    }
-
-    private boolean sharded_ ;
-    /**
-     * <pre>
-     * Shard the save files, one per device that has Variable nodes.
-     * </pre>
-     *
-     * <code>bool sharded = 5;</code>
-     */
-    public boolean getSharded() {
-      return sharded_;
-    }
-    /**
-     * <pre>
-     * Shard the save files, one per device that has Variable nodes.
-     * </pre>
-     *
-     * <code>bool sharded = 5;</code>
-     */
-    public Builder setSharded(boolean value) {
-      
-      sharded_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Shard the save files, one per device that has Variable nodes.
-     * </pre>
-     *
-     * <code>bool sharded = 5;</code>
-     */
-    public Builder clearSharded() {
-      
-      sharded_ = false;
-      onChanged();
-      return this;
-    }
-
-    private float keepCheckpointEveryNHours_ ;
-    /**
-     * <pre>
-     * How often to keep an additional checkpoint. If not specified, only the last
-     * "max_to_keep" checkpoints are kept; if specified, in addition to keeping
-     * the last "max_to_keep" checkpoints, an additional checkpoint will be kept
-     * for every n hours of training.
-     * </pre>
-     *
-     * <code>float keep_checkpoint_every_n_hours = 6;</code>
-     */
-    public float getKeepCheckpointEveryNHours() {
-      return keepCheckpointEveryNHours_;
-    }
-    /**
-     * <pre>
-     * How often to keep an additional checkpoint. If not specified, only the last
-     * "max_to_keep" checkpoints are kept; if specified, in addition to keeping
-     * the last "max_to_keep" checkpoints, an additional checkpoint will be kept
-     * for every n hours of training.
-     * </pre>
-     *
-     * <code>float keep_checkpoint_every_n_hours = 6;</code>
-     */
-    public Builder setKeepCheckpointEveryNHours(float value) {
-      
-      keepCheckpointEveryNHours_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * How often to keep an additional checkpoint. If not specified, only the last
-     * "max_to_keep" checkpoints are kept; if specified, in addition to keeping
-     * the last "max_to_keep" checkpoints, an additional checkpoint will be kept
-     * for every n hours of training.
-     * </pre>
-     *
-     * <code>float keep_checkpoint_every_n_hours = 6;</code>
-     */
-    public Builder clearKeepCheckpointEveryNHours() {
-      
-      keepCheckpointEveryNHours_ = 0F;
-      onChanged();
-      return this;
-    }
-
-    private int version_ = 0;
-    /**
-     * <code>.tensorflow.SaverDef.CheckpointFormatVersion version = 7;</code>
-     */
-    public int getVersionValue() {
-      return version_;
-    }
-    /**
-     * <code>.tensorflow.SaverDef.CheckpointFormatVersion version = 7;</code>
-     */
-    public Builder setVersionValue(int value) {
-      version_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>.tensorflow.SaverDef.CheckpointFormatVersion version = 7;</code>
-     */
-    public org.tensorflow.proto.util.SaverDef.CheckpointFormatVersion getVersion() {
-      @SuppressWarnings("deprecation")
-      org.tensorflow.proto.util.SaverDef.CheckpointFormatVersion result = org.tensorflow.proto.util.SaverDef.CheckpointFormatVersion.valueOf(version_);
-      return result == null ? org.tensorflow.proto.util.SaverDef.CheckpointFormatVersion.UNRECOGNIZED : result;
-    }
-    /**
-     * <code>.tensorflow.SaverDef.CheckpointFormatVersion version = 7;</code>
-     */
-    public Builder setVersion(org.tensorflow.proto.util.SaverDef.CheckpointFormatVersion value) {
-      if (value == null) {
-        throw new NullPointerException();
-      }
-      
-      version_ = value.getNumber();
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>.tensorflow.SaverDef.CheckpointFormatVersion version = 7;</code>
-     */
-    public Builder clearVersion() {
-      
-      version_ = 0;
-      onChanged();
-      return this;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.SaverDef)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.SaverDef)
-  private static final org.tensorflow.proto.util.SaverDef DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.util.SaverDef();
-  }
-
-  public static org.tensorflow.proto.util.SaverDef getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<SaverDef>
-      PARSER = new com.google.protobuf.AbstractParser<SaverDef>() {
-    @java.lang.Override
-    public SaverDef parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new SaverDef(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<SaverDef> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<SaverDef> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.util.SaverDef getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SaverDefOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SaverDefOrBuilder.java
deleted file mode 100644
index 6b2d7fa9e48..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SaverDefOrBuilder.java
+++ /dev/null
@@ -1,104 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/saver.proto
-
-package org.tensorflow.proto.util;
-
-public interface SaverDefOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.SaverDef)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <pre>
-   * The name of the tensor in which to specify the filename when saving or
-   * restoring a model checkpoint.
-   * </pre>
-   *
-   * <code>string filename_tensor_name = 1;</code>
-   */
-  java.lang.String getFilenameTensorName();
-  /**
-   * <pre>
-   * The name of the tensor in which to specify the filename when saving or
-   * restoring a model checkpoint.
-   * </pre>
-   *
-   * <code>string filename_tensor_name = 1;</code>
-   */
-  com.google.protobuf.ByteString
-      getFilenameTensorNameBytes();
-
-  /**
-   * <pre>
-   * The operation to run when saving a model checkpoint.
-   * </pre>
-   *
-   * <code>string save_tensor_name = 2;</code>
-   */
-  java.lang.String getSaveTensorName();
-  /**
-   * <pre>
-   * The operation to run when saving a model checkpoint.
-   * </pre>
-   *
-   * <code>string save_tensor_name = 2;</code>
-   */
-  com.google.protobuf.ByteString
-      getSaveTensorNameBytes();
-
-  /**
-   * <pre>
-   * The operation to run when restoring a model checkpoint.
-   * </pre>
-   *
-   * <code>string restore_op_name = 3;</code>
-   */
-  java.lang.String getRestoreOpName();
-  /**
-   * <pre>
-   * The operation to run when restoring a model checkpoint.
-   * </pre>
-   *
-   * <code>string restore_op_name = 3;</code>
-   */
-  com.google.protobuf.ByteString
-      getRestoreOpNameBytes();
-
-  /**
-   * <pre>
-   * Maximum number of checkpoints to keep.  If 0, no checkpoints are deleted.
-   * </pre>
-   *
-   * <code>int32 max_to_keep = 4;</code>
-   */
-  int getMaxToKeep();
-
-  /**
-   * <pre>
-   * Shard the save files, one per device that has Variable nodes.
-   * </pre>
-   *
-   * <code>bool sharded = 5;</code>
-   */
-  boolean getSharded();
-
-  /**
-   * <pre>
-   * How often to keep an additional checkpoint. If not specified, only the last
-   * "max_to_keep" checkpoints are kept; if specified, in addition to keeping
-   * the last "max_to_keep" checkpoints, an additional checkpoint will be kept
-   * for every n hours of training.
-   * </pre>
-   *
-   * <code>float keep_checkpoint_every_n_hours = 6;</code>
-   */
-  float getKeepCheckpointEveryNHours();
-
-  /**
-   * <code>.tensorflow.SaverDef.CheckpointFormatVersion version = 7;</code>
-   */
-  int getVersionValue();
-  /**
-   * <code>.tensorflow.SaverDef.CheckpointFormatVersion version = 7;</code>
-   */
-  org.tensorflow.proto.util.SaverDef.CheckpointFormatVersion getVersion();
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SaverProtos.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SaverProtos.java
deleted file mode 100644
index 8739b58bf4f..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SaverProtos.java
+++ /dev/null
@@ -1,57 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/saver.proto
-
-package org.tensorflow.proto.util;
-
-public final class SaverProtos {
-  private SaverProtos() {}
-  public static void registerAllExtensions(
-      com.google.protobuf.ExtensionRegistryLite registry) {
-  }
-
-  public static void registerAllExtensions(
-      com.google.protobuf.ExtensionRegistry registry) {
-    registerAllExtensions(
-        (com.google.protobuf.ExtensionRegistryLite) registry);
-  }
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_SaverDef_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_SaverDef_fieldAccessorTable;
-
-  public static com.google.protobuf.Descriptors.FileDescriptor
-      getDescriptor() {
-    return descriptor;
-  }
-  private static  com.google.protobuf.Descriptors.FileDescriptor
-      descriptor;
-  static {
-    java.lang.String[] descriptorData = {
-      "\n$tensorflow/core/protobuf/saver.proto\022\n" +
-      "tensorflow\"\236\002\n\010SaverDef\022\034\n\024filename_tens" +
-      "or_name\030\001 \001(\t\022\030\n\020save_tensor_name\030\002 \001(\t\022" +
-      "\027\n\017restore_op_name\030\003 \001(\t\022\023\n\013max_to_keep\030" +
-      "\004 \001(\005\022\017\n\007sharded\030\005 \001(\010\022%\n\035keep_checkpoin" +
-      "t_every_n_hours\030\006 \001(\002\022=\n\007version\030\007 \001(\0162," +
-      ".tensorflow.SaverDef.CheckpointFormatVer" +
-      "sion\"5\n\027CheckpointFormatVersion\022\n\n\006LEGAC" +
-      "Y\020\000\022\006\n\002V1\020\001\022\006\n\002V2\020\002Bw\n\031org.tensorflow.pr" +
-      "oto.utilB\013SaverProtosP\001ZHgithub.com/tens" +
-      "orflow/tensorflow/tensorflow/go/core/cor" +
-      "e_protos_go_proto\370\001\001b\006proto3"
-    };
-    descriptor = com.google.protobuf.Descriptors.FileDescriptor
-      .internalBuildGeneratedFileFrom(descriptorData,
-        new com.google.protobuf.Descriptors.FileDescriptor[] {
-        });
-    internal_static_tensorflow_SaverDef_descriptor =
-      getDescriptor().getMessageTypes().get(0);
-    internal_static_tensorflow_SaverDef_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_SaverDef_descriptor,
-        new java.lang.String[] { "FilenameTensorName", "SaveTensorName", "RestoreOpName", "MaxToKeep", "Sharded", "KeepCheckpointEveryNHours", "Version", });
-  }
-
-  // @@protoc_insertion_point(outer_class_scope)
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SessionLog.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SessionLog.java
deleted file mode 100644
index 824dba910f9..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SessionLog.java
+++ /dev/null
@@ -1,910 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/util/event.proto
-
-package org.tensorflow.proto.util;
-
-/**
- * <pre>
- * Protocol buffer used for logging session state.
- * </pre>
- *
- * Protobuf type {@code tensorflow.SessionLog}
- */
-public  final class SessionLog extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.SessionLog)
-    SessionLogOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use SessionLog.newBuilder() to construct.
-  private SessionLog(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private SessionLog() {
-    status_ = 0;
-    checkpointPath_ = "";
-    msg_ = "";
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new SessionLog();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private SessionLog(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 8: {
-            int rawValue = input.readEnum();
-
-            status_ = rawValue;
-            break;
-          }
-          case 18: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            checkpointPath_ = s;
-            break;
-          }
-          case 26: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            msg_ = s;
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.util.EventProtos.internal_static_tensorflow_SessionLog_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.util.EventProtos.internal_static_tensorflow_SessionLog_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.util.SessionLog.class, org.tensorflow.proto.util.SessionLog.Builder.class);
-  }
-
-  /**
-   * Protobuf enum {@code tensorflow.SessionLog.SessionStatus}
-   */
-  public enum SessionStatus
-      implements com.google.protobuf.ProtocolMessageEnum {
-    /**
-     * <code>STATUS_UNSPECIFIED = 0;</code>
-     */
-    STATUS_UNSPECIFIED(0),
-    /**
-     * <code>START = 1;</code>
-     */
-    START(1),
-    /**
-     * <code>STOP = 2;</code>
-     */
-    STOP(2),
-    /**
-     * <code>CHECKPOINT = 3;</code>
-     */
-    CHECKPOINT(3),
-    UNRECOGNIZED(-1),
-    ;
-
-    /**
-     * <code>STATUS_UNSPECIFIED = 0;</code>
-     */
-    public static final int STATUS_UNSPECIFIED_VALUE = 0;
-    /**
-     * <code>START = 1;</code>
-     */
-    public static final int START_VALUE = 1;
-    /**
-     * <code>STOP = 2;</code>
-     */
-    public static final int STOP_VALUE = 2;
-    /**
-     * <code>CHECKPOINT = 3;</code>
-     */
-    public static final int CHECKPOINT_VALUE = 3;
-
-
-    public final int getNumber() {
-      if (this == UNRECOGNIZED) {
-        throw new java.lang.IllegalArgumentException(
-            "Can't get the number of an unknown enum value.");
-      }
-      return value;
-    }
-
-    /**
-     * @deprecated Use {@link #forNumber(int)} instead.
-     */
-    @java.lang.Deprecated
-    public static SessionStatus valueOf(int value) {
-      return forNumber(value);
-    }
-
-    public static SessionStatus forNumber(int value) {
-      switch (value) {
-        case 0: return STATUS_UNSPECIFIED;
-        case 1: return START;
-        case 2: return STOP;
-        case 3: return CHECKPOINT;
-        default: return null;
-      }
-    }
-
-    public static com.google.protobuf.Internal.EnumLiteMap<SessionStatus>
-        internalGetValueMap() {
-      return internalValueMap;
-    }
-    private static final com.google.protobuf.Internal.EnumLiteMap<
-        SessionStatus> internalValueMap =
-          new com.google.protobuf.Internal.EnumLiteMap<SessionStatus>() {
-            public SessionStatus findValueByNumber(int number) {
-              return SessionStatus.forNumber(number);
-            }
-          };
-
-    public final com.google.protobuf.Descriptors.EnumValueDescriptor
-        getValueDescriptor() {
-      return getDescriptor().getValues().get(ordinal());
-    }
-    public final com.google.protobuf.Descriptors.EnumDescriptor
-        getDescriptorForType() {
-      return getDescriptor();
-    }
-    public static final com.google.protobuf.Descriptors.EnumDescriptor
-        getDescriptor() {
-      return org.tensorflow.proto.util.SessionLog.getDescriptor().getEnumTypes().get(0);
-    }
-
-    private static final SessionStatus[] VALUES = values();
-
-    public static SessionStatus valueOf(
-        com.google.protobuf.Descriptors.EnumValueDescriptor desc) {
-      if (desc.getType() != getDescriptor()) {
-        throw new java.lang.IllegalArgumentException(
-          "EnumValueDescriptor is not for this type.");
-      }
-      if (desc.getIndex() == -1) {
-        return UNRECOGNIZED;
-      }
-      return VALUES[desc.getIndex()];
-    }
-
-    private final int value;
-
-    private SessionStatus(int value) {
-      this.value = value;
-    }
-
-    // @@protoc_insertion_point(enum_scope:tensorflow.SessionLog.SessionStatus)
-  }
-
-  public static final int STATUS_FIELD_NUMBER = 1;
-  private int status_;
-  /**
-   * <code>.tensorflow.SessionLog.SessionStatus status = 1;</code>
-   */
-  public int getStatusValue() {
-    return status_;
-  }
-  /**
-   * <code>.tensorflow.SessionLog.SessionStatus status = 1;</code>
-   */
-  public org.tensorflow.proto.util.SessionLog.SessionStatus getStatus() {
-    @SuppressWarnings("deprecation")
-    org.tensorflow.proto.util.SessionLog.SessionStatus result = org.tensorflow.proto.util.SessionLog.SessionStatus.valueOf(status_);
-    return result == null ? org.tensorflow.proto.util.SessionLog.SessionStatus.UNRECOGNIZED : result;
-  }
-
-  public static final int CHECKPOINT_PATH_FIELD_NUMBER = 2;
-  private volatile java.lang.Object checkpointPath_;
-  /**
-   * <pre>
-   * This checkpoint_path contains both the path and filename.
-   * </pre>
-   *
-   * <code>string checkpoint_path = 2;</code>
-   */
-  public java.lang.String getCheckpointPath() {
-    java.lang.Object ref = checkpointPath_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      checkpointPath_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * This checkpoint_path contains both the path and filename.
-   * </pre>
-   *
-   * <code>string checkpoint_path = 2;</code>
-   */
-  public com.google.protobuf.ByteString
-      getCheckpointPathBytes() {
-    java.lang.Object ref = checkpointPath_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      checkpointPath_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int MSG_FIELD_NUMBER = 3;
-  private volatile java.lang.Object msg_;
-  /**
-   * <code>string msg = 3;</code>
-   */
-  public java.lang.String getMsg() {
-    java.lang.Object ref = msg_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      msg_ = s;
-      return s;
-    }
-  }
-  /**
-   * <code>string msg = 3;</code>
-   */
-  public com.google.protobuf.ByteString
-      getMsgBytes() {
-    java.lang.Object ref = msg_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      msg_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    if (status_ != org.tensorflow.proto.util.SessionLog.SessionStatus.STATUS_UNSPECIFIED.getNumber()) {
-      output.writeEnum(1, status_);
-    }
-    if (!getCheckpointPathBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 2, checkpointPath_);
-    }
-    if (!getMsgBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 3, msg_);
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (status_ != org.tensorflow.proto.util.SessionLog.SessionStatus.STATUS_UNSPECIFIED.getNumber()) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeEnumSize(1, status_);
-    }
-    if (!getCheckpointPathBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(2, checkpointPath_);
-    }
-    if (!getMsgBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(3, msg_);
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.util.SessionLog)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.util.SessionLog other = (org.tensorflow.proto.util.SessionLog) obj;
-
-    if (status_ != other.status_) return false;
-    if (!getCheckpointPath()
-        .equals(other.getCheckpointPath())) return false;
-    if (!getMsg()
-        .equals(other.getMsg())) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    hash = (37 * hash) + STATUS_FIELD_NUMBER;
-    hash = (53 * hash) + status_;
-    hash = (37 * hash) + CHECKPOINT_PATH_FIELD_NUMBER;
-    hash = (53 * hash) + getCheckpointPath().hashCode();
-    hash = (37 * hash) + MSG_FIELD_NUMBER;
-    hash = (53 * hash) + getMsg().hashCode();
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.util.SessionLog parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.SessionLog parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.SessionLog parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.SessionLog parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.SessionLog parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.SessionLog parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.SessionLog parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.SessionLog parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.SessionLog parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.SessionLog parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.SessionLog parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.SessionLog parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.util.SessionLog prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * <pre>
-   * Protocol buffer used for logging session state.
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.SessionLog}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.SessionLog)
-      org.tensorflow.proto.util.SessionLogOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.util.EventProtos.internal_static_tensorflow_SessionLog_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.util.EventProtos.internal_static_tensorflow_SessionLog_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.util.SessionLog.class, org.tensorflow.proto.util.SessionLog.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.util.SessionLog.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      status_ = 0;
-
-      checkpointPath_ = "";
-
-      msg_ = "";
-
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.util.EventProtos.internal_static_tensorflow_SessionLog_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.SessionLog getDefaultInstanceForType() {
-      return org.tensorflow.proto.util.SessionLog.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.SessionLog build() {
-      org.tensorflow.proto.util.SessionLog result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.SessionLog buildPartial() {
-      org.tensorflow.proto.util.SessionLog result = new org.tensorflow.proto.util.SessionLog(this);
-      result.status_ = status_;
-      result.checkpointPath_ = checkpointPath_;
-      result.msg_ = msg_;
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.util.SessionLog) {
-        return mergeFrom((org.tensorflow.proto.util.SessionLog)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.util.SessionLog other) {
-      if (other == org.tensorflow.proto.util.SessionLog.getDefaultInstance()) return this;
-      if (other.status_ != 0) {
-        setStatusValue(other.getStatusValue());
-      }
-      if (!other.getCheckpointPath().isEmpty()) {
-        checkpointPath_ = other.checkpointPath_;
-        onChanged();
-      }
-      if (!other.getMsg().isEmpty()) {
-        msg_ = other.msg_;
-        onChanged();
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.util.SessionLog parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.util.SessionLog) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-
-    private int status_ = 0;
-    /**
-     * <code>.tensorflow.SessionLog.SessionStatus status = 1;</code>
-     */
-    public int getStatusValue() {
-      return status_;
-    }
-    /**
-     * <code>.tensorflow.SessionLog.SessionStatus status = 1;</code>
-     */
-    public Builder setStatusValue(int value) {
-      status_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>.tensorflow.SessionLog.SessionStatus status = 1;</code>
-     */
-    public org.tensorflow.proto.util.SessionLog.SessionStatus getStatus() {
-      @SuppressWarnings("deprecation")
-      org.tensorflow.proto.util.SessionLog.SessionStatus result = org.tensorflow.proto.util.SessionLog.SessionStatus.valueOf(status_);
-      return result == null ? org.tensorflow.proto.util.SessionLog.SessionStatus.UNRECOGNIZED : result;
-    }
-    /**
-     * <code>.tensorflow.SessionLog.SessionStatus status = 1;</code>
-     */
-    public Builder setStatus(org.tensorflow.proto.util.SessionLog.SessionStatus value) {
-      if (value == null) {
-        throw new NullPointerException();
-      }
-      
-      status_ = value.getNumber();
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>.tensorflow.SessionLog.SessionStatus status = 1;</code>
-     */
-    public Builder clearStatus() {
-      
-      status_ = 0;
-      onChanged();
-      return this;
-    }
-
-    private java.lang.Object checkpointPath_ = "";
-    /**
-     * <pre>
-     * This checkpoint_path contains both the path and filename.
-     * </pre>
-     *
-     * <code>string checkpoint_path = 2;</code>
-     */
-    public java.lang.String getCheckpointPath() {
-      java.lang.Object ref = checkpointPath_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        checkpointPath_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * This checkpoint_path contains both the path and filename.
-     * </pre>
-     *
-     * <code>string checkpoint_path = 2;</code>
-     */
-    public com.google.protobuf.ByteString
-        getCheckpointPathBytes() {
-      java.lang.Object ref = checkpointPath_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        checkpointPath_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * This checkpoint_path contains both the path and filename.
-     * </pre>
-     *
-     * <code>string checkpoint_path = 2;</code>
-     */
-    public Builder setCheckpointPath(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      checkpointPath_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * This checkpoint_path contains both the path and filename.
-     * </pre>
-     *
-     * <code>string checkpoint_path = 2;</code>
-     */
-    public Builder clearCheckpointPath() {
-      
-      checkpointPath_ = getDefaultInstance().getCheckpointPath();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * This checkpoint_path contains both the path and filename.
-     * </pre>
-     *
-     * <code>string checkpoint_path = 2;</code>
-     */
-    public Builder setCheckpointPathBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      checkpointPath_ = value;
-      onChanged();
-      return this;
-    }
-
-    private java.lang.Object msg_ = "";
-    /**
-     * <code>string msg = 3;</code>
-     */
-    public java.lang.String getMsg() {
-      java.lang.Object ref = msg_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        msg_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <code>string msg = 3;</code>
-     */
-    public com.google.protobuf.ByteString
-        getMsgBytes() {
-      java.lang.Object ref = msg_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        msg_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <code>string msg = 3;</code>
-     */
-    public Builder setMsg(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      msg_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>string msg = 3;</code>
-     */
-    public Builder clearMsg() {
-      
-      msg_ = getDefaultInstance().getMsg();
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>string msg = 3;</code>
-     */
-    public Builder setMsgBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      msg_ = value;
-      onChanged();
-      return this;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.SessionLog)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.SessionLog)
-  private static final org.tensorflow.proto.util.SessionLog DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.util.SessionLog();
-  }
-
-  public static org.tensorflow.proto.util.SessionLog getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<SessionLog>
-      PARSER = new com.google.protobuf.AbstractParser<SessionLog>() {
-    @java.lang.Override
-    public SessionLog parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new SessionLog(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<SessionLog> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<SessionLog> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.util.SessionLog getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SessionLogOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SessionLogOrBuilder.java
deleted file mode 100644
index 84ba691a95d..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SessionLogOrBuilder.java
+++ /dev/null
@@ -1,46 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/util/event.proto
-
-package org.tensorflow.proto.util;
-
-public interface SessionLogOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.SessionLog)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <code>.tensorflow.SessionLog.SessionStatus status = 1;</code>
-   */
-  int getStatusValue();
-  /**
-   * <code>.tensorflow.SessionLog.SessionStatus status = 1;</code>
-   */
-  org.tensorflow.proto.util.SessionLog.SessionStatus getStatus();
-
-  /**
-   * <pre>
-   * This checkpoint_path contains both the path and filename.
-   * </pre>
-   *
-   * <code>string checkpoint_path = 2;</code>
-   */
-  java.lang.String getCheckpointPath();
-  /**
-   * <pre>
-   * This checkpoint_path contains both the path and filename.
-   * </pre>
-   *
-   * <code>string checkpoint_path = 2;</code>
-   */
-  com.google.protobuf.ByteString
-      getCheckpointPathBytes();
-
-  /**
-   * <code>string msg = 3;</code>
-   */
-  java.lang.String getMsg();
-  /**
-   * <code>string msg = 3;</code>
-   */
-  com.google.protobuf.ByteString
-      getMsgBytes();
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SnapShot.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SnapShot.java
deleted file mode 100644
index bde929e4c6d..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SnapShot.java
+++ /dev/null
@@ -1,535 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/bfc_memory_map.proto
-
-package org.tensorflow.proto.util;
-
-/**
- * Protobuf type {@code tensorflow.SnapShot}
- */
-public  final class SnapShot extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.SnapShot)
-    SnapShotOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use SnapShot.newBuilder() to construct.
-  private SnapShot(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private SnapShot() {
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new SnapShot();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private SnapShot(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 8: {
-
-            actionCount_ = input.readUInt64();
-            break;
-          }
-          case 16: {
-
-            size_ = input.readInt64();
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.util.BfcMemoryMapProtos.internal_static_tensorflow_SnapShot_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.util.BfcMemoryMapProtos.internal_static_tensorflow_SnapShot_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.util.SnapShot.class, org.tensorflow.proto.util.SnapShot.Builder.class);
-  }
-
-  public static final int ACTION_COUNT_FIELD_NUMBER = 1;
-  private long actionCount_;
-  /**
-   * <code>uint64 action_count = 1;</code>
-   */
-  public long getActionCount() {
-    return actionCount_;
-  }
-
-  public static final int SIZE_FIELD_NUMBER = 2;
-  private long size_;
-  /**
-   * <code>int64 size = 2;</code>
-   */
-  public long getSize() {
-    return size_;
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    if (actionCount_ != 0L) {
-      output.writeUInt64(1, actionCount_);
-    }
-    if (size_ != 0L) {
-      output.writeInt64(2, size_);
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (actionCount_ != 0L) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeUInt64Size(1, actionCount_);
-    }
-    if (size_ != 0L) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt64Size(2, size_);
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.util.SnapShot)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.util.SnapShot other = (org.tensorflow.proto.util.SnapShot) obj;
-
-    if (getActionCount()
-        != other.getActionCount()) return false;
-    if (getSize()
-        != other.getSize()) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    hash = (37 * hash) + ACTION_COUNT_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-        getActionCount());
-    hash = (37 * hash) + SIZE_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-        getSize());
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.util.SnapShot parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.SnapShot parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.SnapShot parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.SnapShot parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.SnapShot parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.SnapShot parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.SnapShot parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.SnapShot parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.SnapShot parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.SnapShot parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.SnapShot parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.SnapShot parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.util.SnapShot prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * Protobuf type {@code tensorflow.SnapShot}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.SnapShot)
-      org.tensorflow.proto.util.SnapShotOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.util.BfcMemoryMapProtos.internal_static_tensorflow_SnapShot_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.util.BfcMemoryMapProtos.internal_static_tensorflow_SnapShot_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.util.SnapShot.class, org.tensorflow.proto.util.SnapShot.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.util.SnapShot.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      actionCount_ = 0L;
-
-      size_ = 0L;
-
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.util.BfcMemoryMapProtos.internal_static_tensorflow_SnapShot_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.SnapShot getDefaultInstanceForType() {
-      return org.tensorflow.proto.util.SnapShot.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.SnapShot build() {
-      org.tensorflow.proto.util.SnapShot result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.SnapShot buildPartial() {
-      org.tensorflow.proto.util.SnapShot result = new org.tensorflow.proto.util.SnapShot(this);
-      result.actionCount_ = actionCount_;
-      result.size_ = size_;
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.util.SnapShot) {
-        return mergeFrom((org.tensorflow.proto.util.SnapShot)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.util.SnapShot other) {
-      if (other == org.tensorflow.proto.util.SnapShot.getDefaultInstance()) return this;
-      if (other.getActionCount() != 0L) {
-        setActionCount(other.getActionCount());
-      }
-      if (other.getSize() != 0L) {
-        setSize(other.getSize());
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.util.SnapShot parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.util.SnapShot) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-
-    private long actionCount_ ;
-    /**
-     * <code>uint64 action_count = 1;</code>
-     */
-    public long getActionCount() {
-      return actionCount_;
-    }
-    /**
-     * <code>uint64 action_count = 1;</code>
-     */
-    public Builder setActionCount(long value) {
-      
-      actionCount_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>uint64 action_count = 1;</code>
-     */
-    public Builder clearActionCount() {
-      
-      actionCount_ = 0L;
-      onChanged();
-      return this;
-    }
-
-    private long size_ ;
-    /**
-     * <code>int64 size = 2;</code>
-     */
-    public long getSize() {
-      return size_;
-    }
-    /**
-     * <code>int64 size = 2;</code>
-     */
-    public Builder setSize(long value) {
-      
-      size_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>int64 size = 2;</code>
-     */
-    public Builder clearSize() {
-      
-      size_ = 0L;
-      onChanged();
-      return this;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.SnapShot)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.SnapShot)
-  private static final org.tensorflow.proto.util.SnapShot DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.util.SnapShot();
-  }
-
-  public static org.tensorflow.proto.util.SnapShot getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<SnapShot>
-      PARSER = new com.google.protobuf.AbstractParser<SnapShot>() {
-    @java.lang.Override
-    public SnapShot parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new SnapShot(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<SnapShot> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<SnapShot> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.util.SnapShot getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SnapShotOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SnapShotOrBuilder.java
deleted file mode 100644
index 787a23ded48..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SnapShotOrBuilder.java
+++ /dev/null
@@ -1,19 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/bfc_memory_map.proto
-
-package org.tensorflow.proto.util;
-
-public interface SnapShotOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.SnapShot)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <code>uint64 action_count = 1;</code>
-   */
-  long getActionCount();
-
-  /**
-   * <code>int64 size = 2;</code>
-   */
-  long getSize();
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SnapshotMetadataRecord.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SnapshotMetadataRecord.java
deleted file mode 100644
index 36e10e21bb1..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SnapshotMetadataRecord.java
+++ /dev/null
@@ -1,1328 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/data/experimental/snapshot.proto
-
-package org.tensorflow.proto.util;
-
-/**
- * <pre>
- * This stores the metadata information present in each snapshot record.
- * </pre>
- *
- * Protobuf type {@code tensorflow.data.experimental.SnapshotMetadataRecord}
- */
-public  final class SnapshotMetadataRecord extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.data.experimental.SnapshotMetadataRecord)
-    SnapshotMetadataRecordOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use SnapshotMetadataRecord.newBuilder() to construct.
-  private SnapshotMetadataRecord(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private SnapshotMetadataRecord() {
-    graphHash_ = "";
-    runId_ = "";
-    dtype_ = java.util.Collections.emptyList();
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new SnapshotMetadataRecord();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private SnapshotMetadataRecord(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    int mutable_bitField0_ = 0;
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 10: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            graphHash_ = s;
-            break;
-          }
-          case 18: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            runId_ = s;
-            break;
-          }
-          case 24: {
-
-            creationTimestamp_ = input.readInt64();
-            break;
-          }
-          case 32: {
-
-            version_ = input.readInt64();
-            break;
-          }
-          case 40: {
-            int rawValue = input.readEnum();
-            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
-              dtype_ = new java.util.ArrayList<java.lang.Integer>();
-              mutable_bitField0_ |= 0x00000001;
-            }
-            dtype_.add(rawValue);
-            break;
-          }
-          case 42: {
-            int length = input.readRawVarint32();
-            int oldLimit = input.pushLimit(length);
-            while(input.getBytesUntilLimit() > 0) {
-              int rawValue = input.readEnum();
-              if (!((mutable_bitField0_ & 0x00000001) != 0)) {
-                dtype_ = new java.util.ArrayList<java.lang.Integer>();
-                mutable_bitField0_ |= 0x00000001;
-              }
-              dtype_.add(rawValue);
-            }
-            input.popLimit(oldLimit);
-            break;
-          }
-          case 48: {
-
-            numElements_ = input.readInt64();
-            break;
-          }
-          case 8000: {
-
-            finalized_ = input.readBool();
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      if (((mutable_bitField0_ & 0x00000001) != 0)) {
-        dtype_ = java.util.Collections.unmodifiableList(dtype_);
-      }
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.util.SnapshotProtos.internal_static_tensorflow_data_experimental_SnapshotMetadataRecord_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.util.SnapshotProtos.internal_static_tensorflow_data_experimental_SnapshotMetadataRecord_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.util.SnapshotMetadataRecord.class, org.tensorflow.proto.util.SnapshotMetadataRecord.Builder.class);
-  }
-
-  public static final int GRAPH_HASH_FIELD_NUMBER = 1;
-  private volatile java.lang.Object graphHash_;
-  /**
-   * <pre>
-   * Stores the fingerprint of the graph that describes the dataset that is
-   * snapshotted.
-   * </pre>
-   *
-   * <code>string graph_hash = 1;</code>
-   */
-  public java.lang.String getGraphHash() {
-    java.lang.Object ref = graphHash_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      graphHash_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * Stores the fingerprint of the graph that describes the dataset that is
-   * snapshotted.
-   * </pre>
-   *
-   * <code>string graph_hash = 1;</code>
-   */
-  public com.google.protobuf.ByteString
-      getGraphHashBytes() {
-    java.lang.Object ref = graphHash_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      graphHash_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int RUN_ID_FIELD_NUMBER = 2;
-  private volatile java.lang.Object runId_;
-  /**
-   * <pre>
-   * Run ID that this snapshot corresponds to.
-   * </pre>
-   *
-   * <code>string run_id = 2;</code>
-   */
-  public java.lang.String getRunId() {
-    java.lang.Object ref = runId_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      runId_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * Run ID that this snapshot corresponds to.
-   * </pre>
-   *
-   * <code>string run_id = 2;</code>
-   */
-  public com.google.protobuf.ByteString
-      getRunIdBytes() {
-    java.lang.Object ref = runId_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      runId_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int CREATION_TIMESTAMP_FIELD_NUMBER = 3;
-  private long creationTimestamp_;
-  /**
-   * <pre>
-   * Time when we started creating this snapshot.
-   * </pre>
-   *
-   * <code>int64 creation_timestamp = 3;</code>
-   */
-  public long getCreationTimestamp() {
-    return creationTimestamp_;
-  }
-
-  public static final int VERSION_FIELD_NUMBER = 4;
-  private long version_;
-  /**
-   * <pre>
-   * Version of the snapshot data file format.
-   * </pre>
-   *
-   * <code>int64 version = 4;</code>
-   */
-  public long getVersion() {
-    return version_;
-  }
-
-  public static final int DTYPE_FIELD_NUMBER = 5;
-  private java.util.List<java.lang.Integer> dtype_;
-  private static final com.google.protobuf.Internal.ListAdapter.Converter<
-      java.lang.Integer, org.tensorflow.proto.framework.DataType> dtype_converter_ =
-          new com.google.protobuf.Internal.ListAdapter.Converter<
-              java.lang.Integer, org.tensorflow.proto.framework.DataType>() {
-            public org.tensorflow.proto.framework.DataType convert(java.lang.Integer from) {
-              @SuppressWarnings("deprecation")
-              org.tensorflow.proto.framework.DataType result = org.tensorflow.proto.framework.DataType.valueOf(from);
-              return result == null ? org.tensorflow.proto.framework.DataType.UNRECOGNIZED : result;
-            }
-          };
-  /**
-   * <pre>
-   * A list of tensor dtype corresponding to each element of the snapshot.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.DataType dtype = 5;</code>
-   */
-  public java.util.List<org.tensorflow.proto.framework.DataType> getDtypeList() {
-    return new com.google.protobuf.Internal.ListAdapter<
-        java.lang.Integer, org.tensorflow.proto.framework.DataType>(dtype_, dtype_converter_);
-  }
-  /**
-   * <pre>
-   * A list of tensor dtype corresponding to each element of the snapshot.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.DataType dtype = 5;</code>
-   */
-  public int getDtypeCount() {
-    return dtype_.size();
-  }
-  /**
-   * <pre>
-   * A list of tensor dtype corresponding to each element of the snapshot.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.DataType dtype = 5;</code>
-   */
-  public org.tensorflow.proto.framework.DataType getDtype(int index) {
-    return dtype_converter_.convert(dtype_.get(index));
-  }
-  /**
-   * <pre>
-   * A list of tensor dtype corresponding to each element of the snapshot.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.DataType dtype = 5;</code>
-   */
-  public java.util.List<java.lang.Integer>
-  getDtypeValueList() {
-    return dtype_;
-  }
-  /**
-   * <pre>
-   * A list of tensor dtype corresponding to each element of the snapshot.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.DataType dtype = 5;</code>
-   */
-  public int getDtypeValue(int index) {
-    return dtype_.get(index);
-  }
-  private int dtypeMemoizedSerializedSize;
-
-  public static final int NUM_ELEMENTS_FIELD_NUMBER = 6;
-  private long numElements_;
-  /**
-   * <pre>
-   * The number of elements in the snapshot.
-   * </pre>
-   *
-   * <code>int64 num_elements = 6;</code>
-   */
-  public long getNumElements() {
-    return numElements_;
-  }
-
-  public static final int FINALIZED_FIELD_NUMBER = 1000;
-  private boolean finalized_;
-  /**
-   * <code>bool finalized = 1000;</code>
-   */
-  public boolean getFinalized() {
-    return finalized_;
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    getSerializedSize();
-    if (!getGraphHashBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 1, graphHash_);
-    }
-    if (!getRunIdBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 2, runId_);
-    }
-    if (creationTimestamp_ != 0L) {
-      output.writeInt64(3, creationTimestamp_);
-    }
-    if (version_ != 0L) {
-      output.writeInt64(4, version_);
-    }
-    if (getDtypeList().size() > 0) {
-      output.writeUInt32NoTag(42);
-      output.writeUInt32NoTag(dtypeMemoizedSerializedSize);
-    }
-    for (int i = 0; i < dtype_.size(); i++) {
-      output.writeEnumNoTag(dtype_.get(i));
-    }
-    if (numElements_ != 0L) {
-      output.writeInt64(6, numElements_);
-    }
-    if (finalized_ != false) {
-      output.writeBool(1000, finalized_);
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (!getGraphHashBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, graphHash_);
-    }
-    if (!getRunIdBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(2, runId_);
-    }
-    if (creationTimestamp_ != 0L) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt64Size(3, creationTimestamp_);
-    }
-    if (version_ != 0L) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt64Size(4, version_);
-    }
-    {
-      int dataSize = 0;
-      for (int i = 0; i < dtype_.size(); i++) {
-        dataSize += com.google.protobuf.CodedOutputStream
-          .computeEnumSizeNoTag(dtype_.get(i));
-      }
-      size += dataSize;
-      if (!getDtypeList().isEmpty()) {  size += 1;
-        size += com.google.protobuf.CodedOutputStream
-          .computeUInt32SizeNoTag(dataSize);
-      }dtypeMemoizedSerializedSize = dataSize;
-    }
-    if (numElements_ != 0L) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt64Size(6, numElements_);
-    }
-    if (finalized_ != false) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeBoolSize(1000, finalized_);
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.util.SnapshotMetadataRecord)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.util.SnapshotMetadataRecord other = (org.tensorflow.proto.util.SnapshotMetadataRecord) obj;
-
-    if (!getGraphHash()
-        .equals(other.getGraphHash())) return false;
-    if (!getRunId()
-        .equals(other.getRunId())) return false;
-    if (getCreationTimestamp()
-        != other.getCreationTimestamp()) return false;
-    if (getVersion()
-        != other.getVersion()) return false;
-    if (!dtype_.equals(other.dtype_)) return false;
-    if (getNumElements()
-        != other.getNumElements()) return false;
-    if (getFinalized()
-        != other.getFinalized()) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    hash = (37 * hash) + GRAPH_HASH_FIELD_NUMBER;
-    hash = (53 * hash) + getGraphHash().hashCode();
-    hash = (37 * hash) + RUN_ID_FIELD_NUMBER;
-    hash = (53 * hash) + getRunId().hashCode();
-    hash = (37 * hash) + CREATION_TIMESTAMP_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-        getCreationTimestamp());
-    hash = (37 * hash) + VERSION_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-        getVersion());
-    if (getDtypeCount() > 0) {
-      hash = (37 * hash) + DTYPE_FIELD_NUMBER;
-      hash = (53 * hash) + dtype_.hashCode();
-    }
-    hash = (37 * hash) + NUM_ELEMENTS_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-        getNumElements());
-    hash = (37 * hash) + FINALIZED_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashBoolean(
-        getFinalized());
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.util.SnapshotMetadataRecord parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.SnapshotMetadataRecord parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.SnapshotMetadataRecord parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.SnapshotMetadataRecord parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.SnapshotMetadataRecord parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.SnapshotMetadataRecord parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.SnapshotMetadataRecord parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.SnapshotMetadataRecord parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.SnapshotMetadataRecord parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.SnapshotMetadataRecord parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.SnapshotMetadataRecord parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.SnapshotMetadataRecord parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.util.SnapshotMetadataRecord prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * <pre>
-   * This stores the metadata information present in each snapshot record.
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.data.experimental.SnapshotMetadataRecord}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.data.experimental.SnapshotMetadataRecord)
-      org.tensorflow.proto.util.SnapshotMetadataRecordOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.util.SnapshotProtos.internal_static_tensorflow_data_experimental_SnapshotMetadataRecord_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.util.SnapshotProtos.internal_static_tensorflow_data_experimental_SnapshotMetadataRecord_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.util.SnapshotMetadataRecord.class, org.tensorflow.proto.util.SnapshotMetadataRecord.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.util.SnapshotMetadataRecord.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      graphHash_ = "";
-
-      runId_ = "";
-
-      creationTimestamp_ = 0L;
-
-      version_ = 0L;
-
-      dtype_ = java.util.Collections.emptyList();
-      bitField0_ = (bitField0_ & ~0x00000001);
-      numElements_ = 0L;
-
-      finalized_ = false;
-
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.util.SnapshotProtos.internal_static_tensorflow_data_experimental_SnapshotMetadataRecord_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.SnapshotMetadataRecord getDefaultInstanceForType() {
-      return org.tensorflow.proto.util.SnapshotMetadataRecord.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.SnapshotMetadataRecord build() {
-      org.tensorflow.proto.util.SnapshotMetadataRecord result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.SnapshotMetadataRecord buildPartial() {
-      org.tensorflow.proto.util.SnapshotMetadataRecord result = new org.tensorflow.proto.util.SnapshotMetadataRecord(this);
-      int from_bitField0_ = bitField0_;
-      result.graphHash_ = graphHash_;
-      result.runId_ = runId_;
-      result.creationTimestamp_ = creationTimestamp_;
-      result.version_ = version_;
-      if (((bitField0_ & 0x00000001) != 0)) {
-        dtype_ = java.util.Collections.unmodifiableList(dtype_);
-        bitField0_ = (bitField0_ & ~0x00000001);
-      }
-      result.dtype_ = dtype_;
-      result.numElements_ = numElements_;
-      result.finalized_ = finalized_;
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.util.SnapshotMetadataRecord) {
-        return mergeFrom((org.tensorflow.proto.util.SnapshotMetadataRecord)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.util.SnapshotMetadataRecord other) {
-      if (other == org.tensorflow.proto.util.SnapshotMetadataRecord.getDefaultInstance()) return this;
-      if (!other.getGraphHash().isEmpty()) {
-        graphHash_ = other.graphHash_;
-        onChanged();
-      }
-      if (!other.getRunId().isEmpty()) {
-        runId_ = other.runId_;
-        onChanged();
-      }
-      if (other.getCreationTimestamp() != 0L) {
-        setCreationTimestamp(other.getCreationTimestamp());
-      }
-      if (other.getVersion() != 0L) {
-        setVersion(other.getVersion());
-      }
-      if (!other.dtype_.isEmpty()) {
-        if (dtype_.isEmpty()) {
-          dtype_ = other.dtype_;
-          bitField0_ = (bitField0_ & ~0x00000001);
-        } else {
-          ensureDtypeIsMutable();
-          dtype_.addAll(other.dtype_);
-        }
-        onChanged();
-      }
-      if (other.getNumElements() != 0L) {
-        setNumElements(other.getNumElements());
-      }
-      if (other.getFinalized() != false) {
-        setFinalized(other.getFinalized());
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.util.SnapshotMetadataRecord parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.util.SnapshotMetadataRecord) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-    private int bitField0_;
-
-    private java.lang.Object graphHash_ = "";
-    /**
-     * <pre>
-     * Stores the fingerprint of the graph that describes the dataset that is
-     * snapshotted.
-     * </pre>
-     *
-     * <code>string graph_hash = 1;</code>
-     */
-    public java.lang.String getGraphHash() {
-      java.lang.Object ref = graphHash_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        graphHash_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Stores the fingerprint of the graph that describes the dataset that is
-     * snapshotted.
-     * </pre>
-     *
-     * <code>string graph_hash = 1;</code>
-     */
-    public com.google.protobuf.ByteString
-        getGraphHashBytes() {
-      java.lang.Object ref = graphHash_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        graphHash_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Stores the fingerprint of the graph that describes the dataset that is
-     * snapshotted.
-     * </pre>
-     *
-     * <code>string graph_hash = 1;</code>
-     */
-    public Builder setGraphHash(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      graphHash_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Stores the fingerprint of the graph that describes the dataset that is
-     * snapshotted.
-     * </pre>
-     *
-     * <code>string graph_hash = 1;</code>
-     */
-    public Builder clearGraphHash() {
-      
-      graphHash_ = getDefaultInstance().getGraphHash();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Stores the fingerprint of the graph that describes the dataset that is
-     * snapshotted.
-     * </pre>
-     *
-     * <code>string graph_hash = 1;</code>
-     */
-    public Builder setGraphHashBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      graphHash_ = value;
-      onChanged();
-      return this;
-    }
-
-    private java.lang.Object runId_ = "";
-    /**
-     * <pre>
-     * Run ID that this snapshot corresponds to.
-     * </pre>
-     *
-     * <code>string run_id = 2;</code>
-     */
-    public java.lang.String getRunId() {
-      java.lang.Object ref = runId_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        runId_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Run ID that this snapshot corresponds to.
-     * </pre>
-     *
-     * <code>string run_id = 2;</code>
-     */
-    public com.google.protobuf.ByteString
-        getRunIdBytes() {
-      java.lang.Object ref = runId_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        runId_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Run ID that this snapshot corresponds to.
-     * </pre>
-     *
-     * <code>string run_id = 2;</code>
-     */
-    public Builder setRunId(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      runId_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Run ID that this snapshot corresponds to.
-     * </pre>
-     *
-     * <code>string run_id = 2;</code>
-     */
-    public Builder clearRunId() {
-      
-      runId_ = getDefaultInstance().getRunId();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Run ID that this snapshot corresponds to.
-     * </pre>
-     *
-     * <code>string run_id = 2;</code>
-     */
-    public Builder setRunIdBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      runId_ = value;
-      onChanged();
-      return this;
-    }
-
-    private long creationTimestamp_ ;
-    /**
-     * <pre>
-     * Time when we started creating this snapshot.
-     * </pre>
-     *
-     * <code>int64 creation_timestamp = 3;</code>
-     */
-    public long getCreationTimestamp() {
-      return creationTimestamp_;
-    }
-    /**
-     * <pre>
-     * Time when we started creating this snapshot.
-     * </pre>
-     *
-     * <code>int64 creation_timestamp = 3;</code>
-     */
-    public Builder setCreationTimestamp(long value) {
-      
-      creationTimestamp_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Time when we started creating this snapshot.
-     * </pre>
-     *
-     * <code>int64 creation_timestamp = 3;</code>
-     */
-    public Builder clearCreationTimestamp() {
-      
-      creationTimestamp_ = 0L;
-      onChanged();
-      return this;
-    }
-
-    private long version_ ;
-    /**
-     * <pre>
-     * Version of the snapshot data file format.
-     * </pre>
-     *
-     * <code>int64 version = 4;</code>
-     */
-    public long getVersion() {
-      return version_;
-    }
-    /**
-     * <pre>
-     * Version of the snapshot data file format.
-     * </pre>
-     *
-     * <code>int64 version = 4;</code>
-     */
-    public Builder setVersion(long value) {
-      
-      version_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Version of the snapshot data file format.
-     * </pre>
-     *
-     * <code>int64 version = 4;</code>
-     */
-    public Builder clearVersion() {
-      
-      version_ = 0L;
-      onChanged();
-      return this;
-    }
-
-    private java.util.List<java.lang.Integer> dtype_ =
-      java.util.Collections.emptyList();
-    private void ensureDtypeIsMutable() {
-      if (!((bitField0_ & 0x00000001) != 0)) {
-        dtype_ = new java.util.ArrayList<java.lang.Integer>(dtype_);
-        bitField0_ |= 0x00000001;
-      }
-    }
-    /**
-     * <pre>
-     * A list of tensor dtype corresponding to each element of the snapshot.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.DataType dtype = 5;</code>
-     */
-    public java.util.List<org.tensorflow.proto.framework.DataType> getDtypeList() {
-      return new com.google.protobuf.Internal.ListAdapter<
-          java.lang.Integer, org.tensorflow.proto.framework.DataType>(dtype_, dtype_converter_);
-    }
-    /**
-     * <pre>
-     * A list of tensor dtype corresponding to each element of the snapshot.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.DataType dtype = 5;</code>
-     */
-    public int getDtypeCount() {
-      return dtype_.size();
-    }
-    /**
-     * <pre>
-     * A list of tensor dtype corresponding to each element of the snapshot.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.DataType dtype = 5;</code>
-     */
-    public org.tensorflow.proto.framework.DataType getDtype(int index) {
-      return dtype_converter_.convert(dtype_.get(index));
-    }
-    /**
-     * <pre>
-     * A list of tensor dtype corresponding to each element of the snapshot.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.DataType dtype = 5;</code>
-     */
-    public Builder setDtype(
-        int index, org.tensorflow.proto.framework.DataType value) {
-      if (value == null) {
-        throw new NullPointerException();
-      }
-      ensureDtypeIsMutable();
-      dtype_.set(index, value.getNumber());
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * A list of tensor dtype corresponding to each element of the snapshot.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.DataType dtype = 5;</code>
-     */
-    public Builder addDtype(org.tensorflow.proto.framework.DataType value) {
-      if (value == null) {
-        throw new NullPointerException();
-      }
-      ensureDtypeIsMutable();
-      dtype_.add(value.getNumber());
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * A list of tensor dtype corresponding to each element of the snapshot.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.DataType dtype = 5;</code>
-     */
-    public Builder addAllDtype(
-        java.lang.Iterable<? extends org.tensorflow.proto.framework.DataType> values) {
-      ensureDtypeIsMutable();
-      for (org.tensorflow.proto.framework.DataType value : values) {
-        dtype_.add(value.getNumber());
-      }
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * A list of tensor dtype corresponding to each element of the snapshot.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.DataType dtype = 5;</code>
-     */
-    public Builder clearDtype() {
-      dtype_ = java.util.Collections.emptyList();
-      bitField0_ = (bitField0_ & ~0x00000001);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * A list of tensor dtype corresponding to each element of the snapshot.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.DataType dtype = 5;</code>
-     */
-    public java.util.List<java.lang.Integer>
-    getDtypeValueList() {
-      return java.util.Collections.unmodifiableList(dtype_);
-    }
-    /**
-     * <pre>
-     * A list of tensor dtype corresponding to each element of the snapshot.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.DataType dtype = 5;</code>
-     */
-    public int getDtypeValue(int index) {
-      return dtype_.get(index);
-    }
-    /**
-     * <pre>
-     * A list of tensor dtype corresponding to each element of the snapshot.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.DataType dtype = 5;</code>
-     */
-    public Builder setDtypeValue(
-        int index, int value) {
-      ensureDtypeIsMutable();
-      dtype_.set(index, value);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * A list of tensor dtype corresponding to each element of the snapshot.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.DataType dtype = 5;</code>
-     */
-    public Builder addDtypeValue(int value) {
-      ensureDtypeIsMutable();
-      dtype_.add(value);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * A list of tensor dtype corresponding to each element of the snapshot.
-     * </pre>
-     *
-     * <code>repeated .tensorflow.DataType dtype = 5;</code>
-     */
-    public Builder addAllDtypeValue(
-        java.lang.Iterable<java.lang.Integer> values) {
-      ensureDtypeIsMutable();
-      for (int value : values) {
-        dtype_.add(value);
-      }
-      onChanged();
-      return this;
-    }
-
-    private long numElements_ ;
-    /**
-     * <pre>
-     * The number of elements in the snapshot.
-     * </pre>
-     *
-     * <code>int64 num_elements = 6;</code>
-     */
-    public long getNumElements() {
-      return numElements_;
-    }
-    /**
-     * <pre>
-     * The number of elements in the snapshot.
-     * </pre>
-     *
-     * <code>int64 num_elements = 6;</code>
-     */
-    public Builder setNumElements(long value) {
-      
-      numElements_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * The number of elements in the snapshot.
-     * </pre>
-     *
-     * <code>int64 num_elements = 6;</code>
-     */
-    public Builder clearNumElements() {
-      
-      numElements_ = 0L;
-      onChanged();
-      return this;
-    }
-
-    private boolean finalized_ ;
-    /**
-     * <code>bool finalized = 1000;</code>
-     */
-    public boolean getFinalized() {
-      return finalized_;
-    }
-    /**
-     * <code>bool finalized = 1000;</code>
-     */
-    public Builder setFinalized(boolean value) {
-      
-      finalized_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>bool finalized = 1000;</code>
-     */
-    public Builder clearFinalized() {
-      
-      finalized_ = false;
-      onChanged();
-      return this;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.data.experimental.SnapshotMetadataRecord)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.data.experimental.SnapshotMetadataRecord)
-  private static final org.tensorflow.proto.util.SnapshotMetadataRecord DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.util.SnapshotMetadataRecord();
-  }
-
-  public static org.tensorflow.proto.util.SnapshotMetadataRecord getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<SnapshotMetadataRecord>
-      PARSER = new com.google.protobuf.AbstractParser<SnapshotMetadataRecord>() {
-    @java.lang.Override
-    public SnapshotMetadataRecord parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new SnapshotMetadataRecord(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<SnapshotMetadataRecord> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<SnapshotMetadataRecord> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.util.SnapshotMetadataRecord getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SnapshotMetadataRecordOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SnapshotMetadataRecordOrBuilder.java
deleted file mode 100644
index 13686e12e57..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SnapshotMetadataRecordOrBuilder.java
+++ /dev/null
@@ -1,121 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/data/experimental/snapshot.proto
-
-package org.tensorflow.proto.util;
-
-public interface SnapshotMetadataRecordOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.data.experimental.SnapshotMetadataRecord)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <pre>
-   * Stores the fingerprint of the graph that describes the dataset that is
-   * snapshotted.
-   * </pre>
-   *
-   * <code>string graph_hash = 1;</code>
-   */
-  java.lang.String getGraphHash();
-  /**
-   * <pre>
-   * Stores the fingerprint of the graph that describes the dataset that is
-   * snapshotted.
-   * </pre>
-   *
-   * <code>string graph_hash = 1;</code>
-   */
-  com.google.protobuf.ByteString
-      getGraphHashBytes();
-
-  /**
-   * <pre>
-   * Run ID that this snapshot corresponds to.
-   * </pre>
-   *
-   * <code>string run_id = 2;</code>
-   */
-  java.lang.String getRunId();
-  /**
-   * <pre>
-   * Run ID that this snapshot corresponds to.
-   * </pre>
-   *
-   * <code>string run_id = 2;</code>
-   */
-  com.google.protobuf.ByteString
-      getRunIdBytes();
-
-  /**
-   * <pre>
-   * Time when we started creating this snapshot.
-   * </pre>
-   *
-   * <code>int64 creation_timestamp = 3;</code>
-   */
-  long getCreationTimestamp();
-
-  /**
-   * <pre>
-   * Version of the snapshot data file format.
-   * </pre>
-   *
-   * <code>int64 version = 4;</code>
-   */
-  long getVersion();
-
-  /**
-   * <pre>
-   * A list of tensor dtype corresponding to each element of the snapshot.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.DataType dtype = 5;</code>
-   */
-  java.util.List<org.tensorflow.proto.framework.DataType> getDtypeList();
-  /**
-   * <pre>
-   * A list of tensor dtype corresponding to each element of the snapshot.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.DataType dtype = 5;</code>
-   */
-  int getDtypeCount();
-  /**
-   * <pre>
-   * A list of tensor dtype corresponding to each element of the snapshot.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.DataType dtype = 5;</code>
-   */
-  org.tensorflow.proto.framework.DataType getDtype(int index);
-  /**
-   * <pre>
-   * A list of tensor dtype corresponding to each element of the snapshot.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.DataType dtype = 5;</code>
-   */
-  java.util.List<java.lang.Integer>
-  getDtypeValueList();
-  /**
-   * <pre>
-   * A list of tensor dtype corresponding to each element of the snapshot.
-   * </pre>
-   *
-   * <code>repeated .tensorflow.DataType dtype = 5;</code>
-   */
-  int getDtypeValue(int index);
-
-  /**
-   * <pre>
-   * The number of elements in the snapshot.
-   * </pre>
-   *
-   * <code>int64 num_elements = 6;</code>
-   */
-  long getNumElements();
-
-  /**
-   * <code>bool finalized = 1000;</code>
-   */
-  boolean getFinalized();
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SnapshotProtos.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SnapshotProtos.java
deleted file mode 100644
index 2d8f207c550..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SnapshotProtos.java
+++ /dev/null
@@ -1,102 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/data/experimental/snapshot.proto
-
-package org.tensorflow.proto.util;
-
-public final class SnapshotProtos {
-  private SnapshotProtos() {}
-  public static void registerAllExtensions(
-      com.google.protobuf.ExtensionRegistryLite registry) {
-  }
-
-  public static void registerAllExtensions(
-      com.google.protobuf.ExtensionRegistry registry) {
-    registerAllExtensions(
-        (com.google.protobuf.ExtensionRegistryLite) registry);
-  }
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_data_experimental_SnapshotRecord_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_data_experimental_SnapshotRecord_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_data_experimental_SnapshotMetadataRecord_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_data_experimental_SnapshotMetadataRecord_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_data_experimental_TensorMetadata_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_data_experimental_TensorMetadata_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_data_experimental_SnapshotTensorMetadata_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_data_experimental_SnapshotTensorMetadata_fieldAccessorTable;
-
-  public static com.google.protobuf.Descriptors.FileDescriptor
-      getDescriptor() {
-    return descriptor;
-  }
-  private static  com.google.protobuf.Descriptors.FileDescriptor
-      descriptor;
-  static {
-    java.lang.String[] descriptorData = {
-      "\n9tensorflow/core/protobuf/data/experime" +
-      "ntal/snapshot.proto\022\034tensorflow.data.exp" +
-      "erimental\032&tensorflow/core/framework/ten" +
-      "sor.proto\032,tensorflow/core/framework/ten" +
-      "sor_shape.proto\032%tensorflow/core/framewo" +
-      "rk/types.proto\"9\n\016SnapshotRecord\022\'\n\006tens" +
-      "or\030\001 \003(\0132\027.tensorflow.TensorProto\"\270\001\n\026Sn" +
-      "apshotMetadataRecord\022\022\n\ngraph_hash\030\001 \001(\t" +
-      "\022\016\n\006run_id\030\002 \001(\t\022\032\n\022creation_timestamp\030\003" +
-      " \001(\003\022\017\n\007version\030\004 \001(\003\022#\n\005dtype\030\005 \003(\0162\024.t" +
-      "ensorflow.DataType\022\024\n\014num_elements\030\006 \001(\003" +
-      "\022\022\n\tfinalized\030\350\007 \001(\010\"_\n\016TensorMetadata\0222" +
-      "\n\014tensor_shape\030\002 \001(\0132\034.tensorflow.Tensor" +
-      "ShapeProto\022\031\n\021tensor_size_bytes\030\003 \001(\003\"_\n" +
-      "\026SnapshotTensorMetadata\022E\n\017tensor_metada" +
-      "ta\030\001 \003(\0132,.tensorflow.data.experimental." +
-      "TensorMetadataB-\n\031org.tensorflow.proto.u" +
-      "tilB\016SnapshotProtosP\001b\006proto3"
-    };
-    descriptor = com.google.protobuf.Descriptors.FileDescriptor
-      .internalBuildGeneratedFileFrom(descriptorData,
-        new com.google.protobuf.Descriptors.FileDescriptor[] {
-          org.tensorflow.proto.framework.TensorProtos.getDescriptor(),
-          org.tensorflow.proto.framework.TensorShapeProtos.getDescriptor(),
-          org.tensorflow.proto.framework.TypesProtos.getDescriptor(),
-        });
-    internal_static_tensorflow_data_experimental_SnapshotRecord_descriptor =
-      getDescriptor().getMessageTypes().get(0);
-    internal_static_tensorflow_data_experimental_SnapshotRecord_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_data_experimental_SnapshotRecord_descriptor,
-        new java.lang.String[] { "Tensor", });
-    internal_static_tensorflow_data_experimental_SnapshotMetadataRecord_descriptor =
-      getDescriptor().getMessageTypes().get(1);
-    internal_static_tensorflow_data_experimental_SnapshotMetadataRecord_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_data_experimental_SnapshotMetadataRecord_descriptor,
-        new java.lang.String[] { "GraphHash", "RunId", "CreationTimestamp", "Version", "Dtype", "NumElements", "Finalized", });
-    internal_static_tensorflow_data_experimental_TensorMetadata_descriptor =
-      getDescriptor().getMessageTypes().get(2);
-    internal_static_tensorflow_data_experimental_TensorMetadata_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_data_experimental_TensorMetadata_descriptor,
-        new java.lang.String[] { "TensorShape", "TensorSizeBytes", });
-    internal_static_tensorflow_data_experimental_SnapshotTensorMetadata_descriptor =
-      getDescriptor().getMessageTypes().get(3);
-    internal_static_tensorflow_data_experimental_SnapshotTensorMetadata_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_data_experimental_SnapshotTensorMetadata_descriptor,
-        new java.lang.String[] { "TensorMetadata", });
-    org.tensorflow.proto.framework.TensorProtos.getDescriptor();
-    org.tensorflow.proto.framework.TensorShapeProtos.getDescriptor();
-    org.tensorflow.proto.framework.TypesProtos.getDescriptor();
-  }
-
-  // @@protoc_insertion_point(outer_class_scope)
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SnapshotRecord.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SnapshotRecord.java
deleted file mode 100644
index 223b22587d7..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SnapshotRecord.java
+++ /dev/null
@@ -1,777 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/data/experimental/snapshot.proto
-
-package org.tensorflow.proto.util;
-
-/**
- * <pre>
- * Each SnapshotRecord represents one batch of pre-processed input data. A batch
- * consists of a list of tensors that we encode as TensorProtos. This message
- * doesn't store the structure of the batch.
- * </pre>
- *
- * Protobuf type {@code tensorflow.data.experimental.SnapshotRecord}
- */
-public  final class SnapshotRecord extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.data.experimental.SnapshotRecord)
-    SnapshotRecordOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use SnapshotRecord.newBuilder() to construct.
-  private SnapshotRecord(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private SnapshotRecord() {
-    tensor_ = java.util.Collections.emptyList();
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new SnapshotRecord();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private SnapshotRecord(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    int mutable_bitField0_ = 0;
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 10: {
-            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
-              tensor_ = new java.util.ArrayList<org.tensorflow.proto.framework.TensorProto>();
-              mutable_bitField0_ |= 0x00000001;
-            }
-            tensor_.add(
-                input.readMessage(org.tensorflow.proto.framework.TensorProto.parser(), extensionRegistry));
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      if (((mutable_bitField0_ & 0x00000001) != 0)) {
-        tensor_ = java.util.Collections.unmodifiableList(tensor_);
-      }
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.util.SnapshotProtos.internal_static_tensorflow_data_experimental_SnapshotRecord_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.util.SnapshotProtos.internal_static_tensorflow_data_experimental_SnapshotRecord_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.util.SnapshotRecord.class, org.tensorflow.proto.util.SnapshotRecord.Builder.class);
-  }
-
-  public static final int TENSOR_FIELD_NUMBER = 1;
-  private java.util.List<org.tensorflow.proto.framework.TensorProto> tensor_;
-  /**
-   * <code>repeated .tensorflow.TensorProto tensor = 1;</code>
-   */
-  public java.util.List<org.tensorflow.proto.framework.TensorProto> getTensorList() {
-    return tensor_;
-  }
-  /**
-   * <code>repeated .tensorflow.TensorProto tensor = 1;</code>
-   */
-  public java.util.List<? extends org.tensorflow.proto.framework.TensorProtoOrBuilder> 
-      getTensorOrBuilderList() {
-    return tensor_;
-  }
-  /**
-   * <code>repeated .tensorflow.TensorProto tensor = 1;</code>
-   */
-  public int getTensorCount() {
-    return tensor_.size();
-  }
-  /**
-   * <code>repeated .tensorflow.TensorProto tensor = 1;</code>
-   */
-  public org.tensorflow.proto.framework.TensorProto getTensor(int index) {
-    return tensor_.get(index);
-  }
-  /**
-   * <code>repeated .tensorflow.TensorProto tensor = 1;</code>
-   */
-  public org.tensorflow.proto.framework.TensorProtoOrBuilder getTensorOrBuilder(
-      int index) {
-    return tensor_.get(index);
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    for (int i = 0; i < tensor_.size(); i++) {
-      output.writeMessage(1, tensor_.get(i));
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    for (int i = 0; i < tensor_.size(); i++) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(1, tensor_.get(i));
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.util.SnapshotRecord)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.util.SnapshotRecord other = (org.tensorflow.proto.util.SnapshotRecord) obj;
-
-    if (!getTensorList()
-        .equals(other.getTensorList())) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    if (getTensorCount() > 0) {
-      hash = (37 * hash) + TENSOR_FIELD_NUMBER;
-      hash = (53 * hash) + getTensorList().hashCode();
-    }
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.util.SnapshotRecord parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.SnapshotRecord parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.SnapshotRecord parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.SnapshotRecord parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.SnapshotRecord parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.SnapshotRecord parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.SnapshotRecord parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.SnapshotRecord parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.SnapshotRecord parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.SnapshotRecord parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.SnapshotRecord parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.SnapshotRecord parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.util.SnapshotRecord prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * <pre>
-   * Each SnapshotRecord represents one batch of pre-processed input data. A batch
-   * consists of a list of tensors that we encode as TensorProtos. This message
-   * doesn't store the structure of the batch.
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.data.experimental.SnapshotRecord}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.data.experimental.SnapshotRecord)
-      org.tensorflow.proto.util.SnapshotRecordOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.util.SnapshotProtos.internal_static_tensorflow_data_experimental_SnapshotRecord_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.util.SnapshotProtos.internal_static_tensorflow_data_experimental_SnapshotRecord_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.util.SnapshotRecord.class, org.tensorflow.proto.util.SnapshotRecord.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.util.SnapshotRecord.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-        getTensorFieldBuilder();
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      if (tensorBuilder_ == null) {
-        tensor_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000001);
-      } else {
-        tensorBuilder_.clear();
-      }
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.util.SnapshotProtos.internal_static_tensorflow_data_experimental_SnapshotRecord_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.SnapshotRecord getDefaultInstanceForType() {
-      return org.tensorflow.proto.util.SnapshotRecord.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.SnapshotRecord build() {
-      org.tensorflow.proto.util.SnapshotRecord result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.SnapshotRecord buildPartial() {
-      org.tensorflow.proto.util.SnapshotRecord result = new org.tensorflow.proto.util.SnapshotRecord(this);
-      int from_bitField0_ = bitField0_;
-      if (tensorBuilder_ == null) {
-        if (((bitField0_ & 0x00000001) != 0)) {
-          tensor_ = java.util.Collections.unmodifiableList(tensor_);
-          bitField0_ = (bitField0_ & ~0x00000001);
-        }
-        result.tensor_ = tensor_;
-      } else {
-        result.tensor_ = tensorBuilder_.build();
-      }
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.util.SnapshotRecord) {
-        return mergeFrom((org.tensorflow.proto.util.SnapshotRecord)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.util.SnapshotRecord other) {
-      if (other == org.tensorflow.proto.util.SnapshotRecord.getDefaultInstance()) return this;
-      if (tensorBuilder_ == null) {
-        if (!other.tensor_.isEmpty()) {
-          if (tensor_.isEmpty()) {
-            tensor_ = other.tensor_;
-            bitField0_ = (bitField0_ & ~0x00000001);
-          } else {
-            ensureTensorIsMutable();
-            tensor_.addAll(other.tensor_);
-          }
-          onChanged();
-        }
-      } else {
-        if (!other.tensor_.isEmpty()) {
-          if (tensorBuilder_.isEmpty()) {
-            tensorBuilder_.dispose();
-            tensorBuilder_ = null;
-            tensor_ = other.tensor_;
-            bitField0_ = (bitField0_ & ~0x00000001);
-            tensorBuilder_ = 
-              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
-                 getTensorFieldBuilder() : null;
-          } else {
-            tensorBuilder_.addAllMessages(other.tensor_);
-          }
-        }
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.util.SnapshotRecord parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.util.SnapshotRecord) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-    private int bitField0_;
-
-    private java.util.List<org.tensorflow.proto.framework.TensorProto> tensor_ =
-      java.util.Collections.emptyList();
-    private void ensureTensorIsMutable() {
-      if (!((bitField0_ & 0x00000001) != 0)) {
-        tensor_ = new java.util.ArrayList<org.tensorflow.proto.framework.TensorProto>(tensor_);
-        bitField0_ |= 0x00000001;
-       }
-    }
-
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.framework.TensorProto, org.tensorflow.proto.framework.TensorProto.Builder, org.tensorflow.proto.framework.TensorProtoOrBuilder> tensorBuilder_;
-
-    /**
-     * <code>repeated .tensorflow.TensorProto tensor = 1;</code>
-     */
-    public java.util.List<org.tensorflow.proto.framework.TensorProto> getTensorList() {
-      if (tensorBuilder_ == null) {
-        return java.util.Collections.unmodifiableList(tensor_);
-      } else {
-        return tensorBuilder_.getMessageList();
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.TensorProto tensor = 1;</code>
-     */
-    public int getTensorCount() {
-      if (tensorBuilder_ == null) {
-        return tensor_.size();
-      } else {
-        return tensorBuilder_.getCount();
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.TensorProto tensor = 1;</code>
-     */
-    public org.tensorflow.proto.framework.TensorProto getTensor(int index) {
-      if (tensorBuilder_ == null) {
-        return tensor_.get(index);
-      } else {
-        return tensorBuilder_.getMessage(index);
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.TensorProto tensor = 1;</code>
-     */
-    public Builder setTensor(
-        int index, org.tensorflow.proto.framework.TensorProto value) {
-      if (tensorBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureTensorIsMutable();
-        tensor_.set(index, value);
-        onChanged();
-      } else {
-        tensorBuilder_.setMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.TensorProto tensor = 1;</code>
-     */
-    public Builder setTensor(
-        int index, org.tensorflow.proto.framework.TensorProto.Builder builderForValue) {
-      if (tensorBuilder_ == null) {
-        ensureTensorIsMutable();
-        tensor_.set(index, builderForValue.build());
-        onChanged();
-      } else {
-        tensorBuilder_.setMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.TensorProto tensor = 1;</code>
-     */
-    public Builder addTensor(org.tensorflow.proto.framework.TensorProto value) {
-      if (tensorBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureTensorIsMutable();
-        tensor_.add(value);
-        onChanged();
-      } else {
-        tensorBuilder_.addMessage(value);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.TensorProto tensor = 1;</code>
-     */
-    public Builder addTensor(
-        int index, org.tensorflow.proto.framework.TensorProto value) {
-      if (tensorBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureTensorIsMutable();
-        tensor_.add(index, value);
-        onChanged();
-      } else {
-        tensorBuilder_.addMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.TensorProto tensor = 1;</code>
-     */
-    public Builder addTensor(
-        org.tensorflow.proto.framework.TensorProto.Builder builderForValue) {
-      if (tensorBuilder_ == null) {
-        ensureTensorIsMutable();
-        tensor_.add(builderForValue.build());
-        onChanged();
-      } else {
-        tensorBuilder_.addMessage(builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.TensorProto tensor = 1;</code>
-     */
-    public Builder addTensor(
-        int index, org.tensorflow.proto.framework.TensorProto.Builder builderForValue) {
-      if (tensorBuilder_ == null) {
-        ensureTensorIsMutable();
-        tensor_.add(index, builderForValue.build());
-        onChanged();
-      } else {
-        tensorBuilder_.addMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.TensorProto tensor = 1;</code>
-     */
-    public Builder addAllTensor(
-        java.lang.Iterable<? extends org.tensorflow.proto.framework.TensorProto> values) {
-      if (tensorBuilder_ == null) {
-        ensureTensorIsMutable();
-        com.google.protobuf.AbstractMessageLite.Builder.addAll(
-            values, tensor_);
-        onChanged();
-      } else {
-        tensorBuilder_.addAllMessages(values);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.TensorProto tensor = 1;</code>
-     */
-    public Builder clearTensor() {
-      if (tensorBuilder_ == null) {
-        tensor_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000001);
-        onChanged();
-      } else {
-        tensorBuilder_.clear();
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.TensorProto tensor = 1;</code>
-     */
-    public Builder removeTensor(int index) {
-      if (tensorBuilder_ == null) {
-        ensureTensorIsMutable();
-        tensor_.remove(index);
-        onChanged();
-      } else {
-        tensorBuilder_.remove(index);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.TensorProto tensor = 1;</code>
-     */
-    public org.tensorflow.proto.framework.TensorProto.Builder getTensorBuilder(
-        int index) {
-      return getTensorFieldBuilder().getBuilder(index);
-    }
-    /**
-     * <code>repeated .tensorflow.TensorProto tensor = 1;</code>
-     */
-    public org.tensorflow.proto.framework.TensorProtoOrBuilder getTensorOrBuilder(
-        int index) {
-      if (tensorBuilder_ == null) {
-        return tensor_.get(index);  } else {
-        return tensorBuilder_.getMessageOrBuilder(index);
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.TensorProto tensor = 1;</code>
-     */
-    public java.util.List<? extends org.tensorflow.proto.framework.TensorProtoOrBuilder> 
-         getTensorOrBuilderList() {
-      if (tensorBuilder_ != null) {
-        return tensorBuilder_.getMessageOrBuilderList();
-      } else {
-        return java.util.Collections.unmodifiableList(tensor_);
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.TensorProto tensor = 1;</code>
-     */
-    public org.tensorflow.proto.framework.TensorProto.Builder addTensorBuilder() {
-      return getTensorFieldBuilder().addBuilder(
-          org.tensorflow.proto.framework.TensorProto.getDefaultInstance());
-    }
-    /**
-     * <code>repeated .tensorflow.TensorProto tensor = 1;</code>
-     */
-    public org.tensorflow.proto.framework.TensorProto.Builder addTensorBuilder(
-        int index) {
-      return getTensorFieldBuilder().addBuilder(
-          index, org.tensorflow.proto.framework.TensorProto.getDefaultInstance());
-    }
-    /**
-     * <code>repeated .tensorflow.TensorProto tensor = 1;</code>
-     */
-    public java.util.List<org.tensorflow.proto.framework.TensorProto.Builder> 
-         getTensorBuilderList() {
-      return getTensorFieldBuilder().getBuilderList();
-    }
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.framework.TensorProto, org.tensorflow.proto.framework.TensorProto.Builder, org.tensorflow.proto.framework.TensorProtoOrBuilder> 
-        getTensorFieldBuilder() {
-      if (tensorBuilder_ == null) {
-        tensorBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
-            org.tensorflow.proto.framework.TensorProto, org.tensorflow.proto.framework.TensorProto.Builder, org.tensorflow.proto.framework.TensorProtoOrBuilder>(
-                tensor_,
-                ((bitField0_ & 0x00000001) != 0),
-                getParentForChildren(),
-                isClean());
-        tensor_ = null;
-      }
-      return tensorBuilder_;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.data.experimental.SnapshotRecord)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.data.experimental.SnapshotRecord)
-  private static final org.tensorflow.proto.util.SnapshotRecord DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.util.SnapshotRecord();
-  }
-
-  public static org.tensorflow.proto.util.SnapshotRecord getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<SnapshotRecord>
-      PARSER = new com.google.protobuf.AbstractParser<SnapshotRecord>() {
-    @java.lang.Override
-    public SnapshotRecord parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new SnapshotRecord(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<SnapshotRecord> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<SnapshotRecord> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.util.SnapshotRecord getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SnapshotRecordOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SnapshotRecordOrBuilder.java
deleted file mode 100644
index e420dac0f6b..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SnapshotRecordOrBuilder.java
+++ /dev/null
@@ -1,33 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/data/experimental/snapshot.proto
-
-package org.tensorflow.proto.util;
-
-public interface SnapshotRecordOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.data.experimental.SnapshotRecord)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <code>repeated .tensorflow.TensorProto tensor = 1;</code>
-   */
-  java.util.List<org.tensorflow.proto.framework.TensorProto> 
-      getTensorList();
-  /**
-   * <code>repeated .tensorflow.TensorProto tensor = 1;</code>
-   */
-  org.tensorflow.proto.framework.TensorProto getTensor(int index);
-  /**
-   * <code>repeated .tensorflow.TensorProto tensor = 1;</code>
-   */
-  int getTensorCount();
-  /**
-   * <code>repeated .tensorflow.TensorProto tensor = 1;</code>
-   */
-  java.util.List<? extends org.tensorflow.proto.framework.TensorProtoOrBuilder> 
-      getTensorOrBuilderList();
-  /**
-   * <code>repeated .tensorflow.TensorProto tensor = 1;</code>
-   */
-  org.tensorflow.proto.framework.TensorProtoOrBuilder getTensorOrBuilder(
-      int index);
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SnapshotTensorMetadata.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SnapshotTensorMetadata.java
deleted file mode 100644
index 2d8d9dc96de..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SnapshotTensorMetadata.java
+++ /dev/null
@@ -1,773 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/data/experimental/snapshot.proto
-
-package org.tensorflow.proto.util;
-
-/**
- * <pre>
- * Metadata for all the tensors in a Snapshot Record.
- * </pre>
- *
- * Protobuf type {@code tensorflow.data.experimental.SnapshotTensorMetadata}
- */
-public  final class SnapshotTensorMetadata extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.data.experimental.SnapshotTensorMetadata)
-    SnapshotTensorMetadataOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use SnapshotTensorMetadata.newBuilder() to construct.
-  private SnapshotTensorMetadata(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private SnapshotTensorMetadata() {
-    tensorMetadata_ = java.util.Collections.emptyList();
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new SnapshotTensorMetadata();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private SnapshotTensorMetadata(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    int mutable_bitField0_ = 0;
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 10: {
-            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
-              tensorMetadata_ = new java.util.ArrayList<org.tensorflow.proto.util.TensorMetadata>();
-              mutable_bitField0_ |= 0x00000001;
-            }
-            tensorMetadata_.add(
-                input.readMessage(org.tensorflow.proto.util.TensorMetadata.parser(), extensionRegistry));
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      if (((mutable_bitField0_ & 0x00000001) != 0)) {
-        tensorMetadata_ = java.util.Collections.unmodifiableList(tensorMetadata_);
-      }
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.util.SnapshotProtos.internal_static_tensorflow_data_experimental_SnapshotTensorMetadata_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.util.SnapshotProtos.internal_static_tensorflow_data_experimental_SnapshotTensorMetadata_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.util.SnapshotTensorMetadata.class, org.tensorflow.proto.util.SnapshotTensorMetadata.Builder.class);
-  }
-
-  public static final int TENSOR_METADATA_FIELD_NUMBER = 1;
-  private java.util.List<org.tensorflow.proto.util.TensorMetadata> tensorMetadata_;
-  /**
-   * <code>repeated .tensorflow.data.experimental.TensorMetadata tensor_metadata = 1;</code>
-   */
-  public java.util.List<org.tensorflow.proto.util.TensorMetadata> getTensorMetadataList() {
-    return tensorMetadata_;
-  }
-  /**
-   * <code>repeated .tensorflow.data.experimental.TensorMetadata tensor_metadata = 1;</code>
-   */
-  public java.util.List<? extends org.tensorflow.proto.util.TensorMetadataOrBuilder> 
-      getTensorMetadataOrBuilderList() {
-    return tensorMetadata_;
-  }
-  /**
-   * <code>repeated .tensorflow.data.experimental.TensorMetadata tensor_metadata = 1;</code>
-   */
-  public int getTensorMetadataCount() {
-    return tensorMetadata_.size();
-  }
-  /**
-   * <code>repeated .tensorflow.data.experimental.TensorMetadata tensor_metadata = 1;</code>
-   */
-  public org.tensorflow.proto.util.TensorMetadata getTensorMetadata(int index) {
-    return tensorMetadata_.get(index);
-  }
-  /**
-   * <code>repeated .tensorflow.data.experimental.TensorMetadata tensor_metadata = 1;</code>
-   */
-  public org.tensorflow.proto.util.TensorMetadataOrBuilder getTensorMetadataOrBuilder(
-      int index) {
-    return tensorMetadata_.get(index);
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    for (int i = 0; i < tensorMetadata_.size(); i++) {
-      output.writeMessage(1, tensorMetadata_.get(i));
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    for (int i = 0; i < tensorMetadata_.size(); i++) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(1, tensorMetadata_.get(i));
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.util.SnapshotTensorMetadata)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.util.SnapshotTensorMetadata other = (org.tensorflow.proto.util.SnapshotTensorMetadata) obj;
-
-    if (!getTensorMetadataList()
-        .equals(other.getTensorMetadataList())) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    if (getTensorMetadataCount() > 0) {
-      hash = (37 * hash) + TENSOR_METADATA_FIELD_NUMBER;
-      hash = (53 * hash) + getTensorMetadataList().hashCode();
-    }
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.util.SnapshotTensorMetadata parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.SnapshotTensorMetadata parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.SnapshotTensorMetadata parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.SnapshotTensorMetadata parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.SnapshotTensorMetadata parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.SnapshotTensorMetadata parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.SnapshotTensorMetadata parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.SnapshotTensorMetadata parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.SnapshotTensorMetadata parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.SnapshotTensorMetadata parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.SnapshotTensorMetadata parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.SnapshotTensorMetadata parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.util.SnapshotTensorMetadata prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * <pre>
-   * Metadata for all the tensors in a Snapshot Record.
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.data.experimental.SnapshotTensorMetadata}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.data.experimental.SnapshotTensorMetadata)
-      org.tensorflow.proto.util.SnapshotTensorMetadataOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.util.SnapshotProtos.internal_static_tensorflow_data_experimental_SnapshotTensorMetadata_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.util.SnapshotProtos.internal_static_tensorflow_data_experimental_SnapshotTensorMetadata_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.util.SnapshotTensorMetadata.class, org.tensorflow.proto.util.SnapshotTensorMetadata.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.util.SnapshotTensorMetadata.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-        getTensorMetadataFieldBuilder();
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      if (tensorMetadataBuilder_ == null) {
-        tensorMetadata_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000001);
-      } else {
-        tensorMetadataBuilder_.clear();
-      }
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.util.SnapshotProtos.internal_static_tensorflow_data_experimental_SnapshotTensorMetadata_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.SnapshotTensorMetadata getDefaultInstanceForType() {
-      return org.tensorflow.proto.util.SnapshotTensorMetadata.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.SnapshotTensorMetadata build() {
-      org.tensorflow.proto.util.SnapshotTensorMetadata result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.SnapshotTensorMetadata buildPartial() {
-      org.tensorflow.proto.util.SnapshotTensorMetadata result = new org.tensorflow.proto.util.SnapshotTensorMetadata(this);
-      int from_bitField0_ = bitField0_;
-      if (tensorMetadataBuilder_ == null) {
-        if (((bitField0_ & 0x00000001) != 0)) {
-          tensorMetadata_ = java.util.Collections.unmodifiableList(tensorMetadata_);
-          bitField0_ = (bitField0_ & ~0x00000001);
-        }
-        result.tensorMetadata_ = tensorMetadata_;
-      } else {
-        result.tensorMetadata_ = tensorMetadataBuilder_.build();
-      }
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.util.SnapshotTensorMetadata) {
-        return mergeFrom((org.tensorflow.proto.util.SnapshotTensorMetadata)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.util.SnapshotTensorMetadata other) {
-      if (other == org.tensorflow.proto.util.SnapshotTensorMetadata.getDefaultInstance()) return this;
-      if (tensorMetadataBuilder_ == null) {
-        if (!other.tensorMetadata_.isEmpty()) {
-          if (tensorMetadata_.isEmpty()) {
-            tensorMetadata_ = other.tensorMetadata_;
-            bitField0_ = (bitField0_ & ~0x00000001);
-          } else {
-            ensureTensorMetadataIsMutable();
-            tensorMetadata_.addAll(other.tensorMetadata_);
-          }
-          onChanged();
-        }
-      } else {
-        if (!other.tensorMetadata_.isEmpty()) {
-          if (tensorMetadataBuilder_.isEmpty()) {
-            tensorMetadataBuilder_.dispose();
-            tensorMetadataBuilder_ = null;
-            tensorMetadata_ = other.tensorMetadata_;
-            bitField0_ = (bitField0_ & ~0x00000001);
-            tensorMetadataBuilder_ = 
-              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
-                 getTensorMetadataFieldBuilder() : null;
-          } else {
-            tensorMetadataBuilder_.addAllMessages(other.tensorMetadata_);
-          }
-        }
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.util.SnapshotTensorMetadata parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.util.SnapshotTensorMetadata) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-    private int bitField0_;
-
-    private java.util.List<org.tensorflow.proto.util.TensorMetadata> tensorMetadata_ =
-      java.util.Collections.emptyList();
-    private void ensureTensorMetadataIsMutable() {
-      if (!((bitField0_ & 0x00000001) != 0)) {
-        tensorMetadata_ = new java.util.ArrayList<org.tensorflow.proto.util.TensorMetadata>(tensorMetadata_);
-        bitField0_ |= 0x00000001;
-       }
-    }
-
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.util.TensorMetadata, org.tensorflow.proto.util.TensorMetadata.Builder, org.tensorflow.proto.util.TensorMetadataOrBuilder> tensorMetadataBuilder_;
-
-    /**
-     * <code>repeated .tensorflow.data.experimental.TensorMetadata tensor_metadata = 1;</code>
-     */
-    public java.util.List<org.tensorflow.proto.util.TensorMetadata> getTensorMetadataList() {
-      if (tensorMetadataBuilder_ == null) {
-        return java.util.Collections.unmodifiableList(tensorMetadata_);
-      } else {
-        return tensorMetadataBuilder_.getMessageList();
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.data.experimental.TensorMetadata tensor_metadata = 1;</code>
-     */
-    public int getTensorMetadataCount() {
-      if (tensorMetadataBuilder_ == null) {
-        return tensorMetadata_.size();
-      } else {
-        return tensorMetadataBuilder_.getCount();
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.data.experimental.TensorMetadata tensor_metadata = 1;</code>
-     */
-    public org.tensorflow.proto.util.TensorMetadata getTensorMetadata(int index) {
-      if (tensorMetadataBuilder_ == null) {
-        return tensorMetadata_.get(index);
-      } else {
-        return tensorMetadataBuilder_.getMessage(index);
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.data.experimental.TensorMetadata tensor_metadata = 1;</code>
-     */
-    public Builder setTensorMetadata(
-        int index, org.tensorflow.proto.util.TensorMetadata value) {
-      if (tensorMetadataBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureTensorMetadataIsMutable();
-        tensorMetadata_.set(index, value);
-        onChanged();
-      } else {
-        tensorMetadataBuilder_.setMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.data.experimental.TensorMetadata tensor_metadata = 1;</code>
-     */
-    public Builder setTensorMetadata(
-        int index, org.tensorflow.proto.util.TensorMetadata.Builder builderForValue) {
-      if (tensorMetadataBuilder_ == null) {
-        ensureTensorMetadataIsMutable();
-        tensorMetadata_.set(index, builderForValue.build());
-        onChanged();
-      } else {
-        tensorMetadataBuilder_.setMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.data.experimental.TensorMetadata tensor_metadata = 1;</code>
-     */
-    public Builder addTensorMetadata(org.tensorflow.proto.util.TensorMetadata value) {
-      if (tensorMetadataBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureTensorMetadataIsMutable();
-        tensorMetadata_.add(value);
-        onChanged();
-      } else {
-        tensorMetadataBuilder_.addMessage(value);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.data.experimental.TensorMetadata tensor_metadata = 1;</code>
-     */
-    public Builder addTensorMetadata(
-        int index, org.tensorflow.proto.util.TensorMetadata value) {
-      if (tensorMetadataBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureTensorMetadataIsMutable();
-        tensorMetadata_.add(index, value);
-        onChanged();
-      } else {
-        tensorMetadataBuilder_.addMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.data.experimental.TensorMetadata tensor_metadata = 1;</code>
-     */
-    public Builder addTensorMetadata(
-        org.tensorflow.proto.util.TensorMetadata.Builder builderForValue) {
-      if (tensorMetadataBuilder_ == null) {
-        ensureTensorMetadataIsMutable();
-        tensorMetadata_.add(builderForValue.build());
-        onChanged();
-      } else {
-        tensorMetadataBuilder_.addMessage(builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.data.experimental.TensorMetadata tensor_metadata = 1;</code>
-     */
-    public Builder addTensorMetadata(
-        int index, org.tensorflow.proto.util.TensorMetadata.Builder builderForValue) {
-      if (tensorMetadataBuilder_ == null) {
-        ensureTensorMetadataIsMutable();
-        tensorMetadata_.add(index, builderForValue.build());
-        onChanged();
-      } else {
-        tensorMetadataBuilder_.addMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.data.experimental.TensorMetadata tensor_metadata = 1;</code>
-     */
-    public Builder addAllTensorMetadata(
-        java.lang.Iterable<? extends org.tensorflow.proto.util.TensorMetadata> values) {
-      if (tensorMetadataBuilder_ == null) {
-        ensureTensorMetadataIsMutable();
-        com.google.protobuf.AbstractMessageLite.Builder.addAll(
-            values, tensorMetadata_);
-        onChanged();
-      } else {
-        tensorMetadataBuilder_.addAllMessages(values);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.data.experimental.TensorMetadata tensor_metadata = 1;</code>
-     */
-    public Builder clearTensorMetadata() {
-      if (tensorMetadataBuilder_ == null) {
-        tensorMetadata_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000001);
-        onChanged();
-      } else {
-        tensorMetadataBuilder_.clear();
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.data.experimental.TensorMetadata tensor_metadata = 1;</code>
-     */
-    public Builder removeTensorMetadata(int index) {
-      if (tensorMetadataBuilder_ == null) {
-        ensureTensorMetadataIsMutable();
-        tensorMetadata_.remove(index);
-        onChanged();
-      } else {
-        tensorMetadataBuilder_.remove(index);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.data.experimental.TensorMetadata tensor_metadata = 1;</code>
-     */
-    public org.tensorflow.proto.util.TensorMetadata.Builder getTensorMetadataBuilder(
-        int index) {
-      return getTensorMetadataFieldBuilder().getBuilder(index);
-    }
-    /**
-     * <code>repeated .tensorflow.data.experimental.TensorMetadata tensor_metadata = 1;</code>
-     */
-    public org.tensorflow.proto.util.TensorMetadataOrBuilder getTensorMetadataOrBuilder(
-        int index) {
-      if (tensorMetadataBuilder_ == null) {
-        return tensorMetadata_.get(index);  } else {
-        return tensorMetadataBuilder_.getMessageOrBuilder(index);
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.data.experimental.TensorMetadata tensor_metadata = 1;</code>
-     */
-    public java.util.List<? extends org.tensorflow.proto.util.TensorMetadataOrBuilder> 
-         getTensorMetadataOrBuilderList() {
-      if (tensorMetadataBuilder_ != null) {
-        return tensorMetadataBuilder_.getMessageOrBuilderList();
-      } else {
-        return java.util.Collections.unmodifiableList(tensorMetadata_);
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.data.experimental.TensorMetadata tensor_metadata = 1;</code>
-     */
-    public org.tensorflow.proto.util.TensorMetadata.Builder addTensorMetadataBuilder() {
-      return getTensorMetadataFieldBuilder().addBuilder(
-          org.tensorflow.proto.util.TensorMetadata.getDefaultInstance());
-    }
-    /**
-     * <code>repeated .tensorflow.data.experimental.TensorMetadata tensor_metadata = 1;</code>
-     */
-    public org.tensorflow.proto.util.TensorMetadata.Builder addTensorMetadataBuilder(
-        int index) {
-      return getTensorMetadataFieldBuilder().addBuilder(
-          index, org.tensorflow.proto.util.TensorMetadata.getDefaultInstance());
-    }
-    /**
-     * <code>repeated .tensorflow.data.experimental.TensorMetadata tensor_metadata = 1;</code>
-     */
-    public java.util.List<org.tensorflow.proto.util.TensorMetadata.Builder> 
-         getTensorMetadataBuilderList() {
-      return getTensorMetadataFieldBuilder().getBuilderList();
-    }
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.util.TensorMetadata, org.tensorflow.proto.util.TensorMetadata.Builder, org.tensorflow.proto.util.TensorMetadataOrBuilder> 
-        getTensorMetadataFieldBuilder() {
-      if (tensorMetadataBuilder_ == null) {
-        tensorMetadataBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
-            org.tensorflow.proto.util.TensorMetadata, org.tensorflow.proto.util.TensorMetadata.Builder, org.tensorflow.proto.util.TensorMetadataOrBuilder>(
-                tensorMetadata_,
-                ((bitField0_ & 0x00000001) != 0),
-                getParentForChildren(),
-                isClean());
-        tensorMetadata_ = null;
-      }
-      return tensorMetadataBuilder_;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.data.experimental.SnapshotTensorMetadata)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.data.experimental.SnapshotTensorMetadata)
-  private static final org.tensorflow.proto.util.SnapshotTensorMetadata DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.util.SnapshotTensorMetadata();
-  }
-
-  public static org.tensorflow.proto.util.SnapshotTensorMetadata getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<SnapshotTensorMetadata>
-      PARSER = new com.google.protobuf.AbstractParser<SnapshotTensorMetadata>() {
-    @java.lang.Override
-    public SnapshotTensorMetadata parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new SnapshotTensorMetadata(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<SnapshotTensorMetadata> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<SnapshotTensorMetadata> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.util.SnapshotTensorMetadata getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SnapshotTensorMetadataOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SnapshotTensorMetadataOrBuilder.java
deleted file mode 100644
index 8c906ffb714..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SnapshotTensorMetadataOrBuilder.java
+++ /dev/null
@@ -1,33 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/data/experimental/snapshot.proto
-
-package org.tensorflow.proto.util;
-
-public interface SnapshotTensorMetadataOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.data.experimental.SnapshotTensorMetadata)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <code>repeated .tensorflow.data.experimental.TensorMetadata tensor_metadata = 1;</code>
-   */
-  java.util.List<org.tensorflow.proto.util.TensorMetadata> 
-      getTensorMetadataList();
-  /**
-   * <code>repeated .tensorflow.data.experimental.TensorMetadata tensor_metadata = 1;</code>
-   */
-  org.tensorflow.proto.util.TensorMetadata getTensorMetadata(int index);
-  /**
-   * <code>repeated .tensorflow.data.experimental.TensorMetadata tensor_metadata = 1;</code>
-   */
-  int getTensorMetadataCount();
-  /**
-   * <code>repeated .tensorflow.data.experimental.TensorMetadata tensor_metadata = 1;</code>
-   */
-  java.util.List<? extends org.tensorflow.proto.util.TensorMetadataOrBuilder> 
-      getTensorMetadataOrBuilderList();
-  /**
-   * <code>repeated .tensorflow.data.experimental.TensorMetadata tensor_metadata = 1;</code>
-   */
-  org.tensorflow.proto.util.TensorMetadataOrBuilder getTensorMetadataOrBuilder(
-      int index);
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SourceFile.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SourceFile.java
deleted file mode 100644
index b8f6c068649..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SourceFile.java
+++ /dev/null
@@ -1,964 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/debug_event.proto
-
-package org.tensorflow.proto.util;
-
-/**
- * <pre>
- * Content of a source file involved in the execution of the debugged TensorFlow
- * program.
- * </pre>
- *
- * Protobuf type {@code tensorflow.SourceFile}
- */
-public  final class SourceFile extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.SourceFile)
-    SourceFileOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use SourceFile.newBuilder() to construct.
-  private SourceFile(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private SourceFile() {
-    filePath_ = "";
-    hostName_ = "";
-    lines_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new SourceFile();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private SourceFile(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    int mutable_bitField0_ = 0;
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 10: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            filePath_ = s;
-            break;
-          }
-          case 18: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            hostName_ = s;
-            break;
-          }
-          case 26: {
-            java.lang.String s = input.readStringRequireUtf8();
-            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
-              lines_ = new com.google.protobuf.LazyStringArrayList();
-              mutable_bitField0_ |= 0x00000001;
-            }
-            lines_.add(s);
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      if (((mutable_bitField0_ & 0x00000001) != 0)) {
-        lines_ = lines_.getUnmodifiableView();
-      }
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.util.DebugEventProtos.internal_static_tensorflow_SourceFile_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.util.DebugEventProtos.internal_static_tensorflow_SourceFile_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.util.SourceFile.class, org.tensorflow.proto.util.SourceFile.Builder.class);
-  }
-
-  public static final int FILE_PATH_FIELD_NUMBER = 1;
-  private volatile java.lang.Object filePath_;
-  /**
-   * <pre>
-   * Path to the file.
-   * </pre>
-   *
-   * <code>string file_path = 1;</code>
-   */
-  public java.lang.String getFilePath() {
-    java.lang.Object ref = filePath_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      filePath_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * Path to the file.
-   * </pre>
-   *
-   * <code>string file_path = 1;</code>
-   */
-  public com.google.protobuf.ByteString
-      getFilePathBytes() {
-    java.lang.Object ref = filePath_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      filePath_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int HOST_NAME_FIELD_NUMBER = 2;
-  private volatile java.lang.Object hostName_;
-  /**
-   * <pre>
-   * Name of the host on which the file is located.
-   * </pre>
-   *
-   * <code>string host_name = 2;</code>
-   */
-  public java.lang.String getHostName() {
-    java.lang.Object ref = hostName_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      hostName_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * Name of the host on which the file is located.
-   * </pre>
-   *
-   * <code>string host_name = 2;</code>
-   */
-  public com.google.protobuf.ByteString
-      getHostNameBytes() {
-    java.lang.Object ref = hostName_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      hostName_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int LINES_FIELD_NUMBER = 3;
-  private com.google.protobuf.LazyStringList lines_;
-  /**
-   * <pre>
-   * Line-by-line content of the file.
-   * </pre>
-   *
-   * <code>repeated string lines = 3;</code>
-   */
-  public com.google.protobuf.ProtocolStringList
-      getLinesList() {
-    return lines_;
-  }
-  /**
-   * <pre>
-   * Line-by-line content of the file.
-   * </pre>
-   *
-   * <code>repeated string lines = 3;</code>
-   */
-  public int getLinesCount() {
-    return lines_.size();
-  }
-  /**
-   * <pre>
-   * Line-by-line content of the file.
-   * </pre>
-   *
-   * <code>repeated string lines = 3;</code>
-   */
-  public java.lang.String getLines(int index) {
-    return lines_.get(index);
-  }
-  /**
-   * <pre>
-   * Line-by-line content of the file.
-   * </pre>
-   *
-   * <code>repeated string lines = 3;</code>
-   */
-  public com.google.protobuf.ByteString
-      getLinesBytes(int index) {
-    return lines_.getByteString(index);
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    if (!getFilePathBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 1, filePath_);
-    }
-    if (!getHostNameBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 2, hostName_);
-    }
-    for (int i = 0; i < lines_.size(); i++) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 3, lines_.getRaw(i));
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (!getFilePathBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, filePath_);
-    }
-    if (!getHostNameBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(2, hostName_);
-    }
-    {
-      int dataSize = 0;
-      for (int i = 0; i < lines_.size(); i++) {
-        dataSize += computeStringSizeNoTag(lines_.getRaw(i));
-      }
-      size += dataSize;
-      size += 1 * getLinesList().size();
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.util.SourceFile)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.util.SourceFile other = (org.tensorflow.proto.util.SourceFile) obj;
-
-    if (!getFilePath()
-        .equals(other.getFilePath())) return false;
-    if (!getHostName()
-        .equals(other.getHostName())) return false;
-    if (!getLinesList()
-        .equals(other.getLinesList())) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    hash = (37 * hash) + FILE_PATH_FIELD_NUMBER;
-    hash = (53 * hash) + getFilePath().hashCode();
-    hash = (37 * hash) + HOST_NAME_FIELD_NUMBER;
-    hash = (53 * hash) + getHostName().hashCode();
-    if (getLinesCount() > 0) {
-      hash = (37 * hash) + LINES_FIELD_NUMBER;
-      hash = (53 * hash) + getLinesList().hashCode();
-    }
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.util.SourceFile parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.SourceFile parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.SourceFile parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.SourceFile parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.SourceFile parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.SourceFile parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.SourceFile parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.SourceFile parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.SourceFile parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.SourceFile parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.SourceFile parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.SourceFile parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.util.SourceFile prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * <pre>
-   * Content of a source file involved in the execution of the debugged TensorFlow
-   * program.
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.SourceFile}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.SourceFile)
-      org.tensorflow.proto.util.SourceFileOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.util.DebugEventProtos.internal_static_tensorflow_SourceFile_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.util.DebugEventProtos.internal_static_tensorflow_SourceFile_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.util.SourceFile.class, org.tensorflow.proto.util.SourceFile.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.util.SourceFile.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      filePath_ = "";
-
-      hostName_ = "";
-
-      lines_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-      bitField0_ = (bitField0_ & ~0x00000001);
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.util.DebugEventProtos.internal_static_tensorflow_SourceFile_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.SourceFile getDefaultInstanceForType() {
-      return org.tensorflow.proto.util.SourceFile.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.SourceFile build() {
-      org.tensorflow.proto.util.SourceFile result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.SourceFile buildPartial() {
-      org.tensorflow.proto.util.SourceFile result = new org.tensorflow.proto.util.SourceFile(this);
-      int from_bitField0_ = bitField0_;
-      result.filePath_ = filePath_;
-      result.hostName_ = hostName_;
-      if (((bitField0_ & 0x00000001) != 0)) {
-        lines_ = lines_.getUnmodifiableView();
-        bitField0_ = (bitField0_ & ~0x00000001);
-      }
-      result.lines_ = lines_;
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.util.SourceFile) {
-        return mergeFrom((org.tensorflow.proto.util.SourceFile)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.util.SourceFile other) {
-      if (other == org.tensorflow.proto.util.SourceFile.getDefaultInstance()) return this;
-      if (!other.getFilePath().isEmpty()) {
-        filePath_ = other.filePath_;
-        onChanged();
-      }
-      if (!other.getHostName().isEmpty()) {
-        hostName_ = other.hostName_;
-        onChanged();
-      }
-      if (!other.lines_.isEmpty()) {
-        if (lines_.isEmpty()) {
-          lines_ = other.lines_;
-          bitField0_ = (bitField0_ & ~0x00000001);
-        } else {
-          ensureLinesIsMutable();
-          lines_.addAll(other.lines_);
-        }
-        onChanged();
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.util.SourceFile parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.util.SourceFile) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-    private int bitField0_;
-
-    private java.lang.Object filePath_ = "";
-    /**
-     * <pre>
-     * Path to the file.
-     * </pre>
-     *
-     * <code>string file_path = 1;</code>
-     */
-    public java.lang.String getFilePath() {
-      java.lang.Object ref = filePath_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        filePath_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Path to the file.
-     * </pre>
-     *
-     * <code>string file_path = 1;</code>
-     */
-    public com.google.protobuf.ByteString
-        getFilePathBytes() {
-      java.lang.Object ref = filePath_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        filePath_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Path to the file.
-     * </pre>
-     *
-     * <code>string file_path = 1;</code>
-     */
-    public Builder setFilePath(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      filePath_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Path to the file.
-     * </pre>
-     *
-     * <code>string file_path = 1;</code>
-     */
-    public Builder clearFilePath() {
-      
-      filePath_ = getDefaultInstance().getFilePath();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Path to the file.
-     * </pre>
-     *
-     * <code>string file_path = 1;</code>
-     */
-    public Builder setFilePathBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      filePath_ = value;
-      onChanged();
-      return this;
-    }
-
-    private java.lang.Object hostName_ = "";
-    /**
-     * <pre>
-     * Name of the host on which the file is located.
-     * </pre>
-     *
-     * <code>string host_name = 2;</code>
-     */
-    public java.lang.String getHostName() {
-      java.lang.Object ref = hostName_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        hostName_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Name of the host on which the file is located.
-     * </pre>
-     *
-     * <code>string host_name = 2;</code>
-     */
-    public com.google.protobuf.ByteString
-        getHostNameBytes() {
-      java.lang.Object ref = hostName_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        hostName_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Name of the host on which the file is located.
-     * </pre>
-     *
-     * <code>string host_name = 2;</code>
-     */
-    public Builder setHostName(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      hostName_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Name of the host on which the file is located.
-     * </pre>
-     *
-     * <code>string host_name = 2;</code>
-     */
-    public Builder clearHostName() {
-      
-      hostName_ = getDefaultInstance().getHostName();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Name of the host on which the file is located.
-     * </pre>
-     *
-     * <code>string host_name = 2;</code>
-     */
-    public Builder setHostNameBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      hostName_ = value;
-      onChanged();
-      return this;
-    }
-
-    private com.google.protobuf.LazyStringList lines_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-    private void ensureLinesIsMutable() {
-      if (!((bitField0_ & 0x00000001) != 0)) {
-        lines_ = new com.google.protobuf.LazyStringArrayList(lines_);
-        bitField0_ |= 0x00000001;
-       }
-    }
-    /**
-     * <pre>
-     * Line-by-line content of the file.
-     * </pre>
-     *
-     * <code>repeated string lines = 3;</code>
-     */
-    public com.google.protobuf.ProtocolStringList
-        getLinesList() {
-      return lines_.getUnmodifiableView();
-    }
-    /**
-     * <pre>
-     * Line-by-line content of the file.
-     * </pre>
-     *
-     * <code>repeated string lines = 3;</code>
-     */
-    public int getLinesCount() {
-      return lines_.size();
-    }
-    /**
-     * <pre>
-     * Line-by-line content of the file.
-     * </pre>
-     *
-     * <code>repeated string lines = 3;</code>
-     */
-    public java.lang.String getLines(int index) {
-      return lines_.get(index);
-    }
-    /**
-     * <pre>
-     * Line-by-line content of the file.
-     * </pre>
-     *
-     * <code>repeated string lines = 3;</code>
-     */
-    public com.google.protobuf.ByteString
-        getLinesBytes(int index) {
-      return lines_.getByteString(index);
-    }
-    /**
-     * <pre>
-     * Line-by-line content of the file.
-     * </pre>
-     *
-     * <code>repeated string lines = 3;</code>
-     */
-    public Builder setLines(
-        int index, java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  ensureLinesIsMutable();
-      lines_.set(index, value);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Line-by-line content of the file.
-     * </pre>
-     *
-     * <code>repeated string lines = 3;</code>
-     */
-    public Builder addLines(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  ensureLinesIsMutable();
-      lines_.add(value);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Line-by-line content of the file.
-     * </pre>
-     *
-     * <code>repeated string lines = 3;</code>
-     */
-    public Builder addAllLines(
-        java.lang.Iterable<java.lang.String> values) {
-      ensureLinesIsMutable();
-      com.google.protobuf.AbstractMessageLite.Builder.addAll(
-          values, lines_);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Line-by-line content of the file.
-     * </pre>
-     *
-     * <code>repeated string lines = 3;</code>
-     */
-    public Builder clearLines() {
-      lines_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-      bitField0_ = (bitField0_ & ~0x00000001);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Line-by-line content of the file.
-     * </pre>
-     *
-     * <code>repeated string lines = 3;</code>
-     */
-    public Builder addLinesBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      ensureLinesIsMutable();
-      lines_.add(value);
-      onChanged();
-      return this;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.SourceFile)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.SourceFile)
-  private static final org.tensorflow.proto.util.SourceFile DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.util.SourceFile();
-  }
-
-  public static org.tensorflow.proto.util.SourceFile getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<SourceFile>
-      PARSER = new com.google.protobuf.AbstractParser<SourceFile>() {
-    @java.lang.Override
-    public SourceFile parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new SourceFile(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<SourceFile> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<SourceFile> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.util.SourceFile getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SourceFileOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SourceFileOrBuilder.java
deleted file mode 100644
index b05d03bfb81..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SourceFileOrBuilder.java
+++ /dev/null
@@ -1,80 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/debug_event.proto
-
-package org.tensorflow.proto.util;
-
-public interface SourceFileOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.SourceFile)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <pre>
-   * Path to the file.
-   * </pre>
-   *
-   * <code>string file_path = 1;</code>
-   */
-  java.lang.String getFilePath();
-  /**
-   * <pre>
-   * Path to the file.
-   * </pre>
-   *
-   * <code>string file_path = 1;</code>
-   */
-  com.google.protobuf.ByteString
-      getFilePathBytes();
-
-  /**
-   * <pre>
-   * Name of the host on which the file is located.
-   * </pre>
-   *
-   * <code>string host_name = 2;</code>
-   */
-  java.lang.String getHostName();
-  /**
-   * <pre>
-   * Name of the host on which the file is located.
-   * </pre>
-   *
-   * <code>string host_name = 2;</code>
-   */
-  com.google.protobuf.ByteString
-      getHostNameBytes();
-
-  /**
-   * <pre>
-   * Line-by-line content of the file.
-   * </pre>
-   *
-   * <code>repeated string lines = 3;</code>
-   */
-  java.util.List<java.lang.String>
-      getLinesList();
-  /**
-   * <pre>
-   * Line-by-line content of the file.
-   * </pre>
-   *
-   * <code>repeated string lines = 3;</code>
-   */
-  int getLinesCount();
-  /**
-   * <pre>
-   * Line-by-line content of the file.
-   * </pre>
-   *
-   * <code>repeated string lines = 3;</code>
-   */
-  java.lang.String getLines(int index);
-  /**
-   * <pre>
-   * Line-by-line content of the file.
-   * </pre>
-   *
-   * <code>repeated string lines = 3;</code>
-   */
-  com.google.protobuf.ByteString
-      getLinesBytes(int index);
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/StackFrameWithId.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/StackFrameWithId.java
deleted file mode 100644
index e41ffbd6cc3..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/StackFrameWithId.java
+++ /dev/null
@@ -1,835 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/debug_event.proto
-
-package org.tensorflow.proto.util;
-
-/**
- * <pre>
- * A stack frame with ID.
- * </pre>
- *
- * Protobuf type {@code tensorflow.StackFrameWithId}
- */
-public  final class StackFrameWithId extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.StackFrameWithId)
-    StackFrameWithIdOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use StackFrameWithId.newBuilder() to construct.
-  private StackFrameWithId(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private StackFrameWithId() {
-    id_ = "";
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new StackFrameWithId();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private StackFrameWithId(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 10: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            id_ = s;
-            break;
-          }
-          case 18: {
-            org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol.Builder subBuilder = null;
-            if (fileLineCol_ != null) {
-              subBuilder = fileLineCol_.toBuilder();
-            }
-            fileLineCol_ = input.readMessage(org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol.parser(), extensionRegistry);
-            if (subBuilder != null) {
-              subBuilder.mergeFrom(fileLineCol_);
-              fileLineCol_ = subBuilder.buildPartial();
-            }
-
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.util.DebugEventProtos.internal_static_tensorflow_StackFrameWithId_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.util.DebugEventProtos.internal_static_tensorflow_StackFrameWithId_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.util.StackFrameWithId.class, org.tensorflow.proto.util.StackFrameWithId.Builder.class);
-  }
-
-  public static final int ID_FIELD_NUMBER = 1;
-  private volatile java.lang.Object id_;
-  /**
-   * <pre>
-   * A unique ID for the stack frame: A UUID-like string.
-   * </pre>
-   *
-   * <code>string id = 1;</code>
-   */
-  public java.lang.String getId() {
-    java.lang.Object ref = id_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      id_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * A unique ID for the stack frame: A UUID-like string.
-   * </pre>
-   *
-   * <code>string id = 1;</code>
-   */
-  public com.google.protobuf.ByteString
-      getIdBytes() {
-    java.lang.Object ref = id_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      id_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int FILE_LINE_COL_FIELD_NUMBER = 2;
-  private org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol fileLineCol_;
-  /**
-   * <pre>
-   * Stack frame, i.e., a frame of a stack trace, containing information
-   * regarding the file name, line number, function name, code content
-   * of the line, and column number (if available).
-   * </pre>
-   *
-   * <code>.tensorflow.GraphDebugInfo.FileLineCol file_line_col = 2;</code>
-   */
-  public boolean hasFileLineCol() {
-    return fileLineCol_ != null;
-  }
-  /**
-   * <pre>
-   * Stack frame, i.e., a frame of a stack trace, containing information
-   * regarding the file name, line number, function name, code content
-   * of the line, and column number (if available).
-   * </pre>
-   *
-   * <code>.tensorflow.GraphDebugInfo.FileLineCol file_line_col = 2;</code>
-   */
-  public org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol getFileLineCol() {
-    return fileLineCol_ == null ? org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol.getDefaultInstance() : fileLineCol_;
-  }
-  /**
-   * <pre>
-   * Stack frame, i.e., a frame of a stack trace, containing information
-   * regarding the file name, line number, function name, code content
-   * of the line, and column number (if available).
-   * </pre>
-   *
-   * <code>.tensorflow.GraphDebugInfo.FileLineCol file_line_col = 2;</code>
-   */
-  public org.tensorflow.proto.framework.GraphDebugInfo.FileLineColOrBuilder getFileLineColOrBuilder() {
-    return getFileLineCol();
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    if (!getIdBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 1, id_);
-    }
-    if (fileLineCol_ != null) {
-      output.writeMessage(2, getFileLineCol());
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (!getIdBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, id_);
-    }
-    if (fileLineCol_ != null) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(2, getFileLineCol());
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.util.StackFrameWithId)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.util.StackFrameWithId other = (org.tensorflow.proto.util.StackFrameWithId) obj;
-
-    if (!getId()
-        .equals(other.getId())) return false;
-    if (hasFileLineCol() != other.hasFileLineCol()) return false;
-    if (hasFileLineCol()) {
-      if (!getFileLineCol()
-          .equals(other.getFileLineCol())) return false;
-    }
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    hash = (37 * hash) + ID_FIELD_NUMBER;
-    hash = (53 * hash) + getId().hashCode();
-    if (hasFileLineCol()) {
-      hash = (37 * hash) + FILE_LINE_COL_FIELD_NUMBER;
-      hash = (53 * hash) + getFileLineCol().hashCode();
-    }
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.util.StackFrameWithId parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.StackFrameWithId parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.StackFrameWithId parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.StackFrameWithId parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.StackFrameWithId parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.StackFrameWithId parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.StackFrameWithId parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.StackFrameWithId parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.StackFrameWithId parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.StackFrameWithId parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.StackFrameWithId parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.StackFrameWithId parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.util.StackFrameWithId prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * <pre>
-   * A stack frame with ID.
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.StackFrameWithId}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.StackFrameWithId)
-      org.tensorflow.proto.util.StackFrameWithIdOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.util.DebugEventProtos.internal_static_tensorflow_StackFrameWithId_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.util.DebugEventProtos.internal_static_tensorflow_StackFrameWithId_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.util.StackFrameWithId.class, org.tensorflow.proto.util.StackFrameWithId.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.util.StackFrameWithId.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      id_ = "";
-
-      if (fileLineColBuilder_ == null) {
-        fileLineCol_ = null;
-      } else {
-        fileLineCol_ = null;
-        fileLineColBuilder_ = null;
-      }
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.util.DebugEventProtos.internal_static_tensorflow_StackFrameWithId_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.StackFrameWithId getDefaultInstanceForType() {
-      return org.tensorflow.proto.util.StackFrameWithId.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.StackFrameWithId build() {
-      org.tensorflow.proto.util.StackFrameWithId result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.StackFrameWithId buildPartial() {
-      org.tensorflow.proto.util.StackFrameWithId result = new org.tensorflow.proto.util.StackFrameWithId(this);
-      result.id_ = id_;
-      if (fileLineColBuilder_ == null) {
-        result.fileLineCol_ = fileLineCol_;
-      } else {
-        result.fileLineCol_ = fileLineColBuilder_.build();
-      }
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.util.StackFrameWithId) {
-        return mergeFrom((org.tensorflow.proto.util.StackFrameWithId)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.util.StackFrameWithId other) {
-      if (other == org.tensorflow.proto.util.StackFrameWithId.getDefaultInstance()) return this;
-      if (!other.getId().isEmpty()) {
-        id_ = other.id_;
-        onChanged();
-      }
-      if (other.hasFileLineCol()) {
-        mergeFileLineCol(other.getFileLineCol());
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.util.StackFrameWithId parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.util.StackFrameWithId) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-
-    private java.lang.Object id_ = "";
-    /**
-     * <pre>
-     * A unique ID for the stack frame: A UUID-like string.
-     * </pre>
-     *
-     * <code>string id = 1;</code>
-     */
-    public java.lang.String getId() {
-      java.lang.Object ref = id_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        id_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * A unique ID for the stack frame: A UUID-like string.
-     * </pre>
-     *
-     * <code>string id = 1;</code>
-     */
-    public com.google.protobuf.ByteString
-        getIdBytes() {
-      java.lang.Object ref = id_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        id_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * A unique ID for the stack frame: A UUID-like string.
-     * </pre>
-     *
-     * <code>string id = 1;</code>
-     */
-    public Builder setId(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      id_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * A unique ID for the stack frame: A UUID-like string.
-     * </pre>
-     *
-     * <code>string id = 1;</code>
-     */
-    public Builder clearId() {
-      
-      id_ = getDefaultInstance().getId();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * A unique ID for the stack frame: A UUID-like string.
-     * </pre>
-     *
-     * <code>string id = 1;</code>
-     */
-    public Builder setIdBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      id_ = value;
-      onChanged();
-      return this;
-    }
-
-    private org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol fileLineCol_;
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol, org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol.Builder, org.tensorflow.proto.framework.GraphDebugInfo.FileLineColOrBuilder> fileLineColBuilder_;
-    /**
-     * <pre>
-     * Stack frame, i.e., a frame of a stack trace, containing information
-     * regarding the file name, line number, function name, code content
-     * of the line, and column number (if available).
-     * </pre>
-     *
-     * <code>.tensorflow.GraphDebugInfo.FileLineCol file_line_col = 2;</code>
-     */
-    public boolean hasFileLineCol() {
-      return fileLineColBuilder_ != null || fileLineCol_ != null;
-    }
-    /**
-     * <pre>
-     * Stack frame, i.e., a frame of a stack trace, containing information
-     * regarding the file name, line number, function name, code content
-     * of the line, and column number (if available).
-     * </pre>
-     *
-     * <code>.tensorflow.GraphDebugInfo.FileLineCol file_line_col = 2;</code>
-     */
-    public org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol getFileLineCol() {
-      if (fileLineColBuilder_ == null) {
-        return fileLineCol_ == null ? org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol.getDefaultInstance() : fileLineCol_;
-      } else {
-        return fileLineColBuilder_.getMessage();
-      }
-    }
-    /**
-     * <pre>
-     * Stack frame, i.e., a frame of a stack trace, containing information
-     * regarding the file name, line number, function name, code content
-     * of the line, and column number (if available).
-     * </pre>
-     *
-     * <code>.tensorflow.GraphDebugInfo.FileLineCol file_line_col = 2;</code>
-     */
-    public Builder setFileLineCol(org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol value) {
-      if (fileLineColBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        fileLineCol_ = value;
-        onChanged();
-      } else {
-        fileLineColBuilder_.setMessage(value);
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Stack frame, i.e., a frame of a stack trace, containing information
-     * regarding the file name, line number, function name, code content
-     * of the line, and column number (if available).
-     * </pre>
-     *
-     * <code>.tensorflow.GraphDebugInfo.FileLineCol file_line_col = 2;</code>
-     */
-    public Builder setFileLineCol(
-        org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol.Builder builderForValue) {
-      if (fileLineColBuilder_ == null) {
-        fileLineCol_ = builderForValue.build();
-        onChanged();
-      } else {
-        fileLineColBuilder_.setMessage(builderForValue.build());
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Stack frame, i.e., a frame of a stack trace, containing information
-     * regarding the file name, line number, function name, code content
-     * of the line, and column number (if available).
-     * </pre>
-     *
-     * <code>.tensorflow.GraphDebugInfo.FileLineCol file_line_col = 2;</code>
-     */
-    public Builder mergeFileLineCol(org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol value) {
-      if (fileLineColBuilder_ == null) {
-        if (fileLineCol_ != null) {
-          fileLineCol_ =
-            org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol.newBuilder(fileLineCol_).mergeFrom(value).buildPartial();
-        } else {
-          fileLineCol_ = value;
-        }
-        onChanged();
-      } else {
-        fileLineColBuilder_.mergeFrom(value);
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Stack frame, i.e., a frame of a stack trace, containing information
-     * regarding the file name, line number, function name, code content
-     * of the line, and column number (if available).
-     * </pre>
-     *
-     * <code>.tensorflow.GraphDebugInfo.FileLineCol file_line_col = 2;</code>
-     */
-    public Builder clearFileLineCol() {
-      if (fileLineColBuilder_ == null) {
-        fileLineCol_ = null;
-        onChanged();
-      } else {
-        fileLineCol_ = null;
-        fileLineColBuilder_ = null;
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Stack frame, i.e., a frame of a stack trace, containing information
-     * regarding the file name, line number, function name, code content
-     * of the line, and column number (if available).
-     * </pre>
-     *
-     * <code>.tensorflow.GraphDebugInfo.FileLineCol file_line_col = 2;</code>
-     */
-    public org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol.Builder getFileLineColBuilder() {
-      
-      onChanged();
-      return getFileLineColFieldBuilder().getBuilder();
-    }
-    /**
-     * <pre>
-     * Stack frame, i.e., a frame of a stack trace, containing information
-     * regarding the file name, line number, function name, code content
-     * of the line, and column number (if available).
-     * </pre>
-     *
-     * <code>.tensorflow.GraphDebugInfo.FileLineCol file_line_col = 2;</code>
-     */
-    public org.tensorflow.proto.framework.GraphDebugInfo.FileLineColOrBuilder getFileLineColOrBuilder() {
-      if (fileLineColBuilder_ != null) {
-        return fileLineColBuilder_.getMessageOrBuilder();
-      } else {
-        return fileLineCol_ == null ?
-            org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol.getDefaultInstance() : fileLineCol_;
-      }
-    }
-    /**
-     * <pre>
-     * Stack frame, i.e., a frame of a stack trace, containing information
-     * regarding the file name, line number, function name, code content
-     * of the line, and column number (if available).
-     * </pre>
-     *
-     * <code>.tensorflow.GraphDebugInfo.FileLineCol file_line_col = 2;</code>
-     */
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol, org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol.Builder, org.tensorflow.proto.framework.GraphDebugInfo.FileLineColOrBuilder> 
-        getFileLineColFieldBuilder() {
-      if (fileLineColBuilder_ == null) {
-        fileLineColBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-            org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol, org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol.Builder, org.tensorflow.proto.framework.GraphDebugInfo.FileLineColOrBuilder>(
-                getFileLineCol(),
-                getParentForChildren(),
-                isClean());
-        fileLineCol_ = null;
-      }
-      return fileLineColBuilder_;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.StackFrameWithId)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.StackFrameWithId)
-  private static final org.tensorflow.proto.util.StackFrameWithId DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.util.StackFrameWithId();
-  }
-
-  public static org.tensorflow.proto.util.StackFrameWithId getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<StackFrameWithId>
-      PARSER = new com.google.protobuf.AbstractParser<StackFrameWithId>() {
-    @java.lang.Override
-    public StackFrameWithId parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new StackFrameWithId(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<StackFrameWithId> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<StackFrameWithId> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.util.StackFrameWithId getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/StackFrameWithIdOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/StackFrameWithIdOrBuilder.java
deleted file mode 100644
index b1132ff9e72..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/StackFrameWithIdOrBuilder.java
+++ /dev/null
@@ -1,58 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/debug_event.proto
-
-package org.tensorflow.proto.util;
-
-public interface StackFrameWithIdOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.StackFrameWithId)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <pre>
-   * A unique ID for the stack frame: A UUID-like string.
-   * </pre>
-   *
-   * <code>string id = 1;</code>
-   */
-  java.lang.String getId();
-  /**
-   * <pre>
-   * A unique ID for the stack frame: A UUID-like string.
-   * </pre>
-   *
-   * <code>string id = 1;</code>
-   */
-  com.google.protobuf.ByteString
-      getIdBytes();
-
-  /**
-   * <pre>
-   * Stack frame, i.e., a frame of a stack trace, containing information
-   * regarding the file name, line number, function name, code content
-   * of the line, and column number (if available).
-   * </pre>
-   *
-   * <code>.tensorflow.GraphDebugInfo.FileLineCol file_line_col = 2;</code>
-   */
-  boolean hasFileLineCol();
-  /**
-   * <pre>
-   * Stack frame, i.e., a frame of a stack trace, containing information
-   * regarding the file name, line number, function name, code content
-   * of the line, and column number (if available).
-   * </pre>
-   *
-   * <code>.tensorflow.GraphDebugInfo.FileLineCol file_line_col = 2;</code>
-   */
-  org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol getFileLineCol();
-  /**
-   * <pre>
-   * Stack frame, i.e., a frame of a stack trace, containing information
-   * regarding the file name, line number, function name, code content
-   * of the line, and column number (if available).
-   * </pre>
-   *
-   * <code>.tensorflow.GraphDebugInfo.FileLineCol file_line_col = 2;</code>
-   */
-  org.tensorflow.proto.framework.GraphDebugInfo.FileLineColOrBuilder getFileLineColOrBuilder();
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/TaggedRunMetadata.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/TaggedRunMetadata.java
deleted file mode 100644
index 4880946d5df..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/TaggedRunMetadata.java
+++ /dev/null
@@ -1,663 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/util/event.proto
-
-package org.tensorflow.proto.util;
-
-/**
- * <pre>
- * For logging the metadata output for a single session.run() call.
- * </pre>
- *
- * Protobuf type {@code tensorflow.TaggedRunMetadata}
- */
-public  final class TaggedRunMetadata extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.TaggedRunMetadata)
-    TaggedRunMetadataOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use TaggedRunMetadata.newBuilder() to construct.
-  private TaggedRunMetadata(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private TaggedRunMetadata() {
-    tag_ = "";
-    runMetadata_ = com.google.protobuf.ByteString.EMPTY;
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new TaggedRunMetadata();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private TaggedRunMetadata(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 10: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            tag_ = s;
-            break;
-          }
-          case 18: {
-
-            runMetadata_ = input.readBytes();
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.util.EventProtos.internal_static_tensorflow_TaggedRunMetadata_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.util.EventProtos.internal_static_tensorflow_TaggedRunMetadata_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.util.TaggedRunMetadata.class, org.tensorflow.proto.util.TaggedRunMetadata.Builder.class);
-  }
-
-  public static final int TAG_FIELD_NUMBER = 1;
-  private volatile java.lang.Object tag_;
-  /**
-   * <pre>
-   * Tag name associated with this metadata.
-   * </pre>
-   *
-   * <code>string tag = 1;</code>
-   */
-  public java.lang.String getTag() {
-    java.lang.Object ref = tag_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      tag_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * Tag name associated with this metadata.
-   * </pre>
-   *
-   * <code>string tag = 1;</code>
-   */
-  public com.google.protobuf.ByteString
-      getTagBytes() {
-    java.lang.Object ref = tag_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      tag_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int RUN_METADATA_FIELD_NUMBER = 2;
-  private com.google.protobuf.ByteString runMetadata_;
-  /**
-   * <pre>
-   * Byte-encoded version of the `RunMetadata` proto in order to allow lazy
-   * deserialization.
-   * </pre>
-   *
-   * <code>bytes run_metadata = 2;</code>
-   */
-  public com.google.protobuf.ByteString getRunMetadata() {
-    return runMetadata_;
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    if (!getTagBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 1, tag_);
-    }
-    if (!runMetadata_.isEmpty()) {
-      output.writeBytes(2, runMetadata_);
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (!getTagBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, tag_);
-    }
-    if (!runMetadata_.isEmpty()) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeBytesSize(2, runMetadata_);
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.util.TaggedRunMetadata)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.util.TaggedRunMetadata other = (org.tensorflow.proto.util.TaggedRunMetadata) obj;
-
-    if (!getTag()
-        .equals(other.getTag())) return false;
-    if (!getRunMetadata()
-        .equals(other.getRunMetadata())) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    hash = (37 * hash) + TAG_FIELD_NUMBER;
-    hash = (53 * hash) + getTag().hashCode();
-    hash = (37 * hash) + RUN_METADATA_FIELD_NUMBER;
-    hash = (53 * hash) + getRunMetadata().hashCode();
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.util.TaggedRunMetadata parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.TaggedRunMetadata parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.TaggedRunMetadata parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.TaggedRunMetadata parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.TaggedRunMetadata parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.TaggedRunMetadata parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.TaggedRunMetadata parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.TaggedRunMetadata parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.TaggedRunMetadata parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.TaggedRunMetadata parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.TaggedRunMetadata parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.TaggedRunMetadata parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.util.TaggedRunMetadata prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * <pre>
-   * For logging the metadata output for a single session.run() call.
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.TaggedRunMetadata}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.TaggedRunMetadata)
-      org.tensorflow.proto.util.TaggedRunMetadataOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.util.EventProtos.internal_static_tensorflow_TaggedRunMetadata_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.util.EventProtos.internal_static_tensorflow_TaggedRunMetadata_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.util.TaggedRunMetadata.class, org.tensorflow.proto.util.TaggedRunMetadata.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.util.TaggedRunMetadata.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      tag_ = "";
-
-      runMetadata_ = com.google.protobuf.ByteString.EMPTY;
-
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.util.EventProtos.internal_static_tensorflow_TaggedRunMetadata_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.TaggedRunMetadata getDefaultInstanceForType() {
-      return org.tensorflow.proto.util.TaggedRunMetadata.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.TaggedRunMetadata build() {
-      org.tensorflow.proto.util.TaggedRunMetadata result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.TaggedRunMetadata buildPartial() {
-      org.tensorflow.proto.util.TaggedRunMetadata result = new org.tensorflow.proto.util.TaggedRunMetadata(this);
-      result.tag_ = tag_;
-      result.runMetadata_ = runMetadata_;
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.util.TaggedRunMetadata) {
-        return mergeFrom((org.tensorflow.proto.util.TaggedRunMetadata)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.util.TaggedRunMetadata other) {
-      if (other == org.tensorflow.proto.util.TaggedRunMetadata.getDefaultInstance()) return this;
-      if (!other.getTag().isEmpty()) {
-        tag_ = other.tag_;
-        onChanged();
-      }
-      if (other.getRunMetadata() != com.google.protobuf.ByteString.EMPTY) {
-        setRunMetadata(other.getRunMetadata());
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.util.TaggedRunMetadata parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.util.TaggedRunMetadata) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-
-    private java.lang.Object tag_ = "";
-    /**
-     * <pre>
-     * Tag name associated with this metadata.
-     * </pre>
-     *
-     * <code>string tag = 1;</code>
-     */
-    public java.lang.String getTag() {
-      java.lang.Object ref = tag_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        tag_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Tag name associated with this metadata.
-     * </pre>
-     *
-     * <code>string tag = 1;</code>
-     */
-    public com.google.protobuf.ByteString
-        getTagBytes() {
-      java.lang.Object ref = tag_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        tag_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Tag name associated with this metadata.
-     * </pre>
-     *
-     * <code>string tag = 1;</code>
-     */
-    public Builder setTag(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      tag_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Tag name associated with this metadata.
-     * </pre>
-     *
-     * <code>string tag = 1;</code>
-     */
-    public Builder clearTag() {
-      
-      tag_ = getDefaultInstance().getTag();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Tag name associated with this metadata.
-     * </pre>
-     *
-     * <code>string tag = 1;</code>
-     */
-    public Builder setTagBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      tag_ = value;
-      onChanged();
-      return this;
-    }
-
-    private com.google.protobuf.ByteString runMetadata_ = com.google.protobuf.ByteString.EMPTY;
-    /**
-     * <pre>
-     * Byte-encoded version of the `RunMetadata` proto in order to allow lazy
-     * deserialization.
-     * </pre>
-     *
-     * <code>bytes run_metadata = 2;</code>
-     */
-    public com.google.protobuf.ByteString getRunMetadata() {
-      return runMetadata_;
-    }
-    /**
-     * <pre>
-     * Byte-encoded version of the `RunMetadata` proto in order to allow lazy
-     * deserialization.
-     * </pre>
-     *
-     * <code>bytes run_metadata = 2;</code>
-     */
-    public Builder setRunMetadata(com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      runMetadata_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Byte-encoded version of the `RunMetadata` proto in order to allow lazy
-     * deserialization.
-     * </pre>
-     *
-     * <code>bytes run_metadata = 2;</code>
-     */
-    public Builder clearRunMetadata() {
-      
-      runMetadata_ = getDefaultInstance().getRunMetadata();
-      onChanged();
-      return this;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.TaggedRunMetadata)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.TaggedRunMetadata)
-  private static final org.tensorflow.proto.util.TaggedRunMetadata DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.util.TaggedRunMetadata();
-  }
-
-  public static org.tensorflow.proto.util.TaggedRunMetadata getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<TaggedRunMetadata>
-      PARSER = new com.google.protobuf.AbstractParser<TaggedRunMetadata>() {
-    @java.lang.Override
-    public TaggedRunMetadata parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new TaggedRunMetadata(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<TaggedRunMetadata> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<TaggedRunMetadata> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.util.TaggedRunMetadata getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/TaggedRunMetadataOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/TaggedRunMetadataOrBuilder.java
deleted file mode 100644
index d364543215c..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/TaggedRunMetadataOrBuilder.java
+++ /dev/null
@@ -1,37 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/util/event.proto
-
-package org.tensorflow.proto.util;
-
-public interface TaggedRunMetadataOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.TaggedRunMetadata)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <pre>
-   * Tag name associated with this metadata.
-   * </pre>
-   *
-   * <code>string tag = 1;</code>
-   */
-  java.lang.String getTag();
-  /**
-   * <pre>
-   * Tag name associated with this metadata.
-   * </pre>
-   *
-   * <code>string tag = 1;</code>
-   */
-  com.google.protobuf.ByteString
-      getTagBytes();
-
-  /**
-   * <pre>
-   * Byte-encoded version of the `RunMetadata` proto in order to allow lazy
-   * deserialization.
-   * </pre>
-   *
-   * <code>bytes run_metadata = 2;</code>
-   */
-  com.google.protobuf.ByteString getRunMetadata();
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/TensorBundleProtos.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/TensorBundleProtos.java
deleted file mode 100644
index 3de209fb5a9..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/TensorBundleProtos.java
+++ /dev/null
@@ -1,83 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/tensor_bundle.proto
-
-package org.tensorflow.proto.util;
-
-public final class TensorBundleProtos {
-  private TensorBundleProtos() {}
-  public static void registerAllExtensions(
-      com.google.protobuf.ExtensionRegistryLite registry) {
-  }
-
-  public static void registerAllExtensions(
-      com.google.protobuf.ExtensionRegistry registry) {
-    registerAllExtensions(
-        (com.google.protobuf.ExtensionRegistryLite) registry);
-  }
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_BundleHeaderProto_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_BundleHeaderProto_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_BundleEntryProto_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_BundleEntryProto_fieldAccessorTable;
-
-  public static com.google.protobuf.Descriptors.FileDescriptor
-      getDescriptor() {
-    return descriptor;
-  }
-  private static  com.google.protobuf.Descriptors.FileDescriptor
-      descriptor;
-  static {
-    java.lang.String[] descriptorData = {
-      "\n,tensorflow/core/protobuf/tensor_bundle" +
-      ".proto\022\ntensorflow\032,tensorflow/core/fram" +
-      "ework/tensor_shape.proto\032,tensorflow/cor" +
-      "e/framework/tensor_slice.proto\032%tensorfl" +
-      "ow/core/framework/types.proto\032(tensorflo" +
-      "w/core/framework/versions.proto\"\261\001\n\021Bund" +
-      "leHeaderProto\022\022\n\nnum_shards\030\001 \001(\005\022<\n\nend" +
-      "ianness\030\002 \001(\0162(.tensorflow.BundleHeaderP" +
-      "roto.Endianness\022\'\n\007version\030\003 \001(\0132\026.tenso" +
-      "rflow.VersionDef\"!\n\nEndianness\022\n\n\006LITTLE" +
-      "\020\000\022\007\n\003BIG\020\001\"\322\001\n\020BundleEntryProto\022#\n\005dtyp" +
-      "e\030\001 \001(\0162\024.tensorflow.DataType\022+\n\005shape\030\002" +
-      " \001(\0132\034.tensorflow.TensorShapeProto\022\020\n\010sh" +
-      "ard_id\030\003 \001(\005\022\016\n\006offset\030\004 \001(\003\022\014\n\004size\030\005 \001" +
-      "(\003\022\016\n\006crc32c\030\006 \001(\007\022,\n\006slices\030\007 \003(\0132\034.ten" +
-      "sorflow.TensorSliceProtoB~\n\031org.tensorfl" +
-      "ow.proto.utilB\022TensorBundleProtosP\001ZHgit" +
-      "hub.com/tensorflow/tensorflow/tensorflow" +
-      "/go/core/core_protos_go_proto\370\001\001b\006proto3"
-    };
-    descriptor = com.google.protobuf.Descriptors.FileDescriptor
-      .internalBuildGeneratedFileFrom(descriptorData,
-        new com.google.protobuf.Descriptors.FileDescriptor[] {
-          org.tensorflow.proto.framework.TensorShapeProtos.getDescriptor(),
-          org.tensorflow.proto.framework.TensorSliceProtos.getDescriptor(),
-          org.tensorflow.proto.framework.TypesProtos.getDescriptor(),
-          org.tensorflow.proto.framework.VersionsProtos.getDescriptor(),
-        });
-    internal_static_tensorflow_BundleHeaderProto_descriptor =
-      getDescriptor().getMessageTypes().get(0);
-    internal_static_tensorflow_BundleHeaderProto_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_BundleHeaderProto_descriptor,
-        new java.lang.String[] { "NumShards", "Endianness", "Version", });
-    internal_static_tensorflow_BundleEntryProto_descriptor =
-      getDescriptor().getMessageTypes().get(1);
-    internal_static_tensorflow_BundleEntryProto_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_BundleEntryProto_descriptor,
-        new java.lang.String[] { "Dtype", "Shape", "ShardId", "Offset", "Size", "Crc32C", "Slices", });
-    org.tensorflow.proto.framework.TensorShapeProtos.getDescriptor();
-    org.tensorflow.proto.framework.TensorSliceProtos.getDescriptor();
-    org.tensorflow.proto.framework.TypesProtos.getDescriptor();
-    org.tensorflow.proto.framework.VersionsProtos.getDescriptor();
-  }
-
-  // @@protoc_insertion_point(outer_class_scope)
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/TensorDebugMode.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/TensorDebugMode.java
deleted file mode 100644
index 074b1ca74c5..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/TensorDebugMode.java
+++ /dev/null
@@ -1,282 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/debug_event.proto
-
-package org.tensorflow.proto.util;
-
-/**
- * <pre>
- * Available modes for extracting debugging information from a Tensor.
- * TODO(cais): Document the detailed column names and semantics in a separate
- * markdown file once the implementation settles.
- * </pre>
- *
- * Protobuf enum {@code tensorflow.TensorDebugMode}
- */
-public enum TensorDebugMode
-    implements com.google.protobuf.ProtocolMessageEnum {
-  /**
-   * <code>UNSPECIFIED = 0;</code>
-   */
-  UNSPECIFIED(0),
-  /**
-   * <pre>
-   * Only records what tensors are computed, eagerly or in graphs.
-   * No information regarding the value of the tensor is available.
-   * </pre>
-   *
-   * <code>NO_TENSOR = 1;</code>
-   */
-  NO_TENSOR(1),
-  /**
-   * <pre>
-   * A minimalist health summary for float-type tensors.
-   * Contains information only about the presence/absence of pathological
-   * values including Infinity and NaN.
-   * Applicable only to float dtypes.
-   * </pre>
-   *
-   * <code>CURT_HEALTH = 2;</code>
-   */
-  CURT_HEALTH(2),
-  /**
-   * <pre>
-   * A concise health summary for float-type tensors.
-   * Contains more information that CURT_HEALTH.
-   * Infinity and NaN are treated differently.
-   * Applicable only to float and integer dtypes.
-   * </pre>
-   *
-   * <code>CONCISE_HEALTH = 3;</code>
-   */
-  CONCISE_HEALTH(3),
-  /**
-   * <pre>
-   * A detailed health summary.
-   * Contains further detailed information than `CONCISE_HEALTH`.
-   * Information about device, dtype and shape are included.
-   * Counts for various types of values (Infinity, NaN, negative, zero,
-   * positive) are included.
-   * Applicable to float, integer and boolean dtypes.
-   * </pre>
-   *
-   * <code>FULL_HEALTH = 4;</code>
-   */
-  FULL_HEALTH(4),
-  /**
-   * <pre>
-   * Provides full runtime shape information, up to a maximum rank, beyond
-   * which the dimension sizes are truncated.
-   * </pre>
-   *
-   * <code>SHAPE = 5;</code>
-   */
-  SHAPE(5),
-  /**
-   * <pre>
-   * Full numeric summary.
-   * Including device, dtype, shape, counts of various types of values
-   * (Infinity, NaN, negative, zero, positive), and summary statistics
-   * (minimum, maximum, mean and variance).
-   * Applicable to float, integer and boolean dtypes.
-   * </pre>
-   *
-   * <code>FULL_NUMERICS = 6;</code>
-   */
-  FULL_NUMERICS(6),
-  /**
-   * <pre>
-   * Full tensor value.
-   * </pre>
-   *
-   * <code>FULL_TENSOR = 7;</code>
-   */
-  FULL_TENSOR(7),
-  /**
-   * <pre>
-   * Reduce the elements of a tensor to a rank-1 tensor of shape [3], in which
-   * - the 1st element is -inf if any element of the tensor is -inf,
-   *   or zero otherwise.
-   * - the 2nd element is +inf if any element of the tensor is +inf,
-   *   or zero otherwise.
-   * - the 3rd element is nan if any element of the tensor is nan, or zero
-   *   otherwise.
-   * </pre>
-   *
-   * <code>REDUCE_INF_NAN_THREE_SLOTS = 8;</code>
-   */
-  REDUCE_INF_NAN_THREE_SLOTS(8),
-  UNRECOGNIZED(-1),
-  ;
-
-  /**
-   * <code>UNSPECIFIED = 0;</code>
-   */
-  public static final int UNSPECIFIED_VALUE = 0;
-  /**
-   * <pre>
-   * Only records what tensors are computed, eagerly or in graphs.
-   * No information regarding the value of the tensor is available.
-   * </pre>
-   *
-   * <code>NO_TENSOR = 1;</code>
-   */
-  public static final int NO_TENSOR_VALUE = 1;
-  /**
-   * <pre>
-   * A minimalist health summary for float-type tensors.
-   * Contains information only about the presence/absence of pathological
-   * values including Infinity and NaN.
-   * Applicable only to float dtypes.
-   * </pre>
-   *
-   * <code>CURT_HEALTH = 2;</code>
-   */
-  public static final int CURT_HEALTH_VALUE = 2;
-  /**
-   * <pre>
-   * A concise health summary for float-type tensors.
-   * Contains more information that CURT_HEALTH.
-   * Infinity and NaN are treated differently.
-   * Applicable only to float and integer dtypes.
-   * </pre>
-   *
-   * <code>CONCISE_HEALTH = 3;</code>
-   */
-  public static final int CONCISE_HEALTH_VALUE = 3;
-  /**
-   * <pre>
-   * A detailed health summary.
-   * Contains further detailed information than `CONCISE_HEALTH`.
-   * Information about device, dtype and shape are included.
-   * Counts for various types of values (Infinity, NaN, negative, zero,
-   * positive) are included.
-   * Applicable to float, integer and boolean dtypes.
-   * </pre>
-   *
-   * <code>FULL_HEALTH = 4;</code>
-   */
-  public static final int FULL_HEALTH_VALUE = 4;
-  /**
-   * <pre>
-   * Provides full runtime shape information, up to a maximum rank, beyond
-   * which the dimension sizes are truncated.
-   * </pre>
-   *
-   * <code>SHAPE = 5;</code>
-   */
-  public static final int SHAPE_VALUE = 5;
-  /**
-   * <pre>
-   * Full numeric summary.
-   * Including device, dtype, shape, counts of various types of values
-   * (Infinity, NaN, negative, zero, positive), and summary statistics
-   * (minimum, maximum, mean and variance).
-   * Applicable to float, integer and boolean dtypes.
-   * </pre>
-   *
-   * <code>FULL_NUMERICS = 6;</code>
-   */
-  public static final int FULL_NUMERICS_VALUE = 6;
-  /**
-   * <pre>
-   * Full tensor value.
-   * </pre>
-   *
-   * <code>FULL_TENSOR = 7;</code>
-   */
-  public static final int FULL_TENSOR_VALUE = 7;
-  /**
-   * <pre>
-   * Reduce the elements of a tensor to a rank-1 tensor of shape [3], in which
-   * - the 1st element is -inf if any element of the tensor is -inf,
-   *   or zero otherwise.
-   * - the 2nd element is +inf if any element of the tensor is +inf,
-   *   or zero otherwise.
-   * - the 3rd element is nan if any element of the tensor is nan, or zero
-   *   otherwise.
-   * </pre>
-   *
-   * <code>REDUCE_INF_NAN_THREE_SLOTS = 8;</code>
-   */
-  public static final int REDUCE_INF_NAN_THREE_SLOTS_VALUE = 8;
-
-
-  public final int getNumber() {
-    if (this == UNRECOGNIZED) {
-      throw new java.lang.IllegalArgumentException(
-          "Can't get the number of an unknown enum value.");
-    }
-    return value;
-  }
-
-  /**
-   * @deprecated Use {@link #forNumber(int)} instead.
-   */
-  @java.lang.Deprecated
-  public static TensorDebugMode valueOf(int value) {
-    return forNumber(value);
-  }
-
-  public static TensorDebugMode forNumber(int value) {
-    switch (value) {
-      case 0: return UNSPECIFIED;
-      case 1: return NO_TENSOR;
-      case 2: return CURT_HEALTH;
-      case 3: return CONCISE_HEALTH;
-      case 4: return FULL_HEALTH;
-      case 5: return SHAPE;
-      case 6: return FULL_NUMERICS;
-      case 7: return FULL_TENSOR;
-      case 8: return REDUCE_INF_NAN_THREE_SLOTS;
-      default: return null;
-    }
-  }
-
-  public static com.google.protobuf.Internal.EnumLiteMap<TensorDebugMode>
-      internalGetValueMap() {
-    return internalValueMap;
-  }
-  private static final com.google.protobuf.Internal.EnumLiteMap<
-      TensorDebugMode> internalValueMap =
-        new com.google.protobuf.Internal.EnumLiteMap<TensorDebugMode>() {
-          public TensorDebugMode findValueByNumber(int number) {
-            return TensorDebugMode.forNumber(number);
-          }
-        };
-
-  public final com.google.protobuf.Descriptors.EnumValueDescriptor
-      getValueDescriptor() {
-    return getDescriptor().getValues().get(ordinal());
-  }
-  public final com.google.protobuf.Descriptors.EnumDescriptor
-      getDescriptorForType() {
-    return getDescriptor();
-  }
-  public static final com.google.protobuf.Descriptors.EnumDescriptor
-      getDescriptor() {
-    return org.tensorflow.proto.util.DebugEventProtos.getDescriptor().getEnumTypes().get(0);
-  }
-
-  private static final TensorDebugMode[] VALUES = values();
-
-  public static TensorDebugMode valueOf(
-      com.google.protobuf.Descriptors.EnumValueDescriptor desc) {
-    if (desc.getType() != getDescriptor()) {
-      throw new java.lang.IllegalArgumentException(
-        "EnumValueDescriptor is not for this type.");
-    }
-    if (desc.getIndex() == -1) {
-      return UNRECOGNIZED;
-    }
-    return VALUES[desc.getIndex()];
-  }
-
-  private final int value;
-
-  private TensorDebugMode(int value) {
-    this.value = value;
-  }
-
-  // @@protoc_insertion_point(enum_scope:tensorflow.TensorDebugMode)
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/TensorMetadata.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/TensorMetadata.java
deleted file mode 100644
index c1ac5388ed4..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/TensorMetadata.java
+++ /dev/null
@@ -1,682 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/data/experimental/snapshot.proto
-
-package org.tensorflow.proto.util;
-
-/**
- * <pre>
- * Metadata for a single tensor in the Snapshot Record.
- * </pre>
- *
- * Protobuf type {@code tensorflow.data.experimental.TensorMetadata}
- */
-public  final class TensorMetadata extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.data.experimental.TensorMetadata)
-    TensorMetadataOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use TensorMetadata.newBuilder() to construct.
-  private TensorMetadata(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private TensorMetadata() {
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new TensorMetadata();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private TensorMetadata(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 18: {
-            org.tensorflow.proto.framework.TensorShapeProto.Builder subBuilder = null;
-            if (tensorShape_ != null) {
-              subBuilder = tensorShape_.toBuilder();
-            }
-            tensorShape_ = input.readMessage(org.tensorflow.proto.framework.TensorShapeProto.parser(), extensionRegistry);
-            if (subBuilder != null) {
-              subBuilder.mergeFrom(tensorShape_);
-              tensorShape_ = subBuilder.buildPartial();
-            }
-
-            break;
-          }
-          case 24: {
-
-            tensorSizeBytes_ = input.readInt64();
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.util.SnapshotProtos.internal_static_tensorflow_data_experimental_TensorMetadata_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.util.SnapshotProtos.internal_static_tensorflow_data_experimental_TensorMetadata_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.util.TensorMetadata.class, org.tensorflow.proto.util.TensorMetadata.Builder.class);
-  }
-
-  public static final int TENSOR_SHAPE_FIELD_NUMBER = 2;
-  private org.tensorflow.proto.framework.TensorShapeProto tensorShape_;
-  /**
-   * <code>.tensorflow.TensorShapeProto tensor_shape = 2;</code>
-   */
-  public boolean hasTensorShape() {
-    return tensorShape_ != null;
-  }
-  /**
-   * <code>.tensorflow.TensorShapeProto tensor_shape = 2;</code>
-   */
-  public org.tensorflow.proto.framework.TensorShapeProto getTensorShape() {
-    return tensorShape_ == null ? org.tensorflow.proto.framework.TensorShapeProto.getDefaultInstance() : tensorShape_;
-  }
-  /**
-   * <code>.tensorflow.TensorShapeProto tensor_shape = 2;</code>
-   */
-  public org.tensorflow.proto.framework.TensorShapeProtoOrBuilder getTensorShapeOrBuilder() {
-    return getTensorShape();
-  }
-
-  public static final int TENSOR_SIZE_BYTES_FIELD_NUMBER = 3;
-  private long tensorSizeBytes_;
-  /**
-   * <pre>
-   * Number of uncompressed bytes used to store the tensor representation.
-   * </pre>
-   *
-   * <code>int64 tensor_size_bytes = 3;</code>
-   */
-  public long getTensorSizeBytes() {
-    return tensorSizeBytes_;
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    if (tensorShape_ != null) {
-      output.writeMessage(2, getTensorShape());
-    }
-    if (tensorSizeBytes_ != 0L) {
-      output.writeInt64(3, tensorSizeBytes_);
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (tensorShape_ != null) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(2, getTensorShape());
-    }
-    if (tensorSizeBytes_ != 0L) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt64Size(3, tensorSizeBytes_);
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.util.TensorMetadata)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.util.TensorMetadata other = (org.tensorflow.proto.util.TensorMetadata) obj;
-
-    if (hasTensorShape() != other.hasTensorShape()) return false;
-    if (hasTensorShape()) {
-      if (!getTensorShape()
-          .equals(other.getTensorShape())) return false;
-    }
-    if (getTensorSizeBytes()
-        != other.getTensorSizeBytes()) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    if (hasTensorShape()) {
-      hash = (37 * hash) + TENSOR_SHAPE_FIELD_NUMBER;
-      hash = (53 * hash) + getTensorShape().hashCode();
-    }
-    hash = (37 * hash) + TENSOR_SIZE_BYTES_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-        getTensorSizeBytes());
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.util.TensorMetadata parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.TensorMetadata parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.TensorMetadata parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.TensorMetadata parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.TensorMetadata parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.TensorMetadata parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.TensorMetadata parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.TensorMetadata parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.TensorMetadata parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.TensorMetadata parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.TensorMetadata parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.TensorMetadata parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.util.TensorMetadata prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * <pre>
-   * Metadata for a single tensor in the Snapshot Record.
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.data.experimental.TensorMetadata}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.data.experimental.TensorMetadata)
-      org.tensorflow.proto.util.TensorMetadataOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.util.SnapshotProtos.internal_static_tensorflow_data_experimental_TensorMetadata_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.util.SnapshotProtos.internal_static_tensorflow_data_experimental_TensorMetadata_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.util.TensorMetadata.class, org.tensorflow.proto.util.TensorMetadata.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.util.TensorMetadata.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      if (tensorShapeBuilder_ == null) {
-        tensorShape_ = null;
-      } else {
-        tensorShape_ = null;
-        tensorShapeBuilder_ = null;
-      }
-      tensorSizeBytes_ = 0L;
-
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.util.SnapshotProtos.internal_static_tensorflow_data_experimental_TensorMetadata_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.TensorMetadata getDefaultInstanceForType() {
-      return org.tensorflow.proto.util.TensorMetadata.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.TensorMetadata build() {
-      org.tensorflow.proto.util.TensorMetadata result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.TensorMetadata buildPartial() {
-      org.tensorflow.proto.util.TensorMetadata result = new org.tensorflow.proto.util.TensorMetadata(this);
-      if (tensorShapeBuilder_ == null) {
-        result.tensorShape_ = tensorShape_;
-      } else {
-        result.tensorShape_ = tensorShapeBuilder_.build();
-      }
-      result.tensorSizeBytes_ = tensorSizeBytes_;
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.util.TensorMetadata) {
-        return mergeFrom((org.tensorflow.proto.util.TensorMetadata)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.util.TensorMetadata other) {
-      if (other == org.tensorflow.proto.util.TensorMetadata.getDefaultInstance()) return this;
-      if (other.hasTensorShape()) {
-        mergeTensorShape(other.getTensorShape());
-      }
-      if (other.getTensorSizeBytes() != 0L) {
-        setTensorSizeBytes(other.getTensorSizeBytes());
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.util.TensorMetadata parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.util.TensorMetadata) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-
-    private org.tensorflow.proto.framework.TensorShapeProto tensorShape_;
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.TensorShapeProto, org.tensorflow.proto.framework.TensorShapeProto.Builder, org.tensorflow.proto.framework.TensorShapeProtoOrBuilder> tensorShapeBuilder_;
-    /**
-     * <code>.tensorflow.TensorShapeProto tensor_shape = 2;</code>
-     */
-    public boolean hasTensorShape() {
-      return tensorShapeBuilder_ != null || tensorShape_ != null;
-    }
-    /**
-     * <code>.tensorflow.TensorShapeProto tensor_shape = 2;</code>
-     */
-    public org.tensorflow.proto.framework.TensorShapeProto getTensorShape() {
-      if (tensorShapeBuilder_ == null) {
-        return tensorShape_ == null ? org.tensorflow.proto.framework.TensorShapeProto.getDefaultInstance() : tensorShape_;
-      } else {
-        return tensorShapeBuilder_.getMessage();
-      }
-    }
-    /**
-     * <code>.tensorflow.TensorShapeProto tensor_shape = 2;</code>
-     */
-    public Builder setTensorShape(org.tensorflow.proto.framework.TensorShapeProto value) {
-      if (tensorShapeBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        tensorShape_ = value;
-        onChanged();
-      } else {
-        tensorShapeBuilder_.setMessage(value);
-      }
-
-      return this;
-    }
-    /**
-     * <code>.tensorflow.TensorShapeProto tensor_shape = 2;</code>
-     */
-    public Builder setTensorShape(
-        org.tensorflow.proto.framework.TensorShapeProto.Builder builderForValue) {
-      if (tensorShapeBuilder_ == null) {
-        tensorShape_ = builderForValue.build();
-        onChanged();
-      } else {
-        tensorShapeBuilder_.setMessage(builderForValue.build());
-      }
-
-      return this;
-    }
-    /**
-     * <code>.tensorflow.TensorShapeProto tensor_shape = 2;</code>
-     */
-    public Builder mergeTensorShape(org.tensorflow.proto.framework.TensorShapeProto value) {
-      if (tensorShapeBuilder_ == null) {
-        if (tensorShape_ != null) {
-          tensorShape_ =
-            org.tensorflow.proto.framework.TensorShapeProto.newBuilder(tensorShape_).mergeFrom(value).buildPartial();
-        } else {
-          tensorShape_ = value;
-        }
-        onChanged();
-      } else {
-        tensorShapeBuilder_.mergeFrom(value);
-      }
-
-      return this;
-    }
-    /**
-     * <code>.tensorflow.TensorShapeProto tensor_shape = 2;</code>
-     */
-    public Builder clearTensorShape() {
-      if (tensorShapeBuilder_ == null) {
-        tensorShape_ = null;
-        onChanged();
-      } else {
-        tensorShape_ = null;
-        tensorShapeBuilder_ = null;
-      }
-
-      return this;
-    }
-    /**
-     * <code>.tensorflow.TensorShapeProto tensor_shape = 2;</code>
-     */
-    public org.tensorflow.proto.framework.TensorShapeProto.Builder getTensorShapeBuilder() {
-      
-      onChanged();
-      return getTensorShapeFieldBuilder().getBuilder();
-    }
-    /**
-     * <code>.tensorflow.TensorShapeProto tensor_shape = 2;</code>
-     */
-    public org.tensorflow.proto.framework.TensorShapeProtoOrBuilder getTensorShapeOrBuilder() {
-      if (tensorShapeBuilder_ != null) {
-        return tensorShapeBuilder_.getMessageOrBuilder();
-      } else {
-        return tensorShape_ == null ?
-            org.tensorflow.proto.framework.TensorShapeProto.getDefaultInstance() : tensorShape_;
-      }
-    }
-    /**
-     * <code>.tensorflow.TensorShapeProto tensor_shape = 2;</code>
-     */
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.framework.TensorShapeProto, org.tensorflow.proto.framework.TensorShapeProto.Builder, org.tensorflow.proto.framework.TensorShapeProtoOrBuilder> 
-        getTensorShapeFieldBuilder() {
-      if (tensorShapeBuilder_ == null) {
-        tensorShapeBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-            org.tensorflow.proto.framework.TensorShapeProto, org.tensorflow.proto.framework.TensorShapeProto.Builder, org.tensorflow.proto.framework.TensorShapeProtoOrBuilder>(
-                getTensorShape(),
-                getParentForChildren(),
-                isClean());
-        tensorShape_ = null;
-      }
-      return tensorShapeBuilder_;
-    }
-
-    private long tensorSizeBytes_ ;
-    /**
-     * <pre>
-     * Number of uncompressed bytes used to store the tensor representation.
-     * </pre>
-     *
-     * <code>int64 tensor_size_bytes = 3;</code>
-     */
-    public long getTensorSizeBytes() {
-      return tensorSizeBytes_;
-    }
-    /**
-     * <pre>
-     * Number of uncompressed bytes used to store the tensor representation.
-     * </pre>
-     *
-     * <code>int64 tensor_size_bytes = 3;</code>
-     */
-    public Builder setTensorSizeBytes(long value) {
-      
-      tensorSizeBytes_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Number of uncompressed bytes used to store the tensor representation.
-     * </pre>
-     *
-     * <code>int64 tensor_size_bytes = 3;</code>
-     */
-    public Builder clearTensorSizeBytes() {
-      
-      tensorSizeBytes_ = 0L;
-      onChanged();
-      return this;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.data.experimental.TensorMetadata)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.data.experimental.TensorMetadata)
-  private static final org.tensorflow.proto.util.TensorMetadata DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.util.TensorMetadata();
-  }
-
-  public static org.tensorflow.proto.util.TensorMetadata getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<TensorMetadata>
-      PARSER = new com.google.protobuf.AbstractParser<TensorMetadata>() {
-    @java.lang.Override
-    public TensorMetadata parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new TensorMetadata(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<TensorMetadata> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<TensorMetadata> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.util.TensorMetadata getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/TensorMetadataOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/TensorMetadataOrBuilder.java
deleted file mode 100644
index 0cfa470c9e6..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/TensorMetadataOrBuilder.java
+++ /dev/null
@@ -1,31 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/protobuf/data/experimental/snapshot.proto
-
-package org.tensorflow.proto.util;
-
-public interface TensorMetadataOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.data.experimental.TensorMetadata)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <code>.tensorflow.TensorShapeProto tensor_shape = 2;</code>
-   */
-  boolean hasTensorShape();
-  /**
-   * <code>.tensorflow.TensorShapeProto tensor_shape = 2;</code>
-   */
-  org.tensorflow.proto.framework.TensorShapeProto getTensorShape();
-  /**
-   * <code>.tensorflow.TensorShapeProto tensor_shape = 2;</code>
-   */
-  org.tensorflow.proto.framework.TensorShapeProtoOrBuilder getTensorShapeOrBuilder();
-
-  /**
-   * <pre>
-   * Number of uncompressed bytes used to store the tensor representation.
-   * </pre>
-   *
-   * <code>int64 tensor_size_bytes = 3;</code>
-   */
-  long getTensorSizeBytes();
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/WatchdogConfig.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/WatchdogConfig.java
deleted file mode 100644
index f818bbca13f..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/WatchdogConfig.java
+++ /dev/null
@@ -1,477 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/util/event.proto
-
-package org.tensorflow.proto.util;
-
-/**
- * Protobuf type {@code tensorflow.WatchdogConfig}
- */
-public  final class WatchdogConfig extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.WatchdogConfig)
-    WatchdogConfigOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use WatchdogConfig.newBuilder() to construct.
-  private WatchdogConfig(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private WatchdogConfig() {
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new WatchdogConfig();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private WatchdogConfig(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 8: {
-
-            timeoutMs_ = input.readInt64();
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.util.EventProtos.internal_static_tensorflow_WatchdogConfig_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.util.EventProtos.internal_static_tensorflow_WatchdogConfig_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.util.WatchdogConfig.class, org.tensorflow.proto.util.WatchdogConfig.Builder.class);
-  }
-
-  public static final int TIMEOUT_MS_FIELD_NUMBER = 1;
-  private long timeoutMs_;
-  /**
-   * <code>int64 timeout_ms = 1;</code>
-   */
-  public long getTimeoutMs() {
-    return timeoutMs_;
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    if (timeoutMs_ != 0L) {
-      output.writeInt64(1, timeoutMs_);
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (timeoutMs_ != 0L) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt64Size(1, timeoutMs_);
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.util.WatchdogConfig)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.util.WatchdogConfig other = (org.tensorflow.proto.util.WatchdogConfig) obj;
-
-    if (getTimeoutMs()
-        != other.getTimeoutMs()) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    hash = (37 * hash) + TIMEOUT_MS_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-        getTimeoutMs());
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.util.WatchdogConfig parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.WatchdogConfig parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.WatchdogConfig parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.WatchdogConfig parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.WatchdogConfig parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.WatchdogConfig parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.WatchdogConfig parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.WatchdogConfig parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.WatchdogConfig parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.WatchdogConfig parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.WatchdogConfig parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.WatchdogConfig parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.util.WatchdogConfig prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * Protobuf type {@code tensorflow.WatchdogConfig}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.WatchdogConfig)
-      org.tensorflow.proto.util.WatchdogConfigOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.util.EventProtos.internal_static_tensorflow_WatchdogConfig_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.util.EventProtos.internal_static_tensorflow_WatchdogConfig_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.util.WatchdogConfig.class, org.tensorflow.proto.util.WatchdogConfig.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.util.WatchdogConfig.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      timeoutMs_ = 0L;
-
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.util.EventProtos.internal_static_tensorflow_WatchdogConfig_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.WatchdogConfig getDefaultInstanceForType() {
-      return org.tensorflow.proto.util.WatchdogConfig.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.WatchdogConfig build() {
-      org.tensorflow.proto.util.WatchdogConfig result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.WatchdogConfig buildPartial() {
-      org.tensorflow.proto.util.WatchdogConfig result = new org.tensorflow.proto.util.WatchdogConfig(this);
-      result.timeoutMs_ = timeoutMs_;
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.util.WatchdogConfig) {
-        return mergeFrom((org.tensorflow.proto.util.WatchdogConfig)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.util.WatchdogConfig other) {
-      if (other == org.tensorflow.proto.util.WatchdogConfig.getDefaultInstance()) return this;
-      if (other.getTimeoutMs() != 0L) {
-        setTimeoutMs(other.getTimeoutMs());
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.util.WatchdogConfig parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.util.WatchdogConfig) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-
-    private long timeoutMs_ ;
-    /**
-     * <code>int64 timeout_ms = 1;</code>
-     */
-    public long getTimeoutMs() {
-      return timeoutMs_;
-    }
-    /**
-     * <code>int64 timeout_ms = 1;</code>
-     */
-    public Builder setTimeoutMs(long value) {
-      
-      timeoutMs_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>int64 timeout_ms = 1;</code>
-     */
-    public Builder clearTimeoutMs() {
-      
-      timeoutMs_ = 0L;
-      onChanged();
-      return this;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.WatchdogConfig)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.WatchdogConfig)
-  private static final org.tensorflow.proto.util.WatchdogConfig DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.util.WatchdogConfig();
-  }
-
-  public static org.tensorflow.proto.util.WatchdogConfig getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<WatchdogConfig>
-      PARSER = new com.google.protobuf.AbstractParser<WatchdogConfig>() {
-    @java.lang.Override
-    public WatchdogConfig parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new WatchdogConfig(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<WatchdogConfig> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<WatchdogConfig> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.util.WatchdogConfig getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/WatchdogConfigOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/WatchdogConfigOrBuilder.java
deleted file mode 100644
index 74e6511d5ad..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/WatchdogConfigOrBuilder.java
+++ /dev/null
@@ -1,14 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/util/event.proto
-
-package org.tensorflow.proto.util;
-
-public interface WatchdogConfigOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.WatchdogConfig)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <code>int64 timeout_ms = 1;</code>
-   */
-  long getTimeoutMs();
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/WorkerHealth.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/WorkerHealth.java
deleted file mode 100644
index 268ecd3a6a0..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/WorkerHealth.java
+++ /dev/null
@@ -1,141 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/util/event.proto
-
-package org.tensorflow.proto.util;
-
-/**
- * <pre>
- * Current health status of a worker.
- * </pre>
- *
- * Protobuf enum {@code tensorflow.WorkerHealth}
- */
-public enum WorkerHealth
-    implements com.google.protobuf.ProtocolMessageEnum {
-  /**
-   * <pre>
-   * By default a worker is healthy.
-   * </pre>
-   *
-   * <code>OK = 0;</code>
-   */
-  OK(0),
-  /**
-   * <code>RECEIVED_SHUTDOWN_SIGNAL = 1;</code>
-   */
-  RECEIVED_SHUTDOWN_SIGNAL(1),
-  /**
-   * <code>INTERNAL_ERROR = 2;</code>
-   */
-  INTERNAL_ERROR(2),
-  /**
-   * <pre>
-   * Worker has been instructed to shutdown after a timeout.
-   * </pre>
-   *
-   * <code>SHUTTING_DOWN = 3;</code>
-   */
-  SHUTTING_DOWN(3),
-  UNRECOGNIZED(-1),
-  ;
-
-  /**
-   * <pre>
-   * By default a worker is healthy.
-   * </pre>
-   *
-   * <code>OK = 0;</code>
-   */
-  public static final int OK_VALUE = 0;
-  /**
-   * <code>RECEIVED_SHUTDOWN_SIGNAL = 1;</code>
-   */
-  public static final int RECEIVED_SHUTDOWN_SIGNAL_VALUE = 1;
-  /**
-   * <code>INTERNAL_ERROR = 2;</code>
-   */
-  public static final int INTERNAL_ERROR_VALUE = 2;
-  /**
-   * <pre>
-   * Worker has been instructed to shutdown after a timeout.
-   * </pre>
-   *
-   * <code>SHUTTING_DOWN = 3;</code>
-   */
-  public static final int SHUTTING_DOWN_VALUE = 3;
-
-
-  public final int getNumber() {
-    if (this == UNRECOGNIZED) {
-      throw new java.lang.IllegalArgumentException(
-          "Can't get the number of an unknown enum value.");
-    }
-    return value;
-  }
-
-  /**
-   * @deprecated Use {@link #forNumber(int)} instead.
-   */
-  @java.lang.Deprecated
-  public static WorkerHealth valueOf(int value) {
-    return forNumber(value);
-  }
-
-  public static WorkerHealth forNumber(int value) {
-    switch (value) {
-      case 0: return OK;
-      case 1: return RECEIVED_SHUTDOWN_SIGNAL;
-      case 2: return INTERNAL_ERROR;
-      case 3: return SHUTTING_DOWN;
-      default: return null;
-    }
-  }
-
-  public static com.google.protobuf.Internal.EnumLiteMap<WorkerHealth>
-      internalGetValueMap() {
-    return internalValueMap;
-  }
-  private static final com.google.protobuf.Internal.EnumLiteMap<
-      WorkerHealth> internalValueMap =
-        new com.google.protobuf.Internal.EnumLiteMap<WorkerHealth>() {
-          public WorkerHealth findValueByNumber(int number) {
-            return WorkerHealth.forNumber(number);
-          }
-        };
-
-  public final com.google.protobuf.Descriptors.EnumValueDescriptor
-      getValueDescriptor() {
-    return getDescriptor().getValues().get(ordinal());
-  }
-  public final com.google.protobuf.Descriptors.EnumDescriptor
-      getDescriptorForType() {
-    return getDescriptor();
-  }
-  public static final com.google.protobuf.Descriptors.EnumDescriptor
-      getDescriptor() {
-    return org.tensorflow.proto.util.EventProtos.getDescriptor().getEnumTypes().get(0);
-  }
-
-  private static final WorkerHealth[] VALUES = values();
-
-  public static WorkerHealth valueOf(
-      com.google.protobuf.Descriptors.EnumValueDescriptor desc) {
-    if (desc.getType() != getDescriptor()) {
-      throw new java.lang.IllegalArgumentException(
-        "EnumValueDescriptor is not for this type.");
-    }
-    if (desc.getIndex() == -1) {
-      return UNRECOGNIZED;
-    }
-    return VALUES[desc.getIndex()];
-  }
-
-  private final int value;
-
-  private WorkerHealth(int value) {
-    this.value = value;
-  }
-
-  // @@protoc_insertion_point(enum_scope:tensorflow.WorkerHealth)
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/WorkerHeartbeatRequest.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/WorkerHeartbeatRequest.java
deleted file mode 100644
index 9be331b3eca..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/WorkerHeartbeatRequest.java
+++ /dev/null
@@ -1,866 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/util/event.proto
-
-package org.tensorflow.proto.util;
-
-/**
- * Protobuf type {@code tensorflow.WorkerHeartbeatRequest}
- */
-public  final class WorkerHeartbeatRequest extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.WorkerHeartbeatRequest)
-    WorkerHeartbeatRequestOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use WorkerHeartbeatRequest.newBuilder() to construct.
-  private WorkerHeartbeatRequest(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private WorkerHeartbeatRequest() {
-    shutdownMode_ = 0;
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new WorkerHeartbeatRequest();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private WorkerHeartbeatRequest(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 8: {
-            int rawValue = input.readEnum();
-
-            shutdownMode_ = rawValue;
-            break;
-          }
-          case 18: {
-            org.tensorflow.proto.util.WatchdogConfig.Builder subBuilder = null;
-            if (watchdogConfig_ != null) {
-              subBuilder = watchdogConfig_.toBuilder();
-            }
-            watchdogConfig_ = input.readMessage(org.tensorflow.proto.util.WatchdogConfig.parser(), extensionRegistry);
-            if (subBuilder != null) {
-              subBuilder.mergeFrom(watchdogConfig_);
-              watchdogConfig_ = subBuilder.buildPartial();
-            }
-
-            break;
-          }
-          case 26: {
-            org.tensorflow.proto.util.RequestedExitCode.Builder subBuilder = null;
-            if (exitCode_ != null) {
-              subBuilder = exitCode_.toBuilder();
-            }
-            exitCode_ = input.readMessage(org.tensorflow.proto.util.RequestedExitCode.parser(), extensionRegistry);
-            if (subBuilder != null) {
-              subBuilder.mergeFrom(exitCode_);
-              exitCode_ = subBuilder.buildPartial();
-            }
-
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.util.EventProtos.internal_static_tensorflow_WorkerHeartbeatRequest_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.util.EventProtos.internal_static_tensorflow_WorkerHeartbeatRequest_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.util.WorkerHeartbeatRequest.class, org.tensorflow.proto.util.WorkerHeartbeatRequest.Builder.class);
-  }
-
-  public static final int SHUTDOWN_MODE_FIELD_NUMBER = 1;
-  private int shutdownMode_;
-  /**
-   * <code>.tensorflow.WorkerShutdownMode shutdown_mode = 1;</code>
-   */
-  public int getShutdownModeValue() {
-    return shutdownMode_;
-  }
-  /**
-   * <code>.tensorflow.WorkerShutdownMode shutdown_mode = 1;</code>
-   */
-  public org.tensorflow.proto.util.WorkerShutdownMode getShutdownMode() {
-    @SuppressWarnings("deprecation")
-    org.tensorflow.proto.util.WorkerShutdownMode result = org.tensorflow.proto.util.WorkerShutdownMode.valueOf(shutdownMode_);
-    return result == null ? org.tensorflow.proto.util.WorkerShutdownMode.UNRECOGNIZED : result;
-  }
-
-  public static final int WATCHDOG_CONFIG_FIELD_NUMBER = 2;
-  private org.tensorflow.proto.util.WatchdogConfig watchdogConfig_;
-  /**
-   * <code>.tensorflow.WatchdogConfig watchdog_config = 2;</code>
-   */
-  public boolean hasWatchdogConfig() {
-    return watchdogConfig_ != null;
-  }
-  /**
-   * <code>.tensorflow.WatchdogConfig watchdog_config = 2;</code>
-   */
-  public org.tensorflow.proto.util.WatchdogConfig getWatchdogConfig() {
-    return watchdogConfig_ == null ? org.tensorflow.proto.util.WatchdogConfig.getDefaultInstance() : watchdogConfig_;
-  }
-  /**
-   * <code>.tensorflow.WatchdogConfig watchdog_config = 2;</code>
-   */
-  public org.tensorflow.proto.util.WatchdogConfigOrBuilder getWatchdogConfigOrBuilder() {
-    return getWatchdogConfig();
-  }
-
-  public static final int EXIT_CODE_FIELD_NUMBER = 3;
-  private org.tensorflow.proto.util.RequestedExitCode exitCode_;
-  /**
-   * <code>.tensorflow.RequestedExitCode exit_code = 3;</code>
-   */
-  public boolean hasExitCode() {
-    return exitCode_ != null;
-  }
-  /**
-   * <code>.tensorflow.RequestedExitCode exit_code = 3;</code>
-   */
-  public org.tensorflow.proto.util.RequestedExitCode getExitCode() {
-    return exitCode_ == null ? org.tensorflow.proto.util.RequestedExitCode.getDefaultInstance() : exitCode_;
-  }
-  /**
-   * <code>.tensorflow.RequestedExitCode exit_code = 3;</code>
-   */
-  public org.tensorflow.proto.util.RequestedExitCodeOrBuilder getExitCodeOrBuilder() {
-    return getExitCode();
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    if (shutdownMode_ != org.tensorflow.proto.util.WorkerShutdownMode.DEFAULT.getNumber()) {
-      output.writeEnum(1, shutdownMode_);
-    }
-    if (watchdogConfig_ != null) {
-      output.writeMessage(2, getWatchdogConfig());
-    }
-    if (exitCode_ != null) {
-      output.writeMessage(3, getExitCode());
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (shutdownMode_ != org.tensorflow.proto.util.WorkerShutdownMode.DEFAULT.getNumber()) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeEnumSize(1, shutdownMode_);
-    }
-    if (watchdogConfig_ != null) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(2, getWatchdogConfig());
-    }
-    if (exitCode_ != null) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(3, getExitCode());
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.util.WorkerHeartbeatRequest)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.util.WorkerHeartbeatRequest other = (org.tensorflow.proto.util.WorkerHeartbeatRequest) obj;
-
-    if (shutdownMode_ != other.shutdownMode_) return false;
-    if (hasWatchdogConfig() != other.hasWatchdogConfig()) return false;
-    if (hasWatchdogConfig()) {
-      if (!getWatchdogConfig()
-          .equals(other.getWatchdogConfig())) return false;
-    }
-    if (hasExitCode() != other.hasExitCode()) return false;
-    if (hasExitCode()) {
-      if (!getExitCode()
-          .equals(other.getExitCode())) return false;
-    }
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    hash = (37 * hash) + SHUTDOWN_MODE_FIELD_NUMBER;
-    hash = (53 * hash) + shutdownMode_;
-    if (hasWatchdogConfig()) {
-      hash = (37 * hash) + WATCHDOG_CONFIG_FIELD_NUMBER;
-      hash = (53 * hash) + getWatchdogConfig().hashCode();
-    }
-    if (hasExitCode()) {
-      hash = (37 * hash) + EXIT_CODE_FIELD_NUMBER;
-      hash = (53 * hash) + getExitCode().hashCode();
-    }
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.util.WorkerHeartbeatRequest parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.WorkerHeartbeatRequest parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.WorkerHeartbeatRequest parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.WorkerHeartbeatRequest parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.WorkerHeartbeatRequest parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.WorkerHeartbeatRequest parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.WorkerHeartbeatRequest parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.WorkerHeartbeatRequest parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.WorkerHeartbeatRequest parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.WorkerHeartbeatRequest parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.WorkerHeartbeatRequest parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.WorkerHeartbeatRequest parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.util.WorkerHeartbeatRequest prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * Protobuf type {@code tensorflow.WorkerHeartbeatRequest}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.WorkerHeartbeatRequest)
-      org.tensorflow.proto.util.WorkerHeartbeatRequestOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.util.EventProtos.internal_static_tensorflow_WorkerHeartbeatRequest_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.util.EventProtos.internal_static_tensorflow_WorkerHeartbeatRequest_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.util.WorkerHeartbeatRequest.class, org.tensorflow.proto.util.WorkerHeartbeatRequest.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.util.WorkerHeartbeatRequest.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      shutdownMode_ = 0;
-
-      if (watchdogConfigBuilder_ == null) {
-        watchdogConfig_ = null;
-      } else {
-        watchdogConfig_ = null;
-        watchdogConfigBuilder_ = null;
-      }
-      if (exitCodeBuilder_ == null) {
-        exitCode_ = null;
-      } else {
-        exitCode_ = null;
-        exitCodeBuilder_ = null;
-      }
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.util.EventProtos.internal_static_tensorflow_WorkerHeartbeatRequest_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.WorkerHeartbeatRequest getDefaultInstanceForType() {
-      return org.tensorflow.proto.util.WorkerHeartbeatRequest.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.WorkerHeartbeatRequest build() {
-      org.tensorflow.proto.util.WorkerHeartbeatRequest result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.WorkerHeartbeatRequest buildPartial() {
-      org.tensorflow.proto.util.WorkerHeartbeatRequest result = new org.tensorflow.proto.util.WorkerHeartbeatRequest(this);
-      result.shutdownMode_ = shutdownMode_;
-      if (watchdogConfigBuilder_ == null) {
-        result.watchdogConfig_ = watchdogConfig_;
-      } else {
-        result.watchdogConfig_ = watchdogConfigBuilder_.build();
-      }
-      if (exitCodeBuilder_ == null) {
-        result.exitCode_ = exitCode_;
-      } else {
-        result.exitCode_ = exitCodeBuilder_.build();
-      }
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.util.WorkerHeartbeatRequest) {
-        return mergeFrom((org.tensorflow.proto.util.WorkerHeartbeatRequest)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.util.WorkerHeartbeatRequest other) {
-      if (other == org.tensorflow.proto.util.WorkerHeartbeatRequest.getDefaultInstance()) return this;
-      if (other.shutdownMode_ != 0) {
-        setShutdownModeValue(other.getShutdownModeValue());
-      }
-      if (other.hasWatchdogConfig()) {
-        mergeWatchdogConfig(other.getWatchdogConfig());
-      }
-      if (other.hasExitCode()) {
-        mergeExitCode(other.getExitCode());
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.util.WorkerHeartbeatRequest parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.util.WorkerHeartbeatRequest) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-
-    private int shutdownMode_ = 0;
-    /**
-     * <code>.tensorflow.WorkerShutdownMode shutdown_mode = 1;</code>
-     */
-    public int getShutdownModeValue() {
-      return shutdownMode_;
-    }
-    /**
-     * <code>.tensorflow.WorkerShutdownMode shutdown_mode = 1;</code>
-     */
-    public Builder setShutdownModeValue(int value) {
-      shutdownMode_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>.tensorflow.WorkerShutdownMode shutdown_mode = 1;</code>
-     */
-    public org.tensorflow.proto.util.WorkerShutdownMode getShutdownMode() {
-      @SuppressWarnings("deprecation")
-      org.tensorflow.proto.util.WorkerShutdownMode result = org.tensorflow.proto.util.WorkerShutdownMode.valueOf(shutdownMode_);
-      return result == null ? org.tensorflow.proto.util.WorkerShutdownMode.UNRECOGNIZED : result;
-    }
-    /**
-     * <code>.tensorflow.WorkerShutdownMode shutdown_mode = 1;</code>
-     */
-    public Builder setShutdownMode(org.tensorflow.proto.util.WorkerShutdownMode value) {
-      if (value == null) {
-        throw new NullPointerException();
-      }
-      
-      shutdownMode_ = value.getNumber();
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>.tensorflow.WorkerShutdownMode shutdown_mode = 1;</code>
-     */
-    public Builder clearShutdownMode() {
-      
-      shutdownMode_ = 0;
-      onChanged();
-      return this;
-    }
-
-    private org.tensorflow.proto.util.WatchdogConfig watchdogConfig_;
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.util.WatchdogConfig, org.tensorflow.proto.util.WatchdogConfig.Builder, org.tensorflow.proto.util.WatchdogConfigOrBuilder> watchdogConfigBuilder_;
-    /**
-     * <code>.tensorflow.WatchdogConfig watchdog_config = 2;</code>
-     */
-    public boolean hasWatchdogConfig() {
-      return watchdogConfigBuilder_ != null || watchdogConfig_ != null;
-    }
-    /**
-     * <code>.tensorflow.WatchdogConfig watchdog_config = 2;</code>
-     */
-    public org.tensorflow.proto.util.WatchdogConfig getWatchdogConfig() {
-      if (watchdogConfigBuilder_ == null) {
-        return watchdogConfig_ == null ? org.tensorflow.proto.util.WatchdogConfig.getDefaultInstance() : watchdogConfig_;
-      } else {
-        return watchdogConfigBuilder_.getMessage();
-      }
-    }
-    /**
-     * <code>.tensorflow.WatchdogConfig watchdog_config = 2;</code>
-     */
-    public Builder setWatchdogConfig(org.tensorflow.proto.util.WatchdogConfig value) {
-      if (watchdogConfigBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        watchdogConfig_ = value;
-        onChanged();
-      } else {
-        watchdogConfigBuilder_.setMessage(value);
-      }
-
-      return this;
-    }
-    /**
-     * <code>.tensorflow.WatchdogConfig watchdog_config = 2;</code>
-     */
-    public Builder setWatchdogConfig(
-        org.tensorflow.proto.util.WatchdogConfig.Builder builderForValue) {
-      if (watchdogConfigBuilder_ == null) {
-        watchdogConfig_ = builderForValue.build();
-        onChanged();
-      } else {
-        watchdogConfigBuilder_.setMessage(builderForValue.build());
-      }
-
-      return this;
-    }
-    /**
-     * <code>.tensorflow.WatchdogConfig watchdog_config = 2;</code>
-     */
-    public Builder mergeWatchdogConfig(org.tensorflow.proto.util.WatchdogConfig value) {
-      if (watchdogConfigBuilder_ == null) {
-        if (watchdogConfig_ != null) {
-          watchdogConfig_ =
-            org.tensorflow.proto.util.WatchdogConfig.newBuilder(watchdogConfig_).mergeFrom(value).buildPartial();
-        } else {
-          watchdogConfig_ = value;
-        }
-        onChanged();
-      } else {
-        watchdogConfigBuilder_.mergeFrom(value);
-      }
-
-      return this;
-    }
-    /**
-     * <code>.tensorflow.WatchdogConfig watchdog_config = 2;</code>
-     */
-    public Builder clearWatchdogConfig() {
-      if (watchdogConfigBuilder_ == null) {
-        watchdogConfig_ = null;
-        onChanged();
-      } else {
-        watchdogConfig_ = null;
-        watchdogConfigBuilder_ = null;
-      }
-
-      return this;
-    }
-    /**
-     * <code>.tensorflow.WatchdogConfig watchdog_config = 2;</code>
-     */
-    public org.tensorflow.proto.util.WatchdogConfig.Builder getWatchdogConfigBuilder() {
-      
-      onChanged();
-      return getWatchdogConfigFieldBuilder().getBuilder();
-    }
-    /**
-     * <code>.tensorflow.WatchdogConfig watchdog_config = 2;</code>
-     */
-    public org.tensorflow.proto.util.WatchdogConfigOrBuilder getWatchdogConfigOrBuilder() {
-      if (watchdogConfigBuilder_ != null) {
-        return watchdogConfigBuilder_.getMessageOrBuilder();
-      } else {
-        return watchdogConfig_ == null ?
-            org.tensorflow.proto.util.WatchdogConfig.getDefaultInstance() : watchdogConfig_;
-      }
-    }
-    /**
-     * <code>.tensorflow.WatchdogConfig watchdog_config = 2;</code>
-     */
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.util.WatchdogConfig, org.tensorflow.proto.util.WatchdogConfig.Builder, org.tensorflow.proto.util.WatchdogConfigOrBuilder> 
-        getWatchdogConfigFieldBuilder() {
-      if (watchdogConfigBuilder_ == null) {
-        watchdogConfigBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-            org.tensorflow.proto.util.WatchdogConfig, org.tensorflow.proto.util.WatchdogConfig.Builder, org.tensorflow.proto.util.WatchdogConfigOrBuilder>(
-                getWatchdogConfig(),
-                getParentForChildren(),
-                isClean());
-        watchdogConfig_ = null;
-      }
-      return watchdogConfigBuilder_;
-    }
-
-    private org.tensorflow.proto.util.RequestedExitCode exitCode_;
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.util.RequestedExitCode, org.tensorflow.proto.util.RequestedExitCode.Builder, org.tensorflow.proto.util.RequestedExitCodeOrBuilder> exitCodeBuilder_;
-    /**
-     * <code>.tensorflow.RequestedExitCode exit_code = 3;</code>
-     */
-    public boolean hasExitCode() {
-      return exitCodeBuilder_ != null || exitCode_ != null;
-    }
-    /**
-     * <code>.tensorflow.RequestedExitCode exit_code = 3;</code>
-     */
-    public org.tensorflow.proto.util.RequestedExitCode getExitCode() {
-      if (exitCodeBuilder_ == null) {
-        return exitCode_ == null ? org.tensorflow.proto.util.RequestedExitCode.getDefaultInstance() : exitCode_;
-      } else {
-        return exitCodeBuilder_.getMessage();
-      }
-    }
-    /**
-     * <code>.tensorflow.RequestedExitCode exit_code = 3;</code>
-     */
-    public Builder setExitCode(org.tensorflow.proto.util.RequestedExitCode value) {
-      if (exitCodeBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        exitCode_ = value;
-        onChanged();
-      } else {
-        exitCodeBuilder_.setMessage(value);
-      }
-
-      return this;
-    }
-    /**
-     * <code>.tensorflow.RequestedExitCode exit_code = 3;</code>
-     */
-    public Builder setExitCode(
-        org.tensorflow.proto.util.RequestedExitCode.Builder builderForValue) {
-      if (exitCodeBuilder_ == null) {
-        exitCode_ = builderForValue.build();
-        onChanged();
-      } else {
-        exitCodeBuilder_.setMessage(builderForValue.build());
-      }
-
-      return this;
-    }
-    /**
-     * <code>.tensorflow.RequestedExitCode exit_code = 3;</code>
-     */
-    public Builder mergeExitCode(org.tensorflow.proto.util.RequestedExitCode value) {
-      if (exitCodeBuilder_ == null) {
-        if (exitCode_ != null) {
-          exitCode_ =
-            org.tensorflow.proto.util.RequestedExitCode.newBuilder(exitCode_).mergeFrom(value).buildPartial();
-        } else {
-          exitCode_ = value;
-        }
-        onChanged();
-      } else {
-        exitCodeBuilder_.mergeFrom(value);
-      }
-
-      return this;
-    }
-    /**
-     * <code>.tensorflow.RequestedExitCode exit_code = 3;</code>
-     */
-    public Builder clearExitCode() {
-      if (exitCodeBuilder_ == null) {
-        exitCode_ = null;
-        onChanged();
-      } else {
-        exitCode_ = null;
-        exitCodeBuilder_ = null;
-      }
-
-      return this;
-    }
-    /**
-     * <code>.tensorflow.RequestedExitCode exit_code = 3;</code>
-     */
-    public org.tensorflow.proto.util.RequestedExitCode.Builder getExitCodeBuilder() {
-      
-      onChanged();
-      return getExitCodeFieldBuilder().getBuilder();
-    }
-    /**
-     * <code>.tensorflow.RequestedExitCode exit_code = 3;</code>
-     */
-    public org.tensorflow.proto.util.RequestedExitCodeOrBuilder getExitCodeOrBuilder() {
-      if (exitCodeBuilder_ != null) {
-        return exitCodeBuilder_.getMessageOrBuilder();
-      } else {
-        return exitCode_ == null ?
-            org.tensorflow.proto.util.RequestedExitCode.getDefaultInstance() : exitCode_;
-      }
-    }
-    /**
-     * <code>.tensorflow.RequestedExitCode exit_code = 3;</code>
-     */
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.util.RequestedExitCode, org.tensorflow.proto.util.RequestedExitCode.Builder, org.tensorflow.proto.util.RequestedExitCodeOrBuilder> 
-        getExitCodeFieldBuilder() {
-      if (exitCodeBuilder_ == null) {
-        exitCodeBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-            org.tensorflow.proto.util.RequestedExitCode, org.tensorflow.proto.util.RequestedExitCode.Builder, org.tensorflow.proto.util.RequestedExitCodeOrBuilder>(
-                getExitCode(),
-                getParentForChildren(),
-                isClean());
-        exitCode_ = null;
-      }
-      return exitCodeBuilder_;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.WorkerHeartbeatRequest)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.WorkerHeartbeatRequest)
-  private static final org.tensorflow.proto.util.WorkerHeartbeatRequest DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.util.WorkerHeartbeatRequest();
-  }
-
-  public static org.tensorflow.proto.util.WorkerHeartbeatRequest getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<WorkerHeartbeatRequest>
-      PARSER = new com.google.protobuf.AbstractParser<WorkerHeartbeatRequest>() {
-    @java.lang.Override
-    public WorkerHeartbeatRequest parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new WorkerHeartbeatRequest(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<WorkerHeartbeatRequest> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<WorkerHeartbeatRequest> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.util.WorkerHeartbeatRequest getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/WorkerHeartbeatRequestOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/WorkerHeartbeatRequestOrBuilder.java
deleted file mode 100644
index 28e78122ba0..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/WorkerHeartbeatRequestOrBuilder.java
+++ /dev/null
@@ -1,44 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/util/event.proto
-
-package org.tensorflow.proto.util;
-
-public interface WorkerHeartbeatRequestOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.WorkerHeartbeatRequest)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <code>.tensorflow.WorkerShutdownMode shutdown_mode = 1;</code>
-   */
-  int getShutdownModeValue();
-  /**
-   * <code>.tensorflow.WorkerShutdownMode shutdown_mode = 1;</code>
-   */
-  org.tensorflow.proto.util.WorkerShutdownMode getShutdownMode();
-
-  /**
-   * <code>.tensorflow.WatchdogConfig watchdog_config = 2;</code>
-   */
-  boolean hasWatchdogConfig();
-  /**
-   * <code>.tensorflow.WatchdogConfig watchdog_config = 2;</code>
-   */
-  org.tensorflow.proto.util.WatchdogConfig getWatchdogConfig();
-  /**
-   * <code>.tensorflow.WatchdogConfig watchdog_config = 2;</code>
-   */
-  org.tensorflow.proto.util.WatchdogConfigOrBuilder getWatchdogConfigOrBuilder();
-
-  /**
-   * <code>.tensorflow.RequestedExitCode exit_code = 3;</code>
-   */
-  boolean hasExitCode();
-  /**
-   * <code>.tensorflow.RequestedExitCode exit_code = 3;</code>
-   */
-  org.tensorflow.proto.util.RequestedExitCode getExitCode();
-  /**
-   * <code>.tensorflow.RequestedExitCode exit_code = 3;</code>
-   */
-  org.tensorflow.proto.util.RequestedExitCodeOrBuilder getExitCodeOrBuilder();
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/WorkerHeartbeatResponse.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/WorkerHeartbeatResponse.java
deleted file mode 100644
index 87262dd0405..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/WorkerHeartbeatResponse.java
+++ /dev/null
@@ -1,977 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/util/event.proto
-
-package org.tensorflow.proto.util;
-
-/**
- * Protobuf type {@code tensorflow.WorkerHeartbeatResponse}
- */
-public  final class WorkerHeartbeatResponse extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.WorkerHeartbeatResponse)
-    WorkerHeartbeatResponseOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use WorkerHeartbeatResponse.newBuilder() to construct.
-  private WorkerHeartbeatResponse(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private WorkerHeartbeatResponse() {
-    healthStatus_ = 0;
-    workerLog_ = java.util.Collections.emptyList();
-    hostname_ = "";
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new WorkerHeartbeatResponse();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private WorkerHeartbeatResponse(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    int mutable_bitField0_ = 0;
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 8: {
-            int rawValue = input.readEnum();
-
-            healthStatus_ = rawValue;
-            break;
-          }
-          case 18: {
-            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
-              workerLog_ = new java.util.ArrayList<org.tensorflow.proto.util.Event>();
-              mutable_bitField0_ |= 0x00000001;
-            }
-            workerLog_.add(
-                input.readMessage(org.tensorflow.proto.util.Event.parser(), extensionRegistry));
-            break;
-          }
-          case 26: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            hostname_ = s;
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      if (((mutable_bitField0_ & 0x00000001) != 0)) {
-        workerLog_ = java.util.Collections.unmodifiableList(workerLog_);
-      }
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.util.EventProtos.internal_static_tensorflow_WorkerHeartbeatResponse_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.util.EventProtos.internal_static_tensorflow_WorkerHeartbeatResponse_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.util.WorkerHeartbeatResponse.class, org.tensorflow.proto.util.WorkerHeartbeatResponse.Builder.class);
-  }
-
-  public static final int HEALTH_STATUS_FIELD_NUMBER = 1;
-  private int healthStatus_;
-  /**
-   * <code>.tensorflow.WorkerHealth health_status = 1;</code>
-   */
-  public int getHealthStatusValue() {
-    return healthStatus_;
-  }
-  /**
-   * <code>.tensorflow.WorkerHealth health_status = 1;</code>
-   */
-  public org.tensorflow.proto.util.WorkerHealth getHealthStatus() {
-    @SuppressWarnings("deprecation")
-    org.tensorflow.proto.util.WorkerHealth result = org.tensorflow.proto.util.WorkerHealth.valueOf(healthStatus_);
-    return result == null ? org.tensorflow.proto.util.WorkerHealth.UNRECOGNIZED : result;
-  }
-
-  public static final int WORKER_LOG_FIELD_NUMBER = 2;
-  private java.util.List<org.tensorflow.proto.util.Event> workerLog_;
-  /**
-   * <code>repeated .tensorflow.Event worker_log = 2;</code>
-   */
-  public java.util.List<org.tensorflow.proto.util.Event> getWorkerLogList() {
-    return workerLog_;
-  }
-  /**
-   * <code>repeated .tensorflow.Event worker_log = 2;</code>
-   */
-  public java.util.List<? extends org.tensorflow.proto.util.EventOrBuilder> 
-      getWorkerLogOrBuilderList() {
-    return workerLog_;
-  }
-  /**
-   * <code>repeated .tensorflow.Event worker_log = 2;</code>
-   */
-  public int getWorkerLogCount() {
-    return workerLog_.size();
-  }
-  /**
-   * <code>repeated .tensorflow.Event worker_log = 2;</code>
-   */
-  public org.tensorflow.proto.util.Event getWorkerLog(int index) {
-    return workerLog_.get(index);
-  }
-  /**
-   * <code>repeated .tensorflow.Event worker_log = 2;</code>
-   */
-  public org.tensorflow.proto.util.EventOrBuilder getWorkerLogOrBuilder(
-      int index) {
-    return workerLog_.get(index);
-  }
-
-  public static final int HOSTNAME_FIELD_NUMBER = 3;
-  private volatile java.lang.Object hostname_;
-  /**
-   * <code>string hostname = 3;</code>
-   */
-  public java.lang.String getHostname() {
-    java.lang.Object ref = hostname_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      hostname_ = s;
-      return s;
-    }
-  }
-  /**
-   * <code>string hostname = 3;</code>
-   */
-  public com.google.protobuf.ByteString
-      getHostnameBytes() {
-    java.lang.Object ref = hostname_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      hostname_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    if (healthStatus_ != org.tensorflow.proto.util.WorkerHealth.OK.getNumber()) {
-      output.writeEnum(1, healthStatus_);
-    }
-    for (int i = 0; i < workerLog_.size(); i++) {
-      output.writeMessage(2, workerLog_.get(i));
-    }
-    if (!getHostnameBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 3, hostname_);
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (healthStatus_ != org.tensorflow.proto.util.WorkerHealth.OK.getNumber()) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeEnumSize(1, healthStatus_);
-    }
-    for (int i = 0; i < workerLog_.size(); i++) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(2, workerLog_.get(i));
-    }
-    if (!getHostnameBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(3, hostname_);
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.util.WorkerHeartbeatResponse)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.util.WorkerHeartbeatResponse other = (org.tensorflow.proto.util.WorkerHeartbeatResponse) obj;
-
-    if (healthStatus_ != other.healthStatus_) return false;
-    if (!getWorkerLogList()
-        .equals(other.getWorkerLogList())) return false;
-    if (!getHostname()
-        .equals(other.getHostname())) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    hash = (37 * hash) + HEALTH_STATUS_FIELD_NUMBER;
-    hash = (53 * hash) + healthStatus_;
-    if (getWorkerLogCount() > 0) {
-      hash = (37 * hash) + WORKER_LOG_FIELD_NUMBER;
-      hash = (53 * hash) + getWorkerLogList().hashCode();
-    }
-    hash = (37 * hash) + HOSTNAME_FIELD_NUMBER;
-    hash = (53 * hash) + getHostname().hashCode();
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.util.WorkerHeartbeatResponse parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.WorkerHeartbeatResponse parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.WorkerHeartbeatResponse parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.WorkerHeartbeatResponse parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.WorkerHeartbeatResponse parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.WorkerHeartbeatResponse parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.WorkerHeartbeatResponse parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.WorkerHeartbeatResponse parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.WorkerHeartbeatResponse parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.WorkerHeartbeatResponse parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.WorkerHeartbeatResponse parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.WorkerHeartbeatResponse parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.util.WorkerHeartbeatResponse prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * Protobuf type {@code tensorflow.WorkerHeartbeatResponse}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.WorkerHeartbeatResponse)
-      org.tensorflow.proto.util.WorkerHeartbeatResponseOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.util.EventProtos.internal_static_tensorflow_WorkerHeartbeatResponse_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.util.EventProtos.internal_static_tensorflow_WorkerHeartbeatResponse_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.util.WorkerHeartbeatResponse.class, org.tensorflow.proto.util.WorkerHeartbeatResponse.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.util.WorkerHeartbeatResponse.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-        getWorkerLogFieldBuilder();
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      healthStatus_ = 0;
-
-      if (workerLogBuilder_ == null) {
-        workerLog_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000001);
-      } else {
-        workerLogBuilder_.clear();
-      }
-      hostname_ = "";
-
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.util.EventProtos.internal_static_tensorflow_WorkerHeartbeatResponse_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.WorkerHeartbeatResponse getDefaultInstanceForType() {
-      return org.tensorflow.proto.util.WorkerHeartbeatResponse.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.WorkerHeartbeatResponse build() {
-      org.tensorflow.proto.util.WorkerHeartbeatResponse result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.WorkerHeartbeatResponse buildPartial() {
-      org.tensorflow.proto.util.WorkerHeartbeatResponse result = new org.tensorflow.proto.util.WorkerHeartbeatResponse(this);
-      int from_bitField0_ = bitField0_;
-      result.healthStatus_ = healthStatus_;
-      if (workerLogBuilder_ == null) {
-        if (((bitField0_ & 0x00000001) != 0)) {
-          workerLog_ = java.util.Collections.unmodifiableList(workerLog_);
-          bitField0_ = (bitField0_ & ~0x00000001);
-        }
-        result.workerLog_ = workerLog_;
-      } else {
-        result.workerLog_ = workerLogBuilder_.build();
-      }
-      result.hostname_ = hostname_;
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.util.WorkerHeartbeatResponse) {
-        return mergeFrom((org.tensorflow.proto.util.WorkerHeartbeatResponse)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.util.WorkerHeartbeatResponse other) {
-      if (other == org.tensorflow.proto.util.WorkerHeartbeatResponse.getDefaultInstance()) return this;
-      if (other.healthStatus_ != 0) {
-        setHealthStatusValue(other.getHealthStatusValue());
-      }
-      if (workerLogBuilder_ == null) {
-        if (!other.workerLog_.isEmpty()) {
-          if (workerLog_.isEmpty()) {
-            workerLog_ = other.workerLog_;
-            bitField0_ = (bitField0_ & ~0x00000001);
-          } else {
-            ensureWorkerLogIsMutable();
-            workerLog_.addAll(other.workerLog_);
-          }
-          onChanged();
-        }
-      } else {
-        if (!other.workerLog_.isEmpty()) {
-          if (workerLogBuilder_.isEmpty()) {
-            workerLogBuilder_.dispose();
-            workerLogBuilder_ = null;
-            workerLog_ = other.workerLog_;
-            bitField0_ = (bitField0_ & ~0x00000001);
-            workerLogBuilder_ = 
-              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
-                 getWorkerLogFieldBuilder() : null;
-          } else {
-            workerLogBuilder_.addAllMessages(other.workerLog_);
-          }
-        }
-      }
-      if (!other.getHostname().isEmpty()) {
-        hostname_ = other.hostname_;
-        onChanged();
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.util.WorkerHeartbeatResponse parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.util.WorkerHeartbeatResponse) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-    private int bitField0_;
-
-    private int healthStatus_ = 0;
-    /**
-     * <code>.tensorflow.WorkerHealth health_status = 1;</code>
-     */
-    public int getHealthStatusValue() {
-      return healthStatus_;
-    }
-    /**
-     * <code>.tensorflow.WorkerHealth health_status = 1;</code>
-     */
-    public Builder setHealthStatusValue(int value) {
-      healthStatus_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>.tensorflow.WorkerHealth health_status = 1;</code>
-     */
-    public org.tensorflow.proto.util.WorkerHealth getHealthStatus() {
-      @SuppressWarnings("deprecation")
-      org.tensorflow.proto.util.WorkerHealth result = org.tensorflow.proto.util.WorkerHealth.valueOf(healthStatus_);
-      return result == null ? org.tensorflow.proto.util.WorkerHealth.UNRECOGNIZED : result;
-    }
-    /**
-     * <code>.tensorflow.WorkerHealth health_status = 1;</code>
-     */
-    public Builder setHealthStatus(org.tensorflow.proto.util.WorkerHealth value) {
-      if (value == null) {
-        throw new NullPointerException();
-      }
-      
-      healthStatus_ = value.getNumber();
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>.tensorflow.WorkerHealth health_status = 1;</code>
-     */
-    public Builder clearHealthStatus() {
-      
-      healthStatus_ = 0;
-      onChanged();
-      return this;
-    }
-
-    private java.util.List<org.tensorflow.proto.util.Event> workerLog_ =
-      java.util.Collections.emptyList();
-    private void ensureWorkerLogIsMutable() {
-      if (!((bitField0_ & 0x00000001) != 0)) {
-        workerLog_ = new java.util.ArrayList<org.tensorflow.proto.util.Event>(workerLog_);
-        bitField0_ |= 0x00000001;
-       }
-    }
-
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.util.Event, org.tensorflow.proto.util.Event.Builder, org.tensorflow.proto.util.EventOrBuilder> workerLogBuilder_;
-
-    /**
-     * <code>repeated .tensorflow.Event worker_log = 2;</code>
-     */
-    public java.util.List<org.tensorflow.proto.util.Event> getWorkerLogList() {
-      if (workerLogBuilder_ == null) {
-        return java.util.Collections.unmodifiableList(workerLog_);
-      } else {
-        return workerLogBuilder_.getMessageList();
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.Event worker_log = 2;</code>
-     */
-    public int getWorkerLogCount() {
-      if (workerLogBuilder_ == null) {
-        return workerLog_.size();
-      } else {
-        return workerLogBuilder_.getCount();
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.Event worker_log = 2;</code>
-     */
-    public org.tensorflow.proto.util.Event getWorkerLog(int index) {
-      if (workerLogBuilder_ == null) {
-        return workerLog_.get(index);
-      } else {
-        return workerLogBuilder_.getMessage(index);
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.Event worker_log = 2;</code>
-     */
-    public Builder setWorkerLog(
-        int index, org.tensorflow.proto.util.Event value) {
-      if (workerLogBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureWorkerLogIsMutable();
-        workerLog_.set(index, value);
-        onChanged();
-      } else {
-        workerLogBuilder_.setMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.Event worker_log = 2;</code>
-     */
-    public Builder setWorkerLog(
-        int index, org.tensorflow.proto.util.Event.Builder builderForValue) {
-      if (workerLogBuilder_ == null) {
-        ensureWorkerLogIsMutable();
-        workerLog_.set(index, builderForValue.build());
-        onChanged();
-      } else {
-        workerLogBuilder_.setMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.Event worker_log = 2;</code>
-     */
-    public Builder addWorkerLog(org.tensorflow.proto.util.Event value) {
-      if (workerLogBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureWorkerLogIsMutable();
-        workerLog_.add(value);
-        onChanged();
-      } else {
-        workerLogBuilder_.addMessage(value);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.Event worker_log = 2;</code>
-     */
-    public Builder addWorkerLog(
-        int index, org.tensorflow.proto.util.Event value) {
-      if (workerLogBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureWorkerLogIsMutable();
-        workerLog_.add(index, value);
-        onChanged();
-      } else {
-        workerLogBuilder_.addMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.Event worker_log = 2;</code>
-     */
-    public Builder addWorkerLog(
-        org.tensorflow.proto.util.Event.Builder builderForValue) {
-      if (workerLogBuilder_ == null) {
-        ensureWorkerLogIsMutable();
-        workerLog_.add(builderForValue.build());
-        onChanged();
-      } else {
-        workerLogBuilder_.addMessage(builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.Event worker_log = 2;</code>
-     */
-    public Builder addWorkerLog(
-        int index, org.tensorflow.proto.util.Event.Builder builderForValue) {
-      if (workerLogBuilder_ == null) {
-        ensureWorkerLogIsMutable();
-        workerLog_.add(index, builderForValue.build());
-        onChanged();
-      } else {
-        workerLogBuilder_.addMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.Event worker_log = 2;</code>
-     */
-    public Builder addAllWorkerLog(
-        java.lang.Iterable<? extends org.tensorflow.proto.util.Event> values) {
-      if (workerLogBuilder_ == null) {
-        ensureWorkerLogIsMutable();
-        com.google.protobuf.AbstractMessageLite.Builder.addAll(
-            values, workerLog_);
-        onChanged();
-      } else {
-        workerLogBuilder_.addAllMessages(values);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.Event worker_log = 2;</code>
-     */
-    public Builder clearWorkerLog() {
-      if (workerLogBuilder_ == null) {
-        workerLog_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000001);
-        onChanged();
-      } else {
-        workerLogBuilder_.clear();
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.Event worker_log = 2;</code>
-     */
-    public Builder removeWorkerLog(int index) {
-      if (workerLogBuilder_ == null) {
-        ensureWorkerLogIsMutable();
-        workerLog_.remove(index);
-        onChanged();
-      } else {
-        workerLogBuilder_.remove(index);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.Event worker_log = 2;</code>
-     */
-    public org.tensorflow.proto.util.Event.Builder getWorkerLogBuilder(
-        int index) {
-      return getWorkerLogFieldBuilder().getBuilder(index);
-    }
-    /**
-     * <code>repeated .tensorflow.Event worker_log = 2;</code>
-     */
-    public org.tensorflow.proto.util.EventOrBuilder getWorkerLogOrBuilder(
-        int index) {
-      if (workerLogBuilder_ == null) {
-        return workerLog_.get(index);  } else {
-        return workerLogBuilder_.getMessageOrBuilder(index);
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.Event worker_log = 2;</code>
-     */
-    public java.util.List<? extends org.tensorflow.proto.util.EventOrBuilder> 
-         getWorkerLogOrBuilderList() {
-      if (workerLogBuilder_ != null) {
-        return workerLogBuilder_.getMessageOrBuilderList();
-      } else {
-        return java.util.Collections.unmodifiableList(workerLog_);
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.Event worker_log = 2;</code>
-     */
-    public org.tensorflow.proto.util.Event.Builder addWorkerLogBuilder() {
-      return getWorkerLogFieldBuilder().addBuilder(
-          org.tensorflow.proto.util.Event.getDefaultInstance());
-    }
-    /**
-     * <code>repeated .tensorflow.Event worker_log = 2;</code>
-     */
-    public org.tensorflow.proto.util.Event.Builder addWorkerLogBuilder(
-        int index) {
-      return getWorkerLogFieldBuilder().addBuilder(
-          index, org.tensorflow.proto.util.Event.getDefaultInstance());
-    }
-    /**
-     * <code>repeated .tensorflow.Event worker_log = 2;</code>
-     */
-    public java.util.List<org.tensorflow.proto.util.Event.Builder> 
-         getWorkerLogBuilderList() {
-      return getWorkerLogFieldBuilder().getBuilderList();
-    }
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.util.Event, org.tensorflow.proto.util.Event.Builder, org.tensorflow.proto.util.EventOrBuilder> 
-        getWorkerLogFieldBuilder() {
-      if (workerLogBuilder_ == null) {
-        workerLogBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
-            org.tensorflow.proto.util.Event, org.tensorflow.proto.util.Event.Builder, org.tensorflow.proto.util.EventOrBuilder>(
-                workerLog_,
-                ((bitField0_ & 0x00000001) != 0),
-                getParentForChildren(),
-                isClean());
-        workerLog_ = null;
-      }
-      return workerLogBuilder_;
-    }
-
-    private java.lang.Object hostname_ = "";
-    /**
-     * <code>string hostname = 3;</code>
-     */
-    public java.lang.String getHostname() {
-      java.lang.Object ref = hostname_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        hostname_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <code>string hostname = 3;</code>
-     */
-    public com.google.protobuf.ByteString
-        getHostnameBytes() {
-      java.lang.Object ref = hostname_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        hostname_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <code>string hostname = 3;</code>
-     */
-    public Builder setHostname(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      hostname_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>string hostname = 3;</code>
-     */
-    public Builder clearHostname() {
-      
-      hostname_ = getDefaultInstance().getHostname();
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>string hostname = 3;</code>
-     */
-    public Builder setHostnameBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      hostname_ = value;
-      onChanged();
-      return this;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.WorkerHeartbeatResponse)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.WorkerHeartbeatResponse)
-  private static final org.tensorflow.proto.util.WorkerHeartbeatResponse DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.util.WorkerHeartbeatResponse();
-  }
-
-  public static org.tensorflow.proto.util.WorkerHeartbeatResponse getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<WorkerHeartbeatResponse>
-      PARSER = new com.google.protobuf.AbstractParser<WorkerHeartbeatResponse>() {
-    @java.lang.Override
-    public WorkerHeartbeatResponse parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new WorkerHeartbeatResponse(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<WorkerHeartbeatResponse> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<WorkerHeartbeatResponse> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.util.WorkerHeartbeatResponse getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/WorkerHeartbeatResponseOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/WorkerHeartbeatResponseOrBuilder.java
deleted file mode 100644
index 5f42f819a51..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/WorkerHeartbeatResponseOrBuilder.java
+++ /dev/null
@@ -1,52 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/util/event.proto
-
-package org.tensorflow.proto.util;
-
-public interface WorkerHeartbeatResponseOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.WorkerHeartbeatResponse)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <code>.tensorflow.WorkerHealth health_status = 1;</code>
-   */
-  int getHealthStatusValue();
-  /**
-   * <code>.tensorflow.WorkerHealth health_status = 1;</code>
-   */
-  org.tensorflow.proto.util.WorkerHealth getHealthStatus();
-
-  /**
-   * <code>repeated .tensorflow.Event worker_log = 2;</code>
-   */
-  java.util.List<org.tensorflow.proto.util.Event> 
-      getWorkerLogList();
-  /**
-   * <code>repeated .tensorflow.Event worker_log = 2;</code>
-   */
-  org.tensorflow.proto.util.Event getWorkerLog(int index);
-  /**
-   * <code>repeated .tensorflow.Event worker_log = 2;</code>
-   */
-  int getWorkerLogCount();
-  /**
-   * <code>repeated .tensorflow.Event worker_log = 2;</code>
-   */
-  java.util.List<? extends org.tensorflow.proto.util.EventOrBuilder> 
-      getWorkerLogOrBuilderList();
-  /**
-   * <code>repeated .tensorflow.Event worker_log = 2;</code>
-   */
-  org.tensorflow.proto.util.EventOrBuilder getWorkerLogOrBuilder(
-      int index);
-
-  /**
-   * <code>string hostname = 3;</code>
-   */
-  java.lang.String getHostname();
-  /**
-   * <code>string hostname = 3;</code>
-   */
-  com.google.protobuf.ByteString
-      getHostnameBytes();
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/WorkerShutdownMode.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/WorkerShutdownMode.java
deleted file mode 100644
index 41c4f45373c..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/WorkerShutdownMode.java
+++ /dev/null
@@ -1,126 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/util/event.proto
-
-package org.tensorflow.proto.util;
-
-/**
- * <pre>
- * Indicates the behavior of the worker when an internal error or shutdown
- * signal is received.
- * </pre>
- *
- * Protobuf enum {@code tensorflow.WorkerShutdownMode}
- */
-public enum WorkerShutdownMode
-    implements com.google.protobuf.ProtocolMessageEnum {
-  /**
-   * <code>DEFAULT = 0;</code>
-   */
-  DEFAULT(0),
-  /**
-   * <code>NOT_CONFIGURED = 1;</code>
-   */
-  NOT_CONFIGURED(1),
-  /**
-   * <code>WAIT_FOR_COORDINATOR = 2;</code>
-   */
-  WAIT_FOR_COORDINATOR(2),
-  /**
-   * <code>SHUTDOWN_AFTER_TIMEOUT = 3;</code>
-   */
-  SHUTDOWN_AFTER_TIMEOUT(3),
-  UNRECOGNIZED(-1),
-  ;
-
-  /**
-   * <code>DEFAULT = 0;</code>
-   */
-  public static final int DEFAULT_VALUE = 0;
-  /**
-   * <code>NOT_CONFIGURED = 1;</code>
-   */
-  public static final int NOT_CONFIGURED_VALUE = 1;
-  /**
-   * <code>WAIT_FOR_COORDINATOR = 2;</code>
-   */
-  public static final int WAIT_FOR_COORDINATOR_VALUE = 2;
-  /**
-   * <code>SHUTDOWN_AFTER_TIMEOUT = 3;</code>
-   */
-  public static final int SHUTDOWN_AFTER_TIMEOUT_VALUE = 3;
-
-
-  public final int getNumber() {
-    if (this == UNRECOGNIZED) {
-      throw new java.lang.IllegalArgumentException(
-          "Can't get the number of an unknown enum value.");
-    }
-    return value;
-  }
-
-  /**
-   * @deprecated Use {@link #forNumber(int)} instead.
-   */
-  @java.lang.Deprecated
-  public static WorkerShutdownMode valueOf(int value) {
-    return forNumber(value);
-  }
-
-  public static WorkerShutdownMode forNumber(int value) {
-    switch (value) {
-      case 0: return DEFAULT;
-      case 1: return NOT_CONFIGURED;
-      case 2: return WAIT_FOR_COORDINATOR;
-      case 3: return SHUTDOWN_AFTER_TIMEOUT;
-      default: return null;
-    }
-  }
-
-  public static com.google.protobuf.Internal.EnumLiteMap<WorkerShutdownMode>
-      internalGetValueMap() {
-    return internalValueMap;
-  }
-  private static final com.google.protobuf.Internal.EnumLiteMap<
-      WorkerShutdownMode> internalValueMap =
-        new com.google.protobuf.Internal.EnumLiteMap<WorkerShutdownMode>() {
-          public WorkerShutdownMode findValueByNumber(int number) {
-            return WorkerShutdownMode.forNumber(number);
-          }
-        };
-
-  public final com.google.protobuf.Descriptors.EnumValueDescriptor
-      getValueDescriptor() {
-    return getDescriptor().getValues().get(ordinal());
-  }
-  public final com.google.protobuf.Descriptors.EnumDescriptor
-      getDescriptorForType() {
-    return getDescriptor();
-  }
-  public static final com.google.protobuf.Descriptors.EnumDescriptor
-      getDescriptor() {
-    return org.tensorflow.proto.util.EventProtos.getDescriptor().getEnumTypes().get(1);
-  }
-
-  private static final WorkerShutdownMode[] VALUES = values();
-
-  public static WorkerShutdownMode valueOf(
-      com.google.protobuf.Descriptors.EnumValueDescriptor desc) {
-    if (desc.getType() != getDescriptor()) {
-      throw new java.lang.IllegalArgumentException(
-        "EnumValueDescriptor is not for this type.");
-    }
-    if (desc.getIndex() == -1) {
-      return UNRECOGNIZED;
-    }
-    return VALUES[desc.getIndex()];
-  }
-
-  private final int value;
-
-  private WorkerShutdownMode(int value) {
-    this.value = value;
-  }
-
-  // @@protoc_insertion_point(enum_scope:tensorflow.WorkerShutdownMode)
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/AvailableDeviceInfo.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/AvailableDeviceInfo.java
deleted file mode 100644
index c80f9c4ff64..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/AvailableDeviceInfo.java
+++ /dev/null
@@ -1,966 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/util/test_log.proto
-
-package org.tensorflow.proto.util.testlog;
-
-/**
- * <pre>
- * Matches DeviceAttributes
- * </pre>
- *
- * Protobuf type {@code tensorflow.AvailableDeviceInfo}
- */
-public  final class AvailableDeviceInfo extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.AvailableDeviceInfo)
-    AvailableDeviceInfoOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use AvailableDeviceInfo.newBuilder() to construct.
-  private AvailableDeviceInfo(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private AvailableDeviceInfo() {
-    name_ = "";
-    type_ = "";
-    physicalDescription_ = "";
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new AvailableDeviceInfo();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private AvailableDeviceInfo(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 10: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            name_ = s;
-            break;
-          }
-          case 18: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            type_ = s;
-            break;
-          }
-          case 24: {
-
-            memoryLimit_ = input.readInt64();
-            break;
-          }
-          case 34: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            physicalDescription_ = s;
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_AvailableDeviceInfo_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_AvailableDeviceInfo_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.util.testlog.AvailableDeviceInfo.class, org.tensorflow.proto.util.testlog.AvailableDeviceInfo.Builder.class);
-  }
-
-  public static final int NAME_FIELD_NUMBER = 1;
-  private volatile java.lang.Object name_;
-  /**
-   * <pre>
-   * Device name.
-   * </pre>
-   *
-   * <code>string name = 1;</code>
-   */
-  public java.lang.String getName() {
-    java.lang.Object ref = name_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      name_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * Device name.
-   * </pre>
-   *
-   * <code>string name = 1;</code>
-   */
-  public com.google.protobuf.ByteString
-      getNameBytes() {
-    java.lang.Object ref = name_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      name_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int TYPE_FIELD_NUMBER = 2;
-  private volatile java.lang.Object type_;
-  /**
-   * <pre>
-   * Device type, e.g. 'CPU' or 'GPU'.
-   * </pre>
-   *
-   * <code>string type = 2;</code>
-   */
-  public java.lang.String getType() {
-    java.lang.Object ref = type_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      type_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * Device type, e.g. 'CPU' or 'GPU'.
-   * </pre>
-   *
-   * <code>string type = 2;</code>
-   */
-  public com.google.protobuf.ByteString
-      getTypeBytes() {
-    java.lang.Object ref = type_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      type_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int MEMORY_LIMIT_FIELD_NUMBER = 3;
-  private long memoryLimit_;
-  /**
-   * <pre>
-   * Memory capacity in bytes.
-   * </pre>
-   *
-   * <code>int64 memory_limit = 3;</code>
-   */
-  public long getMemoryLimit() {
-    return memoryLimit_;
-  }
-
-  public static final int PHYSICAL_DESCRIPTION_FIELD_NUMBER = 4;
-  private volatile java.lang.Object physicalDescription_;
-  /**
-   * <pre>
-   * The physical description of this device.
-   * </pre>
-   *
-   * <code>string physical_description = 4;</code>
-   */
-  public java.lang.String getPhysicalDescription() {
-    java.lang.Object ref = physicalDescription_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      physicalDescription_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * The physical description of this device.
-   * </pre>
-   *
-   * <code>string physical_description = 4;</code>
-   */
-  public com.google.protobuf.ByteString
-      getPhysicalDescriptionBytes() {
-    java.lang.Object ref = physicalDescription_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      physicalDescription_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    if (!getNameBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 1, name_);
-    }
-    if (!getTypeBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 2, type_);
-    }
-    if (memoryLimit_ != 0L) {
-      output.writeInt64(3, memoryLimit_);
-    }
-    if (!getPhysicalDescriptionBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 4, physicalDescription_);
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (!getNameBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, name_);
-    }
-    if (!getTypeBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(2, type_);
-    }
-    if (memoryLimit_ != 0L) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt64Size(3, memoryLimit_);
-    }
-    if (!getPhysicalDescriptionBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(4, physicalDescription_);
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.util.testlog.AvailableDeviceInfo)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.util.testlog.AvailableDeviceInfo other = (org.tensorflow.proto.util.testlog.AvailableDeviceInfo) obj;
-
-    if (!getName()
-        .equals(other.getName())) return false;
-    if (!getType()
-        .equals(other.getType())) return false;
-    if (getMemoryLimit()
-        != other.getMemoryLimit()) return false;
-    if (!getPhysicalDescription()
-        .equals(other.getPhysicalDescription())) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    hash = (37 * hash) + NAME_FIELD_NUMBER;
-    hash = (53 * hash) + getName().hashCode();
-    hash = (37 * hash) + TYPE_FIELD_NUMBER;
-    hash = (53 * hash) + getType().hashCode();
-    hash = (37 * hash) + MEMORY_LIMIT_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-        getMemoryLimit());
-    hash = (37 * hash) + PHYSICAL_DESCRIPTION_FIELD_NUMBER;
-    hash = (53 * hash) + getPhysicalDescription().hashCode();
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.util.testlog.AvailableDeviceInfo parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.testlog.AvailableDeviceInfo parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.testlog.AvailableDeviceInfo parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.testlog.AvailableDeviceInfo parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.testlog.AvailableDeviceInfo parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.testlog.AvailableDeviceInfo parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.testlog.AvailableDeviceInfo parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.testlog.AvailableDeviceInfo parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.testlog.AvailableDeviceInfo parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.testlog.AvailableDeviceInfo parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.testlog.AvailableDeviceInfo parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.testlog.AvailableDeviceInfo parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.util.testlog.AvailableDeviceInfo prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * <pre>
-   * Matches DeviceAttributes
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.AvailableDeviceInfo}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.AvailableDeviceInfo)
-      org.tensorflow.proto.util.testlog.AvailableDeviceInfoOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_AvailableDeviceInfo_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_AvailableDeviceInfo_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.util.testlog.AvailableDeviceInfo.class, org.tensorflow.proto.util.testlog.AvailableDeviceInfo.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.util.testlog.AvailableDeviceInfo.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      name_ = "";
-
-      type_ = "";
-
-      memoryLimit_ = 0L;
-
-      physicalDescription_ = "";
-
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_AvailableDeviceInfo_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.testlog.AvailableDeviceInfo getDefaultInstanceForType() {
-      return org.tensorflow.proto.util.testlog.AvailableDeviceInfo.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.testlog.AvailableDeviceInfo build() {
-      org.tensorflow.proto.util.testlog.AvailableDeviceInfo result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.testlog.AvailableDeviceInfo buildPartial() {
-      org.tensorflow.proto.util.testlog.AvailableDeviceInfo result = new org.tensorflow.proto.util.testlog.AvailableDeviceInfo(this);
-      result.name_ = name_;
-      result.type_ = type_;
-      result.memoryLimit_ = memoryLimit_;
-      result.physicalDescription_ = physicalDescription_;
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.util.testlog.AvailableDeviceInfo) {
-        return mergeFrom((org.tensorflow.proto.util.testlog.AvailableDeviceInfo)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.util.testlog.AvailableDeviceInfo other) {
-      if (other == org.tensorflow.proto.util.testlog.AvailableDeviceInfo.getDefaultInstance()) return this;
-      if (!other.getName().isEmpty()) {
-        name_ = other.name_;
-        onChanged();
-      }
-      if (!other.getType().isEmpty()) {
-        type_ = other.type_;
-        onChanged();
-      }
-      if (other.getMemoryLimit() != 0L) {
-        setMemoryLimit(other.getMemoryLimit());
-      }
-      if (!other.getPhysicalDescription().isEmpty()) {
-        physicalDescription_ = other.physicalDescription_;
-        onChanged();
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.util.testlog.AvailableDeviceInfo parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.util.testlog.AvailableDeviceInfo) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-
-    private java.lang.Object name_ = "";
-    /**
-     * <pre>
-     * Device name.
-     * </pre>
-     *
-     * <code>string name = 1;</code>
-     */
-    public java.lang.String getName() {
-      java.lang.Object ref = name_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        name_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Device name.
-     * </pre>
-     *
-     * <code>string name = 1;</code>
-     */
-    public com.google.protobuf.ByteString
-        getNameBytes() {
-      java.lang.Object ref = name_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        name_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Device name.
-     * </pre>
-     *
-     * <code>string name = 1;</code>
-     */
-    public Builder setName(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      name_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Device name.
-     * </pre>
-     *
-     * <code>string name = 1;</code>
-     */
-    public Builder clearName() {
-      
-      name_ = getDefaultInstance().getName();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Device name.
-     * </pre>
-     *
-     * <code>string name = 1;</code>
-     */
-    public Builder setNameBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      name_ = value;
-      onChanged();
-      return this;
-    }
-
-    private java.lang.Object type_ = "";
-    /**
-     * <pre>
-     * Device type, e.g. 'CPU' or 'GPU'.
-     * </pre>
-     *
-     * <code>string type = 2;</code>
-     */
-    public java.lang.String getType() {
-      java.lang.Object ref = type_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        type_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Device type, e.g. 'CPU' or 'GPU'.
-     * </pre>
-     *
-     * <code>string type = 2;</code>
-     */
-    public com.google.protobuf.ByteString
-        getTypeBytes() {
-      java.lang.Object ref = type_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        type_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Device type, e.g. 'CPU' or 'GPU'.
-     * </pre>
-     *
-     * <code>string type = 2;</code>
-     */
-    public Builder setType(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      type_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Device type, e.g. 'CPU' or 'GPU'.
-     * </pre>
-     *
-     * <code>string type = 2;</code>
-     */
-    public Builder clearType() {
-      
-      type_ = getDefaultInstance().getType();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Device type, e.g. 'CPU' or 'GPU'.
-     * </pre>
-     *
-     * <code>string type = 2;</code>
-     */
-    public Builder setTypeBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      type_ = value;
-      onChanged();
-      return this;
-    }
-
-    private long memoryLimit_ ;
-    /**
-     * <pre>
-     * Memory capacity in bytes.
-     * </pre>
-     *
-     * <code>int64 memory_limit = 3;</code>
-     */
-    public long getMemoryLimit() {
-      return memoryLimit_;
-    }
-    /**
-     * <pre>
-     * Memory capacity in bytes.
-     * </pre>
-     *
-     * <code>int64 memory_limit = 3;</code>
-     */
-    public Builder setMemoryLimit(long value) {
-      
-      memoryLimit_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Memory capacity in bytes.
-     * </pre>
-     *
-     * <code>int64 memory_limit = 3;</code>
-     */
-    public Builder clearMemoryLimit() {
-      
-      memoryLimit_ = 0L;
-      onChanged();
-      return this;
-    }
-
-    private java.lang.Object physicalDescription_ = "";
-    /**
-     * <pre>
-     * The physical description of this device.
-     * </pre>
-     *
-     * <code>string physical_description = 4;</code>
-     */
-    public java.lang.String getPhysicalDescription() {
-      java.lang.Object ref = physicalDescription_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        physicalDescription_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * The physical description of this device.
-     * </pre>
-     *
-     * <code>string physical_description = 4;</code>
-     */
-    public com.google.protobuf.ByteString
-        getPhysicalDescriptionBytes() {
-      java.lang.Object ref = physicalDescription_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        physicalDescription_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * The physical description of this device.
-     * </pre>
-     *
-     * <code>string physical_description = 4;</code>
-     */
-    public Builder setPhysicalDescription(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      physicalDescription_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * The physical description of this device.
-     * </pre>
-     *
-     * <code>string physical_description = 4;</code>
-     */
-    public Builder clearPhysicalDescription() {
-      
-      physicalDescription_ = getDefaultInstance().getPhysicalDescription();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * The physical description of this device.
-     * </pre>
-     *
-     * <code>string physical_description = 4;</code>
-     */
-    public Builder setPhysicalDescriptionBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      physicalDescription_ = value;
-      onChanged();
-      return this;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.AvailableDeviceInfo)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.AvailableDeviceInfo)
-  private static final org.tensorflow.proto.util.testlog.AvailableDeviceInfo DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.util.testlog.AvailableDeviceInfo();
-  }
-
-  public static org.tensorflow.proto.util.testlog.AvailableDeviceInfo getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<AvailableDeviceInfo>
-      PARSER = new com.google.protobuf.AbstractParser<AvailableDeviceInfo>() {
-    @java.lang.Override
-    public AvailableDeviceInfo parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new AvailableDeviceInfo(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<AvailableDeviceInfo> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<AvailableDeviceInfo> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.util.testlog.AvailableDeviceInfo getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/AvailableDeviceInfoOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/AvailableDeviceInfoOrBuilder.java
deleted file mode 100644
index 138fcf086ef..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/AvailableDeviceInfoOrBuilder.java
+++ /dev/null
@@ -1,72 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/util/test_log.proto
-
-package org.tensorflow.proto.util.testlog;
-
-public interface AvailableDeviceInfoOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.AvailableDeviceInfo)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <pre>
-   * Device name.
-   * </pre>
-   *
-   * <code>string name = 1;</code>
-   */
-  java.lang.String getName();
-  /**
-   * <pre>
-   * Device name.
-   * </pre>
-   *
-   * <code>string name = 1;</code>
-   */
-  com.google.protobuf.ByteString
-      getNameBytes();
-
-  /**
-   * <pre>
-   * Device type, e.g. 'CPU' or 'GPU'.
-   * </pre>
-   *
-   * <code>string type = 2;</code>
-   */
-  java.lang.String getType();
-  /**
-   * <pre>
-   * Device type, e.g. 'CPU' or 'GPU'.
-   * </pre>
-   *
-   * <code>string type = 2;</code>
-   */
-  com.google.protobuf.ByteString
-      getTypeBytes();
-
-  /**
-   * <pre>
-   * Memory capacity in bytes.
-   * </pre>
-   *
-   * <code>int64 memory_limit = 3;</code>
-   */
-  long getMemoryLimit();
-
-  /**
-   * <pre>
-   * The physical description of this device.
-   * </pre>
-   *
-   * <code>string physical_description = 4;</code>
-   */
-  java.lang.String getPhysicalDescription();
-  /**
-   * <pre>
-   * The physical description of this device.
-   * </pre>
-   *
-   * <code>string physical_description = 4;</code>
-   */
-  com.google.protobuf.ByteString
-      getPhysicalDescriptionBytes();
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/BenchmarkEntries.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/BenchmarkEntries.java
deleted file mode 100644
index 27fe4cdcc81..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/BenchmarkEntries.java
+++ /dev/null
@@ -1,765 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/util/test_log.proto
-
-package org.tensorflow.proto.util.testlog;
-
-/**
- * Protobuf type {@code tensorflow.BenchmarkEntries}
- */
-public  final class BenchmarkEntries extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.BenchmarkEntries)
-    BenchmarkEntriesOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use BenchmarkEntries.newBuilder() to construct.
-  private BenchmarkEntries(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private BenchmarkEntries() {
-    entry_ = java.util.Collections.emptyList();
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new BenchmarkEntries();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private BenchmarkEntries(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    int mutable_bitField0_ = 0;
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 10: {
-            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
-              entry_ = new java.util.ArrayList<org.tensorflow.proto.util.testlog.BenchmarkEntry>();
-              mutable_bitField0_ |= 0x00000001;
-            }
-            entry_.add(
-                input.readMessage(org.tensorflow.proto.util.testlog.BenchmarkEntry.parser(), extensionRegistry));
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      if (((mutable_bitField0_ & 0x00000001) != 0)) {
-        entry_ = java.util.Collections.unmodifiableList(entry_);
-      }
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_BenchmarkEntries_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_BenchmarkEntries_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.util.testlog.BenchmarkEntries.class, org.tensorflow.proto.util.testlog.BenchmarkEntries.Builder.class);
-  }
-
-  public static final int ENTRY_FIELD_NUMBER = 1;
-  private java.util.List<org.tensorflow.proto.util.testlog.BenchmarkEntry> entry_;
-  /**
-   * <code>repeated .tensorflow.BenchmarkEntry entry = 1;</code>
-   */
-  public java.util.List<org.tensorflow.proto.util.testlog.BenchmarkEntry> getEntryList() {
-    return entry_;
-  }
-  /**
-   * <code>repeated .tensorflow.BenchmarkEntry entry = 1;</code>
-   */
-  public java.util.List<? extends org.tensorflow.proto.util.testlog.BenchmarkEntryOrBuilder> 
-      getEntryOrBuilderList() {
-    return entry_;
-  }
-  /**
-   * <code>repeated .tensorflow.BenchmarkEntry entry = 1;</code>
-   */
-  public int getEntryCount() {
-    return entry_.size();
-  }
-  /**
-   * <code>repeated .tensorflow.BenchmarkEntry entry = 1;</code>
-   */
-  public org.tensorflow.proto.util.testlog.BenchmarkEntry getEntry(int index) {
-    return entry_.get(index);
-  }
-  /**
-   * <code>repeated .tensorflow.BenchmarkEntry entry = 1;</code>
-   */
-  public org.tensorflow.proto.util.testlog.BenchmarkEntryOrBuilder getEntryOrBuilder(
-      int index) {
-    return entry_.get(index);
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    for (int i = 0; i < entry_.size(); i++) {
-      output.writeMessage(1, entry_.get(i));
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    for (int i = 0; i < entry_.size(); i++) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(1, entry_.get(i));
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.util.testlog.BenchmarkEntries)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.util.testlog.BenchmarkEntries other = (org.tensorflow.proto.util.testlog.BenchmarkEntries) obj;
-
-    if (!getEntryList()
-        .equals(other.getEntryList())) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    if (getEntryCount() > 0) {
-      hash = (37 * hash) + ENTRY_FIELD_NUMBER;
-      hash = (53 * hash) + getEntryList().hashCode();
-    }
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.util.testlog.BenchmarkEntries parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.testlog.BenchmarkEntries parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.testlog.BenchmarkEntries parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.testlog.BenchmarkEntries parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.testlog.BenchmarkEntries parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.testlog.BenchmarkEntries parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.testlog.BenchmarkEntries parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.testlog.BenchmarkEntries parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.testlog.BenchmarkEntries parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.testlog.BenchmarkEntries parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.testlog.BenchmarkEntries parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.testlog.BenchmarkEntries parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.util.testlog.BenchmarkEntries prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * Protobuf type {@code tensorflow.BenchmarkEntries}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.BenchmarkEntries)
-      org.tensorflow.proto.util.testlog.BenchmarkEntriesOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_BenchmarkEntries_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_BenchmarkEntries_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.util.testlog.BenchmarkEntries.class, org.tensorflow.proto.util.testlog.BenchmarkEntries.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.util.testlog.BenchmarkEntries.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-        getEntryFieldBuilder();
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      if (entryBuilder_ == null) {
-        entry_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000001);
-      } else {
-        entryBuilder_.clear();
-      }
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_BenchmarkEntries_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.testlog.BenchmarkEntries getDefaultInstanceForType() {
-      return org.tensorflow.proto.util.testlog.BenchmarkEntries.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.testlog.BenchmarkEntries build() {
-      org.tensorflow.proto.util.testlog.BenchmarkEntries result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.testlog.BenchmarkEntries buildPartial() {
-      org.tensorflow.proto.util.testlog.BenchmarkEntries result = new org.tensorflow.proto.util.testlog.BenchmarkEntries(this);
-      int from_bitField0_ = bitField0_;
-      if (entryBuilder_ == null) {
-        if (((bitField0_ & 0x00000001) != 0)) {
-          entry_ = java.util.Collections.unmodifiableList(entry_);
-          bitField0_ = (bitField0_ & ~0x00000001);
-        }
-        result.entry_ = entry_;
-      } else {
-        result.entry_ = entryBuilder_.build();
-      }
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.util.testlog.BenchmarkEntries) {
-        return mergeFrom((org.tensorflow.proto.util.testlog.BenchmarkEntries)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.util.testlog.BenchmarkEntries other) {
-      if (other == org.tensorflow.proto.util.testlog.BenchmarkEntries.getDefaultInstance()) return this;
-      if (entryBuilder_ == null) {
-        if (!other.entry_.isEmpty()) {
-          if (entry_.isEmpty()) {
-            entry_ = other.entry_;
-            bitField0_ = (bitField0_ & ~0x00000001);
-          } else {
-            ensureEntryIsMutable();
-            entry_.addAll(other.entry_);
-          }
-          onChanged();
-        }
-      } else {
-        if (!other.entry_.isEmpty()) {
-          if (entryBuilder_.isEmpty()) {
-            entryBuilder_.dispose();
-            entryBuilder_ = null;
-            entry_ = other.entry_;
-            bitField0_ = (bitField0_ & ~0x00000001);
-            entryBuilder_ = 
-              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
-                 getEntryFieldBuilder() : null;
-          } else {
-            entryBuilder_.addAllMessages(other.entry_);
-          }
-        }
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.util.testlog.BenchmarkEntries parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.util.testlog.BenchmarkEntries) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-    private int bitField0_;
-
-    private java.util.List<org.tensorflow.proto.util.testlog.BenchmarkEntry> entry_ =
-      java.util.Collections.emptyList();
-    private void ensureEntryIsMutable() {
-      if (!((bitField0_ & 0x00000001) != 0)) {
-        entry_ = new java.util.ArrayList<org.tensorflow.proto.util.testlog.BenchmarkEntry>(entry_);
-        bitField0_ |= 0x00000001;
-       }
-    }
-
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.util.testlog.BenchmarkEntry, org.tensorflow.proto.util.testlog.BenchmarkEntry.Builder, org.tensorflow.proto.util.testlog.BenchmarkEntryOrBuilder> entryBuilder_;
-
-    /**
-     * <code>repeated .tensorflow.BenchmarkEntry entry = 1;</code>
-     */
-    public java.util.List<org.tensorflow.proto.util.testlog.BenchmarkEntry> getEntryList() {
-      if (entryBuilder_ == null) {
-        return java.util.Collections.unmodifiableList(entry_);
-      } else {
-        return entryBuilder_.getMessageList();
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.BenchmarkEntry entry = 1;</code>
-     */
-    public int getEntryCount() {
-      if (entryBuilder_ == null) {
-        return entry_.size();
-      } else {
-        return entryBuilder_.getCount();
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.BenchmarkEntry entry = 1;</code>
-     */
-    public org.tensorflow.proto.util.testlog.BenchmarkEntry getEntry(int index) {
-      if (entryBuilder_ == null) {
-        return entry_.get(index);
-      } else {
-        return entryBuilder_.getMessage(index);
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.BenchmarkEntry entry = 1;</code>
-     */
-    public Builder setEntry(
-        int index, org.tensorflow.proto.util.testlog.BenchmarkEntry value) {
-      if (entryBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureEntryIsMutable();
-        entry_.set(index, value);
-        onChanged();
-      } else {
-        entryBuilder_.setMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.BenchmarkEntry entry = 1;</code>
-     */
-    public Builder setEntry(
-        int index, org.tensorflow.proto.util.testlog.BenchmarkEntry.Builder builderForValue) {
-      if (entryBuilder_ == null) {
-        ensureEntryIsMutable();
-        entry_.set(index, builderForValue.build());
-        onChanged();
-      } else {
-        entryBuilder_.setMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.BenchmarkEntry entry = 1;</code>
-     */
-    public Builder addEntry(org.tensorflow.proto.util.testlog.BenchmarkEntry value) {
-      if (entryBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureEntryIsMutable();
-        entry_.add(value);
-        onChanged();
-      } else {
-        entryBuilder_.addMessage(value);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.BenchmarkEntry entry = 1;</code>
-     */
-    public Builder addEntry(
-        int index, org.tensorflow.proto.util.testlog.BenchmarkEntry value) {
-      if (entryBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureEntryIsMutable();
-        entry_.add(index, value);
-        onChanged();
-      } else {
-        entryBuilder_.addMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.BenchmarkEntry entry = 1;</code>
-     */
-    public Builder addEntry(
-        org.tensorflow.proto.util.testlog.BenchmarkEntry.Builder builderForValue) {
-      if (entryBuilder_ == null) {
-        ensureEntryIsMutable();
-        entry_.add(builderForValue.build());
-        onChanged();
-      } else {
-        entryBuilder_.addMessage(builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.BenchmarkEntry entry = 1;</code>
-     */
-    public Builder addEntry(
-        int index, org.tensorflow.proto.util.testlog.BenchmarkEntry.Builder builderForValue) {
-      if (entryBuilder_ == null) {
-        ensureEntryIsMutable();
-        entry_.add(index, builderForValue.build());
-        onChanged();
-      } else {
-        entryBuilder_.addMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.BenchmarkEntry entry = 1;</code>
-     */
-    public Builder addAllEntry(
-        java.lang.Iterable<? extends org.tensorflow.proto.util.testlog.BenchmarkEntry> values) {
-      if (entryBuilder_ == null) {
-        ensureEntryIsMutable();
-        com.google.protobuf.AbstractMessageLite.Builder.addAll(
-            values, entry_);
-        onChanged();
-      } else {
-        entryBuilder_.addAllMessages(values);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.BenchmarkEntry entry = 1;</code>
-     */
-    public Builder clearEntry() {
-      if (entryBuilder_ == null) {
-        entry_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000001);
-        onChanged();
-      } else {
-        entryBuilder_.clear();
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.BenchmarkEntry entry = 1;</code>
-     */
-    public Builder removeEntry(int index) {
-      if (entryBuilder_ == null) {
-        ensureEntryIsMutable();
-        entry_.remove(index);
-        onChanged();
-      } else {
-        entryBuilder_.remove(index);
-      }
-      return this;
-    }
-    /**
-     * <code>repeated .tensorflow.BenchmarkEntry entry = 1;</code>
-     */
-    public org.tensorflow.proto.util.testlog.BenchmarkEntry.Builder getEntryBuilder(
-        int index) {
-      return getEntryFieldBuilder().getBuilder(index);
-    }
-    /**
-     * <code>repeated .tensorflow.BenchmarkEntry entry = 1;</code>
-     */
-    public org.tensorflow.proto.util.testlog.BenchmarkEntryOrBuilder getEntryOrBuilder(
-        int index) {
-      if (entryBuilder_ == null) {
-        return entry_.get(index);  } else {
-        return entryBuilder_.getMessageOrBuilder(index);
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.BenchmarkEntry entry = 1;</code>
-     */
-    public java.util.List<? extends org.tensorflow.proto.util.testlog.BenchmarkEntryOrBuilder> 
-         getEntryOrBuilderList() {
-      if (entryBuilder_ != null) {
-        return entryBuilder_.getMessageOrBuilderList();
-      } else {
-        return java.util.Collections.unmodifiableList(entry_);
-      }
-    }
-    /**
-     * <code>repeated .tensorflow.BenchmarkEntry entry = 1;</code>
-     */
-    public org.tensorflow.proto.util.testlog.BenchmarkEntry.Builder addEntryBuilder() {
-      return getEntryFieldBuilder().addBuilder(
-          org.tensorflow.proto.util.testlog.BenchmarkEntry.getDefaultInstance());
-    }
-    /**
-     * <code>repeated .tensorflow.BenchmarkEntry entry = 1;</code>
-     */
-    public org.tensorflow.proto.util.testlog.BenchmarkEntry.Builder addEntryBuilder(
-        int index) {
-      return getEntryFieldBuilder().addBuilder(
-          index, org.tensorflow.proto.util.testlog.BenchmarkEntry.getDefaultInstance());
-    }
-    /**
-     * <code>repeated .tensorflow.BenchmarkEntry entry = 1;</code>
-     */
-    public java.util.List<org.tensorflow.proto.util.testlog.BenchmarkEntry.Builder> 
-         getEntryBuilderList() {
-      return getEntryFieldBuilder().getBuilderList();
-    }
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.util.testlog.BenchmarkEntry, org.tensorflow.proto.util.testlog.BenchmarkEntry.Builder, org.tensorflow.proto.util.testlog.BenchmarkEntryOrBuilder> 
-        getEntryFieldBuilder() {
-      if (entryBuilder_ == null) {
-        entryBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
-            org.tensorflow.proto.util.testlog.BenchmarkEntry, org.tensorflow.proto.util.testlog.BenchmarkEntry.Builder, org.tensorflow.proto.util.testlog.BenchmarkEntryOrBuilder>(
-                entry_,
-                ((bitField0_ & 0x00000001) != 0),
-                getParentForChildren(),
-                isClean());
-        entry_ = null;
-      }
-      return entryBuilder_;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.BenchmarkEntries)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.BenchmarkEntries)
-  private static final org.tensorflow.proto.util.testlog.BenchmarkEntries DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.util.testlog.BenchmarkEntries();
-  }
-
-  public static org.tensorflow.proto.util.testlog.BenchmarkEntries getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<BenchmarkEntries>
-      PARSER = new com.google.protobuf.AbstractParser<BenchmarkEntries>() {
-    @java.lang.Override
-    public BenchmarkEntries parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new BenchmarkEntries(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<BenchmarkEntries> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<BenchmarkEntries> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.util.testlog.BenchmarkEntries getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/BenchmarkEntriesOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/BenchmarkEntriesOrBuilder.java
deleted file mode 100644
index 25a3905dfdf..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/BenchmarkEntriesOrBuilder.java
+++ /dev/null
@@ -1,33 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/util/test_log.proto
-
-package org.tensorflow.proto.util.testlog;
-
-public interface BenchmarkEntriesOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.BenchmarkEntries)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <code>repeated .tensorflow.BenchmarkEntry entry = 1;</code>
-   */
-  java.util.List<org.tensorflow.proto.util.testlog.BenchmarkEntry> 
-      getEntryList();
-  /**
-   * <code>repeated .tensorflow.BenchmarkEntry entry = 1;</code>
-   */
-  org.tensorflow.proto.util.testlog.BenchmarkEntry getEntry(int index);
-  /**
-   * <code>repeated .tensorflow.BenchmarkEntry entry = 1;</code>
-   */
-  int getEntryCount();
-  /**
-   * <code>repeated .tensorflow.BenchmarkEntry entry = 1;</code>
-   */
-  java.util.List<? extends org.tensorflow.proto.util.testlog.BenchmarkEntryOrBuilder> 
-      getEntryOrBuilderList();
-  /**
-   * <code>repeated .tensorflow.BenchmarkEntry entry = 1;</code>
-   */
-  org.tensorflow.proto.util.testlog.BenchmarkEntryOrBuilder getEntryOrBuilder(
-      int index);
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/BenchmarkEntry.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/BenchmarkEntry.java
deleted file mode 100644
index b83843aa744..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/BenchmarkEntry.java
+++ /dev/null
@@ -1,1676 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/util/test_log.proto
-
-package org.tensorflow.proto.util.testlog;
-
-/**
- * <pre>
- * Each unit test or benchmark in a test or benchmark run provides
- * some set of information.  Here we provide some reasonable keys
- * one would expect to see, with optional key/value pairs for things
- * we haven't considered.
- * This BenchmarkEntry should be emitted by each unit test or benchmark
- * reporter.
- * </pre>
- *
- * Protobuf type {@code tensorflow.BenchmarkEntry}
- */
-public  final class BenchmarkEntry extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.BenchmarkEntry)
-    BenchmarkEntryOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use BenchmarkEntry.newBuilder() to construct.
-  private BenchmarkEntry(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private BenchmarkEntry() {
-    name_ = "";
-    metrics_ = java.util.Collections.emptyList();
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new BenchmarkEntry();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private BenchmarkEntry(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    int mutable_bitField0_ = 0;
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 10: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            name_ = s;
-            break;
-          }
-          case 16: {
-
-            iters_ = input.readInt64();
-            break;
-          }
-          case 25: {
-
-            cpuTime_ = input.readDouble();
-            break;
-          }
-          case 33: {
-
-            wallTime_ = input.readDouble();
-            break;
-          }
-          case 41: {
-
-            throughput_ = input.readDouble();
-            break;
-          }
-          case 50: {
-            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
-              extras_ = com.google.protobuf.MapField.newMapField(
-                  ExtrasDefaultEntryHolder.defaultEntry);
-              mutable_bitField0_ |= 0x00000001;
-            }
-            com.google.protobuf.MapEntry<java.lang.String, org.tensorflow.proto.util.testlog.EntryValue>
-            extras__ = input.readMessage(
-                ExtrasDefaultEntryHolder.defaultEntry.getParserForType(), extensionRegistry);
-            extras_.getMutableMap().put(
-                extras__.getKey(), extras__.getValue());
-            break;
-          }
-          case 58: {
-            if (!((mutable_bitField0_ & 0x00000002) != 0)) {
-              metrics_ = new java.util.ArrayList<org.tensorflow.proto.util.testlog.MetricEntry>();
-              mutable_bitField0_ |= 0x00000002;
-            }
-            metrics_.add(
-                input.readMessage(org.tensorflow.proto.util.testlog.MetricEntry.parser(), extensionRegistry));
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      if (((mutable_bitField0_ & 0x00000002) != 0)) {
-        metrics_ = java.util.Collections.unmodifiableList(metrics_);
-      }
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_BenchmarkEntry_descriptor;
-  }
-
-  @SuppressWarnings({"rawtypes"})
-  @java.lang.Override
-  protected com.google.protobuf.MapField internalGetMapField(
-      int number) {
-    switch (number) {
-      case 6:
-        return internalGetExtras();
-      default:
-        throw new RuntimeException(
-            "Invalid map field number: " + number);
-    }
-  }
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_BenchmarkEntry_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.util.testlog.BenchmarkEntry.class, org.tensorflow.proto.util.testlog.BenchmarkEntry.Builder.class);
-  }
-
-  public static final int NAME_FIELD_NUMBER = 1;
-  private volatile java.lang.Object name_;
-  /**
-   * <pre>
-   * The name of the specific benchmark or test
-   * (e.g. BM_AdjustContrast_gpu_B_W_H)
-   * </pre>
-   *
-   * <code>string name = 1;</code>
-   */
-  public java.lang.String getName() {
-    java.lang.Object ref = name_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      name_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * The name of the specific benchmark or test
-   * (e.g. BM_AdjustContrast_gpu_B_W_H)
-   * </pre>
-   *
-   * <code>string name = 1;</code>
-   */
-  public com.google.protobuf.ByteString
-      getNameBytes() {
-    java.lang.Object ref = name_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      name_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int ITERS_FIELD_NUMBER = 2;
-  private long iters_;
-  /**
-   * <pre>
-   * If a benchmark, how many iterations it was run for
-   * </pre>
-   *
-   * <code>int64 iters = 2;</code>
-   */
-  public long getIters() {
-    return iters_;
-  }
-
-  public static final int CPU_TIME_FIELD_NUMBER = 3;
-  private double cpuTime_;
-  /**
-   * <pre>
-   * Total cpu time used for all iterations (in seconds)
-   * </pre>
-   *
-   * <code>double cpu_time = 3;</code>
-   */
-  public double getCpuTime() {
-    return cpuTime_;
-  }
-
-  public static final int WALL_TIME_FIELD_NUMBER = 4;
-  private double wallTime_;
-  /**
-   * <pre>
-   * Total wall time used for all iterations (in seconds)
-   * </pre>
-   *
-   * <code>double wall_time = 4;</code>
-   */
-  public double getWallTime() {
-    return wallTime_;
-  }
-
-  public static final int THROUGHPUT_FIELD_NUMBER = 5;
-  private double throughput_;
-  /**
-   * <pre>
-   * Throughput (in MB/s)
-   * </pre>
-   *
-   * <code>double throughput = 5;</code>
-   */
-  public double getThroughput() {
-    return throughput_;
-  }
-
-  public static final int EXTRAS_FIELD_NUMBER = 6;
-  private static final class ExtrasDefaultEntryHolder {
-    static final com.google.protobuf.MapEntry<
-        java.lang.String, org.tensorflow.proto.util.testlog.EntryValue> defaultEntry =
-            com.google.protobuf.MapEntry
-            .<java.lang.String, org.tensorflow.proto.util.testlog.EntryValue>newDefaultInstance(
-                org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_BenchmarkEntry_ExtrasEntry_descriptor, 
-                com.google.protobuf.WireFormat.FieldType.STRING,
-                "",
-                com.google.protobuf.WireFormat.FieldType.MESSAGE,
-                org.tensorflow.proto.util.testlog.EntryValue.getDefaultInstance());
-  }
-  private com.google.protobuf.MapField<
-      java.lang.String, org.tensorflow.proto.util.testlog.EntryValue> extras_;
-  private com.google.protobuf.MapField<java.lang.String, org.tensorflow.proto.util.testlog.EntryValue>
-  internalGetExtras() {
-    if (extras_ == null) {
-      return com.google.protobuf.MapField.emptyMapField(
-          ExtrasDefaultEntryHolder.defaultEntry);
-    }
-    return extras_;
-  }
-
-  public int getExtrasCount() {
-    return internalGetExtras().getMap().size();
-  }
-  /**
-   * <pre>
-   * Generic map from result key to value.
-   * </pre>
-   *
-   * <code>map&lt;string, .tensorflow.EntryValue&gt; extras = 6;</code>
-   */
-
-  public boolean containsExtras(
-      java.lang.String key) {
-    if (key == null) { throw new java.lang.NullPointerException(); }
-    return internalGetExtras().getMap().containsKey(key);
-  }
-  /**
-   * Use {@link #getExtrasMap()} instead.
-   */
-  @java.lang.Deprecated
-  public java.util.Map<java.lang.String, org.tensorflow.proto.util.testlog.EntryValue> getExtras() {
-    return getExtrasMap();
-  }
-  /**
-   * <pre>
-   * Generic map from result key to value.
-   * </pre>
-   *
-   * <code>map&lt;string, .tensorflow.EntryValue&gt; extras = 6;</code>
-   */
-
-  public java.util.Map<java.lang.String, org.tensorflow.proto.util.testlog.EntryValue> getExtrasMap() {
-    return internalGetExtras().getMap();
-  }
-  /**
-   * <pre>
-   * Generic map from result key to value.
-   * </pre>
-   *
-   * <code>map&lt;string, .tensorflow.EntryValue&gt; extras = 6;</code>
-   */
-
-  public org.tensorflow.proto.util.testlog.EntryValue getExtrasOrDefault(
-      java.lang.String key,
-      org.tensorflow.proto.util.testlog.EntryValue defaultValue) {
-    if (key == null) { throw new java.lang.NullPointerException(); }
-    java.util.Map<java.lang.String, org.tensorflow.proto.util.testlog.EntryValue> map =
-        internalGetExtras().getMap();
-    return map.containsKey(key) ? map.get(key) : defaultValue;
-  }
-  /**
-   * <pre>
-   * Generic map from result key to value.
-   * </pre>
-   *
-   * <code>map&lt;string, .tensorflow.EntryValue&gt; extras = 6;</code>
-   */
-
-  public org.tensorflow.proto.util.testlog.EntryValue getExtrasOrThrow(
-      java.lang.String key) {
-    if (key == null) { throw new java.lang.NullPointerException(); }
-    java.util.Map<java.lang.String, org.tensorflow.proto.util.testlog.EntryValue> map =
-        internalGetExtras().getMap();
-    if (!map.containsKey(key)) {
-      throw new java.lang.IllegalArgumentException();
-    }
-    return map.get(key);
-  }
-
-  public static final int METRICS_FIELD_NUMBER = 7;
-  private java.util.List<org.tensorflow.proto.util.testlog.MetricEntry> metrics_;
-  /**
-   * <pre>
-   * Metric name, value and expected range. This can include accuracy metrics
-   * typically used to determine whether the accuracy test has passed
-   * </pre>
-   *
-   * <code>repeated .tensorflow.MetricEntry metrics = 7;</code>
-   */
-  public java.util.List<org.tensorflow.proto.util.testlog.MetricEntry> getMetricsList() {
-    return metrics_;
-  }
-  /**
-   * <pre>
-   * Metric name, value and expected range. This can include accuracy metrics
-   * typically used to determine whether the accuracy test has passed
-   * </pre>
-   *
-   * <code>repeated .tensorflow.MetricEntry metrics = 7;</code>
-   */
-  public java.util.List<? extends org.tensorflow.proto.util.testlog.MetricEntryOrBuilder> 
-      getMetricsOrBuilderList() {
-    return metrics_;
-  }
-  /**
-   * <pre>
-   * Metric name, value and expected range. This can include accuracy metrics
-   * typically used to determine whether the accuracy test has passed
-   * </pre>
-   *
-   * <code>repeated .tensorflow.MetricEntry metrics = 7;</code>
-   */
-  public int getMetricsCount() {
-    return metrics_.size();
-  }
-  /**
-   * <pre>
-   * Metric name, value and expected range. This can include accuracy metrics
-   * typically used to determine whether the accuracy test has passed
-   * </pre>
-   *
-   * <code>repeated .tensorflow.MetricEntry metrics = 7;</code>
-   */
-  public org.tensorflow.proto.util.testlog.MetricEntry getMetrics(int index) {
-    return metrics_.get(index);
-  }
-  /**
-   * <pre>
-   * Metric name, value and expected range. This can include accuracy metrics
-   * typically used to determine whether the accuracy test has passed
-   * </pre>
-   *
-   * <code>repeated .tensorflow.MetricEntry metrics = 7;</code>
-   */
-  public org.tensorflow.proto.util.testlog.MetricEntryOrBuilder getMetricsOrBuilder(
-      int index) {
-    return metrics_.get(index);
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    if (!getNameBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 1, name_);
-    }
-    if (iters_ != 0L) {
-      output.writeInt64(2, iters_);
-    }
-    if (cpuTime_ != 0D) {
-      output.writeDouble(3, cpuTime_);
-    }
-    if (wallTime_ != 0D) {
-      output.writeDouble(4, wallTime_);
-    }
-    if (throughput_ != 0D) {
-      output.writeDouble(5, throughput_);
-    }
-    com.google.protobuf.GeneratedMessageV3
-      .serializeStringMapTo(
-        output,
-        internalGetExtras(),
-        ExtrasDefaultEntryHolder.defaultEntry,
-        6);
-    for (int i = 0; i < metrics_.size(); i++) {
-      output.writeMessage(7, metrics_.get(i));
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (!getNameBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, name_);
-    }
-    if (iters_ != 0L) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt64Size(2, iters_);
-    }
-    if (cpuTime_ != 0D) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeDoubleSize(3, cpuTime_);
-    }
-    if (wallTime_ != 0D) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeDoubleSize(4, wallTime_);
-    }
-    if (throughput_ != 0D) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeDoubleSize(5, throughput_);
-    }
-    for (java.util.Map.Entry<java.lang.String, org.tensorflow.proto.util.testlog.EntryValue> entry
-         : internalGetExtras().getMap().entrySet()) {
-      com.google.protobuf.MapEntry<java.lang.String, org.tensorflow.proto.util.testlog.EntryValue>
-      extras__ = ExtrasDefaultEntryHolder.defaultEntry.newBuilderForType()
-          .setKey(entry.getKey())
-          .setValue(entry.getValue())
-          .build();
-      size += com.google.protobuf.CodedOutputStream
-          .computeMessageSize(6, extras__);
-    }
-    for (int i = 0; i < metrics_.size(); i++) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(7, metrics_.get(i));
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.util.testlog.BenchmarkEntry)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.util.testlog.BenchmarkEntry other = (org.tensorflow.proto.util.testlog.BenchmarkEntry) obj;
-
-    if (!getName()
-        .equals(other.getName())) return false;
-    if (getIters()
-        != other.getIters()) return false;
-    if (java.lang.Double.doubleToLongBits(getCpuTime())
-        != java.lang.Double.doubleToLongBits(
-            other.getCpuTime())) return false;
-    if (java.lang.Double.doubleToLongBits(getWallTime())
-        != java.lang.Double.doubleToLongBits(
-            other.getWallTime())) return false;
-    if (java.lang.Double.doubleToLongBits(getThroughput())
-        != java.lang.Double.doubleToLongBits(
-            other.getThroughput())) return false;
-    if (!internalGetExtras().equals(
-        other.internalGetExtras())) return false;
-    if (!getMetricsList()
-        .equals(other.getMetricsList())) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    hash = (37 * hash) + NAME_FIELD_NUMBER;
-    hash = (53 * hash) + getName().hashCode();
-    hash = (37 * hash) + ITERS_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-        getIters());
-    hash = (37 * hash) + CPU_TIME_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-        java.lang.Double.doubleToLongBits(getCpuTime()));
-    hash = (37 * hash) + WALL_TIME_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-        java.lang.Double.doubleToLongBits(getWallTime()));
-    hash = (37 * hash) + THROUGHPUT_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-        java.lang.Double.doubleToLongBits(getThroughput()));
-    if (!internalGetExtras().getMap().isEmpty()) {
-      hash = (37 * hash) + EXTRAS_FIELD_NUMBER;
-      hash = (53 * hash) + internalGetExtras().hashCode();
-    }
-    if (getMetricsCount() > 0) {
-      hash = (37 * hash) + METRICS_FIELD_NUMBER;
-      hash = (53 * hash) + getMetricsList().hashCode();
-    }
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.util.testlog.BenchmarkEntry parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.testlog.BenchmarkEntry parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.testlog.BenchmarkEntry parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.testlog.BenchmarkEntry parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.testlog.BenchmarkEntry parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.testlog.BenchmarkEntry parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.testlog.BenchmarkEntry parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.testlog.BenchmarkEntry parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.testlog.BenchmarkEntry parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.testlog.BenchmarkEntry parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.testlog.BenchmarkEntry parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.testlog.BenchmarkEntry parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.util.testlog.BenchmarkEntry prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * <pre>
-   * Each unit test or benchmark in a test or benchmark run provides
-   * some set of information.  Here we provide some reasonable keys
-   * one would expect to see, with optional key/value pairs for things
-   * we haven't considered.
-   * This BenchmarkEntry should be emitted by each unit test or benchmark
-   * reporter.
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.BenchmarkEntry}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.BenchmarkEntry)
-      org.tensorflow.proto.util.testlog.BenchmarkEntryOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_BenchmarkEntry_descriptor;
-    }
-
-    @SuppressWarnings({"rawtypes"})
-    protected com.google.protobuf.MapField internalGetMapField(
-        int number) {
-      switch (number) {
-        case 6:
-          return internalGetExtras();
-        default:
-          throw new RuntimeException(
-              "Invalid map field number: " + number);
-      }
-    }
-    @SuppressWarnings({"rawtypes"})
-    protected com.google.protobuf.MapField internalGetMutableMapField(
-        int number) {
-      switch (number) {
-        case 6:
-          return internalGetMutableExtras();
-        default:
-          throw new RuntimeException(
-              "Invalid map field number: " + number);
-      }
-    }
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_BenchmarkEntry_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.util.testlog.BenchmarkEntry.class, org.tensorflow.proto.util.testlog.BenchmarkEntry.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.util.testlog.BenchmarkEntry.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-        getMetricsFieldBuilder();
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      name_ = "";
-
-      iters_ = 0L;
-
-      cpuTime_ = 0D;
-
-      wallTime_ = 0D;
-
-      throughput_ = 0D;
-
-      internalGetMutableExtras().clear();
-      if (metricsBuilder_ == null) {
-        metrics_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000002);
-      } else {
-        metricsBuilder_.clear();
-      }
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_BenchmarkEntry_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.testlog.BenchmarkEntry getDefaultInstanceForType() {
-      return org.tensorflow.proto.util.testlog.BenchmarkEntry.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.testlog.BenchmarkEntry build() {
-      org.tensorflow.proto.util.testlog.BenchmarkEntry result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.testlog.BenchmarkEntry buildPartial() {
-      org.tensorflow.proto.util.testlog.BenchmarkEntry result = new org.tensorflow.proto.util.testlog.BenchmarkEntry(this);
-      int from_bitField0_ = bitField0_;
-      result.name_ = name_;
-      result.iters_ = iters_;
-      result.cpuTime_ = cpuTime_;
-      result.wallTime_ = wallTime_;
-      result.throughput_ = throughput_;
-      result.extras_ = internalGetExtras();
-      result.extras_.makeImmutable();
-      if (metricsBuilder_ == null) {
-        if (((bitField0_ & 0x00000002) != 0)) {
-          metrics_ = java.util.Collections.unmodifiableList(metrics_);
-          bitField0_ = (bitField0_ & ~0x00000002);
-        }
-        result.metrics_ = metrics_;
-      } else {
-        result.metrics_ = metricsBuilder_.build();
-      }
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.util.testlog.BenchmarkEntry) {
-        return mergeFrom((org.tensorflow.proto.util.testlog.BenchmarkEntry)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.util.testlog.BenchmarkEntry other) {
-      if (other == org.tensorflow.proto.util.testlog.BenchmarkEntry.getDefaultInstance()) return this;
-      if (!other.getName().isEmpty()) {
-        name_ = other.name_;
-        onChanged();
-      }
-      if (other.getIters() != 0L) {
-        setIters(other.getIters());
-      }
-      if (other.getCpuTime() != 0D) {
-        setCpuTime(other.getCpuTime());
-      }
-      if (other.getWallTime() != 0D) {
-        setWallTime(other.getWallTime());
-      }
-      if (other.getThroughput() != 0D) {
-        setThroughput(other.getThroughput());
-      }
-      internalGetMutableExtras().mergeFrom(
-          other.internalGetExtras());
-      if (metricsBuilder_ == null) {
-        if (!other.metrics_.isEmpty()) {
-          if (metrics_.isEmpty()) {
-            metrics_ = other.metrics_;
-            bitField0_ = (bitField0_ & ~0x00000002);
-          } else {
-            ensureMetricsIsMutable();
-            metrics_.addAll(other.metrics_);
-          }
-          onChanged();
-        }
-      } else {
-        if (!other.metrics_.isEmpty()) {
-          if (metricsBuilder_.isEmpty()) {
-            metricsBuilder_.dispose();
-            metricsBuilder_ = null;
-            metrics_ = other.metrics_;
-            bitField0_ = (bitField0_ & ~0x00000002);
-            metricsBuilder_ = 
-              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
-                 getMetricsFieldBuilder() : null;
-          } else {
-            metricsBuilder_.addAllMessages(other.metrics_);
-          }
-        }
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.util.testlog.BenchmarkEntry parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.util.testlog.BenchmarkEntry) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-    private int bitField0_;
-
-    private java.lang.Object name_ = "";
-    /**
-     * <pre>
-     * The name of the specific benchmark or test
-     * (e.g. BM_AdjustContrast_gpu_B_W_H)
-     * </pre>
-     *
-     * <code>string name = 1;</code>
-     */
-    public java.lang.String getName() {
-      java.lang.Object ref = name_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        name_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * The name of the specific benchmark or test
-     * (e.g. BM_AdjustContrast_gpu_B_W_H)
-     * </pre>
-     *
-     * <code>string name = 1;</code>
-     */
-    public com.google.protobuf.ByteString
-        getNameBytes() {
-      java.lang.Object ref = name_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        name_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * The name of the specific benchmark or test
-     * (e.g. BM_AdjustContrast_gpu_B_W_H)
-     * </pre>
-     *
-     * <code>string name = 1;</code>
-     */
-    public Builder setName(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      name_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * The name of the specific benchmark or test
-     * (e.g. BM_AdjustContrast_gpu_B_W_H)
-     * </pre>
-     *
-     * <code>string name = 1;</code>
-     */
-    public Builder clearName() {
-      
-      name_ = getDefaultInstance().getName();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * The name of the specific benchmark or test
-     * (e.g. BM_AdjustContrast_gpu_B_W_H)
-     * </pre>
-     *
-     * <code>string name = 1;</code>
-     */
-    public Builder setNameBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      name_ = value;
-      onChanged();
-      return this;
-    }
-
-    private long iters_ ;
-    /**
-     * <pre>
-     * If a benchmark, how many iterations it was run for
-     * </pre>
-     *
-     * <code>int64 iters = 2;</code>
-     */
-    public long getIters() {
-      return iters_;
-    }
-    /**
-     * <pre>
-     * If a benchmark, how many iterations it was run for
-     * </pre>
-     *
-     * <code>int64 iters = 2;</code>
-     */
-    public Builder setIters(long value) {
-      
-      iters_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * If a benchmark, how many iterations it was run for
-     * </pre>
-     *
-     * <code>int64 iters = 2;</code>
-     */
-    public Builder clearIters() {
-      
-      iters_ = 0L;
-      onChanged();
-      return this;
-    }
-
-    private double cpuTime_ ;
-    /**
-     * <pre>
-     * Total cpu time used for all iterations (in seconds)
-     * </pre>
-     *
-     * <code>double cpu_time = 3;</code>
-     */
-    public double getCpuTime() {
-      return cpuTime_;
-    }
-    /**
-     * <pre>
-     * Total cpu time used for all iterations (in seconds)
-     * </pre>
-     *
-     * <code>double cpu_time = 3;</code>
-     */
-    public Builder setCpuTime(double value) {
-      
-      cpuTime_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Total cpu time used for all iterations (in seconds)
-     * </pre>
-     *
-     * <code>double cpu_time = 3;</code>
-     */
-    public Builder clearCpuTime() {
-      
-      cpuTime_ = 0D;
-      onChanged();
-      return this;
-    }
-
-    private double wallTime_ ;
-    /**
-     * <pre>
-     * Total wall time used for all iterations (in seconds)
-     * </pre>
-     *
-     * <code>double wall_time = 4;</code>
-     */
-    public double getWallTime() {
-      return wallTime_;
-    }
-    /**
-     * <pre>
-     * Total wall time used for all iterations (in seconds)
-     * </pre>
-     *
-     * <code>double wall_time = 4;</code>
-     */
-    public Builder setWallTime(double value) {
-      
-      wallTime_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Total wall time used for all iterations (in seconds)
-     * </pre>
-     *
-     * <code>double wall_time = 4;</code>
-     */
-    public Builder clearWallTime() {
-      
-      wallTime_ = 0D;
-      onChanged();
-      return this;
-    }
-
-    private double throughput_ ;
-    /**
-     * <pre>
-     * Throughput (in MB/s)
-     * </pre>
-     *
-     * <code>double throughput = 5;</code>
-     */
-    public double getThroughput() {
-      return throughput_;
-    }
-    /**
-     * <pre>
-     * Throughput (in MB/s)
-     * </pre>
-     *
-     * <code>double throughput = 5;</code>
-     */
-    public Builder setThroughput(double value) {
-      
-      throughput_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Throughput (in MB/s)
-     * </pre>
-     *
-     * <code>double throughput = 5;</code>
-     */
-    public Builder clearThroughput() {
-      
-      throughput_ = 0D;
-      onChanged();
-      return this;
-    }
-
-    private com.google.protobuf.MapField<
-        java.lang.String, org.tensorflow.proto.util.testlog.EntryValue> extras_;
-    private com.google.protobuf.MapField<java.lang.String, org.tensorflow.proto.util.testlog.EntryValue>
-    internalGetExtras() {
-      if (extras_ == null) {
-        return com.google.protobuf.MapField.emptyMapField(
-            ExtrasDefaultEntryHolder.defaultEntry);
-      }
-      return extras_;
-    }
-    private com.google.protobuf.MapField<java.lang.String, org.tensorflow.proto.util.testlog.EntryValue>
-    internalGetMutableExtras() {
-      onChanged();;
-      if (extras_ == null) {
-        extras_ = com.google.protobuf.MapField.newMapField(
-            ExtrasDefaultEntryHolder.defaultEntry);
-      }
-      if (!extras_.isMutable()) {
-        extras_ = extras_.copy();
-      }
-      return extras_;
-    }
-
-    public int getExtrasCount() {
-      return internalGetExtras().getMap().size();
-    }
-    /**
-     * <pre>
-     * Generic map from result key to value.
-     * </pre>
-     *
-     * <code>map&lt;string, .tensorflow.EntryValue&gt; extras = 6;</code>
-     */
-
-    public boolean containsExtras(
-        java.lang.String key) {
-      if (key == null) { throw new java.lang.NullPointerException(); }
-      return internalGetExtras().getMap().containsKey(key);
-    }
-    /**
-     * Use {@link #getExtrasMap()} instead.
-     */
-    @java.lang.Deprecated
-    public java.util.Map<java.lang.String, org.tensorflow.proto.util.testlog.EntryValue> getExtras() {
-      return getExtrasMap();
-    }
-    /**
-     * <pre>
-     * Generic map from result key to value.
-     * </pre>
-     *
-     * <code>map&lt;string, .tensorflow.EntryValue&gt; extras = 6;</code>
-     */
-
-    public java.util.Map<java.lang.String, org.tensorflow.proto.util.testlog.EntryValue> getExtrasMap() {
-      return internalGetExtras().getMap();
-    }
-    /**
-     * <pre>
-     * Generic map from result key to value.
-     * </pre>
-     *
-     * <code>map&lt;string, .tensorflow.EntryValue&gt; extras = 6;</code>
-     */
-
-    public org.tensorflow.proto.util.testlog.EntryValue getExtrasOrDefault(
-        java.lang.String key,
-        org.tensorflow.proto.util.testlog.EntryValue defaultValue) {
-      if (key == null) { throw new java.lang.NullPointerException(); }
-      java.util.Map<java.lang.String, org.tensorflow.proto.util.testlog.EntryValue> map =
-          internalGetExtras().getMap();
-      return map.containsKey(key) ? map.get(key) : defaultValue;
-    }
-    /**
-     * <pre>
-     * Generic map from result key to value.
-     * </pre>
-     *
-     * <code>map&lt;string, .tensorflow.EntryValue&gt; extras = 6;</code>
-     */
-
-    public org.tensorflow.proto.util.testlog.EntryValue getExtrasOrThrow(
-        java.lang.String key) {
-      if (key == null) { throw new java.lang.NullPointerException(); }
-      java.util.Map<java.lang.String, org.tensorflow.proto.util.testlog.EntryValue> map =
-          internalGetExtras().getMap();
-      if (!map.containsKey(key)) {
-        throw new java.lang.IllegalArgumentException();
-      }
-      return map.get(key);
-    }
-
-    public Builder clearExtras() {
-      internalGetMutableExtras().getMutableMap()
-          .clear();
-      return this;
-    }
-    /**
-     * <pre>
-     * Generic map from result key to value.
-     * </pre>
-     *
-     * <code>map&lt;string, .tensorflow.EntryValue&gt; extras = 6;</code>
-     */
-
-    public Builder removeExtras(
-        java.lang.String key) {
-      if (key == null) { throw new java.lang.NullPointerException(); }
-      internalGetMutableExtras().getMutableMap()
-          .remove(key);
-      return this;
-    }
-    /**
-     * Use alternate mutation accessors instead.
-     */
-    @java.lang.Deprecated
-    public java.util.Map<java.lang.String, org.tensorflow.proto.util.testlog.EntryValue>
-    getMutableExtras() {
-      return internalGetMutableExtras().getMutableMap();
-    }
-    /**
-     * <pre>
-     * Generic map from result key to value.
-     * </pre>
-     *
-     * <code>map&lt;string, .tensorflow.EntryValue&gt; extras = 6;</code>
-     */
-    public Builder putExtras(
-        java.lang.String key,
-        org.tensorflow.proto.util.testlog.EntryValue value) {
-      if (key == null) { throw new java.lang.NullPointerException(); }
-      if (value == null) { throw new java.lang.NullPointerException(); }
-      internalGetMutableExtras().getMutableMap()
-          .put(key, value);
-      return this;
-    }
-    /**
-     * <pre>
-     * Generic map from result key to value.
-     * </pre>
-     *
-     * <code>map&lt;string, .tensorflow.EntryValue&gt; extras = 6;</code>
-     */
-
-    public Builder putAllExtras(
-        java.util.Map<java.lang.String, org.tensorflow.proto.util.testlog.EntryValue> values) {
-      internalGetMutableExtras().getMutableMap()
-          .putAll(values);
-      return this;
-    }
-
-    private java.util.List<org.tensorflow.proto.util.testlog.MetricEntry> metrics_ =
-      java.util.Collections.emptyList();
-    private void ensureMetricsIsMutable() {
-      if (!((bitField0_ & 0x00000002) != 0)) {
-        metrics_ = new java.util.ArrayList<org.tensorflow.proto.util.testlog.MetricEntry>(metrics_);
-        bitField0_ |= 0x00000002;
-       }
-    }
-
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.util.testlog.MetricEntry, org.tensorflow.proto.util.testlog.MetricEntry.Builder, org.tensorflow.proto.util.testlog.MetricEntryOrBuilder> metricsBuilder_;
-
-    /**
-     * <pre>
-     * Metric name, value and expected range. This can include accuracy metrics
-     * typically used to determine whether the accuracy test has passed
-     * </pre>
-     *
-     * <code>repeated .tensorflow.MetricEntry metrics = 7;</code>
-     */
-    public java.util.List<org.tensorflow.proto.util.testlog.MetricEntry> getMetricsList() {
-      if (metricsBuilder_ == null) {
-        return java.util.Collections.unmodifiableList(metrics_);
-      } else {
-        return metricsBuilder_.getMessageList();
-      }
-    }
-    /**
-     * <pre>
-     * Metric name, value and expected range. This can include accuracy metrics
-     * typically used to determine whether the accuracy test has passed
-     * </pre>
-     *
-     * <code>repeated .tensorflow.MetricEntry metrics = 7;</code>
-     */
-    public int getMetricsCount() {
-      if (metricsBuilder_ == null) {
-        return metrics_.size();
-      } else {
-        return metricsBuilder_.getCount();
-      }
-    }
-    /**
-     * <pre>
-     * Metric name, value and expected range. This can include accuracy metrics
-     * typically used to determine whether the accuracy test has passed
-     * </pre>
-     *
-     * <code>repeated .tensorflow.MetricEntry metrics = 7;</code>
-     */
-    public org.tensorflow.proto.util.testlog.MetricEntry getMetrics(int index) {
-      if (metricsBuilder_ == null) {
-        return metrics_.get(index);
-      } else {
-        return metricsBuilder_.getMessage(index);
-      }
-    }
-    /**
-     * <pre>
-     * Metric name, value and expected range. This can include accuracy metrics
-     * typically used to determine whether the accuracy test has passed
-     * </pre>
-     *
-     * <code>repeated .tensorflow.MetricEntry metrics = 7;</code>
-     */
-    public Builder setMetrics(
-        int index, org.tensorflow.proto.util.testlog.MetricEntry value) {
-      if (metricsBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureMetricsIsMutable();
-        metrics_.set(index, value);
-        onChanged();
-      } else {
-        metricsBuilder_.setMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Metric name, value and expected range. This can include accuracy metrics
-     * typically used to determine whether the accuracy test has passed
-     * </pre>
-     *
-     * <code>repeated .tensorflow.MetricEntry metrics = 7;</code>
-     */
-    public Builder setMetrics(
-        int index, org.tensorflow.proto.util.testlog.MetricEntry.Builder builderForValue) {
-      if (metricsBuilder_ == null) {
-        ensureMetricsIsMutable();
-        metrics_.set(index, builderForValue.build());
-        onChanged();
-      } else {
-        metricsBuilder_.setMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Metric name, value and expected range. This can include accuracy metrics
-     * typically used to determine whether the accuracy test has passed
-     * </pre>
-     *
-     * <code>repeated .tensorflow.MetricEntry metrics = 7;</code>
-     */
-    public Builder addMetrics(org.tensorflow.proto.util.testlog.MetricEntry value) {
-      if (metricsBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureMetricsIsMutable();
-        metrics_.add(value);
-        onChanged();
-      } else {
-        metricsBuilder_.addMessage(value);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Metric name, value and expected range. This can include accuracy metrics
-     * typically used to determine whether the accuracy test has passed
-     * </pre>
-     *
-     * <code>repeated .tensorflow.MetricEntry metrics = 7;</code>
-     */
-    public Builder addMetrics(
-        int index, org.tensorflow.proto.util.testlog.MetricEntry value) {
-      if (metricsBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureMetricsIsMutable();
-        metrics_.add(index, value);
-        onChanged();
-      } else {
-        metricsBuilder_.addMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Metric name, value and expected range. This can include accuracy metrics
-     * typically used to determine whether the accuracy test has passed
-     * </pre>
-     *
-     * <code>repeated .tensorflow.MetricEntry metrics = 7;</code>
-     */
-    public Builder addMetrics(
-        org.tensorflow.proto.util.testlog.MetricEntry.Builder builderForValue) {
-      if (metricsBuilder_ == null) {
-        ensureMetricsIsMutable();
-        metrics_.add(builderForValue.build());
-        onChanged();
-      } else {
-        metricsBuilder_.addMessage(builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Metric name, value and expected range. This can include accuracy metrics
-     * typically used to determine whether the accuracy test has passed
-     * </pre>
-     *
-     * <code>repeated .tensorflow.MetricEntry metrics = 7;</code>
-     */
-    public Builder addMetrics(
-        int index, org.tensorflow.proto.util.testlog.MetricEntry.Builder builderForValue) {
-      if (metricsBuilder_ == null) {
-        ensureMetricsIsMutable();
-        metrics_.add(index, builderForValue.build());
-        onChanged();
-      } else {
-        metricsBuilder_.addMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Metric name, value and expected range. This can include accuracy metrics
-     * typically used to determine whether the accuracy test has passed
-     * </pre>
-     *
-     * <code>repeated .tensorflow.MetricEntry metrics = 7;</code>
-     */
-    public Builder addAllMetrics(
-        java.lang.Iterable<? extends org.tensorflow.proto.util.testlog.MetricEntry> values) {
-      if (metricsBuilder_ == null) {
-        ensureMetricsIsMutable();
-        com.google.protobuf.AbstractMessageLite.Builder.addAll(
-            values, metrics_);
-        onChanged();
-      } else {
-        metricsBuilder_.addAllMessages(values);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Metric name, value and expected range. This can include accuracy metrics
-     * typically used to determine whether the accuracy test has passed
-     * </pre>
-     *
-     * <code>repeated .tensorflow.MetricEntry metrics = 7;</code>
-     */
-    public Builder clearMetrics() {
-      if (metricsBuilder_ == null) {
-        metrics_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000002);
-        onChanged();
-      } else {
-        metricsBuilder_.clear();
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Metric name, value and expected range. This can include accuracy metrics
-     * typically used to determine whether the accuracy test has passed
-     * </pre>
-     *
-     * <code>repeated .tensorflow.MetricEntry metrics = 7;</code>
-     */
-    public Builder removeMetrics(int index) {
-      if (metricsBuilder_ == null) {
-        ensureMetricsIsMutable();
-        metrics_.remove(index);
-        onChanged();
-      } else {
-        metricsBuilder_.remove(index);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Metric name, value and expected range. This can include accuracy metrics
-     * typically used to determine whether the accuracy test has passed
-     * </pre>
-     *
-     * <code>repeated .tensorflow.MetricEntry metrics = 7;</code>
-     */
-    public org.tensorflow.proto.util.testlog.MetricEntry.Builder getMetricsBuilder(
-        int index) {
-      return getMetricsFieldBuilder().getBuilder(index);
-    }
-    /**
-     * <pre>
-     * Metric name, value and expected range. This can include accuracy metrics
-     * typically used to determine whether the accuracy test has passed
-     * </pre>
-     *
-     * <code>repeated .tensorflow.MetricEntry metrics = 7;</code>
-     */
-    public org.tensorflow.proto.util.testlog.MetricEntryOrBuilder getMetricsOrBuilder(
-        int index) {
-      if (metricsBuilder_ == null) {
-        return metrics_.get(index);  } else {
-        return metricsBuilder_.getMessageOrBuilder(index);
-      }
-    }
-    /**
-     * <pre>
-     * Metric name, value and expected range. This can include accuracy metrics
-     * typically used to determine whether the accuracy test has passed
-     * </pre>
-     *
-     * <code>repeated .tensorflow.MetricEntry metrics = 7;</code>
-     */
-    public java.util.List<? extends org.tensorflow.proto.util.testlog.MetricEntryOrBuilder> 
-         getMetricsOrBuilderList() {
-      if (metricsBuilder_ != null) {
-        return metricsBuilder_.getMessageOrBuilderList();
-      } else {
-        return java.util.Collections.unmodifiableList(metrics_);
-      }
-    }
-    /**
-     * <pre>
-     * Metric name, value and expected range. This can include accuracy metrics
-     * typically used to determine whether the accuracy test has passed
-     * </pre>
-     *
-     * <code>repeated .tensorflow.MetricEntry metrics = 7;</code>
-     */
-    public org.tensorflow.proto.util.testlog.MetricEntry.Builder addMetricsBuilder() {
-      return getMetricsFieldBuilder().addBuilder(
-          org.tensorflow.proto.util.testlog.MetricEntry.getDefaultInstance());
-    }
-    /**
-     * <pre>
-     * Metric name, value and expected range. This can include accuracy metrics
-     * typically used to determine whether the accuracy test has passed
-     * </pre>
-     *
-     * <code>repeated .tensorflow.MetricEntry metrics = 7;</code>
-     */
-    public org.tensorflow.proto.util.testlog.MetricEntry.Builder addMetricsBuilder(
-        int index) {
-      return getMetricsFieldBuilder().addBuilder(
-          index, org.tensorflow.proto.util.testlog.MetricEntry.getDefaultInstance());
-    }
-    /**
-     * <pre>
-     * Metric name, value and expected range. This can include accuracy metrics
-     * typically used to determine whether the accuracy test has passed
-     * </pre>
-     *
-     * <code>repeated .tensorflow.MetricEntry metrics = 7;</code>
-     */
-    public java.util.List<org.tensorflow.proto.util.testlog.MetricEntry.Builder> 
-         getMetricsBuilderList() {
-      return getMetricsFieldBuilder().getBuilderList();
-    }
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.util.testlog.MetricEntry, org.tensorflow.proto.util.testlog.MetricEntry.Builder, org.tensorflow.proto.util.testlog.MetricEntryOrBuilder> 
-        getMetricsFieldBuilder() {
-      if (metricsBuilder_ == null) {
-        metricsBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
-            org.tensorflow.proto.util.testlog.MetricEntry, org.tensorflow.proto.util.testlog.MetricEntry.Builder, org.tensorflow.proto.util.testlog.MetricEntryOrBuilder>(
-                metrics_,
-                ((bitField0_ & 0x00000002) != 0),
-                getParentForChildren(),
-                isClean());
-        metrics_ = null;
-      }
-      return metricsBuilder_;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.BenchmarkEntry)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.BenchmarkEntry)
-  private static final org.tensorflow.proto.util.testlog.BenchmarkEntry DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.util.testlog.BenchmarkEntry();
-  }
-
-  public static org.tensorflow.proto.util.testlog.BenchmarkEntry getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<BenchmarkEntry>
-      PARSER = new com.google.protobuf.AbstractParser<BenchmarkEntry>() {
-    @java.lang.Override
-    public BenchmarkEntry parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new BenchmarkEntry(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<BenchmarkEntry> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<BenchmarkEntry> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.util.testlog.BenchmarkEntry getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/BenchmarkEntryOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/BenchmarkEntryOrBuilder.java
deleted file mode 100644
index 475b3e31b7c..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/BenchmarkEntryOrBuilder.java
+++ /dev/null
@@ -1,168 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/util/test_log.proto
-
-package org.tensorflow.proto.util.testlog;
-
-public interface BenchmarkEntryOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.BenchmarkEntry)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <pre>
-   * The name of the specific benchmark or test
-   * (e.g. BM_AdjustContrast_gpu_B_W_H)
-   * </pre>
-   *
-   * <code>string name = 1;</code>
-   */
-  java.lang.String getName();
-  /**
-   * <pre>
-   * The name of the specific benchmark or test
-   * (e.g. BM_AdjustContrast_gpu_B_W_H)
-   * </pre>
-   *
-   * <code>string name = 1;</code>
-   */
-  com.google.protobuf.ByteString
-      getNameBytes();
-
-  /**
-   * <pre>
-   * If a benchmark, how many iterations it was run for
-   * </pre>
-   *
-   * <code>int64 iters = 2;</code>
-   */
-  long getIters();
-
-  /**
-   * <pre>
-   * Total cpu time used for all iterations (in seconds)
-   * </pre>
-   *
-   * <code>double cpu_time = 3;</code>
-   */
-  double getCpuTime();
-
-  /**
-   * <pre>
-   * Total wall time used for all iterations (in seconds)
-   * </pre>
-   *
-   * <code>double wall_time = 4;</code>
-   */
-  double getWallTime();
-
-  /**
-   * <pre>
-   * Throughput (in MB/s)
-   * </pre>
-   *
-   * <code>double throughput = 5;</code>
-   */
-  double getThroughput();
-
-  /**
-   * <pre>
-   * Generic map from result key to value.
-   * </pre>
-   *
-   * <code>map&lt;string, .tensorflow.EntryValue&gt; extras = 6;</code>
-   */
-  int getExtrasCount();
-  /**
-   * <pre>
-   * Generic map from result key to value.
-   * </pre>
-   *
-   * <code>map&lt;string, .tensorflow.EntryValue&gt; extras = 6;</code>
-   */
-  boolean containsExtras(
-      java.lang.String key);
-  /**
-   * Use {@link #getExtrasMap()} instead.
-   */
-  @java.lang.Deprecated
-  java.util.Map<java.lang.String, org.tensorflow.proto.util.testlog.EntryValue>
-  getExtras();
-  /**
-   * <pre>
-   * Generic map from result key to value.
-   * </pre>
-   *
-   * <code>map&lt;string, .tensorflow.EntryValue&gt; extras = 6;</code>
-   */
-  java.util.Map<java.lang.String, org.tensorflow.proto.util.testlog.EntryValue>
-  getExtrasMap();
-  /**
-   * <pre>
-   * Generic map from result key to value.
-   * </pre>
-   *
-   * <code>map&lt;string, .tensorflow.EntryValue&gt; extras = 6;</code>
-   */
-
-  org.tensorflow.proto.util.testlog.EntryValue getExtrasOrDefault(
-      java.lang.String key,
-      org.tensorflow.proto.util.testlog.EntryValue defaultValue);
-  /**
-   * <pre>
-   * Generic map from result key to value.
-   * </pre>
-   *
-   * <code>map&lt;string, .tensorflow.EntryValue&gt; extras = 6;</code>
-   */
-
-  org.tensorflow.proto.util.testlog.EntryValue getExtrasOrThrow(
-      java.lang.String key);
-
-  /**
-   * <pre>
-   * Metric name, value and expected range. This can include accuracy metrics
-   * typically used to determine whether the accuracy test has passed
-   * </pre>
-   *
-   * <code>repeated .tensorflow.MetricEntry metrics = 7;</code>
-   */
-  java.util.List<org.tensorflow.proto.util.testlog.MetricEntry> 
-      getMetricsList();
-  /**
-   * <pre>
-   * Metric name, value and expected range. This can include accuracy metrics
-   * typically used to determine whether the accuracy test has passed
-   * </pre>
-   *
-   * <code>repeated .tensorflow.MetricEntry metrics = 7;</code>
-   */
-  org.tensorflow.proto.util.testlog.MetricEntry getMetrics(int index);
-  /**
-   * <pre>
-   * Metric name, value and expected range. This can include accuracy metrics
-   * typically used to determine whether the accuracy test has passed
-   * </pre>
-   *
-   * <code>repeated .tensorflow.MetricEntry metrics = 7;</code>
-   */
-  int getMetricsCount();
-  /**
-   * <pre>
-   * Metric name, value and expected range. This can include accuracy metrics
-   * typically used to determine whether the accuracy test has passed
-   * </pre>
-   *
-   * <code>repeated .tensorflow.MetricEntry metrics = 7;</code>
-   */
-  java.util.List<? extends org.tensorflow.proto.util.testlog.MetricEntryOrBuilder> 
-      getMetricsOrBuilderList();
-  /**
-   * <pre>
-   * Metric name, value and expected range. This can include accuracy metrics
-   * typically used to determine whether the accuracy test has passed
-   * </pre>
-   *
-   * <code>repeated .tensorflow.MetricEntry metrics = 7;</code>
-   */
-  org.tensorflow.proto.util.testlog.MetricEntryOrBuilder getMetricsOrBuilder(
-      int index);
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/BuildConfiguration.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/BuildConfiguration.java
deleted file mode 100644
index 8d571429f4f..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/BuildConfiguration.java
+++ /dev/null
@@ -1,1021 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/util/test_log.proto
-
-package org.tensorflow.proto.util.testlog;
-
-/**
- * Protobuf type {@code tensorflow.BuildConfiguration}
- */
-public  final class BuildConfiguration extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.BuildConfiguration)
-    BuildConfigurationOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use BuildConfiguration.newBuilder() to construct.
-  private BuildConfiguration(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private BuildConfiguration() {
-    mode_ = "";
-    ccFlags_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-    opts_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new BuildConfiguration();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private BuildConfiguration(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    int mutable_bitField0_ = 0;
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 10: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            mode_ = s;
-            break;
-          }
-          case 18: {
-            java.lang.String s = input.readStringRequireUtf8();
-            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
-              ccFlags_ = new com.google.protobuf.LazyStringArrayList();
-              mutable_bitField0_ |= 0x00000001;
-            }
-            ccFlags_.add(s);
-            break;
-          }
-          case 26: {
-            java.lang.String s = input.readStringRequireUtf8();
-            if (!((mutable_bitField0_ & 0x00000002) != 0)) {
-              opts_ = new com.google.protobuf.LazyStringArrayList();
-              mutable_bitField0_ |= 0x00000002;
-            }
-            opts_.add(s);
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      if (((mutable_bitField0_ & 0x00000001) != 0)) {
-        ccFlags_ = ccFlags_.getUnmodifiableView();
-      }
-      if (((mutable_bitField0_ & 0x00000002) != 0)) {
-        opts_ = opts_.getUnmodifiableView();
-      }
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_BuildConfiguration_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_BuildConfiguration_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.util.testlog.BuildConfiguration.class, org.tensorflow.proto.util.testlog.BuildConfiguration.Builder.class);
-  }
-
-  public static final int MODE_FIELD_NUMBER = 1;
-  private volatile java.lang.Object mode_;
-  /**
-   * <pre>
-   * opt, dbg, etc
-   * </pre>
-   *
-   * <code>string mode = 1;</code>
-   */
-  public java.lang.String getMode() {
-    java.lang.Object ref = mode_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      mode_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * opt, dbg, etc
-   * </pre>
-   *
-   * <code>string mode = 1;</code>
-   */
-  public com.google.protobuf.ByteString
-      getModeBytes() {
-    java.lang.Object ref = mode_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      mode_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int CC_FLAGS_FIELD_NUMBER = 2;
-  private com.google.protobuf.LazyStringList ccFlags_;
-  /**
-   * <pre>
-   * CC compiler flags, if known
-   * </pre>
-   *
-   * <code>repeated string cc_flags = 2;</code>
-   */
-  public com.google.protobuf.ProtocolStringList
-      getCcFlagsList() {
-    return ccFlags_;
-  }
-  /**
-   * <pre>
-   * CC compiler flags, if known
-   * </pre>
-   *
-   * <code>repeated string cc_flags = 2;</code>
-   */
-  public int getCcFlagsCount() {
-    return ccFlags_.size();
-  }
-  /**
-   * <pre>
-   * CC compiler flags, if known
-   * </pre>
-   *
-   * <code>repeated string cc_flags = 2;</code>
-   */
-  public java.lang.String getCcFlags(int index) {
-    return ccFlags_.get(index);
-  }
-  /**
-   * <pre>
-   * CC compiler flags, if known
-   * </pre>
-   *
-   * <code>repeated string cc_flags = 2;</code>
-   */
-  public com.google.protobuf.ByteString
-      getCcFlagsBytes(int index) {
-    return ccFlags_.getByteString(index);
-  }
-
-  public static final int OPTS_FIELD_NUMBER = 3;
-  private com.google.protobuf.LazyStringList opts_;
-  /**
-   * <pre>
-   * Bazel compilation options, if known
-   * </pre>
-   *
-   * <code>repeated string opts = 3;</code>
-   */
-  public com.google.protobuf.ProtocolStringList
-      getOptsList() {
-    return opts_;
-  }
-  /**
-   * <pre>
-   * Bazel compilation options, if known
-   * </pre>
-   *
-   * <code>repeated string opts = 3;</code>
-   */
-  public int getOptsCount() {
-    return opts_.size();
-  }
-  /**
-   * <pre>
-   * Bazel compilation options, if known
-   * </pre>
-   *
-   * <code>repeated string opts = 3;</code>
-   */
-  public java.lang.String getOpts(int index) {
-    return opts_.get(index);
-  }
-  /**
-   * <pre>
-   * Bazel compilation options, if known
-   * </pre>
-   *
-   * <code>repeated string opts = 3;</code>
-   */
-  public com.google.protobuf.ByteString
-      getOptsBytes(int index) {
-    return opts_.getByteString(index);
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    if (!getModeBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 1, mode_);
-    }
-    for (int i = 0; i < ccFlags_.size(); i++) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 2, ccFlags_.getRaw(i));
-    }
-    for (int i = 0; i < opts_.size(); i++) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 3, opts_.getRaw(i));
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (!getModeBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, mode_);
-    }
-    {
-      int dataSize = 0;
-      for (int i = 0; i < ccFlags_.size(); i++) {
-        dataSize += computeStringSizeNoTag(ccFlags_.getRaw(i));
-      }
-      size += dataSize;
-      size += 1 * getCcFlagsList().size();
-    }
-    {
-      int dataSize = 0;
-      for (int i = 0; i < opts_.size(); i++) {
-        dataSize += computeStringSizeNoTag(opts_.getRaw(i));
-      }
-      size += dataSize;
-      size += 1 * getOptsList().size();
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.util.testlog.BuildConfiguration)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.util.testlog.BuildConfiguration other = (org.tensorflow.proto.util.testlog.BuildConfiguration) obj;
-
-    if (!getMode()
-        .equals(other.getMode())) return false;
-    if (!getCcFlagsList()
-        .equals(other.getCcFlagsList())) return false;
-    if (!getOptsList()
-        .equals(other.getOptsList())) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    hash = (37 * hash) + MODE_FIELD_NUMBER;
-    hash = (53 * hash) + getMode().hashCode();
-    if (getCcFlagsCount() > 0) {
-      hash = (37 * hash) + CC_FLAGS_FIELD_NUMBER;
-      hash = (53 * hash) + getCcFlagsList().hashCode();
-    }
-    if (getOptsCount() > 0) {
-      hash = (37 * hash) + OPTS_FIELD_NUMBER;
-      hash = (53 * hash) + getOptsList().hashCode();
-    }
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.util.testlog.BuildConfiguration parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.testlog.BuildConfiguration parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.testlog.BuildConfiguration parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.testlog.BuildConfiguration parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.testlog.BuildConfiguration parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.testlog.BuildConfiguration parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.testlog.BuildConfiguration parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.testlog.BuildConfiguration parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.testlog.BuildConfiguration parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.testlog.BuildConfiguration parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.testlog.BuildConfiguration parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.testlog.BuildConfiguration parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.util.testlog.BuildConfiguration prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * Protobuf type {@code tensorflow.BuildConfiguration}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.BuildConfiguration)
-      org.tensorflow.proto.util.testlog.BuildConfigurationOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_BuildConfiguration_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_BuildConfiguration_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.util.testlog.BuildConfiguration.class, org.tensorflow.proto.util.testlog.BuildConfiguration.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.util.testlog.BuildConfiguration.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      mode_ = "";
-
-      ccFlags_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-      bitField0_ = (bitField0_ & ~0x00000001);
-      opts_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-      bitField0_ = (bitField0_ & ~0x00000002);
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_BuildConfiguration_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.testlog.BuildConfiguration getDefaultInstanceForType() {
-      return org.tensorflow.proto.util.testlog.BuildConfiguration.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.testlog.BuildConfiguration build() {
-      org.tensorflow.proto.util.testlog.BuildConfiguration result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.testlog.BuildConfiguration buildPartial() {
-      org.tensorflow.proto.util.testlog.BuildConfiguration result = new org.tensorflow.proto.util.testlog.BuildConfiguration(this);
-      int from_bitField0_ = bitField0_;
-      result.mode_ = mode_;
-      if (((bitField0_ & 0x00000001) != 0)) {
-        ccFlags_ = ccFlags_.getUnmodifiableView();
-        bitField0_ = (bitField0_ & ~0x00000001);
-      }
-      result.ccFlags_ = ccFlags_;
-      if (((bitField0_ & 0x00000002) != 0)) {
-        opts_ = opts_.getUnmodifiableView();
-        bitField0_ = (bitField0_ & ~0x00000002);
-      }
-      result.opts_ = opts_;
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.util.testlog.BuildConfiguration) {
-        return mergeFrom((org.tensorflow.proto.util.testlog.BuildConfiguration)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.util.testlog.BuildConfiguration other) {
-      if (other == org.tensorflow.proto.util.testlog.BuildConfiguration.getDefaultInstance()) return this;
-      if (!other.getMode().isEmpty()) {
-        mode_ = other.mode_;
-        onChanged();
-      }
-      if (!other.ccFlags_.isEmpty()) {
-        if (ccFlags_.isEmpty()) {
-          ccFlags_ = other.ccFlags_;
-          bitField0_ = (bitField0_ & ~0x00000001);
-        } else {
-          ensureCcFlagsIsMutable();
-          ccFlags_.addAll(other.ccFlags_);
-        }
-        onChanged();
-      }
-      if (!other.opts_.isEmpty()) {
-        if (opts_.isEmpty()) {
-          opts_ = other.opts_;
-          bitField0_ = (bitField0_ & ~0x00000002);
-        } else {
-          ensureOptsIsMutable();
-          opts_.addAll(other.opts_);
-        }
-        onChanged();
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.util.testlog.BuildConfiguration parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.util.testlog.BuildConfiguration) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-    private int bitField0_;
-
-    private java.lang.Object mode_ = "";
-    /**
-     * <pre>
-     * opt, dbg, etc
-     * </pre>
-     *
-     * <code>string mode = 1;</code>
-     */
-    public java.lang.String getMode() {
-      java.lang.Object ref = mode_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        mode_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * opt, dbg, etc
-     * </pre>
-     *
-     * <code>string mode = 1;</code>
-     */
-    public com.google.protobuf.ByteString
-        getModeBytes() {
-      java.lang.Object ref = mode_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        mode_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * opt, dbg, etc
-     * </pre>
-     *
-     * <code>string mode = 1;</code>
-     */
-    public Builder setMode(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      mode_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * opt, dbg, etc
-     * </pre>
-     *
-     * <code>string mode = 1;</code>
-     */
-    public Builder clearMode() {
-      
-      mode_ = getDefaultInstance().getMode();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * opt, dbg, etc
-     * </pre>
-     *
-     * <code>string mode = 1;</code>
-     */
-    public Builder setModeBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      mode_ = value;
-      onChanged();
-      return this;
-    }
-
-    private com.google.protobuf.LazyStringList ccFlags_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-    private void ensureCcFlagsIsMutable() {
-      if (!((bitField0_ & 0x00000001) != 0)) {
-        ccFlags_ = new com.google.protobuf.LazyStringArrayList(ccFlags_);
-        bitField0_ |= 0x00000001;
-       }
-    }
-    /**
-     * <pre>
-     * CC compiler flags, if known
-     * </pre>
-     *
-     * <code>repeated string cc_flags = 2;</code>
-     */
-    public com.google.protobuf.ProtocolStringList
-        getCcFlagsList() {
-      return ccFlags_.getUnmodifiableView();
-    }
-    /**
-     * <pre>
-     * CC compiler flags, if known
-     * </pre>
-     *
-     * <code>repeated string cc_flags = 2;</code>
-     */
-    public int getCcFlagsCount() {
-      return ccFlags_.size();
-    }
-    /**
-     * <pre>
-     * CC compiler flags, if known
-     * </pre>
-     *
-     * <code>repeated string cc_flags = 2;</code>
-     */
-    public java.lang.String getCcFlags(int index) {
-      return ccFlags_.get(index);
-    }
-    /**
-     * <pre>
-     * CC compiler flags, if known
-     * </pre>
-     *
-     * <code>repeated string cc_flags = 2;</code>
-     */
-    public com.google.protobuf.ByteString
-        getCcFlagsBytes(int index) {
-      return ccFlags_.getByteString(index);
-    }
-    /**
-     * <pre>
-     * CC compiler flags, if known
-     * </pre>
-     *
-     * <code>repeated string cc_flags = 2;</code>
-     */
-    public Builder setCcFlags(
-        int index, java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  ensureCcFlagsIsMutable();
-      ccFlags_.set(index, value);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * CC compiler flags, if known
-     * </pre>
-     *
-     * <code>repeated string cc_flags = 2;</code>
-     */
-    public Builder addCcFlags(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  ensureCcFlagsIsMutable();
-      ccFlags_.add(value);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * CC compiler flags, if known
-     * </pre>
-     *
-     * <code>repeated string cc_flags = 2;</code>
-     */
-    public Builder addAllCcFlags(
-        java.lang.Iterable<java.lang.String> values) {
-      ensureCcFlagsIsMutable();
-      com.google.protobuf.AbstractMessageLite.Builder.addAll(
-          values, ccFlags_);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * CC compiler flags, if known
-     * </pre>
-     *
-     * <code>repeated string cc_flags = 2;</code>
-     */
-    public Builder clearCcFlags() {
-      ccFlags_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-      bitField0_ = (bitField0_ & ~0x00000001);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * CC compiler flags, if known
-     * </pre>
-     *
-     * <code>repeated string cc_flags = 2;</code>
-     */
-    public Builder addCcFlagsBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      ensureCcFlagsIsMutable();
-      ccFlags_.add(value);
-      onChanged();
-      return this;
-    }
-
-    private com.google.protobuf.LazyStringList opts_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-    private void ensureOptsIsMutable() {
-      if (!((bitField0_ & 0x00000002) != 0)) {
-        opts_ = new com.google.protobuf.LazyStringArrayList(opts_);
-        bitField0_ |= 0x00000002;
-       }
-    }
-    /**
-     * <pre>
-     * Bazel compilation options, if known
-     * </pre>
-     *
-     * <code>repeated string opts = 3;</code>
-     */
-    public com.google.protobuf.ProtocolStringList
-        getOptsList() {
-      return opts_.getUnmodifiableView();
-    }
-    /**
-     * <pre>
-     * Bazel compilation options, if known
-     * </pre>
-     *
-     * <code>repeated string opts = 3;</code>
-     */
-    public int getOptsCount() {
-      return opts_.size();
-    }
-    /**
-     * <pre>
-     * Bazel compilation options, if known
-     * </pre>
-     *
-     * <code>repeated string opts = 3;</code>
-     */
-    public java.lang.String getOpts(int index) {
-      return opts_.get(index);
-    }
-    /**
-     * <pre>
-     * Bazel compilation options, if known
-     * </pre>
-     *
-     * <code>repeated string opts = 3;</code>
-     */
-    public com.google.protobuf.ByteString
-        getOptsBytes(int index) {
-      return opts_.getByteString(index);
-    }
-    /**
-     * <pre>
-     * Bazel compilation options, if known
-     * </pre>
-     *
-     * <code>repeated string opts = 3;</code>
-     */
-    public Builder setOpts(
-        int index, java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  ensureOptsIsMutable();
-      opts_.set(index, value);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Bazel compilation options, if known
-     * </pre>
-     *
-     * <code>repeated string opts = 3;</code>
-     */
-    public Builder addOpts(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  ensureOptsIsMutable();
-      opts_.add(value);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Bazel compilation options, if known
-     * </pre>
-     *
-     * <code>repeated string opts = 3;</code>
-     */
-    public Builder addAllOpts(
-        java.lang.Iterable<java.lang.String> values) {
-      ensureOptsIsMutable();
-      com.google.protobuf.AbstractMessageLite.Builder.addAll(
-          values, opts_);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Bazel compilation options, if known
-     * </pre>
-     *
-     * <code>repeated string opts = 3;</code>
-     */
-    public Builder clearOpts() {
-      opts_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-      bitField0_ = (bitField0_ & ~0x00000002);
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Bazel compilation options, if known
-     * </pre>
-     *
-     * <code>repeated string opts = 3;</code>
-     */
-    public Builder addOptsBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      ensureOptsIsMutable();
-      opts_.add(value);
-      onChanged();
-      return this;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.BuildConfiguration)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.BuildConfiguration)
-  private static final org.tensorflow.proto.util.testlog.BuildConfiguration DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.util.testlog.BuildConfiguration();
-  }
-
-  public static org.tensorflow.proto.util.testlog.BuildConfiguration getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<BuildConfiguration>
-      PARSER = new com.google.protobuf.AbstractParser<BuildConfiguration>() {
-    @java.lang.Override
-    public BuildConfiguration parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new BuildConfiguration(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<BuildConfiguration> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<BuildConfiguration> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.util.testlog.BuildConfiguration getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/BuildConfigurationOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/BuildConfigurationOrBuilder.java
deleted file mode 100644
index 4fd04ebf65d..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/BuildConfigurationOrBuilder.java
+++ /dev/null
@@ -1,97 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/util/test_log.proto
-
-package org.tensorflow.proto.util.testlog;
-
-public interface BuildConfigurationOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.BuildConfiguration)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <pre>
-   * opt, dbg, etc
-   * </pre>
-   *
-   * <code>string mode = 1;</code>
-   */
-  java.lang.String getMode();
-  /**
-   * <pre>
-   * opt, dbg, etc
-   * </pre>
-   *
-   * <code>string mode = 1;</code>
-   */
-  com.google.protobuf.ByteString
-      getModeBytes();
-
-  /**
-   * <pre>
-   * CC compiler flags, if known
-   * </pre>
-   *
-   * <code>repeated string cc_flags = 2;</code>
-   */
-  java.util.List<java.lang.String>
-      getCcFlagsList();
-  /**
-   * <pre>
-   * CC compiler flags, if known
-   * </pre>
-   *
-   * <code>repeated string cc_flags = 2;</code>
-   */
-  int getCcFlagsCount();
-  /**
-   * <pre>
-   * CC compiler flags, if known
-   * </pre>
-   *
-   * <code>repeated string cc_flags = 2;</code>
-   */
-  java.lang.String getCcFlags(int index);
-  /**
-   * <pre>
-   * CC compiler flags, if known
-   * </pre>
-   *
-   * <code>repeated string cc_flags = 2;</code>
-   */
-  com.google.protobuf.ByteString
-      getCcFlagsBytes(int index);
-
-  /**
-   * <pre>
-   * Bazel compilation options, if known
-   * </pre>
-   *
-   * <code>repeated string opts = 3;</code>
-   */
-  java.util.List<java.lang.String>
-      getOptsList();
-  /**
-   * <pre>
-   * Bazel compilation options, if known
-   * </pre>
-   *
-   * <code>repeated string opts = 3;</code>
-   */
-  int getOptsCount();
-  /**
-   * <pre>
-   * Bazel compilation options, if known
-   * </pre>
-   *
-   * <code>repeated string opts = 3;</code>
-   */
-  java.lang.String getOpts(int index);
-  /**
-   * <pre>
-   * Bazel compilation options, if known
-   * </pre>
-   *
-   * <code>repeated string opts = 3;</code>
-   */
-  com.google.protobuf.ByteString
-      getOptsBytes(int index);
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/CPUInfo.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/CPUInfo.java
deleted file mode 100644
index b4ae942f481..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/CPUInfo.java
+++ /dev/null
@@ -1,1254 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/util/test_log.proto
-
-package org.tensorflow.proto.util.testlog;
-
-/**
- * Protobuf type {@code tensorflow.CPUInfo}
- */
-public  final class CPUInfo extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.CPUInfo)
-    CPUInfoOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use CPUInfo.newBuilder() to construct.
-  private CPUInfo(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private CPUInfo() {
-    cpuInfo_ = "";
-    cpuGovernor_ = "";
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new CPUInfo();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private CPUInfo(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    int mutable_bitField0_ = 0;
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 8: {
-
-            numCores_ = input.readInt64();
-            break;
-          }
-          case 16: {
-
-            numCoresAllowed_ = input.readInt64();
-            break;
-          }
-          case 25: {
-
-            mhzPerCpu_ = input.readDouble();
-            break;
-          }
-          case 34: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            cpuInfo_ = s;
-            break;
-          }
-          case 42: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            cpuGovernor_ = s;
-            break;
-          }
-          case 50: {
-            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
-              cacheSize_ = com.google.protobuf.MapField.newMapField(
-                  CacheSizeDefaultEntryHolder.defaultEntry);
-              mutable_bitField0_ |= 0x00000001;
-            }
-            com.google.protobuf.MapEntry<java.lang.String, java.lang.Long>
-            cacheSize__ = input.readMessage(
-                CacheSizeDefaultEntryHolder.defaultEntry.getParserForType(), extensionRegistry);
-            cacheSize_.getMutableMap().put(
-                cacheSize__.getKey(), cacheSize__.getValue());
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_CPUInfo_descriptor;
-  }
-
-  @SuppressWarnings({"rawtypes"})
-  @java.lang.Override
-  protected com.google.protobuf.MapField internalGetMapField(
-      int number) {
-    switch (number) {
-      case 6:
-        return internalGetCacheSize();
-      default:
-        throw new RuntimeException(
-            "Invalid map field number: " + number);
-    }
-  }
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_CPUInfo_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.util.testlog.CPUInfo.class, org.tensorflow.proto.util.testlog.CPUInfo.Builder.class);
-  }
-
-  public static final int NUM_CORES_FIELD_NUMBER = 1;
-  private long numCores_;
-  /**
-   * <code>int64 num_cores = 1;</code>
-   */
-  public long getNumCores() {
-    return numCores_;
-  }
-
-  public static final int NUM_CORES_ALLOWED_FIELD_NUMBER = 2;
-  private long numCoresAllowed_;
-  /**
-   * <code>int64 num_cores_allowed = 2;</code>
-   */
-  public long getNumCoresAllowed() {
-    return numCoresAllowed_;
-  }
-
-  public static final int MHZ_PER_CPU_FIELD_NUMBER = 3;
-  private double mhzPerCpu_;
-  /**
-   * <pre>
-   * How fast are these cpus?
-   * </pre>
-   *
-   * <code>double mhz_per_cpu = 3;</code>
-   */
-  public double getMhzPerCpu() {
-    return mhzPerCpu_;
-  }
-
-  public static final int CPU_INFO_FIELD_NUMBER = 4;
-  private volatile java.lang.Object cpuInfo_;
-  /**
-   * <pre>
-   * Additional cpu information. For example,
-   * Intel Ivybridge with HyperThreading (24 cores) dL1:32KB dL2:256KB dL3:30MB
-   * </pre>
-   *
-   * <code>string cpu_info = 4;</code>
-   */
-  public java.lang.String getCpuInfo() {
-    java.lang.Object ref = cpuInfo_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      cpuInfo_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * Additional cpu information. For example,
-   * Intel Ivybridge with HyperThreading (24 cores) dL1:32KB dL2:256KB dL3:30MB
-   * </pre>
-   *
-   * <code>string cpu_info = 4;</code>
-   */
-  public com.google.protobuf.ByteString
-      getCpuInfoBytes() {
-    java.lang.Object ref = cpuInfo_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      cpuInfo_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int CPU_GOVERNOR_FIELD_NUMBER = 5;
-  private volatile java.lang.Object cpuGovernor_;
-  /**
-   * <pre>
-   * What kind of cpu scaling is enabled on the host.
-   * Examples include "performance", "ondemand", "conservative", "mixed".
-   * </pre>
-   *
-   * <code>string cpu_governor = 5;</code>
-   */
-  public java.lang.String getCpuGovernor() {
-    java.lang.Object ref = cpuGovernor_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      cpuGovernor_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * What kind of cpu scaling is enabled on the host.
-   * Examples include "performance", "ondemand", "conservative", "mixed".
-   * </pre>
-   *
-   * <code>string cpu_governor = 5;</code>
-   */
-  public com.google.protobuf.ByteString
-      getCpuGovernorBytes() {
-    java.lang.Object ref = cpuGovernor_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      cpuGovernor_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int CACHE_SIZE_FIELD_NUMBER = 6;
-  private static final class CacheSizeDefaultEntryHolder {
-    static final com.google.protobuf.MapEntry<
-        java.lang.String, java.lang.Long> defaultEntry =
-            com.google.protobuf.MapEntry
-            .<java.lang.String, java.lang.Long>newDefaultInstance(
-                org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_CPUInfo_CacheSizeEntry_descriptor, 
-                com.google.protobuf.WireFormat.FieldType.STRING,
-                "",
-                com.google.protobuf.WireFormat.FieldType.INT64,
-                0L);
-  }
-  private com.google.protobuf.MapField<
-      java.lang.String, java.lang.Long> cacheSize_;
-  private com.google.protobuf.MapField<java.lang.String, java.lang.Long>
-  internalGetCacheSize() {
-    if (cacheSize_ == null) {
-      return com.google.protobuf.MapField.emptyMapField(
-          CacheSizeDefaultEntryHolder.defaultEntry);
-    }
-    return cacheSize_;
-  }
-
-  public int getCacheSizeCount() {
-    return internalGetCacheSize().getMap().size();
-  }
-  /**
-   * <pre>
-   * Cache sizes (in bytes), e.g. "L2": 262144 (for 256KB)
-   * </pre>
-   *
-   * <code>map&lt;string, int64&gt; cache_size = 6;</code>
-   */
-
-  public boolean containsCacheSize(
-      java.lang.String key) {
-    if (key == null) { throw new java.lang.NullPointerException(); }
-    return internalGetCacheSize().getMap().containsKey(key);
-  }
-  /**
-   * Use {@link #getCacheSizeMap()} instead.
-   */
-  @java.lang.Deprecated
-  public java.util.Map<java.lang.String, java.lang.Long> getCacheSize() {
-    return getCacheSizeMap();
-  }
-  /**
-   * <pre>
-   * Cache sizes (in bytes), e.g. "L2": 262144 (for 256KB)
-   * </pre>
-   *
-   * <code>map&lt;string, int64&gt; cache_size = 6;</code>
-   */
-
-  public java.util.Map<java.lang.String, java.lang.Long> getCacheSizeMap() {
-    return internalGetCacheSize().getMap();
-  }
-  /**
-   * <pre>
-   * Cache sizes (in bytes), e.g. "L2": 262144 (for 256KB)
-   * </pre>
-   *
-   * <code>map&lt;string, int64&gt; cache_size = 6;</code>
-   */
-
-  public long getCacheSizeOrDefault(
-      java.lang.String key,
-      long defaultValue) {
-    if (key == null) { throw new java.lang.NullPointerException(); }
-    java.util.Map<java.lang.String, java.lang.Long> map =
-        internalGetCacheSize().getMap();
-    return map.containsKey(key) ? map.get(key) : defaultValue;
-  }
-  /**
-   * <pre>
-   * Cache sizes (in bytes), e.g. "L2": 262144 (for 256KB)
-   * </pre>
-   *
-   * <code>map&lt;string, int64&gt; cache_size = 6;</code>
-   */
-
-  public long getCacheSizeOrThrow(
-      java.lang.String key) {
-    if (key == null) { throw new java.lang.NullPointerException(); }
-    java.util.Map<java.lang.String, java.lang.Long> map =
-        internalGetCacheSize().getMap();
-    if (!map.containsKey(key)) {
-      throw new java.lang.IllegalArgumentException();
-    }
-    return map.get(key);
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    if (numCores_ != 0L) {
-      output.writeInt64(1, numCores_);
-    }
-    if (numCoresAllowed_ != 0L) {
-      output.writeInt64(2, numCoresAllowed_);
-    }
-    if (mhzPerCpu_ != 0D) {
-      output.writeDouble(3, mhzPerCpu_);
-    }
-    if (!getCpuInfoBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 4, cpuInfo_);
-    }
-    if (!getCpuGovernorBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 5, cpuGovernor_);
-    }
-    com.google.protobuf.GeneratedMessageV3
-      .serializeStringMapTo(
-        output,
-        internalGetCacheSize(),
-        CacheSizeDefaultEntryHolder.defaultEntry,
-        6);
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (numCores_ != 0L) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt64Size(1, numCores_);
-    }
-    if (numCoresAllowed_ != 0L) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt64Size(2, numCoresAllowed_);
-    }
-    if (mhzPerCpu_ != 0D) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeDoubleSize(3, mhzPerCpu_);
-    }
-    if (!getCpuInfoBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(4, cpuInfo_);
-    }
-    if (!getCpuGovernorBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(5, cpuGovernor_);
-    }
-    for (java.util.Map.Entry<java.lang.String, java.lang.Long> entry
-         : internalGetCacheSize().getMap().entrySet()) {
-      com.google.protobuf.MapEntry<java.lang.String, java.lang.Long>
-      cacheSize__ = CacheSizeDefaultEntryHolder.defaultEntry.newBuilderForType()
-          .setKey(entry.getKey())
-          .setValue(entry.getValue())
-          .build();
-      size += com.google.protobuf.CodedOutputStream
-          .computeMessageSize(6, cacheSize__);
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.util.testlog.CPUInfo)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.util.testlog.CPUInfo other = (org.tensorflow.proto.util.testlog.CPUInfo) obj;
-
-    if (getNumCores()
-        != other.getNumCores()) return false;
-    if (getNumCoresAllowed()
-        != other.getNumCoresAllowed()) return false;
-    if (java.lang.Double.doubleToLongBits(getMhzPerCpu())
-        != java.lang.Double.doubleToLongBits(
-            other.getMhzPerCpu())) return false;
-    if (!getCpuInfo()
-        .equals(other.getCpuInfo())) return false;
-    if (!getCpuGovernor()
-        .equals(other.getCpuGovernor())) return false;
-    if (!internalGetCacheSize().equals(
-        other.internalGetCacheSize())) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    hash = (37 * hash) + NUM_CORES_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-        getNumCores());
-    hash = (37 * hash) + NUM_CORES_ALLOWED_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-        getNumCoresAllowed());
-    hash = (37 * hash) + MHZ_PER_CPU_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-        java.lang.Double.doubleToLongBits(getMhzPerCpu()));
-    hash = (37 * hash) + CPU_INFO_FIELD_NUMBER;
-    hash = (53 * hash) + getCpuInfo().hashCode();
-    hash = (37 * hash) + CPU_GOVERNOR_FIELD_NUMBER;
-    hash = (53 * hash) + getCpuGovernor().hashCode();
-    if (!internalGetCacheSize().getMap().isEmpty()) {
-      hash = (37 * hash) + CACHE_SIZE_FIELD_NUMBER;
-      hash = (53 * hash) + internalGetCacheSize().hashCode();
-    }
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.util.testlog.CPUInfo parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.testlog.CPUInfo parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.testlog.CPUInfo parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.testlog.CPUInfo parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.testlog.CPUInfo parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.testlog.CPUInfo parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.testlog.CPUInfo parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.testlog.CPUInfo parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.testlog.CPUInfo parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.testlog.CPUInfo parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.testlog.CPUInfo parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.testlog.CPUInfo parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.util.testlog.CPUInfo prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * Protobuf type {@code tensorflow.CPUInfo}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.CPUInfo)
-      org.tensorflow.proto.util.testlog.CPUInfoOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_CPUInfo_descriptor;
-    }
-
-    @SuppressWarnings({"rawtypes"})
-    protected com.google.protobuf.MapField internalGetMapField(
-        int number) {
-      switch (number) {
-        case 6:
-          return internalGetCacheSize();
-        default:
-          throw new RuntimeException(
-              "Invalid map field number: " + number);
-      }
-    }
-    @SuppressWarnings({"rawtypes"})
-    protected com.google.protobuf.MapField internalGetMutableMapField(
-        int number) {
-      switch (number) {
-        case 6:
-          return internalGetMutableCacheSize();
-        default:
-          throw new RuntimeException(
-              "Invalid map field number: " + number);
-      }
-    }
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_CPUInfo_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.util.testlog.CPUInfo.class, org.tensorflow.proto.util.testlog.CPUInfo.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.util.testlog.CPUInfo.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      numCores_ = 0L;
-
-      numCoresAllowed_ = 0L;
-
-      mhzPerCpu_ = 0D;
-
-      cpuInfo_ = "";
-
-      cpuGovernor_ = "";
-
-      internalGetMutableCacheSize().clear();
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_CPUInfo_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.testlog.CPUInfo getDefaultInstanceForType() {
-      return org.tensorflow.proto.util.testlog.CPUInfo.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.testlog.CPUInfo build() {
-      org.tensorflow.proto.util.testlog.CPUInfo result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.testlog.CPUInfo buildPartial() {
-      org.tensorflow.proto.util.testlog.CPUInfo result = new org.tensorflow.proto.util.testlog.CPUInfo(this);
-      int from_bitField0_ = bitField0_;
-      result.numCores_ = numCores_;
-      result.numCoresAllowed_ = numCoresAllowed_;
-      result.mhzPerCpu_ = mhzPerCpu_;
-      result.cpuInfo_ = cpuInfo_;
-      result.cpuGovernor_ = cpuGovernor_;
-      result.cacheSize_ = internalGetCacheSize();
-      result.cacheSize_.makeImmutable();
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.util.testlog.CPUInfo) {
-        return mergeFrom((org.tensorflow.proto.util.testlog.CPUInfo)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.util.testlog.CPUInfo other) {
-      if (other == org.tensorflow.proto.util.testlog.CPUInfo.getDefaultInstance()) return this;
-      if (other.getNumCores() != 0L) {
-        setNumCores(other.getNumCores());
-      }
-      if (other.getNumCoresAllowed() != 0L) {
-        setNumCoresAllowed(other.getNumCoresAllowed());
-      }
-      if (other.getMhzPerCpu() != 0D) {
-        setMhzPerCpu(other.getMhzPerCpu());
-      }
-      if (!other.getCpuInfo().isEmpty()) {
-        cpuInfo_ = other.cpuInfo_;
-        onChanged();
-      }
-      if (!other.getCpuGovernor().isEmpty()) {
-        cpuGovernor_ = other.cpuGovernor_;
-        onChanged();
-      }
-      internalGetMutableCacheSize().mergeFrom(
-          other.internalGetCacheSize());
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.util.testlog.CPUInfo parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.util.testlog.CPUInfo) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-    private int bitField0_;
-
-    private long numCores_ ;
-    /**
-     * <code>int64 num_cores = 1;</code>
-     */
-    public long getNumCores() {
-      return numCores_;
-    }
-    /**
-     * <code>int64 num_cores = 1;</code>
-     */
-    public Builder setNumCores(long value) {
-      
-      numCores_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>int64 num_cores = 1;</code>
-     */
-    public Builder clearNumCores() {
-      
-      numCores_ = 0L;
-      onChanged();
-      return this;
-    }
-
-    private long numCoresAllowed_ ;
-    /**
-     * <code>int64 num_cores_allowed = 2;</code>
-     */
-    public long getNumCoresAllowed() {
-      return numCoresAllowed_;
-    }
-    /**
-     * <code>int64 num_cores_allowed = 2;</code>
-     */
-    public Builder setNumCoresAllowed(long value) {
-      
-      numCoresAllowed_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>int64 num_cores_allowed = 2;</code>
-     */
-    public Builder clearNumCoresAllowed() {
-      
-      numCoresAllowed_ = 0L;
-      onChanged();
-      return this;
-    }
-
-    private double mhzPerCpu_ ;
-    /**
-     * <pre>
-     * How fast are these cpus?
-     * </pre>
-     *
-     * <code>double mhz_per_cpu = 3;</code>
-     */
-    public double getMhzPerCpu() {
-      return mhzPerCpu_;
-    }
-    /**
-     * <pre>
-     * How fast are these cpus?
-     * </pre>
-     *
-     * <code>double mhz_per_cpu = 3;</code>
-     */
-    public Builder setMhzPerCpu(double value) {
-      
-      mhzPerCpu_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * How fast are these cpus?
-     * </pre>
-     *
-     * <code>double mhz_per_cpu = 3;</code>
-     */
-    public Builder clearMhzPerCpu() {
-      
-      mhzPerCpu_ = 0D;
-      onChanged();
-      return this;
-    }
-
-    private java.lang.Object cpuInfo_ = "";
-    /**
-     * <pre>
-     * Additional cpu information. For example,
-     * Intel Ivybridge with HyperThreading (24 cores) dL1:32KB dL2:256KB dL3:30MB
-     * </pre>
-     *
-     * <code>string cpu_info = 4;</code>
-     */
-    public java.lang.String getCpuInfo() {
-      java.lang.Object ref = cpuInfo_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        cpuInfo_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Additional cpu information. For example,
-     * Intel Ivybridge with HyperThreading (24 cores) dL1:32KB dL2:256KB dL3:30MB
-     * </pre>
-     *
-     * <code>string cpu_info = 4;</code>
-     */
-    public com.google.protobuf.ByteString
-        getCpuInfoBytes() {
-      java.lang.Object ref = cpuInfo_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        cpuInfo_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Additional cpu information. For example,
-     * Intel Ivybridge with HyperThreading (24 cores) dL1:32KB dL2:256KB dL3:30MB
-     * </pre>
-     *
-     * <code>string cpu_info = 4;</code>
-     */
-    public Builder setCpuInfo(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      cpuInfo_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Additional cpu information. For example,
-     * Intel Ivybridge with HyperThreading (24 cores) dL1:32KB dL2:256KB dL3:30MB
-     * </pre>
-     *
-     * <code>string cpu_info = 4;</code>
-     */
-    public Builder clearCpuInfo() {
-      
-      cpuInfo_ = getDefaultInstance().getCpuInfo();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Additional cpu information. For example,
-     * Intel Ivybridge with HyperThreading (24 cores) dL1:32KB dL2:256KB dL3:30MB
-     * </pre>
-     *
-     * <code>string cpu_info = 4;</code>
-     */
-    public Builder setCpuInfoBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      cpuInfo_ = value;
-      onChanged();
-      return this;
-    }
-
-    private java.lang.Object cpuGovernor_ = "";
-    /**
-     * <pre>
-     * What kind of cpu scaling is enabled on the host.
-     * Examples include "performance", "ondemand", "conservative", "mixed".
-     * </pre>
-     *
-     * <code>string cpu_governor = 5;</code>
-     */
-    public java.lang.String getCpuGovernor() {
-      java.lang.Object ref = cpuGovernor_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        cpuGovernor_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * What kind of cpu scaling is enabled on the host.
-     * Examples include "performance", "ondemand", "conservative", "mixed".
-     * </pre>
-     *
-     * <code>string cpu_governor = 5;</code>
-     */
-    public com.google.protobuf.ByteString
-        getCpuGovernorBytes() {
-      java.lang.Object ref = cpuGovernor_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        cpuGovernor_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * What kind of cpu scaling is enabled on the host.
-     * Examples include "performance", "ondemand", "conservative", "mixed".
-     * </pre>
-     *
-     * <code>string cpu_governor = 5;</code>
-     */
-    public Builder setCpuGovernor(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      cpuGovernor_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * What kind of cpu scaling is enabled on the host.
-     * Examples include "performance", "ondemand", "conservative", "mixed".
-     * </pre>
-     *
-     * <code>string cpu_governor = 5;</code>
-     */
-    public Builder clearCpuGovernor() {
-      
-      cpuGovernor_ = getDefaultInstance().getCpuGovernor();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * What kind of cpu scaling is enabled on the host.
-     * Examples include "performance", "ondemand", "conservative", "mixed".
-     * </pre>
-     *
-     * <code>string cpu_governor = 5;</code>
-     */
-    public Builder setCpuGovernorBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      cpuGovernor_ = value;
-      onChanged();
-      return this;
-    }
-
-    private com.google.protobuf.MapField<
-        java.lang.String, java.lang.Long> cacheSize_;
-    private com.google.protobuf.MapField<java.lang.String, java.lang.Long>
-    internalGetCacheSize() {
-      if (cacheSize_ == null) {
-        return com.google.protobuf.MapField.emptyMapField(
-            CacheSizeDefaultEntryHolder.defaultEntry);
-      }
-      return cacheSize_;
-    }
-    private com.google.protobuf.MapField<java.lang.String, java.lang.Long>
-    internalGetMutableCacheSize() {
-      onChanged();;
-      if (cacheSize_ == null) {
-        cacheSize_ = com.google.protobuf.MapField.newMapField(
-            CacheSizeDefaultEntryHolder.defaultEntry);
-      }
-      if (!cacheSize_.isMutable()) {
-        cacheSize_ = cacheSize_.copy();
-      }
-      return cacheSize_;
-    }
-
-    public int getCacheSizeCount() {
-      return internalGetCacheSize().getMap().size();
-    }
-    /**
-     * <pre>
-     * Cache sizes (in bytes), e.g. "L2": 262144 (for 256KB)
-     * </pre>
-     *
-     * <code>map&lt;string, int64&gt; cache_size = 6;</code>
-     */
-
-    public boolean containsCacheSize(
-        java.lang.String key) {
-      if (key == null) { throw new java.lang.NullPointerException(); }
-      return internalGetCacheSize().getMap().containsKey(key);
-    }
-    /**
-     * Use {@link #getCacheSizeMap()} instead.
-     */
-    @java.lang.Deprecated
-    public java.util.Map<java.lang.String, java.lang.Long> getCacheSize() {
-      return getCacheSizeMap();
-    }
-    /**
-     * <pre>
-     * Cache sizes (in bytes), e.g. "L2": 262144 (for 256KB)
-     * </pre>
-     *
-     * <code>map&lt;string, int64&gt; cache_size = 6;</code>
-     */
-
-    public java.util.Map<java.lang.String, java.lang.Long> getCacheSizeMap() {
-      return internalGetCacheSize().getMap();
-    }
-    /**
-     * <pre>
-     * Cache sizes (in bytes), e.g. "L2": 262144 (for 256KB)
-     * </pre>
-     *
-     * <code>map&lt;string, int64&gt; cache_size = 6;</code>
-     */
-
-    public long getCacheSizeOrDefault(
-        java.lang.String key,
-        long defaultValue) {
-      if (key == null) { throw new java.lang.NullPointerException(); }
-      java.util.Map<java.lang.String, java.lang.Long> map =
-          internalGetCacheSize().getMap();
-      return map.containsKey(key) ? map.get(key) : defaultValue;
-    }
-    /**
-     * <pre>
-     * Cache sizes (in bytes), e.g. "L2": 262144 (for 256KB)
-     * </pre>
-     *
-     * <code>map&lt;string, int64&gt; cache_size = 6;</code>
-     */
-
-    public long getCacheSizeOrThrow(
-        java.lang.String key) {
-      if (key == null) { throw new java.lang.NullPointerException(); }
-      java.util.Map<java.lang.String, java.lang.Long> map =
-          internalGetCacheSize().getMap();
-      if (!map.containsKey(key)) {
-        throw new java.lang.IllegalArgumentException();
-      }
-      return map.get(key);
-    }
-
-    public Builder clearCacheSize() {
-      internalGetMutableCacheSize().getMutableMap()
-          .clear();
-      return this;
-    }
-    /**
-     * <pre>
-     * Cache sizes (in bytes), e.g. "L2": 262144 (for 256KB)
-     * </pre>
-     *
-     * <code>map&lt;string, int64&gt; cache_size = 6;</code>
-     */
-
-    public Builder removeCacheSize(
-        java.lang.String key) {
-      if (key == null) { throw new java.lang.NullPointerException(); }
-      internalGetMutableCacheSize().getMutableMap()
-          .remove(key);
-      return this;
-    }
-    /**
-     * Use alternate mutation accessors instead.
-     */
-    @java.lang.Deprecated
-    public java.util.Map<java.lang.String, java.lang.Long>
-    getMutableCacheSize() {
-      return internalGetMutableCacheSize().getMutableMap();
-    }
-    /**
-     * <pre>
-     * Cache sizes (in bytes), e.g. "L2": 262144 (for 256KB)
-     * </pre>
-     *
-     * <code>map&lt;string, int64&gt; cache_size = 6;</code>
-     */
-    public Builder putCacheSize(
-        java.lang.String key,
-        long value) {
-      if (key == null) { throw new java.lang.NullPointerException(); }
-      
-      internalGetMutableCacheSize().getMutableMap()
-          .put(key, value);
-      return this;
-    }
-    /**
-     * <pre>
-     * Cache sizes (in bytes), e.g. "L2": 262144 (for 256KB)
-     * </pre>
-     *
-     * <code>map&lt;string, int64&gt; cache_size = 6;</code>
-     */
-
-    public Builder putAllCacheSize(
-        java.util.Map<java.lang.String, java.lang.Long> values) {
-      internalGetMutableCacheSize().getMutableMap()
-          .putAll(values);
-      return this;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.CPUInfo)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.CPUInfo)
-  private static final org.tensorflow.proto.util.testlog.CPUInfo DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.util.testlog.CPUInfo();
-  }
-
-  public static org.tensorflow.proto.util.testlog.CPUInfo getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<CPUInfo>
-      PARSER = new com.google.protobuf.AbstractParser<CPUInfo>() {
-    @java.lang.Override
-    public CPUInfo parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new CPUInfo(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<CPUInfo> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<CPUInfo> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.util.testlog.CPUInfo getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/CPUInfoOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/CPUInfoOrBuilder.java
deleted file mode 100644
index be5a9556f41..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/CPUInfoOrBuilder.java
+++ /dev/null
@@ -1,122 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/util/test_log.proto
-
-package org.tensorflow.proto.util.testlog;
-
-public interface CPUInfoOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.CPUInfo)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <code>int64 num_cores = 1;</code>
-   */
-  long getNumCores();
-
-  /**
-   * <code>int64 num_cores_allowed = 2;</code>
-   */
-  long getNumCoresAllowed();
-
-  /**
-   * <pre>
-   * How fast are these cpus?
-   * </pre>
-   *
-   * <code>double mhz_per_cpu = 3;</code>
-   */
-  double getMhzPerCpu();
-
-  /**
-   * <pre>
-   * Additional cpu information. For example,
-   * Intel Ivybridge with HyperThreading (24 cores) dL1:32KB dL2:256KB dL3:30MB
-   * </pre>
-   *
-   * <code>string cpu_info = 4;</code>
-   */
-  java.lang.String getCpuInfo();
-  /**
-   * <pre>
-   * Additional cpu information. For example,
-   * Intel Ivybridge with HyperThreading (24 cores) dL1:32KB dL2:256KB dL3:30MB
-   * </pre>
-   *
-   * <code>string cpu_info = 4;</code>
-   */
-  com.google.protobuf.ByteString
-      getCpuInfoBytes();
-
-  /**
-   * <pre>
-   * What kind of cpu scaling is enabled on the host.
-   * Examples include "performance", "ondemand", "conservative", "mixed".
-   * </pre>
-   *
-   * <code>string cpu_governor = 5;</code>
-   */
-  java.lang.String getCpuGovernor();
-  /**
-   * <pre>
-   * What kind of cpu scaling is enabled on the host.
-   * Examples include "performance", "ondemand", "conservative", "mixed".
-   * </pre>
-   *
-   * <code>string cpu_governor = 5;</code>
-   */
-  com.google.protobuf.ByteString
-      getCpuGovernorBytes();
-
-  /**
-   * <pre>
-   * Cache sizes (in bytes), e.g. "L2": 262144 (for 256KB)
-   * </pre>
-   *
-   * <code>map&lt;string, int64&gt; cache_size = 6;</code>
-   */
-  int getCacheSizeCount();
-  /**
-   * <pre>
-   * Cache sizes (in bytes), e.g. "L2": 262144 (for 256KB)
-   * </pre>
-   *
-   * <code>map&lt;string, int64&gt; cache_size = 6;</code>
-   */
-  boolean containsCacheSize(
-      java.lang.String key);
-  /**
-   * Use {@link #getCacheSizeMap()} instead.
-   */
-  @java.lang.Deprecated
-  java.util.Map<java.lang.String, java.lang.Long>
-  getCacheSize();
-  /**
-   * <pre>
-   * Cache sizes (in bytes), e.g. "L2": 262144 (for 256KB)
-   * </pre>
-   *
-   * <code>map&lt;string, int64&gt; cache_size = 6;</code>
-   */
-  java.util.Map<java.lang.String, java.lang.Long>
-  getCacheSizeMap();
-  /**
-   * <pre>
-   * Cache sizes (in bytes), e.g. "L2": 262144 (for 256KB)
-   * </pre>
-   *
-   * <code>map&lt;string, int64&gt; cache_size = 6;</code>
-   */
-
-  long getCacheSizeOrDefault(
-      java.lang.String key,
-      long defaultValue);
-  /**
-   * <pre>
-   * Cache sizes (in bytes), e.g. "L2": 262144 (for 256KB)
-   * </pre>
-   *
-   * <code>map&lt;string, int64&gt; cache_size = 6;</code>
-   */
-
-  long getCacheSizeOrThrow(
-      java.lang.String key);
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/CommitId.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/CommitId.java
deleted file mode 100644
index beac981e730..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/CommitId.java
+++ /dev/null
@@ -1,964 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/util/test_log.proto
-
-package org.tensorflow.proto.util.testlog;
-
-/**
- * Protobuf type {@code tensorflow.CommitId}
- */
-public  final class CommitId extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.CommitId)
-    CommitIdOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use CommitId.newBuilder() to construct.
-  private CommitId(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private CommitId() {
-    snapshot_ = "";
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new CommitId();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private CommitId(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 8: {
-            kindCase_ = 1;
-            kind_ = input.readInt64();
-            break;
-          }
-          case 18: {
-            java.lang.String s = input.readStringRequireUtf8();
-            kindCase_ = 2;
-            kind_ = s;
-            break;
-          }
-          case 26: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            snapshot_ = s;
-            break;
-          }
-          case 32: {
-
-            pendingChangelist_ = input.readInt64();
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_CommitId_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_CommitId_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.util.testlog.CommitId.class, org.tensorflow.proto.util.testlog.CommitId.Builder.class);
-  }
-
-  private int kindCase_ = 0;
-  private java.lang.Object kind_;
-  public enum KindCase
-      implements com.google.protobuf.Internal.EnumLite {
-    CHANGELIST(1),
-    HASH(2),
-    KIND_NOT_SET(0);
-    private final int value;
-    private KindCase(int value) {
-      this.value = value;
-    }
-    /**
-     * @deprecated Use {@link #forNumber(int)} instead.
-     */
-    @java.lang.Deprecated
-    public static KindCase valueOf(int value) {
-      return forNumber(value);
-    }
-
-    public static KindCase forNumber(int value) {
-      switch (value) {
-        case 1: return CHANGELIST;
-        case 2: return HASH;
-        case 0: return KIND_NOT_SET;
-        default: return null;
-      }
-    }
-    public int getNumber() {
-      return this.value;
-    }
-  };
-
-  public KindCase
-  getKindCase() {
-    return KindCase.forNumber(
-        kindCase_);
-  }
-
-  public static final int CHANGELIST_FIELD_NUMBER = 1;
-  /**
-   * <pre>
-   * Submitted changelist.
-   * </pre>
-   *
-   * <code>int64 changelist = 1;</code>
-   */
-  public long getChangelist() {
-    if (kindCase_ == 1) {
-      return (java.lang.Long) kind_;
-    }
-    return 0L;
-  }
-
-  public static final int HASH_FIELD_NUMBER = 2;
-  /**
-   * <code>string hash = 2;</code>
-   */
-  public java.lang.String getHash() {
-    java.lang.Object ref = "";
-    if (kindCase_ == 2) {
-      ref = kind_;
-    }
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      if (kindCase_ == 2) {
-        kind_ = s;
-      }
-      return s;
-    }
-  }
-  /**
-   * <code>string hash = 2;</code>
-   */
-  public com.google.protobuf.ByteString
-      getHashBytes() {
-    java.lang.Object ref = "";
-    if (kindCase_ == 2) {
-      ref = kind_;
-    }
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      if (kindCase_ == 2) {
-        kind_ = b;
-      }
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int SNAPSHOT_FIELD_NUMBER = 3;
-  private volatile java.lang.Object snapshot_;
-  /**
-   * <pre>
-   * Hash of intermediate change between hash/changelist and what was tested.
-   * Not used if the build is from a commit without modifications.
-   * </pre>
-   *
-   * <code>string snapshot = 3;</code>
-   */
-  public java.lang.String getSnapshot() {
-    java.lang.Object ref = snapshot_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      snapshot_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * Hash of intermediate change between hash/changelist and what was tested.
-   * Not used if the build is from a commit without modifications.
-   * </pre>
-   *
-   * <code>string snapshot = 3;</code>
-   */
-  public com.google.protobuf.ByteString
-      getSnapshotBytes() {
-    java.lang.Object ref = snapshot_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      snapshot_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int PENDING_CHANGELIST_FIELD_NUMBER = 4;
-  private long pendingChangelist_;
-  /**
-   * <pre>
-   * Changelist tested if the change list is not already submitted.
-   * </pre>
-   *
-   * <code>int64 pending_changelist = 4;</code>
-   */
-  public long getPendingChangelist() {
-    return pendingChangelist_;
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    if (kindCase_ == 1) {
-      output.writeInt64(
-          1, (long)((java.lang.Long) kind_));
-    }
-    if (kindCase_ == 2) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 2, kind_);
-    }
-    if (!getSnapshotBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 3, snapshot_);
-    }
-    if (pendingChangelist_ != 0L) {
-      output.writeInt64(4, pendingChangelist_);
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (kindCase_ == 1) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt64Size(
-            1, (long)((java.lang.Long) kind_));
-    }
-    if (kindCase_ == 2) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(2, kind_);
-    }
-    if (!getSnapshotBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(3, snapshot_);
-    }
-    if (pendingChangelist_ != 0L) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt64Size(4, pendingChangelist_);
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.util.testlog.CommitId)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.util.testlog.CommitId other = (org.tensorflow.proto.util.testlog.CommitId) obj;
-
-    if (!getSnapshot()
-        .equals(other.getSnapshot())) return false;
-    if (getPendingChangelist()
-        != other.getPendingChangelist()) return false;
-    if (!getKindCase().equals(other.getKindCase())) return false;
-    switch (kindCase_) {
-      case 1:
-        if (getChangelist()
-            != other.getChangelist()) return false;
-        break;
-      case 2:
-        if (!getHash()
-            .equals(other.getHash())) return false;
-        break;
-      case 0:
-      default:
-    }
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    hash = (37 * hash) + SNAPSHOT_FIELD_NUMBER;
-    hash = (53 * hash) + getSnapshot().hashCode();
-    hash = (37 * hash) + PENDING_CHANGELIST_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-        getPendingChangelist());
-    switch (kindCase_) {
-      case 1:
-        hash = (37 * hash) + CHANGELIST_FIELD_NUMBER;
-        hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-            getChangelist());
-        break;
-      case 2:
-        hash = (37 * hash) + HASH_FIELD_NUMBER;
-        hash = (53 * hash) + getHash().hashCode();
-        break;
-      case 0:
-      default:
-    }
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.util.testlog.CommitId parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.testlog.CommitId parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.testlog.CommitId parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.testlog.CommitId parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.testlog.CommitId parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.testlog.CommitId parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.testlog.CommitId parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.testlog.CommitId parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.testlog.CommitId parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.testlog.CommitId parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.testlog.CommitId parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.testlog.CommitId parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.util.testlog.CommitId prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * Protobuf type {@code tensorflow.CommitId}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.CommitId)
-      org.tensorflow.proto.util.testlog.CommitIdOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_CommitId_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_CommitId_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.util.testlog.CommitId.class, org.tensorflow.proto.util.testlog.CommitId.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.util.testlog.CommitId.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      snapshot_ = "";
-
-      pendingChangelist_ = 0L;
-
-      kindCase_ = 0;
-      kind_ = null;
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_CommitId_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.testlog.CommitId getDefaultInstanceForType() {
-      return org.tensorflow.proto.util.testlog.CommitId.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.testlog.CommitId build() {
-      org.tensorflow.proto.util.testlog.CommitId result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.testlog.CommitId buildPartial() {
-      org.tensorflow.proto.util.testlog.CommitId result = new org.tensorflow.proto.util.testlog.CommitId(this);
-      if (kindCase_ == 1) {
-        result.kind_ = kind_;
-      }
-      if (kindCase_ == 2) {
-        result.kind_ = kind_;
-      }
-      result.snapshot_ = snapshot_;
-      result.pendingChangelist_ = pendingChangelist_;
-      result.kindCase_ = kindCase_;
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.util.testlog.CommitId) {
-        return mergeFrom((org.tensorflow.proto.util.testlog.CommitId)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.util.testlog.CommitId other) {
-      if (other == org.tensorflow.proto.util.testlog.CommitId.getDefaultInstance()) return this;
-      if (!other.getSnapshot().isEmpty()) {
-        snapshot_ = other.snapshot_;
-        onChanged();
-      }
-      if (other.getPendingChangelist() != 0L) {
-        setPendingChangelist(other.getPendingChangelist());
-      }
-      switch (other.getKindCase()) {
-        case CHANGELIST: {
-          setChangelist(other.getChangelist());
-          break;
-        }
-        case HASH: {
-          kindCase_ = 2;
-          kind_ = other.kind_;
-          onChanged();
-          break;
-        }
-        case KIND_NOT_SET: {
-          break;
-        }
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.util.testlog.CommitId parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.util.testlog.CommitId) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-    private int kindCase_ = 0;
-    private java.lang.Object kind_;
-    public KindCase
-        getKindCase() {
-      return KindCase.forNumber(
-          kindCase_);
-    }
-
-    public Builder clearKind() {
-      kindCase_ = 0;
-      kind_ = null;
-      onChanged();
-      return this;
-    }
-
-
-    /**
-     * <pre>
-     * Submitted changelist.
-     * </pre>
-     *
-     * <code>int64 changelist = 1;</code>
-     */
-    public long getChangelist() {
-      if (kindCase_ == 1) {
-        return (java.lang.Long) kind_;
-      }
-      return 0L;
-    }
-    /**
-     * <pre>
-     * Submitted changelist.
-     * </pre>
-     *
-     * <code>int64 changelist = 1;</code>
-     */
-    public Builder setChangelist(long value) {
-      kindCase_ = 1;
-      kind_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Submitted changelist.
-     * </pre>
-     *
-     * <code>int64 changelist = 1;</code>
-     */
-    public Builder clearChangelist() {
-      if (kindCase_ == 1) {
-        kindCase_ = 0;
-        kind_ = null;
-        onChanged();
-      }
-      return this;
-    }
-
-    /**
-     * <code>string hash = 2;</code>
-     */
-    public java.lang.String getHash() {
-      java.lang.Object ref = "";
-      if (kindCase_ == 2) {
-        ref = kind_;
-      }
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        if (kindCase_ == 2) {
-          kind_ = s;
-        }
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <code>string hash = 2;</code>
-     */
-    public com.google.protobuf.ByteString
-        getHashBytes() {
-      java.lang.Object ref = "";
-      if (kindCase_ == 2) {
-        ref = kind_;
-      }
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        if (kindCase_ == 2) {
-          kind_ = b;
-        }
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <code>string hash = 2;</code>
-     */
-    public Builder setHash(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  kindCase_ = 2;
-      kind_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>string hash = 2;</code>
-     */
-    public Builder clearHash() {
-      if (kindCase_ == 2) {
-        kindCase_ = 0;
-        kind_ = null;
-        onChanged();
-      }
-      return this;
-    }
-    /**
-     * <code>string hash = 2;</code>
-     */
-    public Builder setHashBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      kindCase_ = 2;
-      kind_ = value;
-      onChanged();
-      return this;
-    }
-
-    private java.lang.Object snapshot_ = "";
-    /**
-     * <pre>
-     * Hash of intermediate change between hash/changelist and what was tested.
-     * Not used if the build is from a commit without modifications.
-     * </pre>
-     *
-     * <code>string snapshot = 3;</code>
-     */
-    public java.lang.String getSnapshot() {
-      java.lang.Object ref = snapshot_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        snapshot_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Hash of intermediate change between hash/changelist and what was tested.
-     * Not used if the build is from a commit without modifications.
-     * </pre>
-     *
-     * <code>string snapshot = 3;</code>
-     */
-    public com.google.protobuf.ByteString
-        getSnapshotBytes() {
-      java.lang.Object ref = snapshot_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        snapshot_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Hash of intermediate change between hash/changelist and what was tested.
-     * Not used if the build is from a commit without modifications.
-     * </pre>
-     *
-     * <code>string snapshot = 3;</code>
-     */
-    public Builder setSnapshot(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      snapshot_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Hash of intermediate change between hash/changelist and what was tested.
-     * Not used if the build is from a commit without modifications.
-     * </pre>
-     *
-     * <code>string snapshot = 3;</code>
-     */
-    public Builder clearSnapshot() {
-      
-      snapshot_ = getDefaultInstance().getSnapshot();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Hash of intermediate change between hash/changelist and what was tested.
-     * Not used if the build is from a commit without modifications.
-     * </pre>
-     *
-     * <code>string snapshot = 3;</code>
-     */
-    public Builder setSnapshotBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      snapshot_ = value;
-      onChanged();
-      return this;
-    }
-
-    private long pendingChangelist_ ;
-    /**
-     * <pre>
-     * Changelist tested if the change list is not already submitted.
-     * </pre>
-     *
-     * <code>int64 pending_changelist = 4;</code>
-     */
-    public long getPendingChangelist() {
-      return pendingChangelist_;
-    }
-    /**
-     * <pre>
-     * Changelist tested if the change list is not already submitted.
-     * </pre>
-     *
-     * <code>int64 pending_changelist = 4;</code>
-     */
-    public Builder setPendingChangelist(long value) {
-      
-      pendingChangelist_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Changelist tested if the change list is not already submitted.
-     * </pre>
-     *
-     * <code>int64 pending_changelist = 4;</code>
-     */
-    public Builder clearPendingChangelist() {
-      
-      pendingChangelist_ = 0L;
-      onChanged();
-      return this;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.CommitId)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.CommitId)
-  private static final org.tensorflow.proto.util.testlog.CommitId DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.util.testlog.CommitId();
-  }
-
-  public static org.tensorflow.proto.util.testlog.CommitId getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<CommitId>
-      PARSER = new com.google.protobuf.AbstractParser<CommitId>() {
-    @java.lang.Override
-    public CommitId parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new CommitId(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<CommitId> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<CommitId> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.util.testlog.CommitId getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/CommitIdOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/CommitIdOrBuilder.java
deleted file mode 100644
index b63f4353379..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/CommitIdOrBuilder.java
+++ /dev/null
@@ -1,59 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/util/test_log.proto
-
-package org.tensorflow.proto.util.testlog;
-
-public interface CommitIdOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.CommitId)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <pre>
-   * Submitted changelist.
-   * </pre>
-   *
-   * <code>int64 changelist = 1;</code>
-   */
-  long getChangelist();
-
-  /**
-   * <code>string hash = 2;</code>
-   */
-  java.lang.String getHash();
-  /**
-   * <code>string hash = 2;</code>
-   */
-  com.google.protobuf.ByteString
-      getHashBytes();
-
-  /**
-   * <pre>
-   * Hash of intermediate change between hash/changelist and what was tested.
-   * Not used if the build is from a commit without modifications.
-   * </pre>
-   *
-   * <code>string snapshot = 3;</code>
-   */
-  java.lang.String getSnapshot();
-  /**
-   * <pre>
-   * Hash of intermediate change between hash/changelist and what was tested.
-   * Not used if the build is from a commit without modifications.
-   * </pre>
-   *
-   * <code>string snapshot = 3;</code>
-   */
-  com.google.protobuf.ByteString
-      getSnapshotBytes();
-
-  /**
-   * <pre>
-   * Changelist tested if the change list is not already submitted.
-   * </pre>
-   *
-   * <code>int64 pending_changelist = 4;</code>
-   */
-  long getPendingChangelist();
-
-  public org.tensorflow.proto.util.testlog.CommitId.KindCase getKindCase();
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/EntryValue.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/EntryValue.java
deleted file mode 100644
index d8ca9fffd45..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/EntryValue.java
+++ /dev/null
@@ -1,713 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/util/test_log.proto
-
-package org.tensorflow.proto.util.testlog;
-
-/**
- * Protobuf type {@code tensorflow.EntryValue}
- */
-public  final class EntryValue extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.EntryValue)
-    EntryValueOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use EntryValue.newBuilder() to construct.
-  private EntryValue(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private EntryValue() {
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new EntryValue();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private EntryValue(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 9: {
-            kindCase_ = 1;
-            kind_ = input.readDouble();
-            break;
-          }
-          case 18: {
-            java.lang.String s = input.readStringRequireUtf8();
-            kindCase_ = 2;
-            kind_ = s;
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_EntryValue_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_EntryValue_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.util.testlog.EntryValue.class, org.tensorflow.proto.util.testlog.EntryValue.Builder.class);
-  }
-
-  private int kindCase_ = 0;
-  private java.lang.Object kind_;
-  public enum KindCase
-      implements com.google.protobuf.Internal.EnumLite {
-    DOUBLE_VALUE(1),
-    STRING_VALUE(2),
-    KIND_NOT_SET(0);
-    private final int value;
-    private KindCase(int value) {
-      this.value = value;
-    }
-    /**
-     * @deprecated Use {@link #forNumber(int)} instead.
-     */
-    @java.lang.Deprecated
-    public static KindCase valueOf(int value) {
-      return forNumber(value);
-    }
-
-    public static KindCase forNumber(int value) {
-      switch (value) {
-        case 1: return DOUBLE_VALUE;
-        case 2: return STRING_VALUE;
-        case 0: return KIND_NOT_SET;
-        default: return null;
-      }
-    }
-    public int getNumber() {
-      return this.value;
-    }
-  };
-
-  public KindCase
-  getKindCase() {
-    return KindCase.forNumber(
-        kindCase_);
-  }
-
-  public static final int DOUBLE_VALUE_FIELD_NUMBER = 1;
-  /**
-   * <code>double double_value = 1;</code>
-   */
-  public double getDoubleValue() {
-    if (kindCase_ == 1) {
-      return (java.lang.Double) kind_;
-    }
-    return 0D;
-  }
-
-  public static final int STRING_VALUE_FIELD_NUMBER = 2;
-  /**
-   * <code>string string_value = 2;</code>
-   */
-  public java.lang.String getStringValue() {
-    java.lang.Object ref = "";
-    if (kindCase_ == 2) {
-      ref = kind_;
-    }
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      if (kindCase_ == 2) {
-        kind_ = s;
-      }
-      return s;
-    }
-  }
-  /**
-   * <code>string string_value = 2;</code>
-   */
-  public com.google.protobuf.ByteString
-      getStringValueBytes() {
-    java.lang.Object ref = "";
-    if (kindCase_ == 2) {
-      ref = kind_;
-    }
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      if (kindCase_ == 2) {
-        kind_ = b;
-      }
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    if (kindCase_ == 1) {
-      output.writeDouble(
-          1, (double)((java.lang.Double) kind_));
-    }
-    if (kindCase_ == 2) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 2, kind_);
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (kindCase_ == 1) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeDoubleSize(
-            1, (double)((java.lang.Double) kind_));
-    }
-    if (kindCase_ == 2) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(2, kind_);
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.util.testlog.EntryValue)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.util.testlog.EntryValue other = (org.tensorflow.proto.util.testlog.EntryValue) obj;
-
-    if (!getKindCase().equals(other.getKindCase())) return false;
-    switch (kindCase_) {
-      case 1:
-        if (java.lang.Double.doubleToLongBits(getDoubleValue())
-            != java.lang.Double.doubleToLongBits(
-                other.getDoubleValue())) return false;
-        break;
-      case 2:
-        if (!getStringValue()
-            .equals(other.getStringValue())) return false;
-        break;
-      case 0:
-      default:
-    }
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    switch (kindCase_) {
-      case 1:
-        hash = (37 * hash) + DOUBLE_VALUE_FIELD_NUMBER;
-        hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-            java.lang.Double.doubleToLongBits(getDoubleValue()));
-        break;
-      case 2:
-        hash = (37 * hash) + STRING_VALUE_FIELD_NUMBER;
-        hash = (53 * hash) + getStringValue().hashCode();
-        break;
-      case 0:
-      default:
-    }
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.util.testlog.EntryValue parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.testlog.EntryValue parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.testlog.EntryValue parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.testlog.EntryValue parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.testlog.EntryValue parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.testlog.EntryValue parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.testlog.EntryValue parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.testlog.EntryValue parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.testlog.EntryValue parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.testlog.EntryValue parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.testlog.EntryValue parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.testlog.EntryValue parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.util.testlog.EntryValue prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * Protobuf type {@code tensorflow.EntryValue}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.EntryValue)
-      org.tensorflow.proto.util.testlog.EntryValueOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_EntryValue_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_EntryValue_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.util.testlog.EntryValue.class, org.tensorflow.proto.util.testlog.EntryValue.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.util.testlog.EntryValue.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      kindCase_ = 0;
-      kind_ = null;
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_EntryValue_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.testlog.EntryValue getDefaultInstanceForType() {
-      return org.tensorflow.proto.util.testlog.EntryValue.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.testlog.EntryValue build() {
-      org.tensorflow.proto.util.testlog.EntryValue result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.testlog.EntryValue buildPartial() {
-      org.tensorflow.proto.util.testlog.EntryValue result = new org.tensorflow.proto.util.testlog.EntryValue(this);
-      if (kindCase_ == 1) {
-        result.kind_ = kind_;
-      }
-      if (kindCase_ == 2) {
-        result.kind_ = kind_;
-      }
-      result.kindCase_ = kindCase_;
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.util.testlog.EntryValue) {
-        return mergeFrom((org.tensorflow.proto.util.testlog.EntryValue)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.util.testlog.EntryValue other) {
-      if (other == org.tensorflow.proto.util.testlog.EntryValue.getDefaultInstance()) return this;
-      switch (other.getKindCase()) {
-        case DOUBLE_VALUE: {
-          setDoubleValue(other.getDoubleValue());
-          break;
-        }
-        case STRING_VALUE: {
-          kindCase_ = 2;
-          kind_ = other.kind_;
-          onChanged();
-          break;
-        }
-        case KIND_NOT_SET: {
-          break;
-        }
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.util.testlog.EntryValue parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.util.testlog.EntryValue) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-    private int kindCase_ = 0;
-    private java.lang.Object kind_;
-    public KindCase
-        getKindCase() {
-      return KindCase.forNumber(
-          kindCase_);
-    }
-
-    public Builder clearKind() {
-      kindCase_ = 0;
-      kind_ = null;
-      onChanged();
-      return this;
-    }
-
-
-    /**
-     * <code>double double_value = 1;</code>
-     */
-    public double getDoubleValue() {
-      if (kindCase_ == 1) {
-        return (java.lang.Double) kind_;
-      }
-      return 0D;
-    }
-    /**
-     * <code>double double_value = 1;</code>
-     */
-    public Builder setDoubleValue(double value) {
-      kindCase_ = 1;
-      kind_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>double double_value = 1;</code>
-     */
-    public Builder clearDoubleValue() {
-      if (kindCase_ == 1) {
-        kindCase_ = 0;
-        kind_ = null;
-        onChanged();
-      }
-      return this;
-    }
-
-    /**
-     * <code>string string_value = 2;</code>
-     */
-    public java.lang.String getStringValue() {
-      java.lang.Object ref = "";
-      if (kindCase_ == 2) {
-        ref = kind_;
-      }
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        if (kindCase_ == 2) {
-          kind_ = s;
-        }
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <code>string string_value = 2;</code>
-     */
-    public com.google.protobuf.ByteString
-        getStringValueBytes() {
-      java.lang.Object ref = "";
-      if (kindCase_ == 2) {
-        ref = kind_;
-      }
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        if (kindCase_ == 2) {
-          kind_ = b;
-        }
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <code>string string_value = 2;</code>
-     */
-    public Builder setStringValue(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  kindCase_ = 2;
-      kind_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>string string_value = 2;</code>
-     */
-    public Builder clearStringValue() {
-      if (kindCase_ == 2) {
-        kindCase_ = 0;
-        kind_ = null;
-        onChanged();
-      }
-      return this;
-    }
-    /**
-     * <code>string string_value = 2;</code>
-     */
-    public Builder setStringValueBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      kindCase_ = 2;
-      kind_ = value;
-      onChanged();
-      return this;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.EntryValue)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.EntryValue)
-  private static final org.tensorflow.proto.util.testlog.EntryValue DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.util.testlog.EntryValue();
-  }
-
-  public static org.tensorflow.proto.util.testlog.EntryValue getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<EntryValue>
-      PARSER = new com.google.protobuf.AbstractParser<EntryValue>() {
-    @java.lang.Override
-    public EntryValue parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new EntryValue(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<EntryValue> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<EntryValue> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.util.testlog.EntryValue getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/EntryValueOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/EntryValueOrBuilder.java
deleted file mode 100644
index 4a2d71e54ba..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/EntryValueOrBuilder.java
+++ /dev/null
@@ -1,26 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/util/test_log.proto
-
-package org.tensorflow.proto.util.testlog;
-
-public interface EntryValueOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.EntryValue)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <code>double double_value = 1;</code>
-   */
-  double getDoubleValue();
-
-  /**
-   * <code>string string_value = 2;</code>
-   */
-  java.lang.String getStringValue();
-  /**
-   * <code>string string_value = 2;</code>
-   */
-  com.google.protobuf.ByteString
-      getStringValueBytes();
-
-  public org.tensorflow.proto.util.testlog.EntryValue.KindCase getKindCase();
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/GPUInfo.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/GPUInfo.java
deleted file mode 100644
index 7f6878d36ad..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/GPUInfo.java
+++ /dev/null
@@ -1,884 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/util/test_log.proto
-
-package org.tensorflow.proto.util.testlog;
-
-/**
- * Protobuf type {@code tensorflow.GPUInfo}
- */
-public  final class GPUInfo extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.GPUInfo)
-    GPUInfoOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use GPUInfo.newBuilder() to construct.
-  private GPUInfo(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private GPUInfo() {
-    model_ = "";
-    uuid_ = "";
-    busId_ = "";
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new GPUInfo();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private GPUInfo(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 10: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            model_ = s;
-            break;
-          }
-          case 18: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            uuid_ = s;
-            break;
-          }
-          case 26: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            busId_ = s;
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_GPUInfo_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_GPUInfo_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.util.testlog.GPUInfo.class, org.tensorflow.proto.util.testlog.GPUInfo.Builder.class);
-  }
-
-  public static final int MODEL_FIELD_NUMBER = 1;
-  private volatile java.lang.Object model_;
-  /**
-   * <pre>
-   * e.g. "Tesla K40c"
-   * </pre>
-   *
-   * <code>string model = 1;</code>
-   */
-  public java.lang.String getModel() {
-    java.lang.Object ref = model_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      model_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * e.g. "Tesla K40c"
-   * </pre>
-   *
-   * <code>string model = 1;</code>
-   */
-  public com.google.protobuf.ByteString
-      getModelBytes() {
-    java.lang.Object ref = model_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      model_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int UUID_FIELD_NUMBER = 2;
-  private volatile java.lang.Object uuid_;
-  /**
-   * <pre>
-   * Final entry in output of "nvidia-smi -L"
-   * </pre>
-   *
-   * <code>string uuid = 2;</code>
-   */
-  public java.lang.String getUuid() {
-    java.lang.Object ref = uuid_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      uuid_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * Final entry in output of "nvidia-smi -L"
-   * </pre>
-   *
-   * <code>string uuid = 2;</code>
-   */
-  public com.google.protobuf.ByteString
-      getUuidBytes() {
-    java.lang.Object ref = uuid_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      uuid_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int BUS_ID_FIELD_NUMBER = 3;
-  private volatile java.lang.Object busId_;
-  /**
-   * <pre>
-   * e.g. "0000:04:00.0"
-   * </pre>
-   *
-   * <code>string bus_id = 3;</code>
-   */
-  public java.lang.String getBusId() {
-    java.lang.Object ref = busId_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      busId_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * e.g. "0000:04:00.0"
-   * </pre>
-   *
-   * <code>string bus_id = 3;</code>
-   */
-  public com.google.protobuf.ByteString
-      getBusIdBytes() {
-    java.lang.Object ref = busId_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      busId_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    if (!getModelBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 1, model_);
-    }
-    if (!getUuidBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 2, uuid_);
-    }
-    if (!getBusIdBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 3, busId_);
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (!getModelBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, model_);
-    }
-    if (!getUuidBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(2, uuid_);
-    }
-    if (!getBusIdBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(3, busId_);
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.util.testlog.GPUInfo)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.util.testlog.GPUInfo other = (org.tensorflow.proto.util.testlog.GPUInfo) obj;
-
-    if (!getModel()
-        .equals(other.getModel())) return false;
-    if (!getUuid()
-        .equals(other.getUuid())) return false;
-    if (!getBusId()
-        .equals(other.getBusId())) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    hash = (37 * hash) + MODEL_FIELD_NUMBER;
-    hash = (53 * hash) + getModel().hashCode();
-    hash = (37 * hash) + UUID_FIELD_NUMBER;
-    hash = (53 * hash) + getUuid().hashCode();
-    hash = (37 * hash) + BUS_ID_FIELD_NUMBER;
-    hash = (53 * hash) + getBusId().hashCode();
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.util.testlog.GPUInfo parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.testlog.GPUInfo parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.testlog.GPUInfo parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.testlog.GPUInfo parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.testlog.GPUInfo parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.testlog.GPUInfo parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.testlog.GPUInfo parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.testlog.GPUInfo parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.testlog.GPUInfo parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.testlog.GPUInfo parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.testlog.GPUInfo parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.testlog.GPUInfo parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.util.testlog.GPUInfo prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * Protobuf type {@code tensorflow.GPUInfo}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.GPUInfo)
-      org.tensorflow.proto.util.testlog.GPUInfoOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_GPUInfo_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_GPUInfo_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.util.testlog.GPUInfo.class, org.tensorflow.proto.util.testlog.GPUInfo.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.util.testlog.GPUInfo.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      model_ = "";
-
-      uuid_ = "";
-
-      busId_ = "";
-
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_GPUInfo_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.testlog.GPUInfo getDefaultInstanceForType() {
-      return org.tensorflow.proto.util.testlog.GPUInfo.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.testlog.GPUInfo build() {
-      org.tensorflow.proto.util.testlog.GPUInfo result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.testlog.GPUInfo buildPartial() {
-      org.tensorflow.proto.util.testlog.GPUInfo result = new org.tensorflow.proto.util.testlog.GPUInfo(this);
-      result.model_ = model_;
-      result.uuid_ = uuid_;
-      result.busId_ = busId_;
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.util.testlog.GPUInfo) {
-        return mergeFrom((org.tensorflow.proto.util.testlog.GPUInfo)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.util.testlog.GPUInfo other) {
-      if (other == org.tensorflow.proto.util.testlog.GPUInfo.getDefaultInstance()) return this;
-      if (!other.getModel().isEmpty()) {
-        model_ = other.model_;
-        onChanged();
-      }
-      if (!other.getUuid().isEmpty()) {
-        uuid_ = other.uuid_;
-        onChanged();
-      }
-      if (!other.getBusId().isEmpty()) {
-        busId_ = other.busId_;
-        onChanged();
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.util.testlog.GPUInfo parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.util.testlog.GPUInfo) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-
-    private java.lang.Object model_ = "";
-    /**
-     * <pre>
-     * e.g. "Tesla K40c"
-     * </pre>
-     *
-     * <code>string model = 1;</code>
-     */
-    public java.lang.String getModel() {
-      java.lang.Object ref = model_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        model_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * e.g. "Tesla K40c"
-     * </pre>
-     *
-     * <code>string model = 1;</code>
-     */
-    public com.google.protobuf.ByteString
-        getModelBytes() {
-      java.lang.Object ref = model_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        model_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * e.g. "Tesla K40c"
-     * </pre>
-     *
-     * <code>string model = 1;</code>
-     */
-    public Builder setModel(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      model_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * e.g. "Tesla K40c"
-     * </pre>
-     *
-     * <code>string model = 1;</code>
-     */
-    public Builder clearModel() {
-      
-      model_ = getDefaultInstance().getModel();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * e.g. "Tesla K40c"
-     * </pre>
-     *
-     * <code>string model = 1;</code>
-     */
-    public Builder setModelBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      model_ = value;
-      onChanged();
-      return this;
-    }
-
-    private java.lang.Object uuid_ = "";
-    /**
-     * <pre>
-     * Final entry in output of "nvidia-smi -L"
-     * </pre>
-     *
-     * <code>string uuid = 2;</code>
-     */
-    public java.lang.String getUuid() {
-      java.lang.Object ref = uuid_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        uuid_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Final entry in output of "nvidia-smi -L"
-     * </pre>
-     *
-     * <code>string uuid = 2;</code>
-     */
-    public com.google.protobuf.ByteString
-        getUuidBytes() {
-      java.lang.Object ref = uuid_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        uuid_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Final entry in output of "nvidia-smi -L"
-     * </pre>
-     *
-     * <code>string uuid = 2;</code>
-     */
-    public Builder setUuid(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      uuid_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Final entry in output of "nvidia-smi -L"
-     * </pre>
-     *
-     * <code>string uuid = 2;</code>
-     */
-    public Builder clearUuid() {
-      
-      uuid_ = getDefaultInstance().getUuid();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Final entry in output of "nvidia-smi -L"
-     * </pre>
-     *
-     * <code>string uuid = 2;</code>
-     */
-    public Builder setUuidBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      uuid_ = value;
-      onChanged();
-      return this;
-    }
-
-    private java.lang.Object busId_ = "";
-    /**
-     * <pre>
-     * e.g. "0000:04:00.0"
-     * </pre>
-     *
-     * <code>string bus_id = 3;</code>
-     */
-    public java.lang.String getBusId() {
-      java.lang.Object ref = busId_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        busId_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * e.g. "0000:04:00.0"
-     * </pre>
-     *
-     * <code>string bus_id = 3;</code>
-     */
-    public com.google.protobuf.ByteString
-        getBusIdBytes() {
-      java.lang.Object ref = busId_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        busId_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * e.g. "0000:04:00.0"
-     * </pre>
-     *
-     * <code>string bus_id = 3;</code>
-     */
-    public Builder setBusId(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      busId_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * e.g. "0000:04:00.0"
-     * </pre>
-     *
-     * <code>string bus_id = 3;</code>
-     */
-    public Builder clearBusId() {
-      
-      busId_ = getDefaultInstance().getBusId();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * e.g. "0000:04:00.0"
-     * </pre>
-     *
-     * <code>string bus_id = 3;</code>
-     */
-    public Builder setBusIdBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      busId_ = value;
-      onChanged();
-      return this;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.GPUInfo)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.GPUInfo)
-  private static final org.tensorflow.proto.util.testlog.GPUInfo DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.util.testlog.GPUInfo();
-  }
-
-  public static org.tensorflow.proto.util.testlog.GPUInfo getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<GPUInfo>
-      PARSER = new com.google.protobuf.AbstractParser<GPUInfo>() {
-    @java.lang.Override
-    public GPUInfo parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new GPUInfo(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<GPUInfo> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<GPUInfo> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.util.testlog.GPUInfo getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/GPUInfoOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/GPUInfoOrBuilder.java
deleted file mode 100644
index 44985f7d574..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/GPUInfoOrBuilder.java
+++ /dev/null
@@ -1,63 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/util/test_log.proto
-
-package org.tensorflow.proto.util.testlog;
-
-public interface GPUInfoOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.GPUInfo)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <pre>
-   * e.g. "Tesla K40c"
-   * </pre>
-   *
-   * <code>string model = 1;</code>
-   */
-  java.lang.String getModel();
-  /**
-   * <pre>
-   * e.g. "Tesla K40c"
-   * </pre>
-   *
-   * <code>string model = 1;</code>
-   */
-  com.google.protobuf.ByteString
-      getModelBytes();
-
-  /**
-   * <pre>
-   * Final entry in output of "nvidia-smi -L"
-   * </pre>
-   *
-   * <code>string uuid = 2;</code>
-   */
-  java.lang.String getUuid();
-  /**
-   * <pre>
-   * Final entry in output of "nvidia-smi -L"
-   * </pre>
-   *
-   * <code>string uuid = 2;</code>
-   */
-  com.google.protobuf.ByteString
-      getUuidBytes();
-
-  /**
-   * <pre>
-   * e.g. "0000:04:00.0"
-   * </pre>
-   *
-   * <code>string bus_id = 3;</code>
-   */
-  java.lang.String getBusId();
-  /**
-   * <pre>
-   * e.g. "0000:04:00.0"
-   * </pre>
-   *
-   * <code>string bus_id = 3;</code>
-   */
-  com.google.protobuf.ByteString
-      getBusIdBytes();
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/MachineConfiguration.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/MachineConfiguration.java
deleted file mode 100644
index 70b9d4a3d7c..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/MachineConfiguration.java
+++ /dev/null
@@ -1,2241 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/util/test_log.proto
-
-package org.tensorflow.proto.util.testlog;
-
-/**
- * Protobuf type {@code tensorflow.MachineConfiguration}
- */
-public  final class MachineConfiguration extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.MachineConfiguration)
-    MachineConfigurationOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use MachineConfiguration.newBuilder() to construct.
-  private MachineConfiguration(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private MachineConfiguration() {
-    hostname_ = "";
-    serialIdentifier_ = "";
-    deviceInfo_ = java.util.Collections.emptyList();
-    availableDeviceInfo_ = java.util.Collections.emptyList();
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new MachineConfiguration();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private MachineConfiguration(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    int mutable_bitField0_ = 0;
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 10: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            hostname_ = s;
-            break;
-          }
-          case 18: {
-            org.tensorflow.proto.util.testlog.PlatformInfo.Builder subBuilder = null;
-            if (platformInfo_ != null) {
-              subBuilder = platformInfo_.toBuilder();
-            }
-            platformInfo_ = input.readMessage(org.tensorflow.proto.util.testlog.PlatformInfo.parser(), extensionRegistry);
-            if (subBuilder != null) {
-              subBuilder.mergeFrom(platformInfo_);
-              platformInfo_ = subBuilder.buildPartial();
-            }
-
-            break;
-          }
-          case 26: {
-            org.tensorflow.proto.util.testlog.CPUInfo.Builder subBuilder = null;
-            if (cpuInfo_ != null) {
-              subBuilder = cpuInfo_.toBuilder();
-            }
-            cpuInfo_ = input.readMessage(org.tensorflow.proto.util.testlog.CPUInfo.parser(), extensionRegistry);
-            if (subBuilder != null) {
-              subBuilder.mergeFrom(cpuInfo_);
-              cpuInfo_ = subBuilder.buildPartial();
-            }
-
-            break;
-          }
-          case 34: {
-            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
-              deviceInfo_ = new java.util.ArrayList<com.google.protobuf.Any>();
-              mutable_bitField0_ |= 0x00000001;
-            }
-            deviceInfo_.add(
-                input.readMessage(com.google.protobuf.Any.parser(), extensionRegistry));
-            break;
-          }
-          case 42: {
-            if (!((mutable_bitField0_ & 0x00000002) != 0)) {
-              availableDeviceInfo_ = new java.util.ArrayList<org.tensorflow.proto.util.testlog.AvailableDeviceInfo>();
-              mutable_bitField0_ |= 0x00000002;
-            }
-            availableDeviceInfo_.add(
-                input.readMessage(org.tensorflow.proto.util.testlog.AvailableDeviceInfo.parser(), extensionRegistry));
-            break;
-          }
-          case 50: {
-            org.tensorflow.proto.util.testlog.MemoryInfo.Builder subBuilder = null;
-            if (memoryInfo_ != null) {
-              subBuilder = memoryInfo_.toBuilder();
-            }
-            memoryInfo_ = input.readMessage(org.tensorflow.proto.util.testlog.MemoryInfo.parser(), extensionRegistry);
-            if (subBuilder != null) {
-              subBuilder.mergeFrom(memoryInfo_);
-              memoryInfo_ = subBuilder.buildPartial();
-            }
-
-            break;
-          }
-          case 58: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            serialIdentifier_ = s;
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      if (((mutable_bitField0_ & 0x00000001) != 0)) {
-        deviceInfo_ = java.util.Collections.unmodifiableList(deviceInfo_);
-      }
-      if (((mutable_bitField0_ & 0x00000002) != 0)) {
-        availableDeviceInfo_ = java.util.Collections.unmodifiableList(availableDeviceInfo_);
-      }
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_MachineConfiguration_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_MachineConfiguration_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.util.testlog.MachineConfiguration.class, org.tensorflow.proto.util.testlog.MachineConfiguration.Builder.class);
-  }
-
-  public static final int HOSTNAME_FIELD_NUMBER = 1;
-  private volatile java.lang.Object hostname_;
-  /**
-   * <pre>
-   * Host name of machine that ran the benchmark.
-   * </pre>
-   *
-   * <code>string hostname = 1;</code>
-   */
-  public java.lang.String getHostname() {
-    java.lang.Object ref = hostname_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      hostname_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * Host name of machine that ran the benchmark.
-   * </pre>
-   *
-   * <code>string hostname = 1;</code>
-   */
-  public com.google.protobuf.ByteString
-      getHostnameBytes() {
-    java.lang.Object ref = hostname_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      hostname_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int SERIAL_IDENTIFIER_FIELD_NUMBER = 7;
-  private volatile java.lang.Object serialIdentifier_;
-  /**
-   * <pre>
-   * Unique serial number of the machine.
-   * </pre>
-   *
-   * <code>string serial_identifier = 7;</code>
-   */
-  public java.lang.String getSerialIdentifier() {
-    java.lang.Object ref = serialIdentifier_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      serialIdentifier_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * Unique serial number of the machine.
-   * </pre>
-   *
-   * <code>string serial_identifier = 7;</code>
-   */
-  public com.google.protobuf.ByteString
-      getSerialIdentifierBytes() {
-    java.lang.Object ref = serialIdentifier_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      serialIdentifier_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int PLATFORM_INFO_FIELD_NUMBER = 2;
-  private org.tensorflow.proto.util.testlog.PlatformInfo platformInfo_;
-  /**
-   * <pre>
-   * Additional platform information.
-   * </pre>
-   *
-   * <code>.tensorflow.PlatformInfo platform_info = 2;</code>
-   */
-  public boolean hasPlatformInfo() {
-    return platformInfo_ != null;
-  }
-  /**
-   * <pre>
-   * Additional platform information.
-   * </pre>
-   *
-   * <code>.tensorflow.PlatformInfo platform_info = 2;</code>
-   */
-  public org.tensorflow.proto.util.testlog.PlatformInfo getPlatformInfo() {
-    return platformInfo_ == null ? org.tensorflow.proto.util.testlog.PlatformInfo.getDefaultInstance() : platformInfo_;
-  }
-  /**
-   * <pre>
-   * Additional platform information.
-   * </pre>
-   *
-   * <code>.tensorflow.PlatformInfo platform_info = 2;</code>
-   */
-  public org.tensorflow.proto.util.testlog.PlatformInfoOrBuilder getPlatformInfoOrBuilder() {
-    return getPlatformInfo();
-  }
-
-  public static final int CPU_INFO_FIELD_NUMBER = 3;
-  private org.tensorflow.proto.util.testlog.CPUInfo cpuInfo_;
-  /**
-   * <pre>
-   * CPU Information.
-   * </pre>
-   *
-   * <code>.tensorflow.CPUInfo cpu_info = 3;</code>
-   */
-  public boolean hasCpuInfo() {
-    return cpuInfo_ != null;
-  }
-  /**
-   * <pre>
-   * CPU Information.
-   * </pre>
-   *
-   * <code>.tensorflow.CPUInfo cpu_info = 3;</code>
-   */
-  public org.tensorflow.proto.util.testlog.CPUInfo getCpuInfo() {
-    return cpuInfo_ == null ? org.tensorflow.proto.util.testlog.CPUInfo.getDefaultInstance() : cpuInfo_;
-  }
-  /**
-   * <pre>
-   * CPU Information.
-   * </pre>
-   *
-   * <code>.tensorflow.CPUInfo cpu_info = 3;</code>
-   */
-  public org.tensorflow.proto.util.testlog.CPUInfoOrBuilder getCpuInfoOrBuilder() {
-    return getCpuInfo();
-  }
-
-  public static final int DEVICE_INFO_FIELD_NUMBER = 4;
-  private java.util.List<com.google.protobuf.Any> deviceInfo_;
-  /**
-   * <pre>
-   * Other devices that are attached and relevant (e.g. GPUInfo).
-   * </pre>
-   *
-   * <code>repeated .google.protobuf.Any device_info = 4;</code>
-   */
-  public java.util.List<com.google.protobuf.Any> getDeviceInfoList() {
-    return deviceInfo_;
-  }
-  /**
-   * <pre>
-   * Other devices that are attached and relevant (e.g. GPUInfo).
-   * </pre>
-   *
-   * <code>repeated .google.protobuf.Any device_info = 4;</code>
-   */
-  public java.util.List<? extends com.google.protobuf.AnyOrBuilder> 
-      getDeviceInfoOrBuilderList() {
-    return deviceInfo_;
-  }
-  /**
-   * <pre>
-   * Other devices that are attached and relevant (e.g. GPUInfo).
-   * </pre>
-   *
-   * <code>repeated .google.protobuf.Any device_info = 4;</code>
-   */
-  public int getDeviceInfoCount() {
-    return deviceInfo_.size();
-  }
-  /**
-   * <pre>
-   * Other devices that are attached and relevant (e.g. GPUInfo).
-   * </pre>
-   *
-   * <code>repeated .google.protobuf.Any device_info = 4;</code>
-   */
-  public com.google.protobuf.Any getDeviceInfo(int index) {
-    return deviceInfo_.get(index);
-  }
-  /**
-   * <pre>
-   * Other devices that are attached and relevant (e.g. GPUInfo).
-   * </pre>
-   *
-   * <code>repeated .google.protobuf.Any device_info = 4;</code>
-   */
-  public com.google.protobuf.AnyOrBuilder getDeviceInfoOrBuilder(
-      int index) {
-    return deviceInfo_.get(index);
-  }
-
-  public static final int AVAILABLE_DEVICE_INFO_FIELD_NUMBER = 5;
-  private java.util.List<org.tensorflow.proto.util.testlog.AvailableDeviceInfo> availableDeviceInfo_;
-  /**
-   * <pre>
-   * Devices accessible to the test (e.g. as given by list_local_devices).
-   * </pre>
-   *
-   * <code>repeated .tensorflow.AvailableDeviceInfo available_device_info = 5;</code>
-   */
-  public java.util.List<org.tensorflow.proto.util.testlog.AvailableDeviceInfo> getAvailableDeviceInfoList() {
-    return availableDeviceInfo_;
-  }
-  /**
-   * <pre>
-   * Devices accessible to the test (e.g. as given by list_local_devices).
-   * </pre>
-   *
-   * <code>repeated .tensorflow.AvailableDeviceInfo available_device_info = 5;</code>
-   */
-  public java.util.List<? extends org.tensorflow.proto.util.testlog.AvailableDeviceInfoOrBuilder> 
-      getAvailableDeviceInfoOrBuilderList() {
-    return availableDeviceInfo_;
-  }
-  /**
-   * <pre>
-   * Devices accessible to the test (e.g. as given by list_local_devices).
-   * </pre>
-   *
-   * <code>repeated .tensorflow.AvailableDeviceInfo available_device_info = 5;</code>
-   */
-  public int getAvailableDeviceInfoCount() {
-    return availableDeviceInfo_.size();
-  }
-  /**
-   * <pre>
-   * Devices accessible to the test (e.g. as given by list_local_devices).
-   * </pre>
-   *
-   * <code>repeated .tensorflow.AvailableDeviceInfo available_device_info = 5;</code>
-   */
-  public org.tensorflow.proto.util.testlog.AvailableDeviceInfo getAvailableDeviceInfo(int index) {
-    return availableDeviceInfo_.get(index);
-  }
-  /**
-   * <pre>
-   * Devices accessible to the test (e.g. as given by list_local_devices).
-   * </pre>
-   *
-   * <code>repeated .tensorflow.AvailableDeviceInfo available_device_info = 5;</code>
-   */
-  public org.tensorflow.proto.util.testlog.AvailableDeviceInfoOrBuilder getAvailableDeviceInfoOrBuilder(
-      int index) {
-    return availableDeviceInfo_.get(index);
-  }
-
-  public static final int MEMORY_INFO_FIELD_NUMBER = 6;
-  private org.tensorflow.proto.util.testlog.MemoryInfo memoryInfo_;
-  /**
-   * <code>.tensorflow.MemoryInfo memory_info = 6;</code>
-   */
-  public boolean hasMemoryInfo() {
-    return memoryInfo_ != null;
-  }
-  /**
-   * <code>.tensorflow.MemoryInfo memory_info = 6;</code>
-   */
-  public org.tensorflow.proto.util.testlog.MemoryInfo getMemoryInfo() {
-    return memoryInfo_ == null ? org.tensorflow.proto.util.testlog.MemoryInfo.getDefaultInstance() : memoryInfo_;
-  }
-  /**
-   * <code>.tensorflow.MemoryInfo memory_info = 6;</code>
-   */
-  public org.tensorflow.proto.util.testlog.MemoryInfoOrBuilder getMemoryInfoOrBuilder() {
-    return getMemoryInfo();
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    if (!getHostnameBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 1, hostname_);
-    }
-    if (platformInfo_ != null) {
-      output.writeMessage(2, getPlatformInfo());
-    }
-    if (cpuInfo_ != null) {
-      output.writeMessage(3, getCpuInfo());
-    }
-    for (int i = 0; i < deviceInfo_.size(); i++) {
-      output.writeMessage(4, deviceInfo_.get(i));
-    }
-    for (int i = 0; i < availableDeviceInfo_.size(); i++) {
-      output.writeMessage(5, availableDeviceInfo_.get(i));
-    }
-    if (memoryInfo_ != null) {
-      output.writeMessage(6, getMemoryInfo());
-    }
-    if (!getSerialIdentifierBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 7, serialIdentifier_);
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (!getHostnameBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, hostname_);
-    }
-    if (platformInfo_ != null) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(2, getPlatformInfo());
-    }
-    if (cpuInfo_ != null) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(3, getCpuInfo());
-    }
-    for (int i = 0; i < deviceInfo_.size(); i++) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(4, deviceInfo_.get(i));
-    }
-    for (int i = 0; i < availableDeviceInfo_.size(); i++) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(5, availableDeviceInfo_.get(i));
-    }
-    if (memoryInfo_ != null) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(6, getMemoryInfo());
-    }
-    if (!getSerialIdentifierBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(7, serialIdentifier_);
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.util.testlog.MachineConfiguration)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.util.testlog.MachineConfiguration other = (org.tensorflow.proto.util.testlog.MachineConfiguration) obj;
-
-    if (!getHostname()
-        .equals(other.getHostname())) return false;
-    if (!getSerialIdentifier()
-        .equals(other.getSerialIdentifier())) return false;
-    if (hasPlatformInfo() != other.hasPlatformInfo()) return false;
-    if (hasPlatformInfo()) {
-      if (!getPlatformInfo()
-          .equals(other.getPlatformInfo())) return false;
-    }
-    if (hasCpuInfo() != other.hasCpuInfo()) return false;
-    if (hasCpuInfo()) {
-      if (!getCpuInfo()
-          .equals(other.getCpuInfo())) return false;
-    }
-    if (!getDeviceInfoList()
-        .equals(other.getDeviceInfoList())) return false;
-    if (!getAvailableDeviceInfoList()
-        .equals(other.getAvailableDeviceInfoList())) return false;
-    if (hasMemoryInfo() != other.hasMemoryInfo()) return false;
-    if (hasMemoryInfo()) {
-      if (!getMemoryInfo()
-          .equals(other.getMemoryInfo())) return false;
-    }
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    hash = (37 * hash) + HOSTNAME_FIELD_NUMBER;
-    hash = (53 * hash) + getHostname().hashCode();
-    hash = (37 * hash) + SERIAL_IDENTIFIER_FIELD_NUMBER;
-    hash = (53 * hash) + getSerialIdentifier().hashCode();
-    if (hasPlatformInfo()) {
-      hash = (37 * hash) + PLATFORM_INFO_FIELD_NUMBER;
-      hash = (53 * hash) + getPlatformInfo().hashCode();
-    }
-    if (hasCpuInfo()) {
-      hash = (37 * hash) + CPU_INFO_FIELD_NUMBER;
-      hash = (53 * hash) + getCpuInfo().hashCode();
-    }
-    if (getDeviceInfoCount() > 0) {
-      hash = (37 * hash) + DEVICE_INFO_FIELD_NUMBER;
-      hash = (53 * hash) + getDeviceInfoList().hashCode();
-    }
-    if (getAvailableDeviceInfoCount() > 0) {
-      hash = (37 * hash) + AVAILABLE_DEVICE_INFO_FIELD_NUMBER;
-      hash = (53 * hash) + getAvailableDeviceInfoList().hashCode();
-    }
-    if (hasMemoryInfo()) {
-      hash = (37 * hash) + MEMORY_INFO_FIELD_NUMBER;
-      hash = (53 * hash) + getMemoryInfo().hashCode();
-    }
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.util.testlog.MachineConfiguration parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.testlog.MachineConfiguration parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.testlog.MachineConfiguration parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.testlog.MachineConfiguration parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.testlog.MachineConfiguration parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.testlog.MachineConfiguration parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.testlog.MachineConfiguration parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.testlog.MachineConfiguration parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.testlog.MachineConfiguration parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.testlog.MachineConfiguration parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.testlog.MachineConfiguration parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.testlog.MachineConfiguration parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.util.testlog.MachineConfiguration prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * Protobuf type {@code tensorflow.MachineConfiguration}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.MachineConfiguration)
-      org.tensorflow.proto.util.testlog.MachineConfigurationOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_MachineConfiguration_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_MachineConfiguration_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.util.testlog.MachineConfiguration.class, org.tensorflow.proto.util.testlog.MachineConfiguration.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.util.testlog.MachineConfiguration.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-        getDeviceInfoFieldBuilder();
-        getAvailableDeviceInfoFieldBuilder();
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      hostname_ = "";
-
-      serialIdentifier_ = "";
-
-      if (platformInfoBuilder_ == null) {
-        platformInfo_ = null;
-      } else {
-        platformInfo_ = null;
-        platformInfoBuilder_ = null;
-      }
-      if (cpuInfoBuilder_ == null) {
-        cpuInfo_ = null;
-      } else {
-        cpuInfo_ = null;
-        cpuInfoBuilder_ = null;
-      }
-      if (deviceInfoBuilder_ == null) {
-        deviceInfo_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000001);
-      } else {
-        deviceInfoBuilder_.clear();
-      }
-      if (availableDeviceInfoBuilder_ == null) {
-        availableDeviceInfo_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000002);
-      } else {
-        availableDeviceInfoBuilder_.clear();
-      }
-      if (memoryInfoBuilder_ == null) {
-        memoryInfo_ = null;
-      } else {
-        memoryInfo_ = null;
-        memoryInfoBuilder_ = null;
-      }
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_MachineConfiguration_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.testlog.MachineConfiguration getDefaultInstanceForType() {
-      return org.tensorflow.proto.util.testlog.MachineConfiguration.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.testlog.MachineConfiguration build() {
-      org.tensorflow.proto.util.testlog.MachineConfiguration result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.testlog.MachineConfiguration buildPartial() {
-      org.tensorflow.proto.util.testlog.MachineConfiguration result = new org.tensorflow.proto.util.testlog.MachineConfiguration(this);
-      int from_bitField0_ = bitField0_;
-      result.hostname_ = hostname_;
-      result.serialIdentifier_ = serialIdentifier_;
-      if (platformInfoBuilder_ == null) {
-        result.platformInfo_ = platformInfo_;
-      } else {
-        result.platformInfo_ = platformInfoBuilder_.build();
-      }
-      if (cpuInfoBuilder_ == null) {
-        result.cpuInfo_ = cpuInfo_;
-      } else {
-        result.cpuInfo_ = cpuInfoBuilder_.build();
-      }
-      if (deviceInfoBuilder_ == null) {
-        if (((bitField0_ & 0x00000001) != 0)) {
-          deviceInfo_ = java.util.Collections.unmodifiableList(deviceInfo_);
-          bitField0_ = (bitField0_ & ~0x00000001);
-        }
-        result.deviceInfo_ = deviceInfo_;
-      } else {
-        result.deviceInfo_ = deviceInfoBuilder_.build();
-      }
-      if (availableDeviceInfoBuilder_ == null) {
-        if (((bitField0_ & 0x00000002) != 0)) {
-          availableDeviceInfo_ = java.util.Collections.unmodifiableList(availableDeviceInfo_);
-          bitField0_ = (bitField0_ & ~0x00000002);
-        }
-        result.availableDeviceInfo_ = availableDeviceInfo_;
-      } else {
-        result.availableDeviceInfo_ = availableDeviceInfoBuilder_.build();
-      }
-      if (memoryInfoBuilder_ == null) {
-        result.memoryInfo_ = memoryInfo_;
-      } else {
-        result.memoryInfo_ = memoryInfoBuilder_.build();
-      }
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.util.testlog.MachineConfiguration) {
-        return mergeFrom((org.tensorflow.proto.util.testlog.MachineConfiguration)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.util.testlog.MachineConfiguration other) {
-      if (other == org.tensorflow.proto.util.testlog.MachineConfiguration.getDefaultInstance()) return this;
-      if (!other.getHostname().isEmpty()) {
-        hostname_ = other.hostname_;
-        onChanged();
-      }
-      if (!other.getSerialIdentifier().isEmpty()) {
-        serialIdentifier_ = other.serialIdentifier_;
-        onChanged();
-      }
-      if (other.hasPlatformInfo()) {
-        mergePlatformInfo(other.getPlatformInfo());
-      }
-      if (other.hasCpuInfo()) {
-        mergeCpuInfo(other.getCpuInfo());
-      }
-      if (deviceInfoBuilder_ == null) {
-        if (!other.deviceInfo_.isEmpty()) {
-          if (deviceInfo_.isEmpty()) {
-            deviceInfo_ = other.deviceInfo_;
-            bitField0_ = (bitField0_ & ~0x00000001);
-          } else {
-            ensureDeviceInfoIsMutable();
-            deviceInfo_.addAll(other.deviceInfo_);
-          }
-          onChanged();
-        }
-      } else {
-        if (!other.deviceInfo_.isEmpty()) {
-          if (deviceInfoBuilder_.isEmpty()) {
-            deviceInfoBuilder_.dispose();
-            deviceInfoBuilder_ = null;
-            deviceInfo_ = other.deviceInfo_;
-            bitField0_ = (bitField0_ & ~0x00000001);
-            deviceInfoBuilder_ = 
-              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
-                 getDeviceInfoFieldBuilder() : null;
-          } else {
-            deviceInfoBuilder_.addAllMessages(other.deviceInfo_);
-          }
-        }
-      }
-      if (availableDeviceInfoBuilder_ == null) {
-        if (!other.availableDeviceInfo_.isEmpty()) {
-          if (availableDeviceInfo_.isEmpty()) {
-            availableDeviceInfo_ = other.availableDeviceInfo_;
-            bitField0_ = (bitField0_ & ~0x00000002);
-          } else {
-            ensureAvailableDeviceInfoIsMutable();
-            availableDeviceInfo_.addAll(other.availableDeviceInfo_);
-          }
-          onChanged();
-        }
-      } else {
-        if (!other.availableDeviceInfo_.isEmpty()) {
-          if (availableDeviceInfoBuilder_.isEmpty()) {
-            availableDeviceInfoBuilder_.dispose();
-            availableDeviceInfoBuilder_ = null;
-            availableDeviceInfo_ = other.availableDeviceInfo_;
-            bitField0_ = (bitField0_ & ~0x00000002);
-            availableDeviceInfoBuilder_ = 
-              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
-                 getAvailableDeviceInfoFieldBuilder() : null;
-          } else {
-            availableDeviceInfoBuilder_.addAllMessages(other.availableDeviceInfo_);
-          }
-        }
-      }
-      if (other.hasMemoryInfo()) {
-        mergeMemoryInfo(other.getMemoryInfo());
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.util.testlog.MachineConfiguration parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.util.testlog.MachineConfiguration) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-    private int bitField0_;
-
-    private java.lang.Object hostname_ = "";
-    /**
-     * <pre>
-     * Host name of machine that ran the benchmark.
-     * </pre>
-     *
-     * <code>string hostname = 1;</code>
-     */
-    public java.lang.String getHostname() {
-      java.lang.Object ref = hostname_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        hostname_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Host name of machine that ran the benchmark.
-     * </pre>
-     *
-     * <code>string hostname = 1;</code>
-     */
-    public com.google.protobuf.ByteString
-        getHostnameBytes() {
-      java.lang.Object ref = hostname_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        hostname_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Host name of machine that ran the benchmark.
-     * </pre>
-     *
-     * <code>string hostname = 1;</code>
-     */
-    public Builder setHostname(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      hostname_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Host name of machine that ran the benchmark.
-     * </pre>
-     *
-     * <code>string hostname = 1;</code>
-     */
-    public Builder clearHostname() {
-      
-      hostname_ = getDefaultInstance().getHostname();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Host name of machine that ran the benchmark.
-     * </pre>
-     *
-     * <code>string hostname = 1;</code>
-     */
-    public Builder setHostnameBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      hostname_ = value;
-      onChanged();
-      return this;
-    }
-
-    private java.lang.Object serialIdentifier_ = "";
-    /**
-     * <pre>
-     * Unique serial number of the machine.
-     * </pre>
-     *
-     * <code>string serial_identifier = 7;</code>
-     */
-    public java.lang.String getSerialIdentifier() {
-      java.lang.Object ref = serialIdentifier_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        serialIdentifier_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Unique serial number of the machine.
-     * </pre>
-     *
-     * <code>string serial_identifier = 7;</code>
-     */
-    public com.google.protobuf.ByteString
-        getSerialIdentifierBytes() {
-      java.lang.Object ref = serialIdentifier_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        serialIdentifier_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Unique serial number of the machine.
-     * </pre>
-     *
-     * <code>string serial_identifier = 7;</code>
-     */
-    public Builder setSerialIdentifier(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      serialIdentifier_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Unique serial number of the machine.
-     * </pre>
-     *
-     * <code>string serial_identifier = 7;</code>
-     */
-    public Builder clearSerialIdentifier() {
-      
-      serialIdentifier_ = getDefaultInstance().getSerialIdentifier();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Unique serial number of the machine.
-     * </pre>
-     *
-     * <code>string serial_identifier = 7;</code>
-     */
-    public Builder setSerialIdentifierBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      serialIdentifier_ = value;
-      onChanged();
-      return this;
-    }
-
-    private org.tensorflow.proto.util.testlog.PlatformInfo platformInfo_;
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.util.testlog.PlatformInfo, org.tensorflow.proto.util.testlog.PlatformInfo.Builder, org.tensorflow.proto.util.testlog.PlatformInfoOrBuilder> platformInfoBuilder_;
-    /**
-     * <pre>
-     * Additional platform information.
-     * </pre>
-     *
-     * <code>.tensorflow.PlatformInfo platform_info = 2;</code>
-     */
-    public boolean hasPlatformInfo() {
-      return platformInfoBuilder_ != null || platformInfo_ != null;
-    }
-    /**
-     * <pre>
-     * Additional platform information.
-     * </pre>
-     *
-     * <code>.tensorflow.PlatformInfo platform_info = 2;</code>
-     */
-    public org.tensorflow.proto.util.testlog.PlatformInfo getPlatformInfo() {
-      if (platformInfoBuilder_ == null) {
-        return platformInfo_ == null ? org.tensorflow.proto.util.testlog.PlatformInfo.getDefaultInstance() : platformInfo_;
-      } else {
-        return platformInfoBuilder_.getMessage();
-      }
-    }
-    /**
-     * <pre>
-     * Additional platform information.
-     * </pre>
-     *
-     * <code>.tensorflow.PlatformInfo platform_info = 2;</code>
-     */
-    public Builder setPlatformInfo(org.tensorflow.proto.util.testlog.PlatformInfo value) {
-      if (platformInfoBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        platformInfo_ = value;
-        onChanged();
-      } else {
-        platformInfoBuilder_.setMessage(value);
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Additional platform information.
-     * </pre>
-     *
-     * <code>.tensorflow.PlatformInfo platform_info = 2;</code>
-     */
-    public Builder setPlatformInfo(
-        org.tensorflow.proto.util.testlog.PlatformInfo.Builder builderForValue) {
-      if (platformInfoBuilder_ == null) {
-        platformInfo_ = builderForValue.build();
-        onChanged();
-      } else {
-        platformInfoBuilder_.setMessage(builderForValue.build());
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Additional platform information.
-     * </pre>
-     *
-     * <code>.tensorflow.PlatformInfo platform_info = 2;</code>
-     */
-    public Builder mergePlatformInfo(org.tensorflow.proto.util.testlog.PlatformInfo value) {
-      if (platformInfoBuilder_ == null) {
-        if (platformInfo_ != null) {
-          platformInfo_ =
-            org.tensorflow.proto.util.testlog.PlatformInfo.newBuilder(platformInfo_).mergeFrom(value).buildPartial();
-        } else {
-          platformInfo_ = value;
-        }
-        onChanged();
-      } else {
-        platformInfoBuilder_.mergeFrom(value);
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Additional platform information.
-     * </pre>
-     *
-     * <code>.tensorflow.PlatformInfo platform_info = 2;</code>
-     */
-    public Builder clearPlatformInfo() {
-      if (platformInfoBuilder_ == null) {
-        platformInfo_ = null;
-        onChanged();
-      } else {
-        platformInfo_ = null;
-        platformInfoBuilder_ = null;
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Additional platform information.
-     * </pre>
-     *
-     * <code>.tensorflow.PlatformInfo platform_info = 2;</code>
-     */
-    public org.tensorflow.proto.util.testlog.PlatformInfo.Builder getPlatformInfoBuilder() {
-      
-      onChanged();
-      return getPlatformInfoFieldBuilder().getBuilder();
-    }
-    /**
-     * <pre>
-     * Additional platform information.
-     * </pre>
-     *
-     * <code>.tensorflow.PlatformInfo platform_info = 2;</code>
-     */
-    public org.tensorflow.proto.util.testlog.PlatformInfoOrBuilder getPlatformInfoOrBuilder() {
-      if (platformInfoBuilder_ != null) {
-        return platformInfoBuilder_.getMessageOrBuilder();
-      } else {
-        return platformInfo_ == null ?
-            org.tensorflow.proto.util.testlog.PlatformInfo.getDefaultInstance() : platformInfo_;
-      }
-    }
-    /**
-     * <pre>
-     * Additional platform information.
-     * </pre>
-     *
-     * <code>.tensorflow.PlatformInfo platform_info = 2;</code>
-     */
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.util.testlog.PlatformInfo, org.tensorflow.proto.util.testlog.PlatformInfo.Builder, org.tensorflow.proto.util.testlog.PlatformInfoOrBuilder> 
-        getPlatformInfoFieldBuilder() {
-      if (platformInfoBuilder_ == null) {
-        platformInfoBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-            org.tensorflow.proto.util.testlog.PlatformInfo, org.tensorflow.proto.util.testlog.PlatformInfo.Builder, org.tensorflow.proto.util.testlog.PlatformInfoOrBuilder>(
-                getPlatformInfo(),
-                getParentForChildren(),
-                isClean());
-        platformInfo_ = null;
-      }
-      return platformInfoBuilder_;
-    }
-
-    private org.tensorflow.proto.util.testlog.CPUInfo cpuInfo_;
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.util.testlog.CPUInfo, org.tensorflow.proto.util.testlog.CPUInfo.Builder, org.tensorflow.proto.util.testlog.CPUInfoOrBuilder> cpuInfoBuilder_;
-    /**
-     * <pre>
-     * CPU Information.
-     * </pre>
-     *
-     * <code>.tensorflow.CPUInfo cpu_info = 3;</code>
-     */
-    public boolean hasCpuInfo() {
-      return cpuInfoBuilder_ != null || cpuInfo_ != null;
-    }
-    /**
-     * <pre>
-     * CPU Information.
-     * </pre>
-     *
-     * <code>.tensorflow.CPUInfo cpu_info = 3;</code>
-     */
-    public org.tensorflow.proto.util.testlog.CPUInfo getCpuInfo() {
-      if (cpuInfoBuilder_ == null) {
-        return cpuInfo_ == null ? org.tensorflow.proto.util.testlog.CPUInfo.getDefaultInstance() : cpuInfo_;
-      } else {
-        return cpuInfoBuilder_.getMessage();
-      }
-    }
-    /**
-     * <pre>
-     * CPU Information.
-     * </pre>
-     *
-     * <code>.tensorflow.CPUInfo cpu_info = 3;</code>
-     */
-    public Builder setCpuInfo(org.tensorflow.proto.util.testlog.CPUInfo value) {
-      if (cpuInfoBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        cpuInfo_ = value;
-        onChanged();
-      } else {
-        cpuInfoBuilder_.setMessage(value);
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * CPU Information.
-     * </pre>
-     *
-     * <code>.tensorflow.CPUInfo cpu_info = 3;</code>
-     */
-    public Builder setCpuInfo(
-        org.tensorflow.proto.util.testlog.CPUInfo.Builder builderForValue) {
-      if (cpuInfoBuilder_ == null) {
-        cpuInfo_ = builderForValue.build();
-        onChanged();
-      } else {
-        cpuInfoBuilder_.setMessage(builderForValue.build());
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * CPU Information.
-     * </pre>
-     *
-     * <code>.tensorflow.CPUInfo cpu_info = 3;</code>
-     */
-    public Builder mergeCpuInfo(org.tensorflow.proto.util.testlog.CPUInfo value) {
-      if (cpuInfoBuilder_ == null) {
-        if (cpuInfo_ != null) {
-          cpuInfo_ =
-            org.tensorflow.proto.util.testlog.CPUInfo.newBuilder(cpuInfo_).mergeFrom(value).buildPartial();
-        } else {
-          cpuInfo_ = value;
-        }
-        onChanged();
-      } else {
-        cpuInfoBuilder_.mergeFrom(value);
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * CPU Information.
-     * </pre>
-     *
-     * <code>.tensorflow.CPUInfo cpu_info = 3;</code>
-     */
-    public Builder clearCpuInfo() {
-      if (cpuInfoBuilder_ == null) {
-        cpuInfo_ = null;
-        onChanged();
-      } else {
-        cpuInfo_ = null;
-        cpuInfoBuilder_ = null;
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * CPU Information.
-     * </pre>
-     *
-     * <code>.tensorflow.CPUInfo cpu_info = 3;</code>
-     */
-    public org.tensorflow.proto.util.testlog.CPUInfo.Builder getCpuInfoBuilder() {
-      
-      onChanged();
-      return getCpuInfoFieldBuilder().getBuilder();
-    }
-    /**
-     * <pre>
-     * CPU Information.
-     * </pre>
-     *
-     * <code>.tensorflow.CPUInfo cpu_info = 3;</code>
-     */
-    public org.tensorflow.proto.util.testlog.CPUInfoOrBuilder getCpuInfoOrBuilder() {
-      if (cpuInfoBuilder_ != null) {
-        return cpuInfoBuilder_.getMessageOrBuilder();
-      } else {
-        return cpuInfo_ == null ?
-            org.tensorflow.proto.util.testlog.CPUInfo.getDefaultInstance() : cpuInfo_;
-      }
-    }
-    /**
-     * <pre>
-     * CPU Information.
-     * </pre>
-     *
-     * <code>.tensorflow.CPUInfo cpu_info = 3;</code>
-     */
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.util.testlog.CPUInfo, org.tensorflow.proto.util.testlog.CPUInfo.Builder, org.tensorflow.proto.util.testlog.CPUInfoOrBuilder> 
-        getCpuInfoFieldBuilder() {
-      if (cpuInfoBuilder_ == null) {
-        cpuInfoBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-            org.tensorflow.proto.util.testlog.CPUInfo, org.tensorflow.proto.util.testlog.CPUInfo.Builder, org.tensorflow.proto.util.testlog.CPUInfoOrBuilder>(
-                getCpuInfo(),
-                getParentForChildren(),
-                isClean());
-        cpuInfo_ = null;
-      }
-      return cpuInfoBuilder_;
-    }
-
-    private java.util.List<com.google.protobuf.Any> deviceInfo_ =
-      java.util.Collections.emptyList();
-    private void ensureDeviceInfoIsMutable() {
-      if (!((bitField0_ & 0x00000001) != 0)) {
-        deviceInfo_ = new java.util.ArrayList<com.google.protobuf.Any>(deviceInfo_);
-        bitField0_ |= 0x00000001;
-       }
-    }
-
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        com.google.protobuf.Any, com.google.protobuf.Any.Builder, com.google.protobuf.AnyOrBuilder> deviceInfoBuilder_;
-
-    /**
-     * <pre>
-     * Other devices that are attached and relevant (e.g. GPUInfo).
-     * </pre>
-     *
-     * <code>repeated .google.protobuf.Any device_info = 4;</code>
-     */
-    public java.util.List<com.google.protobuf.Any> getDeviceInfoList() {
-      if (deviceInfoBuilder_ == null) {
-        return java.util.Collections.unmodifiableList(deviceInfo_);
-      } else {
-        return deviceInfoBuilder_.getMessageList();
-      }
-    }
-    /**
-     * <pre>
-     * Other devices that are attached and relevant (e.g. GPUInfo).
-     * </pre>
-     *
-     * <code>repeated .google.protobuf.Any device_info = 4;</code>
-     */
-    public int getDeviceInfoCount() {
-      if (deviceInfoBuilder_ == null) {
-        return deviceInfo_.size();
-      } else {
-        return deviceInfoBuilder_.getCount();
-      }
-    }
-    /**
-     * <pre>
-     * Other devices that are attached and relevant (e.g. GPUInfo).
-     * </pre>
-     *
-     * <code>repeated .google.protobuf.Any device_info = 4;</code>
-     */
-    public com.google.protobuf.Any getDeviceInfo(int index) {
-      if (deviceInfoBuilder_ == null) {
-        return deviceInfo_.get(index);
-      } else {
-        return deviceInfoBuilder_.getMessage(index);
-      }
-    }
-    /**
-     * <pre>
-     * Other devices that are attached and relevant (e.g. GPUInfo).
-     * </pre>
-     *
-     * <code>repeated .google.protobuf.Any device_info = 4;</code>
-     */
-    public Builder setDeviceInfo(
-        int index, com.google.protobuf.Any value) {
-      if (deviceInfoBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureDeviceInfoIsMutable();
-        deviceInfo_.set(index, value);
-        onChanged();
-      } else {
-        deviceInfoBuilder_.setMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Other devices that are attached and relevant (e.g. GPUInfo).
-     * </pre>
-     *
-     * <code>repeated .google.protobuf.Any device_info = 4;</code>
-     */
-    public Builder setDeviceInfo(
-        int index, com.google.protobuf.Any.Builder builderForValue) {
-      if (deviceInfoBuilder_ == null) {
-        ensureDeviceInfoIsMutable();
-        deviceInfo_.set(index, builderForValue.build());
-        onChanged();
-      } else {
-        deviceInfoBuilder_.setMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Other devices that are attached and relevant (e.g. GPUInfo).
-     * </pre>
-     *
-     * <code>repeated .google.protobuf.Any device_info = 4;</code>
-     */
-    public Builder addDeviceInfo(com.google.protobuf.Any value) {
-      if (deviceInfoBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureDeviceInfoIsMutable();
-        deviceInfo_.add(value);
-        onChanged();
-      } else {
-        deviceInfoBuilder_.addMessage(value);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Other devices that are attached and relevant (e.g. GPUInfo).
-     * </pre>
-     *
-     * <code>repeated .google.protobuf.Any device_info = 4;</code>
-     */
-    public Builder addDeviceInfo(
-        int index, com.google.protobuf.Any value) {
-      if (deviceInfoBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureDeviceInfoIsMutable();
-        deviceInfo_.add(index, value);
-        onChanged();
-      } else {
-        deviceInfoBuilder_.addMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Other devices that are attached and relevant (e.g. GPUInfo).
-     * </pre>
-     *
-     * <code>repeated .google.protobuf.Any device_info = 4;</code>
-     */
-    public Builder addDeviceInfo(
-        com.google.protobuf.Any.Builder builderForValue) {
-      if (deviceInfoBuilder_ == null) {
-        ensureDeviceInfoIsMutable();
-        deviceInfo_.add(builderForValue.build());
-        onChanged();
-      } else {
-        deviceInfoBuilder_.addMessage(builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Other devices that are attached and relevant (e.g. GPUInfo).
-     * </pre>
-     *
-     * <code>repeated .google.protobuf.Any device_info = 4;</code>
-     */
-    public Builder addDeviceInfo(
-        int index, com.google.protobuf.Any.Builder builderForValue) {
-      if (deviceInfoBuilder_ == null) {
-        ensureDeviceInfoIsMutable();
-        deviceInfo_.add(index, builderForValue.build());
-        onChanged();
-      } else {
-        deviceInfoBuilder_.addMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Other devices that are attached and relevant (e.g. GPUInfo).
-     * </pre>
-     *
-     * <code>repeated .google.protobuf.Any device_info = 4;</code>
-     */
-    public Builder addAllDeviceInfo(
-        java.lang.Iterable<? extends com.google.protobuf.Any> values) {
-      if (deviceInfoBuilder_ == null) {
-        ensureDeviceInfoIsMutable();
-        com.google.protobuf.AbstractMessageLite.Builder.addAll(
-            values, deviceInfo_);
-        onChanged();
-      } else {
-        deviceInfoBuilder_.addAllMessages(values);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Other devices that are attached and relevant (e.g. GPUInfo).
-     * </pre>
-     *
-     * <code>repeated .google.protobuf.Any device_info = 4;</code>
-     */
-    public Builder clearDeviceInfo() {
-      if (deviceInfoBuilder_ == null) {
-        deviceInfo_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000001);
-        onChanged();
-      } else {
-        deviceInfoBuilder_.clear();
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Other devices that are attached and relevant (e.g. GPUInfo).
-     * </pre>
-     *
-     * <code>repeated .google.protobuf.Any device_info = 4;</code>
-     */
-    public Builder removeDeviceInfo(int index) {
-      if (deviceInfoBuilder_ == null) {
-        ensureDeviceInfoIsMutable();
-        deviceInfo_.remove(index);
-        onChanged();
-      } else {
-        deviceInfoBuilder_.remove(index);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Other devices that are attached and relevant (e.g. GPUInfo).
-     * </pre>
-     *
-     * <code>repeated .google.protobuf.Any device_info = 4;</code>
-     */
-    public com.google.protobuf.Any.Builder getDeviceInfoBuilder(
-        int index) {
-      return getDeviceInfoFieldBuilder().getBuilder(index);
-    }
-    /**
-     * <pre>
-     * Other devices that are attached and relevant (e.g. GPUInfo).
-     * </pre>
-     *
-     * <code>repeated .google.protobuf.Any device_info = 4;</code>
-     */
-    public com.google.protobuf.AnyOrBuilder getDeviceInfoOrBuilder(
-        int index) {
-      if (deviceInfoBuilder_ == null) {
-        return deviceInfo_.get(index);  } else {
-        return deviceInfoBuilder_.getMessageOrBuilder(index);
-      }
-    }
-    /**
-     * <pre>
-     * Other devices that are attached and relevant (e.g. GPUInfo).
-     * </pre>
-     *
-     * <code>repeated .google.protobuf.Any device_info = 4;</code>
-     */
-    public java.util.List<? extends com.google.protobuf.AnyOrBuilder> 
-         getDeviceInfoOrBuilderList() {
-      if (deviceInfoBuilder_ != null) {
-        return deviceInfoBuilder_.getMessageOrBuilderList();
-      } else {
-        return java.util.Collections.unmodifiableList(deviceInfo_);
-      }
-    }
-    /**
-     * <pre>
-     * Other devices that are attached and relevant (e.g. GPUInfo).
-     * </pre>
-     *
-     * <code>repeated .google.protobuf.Any device_info = 4;</code>
-     */
-    public com.google.protobuf.Any.Builder addDeviceInfoBuilder() {
-      return getDeviceInfoFieldBuilder().addBuilder(
-          com.google.protobuf.Any.getDefaultInstance());
-    }
-    /**
-     * <pre>
-     * Other devices that are attached and relevant (e.g. GPUInfo).
-     * </pre>
-     *
-     * <code>repeated .google.protobuf.Any device_info = 4;</code>
-     */
-    public com.google.protobuf.Any.Builder addDeviceInfoBuilder(
-        int index) {
-      return getDeviceInfoFieldBuilder().addBuilder(
-          index, com.google.protobuf.Any.getDefaultInstance());
-    }
-    /**
-     * <pre>
-     * Other devices that are attached and relevant (e.g. GPUInfo).
-     * </pre>
-     *
-     * <code>repeated .google.protobuf.Any device_info = 4;</code>
-     */
-    public java.util.List<com.google.protobuf.Any.Builder> 
-         getDeviceInfoBuilderList() {
-      return getDeviceInfoFieldBuilder().getBuilderList();
-    }
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        com.google.protobuf.Any, com.google.protobuf.Any.Builder, com.google.protobuf.AnyOrBuilder> 
-        getDeviceInfoFieldBuilder() {
-      if (deviceInfoBuilder_ == null) {
-        deviceInfoBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
-            com.google.protobuf.Any, com.google.protobuf.Any.Builder, com.google.protobuf.AnyOrBuilder>(
-                deviceInfo_,
-                ((bitField0_ & 0x00000001) != 0),
-                getParentForChildren(),
-                isClean());
-        deviceInfo_ = null;
-      }
-      return deviceInfoBuilder_;
-    }
-
-    private java.util.List<org.tensorflow.proto.util.testlog.AvailableDeviceInfo> availableDeviceInfo_ =
-      java.util.Collections.emptyList();
-    private void ensureAvailableDeviceInfoIsMutable() {
-      if (!((bitField0_ & 0x00000002) != 0)) {
-        availableDeviceInfo_ = new java.util.ArrayList<org.tensorflow.proto.util.testlog.AvailableDeviceInfo>(availableDeviceInfo_);
-        bitField0_ |= 0x00000002;
-       }
-    }
-
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.util.testlog.AvailableDeviceInfo, org.tensorflow.proto.util.testlog.AvailableDeviceInfo.Builder, org.tensorflow.proto.util.testlog.AvailableDeviceInfoOrBuilder> availableDeviceInfoBuilder_;
-
-    /**
-     * <pre>
-     * Devices accessible to the test (e.g. as given by list_local_devices).
-     * </pre>
-     *
-     * <code>repeated .tensorflow.AvailableDeviceInfo available_device_info = 5;</code>
-     */
-    public java.util.List<org.tensorflow.proto.util.testlog.AvailableDeviceInfo> getAvailableDeviceInfoList() {
-      if (availableDeviceInfoBuilder_ == null) {
-        return java.util.Collections.unmodifiableList(availableDeviceInfo_);
-      } else {
-        return availableDeviceInfoBuilder_.getMessageList();
-      }
-    }
-    /**
-     * <pre>
-     * Devices accessible to the test (e.g. as given by list_local_devices).
-     * </pre>
-     *
-     * <code>repeated .tensorflow.AvailableDeviceInfo available_device_info = 5;</code>
-     */
-    public int getAvailableDeviceInfoCount() {
-      if (availableDeviceInfoBuilder_ == null) {
-        return availableDeviceInfo_.size();
-      } else {
-        return availableDeviceInfoBuilder_.getCount();
-      }
-    }
-    /**
-     * <pre>
-     * Devices accessible to the test (e.g. as given by list_local_devices).
-     * </pre>
-     *
-     * <code>repeated .tensorflow.AvailableDeviceInfo available_device_info = 5;</code>
-     */
-    public org.tensorflow.proto.util.testlog.AvailableDeviceInfo getAvailableDeviceInfo(int index) {
-      if (availableDeviceInfoBuilder_ == null) {
-        return availableDeviceInfo_.get(index);
-      } else {
-        return availableDeviceInfoBuilder_.getMessage(index);
-      }
-    }
-    /**
-     * <pre>
-     * Devices accessible to the test (e.g. as given by list_local_devices).
-     * </pre>
-     *
-     * <code>repeated .tensorflow.AvailableDeviceInfo available_device_info = 5;</code>
-     */
-    public Builder setAvailableDeviceInfo(
-        int index, org.tensorflow.proto.util.testlog.AvailableDeviceInfo value) {
-      if (availableDeviceInfoBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureAvailableDeviceInfoIsMutable();
-        availableDeviceInfo_.set(index, value);
-        onChanged();
-      } else {
-        availableDeviceInfoBuilder_.setMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Devices accessible to the test (e.g. as given by list_local_devices).
-     * </pre>
-     *
-     * <code>repeated .tensorflow.AvailableDeviceInfo available_device_info = 5;</code>
-     */
-    public Builder setAvailableDeviceInfo(
-        int index, org.tensorflow.proto.util.testlog.AvailableDeviceInfo.Builder builderForValue) {
-      if (availableDeviceInfoBuilder_ == null) {
-        ensureAvailableDeviceInfoIsMutable();
-        availableDeviceInfo_.set(index, builderForValue.build());
-        onChanged();
-      } else {
-        availableDeviceInfoBuilder_.setMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Devices accessible to the test (e.g. as given by list_local_devices).
-     * </pre>
-     *
-     * <code>repeated .tensorflow.AvailableDeviceInfo available_device_info = 5;</code>
-     */
-    public Builder addAvailableDeviceInfo(org.tensorflow.proto.util.testlog.AvailableDeviceInfo value) {
-      if (availableDeviceInfoBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureAvailableDeviceInfoIsMutable();
-        availableDeviceInfo_.add(value);
-        onChanged();
-      } else {
-        availableDeviceInfoBuilder_.addMessage(value);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Devices accessible to the test (e.g. as given by list_local_devices).
-     * </pre>
-     *
-     * <code>repeated .tensorflow.AvailableDeviceInfo available_device_info = 5;</code>
-     */
-    public Builder addAvailableDeviceInfo(
-        int index, org.tensorflow.proto.util.testlog.AvailableDeviceInfo value) {
-      if (availableDeviceInfoBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        ensureAvailableDeviceInfoIsMutable();
-        availableDeviceInfo_.add(index, value);
-        onChanged();
-      } else {
-        availableDeviceInfoBuilder_.addMessage(index, value);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Devices accessible to the test (e.g. as given by list_local_devices).
-     * </pre>
-     *
-     * <code>repeated .tensorflow.AvailableDeviceInfo available_device_info = 5;</code>
-     */
-    public Builder addAvailableDeviceInfo(
-        org.tensorflow.proto.util.testlog.AvailableDeviceInfo.Builder builderForValue) {
-      if (availableDeviceInfoBuilder_ == null) {
-        ensureAvailableDeviceInfoIsMutable();
-        availableDeviceInfo_.add(builderForValue.build());
-        onChanged();
-      } else {
-        availableDeviceInfoBuilder_.addMessage(builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Devices accessible to the test (e.g. as given by list_local_devices).
-     * </pre>
-     *
-     * <code>repeated .tensorflow.AvailableDeviceInfo available_device_info = 5;</code>
-     */
-    public Builder addAvailableDeviceInfo(
-        int index, org.tensorflow.proto.util.testlog.AvailableDeviceInfo.Builder builderForValue) {
-      if (availableDeviceInfoBuilder_ == null) {
-        ensureAvailableDeviceInfoIsMutable();
-        availableDeviceInfo_.add(index, builderForValue.build());
-        onChanged();
-      } else {
-        availableDeviceInfoBuilder_.addMessage(index, builderForValue.build());
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Devices accessible to the test (e.g. as given by list_local_devices).
-     * </pre>
-     *
-     * <code>repeated .tensorflow.AvailableDeviceInfo available_device_info = 5;</code>
-     */
-    public Builder addAllAvailableDeviceInfo(
-        java.lang.Iterable<? extends org.tensorflow.proto.util.testlog.AvailableDeviceInfo> values) {
-      if (availableDeviceInfoBuilder_ == null) {
-        ensureAvailableDeviceInfoIsMutable();
-        com.google.protobuf.AbstractMessageLite.Builder.addAll(
-            values, availableDeviceInfo_);
-        onChanged();
-      } else {
-        availableDeviceInfoBuilder_.addAllMessages(values);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Devices accessible to the test (e.g. as given by list_local_devices).
-     * </pre>
-     *
-     * <code>repeated .tensorflow.AvailableDeviceInfo available_device_info = 5;</code>
-     */
-    public Builder clearAvailableDeviceInfo() {
-      if (availableDeviceInfoBuilder_ == null) {
-        availableDeviceInfo_ = java.util.Collections.emptyList();
-        bitField0_ = (bitField0_ & ~0x00000002);
-        onChanged();
-      } else {
-        availableDeviceInfoBuilder_.clear();
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Devices accessible to the test (e.g. as given by list_local_devices).
-     * </pre>
-     *
-     * <code>repeated .tensorflow.AvailableDeviceInfo available_device_info = 5;</code>
-     */
-    public Builder removeAvailableDeviceInfo(int index) {
-      if (availableDeviceInfoBuilder_ == null) {
-        ensureAvailableDeviceInfoIsMutable();
-        availableDeviceInfo_.remove(index);
-        onChanged();
-      } else {
-        availableDeviceInfoBuilder_.remove(index);
-      }
-      return this;
-    }
-    /**
-     * <pre>
-     * Devices accessible to the test (e.g. as given by list_local_devices).
-     * </pre>
-     *
-     * <code>repeated .tensorflow.AvailableDeviceInfo available_device_info = 5;</code>
-     */
-    public org.tensorflow.proto.util.testlog.AvailableDeviceInfo.Builder getAvailableDeviceInfoBuilder(
-        int index) {
-      return getAvailableDeviceInfoFieldBuilder().getBuilder(index);
-    }
-    /**
-     * <pre>
-     * Devices accessible to the test (e.g. as given by list_local_devices).
-     * </pre>
-     *
-     * <code>repeated .tensorflow.AvailableDeviceInfo available_device_info = 5;</code>
-     */
-    public org.tensorflow.proto.util.testlog.AvailableDeviceInfoOrBuilder getAvailableDeviceInfoOrBuilder(
-        int index) {
-      if (availableDeviceInfoBuilder_ == null) {
-        return availableDeviceInfo_.get(index);  } else {
-        return availableDeviceInfoBuilder_.getMessageOrBuilder(index);
-      }
-    }
-    /**
-     * <pre>
-     * Devices accessible to the test (e.g. as given by list_local_devices).
-     * </pre>
-     *
-     * <code>repeated .tensorflow.AvailableDeviceInfo available_device_info = 5;</code>
-     */
-    public java.util.List<? extends org.tensorflow.proto.util.testlog.AvailableDeviceInfoOrBuilder> 
-         getAvailableDeviceInfoOrBuilderList() {
-      if (availableDeviceInfoBuilder_ != null) {
-        return availableDeviceInfoBuilder_.getMessageOrBuilderList();
-      } else {
-        return java.util.Collections.unmodifiableList(availableDeviceInfo_);
-      }
-    }
-    /**
-     * <pre>
-     * Devices accessible to the test (e.g. as given by list_local_devices).
-     * </pre>
-     *
-     * <code>repeated .tensorflow.AvailableDeviceInfo available_device_info = 5;</code>
-     */
-    public org.tensorflow.proto.util.testlog.AvailableDeviceInfo.Builder addAvailableDeviceInfoBuilder() {
-      return getAvailableDeviceInfoFieldBuilder().addBuilder(
-          org.tensorflow.proto.util.testlog.AvailableDeviceInfo.getDefaultInstance());
-    }
-    /**
-     * <pre>
-     * Devices accessible to the test (e.g. as given by list_local_devices).
-     * </pre>
-     *
-     * <code>repeated .tensorflow.AvailableDeviceInfo available_device_info = 5;</code>
-     */
-    public org.tensorflow.proto.util.testlog.AvailableDeviceInfo.Builder addAvailableDeviceInfoBuilder(
-        int index) {
-      return getAvailableDeviceInfoFieldBuilder().addBuilder(
-          index, org.tensorflow.proto.util.testlog.AvailableDeviceInfo.getDefaultInstance());
-    }
-    /**
-     * <pre>
-     * Devices accessible to the test (e.g. as given by list_local_devices).
-     * </pre>
-     *
-     * <code>repeated .tensorflow.AvailableDeviceInfo available_device_info = 5;</code>
-     */
-    public java.util.List<org.tensorflow.proto.util.testlog.AvailableDeviceInfo.Builder> 
-         getAvailableDeviceInfoBuilderList() {
-      return getAvailableDeviceInfoFieldBuilder().getBuilderList();
-    }
-    private com.google.protobuf.RepeatedFieldBuilderV3<
-        org.tensorflow.proto.util.testlog.AvailableDeviceInfo, org.tensorflow.proto.util.testlog.AvailableDeviceInfo.Builder, org.tensorflow.proto.util.testlog.AvailableDeviceInfoOrBuilder> 
-        getAvailableDeviceInfoFieldBuilder() {
-      if (availableDeviceInfoBuilder_ == null) {
-        availableDeviceInfoBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
-            org.tensorflow.proto.util.testlog.AvailableDeviceInfo, org.tensorflow.proto.util.testlog.AvailableDeviceInfo.Builder, org.tensorflow.proto.util.testlog.AvailableDeviceInfoOrBuilder>(
-                availableDeviceInfo_,
-                ((bitField0_ & 0x00000002) != 0),
-                getParentForChildren(),
-                isClean());
-        availableDeviceInfo_ = null;
-      }
-      return availableDeviceInfoBuilder_;
-    }
-
-    private org.tensorflow.proto.util.testlog.MemoryInfo memoryInfo_;
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.util.testlog.MemoryInfo, org.tensorflow.proto.util.testlog.MemoryInfo.Builder, org.tensorflow.proto.util.testlog.MemoryInfoOrBuilder> memoryInfoBuilder_;
-    /**
-     * <code>.tensorflow.MemoryInfo memory_info = 6;</code>
-     */
-    public boolean hasMemoryInfo() {
-      return memoryInfoBuilder_ != null || memoryInfo_ != null;
-    }
-    /**
-     * <code>.tensorflow.MemoryInfo memory_info = 6;</code>
-     */
-    public org.tensorflow.proto.util.testlog.MemoryInfo getMemoryInfo() {
-      if (memoryInfoBuilder_ == null) {
-        return memoryInfo_ == null ? org.tensorflow.proto.util.testlog.MemoryInfo.getDefaultInstance() : memoryInfo_;
-      } else {
-        return memoryInfoBuilder_.getMessage();
-      }
-    }
-    /**
-     * <code>.tensorflow.MemoryInfo memory_info = 6;</code>
-     */
-    public Builder setMemoryInfo(org.tensorflow.proto.util.testlog.MemoryInfo value) {
-      if (memoryInfoBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        memoryInfo_ = value;
-        onChanged();
-      } else {
-        memoryInfoBuilder_.setMessage(value);
-      }
-
-      return this;
-    }
-    /**
-     * <code>.tensorflow.MemoryInfo memory_info = 6;</code>
-     */
-    public Builder setMemoryInfo(
-        org.tensorflow.proto.util.testlog.MemoryInfo.Builder builderForValue) {
-      if (memoryInfoBuilder_ == null) {
-        memoryInfo_ = builderForValue.build();
-        onChanged();
-      } else {
-        memoryInfoBuilder_.setMessage(builderForValue.build());
-      }
-
-      return this;
-    }
-    /**
-     * <code>.tensorflow.MemoryInfo memory_info = 6;</code>
-     */
-    public Builder mergeMemoryInfo(org.tensorflow.proto.util.testlog.MemoryInfo value) {
-      if (memoryInfoBuilder_ == null) {
-        if (memoryInfo_ != null) {
-          memoryInfo_ =
-            org.tensorflow.proto.util.testlog.MemoryInfo.newBuilder(memoryInfo_).mergeFrom(value).buildPartial();
-        } else {
-          memoryInfo_ = value;
-        }
-        onChanged();
-      } else {
-        memoryInfoBuilder_.mergeFrom(value);
-      }
-
-      return this;
-    }
-    /**
-     * <code>.tensorflow.MemoryInfo memory_info = 6;</code>
-     */
-    public Builder clearMemoryInfo() {
-      if (memoryInfoBuilder_ == null) {
-        memoryInfo_ = null;
-        onChanged();
-      } else {
-        memoryInfo_ = null;
-        memoryInfoBuilder_ = null;
-      }
-
-      return this;
-    }
-    /**
-     * <code>.tensorflow.MemoryInfo memory_info = 6;</code>
-     */
-    public org.tensorflow.proto.util.testlog.MemoryInfo.Builder getMemoryInfoBuilder() {
-      
-      onChanged();
-      return getMemoryInfoFieldBuilder().getBuilder();
-    }
-    /**
-     * <code>.tensorflow.MemoryInfo memory_info = 6;</code>
-     */
-    public org.tensorflow.proto.util.testlog.MemoryInfoOrBuilder getMemoryInfoOrBuilder() {
-      if (memoryInfoBuilder_ != null) {
-        return memoryInfoBuilder_.getMessageOrBuilder();
-      } else {
-        return memoryInfo_ == null ?
-            org.tensorflow.proto.util.testlog.MemoryInfo.getDefaultInstance() : memoryInfo_;
-      }
-    }
-    /**
-     * <code>.tensorflow.MemoryInfo memory_info = 6;</code>
-     */
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.util.testlog.MemoryInfo, org.tensorflow.proto.util.testlog.MemoryInfo.Builder, org.tensorflow.proto.util.testlog.MemoryInfoOrBuilder> 
-        getMemoryInfoFieldBuilder() {
-      if (memoryInfoBuilder_ == null) {
-        memoryInfoBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-            org.tensorflow.proto.util.testlog.MemoryInfo, org.tensorflow.proto.util.testlog.MemoryInfo.Builder, org.tensorflow.proto.util.testlog.MemoryInfoOrBuilder>(
-                getMemoryInfo(),
-                getParentForChildren(),
-                isClean());
-        memoryInfo_ = null;
-      }
-      return memoryInfoBuilder_;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.MachineConfiguration)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.MachineConfiguration)
-  private static final org.tensorflow.proto.util.testlog.MachineConfiguration DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.util.testlog.MachineConfiguration();
-  }
-
-  public static org.tensorflow.proto.util.testlog.MachineConfiguration getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<MachineConfiguration>
-      PARSER = new com.google.protobuf.AbstractParser<MachineConfiguration>() {
-    @java.lang.Override
-    public MachineConfiguration parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new MachineConfiguration(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<MachineConfiguration> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<MachineConfiguration> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.util.testlog.MachineConfiguration getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/MachineConfigurationOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/MachineConfigurationOrBuilder.java
deleted file mode 100644
index 8b0f5f6f884..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/MachineConfigurationOrBuilder.java
+++ /dev/null
@@ -1,196 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/util/test_log.proto
-
-package org.tensorflow.proto.util.testlog;
-
-public interface MachineConfigurationOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.MachineConfiguration)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <pre>
-   * Host name of machine that ran the benchmark.
-   * </pre>
-   *
-   * <code>string hostname = 1;</code>
-   */
-  java.lang.String getHostname();
-  /**
-   * <pre>
-   * Host name of machine that ran the benchmark.
-   * </pre>
-   *
-   * <code>string hostname = 1;</code>
-   */
-  com.google.protobuf.ByteString
-      getHostnameBytes();
-
-  /**
-   * <pre>
-   * Unique serial number of the machine.
-   * </pre>
-   *
-   * <code>string serial_identifier = 7;</code>
-   */
-  java.lang.String getSerialIdentifier();
-  /**
-   * <pre>
-   * Unique serial number of the machine.
-   * </pre>
-   *
-   * <code>string serial_identifier = 7;</code>
-   */
-  com.google.protobuf.ByteString
-      getSerialIdentifierBytes();
-
-  /**
-   * <pre>
-   * Additional platform information.
-   * </pre>
-   *
-   * <code>.tensorflow.PlatformInfo platform_info = 2;</code>
-   */
-  boolean hasPlatformInfo();
-  /**
-   * <pre>
-   * Additional platform information.
-   * </pre>
-   *
-   * <code>.tensorflow.PlatformInfo platform_info = 2;</code>
-   */
-  org.tensorflow.proto.util.testlog.PlatformInfo getPlatformInfo();
-  /**
-   * <pre>
-   * Additional platform information.
-   * </pre>
-   *
-   * <code>.tensorflow.PlatformInfo platform_info = 2;</code>
-   */
-  org.tensorflow.proto.util.testlog.PlatformInfoOrBuilder getPlatformInfoOrBuilder();
-
-  /**
-   * <pre>
-   * CPU Information.
-   * </pre>
-   *
-   * <code>.tensorflow.CPUInfo cpu_info = 3;</code>
-   */
-  boolean hasCpuInfo();
-  /**
-   * <pre>
-   * CPU Information.
-   * </pre>
-   *
-   * <code>.tensorflow.CPUInfo cpu_info = 3;</code>
-   */
-  org.tensorflow.proto.util.testlog.CPUInfo getCpuInfo();
-  /**
-   * <pre>
-   * CPU Information.
-   * </pre>
-   *
-   * <code>.tensorflow.CPUInfo cpu_info = 3;</code>
-   */
-  org.tensorflow.proto.util.testlog.CPUInfoOrBuilder getCpuInfoOrBuilder();
-
-  /**
-   * <pre>
-   * Other devices that are attached and relevant (e.g. GPUInfo).
-   * </pre>
-   *
-   * <code>repeated .google.protobuf.Any device_info = 4;</code>
-   */
-  java.util.List<com.google.protobuf.Any> 
-      getDeviceInfoList();
-  /**
-   * <pre>
-   * Other devices that are attached and relevant (e.g. GPUInfo).
-   * </pre>
-   *
-   * <code>repeated .google.protobuf.Any device_info = 4;</code>
-   */
-  com.google.protobuf.Any getDeviceInfo(int index);
-  /**
-   * <pre>
-   * Other devices that are attached and relevant (e.g. GPUInfo).
-   * </pre>
-   *
-   * <code>repeated .google.protobuf.Any device_info = 4;</code>
-   */
-  int getDeviceInfoCount();
-  /**
-   * <pre>
-   * Other devices that are attached and relevant (e.g. GPUInfo).
-   * </pre>
-   *
-   * <code>repeated .google.protobuf.Any device_info = 4;</code>
-   */
-  java.util.List<? extends com.google.protobuf.AnyOrBuilder> 
-      getDeviceInfoOrBuilderList();
-  /**
-   * <pre>
-   * Other devices that are attached and relevant (e.g. GPUInfo).
-   * </pre>
-   *
-   * <code>repeated .google.protobuf.Any device_info = 4;</code>
-   */
-  com.google.protobuf.AnyOrBuilder getDeviceInfoOrBuilder(
-      int index);
-
-  /**
-   * <pre>
-   * Devices accessible to the test (e.g. as given by list_local_devices).
-   * </pre>
-   *
-   * <code>repeated .tensorflow.AvailableDeviceInfo available_device_info = 5;</code>
-   */
-  java.util.List<org.tensorflow.proto.util.testlog.AvailableDeviceInfo> 
-      getAvailableDeviceInfoList();
-  /**
-   * <pre>
-   * Devices accessible to the test (e.g. as given by list_local_devices).
-   * </pre>
-   *
-   * <code>repeated .tensorflow.AvailableDeviceInfo available_device_info = 5;</code>
-   */
-  org.tensorflow.proto.util.testlog.AvailableDeviceInfo getAvailableDeviceInfo(int index);
-  /**
-   * <pre>
-   * Devices accessible to the test (e.g. as given by list_local_devices).
-   * </pre>
-   *
-   * <code>repeated .tensorflow.AvailableDeviceInfo available_device_info = 5;</code>
-   */
-  int getAvailableDeviceInfoCount();
-  /**
-   * <pre>
-   * Devices accessible to the test (e.g. as given by list_local_devices).
-   * </pre>
-   *
-   * <code>repeated .tensorflow.AvailableDeviceInfo available_device_info = 5;</code>
-   */
-  java.util.List<? extends org.tensorflow.proto.util.testlog.AvailableDeviceInfoOrBuilder> 
-      getAvailableDeviceInfoOrBuilderList();
-  /**
-   * <pre>
-   * Devices accessible to the test (e.g. as given by list_local_devices).
-   * </pre>
-   *
-   * <code>repeated .tensorflow.AvailableDeviceInfo available_device_info = 5;</code>
-   */
-  org.tensorflow.proto.util.testlog.AvailableDeviceInfoOrBuilder getAvailableDeviceInfoOrBuilder(
-      int index);
-
-  /**
-   * <code>.tensorflow.MemoryInfo memory_info = 6;</code>
-   */
-  boolean hasMemoryInfo();
-  /**
-   * <code>.tensorflow.MemoryInfo memory_info = 6;</code>
-   */
-  org.tensorflow.proto.util.testlog.MemoryInfo getMemoryInfo();
-  /**
-   * <code>.tensorflow.MemoryInfo memory_info = 6;</code>
-   */
-  org.tensorflow.proto.util.testlog.MemoryInfoOrBuilder getMemoryInfoOrBuilder();
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/MemoryInfo.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/MemoryInfo.java
deleted file mode 100644
index e28e00134cc..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/MemoryInfo.java
+++ /dev/null
@@ -1,567 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/util/test_log.proto
-
-package org.tensorflow.proto.util.testlog;
-
-/**
- * Protobuf type {@code tensorflow.MemoryInfo}
- */
-public  final class MemoryInfo extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.MemoryInfo)
-    MemoryInfoOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use MemoryInfo.newBuilder() to construct.
-  private MemoryInfo(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private MemoryInfo() {
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new MemoryInfo();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private MemoryInfo(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 8: {
-
-            total_ = input.readInt64();
-            break;
-          }
-          case 16: {
-
-            available_ = input.readInt64();
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_MemoryInfo_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_MemoryInfo_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.util.testlog.MemoryInfo.class, org.tensorflow.proto.util.testlog.MemoryInfo.Builder.class);
-  }
-
-  public static final int TOTAL_FIELD_NUMBER = 1;
-  private long total_;
-  /**
-   * <pre>
-   * Total virtual memory in bytes
-   * </pre>
-   *
-   * <code>int64 total = 1;</code>
-   */
-  public long getTotal() {
-    return total_;
-  }
-
-  public static final int AVAILABLE_FIELD_NUMBER = 2;
-  private long available_;
-  /**
-   * <pre>
-   * Immediately available memory in bytes
-   * </pre>
-   *
-   * <code>int64 available = 2;</code>
-   */
-  public long getAvailable() {
-    return available_;
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    if (total_ != 0L) {
-      output.writeInt64(1, total_);
-    }
-    if (available_ != 0L) {
-      output.writeInt64(2, available_);
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (total_ != 0L) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt64Size(1, total_);
-    }
-    if (available_ != 0L) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt64Size(2, available_);
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.util.testlog.MemoryInfo)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.util.testlog.MemoryInfo other = (org.tensorflow.proto.util.testlog.MemoryInfo) obj;
-
-    if (getTotal()
-        != other.getTotal()) return false;
-    if (getAvailable()
-        != other.getAvailable()) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    hash = (37 * hash) + TOTAL_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-        getTotal());
-    hash = (37 * hash) + AVAILABLE_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-        getAvailable());
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.util.testlog.MemoryInfo parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.testlog.MemoryInfo parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.testlog.MemoryInfo parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.testlog.MemoryInfo parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.testlog.MemoryInfo parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.testlog.MemoryInfo parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.testlog.MemoryInfo parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.testlog.MemoryInfo parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.testlog.MemoryInfo parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.testlog.MemoryInfo parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.testlog.MemoryInfo parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.testlog.MemoryInfo parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.util.testlog.MemoryInfo prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * Protobuf type {@code tensorflow.MemoryInfo}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.MemoryInfo)
-      org.tensorflow.proto.util.testlog.MemoryInfoOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_MemoryInfo_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_MemoryInfo_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.util.testlog.MemoryInfo.class, org.tensorflow.proto.util.testlog.MemoryInfo.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.util.testlog.MemoryInfo.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      total_ = 0L;
-
-      available_ = 0L;
-
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_MemoryInfo_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.testlog.MemoryInfo getDefaultInstanceForType() {
-      return org.tensorflow.proto.util.testlog.MemoryInfo.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.testlog.MemoryInfo build() {
-      org.tensorflow.proto.util.testlog.MemoryInfo result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.testlog.MemoryInfo buildPartial() {
-      org.tensorflow.proto.util.testlog.MemoryInfo result = new org.tensorflow.proto.util.testlog.MemoryInfo(this);
-      result.total_ = total_;
-      result.available_ = available_;
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.util.testlog.MemoryInfo) {
-        return mergeFrom((org.tensorflow.proto.util.testlog.MemoryInfo)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.util.testlog.MemoryInfo other) {
-      if (other == org.tensorflow.proto.util.testlog.MemoryInfo.getDefaultInstance()) return this;
-      if (other.getTotal() != 0L) {
-        setTotal(other.getTotal());
-      }
-      if (other.getAvailable() != 0L) {
-        setAvailable(other.getAvailable());
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.util.testlog.MemoryInfo parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.util.testlog.MemoryInfo) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-
-    private long total_ ;
-    /**
-     * <pre>
-     * Total virtual memory in bytes
-     * </pre>
-     *
-     * <code>int64 total = 1;</code>
-     */
-    public long getTotal() {
-      return total_;
-    }
-    /**
-     * <pre>
-     * Total virtual memory in bytes
-     * </pre>
-     *
-     * <code>int64 total = 1;</code>
-     */
-    public Builder setTotal(long value) {
-      
-      total_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Total virtual memory in bytes
-     * </pre>
-     *
-     * <code>int64 total = 1;</code>
-     */
-    public Builder clearTotal() {
-      
-      total_ = 0L;
-      onChanged();
-      return this;
-    }
-
-    private long available_ ;
-    /**
-     * <pre>
-     * Immediately available memory in bytes
-     * </pre>
-     *
-     * <code>int64 available = 2;</code>
-     */
-    public long getAvailable() {
-      return available_;
-    }
-    /**
-     * <pre>
-     * Immediately available memory in bytes
-     * </pre>
-     *
-     * <code>int64 available = 2;</code>
-     */
-    public Builder setAvailable(long value) {
-      
-      available_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Immediately available memory in bytes
-     * </pre>
-     *
-     * <code>int64 available = 2;</code>
-     */
-    public Builder clearAvailable() {
-      
-      available_ = 0L;
-      onChanged();
-      return this;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.MemoryInfo)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.MemoryInfo)
-  private static final org.tensorflow.proto.util.testlog.MemoryInfo DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.util.testlog.MemoryInfo();
-  }
-
-  public static org.tensorflow.proto.util.testlog.MemoryInfo getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<MemoryInfo>
-      PARSER = new com.google.protobuf.AbstractParser<MemoryInfo>() {
-    @java.lang.Override
-    public MemoryInfo parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new MemoryInfo(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<MemoryInfo> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<MemoryInfo> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.util.testlog.MemoryInfo getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/MemoryInfoOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/MemoryInfoOrBuilder.java
deleted file mode 100644
index 1700a289e4f..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/MemoryInfoOrBuilder.java
+++ /dev/null
@@ -1,27 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/util/test_log.proto
-
-package org.tensorflow.proto.util.testlog;
-
-public interface MemoryInfoOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.MemoryInfo)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <pre>
-   * Total virtual memory in bytes
-   * </pre>
-   *
-   * <code>int64 total = 1;</code>
-   */
-  long getTotal();
-
-  /**
-   * <pre>
-   * Immediately available memory in bytes
-   * </pre>
-   *
-   * <code>int64 available = 2;</code>
-   */
-  long getAvailable();
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/MetricEntry.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/MetricEntry.java
deleted file mode 100644
index 0ebe3ce31ab..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/MetricEntry.java
+++ /dev/null
@@ -1,1107 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/util/test_log.proto
-
-package org.tensorflow.proto.util.testlog;
-
-/**
- * Protobuf type {@code tensorflow.MetricEntry}
- */
-public  final class MetricEntry extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.MetricEntry)
-    MetricEntryOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use MetricEntry.newBuilder() to construct.
-  private MetricEntry(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private MetricEntry() {
-    name_ = "";
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new MetricEntry();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private MetricEntry(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 10: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            name_ = s;
-            break;
-          }
-          case 17: {
-
-            value_ = input.readDouble();
-            break;
-          }
-          case 26: {
-            com.google.protobuf.DoubleValue.Builder subBuilder = null;
-            if (minValue_ != null) {
-              subBuilder = minValue_.toBuilder();
-            }
-            minValue_ = input.readMessage(com.google.protobuf.DoubleValue.parser(), extensionRegistry);
-            if (subBuilder != null) {
-              subBuilder.mergeFrom(minValue_);
-              minValue_ = subBuilder.buildPartial();
-            }
-
-            break;
-          }
-          case 34: {
-            com.google.protobuf.DoubleValue.Builder subBuilder = null;
-            if (maxValue_ != null) {
-              subBuilder = maxValue_.toBuilder();
-            }
-            maxValue_ = input.readMessage(com.google.protobuf.DoubleValue.parser(), extensionRegistry);
-            if (subBuilder != null) {
-              subBuilder.mergeFrom(maxValue_);
-              maxValue_ = subBuilder.buildPartial();
-            }
-
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_MetricEntry_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_MetricEntry_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.util.testlog.MetricEntry.class, org.tensorflow.proto.util.testlog.MetricEntry.Builder.class);
-  }
-
-  public static final int NAME_FIELD_NUMBER = 1;
-  private volatile java.lang.Object name_;
-  /**
-   * <pre>
-   * Metric name
-   * </pre>
-   *
-   * <code>string name = 1;</code>
-   */
-  public java.lang.String getName() {
-    java.lang.Object ref = name_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      name_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * Metric name
-   * </pre>
-   *
-   * <code>string name = 1;</code>
-   */
-  public com.google.protobuf.ByteString
-      getNameBytes() {
-    java.lang.Object ref = name_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      name_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int VALUE_FIELD_NUMBER = 2;
-  private double value_;
-  /**
-   * <pre>
-   * Metric value
-   * </pre>
-   *
-   * <code>double value = 2;</code>
-   */
-  public double getValue() {
-    return value_;
-  }
-
-  public static final int MIN_VALUE_FIELD_NUMBER = 3;
-  private com.google.protobuf.DoubleValue minValue_;
-  /**
-   * <pre>
-   * The minimum acceptable value for the metric if specified
-   * </pre>
-   *
-   * <code>.google.protobuf.DoubleValue min_value = 3;</code>
-   */
-  public boolean hasMinValue() {
-    return minValue_ != null;
-  }
-  /**
-   * <pre>
-   * The minimum acceptable value for the metric if specified
-   * </pre>
-   *
-   * <code>.google.protobuf.DoubleValue min_value = 3;</code>
-   */
-  public com.google.protobuf.DoubleValue getMinValue() {
-    return minValue_ == null ? com.google.protobuf.DoubleValue.getDefaultInstance() : minValue_;
-  }
-  /**
-   * <pre>
-   * The minimum acceptable value for the metric if specified
-   * </pre>
-   *
-   * <code>.google.protobuf.DoubleValue min_value = 3;</code>
-   */
-  public com.google.protobuf.DoubleValueOrBuilder getMinValueOrBuilder() {
-    return getMinValue();
-  }
-
-  public static final int MAX_VALUE_FIELD_NUMBER = 4;
-  private com.google.protobuf.DoubleValue maxValue_;
-  /**
-   * <pre>
-   * The maximum acceptable value for the metric if specified
-   * </pre>
-   *
-   * <code>.google.protobuf.DoubleValue max_value = 4;</code>
-   */
-  public boolean hasMaxValue() {
-    return maxValue_ != null;
-  }
-  /**
-   * <pre>
-   * The maximum acceptable value for the metric if specified
-   * </pre>
-   *
-   * <code>.google.protobuf.DoubleValue max_value = 4;</code>
-   */
-  public com.google.protobuf.DoubleValue getMaxValue() {
-    return maxValue_ == null ? com.google.protobuf.DoubleValue.getDefaultInstance() : maxValue_;
-  }
-  /**
-   * <pre>
-   * The maximum acceptable value for the metric if specified
-   * </pre>
-   *
-   * <code>.google.protobuf.DoubleValue max_value = 4;</code>
-   */
-  public com.google.protobuf.DoubleValueOrBuilder getMaxValueOrBuilder() {
-    return getMaxValue();
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    if (!getNameBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 1, name_);
-    }
-    if (value_ != 0D) {
-      output.writeDouble(2, value_);
-    }
-    if (minValue_ != null) {
-      output.writeMessage(3, getMinValue());
-    }
-    if (maxValue_ != null) {
-      output.writeMessage(4, getMaxValue());
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (!getNameBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, name_);
-    }
-    if (value_ != 0D) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeDoubleSize(2, value_);
-    }
-    if (minValue_ != null) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(3, getMinValue());
-    }
-    if (maxValue_ != null) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(4, getMaxValue());
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.util.testlog.MetricEntry)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.util.testlog.MetricEntry other = (org.tensorflow.proto.util.testlog.MetricEntry) obj;
-
-    if (!getName()
-        .equals(other.getName())) return false;
-    if (java.lang.Double.doubleToLongBits(getValue())
-        != java.lang.Double.doubleToLongBits(
-            other.getValue())) return false;
-    if (hasMinValue() != other.hasMinValue()) return false;
-    if (hasMinValue()) {
-      if (!getMinValue()
-          .equals(other.getMinValue())) return false;
-    }
-    if (hasMaxValue() != other.hasMaxValue()) return false;
-    if (hasMaxValue()) {
-      if (!getMaxValue()
-          .equals(other.getMaxValue())) return false;
-    }
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    hash = (37 * hash) + NAME_FIELD_NUMBER;
-    hash = (53 * hash) + getName().hashCode();
-    hash = (37 * hash) + VALUE_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-        java.lang.Double.doubleToLongBits(getValue()));
-    if (hasMinValue()) {
-      hash = (37 * hash) + MIN_VALUE_FIELD_NUMBER;
-      hash = (53 * hash) + getMinValue().hashCode();
-    }
-    if (hasMaxValue()) {
-      hash = (37 * hash) + MAX_VALUE_FIELD_NUMBER;
-      hash = (53 * hash) + getMaxValue().hashCode();
-    }
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.util.testlog.MetricEntry parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.testlog.MetricEntry parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.testlog.MetricEntry parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.testlog.MetricEntry parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.testlog.MetricEntry parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.testlog.MetricEntry parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.testlog.MetricEntry parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.testlog.MetricEntry parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.testlog.MetricEntry parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.testlog.MetricEntry parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.testlog.MetricEntry parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.testlog.MetricEntry parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.util.testlog.MetricEntry prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * Protobuf type {@code tensorflow.MetricEntry}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.MetricEntry)
-      org.tensorflow.proto.util.testlog.MetricEntryOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_MetricEntry_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_MetricEntry_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.util.testlog.MetricEntry.class, org.tensorflow.proto.util.testlog.MetricEntry.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.util.testlog.MetricEntry.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      name_ = "";
-
-      value_ = 0D;
-
-      if (minValueBuilder_ == null) {
-        minValue_ = null;
-      } else {
-        minValue_ = null;
-        minValueBuilder_ = null;
-      }
-      if (maxValueBuilder_ == null) {
-        maxValue_ = null;
-      } else {
-        maxValue_ = null;
-        maxValueBuilder_ = null;
-      }
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_MetricEntry_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.testlog.MetricEntry getDefaultInstanceForType() {
-      return org.tensorflow.proto.util.testlog.MetricEntry.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.testlog.MetricEntry build() {
-      org.tensorflow.proto.util.testlog.MetricEntry result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.testlog.MetricEntry buildPartial() {
-      org.tensorflow.proto.util.testlog.MetricEntry result = new org.tensorflow.proto.util.testlog.MetricEntry(this);
-      result.name_ = name_;
-      result.value_ = value_;
-      if (minValueBuilder_ == null) {
-        result.minValue_ = minValue_;
-      } else {
-        result.minValue_ = minValueBuilder_.build();
-      }
-      if (maxValueBuilder_ == null) {
-        result.maxValue_ = maxValue_;
-      } else {
-        result.maxValue_ = maxValueBuilder_.build();
-      }
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.util.testlog.MetricEntry) {
-        return mergeFrom((org.tensorflow.proto.util.testlog.MetricEntry)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.util.testlog.MetricEntry other) {
-      if (other == org.tensorflow.proto.util.testlog.MetricEntry.getDefaultInstance()) return this;
-      if (!other.getName().isEmpty()) {
-        name_ = other.name_;
-        onChanged();
-      }
-      if (other.getValue() != 0D) {
-        setValue(other.getValue());
-      }
-      if (other.hasMinValue()) {
-        mergeMinValue(other.getMinValue());
-      }
-      if (other.hasMaxValue()) {
-        mergeMaxValue(other.getMaxValue());
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.util.testlog.MetricEntry parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.util.testlog.MetricEntry) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-
-    private java.lang.Object name_ = "";
-    /**
-     * <pre>
-     * Metric name
-     * </pre>
-     *
-     * <code>string name = 1;</code>
-     */
-    public java.lang.String getName() {
-      java.lang.Object ref = name_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        name_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Metric name
-     * </pre>
-     *
-     * <code>string name = 1;</code>
-     */
-    public com.google.protobuf.ByteString
-        getNameBytes() {
-      java.lang.Object ref = name_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        name_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Metric name
-     * </pre>
-     *
-     * <code>string name = 1;</code>
-     */
-    public Builder setName(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      name_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Metric name
-     * </pre>
-     *
-     * <code>string name = 1;</code>
-     */
-    public Builder clearName() {
-      
-      name_ = getDefaultInstance().getName();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Metric name
-     * </pre>
-     *
-     * <code>string name = 1;</code>
-     */
-    public Builder setNameBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      name_ = value;
-      onChanged();
-      return this;
-    }
-
-    private double value_ ;
-    /**
-     * <pre>
-     * Metric value
-     * </pre>
-     *
-     * <code>double value = 2;</code>
-     */
-    public double getValue() {
-      return value_;
-    }
-    /**
-     * <pre>
-     * Metric value
-     * </pre>
-     *
-     * <code>double value = 2;</code>
-     */
-    public Builder setValue(double value) {
-      
-      value_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Metric value
-     * </pre>
-     *
-     * <code>double value = 2;</code>
-     */
-    public Builder clearValue() {
-      
-      value_ = 0D;
-      onChanged();
-      return this;
-    }
-
-    private com.google.protobuf.DoubleValue minValue_;
-    private com.google.protobuf.SingleFieldBuilderV3<
-        com.google.protobuf.DoubleValue, com.google.protobuf.DoubleValue.Builder, com.google.protobuf.DoubleValueOrBuilder> minValueBuilder_;
-    /**
-     * <pre>
-     * The minimum acceptable value for the metric if specified
-     * </pre>
-     *
-     * <code>.google.protobuf.DoubleValue min_value = 3;</code>
-     */
-    public boolean hasMinValue() {
-      return minValueBuilder_ != null || minValue_ != null;
-    }
-    /**
-     * <pre>
-     * The minimum acceptable value for the metric if specified
-     * </pre>
-     *
-     * <code>.google.protobuf.DoubleValue min_value = 3;</code>
-     */
-    public com.google.protobuf.DoubleValue getMinValue() {
-      if (minValueBuilder_ == null) {
-        return minValue_ == null ? com.google.protobuf.DoubleValue.getDefaultInstance() : minValue_;
-      } else {
-        return minValueBuilder_.getMessage();
-      }
-    }
-    /**
-     * <pre>
-     * The minimum acceptable value for the metric if specified
-     * </pre>
-     *
-     * <code>.google.protobuf.DoubleValue min_value = 3;</code>
-     */
-    public Builder setMinValue(com.google.protobuf.DoubleValue value) {
-      if (minValueBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        minValue_ = value;
-        onChanged();
-      } else {
-        minValueBuilder_.setMessage(value);
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * The minimum acceptable value for the metric if specified
-     * </pre>
-     *
-     * <code>.google.protobuf.DoubleValue min_value = 3;</code>
-     */
-    public Builder setMinValue(
-        com.google.protobuf.DoubleValue.Builder builderForValue) {
-      if (minValueBuilder_ == null) {
-        minValue_ = builderForValue.build();
-        onChanged();
-      } else {
-        minValueBuilder_.setMessage(builderForValue.build());
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * The minimum acceptable value for the metric if specified
-     * </pre>
-     *
-     * <code>.google.protobuf.DoubleValue min_value = 3;</code>
-     */
-    public Builder mergeMinValue(com.google.protobuf.DoubleValue value) {
-      if (minValueBuilder_ == null) {
-        if (minValue_ != null) {
-          minValue_ =
-            com.google.protobuf.DoubleValue.newBuilder(minValue_).mergeFrom(value).buildPartial();
-        } else {
-          minValue_ = value;
-        }
-        onChanged();
-      } else {
-        minValueBuilder_.mergeFrom(value);
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * The minimum acceptable value for the metric if specified
-     * </pre>
-     *
-     * <code>.google.protobuf.DoubleValue min_value = 3;</code>
-     */
-    public Builder clearMinValue() {
-      if (minValueBuilder_ == null) {
-        minValue_ = null;
-        onChanged();
-      } else {
-        minValue_ = null;
-        minValueBuilder_ = null;
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * The minimum acceptable value for the metric if specified
-     * </pre>
-     *
-     * <code>.google.protobuf.DoubleValue min_value = 3;</code>
-     */
-    public com.google.protobuf.DoubleValue.Builder getMinValueBuilder() {
-      
-      onChanged();
-      return getMinValueFieldBuilder().getBuilder();
-    }
-    /**
-     * <pre>
-     * The minimum acceptable value for the metric if specified
-     * </pre>
-     *
-     * <code>.google.protobuf.DoubleValue min_value = 3;</code>
-     */
-    public com.google.protobuf.DoubleValueOrBuilder getMinValueOrBuilder() {
-      if (minValueBuilder_ != null) {
-        return minValueBuilder_.getMessageOrBuilder();
-      } else {
-        return minValue_ == null ?
-            com.google.protobuf.DoubleValue.getDefaultInstance() : minValue_;
-      }
-    }
-    /**
-     * <pre>
-     * The minimum acceptable value for the metric if specified
-     * </pre>
-     *
-     * <code>.google.protobuf.DoubleValue min_value = 3;</code>
-     */
-    private com.google.protobuf.SingleFieldBuilderV3<
-        com.google.protobuf.DoubleValue, com.google.protobuf.DoubleValue.Builder, com.google.protobuf.DoubleValueOrBuilder> 
-        getMinValueFieldBuilder() {
-      if (minValueBuilder_ == null) {
-        minValueBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-            com.google.protobuf.DoubleValue, com.google.protobuf.DoubleValue.Builder, com.google.protobuf.DoubleValueOrBuilder>(
-                getMinValue(),
-                getParentForChildren(),
-                isClean());
-        minValue_ = null;
-      }
-      return minValueBuilder_;
-    }
-
-    private com.google.protobuf.DoubleValue maxValue_;
-    private com.google.protobuf.SingleFieldBuilderV3<
-        com.google.protobuf.DoubleValue, com.google.protobuf.DoubleValue.Builder, com.google.protobuf.DoubleValueOrBuilder> maxValueBuilder_;
-    /**
-     * <pre>
-     * The maximum acceptable value for the metric if specified
-     * </pre>
-     *
-     * <code>.google.protobuf.DoubleValue max_value = 4;</code>
-     */
-    public boolean hasMaxValue() {
-      return maxValueBuilder_ != null || maxValue_ != null;
-    }
-    /**
-     * <pre>
-     * The maximum acceptable value for the metric if specified
-     * </pre>
-     *
-     * <code>.google.protobuf.DoubleValue max_value = 4;</code>
-     */
-    public com.google.protobuf.DoubleValue getMaxValue() {
-      if (maxValueBuilder_ == null) {
-        return maxValue_ == null ? com.google.protobuf.DoubleValue.getDefaultInstance() : maxValue_;
-      } else {
-        return maxValueBuilder_.getMessage();
-      }
-    }
-    /**
-     * <pre>
-     * The maximum acceptable value for the metric if specified
-     * </pre>
-     *
-     * <code>.google.protobuf.DoubleValue max_value = 4;</code>
-     */
-    public Builder setMaxValue(com.google.protobuf.DoubleValue value) {
-      if (maxValueBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        maxValue_ = value;
-        onChanged();
-      } else {
-        maxValueBuilder_.setMessage(value);
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * The maximum acceptable value for the metric if specified
-     * </pre>
-     *
-     * <code>.google.protobuf.DoubleValue max_value = 4;</code>
-     */
-    public Builder setMaxValue(
-        com.google.protobuf.DoubleValue.Builder builderForValue) {
-      if (maxValueBuilder_ == null) {
-        maxValue_ = builderForValue.build();
-        onChanged();
-      } else {
-        maxValueBuilder_.setMessage(builderForValue.build());
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * The maximum acceptable value for the metric if specified
-     * </pre>
-     *
-     * <code>.google.protobuf.DoubleValue max_value = 4;</code>
-     */
-    public Builder mergeMaxValue(com.google.protobuf.DoubleValue value) {
-      if (maxValueBuilder_ == null) {
-        if (maxValue_ != null) {
-          maxValue_ =
-            com.google.protobuf.DoubleValue.newBuilder(maxValue_).mergeFrom(value).buildPartial();
-        } else {
-          maxValue_ = value;
-        }
-        onChanged();
-      } else {
-        maxValueBuilder_.mergeFrom(value);
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * The maximum acceptable value for the metric if specified
-     * </pre>
-     *
-     * <code>.google.protobuf.DoubleValue max_value = 4;</code>
-     */
-    public Builder clearMaxValue() {
-      if (maxValueBuilder_ == null) {
-        maxValue_ = null;
-        onChanged();
-      } else {
-        maxValue_ = null;
-        maxValueBuilder_ = null;
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * The maximum acceptable value for the metric if specified
-     * </pre>
-     *
-     * <code>.google.protobuf.DoubleValue max_value = 4;</code>
-     */
-    public com.google.protobuf.DoubleValue.Builder getMaxValueBuilder() {
-      
-      onChanged();
-      return getMaxValueFieldBuilder().getBuilder();
-    }
-    /**
-     * <pre>
-     * The maximum acceptable value for the metric if specified
-     * </pre>
-     *
-     * <code>.google.protobuf.DoubleValue max_value = 4;</code>
-     */
-    public com.google.protobuf.DoubleValueOrBuilder getMaxValueOrBuilder() {
-      if (maxValueBuilder_ != null) {
-        return maxValueBuilder_.getMessageOrBuilder();
-      } else {
-        return maxValue_ == null ?
-            com.google.protobuf.DoubleValue.getDefaultInstance() : maxValue_;
-      }
-    }
-    /**
-     * <pre>
-     * The maximum acceptable value for the metric if specified
-     * </pre>
-     *
-     * <code>.google.protobuf.DoubleValue max_value = 4;</code>
-     */
-    private com.google.protobuf.SingleFieldBuilderV3<
-        com.google.protobuf.DoubleValue, com.google.protobuf.DoubleValue.Builder, com.google.protobuf.DoubleValueOrBuilder> 
-        getMaxValueFieldBuilder() {
-      if (maxValueBuilder_ == null) {
-        maxValueBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-            com.google.protobuf.DoubleValue, com.google.protobuf.DoubleValue.Builder, com.google.protobuf.DoubleValueOrBuilder>(
-                getMaxValue(),
-                getParentForChildren(),
-                isClean());
-        maxValue_ = null;
-      }
-      return maxValueBuilder_;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.MetricEntry)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.MetricEntry)
-  private static final org.tensorflow.proto.util.testlog.MetricEntry DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.util.testlog.MetricEntry();
-  }
-
-  public static org.tensorflow.proto.util.testlog.MetricEntry getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<MetricEntry>
-      PARSER = new com.google.protobuf.AbstractParser<MetricEntry>() {
-    @java.lang.Override
-    public MetricEntry parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new MetricEntry(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<MetricEntry> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<MetricEntry> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.util.testlog.MetricEntry getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/MetricEntryOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/MetricEntryOrBuilder.java
deleted file mode 100644
index eeb12bca8a9..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/MetricEntryOrBuilder.java
+++ /dev/null
@@ -1,86 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/util/test_log.proto
-
-package org.tensorflow.proto.util.testlog;
-
-public interface MetricEntryOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.MetricEntry)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <pre>
-   * Metric name
-   * </pre>
-   *
-   * <code>string name = 1;</code>
-   */
-  java.lang.String getName();
-  /**
-   * <pre>
-   * Metric name
-   * </pre>
-   *
-   * <code>string name = 1;</code>
-   */
-  com.google.protobuf.ByteString
-      getNameBytes();
-
-  /**
-   * <pre>
-   * Metric value
-   * </pre>
-   *
-   * <code>double value = 2;</code>
-   */
-  double getValue();
-
-  /**
-   * <pre>
-   * The minimum acceptable value for the metric if specified
-   * </pre>
-   *
-   * <code>.google.protobuf.DoubleValue min_value = 3;</code>
-   */
-  boolean hasMinValue();
-  /**
-   * <pre>
-   * The minimum acceptable value for the metric if specified
-   * </pre>
-   *
-   * <code>.google.protobuf.DoubleValue min_value = 3;</code>
-   */
-  com.google.protobuf.DoubleValue getMinValue();
-  /**
-   * <pre>
-   * The minimum acceptable value for the metric if specified
-   * </pre>
-   *
-   * <code>.google.protobuf.DoubleValue min_value = 3;</code>
-   */
-  com.google.protobuf.DoubleValueOrBuilder getMinValueOrBuilder();
-
-  /**
-   * <pre>
-   * The maximum acceptable value for the metric if specified
-   * </pre>
-   *
-   * <code>.google.protobuf.DoubleValue max_value = 4;</code>
-   */
-  boolean hasMaxValue();
-  /**
-   * <pre>
-   * The maximum acceptable value for the metric if specified
-   * </pre>
-   *
-   * <code>.google.protobuf.DoubleValue max_value = 4;</code>
-   */
-  com.google.protobuf.DoubleValue getMaxValue();
-  /**
-   * <pre>
-   * The maximum acceptable value for the metric if specified
-   * </pre>
-   *
-   * <code>.google.protobuf.DoubleValue max_value = 4;</code>
-   */
-  com.google.protobuf.DoubleValueOrBuilder getMaxValueOrBuilder();
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/PlatformInfo.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/PlatformInfo.java
deleted file mode 100644
index efe3a52d92e..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/PlatformInfo.java
+++ /dev/null
@@ -1,1349 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/util/test_log.proto
-
-package org.tensorflow.proto.util.testlog;
-
-/**
- * Protobuf type {@code tensorflow.PlatformInfo}
- */
-public  final class PlatformInfo extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.PlatformInfo)
-    PlatformInfoOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use PlatformInfo.newBuilder() to construct.
-  private PlatformInfo(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private PlatformInfo() {
-    bits_ = "";
-    linkage_ = "";
-    machine_ = "";
-    release_ = "";
-    system_ = "";
-    version_ = "";
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new PlatformInfo();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private PlatformInfo(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 10: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            bits_ = s;
-            break;
-          }
-          case 18: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            linkage_ = s;
-            break;
-          }
-          case 26: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            machine_ = s;
-            break;
-          }
-          case 34: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            release_ = s;
-            break;
-          }
-          case 42: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            system_ = s;
-            break;
-          }
-          case 50: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            version_ = s;
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_PlatformInfo_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_PlatformInfo_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.util.testlog.PlatformInfo.class, org.tensorflow.proto.util.testlog.PlatformInfo.Builder.class);
-  }
-
-  public static final int BITS_FIELD_NUMBER = 1;
-  private volatile java.lang.Object bits_;
-  /**
-   * <pre>
-   * e.g. '64bit'
-   * </pre>
-   *
-   * <code>string bits = 1;</code>
-   */
-  public java.lang.String getBits() {
-    java.lang.Object ref = bits_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      bits_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * e.g. '64bit'
-   * </pre>
-   *
-   * <code>string bits = 1;</code>
-   */
-  public com.google.protobuf.ByteString
-      getBitsBytes() {
-    java.lang.Object ref = bits_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      bits_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int LINKAGE_FIELD_NUMBER = 2;
-  private volatile java.lang.Object linkage_;
-  /**
-   * <pre>
-   * e.g. 'ELF'
-   * </pre>
-   *
-   * <code>string linkage = 2;</code>
-   */
-  public java.lang.String getLinkage() {
-    java.lang.Object ref = linkage_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      linkage_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * e.g. 'ELF'
-   * </pre>
-   *
-   * <code>string linkage = 2;</code>
-   */
-  public com.google.protobuf.ByteString
-      getLinkageBytes() {
-    java.lang.Object ref = linkage_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      linkage_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int MACHINE_FIELD_NUMBER = 3;
-  private volatile java.lang.Object machine_;
-  /**
-   * <pre>
-   * e.g. 'i386'
-   * </pre>
-   *
-   * <code>string machine = 3;</code>
-   */
-  public java.lang.String getMachine() {
-    java.lang.Object ref = machine_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      machine_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * e.g. 'i386'
-   * </pre>
-   *
-   * <code>string machine = 3;</code>
-   */
-  public com.google.protobuf.ByteString
-      getMachineBytes() {
-    java.lang.Object ref = machine_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      machine_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int RELEASE_FIELD_NUMBER = 4;
-  private volatile java.lang.Object release_;
-  /**
-   * <pre>
-   * e.g. '3.13.0-76-generic'
-   * </pre>
-   *
-   * <code>string release = 4;</code>
-   */
-  public java.lang.String getRelease() {
-    java.lang.Object ref = release_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      release_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * e.g. '3.13.0-76-generic'
-   * </pre>
-   *
-   * <code>string release = 4;</code>
-   */
-  public com.google.protobuf.ByteString
-      getReleaseBytes() {
-    java.lang.Object ref = release_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      release_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int SYSTEM_FIELD_NUMBER = 5;
-  private volatile java.lang.Object system_;
-  /**
-   * <pre>
-   * e.g. 'Linux'
-   * </pre>
-   *
-   * <code>string system = 5;</code>
-   */
-  public java.lang.String getSystem() {
-    java.lang.Object ref = system_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      system_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * e.g. 'Linux'
-   * </pre>
-   *
-   * <code>string system = 5;</code>
-   */
-  public com.google.protobuf.ByteString
-      getSystemBytes() {
-    java.lang.Object ref = system_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      system_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int VERSION_FIELD_NUMBER = 6;
-  private volatile java.lang.Object version_;
-  /**
-   * <pre>
-   * e.g. '#120-Ubuntu SMP Mon Jan 18 15:59:10 UTC 2016'
-   * </pre>
-   *
-   * <code>string version = 6;</code>
-   */
-  public java.lang.String getVersion() {
-    java.lang.Object ref = version_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      version_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * e.g. '#120-Ubuntu SMP Mon Jan 18 15:59:10 UTC 2016'
-   * </pre>
-   *
-   * <code>string version = 6;</code>
-   */
-  public com.google.protobuf.ByteString
-      getVersionBytes() {
-    java.lang.Object ref = version_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      version_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    if (!getBitsBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 1, bits_);
-    }
-    if (!getLinkageBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 2, linkage_);
-    }
-    if (!getMachineBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 3, machine_);
-    }
-    if (!getReleaseBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 4, release_);
-    }
-    if (!getSystemBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 5, system_);
-    }
-    if (!getVersionBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 6, version_);
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (!getBitsBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, bits_);
-    }
-    if (!getLinkageBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(2, linkage_);
-    }
-    if (!getMachineBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(3, machine_);
-    }
-    if (!getReleaseBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(4, release_);
-    }
-    if (!getSystemBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(5, system_);
-    }
-    if (!getVersionBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(6, version_);
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.util.testlog.PlatformInfo)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.util.testlog.PlatformInfo other = (org.tensorflow.proto.util.testlog.PlatformInfo) obj;
-
-    if (!getBits()
-        .equals(other.getBits())) return false;
-    if (!getLinkage()
-        .equals(other.getLinkage())) return false;
-    if (!getMachine()
-        .equals(other.getMachine())) return false;
-    if (!getRelease()
-        .equals(other.getRelease())) return false;
-    if (!getSystem()
-        .equals(other.getSystem())) return false;
-    if (!getVersion()
-        .equals(other.getVersion())) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    hash = (37 * hash) + BITS_FIELD_NUMBER;
-    hash = (53 * hash) + getBits().hashCode();
-    hash = (37 * hash) + LINKAGE_FIELD_NUMBER;
-    hash = (53 * hash) + getLinkage().hashCode();
-    hash = (37 * hash) + MACHINE_FIELD_NUMBER;
-    hash = (53 * hash) + getMachine().hashCode();
-    hash = (37 * hash) + RELEASE_FIELD_NUMBER;
-    hash = (53 * hash) + getRelease().hashCode();
-    hash = (37 * hash) + SYSTEM_FIELD_NUMBER;
-    hash = (53 * hash) + getSystem().hashCode();
-    hash = (37 * hash) + VERSION_FIELD_NUMBER;
-    hash = (53 * hash) + getVersion().hashCode();
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.util.testlog.PlatformInfo parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.testlog.PlatformInfo parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.testlog.PlatformInfo parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.testlog.PlatformInfo parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.testlog.PlatformInfo parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.testlog.PlatformInfo parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.testlog.PlatformInfo parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.testlog.PlatformInfo parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.testlog.PlatformInfo parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.testlog.PlatformInfo parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.testlog.PlatformInfo parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.testlog.PlatformInfo parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.util.testlog.PlatformInfo prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * Protobuf type {@code tensorflow.PlatformInfo}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.PlatformInfo)
-      org.tensorflow.proto.util.testlog.PlatformInfoOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_PlatformInfo_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_PlatformInfo_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.util.testlog.PlatformInfo.class, org.tensorflow.proto.util.testlog.PlatformInfo.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.util.testlog.PlatformInfo.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      bits_ = "";
-
-      linkage_ = "";
-
-      machine_ = "";
-
-      release_ = "";
-
-      system_ = "";
-
-      version_ = "";
-
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_PlatformInfo_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.testlog.PlatformInfo getDefaultInstanceForType() {
-      return org.tensorflow.proto.util.testlog.PlatformInfo.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.testlog.PlatformInfo build() {
-      org.tensorflow.proto.util.testlog.PlatformInfo result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.testlog.PlatformInfo buildPartial() {
-      org.tensorflow.proto.util.testlog.PlatformInfo result = new org.tensorflow.proto.util.testlog.PlatformInfo(this);
-      result.bits_ = bits_;
-      result.linkage_ = linkage_;
-      result.machine_ = machine_;
-      result.release_ = release_;
-      result.system_ = system_;
-      result.version_ = version_;
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.util.testlog.PlatformInfo) {
-        return mergeFrom((org.tensorflow.proto.util.testlog.PlatformInfo)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.util.testlog.PlatformInfo other) {
-      if (other == org.tensorflow.proto.util.testlog.PlatformInfo.getDefaultInstance()) return this;
-      if (!other.getBits().isEmpty()) {
-        bits_ = other.bits_;
-        onChanged();
-      }
-      if (!other.getLinkage().isEmpty()) {
-        linkage_ = other.linkage_;
-        onChanged();
-      }
-      if (!other.getMachine().isEmpty()) {
-        machine_ = other.machine_;
-        onChanged();
-      }
-      if (!other.getRelease().isEmpty()) {
-        release_ = other.release_;
-        onChanged();
-      }
-      if (!other.getSystem().isEmpty()) {
-        system_ = other.system_;
-        onChanged();
-      }
-      if (!other.getVersion().isEmpty()) {
-        version_ = other.version_;
-        onChanged();
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.util.testlog.PlatformInfo parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.util.testlog.PlatformInfo) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-
-    private java.lang.Object bits_ = "";
-    /**
-     * <pre>
-     * e.g. '64bit'
-     * </pre>
-     *
-     * <code>string bits = 1;</code>
-     */
-    public java.lang.String getBits() {
-      java.lang.Object ref = bits_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        bits_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * e.g. '64bit'
-     * </pre>
-     *
-     * <code>string bits = 1;</code>
-     */
-    public com.google.protobuf.ByteString
-        getBitsBytes() {
-      java.lang.Object ref = bits_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        bits_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * e.g. '64bit'
-     * </pre>
-     *
-     * <code>string bits = 1;</code>
-     */
-    public Builder setBits(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      bits_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * e.g. '64bit'
-     * </pre>
-     *
-     * <code>string bits = 1;</code>
-     */
-    public Builder clearBits() {
-      
-      bits_ = getDefaultInstance().getBits();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * e.g. '64bit'
-     * </pre>
-     *
-     * <code>string bits = 1;</code>
-     */
-    public Builder setBitsBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      bits_ = value;
-      onChanged();
-      return this;
-    }
-
-    private java.lang.Object linkage_ = "";
-    /**
-     * <pre>
-     * e.g. 'ELF'
-     * </pre>
-     *
-     * <code>string linkage = 2;</code>
-     */
-    public java.lang.String getLinkage() {
-      java.lang.Object ref = linkage_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        linkage_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * e.g. 'ELF'
-     * </pre>
-     *
-     * <code>string linkage = 2;</code>
-     */
-    public com.google.protobuf.ByteString
-        getLinkageBytes() {
-      java.lang.Object ref = linkage_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        linkage_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * e.g. 'ELF'
-     * </pre>
-     *
-     * <code>string linkage = 2;</code>
-     */
-    public Builder setLinkage(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      linkage_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * e.g. 'ELF'
-     * </pre>
-     *
-     * <code>string linkage = 2;</code>
-     */
-    public Builder clearLinkage() {
-      
-      linkage_ = getDefaultInstance().getLinkage();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * e.g. 'ELF'
-     * </pre>
-     *
-     * <code>string linkage = 2;</code>
-     */
-    public Builder setLinkageBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      linkage_ = value;
-      onChanged();
-      return this;
-    }
-
-    private java.lang.Object machine_ = "";
-    /**
-     * <pre>
-     * e.g. 'i386'
-     * </pre>
-     *
-     * <code>string machine = 3;</code>
-     */
-    public java.lang.String getMachine() {
-      java.lang.Object ref = machine_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        machine_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * e.g. 'i386'
-     * </pre>
-     *
-     * <code>string machine = 3;</code>
-     */
-    public com.google.protobuf.ByteString
-        getMachineBytes() {
-      java.lang.Object ref = machine_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        machine_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * e.g. 'i386'
-     * </pre>
-     *
-     * <code>string machine = 3;</code>
-     */
-    public Builder setMachine(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      machine_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * e.g. 'i386'
-     * </pre>
-     *
-     * <code>string machine = 3;</code>
-     */
-    public Builder clearMachine() {
-      
-      machine_ = getDefaultInstance().getMachine();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * e.g. 'i386'
-     * </pre>
-     *
-     * <code>string machine = 3;</code>
-     */
-    public Builder setMachineBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      machine_ = value;
-      onChanged();
-      return this;
-    }
-
-    private java.lang.Object release_ = "";
-    /**
-     * <pre>
-     * e.g. '3.13.0-76-generic'
-     * </pre>
-     *
-     * <code>string release = 4;</code>
-     */
-    public java.lang.String getRelease() {
-      java.lang.Object ref = release_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        release_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * e.g. '3.13.0-76-generic'
-     * </pre>
-     *
-     * <code>string release = 4;</code>
-     */
-    public com.google.protobuf.ByteString
-        getReleaseBytes() {
-      java.lang.Object ref = release_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        release_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * e.g. '3.13.0-76-generic'
-     * </pre>
-     *
-     * <code>string release = 4;</code>
-     */
-    public Builder setRelease(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      release_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * e.g. '3.13.0-76-generic'
-     * </pre>
-     *
-     * <code>string release = 4;</code>
-     */
-    public Builder clearRelease() {
-      
-      release_ = getDefaultInstance().getRelease();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * e.g. '3.13.0-76-generic'
-     * </pre>
-     *
-     * <code>string release = 4;</code>
-     */
-    public Builder setReleaseBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      release_ = value;
-      onChanged();
-      return this;
-    }
-
-    private java.lang.Object system_ = "";
-    /**
-     * <pre>
-     * e.g. 'Linux'
-     * </pre>
-     *
-     * <code>string system = 5;</code>
-     */
-    public java.lang.String getSystem() {
-      java.lang.Object ref = system_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        system_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * e.g. 'Linux'
-     * </pre>
-     *
-     * <code>string system = 5;</code>
-     */
-    public com.google.protobuf.ByteString
-        getSystemBytes() {
-      java.lang.Object ref = system_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        system_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * e.g. 'Linux'
-     * </pre>
-     *
-     * <code>string system = 5;</code>
-     */
-    public Builder setSystem(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      system_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * e.g. 'Linux'
-     * </pre>
-     *
-     * <code>string system = 5;</code>
-     */
-    public Builder clearSystem() {
-      
-      system_ = getDefaultInstance().getSystem();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * e.g. 'Linux'
-     * </pre>
-     *
-     * <code>string system = 5;</code>
-     */
-    public Builder setSystemBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      system_ = value;
-      onChanged();
-      return this;
-    }
-
-    private java.lang.Object version_ = "";
-    /**
-     * <pre>
-     * e.g. '#120-Ubuntu SMP Mon Jan 18 15:59:10 UTC 2016'
-     * </pre>
-     *
-     * <code>string version = 6;</code>
-     */
-    public java.lang.String getVersion() {
-      java.lang.Object ref = version_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        version_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * e.g. '#120-Ubuntu SMP Mon Jan 18 15:59:10 UTC 2016'
-     * </pre>
-     *
-     * <code>string version = 6;</code>
-     */
-    public com.google.protobuf.ByteString
-        getVersionBytes() {
-      java.lang.Object ref = version_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        version_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * e.g. '#120-Ubuntu SMP Mon Jan 18 15:59:10 UTC 2016'
-     * </pre>
-     *
-     * <code>string version = 6;</code>
-     */
-    public Builder setVersion(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      version_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * e.g. '#120-Ubuntu SMP Mon Jan 18 15:59:10 UTC 2016'
-     * </pre>
-     *
-     * <code>string version = 6;</code>
-     */
-    public Builder clearVersion() {
-      
-      version_ = getDefaultInstance().getVersion();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * e.g. '#120-Ubuntu SMP Mon Jan 18 15:59:10 UTC 2016'
-     * </pre>
-     *
-     * <code>string version = 6;</code>
-     */
-    public Builder setVersionBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      version_ = value;
-      onChanged();
-      return this;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.PlatformInfo)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.PlatformInfo)
-  private static final org.tensorflow.proto.util.testlog.PlatformInfo DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.util.testlog.PlatformInfo();
-  }
-
-  public static org.tensorflow.proto.util.testlog.PlatformInfo getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<PlatformInfo>
-      PARSER = new com.google.protobuf.AbstractParser<PlatformInfo>() {
-    @java.lang.Override
-    public PlatformInfo parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new PlatformInfo(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<PlatformInfo> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<PlatformInfo> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.util.testlog.PlatformInfo getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/PlatformInfoOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/PlatformInfoOrBuilder.java
deleted file mode 100644
index 6a5089fa309..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/PlatformInfoOrBuilder.java
+++ /dev/null
@@ -1,117 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/util/test_log.proto
-
-package org.tensorflow.proto.util.testlog;
-
-public interface PlatformInfoOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.PlatformInfo)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <pre>
-   * e.g. '64bit'
-   * </pre>
-   *
-   * <code>string bits = 1;</code>
-   */
-  java.lang.String getBits();
-  /**
-   * <pre>
-   * e.g. '64bit'
-   * </pre>
-   *
-   * <code>string bits = 1;</code>
-   */
-  com.google.protobuf.ByteString
-      getBitsBytes();
-
-  /**
-   * <pre>
-   * e.g. 'ELF'
-   * </pre>
-   *
-   * <code>string linkage = 2;</code>
-   */
-  java.lang.String getLinkage();
-  /**
-   * <pre>
-   * e.g. 'ELF'
-   * </pre>
-   *
-   * <code>string linkage = 2;</code>
-   */
-  com.google.protobuf.ByteString
-      getLinkageBytes();
-
-  /**
-   * <pre>
-   * e.g. 'i386'
-   * </pre>
-   *
-   * <code>string machine = 3;</code>
-   */
-  java.lang.String getMachine();
-  /**
-   * <pre>
-   * e.g. 'i386'
-   * </pre>
-   *
-   * <code>string machine = 3;</code>
-   */
-  com.google.protobuf.ByteString
-      getMachineBytes();
-
-  /**
-   * <pre>
-   * e.g. '3.13.0-76-generic'
-   * </pre>
-   *
-   * <code>string release = 4;</code>
-   */
-  java.lang.String getRelease();
-  /**
-   * <pre>
-   * e.g. '3.13.0-76-generic'
-   * </pre>
-   *
-   * <code>string release = 4;</code>
-   */
-  com.google.protobuf.ByteString
-      getReleaseBytes();
-
-  /**
-   * <pre>
-   * e.g. 'Linux'
-   * </pre>
-   *
-   * <code>string system = 5;</code>
-   */
-  java.lang.String getSystem();
-  /**
-   * <pre>
-   * e.g. 'Linux'
-   * </pre>
-   *
-   * <code>string system = 5;</code>
-   */
-  com.google.protobuf.ByteString
-      getSystemBytes();
-
-  /**
-   * <pre>
-   * e.g. '#120-Ubuntu SMP Mon Jan 18 15:59:10 UTC 2016'
-   * </pre>
-   *
-   * <code>string version = 6;</code>
-   */
-  java.lang.String getVersion();
-  /**
-   * <pre>
-   * e.g. '#120-Ubuntu SMP Mon Jan 18 15:59:10 UTC 2016'
-   * </pre>
-   *
-   * <code>string version = 6;</code>
-   */
-  com.google.protobuf.ByteString
-      getVersionBytes();
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/RunConfiguration.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/RunConfiguration.java
deleted file mode 100644
index 1ec51529d32..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/RunConfiguration.java
+++ /dev/null
@@ -1,917 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/util/test_log.proto
-
-package org.tensorflow.proto.util.testlog;
-
-/**
- * <pre>
- * Run-specific items such as arguments to the test / benchmark.
- * </pre>
- *
- * Protobuf type {@code tensorflow.RunConfiguration}
- */
-public  final class RunConfiguration extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.RunConfiguration)
-    RunConfigurationOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use RunConfiguration.newBuilder() to construct.
-  private RunConfiguration(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private RunConfiguration() {
-    argument_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new RunConfiguration();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private RunConfiguration(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    int mutable_bitField0_ = 0;
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 10: {
-            java.lang.String s = input.readStringRequireUtf8();
-            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
-              argument_ = new com.google.protobuf.LazyStringArrayList();
-              mutable_bitField0_ |= 0x00000001;
-            }
-            argument_.add(s);
-            break;
-          }
-          case 18: {
-            if (!((mutable_bitField0_ & 0x00000002) != 0)) {
-              envVars_ = com.google.protobuf.MapField.newMapField(
-                  EnvVarsDefaultEntryHolder.defaultEntry);
-              mutable_bitField0_ |= 0x00000002;
-            }
-            com.google.protobuf.MapEntry<java.lang.String, java.lang.String>
-            envVars__ = input.readMessage(
-                EnvVarsDefaultEntryHolder.defaultEntry.getParserForType(), extensionRegistry);
-            envVars_.getMutableMap().put(
-                envVars__.getKey(), envVars__.getValue());
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      if (((mutable_bitField0_ & 0x00000001) != 0)) {
-        argument_ = argument_.getUnmodifiableView();
-      }
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_RunConfiguration_descriptor;
-  }
-
-  @SuppressWarnings({"rawtypes"})
-  @java.lang.Override
-  protected com.google.protobuf.MapField internalGetMapField(
-      int number) {
-    switch (number) {
-      case 2:
-        return internalGetEnvVars();
-      default:
-        throw new RuntimeException(
-            "Invalid map field number: " + number);
-    }
-  }
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_RunConfiguration_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.util.testlog.RunConfiguration.class, org.tensorflow.proto.util.testlog.RunConfiguration.Builder.class);
-  }
-
-  public static final int ARGUMENT_FIELD_NUMBER = 1;
-  private com.google.protobuf.LazyStringList argument_;
-  /**
-   * <code>repeated string argument = 1;</code>
-   */
-  public com.google.protobuf.ProtocolStringList
-      getArgumentList() {
-    return argument_;
-  }
-  /**
-   * <code>repeated string argument = 1;</code>
-   */
-  public int getArgumentCount() {
-    return argument_.size();
-  }
-  /**
-   * <code>repeated string argument = 1;</code>
-   */
-  public java.lang.String getArgument(int index) {
-    return argument_.get(index);
-  }
-  /**
-   * <code>repeated string argument = 1;</code>
-   */
-  public com.google.protobuf.ByteString
-      getArgumentBytes(int index) {
-    return argument_.getByteString(index);
-  }
-
-  public static final int ENV_VARS_FIELD_NUMBER = 2;
-  private static final class EnvVarsDefaultEntryHolder {
-    static final com.google.protobuf.MapEntry<
-        java.lang.String, java.lang.String> defaultEntry =
-            com.google.protobuf.MapEntry
-            .<java.lang.String, java.lang.String>newDefaultInstance(
-                org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_RunConfiguration_EnvVarsEntry_descriptor, 
-                com.google.protobuf.WireFormat.FieldType.STRING,
-                "",
-                com.google.protobuf.WireFormat.FieldType.STRING,
-                "");
-  }
-  private com.google.protobuf.MapField<
-      java.lang.String, java.lang.String> envVars_;
-  private com.google.protobuf.MapField<java.lang.String, java.lang.String>
-  internalGetEnvVars() {
-    if (envVars_ == null) {
-      return com.google.protobuf.MapField.emptyMapField(
-          EnvVarsDefaultEntryHolder.defaultEntry);
-    }
-    return envVars_;
-  }
-
-  public int getEnvVarsCount() {
-    return internalGetEnvVars().getMap().size();
-  }
-  /**
-   * <pre>
-   * Environment variables used to run the test/benchmark.
-   * </pre>
-   *
-   * <code>map&lt;string, string&gt; env_vars = 2;</code>
-   */
-
-  public boolean containsEnvVars(
-      java.lang.String key) {
-    if (key == null) { throw new java.lang.NullPointerException(); }
-    return internalGetEnvVars().getMap().containsKey(key);
-  }
-  /**
-   * Use {@link #getEnvVarsMap()} instead.
-   */
-  @java.lang.Deprecated
-  public java.util.Map<java.lang.String, java.lang.String> getEnvVars() {
-    return getEnvVarsMap();
-  }
-  /**
-   * <pre>
-   * Environment variables used to run the test/benchmark.
-   * </pre>
-   *
-   * <code>map&lt;string, string&gt; env_vars = 2;</code>
-   */
-
-  public java.util.Map<java.lang.String, java.lang.String> getEnvVarsMap() {
-    return internalGetEnvVars().getMap();
-  }
-  /**
-   * <pre>
-   * Environment variables used to run the test/benchmark.
-   * </pre>
-   *
-   * <code>map&lt;string, string&gt; env_vars = 2;</code>
-   */
-
-  public java.lang.String getEnvVarsOrDefault(
-      java.lang.String key,
-      java.lang.String defaultValue) {
-    if (key == null) { throw new java.lang.NullPointerException(); }
-    java.util.Map<java.lang.String, java.lang.String> map =
-        internalGetEnvVars().getMap();
-    return map.containsKey(key) ? map.get(key) : defaultValue;
-  }
-  /**
-   * <pre>
-   * Environment variables used to run the test/benchmark.
-   * </pre>
-   *
-   * <code>map&lt;string, string&gt; env_vars = 2;</code>
-   */
-
-  public java.lang.String getEnvVarsOrThrow(
-      java.lang.String key) {
-    if (key == null) { throw new java.lang.NullPointerException(); }
-    java.util.Map<java.lang.String, java.lang.String> map =
-        internalGetEnvVars().getMap();
-    if (!map.containsKey(key)) {
-      throw new java.lang.IllegalArgumentException();
-    }
-    return map.get(key);
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    for (int i = 0; i < argument_.size(); i++) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 1, argument_.getRaw(i));
-    }
-    com.google.protobuf.GeneratedMessageV3
-      .serializeStringMapTo(
-        output,
-        internalGetEnvVars(),
-        EnvVarsDefaultEntryHolder.defaultEntry,
-        2);
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    {
-      int dataSize = 0;
-      for (int i = 0; i < argument_.size(); i++) {
-        dataSize += computeStringSizeNoTag(argument_.getRaw(i));
-      }
-      size += dataSize;
-      size += 1 * getArgumentList().size();
-    }
-    for (java.util.Map.Entry<java.lang.String, java.lang.String> entry
-         : internalGetEnvVars().getMap().entrySet()) {
-      com.google.protobuf.MapEntry<java.lang.String, java.lang.String>
-      envVars__ = EnvVarsDefaultEntryHolder.defaultEntry.newBuilderForType()
-          .setKey(entry.getKey())
-          .setValue(entry.getValue())
-          .build();
-      size += com.google.protobuf.CodedOutputStream
-          .computeMessageSize(2, envVars__);
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.util.testlog.RunConfiguration)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.util.testlog.RunConfiguration other = (org.tensorflow.proto.util.testlog.RunConfiguration) obj;
-
-    if (!getArgumentList()
-        .equals(other.getArgumentList())) return false;
-    if (!internalGetEnvVars().equals(
-        other.internalGetEnvVars())) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    if (getArgumentCount() > 0) {
-      hash = (37 * hash) + ARGUMENT_FIELD_NUMBER;
-      hash = (53 * hash) + getArgumentList().hashCode();
-    }
-    if (!internalGetEnvVars().getMap().isEmpty()) {
-      hash = (37 * hash) + ENV_VARS_FIELD_NUMBER;
-      hash = (53 * hash) + internalGetEnvVars().hashCode();
-    }
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.util.testlog.RunConfiguration parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.testlog.RunConfiguration parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.testlog.RunConfiguration parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.testlog.RunConfiguration parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.testlog.RunConfiguration parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.testlog.RunConfiguration parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.testlog.RunConfiguration parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.testlog.RunConfiguration parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.testlog.RunConfiguration parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.testlog.RunConfiguration parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.testlog.RunConfiguration parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.testlog.RunConfiguration parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.util.testlog.RunConfiguration prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * <pre>
-   * Run-specific items such as arguments to the test / benchmark.
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.RunConfiguration}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.RunConfiguration)
-      org.tensorflow.proto.util.testlog.RunConfigurationOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_RunConfiguration_descriptor;
-    }
-
-    @SuppressWarnings({"rawtypes"})
-    protected com.google.protobuf.MapField internalGetMapField(
-        int number) {
-      switch (number) {
-        case 2:
-          return internalGetEnvVars();
-        default:
-          throw new RuntimeException(
-              "Invalid map field number: " + number);
-      }
-    }
-    @SuppressWarnings({"rawtypes"})
-    protected com.google.protobuf.MapField internalGetMutableMapField(
-        int number) {
-      switch (number) {
-        case 2:
-          return internalGetMutableEnvVars();
-        default:
-          throw new RuntimeException(
-              "Invalid map field number: " + number);
-      }
-    }
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_RunConfiguration_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.util.testlog.RunConfiguration.class, org.tensorflow.proto.util.testlog.RunConfiguration.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.util.testlog.RunConfiguration.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      argument_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-      bitField0_ = (bitField0_ & ~0x00000001);
-      internalGetMutableEnvVars().clear();
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_RunConfiguration_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.testlog.RunConfiguration getDefaultInstanceForType() {
-      return org.tensorflow.proto.util.testlog.RunConfiguration.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.testlog.RunConfiguration build() {
-      org.tensorflow.proto.util.testlog.RunConfiguration result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.testlog.RunConfiguration buildPartial() {
-      org.tensorflow.proto.util.testlog.RunConfiguration result = new org.tensorflow.proto.util.testlog.RunConfiguration(this);
-      int from_bitField0_ = bitField0_;
-      if (((bitField0_ & 0x00000001) != 0)) {
-        argument_ = argument_.getUnmodifiableView();
-        bitField0_ = (bitField0_ & ~0x00000001);
-      }
-      result.argument_ = argument_;
-      result.envVars_ = internalGetEnvVars();
-      result.envVars_.makeImmutable();
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.util.testlog.RunConfiguration) {
-        return mergeFrom((org.tensorflow.proto.util.testlog.RunConfiguration)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.util.testlog.RunConfiguration other) {
-      if (other == org.tensorflow.proto.util.testlog.RunConfiguration.getDefaultInstance()) return this;
-      if (!other.argument_.isEmpty()) {
-        if (argument_.isEmpty()) {
-          argument_ = other.argument_;
-          bitField0_ = (bitField0_ & ~0x00000001);
-        } else {
-          ensureArgumentIsMutable();
-          argument_.addAll(other.argument_);
-        }
-        onChanged();
-      }
-      internalGetMutableEnvVars().mergeFrom(
-          other.internalGetEnvVars());
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.util.testlog.RunConfiguration parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.util.testlog.RunConfiguration) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-    private int bitField0_;
-
-    private com.google.protobuf.LazyStringList argument_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-    private void ensureArgumentIsMutable() {
-      if (!((bitField0_ & 0x00000001) != 0)) {
-        argument_ = new com.google.protobuf.LazyStringArrayList(argument_);
-        bitField0_ |= 0x00000001;
-       }
-    }
-    /**
-     * <code>repeated string argument = 1;</code>
-     */
-    public com.google.protobuf.ProtocolStringList
-        getArgumentList() {
-      return argument_.getUnmodifiableView();
-    }
-    /**
-     * <code>repeated string argument = 1;</code>
-     */
-    public int getArgumentCount() {
-      return argument_.size();
-    }
-    /**
-     * <code>repeated string argument = 1;</code>
-     */
-    public java.lang.String getArgument(int index) {
-      return argument_.get(index);
-    }
-    /**
-     * <code>repeated string argument = 1;</code>
-     */
-    public com.google.protobuf.ByteString
-        getArgumentBytes(int index) {
-      return argument_.getByteString(index);
-    }
-    /**
-     * <code>repeated string argument = 1;</code>
-     */
-    public Builder setArgument(
-        int index, java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  ensureArgumentIsMutable();
-      argument_.set(index, value);
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>repeated string argument = 1;</code>
-     */
-    public Builder addArgument(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  ensureArgumentIsMutable();
-      argument_.add(value);
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>repeated string argument = 1;</code>
-     */
-    public Builder addAllArgument(
-        java.lang.Iterable<java.lang.String> values) {
-      ensureArgumentIsMutable();
-      com.google.protobuf.AbstractMessageLite.Builder.addAll(
-          values, argument_);
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>repeated string argument = 1;</code>
-     */
-    public Builder clearArgument() {
-      argument_ = com.google.protobuf.LazyStringArrayList.EMPTY;
-      bitField0_ = (bitField0_ & ~0x00000001);
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>repeated string argument = 1;</code>
-     */
-    public Builder addArgumentBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      ensureArgumentIsMutable();
-      argument_.add(value);
-      onChanged();
-      return this;
-    }
-
-    private com.google.protobuf.MapField<
-        java.lang.String, java.lang.String> envVars_;
-    private com.google.protobuf.MapField<java.lang.String, java.lang.String>
-    internalGetEnvVars() {
-      if (envVars_ == null) {
-        return com.google.protobuf.MapField.emptyMapField(
-            EnvVarsDefaultEntryHolder.defaultEntry);
-      }
-      return envVars_;
-    }
-    private com.google.protobuf.MapField<java.lang.String, java.lang.String>
-    internalGetMutableEnvVars() {
-      onChanged();;
-      if (envVars_ == null) {
-        envVars_ = com.google.protobuf.MapField.newMapField(
-            EnvVarsDefaultEntryHolder.defaultEntry);
-      }
-      if (!envVars_.isMutable()) {
-        envVars_ = envVars_.copy();
-      }
-      return envVars_;
-    }
-
-    public int getEnvVarsCount() {
-      return internalGetEnvVars().getMap().size();
-    }
-    /**
-     * <pre>
-     * Environment variables used to run the test/benchmark.
-     * </pre>
-     *
-     * <code>map&lt;string, string&gt; env_vars = 2;</code>
-     */
-
-    public boolean containsEnvVars(
-        java.lang.String key) {
-      if (key == null) { throw new java.lang.NullPointerException(); }
-      return internalGetEnvVars().getMap().containsKey(key);
-    }
-    /**
-     * Use {@link #getEnvVarsMap()} instead.
-     */
-    @java.lang.Deprecated
-    public java.util.Map<java.lang.String, java.lang.String> getEnvVars() {
-      return getEnvVarsMap();
-    }
-    /**
-     * <pre>
-     * Environment variables used to run the test/benchmark.
-     * </pre>
-     *
-     * <code>map&lt;string, string&gt; env_vars = 2;</code>
-     */
-
-    public java.util.Map<java.lang.String, java.lang.String> getEnvVarsMap() {
-      return internalGetEnvVars().getMap();
-    }
-    /**
-     * <pre>
-     * Environment variables used to run the test/benchmark.
-     * </pre>
-     *
-     * <code>map&lt;string, string&gt; env_vars = 2;</code>
-     */
-
-    public java.lang.String getEnvVarsOrDefault(
-        java.lang.String key,
-        java.lang.String defaultValue) {
-      if (key == null) { throw new java.lang.NullPointerException(); }
-      java.util.Map<java.lang.String, java.lang.String> map =
-          internalGetEnvVars().getMap();
-      return map.containsKey(key) ? map.get(key) : defaultValue;
-    }
-    /**
-     * <pre>
-     * Environment variables used to run the test/benchmark.
-     * </pre>
-     *
-     * <code>map&lt;string, string&gt; env_vars = 2;</code>
-     */
-
-    public java.lang.String getEnvVarsOrThrow(
-        java.lang.String key) {
-      if (key == null) { throw new java.lang.NullPointerException(); }
-      java.util.Map<java.lang.String, java.lang.String> map =
-          internalGetEnvVars().getMap();
-      if (!map.containsKey(key)) {
-        throw new java.lang.IllegalArgumentException();
-      }
-      return map.get(key);
-    }
-
-    public Builder clearEnvVars() {
-      internalGetMutableEnvVars().getMutableMap()
-          .clear();
-      return this;
-    }
-    /**
-     * <pre>
-     * Environment variables used to run the test/benchmark.
-     * </pre>
-     *
-     * <code>map&lt;string, string&gt; env_vars = 2;</code>
-     */
-
-    public Builder removeEnvVars(
-        java.lang.String key) {
-      if (key == null) { throw new java.lang.NullPointerException(); }
-      internalGetMutableEnvVars().getMutableMap()
-          .remove(key);
-      return this;
-    }
-    /**
-     * Use alternate mutation accessors instead.
-     */
-    @java.lang.Deprecated
-    public java.util.Map<java.lang.String, java.lang.String>
-    getMutableEnvVars() {
-      return internalGetMutableEnvVars().getMutableMap();
-    }
-    /**
-     * <pre>
-     * Environment variables used to run the test/benchmark.
-     * </pre>
-     *
-     * <code>map&lt;string, string&gt; env_vars = 2;</code>
-     */
-    public Builder putEnvVars(
-        java.lang.String key,
-        java.lang.String value) {
-      if (key == null) { throw new java.lang.NullPointerException(); }
-      if (value == null) { throw new java.lang.NullPointerException(); }
-      internalGetMutableEnvVars().getMutableMap()
-          .put(key, value);
-      return this;
-    }
-    /**
-     * <pre>
-     * Environment variables used to run the test/benchmark.
-     * </pre>
-     *
-     * <code>map&lt;string, string&gt; env_vars = 2;</code>
-     */
-
-    public Builder putAllEnvVars(
-        java.util.Map<java.lang.String, java.lang.String> values) {
-      internalGetMutableEnvVars().getMutableMap()
-          .putAll(values);
-      return this;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.RunConfiguration)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.RunConfiguration)
-  private static final org.tensorflow.proto.util.testlog.RunConfiguration DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.util.testlog.RunConfiguration();
-  }
-
-  public static org.tensorflow.proto.util.testlog.RunConfiguration getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<RunConfiguration>
-      PARSER = new com.google.protobuf.AbstractParser<RunConfiguration>() {
-    @java.lang.Override
-    public RunConfiguration parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new RunConfiguration(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<RunConfiguration> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<RunConfiguration> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.util.testlog.RunConfiguration getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/RunConfigurationOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/RunConfigurationOrBuilder.java
deleted file mode 100644
index 9150f64ee59..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/RunConfigurationOrBuilder.java
+++ /dev/null
@@ -1,82 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/util/test_log.proto
-
-package org.tensorflow.proto.util.testlog;
-
-public interface RunConfigurationOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.RunConfiguration)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <code>repeated string argument = 1;</code>
-   */
-  java.util.List<java.lang.String>
-      getArgumentList();
-  /**
-   * <code>repeated string argument = 1;</code>
-   */
-  int getArgumentCount();
-  /**
-   * <code>repeated string argument = 1;</code>
-   */
-  java.lang.String getArgument(int index);
-  /**
-   * <code>repeated string argument = 1;</code>
-   */
-  com.google.protobuf.ByteString
-      getArgumentBytes(int index);
-
-  /**
-   * <pre>
-   * Environment variables used to run the test/benchmark.
-   * </pre>
-   *
-   * <code>map&lt;string, string&gt; env_vars = 2;</code>
-   */
-  int getEnvVarsCount();
-  /**
-   * <pre>
-   * Environment variables used to run the test/benchmark.
-   * </pre>
-   *
-   * <code>map&lt;string, string&gt; env_vars = 2;</code>
-   */
-  boolean containsEnvVars(
-      java.lang.String key);
-  /**
-   * Use {@link #getEnvVarsMap()} instead.
-   */
-  @java.lang.Deprecated
-  java.util.Map<java.lang.String, java.lang.String>
-  getEnvVars();
-  /**
-   * <pre>
-   * Environment variables used to run the test/benchmark.
-   * </pre>
-   *
-   * <code>map&lt;string, string&gt; env_vars = 2;</code>
-   */
-  java.util.Map<java.lang.String, java.lang.String>
-  getEnvVarsMap();
-  /**
-   * <pre>
-   * Environment variables used to run the test/benchmark.
-   * </pre>
-   *
-   * <code>map&lt;string, string&gt; env_vars = 2;</code>
-   */
-
-  java.lang.String getEnvVarsOrDefault(
-      java.lang.String key,
-      java.lang.String defaultValue);
-  /**
-   * <pre>
-   * Environment variables used to run the test/benchmark.
-   * </pre>
-   *
-   * <code>map&lt;string, string&gt; env_vars = 2;</code>
-   */
-
-  java.lang.String getEnvVarsOrThrow(
-      java.lang.String key);
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/TestLogProtos.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/TestLogProtos.java
deleted file mode 100644
index bb4033abe44..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/TestLogProtos.java
+++ /dev/null
@@ -1,288 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/util/test_log.proto
-
-package org.tensorflow.proto.util.testlog;
-
-public final class TestLogProtos {
-  private TestLogProtos() {}
-  public static void registerAllExtensions(
-      com.google.protobuf.ExtensionRegistryLite registry) {
-  }
-
-  public static void registerAllExtensions(
-      com.google.protobuf.ExtensionRegistry registry) {
-    registerAllExtensions(
-        (com.google.protobuf.ExtensionRegistryLite) registry);
-  }
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_EntryValue_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_EntryValue_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_MetricEntry_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_MetricEntry_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_BenchmarkEntry_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_BenchmarkEntry_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_BenchmarkEntry_ExtrasEntry_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_BenchmarkEntry_ExtrasEntry_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_BenchmarkEntries_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_BenchmarkEntries_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_BuildConfiguration_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_BuildConfiguration_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_CommitId_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_CommitId_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_CPUInfo_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_CPUInfo_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_CPUInfo_CacheSizeEntry_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_CPUInfo_CacheSizeEntry_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_MemoryInfo_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_MemoryInfo_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_GPUInfo_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_GPUInfo_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_PlatformInfo_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_PlatformInfo_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_AvailableDeviceInfo_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_AvailableDeviceInfo_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_MachineConfiguration_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_MachineConfiguration_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_RunConfiguration_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_RunConfiguration_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_RunConfiguration_EnvVarsEntry_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_RunConfiguration_EnvVarsEntry_fieldAccessorTable;
-  static final com.google.protobuf.Descriptors.Descriptor
-    internal_static_tensorflow_TestResults_descriptor;
-  static final 
-    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internal_static_tensorflow_TestResults_fieldAccessorTable;
-
-  public static com.google.protobuf.Descriptors.FileDescriptor
-      getDescriptor() {
-    return descriptor;
-  }
-  private static  com.google.protobuf.Descriptors.FileDescriptor
-      descriptor;
-  static {
-    java.lang.String[] descriptorData = {
-      "\n#tensorflow/core/util/test_log.proto\022\nt" +
-      "ensorflow\032\031google/protobuf/any.proto\032\036go" +
-      "ogle/protobuf/wrappers.proto\"D\n\nEntryVal" +
-      "ue\022\026\n\014double_value\030\001 \001(\001H\000\022\026\n\014string_val" +
-      "ue\030\002 \001(\tH\000B\006\n\004kind\"\214\001\n\013MetricEntry\022\014\n\004na" +
-      "me\030\001 \001(\t\022\r\n\005value\030\002 \001(\001\022/\n\tmin_value\030\003 \001" +
-      "(\0132\034.google.protobuf.DoubleValue\022/\n\tmax_" +
-      "value\030\004 \001(\0132\034.google.protobuf.DoubleValu" +
-      "e\"\217\002\n\016BenchmarkEntry\022\014\n\004name\030\001 \001(\t\022\r\n\005it" +
-      "ers\030\002 \001(\003\022\020\n\010cpu_time\030\003 \001(\001\022\021\n\twall_time" +
-      "\030\004 \001(\001\022\022\n\nthroughput\030\005 \001(\001\0226\n\006extras\030\006 \003" +
-      "(\0132&.tensorflow.BenchmarkEntry.ExtrasEnt" +
-      "ry\022(\n\007metrics\030\007 \003(\0132\027.tensorflow.MetricE" +
-      "ntry\032E\n\013ExtrasEntry\022\013\n\003key\030\001 \001(\t\022%\n\005valu" +
-      "e\030\002 \001(\0132\026.tensorflow.EntryValue:\0028\001\"=\n\020B" +
-      "enchmarkEntries\022)\n\005entry\030\001 \003(\0132\032.tensorf" +
-      "low.BenchmarkEntry\"B\n\022BuildConfiguration" +
-      "\022\014\n\004mode\030\001 \001(\t\022\020\n\010cc_flags\030\002 \003(\t\022\014\n\004opts" +
-      "\030\003 \003(\t\"f\n\010CommitId\022\024\n\nchangelist\030\001 \001(\003H\000" +
-      "\022\016\n\004hash\030\002 \001(\tH\000\022\020\n\010snapshot\030\003 \001(\t\022\032\n\022pe" +
-      "nding_changelist\030\004 \001(\003B\006\n\004kind\"\336\001\n\007CPUIn" +
-      "fo\022\021\n\tnum_cores\030\001 \001(\003\022\031\n\021num_cores_allow" +
-      "ed\030\002 \001(\003\022\023\n\013mhz_per_cpu\030\003 \001(\001\022\020\n\010cpu_inf" +
-      "o\030\004 \001(\t\022\024\n\014cpu_governor\030\005 \001(\t\0226\n\ncache_s" +
-      "ize\030\006 \003(\0132\".tensorflow.CPUInfo.CacheSize" +
-      "Entry\0320\n\016CacheSizeEntry\022\013\n\003key\030\001 \001(\t\022\r\n\005" +
-      "value\030\002 \001(\003:\0028\001\".\n\nMemoryInfo\022\r\n\005total\030\001" +
-      " \001(\003\022\021\n\tavailable\030\002 \001(\003\"6\n\007GPUInfo\022\r\n\005mo" +
-      "del\030\001 \001(\t\022\014\n\004uuid\030\002 \001(\t\022\016\n\006bus_id\030\003 \001(\t\"" +
-      "p\n\014PlatformInfo\022\014\n\004bits\030\001 \001(\t\022\017\n\007linkage" +
-      "\030\002 \001(\t\022\017\n\007machine\030\003 \001(\t\022\017\n\007release\030\004 \001(\t" +
-      "\022\016\n\006system\030\005 \001(\t\022\017\n\007version\030\006 \001(\t\"e\n\023Ava" +
-      "ilableDeviceInfo\022\014\n\004name\030\001 \001(\t\022\014\n\004type\030\002" +
-      " \001(\t\022\024\n\014memory_limit\030\003 \001(\003\022\034\n\024physical_d" +
-      "escription\030\004 \001(\t\"\263\002\n\024MachineConfiguratio" +
-      "n\022\020\n\010hostname\030\001 \001(\t\022\031\n\021serial_identifier" +
-      "\030\007 \001(\t\022/\n\rplatform_info\030\002 \001(\0132\030.tensorfl" +
-      "ow.PlatformInfo\022%\n\010cpu_info\030\003 \001(\0132\023.tens" +
-      "orflow.CPUInfo\022)\n\013device_info\030\004 \003(\0132\024.go" +
-      "ogle.protobuf.Any\022>\n\025available_device_in" +
-      "fo\030\005 \003(\0132\037.tensorflow.AvailableDeviceInf" +
-      "o\022+\n\013memory_info\030\006 \001(\0132\026.tensorflow.Memo" +
-      "ryInfo\"\221\001\n\020RunConfiguration\022\020\n\010argument\030" +
-      "\001 \003(\t\022;\n\010env_vars\030\002 \003(\0132).tensorflow.Run" +
-      "Configuration.EnvVarsEntry\032.\n\014EnvVarsEnt" +
-      "ry\022\013\n\003key\030\001 \001(\t\022\r\n\005value\030\002 \001(\t:\0028\001\"\320\004\n\013T" +
-      "estResults\022\016\n\006target\030\001 \001(\t\022-\n\007entries\030\002 " +
-      "\001(\0132\034.tensorflow.BenchmarkEntries\022;\n\023bui" +
-      "ld_configuration\030\003 \001(\0132\036.tensorflow.Buil" +
-      "dConfiguration\022\'\n\tcommit_id\030\004 \001(\0132\024.tens" +
-      "orflow.CommitId\022\022\n\nstart_time\030\005 \001(\003\022\020\n\010r" +
-      "un_time\030\006 \001(\001\022?\n\025machine_configuration\030\007" +
-      " \001(\0132 .tensorflow.MachineConfiguration\0227" +
-      "\n\021run_configuration\030\010 \001(\0132\034.tensorflow.R" +
-      "unConfiguration\022\014\n\004name\030\t \001(\t\022=\n\016benchma" +
-      "rk_type\030\n \001(\0162%.tensorflow.TestResults.B" +
-      "enchmarkType\022\020\n\010run_mode\030\013 \001(\t\022\022\n\ntf_ver" +
-      "sion\030\014 \001(\t\"\210\001\n\rBenchmarkType\022\013\n\007UNKNOWN\020" +
-      "\000\022\026\n\022CPP_MICROBENCHMARK\020\001\022\024\n\020PYTHON_BENC" +
-      "HMARK\020\002\022\025\n\021ANDROID_BENCHMARK\020\003\022\022\n\016EDGE_B" +
-      "ENCHMARK\020\004\022\021\n\rIOS_BENCHMARK\020\005B7\n!org.ten" +
-      "sorflow.proto.util.testlogB\rTestLogProto" +
-      "sP\001\370\001\001b\006proto3"
-    };
-    descriptor = com.google.protobuf.Descriptors.FileDescriptor
-      .internalBuildGeneratedFileFrom(descriptorData,
-        new com.google.protobuf.Descriptors.FileDescriptor[] {
-          com.google.protobuf.AnyProto.getDescriptor(),
-          com.google.protobuf.WrappersProto.getDescriptor(),
-        });
-    internal_static_tensorflow_EntryValue_descriptor =
-      getDescriptor().getMessageTypes().get(0);
-    internal_static_tensorflow_EntryValue_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_EntryValue_descriptor,
-        new java.lang.String[] { "DoubleValue", "StringValue", "Kind", });
-    internal_static_tensorflow_MetricEntry_descriptor =
-      getDescriptor().getMessageTypes().get(1);
-    internal_static_tensorflow_MetricEntry_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_MetricEntry_descriptor,
-        new java.lang.String[] { "Name", "Value", "MinValue", "MaxValue", });
-    internal_static_tensorflow_BenchmarkEntry_descriptor =
-      getDescriptor().getMessageTypes().get(2);
-    internal_static_tensorflow_BenchmarkEntry_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_BenchmarkEntry_descriptor,
-        new java.lang.String[] { "Name", "Iters", "CpuTime", "WallTime", "Throughput", "Extras", "Metrics", });
-    internal_static_tensorflow_BenchmarkEntry_ExtrasEntry_descriptor =
-      internal_static_tensorflow_BenchmarkEntry_descriptor.getNestedTypes().get(0);
-    internal_static_tensorflow_BenchmarkEntry_ExtrasEntry_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_BenchmarkEntry_ExtrasEntry_descriptor,
-        new java.lang.String[] { "Key", "Value", });
-    internal_static_tensorflow_BenchmarkEntries_descriptor =
-      getDescriptor().getMessageTypes().get(3);
-    internal_static_tensorflow_BenchmarkEntries_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_BenchmarkEntries_descriptor,
-        new java.lang.String[] { "Entry", });
-    internal_static_tensorflow_BuildConfiguration_descriptor =
-      getDescriptor().getMessageTypes().get(4);
-    internal_static_tensorflow_BuildConfiguration_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_BuildConfiguration_descriptor,
-        new java.lang.String[] { "Mode", "CcFlags", "Opts", });
-    internal_static_tensorflow_CommitId_descriptor =
-      getDescriptor().getMessageTypes().get(5);
-    internal_static_tensorflow_CommitId_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_CommitId_descriptor,
-        new java.lang.String[] { "Changelist", "Hash", "Snapshot", "PendingChangelist", "Kind", });
-    internal_static_tensorflow_CPUInfo_descriptor =
-      getDescriptor().getMessageTypes().get(6);
-    internal_static_tensorflow_CPUInfo_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_CPUInfo_descriptor,
-        new java.lang.String[] { "NumCores", "NumCoresAllowed", "MhzPerCpu", "CpuInfo", "CpuGovernor", "CacheSize", });
-    internal_static_tensorflow_CPUInfo_CacheSizeEntry_descriptor =
-      internal_static_tensorflow_CPUInfo_descriptor.getNestedTypes().get(0);
-    internal_static_tensorflow_CPUInfo_CacheSizeEntry_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_CPUInfo_CacheSizeEntry_descriptor,
-        new java.lang.String[] { "Key", "Value", });
-    internal_static_tensorflow_MemoryInfo_descriptor =
-      getDescriptor().getMessageTypes().get(7);
-    internal_static_tensorflow_MemoryInfo_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_MemoryInfo_descriptor,
-        new java.lang.String[] { "Total", "Available", });
-    internal_static_tensorflow_GPUInfo_descriptor =
-      getDescriptor().getMessageTypes().get(8);
-    internal_static_tensorflow_GPUInfo_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_GPUInfo_descriptor,
-        new java.lang.String[] { "Model", "Uuid", "BusId", });
-    internal_static_tensorflow_PlatformInfo_descriptor =
-      getDescriptor().getMessageTypes().get(9);
-    internal_static_tensorflow_PlatformInfo_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_PlatformInfo_descriptor,
-        new java.lang.String[] { "Bits", "Linkage", "Machine", "Release", "System", "Version", });
-    internal_static_tensorflow_AvailableDeviceInfo_descriptor =
-      getDescriptor().getMessageTypes().get(10);
-    internal_static_tensorflow_AvailableDeviceInfo_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_AvailableDeviceInfo_descriptor,
-        new java.lang.String[] { "Name", "Type", "MemoryLimit", "PhysicalDescription", });
-    internal_static_tensorflow_MachineConfiguration_descriptor =
-      getDescriptor().getMessageTypes().get(11);
-    internal_static_tensorflow_MachineConfiguration_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_MachineConfiguration_descriptor,
-        new java.lang.String[] { "Hostname", "SerialIdentifier", "PlatformInfo", "CpuInfo", "DeviceInfo", "AvailableDeviceInfo", "MemoryInfo", });
-    internal_static_tensorflow_RunConfiguration_descriptor =
-      getDescriptor().getMessageTypes().get(12);
-    internal_static_tensorflow_RunConfiguration_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_RunConfiguration_descriptor,
-        new java.lang.String[] { "Argument", "EnvVars", });
-    internal_static_tensorflow_RunConfiguration_EnvVarsEntry_descriptor =
-      internal_static_tensorflow_RunConfiguration_descriptor.getNestedTypes().get(0);
-    internal_static_tensorflow_RunConfiguration_EnvVarsEntry_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_RunConfiguration_EnvVarsEntry_descriptor,
-        new java.lang.String[] { "Key", "Value", });
-    internal_static_tensorflow_TestResults_descriptor =
-      getDescriptor().getMessageTypes().get(13);
-    internal_static_tensorflow_TestResults_fieldAccessorTable = new
-      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
-        internal_static_tensorflow_TestResults_descriptor,
-        new java.lang.String[] { "Target", "Entries", "BuildConfiguration", "CommitId", "StartTime", "RunTime", "MachineConfiguration", "RunConfiguration", "Name", "BenchmarkType", "RunMode", "TfVersion", });
-    com.google.protobuf.AnyProto.getDescriptor();
-    com.google.protobuf.WrappersProto.getDescriptor();
-  }
-
-  // @@protoc_insertion_point(outer_class_scope)
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/TestResults.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/TestResults.java
deleted file mode 100644
index 15ae63f6d31..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/TestResults.java
+++ /dev/null
@@ -1,2624 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/util/test_log.proto
-
-package org.tensorflow.proto.util.testlog;
-
-/**
- * <pre>
- * The output of one benchmark / test run.  Each run contains a list of
- * tests or benchmarks, stored as BenchmarkEntry messages.
- * This message should be emitted by the reporter (which runs the
- * test / BM in a subprocess and then reads the emitted BenchmarkEntry messages;
- * usually from a serialized json file, finally collecting them along
- * with additional information about the test run.
- * </pre>
- *
- * Protobuf type {@code tensorflow.TestResults}
- */
-public  final class TestResults extends
-    com.google.protobuf.GeneratedMessageV3 implements
-    // @@protoc_insertion_point(message_implements:tensorflow.TestResults)
-    TestResultsOrBuilder {
-private static final long serialVersionUID = 0L;
-  // Use TestResults.newBuilder() to construct.
-  private TestResults(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
-    super(builder);
-  }
-  private TestResults() {
-    target_ = "";
-    name_ = "";
-    benchmarkType_ = 0;
-    runMode_ = "";
-    tfVersion_ = "";
-  }
-
-  @java.lang.Override
-  @SuppressWarnings({"unused"})
-  protected java.lang.Object newInstance(
-      UnusedPrivateParameter unused) {
-    return new TestResults();
-  }
-
-  @java.lang.Override
-  public final com.google.protobuf.UnknownFieldSet
-  getUnknownFields() {
-    return this.unknownFields;
-  }
-  private TestResults(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    this();
-    if (extensionRegistry == null) {
-      throw new java.lang.NullPointerException();
-    }
-    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
-        com.google.protobuf.UnknownFieldSet.newBuilder();
-    try {
-      boolean done = false;
-      while (!done) {
-        int tag = input.readTag();
-        switch (tag) {
-          case 0:
-            done = true;
-            break;
-          case 10: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            target_ = s;
-            break;
-          }
-          case 18: {
-            org.tensorflow.proto.util.testlog.BenchmarkEntries.Builder subBuilder = null;
-            if (entries_ != null) {
-              subBuilder = entries_.toBuilder();
-            }
-            entries_ = input.readMessage(org.tensorflow.proto.util.testlog.BenchmarkEntries.parser(), extensionRegistry);
-            if (subBuilder != null) {
-              subBuilder.mergeFrom(entries_);
-              entries_ = subBuilder.buildPartial();
-            }
-
-            break;
-          }
-          case 26: {
-            org.tensorflow.proto.util.testlog.BuildConfiguration.Builder subBuilder = null;
-            if (buildConfiguration_ != null) {
-              subBuilder = buildConfiguration_.toBuilder();
-            }
-            buildConfiguration_ = input.readMessage(org.tensorflow.proto.util.testlog.BuildConfiguration.parser(), extensionRegistry);
-            if (subBuilder != null) {
-              subBuilder.mergeFrom(buildConfiguration_);
-              buildConfiguration_ = subBuilder.buildPartial();
-            }
-
-            break;
-          }
-          case 34: {
-            org.tensorflow.proto.util.testlog.CommitId.Builder subBuilder = null;
-            if (commitId_ != null) {
-              subBuilder = commitId_.toBuilder();
-            }
-            commitId_ = input.readMessage(org.tensorflow.proto.util.testlog.CommitId.parser(), extensionRegistry);
-            if (subBuilder != null) {
-              subBuilder.mergeFrom(commitId_);
-              commitId_ = subBuilder.buildPartial();
-            }
-
-            break;
-          }
-          case 40: {
-
-            startTime_ = input.readInt64();
-            break;
-          }
-          case 49: {
-
-            runTime_ = input.readDouble();
-            break;
-          }
-          case 58: {
-            org.tensorflow.proto.util.testlog.MachineConfiguration.Builder subBuilder = null;
-            if (machineConfiguration_ != null) {
-              subBuilder = machineConfiguration_.toBuilder();
-            }
-            machineConfiguration_ = input.readMessage(org.tensorflow.proto.util.testlog.MachineConfiguration.parser(), extensionRegistry);
-            if (subBuilder != null) {
-              subBuilder.mergeFrom(machineConfiguration_);
-              machineConfiguration_ = subBuilder.buildPartial();
-            }
-
-            break;
-          }
-          case 66: {
-            org.tensorflow.proto.util.testlog.RunConfiguration.Builder subBuilder = null;
-            if (runConfiguration_ != null) {
-              subBuilder = runConfiguration_.toBuilder();
-            }
-            runConfiguration_ = input.readMessage(org.tensorflow.proto.util.testlog.RunConfiguration.parser(), extensionRegistry);
-            if (subBuilder != null) {
-              subBuilder.mergeFrom(runConfiguration_);
-              runConfiguration_ = subBuilder.buildPartial();
-            }
-
-            break;
-          }
-          case 74: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            name_ = s;
-            break;
-          }
-          case 80: {
-            int rawValue = input.readEnum();
-
-            benchmarkType_ = rawValue;
-            break;
-          }
-          case 90: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            runMode_ = s;
-            break;
-          }
-          case 98: {
-            java.lang.String s = input.readStringRequireUtf8();
-
-            tfVersion_ = s;
-            break;
-          }
-          default: {
-            if (!parseUnknownField(
-                input, unknownFields, extensionRegistry, tag)) {
-              done = true;
-            }
-            break;
-          }
-        }
-      }
-    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-      throw e.setUnfinishedMessage(this);
-    } catch (java.io.IOException e) {
-      throw new com.google.protobuf.InvalidProtocolBufferException(
-          e).setUnfinishedMessage(this);
-    } finally {
-      this.unknownFields = unknownFields.build();
-      makeExtensionsImmutable();
-    }
-  }
-  public static final com.google.protobuf.Descriptors.Descriptor
-      getDescriptor() {
-    return org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_TestResults_descriptor;
-  }
-
-  @java.lang.Override
-  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-      internalGetFieldAccessorTable() {
-    return org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_TestResults_fieldAccessorTable
-        .ensureFieldAccessorsInitialized(
-            org.tensorflow.proto.util.testlog.TestResults.class, org.tensorflow.proto.util.testlog.TestResults.Builder.class);
-  }
-
-  /**
-   * <pre>
-   * The type of benchmark.
-   * </pre>
-   *
-   * Protobuf enum {@code tensorflow.TestResults.BenchmarkType}
-   */
-  public enum BenchmarkType
-      implements com.google.protobuf.ProtocolMessageEnum {
-    /**
-     * <pre>
-     * Fallback for protos written before Type was introduced.
-     * </pre>
-     *
-     * <code>UNKNOWN = 0;</code>
-     */
-    UNKNOWN(0),
-    /**
-     * <code>CPP_MICROBENCHMARK = 1;</code>
-     */
-    CPP_MICROBENCHMARK(1),
-    /**
-     * <code>PYTHON_BENCHMARK = 2;</code>
-     */
-    PYTHON_BENCHMARK(2),
-    /**
-     * <code>ANDROID_BENCHMARK = 3;</code>
-     */
-    ANDROID_BENCHMARK(3),
-    /**
-     * <code>EDGE_BENCHMARK = 4;</code>
-     */
-    EDGE_BENCHMARK(4),
-    /**
-     * <code>IOS_BENCHMARK = 5;</code>
-     */
-    IOS_BENCHMARK(5),
-    UNRECOGNIZED(-1),
-    ;
-
-    /**
-     * <pre>
-     * Fallback for protos written before Type was introduced.
-     * </pre>
-     *
-     * <code>UNKNOWN = 0;</code>
-     */
-    public static final int UNKNOWN_VALUE = 0;
-    /**
-     * <code>CPP_MICROBENCHMARK = 1;</code>
-     */
-    public static final int CPP_MICROBENCHMARK_VALUE = 1;
-    /**
-     * <code>PYTHON_BENCHMARK = 2;</code>
-     */
-    public static final int PYTHON_BENCHMARK_VALUE = 2;
-    /**
-     * <code>ANDROID_BENCHMARK = 3;</code>
-     */
-    public static final int ANDROID_BENCHMARK_VALUE = 3;
-    /**
-     * <code>EDGE_BENCHMARK = 4;</code>
-     */
-    public static final int EDGE_BENCHMARK_VALUE = 4;
-    /**
-     * <code>IOS_BENCHMARK = 5;</code>
-     */
-    public static final int IOS_BENCHMARK_VALUE = 5;
-
-
-    public final int getNumber() {
-      if (this == UNRECOGNIZED) {
-        throw new java.lang.IllegalArgumentException(
-            "Can't get the number of an unknown enum value.");
-      }
-      return value;
-    }
-
-    /**
-     * @deprecated Use {@link #forNumber(int)} instead.
-     */
-    @java.lang.Deprecated
-    public static BenchmarkType valueOf(int value) {
-      return forNumber(value);
-    }
-
-    public static BenchmarkType forNumber(int value) {
-      switch (value) {
-        case 0: return UNKNOWN;
-        case 1: return CPP_MICROBENCHMARK;
-        case 2: return PYTHON_BENCHMARK;
-        case 3: return ANDROID_BENCHMARK;
-        case 4: return EDGE_BENCHMARK;
-        case 5: return IOS_BENCHMARK;
-        default: return null;
-      }
-    }
-
-    public static com.google.protobuf.Internal.EnumLiteMap<BenchmarkType>
-        internalGetValueMap() {
-      return internalValueMap;
-    }
-    private static final com.google.protobuf.Internal.EnumLiteMap<
-        BenchmarkType> internalValueMap =
-          new com.google.protobuf.Internal.EnumLiteMap<BenchmarkType>() {
-            public BenchmarkType findValueByNumber(int number) {
-              return BenchmarkType.forNumber(number);
-            }
-          };
-
-    public final com.google.protobuf.Descriptors.EnumValueDescriptor
-        getValueDescriptor() {
-      return getDescriptor().getValues().get(ordinal());
-    }
-    public final com.google.protobuf.Descriptors.EnumDescriptor
-        getDescriptorForType() {
-      return getDescriptor();
-    }
-    public static final com.google.protobuf.Descriptors.EnumDescriptor
-        getDescriptor() {
-      return org.tensorflow.proto.util.testlog.TestResults.getDescriptor().getEnumTypes().get(0);
-    }
-
-    private static final BenchmarkType[] VALUES = values();
-
-    public static BenchmarkType valueOf(
-        com.google.protobuf.Descriptors.EnumValueDescriptor desc) {
-      if (desc.getType() != getDescriptor()) {
-        throw new java.lang.IllegalArgumentException(
-          "EnumValueDescriptor is not for this type.");
-      }
-      if (desc.getIndex() == -1) {
-        return UNRECOGNIZED;
-      }
-      return VALUES[desc.getIndex()];
-    }
-
-    private final int value;
-
-    private BenchmarkType(int value) {
-      this.value = value;
-    }
-
-    // @@protoc_insertion_point(enum_scope:tensorflow.TestResults.BenchmarkType)
-  }
-
-  public static final int TARGET_FIELD_NUMBER = 1;
-  private volatile java.lang.Object target_;
-  /**
-   * <pre>
-   * The target of the run, e.g.:
-   *  //tensorflow/core:kernels_adjust_contrast_op_benchmark_test
-   * </pre>
-   *
-   * <code>string target = 1;</code>
-   */
-  public java.lang.String getTarget() {
-    java.lang.Object ref = target_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      target_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * The target of the run, e.g.:
-   *  //tensorflow/core:kernels_adjust_contrast_op_benchmark_test
-   * </pre>
-   *
-   * <code>string target = 1;</code>
-   */
-  public com.google.protobuf.ByteString
-      getTargetBytes() {
-    java.lang.Object ref = target_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      target_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int ENTRIES_FIELD_NUMBER = 2;
-  private org.tensorflow.proto.util.testlog.BenchmarkEntries entries_;
-  /**
-   * <pre>
-   * The list of tests or benchmarks in this run.
-   * </pre>
-   *
-   * <code>.tensorflow.BenchmarkEntries entries = 2;</code>
-   */
-  public boolean hasEntries() {
-    return entries_ != null;
-  }
-  /**
-   * <pre>
-   * The list of tests or benchmarks in this run.
-   * </pre>
-   *
-   * <code>.tensorflow.BenchmarkEntries entries = 2;</code>
-   */
-  public org.tensorflow.proto.util.testlog.BenchmarkEntries getEntries() {
-    return entries_ == null ? org.tensorflow.proto.util.testlog.BenchmarkEntries.getDefaultInstance() : entries_;
-  }
-  /**
-   * <pre>
-   * The list of tests or benchmarks in this run.
-   * </pre>
-   *
-   * <code>.tensorflow.BenchmarkEntries entries = 2;</code>
-   */
-  public org.tensorflow.proto.util.testlog.BenchmarkEntriesOrBuilder getEntriesOrBuilder() {
-    return getEntries();
-  }
-
-  public static final int BUILD_CONFIGURATION_FIELD_NUMBER = 3;
-  private org.tensorflow.proto.util.testlog.BuildConfiguration buildConfiguration_;
-  /**
-   * <pre>
-   * The configuration of the build (compiled opt? with cuda? any copts?)
-   * </pre>
-   *
-   * <code>.tensorflow.BuildConfiguration build_configuration = 3;</code>
-   */
-  public boolean hasBuildConfiguration() {
-    return buildConfiguration_ != null;
-  }
-  /**
-   * <pre>
-   * The configuration of the build (compiled opt? with cuda? any copts?)
-   * </pre>
-   *
-   * <code>.tensorflow.BuildConfiguration build_configuration = 3;</code>
-   */
-  public org.tensorflow.proto.util.testlog.BuildConfiguration getBuildConfiguration() {
-    return buildConfiguration_ == null ? org.tensorflow.proto.util.testlog.BuildConfiguration.getDefaultInstance() : buildConfiguration_;
-  }
-  /**
-   * <pre>
-   * The configuration of the build (compiled opt? with cuda? any copts?)
-   * </pre>
-   *
-   * <code>.tensorflow.BuildConfiguration build_configuration = 3;</code>
-   */
-  public org.tensorflow.proto.util.testlog.BuildConfigurationOrBuilder getBuildConfigurationOrBuilder() {
-    return getBuildConfiguration();
-  }
-
-  public static final int COMMIT_ID_FIELD_NUMBER = 4;
-  private org.tensorflow.proto.util.testlog.CommitId commitId_;
-  /**
-   * <pre>
-   * The commit id (git hash or changelist)
-   * </pre>
-   *
-   * <code>.tensorflow.CommitId commit_id = 4;</code>
-   */
-  public boolean hasCommitId() {
-    return commitId_ != null;
-  }
-  /**
-   * <pre>
-   * The commit id (git hash or changelist)
-   * </pre>
-   *
-   * <code>.tensorflow.CommitId commit_id = 4;</code>
-   */
-  public org.tensorflow.proto.util.testlog.CommitId getCommitId() {
-    return commitId_ == null ? org.tensorflow.proto.util.testlog.CommitId.getDefaultInstance() : commitId_;
-  }
-  /**
-   * <pre>
-   * The commit id (git hash or changelist)
-   * </pre>
-   *
-   * <code>.tensorflow.CommitId commit_id = 4;</code>
-   */
-  public org.tensorflow.proto.util.testlog.CommitIdOrBuilder getCommitIdOrBuilder() {
-    return getCommitId();
-  }
-
-  public static final int START_TIME_FIELD_NUMBER = 5;
-  private long startTime_;
-  /**
-   * <pre>
-   * The time the run started (in seconds of UTC time since Unix epoch)
-   * </pre>
-   *
-   * <code>int64 start_time = 5;</code>
-   */
-  public long getStartTime() {
-    return startTime_;
-  }
-
-  public static final int RUN_TIME_FIELD_NUMBER = 6;
-  private double runTime_;
-  /**
-   * <pre>
-   * The amount of time the total run took (wall time in seconds)
-   * </pre>
-   *
-   * <code>double run_time = 6;</code>
-   */
-  public double getRunTime() {
-    return runTime_;
-  }
-
-  public static final int MACHINE_CONFIGURATION_FIELD_NUMBER = 7;
-  private org.tensorflow.proto.util.testlog.MachineConfiguration machineConfiguration_;
-  /**
-   * <pre>
-   * Machine-specific parameters (Platform and CPU info)
-   * </pre>
-   *
-   * <code>.tensorflow.MachineConfiguration machine_configuration = 7;</code>
-   */
-  public boolean hasMachineConfiguration() {
-    return machineConfiguration_ != null;
-  }
-  /**
-   * <pre>
-   * Machine-specific parameters (Platform and CPU info)
-   * </pre>
-   *
-   * <code>.tensorflow.MachineConfiguration machine_configuration = 7;</code>
-   */
-  public org.tensorflow.proto.util.testlog.MachineConfiguration getMachineConfiguration() {
-    return machineConfiguration_ == null ? org.tensorflow.proto.util.testlog.MachineConfiguration.getDefaultInstance() : machineConfiguration_;
-  }
-  /**
-   * <pre>
-   * Machine-specific parameters (Platform and CPU info)
-   * </pre>
-   *
-   * <code>.tensorflow.MachineConfiguration machine_configuration = 7;</code>
-   */
-  public org.tensorflow.proto.util.testlog.MachineConfigurationOrBuilder getMachineConfigurationOrBuilder() {
-    return getMachineConfiguration();
-  }
-
-  public static final int RUN_CONFIGURATION_FIELD_NUMBER = 8;
-  private org.tensorflow.proto.util.testlog.RunConfiguration runConfiguration_;
-  /**
-   * <pre>
-   * Run-specific parameters (arguments, etc)
-   * </pre>
-   *
-   * <code>.tensorflow.RunConfiguration run_configuration = 8;</code>
-   */
-  public boolean hasRunConfiguration() {
-    return runConfiguration_ != null;
-  }
-  /**
-   * <pre>
-   * Run-specific parameters (arguments, etc)
-   * </pre>
-   *
-   * <code>.tensorflow.RunConfiguration run_configuration = 8;</code>
-   */
-  public org.tensorflow.proto.util.testlog.RunConfiguration getRunConfiguration() {
-    return runConfiguration_ == null ? org.tensorflow.proto.util.testlog.RunConfiguration.getDefaultInstance() : runConfiguration_;
-  }
-  /**
-   * <pre>
-   * Run-specific parameters (arguments, etc)
-   * </pre>
-   *
-   * <code>.tensorflow.RunConfiguration run_configuration = 8;</code>
-   */
-  public org.tensorflow.proto.util.testlog.RunConfigurationOrBuilder getRunConfigurationOrBuilder() {
-    return getRunConfiguration();
-  }
-
-  public static final int NAME_FIELD_NUMBER = 9;
-  private volatile java.lang.Object name_;
-  /**
-   * <pre>
-   * Benchmark target identifier.
-   * </pre>
-   *
-   * <code>string name = 9;</code>
-   */
-  public java.lang.String getName() {
-    java.lang.Object ref = name_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      name_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * Benchmark target identifier.
-   * </pre>
-   *
-   * <code>string name = 9;</code>
-   */
-  public com.google.protobuf.ByteString
-      getNameBytes() {
-    java.lang.Object ref = name_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      name_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int BENCHMARK_TYPE_FIELD_NUMBER = 10;
-  private int benchmarkType_;
-  /**
-   * <code>.tensorflow.TestResults.BenchmarkType benchmark_type = 10;</code>
-   */
-  public int getBenchmarkTypeValue() {
-    return benchmarkType_;
-  }
-  /**
-   * <code>.tensorflow.TestResults.BenchmarkType benchmark_type = 10;</code>
-   */
-  public org.tensorflow.proto.util.testlog.TestResults.BenchmarkType getBenchmarkType() {
-    @SuppressWarnings("deprecation")
-    org.tensorflow.proto.util.testlog.TestResults.BenchmarkType result = org.tensorflow.proto.util.testlog.TestResults.BenchmarkType.valueOf(benchmarkType_);
-    return result == null ? org.tensorflow.proto.util.testlog.TestResults.BenchmarkType.UNRECOGNIZED : result;
-  }
-
-  public static final int RUN_MODE_FIELD_NUMBER = 11;
-  private volatile java.lang.Object runMode_;
-  /**
-   * <pre>
-   * Used for differentiating between continuous and debug builds.
-   * Must be one of:
-   * * cbuild: results from continuous build.
-   * * presubmit: results from oneshot requests.
-   * * culprit: results from culprit finder rerun.
-   * </pre>
-   *
-   * <code>string run_mode = 11;</code>
-   */
-  public java.lang.String getRunMode() {
-    java.lang.Object ref = runMode_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      runMode_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * Used for differentiating between continuous and debug builds.
-   * Must be one of:
-   * * cbuild: results from continuous build.
-   * * presubmit: results from oneshot requests.
-   * * culprit: results from culprit finder rerun.
-   * </pre>
-   *
-   * <code>string run_mode = 11;</code>
-   */
-  public com.google.protobuf.ByteString
-      getRunModeBytes() {
-    java.lang.Object ref = runMode_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      runMode_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  public static final int TF_VERSION_FIELD_NUMBER = 12;
-  private volatile java.lang.Object tfVersion_;
-  /**
-   * <pre>
-   * TensorFlow version this benchmark runs against.
-   * This can be either set to full version or just the major version.
-   * </pre>
-   *
-   * <code>string tf_version = 12;</code>
-   */
-  public java.lang.String getTfVersion() {
-    java.lang.Object ref = tfVersion_;
-    if (ref instanceof java.lang.String) {
-      return (java.lang.String) ref;
-    } else {
-      com.google.protobuf.ByteString bs = 
-          (com.google.protobuf.ByteString) ref;
-      java.lang.String s = bs.toStringUtf8();
-      tfVersion_ = s;
-      return s;
-    }
-  }
-  /**
-   * <pre>
-   * TensorFlow version this benchmark runs against.
-   * This can be either set to full version or just the major version.
-   * </pre>
-   *
-   * <code>string tf_version = 12;</code>
-   */
-  public com.google.protobuf.ByteString
-      getTfVersionBytes() {
-    java.lang.Object ref = tfVersion_;
-    if (ref instanceof java.lang.String) {
-      com.google.protobuf.ByteString b = 
-          com.google.protobuf.ByteString.copyFromUtf8(
-              (java.lang.String) ref);
-      tfVersion_ = b;
-      return b;
-    } else {
-      return (com.google.protobuf.ByteString) ref;
-    }
-  }
-
-  private byte memoizedIsInitialized = -1;
-  @java.lang.Override
-  public final boolean isInitialized() {
-    byte isInitialized = memoizedIsInitialized;
-    if (isInitialized == 1) return true;
-    if (isInitialized == 0) return false;
-
-    memoizedIsInitialized = 1;
-    return true;
-  }
-
-  @java.lang.Override
-  public void writeTo(com.google.protobuf.CodedOutputStream output)
-                      throws java.io.IOException {
-    if (!getTargetBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 1, target_);
-    }
-    if (entries_ != null) {
-      output.writeMessage(2, getEntries());
-    }
-    if (buildConfiguration_ != null) {
-      output.writeMessage(3, getBuildConfiguration());
-    }
-    if (commitId_ != null) {
-      output.writeMessage(4, getCommitId());
-    }
-    if (startTime_ != 0L) {
-      output.writeInt64(5, startTime_);
-    }
-    if (runTime_ != 0D) {
-      output.writeDouble(6, runTime_);
-    }
-    if (machineConfiguration_ != null) {
-      output.writeMessage(7, getMachineConfiguration());
-    }
-    if (runConfiguration_ != null) {
-      output.writeMessage(8, getRunConfiguration());
-    }
-    if (!getNameBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 9, name_);
-    }
-    if (benchmarkType_ != org.tensorflow.proto.util.testlog.TestResults.BenchmarkType.UNKNOWN.getNumber()) {
-      output.writeEnum(10, benchmarkType_);
-    }
-    if (!getRunModeBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 11, runMode_);
-    }
-    if (!getTfVersionBytes().isEmpty()) {
-      com.google.protobuf.GeneratedMessageV3.writeString(output, 12, tfVersion_);
-    }
-    unknownFields.writeTo(output);
-  }
-
-  @java.lang.Override
-  public int getSerializedSize() {
-    int size = memoizedSize;
-    if (size != -1) return size;
-
-    size = 0;
-    if (!getTargetBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, target_);
-    }
-    if (entries_ != null) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(2, getEntries());
-    }
-    if (buildConfiguration_ != null) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(3, getBuildConfiguration());
-    }
-    if (commitId_ != null) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(4, getCommitId());
-    }
-    if (startTime_ != 0L) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeInt64Size(5, startTime_);
-    }
-    if (runTime_ != 0D) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeDoubleSize(6, runTime_);
-    }
-    if (machineConfiguration_ != null) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(7, getMachineConfiguration());
-    }
-    if (runConfiguration_ != null) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeMessageSize(8, getRunConfiguration());
-    }
-    if (!getNameBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(9, name_);
-    }
-    if (benchmarkType_ != org.tensorflow.proto.util.testlog.TestResults.BenchmarkType.UNKNOWN.getNumber()) {
-      size += com.google.protobuf.CodedOutputStream
-        .computeEnumSize(10, benchmarkType_);
-    }
-    if (!getRunModeBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(11, runMode_);
-    }
-    if (!getTfVersionBytes().isEmpty()) {
-      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(12, tfVersion_);
-    }
-    size += unknownFields.getSerializedSize();
-    memoizedSize = size;
-    return size;
-  }
-
-  @java.lang.Override
-  public boolean equals(final java.lang.Object obj) {
-    if (obj == this) {
-     return true;
-    }
-    if (!(obj instanceof org.tensorflow.proto.util.testlog.TestResults)) {
-      return super.equals(obj);
-    }
-    org.tensorflow.proto.util.testlog.TestResults other = (org.tensorflow.proto.util.testlog.TestResults) obj;
-
-    if (!getTarget()
-        .equals(other.getTarget())) return false;
-    if (hasEntries() != other.hasEntries()) return false;
-    if (hasEntries()) {
-      if (!getEntries()
-          .equals(other.getEntries())) return false;
-    }
-    if (hasBuildConfiguration() != other.hasBuildConfiguration()) return false;
-    if (hasBuildConfiguration()) {
-      if (!getBuildConfiguration()
-          .equals(other.getBuildConfiguration())) return false;
-    }
-    if (hasCommitId() != other.hasCommitId()) return false;
-    if (hasCommitId()) {
-      if (!getCommitId()
-          .equals(other.getCommitId())) return false;
-    }
-    if (getStartTime()
-        != other.getStartTime()) return false;
-    if (java.lang.Double.doubleToLongBits(getRunTime())
-        != java.lang.Double.doubleToLongBits(
-            other.getRunTime())) return false;
-    if (hasMachineConfiguration() != other.hasMachineConfiguration()) return false;
-    if (hasMachineConfiguration()) {
-      if (!getMachineConfiguration()
-          .equals(other.getMachineConfiguration())) return false;
-    }
-    if (hasRunConfiguration() != other.hasRunConfiguration()) return false;
-    if (hasRunConfiguration()) {
-      if (!getRunConfiguration()
-          .equals(other.getRunConfiguration())) return false;
-    }
-    if (!getName()
-        .equals(other.getName())) return false;
-    if (benchmarkType_ != other.benchmarkType_) return false;
-    if (!getRunMode()
-        .equals(other.getRunMode())) return false;
-    if (!getTfVersion()
-        .equals(other.getTfVersion())) return false;
-    if (!unknownFields.equals(other.unknownFields)) return false;
-    return true;
-  }
-
-  @java.lang.Override
-  public int hashCode() {
-    if (memoizedHashCode != 0) {
-      return memoizedHashCode;
-    }
-    int hash = 41;
-    hash = (19 * hash) + getDescriptor().hashCode();
-    hash = (37 * hash) + TARGET_FIELD_NUMBER;
-    hash = (53 * hash) + getTarget().hashCode();
-    if (hasEntries()) {
-      hash = (37 * hash) + ENTRIES_FIELD_NUMBER;
-      hash = (53 * hash) + getEntries().hashCode();
-    }
-    if (hasBuildConfiguration()) {
-      hash = (37 * hash) + BUILD_CONFIGURATION_FIELD_NUMBER;
-      hash = (53 * hash) + getBuildConfiguration().hashCode();
-    }
-    if (hasCommitId()) {
-      hash = (37 * hash) + COMMIT_ID_FIELD_NUMBER;
-      hash = (53 * hash) + getCommitId().hashCode();
-    }
-    hash = (37 * hash) + START_TIME_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-        getStartTime());
-    hash = (37 * hash) + RUN_TIME_FIELD_NUMBER;
-    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
-        java.lang.Double.doubleToLongBits(getRunTime()));
-    if (hasMachineConfiguration()) {
-      hash = (37 * hash) + MACHINE_CONFIGURATION_FIELD_NUMBER;
-      hash = (53 * hash) + getMachineConfiguration().hashCode();
-    }
-    if (hasRunConfiguration()) {
-      hash = (37 * hash) + RUN_CONFIGURATION_FIELD_NUMBER;
-      hash = (53 * hash) + getRunConfiguration().hashCode();
-    }
-    hash = (37 * hash) + NAME_FIELD_NUMBER;
-    hash = (53 * hash) + getName().hashCode();
-    hash = (37 * hash) + BENCHMARK_TYPE_FIELD_NUMBER;
-    hash = (53 * hash) + benchmarkType_;
-    hash = (37 * hash) + RUN_MODE_FIELD_NUMBER;
-    hash = (53 * hash) + getRunMode().hashCode();
-    hash = (37 * hash) + TF_VERSION_FIELD_NUMBER;
-    hash = (53 * hash) + getTfVersion().hashCode();
-    hash = (29 * hash) + unknownFields.hashCode();
-    memoizedHashCode = hash;
-    return hash;
-  }
-
-  public static org.tensorflow.proto.util.testlog.TestResults parseFrom(
-      java.nio.ByteBuffer data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.testlog.TestResults parseFrom(
-      java.nio.ByteBuffer data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.testlog.TestResults parseFrom(
-      com.google.protobuf.ByteString data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.testlog.TestResults parseFrom(
-      com.google.protobuf.ByteString data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.testlog.TestResults parseFrom(byte[] data)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data);
-  }
-  public static org.tensorflow.proto.util.testlog.TestResults parseFrom(
-      byte[] data,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws com.google.protobuf.InvalidProtocolBufferException {
-    return PARSER.parseFrom(data, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.testlog.TestResults parseFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.testlog.TestResults parseFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.testlog.TestResults parseDelimitedFrom(java.io.InputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.testlog.TestResults parseDelimitedFrom(
-      java.io.InputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
-  }
-  public static org.tensorflow.proto.util.testlog.TestResults parseFrom(
-      com.google.protobuf.CodedInputStream input)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input);
-  }
-  public static org.tensorflow.proto.util.testlog.TestResults parseFrom(
-      com.google.protobuf.CodedInputStream input,
-      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-      throws java.io.IOException {
-    return com.google.protobuf.GeneratedMessageV3
-        .parseWithIOException(PARSER, input, extensionRegistry);
-  }
-
-  @java.lang.Override
-  public Builder newBuilderForType() { return newBuilder(); }
-  public static Builder newBuilder() {
-    return DEFAULT_INSTANCE.toBuilder();
-  }
-  public static Builder newBuilder(org.tensorflow.proto.util.testlog.TestResults prototype) {
-    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
-  }
-  @java.lang.Override
-  public Builder toBuilder() {
-    return this == DEFAULT_INSTANCE
-        ? new Builder() : new Builder().mergeFrom(this);
-  }
-
-  @java.lang.Override
-  protected Builder newBuilderForType(
-      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-    Builder builder = new Builder(parent);
-    return builder;
-  }
-  /**
-   * <pre>
-   * The output of one benchmark / test run.  Each run contains a list of
-   * tests or benchmarks, stored as BenchmarkEntry messages.
-   * This message should be emitted by the reporter (which runs the
-   * test / BM in a subprocess and then reads the emitted BenchmarkEntry messages;
-   * usually from a serialized json file, finally collecting them along
-   * with additional information about the test run.
-   * </pre>
-   *
-   * Protobuf type {@code tensorflow.TestResults}
-   */
-  public static final class Builder extends
-      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
-      // @@protoc_insertion_point(builder_implements:tensorflow.TestResults)
-      org.tensorflow.proto.util.testlog.TestResultsOrBuilder {
-    public static final com.google.protobuf.Descriptors.Descriptor
-        getDescriptor() {
-      return org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_TestResults_descriptor;
-    }
-
-    @java.lang.Override
-    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
-        internalGetFieldAccessorTable() {
-      return org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_TestResults_fieldAccessorTable
-          .ensureFieldAccessorsInitialized(
-              org.tensorflow.proto.util.testlog.TestResults.class, org.tensorflow.proto.util.testlog.TestResults.Builder.class);
-    }
-
-    // Construct using org.tensorflow.proto.util.testlog.TestResults.newBuilder()
-    private Builder() {
-      maybeForceBuilderInitialization();
-    }
-
-    private Builder(
-        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
-      super(parent);
-      maybeForceBuilderInitialization();
-    }
-    private void maybeForceBuilderInitialization() {
-      if (com.google.protobuf.GeneratedMessageV3
-              .alwaysUseFieldBuilders) {
-      }
-    }
-    @java.lang.Override
-    public Builder clear() {
-      super.clear();
-      target_ = "";
-
-      if (entriesBuilder_ == null) {
-        entries_ = null;
-      } else {
-        entries_ = null;
-        entriesBuilder_ = null;
-      }
-      if (buildConfigurationBuilder_ == null) {
-        buildConfiguration_ = null;
-      } else {
-        buildConfiguration_ = null;
-        buildConfigurationBuilder_ = null;
-      }
-      if (commitIdBuilder_ == null) {
-        commitId_ = null;
-      } else {
-        commitId_ = null;
-        commitIdBuilder_ = null;
-      }
-      startTime_ = 0L;
-
-      runTime_ = 0D;
-
-      if (machineConfigurationBuilder_ == null) {
-        machineConfiguration_ = null;
-      } else {
-        machineConfiguration_ = null;
-        machineConfigurationBuilder_ = null;
-      }
-      if (runConfigurationBuilder_ == null) {
-        runConfiguration_ = null;
-      } else {
-        runConfiguration_ = null;
-        runConfigurationBuilder_ = null;
-      }
-      name_ = "";
-
-      benchmarkType_ = 0;
-
-      runMode_ = "";
-
-      tfVersion_ = "";
-
-      return this;
-    }
-
-    @java.lang.Override
-    public com.google.protobuf.Descriptors.Descriptor
-        getDescriptorForType() {
-      return org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_TestResults_descriptor;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.testlog.TestResults getDefaultInstanceForType() {
-      return org.tensorflow.proto.util.testlog.TestResults.getDefaultInstance();
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.testlog.TestResults build() {
-      org.tensorflow.proto.util.testlog.TestResults result = buildPartial();
-      if (!result.isInitialized()) {
-        throw newUninitializedMessageException(result);
-      }
-      return result;
-    }
-
-    @java.lang.Override
-    public org.tensorflow.proto.util.testlog.TestResults buildPartial() {
-      org.tensorflow.proto.util.testlog.TestResults result = new org.tensorflow.proto.util.testlog.TestResults(this);
-      result.target_ = target_;
-      if (entriesBuilder_ == null) {
-        result.entries_ = entries_;
-      } else {
-        result.entries_ = entriesBuilder_.build();
-      }
-      if (buildConfigurationBuilder_ == null) {
-        result.buildConfiguration_ = buildConfiguration_;
-      } else {
-        result.buildConfiguration_ = buildConfigurationBuilder_.build();
-      }
-      if (commitIdBuilder_ == null) {
-        result.commitId_ = commitId_;
-      } else {
-        result.commitId_ = commitIdBuilder_.build();
-      }
-      result.startTime_ = startTime_;
-      result.runTime_ = runTime_;
-      if (machineConfigurationBuilder_ == null) {
-        result.machineConfiguration_ = machineConfiguration_;
-      } else {
-        result.machineConfiguration_ = machineConfigurationBuilder_.build();
-      }
-      if (runConfigurationBuilder_ == null) {
-        result.runConfiguration_ = runConfiguration_;
-      } else {
-        result.runConfiguration_ = runConfigurationBuilder_.build();
-      }
-      result.name_ = name_;
-      result.benchmarkType_ = benchmarkType_;
-      result.runMode_ = runMode_;
-      result.tfVersion_ = tfVersion_;
-      onBuilt();
-      return result;
-    }
-
-    @java.lang.Override
-    public Builder clone() {
-      return super.clone();
-    }
-    @java.lang.Override
-    public Builder setField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.setField(field, value);
-    }
-    @java.lang.Override
-    public Builder clearField(
-        com.google.protobuf.Descriptors.FieldDescriptor field) {
-      return super.clearField(field);
-    }
-    @java.lang.Override
-    public Builder clearOneof(
-        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
-      return super.clearOneof(oneof);
-    }
-    @java.lang.Override
-    public Builder setRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        int index, java.lang.Object value) {
-      return super.setRepeatedField(field, index, value);
-    }
-    @java.lang.Override
-    public Builder addRepeatedField(
-        com.google.protobuf.Descriptors.FieldDescriptor field,
-        java.lang.Object value) {
-      return super.addRepeatedField(field, value);
-    }
-    @java.lang.Override
-    public Builder mergeFrom(com.google.protobuf.Message other) {
-      if (other instanceof org.tensorflow.proto.util.testlog.TestResults) {
-        return mergeFrom((org.tensorflow.proto.util.testlog.TestResults)other);
-      } else {
-        super.mergeFrom(other);
-        return this;
-      }
-    }
-
-    public Builder mergeFrom(org.tensorflow.proto.util.testlog.TestResults other) {
-      if (other == org.tensorflow.proto.util.testlog.TestResults.getDefaultInstance()) return this;
-      if (!other.getTarget().isEmpty()) {
-        target_ = other.target_;
-        onChanged();
-      }
-      if (other.hasEntries()) {
-        mergeEntries(other.getEntries());
-      }
-      if (other.hasBuildConfiguration()) {
-        mergeBuildConfiguration(other.getBuildConfiguration());
-      }
-      if (other.hasCommitId()) {
-        mergeCommitId(other.getCommitId());
-      }
-      if (other.getStartTime() != 0L) {
-        setStartTime(other.getStartTime());
-      }
-      if (other.getRunTime() != 0D) {
-        setRunTime(other.getRunTime());
-      }
-      if (other.hasMachineConfiguration()) {
-        mergeMachineConfiguration(other.getMachineConfiguration());
-      }
-      if (other.hasRunConfiguration()) {
-        mergeRunConfiguration(other.getRunConfiguration());
-      }
-      if (!other.getName().isEmpty()) {
-        name_ = other.name_;
-        onChanged();
-      }
-      if (other.benchmarkType_ != 0) {
-        setBenchmarkTypeValue(other.getBenchmarkTypeValue());
-      }
-      if (!other.getRunMode().isEmpty()) {
-        runMode_ = other.runMode_;
-        onChanged();
-      }
-      if (!other.getTfVersion().isEmpty()) {
-        tfVersion_ = other.tfVersion_;
-        onChanged();
-      }
-      this.mergeUnknownFields(other.unknownFields);
-      onChanged();
-      return this;
-    }
-
-    @java.lang.Override
-    public final boolean isInitialized() {
-      return true;
-    }
-
-    @java.lang.Override
-    public Builder mergeFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws java.io.IOException {
-      org.tensorflow.proto.util.testlog.TestResults parsedMessage = null;
-      try {
-        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
-      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
-        parsedMessage = (org.tensorflow.proto.util.testlog.TestResults) e.getUnfinishedMessage();
-        throw e.unwrapIOException();
-      } finally {
-        if (parsedMessage != null) {
-          mergeFrom(parsedMessage);
-        }
-      }
-      return this;
-    }
-
-    private java.lang.Object target_ = "";
-    /**
-     * <pre>
-     * The target of the run, e.g.:
-     *  //tensorflow/core:kernels_adjust_contrast_op_benchmark_test
-     * </pre>
-     *
-     * <code>string target = 1;</code>
-     */
-    public java.lang.String getTarget() {
-      java.lang.Object ref = target_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        target_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * The target of the run, e.g.:
-     *  //tensorflow/core:kernels_adjust_contrast_op_benchmark_test
-     * </pre>
-     *
-     * <code>string target = 1;</code>
-     */
-    public com.google.protobuf.ByteString
-        getTargetBytes() {
-      java.lang.Object ref = target_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        target_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * The target of the run, e.g.:
-     *  //tensorflow/core:kernels_adjust_contrast_op_benchmark_test
-     * </pre>
-     *
-     * <code>string target = 1;</code>
-     */
-    public Builder setTarget(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      target_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * The target of the run, e.g.:
-     *  //tensorflow/core:kernels_adjust_contrast_op_benchmark_test
-     * </pre>
-     *
-     * <code>string target = 1;</code>
-     */
-    public Builder clearTarget() {
-      
-      target_ = getDefaultInstance().getTarget();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * The target of the run, e.g.:
-     *  //tensorflow/core:kernels_adjust_contrast_op_benchmark_test
-     * </pre>
-     *
-     * <code>string target = 1;</code>
-     */
-    public Builder setTargetBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      target_ = value;
-      onChanged();
-      return this;
-    }
-
-    private org.tensorflow.proto.util.testlog.BenchmarkEntries entries_;
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.util.testlog.BenchmarkEntries, org.tensorflow.proto.util.testlog.BenchmarkEntries.Builder, org.tensorflow.proto.util.testlog.BenchmarkEntriesOrBuilder> entriesBuilder_;
-    /**
-     * <pre>
-     * The list of tests or benchmarks in this run.
-     * </pre>
-     *
-     * <code>.tensorflow.BenchmarkEntries entries = 2;</code>
-     */
-    public boolean hasEntries() {
-      return entriesBuilder_ != null || entries_ != null;
-    }
-    /**
-     * <pre>
-     * The list of tests or benchmarks in this run.
-     * </pre>
-     *
-     * <code>.tensorflow.BenchmarkEntries entries = 2;</code>
-     */
-    public org.tensorflow.proto.util.testlog.BenchmarkEntries getEntries() {
-      if (entriesBuilder_ == null) {
-        return entries_ == null ? org.tensorflow.proto.util.testlog.BenchmarkEntries.getDefaultInstance() : entries_;
-      } else {
-        return entriesBuilder_.getMessage();
-      }
-    }
-    /**
-     * <pre>
-     * The list of tests or benchmarks in this run.
-     * </pre>
-     *
-     * <code>.tensorflow.BenchmarkEntries entries = 2;</code>
-     */
-    public Builder setEntries(org.tensorflow.proto.util.testlog.BenchmarkEntries value) {
-      if (entriesBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        entries_ = value;
-        onChanged();
-      } else {
-        entriesBuilder_.setMessage(value);
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * The list of tests or benchmarks in this run.
-     * </pre>
-     *
-     * <code>.tensorflow.BenchmarkEntries entries = 2;</code>
-     */
-    public Builder setEntries(
-        org.tensorflow.proto.util.testlog.BenchmarkEntries.Builder builderForValue) {
-      if (entriesBuilder_ == null) {
-        entries_ = builderForValue.build();
-        onChanged();
-      } else {
-        entriesBuilder_.setMessage(builderForValue.build());
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * The list of tests or benchmarks in this run.
-     * </pre>
-     *
-     * <code>.tensorflow.BenchmarkEntries entries = 2;</code>
-     */
-    public Builder mergeEntries(org.tensorflow.proto.util.testlog.BenchmarkEntries value) {
-      if (entriesBuilder_ == null) {
-        if (entries_ != null) {
-          entries_ =
-            org.tensorflow.proto.util.testlog.BenchmarkEntries.newBuilder(entries_).mergeFrom(value).buildPartial();
-        } else {
-          entries_ = value;
-        }
-        onChanged();
-      } else {
-        entriesBuilder_.mergeFrom(value);
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * The list of tests or benchmarks in this run.
-     * </pre>
-     *
-     * <code>.tensorflow.BenchmarkEntries entries = 2;</code>
-     */
-    public Builder clearEntries() {
-      if (entriesBuilder_ == null) {
-        entries_ = null;
-        onChanged();
-      } else {
-        entries_ = null;
-        entriesBuilder_ = null;
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * The list of tests or benchmarks in this run.
-     * </pre>
-     *
-     * <code>.tensorflow.BenchmarkEntries entries = 2;</code>
-     */
-    public org.tensorflow.proto.util.testlog.BenchmarkEntries.Builder getEntriesBuilder() {
-      
-      onChanged();
-      return getEntriesFieldBuilder().getBuilder();
-    }
-    /**
-     * <pre>
-     * The list of tests or benchmarks in this run.
-     * </pre>
-     *
-     * <code>.tensorflow.BenchmarkEntries entries = 2;</code>
-     */
-    public org.tensorflow.proto.util.testlog.BenchmarkEntriesOrBuilder getEntriesOrBuilder() {
-      if (entriesBuilder_ != null) {
-        return entriesBuilder_.getMessageOrBuilder();
-      } else {
-        return entries_ == null ?
-            org.tensorflow.proto.util.testlog.BenchmarkEntries.getDefaultInstance() : entries_;
-      }
-    }
-    /**
-     * <pre>
-     * The list of tests or benchmarks in this run.
-     * </pre>
-     *
-     * <code>.tensorflow.BenchmarkEntries entries = 2;</code>
-     */
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.util.testlog.BenchmarkEntries, org.tensorflow.proto.util.testlog.BenchmarkEntries.Builder, org.tensorflow.proto.util.testlog.BenchmarkEntriesOrBuilder> 
-        getEntriesFieldBuilder() {
-      if (entriesBuilder_ == null) {
-        entriesBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-            org.tensorflow.proto.util.testlog.BenchmarkEntries, org.tensorflow.proto.util.testlog.BenchmarkEntries.Builder, org.tensorflow.proto.util.testlog.BenchmarkEntriesOrBuilder>(
-                getEntries(),
-                getParentForChildren(),
-                isClean());
-        entries_ = null;
-      }
-      return entriesBuilder_;
-    }
-
-    private org.tensorflow.proto.util.testlog.BuildConfiguration buildConfiguration_;
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.util.testlog.BuildConfiguration, org.tensorflow.proto.util.testlog.BuildConfiguration.Builder, org.tensorflow.proto.util.testlog.BuildConfigurationOrBuilder> buildConfigurationBuilder_;
-    /**
-     * <pre>
-     * The configuration of the build (compiled opt? with cuda? any copts?)
-     * </pre>
-     *
-     * <code>.tensorflow.BuildConfiguration build_configuration = 3;</code>
-     */
-    public boolean hasBuildConfiguration() {
-      return buildConfigurationBuilder_ != null || buildConfiguration_ != null;
-    }
-    /**
-     * <pre>
-     * The configuration of the build (compiled opt? with cuda? any copts?)
-     * </pre>
-     *
-     * <code>.tensorflow.BuildConfiguration build_configuration = 3;</code>
-     */
-    public org.tensorflow.proto.util.testlog.BuildConfiguration getBuildConfiguration() {
-      if (buildConfigurationBuilder_ == null) {
-        return buildConfiguration_ == null ? org.tensorflow.proto.util.testlog.BuildConfiguration.getDefaultInstance() : buildConfiguration_;
-      } else {
-        return buildConfigurationBuilder_.getMessage();
-      }
-    }
-    /**
-     * <pre>
-     * The configuration of the build (compiled opt? with cuda? any copts?)
-     * </pre>
-     *
-     * <code>.tensorflow.BuildConfiguration build_configuration = 3;</code>
-     */
-    public Builder setBuildConfiguration(org.tensorflow.proto.util.testlog.BuildConfiguration value) {
-      if (buildConfigurationBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        buildConfiguration_ = value;
-        onChanged();
-      } else {
-        buildConfigurationBuilder_.setMessage(value);
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * The configuration of the build (compiled opt? with cuda? any copts?)
-     * </pre>
-     *
-     * <code>.tensorflow.BuildConfiguration build_configuration = 3;</code>
-     */
-    public Builder setBuildConfiguration(
-        org.tensorflow.proto.util.testlog.BuildConfiguration.Builder builderForValue) {
-      if (buildConfigurationBuilder_ == null) {
-        buildConfiguration_ = builderForValue.build();
-        onChanged();
-      } else {
-        buildConfigurationBuilder_.setMessage(builderForValue.build());
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * The configuration of the build (compiled opt? with cuda? any copts?)
-     * </pre>
-     *
-     * <code>.tensorflow.BuildConfiguration build_configuration = 3;</code>
-     */
-    public Builder mergeBuildConfiguration(org.tensorflow.proto.util.testlog.BuildConfiguration value) {
-      if (buildConfigurationBuilder_ == null) {
-        if (buildConfiguration_ != null) {
-          buildConfiguration_ =
-            org.tensorflow.proto.util.testlog.BuildConfiguration.newBuilder(buildConfiguration_).mergeFrom(value).buildPartial();
-        } else {
-          buildConfiguration_ = value;
-        }
-        onChanged();
-      } else {
-        buildConfigurationBuilder_.mergeFrom(value);
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * The configuration of the build (compiled opt? with cuda? any copts?)
-     * </pre>
-     *
-     * <code>.tensorflow.BuildConfiguration build_configuration = 3;</code>
-     */
-    public Builder clearBuildConfiguration() {
-      if (buildConfigurationBuilder_ == null) {
-        buildConfiguration_ = null;
-        onChanged();
-      } else {
-        buildConfiguration_ = null;
-        buildConfigurationBuilder_ = null;
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * The configuration of the build (compiled opt? with cuda? any copts?)
-     * </pre>
-     *
-     * <code>.tensorflow.BuildConfiguration build_configuration = 3;</code>
-     */
-    public org.tensorflow.proto.util.testlog.BuildConfiguration.Builder getBuildConfigurationBuilder() {
-      
-      onChanged();
-      return getBuildConfigurationFieldBuilder().getBuilder();
-    }
-    /**
-     * <pre>
-     * The configuration of the build (compiled opt? with cuda? any copts?)
-     * </pre>
-     *
-     * <code>.tensorflow.BuildConfiguration build_configuration = 3;</code>
-     */
-    public org.tensorflow.proto.util.testlog.BuildConfigurationOrBuilder getBuildConfigurationOrBuilder() {
-      if (buildConfigurationBuilder_ != null) {
-        return buildConfigurationBuilder_.getMessageOrBuilder();
-      } else {
-        return buildConfiguration_ == null ?
-            org.tensorflow.proto.util.testlog.BuildConfiguration.getDefaultInstance() : buildConfiguration_;
-      }
-    }
-    /**
-     * <pre>
-     * The configuration of the build (compiled opt? with cuda? any copts?)
-     * </pre>
-     *
-     * <code>.tensorflow.BuildConfiguration build_configuration = 3;</code>
-     */
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.util.testlog.BuildConfiguration, org.tensorflow.proto.util.testlog.BuildConfiguration.Builder, org.tensorflow.proto.util.testlog.BuildConfigurationOrBuilder> 
-        getBuildConfigurationFieldBuilder() {
-      if (buildConfigurationBuilder_ == null) {
-        buildConfigurationBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-            org.tensorflow.proto.util.testlog.BuildConfiguration, org.tensorflow.proto.util.testlog.BuildConfiguration.Builder, org.tensorflow.proto.util.testlog.BuildConfigurationOrBuilder>(
-                getBuildConfiguration(),
-                getParentForChildren(),
-                isClean());
-        buildConfiguration_ = null;
-      }
-      return buildConfigurationBuilder_;
-    }
-
-    private org.tensorflow.proto.util.testlog.CommitId commitId_;
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.util.testlog.CommitId, org.tensorflow.proto.util.testlog.CommitId.Builder, org.tensorflow.proto.util.testlog.CommitIdOrBuilder> commitIdBuilder_;
-    /**
-     * <pre>
-     * The commit id (git hash or changelist)
-     * </pre>
-     *
-     * <code>.tensorflow.CommitId commit_id = 4;</code>
-     */
-    public boolean hasCommitId() {
-      return commitIdBuilder_ != null || commitId_ != null;
-    }
-    /**
-     * <pre>
-     * The commit id (git hash or changelist)
-     * </pre>
-     *
-     * <code>.tensorflow.CommitId commit_id = 4;</code>
-     */
-    public org.tensorflow.proto.util.testlog.CommitId getCommitId() {
-      if (commitIdBuilder_ == null) {
-        return commitId_ == null ? org.tensorflow.proto.util.testlog.CommitId.getDefaultInstance() : commitId_;
-      } else {
-        return commitIdBuilder_.getMessage();
-      }
-    }
-    /**
-     * <pre>
-     * The commit id (git hash or changelist)
-     * </pre>
-     *
-     * <code>.tensorflow.CommitId commit_id = 4;</code>
-     */
-    public Builder setCommitId(org.tensorflow.proto.util.testlog.CommitId value) {
-      if (commitIdBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        commitId_ = value;
-        onChanged();
-      } else {
-        commitIdBuilder_.setMessage(value);
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * The commit id (git hash or changelist)
-     * </pre>
-     *
-     * <code>.tensorflow.CommitId commit_id = 4;</code>
-     */
-    public Builder setCommitId(
-        org.tensorflow.proto.util.testlog.CommitId.Builder builderForValue) {
-      if (commitIdBuilder_ == null) {
-        commitId_ = builderForValue.build();
-        onChanged();
-      } else {
-        commitIdBuilder_.setMessage(builderForValue.build());
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * The commit id (git hash or changelist)
-     * </pre>
-     *
-     * <code>.tensorflow.CommitId commit_id = 4;</code>
-     */
-    public Builder mergeCommitId(org.tensorflow.proto.util.testlog.CommitId value) {
-      if (commitIdBuilder_ == null) {
-        if (commitId_ != null) {
-          commitId_ =
-            org.tensorflow.proto.util.testlog.CommitId.newBuilder(commitId_).mergeFrom(value).buildPartial();
-        } else {
-          commitId_ = value;
-        }
-        onChanged();
-      } else {
-        commitIdBuilder_.mergeFrom(value);
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * The commit id (git hash or changelist)
-     * </pre>
-     *
-     * <code>.tensorflow.CommitId commit_id = 4;</code>
-     */
-    public Builder clearCommitId() {
-      if (commitIdBuilder_ == null) {
-        commitId_ = null;
-        onChanged();
-      } else {
-        commitId_ = null;
-        commitIdBuilder_ = null;
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * The commit id (git hash or changelist)
-     * </pre>
-     *
-     * <code>.tensorflow.CommitId commit_id = 4;</code>
-     */
-    public org.tensorflow.proto.util.testlog.CommitId.Builder getCommitIdBuilder() {
-      
-      onChanged();
-      return getCommitIdFieldBuilder().getBuilder();
-    }
-    /**
-     * <pre>
-     * The commit id (git hash or changelist)
-     * </pre>
-     *
-     * <code>.tensorflow.CommitId commit_id = 4;</code>
-     */
-    public org.tensorflow.proto.util.testlog.CommitIdOrBuilder getCommitIdOrBuilder() {
-      if (commitIdBuilder_ != null) {
-        return commitIdBuilder_.getMessageOrBuilder();
-      } else {
-        return commitId_ == null ?
-            org.tensorflow.proto.util.testlog.CommitId.getDefaultInstance() : commitId_;
-      }
-    }
-    /**
-     * <pre>
-     * The commit id (git hash or changelist)
-     * </pre>
-     *
-     * <code>.tensorflow.CommitId commit_id = 4;</code>
-     */
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.util.testlog.CommitId, org.tensorflow.proto.util.testlog.CommitId.Builder, org.tensorflow.proto.util.testlog.CommitIdOrBuilder> 
-        getCommitIdFieldBuilder() {
-      if (commitIdBuilder_ == null) {
-        commitIdBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-            org.tensorflow.proto.util.testlog.CommitId, org.tensorflow.proto.util.testlog.CommitId.Builder, org.tensorflow.proto.util.testlog.CommitIdOrBuilder>(
-                getCommitId(),
-                getParentForChildren(),
-                isClean());
-        commitId_ = null;
-      }
-      return commitIdBuilder_;
-    }
-
-    private long startTime_ ;
-    /**
-     * <pre>
-     * The time the run started (in seconds of UTC time since Unix epoch)
-     * </pre>
-     *
-     * <code>int64 start_time = 5;</code>
-     */
-    public long getStartTime() {
-      return startTime_;
-    }
-    /**
-     * <pre>
-     * The time the run started (in seconds of UTC time since Unix epoch)
-     * </pre>
-     *
-     * <code>int64 start_time = 5;</code>
-     */
-    public Builder setStartTime(long value) {
-      
-      startTime_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * The time the run started (in seconds of UTC time since Unix epoch)
-     * </pre>
-     *
-     * <code>int64 start_time = 5;</code>
-     */
-    public Builder clearStartTime() {
-      
-      startTime_ = 0L;
-      onChanged();
-      return this;
-    }
-
-    private double runTime_ ;
-    /**
-     * <pre>
-     * The amount of time the total run took (wall time in seconds)
-     * </pre>
-     *
-     * <code>double run_time = 6;</code>
-     */
-    public double getRunTime() {
-      return runTime_;
-    }
-    /**
-     * <pre>
-     * The amount of time the total run took (wall time in seconds)
-     * </pre>
-     *
-     * <code>double run_time = 6;</code>
-     */
-    public Builder setRunTime(double value) {
-      
-      runTime_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * The amount of time the total run took (wall time in seconds)
-     * </pre>
-     *
-     * <code>double run_time = 6;</code>
-     */
-    public Builder clearRunTime() {
-      
-      runTime_ = 0D;
-      onChanged();
-      return this;
-    }
-
-    private org.tensorflow.proto.util.testlog.MachineConfiguration machineConfiguration_;
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.util.testlog.MachineConfiguration, org.tensorflow.proto.util.testlog.MachineConfiguration.Builder, org.tensorflow.proto.util.testlog.MachineConfigurationOrBuilder> machineConfigurationBuilder_;
-    /**
-     * <pre>
-     * Machine-specific parameters (Platform and CPU info)
-     * </pre>
-     *
-     * <code>.tensorflow.MachineConfiguration machine_configuration = 7;</code>
-     */
-    public boolean hasMachineConfiguration() {
-      return machineConfigurationBuilder_ != null || machineConfiguration_ != null;
-    }
-    /**
-     * <pre>
-     * Machine-specific parameters (Platform and CPU info)
-     * </pre>
-     *
-     * <code>.tensorflow.MachineConfiguration machine_configuration = 7;</code>
-     */
-    public org.tensorflow.proto.util.testlog.MachineConfiguration getMachineConfiguration() {
-      if (machineConfigurationBuilder_ == null) {
-        return machineConfiguration_ == null ? org.tensorflow.proto.util.testlog.MachineConfiguration.getDefaultInstance() : machineConfiguration_;
-      } else {
-        return machineConfigurationBuilder_.getMessage();
-      }
-    }
-    /**
-     * <pre>
-     * Machine-specific parameters (Platform and CPU info)
-     * </pre>
-     *
-     * <code>.tensorflow.MachineConfiguration machine_configuration = 7;</code>
-     */
-    public Builder setMachineConfiguration(org.tensorflow.proto.util.testlog.MachineConfiguration value) {
-      if (machineConfigurationBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        machineConfiguration_ = value;
-        onChanged();
-      } else {
-        machineConfigurationBuilder_.setMessage(value);
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Machine-specific parameters (Platform and CPU info)
-     * </pre>
-     *
-     * <code>.tensorflow.MachineConfiguration machine_configuration = 7;</code>
-     */
-    public Builder setMachineConfiguration(
-        org.tensorflow.proto.util.testlog.MachineConfiguration.Builder builderForValue) {
-      if (machineConfigurationBuilder_ == null) {
-        machineConfiguration_ = builderForValue.build();
-        onChanged();
-      } else {
-        machineConfigurationBuilder_.setMessage(builderForValue.build());
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Machine-specific parameters (Platform and CPU info)
-     * </pre>
-     *
-     * <code>.tensorflow.MachineConfiguration machine_configuration = 7;</code>
-     */
-    public Builder mergeMachineConfiguration(org.tensorflow.proto.util.testlog.MachineConfiguration value) {
-      if (machineConfigurationBuilder_ == null) {
-        if (machineConfiguration_ != null) {
-          machineConfiguration_ =
-            org.tensorflow.proto.util.testlog.MachineConfiguration.newBuilder(machineConfiguration_).mergeFrom(value).buildPartial();
-        } else {
-          machineConfiguration_ = value;
-        }
-        onChanged();
-      } else {
-        machineConfigurationBuilder_.mergeFrom(value);
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Machine-specific parameters (Platform and CPU info)
-     * </pre>
-     *
-     * <code>.tensorflow.MachineConfiguration machine_configuration = 7;</code>
-     */
-    public Builder clearMachineConfiguration() {
-      if (machineConfigurationBuilder_ == null) {
-        machineConfiguration_ = null;
-        onChanged();
-      } else {
-        machineConfiguration_ = null;
-        machineConfigurationBuilder_ = null;
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Machine-specific parameters (Platform and CPU info)
-     * </pre>
-     *
-     * <code>.tensorflow.MachineConfiguration machine_configuration = 7;</code>
-     */
-    public org.tensorflow.proto.util.testlog.MachineConfiguration.Builder getMachineConfigurationBuilder() {
-      
-      onChanged();
-      return getMachineConfigurationFieldBuilder().getBuilder();
-    }
-    /**
-     * <pre>
-     * Machine-specific parameters (Platform and CPU info)
-     * </pre>
-     *
-     * <code>.tensorflow.MachineConfiguration machine_configuration = 7;</code>
-     */
-    public org.tensorflow.proto.util.testlog.MachineConfigurationOrBuilder getMachineConfigurationOrBuilder() {
-      if (machineConfigurationBuilder_ != null) {
-        return machineConfigurationBuilder_.getMessageOrBuilder();
-      } else {
-        return machineConfiguration_ == null ?
-            org.tensorflow.proto.util.testlog.MachineConfiguration.getDefaultInstance() : machineConfiguration_;
-      }
-    }
-    /**
-     * <pre>
-     * Machine-specific parameters (Platform and CPU info)
-     * </pre>
-     *
-     * <code>.tensorflow.MachineConfiguration machine_configuration = 7;</code>
-     */
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.util.testlog.MachineConfiguration, org.tensorflow.proto.util.testlog.MachineConfiguration.Builder, org.tensorflow.proto.util.testlog.MachineConfigurationOrBuilder> 
-        getMachineConfigurationFieldBuilder() {
-      if (machineConfigurationBuilder_ == null) {
-        machineConfigurationBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-            org.tensorflow.proto.util.testlog.MachineConfiguration, org.tensorflow.proto.util.testlog.MachineConfiguration.Builder, org.tensorflow.proto.util.testlog.MachineConfigurationOrBuilder>(
-                getMachineConfiguration(),
-                getParentForChildren(),
-                isClean());
-        machineConfiguration_ = null;
-      }
-      return machineConfigurationBuilder_;
-    }
-
-    private org.tensorflow.proto.util.testlog.RunConfiguration runConfiguration_;
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.util.testlog.RunConfiguration, org.tensorflow.proto.util.testlog.RunConfiguration.Builder, org.tensorflow.proto.util.testlog.RunConfigurationOrBuilder> runConfigurationBuilder_;
-    /**
-     * <pre>
-     * Run-specific parameters (arguments, etc)
-     * </pre>
-     *
-     * <code>.tensorflow.RunConfiguration run_configuration = 8;</code>
-     */
-    public boolean hasRunConfiguration() {
-      return runConfigurationBuilder_ != null || runConfiguration_ != null;
-    }
-    /**
-     * <pre>
-     * Run-specific parameters (arguments, etc)
-     * </pre>
-     *
-     * <code>.tensorflow.RunConfiguration run_configuration = 8;</code>
-     */
-    public org.tensorflow.proto.util.testlog.RunConfiguration getRunConfiguration() {
-      if (runConfigurationBuilder_ == null) {
-        return runConfiguration_ == null ? org.tensorflow.proto.util.testlog.RunConfiguration.getDefaultInstance() : runConfiguration_;
-      } else {
-        return runConfigurationBuilder_.getMessage();
-      }
-    }
-    /**
-     * <pre>
-     * Run-specific parameters (arguments, etc)
-     * </pre>
-     *
-     * <code>.tensorflow.RunConfiguration run_configuration = 8;</code>
-     */
-    public Builder setRunConfiguration(org.tensorflow.proto.util.testlog.RunConfiguration value) {
-      if (runConfigurationBuilder_ == null) {
-        if (value == null) {
-          throw new NullPointerException();
-        }
-        runConfiguration_ = value;
-        onChanged();
-      } else {
-        runConfigurationBuilder_.setMessage(value);
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Run-specific parameters (arguments, etc)
-     * </pre>
-     *
-     * <code>.tensorflow.RunConfiguration run_configuration = 8;</code>
-     */
-    public Builder setRunConfiguration(
-        org.tensorflow.proto.util.testlog.RunConfiguration.Builder builderForValue) {
-      if (runConfigurationBuilder_ == null) {
-        runConfiguration_ = builderForValue.build();
-        onChanged();
-      } else {
-        runConfigurationBuilder_.setMessage(builderForValue.build());
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Run-specific parameters (arguments, etc)
-     * </pre>
-     *
-     * <code>.tensorflow.RunConfiguration run_configuration = 8;</code>
-     */
-    public Builder mergeRunConfiguration(org.tensorflow.proto.util.testlog.RunConfiguration value) {
-      if (runConfigurationBuilder_ == null) {
-        if (runConfiguration_ != null) {
-          runConfiguration_ =
-            org.tensorflow.proto.util.testlog.RunConfiguration.newBuilder(runConfiguration_).mergeFrom(value).buildPartial();
-        } else {
-          runConfiguration_ = value;
-        }
-        onChanged();
-      } else {
-        runConfigurationBuilder_.mergeFrom(value);
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Run-specific parameters (arguments, etc)
-     * </pre>
-     *
-     * <code>.tensorflow.RunConfiguration run_configuration = 8;</code>
-     */
-    public Builder clearRunConfiguration() {
-      if (runConfigurationBuilder_ == null) {
-        runConfiguration_ = null;
-        onChanged();
-      } else {
-        runConfiguration_ = null;
-        runConfigurationBuilder_ = null;
-      }
-
-      return this;
-    }
-    /**
-     * <pre>
-     * Run-specific parameters (arguments, etc)
-     * </pre>
-     *
-     * <code>.tensorflow.RunConfiguration run_configuration = 8;</code>
-     */
-    public org.tensorflow.proto.util.testlog.RunConfiguration.Builder getRunConfigurationBuilder() {
-      
-      onChanged();
-      return getRunConfigurationFieldBuilder().getBuilder();
-    }
-    /**
-     * <pre>
-     * Run-specific parameters (arguments, etc)
-     * </pre>
-     *
-     * <code>.tensorflow.RunConfiguration run_configuration = 8;</code>
-     */
-    public org.tensorflow.proto.util.testlog.RunConfigurationOrBuilder getRunConfigurationOrBuilder() {
-      if (runConfigurationBuilder_ != null) {
-        return runConfigurationBuilder_.getMessageOrBuilder();
-      } else {
-        return runConfiguration_ == null ?
-            org.tensorflow.proto.util.testlog.RunConfiguration.getDefaultInstance() : runConfiguration_;
-      }
-    }
-    /**
-     * <pre>
-     * Run-specific parameters (arguments, etc)
-     * </pre>
-     *
-     * <code>.tensorflow.RunConfiguration run_configuration = 8;</code>
-     */
-    private com.google.protobuf.SingleFieldBuilderV3<
-        org.tensorflow.proto.util.testlog.RunConfiguration, org.tensorflow.proto.util.testlog.RunConfiguration.Builder, org.tensorflow.proto.util.testlog.RunConfigurationOrBuilder> 
-        getRunConfigurationFieldBuilder() {
-      if (runConfigurationBuilder_ == null) {
-        runConfigurationBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
-            org.tensorflow.proto.util.testlog.RunConfiguration, org.tensorflow.proto.util.testlog.RunConfiguration.Builder, org.tensorflow.proto.util.testlog.RunConfigurationOrBuilder>(
-                getRunConfiguration(),
-                getParentForChildren(),
-                isClean());
-        runConfiguration_ = null;
-      }
-      return runConfigurationBuilder_;
-    }
-
-    private java.lang.Object name_ = "";
-    /**
-     * <pre>
-     * Benchmark target identifier.
-     * </pre>
-     *
-     * <code>string name = 9;</code>
-     */
-    public java.lang.String getName() {
-      java.lang.Object ref = name_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        name_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Benchmark target identifier.
-     * </pre>
-     *
-     * <code>string name = 9;</code>
-     */
-    public com.google.protobuf.ByteString
-        getNameBytes() {
-      java.lang.Object ref = name_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        name_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Benchmark target identifier.
-     * </pre>
-     *
-     * <code>string name = 9;</code>
-     */
-    public Builder setName(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      name_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Benchmark target identifier.
-     * </pre>
-     *
-     * <code>string name = 9;</code>
-     */
-    public Builder clearName() {
-      
-      name_ = getDefaultInstance().getName();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Benchmark target identifier.
-     * </pre>
-     *
-     * <code>string name = 9;</code>
-     */
-    public Builder setNameBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      name_ = value;
-      onChanged();
-      return this;
-    }
-
-    private int benchmarkType_ = 0;
-    /**
-     * <code>.tensorflow.TestResults.BenchmarkType benchmark_type = 10;</code>
-     */
-    public int getBenchmarkTypeValue() {
-      return benchmarkType_;
-    }
-    /**
-     * <code>.tensorflow.TestResults.BenchmarkType benchmark_type = 10;</code>
-     */
-    public Builder setBenchmarkTypeValue(int value) {
-      benchmarkType_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>.tensorflow.TestResults.BenchmarkType benchmark_type = 10;</code>
-     */
-    public org.tensorflow.proto.util.testlog.TestResults.BenchmarkType getBenchmarkType() {
-      @SuppressWarnings("deprecation")
-      org.tensorflow.proto.util.testlog.TestResults.BenchmarkType result = org.tensorflow.proto.util.testlog.TestResults.BenchmarkType.valueOf(benchmarkType_);
-      return result == null ? org.tensorflow.proto.util.testlog.TestResults.BenchmarkType.UNRECOGNIZED : result;
-    }
-    /**
-     * <code>.tensorflow.TestResults.BenchmarkType benchmark_type = 10;</code>
-     */
-    public Builder setBenchmarkType(org.tensorflow.proto.util.testlog.TestResults.BenchmarkType value) {
-      if (value == null) {
-        throw new NullPointerException();
-      }
-      
-      benchmarkType_ = value.getNumber();
-      onChanged();
-      return this;
-    }
-    /**
-     * <code>.tensorflow.TestResults.BenchmarkType benchmark_type = 10;</code>
-     */
-    public Builder clearBenchmarkType() {
-      
-      benchmarkType_ = 0;
-      onChanged();
-      return this;
-    }
-
-    private java.lang.Object runMode_ = "";
-    /**
-     * <pre>
-     * Used for differentiating between continuous and debug builds.
-     * Must be one of:
-     * * cbuild: results from continuous build.
-     * * presubmit: results from oneshot requests.
-     * * culprit: results from culprit finder rerun.
-     * </pre>
-     *
-     * <code>string run_mode = 11;</code>
-     */
-    public java.lang.String getRunMode() {
-      java.lang.Object ref = runMode_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        runMode_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Used for differentiating between continuous and debug builds.
-     * Must be one of:
-     * * cbuild: results from continuous build.
-     * * presubmit: results from oneshot requests.
-     * * culprit: results from culprit finder rerun.
-     * </pre>
-     *
-     * <code>string run_mode = 11;</code>
-     */
-    public com.google.protobuf.ByteString
-        getRunModeBytes() {
-      java.lang.Object ref = runMode_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        runMode_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * Used for differentiating between continuous and debug builds.
-     * Must be one of:
-     * * cbuild: results from continuous build.
-     * * presubmit: results from oneshot requests.
-     * * culprit: results from culprit finder rerun.
-     * </pre>
-     *
-     * <code>string run_mode = 11;</code>
-     */
-    public Builder setRunMode(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      runMode_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Used for differentiating between continuous and debug builds.
-     * Must be one of:
-     * * cbuild: results from continuous build.
-     * * presubmit: results from oneshot requests.
-     * * culprit: results from culprit finder rerun.
-     * </pre>
-     *
-     * <code>string run_mode = 11;</code>
-     */
-    public Builder clearRunMode() {
-      
-      runMode_ = getDefaultInstance().getRunMode();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * Used for differentiating between continuous and debug builds.
-     * Must be one of:
-     * * cbuild: results from continuous build.
-     * * presubmit: results from oneshot requests.
-     * * culprit: results from culprit finder rerun.
-     * </pre>
-     *
-     * <code>string run_mode = 11;</code>
-     */
-    public Builder setRunModeBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      runMode_ = value;
-      onChanged();
-      return this;
-    }
-
-    private java.lang.Object tfVersion_ = "";
-    /**
-     * <pre>
-     * TensorFlow version this benchmark runs against.
-     * This can be either set to full version or just the major version.
-     * </pre>
-     *
-     * <code>string tf_version = 12;</code>
-     */
-    public java.lang.String getTfVersion() {
-      java.lang.Object ref = tfVersion_;
-      if (!(ref instanceof java.lang.String)) {
-        com.google.protobuf.ByteString bs =
-            (com.google.protobuf.ByteString) ref;
-        java.lang.String s = bs.toStringUtf8();
-        tfVersion_ = s;
-        return s;
-      } else {
-        return (java.lang.String) ref;
-      }
-    }
-    /**
-     * <pre>
-     * TensorFlow version this benchmark runs against.
-     * This can be either set to full version or just the major version.
-     * </pre>
-     *
-     * <code>string tf_version = 12;</code>
-     */
-    public com.google.protobuf.ByteString
-        getTfVersionBytes() {
-      java.lang.Object ref = tfVersion_;
-      if (ref instanceof String) {
-        com.google.protobuf.ByteString b = 
-            com.google.protobuf.ByteString.copyFromUtf8(
-                (java.lang.String) ref);
-        tfVersion_ = b;
-        return b;
-      } else {
-        return (com.google.protobuf.ByteString) ref;
-      }
-    }
-    /**
-     * <pre>
-     * TensorFlow version this benchmark runs against.
-     * This can be either set to full version or just the major version.
-     * </pre>
-     *
-     * <code>string tf_version = 12;</code>
-     */
-    public Builder setTfVersion(
-        java.lang.String value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  
-      tfVersion_ = value;
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * TensorFlow version this benchmark runs against.
-     * This can be either set to full version or just the major version.
-     * </pre>
-     *
-     * <code>string tf_version = 12;</code>
-     */
-    public Builder clearTfVersion() {
-      
-      tfVersion_ = getDefaultInstance().getTfVersion();
-      onChanged();
-      return this;
-    }
-    /**
-     * <pre>
-     * TensorFlow version this benchmark runs against.
-     * This can be either set to full version or just the major version.
-     * </pre>
-     *
-     * <code>string tf_version = 12;</code>
-     */
-    public Builder setTfVersionBytes(
-        com.google.protobuf.ByteString value) {
-      if (value == null) {
-    throw new NullPointerException();
-  }
-  checkByteStringIsUtf8(value);
-      
-      tfVersion_ = value;
-      onChanged();
-      return this;
-    }
-    @java.lang.Override
-    public final Builder setUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.setUnknownFields(unknownFields);
-    }
-
-    @java.lang.Override
-    public final Builder mergeUnknownFields(
-        final com.google.protobuf.UnknownFieldSet unknownFields) {
-      return super.mergeUnknownFields(unknownFields);
-    }
-
-
-    // @@protoc_insertion_point(builder_scope:tensorflow.TestResults)
-  }
-
-  // @@protoc_insertion_point(class_scope:tensorflow.TestResults)
-  private static final org.tensorflow.proto.util.testlog.TestResults DEFAULT_INSTANCE;
-  static {
-    DEFAULT_INSTANCE = new org.tensorflow.proto.util.testlog.TestResults();
-  }
-
-  public static org.tensorflow.proto.util.testlog.TestResults getDefaultInstance() {
-    return DEFAULT_INSTANCE;
-  }
-
-  private static final com.google.protobuf.Parser<TestResults>
-      PARSER = new com.google.protobuf.AbstractParser<TestResults>() {
-    @java.lang.Override
-    public TestResults parsePartialFrom(
-        com.google.protobuf.CodedInputStream input,
-        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
-        throws com.google.protobuf.InvalidProtocolBufferException {
-      return new TestResults(input, extensionRegistry);
-    }
-  };
-
-  public static com.google.protobuf.Parser<TestResults> parser() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public com.google.protobuf.Parser<TestResults> getParserForType() {
-    return PARSER;
-  }
-
-  @java.lang.Override
-  public org.tensorflow.proto.util.testlog.TestResults getDefaultInstanceForType() {
-    return DEFAULT_INSTANCE;
-  }
-
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/TestResultsOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/TestResultsOrBuilder.java
deleted file mode 100644
index 28ef5b7bfd0..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/TestResultsOrBuilder.java
+++ /dev/null
@@ -1,245 +0,0 @@
-// Generated by the protocol buffer compiler.  DO NOT EDIT!
-// source: tensorflow/core/util/test_log.proto
-
-package org.tensorflow.proto.util.testlog;
-
-public interface TestResultsOrBuilder extends
-    // @@protoc_insertion_point(interface_extends:tensorflow.TestResults)
-    com.google.protobuf.MessageOrBuilder {
-
-  /**
-   * <pre>
-   * The target of the run, e.g.:
-   *  //tensorflow/core:kernels_adjust_contrast_op_benchmark_test
-   * </pre>
-   *
-   * <code>string target = 1;</code>
-   */
-  java.lang.String getTarget();
-  /**
-   * <pre>
-   * The target of the run, e.g.:
-   *  //tensorflow/core:kernels_adjust_contrast_op_benchmark_test
-   * </pre>
-   *
-   * <code>string target = 1;</code>
-   */
-  com.google.protobuf.ByteString
-      getTargetBytes();
-
-  /**
-   * <pre>
-   * The list of tests or benchmarks in this run.
-   * </pre>
-   *
-   * <code>.tensorflow.BenchmarkEntries entries = 2;</code>
-   */
-  boolean hasEntries();
-  /**
-   * <pre>
-   * The list of tests or benchmarks in this run.
-   * </pre>
-   *
-   * <code>.tensorflow.BenchmarkEntries entries = 2;</code>
-   */
-  org.tensorflow.proto.util.testlog.BenchmarkEntries getEntries();
-  /**
-   * <pre>
-   * The list of tests or benchmarks in this run.
-   * </pre>
-   *
-   * <code>.tensorflow.BenchmarkEntries entries = 2;</code>
-   */
-  org.tensorflow.proto.util.testlog.BenchmarkEntriesOrBuilder getEntriesOrBuilder();
-
-  /**
-   * <pre>
-   * The configuration of the build (compiled opt? with cuda? any copts?)
-   * </pre>
-   *
-   * <code>.tensorflow.BuildConfiguration build_configuration = 3;</code>
-   */
-  boolean hasBuildConfiguration();
-  /**
-   * <pre>
-   * The configuration of the build (compiled opt? with cuda? any copts?)
-   * </pre>
-   *
-   * <code>.tensorflow.BuildConfiguration build_configuration = 3;</code>
-   */
-  org.tensorflow.proto.util.testlog.BuildConfiguration getBuildConfiguration();
-  /**
-   * <pre>
-   * The configuration of the build (compiled opt? with cuda? any copts?)
-   * </pre>
-   *
-   * <code>.tensorflow.BuildConfiguration build_configuration = 3;</code>
-   */
-  org.tensorflow.proto.util.testlog.BuildConfigurationOrBuilder getBuildConfigurationOrBuilder();
-
-  /**
-   * <pre>
-   * The commit id (git hash or changelist)
-   * </pre>
-   *
-   * <code>.tensorflow.CommitId commit_id = 4;</code>
-   */
-  boolean hasCommitId();
-  /**
-   * <pre>
-   * The commit id (git hash or changelist)
-   * </pre>
-   *
-   * <code>.tensorflow.CommitId commit_id = 4;</code>
-   */
-  org.tensorflow.proto.util.testlog.CommitId getCommitId();
-  /**
-   * <pre>
-   * The commit id (git hash or changelist)
-   * </pre>
-   *
-   * <code>.tensorflow.CommitId commit_id = 4;</code>
-   */
-  org.tensorflow.proto.util.testlog.CommitIdOrBuilder getCommitIdOrBuilder();
-
-  /**
-   * <pre>
-   * The time the run started (in seconds of UTC time since Unix epoch)
-   * </pre>
-   *
-   * <code>int64 start_time = 5;</code>
-   */
-  long getStartTime();
-
-  /**
-   * <pre>
-   * The amount of time the total run took (wall time in seconds)
-   * </pre>
-   *
-   * <code>double run_time = 6;</code>
-   */
-  double getRunTime();
-
-  /**
-   * <pre>
-   * Machine-specific parameters (Platform and CPU info)
-   * </pre>
-   *
-   * <code>.tensorflow.MachineConfiguration machine_configuration = 7;</code>
-   */
-  boolean hasMachineConfiguration();
-  /**
-   * <pre>
-   * Machine-specific parameters (Platform and CPU info)
-   * </pre>
-   *
-   * <code>.tensorflow.MachineConfiguration machine_configuration = 7;</code>
-   */
-  org.tensorflow.proto.util.testlog.MachineConfiguration getMachineConfiguration();
-  /**
-   * <pre>
-   * Machine-specific parameters (Platform and CPU info)
-   * </pre>
-   *
-   * <code>.tensorflow.MachineConfiguration machine_configuration = 7;</code>
-   */
-  org.tensorflow.proto.util.testlog.MachineConfigurationOrBuilder getMachineConfigurationOrBuilder();
-
-  /**
-   * <pre>
-   * Run-specific parameters (arguments, etc)
-   * </pre>
-   *
-   * <code>.tensorflow.RunConfiguration run_configuration = 8;</code>
-   */
-  boolean hasRunConfiguration();
-  /**
-   * <pre>
-   * Run-specific parameters (arguments, etc)
-   * </pre>
-   *
-   * <code>.tensorflow.RunConfiguration run_configuration = 8;</code>
-   */
-  org.tensorflow.proto.util.testlog.RunConfiguration getRunConfiguration();
-  /**
-   * <pre>
-   * Run-specific parameters (arguments, etc)
-   * </pre>
-   *
-   * <code>.tensorflow.RunConfiguration run_configuration = 8;</code>
-   */
-  org.tensorflow.proto.util.testlog.RunConfigurationOrBuilder getRunConfigurationOrBuilder();
-
-  /**
-   * <pre>
-   * Benchmark target identifier.
-   * </pre>
-   *
-   * <code>string name = 9;</code>
-   */
-  java.lang.String getName();
-  /**
-   * <pre>
-   * Benchmark target identifier.
-   * </pre>
-   *
-   * <code>string name = 9;</code>
-   */
-  com.google.protobuf.ByteString
-      getNameBytes();
-
-  /**
-   * <code>.tensorflow.TestResults.BenchmarkType benchmark_type = 10;</code>
-   */
-  int getBenchmarkTypeValue();
-  /**
-   * <code>.tensorflow.TestResults.BenchmarkType benchmark_type = 10;</code>
-   */
-  org.tensorflow.proto.util.testlog.TestResults.BenchmarkType getBenchmarkType();
-
-  /**
-   * <pre>
-   * Used for differentiating between continuous and debug builds.
-   * Must be one of:
-   * * cbuild: results from continuous build.
-   * * presubmit: results from oneshot requests.
-   * * culprit: results from culprit finder rerun.
-   * </pre>
-   *
-   * <code>string run_mode = 11;</code>
-   */
-  java.lang.String getRunMode();
-  /**
-   * <pre>
-   * Used for differentiating between continuous and debug builds.
-   * Must be one of:
-   * * cbuild: results from continuous build.
-   * * presubmit: results from oneshot requests.
-   * * culprit: results from culprit finder rerun.
-   * </pre>
-   *
-   * <code>string run_mode = 11;</code>
-   */
-  com.google.protobuf.ByteString
-      getRunModeBytes();
-
-  /**
-   * <pre>
-   * TensorFlow version this benchmark runs against.
-   * This can be either set to full version or just the major version.
-   * </pre>
-   *
-   * <code>string tf_version = 12;</code>
-   */
-  java.lang.String getTfVersion();
-  /**
-   * <pre>
-   * TensorFlow version this benchmark runs against.
-   * This can be either set to full version or just the major version.
-   * </pre>
-   *
-   * <code>string tf_version = 12;</code>
-   */
-  com.google.protobuf.ByteString
-      getTfVersionBytes();
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/AbstractOperation.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/AbstractOperation.java
index bb83f21fe76..80aa928a09a 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/AbstractOperation.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/AbstractOperation.java
@@ -17,6 +17,7 @@
 
 import org.bytedeco.javacpp.Pointer;
 import org.tensorflow.ndarray.Shape;
+import org.tensorflow.proto.framework.DataType;
 import org.tensorflow.types.family.TType;
 
 /**
@@ -76,7 +77,7 @@ public String toString() {
    * @param outputIdx index of the output of this operation
    * @return datatype native code
    */
-  abstract int dtype(int outputIdx);
+  abstract DataType dtype(int outputIdx);
 
   /**
    * Returns the tensor of the {@code outputIdx}th output of this operation.
@@ -86,5 +87,5 @@ public String toString() {
    * @param outputIdx index of the output of this operation
    * @return output tensor
    */
-  abstract Tensor<?> tensor(int outputIdx);
+  abstract TensorHandle tensor(int outputIdx);
 }
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/ConcreteFunction.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/ConcreteFunction.java
index 872b4b4d16d..a5a1e157e4b 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/ConcreteFunction.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/ConcreteFunction.java
@@ -16,14 +16,13 @@
 package org.tensorflow;
 
 import java.io.IOException;
-import java.util.List;
-import java.util.ListIterator;
-import java.util.HashMap;
 import java.util.Map;
 import java.util.function.Function;
 import org.tensorflow.op.Ops;
 import org.tensorflow.proto.framework.SignatureDef;
 import org.tensorflow.proto.framework.TensorInfo;
+import org.tensorflow.types.family.TType;
+import org.tensorflow.util.TensorMap;
 
 /**
  * A graph that can be invoked as a single function, with an input and output signature.
@@ -163,14 +162,14 @@ public Signature signature() {
    * @return output tensors resulting from the execution of the function,
    *         mapped by their signature name
    */
-  public Map<String, Tensor<?>> call(Map<String, Tensor<?>> arguments)
+  public TensorMap call(Map<String, TType<?>> arguments)
       throws IllegalArgumentException {
 
     final SignatureDef signatureDef = signature.asSignatureDef();
     final Session.Runner runner = session.runner();
 
     signatureDef.getInputsMap().forEach((argName, t) -> {
-      Tensor<?> tensor = arguments.get(argName);
+      TType<?> tensor = arguments.get(argName);
       if (tensor == null) {
         throw new IllegalArgumentException(String.format("Missing argument [%s]", argName));
       }
@@ -180,24 +179,7 @@ public Map<String, Tensor<?>> call(Map<String, Tensor<?>> arguments)
     Map<String, TensorInfo> outputToNode = signatureDef.getOutputsMap();
     outputToNode.values().forEach(t -> runner.fetch(t.getName()));
 
-    List<Tensor<?>> resultTensors = runner.run();
-    try {
-      ListIterator<Tensor<?>> resultTensorIter = resultTensors.listIterator();
-      Map<String, Tensor<?>> returnMap = new HashMap<String, Tensor<?>>();
-
-      // Use the output names as present in the signature definition
-      for (String nodeName: outputToNode.keySet()) {
-        returnMap.put(nodeName, resultTensorIter.next());
-      }
-      return returnMap;
-
-    } catch (Exception e) {
-      // Release tensors before throwing exception
-      for (Tensor<?> t : resultTensors) {
-        t.close();
-      }
-      throw e;
-    }
+    return runner.run().toMap(outputToNode.keySet());
   }
 
   /**
@@ -210,7 +192,7 @@ public Map<String, Tensor<?>> call(Map<String, Tensor<?>> arguments)
    * @throws IllegalArgumentException if there are multiple input or output parameters defined
    *                                  in the function
    */
-  public Tensor<?> call(Tensor<?> tensor) throws IllegalArgumentException {
+  public <T extends TType> T call(TType<?> tensor) throws IllegalArgumentException {
     final SignatureDef signatureDef = signature.asSignatureDef();
 
     if (signatureDef.getInputsCount() != 1) {
@@ -225,7 +207,7 @@ public Tensor<?> call(Tensor<?> tensor) throws IllegalArgumentException {
     }
     String outputNodeName = signatureDef.getOutputsMap().values().iterator().next().getName();
 
-    return session.runner().feed(inputNodeName, tensor).fetch(outputNodeName).run().get(0);
+    return (T)session.runner().feed(inputNodeName, tensor).fetch(outputNodeName).run().get(0);
   }
 
   /**
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/EagerOperation.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/EagerOperation.java
index 65f7ad443f1..db87f7e81c2 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/EagerOperation.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/EagerOperation.java
@@ -29,7 +29,7 @@
 import org.tensorflow.internal.c_api.TF_Status;
 import org.tensorflow.internal.c_api.TF_Tensor;
 import org.tensorflow.ndarray.Shape;
-import org.tensorflow.op.Ops;
+import org.tensorflow.proto.framework.DataType;
 
 /**
  * Implementation of an {@link Operation} executed eagerly.
@@ -90,12 +90,6 @@ public TFE_TensorHandle getUnsafeNativeHandle(int outputIndex) {
 
   @Override
   public Shape shape(int outputIndex) {
-    // If the tensor of this output has already been resolved, return its shape.
-    // Otherwise, retrieve the tensor shape from the native library.
-    Tensor<?> tensor = outputTensors.get(outputIndex);
-    if (tensor != null) {
-      return tensor.shape();
-    }
     TFE_TensorHandle outputNativeHandle = getUnsafeNativeHandle(outputIndex);
     long[] shape = new long[numDims(outputNativeHandle)];
     for (int i = 0; i < shape.length; ++i) {
@@ -105,14 +99,14 @@ public Shape shape(int outputIndex) {
   }
 
   @Override
-  public int dtype(int outputIndex) {
+  public DataType dtype(int outputIndex) {
     TFE_TensorHandle outputNativeHandle = getUnsafeNativeHandle(outputIndex);
-    return dataType(outputNativeHandle);
+    return DataType.forNumber(dataType(outputNativeHandle));
   }
 
   @Override
-  public Tensor<?> tensor(int outputIndex) {
-    Tensor<?> tensor = outputTensors.get(outputIndex);
+  public TensorHandle tensor(int outputIndex) {
+    TensorHandle tensor = outputTensors.get(outputIndex);
     if (tensor == null) {
       tensor = resolveTensor(outputIndex);
     }
@@ -122,15 +116,15 @@ public Tensor<?> tensor(int outputIndex) {
   private final EagerSession session;
   private final String type;
   private final String name;
-  private final AtomicReferenceArray<Tensor<?>> outputTensors;
+  private final AtomicReferenceArray<TensorHandle> outputTensors;
 
-  private Tensor<?> resolveTensor(int outputIndex) {
+  private TensorHandle resolveTensor(int outputIndex) {
     // Take an optimistic approach, where we attempt to resolve the output tensor without locking.
     // If another thread has resolved it meanwhile, release our copy and reuse the existing one
     // instead.
-    Tensor<?> tensor = resolveTensorHandle(getUnsafeNativeHandle(outputIndex), session);
+    TensorHandle tensor = resolveTensorHandle(getUnsafeNativeHandle(outputIndex), session);
     if (!outputTensors.compareAndSet(outputIndex, null, tensor)) {
-      session.detach(((AbstractTensor)tensor).nativeHandle());
+      session.detach(tensor.nativeHandle());
       tensor = outputTensors.get(outputIndex);
     }
     return tensor;
@@ -151,14 +145,14 @@ private static void requireTensorHandle(TFE_TensorHandle handle) {
     }
   }
 
-  private static Tensor<?> resolveTensorHandle(TFE_TensorHandle handle, EagerSession session) {
+  private static TensorHandle resolveTensorHandle(TFE_TensorHandle handle, EagerSession session) {
     requireTensorHandle(handle);
     try (PointerScope scope = new PointerScope()) {
       TF_Status status = TF_Status.newStatus();
       TF_Tensor nativeTensor = TFE_TensorHandleResolve(handle, status).withDeallocator();
       status.throwExceptionIfNotOK();
-      Tensor<?> t = Tensors.fromHandle(nativeTensor);
-      ((AbstractTensor)t).attachTo(session);
+      TensorHandle t = TensorHandle.of(nativeTensor);
+      t.attachTo(session);
       return t;
     }
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/EagerOperationBuilder.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/EagerOperationBuilder.java
index 807585c4d17..7c682fa5383 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/EagerOperationBuilder.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/EagerOperationBuilder.java
@@ -48,6 +48,7 @@
 import org.tensorflow.internal.c_api.TF_Status;
 import org.tensorflow.internal.c_api.TF_Tensor;
 import org.tensorflow.ndarray.Shape;
+import org.tensorflow.types.TypeRegistry;
 import org.tensorflow.types.family.TType;
 
 /**
@@ -161,7 +162,7 @@ public EagerOperationBuilder setAttr(String name, boolean[] values) {
 
   @Override
   public EagerOperationBuilder setAttr(String name, Class<? extends TType> type) {
-    setAttrType(opHandle, name, TensorTypes.numberOf(type));
+    setAttrType(opHandle, name, TypeRegistry.find(type).dataType().getNumber());
     return this;
   }
 
@@ -169,20 +170,20 @@ public EagerOperationBuilder setAttr(String name, Class<? extends TType> type) {
   public EagerOperationBuilder setAttr(String name, Class<? extends TType>[] types) {
     int[] c = new int[types.length];
     for (int i = 0; i < types.length; ++i) {
-      c[i] = TensorTypes.numberOf(types[i]);
+      c[i] = TypeRegistry.find(types[i]).dataType().getNumber();
     }
     setAttrTypeList(opHandle, name, c);
     return this;
   }
 
   @Override
-  public EagerOperationBuilder setAttr(String name, Tensor<?> value) {
-    setAttrTensor(opHandle, name, ((AbstractTensor)value).nativeHandle());
+  public EagerOperationBuilder setAttr(String name, TType<?> value) {
+    setAttrTensor(opHandle, name, value.tensorHandle().nativeHandle());
     return this;
   }
 
   @Override
-  public EagerOperationBuilder setAttr(String name, Tensor<?>[] values) {
+  public EagerOperationBuilder setAttr(String name, TType<?>[] values) {
     // TODO (karllessard) could be supported by adding this attribute type in the eager C API
     throw new UnsupportedOperationException(
         "Tensor list attributes are not supported in eager mode");
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/GraphOperation.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/GraphOperation.java
index 8757bec651b..36e57d5325c 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/GraphOperation.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/GraphOperation.java
@@ -31,6 +31,8 @@
 import org.tensorflow.internal.c_api.TF_Output;
 import org.tensorflow.internal.c_api.TF_Status;
 import org.tensorflow.ndarray.Shape;
+import org.tensorflow.proto.framework.DataType;
+import org.tensorflow.types.family.TType;
 
 /**
  * Implementation for an {@link Operation} added as a node to a {@link Graph}.
@@ -148,17 +150,17 @@ Shape shape(int outputIdx) {
   }
 
   @Override
-  int dtype(int outputIdx) {
+  DataType dtype(int outputIdx) {
     Graph.Reference r = graph.ref();
     try {
-      return dtype(r.nativeHandle(), getUnsafeNativeHandle(), outputIdx);
+      return DataType.forNumber(dtype(r.nativeHandle(), getUnsafeNativeHandle(), outputIdx));
     } finally {
       r.close();
     }
   }
 
   @Override
-  Tensor<?> tensor(int outputIdx) {
+  TensorHandle tensor(int outputIdx) {
     throw new IllegalStateException("Graph tensors must be fetched by running a session");
   }
 
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/GraphOperationBuilder.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/GraphOperationBuilder.java
index ab68d809a4e..80190624e17 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/GraphOperationBuilder.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/GraphOperationBuilder.java
@@ -52,6 +52,7 @@
 import org.tensorflow.internal.c_api.TF_Status;
 import org.tensorflow.internal.c_api.TF_Tensor;
 import org.tensorflow.ndarray.Shape;
+import org.tensorflow.types.TypeRegistry;
 import org.tensorflow.types.family.TType;
 
 /** An {@link OperationBuilder} for adding {@link GraphOperation}s to a {@link Graph}. */
@@ -225,7 +226,7 @@ public GraphOperationBuilder setAttr(String name, boolean[] value) {
   public GraphOperationBuilder setAttr(String name, Class<? extends TType> type) {
     Graph.Reference r = graph.ref();
     try {
-      setAttrType(unsafeNativeHandle, name, TensorTypes.numberOf(type));
+      setAttrType(unsafeNativeHandle, name, TypeRegistry.find(type).dataType().getNumber());
     } finally {
       r.close();
     }
@@ -236,7 +237,7 @@ public GraphOperationBuilder setAttr(String name, Class<? extends TType> type) {
   public GraphOperationBuilder setAttr(String name, Class<? extends TType>[] types) {
     int[] ctypes = new int[types.length];
     for (int i = 0; i < types.length; ++i) {
-      ctypes[i] = TensorTypes.numberOf(types[i]);
+      ctypes[i] = TypeRegistry.find(types[i]).dataType().getNumber();
     }
     Graph.Reference r = graph.ref();
     try {
@@ -248,10 +249,10 @@ public GraphOperationBuilder setAttr(String name, Class<? extends TType>[] types
   }
 
   @Override
-  public GraphOperationBuilder setAttr(String name, Tensor<?> value) {
+  public GraphOperationBuilder setAttr(String name, TType<?> value) {
     Graph.Reference r = graph.ref();
     try {
-      setAttrTensor(unsafeNativeHandle, name, ((AbstractTensor)value).nativeHandle());
+      setAttrTensor(unsafeNativeHandle, name, value.tensorHandle().nativeHandle());
     } finally {
       r.close();
     }
@@ -259,11 +260,11 @@ public GraphOperationBuilder setAttr(String name, Tensor<?> value) {
   }
 
   @Override
-  public GraphOperationBuilder setAttr(String name, Tensor<?>[] value) {
+  public GraphOperationBuilder setAttr(String name, TType<?>[] value) {
     TF_Tensor[] handles = new TF_Tensor[value.length];
     int idx = 0;
-    for (Tensor<?> t : value) {
-      handles[idx++] = ((AbstractTensor)t).nativeHandle();
+    for (TType<?> t : value) {
+      handles[idx++] = t.tensorHandle().nativeHandle();
     }
     Graph.Reference r = graph.ref();
     try {
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/OperationBuilder.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/OperationBuilder.java
index 89a18741846..75787817a07 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/OperationBuilder.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/OperationBuilder.java
@@ -196,7 +196,7 @@ public interface OperationBuilder {
    * @param value attribute value
    * @return the OperationBuilder instance for chaining.
    */
-   OperationBuilder setAttr(String name, Tensor<?> value);
+   OperationBuilder setAttr(String name, TType<?> value);
 
   /**
    * Set the tensor values of an attribute of the operation being built.
@@ -205,7 +205,7 @@ public interface OperationBuilder {
    * @param value attribute values
    * @return the OperationBuilder instance for chaining.
    */
-   OperationBuilder setAttr(String name, Tensor<?>[] value);
+   OperationBuilder setAttr(String name, TType<?>[] value);
 
   /**
    * Set the shape value of an attribute of the operation being built.
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Output.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Output.java
index 5c1c24bc0f4..02ab4fa5a0f 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Output.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Output.java
@@ -19,6 +19,7 @@
 import org.bytedeco.javacpp.Pointer;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.Scope;
+import org.tensorflow.types.TypeRegistry;
 import org.tensorflow.types.family.TType;
 
 /**
@@ -44,7 +45,7 @@ public Shape shape() {
 
   /** Returns the DataType of the tensor referred to by this Output. */
   public Class<T> type() {
-    return (Class)TensorTypes.find(operation.dtype(index)).typeClass();
+    return (Class) TypeRegistry.find(operation.dtype(index)).typeClass();
   }
 
   /**
@@ -81,7 +82,7 @@ public <U extends TType> Output<U> expect(Class<U> tensorType) {
    */
   @SuppressWarnings("unchecked")
   public T tensor() {
-    return (T)operation.tensor(index);
+    return Tensors.fromHandle(operation.tensor(index));
   }
 
   @Override
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/SavedModelBundle.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/SavedModelBundle.java
index 093898ae56c..97090a74276 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/SavedModelBundle.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/SavedModelBundle.java
@@ -46,6 +46,8 @@
 import org.tensorflow.proto.framework.RunOptions;
 import org.tensorflow.proto.framework.SavedModel;
 import org.tensorflow.proto.util.SaverDef;
+import org.tensorflow.types.family.TType;
+import org.tensorflow.util.TensorMap;
 
 /**
  * SavedModelBundle represents a model loaded from storage.
@@ -334,7 +336,7 @@ public ConcreteFunction function(String signatureKey) {
    * @return list of output tensors, mapped by the signature name
    * @throws IllegalArgumentException if no function can be selected by default
    */
-  public Map<String, Tensor<?>> call(Map<String, Tensor<?>> arguments) {
+  public TensorMap call(Map<String, TType<?>> arguments) {
     ConcreteFunction function = null;
     if (functions.size() == 1) {
       function = functions.values().iterator().next();
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Session.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Session.java
index ce2ad5cbbd1..49502c769ea 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Session.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Session.java
@@ -15,7 +15,15 @@
 
 package org.tensorflow;
 
+import static org.tensorflow.Graph.resolveOutputs;
+import static org.tensorflow.internal.c_api.global.tensorflow.TF_CloseSession;
+import static org.tensorflow.internal.c_api.global.tensorflow.TF_DeleteSession;
+import static org.tensorflow.internal.c_api.global.tensorflow.TF_NewSession;
+import static org.tensorflow.internal.c_api.global.tensorflow.TF_SessionRun;
+import static org.tensorflow.internal.c_api.global.tensorflow.TF_SetConfig;
+
 import com.google.protobuf.InvalidProtocolBufferException;
+import java.util.ArrayList;
 import org.bytedeco.javacpp.BytePointer;
 import org.bytedeco.javacpp.Pointer;
 import org.bytedeco.javacpp.PointerPointer;
@@ -33,15 +41,11 @@
 import org.tensorflow.proto.framework.ConfigProto;
 import org.tensorflow.proto.framework.RunMetadata;
 import org.tensorflow.proto.framework.RunOptions;
-
-import java.util.ArrayList;
-import java.util.List;
 import org.tensorflow.proto.util.SaverDef;
+import org.tensorflow.types.family.TType;
+import org.tensorflow.util.TensorList;
 import org.tensorflow.types.TString;
 
-import static org.tensorflow.Graph.resolveOutputs;
-import static org.tensorflow.internal.c_api.global.tensorflow.*;
-
 /**
  * Driver for {@link Graph} execution.
  *
@@ -158,7 +162,7 @@ public final class Runner {
      * @param t the tensor substituting the operation
      * @return this session runner
      */
-    public Runner feed(String operation, Tensor<?> t) {
+    public Runner feed(String operation, TType<?> t) {
       return feed(parseOutput(operation), t);
     }
 
@@ -173,11 +177,11 @@ public Runner feed(String operation, Tensor<?> t) {
      * @param t the tensor substituting the operation
      * @return this session runner
      */
-    public Runner feed(String operation, int index, Tensor<?> t) {
+    public Runner feed(String operation, int index, TType<?> t) {
       Operation op = operationByName(operation);
       if (op != null) {
         inputs.add(op.output(index));
-        inputTensors.add(t);
+        inputTensors.add(t.tensorHandle());
       }
       return this;
     }
@@ -190,9 +194,9 @@ public Runner feed(String operation, int index, Tensor<?> t) {
      * @param t the tensor substituting the operation
      * @return this session runner
      */
-    public Runner feed(Operand<?> operand, Tensor<?> t) {
+    public Runner feed(Operand<?> operand, TType<?> t) {
       inputs.add(operand.asOutput());
-      inputTensors.add(t);
+      inputTensors.add(t.tensorHandle());
       return this;
     }
 
@@ -325,7 +329,7 @@ public Runner setOptions(RunOptions options) {
      *
      * @return list of resulting tensors fetched by this session runner
      */
-    public List<Tensor<?>> run() {
+    public TensorList run() {
       return runHelper(false).outputs;
     }
 
@@ -354,8 +358,8 @@ private Run runHelper(boolean wantMetadata) {
       // It's okay to use Operation.getUnsafeNativeHandle() here since the safety depends on the
       // validity of the Graph and graphRef ensures that.
       int idx = 0;
-      for (Tensor<?> t : inputTensors) {
-        inputTensorHandles[idx++] = ((AbstractTensor)t).nativeHandle();
+      for (TensorHandle t : inputTensors) {
+        inputTensorHandles[idx++] = t.nativeHandle();
       }
       idx = 0;
       for (Output<?> o : inputs) {
@@ -375,7 +379,7 @@ private Run runHelper(boolean wantMetadata) {
       }
       Reference runRef = new Reference();
       RunMetadata metadata = null;
-      List<Tensor<?>> outputs = new ArrayList<>();
+      TensorList outputs = new TensorList();
       try {
         metadata =
             Session.run(
@@ -390,10 +394,7 @@ private Run runHelper(boolean wantMetadata) {
                 wantMetadata,
                 outputs);
       } catch (Exception e) {
-        for (Tensor<?> t : outputs) {
-          t.close();
-        }
-        outputs.clear();
+        outputs.close();
         throw e;
       } finally {
         runRef.close();
@@ -451,7 +452,7 @@ private Output<?> parseOutput(String opName) {
     }
 
     private ArrayList<Output<?>> inputs = new ArrayList<>();
-    private ArrayList<Tensor<?>> inputTensors = new ArrayList<>();
+    private ArrayList<TensorHandle> inputTensors = new ArrayList<>();
     private ArrayList<Output<?>> outputs = new ArrayList<>();
     private ArrayList<GraphOperation> targets = new ArrayList<>();
     private RunOptions runOptions = null;
@@ -518,7 +519,7 @@ public void save(String prefix) {
    */
   public static final class Run {
     /** Tensors from requested fetches. */
-    public List<Tensor<?>> outputs;
+    public TensorList outputs;
 
     /**
      * Metadata about the run.
@@ -627,7 +628,7 @@ private static RunMetadata run(
       int[] outputOpIndices,
       TF_Operation[] targetOpHandles,
       boolean wantRunMetadata,
-      List<Tensor<?>> outputTensors) {
+      TensorList outputTensors) {
     requireHandle(handle);
 
     int ninputs = inputTensorHandles.length;
@@ -667,7 +668,7 @@ private static RunMetadata run(
 
       for (int i = 0; i < noutputs; ++i) {
         TF_Tensor h = outputValues.get(TF_Tensor.class, i).withDeallocator();
-        outputTensors.add(Tensors.fromHandle(h));
+        outputTensors.add(Tensors.fromHandle(TensorHandle.of(h)));
       }
       try {
         return runMetadata != null ? RunMetadata.parseFrom(runMetadata.dataAsByteBuffer()) : null;
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Signature.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Signature.java
index 66bccbb049e..75785ece8ec 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Signature.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Signature.java
@@ -18,11 +18,11 @@
 import java.util.Map;
 import java.util.Set;
 import org.tensorflow.ndarray.Shape;
-import org.tensorflow.proto.framework.DataType;
 import org.tensorflow.proto.framework.SignatureDef;
 import org.tensorflow.proto.framework.TensorInfo;
 import org.tensorflow.proto.framework.TensorShapeProto;
 import org.tensorflow.proto.framework.TensorShapeProto.Dim;
+import org.tensorflow.types.TypeRegistry;
 
 /**
  * Describe the inputs and outputs of an executable entity, such as a {@link ConcreteFunction}, among
@@ -113,7 +113,7 @@ private static TensorInfo toTensorInfo(Output<?> operand) {
         tensorShapeBuilder.addDim(Dim.newBuilder().setSize(shape.size(i)));
       }
       return TensorInfo.newBuilder()
-          .setDtype(DataType.forNumber(TensorTypes.numberOf(operand.type())))
+          .setDtype(TypeRegistry.find(operand.type()).dataType())
           .setTensorShape(tensorShapeBuilder)
           .setName(operand.op().name() + ":" + operand.index())
           .build();
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Tensor.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Tensor.java
index 1344ec36db7..63752b9920a 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Tensor.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Tensor.java
@@ -15,10 +15,8 @@
 
 package org.tensorflow;
 
-import org.tensorflow.ndarray.NdArray;
 import org.tensorflow.ndarray.buffer.ByteDataBuffer;
-import org.tensorflow.ndarray.buffer.DataBuffer;
-import org.tensorflow.types.family.TType;
+import org.tensorflow.proto.framework.DataType;
 
 /**
  * A statically typed multi-dimensional array whose elements are of a type described by T.
@@ -35,7 +33,7 @@
  * }
  * }</pre>
  */
-public interface Tensor<T> extends NdArray<T>, AutoCloseable {
+public interface Tensor extends AutoCloseable {
 
   /**
    * Release resources associated with the Tensor.
@@ -48,9 +46,6 @@ public interface Tensor<T> extends NdArray<T>, AutoCloseable {
   @Override
   void close();
 
-  /** Returns the size, in bytes, of the tensor data. */
-  long numBytes();
-
   /**
    * Returns the raw data of this tensor as a buffer of bytes.
    *
@@ -63,20 +58,10 @@ public interface Tensor<T> extends NdArray<T>, AutoCloseable {
    */
   ByteDataBuffer rawData();
 
-  @Override
-  Tensor<T> set(NdArray<T> src, long... coordinates);
-
-  @Override
-  Tensor<T> setObject(T value, long... coordinates);
-
-  @Override
-  Tensor<T> copyTo(NdArray<T> dst);
-
-  @Override
-  Tensor<T> read(DataBuffer<T> dst);
+  /** Returns the size, in bytes, of the tensor data. */
+  long numBytes();
 
-  @Override
-  Tensor<T> write(DataBuffer<T> src);
+  DataType dataType();
 }
 
 
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/AbstractTensor.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/TensorHandle.java
similarity index 66%
rename from tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/AbstractTensor.java
rename to tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/TensorHandle.java
index 9c2fff4e2e3..0d2369897d1 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/AbstractTensor.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/TensorHandle.java
@@ -16,26 +16,19 @@
 package org.tensorflow;
 
 import static org.tensorflow.internal.c_api.global.tensorflow.TF_TensorByteSize;
+import static org.tensorflow.internal.c_api.global.tensorflow.TF_TensorType;
 
 import org.tensorflow.internal.c_api.TF_Tensor;
-import org.tensorflow.internal.tensor.buffer.TensorBuffers;
-import org.tensorflow.ndarray.NdArray;
+import org.tensorflow.internal.buffer.TensorBuffers;
 import org.tensorflow.ndarray.buffer.ByteDataBuffer;
-import org.tensorflow.types.family.TType;
-
-/**
- * A basic implementation of a dense tensor.
- *
- * <p>Dense tensors are stored as a continuous segment of native memory, mapped in a n-dimensional
- * space, normally by delegating all I/O operations to an {@link NdArray} instance.
- *
- * <p>Usage of this class is meant to be kept internal.
- */
-public abstract class AbstractTensor<T> implements Tensor<T> {
+import org.tensorflow.proto.framework.DataType;
+
+public final class TensorHandle implements Tensor {
 
   @Override
   public void close() {
     nativeHandle().close();
+    this.nativeHandle = null;
   }
 
   @Override
@@ -49,34 +42,44 @@ public ByteDataBuffer rawData() {
   }
 
   @Override
-  public String toString() {
-    return String.format("%s tensor with shape %s", TensorTypes.find(type()).dataType(), shape());
-  }
-
-  protected AbstractTensor(TF_Tensor nativeHandle) {
-    this.nativeHandle = nativeHandle;
-    nativeHandle.retainReference();
+  public DataType dataType() {
+    return DataType.forNumber(dtype(nativeHandle()));
   }
 
   /**
+   * FIXME public??
    * @return native handle to this tensor
    * @throws IllegalStateException if tensor has been closed
    */
-  protected TF_Tensor nativeHandle() {
+  public TF_Tensor nativeHandle() {
     return requireHandle(nativeHandle);
   }
 
+  static TensorHandle of(TF_Tensor nativeHandle) {
+    return new TensorHandle(nativeHandle);
+  }
+
   void attachTo(EagerSession session) {
     session.attach(nativeHandle());
     nativeHandle.releaseReference();
   }
 
-  protected static TF_Tensor requireHandle(TF_Tensor handle) {
+  private TensorHandle(TF_Tensor nativeHandle) {
+    this.nativeHandle = nativeHandle;
+    nativeHandle.retainReference();
+  }
+
+  private static TF_Tensor requireHandle(TF_Tensor handle) {
     if (handle == null || handle.isNull()) {
       throw new IllegalStateException("close() was called on the Tensor");
     }
     return handle;
   }
 
-  protected final TF_Tensor nativeHandle;
+  private static int dtype(TF_Tensor handle) {
+    requireHandle(handle);
+    return TF_TensorType(handle);
+  }
+
+  private TF_Tensor nativeHandle;
 }
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/TensorType.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/TensorType.java
deleted file mode 100644
index 309d0f47225..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/TensorType.java
+++ /dev/null
@@ -1,47 +0,0 @@
-package org.tensorflow;
-
-import java.lang.reflect.Constructor;
-import org.tensorflow.exceptions.TensorFlowException;
-import org.tensorflow.internal.c_api.TF_Tensor;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.proto.framework.DataType;
-import org.tensorflow.types.family.TType;
-
-class TensorType {
-
-  Class<?> typeClass() {
-    return typeClass;
-  }
-
-  DataType dataType() {
-    return dataType;
-  }
-
-  int byteSize() {
-    return byteSize;
-  }
-
-  boolean isVariableLength() {
-    return byteSize < 0;
-  }
-
-  <T extends TType> T newInstance(TF_Tensor nativeHandle, Shape shape) {
-    try {
-      return (T) implConstructor.newInstance(nativeHandle, shape);
-    } catch (ReflectiveOperationException e) {
-      throw new TensorFlowException("Fail to instantiate tensor: " + e.getMessage(), e);
-    }
-  }
-
-  TensorType(Class<?> typeClass, DataType dataType, int byteSize, Constructor<?> implConstructor) {
-    this.typeClass = typeClass;
-    this.dataType = dataType;
-    this.byteSize = byteSize;
-    this.implConstructor = implConstructor;
-  }
-
-  private final Class<?> typeClass;
-  private final DataType dataType;
-  private final int byteSize;
-  private final Constructor<?> implConstructor;
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Tensors.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Tensors.java
index a19d39429d4..09af17471a4 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Tensors.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Tensors.java
@@ -7,9 +7,12 @@
 import java.util.function.Consumer;
 import org.bytedeco.javacpp.PointerScope;
 import org.tensorflow.internal.c_api.TF_Tensor;
-import org.tensorflow.internal.tensor.buffer.TensorBuffers;
+import org.tensorflow.internal.buffer.TensorBuffers;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.ndarray.buffer.ByteDataBuffer;
+import org.tensorflow.proto.framework.DataType;
+import org.tensorflow.types.Type;
+import org.tensorflow.types.TypeRegistry;
 import org.tensorflow.types.family.TType;
 
 public final class Tensors {
@@ -137,9 +140,9 @@ public static <T extends TType> T of(Class<T> type, Shape shape, long size, Cons
    * @throws IllegalStateException if tensor failed to be allocated with the given parameters
    */
   public static <T extends TType> T of(Class<T> type, Shape shape, ByteDataBuffer rawData) {
-    T tensor = of(type, shape, rawData.size());
-    rawData.copyTo(TensorBuffers.toBytes(((AbstractTensor)tensor).nativeHandle()), rawData.size());
-    return tensor;
+    T t = of(type, shape, rawData.size());
+    rawData.copyTo(TensorBuffers.toBytes(t.tensorHandle().nativeHandle()), rawData.size());
+    return t;
   }
 
   /**
@@ -147,28 +150,28 @@ public static <T extends TType> T of(Class<T> type, Shape shape, ByteDataBuffer
    *
    * <p>Takes ownership of the handle.
    */
-  static Tensor<?> fromHandle(TF_Tensor handle) {
-    TensorType type = TensorTypes.find(dtype(handle));
-    Shape shape = Shape.of(shape(handle));
-    return type.newInstance(handle, shape);
+  static <T extends TType> T fromHandle(TensorHandle handle) {
+    Type<T> type = TypeRegistry.find(handle.dataType());
+    Shape shape = Shape.of(shape(handle.nativeHandle()));
+    return type.factory().createDense(handle, shape);
   }
 
-  private static <T extends TType> T allocate(Class<T> type, Shape shape, long size) {
-    TensorType tensorType = TensorTypes.find(type);
+  private static <T extends TType> T allocate(Class<T> typeClass, Shape shape, long size) {
+    Type<T> type = TypeRegistry.find(typeClass);
     long effectiveSize = size;
     if (effectiveSize < 0) {
       // Size of the tensor is by default the sum of the size of all its element
-      effectiveSize = shape.size() * tensorType.byteSize();
+      effectiveSize = shape.size() * type.byteSize();
 
-    } else if (!tensorType.isVariableLength() && shape.size() * tensorType.byteSize() > effectiveSize) {
+    } else if (!type.isVariableLength() && shape.size() * type.byteSize() > effectiveSize) {
       // Minimum requirements for datatypes of variable length cannot be verified in a relevant way
       // so we only validate them for fixed length datatypes
       throw new IllegalArgumentException("Tensor size is not large enough to contain all scalar values");
     }
-    TF_Tensor nativeHandle = allocate(tensorType.dataType().getNumber(), shape.asArray(), effectiveSize);
+    TF_Tensor nativeHandle = allocate(type.dataType().getNumber(), shape.asArray(), effectiveSize);
     try (PointerScope scope = new PointerScope()) {
       scope.attach(nativeHandle);
-      return tensorType.newInstance(nativeHandle, shape);
+      return type.factory().createDense(TensorHandle.of(nativeHandle), shape);
     }
   }
 
@@ -187,11 +190,6 @@ private static TF_Tensor allocate(int dtype, long[] shape, long byteSize) {
     return t;
   }
 
-  private static int dtype(TF_Tensor handle) {
-    requireHandle(handle);
-    return TF_TensorType(handle);
-  }
-
   private static long[] shape(TF_Tensor handle) {
     requireHandle(handle);
     int numDims = TF_NumDims(handle);
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/buffer/ByteSequenceTensorBuffer.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/buffer/ByteSequenceTensorBuffer.java
similarity index 99%
rename from tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/buffer/ByteSequenceTensorBuffer.java
rename to tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/buffer/ByteSequenceTensorBuffer.java
index 1ab7cb55bc8..bc80b2a164c 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/buffer/ByteSequenceTensorBuffer.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/buffer/ByteSequenceTensorBuffer.java
@@ -15,7 +15,7 @@
  *  =======================================================================
  */
 
-package org.tensorflow.internal.tensor.buffer;
+package org.tensorflow.internal.buffer;
 
 import java.nio.ReadOnlyBufferException;
 import java.util.function.Function;
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/buffer/TensorBuffers.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/buffer/TensorBuffers.java
similarity index 99%
rename from tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/buffer/TensorBuffers.java
rename to tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/buffer/TensorBuffers.java
index f06de6631c1..415c5ca35ef 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/buffer/TensorBuffers.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/buffer/TensorBuffers.java
@@ -15,7 +15,7 @@
  *  =======================================================================
  */
 
-package org.tensorflow.internal.tensor.buffer;
+package org.tensorflow.internal.buffer;
 
 import static org.tensorflow.internal.c_api.global.tensorflow.TF_TensorByteSize;
 import static org.tensorflow.internal.c_api.global.tensorflow.TF_TensorData;
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/buffer/TensorRawDataBufferFactory.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/buffer/TensorRawDataBufferFactory.java
similarity index 98%
rename from tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/buffer/TensorRawDataBufferFactory.java
rename to tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/buffer/TensorRawDataBufferFactory.java
index 3a35edf0485..dbaf31f1dcc 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/buffer/TensorRawDataBufferFactory.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/buffer/TensorRawDataBufferFactory.java
@@ -15,7 +15,7 @@
  *  =======================================================================
  */
 
-package org.tensorflow.internal.tensor.buffer;
+package org.tensorflow.internal.buffer;
 
 import org.bytedeco.javacpp.Pointer;
 import org.tensorflow.ndarray.buffer.BooleanDataBuffer;
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/BooleanTensorImpl.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/BooleanTensorImpl.java
deleted file mode 100644
index 1f9a6b37e6d..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/BooleanTensorImpl.java
+++ /dev/null
@@ -1,141 +0,0 @@
-package org.tensorflow.internal.tensor;
-
-import org.tensorflow.AbstractTensor;
-import org.tensorflow.Tensor;
-import org.tensorflow.internal.c_api.TF_Tensor;
-import org.tensorflow.ndarray.BooleanNdArray;
-import org.tensorflow.ndarray.NdArray;
-import org.tensorflow.ndarray.NdArraySequence;
-import org.tensorflow.ndarray.NdArrays;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.ndarray.buffer.DataBuffer;
-import org.tensorflow.ndarray.buffer.BooleanDataBuffer;
-import org.tensorflow.ndarray.index.Index;
-import org.tensorflow.types.tensor.BooleanTensor;
-
-public class BooleanTensorImpl extends AbstractTensor<Boolean> implements BooleanTensor {
-
-  private final BooleanNdArray data;
-
-  public BooleanTensorImpl(TF_Tensor nativeHandle, Shape shape, BooleanDataBuffer buffer) {
-    super(nativeHandle);
-    data = NdArrays.wrap(shape, buffer);
-  }
-
-  @Override
-  public boolean getBoolean(long... coordinates) {
-    requireHandle(nativeHandle);
-    return data.getBoolean(coordinates);
-  }
-
-  @Override
-  public BooleanTensor setBoolean(boolean value, long... coordinates) {
-    requireHandle(nativeHandle);
-    data.setBoolean(value, coordinates);
-    return this;
-  }
-
-  @Override
-  public BooleanNdArray slice(Index... coordinates) {
-    requireHandle(nativeHandle);
-    return data.slice(coordinates);
-  }
-
-  @Override
-  public BooleanTensor get(long... coordinates) {
-    requireHandle(nativeHandle);
-    data.get(coordinates);
-    return this;
-  }
-
-  @Override
-  public BooleanTensor set(NdArray<Boolean> src, long... coordinates) {
-    requireHandle(nativeHandle);
-    data.set(src, coordinates);
-    return this;
-  }
-
-  @Override
-  public Boolean getObject(long... coordinates) {
-    requireHandle(nativeHandle);
-    return data.getObject(coordinates);
-  }
-
-  @Override
-  public BooleanTensor setObject(Boolean value, long... coordinates) {
-    requireHandle(nativeHandle);
-    data.setObject(value, coordinates);
-    return this;
-  }
-
-  @Override
-  public NdArraySequence<BooleanNdArray> elements(int dimensionIdx) {
-    requireHandle(nativeHandle);
-    return data.elements(dimensionIdx);
-  }
-
-  @Override
-  public NdArraySequence<BooleanNdArray> scalars() {
-    requireHandle(nativeHandle);
-    return data.scalars();
-  }
-
-  @Override
-  public BooleanTensor copyTo(NdArray<Boolean> dst) {
-    requireHandle(nativeHandle);
-    data.copyTo(dst);
-    return this;
-  }
-
-  @Override
-  public BooleanTensor read(DataBuffer<Boolean> dst) {
-    requireHandle(nativeHandle);
-    data.read(dst);
-    return this;
-  }
-
-  @Override
-  public BooleanTensor read(BooleanDataBuffer dst) {
-    requireHandle(nativeHandle);
-    data.read(dst);
-    return this;
-  }
-
-  @Override
-  public BooleanTensor write(DataBuffer<Boolean> src) {
-    requireHandle(nativeHandle);
-    data.write(src);
-    return this;
-  }
-
-  @Override
-  public BooleanTensor write(BooleanDataBuffer src) {
-    requireHandle(nativeHandle);
-    data.write(src);
-    return this;
-  }
-
-  @Override
-  public Shape shape() {
-    requireHandle(nativeHandle);
-    return data.shape();
-  }
-
-  @Override
-  public int rank() {
-    requireHandle(nativeHandle);
-    return data.rank();
-  }
-
-  @Override
-  public long size() {
-    requireHandle(nativeHandle);
-    return data.size();
-  }
-
-  @Override
-  public boolean equals(Object obj) {
-    requireHandle(nativeHandle);
-    return data.equals(obj);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/ByteTensorImpl.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/ByteTensorImpl.java
deleted file mode 100644
index f479b8e0bbf..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/ByteTensorImpl.java
+++ /dev/null
@@ -1,141 +0,0 @@
-package org.tensorflow.internal.tensor;
-
-import org.tensorflow.AbstractTensor;
-import org.tensorflow.Tensor;
-import org.tensorflow.internal.c_api.TF_Tensor;
-import org.tensorflow.ndarray.ByteNdArray;
-import org.tensorflow.ndarray.NdArray;
-import org.tensorflow.ndarray.NdArraySequence;
-import org.tensorflow.ndarray.NdArrays;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.ndarray.buffer.DataBuffer;
-import org.tensorflow.ndarray.buffer.ByteDataBuffer;
-import org.tensorflow.ndarray.index.Index;
-import org.tensorflow.types.tensor.ByteTensor;
-
-public class ByteTensorImpl extends AbstractTensor<Byte> implements ByteTensor {
-
-  private final ByteNdArray data;
-
-  public ByteTensorImpl(TF_Tensor nativeHandle, Shape shape, ByteDataBuffer buffer) {
-    super(nativeHandle);
-    data = NdArrays.wrap(shape, buffer);
-  }
-
-  @Override
-  public byte getByte(long... coordinates) {
-    requireHandle(nativeHandle);
-    return data.getByte(coordinates);
-  }
-
-  @Override
-  public ByteTensor setByte(byte value, long... coordinates) {
-    requireHandle(nativeHandle);
-    data.setByte(value, coordinates);
-    return this;
-  }
-
-  @Override
-  public ByteNdArray slice(Index... coordinates) {
-    requireHandle(nativeHandle);
-    return data.slice(coordinates);
-  }
-
-  @Override
-  public ByteTensor get(long... coordinates) {
-    requireHandle(nativeHandle);
-    data.get(coordinates);
-    return this;
-  }
-
-  @Override
-  public ByteTensor set(NdArray<Byte> src, long... coordinates) {
-    requireHandle(nativeHandle);
-    data.set(src, coordinates);
-    return this;
-  }
-
-  @Override
-  public Byte getObject(long... coordinates) {
-    requireHandle(nativeHandle);
-    return data.getObject(coordinates);
-  }
-
-  @Override
-  public ByteTensor setObject(Byte value, long... coordinates) {
-    requireHandle(nativeHandle);
-    data.setObject(value, coordinates);
-    return this;
-  }
-
-  @Override
-  public NdArraySequence<ByteNdArray> elements(int dimensionIdx) {
-    requireHandle(nativeHandle);
-    return data.elements(dimensionIdx);
-  }
-
-  @Override
-  public NdArraySequence<ByteNdArray> scalars() {
-    requireHandle(nativeHandle);
-    return data.scalars();
-  }
-
-  @Override
-  public ByteTensor copyTo(NdArray<Byte> dst) {
-    requireHandle(nativeHandle);
-    data.copyTo(dst);
-    return this;
-  }
-
-  @Override
-  public ByteTensor read(DataBuffer<Byte> dst) {
-    requireHandle(nativeHandle);
-    data.read(dst);
-    return this;
-  }
-
-  @Override
-  public ByteTensor read(ByteDataBuffer dst) {
-    requireHandle(nativeHandle);
-    data.read(dst);
-    return this;
-  }
-
-  @Override
-  public ByteTensor write(DataBuffer<Byte> src) {
-    requireHandle(nativeHandle);
-    data.write(src);
-    return this;
-  }
-
-  @Override
-  public ByteTensor write(ByteDataBuffer src) {
-    requireHandle(nativeHandle);
-    data.write(src);
-    return this;
-  }
-
-  @Override
-  public Shape shape() {
-    requireHandle(nativeHandle);
-    return data.shape();
-  }
-
-  @Override
-  public int rank() {
-    requireHandle(nativeHandle);
-    return data.rank();
-  }
-
-  @Override
-  public long size() {
-    requireHandle(nativeHandle);
-    return data.size();
-  }
-
-  @Override
-  public boolean equals(Object obj) {
-    requireHandle(nativeHandle);
-    return data.equals(obj);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/DoubleTensorImpl.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/DoubleTensorImpl.java
deleted file mode 100644
index 222ca373645..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/DoubleTensorImpl.java
+++ /dev/null
@@ -1,141 +0,0 @@
-package org.tensorflow.internal.tensor;
-
-import org.tensorflow.AbstractTensor;
-import org.tensorflow.Tensor;
-import org.tensorflow.internal.c_api.TF_Tensor;
-import org.tensorflow.ndarray.DoubleNdArray;
-import org.tensorflow.ndarray.NdArray;
-import org.tensorflow.ndarray.NdArraySequence;
-import org.tensorflow.ndarray.NdArrays;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.ndarray.buffer.DataBuffer;
-import org.tensorflow.ndarray.buffer.DoubleDataBuffer;
-import org.tensorflow.ndarray.index.Index;
-import org.tensorflow.types.tensor.DoubleTensor;
-
-public class DoubleTensorImpl extends AbstractTensor<Double> implements DoubleTensor {
-
-  private final DoubleNdArray data;
-
-  public DoubleTensorImpl(TF_Tensor nativeHandle, Shape shape, DoubleDataBuffer buffer) {
-    super(nativeHandle);
-    data = NdArrays.wrap(shape, buffer);
-  }
-
-  @Override
-  public double getDouble(long... coordinates) {
-    requireHandle(nativeHandle);
-    return data.getDouble(coordinates);
-  }
-
-  @Override
-  public DoubleTensor setDouble(double value, long... coordinates) {
-    requireHandle(nativeHandle);
-    data.setDouble(value, coordinates);
-    return this;
-  }
-
-  @Override
-  public DoubleNdArray slice(Index... coordinates) {
-    requireHandle(nativeHandle);
-    return data.slice(coordinates);
-  }
-
-  @Override
-  public DoubleTensor get(long... coordinates) {
-    requireHandle(nativeHandle);
-    data.get(coordinates);
-    return this;
-  }
-
-  @Override
-  public DoubleTensor set(NdArray<Double> src, long... coordinates) {
-    requireHandle(nativeHandle);
-    data.set(src, coordinates);
-    return this;
-  }
-
-  @Override
-  public Double getObject(long... coordinates) {
-    requireHandle(nativeHandle);
-    return data.getObject(coordinates);
-  }
-
-  @Override
-  public DoubleTensor setObject(Double value, long... coordinates) {
-    requireHandle(nativeHandle);
-    data.setObject(value, coordinates);
-    return this;
-  }
-
-  @Override
-  public NdArraySequence<DoubleNdArray> elements(int dimensionIdx) {
-    requireHandle(nativeHandle);
-    return data.elements(dimensionIdx);
-  }
-
-  @Override
-  public NdArraySequence<DoubleNdArray> scalars() {
-    requireHandle(nativeHandle);
-    return data.scalars();
-  }
-
-  @Override
-  public DoubleTensor copyTo(NdArray<Double> dst) {
-    requireHandle(nativeHandle);
-    data.copyTo(dst);
-    return this;
-  }
-
-  @Override
-  public DoubleTensor read(DataBuffer<Double> dst) {
-    requireHandle(nativeHandle);
-    data.read(dst);
-    return this;
-  }
-
-  @Override
-  public DoubleTensor read(DoubleDataBuffer dst) {
-    requireHandle(nativeHandle);
-    data.read(dst);
-    return this;
-  }
-
-  @Override
-  public DoubleTensor write(DataBuffer<Double> src) {
-    requireHandle(nativeHandle);
-    data.write(src);
-    return this;
-  }
-
-  @Override
-  public DoubleTensor write(DoubleDataBuffer src) {
-    requireHandle(nativeHandle);
-    data.write(src);
-    return this;
-  }
-
-  @Override
-  public Shape shape() {
-    requireHandle(nativeHandle);
-    return data.shape();
-  }
-
-  @Override
-  public int rank() {
-    requireHandle(nativeHandle);
-    return data.rank();
-  }
-
-  @Override
-  public long size() {
-    requireHandle(nativeHandle);
-    return data.size();
-  }
-
-  @Override
-  public boolean equals(Object obj) {
-    requireHandle(nativeHandle);
-    return data.equals(obj);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/FloatTensorImpl.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/FloatTensorImpl.java
deleted file mode 100644
index 5d9d942717e..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/FloatTensorImpl.java
+++ /dev/null
@@ -1,140 +0,0 @@
-package org.tensorflow.internal.tensor;
-
-import org.tensorflow.AbstractTensor;
-import org.tensorflow.internal.c_api.TF_Tensor;
-import org.tensorflow.ndarray.FloatNdArray;
-import org.tensorflow.ndarray.NdArray;
-import org.tensorflow.ndarray.NdArraySequence;
-import org.tensorflow.ndarray.NdArrays;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.ndarray.buffer.DataBuffer;
-import org.tensorflow.ndarray.buffer.FloatDataBuffer;
-import org.tensorflow.ndarray.index.Index;
-import org.tensorflow.types.tensor.FloatTensor;
-
-public class FloatTensorImpl extends AbstractTensor<Float> implements FloatTensor {
-
-  private final FloatNdArray data;
-
-  public FloatTensorImpl(TF_Tensor nativeHandle, Shape shape, FloatDataBuffer buffer) {
-    super(nativeHandle);
-    data = NdArrays.wrap(shape, buffer);
-  }
-
-  @Override
-  public float getFloat(long... coordinates) {
-    requireHandle(nativeHandle);
-    return data.getFloat(coordinates);
-  }
-
-  @Override
-  public FloatTensor setFloat(float value, long... coordinates) {
-    requireHandle(nativeHandle);
-    data.setFloat(value, coordinates);
-    return this;
-  }
-
-  @Override
-  public FloatNdArray slice(Index... coordinates) {
-    requireHandle(nativeHandle);
-    return data.slice(coordinates);
-  }
-
-  @Override
-  public FloatTensor get(long... coordinates) {
-    requireHandle(nativeHandle);
-    data.get(coordinates);
-    return this;
-  }
-
-  @Override
-  public FloatTensor set(NdArray<Float> src, long... coordinates) {
-    requireHandle(nativeHandle);
-    data.set(src, coordinates);
-    return this;
-  }
-
-  @Override
-  public Float getObject(long... coordinates) {
-    requireHandle(nativeHandle);
-    return data.getObject(coordinates);
-  }
-
-  @Override
-  public FloatTensor setObject(Float value, long... coordinates) {
-    requireHandle(nativeHandle);
-    data.setObject(value, coordinates);
-    return this;
-  }
-
-  @Override
-  public NdArraySequence<FloatNdArray> elements(int dimensionIdx) {
-    requireHandle(nativeHandle);
-    return data.elements(dimensionIdx);
-  }
-
-  @Override
-  public NdArraySequence<FloatNdArray> scalars() {
-    requireHandle(nativeHandle);
-    return data.scalars();
-  }
-
-  @Override
-  public FloatTensor copyTo(NdArray<Float> dst) {
-    requireHandle(nativeHandle);
-    data.copyTo(dst);
-    return this;
-  }
-
-  @Override
-  public FloatTensor read(DataBuffer<Float> dst) {
-    requireHandle(nativeHandle);
-    data.read(dst);
-    return this;
-  }
-
-  @Override
-  public FloatTensor read(FloatDataBuffer dst) {
-    requireHandle(nativeHandle);
-    data.read(dst);
-    return this;
-  }
-
-  @Override
-  public FloatTensor write(DataBuffer<Float> src) {
-    requireHandle(nativeHandle);
-    data.write(src);
-    return this;
-  }
-
-  @Override
-  public FloatTensor write(FloatDataBuffer src) {
-    requireHandle(nativeHandle);
-    data.write(src);
-    return this;
-  }
-
-  @Override
-  public Shape shape() {
-    requireHandle(nativeHandle);
-    return data.shape();
-  }
-
-  @Override
-  public int rank() {
-    requireHandle(nativeHandle);
-    return data.rank();
-  }
-
-  @Override
-  public long size() {
-    requireHandle(nativeHandle);
-    return data.size();
-  }
-
-  @Override
-  public boolean equals(Object obj) {
-    requireHandle(nativeHandle);
-    return data.equals(obj);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/IntTensorImpl.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/IntTensorImpl.java
deleted file mode 100644
index 63b507676a2..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/IntTensorImpl.java
+++ /dev/null
@@ -1,141 +0,0 @@
-package org.tensorflow.internal.tensor;
-
-import org.tensorflow.AbstractTensor;
-import org.tensorflow.Tensor;
-import org.tensorflow.internal.c_api.TF_Tensor;
-import org.tensorflow.ndarray.IntNdArray;
-import org.tensorflow.ndarray.NdArray;
-import org.tensorflow.ndarray.NdArraySequence;
-import org.tensorflow.ndarray.NdArrays;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.ndarray.buffer.DataBuffer;
-import org.tensorflow.ndarray.buffer.IntDataBuffer;
-import org.tensorflow.ndarray.index.Index;
-import org.tensorflow.types.tensor.IntTensor;
-
-public class IntTensorImpl extends AbstractTensor<Integer> implements IntTensor {
-
-  private final IntNdArray data;
-
-  public IntTensorImpl(TF_Tensor nativeHandle, Shape shape, IntDataBuffer buffer) {
-    super(nativeHandle);
-    data = NdArrays.wrap(shape, buffer);
-  }
-
-  @Override
-  public int getInt(long... coordinates) {
-    requireHandle(nativeHandle);
-    return data.getInt(coordinates);
-  }
-
-  @Override
-  public IntTensor setInt(int value, long... coordinates) {
-    requireHandle(nativeHandle);
-    data.setInt(value, coordinates);
-    return this;
-  }
-
-  @Override
-  public IntNdArray slice(Index... coordinates) {
-    requireHandle(nativeHandle);
-    return data.slice(coordinates);
-  }
-
-  @Override
-  public IntTensor get(long... coordinates) {
-    requireHandle(nativeHandle);
-    data.get(coordinates);
-    return this;
-  }
-
-  @Override
-  public IntTensor set(NdArray<Integer> src, long... coordinates) {
-    requireHandle(nativeHandle);
-    data.set(src, coordinates);
-    return this;
-  }
-
-  @Override
-  public Integer getObject(long... coordinates) {
-    requireHandle(nativeHandle);
-    return data.getObject(coordinates);
-  }
-
-  @Override
-  public IntTensor setObject(Integer value, long... coordinates) {
-    requireHandle(nativeHandle);
-    data.setObject(value, coordinates);
-    return this;
-  }
-
-  @Override
-  public NdArraySequence<IntNdArray> elements(int dimensionIdx) {
-    requireHandle(nativeHandle);
-    return data.elements(dimensionIdx);
-  }
-
-  @Override
-  public NdArraySequence<IntNdArray> scalars() {
-    requireHandle(nativeHandle);
-    return data.scalars();
-  }
-
-  @Override
-  public IntTensor copyTo(NdArray<Integer> dst) {
-    requireHandle(nativeHandle);
-    data.copyTo(dst);
-    return this;
-  }
-
-  @Override
-  public IntTensor read(DataBuffer<Integer> dst) {
-    requireHandle(nativeHandle);
-    data.read(dst);
-    return this;
-  }
-
-  @Override
-  public IntTensor read(IntDataBuffer dst) {
-    requireHandle(nativeHandle);
-    data.read(dst);
-    return this;
-  }
-
-  @Override
-  public IntTensor write(DataBuffer<Integer> src) {
-    requireHandle(nativeHandle);
-    data.write(src);
-    return this;
-  }
-
-  @Override
-  public IntTensor write(IntDataBuffer src) {
-    requireHandle(nativeHandle);
-    data.write(src);
-    return this;
-  }
-
-  @Override
-  public Shape shape() {
-    requireHandle(nativeHandle);
-    return data.shape();
-  }
-
-  @Override
-  public int rank() {
-    requireHandle(nativeHandle);
-    return data.rank();
-  }
-
-  @Override
-  public long size() {
-    requireHandle(nativeHandle);
-    return data.size();
-  }
-
-  @Override
-  public boolean equals(Object obj) {
-    requireHandle(nativeHandle);
-    return data.equals(obj);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/LongTensorImpl.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/LongTensorImpl.java
deleted file mode 100644
index cc235cf6e19..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/LongTensorImpl.java
+++ /dev/null
@@ -1,141 +0,0 @@
-package org.tensorflow.internal.tensor;
-
-import org.tensorflow.AbstractTensor;
-import org.tensorflow.Tensor;
-import org.tensorflow.internal.c_api.TF_Tensor;
-import org.tensorflow.ndarray.LongNdArray;
-import org.tensorflow.ndarray.NdArray;
-import org.tensorflow.ndarray.NdArraySequence;
-import org.tensorflow.ndarray.NdArrays;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.ndarray.buffer.DataBuffer;
-import org.tensorflow.ndarray.buffer.LongDataBuffer;
-import org.tensorflow.ndarray.index.Index;
-import org.tensorflow.types.tensor.LongTensor;
-
-public class LongTensorImpl extends AbstractTensor<Long> implements LongTensor {
-
-  private final LongNdArray data;
-
-  public LongTensorImpl(TF_Tensor nativeHandle, Shape shape, LongDataBuffer buffer) {
-    super(nativeHandle);
-    data = NdArrays.wrap(shape, buffer);
-  }
-
-  @Override
-  public long getLong(long... coordinates) {
-    requireHandle(nativeHandle);
-    return data.getLong(coordinates);
-  }
-
-  @Override
-  public LongTensor setLong(long value, long... coordinates) {
-    requireHandle(nativeHandle);
-    data.setLong(value, coordinates);
-    return this;
-  }
-
-  @Override
-  public LongNdArray slice(Index... coordinates) {
-    requireHandle(nativeHandle);
-    return data.slice(coordinates);
-  }
-
-  @Override
-  public LongTensor get(long... coordinates) {
-    requireHandle(nativeHandle);
-    data.get(coordinates);
-    return this;
-  }
-
-  @Override
-  public LongTensor set(NdArray<Long> src, long... coordinates) {
-    requireHandle(nativeHandle);
-    data.set(src, coordinates);
-    return this;
-  }
-
-  @Override
-  public Long getObject(long... coordinates) {
-    requireHandle(nativeHandle);
-    return data.getObject(coordinates);
-  }
-
-  @Override
-  public LongTensor setObject(Long value, long... coordinates) {
-    requireHandle(nativeHandle);
-    data.setObject(value, coordinates);
-    return this;
-  }
-
-  @Override
-  public NdArraySequence<LongNdArray> elements(int dimensionIdx) {
-    requireHandle(nativeHandle);
-    return data.elements(dimensionIdx);
-  }
-
-  @Override
-  public NdArraySequence<LongNdArray> scalars() {
-    requireHandle(nativeHandle);
-    return data.scalars();
-  }
-
-  @Override
-  public LongTensor copyTo(NdArray<Long> dst) {
-    requireHandle(nativeHandle);
-    data.copyTo(dst);
-    return this;
-  }
-
-  @Override
-  public LongTensor read(DataBuffer<Long> dst) {
-    requireHandle(nativeHandle);
-    data.read(dst);
-    return this;
-  }
-
-  @Override
-  public LongTensor read(LongDataBuffer dst) {
-    requireHandle(nativeHandle);
-    data.read(dst);
-    return this;
-  }
-
-  @Override
-  public LongTensor write(DataBuffer<Long> src) {
-    requireHandle(nativeHandle);
-    data.write(src);
-    return this;
-  }
-
-  @Override
-  public LongTensor write(LongDataBuffer src) {
-    requireHandle(nativeHandle);
-    data.write(src);
-    return this;
-  }
-
-  @Override
-  public Shape shape() {
-    requireHandle(nativeHandle);
-    return data.shape();
-  }
-
-  @Override
-  public int rank() {
-    requireHandle(nativeHandle);
-    return data.rank();
-  }
-
-  @Override
-  public long size() {
-    requireHandle(nativeHandle);
-    return data.size();
-  }
-
-  @Override
-  public boolean equals(Object obj) {
-    requireHandle(nativeHandle);
-    return data.equals(obj);
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/StringTensorImpl.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/StringTensorImpl.java
deleted file mode 100644
index 306d29f3cbe..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/tensor/StringTensorImpl.java
+++ /dev/null
@@ -1,130 +0,0 @@
-package org.tensorflow.internal.tensor;
-
-import org.tensorflow.AbstractTensor;
-import org.tensorflow.Tensor;
-import org.tensorflow.internal.c_api.TF_Tensor;
-import org.tensorflow.internal.tensor.buffer.ByteSequenceTensorBuffer;
-import org.tensorflow.ndarray.NdArray;
-import org.tensorflow.ndarray.NdArraySequence;
-import org.tensorflow.ndarray.NdArrays;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.ndarray.buffer.DataBuffer;
-import org.tensorflow.ndarray.buffer.layout.DataLayout;
-import org.tensorflow.ndarray.index.Index;
-import org.tensorflow.types.tensor.StringTensor;
-
-public class StringTensorImpl extends AbstractTensor<String> implements StringTensor {
-
-  private final ByteSequenceTensorBuffer rawBuffer;
-  private final NdArray<String> data;
-
-  public StringTensorImpl(
-      TF_Tensor nativeHandle,
-      Shape shape,
-      DataLayout<DataBuffer<byte[]>, String> layout,
-      ByteSequenceTensorBuffer buffer
-  ) {
-    super(nativeHandle);
-    this.rawBuffer = buffer;
-    data = NdArrays.wrap(shape, layout.applyTo(buffer));
-  }
-
-  @Override
-  public NdArray<byte[]> asBytes() {
-    requireHandle(nativeHandle);
-    return NdArrays.wrap(shape(), rawBuffer);
-  }
-
-  @Override
-  public NdArraySequence<? extends NdArray<String>> elements(int dimensionIdx) {
-    requireHandle(nativeHandle);
-    return data.elements(dimensionIdx);
-  }
-
-  @Override
-  public NdArraySequence<? extends NdArray<String>> scalars() {
-    requireHandle(nativeHandle);
-    return data.scalars();
-  }
-
-  @Override
-  public NdArray<String> slice(Index... indices) {
-    requireHandle(nativeHandle);
-    return data.slice(indices);
-  }
-
-  @Override
-  public NdArray<String> get(long... coordinates) {
-    requireHandle(nativeHandle);
-    return data.get(coordinates);
-  }
-
-  @Override
-  public StringTensor set(NdArray<String> src, long... coordinates) {
-    requireHandle(nativeHandle);
-    data.set(src, coordinates);
-    return this;
-  }
-
-  @Override
-  public String getObject(long... coordinates) {
-    requireHandle(nativeHandle);
-    return data.getObject(coordinates);
-  }
-
-  @Override
-  public StringTensor setObject(String value, long... coordinates) {
-    requireHandle(nativeHandle);
-    data.setObject(value, coordinates);
-    return this;
-  }
-
-  @Override
-  public StringTensor copyTo(NdArray<String> dst) {
-    requireHandle(nativeHandle);
-    data.copyTo(dst);
-    return this;
-  }
-
-  @Override
-  public StringTensor read(DataBuffer<String> dst) {
-    requireHandle(nativeHandle);
-    data.read(dst);
-    return this;
-  }
-
-  @Override
-  public StringTensor write(DataBuffer<String> src) {
-    requireHandle(nativeHandle);
-    data.write(src);
-    return this;
-  }
-
-  @Override
-  public Shape shape() {
-    requireHandle(nativeHandle);
-    return data.shape();
-  }
-
-  @Override
-  public int rank() {
-    requireHandle(nativeHandle);
-    return data.rank();
-  }
-
-  @Override
-  public long size() {
-    requireHandle(nativeHandle);
-    return data.size();
-  }
-
-  @Override
-  public boolean equals(Object obj) {
-    requireHandle(nativeHandle);
-    return data.equals(obj);
-  }
-
-  protected ByteSequenceTensorBuffer rawBuffer() {
-    return rawBuffer;
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/types/TBfloat16Factory.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/types/TBfloat16Factory.java
new file mode 100644
index 00000000000..7898a9896b9
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/types/TBfloat16Factory.java
@@ -0,0 +1,42 @@
+package org.tensorflow.internal.types;
+
+import org.tensorflow.TensorHandle;
+import org.tensorflow.internal.buffer.TensorBuffers;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.ndarray.buffer.FloatDataBuffer;
+import org.tensorflow.ndarray.buffer.layout.DataLayouts;
+import org.tensorflow.ndarray.impl.dense.FloatDenseNdArray;
+import org.tensorflow.types.TBfloat16;
+import org.tensorflow.types.TypeFactory;
+
+/**
+ * Factory of {@link TBfloat16} tensor instances
+ */
+public class TBfloat16Factory implements TypeFactory<TBfloat16> {
+
+  @Override
+  public TBfloat16 createDense(TensorHandle tensorHandle, Shape shape) {
+    FloatDataBuffer buffer = DataLayouts.BFLOAT16.applyTo(TensorBuffers.toShorts(tensorHandle.nativeHandle()));
+    return new TBfloat16Impl(tensorHandle, buffer, shape);
+  }
+
+  private static final class TBfloat16Impl extends FloatDenseNdArray implements TBfloat16 {
+
+    @Override
+    public Class<TBfloat16> type() {
+      return TBfloat16.class;
+    }
+
+    @Override
+    public TensorHandle tensorHandle() {
+      return tensorHandle;
+    }
+
+    TBfloat16Impl(TensorHandle tensorHandle, FloatDataBuffer buffer, Shape shape) {
+      super(buffer, shape);
+      this.tensorHandle = tensorHandle;
+    }
+
+    private final TensorHandle tensorHandle;
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/types/TBoolFactory.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/types/TBoolFactory.java
new file mode 100644
index 00000000000..4e14bfd07a7
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/types/TBoolFactory.java
@@ -0,0 +1,41 @@
+package org.tensorflow.internal.types;
+
+import org.tensorflow.TensorHandle;
+import org.tensorflow.internal.buffer.TensorBuffers;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.ndarray.buffer.BooleanDataBuffer;
+import org.tensorflow.ndarray.impl.dense.BooleanDenseNdArray;
+import org.tensorflow.types.TBool;
+import org.tensorflow.types.TypeFactory;
+
+/**
+ * Factory of {@link TBool} tensor instances
+ */
+public class TBoolFactory implements TypeFactory<TBool> {
+
+  @Override
+  public TBool createDense(TensorHandle tensorHandle, Shape shape) {
+    BooleanDataBuffer buffer = TensorBuffers.toBooleans(tensorHandle.nativeHandle());
+    return new TBoolImpl(tensorHandle, buffer, shape);
+  }
+
+  private static final class TBoolImpl extends BooleanDenseNdArray implements TBool {
+
+    @Override
+    public Class<TBool> type() {
+      return TBool.class;
+    }
+
+    @Override
+    public TensorHandle tensorHandle() {
+      return tensorHandle;
+    }
+
+    TBoolImpl(TensorHandle tensorHandle, BooleanDataBuffer buffer, Shape shape) {
+      super(buffer, shape);
+      this.tensorHandle = tensorHandle;
+    }
+
+    private final TensorHandle tensorHandle;
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/types/TFloat16Factory.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/types/TFloat16Factory.java
new file mode 100644
index 00000000000..e54fb98c95d
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/types/TFloat16Factory.java
@@ -0,0 +1,42 @@
+package org.tensorflow.internal.types;
+
+import org.tensorflow.TensorHandle;
+import org.tensorflow.internal.buffer.TensorBuffers;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.ndarray.buffer.FloatDataBuffer;
+import org.tensorflow.ndarray.buffer.layout.DataLayouts;
+import org.tensorflow.ndarray.impl.dense.FloatDenseNdArray;
+import org.tensorflow.types.TFloat16;
+import org.tensorflow.types.TypeFactory;
+
+/**
+ * Factory of {@link TFloat16} tensor instances
+ */
+public class TFloat16Factory implements TypeFactory<TFloat16> {
+
+  @Override
+  public TFloat16 createDense(TensorHandle tensorHandle, Shape shape) {
+    FloatDataBuffer buffer = DataLayouts.FLOAT16.applyTo(TensorBuffers.toShorts(tensorHandle.nativeHandle()));
+    return new TFloat16Impl(tensorHandle, buffer, shape);
+  }
+
+  private static final class TFloat16Impl extends FloatDenseNdArray implements TFloat16 {
+
+    @Override
+    public Class<TFloat16> type() {
+      return TFloat16.class;
+    }
+
+    @Override
+    public TensorHandle tensorHandle() {
+      return tensorHandle;
+    }
+
+    TFloat16Impl(TensorHandle tensorHandle, FloatDataBuffer buffer, Shape shape) {
+      super(buffer, shape);
+      this.tensorHandle = tensorHandle;
+    }
+
+    private final TensorHandle tensorHandle;
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/types/TFloat32Factory.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/types/TFloat32Factory.java
new file mode 100644
index 00000000000..33365bcfe31
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/types/TFloat32Factory.java
@@ -0,0 +1,41 @@
+package org.tensorflow.internal.types;
+
+import org.tensorflow.TensorHandle;
+import org.tensorflow.internal.buffer.TensorBuffers;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.ndarray.buffer.FloatDataBuffer;
+import org.tensorflow.ndarray.impl.dense.FloatDenseNdArray;
+import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.TypeFactory;
+
+/**
+ * Factory of {@link TFloat32} tensor instances
+ */
+public class TFloat32Factory implements TypeFactory<TFloat32> {
+
+  @Override
+  public TFloat32 createDense(TensorHandle tensorHandle, Shape shape) {
+    FloatDataBuffer buffer = TensorBuffers.toFloats(tensorHandle.nativeHandle());
+    return new TFloat32Impl(tensorHandle, buffer, shape);
+  }
+
+  private static final class TFloat32Impl extends FloatDenseNdArray implements TFloat32 {
+
+    @Override
+    public Class<TFloat32> type() {
+      return TFloat32.class;
+    }
+
+    @Override
+    public TensorHandle tensorHandle() {
+      return tensorHandle;
+    }
+
+    TFloat32Impl(TensorHandle tensorHandle, FloatDataBuffer buffer, Shape shape) {
+      super(buffer, shape);
+      this.tensorHandle = tensorHandle;
+    }
+
+    private final TensorHandle tensorHandle;
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/types/TFloat64Factory.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/types/TFloat64Factory.java
new file mode 100644
index 00000000000..1223409b799
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/types/TFloat64Factory.java
@@ -0,0 +1,41 @@
+package org.tensorflow.internal.types;
+
+import org.tensorflow.TensorHandle;
+import org.tensorflow.internal.buffer.TensorBuffers;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.ndarray.buffer.DoubleDataBuffer;
+import org.tensorflow.ndarray.impl.dense.DoubleDenseNdArray;
+import org.tensorflow.types.TFloat64;
+import org.tensorflow.types.TypeFactory;
+
+/**
+ * Factory of {@link TFloat64} tensor instances
+ */
+public class TFloat64Factory implements TypeFactory<TFloat64> {
+
+  @Override
+  public TFloat64 createDense(TensorHandle tensorHandle, Shape shape) {
+    DoubleDataBuffer buffer = TensorBuffers.toDoubles(tensorHandle.nativeHandle());
+    return new TFloat64Impl(tensorHandle, buffer, shape);
+  }
+
+  private static final class TFloat64Impl extends DoubleDenseNdArray implements TFloat64 {
+
+    @Override
+    public Class<TFloat64> type() {
+      return TFloat64.class;
+    }
+
+    @Override
+    public TensorHandle tensorHandle() {
+      return tensorHandle;
+    }
+
+    TFloat64Impl(TensorHandle tensorHandle, DoubleDataBuffer buffer, Shape shape) {
+      super(buffer, shape);
+      this.tensorHandle = tensorHandle;
+    }
+
+    private final TensorHandle tensorHandle;
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/types/TInt32Factory.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/types/TInt32Factory.java
new file mode 100644
index 00000000000..7869befa03f
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/types/TInt32Factory.java
@@ -0,0 +1,41 @@
+package org.tensorflow.internal.types;
+
+import org.tensorflow.TensorHandle;
+import org.tensorflow.internal.buffer.TensorBuffers;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.ndarray.buffer.IntDataBuffer;
+import org.tensorflow.ndarray.impl.dense.IntDenseNdArray;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.TypeFactory;
+
+/**
+ * Factory of {@link TInt32} tensor instances
+ */
+public class TInt32Factory implements TypeFactory<TInt32> {
+
+  @Override
+  public TInt32 createDense(TensorHandle tensorHandle, Shape shape) {
+    IntDataBuffer buffer = TensorBuffers.toInts(tensorHandle.nativeHandle());
+    return new TInt32Impl(tensorHandle, buffer, shape);
+  }
+
+  private static final class TInt32Impl extends IntDenseNdArray implements TInt32 {
+
+    @Override
+    public Class<TInt32> type() {
+      return TInt32.class;
+    }
+
+    @Override
+    public TensorHandle tensorHandle() {
+      return tensorHandle;
+    }
+
+    TInt32Impl(TensorHandle tensorHandle, IntDataBuffer buffer, Shape shape) {
+      super(buffer, shape);
+      this.tensorHandle = tensorHandle;
+    }
+
+    private final TensorHandle tensorHandle;
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/types/TInt64Factory.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/types/TInt64Factory.java
new file mode 100644
index 00000000000..cc37cada12d
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/types/TInt64Factory.java
@@ -0,0 +1,41 @@
+package org.tensorflow.internal.types;
+
+import org.tensorflow.TensorHandle;
+import org.tensorflow.internal.buffer.TensorBuffers;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.ndarray.buffer.LongDataBuffer;
+import org.tensorflow.ndarray.impl.dense.LongDenseNdArray;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.TypeFactory;
+
+/**
+ * Factory of {@link TInt64} tensor instances
+ */
+public class TInt64Factory implements TypeFactory<TInt64> {
+
+  @Override
+  public TInt64 createDense(TensorHandle tensorHandle, Shape shape) {
+    LongDataBuffer buffer = TensorBuffers.toLongs(tensorHandle.nativeHandle());
+    return new Int64DenseTensor(tensorHandle, buffer, shape);
+  }
+
+  private static final class Int64DenseTensor extends LongDenseNdArray implements TInt64 {
+
+    @Override
+    public Class<TInt64> type() {
+      return TInt64.class;
+    }
+
+    @Override
+    public TensorHandle tensorHandle() {
+      return tensorHandle;
+    }
+
+    Int64DenseTensor(TensorHandle tensorHandle, LongDataBuffer buffer, Shape shape) {
+      super(buffer, shape);
+      this.tensorHandle = tensorHandle;
+    }
+
+    private final TensorHandle tensorHandle;
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/types/TStringFactory.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/types/TStringFactory.java
new file mode 100644
index 00000000000..69b764fbaa3
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/types/TStringFactory.java
@@ -0,0 +1,78 @@
+package org.tensorflow.internal.types;
+
+import java.nio.charset.Charset;
+import java.nio.charset.StandardCharsets;
+import java.util.function.Function;
+import org.tensorflow.TensorHandle;
+import org.tensorflow.internal.buffer.ByteSequenceTensorBuffer;
+import org.tensorflow.internal.buffer.TensorBuffers;
+import org.tensorflow.ndarray.NdArray;
+import org.tensorflow.ndarray.NdArrays;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.ndarray.buffer.DataBuffer;
+import org.tensorflow.ndarray.buffer.layout.DataLayout;
+import org.tensorflow.ndarray.buffer.layout.DataLayouts;
+import org.tensorflow.ndarray.impl.dense.DenseNdArray;
+import org.tensorflow.types.TString;
+import org.tensorflow.types.TypeFactory;
+
+/**
+ * Factory of {@link TString} tensor instances
+ */
+public class TStringFactory implements TypeFactory<TString> {
+
+  @Override
+  public TString createDense(TensorHandle tensorHandle, Shape shape) {
+    ByteSequenceTensorBuffer buffer = TensorBuffers.toStrings(tensorHandle.nativeHandle(), shape.size());
+    return new TStringImpl(tensorHandle, buffer, shape);
+  }
+
+  private static final DataLayout<DataBuffer<byte[]>, String> UTF_8_LAYOUT =
+      DataLayouts.ofStrings(StandardCharsets.UTF_8);
+
+  private static final class TStringImpl extends DenseNdArray<String> implements TString {
+
+    @Override
+    public NdArray<byte[]> asBytes() {
+      return NdArrays.wrap(shape(), rawBuffer);
+    }
+
+    @Override
+    public Class<TString> type() {
+      return TString.class;
+    }
+
+    @Override
+    public TensorHandle tensorHandle() {
+      return tensorHandle;
+    }
+
+    @Override
+    public TString using(Charset charset) {
+      return new TStringImpl(tensorHandle, rawBuffer, shape(), DataLayouts.ofStrings(charset));
+    }
+
+    @Override
+    public <T> void write(NdArray<T> src, Function<T, byte[]> getBytes) {
+      rawBuffer.init(src, getBytes);
+    }
+
+    TStringImpl(TensorHandle tensorHandle, ByteSequenceTensorBuffer rawBuffer, Shape shape) {
+      this(tensorHandle, rawBuffer, shape, UTF_8_LAYOUT);
+    }
+
+    private TStringImpl(
+        TensorHandle tensorHandle,
+        ByteSequenceTensorBuffer rawBuffer,
+        Shape shape,
+        DataLayout<DataBuffer<byte[]>, String> layout
+    ) {
+      super(layout.applyTo(rawBuffer), shape);
+      this.rawBuffer = rawBuffer;
+      this.tensorHandle = tensorHandle;
+    }
+
+    private final TensorHandle tensorHandle;
+    private final ByteSequenceTensorBuffer rawBuffer;
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/types/TUint8Factory.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/types/TUint8Factory.java
new file mode 100644
index 00000000000..e9a95357d29
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/types/TUint8Factory.java
@@ -0,0 +1,42 @@
+package org.tensorflow.internal.types;
+
+import org.tensorflow.TensorHandle;
+import org.tensorflow.internal.c_api.TF_Tensor;
+import org.tensorflow.internal.buffer.TensorBuffers;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.ndarray.buffer.ByteDataBuffer;
+import org.tensorflow.ndarray.impl.dense.ByteDenseNdArray;
+import org.tensorflow.types.TUint8;
+import org.tensorflow.types.TypeFactory;
+
+/**
+ * Factory of {@link TUint8} tensor instances
+ */
+public class TUint8Factory implements TypeFactory<TUint8> {
+
+  @Override
+  public TUint8 createDense(TensorHandle tensorHandle, Shape shape) {
+    ByteDataBuffer buffer = TensorBuffers.toBytes(tensorHandle.nativeHandle());
+    return new TUint8Impl(tensorHandle, buffer, shape);
+  }
+
+  private static final class TUint8Impl extends ByteDenseNdArray implements TUint8 {
+
+    @Override
+    public Class<TUint8> type() {
+      return TUint8.class;
+    }
+
+    @Override
+    public TensorHandle tensorHandle() {
+      return tensorHandle;
+    }
+
+    TUint8Impl(TensorHandle tensorHandle, ByteDataBuffer buffer, Shape shape) {
+      super(buffer, shape);
+      this.tensorHandle = tensorHandle;
+    }
+
+    private final TensorHandle tensorHandle;
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/op/core/Constant.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/op/core/Constant.java
index 67b08107b95..07aebcb5bf3 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/op/core/Constant.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/op/core/Constant.java
@@ -19,7 +19,6 @@
 import org.tensorflow.Operand;
 import org.tensorflow.Operation;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.Tensors;
 import org.tensorflow.ndarray.BooleanNdArray;
 import org.tensorflow.ndarray.ByteNdArray;
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/op/core/Gradients.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/op/core/Gradients.java
index bda5f97cd96..88b10be5017 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/op/core/Gradients.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/op/core/Gradients.java
@@ -22,7 +22,6 @@
 import org.tensorflow.Graph;
 import org.tensorflow.Operand;
 import org.tensorflow.Output;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.Operands;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/op/core/Shapes.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/op/core/Shapes.java
index 46731e09ea1..a85fdbc1ab2 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/op/core/Shapes.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/op/core/Shapes.java
@@ -16,7 +16,6 @@
 
 import java.util.Arrays;
 import org.tensorflow.Operand;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/op/nn/SigmoidCrossEntropyWithLogits.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/op/nn/SigmoidCrossEntropyWithLogits.java
index cecad2e6fa3..191e4b3c122 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/op/nn/SigmoidCrossEntropyWithLogits.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/op/nn/SigmoidCrossEntropyWithLogits.java
@@ -8,7 +8,13 @@
 import org.tensorflow.op.core.Select;
 import org.tensorflow.op.core.ZerosLike;
 import org.tensorflow.op.dtypes.Cast;
-import org.tensorflow.op.math.*;
+import org.tensorflow.op.math.Add;
+import org.tensorflow.op.math.Exp;
+import org.tensorflow.op.math.GreaterEqual;
+import org.tensorflow.op.math.Log1p;
+import org.tensorflow.op.math.Mul;
+import org.tensorflow.op.math.Neg;
+import org.tensorflow.op.math.Sub;
 import org.tensorflow.types.TBool;
 import org.tensorflow.types.family.TNumber;
 
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/op/nn/SoftmaxCrossEntropyWithLogits.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/op/nn/SoftmaxCrossEntropyWithLogits.java
index 7d5f80d8ba4..d958c1272b1 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/op/nn/SoftmaxCrossEntropyWithLogits.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/op/nn/SoftmaxCrossEntropyWithLogits.java
@@ -1,11 +1,18 @@
 package org.tensorflow.op.nn;
 
+import java.util.Arrays;
+import java.util.List;
 import org.tensorflow.Operand;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.Scope;
 import org.tensorflow.op.annotation.Endpoint;
 import org.tensorflow.op.annotation.Operator;
-import org.tensorflow.op.core.*;
+import org.tensorflow.op.core.Concat;
+import org.tensorflow.op.core.Constant;
+import org.tensorflow.op.core.Range;
+import org.tensorflow.op.core.Rank;
+import org.tensorflow.op.core.Reshape;
+import org.tensorflow.op.core.Slice;
 import org.tensorflow.op.dtypes.Cast;
 import org.tensorflow.op.linalg.Transpose;
 import org.tensorflow.op.math.Sub;
@@ -14,10 +21,6 @@
 import org.tensorflow.types.TFloat32;
 import org.tensorflow.types.TInt64;
 import org.tensorflow.types.family.TNumber;
-import org.tensorflow.types.family.TType;
-
-import java.util.Arrays;
-import java.util.List;
 
 @Operator(group = "nn")
 public class SoftmaxCrossEntropyWithLogits {
@@ -192,7 +195,7 @@ private static <T extends TNumber> Operand<T> flattenOuterDims(Scope scope, Oper
    */
   private static <T extends TNumber, U extends TNumber> Operand<T> moveDimToEnd(
       Scope scope, Operand<T> input, int dimIndex, Operand<U> rank) {
-    Class<? extends TNumber> rankDType = rank.asOutput().type();
+    Class<U> rankDType = rank.asOutput().type();
     Operand one = Cast.create(scope, Constant.scalarOf(scope, 1), rankDType);
     List<Operand<U>> concatList =
         Arrays.asList(
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/op/nn/SparseSoftmaxCrossEntropyWithLogits.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/op/nn/SparseSoftmaxCrossEntropyWithLogits.java
index d99e2c1c00b..79840827c13 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/op/nn/SparseSoftmaxCrossEntropyWithLogits.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/op/nn/SparseSoftmaxCrossEntropyWithLogits.java
@@ -1,5 +1,8 @@
 package org.tensorflow.op.nn;
 
+import java.util.ArrayList;
+import java.util.Collections;
+import java.util.List;
 import org.tensorflow.Operand;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.Op;
@@ -18,10 +21,6 @@
 import org.tensorflow.types.TInt32;
 import org.tensorflow.types.family.TNumber;
 
-import java.util.ArrayList;
-import java.util.Collections;
-import java.util.List;
-
 @Operator(group = "nn")
 public class SparseSoftmaxCrossEntropyWithLogits {
 
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TBfloat16.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TBfloat16.java
index bdee8e2faaf..e9570a780c4 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TBfloat16.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TBfloat16.java
@@ -20,17 +20,14 @@
 import java.util.function.Consumer;
 import org.tensorflow.Tensors;
 import org.tensorflow.exceptions.TensorFlowException;
-import org.tensorflow.internal.c_api.TF_Tensor;
-import org.tensorflow.internal.tensor.FloatTensorImpl;
-import org.tensorflow.internal.tensor.buffer.TensorBuffers;
+import org.tensorflow.internal.types.TBfloat16Factory;
+import org.tensorflow.ndarray.FloatNdArray;
 import org.tensorflow.ndarray.NdArray;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.ndarray.StdArrays;
 import org.tensorflow.ndarray.buffer.FloatDataBuffer;
-import org.tensorflow.ndarray.buffer.layout.DataLayouts;
 import org.tensorflow.proto.framework.DataType;
 import org.tensorflow.types.annotation.TensorType;
-import org.tensorflow.types.tensor.FloatTensor;
 import org.tensorflow.types.family.TFloating;
 
 /**
@@ -49,8 +46,8 @@
  * <p>Note that some CPUs support the bfloat16 format natively, which can result in faster
  * computation compared to {@link TFloat16} when GPUs are not used.
  */
-@TensorType(dataType = DataType.DT_BFLOAT16, byteSize = 2, impl = TBfloat16Impl.class)
-public interface TBfloat16 extends FloatTensor, TFloating<TBfloat16, Float> {
+@TensorType(dataType = DataType.DT_BFLOAT16, byteSize = 2, factory = TBfloat16Factory.class)
+public interface TBfloat16 extends TFloating<Float>, FloatNdArray {
 
   /**
    * Allocates a new tensor for storing a single float value.
@@ -121,22 +118,3 @@ static TBfloat16 tensorOf(Shape shape, Consumer<TBfloat16> tensorInit) {
   }
 }
 
-/**
- * Hidden implementation of a {@code TBfloat16}
- */
-class TBfloat16Impl extends FloatTensorImpl implements TBfloat16 {
-
-  TBfloat16Impl(TF_Tensor nativeTensorHandle, Shape shape) {
-    super(nativeTensorHandle, shape, mapMemory(nativeTensorHandle));
-  }
-
-  private static FloatDataBuffer mapMemory(TF_Tensor nativeTensorHandle) {
-    return DataLayouts.BFLOAT16.applyTo(TensorBuffers.toShorts(nativeTensorHandle));
-  }
-
-  @Override
-  public Class<TBfloat16> type() {
-    return TBfloat16.class;
-  }
-}
-
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TBool.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TBool.java
index a15719f713b..d038c291766 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TBool.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TBool.java
@@ -20,9 +20,8 @@
 import java.util.function.Consumer;
 import org.tensorflow.Tensors;
 import org.tensorflow.exceptions.TensorFlowException;
-import org.tensorflow.internal.c_api.TF_Tensor;
-import org.tensorflow.internal.tensor.BooleanTensorImpl;
-import org.tensorflow.internal.tensor.buffer.TensorBuffers;
+import org.tensorflow.internal.types.TBoolFactory;
+import org.tensorflow.ndarray.BooleanNdArray;
 import org.tensorflow.ndarray.NdArray;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.ndarray.StdArrays;
@@ -30,7 +29,6 @@
 import org.tensorflow.ndarray.buffer.layout.DataLayouts;
 import org.tensorflow.proto.framework.DataType;
 import org.tensorflow.types.annotation.TensorType;
-import org.tensorflow.types.tensor.BooleanTensor;
 import org.tensorflow.types.family.TType;
 
 /**
@@ -40,8 +38,8 @@
  * explicit mapping between Java boolean values and byte buffers using the {@link DataLayouts#BOOL
  * BOOL} layout, which may impact I/O performances.
  */
-@TensorType(dataType = DataType.DT_BOOL, byteSize = 1, impl = TBoolImpl.class)
-public interface TBool extends BooleanTensor, TType<TBool, Boolean> {
+@TensorType(dataType = DataType.DT_BOOL, byteSize = 1, factory = TBoolFactory.class)
+public interface TBool extends TType<Boolean>, BooleanNdArray {
 
   /**
    * Allocates a new tensor for storing a single boolean value.
@@ -110,19 +108,5 @@ static TBool tensorOf(Shape shape, BooleanDataBuffer data) {
   static TBool tensorOf(Shape shape, Consumer<TBool> tensorInit) {
     return Tensors.of(TBool.class, shape, tensorInit);
   }
-}
-
-/**
- * Hidden implementation of a {@code TBool}
- */
-class TBoolImpl extends BooleanTensorImpl implements TBool {
 
-  TBoolImpl(TF_Tensor nativeTensor, Shape shape) {
-    super(nativeTensor, shape, TensorBuffers.toBooleans(nativeTensor));
-  }
-
-  @Override
-  public Class<TBool> type() {
-    return TBool.class;
-  }
 }
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TFloat16.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TFloat16.java
index 0cc483e4724..482704e6823 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TFloat16.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TFloat16.java
@@ -20,17 +20,14 @@
 import java.util.function.Consumer;
 import org.tensorflow.Tensors;
 import org.tensorflow.exceptions.TensorFlowException;
-import org.tensorflow.internal.c_api.TF_Tensor;
-import org.tensorflow.internal.tensor.FloatTensorImpl;
-import org.tensorflow.internal.tensor.buffer.TensorBuffers;
+import org.tensorflow.internal.types.TFloat16Factory;
+import org.tensorflow.ndarray.FloatNdArray;
 import org.tensorflow.ndarray.NdArray;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.ndarray.StdArrays;
 import org.tensorflow.ndarray.buffer.FloatDataBuffer;
-import org.tensorflow.ndarray.buffer.layout.DataLayouts;
 import org.tensorflow.proto.framework.DataType;
 import org.tensorflow.types.annotation.TensorType;
-import org.tensorflow.types.tensor.FloatTensor;
 import org.tensorflow.types.family.TFloating;
 
 /**
@@ -46,8 +43,8 @@
  * most CPUs do not support this format natively. For CPU computation on 16-bit floats, the {@link
  * TBfloat16} tensor type might be a better option.
  */
-@TensorType(dataType = DataType.DT_HALF, byteSize = 2, impl = TFloat16Impl.class)
-public interface TFloat16 extends FloatTensor, TFloating<TFloat16, Float> {
+@TensorType(dataType = DataType.DT_HALF, byteSize = 2, factory = TFloat16Factory.class)
+public interface TFloat16 extends TFloating<Float>, FloatNdArray {
 
   /**
    * Allocates a new tensor for storing a single float value.
@@ -116,24 +113,5 @@ static TFloat16 tensorOf(Shape shape, FloatDataBuffer data) {
   static TFloat16 tensorOf(Shape shape, Consumer<TFloat16> tensorInit) {
     return Tensors.of(TFloat16.class, shape, tensorInit);
   }
-}
-
-/**
- * Hidden implementation of a {@code TFloat16}
- */
-class TFloat16Impl extends FloatTensorImpl implements TFloat16 {
-
-  TFloat16Impl(TF_Tensor nativeTensor, Shape shape) {
-    super(nativeTensor, shape, mapMemory(nativeTensor));
-  }
-
-  private static FloatDataBuffer mapMemory(TF_Tensor nativeTensor) {
-    return DataLayouts.FLOAT16.applyTo(TensorBuffers.toShorts(nativeTensor));
-  }
 
-  @Override
-  public Class<TFloat16> type() {
-    return TFloat16.class;
-  }
 }
-
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TFloat32.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TFloat32.java
index da31526b71d..02a527fd364 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TFloat32.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TFloat32.java
@@ -20,23 +20,21 @@
 import java.util.function.Consumer;
 import org.tensorflow.Tensors;
 import org.tensorflow.exceptions.TensorFlowException;
-import org.tensorflow.internal.c_api.TF_Tensor;
-import org.tensorflow.internal.tensor.FloatTensorImpl;
-import org.tensorflow.internal.tensor.buffer.TensorBuffers;
+import org.tensorflow.internal.types.TFloat32Factory;
+import org.tensorflow.ndarray.FloatNdArray;
 import org.tensorflow.ndarray.NdArray;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.ndarray.StdArrays;
 import org.tensorflow.ndarray.buffer.FloatDataBuffer;
 import org.tensorflow.proto.framework.DataType;
 import org.tensorflow.types.annotation.TensorType;
-import org.tensorflow.types.tensor.FloatTensor;
 import org.tensorflow.types.family.TFloating;
 
 /**
  * IEEE-754 single-precision 32-bit float tensor type.
  */
-@TensorType(dataType = DataType.DT_FLOAT, byteSize = 4, impl = TFloat32Impl.class)
-public interface TFloat32 extends FloatTensor, TFloating<TFloat32, Float> {
+@TensorType(dataType = DataType.DT_FLOAT, byteSize = 4, factory = TFloat32Factory.class)
+public interface TFloat32 extends TFloating<Float>, FloatNdArray {
 
   /**
    * Allocates a new tensor for storing a single float value.
@@ -105,20 +103,5 @@ static TFloat32 tensorOf(Shape shape, FloatDataBuffer data) {
   static TFloat32 tensorOf(Shape shape, Consumer<TFloat32> tensorInit) {
     return Tensors.of(TFloat32.class, shape, tensorInit);
   }
-}
-
-/**
- * Hidden implementation of a {@code TFloat32}
- */
-class TFloat32Impl extends FloatTensorImpl implements TFloat32 {
 
-  TFloat32Impl(TF_Tensor nativeTensor, Shape shape) {
-    super(nativeTensor, shape, TensorBuffers.toFloats(nativeTensor));
-  }
-
-  @Override
-  public Class<TFloat32> type() {
-    return TFloat32.class;
-  }
 }
-
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TFloat64.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TFloat64.java
index 8c53ca2dfbf..53b7a2732a9 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TFloat64.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TFloat64.java
@@ -20,23 +20,21 @@
 import java.util.function.Consumer;
 import org.tensorflow.Tensors;
 import org.tensorflow.exceptions.TensorFlowException;
-import org.tensorflow.internal.c_api.TF_Tensor;
-import org.tensorflow.internal.tensor.DoubleTensorImpl;
-import org.tensorflow.internal.tensor.buffer.TensorBuffers;
+import org.tensorflow.internal.types.TFloat64Factory;
+import org.tensorflow.ndarray.DoubleNdArray;
 import org.tensorflow.ndarray.NdArray;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.ndarray.StdArrays;
 import org.tensorflow.ndarray.buffer.DoubleDataBuffer;
 import org.tensorflow.proto.framework.DataType;
 import org.tensorflow.types.annotation.TensorType;
-import org.tensorflow.types.tensor.DoubleTensor;
 import org.tensorflow.types.family.TFloating;
 
 /**
  * IEEE-754 double-precision 64-bit float tensor type.
  */
-@TensorType(dataType = DataType.DT_DOUBLE, byteSize = 8, impl = TFloat64Impl.class)
-public interface TFloat64 extends DoubleTensor, TFloating<TFloat64, Double> {
+@TensorType(dataType = DataType.DT_DOUBLE, byteSize = 8, factory = TFloat64Factory.class)
+public interface TFloat64 extends TFloating<Double>, DoubleNdArray {
 
   /**
    * Allocates a new tensor for storing a single double value.
@@ -106,18 +104,3 @@ static TFloat64 tensorOf(Shape shape, Consumer<TFloat64> tensorInit) {
     return Tensors.of(TFloat64.class, shape, tensorInit);
   }
 }
-
-/**
- * Hidden implementation of a {@code TFloat64}
- */
-class TFloat64Impl extends DoubleTensorImpl implements TFloat64 {
-
-  TFloat64Impl(TF_Tensor nativeTensor, Shape shape) {
-    super(nativeTensor, shape, TensorBuffers.toDoubles(nativeTensor));
-  }
-
-  @Override
-  public Class<TFloat64> type() {
-    return TFloat64.class;
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TInt32.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TInt32.java
index 657cc134c81..7a77e3f08c4 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TInt32.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TInt32.java
@@ -19,23 +19,21 @@
 
 import java.util.function.Consumer;
 import org.tensorflow.Tensors;
-import org.tensorflow.internal.c_api.TF_Tensor;
-import org.tensorflow.internal.tensor.IntTensorImpl;
-import org.tensorflow.internal.tensor.buffer.TensorBuffers;
+import org.tensorflow.internal.types.TInt32Factory;
+import org.tensorflow.ndarray.IntNdArray;
 import org.tensorflow.ndarray.NdArray;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.ndarray.StdArrays;
 import org.tensorflow.ndarray.buffer.IntDataBuffer;
 import org.tensorflow.proto.framework.DataType;
 import org.tensorflow.types.annotation.TensorType;
-import org.tensorflow.types.tensor.IntTensor;
 import org.tensorflow.types.family.TNumber;
 
 /**
  * 32-bit signed integer tensor type.
  */
-@TensorType(dataType = DataType.DT_INT32, byteSize = 4, impl = TInt32Impl.class)
-public interface TInt32 extends IntTensor, TNumber<TInt32, Integer> {
+@TensorType(dataType = DataType.DT_INT32, byteSize = 4, factory = TInt32Factory.class)
+public interface TInt32 extends TNumber<Integer>, IntNdArray {
 
   /**
    * Allocates a new tensor for storing a single int value.
@@ -105,18 +103,3 @@ static TInt32 tensorOf(Shape shape, Consumer<TInt32> tensorInit) {
     return Tensors.of(TInt32.class, shape, tensorInit);
   }
 }
-
-/**
- * Hidden implementation of a {@code TInt32}
- */
-class TInt32Impl extends IntTensorImpl implements TInt32 {
-
-  TInt32Impl(TF_Tensor nativeTensor, Shape shape) {
-    super(nativeTensor, shape, TensorBuffers.toInts(nativeTensor));
-  }
-
-  @Override
-  public Class<TInt32> type() {
-    return TInt32.class;
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TInt64.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TInt64.java
index 5758eb7f4df..2140c920195 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TInt64.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TInt64.java
@@ -20,23 +20,21 @@
 import java.util.function.Consumer;
 import org.tensorflow.Tensors;
 import org.tensorflow.exceptions.TensorFlowException;
-import org.tensorflow.internal.c_api.TF_Tensor;
-import org.tensorflow.internal.tensor.LongTensorImpl;
-import org.tensorflow.internal.tensor.buffer.TensorBuffers;
+import org.tensorflow.internal.types.TInt64Factory;
+import org.tensorflow.ndarray.LongNdArray;
 import org.tensorflow.ndarray.NdArray;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.ndarray.StdArrays;
 import org.tensorflow.ndarray.buffer.LongDataBuffer;
 import org.tensorflow.proto.framework.DataType;
 import org.tensorflow.types.annotation.TensorType;
-import org.tensorflow.types.tensor.LongTensor;
 import org.tensorflow.types.family.TNumber;
 
 /**
  * 64-bit signed integer tensor type.
  */
-@TensorType(dataType = DataType.DT_INT64, byteSize = 8, impl = TInt64Impl.class)
-public interface TInt64 extends LongTensor, TNumber<TInt64, Long> {
+@TensorType(dataType = DataType.DT_INT64, byteSize = 8, factory = TInt64Factory.class)
+public interface TInt64 extends TNumber<Long>, LongNdArray {
 
   /**
    * Allocates a new tensor for storing a single long value.
@@ -106,18 +104,3 @@ static TInt64 tensorOf(Shape shape, Consumer<TInt64> tensorInit) {
     return Tensors.of(TInt64.class, shape, tensorInit);
   }
 }
-
-/**
- * Hidden implementation of a {@code TInt64}
- */
-class TInt64Impl extends LongTensorImpl implements TInt64 {
-
-  TInt64Impl(TF_Tensor nativeTensor, Shape shape) {
-    super(nativeTensor, shape, TensorBuffers.toLongs(nativeTensor));
-  }
-
-  @Override
-  public Class<TInt64> type() {
-    return TInt64.class;
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TString.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TString.java
index 44cce1363de..667ce36d526 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TString.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TString.java
@@ -21,19 +21,14 @@
 import java.nio.charset.StandardCharsets;
 import java.util.function.Function;
 import org.tensorflow.Tensors;
-import org.tensorflow.internal.c_api.TF_Tensor;
-import org.tensorflow.internal.tensor.StringTensorImpl;
-import org.tensorflow.internal.tensor.buffer.ByteSequenceTensorBuffer;
-import org.tensorflow.internal.tensor.buffer.TensorBuffers;
+import org.tensorflow.internal.buffer.ByteSequenceTensorBuffer;
+import org.tensorflow.internal.types.TStringFactory;
 import org.tensorflow.ndarray.NdArray;
 import org.tensorflow.ndarray.NdArrays;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.ndarray.buffer.DataBuffer;
-import org.tensorflow.ndarray.buffer.layout.DataLayout;
-import org.tensorflow.ndarray.buffer.layout.DataLayouts;
 import org.tensorflow.proto.framework.DataType;
 import org.tensorflow.types.annotation.TensorType;
-import org.tensorflow.types.tensor.StringTensor;
 import org.tensorflow.types.family.TType;
 
 /**
@@ -45,8 +40,8 @@
  * its values initially, so TensorFlow can compute and allocate the right amount of memory. Then the
  * data in the tensor is initialized once and cannot be modified afterwards.
  */
-@TensorType(dataType = DataType.DT_STRING, byteSize = -1, impl = TStringImpl.class)
-public interface TString extends StringTensor, TType<TString, String> {
+@TensorType(dataType = DataType.DT_STRING, byteSize = -1, factory = TStringFactory.class)
+public interface TString extends TType<String> {
 
   /**
    * Allocates a new tensor for storing a string scalar.
@@ -110,7 +105,9 @@ static TString tensorOf(NdArray<String> src) {
    * @return the new tensor
    */
   static TString tensorOf(Charset charset, NdArray<String> src) {
-    return TStringImpl.createTensor(src, s -> s.getBytes(charset));
+    Function<String, byte[]> getBytes = s -> s.getBytes(charset);
+    long size = ByteSequenceTensorBuffer.computeSize(src, getBytes);
+    return Tensors.of(TString.class, src.shape(), size, t -> t.write(src, getBytes));
   }
 
   /**
@@ -170,7 +167,9 @@ static TString tensorOf(Charset charset, Shape shape, DataBuffer<String> data) {
    * @return the new tensor
    */
   static TString tensorOfBytes(NdArray<byte[]> src) {
-    return TStringImpl.createTensor(src, Function.identity());
+    Function<byte[], byte[]> getBytes = Function.identity();
+    long size = ByteSequenceTensorBuffer.computeSize(src, getBytes);
+    return Tensors.of(TString.class, src.shape(), size, t -> t.write(src, getBytes));
   }
 
   /**
@@ -210,38 +209,8 @@ static TString tensorOfBytes(Shape shape, DataBuffer<byte[]> data) {
    * @return string tensor data using this charset
    */
   TString using(Charset charset);
-}
 
-/**
- * Hidden implementation of a {@code TString}
- */
-class TStringImpl extends StringTensorImpl implements TString {
+  <T> void write(NdArray<T> src, Function<T, byte[]> getBytes);
 
-  @Override
-  public TString using(Charset charset) {
-    return new TStringImpl(nativeHandle, shape(), DataLayouts.ofStrings(charset), rawBuffer());
-  }
-
-  static <T> TString createTensor(NdArray<T> src, Function<T, byte[]> getBytes) {
-    long size = ByteSequenceTensorBuffer.computeSize(src, getBytes);
-    return Tensors.of(TString.class, src.shape(), size, t ->
-        ((TStringImpl)t).rawBuffer().init(src, getBytes)
-    );
-  }
-
-  TStringImpl(TF_Tensor nativeTensor, Shape shape) {
-    this(nativeTensor, shape, UTF_8_LAYOUT, TensorBuffers.toStrings(nativeTensor, shape.size()));
-  }
-
-  private static final DataLayout<DataBuffer<byte[]>, String> UTF_8_LAYOUT = DataLayouts.ofStrings(StandardCharsets.UTF_8);
-
-  private TStringImpl(TF_Tensor nativeTensor, Shape shape, DataLayout<DataBuffer<byte[]>, String> layout, ByteSequenceTensorBuffer rawBuffer) {
-    super(nativeTensor, shape, layout, rawBuffer);
-  }
-
-  @Override
-  public Class<TString> type() {
-    return TString.class;
-  }
+  NdArray<byte[]> asBytes();
 }
-
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TUint8.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TUint8.java
index 145f66de745..d60a6436d51 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TUint8.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TUint8.java
@@ -20,23 +20,21 @@
 import java.util.function.Consumer;
 import org.tensorflow.Tensors;
 import org.tensorflow.exceptions.TensorFlowException;
-import org.tensorflow.internal.c_api.TF_Tensor;
-import org.tensorflow.internal.tensor.ByteTensorImpl;
-import org.tensorflow.internal.tensor.buffer.TensorBuffers;
+import org.tensorflow.internal.types.TUint8Factory;
+import org.tensorflow.ndarray.ByteNdArray;
 import org.tensorflow.ndarray.NdArray;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.ndarray.StdArrays;
 import org.tensorflow.ndarray.buffer.ByteDataBuffer;
 import org.tensorflow.proto.framework.DataType;
 import org.tensorflow.types.annotation.TensorType;
-import org.tensorflow.types.tensor.ByteTensor;
 import org.tensorflow.types.family.TNumber;
 
 /**
  * 8-bit unsigned integer tensor type.
  */
-@TensorType(dataType = DataType.DT_UINT8, byteSize = 1, impl = TUint8Impl.class)
-public interface TUint8 extends ByteTensor, TNumber<TUint8, Byte> {
+@TensorType(dataType = DataType.DT_UINT8, byteSize = 1, factory = TUint8Factory.class)
+public interface TUint8 extends TNumber<Byte>, ByteNdArray {
 
   /**
    * Allocates a new tensor for storing a single byte value.
@@ -106,18 +104,3 @@ static TUint8 tensorOf(Shape shape, Consumer<TUint8> tensorInit) {
     return Tensors.of(TUint8.class, shape, tensorInit);
   }
 }
-
-/**
- * Hidden implementation of a {@code TUint8}
- */
-class TUint8Impl extends ByteTensorImpl implements TUint8 {
-
-  TUint8Impl(TF_Tensor nativeTensor, Shape shape) {
-    super(nativeTensor, shape, TensorBuffers.toBytes(nativeTensor));
-  }
-
-  @Override
-  public Class<TUint8> type() {
-    return TUint8.class;
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/Type.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/Type.java
new file mode 100644
index 00000000000..499f393d7cc
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/Type.java
@@ -0,0 +1,39 @@
+package org.tensorflow.types;
+
+import org.tensorflow.proto.framework.DataType;
+import org.tensorflow.types.family.TType;
+
+public class Type<T extends TType> {
+
+  public Class<T> typeClass() {
+    return typeClass;
+  }
+
+  public DataType dataType() {
+    return dataType;
+  }
+
+  public int byteSize() {
+    return byteSize;
+  }
+
+  public boolean isVariableLength() {
+    return byteSize < 0;
+  }
+
+  public TypeFactory<T> factory() {
+    return factory;
+  }
+
+  Type(Class<T> typeClass, DataType dataType, int byteSize, TypeFactory<T> factory) {
+    this.typeClass = typeClass;
+    this.dataType = dataType;
+    this.byteSize = byteSize;
+    this.factory = factory;
+  }
+
+  private final Class<T> typeClass;
+  private final DataType dataType;
+  private final int byteSize;
+  private final TypeFactory<T> factory;
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TypeFactory.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TypeFactory.java
new file mode 100644
index 00000000000..4529657003e
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TypeFactory.java
@@ -0,0 +1,10 @@
+package org.tensorflow.types;
+
+import org.tensorflow.TensorHandle;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.types.family.TType;
+
+public interface TypeFactory<T extends TType> {
+
+  T createDense(TensorHandle tensorHandle, Shape shape);
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/TensorTypes.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TypeRegistry.java
similarity index 53%
rename from tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/TensorTypes.java
rename to tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TypeRegistry.java
index 20345aadd0f..1e82478edf1 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/TensorTypes.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TypeRegistry.java
@@ -15,82 +15,66 @@
  *  =======================================================================
  */
 
-package org.tensorflow;
+package org.tensorflow.types;
 
-import java.lang.reflect.Constructor;
 import java.util.HashMap;
 import java.util.Map;
-import org.tensorflow.internal.c_api.TF_Tensor;
-import org.tensorflow.ndarray.Shape;
-import org.tensorflow.types.TBfloat16;
-import org.tensorflow.types.TBool;
-import org.tensorflow.types.TFloat16;
-import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.TFloat64;
-import org.tensorflow.types.TInt32;
-import org.tensorflow.types.TInt64;
-import org.tensorflow.types.TString;
-import org.tensorflow.types.TUint8;
+import org.tensorflow.proto.framework.DataType;
+import org.tensorflow.types.annotation.TensorType;
 import org.tensorflow.types.family.TType;
 
 /**
  * Utility class for working with {@link DataType} objects.
  */
-final class TensorTypes {
+public final class TypeRegistry {
 
   /**
-   * Find a type registration from the code returned by the native layer (C API).
+   * Find a type registration from a data type
    *
-   * @param nativeCode native type code
+   * @param dataType data type
    * @return type registered to this code
    * @throws IllegalArgumentException if the code matches no registered data type
    */
-  static TensorType find(int nativeCode) {
-    TensorType entry = TYPES_BY_CODE.get(nativeCode);
+  public static <T extends TType> Type<T> find(DataType dataType) {
+    Type entry = TYPES_BY_CODE.get(dataType.getNumber());
     if (entry == null) {
-      throw new IllegalArgumentException(
-          "DataType " + nativeCode + " is not recognized in Java (version " + TensorFlow.version() + ")");
+      throw new IllegalArgumentException("No type has been registered for datatype " + dataType);
     }
     return entry;
   }
 
   /**
-   * Find a type registration.
+   * Find a type registration from a type class
    *
-   * @param typeClass class implementing {@link TType}
+   * @param typeClass class implementing {@link Tensor}
    * @return type registration
    * @throws IllegalArgumentException if the code matches no registered data type
    */
-  static TensorType find(Class<? extends TType> typeClass) {
-    TensorType entry = TYPES_BY_CLASS.get(typeClass);
+  public static <T extends TType> Type<T> find(Class<T> typeClass) {
+    Type entry = TYPES_BY_CLASS.get(typeClass);
     if (entry == null) {
       throw new IllegalArgumentException("Class \"" + typeClass.getName() + "\" is not a valid datatype class");
     }
     return entry;
   }
 
-  static int numberOf(Class<? extends TType> typeClass) {
-    return find(typeClass).dataType().getNumber();
-  }
-
-  private static final Map<Integer, TensorType> TYPES_BY_CODE = new HashMap<>();
-  private static final Map<Class<? extends TType>, TensorType> TYPES_BY_CLASS = new HashMap<>();
+  private static final Map<Integer, Type> TYPES_BY_CODE = new HashMap<>();
+  private static final Map<Class<? extends TType>, Type> TYPES_BY_CLASS = new HashMap<>();
 
-  private static void register(Class<? extends TType> typeClass) {
-    org.tensorflow.types.annotation.TensorType typeAnnot =
-        typeClass.getDeclaredAnnotation(org.tensorflow.types.annotation.TensorType.class);
+  private static <T extends TType> void register(Class<T> typeClass) {
+    TensorType typeAnnot = typeClass.getDeclaredAnnotation(TensorType.class);
     if (typeAnnot == null) {
       throw new IllegalArgumentException("Class \"" + typeClass.getName() + "\" must be annotated "
           + "with @TensorType to be registered as a tensor type");
     }
-    Constructor<?> implConstructor;
+    TypeFactory<T> factory;
     try {
-      implConstructor = typeAnnot.impl().getConstructor(TF_Tensor.class, Shape.class);
-    } catch (NoSuchMethodException e) {
-      throw new IllegalArgumentException("Class \"" + typeClass.getName() + "\" must have a constructor "
-          + "accepting a native `TF_Tensor` handle and a `Shape` to be implement as a tensor type");
+      factory = typeAnnot.factory().newInstance();
+    } catch (ReflectiveOperationException e) {
+      throw new IllegalArgumentException("Class \"" + typeClass.getName() + "\" must have a public "
+          + "parameter-less constructor to be used as a tensor type factory");
     }
-    TensorType type = new TensorType(typeClass, typeAnnot.dataType(), typeAnnot.byteSize(), implConstructor);
+    Type<T> type = new Type<T>(typeClass, typeAnnot.dataType(), typeAnnot.byteSize(), factory);
     TYPES_BY_CLASS.put(typeClass, type);
 
     // If more than one tensor type is mapped to a given native code, the last registered will
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/annotation/TensorType.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/annotation/TensorType.java
index 18130ae7377..b5da7ddcb35 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/annotation/TensorType.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/annotation/TensorType.java
@@ -15,10 +15,18 @@
 
 package org.tensorflow.types.annotation;
 
-import org.tensorflow.Tensor;
+import java.lang.annotation.Documented;
+import java.lang.annotation.ElementType;
+import java.lang.annotation.Retention;
+import java.lang.annotation.RetentionPolicy;
+import java.lang.annotation.Target;
 import org.tensorflow.proto.framework.DataType;
+import org.tensorflow.types.TypeFactory;
 
 /** Represents a type of elements in a {@link Tensor} */
+@Documented
+@Retention(RetentionPolicy.RUNTIME)
+@Target(ElementType.TYPE)
 public @interface TensorType {
 
   DataType dataType();
@@ -28,5 +36,5 @@
   /**
    * The class implementing this tensor type
    */
-  Class<?> impl();
+  Class<? extends TypeFactory> factory();
 }
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/family/TFloating.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/family/TFloating.java
index 92deaffdc68..27e056ef643 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/family/TFloating.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/family/TFloating.java
@@ -16,4 +16,4 @@
  * exp.call(tf.constant(tensor2));  // Compilation failure
  * }</pre>
  */
-public interface TFloating extends TNumber {}
+public interface TFloating<T> extends TNumber<T> {}
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/family/TNumber.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/family/TNumber.java
index 4fa2b563cae..cbc9333dc72 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/family/TNumber.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/family/TNumber.java
@@ -32,4 +32,4 @@
  * tf.nn.softmax(tf.constant(tensor2));  // Compilation failure
  * }</pre>
  */
-public interface TNumber<T extends TNumber, U> extends TType<T, U> {}
+public interface TNumber<T> extends TType<T> {}
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/family/TType.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/family/TType.java
index d4adb8a7b3b..89629974502 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/family/TType.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/family/TType.java
@@ -1,65 +1,34 @@
-/*
- *  Copyright 2019 The TensorFlow Authors. All Rights Reserved.
- *
- *  Licensed under the Apache License, Version 2.0 (the "License");
- *  you may not use this file except in compliance with the License.
- *  You may obtain a copy of the License at
- *
- *      http://www.apache.org/licenses/LICENSE-2.0
- *
- *  Unless required by applicable law or agreed to in writing, software
- *  distributed under the License is distributed on an "AS IS" BASIS,
- *  WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
- *  See the License for the specific language governing permissions and
- *  limitations under the License.
- *  =======================================================================
- */
-
 package org.tensorflow.types.family;
 
-import org.tensorflow.Operand;
-import org.tensorflow.Operation;
-import org.tensorflow.Output;
 import org.tensorflow.Tensor;
-import org.tensorflow.op.Ops;
-import org.tensorflow.op.Scope;
-import org.tensorflow.op.core.Constant;
+import org.tensorflow.TensorHandle;
+import org.tensorflow.ndarray.NdArray;
+import org.tensorflow.ndarray.buffer.ByteDataBuffer;
+import org.tensorflow.proto.framework.DataType;
+
+public interface TType<T> extends NdArray<T>, Tensor {
+
+  Class<? extends TType> type();
 
-/**
- * Marker interface for all tensor types.
- *
- * <p>Tensor types are carried as a generic parameter of the {@link org.tensorflow.Tensor Tensor}
- * class bound by the {@code TType} interface. This generic parameter ensure type-compatibility
- * between operands of a computation at compile-time. For example:
- *
- * <pre>{@code
- * TFloat32 tensor1 = TFloat32.ofShape(2, 3, 2);
- * TFloat32 tensor2 = TFloat32.ofShape(2, 3, 2);
- * TInt32 tensor3 = TInt32.ofShape(2, 3, 2);
- *
- * Ops tf = Ops.create();
- * tf.math.add(tensor1, tensor2);  // OK
- * tf.math.add(tensor1, tensor3);  // Compilation failure
- * }</pre>
- */
-public interface TType<T extends TType, U> extends Tensor<U>, Operand<T> {
+  TensorHandle tensorHandle();
 
-  default Constant<T> toConstant(Scope scope) {
-    if (scope == null) {
-      throw new IllegalArgumentException("Scope is required");
-    }
-    return Constant.create(scope, (T)this);
+  @Override
+  default void close() {
+    tensorHandle().close();
   }
 
   @Override
-  default Output<T> asOutput(Scope scope) {
-    return toConstant(scope).asOutput();
+  default ByteDataBuffer rawData() {
+    return tensorHandle().rawData();
   }
 
   @Override
-  default Operation op() {
-    throw new IllegalStateException("A tensor is not attached to a specific operation");
+  default DataType dataType() {
+    return tensorHandle().dataType();
   }
 
-  Class<T> type();
+  @Override
+  default long numBytes() {
+    return tensorHandle().numBytes();
+  }
 }
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/package-info.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/package-info.java
index afbd69fabe5..b41ce514f00 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/package-info.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/package-info.java
@@ -32,10 +32,10 @@
  *
  * <p>Instances of tensor types must also implement the {@link org.tensorflow.ndarray.NdArray NdArray}
  * interface so a user can access directly the tensor data in a n-dimensional space by invoking
- * {@link org.tensorflow.Tensor#data() Tensor.data()}.
+ * {@link org.tensorflow.util.Tensor#data() Tensor.data()}.
  *
  * <p>Note that while it is always possible to allocate a tensor using the
- * {@link org.tensorflow.Tensor#of(org.tensorflow.DataType, Shape) Tensor.of(...)}
+ * {@link org.tensorflow.util.Tensor#of(org.tensorflow.DataType, Shape) Tensor.of(...)}
  * method, most tensor types expose factory methods that simplify the creation process, like
  * {@code scalarOf(...)}, {@code vectorOf(...)}, {@code tensorOf(...)}, etc.
  */
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/tensor/BooleanTensor.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/tensor/BooleanTensor.java
deleted file mode 100644
index 106e18de8f6..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/tensor/BooleanTensor.java
+++ /dev/null
@@ -1,37 +0,0 @@
-package org.tensorflow.types.tensor;
-
-import org.tensorflow.Tensor;
-import org.tensorflow.ndarray.BooleanNdArray;
-import org.tensorflow.ndarray.NdArray;
-import org.tensorflow.ndarray.buffer.BooleanDataBuffer;
-import org.tensorflow.ndarray.buffer.DataBuffer;
-
-/**
- * A {@link Tensor} of booleans.
- */
-public interface BooleanTensor extends BooleanNdArray, Tensor<Boolean> {
-
-  @Override
-  BooleanTensor setBoolean(boolean value, long... coordinates);
-
-  @Override
-  BooleanTensor set(NdArray<Boolean> src, long... coordinates);
-
-  @Override
-  BooleanTensor setObject(Boolean value, long... coordinates);
-
-  @Override
-  BooleanTensor copyTo(NdArray<Boolean> dst);
-
-  @Override
-  BooleanTensor read(DataBuffer<Boolean> dst);
-
-  @Override
-  BooleanTensor read(BooleanDataBuffer dst);
-
-  @Override
-  BooleanTensor write(DataBuffer<Boolean> src);
-
-  @Override
-  BooleanTensor write(BooleanDataBuffer src);
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/tensor/ByteTensor.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/tensor/ByteTensor.java
deleted file mode 100644
index 16b7825422d..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/tensor/ByteTensor.java
+++ /dev/null
@@ -1,37 +0,0 @@
-package org.tensorflow.types.tensor;
-
-import org.tensorflow.Tensor;
-import org.tensorflow.ndarray.ByteNdArray;
-import org.tensorflow.ndarray.NdArray;
-import org.tensorflow.ndarray.buffer.ByteDataBuffer;
-import org.tensorflow.ndarray.buffer.DataBuffer;
-
-/**
- * A {@link Tensor} of bytes.
- */
-public interface ByteTensor extends ByteNdArray, Tensor<Byte> {
-
-  @Override
-  ByteTensor setByte(byte value, long... coordinates);
-
-  @Override
-  ByteTensor set(NdArray<Byte> src, long... coordinates);
-
-  @Override
-  ByteTensor setObject(Byte value, long... coordinates);
-
-  @Override
-  ByteTensor copyTo(NdArray<Byte> dst);
-
-  @Override
-  ByteTensor read(DataBuffer<Byte> dst);
-
-  @Override
-  ByteTensor read(ByteDataBuffer dst);
-
-  @Override
-  ByteTensor write(DataBuffer<Byte> src);
-
-  @Override
-  ByteTensor write(ByteDataBuffer src);
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/tensor/DoubleTensor.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/tensor/DoubleTensor.java
deleted file mode 100644
index 5f6036fa08f..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/tensor/DoubleTensor.java
+++ /dev/null
@@ -1,37 +0,0 @@
-package org.tensorflow.types.tensor;
-
-import org.tensorflow.Tensor;
-import org.tensorflow.ndarray.DoubleNdArray;
-import org.tensorflow.ndarray.NdArray;
-import org.tensorflow.ndarray.buffer.DataBuffer;
-import org.tensorflow.ndarray.buffer.DoubleDataBuffer;
-
-/**
- * A {@link Tensor} of doubles.
- */
-public interface DoubleTensor extends DoubleNdArray, Tensor<Double> {
-
-  @Override
-  DoubleTensor setDouble(double value, long... coordinates);
-
-  @Override
-  DoubleTensor set(NdArray<Double> src, long... coordinates);
-
-  @Override
-  DoubleTensor setObject(Double value, long... coordinates);
-
-  @Override
-  DoubleTensor copyTo(NdArray<Double> dst);
-
-  @Override
-  DoubleTensor read(DataBuffer<Double> dst);
-
-  @Override
-  DoubleTensor read(DoubleDataBuffer dst);
-
-  @Override
-  DoubleTensor write(DataBuffer<Double> src);
-
-  @Override
-  DoubleTensor write(DoubleDataBuffer src);
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/tensor/FloatTensor.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/tensor/FloatTensor.java
deleted file mode 100644
index c4222d8c1d6..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/tensor/FloatTensor.java
+++ /dev/null
@@ -1,37 +0,0 @@
-package org.tensorflow.types.tensor;
-
-import org.tensorflow.Tensor;
-import org.tensorflow.ndarray.FloatNdArray;
-import org.tensorflow.ndarray.NdArray;
-import org.tensorflow.ndarray.buffer.DataBuffer;
-import org.tensorflow.ndarray.buffer.FloatDataBuffer;
-
-/**
- * A {@link Tensor} of floats.
- */
-public interface FloatTensor extends FloatNdArray, Tensor<Float> {
-
-  @Override
-  FloatTensor setFloat(float value, long... coordinates);
-
-  @Override
-  FloatTensor set(NdArray<Float> src, long... coordinates);
-
-  @Override
-  FloatTensor setObject(Float value, long... coordinates);
-
-  @Override
-  FloatTensor copyTo(NdArray<Float> dst);
-
-  @Override
-  FloatTensor read(DataBuffer<Float> dst);
-
-  @Override
-  FloatTensor read(FloatDataBuffer dst);
-
-  @Override
-  FloatTensor write(DataBuffer<Float> src);
-
-  @Override
-  FloatTensor write(FloatDataBuffer src);
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/tensor/IntTensor.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/tensor/IntTensor.java
deleted file mode 100644
index 67aab65f309..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/tensor/IntTensor.java
+++ /dev/null
@@ -1,37 +0,0 @@
-package org.tensorflow.types.tensor;
-
-import org.tensorflow.Tensor;
-import org.tensorflow.ndarray.IntNdArray;
-import org.tensorflow.ndarray.NdArray;
-import org.tensorflow.ndarray.buffer.DataBuffer;
-import org.tensorflow.ndarray.buffer.IntDataBuffer;
-
-/**
- * A {@link Tensor} of integers.
- */
-public interface IntTensor extends IntNdArray, Tensor<Integer> {
-
-  @Override
-  IntTensor setInt(int value, long... coordinates);
-
-  @Override
-  IntTensor set(NdArray<Integer> src, long... coordinates);
-
-  @Override
-  IntTensor setObject(Integer value, long... coordinates);
-
-  @Override
-  IntTensor copyTo(NdArray<Integer> dst);
-
-  @Override
-  IntTensor read(DataBuffer<Integer> dst);
-
-  @Override
-  IntTensor read(IntDataBuffer dst);
-
-  @Override
-  IntTensor write(DataBuffer<Integer> src);
-
-  @Override
-  IntTensor write(IntDataBuffer src);
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/tensor/LongTensor.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/tensor/LongTensor.java
deleted file mode 100644
index 3d708bf869e..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/tensor/LongTensor.java
+++ /dev/null
@@ -1,37 +0,0 @@
-package org.tensorflow.types.tensor;
-
-import org.tensorflow.Tensor;
-import org.tensorflow.ndarray.LongNdArray;
-import org.tensorflow.ndarray.NdArray;
-import org.tensorflow.ndarray.buffer.DataBuffer;
-import org.tensorflow.ndarray.buffer.LongDataBuffer;
-
-/**
- * A {@link Tensor} of longs.
- */
-public interface LongTensor extends LongNdArray, Tensor<Long> {
-
-  @Override
-  LongTensor setLong(long value, long... coordinates);
-
-  @Override
-  LongTensor set(NdArray<Long> src, long... coordinates);
-
-  @Override
-  LongTensor setObject(Long value, long... coordinates);
-
-  @Override
-  LongTensor copyTo(NdArray<Long> dst);
-
-  @Override
-  LongTensor read(DataBuffer<Long> dst);
-
-  @Override
-  LongTensor read(LongDataBuffer dst);
-
-  @Override
-  LongTensor write(DataBuffer<Long> src);
-
-  @Override
-  LongTensor write(LongDataBuffer src);
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/tensor/StringTensor.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/tensor/StringTensor.java
deleted file mode 100644
index 7a74276e8df..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/tensor/StringTensor.java
+++ /dev/null
@@ -1,31 +0,0 @@
-package org.tensorflow.types.tensor;
-
-import org.tensorflow.Tensor;
-import org.tensorflow.ndarray.NdArray;
-import org.tensorflow.ndarray.buffer.DataBuffer;
-
-/**
- * A {@link Tensor} of strings.
- */
-public interface StringTensor extends Tensor<String> {
-
-  /**
-   * @return the tensor data as a n-dimensional array of raw byte sequences.
-   */
-  NdArray<byte[]> asBytes();
-
-  @Override
-  StringTensor set(NdArray<String> src, long... coordinates);
-
-  @Override
-  StringTensor setObject(String value, long... coordinates);
-
-  @Override
-  StringTensor copyTo(NdArray<String> dst);
-
-  @Override
-  StringTensor read(DataBuffer<String> dst);
-
-  @Override
-  StringTensor write(DataBuffer<String> src);
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/util/TensorList.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/util/TensorList.java
new file mode 100644
index 00000000000..a1d6cd89feb
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/util/TensorList.java
@@ -0,0 +1,60 @@
+package org.tensorflow.util;
+
+import java.util.ArrayList;
+import java.util.Collection;
+import java.util.Iterator;
+import java.util.List;
+import java.util.ListIterator;
+import java.util.stream.Stream;
+import org.tensorflow.types.family.TType;
+
+public class TensorList implements AutoCloseable {
+
+  public boolean isEmpty() {
+    return tensors.isEmpty();
+  }
+
+  public int size() {
+    return tensors.size();
+  }
+
+  public Iterator<? extends TType> iterator() {
+    return tensors.iterator();
+  }
+
+  public ListIterator<? extends TType> listIterator() {
+    return tensors.listIterator();
+  }
+
+  public Stream<? extends TType> stream() {
+    return tensors.stream();
+  }
+
+  public <T extends TType> boolean add(T tensor) {
+    return tensors.add(tensor);
+  }
+
+  public <T extends TType<?>> T single() {
+    if (tensors.size() != 1) {
+      throw new IllegalStateException("List must contain a single tensor to use non-indexed getter");
+    }
+    return get(0);
+  }
+
+  public <T extends TType<?>> T get(int index) {
+    return (T)tensors.get(index);
+  }
+
+  public TensorMap toMap(Collection<String> tensorNames) {
+    return new TensorMap(tensorNames, this);
+  }
+
+  @Override
+  public void close() {
+    for (TType<?> t : tensors) {
+      t.tensorHandle().nativeHandle().releaseReference();
+    }
+  }
+
+  private final List<TType> tensors = new ArrayList<>();
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/util/TensorMap.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/util/TensorMap.java
new file mode 100644
index 00000000000..f335bb99d32
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/util/TensorMap.java
@@ -0,0 +1,73 @@
+package org.tensorflow.util;
+
+import java.util.Collection;
+import java.util.HashMap;
+import java.util.Iterator;
+import java.util.Map;
+import java.util.Set;
+import org.tensorflow.types.family.TType;
+
+public class TensorMap extends HashMap<String, TType<?>> implements AutoCloseable {
+
+  @Override
+  public TType<?> put(String key, TType<?> value) {
+    value.tensorHandle().nativeHandle().retainReference();
+    TType<?> t = super.put(key, value);
+    if (t != null) {
+      t.tensorHandle().nativeHandle().releaseReference();
+    }
+    return t;
+  }
+
+  @Override
+  public TType<?> remove(Object key) {
+    TType<?> t = super.remove(key);
+    if (t != null) {
+      t.tensorHandle().nativeHandle().releaseReference();
+    }
+    return t;
+  }
+
+  @Override
+  public void putAll(Map<? extends String, ? extends TType<?>> m) {
+    super.putAll(m);
+    for (TType<?> t : m.values()) {
+      t.tensorHandle().nativeHandle().retainReference();
+    }
+  }
+
+  @Override
+  public void clear() {
+    for (TType<?> t : values()) {
+      t.tensorHandle().nativeHandle().releaseReference();
+    }
+    super.clear();
+  }
+
+  @Override
+  public void close() {
+    clear();
+  }
+
+  public <T extends TType> T get(String name) {
+    return (T) super.get(name);
+  }
+
+  public <T extends TType<?>> T single() {
+    if (size() != 1) {
+      throw new IllegalStateException("List must contain a single tensor to use non-indexed getter");
+    }
+    return (T) values().iterator().next();
+  }
+
+  public TensorMap() {
+    super();
+  }
+
+  public TensorMap(Collection<String> names, TensorList tensors) {
+    Iterator<? extends TType> tensorIter = tensors.iterator();
+    for (String name : names) {
+      put(name, tensorIter.next());
+    }
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/AutoCloseableList.java b/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/AutoCloseableList.java
deleted file mode 100644
index 330a40bae6b..00000000000
--- a/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/AutoCloseableList.java
+++ /dev/null
@@ -1,27 +0,0 @@
-package org.tensorflow;
-
-import java.util.ArrayList;
-import java.util.Collection;
-
-public final class AutoCloseableList<E extends AutoCloseable> extends ArrayList<E>
-    implements AutoCloseable {
-
-  public AutoCloseableList(Collection<? extends E> c) {
-    super(c);
-  }
-
-  @Override
-  public void close() {
-    Exception toThrow = null;
-    for (AutoCloseable c : this) {
-      try {
-        c.close();
-      } catch (Exception e) {
-        toThrow = e;
-      }
-    }
-    if (toThrow != null) {
-      throw new RuntimeException(toThrow);
-    }
-  }
-}
diff --git a/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/ConcreteFunctionTest.java b/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/ConcreteFunctionTest.java
index 3ea20fcbb46..dd17c891e69 100644
--- a/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/ConcreteFunctionTest.java
+++ b/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/ConcreteFunctionTest.java
@@ -29,14 +29,14 @@
 public class ConcreteFunctionTest {
 
   private static Signature plusFive(Ops tf) {
-    Placeholder<TFloat32> input = tf.placeholder(TFloat32.DTYPE);
+    Placeholder<TFloat32> input = tf.placeholder(TFloat32.class);
     Add<TFloat32> output = tf.math.add(input, tf.constant(5.0f));
     Init init = tf.init();  // for native resource management tests
     return Signature.builder().key("plusFive").input("x", input).output("y", output).build();
   }
 
   private static Signature minusTwo(Ops tf) {
-    Placeholder<TFloat32> input = tf.placeholder(TFloat32.DTYPE);
+    Placeholder<TFloat32> input = tf.placeholder(TFloat32.class);
     Sub<TFloat32> output = tf.math.sub(input, tf.constant(2.0f));
     return Signature.builder().key("minusTwo").input("x", input).output("y", output).build();
   }
@@ -44,8 +44,8 @@ private static Signature minusTwo(Ops tf) {
   @Test
   public void createFunction() {
     try (ConcreteFunction f = ConcreteFunction.create(ConcreteFunctionTest::plusFive);
-        Tensor<TFloat32> x = TFloat32.scalarOf(3.0f)) {
-      assertEquals(8.0f, f.call(x).expect(TFloat32.DTYPE).data().getFloat());
+        TFloat32 x = TFloat32.scalarOf(3.0f)) {
+      assertEquals(8.0f, ((TFloat32)f.call(x)).getFloat());
     }
   }
 
@@ -54,8 +54,8 @@ public void createFunctionFromGraph() {
     try (Graph g = new Graph()) {
       Signature signature = plusFive(Ops.create(g));
       try (ConcreteFunction f = ConcreteFunction.create(signature, g);
-          Tensor<TFloat32> x = TFloat32.scalarOf(3.0f)) {
-        assertEquals(8.0f, f.call(x).expect(TFloat32.DTYPE).data().getFloat());
+          TFloat32 x = TFloat32.scalarOf(3.0f)) {
+        assertEquals(8.0f, ((TFloat32)f.call(x)).getFloat());
       }
     }
   }
@@ -66,8 +66,8 @@ public void createFunctionFromSession() {
       Signature signature = plusFive(Ops.create(g));
       try (Session s = new Session(g)) {
         try (ConcreteFunction f = ConcreteFunction.create(signature, s);
-            Tensor<TFloat32> x = TFloat32.scalarOf(3.0f)) {
-          assertEquals(8.0f, f.call(x).expect(TFloat32.DTYPE).data().getFloat());
+            TFloat32 x = TFloat32.scalarOf(3.0f)) {
+          assertEquals(8.0f, ((TFloat32)f.call(x)).getFloat());
         }
       }
     }
@@ -77,8 +77,9 @@ public void createFunctionFromSession() {
   public void chainFunctions() {
     try (ConcreteFunction f1 = ConcreteFunction.create(ConcreteFunctionTest::plusFive);
         ConcreteFunction f2 = ConcreteFunction.create(ConcreteFunctionTest::minusTwo);
-        Tensor<TFloat32> x = TFloat32.scalarOf(3.0f)) {
-      assertEquals(6.0f, f2.call(f1.call(x)).expect(TFloat32.DTYPE).data().getFloat());
+        TFloat32 x = TFloat32.scalarOf(3.0f)) {
+      TFloat32 y = f1.call(x);
+      assertEquals(6.0f, ((TFloat32)f2.call(y)).getFloat());
     }
   }
 
diff --git a/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/EagerOperationBuilderTest.java b/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/EagerOperationBuilderTest.java
index 8fb3f22cd8d..fda6084b796 100644
--- a/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/EagerOperationBuilderTest.java
+++ b/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/EagerOperationBuilderTest.java
@@ -45,7 +45,7 @@ public void failToBuildOpIfSessionIsClosed() {
       opBuilder = new EagerOperationBuilder(session, "Empty", "empty");
     }
     try {
-      opBuilder.setAttr("dtype", TFloat32.DTYPE);
+      opBuilder.setAttr("dtype", TFloat32.class);
       fail();
     } catch (IllegalStateException e) {
       // expected
@@ -95,7 +95,7 @@ public void setAttrs() {
       // dtype, tensor attributes.
       try (TInt32 t = TInt32.scalarOf(1)) {
         opBuilder(session, "Const", "DataTypeAndTensor")
-            .setAttr("dtype", TInt32.DTYPE)
+            .setAttr("dtype", TInt32.class)
             .setAttr("value", t)
             .build();
       }
@@ -103,7 +103,7 @@ public void setAttrs() {
       opBuilder(session, "RandomUniform", "DataTypeAndInt")
           .addInput(tf.array(1).asOutput())
           .setAttr("seed", 10)
-          .setAttr("dtype", TFloat32.DTYPE)
+          .setAttr("dtype", TFloat32.class)
           .build();
       // list(int), string
       opBuilder(session, "MaxPool", "IntListAndString")
@@ -124,7 +124,7 @@ public void setAttrs() {
           .build();
       // list(shape)
       opBuilder(session, "FIFOQueue", "queue")
-          .setAttr("component_types", new DataType[] {TInt32.DTYPE, TInt32.DTYPE})
+          .setAttr("component_types", new Class[] {TInt32.class, TInt32.class})
           .setAttr("shapes", new Shape[] {Shape.of(2, 2), Shape.of(2, 2, 2)})
           .build();
       // bool
diff --git a/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/EagerOperationTest.java b/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/EagerOperationTest.java
index 6f069506f5f..c702f3e9883 100644
--- a/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/EagerOperationTest.java
+++ b/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/EagerOperationTest.java
@@ -23,6 +23,7 @@
 import org.tensorflow.exceptions.TFInvalidArgumentException;
 import org.tensorflow.op.Ops;
 import org.tensorflow.ndarray.Shape;
+import org.tensorflow.proto.framework.DataType;
 import org.tensorflow.types.TFloat32;
 import org.tensorflow.types.TInt32;
 
@@ -49,10 +50,10 @@ public void outputDataTypeAndShape() {
          TInt32 t = TInt32.tensorOf(Shape.of(2, 3))) {
       EagerOperation op =
           opBuilder(session, "Const", "OutputAttrs")
-              .setAttr("dtype", TInt32.DTYPE)
+              .setAttr("dtype", TInt32.class)
               .setAttr("value", t)
               .build();
-      assertEquals(TInt32.DTYPE, op.dtype(0));
+      assertEquals(DataType.DT_INT32, op.dtype(0));
       assertEquals(2, op.shape(0).size(0));
       assertEquals(3, op.shape(0).size(1));
     }
@@ -67,12 +68,12 @@ public void outputTensor() {
               .addInput(tf.constant(2).asOutput())
               .addInput(tf.constant(4).asOutput())
               .build();
-      assertEquals(6, add.tensor(0).expect(TInt32.DTYPE).getInt());
+      assertEquals(6, ((TInt32)Tensors.fromHandle(add.tensor(0))).getInt());
 
       // Validate that we retrieve the right shape and datatype from the tensor
       // that has been resolved
       assertEquals(0, add.shape(0).numDimensions());
-      assertEquals(TInt32.DTYPE, add.dtype(0));
+      assertEquals(DataType.DT_INT32, add.dtype(0));
     }
   }
 
@@ -123,7 +124,7 @@ public void numOutputs() {
           opBuilder(session, "UniqueWithCountsV2", "unq")
               .addInput(tf.constant(new int[]{1, 2, 1}).asOutput())
               .addInput(tf.constant(new int[]{0}).asOutput())
-              .setAttr("out_idx", TInt32.DTYPE)
+              .setAttr("out_idx", TInt32.class)
               .build();
       assertEquals(3, op.numOutputs());
     }
diff --git a/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/GraphOperationBuilderTest.java b/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/GraphOperationBuilderTest.java
index 7573a25ac13..4d05dd9ad26 100644
--- a/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/GraphOperationBuilderTest.java
+++ b/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/GraphOperationBuilderTest.java
@@ -52,10 +52,10 @@ public void failOnUseAfterBuild() {
     try (Graph g = new Graph();
         TInt32 t = TInt32.scalarOf(1)) {
       OperationBuilder b =
-          g.opBuilder("Const", "Const").setAttr("dtype", t.dataType()).setAttr("value", t);
+          g.opBuilder("Const", "Const").setAttr("dtype", t.type()).setAttr("value", t);
       b.build();
       try {
-        b.setAttr("dtype", t.dataType());
+        b.setAttr("dtype", t.type());
       } catch (IllegalStateException e) {
         // expected exception.
       }
@@ -67,7 +67,7 @@ public void failOnUseAfterGraphClose() {
     OperationBuilder b = null;
     try (Graph g = new Graph();
         TInt32 t = TInt32.scalarOf(1)) {
-      b = g.opBuilder("Const", "Const").setAttr("dtype", t.dataType()).setAttr("value", t);
+      b = g.opBuilder("Const", "Const").setAttr("dtype", t.type()).setAttr("value", t);
     }
     try {
       b.build();
@@ -90,7 +90,7 @@ public void setAttr() {
       // dtype, tensor attributes.
       try (TInt32 t = TInt32.scalarOf(1)) {
         g.opBuilder("Const", "DataTypeAndTensor")
-            .setAttr("dtype", TInt32.DTYPE)
+            .setAttr("dtype", TInt32.class)
             .setAttr("value", t)
             .build()
             .output(0);
@@ -106,7 +106,7 @@ public void setAttr() {
       g.opBuilder("RandomUniform", "Int")
           .addInput(tf.array(1).asOutput())
           .setAttr("seed", 10)
-          .setAttr("dtype", TFloat32.DTYPE)
+          .setAttr("dtype", TFloat32.class)
           .build();
       assertTrue(hasNode(g, "Int"));
       // list(int)
@@ -132,23 +132,23 @@ public void setAttrShape() {
     try (Graph g = new Graph()) {
       Output<?> n =
           g.opBuilder("Placeholder", "unknown")
-              .setAttr("dtype", TFloat32.DTYPE)
+              .setAttr("dtype", TFloat32.class)
               .setAttr("shape", Shape.unknown())
               .build()
               .output(0);
       assertEquals(-1, n.shape().numDimensions());
-      assertEquals(TFloat32.DTYPE, n.dataType());
+      assertEquals(TFloat32.class, n.type());
 
       n =
           g.opBuilder("Placeholder", "batch_of_vectors")
-              .setAttr("dtype", TFloat32.DTYPE)
+              .setAttr("dtype", TFloat32.class)
               .setAttr("shape", Shape.of(-1, 784))
               .build()
               .output(0);
       assertEquals(2, n.shape().numDimensions());
       assertEquals(-1, n.shape().size(0));
       assertEquals(784, n.shape().size(1));
-      assertEquals(TFloat32.DTYPE, n.dataType());
+      assertEquals(TFloat32.class, n.type());
     }
   }
 
@@ -172,7 +172,7 @@ public void addControlInput() {
         TBool yes = TBool.scalarOf(true);
         TBool no = TBool.scalarOf(false)) {
       Ops tf = Ops.create(g);
-      Output<TBool> placeholder = tf.placeholder(TBool.DTYPE).asOutput();
+      Output<TBool> placeholder = tf.placeholder(TBool.class).asOutput();
       GraphOperation check =
           g.opBuilder("Assert", "assert")
               .addInput(placeholder)
@@ -200,7 +200,7 @@ private static void testSetAttrShapeList(Shape[] shapes) {
       int[][] matrix = new int[][] {{0, 0}, {0, 0}};
       Output<?> queue =
           g.opBuilder("FIFOQueue", "queue")
-              .setAttr("component_types", new DataType[] {TInt32.DTYPE, TInt32.DTYPE})
+              .setAttr("component_types", new Class[] {TInt32.class, TInt32.class})
               .setAttr("shapes", shapes)
               .build()
               .output(0);
diff --git a/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/GraphTest.java b/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/GraphTest.java
index efa6c09f2a5..64643c0c58c 100644
--- a/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/GraphTest.java
+++ b/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/GraphTest.java
@@ -28,6 +28,7 @@
 import org.tensorflow.op.Ops;
 import org.tensorflow.op.linalg.MatMul;
 import org.tensorflow.proto.framework.GraphDef;
+import org.tensorflow.util.TensorList;
 import org.tensorflow.types.TFloat32;
 import org.tensorflow.types.TInt32;
 
@@ -42,7 +43,7 @@ public void graphDefRoundTrip() {
       Ops tf = Ops.create(g);
       tf.withName("Y").linalg.matMul(
           tf.withName("A").constant(new int[2][2]),
-          tf.withName("X").placeholder(TInt32.DTYPE),
+          tf.withName("X").placeholder(TInt32.class),
           MatMul.transposeA(true).transposeB(false)
       );
       graphDef = g.toGraphDef();
@@ -140,8 +141,8 @@ public void addGradientsToGraph() {
         Session s = new Session(g)) {
       Ops tf = Ops.create(g);
 
-      Output<TFloat32> x1 = tf.placeholder(TFloat32.DTYPE).output();
-      Output<TFloat32> x2 = tf.placeholder(TFloat32.DTYPE).output();
+      Output<TFloat32> x1 = tf.placeholder(TFloat32.class).output();
+      Output<TFloat32> x2 = tf.placeholder(TFloat32.class).output();
       Output<TFloat32> y0 = tf.math.square(x1).y();
       Output<TFloat32> y1 = tf.math.square(y0).y();
       Output<TFloat32> y2 = tf.math.addN(Arrays.asList(y0, x2)).sum();
@@ -149,29 +150,28 @@ public void addGradientsToGraph() {
       Output<?>[] grads0 = g.addGradients(y1, toArray(x1));
       assertNotNull(grads0);
       assertEquals(1, grads0.length);
-      assertEquals(TFloat32.DTYPE, grads0[0].dataType());
+      assertEquals(TFloat32.class, grads0[0].type());
 
       Output<?>[] grads1 = g.addGradients(y2, toArray(x1, x2));
       assertNotNull(grads1);
       assertEquals(2, grads1.length);
-      assertEquals(TFloat32.DTYPE, grads1[0].dataType());
-      assertEquals(TFloat32.DTYPE, grads1[1].dataType());
+      assertEquals(TFloat32.class, grads1[0].type());
+      assertEquals(TFloat32.class, grads1[1].type());
       
       try (TFloat32 c1 = TFloat32.scalarOf(3.0f);
           TFloat32 c2 = TFloat32.scalarOf(2.0f);
-          AutoCloseableList<Tensor<?>> outputs = new AutoCloseableList<>(
-              s.runner()
+          TensorList outputs = s.runner()
                   .feed(x1, c1)
                   .feed(x2, c2)
                   .fetch(grads0[0])
                   .fetch(grads1[0])
                   .fetch(grads1[1])
-                  .run())) {
+                  .run()) {
      
         assertEquals(3, outputs.size());
-        assertEquals(108.0f, outputs.get(0).expect(TFloat32.DTYPE).getFloat(), 0.0f);
-        assertEquals(6.0f, outputs.get(1).expect(TFloat32.DTYPE).getFloat(), 0.0f);
-        assertEquals(1.0f, outputs.get(2).expect(TFloat32.DTYPE).getFloat(), 0.0f);
+        assertEquals(108.0f, ((TFloat32)outputs.get(0)).getFloat(), 0.0f);
+        assertEquals(6.0f, ((TFloat32)outputs.get(1)).getFloat(), 0.0f);
+        assertEquals(1.0f, ((TFloat32)outputs.get(2)).getFloat(), 0.0f);
       }
     }
   }
@@ -182,22 +182,21 @@ public void addGradientSumsToGraph() {
         Session s = new Session(g)) {
       Ops tf = Ops.create(g);
 
-      Output<TFloat32> x = tf.placeholder(TFloat32.DTYPE).output();
+      Output<TFloat32> x = tf.placeholder(TFloat32.class).output();
       Output<TFloat32> y0 = tf.math.square(x).y();
       Output<TFloat32> y1 = tf.math.square(y0).y();
 
       Output<?>[] grad = g.addGradients(null, toArray(y0, y1), toArray(x), null);
       assertNotNull(grad);
       assertEquals(1, grad.length);
-      assertEquals(TFloat32.DTYPE, grad[0].dataType());
+      assertEquals(TFloat32.class, grad[0].type());
 
       try (TFloat32 c = TFloat32.scalarOf(3.0f);
           TFloat32 output = s.runner()
               .feed(x, c)
               .fetch(grad[0])
               .run()
-              .get(0)
-              .expect(TFloat32.DTYPE)) {
+              .single()) {
         assertEquals(114.0f, output.getFloat(), 0.0f);
       }
     }
@@ -209,27 +208,26 @@ public void addGradientsWithInitialValuesToGraph() {
         Session s = new Session(g)) {
       Ops tf = Ops.create(g);
 
-      Output<TFloat32> x = tf.placeholder(TFloat32.DTYPE).output();
+      Output<TFloat32> x = tf.placeholder(TFloat32.class).output();
       Output<TFloat32> y0 = tf.math.square(x).y();
       Output<TFloat32> y1 = tf.math.square(y0).y();
       
       Output<?>[] grad0 = g.addGradients(y1, toArray(y0));
       assertNotNull(grad0);
       assertEquals(1, grad0.length);
-      assertEquals(TFloat32.DTYPE, grad0[0].dataType());
+      assertEquals(TFloat32.class, grad0[0].type());
 
       Output<?>[] grad1 = g.addGradients(null, toArray(y0), toArray(x), toArray(grad0[0]));
       assertNotNull(grad1);
       assertEquals(1, grad1.length);
-      assertEquals(TFloat32.DTYPE, grad1[0].dataType());
+      assertEquals(TFloat32.class, grad1[0].type());
 
       try (TFloat32 c = TFloat32.scalarOf(3.0f);
           TFloat32 output = s.runner()
               .feed(x, c)
               .fetch(grad1[0])
               .run()
-              .get(0)
-              .expect(TFloat32.DTYPE)) {
+              .single()) {
         assertEquals(108.0f, output.getFloat(), 0.0f);
       }
     }
@@ -240,7 +238,7 @@ public void validateGradientsNames() {
     try (Graph g = new Graph()) {
       Ops tf = Ops.create(g);
 
-      Output<TFloat32> x = tf.placeholder(TFloat32.DTYPE).output();
+      Output<TFloat32> x = tf.placeholder(TFloat32.class).output();
       Output<TFloat32> y0 = tf.math.square(x).y();
 
       Output<?>[] grad0 = g.addGradients(null, toArray(y0), toArray(x), null);
@@ -269,7 +267,7 @@ public void buildWhileLoopSingleInput() {
         Session s = new Session(g)) {
       Ops tf = Ops.create(g);
 
-      Output<?> input = tf.placeholder(TInt32.DTYPE).output();
+      Output<?> input = tf.placeholder(TInt32.class).output();
 
       @SuppressWarnings("unchecked")
       Output<?>[] loopOutputs = g.whileLoop(
@@ -289,8 +287,7 @@ public void buildWhileLoopSingleInput() {
               .feed(input, c)
               .fetch(loopOutputs[0])
               .run()
-              .get(0)
-              .expect(TInt32.DTYPE)) {
+              .single()) {
         assertEquals(16, output.getInt()); // ((2^2)^2)
       }
     }
@@ -302,8 +299,8 @@ public void buildWhileLoopMultipleInputs() {
         Session s = new Session(g)) {
       Ops tf = Ops.create(g);
 
-      Output<?> input1 = tf.placeholder(TInt32.DTYPE).output();
-      Output<?> input2 = tf.placeholder(TInt32.DTYPE).output();
+      Output<?> input1 = tf.placeholder(TInt32.class).output();
+      Output<?> input2 = tf.placeholder(TInt32.class).output();
       Output<?>[] inputs = toArray(input1, input2);
 
       @SuppressWarnings("unchecked")
@@ -322,17 +319,15 @@ public void buildWhileLoopMultipleInputs() {
 
       try (TInt32 c1 = TInt32.scalarOf(2);
           TInt32 c2 = TInt32.scalarOf(5);
-          AutoCloseableList<Tensor<?>> outputs =
-              new AutoCloseableList<>(
-                  s.runner()
-                      .feed(input1, c1)
-                      .feed(input2, c2)
-                      .fetch(loopOutputs[0])
-                      .fetch(loopOutputs[1])
-                      .run())) {
+          TensorList outputs = s.runner()
+                  .feed(input1, c1)
+                  .feed(input2, c2)
+                  .fetch(loopOutputs[0])
+                  .fetch(loopOutputs[1])
+                  .run()) {
         assertEquals(2, outputs.size());
-        assertEquals(16, outputs.get(0).expect(TInt32.DTYPE).getInt()); // ((2^2)^2)
-        assertEquals(625, outputs.get(1).expect(TInt32.DTYPE).getInt()); // ((5^2)^2)
+        assertEquals(16, ((TInt32)outputs.get(0)).getInt()); // ((2^2)^2)
+        assertEquals(625, ((TInt32)outputs.get(1)).getInt()); // ((5^2)^2)
       }
     }
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/SavedModelBundleTest.java b/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/SavedModelBundleTest.java
index d807d13de00..67a0f9d105b 100644
--- a/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/SavedModelBundleTest.java
+++ b/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/SavedModelBundleTest.java
@@ -44,6 +44,7 @@
 import org.tensorflow.proto.framework.RunOptions;
 import org.tensorflow.proto.framework.SignatureDef;
 import org.tensorflow.proto.framework.TensorInfo;
+import org.tensorflow.util.TensorMap;
 import org.tensorflow.types.TFloat32;
 
 /** Unit tests for {@link org.tensorflow.SavedModelBundle}. */
@@ -107,9 +108,9 @@ public void exportFunctionWithVariables() throws IOException {
       f.session().run(Init.DEFAULT_NAME);
 
       // Call the graph and remember the result of computation for later
-      try (Tensor<TFloat32> xTensor = TFloat32.tensorOf(xValue);
-          Tensor<TFloat32> zTensor = f.call(xTensor).expect(TFloat32.DTYPE)) {
-        reducedSum = zTensor.data().getFloat();
+      try (TFloat32 xTensor = TFloat32.tensorOf(xValue);
+          TFloat32 zTensor = f.call(xTensor)) {
+        reducedSum = zTensor.getFloat();
       }
       // Save/export the model (which is a single function in this case)
       f.save(testFolder.toString());
@@ -153,15 +154,15 @@ public void exportFunctionWithVariables() throws IOException {
       assertNotNull(outputInfo);
       assertEquals(0, outputInfo.getTensorShape().getDimCount());
 
-      try (Tensor<TFloat32> xTensor = TFloat32.tensorOf(xValue)) {
+      try (TFloat32 xTensor = TFloat32.tensorOf(xValue)) {
         // Call the saved model function and make sure it returns the same result as before
-        try (Tensor<TFloat32> zTensor = function.call(xTensor).expect(TFloat32.DTYPE)) {
-          assertEquals(reducedSum, zTensor.data().getFloat(), EPSILON);
+        try (TFloat32 zTensor = function.call(xTensor)) {
+          assertEquals(reducedSum, zTensor.getFloat(), EPSILON);
         }
         // Now call the same function directly from the model
-        try (Tensor<TFloat32> zTensor =
-            savedModel.call(Collections.singletonMap("input", xTensor)).get("reducedSum").expect(TFloat32.DTYPE)) {
-          assertEquals(reducedSum, zTensor.data().getFloat(), EPSILON);
+        try (TFloat32 zTensor =
+            savedModel.call(Collections.singletonMap("input", xTensor)).get("reducedSum")) {
+          assertEquals(reducedSum, zTensor.getFloat(), EPSILON);
         }
       }
     }
@@ -179,9 +180,9 @@ public void exportMultipleFunctions() throws IOException {
           ConcreteFunction f1 = ConcreteFunction.create(f1Signature, s);
           ConcreteFunction f2 = ConcreteFunction.create(f2Signature, s)) {
         f1.session().run(Init.DEFAULT_NAME);
-        try (Tensor<TFloat32> x = TFloat32.tensorOf(StdArrays.ndCopyOf(new float[]{2, 2}));
-            Tensor<TFloat32> t = f1.call(x).expect(TFloat32.DTYPE)) {
-          reducedSum = t.data().getFloat();
+        try (TFloat32 x = TFloat32.tensorOf(StdArrays.ndCopyOf(new float[]{2, 2}));
+            TFloat32 t = f1.call(x)) {
+          reducedSum = t.getFloat();
         }
         SavedModelBundle.exporter(testFolder.toString())
             .withFunction(f1)
@@ -193,15 +194,15 @@ public void exportMultipleFunctions() throws IOException {
       assertEquals(2, model.signatures().size());
       ConcreteFunction f1 = model.function(Signature.DEFAULT_KEY);
       assertNotNull(f1);
-      try (Tensor<TFloat32> x = TFloat32.tensorOf(StdArrays.ndCopyOf(new float[]{2, 2}));
-          Tensor<TFloat32> t = f1.call(x).expect(TFloat32.DTYPE)) {
-        assertEquals(reducedSum, t.data().getFloat(), EPSILON);
+      try (TFloat32 x = TFloat32.tensorOf(StdArrays.ndCopyOf(new float[]{2, 2}));
+          TFloat32 t = f1.call(x)) {
+        assertEquals(reducedSum, t.getFloat(), EPSILON);
       }
       ConcreteFunction f2 = model.function("identity");
       assertNotNull(f2);
-      try (Tensor<TFloat32> x = TFloat32.scalarOf(10.0f);
-          Tensor<TFloat32> t = f2.call(x).expect(TFloat32.DTYPE)) {
-        assertEquals(10.0f, t.data().getFloat(), 0.0f);
+      try (TFloat32 x = TFloat32.scalarOf(10.0f);
+          TFloat32 t = f2.call(x)) {
+        assertEquals(10.0f, t.getFloat(), 0.0f);
       }
       try {
         model.function("NoSuchFunction");
@@ -290,31 +291,29 @@ public void pythonTfFunction() {
        *   Signature name used for saving 'add', argument names 'a' and 'b'
        */
       ConcreteFunction add = bundle.function("add");
-      Map<String, Tensor<?>> args = new HashMap();
-      try (Tensor<TFloat32> a = TFloat32.scalarOf(10.0f);
-           Tensor<TFloat32> b = TFloat32.scalarOf(15.5f)) {
-        args.put("a", a);
-        args.put("b", b);
-        Map<String, Tensor<?>> result = add.call(args);
+      try (TensorMap args = new TensorMap()) {
+        args.put("a", TFloat32.scalarOf(10.0f));
+        args.put("b", TFloat32.scalarOf(15.5f));
+        TensorMap result = add.call(args);
         assertEquals(result.size(), 1);
-        try (Tensor<TFloat32> c = result.values().iterator().next().expect(TFloat32.DTYPE)) {
-          assertEquals(25.5f, c.data().getFloat());
+        try (TFloat32 c = result.single()) {
+          assertEquals(25.5f, c.getFloat());
         }
       }
     }
   }
 
   private static Signature buildGraphWithVariables(Ops tf, Shape xShape) {
-    Placeholder<TFloat32> x = tf.placeholder(TFloat32.DTYPE, Placeholder.shape(xShape));
+    Placeholder<TFloat32> x = tf.placeholder(TFloat32.class, Placeholder.shape(xShape));
     Variable<TFloat32> y = tf
-        .variable(tf.random.randomUniform(tf.constant(xShape), TFloat32.DTYPE));
+        .variable(tf.random.randomUniform(tf.constant(xShape), TFloat32.class));
     ReduceSum<TFloat32> z = tf.reduceSum(tf.math.add(x, y), tf.array(0, 1));
     Init init = tf.init();
     return Signature.builder().input("input", x).output("reducedSum", z).build();
   }
 
   private static Signature buildIdentityGraph(Ops tf, String signatureKey) {
-    Placeholder<TFloat32> x = tf.placeholder(TFloat32.DTYPE, Placeholder.shape(Shape.scalar()));
+    Placeholder<TFloat32> x = tf.placeholder(TFloat32.class, Placeholder.shape(Shape.scalar()));
     Identity<TFloat32> xprime = tf.identity(x);
     return Signature.builder().key(signatureKey).input("x", x).output("x", xprime).build();
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/SessionTest.java b/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/SessionTest.java
index 9e8191434bc..cb179ac34ff 100644
--- a/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/SessionTest.java
+++ b/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/SessionTest.java
@@ -23,7 +23,6 @@
 import java.io.IOException;
 import java.nio.file.Files;
 import java.nio.file.Path;
-import java.nio.file.Paths;
 import org.junit.jupiter.api.Test;
 import org.tensorflow.op.Ops;
 import org.tensorflow.op.core.Init;
@@ -36,6 +35,7 @@
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.ndarray.NdArrays;
 import org.tensorflow.ndarray.StdArrays;
+import org.tensorflow.util.TensorList;
 import org.tensorflow.types.TFloat32;
 import org.tensorflow.types.TInt32;
 
@@ -49,10 +49,9 @@ public void runUsingOperationNames() {
       Ops tf = Ops.create(g);
       transpose_A_times_X(tf, new int[][] {{2}, {3}});
       try (TInt32 x = TInt32.tensorOf(StdArrays.ndCopyOf(new int[][] {{5}, {7}}));
-          AutoCloseableList<Tensor<?>> outputs =
-              new AutoCloseableList<>(s.runner().feed("X", x).fetch("Y").run())) {
+          TensorList outputs = s.runner().feed("X", x).fetch("Y").run()) {
         assertEquals(1, outputs.size());
-        assertEquals(31, outputs.get(0).expect(TInt32.DTYPE).getInt(0, 0));
+        assertEquals(31, ((TInt32)outputs.single()).getInt(0, 0));
       }
     }
   }
@@ -66,10 +65,9 @@ public void runUsingOperationHandles() {
       Output<TInt32> feed = g.operation("X").output(0);
       Output<TInt32> fetch = g.operation("Y").output(0);
       try (TInt32 x = TInt32.tensorOf(StdArrays.ndCopyOf(new int[][] {{5}, {7}}));
-          AutoCloseableList<Tensor<?>> outputs =
-              new AutoCloseableList<>(s.runner().feed(feed, x).fetch(fetch).run())) {
+          TensorList outputs = s.runner().feed(feed, x).fetch(fetch).run()) {
         assertEquals(1, outputs.size());
-        assertEquals(31, outputs.get(0).expect(TInt32.DTYPE).getInt(0, 0));
+        assertEquals(31, ((TInt32)outputs.single()).getInt(0, 0));
       }
     }
   }
@@ -83,8 +81,7 @@ public void runUsingColonSeparatedNames() {
       tf.math.add(split.output().get(0), split.output().get(1));
 
       // Fetch using colon separated names.
-      try (TInt32 fetched =
-          s.runner().fetch("Split:1").run().get(0).expect(TInt32.DTYPE)) {
+      try (TInt32 fetched = s.runner().fetch("Split:1").run().get(0)) {
         assertEquals(3, fetched.getInt(0));
         assertEquals(4, fetched.getInt(1));
       }
@@ -96,8 +93,7 @@ public void runUsingColonSeparatedNames() {
                   .feed("Split:1", fed)
                   .fetch("Add")
                   .run()
-                  .get(0)
-                  .expect(TInt32.DTYPE)) {
+                  .single()) {
         assertEquals(NdArrays.vectorOf(8, 6, 4, 2), fetched);
       }
     }
@@ -117,9 +113,9 @@ public void runWithMetadata() {
                 .setOptions(fullTraceRunOptions())
                 .runAndFetchMetadata();
         // Sanity check on outputs.
-        AutoCloseableList<Tensor<?>> outputs = new AutoCloseableList<>(result.outputs);
+        TensorList outputs = result.outputs;
         assertEquals(1, outputs.size());
-        assertEquals(31, outputs.get(0).expect(TInt32.DTYPE).getInt(0, 0));
+        assertEquals(31, ((TInt32)outputs.single()).getInt(0, 0));
         // Sanity check on metadata
         assertNotNull(result.metadata);
         assertTrue(result.metadata.hasStepStats(), result.metadata.toString());
@@ -135,11 +131,10 @@ public void runMultipleOutputs() {
       Ops tf = Ops.create(g);
       tf.withName("c1").constant(2718);
       tf.withName("c2").constant(31415);
-      AutoCloseableList<Tensor<?>> outputs =
-          new AutoCloseableList<>(s.runner().fetch("c2").fetch("c1").run());
+      TensorList outputs = s.runner().fetch("c2").fetch("c1").run();
       assertEquals(2, outputs.size());
-      assertEquals(31415, outputs.get(0).expect(TInt32.DTYPE).getInt());
-      assertEquals(2718, outputs.get(1).expect(TInt32.DTYPE).getInt());
+      assertEquals(31415, ((TInt32)outputs.get(0)).getInt());
+      assertEquals(2718, ((TInt32)outputs.get(1)).getInt());
       outputs.close();
     }
   }
@@ -169,7 +164,7 @@ public void runInit() {
     try (Graph g = new Graph()) {
       Ops tf = Ops.create(g);
 
-      Variable<TInt32> var1 = tf.variable(Shape.scalar(), TInt32.DTYPE);
+      Variable<TInt32> var1 = tf.variable(Shape.scalar(), TInt32.class);
       tf.initAdd(tf.assign(var1, tf.constant(10)));
       Variable<TInt32> var2 = tf.variable(tf.constant(20));
       Add<TInt32> add = tf.math.add(var1, var2);
@@ -177,7 +172,7 @@ public void runInit() {
       try (Session s = new Session(g)) {
         s.run(tf.init());
 
-        try (TInt32 t = s.runner().fetch(add).run().get(0).expect(TInt32.DTYPE)) {
+        try (TInt32 t = s.runner().fetch(add).run().single()) {
           assertEquals(30, t.getInt());
         }
       }
@@ -189,7 +184,7 @@ public void runInitByName() {
     try (Graph g = new Graph()) {
       Ops tf = Ops.create(g);
 
-      Variable<TInt32> var1 = tf.variable(Shape.scalar(), TInt32.DTYPE);
+      Variable<TInt32> var1 = tf.variable(Shape.scalar(), TInt32.class);
       tf.initAdd(tf.assign(var1, tf.constant(10)));
       Variable<TInt32> var2 = tf.variable(tf.constant(20));
       Add<TInt32> add = tf.math.add(var1, var2);
@@ -198,7 +193,7 @@ public void runInitByName() {
       try (Session s = new Session(g)) {
         s.run("init_test");
 
-        try (TInt32 t = s.runner().fetch(add).run().get(0).expect(TInt32.DTYPE)) {
+        try (TInt32 t = s.runner().fetch(add).run().single()) {
           assertEquals(30, t.getInt());
         }
         try {
@@ -216,8 +211,8 @@ public void save() throws IOException  {
     Path testFolder = Files.createTempDirectory("tf-session-save-test");
     try (Graph g = new Graph()) {
       Ops tf = Ops.create(g);
-      Variable<TFloat32> x = tf.variable(tf.random.randomUniform(tf.constant(Shape.of(3, 3L)), TFloat32.DTYPE));
-      Variable<TFloat32> y = tf.variable(tf.random.randomUniform(tf.constant(Shape.of(3, 3L)), TFloat32.DTYPE));
+      Variable<TFloat32> x = tf.variable(tf.random.randomUniform(tf.constant(Shape.of(3, 3L)), TFloat32.class));
+      Variable<TFloat32> y = tf.variable(tf.random.randomUniform(tf.constant(Shape.of(3, 3L)), TFloat32.class));
       Init init = tf.init();
 
       try (Session s = new Session(g)) {
@@ -245,7 +240,7 @@ private static ConfigProto singleThreadConfigProto() {
   private static void transpose_A_times_X(Ops tf, int[][] a) {
     tf.withName("Y").linalg.matMul(
         tf.withName("A").constant(a),
-        tf.withName("X").placeholder(TInt32.DTYPE),
+        tf.withName("X").placeholder(TInt32.class),
         MatMul.transposeA(true).transposeB(false)
     );
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/TensorTest.java b/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/TensorTest.java
index 6aa33ef0a73..11c0cdedde0 100644
--- a/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/TensorTest.java
+++ b/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/TensorTest.java
@@ -48,8 +48,10 @@
 import org.tensorflow.types.TInt64;
 import org.tensorflow.types.TString;
 import org.tensorflow.types.TUint8;
+import org.tensorflow.types.Type;
+import org.tensorflow.types.TypeRegistry;
 
-/** Unit tests for {@link org.tensorflow.Tensor}. */
+/** Unit tests for {@link Tensor}. */
 public class TensorTest {
   private static final double EPSILON = 1e-7;
   private static final float EPSILON_F = 1e-7f;
@@ -71,14 +73,14 @@ public void createWithRawData() {
 
     // validate creating a tensor using a raw data byte buffers
     {
-      try (TBool t = Tensors.of(TBool.DTYPE, bools_shape, DataBuffers.of(bools_))) {
+      try (TBool t = Tensors.of(TBool.class, bools_shape, DataBuffers.of(bools_))) {
         boolean[] actual = new boolean[bools_.length];
         t.read(DataBuffers.of(actual));
         assertArrayEquals(bools, actual);
       }
 
       // note: the buffer is expected to contain raw TF_STRING (as per C API)
-      try (TString t = Tensors.of(TString.DTYPE, strings_shape, DataBuffers.of(strings_))) {
+      try (TString t = Tensors.of(TString.class, strings_shape, DataBuffers.of(strings_))) {
         assertEquals(strings, t.getObject());
       }
     }
@@ -95,7 +97,7 @@ public void createWithRawData() {
     }
 
     // validate shape checking
-    try (TBool t = Tensors.of(TBool.DTYPE, Shape.of(bools_.length * 2), DataBuffers.of(bools_))) {
+    try (TBool t = Tensors.of(TBool.class, Shape.of(bools_.length * 2), DataBuffers.of(bools_))) {
       fail("should have failed on incompatible buffer");
     } catch (IllegalArgumentException e) {
       // expected
@@ -267,44 +269,44 @@ public void readFromRawData() {
   @Test
   public void scalars() {
     try (TFloat32 t = TFloat32.scalarOf(2.718f)) {
-      assertEquals(TFloat32.DTYPE, t.dataType());
+      assertEquals(TFloat32.class, t.type());
       assertEquals(0, t.shape().numDimensions());
       assertEquals(2.718f, t.getFloat(), EPSILON_F);
     }
 
     try (TFloat64 t = TFloat64.scalarOf(3.1415)) {
-      assertEquals(TFloat64.DTYPE, t.dataType());
+      assertEquals(TFloat64.class, t.type());
       assertEquals(0, t.shape().numDimensions());
       assertEquals(3.1415, t.getDouble(), EPSILON);
     }
 
     try (TInt32 t = TInt32.scalarOf(-33)) {
-      assertEquals(TInt32.DTYPE, t.dataType());
+      assertEquals(TInt32.class, t.type());
       assertEquals(0, t.shape().numDimensions());
       assertEquals(-33, t.getInt());
     }
 
     try (TInt64 t = TInt64.scalarOf(8589934592L)) {
-      assertEquals(TInt64.DTYPE, t.dataType());
+      assertEquals(TInt64.class, t.type());
       assertEquals(0, t.shape().numDimensions());
       assertEquals(8589934592L, t.getLong());
     }
 
     try (TBool t = TBool.scalarOf(true)) {
-      assertEquals(TBool.DTYPE, t.dataType());
+      assertEquals(TBool.class, t.type());
       assertEquals(0, t.shape().numDimensions());
       assertTrue(t.getBoolean());
     }
 
     try (TString t = TString.scalarOf("sombrero")) {
-      assertEquals(TString.DTYPE, t.dataType());
+      assertEquals(TString.class, t.type());
       assertEquals(0, t.shape().numDimensions());
       assertEquals("sombrero", t.getObject());
     }
 
     final byte[] bytes = {1, 2, 3, 4};
     try (TString t = TString.tensorOfBytes(NdArrays.scalarOfObject(bytes))) {
-      assertEquals(TString.DTYPE, t.dataType());
+      assertEquals(TString.class, t.type());
       assertEquals(0, t.shape().numDimensions());
       assertArrayEquals(bytes, t.asBytes().getObject());
     }
@@ -314,7 +316,7 @@ public void scalars() {
   public void nDimensional() {
     DoubleNdArray vector = StdArrays.ndCopyOf(new double[]{1.414, 2.718, 3.1415});
     try (TFloat64 t = TFloat64.tensorOf(vector)) {
-      assertEquals(TFloat64.DTYPE, t.dataType());
+      assertEquals(TFloat64.class, t.type());
       assertEquals(1, t.shape().numDimensions());
       assertEquals(3, t.shape().size(0));
       assertEquals(vector, t);
@@ -322,7 +324,7 @@ public void nDimensional() {
 
     IntNdArray matrix = StdArrays.ndCopyOf(new int[][]{{1, 2, 3}, {4, 5, 6}});
     try (TInt32 t = TInt32.tensorOf(matrix)) {
-      assertEquals(TInt32.DTYPE, t.dataType());
+      assertEquals(TInt32.class, t.type());
       assertEquals(2, t.shape().numDimensions());
       assertEquals(2, t.shape().size(0));
       assertEquals(3, t.shape().size(1));
@@ -333,7 +335,7 @@ public void nDimensional() {
       {{1}, {3}, {5}, {7}, {9}}, {{2}, {4}, {6}, {8}, {0}},
     });
     try (TInt64 t = TInt64.tensorOf(threeD)) {
-      assertEquals(TInt64.DTYPE, t.dataType());
+      assertEquals(TInt64.class, t.type());
       assertEquals(3, t.shape().numDimensions());
       assertEquals(2, t.shape().size(0));
       assertEquals(5, t.shape().size(1));
@@ -347,7 +349,7 @@ public void nDimensional() {
       {{{false, true, false, true}, {false, true, true, false}}},
     });
     try (TBool t = TBool.tensorOf(fourD)) {
-      assertEquals(TBool.DTYPE, t.dataType());
+      assertEquals(TBool.class, t.type());
       assertEquals(4, t.shape().numDimensions());
       assertEquals(3, t.shape().size(0));
       assertEquals(1, t.shape().size(1));
@@ -366,7 +368,7 @@ public void testNDimensionalStringTensor() {
       }
     }
     try (TString t = TString.tensorOf(matrix)) {
-      assertEquals(TString.DTYPE, t.dataType());
+      assertEquals(TString.class, t.type());
       assertEquals(2, t.shape().numDimensions());
       assertEquals(4, t.shape().size(0));
       assertEquals(3, t.shape().size(1));
@@ -376,7 +378,7 @@ public void testNDimensionalStringTensor() {
     NdArray<byte[]> byteMatrix = NdArrays.ofObjects(byte[].class, matrix.shape());
     matrix.scalars().forEachIndexed((i, s) -> byteMatrix.setObject(s.getObject().getBytes(UTF_8), i));
     try (TString t = TString.tensorOfBytes(byteMatrix)) {
-      assertEquals(TString.DTYPE, t.dataType());
+      assertEquals(TString.class, t.type());
       assertEquals(2, t.shape().numDimensions());
       assertEquals(4, t.shape().size(0));
       assertEquals(3, t.shape().size(1));
@@ -389,7 +391,7 @@ public void testNDimensionalStringTensor() {
   public void testUint8TensorFromArray() {
     byte[] vector = new byte[] {1, 2, 3, 4};
     try (TUint8 t = TUint8.vectorOf(vector)) {
-      assertEquals(TUint8.DTYPE, t.dataType());
+      assertEquals(TUint8.class, t.type());
       assertEquals(1, t.shape().numDimensions());
       assertEquals(4, t.shape().size(0));
 
@@ -403,7 +405,7 @@ public void testUint8TensorFromArray() {
   public void testCreateFromArrayOfBoxed() {
     Integer[] vector = new Integer[] {1, 2, 3, 4};
     try (TInt32 t = TInt32.tensorOf(Shape.of(4), d -> d.write(DataBuffers.ofObjects(vector)))) {
-      assertEquals(TInt32.DTYPE, t.dataType());
+      assertEquals(TInt32.class, t.type());
       assertEquals(1, t.shape().numDimensions());
       assertEquals(4, t.shape().size(0));
 
@@ -445,14 +447,15 @@ public void tensorWithZeroDimension() {
 
   @Test
   public void allocateTensorWithSize() {
-    try (TInt32 t = Tensors.of(TInt32.DTYPE, Shape.of(2, 2, 2), 8 * TInt32.DTYPE.byteSize())) {
+    Type<TInt32> type = TypeRegistry.find(TInt32.class);
+    try (TInt32 t = Tensors.of(TInt32.class, Shape.of(2, 2, 2), 8 * type.byteSize())) {
       // ok
     }
-    try (TInt32 t = Tensors.of(TInt32.DTYPE, Shape.of(2, 2, 2), 9 * TInt32.DTYPE.byteSize())) {
+    try (TInt32 t = Tensors.of(TInt32.class, Shape.of(2, 2, 2), 9 * type.byteSize())) {
       // ok (size requested is larger that minimum space required)
     }
     try {
-      Tensors.of(TInt32.DTYPE, Shape.of(2, 2, 2), 8 * TInt32.DTYPE.byteSize() - 1);
+      Tensors.of(TInt32.class, Shape.of(2, 2, 2), 8 * type.byteSize() - 1);
       fail();
     } catch (IllegalArgumentException e) {
       // as expected
@@ -465,11 +468,11 @@ public void eagerTensorIsReleasedAfterSessionIsClosed() {
     try (EagerSession session = EagerSession.create()) {
       Ops tf = Ops.create(session);
       sum = tf.math.add(tf.constant(10), tf.constant(20)).asTensor();
-      ((AbstractTensor)sum).nativeHandle(); // does not throw
+      sum.tensorHandle().nativeHandle(); // does not throw
       assertEquals(30, sum.getInt());
     }
     try {
-      ((AbstractTensor)sum).nativeHandle();
+      sum.tensorHandle().nativeHandle();
       fail("Tensor native handle should have been closed by ending eager session");
     } catch (IllegalStateException e) {
       // as expected
@@ -492,8 +495,8 @@ public void fromHandle() {
     // close() on both Tensors.
     final FloatNdArray matrix = StdArrays.ndCopyOf(new float[][]{{1, 2, 3}, {4, 5, 6}});
     try (TFloat32 src = TFloat32.tensorOf(matrix)) {
-      TFloat32 cpy = Tensors.fromHandle(((AbstractTensor)src).nativeHandle()).expect(TFloat32.DTYPE);
-      assertEquals(src.dataType(), cpy.dataType());
+      TFloat32 cpy = Tensors.fromHandle(src.tensorHandle());
+      assertEquals(src.type(), cpy.type());
       assertEquals(src.shape().numDimensions(), cpy.shape().numDimensions());
       assertEquals(src.shape(), cpy.shape());
       assertEquals(matrix, cpy);
diff --git a/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/op/ScopeTest.java b/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/op/ScopeTest.java
index af3523c9e28..e83c956adb7 100644
--- a/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/op/ScopeTest.java
+++ b/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/op/ScopeTest.java
@@ -23,9 +23,8 @@
 import org.tensorflow.Graph;
 import org.tensorflow.Output;
 import org.tensorflow.Session;
-import org.tensorflow.Tensor;
-import org.tensorflow.types.TInt32;
 import org.tensorflow.types.family.TType;
+import org.tensorflow.types.TInt32;
 
 /** Unit tests for {@link org.tensorflow.op.Scope}. */
 public class ScopeTest {
@@ -169,11 +168,12 @@ public void composite() {
       // assertNotNull(g.operation("variance/zero"));
 
       // Verify correct results as well.
-      TInt32 result =
-          sess.runner().fetch(var1.output()).run().get(0).expect(TInt32.DTYPE);
-      assertEquals(21704, result.getInt());
-      result = sess.runner().fetch(var2.output()).run().get(0).expect(TInt32.DTYPE);
-      assertEquals(21704, result.getInt());
+      try (TInt32 result = sess.runner().fetch(var1.output()).run().single()) {
+        assertEquals(21704, result.getInt());
+      }
+      try (TInt32 result = sess.runner().fetch(var2.output()).run().single()) {
+        assertEquals(21704, result.getInt());
+      }
     }
   }
 
@@ -193,7 +193,7 @@ static <T extends TType> Const<T> create(Scope s, T value) {
       return new Const<>(
           s.env()
               .opBuilder("Const", s.makeOpName("Const"))
-              .setAttr("dtype", value.dataType())
+              .setAttr("dtype", value.type())
               .setAttr("value", value)
               .build()
               .output(0));
diff --git a/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/op/core/ConstantTest.java b/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/op/core/ConstantTest.java
index cd9f4c94a24..0f94a546ca8 100644
--- a/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/op/core/ConstantTest.java
+++ b/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/op/core/ConstantTest.java
@@ -19,10 +19,9 @@
 
 import java.io.IOException;
 import org.junit.jupiter.api.Test;
-import org.tensorflow.AutoCloseableList;
 import org.tensorflow.Graph;
 import org.tensorflow.Session;
-import org.tensorflow.Tensor;
+import org.tensorflow.util.TensorList;
 import org.tensorflow.op.Scope;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.ndarray.buffer.DataBuffer;
@@ -57,10 +56,9 @@ public void createInts() {
       Scope scope = new Scope(g);
       Constant<TInt32> op1 = Constant.tensorOf(scope, shape, buffer);
       Constant<TInt32> op2 = Constant.tensorOf(scope, array);
-      try (AutoCloseableList<Tensor<?>> t =
-          new AutoCloseableList<>(sess.runner().fetch(op1).fetch(op2).run())) {
-        assertEquals(array, t.get(0).expect(TInt32.DTYPE));
-        assertEquals(array, t.get(1).expect(TInt32.DTYPE));
+      try (TensorList t = sess.runner().fetch(op1).fetch(op2).run()) {
+        assertEquals(array, t.get(0));
+        assertEquals(array, t.get(1));
       }
     }
   }
@@ -76,10 +74,9 @@ public void createFloats() {
       Scope scope = new Scope(g);
       Constant<TFloat32> op1 = Constant.tensorOf(scope, shape, buffer);
       Constant<TFloat32> op2 = Constant.tensorOf(scope, array);
-      try (AutoCloseableList<Tensor<?>> t =
-          new AutoCloseableList<>(sess.runner().fetch(op1).fetch(op2).run())) {
-        assertEquals(array, t.get(0).expect(TFloat32.DTYPE));
-        assertEquals(array, t.get(1).expect(TFloat32.DTYPE));
+      try (TensorList t = sess.runner().fetch(op1).fetch(op2).run()) {
+        assertEquals(array, t.get(0));
+        assertEquals(array, t.get(1));
       }
     }
   }
@@ -95,10 +92,9 @@ public void createDoubles() {
       Scope scope = new Scope(g);
       Constant<TFloat64> op1 = Constant.tensorOf(scope, shape, buffer);
       Constant<TFloat64> op2 = Constant.tensorOf(scope, array);
-      try (AutoCloseableList<Tensor<?>> t =
-          new AutoCloseableList<>(sess.runner().fetch(op1).fetch(op2).run())) {
-        assertEquals(array, t.get(0).expect(TFloat64.DTYPE));
-        assertEquals(array, t.get(1).expect(TFloat64.DTYPE));
+      try (TensorList t = sess.runner().fetch(op1).fetch(op2).run()) {
+        assertEquals(array, t.get(0));
+        assertEquals(array, t.get(1));
       }
     }
   }
@@ -114,10 +110,9 @@ public void createLongs() {
       Scope scope = new Scope(g);
       Constant<TInt64> op1 = Constant.tensorOf(scope, shape, buffer);
       Constant<TInt64> op2 = Constant.tensorOf(scope, array);
-      try (AutoCloseableList<Tensor<?>> t =
-          new AutoCloseableList<>(sess.runner().fetch(op1).fetch(op2).run())) {
-        assertEquals(array, t.get(0).expect(TInt64.DTYPE));
-        assertEquals(array, t.get(1).expect(TInt64.DTYPE));
+      try (TensorList t = sess.runner().fetch(op1).fetch(op2).run()) {
+        assertEquals(array, t.get(0));
+        assertEquals(array, t.get(1));
       }
     }
   }
@@ -133,10 +128,9 @@ public void createStrings() throws IOException {
       Scope scope = new Scope(g);
       Constant<TString> op1 = Constant.tensorOf(scope, shape, buffer);
       Constant<TString> op2 = Constant.tensorOf(scope, array);
-      try (AutoCloseableList<Tensor<?>> t =
-          new AutoCloseableList<>(sess.runner().fetch(op1).fetch(op2).run())) {
-        assertEquals(array, t.get(0).expect(TString.DTYPE));
-        assertEquals(array, t.get(1).expect(TString.DTYPE));
+      try (TensorList t = sess.runner().fetch(op1).fetch(op2).run()) {
+        assertEquals(array, t.get(0));
+        assertEquals(array, t.get(1));
       }
     }
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/op/core/GeneratedOperationsTest.java b/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/op/core/GeneratedOperationsTest.java
index 8d0c4764a91..e15583cda8c 100644
--- a/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/op/core/GeneratedOperationsTest.java
+++ b/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/op/core/GeneratedOperationsTest.java
@@ -22,7 +22,6 @@
 import org.tensorflow.Graph;
 import org.tensorflow.Operand;
 import org.tensorflow.Session;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.Op;
 import org.tensorflow.op.Ops;
 import org.tensorflow.ndarray.Shape;
@@ -36,7 +35,7 @@ public void tensorInputTensorOutput() {
         Session sess = new Session(g)) {
       Ops ops = Ops.create(g);
       Operand<TInt32> x = ops.math.add(ops.constant(1), ops.constant(2));
-      try (TInt32 result = sess.runner().fetch(x).run().get(0).expect(TInt32.DTYPE)) {
+      try (TInt32 result = sess.runner().fetch(x).run().single()) {
         assertEquals(3, result.getInt());
       }
     }
@@ -52,7 +51,7 @@ public void testListInputTensorOutput() {
       inputs.add(ops.constant(2));
       inputs.add(ops.constant(3));
       Operand<TInt32> x = ops.math.addN(inputs);
-      try (TInt32 result = sess.runner().fetch(x).run().get(0).expect(TInt32.DTYPE)) {
+      try (TInt32 result = sess.runner().fetch(x).run().single()) {
         assertEquals(6, result.getInt());
       }
     }
@@ -70,14 +69,14 @@ public void testControlDependencies() {
     try (Graph g = new Graph();
         Session sess = new Session(g)) {
       Ops ops = Ops.create(g);
-      Operand<TInt32> variable = ops.variable(Shape.scalar(), TInt32.DTYPE);
+      Operand<TInt32> variable = ops.variable(Shape.scalar(), TInt32.class);
       Operand<?> initVariable = ops.assign(variable, ops.constant(0));
       ArrayList<Op> controls = new ArrayList<>();
       controls.add(ops.assign(variable, ops.constant(3)));
       Operand<TInt32> x =
           ops.withControlDependencies(controls).math.add(variable, ops.constant(0));
       sess.runner().addTarget(initVariable).run();
-      try (TInt32 result = sess.runner().fetch(x).run().get(0).expect(TInt32.DTYPE)) {
+      try (TInt32 result = sess.runner().fetch(x).run().single()) {
         assertEquals(3, result.getInt());
       }
     }
diff --git a/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/op/core/GradientsTest.java b/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/op/core/GradientsTest.java
index 3303904f9e6..df2488662fc 100644
--- a/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/op/core/GradientsTest.java
+++ b/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/op/core/GradientsTest.java
@@ -21,12 +21,11 @@
 
 import java.util.Arrays;
 import org.junit.jupiter.api.Test;
-import org.tensorflow.AutoCloseableList;
 import org.tensorflow.Graph;
 import org.tensorflow.Output;
 import org.tensorflow.Session;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.Ops;
+import org.tensorflow.util.TensorList;
 import org.tensorflow.types.TFloat32;
 
 public class GradientsTest {
@@ -37,7 +36,7 @@ public void createGradients() {
         Session sess = new Session(g)) {
       Ops tf = Ops.create(g);
 
-      Output<TFloat32> x = tf.placeholder(TFloat32.DTYPE).output();
+      Output<TFloat32> x = tf.placeholder(TFloat32.class).output();
       Output<TFloat32> y0 = tf.math.square(x).y();
       Output<TFloat32> y1 = tf.math.square(y0).y();
 
@@ -48,12 +47,9 @@ public void createGradients() {
       assertEquals(2, grads.dy().size());
 
       try (TFloat32 c = TFloat32.scalarOf(3.0f);
-          AutoCloseableList<Tensor<?>> outputs =
-              new AutoCloseableList<>(
-                  sess.runner().feed(x, c).fetch(grads.dy(0)).fetch(grads.dy(1)).run())) {
-
-        assertEquals(108.0f, outputs.get(0).expect(TFloat32.DTYPE).getFloat(), 0.0f);
-        assertEquals(18.0f, outputs.get(1).expect(TFloat32.DTYPE).getFloat(), 0.0f);
+          TensorList outputs = sess.runner().feed(x, c).fetch(grads.dy(0)).fetch(grads.dy(1)).run()) {
+        assertEquals(108.0f, ((TFloat32)outputs.get(0)).getFloat(), 0.0f);
+        assertEquals(18.0f, ((TFloat32)outputs.get(1)).getFloat(), 0.0f);
       }
     }
   }
@@ -64,7 +60,7 @@ public void createGradientsWithSum() {
         Session sess = new Session(g)) {
       Ops tf = Ops.create(g);
 
-      Output<TFloat32> x = tf.placeholder(TFloat32.DTYPE).output();
+      Output<TFloat32> x = tf.placeholder(TFloat32.class).output();
       Output<TFloat32> y0 = tf.math.square(x).y();
       Output<TFloat32> y1 = tf.math.square(y0).y();
 
@@ -75,10 +71,8 @@ public void createGradientsWithSum() {
       assertEquals(1, grads.dy().size());
 
       try (TFloat32 c = TFloat32.scalarOf(3.0f);
-          AutoCloseableList<Tensor<?>> outputs =
-              new AutoCloseableList<>(sess.runner().feed(x, c).fetch(grads.dy(0)).run())) {
-
-        assertEquals(114.0f, outputs.get(0).expect(TFloat32.DTYPE).getFloat(), 0.0f);
+          TensorList outputs = sess.runner().feed(x, c).fetch(grads.dy(0)).run()) {
+        assertEquals(114.0f, ((TFloat32)outputs.get(0)).getFloat(), 0.0f);
       }
     }
   }
@@ -89,7 +83,7 @@ public void createGradientsWithInitialValues() {
         Session sess = new Session(g)) {
       Ops tf = Ops.create(g);
 
-      Output<TFloat32> x = tf.placeholder(TFloat32.DTYPE).output();
+      Output<TFloat32> x = tf.placeholder(TFloat32.class).output();
       Output<TFloat32> y0 = tf.math.square(x).y();
       Output<TFloat32> y1 = tf.math.square(y0).y();
 
@@ -101,11 +95,8 @@ public void createGradientsWithInitialValues() {
       assertEquals(1, grads1.dy().size());
 
       try (TFloat32 c = TFloat32.scalarOf(3.0f);
-          AutoCloseableList<Tensor<?>> outputs =
-              new AutoCloseableList<>(
-                  sess.runner().feed(x, c).fetch(grads1.dy(0)).run())) {
-
-        assertEquals(108.0f, outputs.get(0).expect(TFloat32.DTYPE).getFloat(), 0.0f);
+          TensorList outputs = sess.runner().feed(x, c).fetch(grads1.dy(0)).run()) {
+        assertEquals(108.0f, ((TFloat32)outputs.get(0)).getFloat(), 0.0f);
       }
     }
   }
@@ -115,7 +106,7 @@ public void validateGradientsNames() {
     try (Graph g = new Graph()) {
       Ops tf = Ops.create(g).withSubScope("sub");
 
-      Output<TFloat32> x = tf.placeholder(TFloat32.DTYPE).output();
+      Output<TFloat32> x = tf.placeholder(TFloat32.class).output();
       Output<TFloat32> y = tf.math.square(x).y();
 
       Gradients grad0 = Gradients.create(tf.scope(), y, Arrays.asList(x));
diff --git a/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/op/core/ShapesTest.java b/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/op/core/ShapesTest.java
index 583b3d14a0f..665d599145e 100644
--- a/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/op/core/ShapesTest.java
+++ b/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/op/core/ShapesTest.java
@@ -20,12 +20,10 @@
 
 import org.junit.jupiter.api.Test;
 
-import org.junit.jupiter.api.TestTemplate;
 import org.tensorflow.Graph;
 import org.tensorflow.EagerSession;
 import org.tensorflow.Operand;
 import org.tensorflow.Session;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.Scope;
 import org.tensorflow.types.TFloat32;
 import org.tensorflow.types.TInt32;
@@ -40,19 +38,16 @@ public void testFlatten_Operand() {
         Session session = new Session(g)) {
       Scope scope = new Scope(g);
       Operand<TFloat32> operand = Constant.arrayOf(scope, new float[] {1, 2, 3, 4, 5, 6, 7, 8});
-      Shape<TInt64> expResult = Shape.create(scope, operand, TInt64.DTYPE);
+      Shape<TInt64> expResult = Shape.create(scope, operand, TInt64.class);
       Operand<TFloat32> reshaped =
           Reshape.create(scope, operand, Constant.vectorOf(scope, new long[] {4, 2, 1}));
       Operand actual = Shapes.flatten(scope, reshaped);
-      Shape<TInt64> tfshape = Shape.create(scope, actual, TInt64.DTYPE);
+      Shape<TInt64> tfshape = Shape.create(scope, actual, TInt64.class);
 
       AtomicInteger index = new AtomicInteger();
-      try (TInt64 result1 =
-              session.runner().fetch(tfshape.asOutput()).run().get(0).expect(TInt64.DTYPE);
-          TInt64 result2 =
-              session.runner().fetch(expResult.asOutput()).run().get(0).expect(TInt64.DTYPE)) {
+      try (TInt64 result1 = session.runner().fetch(tfshape.asOutput()).run().single();
+          TInt64 result2 = session.runner().fetch(expResult.asOutput()).run().single()) {
         result1
-
             .scalars()
             .forEach(
                 s -> assertEquals(result2.getLong(index.getAndIncrement()), s.getLong()));
@@ -66,11 +61,11 @@ public void testFlatten_Shape() {
     try (EagerSession session = EagerSession.create()) {
       Scope scope = new Scope(session);
       Operand operand = Constant.arrayOf(scope, new float[] {1, 2, 3, 4, 5, 6, 7, 8});
-      Shape<TInt64> expShape = Shape.create(scope, operand, TInt64.DTYPE);
+      Shape<TInt64> expShape = Shape.create(scope, operand, TInt64.class);
       Operand actual =
           Reshape.create(scope, operand, Constant.vectorOf(scope, new long[] {4, 2, 1}));
-      Shape<TInt64> tfshape = Shape.create(scope, actual, TInt64.DTYPE);
-      Operand<TInt64> flattened = Shapes.flatten(scope, tfshape, TInt64.DTYPE);
+      Shape<TInt64> tfshape = Shape.create(scope, actual, TInt64.class);
+      Operand<TInt64> flattened = Shapes.flatten(scope, tfshape, TInt64.class);
 
       AtomicInteger index = new AtomicInteger();
       flattened
@@ -93,12 +88,11 @@ public void testSize_Shape() {
       Operand operand = Constant.arrayOf(scope, new float[] {1, 2, 3, 4, 5, 6, 7, 8});
       Operand actual =
           Reshape.create(scope, operand, Constant.vectorOf(scope, new long[] {4, 2, 1}));
-      Shape<TInt64> tfshape = Shape.create(scope, actual, TInt64.DTYPE);
-      Operand<TInt64> size = Shapes.size(scope, tfshape, TInt64.DTYPE);
+      Shape<TInt64> tfshape = Shape.create(scope, actual, TInt64.class);
+      Operand<TInt64> size = Shapes.size(scope, tfshape, TInt64.class);
 
       AtomicInteger index = new AtomicInteger();
-      try (TInt64 result1 =
-          session.runner().fetch(size.asOutput()).run().get(0).expect(TInt64.DTYPE)) {
+      try (TInt64 result1 = session.runner().fetch(size.asOutput()).run().single()) {
         result1.scalars().forEach(s -> assertEquals(8, s.getLong()));
       }
     }
@@ -117,19 +111,19 @@ public void testSize_Shape_Operand() {
 
       Operand<TInt32> size = Shapes.size(scope, tfshape, Constant.scalarOf(scope, 0));
       try (TInt32 result =
-          session.runner().fetch(size.asOutput()).run().get(0).expect(TInt32.DTYPE)) {
+          session.runner().fetch(size.asOutput()).run().single()) {
         result.scalars().forEach(s -> assertEquals(4, s.getInt()));
       }
 
       size = Shapes.size(scope, tfshape, Constant.scalarOf(scope, 1));
       try (TInt32 result =
-          session.runner().fetch(size.asOutput()).run().get(0).expect(TInt32.DTYPE)) {
+          session.runner().fetch(size.asOutput()).run().single()) {
         result.scalars().forEach(s -> assertEquals(2, s.getInt()));
       }
 
       size = Shapes.size(scope, tfshape, Constant.scalarOf(scope, 2));
       try (TInt32 result =
-          session.runner().fetch(size.asOutput()).run().get(0).expect(TInt32.DTYPE)) {
+          session.runner().fetch(size.asOutput()).run().single()) {
         result.scalars().forEach(s -> assertEquals(1, s.getInt()));
       }
     }
@@ -147,19 +141,19 @@ public void testSize_Operand_Operand() {
 
       Operand<TInt32> size = Shapes.size(scope, actual, Constant.scalarOf(scope, 0));
       try (TInt32 result =
-          session.runner().fetch(size.asOutput()).run().get(0).expect(TInt32.DTYPE)) {
+          session.runner().fetch(size.asOutput()).run().single()) {
         result.scalars().forEach(s -> assertEquals(4, s.getInt()));
       }
 
       size = Shapes.size(scope, actual, Constant.scalarOf(scope, 1));
       try (TInt32 result =
-          session.runner().fetch(size.asOutput()).run().get(0).expect(TInt32.DTYPE)) {
+          session.runner().fetch(size.asOutput()).run().single()) {
         result.scalars().forEach(s -> assertEquals(2, s.getInt()));
       }
 
       size = Shapes.size(scope, actual, Constant.scalarOf(scope, 2));
       try (TInt32 result =
-          session.runner().fetch(size.asOutput()).run().get(0).expect(TInt32.DTYPE)) {
+          session.runner().fetch(size.asOutput()).run().single()) {
         result.scalars().forEach(s -> assertEquals(1, s.getInt()));
       }
     }
@@ -178,7 +172,7 @@ public void testNumDimensions() {
 
       Operand<TInt32> nDims = Shapes.numDimensions(scope, tfshape);
       try (TInt32 result =
-          session.runner().fetch(nDims.asOutput()).run().get(0).expect(TInt32.DTYPE)) {
+          session.runner().fetch(nDims.asOutput()).run().single()) {
         result.scalars().forEach(s -> assertEquals(3, s.getInt()));
       }
     }
@@ -275,7 +269,7 @@ public void testSqueeze() {
       AtomicInteger index = new AtomicInteger();
       int[] expected = {4, 2};
       try (TInt32 result =
-          session.runner().fetch(squeezed.asOutput()).run().get(0).expect(TInt32.DTYPE)) {
+          session.runner().fetch(squeezed.asOutput()).run().single()) {
         result
 
             .scalars()
@@ -302,7 +296,7 @@ public void testHead() {
       AtomicInteger index = new AtomicInteger();
       int[] expected = {4};
       try (TInt32 result =
-          session.runner().fetch(head.asOutput()).run().get(0).expect(TInt32.DTYPE)) {
+          session.runner().fetch(head.asOutput()).run().single()) {
         result
 
             .scalars()
@@ -329,7 +323,7 @@ public void testTake() {
       AtomicInteger index = new AtomicInteger();
       int[] expected = {4, 1};
       try (TInt32 result =
-          session.runner().fetch(take.asOutput()).run().get(0).expect(TInt32.DTYPE)) {
+          session.runner().fetch(take.asOutput()).run().single()) {
         result
 
             .scalars()
@@ -356,7 +350,7 @@ public void testTail() {
       AtomicInteger index = new AtomicInteger();
       int[] expected = {1};
       try (TInt32 result =
-          session.runner().fetch(tail.asOutput()).run().get(0).expect(TInt32.DTYPE)) {
+          session.runner().fetch(tail.asOutput()).run().single()) {
         result
 
             .scalars()
@@ -383,7 +377,7 @@ public void testTakeLast() {
       AtomicInteger index = new AtomicInteger();
       int[] expected = {1, 2, 1};
       try (TInt32 result =
-          session.runner().fetch(takeLast.asOutput()).run().get(0).expect(TInt32.DTYPE)) {
+          session.runner().fetch(takeLast.asOutput()).run().single()) {
         result
 
             .scalars()
@@ -409,7 +403,7 @@ public void testPrependInt() {
       AtomicInteger index = new AtomicInteger();
       int[] expected = {3, 4, 2};
       try (TInt32 result =
-          session.runner().fetch(prepend.asOutput()).run().get(0).expect(TInt32.DTYPE)) {
+          session.runner().fetch(prepend.asOutput()).run().single()) {
         result
 
             .scalars()
@@ -429,13 +423,13 @@ public void testPrependLong() {
       Scope scope = new Scope(g);
       Operand operand = Constant.arrayOf(scope, new float[] {1, 2, 3, 4, 5, 6, 7, 8});
       Operand actual = Reshape.create(scope, operand, Constant.vectorOf(scope, new long[] {4, 2}));
-      Shape<TInt64> tfshape = Shape.create(scope, actual, TInt64.DTYPE);
+      Shape<TInt64> tfshape = Shape.create(scope, actual, TInt64.class);
 
       Operand<TInt64> prepend = Shapes.prepend(scope, tfshape, 1L);
       AtomicInteger index = new AtomicInteger();
       long[] expected = {1, 4, 2};
       try (TInt64 result =
-          session.runner().fetch(prepend.asOutput()).run().get(0).expect(TInt64.DTYPE)) {
+          session.runner().fetch(prepend.asOutput()).run().single()) {
         result
 
             .scalars()
@@ -466,7 +460,7 @@ public void testPrependShapeTInt32() {
       AtomicInteger index = new AtomicInteger();
       int[] expected = {2, 4, 4, 2};
       try (TInt32 result =
-          session.runner().fetch(prepend.asOutput()).run().get(0).expect(TInt32.DTYPE)) {
+          session.runner().fetch(prepend.asOutput()).run().single()) {
         result
 
             .scalars()
@@ -490,14 +484,14 @@ public void testPrependShapeTInt64() {
       Operand operand2 = Constant.arrayOf(scope, new float[] {1, 2, 3, 4, 5, 6, 7, 8});
       Operand actual2 =
           Reshape.create(scope, operand2, Constant.vectorOf(scope, new long[] {2, 4}));
-      Shape<TInt64> tfshape1 = Shape.create(scope, actual1, TInt64.DTYPE);
-      Shape<TInt64> tfshape2 = Shape.create(scope, actual2, TInt64.DTYPE);
+      Shape<TInt64> tfshape1 = Shape.create(scope, actual1, TInt64.class);
+      Shape<TInt64> tfshape2 = Shape.create(scope, actual2, TInt64.class);
 
       Operand<TInt64> prepend = Shapes.prepend(scope, tfshape1, tfshape2);
       AtomicInteger index = new AtomicInteger();
       long[] expected = {2, 4, 4, 2};
       try (TInt64 result =
-          session.runner().fetch(prepend.asOutput()).run().get(0).expect(TInt64.DTYPE)) {
+          session.runner().fetch(prepend.asOutput()).run().single()) {
         result
 
             .scalars()
@@ -517,13 +511,13 @@ public void testAppendLong() {
       Scope scope = new Scope(g);
       Operand operand = Constant.arrayOf(scope, new float[] {1, 2, 3, 4, 5, 6, 7, 8});
       Operand actual = Reshape.create(scope, operand, Constant.vectorOf(scope, new long[] {4, 2}));
-      Shape<TInt64> tfshape = Shape.create(scope, actual, TInt64.DTYPE);
+      Shape<TInt64> tfshape = Shape.create(scope, actual, TInt64.class);
 
       Operand<TInt64> append = Shapes.append(scope, tfshape, 2L);
       AtomicInteger index = new AtomicInteger();
       long[] expected = {4L, 2L, 2L};
       try (TInt64 result =
-          session.runner().fetch(append.asOutput()).run().get(0).expect(TInt64.DTYPE)) {
+          session.runner().fetch(append.asOutput()).run().single()) {
         result
 
             .scalars()
@@ -549,7 +543,7 @@ public void testAppendInt() {
       AtomicInteger index = new AtomicInteger();
       int[] expected = {4, 2, 2};
       try (TInt32 result =
-          session.runner().fetch(append.asOutput()).run().get(0).expect(TInt32.DTYPE)) {
+          session.runner().fetch(append.asOutput()).run().single()) {
         result
 
             .scalars()
@@ -580,7 +574,7 @@ public void testAppendShapeTInt32() {
       AtomicInteger index = new AtomicInteger();
       int[] expected = {4, 2, 2, 4};
       try (TInt32 result =
-          session.runner().fetch(append.asOutput()).run().get(0).expect(TInt32.DTYPE)) {
+          session.runner().fetch(append.asOutput()).run().single()) {
         result
 
             .scalars()
@@ -604,16 +598,15 @@ public void testAppendShapeTInt64() {
       Operand operand2 = Constant.arrayOf(scope, new float[] {1, 2, 3, 4, 5, 6, 7, 8});
       Operand actual2 =
           Reshape.create(scope, operand2, Constant.vectorOf(scope, new long[] {2, 4}));
-      Shape<TInt64> tfshape1 = Shape.create(scope, actual1, TInt64.DTYPE);
-      Shape<TInt64> tfshape2 = Shape.create(scope, actual2, TInt64.DTYPE);
+      Shape<TInt64> tfshape1 = Shape.create(scope, actual1, TInt64.class);
+      Shape<TInt64> tfshape2 = Shape.create(scope, actual2, TInt64.class);
 
       Operand<TInt64> append = Shapes.append(scope, tfshape1, tfshape2);
       AtomicInteger index = new AtomicInteger();
       long[] expected = {4, 2, 2, 4};
       try (TInt64 result =
-          session.runner().fetch(append.asOutput()).run().get(0).expect(TInt64.DTYPE)) {
+          session.runner().fetch(append.asOutput()).run().single()) {
         result
-
             .scalars()
             .forEach(
                 s -> {
diff --git a/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/op/core/ZerosTest.java b/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/op/core/ZerosTest.java
index 55dcef5ae3c..f84c7ce5530 100644
--- a/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/op/core/ZerosTest.java
+++ b/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/op/core/ZerosTest.java
@@ -19,12 +19,11 @@
 import static org.junit.jupiter.api.Assertions.assertFalse;
 import static org.junit.jupiter.api.Assertions.assertTrue;
 
-import java.util.List;
 import org.junit.jupiter.api.Test;
 import org.tensorflow.Graph;
 import org.tensorflow.Session;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.Scope;
+import org.tensorflow.util.TensorList;
 import org.tensorflow.types.TBool;
 import org.tensorflow.types.TFloat32;
 import org.tensorflow.types.TFloat64;
@@ -41,8 +40,8 @@ public void createIntZeros() {
         Session sess = new Session(g)) {
       Scope scope = new Scope(g);
       long[] shape = {2, 2};
-      Zeros<TInt32> op = Zeros.create(scope, Constant.vectorOf(scope, shape), TInt32.DTYPE);
-      try (TInt32 result = sess.runner().fetch(op).run().get(0).expect(TInt32.DTYPE)) {
+      Zeros<TInt32> op = Zeros.create(scope, Constant.vectorOf(scope, shape), TInt32.class);
+      try (TInt32 result = sess.runner().fetch(op).run().single()) {
         result.scalars().forEach(s -> assertEquals(0, s.getInt()));
       }
     }
@@ -54,8 +53,8 @@ public void createFloatZeros() {
         Session sess = new Session(g)) {
       Scope scope = new Scope(g);
       long[] shape = {2, 2};
-      Zeros<TFloat32> op = Zeros.create(scope, Constant.vectorOf(scope, shape), TFloat32.DTYPE);
-      try (TFloat32 result = sess.runner().fetch(op.asOutput()).run().get(0).expect(TFloat32.DTYPE)) {
+      Zeros<TFloat32> op = Zeros.create(scope, Constant.vectorOf(scope, shape), TFloat32.class);
+      try (TFloat32 result = sess.runner().fetch(op.asOutput()).run().single()) {
         result.scalars().forEach(s -> assertEquals(0.0f, s.getFloat(), 0));
       }
     }
@@ -67,8 +66,8 @@ public void createDoubleZeros() {
         Session sess = new Session(g)) {
       Scope scope = new Scope(g);
       long[] shape = {2, 2};
-      Zeros<TFloat64> op = Zeros.create(scope, Constant.vectorOf(scope, shape), TFloat64.DTYPE);
-      try (TFloat64 result = sess.runner().fetch(op.asOutput()).run().get(0).expect(TFloat64.DTYPE)) {
+      Zeros<TFloat64> op = Zeros.create(scope, Constant.vectorOf(scope, shape), TFloat64.class);
+      try (TFloat64 result = sess.runner().fetch(op.asOutput()).run().single()) {
         result.scalars().forEach(s -> assertEquals(0.0f, s.getDouble(), 0));
       }
     }
@@ -80,8 +79,8 @@ public void createLongZeros() {
         Session sess = new Session(g)) {
       Scope scope = new Scope(g);
       long[] shape = {2, 2};
-      Zeros<TInt64> op = Zeros.create(scope, Constant.vectorOf(scope, shape), TInt64.DTYPE);
-      try (TInt64 result = sess.runner().fetch(op.asOutput()).run().get(0).expect(TInt64.DTYPE)) {
+      Zeros<TInt64> op = Zeros.create(scope, Constant.vectorOf(scope, shape), TInt64.class);
+      try (TInt64 result = sess.runner().fetch(op.asOutput()).run().single()) {
         result.scalars().forEach(s -> assertEquals(0L, s.getLong()));
       }
     }
@@ -93,8 +92,8 @@ public void createBooleanZeros() {
         Session sess = new Session(g)) {
       Scope scope = new Scope(g);
       long[] shape = {2, 2};
-      Zeros<TBool> op = Zeros.create(scope, Constant.vectorOf(scope, shape), TBool.DTYPE);
-      try (TBool result = sess.runner().fetch(op.asOutput()).run().get(0).expect(TBool.DTYPE)) {
+      Zeros<TBool> op = Zeros.create(scope, Constant.vectorOf(scope, shape), TBool.class);
+      try (TBool result = sess.runner().fetch(op.asOutput()).run().single()) {
         result.scalars().forEach(s -> assertFalse(s.getBoolean()));
       }
    }
@@ -106,8 +105,8 @@ public void createUint8Zeros() {
         Session sess = new Session(g)) {
       Scope scope = new Scope(g);
       long[] shape = {2, 2};
-      Zeros<TUint8> op = Zeros.create(scope, Constant.vectorOf(scope, shape), TUint8.DTYPE);
-      try (TUint8 result = sess.runner().fetch(op.asOutput()).run().get(0).expect(TUint8.DTYPE)) {
+      Zeros<TUint8> op = Zeros.create(scope, Constant.vectorOf(scope, shape), TUint8.class);
+      try (TUint8 result = sess.runner().fetch(op.asOutput()).run().single()) {
         result.scalars().forEach(s -> assertEquals(0, s.getByte()));
       }
     }
@@ -119,8 +118,8 @@ public void createStringZeros() {
         Session sess = new Session(g)) {
       Scope scope = new Scope(g);
       long[] shape = {2, 2};
-      Zeros<TString> op = Zeros.create(scope, Constant.vectorOf(scope, shape), TString.DTYPE);
-      try (TString result = sess.runner().fetch(op.asOutput()).run().get(0).expect(TString.DTYPE)) {
+      Zeros<TString> op = Zeros.create(scope, Constant.vectorOf(scope, shape), TString.class);
+      try (TString result = sess.runner().fetch(op.asOutput()).run().single()) {
         result.scalars().forEach(s -> assertTrue(s.getObject().isEmpty()));
       }
     }
@@ -132,8 +131,10 @@ public void operationsComposingZerosAreCorrectlyNamed() {
         Session sess = new Session(g)) {
       Scope scope = new Scope(g);
       long[] shape = {2, 2};
-      Zeros<TFloat32> zeros = Zeros.create(scope.withSubScope("test"), Constant.vectorOf(scope, shape), TFloat32.DTYPE);
-      List<Tensor<?>> results = sess.runner().addTarget("test/Zeros/Zero").addTarget("test/Zeros/Fill").run();
+      Zeros<TFloat32> zeros = Zeros.create(scope.withSubScope("test"), Constant.vectorOf(scope, shape), TFloat32.class);
+      try (TensorList results =
+          sess.runner().addTarget("test/Zeros/Zero").addTarget("test/Zeros/Fill").run()) {
+      }
     }
   }
 }
diff --git a/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/types/NumericTypesTestBase.java b/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/types/NumericTypesTestBase.java
index 3c7dbea1450..cff8207a706 100644
--- a/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/types/NumericTypesTestBase.java
+++ b/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/types/NumericTypesTestBase.java
@@ -20,19 +20,14 @@
 import static org.junit.jupiter.api.Assertions.assertEquals;
 
 import org.junit.jupiter.api.Test;
-import org.tensorflow.DataType;
 import org.tensorflow.EagerSession;
-import org.tensorflow.ndarray.IntNdArray;
 import org.tensorflow.ndarray.NdArray;
-import org.tensorflow.ndarray.NdArrays;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.ndarray.index.Indices;
 import org.tensorflow.op.Ops;
 import org.tensorflow.op.core.Constant;
 import org.tensorflow.op.math.Add;
-import org.tensorflow.op.math.Pow;
 import org.tensorflow.op.math.Sub;
-import org.tensorflow.tensor.IntTensor;
 import org.tensorflow.types.family.TNumber;
 
 abstract class NumericTypesTestBase<T extends TNumber, U> {
@@ -97,7 +92,7 @@ public void setAndCompute() {
       try (EagerSession session = EagerSession.create()) {
         Ops tf = Ops.create(session);
 
-        Add<T> add = tf.math.add(tensor, tensor);
+        Add<T> add = tf.math.add(tf.capture(tensor), tf.capture(tensor));
         T result = add.asTensor();
 
         assertEquals(valueOf(0), result.getObject(0, 0));
diff --git a/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/types/TBfloat16Test.java b/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/types/TBfloat16Test.java
index 70051fd1fc7..17a6e0dd2b5 100644
--- a/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/types/TBfloat16Test.java
+++ b/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/types/TBfloat16Test.java
@@ -17,7 +17,6 @@
 
 package org.tensorflow.types;
 
-import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.NdArray;
 import org.tensorflow.ndarray.NdArrays;
 import org.tensorflow.ndarray.Shape;
diff --git a/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/types/TFloat16Test.java b/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/types/TFloat16Test.java
index 7fd65ba9654..c1ae8ad3b6d 100644
--- a/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/types/TFloat16Test.java
+++ b/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/types/TFloat16Test.java
@@ -17,7 +17,6 @@
 
 package org.tensorflow.types;
 
-import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.NdArray;
 import org.tensorflow.ndarray.NdArrays;
 import org.tensorflow.ndarray.Shape;
diff --git a/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/types/TFloat32Test.java b/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/types/TFloat32Test.java
index 938179ab8f3..8df96f2871a 100644
--- a/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/types/TFloat32Test.java
+++ b/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/types/TFloat32Test.java
@@ -17,7 +17,6 @@
 
 package org.tensorflow.types;
 
-import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.NdArray;
 import org.tensorflow.ndarray.NdArrays;
 import org.tensorflow.ndarray.Shape;
diff --git a/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/types/TFloat64Test.java b/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/types/TFloat64Test.java
index 5eaf391692c..47b4b6d936a 100644
--- a/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/types/TFloat64Test.java
+++ b/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/types/TFloat64Test.java
@@ -17,7 +17,6 @@
 
 package org.tensorflow.types;
 
-import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.NdArray;
 import org.tensorflow.ndarray.NdArrays;
 import org.tensorflow.ndarray.Shape;
diff --git a/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/types/TInt32Test.java b/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/types/TInt32Test.java
index 5cae9392122..9ea7f952f04 100644
--- a/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/types/TInt32Test.java
+++ b/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/types/TInt32Test.java
@@ -19,17 +19,9 @@
 
 import static org.junit.jupiter.api.Assertions.assertEquals;
 
-import org.junit.jupiter.api.Test;
-import org.tensorflow.EagerSession;
-import org.tensorflow.Tensor;
-import org.tensorflow.ndarray.IntNdArray;
 import org.tensorflow.ndarray.NdArray;
 import org.tensorflow.ndarray.NdArrays;
 import org.tensorflow.ndarray.Shape;
-import org.tensorflow.ndarray.index.Indices;
-import org.tensorflow.op.Ops;
-import org.tensorflow.op.core.Constant;
-import org.tensorflow.tensor.IntTensor;
 
 public class TInt32Test extends NumericTypesTestBase<TInt32, Integer> {
 
diff --git a/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/types/TInt64Test.java b/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/types/TInt64Test.java
index 23af4f10c6a..a88f3fb4d6d 100644
--- a/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/types/TInt64Test.java
+++ b/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/types/TInt64Test.java
@@ -17,7 +17,6 @@
 
 package org.tensorflow.types;
 
-import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.NdArray;
 import org.tensorflow.ndarray.NdArrays;
 import org.tensorflow.ndarray.Shape;
diff --git a/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/types/TStringTest.java b/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/types/TStringTest.java
index 4872f9be11f..cb4260b1b4c 100644
--- a/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/types/TStringTest.java
+++ b/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/types/TStringTest.java
@@ -23,7 +23,6 @@
 
 import java.nio.charset.StandardCharsets;
 import org.junit.jupiter.api.Test;
-import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.ndarray.NdArray;
 import org.tensorflow.ndarray.NdArrays;
diff --git a/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/types/TUint8Test.java b/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/types/TUint8Test.java
index e2ba52ef10b..ce7397d5878 100644
--- a/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/types/TUint8Test.java
+++ b/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/types/TUint8Test.java
@@ -17,7 +17,6 @@
 
 package org.tensorflow.types;
 
-import org.tensorflow.Tensor;
 import org.tensorflow.ndarray.NdArray;
 import org.tensorflow.ndarray.NdArrays;
 import org.tensorflow.ndarray.Shape;
diff --git a/tensorflow-framework/src/main/java/org/tensorflow/framework/utils/ShapeUtils.java b/tensorflow-framework/src/main/java/org/tensorflow/framework/utils/ShapeUtils.java
index 122de9f21ae..4abf90c4758 100644
--- a/tensorflow-framework/src/main/java/org/tensorflow/framework/utils/ShapeUtils.java
+++ b/tensorflow-framework/src/main/java/org/tensorflow/framework/utils/ShapeUtils.java
@@ -18,6 +18,7 @@
 import org.tensorflow.ndarray.NdArray;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.Scope;
+import org.tensorflow.tensor.Tensor;
 import org.tensorflow.types.TInt32;
 import org.tensorflow.types.TInt64;
 import org.tensorflow.types.TUint8;
diff --git a/tensorflow-framework/src/test/java/org/tensorflow/framework/data/BatchDatasetTest.java b/tensorflow-framework/src/test/java/org/tensorflow/framework/data/BatchDatasetTest.java
index e3c5c6ad06b..137b07e9289 100644
--- a/tensorflow-framework/src/test/java/org/tensorflow/framework/data/BatchDatasetTest.java
+++ b/tensorflow-framework/src/test/java/org/tensorflow/framework/data/BatchDatasetTest.java
@@ -17,7 +17,6 @@
 
 import org.junit.jupiter.api.Test;
 import org.tensorflow.Operand;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.Ops;
 import org.tensorflow.types.TInt32;
 
diff --git a/tensorflow-framework/src/test/java/org/tensorflow/framework/data/DatasetIteratorTest.java b/tensorflow-framework/src/test/java/org/tensorflow/framework/data/DatasetIteratorTest.java
index d89419e8a3d..fd265bf1826 100644
--- a/tensorflow-framework/src/test/java/org/tensorflow/framework/data/DatasetIteratorTest.java
+++ b/tensorflow-framework/src/test/java/org/tensorflow/framework/data/DatasetIteratorTest.java
@@ -20,7 +20,7 @@
 import org.tensorflow.Graph;
 import org.tensorflow.Operand;
 import org.tensorflow.Session;
-import org.tensorflow.Tensor;
+import org.tensorflow.tensor.Tensor;
 import org.tensorflow.exceptions.TFOutOfRangeException;
 import org.tensorflow.op.Ops;
 import org.tensorflow.types.TInt32;
diff --git a/tensorflow-framework/src/test/java/org/tensorflow/framework/data/MapDatasetTest.java b/tensorflow-framework/src/test/java/org/tensorflow/framework/data/MapDatasetTest.java
index c1dc0a98097..3b9f89d05d6 100644
--- a/tensorflow-framework/src/test/java/org/tensorflow/framework/data/MapDatasetTest.java
+++ b/tensorflow-framework/src/test/java/org/tensorflow/framework/data/MapDatasetTest.java
@@ -21,7 +21,7 @@
 import org.tensorflow.Graph;
 import org.tensorflow.Operand;
 import org.tensorflow.Session;
-import org.tensorflow.Tensor;
+import org.tensorflow.tensor.Tensor;
 import org.tensorflow.exceptions.TFOutOfRangeException;
 import org.tensorflow.op.Ops;
 import org.tensorflow.ndarray.IntNdArray;
diff --git a/tensorflow-framework/src/test/java/org/tensorflow/framework/data/SkipDatasetTest.java b/tensorflow-framework/src/test/java/org/tensorflow/framework/data/SkipDatasetTest.java
index eba3a273380..58a4ea93964 100644
--- a/tensorflow-framework/src/test/java/org/tensorflow/framework/data/SkipDatasetTest.java
+++ b/tensorflow-framework/src/test/java/org/tensorflow/framework/data/SkipDatasetTest.java
@@ -17,7 +17,6 @@
 
 import org.junit.jupiter.api.Test;
 import org.tensorflow.Operand;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.Ops;
 import org.tensorflow.types.TInt32;
 
diff --git a/tensorflow-framework/src/test/java/org/tensorflow/framework/data/TakeDatasetTest.java b/tensorflow-framework/src/test/java/org/tensorflow/framework/data/TakeDatasetTest.java
index e07272e3718..a4e94d75ca9 100644
--- a/tensorflow-framework/src/test/java/org/tensorflow/framework/data/TakeDatasetTest.java
+++ b/tensorflow-framework/src/test/java/org/tensorflow/framework/data/TakeDatasetTest.java
@@ -17,7 +17,6 @@
 
 import org.junit.jupiter.api.Test;
 import org.tensorflow.Operand;
-import org.tensorflow.Tensor;
 import org.tensorflow.op.Ops;
 import org.tensorflow.types.TInt32;
 
diff --git a/tensorflow-framework/src/test/java/org/tensorflow/framework/optimizers/AdamTest.java b/tensorflow-framework/src/test/java/org/tensorflow/framework/optimizers/AdamTest.java
index 461fa75397f..a6cec420bd7 100644
--- a/tensorflow-framework/src/test/java/org/tensorflow/framework/optimizers/AdamTest.java
+++ b/tensorflow-framework/src/test/java/org/tensorflow/framework/optimizers/AdamTest.java
@@ -16,7 +16,7 @@
 
 import org.junit.jupiter.api.*;
 import org.tensorflow.Graph;
-import org.tensorflow.Tensor;
+import org.tensorflow.tensor.Tensor;
 import org.tensorflow.framework.utils.ND;
 import org.tensorflow.framework.utils.TestSession;
 import org.tensorflow.ndarray.FloatNdArray;
diff --git a/tensorflow-framework/src/test/java/org/tensorflow/framework/optimizers/AdamaxTest.java b/tensorflow-framework/src/test/java/org/tensorflow/framework/optimizers/AdamaxTest.java
index de17395f76a..541f82dedcc 100644
--- a/tensorflow-framework/src/test/java/org/tensorflow/framework/optimizers/AdamaxTest.java
+++ b/tensorflow-framework/src/test/java/org/tensorflow/framework/optimizers/AdamaxTest.java
@@ -16,7 +16,7 @@
 
 import org.junit.jupiter.api.*;
 import org.tensorflow.Graph;
-import org.tensorflow.Tensor;
+import org.tensorflow.tensor.Tensor;
 import org.tensorflow.framework.utils.ND;
 import org.tensorflow.framework.utils.TestSession;
 import org.tensorflow.ndarray.FloatNdArray;
diff --git a/tensorflow-framework/src/test/java/org/tensorflow/framework/optimizers/NadamTest.java b/tensorflow-framework/src/test/java/org/tensorflow/framework/optimizers/NadamTest.java
index a583d74246b..88acd961d3a 100644
--- a/tensorflow-framework/src/test/java/org/tensorflow/framework/optimizers/NadamTest.java
+++ b/tensorflow-framework/src/test/java/org/tensorflow/framework/optimizers/NadamTest.java
@@ -16,7 +16,7 @@
 
 import org.junit.jupiter.api.*;
 import org.tensorflow.Graph;
-import org.tensorflow.Tensor;
+import org.tensorflow.tensor.Tensor;
 import org.tensorflow.framework.utils.ND;
 import org.tensorflow.framework.utils.TestSession;
 import org.tensorflow.ndarray.FloatNdArray;
diff --git a/tensorflow-framework/src/test/java/org/tensorflow/framework/utils/GraphTestSession.java b/tensorflow-framework/src/test/java/org/tensorflow/framework/utils/GraphTestSession.java
index 33ddec6dce3..11750e8ef47 100644
--- a/tensorflow-framework/src/test/java/org/tensorflow/framework/utils/GraphTestSession.java
+++ b/tensorflow-framework/src/test/java/org/tensorflow/framework/utils/GraphTestSession.java
@@ -19,6 +19,7 @@
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.Op;
 import org.tensorflow.op.Ops;
+import org.tensorflow.tensor.Tensor;
 import org.tensorflow.types.*;
 import org.tensorflow.types.family.TNumber;
 import org.tensorflow.types.family.TType;

From 5e486bd1f8c4c04ada183cd9b8ec73ca8f9b5d5d Mon Sep 17 00:00:00 2001
From: klessard <klessard@expedia.com>
Date: Fri, 30 Oct 2020 17:55:39 -0400
Subject: [PATCH 08/11] Update framework with new type system

---
 .../java/org/tensorflow/ndarray/NdArray.java  |  24 +-
 .../org/tensorflow/ndarray/NdArrayBase.java   |  47 ++
 .../src/bazel/op_generator/op_specs.cc        |  14 +-
 .../org/tensorflow/AbstractTypeFactory.java   |  12 +
 .../java/org/tensorflow/ConcreteFunction.java |   6 +-
 .../java/org/tensorflow/EagerOperation.java   |   2 +-
 .../org/tensorflow/EagerOperationBuilder.java |   6 +-
 .../src/main/java/org/tensorflow/Graph.java   |   7 +-
 .../org/tensorflow/GraphOperationBuilder.java |  10 +-
 .../src/main/java/org/tensorflow/Operand.java |  20 +
 .../java/org/tensorflow/OperationBuilder.java |   4 +-
 .../src/main/java/org/tensorflow/Output.java  |   2 +
 .../java/org/tensorflow/SavedModelBundle.java |   2 +-
 .../src/main/java/org/tensorflow/Session.java |  12 +-
 .../src/main/java/org/tensorflow/Tensor.java  |   8 +-
 .../java/org/tensorflow/TensorHandle.java     |  75 +-
 .../src/main/java/org/tensorflow/Tensors.java |   5 +-
 .../internal/types/TBfloat16Factory.java      |   8 +-
 .../internal/types/TBoolFactory.java          |   8 +-
 .../internal/types/TFloat16Factory.java       |   8 +-
 .../internal/types/TFloat32Factory.java       |   8 +-
 .../internal/types/TFloat64Factory.java       |   8 +-
 .../internal/types/TInt32Factory.java         |   8 +-
 .../internal/types/TInt64Factory.java         |   8 +-
 .../internal/types/TStringFactory.java        |   8 +-
 .../internal/types/TUint8Factory.java         |   9 +-
 .../java/org/tensorflow/types/TBfloat16.java  |   2 +-
 .../main/java/org/tensorflow/types/TBool.java |   2 +-
 .../java/org/tensorflow/types/TFloat16.java   |   2 +-
 .../java/org/tensorflow/types/TFloat32.java   |   2 +-
 .../java/org/tensorflow/types/TFloat64.java   |   2 +-
 .../java/org/tensorflow/types/TInt32.java     |   3 +-
 .../java/org/tensorflow/types/TInt64.java     |   3 +-
 .../java/org/tensorflow/types/TString.java    |   2 +-
 .../java/org/tensorflow/types/TUint8.java     |   3 +-
 .../tensorflow/types/family/TFloating.java    |   2 +-
 .../org/tensorflow/types/family/TInteger.java |   5 +
 .../org/tensorflow/types/family/TNumber.java  |   2 +-
 .../org/tensorflow/types/family/TType.java    |  13 +-
 .../java/org/tensorflow/util/TensorList.java  |   8 +-
 .../java/org/tensorflow/util/TensorMap.java   |  29 +-
 .../test/java/org/tensorflow/TensorTest.java  |  12 +-
 .../types/NumericTypesTestBase.java           |  26 +-
 .../tensorflow/framework/data/Dataset.java    |  12 +-
 .../framework/data/DatasetIterator.java       |  10 +-
 .../framework/data/DatasetOptional.java       |   8 +-
 .../framework/data/impl/BatchDataset.java     |   4 +-
 .../framework/data/impl/SkipDataset.java      |   4 +-
 .../framework/data/impl/TFRecordDataset.java  |   2 +-
 .../framework/data/impl/TakeDataset.java      |   4 +-
 .../data/impl/TensorSliceDataset.java         |   6 +-
 .../framework/data/impl/TextLineDataset.java  |   2 +-
 .../framework/initializers/Constant.java      |  15 +-
 .../framework/initializers/Glorot.java        |   3 +-
 .../tensorflow/framework/initializers/He.java |   3 +-
 .../framework/initializers/Identity.java      |  17 +-
 .../framework/initializers/Initializer.java   |   5 +-
 .../framework/initializers/LeCun.java         |   3 +-
 .../framework/initializers/Ones.java          |  11 +-
 .../framework/initializers/Orthogonal.java    |  20 +-
 .../framework/initializers/RandomNormal.java  |  17 +-
 .../framework/initializers/RandomUniform.java |  35 +-
 .../initializers/TruncatedNormal.java         |  17 +-
 .../initializers/VarianceScaling.java         |  31 +-
 .../framework/initializers/Zeros.java         |   3 +-
 .../framework/optimizers/AdaDelta.java        |  10 +-
 .../framework/optimizers/AdaGrad.java         |   4 +-
 .../framework/optimizers/AdaGradDA.java       |  12 +-
 .../tensorflow/framework/optimizers/Adam.java |  20 +-
 .../framework/optimizers/Adamax.java          |  16 +-
 .../tensorflow/framework/optimizers/Ftrl.java |  14 +-
 .../framework/optimizers/GradientDescent.java |   2 +-
 .../framework/optimizers/Momentum.java        |   6 +-
 .../framework/optimizers/Nadam.java           |  43 +-
 .../framework/optimizers/Optimizer.java       |   2 +-
 .../framework/optimizers/RMSProp.java         |  22 +-
 .../framework/utils/ShapeUtils.java           |  83 +--
 .../framework/data/BatchDatasetTest.java      |  18 +-
 .../framework/data/DatasetIteratorTest.java   |  27 +-
 .../framework/data/MapDatasetTest.java        |  28 +-
 .../framework/data/SkipDatasetTest.java       |   6 +-
 .../framework/data/TakeDatasetTest.java       |   6 +-
 .../framework/initializers/ConstantTest.java  |  18 +-
 .../framework/initializers/GlorotTest.java    |  34 +-
 .../framework/initializers/HeTest.java        |  34 +-
 .../framework/initializers/IdentityTest.java  |  25 +-
 .../framework/initializers/LeCunTest.java     |  34 +-
 .../framework/initializers/OnesTest.java      |  18 +-
 .../initializers/OrthogonalTest.java          |  31 +-
 .../initializers/RandomNormalTest.java        |  14 +-
 .../initializers/RandomUniformTest.java       |  18 +-
 .../initializers/TruncatedNormalTest.java     |  14 +-
 .../initializers/VarianceScalingTest.java     |  48 +-
 .../framework/initializers/ZerosTest.java     |  18 +-
 .../framework/optimizers/AdaDeltaTest.java    |   4 +-
 .../framework/optimizers/AdaGradDATest.java   |   4 +-
 .../framework/optimizers/AdaGradTest.java     |   4 +-
 .../framework/optimizers/AdamTest.java        |  19 +-
 .../framework/optimizers/AdamaxTest.java      |  12 +-
 .../framework/optimizers/FtrlTest.java        |  16 +-
 .../optimizers/GradientDescentTest.java       |   4 +-
 .../framework/optimizers/MomentumTest.java    |   8 +-
 .../framework/optimizers/NadamTest.java       |  19 +-
 .../framework/optimizers/RMSPropTest.java     |   4 +-
 .../framework/utils/EagerTestSession.java     | 320 ++++-----
 .../framework/utils/GraphTestSession.java     | 660 ++++++++----------
 106 files changed, 1154 insertions(+), 1224 deletions(-)
 create mode 100644 ndarray/src/main/java/org/tensorflow/ndarray/NdArrayBase.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/AbstractTypeFactory.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/family/TInteger.java

diff --git a/ndarray/src/main/java/org/tensorflow/ndarray/NdArray.java b/ndarray/src/main/java/org/tensorflow/ndarray/NdArray.java
index 5fe51121b13..dfabc0be8c7 100644
--- a/ndarray/src/main/java/org/tensorflow/ndarray/NdArray.java
+++ b/ndarray/src/main/java/org/tensorflow/ndarray/NdArray.java
@@ -55,29 +55,7 @@
  *
  * @param <T> the type of values to be mapped
  */
-public interface NdArray<T> {
-
-  /**
-   * @return the shape of this N-dimensional array
-   */
-  Shape shape();
-
-  /**
-   * @return the rank of this N-dimensional array
-   */
-  default int rank() {
-    return shape().numDimensions();
-  }
-
-  /**
-   * Computes and returns the total size of this N-dimensional array, in number of values.
-   *
-   * <p>For example, given a 3x3x2 matrix, the return value will be 18.
-   * @return total size of this nd array
-   */
-  default long size() {
-    return shape().size();
-  }
+public interface NdArray<T> extends NdArrayBase {
 
   /**
    * Returns a sequence of all elements at a given dimension.
diff --git a/ndarray/src/main/java/org/tensorflow/ndarray/NdArrayBase.java b/ndarray/src/main/java/org/tensorflow/ndarray/NdArrayBase.java
new file mode 100644
index 00000000000..75112ce2470
--- /dev/null
+++ b/ndarray/src/main/java/org/tensorflow/ndarray/NdArrayBase.java
@@ -0,0 +1,47 @@
+/*
+ Copyright 2019 The TensorFlow Authors. All Rights Reserved.
+
+ Licensed under the Apache License, Version 2.0 (the "License");
+ you may not use this file except in compliance with the License.
+ You may obtain a copy of the License at
+
+     http://www.apache.org/licenses/LICENSE-2.0
+
+ Unless required by applicable law or agreed to in writing, software
+ distributed under the License is distributed on an "AS IS" BASIS,
+ WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ See the License for the specific language governing permissions and
+ limitations under the License.
+ =======================================================================
+ */
+package org.tensorflow.ndarray;
+
+import java.util.function.BiConsumer;
+import java.util.function.Consumer;
+import org.tensorflow.ndarray.buffer.DataBuffer;
+import org.tensorflow.ndarray.index.Index;
+
+public interface NdArrayBase {
+
+  /**
+   * @return the shape of this N-dimensional array
+   */
+  Shape shape();
+
+  /**
+   * @return the rank of this N-dimensional array
+   */
+  default int rank() {
+    return shape().numDimensions();
+  }
+
+  /**
+   * Computes and returns the total size of this N-dimensional array, in number of values.
+   *
+   * <p>For example, given a 3x3x2 matrix, the return value will be 18.
+   * @return total size of this nd array
+   */
+  default long size() {
+    return shape().size();
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/bazel/op_generator/op_specs.cc b/tensorflow-core/tensorflow-core-api/src/bazel/op_generator/op_specs.cc
index 958e641610e..8f676deeead 100644
--- a/tensorflow-core/tensorflow-core-api/src/bazel/op_generator/op_specs.cc
+++ b/tensorflow-core/tensorflow-core-api/src/bazel/op_generator/op_specs.cc
@@ -86,14 +86,14 @@ class TypeResolver {
     if (next_generic_letter_ > 'Z') {
       next_generic_letter_ = 'A';
     }
-    Type generic_type = Type::Generic(string(1, generic_letter));
+    return Type::Generic(string(1, generic_letter));
+  }
+  Type TypeFamilyOf(const OpDef_AttrDef& attr_def) {
     // TODO(karllessard) support more type families
     if (IsRealNumbers(attr_def.allowed_values())) {
-      generic_type.add_supertype(Type::Interface("TNumber", "org.tensorflow.types.family"));
-    } else {
-      generic_type.add_supertype(Type::Interface("TType", "org.tensorflow.types.family"));
+      return Type::Interface("TNumber", "org.tensorflow.types.family");
     }
-    return generic_type;
+    return Type::Interface("TType", "org.tensorflow.types.family");
   }
 };
 
@@ -158,11 +158,11 @@ std::pair<Type, Type> TypeResolver::TypesOf(const OpDef_AttrDef& attr_def,
     types = MakeTypePair(Type::Class("Shape", "org.tensorflow.ndarray"));
 
   } else if (attr_type == "tensor") {
-    types = MakeTypePair(Type::Class("TType", "org.tensorflow.types.family")
-        .add_parameter(Type::Wildcard()));
+    types = MakeTypePair(Type::Class("TType", "org.tensorflow.types.family"));
 
   } else if (attr_type == "type") {
     Type type = *iterable_out ? Type::Wildcard() : NextGeneric(attr_def);
+    type.add_supertype(TypeFamilyOf(attr_def));
     types = MakeTypePair(type, Type::Class("Class"));
 
   } else {
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/AbstractTypeFactory.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/AbstractTypeFactory.java
new file mode 100644
index 00000000000..a1846f3dede
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/AbstractTypeFactory.java
@@ -0,0 +1,12 @@
+package org.tensorflow;
+
+import org.tensorflow.internal.c_api.TF_Tensor;
+import org.tensorflow.types.TypeFactory;
+import org.tensorflow.types.family.TType;
+
+public abstract class AbstractTypeFactory<T extends TType> implements TypeFactory<T> {
+
+  protected static TF_Tensor getNative(TensorHandle tensorHandle) {
+    return tensorHandle.get();
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/ConcreteFunction.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/ConcreteFunction.java
index a5a1e157e4b..de516eba81f 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/ConcreteFunction.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/ConcreteFunction.java
@@ -162,14 +162,14 @@ public Signature signature() {
    * @return output tensors resulting from the execution of the function,
    *         mapped by their signature name
    */
-  public TensorMap call(Map<String, TType<?>> arguments)
+  public TensorMap call(Map<String, TType> arguments)
       throws IllegalArgumentException {
 
     final SignatureDef signatureDef = signature.asSignatureDef();
     final Session.Runner runner = session.runner();
 
     signatureDef.getInputsMap().forEach((argName, t) -> {
-      TType<?> tensor = arguments.get(argName);
+      TType tensor = arguments.get(argName);
       if (tensor == null) {
         throw new IllegalArgumentException(String.format("Missing argument [%s]", argName));
       }
@@ -192,7 +192,7 @@ public TensorMap call(Map<String, TType<?>> arguments)
    * @throws IllegalArgumentException if there are multiple input or output parameters defined
    *                                  in the function
    */
-  public <T extends TType> T call(TType<?> tensor) throws IllegalArgumentException {
+  public <T extends TType> T call(TType tensor) throws IllegalArgumentException {
     final SignatureDef signatureDef = signature.asSignatureDef();
 
     if (signatureDef.getInputsCount() != 1) {
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/EagerOperation.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/EagerOperation.java
index db87f7e81c2..59da4d92dab 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/EagerOperation.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/EagerOperation.java
@@ -124,7 +124,7 @@ private TensorHandle resolveTensor(int outputIndex) {
     // instead.
     TensorHandle tensor = resolveTensorHandle(getUnsafeNativeHandle(outputIndex), session);
     if (!outputTensors.compareAndSet(outputIndex, null, tensor)) {
-      session.detach(tensor.nativeHandle());
+      session.detach(tensor.get());
       tensor = outputTensors.get(outputIndex);
     }
     return tensor;
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/EagerOperationBuilder.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/EagerOperationBuilder.java
index 7c682fa5383..c33d945c2d6 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/EagerOperationBuilder.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/EagerOperationBuilder.java
@@ -177,13 +177,13 @@ public EagerOperationBuilder setAttr(String name, Class<? extends TType>[] types
   }
 
   @Override
-  public EagerOperationBuilder setAttr(String name, TType<?> value) {
-    setAttrTensor(opHandle, name, value.tensorHandle().nativeHandle());
+  public EagerOperationBuilder setAttr(String name, TType value) {
+    setAttrTensor(opHandle, name, value.handle().get());
     return this;
   }
 
   @Override
-  public EagerOperationBuilder setAttr(String name, TType<?>[] values) {
+  public EagerOperationBuilder setAttr(String name, TType[] values) {
     // TODO (karllessard) could be supported by adding this attribute type in the eager C API
     throw new UnsupportedOperationException(
         "Tensor list attributes are not supported in eager mode");
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Graph.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Graph.java
index e9551f6c9b6..8a106983771 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Graph.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Graph.java
@@ -54,6 +54,7 @@
 import org.tensorflow.proto.framework.GraphDef;
 import org.tensorflow.proto.util.SaverDef;
 import org.tensorflow.types.TString;
+import org.tensorflow.types.family.TType;
 
 
 /**
@@ -766,8 +767,8 @@ private static SaverDef addVariableSaver(Graph graph) {
     Ops tf = Ops.create(graph).withSubScope("save");
 
     List<String> varNames = new ArrayList<>();
-    List<Operand<?>> varOutputs = new ArrayList<>();
-    List<Class<?>> varTypes = new ArrayList<>();
+    List<Operand<? extends TType>> varOutputs = new ArrayList<>();
+    List<Class<? extends TType>> varTypes = new ArrayList<>();
 
     for (Iterator<Operation> iter = graph.operations(); iter.hasNext();) {
       Operation op = iter.next();
@@ -798,7 +799,7 @@ private static SaverDef addVariableSaver(Graph graph) {
     );
     List<Op> restoreOps = new ArrayList<>(varOutputs.size());
     for (int i = 0; i < varOutputs.size(); ++i) {
-      restoreOps.add(tf.assign(varOutputs.get(i), (Operand) restoreVariables.tensors().get(i)));
+      restoreOps.add(tf.assign(varOutputs.get(i), (Operand)restoreVariables.tensors().get(i)));
     }
     NoOp restoreAll = tf.withControlDependencies(restoreOps).noOp();
 
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/GraphOperationBuilder.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/GraphOperationBuilder.java
index 80190624e17..a10ce0af010 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/GraphOperationBuilder.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/GraphOperationBuilder.java
@@ -249,10 +249,10 @@ public GraphOperationBuilder setAttr(String name, Class<? extends TType>[] types
   }
 
   @Override
-  public GraphOperationBuilder setAttr(String name, TType<?> value) {
+  public GraphOperationBuilder setAttr(String name, TType value) {
     Graph.Reference r = graph.ref();
     try {
-      setAttrTensor(unsafeNativeHandle, name, value.tensorHandle().nativeHandle());
+      setAttrTensor(unsafeNativeHandle, name, value.handle().get());
     } finally {
       r.close();
     }
@@ -260,11 +260,11 @@ public GraphOperationBuilder setAttr(String name, TType<?> value) {
   }
 
   @Override
-  public GraphOperationBuilder setAttr(String name, TType<?>[] value) {
+  public GraphOperationBuilder setAttr(String name, TType[] value) {
     TF_Tensor[] handles = new TF_Tensor[value.length];
     int idx = 0;
-    for (TType<?> t : value) {
-      handles[idx++] = t.tensorHandle().nativeHandle();
+    for (TType t : value) {
+      handles[idx++] = t.handle().get();
     }
     Graph.Reference r = graph.ref();
     try {
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Operand.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Operand.java
index 18a2385f492..8f7e7791ebf 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Operand.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Operand.java
@@ -15,8 +15,10 @@
 
 package org.tensorflow;
 
+import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.Op;
 import org.tensorflow.op.Scope;
+import org.tensorflow.types.TypeRegistry;
 import org.tensorflow.types.family.TType;
 
 /**
@@ -42,6 +44,16 @@
  */
 public interface Operand<T extends TType> extends Op {
 
+  /** Returns the (possibly partially known) shape of the tensor referred to by this Output. */
+  default Shape shape() {
+    return asOutput().shape();
+  }
+
+  /** Returns the DataType of the tensor referred to by this Output. */
+  default Class<T> type() {
+    return asOutput().type();
+  }
+
   /**
    * Returns the symbolic handle of the tensor.
    *
@@ -81,4 +93,12 @@ default Output<T> asOutput() {
   default T asTensor() {
     return asOutput().tensor();
   }
+
+  default <U extends TType> Operand<U> expect(Class<U> type) {
+    if (asOutput().type() != type) {
+      throw new IllegalArgumentException(
+          "Cannot cast from tensor of " + asOutput().type().getSimpleName() + " to tensor of " + type.getSimpleName());
+    }
+    return (Operand<U>)this;
+  }
 }
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/OperationBuilder.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/OperationBuilder.java
index 75787817a07..df6a80a79ca 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/OperationBuilder.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/OperationBuilder.java
@@ -196,7 +196,7 @@ public interface OperationBuilder {
    * @param value attribute value
    * @return the OperationBuilder instance for chaining.
    */
-   OperationBuilder setAttr(String name, TType<?> value);
+   OperationBuilder setAttr(String name, TType value);
 
   /**
    * Set the tensor values of an attribute of the operation being built.
@@ -205,7 +205,7 @@ public interface OperationBuilder {
    * @param value attribute values
    * @return the OperationBuilder instance for chaining.
    */
-   OperationBuilder setAttr(String name, TType<?>[] value);
+   OperationBuilder setAttr(String name, TType[] value);
 
   /**
    * Set the shape value of an attribute of the operation being built.
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Output.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Output.java
index 02ab4fa5a0f..692daaf6327 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Output.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Output.java
@@ -39,11 +39,13 @@ public int index() {
   }
 
   /** Returns the (possibly partially known) shape of the tensor referred to by this Output. */
+  @Override
   public Shape shape() {
     return operation.shape(index);
   }
 
   /** Returns the DataType of the tensor referred to by this Output. */
+  @Override
   public Class<T> type() {
     return (Class) TypeRegistry.find(operation.dtype(index)).typeClass();
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/SavedModelBundle.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/SavedModelBundle.java
index 97090a74276..6a5ae0b6467 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/SavedModelBundle.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/SavedModelBundle.java
@@ -336,7 +336,7 @@ public ConcreteFunction function(String signatureKey) {
    * @return list of output tensors, mapped by the signature name
    * @throws IllegalArgumentException if no function can be selected by default
    */
-  public TensorMap call(Map<String, TType<?>> arguments) {
+  public TensorMap call(Map<String, TType> arguments) {
     ConcreteFunction function = null;
     if (functions.size() == 1) {
       function = functions.values().iterator().next();
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Session.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Session.java
index 49502c769ea..07e70db6dc9 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Session.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Session.java
@@ -162,7 +162,7 @@ public final class Runner {
      * @param t the tensor substituting the operation
      * @return this session runner
      */
-    public Runner feed(String operation, TType<?> t) {
+    public Runner feed(String operation, TType t) {
       return feed(parseOutput(operation), t);
     }
 
@@ -177,11 +177,11 @@ public Runner feed(String operation, TType<?> t) {
      * @param t the tensor substituting the operation
      * @return this session runner
      */
-    public Runner feed(String operation, int index, TType<?> t) {
+    public Runner feed(String operation, int index, TType t) {
       Operation op = operationByName(operation);
       if (op != null) {
         inputs.add(op.output(index));
-        inputTensors.add(t.tensorHandle());
+        inputTensors.add(t.handle());
       }
       return this;
     }
@@ -194,9 +194,9 @@ public Runner feed(String operation, int index, TType<?> t) {
      * @param t the tensor substituting the operation
      * @return this session runner
      */
-    public Runner feed(Operand<?> operand, TType<?> t) {
+    public Runner feed(Operand<?> operand, TType t) {
       inputs.add(operand.asOutput());
-      inputTensors.add(t.tensorHandle());
+      inputTensors.add(t.handle());
       return this;
     }
 
@@ -359,7 +359,7 @@ private Run runHelper(boolean wantMetadata) {
       // validity of the Graph and graphRef ensures that.
       int idx = 0;
       for (TensorHandle t : inputTensors) {
-        inputTensorHandles[idx++] = t.nativeHandle();
+        inputTensorHandles[idx++] = t.get();
       }
       idx = 0;
       for (Output<?> o : inputs) {
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Tensor.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Tensor.java
index 63752b9920a..4f47bd06dd2 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Tensor.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Tensor.java
@@ -15,6 +15,7 @@
 
 package org.tensorflow;
 
+import org.tensorflow.ndarray.NdArrayBase;
 import org.tensorflow.ndarray.buffer.ByteDataBuffer;
 import org.tensorflow.proto.framework.DataType;
 
@@ -33,7 +34,12 @@
  * }
  * }</pre>
  */
-public interface Tensor extends AutoCloseable {
+public interface Tensor extends NdArrayBase, AutoCloseable {
+
+  /**
+   * Return the handle to the native tensor
+   */
+  TensorHandle handle();
 
   /**
    * Release resources associated with the Tensor.
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/TensorHandle.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/TensorHandle.java
index 0d2369897d1..7ac9ac291a8 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/TensorHandle.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/TensorHandle.java
@@ -15,44 +15,67 @@
 
 package org.tensorflow;
 
+import static org.tensorflow.internal.c_api.global.tensorflow.TF_Dim;
+import static org.tensorflow.internal.c_api.global.tensorflow.TF_NumDims;
 import static org.tensorflow.internal.c_api.global.tensorflow.TF_TensorByteSize;
+import static org.tensorflow.internal.c_api.global.tensorflow.TF_TensorElementCount;
 import static org.tensorflow.internal.c_api.global.tensorflow.TF_TensorType;
 
 import org.tensorflow.internal.c_api.TF_Tensor;
 import org.tensorflow.internal.buffer.TensorBuffers;
+import org.tensorflow.ndarray.Shape;
 import org.tensorflow.ndarray.buffer.ByteDataBuffer;
 import org.tensorflow.proto.framework.DataType;
 
 public final class TensorHandle implements Tensor {
 
+  @Override
+  public TensorHandle handle() {
+    return this;
+  }
+
+  public void retain() {
+    requireHandle(nativeHandle).retainReference();
+  }
+
+  public void release() {
+    requireHandle(nativeHandle).releaseReference();
+  }
+
+  @Override
+  public Shape shape() {
+    return Shape.of(shape(nativeHandle));
+  }
+
+  @Override
+  public int rank() {
+    return numDims(nativeHandle);
+  }
+
+  @Override
+  public long size() {
+    return numElements(nativeHandle);
+  }
+
   @Override
   public void close() {
-    nativeHandle().close();
+    get().close();
     this.nativeHandle = null;
   }
 
   @Override
   public long numBytes() {
-    return TF_TensorByteSize(nativeHandle());
+    return TF_TensorByteSize(get());
   }
 
   @Override
   public ByteDataBuffer rawData() {
-    return TensorBuffers.toBytes(nativeHandle(), true);
+    return TensorBuffers.toBytes(get(), true);
   }
 
   @Override
   public DataType dataType() {
-    return DataType.forNumber(dtype(nativeHandle()));
-  }
-
-  /**
-   * FIXME public??
-   * @return native handle to this tensor
-   * @throws IllegalStateException if tensor has been closed
-   */
-  public TF_Tensor nativeHandle() {
-    return requireHandle(nativeHandle);
+    return DataType.forNumber(dtype(get()));
   }
 
   static TensorHandle of(TF_Tensor nativeHandle) {
@@ -60,10 +83,14 @@ static TensorHandle of(TF_Tensor nativeHandle) {
   }
 
   void attachTo(EagerSession session) {
-    session.attach(nativeHandle());
+    session.attach(get());
     nativeHandle.releaseReference();
   }
 
+  TF_Tensor get() {
+    return requireHandle(nativeHandle);
+  }
+
   private TensorHandle(TF_Tensor nativeHandle) {
     this.nativeHandle = nativeHandle;
     nativeHandle.retainReference();
@@ -81,5 +108,25 @@ private static int dtype(TF_Tensor handle) {
     return TF_TensorType(handle);
   }
 
+  private static int numDims(TF_Tensor handle) {
+    requireHandle(handle);
+    return TF_NumDims(handle);
+  }
+
+  private static long numElements(TF_Tensor handle) {
+    requireHandle(handle);
+    return TF_TensorElementCount(handle);
+  }
+
+  private static long[] shape(TF_Tensor handle) {
+    requireHandle(handle);
+    int numDims = TF_NumDims(handle);
+    long[] dims = new long[numDims];
+    for (int i = 0; i < numDims; ++i) {
+      dims[i] = TF_Dim(handle, i);
+    }
+    return dims;
+  }
+
   private TF_Tensor nativeHandle;
 }
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Tensors.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Tensors.java
index 09af17471a4..22c6a769e0f 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Tensors.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Tensors.java
@@ -2,7 +2,6 @@
 
 import static org.tensorflow.internal.c_api.global.tensorflow.TF_Dim;
 import static org.tensorflow.internal.c_api.global.tensorflow.TF_NumDims;
-import static org.tensorflow.internal.c_api.global.tensorflow.TF_TensorType;
 
 import java.util.function.Consumer;
 import org.bytedeco.javacpp.PointerScope;
@@ -141,7 +140,7 @@ public static <T extends TType> T of(Class<T> type, Shape shape, long size, Cons
    */
   public static <T extends TType> T of(Class<T> type, Shape shape, ByteDataBuffer rawData) {
     T t = of(type, shape, rawData.size());
-    rawData.copyTo(TensorBuffers.toBytes(t.tensorHandle().nativeHandle()), rawData.size());
+    rawData.copyTo(TensorBuffers.toBytes(t.handle().get()), rawData.size());
     return t;
   }
 
@@ -152,7 +151,7 @@ public static <T extends TType> T of(Class<T> type, Shape shape, ByteDataBuffer
    */
   static <T extends TType> T fromHandle(TensorHandle handle) {
     Type<T> type = TypeRegistry.find(handle.dataType());
-    Shape shape = Shape.of(shape(handle.nativeHandle()));
+    Shape shape = Shape.of(shape(handle.get()));
     return type.factory().createDense(handle, shape);
   }
 
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/types/TBfloat16Factory.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/types/TBfloat16Factory.java
index 7898a9896b9..50542dc0dc7 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/types/TBfloat16Factory.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/types/TBfloat16Factory.java
@@ -1,5 +1,6 @@
 package org.tensorflow.internal.types;
 
+import org.tensorflow.AbstractTypeFactory;
 import org.tensorflow.TensorHandle;
 import org.tensorflow.internal.buffer.TensorBuffers;
 import org.tensorflow.ndarray.Shape;
@@ -7,16 +8,15 @@
 import org.tensorflow.ndarray.buffer.layout.DataLayouts;
 import org.tensorflow.ndarray.impl.dense.FloatDenseNdArray;
 import org.tensorflow.types.TBfloat16;
-import org.tensorflow.types.TypeFactory;
 
 /**
  * Factory of {@link TBfloat16} tensor instances
  */
-public class TBfloat16Factory implements TypeFactory<TBfloat16> {
+public class TBfloat16Factory extends AbstractTypeFactory<TBfloat16> {
 
   @Override
   public TBfloat16 createDense(TensorHandle tensorHandle, Shape shape) {
-    FloatDataBuffer buffer = DataLayouts.BFLOAT16.applyTo(TensorBuffers.toShorts(tensorHandle.nativeHandle()));
+    FloatDataBuffer buffer = DataLayouts.BFLOAT16.applyTo(TensorBuffers.toShorts(getNative(tensorHandle)));
     return new TBfloat16Impl(tensorHandle, buffer, shape);
   }
 
@@ -28,7 +28,7 @@ public Class<TBfloat16> type() {
     }
 
     @Override
-    public TensorHandle tensorHandle() {
+    public TensorHandle handle() {
       return tensorHandle;
     }
 
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/types/TBoolFactory.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/types/TBoolFactory.java
index 4e14bfd07a7..1d1552cc291 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/types/TBoolFactory.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/types/TBoolFactory.java
@@ -1,21 +1,21 @@
 package org.tensorflow.internal.types;
 
+import org.tensorflow.AbstractTypeFactory;
 import org.tensorflow.TensorHandle;
 import org.tensorflow.internal.buffer.TensorBuffers;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.ndarray.buffer.BooleanDataBuffer;
 import org.tensorflow.ndarray.impl.dense.BooleanDenseNdArray;
 import org.tensorflow.types.TBool;
-import org.tensorflow.types.TypeFactory;
 
 /**
  * Factory of {@link TBool} tensor instances
  */
-public class TBoolFactory implements TypeFactory<TBool> {
+public class TBoolFactory extends AbstractTypeFactory<TBool> {
 
   @Override
   public TBool createDense(TensorHandle tensorHandle, Shape shape) {
-    BooleanDataBuffer buffer = TensorBuffers.toBooleans(tensorHandle.nativeHandle());
+    BooleanDataBuffer buffer = TensorBuffers.toBooleans(getNative(tensorHandle));
     return new TBoolImpl(tensorHandle, buffer, shape);
   }
 
@@ -27,7 +27,7 @@ public Class<TBool> type() {
     }
 
     @Override
-    public TensorHandle tensorHandle() {
+    public TensorHandle handle() {
       return tensorHandle;
     }
 
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/types/TFloat16Factory.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/types/TFloat16Factory.java
index e54fb98c95d..73301f6894e 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/types/TFloat16Factory.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/types/TFloat16Factory.java
@@ -1,5 +1,6 @@
 package org.tensorflow.internal.types;
 
+import org.tensorflow.AbstractTypeFactory;
 import org.tensorflow.TensorHandle;
 import org.tensorflow.internal.buffer.TensorBuffers;
 import org.tensorflow.ndarray.Shape;
@@ -7,16 +8,15 @@
 import org.tensorflow.ndarray.buffer.layout.DataLayouts;
 import org.tensorflow.ndarray.impl.dense.FloatDenseNdArray;
 import org.tensorflow.types.TFloat16;
-import org.tensorflow.types.TypeFactory;
 
 /**
  * Factory of {@link TFloat16} tensor instances
  */
-public class TFloat16Factory implements TypeFactory<TFloat16> {
+public class TFloat16Factory extends AbstractTypeFactory<TFloat16> {
 
   @Override
   public TFloat16 createDense(TensorHandle tensorHandle, Shape shape) {
-    FloatDataBuffer buffer = DataLayouts.FLOAT16.applyTo(TensorBuffers.toShorts(tensorHandle.nativeHandle()));
+    FloatDataBuffer buffer = DataLayouts.FLOAT16.applyTo(TensorBuffers.toShorts(getNative(tensorHandle)));
     return new TFloat16Impl(tensorHandle, buffer, shape);
   }
 
@@ -28,7 +28,7 @@ public Class<TFloat16> type() {
     }
 
     @Override
-    public TensorHandle tensorHandle() {
+    public TensorHandle handle() {
       return tensorHandle;
     }
 
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/types/TFloat32Factory.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/types/TFloat32Factory.java
index 33365bcfe31..ae3a53f2c8a 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/types/TFloat32Factory.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/types/TFloat32Factory.java
@@ -1,21 +1,21 @@
 package org.tensorflow.internal.types;
 
+import org.tensorflow.AbstractTypeFactory;
 import org.tensorflow.TensorHandle;
 import org.tensorflow.internal.buffer.TensorBuffers;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.ndarray.buffer.FloatDataBuffer;
 import org.tensorflow.ndarray.impl.dense.FloatDenseNdArray;
 import org.tensorflow.types.TFloat32;
-import org.tensorflow.types.TypeFactory;
 
 /**
  * Factory of {@link TFloat32} tensor instances
  */
-public class TFloat32Factory implements TypeFactory<TFloat32> {
+public class TFloat32Factory extends AbstractTypeFactory<TFloat32> {
 
   @Override
   public TFloat32 createDense(TensorHandle tensorHandle, Shape shape) {
-    FloatDataBuffer buffer = TensorBuffers.toFloats(tensorHandle.nativeHandle());
+    FloatDataBuffer buffer = TensorBuffers.toFloats(getNative(tensorHandle));
     return new TFloat32Impl(tensorHandle, buffer, shape);
   }
 
@@ -27,7 +27,7 @@ public Class<TFloat32> type() {
     }
 
     @Override
-    public TensorHandle tensorHandle() {
+    public TensorHandle handle() {
       return tensorHandle;
     }
 
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/types/TFloat64Factory.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/types/TFloat64Factory.java
index 1223409b799..495f743e8cc 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/types/TFloat64Factory.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/types/TFloat64Factory.java
@@ -1,21 +1,21 @@
 package org.tensorflow.internal.types;
 
+import org.tensorflow.AbstractTypeFactory;
 import org.tensorflow.TensorHandle;
 import org.tensorflow.internal.buffer.TensorBuffers;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.ndarray.buffer.DoubleDataBuffer;
 import org.tensorflow.ndarray.impl.dense.DoubleDenseNdArray;
 import org.tensorflow.types.TFloat64;
-import org.tensorflow.types.TypeFactory;
 
 /**
  * Factory of {@link TFloat64} tensor instances
  */
-public class TFloat64Factory implements TypeFactory<TFloat64> {
+public class TFloat64Factory extends AbstractTypeFactory<TFloat64> {
 
   @Override
   public TFloat64 createDense(TensorHandle tensorHandle, Shape shape) {
-    DoubleDataBuffer buffer = TensorBuffers.toDoubles(tensorHandle.nativeHandle());
+    DoubleDataBuffer buffer = TensorBuffers.toDoubles(getNative(tensorHandle));
     return new TFloat64Impl(tensorHandle, buffer, shape);
   }
 
@@ -27,7 +27,7 @@ public Class<TFloat64> type() {
     }
 
     @Override
-    public TensorHandle tensorHandle() {
+    public TensorHandle handle() {
       return tensorHandle;
     }
 
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/types/TInt32Factory.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/types/TInt32Factory.java
index 7869befa03f..368f543d743 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/types/TInt32Factory.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/types/TInt32Factory.java
@@ -1,21 +1,21 @@
 package org.tensorflow.internal.types;
 
+import org.tensorflow.AbstractTypeFactory;
 import org.tensorflow.TensorHandle;
 import org.tensorflow.internal.buffer.TensorBuffers;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.ndarray.buffer.IntDataBuffer;
 import org.tensorflow.ndarray.impl.dense.IntDenseNdArray;
 import org.tensorflow.types.TInt32;
-import org.tensorflow.types.TypeFactory;
 
 /**
  * Factory of {@link TInt32} tensor instances
  */
-public class TInt32Factory implements TypeFactory<TInt32> {
+public class TInt32Factory extends AbstractTypeFactory<TInt32> {
 
   @Override
   public TInt32 createDense(TensorHandle tensorHandle, Shape shape) {
-    IntDataBuffer buffer = TensorBuffers.toInts(tensorHandle.nativeHandle());
+    IntDataBuffer buffer = TensorBuffers.toInts(getNative(tensorHandle));
     return new TInt32Impl(tensorHandle, buffer, shape);
   }
 
@@ -27,7 +27,7 @@ public Class<TInt32> type() {
     }
 
     @Override
-    public TensorHandle tensorHandle() {
+    public TensorHandle handle() {
       return tensorHandle;
     }
 
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/types/TInt64Factory.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/types/TInt64Factory.java
index cc37cada12d..6a6fc0317ec 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/types/TInt64Factory.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/types/TInt64Factory.java
@@ -1,21 +1,21 @@
 package org.tensorflow.internal.types;
 
+import org.tensorflow.AbstractTypeFactory;
 import org.tensorflow.TensorHandle;
 import org.tensorflow.internal.buffer.TensorBuffers;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.ndarray.buffer.LongDataBuffer;
 import org.tensorflow.ndarray.impl.dense.LongDenseNdArray;
 import org.tensorflow.types.TInt64;
-import org.tensorflow.types.TypeFactory;
 
 /**
  * Factory of {@link TInt64} tensor instances
  */
-public class TInt64Factory implements TypeFactory<TInt64> {
+public class TInt64Factory extends AbstractTypeFactory<TInt64> {
 
   @Override
   public TInt64 createDense(TensorHandle tensorHandle, Shape shape) {
-    LongDataBuffer buffer = TensorBuffers.toLongs(tensorHandle.nativeHandle());
+    LongDataBuffer buffer = TensorBuffers.toLongs(getNative(tensorHandle));
     return new Int64DenseTensor(tensorHandle, buffer, shape);
   }
 
@@ -27,7 +27,7 @@ public Class<TInt64> type() {
     }
 
     @Override
-    public TensorHandle tensorHandle() {
+    public TensorHandle handle() {
       return tensorHandle;
     }
 
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/types/TStringFactory.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/types/TStringFactory.java
index 69b764fbaa3..562662579fe 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/types/TStringFactory.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/types/TStringFactory.java
@@ -3,6 +3,7 @@
 import java.nio.charset.Charset;
 import java.nio.charset.StandardCharsets;
 import java.util.function.Function;
+import org.tensorflow.AbstractTypeFactory;
 import org.tensorflow.TensorHandle;
 import org.tensorflow.internal.buffer.ByteSequenceTensorBuffer;
 import org.tensorflow.internal.buffer.TensorBuffers;
@@ -14,16 +15,15 @@
 import org.tensorflow.ndarray.buffer.layout.DataLayouts;
 import org.tensorflow.ndarray.impl.dense.DenseNdArray;
 import org.tensorflow.types.TString;
-import org.tensorflow.types.TypeFactory;
 
 /**
  * Factory of {@link TString} tensor instances
  */
-public class TStringFactory implements TypeFactory<TString> {
+public class TStringFactory extends AbstractTypeFactory<TString> {
 
   @Override
   public TString createDense(TensorHandle tensorHandle, Shape shape) {
-    ByteSequenceTensorBuffer buffer = TensorBuffers.toStrings(tensorHandle.nativeHandle(), shape.size());
+    ByteSequenceTensorBuffer buffer = TensorBuffers.toStrings(getNative(tensorHandle), shape.size());
     return new TStringImpl(tensorHandle, buffer, shape);
   }
 
@@ -43,7 +43,7 @@ public Class<TString> type() {
     }
 
     @Override
-    public TensorHandle tensorHandle() {
+    public TensorHandle handle() {
       return tensorHandle;
     }
 
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/types/TUint8Factory.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/types/TUint8Factory.java
index e9a95357d29..13d3e5bc663 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/types/TUint8Factory.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/internal/types/TUint8Factory.java
@@ -1,22 +1,21 @@
 package org.tensorflow.internal.types;
 
+import org.tensorflow.AbstractTypeFactory;
 import org.tensorflow.TensorHandle;
-import org.tensorflow.internal.c_api.TF_Tensor;
 import org.tensorflow.internal.buffer.TensorBuffers;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.ndarray.buffer.ByteDataBuffer;
 import org.tensorflow.ndarray.impl.dense.ByteDenseNdArray;
 import org.tensorflow.types.TUint8;
-import org.tensorflow.types.TypeFactory;
 
 /**
  * Factory of {@link TUint8} tensor instances
  */
-public class TUint8Factory implements TypeFactory<TUint8> {
+public class TUint8Factory extends AbstractTypeFactory<TUint8> {
 
   @Override
   public TUint8 createDense(TensorHandle tensorHandle, Shape shape) {
-    ByteDataBuffer buffer = TensorBuffers.toBytes(tensorHandle.nativeHandle());
+    ByteDataBuffer buffer = TensorBuffers.toBytes(getNative(tensorHandle));
     return new TUint8Impl(tensorHandle, buffer, shape);
   }
 
@@ -28,7 +27,7 @@ public Class<TUint8> type() {
     }
 
     @Override
-    public TensorHandle tensorHandle() {
+    public TensorHandle handle() {
       return tensorHandle;
     }
 
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TBfloat16.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TBfloat16.java
index e9570a780c4..6b56c310e47 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TBfloat16.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TBfloat16.java
@@ -47,7 +47,7 @@
  * computation compared to {@link TFloat16} when GPUs are not used.
  */
 @TensorType(dataType = DataType.DT_BFLOAT16, byteSize = 2, factory = TBfloat16Factory.class)
-public interface TBfloat16 extends TFloating<Float>, FloatNdArray {
+public interface TBfloat16 extends TFloating, FloatNdArray {
 
   /**
    * Allocates a new tensor for storing a single float value.
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TBool.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TBool.java
index d038c291766..f5e5976bc73 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TBool.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TBool.java
@@ -39,7 +39,7 @@
  * BOOL} layout, which may impact I/O performances.
  */
 @TensorType(dataType = DataType.DT_BOOL, byteSize = 1, factory = TBoolFactory.class)
-public interface TBool extends TType<Boolean>, BooleanNdArray {
+public interface TBool extends TType, BooleanNdArray {
 
   /**
    * Allocates a new tensor for storing a single boolean value.
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TFloat16.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TFloat16.java
index 482704e6823..7add4b25346 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TFloat16.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TFloat16.java
@@ -44,7 +44,7 @@
  * TBfloat16} tensor type might be a better option.
  */
 @TensorType(dataType = DataType.DT_HALF, byteSize = 2, factory = TFloat16Factory.class)
-public interface TFloat16 extends TFloating<Float>, FloatNdArray {
+public interface TFloat16 extends TFloating, FloatNdArray {
 
   /**
    * Allocates a new tensor for storing a single float value.
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TFloat32.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TFloat32.java
index 02a527fd364..32bc85ffa3e 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TFloat32.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TFloat32.java
@@ -34,7 +34,7 @@
  * IEEE-754 single-precision 32-bit float tensor type.
  */
 @TensorType(dataType = DataType.DT_FLOAT, byteSize = 4, factory = TFloat32Factory.class)
-public interface TFloat32 extends TFloating<Float>, FloatNdArray {
+public interface TFloat32 extends TFloating, FloatNdArray {
 
   /**
    * Allocates a new tensor for storing a single float value.
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TFloat64.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TFloat64.java
index 53b7a2732a9..cc5799bf27f 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TFloat64.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TFloat64.java
@@ -34,7 +34,7 @@
  * IEEE-754 double-precision 64-bit float tensor type.
  */
 @TensorType(dataType = DataType.DT_DOUBLE, byteSize = 8, factory = TFloat64Factory.class)
-public interface TFloat64 extends TFloating<Double>, DoubleNdArray {
+public interface TFloat64 extends TFloating, DoubleNdArray {
 
   /**
    * Allocates a new tensor for storing a single double value.
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TInt32.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TInt32.java
index 7a77e3f08c4..ff2f7277ab8 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TInt32.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TInt32.java
@@ -27,13 +27,14 @@
 import org.tensorflow.ndarray.buffer.IntDataBuffer;
 import org.tensorflow.proto.framework.DataType;
 import org.tensorflow.types.annotation.TensorType;
+import org.tensorflow.types.family.TInteger;
 import org.tensorflow.types.family.TNumber;
 
 /**
  * 32-bit signed integer tensor type.
  */
 @TensorType(dataType = DataType.DT_INT32, byteSize = 4, factory = TInt32Factory.class)
-public interface TInt32 extends TNumber<Integer>, IntNdArray {
+public interface TInt32 extends TInteger, IntNdArray {
 
   /**
    * Allocates a new tensor for storing a single int value.
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TInt64.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TInt64.java
index 2140c920195..cbefb248454 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TInt64.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TInt64.java
@@ -28,13 +28,14 @@
 import org.tensorflow.ndarray.buffer.LongDataBuffer;
 import org.tensorflow.proto.framework.DataType;
 import org.tensorflow.types.annotation.TensorType;
+import org.tensorflow.types.family.TInteger;
 import org.tensorflow.types.family.TNumber;
 
 /**
  * 64-bit signed integer tensor type.
  */
 @TensorType(dataType = DataType.DT_INT64, byteSize = 8, factory = TInt64Factory.class)
-public interface TInt64 extends TNumber<Long>, LongNdArray {
+public interface TInt64 extends TInteger, LongNdArray {
 
   /**
    * Allocates a new tensor for storing a single long value.
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TString.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TString.java
index 667ce36d526..8c29ffb5aa6 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TString.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TString.java
@@ -41,7 +41,7 @@
  * data in the tensor is initialized once and cannot be modified afterwards.
  */
 @TensorType(dataType = DataType.DT_STRING, byteSize = -1, factory = TStringFactory.class)
-public interface TString extends TType<String> {
+public interface TString extends TType, NdArray<String> {
 
   /**
    * Allocates a new tensor for storing a string scalar.
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TUint8.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TUint8.java
index d60a6436d51..c92a2db560e 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TUint8.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TUint8.java
@@ -28,13 +28,14 @@
 import org.tensorflow.ndarray.buffer.ByteDataBuffer;
 import org.tensorflow.proto.framework.DataType;
 import org.tensorflow.types.annotation.TensorType;
+import org.tensorflow.types.family.TInteger;
 import org.tensorflow.types.family.TNumber;
 
 /**
  * 8-bit unsigned integer tensor type.
  */
 @TensorType(dataType = DataType.DT_UINT8, byteSize = 1, factory = TUint8Factory.class)
-public interface TUint8 extends TNumber<Byte>, ByteNdArray {
+public interface TUint8 extends TInteger, ByteNdArray {
 
   /**
    * Allocates a new tensor for storing a single byte value.
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/family/TFloating.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/family/TFloating.java
index 27e056ef643..92deaffdc68 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/family/TFloating.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/family/TFloating.java
@@ -16,4 +16,4 @@
  * exp.call(tf.constant(tensor2));  // Compilation failure
  * }</pre>
  */
-public interface TFloating<T> extends TNumber<T> {}
+public interface TFloating extends TNumber {}
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/family/TInteger.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/family/TInteger.java
new file mode 100644
index 00000000000..0112d278e49
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/family/TInteger.java
@@ -0,0 +1,5 @@
+package org.tensorflow.types.family;
+
+public interface TInteger extends TNumber {
+
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/family/TNumber.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/family/TNumber.java
index cbc9333dc72..97ee59af095 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/family/TNumber.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/family/TNumber.java
@@ -32,4 +32,4 @@
  * tf.nn.softmax(tf.constant(tensor2));  // Compilation failure
  * }</pre>
  */
-public interface TNumber<T> extends TType<T> {}
+public interface TNumber extends TType {}
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/family/TType.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/family/TType.java
index 89629974502..8202d22f570 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/family/TType.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/family/TType.java
@@ -2,33 +2,30 @@
 
 import org.tensorflow.Tensor;
 import org.tensorflow.TensorHandle;
-import org.tensorflow.ndarray.NdArray;
 import org.tensorflow.ndarray.buffer.ByteDataBuffer;
 import org.tensorflow.proto.framework.DataType;
 
-public interface TType<T> extends NdArray<T>, Tensor {
+public interface TType extends Tensor {
 
   Class<? extends TType> type();
 
-  TensorHandle tensorHandle();
-
   @Override
   default void close() {
-    tensorHandle().close();
+    handle().close();
   }
 
   @Override
   default ByteDataBuffer rawData() {
-    return tensorHandle().rawData();
+    return handle().rawData();
   }
 
   @Override
   default DataType dataType() {
-    return tensorHandle().dataType();
+    return handle().dataType();
   }
 
   @Override
   default long numBytes() {
-    return tensorHandle().numBytes();
+    return handle().numBytes();
   }
 }
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/util/TensorList.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/util/TensorList.java
index a1d6cd89feb..87a4fb378c9 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/util/TensorList.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/util/TensorList.java
@@ -34,14 +34,14 @@ public <T extends TType> boolean add(T tensor) {
     return tensors.add(tensor);
   }
 
-  public <T extends TType<?>> T single() {
+  public <T extends TType> T single() {
     if (tensors.size() != 1) {
       throw new IllegalStateException("List must contain a single tensor to use non-indexed getter");
     }
     return get(0);
   }
 
-  public <T extends TType<?>> T get(int index) {
+  public <T extends TType> T get(int index) {
     return (T)tensors.get(index);
   }
 
@@ -51,8 +51,8 @@ public TensorMap toMap(Collection<String> tensorNames) {
 
   @Override
   public void close() {
-    for (TType<?> t : tensors) {
-      t.tensorHandle().nativeHandle().releaseReference();
+    for (TType t : tensors) {
+      t.handle().release();
     }
   }
 
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/util/TensorMap.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/util/TensorMap.java
index f335bb99d32..7e821ccf2af 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/util/TensorMap.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/util/TensorMap.java
@@ -4,42 +4,41 @@
 import java.util.HashMap;
 import java.util.Iterator;
 import java.util.Map;
-import java.util.Set;
 import org.tensorflow.types.family.TType;
 
-public class TensorMap extends HashMap<String, TType<?>> implements AutoCloseable {
+public class TensorMap extends HashMap<String, TType> implements AutoCloseable {
 
   @Override
-  public TType<?> put(String key, TType<?> value) {
-    value.tensorHandle().nativeHandle().retainReference();
-    TType<?> t = super.put(key, value);
+  public TType put(String key, TType value) {
+    value.handle().retain();
+    TType t = super.put(key, value);
     if (t != null) {
-      t.tensorHandle().nativeHandle().releaseReference();
+      t.handle().release();
     }
     return t;
   }
 
   @Override
-  public TType<?> remove(Object key) {
-    TType<?> t = super.remove(key);
+  public TType remove(Object key) {
+    TType t = super.remove(key);
     if (t != null) {
-      t.tensorHandle().nativeHandle().releaseReference();
+      t.handle().release();
     }
     return t;
   }
 
   @Override
-  public void putAll(Map<? extends String, ? extends TType<?>> m) {
+  public void putAll(Map<? extends String, ? extends TType> m) {
     super.putAll(m);
-    for (TType<?> t : m.values()) {
-      t.tensorHandle().nativeHandle().retainReference();
+    for (TType t : m.values()) {
+      t.handle().retain();
     }
   }
 
   @Override
   public void clear() {
-    for (TType<?> t : values()) {
-      t.tensorHandle().nativeHandle().releaseReference();
+    for (TType t : values()) {
+      t.handle().release();
     }
     super.clear();
   }
@@ -53,7 +52,7 @@ public <T extends TType> T get(String name) {
     return (T) super.get(name);
   }
 
-  public <T extends TType<?>> T single() {
+  public <T extends TType> T single() {
     if (size() != 1) {
       throw new IllegalStateException("List must contain a single tensor to use non-indexed getter");
     }
diff --git a/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/TensorTest.java b/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/TensorTest.java
index 11c0cdedde0..601cf2e6de5 100644
--- a/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/TensorTest.java
+++ b/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/TensorTest.java
@@ -468,21 +468,15 @@ public void eagerTensorIsReleasedAfterSessionIsClosed() {
     try (EagerSession session = EagerSession.create()) {
       Ops tf = Ops.create(session);
       sum = tf.math.add(tf.constant(10), tf.constant(20)).asTensor();
-      sum.tensorHandle().nativeHandle(); // does not throw
+      sum.handle().get(); // does not throw
       assertEquals(30, sum.getInt());
     }
     try {
-      sum.tensorHandle().nativeHandle();
+      sum.handle().get();
       fail("Tensor native handle should have been closed by ending eager session");
     } catch (IllegalStateException e) {
       // as expected
     }
-    try {
-      sum.getInt();
-      fail("Tensor data should not be accessible after tensor is closed");
-    } catch (IllegalStateException e) {
-      // as expected
-    }
   }
 
   @Test
@@ -495,7 +489,7 @@ public void fromHandle() {
     // close() on both Tensors.
     final FloatNdArray matrix = StdArrays.ndCopyOf(new float[][]{{1, 2, 3}, {4, 5, 6}});
     try (TFloat32 src = TFloat32.tensorOf(matrix)) {
-      TFloat32 cpy = Tensors.fromHandle(src.tensorHandle());
+      TFloat32 cpy = Tensors.fromHandle(src.handle());
       assertEquals(src.type(), cpy.type());
       assertEquals(src.shape().numDimensions(), cpy.shape().numDimensions());
       assertEquals(src.shape(), cpy.shape());
diff --git a/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/types/NumericTypesTestBase.java b/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/types/NumericTypesTestBase.java
index cff8207a706..24390ae0a91 100644
--- a/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/types/NumericTypesTestBase.java
+++ b/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/types/NumericTypesTestBase.java
@@ -39,22 +39,23 @@ public void initializeTensorsWithZeros() {
 
     assertEquals(3, tensor.rank());
     assertEquals(12, tensor.size());
+    NdArray<U> data = (NdArray<U>)tensor;
 
     try (EagerSession session = EagerSession.create()) {
       Ops tf = Ops.create(session);
 
       // Initialize tensor memory with zeros and take a snapshot
-      tensor.scalars().forEach(scalar -> ((NdArray<U>)scalar).setObject(valueOf(0)));
+      data.scalars().forEach(scalar -> ((NdArray<U>)scalar).setObject(valueOf(0)));
       Constant<T> x = tf.capture(tensor);
 
       // Initialize the same tensor memory with ones and take a snapshot
-      tensor.scalars().forEach(scalar -> ((NdArray<U>)scalar).setObject(valueOf(1)));
+      data.scalars().forEach(scalar -> ((NdArray<U>)scalar).setObject(valueOf(1)));
       Constant<T> y = tf.capture(tensor);
 
       // Subtract y from x and validate the result
       Sub<T> sub = tf.math.sub(x, y);
-      sub.asTensor().scalars().forEach(scalar ->
-          assertEquals(valueOf(-1), ((NdArray<U>)scalar).getObject())
+      ((NdArray<U>)sub.asTensor()).scalars().forEach(scalar ->
+          assertEquals(valueOf(-1), scalar.getObject())
       );
     }
   }
@@ -69,20 +70,21 @@ public void setAndCompute() {
 
     // Creates a 2x2 matrix
     try (T tensor = allocateTensor(Shape.of(2, 2))) {
+      NdArray<U> data = (NdArray<U>)tensor;
 
       // Copy first 2 values of the vector to the first row of the matrix
-      tensor.set(heapData.slice(Indices.range(0, 2)), 0);
+      data.set(heapData.slice(Indices.range(0, 2)), 0);
 
       // Copy values at an odd position in the vector as the second row of the matrix
-      tensor.set(heapData.slice(Indices.odd()), 1);
+      data.set(heapData.slice(Indices.odd()), 1);
 
-      assertEquals(valueOf(0), tensor.getObject(0, 0));
-      assertEquals(valueOf(1), tensor.getObject(0, 1));
-      assertEquals(valueOf(1), tensor.getObject(1, 0));
-      assertEquals(valueOf(3), tensor.getObject(1, 1));
+      assertEquals(valueOf(0), data.getObject(0, 0));
+      assertEquals(valueOf(1), data.getObject(0, 1));
+      assertEquals(valueOf(1), data.getObject(1, 0));
+      assertEquals(valueOf(3), data.getObject(1, 1));
 
       // Read rows of the tensor in reverse order
-      NdArray<U> flippedData = tensor.slice(Indices.flip(), Indices.flip());
+      NdArray<U> flippedData = data.slice(Indices.flip(), Indices.flip());
 
       assertEquals(valueOf(3), flippedData.getObject(0, 0));
       assertEquals(valueOf(1), flippedData.getObject(0, 1));
@@ -93,7 +95,7 @@ public void setAndCompute() {
         Ops tf = Ops.create(session);
 
         Add<T> add = tf.math.add(tf.capture(tensor), tf.capture(tensor));
-        T result = add.asTensor();
+        NdArray<U> result = (NdArray<U>)add.asTensor();
 
         assertEquals(valueOf(0), result.getObject(0, 0));
         assertEquals(valueOf(2), result.getObject(0, 1));
diff --git a/tensorflow-framework/src/main/java/org/tensorflow/framework/data/Dataset.java b/tensorflow-framework/src/main/java/org/tensorflow/framework/data/Dataset.java
index 007bcb01a40..c1eca6aeb29 100644
--- a/tensorflow-framework/src/main/java/org/tensorflow/framework/data/Dataset.java
+++ b/tensorflow-framework/src/main/java/org/tensorflow/framework/data/Dataset.java
@@ -15,7 +15,6 @@
  */
 package org.tensorflow.framework.data;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.framework.data.impl.BatchDataset;
 import org.tensorflow.framework.data.impl.MapDataset;
@@ -33,6 +32,7 @@
 import java.util.Iterator;
 import java.util.List;
 import java.util.function.Function;
+import org.tensorflow.types.family.TType;
 
 /**
  * Represents a potentially large list of independent elements (samples), and allows iteration and
@@ -41,11 +41,11 @@
 public abstract class Dataset implements Iterable<List<Operand<?>>> {
   protected Ops tf;
   private Operand<?> variant;
-  private List<DataType<?>> outputTypes;
+  private List<Class<? extends TType>> outputTypes;
   private List<Shape> outputShapes;
 
   public Dataset(
-      Ops tf, Operand<?> variant, List<DataType<?>> outputTypes, List<Shape> outputShapes) {
+      Ops tf, Operand<?> variant, List<Class<? extends TType>> outputTypes, List<Shape> outputShapes) {
     if (tf == null) {
       throw new IllegalArgumentException("Ops accessor cannot be null.");
     }
@@ -266,7 +266,7 @@ public DatasetIterator makeOneShotIterator() {
    * @return A new `Dataset`
    */
   public static Dataset fromTensorSlices(
-      Ops tf, List<Operand<?>> tensors, List<DataType<?>> outputTypes) {
+      Ops tf, List<Operand<?>> tensors, List<Class<? extends TType>> outputTypes) {
     return new TensorSliceDataset(tf, tensors, outputTypes);
   }
 
@@ -288,7 +288,7 @@ public Operand<?> getVariant() {
   }
 
   /** Get a list of output types for each component of this dataset. */
-  public List<DataType<?>> getOutputTypes() {
+  public List<Class<? extends TType>> getOutputTypes() {
     return this.outputTypes;
   }
 
@@ -305,7 +305,7 @@ public Ops getOpsInstance() {
   public String toString() {
     return "Dataset{"
         + "outputTypes="
-        + Arrays.toString(getOutputTypes().stream().map(DataType::name).toArray())
+        + Arrays.toString(getOutputTypes().stream().map(Class::getSimpleName).toArray())
         + ", outputShapes="
         + Arrays.toString(getOutputShapes().stream().map(Shape::toString).toArray())
         + "}";
diff --git a/tensorflow-framework/src/main/java/org/tensorflow/framework/data/DatasetIterator.java b/tensorflow-framework/src/main/java/org/tensorflow/framework/data/DatasetIterator.java
index 0bad6a41214..5dbb2bfacff 100644
--- a/tensorflow-framework/src/main/java/org/tensorflow/framework/data/DatasetIterator.java
+++ b/tensorflow-framework/src/main/java/org/tensorflow/framework/data/DatasetIterator.java
@@ -15,7 +15,6 @@
  */
 package org.tensorflow.framework.data;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Graph;
 import org.tensorflow.Operand;
 import org.tensorflow.op.Op;
@@ -25,6 +24,7 @@
 import java.util.ArrayList;
 import java.util.Iterator;
 import java.util.List;
+import org.tensorflow.types.family.TType;
 
 /**
  * Represents the state of an iteration through a tf.data Datset. DatasetIterator is not a
@@ -106,7 +106,7 @@ public class DatasetIterator implements Iterable<List<Operand<?>>> {
   private Operand<?> iteratorResource;
   private Op initializer;
 
-  protected List<DataType<?>> outputTypes;
+  protected List<Class<? extends TType>> outputTypes;
   protected List<Shape> outputShapes;
 
   /**
@@ -124,7 +124,7 @@ public DatasetIterator(
       Ops tf,
       Operand<?> iteratorResource,
       Op initializer,
-      List<DataType<?>> outputTypes,
+      List<Class<? extends TType>> outputTypes,
       List<Shape> outputShapes) {
 
     this.tf = tf;
@@ -137,7 +137,7 @@ public DatasetIterator(
   public DatasetIterator(
       Ops tf,
       Operand<?> iteratorResource,
-      List<DataType<?>> outputTypes,
+      List<Class<? extends TType>> outputTypes,
       List<Shape> outputShapes) {
     this.tf = tf;
     this.iteratorResource = iteratorResource;
@@ -236,7 +236,7 @@ public Op makeInitializer(Dataset dataset) {
    * @return A new DatasetIterator
    */
   public static DatasetIterator fromStructure(
-      Ops tf, List<DataType<?>> outputTypes, List<Shape> outputShapes) {
+      Ops tf, List<Class<? extends TType>> outputTypes, List<Shape> outputShapes) {
     Operand<?> iteratorResource =
         tf.scope().env() instanceof Graph
             ? tf.data.iterator(EMPTY_SHARED_NAME, "", outputTypes, outputShapes)
diff --git a/tensorflow-framework/src/main/java/org/tensorflow/framework/data/DatasetOptional.java b/tensorflow-framework/src/main/java/org/tensorflow/framework/data/DatasetOptional.java
index 925252c7298..6617c33eaf7 100644
--- a/tensorflow-framework/src/main/java/org/tensorflow/framework/data/DatasetOptional.java
+++ b/tensorflow-framework/src/main/java/org/tensorflow/framework/data/DatasetOptional.java
@@ -15,7 +15,6 @@
  */
 package org.tensorflow.framework.data;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.op.Ops;
 import org.tensorflow.ndarray.Shape;
@@ -23,6 +22,7 @@
 
 import java.util.ArrayList;
 import java.util.List;
+import org.tensorflow.types.family.TType;
 
 /**
  * An optional represents the result of a dataset getNext operation that may fail, when the end of
@@ -36,11 +36,11 @@ public Operand<?> getOptionalVariant() {
   }
 
   private Operand<?> optionalVariant;
-  private List<DataType<?>> outputTypes;
+  private List<Class<? extends TType>> outputTypes;
   private List<Shape> outputShapes;
 
   public DatasetOptional(
-      Ops tf, Operand<?> optionalVariant, List<DataType<?>> outputTypes, List<Shape> outputShapes) {
+      Ops tf, Operand<?> optionalVariant, List<Class<? extends TType>> outputTypes, List<Shape> outputShapes) {
     this.tf = tf;
     this.optionalVariant = optionalVariant;
     this.outputTypes = outputTypes;
@@ -75,7 +75,7 @@ public List<Operand<?>> getValue() {
   public static DatasetOptional fromComponents(
       Ops tf,
       List<Operand<?>> components,
-      List<DataType<?>> outputTypes,
+      List<Class<? extends TType>> outputTypes,
       List<Shape> outputShapes) {
     Operand<?> optionalVariant = tf.data.optionalFromValue(components);
     return new DatasetOptional(tf, optionalVariant, outputTypes, outputShapes);
diff --git a/tensorflow-framework/src/main/java/org/tensorflow/framework/data/impl/BatchDataset.java b/tensorflow-framework/src/main/java/org/tensorflow/framework/data/impl/BatchDataset.java
index 277b049cf6f..f0561b2e61e 100644
--- a/tensorflow-framework/src/main/java/org/tensorflow/framework/data/impl/BatchDataset.java
+++ b/tensorflow-framework/src/main/java/org/tensorflow/framework/data/impl/BatchDataset.java
@@ -15,7 +15,6 @@
  */
 package org.tensorflow.framework.data.impl;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.framework.data.Dataset;
 import org.tensorflow.op.Ops;
@@ -25,6 +24,7 @@
 import org.tensorflow.types.TInt64;
 
 import java.util.List;
+import org.tensorflow.types.family.TType;
 
 public class BatchDataset extends Dataset {
   public BatchDataset(
@@ -32,7 +32,7 @@ public BatchDataset(
       Operand<?> variant,
       Constant<TInt64> batchSize,
       Constant<TBool> dropRemainder,
-      List<DataType<?>> outputTypes,
+      List<Class<? extends TType>> outputTypes,
       List<Shape> outputShapes) {
     super(
         tf,
diff --git a/tensorflow-framework/src/main/java/org/tensorflow/framework/data/impl/SkipDataset.java b/tensorflow-framework/src/main/java/org/tensorflow/framework/data/impl/SkipDataset.java
index 6731bac60b3..63b4208480b 100644
--- a/tensorflow-framework/src/main/java/org/tensorflow/framework/data/impl/SkipDataset.java
+++ b/tensorflow-framework/src/main/java/org/tensorflow/framework/data/impl/SkipDataset.java
@@ -15,7 +15,6 @@
  */
 package org.tensorflow.framework.data.impl;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.framework.data.Dataset;
 import org.tensorflow.op.Ops;
@@ -24,6 +23,7 @@
 import org.tensorflow.types.TInt64;
 
 import java.util.List;
+import org.tensorflow.types.family.TType;
 
 public class SkipDataset extends Dataset {
 
@@ -31,7 +31,7 @@ public SkipDataset(
       Ops tf,
       Operand<?> variant,
       Constant<TInt64> count,
-      List<DataType<?>> outputTypes,
+      List<Class<? extends TType>> outputTypes,
       List<Shape> outputShapes) {
     super(
         tf,
diff --git a/tensorflow-framework/src/main/java/org/tensorflow/framework/data/impl/TFRecordDataset.java b/tensorflow-framework/src/main/java/org/tensorflow/framework/data/impl/TFRecordDataset.java
index ed721b13ebf..00297152e90 100644
--- a/tensorflow-framework/src/main/java/org/tensorflow/framework/data/impl/TFRecordDataset.java
+++ b/tensorflow-framework/src/main/java/org/tensorflow/framework/data/impl/TFRecordDataset.java
@@ -34,7 +34,7 @@ public TFRecordDataset(
     super(
         tf,
         tf.data.tfRecordDataset(filenames, compressionType, bufferSize),
-        Collections.singletonList(TString.DTYPE),
+        Collections.singletonList(TString.class),
         Collections.singletonList(Shape.scalar()));
   }
 }
diff --git a/tensorflow-framework/src/main/java/org/tensorflow/framework/data/impl/TakeDataset.java b/tensorflow-framework/src/main/java/org/tensorflow/framework/data/impl/TakeDataset.java
index 08c57d44a73..39ca9759e74 100644
--- a/tensorflow-framework/src/main/java/org/tensorflow/framework/data/impl/TakeDataset.java
+++ b/tensorflow-framework/src/main/java/org/tensorflow/framework/data/impl/TakeDataset.java
@@ -15,7 +15,6 @@
  */
 package org.tensorflow.framework.data.impl;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.framework.data.Dataset;
 import org.tensorflow.op.Ops;
@@ -24,6 +23,7 @@
 import org.tensorflow.types.TInt64;
 
 import java.util.List;
+import org.tensorflow.types.family.TType;
 
 public class TakeDataset extends Dataset {
 
@@ -31,7 +31,7 @@ public TakeDataset(
       Ops tf,
       Operand<?> variant,
       Constant<TInt64> count,
-      List<DataType<?>> outputTypes,
+      List<Class<? extends TType>> outputTypes,
       List<Shape> outputShapes) {
     super(
         tf,
diff --git a/tensorflow-framework/src/main/java/org/tensorflow/framework/data/impl/TensorSliceDataset.java b/tensorflow-framework/src/main/java/org/tensorflow/framework/data/impl/TensorSliceDataset.java
index 14405ebdaf5..eac0f1cc1ed 100644
--- a/tensorflow-framework/src/main/java/org/tensorflow/framework/data/impl/TensorSliceDataset.java
+++ b/tensorflow-framework/src/main/java/org/tensorflow/framework/data/impl/TensorSliceDataset.java
@@ -15,7 +15,6 @@
  */
 package org.tensorflow.framework.data.impl;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.framework.data.Dataset;
 import org.tensorflow.op.Ops;
@@ -23,10 +22,11 @@
 
 import java.util.List;
 import java.util.stream.Collectors;
+import org.tensorflow.types.family.TType;
 
 public class TensorSliceDataset extends Dataset {
 
-  public TensorSliceDataset(Ops tf, List<Operand<?>> components, List<DataType<?>> outputTypes) {
+  public TensorSliceDataset(Ops tf, List<Operand<?>> components, List<Class<? extends TType>> outputTypes) {
     super(tf, makeVariant(tf, components, outputTypes), outputTypes, outputShapes(components));
   }
 
@@ -35,7 +35,7 @@ private static List<Shape> outputShapes(List<Operand<?>> components) {
   }
 
   private static Operand<?> makeVariant(
-      Ops tf, List<Operand<?>> components, List<DataType<?>> outputTypes) {
+      Ops tf, List<Operand<?>> components, List<Class<? extends TType>> outputTypes) {
     if (!(components.size() == outputTypes.size())) {
       throw new IllegalArgumentException(
           "Lists `tensors` and `dtypes` must have the same number of elements.");
diff --git a/tensorflow-framework/src/main/java/org/tensorflow/framework/data/impl/TextLineDataset.java b/tensorflow-framework/src/main/java/org/tensorflow/framework/data/impl/TextLineDataset.java
index 4ef47825211..c9a26304778 100644
--- a/tensorflow-framework/src/main/java/org/tensorflow/framework/data/impl/TextLineDataset.java
+++ b/tensorflow-framework/src/main/java/org/tensorflow/framework/data/impl/TextLineDataset.java
@@ -34,7 +34,7 @@ public TextLineDataset(
     super(
         tf,
         tf.data.textLineDataset(filenames, compressionType, bufferSize),
-        Collections.singletonList(TString.DTYPE),
+        Collections.singletonList(TString.class),
         Collections.singletonList(Shape.scalar()));
   }
 }
diff --git a/tensorflow-framework/src/main/java/org/tensorflow/framework/initializers/Constant.java b/tensorflow-framework/src/main/java/org/tensorflow/framework/initializers/Constant.java
index b4544de9bd0..16aed3db6ed 100644
--- a/tensorflow-framework/src/main/java/org/tensorflow/framework/initializers/Constant.java
+++ b/tensorflow-framework/src/main/java/org/tensorflow/framework/initializers/Constant.java
@@ -14,10 +14,11 @@
 =======================================================================*/
 package org.tensorflow.framework.initializers;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.op.Ops;
+import org.tensorflow.types.TBool;
 import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TNumber;
 import org.tensorflow.types.family.TType;
 
 /**
@@ -85,17 +86,17 @@ public Constant(Ops tf, boolean value) {
 
   /** {@inheritDoc} */
   @Override
-  public Operand<T> call(Operand<TInt64> dims, DataType<T> dtype) {
-    if (!(dtype.isNumeric() || dtype.isBoolean())) {
-      throw new IllegalArgumentException("DataType must be numeric or boolean: " + dtype.name());
+  public Operand<T> call(Operand<TInt64> dims, Class<T> type) {
+    if (!TNumber.class.isAssignableFrom(type) && type != TBool.class) {
+      throw new IllegalArgumentException("DataType must be numeric or boolean: " + type.getSimpleName());
     }
     switch (valueType) {
       case LONG:
-        return tf.fill(dims, tf.dtypes.cast(tf.constant(longValue), dtype));
+        return tf.fill(dims, tf.dtypes.cast(tf.constant(longValue), type));
       case DOUBLE:
-        return tf.fill(dims, tf.dtypes.cast(tf.constant(doubleValue), dtype));
+        return tf.fill(dims, tf.dtypes.cast(tf.constant(doubleValue), type));
       default:
-        return tf.fill(dims, tf.dtypes.cast(tf.constant(booleanValue), dtype));
+        return tf.fill(dims, tf.dtypes.cast(tf.constant(booleanValue), type));
     }
   }
 
diff --git a/tensorflow-framework/src/main/java/org/tensorflow/framework/initializers/Glorot.java b/tensorflow-framework/src/main/java/org/tensorflow/framework/initializers/Glorot.java
index 3d5d37b91d3..fd6b8542003 100644
--- a/tensorflow-framework/src/main/java/org/tensorflow/framework/initializers/Glorot.java
+++ b/tensorflow-framework/src/main/java/org/tensorflow/framework/initializers/Glorot.java
@@ -16,6 +16,7 @@
 package org.tensorflow.framework.initializers;
 
 import org.tensorflow.op.Ops;
+import org.tensorflow.types.family.TFloating;
 import org.tensorflow.types.family.TNumber;
 import org.tensorflow.types.family.TType;
 
@@ -70,7 +71,7 @@
  * @see VarianceScaling.Distribution
  * @see <a href="http://proceedings.mlr.press/v9/glorot10a.html">Glorot et al., 2010</a>
  */
-public class Glorot<T extends TType, U extends TNumber> extends VarianceScaling<T, U> {
+public class Glorot<T extends TFloating> extends VarianceScaling<T> {
 
   public static final double SCALE = 1.0;
 
diff --git a/tensorflow-framework/src/main/java/org/tensorflow/framework/initializers/He.java b/tensorflow-framework/src/main/java/org/tensorflow/framework/initializers/He.java
index ce99da80bf7..5c0f3bbc8cb 100644
--- a/tensorflow-framework/src/main/java/org/tensorflow/framework/initializers/He.java
+++ b/tensorflow-framework/src/main/java/org/tensorflow/framework/initializers/He.java
@@ -15,6 +15,7 @@
 package org.tensorflow.framework.initializers;
 
 import org.tensorflow.op.Ops;
+import org.tensorflow.types.family.TFloating;
 import org.tensorflow.types.family.TNumber;
 import org.tensorflow.types.family.TType;
 
@@ -66,7 +67,7 @@
  *     href="https://www.cv-foundation.org/openaccess/content_iccv_2015/html/He_Delving_Deep_into_ICCV_2015_paper.html">He
  *     et al., 2015</a>
  */
-public class He<T extends TType, U extends TNumber> extends VarianceScaling<T, U> {
+public class He<T extends TFloating> extends VarianceScaling<T> {
 
   public static final double SCALE = 2.0;
 
diff --git a/tensorflow-framework/src/main/java/org/tensorflow/framework/initializers/Identity.java b/tensorflow-framework/src/main/java/org/tensorflow/framework/initializers/Identity.java
index 34e6cd790f4..9f462dcfc5c 100644
--- a/tensorflow-framework/src/main/java/org/tensorflow/framework/initializers/Identity.java
+++ b/tensorflow-framework/src/main/java/org/tensorflow/framework/initializers/Identity.java
@@ -14,12 +14,12 @@
 =======================================================================*/
 package org.tensorflow.framework.initializers;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.framework.utils.ShapeUtils;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.Ops;
 import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TFloating;
 import org.tensorflow.types.family.TType;
 
 /**
@@ -38,7 +38,7 @@
  *
  * @param <T> The TType for the call operation
  */
-public class Identity<T extends TType> extends BaseInitializer<T> {
+public class Identity<T extends TFloating> extends BaseInitializer<T> {
   public static final double GAIN_DEFAULT = 1.0;
 
   private final double gain;
@@ -66,10 +66,7 @@ public Identity(Ops tf, double gain) {
 
   /** {@inheritDoc} */
   @Override
-  public Operand<T> call(Operand<TInt64> dims, DataType<T> dtype) {
-    if (!dtype.isFloating()) {
-      throw new IllegalArgumentException("DataType must be a float type: " + dtype.name());
-    }
+  public Operand<T> call(Operand<TInt64> dims, Class<T> type) {
     Shape shape = ShapeUtils.toShape(tf.scope(), dims);
     if (shape.numDimensions() != 2) {
       throw new IllegalArgumentException("2D matrix required, got " + shape.numDimensions());
@@ -79,9 +76,9 @@ public Operand<T> call(Operand<TInt64> dims, DataType<T> dtype) {
     Shape diagShape = Shape.of(diagSize);
 
     Operand<T> op;
-    Operand<T> zero = tf.dtypes.cast(tf.constant(0), dtype);
+    Operand<T> zero = tf.dtypes.cast(tf.constant(0), type);
     Operand<T> diagOnes =
-        tf.fill(tf.constant(diagShape.asArray()), tf.dtypes.cast(tf.constant(1.0), dtype));
+        tf.fill(tf.constant(diagShape.asArray()), tf.dtypes.cast(tf.constant(1.0), type));
     if (isSquare) {
       op =
           tf.linalg.matrixDiag(
@@ -91,10 +88,10 @@ public Operand<T> call(Operand<TInt64> dims, DataType<T> dtype) {
               tf.constant((int) shape.size(1)),
               zero);
     } else {
-      Operand<T> zeroMatrix = tf.zeros(dims, dtype);
+      Operand<T> zeroMatrix = tf.zeros(dims, type);
       op = tf.linalg.matrixSetDiag(zeroMatrix, diagOnes, tf.constant(0));
     }
 
-    return tf.math.mul(op, tf.dtypes.cast(tf.constant(gain), dtype));
+    return tf.math.mul(op, tf.dtypes.cast(tf.constant(gain), type));
   }
 }
diff --git a/tensorflow-framework/src/main/java/org/tensorflow/framework/initializers/Initializer.java b/tensorflow-framework/src/main/java/org/tensorflow/framework/initializers/Initializer.java
index 59dce1fc02e..4beb218783b 100644
--- a/tensorflow-framework/src/main/java/org/tensorflow/framework/initializers/Initializer.java
+++ b/tensorflow-framework/src/main/java/org/tensorflow/framework/initializers/Initializer.java
@@ -14,7 +14,6 @@
 =======================================================================*/
 package org.tensorflow.framework.initializers;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.types.TInt64;
 import org.tensorflow.types.family.TType;
@@ -30,8 +29,8 @@ public interface Initializer<T extends TType> {
    * Generates the operation used to perform the initialization.
    *
    * @param dims the shape dimensions
-   * @param dtype the data type
+   * @param type the type of tensor
    * @return An operand for the initialization.
    */
-  Operand<T> call(Operand<TInt64> dims, DataType<T> dtype);
+  Operand<T> call(Operand<TInt64> dims, Class<T> type);
 }
diff --git a/tensorflow-framework/src/main/java/org/tensorflow/framework/initializers/LeCun.java b/tensorflow-framework/src/main/java/org/tensorflow/framework/initializers/LeCun.java
index e2268412fc3..0f29f21284a 100644
--- a/tensorflow-framework/src/main/java/org/tensorflow/framework/initializers/LeCun.java
+++ b/tensorflow-framework/src/main/java/org/tensorflow/framework/initializers/LeCun.java
@@ -15,6 +15,7 @@
 package org.tensorflow.framework.initializers;
 
 import org.tensorflow.op.Ops;
+import org.tensorflow.types.family.TFloating;
 import org.tensorflow.types.family.TNumber;
 import org.tensorflow.types.family.TType;
 
@@ -76,7 +77,7 @@
  *     al., 1998</a>
  * @see VarianceScaling.Distribution
  */
-public class LeCun<T extends TType, U extends TNumber> extends VarianceScaling<T, U> {
+public class LeCun<T extends TFloating> extends VarianceScaling<T> {
 
   /**
    * Creates a LeCunNormal Initializer
diff --git a/tensorflow-framework/src/main/java/org/tensorflow/framework/initializers/Ones.java b/tensorflow-framework/src/main/java/org/tensorflow/framework/initializers/Ones.java
index b78f34e3d35..ef33d37bcb0 100644
--- a/tensorflow-framework/src/main/java/org/tensorflow/framework/initializers/Ones.java
+++ b/tensorflow-framework/src/main/java/org/tensorflow/framework/initializers/Ones.java
@@ -14,10 +14,11 @@
 =======================================================================*/
 package org.tensorflow.framework.initializers;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.op.Ops;
+import org.tensorflow.types.TBool;
 import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TNumber;
 import org.tensorflow.types.family.TType;
 
 /**
@@ -56,10 +57,10 @@ public Ones(Ops tf) {
 
   /** {@inheritDoc} */
   @Override
-  public Operand<T> call(Operand<TInt64> dims, DataType<T> dtype) {
-    if (!(dtype.isNumeric() || dtype.isBoolean())) {
-      throw new IllegalArgumentException("DataType must be numeric or boolean: " + dtype.name());
+  public Operand<T> call(Operand<TInt64> dims, Class<T> type) {
+    if (!TNumber.class.isAssignableFrom(type) && type != TBool.class) {
+      throw new IllegalArgumentException("DataType must be numeric or boolean: " + type.getSimpleName());
     }
-    return tf.fill(dims, tf.dtypes.cast(tf.constant(1.0), dtype));
+    return tf.fill(dims, tf.dtypes.cast(tf.constant(1.0), type));
   }
 }
diff --git a/tensorflow-framework/src/main/java/org/tensorflow/framework/initializers/Orthogonal.java b/tensorflow-framework/src/main/java/org/tensorflow/framework/initializers/Orthogonal.java
index 48e2c56d5be..1aada14d3c4 100644
--- a/tensorflow-framework/src/main/java/org/tensorflow/framework/initializers/Orthogonal.java
+++ b/tensorflow-framework/src/main/java/org/tensorflow/framework/initializers/Orthogonal.java
@@ -14,7 +14,6 @@
 =======================================================================*/
 package org.tensorflow.framework.initializers;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.Output;
 import org.tensorflow.framework.utils.ShapeUtils;
@@ -22,6 +21,7 @@
 import org.tensorflow.op.Ops;
 import org.tensorflow.op.linalg.Qr;
 import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TFloating;
 import org.tensorflow.types.family.TNumber;
 import org.tensorflow.types.family.TType;
 
@@ -50,7 +50,7 @@
  * @param <T> The TType for the call operation
  * @param <U> The TNumber for the call operation
  */
-public class Orthogonal<T extends TType, U extends TNumber> extends BaseInitializer<T> {
+public class Orthogonal<T extends TFloating> extends BaseInitializer<T> {
 
   public static final double GAIN_DEFAULT = 1.0;
 
@@ -84,10 +84,7 @@ public Orthogonal(Ops tf, double gain, long seed) {
 
   /** {@inheritDoc} */
   @Override
-  public Operand<T> call(Operand<TInt64> dims, DataType<T> dtype) {
-    if (!dtype.isFloating()) {
-      throw new IllegalArgumentException("Expected floating point type, got " + dtype.name());
-    }
+  public Operand<T> call(Operand<TInt64> dims, Class<T> type) {
     Shape dimsShape = ShapeUtils.toShape(tf.scope(), dims);
     if (dimsShape.numDimensions() < 2) {
       throw new IllegalArgumentException(
@@ -100,22 +97,17 @@ public Operand<T> call(Operand<TInt64> dims, DataType<T> dtype) {
     long numCols = dimsShape.size(i);
     Shape flatShape = Shape.of(Math.max(numRows, numCols), Math.min(numRows, numCols));
     long[] seeds = {seed, 0};
-    @SuppressWarnings("unchecked")
-    DataType<U> numdType = (DataType<U>) dtype;
-    @SuppressWarnings("unchecked")
     Operand<T> op =
-        (Operand<T>)
-            tf.random.statelessRandomNormal(tf.constant(flatShape), tf.constant(seeds), numdType);
-
+        tf.random.statelessRandomNormal(tf.constant(flatShape), tf.constant(seeds), type);
     Qr.Options qrOptions = Qr.fullMatrices(false);
     Qr<T> qrOp = tf.linalg.qr(op, qrOptions);
     Output<T> qo = qrOp.q();
     Output<T> ro = qrOp.r();
     Operand<T> diagOp =
-        tf.linalg.matrixDiagPart(ro, tf.constant(0), tf.dtypes.cast(tf.constant(0), dtype));
+        tf.linalg.matrixDiagPart(ro, tf.constant(0), tf.dtypes.cast(tf.constant(0), type));
     Operand<T> qop = tf.math.mul(qo, tf.math.sign(diagOp));
     if (numRows < numCols) qop = tf.linalg.transpose(qop, null);
 
-    return tf.math.mul(qop, tf.dtypes.cast(tf.constant(this.gain), dtype));
+    return tf.math.mul(qop, tf.dtypes.cast(tf.constant(this.gain), type));
   }
 }
diff --git a/tensorflow-framework/src/main/java/org/tensorflow/framework/initializers/RandomNormal.java b/tensorflow-framework/src/main/java/org/tensorflow/framework/initializers/RandomNormal.java
index f2d8a0d8e6e..5c2311d7439 100644
--- a/tensorflow-framework/src/main/java/org/tensorflow/framework/initializers/RandomNormal.java
+++ b/tensorflow-framework/src/main/java/org/tensorflow/framework/initializers/RandomNormal.java
@@ -14,7 +14,6 @@
 =======================================================================*/
 package org.tensorflow.framework.initializers;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.op.Ops;
 import org.tensorflow.types.TInt64;
@@ -37,7 +36,7 @@
  * @param <T> The TType for the call operation
  * @param <U> The TNumber for the call operation
  */
-public class RandomNormal<T extends TType, U extends TNumber> extends BaseInitializer<T> {
+public class RandomNormal<T extends TNumber> extends BaseInitializer<T> {
   public static final double MEAN_DEFAULT = 0.0;
   public static final double STDDEV_DEFAULT = 1.0;
 
@@ -87,16 +86,10 @@ public RandomNormal(Ops tf, double mean, double stddev, long seed) {
 
   /** {@inheritDoc} */
   @Override
-  public Operand<T> call(Operand<TInt64> dims, DataType<T> dtype) {
-    if (!dtype.isNumeric())
-      throw new IllegalArgumentException("The data type must be numeric. Found : " + dtype.name());
+  public Operand<T> call(Operand<TInt64> dims, Class<T> type) {
     long[] seeds = {seed, 0};
-    @SuppressWarnings("unchecked")
-    DataType<U> numdType = (DataType<U>) dtype;
-    @SuppressWarnings("unchecked")
-    Operand<T> distOp =
-        (Operand<T>) tf.random.statelessRandomNormal(dims, tf.constant(seeds), numdType);
-    Operand<T> op = tf.math.mul(distOp, tf.dtypes.cast(tf.constant(this.stddev), dtype));
-    return tf.math.add(op, tf.dtypes.cast(tf.constant(mean), dtype));
+    Operand<T> distOp = tf.random.statelessRandomNormal(dims, tf.constant(seeds), type);
+    Operand<T> op = tf.math.mul(distOp, tf.dtypes.cast(tf.constant(this.stddev), type));
+    return tf.math.add(op, tf.dtypes.cast(tf.constant(mean), type));
   }
 }
diff --git a/tensorflow-framework/src/main/java/org/tensorflow/framework/initializers/RandomUniform.java b/tensorflow-framework/src/main/java/org/tensorflow/framework/initializers/RandomUniform.java
index b665729675d..77c1927bdfa 100644
--- a/tensorflow-framework/src/main/java/org/tensorflow/framework/initializers/RandomUniform.java
+++ b/tensorflow-framework/src/main/java/org/tensorflow/framework/initializers/RandomUniform.java
@@ -14,11 +14,11 @@
 =======================================================================*/
 package org.tensorflow.framework.initializers;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.op.Ops;
 import org.tensorflow.op.random.RandomUniformInt;
 import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TInteger;
 import org.tensorflow.types.family.TNumber;
 import org.tensorflow.types.family.TType;
 
@@ -38,7 +38,7 @@
  * @param <T> The TType for the call operation
  * @param <U> The TNumber for the call operation
  */
-public class RandomUniform<T extends TType, U extends TNumber> extends BaseInitializer<T> {
+public class RandomUniform<T extends TNumber> extends BaseInitializer<T> {
 
   public static final double MINVAL_DEFAULT = -0.05;
   public static final double MAXVAL_DEFAULT = 0.05;
@@ -77,39 +77,28 @@ public RandomUniform(Ops tf, double minval, double maxval, long seed) {
 
   /** {@inheritDoc} */
   @Override
-  public Operand<T> call(Operand<TInt64> dims, DataType<T> dtype) {
-    if (!dtype.isNumeric())
-      throw new IllegalArgumentException("The data type must be numeric. Found : " + dtype.name());
-    @SuppressWarnings("unchecked")
-    DataType<U> numdType = (DataType<U>) dtype;
-    Operand<U> distOp;
-
-    if (dtype.isInteger()) {
+  public Operand<T> call(Operand<TInt64> dims, Class<T> type) {
+    Operand<T> distOp;
+    if (TInteger.class.isAssignableFrom(type)) {
       RandomUniformInt.Options options = RandomUniformInt.seed(this.seed);
       distOp =
           tf.random.randomUniformInt(
               dims,
-              tf.dtypes.cast(tf.constant(this.minval), numdType),
-              tf.dtypes.cast(tf.constant(this.maxval), numdType),
+              tf.dtypes.cast(tf.constant(this.minval), type),
+              tf.dtypes.cast(tf.constant(this.maxval), type),
               options);
-      @SuppressWarnings("unchecked")
-      Operand<T> distOpT = (Operand<T>) distOp;
-      return distOpT;
     } else {
       long[] seeds = {seed, 0};
-      distOp = tf.random.statelessRandomUniform(dims, tf.constant(seeds), numdType);
-      @SuppressWarnings("unchecked")
-      Operand<T> distOpT = (Operand<T>) distOp;
+      distOp = tf.random.statelessRandomUniform(dims, tf.constant(seeds), type);
       if (this.minval == 0) {
         if (this.maxval != 1.0) {
-          distOpT = tf.math.mul(distOpT, tf.dtypes.cast(tf.constant(this.maxval), dtype));
+          distOp = tf.math.mul(distOp, tf.dtypes.cast(tf.constant(this.maxval), type));
         }
       } else {
-        distOpT =
-            tf.math.mul(distOpT, tf.dtypes.cast(tf.constant(this.maxval - this.minval), dtype));
-        distOpT = tf.math.add(distOpT, tf.dtypes.cast(tf.constant(this.minval), dtype));
+        distOp = tf.math.mul(distOp, tf.dtypes.cast(tf.constant(this.maxval - this.minval), type));
+        distOp = tf.math.add(distOp, tf.dtypes.cast(tf.constant(this.minval), type));
       }
-      return distOpT;
     }
+    return distOp;
   }
 }
diff --git a/tensorflow-framework/src/main/java/org/tensorflow/framework/initializers/TruncatedNormal.java b/tensorflow-framework/src/main/java/org/tensorflow/framework/initializers/TruncatedNormal.java
index c71cf9a630e..c38c073f282 100644
--- a/tensorflow-framework/src/main/java/org/tensorflow/framework/initializers/TruncatedNormal.java
+++ b/tensorflow-framework/src/main/java/org/tensorflow/framework/initializers/TruncatedNormal.java
@@ -14,7 +14,6 @@
 =======================================================================*/
 package org.tensorflow.framework.initializers;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.op.Ops;
 import org.tensorflow.types.TInt64;
@@ -37,7 +36,7 @@
  * @param <T> The TType for the call operation
  * @param <U> The TNumber for the call operation
  */
-public class TruncatedNormal<T extends TType, U extends TNumber> extends BaseInitializer<T> {
+public class TruncatedNormal<T extends TNumber> extends BaseInitializer<T> {
 
   public static final double MEAN_DEFAULT = 0.0;
   public static final double STDDEV_DEFAULT = 0.05;
@@ -76,17 +75,11 @@ public TruncatedNormal(Ops tf, double mean, double stddev, long seed) {
 
   /** {@inheritDoc} */
   @Override
-  public Operand<T> call(Operand<TInt64> dims, DataType<T> dtype) {
-    if (!dtype.isNumeric())
-      throw new IllegalArgumentException("The data type must be numeric. Found : " + dtype.name());
+  public Operand<T> call(Operand<TInt64> dims, Class<T> type) {
     long[] seeds = {seed,0};
-    @SuppressWarnings("unchecked")
-    DataType<U> numdType = (DataType<U>) dtype;
-    Operand<U> distOp = tf.random.statelessTruncatedNormal(dims, tf.constant(seeds), numdType);
-    @SuppressWarnings("unchecked")
-    Operand<T> distOpT = (Operand<T>) distOp;
+    Operand<T> distOp = tf.random.statelessTruncatedNormal(dims, tf.constant(seeds), type);
     return tf.math.add(
-        tf.math.mul(distOpT, tf.dtypes.cast(tf.constant(stddev), dtype)),
-        tf.dtypes.cast(tf.constant(mean), dtype));
+        tf.math.mul(distOp, tf.dtypes.cast(tf.constant(stddev), type)),
+        tf.dtypes.cast(tf.constant(mean), type));
   }
 }
diff --git a/tensorflow-framework/src/main/java/org/tensorflow/framework/initializers/VarianceScaling.java b/tensorflow-framework/src/main/java/org/tensorflow/framework/initializers/VarianceScaling.java
index fd33adadd5c..fdf8920ce1b 100644
--- a/tensorflow-framework/src/main/java/org/tensorflow/framework/initializers/VarianceScaling.java
+++ b/tensorflow-framework/src/main/java/org/tensorflow/framework/initializers/VarianceScaling.java
@@ -14,12 +14,12 @@
 =======================================================================*/
 package org.tensorflow.framework.initializers;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.framework.utils.ShapeUtils;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.Ops;
 import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TFloating;
 import org.tensorflow.types.family.TNumber;
 import org.tensorflow.types.family.TType;
 
@@ -58,7 +58,7 @@
  * @see VarianceScaling.Mode
  * @see VarianceScaling.Distribution
  */
-public class VarianceScaling<T extends TType, U extends TNumber> extends BaseInitializer<T> {
+public class VarianceScaling<T extends TFloating> extends BaseInitializer<T> {
 
   public static final double SCALE_DEFAULT = 1.0;
   public static final Mode MODE_DEFAULT = Mode.FAN_IN;
@@ -102,10 +102,7 @@ public VarianceScaling(Ops tf, double scale, Mode mode, Distribution distributio
 
   /** {@inheritDoc} */
   @Override
-  public Operand<T> call(Operand<TInt64> dims, DataType<T> dtype) {
-    if (!dtype.isFloating()) {
-      throw new IllegalArgumentException("Expected floating point type, got " + dtype.name());
-    }
+  public Operand<T> call(Operand<TInt64> dims, Class<T> type) {
     Shape shape = ShapeUtils.toShape(this.tf.scope(), dims);
     double lscale = this.scale;
     double[] fans /* fanIn, fanOut */ = computeFans(shape);
@@ -120,32 +117,28 @@ public Operand<T> call(Operand<TInt64> dims, DataType<T> dtype) {
         lscale /= Math.max(1., (fans[0] + fans[1]) / 2.);
         break;
     }
-    Operand<U> distOp;
-    Operand<U> mulOp = null;
-    @SuppressWarnings("unchecked")
-    DataType<U> numdType = (DataType<U>) dtype;
+    Operand<T> distOp;
+    Operand<T> mulOp = null;
     double stddev;
     long[] seeds = {seed, 0};
     switch (distribution) {
       case TRUNCATED_NORMAL:
-        distOp = tf.random.statelessTruncatedNormal(dims, tf.constant(seeds), numdType);
+        distOp = tf.random.statelessTruncatedNormal(dims, tf.constant(seeds), type);
         stddev = Math.sqrt(lscale) / .87962566103423978;
-        mulOp = tf.math.mul(distOp, tf.dtypes.cast(tf.constant(stddev), numdType));
+        mulOp = tf.math.mul(distOp, tf.dtypes.cast(tf.constant(stddev), type));
         break;
       case NORMAL:
-        distOp = tf.random.statelessRandomNormal(dims, tf.constant(seeds), numdType);
+        distOp = tf.random.statelessRandomNormal(dims, tf.constant(seeds), type);
         stddev = Math.sqrt(lscale);
-        mulOp = tf.math.mul(distOp, tf.dtypes.cast(tf.constant(stddev), numdType));
+        mulOp = tf.math.mul(distOp, tf.dtypes.cast(tf.constant(stddev), type));
         break;
       case UNIFORM:
-        distOp = tf.random.statelessRandomUniform(dims, tf.constant(seeds), numdType);
+        distOp = tf.random.statelessRandomUniform(dims, tf.constant(seeds), type);
         stddev = Math.sqrt(3.0 * lscale);
-        mulOp = tf.math.mul(distOp, tf.dtypes.cast(tf.constant(stddev), numdType));
+        mulOp = tf.math.mul(distOp, tf.dtypes.cast(tf.constant(stddev), type));
         break;
     }
-
-    // Need to cast TNumber <U> to TType <T>
-    return (Operand<T>) mulOp;
+    return mulOp;
   }
 
   /**
diff --git a/tensorflow-framework/src/main/java/org/tensorflow/framework/initializers/Zeros.java b/tensorflow-framework/src/main/java/org/tensorflow/framework/initializers/Zeros.java
index 09dd512ffaa..1582750b37b 100644
--- a/tensorflow-framework/src/main/java/org/tensorflow/framework/initializers/Zeros.java
+++ b/tensorflow-framework/src/main/java/org/tensorflow/framework/initializers/Zeros.java
@@ -14,7 +14,6 @@
 =======================================================================*/
 package org.tensorflow.framework.initializers;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.op.Ops;
 import org.tensorflow.types.TInt64;
@@ -46,7 +45,7 @@ public Zeros(Ops tf) {
   }
 
   @Override
-  public Operand<T> call(Operand<TInt64> dims, DataType<T> dtype) {
+  public Operand<T> call(Operand<TInt64> dims, Class<T> dtype) {
     return tf.zeros(dims, dtype);
   }
 }
diff --git a/tensorflow-framework/src/main/java/org/tensorflow/framework/optimizers/AdaDelta.java b/tensorflow-framework/src/main/java/org/tensorflow/framework/optimizers/AdaDelta.java
index 0adf5f58910..822eb490f22 100644
--- a/tensorflow-framework/src/main/java/org/tensorflow/framework/optimizers/AdaDelta.java
+++ b/tensorflow-framework/src/main/java/org/tensorflow/framework/optimizers/AdaDelta.java
@@ -141,10 +141,10 @@ protected void createSlots(List<Output<? extends TType>> variables) {
    */
   private <T extends TType> void createAdaDeltaSlot(Output<T> v) {
     Operand<T> accumulatorInitializer =
-        tf.fill(tf.shape(v), tf.dtypes.cast(tf.constant(0.0f), v.dataType()));
+        tf.fill(tf.shape(v), tf.dtypes.cast(tf.constant(0.0f), v.type()));
     createSlot(v.asOutput(), ACCUMULATOR, accumulatorInitializer);
     Operand<T> updateInitializer =
-        tf.fill(tf.shape(v), tf.dtypes.cast(tf.constant(0.0f), v.dataType()));
+        tf.fill(tf.shape(v), tf.dtypes.cast(tf.constant(0.0f), v.type()));
     createSlot(v.asOutput(), ACCUMULATOR_UPDATE, updateInitializer);
   }
 
@@ -157,9 +157,9 @@ protected <T extends TType> Op applyDense(Output<T> gradient, Output<T> variable
         variable,
         accumSlot,
         accumUpdateSlot,
-        tf.dtypes.cast(tf.constant(learningRate), gradient.dataType()),
-        tf.dtypes.cast(tf.constant(rho), gradient.dataType()),
-        tf.dtypes.cast(tf.constant(epsilon), gradient.dataType()),
+        tf.dtypes.cast(tf.constant(learningRate), gradient.type()),
+        tf.dtypes.cast(tf.constant(rho), gradient.type()),
+        tf.dtypes.cast(tf.constant(epsilon), gradient.type()),
         gradient);
   }
 
diff --git a/tensorflow-framework/src/main/java/org/tensorflow/framework/optimizers/AdaGrad.java b/tensorflow-framework/src/main/java/org/tensorflow/framework/optimizers/AdaGrad.java
index 1a7f4675662..08f5f18a9cd 100644
--- a/tensorflow-framework/src/main/java/org/tensorflow/framework/optimizers/AdaGrad.java
+++ b/tensorflow-framework/src/main/java/org/tensorflow/framework/optimizers/AdaGrad.java
@@ -131,7 +131,7 @@ protected void createSlots(List<Output<? extends TType>> variables) {
    */
   private <T extends TType> void createAdaGradSlot(Output<T> v) {
     Operand<T> initializer =
-        tf.fill(tf.shape(v), tf.dtypes.cast(tf.constant(initialAccumulatorValue), v.dataType()));
+        tf.fill(tf.shape(v), tf.dtypes.cast(tf.constant(initialAccumulatorValue), v.type()));
     createSlot(v.asOutput(), ACCUMULATOR, initializer);
   }
 
@@ -140,7 +140,7 @@ private <T extends TType> void createAdaGradSlot(Output<T> v) {
   protected <T extends TType> Op applyDense(Output<T> gradient, Output<T> variable) {
     Variable<T> slot = getSlot(variable, ACCUMULATOR).get();
     return tf.train.applyAdagrad(
-        variable, slot, tf.dtypes.cast(tf.constant(learningRate), gradient.dataType()), gradient);
+        variable, slot, tf.dtypes.cast(tf.constant(learningRate), gradient.type()), gradient);
   }
 
   /** {@inheritDoc} */
diff --git a/tensorflow-framework/src/main/java/org/tensorflow/framework/optimizers/AdaGradDA.java b/tensorflow-framework/src/main/java/org/tensorflow/framework/optimizers/AdaGradDA.java
index f76217fda85..df624e41c4e 100644
--- a/tensorflow-framework/src/main/java/org/tensorflow/framework/optimizers/AdaGradDA.java
+++ b/tensorflow-framework/src/main/java/org/tensorflow/framework/optimizers/AdaGradDA.java
@@ -187,7 +187,7 @@ protected void createSlots(List<Output<? extends TType>> variables) {
     for (Output<? extends TType> v : variables) {
       createAdaGradDASlot(v);
     }
-    globalStep = tf.withName("adagrad-da-global-step").variable(Shape.scalar(), TInt64.DTYPE);
+    globalStep = tf.withName("adagrad-da-global-step").variable(Shape.scalar(), TInt64.class);
     Assign<TInt64> globalStepInitializer = tf.assign(globalStep, tf.constant(0L));
     graph.addInitializer(globalStepInitializer);
   }
@@ -199,10 +199,10 @@ protected void createSlots(List<Output<? extends TType>> variables) {
    * @param <T> the datatype of the variable.
    */
   private <T extends TType> void createAdaGradDASlot(Output<T> v) {
-    Operand<T> initializer = tf.fill(tf.shape(v), tf.dtypes.cast(tf.constant(0.0f), v.dataType()));
+    Operand<T> initializer = tf.fill(tf.shape(v), tf.dtypes.cast(tf.constant(0.0f), v.type()));
     createSlot(v.asOutput(), ACCUMULATOR, initializer);
     Operand<T> sqInitializer =
-        tf.fill(tf.shape(v), tf.dtypes.cast(tf.constant(initialAccumulatorValue), v.dataType()));
+        tf.fill(tf.shape(v), tf.dtypes.cast(tf.constant(initialAccumulatorValue), v.type()));
     createSlot(v.asOutput(), SQUARED_ACCUMULATOR, sqInitializer);
   }
 
@@ -216,9 +216,9 @@ protected <T extends TType> Op applyDense(Output<T> gradient, Output<T> variable
         gradSlot,
         gradSquaredSlot,
         gradient,
-        tf.dtypes.cast(tf.constant(learningRate), gradient.dataType()),
-        tf.dtypes.cast(tf.constant(l1Strength), gradient.dataType()),
-        tf.dtypes.cast(tf.constant(l2Strength), gradient.dataType()),
+        tf.dtypes.cast(tf.constant(learningRate), gradient.type()),
+        tf.dtypes.cast(tf.constant(l1Strength), gradient.type()),
+        tf.dtypes.cast(tf.constant(l2Strength), gradient.type()),
         globalStep);
   }
 
diff --git a/tensorflow-framework/src/main/java/org/tensorflow/framework/optimizers/Adam.java b/tensorflow-framework/src/main/java/org/tensorflow/framework/optimizers/Adam.java
index 8f620678781..72598d12543 100644
--- a/tensorflow-framework/src/main/java/org/tensorflow/framework/optimizers/Adam.java
+++ b/tensorflow-framework/src/main/java/org/tensorflow/framework/optimizers/Adam.java
@@ -189,10 +189,10 @@ protected void createSlots(List<Output<? extends TType>> variables) {
     for (Output<? extends TType> v : variables) {
       createAdamSlot(v.asOutput());
     }
-    betaOnePower = tf.withName("beta1_power").variable(Shape.scalar(), TFloat32.DTYPE);
+    betaOnePower = tf.withName("beta1_power").variable(Shape.scalar(), TFloat32.class);
     Assign<TFloat32> betaOnePowerInit = tf.assign(betaOnePower, tf.constant(betaOne));
     graph.addInitializer(betaOnePowerInit);
-    betaTwoPower = tf.withName("beta2_power").variable(Shape.scalar(), TFloat32.DTYPE);
+    betaTwoPower = tf.withName("beta2_power").variable(Shape.scalar(), TFloat32.class);
     Assign<TFloat32> betaTwoPowerInit = tf.assign(betaTwoPower, tf.constant(betaTwo));
     graph.addInitializer(betaTwoPowerInit);
   }
@@ -215,10 +215,10 @@ protected Optional<Op> prepare(String scopeName) {
    */
   private <T extends TType> void createAdamSlot(Output<T> v) {
     Operand<T> firstMomentInitializer =
-        tf.fill(tf.shape(v), tf.dtypes.cast(tf.constant(0.0f), v.dataType()));
+        tf.fill(tf.shape(v), tf.dtypes.cast(tf.constant(0.0f), v.type()));
     createSlot(v.asOutput(), FIRST_MOMENT, firstMomentInitializer);
     Operand<T> secondMomentInitializer =
-        tf.fill(tf.shape(v), tf.dtypes.cast(tf.constant(0.0f), v.dataType()));
+        tf.fill(tf.shape(v), tf.dtypes.cast(tf.constant(0.0f), v.type()));
     createSlot(v.asOutput(), SECOND_MOMENT, secondMomentInitializer);
   }
 
@@ -231,12 +231,12 @@ protected <T extends TType> Op applyDense(Output<T> gradient, Output<T> variable
         variable,
         firstMomentSlot,
         secondMomentSlot,
-        tf.dtypes.cast(betaOnePower, gradient.dataType()),
-        tf.dtypes.cast(betaTwoPower, gradient.dataType()),
-        tf.dtypes.cast(learningRateConst, gradient.dataType()),
-        tf.dtypes.cast(betaOneConst, gradient.dataType()),
-        tf.dtypes.cast(betaTwoConst, gradient.dataType()),
-        tf.dtypes.cast(epsilonConst, gradient.dataType()),
+        tf.dtypes.cast(betaOnePower, gradient.type()),
+        tf.dtypes.cast(betaTwoPower, gradient.type()),
+        tf.dtypes.cast(learningRateConst, gradient.type()),
+        tf.dtypes.cast(betaOneConst, gradient.type()),
+        tf.dtypes.cast(betaTwoConst, gradient.type()),
+        tf.dtypes.cast(epsilonConst, gradient.type()),
         gradient);
   }
 
diff --git a/tensorflow-framework/src/main/java/org/tensorflow/framework/optimizers/Adamax.java b/tensorflow-framework/src/main/java/org/tensorflow/framework/optimizers/Adamax.java
index 335d83cedfa..cd95bb3bd07 100644
--- a/tensorflow-framework/src/main/java/org/tensorflow/framework/optimizers/Adamax.java
+++ b/tensorflow-framework/src/main/java/org/tensorflow/framework/optimizers/Adamax.java
@@ -137,7 +137,7 @@ protected void createSlots(List<Output<? extends TType>> variables) {
     for (Output<? extends TType> v : variables) {
       createAdamaxSlot(v.asOutput());
     }
-    betaOnePower = tf.withName("beta1_power").variable(Shape.scalar(), TFloat32.DTYPE);
+    betaOnePower = tf.withName("beta1_power").variable(Shape.scalar(), TFloat32.class);
     Assign<TFloat32> betaOnePowerInit = tf.assign(betaOnePower, tf.constant(betaOne));
     ((Graph) tf.scope().env()).addInitializer(betaOnePowerInit);
   }
@@ -150,10 +150,10 @@ protected void createSlots(List<Output<? extends TType>> variables) {
    */
   private <T extends TType> void createAdamaxSlot(Output<T> v) {
     Operand<T> firstMomentInitializer =
-        tf.fill(tf.shape(v), tf.dtypes.cast(tf.constant(0.0f), v.dataType()));
+        tf.fill(tf.shape(v), tf.dtypes.cast(tf.constant(0.0f), v.type()));
     createSlot(v.asOutput(), FIRST_MOMENT, firstMomentInitializer);
     Operand<T> secondMomentInitializer =
-        tf.fill(tf.shape(v), tf.dtypes.cast(tf.constant(0.0f), v.dataType()));
+        tf.fill(tf.shape(v), tf.dtypes.cast(tf.constant(0.0f), v.type()));
     createSlot(v.asOutput(), SECOND_MOMENT, secondMomentInitializer);
   }
 
@@ -167,11 +167,11 @@ protected <T extends TType> Op applyDense(Output<T> gradient, Output<T> variable
         variable,
         firstMomentSlot,
         secondMomentSlot,
-        tf.dtypes.cast(betaOnePower, gradient.dataType()),
-        tf.dtypes.cast(learningRateConst, gradient.dataType()),
-        tf.dtypes.cast(betaOneConst, gradient.dataType()),
-        tf.dtypes.cast(betaTwoConst, gradient.dataType()),
-        tf.dtypes.cast(epsilonConst, gradient.dataType()),
+        tf.dtypes.cast(betaOnePower, gradient.type()),
+        tf.dtypes.cast(learningRateConst, gradient.type()),
+        tf.dtypes.cast(betaOneConst, gradient.type()),
+        tf.dtypes.cast(betaTwoConst, gradient.type()),
+        tf.dtypes.cast(epsilonConst, gradient.type()),
         gradient);
   }
 
diff --git a/tensorflow-framework/src/main/java/org/tensorflow/framework/optimizers/Ftrl.java b/tensorflow-framework/src/main/java/org/tensorflow/framework/optimizers/Ftrl.java
index 04c34a2535e..66314d2ffe0 100644
--- a/tensorflow-framework/src/main/java/org/tensorflow/framework/optimizers/Ftrl.java
+++ b/tensorflow-framework/src/main/java/org/tensorflow/framework/optimizers/Ftrl.java
@@ -230,10 +230,10 @@ protected void createSlots(List<Output<? extends TType>> variables) {
    */
   private <T extends TType> void createFtrlSlot(Output<T> v) {
     Operand<T> initializer =
-        tf.fill(tf.shape(v), tf.dtypes.cast(tf.constant(initialAccumulatorValue), v.dataType()));
+        tf.fill(tf.shape(v), tf.dtypes.cast(tf.constant(initialAccumulatorValue), v.type()));
     createSlot(v.asOutput(), ACCUMULATOR, initializer);
     Operand<T> linearInitializer =
-        tf.fill(tf.shape(v), tf.dtypes.cast(tf.constant(0.0f), v.dataType()));
+        tf.fill(tf.shape(v), tf.dtypes.cast(tf.constant(0.0f), v.type()));
     createSlot(v.asOutput(), LINEAR_ACCUMULATOR, linearInitializer);
   }
 
@@ -248,12 +248,12 @@ protected <T extends TType> Op applyDense(Output<T> gradient, Output<T> variable
         accumSlot, // accum
         linearSlot, // linear
         gradient, // gradient
-        tf.dtypes.cast(tf.constant(learningRate), gradient.dataType()), // lr
-        tf.dtypes.cast(tf.constant(l1RegularizationStrength), gradient.dataType()), // l1
-        tf.dtypes.cast(tf.constant(l2RegularizationStrength), gradient.dataType()), // l2
+        tf.dtypes.cast(tf.constant(learningRate), gradient.type()), // lr
+        tf.dtypes.cast(tf.constant(l1RegularizationStrength), gradient.type()), // l1
+        tf.dtypes.cast(tf.constant(l2RegularizationStrength), gradient.type()), // l2
         tf.dtypes.cast(
-            tf.constant(l2ShrinkageRegularizationStrength), gradient.dataType()), // l2Shrinkage
-        tf.dtypes.cast(tf.constant(learningRatePower), gradient.dataType()), // lrPower
+            tf.constant(l2ShrinkageRegularizationStrength), gradient.type()), // l2Shrinkage
+        tf.dtypes.cast(tf.constant(learningRatePower), gradient.type()), // lrPower
         options);
   }
 
diff --git a/tensorflow-framework/src/main/java/org/tensorflow/framework/optimizers/GradientDescent.java b/tensorflow-framework/src/main/java/org/tensorflow/framework/optimizers/GradientDescent.java
index e307855e636..a373b2e5b55 100644
--- a/tensorflow-framework/src/main/java/org/tensorflow/framework/optimizers/GradientDescent.java
+++ b/tensorflow-framework/src/main/java/org/tensorflow/framework/optimizers/GradientDescent.java
@@ -66,7 +66,7 @@ public GradientDescent(Graph graph, String name, float learningRate) {
   @Override
   protected <T extends TType> Op applyDense(Output<T> gradient, Output<T> variable) {
     return tf.train.applyGradientDescent(
-        variable, tf.dtypes.cast(tf.constant(learningRate), gradient.dataType()), gradient);
+        variable, tf.dtypes.cast(tf.constant(learningRate), gradient.type()), gradient);
   }
 
   /** {@inheritDoc} */
diff --git a/tensorflow-framework/src/main/java/org/tensorflow/framework/optimizers/Momentum.java b/tensorflow-framework/src/main/java/org/tensorflow/framework/optimizers/Momentum.java
index 111727d26fa..f6640409d60 100644
--- a/tensorflow-framework/src/main/java/org/tensorflow/framework/optimizers/Momentum.java
+++ b/tensorflow-framework/src/main/java/org/tensorflow/framework/optimizers/Momentum.java
@@ -125,7 +125,7 @@ protected void createSlots(List<Output<? extends TType>> variables) {
    * @param <T> the data type of the variable
    */
   private <T extends TType> void createMomentumSlot(Output<T> v) {
-    Operand<T> initializer = tf.fill(tf.shape(v), tf.dtypes.cast(tf.constant(0.0f), v.dataType()));
+    Operand<T> initializer = tf.fill(tf.shape(v), tf.dtypes.cast(tf.constant(0.0f), v.type()));
     createSlot(v.asOutput(), MOMENTUM, initializer);
   }
 
@@ -136,9 +136,9 @@ protected <T extends TType> Op applyDense(Output<T> gradient, Output<T> variable
     return tf.train.applyMomentum(
         variable,
         slot,
-        tf.dtypes.cast(tf.constant(learningRate), gradient.dataType()),
+        tf.dtypes.cast(tf.constant(learningRate), gradient.type()),
         gradient,
-        tf.dtypes.cast(tf.constant(momentum), gradient.dataType()),
+        tf.dtypes.cast(tf.constant(momentum), gradient.type()),
         ApplyMomentum.useNesterov(useNesterov));
   }
 
diff --git a/tensorflow-framework/src/main/java/org/tensorflow/framework/optimizers/Nadam.java b/tensorflow-framework/src/main/java/org/tensorflow/framework/optimizers/Nadam.java
index 48e5135c952..f9900a8ee78 100644
--- a/tensorflow-framework/src/main/java/org/tensorflow/framework/optimizers/Nadam.java
+++ b/tensorflow-framework/src/main/java/org/tensorflow/framework/optimizers/Nadam.java
@@ -1,6 +1,5 @@
 package org.tensorflow.framework.optimizers;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Graph;
 import org.tensorflow.Operand;
 import org.tensorflow.Output;
@@ -142,15 +141,15 @@ protected void createSlots(List<Output<? extends TType>> variables) {
     for (Output<? extends TType> v : variables) {
       createNadamSlot(v.asOutput());
     }
-    betaOnePower = tf.withName("beta1_power").variable(Shape.scalar(), TFloat32.DTYPE);
+    betaOnePower = tf.withName("beta1_power").variable(Shape.scalar(), TFloat32.class);
     Assign<TFloat32> betaOnePowerInit = tf.assign(betaOnePower, tf.constant(betaOne));
     ((Graph) tf.scope().env()).addInitializer(betaOnePowerInit);
 
-    betaTwoPower = tf.withName("beta2_power").variable(Shape.scalar(), TFloat32.DTYPE);
+    betaTwoPower = tf.withName("beta2_power").variable(Shape.scalar(), TFloat32.class);
     Assign<TFloat32> betaTwoPowerInit = tf.assign(betaTwoPower, tf.constant(betaTwo));
     ((Graph) tf.scope().env()).addInitializer(betaTwoPowerInit);
 
-    momentum = tf.withName("momentum").variable(Shape.scalar(), TFloat32.DTYPE);
+    momentum = tf.withName("momentum").variable(Shape.scalar(), TFloat32.class);
     Assign<TFloat32> momentumInit = tf.assign(momentum, tf.constant(1.0F));
     ((Graph) tf.scope().env()).addInitializer(momentumInit);
   }
@@ -163,14 +162,14 @@ protected void createSlots(List<Output<? extends TType>> variables) {
    */
   private <T extends TType> void createNadamSlot(Output<T> v) {
     Operand<T> firstMomentInitializer =
-        tf.fill(tf.shape(v), tf.dtypes.cast(tf.constant(0.0f), v.dataType()));
+        tf.fill(tf.shape(v), tf.dtypes.cast(tf.constant(0.0f), v.type()));
     createSlot(v.asOutput(), FIRST_MOMENT, firstMomentInitializer);
     Operand<T> secondMomentInitializer =
-        tf.fill(tf.shape(v), tf.dtypes.cast(tf.constant(0.0f), v.dataType()));
+        tf.fill(tf.shape(v), tf.dtypes.cast(tf.constant(0.0f), v.type()));
     createSlot(v.asOutput(), SECOND_MOMENT, secondMomentInitializer);
 
     Operand<T> momentumInitializer =
-        tf.fill(tf.shape(v), tf.dtypes.cast(tf.constant(1.0f), v.dataType()));
+        tf.fill(tf.shape(v), tf.dtypes.cast(tf.constant(1.0f), v.type()));
     createSlot(v.asOutput(), MOMENTUM, momentumInitializer);
   }
 
@@ -198,7 +197,7 @@ protected Optional<Op> prepare(String scopeName) {
                     point5,
                     tf.math.pow(
                         decayBaseConst,
-                        tf.math.mul(decayConst, tf.dtypes.cast(localStepConst, TFloat32.DTYPE))))));
+                        tf.math.mul(decayConst, tf.dtypes.cast(localStepConst, TFloat32.class))))));
 
     mT1 =
         tf.math.mul(
@@ -209,7 +208,7 @@ protected Optional<Op> prepare(String scopeName) {
                     point5,
                     tf.math.pow(
                         decayBaseConst,
-                        tf.math.mul(decayConst, tf.dtypes.cast(nextStepConst, TFloat32.DTYPE))))));
+                        tf.math.mul(decayConst, tf.dtypes.cast(nextStepConst, TFloat32.class))))));
 
     Operand<TFloat32> mScheduleNew = tf.math.mul(momentum, mT);
 
@@ -222,57 +221,57 @@ protected Optional<Op> prepare(String scopeName) {
     oneMinusMScheduleNew = tf.math.sub(one, mScheduleNew);
     oneMinusMScheduleNext = tf.math.sub(one, mScheduleNext);
     vTPrimeDenominator =
-        tf.math.sub(one, tf.math.pow(betaTwoConst, tf.dtypes.cast(localStepConst, TFloat32.DTYPE)));
+        tf.math.sub(one, tf.math.pow(betaTwoConst, tf.dtypes.cast(localStepConst, TFloat32.class)));
     return Optional.empty();
   }
 
   /** {@inheritDoc} */
   @Override
   protected <T extends TType> Op applyDense(Output<T> gradient, Output<T> variable) {
-    DataType<T> dType = gradient.dataType();
+    Class<T> type = gradient.type();
     Variable<T> m = getSlot(variable, FIRST_MOMENT).get(); // first Moment
     Variable<T> v = getSlot(variable, SECOND_MOMENT).get(); // Second Moment
 
     //  gPrime = grad / coefficients['oneMinusMScheduleNew']
-    Operand<T> gPrime = tf.math.div(gradient, tf.dtypes.cast(oneMinusMScheduleNew, dType));
+    Operand<T> gPrime = tf.math.div(gradient, tf.dtypes.cast(oneMinusMScheduleNew, type));
     // mT = (coefficients['beta_1_t'] * m + coefficients['one_minus_beta_1_t'] * grad)
     Operand<T> mT =
         tf.math.add(
-            tf.math.mul(tf.dtypes.cast(betaOneConst, dType), m),
-            tf.math.mul(tf.dtypes.cast(oneMinusBeta1, dType), gradient));
+            tf.math.mul(tf.dtypes.cast(betaOneConst, type), m),
+            tf.math.mul(tf.dtypes.cast(oneMinusBeta1, type), gradient));
     // mT = state_ops.assign(m, mT, use_locking=self._use_locking)
     // update m
     mT = tf.assign(m, mT, Assign.useLocking(true));
 
     // mTPrime = mT / coefficients['oneMinusMScheduleNext']
-    Operand<T> mTPrime = tf.math.div(mT, tf.dtypes.cast(oneMinusMScheduleNext, dType));
+    Operand<T> mTPrime = tf.math.div(mT, tf.dtypes.cast(oneMinusMScheduleNext, type));
 
     // vT = (coefficients['beta_2_t'] * v + coefficients['one_minus_beta_2_t'] *
     // math_ops.square(grad))
     Operand<T> vT =
         tf.math.add(
-            tf.math.mul(tf.dtypes.cast(betaTwoConst, dType), v),
-            tf.math.mul(tf.dtypes.cast(oneMinusBeta2, dType), tf.math.square(gradient)));
+            tf.math.mul(tf.dtypes.cast(betaTwoConst, type), v),
+            tf.math.mul(tf.dtypes.cast(oneMinusBeta2, type), tf.math.square(gradient)));
     // vT = state_ops.assign(v, vT, use_locking=self._use_locking)
     // update v
     vT = tf.assign(v, vT, Assign.useLocking(true));
 
     // vTPrime = vT / coefficients['vTPrimeDenominator']
-    Operand<T> vTPrime = tf.math.div(vT, tf.dtypes.cast(vTPrimeDenominator, dType));
+    Operand<T> vTPrime = tf.math.div(vT, tf.dtypes.cast(vTPrimeDenominator, type));
 
     // m_t_bar = (coefficients['oneMinusMT'] * gPrime + coefficients['mT1'] * mTPrime)
     Operand<T> m_t_bar =
         tf.math.add(
-            tf.math.mul(tf.dtypes.cast(oneMinusMT, dType), gPrime),
-            tf.math.mul(tf.dtypes.cast(mT1, dType), mTPrime));
+            tf.math.mul(tf.dtypes.cast(oneMinusMT, type), gPrime),
+            tf.math.mul(tf.dtypes.cast(mT1, type), mTPrime));
     // varT = var - coefficients['lr_t'] * m_t_bar / (math_ops.sqrt(vTPrime) +
     // coefficients['epsilon'])
     Operand<T> varT =
         tf.math.sub(
             variable,
             tf.math.div(
-                tf.math.mul(tf.dtypes.cast(learningRateConst, dType), m_t_bar),
-                tf.math.add(tf.math.sqrt(vTPrime), tf.dtypes.cast(epsilonConst, dType))));
+                tf.math.mul(tf.dtypes.cast(learningRateConst, type), m_t_bar),
+                tf.math.add(tf.math.sqrt(vTPrime), tf.dtypes.cast(epsilonConst, type))));
 
     return tf.assign(variable, varT, Assign.useLocking(true));
   }
diff --git a/tensorflow-framework/src/main/java/org/tensorflow/framework/optimizers/Optimizer.java b/tensorflow-framework/src/main/java/org/tensorflow/framework/optimizers/Optimizer.java
index 70f065814f7..fdf56da4a67 100644
--- a/tensorflow-framework/src/main/java/org/tensorflow/framework/optimizers/Optimizer.java
+++ b/tensorflow-framework/src/main/java/org/tensorflow/framework/optimizers/Optimizer.java
@@ -220,7 +220,7 @@ private <T extends TType> Optional<Variable<T>> getSlot(String varName, String s
   protected <T extends TType> void createSlot(
       Output<T> variable, String slotName, Operand<T> initializer) {
     Variable<T> slot =
-        tf.withName(createName(variable, slotName)).variable(variable.shape(), variable.dataType());
+        tf.withName(createName(variable, slotName)).variable(variable.shape(), variable.type());
     Assign<T> slotInit = tf.assign(slot, initializer);
     graph.addInitializer(slotInit);
     String varName = variable.op().name();
diff --git a/tensorflow-framework/src/main/java/org/tensorflow/framework/optimizers/RMSProp.java b/tensorflow-framework/src/main/java/org/tensorflow/framework/optimizers/RMSProp.java
index 9a48a9b8a7a..b3729dc367f 100644
--- a/tensorflow-framework/src/main/java/org/tensorflow/framework/optimizers/RMSProp.java
+++ b/tensorflow-framework/src/main/java/org/tensorflow/framework/optimizers/RMSProp.java
@@ -175,14 +175,14 @@ protected void createSlots(List<Output<? extends TType>> variables) {
    */
   private <T extends TType> void createRMSPropSlot(Output<T> v) {
     Operand<T> rmsInitializer =
-        tf.fill(tf.shape(v), tf.dtypes.cast(tf.constant(1.0f), v.dataType()));
+        tf.fill(tf.shape(v), tf.dtypes.cast(tf.constant(1.0f), v.type()));
     createSlot(v.asOutput(), RMS, rmsInitializer);
     Operand<T> momentumInitializer =
-        tf.fill(tf.shape(v), tf.dtypes.cast(tf.constant(0.0f), v.dataType()));
+        tf.fill(tf.shape(v), tf.dtypes.cast(tf.constant(0.0f), v.type()));
     createSlot(v.asOutput(), MOMENTUM, momentumInitializer);
     if (centered) {
       Operand<T> mgInitializer =
-          tf.fill(tf.shape(v), tf.dtypes.cast(tf.constant(0.0f), v.dataType()));
+          tf.fill(tf.shape(v), tf.dtypes.cast(tf.constant(0.0f), v.type()));
       createSlot(v.asOutput(), MG, mgInitializer);
     }
   }
@@ -199,20 +199,20 @@ protected <T extends TType> Op applyDense(Output<T> gradient, Output<T> variable
           mgSlot,
           rmsSlot,
           momentumSlot,
-          tf.dtypes.cast(tf.constant(learningRate), gradient.dataType()),
-          tf.dtypes.cast(tf.constant(decay), gradient.dataType()),
-          tf.dtypes.cast(tf.constant(momentum), gradient.dataType()),
-          tf.dtypes.cast(tf.constant(epsilon), gradient.dataType()),
+          tf.dtypes.cast(tf.constant(learningRate), gradient.type()),
+          tf.dtypes.cast(tf.constant(decay), gradient.type()),
+          tf.dtypes.cast(tf.constant(momentum), gradient.type()),
+          tf.dtypes.cast(tf.constant(epsilon), gradient.type()),
           gradient);
     }
     return tf.train.applyRmsProp(
         variable,
         rmsSlot,
         momentumSlot,
-        tf.dtypes.cast(tf.constant(learningRate), gradient.dataType()),
-        tf.dtypes.cast(tf.constant(decay), gradient.dataType()),
-        tf.dtypes.cast(tf.constant(momentum), gradient.dataType()),
-        tf.dtypes.cast(tf.constant(epsilon), gradient.dataType()),
+        tf.dtypes.cast(tf.constant(learningRate), gradient.type()),
+        tf.dtypes.cast(tf.constant(decay), gradient.type()),
+        tf.dtypes.cast(tf.constant(momentum), gradient.type()),
+        tf.dtypes.cast(tf.constant(epsilon), gradient.type()),
         gradient);
   }
 
diff --git a/tensorflow-framework/src/main/java/org/tensorflow/framework/utils/ShapeUtils.java b/tensorflow-framework/src/main/java/org/tensorflow/framework/utils/ShapeUtils.java
index 4abf90c4758..743e8114f6e 100644
--- a/tensorflow-framework/src/main/java/org/tensorflow/framework/utils/ShapeUtils.java
+++ b/tensorflow-framework/src/main/java/org/tensorflow/framework/utils/ShapeUtils.java
@@ -17,11 +17,14 @@
 import org.tensorflow.*;
 import org.tensorflow.ndarray.NdArray;
 import org.tensorflow.ndarray.Shape;
+import org.tensorflow.ndarray.buffer.DataBuffer;
+import org.tensorflow.ndarray.buffer.DataBuffers;
+import org.tensorflow.ndarray.buffer.IntDataBuffer;
 import org.tensorflow.op.Scope;
-import org.tensorflow.tensor.Tensor;
 import org.tensorflow.types.TInt32;
 import org.tensorflow.types.TInt64;
 import org.tensorflow.types.TUint8;
+import org.tensorflow.types.family.TInteger;
 import org.tensorflow.types.family.TNumber;
 
 import java.util.ArrayList;
@@ -37,7 +40,7 @@ public class ShapeUtils {
    * @param dims the Operand containing the shape values
    * @return a new Shape based on an Operand that contains dimensions
    */
-  public static <T extends TNumber> Shape toShape(Scope scope, Operand<T> dims) {
+  public static <T extends TInteger> Shape toShape(Scope scope, Operand<T> dims) {
     long[] longDims = getLongArray(scope, dims);
     return Shape.of(longDims);
   }
@@ -63,65 +66,37 @@ public static int[] getIntArray(Scope scope, Operand<TInt32> dims) {
    * @return the long array
    * @throws java.lang.IllegalArgumentException if the dims type is not an integer
    */
-  public static <T extends TNumber> long[] getLongArray(Scope scope, Operand<T> dims) {
-    DataType<T> dType = dims.asOutput().dataType();
-    if (!dType.isInteger()) {
-      throw new IllegalArgumentException("the data type must be an integer type");
-    }
-    List<Long> result = new ArrayList<>();
-
+  public static <T extends TInteger> long[] getLongArray(Scope scope, Operand<T> dims) {
     if (scope.env().isEager()) {
-      if (dType.equals(TInt32.DTYPE)) {
-        @SuppressWarnings("unchecked")
-        Operand<TInt32> idims = (Operand<TInt32>) dims;
-
-        idims.asOutput().data().scalars().forEach(s -> result.add((long) s.getInt()));
-      } else if (dType.equals(TInt64.DTYPE)) {
-        @SuppressWarnings("unchecked")
-        Operand<TInt64> ldims = (Operand<TInt64>) dims;
-        ldims.asOutput().data().scalars().forEach(s -> result.add(s.getLong()));
-      } else if (dType.equals(TUint8.DTYPE)) {
-        @SuppressWarnings("unchecked")
-        Operand<TUint8> udims = (Operand<TUint8>) dims;
-        udims.asOutput().data().scalars().forEach(s -> result.add(s.getObject().longValue()));
-      } else { // shouldn't happen
-        throw new IllegalArgumentException("the data type must be an integer type");
-      }
-
-    } else {
-      try (Session session = new Session((Graph) scope.env())) {
-        if (dType.equals(TInt32.DTYPE)) {
-          try (Tensor<TInt32> tensorResult =
-              session.runner().fetch(dims).run().get(0).expect(TInt32.DTYPE)) {
-            tensorResult.data().scalars().forEach(s -> result.add((long) s.getInt()));
-          }
-        } else if (dType.equals(TInt64.DTYPE)) {
-          try (Tensor<TInt64> tensorResult =
-              session.runner().fetch(dims).run().get(0).expect(TInt64.DTYPE)) {
-            tensorResult.data().scalars().forEach(s -> result.add(s.getLong()));
-          }
-        } else if (dType.equals(TUint8.DTYPE)) {
-          try (Tensor<TUint8> tensorResult =
-              session.runner().fetch(dims).run().get(0).expect(TUint8.DTYPE)) {
-            tensorResult.data().scalars().forEach(s -> result.add(s.getObject().longValue()));
-          }
-        } else { // shouldn't happen
-          throw new IllegalArgumentException("the data type must be an integer type");
-        }
-      }
+      return getLongArray(dims.asTensor());
+    }
+    try (Session session = new Session((Graph)scope.env());
+        T tensor = session.runner().fetch(dims).run().get(0)) {
+      return getLongArray(tensor);
     }
-    return result.stream().mapToLong(i -> i).toArray();
   }
 
   /**
-   * Gets the shape for the data within a Tensor
+   * Converts a TInt32 or TInt64 to a java long array
    *
-   * @param tensor the tensor
-   * @return the Shape of the tensor's data;
+   * @param scope the TensorFlow scope
+   * @param dims the dimension tensor
+   * @param <T> the type of the dimensions, must either be TInt32 or TInt64 type
+   * @return the long array
+   * @throws java.lang.IllegalArgumentException if the dims type is not an integer
    */
-  public static <T extends TNumber> Shape getShape(Tensor<T> tensor) {
-    NdArray<?> data = (NdArray<?>) tensor.data();
-    return data.shape();
+  public static <T extends TInteger> long[] getLongArray(T dims) {
+    List<Long> result = new ArrayList<>();
+    if (dims instanceof TInt32) {
+      ((TInt32)dims).scalars().forEach(s -> result.add((long) s.getInt()));
+    } else if (dims instanceof TInt64) {
+      ((TInt64)dims).scalars().forEach(s -> result.add(s.getLong()));
+    } else if (dims instanceof TUint8) {
+      ((TUint8)dims).scalars().forEach(s -> result.add(s.getObject().longValue()));
+    } else { // shouldn't happen
+      throw new IllegalArgumentException("the data type must be an integer type");
+    }
+    return result.stream().mapToLong(i -> i).toArray();
   }
 
   /**
diff --git a/tensorflow-framework/src/test/java/org/tensorflow/framework/data/BatchDatasetTest.java b/tensorflow-framework/src/test/java/org/tensorflow/framework/data/BatchDatasetTest.java
index 137b07e9289..41be23a8f6f 100644
--- a/tensorflow-framework/src/test/java/org/tensorflow/framework/data/BatchDatasetTest.java
+++ b/tensorflow-framework/src/test/java/org/tensorflow/framework/data/BatchDatasetTest.java
@@ -39,13 +39,13 @@ public void testEagerBatchDataset() {
             Arrays.asList(
                 tf.constant(testMatrix1),
                 tf.constant(testMatrix2)),
-            Arrays.asList(TInt32.DTYPE, TInt32.DTYPE))
+            Arrays.asList(TInt32.class, TInt32.class))
         .batch(2);
 
     int count = 0;
     for (List<Operand<?>> components : dataset) {
-      try (TInt32 batch1 = components.get(0).asTensor(TInt32.DTYPE);
-           TInt32 batch2 = components.get(1).asTensor(TInt32.DTYPE);) {
+      try (TInt32 batch1 = (TInt32)components.get(0).asTensor();
+           TInt32 batch2 = (TInt32)components.get(1).asTensor()) {
         assertEquals(testMatrix1.slice(range(count, count + 2)), batch1);
         assertEquals(testMatrix2.slice(range(count, count + 2)), batch2);
         count += 2;
@@ -59,13 +59,13 @@ public void testDropLastBatch() {
     Dataset dataset = Dataset
         .fromTensorSlices(tf,
             Arrays.asList(tf.constant(testMatrix1), tf.constant(testMatrix2)),
-            Arrays.asList(TInt32.DTYPE, TInt32.DTYPE))
+            Arrays.asList(TInt32.class, TInt32.class))
         .batch(3, true);
 
     int count = 0;
     for (List<Operand<?>> components : dataset) {
-      try (TInt32 batch1 = components.get(0).asTensor(TInt32.DTYPE);
-          TInt32 batch2 = components.get(1).asTensor(TInt32.DTYPE);) {
+      try (TInt32 batch1 = (TInt32)components.get(0).asTensor();
+          TInt32 batch2 = (TInt32)components.get(1).asTensor()) {
         assertEquals(testMatrix1.slice(range(count, count + 3)), batch1);
         assertEquals(testMatrix2.slice(range(count, count + 3)), batch2);
         count += 3;
@@ -79,15 +79,15 @@ public void testKeepLastBatch() {
     Dataset dataset = Dataset
         .fromTensorSlices(tf,
             Arrays.asList(tf.constant(testMatrix1), tf.constant(testMatrix2)),
-            Arrays.asList(TInt32.DTYPE, TInt32.DTYPE))
+            Arrays.asList(TInt32.class, TInt32.class))
         .batch(3, false);
 
     int count = 0;
     boolean foundLastBatch = false;
 
     for (List<Operand<?>> components : dataset) {
-      try (TInt32 batch1 = components.get(0).asTensor(TInt32.DTYPE);
-           TInt32 batch2 = components.get(1).asTensor(TInt32.DTYPE);) {
+      try (TInt32 batch1 = (TInt32)components.get(0).asTensor();
+           TInt32 batch2 = (TInt32)components.get(1).asTensor();) {
         if (count == 0) {
           assertEquals(testMatrix1.slice(range(count, count + 3)), batch1);
           assertEquals(testMatrix2.slice(range(count, count + 3)), batch2);
diff --git a/tensorflow-framework/src/test/java/org/tensorflow/framework/data/DatasetIteratorTest.java b/tensorflow-framework/src/test/java/org/tensorflow/framework/data/DatasetIteratorTest.java
index fd265bf1826..c5cb3978b4b 100644
--- a/tensorflow-framework/src/test/java/org/tensorflow/framework/data/DatasetIteratorTest.java
+++ b/tensorflow-framework/src/test/java/org/tensorflow/framework/data/DatasetIteratorTest.java
@@ -16,17 +16,17 @@
 package org.tensorflow.framework.data;
 
 import org.junit.jupiter.api.Test;
-import org.tensorflow.DataType;
 import org.tensorflow.Graph;
 import org.tensorflow.Operand;
 import org.tensorflow.Session;
-import org.tensorflow.tensor.Tensor;
+import org.tensorflow.types.family.TType;
 import org.tensorflow.exceptions.TFOutOfRangeException;
 import org.tensorflow.op.Ops;
 import org.tensorflow.types.TInt32;
 
 import java.util.Arrays;
 import java.util.List;
+import org.tensorflow.util.TensorList;
 
 import static org.junit.jupiter.api.Assertions.assertEquals;
 
@@ -39,7 +39,7 @@ public void testGraphIteration() {
 
       List<Operand<?>> tensors = Arrays.asList(tf.constant(testMatrix1), tf.constant(testMatrix2));
 
-      List<DataType<?>> dataTypes = Arrays.asList(TInt32.DTYPE, TInt32.DTYPE);
+      List<Class<? extends TType>> dataTypes = Arrays.asList(TInt32.class, TInt32.class);
 
       Dataset dataset = Dataset.fromTensorSlices(tf, tensors, dataTypes);
       DatasetIterator iterator = dataset.makeOneShotIterator();
@@ -53,15 +53,12 @@ public void testGraphIteration() {
 
         int batches = 0;
         while (true) {
-          try {
-            List<Tensor<?>> outputs = session.runner().fetch(x).fetch(y).run();
-
-            try (TInt32 xBatch = outputs.get(0).expect(TInt32.DTYPE);
-                TInt32 yBatch = outputs.get(1).expect(TInt32.DTYPE)) {
-              assertEquals(testMatrix1.get(batches), xBatch);
-              assertEquals(testMatrix2.get(batches), yBatch);
-              batches++;
-            }
+          try (TensorList outputs = session.runner().fetch(x).fetch(y).run()) {
+            TInt32 xBatch = outputs.get(0);
+            TInt32 yBatch = outputs.get(1);
+            assertEquals(testMatrix1.get(batches), xBatch);
+            assertEquals(testMatrix2.get(batches), yBatch);
+            batches++;
           } catch (TFOutOfRangeException e) {
             break;
           }
@@ -77,13 +74,13 @@ public void testEagerIteration() {
 
     List<Operand<?>> tensors = Arrays.asList(tf.constant(testMatrix1), tf.constant(testMatrix2));
 
-    List<DataType<?>> dataTypes = Arrays.asList(TInt32.DTYPE, TInt32.DTYPE);
+    List<Class<? extends TType>> dataTypes = Arrays.asList(TInt32.class, TInt32.class);
 
     Dataset dataset = Dataset.fromTensorSlices(tf, tensors, dataTypes);
     int count = 0;
     for (List<Operand<?>> outputs : dataset) {
-      try (TInt32 batch1 = outputs.get(0).asTensor(TInt32.DTYPE);
-          TInt32 batch2 = outputs.get(1).asTensor(TInt32.DTYPE); ) {
+      try (TInt32 batch1 = (TInt32)outputs.get(0).asTensor();
+          TInt32 batch2 = (TInt32)outputs.get(1).asTensor()) {
         assertEquals(testMatrix1.get(count), batch1);
         assertEquals(testMatrix2.get(count), batch2);
         count++;
diff --git a/tensorflow-framework/src/test/java/org/tensorflow/framework/data/MapDatasetTest.java b/tensorflow-framework/src/test/java/org/tensorflow/framework/data/MapDatasetTest.java
index 3b9f89d05d6..c6be6960089 100644
--- a/tensorflow-framework/src/test/java/org/tensorflow/framework/data/MapDatasetTest.java
+++ b/tensorflow-framework/src/test/java/org/tensorflow/framework/data/MapDatasetTest.java
@@ -17,11 +17,10 @@
 
 import org.junit.jupiter.api.BeforeEach;
 import org.junit.jupiter.api.Test;
-import org.tensorflow.DataType;
 import org.tensorflow.Graph;
 import org.tensorflow.Operand;
 import org.tensorflow.Session;
-import org.tensorflow.tensor.Tensor;
+import org.tensorflow.types.family.TType;
 import org.tensorflow.exceptions.TFOutOfRangeException;
 import org.tensorflow.op.Ops;
 import org.tensorflow.ndarray.IntNdArray;
@@ -30,6 +29,7 @@
 
 import java.util.Arrays;
 import java.util.List;
+import org.tensorflow.util.TensorList;
 
 import static org.junit.jupiter.api.Assertions.assertEquals;
 
@@ -60,13 +60,13 @@ public void testGraphIteration() {
 
       List<Operand<?>> tensors = Arrays.asList(tf.constant(testMatrix1), tf.constant(testMatrix2));
 
-      List<DataType<?>> dataTypes = Arrays.asList(TInt32.DTYPE, TInt32.DTYPE);
+      List<Class<? extends TType>> dataTypes = Arrays.asList(TInt32.class, TInt32.class);
 
       Dataset dataset =
           Dataset.fromTensorSlices(tf, tensors, dataTypes)
               .mapAllComponents(
                   component ->
-                      tf.math.mul(component.asOutput().expect(TInt32.DTYPE), tf.constant(2)));
+                      tf.math.mul(component.asOutput().expect(TInt32.class), tf.constant(2)));
 
       DatasetIterator iterator = dataset.makeOneShotIterator();
       List<Operand<?>> components = iterator.getNext();
@@ -78,15 +78,12 @@ public void testGraphIteration() {
 
         int batches = 0;
         while (true) {
-          try {
-            List<Tensor<?>> outputs = session.runner().fetch(X).fetch(y).run();
-
-            try (TInt32 XBatch = outputs.get(0).expect(TInt32.DTYPE);
-                TInt32 yBatch = outputs.get(1).expect(TInt32.DTYPE)) {
-              assertEquals(mapped1.get(batches), XBatch);
+          try (TensorList outputs = session.runner().fetch(X).fetch(y).run()) {
+              TInt32 xBatch = outputs.get(0);
+              TInt32 yBatch = outputs.get(1);
+              assertEquals(mapped1.get(batches), xBatch);
               assertEquals(mapped2.get(batches), yBatch);
               batches++;
-            }
           } catch (TFOutOfRangeException e) {
             break;
           }
@@ -103,17 +100,16 @@ public void testEagerIteration() {
 
     List<Operand<?>> tensors = Arrays.asList(tf.constant(testMatrix1), tf.constant(testMatrix2));
 
-    List<DataType<?>> dataTypes = Arrays.asList(TInt32.DTYPE, TInt32.DTYPE);
+    List<Class<? extends TType>> dataTypes = Arrays.asList(TInt32.class, TInt32.class);
 
     Dataset dataset =
         Dataset.fromTensorSlices(tf, tensors, dataTypes)
-            .mapAllComponents(
-                op -> tf.math.mul(op.asOutput().expect(TInt32.DTYPE), tf.constant(2)));
+            .mapAllComponents(op -> tf.math.mul(op.expect(TInt32.class), tf.constant(2)));
 
     int count = 0;
     for (List<Operand<?>> outputs : dataset) {
-      try (TInt32 XBatch = outputs.get(0).asTensor(TInt32.DTYPE);
-          TInt32 yBatch = outputs.get(1).asTensor(TInt32.DTYPE); ) {
+      try (TInt32 XBatch = (TInt32)outputs.get(0).asTensor();
+          TInt32 yBatch = (TInt32)outputs.get(1).asTensor()) {
         assertEquals(mapped1.get(count), XBatch);
         assertEquals(mapped2.get(count), yBatch);
         count++;
diff --git a/tensorflow-framework/src/test/java/org/tensorflow/framework/data/SkipDatasetTest.java b/tensorflow-framework/src/test/java/org/tensorflow/framework/data/SkipDatasetTest.java
index 58a4ea93964..d0cdb4527a5 100644
--- a/tensorflow-framework/src/test/java/org/tensorflow/framework/data/SkipDatasetTest.java
+++ b/tensorflow-framework/src/test/java/org/tensorflow/framework/data/SkipDatasetTest.java
@@ -34,13 +34,13 @@ public void testEagerSkipDataset() {
         Dataset.fromTensorSlices(
                 tf,
                 Arrays.asList(tf.constant(testMatrix1), tf.constant(testMatrix2)),
-                Arrays.asList(TInt32.DTYPE, TInt32.DTYPE))
+                Arrays.asList(TInt32.class, TInt32.class))
             .skip(2);
 
     int count = 2;
     for (List<Operand<?>> components : dataset) {
-      try (TInt32 batch1 = components.get(0).asTensor(TInt32.DTYPE);
-          TInt32 batch2 = components.get(1).asTensor(TInt32.DTYPE); ) {
+      try (TInt32 batch1 = (TInt32)components.get(0).asTensor();
+          TInt32 batch2 = (TInt32)components.get(1).asTensor()) {
         assertEquals(testMatrix1.get(count), batch1);
         assertEquals(testMatrix2.get(count), batch2);
         count++;
diff --git a/tensorflow-framework/src/test/java/org/tensorflow/framework/data/TakeDatasetTest.java b/tensorflow-framework/src/test/java/org/tensorflow/framework/data/TakeDatasetTest.java
index a4e94d75ca9..79a2e79c72e 100644
--- a/tensorflow-framework/src/test/java/org/tensorflow/framework/data/TakeDatasetTest.java
+++ b/tensorflow-framework/src/test/java/org/tensorflow/framework/data/TakeDatasetTest.java
@@ -35,13 +35,13 @@ public void testEagerTakeDataset() {
         Dataset.fromTensorSlices(
                 tf,
                 Arrays.asList(tf.constant(testMatrix1), tf.constant(testMatrix2)),
-                Arrays.asList(TInt32.DTYPE, TInt32.DTYPE))
+                Arrays.asList(TInt32.class, TInt32.class))
             .take(4);
 
     int count = 0;
     for (List<Operand<?>> components : dataset) {
-      try (TInt32 batch1 = components.get(0).asTensor(TInt32.DTYPE);
-          TInt32 batch2 = components.get(1).asTensor(TInt32.DTYPE); ) {
+      try (TInt32 batch1 = (TInt32)components.get(0).asTensor();
+          TInt32 batch2 = (TInt32)components.get(1).asTensor()) {
         assertEquals(testMatrix1.get(count), batch1);
         assertEquals(testMatrix2.get(count), batch2);
         count++;
diff --git a/tensorflow-framework/src/test/java/org/tensorflow/framework/initializers/ConstantTest.java b/tensorflow-framework/src/test/java/org/tensorflow/framework/initializers/ConstantTest.java
index 46e4232d5ae..4e81e0620e6 100644
--- a/tensorflow-framework/src/test/java/org/tensorflow/framework/initializers/ConstantTest.java
+++ b/tensorflow-framework/src/test/java/org/tensorflow/framework/initializers/ConstantTest.java
@@ -52,7 +52,7 @@ public void testCallUInt() {
         Ops tf = session.getTF();
         Shape shape = Shape.of(2, 2);
         Constant<TUint8> instance = new Constant<>(tf, 0xf);
-        Operand<TUint8> operand = instance.call(tf.constant(shape), TUint8.DTYPE);
+        Operand<TUint8> operand = instance.call(tf.constant(shape), TUint8.class);
         session.evaluate(expected, operand);
       }
   }
@@ -68,7 +68,7 @@ public void testCallInt() {
         Ops tf = session.getTF();
         Shape shape = Shape.of(2, 2);
         Constant<TInt32> instance = new Constant<>(tf, 0xf);
-        Operand<TInt32> operand = instance.call(tf.constant(shape), TInt32.DTYPE);
+        Operand<TInt32> operand = instance.call(tf.constant(shape), TInt32.class);
         session.evaluate(expected, operand);
       }
   }
@@ -84,7 +84,7 @@ public void testCallLong() {
         Ops tf = session.getTF();
         Shape shape = Shape.of(2, 2);
         Constant<TInt64> instance = new Constant<>(tf, 0xffL);
-        Operand<TInt64> operand = instance.call(tf.constant(shape), TInt64.DTYPE);
+        Operand<TInt64> operand = instance.call(tf.constant(shape), TInt64.class);
         session.evaluate(expected, operand);
       }
   }
@@ -98,7 +98,7 @@ public void testCallFloat() {
         Ops tf = session.getTF();
         Shape shape = Shape.of(2, 2);
         Constant<TFloat32> instance = new Constant<>(tf, 12.F);
-        Operand<TFloat32> operand = instance.call(tf.constant(shape), TFloat32.DTYPE);
+        Operand<TFloat32> operand = instance.call(tf.constant(shape), TFloat32.class);
         session.evaluate(expected, operand);
       }
   }
@@ -113,7 +113,7 @@ public void testCallDouble() {
         Shape shape = Shape.of(2, 2);
 
         Constant<TFloat64> instance = new Constant<>(tf, 11.);
-        Operand<TFloat64> operand = instance.call(tf.constant(shape), TFloat64.DTYPE);
+        Operand<TFloat64> operand = instance.call(tf.constant(shape), TFloat64.class);
         session.evaluate(expected, operand);
       }
   }
@@ -130,7 +130,7 @@ public void testCallString() {
               Shape shape = Shape.of(2, 2);
 
               Constant<TString> instance = new Constant<>(tf, 22);
-              instance.call(tf.constant(shape), TString.DTYPE);
+              instance.call(tf.constant(shape), TString.class);
               fail("IllegalArgumentException  should have been thrown for TString");
             }
           });
@@ -146,7 +146,7 @@ public void testCallBool() {
         Boolean[] expected = {true, true, true, true};
 
         Constant<TBool> instance = new Constant<>(tf, true);
-        Operand<TBool> operand = instance.call(tf.constant(shape), TBool.DTYPE);
+        Operand<TBool> operand = instance.call(tf.constant(shape), TBool.class);
         session.evaluate(expected, operand);
       }
   }
@@ -159,8 +159,8 @@ public void testReproducible() {
         Shape shape = Shape.of(2, 2);
 
         Constant<TFloat64> instance = new Constant<>(tf, 11.);
-        Operand<TFloat64> operand1 = instance.call(tf.constant(shape), TFloat64.DTYPE);
-        Operand<TFloat64> operand2 = instance.call(tf.constant(shape), TFloat64.DTYPE);
+        Operand<TFloat64> operand1 = instance.call(tf.constant(shape), TFloat64.class);
+        Operand<TFloat64> operand2 = instance.call(tf.constant(shape), TFloat64.class);
         session.evaluate(operand1, operand2);
       }
   }
diff --git a/tensorflow-framework/src/test/java/org/tensorflow/framework/initializers/GlorotTest.java b/tensorflow-framework/src/test/java/org/tensorflow/framework/initializers/GlorotTest.java
index a68bf2a0a98..e9769806928 100644
--- a/tensorflow-framework/src/test/java/org/tensorflow/framework/initializers/GlorotTest.java
+++ b/tensorflow-framework/src/test/java/org/tensorflow/framework/initializers/GlorotTest.java
@@ -51,9 +51,9 @@ public void testCallNormalFloat() {
       try (TestSession session = TestSession.createTestSession(tfMode)) {
         Ops tf = session.getTF();
         Shape shape = Shape.of(2, 2);
-        Glorot<TFloat32, TFloat32> instance = new Glorot<>(tf, Distribution.TRUNCATED_NORMAL, SEED);
+        Glorot<TFloat32> instance = new Glorot<>(tf, Distribution.TRUNCATED_NORMAL, SEED);
 
-        Operand<TFloat32> operand = instance.call(tf.constant(shape), TFloat32.DTYPE);
+        Operand<TFloat32> operand = instance.call(tf.constant(shape), TFloat32.class);
         session.evaluate(expected, operand);
       }
   }
@@ -68,8 +68,8 @@ public void testCallNormalDouble() {
         Ops tf = session.getTF();
         Shape shape = Shape.of(2, 2);
 
-        Glorot<TFloat64, TFloat64> instance = new Glorot<>(tf, Distribution.TRUNCATED_NORMAL, SEED);
-        Operand<TFloat64> operand = instance.call(tf.constant(shape), TFloat64.DTYPE);
+        Glorot<TFloat64> instance = new Glorot<>(tf, Distribution.TRUNCATED_NORMAL, SEED);
+        Operand<TFloat64> operand = instance.call(tf.constant(shape), TFloat64.class);
         session.evaluate(expected, operand);
       }
   }
@@ -82,8 +82,8 @@ public void testCallUniformFloat() {
       try (TestSession session = TestSession.createTestSession(tfMode)) {
         Ops tf = session.getTF();
         Shape shape = Shape.of(2, 2);
-        Glorot<TFloat32, TFloat32> instance = new Glorot<>(tf, Distribution.UNIFORM, SEED);
-        Operand<TFloat32> operand = instance.call(tf.constant(shape), TFloat32.DTYPE);
+        Glorot<TFloat32> instance = new Glorot<>(tf, Distribution.UNIFORM, SEED);
+        Operand<TFloat32> operand = instance.call(tf.constant(shape), TFloat32.class);
         session.evaluate(expected, operand);
       }
   }
@@ -97,8 +97,8 @@ public void testCallUniformDouble() {
       try (TestSession session = TestSession.createTestSession(tfMode)) {
         Ops tf = session.getTF();
         Shape shape = Shape.of(2, 2);
-        Glorot<TFloat64, TFloat64> instance = new Glorot<>(tf, Distribution.UNIFORM, SEED);
-        Operand<TFloat64> operand = instance.call(tf.constant(shape), TFloat64.DTYPE);
+        Glorot<TFloat64> instance = new Glorot<>(tf, Distribution.UNIFORM, SEED);
+        Operand<TFloat64> operand = instance.call(tf.constant(shape), TFloat64.class);
         session.evaluate(expected, operand);
       }
   }
@@ -109,9 +109,9 @@ public void testCallNormalReproducible() {
       try (TestSession session = TestSession.createTestSession(tfMode)) {
         Ops tf = session.getTF();
         Shape shape = Shape.of(2, 2);
-        Glorot<TFloat64, TFloat64> instance = new Glorot<>(tf, Distribution.TRUNCATED_NORMAL, SEED);
-        Operand<TFloat64> operand1 = instance.call(tf.constant(shape), TFloat64.DTYPE);
-        Operand<TFloat64> operand2 = instance.call(tf.constant(shape), TFloat64.DTYPE);
+        Glorot<TFloat64> instance = new Glorot<>(tf, Distribution.TRUNCATED_NORMAL, SEED);
+        Operand<TFloat64> operand1 = instance.call(tf.constant(shape), TFloat64.class);
+        Operand<TFloat64> operand2 = instance.call(tf.constant(shape), TFloat64.class);
         session.evaluate(operand1, operand2);
       }
   }
@@ -122,9 +122,9 @@ public void testCallUniformReproducible() {
       try (TestSession session = TestSession.createTestSession(tfMode)) {
         Ops tf = session.getTF();
         Shape shape = Shape.of(2, 2);
-        Glorot<TFloat64, TFloat64> instance = new Glorot<>(tf, Distribution.UNIFORM, SEED);
-        Operand<TFloat64> operand1 = instance.call(tf.constant(shape), TFloat64.DTYPE);
-        Operand<TFloat64> operand2 = instance.call(tf.constant(shape), TFloat64.DTYPE);
+        Glorot<TFloat64> instance = new Glorot<>(tf, Distribution.UNIFORM, SEED);
+        Operand<TFloat64> operand1 = instance.call(tf.constant(shape), TFloat64.class);
+        Operand<TFloat64> operand2 = instance.call(tf.constant(shape), TFloat64.class);
         session.evaluate(operand1, operand2);
       }
   }
@@ -135,10 +135,10 @@ public void testCallNORMALReproducible() {
       try (TestSession session = TestSession.createTestSession(tfMode)) {
         Ops tf = session.getTF();
         Shape shape = Shape.of(2, 2);
-        Glorot<TFloat64, TFloat64> instance =
+        Glorot<TFloat64> instance =
             new Glorot<>(tf, Distribution.NORMAL, SEED);
-        Operand<TFloat64> operand1 = instance.call(tf.constant(shape), TFloat64.DTYPE);
-        Operand<TFloat64> operand2 = instance.call(tf.constant(shape), TFloat64.DTYPE);
+        Operand<TFloat64> operand1 = instance.call(tf.constant(shape), TFloat64.class);
+        Operand<TFloat64> operand2 = instance.call(tf.constant(shape), TFloat64.class);
         session.evaluate(operand1, operand2);
       }
   }
diff --git a/tensorflow-framework/src/test/java/org/tensorflow/framework/initializers/HeTest.java b/tensorflow-framework/src/test/java/org/tensorflow/framework/initializers/HeTest.java
index 468759d347f..8953fa3005e 100644
--- a/tensorflow-framework/src/test/java/org/tensorflow/framework/initializers/HeTest.java
+++ b/tensorflow-framework/src/test/java/org/tensorflow/framework/initializers/HeTest.java
@@ -51,8 +51,8 @@ public void testCallNormalFloat() {
       try (TestSession session = TestSession.createTestSession(tfMode)) {
         Ops tf = session.getTF();
         Shape shape = Shape.of(2, 2);
-        He<TFloat32, TFloat32> instance = new He<>(tf, Distribution.TRUNCATED_NORMAL, SEED);
-        Operand<TFloat32> operand = instance.call(tf.constant(shape), TFloat32.DTYPE);
+        He<TFloat32> instance = new He<>(tf, Distribution.TRUNCATED_NORMAL, SEED);
+        Operand<TFloat32> operand = instance.call(tf.constant(shape), TFloat32.class);
         session.evaluate(expected, operand);
       }
   }
@@ -66,8 +66,8 @@ public void testCallNormalDouble() {
       try (TestSession session = TestSession.createTestSession(tfMode)) {
         Ops tf = session.getTF();
         Shape shape = Shape.of(2, 2);
-        He<TFloat64, TFloat64> instance = new He<>(tf, Distribution.TRUNCATED_NORMAL, SEED);
-        Operand<TFloat64> operand = instance.call(tf.constant(shape), TFloat64.DTYPE);
+        He<TFloat64> instance = new He<>(tf, Distribution.TRUNCATED_NORMAL, SEED);
+        Operand<TFloat64> operand = instance.call(tf.constant(shape), TFloat64.class);
         session.evaluate(expected, operand);
       }
   }
@@ -80,8 +80,8 @@ public void testCallFloat() {
       try (TestSession session = TestSession.createTestSession(tfMode)) {
         Ops tf = session.getTF();
         Shape shape = Shape.of(2, 2);
-        He<TFloat32, TFloat32> instance = new He<>(tf, Distribution.UNIFORM, SEED);
-        Operand<TFloat32> operand = instance.call(tf.constant(shape), TFloat32.DTYPE);
+        He<TFloat32> instance = new He<>(tf, Distribution.UNIFORM, SEED);
+        Operand<TFloat32> operand = instance.call(tf.constant(shape), TFloat32.class);
         session.evaluate(expected, operand);
       }
   }
@@ -95,8 +95,8 @@ public void testCallDouble() {
       try (TestSession session = TestSession.createTestSession(tfMode)) {
         Ops tf = session.getTF();
         Shape shape = Shape.of(2, 2);
-        He<TFloat64, TFloat64> instance = new He<>(tf, Distribution.UNIFORM, SEED);
-        Operand<TFloat64> operand = instance.call(tf.constant(shape), TFloat64.DTYPE);
+        He<TFloat64> instance = new He<>(tf, Distribution.UNIFORM, SEED);
+        Operand<TFloat64> operand = instance.call(tf.constant(shape), TFloat64.class);
         session.evaluate(expected, operand);
       }
   }
@@ -107,9 +107,9 @@ public void testCallNormalReproducible() {
       try (TestSession session = TestSession.createTestSession(tfMode)) {
         Ops tf = session.getTF();
         Shape shape = Shape.of(2, 2);
-        He<TFloat64, TFloat64> instance = new He<>(tf, Distribution.TRUNCATED_NORMAL, SEED);
-        Operand<TFloat64> operand1 = instance.call(tf.constant(shape), TFloat64.DTYPE);
-        Operand<TFloat64> operand2 = instance.call(tf.constant(shape), TFloat64.DTYPE);
+        He<TFloat64> instance = new He<>(tf, Distribution.TRUNCATED_NORMAL, SEED);
+        Operand<TFloat64> operand1 = instance.call(tf.constant(shape), TFloat64.class);
+        Operand<TFloat64> operand2 = instance.call(tf.constant(shape), TFloat64.class);
         session.evaluate(operand1, operand2);
       }
   }
@@ -120,9 +120,9 @@ public void testCallUniformReproducible() {
       try (TestSession session = TestSession.createTestSession(tfMode)) {
         Ops tf = session.getTF();
         Shape shape = Shape.of(2, 2);
-        He<TFloat64, TFloat64> instance = new He<>(tf, Distribution.UNIFORM, SEED);
-        Operand<TFloat64> operand1 = instance.call(tf.constant(shape), TFloat64.DTYPE);
-        Operand<TFloat64> operand2 = instance.call(tf.constant(shape), TFloat64.DTYPE);
+        He<TFloat64> instance = new He<>(tf, Distribution.UNIFORM, SEED);
+        Operand<TFloat64> operand1 = instance.call(tf.constant(shape), TFloat64.class);
+        Operand<TFloat64> operand2 = instance.call(tf.constant(shape), TFloat64.class);
         session.evaluate(operand1, operand2);
       }
   }
@@ -133,9 +133,9 @@ public void testCallNORMALReproducible() {
       try (TestSession session = TestSession.createTestSession(tfMode)) {
         Ops tf = session.getTF();
         Shape shape = Shape.of(2, 2);
-        He<TFloat64, TFloat64> instance = new He<>(tf, Distribution.NORMAL, SEED);
-        Operand<TFloat64> operand1 = instance.call(tf.constant(shape), TFloat64.DTYPE);
-        Operand<TFloat64> operand2 = instance.call(tf.constant(shape), TFloat64.DTYPE);
+        He<TFloat64> instance = new He<>(tf, Distribution.NORMAL, SEED);
+        Operand<TFloat64> operand1 = instance.call(tf.constant(shape), TFloat64.class);
+        Operand<TFloat64> operand2 = instance.call(tf.constant(shape), TFloat64.class);
         session.evaluate(operand1, operand2);
       }
   }
diff --git a/tensorflow-framework/src/test/java/org/tensorflow/framework/initializers/IdentityTest.java b/tensorflow-framework/src/test/java/org/tensorflow/framework/initializers/IdentityTest.java
index adb6c0c118a..6eee5473937 100644
--- a/tensorflow-framework/src/test/java/org/tensorflow/framework/initializers/IdentityTest.java
+++ b/tensorflow-framework/src/test/java/org/tensorflow/framework/initializers/IdentityTest.java
@@ -45,23 +45,6 @@ public void setUp() {}
   @AfterEach
   public void tearDown() {}
 
-  /** Test of call method, of class Orthogonal. */
-  @Test
-  public void testCallInt() {
-    for (TestSession.Mode tfMode : tfModes)
-      assertThrows(
-          java.lang.IllegalArgumentException.class,
-          () -> {
-            try (TestSession session = TestSession.createTestSession(tfMode)) {
-              Ops tf = session.getTF();
-              Shape shape = Shape.of(10, 10);
-              Identity<TInt32> instance = new Identity<>(tf, 2.);
-              instance.call(tf.constant(shape), TInt32.DTYPE);
-              fail("Should have thrown IllegalArgumentException on Integer type");
-            }
-          });
-  }
-
   /** Test of call method, of class Constant. */
   @Test
   public void testCallFloat() {
@@ -82,7 +65,7 @@ public void testCallFloat() {
         Ops tf = session.getTF();
         Shape shape = Shape.of(10, 10);
         Identity<TFloat32> instance = new Identity<>(tf, 2.);
-        Operand<TFloat32> operand = instance.call(tf.constant(shape), TFloat32.DTYPE);
+        Operand<TFloat32> operand = instance.call(tf.constant(shape), TFloat32.class);
         session.evaluate(expected, operand);
       }
   }
@@ -108,7 +91,7 @@ public void testCallDouble() {
         Ops tf = session.getTF();
         Shape shape = Shape.of(10, 10);
         Identity<TFloat64> instance = new Identity<>(tf, 2.);
-        Operand<TFloat64> operand = instance.call(tf.constant(shape), TFloat64.DTYPE);
+        Operand<TFloat64> operand = instance.call(tf.constant(shape), TFloat64.class);
         session.evaluate(expected, operand);
       }
   }
@@ -121,8 +104,8 @@ public void testReproducible() {
         Shape shape = Shape.of(2, 2);
 
         Identity<TFloat64> instance = new Identity<>(tf, 2.);
-        Operand<TFloat64> operand1 = instance.call(tf.constant(shape), TFloat64.DTYPE);
-        Operand<TFloat64> operand2 = instance.call(tf.constant(shape), TFloat64.DTYPE);
+        Operand<TFloat64> operand1 = instance.call(tf.constant(shape), TFloat64.class);
+        Operand<TFloat64> operand2 = instance.call(tf.constant(shape), TFloat64.class);
         session.evaluate(operand1, operand2);
       }
   }
diff --git a/tensorflow-framework/src/test/java/org/tensorflow/framework/initializers/LeCunTest.java b/tensorflow-framework/src/test/java/org/tensorflow/framework/initializers/LeCunTest.java
index 6033f9e12a5..336850a5549 100644
--- a/tensorflow-framework/src/test/java/org/tensorflow/framework/initializers/LeCunTest.java
+++ b/tensorflow-framework/src/test/java/org/tensorflow/framework/initializers/LeCunTest.java
@@ -51,8 +51,8 @@ public void testCallNormalFloat() {
       try (TestSession session = TestSession.createTestSession(tfMode)) {
         Ops tf = session.getTF();
         Shape shape = Shape.of(2, 2);
-        LeCun<TFloat32, TFloat32> instance = new LeCun<>(tf, Distribution.TRUNCATED_NORMAL, SEED);
-        Operand<TFloat32> operand = instance.call(tf.constant(shape), TFloat32.DTYPE);
+        LeCun<TFloat32> instance = new LeCun<>(tf, Distribution.TRUNCATED_NORMAL, SEED);
+        Operand<TFloat32> operand = instance.call(tf.constant(shape), TFloat32.class);
         session.evaluate(expected, operand);
       }
   }
@@ -66,8 +66,8 @@ public void testCallNormalDouble() {
       try (TestSession session = TestSession.createTestSession(tfMode)) {
         Ops tf = session.getTF();
         Shape shape = Shape.of(2, 2);
-        LeCun<TFloat64, TFloat64> instance = new LeCun<>(tf, Distribution.TRUNCATED_NORMAL, SEED);
-        Operand<TFloat64> operand = instance.call(tf.constant(shape), TFloat64.DTYPE);
+        LeCun<TFloat64> instance = new LeCun<>(tf, Distribution.TRUNCATED_NORMAL, SEED);
+        Operand<TFloat64> operand = instance.call(tf.constant(shape), TFloat64.class);
         session.evaluate(expected, operand);
       }
   }
@@ -80,8 +80,8 @@ public void testCallUniformFloat() {
       try (TestSession session = TestSession.createTestSession(tfMode)) {
         Ops tf = session.getTF();
         Shape shape = Shape.of(2, 2);
-        LeCun<TFloat32, TFloat32> instance = new LeCun<>(tf, Distribution.UNIFORM, SEED);
-        Operand<TFloat32> operand = instance.call(tf.constant(shape), TFloat32.DTYPE);
+        LeCun<TFloat32> instance = new LeCun<>(tf, Distribution.UNIFORM, SEED);
+        Operand<TFloat32> operand = instance.call(tf.constant(shape), TFloat32.class);
         session.evaluate(expected, operand);
       }
   }
@@ -95,8 +95,8 @@ public void testCallUniformDouble() {
       try (TestSession session = TestSession.createTestSession(tfMode)) {
         Ops tf = session.getTF();
         Shape shape = Shape.of(2, 2);
-        LeCun<TFloat64, TFloat64> instance = new LeCun<>(tf, Distribution.UNIFORM, SEED);
-        Operand<TFloat64> operand = instance.call(tf.constant(shape), TFloat64.DTYPE);
+        LeCun<TFloat64> instance = new LeCun<>(tf, Distribution.UNIFORM, SEED);
+        Operand<TFloat64> operand = instance.call(tf.constant(shape), TFloat64.class);
         session.evaluate(expected, operand);
       }
   }
@@ -107,9 +107,9 @@ public void testCallNormalReproducible() {
       try (TestSession session = TestSession.createTestSession(tfMode)) {
         Ops tf = session.getTF();
         Shape shape = Shape.of(2, 2);
-        LeCun<TFloat64, TFloat64> instance = new LeCun<>(tf, Distribution.TRUNCATED_NORMAL, SEED);
-        Operand<TFloat64> operand1 = instance.call(tf.constant(shape), TFloat64.DTYPE);
-        Operand<TFloat64> operand2 = instance.call(tf.constant(shape), TFloat64.DTYPE);
+        LeCun<TFloat64> instance = new LeCun<>(tf, Distribution.TRUNCATED_NORMAL, SEED);
+        Operand<TFloat64> operand1 = instance.call(tf.constant(shape), TFloat64.class);
+        Operand<TFloat64> operand2 = instance.call(tf.constant(shape), TFloat64.class);
         session.evaluate(operand1, operand2);
       }
   }
@@ -120,9 +120,9 @@ public void testCallUniformReproducible() {
       try (TestSession session = TestSession.createTestSession(tfMode)) {
         Ops tf = session.getTF();
         Shape shape = Shape.of(2, 2);
-        LeCun<TFloat64, TFloat64> instance = new LeCun<>(tf, Distribution.UNIFORM, SEED);
-        Operand<TFloat64> operand1 = instance.call(tf.constant(shape), TFloat64.DTYPE);
-        Operand<TFloat64> operand2 = instance.call(tf.constant(shape), TFloat64.DTYPE);
+        LeCun<TFloat64> instance = new LeCun<>(tf, Distribution.UNIFORM, SEED);
+        Operand<TFloat64> operand1 = instance.call(tf.constant(shape), TFloat64.class);
+        Operand<TFloat64> operand2 = instance.call(tf.constant(shape), TFloat64.class);
         session.evaluate(operand1, operand2);
       }
   }
@@ -133,9 +133,9 @@ public void testCallNORMALReproducible() {
       try (TestSession session = TestSession.createTestSession(tfMode)) {
         Ops tf = session.getTF();
         Shape shape = Shape.of(2, 2);
-        LeCun<TFloat64, TFloat64> instance = new LeCun<>(tf, Distribution.NORMAL, SEED);
-        Operand<TFloat64> operand1 = instance.call(tf.constant(shape), TFloat64.DTYPE);
-        Operand<TFloat64> operand2 = instance.call(tf.constant(shape), TFloat64.DTYPE);
+        LeCun<TFloat64> instance = new LeCun<>(tf, Distribution.NORMAL, SEED);
+        Operand<TFloat64> operand1 = instance.call(tf.constant(shape), TFloat64.class);
+        Operand<TFloat64> operand2 = instance.call(tf.constant(shape), TFloat64.class);
         session.evaluate(operand1, operand2);
       }
   }
diff --git a/tensorflow-framework/src/test/java/org/tensorflow/framework/initializers/OnesTest.java b/tensorflow-framework/src/test/java/org/tensorflow/framework/initializers/OnesTest.java
index bbd2ba3d384..053ba5dd7ff 100644
--- a/tensorflow-framework/src/test/java/org/tensorflow/framework/initializers/OnesTest.java
+++ b/tensorflow-framework/src/test/java/org/tensorflow/framework/initializers/OnesTest.java
@@ -52,7 +52,7 @@ public void testCallUInt() {
         Ops tf = session.getTF();
         Shape shape = Shape.of(2, 2);
         Ones<TUint8> instance = new Ones<>(tf);
-        Operand<TUint8> operand = instance.call(tf.constant(shape), TUint8.DTYPE);
+        Operand<TUint8> operand = instance.call(tf.constant(shape), TUint8.class);
         session.evaluate(expected, operand);
       }
   }
@@ -66,7 +66,7 @@ public void testCallInt() {
         Ops tf = session.getTF();
         Shape shape = Shape.of(2, 2);
         Ones<TInt32> instance = new Ones<>(tf);
-        Operand<TInt32> operand = instance.call(tf.constant(shape), TInt32.DTYPE);
+        Operand<TInt32> operand = instance.call(tf.constant(shape), TInt32.class);
         session.evaluate(expected, operand);
       }
   }
@@ -80,7 +80,7 @@ public void testCallLong() {
         Ops tf = session.getTF();
         Shape shape = Shape.of(2, 2);
         Ones<TInt64> instance = new Ones<>(tf);
-        Operand<TInt64> operand = instance.call(tf.constant(shape), TInt64.DTYPE);
+        Operand<TInt64> operand = instance.call(tf.constant(shape), TInt64.class);
         session.evaluate(expected, operand);
       }
   }
@@ -94,7 +94,7 @@ public void testCallFloat() {
         Ops tf = session.getTF();
         Shape shape = Shape.of(2, 2);
         Ones<TFloat32> instance = new Ones<>(tf);
-        Operand<TFloat32> operand = instance.call(tf.constant(shape), TFloat32.DTYPE);
+        Operand<TFloat32> operand = instance.call(tf.constant(shape), TFloat32.class);
         session.evaluate(expected, operand);
       }
   }
@@ -109,7 +109,7 @@ public void testCallDouble() {
         Shape shape = Shape.of(2, 2);
 
         Ones<TFloat64> instance = new Ones<>(tf);
-        Operand<TFloat64> operand = instance.call(tf.constant(shape), TFloat64.DTYPE);
+        Operand<TFloat64> operand = instance.call(tf.constant(shape), TFloat64.class);
         session.evaluate(expected, operand);
       }
   }
@@ -126,7 +126,7 @@ public void testCallString() {
               Shape shape = Shape.of(2, 2);
 
               Ones<TString> instance = new Ones<>(tf);
-              instance.call(tf.constant(shape), TString.DTYPE);
+              instance.call(tf.constant(shape), TString.class);
               fail("IllegalArgumentException should have been thrown for TString");
             }
           });
@@ -141,7 +141,7 @@ public void testCallBool() {
         Shape shape = Shape.of(2, 2);
 
         Ones<TBool> instance = new Ones<>(tf);
-        Operand<TBool> operand = instance.call(tf.constant(shape), TBool.DTYPE);
+        Operand<TBool> operand = instance.call(tf.constant(shape), TBool.class);
         session.evaluate(expected, operand);
       }
   }
@@ -154,8 +154,8 @@ public void testReproducible() {
         Shape shape = Shape.of(2, 2);
 
         Ones<TFloat64> instance = new Ones<>(tf);
-        Operand<TFloat64> operand1 = instance.call(tf.constant(shape), TFloat64.DTYPE);
-        Operand<TFloat64> operand2 = instance.call(tf.constant(shape), TFloat64.DTYPE);
+        Operand<TFloat64> operand1 = instance.call(tf.constant(shape), TFloat64.class);
+        Operand<TFloat64> operand2 = instance.call(tf.constant(shape), TFloat64.class);
         session.evaluate(operand1, operand2);
       }
   }
diff --git a/tensorflow-framework/src/test/java/org/tensorflow/framework/initializers/OrthogonalTest.java b/tensorflow-framework/src/test/java/org/tensorflow/framework/initializers/OrthogonalTest.java
index a4fff5fd19c..22b89d9177c 100644
--- a/tensorflow-framework/src/test/java/org/tensorflow/framework/initializers/OrthogonalTest.java
+++ b/tensorflow-framework/src/test/java/org/tensorflow/framework/initializers/OrthogonalTest.java
@@ -47,23 +47,6 @@ public void setUp() {}
   @AfterEach
   public void tearDown() {}
 
-  /** Test of call method, of class Orthogonal. */
-  @Test
-  public void testCallInt() {
-    for (TestSession.Mode tfMode : tfModes)
-      assertThrows(
-          java.lang.IllegalArgumentException.class,
-          () -> {
-            try (TestSession session = TestSession.createTestSession(tfMode)) {
-              Ops tf = session.getTF();
-              Shape shape = Shape.of(10, 10);
-              Orthogonal<TInt32, TInt32> instance = new Orthogonal<>(tf, GAIN_VALUE, SEED);
-              instance.call(tf.constant(shape), TInt32.DTYPE);
-              fail("Should have thrown IllegalArgumentException on Integer type");
-            }
-          });
-  }
-
   /** Test of call method, of class Orthogonal. */
   @Test
   public void testCallFloat() {
@@ -173,8 +156,8 @@ public void testCallFloat() {
       try (TestSession session = TestSession.createTestSession(tfMode)) {
         Ops tf = session.getTF();
         Shape shape = Shape.of(10, 10);
-        Orthogonal<TFloat32, TFloat32> instance = new Orthogonal<>(tf, GAIN_VALUE, SEED);
-        Operand<TFloat32> operand = instance.call(tf.constant(shape), TFloat32.DTYPE);
+        Orthogonal<TFloat32> instance = new Orthogonal<>(tf, GAIN_VALUE, SEED);
+        Operand<TFloat32> operand = instance.call(tf.constant(shape), TFloat32.class);
         session.evaluate(expected, operand);
       }
   }
@@ -288,8 +271,8 @@ public void testCallDouble() {
       try (TestSession session = TestSession.createTestSession(tfMode)) {
         Ops tf = session.getTF();
         Shape shape = Shape.of(10, 10);
-        Orthogonal<TFloat64, TFloat64> instance = new Orthogonal<>(tf, GAIN_VALUE, SEED);
-        Operand<TFloat64> operand = instance.call(tf.constant(shape), TFloat64.DTYPE);
+        Orthogonal<TFloat64> instance = new Orthogonal<>(tf, GAIN_VALUE, SEED);
+        Operand<TFloat64> operand = instance.call(tf.constant(shape), TFloat64.class);
         session.evaluate(expected, operand);
       }
   }
@@ -301,9 +284,9 @@ public void testReproducible() {
         Ops tf = session.getTF();
         Shape shape = Shape.of(2, 2);
 
-        Orthogonal<TFloat64, TFloat64> instance = new Orthogonal<>(tf, GAIN_VALUE, SEED);
-        Operand<TFloat64> operand1 = instance.call(tf.constant(shape), TFloat64.DTYPE);
-        Operand<TFloat64> operand2 = instance.call(tf.constant(shape), TFloat64.DTYPE);
+        Orthogonal<TFloat64> instance = new Orthogonal<>(tf, GAIN_VALUE, SEED);
+        Operand<TFloat64> operand1 = instance.call(tf.constant(shape), TFloat64.class);
+        Operand<TFloat64> operand2 = instance.call(tf.constant(shape), TFloat64.class);
         session.evaluate(operand1, operand2);
       }
   }
diff --git a/tensorflow-framework/src/test/java/org/tensorflow/framework/initializers/RandomNormalTest.java b/tensorflow-framework/src/test/java/org/tensorflow/framework/initializers/RandomNormalTest.java
index 50aec670503..3b2b3bdb243 100644
--- a/tensorflow-framework/src/test/java/org/tensorflow/framework/initializers/RandomNormalTest.java
+++ b/tensorflow-framework/src/test/java/org/tensorflow/framework/initializers/RandomNormalTest.java
@@ -52,9 +52,9 @@ public void testCalltestSoftmaxFloat() {
       try (TestSession session = TestSession.createTestSession(tfMode)) {
         Ops tf = session.getTF();
         Shape shape = Shape.of(2, 2);
-        RandomNormal<TFloat32, TFloat32> instance =
+        RandomNormal<TFloat32> instance =
             new RandomNormal<>(tf, MEAN_VALUE, STDDEV_VALUE, SEED);
-        Operand<TFloat32> operand = instance.call(tf.constant(shape), TFloat32.DTYPE);
+        Operand<TFloat32> operand = instance.call(tf.constant(shape), TFloat32.class);
         session.evaluate(expected, operand);
       }
   }
@@ -68,9 +68,9 @@ public void testCalltestSoftmaxDouble() {
       try (TestSession session = TestSession.createTestSession(tfMode)) {
         Ops tf = session.getTF();
         Shape shape = Shape.of(2, 2);
-        RandomNormal<TFloat64, TFloat64> instance =
+        RandomNormal<TFloat64> instance =
             new RandomNormal<>(tf, MEAN_VALUE, STDDEV_VALUE, SEED);
-        Operand<TFloat64> operand = instance.call(tf.constant(shape), TFloat64.DTYPE);
+        Operand<TFloat64> operand = instance.call(tf.constant(shape), TFloat64.class);
         session.evaluate(expected, operand);
       }
   }
@@ -82,10 +82,10 @@ public void testReproducible() {
         Ops tf = session.getTF();
         Shape shape = Shape.of(2, 2);
 
-        RandomNormal<TFloat64, TFloat64> instance =
+        RandomNormal<TFloat64> instance =
             new RandomNormal<>(tf, MEAN_VALUE, STDDEV_VALUE, SEED);
-        Operand<TFloat64> operand1 = instance.call(tf.constant(shape), TFloat64.DTYPE);
-        Operand<TFloat64> operand2 = instance.call(tf.constant(shape), TFloat64.DTYPE);
+        Operand<TFloat64> operand1 = instance.call(tf.constant(shape), TFloat64.class);
+        Operand<TFloat64> operand2 = instance.call(tf.constant(shape), TFloat64.class);
         session.evaluate(operand1, operand2);
       }
   }
diff --git a/tensorflow-framework/src/test/java/org/tensorflow/framework/initializers/RandomUniformTest.java b/tensorflow-framework/src/test/java/org/tensorflow/framework/initializers/RandomUniformTest.java
index d3f9af74209..23e26083a9b 100644
--- a/tensorflow-framework/src/test/java/org/tensorflow/framework/initializers/RandomUniformTest.java
+++ b/tensorflow-framework/src/test/java/org/tensorflow/framework/initializers/RandomUniformTest.java
@@ -53,9 +53,9 @@ public void testCallInt() {
       try (TestSession session = TestSession.createTestSession(tfMode)) {
         Ops tf = session.getTF();
         Shape shape = Shape.of(2, 2);
-        RandomUniform<TInt32, TFloat64> instance =
+        RandomUniform<TInt32> instance =
             new RandomUniform<>(tf, MIN_VALUE, MAX_VALUE, SEED);
-        Operand<TInt32> operand = instance.call(tf.constant(shape), TInt32.DTYPE);
+        Operand<TInt32> operand = instance.call(tf.constant(shape), TInt32.class);
         session.evaluate(expected, operand);
       }
   }
@@ -68,9 +68,9 @@ public void testCallFloat() {
       try (TestSession session = TestSession.createTestSession(tfMode)) {
         Ops tf = session.getTF();
         Shape shape = Shape.of(2, 2);
-        RandomUniform<TFloat32, TFloat32> instance =
+        RandomUniform<TFloat32> instance =
             new RandomUniform<>(tf, MIN_VALUE, MAX_VALUE, SEED);
-        Operand<TFloat32> operand = instance.call(tf.constant(shape), TFloat32.DTYPE);
+        Operand<TFloat32> operand = instance.call(tf.constant(shape), TFloat32.class);
         session.evaluate(expected, operand);
       }
   }
@@ -84,9 +84,9 @@ public void testCallDouble() {
       try (TestSession session = TestSession.createTestSession(tfMode)) {
         Ops tf = session.getTF();
         Shape shape = Shape.of(2, 2);
-        RandomUniform<TFloat64, TFloat64> instance =
+        RandomUniform<TFloat64> instance =
             new RandomUniform<>(tf, MIN_VALUE, MAX_VALUE, SEED);
-        Operand<TFloat64> operand = instance.call(tf.constant(shape), TFloat64.DTYPE);
+        Operand<TFloat64> operand = instance.call(tf.constant(shape), TFloat64.class);
         session.evaluate(expected, operand);
       }
   }
@@ -98,10 +98,10 @@ public void testReproducible() {
         Ops tf = session.getTF();
         Shape shape = Shape.of(2, 2);
 
-        RandomUniform<TFloat64, TFloat64> instance =
+        RandomUniform<TFloat64> instance =
             new RandomUniform<>(tf, MIN_VALUE, MAX_VALUE, SEED);
-        Operand<TFloat64> operand1 = instance.call(tf.constant(shape), TFloat64.DTYPE);
-        Operand<TFloat64> operand2 = instance.call(tf.constant(shape), TFloat64.DTYPE);
+        Operand<TFloat64> operand1 = instance.call(tf.constant(shape), TFloat64.class);
+        Operand<TFloat64> operand2 = instance.call(tf.constant(shape), TFloat64.class);
         session.evaluate(operand1, operand2);
       }
   }
diff --git a/tensorflow-framework/src/test/java/org/tensorflow/framework/initializers/TruncatedNormalTest.java b/tensorflow-framework/src/test/java/org/tensorflow/framework/initializers/TruncatedNormalTest.java
index 0a551df2f38..96bf915e199 100644
--- a/tensorflow-framework/src/test/java/org/tensorflow/framework/initializers/TruncatedNormalTest.java
+++ b/tensorflow-framework/src/test/java/org/tensorflow/framework/initializers/TruncatedNormalTest.java
@@ -52,9 +52,9 @@ public void testCallFloat() {
       try (TestSession session = TestSession.createTestSession(tfMode)) {
         Ops tf = session.getTF();
         Shape shape = Shape.of(2, 2);
-        TruncatedNormal<TFloat32, TFloat32> instance =
+        TruncatedNormal<TFloat32> instance =
             new TruncatedNormal<>(tf, MEAN_VALUE, STDDEV_VALUE, SEED);
-        Operand<TFloat32> operand = instance.call(tf.constant(shape), TFloat32.DTYPE);
+        Operand<TFloat32> operand = instance.call(tf.constant(shape), TFloat32.class);
         session.evaluate(expected, operand);
       }
   }
@@ -68,9 +68,9 @@ public void testCallDouble() {
       try (TestSession session = TestSession.createTestSession(tfMode)) {
         Ops tf = session.getTF();
         Shape shape = Shape.of(2, 2);
-        TruncatedNormal<TFloat64, TFloat64> instance =
+        TruncatedNormal<TFloat64> instance =
             new TruncatedNormal<>(tf, MEAN_VALUE, STDDEV_VALUE, SEED);
-        Operand<TFloat64> operand = instance.call(tf.constant(shape), TFloat64.DTYPE);
+        Operand<TFloat64> operand = instance.call(tf.constant(shape), TFloat64.class);
         session.evaluate(expected, operand);
       }
   }
@@ -82,10 +82,10 @@ public void testReproducible() {
         Ops tf = session.getTF();
         Shape shape = Shape.of(2, 2);
 
-        TruncatedNormal<TFloat64, TFloat64> instance =
+        TruncatedNormal<TFloat64> instance =
             new TruncatedNormal<>(tf, MEAN_VALUE, STDDEV_VALUE, SEED);
-        Operand<TFloat64> operand1 = instance.call(tf.constant(shape), TFloat64.DTYPE);
-        Operand<TFloat64> operand2 = instance.call(tf.constant(shape), TFloat64.DTYPE);
+        Operand<TFloat64> operand1 = instance.call(tf.constant(shape), TFloat64.class);
+        Operand<TFloat64> operand2 = instance.call(tf.constant(shape), TFloat64.class);
         session.evaluate(operand1, operand2);
       }
   }
diff --git a/tensorflow-framework/src/test/java/org/tensorflow/framework/initializers/VarianceScalingTest.java b/tensorflow-framework/src/test/java/org/tensorflow/framework/initializers/VarianceScalingTest.java
index 77e0dd7afc7..159affb07e2 100644
--- a/tensorflow-framework/src/test/java/org/tensorflow/framework/initializers/VarianceScalingTest.java
+++ b/tensorflow-framework/src/test/java/org/tensorflow/framework/initializers/VarianceScalingTest.java
@@ -50,14 +50,14 @@ public void testCallFloat1FanInTruncatedNormal() {
       try (TestSession session = TestSession.createTestSession(tfMode)) {
         Ops tf = session.getTF();
         Shape shape = Shape.of(2, 2);
-        VarianceScaling<TFloat32, TFloat32> instance =
+        VarianceScaling<TFloat32> instance =
             new VarianceScaling<>(
                 tf,
                 1.0,
                 VarianceScaling.Mode.FAN_IN,
                 VarianceScaling.Distribution.TRUNCATED_NORMAL,
                 SEED);
-        Operand<TFloat32> operand = instance.call(tf.constant(shape), TFloat32.DTYPE);
+        Operand<TFloat32> operand = instance.call(tf.constant(shape), TFloat32.class);
         session.evaluate(expected, operand);
       }
   }
@@ -71,14 +71,14 @@ public void testCallDouble1FanInTruncatedNormal() {
       try (TestSession session = TestSession.createTestSession(tfMode)) {
         Ops tf = session.getTF();
         Shape shape = Shape.of(2, 2);
-        VarianceScaling<TFloat64, TFloat64> instance =
+        VarianceScaling<TFloat64> instance =
             new VarianceScaling<>(
                 tf,
                 1.0,
                 VarianceScaling.Mode.FAN_IN,
                 VarianceScaling.Distribution.TRUNCATED_NORMAL,
                 SEED);
-        Operand<TFloat64> operand = instance.call(tf.constant(shape), TFloat64.DTYPE);
+        Operand<TFloat64> operand = instance.call(tf.constant(shape), TFloat64.class);
         session.evaluate(expected, operand);
       }
   }
@@ -91,14 +91,14 @@ public void testCallFloat1FanInNormal() {
       try (TestSession session = TestSession.createTestSession(tfMode)) {
         Ops tf = session.getTF();
         Shape shape = Shape.of(2, 2);
-        VarianceScaling<TFloat32, TFloat32> instance =
+        VarianceScaling<TFloat32> instance =
             new VarianceScaling<>(
                 tf,
                 1.0,
                 VarianceScaling.Mode.FAN_IN,
                 VarianceScaling.Distribution.NORMAL,
                 SEED);
-        Operand<TFloat32> operand = instance.call(tf.constant(shape), TFloat32.DTYPE);
+        Operand<TFloat32> operand = instance.call(tf.constant(shape), TFloat32.class);
         session.evaluate(expected, operand);
       }
   }
@@ -112,14 +112,14 @@ public void testCalltestSoftmaxDouble1FanInNormal() {
       try (TestSession session = TestSession.createTestSession(tfMode)) {
         Ops tf = session.getTF();
         Shape shape = Shape.of(2, 2);
-        VarianceScaling<TFloat64, TFloat64> instance =
+        VarianceScaling<TFloat64> instance =
             new VarianceScaling<>(
                 tf,
                 1.0,
                 VarianceScaling.Mode.FAN_IN,
                 VarianceScaling.Distribution.NORMAL,
                 SEED);
-        Operand<TFloat64> operand = instance.call(tf.constant(shape), TFloat64.DTYPE);
+        Operand<TFloat64> operand = instance.call(tf.constant(shape), TFloat64.class);
         session.evaluate(expected, operand);
       }
   }
@@ -132,10 +132,10 @@ public void testCalltestSoftmaxFloat1FanInUNIFORM() {
       try (TestSession session = TestSession.createTestSession(tfMode)) {
         Ops tf = session.getTF();
         Shape shape = Shape.of(2, 2);
-        VarianceScaling<TFloat32, TFloat32> instance =
+        VarianceScaling<TFloat32> instance =
             new VarianceScaling<>(
                 tf, 1.0, VarianceScaling.Mode.FAN_IN, VarianceScaling.Distribution.UNIFORM, SEED);
-        Operand<TFloat32> operand = instance.call(tf.constant(shape), TFloat32.DTYPE);
+        Operand<TFloat32> operand = instance.call(tf.constant(shape), TFloat32.class);
         session.evaluate(expected, operand);
       }
   }
@@ -149,10 +149,10 @@ public void testCalltestSoftmaxDouble1FanInUNIFORM() {
       try (TestSession session = TestSession.createTestSession(tfMode)) {
         Ops tf = session.getTF();
         Shape shape = Shape.of(2, 2);
-        VarianceScaling<TFloat64, TFloat64> instance =
+        VarianceScaling<TFloat64> instance =
             new VarianceScaling<>(
                 tf, 1.0, VarianceScaling.Mode.FAN_IN, VarianceScaling.Distribution.UNIFORM, SEED);
-        Operand<TFloat64> operand = instance.call(tf.constant(shape), TFloat64.DTYPE);
+        Operand<TFloat64> operand = instance.call(tf.constant(shape), TFloat64.class);
         session.evaluate(expected, operand);
       }
   }
@@ -164,11 +164,11 @@ public void testReproducible1() {
         Ops tf = session.getTF();
         Shape shape = Shape.of(2, 2);
 
-        VarianceScaling<TFloat64, TFloat64> instance =
+        VarianceScaling<TFloat64> instance =
             new VarianceScaling<>(
                 tf, 1.0, VarianceScaling.Mode.FAN_IN, VarianceScaling.Distribution.UNIFORM, SEED);
-        Operand<TFloat64> operand1 = instance.call(tf.constant(shape), TFloat64.DTYPE);
-        Operand<TFloat64> operand2 = instance.call(tf.constant(shape), TFloat64.DTYPE);
+        Operand<TFloat64> operand1 = instance.call(tf.constant(shape), TFloat64.class);
+        Operand<TFloat64> operand2 = instance.call(tf.constant(shape), TFloat64.class);
         session.evaluate(operand1, operand2);
       }
   }
@@ -180,15 +180,15 @@ public void testReproducible2() {
         Ops tf = session.getTF();
         Shape shape = Shape.of(2, 2);
 
-        VarianceScaling<TFloat64, TFloat64> instance =
+        VarianceScaling<TFloat64> instance =
             new VarianceScaling<>(
                 tf,
                 1.0,
                 VarianceScaling.Mode.FAN_IN,
                 VarianceScaling.Distribution.NORMAL,
                 SEED);
-        Operand<TFloat64> operand1 = instance.call(tf.constant(shape), TFloat64.DTYPE);
-        Operand<TFloat64> operand2 = instance.call(tf.constant(shape), TFloat64.DTYPE);
+        Operand<TFloat64> operand1 = instance.call(tf.constant(shape), TFloat64.class);
+        Operand<TFloat64> operand2 = instance.call(tf.constant(shape), TFloat64.class);
         session.evaluate(operand1, operand2);
       }
   }
@@ -200,15 +200,15 @@ public void testReproducible3() {
         Ops tf = session.getTF();
         Shape shape = Shape.of(2, 2);
 
-        VarianceScaling<TFloat64, TFloat64> instance =
+        VarianceScaling<TFloat64> instance =
             new VarianceScaling<>(
                 tf,
                 1.0,
                 VarianceScaling.Mode.FAN_OUT,
                 VarianceScaling.Distribution.TRUNCATED_NORMAL,
                 SEED);
-        Operand<TFloat64> operand1 = instance.call(tf.constant(shape), TFloat64.DTYPE);
-        Operand<TFloat64> operand2 = instance.call(tf.constant(shape), TFloat64.DTYPE);
+        Operand<TFloat64> operand1 = instance.call(tf.constant(shape), TFloat64.class);
+        Operand<TFloat64> operand2 = instance.call(tf.constant(shape), TFloat64.class);
         session.evaluate(operand1, operand2);
       }
   }
@@ -220,11 +220,11 @@ public void testReproducible4() {
         Ops tf = session.getTF();
         Shape shape = Shape.of(2, 2);
 
-        VarianceScaling<TFloat64, TFloat64> instance =
+        VarianceScaling<TFloat64> instance =
             new VarianceScaling<>(
                 tf, 1.0, VarianceScaling.Mode.FAN_AVG, VarianceScaling.Distribution.UNIFORM, SEED);
-        Operand<TFloat64> operand1 = instance.call(tf.constant(shape), TFloat64.DTYPE);
-        Operand<TFloat64> operand2 = instance.call(tf.constant(shape), TFloat64.DTYPE);
+        Operand<TFloat64> operand1 = instance.call(tf.constant(shape), TFloat64.class);
+        Operand<TFloat64> operand2 = instance.call(tf.constant(shape), TFloat64.class);
         session.evaluate(operand1, operand2);
       }
   }
diff --git a/tensorflow-framework/src/test/java/org/tensorflow/framework/initializers/ZerosTest.java b/tensorflow-framework/src/test/java/org/tensorflow/framework/initializers/ZerosTest.java
index 975678add19..21bad6ff360 100644
--- a/tensorflow-framework/src/test/java/org/tensorflow/framework/initializers/ZerosTest.java
+++ b/tensorflow-framework/src/test/java/org/tensorflow/framework/initializers/ZerosTest.java
@@ -49,7 +49,7 @@ public void testCallUInt() {
         Ops tf = session.getTF();
         Shape shape = Shape.of(2, 2);
         Zeros<TUint8> instance = new Zeros<>(tf);
-        Operand<TUint8> operand = instance.call(tf.constant(shape), TUint8.DTYPE);
+        Operand<TUint8> operand = instance.call(tf.constant(shape), TUint8.class);
         session.evaluate(expected, operand);
       }
   }
@@ -63,7 +63,7 @@ public void testCallInt() {
         Ops tf = session.getTF();
         Shape shape = Shape.of(2, 2);
         Zeros<TInt32> instance = new Zeros<>(tf);
-        Operand<TInt32> operand = instance.call(tf.constant(shape), TInt32.DTYPE);
+        Operand<TInt32> operand = instance.call(tf.constant(shape), TInt32.class);
         session.evaluate(expected, operand);
       }
   }
@@ -77,7 +77,7 @@ public void testCallLong() {
         Ops tf = session.getTF();
         Shape shape = Shape.of(2, 2);
         Zeros<TInt64> instance = new Zeros<>(tf);
-        Operand<TInt64> operand = instance.call(tf.constant(shape), TInt64.DTYPE);
+        Operand<TInt64> operand = instance.call(tf.constant(shape), TInt64.class);
         session.evaluate(expected, operand);
       }
   }
@@ -91,7 +91,7 @@ public void testCallFloat() {
         Ops tf = session.getTF();
         Shape shape = Shape.of(2, 2);
         Zeros<TFloat32> instance = new Zeros<>(tf);
-        Operand<TFloat32> operand = instance.call(tf.constant(shape), TFloat32.DTYPE);
+        Operand<TFloat32> operand = instance.call(tf.constant(shape), TFloat32.class);
         session.evaluate(expected, operand);
       }
   }
@@ -106,7 +106,7 @@ public void testCallDouble() {
         Shape shape = Shape.of(2, 2);
 
         Zeros<TFloat64> instance = new Zeros<>(tf);
-        Operand<TFloat64> operand = instance.call(tf.constant(shape), TFloat64.DTYPE);
+        Operand<TFloat64> operand = instance.call(tf.constant(shape), TFloat64.class);
         session.evaluate(expected, operand);
       }
   }
@@ -120,7 +120,7 @@ public void testCallString() {
         Shape shape = Shape.of(2, 2);
 
         Zeros<TString> instance = new Zeros<>(tf);
-        Operand<TString> operand = instance.call(tf.constant(shape), TString.DTYPE);
+        Operand<TString> operand = instance.call(tf.constant(shape), TString.class);
         session.evaluateString(operand, String::isEmpty);
       }
   }
@@ -135,7 +135,7 @@ public void testCallBool() {
         Shape shape = Shape.of(2, 2);
 
         Zeros<TBool> instance = new Zeros<>(tf);
-        Operand<TBool> operand = instance.call(tf.constant(shape), TBool.DTYPE);
+        Operand<TBool> operand = instance.call(tf.constant(shape), TBool.class);
         session.evaluate(expected, operand);
       }
   }
@@ -148,8 +148,8 @@ public void testReproducible() {
         Shape shape = Shape.of(2, 2);
 
         Zeros<TFloat64> instance = new Zeros<>(tf);
-        Operand<TFloat64> operand1 = instance.call(tf.constant(shape), TFloat64.DTYPE);
-        Operand<TFloat64> operand2 = instance.call(tf.constant(shape), TFloat64.DTYPE);
+        Operand<TFloat64> operand1 = instance.call(tf.constant(shape), TFloat64.class);
+        Operand<TFloat64> operand2 = instance.call(tf.constant(shape), TFloat64.class);
         session.evaluate(operand1, operand2);
       }
   }
diff --git a/tensorflow-framework/src/test/java/org/tensorflow/framework/optimizers/AdaDeltaTest.java b/tensorflow-framework/src/test/java/org/tensorflow/framework/optimizers/AdaDeltaTest.java
index 5c4ce542c65..bf5e2b1450b 100644
--- a/tensorflow-framework/src/test/java/org/tensorflow/framework/optimizers/AdaDeltaTest.java
+++ b/tensorflow-framework/src/test/java/org/tensorflow/framework/optimizers/AdaDeltaTest.java
@@ -89,8 +89,8 @@ public void testBasic() {
           float[] var1Init = {3.0F, 4.0F};
           float[] fgrads = {grad, grad};
           Shape shape = Shape.of(var0Init.length);
-          Variable<TFloat32> var0 = tf.withName("var0").variable(shape, TFloat32.DTYPE);
-          Variable<TFloat32> var1 = tf.withName("var1").variable(shape, TFloat32.DTYPE);
+          Variable<TFloat32> var0 = tf.withName("var0").variable(shape, TFloat32.class);
+          Variable<TFloat32> var1 = tf.withName("var1").variable(shape, TFloat32.class);
 
           Assign<TFloat32> var0Initializer = tf.assign(var0, tf.constant(var0Init));
           Assign<TFloat32> var1Initializer = tf.assign(var1, tf.constant(var1Init));
diff --git a/tensorflow-framework/src/test/java/org/tensorflow/framework/optimizers/AdaGradDATest.java b/tensorflow-framework/src/test/java/org/tensorflow/framework/optimizers/AdaGradDATest.java
index ef9053ff1eb..d5b2657a4fc 100644
--- a/tensorflow-framework/src/test/java/org/tensorflow/framework/optimizers/AdaGradDATest.java
+++ b/tensorflow-framework/src/test/java/org/tensorflow/framework/optimizers/AdaGradDATest.java
@@ -64,8 +64,8 @@ public void testBasic() {
 
       Shape shape0 = Shape.of(var0Init.length);
       Shape shape1 = Shape.of(var1Init.length);
-      Variable<TFloat32> var0 = tf.withName("var0").variable(shape0, TFloat32.DTYPE);
-      Variable<TFloat32> var1 = tf.withName("var1").variable(shape1, TFloat32.DTYPE);
+      Variable<TFloat32> var0 = tf.withName("var0").variable(shape0, TFloat32.class);
+      Variable<TFloat32> var1 = tf.withName("var1").variable(shape1, TFloat32.class);
 
       Assign<TFloat32> var0Initializer = tf.assign(var0, tf.constant(var0Init));
       Assign<TFloat32> var1Initializer = tf.assign(var1, tf.constant(var1Init));
diff --git a/tensorflow-framework/src/test/java/org/tensorflow/framework/optimizers/AdaGradTest.java b/tensorflow-framework/src/test/java/org/tensorflow/framework/optimizers/AdaGradTest.java
index c5ae178b84c..d2eb63056e8 100644
--- a/tensorflow-framework/src/test/java/org/tensorflow/framework/optimizers/AdaGradTest.java
+++ b/tensorflow-framework/src/test/java/org/tensorflow/framework/optimizers/AdaGradTest.java
@@ -79,8 +79,8 @@ public void testBasic() {
 
       Shape shape0 = Shape.of(var0Init.length);
       Shape shape1 = Shape.of(var1Init.length);
-      Variable<TFloat32> var0 = tf.withName("var0").variable(shape0, TFloat32.DTYPE);
-      Variable<TFloat32> var1 = tf.withName("var1").variable(shape1, TFloat32.DTYPE);
+      Variable<TFloat32> var0 = tf.withName("var0").variable(shape0, TFloat32.class);
+      Variable<TFloat32> var1 = tf.withName("var1").variable(shape1, TFloat32.class);
 
       Assign<TFloat32> var0Initializer = tf.assign(var0, tf.constant(var0Init));
       Assign<TFloat32> var1Initializer = tf.assign(var1, tf.constant(var1Init));
diff --git a/tensorflow-framework/src/test/java/org/tensorflow/framework/optimizers/AdamTest.java b/tensorflow-framework/src/test/java/org/tensorflow/framework/optimizers/AdamTest.java
index a6cec420bd7..4f2afa07972 100644
--- a/tensorflow-framework/src/test/java/org/tensorflow/framework/optimizers/AdamTest.java
+++ b/tensorflow-framework/src/test/java/org/tensorflow/framework/optimizers/AdamTest.java
@@ -16,7 +16,6 @@
 
 import org.junit.jupiter.api.*;
 import org.tensorflow.Graph;
-import org.tensorflow.tensor.Tensor;
 import org.tensorflow.framework.utils.ND;
 import org.tensorflow.framework.utils.TestSession;
 import org.tensorflow.ndarray.FloatNdArray;
@@ -80,8 +79,8 @@ public void testBasic() {
       Ops tf = instance.getTF();
       Shape shape0 = Shape.of(var0Init.length);
       Shape shape1 = Shape.of(var1Init.length);
-      Variable<TFloat32> var0 = tf.withName("var0").variable(shape0, TFloat32.DTYPE);
-      Variable<TFloat32> var1 = tf.withName("var1").variable(shape1, TFloat32.DTYPE);
+      Variable<TFloat32> var0 = tf.withName("var0").variable(shape0, TFloat32.class);
+      Variable<TFloat32> var1 = tf.withName("var1").variable(shape1, TFloat32.class);
 
       Assign<TFloat32> var0Initializer = tf.assign(var0, tf.constant(var0Init));
       Assign<TFloat32> var1Initializer = tf.assign(var1, tf.constant(var1Init));
@@ -140,25 +139,23 @@ public void testBasic() {
           (float) Math.pow(beta1, step + 1), (float) Math.pow(beta2, step + 1)
         };
 
-        try (Tensor<TFloat32> result =
+        try (TFloat32 result =
             session
                 .getGraphSession()
                 .runner()
                 .fetch("beta1_power")
                 .run()
-                .get(0)
-                .expect(TFloat32.DTYPE)) {
-          result.data().scalars().forEach(f -> assertEquals(powers[0], f.getFloat(), epsilon1));
+                .get(0)) {
+          result.scalars().forEach(f -> assertEquals(powers[0], f.getFloat(), epsilon1));
         }
-        try (Tensor<TFloat32> result =
+        try (TFloat32 result =
             session
                 .getGraphSession()
                 .runner()
                 .fetch("beta2_power")
                 .run()
-                .get(0)
-                .expect(TFloat32.DTYPE)) {
-          result.data().scalars().forEach(f -> assertEquals(powers[1], f.getFloat(), epsilon1));
+                .get(0)) {
+          result.scalars().forEach(f -> assertEquals(powers[1], f.getFloat(), epsilon1));
         }
         session.run(update);
 
diff --git a/tensorflow-framework/src/test/java/org/tensorflow/framework/optimizers/AdamaxTest.java b/tensorflow-framework/src/test/java/org/tensorflow/framework/optimizers/AdamaxTest.java
index 541f82dedcc..ea0e96c7fad 100644
--- a/tensorflow-framework/src/test/java/org/tensorflow/framework/optimizers/AdamaxTest.java
+++ b/tensorflow-framework/src/test/java/org/tensorflow/framework/optimizers/AdamaxTest.java
@@ -16,7 +16,6 @@
 
 import org.junit.jupiter.api.*;
 import org.tensorflow.Graph;
-import org.tensorflow.tensor.Tensor;
 import org.tensorflow.framework.utils.ND;
 import org.tensorflow.framework.utils.TestSession;
 import org.tensorflow.ndarray.FloatNdArray;
@@ -101,8 +100,8 @@ public void testBasic() {
 
       Shape shape0 = Shape.of(var0Init.length);
       Shape shape1 = Shape.of(var1Init.length);
-      Variable<TFloat32> var0 = tf.withName("var0").variable(shape0, TFloat32.DTYPE);
-      Variable<TFloat32> var1 = tf.withName("var1").variable(shape1, TFloat32.DTYPE);
+      Variable<TFloat32> var0 = tf.withName("var0").variable(shape0, TFloat32.class);
+      Variable<TFloat32> var1 = tf.withName("var1").variable(shape1, TFloat32.class);
 
       Assign<TFloat32> var0Initializer = tf.assign(var0, tf.constant(var0Init));
       Assign<TFloat32> var1Initializer = tf.assign(var1, tf.constant(var1Init));
@@ -149,15 +148,14 @@ public void testBasic() {
         // Test powers
         final float beta1Power = (float) Math.pow(BETA_ONE_DEFAULT, step + 1);
 
-        try (Tensor<TFloat32> result =
+        try (TFloat32 result =
             session
                 .getGraphSession()
                 .runner()
                 .fetch("beta1_power")
                 .run()
-                .get(0)
-                .expect(TFloat32.DTYPE)) {
-          result.data().scalars().forEach(f -> assertEquals(beta1Power, f.getFloat(), epsilon1));
+                .get(0)) {
+          result.scalars().forEach(f -> assertEquals(beta1Power, f.getFloat(), epsilon1));
         }
         session.run(update);
 
diff --git a/tensorflow-framework/src/test/java/org/tensorflow/framework/optimizers/FtrlTest.java b/tensorflow-framework/src/test/java/org/tensorflow/framework/optimizers/FtrlTest.java
index 597f8e52bcd..7698d76f957 100644
--- a/tensorflow-framework/src/test/java/org/tensorflow/framework/optimizers/FtrlTest.java
+++ b/tensorflow-framework/src/test/java/org/tensorflow/framework/optimizers/FtrlTest.java
@@ -76,8 +76,8 @@ public void testFtrlWithL1L2L2Shrinkage() {
 
       Shape shape0 = Shape.of(var0Init.length);
       Shape shape1 = Shape.of(var1Init.length);
-      Variable<TFloat32> var0 = tf.withName("var0").variable(shape0, TFloat32.DTYPE);
-      Variable<TFloat32> var1 = tf.withName("var1").variable(shape1, TFloat32.DTYPE);
+      Variable<TFloat32> var0 = tf.withName("var0").variable(shape0, TFloat32.class);
+      Variable<TFloat32> var1 = tf.withName("var1").variable(shape1, TFloat32.class);
 
       Assign<TFloat32> var0Initializer = tf.assign(var0, tf.constant(var0Init));
       Assign<TFloat32> var1Initializer = tf.assign(var1, tf.constant(var1Init));
@@ -141,8 +141,8 @@ public void testFtrlWithL1() {
 
       Shape shape0 = Shape.of(var0Init.length);
       Shape shape1 = Shape.of(var1Init.length);
-      Variable<TFloat32> var0 = tf.withName("var0").variable(shape0, TFloat32.DTYPE);
-      Variable<TFloat32> var1 = tf.withName("var1").variable(shape1, TFloat32.DTYPE);
+      Variable<TFloat32> var0 = tf.withName("var0").variable(shape0, TFloat32.class);
+      Variable<TFloat32> var1 = tf.withName("var1").variable(shape1, TFloat32.class);
 
       Assign<TFloat32> var0Initializer = tf.assign(var0, tf.constant(var0Init));
       Assign<TFloat32> var1Initializer = tf.assign(var1, tf.constant(var1Init));
@@ -207,8 +207,8 @@ public void testFtrlWithL1L2() {
 
       Shape shape0 = Shape.of(var0Init.length);
       Shape shape1 = Shape.of(var1Init.length);
-      Variable<TFloat32> var0 = tf.withName("var0").variable(shape0, TFloat32.DTYPE);
-      Variable<TFloat32> var1 = tf.withName("var1").variable(shape1, TFloat32.DTYPE);
+      Variable<TFloat32> var0 = tf.withName("var0").variable(shape0, TFloat32.class);
+      Variable<TFloat32> var1 = tf.withName("var1").variable(shape1, TFloat32.class);
 
       Assign<TFloat32> var0Initializer = tf.assign(var0, tf.constant(var0Init));
       Assign<TFloat32> var1Initializer = tf.assign(var1, tf.constant(var1Init));
@@ -273,8 +273,8 @@ public void doTestFtrlwithoutRegularization() {
 
       Shape shape0 = Shape.of(var0Init.length);
       Shape shape1 = Shape.of(var1Init.length);
-      Variable<TFloat32> var0 = tf.withName("var0").variable(shape0, TFloat32.DTYPE);
-      Variable<TFloat32> var1 = tf.withName("var1").variable(shape1, TFloat32.DTYPE);
+      Variable<TFloat32> var0 = tf.withName("var0").variable(shape0, TFloat32.class);
+      Variable<TFloat32> var1 = tf.withName("var1").variable(shape1, TFloat32.class);
 
       Assign<TFloat32> var0Initializer = tf.assign(var0, tf.constant(var0Init));
       Assign<TFloat32> var1Initializer = tf.assign(var1, tf.constant(var1Init));
diff --git a/tensorflow-framework/src/test/java/org/tensorflow/framework/optimizers/GradientDescentTest.java b/tensorflow-framework/src/test/java/org/tensorflow/framework/optimizers/GradientDescentTest.java
index 4362c54d815..aefcc537979 100644
--- a/tensorflow-framework/src/test/java/org/tensorflow/framework/optimizers/GradientDescentTest.java
+++ b/tensorflow-framework/src/test/java/org/tensorflow/framework/optimizers/GradientDescentTest.java
@@ -61,8 +61,8 @@ public void testBasic() {
 
       Shape shape0 = Shape.of(var0Init.length);
       Shape shape1 = Shape.of(var1Init.length);
-      Variable<TFloat32> var0 = tf.withName("var0").variable(shape0, TFloat32.DTYPE);
-      Variable<TFloat32> var1 = tf.withName("var1").variable(shape1, TFloat32.DTYPE);
+      Variable<TFloat32> var0 = tf.withName("var0").variable(shape0, TFloat32.class);
+      Variable<TFloat32> var1 = tf.withName("var1").variable(shape1, TFloat32.class);
 
       Assign<TFloat32> var0Initializer = tf.assign(var0, tf.constant(var0Init));
       Assign<TFloat32> var1Initializer = tf.assign(var1, tf.constant(var1Init));
diff --git a/tensorflow-framework/src/test/java/org/tensorflow/framework/optimizers/MomentumTest.java b/tensorflow-framework/src/test/java/org/tensorflow/framework/optimizers/MomentumTest.java
index bcfff97773d..ca72c4f415c 100644
--- a/tensorflow-framework/src/test/java/org/tensorflow/framework/optimizers/MomentumTest.java
+++ b/tensorflow-framework/src/test/java/org/tensorflow/framework/optimizers/MomentumTest.java
@@ -77,8 +77,8 @@ public void testBasic() {
 
       Shape shape0 = Shape.of(var0Init.length);
       Shape shape1 = Shape.of(var1Init.length);
-      Variable<TFloat32> var0 = tf.withName("var0").variable(shape0, TFloat32.DTYPE);
-      Variable<TFloat32> var1 = tf.withName("var1").variable(shape1, TFloat32.DTYPE);
+      Variable<TFloat32> var0 = tf.withName("var0").variable(shape0, TFloat32.class);
+      Variable<TFloat32> var1 = tf.withName("var1").variable(shape1, TFloat32.class);
 
       Assign<TFloat32> var0Initializer = tf.assign(var0, tf.constant(var0Init));
       Assign<TFloat32> var1Initializer = tf.assign(var1, tf.constant(var1Init));
@@ -130,8 +130,8 @@ public void testMomentum() {
 
       Shape shape0 = Shape.of(var0Init.length);
       Shape shape1 = Shape.of(var1Init.length);
-      Variable<TFloat32> var0 = tf.withName("var0").variable(shape0, TFloat32.DTYPE);
-      Variable<TFloat32> var1 = tf.withName("var1").variable(shape1, TFloat32.DTYPE);
+      Variable<TFloat32> var0 = tf.withName("var0").variable(shape0, TFloat32.class);
+      Variable<TFloat32> var1 = tf.withName("var1").variable(shape1, TFloat32.class);
 
       Assign<TFloat32> var0Initializer = tf.assign(var0, tf.constant(var0Init));
       Assign<TFloat32> var1Initializer = tf.assign(var1, tf.constant(var1Init));
diff --git a/tensorflow-framework/src/test/java/org/tensorflow/framework/optimizers/NadamTest.java b/tensorflow-framework/src/test/java/org/tensorflow/framework/optimizers/NadamTest.java
index 88acd961d3a..4df0994dffc 100644
--- a/tensorflow-framework/src/test/java/org/tensorflow/framework/optimizers/NadamTest.java
+++ b/tensorflow-framework/src/test/java/org/tensorflow/framework/optimizers/NadamTest.java
@@ -16,7 +16,6 @@
 
 import org.junit.jupiter.api.*;
 import org.tensorflow.Graph;
-import org.tensorflow.tensor.Tensor;
 import org.tensorflow.framework.utils.ND;
 import org.tensorflow.framework.utils.TestSession;
 import org.tensorflow.ndarray.FloatNdArray;
@@ -102,8 +101,8 @@ public void testBasic() {
 
       Shape shape0 = Shape.of(var0Init.length);
       Shape shape1 = Shape.of(var1Init.length);
-      Variable<TFloat32> var0 = tf.withName("var0").variable(shape0, TFloat32.DTYPE);
-      Variable<TFloat32> var1 = tf.withName("var1").variable(shape1, TFloat32.DTYPE);
+      Variable<TFloat32> var0 = tf.withName("var0").variable(shape0, TFloat32.class);
+      Variable<TFloat32> var1 = tf.withName("var1").variable(shape1, TFloat32.class);
 
       Assign<TFloat32> var0Initializer = tf.assign(var0, tf.constant(var0Init));
       Assign<TFloat32> var1Initializer = tf.assign(var1, tf.constant(var1Init));
@@ -147,15 +146,14 @@ public void testBasic() {
       session.evaluate(var0Init, var0);
       session.evaluate(var1Init, var1);
 
-      try (Tensor<TFloat32> result =
+      try (TFloat32 result =
           session
               .getGraphSession()
               .runner()
               .fetch("momentum")
               .run()
-              .get(0)
-              .expect(TFloat32.DTYPE)) {
-        result.data().scalars().forEach(f -> assertEquals(1F, f.getFloat(), epsilon1));
+              .get(0)) {
+        result.scalars().forEach(f -> assertEquals(1F, f.getFloat(), epsilon1));
       }
       momentum = 1F;
 
@@ -167,15 +165,14 @@ public void testBasic() {
             Nadam.BETA_ONE_DEFAULT * (1F - 0.5F * (float) Math.pow(0.96F, (0.004F * (step + 1))));
         momentum = momentum * mut;
 
-        try (Tensor<TFloat32> result =
+        try (TFloat32 result =
             session
                 .getGraphSession()
                 .runner()
                 .fetch("momentum")
                 .run()
-                .get(0)
-                .expect(TFloat32.DTYPE)) {
-          result.data().scalars().forEach(f -> assertEquals(momentum, f.getFloat(), epsilon1));
+                .get(0)) {
+          result.scalars().forEach(f -> assertEquals(momentum, f.getFloat(), epsilon1));
         }
         mcache = ND.mul(mcache, momentum);
         FloatNdArray[] resultsNP = nadamUpdateNdArray(var0Np, grads0Np, step, m0, v0, mcache);
diff --git a/tensorflow-framework/src/test/java/org/tensorflow/framework/optimizers/RMSPropTest.java b/tensorflow-framework/src/test/java/org/tensorflow/framework/optimizers/RMSPropTest.java
index 202fb21ef68..3b002cd1dbe 100644
--- a/tensorflow-framework/src/test/java/org/tensorflow/framework/optimizers/RMSPropTest.java
+++ b/tensorflow-framework/src/test/java/org/tensorflow/framework/optimizers/RMSPropTest.java
@@ -87,8 +87,8 @@ public void testDense() {
 
         Shape shape0 = Shape.of(var0Init.length);
         Shape shape1 = Shape.of(var1Init.length);
-        Variable<TFloat32> var0 = tf.withName("var0").variable(shape0, TFloat32.DTYPE);
-        Variable<TFloat32> var1 = tf.withName("var1").variable(shape1, TFloat32.DTYPE);
+        Variable<TFloat32> var0 = tf.withName("var0").variable(shape0, TFloat32.class);
+        Variable<TFloat32> var1 = tf.withName("var1").variable(shape1, TFloat32.class);
 
         Assign<TFloat32> var0Initializer = tf.assign(var0, tf.constant(var0Init));
         Assign<TFloat32> var1Initializer = tf.assign(var1, tf.constant(var1Init));
diff --git a/tensorflow-framework/src/test/java/org/tensorflow/framework/utils/EagerTestSession.java b/tensorflow-framework/src/test/java/org/tensorflow/framework/utils/EagerTestSession.java
index bca90211e50..69fcee43ab3 100644
--- a/tensorflow-framework/src/test/java/org/tensorflow/framework/utils/EagerTestSession.java
+++ b/tensorflow-framework/src/test/java/org/tensorflow/framework/utils/EagerTestSession.java
@@ -84,57 +84,57 @@ public EagerSession getEagerSession() {
   /** {@inheritDoc} */
   @Override
   public <T extends TNumber> void evaluate(double expected, Operand<T> input) {
-    DataType<T> dtype = input.asOutput().dataType();
-    if (dtype == TFloat32.DTYPE) {
+    Class<T> type = input.asOutput().type();
+    if (type == TFloat32.class) {
       Operand<TFloat32> o = (Operand<TFloat32>) input;
       AtomicInteger index = new AtomicInteger();
       if (debug) {
-        o.data()
+        o.asTensor()
             .scalars()
             .forEach(f -> System.out.printf("%d). %f\n", index.getAndIncrement(), f.getFloat()));
       }
       index.set(0);
-      o.data().scalars().forEach(f -> assertEquals(expected, f.getFloat(), epsilon));
-    } else if (dtype == TFloat64.DTYPE) {
+      o.asTensor().scalars().forEach(f -> assertEquals(expected, f.getFloat(), epsilon));
+    } else if (type == TFloat64.class) {
       Operand<TFloat64> o = (Operand<TFloat64>) input;
       AtomicInteger index = new AtomicInteger();
       if (debug) {
-        o.data()
+        o.asTensor()
             .scalars()
             .forEach(f -> System.out.printf("%d). %f\n", index.getAndIncrement(), f.getDouble()));
       }
       index.set(0);
-      o.data().scalars().forEach(f -> assertEquals(expected, f.getDouble(), epsilon));
-    } else if (dtype == TInt32.DTYPE) {
+      o.asTensor().scalars().forEach(f -> assertEquals(expected, f.getDouble(), epsilon));
+    } else if (type == TInt32.class) {
       Operand<TInt32> o = (Operand<TInt32>) input;
       AtomicInteger index = new AtomicInteger();
       if (debug) {
-        o.data()
+        o.asTensor()
             .scalars()
             .forEach(f -> System.out.printf("%d). %d\n", index.getAndIncrement(), f.getInt()));
       }
       index.set(0);
-      o.data().scalars().forEach(f -> assertEquals((int) expected, f.getInt()));
-    } else if (dtype == TInt64.DTYPE) {
+      o.asTensor().scalars().forEach(f -> assertEquals((int) expected, f.getInt()));
+    } else if (type == TInt64.class) {
       Operand<TInt64> o = (Operand<TInt64>) input;
       AtomicInteger index = new AtomicInteger();
       if (debug) {
-        o.data()
+        o.asTensor()
             .scalars()
             .forEach(f -> System.out.printf("%d). %d\n", index.getAndIncrement(), f.getLong()));
       }
       index.set(0);
-      o.data().scalars().forEach(f -> assertEquals((long) expected, f.getLong()));
-    } else if (dtype == TUint8.DTYPE) {
+      o.asTensor().scalars().forEach(f -> assertEquals((long) expected, f.getLong()));
+    } else if (type == TUint8.class) {
       Operand<TUint8> o = (Operand<TUint8>) input;
       AtomicInteger index = new AtomicInteger();
       if (debug) {
-        o.data()
+        o.asTensor()
             .scalars()
             .forEach(f -> System.out.printf("%d). %x\n", index.getAndIncrement(), f.getByte()));
       }
       index.set(0);
-      o.data().scalars().forEach(f -> assertEquals((long) expected, f.getByte()));
+      o.asTensor().scalars().forEach(f -> assertEquals((long) expected, f.getByte()));
     }
   }
 
@@ -146,71 +146,71 @@ public <T extends TNumber> void evaluate(Number[] expected, Output<T> input) {
         expected.length,
         size,
         () -> String.format("expected length (%d) != to input length (%d)", expected.length, size));
-    DataType<T> dtype = input.dataType();
-    if (dtype == TFloat32.DTYPE) {
+    Class<T> type = input.type();
+    if (type == TFloat32.class) {
       Output<TFloat32> o = (Output<TFloat32>) input;
       AtomicInteger index = new AtomicInteger();
       if (debug) {
-        o.data()
+        o.asTensor()
             .scalars()
             .forEach(f -> System.out.printf("%d). %f\n", index.getAndIncrement(), f.getFloat()));
       }
       index.set(0);
-      o.data()
+      o.asTensor()
           .scalars()
           .forEach(
               f ->
                   assertEquals(
                       expected[index.getAndIncrement()].floatValue(), f.getFloat(), epsilon));
-    } else if (dtype == TFloat64.DTYPE) {
+    } else if (type == TFloat64.class) {
       Output<TFloat64> o = (Output<TFloat64>) input;
       AtomicInteger index = new AtomicInteger();
       if (debug) {
-        o.data()
+        o.asTensor()
             .scalars()
             .forEach(f -> System.out.printf("%d). %f\n", index.getAndIncrement(), f.getDouble()));
       }
       index.set(0);
-      o.data()
+      o.asTensor()
           .scalars()
           .forEach(
               f ->
                   assertEquals(
                       expected[index.getAndIncrement()].doubleValue(), f.getDouble(), epsilon));
-    } else if (dtype == TInt32.DTYPE) {
+    } else if (type == TInt32.class) {
       Output<TInt32> o = (Output<TInt32>) input;
       AtomicInteger index = new AtomicInteger();
       if (debug) {
-        o.data()
+        o.asTensor()
             .scalars()
             .forEach(f -> System.out.printf("%d). %d\n", index.getAndIncrement(), f.getInt()));
       }
       index.set(0);
-      o.data()
+      o.asTensor()
           .scalars()
           .forEach(f -> assertEquals(expected[index.getAndIncrement()].intValue(), f.getInt()));
-    } else if (dtype == TInt64.DTYPE) {
+    } else if (type == TInt64.class) {
       Output<TInt64> o = (Output<TInt64>) input;
       AtomicInteger index = new AtomicInteger();
       if (debug) {
-        o.data()
+        o.asTensor()
             .scalars()
             .forEach(f -> System.out.printf("%d). %d\n", index.getAndIncrement(), f.getLong()));
       }
       index.set(0);
-      o.data()
+      o.asTensor()
           .scalars()
           .forEach(f -> assertEquals(expected[index.getAndIncrement()].longValue(), f.getLong()));
-    } else if (dtype == TUint8.DTYPE) {
+    } else if (type == TUint8.class) {
       Output<TUint8> o = (Output<TUint8>) input;
       AtomicInteger index = new AtomicInteger();
       if (debug) {
-        o.data()
+        o.asTensor()
             .scalars()
             .forEach(f -> System.out.printf("%x). %d\n", index.getAndIncrement(), f.getByte()));
       }
       index.set(0);
-      o.data()
+      o.asTensor()
           .scalars()
           .forEach(f -> assertEquals(expected[index.getAndIncrement()].byteValue(), f.getByte()));
     }
@@ -219,69 +219,69 @@ public <T extends TNumber> void evaluate(Number[] expected, Output<T> input) {
   /** {@inheritDoc} */
   @Override
   public <T extends TType> void evaluate(FloatNdArray expected, Output<T> input) {
-    DataType<T> dtype = input.dataType();
-    if (dtype == TFloat32.DTYPE) {
+    Class<T> type = input.type();
+    if (type == TFloat32.class) {
       Output<TFloat32> o = (Output<TFloat32>) input;
       AtomicLong index = new AtomicLong();
       if (debug) {
-        o.data()
+        o.asTensor()
             .scalars()
             .forEach(f -> System.out.printf("%d). %f\n", index.getAndIncrement(), f.getFloat()));
       }
       index.set(0);
-      o.data()
+      o.asTensor()
           .scalars()
           .forEach(
               f -> assertEquals(expected.getFloat(index.getAndIncrement()), f.getFloat(), epsilon));
-    } else if (dtype == TFloat64.DTYPE) {
+    } else if (type == TFloat64.class) {
       Output<TFloat64> o = (Output<TFloat64>) input;
       AtomicInteger index = new AtomicInteger();
       if (debug) {
-        o.data()
+        o.asTensor()
             .scalars()
             .forEach(f -> System.out.printf("%d). %f\n", index.getAndIncrement(), f.getDouble()));
       }
       index.set(0);
-      o.data()
+      o.asTensor()
           .scalars()
           .forEach(
               f ->
                   assertEquals(expected.getFloat(index.getAndIncrement()), f.getDouble(), epsilon));
-    } else if (dtype == TInt32.DTYPE) {
+    } else if (type == TInt32.class) {
       Output<TInt32> o = (Output<TInt32>) input;
       AtomicInteger index = new AtomicInteger();
       if (debug) {
-        o.data()
+        o.asTensor()
             .scalars()
             .forEach(f -> System.out.printf("%d). %d\n", index.getAndIncrement(), f.getInt()));
       }
       index.set(0);
-      for (IntNdArray f : o.data().scalars()) {
+      for (IntNdArray f : o.asTensor().scalars()) {
         assertEquals((int) expected.getFloat(index.getAndIncrement()), f.getInt());
       }
-    } else if (dtype == TInt64.DTYPE) {
+    } else if (type == TInt64.class) {
       Output<TInt64> o = (Output<TInt64>) input;
       AtomicInteger index = new AtomicInteger();
       if (debug) {
-        o.data()
+        o.asTensor()
             .scalars()
             .forEach(f -> System.out.printf("%d). %d\n", index.getAndIncrement(), f.getLong()));
       }
       index.set(0);
-      o.data()
+      o.asTensor()
           .scalars()
           .forEach(
               f -> assertEquals((long) expected.getFloat(index.getAndIncrement()), f.getLong()));
-    } else if (dtype == TUint8.DTYPE) {
+    } else if (type == TUint8.class) {
       Output<TUint8> o = (Output<TUint8>) input;
       AtomicInteger index = new AtomicInteger();
       if (debug) {
-        o.data()
+        o.asTensor()
             .scalars()
             .forEach(f -> System.out.printf("%d). %x\n", index.getAndIncrement(), f.getByte()));
       }
       index.set(0);
-      o.data()
+      o.asTensor()
           .scalars()
           .forEach(
               f -> assertEquals((long) expected.getFloat(index.getAndIncrement()), f.getByte()));
@@ -296,10 +296,10 @@ public void evaluateString(Output<TString> input, Predicate<String> predicate) {
     if (debug) {
       if (isScalar) {
         System.out.printf(
-            "0). %b <==> %s\n", predicate.test(input.data().getObject()), input.data().getObject());
+            "0). %b <==> %s\n", predicate.test(input.asTensor().getObject()), input.asTensor().getObject());
       } else {
         input
-            .data()
+            .asTensor()
             .scalars()
             .forEachIndexed(
                 (idx, s) ->
@@ -310,9 +310,9 @@ public void evaluateString(Output<TString> input, Predicate<String> predicate) {
     }
     index.set(0);
     if (isScalar) {
-      assertTrue(predicate.test(input.data().getObject()));
+      assertTrue(predicate.test(input.asTensor().getObject()));
     } else {
-      input.data().scalars().forEachIndexed((idx, s) -> assertTrue(predicate.test(s.getObject())));
+      input.asTensor().scalars().forEachIndexed((idx, s) -> assertTrue(predicate.test(s.getObject())));
     }
   }
 
@@ -320,16 +320,16 @@ public void evaluateString(Output<TString> input, Predicate<String> predicate) {
   @Override
   public <T extends TType> void evaluate(Output<T> input, Predicate<Number> predicate) {
     AtomicInteger index = new AtomicInteger();
-    DataType<T> dtype = input.asOutput().dataType();
+    Class<T> type = input.asOutput().type();
     boolean isScalar = input.shape().equals(Shape.scalar());
-    if (dtype == TFloat32.DTYPE) {
+    if (type == TFloat32.class) {
       Output<TFloat32> o = (Output<TFloat32>) input;
       if (debug) {
         if (isScalar) {
           System.out.printf(
-              "0). %b <==> %f\n", predicate.test(o.data().getFloat()), o.data().getFloat());
+              "0). %b <==> %f\n", predicate.test(o.asTensor().getFloat()), o.asTensor().getFloat());
         } else {
-          o.data()
+          o.asTensor()
               .scalars()
               .forEachIndexed(
                   (idx, f) ->
@@ -340,20 +340,20 @@ public <T extends TType> void evaluate(Output<T> input, Predicate<Number> predic
       }
       index.set(0);
       if (isScalar) {
-        assertTrue(predicate.test(o.data().getFloat()));
+        assertTrue(predicate.test(o.asTensor().getFloat()));
       } else {
-        o.data()
+        o.asTensor()
             .scalars()
-            .forEachIndexed((idx, f) -> assertTrue(predicate.test(o.data().getFloat())));
+            .forEachIndexed((idx, f) -> assertTrue(predicate.test(o.asTensor().getFloat())));
       }
-    } else if (dtype == TFloat64.DTYPE) {
+    } else if (type == TFloat64.class) {
       Output<TFloat64> o = (Output<TFloat64>) input;
       if (debug) {
         if (isScalar) {
           System.out.printf(
-              "0). %b <==> %f\n", predicate.test(o.data().getDouble()), o.data().getDouble());
+              "0). %b <==> %f\n", predicate.test(o.asTensor().getDouble()), o.asTensor().getDouble());
         } else {
-          o.data()
+          o.asTensor()
               .scalars()
               .forEachIndexed(
                   (idx, f) ->
@@ -364,20 +364,20 @@ public <T extends TType> void evaluate(Output<T> input, Predicate<Number> predic
       }
       index.set(0);
       if (isScalar) {
-        assertTrue(predicate.test(o.data().getDouble()));
+        assertTrue(predicate.test(o.asTensor().getDouble()));
       } else {
-        o.data()
+        o.asTensor()
             .scalars()
-            .forEachIndexed((idx, f) -> assertTrue(predicate.test(o.data().getDouble())));
+            .forEachIndexed((idx, f) -> assertTrue(predicate.test(o.asTensor().getDouble())));
       }
-    } else if (dtype == TFloat16.DTYPE) {
+    } else if (type == TFloat16.DTYPE) {
       Output<TFloat16> o = (Output<TFloat16>) input;
       if (debug) {
         if (isScalar) {
           System.out.printf(
-              "0). %b <==> %f\n", predicate.test(o.data().getFloat()), o.data().getFloat());
+              "0). %b <==> %f\n", predicate.test(o.asTensor().getFloat()), o.asTensor().getFloat());
         } else {
-          o.data()
+          o.asTensor()
               .scalars()
               .forEachIndexed(
                   (idx, f) ->
@@ -388,20 +388,20 @@ public <T extends TType> void evaluate(Output<T> input, Predicate<Number> predic
       }
       index.set(0);
       if (isScalar) {
-        assertTrue(predicate.test(o.data().getFloat()));
+        assertTrue(predicate.test(o.asTensor().getFloat()));
       } else {
-        o.data()
+        o.asTensor()
             .scalars()
-            .forEachIndexed((idx, f) -> assertTrue(predicate.test(o.data().getFloat())));
+            .forEachIndexed((idx, f) -> assertTrue(predicate.test(o.asTensor().getFloat())));
       }
-    } else if (dtype == TInt32.DTYPE) {
+    } else if (type == TInt32.class) {
       Output<TInt32> o = (Output<TInt32>) input;
       if (debug) {
         if (isScalar) {
           System.out.printf(
-              "0). %b <==> %d\n", predicate.test(o.data().getInt()), o.data().getInt());
+              "0). %b <==> %d\n", predicate.test(o.asTensor().getInt()), o.asTensor().getInt());
         } else {
-          o.data()
+          o.asTensor()
               .scalars()
               .forEachIndexed(
                   (idx, f) ->
@@ -412,20 +412,20 @@ public <T extends TType> void evaluate(Output<T> input, Predicate<Number> predic
       }
       index.set(0);
       if (isScalar) {
-        assertTrue(predicate.test(o.data().getInt()));
+        assertTrue(predicate.test(o.asTensor().getInt()));
       } else {
-        o.data()
+        o.asTensor()
             .scalars()
-            .forEachIndexed((idx, f) -> assertTrue(predicate.test(o.data().getInt())));
+            .forEachIndexed((idx, f) -> assertTrue(predicate.test(o.asTensor().getInt())));
       }
-    } else if (dtype == TInt64.DTYPE) {
+    } else if (type == TInt64.class) {
       Output<TInt64> o = (Output<TInt64>) input;
       if (debug) {
         if (isScalar) {
           System.out.printf(
-              "0). %b <==> %d\n", predicate.test(o.data().getLong()), o.data().getLong());
+              "0). %b <==> %d\n", predicate.test(o.asTensor().getLong()), o.asTensor().getLong());
         } else {
-          o.data()
+          o.asTensor()
               .scalars()
               .forEachIndexed(
                   (idx, f) ->
@@ -436,20 +436,20 @@ public <T extends TType> void evaluate(Output<T> input, Predicate<Number> predic
       }
       index.set(0);
       if (isScalar) {
-        assertTrue(predicate.test(o.data().getLong()));
+        assertTrue(predicate.test(o.asTensor().getLong()));
       } else {
-        o.data()
+        o.asTensor()
             .scalars()
-            .forEachIndexed((idx, f) -> assertTrue(predicate.test(o.data().getLong())));
+            .forEachIndexed((idx, f) -> assertTrue(predicate.test(o.asTensor().getLong())));
       }
-    } else if (dtype == TUint8.DTYPE) {
+    } else if (type == TUint8.class) {
       Output<TUint8> o = (Output<TUint8>) input;
       if (debug) {
         if (isScalar) {
           System.out.printf(
-              "0). %b <==> %x\n", predicate.test(o.data().getByte()), o.data().getByte());
+              "0). %b <==> %x\n", predicate.test(o.asTensor().getByte()), o.asTensor().getByte());
         } else {
-          o.data()
+          o.asTensor()
               .scalars()
               .forEachIndexed(
                   (idx, f) ->
@@ -460,14 +460,14 @@ public <T extends TType> void evaluate(Output<T> input, Predicate<Number> predic
       }
       index.set(0);
       if (isScalar) {
-        assertTrue(predicate.test(o.data().getByte()));
+        assertTrue(predicate.test(o.asTensor().getByte()));
       } else {
-        o.data()
+        o.asTensor()
             .scalars()
-            .forEachIndexed((idx, f) -> assertTrue(predicate.test(o.data().getByte())));
+            .forEachIndexed((idx, f) -> assertTrue(predicate.test(o.asTensor().getByte())));
       }
     } else {
-      fail("Unexpected DataType: " + dtype);
+      fail("Unexpected Class: " + type);
     }
   }
 
@@ -482,13 +482,13 @@ public void evaluate(String[] expected, Output<TString> input) {
     AtomicInteger index = new AtomicInteger();
     if (debug) {
       input
-          .data()
+          .asTensor()
           .scalars()
           .forEach(f -> System.out.printf("%d). %s\n", index.getAndIncrement(), f.getObject()));
     }
     index.set(0);
     input
-        .data()
+        .asTensor()
         .scalars()
         .forEach(f -> assertEquals(expected[index.getAndIncrement()], f.getObject()));
   }
@@ -504,13 +504,13 @@ public void evaluate(Boolean[] expected, Output<TBool> input) {
     AtomicInteger index = new AtomicInteger();
     if (debug) {
       input
-          .data()
+          .asTensor()
           .scalars()
           .forEach(f -> System.out.printf("%d). %b\n", index.getAndIncrement(), f.getBoolean()));
     }
     index.set(0);
     input
-        .data()
+        .asTensor()
         .scalars()
         .forEach(f -> assertEquals(expected[index.getAndIncrement()], f.getBoolean()));
   }
@@ -522,184 +522,184 @@ public <T extends TType> void evaluate(Output<T> expected, Output<T> input) {
         : String.format(
             "expected shape (%s) != to input shape (%s)",
             expected.shape().toString(), input.shape().toString());
-    DataType<T> dtype = input.asOutput().dataType();
+    Class<T> type = input.asOutput().type();
     boolean isScalar = input.shape().equals(Shape.scalar());
-    if (dtype == TFloat32.DTYPE) {
+    if (type == TFloat32.class) {
       Output<TFloat32> x = (Output<TFloat32>) expected;
       Output<TFloat32> o = (Output<TFloat32>) input;
       AtomicInteger index = new AtomicInteger();
       if (debug) {
         if (isScalar) {
-          System.out.printf("0). %f <==> %f\n", x.data().getFloat(), o.data().getFloat());
+          System.out.printf("0). %f <==> %f\n", x.asTensor().getFloat(), o.asTensor().getFloat());
         } else {
-          o.data()
+          o.asTensor()
               .scalars()
               .forEachIndexed(
                   (idx, f) ->
                       System.out.printf(
                           "%d). %f <==> %f\n",
-                          index.getAndIncrement(), x.data().getFloat(idx), f.getFloat()));
+                          index.getAndIncrement(), x.asTensor().getFloat(idx), f.getFloat()));
         }
       }
       index.set(0);
       if (isScalar) {
-        assertEquals(x.data().getFloat(), o.data().getFloat(), epsilon);
+        assertEquals(x.asTensor().getFloat(), o.asTensor().getFloat(), epsilon);
       } else {
-        o.data()
+        o.asTensor()
             .scalars()
             .forEachIndexed(
-                (idx, f) -> assertEquals(x.data().getFloat(idx), f.getFloat(), epsilon));
+                (idx, f) -> assertEquals(x.asTensor().getFloat(idx), f.getFloat(), epsilon));
       }
-    } else if (dtype == TFloat64.DTYPE) {
+    } else if (type == TFloat64.class) {
       Output<TFloat64> x = (Output<TFloat64>) expected;
       Output<TFloat64> o = (Output<TFloat64>) input;
       AtomicInteger index = new AtomicInteger();
       if (debug) {
         if (isScalar) {
-          System.out.printf("0). %f <==> %f\n", x.data().getDouble(), o.data().getDouble());
+          System.out.printf("0). %f <==> %f\n", x.asTensor().getDouble(), o.asTensor().getDouble());
         } else {
-          o.data()
+          o.asTensor()
               .scalars()
               .forEachIndexed(
                   (idx, f) ->
                       System.out.printf(
                           "%d). %f <==> %f\n",
-                          index.getAndIncrement(), x.data().getDouble(idx), f.getDouble()));
+                          index.getAndIncrement(), x.asTensor().getDouble(idx), f.getDouble()));
         }
       }
       index.set(0);
       if (isScalar) {
-        assertEquals(x.data().getDouble(), o.data().getDouble(), epsilon);
+        assertEquals(x.asTensor().getDouble(), o.asTensor().getDouble(), epsilon);
       } else {
-        o.data()
+        o.asTensor()
             .scalars()
             .forEachIndexed(
-                (idx, f) -> assertEquals(x.data().getDouble(idx), f.getDouble(), epsilon));
+                (idx, f) -> assertEquals(x.asTensor().getDouble(idx), f.getDouble(), epsilon));
       }
-    } else if (dtype == TInt32.DTYPE) {
+    } else if (type == TInt32.class) {
       Output<TInt32> x = (Output<TInt32>) expected;
       Output<TInt32> o = (Output<TInt32>) input;
       AtomicInteger index = new AtomicInteger();
       if (debug) {
         if (isScalar) {
-          System.out.printf("0). %d <==> %d\n", x.data().getInt(), o.data().getInt());
+          System.out.printf("0). %d <==> %d\n", x.asTensor().getInt(), o.asTensor().getInt());
         } else {
-          o.data()
+          o.asTensor()
               .scalars()
               .forEachIndexed(
                   (idx, f) ->
                       System.out.printf(
                           "%d). %d <==> %d\n",
-                          index.getAndIncrement(), x.data().getInt(idx), f.getInt()));
+                          index.getAndIncrement(), x.asTensor().getInt(idx), f.getInt()));
         }
       }
       index.set(0);
       if (isScalar) {
-        assertEquals(x.data().getInt(), o.data().getInt());
+        assertEquals(x.asTensor().getInt(), o.asTensor().getInt());
       } else {
-        o.data()
+        o.asTensor()
             .scalars()
-            .forEachIndexed((idx, f) -> assertEquals(x.data().getInt(idx), f.getInt()));
+            .forEachIndexed((idx, f) -> assertEquals(x.asTensor().getInt(idx), f.getInt()));
       }
-    } else if (dtype == TInt64.DTYPE) {
+    } else if (type == TInt64.class) {
       Output<TInt64> x = (Output<TInt64>) expected;
       Output<TInt64> o = (Output<TInt64>) input;
       AtomicInteger index = new AtomicInteger();
       if (debug) {
         if (isScalar) {
-          System.out.printf("0). %d  <==> %d\n", x.data().getLong(), o.data().getLong());
+          System.out.printf("0). %d  <==> %d\n", x.asTensor().getLong(), o.asTensor().getLong());
         } else {
-          o.data()
+          o.asTensor()
               .scalars()
               .forEachIndexed(
                   (idx, f) ->
                       System.out.printf(
                           "%d). %d  <==> %d\n",
-                          index.getAndIncrement(), x.data().getLong(idx), f.getLong()));
+                          index.getAndIncrement(), x.asTensor().getLong(idx), f.getLong()));
         }
       }
       index.set(0);
       if (isScalar) {
-        assertEquals(x.data().getLong(), o.data().getLong());
+        assertEquals(x.asTensor().getLong(), o.asTensor().getLong());
       } else {
-        o.data()
+        o.asTensor()
             .scalars()
-            .forEachIndexed((idx, f) -> assertEquals(x.data().getLong(idx), f.getLong()));
+            .forEachIndexed((idx, f) -> assertEquals(x.asTensor().getLong(idx), f.getLong()));
       }
-    } else if (dtype == TUint8.DTYPE) {
+    } else if (type == TUint8.class) {
       Output<TUint8> x = (Output<TUint8>) expected;
       Output<TUint8> o = (Output<TUint8>) input;
       AtomicInteger index = new AtomicInteger();
       if (debug) {
         if (isScalar) {
-          System.out.printf("0). %x  <==> %x\n", x.data().getByte(), o.data().getByte());
+          System.out.printf("0). %x  <==> %x\n", x.asTensor().getByte(), o.asTensor().getByte());
         } else {
-          o.data()
+          o.asTensor()
               .scalars()
               .forEachIndexed(
                   (idx, f) ->
                       System.out.printf(
                           "%d). %x  <==> %x\n",
-                          index.getAndIncrement(), x.data().getByte(idx), f.getByte()));
+                          index.getAndIncrement(), x.asTensor().getByte(idx), f.getByte()));
         }
       }
       index.set(0);
       if (isScalar) {
-        assertEquals(x.data().getByte(), o.data().getByte());
+        assertEquals(x.asTensor().getByte(), o.asTensor().getByte());
       } else {
-        o.data()
+        o.asTensor()
             .scalars()
-            .forEachIndexed((idx, f) -> assertEquals(x.data().getByte(idx), f.getByte()));
+            .forEachIndexed((idx, f) -> assertEquals(x.asTensor().getByte(idx), f.getByte()));
       }
-    } else if (dtype == TString.DTYPE) {
+    } else if (type == TString.class) {
       Output<TString> x = (Output<TString>) expected;
       Output<TString> o = (Output<TString>) input;
       AtomicInteger index = new AtomicInteger();
       if (debug) {
         if (isScalar) {
-          System.out.printf("0). %s  <==> %s\n", x.data().getObject(), o.data().getObject());
+          System.out.printf("0). %s  <==> %s\n", x.asTensor().getObject(), o.asTensor().getObject());
         } else {
-          o.data()
+          o.asTensor()
               .scalars()
               .forEachIndexed(
                   (idx, f) ->
                       System.out.printf(
                           "%d). %s  <==> %s\n",
-                          index.getAndIncrement(), x.data().getObject(idx), f.getObject()));
+                          index.getAndIncrement(), x.asTensor().getObject(idx), f.getObject()));
         }
       }
       index.set(0);
       if (isScalar) {
-        assertEquals(x.data().getObject(), o.data().getObject());
+        assertEquals(x.asTensor().getObject(), o.asTensor().getObject());
       } else {
-        o.data()
+        o.asTensor()
             .scalars()
-            .forEachIndexed((idx, f) -> assertEquals(x.data().getObject(idx), f.getObject()));
+            .forEachIndexed((idx, f) -> assertEquals(x.asTensor().getObject(idx), f.getObject()));
       }
-    } else if (dtype == TBool.DTYPE) {
+    } else if (type == TBool.class) {
       Output<TBool> x = (Output<TBool>) expected;
       Output<TBool> o = (Output<TBool>) input;
       AtomicInteger index = new AtomicInteger();
       if (debug) {
         if (isScalar) {
-          System.out.printf("0). %b  <==> %b\n", x.data().getBoolean(), o.data().getBoolean());
+          System.out.printf("0). %b  <==> %b\n", x.asTensor().getBoolean(), o.asTensor().getBoolean());
         } else {
-          o.data()
+          o.asTensor()
               .scalars()
               .forEachIndexed(
                   (idx, f) ->
                       System.out.printf(
                           "%d). %b  <==> %b\n",
-                          index.getAndIncrement(), x.data().getBoolean(idx), f.getBoolean()));
+                          index.getAndIncrement(), x.asTensor().getBoolean(idx), f.getBoolean()));
         }
       }
       index.set(0);
       if (isScalar) {
-        assertEquals(x.data().getBoolean(), o.data().getBoolean());
+        assertEquals(x.asTensor().getBoolean(), o.asTensor().getBoolean());
       } else {
-        o.data()
+        o.asTensor()
             .scalars()
-            .forEachIndexed((idx, f) -> assertEquals(x.data().getBoolean(idx), f.getBoolean()));
+            .forEachIndexed((idx, f) -> assertEquals(x.asTensor().getBoolean(idx), f.getBoolean()));
       }
     }
   }
@@ -707,51 +707,51 @@ public <T extends TType> void evaluate(Output<T> expected, Output<T> input) {
   /** {@inheritDoc} */
   @Override
   public <T extends TType> void print(PrintWriter writer, Output<T> input) {
-    DataType<T> dtype = input.asOutput().dataType();
-    if (dtype == TFloat32.DTYPE) {
+    Class<T> type = input.asOutput().type();
+    if (type == TFloat32.class) {
       Output<TFloat32> o = (Output<TFloat32>) input;
       AtomicInteger index = new AtomicInteger();
-      o.data()
+      o.asTensor()
           .scalars()
           .forEach(f -> System.out.printf("%d). %f\n", index.getAndIncrement(), f.getFloat()));
-    } else if (dtype == TFloat64.DTYPE) {
+    } else if (type == TFloat64.class) {
       Output<TFloat64> o = (Output<TFloat64>) input;
       AtomicInteger index = new AtomicInteger();
-      o.data()
+      o.asTensor()
           .scalars()
           .forEach(f -> System.out.printf("%d). %f\n", index.getAndIncrement(), f.getDouble()));
-    } else if (dtype == TInt32.DTYPE) {
+    } else if (type == TInt32.class) {
       Output<TInt32> o = (Output<TInt32>) input;
       AtomicInteger index = new AtomicInteger();
-      o.data()
+      o.asTensor()
           .scalars()
           .forEach(f -> System.out.printf("%d). %d\n", index.getAndIncrement(), f.getInt()));
-    } else if (dtype == TInt64.DTYPE) {
+    } else if (type == TInt64.class) {
       Output<TInt64> o = (Output<TInt64>) input;
       AtomicInteger index = new AtomicInteger();
-      o.data()
+      o.asTensor()
           .scalars()
           .forEach(f -> System.out.printf("%d). %d\n", index.getAndIncrement(), f.getLong()));
-    } else if (dtype == TUint8.DTYPE) {
+    } else if (type == TUint8.class) {
       Output<TUint8> o = (Output<TUint8>) input;
       AtomicInteger index = new AtomicInteger();
-      o.data()
+      o.asTensor()
           .scalars()
           .forEach(f -> System.out.printf("%d). %x\n", index.getAndIncrement(), f.getByte()));
-    } else if (dtype == TString.DTYPE) {
+    } else if (type == TString.class) {
       Output<TString> o = (Output<TString>) input;
       AtomicInteger index = new AtomicInteger();
-      o.data()
+      o.asTensor()
           .scalars()
           .forEach(f -> System.out.printf("%d). %s\n", index.getAndIncrement(), f.getObject()));
-    } else if (dtype == TBool.DTYPE) {
+    } else if (type == TBool.class) {
       Output<TBool> o = (Output<TBool>) input;
       AtomicInteger index = new AtomicInteger();
-      o.data()
+      o.asTensor()
           .scalars()
           .forEach(f -> System.out.printf("%d). %b\n", index.getAndIncrement(), f.getBoolean()));
     } else {
-      writer.println("Unexpected DataType: " + dtype);
+      writer.println("Unexpected Class: " + type);
     }
     writer.flush();
   }
diff --git a/tensorflow-framework/src/test/java/org/tensorflow/framework/utils/GraphTestSession.java b/tensorflow-framework/src/test/java/org/tensorflow/framework/utils/GraphTestSession.java
index 11750e8ef47..42d5f0fe696 100644
--- a/tensorflow-framework/src/test/java/org/tensorflow/framework/utils/GraphTestSession.java
+++ b/tensorflow-framework/src/test/java/org/tensorflow/framework/utils/GraphTestSession.java
@@ -19,7 +19,6 @@
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.Op;
 import org.tensorflow.op.Ops;
-import org.tensorflow.tensor.Tensor;
 import org.tensorflow.types.*;
 import org.tensorflow.types.family.TNumber;
 import org.tensorflow.types.family.TType;
@@ -105,89 +104,84 @@ public void run(Op op) {
   /** {@inheritDoc} */
   @Override
   public <T extends TNumber> void evaluate(double expected, Operand<T> input) {
-    DataType<T> dtype = input.asOutput().dataType();
-    if (dtype == TFloat32.DTYPE) {
+    Class<T> dtype = input.asOutput().type();
+    if (dtype == TFloat32.class) {
       AtomicInteger index = new AtomicInteger();
       if (debug) {
-        try (Tensor<TFloat32> result =
-            this.getGraphSession().runner().fetch(input).run().get(0).expect(TFloat32.DTYPE)) {
+        try (TFloat32 result =
+            this.getGraphSession().runner().fetch(input).run().get(0)) {
           result
-              .data()
               .scalars()
               .forEach(f -> System.out.printf("%d). %f\n", index.getAndIncrement(), f.getFloat()));
         }
       }
       index.set(0);
-      try (Tensor<TFloat32> result =
-          this.getGraphSession().runner().fetch(input).run().get(0).expect(TFloat32.DTYPE)) {
-        result.data().scalars().forEach(f -> assertEquals((float) expected, f.getFloat(), epsilon));
+      try (TFloat32 result =
+          this.getGraphSession().runner().fetch(input).run().get(0)) {
+        result.scalars().forEach(f -> assertEquals((float) expected, f.getFloat(), epsilon));
       }
-    } else if (dtype == TFloat64.DTYPE) {
+    } else if (dtype == TFloat64.class) {
       AtomicInteger index = new AtomicInteger();
       if (debug) {
-        try (Tensor<TFloat64> result =
-            this.getGraphSession().runner().fetch(input).run().get(0).expect(TFloat64.DTYPE)) {
+        try (TFloat64 result =
+            this.getGraphSession().runner().fetch(input).run().get(0)) {
           result
-              .data()
               .scalars()
               .forEach(f -> System.out.printf("%d). %f\n", index.getAndIncrement(), f.getDouble()));
         }
       }
       index.set(0);
-      try (Tensor<TFloat64> result =
-          this.getGraphSession().runner().fetch(input).run().get(0).expect(TFloat64.DTYPE)) {
-        result.data().scalars().forEach(f -> assertEquals(expected, f.getDouble(), epsilon));
+      try (TFloat64 result =
+          this.getGraphSession().runner().fetch(input).run().get(0)) {
+        result.scalars().forEach(f -> assertEquals(expected, f.getDouble(), epsilon));
       }
-    } else if (dtype == TInt32.DTYPE) {
+    } else if (dtype == TInt32.class) {
       AtomicInteger index = new AtomicInteger();
       if (debug) {
-        try (Tensor<TInt32> result =
-            this.getGraphSession().runner().fetch(input).run().get(0).expect(TInt32.DTYPE)) {
+        try (TInt32 result =
+            this.getGraphSession().runner().fetch(input).run().get(0)) {
           result
-              .data()
               .scalars()
               .forEach(f -> System.out.printf("%d). %d\n", index.getAndIncrement(), f.getInt()));
         }
       }
       index.set(0);
-      try (Tensor<TInt32> result =
-          this.getGraphSession().runner().fetch(input).run().get(0).expect(TInt32.DTYPE)) {
-        result.data().scalars().forEach(f -> assertEquals((int) expected, f.getInt()));
+      try (TInt32 result =
+          this.getGraphSession().runner().fetch(input).run().get(0)) {
+        result.scalars().forEach(f -> assertEquals((int) expected, f.getInt()));
       }
-    } else if (dtype == TInt64.DTYPE) {
+    } else if (dtype == TInt64.class) {
       AtomicInteger index = new AtomicInteger();
       if (debug) {
-        try (Tensor<TInt64> result =
-            this.getGraphSession().runner().fetch(input).run().get(0).expect(TInt64.DTYPE)) {
+        try (TInt64 result =
+            this.getGraphSession().runner().fetch(input).run().get(0)) {
           result
-              .data()
               .scalars()
               .forEach(f -> System.out.printf("%d). %d\n", index.getAndIncrement(), f.getLong()));
         }
       }
       index.set(0);
-      try (Tensor<TInt64> result =
-          this.getGraphSession().runner().fetch(input).run().get(0).expect(TInt64.DTYPE)) {
-        result.data().scalars().forEach(f -> assertEquals((long) expected, f.getLong()));
+      try (TInt64 result =
+          this.getGraphSession().runner().fetch(input).run().get(0)) {
+        result.scalars().forEach(f -> assertEquals((long) expected, f.getLong()));
       }
-    } else if (dtype == TUint8.DTYPE) {
+    } else if (dtype == TUint8.class) {
       AtomicInteger index = new AtomicInteger();
       if (debug) {
-        try (Tensor<TUint8> result =
-            this.getGraphSession().runner().fetch(input).run().get(0).expect(TUint8.DTYPE)) {
+        try (TUint8 result =
+            this.getGraphSession().runner().fetch(input).run().get(0)) {
           result
-              .data()
               .scalars()
               .forEach(f -> System.out.printf("%d). %d\n", index.getAndIncrement(), f.getByte()));
         }
       }
       index.set(0);
-      try (Tensor<TUint8> result =
-          this.getGraphSession().runner().fetch(input).run().get(0).expect(TUint8.DTYPE)) {
-        result.data().scalars().forEach(f -> assertEquals((long) expected, f.getByte()));
+      try (TUint8 result =
+          this.getGraphSession().runner().fetch(input).run().get(0)) {
+        result.scalars().forEach(f -> assertEquals((long) expected, f.getByte()));
       }
     } else {
-      fail("Unexpected DataType: " + dtype);
+      fail("Unexpected Class: " + dtype);
     }
   }
 
@@ -199,223 +193,203 @@ public <T extends TNumber> void evaluate(Number[] expected, Output<T> input) {
         expected.length,
         size,
         () -> String.format("expected length (%d) != to input length (%d)", expected.length, size));
-    DataType<T> dtype = input.asOutput().dataType();
-    if (dtype == TFloat32.DTYPE) {
+    Class<T> dtype = input.asOutput().type();
+    if (dtype == TFloat32.class) {
       AtomicInteger index = new AtomicInteger();
       if (debug) {
-        try (Tensor<TFloat32> result =
-            this.getGraphSession().runner().fetch(input).run().get(0).expect(TFloat32.DTYPE)) {
+        try (TFloat32 result =
+            this.getGraphSession().runner().fetch(input).run().get(0)) {
           result
-              .data()
               .scalars()
               .forEach(f -> System.out.printf("%d). %f\n", index.getAndIncrement(), f.getFloat()));
         }
       }
       index.set(0);
-      try (Tensor<TFloat32> result =
-          this.getGraphSession().runner().fetch(input).run().get(0).expect(TFloat32.DTYPE)) {
+      try (TFloat32 result =
+          this.getGraphSession().runner().fetch(input).run().get(0)) {
         result
-            .data()
             .scalars()
             .forEach(
                 f ->
                     assertEquals(
                         expected[index.getAndIncrement()].floatValue(), f.getFloat(), epsilon));
       }
-    } else if (dtype == TFloat64.DTYPE) {
+    } else if (dtype == TFloat64.class) {
       AtomicInteger index = new AtomicInteger();
       if (debug) {
-        try (Tensor<TFloat64> result =
-            this.getGraphSession().runner().fetch(input).run().get(0).expect(TFloat64.DTYPE)) {
+        try (TFloat64 result =
+            this.getGraphSession().runner().fetch(input).run().get(0)) {
           result
-              .data()
               .scalars()
               .forEach(f -> System.out.printf("%d). %f\n", index.getAndIncrement(), f.getDouble()));
         }
       }
       index.set(0);
-      try (Tensor<TFloat64> result =
-          this.getGraphSession().runner().fetch(input).run().get(0).expect(TFloat64.DTYPE)) {
+      try (TFloat64 result =
+          this.getGraphSession().runner().fetch(input).run().get(0)) {
         result
-            .data()
             .scalars()
             .forEach(
                 f ->
                     assertEquals(
                         expected[index.getAndIncrement()].doubleValue(), f.getDouble(), epsilon));
       }
-    } else if (dtype == TInt32.DTYPE) {
+    } else if (dtype == TInt32.class) {
       AtomicInteger index = new AtomicInteger();
       if (debug) {
-        try (Tensor<TInt32> result =
-            this.getGraphSession().runner().fetch(input).run().get(0).expect(TInt32.DTYPE)) {
+        try (TInt32 result =
+            this.getGraphSession().runner().fetch(input).run().get(0)) {
           result
-              .data()
               .scalars()
               .forEach(f -> System.out.printf("%d). %d\n", index.getAndIncrement(), f.getInt()));
         }
       }
       index.set(0);
-      try (Tensor<TInt32> result =
-          this.getGraphSession().runner().fetch(input).run().get(0).expect(TInt32.DTYPE)) {
+      try (TInt32 result =
+          this.getGraphSession().runner().fetch(input).run().get(0)) {
         result
-            .data()
             .scalars()
             .forEach(f -> assertEquals(expected[index.getAndIncrement()].intValue(), f.getInt()));
       }
-    } else if (dtype == TInt64.DTYPE) {
+    } else if (dtype == TInt64.class) {
       AtomicInteger index = new AtomicInteger();
       if (debug) {
-        try (Tensor<TInt64> result =
-            this.getGraphSession().runner().fetch(input).run().get(0).expect(TInt64.DTYPE)) {
+        try (TInt64 result =
+            this.getGraphSession().runner().fetch(input).run().get(0)) {
           result
-              .data()
               .scalars()
               .forEach(f -> System.out.printf("%d). %d\n", index.getAndIncrement(), f.getLong()));
         }
       }
       index.set(0);
-      try (Tensor<TInt64> result =
-          this.getGraphSession().runner().fetch(input).run().get(0).expect(TInt64.DTYPE)) {
+      try (TInt64 result =
+          this.getGraphSession().runner().fetch(input).run().get(0)) {
         result
-            .data()
             .scalars()
             .forEach(f -> assertEquals(expected[index.getAndIncrement()].longValue(), f.getLong()));
       }
-    } else if (dtype == TUint8.DTYPE) {
+    } else if (dtype == TUint8.class) {
       AtomicInteger index = new AtomicInteger();
       if (debug) {
-        try (Tensor<TUint8> result =
-            this.getGraphSession().runner().fetch(input).run().get(0).expect(TUint8.DTYPE)) {
+        try (TUint8 result =
+            this.getGraphSession().runner().fetch(input).run().get(0)) {
           result
-              .data()
               .scalars()
               .forEach(f -> System.out.printf("%d). %d\n", index.getAndIncrement(), f.getByte()));
         }
       }
       index.set(0);
-      try (Tensor<TUint8> result =
-          this.getGraphSession().runner().fetch(input).run().get(0).expect(TUint8.DTYPE)) {
+      try (TUint8 result =
+          this.getGraphSession().runner().fetch(input).run().get(0)) {
         result
-            .data()
             .scalars()
             .forEach(f -> assertEquals(expected[index.getAndIncrement()].longValue(), f.getByte()));
       }
     } else {
-      fail("Unexpected DataType: " + dtype);
+      fail("Unexpected Class: " + dtype);
     }
   }
 
   /** {@inheritDoc} */
   @Override
   public <T extends TType> void evaluate(FloatNdArray expected, Output<T> input) {
-    DataType<T> dtype = input.asOutput().dataType();
-    if (dtype == TFloat32.DTYPE) {
+    Class<T> dtype = input.asOutput().type();
+    if (dtype == TFloat32.class) {
       AtomicLong index = new AtomicLong();
       if (debug) {
-        try (Tensor<TFloat32> result =
-            this.getGraphSession().runner().fetch(input).run().get(0).expect(TFloat32.DTYPE)) {
+        try (TFloat32 result =
+            this.getGraphSession().runner().fetch(input).run().get(0)) {
           result
-              .data()
               .scalars()
               .forEach(f -> System.out.printf("%d). %f\n", index.getAndIncrement(), f.getFloat()));
         }
       }
       index.set(0);
-      try (Tensor<TFloat32> result =
-          this.getGraphSession().runner().fetch(input).run().get(0).expect(TFloat32.DTYPE)) {
+      try (TFloat32 result =
+          this.getGraphSession().runner().fetch(input).run().get(0)) {
         result
-            .data()
             .scalars()
             .forEach(
                 f ->
                     assertEquals(
                         expected.getFloat(index.getAndIncrement()), f.getFloat(), epsilon));
       }
-    } else if (dtype == TFloat64.DTYPE) {
+    } else if (dtype == TFloat64.class) {
       AtomicInteger index = new AtomicInteger();
       if (debug) {
-        try (Tensor<TFloat64> result =
-            this.getGraphSession().runner().fetch(input).run().get(0).expect(TFloat64.DTYPE)) {
+        try (TFloat64 result =
+            this.getGraphSession().runner().fetch(input).run().get(0)) {
           result
-              .data()
               .scalars()
               .forEach(f -> System.out.printf("%d). %f\n", index.getAndIncrement(), f.getDouble()));
         }
       }
       index.set(0);
-      try (Tensor<TFloat64> result =
-          this.getGraphSession().runner().fetch(input).run().get(0).expect(TFloat64.DTYPE)) {
+      try (TFloat64 result =
+          this.getGraphSession().runner().fetch(input).run().get(0)) {
         result
-            .data()
             .scalars()
             .forEach(
                 f ->
                     assertEquals(
                         expected.getFloat(index.getAndIncrement()), f.getDouble(), epsilon));
       }
-    } else if (dtype == TInt32.DTYPE) {
+    } else if (dtype == TInt32.class) {
       AtomicInteger index = new AtomicInteger();
       if (debug) {
-        try (Tensor<TInt32> result =
-            this.getGraphSession().runner().fetch(input).run().get(0).expect(TInt32.DTYPE)) {
+        try (TInt32 result =
+            this.getGraphSession().runner().fetch(input).run().get(0)) {
           result
-              .data()
               .scalars()
               .forEach(f -> System.out.printf("%d). %d\n", index.getAndIncrement(), f.getInt()));
         }
       }
       index.set(0);
-      try (Tensor<TInt32> result =
-          this.getGraphSession().runner().fetch(input).run().get(0).expect(TInt32.DTYPE)) {
+      try (TInt32 result =
+          this.getGraphSession().runner().fetch(input).run().get(0)) {
         result
-            .data()
             .scalars()
             .forEach(
                 f -> assertEquals((int) expected.getFloat(index.getAndIncrement()), f.getInt()));
       }
-    } else if (dtype == TInt64.DTYPE) {
+    } else if (dtype == TInt64.class) {
       AtomicInteger index = new AtomicInteger();
       if (debug) {
-        try (Tensor<TInt64> result =
-            this.getGraphSession().runner().fetch(input).run().get(0).expect(TInt64.DTYPE)) {
+        try (TInt64 result =
+            this.getGraphSession().runner().fetch(input).run().get(0)) {
           result
-              .data()
               .scalars()
               .forEach(f -> System.out.printf("%d). %d\n", index.getAndIncrement(), f.getLong()));
         }
       }
       index.set(0);
-      try (Tensor<TInt64> result =
-          this.getGraphSession().runner().fetch(input).run().get(0).expect(TInt64.DTYPE)) {
+      try (TInt64 result =
+          this.getGraphSession().runner().fetch(input).run().get(0)) {
         result
-            .data()
             .scalars()
             .forEach(
                 f -> assertEquals((long) expected.getFloat(index.getAndIncrement()), f.getLong()));
       }
-    } else if (dtype == TUint8.DTYPE) {
+    } else if (dtype == TUint8.class) {
       AtomicInteger index = new AtomicInteger();
       if (debug) {
-        try (Tensor<TUint8> result =
-            this.getGraphSession().runner().fetch(input).run().get(0).expect(TUint8.DTYPE)) {
+        try (TUint8 result =
+            this.getGraphSession().runner().fetch(input).run().get(0)) {
           result
-              .data()
               .scalars()
               .forEach(f -> System.out.printf("%d). %d\n", index.getAndIncrement(), f.getByte()));
         }
       }
       index.set(0);
-      try (Tensor<TUint8> result =
-          this.getGraphSession().runner().fetch(input).run().get(0).expect(TUint8.DTYPE)) {
+      try (TUint8 result =
+          this.getGraphSession().runner().fetch(input).run().get(0)) {
         result
-            .data()
             .scalars()
             .forEach(
                 f -> assertEquals((long) expected.getFloat(index.getAndIncrement()), f.getByte()));
       }
     } else {
-      fail("Unexpected DataType: " + dtype);
+      fail("Unexpected Class: " + dtype);
     }
   }
 
@@ -429,19 +403,17 @@ public void evaluate(String[] expected, Output<TString> input) {
         () -> String.format("expected length (%d) != to input length (%d)", expected.length, size));
     AtomicInteger index = new AtomicInteger();
     if (debug) {
-      try (Tensor<TString> result =
-          this.getGraphSession().runner().fetch(input).run().get(0).expect(TString.DTYPE)) {
+      try (TString result =
+          this.getGraphSession().runner().fetch(input).run().get(0)) {
         result
-            .data()
             .scalars()
             .forEach(f -> System.out.printf("%d). %s\n", index.getAndIncrement(), f.getObject()));
       }
     }
     index.set(0);
-    try (Tensor<TString> result =
-        this.getGraphSession().runner().fetch(input).run().get(0).expect(TString.DTYPE)) {
+    try (TString result =
+        this.getGraphSession().runner().fetch(input).run().get(0)) {
       result
-          .data()
           .scalars()
           .forEach(f -> assertEquals(expected[index.getAndIncrement()], f.getObject()));
     }
@@ -457,19 +429,17 @@ public void evaluate(Boolean[] expected, Output<TBool> input) {
         () -> String.format("expected length (%d) != to input length (%d)", expected.length, size));
     AtomicInteger index = new AtomicInteger();
     if (debug) {
-      try (Tensor<TBool> result =
-          this.getGraphSession().runner().fetch(input).run().get(0).expect(TBool.DTYPE)) {
+      try (TBool result =
+          this.getGraphSession().runner().fetch(input).run().get(0)) {
         result
-            .data()
             .scalars()
             .forEach(f -> System.out.printf("%d). %b\n", index.getAndIncrement(), f.getObject()));
       }
     }
     index.set(0);
-    try (Tensor<TBool> result =
-        this.getGraphSession().runner().fetch(input).run().get(0).expect(TBool.DTYPE)) {
+    try (TBool result =
+        this.getGraphSession().runner().fetch(input).run().get(0)) {
       result
-          .data()
           .scalars()
           .forEach(f -> assertEquals(expected[index.getAndIncrement()], f.getObject()));
     }
@@ -483,331 +453,314 @@ public <T extends TType> void evaluate(Output<T> expected, Output<T> input) {
             "expected shape (%s) != to input shape (%s)",
             expected.shape().toString(), input.shape().toString());
     AtomicInteger index = new AtomicInteger();
-    DataType<T> dtype = input.asOutput().dataType();
-    if (!dtype.equals(expected.dataType())) {
+    Class<T> dtype = input.asOutput().type();
+    if (!dtype.equals(expected.type())) {
       throw new IllegalArgumentException(
           String.format(
               "Both data type must be equal, inout = %s, expected = %s",
-              dtype, expected.dataType()));
+              dtype, expected.type()));
     }
     boolean isScalar = input.shape().equals(Shape.scalar());
-    if (dtype == TFloat32.DTYPE) {
+    if (dtype == TFloat32.class) {
       final Output<TFloat32> finalExpected = (Output<TFloat32>) expected;
       if (debug) {
-        try (Tensor<TFloat32> result =
-                this.getGraphSession().runner().fetch(input).run().get(0).expect(TFloat32.DTYPE);
-            Tensor<TFloat32> expectedResult =
-                this.getGraphSession().runner().fetch(input).run().get(0).expect(TFloat32.DTYPE)) {
+        try (TFloat32 result =
+                this.getGraphSession().runner().fetch(input).run().get(0);
+            TFloat32 expectedResult =
+                this.getGraphSession().runner().fetch(input).run().get(0)) {
           if (isScalar) {
             System.out.printf(
-                "0). %f <==> %f\n", expectedResult.data().getFloat(), result.data().getFloat());
+                "0). %f <==> %f\n", expectedResult.getFloat(), result.getFloat());
           } else {
             result
-                .data()
                 .scalars()
                 .forEachIndexed(
                     (idx, f) ->
                         System.out.printf(
                             "%d). %f <==> %f\n",
                             index.getAndIncrement(),
-                            finalExpected.data().getFloat(idx),
+                            finalExpected.asTensor().getFloat(idx),
                             f.getFloat()));
           }
         }
       }
       index.set(0);
-      try (Tensor<TFloat32> result =
-              this.getGraphSession().runner().fetch(input).run().get(0).expect(TFloat32.DTYPE);
-          Tensor<TFloat32> expectedResult =
-              this.getGraphSession().runner().fetch(input).run().get(0).expect(TFloat32.DTYPE)) {
+      try (TFloat32 result =
+              this.getGraphSession().runner().fetch(input).run().get(0);
+          TFloat32 expectedResult =
+              this.getGraphSession().runner().fetch(input).run().get(0)) {
         if (isScalar) {
-          assertEquals(expectedResult.data().getFloat(), result.data().getFloat(), epsilon);
+          assertEquals(expectedResult.getFloat(), result.getFloat(), epsilon);
         } else {
           result
-              .data()
               .scalars()
               .forEachIndexed(
                   (idx, f) ->
-                      assertEquals(expectedResult.data().getFloat(idx), f.getFloat(), epsilon));
+                      assertEquals(expectedResult.getFloat(idx), f.getFloat(), epsilon));
         }
       }
-    } else if (dtype == TFloat64.DTYPE) {
+    } else if (dtype == TFloat64.class) {
       final Output<TFloat64> finalExpected = (Output<TFloat64>) expected;
       if (debug) {
-        try (Tensor<TFloat64> result =
-                this.getGraphSession().runner().fetch(input).run().get(0).expect(TFloat64.DTYPE);
-            Tensor<TFloat64> expectedResult =
-                this.getGraphSession().runner().fetch(input).run().get(0).expect(TFloat64.DTYPE)) {
+        try (TFloat64 result =
+                this.getGraphSession().runner().fetch(input).run().get(0);
+            TFloat64 expectedResult =
+                this.getGraphSession().runner().fetch(input).run().get(0)) {
           if (isScalar) {
             System.out.printf(
-                "0). %f <==> %f\n", expectedResult.data().getDouble(), result.data().getDouble());
+                "0). %f <==> %f\n", expectedResult.getDouble(), result.getDouble());
           } else {
             result
-                .data()
                 .scalars()
                 .forEachIndexed(
                     (idx, f) ->
                         System.out.printf(
                             "%d). %f <==> %f\n",
                             index.getAndIncrement(),
-                            finalExpected.data().getDouble(idx),
+                            finalExpected.asTensor().getDouble(idx),
                             f.getDouble()));
           }
         }
       }
       index.set(0);
-      try (Tensor<TFloat64> result =
-              this.getGraphSession().runner().fetch(input).run().get(0).expect(TFloat64.DTYPE);
-          Tensor<TFloat64> expectedResult =
-              this.getGraphSession().runner().fetch(input).run().get(0).expect(TFloat64.DTYPE)) {
+      try (TFloat64 result =
+              this.getGraphSession().runner().fetch(input).run().get(0);
+          TFloat64 expectedResult =
+              this.getGraphSession().runner().fetch(input).run().get(0)) {
         if (isScalar) {
-          assertEquals(expectedResult.data().getDouble(), result.data().getDouble(), epsilon);
+          assertEquals(expectedResult.getDouble(), result.getDouble(), epsilon);
         } else {
           result
-              .data()
               .scalars()
               .forEachIndexed(
                   (idx, f) ->
-                      assertEquals(expectedResult.data().getDouble(idx), f.getDouble(), epsilon));
+                      assertEquals(expectedResult.getDouble(idx), f.getDouble(), epsilon));
         }
       }
     } else if (dtype == TFloat16.DTYPE) {
-      final Output<TFloat16> finalExpected = (Output<TFloat16>) expected;
-      if (debug) {
-        try (Tensor<TFloat16> result =
-                     this.getGraphSession().runner().fetch(input).run().get(0).expect(TFloat16.DTYPE);
-             Tensor<TFloat16> expectedResult =
-                     this.getGraphSession().runner().fetch(input).run().get(0).expect(TFloat16.DTYPE)) {
+        final Output<TFloat16> finalExpected = (Output<TFloat16>) expected;
+        if (debug) {
+          try (TFloat16 result =
+                  this.getGraphSession().runner().fetch(input).run().get(0);
+              TFloat16 expectedResult =
+                  this.getGraphSession().runner().fetch(input).run().get(0)) {
+            if (isScalar) {
+              System.out.printf(
+                  "0). %f <==> %f\n", expectedResult.getFloat(), result.getFloat());
+            } else {
+              result
+                  .scalars()
+                  .forEachIndexed(
+                      (idx, f) ->
+                          System.out.printf(
+                              "%d). %f <==> %f\n",
+                              index.getAndIncrement(),
+                              finalExpected.asTensor().getFloat(idx),
+                              f.getFloat()));
+            }
+          }
+        }
+        index.set(0);
+        try (TFloat16 result =
+                this.getGraphSession().runner().fetch(input).run().get(0);
+            TFloat16 expectedResult =
+                this.getGraphSession().runner().fetch(input).run().get(0)) {
           if (isScalar) {
-            System.out.printf(
-                    "0). %f <==> %f\n", expectedResult.data().getFloat(), result.data().getFloat());
+            assertEquals(expectedResult.getFloat(), result.getFloat(), epsilon);
           } else {
             result
-                    .data()
-                    .scalars()
-                    .forEachIndexed(
-                            (idx, f) ->
-                                    System.out.printf(
-                                            "%d). %f <==> %f\n",
-                                            index.getAndIncrement(),
-                                            finalExpected.data().getFloat(idx),
-                                            f.getFloat()));
+                .scalars()
+                .forEachIndexed(
+                    (idx, f) ->
+                        assertEquals(expectedResult.getFloat(idx), f.getFloat(), epsilon));
           }
         }
       }
-      index.set(0);
-      try (Tensor<TFloat16> result =
-                   this.getGraphSession().runner().fetch(input).run().get(0).expect(TFloat16.DTYPE);
-           Tensor<TFloat16> expectedResult =
-                   this.getGraphSession().runner().fetch(input).run().get(0).expect(TFloat16.DTYPE)) {
-        if (isScalar) {
-          assertEquals(expectedResult.data().getFloat(), result.data().getFloat(), epsilon);
-        } else {
-          result
-                  .data()
-                  .scalars()
-                  .forEachIndexed(
-                          (idx, f) ->
-                                  assertEquals(expectedResult.data().getFloat(idx), f.getFloat(), epsilon));
-        }
-      }
-    } else if (dtype == TInt32.DTYPE) {
+    } else if (dtype == TInt32.class) {
       final Output<TInt32> finalExpected = (Output<TInt32>) expected;
       if (debug) {
-        try (Tensor<TInt32> result =
-                this.getGraphSession().runner().fetch(input).run().get(0).expect(TInt32.DTYPE);
-            Tensor<TInt32> expectedResult =
-                this.getGraphSession().runner().fetch(input).run().get(0).expect(TInt32.DTYPE)) {
+        try (TInt32 result =
+                this.getGraphSession().runner().fetch(input).run().get(0);
+            TInt32 expectedResult =
+                this.getGraphSession().runner().fetch(input).run().get(0)) {
           if (isScalar) {
             System.out.printf(
-                "0). %d <==> %d\n", expectedResult.data().getInt(), result.data().getInt());
+                "0). %d <==> %d\n", expectedResult.getInt(), result.getInt());
           } else {
             result
-                .data()
                 .scalars()
                 .forEachIndexed(
                     (idx, f) ->
                         System.out.printf(
                             "%d). %d <==> %d\n",
-                            index.getAndIncrement(), finalExpected.data().getInt(idx), f.getInt()));
+                            index.getAndIncrement(), finalExpected.asTensor().getInt(idx), f.getInt()));
           }
         }
       }
       index.set(0);
-      try (Tensor<TInt32> result =
-              this.getGraphSession().runner().fetch(input).run().get(0).expect(TInt32.DTYPE);
-          Tensor<TInt32> expectedResult =
-              this.getGraphSession().runner().fetch(input).run().get(0).expect(TInt32.DTYPE)) {
+      try (TInt32 result =
+              this.getGraphSession().runner().fetch(input).run().get(0);
+          TInt32 expectedResult =
+              this.getGraphSession().runner().fetch(input).run().get(0)) {
         if (isScalar) {
-          assertEquals(expectedResult.data().getInt(), result.data().getInt(), epsilon);
+          assertEquals(expectedResult.getInt(), result.getInt(), epsilon);
         } else {
           result
-              .data()
               .scalars()
               .forEachIndexed(
-                  (idx, f) -> assertEquals(expectedResult.data().getInt(idx), f.getInt(), epsilon));
+                  (idx, f) -> assertEquals(expectedResult.getInt(idx), f.getInt(), epsilon));
         }
       }
-    } else if (dtype == TInt64.DTYPE) {
+    } else if (dtype == TInt64.class) {
       final Output<TInt64> finalExpected = (Output<TInt64>) expected;
       if (debug) {
-        try (Tensor<TInt64> result =
-                this.getGraphSession().runner().fetch(input).run().get(0).expect(TInt64.DTYPE);
-            Tensor<TInt64> expectedResult =
-                this.getGraphSession().runner().fetch(input).run().get(0).expect(TInt64.DTYPE)) {
+        try (TInt64 result =
+                this.getGraphSession().runner().fetch(input).run().get(0);
+            TInt64 expectedResult =
+                this.getGraphSession().runner().fetch(input).run().get(0)) {
           if (isScalar) {
             System.out.printf(
-                "0). %d <==> %d\n", expectedResult.data().getLong(), result.data().getLong());
+                "0). %d <==> %d\n", expectedResult.getLong(), result.getLong());
           } else {
             result
-                .data()
                 .scalars()
                 .forEachIndexed(
                     (idx, f) ->
                         System.out.printf(
                             "%d). %d <==> %d\n",
                             index.getAndIncrement(),
-                            finalExpected.data().getLong(idx),
+                            finalExpected.asTensor().getLong(idx),
                             f.getLong()));
           }
         }
       }
       index.set(0);
-      try (Tensor<TInt64> result =
-              this.getGraphSession().runner().fetch(input).run().get(0).expect(TInt64.DTYPE);
-          Tensor<TInt64> expectedResult =
-              this.getGraphSession().runner().fetch(input).run().get(0).expect(TInt64.DTYPE)) {
+      try (TInt64 result =
+              this.getGraphSession().runner().fetch(input).run().get(0);
+          TInt64 expectedResult =
+              this.getGraphSession().runner().fetch(input).run().get(0)) {
         if (isScalar) {
-          assertEquals(expectedResult.data().getLong(), result.data().getLong(), epsilon);
+          assertEquals(expectedResult.getLong(), result.getLong(), epsilon);
         } else {
           result
-              .data()
               .scalars()
               .forEachIndexed(
                   (idx, f) ->
-                      assertEquals(expectedResult.data().getLong(idx), f.getLong(), epsilon));
+                      assertEquals(expectedResult.getLong(idx), f.getLong(), epsilon));
         }
       }
-    } else if (dtype == TUint8.DTYPE) {
+    } else if (dtype == TUint8.class) {
       final Output<TUint8> finalExpected = (Output<TUint8>) expected;
       if (debug) {
-        try (Tensor<TUint8> result =
-                this.getGraphSession().runner().fetch(input).run().get(0).expect(TUint8.DTYPE);
-            Tensor<TUint8> expectedResult =
-                this.getGraphSession().runner().fetch(input).run().get(0).expect(TUint8.DTYPE)) {
+        try (TUint8 result =
+                this.getGraphSession().runner().fetch(input).run().get(0);
+            TUint8 expectedResult =
+                this.getGraphSession().runner().fetch(input).run().get(0)) {
           if (isScalar) {
             System.out.printf(
-                "0). %d <==> %d\n", expectedResult.data().getByte(), result.data().getByte());
+                "0). %d <==> %d\n", expectedResult.getByte(), result.getByte());
           } else {
             result
-                .data()
                 .scalars()
                 .forEachIndexed(
                     (idx, f) ->
                         System.out.printf(
                             "%d). %d <==> %d\n",
                             index.getAndIncrement(),
-                            finalExpected.data().getByte(idx),
+                            finalExpected.asTensor().getByte(idx),
                             f.getByte()));
           }
         }
       }
       index.set(0);
-      try (Tensor<TUint8> result =
-              this.getGraphSession().runner().fetch(input).run().get(0).expect(TUint8.DTYPE);
-          Tensor<TUint8> expectedResult =
-              this.getGraphSession().runner().fetch(input).run().get(0).expect(TUint8.DTYPE)) {
+      try (TUint8 result =
+              this.getGraphSession().runner().fetch(input).run().get(0);
+          TUint8 expectedResult =
+              this.getGraphSession().runner().fetch(input).run().get(0)) {
         if (isScalar) {
-          assertEquals(expectedResult.data().getByte(), result.data().getByte(), epsilon);
+          assertEquals(expectedResult.getByte(), result.getByte(), epsilon);
         } else {
           result
-              .data()
               .scalars()
               .forEachIndexed(
                   (idx, f) ->
-                      assertEquals(expectedResult.data().getByte(idx), f.getByte(), epsilon));
+                      assertEquals(expectedResult.getByte(idx), f.getByte(), epsilon));
         }
       }
-    } else if (dtype == TBool.DTYPE) {
+    } else if (dtype == TBool.class) {
       final Output<TBool> finalExpected = (Output<TBool>) expected;
       if (debug) {
-        try (Tensor<TBool> result =
-                this.getGraphSession().runner().fetch(input).run().get(0).expect(TBool.DTYPE);
-            Tensor<TBool> expectedResult =
-                this.getGraphSession().runner().fetch(input).run().get(0).expect(TBool.DTYPE)) {
+        try (TBool result =
+                this.getGraphSession().runner().fetch(input).run().get(0);
+            TBool expectedResult =
+                this.getGraphSession().runner().fetch(input).run().get(0)) {
           if (isScalar) {
             System.out.printf(
-                "0). %b <==> %b\n", expectedResult.data().getBoolean(), result.data().getBoolean());
+                "0). %b <==> %b\n", expectedResult.getBoolean(), result.getBoolean());
           } else {
             result
-                .data()
                 .scalars()
                 .forEachIndexed(
                     (idx, f) ->
                         System.out.printf(
                             "%d). %b <==> %b\n",
                             index.getAndIncrement(),
-                            finalExpected.data().getBoolean(idx),
+                            finalExpected.asTensor().getBoolean(idx),
                             f.getBoolean()));
           }
         }
       }
       index.set(0);
-      try (Tensor<TBool> result =
-              this.getGraphSession().runner().fetch(input).run().get(0).expect(TBool.DTYPE);
-          Tensor<TBool> expectedResult =
-              this.getGraphSession().runner().fetch(input).run().get(0).expect(TBool.DTYPE)) {
+      try (TBool result = this.getGraphSession().runner().fetch(input).run().get(0);
+          TBool expectedResult = this.getGraphSession().runner().fetch(input).run().get(0)) {
         if (isScalar) {
-          assertEquals(expectedResult.data().getBoolean(), result.data().getBoolean());
+          assertEquals(expectedResult.getBoolean(), result.getBoolean());
         } else {
           result
-              .data()
               .scalars()
               .forEachIndexed(
-                  (idx, f) -> assertEquals(expectedResult.data().getBoolean(idx), f.getBoolean()));
+                  (idx, f) -> assertEquals(expectedResult.getBoolean(idx), f.getBoolean()));
         }
       }
-    } else if (dtype == TString.DTYPE) {
+    } else if (dtype == TString.class) {
       final Output<TString> finalExpected = (Output<TString>) expected;
       if (debug) {
-        try (Tensor<TString> result =
-                this.getGraphSession().runner().fetch(input).run().get(0).expect(TString.DTYPE);
-            Tensor<TString> expectedResult =
-                this.getGraphSession().runner().fetch(input).run().get(0).expect(TString.DTYPE)) {
+        try (TString result =
+                this.getGraphSession().runner().fetch(input).run().get(0);
+            TString expectedResult =
+                this.getGraphSession().runner().fetch(input).run().get(0)) {
           if (isScalar) {
             System.out.printf(
-                "0). %s <==> %s\n", expectedResult.data().getObject(), result.data().getObject());
+                "0). %s <==> %s\n", expectedResult.getObject(), result.getObject());
           } else {
             result
-                .data()
                 .scalars()
                 .forEachIndexed(
                     (idx, f) ->
                         System.out.printf(
                             "%d). %s <==> %s\n",
                             index.getAndIncrement(),
-                            finalExpected.data().getObject(idx),
+                            finalExpected.asTensor().getObject(idx),
                             f.getObject()));
           }
         }
       }
       index.set(0);
-      try (Tensor<TString> result =
-              this.getGraphSession().runner().fetch(input).run().get(0).expect(TString.DTYPE);
-          Tensor<TString> expectedResult =
-              this.getGraphSession().runner().fetch(input).run().get(0).expect(TString.DTYPE)) {
+      try (TString result =
+              this.getGraphSession().runner().fetch(input).run().get(0);
+          TString expectedResult =
+              this.getGraphSession().runner().fetch(input).run().get(0)) {
         if (isScalar) {
-          assertEquals(expectedResult.data().getObject(), result.data().getObject());
+          assertEquals(expectedResult.getObject(), result.getObject());
         } else {
           result
-              .data()
               .scalars()
               .forEachIndexed(
-                  (idx, f) -> assertEquals(expectedResult.data().getObject(idx), f.getObject()));
+                  (idx, f) -> assertEquals(expectedResult.getObject(idx), f.getObject()));
         }
       }
     } else {
-      fail("Unexpected DataType: " + dtype);
+      fail("Unexpected Class: " + dtype);
     }
   }
 
@@ -817,15 +770,14 @@ public void evaluateString(Output<TString> input, Predicate<String> predicate) {
     boolean isScalar = input.shape().equals(Shape.scalar());
     AtomicInteger index = new AtomicInteger();
     if (debug) {
-      try (Tensor<TString> result =
-          this.getGraphSession().runner().fetch(input).run().get(0).expect(TString.DTYPE)) {
+      try (TString result =
+          this.getGraphSession().runner().fetch(input).run().get(0)) {
         if (isScalar) {
           System.out.printf(
               "0). %b <==> %s\n",
-              predicate.test(result.data().getObject()), result.data().getObject());
+              predicate.test(result.getObject()), result.getObject());
         } else {
           result
-              .data()
               .scalars()
               .forEachIndexed(
                   (idx, f) ->
@@ -836,13 +788,12 @@ public void evaluateString(Output<TString> input, Predicate<String> predicate) {
       }
     }
     index.set(0);
-    try (Tensor<TString> result =
-        this.getGraphSession().runner().fetch(input).run().get(0).expect(TString.DTYPE)) {
+    try (TString result =
+        this.getGraphSession().runner().fetch(input).run().get(0)) {
       if (isScalar) {
-        assertTrue(predicate.test(result.data().getObject()));
+        assertTrue(predicate.test(result.getObject()));
       } else {
         result
-            .data()
             .scalars()
             .forEachIndexed((idx, s) -> assertTrue(predicate.test(s.getObject())));
       }
@@ -853,19 +804,18 @@ public void evaluateString(Output<TString> input, Predicate<String> predicate) {
   @Override
   public <T extends TType> void evaluate(Output<T> input, Predicate<Number> predicate) {
     AtomicInteger index = new AtomicInteger();
-    DataType<T> dtype = input.asOutput().dataType();
+    Class<T> dtype = input.asOutput().type();
     boolean isScalar = input.shape().equals(Shape.scalar());
-    if (dtype == TFloat32.DTYPE) {
+    if (dtype == TFloat32.class) {
       if (debug) {
-        try (Tensor<TFloat32> result =
-            this.getGraphSession().runner().fetch(input).run().get(0).expect(TFloat32.DTYPE)) {
+        try (TFloat32 result =
+            this.getGraphSession().runner().fetch(input).run().get(0)) {
           if (isScalar) {
             System.out.printf(
                 "0). %b <==> %f\n",
-                predicate.test(result.data().getFloat()), result.data().getFloat());
+                predicate.test(result.getFloat()), result.getFloat());
           } else {
             result
-                .data()
                 .scalars()
                 .forEachIndexed(
                     (idx, f) ->
@@ -876,28 +826,26 @@ public <T extends TType> void evaluate(Output<T> input, Predicate<Number> predic
         }
       }
       index.set(0);
-      try (Tensor<TFloat32> result =
-          this.getGraphSession().runner().fetch(input).run().get(0).expect(TFloat32.DTYPE)) {
+      try (TFloat32 result =
+          this.getGraphSession().runner().fetch(input).run().get(0)) {
         if (isScalar) {
-          assertTrue(predicate.test(result.data().getFloat()));
+          assertTrue(predicate.test(result.getFloat()));
         } else {
           result
-              .data()
               .scalars()
-              .forEachIndexed((idx, f) -> assertTrue(predicate.test(result.data().getFloat())));
+              .forEachIndexed((idx, f) -> assertTrue(predicate.test(result.getFloat())));
         }
       }
-    } else if (dtype == TFloat64.DTYPE) {
+    } else if (dtype == TFloat64.class) {
       if (debug) {
-        try (Tensor<TFloat64> result =
-            this.getGraphSession().runner().fetch(input).run().get(0).expect(TFloat64.DTYPE)) {
+        try (TFloat64 result =
+            this.getGraphSession().runner().fetch(input).run().get(0)) {
           if (isScalar) {
             System.out.printf(
                 "0). %b <==> %f\n",
-                predicate.test(result.data().getDouble()), result.data().getDouble());
+                predicate.test(result.getDouble()), result.getDouble());
           } else {
             result
-                .data()
                 .scalars()
                 .forEachIndexed(
                     (idx, f) ->
@@ -908,27 +856,25 @@ public <T extends TType> void evaluate(Output<T> input, Predicate<Number> predic
         }
       }
       index.set(0);
-      try (Tensor<TFloat64> result =
-          this.getGraphSession().runner().fetch(input).run().get(0).expect(TFloat64.DTYPE)) {
+      try (TFloat64 result =
+          this.getGraphSession().runner().fetch(input).run().get(0)) {
         if (isScalar) {
-          assertTrue(predicate.test(result.data().getDouble()));
+          assertTrue(predicate.test(result.getDouble()));
         } else {
           result
-              .data()
               .scalars()
-              .forEachIndexed((idx, f) -> assertTrue(predicate.test(result.data().getDouble())));
+              .forEachIndexed((idx, f) -> assertTrue(predicate.test(result.getDouble())));
         }
       }
-    } else if (dtype == TInt32.DTYPE) {
+    } else if (dtype == TInt32.class) {
       if (debug) {
-        try (Tensor<TInt32> result =
-            this.getGraphSession().runner().fetch(input).run().get(0).expect(TInt32.DTYPE)) {
+        try (TInt32 result =
+            this.getGraphSession().runner().fetch(input).run().get(0)) {
           if (isScalar) {
             System.out.printf(
-                "0). %b <==> %d\n", predicate.test(result.data().getInt()), result.data().getInt());
+                "0). %b <==> %d\n", predicate.test(result.getInt()), result.getInt());
           } else {
             result
-                .data()
                 .scalars()
                 .forEachIndexed(
                     (idx, f) ->
@@ -939,28 +885,26 @@ public <T extends TType> void evaluate(Output<T> input, Predicate<Number> predic
         }
       }
       index.set(0);
-      try (Tensor<TInt32> result =
-          this.getGraphSession().runner().fetch(input).run().get(0).expect(TInt32.DTYPE)) {
+      try (TInt32 result =
+          this.getGraphSession().runner().fetch(input).run().get(0)) {
         if (isScalar) {
-          assertTrue(predicate.test(result.data().getInt()));
+          assertTrue(predicate.test(result.getInt()));
         } else {
           result
-              .data()
               .scalars()
-              .forEachIndexed((idx, f) -> assertTrue(predicate.test(result.data().getInt())));
+              .forEachIndexed((idx, f) -> assertTrue(predicate.test(result.getInt())));
         }
       }
-    } else if (dtype == TInt64.DTYPE) {
+    } else if (dtype == TInt64.class) {
       if (debug) {
-        try (Tensor<TInt64> result =
-            this.getGraphSession().runner().fetch(input).run().get(0).expect(TInt64.DTYPE)) {
+        try (TInt64 result =
+            this.getGraphSession().runner().fetch(input).run().get(0)) {
           if (isScalar) {
             System.out.printf(
                 "0). %b <==> %d\n",
-                predicate.test(result.data().getLong()), result.data().getLong());
+                predicate.test(result.getLong()), result.getLong());
           } else {
             result
-                .data()
                 .scalars()
                 .forEachIndexed(
                     (idx, f) ->
@@ -971,28 +915,26 @@ public <T extends TType> void evaluate(Output<T> input, Predicate<Number> predic
         }
       }
       index.set(0);
-      try (Tensor<TInt64> result =
-          this.getGraphSession().runner().fetch(input).run().get(0).expect(TInt64.DTYPE)) {
+      try (TInt64 result =
+          this.getGraphSession().runner().fetch(input).run().get(0)) {
         if (isScalar) {
-          assertTrue(predicate.test(result.data().getLong()));
+          assertTrue(predicate.test(result.getLong()));
         } else {
           result
-              .data()
               .scalars()
-              .forEachIndexed((idx, f) -> assertTrue(predicate.test(result.data().getLong())));
+              .forEachIndexed((idx, f) -> assertTrue(predicate.test(result.getLong())));
         }
       }
-    } else if (dtype == TUint8.DTYPE) {
+    } else if (dtype == TUint8.class) {
       if (debug) {
-        try (Tensor<TUint8> result =
-            this.getGraphSession().runner().fetch(input).run().get(0).expect(TUint8.DTYPE)) {
+        try (TUint8 result =
+            this.getGraphSession().runner().fetch(input).run().get(0)) {
           if (isScalar) {
             System.out.printf(
                 "0). %b <==> %d\n",
-                predicate.test(result.data().getByte()), result.data().getByte());
+                predicate.test(result.getByte()), result.getByte());
           } else {
             result
-                .data()
                 .scalars()
                 .forEachIndexed(
                     (idx, f) ->
@@ -1003,19 +945,18 @@ public <T extends TType> void evaluate(Output<T> input, Predicate<Number> predic
         }
       }
       index.set(0);
-      try (Tensor<TUint8> result =
-          this.getGraphSession().runner().fetch(input).run().get(0).expect(TUint8.DTYPE)) {
+      try (TUint8 result =
+          this.getGraphSession().runner().fetch(input).run().get(0)) {
         if (isScalar) {
-          assertTrue(predicate.test(result.data().getByte()));
+          assertTrue(predicate.test(result.getByte()));
         } else {
           result
-              .data()
               .scalars()
-              .forEachIndexed((idx, f) -> assertTrue(predicate.test(result.data().getByte())));
+              .forEachIndexed((idx, f) -> assertTrue(predicate.test(result.getByte())));
         }
       }
     } else {
-      fail("Unexpected DataType: " + dtype);
+      fail("Unexpected Class: " + dtype);
     }
   }
 
@@ -1024,119 +965,112 @@ public <T extends TType> void evaluate(Output<T> input, Predicate<Number> predic
   public <T extends TType> void print(PrintWriter writer, Output<T> input) {
     boolean isScalar = input.asOutput().shape().size() == 1;
 
-    DataType<T> dtype = input.dataType();
-    if (dtype == TFloat32.DTYPE) {
+    Class<T> dtype = input.type();
+    if (dtype == TFloat32.class) {
       AtomicInteger index = new AtomicInteger();
-      try (Tensor<TFloat32> result =
-          this.getGraphSession().runner().fetch(input).run().get(0).expect(TFloat32.DTYPE)) {
+      try (TFloat32 result =
+          this.getGraphSession().runner().fetch(input).run().get(0)) {
         if (isScalar) {
-          writer.printf("%d). %f\n", index.getAndIncrement(), result.data().getFloat());
+          writer.printf("%d). %f\n", index.getAndIncrement(), result.getFloat());
         } else {
           result
-              .data()
               .scalars()
               .forEachIndexed(
                   (idx, f) -> writer.printf("%d). %f\n", index.getAndIncrement(), f.getFloat()));
         }
       }
-    } else if (dtype == TFloat64.DTYPE) {
+    } else if (dtype == TFloat64.class) {
       AtomicInteger index = new AtomicInteger();
 
-      try (Tensor<TFloat64> result =
-          this.getGraphSession().runner().fetch(input).run().get(0).expect(TFloat64.DTYPE)) {
+      try (TFloat64 result =
+          this.getGraphSession().runner().fetch(input).run().get(0)) {
         if (isScalar) {
           writer.printf(
-              "%d). %f\n", index.getAndIncrement(), ((Output<TFloat64>) input).data().getDouble());
+              "%d). %f\n", index.getAndIncrement(), ((Output<TFloat64>) input).asTensor().getDouble());
         } else {
           result
-              .data()
               .scalars()
               .forEachIndexed(
                   (idx, f) -> writer.printf("%d). %f\n", index.getAndIncrement(), f.getDouble()));
         }
       }
-    } else if (dtype == TInt32.DTYPE) {
+    } else if (dtype == TInt32.class) {
       AtomicInteger index = new AtomicInteger();
 
-      try (Tensor<TInt32> result =
-          this.getGraphSession().runner().fetch(input).run().get(0).expect(TInt32.DTYPE)) {
+      try (TInt32 result =
+          this.getGraphSession().runner().fetch(input).run().get(0)) {
         if (isScalar) {
           writer.printf(
-              "%d). %d\n", index.getAndIncrement(), ((Output<TInt32>) input).data().getInt());
+              "%d). %d\n", index.getAndIncrement(), ((Output<TInt32>) input).asTensor().getInt());
         } else {
           result
-              .data()
               .scalars()
               .forEachIndexed(
                   (idx, f) -> writer.printf("%d). %d\n", index.getAndIncrement(), f.getInt()));
         }
       }
-    } else if (dtype == TInt64.DTYPE) {
+    } else if (dtype == TInt64.class) {
       AtomicInteger index = new AtomicInteger();
 
-      try (Tensor<TInt64> result =
-          this.getGraphSession().runner().fetch(input).run().get(0).expect(TInt64.DTYPE)) {
+      try (TInt64 result =
+          this.getGraphSession().runner().fetch(input).run().get(0)) {
         if (isScalar) {
           writer.printf(
-              "%d). %d\n", index.getAndIncrement(), ((Output<TInt64>) input).data().getLong());
+              "%d). %d\n", index.getAndIncrement(), ((Output<TInt64>) input).asTensor().getLong());
         } else {
           result
-              .data()
               .scalars()
               .forEachIndexed(
                   (idx, f) -> writer.printf("%d). %d\n", index.getAndIncrement(), f.getLong()));
         }
       }
-    } else if (dtype == TUint8.DTYPE) {
+    } else if (dtype == TUint8.class) {
       AtomicInteger index = new AtomicInteger();
 
-      try (Tensor<TUint8> result =
-          this.getGraphSession().runner().fetch(input).run().get(0).expect(TUint8.DTYPE)) {
+      try (TUint8 result =
+          this.getGraphSession().runner().fetch(input).run().get(0)) {
         if (isScalar) {
           writer.printf(
-              "%d). %x\n", index.getAndIncrement(), ((Output<TUint8>) input).data().getByte());
+              "%d). %x\n", index.getAndIncrement(), ((Output<TUint8>) input).asTensor().getByte());
         } else {
           result
-              .data()
               .scalars()
               .forEachIndexed(
                   (idx, f) -> writer.printf("%d). %x\n", index.getAndIncrement(), f.getByte()));
         }
       }
-    } else if (dtype == TBool.DTYPE) {
+    } else if (dtype == TBool.class) {
       AtomicInteger index = new AtomicInteger();
 
-      try (Tensor<TBool> result =
-          this.getGraphSession().runner().fetch(input).run().get(0).expect(TBool.DTYPE)) {
+      try (TBool result =
+          this.getGraphSession().runner().fetch(input).run().get(0)) {
         if (isScalar) {
           writer.printf(
-              "%d). %b\n", index.getAndIncrement(), ((Output<TBool>) input).data().getBoolean());
+              "%d). %b\n", index.getAndIncrement(), ((Output<TBool>) input).asTensor().getBoolean());
         } else {
           result
-              .data()
               .scalars()
               .forEachIndexed(
                   (idx, f) -> writer.printf("%d). %b\n", index.getAndIncrement(), f.getBoolean()));
         }
       }
-    } else if (dtype == TString.DTYPE) {
+    } else if (dtype == TString.class) {
       AtomicInteger index = new AtomicInteger();
 
-      try (Tensor<TString> result =
-          this.getGraphSession().runner().fetch(input).run().get(0).expect(TString.DTYPE)) {
+      try (TString result =
+          this.getGraphSession().runner().fetch(input).run().get(0)) {
         if (isScalar) {
           writer.printf(
-              "%d). %s\n", index.getAndIncrement(), ((Output<TString>) input).data().getObject());
+              "%d). %s\n", index.getAndIncrement(), ((Output<TString>) input).asTensor().getObject());
         } else {
           result
-              .data()
               .scalars()
               .forEachIndexed(
                   (idx, f) -> writer.printf("%d). %s\n", index.getAndIncrement(), f.getObject()));
         }
       }
     } else {
-      writer.println("Unexpected DataType: " + dtype);
+      writer.println("Unexpected Class: " + dtype);
     }
     writer.flush();
   }

From 9473b6c9faa938e25863398b308b41ed65386fcd Mon Sep 17 00:00:00 2001
From: klessard <klessard@expedia.com>
Date: Mon, 2 Nov 2020 13:38:12 -0500
Subject: [PATCH 09/11] Rollback autoconversion of tensor to constant for now

---
 .../src/bazel/op_generator/op_generator.cc     |  5 ++---
 .../src/main/java/org/tensorflow/Operand.java  | 18 +-----------------
 .../src/main/java/org/tensorflow/Output.java   |  7 ++++---
 .../main/java/org/tensorflow/op/Operands.java  |  9 ++-------
 .../java/org/tensorflow/op/core/Constant.java  |  4 ++--
 .../java/org/tensorflow/op/core/Gradients.java |  6 +++---
 .../java/org/tensorflow/op/core/Zeros.java     |  2 +-
 .../org/tensorflow/types/TypeRegistry.java     | 12 ++++++------
 .../java/org/tensorflow/op/OperandsTest.java   |  2 +-
 .../tensorflow/types/NumericTypesTestBase.java |  6 +++---
 10 files changed, 25 insertions(+), 46 deletions(-)

diff --git a/tensorflow-core/tensorflow-core-api/src/bazel/op_generator/op_generator.cc b/tensorflow-core/tensorflow-core-api/src/bazel/op_generator/op_generator.cc
index a3b76b71e70..2bec80a8457 100644
--- a/tensorflow-core/tensorflow-core-api/src/bazel/op_generator/op_generator.cc
+++ b/tensorflow-core/tensorflow-core-api/src/bazel/op_generator/op_generator.cc
@@ -237,12 +237,12 @@ void RenderFactoryMethods(const OpSpec& op, const Type& op_class,
   writer->EndLine();
   for (const ArgumentSpec& input : op.inputs()) {
     if (input.iterable()) {
-      writer->Append("opBuilder.addInputList(Operands.asOutputs(scope, " +
+      writer->Append("opBuilder.addInputList(Operands.asOutputs(" +
                      input.var().name() + "));");
       writer->EndLine();
     } else {
       writer->Append("opBuilder.addInput(" + input.var().name() +
-                     ".asOutput(scope));");
+                     ".asOutput());");
       writer->EndLine();
     }
   }
@@ -350,7 +350,6 @@ void RenderInterfaceImpl(const OpSpec& op, RenderMode mode,
     Type return_type = Type::Class("Output", "org.tensorflow")
         .add_parameter(cast2obj ? Type::Class("TType", "org.tensorflow.types.family") : output.type());
     Method as_output = Method::Create("asOutput", return_type)
-        .add_argument(Variable::Create("scope", Type::Class("Scope", "org.tensorflow.op")))
         .add_annotation(Annotation::Create("Override"));
     if (cast2obj) {
       as_output.add_annotation(
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Operand.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Operand.java
index 8f7e7791ebf..fcb630484d0 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Operand.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Operand.java
@@ -54,20 +54,6 @@ default Class<T> type() {
     return asOutput().type();
   }
 
-  /**
-   * Returns the symbolic handle of the tensor.
-   *
-   * <p>Inputs to TensorFlow operations are outputs of another TensorFlow operation. This method is
-   * used to obtain a symbolic handle that represents the computation of the input.
-   *
-   * <p>If a valid non-null value is provided as the {@code scope} of this operand, the implementor
-   * can program additional computations before returning its output. This might be required if,
-   * for instance, the implementor has not yet been added as a node to the graph.</p>
-   *
-   * @param scope if non-null, scope that can be used by the operand to produce a new output
-   * @see OperationBuilder#addInput(Output)
-   */
-  Output<T> asOutput(Scope scope);
 
   /**
    * Returns the symbolic handle of the tensor.
@@ -77,9 +63,7 @@ default Class<T> type() {
    *
    * @see OperationBuilder#addInput(Output)
    */
-  default Output<T> asOutput() {
-    return asOutput(null);
-  }
+  Output<T> asOutput();
 
   /**
    * Returns this operand as a tensor.
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Output.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Output.java
index 692daaf6327..7397d676393 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Output.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/Output.java
@@ -19,6 +19,7 @@
 import org.bytedeco.javacpp.Pointer;
 import org.tensorflow.ndarray.Shape;
 import org.tensorflow.op.Scope;
+import org.tensorflow.types.Type;
 import org.tensorflow.types.TypeRegistry;
 import org.tensorflow.types.family.TType;
 
@@ -47,7 +48,7 @@ public Shape shape() {
   /** Returns the DataType of the tensor referred to by this Output. */
   @Override
   public Class<T> type() {
-    return (Class) TypeRegistry.find(operation.dtype(index)).typeClass();
+    return ((Type<T>)TypeRegistry.find(operation.dtype(index))).typeClass();
   }
 
   /**
@@ -60,7 +61,7 @@ public Class<T> type() {
    */
   @SuppressWarnings("unchecked")
   public <U extends TType> Output<U> expect(Class<U> tensorType) {
-    if (tensorType == type()) {
+    if (tensorType != type()) {
       throw new IllegalArgumentException(
           "Cannot cast from output of " + type().getSimpleName() + " to output of type " + tensorType.getSimpleName());
     }
@@ -93,7 +94,7 @@ public Operation op() {
   }
 
   @Override
-  public Output<T> asOutput(Scope scope) {
+  public Output<T> asOutput() {
     return this;
   }
 
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/op/Operands.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/op/Operands.java
index 5496eaa3d15..ac48da80326 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/op/Operands.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/op/Operands.java
@@ -30,18 +30,13 @@ public final class Operands {
    * <p>Operation wrappers need to convert back a list of inputs into an array of outputs in order
    * to build an operation, see {@link OperationBuilder#addInputList(Output[])}.
    *
-   * <p>If a valid non-null value is provided as the {@code scope} of this operand, the implementor
-   * can program additional computations before returning its output. This might be required if,
-   * for instance, the implementor has not yet been added as a node to the graph.</p>
-   *
-   * @param scope if non-null, scope that can be used an operand to produce a new output
    * @param inputs an iteration of input operands
    * @return an array of outputs
    */
-  public static Output<?>[] asOutputs(Scope scope, Iterable<? extends Operand<?>> inputs) {
+  public static Output<?>[] asOutputs(Iterable<? extends Operand<?>> inputs) {
     List<Output<?>> outputList = new ArrayList<>();
     for (Operand<?> input : inputs) {
-      outputList.add(input.asOutput(scope));
+      outputList.add(input.asOutput());
     }
     return outputList.toArray(new Output<?>[outputList.size()]);
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/op/core/Constant.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/op/core/Constant.java
index 07aebcb5bf3..3c2d34d3980 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/op/core/Constant.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/op/core/Constant.java
@@ -1262,7 +1262,7 @@ public static Constant<TInt64> tensorOf(Scope scope, Shape shape) {
    * @param tensor a Tensor holding the constant value
    * @return a constant of the same data type as `tensor`
    */
-  @Endpoint(name = "capture")
+  @Endpoint
   public static <T extends TType> Constant<T> create(Scope scope, T tensor) {
     return new Constant<>(
         scope
@@ -1274,7 +1274,7 @@ public static <T extends TType> Constant<T> create(Scope scope, T tensor) {
   }
 
   @Override
-  public Output<T> asOutput(Scope scope) {
+  public Output<T> asOutput() {
     return output;
   }
 
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/op/core/Gradients.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/op/core/Gradients.java
index 88b10be5017..45ac0d1180e 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/op/core/Gradients.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/op/core/Gradients.java
@@ -96,14 +96,14 @@ public static Gradients create(
     if (options != null) {
       for (Options opts : options) {
         if (opts.dx != null) {
-          dx = Operands.asOutputs(scope, opts.dx);
+          dx = Operands.asOutputs(opts.dx);
         }
       }
     }
     Output<?>[] dy = graph.addGradients(
         scope.makeOpName("Gradients"),
-        Operands.asOutputs(scope, y),
-        Operands.asOutputs(scope, x), dx
+        Operands.asOutputs(y),
+        Operands.asOutputs(x), dx
     );
     return new Gradients(Arrays.asList(dy));
   }
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/op/core/Zeros.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/op/core/Zeros.java
index 7a34d14d541..3de8b1d26e8 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/op/core/Zeros.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/op/core/Zeros.java
@@ -67,7 +67,7 @@ public Operation op() {
   }
 
   @Override
-  public Output<T> asOutput(Scope scope) {
+  public Output<T> asOutput() {
     return fill.asOutput();
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TypeRegistry.java b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TypeRegistry.java
index 1e82478edf1..99e9170fa55 100644
--- a/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TypeRegistry.java
+++ b/tensorflow-core/tensorflow-core-api/src/main/java/org/tensorflow/types/TypeRegistry.java
@@ -36,11 +36,11 @@ public final class TypeRegistry {
    * @throws IllegalArgumentException if the code matches no registered data type
    */
   public static <T extends TType> Type<T> find(DataType dataType) {
-    Type entry = TYPES_BY_CODE.get(dataType.getNumber());
+    Type<?> entry = TYPES_BY_CODE.get(dataType.getNumber());
     if (entry == null) {
       throw new IllegalArgumentException("No type has been registered for datatype " + dataType);
     }
-    return entry;
+    return (Type<T>)entry;
   }
 
   /**
@@ -51,15 +51,15 @@ public static <T extends TType> Type<T> find(DataType dataType) {
    * @throws IllegalArgumentException if the code matches no registered data type
    */
   public static <T extends TType> Type<T> find(Class<T> typeClass) {
-    Type entry = TYPES_BY_CLASS.get(typeClass);
+    Type<?> entry = TYPES_BY_CLASS.get(typeClass);
     if (entry == null) {
       throw new IllegalArgumentException("Class \"" + typeClass.getName() + "\" is not a valid datatype class");
     }
-    return entry;
+    return (Type<T>)entry;
   }
 
-  private static final Map<Integer, Type> TYPES_BY_CODE = new HashMap<>();
-  private static final Map<Class<? extends TType>, Type> TYPES_BY_CLASS = new HashMap<>();
+  private static final Map<Integer, Type<?>> TYPES_BY_CODE = new HashMap<>();
+  private static final Map<Class<? extends TType>, Type<?>> TYPES_BY_CLASS = new HashMap<>();
 
   private static <T extends TType> void register(Class<T> typeClass) {
     TensorType typeAnnot = typeClass.getDeclaredAnnotation(TensorType.class);
diff --git a/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/op/OperandsTest.java b/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/op/OperandsTest.java
index f2c27c4cdf0..bc3e418e6f4 100644
--- a/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/op/OperandsTest.java
+++ b/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/op/OperandsTest.java
@@ -31,7 +31,7 @@ public void createOutputArrayFromOperandList() {
     try (Graph g = new Graph()) {
       Ops tf = Ops.create(g);
       Split<?> split = tf.split(tf.constant(0), tf.array(0, 1, 2), 3L);
-      Output<?>[] array = Operands.asOutputs(tf.scope(), split.output());
+      Output<?>[] array = Operands.asOutputs(split.output());
       assertEquals(split.output().size(), array.length);
       assertSame(array[0], split.output().get(0));
       assertSame(array[1], split.output().get(1));
diff --git a/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/types/NumericTypesTestBase.java b/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/types/NumericTypesTestBase.java
index 24390ae0a91..398b1693e14 100644
--- a/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/types/NumericTypesTestBase.java
+++ b/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/types/NumericTypesTestBase.java
@@ -46,11 +46,11 @@ public void initializeTensorsWithZeros() {
 
       // Initialize tensor memory with zeros and take a snapshot
       data.scalars().forEach(scalar -> ((NdArray<U>)scalar).setObject(valueOf(0)));
-      Constant<T> x = tf.capture(tensor);
+      Constant<T> x = tf.constant(tensor);
 
       // Initialize the same tensor memory with ones and take a snapshot
       data.scalars().forEach(scalar -> ((NdArray<U>)scalar).setObject(valueOf(1)));
-      Constant<T> y = tf.capture(tensor);
+      Constant<T> y = tf.constant(tensor);
 
       // Subtract y from x and validate the result
       Sub<T> sub = tf.math.sub(x, y);
@@ -94,7 +94,7 @@ public void setAndCompute() {
       try (EagerSession session = EagerSession.create()) {
         Ops tf = Ops.create(session);
 
-        Add<T> add = tf.math.add(tf.capture(tensor), tf.capture(tensor));
+        Add<T> add = tf.math.add(tf.constant(tensor), tf.constant(tensor));
         NdArray<U> result = (NdArray<U>)add.asTensor();
 
         assertEquals(valueOf(0), result.getObject(0, 0));

From 7897b2578114901257064af88ce3be10bcfeb418 Mon Sep 17 00:00:00 2001
From: klessard <klessard@expedia.com>
Date: Tue, 3 Nov 2020 09:14:36 -0500
Subject: [PATCH 10/11] Add missing generated sources

---
 .../org/tensorflow/op/AudioOps.java           |  147 +
 .../org/tensorflow/op/BitwiseOps.java         |  271 +
 .../tensorflow/op/DataExperimentalOps.java    |   61 +
 .../org/tensorflow/op/DataOps.java            |  413 +
 .../org/tensorflow/op/DtypesOps.java          |  104 +
 .../org/tensorflow/op/ImageOps.java           |  944 ++
 .../annotations/org/tensorflow/op/IoOps.java  | 1037 +++
 .../org/tensorflow/op/LinalgOps.java          | 1608 ++++
 .../org/tensorflow/op/MathOps.java            | 2393 +++++
 .../annotations/org/tensorflow/op/NnOps.java  |   18 +-
 .../org/tensorflow/op/NnRawOps.java           |   74 +
 .../annotations/org/tensorflow/op/Ops.java    | 7728 +++++++++++++++++
 .../org/tensorflow/op/QuantizationOps.java    |  629 ++
 .../org/tensorflow/op/RaggedOps.java          |   63 +
 .../org/tensorflow/op/RandomOps.java          |  585 ++
 .../org/tensorflow/op/ShapeOps.java           |  471 +
 .../org/tensorflow/op/SignalOps.java          |  436 +
 .../org/tensorflow/op/SparseOps.java          | 1499 ++++
 .../org/tensorflow/op/StringsOps.java         |  618 ++
 .../org/tensorflow/op/SummaryOps.java         |  198 +
 .../org/tensorflow/op/TrainOps.java           | 1660 ++++
 .../annotations/org/tensorflow/op/XlaOps.java |  381 +
 ...mpute_func_Pointer_TF_OpKernelContext.java |   19 +
 .../Create_func_TF_OpKernelConstruction.java  |   48 +
 .../Deallocator_Pointer_long_Pointer.java     |   30 +
 .../internal/c_api/Delete_func_Pointer.java   |   19 +
 .../internal/c_api/Listener_BytePointer.java  |   25 +
 .../internal/c_api/Listener_String.java       |   19 +
 ...nc_TF_ShapeInferenceContext_TF_Status.java |   22 +
 .../internal/c_api/TFE_Context.java           |   25 +
 .../internal/c_api/TFE_ContextOptions.java    |   19 +
 .../org/tensorflow/internal/c_api/TFE_Op.java |   29 +
 .../internal/c_api/TFE_TensorDebugInfo.java   |   22 +
 .../internal/c_api/TFE_TensorHandle.java      |   23 +
 .../internal/c_api/TF_ApiDefMap.java          |   26 +
 .../internal/c_api/TF_AttrMetadata.java       |   58 +
 .../tensorflow/internal/c_api/TF_Buffer.java  |   49 +
 .../internal/c_api/TF_DeprecatedSession.java  |   23 +
 .../internal/c_api/TF_DeviceList.java         |   18 +
 .../internal/c_api/TF_DimensionHandle.java    |   19 +
 .../internal/c_api/TF_Function.java           |   21 +
 .../internal/c_api/TF_FunctionOptions.java    |   19 +
 .../tensorflow/internal/c_api/TF_Graph.java   |   26 +
 .../c_api/TF_ImportGraphDefOptions.java       |   20 +
 .../c_api/TF_ImportGraphDefResults.java       |   20 +
 .../tensorflow/internal/c_api/TF_Input.java   |   33 +
 .../internal/c_api/TF_KernelBuilder.java      |   31 +
 .../tensorflow/internal/c_api/TF_Library.java |   22 +
 .../c_api/TF_OpDefinitionBuilder.java         |   17 +
 .../c_api/TF_OpKernelConstruction.java        |   17 +
 .../internal/c_api/TF_OpKernelContext.java    |   17 +
 .../internal/c_api/TF_Operation.java          |   21 +
 .../c_api/TF_OperationDescription.java        |   19 +
 .../tensorflow/internal/c_api/TF_Output.java  |   33 +
 .../tensorflow/internal/c_api/TF_Server.java  |   27 +
 .../tensorflow/internal/c_api/TF_Session.java |   24 +
 .../internal/c_api/TF_SessionOptions.java     |   20 +
 .../internal/c_api/TF_ShapeHandle.java        |   17 +
 .../c_api/TF_ShapeInferenceContext.java       |   17 +
 .../tensorflow/internal/c_api/TF_Status.java  |   19 +
 .../tensorflow/internal/c_api/TF_Tensor.java  |   36 +
 .../internal/c_api/TF_WhileParams.java        |   37 +
 .../org/tensorflow/internal/c_api/Tensor.java |   25 +
 .../internal/c_api/global/tensorflow.java     | 4021 +++++++++
 .../tensorflow/op/audio/AudioSpectrogram.java |  140 +
 .../org/tensorflow/op/audio/DecodeWav.java    |  147 +
 .../org/tensorflow/op/audio/EncodeWav.java    |   85 +
 .../java/org/tensorflow/op/audio/Mfcc.java    |  178 +
 .../org/tensorflow/op/bitwise/BitwiseAnd.java |   96 +
 .../org/tensorflow/op/bitwise/BitwiseOr.java  |   96 +
 .../org/tensorflow/op/bitwise/BitwiseXor.java |   96 +
 .../org/tensorflow/op/bitwise/Invert.java     |  115 +
 .../org/tensorflow/op/bitwise/LeftShift.java  |  107 +
 .../org/tensorflow/op/bitwise/RightShift.java |  110 +
 .../op/cluster/KMC2ChainInitialization.java   |   81 +
 .../cluster/KmeansPlusPlusInitialization.java |   87 +
 .../tensorflow/op/collective/AllReduce.java   |  166 +
 .../op/collective/BroadcastRecv.java          |  135 +
 .../op/collective/BroadcastSend.java          |  135 +
 .../java/org/tensorflow/op/core/Abort.java    |  110 +
 .../gen/java/org/tensorflow/op/core/All.java  |  116 +
 .../gen/java/org/tensorflow/op/core/Any.java  |  116 +
 .../org/tensorflow/op/core/AssertThat.java    |   96 +
 .../java/org/tensorflow/op/core/Assign.java   |  140 +
 .../org/tensorflow/op/core/AssignAdd.java     |  117 +
 .../op/core/AssignAddVariableOp.java          |   61 +
 .../org/tensorflow/op/core/AssignSub.java     |  117 +
 .../op/core/AssignSubVariableOp.java          |   61 +
 .../tensorflow/op/core/AssignVariableOp.java  |   61 +
 .../java/org/tensorflow/op/core/Barrier.java  |  194 +
 .../org/tensorflow/op/core/BarrierClose.java  |  101 +
 .../op/core/BarrierIncompleteSize.java        |   75 +
 .../tensorflow/op/core/BarrierInsertMany.java |   69 +
 .../tensorflow/op/core/BarrierReadySize.java  |   75 +
 .../tensorflow/op/core/BarrierTakeMany.java   |  191 +
 .../java/org/tensorflow/op/core/Batch.java    |  247 +
 .../org/tensorflow/op/core/BatchToSpace.java  |  147 +
 .../tensorflow/op/core/BatchToSpaceNd.java    |  181 +
 .../java/org/tensorflow/op/core/Bitcast.java  |  127 +
 .../op/core/BroadcastDynamicShape.java        |   79 +
 .../op/core/BroadcastGradientArgs.java        |   80 +
 .../org/tensorflow/op/core/BroadcastTo.java   |  106 +
 .../org/tensorflow/op/core/Bucketize.java     |   96 +
 .../org/tensorflow/op/core/ClipByValue.java   |   86 +
 .../tensorflow/op/core/CollectiveGather.java  |  135 +
 .../java/org/tensorflow/op/core/Concat.java   |   83 +
 .../tensorflow/op/core/ConsumeMutexLock.java  |   63 +
 .../tensorflow/op/core/ControlTrigger.java    |   54 +
 .../gen/java/org/tensorflow/op/core/Copy.java |  148 +
 .../java/org/tensorflow/op/core/CopyHost.java |  146 +
 .../org/tensorflow/op/core/CountUpTo.java     |   79 +
 .../org/tensorflow/op/core/DecodeProto.java   |  223 +
 .../java/org/tensorflow/op/core/DeepCopy.java |   76 +
 .../op/core/DeleteSessionTensor.java          |   56 +
 .../tensorflow/op/core/DestroyResourceOp.java |   94 +
 .../op/core/DestroyTemporaryVariable.java     |   85 +
 .../org/tensorflow/op/core/DeviceIndex.java   |   81 +
 .../tensorflow/op/core/DummyMemoryCache.java  |   69 +
 .../tensorflow/op/core/DynamicPartition.java  |  120 +
 .../org/tensorflow/op/core/DynamicStitch.java |  133 +
 .../org/tensorflow/op/core/EditDistance.java  |  164 +
 .../java/org/tensorflow/op/core/Empty.java    |  114 +
 .../tensorflow/op/core/EmptyTensorList.java   |   86 +
 .../org/tensorflow/op/core/EncodeProto.java   |  160 +
 .../org/tensorflow/op/core/EnsureShape.java   |   81 +
 .../java/org/tensorflow/op/core/Enter.java    |  135 +
 .../gen/java/org/tensorflow/op/core/Exit.java |   76 +
 .../org/tensorflow/op/core/ExpandDims.java    |  110 +
 .../op/core/ExtractVolumePatches.java         |  102 +
 .../gen/java/org/tensorflow/op/core/Fill.java |  104 +
 .../org/tensorflow/op/core/Fingerprint.java   |  109 +
 .../java/org/tensorflow/op/core/Gather.java   |  143 +
 .../java/org/tensorflow/op/core/GatherNd.java |  172 +
 .../tensorflow/op/core/GetSessionHandle.java  |   75 +
 .../tensorflow/op/core/GetSessionTensor.java  |   78 +
 .../tensorflow/op/core/GuaranteeConst.java    |   81 +
 .../org/tensorflow/op/core/HashTable.java     |  158 +
 .../op/core/HistogramFixedWidth.java          |  116 +
 .../java/org/tensorflow/op/core/Identity.java |   74 +
 .../org/tensorflow/op/core/IdentityN.java     |   95 +
 .../tensorflow/op/core/ImmutableConst.java    |   82 +
 .../tensorflow/op/core/InitializeTable.java   |   60 +
 .../op/core/InitializeTableFromTextFile.java  |  127 +
 .../org/tensorflow/op/core/InplaceAdd.java    |   82 +
 .../org/tensorflow/op/core/InplaceSub.java    |   82 +
 .../org/tensorflow/op/core/InplaceUpdate.java |   85 +
 .../op/core/IsVariableInitialized.java        |   75 +
 .../java/org/tensorflow/op/core/LinSpace.java |   88 +
 .../tensorflow/op/core/LookupTableExport.java |   84 +
 .../tensorflow/op/core/LookupTableFind.java   |   86 +
 .../tensorflow/op/core/LookupTableImport.java |   63 +
 .../tensorflow/op/core/LookupTableInsert.java |   63 +
 .../tensorflow/op/core/LookupTableRemove.java |   60 +
 .../tensorflow/op/core/LookupTableSize.java   |   73 +
 .../java/org/tensorflow/op/core/LoopCond.java |   76 +
 .../org/tensorflow/op/core/LowerBound.java    |  115 +
 .../java/org/tensorflow/op/core/MapClear.java |  151 +
 .../tensorflow/op/core/MapIncompleteSize.java |  169 +
 .../java/org/tensorflow/op/core/MapPeek.java  |  182 +
 .../java/org/tensorflow/op/core/MapSize.java  |  169 +
 .../java/org/tensorflow/op/core/MapStage.java |  166 +
 .../org/tensorflow/op/core/MapUnstage.java    |  182 +
 .../tensorflow/op/core/MapUnstageNoKey.java   |  181 +
 .../gen/java/org/tensorflow/op/core/Max.java  |  118 +
 .../java/org/tensorflow/op/core/Merge.java    |   87 +
 .../gen/java/org/tensorflow/op/core/Min.java  |  118 +
 .../org/tensorflow/op/core/MirrorPad.java     |  111 +
 .../org/tensorflow/op/core/MirrorPadGrad.java |   99 +
 .../tensorflow/op/core/MlirPassthroughOp.java |  117 +
 .../op/core/MutableDenseHashTable.java        |  224 +
 .../tensorflow/op/core/MutableHashTable.java  |  158 +
 .../op/core/MutableHashTableOfTensors.java    |  176 +
 .../java/org/tensorflow/op/core/Mutex.java    |  129 +
 .../org/tensorflow/op/core/MutexLock.java     |  112 +
 .../org/tensorflow/op/core/NcclAllReduce.java |   92 +
 .../org/tensorflow/op/core/NcclBroadcast.java |   86 +
 .../org/tensorflow/op/core/NcclReduce.java    |   85 +
 .../org/tensorflow/op/core/NextIteration.java |   75 +
 .../gen/java/org/tensorflow/op/core/NoOp.java |   52 +
 .../java/org/tensorflow/op/core/OneHot.java   |  198 +
 .../java/org/tensorflow/op/core/OnesLike.java |   75 +
 .../tensorflow/op/core/OrderedMapClear.java   |  151 +
 .../op/core/OrderedMapIncompleteSize.java     |  169 +
 .../tensorflow/op/core/OrderedMapPeek.java    |  183 +
 .../tensorflow/op/core/OrderedMapSize.java    |  169 +
 .../tensorflow/op/core/OrderedMapStage.java   |  168 +
 .../tensorflow/op/core/OrderedMapUnstage.java |  182 +
 .../op/core/OrderedMapUnstageNoKey.java       |  181 +
 .../gen/java/org/tensorflow/op/core/Pad.java  |  104 +
 .../tensorflow/op/core/ParallelConcat.java    |   96 +
 .../op/core/ParallelDynamicStitch.java        |  132 +
 .../org/tensorflow/op/core/Placeholder.java   |  116 +
 .../op/core/PlaceholderWithDefault.java       |   78 +
 .../java/org/tensorflow/op/core/Print.java    |  111 +
 .../gen/java/org/tensorflow/op/core/Prod.java |  118 +
 .../tensorflow/op/core/QuantizedReshape.java  |   97 +
 .../java/org/tensorflow/op/core/Range.java    |   91 +
 .../gen/java/org/tensorflow/op/core/Rank.java |   85 +
 .../tensorflow/op/core/ReadVariableOp.java    |   83 +
 .../gen/java/org/tensorflow/op/core/Recv.java |  122 +
 .../org/tensorflow/op/core/ReduceAll.java     |  116 +
 .../org/tensorflow/op/core/ReduceAny.java     |  116 +
 .../org/tensorflow/op/core/ReduceMax.java     |  118 +
 .../org/tensorflow/op/core/ReduceMin.java     |  118 +
 .../org/tensorflow/op/core/ReduceProd.java    |  118 +
 .../org/tensorflow/op/core/ReduceSum.java     |  118 +
 .../java/org/tensorflow/op/core/RefEnter.java |  134 +
 .../java/org/tensorflow/op/core/RefExit.java  |   76 +
 .../org/tensorflow/op/core/RefIdentity.java   |   73 +
 .../java/org/tensorflow/op/core/RefMerge.java |   86 +
 .../tensorflow/op/core/RefNextIteration.java  |   75 +
 .../org/tensorflow/op/core/RefSelect.java     |   79 +
 .../org/tensorflow/op/core/RefSwitch.java     |   87 +
 .../op/core/RemoteFusedGraphExecute.java      |   97 +
 .../java/org/tensorflow/op/core/Reshape.java  |  137 +
 .../tensorflow/op/core/ResourceCountUpTo.java |   81 +
 .../tensorflow/op/core/ResourceGather.java    |  146 +
 .../tensorflow/op/core/ResourceGatherNd.java  |   77 +
 .../op/core/ResourceScatterAdd.java           |   81 +
 .../op/core/ResourceScatterDiv.java           |   81 +
 .../op/core/ResourceScatterMax.java           |   81 +
 .../op/core/ResourceScatterMin.java           |   81 +
 .../op/core/ResourceScatterMul.java           |   81 +
 .../op/core/ResourceScatterNdAdd.java         |  131 +
 .../op/core/ResourceScatterNdMax.java         |  100 +
 .../op/core/ResourceScatterNdMin.java         |  100 +
 .../op/core/ResourceScatterNdSub.java         |  131 +
 .../op/core/ResourceScatterNdUpdate.java      |  133 +
 .../op/core/ResourceScatterSub.java           |   81 +
 .../op/core/ResourceScatterUpdate.java        |   72 +
 .../op/core/ResourceStridedSliceAssign.java   |  182 +
 .../java/org/tensorflow/op/core/Reverse.java  |  126 +
 .../tensorflow/op/core/ReverseSequence.java   |  168 +
 .../gen/java/org/tensorflow/op/core/Roll.java |  109 +
 .../gen/java/org/tensorflow/op/core/Rpc.java  |  209 +
 .../org/tensorflow/op/core/ScatterAdd.java    |  140 +
 .../org/tensorflow/op/core/ScatterDiv.java    |  136 +
 .../org/tensorflow/op/core/ScatterMax.java    |  139 +
 .../org/tensorflow/op/core/ScatterMin.java    |  139 +
 .../org/tensorflow/op/core/ScatterMul.java    |  136 +
 .../org/tensorflow/op/core/ScatterNd.java     |  158 +
 .../org/tensorflow/op/core/ScatterNdAdd.java  |  151 +
 .../org/tensorflow/op/core/ScatterNdMax.java  |  120 +
 .../org/tensorflow/op/core/ScatterNdMin.java  |  120 +
 .../op/core/ScatterNdNonAliasingAdd.java      |  117 +
 .../org/tensorflow/op/core/ScatterNdSub.java  |  153 +
 .../tensorflow/op/core/ScatterNdUpdate.java   |  155 +
 .../org/tensorflow/op/core/ScatterSub.java    |  139 +
 .../org/tensorflow/op/core/ScatterUpdate.java |  143 +
 .../java/org/tensorflow/op/core/Select.java   |   77 +
 .../gen/java/org/tensorflow/op/core/Send.java |  103 +
 .../org/tensorflow/op/core/SetDiff1d.java     |  119 +
 .../java/org/tensorflow/op/core/SetSize.java  |  122 +
 .../java/org/tensorflow/op/core/Shape.java    |   99 +
 .../java/org/tensorflow/op/core/ShapeN.java   |  100 +
 .../gen/java/org/tensorflow/op/core/Size.java |  100 +
 .../java/org/tensorflow/op/core/Skipgram.java |  201 +
 .../java/org/tensorflow/op/core/Slice.java    |   90 +
 .../java/org/tensorflow/op/core/Snapshot.java |   74 +
 .../tensorflow/op/core/SpaceToBatchNd.java    |  180 +
 .../java/org/tensorflow/op/core/Split.java    |   91 +
 .../java/org/tensorflow/op/core/SplitV.java   |   95 +
 .../java/org/tensorflow/op/core/Squeeze.java  |  136 +
 .../java/org/tensorflow/op/core/Stack.java    |  130 +
 .../java/org/tensorflow/op/core/Stage.java    |  157 +
 .../org/tensorflow/op/core/StageClear.java    |  151 +
 .../org/tensorflow/op/core/StagePeek.java     |  180 +
 .../org/tensorflow/op/core/StageSize.java     |  169 +
 .../org/tensorflow/op/core/StopGradient.java  |   99 +
 .../org/tensorflow/op/core/StridedSlice.java  |  317 +
 .../op/core/StridedSliceAssign.java           |  200 +
 .../tensorflow/op/core/StridedSliceGrad.java  |  202 +
 .../gen/java/org/tensorflow/op/core/Sum.java  |  118 +
 .../org/tensorflow/op/core/SwitchCond.java    |   87 +
 .../tensorflow/op/core/TemporaryVariable.java |  129 +
 .../org/tensorflow/op/core/TensorArray.java   |  218 +
 .../tensorflow/op/core/TensorArrayClose.java  |   58 +
 .../tensorflow/op/core/TensorArrayConcat.java |  140 +
 .../tensorflow/op/core/TensorArrayGather.java |  125 +
 .../tensorflow/op/core/TensorArrayGrad.java   |  117 +
 .../op/core/TensorArrayGradWithShape.java     |   90 +
 .../tensorflow/op/core/TensorArrayPack.java   |  113 +
 .../tensorflow/op/core/TensorArrayRead.java   |   83 +
 .../op/core/TensorArrayScatter.java           |   83 +
 .../tensorflow/op/core/TensorArraySize.java   |   76 +
 .../tensorflow/op/core/TensorArraySplit.java  |  104 +
 .../tensorflow/op/core/TensorArrayUnpack.java |   77 +
 .../tensorflow/op/core/TensorArrayWrite.java  |   81 +
 .../core/TensorForestCreateTreeVariable.java  |   57 +
 .../op/core/TensorForestTreeDeserialize.java  |   57 +
 .../core/TensorForestTreeIsInitializedOp.java |   72 +
 .../op/core/TensorForestTreePredict.java      |   76 +
 .../TensorForestTreeResourceHandleOp.java     |  123 +
 .../op/core/TensorForestTreeSerialize.java    |   72 +
 .../op/core/TensorForestTreeSize.java         |   72 +
 .../tensorflow/op/core/TensorListConcat.java  |   98 +
 .../op/core/TensorListConcatLists.java        |   76 +
 .../op/core/TensorListElementShape.java       |   79 +
 .../op/core/TensorListFromTensor.java         |   81 +
 .../tensorflow/op/core/TensorListGather.java  |   88 +
 .../tensorflow/op/core/TensorListGetItem.java |   79 +
 .../tensorflow/op/core/TensorListLength.java  |   75 +
 .../tensorflow/op/core/TensorListPopBack.java |   89 +
 .../op/core/TensorListPushBack.java           |   81 +
 .../op/core/TensorListPushBackBatch.java      |   74 +
 .../tensorflow/op/core/TensorListReserve.java |   84 +
 .../tensorflow/op/core/TensorListResize.java  |   81 +
 .../tensorflow/op/core/TensorListScatter.java |   93 +
 .../TensorListScatterIntoExistingList.java    |   86 +
 .../tensorflow/op/core/TensorListSetItem.java |   77 +
 .../tensorflow/op/core/TensorListSplit.java   |   87 +
 .../tensorflow/op/core/TensorListStack.java   |  120 +
 .../op/core/TensorScatterNdAdd.java           |  140 +
 .../op/core/TensorScatterNdMax.java           |   78 +
 .../op/core/TensorScatterNdMin.java           |   78 +
 .../op/core/TensorScatterNdSub.java           |  139 +
 .../op/core/TensorScatterNdUpdate.java        |  155 +
 .../op/core/TensorStridedSliceUpdate.java     |  200 +
 .../gen/java/org/tensorflow/op/core/Tile.java |  104 +
 .../org/tensorflow/op/core/Timestamp.java     |   75 +
 .../java/org/tensorflow/op/core/TryRpc.java   |  226 +
 .../java/org/tensorflow/op/core/Unbatch.java  |  153 +
 .../org/tensorflow/op/core/UnbatchGrad.java   |  149 +
 .../java/org/tensorflow/op/core/Unique.java   |  142 +
 .../tensorflow/op/core/UniqueWithCounts.java  |  155 +
 .../org/tensorflow/op/core/UnravelIndex.java  |   99 +
 .../java/org/tensorflow/op/core/Unstack.java  |  133 +
 .../java/org/tensorflow/op/core/Unstage.java  |  176 +
 .../org/tensorflow/op/core/UpperBound.java    |  115 +
 .../org/tensorflow/op/core/VarHandleOp.java   |  156 +
 .../op/core/VarIsInitializedOp.java           |   74 +
 .../java/org/tensorflow/op/core/Variable.java |  139 +
 .../org/tensorflow/op/core/VariableShape.java |   98 +
 .../java/org/tensorflow/op/core/Where.java    |  133 +
 .../op/core/XlaSpmdFullToShardShape.java      |   81 +
 .../op/core/XlaSpmdShardToFullShape.java      |   83 +
 .../org/tensorflow/op/core/ZerosLike.java     |   75 +
 .../tensorflow/op/data/AnonymousIterator.java |   91 +
 .../op/data/AnonymousMemoryCache.java         |   69 +
 .../op/data/AnonymousMultiDeviceIterator.java |   96 +
 .../tensorflow/op/data/AssertNextDataset.java |  100 +
 .../tensorflow/op/data/AutoShardDataset.java  |  133 +
 .../org/tensorflow/op/data/BatchDataset.java  |  128 +
 .../op/data/BytesProducedStatsDataset.java    |   89 +
 .../org/tensorflow/op/data/CSVDataset.java    |  100 +
 .../org/tensorflow/op/data/CacheDataset.java  |   95 +
 .../tensorflow/op/data/CacheDatasetV2.java    |   90 +
 .../op/data/ChooseFastestDataset.java         |   88 +
 .../op/data/ConcatenateDataset.java           |   89 +
 .../op/data/DatasetCardinality.java           |   75 +
 .../tensorflow/op/data/DatasetFromGraph.java  |   76 +
 .../tensorflow/op/data/DatasetToGraph.java    |  127 +
 .../op/data/DatasetToSingleElement.java       |   91 +
 .../tensorflow/op/data/DatasetToTfRecord.java |   60 +
 .../tensorflow/op/data/DeleteIterator.java    |   57 +
 .../tensorflow/op/data/DeleteMemoryCache.java |   55 +
 .../op/data/DeleteMultiDeviceIterator.java    |   59 +
 .../op/data/DenseToSparseBatchDataset.java    |   94 +
 .../op/data/DeserializeIterator.java          |   58 +
 .../op/data/DirectedInterleaveDataset.java    |   91 +
 .../op/data/FilterByLastComponentDataset.java |   86 +
 .../op/data/FixedLengthRecordDataset.java     |   83 +
 .../op/data/IgnoreErrorsDataset.java          |   86 +
 .../op/data/InitializeTableFromDataset.java   |   55 +
 .../java/org/tensorflow/op/data/Iterator.java |   88 +
 .../op/data/IteratorFromStringHandle.java     |  118 +
 .../tensorflow/op/data/IteratorGetDevice.java |   71 +
 .../tensorflow/op/data/IteratorGetNext.java   |   91 +
 .../op/data/IteratorGetNextAsOptional.java    |   87 +
 .../op/data/IteratorGetNextSync.java          |   96 +
 .../op/data/IteratorToStringHandle.java       |   73 +
 .../org/tensorflow/op/data/LMDBDataset.java   |   99 +
 .../op/data/LatencyStatsDataset.java          |   89 +
 .../org/tensorflow/op/data/LeakyReluGrad.java |  111 +
 .../org/tensorflow/op/data/MakeIterator.java  |   60 +
 .../op/data/MatchingFilesDataset.java         |   72 +
 .../op/data/MaxIntraOpParallelismDataset.java |   89 +
 .../org/tensorflow/op/data/ModelDataset.java  |  141 +
 .../op/data/MultiDeviceIterator.java          |   97 +
 .../MultiDeviceIteratorFromStringHandle.java  |  120 +
 .../MultiDeviceIteratorGetNextFromShard.java  |   97 +
 .../op/data/MultiDeviceIteratorInit.java      |   77 +
 .../MultiDeviceIteratorToStringHandle.java    |   72 +
 .../op/data/NonSerializableDataset.java       |   85 +
 .../tensorflow/op/data/OptimizeDataset.java   |  129 +
 .../tensorflow/op/data/OptionalFromValue.java |   74 +
 .../tensorflow/op/data/OptionalGetValue.java  |   91 +
 .../tensorflow/op/data/OptionalHasValue.java  |   72 +
 .../org/tensorflow/op/data/OptionalNone.java  |   71 +
 .../op/data/PaddedBatchDataset.java           |  131 +
 .../tensorflow/op/data/PrefetchDataset.java   |  143 +
 .../op/data/PrivateThreadPoolDataset.java     |   89 +
 .../org/tensorflow/op/data/RandomDataset.java |  102 +
 .../org/tensorflow/op/data/RangeDataset.java  |   92 +
 .../tensorflow/op/data/RebatchDataset.java    |  128 +
 .../tensorflow/op/data/RegisterDataset.java   |   78 +
 .../org/tensorflow/op/data/RepeatDataset.java |   91 +
 .../tensorflow/op/data/SamplingDataset.java   |  101 +
 .../tensorflow/op/data/SerializeIterator.java |  109 +
 .../op/data/SetStatsAggregatorDataset.java    |   92 +
 .../org/tensorflow/op/data/ShardDataset.java  |  125 +
 .../op/data/ShuffleAndRepeatDataset.java      |  130 +
 .../tensorflow/op/data/ShuffleDataset.java    |  128 +
 .../org/tensorflow/op/data/SkipDataset.java   |   91 +
 .../org/tensorflow/op/data/SleepDataset.java  |   88 +
 .../op/data/SlidingWindowDataset.java         |   96 +
 .../op/data/SparseTensorSliceDataset.java     |   77 +
 .../org/tensorflow/op/data/SqlDataset.java    |   91 +
 .../op/data/StatsAggregatorHandle.java        |  123 +
 .../org/tensorflow/op/data/TakeDataset.java   |   92 +
 .../org/tensorflow/op/data/TensorDataset.java |   81 +
 .../op/data/TensorSliceDataset.java           |   82 +
 .../tensorflow/op/data/TextLineDataset.java   |   81 +
 .../tensorflow/op/data/TfRecordDataset.java   |   82 +
 .../tensorflow/op/data/ThreadPoolDataset.java |   88 +
 .../tensorflow/op/data/ThreadPoolHandle.java  |  152 +
 .../tensorflow/op/data/UnbatchDataset.java    |   86 +
 .../org/tensorflow/op/data/UniqueDataset.java |   86 +
 .../op/data/UnwrapDatasetVariant.java         |   71 +
 .../org/tensorflow/op/data/WindowDataset.java |  143 +
 .../op/data/WrapDatasetVariant.java           |   71 +
 .../org/tensorflow/op/data/ZipDataset.java    |   94 +
 .../AssertCardinalityDataset.java             |   88 +
 .../data/experimental/AssertNextDataset.java  |   88 +
 .../data/experimental/AutoShardDataset.java   |  133 +
 .../BytesProducedStatsDataset.java            |   89 +
 .../op/data/experimental/CSVDataset.java      |   99 +
 .../experimental/ChooseFastestDataset.java    |   88 +
 .../op/data/experimental/CompressElement.java |   73 +
 .../data/experimental/DataServiceDataset.java |  134 +
 .../data/experimental/DatasetCardinality.java |   75 +
 .../data/experimental/DatasetToTFRecord.java  |   60 +
 .../DenseToSparseBatchDataset.java            |   94 +
 .../DirectedInterleaveDataset.java            |   91 +
 .../experimental/DummyIterationCounter.java   |   69 +
 .../experimental/IgnoreErrorsDataset.java     |   86 +
 .../data/experimental/IteratorGetDevice.java  |   71 +
 .../experimental/LatencyStatsDataset.java     |   89 +
 .../op/data/experimental/LmdbDataset.java     |   86 +
 .../experimental/MatchingFilesDataset.java    |   72 +
 .../MaxIntraOpParallelismDataset.java         |   89 +
 .../experimental/NonSerializableDataset.java  |   85 +
 .../experimental/ParseExampleDataset.java     |  207 +
 .../PrivateThreadPoolDataset.java             |   89 +
 .../op/data/experimental/RandomDataset.java   |   91 +
 .../op/data/experimental/RebatchDataset.java  |  128 +
 .../SetStatsAggregatorDataset.java            |   92 +
 .../op/data/experimental/SleepDataset.java    |   88 +
 .../experimental/SlidingWindowDataset.java    |   96 +
 .../op/data/experimental/SqlDataset.java      |   91 +
 .../experimental/StatsAggregatorHandle.java   |  122 +
 .../StatsAggregatorSetSummaryWriter.java      |   56 +
 .../experimental/StatsAggregatorSummary.java  |   71 +
 .../data/experimental/ThreadPoolDataset.java  |   88 +
 .../data/experimental/ThreadPoolHandle.java   |  152 +
 .../op/data/experimental/UnbatchDataset.java  |   86 +
 .../data/experimental/UncompressElement.java  |   90 +
 .../op/data/experimental/UniqueDataset.java   |   86 +
 .../op/debugging/CheckNumerics.java           |   80 +
 .../op/debugging/DebugGradientIdentity.java   |   77 +
 .../debugging/DebugGradientRefIdentity.java   |   77 +
 .../op/debugging/DebugIdentity.java           |  242 +
 .../op/debugging/DebugNanCount.java           |  182 +
 .../op/debugging/DebugNumericsSummary.java    |  253 +
 .../org/tensorflow/op/dtypes/AsString.java    |  206 +
 .../java/org/tensorflow/op/dtypes/Cast.java   |  110 +
 .../org/tensorflow/op/dtypes/Complex.java     |   94 +
 .../java/org/tensorflow/op/dtypes/ToBool.java |   85 +
 .../estimator/BoostedTreesAggregateStats.java |   86 +
 .../op/estimator/BoostedTreesBucketize.java   |   87 +
 ...BoostedTreesCalculateBestFeatureSplit.java |  179 +
 ...ostedTreesCalculateBestFeatureSplitV2.java |  159 +
 ...stedTreesCalculateBestGainsPerFeature.java |  139 +
 .../op/estimator/BoostedTreesCenterBias.java  |   81 +
 .../estimator/BoostedTreesCreateEnsemble.java |   60 +
 ...stedTreesCreateQuantileStreamResource.java |   94 +
 .../BoostedTreesDeserializeEnsemble.java      |   62 +
 .../BoostedTreesEnsembleResourceHandleOp.java |  123 +
 .../BoostedTreesExampleDebugOutputs.java      |   84 +
 .../BoostedTreesFlushQuantileSummaries.java   |   84 +
 .../BoostedTreesGetEnsembleStates.java        |  105 +
 .../BoostedTreesMakeQuantileSummaries.java    |   88 +
 .../BoostedTreesMakeStatsSummary.java         |   86 +
 .../op/estimator/BoostedTreesPredict.java     |   83 +
 ...eesQuantileStreamResourceAddSummaries.java |   62 +
 ...reesQuantileStreamResourceDeserialize.java |   60 +
 ...ostedTreesQuantileStreamResourceFlush.java |  103 +
 ...tileStreamResourceGetBucketBoundaries.java |   84 +
 ...edTreesQuantileStreamResourceHandleOp.java |  123 +
 .../BoostedTreesSerializeEnsemble.java        |   77 +
 .../BoostedTreesSparseAggregateStats.java     |  115 +
 ...dTreesSparseCalculateBestFeatureSplit.java |  184 +
 .../BoostedTreesTrainingPredict.java          |  104 +
 .../estimator/BoostedTreesUpdateEnsemble.java |   85 +
 .../BoostedTreesUpdateEnsembleV2.java         |  124 +
 .../IsBoostedTreesEnsembleInitialized.java    |   72 +
 ...reesQuantileStreamResourceInitialized.java |   74 +
 .../tensorflow/op/image/AdjustContrast.java   |   88 +
 .../org/tensorflow/op/image/AdjustHue.java    |   85 +
 .../tensorflow/op/image/AdjustSaturation.java |   85 +
 .../op/image/CombinedNonMaxSuppression.java   |  197 +
 .../tensorflow/op/image/CropAndResize.java    |  168 +
 .../op/image/CropAndResizeGradBoxes.java      |  128 +
 .../op/image/CropAndResizeGradImage.java      |  133 +
 .../op/image/DecodeAndCropJpeg.java           |  245 +
 .../org/tensorflow/op/image/DecodeBmp.java    |  122 +
 .../org/tensorflow/op/image/DecodeGif.java    |   83 +
 .../org/tensorflow/op/image/DecodeJpeg.java   |  242 +
 .../org/tensorflow/op/image/DecodePng.java    |  150 +
 .../op/image/DrawBoundingBoxes.java           |   94 +
 .../org/tensorflow/op/image/EncodeJpeg.java   |  288 +
 .../op/image/EncodeJpegVariableQuality.java   |   81 +
 .../org/tensorflow/op/image/EncodePng.java    |  128 +
 .../tensorflow/op/image/ExtractGlimpse.java   |  211 +
 .../op/image/ExtractImagePatches.java         |  105 +
 .../tensorflow/op/image/ExtractJpegShape.java |   94 +
 .../image/GenerateBoundingBoxProposals.java   |  138 +
 .../org/tensorflow/op/image/HsvToRgb.java     |   81 +
 .../op/image/ImageProjectiveTransformV2.java  |  125 +
 .../tensorflow/op/image/NearestNeighbors.java |   88 +
 .../op/image/NonMaxSuppression.java           |  169 +
 .../image/NonMaxSuppressionWithOverlaps.java  |  103 +
 .../op/image/QuantizedResizeBilinear.java     |  153 +
 .../org/tensorflow/op/image/RandomCrop.java   |  142 +
 .../org/tensorflow/op/image/ResizeArea.java   |  127 +
 .../tensorflow/op/image/ResizeBicubic.java    |  136 +
 .../op/image/ResizeBicubicGrad.java           |  135 +
 .../tensorflow/op/image/ResizeBilinear.java   |  136 +
 .../op/image/ResizeBilinearGrad.java          |  135 +
 .../op/image/ResizeNearestNeighbor.java       |  135 +
 .../op/image/ResizeNearestNeighborGrad.java   |  134 +
 .../org/tensorflow/op/image/RgbToHsv.java     |   95 +
 .../op/image/SampleDistortedBoundingBox.java  |  291 +
 .../op/image/ScaleAndTranslate.java           |  132 +
 .../op/image/ScaleAndTranslateGrad.java       |  131 +
 .../org/tensorflow/op/io/DecodeBase64.java    |   76 +
 .../tensorflow/op/io/DecodeCompressed.java    |  117 +
 .../java/org/tensorflow/op/io/DecodeCsv.java  |  189 +
 .../tensorflow/op/io/DecodeJsonExample.java   |   82 +
 .../org/tensorflow/op/io/DecodePaddedRaw.java |  120 +
 .../java/org/tensorflow/op/io/DecodeRaw.java  |  118 +
 .../op/io/DeserializeManySparse.java          |  132 +
 .../org/tensorflow/op/io/EncodeBase64.java    |  114 +
 .../java/org/tensorflow/op/io/FifoQueue.java  |  187 +
 .../op/io/FixedLengthRecordReader.java        |  211 +
 .../org/tensorflow/op/io/IdentityReader.java  |  132 +
 .../java/org/tensorflow/op/io/LmdbReader.java |  128 +
 .../org/tensorflow/op/io/MatchingFiles.java   |   77 +
 .../tensorflow/op/io/PaddingFifoQueue.java    |  199 +
 .../org/tensorflow/op/io/ParseExample.java    |  202 +
 .../op/io/ParseSequenceExample.java           |  426 +
 .../tensorflow/op/io/ParseSingleExample.java  |  157 +
 .../op/io/ParseSingleSequenceExample.java     |  283 +
 .../org/tensorflow/op/io/ParseTensor.java     |   79 +
 .../org/tensorflow/op/io/PriorityQueue.java   |  173 +
 .../java/org/tensorflow/op/io/QueueClose.java |   97 +
 .../org/tensorflow/op/io/QueueDequeue.java    |  130 +
 .../tensorflow/op/io/QueueDequeueMany.java    |  140 +
 .../tensorflow/op/io/QueueDequeueUpTo.java    |  144 +
 .../org/tensorflow/op/io/QueueEnqueue.java    |  102 +
 .../tensorflow/op/io/QueueEnqueueMany.java    |  107 +
 .../org/tensorflow/op/io/QueueIsClosed.java   |   75 +
 .../java/org/tensorflow/op/io/QueueSize.java  |   73 +
 .../tensorflow/op/io/RandomShuffleQueue.java  |  250 +
 .../java/org/tensorflow/op/io/ReadFile.java   |   72 +
 .../op/io/ReaderNumRecordsProduced.java       |   75 +
 .../op/io/ReaderNumWorkUnitsCompleted.java    |   72 +
 .../java/org/tensorflow/op/io/ReaderRead.java |   83 +
 .../org/tensorflow/op/io/ReaderReadUpTo.java  |   87 +
 .../org/tensorflow/op/io/ReaderReset.java     |   55 +
 .../tensorflow/op/io/ReaderRestoreState.java  |   62 +
 .../op/io/ReaderSerializeState.java           |   75 +
 .../tensorflow/op/io/SerializeManySparse.java |  105 +
 .../org/tensorflow/op/io/SerializeSparse.java |   97 +
 .../org/tensorflow/op/io/SerializeTensor.java |   74 +
 .../org/tensorflow/op/io/ShardedFilename.java |   79 +
 .../org/tensorflow/op/io/ShardedFilespec.java |   75 +
 .../org/tensorflow/op/io/TextLineReader.java  |  148 +
 .../org/tensorflow/op/io/TfRecordReader.java  |  148 +
 .../org/tensorflow/op/io/WholeFileReader.java |  132 +
 .../java/org/tensorflow/op/io/WriteFile.java  |   60 +
 .../org/tensorflow/op/linalg/BandPart.java    |  118 +
 .../op/linalg/BandedTriangularSolve.java      |  126 +
 .../tensorflow/op/linalg/BatchCholesky.java   |   72 +
 .../op/linalg/BatchCholeskyGrad.java          |   74 +
 .../op/linalg/BatchMatrixBandPart.java        |   77 +
 .../op/linalg/BatchMatrixDeterminant.java     |   72 +
 .../tensorflow/op/linalg/BatchMatrixDiag.java |   72 +
 .../op/linalg/BatchMatrixDiagPart.java        |   72 +
 .../op/linalg/BatchMatrixInverse.java         |  106 +
 .../op/linalg/BatchMatrixSetDiag.java         |   74 +
 .../op/linalg/BatchMatrixSolve.java           |  108 +
 .../op/linalg/BatchMatrixSolveLs.java         |  111 +
 .../op/linalg/BatchMatrixTriangularSolve.java |  127 +
 .../op/linalg/BatchSelfAdjointEig.java        |  109 +
 .../org/tensorflow/op/linalg/BatchSvd.java    |  136 +
 .../org/tensorflow/op/linalg/Cholesky.java    |   89 +
 .../tensorflow/op/linalg/CholeskyGrad.java    |   84 +
 .../op/linalg/ConjugateTranspose.java         |   81 +
 .../java/org/tensorflow/op/linalg/Cross.java  |   81 +
 .../java/org/tensorflow/op/linalg/Det.java    |   79 +
 .../java/org/tensorflow/op/linalg/Eig.java    |  129 +
 .../java/org/tensorflow/op/linalg/Einsum.java |  155 +
 .../tensorflow/op/linalg/EuclideanNorm.java   |  118 +
 .../java/org/tensorflow/op/linalg/Inv.java    |  125 +
 .../op/linalg/LoadAndRemapMatrix.java         |  175 +
 .../op/linalg/LogMatrixDeterminant.java       |   90 +
 .../gen/java/org/tensorflow/op/linalg/Lu.java |  125 +
 .../java/org/tensorflow/op/linalg/MatMul.java |  137 +
 .../org/tensorflow/op/linalg/MatrixDiag.java  |  177 +
 .../tensorflow/op/linalg/MatrixDiagPart.java  |  151 +
 .../op/linalg/MatrixDiagPartV3.java           |  228 +
 .../tensorflow/op/linalg/MatrixDiagV3.java    |  252 +
 .../tensorflow/op/linalg/MatrixLogarithm.java |   94 +
 .../tensorflow/op/linalg/MatrixSetDiag.java   |  232 +
 .../tensorflow/op/linalg/MatrixSolveLs.java   |  153 +
 .../gen/java/org/tensorflow/op/linalg/Qr.java |  130 +
 .../tensorflow/op/linalg/QuantizedMatMul.java |  161 +
 .../op/linalg/QuantizedMatMulWithBias.java    |  181 +
 .../QuantizedMatMulWithBiasAndRelu.java       |  182 +
 ...zedMatMulWithBiasAndReluAndRequantize.java |  187 +
 .../tensorflow/op/linalg/SelfAdjointEig.java  |  127 +
 .../java/org/tensorflow/op/linalg/Solve.java  |  120 +
 .../java/org/tensorflow/op/linalg/Sqrtm.java  |   95 +
 .../java/org/tensorflow/op/linalg/Svd.java    |  166 +
 .../org/tensorflow/op/linalg/TensorDiag.java  |   92 +
 .../tensorflow/op/linalg/TensorDiagPart.java  |   94 +
 .../org/tensorflow/op/linalg/Transpose.java   |   80 +
 .../tensorflow/op/linalg/TriangularSolve.java |  189 +
 .../op/linalg/TridiagonalMatMul.java          |   86 +
 .../op/linalg/TridiagonalSolve.java           |  124 +
 .../sparse/CSRSparseMatrixComponents.java     |   95 +
 .../linalg/sparse/CSRSparseMatrixToDense.java |   76 +
 .../sparse/CSRSparseMatrixToSparseTensor.java |   90 +
 .../linalg/sparse/DenseToCSRSparseMatrix.java |   76 +
 .../op/linalg/sparse/SparseMatrixAdd.java     |   82 +
 .../op/linalg/sparse/SparseMatrixMatMul.java  |  230 +
 .../op/linalg/sparse/SparseMatrixMul.java     |   84 +
 .../op/linalg/sparse/SparseMatrixNNZ.java     |   72 +
 .../sparse/SparseMatrixOrderingAMD.java       |  120 +
 .../op/linalg/sparse/SparseMatrixSoftmax.java |   82 +
 .../sparse/SparseMatrixSoftmaxGrad.java       |   78 +
 .../sparse/SparseMatrixSparseCholesky.java    |  149 +
 .../sparse/SparseMatrixSparseMatMul.java      |  241 +
 .../linalg/sparse/SparseMatrixTranspose.java  |  112 +
 .../op/linalg/sparse/SparseMatrixZeros.java   |   76 +
 .../sparse/SparseTensorToCSRSparseMatrix.java |   78 +
 .../gen/java/org/tensorflow/op/math/Abs.java  |   78 +
 .../org/tensorflow/op/math/AccumulateN.java   |   87 +
 .../gen/java/org/tensorflow/op/math/Acos.java |   74 +
 .../java/org/tensorflow/op/math/Acosh.java    |   82 +
 .../gen/java/org/tensorflow/op/math/Add.java  |   79 +
 .../gen/java/org/tensorflow/op/math/AddN.java |   83 +
 .../java/org/tensorflow/op/math/Angle.java    |  106 +
 .../tensorflow/op/math/ApproximateEqual.java  |  109 +
 .../java/org/tensorflow/op/math/ArgMax.java   |  110 +
 .../java/org/tensorflow/op/math/ArgMin.java   |  110 +
 .../gen/java/org/tensorflow/op/math/Asin.java |   90 +
 .../java/org/tensorflow/op/math/Asinh.java    |   84 +
 .../gen/java/org/tensorflow/op/math/Atan.java |   90 +
 .../java/org/tensorflow/op/math/Atan2.java    |   82 +
 .../java/org/tensorflow/op/math/Atanh.java    |   86 +
 .../java/org/tensorflow/op/math/BesselI0.java |   71 +
 .../org/tensorflow/op/math/BesselI0e.java     |   71 +
 .../java/org/tensorflow/op/math/BesselI1.java |   71 +
 .../org/tensorflow/op/math/BesselI1e.java     |   71 +
 .../java/org/tensorflow/op/math/Betainc.java  |   89 +
 .../java/org/tensorflow/op/math/Bincount.java |   91 +
 .../gen/java/org/tensorflow/op/math/Ceil.java |   74 +
 .../tensorflow/op/math/CompareAndBitpack.java |   98 +
 .../org/tensorflow/op/math/ComplexAbs.java    |   95 +
 .../gen/java/org/tensorflow/op/math/Conj.java |   88 +
 .../gen/java/org/tensorflow/op/math/Cos.java  |   85 +
 .../gen/java/org/tensorflow/op/math/Cosh.java |   84 +
 .../java/org/tensorflow/op/math/Cumprod.java  |  156 +
 .../java/org/tensorflow/op/math/Cumsum.java   |  156 +
 .../op/math/CumulativeLogsumexp.java          |  147 +
 .../org/tensorflow/op/math/DenseBincount.java |  124 +
 .../java/org/tensorflow/op/math/Digamma.java  |   76 +
 .../gen/java/org/tensorflow/op/math/Div.java  |   79 +
 .../java/org/tensorflow/op/math/DivNoNan.java |   80 +
 .../java/org/tensorflow/op/math/Equal.java    |  122 +
 .../gen/java/org/tensorflow/op/math/Erf.java  |   74 +
 .../gen/java/org/tensorflow/op/math/Erfc.java |   74 +
 .../gen/java/org/tensorflow/op/math/Exp.java  |  100 +
 .../java/org/tensorflow/op/math/Expm1.java    |   89 +
 .../gen/java/org/tensorflow/op/math/Fact.java |   70 +
 .../java/org/tensorflow/op/math/Floor.java    |   74 +
 .../java/org/tensorflow/op/math/FloorDiv.java |   79 +
 .../java/org/tensorflow/op/math/FloorMod.java |   82 +
 .../java/org/tensorflow/op/math/Greater.java  |   90 +
 .../org/tensorflow/op/math/GreaterEqual.java  |   90 +
 .../java/org/tensorflow/op/math/Igamma.java   |   89 +
 .../org/tensorflow/op/math/IgammaGradA.java   |   75 +
 .../java/org/tensorflow/op/math/Igammac.java  |   89 +
 .../gen/java/org/tensorflow/op/math/Imag.java |  102 +
 .../tensorflow/op/math/InvertPermutation.java |   91 +
 .../java/org/tensorflow/op/math/IsFinite.java |   84 +
 .../java/org/tensorflow/op/math/IsInf.java    |   84 +
 .../java/org/tensorflow/op/math/IsNan.java    |   84 +
 .../gen/java/org/tensorflow/op/math/Less.java |   90 +
 .../org/tensorflow/op/math/LessEqual.java     |   90 +
 .../java/org/tensorflow/op/math/Lgamma.java   |   84 +
 .../gen/java/org/tensorflow/op/math/Log.java  |   83 +
 .../java/org/tensorflow/op/math/Log1p.java    |   83 +
 .../org/tensorflow/op/math/LogicalAnd.java    |   77 +
 .../org/tensorflow/op/math/LogicalNot.java    |   73 +
 .../org/tensorflow/op/math/LogicalOr.java     |   77 +
 .../java/org/tensorflow/op/math/Maximum.java  |   79 +
 .../gen/java/org/tensorflow/op/math/Mean.java |  118 +
 .../java/org/tensorflow/op/math/Minimum.java  |   79 +
 .../gen/java/org/tensorflow/op/math/Mod.java  |   82 +
 .../gen/java/org/tensorflow/op/math/Mul.java  |   79 +
 .../java/org/tensorflow/op/math/MulNoNan.java |   79 +
 .../java/org/tensorflow/op/math/Ndtri.java    |   72 +
 .../gen/java/org/tensorflow/op/math/Neg.java  |   76 +
 .../org/tensorflow/op/math/NextAfter.java     |   84 +
 .../java/org/tensorflow/op/math/NotEqual.java |  112 +
 .../org/tensorflow/op/math/Polygamma.java     |   83 +
 .../tensorflow/op/math/PopulationCount.java   |   80 +
 .../gen/java/org/tensorflow/op/math/Pow.java  |   85 +
 .../org/tensorflow/op/math/QuantizedAdd.java  |  103 +
 .../org/tensorflow/op/math/QuantizedMul.java  |  103 +
 .../gen/java/org/tensorflow/op/math/Real.java |  102 +
 .../java/org/tensorflow/op/math/RealDiv.java  |   81 +
 .../org/tensorflow/op/math/Reciprocal.java    |   76 +
 .../tensorflow/op/math/ReciprocalGrad.java    |   78 +
 .../math/RequantizationRangePerChannel.java   |   84 +
 .../op/math/RequantizePerChannel.java         |   98 +
 .../gen/java/org/tensorflow/op/math/Rint.java |   84 +
 .../java/org/tensorflow/op/math/Round.java    |   77 +
 .../java/org/tensorflow/op/math/Rsqrt.java    |   76 +
 .../org/tensorflow/op/math/RsqrtGrad.java     |   78 +
 .../org/tensorflow/op/math/SegmentMax.java    |  102 +
 .../org/tensorflow/op/math/SegmentMean.java   |  104 +
 .../org/tensorflow/op/math/SegmentMin.java    |  102 +
 .../org/tensorflow/op/math/SegmentProd.java   |  103 +
 .../org/tensorflow/op/math/SegmentSum.java    |  103 +
 .../java/org/tensorflow/op/math/Sigmoid.java  |   76 +
 .../org/tensorflow/op/math/SigmoidGrad.java   |   78 +
 .../gen/java/org/tensorflow/op/math/Sign.java |   82 +
 .../gen/java/org/tensorflow/op/math/Sin.java  |   84 +
 .../gen/java/org/tensorflow/op/math/Sinh.java |   84 +
 .../org/tensorflow/op/math/SobolSample.java   |  103 +
 .../java/org/tensorflow/op/math/Softplus.java |   74 +
 .../org/tensorflow/op/math/SoftplusGrad.java  |   76 +
 .../gen/java/org/tensorflow/op/math/Sqrt.java |   76 +
 .../java/org/tensorflow/op/math/SqrtGrad.java |   78 +
 .../java/org/tensorflow/op/math/Square.java   |   76 +
 .../tensorflow/op/math/SquaredDifference.java |   79 +
 .../gen/java/org/tensorflow/op/math/Sub.java  |   79 +
 .../gen/java/org/tensorflow/op/math/Tan.java  |   85 +
 .../gen/java/org/tensorflow/op/math/Tanh.java |   84 +
 .../java/org/tensorflow/op/math/TanhGrad.java |   78 +
 .../org/tensorflow/op/math/TruncateDiv.java   |   84 +
 .../org/tensorflow/op/math/TruncateMod.java   |   82 +
 .../op/math/UnsortedSegmentMax.java           |  112 +
 .../op/math/UnsortedSegmentMin.java           |  106 +
 .../op/math/UnsortedSegmentProd.java          |  106 +
 .../op/math/UnsortedSegmentSum.java           |  109 +
 .../java/org/tensorflow/op/math/Xdivy.java    |   76 +
 .../java/org/tensorflow/op/math/Xlog1py.java  |   76 +
 .../java/org/tensorflow/op/math/Xlogy.java    |   76 +
 .../gen/java/org/tensorflow/op/math/Zeta.java |   80 +
 .../java/org/tensorflow/op/math/erfinv.java   |   72 +
 .../tensorflow/op/math/special/BesselJ0.java  |   71 +
 .../tensorflow/op/math/special/BesselJ1.java  |   71 +
 .../tensorflow/op/math/special/BesselK0.java  |   71 +
 .../tensorflow/op/math/special/BesselK0e.java |   71 +
 .../tensorflow/op/math/special/BesselK1.java  |   71 +
 .../tensorflow/op/math/special/BesselK1e.java |   71 +
 .../tensorflow/op/math/special/BesselY0.java  |   71 +
 .../tensorflow/op/math/special/BesselY1.java  |   71 +
 .../org/tensorflow/op/math/special/Dawsn.java |   71 +
 .../tensorflow/op/math/special/Expint.java    |   71 +
 .../op/math/special/FresnelCos.java           |   71 +
 .../op/math/special/FresnelSin.java           |   71 +
 .../tensorflow/op/math/special/Spence.java    |   71 +
 .../java/org/tensorflow/op/nn/AvgPool.java    |  135 +
 .../java/org/tensorflow/op/nn/AvgPool3d.java  |  134 +
 .../org/tensorflow/op/nn/AvgPool3dGrad.java   |  137 +
 .../org/tensorflow/op/nn/AvgPoolGrad.java     |  135 +
 .../nn/BatchNormWithGlobalNormalization.java  |   96 +
 .../BatchNormWithGlobalNormalizationGrad.java |  127 +
 .../java/org/tensorflow/op/nn/BiasAdd.java    |  126 +
 .../org/tensorflow/op/nn/BiasAddGrad.java     |  125 +
 .../java/org/tensorflow/op/nn/BlockLSTM.java  |  216 +
 .../org/tensorflow/op/nn/BlockLSTMGrad.java   |  172 +
 .../java/org/tensorflow/op/nn/CTCLossV2.java  |  173 +
 .../op/nn/ComputeAccidentalHits.java          |  156 +
 .../gen/java/org/tensorflow/op/nn/Conv2d.java |  234 +
 .../op/nn/Conv2dBackpropFilter.java           |  217 +
 .../tensorflow/op/nn/Conv2dBackpropInput.java |  216 +
 .../gen/java/org/tensorflow/op/nn/Conv3d.java |  166 +
 .../op/nn/Conv3dBackpropFilter.java           |  166 +
 .../tensorflow/op/nn/Conv3dBackpropInput.java |  165 +
 .../op/nn/CtcBeamSearchDecoder.java           |  161 +
 .../tensorflow/op/nn/CtcGreedyDecoder.java    |  149 +
 .../java/org/tensorflow/op/nn/CtcLoss.java    |  175 +
 .../java/org/tensorflow/op/nn/CudnnRNN.java   |  333 +
 .../tensorflow/op/nn/CudnnRNNBackprop.java    |  332 +
 .../op/nn/CudnnRNNCanonicalToParams.java      |  263 +
 .../op/nn/CudnnRNNParamsToCanonical.java      |  274 +
 .../tensorflow/op/nn/CudnnRnnParamsSize.java  |  253 +
 .../tensorflow/op/nn/DataFormatDimMap.java    |  131 +
 .../op/nn/DataFormatVecPermute.java           |  130 +
 .../org/tensorflow/op/nn/DepthToSpace.java    |  190 +
 .../op/nn/DepthwiseConv2dNative.java          |  199 +
 .../DepthwiseConv2dNativeBackpropFilter.java  |  195 +
 .../DepthwiseConv2dNativeBackpropInput.java   |  195 +
 .../java/org/tensorflow/op/nn/Dilation2d.java |  118 +
 .../op/nn/Dilation2dBackpropFilter.java       |   96 +
 .../op/nn/Dilation2dBackpropInput.java        |   96 +
 .../gen/java/org/tensorflow/op/nn/Elu.java    |   77 +
 .../java/org/tensorflow/op/nn/EluGrad.java    |   77 +
 .../op/nn/FixedUnigramCandidateSampler.java   |  329 +
 .../tensorflow/op/nn/FractionalAvgPool.java   |  246 +
 .../op/nn/FractionalAvgPoolGrad.java          |  140 +
 .../tensorflow/op/nn/FractionalMaxPool.java   |  270 +
 .../op/nn/FractionalMaxPoolGrad.java          |  136 +
 .../org/tensorflow/op/nn/FusedBatchNorm.java  |  227 +
 .../tensorflow/op/nn/FusedBatchNormGrad.java  |  207 +
 .../org/tensorflow/op/nn/FusedPadConv2d.java  |  105 +
 .../op/nn/FusedResizeAndPadConv2d.java        |  143 +
 .../org/tensorflow/op/nn/GRUBlockCell.java    |  147 +
 .../tensorflow/op/nn/GRUBlockCellGrad.java    |  191 +
 .../gen/java/org/tensorflow/op/nn/InTopK.java |   94 +
 .../java/org/tensorflow/op/nn/InvGrad.java    |   78 +
 .../gen/java/org/tensorflow/op/nn/L2Loss.java |   79 +
 .../org/tensorflow/op/nn/LSTMBlockCell.java   |  234 +
 .../tensorflow/op/nn/LSTMBlockCellGrad.java   |  139 +
 .../java/org/tensorflow/op/nn/LeakyRelu.java  |    4 +-
 .../op/nn/LearnedUnigramCandidateSampler.java |  171 +
 .../op/nn/LocalResponseNormalization.java     |  177 +
 .../op/nn/LocalResponseNormalizationGrad.java |  169 +
 .../java/org/tensorflow/op/nn/LogSoftmax.java |   79 +
 .../java/org/tensorflow/op/nn/MaxPool.java    |  125 +
 .../java/org/tensorflow/op/nn/MaxPool3d.java  |  134 +
 .../org/tensorflow/op/nn/MaxPool3dGrad.java   |  137 +
 .../tensorflow/op/nn/MaxPool3dGradGrad.java   |  138 +
 .../org/tensorflow/op/nn/MaxPoolGrad.java     |  129 +
 .../org/tensorflow/op/nn/MaxPoolGradGrad.java |  129 +
 .../op/nn/MaxPoolGradGradWithArgmax.java      |  130 +
 .../op/nn/MaxPoolGradWithArgmax.java          |  129 +
 .../tensorflow/op/nn/MaxPoolWithArgmax.java   |  160 +
 .../java/org/tensorflow/op/nn/NthElement.java |  123 +
 .../tensorflow/op/nn/QuantizedAvgPool.java    |  109 +
 ...tizedBatchNormWithGlobalNormalization.java |  131 +
 .../tensorflow/op/nn/QuantizedBiasAdd.java    |  102 +
 .../op/nn/QuantizedConv2DAndRelu.java         |  165 +
 .../QuantizedConv2DAndReluAndRequantize.java  |  169 +
 .../op/nn/QuantizedConv2DAndRequantize.java   |  169 +
 .../op/nn/QuantizedConv2DPerChannel.java      |  147 +
 .../op/nn/QuantizedConv2DWithBias.java        |  167 +
 .../op/nn/QuantizedConv2DWithBiasAndRelu.java |  167 +
 ...zedConv2DWithBiasAndReluAndRequantize.java |  171 +
 .../QuantizedConv2DWithBiasAndRequantize.java |  171 +
 ...WithBiasSignedSumAndReluAndRequantize.java |  177 +
 .../nn/QuantizedConv2DWithBiasSumAndRelu.java |  169 +
 ...Conv2DWithBiasSumAndReluAndRequantize.java |  177 +
 .../org/tensorflow/op/nn/QuantizedConv2d.java |  161 +
 .../op/nn/QuantizedDepthwiseConv2D.java       |  147 +
 .../nn/QuantizedDepthwiseConv2DWithBias.java  |  149 +
 ...antizedDepthwiseConv2DWithBiasAndRelu.java |  172 +
 ...iseConv2DWithBiasAndReluAndRequantize.java |  176 +
 .../op/nn/QuantizedInstanceNorm.java          |  207 +
 .../tensorflow/op/nn/QuantizedMaxPool.java    |  109 +
 .../org/tensorflow/op/nn/QuantizedRelu.java   |   95 +
 .../org/tensorflow/op/nn/QuantizedRelu6.java  |   95 +
 .../org/tensorflow/op/nn/QuantizedReluX.java  |   97 +
 .../gen/java/org/tensorflow/op/nn/Relu.java   |   79 +
 .../gen/java/org/tensorflow/op/nn/Relu6.java  |   74 +
 .../java/org/tensorflow/op/nn/Relu6Grad.java  |   78 +
 .../java/org/tensorflow/op/nn/ReluGrad.java   |   77 +
 .../gen/java/org/tensorflow/op/nn/Selu.java   |   82 +
 .../java/org/tensorflow/op/nn/SeluGrad.java   |   77 +
 .../java/org/tensorflow/op/nn/Softmax.java    |   79 +
 .../java/org/tensorflow/op/nn/Softsign.java   |   74 +
 .../org/tensorflow/op/nn/SoftsignGrad.java    |   76 +
 .../org/tensorflow/op/nn/SpaceToBatch.java    |  155 +
 .../org/tensorflow/op/nn/SpaceToDepth.java    |  184 +
 .../gen/java/org/tensorflow/op/nn/TopK.java   |  130 +
 .../nn/raw/SoftmaxCrossEntropyWithLogits.java |   85 +
 .../SparseSoftmaxCrossEntropyWithLogits.java  |   89 +
 .../op/quantization/Dequantize.java           |  219 +
 .../quantization/FakeQuantWithMinMaxArgs.java |  196 +
 .../FakeQuantWithMinMaxArgsGradient.java      |  167 +
 .../quantization/FakeQuantWithMinMaxVars.java |  166 +
 .../FakeQuantWithMinMaxVarsGradient.java      |  149 +
 .../FakeQuantWithMinMaxVarsPerChannel.java    |  167 +
 ...QuantWithMinMaxVarsPerChannelGradient.java |  152 +
 .../tensorflow/op/quantization/Quantize.java  |  327 +
 .../quantization/QuantizeAndDequantize.java   |  175 +
 .../QuantizeDownAndShrinkRange.java           |  117 +
 .../op/quantization/QuantizedConcat.java      |  101 +
 .../QuantizedMatMulWithBiasAndDequantize.java |  163 +
 .../QuantizedMatMulWithBiasAndRequantize.java |  173 +
 .../op/quantization/RequantizationRange.java  |   87 +
 .../op/quantization/Requantize.java           |  106 +
 .../tensorflow/op/ragged/RaggedBincount.java  |  127 +
 .../op/ragged/RaggedCountSparseOutput.java    |  156 +
 .../org/tensorflow/op/ragged/RaggedCross.java |  108 +
 .../tensorflow/op/ragged/RaggedGather.java    |  127 +
 .../org/tensorflow/op/ragged/RaggedRange.java |  115 +
 .../op/ragged/RaggedTensorFromVariant.java    |  127 +
 .../op/ragged/RaggedTensorToSparse.java       |   96 +
 .../op/ragged/RaggedTensorToTensor.java       |  146 +
 .../op/ragged/RaggedTensorToVariant.java      |   93 +
 .../op/random/AllCandidateSampler.java        |  169 +
 .../random/AnonymousRandomSeedGenerator.java  |   75 +
 .../op/random/AnonymousSeedGenerator.java     |   78 +
 .../op/random/DeleteRandomSeedGenerator.java  |   55 +
 .../op/random/DeleteSeedGenerator.java        |   55 +
 .../op/random/LogUniformCandidateSampler.java |  171 +
 .../org/tensorflow/op/random/Multinomial.java |  153 +
 .../op/random/NonDeterministicInts.java       |   91 +
 .../random/ParameterizedTruncatedNormal.java  |  145 +
 .../org/tensorflow/op/random/RandomGamma.java |  142 +
 .../tensorflow/op/random/RandomGammaGrad.java |   75 +
 .../tensorflow/op/random/RandomPoisson.java   |  167 +
 .../tensorflow/op/random/RandomShuffle.java   |  143 +
 .../op/random/RandomStandardNormal.java       |  136 +
 .../tensorflow/op/random/RandomUniform.java   |  137 +
 .../op/random/RandomUniformInt.java           |  144 +
 .../org/tensorflow/op/random/RecordInput.java |  206 +
 .../org/tensorflow/op/random/RngSkip.java     |   64 +
 .../op/random/StatefulRandomBinomial.java     |   99 +
 .../op/random/StatefulStandardNormal.java     |   99 +
 .../op/random/StatefulTruncatedNormal.java    |  100 +
 .../tensorflow/op/random/StatefulUniform.java |   99 +
 .../op/random/StatefulUniformFullInt.java     |   83 +
 .../op/random/StatefulUniformInt.java         |   91 +
 .../op/random/StatelessMultinomial.java       |  100 +
 ...StatelessParameterizedTruncatedNormal.java |   84 +
 .../op/random/StatelessRandomBinomial.java    |  106 +
 .../op/random/StatelessRandomGamma.java       |   83 +
 .../op/random/StatelessRandomNormal.java      |   97 +
 .../op/random/StatelessRandomPoisson.java     |   85 +
 .../op/random/StatelessRandomUniform.java     |   98 +
 .../random/StatelessRandomUniformFullInt.java |   82 +
 .../op/random/StatelessRandomUniformInt.java  |   84 +
 .../op/random/StatelessTruncatedNormal.java   |   99 +
 .../tensorflow/op/random/TruncatedNormal.java |  139 +
 .../op/random/UniformCandidateSampler.java    |  171 +
 .../experimental/DummySeedGenerator.java      |   69 +
 .../org/tensorflow/op/signal/BatchFft.java    |   72 +
 .../org/tensorflow/op/signal/BatchFft2d.java  |   72 +
 .../org/tensorflow/op/signal/BatchFft3d.java  |   72 +
 .../org/tensorflow/op/signal/BatchIfft.java   |   72 +
 .../org/tensorflow/op/signal/BatchIfft2d.java |   72 +
 .../org/tensorflow/op/signal/BatchIfft3d.java |   72 +
 .../java/org/tensorflow/op/signal/Fft.java    |   83 +
 .../java/org/tensorflow/op/signal/Fft2d.java  |   83 +
 .../java/org/tensorflow/op/signal/Fft3d.java  |   83 +
 .../java/org/tensorflow/op/signal/Ifft.java   |   83 +
 .../java/org/tensorflow/op/signal/Ifft2d.java |   83 +
 .../java/org/tensorflow/op/signal/Ifft3d.java |   83 +
 .../java/org/tensorflow/op/signal/Irfft.java  |  115 +
 .../org/tensorflow/op/signal/Irfft2d.java     |  116 +
 .../org/tensorflow/op/signal/Irfft3d.java     |  116 +
 .../java/org/tensorflow/op/signal/Rfft.java   |   98 +
 .../java/org/tensorflow/op/signal/Rfft2d.java |  100 +
 .../java/org/tensorflow/op/signal/Rfft3d.java |  100 +
 .../op/sparse/AddManySparseToTensorsMap.java  |  158 +
 .../op/sparse/AddSparseToTensorsMap.java      |  149 +
 .../op/sparse/DenseCountSparseOutput.java     |  148 +
 .../op/sparse/DenseToDenseSetOperation.java   |  139 +
 .../op/sparse/DenseToSparseSetOperation.java  |  154 +
 .../op/sparse/DeserializeSparse.java          |  131 +
 .../SparseAccumulatorApplyGradient.java       |   75 +
 .../sparse/SparseAccumulatorTakeGradient.java |  103 +
 .../org/tensorflow/op/sparse/SparseAdd.java   |  114 +
 .../tensorflow/op/sparse/SparseAddGrad.java   |   95 +
 .../tensorflow/op/sparse/SparseBincount.java  |  129 +
 .../tensorflow/op/sparse/SparseConcat.java    |  139 +
 .../sparse/SparseConditionalAccumulator.java  |  160 +
 .../op/sparse/SparseCountSparseOutput.java    |  152 +
 .../org/tensorflow/op/sparse/SparseCross.java |  134 +
 .../op/sparse/SparseCrossHashed.java          |  139 +
 .../op/sparse/SparseDenseCwiseAdd.java        |   92 +
 .../op/sparse/SparseDenseCwiseDiv.java        |   86 +
 .../op/sparse/SparseDenseCwiseMul.java        |   90 +
 .../op/sparse/SparseFillEmptyRows.java        |  144 +
 .../op/sparse/SparseFillEmptyRowsGrad.java    |   91 +
 .../tensorflow/op/sparse/SparseMatMul.java    |  176 +
 .../tensorflow/op/sparse/SparseReduceMax.java |  131 +
 .../op/sparse/SparseReduceMaxSparse.java      |  141 +
 .../tensorflow/op/sparse/SparseReduceSum.java |  131 +
 .../op/sparse/SparseReduceSumSparse.java      |  141 +
 .../tensorflow/op/sparse/SparseReorder.java   |   95 +
 .../tensorflow/op/sparse/SparseReshape.java   |  101 +
 .../op/sparse/SparseSegmentMean.java          |   85 +
 .../op/sparse/SparseSegmentMeanGrad.java      |   84 +
 .../SparseSegmentMeanWithNumSegments.java     |   89 +
 .../op/sparse/SparseSegmentSqrtN.java         |   85 +
 .../op/sparse/SparseSegmentSqrtNGrad.java     |   84 +
 .../SparseSegmentSqrtNWithNumSegments.java    |   91 +
 .../op/sparse/SparseSegmentSum.java           |  110 +
 .../SparseSegmentSumWithNumSegments.java      |  110 +
 .../org/tensorflow/op/sparse/SparseSlice.java |  116 +
 .../tensorflow/op/sparse/SparseSliceGrad.java |   87 +
 .../tensorflow/op/sparse/SparseSoftmax.java   |   97 +
 .../op/sparse/SparseSparseMaximum.java        |   93 +
 .../op/sparse/SparseSparseMinimum.java        |   93 +
 .../org/tensorflow/op/sparse/SparseSplit.java |  128 +
 .../op/sparse/SparseTensorDenseAdd.java       |   83 +
 .../op/sparse/SparseTensorDenseMatMul.java    |  149 +
 .../tensorflow/op/sparse/SparseToDense.java   |  139 +
 .../op/sparse/SparseToSparseSetOperation.java |  169 +
 .../sparse/TakeManySparseFromTensorsMap.java  |  195 +
 .../java/org/tensorflow/op/strings/Join.java  |  117 +
 .../java/org/tensorflow/op/strings/Lower.java |  112 +
 .../org/tensorflow/op/strings/ReduceJoin.java |  156 +
 .../tensorflow/op/strings/RegexFullMatch.java |   90 +
 .../tensorflow/op/strings/RegexReplace.java   |  119 +
 .../op/strings/StaticRegexFullMatch.java      |   82 +
 .../op/strings/StaticRegexReplace.java        |  114 +
 .../tensorflow/op/strings/StringFormat.java   |  148 +
 .../tensorflow/op/strings/StringLength.java   |  126 +
 .../tensorflow/op/strings/StringNGrams.java   |  109 +
 .../tensorflow/op/strings/StringSplit.java    |  144 +
 .../java/org/tensorflow/op/strings/Strip.java |   78 +
 .../org/tensorflow/op/strings/Substr.java     |  196 +
 .../tensorflow/op/strings/ToHashBucket.java   |   83 +
 .../op/strings/ToHashBucketFast.java          |   88 +
 .../op/strings/ToHashBucketStrong.java        |  103 +
 .../org/tensorflow/op/strings/ToNumber.java   |  101 +
 .../tensorflow/op/strings/UnicodeDecode.java  |  212 +
 .../op/strings/UnicodeDecodeWithOffsets.java  |  228 +
 .../tensorflow/op/strings/UnicodeEncode.java  |  168 +
 .../tensorflow/op/strings/UnicodeScript.java  |   84 +
 .../op/strings/UnicodeTranscode.java          |  216 +
 .../op/strings/UnsortedSegmentJoin.java       |  139 +
 .../java/org/tensorflow/op/strings/Upper.java |  112 +
 .../tensorflow/op/summary/AudioSummary.java   |  127 +
 .../op/summary/CloseSummaryWriter.java        |   53 +
 .../op/summary/CreateSummaryDbWriter.java     |   62 +
 .../op/summary/CreateSummaryFileWriter.java   |   63 +
 .../op/summary/FlushSummaryWriter.java        |   53 +
 .../op/summary/HistogramSummary.java          |   82 +
 .../tensorflow/op/summary/ImageSummary.java   |  178 +
 .../tensorflow/op/summary/ImportEvent.java    |   56 +
 .../tensorflow/op/summary/MergeSummary.java   |   83 +
 .../tensorflow/op/summary/ScalarSummary.java  |   79 +
 .../op/summary/StatsAggregatorSummary.java    |   71 +
 .../tensorflow/op/summary/SummaryWriter.java  |  122 +
 .../tensorflow/op/summary/TensorSummary.java  |   78 +
 .../op/summary/WriteAudioSummary.java         |   98 +
 .../op/summary/WriteGraphSummary.java         |   59 +
 .../op/summary/WriteHistogramSummary.java     |   62 +
 .../op/summary/WriteImageSummary.java         |   99 +
 .../op/summary/WriteRawProtoSummary.java      |   59 +
 .../op/summary/WriteScalarSummary.java        |   62 +
 .../tensorflow/op/summary/WriteSummary.java   |   64 +
 .../java/org/tensorflow/op/tpu/AllToAll.java  |  103 +
 .../tensorflow/op/tpu/CollectivePermute.java  |   84 +
 .../op/tpu/ConfigureDistributedTPU.java       |  183 +
 .../op/tpu/ConfigureTPUEmbedding.java         |   54 +
 .../tensorflow/op/tpu/CrossReplicaSum.java    |   86 +
 .../tpu/EnqueueTPUEmbeddingIntegerBatch.java  |   99 +
 .../EnqueueTPUEmbeddingRaggedTensorBatch.java |  185 +
 .../tpu/EnqueueTPUEmbeddingSparseBatch.java   |  151 +
 .../EnqueueTPUEmbeddingSparseTensorBatch.java |  183 +
 .../org/tensorflow/op/tpu/InfeedDequeue.java  |   77 +
 .../tensorflow/op/tpu/InfeedDequeueTuple.java |   89 +
 .../org/tensorflow/op/tpu/InfeedEnqueue.java  |  141 +
 .../tpu/InfeedEnqueuePrelinearizedBuffer.java |   90 +
 .../tensorflow/op/tpu/InfeedEnqueueTuple.java |  130 +
 .../tpu/LoadTPUEmbeddingADAMParameters.java   |  141 +
 ...EmbeddingADAMParametersGradAccumDebug.java |  143 +
 .../LoadTPUEmbeddingAdadeltaParameters.java   |  141 +
 ...ddingAdadeltaParametersGradAccumDebug.java |  143 +
 .../LoadTPUEmbeddingAdagradParameters.java    |  139 +
 ...eddingAdagradParametersGradAccumDebug.java |  141 +
 ...TPUEmbeddingCenteredRMSPropParameters.java |  143 +
 .../tpu/LoadTPUEmbeddingFTRLParameters.java   |  141 +
 ...EmbeddingFTRLParametersGradAccumDebug.java |  143 +
 ...TPUEmbeddingMDLAdagradLightParameters.java |  143 +
 .../LoadTPUEmbeddingMomentumParameters.java   |  139 +
 ...ddingMomentumParametersGradAccumDebug.java |  141 +
 ...TPUEmbeddingProximalAdagradParameters.java |  139 +
 ...oximalAdagradParametersGradAccumDebug.java |  141 +
 ...oadTPUEmbeddingProximalYogiParameters.java |  134 +
 ...gProximalYogiParametersGradAccumDebug.java |  136 +
 .../LoadTPUEmbeddingRMSPropParameters.java    |  141 +
 ...eddingRMSPropParametersGradAccumDebug.java |  143 +
 ...ngStochasticGradientDescentParameters.java |  137 +
 ...adientDescentParametersGradAccumDebug.java |  139 +
 .../org/tensorflow/op/tpu/OutfeedDequeue.java |  117 +
 .../op/tpu/OutfeedDequeueTuple.java           |  130 +
 .../org/tensorflow/op/tpu/OutfeedEnqueue.java |   55 +
 .../op/tpu/OutfeedEnqueueTuple.java           |   56 +
 .../org/tensorflow/op/tpu/Prelinearize.java   |  135 +
 .../tensorflow/op/tpu/PrelinearizeTuple.java  |  125 +
 .../op/tpu/RecvTPUEmbeddingActivations.java   |   90 +
 .../RetrieveTPUEmbeddingADAMParameters.java   |  163 +
 ...EmbeddingADAMParametersGradAccumDebug.java |  172 +
 ...etrieveTPUEmbeddingAdadeltaParameters.java |  163 +
 ...ddingAdadeltaParametersGradAccumDebug.java |  172 +
 ...RetrieveTPUEmbeddingAdagradParameters.java |  154 +
 ...eddingAdagradParametersGradAccumDebug.java |  163 +
 ...TPUEmbeddingCenteredRMSPropParameters.java |  172 +
 .../RetrieveTPUEmbeddingFTRLParameters.java   |  163 +
 ...EmbeddingFTRLParametersGradAccumDebug.java |  172 +
 ...TPUEmbeddingMDLAdagradLightParameters.java |  172 +
 ...etrieveTPUEmbeddingMomentumParameters.java |  154 +
 ...ddingMomentumParametersGradAccumDebug.java |  163 +
 ...TPUEmbeddingProximalAdagradParameters.java |  154 +
 ...oximalAdagradParametersGradAccumDebug.java |  163 +
 ...eveTPUEmbeddingProximalYogiParameters.java |  154 +
 ...gProximalYogiParametersGradAccumDebug.java |  162 +
 ...RetrieveTPUEmbeddingRMSPropParameters.java |  163 +
 ...eddingRMSPropParametersGradAccumDebug.java |  172 +
 ...ngStochasticGradientDescentParameters.java |  151 +
 ...adientDescentParametersGradAccumDebug.java |  154 +
 .../op/tpu/SendTPUEmbeddingGradients.java     |   70 +
 .../op/tpu/ShutdownDistributedTPU.java        |   53 +
 .../op/tpu/TPUCompilationResult.java          |   73 +
 .../op/tpu/TPUEmbeddingActivations.java       |   85 +
 .../tensorflow/op/tpu/TPUOrdinalSelector.java |   74 +
 .../op/tpu/TPUReplicateMetadata.java          |  258 +
 .../tensorflow/op/tpu/TPUReplicatedInput.java |  158 +
 .../op/tpu/TPUReplicatedOutput.java           |   92 +
 .../tensorflow/op/tpu/WorkerHeartbeat.java    |   75 +
 .../op/train/AccumulatorApplyGradient.java    |   64 +
 .../op/train/AccumulatorNumAccumulated.java   |   74 +
 .../op/train/AccumulatorSetGlobalStep.java    |   62 +
 .../op/train/AccumulatorTakeGradient.java     |   88 +
 .../org/tensorflow/op/train/ApplyAdaMax.java  |  132 +
 .../tensorflow/op/train/ApplyAdadelta.java    |  128 +
 .../org/tensorflow/op/train/ApplyAdagrad.java |  141 +
 .../tensorflow/op/train/ApplyAdagradDa.java   |  126 +
 .../tensorflow/op/train/ApplyAdagradV2.java   |  142 +
 .../org/tensorflow/op/train/ApplyAdam.java    |  155 +
 .../org/tensorflow/op/train/ApplyAddSign.java |  129 +
 .../op/train/ApplyCenteredRmsProp.java        |  148 +
 .../org/tensorflow/op/train/ApplyFtrl.java    |  156 +
 .../op/train/ApplyGradientDescent.java        |  115 +
 .../tensorflow/op/train/ApplyMomentum.java    |  149 +
 .../tensorflow/op/train/ApplyPowerSign.java   |  129 +
 .../op/train/ApplyProximalAdagrad.java        |  125 +
 .../train/ApplyProximalGradientDescent.java   |  122 +
 .../org/tensorflow/op/train/ApplyRmsProp.java |  138 +
 .../org/tensorflow/op/train/BatchMatMul.java  |  154 +
 .../op/train/ConditionalAccumulator.java      |  160 +
 .../op/train/GenerateVocabRemapping.java      |  152 +
 .../op/train/MergeV2Checkpoints.java          |  102 +
 .../org/tensorflow/op/train/NegTrain.java     |   74 +
 .../tensorflow/op/train/PreventGradient.java  |  119 +
 .../ResourceAccumulatorApplyGradient.java     |   62 +
 .../ResourceAccumulatorNumAccumulated.java    |   72 +
 .../ResourceAccumulatorSetGlobalStep.java     |   60 +
 .../ResourceAccumulatorTakeGradient.java      |   86 +
 .../op/train/ResourceApplyAdaMax.java         |  113 +
 .../op/train/ResourceApplyAdadelta.java       |  109 +
 .../op/train/ResourceApplyAdagrad.java        |  123 +
 .../op/train/ResourceApplyAdagradDa.java      |  107 +
 .../op/train/ResourceApplyAdam.java           |  136 +
 .../train/ResourceApplyAdamWithAmsgrad.java   |  120 +
 .../op/train/ResourceApplyAddSign.java        |  110 +
 .../train/ResourceApplyCenteredRmsProp.java   |  129 +
 .../op/train/ResourceApplyFtrl.java           |  137 +
 .../train/ResourceApplyGradientDescent.java   |   96 +
 .../op/train/ResourceApplyKerasMomentum.java  |  130 +
 .../op/train/ResourceApplyMomentum.java       |  130 +
 .../op/train/ResourceApplyPowerSign.java      |  110 +
 .../train/ResourceApplyProximalAdagrad.java   |  106 +
 .../ResourceApplyProximalGradientDescent.java |  103 +
 .../op/train/ResourceApplyRmsProp.java        |  119 +
 .../train/ResourceConditionalAccumulator.java |  161 +
 .../op/train/ResourceSparseApplyAdadelta.java |  107 +
 .../op/train/ResourceSparseApplyAdagrad.java  |  126 +
 .../train/ResourceSparseApplyAdagradDa.java   |  110 +
 .../train/ResourceSparseApplyAdagradV2.java   |  127 +
 .../ResourceSparseApplyCenteredRmsProp.java   |  130 +
 .../op/train/ResourceSparseApplyFtrl.java     |  141 +
 .../ResourceSparseApplyKerasMomentum.java     |  135 +
 .../op/train/ResourceSparseApplyMomentum.java |  135 +
 .../ResourceSparseApplyProximalAdagrad.java   |  111 +
 ...rceSparseApplyProximalGradientDescent.java |  107 +
 .../op/train/ResourceSparseApplyRmsProp.java  |  122 +
 .../java/org/tensorflow/op/train/Restore.java |  107 +
 .../org/tensorflow/op/train/RestoreSlice.java |  128 +
 .../java/org/tensorflow/op/train/Save.java    |   69 +
 .../org/tensorflow/op/train/SaveSlices.java   |   95 +
 .../org/tensorflow/op/train/SdcaFprint.java   |   75 +
 .../tensorflow/op/train/SdcaOptimizer.java    |  185 +
 .../org/tensorflow/op/train/SdcaShrinkL1.java |   62 +
 .../op/train/SparseApplyAdadelta.java         |  126 +
 .../op/train/SparseApplyAdagrad.java          |  146 +
 .../op/train/SparseApplyAdagradDa.java        |  129 +
 .../op/train/SparseApplyCenteredRmsProp.java  |  149 +
 .../tensorflow/op/train/SparseApplyFtrl.java  |  160 +
 .../op/train/SparseApplyMomentum.java         |  154 +
 .../op/train/SparseApplyProximalAdagrad.java  |  130 +
 .../SparseApplyProximalGradientDescent.java   |  126 +
 .../op/train/SparseApplyRmsProp.java          |  141 +
 .../org/tensorflow/op/train/TileGrad.java     |   81 +
 .../tensorflow/op/xla/BroadcastHelper.java    |   88 +
 .../org/tensorflow/op/xla/ClusterOutput.java  |   74 +
 .../gen/java/org/tensorflow/op/xla/Conv.java  |   94 +
 .../org/tensorflow/op/xla/Dequantize.java     |   86 +
 .../gen/java/org/tensorflow/op/xla/Dot.java   |   83 +
 .../org/tensorflow/op/xla/DynamicSlice.java   |   91 +
 .../tensorflow/op/xla/DynamicUpdateSlice.java |   91 +
 .../java/org/tensorflow/op/xla/Einsum.java    |   81 +
 .../java/org/tensorflow/op/xla/Gather.java    |   85 +
 .../org/tensorflow/op/xla/KeyValueSort.java   |   88 +
 .../gen/java/org/tensorflow/op/xla/Pad.java   |   87 +
 .../gen/java/org/tensorflow/op/xla/Recv.java  |   83 +
 .../java/org/tensorflow/op/xla/ReplicaId.java |   70 +
 .../org/tensorflow/op/xla/SelfAdjointEig.java |   97 +
 .../gen/java/org/tensorflow/op/xla/Send.java  |   61 +
 .../java/org/tensorflow/op/xla/Sharding.java  |   74 +
 .../gen/java/org/tensorflow/op/xla/Sort.java  |   80 +
 .../gen/java/org/tensorflow/op/xla/Svd.java   |  103 +
 .../proto/distruntime/ClusterDef.java         |  865 ++
 .../distruntime/ClusterDefOrBuilder.java      |   53 +
 .../distruntime/ClusterDeviceFilters.java     |  773 ++
 .../ClusterDeviceFiltersOrBuilder.java        |   33 +
 .../proto/distruntime/ClusterProtos.java      |   76 +
 .../distruntime/DeviceFiltersProtos.java      |   91 +
 .../tensorflow/proto/distruntime/JobDef.java  |  935 ++
 .../proto/distruntime/JobDefOrBuilder.java    |   96 +
 .../proto/distruntime/JobDeviceFilters.java   |  902 ++
 .../JobDeviceFiltersOrBuilder.java            |   81 +
 .../proto/distruntime/ServerDef.java          | 1611 ++++
 .../proto/distruntime/ServerDefOrBuilder.java |  149 +
 .../proto/distruntime/ServerProtos.java       |   66 +
 .../proto/distruntime/TaskDeviceFilters.java  |  600 ++
 .../TaskDeviceFiltersOrBuilder.java           |   28 +
 .../proto/distruntime/TransportOptions.java   |  635 ++
 .../tensorflow/proto/example/BytesList.java   |  572 ++
 .../proto/example/BytesListOrBuilder.java     |   22 +
 .../org/tensorflow/proto/example/Example.java |  600 ++
 .../proto/example/ExampleOrBuilder.java       |   22 +
 .../example/ExampleParserConfiguration.java   |  695 ++
 .../ExampleParserConfigurationOrBuilder.java  |   43 +
 .../ExampleParserConfigurationProtos.java     |  123 +
 .../proto/example/ExampleProtos.java          |   68 +
 .../org/tensorflow/proto/example/Feature.java | 1105 +++
 .../proto/example/FeatureConfiguration.java   |  893 ++
 .../FeatureConfigurationOrBuilder.java        |   37 +
 .../tensorflow/proto/example/FeatureList.java |  781 ++
 .../proto/example/FeatureListOrBuilder.java   |   33 +
 .../proto/example/FeatureLists.java           |  739 ++
 .../proto/example/FeatureListsOrBuilder.java  |   63 +
 .../proto/example/FeatureOrBuilder.java       |   50 +
 .../proto/example/FeatureProtos.java          |  153 +
 .../tensorflow/proto/example/Features.java    |  739 ++
 .../proto/example/FeaturesOrBuilder.java      |   63 +
 .../proto/example/FixedLenFeatureProto.java   |  993 +++
 .../FixedLenFeatureProtoOrBuilder.java        |   54 +
 .../tensorflow/proto/example/FloatList.java   |  579 ++
 .../proto/example/FloatListOrBuilder.java     |   22 +
 .../tensorflow/proto/example/Int64List.java   |  582 ++
 .../proto/example/Int64ListOrBuilder.java     |   22 +
 .../proto/example/SequenceExample.java        |  781 ++
 .../example/SequenceExampleOrBuilder.java     |   35 +
 .../proto/example/VarLenFeatureProto.java     |  885 ++
 .../example/VarLenFeatureProtoOrBuilder.java  |   48 +
 .../framework/AllocationDescription.java      |  944 ++
 .../AllocationDescriptionOrBuilder.java       |   72 +
 .../AllocationDescriptionProtos.java          |   56 +
 .../proto/framework/AllocationRecord.java     |  575 ++
 .../framework/AllocationRecordOrBuilder.java  |   27 +
 .../proto/framework/AllocatorMemoryUsed.java  | 1268 +++
 .../AllocatorMemoryUsedOrBuilder.java         |   96 +
 .../tensorflow/proto/framework/ApiDef.java    | 6303 ++++++++++++++
 .../proto/framework/ApiDefOrBuilder.java      |  274 +
 .../proto/framework/ApiDefProtos.java         |  117 +
 .../tensorflow/proto/framework/ApiDefs.java   |  765 ++
 .../proto/framework/ApiDefsOrBuilder.java     |   33 +
 .../proto/framework/AssetFileDef.java         |  827 ++
 .../framework/AssetFileDefOrBuilder.java      |   56 +
 .../tensorflow/proto/framework/AttrValue.java | 5341 ++++++++++++
 .../proto/framework/AttrValueOrBuilder.java   |  203 +
 .../proto/framework/AttrValueProtos.java      |  110 +
 .../proto/framework/AutoParallelOptions.java  |  534 ++
 .../AutoParallelOptionsOrBuilder.java         |   19 +
 .../framework/BoundedTensorSpecProto.java     | 1182 +++
 .../BoundedTensorSpecProtoOrBuilder.java      |   67 +
 .../proto/framework/CallableOptions.java      | 2902 +++++++
 .../framework/CallableOptionsOrBuilder.java   |  485 ++
 .../org/tensorflow/proto/framework/Code.java  |  536 ++
 .../proto/framework/CollectionDef.java        | 4858 +++++++++++
 .../framework/CollectionDefOrBuilder.java     |   76 +
 .../proto/framework/CondContextDef.java       | 1632 ++++
 .../framework/CondContextDefOrBuilder.java    |  141 +
 .../proto/framework/ConfigProto.java          | 6064 +++++++++++++
 .../proto/framework/ConfigProtoOrBuilder.java |  477 +
 .../proto/framework/ConfigProtos.java         |  393 +
 .../framework/ControlFlowContextDef.java      |  903 ++
 .../ControlFlowContextDefOrBuilder.java       |   37 +
 .../proto/framework/ControlFlowProtos.java    |  116 +
 .../proto/framework/CostGraphDef.java         | 5817 +++++++++++++
 .../framework/CostGraphDefOrBuilder.java      |   57 +
 .../proto/framework/CostGraphProtos.java      |  122 +
 .../tensorflow/proto/framework/DataClass.java |  167 +
 .../tensorflow/proto/framework/DataType.java  |  615 ++
 .../proto/framework/DebugOptions.java         | 1033 +++
 .../framework/DebugOptionsOrBuilder.java      |   76 +
 .../proto/framework/DebugProtos.java          |   95 +
 .../proto/framework/DebugTensorWatch.java     | 1444 +++
 .../framework/DebugTensorWatchOrBuilder.java  |  196 +
 .../proto/framework/DebuggedSourceFile.java   | 1102 +++
 .../DebuggedSourceFileOrBuilder.java          |   98 +
 .../proto/framework/DebuggedSourceFiles.java  |  857 ++
 .../DebuggedSourceFilesOrBuilder.java         |   53 +
 .../proto/framework/DeviceAttributes.java     | 1277 +++
 .../framework/DeviceAttributesOrBuilder.java  |  110 +
 .../framework/DeviceAttributesProtos.java     |   94 +
 .../proto/framework/DeviceLocality.java       |  798 ++
 .../framework/DeviceLocalityOrBuilder.java    |   53 +
 .../proto/framework/DeviceProperties.java     | 1885 ++++
 .../framework/DevicePropertiesOrBuilder.java  |  204 +
 .../framework/DevicePropertiesProtos.java     |   85 +
 .../proto/framework/DeviceStepStats.java      | 1209 +++
 .../framework/DeviceStepStatsOrBuilder.java   |   97 +
 .../tensorflow/proto/framework/DictValue.java |  705 ++
 .../proto/framework/DictValueOrBuilder.java   |   43 +
 .../proto/framework/ErrorCodes.java           |   40 +
 .../proto/framework/ErrorCodesProtos.java     |   49 +
 .../proto/framework/FunctionDef.java          | 3415 ++++++++
 .../proto/framework/FunctionDefLibrary.java   | 1116 +++
 .../FunctionDefLibraryOrBuilder.java          |   57 +
 .../proto/framework/FunctionDefOrBuilder.java |  381 +
 .../proto/framework/FunctionProtos.java       |  184 +
 .../proto/framework/FunctionSpec.java         |  961 ++
 .../framework/FunctionSpecOrBuilder.java      |   68 +
 .../proto/framework/GPUOptions.java           | 5061 +++++++++++
 .../proto/framework/GPUOptionsOrBuilder.java  |  206 +
 .../proto/framework/GradientDef.java          |  765 ++
 .../proto/framework/GradientDefOrBuilder.java |   45 +
 .../proto/framework/GraphDebugInfo.java       | 3004 +++++++
 .../framework/GraphDebugInfoOrBuilder.java    |  147 +
 .../proto/framework/GraphDebugInfoProtos.java |   92 +
 .../tensorflow/proto/framework/GraphDef.java  | 1588 ++++
 .../proto/framework/GraphDefOrBuilder.java    |  163 +
 .../proto/framework/GraphOptions.java         | 1462 ++++
 .../framework/GraphOptionsOrBuilder.java      |  139 +
 .../proto/framework/GraphProtos.java          |   63 +
 .../framework/GraphTransferConstNodeInfo.java |  912 ++
 .../GraphTransferConstNodeInfoOrBuilder.java  |   51 +
 .../GraphTransferGraphInputNodeInfo.java      |  794 ++
 ...phTransferGraphInputNodeInfoOrBuilder.java |   41 +
 .../GraphTransferGraphOutputNodeInfo.java     |  794 ++
 ...hTransferGraphOutputNodeInfoOrBuilder.java |   41 +
 .../proto/framework/GraphTransferInfo.java    | 2795 ++++++
 .../framework/GraphTransferInfoOrBuilder.java |  190 +
 .../framework/GraphTransferInfoProto.java     |  163 +
 .../framework/GraphTransferNodeInfo.java      |  958 ++
 .../GraphTransferNodeInfoOrBuilder.java       |   54 +
 .../framework/GraphTransferNodeInput.java     |  533 ++
 .../framework/GraphTransferNodeInputInfo.java |  822 ++
 .../GraphTransferNodeInputInfoOrBuilder.java  |   38 +
 .../GraphTransferNodeInputOrBuilder.java      |   19 +
 .../GraphTransferNodeOutputInfo.java          |  639 ++
 .../GraphTransferNodeOutputInfoOrBuilder.java |   27 +
 .../proto/framework/HistogramProto.java       | 1120 +++
 .../framework/HistogramProtoOrBuilder.java    |   84 +
 .../proto/framework/InterconnectLink.java     |  660 ++
 .../framework/InterconnectLinkOrBuilder.java  |   29 +
 .../tensorflow/proto/framework/KernelDef.java | 2420 ++++++
 .../proto/framework/KernelDefOrBuilder.java   |  141 +
 .../proto/framework/KernelDefProtos.java      |   83 +
 .../proto/framework/KernelList.java           |  773 ++
 .../proto/framework/KernelListOrBuilder.java  |   33 +
 .../tensorflow/proto/framework/ListValue.java |  773 ++
 .../proto/framework/ListValueOrBuilder.java   |   33 +
 .../proto/framework/LocalLinks.java           |  765 ++
 .../proto/framework/LocalLinksOrBuilder.java  |   33 +
 .../proto/framework/LogMemoryProtos.java      |  125 +
 .../framework/MemoryLogRawAllocation.java     | 1029 +++
 .../MemoryLogRawAllocationOrBuilder.java      |   82 +
 .../framework/MemoryLogRawDeallocation.java   |  959 ++
 .../MemoryLogRawDeallocationOrBuilder.java    |   74 +
 .../proto/framework/MemoryLogStep.java        |  648 ++
 .../framework/MemoryLogStepOrBuilder.java     |   36 +
 .../framework/MemoryLogTensorAllocation.java  |  884 ++
 .../MemoryLogTensorAllocationOrBuilder.java   |   63 +
 .../MemoryLogTensorDeallocation.java          |  652 ++
 .../MemoryLogTensorDeallocationOrBuilder.java |   37 +
 .../framework/MemoryLogTensorOutput.java      |  957 ++
 .../MemoryLogTensorOutputOrBuilder.java       |   72 +
 .../proto/framework/MemoryStats.java          |  981 +++
 .../proto/framework/MemoryStatsOrBuilder.java |   55 +
 .../proto/framework/MetaGraphDef.java         | 4702 ++++++++++
 .../framework/MetaGraphDefOrBuilder.java      |  259 +
 .../proto/framework/MetaGraphProtos.java      |  318 +
 .../proto/framework/NameAttrList.java         |  832 ++
 .../framework/NameAttrListOrBuilder.java      |   53 +
 .../proto/framework/NamedDevice.java          |  727 ++
 .../proto/framework/NamedDeviceOrBuilder.java |   32 +
 .../proto/framework/NamedTensorProto.java     |  859 ++
 .../framework/NamedTensorProtoOrBuilder.java  |   64 +
 .../proto/framework/NamedTensorProtos.java    |   55 +
 .../proto/framework/NamedTupleValue.java      |  900 ++
 .../framework/NamedTupleValueOrBuilder.java   |   43 +
 .../tensorflow/proto/framework/NodeDef.java   | 3076 +++++++
 .../proto/framework/NodeDefOrBuilder.java     |  284 +
 .../proto/framework/NodeExecStats.java        | 2580 ++++++
 .../framework/NodeExecStatsOrBuilder.java     |  183 +
 .../proto/framework/NodeOutput.java           |  665 ++
 .../proto/framework/NodeOutputOrBuilder.java  |   27 +
 .../tensorflow/proto/framework/NodeProto.java |   84 +
 .../tensorflow/proto/framework/NoneValue.java |  427 +
 .../proto/framework/NoneValueOrBuilder.java   |    9 +
 .../org/tensorflow/proto/framework/OpDef.java | 6284 ++++++++++++++
 .../proto/framework/OpDefOrBuilder.java       |  296 +
 .../proto/framework/OpDefProtos.java          |  121 +
 .../proto/framework/OpDeprecation.java        |  655 ++
 .../framework/OpDeprecationOrBuilder.java     |   36 +
 .../tensorflow/proto/framework/OpList.java    |  773 ++
 .../proto/framework/OpListOrBuilder.java      |   33 +
 .../proto/framework/OptimizerOptions.java     | 1210 +++
 .../framework/OptimizerOptionsOrBuilder.java  |   77 +
 .../tensorflow/proto/framework/PairValue.java |  735 ++
 .../proto/framework/PairValueOrBuilder.java   |   32 +
 .../proto/framework/QueueRunnerDef.java       | 1435 +++
 .../framework/QueueRunnerDefOrBuilder.java    |  145 +
 .../proto/framework/QueueRunnerProtos.java    |   58 +
 .../proto/framework/RPCOptions.java           |  905 ++
 .../proto/framework/RPCOptionsOrBuilder.java  |   72 +
 .../proto/framework/ReaderBaseProtos.java     |   54 +
 .../proto/framework/ReaderBaseState.java      |  664 ++
 .../framework/ReaderBaseStateOrBuilder.java   |   29 +
 .../RemoteFusedGraphExecuteInfo.java          | 3007 +++++++
 .../RemoteFusedGraphExecuteInfoOrBuilder.java |  239 +
 .../RemoteFusedGraphExecuteInfoProto.java     |   85 +
 .../proto/framework/RemoteTensorHandle.java   | 1441 +++
 .../RemoteTensorHandleOrBuilder.java          |  128 +
 .../framework/RemoteTensorHandleProtos.java   |   76 +
 .../framework/ResourceDtypeAndShape.java      |  685 ++
 .../ResourceDtypeAndShapeOrBuilder.java       |   31 +
 .../proto/framework/ResourceHandle.java       |   75 +
 .../proto/framework/ResourceHandleProto.java  | 2288 +++++
 .../ResourceHandleProtoOrBuilder.java         |  137 +
 .../proto/framework/RewriterConfig.java       | 6064 +++++++++++++
 .../framework/RewriterConfigOrBuilder.java    |  627 ++
 .../proto/framework/RewriterConfigProtos.java |  159 +
 .../proto/framework/RunMetadata.java          | 3277 +++++++
 .../proto/framework/RunMetadataOrBuilder.java |  198 +
 .../proto/framework/RunOptions.java           | 2708 ++++++
 .../proto/framework/RunOptionsOrBuilder.java  |  101 +
 .../proto/framework/SaveSliceInfoDef.java     | 1175 +++
 .../framework/SaveSliceInfoDefOrBuilder.java  |  102 +
 .../proto/framework/SaveableObject.java       |  549 ++
 .../framework/SaveableObjectOrBuilder.java    |   23 +
 .../proto/framework/SavedAsset.java           |  514 ++
 .../proto/framework/SavedAssetOrBuilder.java  |   20 +
 .../framework/SavedBareConcreteFunction.java  |  873 ++
 .../SavedBareConcreteFunctionOrBuilder.java   |   71 +
 .../framework/SavedConcreteFunction.java      | 1156 +++
 .../SavedConcreteFunctionOrBuilder.java       |  102 +
 .../proto/framework/SavedConstant.java        |  574 ++
 .../framework/SavedConstantOrBuilder.java     |   27 +
 .../proto/framework/SavedFunction.java        |  781 ++
 .../framework/SavedFunctionOrBuilder.java     |   41 +
 .../proto/framework/SavedModel.java           |  949 ++
 .../proto/framework/SavedModelOrBuilder.java  |   64 +
 .../proto/framework/SavedModelProtos.java     |   56 +
 .../proto/framework/SavedObject.java          | 3174 +++++++
 .../proto/framework/SavedObjectGraph.java     | 1231 +++
 .../framework/SavedObjectGraphOrBuilder.java  |  122 +
 .../framework/SavedObjectGraphProtos.java     |  263 +
 .../proto/framework/SavedObjectOrBuilder.java |  249 +
 .../proto/framework/SavedResource.java        |  600 ++
 .../framework/SavedResourceOrBuilder.java     |   31 +
 .../proto/framework/SavedUserObject.java      |  974 +++
 .../framework/SavedUserObjectOrBuilder.java   |   70 +
 .../proto/framework/SavedVariable.java        | 1050 +++
 .../framework/SavedVariableOrBuilder.java     |   64 +
 .../framework/ScopedAllocatorOptions.java     |  644 ++
 .../ScopedAllocatorOptionsOrBuilder.java      |   44 +
 .../proto/framework/SessionMetadata.java      |  636 ++
 .../framework/SessionMetadataOrBuilder.java   |   28 +
 .../proto/framework/SignatureDef.java         | 1339 +++
 .../framework/SignatureDefOrBuilder.java      |  147 +
 .../tensorflow/proto/framework/StepStats.java |  765 ++
 .../proto/framework/StepStatsOrBuilder.java   |   33 +
 .../proto/framework/StepStatsProtos.java      |  169 +
 .../proto/framework/StructProtos.java         |  215 +
 .../proto/framework/StructuredValue.java      | 3423 ++++++++
 .../framework/StructuredValueOrBuilder.java   |  309 +
 .../tensorflow/proto/framework/Summary.java   | 4725 ++++++++++
 .../proto/framework/SummaryDescription.java   |  589 ++
 .../SummaryDescriptionOrBuilder.java          |   29 +
 .../proto/framework/SummaryMetadata.java      | 1784 ++++
 .../framework/SummaryMetadataOrBuilder.java   |   93 +
 .../proto/framework/SummaryOrBuilder.java     |   53 +
 .../proto/framework/SummaryProtos.java        |  159 +
 .../proto/framework/TensorConnection.java     |  751 ++
 .../framework/TensorConnectionOrBuilder.java  |   49 +
 .../proto/framework/TensorDescription.java    |  990 +++
 .../framework/TensorDescriptionOrBuilder.java |   76 +
 .../framework/TensorDescriptionProtos.java    |   65 +
 .../proto/framework/TensorInfo.java           | 3680 ++++++++
 .../proto/framework/TensorInfoOrBuilder.java  |  128 +
 .../proto/framework/TensorProto.java          | 3980 +++++++++
 .../proto/framework/TensorProtoOrBuilder.java |  440 +
 .../proto/framework/TensorProtos.java         |   86 +
 .../proto/framework/TensorShapeProto.java     | 1852 ++++
 .../framework/TensorShapeProtoOrBuilder.java  |  108 +
 .../proto/framework/TensorShapeProtos.java    |   65 +
 .../proto/framework/TensorSliceProto.java     | 1589 ++++
 .../framework/TensorSliceProtoOrBuilder.java  |   68 +
 .../proto/framework/TensorSliceProtos.java    |   65 +
 .../proto/framework/TensorSpecProto.java      |  820 ++
 .../framework/TensorSpecProtoOrBuilder.java   |   41 +
 .../framework/ThreadPoolOptionProto.java      |  739 ++
 .../ThreadPoolOptionProtoOrBuilder.java       |   62 +
 .../proto/framework/TrackableObjectGraph.java | 5141 +++++++++++
 .../TrackableObjectGraphOrBuilder.java        |   33 +
 .../framework/TrackableObjectGraphProtos.java |  111 +
 .../proto/framework/TupleValue.java           |  773 ++
 .../proto/framework/TupleValueOrBuilder.java  |   33 +
 .../proto/framework/TypeSpecProto.java        | 1176 +++
 .../framework/TypeSpecProtoOrBuilder.java     |   67 +
 .../proto/framework/TypesProtos.java          |   60 +
 .../tensorflow/proto/framework/ValuesDef.java |  969 +++
 .../proto/framework/ValuesDefOrBuilder.java   |   98 +
 .../proto/framework/VariableAggregation.java  |  165 +
 .../proto/framework/VariableDef.java          | 1650 ++++
 .../proto/framework/VariableDefOrBuilder.java |  158 +
 .../proto/framework/VariableProtos.java       |   81 +
 .../framework/VariableSynchronization.java    |  169 +
 .../framework/VariantTensorDataProto.java     | 1097 +++
 .../VariantTensorDataProtoOrBuilder.java      |   80 +
 .../proto/framework/VerifierConfig.java       |  725 ++
 .../framework/VerifierConfigOrBuilder.java    |   36 +
 .../proto/framework/VerifierConfigProtos.java |   55 +
 .../proto/framework/VersionDef.java           |  792 ++
 .../proto/framework/VersionDefOrBuilder.java  |   52 +
 .../proto/framework/VersionsProtos.java       |   52 +
 .../proto/framework/WhileContextDef.java      | 2534 ++++++
 .../framework/WhileContextDefOrBuilder.java   |  265 +
 .../proto/profiler/ProfileOptions.java        | 1176 +++
 .../profiler/ProfileOptionsOrBuilder.java     |  103 +
 .../proto/profiler/ProfilerOptionsProtos.java |   56 +
 .../org/tensorflow/proto/profiler/XEvent.java | 1263 +++
 .../proto/profiler/XEventMetadata.java        |  890 ++
 .../profiler/XEventMetadataOrBuilder.java     |   63 +
 .../proto/profiler/XEventOrBuilder.java       |   92 +
 .../org/tensorflow/proto/profiler/XLine.java  | 1508 ++++
 .../proto/profiler/XLineOrBuilder.java        |  133 +
 .../org/tensorflow/proto/profiler/XPlane.java | 2168 +++++
 .../proto/profiler/XPlaneOrBuilder.java       |  243 +
 .../proto/profiler/XPlaneProtos.java          |  167 +
 .../org/tensorflow/proto/profiler/XSpace.java | 1219 +++
 .../proto/profiler/XSpaceOrBuilder.java       |  103 +
 .../org/tensorflow/proto/profiler/XStat.java  | 1111 +++
 .../proto/profiler/XStatMetadata.java         |  813 ++
 .../profiler/XStatMetadataOrBuilder.java      |   54 +
 .../proto/profiler/XStatOrBuilder.java        |   59 +
 .../proto/util/BfcMemoryMapProtos.java        |  112 +
 .../org/tensorflow/proto/util/BinSummary.java |  708 ++
 .../proto/util/BinSummaryOrBuilder.java       |   34 +
 .../proto/util/BundleEntryProto.java          | 1540 ++++
 .../proto/util/BundleEntryProtoOrBuilder.java |  137 +
 .../proto/util/BundleHeaderProto.java         |  929 ++
 .../util/BundleHeaderProtoOrBuilder.java      |   52 +
 .../tensorflow/proto/util/CodeLocation.java   |  837 ++
 .../proto/util/CodeLocationOrBuilder.java     |   70 +
 .../org/tensorflow/proto/util/DebugEvent.java | 2883 ++++++
 .../proto/util/DebugEventOrBuilder.java       |  258 +
 .../proto/util/DebugEventProtos.java          |  205 +
 .../tensorflow/proto/util/DebugMetadata.java  |  920 ++
 .../proto/util/DebugMetadataOrBuilder.java    |   71 +
 .../tensorflow/proto/util/DebuggedDevice.java |  667 ++
 .../proto/util/DebuggedDeviceOrBuilder.java   |   39 +
 .../tensorflow/proto/util/DebuggedGraph.java  | 1295 +++
 .../proto/util/DebuggedGraphOrBuilder.java    |  123 +
 .../java/org/tensorflow/proto/util/Event.java | 2049 +++++
 .../tensorflow/proto/util/EventOrBuilder.java |  174 +
 .../tensorflow/proto/util/EventProtos.java    |  161 +
 .../org/tensorflow/proto/util/Execution.java  | 2259 +++++
 .../proto/util/ExecutionOrBuilder.java        |  233 +
 .../proto/util/GraphExecutionTrace.java       | 1359 +++
 .../util/GraphExecutionTraceOrBuilder.java    |  125 +
 .../proto/util/GraphOpCreation.java           | 1936 +++++
 .../proto/util/GraphOpCreationOrBuilder.java  |  195 +
 .../org/tensorflow/proto/util/LogMessage.java |  787 ++
 .../proto/util/LogMessageOrBuilder.java       |   28 +
 .../proto/util/MemAllocatorStats.java         |  718 ++
 .../util/MemAllocatorStatsOrBuilder.java      |   34 +
 .../org/tensorflow/proto/util/MemChunk.java   | 1009 +++
 .../proto/util/MemChunkOrBuilder.java         |   59 +
 .../util/MemmappedFileSystemDirectory.java    |  773 ++
 .../MemmappedFileSystemDirectoryElement.java  |  670 ++
 ...edFileSystemDirectoryElementOrBuilder.java |   29 +
 ...MemmappedFileSystemDirectoryOrBuilder.java |   33 +
 .../proto/util/MemmappedFileSystemProtos.java |   64 +
 .../org/tensorflow/proto/util/MemoryDump.java | 1759 ++++
 .../proto/util/MemoryDumpOrBuilder.java       |  104 +
 .../proto/util/RequestedExitCode.java         |  476 +
 .../util/RequestedExitCodeOrBuilder.java      |   14 +
 .../org/tensorflow/proto/util/SavedSlice.java | 1073 +++
 .../tensorflow/proto/util/SavedSliceMeta.java | 1364 +++
 .../proto/util/SavedSliceMetaOrBuilder.java   |  113 +
 .../proto/util/SavedSliceOrBuilder.java       |   85 +
 .../proto/util/SavedTensorSliceMeta.java      | 1108 +++
 .../util/SavedTensorSliceMetaOrBuilder.java   |   81 +
 .../proto/util/SavedTensorSliceProtos.java    |  109 +
 .../proto/util/SavedTensorSlices.java         |  899 ++
 .../util/SavedTensorSlicesOrBuilder.java      |   62 +
 .../org/tensorflow/proto/util/SaverDef.java   | 1356 +++
 .../proto/util/SaverDefOrBuilder.java         |  104 +
 .../tensorflow/proto/util/SaverProtos.java    |   57 +
 .../org/tensorflow/proto/util/SessionLog.java |  910 ++
 .../proto/util/SessionLogOrBuilder.java       |   46 +
 .../org/tensorflow/proto/util/SnapShot.java   |  535 ++
 .../proto/util/SnapShotOrBuilder.java         |   19 +
 .../proto/util/SnapshotMetadataRecord.java    | 1328 +++
 .../util/SnapshotMetadataRecordOrBuilder.java |  121 +
 .../tensorflow/proto/util/SnapshotProtos.java |  102 +
 .../tensorflow/proto/util/SnapshotRecord.java |  777 ++
 .../proto/util/SnapshotRecordOrBuilder.java   |   33 +
 .../proto/util/SnapshotTensorMetadata.java    |  773 ++
 .../util/SnapshotTensorMetadataOrBuilder.java |   33 +
 .../org/tensorflow/proto/util/SourceFile.java |  964 ++
 .../proto/util/SourceFileOrBuilder.java       |   80 +
 .../proto/util/StackFrameWithId.java          |  835 ++
 .../proto/util/StackFrameWithIdOrBuilder.java |   58 +
 .../proto/util/TaggedRunMetadata.java         |  663 ++
 .../util/TaggedRunMetadataOrBuilder.java      |   37 +
 .../proto/util/TensorBundleProtos.java        |   83 +
 .../proto/util/TensorDebugMode.java           |  282 +
 .../tensorflow/proto/util/TensorMetadata.java |  682 ++
 .../proto/util/TensorMetadataOrBuilder.java   |   31 +
 .../tensorflow/proto/util/WatchdogConfig.java |  477 +
 .../proto/util/WatchdogConfigOrBuilder.java   |   14 +
 .../tensorflow/proto/util/WorkerHealth.java   |  141 +
 .../proto/util/WorkerHeartbeatRequest.java    |  866 ++
 .../util/WorkerHeartbeatRequestOrBuilder.java |   44 +
 .../proto/util/WorkerHeartbeatResponse.java   |  977 +++
 .../WorkerHeartbeatResponseOrBuilder.java     |   52 +
 .../proto/util/WorkerShutdownMode.java        |  126 +
 .../util/testlog/AvailableDeviceInfo.java     |  966 +++
 .../testlog/AvailableDeviceInfoOrBuilder.java |   72 +
 .../proto/util/testlog/BenchmarkEntries.java  |  765 ++
 .../testlog/BenchmarkEntriesOrBuilder.java    |   33 +
 .../proto/util/testlog/BenchmarkEntry.java    | 1676 ++++
 .../util/testlog/BenchmarkEntryOrBuilder.java |  168 +
 .../util/testlog/BuildConfiguration.java      | 1021 +++
 .../testlog/BuildConfigurationOrBuilder.java  |   97 +
 .../proto/util/testlog/CPUInfo.java           | 1254 +++
 .../proto/util/testlog/CPUInfoOrBuilder.java  |  122 +
 .../proto/util/testlog/CommitId.java          |  964 ++
 .../proto/util/testlog/CommitIdOrBuilder.java |   59 +
 .../proto/util/testlog/EntryValue.java        |  713 ++
 .../util/testlog/EntryValueOrBuilder.java     |   26 +
 .../proto/util/testlog/GPUInfo.java           |  884 ++
 .../proto/util/testlog/GPUInfoOrBuilder.java  |   63 +
 .../util/testlog/MachineConfiguration.java    | 2241 +++++
 .../MachineConfigurationOrBuilder.java        |  196 +
 .../proto/util/testlog/MemoryInfo.java        |  567 ++
 .../util/testlog/MemoryInfoOrBuilder.java     |   27 +
 .../proto/util/testlog/MetricEntry.java       | 1107 +++
 .../util/testlog/MetricEntryOrBuilder.java    |   86 +
 .../proto/util/testlog/PlatformInfo.java      | 1349 +++
 .../util/testlog/PlatformInfoOrBuilder.java   |  117 +
 .../proto/util/testlog/RunConfiguration.java  |  917 ++
 .../testlog/RunConfigurationOrBuilder.java    |   82 +
 .../proto/util/testlog/TestLogProtos.java     |  288 +
 .../proto/util/testlog/TestResults.java       | 2624 ++++++
 .../util/testlog/TestResultsOrBuilder.java    |  245 +
 1667 files changed, 460586 insertions(+), 11 deletions(-)
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/AudioOps.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/BitwiseOps.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/DataExperimentalOps.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/DataOps.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/DtypesOps.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/ImageOps.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/IoOps.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/LinalgOps.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/MathOps.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/NnRawOps.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/Ops.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/QuantizationOps.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/RaggedOps.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/RandomOps.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/ShapeOps.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/SignalOps.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/SparseOps.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/StringsOps.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/SummaryOps.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/TrainOps.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/XlaOps.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/Compute_func_Pointer_TF_OpKernelContext.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/Create_func_TF_OpKernelConstruction.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/Deallocator_Pointer_long_Pointer.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/Delete_func_Pointer.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/Listener_BytePointer.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/Listener_String.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/Shape_inference_func_TF_ShapeInferenceContext_TF_Status.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TFE_Context.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TFE_ContextOptions.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TFE_Op.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TFE_TensorDebugInfo.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TFE_TensorHandle.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_ApiDefMap.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_AttrMetadata.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_Buffer.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_DeprecatedSession.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_DeviceList.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_DimensionHandle.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_Function.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_FunctionOptions.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_Graph.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_ImportGraphDefOptions.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_ImportGraphDefResults.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_Input.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_KernelBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_Library.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_OpDefinitionBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_OpKernelConstruction.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_OpKernelContext.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_Operation.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_OperationDescription.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_Output.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_Server.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_Session.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_SessionOptions.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_ShapeHandle.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_ShapeInferenceContext.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_Status.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_Tensor.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_WhileParams.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/Tensor.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/global/tensorflow.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/audio/AudioSpectrogram.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/audio/DecodeWav.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/audio/EncodeWav.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/audio/Mfcc.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/bitwise/BitwiseAnd.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/bitwise/BitwiseOr.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/bitwise/BitwiseXor.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/bitwise/Invert.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/bitwise/LeftShift.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/bitwise/RightShift.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/cluster/KMC2ChainInitialization.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/cluster/KmeansPlusPlusInitialization.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/collective/AllReduce.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/collective/BroadcastRecv.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/collective/BroadcastSend.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Abort.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/All.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Any.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/AssertThat.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Assign.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/AssignAdd.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/AssignAddVariableOp.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/AssignSub.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/AssignSubVariableOp.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/AssignVariableOp.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Barrier.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BarrierClose.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BarrierIncompleteSize.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BarrierInsertMany.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BarrierReadySize.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BarrierTakeMany.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Batch.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BatchToSpace.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BatchToSpaceNd.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Bitcast.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BroadcastDynamicShape.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BroadcastGradientArgs.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BroadcastTo.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Bucketize.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ClipByValue.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/CollectiveGather.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Concat.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ConsumeMutexLock.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ControlTrigger.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Copy.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/CopyHost.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/CountUpTo.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/DecodeProto.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/DeepCopy.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/DeleteSessionTensor.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/DestroyResourceOp.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/DestroyTemporaryVariable.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/DeviceIndex.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/DummyMemoryCache.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/DynamicPartition.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/DynamicStitch.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/EditDistance.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Empty.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/EmptyTensorList.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/EncodeProto.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/EnsureShape.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Enter.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Exit.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ExpandDims.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ExtractVolumePatches.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Fill.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Fingerprint.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Gather.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/GatherNd.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/GetSessionHandle.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/GetSessionTensor.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/GuaranteeConst.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/HashTable.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/HistogramFixedWidth.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Identity.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/IdentityN.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ImmutableConst.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/InitializeTable.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/InitializeTableFromTextFile.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/InplaceAdd.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/InplaceSub.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/InplaceUpdate.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/IsVariableInitialized.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LinSpace.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LookupTableExport.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LookupTableFind.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LookupTableImport.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LookupTableInsert.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LookupTableRemove.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LookupTableSize.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LoopCond.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LowerBound.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MapClear.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MapIncompleteSize.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MapPeek.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MapSize.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MapStage.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MapUnstage.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MapUnstageNoKey.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Max.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Merge.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Min.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MirrorPad.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MirrorPadGrad.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MlirPassthroughOp.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MutableDenseHashTable.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MutableHashTable.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MutableHashTableOfTensors.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Mutex.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MutexLock.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/NcclAllReduce.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/NcclBroadcast.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/NcclReduce.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/NextIteration.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/NoOp.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OneHot.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OnesLike.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OrderedMapClear.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OrderedMapIncompleteSize.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OrderedMapPeek.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OrderedMapSize.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OrderedMapStage.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OrderedMapUnstage.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OrderedMapUnstageNoKey.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Pad.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ParallelConcat.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ParallelDynamicStitch.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Placeholder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/PlaceholderWithDefault.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Print.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Prod.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/QuantizedReshape.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Range.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Rank.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ReadVariableOp.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Recv.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ReduceAll.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ReduceAny.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ReduceMax.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ReduceMin.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ReduceProd.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ReduceSum.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/RefEnter.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/RefExit.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/RefIdentity.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/RefMerge.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/RefNextIteration.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/RefSelect.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/RefSwitch.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/RemoteFusedGraphExecute.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Reshape.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceCountUpTo.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceGather.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceGatherNd.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterAdd.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterDiv.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterMax.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterMin.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterMul.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterNdAdd.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterNdMax.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterNdMin.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterNdSub.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterNdUpdate.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterSub.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterUpdate.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceStridedSliceAssign.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Reverse.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ReverseSequence.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Roll.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Rpc.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterAdd.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterDiv.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterMax.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterMin.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterMul.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterNd.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterNdAdd.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterNdMax.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterNdMin.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterNdNonAliasingAdd.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterNdSub.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterNdUpdate.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterSub.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterUpdate.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Select.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Send.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/SetDiff1d.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/SetSize.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Shape.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ShapeN.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Size.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Skipgram.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Slice.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Snapshot.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/SpaceToBatchNd.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Split.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/SplitV.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Squeeze.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Stack.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Stage.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/StageClear.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/StagePeek.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/StageSize.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/StopGradient.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/StridedSlice.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/StridedSliceAssign.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/StridedSliceGrad.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Sum.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/SwitchCond.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TemporaryVariable.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArray.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayClose.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayConcat.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayGather.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayGrad.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayGradWithShape.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayPack.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayRead.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayScatter.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArraySize.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArraySplit.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayUnpack.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayWrite.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorForestCreateTreeVariable.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorForestTreeDeserialize.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorForestTreeIsInitializedOp.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorForestTreePredict.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorForestTreeResourceHandleOp.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorForestTreeSerialize.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorForestTreeSize.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListConcat.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListConcatLists.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListElementShape.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListFromTensor.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListGather.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListGetItem.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListLength.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListPopBack.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListPushBack.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListPushBackBatch.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListReserve.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListResize.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListScatter.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListScatterIntoExistingList.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListSetItem.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListSplit.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListStack.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorScatterNdAdd.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorScatterNdMax.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorScatterNdMin.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorScatterNdSub.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorScatterNdUpdate.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorStridedSliceUpdate.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Tile.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Timestamp.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TryRpc.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Unbatch.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/UnbatchGrad.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Unique.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/UniqueWithCounts.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/UnravelIndex.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Unstack.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Unstage.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/UpperBound.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/VarHandleOp.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/VarIsInitializedOp.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Variable.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/VariableShape.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Where.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/XlaSpmdFullToShardShape.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/XlaSpmdShardToFullShape.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ZerosLike.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/AnonymousIterator.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/AnonymousMemoryCache.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/AnonymousMultiDeviceIterator.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/AssertNextDataset.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/AutoShardDataset.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/BatchDataset.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/BytesProducedStatsDataset.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/CSVDataset.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/CacheDataset.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/CacheDatasetV2.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ChooseFastestDataset.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ConcatenateDataset.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DatasetCardinality.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DatasetFromGraph.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DatasetToGraph.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DatasetToSingleElement.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DatasetToTfRecord.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DeleteIterator.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DeleteMemoryCache.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DeleteMultiDeviceIterator.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DenseToSparseBatchDataset.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DeserializeIterator.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DirectedInterleaveDataset.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/FilterByLastComponentDataset.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/FixedLengthRecordDataset.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/IgnoreErrorsDataset.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/InitializeTableFromDataset.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/Iterator.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/IteratorFromStringHandle.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/IteratorGetDevice.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/IteratorGetNext.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/IteratorGetNextAsOptional.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/IteratorGetNextSync.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/IteratorToStringHandle.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/LMDBDataset.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/LatencyStatsDataset.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/LeakyReluGrad.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MakeIterator.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MatchingFilesDataset.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MaxIntraOpParallelismDataset.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ModelDataset.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MultiDeviceIterator.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MultiDeviceIteratorFromStringHandle.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MultiDeviceIteratorGetNextFromShard.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MultiDeviceIteratorInit.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MultiDeviceIteratorToStringHandle.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/NonSerializableDataset.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/OptimizeDataset.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/OptionalFromValue.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/OptionalGetValue.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/OptionalHasValue.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/OptionalNone.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/PaddedBatchDataset.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/PrefetchDataset.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/PrivateThreadPoolDataset.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/RandomDataset.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/RangeDataset.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/RebatchDataset.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/RegisterDataset.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/RepeatDataset.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SamplingDataset.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SerializeIterator.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SetStatsAggregatorDataset.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ShardDataset.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ShuffleAndRepeatDataset.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ShuffleDataset.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SkipDataset.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SleepDataset.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SlidingWindowDataset.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SparseTensorSliceDataset.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SqlDataset.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/StatsAggregatorHandle.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/TakeDataset.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/TensorDataset.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/TensorSliceDataset.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/TextLineDataset.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/TfRecordDataset.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ThreadPoolDataset.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ThreadPoolHandle.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/UnbatchDataset.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/UniqueDataset.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/UnwrapDatasetVariant.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/WindowDataset.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/WrapDatasetVariant.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ZipDataset.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/AssertCardinalityDataset.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/AssertNextDataset.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/AutoShardDataset.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/BytesProducedStatsDataset.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/CSVDataset.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/ChooseFastestDataset.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/CompressElement.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/DataServiceDataset.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/DatasetCardinality.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/DatasetToTFRecord.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/DenseToSparseBatchDataset.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/DirectedInterleaveDataset.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/DummyIterationCounter.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/IgnoreErrorsDataset.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/IteratorGetDevice.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/LatencyStatsDataset.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/LmdbDataset.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/MatchingFilesDataset.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/MaxIntraOpParallelismDataset.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/NonSerializableDataset.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/ParseExampleDataset.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/PrivateThreadPoolDataset.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/RandomDataset.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/RebatchDataset.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/SetStatsAggregatorDataset.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/SleepDataset.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/SlidingWindowDataset.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/SqlDataset.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/StatsAggregatorHandle.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/StatsAggregatorSetSummaryWriter.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/StatsAggregatorSummary.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/ThreadPoolDataset.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/ThreadPoolHandle.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/UnbatchDataset.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/UncompressElement.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/UniqueDataset.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/debugging/CheckNumerics.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/debugging/DebugGradientIdentity.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/debugging/DebugGradientRefIdentity.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/debugging/DebugIdentity.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/debugging/DebugNanCount.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/debugging/DebugNumericsSummary.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/dtypes/AsString.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/dtypes/Cast.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/dtypes/Complex.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/dtypes/ToBool.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesAggregateStats.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesBucketize.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesCalculateBestFeatureSplit.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesCalculateBestFeatureSplitV2.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesCalculateBestGainsPerFeature.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesCenterBias.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesCreateEnsemble.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesCreateQuantileStreamResource.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesDeserializeEnsemble.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesEnsembleResourceHandleOp.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesExampleDebugOutputs.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesFlushQuantileSummaries.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesGetEnsembleStates.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesMakeQuantileSummaries.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesMakeStatsSummary.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesPredict.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesQuantileStreamResourceAddSummaries.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesQuantileStreamResourceDeserialize.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesQuantileStreamResourceFlush.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesQuantileStreamResourceGetBucketBoundaries.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesQuantileStreamResourceHandleOp.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesSerializeEnsemble.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesSparseAggregateStats.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesSparseCalculateBestFeatureSplit.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesTrainingPredict.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesUpdateEnsemble.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesUpdateEnsembleV2.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/IsBoostedTreesEnsembleInitialized.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/IsBoostedTreesQuantileStreamResourceInitialized.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/AdjustContrast.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/AdjustHue.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/AdjustSaturation.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/CombinedNonMaxSuppression.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/CropAndResize.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/CropAndResizeGradBoxes.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/CropAndResizeGradImage.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/DecodeAndCropJpeg.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/DecodeBmp.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/DecodeGif.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/DecodeJpeg.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/DecodePng.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/DrawBoundingBoxes.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/EncodeJpeg.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/EncodeJpegVariableQuality.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/EncodePng.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ExtractGlimpse.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ExtractImagePatches.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ExtractJpegShape.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/GenerateBoundingBoxProposals.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/HsvToRgb.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ImageProjectiveTransformV2.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/NearestNeighbors.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/NonMaxSuppression.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/NonMaxSuppressionWithOverlaps.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/QuantizedResizeBilinear.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/RandomCrop.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ResizeArea.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ResizeBicubic.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ResizeBicubicGrad.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ResizeBilinear.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ResizeBilinearGrad.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ResizeNearestNeighbor.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ResizeNearestNeighborGrad.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/RgbToHsv.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/SampleDistortedBoundingBox.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ScaleAndTranslate.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ScaleAndTranslateGrad.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/DecodeBase64.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/DecodeCompressed.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/DecodeCsv.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/DecodeJsonExample.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/DecodePaddedRaw.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/DecodeRaw.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/DeserializeManySparse.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/EncodeBase64.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/FifoQueue.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/FixedLengthRecordReader.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/IdentityReader.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/LmdbReader.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/MatchingFiles.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/PaddingFifoQueue.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ParseExample.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ParseSequenceExample.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ParseSingleExample.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ParseSingleSequenceExample.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ParseTensor.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/PriorityQueue.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/QueueClose.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/QueueDequeue.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/QueueDequeueMany.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/QueueDequeueUpTo.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/QueueEnqueue.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/QueueEnqueueMany.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/QueueIsClosed.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/QueueSize.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/RandomShuffleQueue.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ReadFile.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ReaderNumRecordsProduced.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ReaderNumWorkUnitsCompleted.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ReaderRead.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ReaderReadUpTo.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ReaderReset.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ReaderRestoreState.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ReaderSerializeState.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/SerializeManySparse.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/SerializeSparse.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/SerializeTensor.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ShardedFilename.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ShardedFilespec.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/TextLineReader.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/TfRecordReader.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/WholeFileReader.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/WriteFile.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BandPart.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BandedTriangularSolve.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchCholesky.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchCholeskyGrad.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixBandPart.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixDeterminant.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixDiag.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixDiagPart.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixInverse.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixSetDiag.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixSolve.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixSolveLs.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixTriangularSolve.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchSelfAdjointEig.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchSvd.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Cholesky.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/CholeskyGrad.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/ConjugateTranspose.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Cross.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Det.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Eig.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Einsum.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/EuclideanNorm.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Inv.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/LoadAndRemapMatrix.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/LogMatrixDeterminant.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Lu.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/MatMul.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/MatrixDiag.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/MatrixDiagPart.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/MatrixDiagPartV3.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/MatrixDiagV3.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/MatrixLogarithm.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/MatrixSetDiag.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/MatrixSolveLs.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Qr.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/QuantizedMatMul.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/QuantizedMatMulWithBias.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/QuantizedMatMulWithBiasAndRelu.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/QuantizedMatMulWithBiasAndReluAndRequantize.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/SelfAdjointEig.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Solve.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Sqrtm.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Svd.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/TensorDiag.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/TensorDiagPart.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Transpose.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/TriangularSolve.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/TridiagonalMatMul.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/TridiagonalSolve.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/CSRSparseMatrixComponents.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/CSRSparseMatrixToDense.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/CSRSparseMatrixToSparseTensor.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/DenseToCSRSparseMatrix.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixAdd.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixMatMul.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixMul.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixNNZ.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixOrderingAMD.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixSoftmax.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixSoftmaxGrad.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixSparseCholesky.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixSparseMatMul.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixTranspose.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixZeros.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseTensorToCSRSparseMatrix.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Abs.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/AccumulateN.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Acos.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Acosh.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Add.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/AddN.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Angle.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/ApproximateEqual.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/ArgMax.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/ArgMin.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Asin.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Asinh.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Atan.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Atan2.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Atanh.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/BesselI0.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/BesselI0e.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/BesselI1.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/BesselI1e.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Betainc.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Bincount.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Ceil.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/CompareAndBitpack.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/ComplexAbs.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Conj.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Cos.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Cosh.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Cumprod.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Cumsum.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/CumulativeLogsumexp.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/DenseBincount.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Digamma.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Div.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/DivNoNan.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Equal.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Erf.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Erfc.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Exp.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Expm1.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Fact.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Floor.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/FloorDiv.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/FloorMod.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Greater.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/GreaterEqual.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Igamma.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/IgammaGradA.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Igammac.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Imag.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/InvertPermutation.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/IsFinite.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/IsInf.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/IsNan.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Less.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/LessEqual.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Lgamma.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Log.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Log1p.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/LogicalAnd.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/LogicalNot.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/LogicalOr.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Maximum.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Mean.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Minimum.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Mod.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Mul.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/MulNoNan.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Ndtri.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Neg.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/NextAfter.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/NotEqual.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Polygamma.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/PopulationCount.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Pow.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/QuantizedAdd.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/QuantizedMul.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Real.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/RealDiv.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Reciprocal.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/ReciprocalGrad.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/RequantizationRangePerChannel.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/RequantizePerChannel.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Rint.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Round.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Rsqrt.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/RsqrtGrad.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SegmentMax.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SegmentMean.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SegmentMin.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SegmentProd.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SegmentSum.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Sigmoid.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SigmoidGrad.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Sign.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Sin.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Sinh.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SobolSample.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Softplus.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SoftplusGrad.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Sqrt.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SqrtGrad.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Square.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SquaredDifference.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Sub.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Tan.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Tanh.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/TanhGrad.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/TruncateDiv.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/TruncateMod.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/UnsortedSegmentMax.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/UnsortedSegmentMin.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/UnsortedSegmentProd.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/UnsortedSegmentSum.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Xdivy.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Xlog1py.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Xlogy.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Zeta.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/erfinv.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/BesselJ0.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/BesselJ1.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/BesselK0.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/BesselK0e.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/BesselK1.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/BesselK1e.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/BesselY0.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/BesselY1.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/Dawsn.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/Expint.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/FresnelCos.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/FresnelSin.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/Spence.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/AvgPool.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/AvgPool3d.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/AvgPool3dGrad.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/AvgPoolGrad.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/BatchNormWithGlobalNormalization.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/BatchNormWithGlobalNormalizationGrad.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/BiasAdd.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/BiasAddGrad.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/BlockLSTM.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/BlockLSTMGrad.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CTCLossV2.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/ComputeAccidentalHits.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Conv2d.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Conv2dBackpropFilter.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Conv2dBackpropInput.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Conv3d.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Conv3dBackpropFilter.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Conv3dBackpropInput.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CtcBeamSearchDecoder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CtcGreedyDecoder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CtcLoss.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CudnnRNN.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CudnnRNNBackprop.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CudnnRNNCanonicalToParams.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CudnnRNNParamsToCanonical.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CudnnRnnParamsSize.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/DataFormatDimMap.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/DataFormatVecPermute.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/DepthToSpace.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/DepthwiseConv2dNative.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/DepthwiseConv2dNativeBackpropFilter.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/DepthwiseConv2dNativeBackpropInput.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Dilation2d.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Dilation2dBackpropFilter.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Dilation2dBackpropInput.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Elu.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/EluGrad.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/FixedUnigramCandidateSampler.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/FractionalAvgPool.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/FractionalAvgPoolGrad.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/FractionalMaxPool.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/FractionalMaxPoolGrad.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/FusedBatchNorm.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/FusedBatchNormGrad.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/FusedPadConv2d.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/FusedResizeAndPadConv2d.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/GRUBlockCell.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/GRUBlockCellGrad.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/InTopK.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/InvGrad.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/L2Loss.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/LSTMBlockCell.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/LSTMBlockCellGrad.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/LearnedUnigramCandidateSampler.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/LocalResponseNormalization.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/LocalResponseNormalizationGrad.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/LogSoftmax.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPool.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPool3d.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPool3dGrad.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPool3dGradGrad.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPoolGrad.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPoolGradGrad.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPoolGradGradWithArgmax.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPoolGradWithArgmax.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPoolWithArgmax.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/NthElement.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedAvgPool.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedBatchNormWithGlobalNormalization.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedBiasAdd.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DAndRelu.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DAndReluAndRequantize.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DAndRequantize.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DPerChannel.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DWithBias.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DWithBiasAndRelu.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DWithBiasAndReluAndRequantize.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DWithBiasAndRequantize.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DWithBiasSignedSumAndReluAndRequantize.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DWithBiasSumAndRelu.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DWithBiasSumAndReluAndRequantize.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2d.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedDepthwiseConv2D.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedDepthwiseConv2DWithBias.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedDepthwiseConv2DWithBiasAndRelu.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedDepthwiseConv2DWithBiasAndReluAndRequantize.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedInstanceNorm.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedMaxPool.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedRelu.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedRelu6.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedReluX.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Relu.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Relu6.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Relu6Grad.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/ReluGrad.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Selu.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/SeluGrad.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Softmax.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Softsign.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/SoftsignGrad.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/SpaceToBatch.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/SpaceToDepth.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/TopK.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/raw/SoftmaxCrossEntropyWithLogits.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/raw/SparseSoftmaxCrossEntropyWithLogits.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/Dequantize.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/FakeQuantWithMinMaxArgs.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/FakeQuantWithMinMaxArgsGradient.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/FakeQuantWithMinMaxVars.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/FakeQuantWithMinMaxVarsGradient.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/FakeQuantWithMinMaxVarsPerChannel.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/FakeQuantWithMinMaxVarsPerChannelGradient.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/Quantize.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/QuantizeAndDequantize.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/QuantizeDownAndShrinkRange.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/QuantizedConcat.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/QuantizedMatMulWithBiasAndDequantize.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/QuantizedMatMulWithBiasAndRequantize.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/RequantizationRange.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/Requantize.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/ragged/RaggedBincount.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/ragged/RaggedCountSparseOutput.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/ragged/RaggedCross.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/ragged/RaggedGather.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/ragged/RaggedRange.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/ragged/RaggedTensorFromVariant.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/ragged/RaggedTensorToSparse.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/ragged/RaggedTensorToTensor.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/ragged/RaggedTensorToVariant.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/AllCandidateSampler.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/AnonymousRandomSeedGenerator.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/AnonymousSeedGenerator.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/DeleteRandomSeedGenerator.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/DeleteSeedGenerator.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/LogUniformCandidateSampler.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/Multinomial.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/NonDeterministicInts.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/ParameterizedTruncatedNormal.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RandomGamma.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RandomGammaGrad.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RandomPoisson.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RandomShuffle.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RandomStandardNormal.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RandomUniform.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RandomUniformInt.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RecordInput.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RngSkip.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatefulRandomBinomial.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatefulStandardNormal.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatefulTruncatedNormal.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatefulUniform.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatefulUniformFullInt.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatefulUniformInt.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessMultinomial.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessParameterizedTruncatedNormal.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessRandomBinomial.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessRandomGamma.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessRandomNormal.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessRandomPoisson.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessRandomUniform.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessRandomUniformFullInt.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessRandomUniformInt.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessTruncatedNormal.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/TruncatedNormal.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/UniformCandidateSampler.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/experimental/DummySeedGenerator.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/BatchFft.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/BatchFft2d.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/BatchFft3d.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/BatchIfft.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/BatchIfft2d.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/BatchIfft3d.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Fft.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Fft2d.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Fft3d.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Ifft.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Ifft2d.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Ifft3d.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Irfft.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Irfft2d.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Irfft3d.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Rfft.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Rfft2d.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Rfft3d.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/AddManySparseToTensorsMap.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/AddSparseToTensorsMap.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/DenseCountSparseOutput.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/DenseToDenseSetOperation.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/DenseToSparseSetOperation.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/DeserializeSparse.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseAccumulatorApplyGradient.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseAccumulatorTakeGradient.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseAdd.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseAddGrad.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseBincount.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseConcat.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseConditionalAccumulator.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseCountSparseOutput.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseCross.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseCrossHashed.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseDenseCwiseAdd.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseDenseCwiseDiv.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseDenseCwiseMul.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseFillEmptyRows.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseFillEmptyRowsGrad.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseMatMul.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseReduceMax.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseReduceMaxSparse.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseReduceSum.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseReduceSumSparse.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseReorder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseReshape.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentMean.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentMeanGrad.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentMeanWithNumSegments.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentSqrtN.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentSqrtNGrad.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentSqrtNWithNumSegments.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentSum.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentSumWithNumSegments.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSlice.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSliceGrad.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSoftmax.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSparseMaximum.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSparseMinimum.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSplit.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseTensorDenseAdd.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseTensorDenseMatMul.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseToDense.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseToSparseSetOperation.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/TakeManySparseFromTensorsMap.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/Join.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/Lower.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/ReduceJoin.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/RegexFullMatch.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/RegexReplace.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/StaticRegexFullMatch.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/StaticRegexReplace.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/StringFormat.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/StringLength.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/StringNGrams.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/StringSplit.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/Strip.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/Substr.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/ToHashBucket.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/ToHashBucketFast.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/ToHashBucketStrong.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/ToNumber.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/UnicodeDecode.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/UnicodeDecodeWithOffsets.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/UnicodeEncode.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/UnicodeScript.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/UnicodeTranscode.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/UnsortedSegmentJoin.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/Upper.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/AudioSummary.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/CloseSummaryWriter.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/CreateSummaryDbWriter.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/CreateSummaryFileWriter.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/FlushSummaryWriter.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/HistogramSummary.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/ImageSummary.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/ImportEvent.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/MergeSummary.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/ScalarSummary.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/StatsAggregatorSummary.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/SummaryWriter.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/TensorSummary.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/WriteAudioSummary.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/WriteGraphSummary.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/WriteHistogramSummary.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/WriteImageSummary.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/WriteRawProtoSummary.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/WriteScalarSummary.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/WriteSummary.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/AllToAll.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/CollectivePermute.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/ConfigureDistributedTPU.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/ConfigureTPUEmbedding.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/CrossReplicaSum.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/EnqueueTPUEmbeddingIntegerBatch.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/EnqueueTPUEmbeddingRaggedTensorBatch.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/EnqueueTPUEmbeddingSparseBatch.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/EnqueueTPUEmbeddingSparseTensorBatch.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/InfeedDequeue.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/InfeedDequeueTuple.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/InfeedEnqueue.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/InfeedEnqueuePrelinearizedBuffer.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/InfeedEnqueueTuple.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingADAMParameters.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingADAMParametersGradAccumDebug.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingAdadeltaParameters.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingAdadeltaParametersGradAccumDebug.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingAdagradParameters.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingAdagradParametersGradAccumDebug.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingCenteredRMSPropParameters.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingFTRLParameters.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingFTRLParametersGradAccumDebug.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingMDLAdagradLightParameters.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingMomentumParameters.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingMomentumParametersGradAccumDebug.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingProximalAdagradParameters.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingProximalAdagradParametersGradAccumDebug.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingProximalYogiParameters.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingProximalYogiParametersGradAccumDebug.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingRMSPropParameters.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingRMSPropParametersGradAccumDebug.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingStochasticGradientDescentParameters.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingStochasticGradientDescentParametersGradAccumDebug.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/OutfeedDequeue.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/OutfeedDequeueTuple.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/OutfeedEnqueue.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/OutfeedEnqueueTuple.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/Prelinearize.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/PrelinearizeTuple.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RecvTPUEmbeddingActivations.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingADAMParameters.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingADAMParametersGradAccumDebug.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingAdadeltaParameters.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingAdadeltaParametersGradAccumDebug.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingAdagradParameters.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingAdagradParametersGradAccumDebug.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingCenteredRMSPropParameters.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingFTRLParameters.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingFTRLParametersGradAccumDebug.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingMDLAdagradLightParameters.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingMomentumParameters.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingMomentumParametersGradAccumDebug.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingProximalAdagradParameters.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingProximalAdagradParametersGradAccumDebug.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingProximalYogiParameters.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingProximalYogiParametersGradAccumDebug.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingRMSPropParameters.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingRMSPropParametersGradAccumDebug.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingStochasticGradientDescentParameters.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingStochasticGradientDescentParametersGradAccumDebug.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/SendTPUEmbeddingGradients.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/ShutdownDistributedTPU.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/TPUCompilationResult.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/TPUEmbeddingActivations.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/TPUOrdinalSelector.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/TPUReplicateMetadata.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/TPUReplicatedInput.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/TPUReplicatedOutput.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/WorkerHeartbeat.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/AccumulatorApplyGradient.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/AccumulatorNumAccumulated.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/AccumulatorSetGlobalStep.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/AccumulatorTakeGradient.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyAdaMax.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyAdadelta.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyAdagrad.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyAdagradDa.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyAdagradV2.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyAdam.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyAddSign.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyCenteredRmsProp.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyFtrl.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyGradientDescent.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyMomentum.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyPowerSign.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyProximalAdagrad.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyProximalGradientDescent.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyRmsProp.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/BatchMatMul.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ConditionalAccumulator.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/GenerateVocabRemapping.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/MergeV2Checkpoints.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/NegTrain.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/PreventGradient.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceAccumulatorApplyGradient.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceAccumulatorNumAccumulated.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceAccumulatorSetGlobalStep.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceAccumulatorTakeGradient.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyAdaMax.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyAdadelta.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyAdagrad.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyAdagradDa.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyAdam.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyAdamWithAmsgrad.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyAddSign.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyCenteredRmsProp.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyFtrl.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyGradientDescent.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyKerasMomentum.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyMomentum.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyPowerSign.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyProximalAdagrad.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyProximalGradientDescent.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyRmsProp.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceConditionalAccumulator.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyAdadelta.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyAdagrad.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyAdagradDa.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyAdagradV2.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyCenteredRmsProp.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyFtrl.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyKerasMomentum.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyMomentum.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyProximalAdagrad.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyProximalGradientDescent.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyRmsProp.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/Restore.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/RestoreSlice.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/Save.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SaveSlices.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SdcaFprint.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SdcaOptimizer.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SdcaShrinkL1.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyAdadelta.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyAdagrad.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyAdagradDa.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyCenteredRmsProp.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyFtrl.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyMomentum.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyProximalAdagrad.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyProximalGradientDescent.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyRmsProp.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/TileGrad.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/BroadcastHelper.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/ClusterOutput.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Conv.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Dequantize.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Dot.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/DynamicSlice.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/DynamicUpdateSlice.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Einsum.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Gather.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/KeyValueSort.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Pad.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Recv.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/ReplicaId.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/SelfAdjointEig.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Send.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Sharding.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Sort.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Svd.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/distruntime/ClusterDef.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/distruntime/ClusterDefOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/distruntime/ClusterDeviceFilters.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/distruntime/ClusterDeviceFiltersOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/distruntime/ClusterProtos.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/distruntime/DeviceFiltersProtos.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/distruntime/JobDef.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/distruntime/JobDefOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/distruntime/JobDeviceFilters.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/distruntime/JobDeviceFiltersOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/distruntime/ServerDef.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/distruntime/ServerDefOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/distruntime/ServerProtos.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/distruntime/TaskDeviceFilters.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/distruntime/TaskDeviceFiltersOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/distruntime/TransportOptions.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/BytesList.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/BytesListOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/Example.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/ExampleOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/ExampleParserConfiguration.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/ExampleParserConfigurationOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/ExampleParserConfigurationProtos.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/ExampleProtos.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/Feature.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/FeatureConfiguration.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/FeatureConfigurationOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/FeatureList.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/FeatureListOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/FeatureLists.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/FeatureListsOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/FeatureOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/FeatureProtos.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/Features.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/FeaturesOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/FixedLenFeatureProto.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/FixedLenFeatureProtoOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/FloatList.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/FloatListOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/Int64List.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/Int64ListOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/SequenceExample.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/SequenceExampleOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/VarLenFeatureProto.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/VarLenFeatureProtoOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/AllocationDescription.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/AllocationDescriptionOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/AllocationDescriptionProtos.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/AllocationRecord.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/AllocationRecordOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/AllocatorMemoryUsed.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/AllocatorMemoryUsedOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ApiDef.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ApiDefOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ApiDefProtos.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ApiDefs.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ApiDefsOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/AssetFileDef.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/AssetFileDefOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/AttrValue.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/AttrValueOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/AttrValueProtos.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/AutoParallelOptions.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/AutoParallelOptionsOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/BoundedTensorSpecProto.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/BoundedTensorSpecProtoOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/CallableOptions.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/CallableOptionsOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/Code.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/CollectionDef.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/CollectionDefOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/CondContextDef.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/CondContextDefOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ConfigProto.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ConfigProtoOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ConfigProtos.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ControlFlowContextDef.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ControlFlowContextDefOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ControlFlowProtos.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/CostGraphDef.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/CostGraphDefOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/CostGraphProtos.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DataClass.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DataType.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DebugOptions.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DebugOptionsOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DebugProtos.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DebugTensorWatch.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DebugTensorWatchOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DebuggedSourceFile.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DebuggedSourceFileOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DebuggedSourceFiles.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DebuggedSourceFilesOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DeviceAttributes.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DeviceAttributesOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DeviceAttributesProtos.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DeviceLocality.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DeviceLocalityOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DeviceProperties.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DevicePropertiesOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DevicePropertiesProtos.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DeviceStepStats.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DeviceStepStatsOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DictValue.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DictValueOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ErrorCodes.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ErrorCodesProtos.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/FunctionDef.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/FunctionDefLibrary.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/FunctionDefLibraryOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/FunctionDefOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/FunctionProtos.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/FunctionSpec.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/FunctionSpecOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GPUOptions.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GPUOptionsOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GradientDef.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GradientDefOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphDebugInfo.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphDebugInfoOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphDebugInfoProtos.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphDef.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphDefOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphOptions.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphOptionsOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphProtos.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphTransferConstNodeInfo.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphTransferConstNodeInfoOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphTransferGraphInputNodeInfo.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphTransferGraphInputNodeInfoOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphTransferGraphOutputNodeInfo.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphTransferGraphOutputNodeInfoOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphTransferInfo.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphTransferInfoOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphTransferInfoProto.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphTransferNodeInfo.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphTransferNodeInfoOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphTransferNodeInput.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphTransferNodeInputInfo.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphTransferNodeInputInfoOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphTransferNodeInputOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphTransferNodeOutputInfo.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphTransferNodeOutputInfoOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/HistogramProto.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/HistogramProtoOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/InterconnectLink.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/InterconnectLinkOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/KernelDef.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/KernelDefOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/KernelDefProtos.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/KernelList.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/KernelListOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ListValue.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ListValueOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/LocalLinks.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/LocalLinksOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/LogMemoryProtos.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/MemoryLogRawAllocation.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/MemoryLogRawAllocationOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/MemoryLogRawDeallocation.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/MemoryLogRawDeallocationOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/MemoryLogStep.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/MemoryLogStepOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/MemoryLogTensorAllocation.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/MemoryLogTensorAllocationOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/MemoryLogTensorDeallocation.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/MemoryLogTensorDeallocationOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/MemoryLogTensorOutput.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/MemoryLogTensorOutputOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/MemoryStats.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/MemoryStatsOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/MetaGraphDef.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/MetaGraphDefOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/MetaGraphProtos.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NameAttrList.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NameAttrListOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NamedDevice.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NamedDeviceOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NamedTensorProto.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NamedTensorProtoOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NamedTensorProtos.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NamedTupleValue.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NamedTupleValueOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NodeDef.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NodeDefOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NodeExecStats.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NodeExecStatsOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NodeOutput.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NodeOutputOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NodeProto.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NoneValue.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NoneValueOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/OpDef.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/OpDefOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/OpDefProtos.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/OpDeprecation.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/OpDeprecationOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/OpList.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/OpListOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/OptimizerOptions.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/OptimizerOptionsOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/PairValue.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/PairValueOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/QueueRunnerDef.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/QueueRunnerDefOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/QueueRunnerProtos.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/RPCOptions.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/RPCOptionsOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ReaderBaseProtos.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ReaderBaseState.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ReaderBaseStateOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/RemoteFusedGraphExecuteInfo.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/RemoteFusedGraphExecuteInfoOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/RemoteFusedGraphExecuteInfoProto.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/RemoteTensorHandle.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/RemoteTensorHandleOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/RemoteTensorHandleProtos.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ResourceDtypeAndShape.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ResourceDtypeAndShapeOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ResourceHandle.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ResourceHandleProto.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ResourceHandleProtoOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/RewriterConfig.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/RewriterConfigOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/RewriterConfigProtos.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/RunMetadata.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/RunMetadataOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/RunOptions.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/RunOptionsOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SaveSliceInfoDef.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SaveSliceInfoDefOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SaveableObject.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SaveableObjectOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedAsset.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedAssetOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedBareConcreteFunction.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedBareConcreteFunctionOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedConcreteFunction.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedConcreteFunctionOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedConstant.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedConstantOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedFunction.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedFunctionOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedModel.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedModelOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedModelProtos.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedObject.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedObjectGraph.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedObjectGraphOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedObjectGraphProtos.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedObjectOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedResource.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedResourceOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedUserObject.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedUserObjectOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedVariable.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedVariableOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ScopedAllocatorOptions.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ScopedAllocatorOptionsOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SessionMetadata.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SessionMetadataOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SignatureDef.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SignatureDefOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/StepStats.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/StepStatsOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/StepStatsProtos.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/StructProtos.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/StructuredValue.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/StructuredValueOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/Summary.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SummaryDescription.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SummaryDescriptionOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SummaryMetadata.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SummaryMetadataOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SummaryOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SummaryProtos.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorConnection.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorConnectionOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorDescription.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorDescriptionOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorDescriptionProtos.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorInfo.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorInfoOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorProto.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorProtoOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorProtos.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorShapeProto.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorShapeProtoOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorShapeProtos.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorSliceProto.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorSliceProtoOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorSliceProtos.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorSpecProto.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorSpecProtoOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ThreadPoolOptionProto.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ThreadPoolOptionProtoOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TrackableObjectGraph.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TrackableObjectGraphOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TrackableObjectGraphProtos.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TupleValue.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TupleValueOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TypeSpecProto.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TypeSpecProtoOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TypesProtos.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ValuesDef.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ValuesDefOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/VariableAggregation.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/VariableDef.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/VariableDefOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/VariableProtos.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/VariableSynchronization.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/VariantTensorDataProto.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/VariantTensorDataProtoOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/VerifierConfig.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/VerifierConfigOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/VerifierConfigProtos.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/VersionDef.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/VersionDefOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/VersionsProtos.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/WhileContextDef.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/WhileContextDefOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/ProfileOptions.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/ProfileOptionsOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/ProfilerOptionsProtos.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/XEvent.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/XEventMetadata.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/XEventMetadataOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/XEventOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/XLine.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/XLineOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/XPlane.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/XPlaneOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/XPlaneProtos.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/XSpace.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/XSpaceOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/XStat.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/XStatMetadata.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/XStatMetadataOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/XStatOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/BfcMemoryMapProtos.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/BinSummary.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/BinSummaryOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/BundleEntryProto.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/BundleEntryProtoOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/BundleHeaderProto.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/BundleHeaderProtoOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/CodeLocation.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/CodeLocationOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/DebugEvent.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/DebugEventOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/DebugEventProtos.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/DebugMetadata.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/DebugMetadataOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/DebuggedDevice.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/DebuggedDeviceOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/DebuggedGraph.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/DebuggedGraphOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/Event.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/EventOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/EventProtos.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/Execution.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/ExecutionOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/GraphExecutionTrace.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/GraphExecutionTraceOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/GraphOpCreation.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/GraphOpCreationOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/LogMessage.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/LogMessageOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/MemAllocatorStats.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/MemAllocatorStatsOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/MemChunk.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/MemChunkOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/MemmappedFileSystemDirectory.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/MemmappedFileSystemDirectoryElement.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/MemmappedFileSystemDirectoryElementOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/MemmappedFileSystemDirectoryOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/MemmappedFileSystemProtos.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/MemoryDump.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/MemoryDumpOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/RequestedExitCode.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/RequestedExitCodeOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SavedSlice.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SavedSliceMeta.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SavedSliceMetaOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SavedSliceOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SavedTensorSliceMeta.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SavedTensorSliceMetaOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SavedTensorSliceProtos.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SavedTensorSlices.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SavedTensorSlicesOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SaverDef.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SaverDefOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SaverProtos.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SessionLog.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SessionLogOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SnapShot.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SnapShotOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SnapshotMetadataRecord.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SnapshotMetadataRecordOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SnapshotProtos.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SnapshotRecord.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SnapshotRecordOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SnapshotTensorMetadata.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SnapshotTensorMetadataOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SourceFile.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SourceFileOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/StackFrameWithId.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/StackFrameWithIdOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/TaggedRunMetadata.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/TaggedRunMetadataOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/TensorBundleProtos.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/TensorDebugMode.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/TensorMetadata.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/TensorMetadataOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/WatchdogConfig.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/WatchdogConfigOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/WorkerHealth.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/WorkerHeartbeatRequest.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/WorkerHeartbeatRequestOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/WorkerHeartbeatResponse.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/WorkerHeartbeatResponseOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/WorkerShutdownMode.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/AvailableDeviceInfo.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/AvailableDeviceInfoOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/BenchmarkEntries.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/BenchmarkEntriesOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/BenchmarkEntry.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/BenchmarkEntryOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/BuildConfiguration.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/BuildConfigurationOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/CPUInfo.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/CPUInfoOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/CommitId.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/CommitIdOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/EntryValue.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/EntryValueOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/GPUInfo.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/GPUInfoOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/MachineConfiguration.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/MachineConfigurationOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/MemoryInfo.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/MemoryInfoOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/MetricEntry.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/MetricEntryOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/PlatformInfo.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/PlatformInfoOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/RunConfiguration.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/RunConfigurationOrBuilder.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/TestLogProtos.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/TestResults.java
 create mode 100644 tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/TestResultsOrBuilder.java

diff --git a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/AudioOps.java b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/AudioOps.java
new file mode 100644
index 00000000000..16770394378
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/AudioOps.java
@@ -0,0 +1,147 @@
+// Copyright 2020 The TensorFlow Authors. All Rights Reserved.
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+// ==============================================================================
+//
+// This class has been generated, DO NOT EDIT!
+//
+package org.tensorflow.op;
+
+import org.tensorflow.Operand;
+import org.tensorflow.op.audio.AudioSpectrogram;
+import org.tensorflow.op.audio.DecodeWav;
+import org.tensorflow.op.audio.EncodeWav;
+import org.tensorflow.op.audio.Mfcc;
+import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.TString;
+
+/**
+ * An API for building {@code audio} operations as {@link Op Op}s
+ *
+ * @see {@link Ops}
+ */
+public final class AudioOps {
+  private final Scope scope;
+
+  AudioOps(Scope scope) {
+    this.scope = scope;
+  }
+
+  /**
+   * Produces a visualization of audio data over time.
+   *  <p>
+   *  Spectrograms are a standard way of representing audio information as a series of
+   *  slices of frequency information, one slice for each window of time. By joining
+   *  these together into a sequence, they form a distinctive fingerprint of the sound
+   *  over time.
+   *  <p>
+   *  This op expects to receive audio data as an input, stored as floats in the range
+   *  -1 to 1, together with a window width in samples, and a stride specifying how
+   *  far to move the window between slices. From this it generates a three
+   *  dimensional output. The first dimension is for the channels in the input, so a
+   *  stereo audio input would have two here for example. The second dimension is time,
+   *  with successive frequency slices. The third dimension has an amplitude value for
+   *  each frequency during that time slice.
+   *  <p>
+   *  This means the layout when converted and saved as an image is rotated 90 degrees
+   *  clockwise from a typical spectrogram. Time is descending down the Y axis, and
+   *  the frequency decreases from left to right.
+   *  <p>
+   *  Each value in the result represents the square root of the sum of the real and
+   *  imaginary parts of an FFT on the current window of samples. In this way, the
+   *  lowest dimension represents the power of each frequency in the current window,
+   *  and adjacent windows are concatenated in the next dimension.
+   *  <p>
+   *  To get a more intuitive and visual look at what this operation does, you can run
+   *  tensorflow/examples/wav_to_spectrogram to read in an audio file and save out the
+   *  resulting spectrogram as a PNG image.
+   *
+   * @param input Float representation of audio data.
+   * @param windowSize How wide the input window is in samples. For the highest efficiency
+   *  this should be a power of two, but other values are accepted.
+   * @param stride How widely apart the center of adjacent sample windows should be.
+   * @param options carries optional attributes values
+   * @return a new instance of AudioSpectrogram
+   */
+  public AudioSpectrogram audioSpectrogram(Operand<TFloat32> input, Long windowSize, Long stride,
+      AudioSpectrogram.Options... options) {
+    return AudioSpectrogram.create(scope, input, windowSize, stride, options);
+  }
+
+  /**
+   * Decode a 16-bit PCM WAV file to a float tensor.
+   *  <p>
+   *  The -32768 to 32767 signed 16-bit values will be scaled to -1.0 to 1.0 in float.
+   *  <p>
+   *  When desired_channels is set, if the input contains fewer channels than this
+   *  then the last channel will be duplicated to give the requested number, else if
+   *  the input has more channels than requested then the additional channels will be
+   *  ignored.
+   *  <p>
+   *  If desired_samples is set, then the audio will be cropped or padded with zeroes
+   *  to the requested length.
+   *  <p>
+   *  The first output contains a Tensor with the content of the audio samples. The
+   *  lowest dimension will be the number of channels, and the second will be the
+   *  number of samples. For example, a ten-sample-long stereo WAV file should give an
+   *  output shape of [10, 2].
+   *
+   * @param contents The WAV-encoded audio, usually from a file.
+   * @param options carries optional attributes values
+   * @return a new instance of DecodeWav
+   */
+  public DecodeWav decodeWav(Operand<TString> contents, DecodeWav.Options... options) {
+    return DecodeWav.create(scope, contents, options);
+  }
+
+  /**
+   * Encode audio data using the WAV file format.
+   *  <p>
+   *  This operation will generate a string suitable to be saved out to create a .wav
+   *  audio file. It will be encoded in the 16-bit PCM format. It takes in float
+   *  values in the range -1.0f to 1.0f, and any outside that value will be clamped to
+   *  that range.
+   *  <p>
+   *  `audio` is a 2-D float Tensor of shape `[length, channels]`.
+   *  `sample_rate` is a scalar Tensor holding the rate to use (e.g. 44100).
+   *
+   * @param audio 2-D with shape `[length, channels]`.
+   * @param sampleRate Scalar containing the sample frequency.
+   * @return a new instance of EncodeWav
+   */
+  public EncodeWav encodeWav(Operand<TFloat32> audio, Operand<TInt32> sampleRate) {
+    return EncodeWav.create(scope, audio, sampleRate);
+  }
+
+  /**
+   * Transforms a spectrogram into a form that's useful for speech recognition.
+   *  <p>
+   *  Mel Frequency Cepstral Coefficients are a way of representing audio data that's
+   *  been effective as an input feature for machine learning. They are created by
+   *  taking the spectrum of a spectrogram (a 'cepstrum'), and discarding some of the
+   *  higher frequencies that are less significant to the human ear. They have a long
+   *  history in the speech recognition world, and https://en.wikipedia.org/wiki/Mel-frequency_cepstrum
+   *  is a good resource to learn more.
+   *
+   * @param spectrogram Typically produced by the Spectrogram op, with magnitude_squared
+   *  set to true.
+   * @param sampleRate How many samples per second the source audio used.
+   * @param options carries optional attributes values
+   * @return a new instance of Mfcc
+   */
+  public Mfcc mfcc(Operand<TFloat32> spectrogram, Operand<TInt32> sampleRate,
+      Mfcc.Options... options) {
+    return Mfcc.create(scope, spectrogram, sampleRate, options);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/BitwiseOps.java b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/BitwiseOps.java
new file mode 100644
index 00000000000..8ac6d565e51
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/BitwiseOps.java
@@ -0,0 +1,271 @@
+// Copyright 2020 The TensorFlow Authors. All Rights Reserved.
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+// ==============================================================================
+//
+// This class has been generated, DO NOT EDIT!
+//
+package org.tensorflow.op;
+
+import org.tensorflow.Operand;
+import org.tensorflow.op.bitwise.BitwiseAnd;
+import org.tensorflow.op.bitwise.BitwiseOr;
+import org.tensorflow.op.bitwise.BitwiseXor;
+import org.tensorflow.op.bitwise.Invert;
+import org.tensorflow.op.bitwise.LeftShift;
+import org.tensorflow.op.bitwise.RightShift;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * An API for building {@code bitwise} operations as {@link Op Op}s
+ *
+ * @see {@link Ops}
+ */
+public final class BitwiseOps {
+  private final Scope scope;
+
+  BitwiseOps(Scope scope) {
+    this.scope = scope;
+  }
+
+  /**
+   * Elementwise computes the bitwise AND of `x` and `y`.
+   *  <p>
+   *  The result will have those bits set, that are set in both `x` and `y`. The
+   *  computation is performed on the underlying representations of `x` and `y`.
+   *  <p>
+   *  For example:
+   *  <pre>{@code
+   *  import tensorflow as tf
+   *  from tensorflow.python.ops import bitwise_ops
+   *  dtype_list = [tf.int8, tf.int16, tf.int32, tf.int64,
+   *                tf.uint8, tf.uint16, tf.uint32, tf.uint64]
+   *
+   *  for dtype in dtype_list:
+   *    lhs = tf.constant([0, 5, 3, 14], dtype=dtype)
+   *    rhs = tf.constant([5, 0, 7, 11], dtype=dtype)
+   *    exp = tf.constant([0, 0, 3, 10], dtype=tf.float32)
+   *
+   *    res = bitwise_ops.bitwise_and(lhs, rhs)
+   *    tf.assert_equal(tf.cast(res, tf.float32), exp) # TRUE
+   *  }</pre>
+   *
+   * @param <T> data type for {@code z()} output
+   * @param x
+   * @param y
+   * @return a new instance of BitwiseAnd
+   */
+  public <T extends TNumber> BitwiseAnd<T> bitwiseAnd(Operand<T> x, Operand<T> y) {
+    return BitwiseAnd.create(scope, x, y);
+  }
+
+  /**
+   * Elementwise computes the bitwise OR of `x` and `y`.
+   *  <p>
+   *  The result will have those bits set, that are set in `x`, `y` or both. The
+   *  computation is performed on the underlying representations of `x` and `y`.
+   *  <p>
+   *  For example:
+   *  <pre>{@code
+   *  import tensorflow as tf
+   *  from tensorflow.python.ops import bitwise_ops
+   *  dtype_list = [tf.int8, tf.int16, tf.int32, tf.int64,
+   *                tf.uint8, tf.uint16, tf.uint32, tf.uint64]
+   *
+   *  for dtype in dtype_list:
+   *    lhs = tf.constant([0, 5, 3, 14], dtype=dtype)
+   *    rhs = tf.constant([5, 0, 7, 11], dtype=dtype)
+   *    exp = tf.constant([5, 5, 7, 15], dtype=tf.float32)
+   *
+   *    res = bitwise_ops.bitwise_or(lhs, rhs)
+   *    tf.assert_equal(tf.cast(res,  tf.float32), exp)  # TRUE
+   *  }</pre>
+   *
+   * @param <T> data type for {@code z()} output
+   * @param x
+   * @param y
+   * @return a new instance of BitwiseOr
+   */
+  public <T extends TNumber> BitwiseOr<T> bitwiseOr(Operand<T> x, Operand<T> y) {
+    return BitwiseOr.create(scope, x, y);
+  }
+
+  /**
+   * Elementwise computes the bitwise XOR of `x` and `y`.
+   *  <p>
+   *  The result will have those bits set, that are different in `x` and `y`. The
+   *  computation is performed on the underlying representations of `x` and `y`.
+   *  <p>
+   *  For example:
+   *  <pre>{@code
+   *  import tensorflow as tf
+   *  from tensorflow.python.ops import bitwise_ops
+   *  dtype_list = [tf.int8, tf.int16, tf.int32, tf.int64,
+   *                tf.uint8, tf.uint16, tf.uint32, tf.uint64]
+   *
+   *  for dtype in dtype_list:
+   *    lhs = tf.constant([0, 5, 3, 14], dtype=dtype)
+   *    rhs = tf.constant([5, 0, 7, 11], dtype=dtype)
+   *    exp = tf.constant([5, 5, 4, 5],  dtype=tf.float32)
+   *
+   *    res = bitwise_ops.bitwise_xor(lhs, rhs)
+   *    tf.assert_equal(tf.cast(res, tf.float32), exp) # TRUE
+   *  }</pre>
+   *
+   * @param <T> data type for {@code z()} output
+   * @param x
+   * @param y
+   * @return a new instance of BitwiseXor
+   */
+  public <T extends TNumber> BitwiseXor<T> bitwiseXor(Operand<T> x, Operand<T> y) {
+    return BitwiseXor.create(scope, x, y);
+  }
+
+  /**
+   * Invert (flip) each bit of supported types; for example, type `uint8` value 01010101 becomes 10101010.
+   *  <p>
+   *  Flip each bit of supported types.  For example, type `int8` (decimal 2) binary 00000010 becomes (decimal -3) binary 11111101.
+   *  This operation is performed on each element of the tensor argument `x`.
+   *  <p>
+   *  Example:
+   *  <pre>{@code
+   *  import tensorflow as tf
+   *  from tensorflow.python.ops import bitwise_ops
+   *
+   *  # flip 2 (00000010) to -3 (11111101)
+   *  tf.assert_equal(-3, bitwise_ops.invert(2))
+   *
+   *  dtype_list = [dtypes.int8, dtypes.int16, dtypes.int32, dtypes.int64,
+   *                dtypes.uint8, dtypes.uint16, dtypes.uint32, dtypes.uint64]
+   *
+   *  inputs = [0, 5, 3, 14]
+   *  for dtype in dtype_list:
+   *    # Because of issues with negative numbers, let's test this indirectly.
+   *    # 1. invert(a) and a = 0
+   *    # 2. invert(a) or a = invert(0)
+   *    input_tensor = tf.constant([0, 5, 3, 14], dtype=dtype)
+   *    not_a_and_a, not_a_or_a, not_0 = [bitwise_ops.bitwise_and(
+   *                                        input_tensor, bitwise_ops.invert(input_tensor)),
+   *                                      bitwise_ops.bitwise_or(
+   *                                        input_tensor, bitwise_ops.invert(input_tensor)),
+   *                                      bitwise_ops.invert(
+   *                                        tf.constant(0, dtype=dtype))]
+   *
+   *    expected = tf.constant([0, 0, 0, 0], dtype=tf.float32)
+   *    tf.assert_equal(tf.cast(not_a_and_a, tf.float32), expected)
+   *
+   *    expected = tf.cast([not_0] * 4, tf.float32)
+   *    tf.assert_equal(tf.cast(not_a_or_a, tf.float32), expected)
+   *
+   *    # For unsigned dtypes let's also check the result directly.
+   *    if dtype.is_unsigned:
+   *      inverted = bitwise_ops.invert(input_tensor)
+   *      expected = tf.constant([dtype.max - x for x in inputs], dtype=tf.float32)
+   *      tf.assert_equal(tf.cast(inverted, tf.float32), tf.cast(expected, tf.float32))
+   *  }</pre>
+   *
+   * @param <T> data type for {@code y()} output
+   * @param x
+   * @return a new instance of Invert
+   */
+  public <T extends TNumber> Invert<T> invert(Operand<T> x) {
+    return Invert.create(scope, x);
+  }
+
+  /**
+   * Elementwise computes the bitwise left-shift of `x` and `y`.
+   *  <p>
+   *  If `y` is negative, or greater than or equal to the width of `x` in bits the
+   *  result is implementation defined.
+   *  <p>
+   *  Example:
+   *  <pre>{@code
+   *  import tensorflow as tf
+   *  from tensorflow.python.ops import bitwise_ops
+   *  import numpy as np
+   *  dtype_list = [tf.int8, tf.int16, tf.int32, tf.int64]
+   *
+   *  for dtype in dtype_list:
+   *    lhs = tf.constant([-1, -5, -3, -14], dtype=dtype)
+   *    rhs = tf.constant([5, 0, 7, 11], dtype=dtype)
+   *
+   *    left_shift_result = bitwise_ops.left_shift(lhs, rhs)
+   *
+   *    print(left_shift_result)
+   *
+   *  # This will print:
+   *  # tf.Tensor([ -32   -5 -128    0], shape=(4,), dtype=int8)
+   *  # tf.Tensor([   -32     -5   -384 -28672], shape=(4,), dtype=int16)
+   *  # tf.Tensor([   -32     -5   -384 -28672], shape=(4,), dtype=int32)
+   *  # tf.Tensor([   -32     -5   -384 -28672], shape=(4,), dtype=int64)
+   *
+   *  lhs = np.array([-2, 64, 101, 32], dtype=np.int8)
+   *  rhs = np.array([-1, -5, -3, -14], dtype=np.int8)
+   *  bitwise_ops.left_shift(lhs, rhs)
+   *  # <tf.Tensor: shape=(4,), dtype=int8, numpy=array([ -2,  64, 101,  32], dtype=int8)>
+   *  }</pre>
+   *
+   * @param <T> data type for {@code z()} output
+   * @param x
+   * @param y
+   * @return a new instance of LeftShift
+   */
+  public <T extends TNumber> LeftShift<T> leftShift(Operand<T> x, Operand<T> y) {
+    return LeftShift.create(scope, x, y);
+  }
+
+  /**
+   * Elementwise computes the bitwise right-shift of `x` and `y`.
+   *  <p>
+   *  Performs a logical shift for unsigned integer types, and an arithmetic shift
+   *  for signed integer types.
+   *  <p>
+   *  If `y` is negative, or greater than or equal to than the width of `x` in bits
+   *  the result is implementation defined.
+   *  <p>
+   *  Example:
+   *  <pre>{@code
+   *  import tensorflow as tf
+   *  from tensorflow.python.ops import bitwise_ops
+   *  import numpy as np
+   *  dtype_list = [tf.int8, tf.int16, tf.int32, tf.int64]
+   *
+   *  for dtype in dtype_list:
+   *    lhs = tf.constant([-1, -5, -3, -14], dtype=dtype)
+   *    rhs = tf.constant([5, 0, 7, 11], dtype=dtype)
+   *
+   *    right_shift_result = bitwise_ops.right_shift(lhs, rhs)
+   *
+   *    print(right_shift_result)
+   *
+   *  # This will print:
+   *  # tf.Tensor([-1 -5 -1 -1], shape=(4,), dtype=int8)
+   *  # tf.Tensor([-1 -5 -1 -1], shape=(4,), dtype=int16)
+   *  # tf.Tensor([-1 -5 -1 -1], shape=(4,), dtype=int32)
+   *  # tf.Tensor([-1 -5 -1 -1], shape=(4,), dtype=int64)
+   *
+   *  lhs = np.array([-2, 64, 101, 32], dtype=np.int8)
+   *  rhs = np.array([-1, -5, -3, -14], dtype=np.int8)
+   *  bitwise_ops.right_shift(lhs, rhs)
+   *  # <tf.Tensor: shape=(4,), dtype=int8, numpy=array([ -2,  64, 101,  32], dtype=int8)>
+   *  }</pre>
+   *
+   * @param <T> data type for {@code z()} output
+   * @param x
+   * @param y
+   * @return a new instance of RightShift
+   */
+  public <T extends TNumber> RightShift<T> rightShift(Operand<T> x, Operand<T> y) {
+    return RightShift.create(scope, x, y);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/DataExperimentalOps.java b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/DataExperimentalOps.java
new file mode 100644
index 00000000000..411d8e7969a
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/DataExperimentalOps.java
@@ -0,0 +1,61 @@
+// Copyright 2020 The TensorFlow Authors. All Rights Reserved.
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+// ==============================================================================
+//
+// This class has been generated, DO NOT EDIT!
+//
+package org.tensorflow.op;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.data.experimental.DataServiceDataset;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.TString;
+import org.tensorflow.types.family.TType;
+
+/**
+ * An API for building {@code data.experimental} operations as {@link Op Op}s
+ *
+ * @see {@link Ops}
+ */
+public final class DataExperimentalOps {
+  private final Scope scope;
+
+  DataExperimentalOps(Scope scope) {
+    this.scope = scope;
+  }
+
+  /**
+   *
+   * @param datasetId
+   * @param processingMode
+   * @param address
+   * @param protocol
+   * @param jobName
+   * @param maxOutstandingRequests
+   * @param iterationCounter
+   * @param outputTypes
+   * @param outputShapes
+   * @param options carries optional attributes values
+   * @return a new instance of DataServiceDataset
+   */
+  public DataServiceDataset dataServiceDataset(Operand<TInt64> datasetId,
+      Operand<TString> processingMode, Operand<TString> address, Operand<TString> protocol,
+      Operand<TString> jobName, Operand<TInt64> maxOutstandingRequests, Operand<?> iterationCounter,
+      List<Class<? extends TType>> outputTypes, List<Shape> outputShapes,
+      DataServiceDataset.Options... options) {
+    return DataServiceDataset.create(scope, datasetId, processingMode, address, protocol, jobName, maxOutstandingRequests, iterationCounter, outputTypes, outputShapes, options);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/DataOps.java b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/DataOps.java
new file mode 100644
index 00000000000..d01c9ee6d01
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/DataOps.java
@@ -0,0 +1,413 @@
+// Copyright 2020 The TensorFlow Authors. All Rights Reserved.
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+// ==============================================================================
+//
+// This class has been generated, DO NOT EDIT!
+//
+package org.tensorflow.op;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.data.AnonymousIterator;
+import org.tensorflow.op.data.BatchDataset;
+import org.tensorflow.op.data.CSVDataset;
+import org.tensorflow.op.data.ConcatenateDataset;
+import org.tensorflow.op.data.DeleteIterator;
+import org.tensorflow.op.data.DeserializeIterator;
+import org.tensorflow.op.data.Iterator;
+import org.tensorflow.op.data.IteratorGetNext;
+import org.tensorflow.op.data.IteratorGetNextAsOptional;
+import org.tensorflow.op.data.IteratorGetNextSync;
+import org.tensorflow.op.data.IteratorToStringHandle;
+import org.tensorflow.op.data.MakeIterator;
+import org.tensorflow.op.data.OptionalFromValue;
+import org.tensorflow.op.data.OptionalGetValue;
+import org.tensorflow.op.data.OptionalHasValue;
+import org.tensorflow.op.data.OptionalNone;
+import org.tensorflow.op.data.RangeDataset;
+import org.tensorflow.op.data.RepeatDataset;
+import org.tensorflow.op.data.SerializeIterator;
+import org.tensorflow.op.data.SkipDataset;
+import org.tensorflow.op.data.TakeDataset;
+import org.tensorflow.op.data.TensorSliceDataset;
+import org.tensorflow.op.data.TextLineDataset;
+import org.tensorflow.op.data.TfRecordDataset;
+import org.tensorflow.op.data.ZipDataset;
+import org.tensorflow.types.TBool;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.TString;
+import org.tensorflow.types.family.TType;
+
+/**
+ * An API for building {@code data} operations as {@link Op Op}s
+ *
+ * @see {@link Ops}
+ */
+public final class DataOps {
+  public final DataExperimentalOps experimental;
+
+  private final Scope scope;
+
+  DataOps(Scope scope) {
+    this.scope = scope;
+    experimental = new DataExperimentalOps(scope);
+  }
+
+  /**
+   * A container for an iterator resource.
+   *
+   * @param outputTypes
+   * @param outputShapes
+   * @return a new instance of AnonymousIterator
+   */
+  public AnonymousIterator anonymousIterator(List<Class<? extends TType>> outputTypes,
+      List<Shape> outputShapes) {
+    return AnonymousIterator.create(scope, outputTypes, outputShapes);
+  }
+
+  /**
+   * Creates a dataset that batches `batch_size` elements from `input_dataset`.
+   *
+   * @param inputDataset
+   * @param batchSize A scalar representing the number of elements to accumulate in a batch.
+   * @param dropRemainder A scalar representing whether the last batch should be dropped in case its size
+   *  is smaller than desired.
+   * @param outputTypes
+   * @param outputShapes
+   * @param options carries optional attributes values
+   * @return a new instance of BatchDataset
+   */
+  public BatchDataset batchDataset(Operand<?> inputDataset, Operand<TInt64> batchSize,
+      Operand<TBool> dropRemainder, List<Class<? extends TType>> outputTypes,
+      List<Shape> outputShapes, BatchDataset.Options... options) {
+    return BatchDataset.create(scope, inputDataset, batchSize, dropRemainder, outputTypes, outputShapes, options);
+  }
+
+  /**
+   *
+   * @param filenames
+   * @param compressionType
+   * @param bufferSize
+   * @param header
+   * @param fieldDelim
+   * @param useQuoteDelim
+   * @param naValue
+   * @param selectCols
+   * @param recordDefaults
+   * @param outputShapes
+   * @return a new instance of CSVDataset
+   */
+  public CSVDataset cSVDataset(Operand<TString> filenames, Operand<TString> compressionType,
+      Operand<TInt64> bufferSize, Operand<TBool> header, Operand<TString> fieldDelim,
+      Operand<TBool> useQuoteDelim, Operand<TString> naValue, Operand<TInt64> selectCols,
+      Iterable<Operand<?>> recordDefaults, List<Shape> outputShapes) {
+    return CSVDataset.create(scope, filenames, compressionType, bufferSize, header, fieldDelim, useQuoteDelim, naValue, selectCols, recordDefaults, outputShapes);
+  }
+
+  /**
+   * Creates a dataset that concatenates `input_dataset` with `another_dataset`.
+   *
+   * @param inputDataset
+   * @param anotherDataset
+   * @param outputTypes
+   * @param outputShapes
+   * @return a new instance of ConcatenateDataset
+   */
+  public ConcatenateDataset concatenateDataset(Operand<?> inputDataset, Operand<?> anotherDataset,
+      List<Class<? extends TType>> outputTypes, List<Shape> outputShapes) {
+    return ConcatenateDataset.create(scope, inputDataset, anotherDataset, outputTypes, outputShapes);
+  }
+
+  /**
+   * A container for an iterator resource.
+   *
+   * @param handle A handle to the iterator to delete.
+   * @param deleter A variant deleter.
+   * @return a new instance of DeleteIterator
+   */
+  public DeleteIterator deleteIterator(Operand<?> handle, Operand<?> deleter) {
+    return DeleteIterator.create(scope, handle, deleter);
+  }
+
+  /**
+   * Converts the given variant tensor to an iterator and stores it in the given resource.
+   *
+   * @param resourceHandle A handle to an iterator resource.
+   * @param serialized A variant tensor storing the state of the iterator contained in the
+   *  resource.
+   * @return a new instance of DeserializeIterator
+   */
+  public DeserializeIterator deserializeIterator(Operand<?> resourceHandle, Operand<?> serialized) {
+    return DeserializeIterator.create(scope, resourceHandle, serialized);
+  }
+
+  /**
+   *
+   * @param sharedName
+   * @param container
+   * @param outputTypes
+   * @param outputShapes
+   * @return a new instance of Iterator
+   */
+  public Iterator iterator(String sharedName, String container,
+      List<Class<? extends TType>> outputTypes, List<Shape> outputShapes) {
+    return Iterator.create(scope, sharedName, container, outputTypes, outputShapes);
+  }
+
+  /**
+   * Gets the next output from the given iterator .
+   *
+   * @param iterator
+   * @param outputTypes
+   * @param outputShapes
+   * @return a new instance of IteratorGetNext
+   */
+  public IteratorGetNext iteratorGetNext(Operand<?> iterator,
+      List<Class<? extends TType>> outputTypes, List<Shape> outputShapes) {
+    return IteratorGetNext.create(scope, iterator, outputTypes, outputShapes);
+  }
+
+  /**
+   * Gets the next output from the given iterator as an Optional variant.
+   *
+   * @param iterator
+   * @param outputTypes
+   * @param outputShapes
+   * @return a new instance of IteratorGetNextAsOptional
+   */
+  public IteratorGetNextAsOptional iteratorGetNextAsOptional(Operand<?> iterator,
+      List<Class<? extends TType>> outputTypes, List<Shape> outputShapes) {
+    return IteratorGetNextAsOptional.create(scope, iterator, outputTypes, outputShapes);
+  }
+
+  /**
+   * Gets the next output from the given iterator.
+   *  <p>
+   *  This operation is a synchronous version IteratorGetNext. It should only be used
+   *  in situations where the iterator does not block the calling thread, or where
+   *  the calling thread is not a member of the thread pool used to execute parallel
+   *  operations (e.g. in eager mode).
+   *
+   * @param iterator
+   * @param outputTypes
+   * @param outputShapes
+   * @return a new instance of IteratorGetNextSync
+   */
+  public IteratorGetNextSync iteratorGetNextSync(Operand<?> iterator,
+      List<Class<? extends TType>> outputTypes, List<Shape> outputShapes) {
+    return IteratorGetNextSync.create(scope, iterator, outputTypes, outputShapes);
+  }
+
+  /**
+   * Converts the given `resource_handle` representing an iterator to a string.
+   *
+   * @param resourceHandle A handle to an iterator resource.
+   * @return a new instance of IteratorToStringHandle
+   */
+  public IteratorToStringHandle iteratorToStringHandle(Operand<?> resourceHandle) {
+    return IteratorToStringHandle.create(scope, resourceHandle);
+  }
+
+  /**
+   * Makes a new iterator from the given `dataset` and stores it in `iterator`.
+   *  <p>
+   *  This operation may be executed multiple times. Each execution will reset the
+   *  iterator in `iterator` to the first element of `dataset`.
+   *
+   * @param dataset
+   * @param iterator
+   * @return a new instance of MakeIterator
+   */
+  public MakeIterator makeIterator(Operand<?> dataset, Operand<?> iterator) {
+    return MakeIterator.create(scope, dataset, iterator);
+  }
+
+  /**
+   * Constructs an Optional variant from a tuple of tensors.
+   *
+   * @param components
+   * @return a new instance of OptionalFromValue
+   */
+  public OptionalFromValue optionalFromValue(Iterable<Operand<?>> components) {
+    return OptionalFromValue.create(scope, components);
+  }
+
+  /**
+   * Returns the value stored in an Optional variant or raises an error if none exists.
+   *
+   * @param optional
+   * @param outputTypes
+   * @param outputShapes
+   * @return a new instance of OptionalGetValue
+   */
+  public OptionalGetValue optionalGetValue(Operand<?> optional,
+      List<Class<? extends TType>> outputTypes, List<Shape> outputShapes) {
+    return OptionalGetValue.create(scope, optional, outputTypes, outputShapes);
+  }
+
+  /**
+   * Returns true if and only if the given Optional variant has a value.
+   *
+   * @param optional
+   * @return a new instance of OptionalHasValue
+   */
+  public OptionalHasValue optionalHasValue(Operand<?> optional) {
+    return OptionalHasValue.create(scope, optional);
+  }
+
+  /**
+   * Creates an Optional variant with no value.
+   *
+   * @return a new instance of OptionalNone
+   */
+  public OptionalNone optionalNone() {
+    return OptionalNone.create(scope);
+  }
+
+  /**
+   * Creates a dataset with a range of values. Corresponds to python's xrange.
+   *
+   * @param start corresponds to start in python's xrange().
+   * @param stop corresponds to stop in python's xrange().
+   * @param step corresponds to step in python's xrange().
+   * @param outputTypes
+   * @param outputShapes
+   * @return a new instance of RangeDataset
+   */
+  public RangeDataset rangeDataset(Operand<TInt64> start, Operand<TInt64> stop,
+      Operand<TInt64> step, List<Class<? extends TType>> outputTypes, List<Shape> outputShapes) {
+    return RangeDataset.create(scope, start, stop, step, outputTypes, outputShapes);
+  }
+
+  /**
+   * Creates a dataset that emits the outputs of `input_dataset` `count` times.
+   *
+   * @param inputDataset
+   * @param count A scalar representing the number of times that `input_dataset` should
+   *  be repeated. A value of `-1` indicates that it should be repeated infinitely.
+   * @param outputTypes
+   * @param outputShapes
+   * @return a new instance of RepeatDataset
+   */
+  public RepeatDataset repeatDataset(Operand<?> inputDataset, Operand<TInt64> count,
+      List<Class<? extends TType>> outputTypes, List<Shape> outputShapes) {
+    return RepeatDataset.create(scope, inputDataset, count, outputTypes, outputShapes);
+  }
+
+  /**
+   * Converts the given `resource_handle` representing an iterator to a variant tensor.
+   *
+   * @param resourceHandle A handle to an iterator resource.
+   * @param options carries optional attributes values
+   * @return a new instance of SerializeIterator
+   */
+  public SerializeIterator serializeIterator(Operand<?> resourceHandle,
+      SerializeIterator.Options... options) {
+    return SerializeIterator.create(scope, resourceHandle, options);
+  }
+
+  /**
+   * Creates a dataset that skips `count` elements from the `input_dataset`.
+   *
+   * @param inputDataset
+   * @param count A scalar representing the number of elements from the `input_dataset`
+   *  that should be skipped.  If count is -1, skips everything.
+   * @param outputTypes
+   * @param outputShapes
+   * @return a new instance of SkipDataset
+   */
+  public SkipDataset skipDataset(Operand<?> inputDataset, Operand<TInt64> count,
+      List<Class<? extends TType>> outputTypes, List<Shape> outputShapes) {
+    return SkipDataset.create(scope, inputDataset, count, outputTypes, outputShapes);
+  }
+
+  /**
+   * Creates a dataset that contains `count` elements from the `input_dataset`.
+   *
+   * @param inputDataset
+   * @param count A scalar representing the number of elements from the `input_dataset`
+   *  that should be taken. A value of `-1` indicates that all of `input_dataset`
+   *  is taken.
+   * @param outputTypes
+   * @param outputShapes
+   * @return a new instance of TakeDataset
+   */
+  public TakeDataset takeDataset(Operand<?> inputDataset, Operand<TInt64> count,
+      List<Class<? extends TType>> outputTypes, List<Shape> outputShapes) {
+    return TakeDataset.create(scope, inputDataset, count, outputTypes, outputShapes);
+  }
+
+  /**
+   * Creates a dataset that emits each dim-0 slice of `components` once.
+   *
+   * @param components
+   * @param outputShapes
+   * @return a new instance of TensorSliceDataset
+   */
+  public TensorSliceDataset tensorSliceDataset(Iterable<Operand<?>> components,
+      List<Shape> outputShapes) {
+    return TensorSliceDataset.create(scope, components, outputShapes);
+  }
+
+  /**
+   * Creates a dataset that emits the lines of one or more text files.
+   *
+   * @param filenames A scalar or a vector containing the name(s) of the file(s) to be
+   *  read.
+   * @param compressionType A scalar containing either (i) the empty string (no
+   *  compression), (ii) "ZLIB", or (iii) "GZIP".
+   * @param bufferSize A scalar containing the number of bytes to buffer.
+   * @return a new instance of TextLineDataset
+   */
+  public TextLineDataset textLineDataset(Operand<TString> filenames,
+      Operand<TString> compressionType, Operand<TInt64> bufferSize) {
+    return TextLineDataset.create(scope, filenames, compressionType, bufferSize);
+  }
+
+  /**
+   * Creates a dataset that emits the records from one or more TFRecord files.
+   *
+   * @param filenames A scalar or vector containing the name(s) of the file(s) to be
+   *  read.
+   * @param compressionType A scalar containing either (i) the empty string (no
+   *  compression), (ii) "ZLIB", or (iii) "GZIP".
+   * @param bufferSize A scalar representing the number of bytes to buffer. A value of
+   *  0 means no buffering will be performed.
+   * @return a new instance of TfRecordDataset
+   */
+  public TfRecordDataset tfRecordDataset(Operand<TString> filenames,
+      Operand<TString> compressionType, Operand<TInt64> bufferSize) {
+    return TfRecordDataset.create(scope, filenames, compressionType, bufferSize);
+  }
+
+  /**
+   * Creates a dataset that zips together `input_datasets`.
+   *  <p>
+   *  The elements of the resulting dataset are created by zipping corresponding
+   *  elements from each of the input datasets.
+   *  <p>
+   *  The size of the resulting dataset will match the size of the smallest input
+   *  dataset, and no error will be raised if input datasets have different sizes.
+   *
+   * @param inputDatasets List of `N` variant Tensors representing datasets to be zipped together.
+   * @param outputTypes
+   * @param outputShapes
+   * @return a new instance of ZipDataset
+   */
+  public ZipDataset zipDataset(Iterable<Operand<?>> inputDatasets,
+      List<Class<? extends TType>> outputTypes, List<Shape> outputShapes) {
+    return ZipDataset.create(scope, inputDatasets, outputTypes, outputShapes);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/DtypesOps.java b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/DtypesOps.java
new file mode 100644
index 00000000000..824f3383f2b
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/DtypesOps.java
@@ -0,0 +1,104 @@
+// Copyright 2020 The TensorFlow Authors. All Rights Reserved.
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+// ==============================================================================
+//
+// This class has been generated, DO NOT EDIT!
+//
+package org.tensorflow.op;
+
+import org.tensorflow.Operand;
+import org.tensorflow.op.dtypes.AsString;
+import org.tensorflow.op.dtypes.Cast;
+import org.tensorflow.op.dtypes.Complex;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * An API for building {@code dtypes} operations as {@link Op Op}s
+ *
+ * @see {@link Ops}
+ */
+public final class DtypesOps {
+  private final Scope scope;
+
+  DtypesOps(Scope scope) {
+    this.scope = scope;
+  }
+
+  /**
+   * Converts each entry in the given tensor to strings.
+   *  <p>
+   *  Supports many numeric types and boolean.
+   *  <p>
+   *  For Unicode, see the
+   *  [https://www.tensorflow.org/tutorials/representation/unicode](Working with Unicode text)
+   *  tutorial.
+   *  <p>
+   *  Examples:
+   *  <p>
+   *  >>> tf.strings.as_string([3, 2])
+   *  <tf.Tensor: shape=(2,), dtype=string, numpy=array([b'3', b'2'], dtype=object)>
+   *  >>> tf.strings.as_string([3.1415926, 2.71828], precision=2).numpy()
+   *  array([b'3.14', b'2.72'], dtype=object)
+   *
+   * @param input
+   * @param options carries optional attributes values
+   * @return a new instance of AsString
+   */
+  public <T extends TType> AsString asString(Operand<T> input, AsString.Options... options) {
+    return AsString.create(scope, input, options);
+  }
+
+  /**
+   * Cast x of type SrcT to y of DstT.
+   *
+   * @param <U> data type for {@code y()} output
+   * @param x
+   * @param DstT
+   * @param options carries optional attributes values
+   * @return a new instance of Cast
+   */
+  public <U extends TType, T extends TType> Cast<U> cast(Operand<T> x, Class<U> DstT,
+      Cast.Options... options) {
+    return Cast.create(scope, x, DstT, options);
+  }
+
+  /**
+   * Converts two real numbers to a complex number.
+   *  <p>
+   *  Given a tensor `real` representing the real part of a complex number, and a
+   *  tensor `imag` representing the imaginary part of a complex number, this
+   *  operation returns complex numbers elementwise of the form \\(a + bj\\), where
+   *  <i>a</i> represents the `real` part and <i>b</i> represents the `imag` part.
+   *  <p>
+   *  The input tensors `real` and `imag` must have the same shape.
+   *  <p>
+   *  For example:
+   *  <pre>{@code
+   *  # tensor 'real' is [2.25, 3.25]
+   *  # tensor `imag` is [4.75, 5.75]
+   *  tf.complex(real, imag) ==> [[2.25 + 4.75j], [3.25 + 5.75j]]
+   *  }</pre>
+   *
+   * @param <U> data type for {@code out()} output
+   * @param real
+   * @param imag
+   * @param Tout
+   * @return a new instance of Complex
+   */
+  public <U extends TType, T extends TNumber> Complex<U> complex(Operand<T> real, Operand<T> imag,
+      Class<U> Tout) {
+    return Complex.create(scope, real, imag, Tout);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/ImageOps.java b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/ImageOps.java
new file mode 100644
index 00000000000..b0fa7751b8a
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/ImageOps.java
@@ -0,0 +1,944 @@
+// Copyright 2020 The TensorFlow Authors. All Rights Reserved.
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+// ==============================================================================
+//
+// This class has been generated, DO NOT EDIT!
+//
+package org.tensorflow.op;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.op.image.AdjustContrast;
+import org.tensorflow.op.image.AdjustHue;
+import org.tensorflow.op.image.AdjustSaturation;
+import org.tensorflow.op.image.CombinedNonMaxSuppression;
+import org.tensorflow.op.image.CropAndResize;
+import org.tensorflow.op.image.CropAndResizeGradBoxes;
+import org.tensorflow.op.image.CropAndResizeGradImage;
+import org.tensorflow.op.image.DecodeAndCropJpeg;
+import org.tensorflow.op.image.DecodeBmp;
+import org.tensorflow.op.image.DecodeGif;
+import org.tensorflow.op.image.DecodeJpeg;
+import org.tensorflow.op.image.DecodePng;
+import org.tensorflow.op.image.DrawBoundingBoxes;
+import org.tensorflow.op.image.EncodeJpeg;
+import org.tensorflow.op.image.EncodeJpegVariableQuality;
+import org.tensorflow.op.image.EncodePng;
+import org.tensorflow.op.image.ExtractImagePatches;
+import org.tensorflow.op.image.ExtractJpegShape;
+import org.tensorflow.op.image.HsvToRgb;
+import org.tensorflow.op.image.NonMaxSuppression;
+import org.tensorflow.op.image.NonMaxSuppressionWithOverlaps;
+import org.tensorflow.op.image.QuantizedResizeBilinear;
+import org.tensorflow.op.image.RandomCrop;
+import org.tensorflow.op.image.ResizeArea;
+import org.tensorflow.op.image.ResizeBicubic;
+import org.tensorflow.op.image.ResizeBilinear;
+import org.tensorflow.op.image.ResizeNearestNeighbor;
+import org.tensorflow.op.image.RgbToHsv;
+import org.tensorflow.op.image.SampleDistortedBoundingBox;
+import org.tensorflow.op.image.ScaleAndTranslate;
+import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.TString;
+import org.tensorflow.types.TUint8;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * An API for building {@code image} operations as {@link Op Op}s
+ *
+ * @see {@link Ops}
+ */
+public final class ImageOps {
+  private final Scope scope;
+
+  ImageOps(Scope scope) {
+    this.scope = scope;
+  }
+
+  /**
+   * Adjust the contrast of one or more images.
+   *  <p>
+   *  `images` is a tensor of at least 3 dimensions.  The last 3 dimensions are
+   *  interpreted as `[height, width, channels]`.  The other dimensions only
+   *  represent a collection of images, such as `[batch, height, width, channels].`
+   *  <p>
+   *  Contrast is adjusted independently for each channel of each image.
+   *  <p>
+   *  For each channel, the Op first computes the mean of the image pixels in the
+   *  channel and then adjusts each component of each pixel to
+   *  `(x - mean) * contrast_factor + mean`.
+   *
+   * @param <T> data type for {@code output()} output
+   * @param images Images to adjust.  At least 3-D.
+   * @param contrastFactor A float multiplier for adjusting contrast.
+   * @return a new instance of AdjustContrast
+   */
+  public <T extends TNumber> AdjustContrast<T> adjustContrast(Operand<T> images,
+      Operand<TFloat32> contrastFactor) {
+    return AdjustContrast.create(scope, images, contrastFactor);
+  }
+
+  /**
+   * Adjust the hue of one or more images.
+   *  <p>
+   *  `images` is a tensor of at least 3 dimensions.  The last dimension is
+   *  interpreted as channels, and must be three.
+   *  <p>
+   *  The input image is considered in the RGB colorspace. Conceptually, the RGB
+   *  colors are first mapped into HSV. A delta is then applied all the hue values,
+   *  and then remapped back to RGB colorspace.
+   *
+   * @param <T> data type for {@code output()} output
+   * @param images Images to adjust.  At least 3-D.
+   * @param delta A float delta to add to the hue.
+   * @return a new instance of AdjustHue
+   */
+  public <T extends TNumber> AdjustHue<T> adjustHue(Operand<T> images, Operand<TFloat32> delta) {
+    return AdjustHue.create(scope, images, delta);
+  }
+
+  /**
+   * Adjust the saturation of one or more images.
+   *  <p>
+   *  `images` is a tensor of at least 3 dimensions.  The last dimension is
+   *  interpreted as channels, and must be three.
+   *  <p>
+   *  The input image is considered in the RGB colorspace. Conceptually, the RGB
+   *  colors are first mapped into HSV. A scale is then applied all the saturation
+   *  values, and then remapped back to RGB colorspace.
+   *
+   * @param <T> data type for {@code output()} output
+   * @param images Images to adjust.  At least 3-D.
+   * @param scale A float scale to add to the saturation.
+   * @return a new instance of AdjustSaturation
+   */
+  public <T extends TNumber> AdjustSaturation<T> adjustSaturation(Operand<T> images,
+      Operand<TFloat32> scale) {
+    return AdjustSaturation.create(scope, images, scale);
+  }
+
+  /**
+   * Greedily selects a subset of bounding boxes in descending order of score,
+   *  <p>
+   *  This operation performs non_max_suppression on the inputs per batch, across
+   *  all classes.
+   *  Prunes away boxes that have high intersection-over-union (IOU) overlap
+   *  with previously selected boxes.  Bounding boxes are supplied as
+   *  [y1, x1, y2, x2], where (y1, x1) and (y2, x2) are the coordinates of any
+   *  diagonal pair of box corners and the coordinates can be provided as normalized
+   *  (i.e., lying in the interval [0, 1]) or absolute.  Note that this algorithm
+   *  is agnostic to where the origin is in the coordinate system. Also note that
+   *  this algorithm is invariant to orthogonal transformations and translations
+   *  of the coordinate system; thus translating or reflections of the coordinate
+   *  system result in the same boxes being selected by the algorithm.
+   *  The output of this operation is the final boxes, scores and classes tensor
+   *  returned after performing non_max_suppression.
+   *
+   * @param boxes A 4-D float tensor of shape `[batch_size, num_boxes, q, 4]`. If `q` is 1 then
+   *  same boxes are used for all classes otherwise, if `q` is equal to number of
+   *  classes, class-specific boxes are used.
+   * @param scores A 3-D float tensor of shape `[batch_size, num_boxes, num_classes]`
+   *  representing a single score corresponding to each box (each row of boxes).
+   * @param maxOutputSizePerClass A scalar integer tensor representing the maximum number of
+   *  boxes to be selected by non max suppression per class
+   * @param maxTotalSize A scalar representing maximum number of boxes retained over all classes.
+   * @param iouThreshold A 0-D float tensor representing the threshold for deciding whether
+   *  boxes overlap too much with respect to IOU.
+   * @param scoreThreshold A 0-D float tensor representing the threshold for deciding when to remove
+   *  boxes based on score.
+   * @param options carries optional attributes values
+   * @return a new instance of CombinedNonMaxSuppression
+   */
+  public CombinedNonMaxSuppression combinedNonMaxSuppression(Operand<TFloat32> boxes,
+      Operand<TFloat32> scores, Operand<TInt32> maxOutputSizePerClass, Operand<TInt32> maxTotalSize,
+      Operand<TFloat32> iouThreshold, Operand<TFloat32> scoreThreshold,
+      CombinedNonMaxSuppression.Options... options) {
+    return CombinedNonMaxSuppression.create(scope, boxes, scores, maxOutputSizePerClass, maxTotalSize, iouThreshold, scoreThreshold, options);
+  }
+
+  /**
+   * Extracts crops from the input image tensor and resizes them.
+   *  <p>
+   *  Extracts crops from the input image tensor and resizes them using bilinear
+   *  sampling or nearest neighbor sampling (possibly with aspect ratio change) to a
+   *  common output size specified by `crop_size`. This is more general than the
+   *  `crop_to_bounding_box` op which extracts a fixed size slice from the input image
+   *  and does not allow resizing or aspect ratio change.
+   *  <p>
+   *  Returns a tensor with `crops` from the input `image` at positions defined at the
+   *  bounding box locations in `boxes`. The cropped boxes are all resized (with
+   *  bilinear or nearest neighbor interpolation) to a fixed
+   *  `size = [crop_height, crop_width]`. The result is a 4-D tensor
+   *  `[num_boxes, crop_height, crop_width, depth]`. The resizing is corner aligned.
+   *  In particular, if `boxes = [[0, 0, 1, 1]]`, the method will give identical
+   *  results to using `tf.image.resize_bilinear()` or
+   *  `tf.image.resize_nearest_neighbor()`(depends on the `method` argument) with
+   *  `align_corners=True`.
+   *
+   * @param image A 4-D tensor of shape `[batch, image_height, image_width, depth]`.
+   *  Both `image_height` and `image_width` need to be positive.
+   * @param boxes A 2-D tensor of shape `[num_boxes, 4]`. The `i`-th row of the tensor
+   *  specifies the coordinates of a box in the `box_ind[i]` image and is specified
+   *  in normalized coordinates `[y1, x1, y2, x2]`. A normalized coordinate value of
+   *  `y` is mapped to the image coordinate at `y * (image_height - 1)`, so as the
+   *  `[0, 1]` interval of normalized image height is mapped to
+   *  `[0, image_height - 1]` in image height coordinates. We do allow `y1` > `y2`, in
+   *  which case the sampled crop is an up-down flipped version of the original
+   *  image. The width dimension is treated similarly. Normalized coordinates
+   *  outside the `[0, 1]` range are allowed, in which case we use
+   *  `extrapolation_value` to extrapolate the input image values.
+   * @param boxInd A 1-D tensor of shape `[num_boxes]` with int32 values in `[0, batch)`.
+   *  The value of `box_ind[i]` specifies the image that the `i`-th box refers to.
+   * @param cropSize A 1-D tensor of 2 elements, `size = [crop_height, crop_width]`. All
+   *  cropped image patches are resized to this size. The aspect ratio of the image
+   *  content is not preserved. Both `crop_height` and `crop_width` need to be
+   *  positive.
+   * @param options carries optional attributes values
+   * @return a new instance of CropAndResize
+   */
+  public <T extends TNumber> CropAndResize cropAndResize(Operand<T> image, Operand<TFloat32> boxes,
+      Operand<TInt32> boxInd, Operand<TInt32> cropSize, CropAndResize.Options... options) {
+    return CropAndResize.create(scope, image, boxes, boxInd, cropSize, options);
+  }
+
+  /**
+   * Computes the gradient of the crop_and_resize op wrt the input boxes tensor.
+   *
+   * @param grads A 4-D tensor of shape `[num_boxes, crop_height, crop_width, depth]`.
+   * @param image A 4-D tensor of shape `[batch, image_height, image_width, depth]`.
+   *  Both `image_height` and `image_width` need to be positive.
+   * @param boxes A 2-D tensor of shape `[num_boxes, 4]`. The `i`-th row of the tensor
+   *  specifies the coordinates of a box in the `box_ind[i]` image and is specified
+   *  in normalized coordinates `[y1, x1, y2, x2]`. A normalized coordinate value of
+   *  `y` is mapped to the image coordinate at `y * (image_height - 1)`, so as the
+   *  `[0, 1]` interval of normalized image height is mapped to
+   *  `[0, image_height - 1] in image height coordinates. We do allow y1 > y2, in
+   *  which case the sampled crop is an up-down flipped version of the original
+   *  image. The width dimension is treated similarly. Normalized coordinates
+   *  outside the `[0, 1]` range are allowed, in which case we use
+   *  `extrapolation_value` to extrapolate the input image values.
+   * @param boxInd A 1-D tensor of shape `[num_boxes]` with int32 values in `[0, batch)`.
+   *  The value of `box_ind[i]` specifies the image that the `i`-th box refers to.
+   * @param options carries optional attributes values
+   * @return a new instance of CropAndResizeGradBoxes
+   */
+  public <T extends TNumber> CropAndResizeGradBoxes cropAndResizeGradBoxes(Operand<TFloat32> grads,
+      Operand<T> image, Operand<TFloat32> boxes, Operand<TInt32> boxInd,
+      CropAndResizeGradBoxes.Options... options) {
+    return CropAndResizeGradBoxes.create(scope, grads, image, boxes, boxInd, options);
+  }
+
+  /**
+   * Computes the gradient of the crop_and_resize op wrt the input image tensor.
+   *
+   * @param <T> data type for {@code output()} output
+   * @param grads A 4-D tensor of shape `[num_boxes, crop_height, crop_width, depth]`.
+   * @param boxes A 2-D tensor of shape `[num_boxes, 4]`. The `i`-th row of the tensor
+   *  specifies the coordinates of a box in the `box_ind[i]` image and is specified
+   *  in normalized coordinates `[y1, x1, y2, x2]`. A normalized coordinate value of
+   *  `y` is mapped to the image coordinate at `y * (image_height - 1)`, so as the
+   *  `[0, 1]` interval of normalized image height is mapped to
+   *  `[0, image_height - 1] in image height coordinates. We do allow y1 > y2, in
+   *  which case the sampled crop is an up-down flipped version of the original
+   *  image. The width dimension is treated similarly. Normalized coordinates
+   *  outside the `[0, 1]` range are allowed, in which case we use
+   *  `extrapolation_value` to extrapolate the input image values.
+   * @param boxInd A 1-D tensor of shape `[num_boxes]` with int32 values in `[0, batch)`.
+   *  The value of `box_ind[i]` specifies the image that the `i`-th box refers to.
+   * @param imageSize A 1-D tensor with value `[batch, image_height, image_width, depth]`
+   *  containing the original image size. Both `image_height` and `image_width` need
+   *  to be positive.
+   * @param T
+   * @param options carries optional attributes values
+   * @return a new instance of CropAndResizeGradImage
+   */
+  public <T extends TNumber> CropAndResizeGradImage<T> cropAndResizeGradImage(
+      Operand<TFloat32> grads, Operand<TFloat32> boxes, Operand<TInt32> boxInd,
+      Operand<TInt32> imageSize, Class<T> T, CropAndResizeGradImage.Options... options) {
+    return CropAndResizeGradImage.create(scope, grads, boxes, boxInd, imageSize, T, options);
+  }
+
+  /**
+   * Decode and Crop a JPEG-encoded image to a uint8 tensor.
+   *  <p>
+   *  The attr `channels` indicates the desired number of color channels for the
+   *  decoded image.
+   *  <p>
+   *  Accepted values are:
+   *  <ul>
+   *  <li>
+   *  0: Use the number of channels in the JPEG-encoded image.
+   *  </li>
+   *  <li>
+   *  1: output a grayscale image.
+   *  </li>
+   *  <li>
+   *  3: output an RGB image.
+   *  </li>
+   *  </ul>
+   *  If needed, the JPEG-encoded image is transformed to match the requested number
+   *  of color channels.
+   *  <p>
+   *  The attr `ratio` allows downscaling the image by an integer factor during
+   *  decoding.  Allowed values are: 1, 2, 4, and 8.  This is much faster than
+   *  downscaling the image later.
+   *  <p>
+   *  It is equivalent to a combination of decode and crop, but much faster by only
+   *  decoding partial jpeg image.
+   *
+   * @param contents 0-D.  The JPEG-encoded image.
+   * @param cropWindow 1-D.  The crop window: [crop_y, crop_x, crop_height, crop_width].
+   * @param options carries optional attributes values
+   * @return a new instance of DecodeAndCropJpeg
+   */
+  public DecodeAndCropJpeg decodeAndCropJpeg(Operand<TString> contents, Operand<TInt32> cropWindow,
+      DecodeAndCropJpeg.Options... options) {
+    return DecodeAndCropJpeg.create(scope, contents, cropWindow, options);
+  }
+
+  /**
+   * Decode the first frame of a BMP-encoded image to a uint8 tensor.
+   *  <p>
+   *  The attr `channels` indicates the desired number of color channels for the
+   *  decoded image.
+   *  <p>
+   *  Accepted values are:
+   *  <ul>
+   *  <li>
+   *  0: Use the number of channels in the BMP-encoded image.
+   *  </li>
+   *  <li>
+   *  3: output an RGB image.
+   *  </li>
+   *  <li>
+   *  4: output an RGBA image.
+   *
+   * @param contents 0-D.  The BMP-encoded image.
+   * @param options carries optional attributes values
+   * @return a new instance of DecodeBmp
+   */
+  public DecodeBmp decodeBmp(Operand<TString> contents, DecodeBmp.Options... options) {
+    return DecodeBmp.create(scope, contents, options);
+  }
+
+  /**
+   * Decode the frame(s) of a GIF-encoded image to a uint8 tensor.
+   *  <p>
+   *  GIF images with frame or transparency compression are not supported.
+   *  On Linux and MacOS systems, convert animated GIFs from compressed to
+   *  uncompressed by running:
+   *  <p>
+   *      convert $src.gif -coalesce $dst.gif
+   *  <p>
+   *  This op also supports decoding JPEGs and PNGs, though it is cleaner to use
+   *  `tf.io.decode_image`.
+   *
+   * @param contents 0-D.  The GIF-encoded image.
+   * @return a new instance of DecodeGif
+   */
+  public DecodeGif decodeGif(Operand<TString> contents) {
+    return DecodeGif.create(scope, contents);
+  }
+
+  /**
+   * Decode a JPEG-encoded image to a uint8 tensor.
+   *  <p>
+   *  The attr `channels` indicates the desired number of color channels for the
+   *  decoded image.
+   *  <p>
+   *  Accepted values are:
+   *  <ul>
+   *  <li>
+   *  0: Use the number of channels in the JPEG-encoded image.
+   *  </li>
+   *  <li>
+   *  1: output a grayscale image.
+   *  </li>
+   *  <li>
+   *  3: output an RGB image.
+   *  </li>
+   *  </ul>
+   *  If needed, the JPEG-encoded image is transformed to match the requested number
+   *  of color channels.
+   *  <p>
+   *  The attr `ratio` allows downscaling the image by an integer factor during
+   *  decoding.  Allowed values are: 1, 2, 4, and 8.  This is much faster than
+   *  downscaling the image later.
+   *  <p>
+   *  This op also supports decoding PNGs and non-animated GIFs since the interface is
+   *  the same, though it is cleaner to use `tf.io.decode_image`.
+   *
+   * @param contents 0-D.  The JPEG-encoded image.
+   * @param options carries optional attributes values
+   * @return a new instance of DecodeJpeg
+   */
+  public DecodeJpeg decodeJpeg(Operand<TString> contents, DecodeJpeg.Options... options) {
+    return DecodeJpeg.create(scope, contents, options);
+  }
+
+  /**
+   * Decode a PNG-encoded image to a uint8 or uint16 tensor.
+   *  <p>
+   *  The attr `channels` indicates the desired number of color channels for the
+   *  decoded image.
+   *  <p>
+   *  Accepted values are:
+   *  <ul>
+   *  <li>
+   *  0: Use the number of channels in the PNG-encoded image.
+   *  </li>
+   *  <li>
+   *  1: output a grayscale image.
+   *  </li>
+   *  <li>
+   *  3: output an RGB image.
+   *  </li>
+   *  <li>
+   *  4: output an RGBA image.
+   *  </li>
+   *  </ul>
+   *  If needed, the PNG-encoded image is transformed to match the requested number
+   *  of color channels.
+   *  <p>
+   *  This op also supports decoding JPEGs and non-animated GIFs since the interface
+   *  is the same, though it is cleaner to use `tf.io.decode_image`.
+   *
+   * @param <T> data type for {@code image()} output
+   * @param contents 0-D.  The PNG-encoded image.
+   * @param options carries optional attributes values
+   * @return a new instance of DecodePng
+   */
+  public DecodePng<TUint8> decodePng(Operand<TString> contents, DecodePng.Options... options) {
+    return DecodePng.create(scope, contents, options);
+  }
+
+  /**
+   * Decode a PNG-encoded image to a uint8 or uint16 tensor.
+   *  <p>
+   *  The attr `channels` indicates the desired number of color channels for the
+   *  decoded image.
+   *  <p>
+   *  Accepted values are:
+   *  <ul>
+   *  <li>
+   *  0: Use the number of channels in the PNG-encoded image.
+   *  </li>
+   *  <li>
+   *  1: output a grayscale image.
+   *  </li>
+   *  <li>
+   *  3: output an RGB image.
+   *  </li>
+   *  <li>
+   *  4: output an RGBA image.
+   *  </li>
+   *  </ul>
+   *  If needed, the PNG-encoded image is transformed to match the requested number
+   *  of color channels.
+   *  <p>
+   *  This op also supports decoding JPEGs and non-animated GIFs since the interface
+   *  is the same, though it is cleaner to use `tf.io.decode_image`.
+   *
+   * @param <T> data type for {@code image()} output
+   * @param contents 0-D.  The PNG-encoded image.
+   * @param dtype
+   * @param options carries optional attributes values
+   * @return a new instance of DecodePng
+   */
+  public <T extends TNumber> DecodePng<T> decodePng(Operand<TString> contents, Class<T> dtype,
+      DecodePng.Options... options) {
+    return DecodePng.create(scope, contents, dtype, options);
+  }
+
+  /**
+   * Draw bounding boxes on a batch of images.
+   *  <p>
+   *  Outputs a copy of `images` but draws on top of the pixels zero or more bounding
+   *  boxes specified by the locations in `boxes`. The coordinates of the each
+   *  bounding box in `boxes` are encoded as `[y_min, x_min, y_max, x_max]`. The
+   *  bounding box coordinates are floats in `[0.0, 1.0]` relative to the width and
+   *  height of the underlying image.
+   *  <p>
+   *  For example, if an image is 100 x 200 pixels (height x width) and the bounding
+   *  box is `[0.1, 0.2, 0.5, 0.9]`, the upper-left and bottom-right coordinates of
+   *  the bounding box will be `(40, 10)` to `(100, 50)` (in (x,y) coordinates).
+   *  <p>
+   *  Parts of the bounding box may fall outside the image.
+   *
+   * @param <T> data type for {@code output()} output
+   * @param images 4-D with shape `[batch, height, width, depth]`. A batch of images.
+   * @param boxes 3-D with shape `[batch, num_bounding_boxes, 4]` containing bounding
+   *  boxes.
+   * @param colors 2-D. A list of RGBA colors to cycle through for the boxes.
+   * @return a new instance of DrawBoundingBoxes
+   */
+  public <T extends TNumber> DrawBoundingBoxes<T> drawBoundingBoxes(Operand<T> images,
+      Operand<TFloat32> boxes, Operand<TFloat32> colors) {
+    return DrawBoundingBoxes.create(scope, images, boxes, colors);
+  }
+
+  /**
+   * JPEG-encode an image.
+   *  <p>
+   *  `image` is a 3-D uint8 Tensor of shape `[height, width, channels]`.
+   *  <p>
+   *  The attr `format` can be used to override the color format of the encoded
+   *  output.  Values can be:
+   *  <ul>
+   *  <li>
+   *  `''`: Use a default format based on the number of channels in the image.
+   *  </li>
+   *  <li>
+   *  `grayscale`: Output a grayscale JPEG image.  The `channels` dimension
+   *      of `image` must be 1.
+   *  </li>
+   *  <li>
+   *  `rgb`: Output an RGB JPEG image. The `channels` dimension
+   *      of `image` must be 3.
+   *  </li>
+   *  </ul>
+   *  If `format` is not specified or is the empty string, a default format is picked
+   *  in function of the number of channels in `image`:
+   *  <ul>
+   *  <li>
+   *  1: Output a grayscale image.
+   *  </li>
+   *  <li>
+   *  3: Output an RGB image.
+   *
+   * @param image 3-D with shape `[height, width, channels]`.
+   * @param options carries optional attributes values
+   * @return a new instance of EncodeJpeg
+   */
+  public EncodeJpeg encodeJpeg(Operand<TUint8> image, EncodeJpeg.Options... options) {
+    return EncodeJpeg.create(scope, image, options);
+  }
+
+  /**
+   * JPEG encode input image with provided compression quality.
+   *  <p>
+   *  `image` is a 3-D uint8 Tensor of shape `[height, width, channels]`.
+   *  `quality` is an int32 jpeg compression quality value between 0 and 100.
+   *
+   * @param images Images to adjust.  At least 3-D.
+   * @param quality An int quality to encode to.
+   * @return a new instance of EncodeJpegVariableQuality
+   */
+  public EncodeJpegVariableQuality encodeJpegVariableQuality(Operand<TUint8> images,
+      Operand<TInt32> quality) {
+    return EncodeJpegVariableQuality.create(scope, images, quality);
+  }
+
+  /**
+   * PNG-encode an image.
+   *  <p>
+   *  `image` is a 3-D uint8 or uint16 Tensor of shape `[height, width, channels]`
+   *  where `channels` is:
+   *  <ul>
+   *  <li>
+   *  1: for grayscale.
+   *  </li>
+   *  <li>
+   *  2: for grayscale + alpha.
+   *  </li>
+   *  <li>
+   *  3: for RGB.
+   *  </li>
+   *  <li>
+   *  4: for RGBA.
+   *  </li>
+   *  </ul>
+   *  The ZLIB compression level, `compression`, can be -1 for the PNG-encoder
+   *  default or a value from 0 to 9.  9 is the highest compression level, generating
+   *  the smallest output, but is slower.
+   *
+   * @param image 3-D with shape `[height, width, channels]`.
+   * @param options carries optional attributes values
+   * @return a new instance of EncodePng
+   */
+  public <T extends TNumber> EncodePng encodePng(Operand<T> image, EncodePng.Options... options) {
+    return EncodePng.create(scope, image, options);
+  }
+
+  /**
+   * Extract `patches` from `images` and put them in the "depth" output dimension.
+   *
+   * @param <T> data type for {@code patches()} output
+   * @param images 4-D Tensor with shape `[batch, in_rows, in_cols, depth]`.
+   * @param ksizes The size of the sliding window for each dimension of `images`.
+   * @param strides How far the centers of two consecutive patches are in
+   *  the images. Must be: `[1, stride_rows, stride_cols, 1]`.
+   * @param rates Must be: `[1, rate_rows, rate_cols, 1]`. This is the
+   *  input stride, specifying how far two consecutive patch samples are in the
+   *  input. Equivalent to extracting patches with
+   *  `patch_sizes_eff = patch_sizes + (patch_sizes - 1) * (rates - 1)`, followed by
+   *  subsampling them spatially by a factor of `rates`. This is equivalent to
+   *  `rate` in dilated (a.k.a. Atrous) convolutions.
+   * @param padding The type of padding algorithm to use.
+   * @return a new instance of ExtractImagePatches
+   */
+  public <T extends TType> ExtractImagePatches<T> extractImagePatches(Operand<T> images,
+      List<Long> ksizes, List<Long> strides, List<Long> rates, String padding) {
+    return ExtractImagePatches.create(scope, images, ksizes, strides, rates, padding);
+  }
+
+  /**
+   * Extract the shape information of a JPEG-encoded image.
+   *  <p>
+   *  This op only parses the image header, so it is much faster than DecodeJpeg.
+   *
+   * @param <T> data type for {@code imageShape()} output
+   * @param contents 0-D. The JPEG-encoded image.
+   * @return a new instance of ExtractJpegShape
+   */
+  public ExtractJpegShape<TInt32> extractJpegShape(Operand<TString> contents) {
+    return ExtractJpegShape.create(scope, contents);
+  }
+
+  /**
+   * Extract the shape information of a JPEG-encoded image.
+   *  <p>
+   *  This op only parses the image header, so it is much faster than DecodeJpeg.
+   *
+   * @param <T> data type for {@code imageShape()} output
+   * @param contents 0-D. The JPEG-encoded image.
+   * @param outputType (Optional) The output type of the operation (int32 or int64).
+   *  Defaults to int32.
+   * @return a new instance of ExtractJpegShape
+   */
+  public <T extends TNumber> ExtractJpegShape<T> extractJpegShape(Operand<TString> contents,
+      Class<T> outputType) {
+    return ExtractJpegShape.create(scope, contents, outputType);
+  }
+
+  /**
+   * Convert one or more images from HSV to RGB.
+   *  <p>
+   *  Outputs a tensor of the same shape as the `images` tensor, containing the RGB
+   *  value of the pixels. The output is only well defined if the value in `images`
+   *  are in `[0,1]`.
+   *  <p>
+   *  See `rgb_to_hsv` for a description of the HSV encoding.
+   *
+   * @param <T> data type for {@code output()} output
+   * @param images 1-D or higher rank. HSV data to convert. Last dimension must be size 3.
+   * @return a new instance of HsvToRgb
+   */
+  public <T extends TNumber> HsvToRgb<T> hsvToRgb(Operand<T> images) {
+    return HsvToRgb.create(scope, images);
+  }
+
+  /**
+   * Greedily selects a subset of bounding boxes in descending order of score,
+   *  <p>
+   *  pruning away boxes that have high intersection-over-union (IOU) overlap
+   *  with previously selected boxes.  Bounding boxes with score less than
+   *  `score_threshold` are removed.  Bounding boxes are supplied as
+   *  [y1, x1, y2, x2], where (y1, x1) and (y2, x2) are the coordinates of any
+   *  diagonal pair of box corners and the coordinates can be provided as normalized
+   *  (i.e., lying in the interval [0, 1]) or absolute.  Note that this algorithm
+   *  is agnostic to where the origin is in the coordinate system and more
+   *  generally is invariant to orthogonal transformations and translations
+   *  of the coordinate system; thus translating or reflections of the coordinate
+   *  system result in the same boxes being selected by the algorithm.
+   *  The output of this operation is a set of integers indexing into the input
+   *  collection of bounding boxes representing the selected boxes.  The bounding
+   *  box coordinates corresponding to the selected indices can then be obtained
+   *  using the `tf.gather operation`.  For example:
+   *    selected_indices = tf.image.non_max_suppression_v2(
+   *        boxes, scores, max_output_size, iou_threshold, score_threshold)
+   *    selected_boxes = tf.gather(boxes, selected_indices)
+   *  This op also supports a Soft-NMS (with Gaussian weighting) mode (c.f.
+   *  Bodla et al, https://arxiv.org/abs/1704.04503) where boxes reduce the score
+   *  of other overlapping boxes instead of directly causing them to be pruned.
+   *  To enable this Soft-NMS mode, set the `soft_nms_sigma` parameter to be
+   *  larger than 0.
+   *
+   * @param <T> data type for {@code selectedScores()} output
+   * @param boxes A 2-D float tensor of shape `[num_boxes, 4]`.
+   * @param scores A 1-D float tensor of shape `[num_boxes]` representing a single
+   *  score corresponding to each box (each row of boxes).
+   * @param maxOutputSize A scalar integer tensor representing the maximum number of
+   *  boxes to be selected by non max suppression.
+   * @param iouThreshold A 0-D float tensor representing the threshold for deciding whether
+   *  boxes overlap too much with respect to IOU.
+   * @param scoreThreshold A 0-D float tensor representing the threshold for deciding when to remove
+   *  boxes based on score.
+   * @param softNmsSigma A 0-D float tensor representing the sigma parameter for Soft NMS; see Bodla et
+   *  al (c.f. https://arxiv.org/abs/1704.04503).  When `soft_nms_sigma=0.0` (which
+   *  is default), we fall back to standard (hard) NMS.
+   * @param options carries optional attributes values
+   * @return a new instance of NonMaxSuppression
+   */
+  public <T extends TNumber> NonMaxSuppression<T> nonMaxSuppression(Operand<T> boxes,
+      Operand<T> scores, Operand<TInt32> maxOutputSize, Operand<T> iouThreshold,
+      Operand<T> scoreThreshold, Operand<T> softNmsSigma, NonMaxSuppression.Options... options) {
+    return NonMaxSuppression.create(scope, boxes, scores, maxOutputSize, iouThreshold, scoreThreshold, softNmsSigma, options);
+  }
+
+  /**
+   * Greedily selects a subset of bounding boxes in descending order of score,
+   *  <p>
+   *  pruning away boxes that have high overlaps
+   *  with previously selected boxes.  Bounding boxes with score less than
+   *  `score_threshold` are removed. N-by-n overlap values are supplied as square matrix,
+   *  which allows for defining a custom overlap criterium (eg. intersection over union,
+   *  intersection over area, etc.).
+   *  <p>
+   *  The output of this operation is a set of integers indexing into the input
+   *  collection of bounding boxes representing the selected boxes.  The bounding
+   *  box coordinates corresponding to the selected indices can then be obtained
+   *  using the `tf.gather operation`.  For example:
+   *  <p>
+   *    selected_indices = tf.image.non_max_suppression_with_overlaps(
+   *        overlaps, scores, max_output_size, overlap_threshold, score_threshold)
+   *    selected_boxes = tf.gather(boxes, selected_indices)
+   *
+   * @param overlaps A 2-D float tensor of shape `[num_boxes, num_boxes]` representing
+   *  the n-by-n box overlap values.
+   * @param scores A 1-D float tensor of shape `[num_boxes]` representing a single
+   *  score corresponding to each box (each row of boxes).
+   * @param maxOutputSize A scalar integer tensor representing the maximum number of
+   *  boxes to be selected by non max suppression.
+   * @param overlapThreshold A 0-D float tensor representing the threshold for deciding whether
+   *  boxes overlap too.
+   * @param scoreThreshold A 0-D float tensor representing the threshold for deciding when to remove
+   *  boxes based on score.
+   * @return a new instance of NonMaxSuppressionWithOverlaps
+   */
+  public NonMaxSuppressionWithOverlaps nonMaxSuppressionWithOverlaps(Operand<TFloat32> overlaps,
+      Operand<TFloat32> scores, Operand<TInt32> maxOutputSize, Operand<TFloat32> overlapThreshold,
+      Operand<TFloat32> scoreThreshold) {
+    return NonMaxSuppressionWithOverlaps.create(scope, overlaps, scores, maxOutputSize, overlapThreshold, scoreThreshold);
+  }
+
+  /**
+   * Resize quantized `images` to `size` using quantized bilinear interpolation.
+   *  <p>
+   *  Input images and output images must be quantized types.
+   *
+   * @param <T> data type for {@code resizedImages()} output
+   * @param images 4-D with shape `[batch, height, width, channels]`.
+   * @param size = A 1-D int32 Tensor of 2 elements: `new_height, new_width`.  The
+   *  new size for the images.
+   * @param min
+   * @param max
+   * @param options carries optional attributes values
+   * @return a new instance of QuantizedResizeBilinear
+   */
+  public <T extends TType> QuantizedResizeBilinear<T> quantizedResizeBilinear(Operand<T> images,
+      Operand<TInt32> size, Operand<TFloat32> min, Operand<TFloat32> max,
+      QuantizedResizeBilinear.Options... options) {
+    return QuantizedResizeBilinear.create(scope, images, size, min, max, options);
+  }
+
+  /**
+   * Randomly crop `image`.
+   *  <p>
+   *  `size` is a 1-D int64 tensor with 2 elements representing the crop height and
+   *  width.  The values must be non negative.
+   *  <p>
+   *  This Op picks a random location in `image` and crops a `height` by `width`
+   *  rectangle from that location.  The random location is picked so the cropped
+   *  area will fit inside the original image.
+   *
+   * @param <T> data type for {@code output()} output
+   * @param image 3-D of shape `[height, width, channels]`.
+   * @param size 1-D of length 2 containing: `crop_height`, `crop_width`..
+   * @param options carries optional attributes values
+   * @return a new instance of RandomCrop
+   */
+  public <T extends TNumber> RandomCrop<T> randomCrop(Operand<T> image, Operand<TInt64> size,
+      RandomCrop.Options... options) {
+    return RandomCrop.create(scope, image, size, options);
+  }
+
+  /**
+   * Resize `images` to `size` using area interpolation.
+   *  <p>
+   *  Input images can be of different types but output images are always float.
+   *  <p>
+   *  The range of pixel values for the output image might be slightly different
+   *  from the range for the input image because of limited numerical precision.
+   *  To guarantee an output range, for example `[0.0, 1.0]`, apply
+   *  `tf.clip_by_value` to the output.
+   *  <p>
+   *  Each output pixel is computed by first transforming the pixel's footprint into
+   *  the input tensor and then averaging the pixels that intersect the footprint. An
+   *  input pixel's contribution to the average is weighted by the fraction of its
+   *  area that intersects the footprint.  This is the same as OpenCV's INTER_AREA.
+   *
+   * @param images 4-D with shape `[batch, height, width, channels]`.
+   * @param size = A 1-D int32 Tensor of 2 elements: `new_height, new_width`.  The
+   *  new size for the images.
+   * @param options carries optional attributes values
+   * @return a new instance of ResizeArea
+   */
+  public <T extends TNumber> ResizeArea resizeArea(Operand<T> images, Operand<TInt32> size,
+      ResizeArea.Options... options) {
+    return ResizeArea.create(scope, images, size, options);
+  }
+
+  /**
+   * Resize `images` to `size` using bicubic interpolation.
+   *  <p>
+   *  Input images can be of different types but output images are always float.
+   *
+   * @param images 4-D with shape `[batch, height, width, channels]`.
+   * @param size = A 1-D int32 Tensor of 2 elements: `new_height, new_width`.  The
+   *  new size for the images.
+   * @param options carries optional attributes values
+   * @return a new instance of ResizeBicubic
+   */
+  public <T extends TNumber> ResizeBicubic resizeBicubic(Operand<T> images, Operand<TInt32> size,
+      ResizeBicubic.Options... options) {
+    return ResizeBicubic.create(scope, images, size, options);
+  }
+
+  /**
+   * Resize `images` to `size` using bilinear interpolation.
+   *  <p>
+   *  Input images can be of different types but output images are always float.
+   *
+   * @param images 4-D with shape `[batch, height, width, channels]`.
+   * @param size = A 1-D int32 Tensor of 2 elements: `new_height, new_width`.  The
+   *  new size for the images.
+   * @param options carries optional attributes values
+   * @return a new instance of ResizeBilinear
+   */
+  public <T extends TNumber> ResizeBilinear resizeBilinear(Operand<T> images, Operand<TInt32> size,
+      ResizeBilinear.Options... options) {
+    return ResizeBilinear.create(scope, images, size, options);
+  }
+
+  /**
+   * Resize `images` to `size` using nearest neighbor interpolation.
+   *
+   * @param <T> data type for {@code resizedImages()} output
+   * @param images 4-D with shape `[batch, height, width, channels]`.
+   * @param size = A 1-D int32 Tensor of 2 elements: `new_height, new_width`.  The
+   *  new size for the images.
+   * @param options carries optional attributes values
+   * @return a new instance of ResizeNearestNeighbor
+   */
+  public <T extends TNumber> ResizeNearestNeighbor<T> resizeNearestNeighbor(Operand<T> images,
+      Operand<TInt32> size, ResizeNearestNeighbor.Options... options) {
+    return ResizeNearestNeighbor.create(scope, images, size, options);
+  }
+
+  /**
+   * Converts one or more images from RGB to HSV.
+   *  <p>
+   *  Outputs a tensor of the same shape as the `images` tensor, containing the HSV
+   *  value of the pixels. The output is only well defined if the value in `images`
+   *  are in `[0,1]`.
+   *  <p>
+   *  `output[..., 0]` contains hue, `output[..., 1]` contains saturation, and
+   *  `output[..., 2]` contains value. All HSV values are in `[0,1]`. A hue of 0
+   *  corresponds to pure red, hue 1/3 is pure green, and 2/3 is pure blue.
+   *  <p>
+   *  Usage Example:
+   *  <p>
+   *  >>> blue_image = tf.stack([
+   *  ...    tf.zeros([5,5]),
+   *  ...    tf.zeros([5,5]),
+   *  ...    tf.ones([5,5])],
+   *  ...    axis=-1)
+   *  >>> blue_hsv_image = tf.image.rgb_to_hsv(blue_image)
+   *  >>> blue_hsv_image[0,0].numpy()
+   *  array([0.6666667, 1. , 1. ], dtype=float32)
+   *
+   * @param <T> data type for {@code output()} output
+   * @param images 1-D or higher rank. RGB data to convert. Last dimension must be size 3.
+   * @return a new instance of RgbToHsv
+   */
+  public <T extends TNumber> RgbToHsv<T> rgbToHsv(Operand<T> images) {
+    return RgbToHsv.create(scope, images);
+  }
+
+  /**
+   * Generate a single randomly distorted bounding box for an image.
+   *  <p>
+   *  Bounding box annotations are often supplied in addition to ground-truth labels
+   *  in image recognition or object localization tasks. A common technique for
+   *  training such a system is to randomly distort an image while preserving
+   *  its content, i.e. <i>data augmentation</i>. This Op outputs a randomly distorted
+   *  localization of an object, i.e. bounding box, given an `image_size`,
+   *  `bounding_boxes` and a series of constraints.
+   *  <p>
+   *  The output of this Op is a single bounding box that may be used to crop the
+   *  original image. The output is returned as 3 tensors: `begin`, `size` and
+   *  `bboxes`. The first 2 tensors can be fed directly into `tf.slice` to crop the
+   *  image. The latter may be supplied to `tf.image.draw_bounding_boxes` to visualize
+   *  what the bounding box looks like.
+   *  <p>
+   *  Bounding boxes are supplied and returned as `[y_min, x_min, y_max, x_max]`. The
+   *  bounding box coordinates are floats in `[0.0, 1.0]` relative to the width and
+   *  height of the underlying image.
+   *  <p>
+   *  For example,
+   *  <pre>{@code
+   *      # Generate a single distorted bounding box.
+   *      begin, size, bbox_for_draw = tf.image.sample_distorted_bounding_box(
+   *          tf.shape(image),
+   *          bounding_boxes=bounding_boxes)
+   *
+   *      # Draw the bounding box in an image summary.
+   *      image_with_box = tf.image.draw_bounding_boxes(tf.expand_dims(image, 0),
+   *                                                    bbox_for_draw)
+   *      tf.summary.image('images_with_box', image_with_box)
+   *
+   *      # Employ the bounding box to distort the image.
+   *      distorted_image = tf.slice(image, begin, size)
+   *  }</pre>
+   *  Note that if no bounding box information is available, setting
+   *  `use_image_if_no_bounding_boxes = true` will assume there is a single implicit
+   *  bounding box covering the whole image. If `use_image_if_no_bounding_boxes` is
+   *  false and no bounding boxes are supplied, an error is raised.
+   *
+   * @param <T> data type for {@code begin()} output
+   * @param imageSize 1-D, containing `[height, width, channels]`.
+   * @param boundingBoxes 3-D with shape `[batch, N, 4]` describing the N bounding boxes
+   *  associated with the image.
+   * @param minObjectCovered The cropped area of the image must contain at least this
+   *  fraction of any bounding box supplied. The value of this parameter should be
+   *  non-negative. In the case of 0, the cropped area does not need to overlap
+   *  any of the bounding boxes supplied.
+   * @param options carries optional attributes values
+   * @return a new instance of SampleDistortedBoundingBox
+   */
+  public <T extends TNumber> SampleDistortedBoundingBox<T> sampleDistortedBoundingBox(
+      Operand<T> imageSize, Operand<TFloat32> boundingBoxes, Operand<TFloat32> minObjectCovered,
+      SampleDistortedBoundingBox.Options... options) {
+    return SampleDistortedBoundingBox.create(scope, imageSize, boundingBoxes, minObjectCovered, options);
+  }
+
+  /**
+   *
+   * @param images
+   * @param size
+   * @param scale
+   * @param translation
+   * @param options carries optional attributes values
+   * @return a new instance of ScaleAndTranslate
+   */
+  public <T extends TNumber> ScaleAndTranslate scaleAndTranslate(Operand<T> images,
+      Operand<TInt32> size, Operand<TFloat32> scale, Operand<TFloat32> translation,
+      ScaleAndTranslate.Options... options) {
+    return ScaleAndTranslate.create(scope, images, size, scale, translation, options);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/IoOps.java b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/IoOps.java
new file mode 100644
index 00000000000..78aa5897a97
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/IoOps.java
@@ -0,0 +1,1037 @@
+// Copyright 2020 The TensorFlow Authors. All Rights Reserved.
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+// ==============================================================================
+//
+// This class has been generated, DO NOT EDIT!
+//
+package org.tensorflow.op;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.io.DecodeBase64;
+import org.tensorflow.op.io.DecodeCompressed;
+import org.tensorflow.op.io.DecodeCsv;
+import org.tensorflow.op.io.DecodeJsonExample;
+import org.tensorflow.op.io.DecodePaddedRaw;
+import org.tensorflow.op.io.DecodeRaw;
+import org.tensorflow.op.io.DeserializeManySparse;
+import org.tensorflow.op.io.EncodeBase64;
+import org.tensorflow.op.io.FifoQueue;
+import org.tensorflow.op.io.FixedLengthRecordReader;
+import org.tensorflow.op.io.IdentityReader;
+import org.tensorflow.op.io.LmdbReader;
+import org.tensorflow.op.io.MatchingFiles;
+import org.tensorflow.op.io.PaddingFifoQueue;
+import org.tensorflow.op.io.ParseExample;
+import org.tensorflow.op.io.ParseSequenceExample;
+import org.tensorflow.op.io.ParseSingleExample;
+import org.tensorflow.op.io.ParseSingleSequenceExample;
+import org.tensorflow.op.io.ParseTensor;
+import org.tensorflow.op.io.PriorityQueue;
+import org.tensorflow.op.io.QueueClose;
+import org.tensorflow.op.io.QueueDequeue;
+import org.tensorflow.op.io.QueueDequeueMany;
+import org.tensorflow.op.io.QueueDequeueUpTo;
+import org.tensorflow.op.io.QueueEnqueue;
+import org.tensorflow.op.io.QueueEnqueueMany;
+import org.tensorflow.op.io.QueueIsClosed;
+import org.tensorflow.op.io.QueueSize;
+import org.tensorflow.op.io.RandomShuffleQueue;
+import org.tensorflow.op.io.ReadFile;
+import org.tensorflow.op.io.ReaderNumRecordsProduced;
+import org.tensorflow.op.io.ReaderNumWorkUnitsCompleted;
+import org.tensorflow.op.io.ReaderRead;
+import org.tensorflow.op.io.ReaderReadUpTo;
+import org.tensorflow.op.io.ReaderReset;
+import org.tensorflow.op.io.ReaderRestoreState;
+import org.tensorflow.op.io.ReaderSerializeState;
+import org.tensorflow.op.io.SerializeManySparse;
+import org.tensorflow.op.io.SerializeSparse;
+import org.tensorflow.op.io.SerializeTensor;
+import org.tensorflow.op.io.ShardedFilename;
+import org.tensorflow.op.io.ShardedFilespec;
+import org.tensorflow.op.io.TextLineReader;
+import org.tensorflow.op.io.TfRecordReader;
+import org.tensorflow.op.io.WholeFileReader;
+import org.tensorflow.op.io.WriteFile;
+import org.tensorflow.types.TBool;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.TString;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * An API for building {@code io} operations as {@link Op Op}s
+ *
+ * @see {@link Ops}
+ */
+public final class IoOps {
+  private final Scope scope;
+
+  IoOps(Scope scope) {
+    this.scope = scope;
+  }
+
+  /**
+   * Decode web-safe base64-encoded strings.
+   *  <p>
+   *  Input may or may not have padding at the end. See EncodeBase64 for padding.
+   *  Web-safe means that input must use - and _ instead of + and /.
+   *
+   * @param input Base64 strings to decode.
+   * @return a new instance of DecodeBase64
+   */
+  public DecodeBase64 decodeBase64(Operand<TString> input) {
+    return DecodeBase64.create(scope, input);
+  }
+
+  /**
+   * Decompress strings.
+   *  <p>
+   *  This op decompresses each element of the `bytes` input `Tensor`, which
+   *  is assumed to be compressed using the given `compression_type`.
+   *  <p>
+   *  The `output` is a string `Tensor` of the same shape as `bytes`,
+   *  each element containing the decompressed data from the corresponding
+   *  element in `bytes`.
+   *
+   * @param bytes A Tensor of string which is compressed.
+   * @param options carries optional attributes values
+   * @return a new instance of DecodeCompressed
+   */
+  public DecodeCompressed decodeCompressed(Operand<TString> bytes,
+      DecodeCompressed.Options... options) {
+    return DecodeCompressed.create(scope, bytes, options);
+  }
+
+  /**
+   * Convert CSV records to tensors. Each column maps to one tensor.
+   *  <p>
+   *  RFC 4180 format is expected for the CSV records.
+   *  (https://tools.ietf.org/html/rfc4180)
+   *  Note that we allow leading and trailing spaces with int or float field.
+   *
+   * @param records Each string is a record/row in the csv and all records should have
+   *  the same format.
+   * @param recordDefaults One tensor per column of the input record, with either a
+   *  scalar default value for that column or an empty vector if the column is
+   *  required.
+   * @param options carries optional attributes values
+   * @return a new instance of DecodeCsv
+   */
+  public DecodeCsv decodeCsv(Operand<TString> records, Iterable<Operand<?>> recordDefaults,
+      DecodeCsv.Options... options) {
+    return DecodeCsv.create(scope, records, recordDefaults, options);
+  }
+
+  /**
+   * Convert JSON-encoded Example records to binary protocol buffer strings.
+   *  <p>
+   *  This op translates a tensor containing Example records, encoded using
+   *  the [standard JSON
+   *  mapping](https://developers.google.com/protocol-buffers/docs/proto3#json),
+   *  into a tensor containing the same records encoded as binary protocol
+   *  buffers. The resulting tensor can then be fed to any of the other
+   *  Example-parsing ops.
+   *
+   * @param jsonExamples Each string is a JSON object serialized according to the JSON
+   *  mapping of the Example proto.
+   * @return a new instance of DecodeJsonExample
+   */
+  public DecodeJsonExample decodeJsonExample(Operand<TString> jsonExamples) {
+    return DecodeJsonExample.create(scope, jsonExamples);
+  }
+
+  /**
+   * Reinterpret the bytes of a string as a vector of numbers.
+   *
+   * @param <T> data type for {@code output()} output
+   * @param inputBytes Tensor of string to be decoded.
+   * @param fixedLength Length in bytes for each element of the decoded output. Must be a multiple
+   *  of the size of the output type.
+   * @param outType
+   * @param options carries optional attributes values
+   * @return a new instance of DecodePaddedRaw
+   */
+  public <T extends TNumber> DecodePaddedRaw<T> decodePaddedRaw(Operand<TString> inputBytes,
+      Operand<TInt32> fixedLength, Class<T> outType, DecodePaddedRaw.Options... options) {
+    return DecodePaddedRaw.create(scope, inputBytes, fixedLength, outType, options);
+  }
+
+  /**
+   * Reinterpret the bytes of a string as a vector of numbers.
+   *
+   * @param <T> data type for {@code output()} output
+   * @param bytes All the elements must have the same length.
+   * @param outType
+   * @param options carries optional attributes values
+   * @return a new instance of DecodeRaw
+   */
+  public <T extends TType> DecodeRaw<T> decodeRaw(Operand<TString> bytes, Class<T> outType,
+      DecodeRaw.Options... options) {
+    return DecodeRaw.create(scope, bytes, outType, options);
+  }
+
+  /**
+   * Deserialize and concatenate `SparseTensors` from a serialized minibatch.
+   *  <p>
+   *  The input `serialized_sparse` must be a string matrix of shape `[N x 3]` where
+   *  `N` is the minibatch size and the rows correspond to packed outputs of
+   *  `SerializeSparse`.  The ranks of the original `SparseTensor` objects
+   *  must all match.  When the final `SparseTensor` is created, it has rank one
+   *  higher than the ranks of the incoming `SparseTensor` objects
+   *  (they have been concatenated along a new row dimension).
+   *  <p>
+   *  The output `SparseTensor` object's shape values for all dimensions but the
+   *  first are the max across the input `SparseTensor` objects' shape values
+   *  for the corresponding dimensions.  Its first shape value is `N`, the minibatch
+   *  size.
+   *  <p>
+   *  The input `SparseTensor` objects' indices are assumed ordered in
+   *  standard lexicographic order.  If this is not the case, after this
+   *  step run `SparseReorder` to restore index ordering.
+   *  <p>
+   *  For example, if the serialized input is a `[2 x 3]` matrix representing two
+   *  original `SparseTensor` objects:
+   *  <p>
+   *      index = [ 0]
+   *              [10]
+   *              [20]
+   *      values = [1, 2, 3]
+   *      shape = [50]
+   *  <p>
+   *  and
+   *  <p>
+   *      index = [ 2]
+   *              [10]
+   *      values = [4, 5]
+   *      shape = [30]
+   *  <p>
+   *  then the final deserialized `SparseTensor` will be:
+   *  <p>
+   *      index = [0  0]
+   *              [0 10]
+   *              [0 20]
+   *              [1  2]
+   *              [1 10]
+   *      values = [1, 2, 3, 4, 5]
+   *      shape = [2 50]
+   *
+   * @param <T> data type for {@code sparseValues()} output
+   * @param serializedSparse 2-D, The `N` serialized `SparseTensor` objects.
+   *  Must have 3 columns.
+   * @param dtype The `dtype` of the serialized `SparseTensor` objects.
+   * @return a new instance of DeserializeManySparse
+   */
+  public <T extends TType> DeserializeManySparse<T> deserializeManySparse(
+      Operand<TString> serializedSparse, Class<T> dtype) {
+    return DeserializeManySparse.create(scope, serializedSparse, dtype);
+  }
+
+  /**
+   * Encode strings into web-safe base64 format.
+   *  <p>
+   *  Refer to the following article for more information on base64 format:
+   *  en.wikipedia.org/wiki/Base64. Base64 strings may have padding with '=' at the
+   *  end so that the encoded has length multiple of 4. See Padding section of the
+   *  link above.
+   *  <p>
+   *  Web-safe means that the encoder uses - and _ instead of + and /.
+   *
+   * @param input Strings to be encoded.
+   * @param options carries optional attributes values
+   * @return a new instance of EncodeBase64
+   */
+  public EncodeBase64 encodeBase64(Operand<TString> input, EncodeBase64.Options... options) {
+    return EncodeBase64.create(scope, input, options);
+  }
+
+  /**
+   * A queue that produces elements in first-in first-out order.
+   *
+   * @param componentTypes The type of each component in a value.
+   * @param options carries optional attributes values
+   * @return a new instance of FifoQueue
+   */
+  public FifoQueue fifoQueue(List<Class<? extends TType>> componentTypes,
+      FifoQueue.Options... options) {
+    return FifoQueue.create(scope, componentTypes, options);
+  }
+
+  /**
+   * A Reader that outputs fixed-length records from a file.
+   *
+   * @param recordBytes Number of bytes in the record.
+   * @param options carries optional attributes values
+   * @return a new instance of FixedLengthRecordReader
+   */
+  public FixedLengthRecordReader fixedLengthRecordReader(Long recordBytes,
+      FixedLengthRecordReader.Options... options) {
+    return FixedLengthRecordReader.create(scope, recordBytes, options);
+  }
+
+  /**
+   * A Reader that outputs the queued work as both the key and value.
+   *  <p>
+   *  To use, enqueue strings in a Queue.  ReaderRead will take the front
+   *  work string and output (work, work).
+   *
+   * @param options carries optional attributes values
+   * @return a new instance of IdentityReader
+   */
+  public IdentityReader identityReader(IdentityReader.Options... options) {
+    return IdentityReader.create(scope, options);
+  }
+
+  /**
+   * A Reader that outputs the records from a LMDB file.
+   *
+   * @param options carries optional attributes values
+   * @return a new instance of LmdbReader
+   */
+  public LmdbReader lmdbReader(LmdbReader.Options... options) {
+    return LmdbReader.create(scope, options);
+  }
+
+  /**
+   * Returns the set of files matching one or more glob patterns.
+   *  <p>
+   *  Note that this routine only supports wildcard characters in the
+   *  basename portion of the pattern, not in the directory portion.
+   *  Note also that the order of filenames returned is deterministic.
+   *
+   * @param pattern Shell wildcard pattern(s). Scalar or vector of type string.
+   * @return a new instance of MatchingFiles
+   */
+  public MatchingFiles matchingFiles(Operand<TString> pattern) {
+    return MatchingFiles.create(scope, pattern);
+  }
+
+  /**
+   * A queue that produces elements in first-in first-out order.
+   *  <p>
+   *  Variable-size shapes are allowed by setting the corresponding shape dimensions
+   *  to 0 in the shape attr.  In this case DequeueMany will pad up to the maximum
+   *  size of any given element in the minibatch.  See below for details.
+   *
+   * @param componentTypes The type of each component in a value.
+   * @param options carries optional attributes values
+   * @return a new instance of PaddingFifoQueue
+   */
+  public PaddingFifoQueue paddingFifoQueue(List<Class<? extends TType>> componentTypes,
+      PaddingFifoQueue.Options... options) {
+    return PaddingFifoQueue.create(scope, componentTypes, options);
+  }
+
+  /**
+   * Transforms a vector of tf.Example protos (as strings) into typed tensors.
+   *
+   * @param serialized A scalar or vector containing binary serialized Example protos.
+   * @param names A tensor containing the names of the serialized protos.
+   *  Corresponds 1:1 with the `serialized` tensor.
+   *  May contain, for example, table key (descriptive) names for the
+   *  corresponding serialized protos.  These are purely useful for debugging
+   *  purposes, and the presence of values here has no effect on the output.
+   *  May also be an empty vector if no names are available.
+   *  If non-empty, this tensor must have the same shape as "serialized".
+   * @param sparseKeys Vector of strings.
+   *  The keys expected in the Examples' features associated with sparse values.
+   * @param denseKeys Vector of strings.
+   *  The keys expected in the Examples' features associated with dense values.
+   * @param raggedKeys Vector of strings.
+   *  The keys expected in the Examples' features associated with ragged values.
+   * @param denseDefaults A list of Tensors (some may be empty).  Corresponds 1:1 with `dense_keys`.
+   *  dense_defaults[j] provides default values
+   *  when the example's feature_map lacks dense_key[j].  If an empty Tensor is
+   *  provided for dense_defaults[j], then the Feature dense_keys[j] is required.
+   *  The input type is inferred from dense_defaults[j], even when it's empty.
+   *  If dense_defaults[j] is not empty, and dense_shapes[j] is fully defined,
+   *  then the shape of dense_defaults[j] must match that of dense_shapes[j].
+   *  If dense_shapes[j] has an undefined major dimension (variable strides dense
+   *  feature), dense_defaults[j] must contain a single element:
+   *  the padding element.
+   * @param numSparse The number of sparse keys.
+   * @param sparseTypes A list of `num_sparse` types; the data types of data in each Feature
+   *  given in sparse_keys.
+   *  Currently the ParseExample supports DT_FLOAT (FloatList),
+   *  DT_INT64 (Int64List), and DT_STRING (BytesList).
+   * @param raggedValueTypes A list of `num_ragged` types; the data types of data in each Feature
+   *  given in ragged_keys (where `num_ragged = sparse_keys.size()`).
+   *  Currently the ParseExample supports DT_FLOAT (FloatList),
+   *  DT_INT64 (Int64List), and DT_STRING (BytesList).
+   * @param raggedSplitTypes A list of `num_ragged` types; the data types of row_splits in each Feature
+   *  given in ragged_keys (where `num_ragged = sparse_keys.size()`).
+   *  May be DT_INT32 or DT_INT64.
+   * @param denseShapes A list of `num_dense` shapes; the shapes of data in each Feature
+   *  given in dense_keys (where `num_dense = dense_keys.size()`).
+   *  The number of elements in the Feature corresponding to dense_key[j]
+   *  must always equal dense_shapes[j].NumEntries().
+   *  If dense_shapes[j] == (D0, D1, ..., DN) then the shape of output
+   *  Tensor dense_values[j] will be (|serialized|, D0, D1, ..., DN):
+   *  The dense outputs are just the inputs row-stacked by batch.
+   *  This works for dense_shapes[j] = (-1, D1, ..., DN).  In this case
+   *  the shape of the output Tensor dense_values[j] will be
+   *  (|serialized|, M, D1, .., DN), where M is the maximum number of blocks
+   *  of elements of length D1 * .... * DN, across all minibatch entries
+   *  in the input.  Any minibatch entry with less than M blocks of elements of
+   *  length D1 * ... * DN will be padded with the corresponding default_value
+   *  scalar element along the second dimension.
+   * @return a new instance of ParseExample
+   */
+  public ParseExample parseExample(Operand<TString> serialized, Operand<TString> names,
+      Operand<TString> sparseKeys, Operand<TString> denseKeys, Operand<TString> raggedKeys,
+      Iterable<Operand<?>> denseDefaults, Long numSparse, List<Class<? extends TType>> sparseTypes,
+      List<Class<? extends TType>> raggedValueTypes,
+      List<Class<? extends TNumber>> raggedSplitTypes, List<Shape> denseShapes) {
+    return ParseExample.create(scope, serialized, names, sparseKeys, denseKeys, raggedKeys, denseDefaults, numSparse, sparseTypes, raggedValueTypes, raggedSplitTypes, denseShapes);
+  }
+
+  /**
+   * Transforms a vector of tf.io.SequenceExample protos (as strings) into
+   *  typed tensors.
+   *
+   * @param serialized A scalar or vector containing binary serialized SequenceExample protos.
+   * @param debugName A scalar or vector containing the names of the serialized protos.
+   *  May contain, for example, table key (descriptive) name for the
+   *  corresponding serialized proto.  This is purely useful for debugging
+   *  purposes, and the presence of values here has no effect on the output.
+   *  May also be an empty vector if no name is available.
+   * @param contextSparseKeys The keys expected in the Examples' features associated with context_sparse
+   *  values.
+   * @param contextDenseKeys The keys expected in the SequenceExamples' context features associated with
+   *  dense values.
+   * @param contextRaggedKeys The keys expected in the Examples' features associated with context_ragged
+   *  values.
+   * @param featureListSparseKeys The keys expected in the FeatureLists associated with sparse values.
+   * @param featureListDenseKeys The keys expected in the SequenceExamples' feature_lists associated
+   *  with lists of dense values.
+   * @param featureListRaggedKeys The keys expected in the FeatureLists associated with ragged values.
+   * @param featureListDenseMissingAssumedEmpty A vector corresponding 1:1 with feature_list_dense_keys, indicating which
+   *  features may be missing from the SequenceExamples.  If the associated
+   *  FeatureList is missing, it is treated as empty.
+   * @param contextDenseDefaults A list of Ncontext_dense Tensors (some may be empty).
+   *  context_dense_defaults[j] provides default values
+   *  when the SequenceExample's context map lacks context_dense_key[j].
+   *  If an empty Tensor is provided for context_dense_defaults[j],
+   *  then the Feature context_dense_keys[j] is required.
+   *  The input type is inferred from context_dense_defaults[j], even when it's
+   *  empty.  If context_dense_defaults[j] is not empty, its shape must match
+   *  context_dense_shapes[j].
+   * @param contextSparseTypes A list of Ncontext_sparse types; the data types of data in
+   *  each context Feature given in context_sparse_keys.
+   *  Currently the ParseSingleSequenceExample supports DT_FLOAT (FloatList),
+   *  DT_INT64 (Int64List), and DT_STRING (BytesList).
+   * @param contextRaggedValueTypes RaggedTensor.value dtypes for the ragged context features.
+   * @param contextRaggedSplitTypes RaggedTensor.row_split dtypes for the ragged context features.
+   * @param featureListDenseTypes
+   * @param featureListSparseTypes A list of Nfeature_list_sparse types; the data types
+   *  of data in each FeatureList given in feature_list_sparse_keys.
+   *  Currently the ParseSingleSequenceExample supports DT_FLOAT (FloatList),
+   *  DT_INT64 (Int64List), and DT_STRING (BytesList).
+   * @param featureListRaggedValueTypes RaggedTensor.value dtypes for the ragged FeatureList features.
+   * @param featureListRaggedSplitTypes RaggedTensor.row_split dtypes for the ragged FeatureList features.
+   * @param options carries optional attributes values
+   * @return a new instance of ParseSequenceExample
+   */
+  public ParseSequenceExample parseSequenceExample(Operand<TString> serialized,
+      Operand<TString> debugName, Operand<TString> contextSparseKeys,
+      Operand<TString> contextDenseKeys, Operand<TString> contextRaggedKeys,
+      Operand<TString> featureListSparseKeys, Operand<TString> featureListDenseKeys,
+      Operand<TString> featureListRaggedKeys, Operand<TBool> featureListDenseMissingAssumedEmpty,
+      Iterable<Operand<?>> contextDenseDefaults, List<Class<? extends TType>> contextSparseTypes,
+      List<Class<? extends TType>> contextRaggedValueTypes,
+      List<Class<? extends TNumber>> contextRaggedSplitTypes,
+      List<Class<? extends TType>> featureListDenseTypes,
+      List<Class<? extends TType>> featureListSparseTypes,
+      List<Class<? extends TType>> featureListRaggedValueTypes,
+      List<Class<? extends TNumber>> featureListRaggedSplitTypes,
+      ParseSequenceExample.Options... options) {
+    return ParseSequenceExample.create(scope, serialized, debugName, contextSparseKeys, contextDenseKeys, contextRaggedKeys, featureListSparseKeys, featureListDenseKeys, featureListRaggedKeys, featureListDenseMissingAssumedEmpty, contextDenseDefaults, contextSparseTypes, contextRaggedValueTypes, contextRaggedSplitTypes, featureListDenseTypes, featureListSparseTypes, featureListRaggedValueTypes, featureListRaggedSplitTypes, options);
+  }
+
+  /**
+   * Transforms a tf.Example proto (as a string) into typed tensors.
+   *
+   * @param serialized A vector containing a batch of binary serialized Example protos.
+   * @param denseDefaults A list of Tensors (some may be empty), whose length matches
+   *  the length of `dense_keys`. dense_defaults[j] provides default values
+   *  when the example's feature_map lacks dense_key[j].  If an empty Tensor is
+   *  provided for dense_defaults[j], then the Feature dense_keys[j] is required.
+   *  The input type is inferred from dense_defaults[j], even when it's empty.
+   *  If dense_defaults[j] is not empty, and dense_shapes[j] is fully defined,
+   *  then the shape of dense_defaults[j] must match that of dense_shapes[j].
+   *  If dense_shapes[j] has an undefined major dimension (variable strides dense
+   *  feature), dense_defaults[j] must contain a single element:
+   *  the padding element.
+   * @param numSparse The number of sparse features to be parsed from the example. This
+   *  must match the lengths of `sparse_keys` and `sparse_types`.
+   * @param sparseKeys A list of `num_sparse` strings.
+   *  The keys expected in the Examples' features associated with sparse values.
+   * @param denseKeys The keys expected in the Examples' features associated with dense
+   *  values.
+   * @param sparseTypes A list of `num_sparse` types; the data types of data in each
+   *  Feature given in sparse_keys.
+   *  Currently the ParseSingleExample op supports DT_FLOAT (FloatList),
+   *  DT_INT64 (Int64List), and DT_STRING (BytesList).
+   * @param denseShapes The shapes of data in each Feature given in dense_keys.
+   *  The length of this list must match the length of `dense_keys`.  The
+   *  number of elements in the Feature corresponding to dense_key[j] must
+   *  always equal dense_shapes[j].NumEntries().  If dense_shapes[j] ==
+   *  (D0, D1, ..., DN) then the shape of output Tensor dense_values[j]
+   *  will be (D0, D1, ..., DN): In the case dense_shapes[j] = (-1, D1,
+   *  ..., DN), the shape of the output Tensor dense_values[j] will be (M,
+   *  D1, .., DN), where M is the number of blocks of elements of length
+   *  D1 * .... * DN, in the input.
+   * @return a new instance of ParseSingleExample
+   */
+  public ParseSingleExample parseSingleExample(Operand<TString> serialized,
+      Iterable<Operand<?>> denseDefaults, Long numSparse, List<String> sparseKeys,
+      List<String> denseKeys, List<Class<? extends TType>> sparseTypes, List<Shape> denseShapes) {
+    return ParseSingleExample.create(scope, serialized, denseDefaults, numSparse, sparseKeys, denseKeys, sparseTypes, denseShapes);
+  }
+
+  /**
+   * Transforms a scalar brain.SequenceExample proto (as strings) into typed tensors.
+   *
+   * @param serialized A scalar containing a binary serialized SequenceExample proto.
+   * @param featureListDenseMissingAssumedEmpty A vector listing the
+   *  FeatureList keys which may be missing from the SequenceExample.  If the
+   *  associated FeatureList is missing, it is treated as empty.  By default,
+   *  any FeatureList not listed in this vector must exist in the SequenceExample.
+   * @param contextSparseKeys A list of Ncontext_sparse string Tensors (scalars).
+   *  The keys expected in the Examples' features associated with context_sparse
+   *  values.
+   * @param contextDenseKeys A list of Ncontext_dense string Tensors (scalars).
+   *  The keys expected in the SequenceExamples' context features associated with
+   *  dense values.
+   * @param featureListSparseKeys A list of Nfeature_list_sparse string Tensors
+   *  (scalars).  The keys expected in the FeatureLists associated with sparse
+   *  values.
+   * @param featureListDenseKeys A list of Nfeature_list_dense string Tensors (scalars).
+   *  The keys expected in the SequenceExamples' feature_lists associated
+   *  with lists of dense values.
+   * @param contextDenseDefaults A list of Ncontext_dense Tensors (some may be empty).
+   *  context_dense_defaults[j] provides default values
+   *  when the SequenceExample's context map lacks context_dense_key[j].
+   *  If an empty Tensor is provided for context_dense_defaults[j],
+   *  then the Feature context_dense_keys[j] is required.
+   *  The input type is inferred from context_dense_defaults[j], even when it's
+   *  empty.  If context_dense_defaults[j] is not empty, its shape must match
+   *  context_dense_shapes[j].
+   * @param debugName A scalar containing the name of the serialized proto.
+   *  May contain, for example, table key (descriptive) name for the
+   *  corresponding serialized proto.  This is purely useful for debugging
+   *  purposes, and the presence of values here has no effect on the output.
+   *  May also be an empty scalar if no name is available.
+   * @param contextSparseTypes A list of Ncontext_sparse types; the data types of data in
+   *  each context Feature given in context_sparse_keys.
+   *  Currently the ParseSingleSequenceExample supports DT_FLOAT (FloatList),
+   *  DT_INT64 (Int64List), and DT_STRING (BytesList).
+   * @param featureListDenseTypes
+   * @param featureListSparseTypes A list of Nfeature_list_sparse types; the data types
+   *  of data in each FeatureList given in feature_list_sparse_keys.
+   *  Currently the ParseSingleSequenceExample supports DT_FLOAT (FloatList),
+   *  DT_INT64 (Int64List), and DT_STRING (BytesList).
+   * @param options carries optional attributes values
+   * @return a new instance of ParseSingleSequenceExample
+   */
+  public ParseSingleSequenceExample parseSingleSequenceExample(Operand<TString> serialized,
+      Operand<TString> featureListDenseMissingAssumedEmpty,
+      Iterable<Operand<TString>> contextSparseKeys, Iterable<Operand<TString>> contextDenseKeys,
+      Iterable<Operand<TString>> featureListSparseKeys,
+      Iterable<Operand<TString>> featureListDenseKeys, Iterable<Operand<?>> contextDenseDefaults,
+      Operand<TString> debugName, List<Class<? extends TType>> contextSparseTypes,
+      List<Class<? extends TType>> featureListDenseTypes,
+      List<Class<? extends TType>> featureListSparseTypes,
+      ParseSingleSequenceExample.Options... options) {
+    return ParseSingleSequenceExample.create(scope, serialized, featureListDenseMissingAssumedEmpty, contextSparseKeys, contextDenseKeys, featureListSparseKeys, featureListDenseKeys, contextDenseDefaults, debugName, contextSparseTypes, featureListDenseTypes, featureListSparseTypes, options);
+  }
+
+  /**
+   * Transforms a serialized tensorflow.TensorProto proto into a Tensor.
+   *
+   * @param <T> data type for {@code output()} output
+   * @param serialized A scalar string containing a serialized TensorProto proto.
+   * @param outType The type of the serialized tensor.  The provided type must match the
+   *  type of the serialized tensor and no implicit conversion will take place.
+   * @return a new instance of ParseTensor
+   */
+  public <T extends TType> ParseTensor<T> parseTensor(Operand<TString> serialized,
+      Class<T> outType) {
+    return ParseTensor.create(scope, serialized, outType);
+  }
+
+  /**
+   * A queue that produces elements sorted by the first component value.
+   *  <p>
+   *  Note that the PriorityQueue requires the first component of any element
+   *  to be a scalar int64, in addition to the other elements declared by
+   *  component_types.  Therefore calls to Enqueue and EnqueueMany (resp. Dequeue
+   *  and DequeueMany) on a PriorityQueue will all require (resp. output) one extra
+   *  entry in their input (resp. output) lists.
+   *
+   * @param componentTypes The type of each component in a value.
+   * @param shapes The shape of each component in a value. The length of this attr must
+   *  be either 0 or the same as the length of component_types. If the length of
+   *  this attr is 0, the shapes of queue elements are not constrained, and
+   *  only one element may be dequeued at a time.
+   * @param options carries optional attributes values
+   * @return a new instance of PriorityQueue
+   */
+  public PriorityQueue priorityQueue(List<Class<? extends TType>> componentTypes,
+      List<Shape> shapes, PriorityQueue.Options... options) {
+    return PriorityQueue.create(scope, componentTypes, shapes, options);
+  }
+
+  /**
+   * Closes the given queue.
+   *  <p>
+   *  This operation signals that no more elements will be enqueued in the
+   *  given queue. Subsequent Enqueue(Many) operations will fail.
+   *  Subsequent Dequeue(Many) operations will continue to succeed if
+   *  sufficient elements remain in the queue. Subsequent Dequeue(Many)
+   *  operations that would block will fail immediately.
+   *
+   * @param handle The handle to a queue.
+   * @param options carries optional attributes values
+   * @return a new instance of QueueClose
+   */
+  public QueueClose queueClose(Operand<?> handle, QueueClose.Options... options) {
+    return QueueClose.create(scope, handle, options);
+  }
+
+  /**
+   * Dequeues a tuple of one or more tensors from the given queue.
+   *  <p>
+   *  This operation has k outputs, where k is the number of components
+   *  in the tuples stored in the given queue, and output i is the ith
+   *  component of the dequeued tuple.
+   *  <p>
+   *  N.B. If the queue is empty, this operation will block until an element
+   *  has been dequeued (or 'timeout_ms' elapses, if specified).
+   *
+   * @param handle The handle to a queue.
+   * @param componentTypes The type of each component in a tuple.
+   * @param options carries optional attributes values
+   * @return a new instance of QueueDequeue
+   */
+  public QueueDequeue queueDequeue(Operand<?> handle, List<Class<? extends TType>> componentTypes,
+      QueueDequeue.Options... options) {
+    return QueueDequeue.create(scope, handle, componentTypes, options);
+  }
+
+  /**
+   * Dequeues `n` tuples of one or more tensors from the given queue.
+   *  <p>
+   *  If the queue is closed and there are fewer than `n` elements, then an
+   *  OutOfRange error is returned.
+   *  <p>
+   *  This operation concatenates queue-element component tensors along the
+   *  0th dimension to make a single component tensor.  All of the components
+   *  in the dequeued tuple will have size `n` in the 0th dimension.
+   *  <p>
+   *  This operation has `k` outputs, where `k` is the number of components in
+   *  the tuples stored in the given queue, and output `i` is the ith
+   *  component of the dequeued tuple.
+   *  <p>
+   *  N.B. If the queue is empty, this operation will block until `n` elements
+   *  have been dequeued (or 'timeout_ms' elapses, if specified).
+   *
+   * @param handle The handle to a queue.
+   * @param n The number of tuples to dequeue.
+   * @param componentTypes The type of each component in a tuple.
+   * @param options carries optional attributes values
+   * @return a new instance of QueueDequeueMany
+   */
+  public QueueDequeueMany queueDequeueMany(Operand<?> handle, Operand<TInt32> n,
+      List<Class<? extends TType>> componentTypes, QueueDequeueMany.Options... options) {
+    return QueueDequeueMany.create(scope, handle, n, componentTypes, options);
+  }
+
+  /**
+   * Dequeues `n` tuples of one or more tensors from the given queue.
+   *  <p>
+   *  This operation is not supported by all queues.  If a queue does not support
+   *  DequeueUpTo, then an Unimplemented error is returned.
+   *  <p>
+   *  If the queue is closed and there are more than 0 but less than `n`
+   *  elements remaining, then instead of returning an OutOfRange error like
+   *  QueueDequeueMany, less than `n` elements are returned immediately.  If
+   *  the queue is closed and there are 0 elements left in the queue, then
+   *  an OutOfRange error is returned just like in QueueDequeueMany.
+   *  Otherwise the behavior is identical to QueueDequeueMany:
+   *  <p>
+   *  This operation concatenates queue-element component tensors along the
+   *  0th dimension to make a single component tensor.  All of the components
+   *  in the dequeued tuple will have size n in the 0th dimension.
+   *  <p>
+   *  This operation has `k` outputs, where `k` is the number of components in
+   *  the tuples stored in the given queue, and output `i` is the ith
+   *  component of the dequeued tuple.
+   *
+   * @param handle The handle to a queue.
+   * @param n The number of tuples to dequeue.
+   * @param componentTypes The type of each component in a tuple.
+   * @param options carries optional attributes values
+   * @return a new instance of QueueDequeueUpTo
+   */
+  public QueueDequeueUpTo queueDequeueUpTo(Operand<?> handle, Operand<TInt32> n,
+      List<Class<? extends TType>> componentTypes, QueueDequeueUpTo.Options... options) {
+    return QueueDequeueUpTo.create(scope, handle, n, componentTypes, options);
+  }
+
+  /**
+   * Enqueues a tuple of one or more tensors in the given queue.
+   *  <p>
+   *  The components input has k elements, which correspond to the components of
+   *  tuples stored in the given queue.
+   *  <p>
+   *  N.B. If the queue is full, this operation will block until the given
+   *  element has been enqueued (or 'timeout_ms' elapses, if specified).
+   *
+   * @param handle The handle to a queue.
+   * @param components One or more tensors from which the enqueued tensors should be taken.
+   * @param options carries optional attributes values
+   * @return a new instance of QueueEnqueue
+   */
+  public QueueEnqueue queueEnqueue(Operand<?> handle, Iterable<Operand<?>> components,
+      QueueEnqueue.Options... options) {
+    return QueueEnqueue.create(scope, handle, components, options);
+  }
+
+  /**
+   * Enqueues zero or more tuples of one or more tensors in the given queue.
+   *  <p>
+   *  This operation slices each component tensor along the 0th dimension to
+   *  make multiple queue elements. All of the tuple components must have the
+   *  same size in the 0th dimension.
+   *  <p>
+   *  The components input has k elements, which correspond to the components of
+   *  tuples stored in the given queue.
+   *  <p>
+   *  N.B. If the queue is full, this operation will block until the given
+   *  elements have been enqueued (or 'timeout_ms' elapses, if specified).
+   *
+   * @param handle The handle to a queue.
+   * @param components One or more tensors from which the enqueued tensors should
+   *  be taken.
+   * @param options carries optional attributes values
+   * @return a new instance of QueueEnqueueMany
+   */
+  public QueueEnqueueMany queueEnqueueMany(Operand<?> handle, Iterable<Operand<?>> components,
+      QueueEnqueueMany.Options... options) {
+    return QueueEnqueueMany.create(scope, handle, components, options);
+  }
+
+  /**
+   * Returns true if queue is closed.
+   *  <p>
+   *  This operation returns true if the queue is closed and false if the queue
+   *  is open.
+   *
+   * @param handle The handle to a queue.
+   * @return a new instance of QueueIsClosed
+   */
+  public QueueIsClosed queueIsClosed(Operand<?> handle) {
+    return QueueIsClosed.create(scope, handle);
+  }
+
+  /**
+   * Computes the number of elements in the given queue.
+   *
+   * @param handle The handle to a queue.
+   * @return a new instance of QueueSize
+   */
+  public QueueSize queueSize(Operand<?> handle) {
+    return QueueSize.create(scope, handle);
+  }
+
+  /**
+   * A queue that randomizes the order of elements.
+   *
+   * @param componentTypes The type of each component in a value.
+   * @param options carries optional attributes values
+   * @return a new instance of RandomShuffleQueue
+   */
+  public RandomShuffleQueue randomShuffleQueue(List<Class<? extends TType>> componentTypes,
+      RandomShuffleQueue.Options... options) {
+    return RandomShuffleQueue.create(scope, componentTypes, options);
+  }
+
+  /**
+   * Reads and outputs the entire contents of the input filename.
+   *
+   * @param filename
+   * @return a new instance of ReadFile
+   */
+  public ReadFile readFile(Operand<TString> filename) {
+    return ReadFile.create(scope, filename);
+  }
+
+  /**
+   * Returns the number of records this Reader has produced.
+   *  <p>
+   *  This is the same as the number of ReaderRead executions that have
+   *  succeeded.
+   *
+   * @param readerHandle Handle to a Reader.
+   * @return a new instance of ReaderNumRecordsProduced
+   */
+  public ReaderNumRecordsProduced readerNumRecordsProduced(Operand<?> readerHandle) {
+    return ReaderNumRecordsProduced.create(scope, readerHandle);
+  }
+
+  /**
+   * Returns the number of work units this Reader has finished processing.
+   *
+   * @param readerHandle Handle to a Reader.
+   * @return a new instance of ReaderNumWorkUnitsCompleted
+   */
+  public ReaderNumWorkUnitsCompleted readerNumWorkUnitsCompleted(Operand<?> readerHandle) {
+    return ReaderNumWorkUnitsCompleted.create(scope, readerHandle);
+  }
+
+  /**
+   * Returns the next record (key, value pair) produced by a Reader.
+   *  <p>
+   *  Will dequeue from the input queue if necessary (e.g. when the
+   *  Reader needs to start reading from a new file since it has finished
+   *  with the previous file).
+   *
+   * @param readerHandle Handle to a Reader.
+   * @param queueHandle Handle to a Queue, with string work items.
+   * @return a new instance of ReaderRead
+   */
+  public ReaderRead readerRead(Operand<?> readerHandle, Operand<?> queueHandle) {
+    return ReaderRead.create(scope, readerHandle, queueHandle);
+  }
+
+  /**
+   * Returns up to `num_records` (key, value) pairs produced by a Reader.
+   *  <p>
+   *  Will dequeue from the input queue if necessary (e.g. when the
+   *  Reader needs to start reading from a new file since it has finished
+   *  with the previous file).
+   *  It may return less than `num_records` even before the last batch.
+   *
+   * @param readerHandle Handle to a `Reader`.
+   * @param queueHandle Handle to a `Queue`, with string work items.
+   * @param numRecords number of records to read from `Reader`.
+   * @return a new instance of ReaderReadUpTo
+   */
+  public ReaderReadUpTo readerReadUpTo(Operand<?> readerHandle, Operand<?> queueHandle,
+      Operand<TInt64> numRecords) {
+    return ReaderReadUpTo.create(scope, readerHandle, queueHandle, numRecords);
+  }
+
+  /**
+   * Restore a Reader to its initial clean state.
+   *
+   * @param readerHandle Handle to a Reader.
+   * @return a new instance of ReaderReset
+   */
+  public ReaderReset readerReset(Operand<?> readerHandle) {
+    return ReaderReset.create(scope, readerHandle);
+  }
+
+  /**
+   * Restore a reader to a previously saved state.
+   *  <p>
+   *  Not all Readers support being restored, so this can produce an
+   *  Unimplemented error.
+   *
+   * @param readerHandle Handle to a Reader.
+   * @param state Result of a ReaderSerializeState of a Reader with type
+   *  matching reader_handle.
+   * @return a new instance of ReaderRestoreState
+   */
+  public ReaderRestoreState readerRestoreState(Operand<?> readerHandle, Operand<TString> state) {
+    return ReaderRestoreState.create(scope, readerHandle, state);
+  }
+
+  /**
+   * Produce a string tensor that encodes the state of a Reader.
+   *  <p>
+   *  Not all Readers support being serialized, so this can produce an
+   *  Unimplemented error.
+   *
+   * @param readerHandle Handle to a Reader.
+   * @return a new instance of ReaderSerializeState
+   */
+  public ReaderSerializeState readerSerializeState(Operand<?> readerHandle) {
+    return ReaderSerializeState.create(scope, readerHandle);
+  }
+
+  /**
+   * Serialize an `N`-minibatch `SparseTensor` into an `[N, 3]` `Tensor` object.
+   *  <p>
+   *  The `SparseTensor` must have rank `R` greater than 1, and the first dimension
+   *  is treated as the minibatch dimension.  Elements of the `SparseTensor`
+   *  must be sorted in increasing order of this first dimension.  The serialized
+   *  `SparseTensor` objects going into each row of `serialized_sparse` will have
+   *  rank `R-1`.
+   *  <p>
+   *  The minibatch size `N` is extracted from `sparse_shape[0]`.
+   *
+   * @param <U> data type for {@code serializedSparse()} output
+   * @param sparseIndices 2-D.  The `indices` of the minibatch `SparseTensor`.
+   * @param sparseValues 1-D.  The `values` of the minibatch `SparseTensor`.
+   * @param sparseShape 1-D.  The `shape` of the minibatch `SparseTensor`.
+   * @return a new instance of SerializeManySparse
+   */
+  public <T extends TType> SerializeManySparse<TString> serializeManySparse(
+      Operand<TInt64> sparseIndices, Operand<T> sparseValues, Operand<TInt64> sparseShape) {
+    return SerializeManySparse.create(scope, sparseIndices, sparseValues, sparseShape);
+  }
+
+  /**
+   * Serialize an `N`-minibatch `SparseTensor` into an `[N, 3]` `Tensor` object.
+   *  <p>
+   *  The `SparseTensor` must have rank `R` greater than 1, and the first dimension
+   *  is treated as the minibatch dimension.  Elements of the `SparseTensor`
+   *  must be sorted in increasing order of this first dimension.  The serialized
+   *  `SparseTensor` objects going into each row of `serialized_sparse` will have
+   *  rank `R-1`.
+   *  <p>
+   *  The minibatch size `N` is extracted from `sparse_shape[0]`.
+   *
+   * @param <U> data type for {@code serializedSparse()} output
+   * @param sparseIndices 2-D.  The `indices` of the minibatch `SparseTensor`.
+   * @param sparseValues 1-D.  The `values` of the minibatch `SparseTensor`.
+   * @param sparseShape 1-D.  The `shape` of the minibatch `SparseTensor`.
+   * @param outType The `dtype` to use for serialization; the supported types are `string`
+   *  (default) and `variant`.
+   * @return a new instance of SerializeManySparse
+   */
+  public <U extends TType, T extends TType> SerializeManySparse<U> serializeManySparse(
+      Operand<TInt64> sparseIndices, Operand<T> sparseValues, Operand<TInt64> sparseShape,
+      Class<U> outType) {
+    return SerializeManySparse.create(scope, sparseIndices, sparseValues, sparseShape, outType);
+  }
+
+  /**
+   * Serialize a `SparseTensor` into a `[3]` `Tensor` object.
+   *
+   * @param <U> data type for {@code serializedSparse()} output
+   * @param sparseIndices 2-D.  The `indices` of the `SparseTensor`.
+   * @param sparseValues 1-D.  The `values` of the `SparseTensor`.
+   * @param sparseShape 1-D.  The `shape` of the `SparseTensor`.
+   * @return a new instance of SerializeSparse
+   */
+  public <T extends TType> SerializeSparse<TString> serializeSparse(Operand<TInt64> sparseIndices,
+      Operand<T> sparseValues, Operand<TInt64> sparseShape) {
+    return SerializeSparse.create(scope, sparseIndices, sparseValues, sparseShape);
+  }
+
+  /**
+   * Serialize a `SparseTensor` into a `[3]` `Tensor` object.
+   *
+   * @param <U> data type for {@code serializedSparse()} output
+   * @param sparseIndices 2-D.  The `indices` of the `SparseTensor`.
+   * @param sparseValues 1-D.  The `values` of the `SparseTensor`.
+   * @param sparseShape 1-D.  The `shape` of the `SparseTensor`.
+   * @param outType The `dtype` to use for serialization; the supported types are `string`
+   *  (default) and `variant`.
+   * @return a new instance of SerializeSparse
+   */
+  public <U extends TType, T extends TType> SerializeSparse<U> serializeSparse(
+      Operand<TInt64> sparseIndices, Operand<T> sparseValues, Operand<TInt64> sparseShape,
+      Class<U> outType) {
+    return SerializeSparse.create(scope, sparseIndices, sparseValues, sparseShape, outType);
+  }
+
+  /**
+   * Transforms a Tensor into a serialized TensorProto proto.
+   *
+   * @param tensor A Tensor of type `T`.
+   * @return a new instance of SerializeTensor
+   */
+  public <T extends TType> SerializeTensor serializeTensor(Operand<T> tensor) {
+    return SerializeTensor.create(scope, tensor);
+  }
+
+  /**
+   * Generate a sharded filename. The filename is printf formatted as
+   *  <p>
+   *     %s-%05d-of-%05d, basename, shard, num_shards.
+   *
+   * @param basename
+   * @param shard
+   * @param numShards
+   * @return a new instance of ShardedFilename
+   */
+  public ShardedFilename shardedFilename(Operand<TString> basename, Operand<TInt32> shard,
+      Operand<TInt32> numShards) {
+    return ShardedFilename.create(scope, basename, shard, numShards);
+  }
+
+  /**
+   * Generate a glob pattern matching all sharded file names.
+   *
+   * @param basename
+   * @param numShards
+   * @return a new instance of ShardedFilespec
+   */
+  public ShardedFilespec shardedFilespec(Operand<TString> basename, Operand<TInt32> numShards) {
+    return ShardedFilespec.create(scope, basename, numShards);
+  }
+
+  /**
+   * A Reader that outputs the lines of a file delimited by '\n'.
+   *
+   * @param options carries optional attributes values
+   * @return a new instance of TextLineReader
+   */
+  public TextLineReader textLineReader(TextLineReader.Options... options) {
+    return TextLineReader.create(scope, options);
+  }
+
+  /**
+   * A Reader that outputs the records from a TensorFlow Records file.
+   *
+   * @param options carries optional attributes values
+   * @return a new instance of TfRecordReader
+   */
+  public TfRecordReader tfRecordReader(TfRecordReader.Options... options) {
+    return TfRecordReader.create(scope, options);
+  }
+
+  /**
+   * A Reader that outputs the entire contents of a file as a value.
+   *  <p>
+   *  To use, enqueue filenames in a Queue.  The output of ReaderRead will
+   *  be a filename (key) and the contents of that file (value).
+   *
+   * @param options carries optional attributes values
+   * @return a new instance of WholeFileReader
+   */
+  public WholeFileReader wholeFileReader(WholeFileReader.Options... options) {
+    return WholeFileReader.create(scope, options);
+  }
+
+  /**
+   * Writes contents to the file at input filename. Creates file and recursively
+   *  <p>
+   *  creates directory if not existing.
+   *
+   * @param filename scalar. The name of the file to which we write the contents.
+   * @param contents scalar. The content to be written to the output file.
+   * @return a new instance of WriteFile
+   */
+  public WriteFile writeFile(Operand<TString> filename, Operand<TString> contents) {
+    return WriteFile.create(scope, filename, contents);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/LinalgOps.java b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/LinalgOps.java
new file mode 100644
index 00000000000..8163e0a50b3
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/LinalgOps.java
@@ -0,0 +1,1608 @@
+// Copyright 2020 The TensorFlow Authors. All Rights Reserved.
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+// ==============================================================================
+//
+// This class has been generated, DO NOT EDIT!
+//
+package org.tensorflow.op;
+
+import org.tensorflow.Operand;
+import org.tensorflow.op.linalg.BandPart;
+import org.tensorflow.op.linalg.BatchCholesky;
+import org.tensorflow.op.linalg.BatchCholeskyGrad;
+import org.tensorflow.op.linalg.BatchMatrixBandPart;
+import org.tensorflow.op.linalg.BatchMatrixDeterminant;
+import org.tensorflow.op.linalg.BatchMatrixDiag;
+import org.tensorflow.op.linalg.BatchMatrixDiagPart;
+import org.tensorflow.op.linalg.BatchMatrixInverse;
+import org.tensorflow.op.linalg.BatchMatrixSetDiag;
+import org.tensorflow.op.linalg.BatchMatrixSolve;
+import org.tensorflow.op.linalg.BatchMatrixSolveLs;
+import org.tensorflow.op.linalg.BatchMatrixTriangularSolve;
+import org.tensorflow.op.linalg.BatchSelfAdjointEig;
+import org.tensorflow.op.linalg.BatchSvd;
+import org.tensorflow.op.linalg.Cholesky;
+import org.tensorflow.op.linalg.CholeskyGrad;
+import org.tensorflow.op.linalg.ConjugateTranspose;
+import org.tensorflow.op.linalg.Cross;
+import org.tensorflow.op.linalg.Det;
+import org.tensorflow.op.linalg.Eig;
+import org.tensorflow.op.linalg.Einsum;
+import org.tensorflow.op.linalg.EuclideanNorm;
+import org.tensorflow.op.linalg.Inv;
+import org.tensorflow.op.linalg.LoadAndRemapMatrix;
+import org.tensorflow.op.linalg.LogMatrixDeterminant;
+import org.tensorflow.op.linalg.Lu;
+import org.tensorflow.op.linalg.MatMul;
+import org.tensorflow.op.linalg.MatrixDiag;
+import org.tensorflow.op.linalg.MatrixDiagPart;
+import org.tensorflow.op.linalg.MatrixDiagPartV3;
+import org.tensorflow.op.linalg.MatrixDiagV3;
+import org.tensorflow.op.linalg.MatrixSetDiag;
+import org.tensorflow.op.linalg.MatrixSolveLs;
+import org.tensorflow.op.linalg.Qr;
+import org.tensorflow.op.linalg.QuantizedMatMul;
+import org.tensorflow.op.linalg.SelfAdjointEig;
+import org.tensorflow.op.linalg.Solve;
+import org.tensorflow.op.linalg.Sqrtm;
+import org.tensorflow.op.linalg.Svd;
+import org.tensorflow.op.linalg.TensorDiag;
+import org.tensorflow.op.linalg.TensorDiagPart;
+import org.tensorflow.op.linalg.Transpose;
+import org.tensorflow.op.linalg.TriangularSolve;
+import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.TFloat64;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.TString;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * An API for building {@code linalg} operations as {@link Op Op}s
+ *
+ * @see {@link Ops}
+ */
+public final class LinalgOps {
+  private final Scope scope;
+
+  LinalgOps(Scope scope) {
+    this.scope = scope;
+  }
+
+  /**
+   * Copy a tensor setting everything outside a central band in each innermost matrix to zero.
+   *  <p>
+   *  The `band` part is computed as follows:
+   *  Assume `input` has `k` dimensions `[I, J, K, ..., M, N]`, then the output is a
+   *  tensor with the same shape where
+   *  <p>
+   *  `band[i, j, k, ..., m, n] = in_band(m, n) * input[i, j, k, ..., m, n]`.
+   *  <p>
+   *  The indicator function
+   *  <p>
+   *  `in_band(m, n) = (num_lower < 0 || (m-n) <= num_lower)) &&
+   *                   (num_upper < 0 || (n-m) <= num_upper)`.
+   *  <p>
+   *  For example:
+   *  <pre>{@code
+   *  # if 'input' is [[ 0,  1,  2, 3]
+   *                   [-1,  0,  1, 2]
+   *                   [-2, -1,  0, 1]
+   *                   [-3, -2, -1, 0]],
+   *
+   *  tf.matrix_band_part(input, 1, -1) ==> [[ 0,  1,  2, 3]
+   *                                         [-1,  0,  1, 2]
+   *                                         [ 0, -1,  0, 1]
+   *                                         [ 0,  0, -1, 0]],
+   *
+   *  tf.matrix_band_part(input, 2, 1) ==> [[ 0,  1,  0, 0]
+   *                                        [-1,  0,  1, 0]
+   *                                        [-2, -1,  0, 1]
+   *                                        [ 0, -2, -1, 0]]
+   *  }</pre>
+   *  Useful special cases:
+   *  <pre>{@code
+   *   tf.matrix_band_part(input, 0, -1) ==> Upper triangular part.
+   *   tf.matrix_band_part(input, -1, 0) ==> Lower triangular part.
+   *   tf.matrix_band_part(input, 0, 0) ==> Diagonal.
+   *  }</pre>
+   *
+   * @param <T> data type for {@code band()} output
+   * @param input Rank `k` tensor.
+   * @param numLower 0-D tensor. Number of subdiagonals to keep. If negative, keep entire
+   *  lower triangle.
+   * @param numUpper 0-D tensor. Number of superdiagonals to keep. If negative, keep
+   *  entire upper triangle.
+   * @return a new instance of BandPart
+   */
+  public <T extends TType, U extends TNumber> BandPart<T> bandPart(Operand<T> input,
+      Operand<U> numLower, Operand<U> numUpper) {
+    return BandPart.create(scope, input, numLower, numUpper);
+  }
+
+  /**
+   *
+   * @param <T> data type for {@code output()} output
+   * @param input
+   * @return a new instance of BatchCholesky
+   */
+  public <T extends TNumber> BatchCholesky<T> batchCholesky(Operand<T> input) {
+    return BatchCholesky.create(scope, input);
+  }
+
+  /**
+   *
+   * @param <T> data type for {@code output()} output
+   * @param l
+   * @param grad
+   * @return a new instance of BatchCholeskyGrad
+   */
+  public <T extends TNumber> BatchCholeskyGrad<T> batchCholeskyGrad(Operand<T> l, Operand<T> grad) {
+    return BatchCholeskyGrad.create(scope, l, grad);
+  }
+
+  /**
+   *
+   * @param <T> data type for {@code band()} output
+   * @param input
+   * @param numLower
+   * @param numUpper
+   * @return a new instance of BatchMatrixBandPart
+   */
+  public <T extends TType> BatchMatrixBandPart<T> batchMatrixBandPart(Operand<T> input,
+      Operand<TInt64> numLower, Operand<TInt64> numUpper) {
+    return BatchMatrixBandPart.create(scope, input, numLower, numUpper);
+  }
+
+  /**
+   *
+   * @param <T> data type for {@code output()} output
+   * @param input
+   * @return a new instance of BatchMatrixDeterminant
+   */
+  public <T extends TType> BatchMatrixDeterminant<T> batchMatrixDeterminant(Operand<T> input) {
+    return BatchMatrixDeterminant.create(scope, input);
+  }
+
+  /**
+   *
+   * @param <T> data type for {@code output()} output
+   * @param diagonal
+   * @return a new instance of BatchMatrixDiag
+   */
+  public <T extends TType> BatchMatrixDiag<T> batchMatrixDiag(Operand<T> diagonal) {
+    return BatchMatrixDiag.create(scope, diagonal);
+  }
+
+  /**
+   *
+   * @param <T> data type for {@code diagonal()} output
+   * @param input
+   * @return a new instance of BatchMatrixDiagPart
+   */
+  public <T extends TType> BatchMatrixDiagPart<T> batchMatrixDiagPart(Operand<T> input) {
+    return BatchMatrixDiagPart.create(scope, input);
+  }
+
+  /**
+   *
+   * @param <T> data type for {@code output()} output
+   * @param input
+   * @param options carries optional attributes values
+   * @return a new instance of BatchMatrixInverse
+   */
+  public <T extends TNumber> BatchMatrixInverse<T> batchMatrixInverse(Operand<T> input,
+      BatchMatrixInverse.Options... options) {
+    return BatchMatrixInverse.create(scope, input, options);
+  }
+
+  /**
+   *
+   * @param <T> data type for {@code output()} output
+   * @param input
+   * @param diagonal
+   * @return a new instance of BatchMatrixSetDiag
+   */
+  public <T extends TType> BatchMatrixSetDiag<T> batchMatrixSetDiag(Operand<T> input,
+      Operand<T> diagonal) {
+    return BatchMatrixSetDiag.create(scope, input, diagonal);
+  }
+
+  /**
+   *
+   * @param <T> data type for {@code output()} output
+   * @param matrix
+   * @param rhs
+   * @param options carries optional attributes values
+   * @return a new instance of BatchMatrixSolve
+   */
+  public <T extends TNumber> BatchMatrixSolve<T> batchMatrixSolve(Operand<T> matrix, Operand<T> rhs,
+      BatchMatrixSolve.Options... options) {
+    return BatchMatrixSolve.create(scope, matrix, rhs, options);
+  }
+
+  /**
+   *
+   * @param <T> data type for {@code output()} output
+   * @param matrix
+   * @param rhs
+   * @param l2Regularizer
+   * @param options carries optional attributes values
+   * @return a new instance of BatchMatrixSolveLs
+   */
+  public <T extends TNumber> BatchMatrixSolveLs<T> batchMatrixSolveLs(Operand<T> matrix,
+      Operand<T> rhs, Operand<TFloat64> l2Regularizer, BatchMatrixSolveLs.Options... options) {
+    return BatchMatrixSolveLs.create(scope, matrix, rhs, l2Regularizer, options);
+  }
+
+  /**
+   *
+   * @param <T> data type for {@code output()} output
+   * @param matrix
+   * @param rhs
+   * @param options carries optional attributes values
+   * @return a new instance of BatchMatrixTriangularSolve
+   */
+  public <T extends TNumber> BatchMatrixTriangularSolve<T> batchMatrixTriangularSolve(
+      Operand<T> matrix, Operand<T> rhs, BatchMatrixTriangularSolve.Options... options) {
+    return BatchMatrixTriangularSolve.create(scope, matrix, rhs, options);
+  }
+
+  /**
+   *
+   * @param <T> data type for {@code e()} output
+   * @param input
+   * @param options carries optional attributes values
+   * @return a new instance of BatchSelfAdjointEig
+   */
+  public <T extends TNumber> BatchSelfAdjointEig<T> batchSelfAdjointEig(Operand<T> input,
+      BatchSelfAdjointEig.Options... options) {
+    return BatchSelfAdjointEig.create(scope, input, options);
+  }
+
+  /**
+   *
+   * @param <T> data type for {@code s()} output
+   * @param input
+   * @param options carries optional attributes values
+   * @return a new instance of BatchSvd
+   */
+  public <T extends TType> BatchSvd<T> batchSvd(Operand<T> input, BatchSvd.Options... options) {
+    return BatchSvd.create(scope, input, options);
+  }
+
+  /**
+   * Computes the Cholesky decomposition of one or more square matrices.
+   *  <p>
+   *  The input is a tensor of shape `[..., M, M]` whose inner-most 2 dimensions
+   *  form square matrices.
+   *  <p>
+   *  The input has to be symmetric and positive definite. Only the lower-triangular
+   *  part of the input will be used for this operation. The upper-triangular part
+   *  will not be read.
+   *  <p>
+   *  The output is a tensor of the same shape as the input
+   *  containing the Cholesky decompositions for all input submatrices `[..., :, :]`.
+   *  <p>
+   *  <b>Note</b>: The gradient computation on GPU is faster for large matrices but
+   *  not for large batch dimensions when the submatrices are small. In this
+   *  case it might be faster to use the CPU.
+   *
+   * @param <T> data type for {@code output()} output
+   * @param input Shape is `[..., M, M]`.
+   * @return a new instance of Cholesky
+   */
+  public <T extends TType> Cholesky<T> cholesky(Operand<T> input) {
+    return Cholesky.create(scope, input);
+  }
+
+  /**
+   * Computes the reverse mode backpropagated gradient of the Cholesky algorithm.
+   *  <p>
+   *  For an explanation see "Differentiation of the Cholesky algorithm" by
+   *  Iain Murray http://arxiv.org/abs/1602.07527.
+   *
+   * @param <T> data type for {@code output()} output
+   * @param l Output of batch Cholesky algorithm l = cholesky(A). Shape is `[..., M, M]`.
+   *  Algorithm depends only on lower triangular part of the innermost matrices of
+   *  this tensor.
+   * @param grad df/dl where f is some scalar function. Shape is `[..., M, M]`.
+   *  Algorithm depends only on lower triangular part of the innermost matrices of
+   *  this tensor.
+   * @return a new instance of CholeskyGrad
+   */
+  public <T extends TNumber> CholeskyGrad<T> choleskyGrad(Operand<T> l, Operand<T> grad) {
+    return CholeskyGrad.create(scope, l, grad);
+  }
+
+  /**
+   * Shuffle dimensions of x according to a permutation and conjugate the result.
+   *  <p>
+   *  The output `y` has the same rank as `x`. The shapes of `x` and `y` satisfy:
+   *    `y.shape[i] == x.shape[perm[i]] for i in [0, 1, ..., rank(x) - 1]`
+   *    `y[i,j,k,...,s,t,u] == conj(x[perm[i], perm[j], perm[k],...,perm[s], perm[t], perm[u]])`
+   *
+   * @param <T> data type for {@code y()} output
+   * @param x
+   * @param perm
+   * @return a new instance of ConjugateTranspose
+   */
+  public <T extends TType, U extends TNumber> ConjugateTranspose<T> conjugateTranspose(Operand<T> x,
+      Operand<U> perm) {
+    return ConjugateTranspose.create(scope, x, perm);
+  }
+
+  /**
+   * Compute the pairwise cross product.
+   *  <p>
+   *  `a` and `b` must be the same shape; they can either be simple 3-element vectors,
+   *  or any shape where the innermost dimension is 3. In the latter case, each pair
+   *  of corresponding 3-element vectors is cross-multiplied independently.
+   *
+   * @param <T> data type for {@code product()} output
+   * @param a A tensor containing 3-element vectors.
+   * @param b Another tensor, of same type and shape as `a`.
+   * @return a new instance of Cross
+   */
+  public <T extends TNumber> Cross<T> cross(Operand<T> a, Operand<T> b) {
+    return Cross.create(scope, a, b);
+  }
+
+  /**
+   * Computes the determinant of one or more square matrices.
+   *  <p>
+   *  The input is a tensor of shape `[..., M, M]` whose inner-most 2 dimensions
+   *  form square matrices. The output is a tensor containing the determinants
+   *  for all input submatrices `[..., :, :]`.
+   *
+   * @param <T> data type for {@code output()} output
+   * @param input Shape is `[..., M, M]`.
+   * @return a new instance of Det
+   */
+  public <T extends TType> Det<T> det(Operand<T> input) {
+    return Det.create(scope, input);
+  }
+
+  /**
+   * Computes the eigen decomposition of one or more square matrices.
+   *  <p>
+   *  Computes the eigenvalues and (optionally) right eigenvectors of each inner matrix in
+   *  `input` such that `input[..., :, :] = v[..., :, :] * diag(e[..., :])`. The eigenvalues
+   *  are sorted in non-decreasing order.
+   *  <pre>{@code
+   *  # a is a tensor.
+   *  # e is a tensor of eigenvalues.
+   *  # v is a tensor of eigenvectors.
+   *  e, v = eig(a)
+   *  e = eig(a, compute_v=False)
+   *  }</pre>
+   *
+   * @param <U> data type for {@code e()} output
+   * @param input `Tensor` input of shape `[N, N]`.
+   * @param Tout
+   * @param options carries optional attributes values
+   * @return a new instance of Eig
+   */
+  public <U extends TType, T extends TType> Eig<U> eig(Operand<T> input, Class<U> Tout,
+      Eig.Options... options) {
+    return Eig.create(scope, input, Tout, options);
+  }
+
+  /**
+   * Tensor contraction according to Einstein summation convention.
+   *  <p>
+   *  Implements generalized Tensor contraction and reduction. Each input Tensor must
+   *  have a corresponding input subscript appearing in the comma-separated left-hand
+   *  side of the equation. The right-hand side of the equation consists of the
+   *  output subscript. The input subscripts and the output subscript should consist
+   *  of zero or more named axis labels and at most one ellipsis (`...`).
+   *  <p>
+   *  The named axis labels may be any single character other than those having
+   *  special meaning, namely `,.->`. The behavior of this Op is undefined if it
+   *  receives an ill-formatted equation; since the validation is done at
+   *  graph-building time, we omit format validation checks at runtime.
+   *  <p>
+   *  Note: This Op is <i>not</i> intended to be called by the user; instead users should
+   *  call `tf.einsum` directly. It is a hidden Op used by `tf.einsum`.
+   *  <p>
+   *  Operations are applied to the input(s) according to the following rules:
+   *  <p>
+   *   (a) Generalized Diagonals: For input dimensions corresponding to axis labels
+   *       appearing more than once in the same input subscript, we take the
+   *       generalized (`k`-dimensional) diagonal.
+   *       For example, in the equation `iii->i` with input shape `[3, 3, 3]`, the
+   *       generalized diagonal would consist of `3` elements at indices `(0, 0, 0)`,
+   *       `(1, 1, 1)` and `(2, 2, 2)` to create a Tensor of shape `[3]`.
+   *  <p>
+   *   (b) Reduction: Axes corresponding to labels appearing only in one input
+   *       subscript but not in the output subscript are summed over prior to Tensor
+   *       contraction.
+   *       For example, in the equation `ab,bc->b`, the axis labels `a` and `c` are
+   *       the reduction axis labels.
+   *  <p>
+   *   (c) Batch Dimensions: Axes corresponding to labels appearing in each of the
+   *       input subscripts and also in the output subscript make up the batch
+   *       dimensions in Tensor contraction. Unnamed axis labels corresponding to
+   *       ellipsis (`...`) also correspond to batch dimensions.
+   *       For example, for the equation denoting batch matrix multiplication,
+   *       `bij,bjk->bik`, the axis label `b` corresponds to a batch dimension.
+   *  <p>
+   *   (d) Contraction: In case of binary einsum, axes corresponding to labels
+   *       appearing in two different inputs (and not in the output) are contracted
+   *       against each other.
+   *       Considering the batch matrix multiplication equation again
+   *       (`bij,bjk->bik`), the contracted axis label is `j`.
+   *  <p>
+   *   (e) Expand Diagonal: If the output subscripts contain repeated (explicit) axis
+   *       labels, the opposite operation of (a) is applied. For example, in the
+   *       equation `i->iii`, and input shape `[3]`, the output of shape `[3, 3, 3]`
+   *       are all zeros, except for the (generalized) diagonal which is populated
+   *       with values from the input.
+   *       Note: This operation is not supported by `np.einsum` or `tf.einsum`; it is
+   *       provided to enable computing the symbolic gradient of `tf.einsum`.
+   *  <p>
+   *  The output subscripts must contain only labels appearing in at least one of the
+   *  input subscripts. Furthermore, all dimensions mapping to the same axis label
+   *  must be equal.
+   *  <p>
+   *  Any of the input and output subscripts may contain at most a single ellipsis
+   *  (`...`). These ellipsis are mapped against dimensions not corresponding to any
+   *  named axis label. If two inputs contain ellipsis, then they are broadcasted
+   *  according to standard NumPy broadcasting
+   *  [rules](http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html).
+   *  <p>
+   *  The broadcasted dimensions are placed in the corresponding location of the
+   *  ellipsis in the output subscript. If the broadcasted dimensions are non-empty
+   *  and the output subscripts do not contain ellipsis, then an InvalidArgument error
+   *  is raised.
+   *  <p>
+   *
+   * @compatibility(numpy) Similar to [`numpy.einsum`](https://docs.scipy.org/doc/numpy/reference/generated/numpy.einsum.html).
+   *  <p>
+   *  Comparison with `numpy.einsum`:
+   *  <p>
+   * This Op only supports unary and binary forms of `numpy.einsum`.
+   * This Op does not support implicit form. (i.e. equations without `->`).
+   * This Op also supports repeated indices in the output subscript, which is not
+   *     supported by `numpy.einsum`.
+   * @end_compatibility
+   * @param <T> data type for {@code output()} output
+   * @param inputs List of 1 or 2 Tensors.
+   * @param equation String describing the Einstein Summation operation; in the format of np.einsum.
+   * @return a new instance of Einsum
+   */
+  public <T extends TType> Einsum<T> einsum(Iterable<Operand<T>> inputs, String equation) {
+    return Einsum.create(scope, inputs, equation);
+  }
+
+  /**
+   * Computes the euclidean norm of elements across dimensions of a tensor.
+   *  <p>
+   *  Reduces `input` along the dimensions given in `axis`. Unless
+   *  `keep_dims` is true, the rank of the tensor is reduced by 1 for each entry in
+   *  `axis`. If `keep_dims` is true, the reduced dimensions are
+   *  retained with length 1.
+   *
+   * @param <T> data type for {@code output()} output
+   * @param input The tensor to reduce.
+   * @param axis The dimensions to reduce. Must be in the range
+   *  `[-rank(input), rank(input))`.
+   * @param options carries optional attributes values
+   * @return a new instance of EuclideanNorm
+   */
+  public <T extends TType, U extends TNumber> EuclideanNorm<T> euclideanNorm(Operand<T> input,
+      Operand<U> axis, EuclideanNorm.Options... options) {
+    return EuclideanNorm.create(scope, input, axis, options);
+  }
+
+  /**
+   * Computes the inverse of one or more square invertible matrices or their
+   *  <p>
+   *  adjoints (conjugate transposes).
+   *  <p>
+   *  The input is a tensor of shape `[..., M, M]` whose inner-most 2 dimensions
+   *  form square matrices. The output is a tensor of the same shape as the input
+   *  containing the inverse for all input submatrices `[..., :, :]`.
+   *  <p>
+   *  The op uses LU decomposition with partial pivoting to compute the inverses.
+   *  <p>
+   *  If a matrix is not invertible there is no guarantee what the op does. It
+   *  may detect the condition and raise an exception or it may simply return a
+   *  garbage result.
+   *
+   * @param <T> data type for {@code output()} output
+   * @param input Shape is `[..., M, M]`.
+   * @param options carries optional attributes values
+   * @return a new instance of Inv
+   */
+  public <T extends TType> Inv<T> inv(Operand<T> input, Inv.Options... options) {
+    return Inv.create(scope, input, options);
+  }
+
+  /**
+   * Loads a 2-D (matrix) `Tensor` with name `old_tensor_name` from the checkpoint
+   *  <p>
+   *  at `ckpt_path` and potentially reorders its rows and columns using the
+   *  specified remappings.
+   *  <p>
+   *  Most users should use one of the wrapper initializers (such as
+   *  `tf.contrib.framework.load_and_remap_matrix_initializer`) instead of this
+   *  function directly.
+   *  <p>
+   *  The remappings are 1-D tensors with the following properties:
+   *  <ul>
+   *  <li>
+   *  `row_remapping` must have exactly `num_rows` entries. Row `i` of the output
+   *    matrix will be initialized from the row corresponding to index
+   *    `row_remapping[i]` in the old `Tensor` from the checkpoint.
+   *  </li>
+   *  <li>
+   *  `col_remapping` must have either 0 entries (indicating that no column
+   *    reordering is needed) or `num_cols` entries. If specified, column `j` of the
+   *    output matrix will be initialized from the column corresponding to index
+   *    `col_remapping[j]` in the old `Tensor` from the checkpoint.
+   *  </li>
+   *  <li>
+   *  A value of -1 in either of the remappings signifies a "missing" entry. In that
+   *    case, values from the `initializing_values` tensor will be used to fill that
+   *    missing row or column. If `row_remapping` has `r` missing entries and
+   *    `col_remapping` has `c` missing entries, then the following condition must be
+   *    true:
+   *  </li>
+   *  </ul>
+   *  `(r * num_cols) + (c * num_rows) - (r * c) == len(initializing_values)`
+   *  <p>
+   *  The remapping tensors can be generated using the GenerateVocabRemapping op.
+   *  <p>
+   *  As an example, with row_remapping = [1, 0, -1], col_remapping = [0, 2, -1],
+   *  initializing_values = [0.5, -0.5, 0.25, -0.25, 42], and w(i, j) representing
+   *  the value from row i, column j of the old tensor in the checkpoint, the output
+   *  matrix will look like the following:
+   *  <p>
+   *  [[w(1, 0),  w(1, 2),  0.5],
+   *   [w(0, 0),  w(0, 2), -0.5],
+   *   [0.25,    -0.25,      42]]
+   *
+   * @param ckptPath Path to the TensorFlow checkpoint (version 2, `TensorBundle`) from
+   *  which the old matrix `Tensor` will be loaded.
+   * @param oldTensorName Name of the 2-D `Tensor` to load from checkpoint.
+   * @param rowRemapping An int `Tensor` of row remappings (generally created by
+   *  `generate_vocab_remapping`).  Even if no row remapping is needed, this must
+   *  still be an index-valued Tensor (e.g. [0, 1, 2, ...]), or a shifted
+   *  index-valued `Tensor` (e.g. [8, 9, 10, ...], for partitioned `Variables`).
+   * @param colRemapping An int `Tensor` of column remappings (generally created by
+   *  `generate_vocab_remapping`).  May be a size-0 `Tensor` if only row remapping
+   *  is to be done (e.g. column ordering is the same).
+   * @param initializingValues A float `Tensor` containing  values to fill in for cells
+   *  in the output matrix that are not loaded from the checkpoint. Length must be
+   *  exactly the same as the number of missing / new cells.
+   * @param numRows Number of rows (length of the 1st dimension) in the output matrix.
+   * @param numCols Number of columns (length of the 2nd dimension) in the output matrix.
+   * @param options carries optional attributes values
+   * @return a new instance of LoadAndRemapMatrix
+   */
+  public LoadAndRemapMatrix loadAndRemapMatrix(Operand<TString> ckptPath,
+      Operand<TString> oldTensorName, Operand<TInt64> rowRemapping, Operand<TInt64> colRemapping,
+      Operand<TFloat32> initializingValues, Long numRows, Long numCols,
+      LoadAndRemapMatrix.Options... options) {
+    return LoadAndRemapMatrix.create(scope, ckptPath, oldTensorName, rowRemapping, colRemapping, initializingValues, numRows, numCols, options);
+  }
+
+  /**
+   * Computes the sign and the log of the absolute value of the determinant of
+   *  <p>
+   *  one or more square matrices.
+   *  <p>
+   *  The input is a tensor of shape `[N, M, M]` whose inner-most 2 dimensions
+   *  form square matrices. The outputs are two tensors containing the signs and
+   *  absolute values of the log determinants for all N input submatrices
+   *  `[..., :, :]` such that the determinant = sign*exp(log_abs_determinant).
+   *  The log_abs_determinant is computed as det(P)*sum(log(diag(LU))) where LU
+   *  is the LU decomposition of the input and P is the corresponding
+   *  permutation matrix.
+   *
+   * @param <T> data type for {@code sign()} output
+   * @param input Shape is `[N, M, M]`.
+   * @return a new instance of LogMatrixDeterminant
+   */
+  public <T extends TType> LogMatrixDeterminant<T> logMatrixDeterminant(Operand<T> input) {
+    return LogMatrixDeterminant.create(scope, input);
+  }
+
+  /**
+   * Computes the LU decomposition of one or more square matrices.
+   *  <p>
+   *  The input is a tensor of shape `[..., M, M]` whose inner-most 2 dimensions
+   *  form square matrices.
+   *  <p>
+   *  The input has to be invertible.
+   *  <p>
+   *  The output consists of two tensors LU and P containing the LU decomposition
+   *  of all input submatrices `[..., :, :]`. LU encodes the lower triangular and
+   *  upper triangular factors.
+   *  <p>
+   *  For each input submatrix of shape `[M, M]`, L is a lower triangular matrix of
+   *  shape `[M, M]` with unit diagonal whose entries correspond to the strictly lower
+   *  triangular part of LU. U is a upper triangular matrix of shape `[M, M]` whose
+   *  entries correspond to the upper triangular part, including the diagonal, of LU.
+   *  <p>
+   *  P represents a permutation matrix encoded as a list of indices each between `0`
+   *  and `M-1`, inclusive. If P_mat denotes the permutation matrix corresponding to
+   *  P, then the L, U and P satisfies P_mat * input = L * U.
+   *
+   * @param <T> data type for {@code lu()} output
+   * @param <U> data type for {@code p()} output
+   * @param input A tensor of shape `[..., M, M]` whose inner-most 2 dimensions form matrices of
+   *  size `[M, M]`.
+   * @return a new instance of Lu
+   */
+  public <T extends TType> Lu<T, TInt32> lu(Operand<T> input) {
+    return Lu.create(scope, input);
+  }
+
+  /**
+   * Computes the LU decomposition of one or more square matrices.
+   *  <p>
+   *  The input is a tensor of shape `[..., M, M]` whose inner-most 2 dimensions
+   *  form square matrices.
+   *  <p>
+   *  The input has to be invertible.
+   *  <p>
+   *  The output consists of two tensors LU and P containing the LU decomposition
+   *  of all input submatrices `[..., :, :]`. LU encodes the lower triangular and
+   *  upper triangular factors.
+   *  <p>
+   *  For each input submatrix of shape `[M, M]`, L is a lower triangular matrix of
+   *  shape `[M, M]` with unit diagonal whose entries correspond to the strictly lower
+   *  triangular part of LU. U is a upper triangular matrix of shape `[M, M]` whose
+   *  entries correspond to the upper triangular part, including the diagonal, of LU.
+   *  <p>
+   *  P represents a permutation matrix encoded as a list of indices each between `0`
+   *  and `M-1`, inclusive. If P_mat denotes the permutation matrix corresponding to
+   *  P, then the L, U and P satisfies P_mat * input = L * U.
+   *
+   * @param <T> data type for {@code lu()} output
+   * @param <U> data type for {@code p()} output
+   * @param input A tensor of shape `[..., M, M]` whose inner-most 2 dimensions form matrices of
+   *  size `[M, M]`.
+   * @param outputIdxType
+   * @return a new instance of Lu
+   */
+  public <T extends TType, U extends TNumber> Lu<T, U> lu(Operand<T> input,
+      Class<U> outputIdxType) {
+    return Lu.create(scope, input, outputIdxType);
+  }
+
+  /**
+   * Multiply the matrix "a" by the matrix "b".
+   *  <p>
+   *  The inputs must be two-dimensional matrices and the inner dimension of
+   *  "a" (after being transposed if transpose_a is true) must match the
+   *  outer dimension of "b" (after being transposed if transposed_b is
+   *  true).
+   *  <p>
+   *  <i>Note</i>: The default kernel implementation for MatMul on GPUs uses
+   *  cublas.
+   *
+   * @param <T> data type for {@code product()} output
+   * @param a
+   * @param b
+   * @param options carries optional attributes values
+   * @return a new instance of MatMul
+   */
+  public <T extends TType> MatMul<T> matMul(Operand<T> a, Operand<T> b, MatMul.Options... options) {
+    return MatMul.create(scope, a, b, options);
+  }
+
+  /**
+   * Returns a batched diagonal tensor with given batched diagonal values.
+   *  <p>
+   *  Returns a tensor with the contents in `diagonal` as `k[0]`-th to `k[1]`-th
+   *  diagonals of a matrix, with everything else padded with `padding`. `num_rows`
+   *  and `num_cols` specify the dimension of the innermost matrix of the output. If
+   *  both are not specified, the op assumes the innermost matrix is square and infers
+   *  its size from `k` and the innermost dimension of `diagonal`. If only one of them
+   *  is specified, the op assumes the unspecified value is the smallest possible
+   *  based on other criteria.
+   *  <p>
+   *  Let `diagonal` have `r` dimensions `[I, J, ..., L, M, N]`. The output tensor has
+   *  rank `r+1` with shape `[I, J, ..., L, M, num_rows, num_cols]` when only one
+   *  diagonal is given (`k` is an integer or `k[0] == k[1]`). Otherwise, it has rank
+   *  `r` with shape `[I, J, ..., L, num_rows, num_cols]`.
+   *  <p>
+   *  The second innermost dimension of `diagonal` has double meaning.
+   *  When `k` is scalar or `k[0] == k[1]`, `M` is part of the batch size
+   *  [I, J, ..., M], and the output tensor is:
+   *  <pre>{@code
+   *  output[i, j, ..., l, m, n]
+   *    = diagonal[i, j, ..., l, n-max(d_upper, 0)] ; if n - m == d_upper
+   *      padding_value                             ; otherwise
+   *  }</pre>
+   *  Otherwise, `M` is treated as the number of diagonals for the matrix in the
+   *  same batch (`M = k[1]-k[0]+1`), and the output tensor is:
+   *  <pre>{@code
+   *  output[i, j, ..., l, m, n]
+   *    = diagonal[i, j, ..., l, diag_index, index_in_diag] ; if k[0] <= d <= k[1]
+   *      padding_value                                     ; otherwise
+   *  }</pre>
+   *  where `d = n - m`, `diag_index = k[1] - d`, and `index_in_diag = n - max(d, 0)`.
+   *  <p>
+   *  For example:
+   *  <pre>{@code
+   *  # The main diagonal.
+   *  diagonal = np.array([[1, 2, 3, 4],            # Input shape: (2, 4)
+   *                       [5, 6, 7, 8]])
+   *  tf.matrix_diag(diagonal) ==> [[[1, 0, 0, 0],  # Output shape: (2, 4, 4)
+   *                                 [0, 2, 0, 0],
+   *                                 [0, 0, 3, 0],
+   *                                 [0, 0, 0, 4]],
+   *                                [[5, 0, 0, 0],
+   *                                 [0, 6, 0, 0],
+   *                                 [0, 0, 7, 0],
+   *                                 [0, 0, 0, 8]]]
+   *
+   *  # A superdiagonal (per batch).
+   *  diagonal = np.array([[1, 2, 3],  # Input shape: (2, 3)
+   *                       [4, 5, 6]])
+   *  tf.matrix_diag(diagonal, k = 1)
+   *    ==> [[[0, 1, 0, 0],  # Output shape: (2, 4, 4)
+   *          [0, 0, 2, 0],
+   *          [0, 0, 0, 3],
+   *          [0, 0, 0, 0]],
+   *         [[0, 4, 0, 0],
+   *          [0, 0, 5, 0],
+   *          [0, 0, 0, 6],
+   *          [0, 0, 0, 0]]]
+   *
+   *  # A band of diagonals.
+   *  diagonals = np.array([[[1, 2, 3],  # Input shape: (2, 2, 3)
+   *                         [4, 5, 0]],
+   *                        [[6, 7, 9],
+   *                         [9, 1, 0]]])
+   *  tf.matrix_diag(diagonals, k = (-1, 0))
+   *    ==> [[[1, 0, 0],  # Output shape: (2, 3, 3)
+   *          [4, 2, 0],
+   *          [0, 5, 3]],
+   *         [[6, 0, 0],
+   *          [9, 7, 0],
+   *          [0, 1, 9]]]
+   *
+   *  # Rectangular matrix.
+   *  diagonal = np.array([1, 2])  # Input shape: (2)
+   *  tf.matrix_diag(diagonal, k = -1, num_rows = 3, num_cols = 4)
+   *    ==> [[0, 0, 0, 0],  # Output shape: (3, 4)
+   *         [1, 0, 0, 0],
+   *         [0, 2, 0, 0]]
+   *
+   *  # Rectangular matrix with inferred num_cols and padding_value = 9.
+   *  tf.matrix_diag(diagonal, k = -1, num_rows = 3, padding_value = 9)
+   *    ==> [[9, 9],  # Output shape: (3, 2)
+   *         [1, 9],
+   *         [9, 2]]
+   *  }</pre>
+   *
+   * @param <T> data type for {@code output()} output
+   * @param diagonal Rank `r`, where `r >= 1`
+   * @param k Diagonal offset(s). Positive value means superdiagonal, 0 refers to the main
+   *  diagonal, and negative value means subdiagonals. `k` can be a single integer
+   *  (for a single diagonal) or a pair of integers specifying the low and high ends
+   *  of a matrix band. `k[0]` must not be larger than `k[1]`.
+   * @param numRows The number of rows of the output matrix. If it is not provided, the op assumes
+   *  the output matrix is a square matrix and infers the matrix size from k and the
+   *  innermost dimension of `diagonal`.
+   * @param numCols The number of columns of the output matrix. If it is not provided, the op
+   *  assumes the output matrix is a square matrix and infers the matrix size from
+   *  k and the innermost dimension of `diagonal`.
+   * @param paddingValue The number to fill the area outside the specified diagonal band with.
+   *  Default is 0.
+   * @return a new instance of MatrixDiag
+   */
+  public <T extends TType> MatrixDiag<T> matrixDiag(Operand<T> diagonal, Operand<TInt32> k,
+      Operand<TInt32> numRows, Operand<TInt32> numCols, Operand<T> paddingValue) {
+    return MatrixDiag.create(scope, diagonal, k, numRows, numCols, paddingValue);
+  }
+
+  /**
+   * Returns the batched diagonal part of a batched tensor.
+   *  <p>
+   *  Returns a tensor with the `k[0]`-th to `k[1]`-th diagonals of the batched
+   *  `input`.
+   *  <p>
+   *  Assume `input` has `r` dimensions `[I, J, ..., L, M, N]`.
+   *  Let `max_diag_len` be the maximum length among all diagonals to be extracted,
+   *  `max_diag_len = min(M + min(k[1], 0), N + min(-k[0], 0))`
+   *  Let `num_diags` be the number of diagonals to extract,
+   *  `num_diags = k[1] - k[0] + 1`.
+   *  <p>
+   *  If `num_diags == 1`, the output tensor is of rank `r - 1` with shape
+   *  `[I, J, ..., L, max_diag_len]` and values:
+   *  <pre>{@code
+   *  diagonal[i, j, ..., l, n]
+   *    = input[i, j, ..., l, n+y, n+x] ; if 0 <= n+y < M and 0 <= n+x < N,
+   *      padding_value                 ; otherwise.
+   *  }</pre>
+   *  where `y = max(-k[1], 0)`, `x = max(k[1], 0)`.
+   *  <p>
+   *  Otherwise, the output tensor has rank `r` with dimensions
+   *  `[I, J, ..., L, num_diags, max_diag_len]` with values:
+   *  <pre>{@code
+   *  diagonal[i, j, ..., l, m, n]
+   *    = input[i, j, ..., l, n+y, n+x] ; if 0 <= n+y < M and 0 <= n+x < N,
+   *      padding_value                 ; otherwise.
+   *  }</pre>
+   *  where `d = k[1] - m`, `y = max(-d, 0)`, and `x = max(d, 0)`.
+   *  <p>
+   *  The input must be at least a matrix.
+   *  <p>
+   *  For example:
+   *  <pre>{@code
+   *  input = np.array([[[1, 2, 3, 4],  # Input shape: (2, 3, 4)
+   *                     [5, 6, 7, 8],
+   *                     [9, 8, 7, 6]],
+   *                    [[5, 4, 3, 2],
+   *                     [1, 2, 3, 4],
+   *                     [5, 6, 7, 8]]])
+   *
+   *  # A main diagonal from each batch.
+   *  tf.matrix_diag_part(input) ==> [[1, 6, 7],  # Output shape: (2, 3)
+   *                                  [5, 2, 7]]
+   *
+   *  # A superdiagonal from each batch.
+   *  tf.matrix_diag_part(input, k = 1)
+   *    ==> [[2, 7, 6],  # Output shape: (2, 3)
+   *         [4, 3, 8]]
+   *
+   *  # A tridiagonal band from each batch.
+   *  tf.matrix_diag_part(input, k = (-1, 1))
+   *    ==> [[[2, 7, 6],  # Output shape: (2, 3, 3)
+   *          [1, 6, 7],
+   *          [5, 8, 0]],
+   *         [[4, 3, 8],
+   *          [5, 2, 7],
+   *          [1, 6, 0]]]
+   *
+   *  # Padding value = 9
+   *  tf.matrix_diag_part(input, k = (1, 3), padding_value = 9)
+   *    ==> [[[4, 9, 9],  # Output shape: (2, 3, 3)
+   *          [3, 8, 9],
+   *          [2, 7, 6]],
+   *         [[2, 9, 9],
+   *          [3, 4, 9],
+   *          [4, 3, 8]]]
+   *  }</pre>
+   *
+   * @param <T> data type for {@code diagonal()} output
+   * @param input Rank `r` tensor where `r >= 2`.
+   * @param k Diagonal offset(s). Positive value means superdiagonal, 0 refers to the main
+   *  diagonal, and negative value means subdiagonals. `k` can be a single integer
+   *  (for a single diagonal) or a pair of integers specifying the low and high ends
+   *  of a matrix band. `k[0]` must not be larger than `k[1]`.
+   * @param paddingValue The value to fill the area outside the specified diagonal band with.
+   *  Default is 0.
+   * @return a new instance of MatrixDiagPart
+   */
+  public <T extends TType> MatrixDiagPart<T> matrixDiagPart(Operand<T> input, Operand<TInt32> k,
+      Operand<T> paddingValue) {
+    return MatrixDiagPart.create(scope, input, k, paddingValue);
+  }
+
+  /**
+   * Returns the batched diagonal part of a batched tensor.
+   *  <p>
+   *  Returns a tensor with the `k[0]`-th to `k[1]`-th diagonals of the batched
+   *  `input`.
+   *  <p>
+   *  Assume `input` has `r` dimensions `[I, J, ..., L, M, N]`.
+   *  Let `max_diag_len` be the maximum length among all diagonals to be extracted,
+   *  `max_diag_len = min(M + min(k[1], 0), N + min(-k[0], 0))`
+   *  Let `num_diags` be the number of diagonals to extract,
+   *  `num_diags = k[1] - k[0] + 1`.
+   *  <p>
+   *  If `num_diags == 1`, the output tensor is of rank `r - 1` with shape
+   *  `[I, J, ..., L, max_diag_len]` and values:
+   *  <pre>{@code
+   *  diagonal[i, j, ..., l, n]
+   *    = input[i, j, ..., l, n+y, n+x] ; if 0 <= n+y < M and 0 <= n+x < N,
+   *      padding_value                 ; otherwise.
+   *  }</pre>
+   *  where `y = max(-k[1], 0)`, `x = max(k[1], 0)`.
+   *  <p>
+   *  Otherwise, the output tensor has rank `r` with dimensions
+   *  `[I, J, ..., L, num_diags, max_diag_len]` with values:
+   *  <pre>{@code
+   *  diagonal[i, j, ..., l, m, n]
+   *    = input[i, j, ..., l, n+y, n+x] ; if 0 <= n+y < M and 0 <= n+x < N,
+   *      padding_value                 ; otherwise.
+   *  }</pre>
+   *  where `d = k[1] - m`, `y = max(-d, 0) - offset`, and `x = max(d, 0) - offset`.
+   *  <p>
+   *  `offset` is zero except when the alignment of the diagonal is to the right.
+   *  <pre>{@code
+   *  offset = max_diag_len - diag_len(d) ; if (`align` in {RIGHT_LEFT, RIGHT_RIGHT}
+   *                                             and `d >= 0`) or
+   *                                           (`align` in {LEFT_RIGHT, RIGHT_RIGHT}
+   *                                             and `d <= 0`)
+   *           0                          ; otherwise
+   *  }</pre>
+   *  where `diag_len(d) = min(cols - max(d, 0), rows + min(d, 0))`.
+   *  <p>
+   *  The input must be at least a matrix.
+   *  <p>
+   *  For example:
+   *  <pre>{@code
+   *  input = np.array([[[1, 2, 3, 4],  # Input shape: (2, 3, 4)
+   *                     [5, 6, 7, 8],
+   *                     [9, 8, 7, 6]],
+   *                    [[5, 4, 3, 2],
+   *                     [1, 2, 3, 4],
+   *                     [5, 6, 7, 8]]])
+   *
+   *  # A main diagonal from each batch.
+   *  tf.matrix_diag_part(input) ==> [[1, 6, 7],  # Output shape: (2, 3)
+   *                                  [5, 2, 7]]
+   *
+   *  # A superdiagonal from each batch.
+   *  tf.matrix_diag_part(input, k = 1)
+   *    ==> [[2, 7, 6],  # Output shape: (2, 3)
+   *         [4, 3, 8]]
+   *
+   *  # A band from each batch.
+   *  tf.matrix_diag_part(input, k = (-1, 2))
+   *    ==> [[[0, 3, 8],  # Output shape: (2, 4, 3)
+   *          [2, 7, 6],
+   *          [1, 6, 7],
+   *          [5, 8, 0]],
+   *         [[0, 3, 4],
+   *          [4, 3, 8],
+   *          [5, 2, 7],
+   *          [1, 6, 0]]]
+   *
+   *  # LEFT_RIGHT alignment.
+   *  tf.matrix_diag_part(input, k = (-1, 2), align="LEFT_RIGHT")
+   *    ==> [[[3, 8, 0],  # Output shape: (2, 4, 3)
+   *          [2, 7, 6],
+   *          [1, 6, 7],
+   *          [0, 5, 8]],
+   *         [[3, 4, 0],
+   *          [4, 3, 8],
+   *          [5, 2, 7],
+   *          [0, 1, 6]]]
+   *
+   *  # max_diag_len can be shorter than the main diagonal.
+   *  tf.matrix_diag_part(input, k = (-2, -1))
+   *    ==> [[[5, 8],
+   *          [9, 0]],
+   *         [[1, 6],
+   *          [5, 0]]]
+   *
+   *  # padding_value = 9
+   *  tf.matrix_diag_part(input, k = (1, 3), padding_value = 9)
+   *    ==> [[[9, 9, 4],  # Output shape: (2, 3, 3)
+   *          [9, 3, 8],
+   *          [2, 7, 6]],
+   *         [[9, 9, 2],
+   *          [9, 3, 4],
+   *          [4, 3, 8]]]
+   *
+   *  }</pre>
+   *
+   * @param <T> data type for {@code diagonal()} output
+   * @param input Rank `r` tensor where `r >= 2`.
+   * @param k Diagonal offset(s). Positive value means superdiagonal, 0 refers to the main
+   *  diagonal, and negative value means subdiagonals. `k` can be a single integer
+   *  (for a single diagonal) or a pair of integers specifying the low and high ends
+   *  of a matrix band. `k[0]` must not be larger than `k[1]`.
+   * @param paddingValue The value to fill the area outside the specified diagonal band with.
+   *  Default is 0.
+   * @param options carries optional attributes values
+   * @return a new instance of MatrixDiagPartV3
+   */
+  public <T extends TType> MatrixDiagPartV3<T> matrixDiagPartV3(Operand<T> input, Operand<TInt32> k,
+      Operand<T> paddingValue, MatrixDiagPartV3.Options... options) {
+    return MatrixDiagPartV3.create(scope, input, k, paddingValue, options);
+  }
+
+  /**
+   * Returns a batched diagonal tensor with given batched diagonal values.
+   *  <p>
+   *  Returns a tensor with the contents in `diagonal` as `k[0]`-th to `k[1]`-th
+   *  diagonals of a matrix, with everything else padded with `padding`. `num_rows`
+   *  and `num_cols` specify the dimension of the innermost matrix of the output. If
+   *  both are not specified, the op assumes the innermost matrix is square and infers
+   *  its size from `k` and the innermost dimension of `diagonal`. If only one of them
+   *  is specified, the op assumes the unspecified value is the smallest possible
+   *  based on other criteria.
+   *  <p>
+   *  Let `diagonal` have `r` dimensions `[I, J, ..., L, M, N]`. The output tensor has
+   *  rank `r+1` with shape `[I, J, ..., L, M, num_rows, num_cols]` when only one
+   *  diagonal is given (`k` is an integer or `k[0] == k[1]`). Otherwise, it has rank
+   *  `r` with shape `[I, J, ..., L, num_rows, num_cols]`.
+   *  <p>
+   *  The second innermost dimension of `diagonal` has double meaning.
+   *  When `k` is scalar or `k[0] == k[1]`, `M` is part of the batch size
+   *  [I, J, ..., M], and the output tensor is:
+   *  <pre>{@code
+   *  output[i, j, ..., l, m, n]
+   *    = diagonal[i, j, ..., l, n-max(d_upper, 0)] ; if n - m == d_upper
+   *      padding_value                             ; otherwise
+   *  }</pre>
+   *  Otherwise, `M` is treated as the number of diagonals for the matrix in the
+   *  same batch (`M = k[1]-k[0]+1`), and the output tensor is:
+   *  <pre>{@code
+   *  output[i, j, ..., l, m, n]
+   *    = diagonal[i, j, ..., l, diag_index, index_in_diag] ; if k[0] <= d <= k[1]
+   *      padding_value                                     ; otherwise
+   *  }</pre>
+   *  where `d = n - m`, `diag_index = [k] - d`, and
+   *  `index_in_diag = n - max(d, 0) + offset`.
+   *  <p>
+   *  `offset` is zero except when the alignment of the diagonal is to the right.
+   *  <pre>{@code
+   *  offset = max_diag_len - diag_len(d) ; if (`align` in {RIGHT_LEFT, RIGHT_RIGHT}
+   *                                             and `d >= 0`) or
+   *                                           (`align` in {LEFT_RIGHT, RIGHT_RIGHT}
+   *                                             and `d <= 0`)
+   *           0                          ; otherwise
+   *  }</pre>
+   *  where `diag_len(d) = min(cols - max(d, 0), rows + min(d, 0))`.
+   *  <p>
+   *  For example:
+   *  <pre>{@code
+   *  # The main diagonal.
+   *  diagonal = np.array([[1, 2, 3, 4],            # Input shape: (2, 4)
+   *                       [5, 6, 7, 8]])
+   *  tf.matrix_diag(diagonal) ==> [[[1, 0, 0, 0],  # Output shape: (2, 4, 4)
+   *                                 [0, 2, 0, 0],
+   *                                 [0, 0, 3, 0],
+   *                                 [0, 0, 0, 4]],
+   *                                [[5, 0, 0, 0],
+   *                                 [0, 6, 0, 0],
+   *                                 [0, 0, 7, 0],
+   *                                 [0, 0, 0, 8]]]
+   *
+   *  # A superdiagonal (per batch).
+   *  diagonal = np.array([[1, 2, 3],  # Input shape: (2, 3)
+   *                       [4, 5, 6]])
+   *  tf.matrix_diag(diagonal, k = 1)
+   *    ==> [[[0, 1, 0, 0],  # Output shape: (2, 4, 4)
+   *          [0, 0, 2, 0],
+   *          [0, 0, 0, 3],
+   *          [0, 0, 0, 0]],
+   *         [[0, 4, 0, 0],
+   *          [0, 0, 5, 0],
+   *          [0, 0, 0, 6],
+   *          [0, 0, 0, 0]]]
+   *
+   *  # A tridiagonal band (per batch).
+   *  diagonals = np.array([[[0, 8, 9],  # Input shape: (2, 2, 3)
+   *                         [1, 2, 3],
+   *                         [4, 5, 0]],
+   *                        [[0, 2, 3],
+   *                         [6, 7, 9],
+   *                         [9, 1, 0]]])
+   *  tf.matrix_diag(diagonals, k = (-1, 1))
+   *    ==> [[[1, 8, 0],  # Output shape: (2, 3, 3)
+   *          [4, 2, 9],
+   *          [0, 5, 3]],
+   *         [[6, 2, 0],
+   *          [9, 7, 3],
+   *          [0, 1, 9]]]
+   *
+   *  # LEFT_RIGHT alignment.
+   *  diagonals = np.array([[[8, 9, 0],  # Input shape: (2, 2, 3)
+   *                         [1, 2, 3],
+   *                         [0, 4, 5]],
+   *                        [[2, 3, 0],
+   *                         [6, 7, 9],
+   *                         [0, 9, 1]]])
+   *  tf.matrix_diag(diagonals, k = (-1, 1), align="LEFT_RIGHT")
+   *    ==> [[[1, 8, 0],  # Output shape: (2, 3, 3)
+   *          [4, 2, 9],
+   *          [0, 5, 3]],
+   *         [[6, 2, 0],
+   *          [9, 7, 3],
+   *          [0, 1, 9]]]
+   *
+   *  # Rectangular matrix.
+   *  diagonal = np.array([1, 2])  # Input shape: (2)
+   *  tf.matrix_diag(diagonal, k = -1, num_rows = 3, num_cols = 4)
+   *    ==> [[0, 0, 0, 0],  # Output shape: (3, 4)
+   *         [1, 0, 0, 0],
+   *         [0, 2, 0, 0]]
+   *
+   *  # Rectangular matrix with inferred num_cols and padding_value = 9.
+   *  tf.matrix_diag(diagonal, k = -1, num_rows = 3, padding_value = 9)
+   *    ==> [[9, 9],  # Output shape: (3, 2)
+   *         [1, 9],
+   *         [9, 2]]
+   *
+   *  }</pre>
+   *
+   * @param <T> data type for {@code output()} output
+   * @param diagonal Rank `r`, where `r >= 1`
+   * @param k Diagonal offset(s). Positive value means superdiagonal, 0 refers to the main
+   *  diagonal, and negative value means subdiagonals. `k` can be a single integer
+   *  (for a single diagonal) or a pair of integers specifying the low and high ends
+   *  of a matrix band. `k[0]` must not be larger than `k[1]`.
+   * @param numRows The number of rows of the output matrix. If it is not provided, the op assumes
+   *  the output matrix is a square matrix and infers the matrix size from k and the
+   *  innermost dimension of `diagonal`.
+   * @param numCols The number of columns of the output matrix. If it is not provided, the op
+   *  assumes the output matrix is a square matrix and infers the matrix size from
+   *  k and the innermost dimension of `diagonal`.
+   * @param paddingValue The number to fill the area outside the specified diagonal band with.
+   *  Default is 0.
+   * @param options carries optional attributes values
+   * @return a new instance of MatrixDiagV3
+   */
+  public <T extends TType> MatrixDiagV3<T> matrixDiagV3(Operand<T> diagonal, Operand<TInt32> k,
+      Operand<TInt32> numRows, Operand<TInt32> numCols, Operand<T> paddingValue,
+      MatrixDiagV3.Options... options) {
+    return MatrixDiagV3.create(scope, diagonal, k, numRows, numCols, paddingValue, options);
+  }
+
+  /**
+   * Returns a batched matrix tensor with new batched diagonal values.
+   *  <p>
+   *  Given `input` and `diagonal`, this operation returns a tensor with the
+   *  same shape and values as `input`, except for the specified diagonals of the
+   *  innermost matrices. These will be overwritten by the values in `diagonal`.
+   *  <p>
+   *  `input` has `r+1` dimensions `[I, J, ..., L, M, N]`. When `k` is scalar or
+   *  `k[0] == k[1]`, `diagonal` has `r` dimensions `[I, J, ..., L, max_diag_len]`.
+   *  Otherwise, it has `r+1` dimensions `[I, J, ..., L, num_diags, max_diag_len]`.
+   *  `num_diags` is the number of diagonals, `num_diags = k[1] - k[0] + 1`.
+   *  `max_diag_len` is the longest diagonal in the range `[k[0], k[1]]`,
+   *  `max_diag_len = min(M + min(k[1], 0), N + min(-k[0], 0))`
+   *  <p>
+   *  The output is a tensor of rank `k+1` with dimensions `[I, J, ..., L, M, N]`.
+   *  If `k` is scalar or `k[0] == k[1]`:
+   *  <pre>{@code
+   *  output[i, j, ..., l, m, n]
+   *    = diagonal[i, j, ..., l, n-max(k[1], 0)] ; if n - m == k[1]
+   *      input[i, j, ..., l, m, n]              ; otherwise
+   *  }</pre>
+   *  Otherwise,
+   *  <pre>{@code
+   *  output[i, j, ..., l, m, n]
+   *    = diagonal[i, j, ..., l, diag_index, index_in_diag] ; if k[0] <= d <= k[1]
+   *      input[i, j, ..., l, m, n]                         ; otherwise
+   *  }</pre>
+   *  where `d = n - m`, `diag_index = k[1] - d`, and
+   *  `index_in_diag = n - max(d, 0) + offset`.
+   *  <p>
+   *  `offset` is zero except when the alignment of the diagonal is to the right.
+   *  <pre>{@code
+   *  offset = max_diag_len - diag_len(d) ; if (`align` in {RIGHT_LEFT, RIGHT_RIGHT}
+   *                                             and `d >= 0`) or
+   *                                           (`align` in {LEFT_RIGHT, RIGHT_RIGHT}
+   *                                             and `d <= 0`)
+   *           0                          ; otherwise
+   *  }</pre>
+   *  where `diag_len(d) = min(cols - max(d, 0), rows + min(d, 0))`.
+   *  <p>
+   *  For example:
+   *  <pre>{@code
+   *  # The main diagonal.
+   *  input = np.array([[[7, 7, 7, 7],              # Input shape: (2, 3, 4)
+   *                     [7, 7, 7, 7],
+   *                     [7, 7, 7, 7]],
+   *                    [[7, 7, 7, 7],
+   *                     [7, 7, 7, 7],
+   *                     [7, 7, 7, 7]]])
+   *  diagonal = np.array([[1, 2, 3],               # Diagonal shape: (2, 3)
+   *                       [4, 5, 6]])
+   *  tf.matrix_set_diag(input, diagonal)
+   *    ==> [[[1, 7, 7, 7],  # Output shape: (2, 3, 4)
+   *          [7, 2, 7, 7],
+   *          [7, 7, 3, 7]],
+   *         [[4, 7, 7, 7],
+   *          [7, 5, 7, 7],
+   *          [7, 7, 6, 7]]]
+   *
+   *  # A superdiagonal (per batch).
+   *  tf.matrix_set_diag(input, diagonal, k = 1)
+   *    ==> [[[7, 1, 7, 7],  # Output shape: (2, 3, 4)
+   *          [7, 7, 2, 7],
+   *          [7, 7, 7, 3]],
+   *         [[7, 4, 7, 7],
+   *          [7, 7, 5, 7],
+   *          [7, 7, 7, 6]]]
+   *
+   *  # A band of diagonals.
+   *  diagonals = np.array([[[0, 9, 1],  # Diagonal shape: (2, 4, 3)
+   *                         [6, 5, 8],
+   *                         [1, 2, 3],
+   *                         [4, 5, 0]],
+   *                        [[0, 1, 2],
+   *                         [5, 6, 4],
+   *                         [6, 1, 2],
+   *                         [3, 4, 0]]])
+   *  tf.matrix_set_diag(input, diagonals, k = (-1, 2))
+   *    ==> [[[1, 6, 9, 7],  # Output shape: (2, 3, 4)
+   *          [4, 2, 5, 1],
+   *          [7, 5, 3, 8]],
+   *         [[6, 5, 1, 7],
+   *          [3, 1, 6, 2],
+   *          [7, 4, 2, 4]]]
+   *
+   *  # LEFT_RIGHT alignment.
+   *  diagonals = np.array([[[9, 1, 0],  # Diagonal shape: (2, 4, 3)
+   *                         [6, 5, 8],
+   *                         [1, 2, 3],
+   *                         [0, 4, 5]],
+   *                        [[1, 2, 0],
+   *                         [5, 6, 4],
+   *                         [6, 1, 2],
+   *                         [0, 3, 4]]])
+   *  tf.matrix_set_diag(input, diagonals, k = (-1, 2), align="LEFT_RIGHT")
+   *    ==> [[[1, 6, 9, 7],  # Output shape: (2, 3, 4)
+   *          [4, 2, 5, 1],
+   *          [7, 5, 3, 8]],
+   *         [[6, 5, 1, 7],
+   *          [3, 1, 6, 2],
+   *          [7, 4, 2, 4]]]
+   *
+   *  }</pre>
+   *
+   * @param <T> data type for {@code output()} output
+   * @param input Rank `r+1`, where `r >= 1`.
+   * @param diagonal Rank `r` when `k` is an integer or `k[0] == k[1]`. Otherwise, it has rank `r+1`.
+   *  `k >= 1`.
+   * @param k Diagonal offset(s). Positive value means superdiagonal, 0 refers to the main
+   *  diagonal, and negative value means subdiagonals. `k` can be a single integer
+   *  (for a single diagonal) or a pair of integers specifying the low and high ends
+   *  of a matrix band. `k[0]` must not be larger than `k[1]`.
+   * @param options carries optional attributes values
+   * @return a new instance of MatrixSetDiag
+   */
+  public <T extends TType> MatrixSetDiag<T> matrixSetDiag(Operand<T> input, Operand<T> diagonal,
+      Operand<TInt32> k, MatrixSetDiag.Options... options) {
+    return MatrixSetDiag.create(scope, input, diagonal, k, options);
+  }
+
+  /**
+   * Solves one or more linear least-squares problems.
+   *  <p>
+   *  `matrix` is a tensor of shape `[..., M, N]` whose inner-most 2 dimensions
+   *  form real or complex matrices of size `[M, N]`. `Rhs` is a tensor of the same
+   *  type as `matrix` and shape `[..., M, K]`.
+   *  The output is a tensor shape `[..., N, K]` where each output matrix solves
+   *  each of the equations
+   *  `matrix[..., :, :]` * `output[..., :, :]` = `rhs[..., :, :]`
+   *  in the least squares sense.
+   *  <p>
+   *  We use the following notation for (complex) matrix and right-hand sides
+   *  in the batch:
+   *  <p>
+   *  `matrix`=\\(A \in \mathbb{C}^{m \times n}\\),
+   *  `rhs`=\\(B  \in \mathbb{C}^{m \times k}\\),
+   *  `output`=\\(X  \in \mathbb{C}^{n \times k}\\),
+   *  `l2_regularizer`=\\(\lambda \in \mathbb{R}\\).
+   *  <p>
+   *  If `fast` is `True`, then the solution is computed by solving the normal
+   *  equations using Cholesky decomposition. Specifically, if \\(m \ge n\\) then
+   *  \\(X = (A^H A + \lambda I)^{-1} A^H B\\), which solves the least-squares
+   *  problem \\(X = \mathrm{argmin}_{Z \in \Re^{n \times k} } ||A Z - B||_F^2 + \lambda ||Z||_F^2\\).
+   *  If \\(m \lt n\\) then `output` is computed as
+   *  \\(X = A^H (A A^H + \lambda I)^{-1} B\\), which (for \\(\lambda = 0\\)) is the
+   *  minimum-norm solution to the under-determined linear system, i.e.
+   *  \\(X = \mathrm{argmin}_{Z \in \mathbb{C}^{n \times k} } ||Z||_F^2 \\),
+   *  subject to \\(A Z = B\\). Notice that the fast path is only numerically stable
+   *  when \\(A\\) is numerically full rank and has a condition number
+   *  \\(\mathrm{cond}(A) \lt \frac{1}{\sqrt{\epsilon_{mach} } }\\) or \\(\lambda\\) is
+   *  sufficiently large.
+   *  <p>
+   *  If `fast` is `False` an algorithm based on the numerically robust complete
+   *  orthogonal decomposition is used. This computes the minimum-norm
+   *  least-squares solution, even when \\(A\\) is rank deficient. This path is
+   *  typically 6-7 times slower than the fast path. If `fast` is `False` then
+   *  `l2_regularizer` is ignored.
+   *
+   * @param <T> data type for {@code output()} output
+   * @param matrix Shape is `[..., M, N]`.
+   * @param rhs Shape is `[..., M, K]`.
+   * @param l2Regularizer Scalar tensor.
+   *  <p>
+   * @compatibility(numpy) Equivalent to np.linalg.lstsq
+   * @end_compatibility
+   * @param options carries optional attributes values
+   * @return a new instance of MatrixSolveLs
+   */
+  public <T extends TType> MatrixSolveLs<T> matrixSolveLs(Operand<T> matrix, Operand<T> rhs,
+      Operand<TFloat64> l2Regularizer, MatrixSolveLs.Options... options) {
+    return MatrixSolveLs.create(scope, matrix, rhs, l2Regularizer, options);
+  }
+
+  /**
+   * Computes the QR decompositions of one or more matrices.
+   *  <p>
+   *  Computes the QR decomposition of each inner matrix in `tensor` such that
+   *  `tensor[..., :, :] = q[..., :, :] * r[..., :,:])`
+   *  <pre>{@code
+   *  # a is a tensor.
+   *  # q is a tensor of orthonormal matrices.
+   *  # r is a tensor of upper triangular matrices.
+   *  q, r = qr(a)
+   *  q_full, r_full = qr(a, full_matrices=True)
+   *  }</pre>
+   *
+   * @param <T> data type for {@code q()} output
+   * @param input A tensor of shape `[..., M, N]` whose inner-most 2 dimensions
+   *  form matrices of size `[M, N]`. Let `P` be the minimum of `M` and `N`.
+   * @param options carries optional attributes values
+   * @return a new instance of Qr
+   */
+  public <T extends TType> Qr<T> qr(Operand<T> input, Qr.Options... options) {
+    return Qr.create(scope, input, options);
+  }
+
+  /**
+   * Perform a quantized matrix multiplication of  `a` by the matrix `b`.
+   *  <p>
+   *  The inputs must be two-dimensional matrices and the inner dimension of
+   *  `a` (after being transposed if `transpose_a` is non-zero) must match the
+   *  outer dimension of `b` (after being transposed if `transposed_b` is
+   *  non-zero).
+   *
+   * @param <V> data type for {@code out()} output
+   * @param a Must be a two-dimensional tensor.
+   * @param b Must be a two-dimensional tensor.
+   * @param minA The float value that the lowest quantized `a` value represents.
+   * @param maxA The float value that the highest quantized `a` value represents.
+   * @param minB The float value that the lowest quantized `b` value represents.
+   * @param maxB The float value that the highest quantized `b` value represents.
+   * @param Toutput
+   * @param Tactivation The type of output produced by activation function
+   *  following this operation.
+   * @param options carries optional attributes values
+   * @return a new instance of QuantizedMatMul
+   */
+  public <V extends TType, T extends TType, U extends TType, W extends TType> QuantizedMatMul<V> quantizedMatMul(
+      Operand<T> a, Operand<U> b, Operand<TFloat32> minA, Operand<TFloat32> maxA,
+      Operand<TFloat32> minB, Operand<TFloat32> maxB, Class<V> Toutput, Class<W> Tactivation,
+      QuantizedMatMul.Options... options) {
+    return QuantizedMatMul.create(scope, a, b, minA, maxA, minB, maxB, Toutput, Tactivation, options);
+  }
+
+  /**
+   * Computes the eigen decomposition of one or more square self-adjoint matrices.
+   *  <p>
+   *  Computes the eigenvalues and (optionally) eigenvectors of each inner matrix in
+   *  `input` such that `input[..., :, :] = v[..., :, :] * diag(e[..., :])`. The eigenvalues
+   *  are sorted in non-decreasing order.
+   *  <pre>{@code
+   *  # a is a tensor.
+   *  # e is a tensor of eigenvalues.
+   *  # v is a tensor of eigenvectors.
+   *  e, v = self_adjoint_eig(a)
+   *  e = self_adjoint_eig(a, compute_v=False)
+   *  }</pre>
+   *
+   * @param <T> data type for {@code e()} output
+   * @param input `Tensor` input of shape `[N, N]`.
+   * @param options carries optional attributes values
+   * @return a new instance of SelfAdjointEig
+   */
+  public <T extends TType> SelfAdjointEig<T> selfAdjointEig(Operand<T> input,
+      SelfAdjointEig.Options... options) {
+    return SelfAdjointEig.create(scope, input, options);
+  }
+
+  /**
+   * Solves systems of linear equations.
+   *  <p>
+   *  `Matrix` is a tensor of shape `[..., M, M]` whose inner-most 2 dimensions
+   *  form square matrices. `Rhs` is a tensor of shape `[..., M, K]`. The `output` is
+   *  a tensor shape `[..., M, K]`.  If `adjoint` is `False` then each output matrix
+   *  satisfies `matrix[..., :, :] * output[..., :, :] = rhs[..., :, :]`.
+   *  If `adjoint` is `True` then each output matrix satisfies
+   *  `adjoint(matrix[..., :, :]) * output[..., :, :] = rhs[..., :, :]`.
+   *
+   * @param <T> data type for {@code output()} output
+   * @param matrix Shape is `[..., M, M]`.
+   * @param rhs Shape is `[..., M, K]`.
+   * @param options carries optional attributes values
+   * @return a new instance of Solve
+   */
+  public <T extends TType> Solve<T> solve(Operand<T> matrix, Operand<T> rhs,
+      Solve.Options... options) {
+    return Solve.create(scope, matrix, rhs, options);
+  }
+
+  /**
+   * Computes the matrix square root of one or more square matrices:
+   *  <p>
+   *  matmul(sqrtm(A), sqrtm(A)) = A
+   *  <p>
+   *  The input matrix should be invertible. If the input matrix is real, it should
+   *  have no eigenvalues which are real and negative (pairs of complex conjugate
+   *  eigenvalues are allowed).
+   *  <p>
+   *  The matrix square root is computed by first reducing the matrix to
+   *  quasi-triangular form with the real Schur decomposition. The square root
+   *  of the quasi-triangular matrix is then computed directly. Details of
+   *  the algorithm can be found in: Nicholas J. Higham, "Computing real
+   *  square roots of a real matrix", Linear Algebra Appl., 1987.
+   *  <p>
+   *  The input is a tensor of shape `[..., M, M]` whose inner-most 2 dimensions
+   *  form square matrices. The output is a tensor of the same shape as the input
+   *  containing the matrix square root for all input submatrices `[..., :, :]`.
+   *
+   * @param <T> data type for {@code output()} output
+   * @param input Shape is `[..., M, M]`.
+   * @return a new instance of Sqrtm
+   */
+  public <T extends TType> Sqrtm<T> sqrtm(Operand<T> input) {
+    return Sqrtm.create(scope, input);
+  }
+
+  /**
+   * Computes the singular value decompositions of one or more matrices.
+   *  <p>
+   *  Computes the SVD of each inner matrix in `input` such that
+   *  `input[..., :, :] = u[..., :, :] * diag(s[..., :, :]) * transpose(v[..., :, :])`
+   *  <pre>{@code
+   *  # a is a tensor containing a batch of matrices.
+   *  # s is a tensor of singular values for each matrix.
+   *  # u is the tensor containing the left singular vectors for each matrix.
+   *  # v is the tensor containing the right singular vectors for each matrix.
+   *  s, u, v = svd(a)
+   *  s, _, _ = svd(a, compute_uv=False)
+   *  }</pre>
+   *
+   * @param <T> data type for {@code s()} output
+   * @param input A tensor of shape `[..., M, N]` whose inner-most 2 dimensions
+   *  form matrices of size `[M, N]`. Let `P` be the minimum of `M` and `N`.
+   * @param options carries optional attributes values
+   * @return a new instance of Svd
+   */
+  public <T extends TType> Svd<T> svd(Operand<T> input, Svd.Options... options) {
+    return Svd.create(scope, input, options);
+  }
+
+  /**
+   * Returns a diagonal tensor with a given diagonal values.
+   *  <p>
+   *  Given a `diagonal`, this operation returns a tensor with the `diagonal` and
+   *  everything else padded with zeros. The diagonal is computed as follows:
+   *  <p>
+   *  Assume `diagonal` has dimensions [D1,..., Dk], then the output is a tensor of
+   *  rank 2k with dimensions [D1,..., Dk, D1,..., Dk] where:
+   *  <p>
+   *  `output[i1,..., ik, i1,..., ik] = diagonal[i1, ..., ik]` and 0 everywhere else.
+   *  <p>
+   *  For example:
+   *  <pre>{@code
+   *  # 'diagonal' is [1, 2, 3, 4]
+   *  tf.diag(diagonal) ==> [[1, 0, 0, 0]
+   *                         [0, 2, 0, 0]
+   *                         [0, 0, 3, 0]
+   *                         [0, 0, 0, 4]]
+   *  }</pre>
+   *
+   * @param <T> data type for {@code output()} output
+   * @param diagonal Rank k tensor where k is at most 1.
+   * @return a new instance of TensorDiag
+   */
+  public <T extends TType> TensorDiag<T> tensorDiag(Operand<T> diagonal) {
+    return TensorDiag.create(scope, diagonal);
+  }
+
+  /**
+   * Returns the diagonal part of the tensor.
+   *  <p>
+   *  This operation returns a tensor with the `diagonal` part
+   *  of the `input`. The `diagonal` part is computed as follows:
+   *  <p>
+   *  Assume `input` has dimensions `[D1,..., Dk, D1,..., Dk]`, then the output is a
+   *  tensor of rank `k` with dimensions `[D1,..., Dk]` where:
+   *  <p>
+   *  `diagonal[i1,..., ik] = input[i1, ..., ik, i1,..., ik]`.
+   *  <p>
+   *  For example:
+   *  <pre>{@code
+   *  # 'input' is [[1, 0, 0, 0]
+   *                [0, 2, 0, 0]
+   *                [0, 0, 3, 0]
+   *                [0, 0, 0, 4]]
+   *
+   *  tf.diag_part(input) ==> [1, 2, 3, 4]
+   *  }</pre>
+   *
+   * @param <T> data type for {@code diagonal()} output
+   * @param input Rank k tensor where k is even and not zero.
+   * @return a new instance of TensorDiagPart
+   */
+  public <T extends TType> TensorDiagPart<T> tensorDiagPart(Operand<T> input) {
+    return TensorDiagPart.create(scope, input);
+  }
+
+  /**
+   * Shuffle dimensions of x according to a permutation.
+   *  <p>
+   *  The output `y` has the same rank as `x`. The shapes of `x` and `y` satisfy:
+   *    `y.shape[i] == x.shape[perm[i]] for i in [0, 1, ..., rank(x) - 1]`
+   *
+   * @param <T> data type for {@code y()} output
+   * @param x
+   * @param perm
+   * @return a new instance of Transpose
+   */
+  public <T extends TType, U extends TNumber> Transpose<T> transpose(Operand<T> x,
+      Operand<U> perm) {
+    return Transpose.create(scope, x, perm);
+  }
+
+  /**
+   * Solves systems of linear equations with upper or lower triangular matrices by backsubstitution.
+   *  <p>
+   *
+   *  `matrix` is a tensor of shape `[..., M, M]` whose inner-most 2 dimensions form
+   *  square matrices. If `lower` is `True` then the strictly upper triangular part
+   *  of each inner-most matrix is assumed to be zero and not accessed.
+   *  If `lower` is False then the strictly lower triangular part of each inner-most
+   *  matrix is assumed to be zero and not accessed.
+   *  `rhs` is a tensor of shape `[..., M, N]`.
+   *  <p>
+   *  The output is a tensor of shape `[..., M, N]`. If `adjoint` is
+   *  `True` then the innermost matrices in `output` satisfy matrix equations
+   *  `matrix[..., :, :] * output[..., :, :] = rhs[..., :, :]`.
+   *  If `adjoint` is `False` then the strictly then the  innermost matrices in
+   *  `output` satisfy matrix equations
+   *  `adjoint(matrix[..., i, k]) * output[..., k, j] = rhs[..., i, j]`.
+   *  <p>
+   *  Note, the batch shapes for the inputs only need to broadcast.
+   *  <p>
+   *  Example:
+   *  <pre>{@code
+   *  a = tf.constant([[3,  0,  0,  0],
+   *                   [2,  1,  0,  0],
+   *                   [1,  0,  1,  0],
+   *                   [1,  1,  1,  1]], dtype=tf.float32)
+   *
+   *  b = tf.constant([[4],
+   *                   [2],
+   *                   [4],
+   *                   [2]], dtype=tf.float32)
+   *
+   *  x = tf.linalg.triangular_solve(a, b, lower=True)
+   *  x
+   *  # <tf.Tensor: shape=(4, 1), dtype=float32, numpy=
+   *  # array([[ 1.3333334 ],
+   *  #        [-0.66666675],
+   *  #        [ 2.6666665 ],
+   *  #        [-1.3333331 ]], dtype=float32)>
+   *
+   *  # in python3 one can use `a@x`
+   *  tf.matmul(a, x)
+   *  # <tf.Tensor: shape=(4, 1), dtype=float32, numpy=
+   *  # array([[4.       ],
+   *  #        [2.       ],
+   *  #        [4.       ],
+   *  #        [1.9999999]], dtype=float32)>
+   *  }</pre>
+   *
+   * @param <T> data type for {@code output()} output
+   * @param matrix Shape is `[..., M, M]`.
+   * @param rhs Shape is `[..., M, K]`.
+   * @param options carries optional attributes values
+   * @return a new instance of TriangularSolve
+   */
+  public <T extends TType> TriangularSolve<T> triangularSolve(Operand<T> matrix, Operand<T> rhs,
+      TriangularSolve.Options... options) {
+    return TriangularSolve.create(scope, matrix, rhs, options);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/MathOps.java b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/MathOps.java
new file mode 100644
index 00000000000..d89679c5084
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/MathOps.java
@@ -0,0 +1,2393 @@
+// Copyright 2020 The TensorFlow Authors. All Rights Reserved.
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+// ==============================================================================
+//
+// This class has been generated, DO NOT EDIT!
+//
+package org.tensorflow.op;
+
+import org.tensorflow.Operand;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.math.Abs;
+import org.tensorflow.op.math.AccumulateN;
+import org.tensorflow.op.math.Acos;
+import org.tensorflow.op.math.Acosh;
+import org.tensorflow.op.math.Add;
+import org.tensorflow.op.math.AddN;
+import org.tensorflow.op.math.Angle;
+import org.tensorflow.op.math.ApproximateEqual;
+import org.tensorflow.op.math.ArgMax;
+import org.tensorflow.op.math.ArgMin;
+import org.tensorflow.op.math.Asin;
+import org.tensorflow.op.math.Asinh;
+import org.tensorflow.op.math.Atan;
+import org.tensorflow.op.math.Atan2;
+import org.tensorflow.op.math.Atanh;
+import org.tensorflow.op.math.Betainc;
+import org.tensorflow.op.math.Bincount;
+import org.tensorflow.op.math.Ceil;
+import org.tensorflow.op.math.CompareAndBitpack;
+import org.tensorflow.op.math.ComplexAbs;
+import org.tensorflow.op.math.Conj;
+import org.tensorflow.op.math.Cos;
+import org.tensorflow.op.math.Cosh;
+import org.tensorflow.op.math.Cumprod;
+import org.tensorflow.op.math.Cumsum;
+import org.tensorflow.op.math.DenseBincount;
+import org.tensorflow.op.math.Digamma;
+import org.tensorflow.op.math.Div;
+import org.tensorflow.op.math.DivNoNan;
+import org.tensorflow.op.math.Equal;
+import org.tensorflow.op.math.Erf;
+import org.tensorflow.op.math.Erfc;
+import org.tensorflow.op.math.Exp;
+import org.tensorflow.op.math.Expm1;
+import org.tensorflow.op.math.Fact;
+import org.tensorflow.op.math.Floor;
+import org.tensorflow.op.math.FloorDiv;
+import org.tensorflow.op.math.FloorMod;
+import org.tensorflow.op.math.Greater;
+import org.tensorflow.op.math.GreaterEqual;
+import org.tensorflow.op.math.Igamma;
+import org.tensorflow.op.math.Igammac;
+import org.tensorflow.op.math.Imag;
+import org.tensorflow.op.math.InvertPermutation;
+import org.tensorflow.op.math.IsFinite;
+import org.tensorflow.op.math.IsInf;
+import org.tensorflow.op.math.IsNan;
+import org.tensorflow.op.math.Less;
+import org.tensorflow.op.math.LessEqual;
+import org.tensorflow.op.math.Lgamma;
+import org.tensorflow.op.math.Log;
+import org.tensorflow.op.math.Log1p;
+import org.tensorflow.op.math.LogicalAnd;
+import org.tensorflow.op.math.LogicalNot;
+import org.tensorflow.op.math.LogicalOr;
+import org.tensorflow.op.math.Maximum;
+import org.tensorflow.op.math.Mean;
+import org.tensorflow.op.math.Minimum;
+import org.tensorflow.op.math.Mod;
+import org.tensorflow.op.math.Mul;
+import org.tensorflow.op.math.MulNoNan;
+import org.tensorflow.op.math.Ndtri;
+import org.tensorflow.op.math.Neg;
+import org.tensorflow.op.math.NextAfter;
+import org.tensorflow.op.math.NotEqual;
+import org.tensorflow.op.math.Polygamma;
+import org.tensorflow.op.math.PopulationCount;
+import org.tensorflow.op.math.Pow;
+import org.tensorflow.op.math.QuantizedAdd;
+import org.tensorflow.op.math.QuantizedMul;
+import org.tensorflow.op.math.Real;
+import org.tensorflow.op.math.RealDiv;
+import org.tensorflow.op.math.Reciprocal;
+import org.tensorflow.op.math.Rint;
+import org.tensorflow.op.math.Round;
+import org.tensorflow.op.math.Rsqrt;
+import org.tensorflow.op.math.SegmentMax;
+import org.tensorflow.op.math.SegmentMean;
+import org.tensorflow.op.math.SegmentMin;
+import org.tensorflow.op.math.SegmentProd;
+import org.tensorflow.op.math.SegmentSum;
+import org.tensorflow.op.math.Sigmoid;
+import org.tensorflow.op.math.Sign;
+import org.tensorflow.op.math.Sin;
+import org.tensorflow.op.math.Sinh;
+import org.tensorflow.op.math.Softplus;
+import org.tensorflow.op.math.Sqrt;
+import org.tensorflow.op.math.Square;
+import org.tensorflow.op.math.SquaredDifference;
+import org.tensorflow.op.math.Sub;
+import org.tensorflow.op.math.Tan;
+import org.tensorflow.op.math.Tanh;
+import org.tensorflow.op.math.TruncateDiv;
+import org.tensorflow.op.math.TruncateMod;
+import org.tensorflow.op.math.UnsortedSegmentMax;
+import org.tensorflow.op.math.UnsortedSegmentMin;
+import org.tensorflow.op.math.UnsortedSegmentProd;
+import org.tensorflow.op.math.UnsortedSegmentSum;
+import org.tensorflow.op.math.Xdivy;
+import org.tensorflow.op.math.Xlog1py;
+import org.tensorflow.op.math.Xlogy;
+import org.tensorflow.op.math.Zeta;
+import org.tensorflow.op.math.erfinv;
+import org.tensorflow.types.TBool;
+import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * An API for building {@code math} operations as {@link Op Op}s
+ *
+ * @see {@link Ops}
+ */
+public final class MathOps {
+  private final Scope scope;
+
+  MathOps(Scope scope) {
+    this.scope = scope;
+  }
+
+  /**
+   * Computes the absolute value of a tensor.
+   *  <p>
+   *  Given a tensor `x`, this operation returns a tensor containing the absolute
+   *  value of each element in `x`. For example, if x is an input element and y is
+   *  an output element, this operation computes \\(y = |x|\\).
+   *
+   * @param <T> data type for {@code y()} output
+   * @param x
+   * @return a new instance of Abs
+   */
+  public <T extends TNumber> Abs<T> abs(Operand<T> x) {
+    return Abs.create(scope, x);
+  }
+
+  /**
+   * Returns the element-wise sum of a list of tensors.
+   *  <p>
+   *  `tf.accumulate_n_v2` performs the same operation as `tf.add_n`, but does not
+   *  wait for all of its inputs to be ready before beginning to sum. This can
+   *  save memory if inputs are ready at different times, since minimum temporary
+   *  storage is proportional to the output size rather than the inputs size.
+   *  <p>
+   *  Unlike the original `accumulate_n`, `accumulate_n_v2` is differentiable.
+   *  <p>
+   *  Returns a `Tensor` of same shape and type as the elements of `inputs`.
+   *
+   * @param <T> data type for {@code sum()} output
+   * @param inputs A list of `Tensor` objects, each with same shape and type.
+   * @param shape Shape of elements of `inputs`.
+   * @return a new instance of AccumulateN
+   */
+  public <T extends TType> AccumulateN<T> accumulateN(Iterable<Operand<T>> inputs, Shape shape) {
+    return AccumulateN.create(scope, inputs, shape);
+  }
+
+  /**
+   * Computes acos of x element-wise.
+   *
+   * @param <T> data type for {@code y()} output
+   * @param x
+   * @return a new instance of Acos
+   */
+  public <T extends TType> Acos<T> acos(Operand<T> x) {
+    return Acos.create(scope, x);
+  }
+
+  /**
+   * Computes inverse hyperbolic cosine of x element-wise.
+   *  <p>
+   *  Given an input tensor, the function computes inverse hyperbolic cosine of every element.
+   *  Input range is `[1, inf]`. It returns `nan` if the input lies outside the range.
+   *  <pre>{@code
+   *  x = tf.constant([-2, -0.5, 1, 1.2, 200, 10000, float("inf")])
+   *  tf.math.acosh(x) ==> [nan nan 0. 0.62236255 5.9914584 9.903487 inf]
+   *  }</pre>
+   *
+   * @param <T> data type for {@code y()} output
+   * @param x
+   * @return a new instance of Acosh
+   */
+  public <T extends TType> Acosh<T> acosh(Operand<T> x) {
+    return Acosh.create(scope, x);
+  }
+
+  /**
+   * Returns x + y element-wise.
+   *  <p>
+   *  <i>NOTE</i>: `math.Add` supports broadcasting. `AddN` does not. More about broadcasting
+   *  [here](http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html)
+   *
+   * @param <T> data type for {@code z()} output
+   * @param x
+   * @param y
+   * @return a new instance of Add
+   */
+  public <T extends TType> Add<T> add(Operand<T> x, Operand<T> y) {
+    return Add.create(scope, x, y);
+  }
+
+  /**
+   * Add all input tensors element wise.
+   *  <p>
+   *    Inputs must be of same size and shape.
+   *  <p>
+   *    <pre>{@code
+   *    x = [9, 7, 10]
+   *    tf.math.add_n(x) ==> 26
+   *    }</pre>
+   *
+   * @param <T> data type for {@code sum()} output
+   * @param inputs
+   * @return a new instance of AddN
+   */
+  public <T extends TType> AddN<T> addN(Iterable<Operand<T>> inputs) {
+    return AddN.create(scope, inputs);
+  }
+
+  /**
+   * Returns the argument of a complex number.
+   *  <p>
+   *  Given a tensor `input` of complex numbers, this operation returns a tensor of
+   *  type `float` that is the argument of each element in `input`. All elements in
+   *  `input` must be complex numbers of the form \\(a + bj\\), where <i>a</i>
+   *  is the real part and <i>b</i> is the imaginary part.
+   *  <p>
+   *  The argument returned by this operation is of the form \\(atan2(b, a)\\).
+   *  <p>
+   *  For example:
+   *  <pre>{@code
+   *  # tensor 'input' is [-2.25 + 4.75j, 3.25 + 5.75j]
+   *  tf.angle(input) ==> [2.0132, 1.056]
+   *  }</pre>
+   *
+   * @compatibility(numpy) Equivalent to np.angle.
+   * @end_compatibility
+   * @param <U> data type for {@code output()} output
+   * @param input
+   * @return a new instance of Angle
+   */
+  public <T extends TType> Angle<TFloat32> angle(Operand<T> input) {
+    return Angle.create(scope, input);
+  }
+
+  /**
+   * Returns the argument of a complex number.
+   *  <p>
+   *  Given a tensor `input` of complex numbers, this operation returns a tensor of
+   *  type `float` that is the argument of each element in `input`. All elements in
+   *  `input` must be complex numbers of the form \\(a + bj\\), where <i>a</i>
+   *  is the real part and <i>b</i> is the imaginary part.
+   *  <p>
+   *  The argument returned by this operation is of the form \\(atan2(b, a)\\).
+   *  <p>
+   *  For example:
+   *  <pre>{@code
+   *  # tensor 'input' is [-2.25 + 4.75j, 3.25 + 5.75j]
+   *  tf.angle(input) ==> [2.0132, 1.056]
+   *  }</pre>
+   *
+   * @compatibility(numpy) Equivalent to np.angle.
+   * @end_compatibility
+   * @param <U> data type for {@code output()} output
+   * @param input
+   * @param Tout
+   * @return a new instance of Angle
+   */
+  public <U extends TNumber, T extends TType> Angle<U> angle(Operand<T> input, Class<U> Tout) {
+    return Angle.create(scope, input, Tout);
+  }
+
+  /**
+   * Returns the truth value of abs(x-y) < tolerance element-wise.
+   *
+   * @param x
+   * @param y
+   * @param options carries optional attributes values
+   * @return a new instance of ApproximateEqual
+   */
+  public <T extends TType> ApproximateEqual approximateEqual(Operand<T> x, Operand<T> y,
+      ApproximateEqual.Options... options) {
+    return ApproximateEqual.create(scope, x, y, options);
+  }
+
+  /**
+   * Returns the index with the largest value across dimensions of a tensor.
+   *  <p>
+   *  Note that in case of ties the identity of the return value is not guaranteed.
+   *  <p>
+   *  Usage:
+   *    <pre>{@code
+   *    import tensorflow as tf
+   *    a = [1, 10, 26.9, 2.8, 166.32, 62.3]
+   *    b = tf.math.argmax(input = a)
+   *    c = tf.keras.backend.eval(b)
+   *    # c = 4
+   *    # here a[4] = 166.32 which is the largest element of a across axis 0
+   *    }</pre>
+   *
+   * @param <V> data type for {@code output()} output
+   * @param input
+   * @param dimension int32 or int64, must be in the range `[-rank(input), rank(input))`.
+   *  Describes which dimension of the input Tensor to reduce across. For vectors,
+   *  use dimension = 0.
+   * @return a new instance of ArgMax
+   */
+  public <T extends TType, U extends TNumber> ArgMax<TInt64> argMax(Operand<T> input,
+      Operand<U> dimension) {
+    return ArgMax.create(scope, input, dimension);
+  }
+
+  /**
+   * Returns the index with the largest value across dimensions of a tensor.
+   *  <p>
+   *  Note that in case of ties the identity of the return value is not guaranteed.
+   *  <p>
+   *  Usage:
+   *    <pre>{@code
+   *    import tensorflow as tf
+   *    a = [1, 10, 26.9, 2.8, 166.32, 62.3]
+   *    b = tf.math.argmax(input = a)
+   *    c = tf.keras.backend.eval(b)
+   *    # c = 4
+   *    # here a[4] = 166.32 which is the largest element of a across axis 0
+   *    }</pre>
+   *
+   * @param <V> data type for {@code output()} output
+   * @param input
+   * @param dimension int32 or int64, must be in the range `[-rank(input), rank(input))`.
+   *  Describes which dimension of the input Tensor to reduce across. For vectors,
+   *  use dimension = 0.
+   * @param outputType
+   * @return a new instance of ArgMax
+   */
+  public <V extends TNumber, T extends TType, U extends TNumber> ArgMax<V> argMax(Operand<T> input,
+      Operand<U> dimension, Class<V> outputType) {
+    return ArgMax.create(scope, input, dimension, outputType);
+  }
+
+  /**
+   * Returns the index with the smallest value across dimensions of a tensor.
+   *  <p>
+   *  Note that in case of ties the identity of the return value is not guaranteed.
+   *  <p>
+   *  Usage:
+   *    <pre>{@code
+   *    import tensorflow as tf
+   *    a = [1, 10, 26.9, 2.8, 166.32, 62.3]
+   *    b = tf.math.argmin(input = a)
+   *    c = tf.keras.backend.eval(b)
+   *    # c = 0
+   *    # here a[0] = 1 which is the smallest element of a across axis 0
+   *    }</pre>
+   *
+   * @param <V> data type for {@code output()} output
+   * @param input
+   * @param dimension int32 or int64, must be in the range `[-rank(input), rank(input))`.
+   *  Describes which dimension of the input Tensor to reduce across. For vectors,
+   *  use dimension = 0.
+   * @return a new instance of ArgMin
+   */
+  public <T extends TType, U extends TNumber> ArgMin<TInt64> argMin(Operand<T> input,
+      Operand<U> dimension) {
+    return ArgMin.create(scope, input, dimension);
+  }
+
+  /**
+   * Returns the index with the smallest value across dimensions of a tensor.
+   *  <p>
+   *  Note that in case of ties the identity of the return value is not guaranteed.
+   *  <p>
+   *  Usage:
+   *    <pre>{@code
+   *    import tensorflow as tf
+   *    a = [1, 10, 26.9, 2.8, 166.32, 62.3]
+   *    b = tf.math.argmin(input = a)
+   *    c = tf.keras.backend.eval(b)
+   *    # c = 0
+   *    # here a[0] = 1 which is the smallest element of a across axis 0
+   *    }</pre>
+   *
+   * @param <V> data type for {@code output()} output
+   * @param input
+   * @param dimension int32 or int64, must be in the range `[-rank(input), rank(input))`.
+   *  Describes which dimension of the input Tensor to reduce across. For vectors,
+   *  use dimension = 0.
+   * @param outputType
+   * @return a new instance of ArgMin
+   */
+  public <V extends TNumber, T extends TType, U extends TNumber> ArgMin<V> argMin(Operand<T> input,
+      Operand<U> dimension, Class<V> outputType) {
+    return ArgMin.create(scope, input, dimension, outputType);
+  }
+
+  /**
+   * Computes the trignometric inverse sine of x element-wise.
+   *  <p>
+   *  The `tf.math.asin` operation returns the inverse of `tf.math.sin`, such that
+   *  if `y = tf.math.sin(x)` then, `x = tf.math.asin(y)`.
+   *  <p>
+   *  <b>Note</b>: The output of `tf.math.asin` will lie within the invertible range
+   *  of sine, i.e [-pi/2, pi/2].
+   *  <p>
+   *  For example:
+   *  <pre>{@code
+   *  # Note: [1.047, 0.785] ~= [(pi/3), (pi/4)]
+   *  x = tf.constant([1.047, 0.785])
+   *  y = tf.math.sin(x) # [0.8659266, 0.7068252]
+   *
+   *  tf.math.asin(y) # [1.047, 0.785] = x
+   *  }</pre>
+   *
+   * @param <T> data type for {@code y()} output
+   * @param x
+   * @return a new instance of Asin
+   */
+  public <T extends TType> Asin<T> asin(Operand<T> x) {
+    return Asin.create(scope, x);
+  }
+
+  /**
+   * Computes inverse hyperbolic sine of x element-wise.
+   *  <p>
+   *    Given an input tensor, this function computes inverse hyperbolic sine
+   *    for every element in the tensor. Both input and output has a range of
+   *    `[-inf, inf]`.
+   *  <p>
+   *    <pre>{@code
+   *    x = tf.constant([-float("inf"), -2, -0.5, 1, 1.2, 200, 10000, float("inf")])
+   *    tf.math.asinh(x) ==> [-inf -1.4436355 -0.4812118 0.8813736 1.0159732 5.991471 9.903487 inf]
+   *    }</pre>
+   *
+   * @param <T> data type for {@code y()} output
+   * @param x
+   * @return a new instance of Asinh
+   */
+  public <T extends TType> Asinh<T> asinh(Operand<T> x) {
+    return Asinh.create(scope, x);
+  }
+
+  /**
+   * Computes the trignometric inverse tangent of x element-wise.
+   *  <p>
+   *  The `tf.math.atan` operation returns the inverse of `tf.math.tan`, such that
+   *  if `y = tf.math.tan(x)` then, `x = tf.math.atan(y)`.
+   *  <p>
+   *  <b>Note</b>: The output of `tf.math.atan` will lie within the invertible range
+   *  of tan, i.e (-pi/2, pi/2).
+   *  <p>
+   *  For example:
+   *  <pre>{@code
+   *  # Note: [1.047, 0.785] ~= [(pi/3), (pi/4)]
+   *  x = tf.constant([1.047, 0.785])
+   *  y = tf.math.tan(x) # [1.731261, 0.99920404]
+   *
+   *  tf.math.atan(y) # [1.047, 0.785] = x
+   *  }</pre>
+   *
+   * @param <T> data type for {@code y()} output
+   * @param x
+   * @return a new instance of Atan
+   */
+  public <T extends TType> Atan<T> atan(Operand<T> x) {
+    return Atan.create(scope, x);
+  }
+
+  /**
+   * Computes arctangent of `y/x` element-wise, respecting signs of the arguments.
+   *  <p>
+   *  This is the angle \( \theta \in [-\pi, \pi] \) such that
+   *  \[ x = r \cos(\theta) \]
+   *  and
+   *  \[ y = r \sin(\theta) \]
+   *  where \(r = \sqrt(x^2 + y^2) \).
+   *
+   * @param <T> data type for {@code z()} output
+   * @param y
+   * @param x
+   * @return a new instance of Atan2
+   */
+  public <T extends TNumber> Atan2<T> atan2(Operand<T> y, Operand<T> x) {
+    return Atan2.create(scope, y, x);
+  }
+
+  /**
+   * Computes inverse hyperbolic tangent of x element-wise.
+   *  <p>
+   *    Given an input tensor, this function computes inverse hyperbolic tangent
+   *    for every element in the tensor. Input range is `[-1,1]` and output range is
+   *    `[-inf, inf]`. If input is `-1`, output will be `-inf` and if the
+   *    input is `1`, output will be `inf`. Values outside the range will have
+   *    `nan` as output.
+   *  <p>
+   *    <pre>{@code
+   *    x = tf.constant([-float("inf"), -1, -0.5, 1, 0, 0.5, 10, float("inf")])
+   *    tf.math.atanh(x) ==> [nan -inf -0.54930615 inf  0. 0.54930615 nan nan]
+   *    }</pre>
+   *
+   * @param <T> data type for {@code y()} output
+   * @param x
+   * @return a new instance of Atanh
+   */
+  public <T extends TType> Atanh<T> atanh(Operand<T> x) {
+    return Atanh.create(scope, x);
+  }
+
+  /**
+   * Compute the regularized incomplete beta integral \\(I_x(a, b)\\).
+   *  <p>
+   *  The regularized incomplete beta integral is defined as:
+   *  <p>
+   *  \\(I_x(a, b) = \frac{B(x; a, b)}{B(a, b)}\\)
+   *  <p>
+   *  where
+   *  <p>
+   *  \\(B(x; a, b) = \int_0^x t^{a-1} (1 - t)^{b-1} dt\\)
+   *  <p>
+   *  is the incomplete beta function and \\(B(a, b)\\) is the <i>complete</i>
+   *  beta function.
+   *
+   * @param <T> data type for {@code z()} output
+   * @param a
+   * @param b
+   * @param x
+   * @return a new instance of Betainc
+   */
+  public <T extends TNumber> Betainc<T> betainc(Operand<T> a, Operand<T> b, Operand<T> x) {
+    return Betainc.create(scope, a, b, x);
+  }
+
+  /**
+   * Counts the number of occurrences of each value in an integer array.
+   *  <p>
+   *  Outputs a vector with length `size` and the same dtype as `weights`. If
+   *  `weights` are empty, then index `i` stores the number of times the value `i` is
+   *  counted in `arr`. If `weights` are non-empty, then index `i` stores the sum of
+   *  the value in `weights` at each index where the corresponding value in `arr` is
+   *  `i`.
+   *  <p>
+   *  Values in `arr` outside of the range [0, size) are ignored.
+   *
+   * @param <T> data type for {@code bins()} output
+   * @param arr int32 `Tensor`.
+   * @param size non-negative int32 scalar `Tensor`.
+   * @param weights is an int32, int64, float32, or float64 `Tensor` with the same
+   *  shape as `arr`, or a length-0 `Tensor`, in which case it acts as all weights
+   *  equal to 1.
+   * @return a new instance of Bincount
+   */
+  public <T extends TNumber> Bincount<T> bincount(Operand<TInt32> arr, Operand<TInt32> size,
+      Operand<T> weights) {
+    return Bincount.create(scope, arr, size, weights);
+  }
+
+  /**
+   * Returns element-wise smallest integer not less than x.
+   *
+   * @param <T> data type for {@code y()} output
+   * @param x
+   * @return a new instance of Ceil
+   */
+  public <T extends TNumber> Ceil<T> ceil(Operand<T> x) {
+    return Ceil.create(scope, x);
+  }
+
+  /**
+   * Compare values of `input` to `threshold` and pack resulting bits into a `uint8`.
+   *  <p>
+   *  Each comparison returns a boolean `true` (if `input_value > threshold`)
+   *  or and `false` otherwise.
+   *  <p>
+   *  This operation is useful for Locality-Sensitive-Hashing (LSH) and other
+   *  algorithms that use hashing approximations of cosine and `L2` distances;
+   *  codes can be generated from an input via:
+   *  <pre>{@code
+   *  codebook_size = 50
+   *  codebook_bits = codebook_size * 32
+   *  codebook = tf.get_variable('codebook', [x.shape[-1].value, codebook_bits],
+   *                             dtype=x.dtype,
+   *                             initializer=tf.orthogonal_initializer())
+   *  codes = compare_and_threshold(tf.matmul(x, codebook), threshold=0.)
+   *  codes = tf.bitcast(codes, tf.int32)  # go from uint8 to int32
+   *  # now codes has shape x.shape[:-1] + [codebook_size]
+   *  }</pre>
+   *  <b>NOTE</b>: Currently, the innermost dimension of the tensor must be divisible
+   *  by 8.
+   *  <p>
+   *  Given an `input` shaped `[s0, s1, ..., s_n]`, the output is
+   *  a `uint8` tensor shaped `[s0, s1, ..., s_n / 8]`.
+   *
+   * @param input Values to compare against `threshold` and bitpack.
+   * @param threshold Threshold to compare against.
+   * @return a new instance of CompareAndBitpack
+   */
+  public <T extends TType> CompareAndBitpack compareAndBitpack(Operand<T> input,
+      Operand<T> threshold) {
+    return CompareAndBitpack.create(scope, input, threshold);
+  }
+
+  /**
+   * Computes the complex absolute value of a tensor.
+   *  <p>
+   *  Given a tensor `x` of complex numbers, this operation returns a tensor of type
+   *  `float` or `double` that is the absolute value of each element in `x`. All
+   *  elements in `x` must be complex numbers of the form \\(a + bj\\). The absolute
+   *  value is computed as \\( \sqrt{a^2 + b^2}\\).
+   *
+   * @param <U> data type for {@code y()} output
+   * @param x
+   * @return a new instance of ComplexAbs
+   */
+  public <T extends TType> ComplexAbs<TFloat32> complexAbs(Operand<T> x) {
+    return ComplexAbs.create(scope, x);
+  }
+
+  /**
+   * Computes the complex absolute value of a tensor.
+   *  <p>
+   *  Given a tensor `x` of complex numbers, this operation returns a tensor of type
+   *  `float` or `double` that is the absolute value of each element in `x`. All
+   *  elements in `x` must be complex numbers of the form \\(a + bj\\). The absolute
+   *  value is computed as \\( \sqrt{a^2 + b^2}\\).
+   *
+   * @param <U> data type for {@code y()} output
+   * @param x
+   * @param Tout
+   * @return a new instance of ComplexAbs
+   */
+  public <U extends TNumber, T extends TType> ComplexAbs<U> complexAbs(Operand<T> x,
+      Class<U> Tout) {
+    return ComplexAbs.create(scope, x, Tout);
+  }
+
+  /**
+   * Returns the complex conjugate of a complex number.
+   *  <p>
+   *  Given a tensor `input` of complex numbers, this operation returns a tensor of
+   *  complex numbers that are the complex conjugate of each element in `input`. The
+   *  complex numbers in `input` must be of the form \\(a + bj\\), where <i>a</i> is the
+   *  real part and <i>b</i> is the imaginary part.
+   *  <p>
+   *  The complex conjugate returned by this operation is of the form \\(a - bj\\).
+   *  <p>
+   *  For example:
+   *  <pre>{@code
+   *  # tensor 'input' is [-2.25 + 4.75j, 3.25 + 5.75j]
+   *  tf.conj(input) ==> [-2.25 - 4.75j, 3.25 - 5.75j]
+   *  }</pre>
+   *
+   * @param <T> data type for {@code output()} output
+   * @param input
+   * @return a new instance of Conj
+   */
+  public <T extends TType> Conj<T> conj(Operand<T> input) {
+    return Conj.create(scope, input);
+  }
+
+  /**
+   * Computes cos of x element-wise.
+   *  <p>
+   *    Given an input tensor, this function computes cosine of every
+   *    element in the tensor. Input range is `(-inf, inf)` and
+   *    output range is `[-1,1]`. If input lies outside the boundary, `nan`
+   *    is returned.
+   *  <p>
+   *    <pre>{@code
+   *    x = tf.constant([-float("inf"), -9, -0.5, 1, 1.2, 200, 10000, float("inf")])
+   *    tf.math.cos(x) ==> [nan -0.91113025 0.87758255 0.5403023 0.36235774 0.48718765 -0.95215535 nan]
+   *    }</pre>
+   *
+   * @param <T> data type for {@code y()} output
+   * @param x
+   * @return a new instance of Cos
+   */
+  public <T extends TType> Cos<T> cos(Operand<T> x) {
+    return Cos.create(scope, x);
+  }
+
+  /**
+   * Computes hyperbolic cosine of x element-wise.
+   *  <p>
+   *    Given an input tensor, this function computes hyperbolic cosine of every
+   *    element in the tensor. Input range is `[-inf, inf]` and output range
+   *    is `[1, inf]`.
+   *  <p>
+   *    <pre>{@code
+   *    x = tf.constant([-float("inf"), -9, -0.5, 1, 1.2, 2, 10, float("inf")])
+   *    tf.math.cosh(x) ==> [inf 4.0515420e+03 1.1276259e+00 1.5430807e+00 1.8106556e+00 3.7621956e+00 1.1013233e+04 inf]
+   *    }</pre>
+   *
+   * @param <T> data type for {@code y()} output
+   * @param x
+   * @return a new instance of Cosh
+   */
+  public <T extends TType> Cosh<T> cosh(Operand<T> x) {
+    return Cosh.create(scope, x);
+  }
+
+  /**
+   * Compute the cumulative product of the tensor `x` along `axis`.
+   *  <p>
+   *  By default, this op performs an inclusive cumprod, which means that the first
+   *  element of the input is identical to the first element of the output:
+   *  <pre>{@code
+   *  tf.cumprod([a, b, c])  # => [a, a * b, a * b * c]
+   *  }</pre>
+   *  By setting the `exclusive` kwarg to `True`, an exclusive cumprod is
+   *  performed instead:
+   *  <pre>{@code
+   *  tf.cumprod([a, b, c], exclusive=True)  # => [1, a, a * b]
+   *  }</pre>
+   *  By setting the `reverse` kwarg to `True`, the cumprod is performed in the
+   *  opposite direction:
+   *  <pre>{@code
+   *  tf.cumprod([a, b, c], reverse=True)  # => [a * b * c, b * c, c]
+   *  }</pre>
+   *  This is more efficient than using separate `tf.reverse` ops.
+   *  <p>
+   *  The `reverse` and `exclusive` kwargs can also be combined:
+   *  <pre>{@code
+   *  tf.cumprod([a, b, c], exclusive=True, reverse=True)  # => [b * c, c, 1]
+   *  }</pre>
+   *
+   * @param <T> data type for {@code out()} output
+   * @param x A `Tensor`. Must be one of the following types: `float32`, `float64`,
+   *  `int64`, `int32`, `uint8`, `uint16`, `int16`, `int8`, `complex64`,
+   *  `complex128`, `qint8`, `quint8`, `qint32`, `half`.
+   * @param axis A `Tensor` of type `int32` (default: 0). Must be in the range
+   *  `[-rank(x), rank(x))`.
+   * @param options carries optional attributes values
+   * @return a new instance of Cumprod
+   */
+  public <T extends TType, U extends TNumber> Cumprod<T> cumprod(Operand<T> x, Operand<U> axis,
+      Cumprod.Options... options) {
+    return Cumprod.create(scope, x, axis, options);
+  }
+
+  /**
+   * Compute the cumulative sum of the tensor `x` along `axis`.
+   *  <p>
+   *  By default, this op performs an inclusive cumsum, which means that the first
+   *  element of the input is identical to the first element of the output:
+   *  <pre>{@code
+   *  tf.cumsum([a, b, c])  # => [a, a + b, a + b + c]
+   *  }</pre>
+   *  By setting the `exclusive` kwarg to `True`, an exclusive cumsum is
+   *  performed instead:
+   *  <pre>{@code
+   *  tf.cumsum([a, b, c], exclusive=True)  # => [0, a, a + b]
+   *  }</pre>
+   *  By setting the `reverse` kwarg to `True`, the cumsum is performed in the
+   *  opposite direction:
+   *  <pre>{@code
+   *  tf.cumsum([a, b, c], reverse=True)  # => [a + b + c, b + c, c]
+   *  }</pre>
+   *  This is more efficient than using separate `tf.reverse` ops.
+   *  <p>
+   *  The `reverse` and `exclusive` kwargs can also be combined:
+   *  <pre>{@code
+   *  tf.cumsum([a, b, c], exclusive=True, reverse=True)  # => [b + c, c, 0]
+   *  }</pre>
+   *
+   * @param <T> data type for {@code out()} output
+   * @param x A `Tensor`. Must be one of the following types: `float32`, `float64`,
+   *  `int64`, `int32`, `uint8`, `uint16`, `int16`, `int8`, `complex64`,
+   *  `complex128`, `qint8`, `quint8`, `qint32`, `half`.
+   * @param axis A `Tensor` of type `int32` (default: 0). Must be in the range
+   *  `[-rank(x), rank(x))`.
+   * @param options carries optional attributes values
+   * @return a new instance of Cumsum
+   */
+  public <T extends TType, U extends TNumber> Cumsum<T> cumsum(Operand<T> x, Operand<U> axis,
+      Cumsum.Options... options) {
+    return Cumsum.create(scope, x, axis, options);
+  }
+
+  /**
+   * Counts the number of occurrences of each value in an integer array.
+   *  <p>
+   *  Outputs a vector with length `size` and the same dtype as `weights`. If
+   *  `weights` are empty, then index `i` stores the number of times the value `i` is
+   *  counted in `arr`. If `weights` are non-empty, then index `i` stores the sum of
+   *  the value in `weights` at each index where the corresponding value in `arr` is
+   *  `i`.
+   *  <p>
+   *  Values in `arr` outside of the range [0, size) are ignored.
+   *
+   * @param <U> data type for {@code output()} output
+   * @param input 1D or 2D int `Tensor`.
+   * @param size non-negative int scalar `Tensor`.
+   * @param weights is an int32, int64, float32, or float64 `Tensor` with the same
+   *  shape as `arr`, or a length-0 `Tensor`, in which case it acts as all weights
+   *  equal to 1.
+   * @param options carries optional attributes values
+   * @return a new instance of DenseBincount
+   */
+  public <U extends TNumber, T extends TNumber> DenseBincount<U> denseBincount(Operand<T> input,
+      Operand<T> size, Operand<U> weights, DenseBincount.Options... options) {
+    return DenseBincount.create(scope, input, size, weights, options);
+  }
+
+  /**
+   * Computes Psi, the derivative of Lgamma (the log of the absolute value of
+   *  <p>
+   *  `Gamma(x)`), element-wise.
+   *
+   * @param <T> data type for {@code y()} output
+   * @param x
+   * @return a new instance of Digamma
+   */
+  public <T extends TNumber> Digamma<T> digamma(Operand<T> x) {
+    return Digamma.create(scope, x);
+  }
+
+  /**
+   * Returns x / y element-wise.
+   *  <p>
+   *  <i>NOTE</i>: `math.Div` supports broadcasting. More about broadcasting
+   *  [here](http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html)
+   *
+   * @param <T> data type for {@code z()} output
+   * @param x
+   * @param y
+   * @return a new instance of Div
+   */
+  public <T extends TType> Div<T> div(Operand<T> x, Operand<T> y) {
+    return Div.create(scope, x, y);
+  }
+
+  /**
+   * Returns 0 if the denominator is zero.
+   *  <p>
+   *
+   *  <i>NOTE</i>: `math.DivNoNan` supports broadcasting. More about broadcasting
+   *  [here](http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html)
+   *
+   * @param <T> data type for {@code z()} output
+   * @param x
+   * @param y
+   * @return a new instance of DivNoNan
+   */
+  public <T extends TType> DivNoNan<T> divNoNan(Operand<T> x, Operand<T> y) {
+    return DivNoNan.create(scope, x, y);
+  }
+
+  /**
+   * Returns the truth value of (x == y) element-wise.
+   *  <p>
+   *  <i>NOTE</i>: `math.Equal` supports broadcasting. More about broadcasting
+   *  [here](http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html)
+   *  <pre>{@code
+   *  x = tf.constant([2, 4])
+   *  y = tf.constant(2)
+   *  tf.math.equal(x, y) ==> array([True, False])
+   *
+   *  x = tf.constant([2, 4])
+   *  y = tf.constant([2, 4])
+   *  tf.math.equal(x, y) ==> array([True,  True])
+   *  }</pre>
+   *
+   * @param x
+   * @param y
+   * @param options carries optional attributes values
+   * @return a new instance of Equal
+   */
+  public <T extends TType> Equal equal(Operand<T> x, Operand<T> y, Equal.Options... options) {
+    return Equal.create(scope, x, y, options);
+  }
+
+  /**
+   * Computes the Gauss error function of `x` element-wise.
+   *
+   * @param <T> data type for {@code y()} output
+   * @param x
+   * @return a new instance of Erf
+   */
+  public <T extends TNumber> Erf<T> erf(Operand<T> x) {
+    return Erf.create(scope, x);
+  }
+
+  /**
+   * Computes the complementary error function of `x` element-wise.
+   *
+   * @param <T> data type for {@code y()} output
+   * @param x
+   * @return a new instance of Erfc
+   */
+  public <T extends TNumber> Erfc<T> erfc(Operand<T> x) {
+    return Erfc.create(scope, x);
+  }
+
+  /**
+   *
+   * @param <T> data type for {@code y()} output
+   * @param x
+   * @return a new instance of erfinv
+   */
+  public <T extends TNumber> erfinv<T> erfinv(Operand<T> x) {
+    return erfinv.create(scope, x);
+  }
+
+  /**
+   * Computes exponential of x element-wise.  \\(y = e^x\\).
+   *  <p>
+   *    This function computes the exponential of every element in the input tensor.
+   *    i.e. `exp(x)` or `e^(x)`, where `x` is the input tensor.
+   *    `e` denotes Euler's number and is approximately equal to 2.718281.
+   *    Output is positive for any real input.
+   *  <p>
+   *    <pre>{@code
+   *    x = tf.constant(2.0)
+   *    tf.math.exp(x) ==> 7.389056
+   *
+   *    x = tf.constant([2.0, 8.0])
+   *    tf.math.exp(x) ==> array([7.389056, 2980.958], dtype=float32)
+   *    }</pre>
+   *  For complex numbers, the exponential value is calculated as follows:
+   *  <p>
+   *    <pre>{@code
+   *    e^(x+iy) = e^x * e^iy = e^x * (cos y + i sin y)
+   *    }</pre>
+   *  Let's consider complex number 1+1j as an example.
+   *    e^1 * (cos 1 + i sin 1) = 2.7182818284590 * (0.54030230586+0.8414709848j)
+   *  <p>
+   *    <pre>{@code
+   *    x = tf.constant(1 + 1j)
+   *    tf.math.exp(x) ==> 1.4686939399158851+2.2873552871788423j
+   *    }</pre>
+   *
+   * @param <T> data type for {@code y()} output
+   * @param x
+   * @return a new instance of Exp
+   */
+  public <T extends TType> Exp<T> exp(Operand<T> x) {
+    return Exp.create(scope, x);
+  }
+
+  /**
+   * Computes `exp(x) - 1` element-wise.
+   *  <p>
+   *    i.e. `exp(x) - 1` or `e^(x) - 1`, where `x` is the input tensor.
+   *    `e` denotes Euler's number and is approximately equal to 2.718281.
+   *  <p>
+   *    <pre>{@code
+   *    x = tf.constant(2.0)
+   *    tf.math.expm1(x) ==> 6.389056
+   *
+   *    x = tf.constant([2.0, 8.0])
+   *    tf.math.expm1(x) ==> array([6.389056, 2979.958], dtype=float32)
+   *
+   *    x = tf.constant(1 + 1j)
+   *    tf.math.expm1(x) ==> (0.46869393991588515+2.2873552871788423j)
+   *    }</pre>
+   *
+   * @param <T> data type for {@code y()} output
+   * @param x
+   * @return a new instance of Expm1
+   */
+  public <T extends TType> Expm1<T> expm1(Operand<T> x) {
+    return Expm1.create(scope, x);
+  }
+
+  /**
+   * Output a fact about factorials.
+   *
+   * @return a new instance of Fact
+   */
+  public Fact fact() {
+    return Fact.create(scope);
+  }
+
+  /**
+   * Returns element-wise largest integer not greater than x.
+   *
+   * @param <T> data type for {@code y()} output
+   * @param x
+   * @return a new instance of Floor
+   */
+  public <T extends TNumber> Floor<T> floor(Operand<T> x) {
+    return Floor.create(scope, x);
+  }
+
+  /**
+   * Returns x // y element-wise.
+   *  <p>
+   *  <i>NOTE</i>: `math.FloorDiv` supports broadcasting. More about broadcasting
+   *  [here](http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html)
+   *
+   * @param <T> data type for {@code z()} output
+   * @param x
+   * @param y
+   * @return a new instance of FloorDiv
+   */
+  public <T extends TType> FloorDiv<T> floorDiv(Operand<T> x, Operand<T> y) {
+    return FloorDiv.create(scope, x, y);
+  }
+
+  /**
+   * Returns element-wise remainder of division. When `x < 0` xor `y < 0` is
+   *  <p>
+   *  true, this follows Python semantics in that the result here is consistent
+   *  with a flooring divide. E.g. `floor(x / y) * y + mod(x, y) = x`.
+   *  <p>
+   *  <i>NOTE</i>: `math.FloorMod` supports broadcasting. More about broadcasting
+   *  [here](http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html)
+   *
+   * @param <T> data type for {@code z()} output
+   * @param x
+   * @param y
+   * @return a new instance of FloorMod
+   */
+  public <T extends TNumber> FloorMod<T> floorMod(Operand<T> x, Operand<T> y) {
+    return FloorMod.create(scope, x, y);
+  }
+
+  /**
+   * Returns the truth value of (x > y) element-wise.
+   *  <p>
+   *  <i>NOTE</i>: `math.Greater` supports broadcasting. More about broadcasting
+   *  [here](http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html)
+   *  <p>
+   *  Example:
+   *  <pre>{@code
+   *  x = tf.constant([5, 4, 6])
+   *  y = tf.constant([5, 2, 5])
+   *  tf.math.greater(x, y) ==> [False, True, True]
+   *
+   *  x = tf.constant([5, 4, 6])
+   *  y = tf.constant([5])
+   *  tf.math.greater(x, y) ==> [False, False, True]
+   *  }</pre>
+   *
+   * @param x
+   * @param y
+   * @return a new instance of Greater
+   */
+  public <T extends TNumber> Greater greater(Operand<T> x, Operand<T> y) {
+    return Greater.create(scope, x, y);
+  }
+
+  /**
+   * Returns the truth value of (x >= y) element-wise.
+   *  <p>
+   *  <i>NOTE</i>: `math.GreaterEqual` supports broadcasting. More about broadcasting
+   *  [here](http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html)
+   *  <p>
+   *  Example:
+   *  <pre>{@code
+   *  x = tf.constant([5, 4, 6, 7])
+   *  y = tf.constant([5, 2, 5, 10])
+   *  tf.math.greater_equal(x, y) ==> [True, True, True, False]
+   *
+   *  x = tf.constant([5, 4, 6, 7])
+   *  y = tf.constant([5])
+   *  tf.math.greater_equal(x, y) ==> [True, False, True, True]
+   *  }</pre>
+   *
+   * @param x
+   * @param y
+   * @return a new instance of GreaterEqual
+   */
+  public <T extends TNumber> GreaterEqual greaterEqual(Operand<T> x, Operand<T> y) {
+    return GreaterEqual.create(scope, x, y);
+  }
+
+  /**
+   * Compute the lower regularized incomplete Gamma function `P(a, x)`.
+   *  <p>
+   *  The lower regularized incomplete Gamma function is defined as:
+   *  <p>
+   *  \\(P(a, x) = gamma(a, x) / Gamma(a) = 1 - Q(a, x)\\)
+   *  <p>
+   *  where
+   *  <p>
+   *  \\(gamma(a, x) = \\int_{0}^{x} t^{a-1} exp(-t) dt\\)
+   *  <p>
+   *  is the lower incomplete Gamma function.
+   *  <p>
+   *  Note, above `Q(a, x)` (`Igammac`) is the upper regularized complete
+   *  Gamma function.
+   *
+   * @param <T> data type for {@code z()} output
+   * @param a
+   * @param x
+   * @return a new instance of Igamma
+   */
+  public <T extends TNumber> Igamma<T> igamma(Operand<T> a, Operand<T> x) {
+    return Igamma.create(scope, a, x);
+  }
+
+  /**
+   * Compute the upper regularized incomplete Gamma function `Q(a, x)`.
+   *  <p>
+   *  The upper regularized incomplete Gamma function is defined as:
+   *  <p>
+   *  \\(Q(a, x) = Gamma(a, x) / Gamma(a) = 1 - P(a, x)\\)
+   *  <p>
+   *  where
+   *  <p>
+   *  \\(Gamma(a, x) = int_{x}^{\infty} t^{a-1} exp(-t) dt\\)
+   *  <p>
+   *  is the upper incomplete Gama function.
+   *  <p>
+   *  Note, above `P(a, x)` (`Igamma`) is the lower regularized complete
+   *  Gamma function.
+   *
+   * @param <T> data type for {@code z()} output
+   * @param a
+   * @param x
+   * @return a new instance of Igammac
+   */
+  public <T extends TNumber> Igammac<T> igammac(Operand<T> a, Operand<T> x) {
+    return Igammac.create(scope, a, x);
+  }
+
+  /**
+   * Returns the imaginary part of a complex number.
+   *  <p>
+   *  Given a tensor `input` of complex numbers, this operation returns a tensor of
+   *  type `float` that is the imaginary part of each element in `input`. All
+   *  elements in `input` must be complex numbers of the form \\(a + bj\\), where <i>a</i>
+   *  is the real part and <i>b</i> is the imaginary part returned by this operation.
+   *  <p>
+   *  For example:
+   *  <pre>{@code
+   *  # tensor 'input' is [-2.25 + 4.75j, 3.25 + 5.75j]
+   *  tf.imag(input) ==> [4.75, 5.75]
+   *  }</pre>
+   *
+   * @param <U> data type for {@code output()} output
+   * @param input
+   * @return a new instance of Imag
+   */
+  public <T extends TType> Imag<TFloat32> imag(Operand<T> input) {
+    return Imag.create(scope, input);
+  }
+
+  /**
+   * Returns the imaginary part of a complex number.
+   *  <p>
+   *  Given a tensor `input` of complex numbers, this operation returns a tensor of
+   *  type `float` that is the imaginary part of each element in `input`. All
+   *  elements in `input` must be complex numbers of the form \\(a + bj\\), where <i>a</i>
+   *  is the real part and <i>b</i> is the imaginary part returned by this operation.
+   *  <p>
+   *  For example:
+   *  <pre>{@code
+   *  # tensor 'input' is [-2.25 + 4.75j, 3.25 + 5.75j]
+   *  tf.imag(input) ==> [4.75, 5.75]
+   *  }</pre>
+   *
+   * @param <U> data type for {@code output()} output
+   * @param input
+   * @param Tout
+   * @return a new instance of Imag
+   */
+  public <U extends TNumber, T extends TType> Imag<U> imag(Operand<T> input, Class<U> Tout) {
+    return Imag.create(scope, input, Tout);
+  }
+
+  /**
+   * Computes the inverse permutation of a tensor.
+   *  <p>
+   *  This operation computes the inverse of an index permutation. It takes a 1-D
+   *  integer tensor `x`, which represents the indices of a zero-based array, and
+   *  swaps each value with its index position. In other words, for an output tensor
+   *  `y` and an input tensor `x`, this operation computes the following:
+   *  <p>
+   *  `y[x[i]] = i for i in [0, 1, ..., len(x) - 1]`
+   *  <p>
+   *  The values must include 0. There can be no duplicate values or negative values.
+   *  <p>
+   *  For example:
+   *  <pre>{@code
+   *  # tensor `x` is [3, 4, 0, 2, 1]
+   *  invert_permutation(x) ==> [2, 4, 3, 0, 1]
+   *  }</pre>
+   *
+   * @param <T> data type for {@code y()} output
+   * @param x 1-D.
+   * @return a new instance of InvertPermutation
+   */
+  public <T extends TNumber> InvertPermutation<T> invertPermutation(Operand<T> x) {
+    return InvertPermutation.create(scope, x);
+  }
+
+  /**
+   * Returns which elements of x are finite.
+   *  <p>
+   *
+   * @compatibility(numpy) Equivalent to np.isfinite
+   * @end_compatibility <p>
+   *  Example:
+   *  <pre>{@code
+   *  x = tf.constant([5.0, 4.8, 6.8, np.inf, np.nan])
+   *  tf.math.is_finite(x) ==> [True, True, True, False, False]
+   *  }</pre>
+   * @param x
+   * @return a new instance of IsFinite
+   */
+  public <T extends TNumber> IsFinite isFinite(Operand<T> x) {
+    return IsFinite.create(scope, x);
+  }
+
+  /**
+   * Returns which elements of x are Inf.
+   *  <p>
+   *
+   * @compatibility(numpy) Equivalent to np.isinf
+   * @end_compatibility <p>
+   *  Example:
+   *  <pre>{@code
+   *  x = tf.constant([5.0, np.inf, 6.8, np.inf])
+   *  tf.math.is_inf(x) ==> [False, True, False, True]
+   *  }</pre>
+   * @param x
+   * @return a new instance of IsInf
+   */
+  public <T extends TNumber> IsInf isInf(Operand<T> x) {
+    return IsInf.create(scope, x);
+  }
+
+  /**
+   * Returns which elements of x are NaN.
+   *  <p>
+   *
+   * @compatibility(numpy) Equivalent to np.isnan
+   * @end_compatibility <p>
+   *  Example:
+   *  <pre>{@code
+   *  x = tf.constant([5.0, np.nan, 6.8, np.nan, np.inf])
+   *  tf.math.is_nan(x) ==> [False, True, False, True, False]
+   *  }</pre>
+   * @param x
+   * @return a new instance of IsNan
+   */
+  public <T extends TNumber> IsNan isNan(Operand<T> x) {
+    return IsNan.create(scope, x);
+  }
+
+  /**
+   * Returns the truth value of (x < y) element-wise.
+   *  <p>
+   *  <i>NOTE</i>: `math.Less` supports broadcasting. More about broadcasting
+   *  [here](http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html)
+   *  <p>
+   *  Example:
+   *  <pre>{@code
+   *  x = tf.constant([5, 4, 6])
+   *  y = tf.constant([5])
+   *  tf.math.less(x, y) ==> [False, True, False]
+   *
+   *  x = tf.constant([5, 4, 6])
+   *  y = tf.constant([5, 6, 7])
+   *  tf.math.less(x, y) ==> [False, True, True]
+   *  }</pre>
+   *
+   * @param x
+   * @param y
+   * @return a new instance of Less
+   */
+  public <T extends TNumber> Less less(Operand<T> x, Operand<T> y) {
+    return Less.create(scope, x, y);
+  }
+
+  /**
+   * Returns the truth value of (x <= y) element-wise.
+   *  <p>
+   *  <i>NOTE</i>: `math.LessEqual` supports broadcasting. More about broadcasting
+   *  [here](http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html)
+   *  <p>
+   *  Example:
+   *  <pre>{@code
+   *  x = tf.constant([5, 4, 6])
+   *  y = tf.constant([5])
+   *  tf.math.less_equal(x, y) ==> [True, True, False]
+   *
+   *  x = tf.constant([5, 4, 6])
+   *  y = tf.constant([5, 6, 6])
+   *  tf.math.less_equal(x, y) ==> [True, True, True]
+   *  }</pre>
+   *
+   * @param x
+   * @param y
+   * @return a new instance of LessEqual
+   */
+  public <T extends TNumber> LessEqual lessEqual(Operand<T> x, Operand<T> y) {
+    return LessEqual.create(scope, x, y);
+  }
+
+  /**
+   * Computes the log of the absolute value of `Gamma(x)` element-wise.
+   *  <p>
+   *    For positive numbers, this function computes log((input - 1)!) for every element in the tensor.
+   *    `lgamma(5) = log((5-1)!) = log(4!) = log(24) = 3.1780539`
+   *  <p>
+   *  Example:
+   *  <pre>{@code
+   *  x = tf.constant([0, 0.5, 1, 4.5, -4, -5.6])
+   *  tf.math.lgamma(x) ==> [inf, 0.5723649, 0., 2.4537368, inf, -4.6477685]
+   *  }</pre>
+   *
+   * @param <T> data type for {@code y()} output
+   * @param x
+   * @return a new instance of Lgamma
+   */
+  public <T extends TNumber> Lgamma<T> lgamma(Operand<T> x) {
+    return Lgamma.create(scope, x);
+  }
+
+  /**
+   * Computes natural logarithm of x element-wise.
+   *  <p>
+   *  I.e., \\(y = \log_e x\\).
+   *  <p>
+   *  Example:
+   *  <pre>{@code
+   *  x = tf.constant([0, 0.5, 1, 5])
+   *  tf.math.log(x) ==> [-inf, -0.6931472,  0. ,  1.609438]
+   *  }</pre>
+   *
+   * @param <T> data type for {@code y()} output
+   * @param x
+   * @return a new instance of Log
+   */
+  public <T extends TType> Log<T> log(Operand<T> x) {
+    return Log.create(scope, x);
+  }
+
+  /**
+   * Computes natural logarithm of (1 + x) element-wise.
+   *  <p>
+   *  I.e., \\(y = \log_e (1 + x)\\).
+   *  <p>
+   *  Example:
+   *  <pre>{@code
+   *  x = tf.constant([0, 0.5, 1, 5])
+   *  tf.math.log1p(x) ==> [0., 0.4054651, 0.6931472, 1.7917595]
+   *  }</pre>
+   *
+   * @param <T> data type for {@code y()} output
+   * @param x
+   * @return a new instance of Log1p
+   */
+  public <T extends TType> Log1p<T> log1p(Operand<T> x) {
+    return Log1p.create(scope, x);
+  }
+
+  /**
+   * Returns the truth value of x AND y element-wise.
+   *  <p>
+   *  <i>NOTE</i>: `math.LogicalAnd` supports broadcasting. More about broadcasting
+   *  [here](http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html)
+   *
+   * @param x
+   * @param y
+   * @return a new instance of LogicalAnd
+   */
+  public LogicalAnd logicalAnd(Operand<TBool> x, Operand<TBool> y) {
+    return LogicalAnd.create(scope, x, y);
+  }
+
+  /**
+   * Returns the truth value of `NOT x` element-wise.
+   *
+   * @param x A `Tensor` of type `bool`.
+   * @return a new instance of LogicalNot
+   */
+  public LogicalNot logicalNot(Operand<TBool> x) {
+    return LogicalNot.create(scope, x);
+  }
+
+  /**
+   * Returns the truth value of x OR y element-wise.
+   *  <p>
+   *  <i>NOTE</i>: `math.LogicalOr` supports broadcasting. More about broadcasting
+   *  [here](http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html)
+   *
+   * @param x
+   * @param y
+   * @return a new instance of LogicalOr
+   */
+  public LogicalOr logicalOr(Operand<TBool> x, Operand<TBool> y) {
+    return LogicalOr.create(scope, x, y);
+  }
+
+  /**
+   * Returns the max of x and y (i.e. x > y ? x : y) element-wise.
+   *  <p>
+   *  <i>NOTE</i>: `math.Maximum` supports broadcasting. More about broadcasting
+   *  [here](http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html)
+   *
+   * @param <T> data type for {@code z()} output
+   * @param x
+   * @param y
+   * @return a new instance of Maximum
+   */
+  public <T extends TNumber> Maximum<T> maximum(Operand<T> x, Operand<T> y) {
+    return Maximum.create(scope, x, y);
+  }
+
+  /**
+   * Computes the mean of elements across dimensions of a tensor.
+   *  <p>
+   *  Reduces `input` along the dimensions given in `axis`. Unless
+   *  `keep_dims` is true, the rank of the tensor is reduced by 1 for each entry in
+   *  `axis`. If `keep_dims` is true, the reduced dimensions are
+   *  retained with length 1.
+   *
+   * @param <T> data type for {@code output()} output
+   * @param input The tensor to reduce.
+   * @param axis The dimensions to reduce. Must be in the range
+   *  `[-rank(input), rank(input))`.
+   * @param options carries optional attributes values
+   * @return a new instance of Mean
+   */
+  public <T extends TType, U extends TNumber> Mean<T> mean(Operand<T> input, Operand<U> axis,
+      Mean.Options... options) {
+    return Mean.create(scope, input, axis, options);
+  }
+
+  /**
+   * Returns the min of x and y (i.e. x < y ? x : y) element-wise.
+   *  <p>
+   *  <i>NOTE</i>: `math.Minimum` supports broadcasting. More about broadcasting
+   *  [here](http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html)
+   *
+   * @param <T> data type for {@code z()} output
+   * @param x
+   * @param y
+   * @return a new instance of Minimum
+   */
+  public <T extends TNumber> Minimum<T> minimum(Operand<T> x, Operand<T> y) {
+    return Minimum.create(scope, x, y);
+  }
+
+  /**
+   * Returns element-wise remainder of division. This emulates C semantics in that
+   *  <p>
+   *  the result here is consistent with a truncating divide. E.g.
+   *  `tf.truncatediv(x, y) * y + truncate_mod(x, y) = x`.
+   *  <p>
+   *  <i>NOTE</i>: `math.Mod` supports broadcasting. More about broadcasting
+   *  [here](http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html)
+   *
+   * @param <T> data type for {@code z()} output
+   * @param x
+   * @param y
+   * @return a new instance of Mod
+   */
+  public <T extends TNumber> Mod<T> mod(Operand<T> x, Operand<T> y) {
+    return Mod.create(scope, x, y);
+  }
+
+  /**
+   * Returns x * y element-wise.
+   *  <p>
+   *  <i>NOTE</i>: `math.Mul` supports broadcasting. More about broadcasting
+   *  [here](http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html)
+   *
+   * @param <T> data type for {@code z()} output
+   * @param x
+   * @param y
+   * @return a new instance of Mul
+   */
+  public <T extends TType> Mul<T> mul(Operand<T> x, Operand<T> y) {
+    return Mul.create(scope, x, y);
+  }
+
+  /**
+   * Returns x * y element-wise. Returns zero if y is zero, even if x if infinite or NaN.
+   *  <p>
+   *  <i>NOTE</i>: `math.MulNoNan` supports broadcasting. More about broadcasting
+   *  [here](http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html)
+   *
+   * @param <T> data type for {@code z()} output
+   * @param x
+   * @param y
+   * @return a new instance of MulNoNan
+   */
+  public <T extends TType> MulNoNan<T> mulNoNan(Operand<T> x, Operand<T> y) {
+    return MulNoNan.create(scope, x, y);
+  }
+
+  /**
+   *
+   * @param <T> data type for {@code y()} output
+   * @param x
+   * @return a new instance of Ndtri
+   */
+  public <T extends TNumber> Ndtri<T> ndtri(Operand<T> x) {
+    return Ndtri.create(scope, x);
+  }
+
+  /**
+   * Computes numerical negative value element-wise.
+   *  <p>
+   *  I.e., \\(y = -x\\).
+   *
+   * @param <T> data type for {@code y()} output
+   * @param x
+   * @return a new instance of Neg
+   */
+  public <T extends TType> Neg<T> neg(Operand<T> x) {
+    return Neg.create(scope, x);
+  }
+
+  /**
+   * Returns the next representable value of `x1` in the direction of `x2`, element-wise.
+   *  <p>
+   *  This operation returns the same result as the C++ std::nextafter function.
+   *  <p>
+   *  It can also return a subnormal number.
+   *  <p>
+   *
+   * @compatibility(cpp) Equivalent to C++ std::nextafter function.
+   * @end_compatibility
+   * @param <T> data type for {@code output()} output
+   * @param x1
+   * @param x2
+   * @return a new instance of NextAfter
+   */
+  public <T extends TNumber> NextAfter<T> nextAfter(Operand<T> x1, Operand<T> x2) {
+    return NextAfter.create(scope, x1, x2);
+  }
+
+  /**
+   * Returns the truth value of (x != y) element-wise.
+   *  <p>
+   *  <i>NOTE</i>: `math.NotEqual` supports broadcasting. More about broadcasting
+   *  [here](http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html)
+   *
+   * @param x
+   * @param y
+   * @param options carries optional attributes values
+   * @return a new instance of NotEqual
+   */
+  public <T extends TType> NotEqual notEqual(Operand<T> x, Operand<T> y,
+      NotEqual.Options... options) {
+    return NotEqual.create(scope, x, y, options);
+  }
+
+  /**
+   * Compute the polygamma function \\(\psi^{(n)}(x)\\).
+   *  <p>
+   *  The polygamma function is defined as:
+   *  <p>
+   *  \\(\psi^{(a)}(x) = \frac{d^a}{dx^a} \psi(x)\\)
+   *  <p>
+   *  where \\(\psi(x)\\) is the digamma function.
+   *  The polygamma function is defined only for non-negative integer orders \\a\\.
+   *
+   * @param <T> data type for {@code z()} output
+   * @param a
+   * @param x
+   * @return a new instance of Polygamma
+   */
+  public <T extends TNumber> Polygamma<T> polygamma(Operand<T> a, Operand<T> x) {
+    return Polygamma.create(scope, a, x);
+  }
+
+  /**
+   * Computes element-wise population count (a.k.a. popcount, bitsum, bitcount).
+   *  <p>
+   *  For each entry in `x`, calculates the number of `1` (on) bits in the binary
+   *  representation of that entry.
+   *  <p>
+   *  <b>NOTE</b>: It is more efficient to first `tf.bitcast` your tensors into
+   *  `int32` or `int64` and perform the bitcount on the result, than to feed in
+   *  8- or 16-bit inputs and then aggregate the resulting counts.
+   *
+   * @param x
+   * @return a new instance of PopulationCount
+   */
+  public <T extends TNumber> PopulationCount populationCount(Operand<T> x) {
+    return PopulationCount.create(scope, x);
+  }
+
+  /**
+   * Computes the power of one value to another.
+   *  <p>
+   *  Given a tensor `x` and a tensor `y`, this operation computes \\(x^y\\) for
+   *  corresponding elements in `x` and `y`. For example:
+   *  <pre>{@code
+   *  # tensor 'x' is [[2, 2]], [3, 3]]
+   *  # tensor 'y' is [[8, 16], [2, 3]]
+   *  tf.pow(x, y) ==> [[256, 65536], [9, 27]]
+   *  }</pre>
+   *
+   * @param <T> data type for {@code z()} output
+   * @param x
+   * @param y
+   * @return a new instance of Pow
+   */
+  public <T extends TType> Pow<T> pow(Operand<T> x, Operand<T> y) {
+    return Pow.create(scope, x, y);
+  }
+
+  /**
+   * Returns x + y element-wise, working on quantized buffers.
+   *
+   * @param <V> data type for {@code z()} output
+   * @param x
+   * @param y
+   * @param minX The float value that the lowest quantized `x` value represents.
+   * @param maxX The float value that the highest quantized `x` value represents.
+   * @param minY The float value that the lowest quantized `y` value represents.
+   * @param maxY The float value that the highest quantized `y` value represents.
+   * @param Toutput
+   * @return a new instance of QuantizedAdd
+   */
+  public <V extends TType, T extends TType, U extends TType> QuantizedAdd<V> quantizedAdd(
+      Operand<T> x, Operand<U> y, Operand<TFloat32> minX, Operand<TFloat32> maxX,
+      Operand<TFloat32> minY, Operand<TFloat32> maxY, Class<V> Toutput) {
+    return QuantizedAdd.create(scope, x, y, minX, maxX, minY, maxY, Toutput);
+  }
+
+  /**
+   * Returns x * y element-wise, working on quantized buffers.
+   *
+   * @param <V> data type for {@code z()} output
+   * @param x
+   * @param y
+   * @param minX The float value that the lowest quantized `x` value represents.
+   * @param maxX The float value that the highest quantized `x` value represents.
+   * @param minY The float value that the lowest quantized `y` value represents.
+   * @param maxY The float value that the highest quantized `y` value represents.
+   * @param Toutput
+   * @return a new instance of QuantizedMul
+   */
+  public <V extends TType, T extends TType, U extends TType> QuantizedMul<V> quantizedMul(
+      Operand<T> x, Operand<U> y, Operand<TFloat32> minX, Operand<TFloat32> maxX,
+      Operand<TFloat32> minY, Operand<TFloat32> maxY, Class<V> Toutput) {
+    return QuantizedMul.create(scope, x, y, minX, maxX, minY, maxY, Toutput);
+  }
+
+  /**
+   * Returns the real part of a complex number.
+   *  <p>
+   *  Given a tensor `input` of complex numbers, this operation returns a tensor of
+   *  type `float` that is the real part of each element in `input`. All elements in
+   *  `input` must be complex numbers of the form \\(a + bj\\), where <i>a</i> is the real
+   *   part returned by this operation and <i>b</i> is the imaginary part.
+   *  <p>
+   *  For example:
+   *  <pre>{@code
+   *  # tensor 'input' is [-2.25 + 4.75j, 3.25 + 5.75j]
+   *  tf.real(input) ==> [-2.25, 3.25]
+   *  }</pre>
+   *
+   * @param <U> data type for {@code output()} output
+   * @param input
+   * @return a new instance of Real
+   */
+  public <T extends TType> Real<TFloat32> real(Operand<T> input) {
+    return Real.create(scope, input);
+  }
+
+  /**
+   * Returns the real part of a complex number.
+   *  <p>
+   *  Given a tensor `input` of complex numbers, this operation returns a tensor of
+   *  type `float` that is the real part of each element in `input`. All elements in
+   *  `input` must be complex numbers of the form \\(a + bj\\), where <i>a</i> is the real
+   *   part returned by this operation and <i>b</i> is the imaginary part.
+   *  <p>
+   *  For example:
+   *  <pre>{@code
+   *  # tensor 'input' is [-2.25 + 4.75j, 3.25 + 5.75j]
+   *  tf.real(input) ==> [-2.25, 3.25]
+   *  }</pre>
+   *
+   * @param <U> data type for {@code output()} output
+   * @param input
+   * @param Tout
+   * @return a new instance of Real
+   */
+  public <U extends TNumber, T extends TType> Real<U> real(Operand<T> input, Class<U> Tout) {
+    return Real.create(scope, input, Tout);
+  }
+
+  /**
+   * Returns x / y element-wise for real types.
+   *  <p>
+   *  If `x` and `y` are reals, this will return the floating-point division.
+   *  <p>
+   *  <i>NOTE</i>: `Div` supports broadcasting. More about broadcasting
+   *  [here](http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html)
+   *
+   * @param <T> data type for {@code z()} output
+   * @param x
+   * @param y
+   * @return a new instance of RealDiv
+   */
+  public <T extends TType> RealDiv<T> realDiv(Operand<T> x, Operand<T> y) {
+    return RealDiv.create(scope, x, y);
+  }
+
+  /**
+   * Computes the reciprocal of x element-wise.
+   *  <p>
+   *  I.e., \\(y = 1 / x\\).
+   *
+   * @param <T> data type for {@code y()} output
+   * @param x
+   * @return a new instance of Reciprocal
+   */
+  public <T extends TType> Reciprocal<T> reciprocal(Operand<T> x) {
+    return Reciprocal.create(scope, x);
+  }
+
+  /**
+   * Returns element-wise integer closest to x.
+   *  <p>
+   *  If the result is midway between two representable values,
+   *  the even representable is chosen.
+   *  For example:
+   *  <pre>{@code
+   *  rint(-1.5) ==> -2.0
+   *  rint(0.5000001) ==> 1.0
+   *  rint([-1.7, -1.5, -0.2, 0.2, 1.5, 1.7, 2.0]) ==> [-2., -2., -0., 0., 2., 2., 2.]
+   *  }</pre>
+   *
+   * @param <T> data type for {@code y()} output
+   * @param x
+   * @return a new instance of Rint
+   */
+  public <T extends TNumber> Rint<T> rint(Operand<T> x) {
+    return Rint.create(scope, x);
+  }
+
+  /**
+   * Rounds the values of a tensor to the nearest integer, element-wise.
+   *  <p>
+   *  Rounds half to even.  Also known as bankers rounding. If you want to round
+   *  according to the current system rounding mode use std::cint.
+   *
+   * @param <T> data type for {@code y()} output
+   * @param x
+   * @return a new instance of Round
+   */
+  public <T extends TType> Round<T> round(Operand<T> x) {
+    return Round.create(scope, x);
+  }
+
+  /**
+   * Computes reciprocal of square root of x element-wise.
+   *  <p>
+   *  I.e., \\(y = 1 / \sqrt{x}\\).
+   *
+   * @param <T> data type for {@code y()} output
+   * @param x
+   * @return a new instance of Rsqrt
+   */
+  public <T extends TType> Rsqrt<T> rsqrt(Operand<T> x) {
+    return Rsqrt.create(scope, x);
+  }
+
+  /**
+   * Computes the maximum along segments of a tensor.
+   *  <p>
+   *  Read
+   *  [the section on segmentation](https://tensorflow.org/api_docs/python/tf/math#Segmentation)
+   *  for an explanation of segments.
+   *  <p>
+   *  Computes a tensor such that
+   *  \\(output_i = \max_j(data_j)\\) where `max` is over `j` such
+   *  that `segment_ids[j] == i`.
+   *  <p>
+   *  If the max is empty for a given segment ID `i`, `output[i] = 0`.
+   *  <p>
+   *  <div style="width:70%; margin:auto; margin-bottom:10px; margin-top:20px;">
+   *  <img style="width:100%" src="https://www.tensorflow.org/images/SegmentMax.png" alt>
+   *  </div>
+   *  <p>
+   *  For example:
+   *  <pre>{@code
+   *  c = tf.constant([[1,2,3,4], [4, 3, 2, 1], [5,6,7,8]])
+   *  tf.segment_max(c, tf.constant([0, 0, 1]))
+   *  # ==> [[4, 3, 3, 4],
+   *  #      [5, 6, 7, 8]]
+   *  }</pre>
+   *
+   * @param <T> data type for {@code output()} output
+   * @param data
+   * @param segmentIds A 1-D tensor whose size is equal to the size of `data`'s
+   *  first dimension.  Values should be sorted and can be repeated.
+   * @return a new instance of SegmentMax
+   */
+  public <T extends TNumber, U extends TNumber> SegmentMax<T> segmentMax(Operand<T> data,
+      Operand<U> segmentIds) {
+    return SegmentMax.create(scope, data, segmentIds);
+  }
+
+  /**
+   * Computes the mean along segments of a tensor.
+   *  <p>
+   *  Read
+   *  [the section on segmentation](https://tensorflow.org/api_docs/python/tf/math#Segmentation)
+   *  for an explanation of segments.
+   *  <p>
+   *  Computes a tensor such that
+   *  \\(output_i = \frac{\sum_j data_j}{N}\\) where `mean` is
+   *  over `j` such that `segment_ids[j] == i` and `N` is the total number of
+   *  values summed.
+   *  <p>
+   *  If the mean is empty for a given segment ID `i`, `output[i] = 0`.
+   *  <p>
+   *  <div style="width:70%; margin:auto; margin-bottom:10px; margin-top:20px;">
+   *  <img style="width:100%" src="https://www.tensorflow.org/images/SegmentMean.png" alt>
+   *  </div>
+   *  <p>
+   *  For example:
+   *  <pre>{@code
+   *  c = tf.constant([[1.0,2,3,4], [4, 3, 2, 1], [5,6,7,8]])
+   *  tf.segment_mean(c, tf.constant([0, 0, 1]))
+   *  # ==> [[2.5, 2.5, 2.5, 2.5],
+   *  #      [5, 6, 7, 8]]
+   *  }</pre>
+   *
+   * @param <T> data type for {@code output()} output
+   * @param data
+   * @param segmentIds A 1-D tensor whose size is equal to the size of `data`'s
+   *  first dimension.  Values should be sorted and can be repeated.
+   * @return a new instance of SegmentMean
+   */
+  public <T extends TType, U extends TNumber> SegmentMean<T> segmentMean(Operand<T> data,
+      Operand<U> segmentIds) {
+    return SegmentMean.create(scope, data, segmentIds);
+  }
+
+  /**
+   * Computes the minimum along segments of a tensor.
+   *  <p>
+   *  Read
+   *  [the section on segmentation](https://tensorflow.org/api_docs/python/tf/math#Segmentation)
+   *  for an explanation of segments.
+   *  <p>
+   *  Computes a tensor such that
+   *  \\(output_i = \min_j(data_j)\\) where `min` is over `j` such
+   *  that `segment_ids[j] == i`.
+   *  <p>
+   *  If the min is empty for a given segment ID `i`, `output[i] = 0`.
+   *  <p>
+   *  <div style="width:70%; margin:auto; margin-bottom:10px; margin-top:20px;">
+   *  <img style="width:100%" src="https://www.tensorflow.org/images/SegmentMin.png" alt>
+   *  </div>
+   *  <p>
+   *  For example:
+   *  <pre>{@code
+   *  c = tf.constant([[1,2,3,4], [4, 3, 2, 1], [5,6,7,8]])
+   *  tf.segment_min(c, tf.constant([0, 0, 1]))
+   *  # ==> [[1, 2, 2, 1],
+   *  #      [5, 6, 7, 8]]
+   *  }</pre>
+   *
+   * @param <T> data type for {@code output()} output
+   * @param data
+   * @param segmentIds A 1-D tensor whose size is equal to the size of `data`'s
+   *  first dimension.  Values should be sorted and can be repeated.
+   * @return a new instance of SegmentMin
+   */
+  public <T extends TNumber, U extends TNumber> SegmentMin<T> segmentMin(Operand<T> data,
+      Operand<U> segmentIds) {
+    return SegmentMin.create(scope, data, segmentIds);
+  }
+
+  /**
+   * Computes the product along segments of a tensor.
+   *  <p>
+   *  Read
+   *  [the section on segmentation](https://tensorflow.org/api_docs/python/tf/math#Segmentation)
+   *  for an explanation of segments.
+   *  <p>
+   *  Computes a tensor such that
+   *  \\(output_i = \prod_j data_j\\) where the product is over `j` such
+   *  that `segment_ids[j] == i`.
+   *  <p>
+   *  If the product is empty for a given segment ID `i`, `output[i] = 1`.
+   *  <p>
+   *  <div style="width:70%; margin:auto; margin-bottom:10px; margin-top:20px;">
+   *  <img style="width:100%" src="https://www.tensorflow.org/images/SegmentProd.png" alt>
+   *  </div>
+   *  <p>
+   *  For example:
+   *  <pre>{@code
+   *  c = tf.constant([[1,2,3,4], [4, 3, 2, 1], [5,6,7,8]])
+   *  tf.segment_prod(c, tf.constant([0, 0, 1]))
+   *  # ==> [[4, 6, 6, 4],
+   *  #      [5, 6, 7, 8]]
+   *  }</pre>
+   *
+   * @param <T> data type for {@code output()} output
+   * @param data
+   * @param segmentIds A 1-D tensor whose size is equal to the size of `data`'s
+   *  first dimension.  Values should be sorted and can be repeated.
+   * @return a new instance of SegmentProd
+   */
+  public <T extends TType, U extends TNumber> SegmentProd<T> segmentProd(Operand<T> data,
+      Operand<U> segmentIds) {
+    return SegmentProd.create(scope, data, segmentIds);
+  }
+
+  /**
+   * Computes the sum along segments of a tensor.
+   *  <p>
+   *  Read
+   *  [the section on segmentation](https://tensorflow.org/api_docs/python/tf/math#Segmentation)
+   *  for an explanation of segments.
+   *  <p>
+   *  Computes a tensor such that
+   *  \\(output_i = \sum_j data_j\\) where sum is over `j` such
+   *  that `segment_ids[j] == i`.
+   *  <p>
+   *  If the sum is empty for a given segment ID `i`, `output[i] = 0`.
+   *  <p>
+   *  <div style="width:70%; margin:auto; margin-bottom:10px; margin-top:20px;">
+   *  <img style="width:100%" src="https://www.tensorflow.org/images/SegmentSum.png" alt>
+   *  </div>
+   *  <p>
+   *  For example:
+   *  <pre>{@code
+   *  c = tf.constant([[1,2,3,4], [4, 3, 2, 1], [5,6,7,8]])
+   *  tf.segment_sum(c, tf.constant([0, 0, 1]))
+   *  # ==> [[5, 5, 5, 5],
+   *  #      [5, 6, 7, 8]]
+   *  }</pre>
+   *
+   * @param <T> data type for {@code output()} output
+   * @param data
+   * @param segmentIds A 1-D tensor whose size is equal to the size of `data`'s
+   *  first dimension.  Values should be sorted and can be repeated.
+   * @return a new instance of SegmentSum
+   */
+  public <T extends TType, U extends TNumber> SegmentSum<T> segmentSum(Operand<T> data,
+      Operand<U> segmentIds) {
+    return SegmentSum.create(scope, data, segmentIds);
+  }
+
+  /**
+   * Computes sigmoid of `x` element-wise.
+   *  <p>
+   *  Specifically, `y = 1 / (1 + exp(-x))`.
+   *
+   * @param <T> data type for {@code y()} output
+   * @param x
+   * @return a new instance of Sigmoid
+   */
+  public <T extends TType> Sigmoid<T> sigmoid(Operand<T> x) {
+    return Sigmoid.create(scope, x);
+  }
+
+  /**
+   * Returns an element-wise indication of the sign of a number.
+   *  <p>
+   *  `y = sign(x) = -1` if `x < 0`; 0 if `x == 0`; 1 if `x > 0`.
+   *  <p>
+   *  For complex numbers, `y = sign(x) = x / |x|` if `x != 0`, otherwise `y = 0`.
+   *  <p>
+   *  Example usage:
+   *  >>> tf.math.sign([0., 2., -3.])
+   *  <tf.Tensor: shape=(3,), dtype=float32, numpy=array([ 0.,  1., -1.], dtype=float32)>
+   *
+   * @param <T> data type for {@code y()} output
+   * @param x
+   * @return a new instance of Sign
+   */
+  public <T extends TType> Sign<T> sign(Operand<T> x) {
+    return Sign.create(scope, x);
+  }
+
+  /**
+   * Computes sine of x element-wise.
+   *  <p>
+   *    Given an input tensor, this function computes sine of every
+   *    element in the tensor. Input range is `(-inf, inf)` and
+   *    output range is `[-1,1]`.
+   *  <p>
+   *    <pre>{@code
+   *    x = tf.constant([-float("inf"), -9, -0.5, 1, 1.2, 200, 10, float("inf")])
+   *    tf.math.sin(x) ==> [nan -0.4121185 -0.47942555 0.84147096 0.9320391 -0.87329733 -0.54402107 nan]
+   *    }</pre>
+   *
+   * @param <T> data type for {@code y()} output
+   * @param x
+   * @return a new instance of Sin
+   */
+  public <T extends TType> Sin<T> sin(Operand<T> x) {
+    return Sin.create(scope, x);
+  }
+
+  /**
+   * Computes hyperbolic sine of x element-wise.
+   *  <p>
+   *    Given an input tensor, this function computes hyperbolic sine of every
+   *    element in the tensor. Input range is `[-inf,inf]` and output range
+   *    is `[-inf,inf]`.
+   *  <p>
+   *    <pre>{@code
+   *    x = tf.constant([-float("inf"), -9, -0.5, 1, 1.2, 2, 10, float("inf")])
+   *    tf.math.sinh(x) ==> [-inf -4.0515420e+03 -5.2109528e-01 1.1752012e+00 1.5094614e+00 3.6268604e+00 1.1013232e+04 inf]
+   *    }</pre>
+   *
+   * @param <T> data type for {@code y()} output
+   * @param x
+   * @return a new instance of Sinh
+   */
+  public <T extends TType> Sinh<T> sinh(Operand<T> x) {
+    return Sinh.create(scope, x);
+  }
+
+  /**
+   * Computes softplus: `log(exp(features) + 1)`.
+   *
+   * @param <T> data type for {@code activations()} output
+   * @param features
+   * @return a new instance of Softplus
+   */
+  public <T extends TNumber> Softplus<T> softplus(Operand<T> features) {
+    return Softplus.create(scope, features);
+  }
+
+  /**
+   * Computes square root of x element-wise.
+   *  <p>
+   *  I.e., \\(y = \sqrt{x} = x^{1/2}\\).
+   *
+   * @param <T> data type for {@code y()} output
+   * @param x
+   * @return a new instance of Sqrt
+   */
+  public <T extends TType> Sqrt<T> sqrt(Operand<T> x) {
+    return Sqrt.create(scope, x);
+  }
+
+  /**
+   * Computes square of x element-wise.
+   *  <p>
+   *  I.e., \\(y = x * x = x^2\\).
+   *
+   * @param <T> data type for {@code y()} output
+   * @param x
+   * @return a new instance of Square
+   */
+  public <T extends TType> Square<T> square(Operand<T> x) {
+    return Square.create(scope, x);
+  }
+
+  /**
+   * Returns (x - y)(x - y) element-wise.
+   *  <p>
+   *  <i>NOTE</i>: `math.SquaredDifference` supports broadcasting. More about broadcasting
+   *  [here](http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html)
+   *
+   * @param <T> data type for {@code z()} output
+   * @param x
+   * @param y
+   * @return a new instance of SquaredDifference
+   */
+  public <T extends TType> SquaredDifference<T> squaredDifference(Operand<T> x, Operand<T> y) {
+    return SquaredDifference.create(scope, x, y);
+  }
+
+  /**
+   * Returns x - y element-wise.
+   *  <p>
+   *  <i>NOTE</i>: `math.Sub` supports broadcasting. More about broadcasting
+   *  [here](http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html)
+   *
+   * @param <T> data type for {@code z()} output
+   * @param x
+   * @param y
+   * @return a new instance of Sub
+   */
+  public <T extends TType> Sub<T> sub(Operand<T> x, Operand<T> y) {
+    return Sub.create(scope, x, y);
+  }
+
+  /**
+   * Computes tan of x element-wise.
+   *  <p>
+   *    Given an input tensor, this function computes tangent of every
+   *    element in the tensor. Input range is `(-inf, inf)` and
+   *    output range is `(-inf, inf)`. If input lies outside the boundary, `nan`
+   *    is returned.
+   *  <p>
+   *    <pre>{@code
+   *    x = tf.constant([-float("inf"), -9, -0.5, 1, 1.2, 200, 10000, float("inf")])
+   *    tf.math.tan(x) ==> [nan 0.45231566 -0.5463025 1.5574077 2.572152 -1.7925274 0.32097113 nan]
+   *    }</pre>
+   *
+   * @param <T> data type for {@code y()} output
+   * @param x
+   * @return a new instance of Tan
+   */
+  public <T extends TType> Tan<T> tan(Operand<T> x) {
+    return Tan.create(scope, x);
+  }
+
+  /**
+   * Computes hyperbolic tangent of `x` element-wise.
+   *  <p>
+   *    Given an input tensor, this function computes hyperbolic tangent of every
+   *    element in the tensor. Input range is `[-inf, inf]` and
+   *    output range is `[-1,1]`.
+   *  <p>
+   *    <pre>{@code
+   *    x = tf.constant([-float("inf"), -5, -0.5, 1, 1.2, 2, 3, float("inf")])
+   *    tf.math.tanh(x) ==> [-1. -0.99990916 -0.46211717 0.7615942 0.8336547 0.9640276 0.9950547 1.]
+   *    }</pre>
+   *
+   * @param <T> data type for {@code y()} output
+   * @param x
+   * @return a new instance of Tanh
+   */
+  public <T extends TType> Tanh<T> tanh(Operand<T> x) {
+    return Tanh.create(scope, x);
+  }
+
+  /**
+   * Returns x / y element-wise for integer types.
+   *  <p>
+   *  Truncation designates that negative numbers will round fractional quantities
+   *  toward zero. I.e. -7 / 5 = -1. This matches C semantics but it is different
+   *  than Python semantics. See `FloorDiv` for a division function that matches
+   *  Python Semantics.
+   *  <p>
+   *  <i>NOTE</i>: `math.TruncateDiv` supports broadcasting. More about broadcasting
+   *  [here](http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html)
+   *
+   * @param <T> data type for {@code z()} output
+   * @param x
+   * @param y
+   * @return a new instance of TruncateDiv
+   */
+  public <T extends TType> TruncateDiv<T> truncateDiv(Operand<T> x, Operand<T> y) {
+    return TruncateDiv.create(scope, x, y);
+  }
+
+  /**
+   * Returns element-wise remainder of division. This emulates C semantics in that
+   *  <p>
+   *  the result here is consistent with a truncating divide. E.g. `truncate(x / y) *
+   *  y + truncate_mod(x, y) = x`.
+   *  <p>
+   *  <i>NOTE</i>: `math.TruncateMod` supports broadcasting. More about broadcasting
+   *  [here](http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html)
+   *
+   * @param <T> data type for {@code z()} output
+   * @param x
+   * @param y
+   * @return a new instance of TruncateMod
+   */
+  public <T extends TNumber> TruncateMod<T> truncateMod(Operand<T> x, Operand<T> y) {
+    return TruncateMod.create(scope, x, y);
+  }
+
+  /**
+   * Computes the maximum along segments of a tensor.
+   *  <p>
+   *  Read
+   *  [the section on segmentation](https://tensorflow.org/api_docs/python/tf/math#Segmentation)
+   *  for an explanation of segments.
+   *  <p>
+   *  This operator is similar to the unsorted segment sum operator found
+   *  [(here)](../../../api_docs/python/math_ops.md#UnsortedSegmentSum).
+   *  Instead of computing the sum over segments, it computes the maximum such that:
+   *  <p>
+   *  \\(output_i = \max_{j...} data[j...]\\) where max is over tuples `j...` such
+   *  that `segment_ids[j...] == i`.
+   *  <p>
+   *  If the maximum is empty for a given segment ID `i`, it outputs the smallest
+   *  possible value for the specific numeric type,
+   *  `output[i] = numeric_limits<T>::lowest()`.
+   *  <p>
+   *  If the given segment ID `i` is negative, then the corresponding value is
+   *  dropped, and will not be included in the result.
+   *  <p>
+   *  <div style="width:70%; margin:auto; margin-bottom:10px; margin-top:20px;">
+   *  <img style="width:100%" src="https://www.tensorflow.org/images/UnsortedSegmentMax.png" alt>
+   *  </div>
+   *  <p>
+   *  For example:
+   *  <pre>{@code
+   *  c = tf.constant([[1,2,3,4], [5,6,7,8], [4,3,2,1]])
+   *  tf.unsorted_segment_max(c, tf.constant([0, 1, 0]), num_segments=2)
+   *  # ==> [[ 4,  3, 3, 4],
+   *  #       [5,  6, 7, 8]]
+   *  }</pre>
+   *
+   * @param <T> data type for {@code output()} output
+   * @param data
+   * @param segmentIds A tensor whose shape is a prefix of `data.shape`.
+   * @param numSegments
+   * @return a new instance of UnsortedSegmentMax
+   */
+  public <T extends TNumber, U extends TNumber, V extends TNumber> UnsortedSegmentMax<T> unsortedSegmentMax(
+      Operand<T> data, Operand<U> segmentIds, Operand<V> numSegments) {
+    return UnsortedSegmentMax.create(scope, data, segmentIds, numSegments);
+  }
+
+  /**
+   * Computes the minimum along segments of a tensor.
+   *  <p>
+   *  Read
+   *  [the section on segmentation](https://tensorflow.org/api_docs/python/tf/math#Segmentation)
+   *  for an explanation of segments.
+   *  <p>
+   *  This operator is similar to the unsorted segment sum operator found
+   *  [(here)](../../../api_docs/python/math_ops.md#UnsortedSegmentSum).
+   *  Instead of computing the sum over segments, it computes the minimum such that:
+   *  <p>
+   *  \\(output_i = \min_{j...} data_[j...]\\) where min is over tuples `j...` such
+   *  that `segment_ids[j...] == i`.
+   *  <p>
+   *  If the minimum is empty for a given segment ID `i`, it outputs the largest
+   *  possible value for the specific numeric type,
+   *  `output[i] = numeric_limits<T>::max()`.
+   *  <p>
+   *  For example:
+   *  <pre>{@code
+   *  c = tf.constant([[1,2,3,4], [5,6,7,8], [4,3,2,1]])
+   *  tf.unsorted_segment_min(c, tf.constant([0, 1, 0]), num_segments=2)
+   *  # ==> [[ 1,  2, 2, 1],
+   *  #       [5,  6, 7, 8]]
+   *  }</pre>
+   *  If the given segment ID `i` is negative, then the corresponding value is
+   *  dropped, and will not be included in the result.
+   *
+   * @param <T> data type for {@code output()} output
+   * @param data
+   * @param segmentIds A tensor whose shape is a prefix of `data.shape`.
+   * @param numSegments
+   * @return a new instance of UnsortedSegmentMin
+   */
+  public <T extends TNumber, U extends TNumber, V extends TNumber> UnsortedSegmentMin<T> unsortedSegmentMin(
+      Operand<T> data, Operand<U> segmentIds, Operand<V> numSegments) {
+    return UnsortedSegmentMin.create(scope, data, segmentIds, numSegments);
+  }
+
+  /**
+   * Computes the product along segments of a tensor.
+   *  <p>
+   *  Read
+   *  [the section on segmentation](https://tensorflow.org/api_docs/python/tf/math#Segmentation)
+   *  for an explanation of segments.
+   *  <p>
+   *  This operator is similar to the unsorted segment sum operator found
+   *  [(here)](../../../api_docs/python/math_ops.md#UnsortedSegmentSum).
+   *  Instead of computing the sum over segments, it computes the product of all
+   *  entries belonging to a segment such that:
+   *  <p>
+   *  \\(output_i = \prod_{j...} data[j...]\\) where the product is over tuples
+   *  `j...` such that `segment_ids[j...] == i`.
+   *  <p>
+   *  For example:
+   *  <pre>{@code
+   *  c = tf.constant([[1,2,3,4], [5,6,7,8], [4,3,2,1]])
+   *  tf.unsorted_segment_prod(c, tf.constant([0, 1, 0]), num_segments=2)
+   *  # ==> [[ 4,  6, 6, 4],
+   *  #       [5,  6, 7, 8]]
+   *  }</pre>
+   *  If there is no entry for a given segment ID `i`, it outputs 1.
+   *  <p>
+   *  If the given segment ID `i` is negative, then the corresponding value is
+   *  dropped, and will not be included in the result.
+   *
+   * @param <T> data type for {@code output()} output
+   * @param data
+   * @param segmentIds A tensor whose shape is a prefix of `data.shape`.
+   * @param numSegments
+   * @return a new instance of UnsortedSegmentProd
+   */
+  public <T extends TType, U extends TNumber, V extends TNumber> UnsortedSegmentProd<T> unsortedSegmentProd(
+      Operand<T> data, Operand<U> segmentIds, Operand<V> numSegments) {
+    return UnsortedSegmentProd.create(scope, data, segmentIds, numSegments);
+  }
+
+  /**
+   * Computes the sum along segments of a tensor.
+   *  <p>
+   *  Read
+   *  [the section on segmentation](https://tensorflow.org/api_docs/python/tf/math#Segmentation)
+   *  for an explanation of segments.
+   *  <p>
+   *  Computes a tensor such that
+   *  \\(output[i] = \sum_{j...} data[j...]\\) where the sum is over tuples `j...` such
+   *  that `segment_ids[j...] == i`.  Unlike `SegmentSum`, `segment_ids`
+   *  need not be sorted and need not cover all values in the full
+   *  range of valid values.
+   *  <p>
+   *  If the sum is empty for a given segment ID `i`, `output[i] = 0`.
+   *  If the given segment ID `i` is negative, the value is dropped and will not be
+   *  added to the sum of the segment.
+   *  <p>
+   *  `num_segments` should equal the number of distinct segment IDs.
+   *  <p>
+   *  <div style="width:70%; margin:auto; margin-bottom:10px; margin-top:20px;">
+   *  <img style="width:100%" src="https://www.tensorflow.org/images/UnsortedSegmentSum.png" alt>
+   *  </div>
+   *  <pre>{@code
+   *  c = tf.constant([[1,2,3,4], [5,6,7,8], [4,3,2,1]])
+   *  tf.unsorted_segment_sum(c, tf.constant([0, 1, 0]), num_segments=2)
+   *  # ==> [[ 5,  5, 5, 5],
+   *  #       [5,  6, 7, 8]]
+   *  }</pre>
+   *
+   * @param <T> data type for {@code output()} output
+   * @param data
+   * @param segmentIds A tensor whose shape is a prefix of `data.shape`.
+   * @param numSegments
+   * @return a new instance of UnsortedSegmentSum
+   */
+  public <T extends TType, U extends TNumber, V extends TNumber> UnsortedSegmentSum<T> unsortedSegmentSum(
+      Operand<T> data, Operand<U> segmentIds, Operand<V> numSegments) {
+    return UnsortedSegmentSum.create(scope, data, segmentIds, numSegments);
+  }
+
+  /**
+   * Returns 0 if x == 0, and x / y otherwise, elementwise.
+   *
+   * @param <T> data type for {@code z()} output
+   * @param x
+   * @param y
+   * @return a new instance of Xdivy
+   */
+  public <T extends TType> Xdivy<T> xdivy(Operand<T> x, Operand<T> y) {
+    return Xdivy.create(scope, x, y);
+  }
+
+  /**
+   * Returns 0 if x == 0, and x * log1p(y) otherwise, elementwise.
+   *
+   * @param <T> data type for {@code z()} output
+   * @param x
+   * @param y
+   * @return a new instance of Xlog1py
+   */
+  public <T extends TType> Xlog1py<T> xlog1py(Operand<T> x, Operand<T> y) {
+    return Xlog1py.create(scope, x, y);
+  }
+
+  /**
+   * Returns 0 if x == 0, and x * log(y) otherwise, elementwise.
+   *
+   * @param <T> data type for {@code z()} output
+   * @param x
+   * @param y
+   * @return a new instance of Xlogy
+   */
+  public <T extends TType> Xlogy<T> xlogy(Operand<T> x, Operand<T> y) {
+    return Xlogy.create(scope, x, y);
+  }
+
+  /**
+   * Compute the Hurwitz zeta function \\(\zeta(x, q)\\).
+   *  <p>
+   *  The Hurwitz zeta function is defined as:
+   *  <p>
+   *  \\(\zeta(x, q) = \sum_{n=0}^{\infty} (q + n)^{-x}\\)
+   *
+   * @param <T> data type for {@code z()} output
+   * @param x
+   * @param q
+   * @return a new instance of Zeta
+   */
+  public <T extends TNumber> Zeta<T> zeta(Operand<T> x, Operand<T> q) {
+    return Zeta.create(scope, x, q);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/NnOps.java b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/NnOps.java
index fb57feee8c4..d6984bb1a95 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/NnOps.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/NnOps.java
@@ -638,8 +638,8 @@ public <T extends TNumber> CudnnRNNParamsToCanonical<T> cudnnRNNParamsToCanonica
    * @return a new instance of CudnnRnnParamsSize
    */
   public <U extends TNumber, T extends TNumber> CudnnRnnParamsSize<U> cudnnRnnParamsSize(
-      Operand<TInt32> numLayers, Operand<TInt32> numUnits, Operand<TInt32> inputSize, DataType<T> T,
-      DataType<U> S, CudnnRnnParamsSize.Options... options) {
+      Operand<TInt32> numLayers, Operand<TInt32> numUnits, Operand<TInt32> inputSize, Class<T> T,
+      Class<U> S, CudnnRnnParamsSize.Options... options) {
     return CudnnRnnParamsSize.create(scope, numLayers, numUnits, inputSize, T, S, options);
   }
 
@@ -1497,7 +1497,7 @@ public <T extends TNumber> MaxPoolWithArgmax<T, TInt64> maxPoolWithArgmax(Operan
    * @return a new instance of MaxPoolWithArgmax
    */
   public <T extends TNumber, U extends TNumber> MaxPoolWithArgmax<T, U> maxPoolWithArgmax(
-      Operand<T> input, List<Long> ksize, List<Long> strides, DataType<U> Targmax, String padding,
+      Operand<T> input, List<Long> ksize, List<Long> strides, Class<U> Targmax, String padding,
       MaxPoolWithArgmax.Options... options) {
     return MaxPoolWithArgmax.create(scope, input, ksize, strides, Targmax, padding, options);
   }
@@ -1584,7 +1584,7 @@ public <U extends TType, T extends TType> QuantizedBatchNormWithGlobalNormalizat
       Operand<T> t, Operand<TFloat32> tMin, Operand<TFloat32> tMax, Operand<T> m,
       Operand<TFloat32> mMin, Operand<TFloat32> mMax, Operand<T> v, Operand<TFloat32> vMin,
       Operand<TFloat32> vMax, Operand<T> beta, Operand<TFloat32> betaMin, Operand<TFloat32> betaMax,
-      Operand<T> gamma, Operand<TFloat32> gammaMin, Operand<TFloat32> gammaMax, DataType<U> outType,
+      Operand<T> gamma, Operand<TFloat32> gammaMin, Operand<TFloat32> gammaMax, Class<U> outType,
       Float varianceEpsilon, Boolean scaleAfterNormalization) {
     return QuantizedBatchNormWithGlobalNormalization.create(scope, t, tMin, tMax, m, mMin, mMax, v, vMin, vMax, beta, betaMin, betaMax, gamma, gammaMin, gammaMax, outType, varianceEpsilon, scaleAfterNormalization);
   }
@@ -1606,7 +1606,7 @@ public <U extends TType, T extends TType> QuantizedBatchNormWithGlobalNormalizat
    */
   public <V extends TType, T extends TType, U extends TType> QuantizedBiasAdd<V> quantizedBiasAdd(
       Operand<T> input, Operand<U> bias, Operand<TFloat32> minInput, Operand<TFloat32> maxInput,
-      Operand<TFloat32> minBias, Operand<TFloat32> maxBias, DataType<V> outType) {
+      Operand<TFloat32> minBias, Operand<TFloat32> maxBias, Class<V> outType) {
     return QuantizedBiasAdd.create(scope, input, bias, minInput, maxInput, minBias, maxBias, outType);
   }
 
@@ -1634,7 +1634,7 @@ public <V extends TType, T extends TType, U extends TType> QuantizedBiasAdd<V> q
    */
   public <V extends TType, T extends TType, U extends TType> QuantizedConv2d<V> quantizedConv2d(
       Operand<T> input, Operand<U> filter, Operand<TFloat32> minInput, Operand<TFloat32> maxInput,
-      Operand<TFloat32> minFilter, Operand<TFloat32> maxFilter, DataType<V> outType,
+      Operand<TFloat32> minFilter, Operand<TFloat32> maxFilter, Class<V> outType,
       List<Long> strides, String padding, QuantizedConv2d.Options... options) {
     return QuantizedConv2d.create(scope, input, filter, minInput, maxInput, minFilter, maxFilter, outType, strides, padding, options);
   }
@@ -1685,7 +1685,7 @@ public <T extends TType> QuantizedMaxPool<T> quantizedMaxPool(Operand<T> input,
    * @return a new instance of QuantizedRelu
    */
   public <U extends TType, T extends TType> QuantizedRelu<U> quantizedRelu(Operand<T> features,
-      Operand<TFloat32> minFeatures, Operand<TFloat32> maxFeatures, DataType<U> outType) {
+      Operand<TFloat32> minFeatures, Operand<TFloat32> maxFeatures, Class<U> outType) {
     return QuantizedRelu.create(scope, features, minFeatures, maxFeatures, outType);
   }
 
@@ -1700,7 +1700,7 @@ public <U extends TType, T extends TType> QuantizedRelu<U> quantizedRelu(Operand
    * @return a new instance of QuantizedRelu6
    */
   public <U extends TType, T extends TType> QuantizedRelu6<U> quantizedRelu6(Operand<T> features,
-      Operand<TFloat32> minFeatures, Operand<TFloat32> maxFeatures, DataType<U> outType) {
+      Operand<TFloat32> minFeatures, Operand<TFloat32> maxFeatures, Class<U> outType) {
     return QuantizedRelu6.create(scope, features, minFeatures, maxFeatures, outType);
   }
 
@@ -1717,7 +1717,7 @@ public <U extends TType, T extends TType> QuantizedRelu6<U> quantizedRelu6(Opera
    */
   public <U extends TType, T extends TType> QuantizedReluX<U> quantizedReluX(Operand<T> features,
       Operand<TFloat32> maxValue, Operand<TFloat32> minFeatures, Operand<TFloat32> maxFeatures,
-      DataType<U> outType) {
+      Class<U> outType) {
     return QuantizedReluX.create(scope, features, maxValue, minFeatures, maxFeatures, outType);
   }
 
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/NnRawOps.java b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/NnRawOps.java
new file mode 100644
index 00000000000..d9147af3934
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/NnRawOps.java
@@ -0,0 +1,74 @@
+// Copyright 2020 The TensorFlow Authors. All Rights Reserved.
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+// ==============================================================================
+//
+// This class has been generated, DO NOT EDIT!
+//
+package org.tensorflow.op;
+
+import org.tensorflow.Operand;
+import org.tensorflow.op.nn.raw.SoftmaxCrossEntropyWithLogits;
+import org.tensorflow.op.nn.raw.SparseSoftmaxCrossEntropyWithLogits;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * An API for building {@code nn.raw} operations as {@link Op Op}s
+ *
+ * @see {@link Ops}
+ */
+public final class NnRawOps {
+  private final Scope scope;
+
+  NnRawOps(Scope scope) {
+    this.scope = scope;
+  }
+
+  /**
+   * Computes softmax cross entropy cost and gradients to backpropagate.
+   *  <p>
+   *  Inputs are the logits, not probabilities.
+   *
+   * @param <T> data type for {@code loss()} output
+   * @param features batch_size x num_classes matrix
+   * @param labels batch_size x num_classes matrix
+   *  The caller must ensure that each batch of labels represents a valid
+   *  probability distribution.
+   * @return a new instance of SoftmaxCrossEntropyWithLogits
+   */
+  public <T extends TNumber> SoftmaxCrossEntropyWithLogits<T> softmaxCrossEntropyWithLogits(
+      Operand<T> features, Operand<T> labels) {
+    return SoftmaxCrossEntropyWithLogits.create(scope, features, labels);
+  }
+
+  /**
+   * Computes softmax cross entropy cost and gradients to backpropagate.
+   *  <p>
+   *  Unlike `SoftmaxCrossEntropyWithLogits`, this operation does not accept
+   *  a matrix of label probabilities, but rather a single label per row
+   *  of features.  This label is considered to have probability 1.0 for the
+   *  given row.
+   *  <p>
+   *  Inputs are the logits, not probabilities.
+   *
+   * @param <T> data type for {@code loss()} output
+   * @param features batch_size x num_classes matrix
+   * @param labels batch_size vector with values in [0, num_classes).
+   *  This is the label for the given minibatch entry.
+   * @return a new instance of SparseSoftmaxCrossEntropyWithLogits
+   */
+  public <T extends TNumber, U extends TNumber> SparseSoftmaxCrossEntropyWithLogits<T> sparseSoftmaxCrossEntropyWithLogits(
+      Operand<T> features, Operand<U> labels) {
+    return SparseSoftmaxCrossEntropyWithLogits.create(scope, features, labels);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/Ops.java b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/Ops.java
new file mode 100644
index 00000000000..f1abd40eb39
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/Ops.java
@@ -0,0 +1,7728 @@
+// Copyright 2020 The TensorFlow Authors. All Rights Reserved.
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+// ==============================================================================
+//
+// This class has been generated, DO NOT EDIT!
+//
+package org.tensorflow.op;
+
+import java.nio.charset.Charset;
+import java.util.List;
+import org.tensorflow.EagerSession;
+import org.tensorflow.ExecutionEnvironment;
+import org.tensorflow.Operand;
+import org.tensorflow.ndarray.BooleanNdArray;
+import org.tensorflow.ndarray.ByteNdArray;
+import org.tensorflow.ndarray.DoubleNdArray;
+import org.tensorflow.ndarray.FloatNdArray;
+import org.tensorflow.ndarray.IntNdArray;
+import org.tensorflow.ndarray.LongNdArray;
+import org.tensorflow.ndarray.NdArray;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.ndarray.buffer.BooleanDataBuffer;
+import org.tensorflow.ndarray.buffer.ByteDataBuffer;
+import org.tensorflow.ndarray.buffer.DataBuffer;
+import org.tensorflow.ndarray.buffer.DoubleDataBuffer;
+import org.tensorflow.ndarray.buffer.FloatDataBuffer;
+import org.tensorflow.ndarray.buffer.IntDataBuffer;
+import org.tensorflow.ndarray.buffer.LongDataBuffer;
+import org.tensorflow.op.core.Abort;
+import org.tensorflow.op.core.All;
+import org.tensorflow.op.core.Any;
+import org.tensorflow.op.core.AssertThat;
+import org.tensorflow.op.core.Assign;
+import org.tensorflow.op.core.AssignAdd;
+import org.tensorflow.op.core.AssignAddVariableOp;
+import org.tensorflow.op.core.AssignSub;
+import org.tensorflow.op.core.AssignSubVariableOp;
+import org.tensorflow.op.core.AssignVariableOp;
+import org.tensorflow.op.core.Barrier;
+import org.tensorflow.op.core.BarrierClose;
+import org.tensorflow.op.core.BarrierIncompleteSize;
+import org.tensorflow.op.core.BarrierInsertMany;
+import org.tensorflow.op.core.BarrierReadySize;
+import org.tensorflow.op.core.BarrierTakeMany;
+import org.tensorflow.op.core.Batch;
+import org.tensorflow.op.core.BatchToSpace;
+import org.tensorflow.op.core.BatchToSpaceNd;
+import org.tensorflow.op.core.Bitcast;
+import org.tensorflow.op.core.BroadcastDynamicShape;
+import org.tensorflow.op.core.BroadcastTo;
+import org.tensorflow.op.core.Bucketize;
+import org.tensorflow.op.core.ClipByValue;
+import org.tensorflow.op.core.Concat;
+import org.tensorflow.op.core.Constant;
+import org.tensorflow.op.core.ConsumeMutexLock;
+import org.tensorflow.op.core.ControlTrigger;
+import org.tensorflow.op.core.CountUpTo;
+import org.tensorflow.op.core.DeepCopy;
+import org.tensorflow.op.core.DeleteSessionTensor;
+import org.tensorflow.op.core.DestroyResourceOp;
+import org.tensorflow.op.core.DestroyTemporaryVariable;
+import org.tensorflow.op.core.DynamicPartition;
+import org.tensorflow.op.core.DynamicStitch;
+import org.tensorflow.op.core.EditDistance;
+import org.tensorflow.op.core.Empty;
+import org.tensorflow.op.core.EmptyTensorList;
+import org.tensorflow.op.core.EnsureShape;
+import org.tensorflow.op.core.ExpandDims;
+import org.tensorflow.op.core.ExtractVolumePatches;
+import org.tensorflow.op.core.Fill;
+import org.tensorflow.op.core.Fingerprint;
+import org.tensorflow.op.core.Gather;
+import org.tensorflow.op.core.GatherNd;
+import org.tensorflow.op.core.GetSessionHandle;
+import org.tensorflow.op.core.GetSessionTensor;
+import org.tensorflow.op.core.Gradients;
+import org.tensorflow.op.core.GuaranteeConst;
+import org.tensorflow.op.core.HashTable;
+import org.tensorflow.op.core.Helpers;
+import org.tensorflow.op.core.HistogramFixedWidth;
+import org.tensorflow.op.core.Identity;
+import org.tensorflow.op.core.IdentityN;
+import org.tensorflow.op.core.ImmutableConst;
+import org.tensorflow.op.core.Init;
+import org.tensorflow.op.core.InitializeTable;
+import org.tensorflow.op.core.InitializeTableFromTextFile;
+import org.tensorflow.op.core.InplaceAdd;
+import org.tensorflow.op.core.InplaceSub;
+import org.tensorflow.op.core.InplaceUpdate;
+import org.tensorflow.op.core.IsVariableInitialized;
+import org.tensorflow.op.core.LookupTableExport;
+import org.tensorflow.op.core.LookupTableFind;
+import org.tensorflow.op.core.LookupTableImport;
+import org.tensorflow.op.core.LookupTableInsert;
+import org.tensorflow.op.core.LookupTableSize;
+import org.tensorflow.op.core.LoopCond;
+import org.tensorflow.op.core.MapClear;
+import org.tensorflow.op.core.MapIncompleteSize;
+import org.tensorflow.op.core.MapPeek;
+import org.tensorflow.op.core.MapSize;
+import org.tensorflow.op.core.MapStage;
+import org.tensorflow.op.core.MapUnstage;
+import org.tensorflow.op.core.MapUnstageNoKey;
+import org.tensorflow.op.core.Max;
+import org.tensorflow.op.core.Merge;
+import org.tensorflow.op.core.Min;
+import org.tensorflow.op.core.MirrorPad;
+import org.tensorflow.op.core.MlirPassthroughOp;
+import org.tensorflow.op.core.MutableDenseHashTable;
+import org.tensorflow.op.core.MutableHashTable;
+import org.tensorflow.op.core.MutableHashTableOfTensors;
+import org.tensorflow.op.core.Mutex;
+import org.tensorflow.op.core.MutexLock;
+import org.tensorflow.op.core.NextIteration;
+import org.tensorflow.op.core.NoOp;
+import org.tensorflow.op.core.OneHot;
+import org.tensorflow.op.core.OnesLike;
+import org.tensorflow.op.core.OrderedMapClear;
+import org.tensorflow.op.core.OrderedMapIncompleteSize;
+import org.tensorflow.op.core.OrderedMapPeek;
+import org.tensorflow.op.core.OrderedMapSize;
+import org.tensorflow.op.core.OrderedMapStage;
+import org.tensorflow.op.core.OrderedMapUnstage;
+import org.tensorflow.op.core.OrderedMapUnstageNoKey;
+import org.tensorflow.op.core.Pad;
+import org.tensorflow.op.core.ParallelConcat;
+import org.tensorflow.op.core.ParallelDynamicStitch;
+import org.tensorflow.op.core.Placeholder;
+import org.tensorflow.op.core.PlaceholderWithDefault;
+import org.tensorflow.op.core.Print;
+import org.tensorflow.op.core.Prod;
+import org.tensorflow.op.core.QuantizedReshape;
+import org.tensorflow.op.core.Range;
+import org.tensorflow.op.core.Rank;
+import org.tensorflow.op.core.ReadVariableOp;
+import org.tensorflow.op.core.ReduceAll;
+import org.tensorflow.op.core.ReduceAny;
+import org.tensorflow.op.core.ReduceMax;
+import org.tensorflow.op.core.ReduceMin;
+import org.tensorflow.op.core.ReduceProd;
+import org.tensorflow.op.core.ReduceSum;
+import org.tensorflow.op.core.RefNextIteration;
+import org.tensorflow.op.core.RefSelect;
+import org.tensorflow.op.core.RefSwitch;
+import org.tensorflow.op.core.RemoteFusedGraphExecute;
+import org.tensorflow.op.core.Reshape;
+import org.tensorflow.op.core.ResourceCountUpTo;
+import org.tensorflow.op.core.ResourceGather;
+import org.tensorflow.op.core.ResourceGatherNd;
+import org.tensorflow.op.core.ResourceScatterAdd;
+import org.tensorflow.op.core.ResourceScatterDiv;
+import org.tensorflow.op.core.ResourceScatterMax;
+import org.tensorflow.op.core.ResourceScatterMin;
+import org.tensorflow.op.core.ResourceScatterMul;
+import org.tensorflow.op.core.ResourceScatterNdAdd;
+import org.tensorflow.op.core.ResourceScatterNdMax;
+import org.tensorflow.op.core.ResourceScatterNdMin;
+import org.tensorflow.op.core.ResourceScatterNdSub;
+import org.tensorflow.op.core.ResourceScatterNdUpdate;
+import org.tensorflow.op.core.ResourceScatterSub;
+import org.tensorflow.op.core.ResourceScatterUpdate;
+import org.tensorflow.op.core.ResourceStridedSliceAssign;
+import org.tensorflow.op.core.Reverse;
+import org.tensorflow.op.core.ReverseSequence;
+import org.tensorflow.op.core.Roll;
+import org.tensorflow.op.core.Rpc;
+import org.tensorflow.op.core.ScatterAdd;
+import org.tensorflow.op.core.ScatterDiv;
+import org.tensorflow.op.core.ScatterMax;
+import org.tensorflow.op.core.ScatterMin;
+import org.tensorflow.op.core.ScatterMul;
+import org.tensorflow.op.core.ScatterNd;
+import org.tensorflow.op.core.ScatterNdAdd;
+import org.tensorflow.op.core.ScatterNdNonAliasingAdd;
+import org.tensorflow.op.core.ScatterNdSub;
+import org.tensorflow.op.core.ScatterNdUpdate;
+import org.tensorflow.op.core.ScatterSub;
+import org.tensorflow.op.core.ScatterUpdate;
+import org.tensorflow.op.core.Select;
+import org.tensorflow.op.core.SetDiff1d;
+import org.tensorflow.op.core.SetSize;
+import org.tensorflow.op.core.ShapeN;
+import org.tensorflow.op.core.Size;
+import org.tensorflow.op.core.Skipgram;
+import org.tensorflow.op.core.Slice;
+import org.tensorflow.op.core.Snapshot;
+import org.tensorflow.op.core.SpaceToBatchNd;
+import org.tensorflow.op.core.Split;
+import org.tensorflow.op.core.SplitV;
+import org.tensorflow.op.core.Squeeze;
+import org.tensorflow.op.core.Stack;
+import org.tensorflow.op.core.Stage;
+import org.tensorflow.op.core.StageClear;
+import org.tensorflow.op.core.StagePeek;
+import org.tensorflow.op.core.StageSize;
+import org.tensorflow.op.core.StopGradient;
+import org.tensorflow.op.core.StridedSlice;
+import org.tensorflow.op.core.StridedSliceAssign;
+import org.tensorflow.op.core.StridedSliceGrad;
+import org.tensorflow.op.core.Sum;
+import org.tensorflow.op.core.SwitchCond;
+import org.tensorflow.op.core.TemporaryVariable;
+import org.tensorflow.op.core.TensorArray;
+import org.tensorflow.op.core.TensorArrayClose;
+import org.tensorflow.op.core.TensorArrayConcat;
+import org.tensorflow.op.core.TensorArrayGather;
+import org.tensorflow.op.core.TensorArrayGrad;
+import org.tensorflow.op.core.TensorArrayGradWithShape;
+import org.tensorflow.op.core.TensorArrayPack;
+import org.tensorflow.op.core.TensorArrayRead;
+import org.tensorflow.op.core.TensorArrayScatter;
+import org.tensorflow.op.core.TensorArraySize;
+import org.tensorflow.op.core.TensorArraySplit;
+import org.tensorflow.op.core.TensorArrayUnpack;
+import org.tensorflow.op.core.TensorArrayWrite;
+import org.tensorflow.op.core.TensorListConcat;
+import org.tensorflow.op.core.TensorListConcatLists;
+import org.tensorflow.op.core.TensorListElementShape;
+import org.tensorflow.op.core.TensorListFromTensor;
+import org.tensorflow.op.core.TensorListGather;
+import org.tensorflow.op.core.TensorListGetItem;
+import org.tensorflow.op.core.TensorListLength;
+import org.tensorflow.op.core.TensorListPopBack;
+import org.tensorflow.op.core.TensorListPushBack;
+import org.tensorflow.op.core.TensorListPushBackBatch;
+import org.tensorflow.op.core.TensorListReserve;
+import org.tensorflow.op.core.TensorListResize;
+import org.tensorflow.op.core.TensorListScatter;
+import org.tensorflow.op.core.TensorListScatterIntoExistingList;
+import org.tensorflow.op.core.TensorListSetItem;
+import org.tensorflow.op.core.TensorListSplit;
+import org.tensorflow.op.core.TensorListStack;
+import org.tensorflow.op.core.TensorScatterNdAdd;
+import org.tensorflow.op.core.TensorScatterNdMax;
+import org.tensorflow.op.core.TensorScatterNdMin;
+import org.tensorflow.op.core.TensorScatterNdSub;
+import org.tensorflow.op.core.TensorScatterNdUpdate;
+import org.tensorflow.op.core.TensorStridedSliceUpdate;
+import org.tensorflow.op.core.Tile;
+import org.tensorflow.op.core.Timestamp;
+import org.tensorflow.op.core.TryRpc;
+import org.tensorflow.op.core.Unbatch;
+import org.tensorflow.op.core.UnbatchGrad;
+import org.tensorflow.op.core.Unique;
+import org.tensorflow.op.core.UniqueWithCounts;
+import org.tensorflow.op.core.UnravelIndex;
+import org.tensorflow.op.core.Unstack;
+import org.tensorflow.op.core.Unstage;
+import org.tensorflow.op.core.VarHandleOp;
+import org.tensorflow.op.core.VarIsInitializedOp;
+import org.tensorflow.op.core.Variable;
+import org.tensorflow.op.core.VariableShape;
+import org.tensorflow.op.core.Where;
+import org.tensorflow.op.core.XlaSpmdFullToShardShape;
+import org.tensorflow.op.core.XlaSpmdShardToFullShape;
+import org.tensorflow.op.core.Zeros;
+import org.tensorflow.op.core.ZerosLike;
+import org.tensorflow.types.TBool;
+import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.TFloat64;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.TString;
+import org.tensorflow.types.TUint8;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * An API for building operations as {@link Op Op}s
+ * <p>
+ * Any operation wrapper found in the classpath properly annotated as an{@link org.tensorflow.op.annotation.Operator @Operator} is exposed
+ * by this API or one of its subgroup.
+ * <p>Example usage:
+ * <pre>{@code
+ * try (Graph g = new Graph()) {
+ *   Ops tf = Ops.create(g);
+ *   // Operations are typed classes with convenience
+ *   // builders in Ops.
+ *   Constant<TInt32> three = tf.constant(3);
+ *   // Single-result operations implement the Operand
+ *   // interface, so this works too.
+ *   Operand<TInt32> four = tf.constant(4);
+ *   // Most builders are found within a group, and accept
+ *   // Operand types as operands
+ *   Operand<TInt32> nine = tf.math.add(four, tf.constant(5));
+ *   // Multi-result operations however offer methods to
+ *   // select a particular result for use.
+ *   Operand<TInt32> result = 
+ *       tf.math.add(tf.unique(s, a).y(), b);
+ *   // Optional attributes
+ *   tf.linalg.matMul(a, b, MatMul.transposeA(true));
+ *   // Naming operators
+ *   tf.withName("foo").constant(5); // name "foo"
+ *   // Names can exist in a hierarchy
+ *   Ops sub = tf.withSubScope("sub");
+ *   sub.withName("bar").constant(4); // "sub/bar"
+ * }
+ * }</pre>
+ */
+public final class Ops {
+  public final NnOps nn;
+
+  public final SummaryOps summary;
+
+  public final ImageOps image;
+
+  public final RaggedOps ragged;
+
+  public final DataOps data;
+
+  public final ShapeOps shape;
+
+  public final IoOps io;
+
+  public final DtypesOps dtypes;
+
+  public final XlaOps xla;
+
+  public final LinalgOps linalg;
+
+  public final RandomOps random;
+
+  public final StringsOps strings;
+
+  public final SparseOps sparse;
+
+  public final BitwiseOps bitwise;
+
+  public final MathOps math;
+
+  public final AudioOps audio;
+
+  public final SignalOps signal;
+
+  public final QuantizationOps quantization;
+
+  public final TrainOps train;
+
+  private final Scope scope;
+
+  private Ops(Scope scope) {
+    this.scope = scope;
+    nn = new NnOps(scope);
+    summary = new SummaryOps(scope);
+    image = new ImageOps(scope);
+    ragged = new RaggedOps(scope);
+    data = new DataOps(scope);
+    shape = new ShapeOps(scope);
+    io = new IoOps(scope);
+    dtypes = new DtypesOps(scope);
+    xla = new XlaOps(scope);
+    linalg = new LinalgOps(scope);
+    random = new RandomOps(scope);
+    strings = new StringsOps(scope);
+    sparse = new SparseOps(scope);
+    bitwise = new BitwiseOps(scope);
+    math = new MathOps(scope);
+    audio = new AudioOps(scope);
+    signal = new SignalOps(scope);
+    quantization = new QuantizationOps(scope);
+    train = new TrainOps(scope);
+  }
+
+  /**
+   * Raise a exception to abort the process when called.
+   *  <p>
+   *  If exit_without_error is true, the process will exit normally,
+   *  otherwise it will exit with a SIGABORT signal.
+   *  <p>
+   *  Returns nothing but an exception.
+   *
+   * @param options carries optional attributes values
+   * @return a new instance of Abort
+   */
+  public Abort abort(Abort.Options... options) {
+    return Abort.create(scope, options);
+  }
+
+  /**
+   * Computes the "logical and" of elements across dimensions of a tensor.
+   *  <p>
+   *  Reduces `input` along the dimensions given in `axis`. Unless
+   *  `keep_dims` is true, the rank of the tensor is reduced by 1 for each entry in
+   *  `axis`. If `keep_dims` is true, the reduced dimensions are
+   *  retained with length 1.
+   *
+   * @param input The tensor to reduce.
+   * @param axis The dimensions to reduce. Must be in the range
+   *  `[-rank(input), rank(input))`.
+   * @param options carries optional attributes values
+   * @return a new instance of All
+   */
+  public <T extends TNumber> All all(Operand<TBool> input, Operand<T> axis,
+      All.Options... options) {
+    return All.create(scope, input, axis, options);
+  }
+
+  /**
+   * Computes the "logical or" of elements across dimensions of a tensor.
+   *  <p>
+   *  Reduces `input` along the dimensions given in `axis`. Unless
+   *  `keep_dims` is true, the rank of the tensor is reduced by 1 for each entry in
+   *  `axis`. If `keep_dims` is true, the reduced dimensions are
+   *  retained with length 1.
+   *
+   * @param input The tensor to reduce.
+   * @param axis The dimensions to reduce. Must be in the range
+   *  `[-rank(input), rank(input))`.
+   * @param options carries optional attributes values
+   * @return a new instance of Any
+   */
+  public <T extends TNumber> Any any(Operand<TBool> input, Operand<T> axis,
+      Any.Options... options) {
+    return Any.create(scope, input, axis, options);
+  }
+
+  /**
+   * Creates a constant of {@code int} elements.
+   *
+   * @param scope is a scope used to add the underlying operation.
+   * @param data An array containing the values to put into the new constant.
+   * @return a float constant
+   */
+  public Constant<TInt32> array(int... data) {
+    return Constant.arrayOf(scope, data);
+  }
+
+  /**
+   * Creates a constant of {@code String} elements, using the default UTF-8 charset.
+   *
+   * @param scope is a scope used to add the underlying operation.
+   * @param data An array containing the values to put into the new constant.
+   * @return the {@code String} constant
+   */
+  public Constant<TString> array(String... data) {
+    return Constant.arrayOf(scope, data);
+  }
+
+  /**
+   * Creates a constant of {@code boolean} elements.
+   *
+   * @param scope is a scope used to add the underlying operation.
+   * @param data An array containing the values to put into the new constant.
+   * @return a boolean constant
+   */
+  public Constant<TBool> array(boolean... data) {
+    return Constant.arrayOf(scope, data);
+  }
+
+  /**
+   * Creates a constant of {@code long} elements.
+   *
+   * @param scope is a scope used to add the underlying operation.
+   * @param data An array containing the values to put into the new constant.
+   * @return a long constant
+   */
+  public Constant<TInt64> array(long... data) {
+    return Constant.arrayOf(scope, data);
+  }
+
+  /**
+   * Creates a constant of {@code float} elements.
+   *
+   * @param scope is a scope used to add the underlying operation.
+   * @param data An array containing the values to put into the new constant.
+   * @return a float constant
+   */
+  public Constant<TFloat32> array(float... data) {
+    return Constant.arrayOf(scope, data);
+  }
+
+  /**
+   * Creates a constant of {@code double} elements.
+   *
+   * @param scope is a scope used to add the underlying operation.
+   * @param data An array containing the values to put into the new constant.
+   * @return a double constant
+   */
+  public Constant<TFloat64> array(double... data) {
+    return Constant.arrayOf(scope, data);
+  }
+
+  /**
+   * Creates a constant of {@code byte} elements.
+   *
+   * @param scope is a scope used to add the underlying operation.
+   * @param data An array containing the values to put into the new constant.
+   * @return a byte constant
+   */
+  public Constant<TUint8> array(byte... data) {
+    return Constant.arrayOf(scope, data);
+  }
+
+  /**
+   * Creates a constant of {@code String} elements, using the given charset.
+   *
+   * @param scope is a scope used to add the underlying operation.
+   * @param charset charset for encoding/decoding strings bytes.
+   * @param data An array containing the values to put into the new constant. String elements are
+   *      sequences of bytes from the last array dimension.
+   * @return the {@code String} constant
+   */
+  public Constant<TString> array(Charset charset, String... data) {
+    return Constant.arrayOf(scope, charset, data);
+  }
+
+  /**
+   * Asserts that the given condition is true.
+   *  <p>
+   *  If `condition` evaluates to false, print the list of tensors in `data`.
+   *  `summarize` determines how many entries of the tensors to print.
+   *
+   * @param condition The condition to evaluate.
+   * @param data The tensors to print out when condition is false.
+   * @param options carries optional attributes values
+   * @return a new instance of AssertThat
+   */
+  public AssertThat assertThat(Operand<TBool> condition, Iterable<Operand<?>> data,
+      AssertThat.Options... options) {
+    return AssertThat.create(scope, condition, data, options);
+  }
+
+  /**
+   * Update 'ref' by assigning 'value' to it.
+   *  <p>
+   *  This operation outputs "ref" after the assignment is done.
+   *  This makes it easier to chain operations that need to use the reset value.
+   *
+   * @param <T> data type for {@code outputRef()} output
+   * @param ref Should be from a `Variable` node. May be uninitialized.
+   * @param value The value to be assigned to the variable.
+   * @param options carries optional attributes values
+   * @return a new instance of Assign
+   */
+  public <T extends TType> Assign<T> assign(Operand<T> ref, Operand<T> value,
+      Assign.Options... options) {
+    return Assign.create(scope, ref, value, options);
+  }
+
+  /**
+   * Update 'ref' by adding 'value' to it.
+   *  <p>
+   *  This operation outputs "ref" after the update is done.
+   *  This makes it easier to chain operations that need to use the reset value.
+   *
+   * @param <T> data type for {@code outputRef()} output
+   * @param ref Should be from a `Variable` node.
+   * @param value The value to be added to the variable.
+   * @param options carries optional attributes values
+   * @return a new instance of AssignAdd
+   */
+  public <T extends TType> AssignAdd<T> assignAdd(Operand<T> ref, Operand<T> value,
+      AssignAdd.Options... options) {
+    return AssignAdd.create(scope, ref, value, options);
+  }
+
+  /**
+   * Adds a value to the current value of a variable.
+   *  <p>
+   *  Any ReadVariableOp with a control dependency on this op is guaranteed to
+   *  see the incremented value or a subsequent newer one.
+   *
+   * @param resource handle to the resource in which to store the variable.
+   * @param value the value by which the variable will be incremented.
+   * @return a new instance of AssignAddVariableOp
+   */
+  public <T extends TType> AssignAddVariableOp assignAddVariableOp(Operand<?> resource,
+      Operand<T> value) {
+    return AssignAddVariableOp.create(scope, resource, value);
+  }
+
+  /**
+   * Update 'ref' by subtracting 'value' from it.
+   *  <p>
+   *  This operation outputs "ref" after the update is done.
+   *  This makes it easier to chain operations that need to use the reset value.
+   *
+   * @param <T> data type for {@code outputRef()} output
+   * @param ref Should be from a `Variable` node.
+   * @param value The value to be subtracted to the variable.
+   * @param options carries optional attributes values
+   * @return a new instance of AssignSub
+   */
+  public <T extends TType> AssignSub<T> assignSub(Operand<T> ref, Operand<T> value,
+      AssignSub.Options... options) {
+    return AssignSub.create(scope, ref, value, options);
+  }
+
+  /**
+   * Subtracts a value from the current value of a variable.
+   *  <p>
+   *  Any ReadVariableOp with a control dependency on this op is guaranteed to
+   *  see the decremented value or a subsequent newer one.
+   *
+   * @param resource handle to the resource in which to store the variable.
+   * @param value the value by which the variable will be incremented.
+   * @return a new instance of AssignSubVariableOp
+   */
+  public <T extends TType> AssignSubVariableOp assignSubVariableOp(Operand<?> resource,
+      Operand<T> value) {
+    return AssignSubVariableOp.create(scope, resource, value);
+  }
+
+  /**
+   * Assigns a new value to a variable.
+   *  <p>
+   *  Any ReadVariableOp with a control dependency on this op is guaranteed to return
+   *  this value or a subsequent newer value of the variable.
+   *
+   * @param resource handle to the resource in which to store the variable.
+   * @param value the value to set the new tensor to use.
+   * @return a new instance of AssignVariableOp
+   */
+  public <T extends TType> AssignVariableOp assignVariableOp(Operand<?> resource,
+      Operand<T> value) {
+    return AssignVariableOp.create(scope, resource, value);
+  }
+
+  /**
+   * Defines a barrier that persists across different graph executions.
+   *  <p>
+   *  A barrier represents a key-value map, where each key is a string, and
+   *  each value is a tuple of tensors.
+   *  <p>
+   *  At runtime, the barrier contains 'complete' and 'incomplete'
+   *  elements. A complete element has defined tensors for all components of
+   *  its value tuple, and may be accessed using BarrierTakeMany. An
+   *  incomplete element has some undefined components in its value tuple,
+   *  and may be updated using BarrierInsertMany.
+   *
+   * @param componentTypes The type of each component in a value.
+   * @param options carries optional attributes values
+   * @return a new instance of Barrier
+   */
+  public Barrier barrier(List<Class<? extends TType>> componentTypes, Barrier.Options... options) {
+    return Barrier.create(scope, componentTypes, options);
+  }
+
+  /**
+   * Closes the given barrier.
+   *  <p>
+   *  This operation signals that no more new elements will be inserted in the
+   *  given barrier. Subsequent InsertMany that try to introduce a new key will fail.
+   *  Subsequent InsertMany operations that just add missing components to already
+   *  existing elements will continue to succeed. Subsequent TakeMany operations will
+   *  continue to succeed if sufficient completed elements remain in the barrier.
+   *  Subsequent TakeMany operations that would block will fail immediately.
+   *
+   * @param handle The handle to a barrier.
+   * @param options carries optional attributes values
+   * @return a new instance of BarrierClose
+   */
+  public BarrierClose barrierClose(Operand<TString> handle, BarrierClose.Options... options) {
+    return BarrierClose.create(scope, handle, options);
+  }
+
+  /**
+   * Computes the number of incomplete elements in the given barrier.
+   *
+   * @param handle The handle to a barrier.
+   * @return a new instance of BarrierIncompleteSize
+   */
+  public BarrierIncompleteSize barrierIncompleteSize(Operand<TString> handle) {
+    return BarrierIncompleteSize.create(scope, handle);
+  }
+
+  /**
+   * For each key, assigns the respective value to the specified component.
+   *  <p>
+   *  If a key is not found in the barrier, this operation will create a new
+   *  incomplete element. If a key is found in the barrier, and the element
+   *  already has a value at component_index, this operation will fail with
+   *  INVALID_ARGUMENT, and leave the barrier in an undefined state.
+   *
+   * @param handle The handle to a barrier.
+   * @param keys A one-dimensional tensor of keys, with length n.
+   * @param values An any-dimensional tensor of values, which are associated with the
+   *  respective keys. The 0th dimension must have length n.
+   * @param componentIndex The component of the barrier elements that is being assigned.
+   * @return a new instance of BarrierInsertMany
+   */
+  public <T extends TType> BarrierInsertMany barrierInsertMany(Operand<TString> handle,
+      Operand<TString> keys, Operand<T> values, Long componentIndex) {
+    return BarrierInsertMany.create(scope, handle, keys, values, componentIndex);
+  }
+
+  /**
+   * Computes the number of complete elements in the given barrier.
+   *
+   * @param handle The handle to a barrier.
+   * @return a new instance of BarrierReadySize
+   */
+  public BarrierReadySize barrierReadySize(Operand<TString> handle) {
+    return BarrierReadySize.create(scope, handle);
+  }
+
+  /**
+   * Takes the given number of completed elements from a barrier.
+   *  <p>
+   *  This operation concatenates completed-element component tensors along
+   *  the 0th dimension to make a single component tensor.
+   *  <p>
+   *  Elements come out of the barrier when they are complete, and in the order
+   *  in which they were placed into the barrier.  The indices output provides
+   *  information about the batch in which each element was originally inserted
+   *  into the barrier.
+   *
+   * @param handle The handle to a barrier.
+   * @param numElements A single-element tensor containing the number of elements to
+   *  take.
+   * @param componentTypes The type of each component in a value.
+   * @param options carries optional attributes values
+   * @return a new instance of BarrierTakeMany
+   */
+  public BarrierTakeMany barrierTakeMany(Operand<TString> handle, Operand<TInt32> numElements,
+      List<Class<? extends TType>> componentTypes, BarrierTakeMany.Options... options) {
+    return BarrierTakeMany.create(scope, handle, numElements, componentTypes, options);
+  }
+
+  /**
+   * Batches all input tensors nondeterministically.
+   *  <p>
+   *  When many instances of this Op are being run concurrently with the same
+   *  container/shared_name in the same device, some will output zero-shaped Tensors
+   *  and others will output Tensors of size up to max_batch_size.
+   *  <p>
+   *  All Tensors in in_tensors are batched together (so, for example, labels and
+   *  features should be batched with a single instance of this operation.
+   *  <p>
+   *  Each invocation of batch emits an `id` scalar which will be used to identify
+   *  this particular invocation when doing unbatch or its gradient.
+   *  <p>
+   *  Each op which emits a non-empty batch will also emit a non-empty batch_index
+   *  Tensor, which, is a [K, 3] matrix where each row contains the invocation's id,
+   *  start, and length of elements of each set of Tensors present in batched_tensors.
+   *  <p>
+   *  Batched tensors are concatenated along the first dimension, and all tensors in
+   *  in_tensors must have the first dimension of the same size.
+   *  <p>
+   *  in_tensors: The tensors to be batched.
+   *  num_batch_threads: Number of scheduling threads for processing batches of work.
+   *   Determines the number of batches processed in parallel.
+   *  max_batch_size: Batch sizes will never be bigger than this.
+   *  batch_timeout_micros: Maximum number of microseconds to wait before outputting
+   *   an incomplete batch.
+   *  allowed_batch_sizes: Optional list of allowed batch sizes. If left empty, does
+   *   nothing. Otherwise, supplies a list of batch sizes, causing the op to pad
+   *   batches up to one of those sizes. The entries must increase monotonically, and
+   *   the final entry must equal max_batch_size.
+   *  grad_timeout_micros: The timeout to use for the gradient. See Unbatch.
+   *  batched_tensors: Either empty tensors or a batch of concatenated Tensors.
+   *  batch_index: If out_tensors is non-empty, has information to invert it.
+   *  container: Controls the scope of sharing of this batch.
+   *  id: always contains a scalar with a unique ID for this invocation of Batch.
+   *  shared_name: Concurrently running instances of batch in the same device with the
+   *   same container and shared_name will batch their elements together. If left
+   *   empty, the op name will be used as the shared name.
+   *  T: the types of tensors to be batched.
+   *
+   * @param inTensors
+   * @param numBatchThreads
+   * @param maxBatchSize
+   * @param batchTimeoutMicros
+   * @param gradTimeoutMicros
+   * @param options carries optional attributes values
+   * @return a new instance of Batch
+   */
+  public Batch batch(Iterable<Operand<?>> inTensors, Long numBatchThreads, Long maxBatchSize,
+      Long batchTimeoutMicros, Long gradTimeoutMicros, Batch.Options... options) {
+    return Batch.create(scope, inTensors, numBatchThreads, maxBatchSize, batchTimeoutMicros, gradTimeoutMicros, options);
+  }
+
+  /**
+   * BatchToSpace for 4-D tensors of type T.
+   *  <p>
+   *  This is a legacy version of the more general BatchToSpaceND.
+   *  <p>
+   *  Rearranges (permutes) data from batch into blocks of spatial data, followed by
+   *  cropping. This is the reverse transformation of SpaceToBatch. More specifically,
+   *  this op outputs a copy of the input tensor where values from the `batch`
+   *  dimension are moved in spatial blocks to the `height` and `width` dimensions,
+   *  followed by cropping along the `height` and `width` dimensions.
+   *
+   * @param <T> data type for {@code output()} output
+   * @param input 4-D tensor with shape
+   *  `[batch<i>block_size</i>block_size, height_pad/block_size, width_pad/block_size,
+   *    depth]`. Note that the batch size of the input tensor must be divisible by
+   *  `block_size * block_size`.
+   * @param crops 2-D tensor of non-negative integers with shape `[2, 2]`. It specifies
+   *  how many elements to crop from the intermediate result across the spatial
+   *  dimensions as follows:
+   *  <p>
+   *      crops = [[crop_top, crop_bottom], [crop_left, crop_right]]
+   * @param blockSize
+   * @return a new instance of BatchToSpace
+   */
+  public <T extends TType, U extends TNumber> BatchToSpace<T> batchToSpace(Operand<T> input,
+      Operand<U> crops, Long blockSize) {
+    return BatchToSpace.create(scope, input, crops, blockSize);
+  }
+
+  /**
+   * BatchToSpace for N-D tensors of type T.
+   *  <p>
+   *  This operation reshapes the "batch" dimension 0 into `M + 1` dimensions of shape
+   *  `block_shape + [batch]`, interleaves these blocks back into the grid defined by
+   *  the spatial dimensions `[1, ..., M]`, to obtain a result with the same rank as
+   *  the input.  The spatial dimensions of this intermediate result are then
+   *  optionally cropped according to `crops` to produce the output.  This is the
+   *  reverse of SpaceToBatch.  See below for a precise description.
+   *
+   * @param <T> data type for {@code output()} output
+   * @param input N-D with shape `input_shape = [batch] + spatial_shape + remaining_shape`,
+   *  where spatial_shape has M dimensions.
+   * @param blockShape 1-D with shape `[M]`, all values must be >= 1.
+   * @param crops 2-D with shape `[M, 2]`, all values must be >= 0.
+   *    `crops[i] = [crop_start, crop_end]` specifies the amount to crop from input
+   *    dimension `i + 1`, which corresponds to spatial dimension `i`.  It is
+   *    required that
+   *    `crop_start[i] + crop_end[i] <= block_shape[i] * input_shape[i + 1]`.
+   *  <p>
+   *  This operation is equivalent to the following steps:
+   *  <p>
+   *  1. Reshape `input` to `reshaped` of shape:
+   *       [block_shape[0], ..., block_shape[M-1],
+   *        batch / prod(block_shape),
+   *        input_shape[1], ..., input_shape[N-1]]
+   *  <p>
+   *  2. Permute dimensions of `reshaped` to produce `permuted` of shape
+   *       [batch / prod(block_shape),
+   *  <p>
+   *        input_shape[1], block_shape[0],
+   *        ...,
+   *        input_shape[M], block_shape[M-1],
+   *  <p>
+   *        input_shape[M+1], ..., input_shape[N-1]]
+   *  <p>
+   *  3. Reshape `permuted` to produce `reshaped_permuted` of shape
+   *       [batch / prod(block_shape),
+   *  <p>
+   *        input_shape[1] * block_shape[0],
+   *        ...,
+   *        input_shape[M] * block_shape[M-1],
+   *  <p>
+   *        input_shape[M+1],
+   *        ...,
+   *        input_shape[N-1]]
+   *  <p>
+   *  4. Crop the start and end of dimensions `[1, ..., M]` of
+   *     `reshaped_permuted` according to `crops` to produce the output of shape:
+   *       [batch / prod(block_shape),
+   *  <p>
+   *        input_shape[1] * block_shape[0] - crops[0,0] - crops[0,1],
+   *        ...,
+   *        input_shape[M] * block_shape[M-1] - crops[M-1,0] - crops[M-1,1],
+   *  <p>
+   *        input_shape[M+1], ..., input_shape[N-1]]
+   *  <p>
+   *  Some examples:
+   *  <p>
+   *  (1) For the following input of shape `[4, 1, 1, 1]`, `block_shape = [2, 2]`, and
+   *      `crops = [[0, 0], [0, 0]]`:
+   *  <pre>{@code
+   *  [[[[1]]], [[[2]]], [[[3]]], [[[4]]]]
+   *  }</pre>
+   *  The output tensor has shape `[1, 2, 2, 1]` and value:
+   *  <pre>{@code
+   *  x = [[[[1], [2]], [[3], [4]]]]
+   *  }</pre>
+   *  (2) For the following input of shape `[4, 1, 1, 3]`, `block_shape = [2, 2]`, and
+   *      `crops = [[0, 0], [0, 0]]`:
+   *  <pre>{@code
+   *  [[[[1, 2, 3]]], [[[4, 5, 6]]], [[[7, 8, 9]]], [[[10, 11, 12]]]]
+   *  }</pre>
+   *  The output tensor has shape `[1, 2, 2, 3]` and value:
+   *  <pre>{@code
+   *  x = [[[[1, 2, 3], [4, 5, 6]],
+   *        [[7, 8, 9], [10, 11, 12]]]]
+   *  }</pre>
+   *  (3) For the following input of shape `[4, 2, 2, 1]`, `block_shape = [2, 2]`, and
+   *      `crops = [[0, 0], [0, 0]]`:
+   *  <pre>{@code
+   *  x = [[[[1], [3]], [[9], [11]]],
+   *       [[[2], [4]], [[10], [12]]],
+   *       [[[5], [7]], [[13], [15]]],
+   *       [[[6], [8]], [[14], [16]]]]
+   *  }</pre>
+   *  The output tensor has shape `[1, 4, 4, 1]` and value:
+   *  <pre>{@code
+   *  x = [[[[1],   [2],  [3],  [4]],
+   *       [[5],   [6],  [7],  [8]],
+   *       [[9],  [10], [11],  [12]],
+   *       [[13], [14], [15],  [16]]]]
+   *  }</pre>
+   *  (4) For the following input of shape `[8, 1, 3, 1]`, `block_shape = [2, 2]`, and
+   *      `crops = [[0, 0], [2, 0]]`:
+   *  <pre>{@code
+   *  x = [[[[0], [1], [3]]], [[[0], [9], [11]]],
+   *       [[[0], [2], [4]]], [[[0], [10], [12]]],
+   *       [[[0], [5], [7]]], [[[0], [13], [15]]],
+   *       [[[0], [6], [8]]], [[[0], [14], [16]]]]
+   *  }</pre>
+   *  The output tensor has shape `[2, 2, 4, 1]` and value:
+   *  <pre>{@code
+   *  x = [[[[1],   [2],  [3],  [4]],
+   *        [[5],   [6],  [7],  [8]]],
+   *       [[[9],  [10], [11],  [12]],
+   *        [[13], [14], [15],  [16]]]]
+   *  }</pre>
+   * @return a new instance of BatchToSpaceNd
+   */
+  public <T extends TType, U extends TNumber, V extends TNumber> BatchToSpaceNd<T> batchToSpaceNd(
+      Operand<T> input, Operand<U> blockShape, Operand<V> crops) {
+    return BatchToSpaceNd.create(scope, input, blockShape, crops);
+  }
+
+  /**
+   * Bitcasts a tensor from one type to another without copying data.
+   *  <p>
+   *  Given a tensor `input`, this operation returns a tensor that has the same buffer
+   *  data as `input` with datatype `type`.
+   *  <p>
+   *  If the input datatype `T` is larger than the output datatype `type` then the
+   *  shape changes from [...] to [..., sizeof(`T`)/sizeof(`type`)].
+   *  <p>
+   *  If `T` is smaller than `type`, the operator requires that the rightmost
+   *  dimension be equal to sizeof(`type`)/sizeof(`T`). The shape then goes from
+   *  [..., sizeof(`type`)/sizeof(`T`)] to [...].
+   *  <p>
+   *  tf.bitcast() and tf.cast() work differently when real dtype is casted as a complex dtype
+   *  (e.g. tf.complex64 or tf.complex128) as tf.cast() make imaginary part 0 while tf.bitcast()
+   *  gives module error.
+   *  For example,
+   *  <p>
+   *  Example 1:
+   *  <p>
+   *  >>> a = [1., 2., 3.]
+   *  >>> equality_bitcast = tf.bitcast(a, tf.complex128)
+   *  Traceback (most recent call last):
+   *  ...
+   *  InvalidArgumentError: Cannot bitcast from 1 to 18 [Op:Bitcast]
+   *  >>> equality_cast = tf.cast(a, tf.complex128)
+   *  >>> print(equality_cast)
+   *  tf.Tensor([1.+0.j 2.+0.j 3.+0.j], shape=(3,), dtype=complex128)
+   *  <p>
+   *  Example 2:
+   *  <p>
+   *  >>> tf.bitcast(tf.constant(0xffffffff, dtype=tf.uint32), tf.uint8)
+   *  <tf.Tensor: shape=(4,), dtype=uint8, numpy=array([255, 255, 255, 255], dtype=uint8)>
+   *  <p>
+   *  Example 3:
+   *  <p>
+   *  >>> x = [1., 2., 3.]
+   *  >>> y = [0., 2., 3.]
+   *  >>> equality= tf.equal(x,y)
+   *  >>> equality_cast = tf.cast(equality,tf.float32)
+   *  >>> equality_bitcast = tf.bitcast(equality_cast,tf.uint8)
+   *  >>> print(equality)
+   *  tf.Tensor([False True True], shape=(3,), dtype=bool)
+   *  >>> print(equality_cast)
+   *  tf.Tensor([0. 1. 1.], shape=(3,), dtype=float32)
+   *  >>> print(equality_bitcast)
+   *  tf.Tensor(
+   *      [[  0   0   0   0]
+   *       [  0   0 128  63]
+   *       [  0   0 128  63]], shape=(3, 4), dtype=uint8)
+   *  <p>
+   *  <i>NOTE</i>: Bitcast is implemented as a low-level cast, so machines with different
+   *  endian orderings will give different results.
+   *
+   * @param <U> data type for {@code output()} output
+   * @param input
+   * @param type
+   * @return a new instance of Bitcast
+   */
+  public <U extends TType, T extends TType> Bitcast<U> bitcast(Operand<T> input, Class<U> type) {
+    return Bitcast.create(scope, input, type);
+  }
+
+  /**
+   * Return the shape of s0 op s1 with broadcast.
+   *  <p>
+   *  Given `s0` and `s1`, tensors that represent shapes, compute `r0`, the
+   *  broadcasted shape. `s0`, `s1` and `r0` are all integer vectors.
+   *
+   * @param <T> data type for {@code r0()} output
+   * @param s0
+   * @param s1
+   * @return a new instance of BroadcastDynamicShape
+   */
+  public <T extends TNumber> BroadcastDynamicShape<T> broadcastDynamicShape(Operand<T> s0,
+      Operand<T> s1) {
+    return BroadcastDynamicShape.create(scope, s0, s1);
+  }
+
+  /**
+   * Broadcast an array for a compatible shape.
+   *  <p>
+   *  Broadcasting is the process of making arrays to have compatible shapes
+   *  for arithmetic operations. Two shapes are compatible if for each
+   *  dimension pair they are either equal or one of them is one. When trying
+   *  to broadcast a Tensor to a shape, it starts with the trailing dimensions,
+   *  and works its way forward.
+   *  <p>
+   *  For example,
+   *  <p>
+   *  >>> x = tf.constant([1, 2, 3])
+   *  >>> y = tf.broadcast_to(x, [3, 3])
+   *  >>> print(y)
+   *  tf.Tensor(
+   *      [[1 2 3]
+   *       [1 2 3]
+   *       [1 2 3]], shape=(3, 3), dtype=int32)
+   *  <p>
+   *  In the above example, the input Tensor with the shape of `[1, 3]`
+   *  is broadcasted to output Tensor with shape of `[3, 3]`.
+   *  <p>
+   *  When doing broadcasted operations such as multiplying a tensor
+   *  by a scalar, broadcasting (usually) confers some time or space
+   *  benefit, as the broadcasted tensor is never materialized.
+   *  <p>
+   *  However, `broadcast_to` does not carry with it any such benefits.
+   *  The newly-created tensor takes the full memory of the broadcasted
+   *  shape. (In a graph context, `broadcast_to` might be fused to
+   *  subsequent operation and then be optimized away, however.)
+   *
+   * @param <T> data type for {@code output()} output
+   * @param input A Tensor to broadcast.
+   * @param shape An 1-D `int` Tensor. The shape of the desired output.
+   * @return a new instance of BroadcastTo
+   */
+  public <T extends TType, U extends TNumber> BroadcastTo<T> broadcastTo(Operand<T> input,
+      Operand<U> shape) {
+    return BroadcastTo.create(scope, input, shape);
+  }
+
+  /**
+   * Bucketizes 'input' based on 'boundaries'.
+   *  <p>
+   *  For example, if the inputs are
+   *      boundaries = [0, 10, 100]
+   *      input = [[-5, 10000]
+   *               [150,   10]
+   *               [5,    100]]
+   *  <p>
+   *  then the output will be
+   *      output = [[0, 3]
+   *                [3, 2]
+   *                [1, 3]]
+   *
+   * @param input Any shape of Tensor contains with int or float type.
+   * @param boundaries A sorted list of floats gives the boundary of the buckets.
+   * @return a new instance of Bucketize
+   */
+  public <T extends TNumber> Bucketize bucketize(Operand<T> input, List<Float> boundaries) {
+    return Bucketize.create(scope, input, boundaries);
+  }
+
+  /**
+   * Clips tensor values to a specified min and max.
+   *  <p>
+   *  Given a tensor `t`, this operation returns a tensor of the same type and
+   *  shape as `t` with its values clipped to `clip_value_min` and `clip_value_max`.
+   *  Any values less than `clip_value_min` are set to `clip_value_min`. Any values
+   *  greater than `clip_value_max` are set to `clip_value_max`.
+   *
+   * @param <T> data type for {@code output()} output
+   * @param t A `Tensor`.
+   * @param clipValueMin A 0-D (scalar) `Tensor`, or a `Tensor` with the same shape
+   *  as `t`. The minimum value to clip by.
+   * @param clipValueMax A 0-D (scalar) `Tensor`, or a `Tensor` with the same shape
+   *  as `t`. The maximum value to clip by.
+   * @return a new instance of ClipByValue
+   */
+  public <T extends TType> ClipByValue<T> clipByValue(Operand<T> t, Operand<T> clipValueMin,
+      Operand<T> clipValueMax) {
+    return ClipByValue.create(scope, t, clipValueMin, clipValueMax);
+  }
+
+  /**
+   * Concatenates tensors along one dimension.
+   *
+   * @param <T> data type for {@code output()} output
+   * @param values List of `N` Tensors to concatenate. Their ranks and types must match,
+   *  and their sizes must match in all dimensions except `concat_dim`.
+   * @param axis 0-D.  The dimension along which to concatenate.  Must be in the
+   *  range [-rank(values), rank(values)).
+   * @return a new instance of Concat
+   */
+  public <T extends TType, U extends TNumber> Concat<T> concat(Iterable<Operand<T>> values,
+      Operand<U> axis) {
+    return Concat.create(scope, values, axis);
+  }
+
+  /**
+   * Creates a constant of {@code long} elements that is a copy of a given n-dimensional array.
+   *
+   * @param scope is a scope used to add the underlying operation.
+   * @param data an n-dimensional array of {@code long} elements.
+   * @return a long constant
+   */
+  public Constant<TInt64> constant(LongNdArray data) {
+    return Constant.tensorOf(scope, data);
+  }
+
+  /**
+   * Creates a rank-1 constant of {@code int} elements.
+   *
+   * @param scope is a scope used to add the underlying operation.
+   * @param data An array containing the values to put into the new constant. The dimensions of the
+   *      new constant will match those of the array.
+   * @return an integer constant
+   */
+  public Constant<TInt32> constant(int[] data) {
+    return Constant.vectorOf(scope, data);
+  }
+
+  /**
+   * Creates a rank-3 constant of {@code int} elements.
+   *
+   * @param scope is a scope used to add the underlying operation.
+   * @param data An array containing the values to put into the new constant. The dimensions of the
+   *      new constant will match those of the array.
+   * @return an integer constant
+   */
+  public Constant<TInt32> constant(int[][][] data) {
+    return Constant.tensorOf(scope, data);
+  }
+
+  /**
+   * Creates a constant containing a single {@code double} element.
+   *
+   * @param scope is a scope used to add the underlying operation.
+   * @param data The value to put into the new constant.
+   * @return a double constant
+   */
+  public Constant<TFloat64> constant(double data) {
+    return Constant.scalarOf(scope, data);
+  }
+
+  /**
+   * Creates a rank-5 constant of {@code long} elements.
+   *
+   * @param scope is a scope used to add the underlying operation.
+   * @param data An array containing the values to put into the new constant. The dimensions of the
+   *      new constant will match those of the array.
+   * @return a long constant
+   */
+  public Constant<TInt64> constant(long[][][][][] data) {
+    return Constant.tensorOf(scope, data);
+  }
+
+  /**
+   * Creates a rank-5 constant of {@code boolean} elements.
+   *
+   * @param scope is a scope used to add the underlying operation.
+   * @param data An array containing the values to put into the new constant. The dimensions of the
+   *      new constant will match those of the array.
+   * @return a boolean constant
+   */
+  public Constant<TBool> constant(boolean[][][][][] data) {
+    return Constant.tensorOf(scope, data);
+  }
+
+  /**
+   * Creates a constant of {@code int} elements that is a copy of a given n-dimensional array.
+   *
+   * @param scope is a scope used to add the underlying operation.
+   * @param data an n-dimensional array of {@code int} elements.
+   * @return an integer constant
+   */
+  public Constant<TInt32> constant(IntNdArray data) {
+    return Constant.tensorOf(scope, data);
+  }
+
+  /**
+   * Creates a constant of {@code double} elements that is a copy of a given n-dimensional array.
+   *
+   * @param scope is a scope used to add the underlying operation.
+   * @param data an n-dimensional array of {@code double} elements.
+   * @return a double constant
+   */
+  public Constant<TFloat64> constant(DoubleNdArray data) {
+    return Constant.tensorOf(scope, data);
+  }
+
+  /**
+   * Creates a rank-4 constant of {@code int} elements.
+   *
+   * @param scope is a scope used to add the underlying operation.
+   * @param data An array containing the values to put into the new constant. The dimensions of the
+   *      new constant will match those of the array.
+   * @return an integer constant
+   */
+  public Constant<TInt32> constant(int[][][][] data) {
+    return Constant.tensorOf(scope, data);
+  }
+
+  /**
+   * Creates a rank-6 constant of {@code float} elements.
+   *
+   * @param scope is a scope used to add the underlying operation.
+   * @param data An array containing the values to put into the new constant. The dimensions of the
+   *      new constant will match those of the array.
+   * @return a float constant
+   */
+  public Constant<TFloat32> constant(float[][][][][][] data) {
+    return Constant.tensorOf(scope, data);
+  }
+
+  /**
+   * Creates a constant containing a single {@code byte} element.
+   *
+   * @param scope is a scope used to add the underlying operation.
+   * @param data The value to put into the new constant.
+   * @return a byte constant
+   */
+  public Constant<TUint8> constant(byte data) {
+    return Constant.scalarOf(scope, data);
+  }
+
+  /**
+   * Creates a rank-3 constant of {@code boolean} elements.
+   *
+   * @param scope is a scope used to add the underlying operation.
+   * @param data An array containing the values to put into the new constant. The dimensions of the
+   *      new constant will match those of the array.
+   * @return a boolean constant
+   */
+  public Constant<TBool> constant(boolean[][][] data) {
+    return Constant.tensorOf(scope, data);
+  }
+
+  /**
+   * Creates a rank-4 constant of {@code float} elements.
+   *
+   * @param scope is a scope used to add the underlying operation.
+   * @param data An array containing the values to put into the new constant. The dimensions of the
+   *      new constant will match those of the array.
+   * @return a float constant
+   */
+  public Constant<TFloat32> constant(float[][][][] data) {
+    return Constant.tensorOf(scope, data);
+  }
+
+  /**
+   * Creates a rank-2 constant of {@code long} elements.
+   *
+   * @param scope is a scope used to add the underlying operation.
+   * @param data An array containing the values to put into the new constant. The dimensions of the
+   *      new constant will match those of the array.
+   * @return a long constant
+   */
+  public Constant<TInt64> constant(long[][] data) {
+    return Constant.tensorOf(scope, data);
+  }
+
+  /**
+   * Creates a rank-5 constant of {@code byte} elements.
+   *
+   * @param scope is a scope used to add the underlying operation.
+   * @param data An array containing the values to put into the new constant. The dimensions of the
+   *      new constant will match those of the array.
+   * @return a byte constant
+   */
+  public Constant<TUint8> constant(byte[][][][][] data) {
+    return Constant.tensorOf(scope, data);
+  }
+
+  /**
+   * Creates a constant of {@code boolean} elements that is a copy of a given n-dimensional array.
+   *
+   * @param scope is a scope used to add the underlying operation.
+   * @param data an n-dimensional array of {@code boolean} elements.
+   * @return a boolean constant
+   */
+  public Constant<TBool> constant(BooleanNdArray data) {
+    return Constant.tensorOf(scope, data);
+  }
+
+  /**
+   * Creates a rank-2 constant of {@code float} elements.
+   *
+   * @param scope is a scope used to add the underlying operation.
+   * @param data An array containing the values to put into the new constant. The dimensions of the
+   *      new constant will match those of the array.
+   * @return a float constant
+   */
+  public Constant<TFloat32> constant(float[][] data) {
+    return Constant.tensorOf(scope, data);
+  }
+
+  /**
+   * Creates a constant of {@code byte} elements that is a copy of a given n-dimensional array.
+   *
+   * @param scope is a scope used to add the underlying operation.
+   * @param data an n-dimensional array of {@code byte} elements.
+   * @return a byte constant
+   */
+  public Constant<TUint8> constant(ByteNdArray data) {
+    return Constant.tensorOf(scope, data);
+  }
+
+  /**
+   * Creates a rank-2 constant of {@code byte} elements.
+   *
+   * @param scope is a scope used to add the underlying operation.
+   * @param data An array containing the values to put into the new constant. The dimensions of the
+   *      new constant will match those of the array.
+   * @return a byte constant
+   */
+  public Constant<TUint8> constant(byte[][] data) {
+    return Constant.tensorOf(scope, data);
+  }
+
+  /**
+   * Creates a rank-5 constant of {@code double} elements.
+   *
+   * @param scope is a scope used to add the underlying operation.
+   * @param data An array containing the values to put into the new constant. The dimensions of the
+   *      new constant will match those of the array.
+   * @return a double constant
+   */
+  public Constant<TFloat64> constant(double[][][][][] data) {
+    return Constant.tensorOf(scope, data);
+  }
+
+  /**
+   * Creates a rank-3 constant of {@code float} elements.
+   *
+   * @param scope is a scope used to add the underlying operation.
+   * @param data An array containing the values to put into the new constant. The dimensions of the
+   *      new constant will match those of the array.
+   * @return a float constant
+   */
+  public Constant<TFloat32> constant(float[][][] data) {
+    return Constant.tensorOf(scope, data);
+  }
+
+  /**
+   * Creates a rank-1 constant of {@code byte} elements.
+   *
+   * @param scope is a scope used to add the underlying operation.
+   * @param data An array containing the values to put into the new constant. The dimensions of the
+   *      new constant will match those of the array.
+   * @return a byte constant
+   */
+  public Constant<TUint8> constant(byte[] data) {
+    return Constant.vectorOf(scope, data);
+  }
+
+  /**
+   * Creates a rank-1 constant of {@code float} elements.
+   *
+   * @param scope is a scope used to add the underlying operation.
+   * @param data An array containing the values to put into the new constant. The dimensions of the
+   *      new constant will match those of the array.
+   * @return a float constant
+   */
+  public Constant<TFloat32> constant(float[] data) {
+    return Constant.vectorOf(scope, data);
+  }
+
+  /**
+   * Creates a rank-2 constant of {@code boolean} elements.
+   *
+   * @param scope is a scope used to add the underlying operation.
+   * @param data An array containing the values to put into the new constant. The dimensions of the
+   *      new constant will match those of the array.
+   * @return a boolean constant
+   */
+  public Constant<TBool> constant(boolean[][] data) {
+    return Constant.tensorOf(scope, data);
+  }
+
+  /**
+   * Creates a constant of {@code String} elements that is a copy of a given n-dimensional array,
+   *  using the default UTF-8 encoding.
+   *
+   * @param scope is a scope used to add the underlying operation.
+   * @param data an n-dimensional array of {@code String} elements.
+   * @return a string constant
+   */
+  public Constant<TString> constant(NdArray<String> data) {
+    return Constant.tensorOf(scope, data);
+  }
+
+  /**
+   * Creates a {@code String} constant using the default, UTF-8 encoding.
+   *
+   * @param scope is a scope used to add the underlying operation.
+   * @param data The string to put into the new constant.
+   * @return a string constant
+   */
+  public Constant<TString> constant(String data) {
+    return Constant.scalarOf(scope, data);
+  }
+
+  /**
+   * Creates a rank-4 constant of {@code double} elements.
+   *
+   * @param scope is a scope used to add the underlying operation.
+   * @param data An array containing the values to put into the new constant. The dimensions of the
+   *      new constant will match those of the array.
+   * @return a double constant
+   */
+  public Constant<TFloat64> constant(double[][][][] data) {
+    return Constant.tensorOf(scope, data);
+  }
+
+  /**
+   * Creates a rank-2 constant of {@code double} elements.
+   *
+   * @param scope is a scope used to add the underlying operation.
+   * @param data An array containing the values to put into the new constant. The dimensions of the
+   *      new constant will match those of the array.
+   * @return a double constant
+   */
+  public Constant<TFloat64> constant(double[][] data) {
+    return Constant.tensorOf(scope, data);
+  }
+
+  /**
+   * Creates a constant containing a single {@code int} element.
+   *
+   * @param scope is a scope used to add the underlying operation.
+   * @param data The value to put into the new constant.
+   * @return an integer constant
+   */
+  public Constant<TInt32> constant(int data) {
+    return Constant.scalarOf(scope, data);
+  }
+
+  /**
+   * Creates a rank-4 constant of {@code byte} elements.
+   *
+   * @param scope is a scope used to add the underlying operation.
+   * @param data An array containing the values to put into the new constant. The dimensions of the
+   *      new constant will match those of the array.
+   * @return a byte constant
+   */
+  public Constant<TUint8> constant(byte[][][][] data) {
+    return Constant.tensorOf(scope, data);
+  }
+
+  /**
+   * Creates a rank-6 constant of {@code int} elements.
+   *
+   * @param scope is a scope used to add the underlying operation.
+   * @param data An array containing the values to put into the new constant. The dimensions of the
+   *      new constant will match those of the array.
+   * @return an integer constant
+   */
+  public Constant<TInt32> constant(int[][][][][][] data) {
+    return Constant.tensorOf(scope, data);
+  }
+
+  /**
+   * Creates a constant containing a single {@code long} element.
+   *
+   * @param scope is a scope used to add the underlying operation.
+   * @param data The value to put into the new constant.
+   * @return a long constant
+   */
+  public Constant<TInt64> constant(long data) {
+    return Constant.scalarOf(scope, data);
+  }
+
+  /**
+   * Creates a constant containing a single {@code float} element.
+   *
+   * @param scope is a scope used to add the underlying operation.
+   * @param data The value to put into the new constant.
+   * @return a float constant
+   */
+  public Constant<TFloat32> constant(float data) {
+    return Constant.scalarOf(scope, data);
+  }
+
+  /**
+   * Creates a rank-5 constant of {@code float} elements.
+   *
+   * @param scope is a scope used to add the underlying operation.
+   * @param data An array containing the values to put into the new constant. The dimensions of the
+   *      new constant will match those of the array.
+   * @return a float constant
+   */
+  public Constant<TFloat32> constant(float[][][][][] data) {
+    return Constant.tensorOf(scope, data);
+  }
+
+  /**
+   * Creates a rank-3 constant of {@code double} elements.
+   *
+   * @param scope is a scope used to add the underlying operation.
+   * @param data An array containing the values to put into the new constant. The dimensions of the
+   *      new constant will match those of the array.
+   * @return a double constant
+   */
+  public Constant<TFloat64> constant(double[][][] data) {
+    return Constant.tensorOf(scope, data);
+  }
+
+  /**
+   * Creates a rank-6 constant of {@code long} elements.
+   *
+   * @param scope is a scope used to add the underlying operation.
+   * @param data An array containing the values to put into the new constant. The dimensions of the
+   *      new constant will match those of the array.
+   * @return a long constant
+   */
+  public Constant<TInt64> constant(long[][][][][][] data) {
+    return Constant.tensorOf(scope, data);
+  }
+
+  /**
+   * Creates a rank-4 constant of {@code long} elements.
+   *
+   * @param scope is a scope used to add the underlying operation.
+   * @param data An array containing the values to put into the new constant. The dimensions of the
+   *      new constant will match those of the array.
+   * @return a long constant
+   */
+  public Constant<TInt64> constant(long[][][][] data) {
+    return Constant.tensorOf(scope, data);
+  }
+
+  /**
+   * Creates a rank-1 constant of {@code long} elements.
+   *
+   * @param scope is a scope used to add the underlying operation.
+   * @param data An array containing the values to put into the new constant. The dimensions of the
+   *      new constant will match those of the array.
+   * @return a long constant
+   */
+  public Constant<TInt64> constant(long[] data) {
+    return Constant.vectorOf(scope, data);
+  }
+
+  /**
+   * Creates a rank-1 constant of {@code boolean} elements.
+   *
+   * @param scope is a scope used to add the underlying operation.
+   * @param data An array containing the values to put into the new constant. The dimensions of the
+   *      new constant will match those of the array.
+   * @return a boolean constant
+   */
+  public Constant<TBool> constant(boolean[] data) {
+    return Constant.vectorOf(scope, data);
+  }
+
+  /**
+   * Creates a rank-3 constant of {@code byte} elements.
+   *
+   * @param scope is a scope used to add the underlying operation.
+   * @param data An array containing the values to put into the new constant. The dimensions of the
+   *      new constant will match those of the array.
+   * @return a byte constant
+   */
+  public Constant<TUint8> constant(byte[][][] data) {
+    return Constant.tensorOf(scope, data);
+  }
+
+  /**
+   * Creates a rank-6 constant of {@code byte} elements.
+   *
+   * @param scope is a scope used to add the underlying operation.
+   * @param data An array containing the values to put into the new constant. The dimensions of the
+   *      new constant will match those of the array.
+   * @return a byte constant
+   */
+  public Constant<TUint8> constant(byte[][][][][][] data) {
+    return Constant.tensorOf(scope, data);
+  }
+
+  /**
+   * Creates a rank-2 constant of {@code int} elements.
+   *
+   * @param scope is a scope used to add the underlying operation.
+   * @param data An array containing the values to put into the new constant. The dimensions of the
+   *      new constant will match those of the array.
+   * @return an integer constant
+   */
+  public Constant<TInt32> constant(int[][] data) {
+    return Constant.tensorOf(scope, data);
+  }
+
+  /**
+   * Creates a constant of {@code float} elements that is a copy of a given n-dimensional array.
+   *
+   * @param scope is a scope used to add the underlying operation.
+   * @param data an n-dimensional array of {@code float} elements.
+   * @return a float constant
+   */
+  public Constant<TFloat32> constant(FloatNdArray data) {
+    return Constant.tensorOf(scope, data);
+  }
+
+  /**
+   * Creates a rank-5 constant of {@code int} elements.
+   *
+   * @param scope is a scope used to add the underlying operation.
+   * @param data An array containing the values to put into the new constant. The dimensions of the
+   *      new constant will match those of the array.
+   * @return an integer constant
+   */
+  public Constant<TInt32> constant(int[][][][][] data) {
+    return Constant.tensorOf(scope, data);
+  }
+
+  /**
+   * Creates a rank-1 constant of {@code double} elements.
+   *
+   * @param scope is a scope used to add the underlying operation.
+   * @param data An array containing the values to put into the new constant. The dimensions of the
+   *      new constant will match those of the array.
+   * @return a double constant
+   */
+  public Constant<TFloat64> constant(double[] data) {
+    return Constant.vectorOf(scope, data);
+  }
+
+  /**
+   * Creates a rank-6 constant of {@code boolean} elements.
+   *
+   * @param scope is a scope used to add the underlying operation.
+   * @param data An array containing the values to put into the new constant. The dimensions of the
+   *      new constant will match those of the array.
+   * @return a boolean constant
+   */
+  public Constant<TBool> constant(boolean[][][][][][] data) {
+    return Constant.tensorOf(scope, data);
+  }
+
+  /**
+   * Creates a rank-6 constant of {@code double} elements.
+   *
+   * @param scope is a scope used to add the underlying operation.
+   * @param data An array containing the values to put into the new constant. The dimensions of the
+   *      new constant will match those of the array.
+   * @return a double constant
+   */
+  public Constant<TFloat64> constant(double[][][][][][] data) {
+    return Constant.tensorOf(scope, data);
+  }
+
+  /**
+   * Creates a constant containing a single {@code boolean} element.
+   *
+   * @param scope is a scope used to add the underlying operation.
+   * @param data The value to put into the new constant.
+   * @return a boolean constant
+   */
+  public Constant<TBool> constant(boolean data) {
+    return Constant.scalarOf(scope, data);
+  }
+
+  /**
+   * Creates a rank-4 constant of {@code boolean} elements.
+   *
+   * @param scope is a scope used to add the underlying operation.
+   * @param data An array containing the values to put into the new constant. The dimensions of the
+   *      new constant will match those of the array.
+   * @return a boolean constant
+   */
+  public Constant<TBool> constant(boolean[][][][] data) {
+    return Constant.tensorOf(scope, data);
+  }
+
+  /**
+   * Creates a rank-3 constant of {@code long} elements.
+   *
+   * @param scope is a scope used to add the underlying operation.
+   * @param data An array containing the values to put into the new constant. The dimensions of the
+   *      new constant will match those of the array.
+   * @return a long constant
+   */
+  public Constant<TInt64> constant(long[][][] data) {
+    return Constant.tensorOf(scope, data);
+  }
+
+  /**
+   * Creates a rank-1 constant of {@code long} elements representing the size of each dimensions of
+   *  the given shape.
+   *
+   * @param scope is a scope used to add the underlying operation.
+   * @param shape a shape
+   * @return a long constant
+   */
+  public Constant<TInt64> constant(Shape shape) {
+    return Constant.tensorOf(scope, shape);
+  }
+
+  /**
+   * Create a constant from a Tensor.
+   *
+   * @param scope is a scope used to add the underlying operation.
+   * @param tensor a Tensor holding the constant value
+   * @return a constant of the same data type as `tensor`
+   */
+  public <T extends TType> Constant<T> constant(T tensor) {
+    return Constant.create(scope, tensor);
+  }
+
+  /**
+   * Creates a constant of {@code String} elements, using the given charset.
+   *
+   * @param scope is a scope used to add the underlying operation.
+   * @param charset charset for encoding/decoding strings bytes.
+   * @param data An array containing the values to put into the new constant. String elements are
+   *      sequences of bytes from the last array dimension.
+   * @return the {@code String} constant
+   */
+  public Constant<TString> constant(Charset charset, String[] data) {
+    return Constant.vectorOf(scope, charset, data);
+  }
+
+  /**
+   * Creates a {@code String} constant using a specified encoding.
+   *
+   * @param scope is a scope used to add the underlying operation.
+   * @param charset The encoding from String to bytes.
+   * @param data The string to put into the new constant.
+   * @return a string constant
+   */
+  public Constant<TString> constant(Charset charset, String data) {
+    return Constant.scalarOf(scope, charset, data);
+  }
+
+  /**
+   * Creates a constant of {@code String} elements that is a copy of a given n-dimensional array,
+   *  using the given encoding.
+   *
+   * @param scope is a scope used to add the underlying operation.
+   * @param charset charset used to encode/decode string bytes.
+   * @param data an n-dimensional array of {@code String} elements.
+   * @return a string constant
+   */
+  public Constant<TString> constant(Charset charset, NdArray<String> data) {
+    return Constant.tensorOf(scope, charset, data);
+  }
+
+  /**
+   * Create a {@link TFloat32} constant with data from the given buffer.
+   *
+   * @param scope is a scope used to add the underlying operation.
+   * @param shape the tensor shape.
+   * @param data a buffer containing the tensor data.
+   * @return a float constant
+   * @throws IllegalArgumentException If the tensor shape is not compatible with the buffer
+   */
+  public Constant<TFloat32> constant(Shape shape, FloatDataBuffer data) {
+    return Constant.tensorOf(scope, shape, data);
+  }
+
+  /**
+   * Create a {@link TBool} constant with data from the given buffer.
+   *
+   * @param scope is a scope used to add the underlying operation.
+   * @param shape the tensor shape.
+   * @param data a buffer containing the tensor data.
+   * @return an boolean constant
+   * @throws IllegalArgumentException If the tensor shape is not compatible with the buffer
+   */
+  public Constant<TBool> constant(Shape shape, BooleanDataBuffer data) {
+    return Constant.tensorOf(scope, shape, data);
+  }
+
+  /**
+   * Create a {@link TUint8} constant with data from the given buffer.
+   *
+   * @param scope is a scope used to add the underlying operation.
+   * @param shape the tensor shape.
+   * @param data a buffer containing the tensor data.
+   * @return a byte constant
+   * @throws IllegalArgumentException If the tensor shape is not compatible with the buffer
+   */
+  public Constant<TUint8> constant(Shape shape, ByteDataBuffer data) {
+    return Constant.tensorOf(scope, shape, data);
+  }
+
+  /**
+   * Create a {@link TInt64} constant with data from the given buffer.
+   *
+   * @param scope is a scope used to add the underlying operation.
+   * @param shape the tensor shape.
+   * @param data a buffer containing the tensor data.
+   * @return a long constant
+   * @throws IllegalArgumentException If the tensor shape is not compatible with the buffer
+   */
+  public Constant<TInt64> constant(Shape shape, LongDataBuffer data) {
+    return Constant.tensorOf(scope, shape, data);
+  }
+
+  /**
+   * Create a {@link TString} constant with data from the given buffer, using the default UTF-8
+   *  encoding.
+   *
+   * @param scope is a scope used to add the underlying operation.
+   * @param shape the tensor shape.
+   * @param data a buffer containing the tensor data.
+   * @return a string constant
+   * @throws IllegalArgumentException If the tensor shape is not compatible with the buffer
+   */
+  public Constant<TString> constant(Shape shape, DataBuffer<String> data) {
+    return Constant.tensorOf(scope, shape, data);
+  }
+
+  /**
+   * Create a {@link TFloat64} constant with data from the given buffer.
+   *
+   * @param scope is a scope used to add the underlying operation.
+   * @param shape the tensor shape.
+   * @param data a buffer containing the tensor data.
+   * @return a double constant
+   * @throws IllegalArgumentException If the tensor shape is not compatible with the buffer
+   */
+  public Constant<TFloat64> constant(Shape shape, DoubleDataBuffer data) {
+    return Constant.tensorOf(scope, shape, data);
+  }
+
+  /**
+   * Create a {@link TInt32} constant with data from the given buffer.
+   *
+   * @param scope is a scope used to add the underlying operation.
+   * @param shape the tensor shape.
+   * @param data a buffer containing the tensor data.
+   * @return an integer constant
+   * @throws IllegalArgumentException If the tensor shape is not compatible with the buffer
+   */
+  public Constant<TInt32> constant(Shape shape, IntDataBuffer data) {
+    return Constant.tensorOf(scope, shape, data);
+  }
+
+  /**
+   * Create a {@link TString} constant with data from the given buffer, using the given encoding.
+   *
+   * @param scope is a scope used to add the underlying operation.
+   * @param charset charset used to encode/decode string bytes.
+   * @param shape the tensor shape.
+   * @param data a buffer containing the tensor data.
+   * @return a string constant
+   * @throws IllegalArgumentException If the tensor shape is not compatible with the buffer
+   */
+  public Constant<TString> constant(Charset charset, Shape shape, DataBuffer<String> data) {
+    return Constant.tensorOf(scope, charset, shape, data);
+  }
+
+  /**
+   * Create a constant with data from the given buffer.
+   *
+   * @param scope is a scope used to add the underlying operation.
+   * @param type the tensor type.
+   * @param shape the tensor shape.
+   * @param data a buffer containing the tensor data.
+   * @return a constant of type `type`
+   * @throws IllegalArgumentException If the tensor datatype or shape is not compatible with the
+   *      buffer
+   */
+  public <T extends TType> Constant<T> constant(Class<T> type, Shape shape, ByteDataBuffer data) {
+    return Constant.tensorOf(scope, type, shape, data);
+  }
+
+  /**
+   * This op consumes a lock created by `MutexLock`.
+   *  <p>
+   *  This op exists to consume a tensor created by `MutexLock` (other than
+   *  direct control dependencies).  It should be the only that consumes the tensor,
+   *  and will raise an error if it is not.  Its only purpose is to keep the
+   *  mutex lock tensor alive until it is consumed by this op.
+   *  <p>
+   *  <b>NOTE</b>: This operation must run on the same device as its input.  This may
+   *  be enforced via the `colocate_with` mechanism.
+   *
+   * @param mutexLock A tensor returned by `MutexLock`.
+   * @return a new instance of ConsumeMutexLock
+   */
+  public ConsumeMutexLock consumeMutexLock(Operand<?> mutexLock) {
+    return ConsumeMutexLock.create(scope, mutexLock);
+  }
+
+  /**
+   * Does nothing. Serves as a control trigger for scheduling.
+   *  <p>
+   *  Only useful as a placeholder for control edges.
+   *
+   * @return a new instance of ControlTrigger
+   */
+  public ControlTrigger controlTrigger() {
+    return ControlTrigger.create(scope);
+  }
+
+  /**
+   * Increments 'ref' until it reaches 'limit'.
+   *
+   * @param <T> data type for {@code output()} output
+   * @param ref Should be from a scalar `Variable` node.
+   * @param limit If incrementing ref would bring it above limit, instead generates an
+   *  'OutOfRange' error.
+   * @return a new instance of CountUpTo
+   */
+  public <T extends TNumber> CountUpTo<T> countUpTo(Operand<T> ref, Long limit) {
+    return CountUpTo.create(scope, ref, limit);
+  }
+
+  /**
+   * Makes a copy of `x`.
+   *
+   * @param <T> data type for {@code y()} output
+   * @param x The source tensor of type `T`.
+   * @return a new instance of DeepCopy
+   */
+  public <T extends TType> DeepCopy<T> deepCopy(Operand<T> x) {
+    return DeepCopy.create(scope, x);
+  }
+
+  /**
+   * Delete the tensor specified by its handle in the session.
+   *
+   * @param handle The handle for a tensor stored in the session state.
+   * @return a new instance of DeleteSessionTensor
+   */
+  public DeleteSessionTensor deleteSessionTensor(Operand<TString> handle) {
+    return DeleteSessionTensor.create(scope, handle);
+  }
+
+  /**
+   * Deletes the resource specified by the handle.
+   *  <p>
+   *  All subsequent operations using the resource will result in a NotFound
+   *  error status.
+   *
+   * @param resource handle to the resource to delete.
+   * @param options carries optional attributes values
+   * @return a new instance of DestroyResourceOp
+   */
+  public DestroyResourceOp destroyResourceOp(Operand<?> resource,
+      DestroyResourceOp.Options... options) {
+    return DestroyResourceOp.create(scope, resource, options);
+  }
+
+  /**
+   * Destroys the temporary variable and returns its final value.
+   *  <p>
+   *  Sets output to the value of the Tensor pointed to by 'ref', then destroys
+   *  the temporary variable called 'var_name'.
+   *  All other uses of 'ref' <i>must</i> have executed before this op.
+   *  This is typically achieved by chaining the ref through each assign op, or by
+   *  using control dependencies.
+   *  <p>
+   *  Outputs the final value of the tensor pointed to by 'ref'.
+   *
+   * @param <T> data type for {@code value()} output
+   * @param ref A reference to the temporary variable tensor.
+   * @param varName Name of the temporary variable, usually the name of the matching
+   *  'TemporaryVariable' op.
+   * @return a new instance of DestroyTemporaryVariable
+   */
+  public <T extends TType> DestroyTemporaryVariable<T> destroyTemporaryVariable(Operand<T> ref,
+      String varName) {
+    return DestroyTemporaryVariable.create(scope, ref, varName);
+  }
+
+  /**
+   * Partitions `data` into `num_partitions` tensors using indices from `partitions`.
+   *  <p>
+   *  For each index tuple `js` of size `partitions.ndim`, the slice `data[js, ...]`
+   *  becomes part of `outputs[partitions[js]]`.  The slices with `partitions[js] = i`
+   *  are placed in `outputs[i]` in lexicographic order of `js`, and the first
+   *  dimension of `outputs[i]` is the number of entries in `partitions` equal to `i`.
+   *  In detail,
+   *  <pre>{@code
+   *      outputs[i].shape = [sum(partitions == i)] + data.shape[partitions.ndim:]
+   *
+   *      outputs[i] = pack([data[js, ...] for js if partitions[js] == i])
+   *  }</pre>
+   *  `data.shape` must start with `partitions.shape`.
+   *  <p>
+   *  For example:
+   *  <pre>{@code
+   *      # Scalar partitions.
+   *      partitions = 1
+   *      num_partitions = 2
+   *      data = [10, 20]
+   *      outputs[0] = []  # Empty with shape [0, 2]
+   *      outputs[1] = [[10, 20]]
+   *
+   *      # Vector partitions.
+   *      partitions = [0, 0, 1, 1, 0]
+   *      num_partitions = 2
+   *      data = [10, 20, 30, 40, 50]
+   *      outputs[0] = [10, 20, 50]
+   *      outputs[1] = [30, 40]
+   *  }</pre>
+   *  See `dynamic_stitch` for an example on how to merge partitions back.
+   *  <p>
+   *  <div style="width:70%; margin:auto; margin-bottom:10px; margin-top:20px;">
+   *  <img style="width:100%" src="https://www.tensorflow.org/images/DynamicPartition.png" alt>
+   *  </div>
+   *
+   * @param <T> data type for {@code outputs()} output
+   * @param data
+   * @param partitions Any shape.  Indices in the range `[0, num_partitions)`.
+   * @param numPartitions The number of partitions to output.
+   * @return a new instance of DynamicPartition
+   */
+  public <T extends TType> DynamicPartition<T> dynamicPartition(Operand<T> data,
+      Operand<TInt32> partitions, Long numPartitions) {
+    return DynamicPartition.create(scope, data, partitions, numPartitions);
+  }
+
+  /**
+   * Interleave the values from the `data` tensors into a single tensor.
+   *  <p>
+   *  Builds a merged tensor such that
+   *  <pre>{@code
+   *      merged[indices[m][i, ..., j], ...] = data[m][i, ..., j, ...]
+   *  }</pre>
+   *  For example, if each `indices[m]` is scalar or vector, we have
+   *  <pre>{@code
+   *      # Scalar indices:
+   *      merged[indices[m], ...] = data[m][...]
+   *
+   *      # Vector indices:
+   *      merged[indices[m][i], ...] = data[m][i, ...]
+   *  }</pre>
+   *  Each `data[i].shape` must start with the corresponding `indices[i].shape`,
+   *  and the rest of `data[i].shape` must be constant w.r.t. `i`.  That is, we
+   *  must have `data[i].shape = indices[i].shape + constant`.  In terms of this
+   *  `constant`, the output shape is
+   *  <p>
+   *      merged.shape = [max(indices)] + constant
+   *  <p>
+   *  Values are merged in order, so if an index appears in both `indices[m][i]` and
+   *  `indices[n][j]` for `(m,i) < (n,j)` the slice `data[n][j]` will appear in the
+   *  merged result. If you do not need this guarantee, ParallelDynamicStitch might
+   *  perform better on some devices.
+   *  <p>
+   *  For example:
+   *  <pre>{@code
+   *      indices[0] = 6
+   *      indices[1] = [4, 1]
+   *      indices[2] = [[5, 2], [0, 3]]
+   *      data[0] = [61, 62]
+   *      data[1] = [[41, 42], [11, 12]]
+   *      data[2] = [[[51, 52], [21, 22]], [[1, 2], [31, 32]]]
+   *      merged = [[1, 2], [11, 12], [21, 22], [31, 32], [41, 42],
+   *                [51, 52], [61, 62]]
+   *  }</pre>
+   *  This method can be used to merge partitions created by `dynamic_partition`
+   *  as illustrated on the following example:
+   *  <pre>{@code
+   *      # Apply function (increments x_i) on elements for which a certain condition
+   *      # apply (x_i != -1 in this example).
+   *      x=tf.constant([0.1, -1., 5.2, 4.3, -1., 7.4])
+   *      condition_mask=tf.not_equal(x,tf.constant(-1.))
+   *      partitioned_data = tf.dynamic_partition(
+   *          x, tf.cast(condition_mask, tf.int32) , 2)
+   *      partitioned_data[1] = partitioned_data[1] + 1.0
+   *      condition_indices = tf.dynamic_partition(
+   *          tf.range(tf.shape(x)[0]), tf.cast(condition_mask, tf.int32) , 2)
+   *      x = tf.dynamic_stitch(condition_indices, partitioned_data)
+   *      # Here x=[1.1, -1., 6.2, 5.3, -1, 8.4], the -1. values remain
+   *      # unchanged.
+   *  }</pre>
+   *  <div style="width:70%; margin:auto; margin-bottom:10px; margin-top:20px;">
+   *  <img style="width:100%" src="https://www.tensorflow.org/images/DynamicStitch.png" alt>
+   *  </div>
+   *
+   * @param <T> data type for {@code merged()} output
+   * @param indices
+   * @param data
+   * @return a new instance of DynamicStitch
+   */
+  public <T extends TType> DynamicStitch<T> dynamicStitch(Iterable<Operand<TInt32>> indices,
+      Iterable<Operand<T>> data) {
+    return DynamicStitch.create(scope, indices, data);
+  }
+
+  /**
+   * Computes the (possibly normalized) Levenshtein Edit Distance.
+   *  <p>
+   *  The inputs are variable-length sequences provided by SparseTensors
+   *    (hypothesis_indices, hypothesis_values, hypothesis_shape)
+   *  and
+   *    (truth_indices, truth_values, truth_shape).
+   *  <p>
+   *  The inputs are:
+   *
+   * @param hypothesisIndices The indices of the hypothesis list SparseTensor.
+   *  This is an N x R int64 matrix.
+   * @param hypothesisValues The values of the hypothesis list SparseTensor.
+   *  This is an N-length vector.
+   * @param hypothesisShape The shape of the hypothesis list SparseTensor.
+   *  This is an R-length vector.
+   * @param truthIndices The indices of the truth list SparseTensor.
+   *  This is an M x R int64 matrix.
+   * @param truthValues The values of the truth list SparseTensor.
+   *  This is an M-length vector.
+   * @param truthShape truth indices, vector.
+   * @param options carries optional attributes values
+   * @return a new instance of EditDistance
+   */
+  public <T extends TType> EditDistance editDistance(Operand<TInt64> hypothesisIndices,
+      Operand<T> hypothesisValues, Operand<TInt64> hypothesisShape, Operand<TInt64> truthIndices,
+      Operand<T> truthValues, Operand<TInt64> truthShape, EditDistance.Options... options) {
+    return EditDistance.create(scope, hypothesisIndices, hypothesisValues, hypothesisShape, truthIndices, truthValues, truthShape, options);
+  }
+
+  /**
+   * Creates a tensor with the given shape.
+   *  <p>
+   *  This operation creates a tensor of `shape` and `dtype`.
+   *
+   * @param <T> data type for {@code output()} output
+   * @param shape 1-D. Represents the shape of the output tensor.
+   * @param dtype
+   * @param options carries optional attributes values
+   * @return a new instance of Empty
+   */
+  public <T extends TType> Empty<T> empty(Operand<TInt32> shape, Class<T> dtype,
+      Empty.Options... options) {
+    return Empty.create(scope, shape, dtype, options);
+  }
+
+  /**
+   * Creates and returns an empty tensor list.
+   *  <p>
+   *  All list elements must be tensors of dtype element_dtype and shape compatible
+   *  with element_shape.
+   *  <p>
+   *  handle: an empty tensor list.
+   *  element_dtype: the type of elements in the list.
+   *  element_shape: a shape compatible with that of elements in the list.
+   *
+   * @param elementShape
+   * @param maxNumElements
+   * @param elementDtype
+   * @return a new instance of EmptyTensorList
+   */
+  public <T extends TNumber, U extends TType> EmptyTensorList emptyTensorList(
+      Operand<T> elementShape, Operand<TInt32> maxNumElements, Class<U> elementDtype) {
+    return EmptyTensorList.create(scope, elementShape, maxNumElements, elementDtype);
+  }
+
+  /**
+   * Ensures that the tensor's shape matches the expected shape.
+   *  <p>
+   *  Raises an error if the input tensor's shape does not match the specified shape.
+   *  Returns the input tensor otherwise.
+   *
+   * @param <T> data type for {@code output()} output
+   * @param input A tensor, whose shape is to be validated.
+   * @param shape The expected (possibly partially specified) shape of the input tensor.
+   * @return a new instance of EnsureShape
+   */
+  public <T extends TType> EnsureShape<T> ensureShape(Operand<T> input, Shape shape) {
+    return EnsureShape.create(scope, input, shape);
+  }
+
+  /**
+   * Inserts a dimension of 1 into a tensor's shape.
+   *  <p>
+   *  Given a tensor `input`, this operation inserts a dimension of 1 at the
+   *  dimension index `axis` of `input`'s shape. The dimension index `axis` starts at
+   *  zero; if you specify a negative number for `axis` it is counted backward from
+   *  the end.
+   *  <p>
+   *  This operation is useful if you want to add a batch dimension to a single
+   *  element. For example, if you have a single image of shape `[height, width,
+   *  channels]`, you can make it a batch of 1 image with `expand_dims(image, 0)`,
+   *  which will make the shape `[1, height, width, channels]`.
+   *  <p>
+   *  Other examples:
+   *  <pre>{@code
+   *  # 't' is a tensor of shape [2]
+   *  shape(expand_dims(t, 0)) ==> [1, 2]
+   *  shape(expand_dims(t, 1)) ==> [2, 1]
+   *  shape(expand_dims(t, -1)) ==> [2, 1]
+   *
+   *  # 't2' is a tensor of shape [2, 3, 5]
+   *  shape(expand_dims(t2, 0)) ==> [1, 2, 3, 5]
+   *  shape(expand_dims(t2, 2)) ==> [2, 3, 1, 5]
+   *  shape(expand_dims(t2, 3)) ==> [2, 3, 5, 1]
+   *  }</pre>
+   *  This operation requires that:
+   *  <p>
+   *  `-1-input.dims() <= dim <= input.dims()`
+   *  <p>
+   *  This operation is related to `squeeze()`, which removes dimensions of
+   *  size 1.
+   *
+   * @param <T> data type for {@code output()} output
+   * @param input
+   * @param axis 0-D (scalar). Specifies the dimension index at which to
+   *  expand the shape of `input`. Must be in the range
+   *  `[-rank(input) - 1, rank(input)]`.
+   * @return a new instance of ExpandDims
+   */
+  public <T extends TType, U extends TNumber> ExpandDims<T> expandDims(Operand<T> input,
+      Operand<U> axis) {
+    return ExpandDims.create(scope, input, axis);
+  }
+
+  /**
+   * Extract `patches` from `input` and put them in the "depth" output dimension. 3D extension of `extract_image_patches`.
+   *
+   * @param <T> data type for {@code patches()} output
+   * @param input 5-D Tensor with shape `[batch, in_planes, in_rows, in_cols, depth]`.
+   * @param ksizes The size of the sliding window for each dimension of `input`.
+   * @param strides 1-D of length 5. How far the centers of two consecutive patches are in
+   *  `input`. Must be: `[1, stride_planes, stride_rows, stride_cols, 1]`.
+   * @param padding The type of padding algorithm to use.
+   *  <p>
+   *  We specify the size-related attributes as:
+   *  <pre>{@code
+   *        ksizes = [1, ksize_planes, ksize_rows, ksize_cols, 1]
+   *        strides = [1, stride_planes, strides_rows, strides_cols, 1]
+   *  }</pre>
+   * @return a new instance of ExtractVolumePatches
+   */
+  public <T extends TNumber> ExtractVolumePatches<T> extractVolumePatches(Operand<T> input,
+      List<Long> ksizes, List<Long> strides, String padding) {
+    return ExtractVolumePatches.create(scope, input, ksizes, strides, padding);
+  }
+
+  /**
+   * Creates a tensor filled with a scalar value.
+   *  <p>
+   *  This operation creates a tensor of shape `dims` and fills it with `value`.
+   *  <p>
+   *  For example:
+   *  <pre>{@code
+   *  # Output tensor has shape [2, 3].
+   *  fill([2, 3], 9) ==> [[9, 9, 9]
+   *                       [9, 9, 9]]
+   *  }</pre>
+   *  `tf.fill` differs from `tf.constant` in a few ways:
+   *  <ul>
+   *  <li>
+   *  `tf.fill` only supports scalar contents, whereas `tf.constant` supports
+   *      Tensor values.
+   *  </li>
+   *  <li>
+   *  `tf.fill` creates an Op in the computation graph that constructs the actual
+   *      Tensor value at runtime. This is in contrast to `tf.constant` which embeds
+   *      the entire Tensor into the graph with a `Const` node.
+   *  </li>
+   *  <li>
+   *  Because `tf.fill` evaluates at graph runtime, it supports dynamic shapes
+   *      based on other runtime Tensors, unlike `tf.constant`.
+   *
+   * @param <U> data type for {@code output()} output
+   * @param dims 1-D. Represents the shape of the output tensor.
+   * @param value 0-D (scalar). Value to fill the returned tensor.
+   *  <p>
+   * @compatibility(numpy) Equivalent to np.full
+   * @end_compatibility
+   * @return a new instance of Fill
+   */
+  public <U extends TType, T extends TNumber> Fill<U> fill(Operand<T> dims, Operand<U> value) {
+    return Fill.create(scope, dims, value);
+  }
+
+  /**
+   * Generates fingerprint values.
+   *  <p>
+   *  Generates fingerprint values of `data`.
+   *  <p>
+   *  Fingerprint op considers the first dimension of `data` as the batch dimension,
+   *  and `output[i]` contains the fingerprint value generated from contents in
+   *  `data[i, ...]` for all `i`.
+   *  <p>
+   *  Fingerprint op writes fingerprint values as byte arrays. For example, the
+   *  default method `farmhash64` generates a 64-bit fingerprint value at a time.
+   *  This 8-byte value is written out as an `uint8` array of size 8, in little-endian
+   *  order.
+   *  <p>
+   *  For example, suppose that `data` has data type `DT_INT32` and shape (2, 3, 4),
+   *  and that the fingerprint method is `farmhash64`. In this case, the output shape
+   *  is (2, 8), where 2 is the batch dimension size of `data`, and 8 is the size of
+   *  each fingerprint value in bytes. `output[0, :]` is generated from 12 integers in
+   *  `data[0, :, :]` and similarly `output[1, :]` is generated from other 12 integers
+   *  in `data[1, :, :]`.
+   *  <p>
+   *  Note that this op fingerprints the raw underlying buffer, and it does not
+   *  fingerprint Tensor's metadata such as data type and/or shape. For example, the
+   *  fingerprint values are invariant under reshapes and bitcasts as long as the
+   *  batch dimension remain the same:
+   *  <pre>{@code
+   *  Fingerprint(data) == Fingerprint(Reshape(data, ...))
+   *  Fingerprint(data) == Fingerprint(Bitcast(data, ...))
+   *  }</pre>
+   *  For string data, one should expect `Fingerprint(data) !=
+   *  Fingerprint(ReduceJoin(data))` in general.
+   *
+   * @param data Must have rank 1 or higher.
+   * @param method Fingerprint method used by this op. Currently available method is
+   *  `farmhash::fingerprint64`.
+   * @return a new instance of Fingerprint
+   */
+  public <T extends TType> Fingerprint fingerprint(Operand<T> data, Operand<TString> method) {
+    return Fingerprint.create(scope, data, method);
+  }
+
+  /**
+   * Gather slices from `params` axis `axis` according to `indices`.
+   *  <p>
+   *  `indices` must be an integer tensor of any dimension (usually 0-D or 1-D).
+   *  Produces an output tensor with shape `params.shape[:axis] +
+   *  indices.shape[batch_dims:] + params.shape[axis + 1:]` where:
+   *  <pre>{@code
+   *      # Scalar indices (output is rank(params) - 1).
+   *      output[a_0, ..., a_n, b_0, ..., b_n] =
+   *        params[a_0, ..., a_n, indices, b_0, ..., b_n]
+   *
+   *      # Vector indices (output is rank(params)).
+   *      output[a_0, ..., a_n, i, b_0, ..., b_n] =
+   *        params[a_0, ..., a_n, indices[i], b_0, ..., b_n]
+   *
+   *      # Higher rank indices (output is rank(params) + rank(indices) - 1).
+   *      output[a_0, ..., a_n, i, ..., j, b_0, ... b_n] =
+   *        params[a_0, ..., a_n, indices[i, ..., j], b_0, ..., b_n]
+   *  }</pre>
+   *  <div style="width:70%; margin:auto; margin-bottom:10px; margin-top:20px;">
+   *  <img style="width:100%" src="https://www.tensorflow.org/images/Gather.png" alt>
+   *  </div>
+   *  <p>
+   *  Note that on CPU, if an out of bound index is found, an error is returned.
+   *  On GPU, if an out of bound index is found, a 0 is stored in the
+   *  corresponding output value.
+   *  <p>
+   *  See also `tf.batch_gather` and `tf.gather_nd`.
+   *
+   * @param <T> data type for {@code output()} output
+   * @param params The tensor from which to gather values. Must be at least rank
+   *  `axis + 1`.
+   * @param indices Index tensor. Must be in range `[0, params.shape[axis])`.
+   * @param axis The axis in `params` to gather `indices` from. Defaults to the first
+   *  dimension. Supports negative indexes.
+   * @param options carries optional attributes values
+   * @return a new instance of Gather
+   */
+  public <T extends TType, U extends TNumber, V extends TNumber> Gather<T> gather(Operand<T> params,
+      Operand<U> indices, Operand<V> axis, Gather.Options... options) {
+    return Gather.create(scope, params, indices, axis, options);
+  }
+
+  /**
+   * Gather slices from `params` into a Tensor with shape specified by `indices`.
+   *  <p>
+   *  `indices` is a K-dimensional integer tensor, best thought of as a
+   *  (K-1)-dimensional tensor of indices into `params`, where each element defines a
+   *  slice of `params`:
+   *  <p>
+   *      output[\\(i_0, ..., i_{K-2}\\)] = params[indices[\\(i_0, ..., i_{K-2}\\)]]
+   *  <p>
+   *  Whereas in `tf.gather` `indices` defines slices into the `axis`
+   *  dimension of `params`, in `tf.gather_nd`, `indices` defines slices into the
+   *  first `N` dimensions of `params`, where `N = indices.shape[-1]`.
+   *  <p>
+   *  The last dimension of `indices` can be at most the rank of
+   *  `params`:
+   *  <p>
+   *      indices.shape[-1] <= params.rank
+   *  <p>
+   *  The last dimension of `indices` corresponds to elements
+   *  (if `indices.shape[-1] == params.rank`) or slices
+   *  (if `indices.shape[-1] < params.rank`) along dimension `indices.shape[-1]`
+   *  of `params`.  The output tensor has shape
+   *  <p>
+   *      indices.shape[:-1] + params.shape[indices.shape[-1]:]
+   *  <p>
+   *  Note that on CPU, if an out of bound index is found, an error is returned.
+   *  On GPU, if an out of bound index is found, a 0 is stored in the
+   *  corresponding output value.
+   *  <p>
+   *  Some examples below.
+   *  <p>
+   *  Simple indexing into a matrix:
+   *  <pre>{@code
+   *      indices = [[0, 0], [1, 1]]
+   *      params = [['a', 'b'], ['c', 'd']]
+   *      output = ['a', 'd']
+   *  }</pre>
+   *  Slice indexing into a matrix:
+   *  <pre>{@code
+   *      indices = [[1], [0]]
+   *      params = [['a', 'b'], ['c', 'd']]
+   *      output = [['c', 'd'], ['a', 'b']]
+   *  }</pre>
+   *  Indexing into a 3-tensor:
+   *  <pre>{@code
+   *      indices = [[1]]
+   *      params = [[['a0', 'b0'], ['c0', 'd0']],
+   *                [['a1', 'b1'], ['c1', 'd1']]]
+   *      output = [[['a1', 'b1'], ['c1', 'd1']]]
+   *
+   *
+   *      indices = [[0, 1], [1, 0]]
+   *      params = [[['a0', 'b0'], ['c0', 'd0']],
+   *                [['a1', 'b1'], ['c1', 'd1']]]
+   *      output = [['c0', 'd0'], ['a1', 'b1']]
+   *
+   *
+   *      indices = [[0, 0, 1], [1, 0, 1]]
+   *      params = [[['a0', 'b0'], ['c0', 'd0']],
+   *                [['a1', 'b1'], ['c1', 'd1']]]
+   *      output = ['b0', 'b1']
+   *  }</pre>
+   *  Batched indexing into a matrix:
+   *  <pre>{@code
+   *      indices = [[[0, 0]], [[0, 1]]]
+   *      params = [['a', 'b'], ['c', 'd']]
+   *      output = [['a'], ['b']]
+   *  }</pre>
+   *  Batched slice indexing into a matrix:
+   *  <pre>{@code
+   *      indices = [[[1]], [[0]]]
+   *      params = [['a', 'b'], ['c', 'd']]
+   *      output = [[['c', 'd']], [['a', 'b']]]
+   *  }</pre>
+   *  Batched indexing into a 3-tensor:
+   *  <pre>{@code
+   *      indices = [[[1]], [[0]]]
+   *      params = [[['a0', 'b0'], ['c0', 'd0']],
+   *                [['a1', 'b1'], ['c1', 'd1']]]
+   *      output = [[[['a1', 'b1'], ['c1', 'd1']]],
+   *                [[['a0', 'b0'], ['c0', 'd0']]]]
+   *
+   *      indices = [[[0, 1], [1, 0]], [[0, 0], [1, 1]]]
+   *      params = [[['a0', 'b0'], ['c0', 'd0']],
+   *                [['a1', 'b1'], ['c1', 'd1']]]
+   *      output = [[['c0', 'd0'], ['a1', 'b1']],
+   *                [['a0', 'b0'], ['c1', 'd1']]]
+   *
+   *
+   *      indices = [[[0, 0, 1], [1, 0, 1]], [[0, 1, 1], [1, 1, 0]]]
+   *      params = [[['a0', 'b0'], ['c0', 'd0']],
+   *                [['a1', 'b1'], ['c1', 'd1']]]
+   *      output = [['b0', 'b1'], ['d0', 'c1']]
+   *  }</pre>
+   *  See also `tf.gather` and `tf.batch_gather`.
+   *
+   * @param <T> data type for {@code output()} output
+   * @param params The tensor from which to gather values.
+   * @param indices Index tensor.
+   * @return a new instance of GatherNd
+   */
+  public <T extends TType, U extends TNumber> GatherNd<T> gatherNd(Operand<T> params,
+      Operand<U> indices) {
+    return GatherNd.create(scope, params, indices);
+  }
+
+  /**
+   * Store the input tensor in the state of the current session.
+   *
+   * @param value The tensor to be stored.
+   * @return a new instance of GetSessionHandle
+   */
+  public <T extends TType> GetSessionHandle getSessionHandle(Operand<T> value) {
+    return GetSessionHandle.create(scope, value);
+  }
+
+  /**
+   * Get the value of the tensor specified by its handle.
+   *
+   * @param <T> data type for {@code value()} output
+   * @param handle The handle for a tensor stored in the session state.
+   * @param dtype The type of the output value.
+   * @return a new instance of GetSessionTensor
+   */
+  public <T extends TType> GetSessionTensor<T> getSessionTensor(Operand<TString> handle,
+      Class<T> dtype) {
+    return GetSessionTensor.create(scope, handle, dtype);
+  }
+
+  /**
+   * Adds gradients computation ops to the graph according to scope.
+   *
+   * @param scope current graph scope
+   * @param y outputs of the function to derive
+   * @param x inputs of the function for which partial derivatives are computed
+   * @param options carries optional attributes values
+   * @return a new instance of {@code Gradients}
+   * @throws IllegalArgumentException if execution environment is not a graph
+   */
+  public Gradients gradients(Iterable<? extends Operand<?>> y, Iterable<? extends Operand<?>> x,
+      Gradients.Options... options) {
+    return Gradients.create(scope, y, x, options);
+  }
+
+  /**
+   * Adds operations to compute the partial derivatives of sum of {@code y}s w.r.t {@code x}s,
+   *  i.e., {@code d(y_1 + y_2 + ...)/dx_1, d(y_1 + y_2 + ...)/dx_2...}
+   *  <p> 
+   *  If {@code Options.dx()} values are set, they are as the initial symbolic partial derivatives of some loss 
+   *  function {@code L} w.r.t. {@code y}. {@code Options.dx()} must have the size of {@code y}.
+   *  <p>
+   *  If {@code Options.dx()} is not set, the implementation will use dx of {@code OnesLike} for all
+   *  shapes in {@code y}.
+   *  <p>
+   *  The partial derivatives are returned in output {@code dy}, with the size of {@code x}.
+   *  <p>
+   *  Example of usage:
+   *  <pre>{@code
+   *  Gradients gradients = tf.gradients(loss, Arrays.asList(w, b));
+   *  Constant<TFloat32> alpha = tf.constant(1.0f);
+   *  tf.train.applyGradientDescent(w, alpha, gradients.<Float>dy(0));
+   *  tf.train.applyGradientDescent(b, alpha, gradients.<Float>dy(1));
+   *  }</pre>
+   *
+   * @param y output of the function to derive
+   * @param x inputs of the function for which partial derivatives are computed
+   * @param options carries optional attributes values
+   * @return a new instance of {@code Gradients}
+   * @throws IllegalArgumentException if execution environment is not a graph
+   */
+  public Gradients gradients(Operand<?> y, Iterable<? extends Operand<?>> x,
+      Gradients.Options... options) {
+    return Gradients.create(scope, y, x, options);
+  }
+
+  /**
+   * Gives a guarantee to the TF runtime that the input tensor is a constant.
+   *  <p>
+   *  The runtime is then free to make optimizations based on this.
+   *  <p>
+   *  Only accepts value typed tensors as inputs and rejects resource variable handles
+   *  as input.
+   *  <p>
+   *  Returns the input tensor without modification.
+   *
+   * @param <T> data type for {@code output()} output
+   * @param input
+   * @return a new instance of GuaranteeConst
+   */
+  public <T extends TType> GuaranteeConst<T> guaranteeConst(Operand<T> input) {
+    return GuaranteeConst.create(scope, input);
+  }
+
+  /**
+   * Creates a non-initialized hash table.
+   *  <p>
+   *  This op creates a hash table, specifying the type of its keys and values.
+   *  Before using the table you will have to initialize it.  After initialization the
+   *  table will be immutable.
+   *
+   * @param keyDtype Type of the table keys.
+   * @param valueDtype Type of the table values.
+   * @param options carries optional attributes values
+   * @return a new instance of HashTable
+   */
+  public <T extends TType, U extends TType> HashTable hashTable(Class<T> keyDtype,
+      Class<U> valueDtype, HashTable.Options... options) {
+    return HashTable.create(scope, keyDtype, valueDtype, options);
+  }
+
+  /**
+   * Return histogram of values.
+   *  <p>
+   *  Given the tensor `values`, this operation returns a rank 1 histogram counting
+   *  the number of entries in `values` that fall into every bin.  The bins are
+   *  equal width and determined by the arguments `value_range` and `nbins`.
+   *  <pre>{@code
+   *  # Bins will be:  (-inf, 1), [1, 2), [2, 3), [3, 4), [4, inf)
+   *  nbins = 5
+   *  value_range = [0.0, 5.0]
+   *  new_values = [-1.0, 0.0, 1.5, 2.0, 5.0, 15]
+   *
+   *  with tf.get_default_session() as sess:
+   *    hist = tf.histogram_fixed_width(new_values, value_range, nbins=5)
+   *    variables.global_variables_initializer().run()
+   *    sess.run(hist) => [2, 1, 1, 0, 2]
+   *  }</pre>
+   *
+   * @param <U> data type for {@code out()} output
+   * @param values Numeric `Tensor`.
+   * @param valueRange Shape [2] `Tensor` of same `dtype` as `values`.
+   *  values <= value_range[0] will be mapped to hist[0],
+   *  values >= value_range[1] will be mapped to hist[-1].
+   * @param nbins Scalar `int32 Tensor`.  Number of histogram bins.
+   * @return a new instance of HistogramFixedWidth
+   */
+  public <T extends TNumber> HistogramFixedWidth<TInt32> histogramFixedWidth(Operand<T> values,
+      Operand<T> valueRange, Operand<TInt32> nbins) {
+    return HistogramFixedWidth.create(scope, values, valueRange, nbins);
+  }
+
+  /**
+   * Return histogram of values.
+   *  <p>
+   *  Given the tensor `values`, this operation returns a rank 1 histogram counting
+   *  the number of entries in `values` that fall into every bin.  The bins are
+   *  equal width and determined by the arguments `value_range` and `nbins`.
+   *  <pre>{@code
+   *  # Bins will be:  (-inf, 1), [1, 2), [2, 3), [3, 4), [4, inf)
+   *  nbins = 5
+   *  value_range = [0.0, 5.0]
+   *  new_values = [-1.0, 0.0, 1.5, 2.0, 5.0, 15]
+   *
+   *  with tf.get_default_session() as sess:
+   *    hist = tf.histogram_fixed_width(new_values, value_range, nbins=5)
+   *    variables.global_variables_initializer().run()
+   *    sess.run(hist) => [2, 1, 1, 0, 2]
+   *  }</pre>
+   *
+   * @param <U> data type for {@code out()} output
+   * @param values Numeric `Tensor`.
+   * @param valueRange Shape [2] `Tensor` of same `dtype` as `values`.
+   *  values <= value_range[0] will be mapped to hist[0],
+   *  values >= value_range[1] will be mapped to hist[-1].
+   * @param nbins Scalar `int32 Tensor`.  Number of histogram bins.
+   * @param dtype
+   * @return a new instance of HistogramFixedWidth
+   */
+  public <U extends TNumber, T extends TNumber> HistogramFixedWidth<U> histogramFixedWidth(
+      Operand<T> values, Operand<T> valueRange, Operand<TInt32> nbins, Class<U> dtype) {
+    return HistogramFixedWidth.create(scope, values, valueRange, nbins, dtype);
+  }
+
+  /**
+   * Return a tensor with the same shape and contents as the input tensor or value.
+   *
+   * @param <T> data type for {@code output()} output
+   * @param input
+   * @return a new instance of Identity
+   */
+  public <T extends TType> Identity<T> identity(Operand<T> input) {
+    return Identity.create(scope, input);
+  }
+
+  /**
+   * Returns a list of tensors with the same shapes and contents as the input
+   *  <p>
+   *  tensors.
+   *  <p>
+   *  This op can be used to override the gradient for complicated functions. For
+   *  example, suppose y = f(x) and we wish to apply a custom function g for backprop
+   *  such that dx = g(dy). In Python,
+   *  <pre>{@code
+   *  with tf.get_default_graph().gradient_override_map(
+   *      {'IdentityN': 'OverrideGradientWithG'}):
+   *    y, _ = identity_n([f(x), x])
+   *
+   * @tf.RegisterGradient('OverrideGradientWithG') def ApplyG(op, dy, _):
+   *    return [None, g(dy)]  # Do not backprop to f(x).
+   *  }</pre>
+   * @param input
+   * @return a new instance of IdentityN
+   */
+  public IdentityN identityN(Iterable<Operand<?>> input) {
+    return IdentityN.create(scope, input);
+  }
+
+  /**
+   * Returns immutable tensor from memory region.
+   *  <p>
+   *  The current implementation memmaps the tensor from a file.
+   *
+   * @param <T> data type for {@code tensor()} output
+   * @param dtype Type of the returned tensor.
+   * @param shape Shape of the returned tensor.
+   * @param memoryRegionName Name of readonly memory region used by the tensor, see
+   *  NewReadOnlyMemoryRegionFromFile in tensorflow::Env.
+   * @return a new instance of ImmutableConst
+   */
+  public <T extends TType> ImmutableConst<T> immutableConst(Class<T> dtype, Shape shape,
+      String memoryRegionName) {
+    return ImmutableConst.create(scope, dtype, shape, memoryRegionName);
+  }
+
+  /**
+   * Factory method to create an operation executing all initializers of a graph.
+   *
+   *  <p>All initializers added to a graph via
+   *  {@link org.tensorflow.op.core.Init#add(Scope, Op) tf.initAdd} are grouped together as a single
+   *  unit of computation in the graph. This operation must then be added to any graph using one or
+   *  more {@link Variable variables} and executed once before running the graph so the variable
+   *  states are initialized properly.</p>
+   *
+   *  <p>When the graph is built by the same process that is running the session, the initializers
+   *  can be invoked by executing this single endpoint. For example:</p>
+   *  <pre>{@code
+   *  try (Graph g = new Graph()) {
+   *    Variable<TInt32> x = tf.variable(tf.constant(10));  // initAdd is called implicitly
+   *    Variable<TInt32> y = tf.variable(tf.constant(20));  // idem
+   *    Add<TInt32> z = tf.math.add(x, y);
+   *
+   *    try (Session s = new Session(g)) {
+   *      s.run(tf.init());  // initialize all variables
+   *
+   *      try (Tensor<TInt32> t = s.runner().fetch(z).run().get(0).expect(TInt32.DTYPE)) {
+   *        assertEquals(30, t.data().getInt());
+   *      }
+   *    }
+   *  }
+   *  }</pre>
+   *
+   *  <p>When the graph is built by a separate process, the initializers can be invoked by running
+   *  the init op by its name, which defaults to {@link org.tensorflow.op.core.Init#DEFAULT_NAME}.
+   *  For example:</p>
+   *  <pre>{@code
+   *  // Building the model
+   *  try (Graph g = new Graph()) {
+   *    Variable<TInt32> x = tf.variable(tf.constant(10));  // initAdd is called implicitly
+   *    Variable<TInt32> y = tf.variable(tf.constant(20));  // idem
+   *    Add<TInt32> z = tf.withName("z").math.add(x, y);
+   *
+   *    tf.init();  // add variables initializers to the graph, as Init.DEFAULT_NAME
+   *    // ...exporting graph as a saved model...
+   *  }
+   *
+   *  ...
+   *
+   *  // Running the model
+   *  try (SavedModelBundle model = SavedModelBundle.load("/path/to/model", "train")) {
+   *    model.session().run(Init.DEFAULT_NAME);
+   *
+   *    try (Tensor<TInt32> t = s.runner().fetch("z").run().get(0).expect(TInt32.DTYPE)) {
+   *      assertEquals(30, t.data().getInt());
+   *    }
+   *  }
+   *  }</pre>
+   *
+   * @param scope current scope
+   * @return an op grouping all initializers added to the graph
+   * @throws IllegalArgumentException if the execution environment in scope is not a graph
+   */
+  public Init init() {
+    return Init.create(scope);
+  }
+
+  /**
+   * Register an op as an initializer of the graph.
+   *
+   *  <p>Registered initializers are then grouped as a single unit of computation by adding
+   *  and executing an {@link org.tensorflow.op.core.Init#create(Scope) init} operation from a graph
+   *  session.
+   *
+   * @param scope
+   * @param initializer
+   * @throws IllegalArgumentException if the execution environment in scope is not a graph
+   * @see org.tensorflow.op.core.Init#create(Scope) init
+   */
+  public void initAdd(Op initializer) {
+    Init.add(scope, initializer);
+  }
+
+  /**
+   * Table initializer that takes two tensors for keys and values respectively.
+   *
+   * @param tableHandle Handle to a table which will be initialized.
+   * @param keys Keys of type Tkey.
+   * @param values Values of type Tval.
+   * @return a new instance of InitializeTable
+   */
+  public <T extends TType, U extends TType> InitializeTable initializeTable(Operand<?> tableHandle,
+      Operand<T> keys, Operand<U> values) {
+    return InitializeTable.create(scope, tableHandle, keys, values);
+  }
+
+  /**
+   * Initializes a table from a text file.
+   *  <p>
+   *  It inserts one key-value pair into the table for each line of the file.
+   *  The key and value is extracted from the whole line content, elements from the
+   *  split line based on `delimiter` or the line number (starting from zero).
+   *  Where to extract the key and value from a line is specified by `key_index` and
+   *  `value_index`.
+   *  <p>
+   *  - A value of -1 means use the line number(starting from zero), expects `int64`.
+   *  - A value of -2 means use the whole line content, expects `string`.
+   *  - A value >= 0 means use the index (starting at zero) of the split line based
+   *    on `delimiter`.
+   *
+   * @param tableHandle Handle to a table which will be initialized.
+   * @param filename Filename of a vocabulary text file.
+   * @param keyIndex Column index in a line to get the table `key` values from.
+   * @param valueIndex Column index that represents information of a line to get the table
+   *  `value` values from.
+   * @param options carries optional attributes values
+   * @return a new instance of InitializeTableFromTextFile
+   */
+  public InitializeTableFromTextFile initializeTableFromTextFile(Operand<?> tableHandle,
+      Operand<TString> filename, Long keyIndex, Long valueIndex,
+      InitializeTableFromTextFile.Options... options) {
+    return InitializeTableFromTextFile.create(scope, tableHandle, filename, keyIndex, valueIndex, options);
+  }
+
+  /**
+   *     Adds v into specified rows of x.
+   *  <p>
+   *      Computes y = x; y[i, :] += v; return y.
+   *
+   * @param <T> data type for {@code y()} output
+   * @param x A `Tensor` of type T.
+   * @param i A vector. Indices into the left-most dimension of `x`.
+   * @param v A `Tensor` of type T. Same dimension sizes as x except the first dimension, which must be the same as i's size.
+   * @return a new instance of InplaceAdd
+   */
+  public <T extends TType> InplaceAdd<T> inplaceAdd(Operand<T> x, Operand<TInt32> i, Operand<T> v) {
+    return InplaceAdd.create(scope, x, i, v);
+  }
+
+  /**
+   *     Subtracts `v` into specified rows of `x`.
+   *  <p>
+   *      Computes y = x; y[i, :] -= v; return y.
+   *
+   * @param <T> data type for {@code y()} output
+   * @param x A `Tensor` of type T.
+   * @param i A vector. Indices into the left-most dimension of `x`.
+   * @param v A `Tensor` of type T. Same dimension sizes as x except the first dimension, which must be the same as i's size.
+   * @return a new instance of InplaceSub
+   */
+  public <T extends TType> InplaceSub<T> inplaceSub(Operand<T> x, Operand<TInt32> i, Operand<T> v) {
+    return InplaceSub.create(scope, x, i, v);
+  }
+
+  /**
+   * Updates specified rows 'i' with values 'v'.
+   *  <p>
+   *  Computes `x[i, :] = v; return x`.
+   *  <p>
+   *  Originally this function is mutative however for compilation we make this
+   *  operation create / operate on a copy of `x`.
+   *
+   * @param <T> data type for {@code y()} output
+   * @param x A tensor of type `T`.
+   * @param i A vector. Indices into the left-most dimension of `x`.
+   * @param v A `Tensor` of type T. Same dimension sizes as x except the first dimension, which must be the same as i's size.
+   * @return a new instance of InplaceUpdate
+   */
+  public <T extends TType> InplaceUpdate<T> inplaceUpdate(Operand<T> x, Operand<TInt32> i,
+      Operand<T> v) {
+    return InplaceUpdate.create(scope, x, i, v);
+  }
+
+  /**
+   * Checks whether a tensor has been initialized.
+   *  <p>
+   *  Outputs boolean scalar indicating whether the tensor has been initialized.
+   *
+   * @param ref Should be from a `Variable` node. May be uninitialized.
+   * @return a new instance of IsVariableInitialized
+   */
+  public <T extends TType> IsVariableInitialized isVariableInitialized(Operand<T> ref) {
+    return IsVariableInitialized.create(scope, ref);
+  }
+
+  /**
+   * Outputs all keys and values in the table.
+   *
+   * @param <T> data type for {@code keys()} output
+   * @param <U> data type for {@code values()} output
+   * @param tableHandle Handle to the table.
+   * @param Tkeys
+   * @param Tvalues
+   * @return a new instance of LookupTableExport
+   */
+  public <T extends TType, U extends TType> LookupTableExport<T, U> lookupTableExport(
+      Operand<?> tableHandle, Class<T> Tkeys, Class<U> Tvalues) {
+    return LookupTableExport.create(scope, tableHandle, Tkeys, Tvalues);
+  }
+
+  /**
+   * Looks up keys in a table, outputs the corresponding values.
+   *  <p>
+   *  The tensor `keys` must of the same type as the keys of the table.
+   *  The output `values` is of the type of the table values.
+   *  <p>
+   *  The scalar `default_value` is the value output for keys not present in the
+   *  table. It must also be of the same type as the table values.
+   *
+   * @param <U> data type for {@code values()} output
+   * @param tableHandle Handle to the table.
+   * @param keys Any shape.  Keys to look up.
+   * @param defaultValue
+   * @return a new instance of LookupTableFind
+   */
+  public <U extends TType, T extends TType> LookupTableFind<U> lookupTableFind(
+      Operand<?> tableHandle, Operand<T> keys, Operand<U> defaultValue) {
+    return LookupTableFind.create(scope, tableHandle, keys, defaultValue);
+  }
+
+  /**
+   * Replaces the contents of the table with the specified keys and values.
+   *  <p>
+   *  The tensor `keys` must be of the same type as the keys of the table.
+   *  The tensor `values` must be of the type of the table values.
+   *
+   * @param tableHandle Handle to the table.
+   * @param keys Any shape.  Keys to look up.
+   * @param values Values to associate with keys.
+   * @return a new instance of LookupTableImport
+   */
+  public <T extends TType, U extends TType> LookupTableImport lookupTableImport(
+      Operand<?> tableHandle, Operand<T> keys, Operand<U> values) {
+    return LookupTableImport.create(scope, tableHandle, keys, values);
+  }
+
+  /**
+   * Updates the table to associates keys with values.
+   *  <p>
+   *  The tensor `keys` must be of the same type as the keys of the table.
+   *  The tensor `values` must be of the type of the table values.
+   *
+   * @param tableHandle Handle to the table.
+   * @param keys Any shape.  Keys to look up.
+   * @param values Values to associate with keys.
+   * @return a new instance of LookupTableInsert
+   */
+  public <T extends TType, U extends TType> LookupTableInsert lookupTableInsert(
+      Operand<?> tableHandle, Operand<T> keys, Operand<U> values) {
+    return LookupTableInsert.create(scope, tableHandle, keys, values);
+  }
+
+  /**
+   * Computes the number of elements in the given table.
+   *
+   * @param tableHandle Handle to the table.
+   * @return a new instance of LookupTableSize
+   */
+  public LookupTableSize lookupTableSize(Operand<?> tableHandle) {
+    return LookupTableSize.create(scope, tableHandle);
+  }
+
+  /**
+   * Forwards the input to the output.
+   *  <p>
+   *  This operator represents the loop termination condition used by the
+   *  "pivot" switches of a loop.
+   *
+   * @param input A boolean scalar, representing the branch predicate of the Switch op.
+   * @return a new instance of LoopCond
+   */
+  public LoopCond loopCond(Operand<TBool> input) {
+    return LoopCond.create(scope, input);
+  }
+
+  /**
+   * Op removes all elements in the underlying container.
+   *
+   * @param dtypes
+   * @param options carries optional attributes values
+   * @return a new instance of MapClear
+   */
+  public MapClear mapClear(List<Class<? extends TType>> dtypes, MapClear.Options... options) {
+    return MapClear.create(scope, dtypes, options);
+  }
+
+  /**
+   * Op returns the number of incomplete elements in the underlying container.
+   *
+   * @param dtypes
+   * @param options carries optional attributes values
+   * @return a new instance of MapIncompleteSize
+   */
+  public MapIncompleteSize mapIncompleteSize(List<Class<? extends TType>> dtypes,
+      MapIncompleteSize.Options... options) {
+    return MapIncompleteSize.create(scope, dtypes, options);
+  }
+
+  /**
+   * Op peeks at the values at the specified key.  If the
+   *  <p>
+   *  underlying container does not contain this key
+   *  this op will block until it does.
+   *
+   * @param key
+   * @param indices
+   * @param dtypes
+   * @param options carries optional attributes values
+   * @return a new instance of MapPeek
+   */
+  public MapPeek mapPeek(Operand<TInt64> key, Operand<TInt32> indices,
+      List<Class<? extends TType>> dtypes, MapPeek.Options... options) {
+    return MapPeek.create(scope, key, indices, dtypes, options);
+  }
+
+  /**
+   * Op returns the number of elements in the underlying container.
+   *
+   * @param dtypes
+   * @param options carries optional attributes values
+   * @return a new instance of MapSize
+   */
+  public MapSize mapSize(List<Class<? extends TType>> dtypes, MapSize.Options... options) {
+    return MapSize.create(scope, dtypes, options);
+  }
+
+  /**
+   * Stage (key, values) in the underlying container which behaves like a hashtable.
+   *
+   * @param key int64
+   * @param indices
+   * @param values a list of tensors
+   *  dtypes A list of data types that inserted values should adhere to.
+   * @param dtypes
+   * @param options carries optional attributes values
+   * @return a new instance of MapStage
+   */
+  public MapStage mapStage(Operand<TInt64> key, Operand<TInt32> indices,
+      Iterable<Operand<?>> values, List<Class<? extends TType>> dtypes,
+      MapStage.Options... options) {
+    return MapStage.create(scope, key, indices, values, dtypes, options);
+  }
+
+  /**
+   * Op removes and returns the values associated with the key
+   *  <p>
+   *  from the underlying container.   If the underlying container
+   *  does not contain this key, the op will block until it does.
+   *
+   * @param key
+   * @param indices
+   * @param dtypes
+   * @param options carries optional attributes values
+   * @return a new instance of MapUnstage
+   */
+  public MapUnstage mapUnstage(Operand<TInt64> key, Operand<TInt32> indices,
+      List<Class<? extends TType>> dtypes, MapUnstage.Options... options) {
+    return MapUnstage.create(scope, key, indices, dtypes, options);
+  }
+
+  /**
+   * Op removes and returns a random (key, value)
+   *  <p>
+   *  from the underlying container.   If the underlying container
+   *  does not contain elements, the op will block until it does.
+   *
+   * @param indices
+   * @param dtypes
+   * @param options carries optional attributes values
+   * @return a new instance of MapUnstageNoKey
+   */
+  public MapUnstageNoKey mapUnstageNoKey(Operand<TInt32> indices,
+      List<Class<? extends TType>> dtypes, MapUnstageNoKey.Options... options) {
+    return MapUnstageNoKey.create(scope, indices, dtypes, options);
+  }
+
+  /**
+   * Computes the maximum of elements across dimensions of a tensor.
+   *  <p>
+   *  Reduces `input` along the dimensions given in `axis`. Unless
+   *  `keep_dims` is true, the rank of the tensor is reduced by 1 for each entry in
+   *  `axis`. If `keep_dims` is true, the reduced dimensions are
+   *  retained with length 1.
+   *
+   * @param <T> data type for {@code output()} output
+   * @param input The tensor to reduce.
+   * @param axis The dimensions to reduce. Must be in the range
+   *  `[-rank(input), rank(input))`.
+   * @param options carries optional attributes values
+   * @return a new instance of Max
+   */
+  public <T extends TType, U extends TNumber> Max<T> max(Operand<T> input, Operand<U> axis,
+      Max.Options... options) {
+    return Max.create(scope, input, axis, options);
+  }
+
+  /**
+   * Forwards the value of an available tensor from `inputs` to `output`.
+   *  <p>
+   *  `Merge` waits for at least one of the tensors in `inputs` to become available.
+   *  It is usually combined with `Switch` to implement branching.
+   *  <p>
+   *  `Merge` forwards the first tensor to become available to `output`, and sets
+   *  `value_index` to its index in `inputs`.
+   *
+   * @param <T> data type for {@code output()} output
+   * @param inputs The input tensors, exactly one of which will become available.
+   * @return a new instance of Merge
+   */
+  public <T extends TType> Merge<T> merge(Iterable<Operand<T>> inputs) {
+    return Merge.create(scope, inputs);
+  }
+
+  /**
+   * Computes the minimum of elements across dimensions of a tensor.
+   *  <p>
+   *  Reduces `input` along the dimensions given in `axis`. Unless
+   *  `keep_dims` is true, the rank of the tensor is reduced by 1 for each entry in
+   *  `axis`. If `keep_dims` is true, the reduced dimensions are
+   *  retained with length 1.
+   *
+   * @param <T> data type for {@code output()} output
+   * @param input The tensor to reduce.
+   * @param axis The dimensions to reduce. Must be in the range
+   *  `[-rank(input), rank(input))`.
+   * @param options carries optional attributes values
+   * @return a new instance of Min
+   */
+  public <T extends TType, U extends TNumber> Min<T> min(Operand<T> input, Operand<U> axis,
+      Min.Options... options) {
+    return Min.create(scope, input, axis, options);
+  }
+
+  /**
+   * Pads a tensor with mirrored values.
+   *  <p>
+   *  This operation pads a `input` with mirrored values according to the `paddings`
+   *  you specify. `paddings` is an integer tensor with shape `[n, 2]`, where n is
+   *  the rank of `input`. For each dimension D of `input`, `paddings[D, 0]` indicates
+   *  how many values to add before the contents of `input` in that dimension, and
+   *  `paddings[D, 1]` indicates how many values to add after the contents of `input`
+   *  in that dimension. Both `paddings[D, 0]` and `paddings[D, 1]` must be no greater
+   *  than `input.dim_size(D)` (or `input.dim_size(D) - 1`) if `copy_border` is true
+   *  (if false, respectively).
+   *  <p>
+   *  The padded size of each dimension D of the output is:
+   *  <p>
+   *  `paddings(D, 0) + input.dim_size(D) + paddings(D, 1)`
+   *  <p>
+   *  For example:
+   *  <pre>{@code
+   *  # 't' is [[1, 2, 3], [4, 5, 6]].
+   *  # 'paddings' is [[1, 1]], [2, 2]].
+   *  # 'mode' is SYMMETRIC.
+   *  # rank of 't' is 2.
+   *  pad(t, paddings) ==> [[2, 1, 1, 2, 3, 3, 2]
+   *                        [2, 1, 1, 2, 3, 3, 2]
+   *                        [5, 4, 4, 5, 6, 6, 5]
+   *                        [5, 4, 4, 5, 6, 6, 5]]
+   *  }</pre>
+   *
+   * @param <T> data type for {@code output()} output
+   * @param input The input tensor to be padded.
+   * @param paddings A two-column matrix specifying the padding sizes. The number of
+   *  rows must be the same as the rank of `input`.
+   * @param mode Either `REFLECT` or `SYMMETRIC`. In reflect mode the padded regions
+   *  do not include the borders, while in symmetric mode the padded regions
+   *  do include the borders. For example, if `input` is `[1, 2, 3]` and `paddings`
+   *  is `[0, 2]`, then the output is `[1, 2, 3, 2, 1]` in reflect mode, and
+   *  it is `[1, 2, 3, 3, 2]` in symmetric mode.
+   * @return a new instance of MirrorPad
+   */
+  public <T extends TType, U extends TNumber> MirrorPad<T> mirrorPad(Operand<T> input,
+      Operand<U> paddings, String mode) {
+    return MirrorPad.create(scope, input, paddings, mode);
+  }
+
+  /**
+   * Wraps an arbitrary MLIR computation expressed as a module with a main() function.
+   *  <p>
+   *  This operation does not have an associated kernel and is not intended to be
+   *  executed in a regular TensorFlow session. Instead it is intended to be used for
+   *  testing or for special case where a user intends to pass custom MLIR computation
+   *  through a TensorFlow graph with the intent of having custom tooling processing
+   *  it downstream (when targeting a different environment, like TensorFlow lite for
+   *  example).
+   *  The MLIR module is expected to have a main() function that will be used as an
+   *  entry point. The inputs to the operations will be passed as argument to the
+   *  main() function and the returned values of the main function mapped to the
+   *  outputs.
+   *  Example usage:
+   *  <pre>{@code
+   *  import tensorflow as tf
+   *  from tensorflow.compiler.mlir.tensorflow.gen_mlir_passthrough_op import mlir_passthrough_op
+   *
+   *  mlir_module = '''python
+   *  func @main(%arg0 : tensor<10xf32>, %arg1 : tensor<10xf32>) -> tensor<10x10xf32> {
+   *     %add = "magic.op"(%arg0, %arg1) : (tensor<10xf32>, tensor<10xf32>) -> tensor<10x10xf32>
+   *     return %ret : tensor<10x10xf32>
+   *  }
+   *  '''
+   *
+   * @tf.function def foo(x, y):
+   *    return mlir_passthrough_op([x, y], mlir_module, Toutputs=[tf.float32])
+   *
+   *  graph_def = foo.get_concrete_function(tf.TensorSpec([10], tf.float32), tf.TensorSpec([10], tf.float32)).graph.as_graph_def()
+   *  }</pre>
+   * @param inputs
+   * @param mlirModule
+   * @param Toutputs
+   * @return a new instance of MlirPassthroughOp
+   */
+  public MlirPassthroughOp mlirPassthroughOp(Iterable<Operand<?>> inputs, String mlirModule,
+      List<Class<? extends TType>> Toutputs) {
+    return MlirPassthroughOp.create(scope, inputs, mlirModule, Toutputs);
+  }
+
+  /**
+   * Creates an empty hash table that uses tensors as the backing store.
+   *  <p>
+   *  It uses "open addressing" with quadratic reprobing to resolve
+   *  collisions.
+   *  <p>
+   *  This op creates a mutable hash table, specifying the type of its keys and
+   *  values. Each value must be a scalar. Data can be inserted into the table using
+   *  the insert operations. It does not support the initialization operation.
+   *
+   * @param emptyKey The key used to represent empty key buckets internally. Must not
+   *  be used in insert or lookup operations.
+   * @param deletedKey
+   * @param valueDtype Type of the table values.
+   * @param options carries optional attributes values
+   * @return a new instance of MutableDenseHashTable
+   */
+  public <T extends TType, U extends TType> MutableDenseHashTable mutableDenseHashTable(
+      Operand<T> emptyKey, Operand<T> deletedKey, Class<U> valueDtype,
+      MutableDenseHashTable.Options... options) {
+    return MutableDenseHashTable.create(scope, emptyKey, deletedKey, valueDtype, options);
+  }
+
+  /**
+   * Creates an empty hash table.
+   *  <p>
+   *  This op creates a mutable hash table, specifying the type of its keys and
+   *  values. Each value must be a scalar. Data can be inserted into the table using
+   *  the insert operations. It does not support the initialization operation.
+   *
+   * @param keyDtype Type of the table keys.
+   * @param valueDtype Type of the table values.
+   * @param options carries optional attributes values
+   * @return a new instance of MutableHashTable
+   */
+  public <T extends TType, U extends TType> MutableHashTable mutableHashTable(Class<T> keyDtype,
+      Class<U> valueDtype, MutableHashTable.Options... options) {
+    return MutableHashTable.create(scope, keyDtype, valueDtype, options);
+  }
+
+  /**
+   * Creates an empty hash table.
+   *  <p>
+   *  This op creates a mutable hash table, specifying the type of its keys and
+   *  values. Each value must be a vector. Data can be inserted into the table using
+   *  the insert operations. It does not support the initialization operation.
+   *
+   * @param keyDtype Type of the table keys.
+   * @param valueDtype Type of the table values.
+   * @param options carries optional attributes values
+   * @return a new instance of MutableHashTableOfTensors
+   */
+  public <T extends TType, U extends TType> MutableHashTableOfTensors mutableHashTableOfTensors(
+      Class<T> keyDtype, Class<U> valueDtype, MutableHashTableOfTensors.Options... options) {
+    return MutableHashTableOfTensors.create(scope, keyDtype, valueDtype, options);
+  }
+
+  /**
+   * Creates a Mutex resource that can be locked by `MutexLock`.
+   *
+   * @param options carries optional attributes values
+   * @return a new instance of Mutex
+   */
+  public Mutex mutex(Mutex.Options... options) {
+    return Mutex.create(scope, options);
+  }
+
+  /**
+   * Locks a mutex resource.  The output is the lock.  So long as the lock tensor
+   *  <p>
+   *  is alive, any other request to use `MutexLock` with this mutex will wait.
+   *  <p>
+   *  This is particularly useful for creating a critical section when used in
+   *  conjunction with `MutexLockIdentity`:
+   *  <pre>{@code
+   *  mutex = mutex_v2(
+   *    shared_name=handle_name, container=container, name=name)
+   *
+   *  def execute_in_critical_section(fn, *args, **kwargs):
+   *    lock = gen_resource_variable_ops.mutex_lock(mutex)
+   *
+   *    with ops.control_dependencies([lock]):
+   *      r = fn(*args, **kwargs)
+   *
+   *    with ops.control_dependencies(nest.flatten(r)):
+   *      with ops.colocate_with(mutex):
+   *        ensure_lock_exists = mutex_lock_identity(lock)
+   *
+   *      # Make sure that if any element of r is accessed, all of
+   *      # them are executed together.
+   *      r = nest.map_structure(tf.identity, r)
+   *
+   *    with ops.control_dependencies([ensure_lock_exists]):
+   *      return nest.map_structure(tf.identity, r)
+   *  }</pre>
+   *  While `fn` is running in the critical section, no other functions which wish to
+   *  use this critical section may run.
+   *  <p>
+   *  Often the use case is that two executions of the same graph, in parallel,
+   *  wish to run `fn`; and we wish to ensure that only one of them executes
+   *  at a time.  This is especially important if `fn` modifies one or more
+   *  variables at a time.
+   *  <p>
+   *  It is also useful if two separate functions must share a resource, but we
+   *  wish to ensure the usage is exclusive.
+   *
+   * @param mutex The mutex resource to lock.
+   * @return a new instance of MutexLock
+   */
+  public MutexLock mutexLock(Operand<?> mutex) {
+    return MutexLock.create(scope, mutex);
+  }
+
+  /**
+   * Makes its input available to the next iteration.
+   *
+   * @param <T> data type for {@code output()} output
+   * @param data The tensor to be made available to the next iteration.
+   * @return a new instance of NextIteration
+   */
+  public <T extends TType> NextIteration<T> nextIteration(Operand<T> data) {
+    return NextIteration.create(scope, data);
+  }
+
+  /**
+   * Does nothing. Only useful as a placeholder for control edges.
+   *
+   * @return a new instance of NoOp
+   */
+  public NoOp noOp() {
+    return NoOp.create(scope);
+  }
+
+  /**
+   * Returns a one-hot tensor.
+   *  <p>
+   *  The locations represented by indices in `indices` take value `on_value`,
+   *  while all other locations take value `off_value`.
+   *  <p>
+   *  If the input `indices` is rank `N`, the output will have rank `N+1`,
+   *  The new axis is created at dimension `axis` (default: the new axis is
+   *  appended at the end).
+   *  <p>
+   *  If `indices` is a scalar the output shape will be a vector of length `depth`.
+   *  <p>
+   *  If `indices` is a vector of length `features`, the output shape will be:
+   *  <pre>{@code
+   *    features x depth if axis == -1
+   *    depth x features if axis == 0
+   *  }</pre>
+   *  If `indices` is a matrix (batch) with shape `[batch, features]`,
+   *  the output shape will be:
+   *  <pre>{@code
+   *    batch x features x depth if axis == -1
+   *    batch x depth x features if axis == 1
+   *    depth x batch x features if axis == 0
+   *  }</pre>
+   *  Examples
+   *  =========
+   *  <p>
+   *  Suppose that
+   *  <pre>{@code
+   *    indices = [0, 2, -1, 1]
+   *    depth = 3
+   *    on_value = 5.0
+   *    off_value = 0.0
+   *    axis = -1
+   *  }</pre>
+   *  Then output is `[4 x 3]`:
+   *  <pre>{@code
+   *  output =
+   *    [5.0 0.0 0.0]  // one_hot(0)
+   *    [0.0 0.0 5.0]  // one_hot(2)
+   *    [0.0 0.0 0.0]  // one_hot(-1)
+   *    [0.0 5.0 0.0]  // one_hot(1)
+   *  }</pre>
+   *  Suppose that
+   *  <pre>{@code
+   *    indices = [0, 2, -1, 1]
+   *    depth = 3
+   *    on_value = 0.0
+   *    off_value = 3.0
+   *    axis = 0
+   *  }</pre>
+   *  Then output is `[3 x 4]`:
+   *  <pre>{@code
+   *  output =
+   *    [0.0 3.0 3.0 3.0]
+   *    [3.0 3.0 3.0 0.0]
+   *    [3.0 3.0 3.0 3.0]
+   *    [3.0 0.0 3.0 3.0]
+   *  //  ^                one_hot(0)
+   *  //      ^            one_hot(2)
+   *  //          ^        one_hot(-1)
+   *  //              ^    one_hot(1)
+   *  }</pre>
+   *  Suppose that
+   *  <pre>{@code
+   *    indices = [[0, 2], [1, -1]]
+   *    depth = 3
+   *    on_value = 1.0
+   *    off_value = 0.0
+   *    axis = -1
+   *  }</pre>
+   *  Then output is `[2 x 2 x 3]`:
+   *  <pre>{@code
+   *  output =
+   *    [
+   *      [1.0, 0.0, 0.0]  // one_hot(0)
+   *      [0.0, 0.0, 1.0]  // one_hot(2)
+   *    ][
+   *      [0.0, 1.0, 0.0]  // one_hot(1)
+   *      [0.0, 0.0, 0.0]  // one_hot(-1)
+   *    ]
+   *  }</pre>
+   *
+   * @param <U> data type for {@code output()} output
+   * @param indices A tensor of indices.
+   * @param depth A scalar defining the depth of the one hot dimension.
+   * @param onValue A scalar defining the value to fill in output when `indices[j] = i`.
+   * @param offValue A scalar defining the value to fill in output when `indices[j] != i`.
+   * @param options carries optional attributes values
+   * @return a new instance of OneHot
+   */
+  public <U extends TType, T extends TNumber> OneHot<U> oneHot(Operand<T> indices,
+      Operand<TInt32> depth, Operand<U> onValue, Operand<U> offValue, OneHot.Options... options) {
+    return OneHot.create(scope, indices, depth, onValue, offValue, options);
+  }
+
+  /**
+   * Returns a tensor of ones with the same shape and type as x.
+   *
+   * @param <T> data type for {@code y()} output
+   * @param x a tensor of type T.
+   * @return a new instance of OnesLike
+   */
+  public <T extends TType> OnesLike<T> onesLike(Operand<T> x) {
+    return OnesLike.create(scope, x);
+  }
+
+  /**
+   * Op removes all elements in the underlying container.
+   *
+   * @param dtypes
+   * @param options carries optional attributes values
+   * @return a new instance of OrderedMapClear
+   */
+  public OrderedMapClear orderedMapClear(List<Class<? extends TType>> dtypes,
+      OrderedMapClear.Options... options) {
+    return OrderedMapClear.create(scope, dtypes, options);
+  }
+
+  /**
+   * Op returns the number of incomplete elements in the underlying container.
+   *
+   * @param dtypes
+   * @param options carries optional attributes values
+   * @return a new instance of OrderedMapIncompleteSize
+   */
+  public OrderedMapIncompleteSize orderedMapIncompleteSize(List<Class<? extends TType>> dtypes,
+      OrderedMapIncompleteSize.Options... options) {
+    return OrderedMapIncompleteSize.create(scope, dtypes, options);
+  }
+
+  /**
+   * Op peeks at the values at the specified key.  If the
+   *  <p>
+   *  underlying container does not contain this key
+   *  this op will block until it does.   This Op is optimized for
+   *  performance.
+   *
+   * @param key
+   * @param indices
+   * @param dtypes
+   * @param options carries optional attributes values
+   * @return a new instance of OrderedMapPeek
+   */
+  public OrderedMapPeek orderedMapPeek(Operand<TInt64> key, Operand<TInt32> indices,
+      List<Class<? extends TType>> dtypes, OrderedMapPeek.Options... options) {
+    return OrderedMapPeek.create(scope, key, indices, dtypes, options);
+  }
+
+  /**
+   * Op returns the number of elements in the underlying container.
+   *
+   * @param dtypes
+   * @param options carries optional attributes values
+   * @return a new instance of OrderedMapSize
+   */
+  public OrderedMapSize orderedMapSize(List<Class<? extends TType>> dtypes,
+      OrderedMapSize.Options... options) {
+    return OrderedMapSize.create(scope, dtypes, options);
+  }
+
+  /**
+   * Stage (key, values) in the underlying container which behaves like a ordered
+   *  <p>
+   *  associative container.   Elements are ordered by key.
+   *
+   * @param key int64
+   * @param indices
+   * @param values a list of tensors
+   *  dtypes A list of data types that inserted values should adhere to.
+   * @param dtypes
+   * @param options carries optional attributes values
+   * @return a new instance of OrderedMapStage
+   */
+  public OrderedMapStage orderedMapStage(Operand<TInt64> key, Operand<TInt32> indices,
+      Iterable<Operand<?>> values, List<Class<? extends TType>> dtypes,
+      OrderedMapStage.Options... options) {
+    return OrderedMapStage.create(scope, key, indices, values, dtypes, options);
+  }
+
+  /**
+   * Op removes and returns the values associated with the key
+   *  <p>
+   *  from the underlying container.   If the underlying container
+   *  does not contain this key, the op will block until it does.
+   *
+   * @param key
+   * @param indices
+   * @param dtypes
+   * @param options carries optional attributes values
+   * @return a new instance of OrderedMapUnstage
+   */
+  public OrderedMapUnstage orderedMapUnstage(Operand<TInt64> key, Operand<TInt32> indices,
+      List<Class<? extends TType>> dtypes, OrderedMapUnstage.Options... options) {
+    return OrderedMapUnstage.create(scope, key, indices, dtypes, options);
+  }
+
+  /**
+   * Op removes and returns the (key, value) element with the smallest
+   *  <p>
+   *  key from the underlying container.   If the underlying container
+   *  does not contain elements, the op will block until it does.
+   *
+   * @param indices
+   * @param dtypes
+   * @param options carries optional attributes values
+   * @return a new instance of OrderedMapUnstageNoKey
+   */
+  public OrderedMapUnstageNoKey orderedMapUnstageNoKey(Operand<TInt32> indices,
+      List<Class<? extends TType>> dtypes, OrderedMapUnstageNoKey.Options... options) {
+    return OrderedMapUnstageNoKey.create(scope, indices, dtypes, options);
+  }
+
+  /**
+   * Pads a tensor.
+   *  <p>
+   *  This operation pads `input` according to the `paddings` and `constant_values`
+   *  you specify. `paddings` is an integer tensor with shape `[Dn, 2]`, where n is
+   *  the rank of `input`. For each dimension D of `input`, `paddings[D, 0]` indicates
+   *  how many padding values to add before the contents of `input` in that dimension,
+   *  and `paddings[D, 1]` indicates how many padding values to add after the contents
+   *  of `input` in that dimension. `constant_values` is a scalar tensor of the same
+   *  type as `input` that indicates the value to use for padding `input`.
+   *  <p>
+   *  The padded size of each dimension D of the output is:
+   *  <p>
+   *  `paddings(D, 0) + input.dim_size(D) + paddings(D, 1)`
+   *  <p>
+   *  For example:
+   *  <pre>{@code
+   *  # 't' is [[1, 1], [2, 2]]
+   *  # 'paddings' is [[1, 1], [2, 2]]
+   *  # 'constant_values' is 0
+   *  # rank of 't' is 2
+   *  pad(t, paddings) ==> [[0, 0, 0, 0, 0, 0]
+   *                        [0, 0, 1, 1, 0, 0]
+   *                        [0, 0, 2, 2, 0, 0]
+   *                        [0, 0, 0, 0, 0, 0]]
+   *  }</pre>
+   *
+   * @param <T> data type for {@code output()} output
+   * @param input
+   * @param paddings
+   * @param constantValues
+   * @return a new instance of Pad
+   */
+  public <T extends TType, U extends TNumber> Pad<T> pad(Operand<T> input, Operand<U> paddings,
+      Operand<T> constantValues) {
+    return Pad.create(scope, input, paddings, constantValues);
+  }
+
+  /**
+   * Concatenates a list of `N` tensors along the first dimension.
+   *  <p>
+   *  The input tensors are all required to have size 1 in the first dimension.
+   *  <p>
+   *  For example:
+   *  <pre>{@code
+   *  # 'x' is [[1, 4]]
+   *  # 'y' is [[2, 5]]
+   *  # 'z' is [[3, 6]]
+   *  parallel_concat([x, y, z]) => [[1, 4], [2, 5], [3, 6]]  # Pack along first dim.
+   *  }</pre>
+   *  The difference between concat and parallel_concat is that concat requires all
+   *  of the inputs be computed before the operation will begin but doesn't require
+   *  that the input shapes be known during graph construction.  Parallel concat
+   *  will copy pieces of the input into the output as they become available, in
+   *  some situations this can provide a performance benefit.
+   *
+   * @param <T> data type for {@code output()} output
+   * @param values Tensors to be concatenated. All must have size 1 in the first dimension
+   *  and same shape.
+   * @param shape the final shape of the result; should be equal to the shapes of any input
+   *  but with the number of input values in the first dimension.
+   * @return a new instance of ParallelConcat
+   */
+  public <T extends TType> ParallelConcat<T> parallelConcat(Iterable<Operand<T>> values,
+      Shape shape) {
+    return ParallelConcat.create(scope, values, shape);
+  }
+
+  /**
+   * Interleave the values from the `data` tensors into a single tensor.
+   *  <p>
+   *  Builds a merged tensor such that
+   *  <pre>{@code
+   *      merged[indices[m][i, ..., j], ...] = data[m][i, ..., j, ...]
+   *  }</pre>
+   *  For example, if each `indices[m]` is scalar or vector, we have
+   *  <pre>{@code
+   *      # Scalar indices:
+   *      merged[indices[m], ...] = data[m][...]
+   *
+   *      # Vector indices:
+   *      merged[indices[m][i], ...] = data[m][i, ...]
+   *  }</pre>
+   *  Each `data[i].shape` must start with the corresponding `indices[i].shape`,
+   *  and the rest of `data[i].shape` must be constant w.r.t. `i`.  That is, we
+   *  must have `data[i].shape = indices[i].shape + constant`.  In terms of this
+   *  `constant`, the output shape is
+   *  <p>
+   *      merged.shape = [max(indices)] + constant
+   *  <p>
+   *  Values may be merged in parallel, so if an index appears in both `indices[m][i]`
+   *  and `indices[n][j]`, the result may be invalid. This differs from the normal
+   *  DynamicStitch operator that defines the behavior in that case.
+   *  <p>
+   *  For example:
+   *  <pre>{@code
+   *      indices[0] = 6
+   *      indices[1] = [4, 1]
+   *      indices[2] = [[5, 2], [0, 3]]
+   *      data[0] = [61, 62]
+   *      data[1] = [[41, 42], [11, 12]]
+   *      data[2] = [[[51, 52], [21, 22]], [[1, 2], [31, 32]]]
+   *      merged = [[1, 2], [11, 12], [21, 22], [31, 32], [41, 42],
+   *                [51, 52], [61, 62]]
+   *  }</pre>
+   *  This method can be used to merge partitions created by `dynamic_partition`
+   *  as illustrated on the following example:
+   *  <pre>{@code
+   *      # Apply function (increments x_i) on elements for which a certain condition
+   *      # apply (x_i != -1 in this example).
+   *      x=tf.constant([0.1, -1., 5.2, 4.3, -1., 7.4])
+   *      condition_mask=tf.not_equal(x,tf.constant(-1.))
+   *      partitioned_data = tf.dynamic_partition(
+   *          x, tf.cast(condition_mask, tf.int32) , 2)
+   *      partitioned_data[1] = partitioned_data[1] + 1.0
+   *      condition_indices = tf.dynamic_partition(
+   *          tf.range(tf.shape(x)[0]), tf.cast(condition_mask, tf.int32) , 2)
+   *      x = tf.dynamic_stitch(condition_indices, partitioned_data)
+   *      # Here x=[1.1, -1., 6.2, 5.3, -1, 8.4], the -1. values remain
+   *      # unchanged.
+   *  }</pre>
+   *  <div style="width:70%; margin:auto; margin-bottom:10px; margin-top:20px;">
+   *  <img style="width:100%" src="https://www.tensorflow.org/images/DynamicStitch.png" alt>
+   *  </div>
+   *
+   * @param <T> data type for {@code merged()} output
+   * @param indices
+   * @param data
+   * @return a new instance of ParallelDynamicStitch
+   */
+  public <T extends TType> ParallelDynamicStitch<T> parallelDynamicStitch(
+      Iterable<Operand<TInt32>> indices, Iterable<Operand<T>> data) {
+    return ParallelDynamicStitch.create(scope, indices, data);
+  }
+
+  /**
+   * A placeholder op for a value that will be fed into the computation.
+   *  <p>
+   *  N.B. This operation will fail with an error if it is executed. It is
+   *  intended as a way to represent a value that will always be fed, and to
+   *  provide attrs that enable the fed value to be checked at runtime.
+   *
+   * @param <T> data type for {@code output()} output
+   * @param dtype The type of elements in the tensor.
+   * @param options carries optional attributes values
+   * @return a new instance of Placeholder
+   */
+  public <T extends TType> Placeholder<T> placeholder(Class<T> dtype,
+      Placeholder.Options... options) {
+    return Placeholder.create(scope, dtype, options);
+  }
+
+  /**
+   * A placeholder op that passes through `input` when its output is not fed.
+   *
+   * @param <T> data type for {@code output()} output
+   * @param input The default value to produce when `output` is not fed.
+   * @param shape The (possibly partial) shape of the tensor.
+   * @return a new instance of PlaceholderWithDefault
+   */
+  public <T extends TType> PlaceholderWithDefault<T> placeholderWithDefault(Operand<T> input,
+      Shape shape) {
+    return PlaceholderWithDefault.create(scope, input, shape);
+  }
+
+  /**
+   * Prints a string scalar.
+   *  <p>
+   *  Prints a string scalar to the desired output_stream.
+   *
+   * @param input The string scalar to print.
+   * @param options carries optional attributes values
+   * @return a new instance of Print
+   */
+  public Print print(Operand<TString> input, Print.Options... options) {
+    return Print.create(scope, input, options);
+  }
+
+  /**
+   * Computes the product of elements across dimensions of a tensor.
+   *  <p>
+   *  Reduces `input` along the dimensions given in `axis`. Unless
+   *  `keep_dims` is true, the rank of the tensor is reduced by 1 for each entry in
+   *  `axis`. If `keep_dims` is true, the reduced dimensions are
+   *  retained with length 1.
+   *
+   * @param <T> data type for {@code output()} output
+   * @param input The tensor to reduce.
+   * @param axis The dimensions to reduce. Must be in the range
+   *  `[-rank(input), rank(input))`.
+   * @param options carries optional attributes values
+   * @return a new instance of Prod
+   */
+  public <T extends TType, U extends TNumber> Prod<T> prod(Operand<T> input, Operand<U> axis,
+      Prod.Options... options) {
+    return Prod.create(scope, input, axis, options);
+  }
+
+  /**
+   * Reshapes a quantized tensor as per the Reshape op.
+   *  <p>
+   *  ```
+   *
+   * @param <T> data type for {@code output()} output
+   * @param tensor
+   * @param shape Defines the shape of the output tensor.
+   * @param inputMin The minimum value of the input.
+   * @param inputMax The maximum value of the input.
+   * @return a new instance of QuantizedReshape
+   */
+  public <T extends TType, U extends TNumber> QuantizedReshape<T> quantizedReshape(
+      Operand<T> tensor, Operand<U> shape, Operand<TFloat32> inputMin, Operand<TFloat32> inputMax) {
+    return QuantizedReshape.create(scope, tensor, shape, inputMin, inputMax);
+  }
+
+  /**
+   * Creates a sequence of numbers.
+   *  <p>
+   *  This operation creates a sequence of numbers that begins at `start` and
+   *  extends by increments of `delta` up to but not including `limit`.
+   *  <p>
+   *  For example:
+   *  <pre>{@code
+   *  # 'start' is 3
+   *  # 'limit' is 18
+   *  # 'delta' is 3
+   *  tf.range(start, limit, delta) ==> [3, 6, 9, 12, 15]
+   *  }</pre>
+   *
+   * @param <T> data type for {@code output()} output
+   * @param start 0-D (scalar). First entry in the sequence.
+   * @param limit 0-D (scalar). Upper limit of sequence, exclusive.
+   * @param delta 0-D (scalar). Optional. Default is 1. Number that increments `start`.
+   * @return a new instance of Range
+   */
+  public <T extends TNumber> Range<T> range(Operand<T> start, Operand<T> limit, Operand<T> delta) {
+    return Range.create(scope, start, limit, delta);
+  }
+
+  /**
+   * Returns the rank of a tensor.
+   *  <p>
+   *  This operation returns an integer representing the rank of `input`.
+   *  <p>
+   *  For example:
+   *  <pre>{@code
+   *  # 't' is [[[1, 1, 1], [2, 2, 2]], [[3, 3, 3], [4, 4, 4]]]
+   *  # shape of tensor 't' is [2, 2, 3]
+   *  rank(t) ==> 3
+   *  }</pre>
+   *  <b>Note</b>: The rank of a tensor is not the same as the rank of a matrix. The rank
+   *  of a tensor is the number of indices required to uniquely select each element
+   *  of the tensor. Rank is also known as "order", "degree", or "ndims."
+   *
+   * @param input
+   * @return a new instance of Rank
+   */
+  public <T extends TType> Rank rank(Operand<T> input) {
+    return Rank.create(scope, input);
+  }
+
+  /**
+   * Reads the value of a variable.
+   *  <p>
+   *  The tensor returned by this operation is immutable.
+   *  <p>
+   *  The value returned by this operation is guaranteed to be influenced by all the
+   *  writes on which this operation depends directly or indirectly, and to not be
+   *  influenced by any of the writes which depend directly or indirectly on this
+   *  operation.
+   *
+   * @param <T> data type for {@code value()} output
+   * @param resource handle to the resource in which to store the variable.
+   * @param dtype the dtype of the value.
+   * @return a new instance of ReadVariableOp
+   */
+  public <T extends TType> ReadVariableOp<T> readVariableOp(Operand<?> resource, Class<T> dtype) {
+    return ReadVariableOp.create(scope, resource, dtype);
+  }
+
+  /**
+   * Computes the "logical and" of elements across dimensions of a tensor.
+   *  <p>
+   *  Reduces `input` along the dimensions given in `axis`. Unless
+   *  `keep_dims` is true, the rank of the tensor is reduced by 1 for each entry in
+   *  `axis`. If `keep_dims` is true, the reduced dimensions are
+   *  retained with length 1.
+   *
+   * @param input The tensor to reduce.
+   * @param axis The dimensions to reduce. Must be in the range
+   *  `[-rank(input), rank(input))`.
+   * @param options carries optional attributes values
+   * @return a new instance of ReduceAll
+   */
+  public <T extends TNumber> ReduceAll reduceAll(Operand<TBool> input, Operand<T> axis,
+      ReduceAll.Options... options) {
+    return ReduceAll.create(scope, input, axis, options);
+  }
+
+  /**
+   * Computes the "logical or" of elements across dimensions of a tensor.
+   *  <p>
+   *  Reduces `input` along the dimensions given in `axis`. Unless
+   *  `keep_dims` is true, the rank of the tensor is reduced by 1 for each entry in
+   *  `axis`. If `keep_dims` is true, the reduced dimensions are
+   *  retained with length 1.
+   *
+   * @param input The tensor to reduce.
+   * @param axis The dimensions to reduce. Must be in the range
+   *  `[-rank(input), rank(input))`.
+   * @param options carries optional attributes values
+   * @return a new instance of ReduceAny
+   */
+  public <T extends TNumber> ReduceAny reduceAny(Operand<TBool> input, Operand<T> axis,
+      ReduceAny.Options... options) {
+    return ReduceAny.create(scope, input, axis, options);
+  }
+
+  /**
+   * Computes the maximum of elements across dimensions of a tensor.
+   *  <p>
+   *  Reduces `input` along the dimensions given in `axis`. Unless
+   *  `keep_dims` is true, the rank of the tensor is reduced by 1 for each entry in
+   *  `axis`. If `keep_dims` is true, the reduced dimensions are
+   *  retained with length 1.
+   *
+   * @param <T> data type for {@code output()} output
+   * @param input The tensor to reduce.
+   * @param axis The dimensions to reduce. Must be in the range
+   *  `[-rank(input), rank(input))`.
+   * @param options carries optional attributes values
+   * @return a new instance of ReduceMax
+   */
+  public <T extends TType, U extends TNumber> ReduceMax<T> reduceMax(Operand<T> input,
+      Operand<U> axis, ReduceMax.Options... options) {
+    return ReduceMax.create(scope, input, axis, options);
+  }
+
+  /**
+   * Computes the minimum of elements across dimensions of a tensor.
+   *  <p>
+   *  Reduces `input` along the dimensions given in `axis`. Unless
+   *  `keep_dims` is true, the rank of the tensor is reduced by 1 for each entry in
+   *  `axis`. If `keep_dims` is true, the reduced dimensions are
+   *  retained with length 1.
+   *
+   * @param <T> data type for {@code output()} output
+   * @param input The tensor to reduce.
+   * @param axis The dimensions to reduce. Must be in the range
+   *  `[-rank(input), rank(input))`.
+   * @param options carries optional attributes values
+   * @return a new instance of ReduceMin
+   */
+  public <T extends TType, U extends TNumber> ReduceMin<T> reduceMin(Operand<T> input,
+      Operand<U> axis, ReduceMin.Options... options) {
+    return ReduceMin.create(scope, input, axis, options);
+  }
+
+  /**
+   * Computes the product of elements across dimensions of a tensor.
+   *  <p>
+   *  Reduces `input` along the dimensions given in `axis`. Unless
+   *  `keep_dims` is true, the rank of the tensor is reduced by 1 for each entry in
+   *  `axis`. If `keep_dims` is true, the reduced dimensions are
+   *  retained with length 1.
+   *
+   * @param <T> data type for {@code output()} output
+   * @param input The tensor to reduce.
+   * @param axis The dimensions to reduce. Must be in the range
+   *  `[-rank(input), rank(input))`.
+   * @param options carries optional attributes values
+   * @return a new instance of ReduceProd
+   */
+  public <T extends TType, U extends TNumber> ReduceProd<T> reduceProd(Operand<T> input,
+      Operand<U> axis, ReduceProd.Options... options) {
+    return ReduceProd.create(scope, input, axis, options);
+  }
+
+  /**
+   * Computes the sum of elements across dimensions of a tensor.
+   *  <p>
+   *  Reduces `input` along the dimensions given in `axis`. Unless
+   *  `keep_dims` is true, the rank of the tensor is reduced by 1 for each entry in
+   *  `axis`. If `keep_dims` is true, the reduced dimensions are
+   *  retained with length 1.
+   *
+   * @param <T> data type for {@code output()} output
+   * @param input The tensor to reduce.
+   * @param axis The dimensions to reduce. Must be in the range
+   *  `[-rank(input), rank(input))`.
+   * @param options carries optional attributes values
+   * @return a new instance of ReduceSum
+   */
+  public <T extends TType, U extends TNumber> ReduceSum<T> reduceSum(Operand<T> input,
+      Operand<U> axis, ReduceSum.Options... options) {
+    return ReduceSum.create(scope, input, axis, options);
+  }
+
+  /**
+   * Makes its input available to the next iteration.
+   *
+   * @param <T> data type for {@code output()} output
+   * @param data The tensor to be made available to the next iteration.
+   * @return a new instance of RefNextIteration
+   */
+  public <T extends TType> RefNextIteration<T> refNextIteration(Operand<T> data) {
+    return RefNextIteration.create(scope, data);
+  }
+
+  /**
+   * Forwards the `index`th element of `inputs` to `output`.
+   *
+   * @param <T> data type for {@code output()} output
+   * @param index A scalar that determines the input that gets selected.
+   * @param inputs A list of ref tensors, one of which will be forwarded to `output`.
+   * @return a new instance of RefSelect
+   */
+  public <T extends TType> RefSelect<T> refSelect(Operand<TInt32> index,
+      Iterable<Operand<T>> inputs) {
+    return RefSelect.create(scope, index, inputs);
+  }
+
+  /**
+   * Forwards the ref tensor `data` to the output port determined by `pred`.
+   *  <p>
+   *  If `pred` is true, the `data` input is forwarded to `output_true`. Otherwise,
+   *  the data goes to `output_false`.
+   *  <p>
+   *  See also `Switch` and `Merge`.
+   *
+   * @param <T> data type for {@code outputFalse()} output
+   * @param data The ref tensor to be forwarded to the appropriate output.
+   * @param pred A scalar that specifies which output port will receive data.
+   * @return a new instance of RefSwitch
+   */
+  public <T extends TType> RefSwitch<T> refSwitch(Operand<T> data, Operand<TBool> pred) {
+    return RefSwitch.create(scope, data, pred);
+  }
+
+  /**
+   * Execute a sub graph on a remote processor.
+   *  <p>
+   *  The graph specifications(such as graph itself, input tensors and output names)
+   *  are stored as a serialized protocol buffer of RemoteFusedGraphExecuteInfo
+   *  as serialized_remote_fused_graph_execute_info.
+   *  The specifications will be passed to a dedicated registered
+   *  remote fused graph executor.  The executor will send the graph specifications
+   *  to a remote processor and execute that graph.  The execution results
+   *  will be passed to consumer nodes as outputs of this node.
+   *
+   * @param inputs Arbitrary number of tensors with arbitrary data types
+   * @param Toutputs
+   * @param serializedRemoteFusedGraphExecuteInfo Serialized protocol buffer
+   *  of RemoteFusedGraphExecuteInfo which contains graph specifications.
+   * @return a new instance of RemoteFusedGraphExecute
+   */
+  public RemoteFusedGraphExecute remoteFusedGraphExecute(Iterable<Operand<?>> inputs,
+      List<Class<? extends TType>> Toutputs, String serializedRemoteFusedGraphExecuteInfo) {
+    return RemoteFusedGraphExecute.create(scope, inputs, Toutputs, serializedRemoteFusedGraphExecuteInfo);
+  }
+
+  /**
+   * Reshapes a tensor.
+   *  <p>
+   *  Given `tensor`, this operation returns a tensor that has the same values
+   *  as `tensor` with shape `shape`.
+   *  <p>
+   *  If one component of 1-D tensor `shape` is the special value -1, the size of that
+   *  dimension is computed so that the total size remains constant.  In particular, a
+   *  `shape` of `[-1]` flattens into 1-D.  At most one component of `shape` may be
+   *  unknown.
+   *  <p>
+   *  The `shape` must be 1-D and the operation returns a tensor with shape
+   *  `shape` filled with the values of `tensor`. In this case, the number of elements
+   *  implied by `shape` must be the same as the number of elements in `tensor`.
+   *  <p>
+   *  It is an error if `shape` is not 1-D.
+   *  <p>
+   *  For example:
+   *  <pre>{@code
+   *  # tensor 't' is [1, 2, 3, 4, 5, 6, 7, 8, 9]
+   *  # tensor 't' has shape [9]
+   *  reshape(t, [3, 3]) ==> [[1, 2, 3],
+   *                          [4, 5, 6],
+   *                          [7, 8, 9]]
+   *
+   *  # tensor 't' is [[[1, 1], [2, 2]],
+   *  #                [[3, 3], [4, 4]]]
+   *  # tensor 't' has shape [2, 2, 2]
+   *  reshape(t, [2, 4]) ==> [[1, 1, 2, 2],
+   *                          [3, 3, 4, 4]]
+   *
+   *  # tensor 't' is [[[1, 1, 1],
+   *  #                 [2, 2, 2]],
+   *  #                [[3, 3, 3],
+   *  #                 [4, 4, 4]],
+   *  #                [[5, 5, 5],
+   *  #                 [6, 6, 6]]]
+   *  # tensor 't' has shape [3, 2, 3]
+   *  # pass '[-1]' to flatten 't'
+   *  reshape(t, [-1]) ==> [1, 1, 1, 2, 2, 2, 3, 3, 3, 4, 4, 4, 5, 5, 5, 6, 6, 6]
+   *
+   *  # -1 can also be used to infer the shape
+   *
+   *  # -1 is inferred to be 9:
+   *  reshape(t, [2, -1]) ==> [[1, 1, 1, 2, 2, 2, 3, 3, 3],
+   *                           [4, 4, 4, 5, 5, 5, 6, 6, 6]]
+   *  # -1 is inferred to be 2:
+   *  reshape(t, [-1, 9]) ==> [[1, 1, 1, 2, 2, 2, 3, 3, 3],
+   *                           [4, 4, 4, 5, 5, 5, 6, 6, 6]]
+   *  # -1 is inferred to be 3:
+   *  reshape(t, [ 2, -1, 3]) ==> [[[1, 1, 1],
+   *                                [2, 2, 2],
+   *                                [3, 3, 3]],
+   *                               [[4, 4, 4],
+   *                                [5, 5, 5],
+   *                                [6, 6, 6]]]
+   *
+   *  # tensor 't' is [7]
+   *  # shape `[]` reshapes to a scalar
+   *  reshape(t, []) ==> 7
+   *  }</pre>
+   *
+   * @param <T> data type for {@code output()} output
+   * @param tensor
+   * @param shape Defines the shape of the output tensor.
+   * @return a new instance of Reshape
+   */
+  public <T extends TType, U extends TNumber> Reshape<T> reshape(Operand<T> tensor,
+      Operand<U> shape) {
+    return Reshape.create(scope, tensor, shape);
+  }
+
+  /**
+   * Increments variable pointed to by 'resource' until it reaches 'limit'.
+   *
+   * @param <T> data type for {@code output()} output
+   * @param resource Should be from a scalar `Variable` node.
+   * @param limit If incrementing ref would bring it above limit, instead generates an
+   *  'OutOfRange' error.
+   * @param T
+   * @return a new instance of ResourceCountUpTo
+   */
+  public <T extends TNumber> ResourceCountUpTo<T> resourceCountUpTo(Operand<?> resource, Long limit,
+      Class<T> T) {
+    return ResourceCountUpTo.create(scope, resource, limit, T);
+  }
+
+  /**
+   * Gather slices from the variable pointed to by `resource` according to `indices`.
+   *  <p>
+   *  `indices` must be an integer tensor of any dimension (usually 0-D or 1-D).
+   *  Produces an output tensor with shape `indices.shape + params.shape[1:]` where:
+   *  <pre>{@code
+   *      # Scalar indices
+   *      output[:, ..., :] = params[indices, :, ... :]
+   *
+   *      # Vector indices
+   *      output[i, :, ..., :] = params[indices[i], :, ... :]
+   *
+   *      # Higher rank indices
+   *      output[i, ..., j, :, ... :] = params[indices[i, ..., j], :, ..., :]
+   *  }</pre>
+   *
+   * @param <U> data type for {@code output()} output
+   * @param resource
+   * @param indices
+   * @param dtype
+   * @param options carries optional attributes values
+   * @return a new instance of ResourceGather
+   */
+  public <U extends TType, T extends TNumber> ResourceGather<U> resourceGather(Operand<?> resource,
+      Operand<T> indices, Class<U> dtype, ResourceGather.Options... options) {
+    return ResourceGather.create(scope, resource, indices, dtype, options);
+  }
+
+  /**
+   *
+   * @param <U> data type for {@code output()} output
+   * @param resource
+   * @param indices
+   * @param dtype
+   * @return a new instance of ResourceGatherNd
+   */
+  public <U extends TType, T extends TNumber> ResourceGatherNd<U> resourceGatherNd(
+      Operand<?> resource, Operand<T> indices, Class<U> dtype) {
+    return ResourceGatherNd.create(scope, resource, indices, dtype);
+  }
+
+  /**
+   * Adds sparse updates to the variable referenced by `resource`.
+   *  <p>
+   *  This operation computes
+   *  <p>
+   *      # Scalar indices
+   *      ref[indices, ...] += updates[...]
+   *  <p>
+   *      # Vector indices (for each i)
+   *      ref[indices[i], ...] += updates[i, ...]
+   *  <p>
+   *      # High rank indices (for each i, ..., j)
+   *      ref[indices[i, ..., j], ...] += updates[i, ..., j, ...]
+   *  <p>
+   *  Duplicate entries are handled correctly: if multiple `indices` reference
+   *  the same location, their contributions add.
+   *  <p>
+   *  Requires `updates.shape = indices.shape + ref.shape[1:]` or `updates.shape = []`.
+   *  <p>
+   *  <div style="width:70%; margin:auto; margin-bottom:10px; margin-top:20px;">
+   *  <img style="width:100%" src='https://www.tensorflow.org/images/ScatterAdd.png' alt>
+   *  </div>
+   *
+   * @param resource Should be from a `Variable` node.
+   * @param indices A tensor of indices into the first dimension of `ref`.
+   * @param updates A tensor of updated values to add to `ref`.
+   * @return a new instance of ResourceScatterAdd
+   */
+  public <T extends TNumber, U extends TType> ResourceScatterAdd resourceScatterAdd(
+      Operand<?> resource, Operand<T> indices, Operand<U> updates) {
+    return ResourceScatterAdd.create(scope, resource, indices, updates);
+  }
+
+  /**
+   * Divides sparse updates into the variable referenced by `resource`.
+   *  <p>
+   *  This operation computes
+   *  <p>
+   *      # Scalar indices
+   *      ref[indices, ...] /= updates[...]
+   *  <p>
+   *      # Vector indices (for each i)
+   *      ref[indices[i], ...] /= updates[i, ...]
+   *  <p>
+   *      # High rank indices (for each i, ..., j)
+   *      ref[indices[i, ..., j], ...] /= updates[i, ..., j, ...]
+   *  <p>
+   *  Duplicate entries are handled correctly: if multiple `indices` reference
+   *  the same location, their contributions multiply.
+   *  <p>
+   *  Requires `updates.shape = indices.shape + ref.shape[1:]` or `updates.shape = []`.
+   *  <p>
+   *  <div style="width:70%; margin:auto; margin-bottom:10px; margin-top:20px;">
+   *  <img style="width:100%" src='https://www.tensorflow.org/images/ScatterAdd.png' alt>
+   *  </div>
+   *
+   * @param resource Should be from a `Variable` node.
+   * @param indices A tensor of indices into the first dimension of `ref`.
+   * @param updates A tensor of updated values to add to `ref`.
+   * @return a new instance of ResourceScatterDiv
+   */
+  public <T extends TNumber, U extends TType> ResourceScatterDiv resourceScatterDiv(
+      Operand<?> resource, Operand<T> indices, Operand<U> updates) {
+    return ResourceScatterDiv.create(scope, resource, indices, updates);
+  }
+
+  /**
+   * Reduces sparse updates into the variable referenced by `resource` using the `max` operation.
+   *  <p>
+   *  This operation computes
+   *  <p>
+   *      # Scalar indices
+   *      ref[indices, ...] = max(ref[indices, ...], updates[...])
+   *  <p>
+   *      # Vector indices (for each i)
+   *      ref[indices[i], ...] = max(ref[indices[i], ...], updates[i, ...])
+   *  <p>
+   *      # High rank indices (for each i, ..., j)
+   *      ref[indices[i, ..., j], ...] = max(ref[indices[i, ..., j], ...], updates[i, ..., j, ...])
+   *  <p>
+   *  Duplicate entries are handled correctly: if multiple `indices` reference
+   *  the same location, their contributions are combined.
+   *  <p>
+   *  Requires `updates.shape = indices.shape + ref.shape[1:]` or `updates.shape = []`.
+   *  <p>
+   *  <div style="width:70%; margin:auto; margin-bottom:10px; margin-top:20px;">
+   *  <img style="width:100%" src='https://www.tensorflow.org/images/ScatterAdd.png' alt>
+   *  </div>
+   *
+   * @param resource Should be from a `Variable` node.
+   * @param indices A tensor of indices into the first dimension of `ref`.
+   * @param updates A tensor of updated values to add to `ref`.
+   * @return a new instance of ResourceScatterMax
+   */
+  public <T extends TNumber, U extends TType> ResourceScatterMax resourceScatterMax(
+      Operand<?> resource, Operand<T> indices, Operand<U> updates) {
+    return ResourceScatterMax.create(scope, resource, indices, updates);
+  }
+
+  /**
+   * Reduces sparse updates into the variable referenced by `resource` using the `min` operation.
+   *  <p>
+   *  This operation computes
+   *  <p>
+   *      # Scalar indices
+   *      ref[indices, ...] = min(ref[indices, ...], updates[...])
+   *  <p>
+   *      # Vector indices (for each i)
+   *      ref[indices[i], ...] = min(ref[indices[i], ...], updates[i, ...])
+   *  <p>
+   *      # High rank indices (for each i, ..., j)
+   *      ref[indices[i, ..., j], ...] = min(ref[indices[i, ..., j], ...], updates[i, ..., j, ...])
+   *  <p>
+   *  Duplicate entries are handled correctly: if multiple `indices` reference
+   *  the same location, their contributions are combined.
+   *  <p>
+   *  Requires `updates.shape = indices.shape + ref.shape[1:]` or `updates.shape = []`.
+   *  <p>
+   *  <div style="width:70%; margin:auto; margin-bottom:10px; margin-top:20px;">
+   *  <img style="width:100%" src='https://www.tensorflow.org/images/ScatterAdd.png' alt>
+   *  </div>
+   *
+   * @param resource Should be from a `Variable` node.
+   * @param indices A tensor of indices into the first dimension of `ref`.
+   * @param updates A tensor of updated values to add to `ref`.
+   * @return a new instance of ResourceScatterMin
+   */
+  public <T extends TNumber, U extends TType> ResourceScatterMin resourceScatterMin(
+      Operand<?> resource, Operand<T> indices, Operand<U> updates) {
+    return ResourceScatterMin.create(scope, resource, indices, updates);
+  }
+
+  /**
+   * Multiplies sparse updates into the variable referenced by `resource`.
+   *  <p>
+   *  This operation computes
+   *  <p>
+   *      # Scalar indices
+   *      ref[indices, ...] *= updates[...]
+   *  <p>
+   *      # Vector indices (for each i)
+   *      ref[indices[i], ...] *= updates[i, ...]
+   *  <p>
+   *      # High rank indices (for each i, ..., j)
+   *      ref[indices[i, ..., j], ...] *= updates[i, ..., j, ...]
+   *  <p>
+   *  Duplicate entries are handled correctly: if multiple `indices` reference
+   *  the same location, their contributions multiply.
+   *  <p>
+   *  Requires `updates.shape = indices.shape + ref.shape[1:]` or `updates.shape = []`.
+   *  <p>
+   *  <div style="width:70%; margin:auto; margin-bottom:10px; margin-top:20px;">
+   *  <img style="width:100%" src='https://www.tensorflow.org/images/ScatterAdd.png' alt>
+   *  </div>
+   *
+   * @param resource Should be from a `Variable` node.
+   * @param indices A tensor of indices into the first dimension of `ref`.
+   * @param updates A tensor of updated values to add to `ref`.
+   * @return a new instance of ResourceScatterMul
+   */
+  public <T extends TNumber, U extends TType> ResourceScatterMul resourceScatterMul(
+      Operand<?> resource, Operand<T> indices, Operand<U> updates) {
+    return ResourceScatterMul.create(scope, resource, indices, updates);
+  }
+
+  /**
+   * Applies sparse addition to individual values or slices in a Variable.
+   *  <p>
+   *  `ref` is a `Tensor` with rank `P` and `indices` is a `Tensor` of rank `Q`.
+   *  <p>
+   *  `indices` must be integer tensor, containing indices into `ref`.
+   *  It must be shape `[d_0, ..., d_{Q-2}, K]` where `0 < K <= P`.
+   *  <p>
+   *  The innermost dimension of `indices` (with length `K`) corresponds to
+   *  indices into elements (if `K = P`) or slices (if `K < P`) along the `K`th
+   *  dimension of `ref`.
+   *  <p>
+   *  `updates` is `Tensor` of rank `Q-1+P-K` with shape:
+   *  <pre>{@code
+   *  [d_0, ..., d_{Q-2}, ref.shape[K], ..., ref.shape[P-1]]
+   *  }</pre>
+   *  For example, say we want to add 4 scattered elements to a rank-1 tensor to
+   *  8 elements. In Python, that addition would look like this:
+   *  <pre>{@code
+   *  ref = tf.Variable([1, 2, 3, 4, 5, 6, 7, 8], use_resource=True)
+   *  indices = tf.constant([[4], [3], [1], [7]])
+   *  updates = tf.constant([9, 10, 11, 12])
+   *  add = tf.scatter_nd_add(ref, indices, updates)
+   *  with tf.Session() as sess:
+   *    print sess.run(add)
+   *  }</pre>
+   *  The resulting update to ref would look like this:
+   *  <p>
+   *      [1, 13, 3, 14, 14, 6, 7, 20]
+   *  <p>
+   *  See `tf.scatter_nd` for more details about how to make updates to
+   *  slices.
+   *
+   * @param ref A resource handle. Must be from a VarHandleOp.
+   * @param indices A Tensor. Must be one of the following types: int32, int64.
+   *  A tensor of indices into ref.
+   * @param updates A Tensor. Must have the same type as ref. A tensor of
+   *  values to add to ref.
+   * @param options carries optional attributes values
+   * @return a new instance of ResourceScatterNdAdd
+   */
+  public <T extends TNumber, U extends TType> ResourceScatterNdAdd resourceScatterNdAdd(
+      Operand<?> ref, Operand<T> indices, Operand<U> updates,
+      ResourceScatterNdAdd.Options... options) {
+    return ResourceScatterNdAdd.create(scope, ref, indices, updates, options);
+  }
+
+  /**
+   *
+   * @param ref A resource handle. Must be from a VarHandleOp.
+   * @param indices A Tensor. Must be one of the following types: int32, int64.
+   *  A tensor of indices into ref.
+   * @param updates A Tensor. Must have the same type as ref. A tensor of
+   *  values whose element wise max is taken with ref
+   * @param options carries optional attributes values
+   * @return a new instance of ResourceScatterNdMax
+   */
+  public <T extends TNumber, U extends TType> ResourceScatterNdMax resourceScatterNdMax(
+      Operand<?> ref, Operand<T> indices, Operand<U> updates,
+      ResourceScatterNdMax.Options... options) {
+    return ResourceScatterNdMax.create(scope, ref, indices, updates, options);
+  }
+
+  /**
+   *
+   * @param ref A resource handle. Must be from a VarHandleOp.
+   * @param indices A Tensor. Must be one of the following types: int32, int64.
+   *  A tensor of indices into ref.
+   * @param updates A Tensor. Must have the same type as ref. A tensor of
+   *  values whose element wise min is taken with ref.
+   * @param options carries optional attributes values
+   * @return a new instance of ResourceScatterNdMin
+   */
+  public <T extends TNumber, U extends TType> ResourceScatterNdMin resourceScatterNdMin(
+      Operand<?> ref, Operand<T> indices, Operand<U> updates,
+      ResourceScatterNdMin.Options... options) {
+    return ResourceScatterNdMin.create(scope, ref, indices, updates, options);
+  }
+
+  /**
+   * Applies sparse subtraction to individual values or slices in a Variable.
+   *  <p>
+   *  `ref` is a `Tensor` with rank `P` and `indices` is a `Tensor` of rank `Q`.
+   *  <p>
+   *  `indices` must be integer tensor, containing indices into `ref`.
+   *  It must be shape `[d_0, ..., d_{Q-2}, K]` where `0 < K <= P`.
+   *  <p>
+   *  The innermost dimension of `indices` (with length `K`) corresponds to
+   *  indices into elements (if `K = P`) or slices (if `K < P`) along the `K`th
+   *  dimension of `ref`.
+   *  <p>
+   *  `updates` is `Tensor` of rank `Q-1+P-K` with shape:
+   *  <pre>{@code
+   *  [d_0, ..., d_{Q-2}, ref.shape[K], ..., ref.shape[P-1]]
+   *  }</pre>
+   *  For example, say we want to subtract 4 scattered elements from a rank-1 tensor
+   *  with 8 elements. In Python, that subtraction would look like this:
+   *  <pre>{@code
+   *  ref = tf.Variable([1, 2, 3, 4, 5, 6, 7, 8], use_resource=True)
+   *  indices = tf.constant([[4], [3], [1], [7]])
+   *  updates = tf.constant([9, 10, 11, 12])
+   *  sub = tf.scatter_nd_sub(ref, indices, updates)
+   *  with tf.Session() as sess:
+   *    print sess.run(sub)
+   *  }</pre>
+   *  The resulting update to ref would look like this:
+   *  <p>
+   *      [1, -9, 3, -6, -4, 6, 7, -4]
+   *  <p>
+   *  See `tf.scatter_nd` for more details about how to make updates to
+   *  slices.
+   *
+   * @param ref A resource handle. Must be from a VarHandleOp.
+   * @param indices A Tensor. Must be one of the following types: int32, int64.
+   *  A tensor of indices into ref.
+   * @param updates A Tensor. Must have the same type as ref. A tensor of
+   *  values to add to ref.
+   * @param options carries optional attributes values
+   * @return a new instance of ResourceScatterNdSub
+   */
+  public <T extends TNumber, U extends TType> ResourceScatterNdSub resourceScatterNdSub(
+      Operand<?> ref, Operand<T> indices, Operand<U> updates,
+      ResourceScatterNdSub.Options... options) {
+    return ResourceScatterNdSub.create(scope, ref, indices, updates, options);
+  }
+
+  /**
+   * Applies sparse `updates` to individual values or slices within a given
+   *  <p>
+   *  variable according to `indices`.
+   *  <p>
+   *  `ref` is a `Tensor` with rank `P` and `indices` is a `Tensor` of rank `Q`.
+   *  <p>
+   *  `indices` must be integer tensor, containing indices into `ref`.
+   *  It must be shape `[d_0, ..., d_{Q-2}, K]` where `0 < K <= P`.
+   *  <p>
+   *  The innermost dimension of `indices` (with length `K`) corresponds to
+   *  indices into elements (if `K = P`) or slices (if `K < P`) along the `K`th
+   *  dimension of `ref`.
+   *  <p>
+   *  `updates` is `Tensor` of rank `Q-1+P-K` with shape:
+   *  <pre>{@code
+   *  [d_0, ..., d_{Q-2}, ref.shape[K], ..., ref.shape[P-1]].
+   *  }</pre>
+   *  For example, say we want to update 4 scattered elements to a rank-1 tensor to
+   *  8 elements. In Python, that update would look like this:
+   *  <pre>{@code
+   *      ref = tf.Variable([1, 2, 3, 4, 5, 6, 7, 8])
+   *      indices = tf.constant([[4], [3], [1] ,[7]])
+   *      updates = tf.constant([9, 10, 11, 12])
+   *      update = tf.scatter_nd_update(ref, indices, updates)
+   *      with tf.Session() as sess:
+   *        print sess.run(update)
+   *  }</pre>
+   *  The resulting update to ref would look like this:
+   *  <p>
+   *      [1, 11, 3, 10, 9, 6, 7, 12]
+   *  <p>
+   *  See `tf.scatter_nd` for more details about how to make updates to
+   *  slices.
+   *
+   * @param ref A resource handle. Must be from a VarHandleOp.
+   * @param indices A Tensor. Must be one of the following types: int32, int64.
+   *  A tensor of indices into ref.
+   * @param updates A Tensor. Must have the same type as ref. A tensor of updated
+   *  values to add to ref.
+   * @param options carries optional attributes values
+   * @return a new instance of ResourceScatterNdUpdate
+   */
+  public <T extends TNumber, U extends TType> ResourceScatterNdUpdate resourceScatterNdUpdate(
+      Operand<?> ref, Operand<T> indices, Operand<U> updates,
+      ResourceScatterNdUpdate.Options... options) {
+    return ResourceScatterNdUpdate.create(scope, ref, indices, updates, options);
+  }
+
+  /**
+   * Subtracts sparse updates from the variable referenced by `resource`.
+   *  <p>
+   *  This operation computes
+   *  <p>
+   *      # Scalar indices
+   *      ref[indices, ...] -= updates[...]
+   *  <p>
+   *      # Vector indices (for each i)
+   *      ref[indices[i], ...] -= updates[i, ...]
+   *  <p>
+   *      # High rank indices (for each i, ..., j)
+   *      ref[indices[i, ..., j], ...] -= updates[i, ..., j, ...]
+   *  <p>
+   *  Duplicate entries are handled correctly: if multiple `indices` reference
+   *  the same location, their contributions add.
+   *  <p>
+   *  Requires `updates.shape = indices.shape + ref.shape[1:]` or `updates.shape = []`.
+   *  <p>
+   *  <div style="width:70%; margin:auto; margin-bottom:10px; margin-top:20px;">
+   *  <img style="width:100%" src='https://www.tensorflow.org/images/ScatterAdd.png' alt>
+   *  </div>
+   *
+   * @param resource Should be from a `Variable` node.
+   * @param indices A tensor of indices into the first dimension of `ref`.
+   * @param updates A tensor of updated values to add to `ref`.
+   * @return a new instance of ResourceScatterSub
+   */
+  public <T extends TNumber, U extends TType> ResourceScatterSub resourceScatterSub(
+      Operand<?> resource, Operand<T> indices, Operand<U> updates) {
+    return ResourceScatterSub.create(scope, resource, indices, updates);
+  }
+
+  /**
+   * Assigns sparse updates to the variable referenced by `resource`.
+   *  <p>
+   *  This operation computes
+   *  <p>
+   *      # Scalar indices
+   *      ref[indices, ...] = updates[...]
+   *  <p>
+   *      # Vector indices (for each i)
+   *      ref[indices[i], ...] = updates[i, ...]
+   *  <p>
+   *      # High rank indices (for each i, ..., j)
+   *      ref[indices[i, ..., j], ...] = updates[i, ..., j, ...]
+   *
+   * @param resource Should be from a `Variable` node.
+   * @param indices A tensor of indices into the first dimension of `ref`.
+   * @param updates A tensor of updated values to add to `ref`.
+   * @return a new instance of ResourceScatterUpdate
+   */
+  public <T extends TNumber, U extends TType> ResourceScatterUpdate resourceScatterUpdate(
+      Operand<?> resource, Operand<T> indices, Operand<U> updates) {
+    return ResourceScatterUpdate.create(scope, resource, indices, updates);
+  }
+
+  /**
+   * Assign `value` to the sliced l-value reference of `ref`.
+   *  <p>
+   *  The values of `value` are assigned to the positions in the variable
+   *  `ref` that are selected by the slice parameters. The slice parameters
+   *  `begin, `end`, `strides`, etc. work exactly as in `StridedSlice`.
+   *  <p>
+   *  NOTE this op currently does not support broadcasting and so `value`'s
+   *  shape must be exactly the shape produced by the slice of `ref`.
+   *
+   * @param ref
+   * @param begin
+   * @param end
+   * @param strides
+   * @param value
+   * @param options carries optional attributes values
+   * @return a new instance of ResourceStridedSliceAssign
+   */
+  public <T extends TNumber, U extends TType> ResourceStridedSliceAssign resourceStridedSliceAssign(
+      Operand<?> ref, Operand<T> begin, Operand<T> end, Operand<T> strides, Operand<U> value,
+      ResourceStridedSliceAssign.Options... options) {
+    return ResourceStridedSliceAssign.create(scope, ref, begin, end, strides, value, options);
+  }
+
+  /**
+   * Reverses specific dimensions of a tensor.
+   *  <p>
+   *  NOTE `tf.reverse` has now changed behavior in preparation for 1.0.
+   *  `tf.reverse_v2` is currently an alias that will be deprecated before TF 1.0.
+   *  <p>
+   *  Given a `tensor`, and a `int32` tensor `axis` representing the set of
+   *  dimensions of `tensor` to reverse. This operation reverses each dimension
+   *  `i` for which there exists `j` s.t. `axis[j] == i`.
+   *  <p>
+   *  `tensor` can have up to 8 dimensions. The number of dimensions specified
+   *  in `axis` may be 0 or more entries. If an index is specified more than
+   *  once, a InvalidArgument error is raised.
+   *  <p>
+   *  For example:
+   *  <pre>{@code
+   *  # tensor 't' is [[[[ 0,  1,  2,  3],
+   *  #                  [ 4,  5,  6,  7],
+   *  #                  [ 8,  9, 10, 11]],
+   *  #                 [[12, 13, 14, 15],
+   *  #                  [16, 17, 18, 19],
+   *  #                  [20, 21, 22, 23]]]]
+   *  # tensor 't' shape is [1, 2, 3, 4]
+   *
+   *  # 'dims' is [3] or 'dims' is [-1]
+   *  reverse(t, dims) ==> [[[[ 3,  2,  1,  0],
+   *                          [ 7,  6,  5,  4],
+   *                          [ 11, 10, 9, 8]],
+   *                         [[15, 14, 13, 12],
+   *                          [19, 18, 17, 16],
+   *                          [23, 22, 21, 20]]]]
+   *
+   *  # 'dims' is '[1]' (or 'dims' is '[-3]')
+   *  reverse(t, dims) ==> [[[[12, 13, 14, 15],
+   *                          [16, 17, 18, 19],
+   *                          [20, 21, 22, 23]
+   *                         [[ 0,  1,  2,  3],
+   *                          [ 4,  5,  6,  7],
+   *                          [ 8,  9, 10, 11]]]]
+   *
+   *  # 'dims' is '[2]' (or 'dims' is '[-2]')
+   *  reverse(t, dims) ==> [[[[8, 9, 10, 11],
+   *                          [4, 5, 6, 7],
+   *                          [0, 1, 2, 3]]
+   *                         [[20, 21, 22, 23],
+   *                          [16, 17, 18, 19],
+   *                          [12, 13, 14, 15]]]]
+   *  }</pre>
+   *
+   * @param <T> data type for {@code output()} output
+   * @param tensor Up to 8-D.
+   * @param axis 1-D. The indices of the dimensions to reverse. Must be in the range
+   *  `[-rank(tensor), rank(tensor))`.
+   * @return a new instance of Reverse
+   */
+  public <T extends TType, U extends TNumber> Reverse<T> reverse(Operand<T> tensor,
+      Operand<U> axis) {
+    return Reverse.create(scope, tensor, axis);
+  }
+
+  /**
+   * Reverses variable length slices.
+   *  <p>
+   *  This op first slices `input` along the dimension `batch_dim`, and for each
+   *  slice `i`, reverses the first `seq_lengths[i]` elements along
+   *  the dimension `seq_dim`.
+   *  <p>
+   *  The elements of `seq_lengths` must obey `seq_lengths[i] <= input.dims[seq_dim]`,
+   *  and `seq_lengths` must be a vector of length `input.dims[batch_dim]`.
+   *  <p>
+   *  The output slice `i` along dimension `batch_dim` is then given by input
+   *  slice `i`, with the first `seq_lengths[i]` slices along dimension
+   *  `seq_dim` reversed.
+   *  <p>
+   *  For example:
+   *  <pre>{@code
+   *  # Given this:
+   *  batch_dim = 0
+   *  seq_dim = 1
+   *  input.dims = (4, 8, ...)
+   *  seq_lengths = [7, 2, 3, 5]
+   *
+   *  # then slices of input are reversed on seq_dim, but only up to seq_lengths:
+   *  output[0, 0:7, :, ...] = input[0, 7:0:-1, :, ...]
+   *  output[1, 0:2, :, ...] = input[1, 2:0:-1, :, ...]
+   *  output[2, 0:3, :, ...] = input[2, 3:0:-1, :, ...]
+   *  output[3, 0:5, :, ...] = input[3, 5:0:-1, :, ...]
+   *
+   *  # while entries past seq_lens are copied through:
+   *  output[0, 7:, :, ...] = input[0, 7:, :, ...]
+   *  output[1, 2:, :, ...] = input[1, 2:, :, ...]
+   *  output[2, 3:, :, ...] = input[2, 3:, :, ...]
+   *  output[3, 2:, :, ...] = input[3, 2:, :, ...]
+   *  }</pre>
+   *  In contrast, if:
+   *  <pre>{@code
+   *  # Given this:
+   *  batch_dim = 2
+   *  seq_dim = 0
+   *  input.dims = (8, ?, 4, ...)
+   *  seq_lengths = [7, 2, 3, 5]
+   *
+   *  # then slices of input are reversed on seq_dim, but only up to seq_lengths:
+   *  output[0:7, :, 0, :, ...] = input[7:0:-1, :, 0, :, ...]
+   *  output[0:2, :, 1, :, ...] = input[2:0:-1, :, 1, :, ...]
+   *  output[0:3, :, 2, :, ...] = input[3:0:-1, :, 2, :, ...]
+   *  output[0:5, :, 3, :, ...] = input[5:0:-1, :, 3, :, ...]
+   *
+   *  # while entries past seq_lens are copied through:
+   *  output[7:, :, 0, :, ...] = input[7:, :, 0, :, ...]
+   *  output[2:, :, 1, :, ...] = input[2:, :, 1, :, ...]
+   *  output[3:, :, 2, :, ...] = input[3:, :, 2, :, ...]
+   *  output[2:, :, 3, :, ...] = input[2:, :, 3, :, ...]
+   *  }</pre>
+   *
+   * @param <T> data type for {@code output()} output
+   * @param input The input to reverse.
+   * @param seqLengths 1-D with length `input.dims(batch_dim)` and
+   *  `max(seq_lengths) <= input.dims(seq_dim)`
+   * @param seqDim The dimension which is partially reversed.
+   * @param options carries optional attributes values
+   * @return a new instance of ReverseSequence
+   */
+  public <T extends TType, U extends TNumber> ReverseSequence<T> reverseSequence(Operand<T> input,
+      Operand<U> seqLengths, Long seqDim, ReverseSequence.Options... options) {
+    return ReverseSequence.create(scope, input, seqLengths, seqDim, options);
+  }
+
+  /**
+   * Rolls the elements of a tensor along an axis.
+   *  <p>
+   *  The elements are shifted positively (towards larger indices) by the offset of
+   *  `shift` along the dimension of `axis`. Negative `shift` values will shift
+   *  elements in the opposite direction. Elements that roll passed the last position
+   *  will wrap around to the first and vice versa. Multiple shifts along multiple
+   *  axes may be specified.
+   *  <p>
+   *  For example:
+   *  <pre>{@code
+   *  # 't' is [0, 1, 2, 3, 4]
+   *  roll(t, shift=2, axis=0) ==> [3, 4, 0, 1, 2]
+   *
+   *  # shifting along multiple dimensions
+   *  # 't' is [[0, 1, 2, 3, 4], [5, 6, 7, 8, 9]]
+   *  roll(t, shift=[1, -2], axis=[0, 1]) ==> [[7, 8, 9, 5, 6], [2, 3, 4, 0, 1]]
+   *
+   *  # shifting along the same axis multiple times
+   *  # 't' is [[0, 1, 2, 3, 4], [5, 6, 7, 8, 9]]
+   *  roll(t, shift=[2, -3], axis=[1, 1]) ==> [[1, 2, 3, 4, 0], [6, 7, 8, 9, 5]]
+   *  }</pre>
+   *
+   * @param <T> data type for {@code output()} output
+   * @param input
+   * @param shift Dimension must be 0-D or 1-D. `shift[i]` specifies the number of places by which
+   *  elements are shifted positively (towards larger indices) along the dimension
+   *  specified by `axis[i]`. Negative shifts will roll the elements in the opposite
+   *  direction.
+   * @param axis Dimension must be 0-D or 1-D. `axis[i]` specifies the dimension that the shift
+   *  `shift[i]` should occur. If the same axis is referenced more than once, the
+   *  total shift for that axis will be the sum of all the shifts that belong to that
+   *  axis.
+   * @return a new instance of Roll
+   */
+  public <T extends TType, U extends TNumber, V extends TNumber> Roll<T> roll(Operand<T> input,
+      Operand<U> shift, Operand<V> axis) {
+    return Roll.create(scope, input, shift, axis);
+  }
+
+  /**
+   * Perform batches of RPC requests.
+   *  <p>
+   *  This op asynchronously performs either a single RPC request, or a batch
+   *  of requests.  RPC requests are defined by three main parameters:
+   *  <p>
+   *    - `address` (the host+port or BNS address of the request)
+   *    - `method` (the RPC method name for the request)
+   *    - `request` (the serialized proto string, or vector of strings,
+   *       of the RPC request argument).
+   *  <p>
+   *  For example, if you have an RPC service running on port localhost:2345,
+   *  and its interface is configured with the following proto declaration:
+   *  <pre>{@code
+   *  service MyService {
+   *    rpc MyMethod(MyRequestProto) returns (MyResponseProto) {
+   *    }
+   *  };
+   *  }</pre>
+   *  then call this op with arguments:
+   *  <pre>{@code
+   *  address = "localhost:2345"
+   *  method = "MyService/MyMethod"
+   *  }</pre>
+   *  The `request` tensor is a string tensor representing serialized `MyRequestProto`
+   *  strings; and the output string tensor `response` will have the same shape
+   *  and contain (upon successful completion) corresponding serialized
+   *  `MyResponseProto` strings.
+   *  <p>
+   *  For example, to send a single, empty, `MyRequestProto`, call
+   *  this op with `request = ""`.  To send 5 <b>parallel</b> empty requests,
+   *  call this op with `request = ["", "", "", "", ""]`.
+   *  <p>
+   *  More generally, one can create a batch of `MyRequestProto` serialized protos
+   *  from regular batched tensors using the `encode_proto` op, and convert
+   *  the response `MyResponseProto` serialized protos to batched tensors
+   *  using the `decode_proto` op.
+   *  <p>
+   *  <b>NOTE</b> Working with serialized proto strings is faster than instantiating
+   *  actual proto objects in memory, so no performance degradation is expected
+   *  compared to writing custom kernels for this workflow.
+   *  <p>
+   *  If the connection fails or the remote worker returns an error
+   *  status, the op reraises this exception locally.
+   *  <p>
+   *  See the `TryRpc` op if you prefer to handle RPC failures manually in the graph.
+   *
+   * @param address `0-D` or `1-D`.  The address (i.e. host_name:port) of the RPC server.
+   *  If this tensor has more than 1 element, then multiple parallel rpc requests
+   *  are sent.  This argument broadcasts with `method` and `request`.
+   * @param method `0-D` or `1-D`.  The method address on the RPC server.
+   *  If this tensor has more than 1 element, then multiple parallel rpc requests
+   *  are sent.  This argument broadcasts with `address` and `request`.
+   * @param request `0-D` or `1-D`.  Serialized proto strings: the rpc request argument.
+   *  If this tensor has more than 1 element, then multiple parallel rpc requests
+   *  are sent.  This argument broadcasts with `address` and `method`.
+   * @param options carries optional attributes values
+   * @return a new instance of Rpc
+   */
+  public Rpc rpc(Operand<TString> address, Operand<TString> method, Operand<TString> request,
+      Rpc.Options... options) {
+    return Rpc.create(scope, address, method, request, options);
+  }
+
+  /**
+   * Adds sparse updates to a variable reference.
+   *  <p>
+   *  This operation computes
+   *  <p>
+   *      # Scalar indices
+   *      ref[indices, ...] += updates[...]
+   *  <p>
+   *      # Vector indices (for each i)
+   *      ref[indices[i], ...] += updates[i, ...]
+   *  <p>
+   *      # High rank indices (for each i, ..., j)
+   *      ref[indices[i, ..., j], ...] += updates[i, ..., j, ...]
+   *  <p>
+   *  This operation outputs `ref` after the update is done.
+   *  This makes it easier to chain operations that need to use the reset value.
+   *  <p>
+   *  Duplicate entries are handled correctly: if multiple `indices` reference
+   *  the same location, their contributions add.
+   *  <p>
+   *  Requires `updates.shape = indices.shape + ref.shape[1:]` or `updates.shape = []`.
+   *  <p>
+   *  <div style="width:70%; margin:auto; margin-bottom:10px; margin-top:20px;">
+   *  <img style="width:100%" src="https://www.tensorflow.org/images/ScatterAdd.png" alt>
+   *  </div>
+   *
+   * @param <T> data type for {@code outputRef()} output
+   * @param ref Should be from a `Variable` node.
+   * @param indices A tensor of indices into the first dimension of `ref`.
+   * @param updates A tensor of updated values to add to `ref`.
+   * @param options carries optional attributes values
+   * @return a new instance of ScatterAdd
+   */
+  public <T extends TType, U extends TNumber> ScatterAdd<T> scatterAdd(Operand<T> ref,
+      Operand<U> indices, Operand<T> updates, ScatterAdd.Options... options) {
+    return ScatterAdd.create(scope, ref, indices, updates, options);
+  }
+
+  /**
+   * Divides a variable reference by sparse updates.
+   *  <p>
+   *  This operation computes
+   *  <pre>{@code
+   *      # Scalar indices
+   *      ref[indices, ...] /= updates[...]
+   *
+   *      # Vector indices (for each i)
+   *      ref[indices[i], ...] /= updates[i, ...]
+   *
+   *      # High rank indices (for each i, ..., j)
+   *      ref[indices[i, ..., j], ...] /= updates[i, ..., j, ...]
+   *  }</pre>
+   *  This operation outputs `ref` after the update is done.
+   *  This makes it easier to chain operations that need to use the reset value.
+   *  <p>
+   *  Duplicate entries are handled correctly: if multiple `indices` reference
+   *  the same location, their contributions divide.
+   *  <p>
+   *  Requires `updates.shape = indices.shape + ref.shape[1:]` or `updates.shape = []`.
+   *
+   * @param <T> data type for {@code outputRef()} output
+   * @param ref Should be from a `Variable` node.
+   * @param indices A tensor of indices into the first dimension of `ref`.
+   * @param updates A tensor of values that `ref` is divided by.
+   * @param options carries optional attributes values
+   * @return a new instance of ScatterDiv
+   */
+  public <T extends TType, U extends TNumber> ScatterDiv<T> scatterDiv(Operand<T> ref,
+      Operand<U> indices, Operand<T> updates, ScatterDiv.Options... options) {
+    return ScatterDiv.create(scope, ref, indices, updates, options);
+  }
+
+  /**
+   * Reduces sparse updates into a variable reference using the `max` operation.
+   *  <p>
+   *  This operation computes
+   *  <p>
+   *      # Scalar indices
+   *      ref[indices, ...] = max(ref[indices, ...], updates[...])
+   *  <p>
+   *      # Vector indices (for each i)
+   *      ref[indices[i], ...] = max(ref[indices[i], ...], updates[i, ...])
+   *  <p>
+   *      # High rank indices (for each i, ..., j)
+   *      ref[indices[i, ..., j], ...] = max(ref[indices[i, ..., j], ...], updates[i, ..., j, ...])
+   *  <p>
+   *  This operation outputs `ref` after the update is done.
+   *  This makes it easier to chain operations that need to use the reset value.
+   *  <p>
+   *  Duplicate entries are handled correctly: if multiple `indices` reference
+   *  the same location, their contributions combine.
+   *  <p>
+   *  Requires `updates.shape = indices.shape + ref.shape[1:]` or `updates.shape = []`.
+   *  <p>
+   *  <div style="width:70%; margin:auto; margin-bottom:10px; margin-top:20px;">
+   *  <img style="width:100%" src="https://www.tensorflow.org/images/ScatterAdd.png" alt>
+   *  </div>
+   *
+   * @param <T> data type for {@code outputRef()} output
+   * @param ref Should be from a `Variable` node.
+   * @param indices A tensor of indices into the first dimension of `ref`.
+   * @param updates A tensor of updated values to reduce into `ref`.
+   * @param options carries optional attributes values
+   * @return a new instance of ScatterMax
+   */
+  public <T extends TNumber, U extends TNumber> ScatterMax<T> scatterMax(Operand<T> ref,
+      Operand<U> indices, Operand<T> updates, ScatterMax.Options... options) {
+    return ScatterMax.create(scope, ref, indices, updates, options);
+  }
+
+  /**
+   * Reduces sparse updates into a variable reference using the `min` operation.
+   *  <p>
+   *  This operation computes
+   *  <p>
+   *      # Scalar indices
+   *      ref[indices, ...] = min(ref[indices, ...], updates[...])
+   *  <p>
+   *      # Vector indices (for each i)
+   *      ref[indices[i], ...] = min(ref[indices[i], ...], updates[i, ...])
+   *  <p>
+   *      # High rank indices (for each i, ..., j)
+   *      ref[indices[i, ..., j], ...] = min(ref[indices[i, ..., j], ...], updates[i, ..., j, ...])
+   *  <p>
+   *  This operation outputs `ref` after the update is done.
+   *  This makes it easier to chain operations that need to use the reset value.
+   *  <p>
+   *  Duplicate entries are handled correctly: if multiple `indices` reference
+   *  the same location, their contributions combine.
+   *  <p>
+   *  Requires `updates.shape = indices.shape + ref.shape[1:]` or `updates.shape = []`.
+   *  <p>
+   *  <div style="width:70%; margin:auto; margin-bottom:10px; margin-top:20px;">
+   *  <img style="width:100%" src="https://www.tensorflow.org/images/ScatterAdd.png" alt>
+   *  </div>
+   *
+   * @param <T> data type for {@code outputRef()} output
+   * @param ref Should be from a `Variable` node.
+   * @param indices A tensor of indices into the first dimension of `ref`.
+   * @param updates A tensor of updated values to reduce into `ref`.
+   * @param options carries optional attributes values
+   * @return a new instance of ScatterMin
+   */
+  public <T extends TNumber, U extends TNumber> ScatterMin<T> scatterMin(Operand<T> ref,
+      Operand<U> indices, Operand<T> updates, ScatterMin.Options... options) {
+    return ScatterMin.create(scope, ref, indices, updates, options);
+  }
+
+  /**
+   * Multiplies sparse updates into a variable reference.
+   *  <p>
+   *  This operation computes
+   *  <pre>{@code
+   *      # Scalar indices
+   *      ref[indices, ...] *= updates[...]
+   *
+   *      # Vector indices (for each i)
+   *      ref[indices[i], ...] *= updates[i, ...]
+   *
+   *      # High rank indices (for each i, ..., j)
+   *      ref[indices[i, ..., j], ...] *= updates[i, ..., j, ...]
+   *  }</pre>
+   *  This operation outputs `ref` after the update is done.
+   *  This makes it easier to chain operations that need to use the reset value.
+   *  <p>
+   *  Duplicate entries are handled correctly: if multiple `indices` reference
+   *  the same location, their contributions multiply.
+   *  <p>
+   *  Requires `updates.shape = indices.shape + ref.shape[1:]` or `updates.shape = []`.
+   *
+   * @param <T> data type for {@code outputRef()} output
+   * @param ref Should be from a `Variable` node.
+   * @param indices A tensor of indices into the first dimension of `ref`.
+   * @param updates A tensor of updated values to multiply to `ref`.
+   * @param options carries optional attributes values
+   * @return a new instance of ScatterMul
+   */
+  public <T extends TType, U extends TNumber> ScatterMul<T> scatterMul(Operand<T> ref,
+      Operand<U> indices, Operand<T> updates, ScatterMul.Options... options) {
+    return ScatterMul.create(scope, ref, indices, updates, options);
+  }
+
+  /**
+   * Scatter `updates` into a new tensor according to `indices`.
+   *  <p>
+   *  Creates a new tensor by applying sparse `updates` to individual values or
+   *  slices within a tensor (initially zero for numeric, empty for string) of
+   *  the given `shape` according to indices.  This operator is the inverse of the
+   *  `tf.gather_nd` operator which extracts values or slices from a given tensor.
+   *  <p>
+   *  This operation is similar to tensor_scatter_add, except that the tensor is
+   *  zero-initialized. Calling `tf.scatter_nd(indices, values, shape)` is identical
+   *  to `tensor_scatter_add(tf.zeros(shape, values.dtype), indices, values)`
+   *  <p>
+   *  If `indices` contains duplicates, then their updates are accumulated (summed).
+   *  <p>
+   *  <b>WARNING</b>: The order in which updates are applied is nondeterministic, so the
+   *  output will be nondeterministic if `indices` contains duplicates -- because
+   *  of some numerical approximation issues, numbers summed in different order
+   *  may yield different results.
+   *  <p>
+   *  `indices` is an integer tensor containing indices into a new tensor of shape
+   *  `shape`.  The last dimension of `indices` can be at most the rank of `shape`:
+   *  <p>
+   *      indices.shape[-1] <= shape.rank
+   *  <p>
+   *  The last dimension of `indices` corresponds to indices into elements
+   *  (if `indices.shape[-1] = shape.rank`) or slices
+   *  (if `indices.shape[-1] < shape.rank`) along dimension `indices.shape[-1]` of
+   *  `shape`.  `updates` is a tensor with shape
+   *  <p>
+   *      indices.shape[:-1] + shape[indices.shape[-1]:]
+   *  <p>
+   *  The simplest form of scatter is to insert individual elements in a tensor by
+   *  index. For example, say we want to insert 4 scattered elements in a rank-1
+   *  tensor with 8 elements.
+   *  <p>
+   *  <div style="width:70%; margin:auto; margin-bottom:10px; margin-top:20px;">
+   *  <img style="width:100%" src="https://www.tensorflow.org/images/ScatterNd1.png" alt>
+   *  </div>
+   *  <p>
+   *  In Python, this scatter operation would look like this:
+   *  <pre>{@code
+   *      indices = tf.constant([[4], [3], [1], [7]])
+   *      updates = tf.constant([9, 10, 11, 12])
+   *      shape = tf.constant([8])
+   *      scatter = tf.scatter_nd(indices, updates, shape)
+   *      print(scatter)
+   *  }</pre>
+   *  The resulting tensor would look like this:
+   *  <p>
+   *      [0, 11, 0, 10, 9, 0, 0, 12]
+   *  <p>
+   *  We can also, insert entire slices of a higher rank tensor all at once. For
+   *  example, if we wanted to insert two slices in the first dimension of a
+   *  rank-3 tensor with two matrices of new values.
+   *  <p>
+   *  <div style="width:70%; margin:auto; margin-bottom:10px; margin-top:20px;">
+   *  <img style="width:100%" src="https://www.tensorflow.org/images/ScatterNd2.png" alt>
+   *  </div>
+   *  <p>
+   *  In Python, this scatter operation would look like this:
+   *  <pre>{@code
+   *      indices = tf.constant([[0], [2]])
+   *      updates = tf.constant([[[5, 5, 5, 5], [6, 6, 6, 6],
+   *                              [7, 7, 7, 7], [8, 8, 8, 8]],
+   *                             [[5, 5, 5, 5], [6, 6, 6, 6],
+   *                              [7, 7, 7, 7], [8, 8, 8, 8]]])
+   *      shape = tf.constant([4, 4, 4])
+   *      scatter = tf.scatter_nd(indices, updates, shape)
+   *      print(scatter)
+   *  }</pre>
+   *  The resulting tensor would look like this:
+   *  <p>
+   *      [[[5, 5, 5, 5], [6, 6, 6, 6], [7, 7, 7, 7], [8, 8, 8, 8]],
+   *       [[0, 0, 0, 0], [0, 0, 0, 0], [0, 0, 0, 0], [0, 0, 0, 0]],
+   *       [[5, 5, 5, 5], [6, 6, 6, 6], [7, 7, 7, 7], [8, 8, 8, 8]],
+   *       [[0, 0, 0, 0], [0, 0, 0, 0], [0, 0, 0, 0], [0, 0, 0, 0]]]
+   *  <p>
+   *  Note that on CPU, if an out of bound index is found, an error is returned.
+   *  On GPU, if an out of bound index is found, the index is ignored.
+   *
+   * @param <U> data type for {@code output()} output
+   * @param indices Index tensor.
+   * @param updates Updates to scatter into output.
+   * @param shape 1-D. The shape of the resulting tensor.
+   * @return a new instance of ScatterNd
+   */
+  public <U extends TType, T extends TNumber> ScatterNd<U> scatterNd(Operand<T> indices,
+      Operand<U> updates, Operand<T> shape) {
+    return ScatterNd.create(scope, indices, updates, shape);
+  }
+
+  /**
+   * Applies sparse addition to individual values or slices in a Variable.
+   *  <p>
+   *  `ref` is a `Tensor` with rank `P` and `indices` is a `Tensor` of rank `Q`.
+   *  <p>
+   *  `indices` must be integer tensor, containing indices into `ref`.
+   *  It must be shape `[d_0, ..., d_{Q-2}, K]` where `0 < K <= P`.
+   *  <p>
+   *  The innermost dimension of `indices` (with length `K`) corresponds to
+   *  indices into elements (if `K = P`) or slices (if `K < P`) along the `K`th
+   *  dimension of `ref`.
+   *  <p>
+   *  `updates` is `Tensor` of rank `Q-1+P-K` with shape:
+   *  <pre>{@code
+   *  [d_0, ..., d_{Q-2}, ref.shape[K], ..., ref.shape[P-1]]
+   *  }</pre>
+   *  For example, say we want to add 4 scattered elements to a rank-1 tensor to
+   *  8 elements. In Python, that addition would look like this:
+   *  <pre>{@code
+   *  ref = tf.Variable([1, 2, 3, 4, 5, 6, 7, 8])
+   *  indices = tf.constant([[4], [3], [1], [7]])
+   *  updates = tf.constant([9, 10, 11, 12])
+   *  add = tf.scatter_nd_add(ref, indices, updates)
+   *  with tf.Session() as sess:
+   *    print sess.run(add)
+   *  }</pre>
+   *  The resulting update to ref would look like this:
+   *  <p>
+   *      [1, 13, 3, 14, 14, 6, 7, 20]
+   *  <p>
+   *  See `tf.scatter_nd` for more details about how to make updates to
+   *  slices.
+   *
+   * @param <T> data type for {@code outputRef()} output
+   * @param ref A mutable Tensor. Should be from a Variable node.
+   * @param indices A Tensor. Must be one of the following types: int32, int64.
+   *  A tensor of indices into ref.
+   * @param updates A Tensor. Must have the same type as ref. A tensor of updated values
+   *  to add to ref.
+   * @param options carries optional attributes values
+   * @return a new instance of ScatterNdAdd
+   */
+  public <T extends TType, U extends TNumber> ScatterNdAdd<T> scatterNdAdd(Operand<T> ref,
+      Operand<U> indices, Operand<T> updates, ScatterNdAdd.Options... options) {
+    return ScatterNdAdd.create(scope, ref, indices, updates, options);
+  }
+
+  /**
+   * Applies sparse addition to `input` using individual values or slices
+   *  <p>
+   *  from `updates` according to indices `indices`.  The updates are non-aliasing:
+   *  `input` is only modified in-place if no other operations will use it.
+   *  Otherwise, a copy of `input` is made.  This operation has a gradient with
+   *  respect to both `input` and `updates`.
+   *  <p>
+   *  `input` is a `Tensor` with rank `P` and `indices` is a `Tensor` of rank `Q`.
+   *  <p>
+   *  `indices` must be integer tensor, containing indices into `input`.
+   *  It must be shape \\([d_0, ..., d_{Q-2}, K]\\) where `0 < K <= P`.
+   *  <p>
+   *  The innermost dimension of `indices` (with length `K`) corresponds to
+   *  indices into elements (if `K = P`) or `(P-K)`-dimensional slices
+   *  (if `K < P`) along the `K`th dimension of `input`.
+   *  <p>
+   *  `updates` is `Tensor` of rank `Q-1+P-K` with shape:
+   *  <p>
+   *  $$[d_0, ..., d_{Q-2}, input.shape[K], ..., input.shape[P-1]].$$
+   *  <p>
+   *  For example, say we want to add 4 scattered elements to a rank-1 tensor to 8
+   *  elements. In Python, that addition would look like this:
+   *  <p>
+   *      input = tf.constant([1, 2, 3, 4, 5, 6, 7, 8])
+   *      indices = tf.constant([[4], [3], [1], [7]])
+   *      updates = tf.constant([9, 10, 11, 12])
+   *      output = tf.scatter_nd_non_aliasing_add(input, indices, updates)
+   *      with tf.Session() as sess:
+   *        print(sess.run(output))
+   *  <p>
+   *  The resulting value `output` would look like this:
+   *  <p>
+   *      [1, 13, 3, 14, 14, 6, 7, 20]
+   *  <p>
+   *  See `tf.scatter_nd` for more details about how to make updates to slices.
+   *
+   * @param <T> data type for {@code output()} output
+   * @param input A Tensor.
+   * @param indices A Tensor. Must be one of the following types: `int32`, `int64`.
+   *  A tensor of indices into `input`.
+   * @param updates A Tensor. Must have the same type as ref. A tensor of updated values
+   *  to add to `input`.
+   * @return a new instance of ScatterNdNonAliasingAdd
+   */
+  public <T extends TType, U extends TNumber> ScatterNdNonAliasingAdd<T> scatterNdNonAliasingAdd(
+      Operand<T> input, Operand<U> indices, Operand<T> updates) {
+    return ScatterNdNonAliasingAdd.create(scope, input, indices, updates);
+  }
+
+  /**
+   * Applies sparse subtraction to individual values or slices in a Variable.
+   *  <p>
+   *  within a given variable according to `indices`.
+   *  <p>
+   *  `ref` is a `Tensor` with rank `P` and `indices` is a `Tensor` of rank `Q`.
+   *  <p>
+   *  `indices` must be integer tensor, containing indices into `ref`.
+   *  It must be shape `[d_0, ..., d_{Q-2}, K]` where `0 < K <= P`.
+   *  <p>
+   *  The innermost dimension of `indices` (with length `K`) corresponds to
+   *  indices into elements (if `K = P`) or slices (if `K < P`) along the `K`th
+   *  dimension of `ref`.
+   *  <p>
+   *  `updates` is `Tensor` of rank `Q-1+P-K` with shape:
+   *  <pre>{@code
+   *  [d_0, ..., d_{Q-2}, ref.shape[K], ..., ref.shape[P-1]]
+   *  }</pre>
+   *  For example, say we want to subtract 4 scattered elements from a rank-1 tensor
+   *  with 8 elements. In Python, that subtraction would look like this:
+   *  <pre>{@code
+   *  ref = tf.Variable([1, 2, 3, 4, 5, 6, 7, 8])
+   *  indices = tf.constant([[4], [3], [1], [7]])
+   *  updates = tf.constant([9, 10, 11, 12])
+   *  sub = tf.scatter_nd_sub(ref, indices, updates)
+   *  with tf.Session() as sess:
+   *    print sess.run(sub)
+   *  }</pre>
+   *  The resulting update to ref would look like this:
+   *  <p>
+   *      [1, -9, 3, -6, -4, 6, 7, -4]
+   *  <p>
+   *  See `tf.scatter_nd` for more details about how to make updates to
+   *  slices.
+   *
+   * @param <T> data type for {@code outputRef()} output
+   * @param ref A mutable Tensor. Should be from a Variable node.
+   * @param indices A Tensor. Must be one of the following types: int32, int64.
+   *  A tensor of indices into ref.
+   * @param updates A Tensor. Must have the same type as ref. A tensor of updated values
+   *  to subtract from ref.
+   * @param options carries optional attributes values
+   * @return a new instance of ScatterNdSub
+   */
+  public <T extends TType, U extends TNumber> ScatterNdSub<T> scatterNdSub(Operand<T> ref,
+      Operand<U> indices, Operand<T> updates, ScatterNdSub.Options... options) {
+    return ScatterNdSub.create(scope, ref, indices, updates, options);
+  }
+
+  /**
+   * Applies sparse `updates` to individual values or slices within a given
+   *  <p>
+   *  variable according to `indices`.
+   *  <p>
+   *  `ref` is a `Tensor` with rank `P` and `indices` is a `Tensor` of rank `Q`.
+   *  <p>
+   *  `indices` must be integer tensor, containing indices into `ref`.
+   *  It must be shape \\([d_0, ..., d_{Q-2}, K]\\) where `0 < K <= P`.
+   *  <p>
+   *  The innermost dimension of `indices` (with length `K`) corresponds to
+   *  indices into elements (if `K = P`) or slices (if `K < P`) along the `K`th
+   *  dimension of `ref`.
+   *  <p>
+   *  `updates` is `Tensor` of rank `Q-1+P-K` with shape:
+   *  <p>
+   *  $$[d_0, ..., d_{Q-2}, ref.shape[K], ..., ref.shape[P-1]].$$
+   *  <p>
+   *  For example, say we want to update 4 scattered elements to a rank-1 tensor to
+   *  8 elements. In Python, that update would look like this:
+   *  <pre>{@code
+   *      ref = tf.Variable([1, 2, 3, 4, 5, 6, 7, 8])
+   *      indices = tf.constant([[4], [3], [1] ,[7]])
+   *      updates = tf.constant([9, 10, 11, 12])
+   *      update = tf.scatter_nd_update(ref, indices, updates)
+   *      with tf.Session() as sess:
+   *        print sess.run(update)
+   *  }</pre>
+   *  The resulting update to ref would look like this:
+   *  <p>
+   *      [1, 11, 3, 10, 9, 6, 7, 12]
+   *  <p>
+   *  See `tf.scatter_nd` for more details about how to make updates to
+   *  slices.
+   *  <p>
+   *  See also `tf.scatter_update` and `tf.batch_scatter_update`.
+   *
+   * @param <T> data type for {@code outputRef()} output
+   * @param ref A mutable Tensor. Should be from a Variable node.
+   * @param indices A Tensor. Must be one of the following types: int32, int64.
+   *  A tensor of indices into ref.
+   * @param updates A Tensor. Must have the same type as ref. A tensor of updated
+   *  values to add to ref.
+   * @param options carries optional attributes values
+   * @return a new instance of ScatterNdUpdate
+   */
+  public <T extends TType, U extends TNumber> ScatterNdUpdate<T> scatterNdUpdate(Operand<T> ref,
+      Operand<U> indices, Operand<T> updates, ScatterNdUpdate.Options... options) {
+    return ScatterNdUpdate.create(scope, ref, indices, updates, options);
+  }
+
+  /**
+   * Subtracts sparse updates to a variable reference.
+   *  <p>
+   *  <pre>{@code
+   *      # Scalar indices
+   *      ref[indices, ...] -= updates[...]
+   *
+   *      # Vector indices (for each i)
+   *      ref[indices[i], ...] -= updates[i, ...]
+   *
+   *      # High rank indices (for each i, ..., j)
+   *      ref[indices[i, ..., j], ...] -= updates[i, ..., j, ...]
+   *  }</pre>
+   *  This operation outputs `ref` after the update is done.
+   *  This makes it easier to chain operations that need to use the reset value.
+   *  <p>
+   *  Duplicate entries are handled correctly: if multiple `indices` reference
+   *  the same location, their (negated) contributions add.
+   *  <p>
+   *  Requires `updates.shape = indices.shape + ref.shape[1:]` or `updates.shape = []`.
+   *  <p>
+   *  <div style="width:70%; margin:auto; margin-bottom:10px; margin-top:20px;">
+   *  <img style="width:100%" src="https://www.tensorflow.org/images/ScatterSub.png" alt>
+   *  </div>
+   *
+   * @param <T> data type for {@code outputRef()} output
+   * @param ref Should be from a `Variable` node.
+   * @param indices A tensor of indices into the first dimension of `ref`.
+   * @param updates A tensor of updated values to subtract from `ref`.
+   * @param options carries optional attributes values
+   * @return a new instance of ScatterSub
+   */
+  public <T extends TType, U extends TNumber> ScatterSub<T> scatterSub(Operand<T> ref,
+      Operand<U> indices, Operand<T> updates, ScatterSub.Options... options) {
+    return ScatterSub.create(scope, ref, indices, updates, options);
+  }
+
+  /**
+   * Applies sparse updates to a variable reference.
+   *  <p>
+   *  This operation computes
+   *  <pre>{@code
+   *      # Scalar indices
+   *      ref[indices, ...] = updates[...]
+   *
+   *      # Vector indices (for each i)
+   *      ref[indices[i], ...] = updates[i, ...]
+   *
+   *      # High rank indices (for each i, ..., j)
+   *      ref[indices[i, ..., j], ...] = updates[i, ..., j, ...]
+   *  }</pre>
+   *  This operation outputs `ref` after the update is done.
+   *  This makes it easier to chain operations that need to use the reset value.
+   *  <p>
+   *  If values in `ref` is to be updated more than once, because there are
+   *  duplicate entries in `indices`, the order at which the updates happen
+   *  for each value is undefined.
+   *  <p>
+   *  Requires `updates.shape = indices.shape + ref.shape[1:]` or `updates.shape = []`.
+   *  <p>
+   *  <div style="width:70%; margin:auto; margin-bottom:10px; margin-top:20px;">
+   *  <img style="width:100%" src="https://www.tensorflow.org/images/ScatterUpdate.png" alt>
+   *  </div>
+   *  <p>
+   *  See also `tf.batch_scatter_update` and `tf.scatter_nd_update`.
+   *
+   * @param <T> data type for {@code outputRef()} output
+   * @param ref Should be from a `Variable` node.
+   * @param indices A tensor of indices into the first dimension of `ref`.
+   * @param updates A tensor of updated values to store in `ref`.
+   * @param options carries optional attributes values
+   * @return a new instance of ScatterUpdate
+   */
+  public <T extends TType, U extends TNumber> ScatterUpdate<T> scatterUpdate(Operand<T> ref,
+      Operand<U> indices, Operand<T> updates, ScatterUpdate.Options... options) {
+    return ScatterUpdate.create(scope, ref, indices, updates, options);
+  }
+
+  /**
+   *
+   * @param <T> data type for {@code output()} output
+   * @param condition
+   * @param t
+   * @param e
+   * @return a new instance of Select
+   */
+  public <T extends TType> Select<T> select(Operand<TBool> condition, Operand<T> t, Operand<T> e) {
+    return Select.create(scope, condition, t, e);
+  }
+
+  /**
+   * Computes the difference between two lists of numbers or strings.
+   *  <p>
+   *  Given a list `x` and a list `y`, this operation returns a list `out` that
+   *  represents all values that are in `x` but not in `y`. The returned list `out`
+   *  is sorted in the same order that the numbers appear in `x` (duplicates are
+   *  preserved). This operation also returns a list `idx` that represents the
+   *  position of each `out` element in `x`. In other words:
+   *  <p>
+   *  `out[i] = x[idx[i]] for i in [0, 1, ..., len(out) - 1]`
+   *  <p>
+   *  For example, given this input:
+   *  <pre>{@code
+   *  x = [1, 2, 3, 4, 5, 6]
+   *  y = [1, 3, 5]
+   *  }</pre>
+   *  This operation would return:
+   *  <pre>{@code
+   *  out ==> [2, 4, 6]
+   *  idx ==> [1, 3, 5]
+   *  }</pre>
+   *
+   * @param <T> data type for {@code out()} output
+   * @param <U> data type for {@code idx()} output
+   * @param x 1-D. Values to keep.
+   * @param y 1-D. Values to remove.
+   * @return a new instance of SetDiff1d
+   */
+  public <T extends TType> SetDiff1d<T, TInt32> setDiff1d(Operand<T> x, Operand<T> y) {
+    return SetDiff1d.create(scope, x, y);
+  }
+
+  /**
+   * Computes the difference between two lists of numbers or strings.
+   *  <p>
+   *  Given a list `x` and a list `y`, this operation returns a list `out` that
+   *  represents all values that are in `x` but not in `y`. The returned list `out`
+   *  is sorted in the same order that the numbers appear in `x` (duplicates are
+   *  preserved). This operation also returns a list `idx` that represents the
+   *  position of each `out` element in `x`. In other words:
+   *  <p>
+   *  `out[i] = x[idx[i]] for i in [0, 1, ..., len(out) - 1]`
+   *  <p>
+   *  For example, given this input:
+   *  <pre>{@code
+   *  x = [1, 2, 3, 4, 5, 6]
+   *  y = [1, 3, 5]
+   *  }</pre>
+   *  This operation would return:
+   *  <pre>{@code
+   *  out ==> [2, 4, 6]
+   *  idx ==> [1, 3, 5]
+   *  }</pre>
+   *
+   * @param <T> data type for {@code out()} output
+   * @param <U> data type for {@code idx()} output
+   * @param x 1-D. Values to keep.
+   * @param y 1-D. Values to remove.
+   * @param outIdx
+   * @return a new instance of SetDiff1d
+   */
+  public <T extends TType, U extends TNumber> SetDiff1d<T, U> setDiff1d(Operand<T> x, Operand<T> y,
+      Class<U> outIdx) {
+    return SetDiff1d.create(scope, x, y, outIdx);
+  }
+
+  /**
+   * Number of unique elements along last dimension of input `set`.
+   *  <p>
+   *  Input `set` is a `SparseTensor` represented by `set_indices`, `set_values`,
+   *  and `set_shape`. The last dimension contains values in a set, duplicates are
+   *  allowed but ignored.
+   *  <p>
+   *  If `validate_indices` is `True`, this op validates the order and range of `set`
+   *  indices.
+   *
+   * @param setIndices 2D `Tensor`, indices of a `SparseTensor`.
+   * @param setValues 1D `Tensor`, values of a `SparseTensor`.
+   * @param setShape 1D `Tensor`, shape of a `SparseTensor`.
+   * @param options carries optional attributes values
+   * @return a new instance of SetSize
+   */
+  public <T extends TType> SetSize setSize(Operand<TInt64> setIndices, Operand<T> setValues,
+      Operand<TInt64> setShape, SetSize.Options... options) {
+    return SetSize.create(scope, setIndices, setValues, setShape, options);
+  }
+
+  /**
+   * Returns the shape of a tensor.
+   *  <p>
+   *  This operation returns a 1-D integer tensor representing the shape of `input`.
+   *  <p>
+   *  For example:
+   *  <pre>{@code
+   *  # 't' is [[[1, 1, 1], [2, 2, 2]], [[3, 3, 3], [4, 4, 4]]]
+   *  shape(t) ==> [2, 2, 3]
+   *  }</pre>
+   *
+   * @param <U> data type for {@code output()} output
+   * @param input
+   * @return a new instance of Shape
+   */
+  public <T extends TType> org.tensorflow.op.core.Shape<TInt32> shape(Operand<T> input) {
+    return org.tensorflow.op.core.Shape.create(scope, input);
+  }
+
+  /**
+   * Returns the shape of a tensor.
+   *  <p>
+   *  This operation returns a 1-D integer tensor representing the shape of `input`.
+   *  <p>
+   *  For example:
+   *  <pre>{@code
+   *  # 't' is [[[1, 1, 1], [2, 2, 2]], [[3, 3, 3], [4, 4, 4]]]
+   *  shape(t) ==> [2, 2, 3]
+   *  }</pre>
+   *
+   * @param <U> data type for {@code output()} output
+   * @param input
+   * @param outType
+   * @return a new instance of Shape
+   */
+  public <U extends TNumber, T extends TType> org.tensorflow.op.core.Shape<U> shape(
+      Operand<T> input, Class<U> outType) {
+    return org.tensorflow.op.core.Shape.create(scope, input, outType);
+  }
+
+  /**
+   * Returns shape of tensors.
+   *  <p>
+   *  This operation returns N 1-D integer tensors representing shape of `input[i]s`.
+   *
+   * @param <U> data type for {@code output()} output
+   * @param input
+   * @return a new instance of ShapeN
+   */
+  public <T extends TType> ShapeN<TInt32> shapeN(Iterable<Operand<T>> input) {
+    return ShapeN.create(scope, input);
+  }
+
+  /**
+   * Returns shape of tensors.
+   *  <p>
+   *  This operation returns N 1-D integer tensors representing shape of `input[i]s`.
+   *
+   * @param <U> data type for {@code output()} output
+   * @param input
+   * @param outType
+   * @return a new instance of ShapeN
+   */
+  public <U extends TNumber, T extends TType> ShapeN<U> shapeN(Iterable<Operand<T>> input,
+      Class<U> outType) {
+    return ShapeN.create(scope, input, outType);
+  }
+
+  /**
+   * Returns the size of a tensor.
+   *  <p>
+   *  This operation returns an integer representing the number of elements in
+   *  `input`.
+   *  <p>
+   *  For example:
+   *  <pre>{@code
+   *  # 't' is [[[1, 1,, 1], [2, 2, 2]], [[3, 3, 3], [4, 4, 4]]]]
+   *  size(t) ==> 12
+   *  }</pre>
+   *
+   * @param <U> data type for {@code output()} output
+   * @param input
+   * @return a new instance of Size
+   */
+  public <T extends TType> Size<TInt32> size(Operand<T> input) {
+    return Size.create(scope, input);
+  }
+
+  /**
+   * Returns the size of a tensor.
+   *  <p>
+   *  This operation returns an integer representing the number of elements in
+   *  `input`.
+   *  <p>
+   *  For example:
+   *  <pre>{@code
+   *  # 't' is [[[1, 1,, 1], [2, 2, 2]], [[3, 3, 3], [4, 4, 4]]]]
+   *  size(t) ==> 12
+   *  }</pre>
+   *
+   * @param <U> data type for {@code output()} output
+   * @param input
+   * @param outType
+   * @return a new instance of Size
+   */
+  public <U extends TNumber, T extends TType> Size<U> size(Operand<T> input, Class<U> outType) {
+    return Size.create(scope, input, outType);
+  }
+
+  /**
+   * Parses a text file and creates a batch of examples.
+   *
+   * @param filename The corpus's text file name.
+   * @param batchSize The size of produced batch.
+   * @param options carries optional attributes values
+   * @return a new instance of Skipgram
+   */
+  public Skipgram skipgram(String filename, Long batchSize, Skipgram.Options... options) {
+    return Skipgram.create(scope, filename, batchSize, options);
+  }
+
+  /**
+   * Return a slice from 'input'.
+   *  <p>
+   *  The output tensor is a tensor with dimensions described by 'size'
+   *  whose values are extracted from 'input' starting at the offsets in
+   *  'begin'.
+   *  <p>
+   *  <i>Requirements</i>:
+   *    0 <= begin[i] <= begin[i] + size[i] <= Di  for i in [0, n)
+   *
+   * @param <T> data type for {@code output()} output
+   * @param input
+   * @param begin begin[i] specifies the offset into the 'i'th dimension of
+   *  'input' to slice from.
+   * @param size size[i] specifies the number of elements of the 'i'th dimension
+   *  of 'input' to slice. If size[i] is -1, all remaining elements in dimension
+   *  i are included in the slice (i.e. this is equivalent to setting
+   *  size[i] = input.dim_size(i) - begin[i]).
+   * @return a new instance of Slice
+   */
+  public <T extends TType, U extends TNumber> Slice<T> slice(Operand<T> input, Operand<U> begin,
+      Operand<U> size) {
+    return Slice.create(scope, input, begin, size);
+  }
+
+  /**
+   * Returns a copy of the input tensor.
+   *
+   * @param <T> data type for {@code output()} output
+   * @param input
+   * @return a new instance of Snapshot
+   */
+  public <T extends TType> Snapshot<T> snapshot(Operand<T> input) {
+    return Snapshot.create(scope, input);
+  }
+
+  /**
+   * SpaceToBatch for N-D tensors of type T.
+   *  <p>
+   *  This operation divides "spatial" dimensions `[1, ..., M]` of the input into a
+   *  grid of blocks of shape `block_shape`, and interleaves these blocks with the
+   *  "batch" dimension (0) such that in the output, the spatial dimensions
+   *  `[1, ..., M]` correspond to the position within the grid, and the batch
+   *  dimension combines both the position within a spatial block and the original
+   *  batch position.  Prior to division into blocks, the spatial dimensions of the
+   *  input are optionally zero padded according to `paddings`.  See below for a
+   *  precise description.
+   *
+   * @param <T> data type for {@code output()} output
+   * @param input N-D with shape `input_shape = [batch] + spatial_shape + remaining_shape`,
+   *  where spatial_shape has `M` dimensions.
+   * @param blockShape 1-D with shape `[M]`, all values must be >= 1.
+   * @param paddings 2-D with shape `[M, 2]`, all values must be >= 0.
+   *    `paddings[i] = [pad_start, pad_end]` specifies the padding for input dimension
+   *    `i + 1`, which corresponds to spatial dimension `i`.  It is required that
+   *    `block_shape[i]` divides `input_shape[i + 1] + pad_start + pad_end`.
+   *  <p>
+   *  This operation is equivalent to the following steps:
+   *  <p>
+   *  1. Zero-pad the start and end of dimensions `[1, ..., M]` of the
+   *     input according to `paddings` to produce `padded` of shape `padded_shape`.
+   *  <p>
+   *  2. Reshape `padded` to `reshaped_padded` of shape:
+   *  <p>
+   *       [batch] +
+   *       [padded_shape[1] / block_shape[0],
+   *         block_shape[0],
+   *        ...,
+   *        padded_shape[M] / block_shape[M-1],
+   *        block_shape[M-1]] +
+   *       remaining_shape
+   *  <p>
+   *  3. Permute dimensions of `reshaped_padded` to produce
+   *     `permuted_reshaped_padded` of shape:
+   *  <p>
+   *       block_shape +
+   *       [batch] +
+   *       [padded_shape[1] / block_shape[0],
+   *        ...,
+   *        padded_shape[M] / block_shape[M-1]] +
+   *       remaining_shape
+   *  <p>
+   *  4. Reshape `permuted_reshaped_padded` to flatten `block_shape` into the batch
+   *     dimension, producing an output tensor of shape:
+   *  <p>
+   *       [batch * prod(block_shape)] +
+   *       [padded_shape[1] / block_shape[0],
+   *        ...,
+   *        padded_shape[M] / block_shape[M-1]] +
+   *       remaining_shape
+   *  <p>
+   *  Some examples:
+   *  <p>
+   *  (1) For the following input of shape `[1, 2, 2, 1]`, `block_shape = [2, 2]`, and
+   *      `paddings = [[0, 0], [0, 0]]`:
+   *  <pre>{@code
+   *  x = [[[[1], [2]], [[3], [4]]]]
+   *  }</pre>
+   *  The output tensor has shape `[4, 1, 1, 1]` and value:
+   *  <pre>{@code
+   *  [[[[1]]], [[[2]]], [[[3]]], [[[4]]]]
+   *  }</pre>
+   *  (2) For the following input of shape `[1, 2, 2, 3]`, `block_shape = [2, 2]`, and
+   *      `paddings = [[0, 0], [0, 0]]`:
+   *  <pre>{@code
+   *  x = [[[[1, 2, 3], [4, 5, 6]],
+   *        [[7, 8, 9], [10, 11, 12]]]]
+   *  }</pre>
+   *  The output tensor has shape `[4, 1, 1, 3]` and value:
+   *  <pre>{@code
+   *  [[[[1, 2, 3]]], [[[4, 5, 6]]], [[[7, 8, 9]]], [[[10, 11, 12]]]]
+   *  }</pre>
+   *  (3) For the following input of shape `[1, 4, 4, 1]`, `block_shape = [2, 2]`, and
+   *      `paddings = [[0, 0], [0, 0]]`:
+   *  <pre>{@code
+   *  x = [[[[1],   [2],  [3],  [4]],
+   *        [[5],   [6],  [7],  [8]],
+   *        [[9],  [10], [11],  [12]],
+   *        [[13], [14], [15],  [16]]]]
+   *  }</pre>
+   *  The output tensor has shape `[4, 2, 2, 1]` and value:
+   *  <pre>{@code
+   *  x = [[[[1], [3]], [[9], [11]]],
+   *       [[[2], [4]], [[10], [12]]],
+   *       [[[5], [7]], [[13], [15]]],
+   *       [[[6], [8]], [[14], [16]]]]
+   *  }</pre>
+   *  (4) For the following input of shape `[2, 2, 4, 1]`, block_shape = `[2, 2]`, and
+   *      paddings = `[[0, 0], [2, 0]]`:
+   *  <pre>{@code
+   *  x = [[[[1],   [2],  [3],  [4]],
+   *        [[5],   [6],  [7],  [8]]],
+   *       [[[9],  [10], [11],  [12]],
+   *        [[13], [14], [15],  [16]]]]
+   *  }</pre>
+   *  The output tensor has shape `[8, 1, 3, 1]` and value:
+   *  <pre>{@code
+   *  x = [[[[0], [1], [3]]], [[[0], [9], [11]]],
+   *       [[[0], [2], [4]]], [[[0], [10], [12]]],
+   *       [[[0], [5], [7]]], [[[0], [13], [15]]],
+   *       [[[0], [6], [8]]], [[[0], [14], [16]]]]
+   *  }</pre>
+   *  Among others, this operation is useful for reducing atrous convolution into
+   *  regular convolution.
+   * @return a new instance of SpaceToBatchNd
+   */
+  public <T extends TType, U extends TNumber, V extends TNumber> SpaceToBatchNd<T> spaceToBatchNd(
+      Operand<T> input, Operand<U> blockShape, Operand<V> paddings) {
+    return SpaceToBatchNd.create(scope, input, blockShape, paddings);
+  }
+
+  /**
+   * Splits a tensor into `num_split` tensors along one dimension.
+   *
+   * @param <T> data type for {@code output()} output
+   * @param axis 0-D.  The dimension along which to split.  Must be in the range
+   *  `[-rank(value), rank(value))`.
+   * @param value The tensor to split.
+   * @param numSplit The number of ways to split.  Must evenly divide
+   *  `value.shape[split_dim]`.
+   * @return a new instance of Split
+   */
+  public <T extends TType> Split<T> split(Operand<TInt32> axis, Operand<T> value, Long numSplit) {
+    return Split.create(scope, axis, value, numSplit);
+  }
+
+  /**
+   * Splits a tensor into `num_split` tensors along one dimension.
+   *
+   * @param <T> data type for {@code output()} output
+   * @param value The tensor to split.
+   * @param sizeSplits list containing the sizes of each output tensor along the split
+   *  dimension. Must sum to the dimension of value along split_dim.
+   *  Can contain one -1 indicating that dimension is to be inferred.
+   * @param axis 0-D.  The dimension along which to split.  Must be in the range
+   *  `[-rank(value), rank(value))`.
+   * @param numSplit
+   * @return a new instance of SplitV
+   */
+  public <T extends TType, U extends TNumber> SplitV<T> splitV(Operand<T> value,
+      Operand<U> sizeSplits, Operand<TInt32> axis, Long numSplit) {
+    return SplitV.create(scope, value, sizeSplits, axis, numSplit);
+  }
+
+  /**
+   * Removes dimensions of size 1 from the shape of a tensor.
+   *  <p>
+   *  Given a tensor `input`, this operation returns a tensor of the same type with
+   *  all dimensions of size 1 removed. If you don't want to remove all size 1
+   *  dimensions, you can remove specific size 1 dimensions by specifying
+   *  `axis`.
+   *  <p>
+   *  For example:
+   *  <pre>{@code
+   *  # 't' is a tensor of shape [1, 2, 1, 3, 1, 1]
+   *  shape(squeeze(t)) ==> [2, 3]
+   *  }</pre>
+   *  Or, to remove specific size 1 dimensions:
+   *  <pre>{@code
+   *  # 't' is a tensor of shape [1, 2, 1, 3, 1, 1]
+   *  shape(squeeze(t, [2, 4])) ==> [1, 2, 3, 1]
+   *  }</pre>
+   *
+   * @param <T> data type for {@code output()} output
+   * @param input The `input` to squeeze.
+   * @param options carries optional attributes values
+   * @return a new instance of Squeeze
+   */
+  public <T extends TType> Squeeze<T> squeeze(Operand<T> input, Squeeze.Options... options) {
+    return Squeeze.create(scope, input, options);
+  }
+
+  /**
+   * Packs a list of `N` rank-`R` tensors into one rank-`(R+1)` tensor.
+   *  <p>
+   *  Packs the `N` tensors in `values` into a tensor with rank one higher than each
+   *  tensor in `values`, by packing them along the `axis` dimension.
+   *  Given a list of tensors of shape `(A, B, C)`;
+   *  <p>
+   *  if `axis == 0` then the `output` tensor will have the shape `(N, A, B, C)`.
+   *  if `axis == 1` then the `output` tensor will have the shape `(A, N, B, C)`.
+   *  Etc.
+   *  <p>
+   *  For example:
+   *  <pre>{@code
+   *  # 'x' is [1, 4]
+   *  # 'y' is [2, 5]
+   *  # 'z' is [3, 6]
+   *  pack([x, y, z]) => [[1, 4], [2, 5], [3, 6]]  # Pack along first dim.
+   *  pack([x, y, z], axis=1) => [[1, 2, 3], [4, 5, 6]]
+   *  }</pre>
+   *  This is the opposite of `unpack`.
+   *
+   * @param <T> data type for {@code output()} output
+   * @param values Must be of same shape and type.
+   * @param options carries optional attributes values
+   * @return a new instance of Stack
+   */
+  public <T extends TType> Stack<T> stack(Iterable<Operand<T>> values, Stack.Options... options) {
+    return Stack.create(scope, values, options);
+  }
+
+  /**
+   * Stage values similar to a lightweight Enqueue.
+   *  <p>
+   *  The basic functionality of this Op is similar to a queue with many
+   *  fewer capabilities and options.  This Op is optimized for performance.
+   *
+   * @param values a list of tensors
+   *  dtypes A list of data types that inserted values should adhere to.
+   * @param options carries optional attributes values
+   * @return a new instance of Stage
+   */
+  public Stage stage(Iterable<Operand<?>> values, Stage.Options... options) {
+    return Stage.create(scope, values, options);
+  }
+
+  /**
+   * Op removes all elements in the underlying container.
+   *
+   * @param dtypes
+   * @param options carries optional attributes values
+   * @return a new instance of StageClear
+   */
+  public StageClear stageClear(List<Class<? extends TType>> dtypes, StageClear.Options... options) {
+    return StageClear.create(scope, dtypes, options);
+  }
+
+  /**
+   * Op peeks at the values at the specified index.  If the
+   *  <p>
+   *  underlying container does not contain sufficient elements
+   *  this op will block until it does.   This Op is optimized for
+   *  performance.
+   *
+   * @param index
+   * @param dtypes
+   * @param options carries optional attributes values
+   * @return a new instance of StagePeek
+   */
+  public StagePeek stagePeek(Operand<TInt32> index, List<Class<? extends TType>> dtypes,
+      StagePeek.Options... options) {
+    return StagePeek.create(scope, index, dtypes, options);
+  }
+
+  /**
+   * Op returns the number of elements in the underlying container.
+   *
+   * @param dtypes
+   * @param options carries optional attributes values
+   * @return a new instance of StageSize
+   */
+  public StageSize stageSize(List<Class<? extends TType>> dtypes, StageSize.Options... options) {
+    return StageSize.create(scope, dtypes, options);
+  }
+
+  /**
+   * Stops gradient computation.
+   *  <p>
+   *  When executed in a graph, this op outputs its input tensor as-is.
+   *  <p>
+   *  When building ops to compute gradients, this op prevents the contribution of
+   *  its inputs to be taken into account.  Normally, the gradient generator adds ops
+   *  to a graph to compute the derivatives of a specified 'loss' by recursively
+   *  finding out inputs that contributed to its computation.  If you insert this op
+   *  in the graph it inputs are masked from the gradient generator.  They are not
+   *  taken into account for computing gradients.
+   *  <p>
+   *  This is useful any time you want to compute a value with TensorFlow but need
+   *  to pretend that the value was a constant. Some examples include:
+   *  <ul>
+   *  <li>
+   *  The <i>EM</i> algorithm where the <i>M-step</i> should not involve backpropagation
+   *     through the output of the <i>E-step</i>.
+   *  </li>
+   *  <li>
+   *  Contrastive divergence training of Boltzmann machines where, when
+   *     differentiating the energy function, the training must not backpropagate
+   *     through the graph that generated the samples from the model.
+   *  </li>
+   *  <li>
+   *  Adversarial training, where no backprop should happen through the adversarial
+   *     example generation process.
+   *
+   * @param <T> data type for {@code output()} output
+   * @param input
+   * @return a new instance of StopGradient
+   */
+  public <T extends TType> StopGradient<T> stopGradient(Operand<T> input) {
+    return StopGradient.create(scope, input);
+  }
+
+  /**
+   * Return a strided slice from `input`.
+   *  <p>
+   *  Note, most python users will want to use the Python `Tensor.__getitem__`
+   *  or `Variable.__getitem__` rather than this op directly.
+   *  <p>
+   *  The goal of this op is to produce a new tensor with a subset of
+   *  the elements from the `n` dimensional `input` tensor. The subset is chosen using
+   *  a sequence of `m` sparse range specifications encoded into the arguments
+   *  of this function. Note, in some cases
+   *  `m` could be equal to `n`, but this need not be the case. Each
+   *  range specification entry can be one of the following:
+   *  <p>
+   *  - An ellipsis (...). Ellipses are used to imply zero or more
+   *    dimensions of full-dimension selection and are produced using
+   *    `ellipsis_mask`. For example, `foo[...]` is the identity slice.
+   *  <p>
+   *  - A new axis. This is used to insert a new shape=1 dimension and is
+   *    produced using `new_axis_mask`. For example, `foo[:, ...]` where
+   *    `foo` is shape `(3, 4)` produces a `(1, 3, 4)` tensor.
+   *  <p>
+   *  - A range `begin:end:stride`. This is used to specify how much to choose from
+   *    a given dimension. `stride` can be any integer but 0.  `begin` is an integer
+   *    which represents the index of the first value to select while `end` represents
+   *    the index of the last value to select. The number of values selected in each
+   *    dimension is `end - begin` if `stride > 0` and `begin - end` if `stride < 0`.
+   *    `begin` and `end` can be negative where `-1` is the last element, `-2` is
+   *    the second to last. `begin_mask` controls whether to replace the explicitly
+   *    given `begin` with an implicit effective value of `0` if `stride > 0` and
+   *    `-1` if `stride < 0`. `end_mask` is analogous but produces the number
+   *    required to create the largest open interval. For example, given a shape
+   *    `(3,)` tensor `foo[:]`, the effective `begin` and `end` are `0` and `3`. Do
+   *    not assume this is equivalent to `foo[0:-1]` which has an effective `begin`
+   *    and `end` of `0` and `2`. Another example is `foo[-2::-1]` which reverses the
+   *    first dimension of a tensor while dropping the last two (in the original
+   *    order elements). For example `foo = [1,2,3,4]; foo[-2::-1]` is `[4,3]`.
+   *  <p>
+   *  - A single index. This is used to keep only elements that have a given
+   *    index. For example (`foo[2, :]` on a shape `(5,6)` tensor produces a
+   *    shape `(6,)` tensor. This is encoded in `begin` and `end` and
+   *    `shrink_axis_mask`.
+   *  <p>
+   *  Each conceptual range specification is encoded in the op's argument. This
+   *  encoding is best understand by considering a non-trivial example. In
+   *  particular,
+   *  `foo[1, 2:4, None, ..., :-3:-1, :]` will be encoded as
+   *  <pre>{@code
+   *  begin = [1, 2, x, x, 0, x] # x denotes don't care (usually 0)
+   *  end = [2, 4, x, x, -3, x]
+   *  strides = [1, 1, x, x, -1, 1]
+   *  begin_mask = 1<<4 | 1<<5 = 48
+   *  end_mask = 1<<5 = 32
+   *  ellipsis_mask = 1<<3 = 8
+   *  new_axis_mask = 1<<2 = 4
+   *  shrink_axis_mask = 1<<0 = 1
+   *  }</pre>
+   *  In this case if `foo.shape` is (5, 5, 5, 5, 5, 5) the final shape of
+   *  the slice becomes (2, 1, 5, 5, 2, 5).
+   *  Let us walk step by step through each argument specification.
+   *  <p>
+   *  1.  The first argument in the example slice is turned into `begin = 1` and
+   *  `end = begin + 1 = 2`. To disambiguate from the original spec `2:4` we
+   *  also set the appropriate bit in `shrink_axis_mask`.
+   *  <p>
+   *  2. `2:4` is contributes 2, 4, 1 to begin, end, and stride. All masks have
+   *  zero bits contributed.
+   *  <p>
+   *  3. None is a synonym for `tf.newaxis`. This means insert a dimension of size 1
+   *  dimension in the final shape. Dummy values are contributed to begin,
+   *  end and stride, while the new_axis_mask bit is set.
+   *  <p>
+   *  4. `...` grab the full ranges from as many dimensions as needed to
+   *  fully specify a slice for every dimension of the input shape.
+   *  <p>
+   *  5. `:-3:-1` shows the use of negative indices. A negative index `i` associated
+   *  with a dimension that has shape `s` is converted to a positive index
+   *  `s + i`. So `-1` becomes `s-1` (i.e. the last element). This conversion
+   *  is done internally so begin, end and strides receive x, -3, and -1.
+   *  The appropriate begin_mask bit is set to indicate the start range is the
+   *  full range (ignoring the x).
+   *  <p>
+   *  6. `:` indicates that the entire contents of the corresponding dimension
+   *  is selected. This is equivalent to `::` or `0::1`. begin, end, and strides
+   *  receive 0, 0, and 1, respectively. The appropriate bits in `begin_mask` and
+   *  `end_mask` are also set.
+   *  <p>
+   *  <i>Requirements</i>:
+   *    `0 != strides[i] for i in [0, m)`
+   *    `ellipsis_mask must be a power of two (only one ellipsis)`
+   *
+   * @param <T> data type for {@code output()} output
+   * @param input
+   * @param begin `begin[k]` specifies the offset into the `k`th range specification.
+   *  The exact dimension this corresponds to will be determined by context.
+   *  Out-of-bounds values will be silently clamped. If the `k`th bit of
+   *  `begin_mask` then `begin[k]` is ignored and the full range of the
+   *  appropriate dimension is used instead. Negative values causes indexing
+   *  to start from the highest element e.g. If `foo==[1,2,3]` then `foo[-1]==3`.
+   * @param end `end[i]` is like `begin` with the exception that `end_mask` is
+   *  used to determine full ranges.
+   * @param strides `strides[i]` specifies the increment in the `i`th specification
+   *  after extracting a given element. Negative indices will reverse
+   *  the original order. Out or range values are
+   *  clamped to `[0,dim[i]) if slice[i]>0` or `[-1,dim[i]-1] if slice[i] < 0`
+   * @param options carries optional attributes values
+   * @return a new instance of StridedSlice
+   */
+  public <T extends TType, U extends TNumber> StridedSlice<T> stridedSlice(Operand<T> input,
+      Operand<U> begin, Operand<U> end, Operand<U> strides, StridedSlice.Options... options) {
+    return StridedSlice.create(scope, input, begin, end, strides, options);
+  }
+
+  /**
+   * Assign `value` to the sliced l-value reference of `ref`.
+   *  <p>
+   *  The values of `value` are assigned to the positions in the variable
+   *  `ref` that are selected by the slice parameters. The slice parameters
+   *  `begin`, `end`, `strides`, etc. work exactly as in `StridedSlice`.
+   *  <p>
+   *  NOTE this op currently does not support broadcasting and so `value`'s
+   *  shape must be exactly the shape produced by the slice of `ref`.
+   *
+   * @param <T> data type for {@code outputRef()} output
+   * @param ref
+   * @param begin
+   * @param end
+   * @param strides
+   * @param value
+   * @param options carries optional attributes values
+   * @return a new instance of StridedSliceAssign
+   */
+  public <T extends TType, U extends TNumber> StridedSliceAssign<T> stridedSliceAssign(
+      Operand<T> ref, Operand<U> begin, Operand<U> end, Operand<U> strides, Operand<T> value,
+      StridedSliceAssign.Options... options) {
+    return StridedSliceAssign.create(scope, ref, begin, end, strides, value, options);
+  }
+
+  /**
+   * Returns the gradient of `StridedSlice`.
+   *  <p>
+   *  Since `StridedSlice` cuts out pieces of its `input` which is size
+   *  `shape`, its gradient will have the same shape (which is passed here
+   *  as `shape`). The gradient will be zero in any element that the slice
+   *  does not select.
+   *  <p>
+   *  Arguments are the same as StridedSliceGrad with the exception that
+   *  `dy` is the input gradient to be propagated and `shape` is the
+   *  shape of `StridedSlice`'s `input`.
+   *
+   * @param <U> data type for {@code output()} output
+   * @param shape
+   * @param begin
+   * @param end
+   * @param strides
+   * @param dy
+   * @param options carries optional attributes values
+   * @return a new instance of StridedSliceGrad
+   */
+  public <U extends TType, T extends TNumber> StridedSliceGrad<U> stridedSliceGrad(Operand<T> shape,
+      Operand<T> begin, Operand<T> end, Operand<T> strides, Operand<U> dy,
+      StridedSliceGrad.Options... options) {
+    return StridedSliceGrad.create(scope, shape, begin, end, strides, dy, options);
+  }
+
+  /**
+   * Computes the sum of elements across dimensions of a tensor.
+   *  <p>
+   *  Reduces `input` along the dimensions given in `axis`. Unless
+   *  `keep_dims` is true, the rank of the tensor is reduced by 1 for each entry in
+   *  `axis`. If `keep_dims` is true, the reduced dimensions are
+   *  retained with length 1.
+   *
+   * @param <T> data type for {@code output()} output
+   * @param input The tensor to reduce.
+   * @param axis The dimensions to reduce. Must be in the range
+   *  `[-rank(input), rank(input))`.
+   * @param options carries optional attributes values
+   * @return a new instance of Sum
+   */
+  public <T extends TType, U extends TNumber> Sum<T> sum(Operand<T> input, Operand<U> axis,
+      Sum.Options... options) {
+    return Sum.create(scope, input, axis, options);
+  }
+
+  /**
+   * Forwards `data` to the output port determined by `pred`.
+   *  <p>
+   *  If `pred` is true, the `data` input is forwarded to `output_true`. Otherwise,
+   *  the data goes to `output_false`.
+   *  <p>
+   *  See also `RefSwitch` and `Merge`.
+   *
+   * @param <T> data type for {@code outputFalse()} output
+   * @param data The tensor to be forwarded to the appropriate output.
+   * @param pred A scalar that specifies which output port will receive data.
+   * @return a new instance of SwitchCond
+   */
+  public <T extends TType> SwitchCond<T> switchCond(Operand<T> data, Operand<TBool> pred) {
+    return SwitchCond.create(scope, data, pred);
+  }
+
+  /**
+   * Returns a tensor that may be mutated, but only persists within a single step.
+   *  <p>
+   *  This is an experimental op for internal use only and it is possible to use this
+   *  op in unsafe ways.  DO NOT USE unless you fully understand the risks.
+   *  <p>
+   *  It is the caller's responsibility to ensure that 'ref' is eventually passed to a
+   *  matching 'DestroyTemporaryVariable' op after all other uses have completed.
+   *  <p>
+   *  Outputs a ref to the tensor state so it may be read or modified.
+   *  <p>
+   *    E.g.
+   *        var = state_ops._temporary_variable([1, 2], types.float_)
+   *        var_name = var.op.name
+   *        var = state_ops.assign(var, [[4.0, 5.0]])
+   *        var = state_ops.assign_add(var, [[6.0, 7.0]])
+   *        final = state_ops._destroy_temporary_variable(var, var_name=var_name)
+   *
+   * @param <T> data type for {@code ref()} output
+   * @param shape The shape of the variable tensor.
+   * @param dtype The type of elements in the variable tensor.
+   * @param options carries optional attributes values
+   * @return a new instance of TemporaryVariable
+   */
+  public <T extends TType> TemporaryVariable<T> temporaryVariable(Shape shape, Class<T> dtype,
+      TemporaryVariable.Options... options) {
+    return TemporaryVariable.create(scope, shape, dtype, options);
+  }
+
+  /**
+   * An array of Tensors of given size.
+   *  <p>
+   *  Write data via Write and read via Read or Pack.
+   *
+   * @param size The size of the array.
+   * @param dtype The type of the elements on the tensor_array.
+   * @param options carries optional attributes values
+   * @return a new instance of TensorArray
+   */
+  public <T extends TType> TensorArray tensorArray(Operand<TInt32> size, Class<T> dtype,
+      TensorArray.Options... options) {
+    return TensorArray.create(scope, size, dtype, options);
+  }
+
+  /**
+   * Delete the TensorArray from its resource container.
+   *  <p>
+   *  This enables the user to close and release the resource in the middle
+   *  of a step/run.
+   *
+   * @param handle The handle to a TensorArray (output of TensorArray or TensorArrayGrad).
+   * @return a new instance of TensorArrayClose
+   */
+  public TensorArrayClose tensorArrayClose(Operand<?> handle) {
+    return TensorArrayClose.create(scope, handle);
+  }
+
+  /**
+   * Concat the elements from the TensorArray into value `value`.
+   *  <p>
+   *  Takes `T` elements of shapes
+   *  <p>
+   *    <pre>{@code
+   *    (n0 x d0 x d1 x ...), (n1 x d0 x d1 x ...), ..., (n(T-1) x d0 x d1 x ...)
+   *    }</pre>
+   *  and concatenates them into a Tensor of shape:
+   *  <p>
+   *    <pre>{@code
+   *  (n0 + n1 + ... + n(T-1) x d0 x d1 x ...)}</pre>
+   *  All elements must have the same shape (excepting the first dimension).
+   *
+   * @param <T> data type for {@code value()} output
+   * @param handle The handle to a TensorArray.
+   * @param flowIn A float scalar that enforces proper chaining of operations.
+   * @param dtype The type of the elem that is returned.
+   * @param options carries optional attributes values
+   * @return a new instance of TensorArrayConcat
+   */
+  public <T extends TType> TensorArrayConcat<T> tensorArrayConcat(Operand<?> handle,
+      Operand<TFloat32> flowIn, Class<T> dtype, TensorArrayConcat.Options... options) {
+    return TensorArrayConcat.create(scope, handle, flowIn, dtype, options);
+  }
+
+  /**
+   * Gather specific elements from the TensorArray into output `value`.
+   *  <p>
+   *  All elements selected by `indices` must have the same shape.
+   *
+   * @param <T> data type for {@code value()} output
+   * @param handle The handle to a TensorArray.
+   * @param indices The locations in the TensorArray from which to read tensor elements.
+   * @param flowIn A float scalar that enforces proper chaining of operations.
+   * @param dtype The type of the elem that is returned.
+   * @param options carries optional attributes values
+   * @return a new instance of TensorArrayGather
+   */
+  public <T extends TType> TensorArrayGather<T> tensorArrayGather(Operand<?> handle,
+      Operand<TInt32> indices, Operand<TFloat32> flowIn, Class<T> dtype,
+      TensorArrayGather.Options... options) {
+    return TensorArrayGather.create(scope, handle, indices, flowIn, dtype, options);
+  }
+
+  /**
+   * Creates a TensorArray for storing the gradients of values in the given handle.
+   *  <p>
+   *  If the given TensorArray gradient already exists, returns a reference to it.
+   *  <p>
+   *  Locks the size of the original TensorArray by disabling its dynamic size flag.
+   *  <p>
+   * *A note about the input flow_in:**
+   *  <p>
+   *  The handle flow_in forces the execution of the gradient lookup to occur
+   *  only after certain other operations have occurred.  For example, when
+   *  the forward TensorArray is dynamically sized, writes to this TensorArray
+   *  may resize the object.  The gradient TensorArray is statically sized based
+   *  on the size of the forward TensorArray when this operation executes.
+   *  Furthermore, the size of the forward TensorArray is frozen by this call.
+   *  As a result, the flow is used to ensure that the call to generate the gradient
+   *  TensorArray only happens after all writes are executed.
+   *  <p>
+   *  In the case of dynamically sized TensorArrays, gradient computation should
+   *  only be performed on read operations that have themselves been chained via
+   *  flow to occur only after all writes have executed. That way the final size
+   *  of the forward TensorArray is known when this operation is called.
+   *  <p>
+   * *A note about the source attribute:**
+   *  <p>
+   *  TensorArray gradient calls use an accumulator TensorArray object.  If
+   *  multiple gradients are calculated and run in the same session, the multiple
+   *  gradient nodes may accidentally flow through the same accumulator TensorArray.
+   *  This double counts and generally breaks the TensorArray gradient flow.
+   *  <p>
+   *  The solution is to identify which gradient call this particular
+   *  TensorArray gradient is being called in.  This is performed by identifying
+   *  a unique string (e.g. "gradients", "gradients_1", ...) from the input
+   *  gradient Tensor's name.  This string is used as a suffix when creating
+   *  the TensorArray gradient object here (the attribute `source`).
+   *  <p>
+   *  The attribute `source` is added as a suffix to the forward TensorArray's
+   *  name when performing the creation / lookup, so that each separate gradient
+   *  calculation gets its own TensorArray accumulator.
+   *
+   * @param handle The handle to the forward TensorArray.
+   * @param flowIn A float scalar that enforces proper chaining of operations.
+   * @param source The gradient source string, used to decide which gradient TensorArray
+   *  to return.
+   * @return a new instance of TensorArrayGrad
+   */
+  public TensorArrayGrad tensorArrayGrad(Operand<?> handle, Operand<TFloat32> flowIn,
+      String source) {
+    return TensorArrayGrad.create(scope, handle, flowIn, source);
+  }
+
+  /**
+   * Creates a TensorArray for storing multiple gradients of values in the given handle.
+   *  <p>
+   *  Similar to TensorArrayGradV3. However it creates an accumulator with an
+   *  expanded shape compared to the input TensorArray whose gradient is being
+   *  computed. This enables multiple gradients for the same TensorArray to be
+   *  calculated using the same accumulator.
+   *
+   * @param handle The handle to the forward TensorArray.
+   * @param flowIn A float scalar that enforces proper chaining of operations.
+   * @param shapeToPrepend An int32 vector representing a shape. Elements in the gradient accumulator will
+   *  have shape which is this shape_to_prepend value concatenated with shape of the
+   *  elements in the TensorArray corresponding to the input handle.
+   * @param source The gradient source string, used to decide which gradient TensorArray
+   *  to return.
+   * @return a new instance of TensorArrayGradWithShape
+   */
+  public TensorArrayGradWithShape tensorArrayGradWithShape(Operand<?> handle,
+      Operand<TFloat32> flowIn, Operand<TInt32> shapeToPrepend, String source) {
+    return TensorArrayGradWithShape.create(scope, handle, flowIn, shapeToPrepend, source);
+  }
+
+  /**
+   *
+   * @param <T> data type for {@code value()} output
+   * @param handle
+   * @param flowIn
+   * @param dtype
+   * @param options carries optional attributes values
+   * @return a new instance of TensorArrayPack
+   */
+  public <T extends TType> TensorArrayPack<T> tensorArrayPack(Operand<TString> handle,
+      Operand<TFloat32> flowIn, Class<T> dtype, TensorArrayPack.Options... options) {
+    return TensorArrayPack.create(scope, handle, flowIn, dtype, options);
+  }
+
+  /**
+   * Read an element from the TensorArray into output `value`.
+   *
+   * @param <T> data type for {@code value()} output
+   * @param handle The handle to a TensorArray.
+   * @param index
+   * @param flowIn A float scalar that enforces proper chaining of operations.
+   * @param dtype The type of the elem that is returned.
+   * @return a new instance of TensorArrayRead
+   */
+  public <T extends TType> TensorArrayRead<T> tensorArrayRead(Operand<?> handle,
+      Operand<TInt32> index, Operand<TFloat32> flowIn, Class<T> dtype) {
+    return TensorArrayRead.create(scope, handle, index, flowIn, dtype);
+  }
+
+  /**
+   * Scatter the data from the input value into specific TensorArray elements.
+   *  <p>
+   *  `indices` must be a vector, its length must match the first dim of `value`.
+   *
+   * @param handle The handle to a TensorArray.
+   * @param indices The locations at which to write the tensor elements.
+   * @param value The concatenated tensor to write to the TensorArray.
+   * @param flowIn A float scalar that enforces proper chaining of operations.
+   * @return a new instance of TensorArrayScatter
+   */
+  public <T extends TType> TensorArrayScatter tensorArrayScatter(Operand<?> handle,
+      Operand<TInt32> indices, Operand<T> value, Operand<TFloat32> flowIn) {
+    return TensorArrayScatter.create(scope, handle, indices, value, flowIn);
+  }
+
+  /**
+   * Get the current size of the TensorArray.
+   *
+   * @param handle The handle to a TensorArray (output of TensorArray or TensorArrayGrad).
+   * @param flowIn A float scalar that enforces proper chaining of operations.
+   * @return a new instance of TensorArraySize
+   */
+  public TensorArraySize tensorArraySize(Operand<?> handle, Operand<TFloat32> flowIn) {
+    return TensorArraySize.create(scope, handle, flowIn);
+  }
+
+  /**
+   * Split the data from the input value into TensorArray elements.
+   *  <p>
+   *  Assuming that `lengths` takes on values
+   *  <p>
+   *    <pre>{@code
+   *  (n0, n1, ..., n(T-1))}</pre>
+   *  and that `value` has shape
+   *  <p>
+   *    <pre>{@code
+   *  (n0 + n1 + ... + n(T-1) x d0 x d1 x ...)}</pre>
+   *  ,
+   *  <p>
+   *  this splits values into a TensorArray with T tensors.
+   *  <p>
+   *  TensorArray index t will be the subtensor of values with starting position
+   *  <p>
+   *    <pre>{@code
+   *  (n0 + n1 + ... + n(t-1), 0, 0, ...)}</pre>
+   *  and having size
+   *  <p>
+   *    <pre>{@code
+   *  nt x d0 x d1 x ...}</pre>
+   *
+   * @param handle The handle to a TensorArray.
+   * @param value The concatenated tensor to write to the TensorArray.
+   * @param lengths The vector of lengths, how to split the rows of value into the
+   *  TensorArray.
+   * @param flowIn A float scalar that enforces proper chaining of operations.
+   * @return a new instance of TensorArraySplit
+   */
+  public <T extends TType> TensorArraySplit tensorArraySplit(Operand<?> handle, Operand<T> value,
+      Operand<TInt64> lengths, Operand<TFloat32> flowIn) {
+    return TensorArraySplit.create(scope, handle, value, lengths, flowIn);
+  }
+
+  /**
+   *
+   * @param handle
+   * @param value
+   * @param flowIn
+   * @return a new instance of TensorArrayUnpack
+   */
+  public <T extends TType> TensorArrayUnpack tensorArrayUnpack(Operand<TString> handle,
+      Operand<T> value, Operand<TFloat32> flowIn) {
+    return TensorArrayUnpack.create(scope, handle, value, flowIn);
+  }
+
+  /**
+   * Push an element onto the tensor_array.
+   *
+   * @param handle The handle to a TensorArray.
+   * @param index The position to write to inside the TensorArray.
+   * @param value The tensor to write to the TensorArray.
+   * @param flowIn A float scalar that enforces proper chaining of operations.
+   * @return a new instance of TensorArrayWrite
+   */
+  public <T extends TType> TensorArrayWrite tensorArrayWrite(Operand<?> handle,
+      Operand<TInt32> index, Operand<T> value, Operand<TFloat32> flowIn) {
+    return TensorArrayWrite.create(scope, handle, index, value, flowIn);
+  }
+
+  /**
+   * Concats all tensors in the list along the 0th dimension.
+   *  <p>
+   *  Requires that all tensors have the same shape except the first dimension.
+   *  <p>
+   *  input_handle: The input list.
+   *  element_shape: The shape of the uninitialized elements in the list. If the first
+   *    dimension is not -1, it is assumed that all list elements have the same
+   *    leading dim.
+   *  leading_dims: The list of leading dims of uninitialized list elements. Used if
+   *    the leading dim of input_handle.element_shape or the element_shape input arg
+   *    is not already set.
+   *  tensor: The concated result.
+   *  lengths: Output tensor containing sizes of the 0th dimension of tensors in the list, used for computing the gradient.
+   *
+   * @param <U> data type for {@code tensor()} output
+   * @param inputHandle
+   * @param elementShape
+   * @param leadingDims
+   * @param elementDtype
+   * @return a new instance of TensorListConcat
+   */
+  public <U extends TType, T extends TNumber> TensorListConcat<U> tensorListConcat(
+      Operand<?> inputHandle, Operand<T> elementShape, Operand<TInt64> leadingDims,
+      Class<U> elementDtype) {
+    return TensorListConcat.create(scope, inputHandle, elementShape, leadingDims, elementDtype);
+  }
+
+  /**
+   *
+   * @param inputA
+   * @param inputB
+   * @param elementDtype
+   * @return a new instance of TensorListConcatLists
+   */
+  public <T extends TType> TensorListConcatLists tensorListConcatLists(Operand<?> inputA,
+      Operand<?> inputB, Class<T> elementDtype) {
+    return TensorListConcatLists.create(scope, inputA, inputB, elementDtype);
+  }
+
+  /**
+   * The shape of the elements of the given list, as a tensor.
+   *  <p>
+   *    input_handle: the list
+   *    element_shape: the shape of elements of the list
+   *
+   * @param <T> data type for {@code elementShape()} output
+   * @param inputHandle
+   * @param shapeType
+   * @return a new instance of TensorListElementShape
+   */
+  public <T extends TNumber> TensorListElementShape<T> tensorListElementShape(
+      Operand<?> inputHandle, Class<T> shapeType) {
+    return TensorListElementShape.create(scope, inputHandle, shapeType);
+  }
+
+  /**
+   * Creates a TensorList which, when stacked, has the value of `tensor`.
+   *  <p>
+   *  Each tensor in the result list corresponds to one row of the input tensor.
+   *  <p>
+   *  tensor: The input tensor.
+   *  output_handle: The list.
+   *
+   * @param tensor
+   * @param elementShape
+   * @return a new instance of TensorListFromTensor
+   */
+  public <T extends TType, U extends TNumber> TensorListFromTensor tensorListFromTensor(
+      Operand<T> tensor, Operand<U> elementShape) {
+    return TensorListFromTensor.create(scope, tensor, elementShape);
+  }
+
+  /**
+   * Creates a Tensor by indexing into the TensorList.
+   *  <p>
+   *  Each row in the produced Tensor corresponds to the element in the TensorList
+   *  specified by the given index (see `tf.gather`).
+   *  <p>
+   *  input_handle: The input tensor list.
+   *  indices: The indices used to index into the list.
+   *  values: The tensor.
+   *
+   * @param <T> data type for {@code values()} output
+   * @param inputHandle
+   * @param indices
+   * @param elementShape
+   * @param elementDtype
+   * @return a new instance of TensorListGather
+   */
+  public <T extends TType> TensorListGather<T> tensorListGather(Operand<?> inputHandle,
+      Operand<TInt32> indices, Operand<TInt32> elementShape, Class<T> elementDtype) {
+    return TensorListGather.create(scope, inputHandle, indices, elementShape, elementDtype);
+  }
+
+  /**
+   *
+   * @param <T> data type for {@code item()} output
+   * @param inputHandle
+   * @param index
+   * @param elementShape
+   * @param elementDtype
+   * @return a new instance of TensorListGetItem
+   */
+  public <T extends TType> TensorListGetItem<T> tensorListGetItem(Operand<?> inputHandle,
+      Operand<TInt32> index, Operand<TInt32> elementShape, Class<T> elementDtype) {
+    return TensorListGetItem.create(scope, inputHandle, index, elementShape, elementDtype);
+  }
+
+  /**
+   * Returns the number of tensors in the input tensor list.
+   *  <p>
+   *  input_handle: the input list
+   *  length: the number of tensors in the list
+   *
+   * @param inputHandle
+   * @return a new instance of TensorListLength
+   */
+  public TensorListLength tensorListLength(Operand<?> inputHandle) {
+    return TensorListLength.create(scope, inputHandle);
+  }
+
+  /**
+   * Returns the last element of the input list as well as a list with all but that element.
+   *  <p>
+   *  Fails if the list is empty.
+   *  <p>
+   *  input_handle: the input list
+   *  tensor: the withdrawn last element of the list
+   *  element_dtype: the type of elements in the list
+   *  element_shape: the shape of the output tensor
+   *
+   * @param <T> data type for {@code tensor()} output
+   * @param inputHandle
+   * @param elementShape
+   * @param elementDtype
+   * @return a new instance of TensorListPopBack
+   */
+  public <T extends TType> TensorListPopBack<T> tensorListPopBack(Operand<?> inputHandle,
+      Operand<TInt32> elementShape, Class<T> elementDtype) {
+    return TensorListPopBack.create(scope, inputHandle, elementShape, elementDtype);
+  }
+
+  /**
+   * Returns a list which has the passed-in `Tensor` as last element and the other elements of the given list in `input_handle`.
+   *  <p>
+   *  tensor: The tensor to put on the list.
+   *  input_handle: The old list.
+   *  output_handle: A list with the elements of the old list followed by tensor.
+   *  element_dtype: the type of elements in the list.
+   *  element_shape: a shape compatible with that of elements in the list.
+   *
+   * @param inputHandle
+   * @param tensor
+   * @return a new instance of TensorListPushBack
+   */
+  public <T extends TType> TensorListPushBack tensorListPushBack(Operand<?> inputHandle,
+      Operand<T> tensor) {
+    return TensorListPushBack.create(scope, inputHandle, tensor);
+  }
+
+  /**
+   *
+   * @param inputHandles
+   * @param tensor
+   * @return a new instance of TensorListPushBackBatch
+   */
+  public <T extends TType> TensorListPushBackBatch tensorListPushBackBatch(Operand<?> inputHandles,
+      Operand<T> tensor) {
+    return TensorListPushBackBatch.create(scope, inputHandles, tensor);
+  }
+
+  /**
+   * List of the given size with empty elements.
+   *  <p>
+   *  element_shape: the shape of the future elements of the list
+   *  num_elements: the number of elements to reserve
+   *  handle: the output list
+   *  element_dtype: the desired type of elements in the list.
+   *
+   * @param elementShape
+   * @param numElements
+   * @param elementDtype
+   * @return a new instance of TensorListReserve
+   */
+  public <T extends TNumber, U extends TType> TensorListReserve tensorListReserve(
+      Operand<T> elementShape, Operand<TInt32> numElements, Class<U> elementDtype) {
+    return TensorListReserve.create(scope, elementShape, numElements, elementDtype);
+  }
+
+  /**
+   * Resizes the list.
+   *  <p>
+   *
+   *  input_handle: the input list
+   *  size: size of the output list
+   *
+   * @param inputHandle
+   * @param size
+   * @return a new instance of TensorListResize
+   */
+  public TensorListResize tensorListResize(Operand<?> inputHandle, Operand<TInt32> size) {
+    return TensorListResize.create(scope, inputHandle, size);
+  }
+
+  /**
+   * Creates a TensorList by indexing into a Tensor.
+   *  <p>
+   *  Each member of the TensorList corresponds to one row of the input tensor,
+   *  specified by the given index (see `tf.gather`).
+   *  <p>
+   *  tensor: The input tensor.
+   *  indices: The indices used to index into the list.
+   *  element_shape: The shape of the elements in the list (can be less specified than
+   *    the shape of the tensor).
+   *  num_elements: The size of the output list. Must be large enough to accommodate
+   *    the largest index in indices. If -1, the list is just large enough to include
+   *    the largest index in indices.
+   *  output_handle: The TensorList.
+   *
+   * @param tensor
+   * @param indices
+   * @param elementShape
+   * @param numElements
+   * @return a new instance of TensorListScatter
+   */
+  public <T extends TType, U extends TNumber> TensorListScatter tensorListScatter(Operand<T> tensor,
+      Operand<TInt32> indices, Operand<U> elementShape, Operand<TInt32> numElements) {
+    return TensorListScatter.create(scope, tensor, indices, elementShape, numElements);
+  }
+
+  /**
+   * Scatters tensor at indices in an input list.
+   *  <p>
+   *  Each member of the TensorList corresponds to one row of the input tensor,
+   *  specified by the given index (see `tf.gather`).
+   *  <p>
+   *  input_handle: The list to scatter into.
+   *  tensor: The input tensor.
+   *  indices: The indices used to index into the list.
+   *  output_handle: The TensorList.
+   *
+   * @param inputHandle
+   * @param tensor
+   * @param indices
+   * @return a new instance of TensorListScatterIntoExistingList
+   */
+  public <T extends TType> TensorListScatterIntoExistingList tensorListScatterIntoExistingList(
+      Operand<?> inputHandle, Operand<T> tensor, Operand<TInt32> indices) {
+    return TensorListScatterIntoExistingList.create(scope, inputHandle, tensor, indices);
+  }
+
+  /**
+   *
+   * @param inputHandle
+   * @param index
+   * @param item
+   * @return a new instance of TensorListSetItem
+   */
+  public <T extends TType> TensorListSetItem tensorListSetItem(Operand<?> inputHandle,
+      Operand<TInt32> index, Operand<T> item) {
+    return TensorListSetItem.create(scope, inputHandle, index, item);
+  }
+
+  /**
+   * Splits a tensor into a list.
+   *  <p>
+   *  list[i] corresponds to lengths[i] tensors from the input tensor.
+   *  The tensor must have rank at least 1 and contain exactly sum(lengths) elements.
+   *  <p>
+   *  tensor: The input tensor.
+   *  element_shape: A shape compatible with that of elements in the tensor.
+   *  lengths: Vector of sizes of the 0th dimension of tensors in the list.
+   *  output_handle: The list.
+   *
+   * @param tensor
+   * @param elementShape
+   * @param lengths
+   * @return a new instance of TensorListSplit
+   */
+  public <T extends TType, U extends TNumber> TensorListSplit tensorListSplit(Operand<T> tensor,
+      Operand<U> elementShape, Operand<TInt64> lengths) {
+    return TensorListSplit.create(scope, tensor, elementShape, lengths);
+  }
+
+  /**
+   * Stacks all tensors in the list.
+   *  <p>
+   *  Requires that all tensors have the same shape.
+   *  <p>
+   *  input_handle: the input list
+   *  tensor: the gathered result
+   *  num_elements: optional. If not -1, the number of elements in the list.
+   *
+   * @param <T> data type for {@code tensor()} output
+   * @param inputHandle
+   * @param elementShape
+   * @param elementDtype
+   * @param options carries optional attributes values
+   * @return a new instance of TensorListStack
+   */
+  public <T extends TType> TensorListStack<T> tensorListStack(Operand<?> inputHandle,
+      Operand<TInt32> elementShape, Class<T> elementDtype, TensorListStack.Options... options) {
+    return TensorListStack.create(scope, inputHandle, elementShape, elementDtype, options);
+  }
+
+  /**
+   * Adds sparse `updates` to an existing tensor according to `indices`.
+   *  <p>
+   *  This operation creates a new tensor by adding sparse `updates` to the passed
+   *  in `tensor`.
+   *  This operation is very similar to `tf.scatter_nd_add`, except that the updates
+   *  are added onto an existing tensor (as opposed to a variable). If the memory
+   *  for the existing tensor cannot be re-used, a copy is made and updated.
+   *  <p>
+   *  `indices` is an integer tensor containing indices into a new tensor of shape
+   *  `tensor.shape`.  The last dimension of `indices` can be at most the rank of
+   *  `tensor.shape`:
+   *  <p>
+   *      indices.shape[-1] <= tensor.shape.rank
+   *  <p>
+   *  The last dimension of `indices` corresponds to indices into elements
+   *  (if `indices.shape[-1] = tensor.shape.rank`) or slices
+   *  (if `indices.shape[-1] < tensor.shape.rank`) along dimension
+   *  `indices.shape[-1]` of `tensor.shape`.  `updates` is a tensor with shape
+   *  <p>
+   *      indices.shape[:-1] + tensor.shape[indices.shape[-1]:]
+   *  <p>
+   *  The simplest form of tensor_scatter_add is to add individual elements to a
+   *  tensor by index. For example, say we want to add 4 elements in a rank-1
+   *  tensor with 8 elements.
+   *  <p>
+   *  In Python, this scatter add operation would look like this:
+   *  <pre>{@code
+   *      indices = tf.constant([[4], [3], [1], [7]])
+   *      updates = tf.constant([9, 10, 11, 12])
+   *      tensor = tf.ones([8], dtype=tf.int32)
+   *      updated = tf.tensor_scatter_nd_add(tensor, indices, updates)
+   *      print(updated)
+   *  }</pre>
+   *  The resulting tensor would look like this:
+   *  <p>
+   *      [1, 12, 1, 11, 10, 1, 1, 13]
+   *  <p>
+   *  We can also, insert entire slices of a higher rank tensor all at once. For
+   *  example, if we wanted to insert two slices in the first dimension of a
+   *  rank-3 tensor with two matrices of new values.
+   *  <p>
+   *  In Python, this scatter add operation would look like this:
+   *  <pre>{@code
+   *      indices = tf.constant([[0], [2]])
+   *      updates = tf.constant([[[5, 5, 5, 5], [6, 6, 6, 6],
+   *                              [7, 7, 7, 7], [8, 8, 8, 8]],
+   *                             [[5, 5, 5, 5], [6, 6, 6, 6],
+   *                              [7, 7, 7, 7], [8, 8, 8, 8]]])
+   *      tensor = tf.ones([4, 4, 4],dtype=tf.int32)
+   *      updated = tf.tensor_scatter_nd_add(tensor, indices, updates)
+   *      print(updated)
+   *  }</pre>
+   *  The resulting tensor would look like this:
+   *  <p>
+   *      [[[6, 6, 6, 6], [7, 7, 7, 7], [8, 8, 8, 8], [9, 9, 9, 9]],
+   *       [[1, 1, 1, 1], [1, 1, 1, 1], [1, 1, 1, 1], [1, 1, 1, 1]],
+   *       [[6, 6, 6, 6], [7, 7, 7, 7], [8, 8, 8, 8], [9, 9, 9, 9]],
+   *       [[1, 1, 1, 1], [1, 1, 1, 1], [1, 1, 1, 1], [1, 1, 1, 1]]]
+   *  <p>
+   *  Note that on CPU, if an out of bound index is found, an error is returned.
+   *  On GPU, if an out of bound index is found, the index is ignored.
+   *
+   * @param <T> data type for {@code output()} output
+   * @param tensor Tensor to copy/update.
+   * @param indices Index tensor.
+   * @param updates Updates to scatter into output.
+   * @return a new instance of TensorScatterNdAdd
+   */
+  public <T extends TType, U extends TNumber> TensorScatterNdAdd<T> tensorScatterNdAdd(
+      Operand<T> tensor, Operand<U> indices, Operand<T> updates) {
+    return TensorScatterNdAdd.create(scope, tensor, indices, updates);
+  }
+
+  /**
+   *
+   * @param <T> data type for {@code output()} output
+   * @param tensor Tensor to update.
+   * @param indices Index tensor.
+   * @param updates Updates to scatter into output.
+   * @return a new instance of TensorScatterNdMax
+   */
+  public <T extends TType, U extends TNumber> TensorScatterNdMax<T> tensorScatterNdMax(
+      Operand<T> tensor, Operand<U> indices, Operand<T> updates) {
+    return TensorScatterNdMax.create(scope, tensor, indices, updates);
+  }
+
+  /**
+   *
+   * @param <T> data type for {@code output()} output
+   * @param tensor Tensor to update.
+   * @param indices Index tensor.
+   * @param updates Updates to scatter into output.
+   * @return a new instance of TensorScatterNdMin
+   */
+  public <T extends TType, U extends TNumber> TensorScatterNdMin<T> tensorScatterNdMin(
+      Operand<T> tensor, Operand<U> indices, Operand<T> updates) {
+    return TensorScatterNdMin.create(scope, tensor, indices, updates);
+  }
+
+  /**
+   * Subtracts sparse `updates` from an existing tensor according to `indices`.
+   *  <p>
+   *  This operation creates a new tensor by subtracting sparse `updates` from the
+   *  passed in `tensor`.
+   *  This operation is very similar to `tf.scatter_nd_sub`, except that the updates
+   *  are subtracted from an existing tensor (as opposed to a variable). If the memory
+   *  for the existing tensor cannot be re-used, a copy is made and updated.
+   *  <p>
+   *  `indices` is an integer tensor containing indices into a new tensor of shape
+   *  `shape`.  The last dimension of `indices` can be at most the rank of `shape`:
+   *  <p>
+   *      indices.shape[-1] <= shape.rank
+   *  <p>
+   *  The last dimension of `indices` corresponds to indices into elements
+   *  (if `indices.shape[-1] = shape.rank`) or slices
+   *  (if `indices.shape[-1] < shape.rank`) along dimension `indices.shape[-1]` of
+   *  `shape`.  `updates` is a tensor with shape
+   *  <p>
+   *      indices.shape[:-1] + shape[indices.shape[-1]:]
+   *  <p>
+   *  The simplest form of tensor_scatter_sub is to subtract individual elements
+   *  from a tensor by index. For example, say we want to insert 4 scattered elements
+   *  in a rank-1 tensor with 8 elements.
+   *  <p>
+   *  In Python, this scatter subtract operation would look like this:
+   *  <pre>{@code
+   *      indices = tf.constant([[4], [3], [1], [7]])
+   *      updates = tf.constant([9, 10, 11, 12])
+   *      tensor = tf.ones([8], dtype=tf.int32)
+   *      updated = tf.tensor_scatter_nd_sub(tensor, indices, updates)
+   *      print(updated)
+   *  }</pre>
+   *  The resulting tensor would look like this:
+   *  <p>
+   *      [1, -10, 1, -9, -8, 1, 1, -11]
+   *  <p>
+   *  We can also, insert entire slices of a higher rank tensor all at once. For
+   *  example, if we wanted to insert two slices in the first dimension of a
+   *  rank-3 tensor with two matrices of new values.
+   *  <p>
+   *  In Python, this scatter add operation would look like this:
+   *  <pre>{@code
+   *      indices = tf.constant([[0], [2]])
+   *      updates = tf.constant([[[5, 5, 5, 5], [6, 6, 6, 6],
+   *                              [7, 7, 7, 7], [8, 8, 8, 8]],
+   *                             [[5, 5, 5, 5], [6, 6, 6, 6],
+   *                              [7, 7, 7, 7], [8, 8, 8, 8]]])
+   *      tensor = tf.ones([4, 4, 4],dtype=tf.int32)
+   *      updated = tf.tensor_scatter_nd_sub(tensor, indices, updates)
+   *      print(updated)
+   *  }</pre>
+   *  The resulting tensor would look like this:
+   *  <p>
+   *      [[[-4, -4, -4, -4], [-5, -5, -5, -5], [-6, -6, -6, -6], [-7, -7, -7, -7]],
+   *       [[1, 1, 1, 1], [1, 1, 1, 1], [1, 1, 1, 1], [1, 1, 1, 1]],
+   *       [[-4, -4, -4, -4], [-5, -5, -5, -5], [-6, -6, -6, -6], [-7, -7, -7, -7]],
+   *       [[1, 1, 1, 1], [1, 1, 1, 1], [1, 1, 1, 1], [1, 1, 1, 1]]]
+   *  <p>
+   *  Note that on CPU, if an out of bound index is found, an error is returned.
+   *  On GPU, if an out of bound index is found, the index is ignored.
+   *
+   * @param <T> data type for {@code output()} output
+   * @param tensor Tensor to copy/update.
+   * @param indices Index tensor.
+   * @param updates Updates to scatter into output.
+   * @return a new instance of TensorScatterNdSub
+   */
+  public <T extends TType, U extends TNumber> TensorScatterNdSub<T> tensorScatterNdSub(
+      Operand<T> tensor, Operand<U> indices, Operand<T> updates) {
+    return TensorScatterNdSub.create(scope, tensor, indices, updates);
+  }
+
+  /**
+   * Scatter `updates` into an existing tensor according to `indices`.
+   *  <p>
+   *  This operation creates a new tensor by applying sparse `updates` to the passed
+   *  in `tensor`.
+   *  This operation is very similar to `tf.scatter_nd`, except that the updates are
+   *  scattered onto an existing tensor (as opposed to a zero-tensor). If the memory
+   *  for the existing tensor cannot be re-used, a copy is made and updated.
+   *  <p>
+   *  If `indices` contains duplicates, then their updates are accumulated (summed).
+   *  <p>
+   *  <b>WARNING</b>: The order in which updates are applied is nondeterministic, so the
+   *  output will be nondeterministic if `indices` contains duplicates -- because
+   *  of some numerical approximation issues, numbers summed in different order
+   *  may yield different results.
+   *  <p>
+   *  `indices` is an integer tensor containing indices into a new tensor of shape
+   *  `shape`.  The last dimension of `indices` can be at most the rank of `shape`:
+   *  <p>
+   *      indices.shape[-1] <= shape.rank
+   *  <p>
+   *  The last dimension of `indices` corresponds to indices into elements
+   *  (if `indices.shape[-1] = shape.rank`) or slices
+   *  (if `indices.shape[-1] < shape.rank`) along dimension `indices.shape[-1]` of
+   *  `shape`.  `updates` is a tensor with shape
+   *  <p>
+   *      indices.shape[:-1] + shape[indices.shape[-1]:]
+   *  <p>
+   *  The simplest form of scatter is to insert individual elements in a tensor by
+   *  index. For example, say we want to insert 4 scattered elements in a rank-1
+   *  tensor with 8 elements.
+   *  <p>
+   *  <div style="width:70%; margin:auto; margin-bottom:10px; margin-top:20px;">
+   *  <img style="width:100%" src="https://www.tensorflow.org/images/ScatterNd1.png" alt>
+   *  </div>
+   *  <p>
+   *  In Python, this scatter operation would look like this:
+   *  <p>
+   *      >>> indices = tf.constant([[4], [3], [1], [7]])
+   *      >>> updates = tf.constant([9, 10, 11, 12])
+   *      >>> tensor = tf.ones([8], dtype=tf.int32)
+   *      >>> print(tf.tensor_scatter_nd_update(tensor, indices, updates))
+   *      tf.Tensor([ 1 11  1 10  9  1  1 12], shape=(8,), dtype=int32)
+   *  <p>
+   *  We can also, insert entire slices of a higher rank tensor all at once. For
+   *  example, if we wanted to insert two slices in the first dimension of a
+   *  rank-3 tensor with two matrices of new values.
+   *  <p>
+   *  In Python, this scatter operation would look like this:
+   *  <p>
+   *      >>> indices = tf.constant([[0], [2]])
+   *      >>> updates = tf.constant([[[5, 5, 5, 5], [6, 6, 6, 6],
+   *      ...                         [7, 7, 7, 7], [8, 8, 8, 8]],
+   *      ...                        [[5, 5, 5, 5], [6, 6, 6, 6],
+   *      ...                         [7, 7, 7, 7], [8, 8, 8, 8]]])
+   *      >>> tensor = tf.ones([4, 4, 4], dtype=tf.int32)
+   *      >>> print(tf.tensor_scatter_nd_update(tensor, indices, updates).numpy())
+   *      [[[5 5 5 5]
+   *        [6 6 6 6]
+   *        [7 7 7 7]
+   *        [8 8 8 8]]
+   *       [[1 1 1 1]
+   *        [1 1 1 1]
+   *        [1 1 1 1]
+   *        [1 1 1 1]]
+   *       [[5 5 5 5]
+   *        [6 6 6 6]
+   *        [7 7 7 7]
+   *        [8 8 8 8]]
+   *       [[1 1 1 1]
+   *        [1 1 1 1]
+   *        [1 1 1 1]
+   *        [1 1 1 1]]]
+   *  <p>
+   *  Note that on CPU, if an out of bound index is found, an error is returned.
+   *  On GPU, if an out of bound index is found, the index is ignored.
+   *
+   * @param <T> data type for {@code output()} output
+   * @param tensor Tensor to copy/update.
+   * @param indices Index tensor.
+   * @param updates Updates to scatter into output.
+   * @return a new instance of TensorScatterNdUpdate
+   */
+  public <T extends TType, U extends TNumber> TensorScatterNdUpdate<T> tensorScatterNdUpdate(
+      Operand<T> tensor, Operand<U> indices, Operand<T> updates) {
+    return TensorScatterNdUpdate.create(scope, tensor, indices, updates);
+  }
+
+  /**
+   * Assign `value` to the sliced l-value reference of `input`.
+   *  <p>
+   *  The values of `value` are assigned to the positions in the tensor `input` that
+   *  are selected by the slice parameters. The slice parameters `begin` `end`
+   *  `strides` etc. work exactly as in `StridedSlice`.
+   *  <p>
+   *  NOTE this op currently does not support broadcasting and so `value`'s shape
+   *  must be exactly the shape produced by the slice of `input`.
+   *
+   * @param <T> data type for {@code output()} output
+   * @param input
+   * @param begin
+   * @param end
+   * @param strides
+   * @param value
+   * @param options carries optional attributes values
+   * @return a new instance of TensorStridedSliceUpdate
+   */
+  public <T extends TType, U extends TNumber> TensorStridedSliceUpdate<T> tensorStridedSliceUpdate(
+      Operand<T> input, Operand<U> begin, Operand<U> end, Operand<U> strides, Operand<T> value,
+      TensorStridedSliceUpdate.Options... options) {
+    return TensorStridedSliceUpdate.create(scope, input, begin, end, strides, value, options);
+  }
+
+  /**
+   * Constructs a tensor by tiling a given tensor.
+   *  <p>
+   *  This operation creates a new tensor by replicating `input` `multiples` times.
+   *  The output tensor's i'th dimension has `input.dims(i) * multiples[i]` elements,
+   *  and the values of `input` are replicated `multiples[i]` times along the 'i'th
+   *  dimension. For example, tiling `[a b c d]` by `[2]` produces
+   *  `[a b c d a b c d]`.
+   *  <p>
+   *  >>> a = tf.constant([[1,2,3],[4,5,6]], tf.int32)
+   *  >>> b = tf.constant([1,2], tf.int32)
+   *  >>> tf.tile(a, b)
+   *  <tf.Tensor: shape=(2, 6), dtype=int32, numpy=
+   *  array([[1, 2, 3, 1, 2, 3],
+   *         [4, 5, 6, 4, 5, 6]], dtype=int32)>
+   *  >>> c = tf.constant([2,1], tf.int32)
+   *  >>> tf.tile(a, c)
+   *  <tf.Tensor: shape=(4, 3), dtype=int32, numpy=
+   *  array([[1, 2, 3],
+   *         [4, 5, 6],
+   *         [1, 2, 3],
+   *         [4, 5, 6]], dtype=int32)>
+   *  >>> d = tf.constant([2,2], tf.int32)
+   *  >>> tf.tile(a, d)
+   *  <tf.Tensor: shape=(4, 6), dtype=int32, numpy=
+   *  array([[1, 2, 3, 1, 2, 3],
+   *         [4, 5, 6, 4, 5, 6],
+   *         [1, 2, 3, 1, 2, 3],
+   *         [4, 5, 6, 4, 5, 6]], dtype=int32)>
+   *
+   * @param <T> data type for {@code output()} output
+   * @param input 1-D or higher.
+   * @param multiples 1-D. Length must be the same as the number of dimensions in `input`
+   * @return a new instance of Tile
+   */
+  public <T extends TType, U extends TNumber> Tile<T> tile(Operand<T> input, Operand<U> multiples) {
+    return Tile.create(scope, input, multiples);
+  }
+
+  /**
+   * Provides the time since epoch in seconds.
+   *  <p>
+   *  Returns the timestamp as a `float64` for seconds since the Unix epoch.
+   *  <p>
+   *  Note: the timestamp is computed when the op is executed, not when it is added
+   *  to the graph.
+   *
+   * @return a new instance of Timestamp
+   */
+  public Timestamp timestamp() {
+    return Timestamp.create(scope);
+  }
+
+  /**
+   * Perform batches of RPC requests.
+   *  <p>
+   *  This op asynchronously performs either a single RPC request, or a batch
+   *  of requests.  RPC requests are defined by three main parameters:
+   *  <p>
+   *    - `address` (the host+port or BNS address of the request)
+   *    - `method` (the method name for the request)
+   *    - `request` (the serialized proto string, or vector of strings,
+   *       of the RPC request argument).
+   *  <p>
+   *  For example, if you have an RPC service running on port localhost:2345,
+   *  and its interface is configured with the following proto declaration:
+   *  <pre>{@code
+   *  service MyService {
+   *    rpc MyMethod(MyRequestProto) returns (MyResponseProto) {
+   *    }
+   *  };
+   *  }</pre>
+   *  then call this op with arguments:
+   *  <pre>{@code
+   *  address = "localhost:2345"
+   *  method = "MyService/MyMethod"
+   *  }</pre>
+   *  The `request` tensor is a string tensor representing serialized `MyRequestProto`
+   *  strings; and the output string tensor `response` will have the same shape
+   *  and contain (upon successful completion) corresponding serialized
+   *  `MyResponseProto` strings.
+   *  <p>
+   *  For example, to send a single, empty, `MyRequestProto`, call
+   *  this op with `request = ""`.  To send 5 <b>parallel</b> empty requests,
+   *  call this op with `request = ["", "", "", "", ""]`.
+   *  <p>
+   *  More generally, one can create a batch of `MyRequestProto` serialized protos
+   *  from regular batched tensors using the `encode_proto` op, and convert
+   *  the response `MyResponseProto` serialized protos to batched tensors
+   *  using the `decode_proto` op.
+   *  <p>
+   *  <b>NOTE</b> Working with serialized proto strings is faster than instantiating
+   *  actual proto objects in memory, so no performance degradation is expected
+   *  compared to writing custom kernels for this workflow.
+   *  <p>
+   *  Unlike the standard `Rpc` op, if the connection fails or the remote worker
+   *  returns an error status, this op does <b>not</b> reraise the exception.
+   *  Instead, the `status_code` and `status_message` entry for the corresponding RPC
+   *  call is set with the error returned from the RPC call.  The `response` tensor
+   *  will contain valid response values for those minibatch entries whose RPCs did
+   *  not fail; the rest of the entries will have empty strings.
+   *
+   * @param address `0-D` or `1-D`.  The address (i.e. host_name:port) of the RPC server.
+   *  If this tensor has more than 1 element, then multiple parallel rpc requests
+   *  are sent.  This argument broadcasts with `method` and `request`.
+   * @param method `0-D` or `1-D`.  The method address on the RPC server.
+   *  If this tensor has more than 1 element, then multiple parallel rpc requests
+   *  are sent.  This argument broadcasts with `address` and `request`.
+   * @param request `0-D` or `1-D`.  Serialized proto strings: the rpc request argument.
+   *  If this tensor has more than 1 element, then multiple parallel rpc requests
+   *  are sent.  This argument broadcasts with `address` and `method`.
+   * @param options carries optional attributes values
+   * @return a new instance of TryRpc
+   */
+  public TryRpc tryRpc(Operand<TString> address, Operand<TString> method, Operand<TString> request,
+      TryRpc.Options... options) {
+    return TryRpc.create(scope, address, method, request, options);
+  }
+
+  /**
+   * Reverses the operation of Batch for a single output Tensor.
+   *  <p>
+   *  An instance of Unbatch either receives an empty batched_tensor, in which case it
+   *  asynchronously waits until the values become available from a concurrently
+   *  running instance of Unbatch with the same container and shared_name, or receives
+   *  a non-empty batched_tensor in which case it finalizes all other concurrently
+   *  running instances and outputs its own element from the batch.
+   *  <p>
+   *  batched_tensor: The possibly transformed output of Batch. The size of the first
+   *   dimension should remain unchanged by the transformations for the operation to
+   *   work.
+   *  batch_index: The matching batch_index obtained from Batch.
+   *  id: The id scalar emitted by Batch.
+   *  unbatched_tensor: The Tensor corresponding to this execution.
+   *  timeout_micros: Maximum amount of time (in microseconds) to wait to receive the
+   *   batched input tensor associated with a given invocation of the op.
+   *  container: Container to control resource sharing.
+   *  shared_name: Instances of Unbatch with the same container and shared_name are
+   *   assumed to possibly belong to the same batch. If left empty, the op name will
+   *   be used as the shared name.
+   *
+   * @param <T> data type for {@code unbatchedTensor()} output
+   * @param batchedTensor
+   * @param batchIndex
+   * @param id
+   * @param timeoutMicros
+   * @param options carries optional attributes values
+   * @return a new instance of Unbatch
+   */
+  public <T extends TType> Unbatch<T> unbatch(Operand<T> batchedTensor, Operand<TInt64> batchIndex,
+      Operand<TInt64> id, Long timeoutMicros, Unbatch.Options... options) {
+    return Unbatch.create(scope, batchedTensor, batchIndex, id, timeoutMicros, options);
+  }
+
+  /**
+   * Gradient of Unbatch.
+   *  <p>
+   *  Acts like Batch but using the given batch_index index of batching things as they
+   *  become available. This ensures that the gradients are propagated back in the
+   *  same session which did the forward pass.
+   *  <p>
+   *  original_input: The input to the Unbatch operation this is the gradient of.
+   *  batch_index: The batch_index given to the Unbatch operation this is the gradient
+   *  of.
+   *  grad: The downstream gradient.
+   *  id: The id scalar emitted by Batch.
+   *  batched_grad: The return value, either an empty tensor or the batched gradient.
+   *  container: Container to control resource sharing.
+   *  shared_name: Instances of UnbatchGrad with the same container and shared_name
+   *   are assumed to possibly belong to the same batch. If left empty, the op name
+   *   will be used as the shared name.
+   *
+   * @param <T> data type for {@code batchedGrad()} output
+   * @param originalInput
+   * @param batchIndex
+   * @param grad
+   * @param id
+   * @param options carries optional attributes values
+   * @return a new instance of UnbatchGrad
+   */
+  public <T extends TType> UnbatchGrad<T> unbatchGrad(Operand<T> originalInput,
+      Operand<TInt64> batchIndex, Operand<T> grad, Operand<TInt64> id,
+      UnbatchGrad.Options... options) {
+    return UnbatchGrad.create(scope, originalInput, batchIndex, grad, id, options);
+  }
+
+  /**
+   * Finds unique elements along an axis of a tensor.
+   *  <p>
+   *  This operation either returns a tensor `y` containing unique elements
+   *  along the `axis` of a tensor. The returned unique elements is sorted
+   *  in the same order as they occur along `axis` in `x`.
+   *  This operation also returns a tensor `idx` that is the same size as
+   *  the number of the elements in `x` along the `axis` dimension. It
+   *  contains the index in the unique output `y`.
+   *  In other words, for an `1-D` tensor `x` with `axis = None:
+   *  <p>
+   *  `y[idx[i]] = x[i] for i in [0, 1,...,rank(x) - 1]`
+   *  <p>
+   *  For example:
+   *  <pre>{@code
+   *  # tensor 'x' is [1, 1, 2, 4, 4, 4, 7, 8, 8]
+   *  y, idx = unique(x)
+   *  y ==> [1, 2, 4, 7, 8]
+   *  idx ==> [0, 0, 1, 2, 2, 2, 3, 4, 4]
+   *  }</pre>
+   *  For an `2-D` tensor `x` with `axis = 0`:
+   *  <pre>{@code
+   *  # tensor 'x' is [[1, 0, 0],
+   *  #                [1, 0, 0],
+   *  #                [2, 0, 0]]
+   *  y, idx = unique(x, axis=0)
+   *  y ==> [[1, 0, 0],
+   *         [2, 0, 0]]
+   *  idx ==> [0, 0, 1]
+   *  }</pre>
+   *  For an `2-D` tensor `x` with `axis = 1`:
+   *  <pre>{@code
+   *  # tensor 'x' is [[1, 0, 0],
+   *  #                [1, 0, 0],
+   *  #                [2, 0, 0]]
+   *  y, idx = unique(x, axis=1)
+   *  y ==> [[1, 0],
+   *         [1, 0],
+   *         [2, 0]]
+   *  idx ==> [0, 1, 1]
+   *  }</pre>
+   *
+   * @param <T> data type for {@code y()} output
+   * @param <V> data type for {@code idx()} output
+   * @param x A `Tensor`.
+   * @param axis A `Tensor` of type `int32` (default: None). The axis of the Tensor to
+   *  find the unique elements.
+   * @return a new instance of Unique
+   */
+  public <T extends TType, U extends TNumber> Unique<T, TInt32> unique(Operand<T> x,
+      Operand<U> axis) {
+    return Unique.create(scope, x, axis);
+  }
+
+  /**
+   * Finds unique elements along an axis of a tensor.
+   *  <p>
+   *  This operation either returns a tensor `y` containing unique elements
+   *  along the `axis` of a tensor. The returned unique elements is sorted
+   *  in the same order as they occur along `axis` in `x`.
+   *  This operation also returns a tensor `idx` that is the same size as
+   *  the number of the elements in `x` along the `axis` dimension. It
+   *  contains the index in the unique output `y`.
+   *  In other words, for an `1-D` tensor `x` with `axis = None:
+   *  <p>
+   *  `y[idx[i]] = x[i] for i in [0, 1,...,rank(x) - 1]`
+   *  <p>
+   *  For example:
+   *  <pre>{@code
+   *  # tensor 'x' is [1, 1, 2, 4, 4, 4, 7, 8, 8]
+   *  y, idx = unique(x)
+   *  y ==> [1, 2, 4, 7, 8]
+   *  idx ==> [0, 0, 1, 2, 2, 2, 3, 4, 4]
+   *  }</pre>
+   *  For an `2-D` tensor `x` with `axis = 0`:
+   *  <pre>{@code
+   *  # tensor 'x' is [[1, 0, 0],
+   *  #                [1, 0, 0],
+   *  #                [2, 0, 0]]
+   *  y, idx = unique(x, axis=0)
+   *  y ==> [[1, 0, 0],
+   *         [2, 0, 0]]
+   *  idx ==> [0, 0, 1]
+   *  }</pre>
+   *  For an `2-D` tensor `x` with `axis = 1`:
+   *  <pre>{@code
+   *  # tensor 'x' is [[1, 0, 0],
+   *  #                [1, 0, 0],
+   *  #                [2, 0, 0]]
+   *  y, idx = unique(x, axis=1)
+   *  y ==> [[1, 0],
+   *         [1, 0],
+   *         [2, 0]]
+   *  idx ==> [0, 1, 1]
+   *  }</pre>
+   *
+   * @param <T> data type for {@code y()} output
+   * @param <V> data type for {@code idx()} output
+   * @param x A `Tensor`.
+   * @param axis A `Tensor` of type `int32` (default: None). The axis of the Tensor to
+   *  find the unique elements.
+   * @param outIdx
+   * @return a new instance of Unique
+   */
+  public <T extends TType, V extends TNumber, U extends TNumber> Unique<T, V> unique(Operand<T> x,
+      Operand<U> axis, Class<V> outIdx) {
+    return Unique.create(scope, x, axis, outIdx);
+  }
+
+  /**
+   * Finds unique elements along an axis of a tensor.
+   *  <p>
+   *  This operation either returns a tensor `y` containing unique elements
+   *  along the `axis` of a tensor. The returned unique elements is sorted
+   *  in the same order as they occur along `axis` in `x`.
+   *  This operation also returns a tensor `idx` and a tensor `count`
+   *  that are the same size as the number of the elements in `x` along the
+   *  `axis` dimension. The `idx` contains the index in the unique output `y`
+   *  and the `count` contains the count in the unique output `y`.
+   *  In other words, for an `1-D` tensor `x` with `axis = None:
+   *  <p>
+   *  `y[idx[i]] = x[i] for i in [0, 1,...,rank(x) - 1]`
+   *  <p>
+   *  For example:
+   *  <pre>{@code
+   *  # tensor 'x' is [1, 1, 2, 4, 4, 4, 7, 8, 8]
+   *  y, idx, count = unique_with_counts(x)
+   *  y ==> [1, 2, 4, 7, 8]
+   *  idx ==> [0, 0, 1, 2, 2, 2, 3, 4, 4]
+   *  count ==> [2, 1, 3, 1, 2]
+   *  }</pre>
+   *  For an `2-D` tensor `x` with `axis = 0`:
+   *  <pre>{@code
+   *  # tensor 'x' is [[1, 0, 0],
+   *  #                [1, 0, 0],
+   *  #                [2, 0, 0]]
+   *  y, idx, count = unique_with_counts(x, axis=0)
+   *  y ==> [[1, 0, 0],
+   *         [2, 0, 0]]
+   *  idx ==> [0, 0, 1]
+   *  count ==> [2, 1]
+   *  }</pre>
+   *  For an `2-D` tensor `x` with `axis = 1`:
+   *  <pre>{@code
+   *  # tensor 'x' is [[1, 0, 0],
+   *  #                [1, 0, 0],
+   *  #                [2, 0, 0]]
+   *  y, idx, count = unique_with_counts(x, axis=1)
+   *  y ==> [[1, 0],
+   *         [1, 0],
+   *         [2, 0]]
+   *  idx ==> [0, 1, 1]
+   *  count ==> [1, 2]
+   *  }</pre>
+   *
+   * @param <T> data type for {@code y()} output
+   * @param <V> data type for {@code idx()} output
+   * @param x A `Tensor`.
+   * @param axis A `Tensor` of type `int32` (default: None). The axis of the Tensor to
+   *  find the unique elements.
+   * @return a new instance of UniqueWithCounts
+   */
+  public <T extends TType, U extends TNumber> UniqueWithCounts<T, TInt32> uniqueWithCounts(
+      Operand<T> x, Operand<U> axis) {
+    return UniqueWithCounts.create(scope, x, axis);
+  }
+
+  /**
+   * Finds unique elements along an axis of a tensor.
+   *  <p>
+   *  This operation either returns a tensor `y` containing unique elements
+   *  along the `axis` of a tensor. The returned unique elements is sorted
+   *  in the same order as they occur along `axis` in `x`.
+   *  This operation also returns a tensor `idx` and a tensor `count`
+   *  that are the same size as the number of the elements in `x` along the
+   *  `axis` dimension. The `idx` contains the index in the unique output `y`
+   *  and the `count` contains the count in the unique output `y`.
+   *  In other words, for an `1-D` tensor `x` with `axis = None:
+   *  <p>
+   *  `y[idx[i]] = x[i] for i in [0, 1,...,rank(x) - 1]`
+   *  <p>
+   *  For example:
+   *  <pre>{@code
+   *  # tensor 'x' is [1, 1, 2, 4, 4, 4, 7, 8, 8]
+   *  y, idx, count = unique_with_counts(x)
+   *  y ==> [1, 2, 4, 7, 8]
+   *  idx ==> [0, 0, 1, 2, 2, 2, 3, 4, 4]
+   *  count ==> [2, 1, 3, 1, 2]
+   *  }</pre>
+   *  For an `2-D` tensor `x` with `axis = 0`:
+   *  <pre>{@code
+   *  # tensor 'x' is [[1, 0, 0],
+   *  #                [1, 0, 0],
+   *  #                [2, 0, 0]]
+   *  y, idx, count = unique_with_counts(x, axis=0)
+   *  y ==> [[1, 0, 0],
+   *         [2, 0, 0]]
+   *  idx ==> [0, 0, 1]
+   *  count ==> [2, 1]
+   *  }</pre>
+   *  For an `2-D` tensor `x` with `axis = 1`:
+   *  <pre>{@code
+   *  # tensor 'x' is [[1, 0, 0],
+   *  #                [1, 0, 0],
+   *  #                [2, 0, 0]]
+   *  y, idx, count = unique_with_counts(x, axis=1)
+   *  y ==> [[1, 0],
+   *         [1, 0],
+   *         [2, 0]]
+   *  idx ==> [0, 1, 1]
+   *  count ==> [1, 2]
+   *  }</pre>
+   *
+   * @param <T> data type for {@code y()} output
+   * @param <V> data type for {@code idx()} output
+   * @param x A `Tensor`.
+   * @param axis A `Tensor` of type `int32` (default: None). The axis of the Tensor to
+   *  find the unique elements.
+   * @param outIdx
+   * @return a new instance of UniqueWithCounts
+   */
+  public <T extends TType, V extends TNumber, U extends TNumber> UniqueWithCounts<T, V> uniqueWithCounts(
+      Operand<T> x, Operand<U> axis, Class<V> outIdx) {
+    return UniqueWithCounts.create(scope, x, axis, outIdx);
+  }
+
+  /**
+   * Converts an array of flat indices into a tuple of coordinate arrays.
+   *  <p>
+   *
+   *  Example:
+   *  <pre>{@code
+   *  y = tf.unravel_index(indices=[2, 5, 7], dims=[3, 3])
+   *  # 'dims' represent a hypothetical (3, 3) tensor of indices:
+   *  # [[0, 1, *2*],
+   *  #  [3, 4, *5*],
+   *  #  [6, *7*, 8]]
+   *  # For each entry from 'indices', this operation returns
+   *  # its coordinates (marked with '*'), such as
+   *  # 2 ==> (0, 2)
+   *  # 5 ==> (1, 2)
+   *  # 7 ==> (2, 1)
+   *  y ==> [[0, 1, 2], [2, 2, 1]]
+   *  }</pre>
+   *
+   * @compatibility(numpy) Equivalent to np.unravel_index
+   * @end_compatibility
+   * @param <T> data type for {@code output()} output
+   * @param indices An 0-D or 1-D `int` Tensor whose elements are indices into the
+   *  flattened version of an array of dimensions dims.
+   * @param dims An 1-D `int` Tensor. The shape of the array to use for unraveling
+   *  indices.
+   * @return a new instance of UnravelIndex
+   */
+  public <T extends TNumber> UnravelIndex<T> unravelIndex(Operand<T> indices, Operand<T> dims) {
+    return UnravelIndex.create(scope, indices, dims);
+  }
+
+  /**
+   * Unpacks a given dimension of a rank-`R` tensor into `num` rank-`(R-1)` tensors.
+   *  <p>
+   *  Unpacks `num` tensors from `value` by chipping it along the `axis` dimension.
+   *  For example, given a tensor of shape `(A, B, C, D)`;
+   *  <p>
+   *  If `axis == 0` then the i'th tensor in `output` is the slice `value[i, :, :, :]`
+   *    and each tensor in `output` will have shape `(B, C, D)`. (Note that the
+   *    dimension unpacked along is gone, unlike `split`).
+   *  <p>
+   *  If `axis == 1` then the i'th tensor in `output` is the slice `value[:, i, :, :]`
+   *    and each tensor in `output` will have shape `(A, C, D)`.
+   *  Etc.
+   *  <p>
+   *  This is the opposite of `pack`.
+   *
+   * @param <T> data type for {@code output()} output
+   * @param value 1-D or higher, with `axis` dimension size equal to `num`.
+   * @param num
+   * @param options carries optional attributes values
+   * @return a new instance of Unstack
+   */
+  public <T extends TType> Unstack<T> unstack(Operand<T> value, Long num,
+      Unstack.Options... options) {
+    return Unstack.create(scope, value, num, options);
+  }
+
+  /**
+   * Op is similar to a lightweight Dequeue.
+   *  <p>
+   *  The basic functionality is similar to dequeue with many fewer
+   *  capabilities and options.  This Op is optimized for performance.
+   *
+   * @param dtypes
+   * @param options carries optional attributes values
+   * @return a new instance of Unstage
+   */
+  public Unstage unstage(List<Class<? extends TType>> dtypes, Unstage.Options... options) {
+    return Unstage.create(scope, dtypes, options);
+  }
+
+  /**
+   * Creates a handle to a Variable resource.
+   *
+   * @param dtype the type of this variable. Must agree with the dtypes
+   *  of all ops using this variable.
+   * @param shape The (possibly partially specified) shape of this variable.
+   * @param options carries optional attributes values
+   * @return a new instance of VarHandleOp
+   */
+  public <T extends TType> VarHandleOp varHandleOp(Class<T> dtype, Shape shape,
+      VarHandleOp.Options... options) {
+    return VarHandleOp.create(scope, dtype, shape, options);
+  }
+
+  /**
+   * Checks whether a resource handle-based variable has been initialized.
+   *
+   * @param resource the input resource handle.
+   * @return a new instance of VarIsInitializedOp
+   */
+  public VarIsInitializedOp varIsInitializedOp(Operand<?> resource) {
+    return VarIsInitializedOp.create(scope, resource);
+  }
+
+  /**
+   * Factory method to create a new Variable with it's initializer.
+   *  <p>
+   *  Only supported on Graph sessions as the {@link org.tensorflow.op.core.Assign} op
+   *  does not work in an EagerSession.
+   *
+   * @param scope current scope
+   * @param init The op to use to initialise this variable.
+   * @param options carries optional attributes values
+   * @return a new instance of Variable
+   */
+  public <T extends TType> Variable<T> variable(Operand<T> init, Variable.Options... options) {
+    return Helpers.createVariableWithInit(scope, init, options);
+  }
+
+  /**
+   * Holds state in the form of a tensor that persists across steps.
+   *  <p>
+   *  Outputs a ref to the tensor state so it may be read or modified.
+   *  TODO(zhifengc/mrry): Adds a pointer to a more detail document
+   *  about sharing states in tensorflow.
+   *
+   * @param <T> data type for {@code ref()} output
+   * @param shape The shape of the variable tensor.
+   * @param dtype The type of elements in the variable tensor.
+   * @param options carries optional attributes values
+   * @return a new instance of Variable
+   */
+  public <T extends TType> Variable<T> variable(Shape shape, Class<T> dtype,
+      Variable.Options... options) {
+    return Variable.create(scope, shape, dtype, options);
+  }
+
+  /**
+   * Returns the shape of the variable pointed to by `resource`.
+   *  <p>
+   *  This operation returns a 1-D integer tensor representing the shape of `input`.
+   *  <p>
+   *  For example:
+   *  <pre>{@code
+   *  # 't' is [[[1, 1, 1], [2, 2, 2]], [[3, 3, 3], [4, 4, 4]]]
+   *  shape(t) ==> [2, 2, 3]
+   *  }</pre>
+   *
+   * @param <T> data type for {@code output()} output
+   * @param input
+   * @return a new instance of VariableShape
+   */
+  public VariableShape<TInt32> variableShape(Operand<?> input) {
+    return VariableShape.create(scope, input);
+  }
+
+  /**
+   * Returns the shape of the variable pointed to by `resource`.
+   *  <p>
+   *  This operation returns a 1-D integer tensor representing the shape of `input`.
+   *  <p>
+   *  For example:
+   *  <pre>{@code
+   *  # 't' is [[[1, 1, 1], [2, 2, 2]], [[3, 3, 3], [4, 4, 4]]]
+   *  shape(t) ==> [2, 2, 3]
+   *  }</pre>
+   *
+   * @param <T> data type for {@code output()} output
+   * @param input
+   * @param outType
+   * @return a new instance of VariableShape
+   */
+  public <T extends TNumber> VariableShape<T> variableShape(Operand<?> input, Class<T> outType) {
+    return VariableShape.create(scope, input, outType);
+  }
+
+  /**
+   * Returns locations of nonzero / true values in a tensor.
+   *  <p>
+   *  This operation returns the coordinates of true elements in `condition`. The
+   *  coordinates are returned in a 2-D tensor where the first dimension (rows)
+   *  represents the number of true elements, and the second dimension (columns)
+   *  represents the coordinates of the true elements. Keep in mind, the shape of
+   *  the output tensor can vary depending on how many true values there are in
+   *  `condition`. Indices are output in row-major order.
+   *  <p>
+   *  For example:
+   *  <pre>{@code
+   *  # 'input' tensor is [[True, False]
+   *  #                    [True, False]]
+   *  # 'input' has two true values, so output has two coordinates.
+   *  # 'input' has rank of 2, so coordinates have two indices.
+   *  where(input) ==> [[0, 0],
+   *                    [1, 0]]
+   *
+   *  # `condition` tensor is [[[True, False]
+   *  #                     [True, False]]
+   *  #                    [[False, True]
+   *  #                     [False, True]]
+   *  #                    [[False, False]
+   *  #                     [False, True]]]
+   *  # 'input' has 5 true values, so output has 5 coordinates.
+   *  # 'input' has rank of 3, so coordinates have three indices.
+   *  where(input) ==> [[0, 0, 0],
+   *                    [0, 1, 0],
+   *                    [1, 0, 1],
+   *                    [1, 1, 1],
+   *                    [2, 1, 1]]
+   *
+   *  # `condition` tensor is [[[1.5,  0.0]
+   *  #                     [-0.5, 0.0]]
+   *  #                    [[0.0,  0.25]
+   *  #                     [0.0,  0.75]]
+   *  #                    [[0.0,  0.0]
+   *  #                     [0.0,  0.01]]]
+   *  # 'input' has 5 nonzero values, so output has 5 coordinates.
+   *  # 'input' has rank of 3, so coordinates have three indices.
+   *  where(input) ==> [[0, 0, 0],
+   *                    [0, 1, 0],
+   *                    [1, 0, 1],
+   *                    [1, 1, 1],
+   *                    [2, 1, 1]]
+   *
+   *  # `condition` tensor is [[[1.5 + 0.0j, 0.0  + 0.0j]
+   *  #                     [0.0 + 0.5j, 0.0  + 0.0j]]
+   *  #                    [[0.0 + 0.0j, 0.25 + 1.5j]
+   *  #                     [0.0 + 0.0j, 0.75 + 0.0j]]
+   *  #                    [[0.0 + 0.0j, 0.0  + 0.0j]
+   *  #                     [0.0 + 0.0j, 0.01 + 0.0j]]]
+   *  # 'input' has 5 nonzero magnitude values, so output has 5 coordinates.
+   *  # 'input' has rank of 3, so coordinates have three indices.
+   *  where(input) ==> [[0, 0, 0],
+   *                    [0, 1, 0],
+   *                    [1, 0, 1],
+   *                    [1, 1, 1],
+   *                    [2, 1, 1]]
+   *  }</pre>
+   *
+   * @param condition
+   * @return a new instance of Where
+   */
+  public <T extends TType> Where where(Operand<T> condition) {
+    return Where.create(scope, condition);
+  }
+
+  /**
+   * An op used by XLA SPMD partitioner to switch from automatic partitioning to
+   *  <p>
+   *  manual partitioning. It annotates the input (full-shape, to be automatically
+   *  partitioned) with the same sharding used by manual partitioning, and outputs a
+   *  shard-shaped tensor to be consumed by later manually-partitioned ops. If the
+   *  shape is not evenly partitionable, the padding region will be masked with 0s.
+   *
+   * @param <T> data type for {@code output()} output
+   * @param input
+   * @param manualSharding
+   * @return a new instance of XlaSpmdFullToShardShape
+   */
+  public <T extends TType> XlaSpmdFullToShardShape<T> xlaSpmdFullToShardShape(Operand<T> input,
+      String manualSharding) {
+    return XlaSpmdFullToShardShape.create(scope, input, manualSharding);
+  }
+
+  /**
+   * An op used by XLA SPMD partitioner to switch from manual partitioning to
+   *  <p>
+   *  automatic partitioning. It converts the shard-shaped, manually partitioned input
+   *  into full-shaped tensor to be partitioned automatically with the same sharding
+   *  used by manual partitioning.
+   *
+   * @param <T> data type for {@code output()} output
+   * @param input
+   * @param manualSharding
+   * @param fullShape
+   * @return a new instance of XlaSpmdShardToFullShape
+   */
+  public <T extends TType> XlaSpmdShardToFullShape<T> xlaSpmdShardToFullShape(Operand<T> input,
+      String manualSharding, Shape fullShape) {
+    return XlaSpmdShardToFullShape.create(scope, input, manualSharding, fullShape);
+  }
+
+  /**
+   * Creates a zeroed tensor given its type and shape.
+   *
+   * @param scope is a scope used to add the underlying operation
+   * @param dims a 1-D operand that represents the shape of the output tensor
+   * @param type the output tensor datatype
+   * @return a constant tensor initialized with zeros
+   * @throws IllegalArgumentException if the tensor type or shape cannot be initialized with zeros.
+   */
+  public <T extends TType, U extends TNumber> Zeros<T> zeros(Operand<U> dims, Class<T> type) {
+    return Zeros.create(scope, dims, type);
+  }
+
+  /**
+   * Returns a tensor of zeros with the same shape and type as x.
+   *
+   * @param <T> data type for {@code y()} output
+   * @param x a tensor of type T.
+   * @return a new instance of ZerosLike
+   */
+  public <T extends TType> ZerosLike<T> zerosLike(Operand<T> x) {
+    return ZerosLike.create(scope, x);
+  }
+
+  /**
+   * Returns an API that builds operations with the provided name prefix.
+   *
+   * @see {@link Scope#withSubScope(String)}
+   */
+  public Ops withSubScope(String childScopeName) {
+    return new Ops(scope.withSubScope(childScopeName));
+  }
+
+  /**
+   * Returns an API that uses the provided name for an op.
+   *
+   * @see {@link Scope#withName(String)}
+   */
+  public Ops withName(String opName) {
+    return new Ops(scope.withName(opName));
+  }
+
+  /**
+   * Returns an API that adds operations to the graph with the provided control dependencies.
+   *
+   * @see {@link Scope#withControlDependencies(Iterable<Op<?>>)}
+   */
+  public Ops withControlDependencies(Iterable<Op> controls) {
+    return new Ops(scope.withControlDependencies(controls));
+  }
+
+  /**
+   * Returns the current {@link Scope scope} of this API
+   */
+  public final Scope scope() {
+    return scope;
+  }
+
+  /**
+   * Creates an API for building operations in the provided execution environment
+   */
+  public static Ops create(ExecutionEnvironment env) {
+    return new Ops(new Scope(env));
+  }
+
+  /**
+   * Creates an API for building operations in the default eager execution environment
+   *
+   * <p>Invoking this method is equivalent to {@code Ops.create(EagerSession.getDefault())}.
+   */
+  public static Ops create() {
+    return new Ops(new Scope(EagerSession.getDefault()));
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/QuantizationOps.java b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/QuantizationOps.java
new file mode 100644
index 00000000000..045ed3d4b03
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/QuantizationOps.java
@@ -0,0 +1,629 @@
+// Copyright 2020 The TensorFlow Authors. All Rights Reserved.
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+// ==============================================================================
+//
+// This class has been generated, DO NOT EDIT!
+//
+package org.tensorflow.op;
+
+import org.tensorflow.Operand;
+import org.tensorflow.op.quantization.Dequantize;
+import org.tensorflow.op.quantization.FakeQuantWithMinMaxArgs;
+import org.tensorflow.op.quantization.FakeQuantWithMinMaxArgsGradient;
+import org.tensorflow.op.quantization.FakeQuantWithMinMaxVars;
+import org.tensorflow.op.quantization.FakeQuantWithMinMaxVarsGradient;
+import org.tensorflow.op.quantization.FakeQuantWithMinMaxVarsPerChannel;
+import org.tensorflow.op.quantization.FakeQuantWithMinMaxVarsPerChannelGradient;
+import org.tensorflow.op.quantization.Quantize;
+import org.tensorflow.op.quantization.QuantizeAndDequantize;
+import org.tensorflow.op.quantization.QuantizeDownAndShrinkRange;
+import org.tensorflow.op.quantization.QuantizedConcat;
+import org.tensorflow.op.quantization.RequantizationRange;
+import org.tensorflow.op.quantization.Requantize;
+import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * An API for building {@code quantization} operations as {@link Op Op}s
+ *
+ * @see {@link Ops}
+ */
+public final class QuantizationOps {
+  private final Scope scope;
+
+  QuantizationOps(Scope scope) {
+    this.scope = scope;
+  }
+
+  /**
+   * Dequantize the 'input' tensor into a float or bfloat16 Tensor.
+   *  <p>
+   *  [min_range, max_range] are scalar floats that specify the range for
+   *  the output. The 'mode' attribute controls exactly which calculations are
+   *  used to convert the float values to their quantized equivalents.
+   *  <p>
+   *  In 'MIN_COMBINED' mode, each value of the tensor will undergo the following:
+   *  <pre>{@code
+   *  if T == qint8: in[i] += (range(T) + 1)/ 2.0
+   *  out[i] = min_range + (in[i]* (max_range - min_range) / range(T))
+   *  }</pre>
+   *  here `range(T) = numeric_limits<T>::max() - numeric_limits<T>::min()`
+   *  <p>
+   *  <i>MIN_COMBINED Mode Example</i>
+   *  <p>
+   *  If the input comes from a QuantizedRelu6, the output type is
+   *  quint8 (range of 0-255) but the possible range of QuantizedRelu6 is
+   *  0-6.  The min_range and max_range values are therefore 0.0 and 6.0.
+   *  Dequantize on quint8 will take each value, cast to float, and multiply
+   *  by 6 / 255.
+   *  Note that if quantizedtype is qint8, the operation will additionally add
+   *  each value by 128 prior to casting.
+   *  <p>
+   *  If the mode is 'MIN_FIRST', then this approach is used:
+   *  <pre>{@code
+   *  num_discrete_values = 1 << (# of bits in T)
+   *  range_adjust = num_discrete_values / (num_discrete_values - 1)
+   *  range = (range_max - range_min) * range_adjust
+   *  range_scale = range / num_discrete_values
+   *  const double offset_input = static_cast<double>(input) - lowest_quantized;
+   *  result = range_min + ((input - numeric_limits<T>::min()) * range_scale)
+   *  }</pre>
+   *  If the mode is `SCALED`, dequantization is performed by multiplying each
+   *  input value by a scaling_factor. (Thus an input of 0 always maps to 0.0).
+   *  <p>
+   *  The scaling_factor is determined from `min_range`, `max_range`, and
+   *  `narrow_range` in a way that is compatible with `QuantizeAndDequantize{V2|V3}`
+   *  and `QuantizeV2`, using the following algorithm:
+   *  <pre>{@code
+   *    const int min_expected_T = std::numeric_limits<T>::min() +
+   *      (narrow_range ? 1 : 0);
+   *    const int max_expected_T = std::numeric_limits<T>::max();
+   *    const float max_expected_T = std::numeric_limits<float>::max();
+   *
+   *    const float scale_factor =
+   *      (std::numeric_limits<T>::min() == 0) ? (max_range / max_expected_T)
+   *                                           : std::max(min_range / min_expected_T,
+   *                                                      max_range / max_expected_T);
+   *  }</pre>
+   *
+   * @param <U> data type for {@code output()} output
+   * @param input
+   * @param minRange The minimum scalar value possibly produced for the input.
+   * @param maxRange The maximum scalar value possibly produced for the input.
+   * @param options carries optional attributes values
+   * @return a new instance of Dequantize
+   */
+  public <T extends TType> Dequantize<TFloat32> dequantize(Operand<T> input,
+      Operand<TFloat32> minRange, Operand<TFloat32> maxRange, Dequantize.Options... options) {
+    return Dequantize.create(scope, input, minRange, maxRange, options);
+  }
+
+  /**
+   * Dequantize the 'input' tensor into a float or bfloat16 Tensor.
+   *  <p>
+   *  [min_range, max_range] are scalar floats that specify the range for
+   *  the output. The 'mode' attribute controls exactly which calculations are
+   *  used to convert the float values to their quantized equivalents.
+   *  <p>
+   *  In 'MIN_COMBINED' mode, each value of the tensor will undergo the following:
+   *  <pre>{@code
+   *  if T == qint8: in[i] += (range(T) + 1)/ 2.0
+   *  out[i] = min_range + (in[i]* (max_range - min_range) / range(T))
+   *  }</pre>
+   *  here `range(T) = numeric_limits<T>::max() - numeric_limits<T>::min()`
+   *  <p>
+   *  <i>MIN_COMBINED Mode Example</i>
+   *  <p>
+   *  If the input comes from a QuantizedRelu6, the output type is
+   *  quint8 (range of 0-255) but the possible range of QuantizedRelu6 is
+   *  0-6.  The min_range and max_range values are therefore 0.0 and 6.0.
+   *  Dequantize on quint8 will take each value, cast to float, and multiply
+   *  by 6 / 255.
+   *  Note that if quantizedtype is qint8, the operation will additionally add
+   *  each value by 128 prior to casting.
+   *  <p>
+   *  If the mode is 'MIN_FIRST', then this approach is used:
+   *  <pre>{@code
+   *  num_discrete_values = 1 << (# of bits in T)
+   *  range_adjust = num_discrete_values / (num_discrete_values - 1)
+   *  range = (range_max - range_min) * range_adjust
+   *  range_scale = range / num_discrete_values
+   *  const double offset_input = static_cast<double>(input) - lowest_quantized;
+   *  result = range_min + ((input - numeric_limits<T>::min()) * range_scale)
+   *  }</pre>
+   *  If the mode is `SCALED`, dequantization is performed by multiplying each
+   *  input value by a scaling_factor. (Thus an input of 0 always maps to 0.0).
+   *  <p>
+   *  The scaling_factor is determined from `min_range`, `max_range`, and
+   *  `narrow_range` in a way that is compatible with `QuantizeAndDequantize{V2|V3}`
+   *  and `QuantizeV2`, using the following algorithm:
+   *  <pre>{@code
+   *    const int min_expected_T = std::numeric_limits<T>::min() +
+   *      (narrow_range ? 1 : 0);
+   *    const int max_expected_T = std::numeric_limits<T>::max();
+   *    const float max_expected_T = std::numeric_limits<float>::max();
+   *
+   *    const float scale_factor =
+   *      (std::numeric_limits<T>::min() == 0) ? (max_range / max_expected_T)
+   *                                           : std::max(min_range / min_expected_T,
+   *                                                      max_range / max_expected_T);
+   *  }</pre>
+   *
+   * @param <U> data type for {@code output()} output
+   * @param input
+   * @param minRange The minimum scalar value possibly produced for the input.
+   * @param maxRange The maximum scalar value possibly produced for the input.
+   * @param dtype Type of the output tensor. Currently Dequantize supports float and bfloat16.
+   *  If 'dtype' is 'bfloat16', it only supports 'MIN_COMBINED' mode.
+   * @param options carries optional attributes values
+   * @return a new instance of Dequantize
+   */
+  public <U extends TNumber, T extends TType> Dequantize<U> dequantize(Operand<T> input,
+      Operand<TFloat32> minRange, Operand<TFloat32> maxRange, Class<U> dtype,
+      Dequantize.Options... options) {
+    return Dequantize.create(scope, input, minRange, maxRange, dtype, options);
+  }
+
+  /**
+   * Fake-quantize the 'inputs' tensor, type float to 'outputs' tensor of same type.
+   *  <p>
+   *  Attributes
+   *  <ul>
+   *  <li>
+   *  `[min; max]` define the clamping range for the `inputs` data.
+   *  </li>
+   *  <li>
+   *  `inputs` values are quantized into the quantization range (
+   *  `[0; 2^num_bits - 1]` when `narrow_range` is false and `[1; 2^num_bits - 1]`
+   *  when it is true) and then de-quantized and output as floats in `[min; max]`
+   *  interval.
+   *  </li>
+   *  <li>
+   *  `num_bits` is the bitwidth of the quantization; between 2 and 16, inclusive.
+   *  </li>
+   *  </ul>
+   *  Before quantization, `min` and `max` values are adjusted with the following
+   *  logic.
+   *  It is suggested to have `min <= 0 <= max`. If `0` is not in the range of values,
+   *  the behavior can be unexpected:
+   *  <ul>
+   *  <li>
+   *  If `0 < min < max`: `min_adj = 0` and `max_adj = max - min`.
+   *  </li>
+   *  <li>
+   *  If `min < max < 0`: `min_adj = min - max` and `max_adj = 0`.
+   *  </li>
+   *  <li>
+   *  If `min <= 0 <= max`: `scale = (max - min) / (2^num_bits - 1) `,
+   *  `min_adj = scale * round(min / scale)` and `max_adj = max + min_adj - min`.
+   *  </li>
+   *  </ul>
+   *  Quantization is called fake since the output is still in floating point.
+   *
+   * @param inputs
+   * @param options carries optional attributes values
+   * @return a new instance of FakeQuantWithMinMaxArgs
+   */
+  public FakeQuantWithMinMaxArgs fakeQuantWithMinMaxArgs(Operand<TFloat32> inputs,
+      FakeQuantWithMinMaxArgs.Options... options) {
+    return FakeQuantWithMinMaxArgs.create(scope, inputs, options);
+  }
+
+  /**
+   * Compute gradients for a FakeQuantWithMinMaxArgs operation.
+   *
+   * @param gradients Backpropagated gradients above the FakeQuantWithMinMaxArgs operation.
+   * @param inputs Values passed as inputs to the FakeQuantWithMinMaxArgs operation.
+   * @param options carries optional attributes values
+   * @return a new instance of FakeQuantWithMinMaxArgsGradient
+   */
+  public FakeQuantWithMinMaxArgsGradient fakeQuantWithMinMaxArgsGradient(
+      Operand<TFloat32> gradients, Operand<TFloat32> inputs,
+      FakeQuantWithMinMaxArgsGradient.Options... options) {
+    return FakeQuantWithMinMaxArgsGradient.create(scope, gradients, inputs, options);
+  }
+
+  /**
+   * Fake-quantize the 'inputs' tensor of type float via global float scalars
+   *  <p>
+   *  Fake-quantize the `inputs` tensor of type float via global float scalars
+   *  `min` and `max` to `outputs` tensor of same shape as `inputs`.
+   *  <p>
+   *  Attributes
+   *  <ul>
+   *  <li>
+   *  `[min; max]` define the clamping range for the `inputs` data.
+   *  </li>
+   *  <li>
+   *  `inputs` values are quantized into the quantization range (
+   *  `[0; 2^num_bits - 1]` when `narrow_range` is false and `[1; 2^num_bits - 1]`
+   *  when it is true) and then de-quantized and output as floats in `[min; max]`
+   *  interval.
+   *  </li>
+   *  <li>
+   *  `num_bits` is the bitwidth of the quantization; between 2 and 16, inclusive.
+   *  </li>
+   *  </ul>
+   *  Before quantization, `min` and `max` values are adjusted with the following
+   *  logic.
+   *  It is suggested to have `min <= 0 <= max`. If `0` is not in the range of values,
+   *  the behavior can be unexpected:
+   *  <ul>
+   *  <li>
+   *  If `0 < min < max`: `min_adj = 0` and `max_adj = max - min`.
+   *  </li>
+   *  <li>
+   *  If `min < max < 0`: `min_adj = min - max` and `max_adj = 0`.
+   *  </li>
+   *  <li>
+   *  If `min <= 0 <= max`: `scale = (max - min) / (2^num_bits - 1) `,
+   *  `min_adj = scale * round(min / scale)` and `max_adj = max + min_adj - min`.
+   *  </li>
+   *  </ul>
+   *  This operation has a gradient and thus allows for training `min` and `max`
+   *  values.
+   *
+   * @param inputs
+   * @param min
+   * @param max
+   * @param options carries optional attributes values
+   * @return a new instance of FakeQuantWithMinMaxVars
+   */
+  public FakeQuantWithMinMaxVars fakeQuantWithMinMaxVars(Operand<TFloat32> inputs,
+      Operand<TFloat32> min, Operand<TFloat32> max, FakeQuantWithMinMaxVars.Options... options) {
+    return FakeQuantWithMinMaxVars.create(scope, inputs, min, max, options);
+  }
+
+  /**
+   * Compute gradients for a FakeQuantWithMinMaxVars operation.
+   *
+   * @param gradients Backpropagated gradients above the FakeQuantWithMinMaxVars operation.
+   * @param inputs Values passed as inputs to the FakeQuantWithMinMaxVars operation.
+   *  min, max: Quantization interval, scalar floats.
+   * @param min
+   * @param max
+   * @param options carries optional attributes values
+   * @return a new instance of FakeQuantWithMinMaxVarsGradient
+   */
+  public FakeQuantWithMinMaxVarsGradient fakeQuantWithMinMaxVarsGradient(
+      Operand<TFloat32> gradients, Operand<TFloat32> inputs, Operand<TFloat32> min,
+      Operand<TFloat32> max, FakeQuantWithMinMaxVarsGradient.Options... options) {
+    return FakeQuantWithMinMaxVarsGradient.create(scope, gradients, inputs, min, max, options);
+  }
+
+  /**
+   * Fake-quantize the 'inputs' tensor of type float via per-channel floats
+   *  <p>
+   *  Fake-quantize the `inputs` tensor of type float per-channel and one of the
+   *  shapes: `[d]`, `[b, d]` `[b, h, w, d]` via per-channel floats `min` and `max`
+   *  of shape `[d]` to `outputs` tensor of same shape as `inputs`.
+   *  <p>
+   *  Attributes
+   *  <ul>
+   *  <li>
+   *  `[min; max]` define the clamping range for the `inputs` data.
+   *  </li>
+   *  <li>
+   *  `inputs` values are quantized into the quantization range (
+   *  `[0; 2^num_bits - 1]` when `narrow_range` is false and `[1; 2^num_bits - 1]`
+   *  when it is true) and then de-quantized and output as floats in `[min; max]`
+   *  interval.
+   *  </li>
+   *  <li>
+   *  `num_bits` is the bitwidth of the quantization; between 2 and 16, inclusive.
+   *  </li>
+   *  </ul>
+   *  Before quantization, `min` and `max` values are adjusted with the following
+   *  logic.
+   *  It is suggested to have `min <= 0 <= max`. If `0` is not in the range of values,
+   *  the behavior can be unexpected:
+   *  <ul>
+   *  <li>
+   *  If `0 < min < max`: `min_adj = 0` and `max_adj = max - min`.
+   *  </li>
+   *  <li>
+   *  If `min < max < 0`: `min_adj = min - max` and `max_adj = 0`.
+   *  </li>
+   *  <li>
+   *  If `min <= 0 <= max`: `scale = (max - min) / (2^num_bits - 1) `,
+   *  `min_adj = scale * round(min / scale)` and `max_adj = max + min_adj - min`.
+   *  </li>
+   *  </ul>
+   *  This operation has a gradient and thus allows for training `min` and `max`
+   *  values.
+   *
+   * @param inputs
+   * @param min
+   * @param max
+   * @param options carries optional attributes values
+   * @return a new instance of FakeQuantWithMinMaxVarsPerChannel
+   */
+  public FakeQuantWithMinMaxVarsPerChannel fakeQuantWithMinMaxVarsPerChannel(
+      Operand<TFloat32> inputs, Operand<TFloat32> min, Operand<TFloat32> max,
+      FakeQuantWithMinMaxVarsPerChannel.Options... options) {
+    return FakeQuantWithMinMaxVarsPerChannel.create(scope, inputs, min, max, options);
+  }
+
+  /**
+   * Compute gradients for a FakeQuantWithMinMaxVarsPerChannel operation.
+   *
+   * @param gradients Backpropagated gradients above the FakeQuantWithMinMaxVars operation,
+   *  shape one of: `[d]`, `[b, d]`,  `[b, h, w, d]`.
+   * @param inputs Values passed as inputs to the FakeQuantWithMinMaxVars operation, shape
+   *    same as `gradients`.
+   *  min, max: Quantization interval, floats of shape `[d]`.
+   * @param min
+   * @param max
+   * @param options carries optional attributes values
+   * @return a new instance of FakeQuantWithMinMaxVarsPerChannelGradient
+   */
+  public FakeQuantWithMinMaxVarsPerChannelGradient fakeQuantWithMinMaxVarsPerChannelGradient(
+      Operand<TFloat32> gradients, Operand<TFloat32> inputs, Operand<TFloat32> min,
+      Operand<TFloat32> max, FakeQuantWithMinMaxVarsPerChannelGradient.Options... options) {
+    return FakeQuantWithMinMaxVarsPerChannelGradient.create(scope, gradients, inputs, min, max, options);
+  }
+
+  /**
+   * Quantize the 'input' tensor of type float to 'output' tensor of type 'T'.
+   *  <p>
+   *  [min_range, max_range] are scalar floats that specify the range for
+   *  the 'input' data. The 'mode' attribute controls exactly which calculations are
+   *  used to convert the float values to their quantized equivalents.  The
+   *  'round_mode' attribute controls which rounding tie-breaking algorithm is used
+   *  when rounding float values to their quantized equivalents.
+   *  <p>
+   *  In 'MIN_COMBINED' mode, each value of the tensor will undergo the following:
+   *  <pre>{@code
+   *  out[i] = (in[i] - min_range) * range(T) / (max_range - min_range)
+   *  if T == qint8: out[i] -= (range(T) + 1) / 2.0
+   *  }</pre>
+   *  here `range(T) = numeric_limits<T>::max() - numeric_limits<T>::min()`
+   *  <p>
+   *  <i>MIN_COMBINED Mode Example</i>
+   *  <p>
+   *  Assume the input is type float and has a possible range of [0.0, 6.0] and the
+   *  output type is quint8 ([0, 255]). The min_range and max_range values should be
+   *  specified as 0.0 and 6.0. Quantizing from float to quint8 will multiply each
+   *  value of the input by 255/6 and cast to quint8.
+   *  <p>
+   *  If the output type was qint8 ([-128, 127]), the operation will additionally
+   *  subtract each value by 128 prior to casting, so that the range of values aligns
+   *  with the range of qint8.
+   *  <p>
+   *  If the mode is 'MIN_FIRST', then this approach is used:
+   *  <pre>{@code
+   *  num_discrete_values = 1 << (# of bits in T)
+   *  range_adjust = num_discrete_values / (num_discrete_values - 1)
+   *  range = (range_max - range_min) * range_adjust
+   *  range_scale = num_discrete_values / range
+   *  quantized = round(input * range_scale) - round(range_min * range_scale) +
+   *    numeric_limits<T>::min()
+   *  quantized = max(quantized, numeric_limits<T>::min())
+   *  quantized = min(quantized, numeric_limits<T>::max())
+   *  }</pre>
+   *  The biggest difference between this and MIN_COMBINED is that the minimum range
+   *  is rounded first, before it's subtracted from the rounded value. With
+   *  MIN_COMBINED, a small bias is introduced where repeated iterations of quantizing
+   *  and dequantizing will introduce a larger and larger error.
+   *  <p>
+   *  <i>SCALED mode Example</i>
+   *  <p>
+   *  `SCALED` mode matches the quantization approach used in
+   *  `QuantizeAndDequantize{V2|V3}`.
+   *  <p>
+   *  If the mode is `SCALED`, the quantization is performed by multiplying each
+   *  input value by a scaling_factor.
+   *  The scaling_factor is determined from `min_range` and `max_range` to be as large
+   *  as possible such that the range from `min_range` to `max_range` is representable
+   *  within values of type T.
+   *  <pre>{@code
+   *    const int min_T = std::numeric_limits<T>::min();
+   *    const int max_T = std::numeric_limits<T>::max();
+   *    const float max_float = std::numeric_limits<float>::max();
+   *
+   *    const float scale_factor_from_min_side =
+   *        (min_T * min_range > 0) ? min_T / min_range : max_float;
+   *    const float scale_factor_from_max_side =
+   *        (max_T * max_range > 0) ? max_T / max_range : max_float;
+   *
+   *    const float scale_factor = std::min(scale_factor_from_min_side,
+   *                                        scale_factor_from_max_side);
+   *  }</pre>
+   *  We next use the scale_factor to adjust min_range and max_range as follows:
+   *  <pre>{@code
+   *        min_range = min_T / scale_factor;
+   *        max_range = max_T / scale_factor;
+   *  }</pre>
+   *  e.g. if T = qint8, and initially min_range = -10, and max_range = 9, we would
+   *  compare -128/-10.0 = 12.8 to 127/9.0 = 14.11, and set scaling_factor = 12.8
+   *  In this case, min_range would remain -10, but max_range would be adjusted to
+   *  127 / 12.8 = 9.921875
+   *  <p>
+   *  So we will quantize input values in the range (-10, 9.921875) to (-128, 127).
+   *  <p>
+   *  The input tensor can now be quantized by clipping values to the range
+   *  `min_range` to `max_range`, then multiplying by scale_factor as follows:
+   *  <pre>{@code
+   *  result = round(min(max_range, max(min_range, input)) * scale_factor)
+   *  }</pre>
+   *  The adjusted `min_range` and `max_range` are returned as outputs 2 and 3 of
+   *  this operation. These outputs should be used as the range for any further
+   *  calculations.
+   *  <p>
+   *  <i>narrow_range (bool) attribute</i>
+   *  <p>
+   *  If true, we do not use the minimum quantized value.
+   *  i.e. for int8 the quantized output, it would be restricted to the range
+   *  -127..127 instead of the full -128..127 range.
+   *  This is provided for compatibility with certain inference backends.
+   *  (Only applies to SCALED mode)
+   *  <p>
+   *  <i>axis (int) attribute</i>
+   *  <p>
+   *  An optional `axis` attribute can specify a dimension index of the input tensor,
+   *  such that quantization ranges will be calculated and applied separately for each
+   *  slice of the tensor along that dimension. This is useful for per-channel
+   *  quantization.
+   *  <p>
+   *  If axis is specified, min_range and max_range
+   *  <p>
+   *  if `axis`=None, per-tensor quantization is performed as normal.
+   *  <p>
+   *  <i>ensure_minimum_range (float) attribute</i>
+   *  <p>
+   *  Ensures the minimum quantization range is at least this value.
+   *  The legacy default value for this is 0.01, but it is strongly suggested to
+   *  set it to 0 for new uses.
+   *
+   * @param <T> data type for {@code output()} output
+   * @param input
+   * @param minRange The minimum value of the quantization range. This value may be adjusted by the
+   *  op depending on other parameters. The adjusted value is written to `output_min`.
+   *  If the `axis` attribute is specified, this must be a 1-D tensor whose size
+   *  matches the `axis` dimension of the input and output tensors.
+   * @param maxRange The maximum value of the quantization range. This value may be adjusted by the
+   *  op depending on other parameters. The adjusted value is written to `output_max`.
+   *  If the `axis` attribute is specified, this must be a 1-D tensor whose size
+   *  matches the `axis` dimension of the input and output tensors.
+   * @param T
+   * @param options carries optional attributes values
+   * @return a new instance of Quantize
+   */
+  public <T extends TType> Quantize<T> quantize(Operand<TFloat32> input, Operand<TFloat32> minRange,
+      Operand<TFloat32> maxRange, Class<T> T, Quantize.Options... options) {
+    return Quantize.create(scope, input, minRange, maxRange, T, options);
+  }
+
+  /**
+   * Quantizes then dequantizes a tensor.
+   *  <p>
+   *  This is almost identical to QuantizeAndDequantizeV2, except that num_bits is a
+   *  tensor, so its value can change during training.
+   *
+   * @param <T> data type for {@code output()} output
+   * @param input
+   * @param inputMin
+   * @param inputMax
+   * @param numBits
+   * @param options carries optional attributes values
+   * @return a new instance of QuantizeAndDequantize
+   */
+  public <T extends TNumber> QuantizeAndDequantize<T> quantizeAndDequantize(Operand<T> input,
+      Operand<T> inputMin, Operand<T> inputMax, Operand<TInt32> numBits,
+      QuantizeAndDequantize.Options... options) {
+    return QuantizeAndDequantize.create(scope, input, inputMin, inputMax, numBits, options);
+  }
+
+  /**
+   * Convert the quantized 'input' tensor into a lower-precision 'output', using the
+   *  <p>
+   *  actual distribution of the values to maximize the usage of the lower bit depth
+   *  and adjusting the output min and max ranges accordingly.
+   *  <p>
+   *  [input_min, input_max] are scalar floats that specify the range for the float
+   *  interpretation of the 'input' data. For example, if input_min is -1.0f and
+   *  input_max is 1.0f, and we are dealing with quint16 quantized data, then a 0
+   *  value in the 16-bit data should be interpreted as -1.0f, and a 65535 means 1.0f.
+   *  <p>
+   *  This operator tries to squeeze as much precision as possible into an output with
+   *  a lower bit depth by calculating the actual min and max values found in the
+   *  data. For example, maybe that quint16 input has no values lower than 16,384 and
+   *  none higher than 49,152. That means only half the range is actually needed, all
+   *  the float interpretations are between -0.5f and 0.5f, so if we want to compress
+   *  the data into a quint8 output, we can use that range rather than the theoretical
+   *  -1.0f to 1.0f that is suggested by the input min and max.
+   *  <p>
+   *  In practice, this is most useful for taking output from operations like
+   *  QuantizedMatMul that can produce higher bit-depth outputs than their inputs and
+   *  may have large potential output ranges, but in practice have a distribution of
+   *  input values that only uses a small fraction of the possible range. By feeding
+   *  that output into this operator, we can reduce it from 32 bits down to 8 with
+   *  minimal loss of accuracy.
+   *
+   * @param <U> data type for {@code output()} output
+   * @param input
+   * @param inputMin The float value that the minimum quantized input value represents.
+   * @param inputMax The float value that the maximum quantized input value represents.
+   * @param outType The type of the output. Should be a lower bit depth than Tinput.
+   * @return a new instance of QuantizeDownAndShrinkRange
+   */
+  public <U extends TType, T extends TType> QuantizeDownAndShrinkRange<U> quantizeDownAndShrinkRange(
+      Operand<T> input, Operand<TFloat32> inputMin, Operand<TFloat32> inputMax, Class<U> outType) {
+    return QuantizeDownAndShrinkRange.create(scope, input, inputMin, inputMax, outType);
+  }
+
+  /**
+   * Concatenates quantized tensors along one dimension.
+   *
+   * @param <T> data type for {@code output()} output
+   * @param concatDim 0-D.  The dimension along which to concatenate.  Must be in the
+   *  range [0, rank(values)).
+   * @param values The `N` Tensors to concatenate. Their ranks and types must match,
+   *  and their sizes must match in all dimensions except `concat_dim`.
+   * @param inputMins The minimum scalar values for each of the input tensors.
+   * @param inputMaxes The maximum scalar values for each of the input tensors.
+   * @return a new instance of QuantizedConcat
+   */
+  public <T extends TType> QuantizedConcat<T> quantizedConcat(Operand<TInt32> concatDim,
+      Iterable<Operand<T>> values, Iterable<Operand<TFloat32>> inputMins,
+      Iterable<Operand<TFloat32>> inputMaxes) {
+    return QuantizedConcat.create(scope, concatDim, values, inputMins, inputMaxes);
+  }
+
+  /**
+   * Computes a range that covers the actual values present in a quantized tensor.
+   *  <p>
+   *  Given a quantized tensor described by `(input, input_min, input_max)`, outputs a
+   *  range that covers the actual values present in that tensor. This op is typically
+   *  used to produce the `requested_output_min` and `requested_output_max` for
+   *  `Requantize`.
+   *
+   * @param input
+   * @param inputMin The float value that the minimum quantized input value represents.
+   * @param inputMax The float value that the maximum quantized input value represents.
+   * @return a new instance of RequantizationRange
+   */
+  public <T extends TType> RequantizationRange requantizationRange(Operand<T> input,
+      Operand<TFloat32> inputMin, Operand<TFloat32> inputMax) {
+    return RequantizationRange.create(scope, input, inputMin, inputMax);
+  }
+
+  /**
+   * Converts the quantized `input` tensor into a lower-precision `output`.
+   *  <p>
+   *  Converts the quantized `input` tensor into a lower-precision `output`, using the
+   *  output range specified with `requested_output_min` and `requested_output_max`.
+   *  <p>
+   *  `[input_min, input_max]` are scalar floats that specify the range for the float
+   *  interpretation of the `input` data. For example, if `input_min` is -1.0f and
+   *  `input_max` is 1.0f, and we are dealing with `quint16` quantized data, then a 0
+   *  value in the 16-bit data should be interpreted as -1.0f, and a 65535 means 1.0f.
+   *
+   * @param <U> data type for {@code output()} output
+   * @param input
+   * @param inputMin The float value that the minimum quantized input value represents.
+   * @param inputMax The float value that the maximum quantized input value represents.
+   * @param requestedOutputMin The float value that the minimum quantized output value represents.
+   * @param requestedOutputMax The float value that the maximum quantized output value represents.
+   * @param outType The type of the output. Should be a lower bit depth than Tinput.
+   * @return a new instance of Requantize
+   */
+  public <U extends TType, T extends TType> Requantize<U> requantize(Operand<T> input,
+      Operand<TFloat32> inputMin, Operand<TFloat32> inputMax, Operand<TFloat32> requestedOutputMin,
+      Operand<TFloat32> requestedOutputMax, Class<U> outType) {
+    return Requantize.create(scope, input, inputMin, inputMax, requestedOutputMin, requestedOutputMax, outType);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/RaggedOps.java b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/RaggedOps.java
new file mode 100644
index 00000000000..7d585455bdd
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/RaggedOps.java
@@ -0,0 +1,63 @@
+// Copyright 2020 The TensorFlow Authors. All Rights Reserved.
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+// ==============================================================================
+//
+// This class has been generated, DO NOT EDIT!
+//
+package org.tensorflow.op;
+
+import org.tensorflow.Operand;
+import org.tensorflow.op.ragged.RaggedBincount;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * An API for building {@code ragged} operations as {@link Op Op}s
+ *
+ * @see {@link Ops}
+ */
+public final class RaggedOps {
+  private final Scope scope;
+
+  RaggedOps(Scope scope) {
+    this.scope = scope;
+  }
+
+  /**
+   * Counts the number of occurrences of each value in an integer array.
+   *  <p>
+   *  Outputs a vector with length `size` and the same dtype as `weights`. If
+   *  `weights` are empty, then index `i` stores the number of times the value `i` is
+   *  counted in `arr`. If `weights` are non-empty, then index `i` stores the sum of
+   *  the value in `weights` at each index where the corresponding value in `arr` is
+   *  `i`.
+   *  <p>
+   *  Values in `arr` outside of the range [0, size) are ignored.
+   *
+   * @param <U> data type for {@code output()} output
+   * @param splits 1D int64 `Tensor`.
+   * @param values 2D int `Tensor`.
+   * @param size non-negative int scalar `Tensor`.
+   * @param weights is an int32, int64, float32, or float64 `Tensor` with the same
+   *  shape as `input`, or a length-0 `Tensor`, in which case it acts as all weights
+   *  equal to 1.
+   * @param options carries optional attributes values
+   * @return a new instance of RaggedBincount
+   */
+  public <U extends TNumber, T extends TNumber> RaggedBincount<U> raggedBincount(
+      Operand<TInt64> splits, Operand<T> values, Operand<T> size, Operand<U> weights,
+      RaggedBincount.Options... options) {
+    return RaggedBincount.create(scope, splits, values, size, weights, options);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/RandomOps.java b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/RandomOps.java
new file mode 100644
index 00000000000..5fe5b6268fe
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/RandomOps.java
@@ -0,0 +1,585 @@
+// Copyright 2020 The TensorFlow Authors. All Rights Reserved.
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+// ==============================================================================
+//
+// This class has been generated, DO NOT EDIT!
+//
+package org.tensorflow.op;
+
+import org.tensorflow.Operand;
+import org.tensorflow.op.random.AllCandidateSampler;
+import org.tensorflow.op.random.LogUniformCandidateSampler;
+import org.tensorflow.op.random.Multinomial;
+import org.tensorflow.op.random.ParameterizedTruncatedNormal;
+import org.tensorflow.op.random.RandomGamma;
+import org.tensorflow.op.random.RandomPoisson;
+import org.tensorflow.op.random.RandomShuffle;
+import org.tensorflow.op.random.RandomStandardNormal;
+import org.tensorflow.op.random.RandomUniform;
+import org.tensorflow.op.random.RandomUniformInt;
+import org.tensorflow.op.random.RecordInput;
+import org.tensorflow.op.random.StatefulRandomBinomial;
+import org.tensorflow.op.random.StatefulStandardNormal;
+import org.tensorflow.op.random.StatelessMultinomial;
+import org.tensorflow.op.random.StatelessRandomNormal;
+import org.tensorflow.op.random.StatelessRandomUniform;
+import org.tensorflow.op.random.StatelessTruncatedNormal;
+import org.tensorflow.op.random.TruncatedNormal;
+import org.tensorflow.op.random.UniformCandidateSampler;
+import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * An API for building {@code random} operations as {@link Op Op}s
+ *
+ * @see {@link Ops}
+ */
+public final class RandomOps {
+  private final Scope scope;
+
+  RandomOps(Scope scope) {
+    this.scope = scope;
+  }
+
+  /**
+   * Generates labels for candidate sampling with a learned unigram distribution.
+   *  <p>
+   *  See explanations of candidate sampling and the data formats at
+   *  go/candidate-sampling.
+   *  <p>
+   *  For each batch, this op picks a single set of sampled candidate labels.
+   *  <p>
+   *  The advantages of sampling candidates per-batch are simplicity and the
+   *  possibility of efficient dense matrix multiplication. The disadvantage is that
+   *  the sampled candidates must be chosen independently of the context and of the
+   *  true labels.
+   *
+   * @param trueClasses A batch_size * num_true matrix, in which each row contains the
+   *  IDs of the num_true target_classes in the corresponding original label.
+   * @param numTrue Number of true labels per context.
+   * @param numSampled Number of candidates to produce.
+   * @param unique If unique is true, we sample with rejection, so that all sampled
+   *  candidates in a batch are unique. This requires some approximation to
+   *  estimate the post-rejection sampling probabilities.
+   * @param options carries optional attributes values
+   * @return a new instance of AllCandidateSampler
+   */
+  public AllCandidateSampler allCandidateSampler(Operand<TInt64> trueClasses, Long numTrue,
+      Long numSampled, Boolean unique, AllCandidateSampler.Options... options) {
+    return AllCandidateSampler.create(scope, trueClasses, numTrue, numSampled, unique, options);
+  }
+
+  /**
+   * Generates labels for candidate sampling with a log-uniform distribution.
+   *  <p>
+   *  See explanations of candidate sampling and the data formats at
+   *  go/candidate-sampling.
+   *  <p>
+   *  For each batch, this op picks a single set of sampled candidate labels.
+   *  <p>
+   *  The advantages of sampling candidates per-batch are simplicity and the
+   *  possibility of efficient dense matrix multiplication. The disadvantage is that
+   *  the sampled candidates must be chosen independently of the context and of the
+   *  true labels.
+   *
+   * @param trueClasses A batch_size * num_true matrix, in which each row contains the
+   *  IDs of the num_true target_classes in the corresponding original label.
+   * @param numTrue Number of true labels per context.
+   * @param numSampled Number of candidates to randomly sample.
+   * @param unique If unique is true, we sample with rejection, so that all sampled
+   *  candidates in a batch are unique. This requires some approximation to
+   *  estimate the post-rejection sampling probabilities.
+   * @param rangeMax The sampler will sample integers from the interval [0, range_max).
+   * @param options carries optional attributes values
+   * @return a new instance of LogUniformCandidateSampler
+   */
+  public LogUniformCandidateSampler logUniformCandidateSampler(Operand<TInt64> trueClasses,
+      Long numTrue, Long numSampled, Boolean unique, Long rangeMax,
+      LogUniformCandidateSampler.Options... options) {
+    return LogUniformCandidateSampler.create(scope, trueClasses, numTrue, numSampled, unique, rangeMax, options);
+  }
+
+  /**
+   * Draws samples from a multinomial distribution.
+   *
+   * @param <U> data type for {@code output()} output
+   * @param logits 2-D Tensor with shape `[batch_size, num_classes]`.  Each slice `[i, :]`
+   *  represents the unnormalized log probabilities for all classes.
+   * @param numSamples 0-D.  Number of independent samples to draw for each row slice.
+   * @param options carries optional attributes values
+   * @return a new instance of Multinomial
+   */
+  public <T extends TNumber> Multinomial<TInt64> multinomial(Operand<T> logits,
+      Operand<TInt32> numSamples, Multinomial.Options... options) {
+    return Multinomial.create(scope, logits, numSamples, options);
+  }
+
+  /**
+   * Draws samples from a multinomial distribution.
+   *
+   * @param <U> data type for {@code output()} output
+   * @param logits 2-D Tensor with shape `[batch_size, num_classes]`.  Each slice `[i, :]`
+   *  represents the unnormalized log probabilities for all classes.
+   * @param numSamples 0-D.  Number of independent samples to draw for each row slice.
+   * @param outputDtype
+   * @param options carries optional attributes values
+   * @return a new instance of Multinomial
+   */
+  public <U extends TNumber, T extends TNumber> Multinomial<U> multinomial(Operand<T> logits,
+      Operand<TInt32> numSamples, Class<U> outputDtype, Multinomial.Options... options) {
+    return Multinomial.create(scope, logits, numSamples, outputDtype, options);
+  }
+
+  /**
+   * Outputs random values from a normal distribution. The parameters may each be a
+   *  <p>
+   *  scalar which applies to the entire output, or a vector of length shape[0] which
+   *  stores the parameters for each batch.
+   *
+   * @param <U> data type for {@code output()} output
+   * @param shape The shape of the output tensor. Batches are indexed by the 0th dimension.
+   * @param means The mean parameter of each batch.
+   * @param stdevs The standard deviation parameter of each batch. Must be greater than 0.
+   * @param minvals The minimum cutoff. May be -infinity.
+   * @param maxvals The maximum cutoff. May be +infinity, and must be more than the minval
+   *  for each batch.
+   * @param options carries optional attributes values
+   * @return a new instance of ParameterizedTruncatedNormal
+   */
+  public <U extends TNumber, T extends TNumber> ParameterizedTruncatedNormal<U> parameterizedTruncatedNormal(
+      Operand<T> shape, Operand<U> means, Operand<U> stdevs, Operand<U> minvals, Operand<U> maxvals,
+      ParameterizedTruncatedNormal.Options... options) {
+    return ParameterizedTruncatedNormal.create(scope, shape, means, stdevs, minvals, maxvals, options);
+  }
+
+  /**
+   * Outputs random values from the Gamma distribution(s) described by alpha.
+   *  <p>
+   *  This op uses the algorithm by Marsaglia et al. to acquire samples via
+   *  transformation-rejection from pairs of uniform and normal random variables.
+   *  See http://dl.acm.org/citation.cfm?id=358414
+   *
+   * @param <U> data type for {@code output()} output
+   * @param shape 1-D integer tensor. Shape of independent samples to draw from each
+   *  distribution described by the shape parameters given in alpha.
+   * @param alpha A tensor in which each scalar is a "shape" parameter describing the
+   *  associated gamma distribution.
+   * @param options carries optional attributes values
+   * @return a new instance of RandomGamma
+   */
+  public <U extends TNumber, T extends TNumber> RandomGamma<U> randomGamma(Operand<T> shape,
+      Operand<U> alpha, RandomGamma.Options... options) {
+    return RandomGamma.create(scope, shape, alpha, options);
+  }
+
+  /**
+   * Outputs random values from the Poisson distribution(s) described by rate.
+   *  <p>
+   *  This op uses two algorithms, depending on rate. If rate >= 10, then
+   *  the algorithm by Hormann is used to acquire samples via
+   *  transformation-rejection.
+   *  See http://www.sciencedirect.com/science/article/pii/0167668793909974.
+   *  <p>
+   *  Otherwise, Knuth's algorithm is used to acquire samples via multiplying uniform
+   *  random variables.
+   *  See Donald E. Knuth (1969). Seminumerical Algorithms. The Art of Computer
+   *  Programming, Volume 2. Addison Wesley
+   *
+   * @param <V> data type for {@code output()} output
+   * @param shape 1-D integer tensor. Shape of independent samples to draw from each
+   *  distribution described by the shape parameters given in rate.
+   * @param rate A tensor in which each scalar is a "rate" parameter describing the
+   *  associated poisson distribution.
+   * @param options carries optional attributes values
+   * @return a new instance of RandomPoisson
+   */
+  public <T extends TNumber, U extends TNumber> RandomPoisson<TInt64> randomPoisson(
+      Operand<T> shape, Operand<U> rate, RandomPoisson.Options... options) {
+    return RandomPoisson.create(scope, shape, rate, options);
+  }
+
+  /**
+   * Outputs random values from the Poisson distribution(s) described by rate.
+   *  <p>
+   *  This op uses two algorithms, depending on rate. If rate >= 10, then
+   *  the algorithm by Hormann is used to acquire samples via
+   *  transformation-rejection.
+   *  See http://www.sciencedirect.com/science/article/pii/0167668793909974.
+   *  <p>
+   *  Otherwise, Knuth's algorithm is used to acquire samples via multiplying uniform
+   *  random variables.
+   *  See Donald E. Knuth (1969). Seminumerical Algorithms. The Art of Computer
+   *  Programming, Volume 2. Addison Wesley
+   *
+   * @param <V> data type for {@code output()} output
+   * @param shape 1-D integer tensor. Shape of independent samples to draw from each
+   *  distribution described by the shape parameters given in rate.
+   * @param rate A tensor in which each scalar is a "rate" parameter describing the
+   *  associated poisson distribution.
+   * @param dtype
+   * @param options carries optional attributes values
+   * @return a new instance of RandomPoisson
+   */
+  public <V extends TNumber, T extends TNumber, U extends TNumber> RandomPoisson<V> randomPoisson(
+      Operand<T> shape, Operand<U> rate, Class<V> dtype, RandomPoisson.Options... options) {
+    return RandomPoisson.create(scope, shape, rate, dtype, options);
+  }
+
+  /**
+   * Randomly shuffles a tensor along its first dimension.
+   *  <p>
+   *    The tensor is shuffled along dimension 0, such that each `value[j]` is mapped
+   *    to one and only one `output[i]`. For example, a mapping that might occur for a
+   *    3x2 tensor is:
+   *  <pre>{@code
+   *  [[1, 2],       [[5, 6],
+   *   [3, 4],  ==>   [1, 2],
+   *   [5, 6]]        [3, 4]]
+   *  }</pre>
+   *
+   * @param <T> data type for {@code output()} output
+   * @param value The tensor to be shuffled.
+   * @param options carries optional attributes values
+   * @return a new instance of RandomShuffle
+   */
+  public <T extends TType> RandomShuffle<T> randomShuffle(Operand<T> value,
+      RandomShuffle.Options... options) {
+    return RandomShuffle.create(scope, value, options);
+  }
+
+  /**
+   * Outputs random values from a normal distribution.
+   *  <p>
+   *  The generated values will have mean 0 and standard deviation 1.
+   *
+   * @param <U> data type for {@code output()} output
+   * @param shape The shape of the output tensor.
+   * @param dtype The type of the output.
+   * @param options carries optional attributes values
+   * @return a new instance of RandomStandardNormal
+   */
+  public <U extends TNumber, T extends TNumber> RandomStandardNormal<U> randomStandardNormal(
+      Operand<T> shape, Class<U> dtype, RandomStandardNormal.Options... options) {
+    return RandomStandardNormal.create(scope, shape, dtype, options);
+  }
+
+  /**
+   * Outputs random values from a uniform distribution.
+   *  <p>
+   *  The generated values follow a uniform distribution in the range `[0, 1)`. The
+   *  lower bound 0 is included in the range, while the upper bound 1 is excluded.
+   *
+   * @param <U> data type for {@code output()} output
+   * @param shape The shape of the output tensor.
+   * @param dtype The type of the output.
+   * @param options carries optional attributes values
+   * @return a new instance of RandomUniform
+   */
+  public <U extends TNumber, T extends TNumber> RandomUniform<U> randomUniform(Operand<T> shape,
+      Class<U> dtype, RandomUniform.Options... options) {
+    return RandomUniform.create(scope, shape, dtype, options);
+  }
+
+  /**
+   * Outputs random integers from a uniform distribution.
+   *  <p>
+   *  The generated values are uniform integers in the range `[minval, maxval)`.
+   *  The lower bound `minval` is included in the range, while the upper bound
+   *  `maxval` is excluded.
+   *  <p>
+   *  The random integers are slightly biased unless `maxval - minval` is an exact
+   *  power of two.  The bias is small for values of `maxval - minval` significantly
+   *  smaller than the range of the output (either `2^32` or `2^64`).
+   *
+   * @param <U> data type for {@code output()} output
+   * @param shape The shape of the output tensor.
+   * @param minval 0-D.  Inclusive lower bound on the generated integers.
+   * @param maxval 0-D.  Exclusive upper bound on the generated integers.
+   * @param options carries optional attributes values
+   * @return a new instance of RandomUniformInt
+   */
+  public <U extends TNumber, T extends TNumber> RandomUniformInt<U> randomUniformInt(
+      Operand<T> shape, Operand<U> minval, Operand<U> maxval, RandomUniformInt.Options... options) {
+    return RandomUniformInt.create(scope, shape, minval, maxval, options);
+  }
+
+  /**
+   * Emits randomized records.
+   *
+   * @param filePattern Glob pattern for the data files.
+   * @param options carries optional attributes values
+   * @return a new instance of RecordInput
+   */
+  public RecordInput recordInput(String filePattern, RecordInput.Options... options) {
+    return RecordInput.create(scope, filePattern, options);
+  }
+
+  /**
+   *
+   * @param <V> data type for {@code output()} output
+   * @param resource
+   * @param algorithm
+   * @param shape
+   * @param counts
+   * @param probs
+   * @return a new instance of StatefulRandomBinomial
+   */
+  public <T extends TNumber, U extends TNumber> StatefulRandomBinomial<TInt64> statefulRandomBinomial(
+      Operand<?> resource, Operand<TInt64> algorithm, Operand<T> shape, Operand<U> counts,
+      Operand<U> probs) {
+    return StatefulRandomBinomial.create(scope, resource, algorithm, shape, counts, probs);
+  }
+
+  /**
+   *
+   * @param <V> data type for {@code output()} output
+   * @param resource
+   * @param algorithm
+   * @param shape
+   * @param counts
+   * @param probs
+   * @param dtype
+   * @return a new instance of StatefulRandomBinomial
+   */
+  public <V extends TNumber, T extends TNumber, U extends TNumber> StatefulRandomBinomial<V> statefulRandomBinomial(
+      Operand<?> resource, Operand<TInt64> algorithm, Operand<T> shape, Operand<U> counts,
+      Operand<U> probs, Class<V> dtype) {
+    return StatefulRandomBinomial.create(scope, resource, algorithm, shape, counts, probs, dtype);
+  }
+
+  /**
+   * Outputs random values from a normal distribution.
+   *  <p>
+   *  The generated values will have mean 0 and standard deviation 1.
+   *
+   * @param <U> data type for {@code output()} output
+   * @param resource The handle of the resource variable that stores the state of the RNG.
+   * @param algorithm The RNG algorithm.
+   * @param shape The shape of the output tensor.
+   * @return a new instance of StatefulStandardNormal
+   */
+  public <T extends TType> StatefulStandardNormal<TFloat32> statefulStandardNormal(
+      Operand<?> resource, Operand<TInt64> algorithm, Operand<T> shape) {
+    return StatefulStandardNormal.create(scope, resource, algorithm, shape);
+  }
+
+  /**
+   * Outputs random values from a normal distribution.
+   *  <p>
+   *  The generated values will have mean 0 and standard deviation 1.
+   *
+   * @param <U> data type for {@code output()} output
+   * @param resource The handle of the resource variable that stores the state of the RNG.
+   * @param algorithm The RNG algorithm.
+   * @param shape The shape of the output tensor.
+   * @param dtype The type of the output.
+   * @return a new instance of StatefulStandardNormal
+   */
+  public <U extends TType, T extends TType> StatefulStandardNormal<U> statefulStandardNormal(
+      Operand<?> resource, Operand<TInt64> algorithm, Operand<T> shape, Class<U> dtype) {
+    return StatefulStandardNormal.create(scope, resource, algorithm, shape, dtype);
+  }
+
+  /**
+   * Draws samples from a multinomial distribution.
+   *
+   * @param <V> data type for {@code output()} output
+   * @param logits 2-D Tensor with shape `[batch_size, num_classes]`.  Each slice `[i, :]`
+   *  represents the unnormalized log probabilities for all classes.
+   * @param numSamples 0-D.  Number of independent samples to draw for each row slice.
+   * @param seed 2 seeds (shape [2]).
+   * @return a new instance of StatelessMultinomial
+   */
+  public <T extends TNumber, U extends TNumber> StatelessMultinomial<TInt64> statelessMultinomial(
+      Operand<T> logits, Operand<TInt32> numSamples, Operand<U> seed) {
+    return StatelessMultinomial.create(scope, logits, numSamples, seed);
+  }
+
+  /**
+   * Draws samples from a multinomial distribution.
+   *
+   * @param <V> data type for {@code output()} output
+   * @param logits 2-D Tensor with shape `[batch_size, num_classes]`.  Each slice `[i, :]`
+   *  represents the unnormalized log probabilities for all classes.
+   * @param numSamples 0-D.  Number of independent samples to draw for each row slice.
+   * @param seed 2 seeds (shape [2]).
+   * @param outputDtype
+   * @return a new instance of StatelessMultinomial
+   */
+  public <V extends TNumber, T extends TNumber, U extends TNumber> StatelessMultinomial<V> statelessMultinomial(
+      Operand<T> logits, Operand<TInt32> numSamples, Operand<U> seed, Class<V> outputDtype) {
+    return StatelessMultinomial.create(scope, logits, numSamples, seed, outputDtype);
+  }
+
+  /**
+   * Outputs deterministic pseudorandom values from a normal distribution.
+   *  <p>
+   *  The generated values will have mean 0 and standard deviation 1.
+   *  <p>
+   *  The outputs are a deterministic function of `shape` and `seed`.
+   *
+   * @param <V> data type for {@code output()} output
+   * @param shape The shape of the output tensor.
+   * @param seed 2 seeds (shape [2]).
+   * @return a new instance of StatelessRandomNormal
+   */
+  public <T extends TNumber, U extends TNumber> StatelessRandomNormal<TFloat32> statelessRandomNormal(
+      Operand<T> shape, Operand<U> seed) {
+    return StatelessRandomNormal.create(scope, shape, seed);
+  }
+
+  /**
+   * Outputs deterministic pseudorandom values from a normal distribution.
+   *  <p>
+   *  The generated values will have mean 0 and standard deviation 1.
+   *  <p>
+   *  The outputs are a deterministic function of `shape` and `seed`.
+   *
+   * @param <V> data type for {@code output()} output
+   * @param shape The shape of the output tensor.
+   * @param seed 2 seeds (shape [2]).
+   * @param dtype The type of the output.
+   * @return a new instance of StatelessRandomNormal
+   */
+  public <V extends TNumber, T extends TNumber, U extends TNumber> StatelessRandomNormal<V> statelessRandomNormal(
+      Operand<T> shape, Operand<U> seed, Class<V> dtype) {
+    return StatelessRandomNormal.create(scope, shape, seed, dtype);
+  }
+
+  /**
+   * Outputs deterministic pseudorandom random values from a uniform distribution.
+   *  <p>
+   *  The generated values follow a uniform distribution in the range `[0, 1)`. The
+   *  lower bound 0 is included in the range, while the upper bound 1 is excluded.
+   *  <p>
+   *  The outputs are a deterministic function of `shape` and `seed`.
+   *
+   * @param <V> data type for {@code output()} output
+   * @param shape The shape of the output tensor.
+   * @param seed 2 seeds (shape [2]).
+   * @return a new instance of StatelessRandomUniform
+   */
+  public <T extends TNumber, U extends TNumber> StatelessRandomUniform<TFloat32> statelessRandomUniform(
+      Operand<T> shape, Operand<U> seed) {
+    return StatelessRandomUniform.create(scope, shape, seed);
+  }
+
+  /**
+   * Outputs deterministic pseudorandom random values from a uniform distribution.
+   *  <p>
+   *  The generated values follow a uniform distribution in the range `[0, 1)`. The
+   *  lower bound 0 is included in the range, while the upper bound 1 is excluded.
+   *  <p>
+   *  The outputs are a deterministic function of `shape` and `seed`.
+   *
+   * @param <V> data type for {@code output()} output
+   * @param shape The shape of the output tensor.
+   * @param seed 2 seeds (shape [2]).
+   * @param dtype The type of the output.
+   * @return a new instance of StatelessRandomUniform
+   */
+  public <V extends TNumber, T extends TNumber, U extends TNumber> StatelessRandomUniform<V> statelessRandomUniform(
+      Operand<T> shape, Operand<U> seed, Class<V> dtype) {
+    return StatelessRandomUniform.create(scope, shape, seed, dtype);
+  }
+
+  /**
+   * Outputs deterministic pseudorandom values from a truncated normal distribution.
+   *  <p>
+   *  The generated values follow a normal distribution with mean 0 and standard
+   *  deviation 1, except that values whose magnitude is more than 2 standard
+   *  deviations from the mean are dropped and re-picked.
+   *  <p>
+   *  The outputs are a deterministic function of `shape` and `seed`.
+   *
+   * @param <V> data type for {@code output()} output
+   * @param shape The shape of the output tensor.
+   * @param seed 2 seeds (shape [2]).
+   * @return a new instance of StatelessTruncatedNormal
+   */
+  public <T extends TNumber, U extends TNumber> StatelessTruncatedNormal<TFloat32> statelessTruncatedNormal(
+      Operand<T> shape, Operand<U> seed) {
+    return StatelessTruncatedNormal.create(scope, shape, seed);
+  }
+
+  /**
+   * Outputs deterministic pseudorandom values from a truncated normal distribution.
+   *  <p>
+   *  The generated values follow a normal distribution with mean 0 and standard
+   *  deviation 1, except that values whose magnitude is more than 2 standard
+   *  deviations from the mean are dropped and re-picked.
+   *  <p>
+   *  The outputs are a deterministic function of `shape` and `seed`.
+   *
+   * @param <V> data type for {@code output()} output
+   * @param shape The shape of the output tensor.
+   * @param seed 2 seeds (shape [2]).
+   * @param dtype The type of the output.
+   * @return a new instance of StatelessTruncatedNormal
+   */
+  public <V extends TNumber, T extends TNumber, U extends TNumber> StatelessTruncatedNormal<V> statelessTruncatedNormal(
+      Operand<T> shape, Operand<U> seed, Class<V> dtype) {
+    return StatelessTruncatedNormal.create(scope, shape, seed, dtype);
+  }
+
+  /**
+   * Outputs random values from a truncated normal distribution.
+   *  <p>
+   *  The generated values follow a normal distribution with mean 0 and standard
+   *  deviation 1, except that values whose magnitude is more than 2 standard
+   *  deviations from the mean are dropped and re-picked.
+   *
+   * @param <U> data type for {@code output()} output
+   * @param shape The shape of the output tensor.
+   * @param dtype The type of the output.
+   * @param options carries optional attributes values
+   * @return a new instance of TruncatedNormal
+   */
+  public <U extends TNumber, T extends TNumber> TruncatedNormal<U> truncatedNormal(Operand<T> shape,
+      Class<U> dtype, TruncatedNormal.Options... options) {
+    return TruncatedNormal.create(scope, shape, dtype, options);
+  }
+
+  /**
+   * Generates labels for candidate sampling with a uniform distribution.
+   *  <p>
+   *  See explanations of candidate sampling and the data formats at
+   *  go/candidate-sampling.
+   *  <p>
+   *  For each batch, this op picks a single set of sampled candidate labels.
+   *  <p>
+   *  The advantages of sampling candidates per-batch are simplicity and the
+   *  possibility of efficient dense matrix multiplication. The disadvantage is that
+   *  the sampled candidates must be chosen independently of the context and of the
+   *  true labels.
+   *
+   * @param trueClasses A batch_size * num_true matrix, in which each row contains the
+   *  IDs of the num_true target_classes in the corresponding original label.
+   * @param numTrue Number of true labels per context.
+   * @param numSampled Number of candidates to randomly sample.
+   * @param unique If unique is true, we sample with rejection, so that all sampled
+   *  candidates in a batch are unique. This requires some approximation to
+   *  estimate the post-rejection sampling probabilities.
+   * @param rangeMax The sampler will sample integers from the interval [0, range_max).
+   * @param options carries optional attributes values
+   * @return a new instance of UniformCandidateSampler
+   */
+  public UniformCandidateSampler uniformCandidateSampler(Operand<TInt64> trueClasses, Long numTrue,
+      Long numSampled, Boolean unique, Long rangeMax, UniformCandidateSampler.Options... options) {
+    return UniformCandidateSampler.create(scope, trueClasses, numTrue, numSampled, unique, rangeMax, options);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/ShapeOps.java b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/ShapeOps.java
new file mode 100644
index 00000000000..3683805eb2e
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/ShapeOps.java
@@ -0,0 +1,471 @@
+// Copyright 2020 The TensorFlow Authors. All Rights Reserved.
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+// ==============================================================================
+//
+// This class has been generated, DO NOT EDIT!
+//
+package org.tensorflow.op;
+
+import org.tensorflow.Operand;
+import org.tensorflow.op.core.Shape;
+import org.tensorflow.op.core.Shapes;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * An API for building {@code shape} operations as {@link Op Op}s
+ *
+ * @see {@link Ops}
+ */
+public final class ShapeOps {
+  private final Scope scope;
+
+  ShapeOps(Scope scope) {
+    this.scope = scope;
+  }
+
+  /**
+   * Creates a 1-dimensional operand containing the dimensions of a shape followed by the last
+   *  dimension.
+   *
+   * @param scope current scope
+   * @param shape the TensorFlow shape
+   * @param lastDimension the dimension(s) to append
+   * @return a 1-dimensional operand containing the dimensions of a shape followed by the last
+   *      dimension
+   */
+  public Operand<TInt64> append(Shape<TInt64> shape, long lastDimension) {
+    return Shapes.append(scope, shape, lastDimension);
+  }
+
+  /**
+   * Creates a 1-dimensional operand containing the dimensions of a shape followed by the last
+   *  dimension.
+   *
+   * @param scope current scope
+   * @param shape the TensorFlow shape
+   * @param lastDimension the dimension(s) to append
+   * @return a 1-dimensional operand containing the dimensions of a shape followed by the last
+   *      dimension
+   */
+  public Operand<TInt32> append(Shape<TInt32> shape, int lastDimension) {
+    return Shapes.append(scope, shape, lastDimension);
+  }
+
+  /**
+   * Creates a 1-dimensional operand that represents a new shape containing the dimensions of the
+   *  operand representing a shape, followed by the dimensions of an operand representing a shape to
+   *  append.
+   *
+   * @param scope current scope
+   * @param shape the TensorFlow shape
+   * @param shapeToAppend the other shape to append
+   * @return a 1-dimensional operand that represents a new shape containing the dimensions of the
+   *      operand representing a shape, followed by the dimensions of an operand representing a shape
+   *      to append
+   */
+  public <T extends TNumber> Operand<T> append(Operand<T> shape, Operand<T> shapeToAppend) {
+    return Shapes.append(scope, shape, shapeToAppend);
+  }
+
+  /**
+   * Flatten the operand to 1 dimension.
+   *
+   * @param <T> the type of operand
+   * @param scope current scope
+   * @param operand the operand to flatten
+   * @return the reshaped operand
+   */
+  public <T extends TType> Operand<T> flatten(Operand<T> operand) {
+    return Shapes.flatten(scope, operand);
+  }
+
+  /**
+   * Flatten the shape to 1 dimension.
+   *
+   * @param scope current scope
+   * @param shape the TensorFlow shape
+   * @return the flattened shape
+   */
+  public Operand<TInt32> flatten(Shape<TInt32> shape) {
+    return Shapes.flatten(scope, shape);
+  }
+
+  /**
+   * Flatten the operand to 1 dimension
+   *
+   * @param <T> the type of operand
+   * @param <U> the shape datatype
+   * @param scope current scope
+   * @param operand the operand to flatten
+   * @param dType the shape datatype
+   * @return the reshaped operand
+   */
+  public <T extends TType, U extends TNumber> Operand<T> flatten(Operand<T> operand,
+      Class<U> dType) {
+    return Shapes.flatten(scope, operand, dType);
+  }
+
+  /**
+   * Flatten the shape to 1 dimension.
+   *
+   * @param <U> the shape datatype
+   * @param scope current scope
+   * @param shape the TensorFlow shape
+   * @param dType the shape datatype
+   * @return the flattened shape
+   */
+  public <U extends TNumber> Operand<U> flatten(Shape<U> shape, Class<U> dType) {
+    return Shapes.flatten(scope, shape, dType);
+  }
+
+  /**
+   * Creates a 1-dimensional Operand containing the Shape's first dimension.
+   *
+   * @param scope current scope
+   * @param shape the TensorFlow shape
+   * @return a 1-dimensional Operand containing the Shape's first dimension
+   */
+  public Operand<TInt32> head(Shape<TInt32> shape) {
+    return Shapes.head(scope, shape);
+  }
+
+  /**
+   * Creates a 1-dimensional Operand containing the Shape's first dimension.
+   *
+   * @param scope current scope
+   * @param shape the TensorFlow shape
+   * @param dType the shape datatype.
+   * @param <U> the shape datatype.
+   * @return a 1-dimensional Operand containing the Shape's first dimension
+   */
+  public <U extends TNumber> Operand<U> head(Shape<U> shape, Class<U> dType) {
+    return Shapes.head(scope, shape, dType);
+  }
+
+  /**
+   * Get the number of dimensions of the shape object.
+   *
+   * @param scope current scope
+   * @param shape the shape
+   * @return the number of dimensions
+   */
+  public Operand<TInt32> numDimensions(Shape<TInt32> shape) {
+    return Shapes.numDimensions(scope, shape);
+  }
+
+  /**
+   * Get the number of dimensions of the shape object.
+   *
+   * @param <U> the shape datatype
+   * @param scope the curren scope
+   * @param shape the shape
+   * @param dType the shape datatype
+   * @return the number of dimensions
+   */
+  public <U extends TNumber> Operand<U> numDimensions(Shape<U> shape, Class<U> dType) {
+    return Shapes.numDimensions(scope, shape, dType);
+  }
+
+  /**
+   * Creates a 1-dimensional operand containing the first dimension followed by the dimensions of
+   *  the shape.
+   *
+   * @param scope current scope
+   * @param shape the TensorFlow shape
+   * @param firstDimension the dimension to prepend
+   * @return a 1-dimensional operand containing the first dimension followed by the dimensions of
+   *      the shape
+   */
+  public Operand<TInt64> prepend(Shape<TInt64> shape, long firstDimension) {
+    return Shapes.prepend(scope, shape, firstDimension);
+  }
+
+  /**
+   * Creates a 1-dimensional operand containing the first dimension followed by the dimensions of
+   *  the shape.
+   *
+   * @param scope current scope
+   * @param shape the TensorFlow shape
+   * @param firstDimension the dimension to prepend
+   * @return a 1-dimensional operand containing the first dimension followed by the dimensions of
+   *      the shape
+   */
+  public Operand<TInt32> prepend(Shape<TInt32> shape, int firstDimension) {
+    return Shapes.prepend(scope, shape, firstDimension);
+  }
+
+  /**
+   * Creates a 1-dimensional operand that represents a new shape containing the dimensions of an
+   *  operand representing the shape to prepend, followed by the dimensions of an operand
+   *  representing a shape.
+   *
+   * @param scope current scope
+   * @param shape an operand containing the dimensions of a shape
+   * @param shapeToPrepend an operand containing the dimensions of the shape to prepend
+   * @return a 1-dimensional operand that represents a new shape containing the dimensions of an
+   *      operand representing the shape to prepend, followed by the dimensions of an operand
+   *      representing the shape
+   */
+  public <T extends TNumber> Operand<T> prepend(Operand<T> shape, Operand<T> shapeToPrepend) {
+    return Shapes.prepend(scope, shape, shapeToPrepend);
+  }
+
+  /**
+   * Reshapes the operand by reducing the shape to the specified axis.
+   *
+   * @param <T> the type of Operand
+   * @param scope current scope
+   * @param operand the operand
+   * @param axis the axis
+   * @return the reshaped operand
+   */
+  public <T extends TType> Operand<T> reduceDims(Operand<T> operand, Operand<TInt32> axis) {
+    return Shapes.reduceDims(scope, operand, axis);
+  }
+
+  /**
+   * Reduces the shape to the specified axis.
+   *
+   * @param scope current scope
+   * @param shape the TensorFlow shape
+   * @param axis the axis
+   * @return an operand containing the dimensions for the reduced shape
+   */
+  public Operand<TInt32> reduceDims(Shape<TInt32> shape, Operand<TInt32> axis) {
+    return Shapes.reduceDims(scope, shape, axis);
+  }
+
+  /**
+   * Reshapes the operand by reducing the shape to the specified axis.
+   *
+   * @param <T> the type of Operand
+   * @param <U> the shape datatype
+   * @param scope current scope
+   * @param operand the operand
+   * @param axis the axis
+   * @param dType the shape datatype
+   * @return the reshaped operand
+   */
+  public <T extends TType, U extends TNumber> Operand<T> reduceDims(Operand<T> operand,
+      Operand<U> axis, Class<U> dType) {
+    return Shapes.reduceDims(scope, operand, axis, dType);
+  }
+
+  /**
+   * Reduces the shape to the specified axis.
+   *
+   * @param <U> the shape datatype
+   * @param scope current scope
+   * @param shape the TensorFlow shape
+   * @param axis the axis
+   * @param dType the shape datatype
+   * @return the reduced shape
+   */
+  public <U extends TNumber> Operand<U> reduceDims(Shape<U> shape, Operand<U> axis,
+      Class<U> dType) {
+    return Shapes.reduceDims(scope, shape, axis, dType);
+  }
+
+  /**
+   * Get the size represented by the TensorFlow shape.
+   *
+   * @param scope current scope
+   * @param shape the TensorFlow shape
+   * @return the size
+   */
+  public Operand<TInt32> size(Shape<TInt32> shape) {
+    return Shapes.size(scope, shape);
+  }
+
+  /**
+   * Get the size of the specified dimension for the shape of the tensor.
+   *
+   * @param scope current scope
+   * @param input the operand
+   * @param dim the dimension
+   * @return the size of the specified dimension
+   */
+  public <T extends TType> Operand<TInt32> size(Operand<T> input, Operand<TInt32> dim) {
+    return Shapes.size(scope, input, dim);
+  }
+
+  /**
+   * Get the size represented by the TensorFlow shape.
+   *
+   * @param <U> the type of the shape
+   * @param scope current scope
+   * @param shape the TensorFlow shape
+   * @param dType the shape datatype
+   * @return the size
+   */
+  public <U extends TNumber> Operand<U> size(Shape<U> shape, Class<U> dType) {
+    return Shapes.size(scope, shape, dType);
+  }
+
+  /**
+   * Get the size of the specified dimension in the shape.
+   *
+   * @param scope current scope
+   * @param shape the TensorFlow shape
+   * @param dim the dimension
+   * @return the size of the specified dimension
+   */
+  public Operand<TInt32> size(Shape<TInt32> shape, Operand<TInt32> dim) {
+    return Shapes.size(scope, shape, dim);
+  }
+
+  /**
+   * Get the size of the specified dimension for the shape of the tensor.
+   *
+   * @param <U> the shape datatype
+   * @param scope current scope
+   * @param input the operand
+   * @param dim the dimension
+   * @param dType the shape datatype
+   * @return the size of the specified dimension
+   */
+  public <T extends TType, U extends TNumber> Operand<U> size(Operand<T> input, Operand<U> dim,
+      Class<U> dType) {
+    return Shapes.size(scope, input, dim, dType);
+  }
+
+  /**
+   * Get the size of the specified dimension in the shape.
+   *
+   * @param <U> the shape datatype
+   * @param scope current scope
+   * @param shape the TensorFlow shape
+   * @param dim the dimension
+   * @param dType the shape datatype
+   * @return the size of the specified dimension
+   */
+  public <U extends TNumber> Operand<U> size(Shape<U> shape, Operand<U> dim, Class<U> dType) {
+    return Shapes.size(scope, shape, dim, dType);
+  }
+
+  /**
+   * Removes dimensions of size 1 from the shape.
+   *
+   * @param scope current scope
+   * @param shape the TensorFlow shape
+   * @return the squeezed shape
+   */
+  public Operand<TInt32> squeeze(Shape<TInt32> shape) {
+    return Shapes.squeeze(scope, shape);
+  }
+
+  /**
+   * Removes dimensions of size 1 from the shape.
+   *
+   * @param <U> the shape datatype.
+   * @param scope current scope
+   * @param shape the TensorFlow shape
+   * @param dType the shape datatype.
+   * @return the squeezed shape
+   */
+  public <U extends TNumber> Operand<U> squeeze(Shape<U> shape, Class<U> dType) {
+    return Shapes.squeeze(scope, shape, dType);
+  }
+
+  /**
+   * Creates a 1-dimensional Operand that contains the dimension matching the last dimension of the
+   *  Shape.
+   *
+   * @param scope current scope
+   * @param shape the TensorFlow shape
+   * @return a 1-dimensional Operand that contains the dimension matching the last dimension of the
+   *      Shape
+   */
+  public Operand<TInt32> tail(Shape<TInt32> shape) {
+    return Shapes.tail(scope, shape);
+  }
+
+  /**
+   * Creates a 1-dimensional Operand that contains the dimension matching the last dimension of *
+   *  the Shape.
+   *
+   * @param scope current scope
+   * @param shape the TensorFlow shape
+   * @param dType the shape datatype.
+   * @param <U> the shape datatype.
+   * @return a 1-dimensional Operand that contains the dimension matching the last dimension of the
+   *      Shape
+   */
+  public <U extends TNumber> Operand<U> tail(Shape<U> shape, Class<U> dType) {
+    return Shapes.tail(scope, shape, dType);
+  }
+
+  /**
+   * Creates a 1-dimensional operand with the dimensions matching the first n dimensions of the
+   *  shape.
+   *
+   * @param scope current scope
+   * @param shape the TensorFlow shape
+   * @param n the number of leading dimensions to get, must be <= than the shape's numDimensions()
+   * @return a 1-dimensional operand with the dimensions matching the first n dimensions of the
+   *      shape
+   */
+  public Operand<TInt32> take(Shape<TInt32> shape, Operand<TInt32> n) {
+    return Shapes.take(scope, shape, n);
+  }
+
+  /**
+   * Creates a 1-dimensional operand containin the dimensions matching the first n dimensions of the
+   *  shape.
+   *
+   * @param scope current scope
+   * @param shape the TensorFlow shape
+   * @param n the number of leading dimensions to get, must be <= than the shape's numDimensions()
+   * @param dType the shape datatype.
+   * @param <U> the shape datatype.
+   * @return a 1-dimensional operand with the dimensions matching * the first n dimensions of the
+   *      shape
+   */
+  public <U extends TNumber> Operand<U> take(Shape<U> shape, Operand<U> n, Class<U> dType) {
+    return Shapes.take(scope, shape, n, dType);
+  }
+
+  /**
+   * Creates a 1-dimensional operand containing the dimensions matching the last n dimensions of the
+   *  shape.
+   *
+   * @param scope current scope
+   * @param shape the TensorFlow shape
+   * @param n the number of leading dimensions to get, must be <= than the shape's numDimensions()
+   * @return a 1-dimensional operand containing the dimensions matching the last n dimensions of the
+   *      shape
+   */
+  public <U extends TNumber> Operand<TInt32> takeLast(Shape<TInt32> shape, Operand<TInt32> n) {
+    return Shapes.takeLast(scope, shape, n);
+  }
+
+  /**
+   * Creates a 1-dimensional operand containing the dimensions matching the last n dimensions of the
+   *  shape.
+   *
+   * @param scope current scope
+   * @param shape the TensorFlow shape
+   * @param n the number of leading dimensions to get, must be <= than the shape's numDimensions()
+   * @param dType the shape datatype.
+   * @param <U> the shape datatype.
+   * @return a 1-dimensional operand containing the dimensions matching the last n dimensions of the
+   *      shape
+   */
+  public <U extends TNumber> Operand<U> takeLast(Shape<U> shape, Operand<U> n, Class<U> dType) {
+    return Shapes.takeLast(scope, shape, n, dType);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/SignalOps.java b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/SignalOps.java
new file mode 100644
index 00000000000..1466dcd8de4
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/SignalOps.java
@@ -0,0 +1,436 @@
+// Copyright 2020 The TensorFlow Authors. All Rights Reserved.
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+// ==============================================================================
+//
+// This class has been generated, DO NOT EDIT!
+//
+package org.tensorflow.op;
+
+import org.tensorflow.Operand;
+import org.tensorflow.op.signal.BatchFft;
+import org.tensorflow.op.signal.BatchFft2d;
+import org.tensorflow.op.signal.BatchFft3d;
+import org.tensorflow.op.signal.BatchIfft;
+import org.tensorflow.op.signal.BatchIfft2d;
+import org.tensorflow.op.signal.BatchIfft3d;
+import org.tensorflow.op.signal.Fft;
+import org.tensorflow.op.signal.Fft2d;
+import org.tensorflow.op.signal.Fft3d;
+import org.tensorflow.op.signal.Ifft;
+import org.tensorflow.op.signal.Ifft2d;
+import org.tensorflow.op.signal.Ifft3d;
+import org.tensorflow.op.signal.Irfft;
+import org.tensorflow.op.signal.Irfft2d;
+import org.tensorflow.op.signal.Irfft3d;
+import org.tensorflow.op.signal.Rfft;
+import org.tensorflow.op.signal.Rfft2d;
+import org.tensorflow.op.signal.Rfft3d;
+import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * An API for building {@code signal} operations as {@link Op Op}s
+ *
+ * @see {@link Ops}
+ */
+public final class SignalOps {
+  private final Scope scope;
+
+  SignalOps(Scope scope) {
+    this.scope = scope;
+  }
+
+  /**
+   *
+   * @param input
+   * @return a new instance of BatchFft
+   */
+  public BatchFft batchFft(Operand<?> input) {
+    return BatchFft.create(scope, input);
+  }
+
+  /**
+   *
+   * @param input
+   * @return a new instance of BatchFft2d
+   */
+  public BatchFft2d batchFft2d(Operand<?> input) {
+    return BatchFft2d.create(scope, input);
+  }
+
+  /**
+   *
+   * @param input
+   * @return a new instance of BatchFft3d
+   */
+  public BatchFft3d batchFft3d(Operand<?> input) {
+    return BatchFft3d.create(scope, input);
+  }
+
+  /**
+   *
+   * @param input
+   * @return a new instance of BatchIfft
+   */
+  public BatchIfft batchIfft(Operand<?> input) {
+    return BatchIfft.create(scope, input);
+  }
+
+  /**
+   *
+   * @param input
+   * @return a new instance of BatchIfft2d
+   */
+  public BatchIfft2d batchIfft2d(Operand<?> input) {
+    return BatchIfft2d.create(scope, input);
+  }
+
+  /**
+   *
+   * @param input
+   * @return a new instance of BatchIfft3d
+   */
+  public BatchIfft3d batchIfft3d(Operand<?> input) {
+    return BatchIfft3d.create(scope, input);
+  }
+
+  /**
+   * Fast Fourier transform.
+   *  <p>
+   *  Computes the 1-dimensional discrete Fourier transform over the inner-most
+   *  dimension of `input`.
+   *
+   * @param <T> data type for {@code output()} output
+   * @param input A complex tensor.
+   * @return a new instance of Fft
+   */
+  public <T extends TType> Fft<T> fft(Operand<T> input) {
+    return Fft.create(scope, input);
+  }
+
+  /**
+   * 2D fast Fourier transform.
+   *  <p>
+   *  Computes the 2-dimensional discrete Fourier transform over the inner-most
+   *  2 dimensions of `input`.
+   *
+   * @param <T> data type for {@code output()} output
+   * @param input A complex tensor.
+   * @return a new instance of Fft2d
+   */
+  public <T extends TType> Fft2d<T> fft2d(Operand<T> input) {
+    return Fft2d.create(scope, input);
+  }
+
+  /**
+   * 3D fast Fourier transform.
+   *  <p>
+   *  Computes the 3-dimensional discrete Fourier transform over the inner-most 3
+   *  dimensions of `input`.
+   *
+   * @param <T> data type for {@code output()} output
+   * @param input A complex tensor.
+   * @return a new instance of Fft3d
+   */
+  public <T extends TType> Fft3d<T> fft3d(Operand<T> input) {
+    return Fft3d.create(scope, input);
+  }
+
+  /**
+   * Inverse fast Fourier transform.
+   *  <p>
+   *  Computes the inverse 1-dimensional discrete Fourier transform over the
+   *  inner-most dimension of `input`.
+   *
+   * @param <T> data type for {@code output()} output
+   * @param input A complex tensor.
+   * @return a new instance of Ifft
+   */
+  public <T extends TType> Ifft<T> ifft(Operand<T> input) {
+    return Ifft.create(scope, input);
+  }
+
+  /**
+   * Inverse 2D fast Fourier transform.
+   *  <p>
+   *  Computes the inverse 2-dimensional discrete Fourier transform over the
+   *  inner-most 2 dimensions of `input`.
+   *
+   * @param <T> data type for {@code output()} output
+   * @param input A complex tensor.
+   * @return a new instance of Ifft2d
+   */
+  public <T extends TType> Ifft2d<T> ifft2d(Operand<T> input) {
+    return Ifft2d.create(scope, input);
+  }
+
+  /**
+   * Inverse 3D fast Fourier transform.
+   *  <p>
+   *  Computes the inverse 3-dimensional discrete Fourier transform over the
+   *  inner-most 3 dimensions of `input`.
+   *
+   * @param <T> data type for {@code output()} output
+   * @param input A complex tensor.
+   * @return a new instance of Ifft3d
+   */
+  public <T extends TType> Ifft3d<T> ifft3d(Operand<T> input) {
+    return Ifft3d.create(scope, input);
+  }
+
+  /**
+   * Inverse real-valued fast Fourier transform.
+   *  <p>
+   *  Computes the inverse 1-dimensional discrete Fourier transform of a real-valued
+   *  signal over the inner-most dimension of `input`.
+   *  <p>
+   *  The inner-most dimension of `input` is assumed to be the result of `RFFT`: the
+   *  `fft_length / 2 + 1` unique components of the DFT of a real-valued signal. If
+   *  `fft_length` is not provided, it is computed from the size of the inner-most
+   *  dimension of `input` (`fft_length = 2 * (inner - 1)`). If the FFT length used to
+   *  compute `input` is odd, it should be provided since it cannot be inferred
+   *  properly.
+   *  <p>
+   *  Along the axis `signal.Irfft` is computed on, if `fft_length / 2 + 1` is smaller
+   *  than the corresponding dimension of `input`, the dimension is cropped. If it is
+   *  larger, the dimension is padded with zeros.
+   *
+   * @param <U> data type for {@code output()} output
+   * @param input A complex tensor.
+   * @param fftLength An int32 tensor of shape [1]. The FFT length.
+   * @return a new instance of Irfft
+   */
+  public <T extends TType> Irfft<TFloat32> irfft(Operand<T> input, Operand<TInt32> fftLength) {
+    return Irfft.create(scope, input, fftLength);
+  }
+
+  /**
+   * Inverse real-valued fast Fourier transform.
+   *  <p>
+   *  Computes the inverse 1-dimensional discrete Fourier transform of a real-valued
+   *  signal over the inner-most dimension of `input`.
+   *  <p>
+   *  The inner-most dimension of `input` is assumed to be the result of `RFFT`: the
+   *  `fft_length / 2 + 1` unique components of the DFT of a real-valued signal. If
+   *  `fft_length` is not provided, it is computed from the size of the inner-most
+   *  dimension of `input` (`fft_length = 2 * (inner - 1)`). If the FFT length used to
+   *  compute `input` is odd, it should be provided since it cannot be inferred
+   *  properly.
+   *  <p>
+   *  Along the axis `signal.Irfft` is computed on, if `fft_length / 2 + 1` is smaller
+   *  than the corresponding dimension of `input`, the dimension is cropped. If it is
+   *  larger, the dimension is padded with zeros.
+   *
+   * @param <U> data type for {@code output()} output
+   * @param input A complex tensor.
+   * @param fftLength An int32 tensor of shape [1]. The FFT length.
+   * @param Treal
+   * @return a new instance of Irfft
+   */
+  public <U extends TNumber, T extends TType> Irfft<U> irfft(Operand<T> input,
+      Operand<TInt32> fftLength, Class<U> Treal) {
+    return Irfft.create(scope, input, fftLength, Treal);
+  }
+
+  /**
+   * Inverse 2D real-valued fast Fourier transform.
+   *  <p>
+   *  Computes the inverse 2-dimensional discrete Fourier transform of a real-valued
+   *  signal over the inner-most 2 dimensions of `input`.
+   *  <p>
+   *  The inner-most 2 dimensions of `input` are assumed to be the result of `RFFT2D`:
+   *  The inner-most dimension contains the `fft_length / 2 + 1` unique components of
+   *  the DFT of a real-valued signal. If `fft_length` is not provided, it is computed
+   *  from the size of the inner-most 2 dimensions of `input`. If the FFT length used
+   *  to compute `input` is odd, it should be provided since it cannot be inferred
+   *  properly.
+   *  <p>
+   *  Along each axis `signal.Irfft2d` is computed on, if `fft_length` (or
+   *  `fft_length / 2 + 1` for the inner-most dimension) is smaller than the
+   *  corresponding dimension of `input`, the dimension is cropped. If it is larger,
+   *  the dimension is padded with zeros.
+   *
+   * @param <U> data type for {@code output()} output
+   * @param input A complex tensor.
+   * @param fftLength An int32 tensor of shape [2]. The FFT length for each dimension.
+   * @return a new instance of Irfft2d
+   */
+  public <T extends TType> Irfft2d<TFloat32> irfft2d(Operand<T> input, Operand<TInt32> fftLength) {
+    return Irfft2d.create(scope, input, fftLength);
+  }
+
+  /**
+   * Inverse 2D real-valued fast Fourier transform.
+   *  <p>
+   *  Computes the inverse 2-dimensional discrete Fourier transform of a real-valued
+   *  signal over the inner-most 2 dimensions of `input`.
+   *  <p>
+   *  The inner-most 2 dimensions of `input` are assumed to be the result of `RFFT2D`:
+   *  The inner-most dimension contains the `fft_length / 2 + 1` unique components of
+   *  the DFT of a real-valued signal. If `fft_length` is not provided, it is computed
+   *  from the size of the inner-most 2 dimensions of `input`. If the FFT length used
+   *  to compute `input` is odd, it should be provided since it cannot be inferred
+   *  properly.
+   *  <p>
+   *  Along each axis `signal.Irfft2d` is computed on, if `fft_length` (or
+   *  `fft_length / 2 + 1` for the inner-most dimension) is smaller than the
+   *  corresponding dimension of `input`, the dimension is cropped. If it is larger,
+   *  the dimension is padded with zeros.
+   *
+   * @param <U> data type for {@code output()} output
+   * @param input A complex tensor.
+   * @param fftLength An int32 tensor of shape [2]. The FFT length for each dimension.
+   * @param Treal
+   * @return a new instance of Irfft2d
+   */
+  public <U extends TNumber, T extends TType> Irfft2d<U> irfft2d(Operand<T> input,
+      Operand<TInt32> fftLength, Class<U> Treal) {
+    return Irfft2d.create(scope, input, fftLength, Treal);
+  }
+
+  /**
+   * Inverse 3D real-valued fast Fourier transform.
+   *  <p>
+   *  Computes the inverse 3-dimensional discrete Fourier transform of a real-valued
+   *  signal over the inner-most 3 dimensions of `input`.
+   *  <p>
+   *  The inner-most 3 dimensions of `input` are assumed to be the result of `RFFT3D`:
+   *  The inner-most dimension contains the `fft_length / 2 + 1` unique components of
+   *  the DFT of a real-valued signal. If `fft_length` is not provided, it is computed
+   *  from the size of the inner-most 3 dimensions of `input`. If the FFT length used
+   *  to compute `input` is odd, it should be provided since it cannot be inferred
+   *  properly.
+   *  <p>
+   *  Along each axis `signal.Irfft3d` is computed on, if `fft_length` (or
+   *  `fft_length / 2 + 1` for the inner-most dimension) is smaller than the
+   *  corresponding dimension of `input`, the dimension is cropped. If it is larger,
+   *  the dimension is padded with zeros.
+   *
+   * @param <U> data type for {@code output()} output
+   * @param input A complex tensor.
+   * @param fftLength An int32 tensor of shape [3]. The FFT length for each dimension.
+   * @return a new instance of Irfft3d
+   */
+  public <T extends TType> Irfft3d<TFloat32> irfft3d(Operand<T> input, Operand<TInt32> fftLength) {
+    return Irfft3d.create(scope, input, fftLength);
+  }
+
+  /**
+   * Inverse 3D real-valued fast Fourier transform.
+   *  <p>
+   *  Computes the inverse 3-dimensional discrete Fourier transform of a real-valued
+   *  signal over the inner-most 3 dimensions of `input`.
+   *  <p>
+   *  The inner-most 3 dimensions of `input` are assumed to be the result of `RFFT3D`:
+   *  The inner-most dimension contains the `fft_length / 2 + 1` unique components of
+   *  the DFT of a real-valued signal. If `fft_length` is not provided, it is computed
+   *  from the size of the inner-most 3 dimensions of `input`. If the FFT length used
+   *  to compute `input` is odd, it should be provided since it cannot be inferred
+   *  properly.
+   *  <p>
+   *  Along each axis `signal.Irfft3d` is computed on, if `fft_length` (or
+   *  `fft_length / 2 + 1` for the inner-most dimension) is smaller than the
+   *  corresponding dimension of `input`, the dimension is cropped. If it is larger,
+   *  the dimension is padded with zeros.
+   *
+   * @param <U> data type for {@code output()} output
+   * @param input A complex tensor.
+   * @param fftLength An int32 tensor of shape [3]. The FFT length for each dimension.
+   * @param Treal
+   * @return a new instance of Irfft3d
+   */
+  public <U extends TNumber, T extends TType> Irfft3d<U> irfft3d(Operand<T> input,
+      Operand<TInt32> fftLength, Class<U> Treal) {
+    return Irfft3d.create(scope, input, fftLength, Treal);
+  }
+
+  /**
+   * Real-valued fast Fourier transform.
+   *  <p>
+   *  Computes the 1-dimensional discrete Fourier transform of a real-valued signal
+   *  over the inner-most dimension of `input`.
+   *  <p>
+   *  Since the DFT of a real signal is Hermitian-symmetric, `signal.Rfft` only returns the
+   *  `fft_length / 2 + 1` unique components of the FFT: the zero-frequency term,
+   *  followed by the `fft_length / 2` positive-frequency terms.
+   *  <p>
+   *  Along the axis `signal.Rfft` is computed on, if `fft_length` is smaller than the
+   *  corresponding dimension of `input`, the dimension is cropped. If it is larger,
+   *  the dimension is padded with zeros.
+   *
+   * @param <U> data type for {@code output()} output
+   * @param input A float32 tensor.
+   * @param fftLength An int32 tensor of shape [1]. The FFT length.
+   * @param Tcomplex
+   * @return a new instance of Rfft
+   */
+  public <U extends TType, T extends TNumber> Rfft<U> rfft(Operand<T> input,
+      Operand<TInt32> fftLength, Class<U> Tcomplex) {
+    return Rfft.create(scope, input, fftLength, Tcomplex);
+  }
+
+  /**
+   * 2D real-valued fast Fourier transform.
+   *  <p>
+   *  Computes the 2-dimensional discrete Fourier transform of a real-valued signal
+   *  over the inner-most 2 dimensions of `input`.
+   *  <p>
+   *  Since the DFT of a real signal is Hermitian-symmetric, `signal.Rfft2d` only returns the
+   *  `fft_length / 2 + 1` unique components of the FFT for the inner-most dimension
+   *  of `output`: the zero-frequency term, followed by the `fft_length / 2`
+   *  positive-frequency terms.
+   *  <p>
+   *  Along each axis `signal.Rfft2d` is computed on, if `fft_length` is smaller than the
+   *  corresponding dimension of `input`, the dimension is cropped. If it is larger,
+   *  the dimension is padded with zeros.
+   *
+   * @param <U> data type for {@code output()} output
+   * @param input A float32 tensor.
+   * @param fftLength An int32 tensor of shape [2]. The FFT length for each dimension.
+   * @param Tcomplex
+   * @return a new instance of Rfft2d
+   */
+  public <U extends TType, T extends TNumber> Rfft2d<U> rfft2d(Operand<T> input,
+      Operand<TInt32> fftLength, Class<U> Tcomplex) {
+    return Rfft2d.create(scope, input, fftLength, Tcomplex);
+  }
+
+  /**
+   * 3D real-valued fast Fourier transform.
+   *  <p>
+   *  Computes the 3-dimensional discrete Fourier transform of a real-valued signal
+   *  over the inner-most 3 dimensions of `input`.
+   *  <p>
+   *  Since the DFT of a real signal is Hermitian-symmetric, `signal.Rfft3d` only returns the
+   *  `fft_length / 2 + 1` unique components of the FFT for the inner-most dimension
+   *  of `output`: the zero-frequency term, followed by the `fft_length / 2`
+   *  positive-frequency terms.
+   *  <p>
+   *  Along each axis `signal.Rfft3d` is computed on, if `fft_length` is smaller than the
+   *  corresponding dimension of `input`, the dimension is cropped. If it is larger,
+   *  the dimension is padded with zeros.
+   *
+   * @param <U> data type for {@code output()} output
+   * @param input A float32 tensor.
+   * @param fftLength An int32 tensor of shape [3]. The FFT length for each dimension.
+   * @param Tcomplex
+   * @return a new instance of Rfft3d
+   */
+  public <U extends TType, T extends TNumber> Rfft3d<U> rfft3d(Operand<T> input,
+      Operand<TInt32> fftLength, Class<U> Tcomplex) {
+    return Rfft3d.create(scope, input, fftLength, Tcomplex);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/SparseOps.java b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/SparseOps.java
new file mode 100644
index 00000000000..d7690dacdb4
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/SparseOps.java
@@ -0,0 +1,1499 @@
+// Copyright 2020 The TensorFlow Authors. All Rights Reserved.
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+// ==============================================================================
+//
+// This class has been generated, DO NOT EDIT!
+//
+package org.tensorflow.op;
+
+import org.tensorflow.Operand;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.sparse.AddManySparseToTensorsMap;
+import org.tensorflow.op.sparse.AddSparseToTensorsMap;
+import org.tensorflow.op.sparse.DenseToDenseSetOperation;
+import org.tensorflow.op.sparse.DenseToSparseSetOperation;
+import org.tensorflow.op.sparse.DeserializeSparse;
+import org.tensorflow.op.sparse.SparseAccumulatorApplyGradient;
+import org.tensorflow.op.sparse.SparseAccumulatorTakeGradient;
+import org.tensorflow.op.sparse.SparseAdd;
+import org.tensorflow.op.sparse.SparseAddGrad;
+import org.tensorflow.op.sparse.SparseBincount;
+import org.tensorflow.op.sparse.SparseConcat;
+import org.tensorflow.op.sparse.SparseConditionalAccumulator;
+import org.tensorflow.op.sparse.SparseCross;
+import org.tensorflow.op.sparse.SparseCrossHashed;
+import org.tensorflow.op.sparse.SparseDenseCwiseAdd;
+import org.tensorflow.op.sparse.SparseDenseCwiseDiv;
+import org.tensorflow.op.sparse.SparseDenseCwiseMul;
+import org.tensorflow.op.sparse.SparseFillEmptyRows;
+import org.tensorflow.op.sparse.SparseFillEmptyRowsGrad;
+import org.tensorflow.op.sparse.SparseMatMul;
+import org.tensorflow.op.sparse.SparseReduceMax;
+import org.tensorflow.op.sparse.SparseReduceMaxSparse;
+import org.tensorflow.op.sparse.SparseReduceSum;
+import org.tensorflow.op.sparse.SparseReduceSumSparse;
+import org.tensorflow.op.sparse.SparseReorder;
+import org.tensorflow.op.sparse.SparseReshape;
+import org.tensorflow.op.sparse.SparseSegmentMean;
+import org.tensorflow.op.sparse.SparseSegmentMeanGrad;
+import org.tensorflow.op.sparse.SparseSegmentMeanWithNumSegments;
+import org.tensorflow.op.sparse.SparseSegmentSqrtN;
+import org.tensorflow.op.sparse.SparseSegmentSqrtNGrad;
+import org.tensorflow.op.sparse.SparseSegmentSqrtNWithNumSegments;
+import org.tensorflow.op.sparse.SparseSegmentSum;
+import org.tensorflow.op.sparse.SparseSegmentSumWithNumSegments;
+import org.tensorflow.op.sparse.SparseSlice;
+import org.tensorflow.op.sparse.SparseSliceGrad;
+import org.tensorflow.op.sparse.SparseSoftmax;
+import org.tensorflow.op.sparse.SparseSparseMaximum;
+import org.tensorflow.op.sparse.SparseSparseMinimum;
+import org.tensorflow.op.sparse.SparseSplit;
+import org.tensorflow.op.sparse.SparseTensorDenseAdd;
+import org.tensorflow.op.sparse.SparseTensorDenseMatMul;
+import org.tensorflow.op.sparse.SparseToDense;
+import org.tensorflow.op.sparse.SparseToSparseSetOperation;
+import org.tensorflow.op.sparse.TakeManySparseFromTensorsMap;
+import org.tensorflow.types.TBool;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.TString;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * An API for building {@code sparse} operations as {@link Op Op}s
+ *
+ * @see {@link Ops}
+ */
+public final class SparseOps {
+  private final Scope scope;
+
+  SparseOps(Scope scope) {
+    this.scope = scope;
+  }
+
+  /**
+   * Add an `N`-minibatch `SparseTensor` to a `SparseTensorsMap`, return `N` handles.
+   *  <p>
+   *  A `SparseTensor` of rank `R` is represented by three tensors: `sparse_indices`,
+   *  `sparse_values`, and `sparse_shape`, where
+   *  <pre>{@code
+   *  sparse_indices.shape[1] == sparse_shape.shape[0] == R}</pre>
+   *  An `N`-minibatch of `SparseTensor` objects is represented as a `SparseTensor`
+   *  having a first `sparse_indices` column taking values between `[0, N)`, where
+   *  the minibatch size `N == sparse_shape[0]`.
+   *  <p>
+   *  The input `SparseTensor` must have rank `R` greater than 1, and the first
+   *  dimension is treated as the minibatch dimension.  Elements of the `SparseTensor`
+   *  must be sorted in increasing order of this first dimension.  The stored
+   *  `SparseTensor` objects pointed to by each row of the output `sparse_handles`
+   *  will have rank `R-1`.
+   *  <p>
+   *  The `SparseTensor` values can then be read out as part of a minibatch by passing
+   *  the given keys as vector elements to `TakeManySparseFromTensorsMap`.  To ensure
+   *  the correct `SparseTensorsMap` is accessed, ensure that the same
+   *  `container` and `shared_name` are passed to that Op.  If no `shared_name`
+   *  is provided here, instead use the <i>name</i> of the Operation created by calling
+   *  `sparse.AddManySparseToTensorsMap` as the `shared_name` passed to
+   *  `TakeManySparseFromTensorsMap`.  Ensure the Operations are colocated.
+   *
+   * @param sparseIndices 2-D.  The `indices` of the minibatch `SparseTensor`.
+   *  `sparse_indices[:, 0]` must be ordered values in `[0, N)`.
+   * @param sparseValues 1-D.  The `values` of the minibatch `SparseTensor`.
+   * @param sparseShape 1-D.  The `shape` of the minibatch `SparseTensor`.
+   *  The minibatch size `N == sparse_shape[0]`.
+   * @param options carries optional attributes values
+   * @return a new instance of AddManySparseToTensorsMap
+   */
+  public <T extends TType> AddManySparseToTensorsMap addManySparseToTensorsMap(
+      Operand<TInt64> sparseIndices, Operand<T> sparseValues, Operand<TInt64> sparseShape,
+      AddManySparseToTensorsMap.Options... options) {
+    return AddManySparseToTensorsMap.create(scope, sparseIndices, sparseValues, sparseShape, options);
+  }
+
+  /**
+   * Add a `SparseTensor` to a `SparseTensorsMap` return its handle.
+   *  <p>
+   *  A `SparseTensor` is represented by three tensors: `sparse_indices`,
+   *  `sparse_values`, and `sparse_shape`.
+   *  <p>
+   *  This operator takes the given `SparseTensor` and adds it to a container
+   *  object (a `SparseTensorsMap`).  A unique key within this container is generated
+   *  in the form of an `int64`, and this is the value that is returned.
+   *  <p>
+   *  The `SparseTensor` can then be read out as part of a minibatch by passing
+   *  the key as a vector element to `TakeManySparseFromTensorsMap`.  To ensure
+   *  the correct `SparseTensorsMap` is accessed, ensure that the same
+   *  `container` and `shared_name` are passed to that Op.  If no `shared_name`
+   *  is provided here, instead use the <i>name</i> of the Operation created by calling
+   *  `sparse.AddSparseToTensorsMap` as the `shared_name` passed to
+   *  `TakeManySparseFromTensorsMap`.  Ensure the Operations are colocated.
+   *
+   * @param sparseIndices 2-D.  The `indices` of the `SparseTensor`.
+   * @param sparseValues 1-D.  The `values` of the `SparseTensor`.
+   * @param sparseShape 1-D.  The `shape` of the `SparseTensor`.
+   * @param options carries optional attributes values
+   * @return a new instance of AddSparseToTensorsMap
+   */
+  public <T extends TType> AddSparseToTensorsMap addSparseToTensorsMap(
+      Operand<TInt64> sparseIndices, Operand<T> sparseValues, Operand<TInt64> sparseShape,
+      AddSparseToTensorsMap.Options... options) {
+    return AddSparseToTensorsMap.create(scope, sparseIndices, sparseValues, sparseShape, options);
+  }
+
+  /**
+   * Applies set operation along last dimension of 2 `Tensor` inputs.
+   *  <p>
+   *  See SetOperationOp::SetOperationFromContext for values of `set_operation`.
+   *  <p>
+   *  Output `result` is a `SparseTensor` represented by `result_indices`,
+   *  `result_values`, and `result_shape`. For `set1` and `set2` ranked `n`, this
+   *  has rank `n` and the same 1st `n-1` dimensions as `set1` and `set2`. The `nth`
+   *  dimension contains the result of `set_operation` applied to the corresponding
+   *  `[0...n-1]` dimension of `set`.
+   *
+   * @param <T> data type for {@code resultValues()} output
+   * @param set1 `Tensor` with rank `n`. 1st `n-1` dimensions must be the same as `set2`.
+   *  Dimension `n` contains values in a set, duplicates are allowed but ignored.
+   * @param set2 `Tensor` with rank `n`. 1st `n-1` dimensions must be the same as `set1`.
+   *  Dimension `n` contains values in a set, duplicates are allowed but ignored.
+   * @param setOperation
+   * @param options carries optional attributes values
+   * @return a new instance of DenseToDenseSetOperation
+   */
+  public <T extends TType> DenseToDenseSetOperation<T> denseToDenseSetOperation(Operand<T> set1,
+      Operand<T> set2, String setOperation, DenseToDenseSetOperation.Options... options) {
+    return DenseToDenseSetOperation.create(scope, set1, set2, setOperation, options);
+  }
+
+  /**
+   * Applies set operation along last dimension of `Tensor` and `SparseTensor`.
+   *  <p>
+   *  See SetOperationOp::SetOperationFromContext for values of `set_operation`.
+   *  <p>
+   *  Input `set2` is a `SparseTensor` represented by `set2_indices`, `set2_values`,
+   *  and `set2_shape`. For `set2` ranked `n`, 1st `n-1` dimensions must be the same
+   *  as `set1`. Dimension `n` contains values in a set, duplicates are allowed but
+   *  ignored.
+   *  <p>
+   *  If `validate_indices` is `True`, this op validates the order and range of `set2`
+   *  indices.
+   *  <p>
+   *  Output `result` is a `SparseTensor` represented by `result_indices`,
+   *  `result_values`, and `result_shape`. For `set1` and `set2` ranked `n`, this
+   *  has rank `n` and the same 1st `n-1` dimensions as `set1` and `set2`. The `nth`
+   *  dimension contains the result of `set_operation` applied to the corresponding
+   *  `[0...n-1]` dimension of `set`.
+   *
+   * @param <T> data type for {@code resultValues()} output
+   * @param set1 `Tensor` with rank `n`. 1st `n-1` dimensions must be the same as `set2`.
+   *  Dimension `n` contains values in a set, duplicates are allowed but ignored.
+   * @param set2Indices 2D `Tensor`, indices of a `SparseTensor`. Must be in row-major
+   *  order.
+   * @param set2Values 1D `Tensor`, values of a `SparseTensor`. Must be in row-major
+   *  order.
+   * @param set2Shape 1D `Tensor`, shape of a `SparseTensor`. `set2_shape[0...n-1]` must
+   *  be the same as the 1st `n-1` dimensions of `set1`, `result_shape[n]` is the
+   *  max set size across `n-1` dimensions.
+   * @param setOperation
+   * @param options carries optional attributes values
+   * @return a new instance of DenseToSparseSetOperation
+   */
+  public <T extends TType> DenseToSparseSetOperation<T> denseToSparseSetOperation(Operand<T> set1,
+      Operand<TInt64> set2Indices, Operand<T> set2Values, Operand<TInt64> set2Shape,
+      String setOperation, DenseToSparseSetOperation.Options... options) {
+    return DenseToSparseSetOperation.create(scope, set1, set2Indices, set2Values, set2Shape, setOperation, options);
+  }
+
+  /**
+   * Deserialize `SparseTensor` objects.
+   *  <p>
+   *  The input `serialized_sparse` must have the shape `[?, ?, ..., ?, 3]` where
+   *  the last dimension stores serialized `SparseTensor` objects and the other N
+   *  dimensions (N >= 0) correspond to a batch. The ranks of the original
+   *  `SparseTensor` objects must all match. When the final `SparseTensor` is
+   *  created, its rank is the rank of the incoming `SparseTensor` objects plus N;
+   *  the sparse tensors have been concatenated along new dimensions, one for each
+   *  batch.
+   *  <p>
+   *  The output `SparseTensor` object's shape values for the original dimensions
+   *  are the max across the input `SparseTensor` objects' shape values for the
+   *  corresponding dimensions. The new dimensions match the size of the batch.
+   *  <p>
+   *  The input `SparseTensor` objects' indices are assumed ordered in
+   *  standard lexicographic order.  If this is not the case, after this
+   *  step run `SparseReorder` to restore index ordering.
+   *  <p>
+   *  For example, if the serialized input is a `[2 x 3]` matrix representing two
+   *  original `SparseTensor` objects:
+   *  <p>
+   *      index = [ 0]
+   *              [10]
+   *              [20]
+   *      values = [1, 2, 3]
+   *      shape = [50]
+   *  <p>
+   *  and
+   *  <p>
+   *      index = [ 2]
+   *              [10]
+   *      values = [4, 5]
+   *      shape = [30]
+   *  <p>
+   *  then the final deserialized `SparseTensor` will be:
+   *  <p>
+   *      index = [0  0]
+   *              [0 10]
+   *              [0 20]
+   *              [1  2]
+   *              [1 10]
+   *      values = [1, 2, 3, 4, 5]
+   *      shape = [2 50]
+   *
+   * @param <U> data type for {@code sparseValues()} output
+   * @param serializedSparse The serialized `SparseTensor` objects. The last dimension
+   *  must have 3 columns.
+   * @param dtype The `dtype` of the serialized `SparseTensor` objects.
+   * @return a new instance of DeserializeSparse
+   */
+  public <U extends TType, T extends TType> DeserializeSparse<U> deserializeSparse(
+      Operand<T> serializedSparse, Class<U> dtype) {
+    return DeserializeSparse.create(scope, serializedSparse, dtype);
+  }
+
+  /**
+   * Applies a sparse gradient to a given accumulator.
+   *  <p>
+   *  Does not add if local_step is smaller than the accumulator's
+   *  global_step.
+   *
+   * @param handle The handle to a accumulator.
+   * @param localStep The local_step value at which the sparse gradient was computed.
+   * @param gradientIndices Indices of the sparse gradient to be accumulated. Must be a
+   *  vector.
+   * @param gradientValues Values are the non-zero slices of the gradient, and must have
+   *  the same first dimension as indices, i.e., the nnz represented by indices and
+   *  values must be consistent.
+   * @param gradientShape Shape of the sparse gradient to be accumulated.
+   * @param hasKnownShape Boolean indicating whether gradient_shape is unknown, in which
+   *  case the input is ignored during validation.
+   * @return a new instance of SparseAccumulatorApplyGradient
+   */
+  public <T extends TType> SparseAccumulatorApplyGradient sparseAccumulatorApplyGradient(
+      Operand<TString> handle, Operand<TInt64> localStep, Operand<TInt64> gradientIndices,
+      Operand<T> gradientValues, Operand<TInt64> gradientShape, Boolean hasKnownShape) {
+    return SparseAccumulatorApplyGradient.create(scope, handle, localStep, gradientIndices, gradientValues, gradientShape, hasKnownShape);
+  }
+
+  /**
+   * Extracts the average sparse gradient in a SparseConditionalAccumulator.
+   *  <p>
+   *  The op will blocks until sufficient (i.e., more than num_required)
+   *  gradients have been accumulated. If the accumulator has already
+   *  aggregated more than num_required gradients, it will return its
+   *  average of the accumulated gradients.  Also automatically increments
+   *  the recorded global_step in the accumulator by 1, and resets the
+   *  aggregate to 0.
+   *
+   * @param <T> data type for {@code values()} output
+   * @param handle The handle to a SparseConditionalAccumulator.
+   * @param numRequired Number of gradients required before we return an aggregate.
+   * @param dtype The data type of accumulated gradients. Needs to correspond to the type
+   *  of the accumulator.
+   * @return a new instance of SparseAccumulatorTakeGradient
+   */
+  public <T extends TType> SparseAccumulatorTakeGradient<T> sparseAccumulatorTakeGradient(
+      Operand<TString> handle, Operand<TInt32> numRequired, Class<T> dtype) {
+    return SparseAccumulatorTakeGradient.create(scope, handle, numRequired, dtype);
+  }
+
+  /**
+   * Adds two `SparseTensor` objects to produce another `SparseTensor`.
+   *  <p>
+   *  The input `SparseTensor` objects' indices are assumed ordered in standard
+   *  lexicographic order.  If this is not the case, before this step run
+   *  `SparseReorder` to restore index ordering.
+   *  <p>
+   *  By default, if two values sum to zero at some index, the output `SparseTensor`
+   *  would still include that particular location in its index, storing a zero in the
+   *  corresponding value slot.  To override this, callers can specify `thresh`,
+   *  indicating that if the sum has a magnitude strictly smaller than `thresh`, its
+   *  corresponding value and index would then not be included.  In particular,
+   *  `thresh == 0` (default) means everything is kept and actual thresholding happens
+   *  only for a positive value.
+   *  <p>
+   *  In the following shapes, `nnz` is the count after taking `thresh` into account.
+   *
+   * @param <T> data type for {@code sumValues()} output
+   * @param aIndices 2-D.  The `indices` of the first `SparseTensor`, size `[nnz, ndims]` Matrix.
+   * @param aValues 1-D.  The `values` of the first `SparseTensor`, size `[nnz]` Vector.
+   * @param aShape 1-D.  The `shape` of the first `SparseTensor`, size `[ndims]` Vector.
+   * @param bIndices 2-D.  The `indices` of the second `SparseTensor`, size `[nnz, ndims]` Matrix.
+   * @param bValues 1-D.  The `values` of the second `SparseTensor`, size `[nnz]` Vector.
+   * @param bShape 1-D.  The `shape` of the second `SparseTensor`, size `[ndims]` Vector.
+   * @param thresh 0-D.  The magnitude threshold that determines if an output value/index
+   *  pair takes space.
+   * @return a new instance of SparseAdd
+   */
+  public <T extends TType, U extends TNumber> SparseAdd<T> sparseAdd(Operand<TInt64> aIndices,
+      Operand<T> aValues, Operand<TInt64> aShape, Operand<TInt64> bIndices, Operand<T> bValues,
+      Operand<TInt64> bShape, Operand<U> thresh) {
+    return SparseAdd.create(scope, aIndices, aValues, aShape, bIndices, bValues, bShape, thresh);
+  }
+
+  /**
+   * The gradient operator for the SparseAdd op.
+   *  <p>
+   *  The SparseAdd op calculates A + B, where A, B, and the sum are all represented
+   *  as `SparseTensor` objects.  This op takes in the upstream gradient w.r.t.
+   *  non-empty values of the sum, and outputs the gradients w.r.t. the non-empty
+   *  values of A and B.
+   *
+   * @param <T> data type for {@code aValGrad()} output
+   * @param backpropValGrad 1-D with shape `[nnz(sum)]`.  The gradient with respect to
+   *  the non-empty values of the sum.
+   * @param aIndices 2-D.  The `indices` of the `SparseTensor` A, size `[nnz(A), ndims]`.
+   * @param bIndices 2-D.  The `indices` of the `SparseTensor` B, size `[nnz(B), ndims]`.
+   * @param sumIndices 2-D.  The `indices` of the sum `SparseTensor`, size
+   *  `[nnz(sum), ndims]`.
+   * @return a new instance of SparseAddGrad
+   */
+  public <T extends TType> SparseAddGrad<T> sparseAddGrad(Operand<T> backpropValGrad,
+      Operand<TInt64> aIndices, Operand<TInt64> bIndices, Operand<TInt64> sumIndices) {
+    return SparseAddGrad.create(scope, backpropValGrad, aIndices, bIndices, sumIndices);
+  }
+
+  /**
+   * Counts the number of occurrences of each value in an integer array.
+   *  <p>
+   *  Outputs a vector with length `size` and the same dtype as `weights`. If
+   *  `weights` are empty, then index `i` stores the number of times the value `i` is
+   *  counted in `arr`. If `weights` are non-empty, then index `i` stores the sum of
+   *  the value in `weights` at each index where the corresponding value in `arr` is
+   *  `i`.
+   *  <p>
+   *  Values in `arr` outside of the range [0, size) are ignored.
+   *
+   * @param <U> data type for {@code output()} output
+   * @param indices 2D int64 `Tensor`.
+   * @param values 1D int `Tensor`.
+   * @param denseShape 1D int64 `Tensor`.
+   * @param size non-negative int scalar `Tensor`.
+   * @param weights is an int32, int64, float32, or float64 `Tensor` with the same
+   *  shape as `input`, or a length-0 `Tensor`, in which case it acts as all weights
+   *  equal to 1.
+   * @param options carries optional attributes values
+   * @return a new instance of SparseBincount
+   */
+  public <U extends TNumber, T extends TNumber> SparseBincount<U> sparseBincount(
+      Operand<TInt64> indices, Operand<T> values, Operand<TInt64> denseShape, Operand<T> size,
+      Operand<U> weights, SparseBincount.Options... options) {
+    return SparseBincount.create(scope, indices, values, denseShape, size, weights, options);
+  }
+
+  /**
+   * Concatenates a list of `SparseTensor` along the specified dimension.
+   *  <p>
+   *  Concatenation is with respect to the dense versions of these sparse tensors.
+   *  It is assumed that each input is a `SparseTensor` whose elements are ordered
+   *  along increasing dimension number.
+   *  <p>
+   *  All inputs' shapes must match, except for the concat dimension.  The
+   *  `indices`, `values`, and `shapes` lists must have the same length.
+   *  <p>
+   *  The output shape is identical to the inputs', except along the concat
+   *  dimension, where it is the sum of the inputs' sizes along that dimension.
+   *  <p>
+   *  The output elements will be resorted to preserve the sort order along
+   *  increasing dimension number.
+   *  <p>
+   *  This op runs in `O(M log M)` time, where `M` is the total number of non-empty
+   *  values across all inputs. This is due to the need for an internal sort in
+   *  order to concatenate efficiently across an arbitrary dimension.
+   *  <p>
+   *  For example, if `concat_dim = 1` and the inputs are
+   *  <p>
+   *      sp_inputs[0]: shape = [2, 3]
+   *      [0, 2]: "a"
+   *      [1, 0]: "b"
+   *      [1, 1]: "c"
+   *  <p>
+   *      sp_inputs[1]: shape = [2, 4]
+   *      [0, 1]: "d"
+   *      [0, 2]: "e"
+   *  <p>
+   *  then the output will be
+   *  <p>
+   *      shape = [2, 7]
+   *      [0, 2]: "a"
+   *      [0, 4]: "d"
+   *      [0, 5]: "e"
+   *      [1, 0]: "b"
+   *      [1, 1]: "c"
+   *  <p>
+   *  Graphically this is equivalent to doing
+   *  <p>
+   *      [    a] concat [  d e  ] = [    a   d e  ]
+   *      [b c  ]        [       ]   [b c          ]
+   *
+   * @param <T> data type for {@code outputValues()} output
+   * @param indices 2-D.  Indices of each input `SparseTensor`.
+   * @param values 1-D.  Non-empty values of each `SparseTensor`.
+   * @param shapes 1-D.  Shapes of each `SparseTensor`.
+   * @param concatDim Dimension to concatenate along. Must be in range [-rank, rank),
+   *  where rank is the number of dimensions in each input `SparseTensor`.
+   * @return a new instance of SparseConcat
+   */
+  public <T extends TType> SparseConcat<T> sparseConcat(Iterable<Operand<TInt64>> indices,
+      Iterable<Operand<T>> values, Iterable<Operand<TInt64>> shapes, Long concatDim) {
+    return SparseConcat.create(scope, indices, values, shapes, concatDim);
+  }
+
+  /**
+   * A conditional accumulator for aggregating sparse gradients.
+   *  <p>
+   *  The accumulator accepts gradients marked with local_step greater or
+   *  equal to the most recent global_step known to the accumulator. The
+   *  average can be extracted from the accumulator, provided sufficient
+   *  gradients have been accumulated. Extracting the average automatically
+   *  resets the aggregate to 0, and increments the global_step recorded by
+   *  the accumulator.
+   *
+   * @param dtype The type of the value being accumulated.
+   * @param shape The shape of the values.
+   * @param options carries optional attributes values
+   * @return a new instance of SparseConditionalAccumulator
+   */
+  public <T extends TType> SparseConditionalAccumulator sparseConditionalAccumulator(Class<T> dtype,
+      Shape shape, SparseConditionalAccumulator.Options... options) {
+    return SparseConditionalAccumulator.create(scope, dtype, shape, options);
+  }
+
+  /**
+   * Generates sparse cross from a list of sparse and dense tensors.
+   *  <p>
+   *  The op takes two lists, one of 2D `SparseTensor` and one of 2D `Tensor`, each
+   *  representing features of one feature column. It outputs a 2D `SparseTensor` with
+   *  the batchwise crosses of these features.
+   *  <p>
+   *  For example, if the inputs are
+   *  <p>
+   *      inputs[0]: SparseTensor with shape = [2, 2]
+   *      [0, 0]: "a"
+   *      [1, 0]: "b"
+   *      [1, 1]: "c"
+   *  <p>
+   *      inputs[1]: SparseTensor with shape = [2, 1]
+   *      [0, 0]: "d"
+   *      [1, 0]: "e"
+   *  <p>
+   *      inputs[2]: Tensor [["f"], ["g"]]
+   *  <p>
+   *  then the output will be
+   *  <p>
+   *      shape = [2, 2]
+   *      [0, 0]: "a_X_d_X_f"
+   *      [1, 0]: "b_X_e_X_g"
+   *      [1, 1]: "c_X_e_X_g"
+   *  <p>
+   *  if hashed_output=true then the output will be
+   *  <p>
+   *      shape = [2, 2]
+   *      [0, 0]: FingerprintCat64(
+   *                  Fingerprint64("f"), FingerprintCat64(
+   *                      Fingerprint64("d"), Fingerprint64("a")))
+   *      [1, 0]: FingerprintCat64(
+   *                  Fingerprint64("g"), FingerprintCat64(
+   *                      Fingerprint64("e"), Fingerprint64("b")))
+   *      [1, 1]: FingerprintCat64(
+   *                  Fingerprint64("g"), FingerprintCat64(
+   *                      Fingerprint64("e"), Fingerprint64("c")))
+   *
+   * @param indices 2-D.  Indices of each input `SparseTensor`.
+   * @param values 1-D.   values of each `SparseTensor`.
+   * @param shapes 1-D.   Shapes of each `SparseTensor`.
+   * @param denseInputs 2-D.    Columns represented by dense `Tensor`.
+   * @param sep string used when joining a list of string inputs, can be used as separator later.
+   * @return a new instance of SparseCross
+   */
+  public SparseCross sparseCross(Iterable<Operand<TInt64>> indices, Iterable<Operand<?>> values,
+      Iterable<Operand<TInt64>> shapes, Iterable<Operand<?>> denseInputs, Operand<TString> sep) {
+    return SparseCross.create(scope, indices, values, shapes, denseInputs, sep);
+  }
+
+  /**
+   * Generates sparse cross from a list of sparse and dense tensors.
+   *  <p>
+   *  The op takes two lists, one of 2D `SparseTensor` and one of 2D `Tensor`, each
+   *  representing features of one feature column. It outputs a 2D `SparseTensor` with
+   *  the batchwise crosses of these features.
+   *  <p>
+   *  For example, if the inputs are
+   *  <p>
+   *      inputs[0]: SparseTensor with shape = [2, 2]
+   *      [0, 0]: "a"
+   *      [1, 0]: "b"
+   *      [1, 1]: "c"
+   *  <p>
+   *      inputs[1]: SparseTensor with shape = [2, 1]
+   *      [0, 0]: "d"
+   *      [1, 0]: "e"
+   *  <p>
+   *      inputs[2]: Tensor [["f"], ["g"]]
+   *  <p>
+   *  then the output will be
+   *  <p>
+   *      shape = [2, 2]
+   *      [0, 0]: "a_X_d_X_f"
+   *      [1, 0]: "b_X_e_X_g"
+   *      [1, 1]: "c_X_e_X_g"
+   *  <p>
+   *  if hashed_output=true then the output will be
+   *  <p>
+   *      shape = [2, 2]
+   *      [0, 0]: FingerprintCat64(
+   *                  Fingerprint64("f"), FingerprintCat64(
+   *                      Fingerprint64("d"), Fingerprint64("a")))
+   *      [1, 0]: FingerprintCat64(
+   *                  Fingerprint64("g"), FingerprintCat64(
+   *                      Fingerprint64("e"), Fingerprint64("b")))
+   *      [1, 1]: FingerprintCat64(
+   *                  Fingerprint64("g"), FingerprintCat64(
+   *                      Fingerprint64("e"), Fingerprint64("c")))
+   *
+   * @param indices 2-D.  Indices of each input `SparseTensor`.
+   * @param values 1-D.   values of each `SparseTensor`.
+   * @param shapes 1-D.   Shapes of each `SparseTensor`.
+   * @param denseInputs 2-D.    Columns represented by dense `Tensor`.
+   * @param numBuckets It is used if hashed_output is true.
+   *  output = hashed_value%num_buckets if num_buckets > 0 else hashed_value.
+   * @param strongHash boolean, if true, siphash with salt will be used instead of farmhash.
+   * @param salt Specify the salt that will be used by the siphash function.
+   * @return a new instance of SparseCrossHashed
+   */
+  public SparseCrossHashed sparseCrossHashed(Iterable<Operand<TInt64>> indices,
+      Iterable<Operand<?>> values, Iterable<Operand<TInt64>> shapes,
+      Iterable<Operand<?>> denseInputs, Operand<TInt64> numBuckets, Operand<TBool> strongHash,
+      Operand<TInt64> salt) {
+    return SparseCrossHashed.create(scope, indices, values, shapes, denseInputs, numBuckets, strongHash, salt);
+  }
+
+  /**
+   * Adds up a SparseTensor and a dense Tensor, using these special rules:
+   *  <p>
+   *  (1) Broadcasts the dense side to have the same shape as the sparse side, if
+   *      eligible;
+   *  (2) Then, only the dense values pointed to by the indices of the SparseTensor
+   *      participate in the cwise addition.
+   *  <p>
+   *  By these rules, the result is a logical SparseTensor with exactly the same
+   *  indices and shape, but possibly with different non-zero values.  The output of
+   *  this Op is the resultant non-zero values.
+   *
+   * @param <T> data type for {@code output()} output
+   * @param spIndices 2-D.  `N x R` matrix with the indices of non-empty values in a
+   *  SparseTensor, possibly not in canonical ordering.
+   * @param spValues 1-D.  `N` non-empty values corresponding to `sp_indices`.
+   * @param spShape 1-D.  Shape of the input SparseTensor.
+   * @param dense `R`-D.  The dense Tensor operand.
+   * @return a new instance of SparseDenseCwiseAdd
+   */
+  public <T extends TType> SparseDenseCwiseAdd<T> sparseDenseCwiseAdd(Operand<TInt64> spIndices,
+      Operand<T> spValues, Operand<TInt64> spShape, Operand<T> dense) {
+    return SparseDenseCwiseAdd.create(scope, spIndices, spValues, spShape, dense);
+  }
+
+  /**
+   * Component-wise divides a SparseTensor by a dense Tensor.
+   *  <p>
+   *  <i>Limitation</i>: this Op only broadcasts the dense side to the sparse side, but not
+   *  the other direction.
+   *
+   * @param <T> data type for {@code output()} output
+   * @param spIndices 2-D.  `N x R` matrix with the indices of non-empty values in a
+   *  SparseTensor, possibly not in canonical ordering.
+   * @param spValues 1-D.  `N` non-empty values corresponding to `sp_indices`.
+   * @param spShape 1-D.  Shape of the input SparseTensor.
+   * @param dense `R`-D.  The dense Tensor operand.
+   * @return a new instance of SparseDenseCwiseDiv
+   */
+  public <T extends TType> SparseDenseCwiseDiv<T> sparseDenseCwiseDiv(Operand<TInt64> spIndices,
+      Operand<T> spValues, Operand<TInt64> spShape, Operand<T> dense) {
+    return SparseDenseCwiseDiv.create(scope, spIndices, spValues, spShape, dense);
+  }
+
+  /**
+   * Component-wise multiplies a SparseTensor by a dense Tensor.
+   *  <p>
+   *  The output locations corresponding to the implicitly zero elements in the sparse
+   *  tensor will be zero (i.e., will not take up storage space), regardless of the
+   *  contents of the dense tensor (even if it's +/-INF and that INF<i>0 == NaN).
+   *  <p>
+   *  </i>Limitation*: this Op only broadcasts the dense side to the sparse side, but not
+   *  the other direction.
+   *
+   * @param <T> data type for {@code output()} output
+   * @param spIndices 2-D.  `N x R` matrix with the indices of non-empty values in a
+   *  SparseTensor, possibly not in canonical ordering.
+   * @param spValues 1-D.  `N` non-empty values corresponding to `sp_indices`.
+   * @param spShape 1-D.  Shape of the input SparseTensor.
+   * @param dense `R`-D.  The dense Tensor operand.
+   * @return a new instance of SparseDenseCwiseMul
+   */
+  public <T extends TType> SparseDenseCwiseMul<T> sparseDenseCwiseMul(Operand<TInt64> spIndices,
+      Operand<T> spValues, Operand<TInt64> spShape, Operand<T> dense) {
+    return SparseDenseCwiseMul.create(scope, spIndices, spValues, spShape, dense);
+  }
+
+  /**
+   * Fills empty rows in the input 2-D `SparseTensor` with a default value.
+   *  <p>
+   *  The input `SparseTensor` is represented via the tuple of inputs
+   *  (`indices`, `values`, `dense_shape`).  The output `SparseTensor` has the
+   *  same `dense_shape` but with indices `output_indices` and values
+   *  `output_values`.
+   *  <p>
+   *  This op inserts a single entry for every row that doesn't have any values.
+   *  The index is created as `[row, 0, ..., 0]` and the inserted value
+   *  is `default_value`.
+   *  <p>
+   *  For example, suppose `sp_input` has shape `[5, 6]` and non-empty values:
+   *  <p>
+   *      [0, 1]: a
+   *      [0, 3]: b
+   *      [2, 0]: c
+   *      [3, 1]: d
+   *  <p>
+   *  Rows 1 and 4 are empty, so the output will be of shape `[5, 6]` with values:
+   *  <p>
+   *      [0, 1]: a
+   *      [0, 3]: b
+   *      [1, 0]: default_value
+   *      [2, 0]: c
+   *      [3, 1]: d
+   *      [4, 0]: default_value
+   *  <p>
+   *  The output `SparseTensor` will be in row-major order and will have the
+   *  same shape as the input.
+   *  <p>
+   *  This op also returns an indicator vector shaped `[dense_shape[0]]` such that
+   *  <p>
+   *      empty_row_indicator[i] = True iff row i was an empty row.
+   *  <p>
+   *  And a reverse index map vector shaped `[indices.shape[0]]` that is used during
+   *  backpropagation,
+   *  <p>
+   *      reverse_index_map[j] = out_j s.t. indices[j, :] == output_indices[out_j, :]
+   *
+   * @param <T> data type for {@code outputValues()} output
+   * @param indices 2-D. the indices of the sparse tensor.
+   * @param values 1-D. the values of the sparse tensor.
+   * @param denseShape 1-D. the shape of the sparse tensor.
+   * @param defaultValue 0-D. default value to insert into location `[row, 0, ..., 0]`
+   *    for rows missing from the input sparse tensor.
+   *  output indices: 2-D. the indices of the filled sparse tensor.
+   * @return a new instance of SparseFillEmptyRows
+   */
+  public <T extends TType> SparseFillEmptyRows<T> sparseFillEmptyRows(Operand<TInt64> indices,
+      Operand<T> values, Operand<TInt64> denseShape, Operand<T> defaultValue) {
+    return SparseFillEmptyRows.create(scope, indices, values, denseShape, defaultValue);
+  }
+
+  /**
+   * The gradient of SparseFillEmptyRows.
+   *  <p>
+   *  Takes vectors reverse_index_map, shaped `[N]`, and grad_values,
+   *  shaped `[N_full]`, where `N_full >= N` and copies data into either
+   *  `d_values` or `d_default_value`.  Here `d_values` is shaped `[N]` and
+   *  `d_default_value` is a scalar.
+   *  <p>
+   *    d_values[j] = grad_values[reverse_index_map[j]]
+   *    d_default_value = sum_{k : 0 .. N_full - 1} (
+   *       grad_values[k] * 1{k not in reverse_index_map})
+   *
+   * @param <T> data type for {@code dValues()} output
+   * @param reverseIndexMap 1-D.  The reverse index map from SparseFillEmptyRows.
+   * @param gradValues 1-D.  The gradients from backprop.
+   * @return a new instance of SparseFillEmptyRowsGrad
+   */
+  public <T extends TType> SparseFillEmptyRowsGrad<T> sparseFillEmptyRowsGrad(
+      Operand<TInt64> reverseIndexMap, Operand<T> gradValues) {
+    return SparseFillEmptyRowsGrad.create(scope, reverseIndexMap, gradValues);
+  }
+
+  /**
+   * Multiply matrix "a" by matrix "b".
+   *  <p>
+   *  The inputs must be two-dimensional matrices and the inner dimension of "a" must
+   *  match the outer dimension of "b". Both "a" and "b" must be `Tensor`s not
+   *  `SparseTensor`s.  This op is optimized for the case where at least one of "a" or
+   *  "b" is sparse, in the sense that they have a large proportion of zero values.
+   *  The breakeven for using this versus a dense matrix multiply on one platform was
+   *  30% zero values in the sparse matrix.
+   *  <p>
+   *  The gradient computation of this operation will only take advantage of sparsity
+   *  in the input gradient when that gradient comes from a Relu.
+   *
+   * @param a
+   * @param b
+   * @param options carries optional attributes values
+   * @return a new instance of SparseMatMul
+   */
+  public <T extends TNumber, U extends TNumber> SparseMatMul sparseMatMul(Operand<T> a,
+      Operand<U> b, SparseMatMul.Options... options) {
+    return SparseMatMul.create(scope, a, b, options);
+  }
+
+  /**
+   * Computes the max of elements across dimensions of a SparseTensor.
+   *  <p>
+   *  This Op takes a SparseTensor and is the sparse counterpart to
+   *  `tf.reduce_max()`.  In particular, this Op also returns a dense `Tensor`
+   *  instead of a sparse one.
+   *  <p>
+   *  Reduces `sp_input` along the dimensions given in `reduction_axes`.  Unless
+   *  `keep_dims` is true, the rank of the tensor is reduced by 1 for each entry in
+   *  `reduction_axes`. If `keep_dims` is true, the reduced dimensions are retained
+   *  with length 1.
+   *  <p>
+   *  If `reduction_axes` has no entries, all dimensions are reduced, and a tensor
+   *  with a single element is returned.  Additionally, the axes can be negative,
+   *  which are interpreted according to the indexing rules in Python.
+   *
+   * @param <T> data type for {@code output()} output
+   * @param inputIndices 2-D.  `N x R` matrix with the indices of non-empty values in a
+   *  SparseTensor, possibly not in canonical ordering.
+   * @param inputValues 1-D.  `N` non-empty values corresponding to `input_indices`.
+   * @param inputShape 1-D.  Shape of the input SparseTensor.
+   * @param reductionAxes 1-D.  Length-`K` vector containing the reduction axes.
+   * @param options carries optional attributes values
+   * @return a new instance of SparseReduceMax
+   */
+  public <T extends TNumber> SparseReduceMax<T> sparseReduceMax(Operand<TInt64> inputIndices,
+      Operand<T> inputValues, Operand<TInt64> inputShape, Operand<TInt32> reductionAxes,
+      SparseReduceMax.Options... options) {
+    return SparseReduceMax.create(scope, inputIndices, inputValues, inputShape, reductionAxes, options);
+  }
+
+  /**
+   * Computes the max of elements across dimensions of a SparseTensor.
+   *  <p>
+   *  This Op takes a SparseTensor and is the sparse counterpart to
+   *  `tf.reduce_max()`.  In contrast to SparseReduceMax, this Op returns a
+   *  SparseTensor.
+   *  <p>
+   *  Reduces `sp_input` along the dimensions given in `reduction_axes`.  Unless
+   *  `keep_dims` is true, the rank of the tensor is reduced by 1 for each entry in
+   *  `reduction_axes`. If `keep_dims` is true, the reduced dimensions are retained
+   *  with length 1.
+   *  <p>
+   *  If `reduction_axes` has no entries, all dimensions are reduced, and a tensor
+   *  with a single element is returned.  Additionally, the axes can be negative,
+   *  which are interpreted according to the indexing rules in Python.
+   *
+   * @param <T> data type for {@code outputValues()} output
+   * @param inputIndices 2-D.  `N x R` matrix with the indices of non-empty values in a
+   *  SparseTensor, possibly not in canonical ordering.
+   * @param inputValues 1-D.  `N` non-empty values corresponding to `input_indices`.
+   * @param inputShape 1-D.  Shape of the input SparseTensor.
+   * @param reductionAxes 1-D.  Length-`K` vector containing the reduction axes.
+   * @param options carries optional attributes values
+   * @return a new instance of SparseReduceMaxSparse
+   */
+  public <T extends TNumber> SparseReduceMaxSparse<T> sparseReduceMaxSparse(
+      Operand<TInt64> inputIndices, Operand<T> inputValues, Operand<TInt64> inputShape,
+      Operand<TInt32> reductionAxes, SparseReduceMaxSparse.Options... options) {
+    return SparseReduceMaxSparse.create(scope, inputIndices, inputValues, inputShape, reductionAxes, options);
+  }
+
+  /**
+   * Computes the sum of elements across dimensions of a SparseTensor.
+   *  <p>
+   *  This Op takes a SparseTensor and is the sparse counterpart to
+   *  `tf.reduce_sum()`.  In particular, this Op also returns a dense `Tensor`
+   *  instead of a sparse one.
+   *  <p>
+   *  Reduces `sp_input` along the dimensions given in `reduction_axes`.  Unless
+   *  `keep_dims` is true, the rank of the tensor is reduced by 1 for each entry in
+   *  `reduction_axes`. If `keep_dims` is true, the reduced dimensions are retained
+   *  with length 1.
+   *  <p>
+   *  If `reduction_axes` has no entries, all dimensions are reduced, and a tensor
+   *  with a single element is returned.  Additionally, the axes can be negative,
+   *  which are interpreted according to the indexing rules in Python.
+   *
+   * @param <T> data type for {@code output()} output
+   * @param inputIndices 2-D.  `N x R` matrix with the indices of non-empty values in a
+   *  SparseTensor, possibly not in canonical ordering.
+   * @param inputValues 1-D.  `N` non-empty values corresponding to `input_indices`.
+   * @param inputShape 1-D.  Shape of the input SparseTensor.
+   * @param reductionAxes 1-D.  Length-`K` vector containing the reduction axes.
+   * @param options carries optional attributes values
+   * @return a new instance of SparseReduceSum
+   */
+  public <T extends TType> SparseReduceSum<T> sparseReduceSum(Operand<TInt64> inputIndices,
+      Operand<T> inputValues, Operand<TInt64> inputShape, Operand<TInt32> reductionAxes,
+      SparseReduceSum.Options... options) {
+    return SparseReduceSum.create(scope, inputIndices, inputValues, inputShape, reductionAxes, options);
+  }
+
+  /**
+   * Computes the sum of elements across dimensions of a SparseTensor.
+   *  <p>
+   *  This Op takes a SparseTensor and is the sparse counterpart to
+   *  `tf.reduce_sum()`.  In contrast to SparseReduceSum, this Op returns a
+   *  SparseTensor.
+   *  <p>
+   *  Reduces `sp_input` along the dimensions given in `reduction_axes`.  Unless
+   *  `keep_dims` is true, the rank of the tensor is reduced by 1 for each entry in
+   *  `reduction_axes`. If `keep_dims` is true, the reduced dimensions are retained
+   *  with length 1.
+   *  <p>
+   *  If `reduction_axes` has no entries, all dimensions are reduced, and a tensor
+   *  with a single element is returned.  Additionally, the axes can be negative,
+   *  which are interpreted according to the indexing rules in Python.
+   *
+   * @param <T> data type for {@code outputValues()} output
+   * @param inputIndices 2-D.  `N x R` matrix with the indices of non-empty values in a
+   *  SparseTensor, possibly not in canonical ordering.
+   * @param inputValues 1-D.  `N` non-empty values corresponding to `input_indices`.
+   * @param inputShape 1-D.  Shape of the input SparseTensor.
+   * @param reductionAxes 1-D.  Length-`K` vector containing the reduction axes.
+   * @param options carries optional attributes values
+   * @return a new instance of SparseReduceSumSparse
+   */
+  public <T extends TType> SparseReduceSumSparse<T> sparseReduceSumSparse(
+      Operand<TInt64> inputIndices, Operand<T> inputValues, Operand<TInt64> inputShape,
+      Operand<TInt32> reductionAxes, SparseReduceSumSparse.Options... options) {
+    return SparseReduceSumSparse.create(scope, inputIndices, inputValues, inputShape, reductionAxes, options);
+  }
+
+  /**
+   * Reorders a SparseTensor into the canonical, row-major ordering.
+   *  <p>
+   *  Note that by convention, all sparse ops preserve the canonical ordering along
+   *  increasing dimension number. The only time ordering can be violated is during
+   *  manual manipulation of the indices and values vectors to add entries.
+   *  <p>
+   *  Reordering does not affect the shape of the SparseTensor.
+   *  <p>
+   *  If the tensor has rank `R` and `N` non-empty values, `input_indices` has
+   *  shape `[N, R]`, input_values has length `N`, and input_shape has length `R`.
+   *
+   * @param <T> data type for {@code outputValues()} output
+   * @param inputIndices 2-D.  `N x R` matrix with the indices of non-empty values in a
+   *  SparseTensor, possibly not in canonical ordering.
+   * @param inputValues 1-D.  `N` non-empty values corresponding to `input_indices`.
+   * @param inputShape 1-D.  Shape of the input SparseTensor.
+   * @return a new instance of SparseReorder
+   */
+  public <T extends TType> SparseReorder<T> sparseReorder(Operand<TInt64> inputIndices,
+      Operand<T> inputValues, Operand<TInt64> inputShape) {
+    return SparseReorder.create(scope, inputIndices, inputValues, inputShape);
+  }
+
+  /**
+   * Reshapes a SparseTensor to represent values in a new dense shape.
+   *  <p>
+   *  This operation has the same semantics as reshape on the represented dense
+   *  tensor.  The `input_indices` are recomputed based on the requested `new_shape`.
+   *  <p>
+   *  If one component of `new_shape` is the special value -1, the size of that
+   *  dimension is computed so that the total dense size remains constant.  At
+   *  most one component of `new_shape` can be -1.  The number of dense elements
+   *  implied by `new_shape` must be the same as the number of dense elements
+   *  originally implied by `input_shape`.
+   *  <p>
+   *  Reshaping does not affect the order of values in the SparseTensor.
+   *  <p>
+   *  If the input tensor has rank `R_in` and `N` non-empty values, and `new_shape`
+   *  has length `R_out`, then `input_indices` has shape `[N, R_in]`,
+   *  `input_shape` has length `R_in`, `output_indices` has shape `[N, R_out]`, and
+   *  `output_shape` has length `R_out`.
+   *
+   * @param inputIndices 2-D.  `N x R_in` matrix with the indices of non-empty values in a
+   *  SparseTensor.
+   * @param inputShape 1-D.  `R_in` vector with the input SparseTensor's dense shape.
+   * @param newShape 1-D.  `R_out` vector with the requested new dense shape.
+   * @return a new instance of SparseReshape
+   */
+  public SparseReshape sparseReshape(Operand<TInt64> inputIndices, Operand<TInt64> inputShape,
+      Operand<TInt64> newShape) {
+    return SparseReshape.create(scope, inputIndices, inputShape, newShape);
+  }
+
+  /**
+   * Computes the mean along sparse segments of a tensor.
+   *  <p>
+   *  See `tf.sparse.segment_sum` for usage examples.
+   *  <p>
+   *  Like `SegmentMean`, but `segment_ids` can have rank less than `data`'s first
+   *  dimension, selecting a subset of dimension 0, specified by `indices`.
+   *
+   * @param <T> data type for {@code output()} output
+   * @param data
+   * @param indices A 1-D tensor. Has same rank as `segment_ids`.
+   * @param segmentIds A 1-D tensor. Values should be sorted and can be repeated.
+   * @return a new instance of SparseSegmentMean
+   */
+  public <T extends TNumber, U extends TNumber, V extends TNumber> SparseSegmentMean<T> sparseSegmentMean(
+      Operand<T> data, Operand<U> indices, Operand<V> segmentIds) {
+    return SparseSegmentMean.create(scope, data, indices, segmentIds);
+  }
+
+  /**
+   * Computes gradients for SparseSegmentMean.
+   *  <p>
+   *  Returns tensor "output" with same shape as grad, except for dimension 0 whose
+   *  value is output_dim0.
+   *
+   * @param <T> data type for {@code output()} output
+   * @param grad gradient propagated to the SparseSegmentMean op.
+   * @param indices indices passed to the corresponding SparseSegmentMean op.
+   * @param segmentIds segment_ids passed to the corresponding SparseSegmentMean op.
+   * @param outputDim0 dimension 0 of "data" passed to SparseSegmentMean op.
+   * @return a new instance of SparseSegmentMeanGrad
+   */
+  public <T extends TNumber, U extends TNumber, V extends TNumber> SparseSegmentMeanGrad<T> sparseSegmentMeanGrad(
+      Operand<T> grad, Operand<U> indices, Operand<V> segmentIds, Operand<TInt32> outputDim0) {
+    return SparseSegmentMeanGrad.create(scope, grad, indices, segmentIds, outputDim0);
+  }
+
+  /**
+   * Computes the mean along sparse segments of a tensor.
+   *  <p>
+   *  Like `SparseSegmentMean`, but allows missing ids in `segment_ids`. If an id is
+   *  missing, the `output` tensor at that position will be zeroed.
+   *  <p>
+   *  Read
+   *  [the section on segmentation](https://tensorflow.org/api_docs/python/tf/math#Segmentation)
+   *  for an explanation of segments.
+   *
+   * @param <T> data type for {@code output()} output
+   * @param data
+   * @param indices A 1-D tensor. Has same rank as `segment_ids`.
+   * @param segmentIds A 1-D tensor. Values should be sorted and can be repeated.
+   * @param numSegments Should equal the number of distinct segment IDs.
+   * @return a new instance of SparseSegmentMeanWithNumSegments
+   */
+  public <T extends TNumber, U extends TNumber, V extends TNumber, W extends TNumber> SparseSegmentMeanWithNumSegments<T> sparseSegmentMeanWithNumSegments(
+      Operand<T> data, Operand<U> indices, Operand<V> segmentIds, Operand<W> numSegments) {
+    return SparseSegmentMeanWithNumSegments.create(scope, data, indices, segmentIds, numSegments);
+  }
+
+  /**
+   * Computes the sum along sparse segments of a tensor divided by the sqrt of N.
+   *  <p>
+   *  N is the size of the segment being reduced.
+   *  <p>
+   *  See `tf.sparse.segment_sum` for usage examples.
+   *
+   * @param <T> data type for {@code output()} output
+   * @param data
+   * @param indices A 1-D tensor. Has same rank as `segment_ids`.
+   * @param segmentIds A 1-D tensor. Values should be sorted and can be repeated.
+   * @return a new instance of SparseSegmentSqrtN
+   */
+  public <T extends TNumber, U extends TNumber, V extends TNumber> SparseSegmentSqrtN<T> sparseSegmentSqrtN(
+      Operand<T> data, Operand<U> indices, Operand<V> segmentIds) {
+    return SparseSegmentSqrtN.create(scope, data, indices, segmentIds);
+  }
+
+  /**
+   * Computes gradients for SparseSegmentSqrtN.
+   *  <p>
+   *  Returns tensor "output" with same shape as grad, except for dimension 0 whose
+   *  value is output_dim0.
+   *
+   * @param <T> data type for {@code output()} output
+   * @param grad gradient propagated to the SparseSegmentSqrtN op.
+   * @param indices indices passed to the corresponding SparseSegmentSqrtN op.
+   * @param segmentIds segment_ids passed to the corresponding SparseSegmentSqrtN op.
+   * @param outputDim0 dimension 0 of "data" passed to SparseSegmentSqrtN op.
+   * @return a new instance of SparseSegmentSqrtNGrad
+   */
+  public <T extends TNumber, U extends TNumber, V extends TNumber> SparseSegmentSqrtNGrad<T> sparseSegmentSqrtNGrad(
+      Operand<T> grad, Operand<U> indices, Operand<V> segmentIds, Operand<TInt32> outputDim0) {
+    return SparseSegmentSqrtNGrad.create(scope, grad, indices, segmentIds, outputDim0);
+  }
+
+  /**
+   * Computes the sum along sparse segments of a tensor divided by the sqrt of N.
+   *  <p>
+   *  N is the size of the segment being reduced.
+   *  <p>
+   *  Like `SparseSegmentSqrtN`, but allows missing ids in `segment_ids`. If an id is
+   *  missing, the `output` tensor at that position will be zeroed.
+   *  <p>
+   *  Read
+   *  [the section on segmentation](https://tensorflow.org/api_docs/python/tf/math#Segmentation)
+   *  for an explanation of segments.
+   *
+   * @param <T> data type for {@code output()} output
+   * @param data
+   * @param indices A 1-D tensor. Has same rank as `segment_ids`.
+   * @param segmentIds A 1-D tensor. Values should be sorted and can be repeated.
+   * @param numSegments Should equal the number of distinct segment IDs.
+   * @return a new instance of SparseSegmentSqrtNWithNumSegments
+   */
+  public <T extends TNumber, U extends TNumber, V extends TNumber, W extends TNumber> SparseSegmentSqrtNWithNumSegments<T> sparseSegmentSqrtNWithNumSegments(
+      Operand<T> data, Operand<U> indices, Operand<V> segmentIds, Operand<W> numSegments) {
+    return SparseSegmentSqrtNWithNumSegments.create(scope, data, indices, segmentIds, numSegments);
+  }
+
+  /**
+   * Computes the sum along sparse segments of a tensor.
+   *  <p>
+   *  Read
+   *  [the section on segmentation](https://tensorflow.org/api_docs/python/tf/math#Segmentation)
+   *  for an explanation of segments.
+   *  <p>
+   *  Like `SegmentSum`, but `segment_ids` can have rank less than `data`'s first
+   *  dimension, selecting a subset of dimension 0, specified by `indices`.
+   *  <p>
+   *  For example:
+   *  <pre>{@code
+   *  c = tf.constant([[1,2,3,4], [-1,-2,-3,-4], [5,6,7,8]])
+   *
+   *  # Select two rows, one segment.
+   *  tf.sparse_segment_sum(c, tf.constant([0, 1]), tf.constant([0, 0]))
+   *  # => [[0 0 0 0]]
+   *
+   *  # Select two rows, two segment.
+   *  tf.sparse_segment_sum(c, tf.constant([0, 1]), tf.constant([0, 1]))
+   *  # => [[ 1  2  3  4]
+   *  #     [-1 -2 -3 -4]]
+   *
+   *  # Select all rows, two segments.
+   *  tf.sparse_segment_sum(c, tf.constant([0, 1, 2]), tf.constant([0, 0, 1]))
+   *  # => [[0 0 0 0]
+   *  #     [5 6 7 8]]
+   *
+   *  # Which is equivalent to:
+   *  tf.segment_sum(c, tf.constant([0, 0, 1]))
+   *  }</pre>
+   *
+   * @param <T> data type for {@code output()} output
+   * @param data
+   * @param indices A 1-D tensor. Has same rank as `segment_ids`.
+   * @param segmentIds A 1-D tensor. Values should be sorted and can be repeated.
+   * @return a new instance of SparseSegmentSum
+   */
+  public <T extends TNumber, U extends TNumber, V extends TNumber> SparseSegmentSum<T> sparseSegmentSum(
+      Operand<T> data, Operand<U> indices, Operand<V> segmentIds) {
+    return SparseSegmentSum.create(scope, data, indices, segmentIds);
+  }
+
+  /**
+   * Computes the sum along sparse segments of a tensor.
+   *  <p>
+   *  Like `SparseSegmentSum`, but allows missing ids in `segment_ids`. If an id is
+   *  missing, the `output` tensor at that position will be zeroed.
+   *  <p>
+   *  Read
+   *  [the section on segmentation](https://tensorflow.org/api_docs/python/tf/sparse#Segmentation)
+   *  for an explanation of segments.
+   *  <p>
+   *  For example:
+   *  <pre>{@code
+   *  c = tf.constant([[1,2,3,4], [-1,-2,-3,-4], [5,6,7,8]])
+   *
+   *  tf.sparse_segment_sum_with_num_segments(
+   *      c, tf.constant([0, 1]), tf.constant([0, 0]), num_segments=3)
+   *  # => [[0 0 0 0]
+   *  #     [0 0 0 0]
+   *  #     [0 0 0 0]]
+   *
+   *  tf.sparse_segment_sum_with_num_segments(c,
+   *                                          tf.constant([0, 1]),
+   *                                          tf.constant([0, 2],
+   *                                          num_segments=4))
+   *  # => [[ 1  2  3  4]
+   *  #     [ 0  0  0  0]
+   *  #     [-1 -2 -3 -4]
+   *  #     [ 0  0  0  0]]
+   *  }</pre>
+   *
+   * @param <T> data type for {@code output()} output
+   * @param data
+   * @param indices A 1-D tensor. Has same rank as `segment_ids`.
+   * @param segmentIds A 1-D tensor. Values should be sorted and can be repeated.
+   * @param numSegments Should equal the number of distinct segment IDs.
+   * @return a new instance of SparseSegmentSumWithNumSegments
+   */
+  public <T extends TNumber, U extends TNumber, V extends TNumber, W extends TNumber> SparseSegmentSumWithNumSegments<T> sparseSegmentSumWithNumSegments(
+      Operand<T> data, Operand<U> indices, Operand<V> segmentIds, Operand<W> numSegments) {
+    return SparseSegmentSumWithNumSegments.create(scope, data, indices, segmentIds, numSegments);
+  }
+
+  /**
+   * Slice a `SparseTensor` based on the `start` and `size`.
+   *  <p>
+   *  For example, if the input is
+   *  <p>
+   *      input_tensor = shape = [2, 7]
+   *      [    a   d e  ]
+   *      [b c          ]
+   *  <p>
+   *  Graphically the output tensors are:
+   *  <p>
+   *      sparse_slice([0, 0], [2, 4]) = shape = [2, 4]
+   *      [    a  ]
+   *      [b c    ]
+   *  <p>
+   *      sparse_slice([0, 4], [2, 3]) = shape = [2, 3]
+   *      [ d e  ]
+   *      [      ]
+   *
+   * @param <T> data type for {@code outputValues()} output
+   * @param indices 2-D tensor represents the indices of the sparse tensor.
+   * @param values 1-D tensor represents the values of the sparse tensor.
+   * @param shape 1-D. tensor represents the shape of the sparse tensor.
+   * @param start 1-D. tensor represents the start of the slice.
+   * @param size 1-D. tensor represents the size of the slice.
+   *  output indices: A list of 1-D tensors represents the indices of the output
+   *  sparse tensors.
+   * @return a new instance of SparseSlice
+   */
+  public <T extends TType> SparseSlice<T> sparseSlice(Operand<TInt64> indices, Operand<T> values,
+      Operand<TInt64> shape, Operand<TInt64> start, Operand<TInt64> size) {
+    return SparseSlice.create(scope, indices, values, shape, start, size);
+  }
+
+  /**
+   * The gradient operator for the SparseSlice op.
+   *  <p>
+   *  This op takes in the upstream gradient w.r.t. non-empty values of
+   *  the sliced `SparseTensor`, and outputs the gradients w.r.t.
+   *  the non-empty values of input `SparseTensor`.
+   *
+   * @param <T> data type for {@code valGrad()} output
+   * @param backpropValGrad 1-D. The gradient with respect to
+   *  the non-empty values of the sliced `SparseTensor`.
+   * @param inputIndices 2-D.  The `indices` of the input `SparseTensor`.
+   * @param inputStart 1-D. tensor represents the start of the slice.
+   * @param outputIndices 2-D.  The `indices` of the sliced `SparseTensor`.
+   * @return a new instance of SparseSliceGrad
+   */
+  public <T extends TType> SparseSliceGrad<T> sparseSliceGrad(Operand<T> backpropValGrad,
+      Operand<TInt64> inputIndices, Operand<TInt64> inputStart, Operand<TInt64> outputIndices) {
+    return SparseSliceGrad.create(scope, backpropValGrad, inputIndices, inputStart, outputIndices);
+  }
+
+  /**
+   * Applies softmax to a batched N-D `SparseTensor`.
+   *  <p>
+   *  The inputs represent an N-D SparseTensor  with logical shape `[..., B, C]`
+   *  (where `N >= 2`), and with indices sorted in the canonical lexicographic order.
+   *  <p>
+   *  This op is equivalent to applying the normal `tf.nn.softmax()` to each innermost
+   *  logical submatrix with shape `[B, C]`, but with the catch that <i>the implicitly
+   *  zero elements do not participate</i>.  Specifically, the algorithm is equivalent
+   *  to the following:
+   *  <p>
+   *    (1) Applies `tf.nn.softmax()` to a densified view of each innermost submatrix
+   *        with shape `[B, C]`, along the size-C dimension;
+   *    (2) Masks out the original implicitly-zero locations;
+   *    (3) Renormalizes the remaining elements.
+   *  <p>
+   *  Hence, the `SparseTensor` result has exactly the same non-zero indices and
+   *  shape.
+   *
+   * @param <T> data type for {@code output()} output
+   * @param spIndices 2-D.  `NNZ x R` matrix with the indices of non-empty values in a
+   *  SparseTensor, in canonical ordering.
+   * @param spValues 1-D.  `NNZ` non-empty values corresponding to `sp_indices`.
+   * @param spShape 1-D.  Shape of the input SparseTensor.
+   * @return a new instance of SparseSoftmax
+   */
+  public <T extends TNumber> SparseSoftmax<T> sparseSoftmax(Operand<TInt64> spIndices,
+      Operand<T> spValues, Operand<TInt64> spShape) {
+    return SparseSoftmax.create(scope, spIndices, spValues, spShape);
+  }
+
+  /**
+   * Returns the element-wise max of two SparseTensors.
+   *  <p>
+   *  Assumes the two SparseTensors have the same shape, i.e., no broadcasting.
+   *
+   * @param <T> data type for {@code outputValues()} output
+   * @param aIndices 2-D.  `N x R` matrix with the indices of non-empty values in a
+   *  SparseTensor, in the canonical lexicographic ordering.
+   * @param aValues 1-D.  `N` non-empty values corresponding to `a_indices`.
+   * @param aShape 1-D.  Shape of the input SparseTensor.
+   * @param bIndices counterpart to `a_indices` for the other operand.
+   * @param bValues counterpart to `a_values` for the other operand; must be of the same dtype.
+   * @param bShape counterpart to `a_shape` for the other operand; the two shapes must be equal.
+   * @return a new instance of SparseSparseMaximum
+   */
+  public <T extends TNumber> SparseSparseMaximum<T> sparseSparseMaximum(Operand<TInt64> aIndices,
+      Operand<T> aValues, Operand<TInt64> aShape, Operand<TInt64> bIndices, Operand<T> bValues,
+      Operand<TInt64> bShape) {
+    return SparseSparseMaximum.create(scope, aIndices, aValues, aShape, bIndices, bValues, bShape);
+  }
+
+  /**
+   * Returns the element-wise min of two SparseTensors.
+   *  <p>
+   *  Assumes the two SparseTensors have the same shape, i.e., no broadcasting.
+   *
+   * @param <T> data type for {@code outputValues()} output
+   * @param aIndices 2-D.  `N x R` matrix with the indices of non-empty values in a
+   *  SparseTensor, in the canonical lexicographic ordering.
+   * @param aValues 1-D.  `N` non-empty values corresponding to `a_indices`.
+   * @param aShape 1-D.  Shape of the input SparseTensor.
+   * @param bIndices counterpart to `a_indices` for the other operand.
+   * @param bValues counterpart to `a_values` for the other operand; must be of the same dtype.
+   * @param bShape counterpart to `a_shape` for the other operand; the two shapes must be equal.
+   * @return a new instance of SparseSparseMinimum
+   */
+  public <T extends TType> SparseSparseMinimum<T> sparseSparseMinimum(Operand<TInt64> aIndices,
+      Operand<T> aValues, Operand<TInt64> aShape, Operand<TInt64> bIndices, Operand<T> bValues,
+      Operand<TInt64> bShape) {
+    return SparseSparseMinimum.create(scope, aIndices, aValues, aShape, bIndices, bValues, bShape);
+  }
+
+  /**
+   * Split a `SparseTensor` into `num_split` tensors along one dimension.
+   *  <p>
+   *  If the `shape[split_dim]` is not an integer multiple of `num_split`. Slices
+   *  `[0 : shape[split_dim] % num_split]` gets one extra dimension.
+   *  For example, if `split_dim = 1` and `num_split = 2` and the input is
+   *  <p>
+   *      input_tensor = shape = [2, 7]
+   *      [    a   d e  ]
+   *      [b c          ]
+   *  <p>
+   *  Graphically the output tensors are:
+   *  <p>
+   *      output_tensor[0] = shape = [2, 4]
+   *      [    a  ]
+   *      [b c    ]
+   *  <p>
+   *      output_tensor[1] = shape = [2, 3]
+   *      [ d e  ]
+   *      [      ]
+   *
+   * @param <T> data type for {@code outputValues()} output
+   * @param splitDim 0-D.  The dimension along which to split.  Must be in the range
+   *  `[0, rank(shape))`.
+   * @param indices 2-D tensor represents the indices of the sparse tensor.
+   * @param values 1-D tensor represents the values of the sparse tensor.
+   * @param shape 1-D. tensor represents the shape of the sparse tensor.
+   *  output indices: A list of 1-D tensors represents the indices of the output
+   *  sparse tensors.
+   * @param numSplit The number of ways to split.
+   * @return a new instance of SparseSplit
+   */
+  public <T extends TType> SparseSplit<T> sparseSplit(Operand<TInt64> splitDim,
+      Operand<TInt64> indices, Operand<T> values, Operand<TInt64> shape, Long numSplit) {
+    return SparseSplit.create(scope, splitDim, indices, values, shape, numSplit);
+  }
+
+  /**
+   * Adds up a `SparseTensor` and a dense `Tensor`, producing a dense `Tensor`.
+   *  <p>
+   *  This Op does not require `a_indices` be sorted in standard lexicographic order.
+   *
+   * @param <U> data type for {@code output()} output
+   * @param aIndices 2-D.  The `indices` of the `SparseTensor`, with shape `[nnz, ndims]`.
+   * @param aValues 1-D.  The `values` of the `SparseTensor`, with shape `[nnz]`.
+   * @param aShape 1-D.  The `shape` of the `SparseTensor`, with shape `[ndims]`.
+   * @param b `ndims`-D Tensor.  With shape `a_shape`.
+   * @return a new instance of SparseTensorDenseAdd
+   */
+  public <U extends TType, T extends TNumber> SparseTensorDenseAdd<U> sparseTensorDenseAdd(
+      Operand<T> aIndices, Operand<U> aValues, Operand<T> aShape, Operand<U> b) {
+    return SparseTensorDenseAdd.create(scope, aIndices, aValues, aShape, b);
+  }
+
+  /**
+   * Multiply SparseTensor (of rank 2) "A" by dense matrix "B".
+   *  <p>
+   *  No validity checking is performed on the indices of A.  However, the following
+   *  input format is recommended for optimal behavior:
+   *  <p>
+   *  if adjoint_a == false:
+   *    A should be sorted in lexicographically increasing order.  Use SparseReorder
+   *    if you're not sure.
+   *  if adjoint_a == true:
+   *    A should be sorted in order of increasing dimension 1 (i.e., "column major"
+   *    order instead of "row major" order).
+   *
+   * @param <U> data type for {@code product()} output
+   * @param aIndices 2-D.  The `indices` of the `SparseTensor`, size `[nnz, 2]` Matrix.
+   * @param aValues 1-D.  The `values` of the `SparseTensor`, size `[nnz]` Vector.
+   * @param aShape 1-D.  The `shape` of the `SparseTensor`, size `[2]` Vector.
+   * @param b 2-D.  A dense Matrix.
+   * @param options carries optional attributes values
+   * @return a new instance of SparseTensorDenseMatMul
+   */
+  public <U extends TType, T extends TNumber> SparseTensorDenseMatMul<U> sparseTensorDenseMatMul(
+      Operand<T> aIndices, Operand<U> aValues, Operand<TInt64> aShape, Operand<U> b,
+      SparseTensorDenseMatMul.Options... options) {
+    return SparseTensorDenseMatMul.create(scope, aIndices, aValues, aShape, b, options);
+  }
+
+  /**
+   * Converts a sparse representation into a dense tensor.
+   *  <p>
+   *  Builds an array `dense` with shape `output_shape` such that
+   *  <pre>{@code
+   *  # If sparse_indices is scalar
+   *  dense[i] = (i == sparse_indices ? sparse_values : default_value)
+   *
+   *  # If sparse_indices is a vector, then for each i
+   *  dense[sparse_indices[i]] = sparse_values[i]
+   *
+   *  # If sparse_indices is an n by d matrix, then for each i in [0, n)
+   *  dense[sparse_indices[i][0], ..., sparse_indices[i][d-1]] = sparse_values[i]
+   *  }</pre>
+   *  All other values in `dense` are set to `default_value`.  If `sparse_values` is a
+   *  scalar, all sparse indices are set to this single value.
+   *  <p>
+   *  Indices should be sorted in lexicographic order, and indices must not
+   *  contain any repeats. If `validate_indices` is true, these properties
+   *  are checked during execution.
+   *
+   * @param <U> data type for {@code dense()} output
+   * @param sparseIndices 0-D, 1-D, or 2-D.  `sparse_indices[i]` contains the complete
+   *  index where `sparse_values[i]` will be placed.
+   * @param outputShape 1-D.  Shape of the dense output tensor.
+   * @param sparseValues 1-D.  Values corresponding to each row of `sparse_indices`,
+   *  or a scalar value to be used for all sparse indices.
+   * @param defaultValue Scalar value to set for indices not specified in
+   *  `sparse_indices`.
+   * @param options carries optional attributes values
+   * @return a new instance of SparseToDense
+   */
+  public <U extends TType, T extends TNumber> SparseToDense<U> sparseToDense(
+      Operand<T> sparseIndices, Operand<T> outputShape, Operand<U> sparseValues,
+      Operand<U> defaultValue, SparseToDense.Options... options) {
+    return SparseToDense.create(scope, sparseIndices, outputShape, sparseValues, defaultValue, options);
+  }
+
+  /**
+   * Applies set operation along last dimension of 2 `SparseTensor` inputs.
+   *  <p>
+   *  See SetOperationOp::SetOperationFromContext for values of `set_operation`.
+   *  <p>
+   *  If `validate_indices` is `True`, `sparse.SparseToSparseSetOperation` validates the
+   *  order and range of `set1` and `set2` indices.
+   *  <p>
+   *  Input `set1` is a `SparseTensor` represented by `set1_indices`, `set1_values`,
+   *  and `set1_shape`. For `set1` ranked `n`, 1st `n-1` dimensions must be the same
+   *  as `set2`. Dimension `n` contains values in a set, duplicates are allowed but
+   *  ignored.
+   *  <p>
+   *  Input `set2` is a `SparseTensor` represented by `set2_indices`, `set2_values`,
+   *  and `set2_shape`. For `set2` ranked `n`, 1st `n-1` dimensions must be the same
+   *  as `set1`. Dimension `n` contains values in a set, duplicates are allowed but
+   *  ignored.
+   *  <p>
+   *  If `validate_indices` is `True`, this op validates the order and range of `set1`
+   *  and `set2` indices.
+   *  <p>
+   *  Output `result` is a `SparseTensor` represented by `result_indices`,
+   *  `result_values`, and `result_shape`. For `set1` and `set2` ranked `n`, this
+   *  has rank `n` and the same 1st `n-1` dimensions as `set1` and `set2`. The `nth`
+   *  dimension contains the result of `set_operation` applied to the corresponding
+   *  `[0...n-1]` dimension of `set`.
+   *
+   * @param <T> data type for {@code resultValues()} output
+   * @param set1Indices 2D `Tensor`, indices of a `SparseTensor`. Must be in row-major
+   *  order.
+   * @param set1Values 1D `Tensor`, values of a `SparseTensor`. Must be in row-major
+   *  order.
+   * @param set1Shape 1D `Tensor`, shape of a `SparseTensor`. `set1_shape[0...n-1]` must
+   *  be the same as `set2_shape[0...n-1]`, `set1_shape[n]` is the
+   *  max set size across `0...n-1` dimensions.
+   * @param set2Indices 2D `Tensor`, indices of a `SparseTensor`. Must be in row-major
+   *  order.
+   * @param set2Values 1D `Tensor`, values of a `SparseTensor`. Must be in row-major
+   *  order.
+   * @param set2Shape 1D `Tensor`, shape of a `SparseTensor`. `set2_shape[0...n-1]` must
+   *  be the same as `set1_shape[0...n-1]`, `set2_shape[n]` is the
+   *  max set size across `0...n-1` dimensions.
+   * @param setOperation
+   * @param options carries optional attributes values
+   * @return a new instance of SparseToSparseSetOperation
+   */
+  public <T extends TType> SparseToSparseSetOperation<T> sparseToSparseSetOperation(
+      Operand<TInt64> set1Indices, Operand<T> set1Values, Operand<TInt64> set1Shape,
+      Operand<TInt64> set2Indices, Operand<T> set2Values, Operand<TInt64> set2Shape,
+      String setOperation, SparseToSparseSetOperation.Options... options) {
+    return SparseToSparseSetOperation.create(scope, set1Indices, set1Values, set1Shape, set2Indices, set2Values, set2Shape, setOperation, options);
+  }
+
+  /**
+   * Read `SparseTensors` from a `SparseTensorsMap` and concatenate them.
+   *  <p>
+   *  The input `sparse_handles` must be an `int64` matrix of shape `[N, 1]` where
+   *  `N` is the minibatch size and the rows correspond to the output handles of
+   *  `AddSparseToTensorsMap` or `AddManySparseToTensorsMap`.  The ranks of the
+   *  original `SparseTensor` objects that went into the given input ops must all
+   *  match.  When the final `SparseTensor` is created, it has rank one
+   *  higher than the ranks of the incoming `SparseTensor` objects
+   *  (they have been concatenated along a new row dimension on the left).
+   *  <p>
+   *  The output `SparseTensor` object's shape values for all dimensions but the
+   *  first are the max across the input `SparseTensor` objects' shape values
+   *  for the corresponding dimensions.  Its first shape value is `N`, the minibatch
+   *  size.
+   *  <p>
+   *  The input `SparseTensor` objects' indices are assumed ordered in
+   *  standard lexicographic order.  If this is not the case, after this
+   *  step run `SparseReorder` to restore index ordering.
+   *  <p>
+   *  For example, if the handles represent an input, which is a `[2, 3]` matrix
+   *  representing two original `SparseTensor` objects:
+   *  <pre>{@code
+   *      index = [ 0]
+   *              [10]
+   *              [20]
+   *      values = [1, 2, 3]
+   *      shape = [50]
+   *  }</pre>
+   *  and
+   *  <pre>{@code
+   *      index = [ 2]
+   *              [10]
+   *      values = [4, 5]
+   *      shape = [30]
+   *  }</pre>
+   *  then the final `SparseTensor` will be:
+   *  <pre>{@code
+   *      index = [0  0]
+   *              [0 10]
+   *              [0 20]
+   *              [1  2]
+   *              [1 10]
+   *      values = [1, 2, 3, 4, 5]
+   *      shape = [2 50]
+   *  }</pre>
+   *
+   * @param <T> data type for {@code sparseValues()} output
+   * @param sparseHandles 1-D, The `N` serialized `SparseTensor` objects.
+   *  Shape: `[N]`.
+   * @param dtype The `dtype` of the `SparseTensor` objects stored in the
+   *  `SparseTensorsMap`.
+   * @param options carries optional attributes values
+   * @return a new instance of TakeManySparseFromTensorsMap
+   */
+  public <T extends TType> TakeManySparseFromTensorsMap<T> takeManySparseFromTensorsMap(
+      Operand<TInt64> sparseHandles, Class<T> dtype,
+      TakeManySparseFromTensorsMap.Options... options) {
+    return TakeManySparseFromTensorsMap.create(scope, sparseHandles, dtype, options);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/StringsOps.java b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/StringsOps.java
new file mode 100644
index 00000000000..a09884817f8
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/StringsOps.java
@@ -0,0 +1,618 @@
+// Copyright 2020 The TensorFlow Authors. All Rights Reserved.
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+// ==============================================================================
+//
+// This class has been generated, DO NOT EDIT!
+//
+package org.tensorflow.op;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.op.strings.Join;
+import org.tensorflow.op.strings.Lower;
+import org.tensorflow.op.strings.ReduceJoin;
+import org.tensorflow.op.strings.RegexFullMatch;
+import org.tensorflow.op.strings.RegexReplace;
+import org.tensorflow.op.strings.StringFormat;
+import org.tensorflow.op.strings.StringLength;
+import org.tensorflow.op.strings.StringNGrams;
+import org.tensorflow.op.strings.StringSplit;
+import org.tensorflow.op.strings.Strip;
+import org.tensorflow.op.strings.Substr;
+import org.tensorflow.op.strings.ToHashBucket;
+import org.tensorflow.op.strings.ToHashBucketFast;
+import org.tensorflow.op.strings.ToHashBucketStrong;
+import org.tensorflow.op.strings.ToNumber;
+import org.tensorflow.op.strings.UnicodeScript;
+import org.tensorflow.op.strings.UnicodeTranscode;
+import org.tensorflow.op.strings.UnsortedSegmentJoin;
+import org.tensorflow.op.strings.Upper;
+import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.TString;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * An API for building {@code strings} operations as {@link Op Op}s
+ *
+ * @see {@link Ops}
+ */
+public final class StringsOps {
+  private final Scope scope;
+
+  StringsOps(Scope scope) {
+    this.scope = scope;
+  }
+
+  /**
+   * Joins the strings in the given list of string tensors into one tensor;
+   *  <p>
+   *  with the given separator (default is an empty separator).
+   *  <p>
+   *  Examples:
+   *  <p>
+   *  >>> s = ["hello", "world", "tensorflow"]
+   *  >>> tf.strings.join(s, " ")
+   *  <tf.Tensor: shape=(), dtype=string, numpy=b'hello world tensorflow'>
+   *
+   * @param inputs A list of string tensors.  The tensors must all have the same shape,
+   *  or be scalars.  Scalars may be mixed in; these will be broadcast to the shape
+   *  of non-scalar inputs.
+   * @param options carries optional attributes values
+   * @return a new instance of Join
+   */
+  public Join join(Iterable<Operand<TString>> inputs, Join.Options... options) {
+    return Join.create(scope, inputs, options);
+  }
+
+  /**
+   * Converts all uppercase characters into their respective lowercase replacements.
+   *  <p>
+   *  Example:
+   *  <p>
+   *  >>> tf.strings.lower("CamelCase string and ALL CAPS")
+   *  <tf.Tensor: shape=(), dtype=string, numpy=b'camelcase string and all caps'>
+   *
+   * @param input
+   * @param options carries optional attributes values
+   * @return a new instance of Lower
+   */
+  public Lower lower(Operand<TString> input, Lower.Options... options) {
+    return Lower.create(scope, input, options);
+  }
+
+  /**
+   * Joins a string Tensor across the given dimensions.
+   *  <p>
+   *  Computes the string join across dimensions in the given string Tensor of shape
+   *  `[\\(d_0, d_1, ..., d_{n-1}\\)]`.  Returns a new Tensor created by joining the input
+   *  strings with the given separator (default: empty string).  Negative indices are
+   *  counted backwards from the end, with `-1` being equivalent to `n - 1`.  If
+   *  indices are not specified, joins across all dimensions beginning from `n - 1`
+   *  through `0`.
+   *  <p>
+   *  For example:
+   *  <pre>{@code
+   *  # tensor `a` is [["a", "b"], ["c", "d"]]
+   *  tf.reduce_join(a, 0) ==> ["ac", "bd"]
+   *  tf.reduce_join(a, 1) ==> ["ab", "cd"]
+   *  tf.reduce_join(a, -2) = tf.reduce_join(a, 0) ==> ["ac", "bd"]
+   *  tf.reduce_join(a, -1) = tf.reduce_join(a, 1) ==> ["ab", "cd"]
+   *  tf.reduce_join(a, 0, keep_dims=True) ==> [["ac", "bd"]]
+   *  tf.reduce_join(a, 1, keep_dims=True) ==> [["ab"], ["cd"]]
+   *  tf.reduce_join(a, 0, separator=".") ==> ["a.c", "b.d"]
+   *  tf.reduce_join(a, [0, 1]) ==> "acbd"
+   *  tf.reduce_join(a, [1, 0]) ==> "abcd"
+   *  tf.reduce_join(a, []) ==> [["a", "b"], ["c", "d"]]
+   *  tf.reduce_join(a) = tf.reduce_join(a, [1, 0]) ==> "abcd"
+   *  }</pre>
+   *
+   * @param inputs The input to be joined.  All reduced indices must have non-zero size.
+   * @param reductionIndices The dimensions to reduce over.  Dimensions are reduced in the
+   *  order specified.  Omitting `reduction_indices` is equivalent to passing
+   *  `[n-1, n-2, ..., 0]`.  Negative indices from `-n` to `-1` are supported.
+   * @param options carries optional attributes values
+   * @return a new instance of ReduceJoin
+   */
+  public ReduceJoin reduceJoin(Operand<TString> inputs, Operand<TInt32> reductionIndices,
+      ReduceJoin.Options... options) {
+    return ReduceJoin.create(scope, inputs, reductionIndices, options);
+  }
+
+  /**
+   * Check if the input matches the regex pattern.
+   *  <p>
+   *  The input is a string tensor of any shape. The pattern is a scalar
+   *  string tensor which is applied to every element of the input tensor.
+   *  The boolean values (True or False) of the output tensor indicate
+   *  if the input matches the regex pattern provided.
+   *  <p>
+   *  The pattern follows the re2 syntax (https://github.com/google/re2/wiki/Syntax)
+   *  <p>
+   *  Examples:
+   *  <p>
+   *  >>> tf.strings.regex_full_match(["TF lib", "lib TF"], ".*lib$")
+   *  <tf.Tensor: shape=(2,), dtype=bool, numpy=array([ True, False])>
+   *  >>> tf.strings.regex_full_match(["TF lib", "lib TF"], ".*TF$")
+   *  <tf.Tensor: shape=(2,), dtype=bool, numpy=array([False,  True])>
+   *
+   * @param input A string tensor of the text to be processed.
+   * @param pattern A scalar string tensor containing the regular expression to match the input.
+   * @return a new instance of RegexFullMatch
+   */
+  public RegexFullMatch regexFullMatch(Operand<TString> input, Operand<TString> pattern) {
+    return RegexFullMatch.create(scope, input, pattern);
+  }
+
+  /**
+   * Replaces matches of the `pattern` regular expression in `input` with the
+   *  replacement string provided in `rewrite`.
+   *  <p>
+   *  It follows the re2 syntax (https://github.com/google/re2/wiki/Syntax)
+   *
+   * @param input The text to be processed.
+   * @param pattern The regular expression to be matched in the `input` strings.
+   * @param rewrite The rewrite string to be substituted for the `pattern` expression where it is
+   *  matched in the `input` strings.
+   * @param options carries optional attributes values
+   * @return a new instance of RegexReplace
+   */
+  public RegexReplace regexReplace(Operand<TString> input, Operand<TString> pattern,
+      Operand<TString> rewrite, RegexReplace.Options... options) {
+    return RegexReplace.create(scope, input, pattern, rewrite, options);
+  }
+
+  /**
+   * Formats a string template using a list of tensors.
+   *  <p>
+   *  Formats a string template using a list of tensors, pretty-printing tensor summaries.
+   *
+   * @param inputs The list of tensors to format into the placeholder string.
+   * @param options carries optional attributes values
+   * @return a new instance of StringFormat
+   */
+  public StringFormat stringFormat(Iterable<Operand<?>> inputs, StringFormat.Options... options) {
+    return StringFormat.create(scope, inputs, options);
+  }
+
+  /**
+   * String lengths of `input`.
+   *  <p>
+   *  Computes the length of each string given in the input tensor.
+   *  <p>
+   *  >>> strings = tf.constant(['Hello','TensorFlow', '\U0001F642'])
+   *  >>> tf.strings.length(strings).numpy() # default counts bytes
+   *  array([ 5, 10, 4], dtype=int32)
+   *  >>> tf.strings.length(strings, unit="UTF8_CHAR").numpy()
+   *  array([ 5, 10, 1], dtype=int32)
+   *
+   * @param input The strings for which to compute the length for each element.
+   * @param options carries optional attributes values
+   * @return a new instance of StringLength
+   */
+  public StringLength stringLength(Operand<TString> input, StringLength.Options... options) {
+    return StringLength.create(scope, input, options);
+  }
+
+  /**
+   * Creates ngrams from ragged string data.
+   *  <p>
+   *  This op accepts a ragged tensor with 1 ragged dimension containing only
+   *  strings and outputs a ragged tensor with 1 ragged dimension containing ngrams
+   *  of that string, joined along the innermost axis.
+   *
+   * @param <T> data type for {@code ngramsSplits()} output
+   * @param data The values tensor of the ragged string tensor to make ngrams out of. Must be a
+   *  1D string tensor.
+   * @param dataSplits The splits tensor of the ragged string tensor to make ngrams out of.
+   * @param separator The string to append between elements of the token. Use "" for no separator.
+   * @param ngramWidths The sizes of the ngrams to create.
+   * @param leftPad The string to use to pad the left side of the ngram sequence. Only used if
+   *  pad_width != 0.
+   * @param rightPad The string to use to pad the right side of the ngram sequence. Only used if
+   *  pad_width != 0.
+   * @param padWidth The number of padding elements to add to each side of each
+   *  sequence. Note that padding will never be greater than 'ngram_widths'-1
+   *  regardless of this value. If `pad_width=-1`, then add `max(ngram_widths)-1`
+   *  elements.
+   * @param preserveShortSequences
+   * @return a new instance of StringNGrams
+   */
+  public <T extends TNumber> StringNGrams<T> stringNGrams(Operand<TString> data,
+      Operand<T> dataSplits, String separator, List<Long> ngramWidths, String leftPad,
+      String rightPad, Long padWidth, Boolean preserveShortSequences) {
+    return StringNGrams.create(scope, data, dataSplits, separator, ngramWidths, leftPad, rightPad, padWidth, preserveShortSequences);
+  }
+
+  /**
+   * Split elements of `source` based on `sep` into a `SparseTensor`.
+   *  <p>
+   *  Let N be the size of source (typically N will be the batch size). Split each
+   *  element of `source` based on `sep` and return a `SparseTensor`
+   *  containing the split tokens. Empty tokens are ignored.
+   *  <p>
+   *  For example, N = 2, source[0] is 'hello world' and source[1] is 'a b c',
+   *  then the output will be
+   *  <pre>{@code
+   *  st.indices = [0, 0;
+   *                0, 1;
+   *                1, 0;
+   *                1, 1;
+   *                1, 2]
+   *  st.shape = [2, 3]
+   *  st.values = ['hello', 'world', 'a', 'b', 'c']
+   *  }</pre>
+   *  If `sep` is given, consecutive delimiters are not grouped together and are
+   *  deemed to delimit empty strings. For example, source of `"1<>2<><>3"` and
+   *  sep of `"<>"` returns `["1", "2", "", "3"]`. If `sep` is None or an empty
+   *  string, consecutive whitespace are regarded as a single separator, and the
+   *  result will contain no empty strings at the startor end if the string has
+   *  leading or trailing whitespace.
+   *  <p>
+   *  Note that the above mentioned behavior matches python's str.split.
+   *
+   * @param input `1-D` string `Tensor`, the strings to split.
+   * @param sep `0-D` string `Tensor`, the delimiter character.
+   * @param options carries optional attributes values
+   * @return a new instance of StringSplit
+   */
+  public StringSplit stringSplit(Operand<TString> input, Operand<TString> sep,
+      StringSplit.Options... options) {
+    return StringSplit.create(scope, input, sep, options);
+  }
+
+  /**
+   * Strip leading and trailing whitespaces from the Tensor.
+   *
+   * @param input A string `Tensor` of any shape.
+   * @return a new instance of Strip
+   */
+  public Strip strip(Operand<TString> input) {
+    return Strip.create(scope, input);
+  }
+
+  /**
+   * Return substrings from `Tensor` of strings.
+   *  <p>
+   *  For each string in the input `Tensor`, creates a substring starting at index
+   *  `pos` with a total length of `len`.
+   *  <p>
+   *  If `len` defines a substring that would extend beyond the length of the input
+   *  string, or if `len` is negative, then as many characters as possible are used.
+   *  <p>
+   *  A negative `pos` indicates distance within the string backwards from the end.
+   *  <p>
+   *  If `pos` specifies an index which is out of range for any of the input strings,
+   *  then an `InvalidArgumentError` is thrown.
+   *  <p>
+   *  `pos` and `len` must have the same shape, otherwise a `ValueError` is thrown on
+   *  Op creation.
+   *  <p>
+   *  <i>NOTE</i>: `strings.Substr` supports broadcasting up to two dimensions. More about
+   *  broadcasting
+   *  [here](http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html)
+   *  <p>
+   *  ---
+   *  <p>
+   *  Examples
+   *  <p>
+   *  Using scalar `pos` and `len`:
+   *  <pre>{@code
+   *  input = [b'Hello', b'World']
+   *  position = 1
+   *  length = 3
+   *
+   *  output = [b'ell', b'orl']
+   *  }</pre>
+   *  Using `pos` and `len` with same shape as `input`:
+   *  <pre>{@code
+   *  input = [[b'ten', b'eleven', b'twelve'],
+   *           [b'thirteen', b'fourteen', b'fifteen'],
+   *           [b'sixteen', b'seventeen', b'eighteen']]
+   *  position = [[1, 2, 3],
+   *              [1, 2, 3],
+   *              [1, 2, 3]]
+   *  length =   [[2, 3, 4],
+   *              [4, 3, 2],
+   *              [5, 5, 5]]
+   *
+   *  output = [[b'en', b'eve', b'lve'],
+   *            [b'hirt', b'urt', b'te'],
+   *            [b'ixtee', b'vente', b'hteen']]
+   *  }</pre>
+   *  Broadcasting `pos` and `len` onto `input`:
+   *  <pre>{@code
+   *  input = [[b'ten', b'eleven', b'twelve'],
+   *           [b'thirteen', b'fourteen', b'fifteen'],
+   *           [b'sixteen', b'seventeen', b'eighteen'],
+   *           [b'nineteen', b'twenty', b'twentyone']]
+   *  position = [1, 2, 3]
+   *  length =   [1, 2, 3]
+   *
+   *  output = [[b'e', b'ev', b'lve'],
+   *            [b'h', b'ur', b'tee'],
+   *            [b'i', b've', b'hte'],
+   *            [b'i', b'en', b'nty']]
+   *  }</pre>
+   *  Broadcasting `input` onto `pos` and `len`:
+   *  <pre>{@code
+   *  input = b'thirteen'
+   *  position = [1, 5, 7]
+   *  length =   [3, 2, 1]
+   *
+   *  output = [b'hir', b'ee', b'n']
+   *  }</pre>
+   *  Raises:
+   *  <p>
+   * `ValueError`: If the first argument cannot be converted to a
+   *       Tensor of `dtype string`.
+   * `InvalidArgumentError`: If indices are out of range.
+   * `ValueError`: If `pos` and `len` are not the same shape.
+   *
+   * @param input Tensor of strings
+   * @param pos Scalar defining the position of first character in each substring
+   * @param len Scalar defining the number of characters to include in each substring
+   * @param options carries optional attributes values
+   * @return a new instance of Substr
+   */
+  public <T extends TNumber> Substr substr(Operand<TString> input, Operand<T> pos, Operand<T> len,
+      Substr.Options... options) {
+    return Substr.create(scope, input, pos, len, options);
+  }
+
+  /**
+   * Converts each string in the input Tensor to its hash mod by a number of buckets.
+   *  <p>
+   *  The hash function is deterministic on the content of the string within the
+   *  process.
+   *  <p>
+   *  Note that the hash function may change from time to time.
+   *  This functionality will be deprecated and it's recommended to use
+   *  `tf.string_to_hash_bucket_fast()` or `tf.string_to_hash_bucket_strong()`.
+   *
+   * @param stringTensor
+   * @param numBuckets The number of buckets.
+   * @return a new instance of ToHashBucket
+   */
+  public ToHashBucket toHashBucket(Operand<TString> stringTensor, Long numBuckets) {
+    return ToHashBucket.create(scope, stringTensor, numBuckets);
+  }
+
+  /**
+   * Converts each string in the input Tensor to its hash mod by a number of buckets.
+   *  <p>
+   *  The hash function is deterministic on the content of the string within the
+   *  process and will never change. However, it is not suitable for cryptography.
+   *  This function may be used when CPU time is scarce and inputs are trusted or
+   *  unimportant. There is a risk of adversaries constructing inputs that all hash
+   *  to the same bucket. To prevent this problem, use a strong hash function with
+   *  `tf.string_to_hash_bucket_strong`.
+   *  <p>
+   *  Examples:
+   *  <p>
+   *  >>> tf.strings.to_hash_bucket_fast(["Hello", "TensorFlow", "2.x"], 3).numpy()
+   *  array([0, 2, 2])
+   *
+   * @param input The strings to assign a hash bucket.
+   * @param numBuckets The number of buckets.
+   * @return a new instance of ToHashBucketFast
+   */
+  public ToHashBucketFast toHashBucketFast(Operand<TString> input, Long numBuckets) {
+    return ToHashBucketFast.create(scope, input, numBuckets);
+  }
+
+  /**
+   * Converts each string in the input Tensor to its hash mod by a number of buckets.
+   *  <p>
+   *  The hash function is deterministic on the content of the string within the
+   *  process. The hash function is a keyed hash function, where attribute `key`
+   *  defines the key of the hash function. `key` is an array of 2 elements.
+   *  <p>
+   *  A strong hash is important when inputs may be malicious, e.g. URLs with
+   *  additional components. Adversaries could try to make their inputs hash to the
+   *  same bucket for a denial-of-service attack or to skew the results. A strong
+   *  hash can be used to make it difficult to find inputs with a skewed hash value
+   *  distribution over buckets. This requires that the hash function is
+   *  seeded by a high-entropy (random) "key" unknown to the adversary.
+   *  <p>
+   *  The additional robustness comes at a cost of roughly 4x higher compute
+   *  time than `tf.string_to_hash_bucket_fast`.
+   *  <p>
+   *  Examples:
+   *  <p>
+   *  >>> tf.strings.to_hash_bucket_strong(["Hello", "TF"], 3, [1, 2]).numpy()
+   *  array([2, 0])
+   *
+   * @param input The strings to assign a hash bucket.
+   * @param numBuckets The number of buckets.
+   * @param key The key used to seed the hash function, passed as a list of two uint64
+   *  elements.
+   * @return a new instance of ToHashBucketStrong
+   */
+  public ToHashBucketStrong toHashBucketStrong(Operand<TString> input, Long numBuckets,
+      List<Long> key) {
+    return ToHashBucketStrong.create(scope, input, numBuckets, key);
+  }
+
+  /**
+   * Converts each string in the input Tensor to the specified numeric type.
+   *  <p>
+   *  (Note that int32 overflow results in an error while float overflow
+   *  results in a rounded value.)
+   *  <p>
+   *  Example:
+   *  <p>
+   *  >>> strings = ["5.0", "3.0", "7.0"]
+   *  >>> tf.strings.to_number(strings)
+   *  <tf.Tensor: shape=(3,), dtype=float32, numpy=array([5., 3., 7.], dtype=float32)>
+   *
+   * @param <T> data type for {@code output()} output
+   * @param stringTensor
+   * @return a new instance of ToNumber
+   */
+  public ToNumber<TFloat32> toNumber(Operand<TString> stringTensor) {
+    return ToNumber.create(scope, stringTensor);
+  }
+
+  /**
+   * Converts each string in the input Tensor to the specified numeric type.
+   *  <p>
+   *  (Note that int32 overflow results in an error while float overflow
+   *  results in a rounded value.)
+   *  <p>
+   *  Example:
+   *  <p>
+   *  >>> strings = ["5.0", "3.0", "7.0"]
+   *  >>> tf.strings.to_number(strings)
+   *  <tf.Tensor: shape=(3,), dtype=float32, numpy=array([5., 3., 7.], dtype=float32)>
+   *
+   * @param <T> data type for {@code output()} output
+   * @param stringTensor
+   * @param outType The numeric type to interpret each string in `string_tensor` as.
+   * @return a new instance of ToNumber
+   */
+  public <T extends TNumber> ToNumber<T> toNumber(Operand<TString> stringTensor, Class<T> outType) {
+    return ToNumber.create(scope, stringTensor, outType);
+  }
+
+  /**
+   * Determine the script codes of a given tensor of Unicode integer code points.
+   *  <p>
+   *  This operation converts Unicode code points to script codes corresponding to
+   *  each code point. Script codes correspond to International Components for
+   *  Unicode (ICU) UScriptCode values. See http://icu-project.org/apiref/icu4c/uscript_8h.html.
+   *  Returns -1 (USCRIPT_INVALID_CODE) for invalid codepoints. Output shape will
+   *  match input shape.
+   *  <p>
+   *  Examples:
+   *  <p>
+   *  >>> tf.strings.unicode_script([1, 31, 38])
+   *  <tf.Tensor: shape=(3,), dtype=int32, numpy=array([0, 0, 0], dtype=int32)>
+   *
+   * @param input A Tensor of int32 Unicode code points.
+   * @return a new instance of UnicodeScript
+   */
+  public UnicodeScript unicodeScript(Operand<TInt32> input) {
+    return UnicodeScript.create(scope, input);
+  }
+
+  /**
+   * Transcode the input text from a source encoding to a destination encoding.
+   *  <p>
+   *  The input is a string tensor of any shape. The output is a string tensor of
+   *  the same shape containing the transcoded strings. Output strings are always
+   *  valid unicode. If the input contains invalid encoding positions, the
+   *  `errors` attribute sets the policy for how to deal with them. If the default
+   *  error-handling policy is used, invalid formatting will be substituted in the
+   *  output by the `replacement_char`. If the errors policy is to `ignore`, any
+   *  invalid encoding positions in the input are skipped and not included in the
+   *  output. If it set to `strict` then any invalid formatting will result in an
+   *  InvalidArgument error.
+   *  <p>
+   *  This operation can be used with `output_encoding = input_encoding` to enforce
+   *  correct formatting for inputs even if they are already in the desired encoding.
+   *  <p>
+   *  If the input is prefixed by a Byte Order Mark needed to determine encoding
+   *  (e.g. if the encoding is UTF-16 and the BOM indicates big-endian), then that
+   *  BOM will be consumed and not emitted into the output. If the input encoding
+   *  is marked with an explicit endianness (e.g. UTF-16-BE), then the BOM is
+   *  interpreted as a non-breaking-space and is preserved in the output (including
+   *  always for UTF-8).
+   *  <p>
+   *  The end result is that if the input is marked as an explicit endianness the
+   *  transcoding is faithful to all codepoints in the source. If it is not marked
+   *  with an explicit endianness, the BOM is not considered part of the string itself
+   *  but as metadata, and so is not preserved in the output.
+   *  <p>
+   *  Examples:
+   *  <p>
+   *  >>> tf.strings.unicode_transcode(["Hello", "TensorFlow", "2.x"], "UTF-8", "UTF-16-BE")
+   *  <tf.Tensor: shape=(3,), dtype=string, numpy=
+   *  array([b'\x00H\x00e\x00l\x00l\x00o',
+   *         b'\x00T\x00e\x00n\x00s\x00o\x00r\x00F\x00l\x00o\x00w',
+   *         b'\x002\x00.\x00x'], dtype=object)>
+   *  >>> tf.strings.unicode_transcode(["A", "B", "C"], "US ASCII", "UTF-8").numpy()
+   *  array([b'A', b'B', b'C'], dtype=object)
+   *
+   * @param input The text to be processed. Can have any shape.
+   * @param inputEncoding Text encoding of the input strings. This is any of the encodings supported
+   *  by ICU ucnv algorithmic converters. Examples: `"UTF-16", "US ASCII", "UTF-8"`.
+   * @param outputEncoding The unicode encoding to use in the output. Must be one of
+   *  `"UTF-8", "UTF-16-BE", "UTF-32-BE"`. Multi-byte encodings will be big-endian.
+   * @param options carries optional attributes values
+   * @return a new instance of UnicodeTranscode
+   */
+  public UnicodeTranscode unicodeTranscode(Operand<TString> input, String inputEncoding,
+      String outputEncoding, UnicodeTranscode.Options... options) {
+    return UnicodeTranscode.create(scope, input, inputEncoding, outputEncoding, options);
+  }
+
+  /**
+   * Joins the elements of `inputs` based on `segment_ids`.
+   *  <p>
+   *  Computes the string join along segments of a tensor.
+   *  Given `segment_ids` with rank `N` and `data` with rank `N+M`:
+   *  <p>
+   *      `output[i, k1...kM] = strings.join([data[j1...jN, k1...kM])`
+   *  <p>
+   *  where the join is over all [j1...jN] such that segment_ids[j1...jN] = i.
+   *  Strings are joined in row-major order.
+   *  <p>
+   *  For example:
+   *  <pre>{@code
+   *  inputs = [['Y', 'q', 'c'], ['Y', '6', '6'], ['p', 'G', 'a']]
+   *  output_array = string_ops.unsorted_segment_join(inputs=inputs,
+   *                                                  segment_ids=[1, 0, 1],
+   *                                                  num_segments=2,
+   *                                                  separator=':'))
+   *  # output_array ==> [['Y', '6', '6'], ['Y:p', 'q:G', 'c:a']]
+   *
+   *
+   *  inputs = ['this', 'is', 'a', 'test']
+   *  output_array = string_ops.unsorted_segment_join(inputs=inputs,
+   *                                                  segment_ids=[0, 0, 0, 0],
+   *                                                  num_segments=1,
+   *                                                  separator=':'))
+   *  # output_array ==> ['this:is:a:test']
+   *  }</pre>
+   *
+   * @param inputs The input to be joined.
+   * @param segmentIds A tensor whose shape is a prefix of data.shape.  Negative segment ids are not
+   *  supported.
+   * @param numSegments A scalar.
+   * @param options carries optional attributes values
+   * @return a new instance of UnsortedSegmentJoin
+   */
+  public <T extends TNumber, U extends TNumber> UnsortedSegmentJoin unsortedSegmentJoin(
+      Operand<TString> inputs, Operand<T> segmentIds, Operand<U> numSegments,
+      UnsortedSegmentJoin.Options... options) {
+    return UnsortedSegmentJoin.create(scope, inputs, segmentIds, numSegments, options);
+  }
+
+  /**
+   * Converts all lowercase characters into their respective uppercase replacements.
+   *  <p>
+   *  Example:
+   *  <p>
+   *  >>> tf.strings.upper("CamelCase string and ALL CAPS")
+   *  <tf.Tensor: shape=(), dtype=string, numpy=b'CAMELCASE STRING AND ALL CAPS'>
+   *
+   * @param input
+   * @param options carries optional attributes values
+   * @return a new instance of Upper
+   */
+  public Upper upper(Operand<TString> input, Upper.Options... options) {
+    return Upper.create(scope, input, options);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/SummaryOps.java b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/SummaryOps.java
new file mode 100644
index 00000000000..6143335b8fd
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/SummaryOps.java
@@ -0,0 +1,198 @@
+// Copyright 2020 The TensorFlow Authors. All Rights Reserved.
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+// ==============================================================================
+//
+// This class has been generated, DO NOT EDIT!
+//
+package org.tensorflow.op;
+
+import org.tensorflow.Operand;
+import org.tensorflow.op.summary.AudioSummary;
+import org.tensorflow.op.summary.HistogramSummary;
+import org.tensorflow.op.summary.ImageSummary;
+import org.tensorflow.op.summary.MergeSummary;
+import org.tensorflow.op.summary.ScalarSummary;
+import org.tensorflow.op.summary.TensorSummary;
+import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.TString;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * An API for building {@code summary} operations as {@link Op Op}s
+ *
+ * @see {@link Ops}
+ */
+public final class SummaryOps {
+  private final Scope scope;
+
+  SummaryOps(Scope scope) {
+    this.scope = scope;
+  }
+
+  /**
+   * Outputs a `Summary` protocol buffer with audio.
+   *  <p>
+   *  The summary has up to `max_outputs` summary values containing audio. The
+   *  audio is built from `tensor` which must be 3-D with shape `[batch_size,
+   *  frames, channels]` or 2-D with shape `[batch_size, frames]`. The values are
+   *  assumed to be in the range of `[-1.0, 1.0]` with a sample rate of `sample_rate`.
+   *  <p>
+   *  The `tag` argument is a scalar `Tensor` of type `string`.  It is used to
+   *  build the `tag` of the summary values:
+   *  <ul>
+   *  <li>
+   *  If `max_outputs` is 1, the summary value tag is '<i>tag</i>/audio'.
+   *  </li>
+   *  <li>
+   *  If `max_outputs` is greater than 1, the summary value tags are
+   *     generated sequentially as '<i>tag</i>/audio/0', '<i>tag</i>/audio/1', etc.
+   *
+   * @param tag Scalar. Used to build the `tag` attribute of the summary values.
+   * @param tensor 2-D of shape `[batch_size, frames]`.
+   * @param sampleRate The sample rate of the signal in hertz.
+   * @param options carries optional attributes values
+   * @return a new instance of AudioSummary
+   */
+  public AudioSummary audioSummary(Operand<TString> tag, Operand<TFloat32> tensor,
+      Operand<TFloat32> sampleRate, AudioSummary.Options... options) {
+    return AudioSummary.create(scope, tag, tensor, sampleRate, options);
+  }
+
+  /**
+   * Outputs a `Summary` protocol buffer with a histogram.
+   *  <p>
+   *  The generated
+   *  [`Summary`](https://www.tensorflow.org/code/tensorflow/core/framework/summary.proto)
+   *  has one summary value containing a histogram for `values`.
+   *  <p>
+   *  This op reports an `InvalidArgument` error if any value is not finite.
+   *
+   * @param tag Scalar.  Tag to use for the `Summary.Value`.
+   * @param values Any shape. Values to use to build the histogram.
+   * @return a new instance of HistogramSummary
+   */
+  public <T extends TNumber> HistogramSummary histogramSummary(Operand<TString> tag,
+      Operand<T> values) {
+    return HistogramSummary.create(scope, tag, values);
+  }
+
+  /**
+   * Outputs a `Summary` protocol buffer with images.
+   *  <p>
+   *  The summary has up to `max_images` summary values containing images. The
+   *  images are built from `tensor` which must be 4-D with shape `[batch_size,
+   *  height, width, channels]` and where `channels` can be:
+   *  <ul>
+   *  <li>
+   *  1: `tensor` is interpreted as Grayscale.
+   *  </li>
+   *  <li>
+   *  3: `tensor` is interpreted as RGB.
+   *  </li>
+   *  <li>
+   *  4: `tensor` is interpreted as RGBA.
+   *  </li>
+   *  </ul>
+   *  The images have the same number of channels as the input tensor. For float
+   *  input, the values are normalized one image at a time to fit in the range
+   *  `[0, 255]`.  `uint8` values are unchanged.  The op uses two different
+   *  normalization algorithms:
+   *  <ul>
+   *  <li>
+   *  If the input values are all positive, they are rescaled so the largest one
+   *     is 255.
+   *  </li>
+   *  <li>
+   *  If any input value is negative, the values are shifted so input value 0.0
+   *     is at 127.  They are then rescaled so that either the smallest value is 0,
+   *     or the largest one is 255.
+   *  </li>
+   *  </ul>
+   *  The `tag` argument is a scalar `Tensor` of type `string`.  It is used to
+   *  build the `tag` of the summary values:
+   *  <ul>
+   *  <li>
+   *  If `max_images` is 1, the summary value tag is '<i>tag</i>/image'.
+   *  </li>
+   *  <li>
+   *  If `max_images` is greater than 1, the summary value tags are
+   *     generated sequentially as '<i>tag</i>/image/0', '<i>tag</i>/image/1', etc.
+   *  </li>
+   *  </ul>
+   *  The `bad_color` argument is the color to use in the generated images for
+   *  non-finite input values.  It is a `uint8` 1-D tensor of length `channels`.
+   *  Each element must be in the range `[0, 255]` (It represents the value of a
+   *  pixel in the output image).  Non-finite values in the input tensor are
+   *  replaced by this tensor in the output image.  The default value is the color
+   *  red.
+   *
+   * @param tag Scalar. Used to build the `tag` attribute of the summary values.
+   * @param tensor 4-D of shape `[batch_size, height, width, channels]` where
+   *  `channels` is 1, 3, or 4.
+   * @param options carries optional attributes values
+   * @return a new instance of ImageSummary
+   */
+  public <T extends TNumber> ImageSummary imageSummary(Operand<TString> tag, Operand<T> tensor,
+      ImageSummary.Options... options) {
+    return ImageSummary.create(scope, tag, tensor, options);
+  }
+
+  /**
+   * Merges summaries.
+   *  <p>
+   *  This op creates a
+   *  [`Summary`](https://www.tensorflow.org/code/tensorflow/core/framework/summary.proto)
+   *  protocol buffer that contains the union of all the values in the input
+   *  summaries.
+   *  <p>
+   *  When the Op is run, it reports an `InvalidArgument` error if multiple values
+   *  in the summaries to merge use the same tag.
+   *
+   * @param inputs Can be of any shape.  Each must contain serialized `Summary` protocol
+   *  buffers.
+   * @return a new instance of MergeSummary
+   */
+  public MergeSummary mergeSummary(Iterable<Operand<TString>> inputs) {
+    return MergeSummary.create(scope, inputs);
+  }
+
+  /**
+   * Outputs a `Summary` protocol buffer with scalar values.
+   *  <p>
+   *  The input `tags` and `values` must have the same shape.  The generated summary
+   *  has a summary value for each tag-value pair in `tags` and `values`.
+   *
+   * @param tags Tags for the summary.
+   * @param values Same shape as `tags.  Values for the summary.
+   * @return a new instance of ScalarSummary
+   */
+  public <T extends TNumber> ScalarSummary scalarSummary(Operand<TString> tags, Operand<T> values) {
+    return ScalarSummary.create(scope, tags, values);
+  }
+
+  /**
+   * Outputs a `Summary` protocol buffer with a tensor and per-plugin data.
+   *
+   * @param tag A string attached to this summary. Used for organization in TensorBoard.
+   * @param tensor A tensor to serialize.
+   * @param serializedSummaryMetadata A serialized SummaryMetadata proto. Contains plugin
+   *  data.
+   * @return a new instance of TensorSummary
+   */
+  public <T extends TType> TensorSummary tensorSummary(Operand<TString> tag, Operand<T> tensor,
+      Operand<TString> serializedSummaryMetadata) {
+    return TensorSummary.create(scope, tag, tensor, serializedSummaryMetadata);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/TrainOps.java b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/TrainOps.java
new file mode 100644
index 00000000000..1e253ac9956
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/TrainOps.java
@@ -0,0 +1,1660 @@
+// Copyright 2020 The TensorFlow Authors. All Rights Reserved.
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+// ==============================================================================
+//
+// This class has been generated, DO NOT EDIT!
+//
+package org.tensorflow.op;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.train.AccumulatorApplyGradient;
+import org.tensorflow.op.train.AccumulatorNumAccumulated;
+import org.tensorflow.op.train.AccumulatorSetGlobalStep;
+import org.tensorflow.op.train.AccumulatorTakeGradient;
+import org.tensorflow.op.train.ApplyAdadelta;
+import org.tensorflow.op.train.ApplyAdagrad;
+import org.tensorflow.op.train.ApplyAdagradDa;
+import org.tensorflow.op.train.ApplyAdam;
+import org.tensorflow.op.train.ApplyAddSign;
+import org.tensorflow.op.train.ApplyCenteredRmsProp;
+import org.tensorflow.op.train.ApplyFtrl;
+import org.tensorflow.op.train.ApplyGradientDescent;
+import org.tensorflow.op.train.ApplyMomentum;
+import org.tensorflow.op.train.ApplyPowerSign;
+import org.tensorflow.op.train.ApplyProximalAdagrad;
+import org.tensorflow.op.train.ApplyProximalGradientDescent;
+import org.tensorflow.op.train.ApplyRmsProp;
+import org.tensorflow.op.train.BatchMatMul;
+import org.tensorflow.op.train.ConditionalAccumulator;
+import org.tensorflow.op.train.GenerateVocabRemapping;
+import org.tensorflow.op.train.MergeV2Checkpoints;
+import org.tensorflow.op.train.NegTrain;
+import org.tensorflow.op.train.PreventGradient;
+import org.tensorflow.op.train.ResourceApplyAdadelta;
+import org.tensorflow.op.train.ResourceApplyAdagradDa;
+import org.tensorflow.op.train.ResourceApplyAdam;
+import org.tensorflow.op.train.ResourceApplyAdamWithAmsgrad;
+import org.tensorflow.op.train.ResourceApplyAddSign;
+import org.tensorflow.op.train.ResourceApplyCenteredRmsProp;
+import org.tensorflow.op.train.ResourceApplyFtrl;
+import org.tensorflow.op.train.ResourceApplyGradientDescent;
+import org.tensorflow.op.train.ResourceApplyKerasMomentum;
+import org.tensorflow.op.train.ResourceApplyMomentum;
+import org.tensorflow.op.train.ResourceApplyPowerSign;
+import org.tensorflow.op.train.ResourceApplyProximalAdagrad;
+import org.tensorflow.op.train.ResourceApplyProximalGradientDescent;
+import org.tensorflow.op.train.ResourceApplyRmsProp;
+import org.tensorflow.op.train.ResourceSparseApplyAdadelta;
+import org.tensorflow.op.train.ResourceSparseApplyAdagrad;
+import org.tensorflow.op.train.ResourceSparseApplyAdagradDa;
+import org.tensorflow.op.train.ResourceSparseApplyCenteredRmsProp;
+import org.tensorflow.op.train.ResourceSparseApplyFtrl;
+import org.tensorflow.op.train.ResourceSparseApplyKerasMomentum;
+import org.tensorflow.op.train.ResourceSparseApplyMomentum;
+import org.tensorflow.op.train.ResourceSparseApplyProximalAdagrad;
+import org.tensorflow.op.train.ResourceSparseApplyProximalGradientDescent;
+import org.tensorflow.op.train.ResourceSparseApplyRmsProp;
+import org.tensorflow.op.train.Restore;
+import org.tensorflow.op.train.RestoreSlice;
+import org.tensorflow.op.train.Save;
+import org.tensorflow.op.train.SaveSlices;
+import org.tensorflow.op.train.SdcaFprint;
+import org.tensorflow.op.train.SdcaShrinkL1;
+import org.tensorflow.op.train.SparseApplyAdadelta;
+import org.tensorflow.op.train.SparseApplyAdagradDa;
+import org.tensorflow.op.train.SparseApplyCenteredRmsProp;
+import org.tensorflow.op.train.SparseApplyFtrl;
+import org.tensorflow.op.train.SparseApplyMomentum;
+import org.tensorflow.op.train.SparseApplyProximalAdagrad;
+import org.tensorflow.op.train.SparseApplyProximalGradientDescent;
+import org.tensorflow.op.train.SparseApplyRmsProp;
+import org.tensorflow.op.train.TileGrad;
+import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.TString;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * An API for building {@code train} operations as {@link Op Op}s
+ *
+ * @see {@link Ops}
+ */
+public final class TrainOps {
+  private final Scope scope;
+
+  TrainOps(Scope scope) {
+    this.scope = scope;
+  }
+
+  /**
+   * Applies a gradient to a given accumulator.
+   *  <p>
+   *  Does not add if local_step is lesser than the accumulator's global_step.
+   *
+   * @param handle The handle to a accumulator.
+   * @param localStep The local_step value at which the gradient was computed.
+   * @param gradient A tensor of the gradient to be accumulated.
+   * @return a new instance of AccumulatorApplyGradient
+   */
+  public <T extends TType> AccumulatorApplyGradient accumulatorApplyGradient(
+      Operand<TString> handle, Operand<TInt64> localStep, Operand<T> gradient) {
+    return AccumulatorApplyGradient.create(scope, handle, localStep, gradient);
+  }
+
+  /**
+   * Returns the number of gradients aggregated in the given accumulators.
+   *
+   * @param handle The handle to an accumulator.
+   * @return a new instance of AccumulatorNumAccumulated
+   */
+  public AccumulatorNumAccumulated accumulatorNumAccumulated(Operand<TString> handle) {
+    return AccumulatorNumAccumulated.create(scope, handle);
+  }
+
+  /**
+   * Updates the accumulator with a new value for global_step.
+   *  <p>
+   *  Logs warning if the accumulator's value is already higher than
+   *  new_global_step.
+   *
+   * @param handle The handle to an accumulator.
+   * @param newGlobalStep The new global_step value to set.
+   * @return a new instance of AccumulatorSetGlobalStep
+   */
+  public AccumulatorSetGlobalStep accumulatorSetGlobalStep(Operand<TString> handle,
+      Operand<TInt64> newGlobalStep) {
+    return AccumulatorSetGlobalStep.create(scope, handle, newGlobalStep);
+  }
+
+  /**
+   * Extracts the average gradient in the given ConditionalAccumulator.
+   *  <p>
+   *  The op blocks until sufficient (i.e., more than num_required)
+   *  gradients have been accumulated.  If the accumulator has already
+   *  aggregated more than num_required gradients, it returns the average of
+   *  the accumulated gradients.  Also automatically increments the recorded
+   *  global_step in the accumulator by 1, and resets the aggregate to 0.
+   *
+   * @param <T> data type for {@code average()} output
+   * @param handle The handle to an accumulator.
+   * @param numRequired Number of gradients required before we return an aggregate.
+   * @param dtype The data type of accumulated gradients. Needs to correspond to the type
+   *  of the accumulator.
+   * @return a new instance of AccumulatorTakeGradient
+   */
+  public <T extends TType> AccumulatorTakeGradient<T> accumulatorTakeGradient(
+      Operand<TString> handle, Operand<TInt32> numRequired, Class<T> dtype) {
+    return AccumulatorTakeGradient.create(scope, handle, numRequired, dtype);
+  }
+
+  /**
+   * Update '*var' according to the adadelta scheme.
+   *  <p>
+   *  accum = rho() * accum + (1 - rho()) * grad.square();
+   *  update = (update_accum + epsilon).sqrt() * (accum + epsilon()).rsqrt() * grad;
+   *  update_accum = rho() * update_accum + (1 - rho()) * update.square();
+   *  var -= update;
+   *
+   * @param <T> data type for {@code out()} output
+   * @param var Should be from a Variable().
+   * @param accum Should be from a Variable().
+   * @param accumUpdate Should be from a Variable().
+   * @param lr Scaling factor. Must be a scalar.
+   * @param rho Decay factor. Must be a scalar.
+   * @param epsilon Constant factor. Must be a scalar.
+   * @param grad The gradient.
+   * @param options carries optional attributes values
+   * @return a new instance of ApplyAdadelta
+   */
+  public <T extends TType> ApplyAdadelta<T> applyAdadelta(Operand<T> var, Operand<T> accum,
+      Operand<T> accumUpdate, Operand<T> lr, Operand<T> rho, Operand<T> epsilon, Operand<T> grad,
+      ApplyAdadelta.Options... options) {
+    return ApplyAdadelta.create(scope, var, accum, accumUpdate, lr, rho, epsilon, grad, options);
+  }
+
+  /**
+   * Update '*var' according to the adagrad scheme.
+   *  <p>
+   *  accum += grad * grad
+   *  var -= lr * grad * (1 / sqrt(accum))
+   *
+   * @param <T> data type for {@code out()} output
+   * @param var Should be from a Variable().
+   * @param accum Should be from a Variable().
+   * @param lr Scaling factor. Must be a scalar.
+   * @param grad The gradient.
+   * @param options carries optional attributes values
+   * @return a new instance of ApplyAdagrad
+   */
+  public <T extends TType> ApplyAdagrad<T> applyAdagrad(Operand<T> var, Operand<T> accum,
+      Operand<T> lr, Operand<T> grad, ApplyAdagrad.Options... options) {
+    return ApplyAdagrad.create(scope, var, accum, lr, grad, options);
+  }
+
+  /**
+   * Update '*var' according to the proximal adagrad scheme.
+   *
+   * @param <T> data type for {@code out()} output
+   * @param var Should be from a Variable().
+   * @param gradientAccumulator Should be from a Variable().
+   * @param gradientSquaredAccumulator Should be from a Variable().
+   * @param grad The gradient.
+   * @param lr Scaling factor. Must be a scalar.
+   * @param l1 L1 regularization. Must be a scalar.
+   * @param l2 L2 regularization. Must be a scalar.
+   * @param globalStep Training step number. Must be a scalar.
+   * @param options carries optional attributes values
+   * @return a new instance of ApplyAdagradDa
+   */
+  public <T extends TType> ApplyAdagradDa<T> applyAdagradDa(Operand<T> var,
+      Operand<T> gradientAccumulator, Operand<T> gradientSquaredAccumulator, Operand<T> grad,
+      Operand<T> lr, Operand<T> l1, Operand<T> l2, Operand<TInt64> globalStep,
+      ApplyAdagradDa.Options... options) {
+    return ApplyAdagradDa.create(scope, var, gradientAccumulator, gradientSquaredAccumulator, grad, lr, l1, l2, globalStep, options);
+  }
+
+  /**
+   * Update '*var' according to the Adam algorithm.
+   *  <p>
+   *  $$lr_t := \text{learning\_rate} * \sqrt{1 - beta_2^t} / (1 - beta_1^t)$$
+   *  $$m_t := beta_1 * m_{t-1} + (1 - beta_1) * g$$
+   *  $$v_t := beta_2 * v_{t-1} + (1 - beta_2) * g * g$$
+   *  $$variable := variable - lr_t * m_t / (\sqrt{v_t} + \epsilon)$$
+   *
+   * @param <T> data type for {@code out()} output
+   * @param var Should be from a Variable().
+   * @param m Should be from a Variable().
+   * @param v Should be from a Variable().
+   * @param beta1Power Must be a scalar.
+   * @param beta2Power Must be a scalar.
+   * @param lr Scaling factor. Must be a scalar.
+   * @param beta1 Momentum factor. Must be a scalar.
+   * @param beta2 Momentum factor. Must be a scalar.
+   * @param epsilon Ridge term. Must be a scalar.
+   * @param grad The gradient.
+   * @param options carries optional attributes values
+   * @return a new instance of ApplyAdam
+   */
+  public <T extends TType> ApplyAdam<T> applyAdam(Operand<T> var, Operand<T> m, Operand<T> v,
+      Operand<T> beta1Power, Operand<T> beta2Power, Operand<T> lr, Operand<T> beta1,
+      Operand<T> beta2, Operand<T> epsilon, Operand<T> grad, ApplyAdam.Options... options) {
+    return ApplyAdam.create(scope, var, m, v, beta1Power, beta2Power, lr, beta1, beta2, epsilon, grad, options);
+  }
+
+  /**
+   * Update '*var' according to the AddSign update.
+   *  <p>
+   *  m_t <- beta1 * m_{t-1} + (1 - beta1) * g
+   *  update <- (alpha + sign_decay * sign(g) *sign(m)) * g
+   *  variable <- variable - lr_t * update
+   *
+   * @param <T> data type for {@code out()} output
+   * @param var Should be from a Variable().
+   * @param m Should be from a Variable().
+   * @param lr Scaling factor. Must be a scalar.
+   * @param alpha Must be a scalar.
+   * @param signDecay Must be a scalar.
+   * @param beta Must be a scalar.
+   * @param grad The gradient.
+   * @param options carries optional attributes values
+   * @return a new instance of ApplyAddSign
+   */
+  public <T extends TType> ApplyAddSign<T> applyAddSign(Operand<T> var, Operand<T> m, Operand<T> lr,
+      Operand<T> alpha, Operand<T> signDecay, Operand<T> beta, Operand<T> grad,
+      ApplyAddSign.Options... options) {
+    return ApplyAddSign.create(scope, var, m, lr, alpha, signDecay, beta, grad, options);
+  }
+
+  /**
+   * Update '*var' according to the centered RMSProp algorithm.
+   *  <p>
+   *  The centered RMSProp algorithm uses an estimate of the centered second moment
+   *  (i.e., the variance) for normalization, as opposed to regular RMSProp, which
+   *  uses the (uncentered) second moment. This often helps with training, but is
+   *  slightly more expensive in terms of computation and memory.
+   *  <p>
+   *  Note that in dense implementation of this algorithm, mg, ms, and mom will
+   *  update even if the grad is zero, but in this sparse implementation, mg, ms,
+   *  and mom will not update in iterations during which the grad is zero.
+   *  <p>
+   *  mean_square = decay * mean_square + (1-decay) * gradient ** 2
+   *  mean_grad = decay * mean_grad + (1-decay) * gradient
+   *  <p>
+   *  Delta = learning_rate * gradient / sqrt(mean_square + epsilon - mean_grad ** 2)
+   *  <p>
+   *  mg <- rho * mg_{t-1} + (1-rho) * grad
+   *  ms <- rho * ms_{t-1} + (1-rho) * grad * grad
+   *  mom <- momentum * mom_{t-1} + lr * grad / sqrt(ms - mg * mg + epsilon)
+   *  var <- var - mom
+   *
+   * @param <T> data type for {@code out()} output
+   * @param var Should be from a Variable().
+   * @param mg Should be from a Variable().
+   * @param ms Should be from a Variable().
+   * @param mom Should be from a Variable().
+   * @param lr Scaling factor. Must be a scalar.
+   * @param rho Decay rate. Must be a scalar.
+   * @param momentum
+   * @param epsilon Ridge term. Must be a scalar.
+   * @param grad The gradient.
+   * @param options carries optional attributes values
+   * @return a new instance of ApplyCenteredRmsProp
+   */
+  public <T extends TType> ApplyCenteredRmsProp<T> applyCenteredRmsProp(Operand<T> var,
+      Operand<T> mg, Operand<T> ms, Operand<T> mom, Operand<T> lr, Operand<T> rho,
+      Operand<T> momentum, Operand<T> epsilon, Operand<T> grad,
+      ApplyCenteredRmsProp.Options... options) {
+    return ApplyCenteredRmsProp.create(scope, var, mg, ms, mom, lr, rho, momentum, epsilon, grad, options);
+  }
+
+  /**
+   * Update '*var' according to the Ftrl-proximal scheme.
+   *  <p>
+   *  grad_with_shrinkage = grad + 2 * l2_shrinkage * var
+   *  accum_new = accum + grad * grad
+   *  linear += grad_with_shrinkage -
+   *      (accum_new^(-lr_power) - accum^(-lr_power)) / lr * var
+   *  quadratic = 1.0 / (accum_new^(lr_power) * lr) + 2 * l2
+   *  var = (sign(linear) * l1 - linear) / quadratic if |linear| > l1 else 0.0
+   *  accum = accum_new
+   *
+   * @param <T> data type for {@code out()} output
+   * @param var Should be from a Variable().
+   * @param accum Should be from a Variable().
+   * @param linear Should be from a Variable().
+   * @param grad The gradient.
+   * @param lr Scaling factor. Must be a scalar.
+   * @param l1 L1 regularization. Must be a scalar.
+   * @param l2 L2 shrinkage regularization. Must be a scalar.
+   * @param l2Shrinkage
+   * @param lrPower Scaling factor. Must be a scalar.
+   * @param options carries optional attributes values
+   * @return a new instance of ApplyFtrl
+   */
+  public <T extends TType> ApplyFtrl<T> applyFtrl(Operand<T> var, Operand<T> accum,
+      Operand<T> linear, Operand<T> grad, Operand<T> lr, Operand<T> l1, Operand<T> l2,
+      Operand<T> l2Shrinkage, Operand<T> lrPower, ApplyFtrl.Options... options) {
+    return ApplyFtrl.create(scope, var, accum, linear, grad, lr, l1, l2, l2Shrinkage, lrPower, options);
+  }
+
+  /**
+   * Update '*var' by subtracting 'alpha' * 'delta' from it.
+   *
+   * @param <T> data type for {@code out()} output
+   * @param var Should be from a Variable().
+   * @param alpha Scaling factor. Must be a scalar.
+   * @param delta The change.
+   * @param options carries optional attributes values
+   * @return a new instance of ApplyGradientDescent
+   */
+  public <T extends TType> ApplyGradientDescent<T> applyGradientDescent(Operand<T> var,
+      Operand<T> alpha, Operand<T> delta, ApplyGradientDescent.Options... options) {
+    return ApplyGradientDescent.create(scope, var, alpha, delta, options);
+  }
+
+  /**
+   * Update '*var' according to the momentum scheme.
+   *  <p>
+   *  Set use_nesterov = True if you want to use Nesterov momentum.
+   *  <p>
+   *  accum = accum * momentum + grad
+   *  var -= lr * accum
+   *
+   * @param <T> data type for {@code out()} output
+   * @param var Should be from a Variable().
+   * @param accum Should be from a Variable().
+   * @param lr Scaling factor. Must be a scalar.
+   * @param grad The gradient.
+   * @param momentum Momentum. Must be a scalar.
+   * @param options carries optional attributes values
+   * @return a new instance of ApplyMomentum
+   */
+  public <T extends TType> ApplyMomentum<T> applyMomentum(Operand<T> var, Operand<T> accum,
+      Operand<T> lr, Operand<T> grad, Operand<T> momentum, ApplyMomentum.Options... options) {
+    return ApplyMomentum.create(scope, var, accum, lr, grad, momentum, options);
+  }
+
+  /**
+   * Update '*var' according to the AddSign update.
+   *  <p>
+   *  m_t <- beta1 * m_{t-1} + (1 - beta1) * g
+   *  update <- exp(logbase * sign_decay * sign(g) * sign(m_t)) * g
+   *  variable <- variable - lr_t * update
+   *
+   * @param <T> data type for {@code out()} output
+   * @param var Should be from a Variable().
+   * @param m Should be from a Variable().
+   * @param lr Scaling factor. Must be a scalar.
+   * @param logbase Must be a scalar.
+   * @param signDecay Must be a scalar.
+   * @param beta Must be a scalar.
+   * @param grad The gradient.
+   * @param options carries optional attributes values
+   * @return a new instance of ApplyPowerSign
+   */
+  public <T extends TType> ApplyPowerSign<T> applyPowerSign(Operand<T> var, Operand<T> m,
+      Operand<T> lr, Operand<T> logbase, Operand<T> signDecay, Operand<T> beta, Operand<T> grad,
+      ApplyPowerSign.Options... options) {
+    return ApplyPowerSign.create(scope, var, m, lr, logbase, signDecay, beta, grad, options);
+  }
+
+  /**
+   * Update '*var' and '*accum' according to FOBOS with Adagrad learning rate.
+   *  <p>
+   *  accum += grad <i> grad
+   *  prox_v = var - lr </i> grad <i> (1 / sqrt(accum))
+   *  var = sign(prox_v)/(1+lr</i>l2) <i> max{|prox_v|-lr</i>l1,0}
+   *
+   * @param <T> data type for {@code out()} output
+   * @param var Should be from a Variable().
+   * @param accum Should be from a Variable().
+   * @param lr Scaling factor. Must be a scalar.
+   * @param l1 L1 regularization. Must be a scalar.
+   * @param l2 L2 regularization. Must be a scalar.
+   * @param grad The gradient.
+   * @param options carries optional attributes values
+   * @return a new instance of ApplyProximalAdagrad
+   */
+  public <T extends TType> ApplyProximalAdagrad<T> applyProximalAdagrad(Operand<T> var,
+      Operand<T> accum, Operand<T> lr, Operand<T> l1, Operand<T> l2, Operand<T> grad,
+      ApplyProximalAdagrad.Options... options) {
+    return ApplyProximalAdagrad.create(scope, var, accum, lr, l1, l2, grad, options);
+  }
+
+  /**
+   * Update '*var' as FOBOS algorithm with fixed learning rate.
+   *  <p>
+   *  prox_v = var - alpha <i> delta
+   *  var = sign(prox_v)/(1+alpha</i>l2) <i> max{|prox_v|-alpha</i>l1,0}
+   *
+   * @param <T> data type for {@code out()} output
+   * @param var Should be from a Variable().
+   * @param alpha Scaling factor. Must be a scalar.
+   * @param l1 L1 regularization. Must be a scalar.
+   * @param l2 L2 regularization. Must be a scalar.
+   * @param delta The change.
+   * @param options carries optional attributes values
+   * @return a new instance of ApplyProximalGradientDescent
+   */
+  public <T extends TType> ApplyProximalGradientDescent<T> applyProximalGradientDescent(
+      Operand<T> var, Operand<T> alpha, Operand<T> l1, Operand<T> l2, Operand<T> delta,
+      ApplyProximalGradientDescent.Options... options) {
+    return ApplyProximalGradientDescent.create(scope, var, alpha, l1, l2, delta, options);
+  }
+
+  /**
+   * Update '*var' according to the RMSProp algorithm.
+   *  <p>
+   *  Note that in dense implementation of this algorithm, ms and mom will
+   *  update even if the grad is zero, but in this sparse implementation, ms
+   *  and mom will not update in iterations during which the grad is zero.
+   *  <p>
+   *  mean_square = decay * mean_square + (1-decay) * gradient ** 2
+   *  Delta = learning_rate * gradient / sqrt(mean_square + epsilon)
+   *  <p>
+   *  ms <- rho * ms_{t-1} + (1-rho) * grad * grad
+   *  mom <- momentum * mom_{t-1} + lr * grad / sqrt(ms + epsilon)
+   *  var <- var - mom
+   *
+   * @param <T> data type for {@code out()} output
+   * @param var Should be from a Variable().
+   * @param ms Should be from a Variable().
+   * @param mom Should be from a Variable().
+   * @param lr Scaling factor. Must be a scalar.
+   * @param rho Decay rate. Must be a scalar.
+   * @param momentum
+   * @param epsilon Ridge term. Must be a scalar.
+   * @param grad The gradient.
+   * @param options carries optional attributes values
+   * @return a new instance of ApplyRmsProp
+   */
+  public <T extends TType> ApplyRmsProp<T> applyRmsProp(Operand<T> var, Operand<T> ms,
+      Operand<T> mom, Operand<T> lr, Operand<T> rho, Operand<T> momentum, Operand<T> epsilon,
+      Operand<T> grad, ApplyRmsProp.Options... options) {
+    return ApplyRmsProp.create(scope, var, ms, mom, lr, rho, momentum, epsilon, grad, options);
+  }
+
+  /**
+   * Multiplies slices of two tensors in batches.
+   *  <p>
+   *  Multiplies all slices of `Tensor` `x` and `y` (each slice can be
+   *  viewed as an element of a batch), and arranges the individual results
+   *  in a single output tensor of the same batch size. Each of the
+   *  individual slices can optionally be adjointed (to adjoint a matrix
+   *  means to transpose and conjugate it) before multiplication by setting
+   *  the `adj_x` or `adj_y` flag to `True`, which are by default `False`.
+   *  <p>
+   *  The input tensors `x` and `y` are 2-D or higher with shape `[..., r_x, c_x]`
+   *  and `[..., r_y, c_y]`.
+   *  <p>
+   *  The output tensor is 2-D or higher with shape `[..., r_o, c_o]`, where:
+   *  <p>
+   *      r_o = c_x if adj_x else r_x
+   *      c_o = r_y if adj_y else c_y
+   *  <p>
+   *  It is computed as:
+   *  <p>
+   *      output[..., :, :] = matrix(x[..., :, :]) * matrix(y[..., :, :])
+   *  <p>
+   *  <i>NOTE</i>: `train.BatchMatMul` supports broadcasting in the batch dimensions. More
+   *  about broadcasting
+   *  [here](http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html).
+   *
+   * @param <T> data type for {@code output()} output
+   * @param x 2-D or higher with shape `[..., r_x, c_x]`.
+   * @param y 2-D or higher with shape `[..., r_y, c_y]`.
+   * @param options carries optional attributes values
+   * @return a new instance of BatchMatMul
+   */
+  public <T extends TType> BatchMatMul<T> batchMatMul(Operand<T> x, Operand<T> y,
+      BatchMatMul.Options... options) {
+    return BatchMatMul.create(scope, x, y, options);
+  }
+
+  /**
+   * A conditional accumulator for aggregating gradients.
+   *  <p>
+   *  The accumulator accepts gradients marked with local_step greater or
+   *  equal to the most recent global_step known to the accumulator. The
+   *  average can be extracted from the accumulator, provided sufficient
+   *  gradients have been accumulated. Extracting the average automatically
+   *  resets the aggregate to 0, and increments the global_step recorded by
+   *  the accumulator.
+   *
+   * @param dtype The type of the value being accumulated.
+   * @param shape The shape of the values, can be [], in which case shape is unknown.
+   * @param options carries optional attributes values
+   * @return a new instance of ConditionalAccumulator
+   */
+  public <T extends TType> ConditionalAccumulator conditionalAccumulator(Class<T> dtype,
+      Shape shape, ConditionalAccumulator.Options... options) {
+    return ConditionalAccumulator.create(scope, dtype, shape, options);
+  }
+
+  /**
+   * Given a path to new and old vocabulary files, returns a remapping Tensor of
+   *  <p>
+   *  length `num_new_vocab`, where `remapping[i]` contains the row number in the old
+   *  vocabulary that corresponds to row `i` in the new vocabulary (starting at line
+   *  `new_vocab_offset` and up to `num_new_vocab` entities), or `-1` if entry `i`
+   *  in the new vocabulary is not in the old vocabulary.  The old vocabulary is
+   *  constrained to the first `old_vocab_size` entries if `old_vocab_size` is not the
+   *  default value of -1.
+   *  <p>
+   *  `num_vocab_offset` enables
+   *  use in the partitioned variable case, and should generally be set through
+   *  examining partitioning info.  The format of the files should be a text file,
+   *  with each line containing a single entity within the vocabulary.
+   *  <p>
+   *  For example, with `new_vocab_file` a text file containing each of the following
+   *  elements on a single line: `[f0, f1, f2, f3]`, old_vocab_file = [f1, f0, f3],
+   *  `num_new_vocab = 3, new_vocab_offset = 1`, the returned remapping would be
+   *  `[0, -1, 2]`.
+   *  <p>
+   *  The op also returns a count of how many entries in the new vocabulary
+   *  were present in the old vocabulary, which is used to calculate the number of
+   *  values to initialize in a weight matrix remapping
+   *  <p>
+   *  This functionality can be used to remap both row vocabularies (typically,
+   *  features) and column vocabularies (typically, classes) from TensorFlow
+   *  checkpoints.  Note that the partitioning logic relies on contiguous vocabularies
+   *  corresponding to div-partitioned variables.  Moreover, the underlying remapping
+   *  uses an IndexTable (as opposed to an inexact CuckooTable), so client code should
+   *  use the corresponding index_table_from_file() as the FeatureColumn framework
+   *  does (as opposed to tf.feature_to_id(), which uses a CuckooTable).
+   *
+   * @param newVocabFile Path to the new vocab file.
+   * @param oldVocabFile Path to the old vocab file.
+   * @param newVocabOffset How many entries into the new vocab file to start reading.
+   * @param numNewVocab Number of entries in the new vocab file to remap.
+   * @param options carries optional attributes values
+   * @return a new instance of GenerateVocabRemapping
+   */
+  public GenerateVocabRemapping generateVocabRemapping(Operand<TString> newVocabFile,
+      Operand<TString> oldVocabFile, Long newVocabOffset, Long numNewVocab,
+      GenerateVocabRemapping.Options... options) {
+    return GenerateVocabRemapping.create(scope, newVocabFile, oldVocabFile, newVocabOffset, numNewVocab, options);
+  }
+
+  /**
+   * V2 format specific: merges the metadata files of sharded checkpoints.  The
+   *  <p>
+   *  result is one logical checkpoint, with one physical metadata file and renamed
+   *  data files.
+   *  <p>
+   *  Intended for "grouping" multiple checkpoints in a sharded checkpoint setup.
+   *  <p>
+   *  If delete_old_dirs is true, attempts to delete recursively the dirname of each
+   *  path in the input checkpoint_prefixes.  This is useful when those paths are non
+   *  user-facing temporary locations.
+   *
+   * @param checkpointPrefixes prefixes of V2 checkpoints to merge.
+   * @param destinationPrefix scalar.  The desired final prefix.  Allowed to be the same
+   *  as one of the checkpoint_prefixes.
+   * @param options carries optional attributes values
+   * @return a new instance of MergeV2Checkpoints
+   */
+  public MergeV2Checkpoints mergeV2Checkpoints(Operand<TString> checkpointPrefixes,
+      Operand<TString> destinationPrefix, MergeV2Checkpoints.Options... options) {
+    return MergeV2Checkpoints.create(scope, checkpointPrefixes, destinationPrefix, options);
+  }
+
+  /**
+   * Training via negative sampling.
+   *
+   * @param wIn input word embedding.
+   * @param wOut output word embedding.
+   * @param examples A vector of word ids.
+   * @param labels A vector of word ids.
+   * @param lr
+   * @param vocabCount Count of words in the vocabulary.
+   * @param numNegativeSamples Number of negative samples per example.
+   * @return a new instance of NegTrain
+   */
+  public NegTrain negTrain(Operand<TFloat32> wIn, Operand<TFloat32> wOut, Operand<TInt32> examples,
+      Operand<TInt32> labels, Operand<TFloat32> lr, List<Long> vocabCount,
+      Long numNegativeSamples) {
+    return NegTrain.create(scope, wIn, wOut, examples, labels, lr, vocabCount, numNegativeSamples);
+  }
+
+  /**
+   * An identity op that triggers an error if a gradient is requested.
+   *  <p>
+   *  When executed in a graph, this op outputs its input tensor as-is.
+   *  <p>
+   *  When building ops to compute gradients, the TensorFlow gradient system
+   *  will return an error when trying to lookup the gradient of this op,
+   *  because no gradient must ever be registered for this function.  This
+   *  op exists to prevent subtle bugs from silently returning unimplemented
+   *  gradients in some corner cases.
+   *
+   * @param <T> data type for {@code output()} output
+   * @param input any tensor.
+   * @param options carries optional attributes values
+   * @return a new instance of PreventGradient
+   */
+  public <T extends TType> PreventGradient<T> preventGradient(Operand<T> input,
+      PreventGradient.Options... options) {
+    return PreventGradient.create(scope, input, options);
+  }
+
+  /**
+   * Update '*var' according to the adadelta scheme.
+   *  <p>
+   *  accum = rho() * accum + (1 - rho()) * grad.square();
+   *  update = (update_accum + epsilon).sqrt() * (accum + epsilon()).rsqrt() * grad;
+   *  update_accum = rho() * update_accum + (1 - rho()) * update.square();
+   *  var -= update;
+   *
+   * @param var Should be from a Variable().
+   * @param accum Should be from a Variable().
+   * @param accumUpdate Should be from a Variable().
+   * @param lr Scaling factor. Must be a scalar.
+   * @param rho Decay factor. Must be a scalar.
+   * @param epsilon Constant factor. Must be a scalar.
+   * @param grad The gradient.
+   * @param options carries optional attributes values
+   * @return a new instance of ResourceApplyAdadelta
+   */
+  public <T extends TType> ResourceApplyAdadelta resourceApplyAdadelta(Operand<?> var,
+      Operand<?> accum, Operand<?> accumUpdate, Operand<T> lr, Operand<T> rho, Operand<T> epsilon,
+      Operand<T> grad, ResourceApplyAdadelta.Options... options) {
+    return ResourceApplyAdadelta.create(scope, var, accum, accumUpdate, lr, rho, epsilon, grad, options);
+  }
+
+  /**
+   * Update '*var' according to the proximal adagrad scheme.
+   *
+   * @param var Should be from a Variable().
+   * @param gradientAccumulator Should be from a Variable().
+   * @param gradientSquaredAccumulator Should be from a Variable().
+   * @param grad The gradient.
+   * @param lr Scaling factor. Must be a scalar.
+   * @param l1 L1 regularization. Must be a scalar.
+   * @param l2 L2 regularization. Must be a scalar.
+   * @param globalStep Training step number. Must be a scalar.
+   * @param options carries optional attributes values
+   * @return a new instance of ResourceApplyAdagradDa
+   */
+  public <T extends TType> ResourceApplyAdagradDa resourceApplyAdagradDa(Operand<?> var,
+      Operand<?> gradientAccumulator, Operand<?> gradientSquaredAccumulator, Operand<T> grad,
+      Operand<T> lr, Operand<T> l1, Operand<T> l2, Operand<TInt64> globalStep,
+      ResourceApplyAdagradDa.Options... options) {
+    return ResourceApplyAdagradDa.create(scope, var, gradientAccumulator, gradientSquaredAccumulator, grad, lr, l1, l2, globalStep, options);
+  }
+
+  /**
+   * Update '*var' according to the Adam algorithm.
+   *  <p>
+   *  $$\text{lr}_t := \mathrm{learning_rate} * \sqrt{1 - \beta_2^t} / (1 - \beta_1^t)$$
+   *  $$m_t := \beta_1 * m_{t-1} + (1 - \beta_1) * g$$
+   *  $$v_t := \beta_2 * v_{t-1} + (1 - \beta_2) * g * g$$
+   *  $$\text{variable} := \text{variable} - \text{lr}_t * m_t / (\sqrt{v_t} + \epsilon)$$
+   *
+   * @param var Should be from a Variable().
+   * @param m Should be from a Variable().
+   * @param v Should be from a Variable().
+   * @param beta1Power Must be a scalar.
+   * @param beta2Power Must be a scalar.
+   * @param lr Scaling factor. Must be a scalar.
+   * @param beta1 Momentum factor. Must be a scalar.
+   * @param beta2 Momentum factor. Must be a scalar.
+   * @param epsilon Ridge term. Must be a scalar.
+   * @param grad The gradient.
+   * @param options carries optional attributes values
+   * @return a new instance of ResourceApplyAdam
+   */
+  public <T extends TType> ResourceApplyAdam resourceApplyAdam(Operand<?> var, Operand<?> m,
+      Operand<?> v, Operand<T> beta1Power, Operand<T> beta2Power, Operand<T> lr, Operand<T> beta1,
+      Operand<T> beta2, Operand<T> epsilon, Operand<T> grad, ResourceApplyAdam.Options... options) {
+    return ResourceApplyAdam.create(scope, var, m, v, beta1Power, beta2Power, lr, beta1, beta2, epsilon, grad, options);
+  }
+
+  /**
+   * Update '*var' according to the Adam algorithm.
+   *  <p>
+   *  $$\text{lr}_t := \mathrm{learning_rate} * \sqrt{1 - \beta_2^t} / (1 - \beta_1^t)$$
+   *  $$m_t := \beta_1 * m_{t-1} + (1 - \beta_1) * g$$
+   *  $$v_t := \beta_2 * v_{t-1} + (1 - \beta_2) * g * g$$
+   *  $$\hat{v}_t := max{\hat{v}_{t-1}, v_t}$$
+   *  $$\text{variable} := \text{variable} - \text{lr}_t * m_t / (\sqrt{\hat{v}_t} + \epsilon)$$
+   *
+   * @param var Should be from a Variable().
+   * @param m Should be from a Variable().
+   * @param v Should be from a Variable().
+   * @param vhat Should be from a Variable().
+   * @param beta1Power Must be a scalar.
+   * @param beta2Power Must be a scalar.
+   * @param lr Scaling factor. Must be a scalar.
+   * @param beta1 Momentum factor. Must be a scalar.
+   * @param beta2 Momentum factor. Must be a scalar.
+   * @param epsilon Ridge term. Must be a scalar.
+   * @param grad The gradient.
+   * @param options carries optional attributes values
+   * @return a new instance of ResourceApplyAdamWithAmsgrad
+   */
+  public <T extends TType> ResourceApplyAdamWithAmsgrad resourceApplyAdamWithAmsgrad(Operand<?> var,
+      Operand<?> m, Operand<?> v, Operand<?> vhat, Operand<T> beta1Power, Operand<T> beta2Power,
+      Operand<T> lr, Operand<T> beta1, Operand<T> beta2, Operand<T> epsilon, Operand<T> grad,
+      ResourceApplyAdamWithAmsgrad.Options... options) {
+    return ResourceApplyAdamWithAmsgrad.create(scope, var, m, v, vhat, beta1Power, beta2Power, lr, beta1, beta2, epsilon, grad, options);
+  }
+
+  /**
+   * Update '*var' according to the AddSign update.
+   *  <p>
+   *  m_t <- beta1 * m_{t-1} + (1 - beta1) * g
+   *  update <- (alpha + sign_decay * sign(g) *sign(m)) * g
+   *  variable <- variable - lr_t * update
+   *
+   * @param var Should be from a Variable().
+   * @param m Should be from a Variable().
+   * @param lr Scaling factor. Must be a scalar.
+   * @param alpha Must be a scalar.
+   * @param signDecay Must be a scalar.
+   * @param beta Must be a scalar.
+   * @param grad The gradient.
+   * @param options carries optional attributes values
+   * @return a new instance of ResourceApplyAddSign
+   */
+  public <T extends TType> ResourceApplyAddSign resourceApplyAddSign(Operand<?> var, Operand<?> m,
+      Operand<T> lr, Operand<T> alpha, Operand<T> signDecay, Operand<T> beta, Operand<T> grad,
+      ResourceApplyAddSign.Options... options) {
+    return ResourceApplyAddSign.create(scope, var, m, lr, alpha, signDecay, beta, grad, options);
+  }
+
+  /**
+   * Update '*var' according to the centered RMSProp algorithm.
+   *  <p>
+   *  The centered RMSProp algorithm uses an estimate of the centered second moment
+   *  (i.e., the variance) for normalization, as opposed to regular RMSProp, which
+   *  uses the (uncentered) second moment. This often helps with training, but is
+   *  slightly more expensive in terms of computation and memory.
+   *  <p>
+   *  Note that in dense implementation of this algorithm, mg, ms, and mom will
+   *  update even if the grad is zero, but in this sparse implementation, mg, ms,
+   *  and mom will not update in iterations during which the grad is zero.
+   *  <p>
+   *  mean_square = decay * mean_square + (1-decay) * gradient ** 2
+   *  mean_grad = decay * mean_grad + (1-decay) * gradient
+   *  <p>
+   *  Delta = learning_rate * gradient / sqrt(mean_square + epsilon - mean_grad ** 2)
+   *  <p>
+   *  mg <- rho * mg_{t-1} + (1-rho) * grad
+   *  ms <- rho * ms_{t-1} + (1-rho) * grad * grad
+   *  mom <- momentum * mom_{t-1} + lr * grad / sqrt(ms - mg * mg + epsilon)
+   *  var <- var - mom
+   *
+   * @param var Should be from a Variable().
+   * @param mg Should be from a Variable().
+   * @param ms Should be from a Variable().
+   * @param mom Should be from a Variable().
+   * @param lr Scaling factor. Must be a scalar.
+   * @param rho Decay rate. Must be a scalar.
+   * @param momentum
+   * @param epsilon Ridge term. Must be a scalar.
+   * @param grad The gradient.
+   * @param options carries optional attributes values
+   * @return a new instance of ResourceApplyCenteredRmsProp
+   */
+  public <T extends TType> ResourceApplyCenteredRmsProp resourceApplyCenteredRmsProp(Operand<?> var,
+      Operand<?> mg, Operand<?> ms, Operand<?> mom, Operand<T> lr, Operand<T> rho,
+      Operand<T> momentum, Operand<T> epsilon, Operand<T> grad,
+      ResourceApplyCenteredRmsProp.Options... options) {
+    return ResourceApplyCenteredRmsProp.create(scope, var, mg, ms, mom, lr, rho, momentum, epsilon, grad, options);
+  }
+
+  /**
+   * Update '*var' according to the Ftrl-proximal scheme.
+   *  <p>
+   *  grad_with_shrinkage = grad + 2 * l2_shrinkage * var
+   *  accum_new = accum + grad_with_shrinkage * grad_with_shrinkage
+   *  linear += grad_with_shrinkage +
+   *      (accum_new^(-lr_power) - accum^(-lr_power)) / lr * var
+   *  quadratic = 1.0 / (accum_new^(lr_power) * lr) + 2 * l2
+   *  var = (sign(linear) * l1 - linear) / quadratic if |linear| > l1 else 0.0
+   *  accum = accum_new
+   *
+   * @param var Should be from a Variable().
+   * @param accum Should be from a Variable().
+   * @param linear Should be from a Variable().
+   * @param grad The gradient.
+   * @param lr Scaling factor. Must be a scalar.
+   * @param l1 L1 regularization. Must be a scalar.
+   * @param l2 L2 shrinkage regularization. Must be a scalar.
+   * @param l2Shrinkage
+   * @param lrPower Scaling factor. Must be a scalar.
+   * @param options carries optional attributes values
+   * @return a new instance of ResourceApplyFtrl
+   */
+  public <T extends TType> ResourceApplyFtrl resourceApplyFtrl(Operand<?> var, Operand<?> accum,
+      Operand<?> linear, Operand<T> grad, Operand<T> lr, Operand<T> l1, Operand<T> l2,
+      Operand<T> l2Shrinkage, Operand<T> lrPower, ResourceApplyFtrl.Options... options) {
+    return ResourceApplyFtrl.create(scope, var, accum, linear, grad, lr, l1, l2, l2Shrinkage, lrPower, options);
+  }
+
+  /**
+   * Update '*var' by subtracting 'alpha' * 'delta' from it.
+   *
+   * @param var Should be from a Variable().
+   * @param alpha Scaling factor. Must be a scalar.
+   * @param delta The change.
+   * @param options carries optional attributes values
+   * @return a new instance of ResourceApplyGradientDescent
+   */
+  public <T extends TType> ResourceApplyGradientDescent resourceApplyGradientDescent(Operand<?> var,
+      Operand<T> alpha, Operand<T> delta, ResourceApplyGradientDescent.Options... options) {
+    return ResourceApplyGradientDescent.create(scope, var, alpha, delta, options);
+  }
+
+  /**
+   * Update '*var' according to the momentum scheme.
+   *  <p>
+   *  Set use_nesterov = True if you want to use Nesterov momentum.
+   *  <p>
+   *  accum = accum * momentum - lr * grad
+   *  var += accum
+   *
+   * @param var Should be from a Variable().
+   * @param accum Should be from a Variable().
+   * @param lr Scaling factor. Must be a scalar.
+   * @param grad The gradient.
+   * @param momentum Momentum. Must be a scalar.
+   * @param options carries optional attributes values
+   * @return a new instance of ResourceApplyKerasMomentum
+   */
+  public <T extends TType> ResourceApplyKerasMomentum resourceApplyKerasMomentum(Operand<?> var,
+      Operand<?> accum, Operand<T> lr, Operand<T> grad, Operand<T> momentum,
+      ResourceApplyKerasMomentum.Options... options) {
+    return ResourceApplyKerasMomentum.create(scope, var, accum, lr, grad, momentum, options);
+  }
+
+  /**
+   * Update '*var' according to the momentum scheme.
+   *  <p>
+   *  Set use_nesterov = True if you want to use Nesterov momentum.
+   *  <p>
+   *  accum = accum * momentum + grad
+   *  var -= lr * accum
+   *
+   * @param var Should be from a Variable().
+   * @param accum Should be from a Variable().
+   * @param lr Scaling factor. Must be a scalar.
+   * @param grad The gradient.
+   * @param momentum Momentum. Must be a scalar.
+   * @param options carries optional attributes values
+   * @return a new instance of ResourceApplyMomentum
+   */
+  public <T extends TType> ResourceApplyMomentum resourceApplyMomentum(Operand<?> var,
+      Operand<?> accum, Operand<T> lr, Operand<T> grad, Operand<T> momentum,
+      ResourceApplyMomentum.Options... options) {
+    return ResourceApplyMomentum.create(scope, var, accum, lr, grad, momentum, options);
+  }
+
+  /**
+   * Update '*var' according to the AddSign update.
+   *  <p>
+   *  m_t <- beta1 * m_{t-1} + (1 - beta1) * g
+   *  update <- exp(logbase * sign_decay * sign(g) * sign(m_t)) * g
+   *  variable <- variable - lr_t * update
+   *
+   * @param var Should be from a Variable().
+   * @param m Should be from a Variable().
+   * @param lr Scaling factor. Must be a scalar.
+   * @param logbase Must be a scalar.
+   * @param signDecay Must be a scalar.
+   * @param beta Must be a scalar.
+   * @param grad The gradient.
+   * @param options carries optional attributes values
+   * @return a new instance of ResourceApplyPowerSign
+   */
+  public <T extends TType> ResourceApplyPowerSign resourceApplyPowerSign(Operand<?> var,
+      Operand<?> m, Operand<T> lr, Operand<T> logbase, Operand<T> signDecay, Operand<T> beta,
+      Operand<T> grad, ResourceApplyPowerSign.Options... options) {
+    return ResourceApplyPowerSign.create(scope, var, m, lr, logbase, signDecay, beta, grad, options);
+  }
+
+  /**
+   * Update '*var' and '*accum' according to FOBOS with Adagrad learning rate.
+   *  <p>
+   *  accum += grad <i> grad
+   *  prox_v = var - lr </i> grad <i> (1 / sqrt(accum))
+   *  var = sign(prox_v)/(1+lr</i>l2) <i> max{|prox_v|-lr</i>l1,0}
+   *
+   * @param var Should be from a Variable().
+   * @param accum Should be from a Variable().
+   * @param lr Scaling factor. Must be a scalar.
+   * @param l1 L1 regularization. Must be a scalar.
+   * @param l2 L2 regularization. Must be a scalar.
+   * @param grad The gradient.
+   * @param options carries optional attributes values
+   * @return a new instance of ResourceApplyProximalAdagrad
+   */
+  public <T extends TType> ResourceApplyProximalAdagrad resourceApplyProximalAdagrad(Operand<?> var,
+      Operand<?> accum, Operand<T> lr, Operand<T> l1, Operand<T> l2, Operand<T> grad,
+      ResourceApplyProximalAdagrad.Options... options) {
+    return ResourceApplyProximalAdagrad.create(scope, var, accum, lr, l1, l2, grad, options);
+  }
+
+  /**
+   * Update '*var' as FOBOS algorithm with fixed learning rate.
+   *  <p>
+   *  prox_v = var - alpha <i> delta
+   *  var = sign(prox_v)/(1+alpha</i>l2) <i> max{|prox_v|-alpha</i>l1,0}
+   *
+   * @param var Should be from a Variable().
+   * @param alpha Scaling factor. Must be a scalar.
+   * @param l1 L1 regularization. Must be a scalar.
+   * @param l2 L2 regularization. Must be a scalar.
+   * @param delta The change.
+   * @param options carries optional attributes values
+   * @return a new instance of ResourceApplyProximalGradientDescent
+   */
+  public <T extends TType> ResourceApplyProximalGradientDescent resourceApplyProximalGradientDescent(
+      Operand<?> var, Operand<T> alpha, Operand<T> l1, Operand<T> l2, Operand<T> delta,
+      ResourceApplyProximalGradientDescent.Options... options) {
+    return ResourceApplyProximalGradientDescent.create(scope, var, alpha, l1, l2, delta, options);
+  }
+
+  /**
+   * Update '*var' according to the RMSProp algorithm.
+   *  <p>
+   *  Note that in dense implementation of this algorithm, ms and mom will
+   *  update even if the grad is zero, but in this sparse implementation, ms
+   *  and mom will not update in iterations during which the grad is zero.
+   *  <p>
+   *  mean_square = decay * mean_square + (1-decay) * gradient ** 2
+   *  Delta = learning_rate * gradient / sqrt(mean_square + epsilon)
+   *  <p>
+   *  ms <- rho * ms_{t-1} + (1-rho) * grad * grad
+   *  mom <- momentum * mom_{t-1} + lr * grad / sqrt(ms + epsilon)
+   *  var <- var - mom
+   *
+   * @param var Should be from a Variable().
+   * @param ms Should be from a Variable().
+   * @param mom Should be from a Variable().
+   * @param lr Scaling factor. Must be a scalar.
+   * @param rho Decay rate. Must be a scalar.
+   * @param momentum
+   * @param epsilon Ridge term. Must be a scalar.
+   * @param grad The gradient.
+   * @param options carries optional attributes values
+   * @return a new instance of ResourceApplyRmsProp
+   */
+  public <T extends TType> ResourceApplyRmsProp resourceApplyRmsProp(Operand<?> var, Operand<?> ms,
+      Operand<?> mom, Operand<T> lr, Operand<T> rho, Operand<T> momentum, Operand<T> epsilon,
+      Operand<T> grad, ResourceApplyRmsProp.Options... options) {
+    return ResourceApplyRmsProp.create(scope, var, ms, mom, lr, rho, momentum, epsilon, grad, options);
+  }
+
+  /**
+   * var: Should be from a Variable().
+   *
+   * @param var
+   * @param accum Should be from a Variable().
+   * @param accumUpdate : Should be from a Variable().
+   * @param lr Learning rate. Must be a scalar.
+   * @param rho Decay factor. Must be a scalar.
+   * @param epsilon Constant factor. Must be a scalar.
+   * @param grad The gradient.
+   * @param indices A vector of indices into the first dimension of var and accum.
+   * @param options carries optional attributes values
+   * @return a new instance of ResourceSparseApplyAdadelta
+   */
+  public <T extends TType, U extends TNumber> ResourceSparseApplyAdadelta resourceSparseApplyAdadelta(
+      Operand<?> var, Operand<?> accum, Operand<?> accumUpdate, Operand<T> lr, Operand<T> rho,
+      Operand<T> epsilon, Operand<T> grad, Operand<U> indices,
+      ResourceSparseApplyAdadelta.Options... options) {
+    return ResourceSparseApplyAdadelta.create(scope, var, accum, accumUpdate, lr, rho, epsilon, grad, indices, options);
+  }
+
+  /**
+   * Update relevant entries in '*var' and '*accum' according to the adagrad scheme.
+   *  <p>
+   *  That is for rows we have grad for, we update var and accum as follows:
+   *  accum += grad * grad
+   *  var -= lr * grad * (1 / sqrt(accum))
+   *
+   * @param var Should be from a Variable().
+   * @param accum Should be from a Variable().
+   * @param lr Learning rate. Must be a scalar.
+   * @param grad The gradient.
+   * @param indices A vector of indices into the first dimension of var and accum.
+   * @param options carries optional attributes values
+   * @return a new instance of ResourceSparseApplyAdagrad
+   */
+  public <T extends TType, U extends TNumber> ResourceSparseApplyAdagrad resourceSparseApplyAdagrad(
+      Operand<?> var, Operand<?> accum, Operand<T> lr, Operand<T> grad, Operand<U> indices,
+      ResourceSparseApplyAdagrad.Options... options) {
+    return ResourceSparseApplyAdagrad.create(scope, var, accum, lr, grad, indices, options);
+  }
+
+  /**
+   * Update entries in '*var' and '*accum' according to the proximal adagrad scheme.
+   *
+   * @param var Should be from a Variable().
+   * @param gradientAccumulator Should be from a Variable().
+   * @param gradientSquaredAccumulator Should be from a Variable().
+   * @param grad The gradient.
+   * @param indices A vector of indices into the first dimension of var and accum.
+   * @param lr Learning rate. Must be a scalar.
+   * @param l1 L1 regularization. Must be a scalar.
+   * @param l2 L2 regularization. Must be a scalar.
+   * @param globalStep Training step number. Must be a scalar.
+   * @param options carries optional attributes values
+   * @return a new instance of ResourceSparseApplyAdagradDa
+   */
+  public <T extends TType, U extends TNumber> ResourceSparseApplyAdagradDa resourceSparseApplyAdagradDa(
+      Operand<?> var, Operand<?> gradientAccumulator, Operand<?> gradientSquaredAccumulator,
+      Operand<T> grad, Operand<U> indices, Operand<T> lr, Operand<T> l1, Operand<T> l2,
+      Operand<TInt64> globalStep, ResourceSparseApplyAdagradDa.Options... options) {
+    return ResourceSparseApplyAdagradDa.create(scope, var, gradientAccumulator, gradientSquaredAccumulator, grad, indices, lr, l1, l2, globalStep, options);
+  }
+
+  /**
+   * Update '*var' according to the centered RMSProp algorithm.
+   *  <p>
+   *  The centered RMSProp algorithm uses an estimate of the centered second moment
+   *  (i.e., the variance) for normalization, as opposed to regular RMSProp, which
+   *  uses the (uncentered) second moment. This often helps with training, but is
+   *  slightly more expensive in terms of computation and memory.
+   *  <p>
+   *  Note that in dense implementation of this algorithm, mg, ms, and mom will
+   *  update even if the grad is zero, but in this sparse implementation, mg, ms,
+   *  and mom will not update in iterations during which the grad is zero.
+   *  <p>
+   *  mean_square = decay * mean_square + (1-decay) * gradient ** 2
+   *  mean_grad = decay * mean_grad + (1-decay) * gradient
+   *  Delta = learning_rate * gradient / sqrt(mean_square + epsilon - mean_grad ** 2)
+   *  <p>
+   *  ms <- rho * ms_{t-1} + (1-rho) * grad * grad
+   *  mom <- momentum * mom_{t-1} + lr * grad / sqrt(ms + epsilon)
+   *  var <- var - mom
+   *
+   * @param var Should be from a Variable().
+   * @param mg Should be from a Variable().
+   * @param ms Should be from a Variable().
+   * @param mom Should be from a Variable().
+   * @param lr Scaling factor. Must be a scalar.
+   * @param rho Decay rate. Must be a scalar.
+   * @param momentum
+   * @param epsilon Ridge term. Must be a scalar.
+   * @param grad The gradient.
+   * @param indices A vector of indices into the first dimension of var, ms and mom.
+   * @param options carries optional attributes values
+   * @return a new instance of ResourceSparseApplyCenteredRmsProp
+   */
+  public <T extends TType, U extends TNumber> ResourceSparseApplyCenteredRmsProp resourceSparseApplyCenteredRmsProp(
+      Operand<?> var, Operand<?> mg, Operand<?> ms, Operand<?> mom, Operand<T> lr, Operand<T> rho,
+      Operand<T> momentum, Operand<T> epsilon, Operand<T> grad, Operand<U> indices,
+      ResourceSparseApplyCenteredRmsProp.Options... options) {
+    return ResourceSparseApplyCenteredRmsProp.create(scope, var, mg, ms, mom, lr, rho, momentum, epsilon, grad, indices, options);
+  }
+
+  /**
+   * Update relevant entries in '*var' according to the Ftrl-proximal scheme.
+   *  <p>
+   *  That is for rows we have grad for, we update var, accum and linear as follows:
+   *  grad_with_shrinkage = grad + 2 * l2_shrinkage * var
+   *  accum_new = accum + grad_with_shrinkage * grad_with_shrinkage
+   *  linear += grad_with_shrinkage +
+   *      (accum_new^(-lr_power) - accum^(-lr_power)) / lr * var
+   *  quadratic = 1.0 / (accum_new^(lr_power) * lr) + 2 * l2
+   *  var = (sign(linear) * l1 - linear) / quadratic if |linear| > l1 else 0.0
+   *  accum = accum_new
+   *
+   * @param var Should be from a Variable().
+   * @param accum Should be from a Variable().
+   * @param linear Should be from a Variable().
+   * @param grad The gradient.
+   * @param indices A vector of indices into the first dimension of var and accum.
+   * @param lr Scaling factor. Must be a scalar.
+   * @param l1 L1 regularization. Must be a scalar.
+   * @param l2 L2 shrinkage regularization. Must be a scalar.
+   * @param l2Shrinkage
+   * @param lrPower Scaling factor. Must be a scalar.
+   * @param options carries optional attributes values
+   * @return a new instance of ResourceSparseApplyFtrl
+   */
+  public <T extends TType, U extends TNumber> ResourceSparseApplyFtrl resourceSparseApplyFtrl(
+      Operand<?> var, Operand<?> accum, Operand<?> linear, Operand<T> grad, Operand<U> indices,
+      Operand<T> lr, Operand<T> l1, Operand<T> l2, Operand<T> l2Shrinkage, Operand<T> lrPower,
+      ResourceSparseApplyFtrl.Options... options) {
+    return ResourceSparseApplyFtrl.create(scope, var, accum, linear, grad, indices, lr, l1, l2, l2Shrinkage, lrPower, options);
+  }
+
+  /**
+   * Update relevant entries in '*var' and '*accum' according to the momentum scheme.
+   *  <p>
+   *  Set use_nesterov = True if you want to use Nesterov momentum.
+   *  <p>
+   *  That is for rows we have grad for, we update var and accum as follows:
+   *  <p>
+   *  accum = accum * momentum - lr * grad
+   *  var += accum
+   *
+   * @param var Should be from a Variable().
+   * @param accum Should be from a Variable().
+   * @param lr Learning rate. Must be a scalar.
+   * @param grad The gradient.
+   * @param indices A vector of indices into the first dimension of var and accum.
+   * @param momentum Momentum. Must be a scalar.
+   * @param options carries optional attributes values
+   * @return a new instance of ResourceSparseApplyKerasMomentum
+   */
+  public <T extends TType, U extends TNumber> ResourceSparseApplyKerasMomentum resourceSparseApplyKerasMomentum(
+      Operand<?> var, Operand<?> accum, Operand<T> lr, Operand<T> grad, Operand<U> indices,
+      Operand<T> momentum, ResourceSparseApplyKerasMomentum.Options... options) {
+    return ResourceSparseApplyKerasMomentum.create(scope, var, accum, lr, grad, indices, momentum, options);
+  }
+
+  /**
+   * Update relevant entries in '*var' and '*accum' according to the momentum scheme.
+   *  <p>
+   *  Set use_nesterov = True if you want to use Nesterov momentum.
+   *  <p>
+   *  That is for rows we have grad for, we update var and accum as follows:
+   *  <p>
+   *  accum = accum * momentum + grad
+   *  var -= lr * accum
+   *
+   * @param var Should be from a Variable().
+   * @param accum Should be from a Variable().
+   * @param lr Learning rate. Must be a scalar.
+   * @param grad The gradient.
+   * @param indices A vector of indices into the first dimension of var and accum.
+   * @param momentum Momentum. Must be a scalar.
+   * @param options carries optional attributes values
+   * @return a new instance of ResourceSparseApplyMomentum
+   */
+  public <T extends TType, U extends TNumber> ResourceSparseApplyMomentum resourceSparseApplyMomentum(
+      Operand<?> var, Operand<?> accum, Operand<T> lr, Operand<T> grad, Operand<U> indices,
+      Operand<T> momentum, ResourceSparseApplyMomentum.Options... options) {
+    return ResourceSparseApplyMomentum.create(scope, var, accum, lr, grad, indices, momentum, options);
+  }
+
+  /**
+   * Sparse update entries in '*var' and '*accum' according to FOBOS algorithm.
+   *  <p>
+   *  That is for rows we have grad for, we update var and accum as follows:
+   *  accum += grad <i> grad
+   *  prox_v = var
+   *  prox_v -= lr </i> grad <i> (1 / sqrt(accum))
+   *  var = sign(prox_v)/(1+lr</i>l2) <i> max{|prox_v|-lr</i>l1,0}
+   *
+   * @param var Should be from a Variable().
+   * @param accum Should be from a Variable().
+   * @param lr Learning rate. Must be a scalar.
+   * @param l1 L1 regularization. Must be a scalar.
+   * @param l2 L2 regularization. Must be a scalar.
+   * @param grad The gradient.
+   * @param indices A vector of indices into the first dimension of var and accum.
+   * @param options carries optional attributes values
+   * @return a new instance of ResourceSparseApplyProximalAdagrad
+   */
+  public <T extends TType, U extends TNumber> ResourceSparseApplyProximalAdagrad resourceSparseApplyProximalAdagrad(
+      Operand<?> var, Operand<?> accum, Operand<T> lr, Operand<T> l1, Operand<T> l2,
+      Operand<T> grad, Operand<U> indices, ResourceSparseApplyProximalAdagrad.Options... options) {
+    return ResourceSparseApplyProximalAdagrad.create(scope, var, accum, lr, l1, l2, grad, indices, options);
+  }
+
+  /**
+   * Sparse update '*var' as FOBOS algorithm with fixed learning rate.
+   *  <p>
+   *  That is for rows we have grad for, we update var as follows:
+   *  prox_v = var - alpha <i> grad
+   *  var = sign(prox_v)/(1+alpha</i>l2) <i> max{|prox_v|-alpha</i>l1,0}
+   *
+   * @param var Should be from a Variable().
+   * @param alpha Scaling factor. Must be a scalar.
+   * @param l1 L1 regularization. Must be a scalar.
+   * @param l2 L2 regularization. Must be a scalar.
+   * @param grad The gradient.
+   * @param indices A vector of indices into the first dimension of var and accum.
+   * @param options carries optional attributes values
+   * @return a new instance of ResourceSparseApplyProximalGradientDescent
+   */
+  public <T extends TType, U extends TNumber> ResourceSparseApplyProximalGradientDescent resourceSparseApplyProximalGradientDescent(
+      Operand<?> var, Operand<T> alpha, Operand<T> l1, Operand<T> l2, Operand<T> grad,
+      Operand<U> indices, ResourceSparseApplyProximalGradientDescent.Options... options) {
+    return ResourceSparseApplyProximalGradientDescent.create(scope, var, alpha, l1, l2, grad, indices, options);
+  }
+
+  /**
+   * Update '*var' according to the RMSProp algorithm.
+   *  <p>
+   *  Note that in dense implementation of this algorithm, ms and mom will
+   *  update even if the grad is zero, but in this sparse implementation, ms
+   *  and mom will not update in iterations during which the grad is zero.
+   *  <p>
+   *  mean_square = decay * mean_square + (1-decay) * gradient ** 2
+   *  Delta = learning_rate * gradient / sqrt(mean_square + epsilon)
+   *  <p>
+   *  ms <- rho * ms_{t-1} + (1-rho) * grad * grad
+   *  mom <- momentum * mom_{t-1} + lr * grad / sqrt(ms + epsilon)
+   *  var <- var - mom
+   *
+   * @param var Should be from a Variable().
+   * @param ms Should be from a Variable().
+   * @param mom Should be from a Variable().
+   * @param lr Scaling factor. Must be a scalar.
+   * @param rho Decay rate. Must be a scalar.
+   * @param momentum
+   * @param epsilon Ridge term. Must be a scalar.
+   * @param grad The gradient.
+   * @param indices A vector of indices into the first dimension of var, ms and mom.
+   * @param options carries optional attributes values
+   * @return a new instance of ResourceSparseApplyRmsProp
+   */
+  public <T extends TType, U extends TNumber> ResourceSparseApplyRmsProp resourceSparseApplyRmsProp(
+      Operand<?> var, Operand<?> ms, Operand<?> mom, Operand<T> lr, Operand<T> rho,
+      Operand<T> momentum, Operand<T> epsilon, Operand<T> grad, Operand<U> indices,
+      ResourceSparseApplyRmsProp.Options... options) {
+    return ResourceSparseApplyRmsProp.create(scope, var, ms, mom, lr, rho, momentum, epsilon, grad, indices, options);
+  }
+
+  /**
+   * Restores tensors from a V2 checkpoint.
+   *  <p>
+   *  For backward compatibility with the V1 format, this Op currently allows
+   *  restoring from a V1 checkpoint as well:
+   *    - This Op first attempts to find the V2 index file pointed to by "prefix", and
+   *      if found proceed to read it as a V2 checkpoint;
+   *    - Otherwise the V1 read path is invoked.
+   *  Relying on this behavior is not recommended, as the ability to fall back to read
+   *  V1 might be deprecated and eventually removed.
+   *  <p>
+   *  By default, restores the named tensors in full.  If the caller wishes to restore
+   *  specific slices of stored tensors, "shape_and_slices" should be non-empty
+   *  strings and correspondingly well-formed.
+   *  <p>
+   *  Callers must ensure all the named tensors are indeed stored in the checkpoint.
+   *
+   * @param prefix Must have a single element.  The prefix of a V2 checkpoint.
+   * @param tensorNames shape {N}.  The names of the tensors to be restored.
+   * @param shapeAndSlices shape {N}.  The slice specs of the tensors to be restored.
+   *  Empty strings indicate that they are non-partitioned tensors.
+   * @param dtypes shape {N}.  The list of expected dtype for the tensors.  Must match
+   *  those stored in the checkpoint.
+   * @return a new instance of Restore
+   */
+  public Restore restore(Operand<TString> prefix, Operand<TString> tensorNames,
+      Operand<TString> shapeAndSlices, List<Class<? extends TType>> dtypes) {
+    return Restore.create(scope, prefix, tensorNames, shapeAndSlices, dtypes);
+  }
+
+  /**
+   * Restores a tensor from checkpoint files.
+   *  <p>
+   *  This is like `Restore` except that restored tensor can be listed as filling
+   *  only a slice of a larger tensor.  `shape_and_slice` specifies the shape of the
+   *  larger tensor and the slice that the restored tensor covers.
+   *  <p>
+   *  The `shape_and_slice` input has the same format as the
+   *  elements of the `shapes_and_slices` input of the `SaveSlices` op.
+   *
+   * @param <T> data type for {@code tensor()} output
+   * @param filePattern Must have a single element. The pattern of the files from
+   *  which we read the tensor.
+   * @param tensorName Must have a single element. The name of the tensor to be
+   *  restored.
+   * @param shapeAndSlice Scalar. The shapes and slice specifications to use when
+   *  restoring a tensors.
+   * @param dt The type of the tensor to be restored.
+   * @param options carries optional attributes values
+   * @return a new instance of RestoreSlice
+   */
+  public <T extends TType> RestoreSlice<T> restoreSlice(Operand<TString> filePattern,
+      Operand<TString> tensorName, Operand<TString> shapeAndSlice, Class<T> dt,
+      RestoreSlice.Options... options) {
+    return RestoreSlice.create(scope, filePattern, tensorName, shapeAndSlice, dt, options);
+  }
+
+  /**
+   * Saves tensors in V2 checkpoint format.
+   *  <p>
+   *  By default, saves the named tensors in full.  If the caller wishes to save
+   *  specific slices of full tensors, "shape_and_slices" should be non-empty strings
+   *  and correspondingly well-formed.
+   *
+   * @param prefix Must have a single element. The prefix of the V2 checkpoint to which we
+   *  write the tensors.
+   * @param tensorNames shape {N}. The names of the tensors to be saved.
+   * @param shapeAndSlices shape {N}.  The slice specs of the tensors to be saved.
+   *  Empty strings indicate that they are non-partitioned tensors.
+   * @param tensors `N` tensors to save.
+   * @return a new instance of Save
+   */
+  public Save save(Operand<TString> prefix, Operand<TString> tensorNames,
+      Operand<TString> shapeAndSlices, Iterable<Operand<?>> tensors) {
+    return Save.create(scope, prefix, tensorNames, shapeAndSlices, tensors);
+  }
+
+  /**
+   * Saves input tensors slices to disk.
+   *  <p>
+   *  This is like `Save` except that tensors can be listed in the saved file as being
+   *  a slice of a larger tensor.  `shapes_and_slices` specifies the shape of the
+   *  larger tensor and the slice that this tensor covers. `shapes_and_slices` must
+   *  have as many elements as `tensor_names`.
+   *  <p>
+   *  Elements of the `shapes_and_slices` input must either be:
+   *  <ul>
+   *  <li>
+   *  The empty string, in which case the corresponding tensor is
+   *     saved normally.
+   *  </li>
+   *  <li>
+   *  A string of the form `dim0 dim1 ... dimN-1 slice-spec` where the
+   *     `dimI` are the dimensions of the larger tensor and `slice-spec`
+   *     specifies what part is covered by the tensor to save.
+   *  </li>
+   *  </ul>
+   *  `slice-spec` itself is a `:`-separated list: `slice0:slice1:...:sliceN-1`
+   *  where each `sliceI` is either:
+   *  <ul>
+   *  <li>
+   *  The string `-` meaning that the slice covers all indices of this dimension
+   *  </li>
+   *  <li>
+   *  `start,length` where `start` and `length` are integers.  In that
+   *     case the slice covers `length` indices starting at `start`.
+   *  </li>
+   *  </ul>
+   *  See also `Save`.
+   *
+   * @param filename Must have a single element. The name of the file to which we write the
+   *  tensor.
+   * @param tensorNames Shape `[N]`. The names of the tensors to be saved.
+   * @param shapesAndSlices Shape `[N]`.  The shapes and slice specifications to use when
+   *  saving the tensors.
+   * @param data `N` tensors to save.
+   * @return a new instance of SaveSlices
+   */
+  public SaveSlices saveSlices(Operand<TString> filename, Operand<TString> tensorNames,
+      Operand<TString> shapesAndSlices, Iterable<Operand<?>> data) {
+    return SaveSlices.create(scope, filename, tensorNames, shapesAndSlices, data);
+  }
+
+  /**
+   * Computes fingerprints of the input strings.
+   *
+   * @param input vector of strings to compute fingerprints on.
+   * @return a new instance of SdcaFprint
+   */
+  public SdcaFprint sdcaFprint(Operand<TString> input) {
+    return SdcaFprint.create(scope, input);
+  }
+
+  /**
+   * Applies L1 regularization shrink step on the parameters.
+   *
+   * @param weights a list of vectors where each value is the weight associated with a
+   *  feature group.
+   * @param l1 Symmetric l1 regularization strength.
+   * @param l2 Symmetric l2 regularization strength. Should be a positive float.
+   * @return a new instance of SdcaShrinkL1
+   */
+  public SdcaShrinkL1 sdcaShrinkL1(Iterable<Operand<TFloat32>> weights, Float l1, Float l2) {
+    return SdcaShrinkL1.create(scope, weights, l1, l2);
+  }
+
+  /**
+   * var: Should be from a Variable().
+   *
+   * @param <T> data type for {@code out()} output
+   * @param var
+   * @param accum Should be from a Variable().
+   * @param accumUpdate : Should be from a Variable().
+   * @param lr Learning rate. Must be a scalar.
+   * @param rho Decay factor. Must be a scalar.
+   * @param epsilon Constant factor. Must be a scalar.
+   * @param grad The gradient.
+   * @param indices A vector of indices into the first dimension of var and accum.
+   * @param options carries optional attributes values
+   * @return a new instance of SparseApplyAdadelta
+   */
+  public <T extends TType, U extends TNumber> SparseApplyAdadelta<T> sparseApplyAdadelta(
+      Operand<T> var, Operand<T> accum, Operand<T> accumUpdate, Operand<T> lr, Operand<T> rho,
+      Operand<T> epsilon, Operand<T> grad, Operand<U> indices,
+      SparseApplyAdadelta.Options... options) {
+    return SparseApplyAdadelta.create(scope, var, accum, accumUpdate, lr, rho, epsilon, grad, indices, options);
+  }
+
+  /**
+   * Update entries in '*var' and '*accum' according to the proximal adagrad scheme.
+   *
+   * @param <T> data type for {@code out()} output
+   * @param var Should be from a Variable().
+   * @param gradientAccumulator Should be from a Variable().
+   * @param gradientSquaredAccumulator Should be from a Variable().
+   * @param grad The gradient.
+   * @param indices A vector of indices into the first dimension of var and accum.
+   * @param lr Learning rate. Must be a scalar.
+   * @param l1 L1 regularization. Must be a scalar.
+   * @param l2 L2 regularization. Must be a scalar.
+   * @param globalStep Training step number. Must be a scalar.
+   * @param options carries optional attributes values
+   * @return a new instance of SparseApplyAdagradDa
+   */
+  public <T extends TType, U extends TNumber> SparseApplyAdagradDa<T> sparseApplyAdagradDa(
+      Operand<T> var, Operand<T> gradientAccumulator, Operand<T> gradientSquaredAccumulator,
+      Operand<T> grad, Operand<U> indices, Operand<T> lr, Operand<T> l1, Operand<T> l2,
+      Operand<TInt64> globalStep, SparseApplyAdagradDa.Options... options) {
+    return SparseApplyAdagradDa.create(scope, var, gradientAccumulator, gradientSquaredAccumulator, grad, indices, lr, l1, l2, globalStep, options);
+  }
+
+  /**
+   * Update '*var' according to the centered RMSProp algorithm.
+   *  <p>
+   *  The centered RMSProp algorithm uses an estimate of the centered second moment
+   *  (i.e., the variance) for normalization, as opposed to regular RMSProp, which
+   *  uses the (uncentered) second moment. This often helps with training, but is
+   *  slightly more expensive in terms of computation and memory.
+   *  <p>
+   *  Note that in dense implementation of this algorithm, mg, ms, and mom will
+   *  update even if the grad is zero, but in this sparse implementation, mg, ms,
+   *  and mom will not update in iterations during which the grad is zero.
+   *  <p>
+   *  mean_square = decay * mean_square + (1-decay) * gradient ** 2
+   *  mean_grad = decay * mean_grad + (1-decay) * gradient
+   *  Delta = learning_rate * gradient / sqrt(mean_square + epsilon - mean_grad ** 2)
+   *  <p>
+   *  $$ms <- rho * ms_{t-1} + (1-rho) * grad * grad$$
+   *  $$mom <- momentum * mom_{t-1} + lr * grad / sqrt(ms + epsilon)$$
+   *  $$var <- var - mom$$
+   *
+   * @param <T> data type for {@code out()} output
+   * @param var Should be from a Variable().
+   * @param mg Should be from a Variable().
+   * @param ms Should be from a Variable().
+   * @param mom Should be from a Variable().
+   * @param lr Scaling factor. Must be a scalar.
+   * @param rho Decay rate. Must be a scalar.
+   * @param momentum
+   * @param epsilon Ridge term. Must be a scalar.
+   * @param grad The gradient.
+   * @param indices A vector of indices into the first dimension of var, ms and mom.
+   * @param options carries optional attributes values
+   * @return a new instance of SparseApplyCenteredRmsProp
+   */
+  public <T extends TType, U extends TNumber> SparseApplyCenteredRmsProp<T> sparseApplyCenteredRmsProp(
+      Operand<T> var, Operand<T> mg, Operand<T> ms, Operand<T> mom, Operand<T> lr, Operand<T> rho,
+      Operand<T> momentum, Operand<T> epsilon, Operand<T> grad, Operand<U> indices,
+      SparseApplyCenteredRmsProp.Options... options) {
+    return SparseApplyCenteredRmsProp.create(scope, var, mg, ms, mom, lr, rho, momentum, epsilon, grad, indices, options);
+  }
+
+  /**
+   * Update relevant entries in '*var' according to the Ftrl-proximal scheme.
+   *  <p>
+   *  That is for rows we have grad for, we update var, accum and linear as follows:
+   *  grad_with_shrinkage = grad + 2 * l2_shrinkage * var
+   *  accum_new = accum + grad * grad
+   *  linear += grad_with_shrinkage -
+   *      (accum_new^(-lr_power) - accum^(-lr_power)) / lr * var
+   *  quadratic = 1.0 / (accum_new^(lr_power) * lr) + 2 * l2
+   *  var = (sign(linear) * l1 - linear) / quadratic if |linear| > l1 else 0.0
+   *  accum = accum_new
+   *
+   * @param <T> data type for {@code out()} output
+   * @param var Should be from a Variable().
+   * @param accum Should be from a Variable().
+   * @param linear Should be from a Variable().
+   * @param grad The gradient.
+   * @param indices A vector of indices into the first dimension of var and accum.
+   * @param lr Scaling factor. Must be a scalar.
+   * @param l1 L1 regularization. Must be a scalar.
+   * @param l2 L2 shrinkage regularization. Must be a scalar.
+   * @param l2Shrinkage
+   * @param lrPower Scaling factor. Must be a scalar.
+   * @param options carries optional attributes values
+   * @return a new instance of SparseApplyFtrl
+   */
+  public <T extends TType, U extends TNumber> SparseApplyFtrl<T> sparseApplyFtrl(Operand<T> var,
+      Operand<T> accum, Operand<T> linear, Operand<T> grad, Operand<U> indices, Operand<T> lr,
+      Operand<T> l1, Operand<T> l2, Operand<T> l2Shrinkage, Operand<T> lrPower,
+      SparseApplyFtrl.Options... options) {
+    return SparseApplyFtrl.create(scope, var, accum, linear, grad, indices, lr, l1, l2, l2Shrinkage, lrPower, options);
+  }
+
+  /**
+   * Update relevant entries in '*var' and '*accum' according to the momentum scheme.
+   *  <p>
+   *  Set use_nesterov = True if you want to use Nesterov momentum.
+   *  <p>
+   *  That is for rows we have grad for, we update var and accum as follows:
+   *  <p>
+   *  $$accum = accum * momentum + grad$$
+   *  $$var -= lr * accum$$
+   *
+   * @param <T> data type for {@code out()} output
+   * @param var Should be from a Variable().
+   * @param accum Should be from a Variable().
+   * @param lr Learning rate. Must be a scalar.
+   * @param grad The gradient.
+   * @param indices A vector of indices into the first dimension of var and accum.
+   * @param momentum Momentum. Must be a scalar.
+   * @param options carries optional attributes values
+   * @return a new instance of SparseApplyMomentum
+   */
+  public <T extends TType, U extends TNumber> SparseApplyMomentum<T> sparseApplyMomentum(
+      Operand<T> var, Operand<T> accum, Operand<T> lr, Operand<T> grad, Operand<U> indices,
+      Operand<T> momentum, SparseApplyMomentum.Options... options) {
+    return SparseApplyMomentum.create(scope, var, accum, lr, grad, indices, momentum, options);
+  }
+
+  /**
+   * Sparse update entries in '*var' and '*accum' according to FOBOS algorithm.
+   *  <p>
+   *  That is for rows we have grad for, we update var and accum as follows:
+   *  $$accum += grad <i> grad$$
+   *  $$prox_v = var$$
+   *  $$prox_v -= lr </i> grad <i> (1 / sqrt(accum))$$
+   *  $$var = sign(prox_v)/(1+lr</i>l2) <i> max{|prox_v|-lr</i>l1,0}$$
+   *
+   * @param <T> data type for {@code out()} output
+   * @param var Should be from a Variable().
+   * @param accum Should be from a Variable().
+   * @param lr Learning rate. Must be a scalar.
+   * @param l1 L1 regularization. Must be a scalar.
+   * @param l2 L2 regularization. Must be a scalar.
+   * @param grad The gradient.
+   * @param indices A vector of indices into the first dimension of var and accum.
+   * @param options carries optional attributes values
+   * @return a new instance of SparseApplyProximalAdagrad
+   */
+  public <T extends TType, U extends TNumber> SparseApplyProximalAdagrad<T> sparseApplyProximalAdagrad(
+      Operand<T> var, Operand<T> accum, Operand<T> lr, Operand<T> l1, Operand<T> l2,
+      Operand<T> grad, Operand<U> indices, SparseApplyProximalAdagrad.Options... options) {
+    return SparseApplyProximalAdagrad.create(scope, var, accum, lr, l1, l2, grad, indices, options);
+  }
+
+  /**
+   * Sparse update '*var' as FOBOS algorithm with fixed learning rate.
+   *  <p>
+   *  That is for rows we have grad for, we update var as follows:
+   *  $$prox_v = var - alpha <i> grad$$
+   *  $$var = sign(prox_v)/(1+alpha</i>l2) <i> max{|prox_v|-alpha</i>l1,0}$$
+   *
+   * @param <T> data type for {@code out()} output
+   * @param var Should be from a Variable().
+   * @param alpha Scaling factor. Must be a scalar.
+   * @param l1 L1 regularization. Must be a scalar.
+   * @param l2 L2 regularization. Must be a scalar.
+   * @param grad The gradient.
+   * @param indices A vector of indices into the first dimension of var and accum.
+   * @param options carries optional attributes values
+   * @return a new instance of SparseApplyProximalGradientDescent
+   */
+  public <T extends TType, U extends TNumber> SparseApplyProximalGradientDescent<T> sparseApplyProximalGradientDescent(
+      Operand<T> var, Operand<T> alpha, Operand<T> l1, Operand<T> l2, Operand<T> grad,
+      Operand<U> indices, SparseApplyProximalGradientDescent.Options... options) {
+    return SparseApplyProximalGradientDescent.create(scope, var, alpha, l1, l2, grad, indices, options);
+  }
+
+  /**
+   * Update '*var' according to the RMSProp algorithm.
+   *  <p>
+   *  Note that in dense implementation of this algorithm, ms and mom will
+   *  update even if the grad is zero, but in this sparse implementation, ms
+   *  and mom will not update in iterations during which the grad is zero.
+   *  <p>
+   *  mean_square = decay * mean_square + (1-decay) * gradient ** 2
+   *  Delta = learning_rate * gradient / sqrt(mean_square + epsilon)
+   *  <p>
+   *  $$ms <- rho * ms_{t-1} + (1-rho) * grad * grad$$
+   *  $$mom <- momentum * mom_{t-1} + lr * grad / sqrt(ms + epsilon)$$
+   *  $$var <- var - mom$$
+   *
+   * @param <T> data type for {@code out()} output
+   * @param var Should be from a Variable().
+   * @param ms Should be from a Variable().
+   * @param mom Should be from a Variable().
+   * @param lr Scaling factor. Must be a scalar.
+   * @param rho Decay rate. Must be a scalar.
+   * @param momentum
+   * @param epsilon Ridge term. Must be a scalar.
+   * @param grad The gradient.
+   * @param indices A vector of indices into the first dimension of var, ms and mom.
+   * @param options carries optional attributes values
+   * @return a new instance of SparseApplyRmsProp
+   */
+  public <T extends TType, U extends TNumber> SparseApplyRmsProp<T> sparseApplyRmsProp(
+      Operand<T> var, Operand<T> ms, Operand<T> mom, Operand<T> lr, Operand<T> rho,
+      Operand<T> momentum, Operand<T> epsilon, Operand<T> grad, Operand<U> indices,
+      SparseApplyRmsProp.Options... options) {
+    return SparseApplyRmsProp.create(scope, var, ms, mom, lr, rho, momentum, epsilon, grad, indices, options);
+  }
+
+  /**
+   * Returns the gradient of `Tile`.
+   *  <p>
+   *  Since `Tile` takes an input and repeats the input `multiples` times
+   *  along each dimension, `train.TileGrad` takes in `multiples` and aggregates
+   *  each repeated tile of `input` into `output`.
+   *
+   * @param <T> data type for {@code output()} output
+   * @param input
+   * @param multiples
+   * @return a new instance of TileGrad
+   */
+  public <T extends TType> TileGrad<T> tileGrad(Operand<T> input, Operand<TInt32> multiples) {
+    return TileGrad.create(scope, input, multiples);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/XlaOps.java b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/XlaOps.java
new file mode 100644
index 00000000000..6c16e3cb7ff
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/XlaOps.java
@@ -0,0 +1,381 @@
+// Copyright 2020 The TensorFlow Authors. All Rights Reserved.
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+// ==============================================================================
+//
+// This class has been generated, DO NOT EDIT!
+//
+package org.tensorflow.op;
+
+import org.tensorflow.Operand;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.xla.BroadcastHelper;
+import org.tensorflow.op.xla.ClusterOutput;
+import org.tensorflow.op.xla.Conv;
+import org.tensorflow.op.xla.Dequantize;
+import org.tensorflow.op.xla.Dot;
+import org.tensorflow.op.xla.DynamicSlice;
+import org.tensorflow.op.xla.DynamicUpdateSlice;
+import org.tensorflow.op.xla.Einsum;
+import org.tensorflow.op.xla.Gather;
+import org.tensorflow.op.xla.KeyValueSort;
+import org.tensorflow.op.xla.Pad;
+import org.tensorflow.op.xla.Recv;
+import org.tensorflow.op.xla.ReplicaId;
+import org.tensorflow.op.xla.SelfAdjointEig;
+import org.tensorflow.op.xla.Send;
+import org.tensorflow.op.xla.Sharding;
+import org.tensorflow.op.xla.Sort;
+import org.tensorflow.op.xla.Svd;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * An API for building {@code xla} operations as {@link Op Op}s
+ *
+ * @see {@link Ops}
+ */
+public final class XlaOps {
+  private final Scope scope;
+
+  XlaOps(Scope scope) {
+    this.scope = scope;
+  }
+
+  /**
+   * Helper operator for performing XLA-style broadcasts
+   *  <p>
+   *  Broadcasts `lhs` and `rhs` to the same rank, by adding size 1 dimensions to
+   *  whichever of `lhs` and `rhs` has the lower rank, using XLA's broadcasting rules
+   *  for binary operators.
+   *
+   * @param <T> data type for {@code lhsOutput()} output
+   * @param lhs the LHS input tensor
+   * @param rhs the RHS input tensor
+   * @param broadcastDims an XLA-style broadcast dimension specification
+   * @return a new instance of BroadcastHelper
+   */
+  public <T extends TType, U extends TNumber> BroadcastHelper<T> broadcastHelper(Operand<T> lhs,
+      Operand<T> rhs, Operand<U> broadcastDims) {
+    return BroadcastHelper.create(scope, lhs, rhs, broadcastDims);
+  }
+
+  /**
+   * Operator that connects the output of an XLA computation to other consumer graph nodes.
+   *
+   * @param <T> data type for {@code outputs()} output
+   * @param input
+   * @return a new instance of ClusterOutput
+   */
+  public <T extends TType> ClusterOutput<T> clusterOutput(Operand<T> input) {
+    return ClusterOutput.create(scope, input);
+  }
+
+  /**
+   * Wraps the XLA ConvGeneralDilated operator, documented at
+   *  <p>
+   *   https://www.tensorflow.org/performance/xla/operation_semantics#conv_convolution
+   *  .
+   *
+   * @param <T> data type for {@code output()} output
+   * @param lhs the input tensor
+   * @param rhs the kernel tensor
+   * @param windowStrides the inter-window strides
+   * @param padding the padding to apply at the start and end of each input dimensions
+   * @param lhsDilation dilation to apply between input elements
+   * @param rhsDilation dilation to apply between kernel elements
+   * @param featureGroupCount number of feature groups for grouped convolution.
+   * @param dimensionNumbers a serialized xla::ConvolutionDimensionNumbers proto.
+   * @param precisionConfig a serialized xla::PrecisionConfig proto.
+   * @return a new instance of Conv
+   */
+  public <T extends TType, U extends TNumber> Conv<T> conv(Operand<T> lhs, Operand<T> rhs,
+      Operand<U> windowStrides, Operand<U> padding, Operand<U> lhsDilation, Operand<U> rhsDilation,
+      Operand<U> featureGroupCount, String dimensionNumbers, String precisionConfig) {
+    return Conv.create(scope, lhs, rhs, windowStrides, padding, lhsDilation, rhsDilation, featureGroupCount, dimensionNumbers, precisionConfig);
+  }
+
+  /**
+   * Takes the packed uint32 input and unpacks the input to uint8 to do
+   *  <p>
+   *  Dequantization on device.
+   *
+   * @param input Input tensors whose types is uint32, shape is [d0, ..., dn].
+   * @param minRange The minimum scalar value possibly produced for the input.
+   * @param maxRange The maximum scalar value possibly produced for the input.
+   * @param mode String to determine the dequantize mode in {"MIN_COMBINED", "MIN_FIRST", "SCALED"}.
+   * @param transposeOutput Boolean to determine if output is transposed. transpose_output
+   *  is faster when input is large and rank of input is higher than 1.
+   * @return a new instance of Dequantize
+   */
+  public Dequantize dequantize(Operand<?> input, Float minRange, Float maxRange, String mode,
+      Boolean transposeOutput) {
+    return Dequantize.create(scope, input, minRange, maxRange, mode, transposeOutput);
+  }
+
+  /**
+   * Wraps the XLA DotGeneral operator, documented at
+   *  <p>
+   *   https://www.tensorflow.org/performance/xla/operation_semantics#dotgeneral
+   *  .
+   *
+   * @param <T> data type for {@code output()} output
+   * @param lhs the LHS tensor
+   * @param rhs the RHS tensor
+   * @param dimensionNumbers a serialized xla::DotDimensionNumbers proto.
+   * @param precisionConfig a serialized xla::PrecisionConfig proto.
+   * @return a new instance of Dot
+   */
+  public <T extends TType> Dot<T> dot(Operand<T> lhs, Operand<T> rhs, String dimensionNumbers,
+      String precisionConfig) {
+    return Dot.create(scope, lhs, rhs, dimensionNumbers, precisionConfig);
+  }
+
+  /**
+   * Wraps the XLA DynamicSlice operator, documented at
+   *  <p>
+   *   https://www.tensorflow.org/performance/xla/operation_semantics#dynamicslice
+   *  .
+   *  <p>
+   *  DynamicSlice extracts a sub-array from the input array at dynamic
+   *  start_indices. The size of the slice in each dimension is passed in
+   *  size_indices, which specify the end point of exclusive slice intervals in each
+   *  dimension -- [start, start + size). The shape of start_indices must have rank 1,
+   *  with dimension size equal to the rank of operand.
+   *
+   * @param <T> data type for {@code output()} output
+   * @param input A `Tensor` of type T.
+   * @param startIndices List of N integers containing the slice size for each
+   *  dimension. Each value must be strictly greater than zero, and start + size
+   *  must be less than or equal to the size of the dimension to avoid
+   *  implementation defined behavior.
+   * @param sizeIndices
+   * @return a new instance of DynamicSlice
+   */
+  public <T extends TType, U extends TNumber> DynamicSlice<T> dynamicSlice(Operand<T> input,
+      Operand<U> startIndices, Operand<U> sizeIndices) {
+    return DynamicSlice.create(scope, input, startIndices, sizeIndices);
+  }
+
+  /**
+   * Wraps the XLA DynamicUpdateSlice operator, documented at
+   *  <p>
+   *   https://www.tensorflow.org/performance/xla/operation_semantics#dynamicupdateslice
+   *  .
+   *  <p>
+   *  XlaDynamicUpdateSlice generates a result which is the value of the `input`
+   *  operand, with a slice update overwritten at `indices`. The shape of `update`
+   *  determines the shape of the sub-array of the result which is updated. The shape
+   *  of indices must be rank == 1, with dimension size equal to the rank of `input`.
+   *  <p>
+   *  Handling of out-of-bounds slice indices is implementation-defined.
+   *
+   * @param <T> data type for {@code output()} output
+   * @param input A `Tensor` of type T.
+   * @param update A `Tensor` of type T. Same rank as `input`.
+   * @param indices A vector of indices into `input`. Must have length equal to the rank of
+   *  `input`.
+   * @return a new instance of DynamicUpdateSlice
+   */
+  public <T extends TType, U extends TNumber> DynamicUpdateSlice<T> dynamicUpdateSlice(
+      Operand<T> input, Operand<T> update, Operand<U> indices) {
+    return DynamicUpdateSlice.create(scope, input, update, indices);
+  }
+
+  /**
+   * An op which supports basic einsum op with 2 inputs and 1 output.
+   *  <p>
+   *  This op has better TPU performance since it doesn't have explicitly reshape and
+   *  transpose operations as tf.einsum does.
+   *
+   * @param <T> data type for {@code product()} output
+   * @param a
+   * @param b
+   * @param equation
+   * @return a new instance of Einsum
+   */
+  public <T extends TType> Einsum<T> einsum(Operand<T> a, Operand<T> b, String equation) {
+    return Einsum.create(scope, a, b, equation);
+  }
+
+  /**
+   * Wraps the XLA Gather operator documented at
+   *  <p>
+   *    https://www.tensorflow.org/xla/operation_semantics#gather
+   *
+   * @param <T> data type for {@code output()} output
+   * @param operand The array we're gathering from.
+   * @param startIndices Array containing the starting indices of the slices we gather.
+   * @param sliceSizes slice_sizes[i] is the bounds for the slice on dimension i.
+   * @param dimensionNumbers A serialized xla::GatherDimensionNumbers proto.
+   * @param indicesAreSorted Boolean indicating if the indices are sorted.
+   * @return a new instance of Gather
+   */
+  public <T extends TType, U extends TNumber> Gather<T> gather(Operand<T> operand,
+      Operand<U> startIndices, Operand<U> sliceSizes, String dimensionNumbers,
+      Boolean indicesAreSorted) {
+    return Gather.create(scope, operand, startIndices, sliceSizes, dimensionNumbers, indicesAreSorted);
+  }
+
+  /**
+   * Wraps the XLA Sort operator, documented at
+   *  <p>
+   *   https://www.tensorflow.org/performance/xla/operation_semantics#sort
+   *  .
+   *  <p>
+   *  Sorts a tensor. Currently only sorts in ascending order are supported.
+   *
+   * @param <T> data type for {@code sortedKeys()} output
+   * @param <U> data type for {@code sortedValues()} output
+   * @param keys A `Tensor` of type K.
+   * @param values A `Tensor` of type V.
+   * @return a new instance of KeyValueSort
+   */
+  public <T extends TNumber, U extends TType> KeyValueSort<T, U> keyValueSort(Operand<T> keys,
+      Operand<U> values) {
+    return KeyValueSort.create(scope, keys, values);
+  }
+
+  /**
+   * Wraps the XLA Pad operator, documented at
+   *  <p>
+   *   https://www.tensorflow.org/performance/xla/operation_semantics#pad
+   *  .
+   *
+   * @param <T> data type for {@code output()} output
+   * @param input A `Tensor` of type T.
+   * @param paddingValue A scalar `Tensor` of type T.
+   * @param paddingLow the padding to apply at the start of each input dimensions
+   * @param paddingHigh the padding to apply at the end of each input dimension.
+   * @param paddingInterior the padding to apply between each input element.
+   * @return a new instance of Pad
+   */
+  public <T extends TType, U extends TNumber> Pad<T> pad(Operand<T> input, Operand<T> paddingValue,
+      Operand<U> paddingLow, Operand<U> paddingHigh, Operand<U> paddingInterior) {
+    return Pad.create(scope, input, paddingValue, paddingLow, paddingHigh, paddingInterior);
+  }
+
+  /**
+   * Receives the named tensor from another XLA computation. Wraps the XLA Recv
+   *  <p>
+   *  operator documented at
+   *   https://www.tensorflow.org/performance/xla/operation_semantics#recv .
+   *
+   * @param <T> data type for {@code tensor()} output
+   * @param dtype The type of the tensor.
+   * @param tensorName A string key that identifies the channel.
+   * @param shape The shape of the tensor.
+   * @return a new instance of Recv
+   */
+  public <T extends TType> Recv<T> recv(Class<T> dtype, String tensorName, Shape shape) {
+    return Recv.create(scope, dtype, tensorName, shape);
+  }
+
+  /**
+   * Replica ID.
+   *
+   * @return a new instance of ReplicaId
+   */
+  public ReplicaId replicaId() {
+    return ReplicaId.create(scope);
+  }
+
+  /**
+   * Computes the eigen decomposition of a batch of self-adjoint matrices
+   *  <p>
+   *  (Note: Only real inputs are supported).
+   *  <p>
+   *  Computes the eigenvalues and eigenvectors of the innermost N-by-N matrices in
+   *  tensor such that tensor[...,:,:] * v[..., :,i] = e[..., i] * v[...,:,i], for
+   *  i=0...N-1.
+   *
+   * @param <T> data type for {@code w()} output
+   * @param a the input tensor.
+   * @param lower a boolean specifies whether the calculation is done with the lower
+   *  triangular part or the upper triangular part.
+   * @param maxIter maximum number of sweep update, i.e., the whole lower triangular
+   *  part or upper triangular part based on parameter lower. Heuristically, it has
+   *  been argued that approximately logN sweeps are needed in practice (Ref: Golub &
+   *  van Loan "Matrix Computation").
+   * @param epsilon the tolerance ratio.
+   * @return a new instance of SelfAdjointEig
+   */
+  public <T extends TType> SelfAdjointEig<T> selfAdjointEig(Operand<T> a, Boolean lower,
+      Long maxIter, Float epsilon) {
+    return SelfAdjointEig.create(scope, a, lower, maxIter, epsilon);
+  }
+
+  /**
+   * Sends the named tensor to another XLA computation. Wraps the XLA Send operator
+   *  <p>
+   *  documented at
+   *   https://www.tensorflow.org/performance/xla/operation_semantics#send .
+   *
+   * @param tensor The tensor to send.
+   * @param tensorName A string key that identifies the channel.
+   * @return a new instance of Send
+   */
+  public <T extends TType> Send send(Operand<T> tensor, String tensorName) {
+    return Send.create(scope, tensor, tensorName);
+  }
+
+  /**
+   * An op which shards the input based on the given sharding attribute.
+   *
+   * @param <T> data type for {@code output()} output
+   * @param input
+   * @return a new instance of Sharding
+   */
+  public <T extends TType> Sharding<T> sharding(Operand<T> input) {
+    return Sharding.create(scope, input);
+  }
+
+  /**
+   * Wraps the XLA Sort operator, documented at
+   *  <p>
+   *   https://www.tensorflow.org/performance/xla/operation_semantics#sort
+   *  .
+   *  <p>
+   *  Sorts a tensor. Currently only sorts in ascending order are supported.
+   *
+   * @param <T> data type for {@code output()} output
+   * @param input A `Tensor` of type T.
+   * @return a new instance of Sort
+   */
+  public <T extends TType> Sort<T> sort(Operand<T> input) {
+    return Sort.create(scope, input);
+  }
+
+  /**
+   * Computes the eigen decomposition of a batch of self-adjoint matrices
+   *  <p>
+   *  (Note: Only real inputs are supported).
+   *  <p>
+   *  Computes the eigenvalues and eigenvectors of the innermost M-by-N matrices in
+   *  tensor such that tensor[...,:,:] = u[..., :, :] * Diag(s[..., :]) * Transpose(v[...,:,:]).
+   *
+   * @param <T> data type for {@code s()} output
+   * @param a the input tensor.
+   * @param maxIter maximum number of sweep update, i.e., the whole lower triangular
+   *  part or upper triangular part based on parameter lower. Heuristically, it has
+   *  been argued that approximately log(min (M, N)) sweeps are needed in practice
+   *  (Ref: Golub & van Loan "Matrix Computation").
+   * @param epsilon the tolerance ratio.
+   * @param precisionConfig a serialized xla::PrecisionConfig proto.
+   * @return a new instance of Svd
+   */
+  public <T extends TType> Svd<T> svd(Operand<T> a, Long maxIter, Float epsilon,
+      String precisionConfig) {
+    return Svd.create(scope, a, maxIter, epsilon, precisionConfig);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/Compute_func_Pointer_TF_OpKernelContext.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/Compute_func_Pointer_TF_OpKernelContext.java
new file mode 100644
index 00000000000..fd70471e100
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/Compute_func_Pointer_TF_OpKernelContext.java
@@ -0,0 +1,19 @@
+// Targeted by JavaCPP version 1.5.4: DO NOT EDIT THIS FILE
+
+package org.tensorflow.internal.c_api;
+
+import java.nio.*;
+import org.bytedeco.javacpp.*;
+import org.bytedeco.javacpp.annotation.*;
+
+import static org.tensorflow.internal.c_api.global.tensorflow.*;
+
+@Properties(inherit = org.tensorflow.internal.c_api.presets.tensorflow.class)
+public class Compute_func_Pointer_TF_OpKernelContext extends FunctionPointer {
+    static { Loader.load(); }
+    /** Pointer cast constructor. Invokes {@link Pointer#Pointer(Pointer)}. */
+    public    Compute_func_Pointer_TF_OpKernelContext(Pointer p) { super(p); }
+    protected Compute_func_Pointer_TF_OpKernelContext() { allocate(); }
+    private native void allocate();
+    public native void call(Pointer arg0, TF_OpKernelContext arg1);
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/Create_func_TF_OpKernelConstruction.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/Create_func_TF_OpKernelConstruction.java
new file mode 100644
index 00000000000..8f951ea6a73
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/Create_func_TF_OpKernelConstruction.java
@@ -0,0 +1,48 @@
+// Targeted by JavaCPP version 1.5.4: DO NOT EDIT THIS FILE
+
+package org.tensorflow.internal.c_api;
+
+import java.nio.*;
+import org.bytedeco.javacpp.*;
+import org.bytedeco.javacpp.annotation.*;
+
+import static org.tensorflow.internal.c_api.global.tensorflow.*;
+
+
+// Allocates a new kernel builder and returns a pointer to it.
+//
+// If non-null, TensorFlow will call create_func when it needs to instantiate
+// the kernel. The pointer returned by create_func will be passed to
+// compute_func and delete_func, thereby functioning as a "this" pointer for
+// referring to kernel instances.
+//
+// The TF_OpKernelConstruction pointer passed to create_func is owned by
+// TensorFlow and will be deleted once create_func returns. It must not be used
+// after this.
+//
+// When TensorFlow needs to perform a computation with this kernel, it will
+// call compute_func. This function will receive the pointer returned by
+// create_func (or null if no create_func was provided), along with the inputs
+// to the computation.
+//
+// The TF_OpKernelContext pointer received by compute_func is owned by
+// TensorFlow and will be deleted once compute_func returns. It must not be used
+// after this.
+//
+// Finally, when TensorFlow no longer needs the kernel, it will call
+// delete_func if one is provided. This function will receive the pointer
+// returned in `create_func` or nullptr if no `create_func` was provided.
+//
+// The caller should pass the result of this function to
+// TF_RegisterKernelBuilder, which will take ownership of the pointer. If, for
+// some reason, the kernel builder will not be registered, the caller should
+// delete it with TF_DeleteKernelBuilder.
+@Properties(inherit = org.tensorflow.internal.c_api.presets.tensorflow.class)
+public class Create_func_TF_OpKernelConstruction extends FunctionPointer {
+    static { Loader.load(); }
+    /** Pointer cast constructor. Invokes {@link Pointer#Pointer(Pointer)}. */
+    public    Create_func_TF_OpKernelConstruction(Pointer p) { super(p); }
+    protected Create_func_TF_OpKernelConstruction() { allocate(); }
+    private native void allocate();
+    public native Pointer call(TF_OpKernelConstruction arg0);
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/Deallocator_Pointer_long_Pointer.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/Deallocator_Pointer_long_Pointer.java
new file mode 100644
index 00000000000..168135a5b14
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/Deallocator_Pointer_long_Pointer.java
@@ -0,0 +1,30 @@
+// Targeted by JavaCPP version 1.5.4: DO NOT EDIT THIS FILE
+
+package org.tensorflow.internal.c_api;
+
+import java.nio.*;
+import org.bytedeco.javacpp.*;
+import org.bytedeco.javacpp.annotation.*;
+
+import static org.tensorflow.internal.c_api.global.tensorflow.*;
+
+
+// Return a new tensor that holds the bytes data[0,len-1].
+//
+// The data will be deallocated by a subsequent call to TF_DeleteTensor via:
+//      (*deallocator)(data, len, deallocator_arg)
+// Clients must provide a custom deallocator function so they can pass in
+// memory managed by something like numpy.
+//
+// May return NULL (and invoke the deallocator) if the provided data buffer
+// (data, len) is inconsistent with a tensor of the given TF_DataType
+// and the shape specified by (dima, num_dims).
+@Properties(inherit = org.tensorflow.internal.c_api.presets.tensorflow.class)
+public class Deallocator_Pointer_long_Pointer extends FunctionPointer {
+    static { Loader.load(); }
+    /** Pointer cast constructor. Invokes {@link Pointer#Pointer(Pointer)}. */
+    public    Deallocator_Pointer_long_Pointer(Pointer p) { super(p); }
+    protected Deallocator_Pointer_long_Pointer() { allocate(); }
+    private native void allocate();
+    public native void call(Pointer data, @Cast("size_t") long len, Pointer arg);
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/Delete_func_Pointer.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/Delete_func_Pointer.java
new file mode 100644
index 00000000000..734040f20f4
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/Delete_func_Pointer.java
@@ -0,0 +1,19 @@
+// Targeted by JavaCPP version 1.5.4: DO NOT EDIT THIS FILE
+
+package org.tensorflow.internal.c_api;
+
+import java.nio.*;
+import org.bytedeco.javacpp.*;
+import org.bytedeco.javacpp.annotation.*;
+
+import static org.tensorflow.internal.c_api.global.tensorflow.*;
+
+@Properties(inherit = org.tensorflow.internal.c_api.presets.tensorflow.class)
+public class Delete_func_Pointer extends FunctionPointer {
+    static { Loader.load(); }
+    /** Pointer cast constructor. Invokes {@link Pointer#Pointer(Pointer)}. */
+    public    Delete_func_Pointer(Pointer p) { super(p); }
+    protected Delete_func_Pointer() { allocate(); }
+    private native void allocate();
+    public native void call(Pointer arg0);
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/Listener_BytePointer.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/Listener_BytePointer.java
new file mode 100644
index 00000000000..f1775998256
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/Listener_BytePointer.java
@@ -0,0 +1,25 @@
+// Targeted by JavaCPP version 1.5.4: DO NOT EDIT THIS FILE
+
+package org.tensorflow.internal.c_api;
+
+import java.nio.*;
+import org.bytedeco.javacpp.*;
+import org.bytedeco.javacpp.annotation.*;
+
+import static org.tensorflow.internal.c_api.global.tensorflow.*;
+
+
+// Register a listener method that processes printed messages.
+//
+// If any listeners are registered, the print operator will call all listeners
+// with the printed messages and immediately return without writing to the
+// logs.
+@Properties(inherit = org.tensorflow.internal.c_api.presets.tensorflow.class)
+public class Listener_BytePointer extends FunctionPointer {
+    static { Loader.load(); }
+    /** Pointer cast constructor. Invokes {@link Pointer#Pointer(Pointer)}. */
+    public    Listener_BytePointer(Pointer p) { super(p); }
+    protected Listener_BytePointer() { allocate(); }
+    private native void allocate();
+    public native void call(@Cast("const char*") BytePointer arg0);
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/Listener_String.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/Listener_String.java
new file mode 100644
index 00000000000..a4114f23dbe
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/Listener_String.java
@@ -0,0 +1,19 @@
+// Targeted by JavaCPP version 1.5.4: DO NOT EDIT THIS FILE
+
+package org.tensorflow.internal.c_api;
+
+import java.nio.*;
+import org.bytedeco.javacpp.*;
+import org.bytedeco.javacpp.annotation.*;
+
+import static org.tensorflow.internal.c_api.global.tensorflow.*;
+
+@Properties(inherit = org.tensorflow.internal.c_api.presets.tensorflow.class)
+public class Listener_String extends FunctionPointer {
+    static { Loader.load(); }
+    /** Pointer cast constructor. Invokes {@link Pointer#Pointer(Pointer)}. */
+    public    Listener_String(Pointer p) { super(p); }
+    protected Listener_String() { allocate(); }
+    private native void allocate();
+    public native void call(String arg0);
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/Shape_inference_func_TF_ShapeInferenceContext_TF_Status.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/Shape_inference_func_TF_ShapeInferenceContext_TF_Status.java
new file mode 100644
index 00000000000..56e99923a0b
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/Shape_inference_func_TF_ShapeInferenceContext_TF_Status.java
@@ -0,0 +1,22 @@
+// Targeted by JavaCPP version 1.5.4: DO NOT EDIT THIS FILE
+
+package org.tensorflow.internal.c_api;
+
+import java.nio.*;
+import org.bytedeco.javacpp.*;
+import org.bytedeco.javacpp.annotation.*;
+
+import static org.tensorflow.internal.c_api.global.tensorflow.*;
+
+
+// Sets the shape inference function for the op.
+@Properties(inherit = org.tensorflow.internal.c_api.presets.tensorflow.class)
+public class Shape_inference_func_TF_ShapeInferenceContext_TF_Status extends FunctionPointer {
+    static { Loader.load(); }
+    /** Pointer cast constructor. Invokes {@link Pointer#Pointer(Pointer)}. */
+    public    Shape_inference_func_TF_ShapeInferenceContext_TF_Status(Pointer p) { super(p); }
+    protected Shape_inference_func_TF_ShapeInferenceContext_TF_Status() { allocate(); }
+    private native void allocate();
+    public native void call(TF_ShapeInferenceContext ctx,
+                                 TF_Status status);
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TFE_Context.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TFE_Context.java
new file mode 100644
index 00000000000..886782d2e0b
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TFE_Context.java
@@ -0,0 +1,25 @@
+// Targeted by JavaCPP version 1.5.4: DO NOT EDIT THIS FILE
+
+package org.tensorflow.internal.c_api;
+
+import java.nio.*;
+import org.bytedeco.javacpp.*;
+import org.bytedeco.javacpp.annotation.*;
+
+import static org.tensorflow.internal.c_api.global.tensorflow.*;
+
+
+// "Context" under which operations/functions are executed. It encapsulates
+// things like the available devices, resource manager etc.
+// TFE_Context must outlive all tensor handles created using it. In other
+// words, TFE_DeleteContext() must be called after all tensor handles have
+// been deleted (with TFE_DeleteTensorHandle).
+//
+// TODO(ashankar): Merge with TF_Session?
+@Opaque @Properties(inherit = org.tensorflow.internal.c_api.presets.tensorflow.class)
+public class TFE_Context extends org.tensorflow.internal.c_api.AbstractTFE_Context {
+    /** Empty constructor. Calls {@code super((Pointer)null)}. */
+    public TFE_Context() { super((Pointer)null); }
+    /** Pointer cast constructor. Invokes {@link Pointer#Pointer(Pointer)}. */
+    public TFE_Context(Pointer p) { super(p); }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TFE_ContextOptions.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TFE_ContextOptions.java
new file mode 100644
index 00000000000..7d986372863
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TFE_ContextOptions.java
@@ -0,0 +1,19 @@
+// Targeted by JavaCPP version 1.5.4: DO NOT EDIT THIS FILE
+
+package org.tensorflow.internal.c_api;
+
+import java.nio.*;
+import org.bytedeco.javacpp.*;
+import org.bytedeco.javacpp.annotation.*;
+
+import static org.tensorflow.internal.c_api.global.tensorflow.*;
+
+// #endif
+
+@Opaque @Properties(inherit = org.tensorflow.internal.c_api.presets.tensorflow.class)
+public class TFE_ContextOptions extends org.tensorflow.internal.c_api.AbstractTFE_ContextOptions {
+    /** Empty constructor. Calls {@code super((Pointer)null)}. */
+    public TFE_ContextOptions() { super((Pointer)null); }
+    /** Pointer cast constructor. Invokes {@link Pointer#Pointer(Pointer)}. */
+    public TFE_ContextOptions(Pointer p) { super(p); }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TFE_Op.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TFE_Op.java
new file mode 100644
index 00000000000..b960617ab40
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TFE_Op.java
@@ -0,0 +1,29 @@
+// Targeted by JavaCPP version 1.5.4: DO NOT EDIT THIS FILE
+
+package org.tensorflow.internal.c_api;
+
+import java.nio.*;
+import org.bytedeco.javacpp.*;
+import org.bytedeco.javacpp.annotation.*;
+
+import static org.tensorflow.internal.c_api.global.tensorflow.*;
+
+
+// Description of the TensorFlow op to execute.
+//
+// Assumes that the provided 'ctx' outlives the returned TFE_Op, i.e.,
+// TFE_DeleteOp() is called before TFE_DeleteContext().
+//
+// Very similar to TF_OperationDescription with some differences:
+// (1) TF_Output or TFE_TensorHandle* as arguments to TF_AddInput,
+//     TF_AddInputList
+// (2) TF_ColocateWith, TF_AddControlInput etc. do not make sense.
+// (3) Implementation detail: Avoid use of NodeBuilder/NodeDefBuilder since
+//     the additional sanity checks there seem unnecessary;
+@Opaque @Properties(inherit = org.tensorflow.internal.c_api.presets.tensorflow.class)
+public class TFE_Op extends org.tensorflow.internal.c_api.AbstractTFE_Op {
+    /** Empty constructor. Calls {@code super((Pointer)null)}. */
+    public TFE_Op() { super((Pointer)null); }
+    /** Pointer cast constructor. Invokes {@link Pointer#Pointer(Pointer)}. */
+    public TFE_Op(Pointer p) { super(p); }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TFE_TensorDebugInfo.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TFE_TensorDebugInfo.java
new file mode 100644
index 00000000000..e42ebf581b8
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TFE_TensorDebugInfo.java
@@ -0,0 +1,22 @@
+// Targeted by JavaCPP version 1.5.4: DO NOT EDIT THIS FILE
+
+package org.tensorflow.internal.c_api;
+
+import java.nio.*;
+import org.bytedeco.javacpp.*;
+import org.bytedeco.javacpp.annotation.*;
+
+import static org.tensorflow.internal.c_api.global.tensorflow.*;
+
+
+// Debugging/Profiling information for TFE_TensorHandle
+//
+// TFE_TensorDebugInfo contains information useful for debugging and
+// profiling tensors.
+@Opaque @Properties(inherit = org.tensorflow.internal.c_api.presets.tensorflow.class)
+public class TFE_TensorDebugInfo extends Pointer {
+    /** Empty constructor. Calls {@code super((Pointer)null)}. */
+    public TFE_TensorDebugInfo() { super((Pointer)null); }
+    /** Pointer cast constructor. Invokes {@link Pointer#Pointer(Pointer)}. */
+    public TFE_TensorDebugInfo(Pointer p) { super(p); }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TFE_TensorHandle.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TFE_TensorHandle.java
new file mode 100644
index 00000000000..00fab08ff54
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TFE_TensorHandle.java
@@ -0,0 +1,23 @@
+// Targeted by JavaCPP version 1.5.4: DO NOT EDIT THIS FILE
+
+package org.tensorflow.internal.c_api;
+
+import java.nio.*;
+import org.bytedeco.javacpp.*;
+import org.bytedeco.javacpp.annotation.*;
+
+import static org.tensorflow.internal.c_api.global.tensorflow.*;
+
+
+// A handle to a tensor on a device.
+//
+// Like a TF_Tensor, a TFE_TensorHandle refers to a tensor with a value, shape,
+// type etc. Unlike a TF_Tensor, a TFE_TensorHandle may refer to such tensors
+// placed in memory of different devices or remote address spaces.
+@Opaque @Properties(inherit = org.tensorflow.internal.c_api.presets.tensorflow.class)
+public class TFE_TensorHandle extends org.tensorflow.internal.c_api.AbstractTFE_TensorHandle {
+    /** Empty constructor. Calls {@code super((Pointer)null)}. */
+    public TFE_TensorHandle() { super((Pointer)null); }
+    /** Pointer cast constructor. Invokes {@link Pointer#Pointer(Pointer)}. */
+    public TFE_TensorHandle(Pointer p) { super(p); }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_ApiDefMap.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_ApiDefMap.java
new file mode 100644
index 00000000000..cfe167e125a
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_ApiDefMap.java
@@ -0,0 +1,26 @@
+// Targeted by JavaCPP version 1.5.4: DO NOT EDIT THIS FILE
+
+package org.tensorflow.internal.c_api;
+
+import java.nio.*;
+import org.bytedeco.javacpp.*;
+import org.bytedeco.javacpp.annotation.*;
+
+import static org.tensorflow.internal.c_api.global.tensorflow.*;
+
+
+// TF_ApiDefMap encapsulates a collection of API definitions for an operation.
+//
+// This object maps the name of a TensorFlow operation to a description of the
+// API to generate for it, as defined by the ApiDef protocol buffer (
+// https://www.tensorflow.org/code/tensorflow/core/framework/api_def.proto)
+//
+// The ApiDef messages are typically used to generate convenience wrapper
+// functions for TensorFlow operations in various language bindings.
+@Opaque @Properties(inherit = org.tensorflow.internal.c_api.presets.tensorflow.class)
+public class TF_ApiDefMap extends Pointer {
+    /** Empty constructor. Calls {@code super((Pointer)null)}. */
+    public TF_ApiDefMap() { super((Pointer)null); }
+    /** Pointer cast constructor. Invokes {@link Pointer#Pointer(Pointer)}. */
+    public TF_ApiDefMap(Pointer p) { super(p); }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_AttrMetadata.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_AttrMetadata.java
new file mode 100644
index 00000000000..50d82309003
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_AttrMetadata.java
@@ -0,0 +1,58 @@
+// Targeted by JavaCPP version 1.5.4: DO NOT EDIT THIS FILE
+
+package org.tensorflow.internal.c_api;
+
+import java.nio.*;
+import org.bytedeco.javacpp.*;
+import org.bytedeco.javacpp.annotation.*;
+
+import static org.tensorflow.internal.c_api.global.tensorflow.*;
+
+
+// TF_AttrMetadata describes the value of an attribute on an operation.
+@Properties(inherit = org.tensorflow.internal.c_api.presets.tensorflow.class)
+public class TF_AttrMetadata extends Pointer {
+    static { Loader.load(); }
+    /** Default native constructor. */
+    public TF_AttrMetadata() { super((Pointer)null); allocate(); }
+    /** Native array allocator. Access with {@link Pointer#position(long)}. */
+    public TF_AttrMetadata(long size) { super((Pointer)null); allocateArray(size); }
+    /** Pointer cast constructor. Invokes {@link Pointer#Pointer(Pointer)}. */
+    public TF_AttrMetadata(Pointer p) { super(p); }
+    private native void allocate();
+    private native void allocateArray(long size);
+    @Override public TF_AttrMetadata position(long position) {
+        return (TF_AttrMetadata)super.position(position);
+    }
+    @Override public TF_AttrMetadata getPointer(long i) {
+        return new TF_AttrMetadata(this).position(position + i);
+    }
+
+  // A boolean: 1 if the attribute value is a list, 0 otherwise.
+  public native @Cast("unsigned char") byte is_list(); public native TF_AttrMetadata is_list(byte setter);
+
+  // Length of the list if is_list is true. Undefined otherwise.
+  public native @Cast("int64_t") long list_size(); public native TF_AttrMetadata list_size(long setter);
+
+  // Type of elements of the list if is_list != 0.
+  // Type of the single value stored in the attribute if is_list == 0.
+  public native @Cast("TF_AttrType") int type(); public native TF_AttrMetadata type(int setter);
+
+  // Total size the attribute value.
+  // The units of total_size depend on is_list and type.
+  // (1) If type == TF_ATTR_STRING and is_list == 0
+  //     then total_size is the byte size of the string
+  //     valued attribute.
+  // (2) If type == TF_ATTR_STRING and is_list == 1
+  //     then total_size is the cumulative byte size
+  //     of all the strings in the list.
+  // (3) If type == TF_ATTR_SHAPE and is_list == 0
+  //     then total_size is the number of dimensions
+  //     of the shape valued attribute, or -1
+  //     if its rank is unknown.
+  // (4) If type == TF_ATTR_SHAPE and is_list == 1
+  //     then total_size is the cumulative number
+  //     of dimensions of all shapes in the list.
+  // (5) Otherwise, total_size is undefined.
+  public native @Cast("int64_t") long total_size(); public native TF_AttrMetadata total_size(long setter);
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_Buffer.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_Buffer.java
new file mode 100644
index 00000000000..a3d80f3b50e
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_Buffer.java
@@ -0,0 +1,49 @@
+// Targeted by JavaCPP version 1.5.4: DO NOT EDIT THIS FILE
+
+package org.tensorflow.internal.c_api;
+
+import java.nio.*;
+import org.bytedeco.javacpp.*;
+import org.bytedeco.javacpp.annotation.*;
+
+import static org.tensorflow.internal.c_api.global.tensorflow.*;
+
+
+// --------------------------------------------------------------------------
+// TF_Buffer holds a pointer to a block of data and its associated length.
+// Typically, the data consists of a serialized protocol buffer, but other data
+// may also be held in a buffer.
+//
+// By default, TF_Buffer itself does not do any memory management of the
+// pointed-to block.  If need be, users of this struct should specify how to
+// deallocate the block by setting the `data_deallocator` function pointer.
+@Properties(inherit = org.tensorflow.internal.c_api.presets.tensorflow.class)
+public class TF_Buffer extends org.tensorflow.internal.c_api.AbstractTF_Buffer {
+    static { Loader.load(); }
+    /** Default native constructor. */
+    public TF_Buffer() { super((Pointer)null); allocate(); }
+    /** Native array allocator. Access with {@link Pointer#position(long)}. */
+    public TF_Buffer(long size) { super((Pointer)null); allocateArray(size); }
+    /** Pointer cast constructor. Invokes {@link Pointer#Pointer(Pointer)}. */
+    public TF_Buffer(Pointer p) { super(p); }
+    private native void allocate();
+    private native void allocateArray(long size);
+    @Override public TF_Buffer position(long position) {
+        return (TF_Buffer)super.position(position);
+    }
+    @Override public TF_Buffer getPointer(long i) {
+        return new TF_Buffer(this).position(position + i);
+    }
+
+  public native @Const Pointer data(); public native TF_Buffer data(Pointer data);
+  public native @Cast("size_t") long length(); public native TF_Buffer length(long setter);
+  public static class Data_deallocator_Pointer_long extends FunctionPointer {
+      static { Loader.load(); }
+      /** Pointer cast constructor. Invokes {@link Pointer#Pointer(Pointer)}. */
+      public    Data_deallocator_Pointer_long(Pointer p) { super(p); }
+      protected Data_deallocator_Pointer_long() { allocate(); }
+      private native void allocate();
+      public native void call(Pointer data, @Cast("size_t") long length);
+  }
+  public native Data_deallocator_Pointer_long data_deallocator(); public native TF_Buffer data_deallocator(Data_deallocator_Pointer_long setter);
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_DeprecatedSession.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_DeprecatedSession.java
new file mode 100644
index 00000000000..fdaab8327a9
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_DeprecatedSession.java
@@ -0,0 +1,23 @@
+// Targeted by JavaCPP version 1.5.4: DO NOT EDIT THIS FILE
+
+package org.tensorflow.internal.c_api;
+
+import java.nio.*;
+import org.bytedeco.javacpp.*;
+import org.bytedeco.javacpp.annotation.*;
+
+import static org.tensorflow.internal.c_api.global.tensorflow.*;
+
+
+// --------------------------------------------------------------------------
+// The deprecated session API.  Please switch to the above instead of
+// TF_ExtendGraph(). This deprecated API can be removed at any time without
+// notice.
+
+@Opaque @Properties(inherit = org.tensorflow.internal.c_api.presets.tensorflow.class)
+public class TF_DeprecatedSession extends Pointer {
+    /** Empty constructor. Calls {@code super((Pointer)null)}. */
+    public TF_DeprecatedSession() { super((Pointer)null); }
+    /** Pointer cast constructor. Invokes {@link Pointer#Pointer(Pointer)}. */
+    public TF_DeprecatedSession(Pointer p) { super(p); }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_DeviceList.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_DeviceList.java
new file mode 100644
index 00000000000..e56d93340c1
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_DeviceList.java
@@ -0,0 +1,18 @@
+// Targeted by JavaCPP version 1.5.4: DO NOT EDIT THIS FILE
+
+package org.tensorflow.internal.c_api;
+
+import java.nio.*;
+import org.bytedeco.javacpp.*;
+import org.bytedeco.javacpp.annotation.*;
+
+import static org.tensorflow.internal.c_api.global.tensorflow.*;
+
+
+@Opaque @Properties(inherit = org.tensorflow.internal.c_api.presets.tensorflow.class)
+public class TF_DeviceList extends Pointer {
+    /** Empty constructor. Calls {@code super((Pointer)null)}. */
+    public TF_DeviceList() { super((Pointer)null); }
+    /** Pointer cast constructor. Invokes {@link Pointer#Pointer(Pointer)}. */
+    public TF_DeviceList(Pointer p) { super(p); }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_DimensionHandle.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_DimensionHandle.java
new file mode 100644
index 00000000000..2eb78c52b75
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_DimensionHandle.java
@@ -0,0 +1,19 @@
+// Targeted by JavaCPP version 1.5.4: DO NOT EDIT THIS FILE
+
+package org.tensorflow.internal.c_api;
+
+import java.nio.*;
+import org.bytedeco.javacpp.*;
+import org.bytedeco.javacpp.annotation.*;
+
+import static org.tensorflow.internal.c_api.global.tensorflow.*;
+
+// #endif
+
+@Opaque @Properties(inherit = org.tensorflow.internal.c_api.presets.tensorflow.class)
+public class TF_DimensionHandle extends Pointer {
+    /** Empty constructor. Calls {@code super((Pointer)null)}. */
+    public TF_DimensionHandle() { super((Pointer)null); }
+    /** Pointer cast constructor. Invokes {@link Pointer#Pointer(Pointer)}. */
+    public TF_DimensionHandle(Pointer p) { super(p); }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_Function.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_Function.java
new file mode 100644
index 00000000000..e370b2f9f08
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_Function.java
@@ -0,0 +1,21 @@
+// Targeted by JavaCPP version 1.5.4: DO NOT EDIT THIS FILE
+
+package org.tensorflow.internal.c_api;
+
+import java.nio.*;
+import org.bytedeco.javacpp.*;
+import org.bytedeco.javacpp.annotation.*;
+
+import static org.tensorflow.internal.c_api.global.tensorflow.*;
+
+
+// TF_Function is a grouping of operations with defined inputs and outputs.
+// Once created and added to graphs, functions can be invoked by creating an
+// operation whose operation type matches the function name.
+@Opaque @Properties(inherit = org.tensorflow.internal.c_api.presets.tensorflow.class)
+public class TF_Function extends Pointer {
+    /** Empty constructor. Calls {@code super((Pointer)null)}. */
+    public TF_Function() { super((Pointer)null); }
+    /** Pointer cast constructor. Invokes {@link Pointer#Pointer(Pointer)}. */
+    public TF_Function(Pointer p) { super(p); }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_FunctionOptions.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_FunctionOptions.java
new file mode 100644
index 00000000000..5610e784a6f
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_FunctionOptions.java
@@ -0,0 +1,19 @@
+// Targeted by JavaCPP version 1.5.4: DO NOT EDIT THIS FILE
+
+package org.tensorflow.internal.c_api;
+
+import java.nio.*;
+import org.bytedeco.javacpp.*;
+import org.bytedeco.javacpp.annotation.*;
+
+import static org.tensorflow.internal.c_api.global.tensorflow.*;
+
+
+// Function definition options. TODO(iga): Define and implement
+@Opaque @Properties(inherit = org.tensorflow.internal.c_api.presets.tensorflow.class)
+public class TF_FunctionOptions extends Pointer {
+    /** Empty constructor. Calls {@code super((Pointer)null)}. */
+    public TF_FunctionOptions() { super((Pointer)null); }
+    /** Pointer cast constructor. Invokes {@link Pointer#Pointer(Pointer)}. */
+    public TF_FunctionOptions(Pointer p) { super(p); }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_Graph.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_Graph.java
new file mode 100644
index 00000000000..0a287cd5642
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_Graph.java
@@ -0,0 +1,26 @@
+// Targeted by JavaCPP version 1.5.4: DO NOT EDIT THIS FILE
+
+package org.tensorflow.internal.c_api;
+
+import java.nio.*;
+import org.bytedeco.javacpp.*;
+import org.bytedeco.javacpp.annotation.*;
+
+import static org.tensorflow.internal.c_api.global.tensorflow.*;
+
+
+// TODO(jeff,sanjay):
+// - export functions to set Config fields
+
+// --------------------------------------------------------------------------
+// The new graph construction API, still under development.
+
+// Represents a computation graph.  Graphs may be shared between sessions.
+// Graphs are thread-safe when used as directed below.
+@Opaque @Properties(inherit = org.tensorflow.internal.c_api.presets.tensorflow.class)
+public class TF_Graph extends org.tensorflow.internal.c_api.AbstractTF_Graph {
+    /** Empty constructor. Calls {@code super((Pointer)null)}. */
+    public TF_Graph() { super((Pointer)null); }
+    /** Pointer cast constructor. Invokes {@link Pointer#Pointer(Pointer)}. */
+    public TF_Graph(Pointer p) { super(p); }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_ImportGraphDefOptions.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_ImportGraphDefOptions.java
new file mode 100644
index 00000000000..442488d561d
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_ImportGraphDefOptions.java
@@ -0,0 +1,20 @@
+// Targeted by JavaCPP version 1.5.4: DO NOT EDIT THIS FILE
+
+package org.tensorflow.internal.c_api;
+
+import java.nio.*;
+import org.bytedeco.javacpp.*;
+import org.bytedeco.javacpp.annotation.*;
+
+import static org.tensorflow.internal.c_api.global.tensorflow.*;
+
+
+// TF_ImportGraphDefOptions holds options that can be passed to
+// TF_GraphImportGraphDef.
+@Opaque @Properties(inherit = org.tensorflow.internal.c_api.presets.tensorflow.class)
+public class TF_ImportGraphDefOptions extends org.tensorflow.internal.c_api.AbstractTF_ImportGraphDefOptions {
+    /** Empty constructor. Calls {@code super((Pointer)null)}. */
+    public TF_ImportGraphDefOptions() { super((Pointer)null); }
+    /** Pointer cast constructor. Invokes {@link Pointer#Pointer(Pointer)}. */
+    public TF_ImportGraphDefOptions(Pointer p) { super(p); }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_ImportGraphDefResults.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_ImportGraphDefResults.java
new file mode 100644
index 00000000000..ac800f534e1
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_ImportGraphDefResults.java
@@ -0,0 +1,20 @@
+// Targeted by JavaCPP version 1.5.4: DO NOT EDIT THIS FILE
+
+package org.tensorflow.internal.c_api;
+
+import java.nio.*;
+import org.bytedeco.javacpp.*;
+import org.bytedeco.javacpp.annotation.*;
+
+import static org.tensorflow.internal.c_api.global.tensorflow.*;
+
+
+// TF_ImportGraphDefResults holds results that are generated by
+// TF_GraphImportGraphDefWithResults().
+@Opaque @Properties(inherit = org.tensorflow.internal.c_api.presets.tensorflow.class)
+public class TF_ImportGraphDefResults extends Pointer {
+    /** Empty constructor. Calls {@code super((Pointer)null)}. */
+    public TF_ImportGraphDefResults() { super((Pointer)null); }
+    /** Pointer cast constructor. Invokes {@link Pointer#Pointer(Pointer)}. */
+    public TF_ImportGraphDefResults(Pointer p) { super(p); }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_Input.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_Input.java
new file mode 100644
index 00000000000..aa03e503d3a
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_Input.java
@@ -0,0 +1,33 @@
+// Targeted by JavaCPP version 1.5.4: DO NOT EDIT THIS FILE
+
+package org.tensorflow.internal.c_api;
+
+import java.nio.*;
+import org.bytedeco.javacpp.*;
+import org.bytedeco.javacpp.annotation.*;
+
+import static org.tensorflow.internal.c_api.global.tensorflow.*;
+
+
+// Represents a specific input of an operation.
+@Properties(inherit = org.tensorflow.internal.c_api.presets.tensorflow.class)
+public class TF_Input extends Pointer {
+    static { Loader.load(); }
+    /** Default native constructor. */
+    public TF_Input() { super((Pointer)null); allocate(); }
+    /** Native array allocator. Access with {@link Pointer#position(long)}. */
+    public TF_Input(long size) { super((Pointer)null); allocateArray(size); }
+    /** Pointer cast constructor. Invokes {@link Pointer#Pointer(Pointer)}. */
+    public TF_Input(Pointer p) { super(p); }
+    private native void allocate();
+    private native void allocateArray(long size);
+    @Override public TF_Input position(long position) {
+        return (TF_Input)super.position(position);
+    }
+    @Override public TF_Input getPointer(long i) {
+        return new TF_Input(this).position(position + i);
+    }
+
+  public native TF_Operation oper(); public native TF_Input oper(TF_Operation setter);
+  public native int index(); public native TF_Input index(int setter);  // The index of the input within oper.
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_KernelBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_KernelBuilder.java
new file mode 100644
index 00000000000..796bfa5aefc
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_KernelBuilder.java
@@ -0,0 +1,31 @@
+// Targeted by JavaCPP version 1.5.4: DO NOT EDIT THIS FILE
+
+package org.tensorflow.internal.c_api;
+
+import java.nio.*;
+import org.bytedeco.javacpp.*;
+import org.bytedeco.javacpp.annotation.*;
+
+import static org.tensorflow.internal.c_api.global.tensorflow.*;
+
+
+// --------------------------------------------------------------------------
+// C API for TensorFlow Kernels.
+//
+// This API allows developers to register custom kernel implementations for
+// TensorFlow.
+//
+// See c_api.h header comments for a discussion about API conventions.
+//
+// Users wishing to extend TensorFlow with new kernels will call
+// `TF_NewKernelBuilder`. The resulting kernel builder can be registered with
+// `TF_RegisterKernelBuilder`, which will allow TF to construct user-provided
+// kernels when necessary.
+
+@Opaque @Properties(inherit = org.tensorflow.internal.c_api.presets.tensorflow.class)
+public class TF_KernelBuilder extends Pointer {
+    /** Empty constructor. Calls {@code super((Pointer)null)}. */
+    public TF_KernelBuilder() { super((Pointer)null); }
+    /** Pointer cast constructor. Invokes {@link Pointer#Pointer(Pointer)}. */
+    public TF_KernelBuilder(Pointer p) { super(p); }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_Library.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_Library.java
new file mode 100644
index 00000000000..128efd99025
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_Library.java
@@ -0,0 +1,22 @@
+// Targeted by JavaCPP version 1.5.4: DO NOT EDIT THIS FILE
+
+package org.tensorflow.internal.c_api;
+
+import java.nio.*;
+import org.bytedeco.javacpp.*;
+import org.bytedeco.javacpp.annotation.*;
+
+import static org.tensorflow.internal.c_api.global.tensorflow.*;
+
+
+// --------------------------------------------------------------------------
+// Load plugins containing custom ops and kernels
+
+// TF_Library holds information about dynamically loaded TensorFlow plugins.
+@Opaque @Properties(inherit = org.tensorflow.internal.c_api.presets.tensorflow.class)
+public class TF_Library extends Pointer {
+    /** Empty constructor. Calls {@code super((Pointer)null)}. */
+    public TF_Library() { super((Pointer)null); }
+    /** Pointer cast constructor. Invokes {@link Pointer#Pointer(Pointer)}. */
+    public TF_Library(Pointer p) { super(p); }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_OpDefinitionBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_OpDefinitionBuilder.java
new file mode 100644
index 00000000000..ef8a6b7d7f1
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_OpDefinitionBuilder.java
@@ -0,0 +1,17 @@
+// Targeted by JavaCPP version 1.5.4: DO NOT EDIT THIS FILE
+
+package org.tensorflow.internal.c_api;
+
+import java.nio.*;
+import org.bytedeco.javacpp.*;
+import org.bytedeco.javacpp.annotation.*;
+
+import static org.tensorflow.internal.c_api.global.tensorflow.*;
+
+@Opaque @Properties(inherit = org.tensorflow.internal.c_api.presets.tensorflow.class)
+public class TF_OpDefinitionBuilder extends Pointer {
+    /** Empty constructor. Calls {@code super((Pointer)null)}. */
+    public TF_OpDefinitionBuilder() { super((Pointer)null); }
+    /** Pointer cast constructor. Invokes {@link Pointer#Pointer(Pointer)}. */
+    public TF_OpDefinitionBuilder(Pointer p) { super(p); }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_OpKernelConstruction.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_OpKernelConstruction.java
new file mode 100644
index 00000000000..6a984f2a25d
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_OpKernelConstruction.java
@@ -0,0 +1,17 @@
+// Targeted by JavaCPP version 1.5.4: DO NOT EDIT THIS FILE
+
+package org.tensorflow.internal.c_api;
+
+import java.nio.*;
+import org.bytedeco.javacpp.*;
+import org.bytedeco.javacpp.annotation.*;
+
+import static org.tensorflow.internal.c_api.global.tensorflow.*;
+
+@Opaque @Properties(inherit = org.tensorflow.internal.c_api.presets.tensorflow.class)
+public class TF_OpKernelConstruction extends Pointer {
+    /** Empty constructor. Calls {@code super((Pointer)null)}. */
+    public TF_OpKernelConstruction() { super((Pointer)null); }
+    /** Pointer cast constructor. Invokes {@link Pointer#Pointer(Pointer)}. */
+    public TF_OpKernelConstruction(Pointer p) { super(p); }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_OpKernelContext.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_OpKernelContext.java
new file mode 100644
index 00000000000..9c145e89bdf
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_OpKernelContext.java
@@ -0,0 +1,17 @@
+// Targeted by JavaCPP version 1.5.4: DO NOT EDIT THIS FILE
+
+package org.tensorflow.internal.c_api;
+
+import java.nio.*;
+import org.bytedeco.javacpp.*;
+import org.bytedeco.javacpp.annotation.*;
+
+import static org.tensorflow.internal.c_api.global.tensorflow.*;
+
+@Opaque @Properties(inherit = org.tensorflow.internal.c_api.presets.tensorflow.class)
+public class TF_OpKernelContext extends Pointer {
+    /** Empty constructor. Calls {@code super((Pointer)null)}. */
+    public TF_OpKernelContext() { super((Pointer)null); }
+    /** Pointer cast constructor. Invokes {@link Pointer#Pointer(Pointer)}. */
+    public TF_OpKernelContext(Pointer p) { super(p); }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_Operation.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_Operation.java
new file mode 100644
index 00000000000..0cd1c90ecae
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_Operation.java
@@ -0,0 +1,21 @@
+// Targeted by JavaCPP version 1.5.4: DO NOT EDIT THIS FILE
+
+package org.tensorflow.internal.c_api;
+
+import java.nio.*;
+import org.bytedeco.javacpp.*;
+import org.bytedeco.javacpp.annotation.*;
+
+import static org.tensorflow.internal.c_api.global.tensorflow.*;
+
+
+// Operation that has been added to the graph. Valid until the graph is
+// deleted -- in particular adding a new operation to the graph does not
+// invalidate old TF_Operation* pointers.
+@Opaque @Properties(inherit = org.tensorflow.internal.c_api.presets.tensorflow.class)
+public class TF_Operation extends Pointer {
+    /** Empty constructor. Calls {@code super((Pointer)null)}. */
+    public TF_Operation() { super((Pointer)null); }
+    /** Pointer cast constructor. Invokes {@link Pointer#Pointer(Pointer)}. */
+    public TF_Operation(Pointer p) { super(p); }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_OperationDescription.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_OperationDescription.java
new file mode 100644
index 00000000000..64a577c28de
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_OperationDescription.java
@@ -0,0 +1,19 @@
+// Targeted by JavaCPP version 1.5.4: DO NOT EDIT THIS FILE
+
+package org.tensorflow.internal.c_api;
+
+import java.nio.*;
+import org.bytedeco.javacpp.*;
+import org.bytedeco.javacpp.annotation.*;
+
+import static org.tensorflow.internal.c_api.global.tensorflow.*;
+
+
+// Operation being built. The underlying graph must outlive this.
+@Opaque @Properties(inherit = org.tensorflow.internal.c_api.presets.tensorflow.class)
+public class TF_OperationDescription extends Pointer {
+    /** Empty constructor. Calls {@code super((Pointer)null)}. */
+    public TF_OperationDescription() { super((Pointer)null); }
+    /** Pointer cast constructor. Invokes {@link Pointer#Pointer(Pointer)}. */
+    public TF_OperationDescription(Pointer p) { super(p); }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_Output.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_Output.java
new file mode 100644
index 00000000000..b990302d373
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_Output.java
@@ -0,0 +1,33 @@
+// Targeted by JavaCPP version 1.5.4: DO NOT EDIT THIS FILE
+
+package org.tensorflow.internal.c_api;
+
+import java.nio.*;
+import org.bytedeco.javacpp.*;
+import org.bytedeco.javacpp.annotation.*;
+
+import static org.tensorflow.internal.c_api.global.tensorflow.*;
+
+
+// Represents a specific output of an operation.
+@Properties(inherit = org.tensorflow.internal.c_api.presets.tensorflow.class)
+public class TF_Output extends Pointer {
+    static { Loader.load(); }
+    /** Default native constructor. */
+    public TF_Output() { super((Pointer)null); allocate(); }
+    /** Native array allocator. Access with {@link Pointer#position(long)}. */
+    public TF_Output(long size) { super((Pointer)null); allocateArray(size); }
+    /** Pointer cast constructor. Invokes {@link Pointer#Pointer(Pointer)}. */
+    public TF_Output(Pointer p) { super(p); }
+    private native void allocate();
+    private native void allocateArray(long size);
+    @Override public TF_Output position(long position) {
+        return (TF_Output)super.position(position);
+    }
+    @Override public TF_Output getPointer(long i) {
+        return new TF_Output(this).position(position + i);
+    }
+
+  public native TF_Operation oper(); public native TF_Output oper(TF_Operation setter);
+  public native int index(); public native TF_Output index(int setter);  // The index of the output within oper.
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_Server.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_Server.java
new file mode 100644
index 00000000000..c34d6dd2eab
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_Server.java
@@ -0,0 +1,27 @@
+// Targeted by JavaCPP version 1.5.4: DO NOT EDIT THIS FILE
+
+package org.tensorflow.internal.c_api;
+
+import java.nio.*;
+import org.bytedeco.javacpp.*;
+import org.bytedeco.javacpp.annotation.*;
+
+import static org.tensorflow.internal.c_api.global.tensorflow.*;
+
+
+// --------------------------------------------------------------------------
+// In-process TensorFlow server functionality, for use in distributed training.
+// A Server instance encapsulates a set of devices and a Session target that
+// can participate in distributed training. A server belongs to a cluster
+// (specified by a ClusterSpec), and corresponds to a particular task in a
+// named job. The server can communicate with any other server in the same
+// cluster.
+
+// In-process TensorFlow server.
+@Opaque @Properties(inherit = org.tensorflow.internal.c_api.presets.tensorflow.class)
+public class TF_Server extends Pointer {
+    /** Empty constructor. Calls {@code super((Pointer)null)}. */
+    public TF_Server() { super((Pointer)null); }
+    /** Pointer cast constructor. Invokes {@link Pointer#Pointer(Pointer)}. */
+    public TF_Server(Pointer p) { super(p); }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_Session.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_Session.java
new file mode 100644
index 00000000000..10818f6b59f
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_Session.java
@@ -0,0 +1,24 @@
+// Targeted by JavaCPP version 1.5.4: DO NOT EDIT THIS FILE
+
+package org.tensorflow.internal.c_api;
+
+import java.nio.*;
+import org.bytedeco.javacpp.*;
+import org.bytedeco.javacpp.annotation.*;
+
+import static org.tensorflow.internal.c_api.global.tensorflow.*;
+
+
+// TODO(josh11b): Register OpDef, available to all operations added
+// to this graph.
+
+// --------------------------------------------------------------------------
+// API for driving Graph execution.
+
+@Opaque @Properties(inherit = org.tensorflow.internal.c_api.presets.tensorflow.class)
+public class TF_Session extends org.tensorflow.internal.c_api.AbstractTF_Session {
+    /** Empty constructor. Calls {@code super((Pointer)null)}. */
+    public TF_Session() { super((Pointer)null); }
+    /** Pointer cast constructor. Invokes {@link Pointer#Pointer(Pointer)}. */
+    public TF_Session(Pointer p) { super(p); }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_SessionOptions.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_SessionOptions.java
new file mode 100644
index 00000000000..c96017ab554
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_SessionOptions.java
@@ -0,0 +1,20 @@
+// Targeted by JavaCPP version 1.5.4: DO NOT EDIT THIS FILE
+
+package org.tensorflow.internal.c_api;
+
+import java.nio.*;
+import org.bytedeco.javacpp.*;
+import org.bytedeco.javacpp.annotation.*;
+
+import static org.tensorflow.internal.c_api.global.tensorflow.*;
+
+
+// --------------------------------------------------------------------------
+// TF_SessionOptions holds options that can be passed during session creation.
+@Opaque @Properties(inherit = org.tensorflow.internal.c_api.presets.tensorflow.class)
+public class TF_SessionOptions extends org.tensorflow.internal.c_api.AbstractTF_SessionOptions {
+    /** Empty constructor. Calls {@code super((Pointer)null)}. */
+    public TF_SessionOptions() { super((Pointer)null); }
+    /** Pointer cast constructor. Invokes {@link Pointer#Pointer(Pointer)}. */
+    public TF_SessionOptions(Pointer p) { super(p); }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_ShapeHandle.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_ShapeHandle.java
new file mode 100644
index 00000000000..7fc4dd37276
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_ShapeHandle.java
@@ -0,0 +1,17 @@
+// Targeted by JavaCPP version 1.5.4: DO NOT EDIT THIS FILE
+
+package org.tensorflow.internal.c_api;
+
+import java.nio.*;
+import org.bytedeco.javacpp.*;
+import org.bytedeco.javacpp.annotation.*;
+
+import static org.tensorflow.internal.c_api.global.tensorflow.*;
+
+@Opaque @Properties(inherit = org.tensorflow.internal.c_api.presets.tensorflow.class)
+public class TF_ShapeHandle extends Pointer {
+    /** Empty constructor. Calls {@code super((Pointer)null)}. */
+    public TF_ShapeHandle() { super((Pointer)null); }
+    /** Pointer cast constructor. Invokes {@link Pointer#Pointer(Pointer)}. */
+    public TF_ShapeHandle(Pointer p) { super(p); }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_ShapeInferenceContext.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_ShapeInferenceContext.java
new file mode 100644
index 00000000000..4d2bd0c7441
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_ShapeInferenceContext.java
@@ -0,0 +1,17 @@
+// Targeted by JavaCPP version 1.5.4: DO NOT EDIT THIS FILE
+
+package org.tensorflow.internal.c_api;
+
+import java.nio.*;
+import org.bytedeco.javacpp.*;
+import org.bytedeco.javacpp.annotation.*;
+
+import static org.tensorflow.internal.c_api.global.tensorflow.*;
+
+@Opaque @Properties(inherit = org.tensorflow.internal.c_api.presets.tensorflow.class)
+public class TF_ShapeInferenceContext extends Pointer {
+    /** Empty constructor. Calls {@code super((Pointer)null)}. */
+    public TF_ShapeInferenceContext() { super((Pointer)null); }
+    /** Pointer cast constructor. Invokes {@link Pointer#Pointer(Pointer)}. */
+    public TF_ShapeInferenceContext(Pointer p) { super(p); }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_Status.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_Status.java
new file mode 100644
index 00000000000..0ea4e3415c3
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_Status.java
@@ -0,0 +1,19 @@
+// Targeted by JavaCPP version 1.5.4: DO NOT EDIT THIS FILE
+
+package org.tensorflow.internal.c_api;
+
+import java.nio.*;
+import org.bytedeco.javacpp.*;
+import org.bytedeco.javacpp.annotation.*;
+
+import static org.tensorflow.internal.c_api.global.tensorflow.*;
+
+// #endif
+
+@Opaque @Properties(inherit = org.tensorflow.internal.c_api.presets.tensorflow.class)
+public class TF_Status extends org.tensorflow.internal.c_api.AbstractTF_Status {
+    /** Empty constructor. Calls {@code super((Pointer)null)}. */
+    public TF_Status() { super((Pointer)null); }
+    /** Pointer cast constructor. Invokes {@link Pointer#Pointer(Pointer)}. */
+    public TF_Status(Pointer p) { super(p); }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_Tensor.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_Tensor.java
new file mode 100644
index 00000000000..b1839dadd2d
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_Tensor.java
@@ -0,0 +1,36 @@
+// Targeted by JavaCPP version 1.5.4: DO NOT EDIT THIS FILE
+
+package org.tensorflow.internal.c_api;
+
+import java.nio.*;
+import org.bytedeco.javacpp.*;
+import org.bytedeco.javacpp.annotation.*;
+
+import static org.tensorflow.internal.c_api.global.tensorflow.*;
+
+// #endif
+
+// --------------------------------------------------------------------------
+// TF_Tensor holds a multi-dimensional array of elements of a single data type.
+// For all types other than TF_STRING, the data buffer stores elements
+// in row major order.  E.g. if data is treated as a vector of TF_DataType:
+//
+//   element 0:   index (0, ..., 0)
+//   element 1:   index (0, ..., 1)
+//   ...
+//
+// The format for TF_STRING tensors is:
+//   start_offset: array[uint64]
+//   data:         byte[...]
+//
+//   The string length (as a varint, start_offset[i + 1] - start_offset[i]),
+//   followed by the contents of the string is encoded at data[start_offset[i]].
+//   TF_StringEncode and TF_StringDecode facilitate this encoding.
+
+@Opaque @Properties(inherit = org.tensorflow.internal.c_api.presets.tensorflow.class)
+public class TF_Tensor extends org.tensorflow.internal.c_api.AbstractTF_Tensor {
+    /** Empty constructor. Calls {@code super((Pointer)null)}. */
+    public TF_Tensor() { super((Pointer)null); }
+    /** Pointer cast constructor. Invokes {@link Pointer#Pointer(Pointer)}. */
+    public TF_Tensor(Pointer p) { super(p); }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_WhileParams.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_WhileParams.java
new file mode 100644
index 00000000000..4c6dc486d52
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/TF_WhileParams.java
@@ -0,0 +1,37 @@
+// Targeted by JavaCPP version 1.5.4: DO NOT EDIT THIS FILE
+
+package org.tensorflow.internal.c_api;
+
+import java.nio.*;
+import org.bytedeco.javacpp.*;
+import org.bytedeco.javacpp.annotation.*;
+
+import static org.tensorflow.internal.c_api.global.tensorflow.*;
+
+
+@Properties(inherit = org.tensorflow.internal.c_api.presets.tensorflow.class)
+public class TF_WhileParams extends Pointer {
+    static { Loader.load(); }
+    /** Pointer cast constructor. Invokes {@link Pointer#Pointer(Pointer)}. */
+    public TF_WhileParams(Pointer p) { super(p); }
+
+  // The number of inputs to the while loop, i.e. the number of loop variables.
+  // This is the size of cond_inputs, body_inputs, and body_outputs.
+  @MemberGetter public native int ninputs();
+
+  // The while condition graph. The inputs are the current values of the loop
+  // variables. The output should be a scalar boolean.
+  @MemberGetter public native TF_Graph cond_graph();
+  @MemberGetter public native @Const TF_Output cond_inputs();
+  public native @ByRef TF_Output cond_output(); public native TF_WhileParams cond_output(TF_Output setter);
+
+  // The loop body graph. The inputs are the current values of the loop
+  // variables. The outputs are the updated values of the loop variables.
+  @MemberGetter public native TF_Graph body_graph();
+  @MemberGetter public native @Const TF_Output body_inputs();
+  @MemberGetter public native TF_Output body_outputs();
+
+  // Unique null-terminated name for this while loop. This is used as a prefix
+  // for created operations.
+  public native @Cast("const char*") BytePointer name(); public native TF_WhileParams name(BytePointer setter);
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/Tensor.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/Tensor.java
new file mode 100644
index 00000000000..0a9efa65762
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/Tensor.java
@@ -0,0 +1,25 @@
+// Targeted by JavaCPP version 1.5.4: DO NOT EDIT THIS FILE
+
+package org.tensorflow.internal.c_api;
+
+import java.nio.*;
+import org.bytedeco.javacpp.*;
+import org.bytedeco.javacpp.annotation.*;
+
+import static org.tensorflow.internal.c_api.global.tensorflow.*;
+ /* end extern "C" */
+// #endif
+
+// #ifdef __cplusplus
+// A workaround to ease conversion to and from numpy objects and
+// TFE_TensorHandle's.
+//
+// TODO(ashankar): Figure out an alternative scheme that precludes the need for
+// these API-boundary breaking methods.
+@Namespace("tensorflow") @Opaque @Properties(inherit = org.tensorflow.internal.c_api.presets.tensorflow.class)
+public class Tensor extends Pointer {
+    /** Empty constructor. Calls {@code super((Pointer)null)}. */
+    public Tensor() { super((Pointer)null); }
+    /** Pointer cast constructor. Invokes {@link Pointer#Pointer(Pointer)}. */
+    public Tensor(Pointer p) { super(p); }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/global/tensorflow.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/global/tensorflow.java
new file mode 100644
index 00000000000..bdd9cc618ac
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/internal/c_api/global/tensorflow.java
@@ -0,0 +1,4021 @@
+// Targeted by JavaCPP version 1.5.4: DO NOT EDIT THIS FILE
+
+package org.tensorflow.internal.c_api.global;
+
+import org.tensorflow.internal.c_api.*;
+
+import java.nio.*;
+import org.bytedeco.javacpp.*;
+import org.bytedeco.javacpp.annotation.*;
+
+public class tensorflow extends org.tensorflow.internal.c_api.presets.tensorflow {
+    static { Loader.load(); }
+
+// Parsed from tensorflow/core/util/port.h
+
+/* Copyright 2015 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+==============================================================================*/
+
+// #ifndef TENSORFLOW_CORE_UTIL_PORT_H_
+// #define TENSORFLOW_CORE_UTIL_PORT_H_
+
+// Returns true if GOOGLE_CUDA is defined.
+@Namespace("tensorflow") public static native @Cast("bool") boolean IsGoogleCudaEnabled();
+
+// Returns true if TENSORFLOW_USE_ROCM is defined. (i.e. TF is built with ROCm)
+@Namespace("tensorflow") public static native @Cast("bool") boolean IsBuiltWithROCm();
+
+// Returns true if TENSORFLOW_USE_XLA is defined. (i.e. TF is built with XLA)
+@Namespace("tensorflow") public static native @Cast("bool") boolean IsBuiltWithXLA();
+
+// Returns true if TENSORFLOW_USE_NVCC is defined. (i.e. TF is built with nvcc)
+@Namespace("tensorflow") public static native @Cast("bool") boolean IsBuiltWithNvcc();
+
+// Returns true if either
+//
+//   GOOGLE_CUDA is defined, and the given CUDA version supports
+//   half-precision matrix multiplications and convolution operations.
+//
+//     OR
+//
+//   TENSORFLOW_USE_ROCM is defined
+//
+@Namespace("tensorflow") public static native @Cast("bool") boolean GpuSupportsHalfMatMulAndConv();
+
+// Returns true if INTEL_MKL is defined
+@Namespace("tensorflow") public static native @Cast("bool") boolean IsMklEnabled();
+
+  // end namespace tensorflow
+
+// #endif  // TENSORFLOW_CORE_UTIL_PORT_H_
+
+
+// Parsed from tensorflow/c/tf_attrtype.h
+
+/* Copyright 2019 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+==============================================================================*/
+// #ifndef TENSORFLOW_C_TF_ATTRTYPE_H_
+// #define TENSORFLOW_C_TF_ATTRTYPE_H_
+
+// #ifdef __cplusplus
+// #endif
+
+// TF_AttrType describes the type of the value of an attribute on an operation.
+/** enum TF_AttrType */
+public static final int
+  TF_ATTR_STRING = 0,
+  TF_ATTR_INT = 1,
+  TF_ATTR_FLOAT = 2,
+  TF_ATTR_BOOL = 3,
+  TF_ATTR_TYPE = 4,
+  TF_ATTR_SHAPE = 5,
+  TF_ATTR_TENSOR = 6,
+  TF_ATTR_PLACEHOLDER = 7,
+  TF_ATTR_FUNC = 8;
+
+// #ifdef __cplusplus /* end extern "C" */
+// #endif
+
+// #endif  // TENSORFLOW_C_TF_ATTRTYPE_H_
+
+
+// Parsed from tensorflow/c/tf_datatype.h
+
+/* Copyright 2019 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+==============================================================================*/
+
+// #ifndef TENSORFLOW_C_TF_DATATYPE_H_
+// #define TENSORFLOW_C_TF_DATATYPE_H_
+
+// #include <stddef.h>
+
+// Macro to control visibility of exported symbols in the shared library (.so,
+// .dylib, .dll).
+// This duplicates the TF_EXPORT macro definition in
+// tensorflow/core/platform/macros.h in order to keep this .h file independent
+// of any other includes.
+// #ifdef SWIG
+// #define TF_CAPI_EXPORT
+// #else
+// #if defined(_WIN32)
+// #ifdef TF_COMPILE_LIBRARY
+// #define TF_CAPI_EXPORT __declspec(dllexport)
+// #else
+// #define TF_CAPI_EXPORT __declspec(dllimport)
+// #endif  // TF_COMPILE_LIBRARY
+// #else
+// #define TF_CAPI_EXPORT __attribute__((visibility("default")))
+// #endif  // _WIN32
+// #endif  // SWIG
+
+// #ifdef __cplusplus
+// #endif
+
+// --------------------------------------------------------------------------
+// TF_DataType holds the type for a scalar value.  E.g., one slot in a tensor.
+// The enum values here are identical to corresponding values in types.proto.
+/** enum TF_DataType */
+public static final int
+  TF_FLOAT = 1,
+  TF_DOUBLE = 2,
+  TF_INT32 = 3,  // Int32 tensors are always in 'host' memory.
+  TF_UINT8 = 4,
+  TF_INT16 = 5,
+  TF_INT8 = 6,
+  TF_STRING = 7,
+  TF_COMPLEX64 = 8,  // Single-precision complex
+  TF_COMPLEX = 8,    // Old identifier kept for API backwards compatibility
+  TF_INT64 = 9,
+  TF_BOOL = 10,
+  TF_QINT8 = 11,     // Quantized int8
+  TF_QUINT8 = 12,    // Quantized uint8
+  TF_QINT32 = 13,    // Quantized int32
+  TF_BFLOAT16 = 14,  // Float32 truncated to 16 bits.  Only for cast ops.
+  TF_QINT16 = 15,    // Quantized int16
+  TF_QUINT16 = 16,   // Quantized uint16
+  TF_UINT16 = 17,
+  TF_COMPLEX128 = 18,  // Double-precision complex
+  TF_HALF = 19,
+  TF_RESOURCE = 20,
+  TF_VARIANT = 21,
+  TF_UINT32 = 22,
+  TF_UINT64 = 23;
+
+// TF_DataTypeSize returns the sizeof() for the underlying type corresponding
+// to the given TF_DataType enum value. Returns 0 for variable length types
+// (eg. TF_STRING) or on failure.
+public static native @Cast("size_t") long TF_DataTypeSize(@Cast("TF_DataType") int dt);
+
+// #ifdef __cplusplus /* end extern "C" */
+// #endif
+
+// #endif  // TENSORFLOW_C_TF_DATATYPE_H_
+
+
+// Parsed from tensorflow/c/tf_status.h
+
+/* Copyright 2019 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+==============================================================================*/
+
+// #ifndef TENSORFLOW_C_TF_STATUS_H_
+// #define TENSORFLOW_C_TF_STATUS_H_
+
+// #ifdef SWIG
+// #define TF_CAPI_EXPORT
+// #else
+// #if defined(_WIN32)
+// #ifdef TF_COMPILE_LIBRARY
+// #define TF_CAPI_EXPORT __declspec(dllexport)
+// #else
+// #define TF_CAPI_EXPORT __declspec(dllimport)
+// #endif  // TF_COMPILE_LIBRARY
+// #else
+// #define TF_CAPI_EXPORT __attribute__((visibility("default")))
+// #endif  // _WIN32
+// #endif  // SWIG
+
+// #ifdef __cplusplus
+// Targeting ../TF_Status.java
+
+
+
+// --------------------------------------------------------------------------
+// TF_Code holds an error code.  The enum values here are identical to
+// corresponding values in error_codes.proto.
+/** enum TF_Code */
+public static final int
+  TF_OK = 0,
+  TF_CANCELLED = 1,
+  TF_UNKNOWN = 2,
+  TF_INVALID_ARGUMENT = 3,
+  TF_DEADLINE_EXCEEDED = 4,
+  TF_NOT_FOUND = 5,
+  TF_ALREADY_EXISTS = 6,
+  TF_PERMISSION_DENIED = 7,
+  TF_UNAUTHENTICATED = 16,
+  TF_RESOURCE_EXHAUSTED = 8,
+  TF_FAILED_PRECONDITION = 9,
+  TF_ABORTED = 10,
+  TF_OUT_OF_RANGE = 11,
+  TF_UNIMPLEMENTED = 12,
+  TF_INTERNAL = 13,
+  TF_UNAVAILABLE = 14,
+  TF_DATA_LOSS = 15;
+
+// --------------------------------------------------------------------------
+
+// Return a new status object.
+public static native TF_Status TF_NewStatus();
+
+// Delete a previously created status object.
+public static native void TF_DeleteStatus(TF_Status arg0);
+
+// Record <code, msg> in *s.  Any previous information is lost.
+// A common use is to clear a status: TF_SetStatus(s, TF_OK, "");
+public static native void TF_SetStatus(TF_Status s, @Cast("TF_Code") int code,
+                                        @Cast("const char*") BytePointer msg);
+public static native void TF_SetStatus(TF_Status s, @Cast("TF_Code") int code,
+                                        String msg);
+
+// Convert from an I/O error code (e.g., errno) to a TF_Status value.
+// Any previous information is lost. Prefer to use this instead of TF_SetStatus
+// when the error comes from I/O operations.
+public static native void TF_SetStatusFromIOError(TF_Status s, int error_code,
+                                                   @Cast("const char*") BytePointer context);
+public static native void TF_SetStatusFromIOError(TF_Status s, int error_code,
+                                                   String context);
+
+// Return the code record in *s.
+public static native @Cast("TF_Code") int TF_GetCode(@Const TF_Status s);
+
+// Return a pointer to the (null-terminated) error message in *s.  The
+// return value points to memory that is only usable until the next
+// mutation to *s.  Always returns an empty string if TF_GetCode(s) is
+// TF_OK.
+public static native @Cast("const char*") BytePointer TF_Message(@Const TF_Status s);
+
+// #ifdef __cplusplus /* end extern "C" */
+// #endif
+
+// #endif  // TENSORFLOW_C_TF_STATUS_H_
+
+
+// Parsed from tensorflow/c/tf_tensor.h
+
+/* Copyright 2019 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+==============================================================================*/
+
+// #ifndef TENSORFLOW_C_TF_TENSOR_H_
+// #define TENSORFLOW_C_TF_TENSOR_H_
+
+// #include <stdbool.h>
+// #include <stdint.h>
+
+// #include "tensorflow/c/tf_datatype.h"
+// #include "tensorflow/c/tf_status.h"
+
+// Macro to control visibility of exported symbols in the shared library (.so,
+// .dylib, .dll).
+// This duplicates the TF_EXPORT macro definition in
+// tensorflow/core/platform/macros.h in order to keep this .h file independent
+// of any other includes.
+// #ifdef SWIG
+// #define TF_CAPI_EXPORT
+// #else
+// #if defined(_WIN32)
+// #ifdef TF_COMPILE_LIBRARY
+// #define TF_CAPI_EXPORT __declspec(dllexport)
+// #else
+// #define TF_CAPI_EXPORT __declspec(dllimport)
+// #endif  // TF_COMPILE_LIBRARY
+// #else
+// #define TF_CAPI_EXPORT __attribute__((visibility("default")))
+// #endif  // _WIN32
+// #endif  // SWIG
+
+// #ifdef __cplusplus
+// Targeting ../TF_Tensor.java
+
+
+// Targeting ../Deallocator_Pointer_long_Pointer.java
+
+
+public static native TF_Tensor TF_NewTensor(
+    @Cast("TF_DataType") int arg0, @Cast("const int64_t*") LongPointer dims, int num_dims, Pointer data, @Cast("size_t") long len,
+    Deallocator_Pointer_long_Pointer deallocator,
+    Pointer deallocator_arg);
+public static native TF_Tensor TF_NewTensor(
+    @Cast("TF_DataType") int arg0, @Cast("const int64_t*") LongBuffer dims, int num_dims, Pointer data, @Cast("size_t") long len,
+    Deallocator_Pointer_long_Pointer deallocator,
+    Pointer deallocator_arg);
+public static native TF_Tensor TF_NewTensor(
+    @Cast("TF_DataType") int arg0, @Cast("const int64_t*") long[] dims, int num_dims, Pointer data, @Cast("size_t") long len,
+    Deallocator_Pointer_long_Pointer deallocator,
+    Pointer deallocator_arg);
+
+// Allocate and return a new Tensor.
+//
+// This function is an alternative to TF_NewTensor and should be used when
+// memory is allocated to pass the Tensor to the C API. The allocated memory
+// satisfies TensorFlow's memory alignment preferences and should be preferred
+// over calling malloc and free.
+//
+// The caller must set the Tensor values by writing them to the pointer returned
+// by TF_TensorData with length TF_TensorByteSize.
+public static native TF_Tensor TF_AllocateTensor(@Cast("TF_DataType") int arg0,
+                                                   @Cast("const int64_t*") LongPointer dims,
+                                                   int num_dims, @Cast("size_t") long len);
+public static native TF_Tensor TF_AllocateTensor(@Cast("TF_DataType") int arg0,
+                                                   @Cast("const int64_t*") LongBuffer dims,
+                                                   int num_dims, @Cast("size_t") long len);
+public static native TF_Tensor TF_AllocateTensor(@Cast("TF_DataType") int arg0,
+                                                   @Cast("const int64_t*") long[] dims,
+                                                   int num_dims, @Cast("size_t") long len);
+
+// Deletes `tensor` and returns a new TF_Tensor with the same content if
+// possible. Returns nullptr and leaves `tensor` untouched if not.
+public static native TF_Tensor TF_TensorMaybeMove(TF_Tensor tensor);
+
+// Destroy a tensor.
+public static native void TF_DeleteTensor(TF_Tensor arg0);
+
+// Return the type of a tensor element.
+public static native @Cast("TF_DataType") int TF_TensorType(@Const TF_Tensor arg0);
+
+// Return the number of dimensions that the tensor has.
+public static native int TF_NumDims(@Const TF_Tensor arg0);
+
+// Return the length of the tensor in the "dim_index" dimension.
+// REQUIRES: 0 <= dim_index < TF_NumDims(tensor)
+public static native @Cast("int64_t") long TF_Dim(@Const TF_Tensor tensor, int dim_index);
+
+// Return the size of the underlying data in bytes.
+public static native @Cast("size_t") long TF_TensorByteSize(@Const TF_Tensor arg0);
+
+// Return a pointer to the underlying data buffer.
+public static native Pointer TF_TensorData(@Const TF_Tensor arg0);
+
+// Returns the number of elements in the tensor.
+public static native @Cast("int64_t") long TF_TensorElementCount(@Const TF_Tensor tensor);
+
+// Copy the internal data representation of `from` to `to`. `new_dims` and
+// `num_new_dims` specify the new shape of the `to` tensor, `type` specifies its
+// data type. On success, *status is set to TF_OK and the two tensors share the
+// same data buffer.
+//
+// This call requires that the `from` tensor and the given type and shape (dims
+// and num_dims) are "compatible" (i.e. they occupy the same number of bytes).
+// Specifically, given from_type_size = TF_DataTypeSize(TF_TensorType(from)):
+//
+// ShapeElementCount(dims, num_dims) * TF_DataTypeSize(type)
+//
+// must equal
+//
+// TF_TensorElementCount(from) * from_type_size
+//
+// where TF_ShapeElementCount would be the number of elements in a tensor with
+// the given shape.
+//
+// In addition, this function requires:
+//   * TF_DataTypeSize(TF_TensorType(from)) != 0
+//   * TF_DataTypeSize(type) != 0
+//
+// If any of the requirements are not met, *status is set to
+// TF_INVALID_ARGUMENT.
+public static native void TF_TensorBitcastFrom(@Const TF_Tensor from,
+                                                @Cast("TF_DataType") int type, TF_Tensor to,
+                                                @Cast("const int64_t*") LongPointer new_dims,
+                                                int num_new_dims,
+                                                TF_Status status);
+public static native void TF_TensorBitcastFrom(@Const TF_Tensor from,
+                                                @Cast("TF_DataType") int type, TF_Tensor to,
+                                                @Cast("const int64_t*") LongBuffer new_dims,
+                                                int num_new_dims,
+                                                TF_Status status);
+public static native void TF_TensorBitcastFrom(@Const TF_Tensor from,
+                                                @Cast("TF_DataType") int type, TF_Tensor to,
+                                                @Cast("const int64_t*") long[] new_dims,
+                                                int num_new_dims,
+                                                TF_Status status);
+
+// --------------------------------------------------------------------------
+// Encode the string `src` (`src_len` bytes long) into `dst` in the format
+// required by TF_STRING tensors. Does not write to memory more than `dst_len`
+// bytes beyond `*dst`. `dst_len` should be at least
+// TF_StringEncodedSize(src_len).
+//
+// On success returns the size in bytes of the encoded string.
+// Returns an error into `status` otherwise.
+public static native @Cast("size_t") long TF_StringEncode(@Cast("const char*") BytePointer src, @Cast("size_t") long src_len,
+                                             @Cast("char*") BytePointer dst, @Cast("size_t") long dst_len,
+                                             TF_Status status);
+public static native @Cast("size_t") long TF_StringEncode(String src, @Cast("size_t") long src_len,
+                                             @Cast("char*") ByteBuffer dst, @Cast("size_t") long dst_len,
+                                             TF_Status status);
+public static native @Cast("size_t") long TF_StringEncode(@Cast("const char*") BytePointer src, @Cast("size_t") long src_len,
+                                             @Cast("char*") byte[] dst, @Cast("size_t") long dst_len,
+                                             TF_Status status);
+public static native @Cast("size_t") long TF_StringEncode(String src, @Cast("size_t") long src_len,
+                                             @Cast("char*") BytePointer dst, @Cast("size_t") long dst_len,
+                                             TF_Status status);
+public static native @Cast("size_t") long TF_StringEncode(@Cast("const char*") BytePointer src, @Cast("size_t") long src_len,
+                                             @Cast("char*") ByteBuffer dst, @Cast("size_t") long dst_len,
+                                             TF_Status status);
+public static native @Cast("size_t") long TF_StringEncode(String src, @Cast("size_t") long src_len,
+                                             @Cast("char*") byte[] dst, @Cast("size_t") long dst_len,
+                                             TF_Status status);
+
+// Decode a string encoded using TF_StringEncode.
+//
+// On success, sets `*dst` to the start of the decoded string and `*dst_len` to
+// its length. Returns the number of bytes starting at `src` consumed while
+// decoding. `*dst` points to memory within the encoded buffer.  On failure,
+// `*dst` and `*dst_len` are undefined and an error is set in `status`.
+//
+// Does not read memory more than `src_len` bytes beyond `src`.
+public static native @Cast("size_t") long TF_StringDecode(@Cast("const char*") BytePointer src, @Cast("size_t") long src_len,
+                                             @Cast("const char**") PointerPointer dst, @Cast("size_t*") SizeTPointer dst_len,
+                                             TF_Status status);
+public static native @Cast("size_t") long TF_StringDecode(@Cast("const char*") BytePointer src, @Cast("size_t") long src_len,
+                                             @Cast("const char**") @ByPtrPtr BytePointer dst, @Cast("size_t*") SizeTPointer dst_len,
+                                             TF_Status status);
+public static native @Cast("size_t") long TF_StringDecode(String src, @Cast("size_t") long src_len,
+                                             @Cast("const char**") @ByPtrPtr ByteBuffer dst, @Cast("size_t*") SizeTPointer dst_len,
+                                             TF_Status status);
+public static native @Cast("size_t") long TF_StringDecode(@Cast("const char*") BytePointer src, @Cast("size_t") long src_len,
+                                             @Cast("const char**") @ByPtrPtr byte[] dst, @Cast("size_t*") SizeTPointer dst_len,
+                                             TF_Status status);
+public static native @Cast("size_t") long TF_StringDecode(String src, @Cast("size_t") long src_len,
+                                             @Cast("const char**") @ByPtrPtr BytePointer dst, @Cast("size_t*") SizeTPointer dst_len,
+                                             TF_Status status);
+public static native @Cast("size_t") long TF_StringDecode(@Cast("const char*") BytePointer src, @Cast("size_t") long src_len,
+                                             @Cast("const char**") @ByPtrPtr ByteBuffer dst, @Cast("size_t*") SizeTPointer dst_len,
+                                             TF_Status status);
+public static native @Cast("size_t") long TF_StringDecode(String src, @Cast("size_t") long src_len,
+                                             @Cast("const char**") @ByPtrPtr byte[] dst, @Cast("size_t*") SizeTPointer dst_len,
+                                             TF_Status status);
+
+// Return the size in bytes required to encode a string `len` bytes long into a
+// TF_STRING tensor.
+public static native @Cast("size_t") long TF_StringEncodedSize(@Cast("size_t") long len);
+
+// Returns bool iff this tensor is aligned.
+public static native @Cast("bool") boolean TF_TensorIsAligned(@Const TF_Tensor arg0);
+
+// #ifdef __cplusplus /* end extern "C" */
+// #endif
+
+// #endif  // TENSORFLOW_C_TF_TENSOR_H_
+
+
+// Parsed from tensorflow/c/c_api.h
+
+/* Copyright 2015 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+==============================================================================*/
+
+// #ifndef TENSORFLOW_C_C_API_H_
+// #define TENSORFLOW_C_C_API_H_
+
+// #include <stddef.h>
+// #include <stdint.h>
+
+// #include "tensorflow/c/tf_attrtype.h"
+// #include "tensorflow/c/tf_datatype.h"
+// #include "tensorflow/c/tf_status.h"
+// #include "tensorflow/c/tf_tensor.h"
+
+// --------------------------------------------------------------------------
+// C API for TensorFlow.
+//
+// The API leans towards simplicity and uniformity instead of convenience
+// since most usage will be by language specific wrappers.
+//
+// Conventions:
+// * We use the prefix TF_ for everything in the API.
+// * Objects are always passed around as pointers to opaque structs
+//   and these structs are allocated/deallocated via the API.
+// * TF_Status holds error information.  It is an object type
+//   and therefore is passed around as a pointer to an opaque
+//   struct as mentioned above.
+// * Every call that has a TF_Status* argument clears it on success
+//   and fills it with error info on failure.
+// * unsigned char is used for booleans (instead of the 'bool' type).
+//   In C++ bool is a keyword while in C99 bool is a macro defined
+//   in stdbool.h. It is possible for the two to be inconsistent.
+//   For example, neither the C99 nor the C++11 standard force a byte
+//   size on the bool type, so the macro defined in stdbool.h could
+//   be inconsistent with the bool keyword in C++. Thus, the use
+//   of stdbool.h is avoided and unsigned char is used instead.
+// * size_t is used to represent byte sizes of objects that are
+//   materialized in the address space of the calling process.
+// * int is used as an index into arrays.
+// * Deletion functions are safe to call on nullptr.
+//
+// Questions left to address:
+// * Might at some point need a way for callers to provide their own Env.
+// * Maybe add TF_TensorShape that encapsulates dimension info.
+//
+// Design decisions made:
+// * Backing store for tensor memory has an associated deallocation
+//   function.  This deallocation function will point to client code
+//   for tensors populated by the client.  So the client can do things
+//   like shadowing a numpy array.
+// * We do not provide TF_OK since it is not strictly necessary and we
+//   are not optimizing for convenience.
+// * We make assumption that one session has one graph.  This should be
+//   fine since we have the ability to run sub-graphs.
+// * We could allow NULL for some arguments (e.g., NULL options arg).
+//   However since convenience is not a primary goal, we don't do this.
+// * Devices are not in this API.  Instead, they are created/used internally
+//   and the API just provides high level controls over the number of
+//   devices of each type.
+
+// Macro to control visibility of exported symbols in the shared library (.so,
+// .dylib, .dll).
+// This duplicates the TF_EXPORT macro definition in
+// tensorflow/core/platform/macros.h in order to keep this .h file independent
+// of any other includes.
+// #ifdef SWIG
+// #define TF_CAPI_EXPORT
+// #else
+// #if defined(_WIN32)
+// #ifdef TF_COMPILE_LIBRARY
+// #define TF_CAPI_EXPORT __declspec(dllexport)
+// #else
+// #define TF_CAPI_EXPORT __declspec(dllimport)
+// #endif  // TF_COMPILE_LIBRARY
+// #else
+// #define TF_CAPI_EXPORT __attribute__((visibility("default")))
+// #endif  // _WIN32
+// #endif  // SWIG
+
+// #ifdef __cplusplus
+// #endif
+
+// --------------------------------------------------------------------------
+// TF_Version returns a string describing version information of the
+// TensorFlow library. TensorFlow using semantic versioning.
+public static native @Cast("const char*") BytePointer TF_Version();
+// Targeting ../TF_Buffer.java
+
+
+
+// Makes a copy of the input and sets an appropriate deallocator.  Useful for
+// passing in read-only, input protobufs.
+public static native TF_Buffer TF_NewBufferFromString(@Const Pointer proto,
+                                                        @Cast("size_t") long proto_len);
+
+// Useful for passing *out* a protobuf.
+public static native TF_Buffer TF_NewBuffer();
+
+public static native void TF_DeleteBuffer(TF_Buffer arg0);
+
+public static native @ByVal TF_Buffer TF_GetBuffer(TF_Buffer buffer);
+// Targeting ../TF_SessionOptions.java
+
+
+
+// Return a new options object.
+public static native TF_SessionOptions TF_NewSessionOptions();
+
+// Set the target in TF_SessionOptions.options.
+// target can be empty, a single entry, or a comma separated list of entries.
+// Each entry is in one of the following formats :
+// "local"
+// ip:port
+// host:port
+public static native void TF_SetTarget(TF_SessionOptions options,
+                                        @Cast("const char*") BytePointer target);
+public static native void TF_SetTarget(TF_SessionOptions options,
+                                        String target);
+
+// Set the config in TF_SessionOptions.options.
+// config should be a serialized tensorflow.ConfigProto proto.
+// If config was not parsed successfully as a ConfigProto, record the
+// error information in *status.
+public static native void TF_SetConfig(TF_SessionOptions options,
+                                        @Const Pointer proto, @Cast("size_t") long proto_len,
+                                        TF_Status status);
+
+// Destroy an options object.
+public static native void TF_DeleteSessionOptions(TF_SessionOptions arg0);
+// Targeting ../TF_Graph.java
+
+
+
+// Return a new graph object.
+public static native TF_Graph TF_NewGraph();
+
+// Destroy an options object.  Graph will be deleted once no more
+// TFSession's are referencing it.
+public static native void TF_DeleteGraph(TF_Graph arg0);
+// Targeting ../TF_OperationDescription.java
+
+
+// Targeting ../TF_Operation.java
+
+
+// Targeting ../TF_Input.java
+
+
+// Targeting ../TF_Output.java
+
+
+// Targeting ../TF_Function.java
+
+
+// Targeting ../TF_FunctionOptions.java
+
+
+
+// Sets the shape of the Tensor referenced by `output` in `graph` to
+// the shape described by `dims` and `num_dims`.
+//
+// If the number of dimensions is unknown, `num_dims` must be set to
+// -1 and `dims` can be null. If a dimension is unknown, the
+// corresponding entry in the `dims` array must be -1.
+//
+// This does not overwrite the existing shape associated with `output`,
+// but merges the input shape with the existing shape.  For example,
+// setting a shape of [-1, 2] with an existing shape [2, -1] would set
+// a final shape of [2, 2] based on shape merging semantics.
+//
+// Returns an error into `status` if:
+//   * `output` is not in `graph`.
+//   * An invalid shape is being set (e.g., the shape being set
+//     is incompatible with the existing shape).
+public static native void TF_GraphSetTensorShape(TF_Graph graph,
+                                                  @ByVal TF_Output output,
+                                                  @Cast("const int64_t*") LongPointer dims,
+                                                  int num_dims,
+                                                  TF_Status status);
+public static native void TF_GraphSetTensorShape(TF_Graph graph,
+                                                  @ByVal TF_Output output,
+                                                  @Cast("const int64_t*") LongBuffer dims,
+                                                  int num_dims,
+                                                  TF_Status status);
+public static native void TF_GraphSetTensorShape(TF_Graph graph,
+                                                  @ByVal TF_Output output,
+                                                  @Cast("const int64_t*") long[] dims,
+                                                  int num_dims,
+                                                  TF_Status status);
+
+// Returns the number of dimensions of the Tensor referenced by `output`
+// in `graph`.
+//
+// If the number of dimensions in the shape is unknown, returns -1.
+//
+// Returns an error into `status` if:
+//   * `output` is not in `graph`.
+public static native int TF_GraphGetTensorNumDims(TF_Graph graph,
+                                                   @ByVal TF_Output output,
+                                                   TF_Status status);
+
+// Returns the shape of the Tensor referenced by `output` in `graph`
+// into `dims`. `dims` must be an array large enough to hold `num_dims`
+// entries (e.g., the return value of TF_GraphGetTensorNumDims).
+//
+// If the number of dimensions in the shape is unknown or the shape is
+// a scalar, `dims` will remain untouched. Otherwise, each element of
+// `dims` will be set corresponding to the size of the dimension. An
+// unknown dimension is represented by `-1`.
+//
+// Returns an error into `status` if:
+//   * `output` is not in `graph`.
+//   * `num_dims` does not match the actual number of dimensions.
+public static native void TF_GraphGetTensorShape(TF_Graph graph,
+                                                  @ByVal TF_Output output,
+                                                  @Cast("int64_t*") LongPointer dims, int num_dims,
+                                                  TF_Status status);
+public static native void TF_GraphGetTensorShape(TF_Graph graph,
+                                                  @ByVal TF_Output output,
+                                                  @Cast("int64_t*") LongBuffer dims, int num_dims,
+                                                  TF_Status status);
+public static native void TF_GraphGetTensorShape(TF_Graph graph,
+                                                  @ByVal TF_Output output,
+                                                  @Cast("int64_t*") long[] dims, int num_dims,
+                                                  TF_Status status);
+
+// Operation will only be added to *graph when TF_FinishOperation() is
+// called (assuming TF_FinishOperation() does not return an error).
+// *graph must not be deleted until after TF_FinishOperation() is
+// called.
+public static native TF_OperationDescription TF_NewOperation(
+    TF_Graph graph, @Cast("const char*") BytePointer op_type, @Cast("const char*") BytePointer oper_name);
+public static native TF_OperationDescription TF_NewOperation(
+    TF_Graph graph, String op_type, String oper_name);
+
+// Specify the device for `desc`.  Defaults to empty, meaning unconstrained.
+public static native void TF_SetDevice(TF_OperationDescription desc,
+                                        @Cast("const char*") BytePointer device);
+public static native void TF_SetDevice(TF_OperationDescription desc,
+                                        String device);
+
+// The calls to TF_AddInput and TF_AddInputList must match (in number,
+// order, and type) the op declaration.  For example, the "Concat" op
+// has registration:
+//   REGISTER_OP("Concat")
+//       .Input("concat_dim: int32")
+//       .Input("values: N * T")
+//       .Output("output: T")
+//       .Attr("N: int >= 2")
+//       .Attr("T: type");
+// that defines two inputs, "concat_dim" and "values" (in that order).
+// You must use TF_AddInput() for the first input (since it takes a
+// single tensor), and TF_AddInputList() for the second input (since
+// it takes a list, even if you were to pass a list with a single
+// tensor), as in:
+//   TF_OperationDescription* desc = TF_NewOperation(graph, "Concat", "c");
+//   TF_Output concat_dim_input = {...};
+//   TF_AddInput(desc, concat_dim_input);
+//   TF_Output values_inputs[5] = {{...}, ..., {...}};
+//   TF_AddInputList(desc, values_inputs, 5);
+
+// For inputs that take a single tensor.
+public static native void TF_AddInput(TF_OperationDescription desc,
+                                       @ByVal TF_Output input);
+
+// For inputs that take a list of tensors.
+// inputs must point to TF_Output[num_inputs].
+public static native void TF_AddInputList(TF_OperationDescription desc,
+                                           @Const TF_Output inputs,
+                                           int num_inputs);
+
+// Call once per control input to `desc`.
+public static native void TF_AddControlInput(TF_OperationDescription desc,
+                                              TF_Operation input);
+
+// Request that `desc` be co-located on the device where `op`
+// is placed.
+//
+// Use of this is discouraged since the implementation of device placement is
+// subject to change. Primarily intended for internal libraries
+public static native void TF_ColocateWith(TF_OperationDescription desc,
+                                           TF_Operation op);
+
+// Call some TF_SetAttr*() function for every attr that is not
+// inferred from an input and doesn't have a default value you wish to
+// keep.
+
+// `value` must point to a string of length `length` bytes.
+public static native void TF_SetAttrString(TF_OperationDescription desc,
+                                            @Cast("const char*") BytePointer attr_name,
+                                            @Const Pointer value, @Cast("size_t") long length);
+public static native void TF_SetAttrString(TF_OperationDescription desc,
+                                            String attr_name,
+                                            @Const Pointer value, @Cast("size_t") long length);
+// `values` and `lengths` each must have lengths `num_values`.
+// `values[i]` must point to a string of length `lengths[i]` bytes.
+public static native void TF_SetAttrStringList(TF_OperationDescription desc,
+                                                @Cast("const char*") BytePointer attr_name,
+                                                @Cast("const void*const*") PointerPointer values,
+                                                @Cast("const size_t*") SizeTPointer lengths,
+                                                int num_values);
+public static native void TF_SetAttrStringList(TF_OperationDescription desc,
+                                                @Cast("const char*") BytePointer attr_name,
+                                                @Cast("const void*const*") @ByPtrPtr Pointer values,
+                                                @Cast("const size_t*") SizeTPointer lengths,
+                                                int num_values);
+public static native void TF_SetAttrStringList(TF_OperationDescription desc,
+                                                String attr_name,
+                                                @Cast("const void*const*") @ByPtrPtr Pointer values,
+                                                @Cast("const size_t*") SizeTPointer lengths,
+                                                int num_values);
+public static native void TF_SetAttrInt(TF_OperationDescription desc,
+                                         @Cast("const char*") BytePointer attr_name, @Cast("int64_t") long value);
+public static native void TF_SetAttrInt(TF_OperationDescription desc,
+                                         String attr_name, @Cast("int64_t") long value);
+public static native void TF_SetAttrIntList(TF_OperationDescription desc,
+                                             @Cast("const char*") BytePointer attr_name,
+                                             @Cast("const int64_t*") LongPointer values,
+                                             int num_values);
+public static native void TF_SetAttrIntList(TF_OperationDescription desc,
+                                             String attr_name,
+                                             @Cast("const int64_t*") LongBuffer values,
+                                             int num_values);
+public static native void TF_SetAttrIntList(TF_OperationDescription desc,
+                                             @Cast("const char*") BytePointer attr_name,
+                                             @Cast("const int64_t*") long[] values,
+                                             int num_values);
+public static native void TF_SetAttrIntList(TF_OperationDescription desc,
+                                             String attr_name,
+                                             @Cast("const int64_t*") LongPointer values,
+                                             int num_values);
+public static native void TF_SetAttrIntList(TF_OperationDescription desc,
+                                             @Cast("const char*") BytePointer attr_name,
+                                             @Cast("const int64_t*") LongBuffer values,
+                                             int num_values);
+public static native void TF_SetAttrIntList(TF_OperationDescription desc,
+                                             String attr_name,
+                                             @Cast("const int64_t*") long[] values,
+                                             int num_values);
+public static native void TF_SetAttrFloat(TF_OperationDescription desc,
+                                           @Cast("const char*") BytePointer attr_name, float value);
+public static native void TF_SetAttrFloat(TF_OperationDescription desc,
+                                           String attr_name, float value);
+public static native void TF_SetAttrFloatList(TF_OperationDescription desc,
+                                               @Cast("const char*") BytePointer attr_name,
+                                               @Const FloatPointer values,
+                                               int num_values);
+public static native void TF_SetAttrFloatList(TF_OperationDescription desc,
+                                               String attr_name,
+                                               @Const FloatBuffer values,
+                                               int num_values);
+public static native void TF_SetAttrFloatList(TF_OperationDescription desc,
+                                               @Cast("const char*") BytePointer attr_name,
+                                               @Const float[] values,
+                                               int num_values);
+public static native void TF_SetAttrFloatList(TF_OperationDescription desc,
+                                               String attr_name,
+                                               @Const FloatPointer values,
+                                               int num_values);
+public static native void TF_SetAttrFloatList(TF_OperationDescription desc,
+                                               @Cast("const char*") BytePointer attr_name,
+                                               @Const FloatBuffer values,
+                                               int num_values);
+public static native void TF_SetAttrFloatList(TF_OperationDescription desc,
+                                               String attr_name,
+                                               @Const float[] values,
+                                               int num_values);
+public static native void TF_SetAttrBool(TF_OperationDescription desc,
+                                          @Cast("const char*") BytePointer attr_name,
+                                          @Cast("unsigned char") byte value);
+public static native void TF_SetAttrBool(TF_OperationDescription desc,
+                                          String attr_name,
+                                          @Cast("unsigned char") byte value);
+public static native void TF_SetAttrBoolList(TF_OperationDescription desc,
+                                              @Cast("const char*") BytePointer attr_name,
+                                              @Cast("const unsigned char*") BytePointer values,
+                                              int num_values);
+public static native void TF_SetAttrBoolList(TF_OperationDescription desc,
+                                              String attr_name,
+                                              @Cast("const unsigned char*") ByteBuffer values,
+                                              int num_values);
+public static native void TF_SetAttrBoolList(TF_OperationDescription desc,
+                                              @Cast("const char*") BytePointer attr_name,
+                                              @Cast("const unsigned char*") byte[] values,
+                                              int num_values);
+public static native void TF_SetAttrBoolList(TF_OperationDescription desc,
+                                              String attr_name,
+                                              @Cast("const unsigned char*") BytePointer values,
+                                              int num_values);
+public static native void TF_SetAttrBoolList(TF_OperationDescription desc,
+                                              @Cast("const char*") BytePointer attr_name,
+                                              @Cast("const unsigned char*") ByteBuffer values,
+                                              int num_values);
+public static native void TF_SetAttrBoolList(TF_OperationDescription desc,
+                                              String attr_name,
+                                              @Cast("const unsigned char*") byte[] values,
+                                              int num_values);
+public static native void TF_SetAttrType(TF_OperationDescription desc,
+                                          @Cast("const char*") BytePointer attr_name,
+                                          @Cast("TF_DataType") int value);
+public static native void TF_SetAttrType(TF_OperationDescription desc,
+                                          String attr_name,
+                                          @Cast("TF_DataType") int value);
+public static native void TF_SetAttrTypeList(TF_OperationDescription desc,
+                                              @Cast("const char*") BytePointer attr_name,
+                                              @Cast("const TF_DataType*") IntPointer values,
+                                              int num_values);
+public static native void TF_SetAttrTypeList(TF_OperationDescription desc,
+                                              String attr_name,
+                                              @Cast("const TF_DataType*") IntBuffer values,
+                                              int num_values);
+public static native void TF_SetAttrTypeList(TF_OperationDescription desc,
+                                              @Cast("const char*") BytePointer attr_name,
+                                              @Cast("const TF_DataType*") int[] values,
+                                              int num_values);
+public static native void TF_SetAttrTypeList(TF_OperationDescription desc,
+                                              String attr_name,
+                                              @Cast("const TF_DataType*") IntPointer values,
+                                              int num_values);
+public static native void TF_SetAttrTypeList(TF_OperationDescription desc,
+                                              @Cast("const char*") BytePointer attr_name,
+                                              @Cast("const TF_DataType*") IntBuffer values,
+                                              int num_values);
+public static native void TF_SetAttrTypeList(TF_OperationDescription desc,
+                                              String attr_name,
+                                              @Cast("const TF_DataType*") int[] values,
+                                              int num_values);
+public static native void TF_SetAttrPlaceholder(TF_OperationDescription desc,
+                                                 @Cast("const char*") BytePointer attr_name,
+                                                 @Cast("const char*") BytePointer placeholder);
+public static native void TF_SetAttrPlaceholder(TF_OperationDescription desc,
+                                                 String attr_name,
+                                                 String placeholder);
+
+// Set a 'func' attribute to the specified name.
+// `value` must point to a string of length `length` bytes.
+public static native void TF_SetAttrFuncName(TF_OperationDescription desc,
+                                              @Cast("const char*") BytePointer attr_name,
+                                              @Cast("const char*") BytePointer value, @Cast("size_t") long length);
+public static native void TF_SetAttrFuncName(TF_OperationDescription desc,
+                                              String attr_name,
+                                              String value, @Cast("size_t") long length);
+
+// Set `num_dims` to -1 to represent "unknown rank".  Otherwise,
+// `dims` points to an array of length `num_dims`.  `dims[i]` must be
+// >= -1, with -1 meaning "unknown dimension".
+public static native void TF_SetAttrShape(TF_OperationDescription desc,
+                                           @Cast("const char*") BytePointer attr_name,
+                                           @Cast("const int64_t*") LongPointer dims, int num_dims);
+public static native void TF_SetAttrShape(TF_OperationDescription desc,
+                                           String attr_name,
+                                           @Cast("const int64_t*") LongBuffer dims, int num_dims);
+public static native void TF_SetAttrShape(TF_OperationDescription desc,
+                                           @Cast("const char*") BytePointer attr_name,
+                                           @Cast("const int64_t*") long[] dims, int num_dims);
+public static native void TF_SetAttrShape(TF_OperationDescription desc,
+                                           String attr_name,
+                                           @Cast("const int64_t*") LongPointer dims, int num_dims);
+public static native void TF_SetAttrShape(TF_OperationDescription desc,
+                                           @Cast("const char*") BytePointer attr_name,
+                                           @Cast("const int64_t*") LongBuffer dims, int num_dims);
+public static native void TF_SetAttrShape(TF_OperationDescription desc,
+                                           String attr_name,
+                                           @Cast("const int64_t*") long[] dims, int num_dims);
+// `dims` and `num_dims` must point to arrays of length `num_shapes`.
+// Set `num_dims[i]` to -1 to represent "unknown rank".  Otherwise,
+// `dims[i]` points to an array of length `num_dims[i]`.  `dims[i][j]`
+// must be >= -1, with -1 meaning "unknown dimension".
+public static native void TF_SetAttrShapeList(TF_OperationDescription desc,
+                                               @Cast("const char*") BytePointer attr_name,
+                                               @Cast("const int64_t*const*") PointerPointer dims,
+                                               @Const IntPointer num_dims,
+                                               int num_shapes);
+public static native void TF_SetAttrShapeList(TF_OperationDescription desc,
+                                               @Cast("const char*") BytePointer attr_name,
+                                               @Cast("const int64_t*const*") @ByPtrPtr LongPointer dims,
+                                               @Const IntPointer num_dims,
+                                               int num_shapes);
+public static native void TF_SetAttrShapeList(TF_OperationDescription desc,
+                                               String attr_name,
+                                               @Cast("const int64_t*const*") @ByPtrPtr LongBuffer dims,
+                                               @Const IntBuffer num_dims,
+                                               int num_shapes);
+public static native void TF_SetAttrShapeList(TF_OperationDescription desc,
+                                               @Cast("const char*") BytePointer attr_name,
+                                               @Cast("const int64_t*const*") @ByPtrPtr long[] dims,
+                                               @Const int[] num_dims,
+                                               int num_shapes);
+public static native void TF_SetAttrShapeList(TF_OperationDescription desc,
+                                               String attr_name,
+                                               @Cast("const int64_t*const*") @ByPtrPtr LongPointer dims,
+                                               @Const IntPointer num_dims,
+                                               int num_shapes);
+public static native void TF_SetAttrShapeList(TF_OperationDescription desc,
+                                               @Cast("const char*") BytePointer attr_name,
+                                               @Cast("const int64_t*const*") @ByPtrPtr LongBuffer dims,
+                                               @Const IntBuffer num_dims,
+                                               int num_shapes);
+public static native void TF_SetAttrShapeList(TF_OperationDescription desc,
+                                               String attr_name,
+                                               @Cast("const int64_t*const*") @ByPtrPtr long[] dims,
+                                               @Const int[] num_dims,
+                                               int num_shapes);
+// `proto` must point to an array of `proto_len` bytes representing a
+// binary-serialized TensorShapeProto.
+public static native void TF_SetAttrTensorShapeProto(
+    TF_OperationDescription desc, @Cast("const char*") BytePointer attr_name, @Const Pointer proto,
+    @Cast("size_t") long proto_len, TF_Status status);
+public static native void TF_SetAttrTensorShapeProto(
+    TF_OperationDescription desc, String attr_name, @Const Pointer proto,
+    @Cast("size_t") long proto_len, TF_Status status);
+// `protos` and `proto_lens` must point to arrays of length `num_shapes`.
+// `protos[i]` must point to an array of `proto_lens[i]` bytes
+// representing a binary-serialized TensorShapeProto.
+public static native void TF_SetAttrTensorShapeProtoList(
+    TF_OperationDescription desc, @Cast("const char*") BytePointer attr_name,
+    @Cast("const void*const*") PointerPointer protos, @Cast("const size_t*") SizeTPointer proto_lens, int num_shapes,
+    TF_Status status);
+public static native void TF_SetAttrTensorShapeProtoList(
+    TF_OperationDescription desc, @Cast("const char*") BytePointer attr_name,
+    @Cast("const void*const*") @ByPtrPtr Pointer protos, @Cast("const size_t*") SizeTPointer proto_lens, int num_shapes,
+    TF_Status status);
+public static native void TF_SetAttrTensorShapeProtoList(
+    TF_OperationDescription desc, String attr_name,
+    @Cast("const void*const*") @ByPtrPtr Pointer protos, @Cast("const size_t*") SizeTPointer proto_lens, int num_shapes,
+    TF_Status status);
+
+public static native void TF_SetAttrTensor(TF_OperationDescription desc,
+                                            @Cast("const char*") BytePointer attr_name,
+                                            TF_Tensor value,
+                                            TF_Status status);
+public static native void TF_SetAttrTensor(TF_OperationDescription desc,
+                                            String attr_name,
+                                            TF_Tensor value,
+                                            TF_Status status);
+public static native void TF_SetAttrTensorList(TF_OperationDescription desc,
+                                                @Cast("const char*") BytePointer attr_name,
+                                                @Cast("TF_Tensor*const*") PointerPointer values,
+                                                int num_values,
+                                                TF_Status status);
+public static native void TF_SetAttrTensorList(TF_OperationDescription desc,
+                                                @Cast("const char*") BytePointer attr_name,
+                                                @ByPtrPtr TF_Tensor values,
+                                                int num_values,
+                                                TF_Status status);
+public static native void TF_SetAttrTensorList(TF_OperationDescription desc,
+                                                String attr_name,
+                                                @ByPtrPtr TF_Tensor values,
+                                                int num_values,
+                                                TF_Status status);
+
+// `proto` should point to a sequence of bytes of length `proto_len`
+// representing a binary serialization of an AttrValue protocol
+// buffer.
+public static native void TF_SetAttrValueProto(TF_OperationDescription desc,
+                                                @Cast("const char*") BytePointer attr_name,
+                                                @Const Pointer proto,
+                                                @Cast("size_t") long proto_len,
+                                                TF_Status status);
+public static native void TF_SetAttrValueProto(TF_OperationDescription desc,
+                                                String attr_name,
+                                                @Const Pointer proto,
+                                                @Cast("size_t") long proto_len,
+                                                TF_Status status);
+
+// If this function succeeds:
+//   * *status is set to an OK value,
+//   * a TF_Operation is added to the graph,
+//   * a non-null value pointing to the added operation is returned --
+//     this value is valid until the underlying graph is deleted.
+// Otherwise:
+//   * *status is set to a non-OK value,
+//   * the graph is not modified,
+//   * a null value is returned.
+// In either case, it deletes `desc`.
+public static native TF_Operation TF_FinishOperation(
+    TF_OperationDescription desc, TF_Status status);
+
+// TF_Operation functions.  Operations are immutable once created, so
+// these are all query functions.
+
+public static native @Cast("const char*") BytePointer TF_OperationName(TF_Operation oper);
+public static native @Cast("const char*") BytePointer TF_OperationOpType(TF_Operation oper);
+public static native @Cast("const char*") BytePointer TF_OperationDevice(TF_Operation oper);
+
+public static native int TF_OperationNumOutputs(TF_Operation oper);
+public static native @Cast("TF_DataType") int TF_OperationOutputType(@ByVal TF_Output oper_out);
+public static native int TF_OperationOutputListLength(TF_Operation oper,
+                                                       @Cast("const char*") BytePointer arg_name,
+                                                       TF_Status status);
+public static native int TF_OperationOutputListLength(TF_Operation oper,
+                                                       String arg_name,
+                                                       TF_Status status);
+
+public static native int TF_OperationNumInputs(TF_Operation oper);
+public static native @Cast("TF_DataType") int TF_OperationInputType(@ByVal TF_Input oper_in);
+public static native int TF_OperationInputListLength(TF_Operation oper,
+                                                      @Cast("const char*") BytePointer arg_name,
+                                                      TF_Status status);
+public static native int TF_OperationInputListLength(TF_Operation oper,
+                                                      String arg_name,
+                                                      TF_Status status);
+
+// In this code:
+//   TF_Output producer = TF_OperationInput(consumer);
+// There is an edge from producer.oper's output (given by
+// producer.index) to consumer.oper's input (given by consumer.index).
+public static native @ByVal TF_Output TF_OperationInput(@ByVal TF_Input oper_in);
+
+// Get list of all inputs of a specific operation.  `inputs` must point to
+// an array of length at least `max_inputs` (ideally set to
+// TF_OperationNumInputs(oper)).  Beware that a concurrent
+// modification of the graph can increase the number of inputs of
+// an operation.
+public static native void TF_OperationAllInputs(TF_Operation oper,
+                                                 TF_Output inputs,
+                                                 int max_inputs);
+
+// Get the number of current consumers of a specific output of an
+// operation.  Note that this number can change when new operations
+// are added to the graph.
+public static native int TF_OperationOutputNumConsumers(@ByVal TF_Output oper_out);
+
+// Get list of all current consumers of a specific output of an
+// operation.  `consumers` must point to an array of length at least
+// `max_consumers` (ideally set to
+// TF_OperationOutputNumConsumers(oper_out)).  Beware that a concurrent
+// modification of the graph can increase the number of consumers of
+// an operation.  Returns the number of output consumers (should match
+// TF_OperationOutputNumConsumers(oper_out)).
+public static native int TF_OperationOutputConsumers(@ByVal TF_Output oper_out,
+                                                      TF_Input consumers,
+                                                      int max_consumers);
+
+// Get the number of control inputs to an operation.
+public static native int TF_OperationNumControlInputs(TF_Operation oper);
+
+// Get list of all control inputs to an operation.  `control_inputs` must
+// point to an array of length `max_control_inputs` (ideally set to
+// TF_OperationNumControlInputs(oper)).  Returns the number of control
+// inputs (should match TF_OperationNumControlInputs(oper)).
+public static native int TF_OperationGetControlInputs(
+    TF_Operation oper, @Cast("TF_Operation**") PointerPointer control_inputs, int max_control_inputs);
+public static native int TF_OperationGetControlInputs(
+    TF_Operation oper, @ByPtrPtr TF_Operation control_inputs, int max_control_inputs);
+
+// Get the number of operations that have `*oper` as a control input.
+// Note that this number can change when new operations are added to
+// the graph.
+public static native int TF_OperationNumControlOutputs(TF_Operation oper);
+
+// Get the list of operations that have `*oper` as a control input.
+// `control_outputs` must point to an array of length at least
+// `max_control_outputs` (ideally set to
+// TF_OperationNumControlOutputs(oper)). Beware that a concurrent
+// modification of the graph can increase the number of control
+// outputs.  Returns the number of control outputs (should match
+// TF_OperationNumControlOutputs(oper)).
+public static native int TF_OperationGetControlOutputs(
+    TF_Operation oper, @Cast("TF_Operation**") PointerPointer control_outputs,
+    int max_control_outputs);
+public static native int TF_OperationGetControlOutputs(
+    TF_Operation oper, @ByPtrPtr TF_Operation control_outputs,
+    int max_control_outputs);
+// Targeting ../TF_AttrMetadata.java
+
+
+
+// Returns metadata about the value of the attribute `attr_name` of `oper`.
+public static native @ByVal TF_AttrMetadata TF_OperationGetAttrMetadata(
+    TF_Operation oper, @Cast("const char*") BytePointer attr_name, TF_Status status);
+public static native @ByVal TF_AttrMetadata TF_OperationGetAttrMetadata(
+    TF_Operation oper, String attr_name, TF_Status status);
+
+// Fills in `value` with the value of the attribute `attr_name`.  `value` must
+// point to an array of length at least `max_length` (ideally set to
+// TF_AttrMetadata.total_size from TF_OperationGetAttrMetadata(oper,
+// attr_name)).
+public static native void TF_OperationGetAttrString(TF_Operation oper,
+                                                     @Cast("const char*") BytePointer attr_name,
+                                                     Pointer value,
+                                                     @Cast("size_t") long max_length,
+                                                     TF_Status status);
+public static native void TF_OperationGetAttrString(TF_Operation oper,
+                                                     String attr_name,
+                                                     Pointer value,
+                                                     @Cast("size_t") long max_length,
+                                                     TF_Status status);
+
+// Get the list of strings in the value of the attribute `attr_name`.  Fills in
+// `values` and `lengths`, each of which must point to an array of length at
+// least `max_values`.
+//
+// The elements of values will point to addresses in `storage` which must be at
+// least `storage_size` bytes in length.  Ideally, max_values would be set to
+// TF_AttrMetadata.list_size and `storage` would be at least
+// TF_AttrMetadata.total_size, obtained from TF_OperationGetAttrMetadata(oper,
+// attr_name).
+//
+// Fails if storage_size is too small to hold the requested number of strings.
+public static native void TF_OperationGetAttrStringList(
+    TF_Operation oper, @Cast("const char*") BytePointer attr_name, @Cast("void**") PointerPointer values, @Cast("size_t*") SizeTPointer lengths,
+    int max_values, Pointer storage, @Cast("size_t") long storage_size, TF_Status status);
+public static native void TF_OperationGetAttrStringList(
+    TF_Operation oper, @Cast("const char*") BytePointer attr_name, @Cast("void**") @ByPtrPtr Pointer values, @Cast("size_t*") SizeTPointer lengths,
+    int max_values, Pointer storage, @Cast("size_t") long storage_size, TF_Status status);
+public static native void TF_OperationGetAttrStringList(
+    TF_Operation oper, String attr_name, @Cast("void**") @ByPtrPtr Pointer values, @Cast("size_t*") SizeTPointer lengths,
+    int max_values, Pointer storage, @Cast("size_t") long storage_size, TF_Status status);
+
+public static native void TF_OperationGetAttrInt(TF_Operation oper,
+                                                  @Cast("const char*") BytePointer attr_name,
+                                                  @Cast("int64_t*") LongPointer value,
+                                                  TF_Status status);
+public static native void TF_OperationGetAttrInt(TF_Operation oper,
+                                                  String attr_name,
+                                                  @Cast("int64_t*") LongBuffer value,
+                                                  TF_Status status);
+public static native void TF_OperationGetAttrInt(TF_Operation oper,
+                                                  @Cast("const char*") BytePointer attr_name,
+                                                  @Cast("int64_t*") long[] value,
+                                                  TF_Status status);
+public static native void TF_OperationGetAttrInt(TF_Operation oper,
+                                                  String attr_name,
+                                                  @Cast("int64_t*") LongPointer value,
+                                                  TF_Status status);
+public static native void TF_OperationGetAttrInt(TF_Operation oper,
+                                                  @Cast("const char*") BytePointer attr_name,
+                                                  @Cast("int64_t*") LongBuffer value,
+                                                  TF_Status status);
+public static native void TF_OperationGetAttrInt(TF_Operation oper,
+                                                  String attr_name,
+                                                  @Cast("int64_t*") long[] value,
+                                                  TF_Status status);
+
+// Fills in `values` with the value of the attribute `attr_name` of `oper`.
+// `values` must point to an array of length at least `max_values` (ideally set
+// TF_AttrMetadata.list_size from TF_OperationGetAttrMetadata(oper,
+// attr_name)).
+public static native void TF_OperationGetAttrIntList(TF_Operation oper,
+                                                      @Cast("const char*") BytePointer attr_name,
+                                                      @Cast("int64_t*") LongPointer values,
+                                                      int max_values,
+                                                      TF_Status status);
+public static native void TF_OperationGetAttrIntList(TF_Operation oper,
+                                                      String attr_name,
+                                                      @Cast("int64_t*") LongBuffer values,
+                                                      int max_values,
+                                                      TF_Status status);
+public static native void TF_OperationGetAttrIntList(TF_Operation oper,
+                                                      @Cast("const char*") BytePointer attr_name,
+                                                      @Cast("int64_t*") long[] values,
+                                                      int max_values,
+                                                      TF_Status status);
+public static native void TF_OperationGetAttrIntList(TF_Operation oper,
+                                                      String attr_name,
+                                                      @Cast("int64_t*") LongPointer values,
+                                                      int max_values,
+                                                      TF_Status status);
+public static native void TF_OperationGetAttrIntList(TF_Operation oper,
+                                                      @Cast("const char*") BytePointer attr_name,
+                                                      @Cast("int64_t*") LongBuffer values,
+                                                      int max_values,
+                                                      TF_Status status);
+public static native void TF_OperationGetAttrIntList(TF_Operation oper,
+                                                      String attr_name,
+                                                      @Cast("int64_t*") long[] values,
+                                                      int max_values,
+                                                      TF_Status status);
+
+public static native void TF_OperationGetAttrFloat(TF_Operation oper,
+                                                    @Cast("const char*") BytePointer attr_name,
+                                                    FloatPointer value,
+                                                    TF_Status status);
+public static native void TF_OperationGetAttrFloat(TF_Operation oper,
+                                                    String attr_name,
+                                                    FloatBuffer value,
+                                                    TF_Status status);
+public static native void TF_OperationGetAttrFloat(TF_Operation oper,
+                                                    @Cast("const char*") BytePointer attr_name,
+                                                    float[] value,
+                                                    TF_Status status);
+public static native void TF_OperationGetAttrFloat(TF_Operation oper,
+                                                    String attr_name,
+                                                    FloatPointer value,
+                                                    TF_Status status);
+public static native void TF_OperationGetAttrFloat(TF_Operation oper,
+                                                    @Cast("const char*") BytePointer attr_name,
+                                                    FloatBuffer value,
+                                                    TF_Status status);
+public static native void TF_OperationGetAttrFloat(TF_Operation oper,
+                                                    String attr_name,
+                                                    float[] value,
+                                                    TF_Status status);
+
+// Fills in `values` with the value of the attribute `attr_name` of `oper`.
+// `values` must point to an array of length at least `max_values` (ideally set
+// to TF_AttrMetadata.list_size from TF_OperationGetAttrMetadata(oper,
+// attr_name)).
+public static native void TF_OperationGetAttrFloatList(TF_Operation oper,
+                                                        @Cast("const char*") BytePointer attr_name,
+                                                        FloatPointer values,
+                                                        int max_values,
+                                                        TF_Status status);
+public static native void TF_OperationGetAttrFloatList(TF_Operation oper,
+                                                        String attr_name,
+                                                        FloatBuffer values,
+                                                        int max_values,
+                                                        TF_Status status);
+public static native void TF_OperationGetAttrFloatList(TF_Operation oper,
+                                                        @Cast("const char*") BytePointer attr_name,
+                                                        float[] values,
+                                                        int max_values,
+                                                        TF_Status status);
+public static native void TF_OperationGetAttrFloatList(TF_Operation oper,
+                                                        String attr_name,
+                                                        FloatPointer values,
+                                                        int max_values,
+                                                        TF_Status status);
+public static native void TF_OperationGetAttrFloatList(TF_Operation oper,
+                                                        @Cast("const char*") BytePointer attr_name,
+                                                        FloatBuffer values,
+                                                        int max_values,
+                                                        TF_Status status);
+public static native void TF_OperationGetAttrFloatList(TF_Operation oper,
+                                                        String attr_name,
+                                                        float[] values,
+                                                        int max_values,
+                                                        TF_Status status);
+
+public static native void TF_OperationGetAttrBool(TF_Operation oper,
+                                                   @Cast("const char*") BytePointer attr_name,
+                                                   @Cast("unsigned char*") BytePointer value,
+                                                   TF_Status status);
+public static native void TF_OperationGetAttrBool(TF_Operation oper,
+                                                   String attr_name,
+                                                   @Cast("unsigned char*") ByteBuffer value,
+                                                   TF_Status status);
+public static native void TF_OperationGetAttrBool(TF_Operation oper,
+                                                   @Cast("const char*") BytePointer attr_name,
+                                                   @Cast("unsigned char*") byte[] value,
+                                                   TF_Status status);
+public static native void TF_OperationGetAttrBool(TF_Operation oper,
+                                                   String attr_name,
+                                                   @Cast("unsigned char*") BytePointer value,
+                                                   TF_Status status);
+public static native void TF_OperationGetAttrBool(TF_Operation oper,
+                                                   @Cast("const char*") BytePointer attr_name,
+                                                   @Cast("unsigned char*") ByteBuffer value,
+                                                   TF_Status status);
+public static native void TF_OperationGetAttrBool(TF_Operation oper,
+                                                   String attr_name,
+                                                   @Cast("unsigned char*") byte[] value,
+                                                   TF_Status status);
+
+// Fills in `values` with the value of the attribute `attr_name` of `oper`.
+// `values` must point to an array of length at least `max_values` (ideally set
+// to TF_AttrMetadata.list_size from TF_OperationGetAttrMetadata(oper,
+// attr_name)).
+public static native void TF_OperationGetAttrBoolList(TF_Operation oper,
+                                                       @Cast("const char*") BytePointer attr_name,
+                                                       @Cast("unsigned char*") BytePointer values,
+                                                       int max_values,
+                                                       TF_Status status);
+public static native void TF_OperationGetAttrBoolList(TF_Operation oper,
+                                                       String attr_name,
+                                                       @Cast("unsigned char*") ByteBuffer values,
+                                                       int max_values,
+                                                       TF_Status status);
+public static native void TF_OperationGetAttrBoolList(TF_Operation oper,
+                                                       @Cast("const char*") BytePointer attr_name,
+                                                       @Cast("unsigned char*") byte[] values,
+                                                       int max_values,
+                                                       TF_Status status);
+public static native void TF_OperationGetAttrBoolList(TF_Operation oper,
+                                                       String attr_name,
+                                                       @Cast("unsigned char*") BytePointer values,
+                                                       int max_values,
+                                                       TF_Status status);
+public static native void TF_OperationGetAttrBoolList(TF_Operation oper,
+                                                       @Cast("const char*") BytePointer attr_name,
+                                                       @Cast("unsigned char*") ByteBuffer values,
+                                                       int max_values,
+                                                       TF_Status status);
+public static native void TF_OperationGetAttrBoolList(TF_Operation oper,
+                                                       String attr_name,
+                                                       @Cast("unsigned char*") byte[] values,
+                                                       int max_values,
+                                                       TF_Status status);
+
+public static native void TF_OperationGetAttrType(TF_Operation oper,
+                                                   @Cast("const char*") BytePointer attr_name,
+                                                   @Cast("TF_DataType*") IntPointer value,
+                                                   TF_Status status);
+public static native void TF_OperationGetAttrType(TF_Operation oper,
+                                                   String attr_name,
+                                                   @Cast("TF_DataType*") IntBuffer value,
+                                                   TF_Status status);
+public static native void TF_OperationGetAttrType(TF_Operation oper,
+                                                   @Cast("const char*") BytePointer attr_name,
+                                                   @Cast("TF_DataType*") int[] value,
+                                                   TF_Status status);
+public static native void TF_OperationGetAttrType(TF_Operation oper,
+                                                   String attr_name,
+                                                   @Cast("TF_DataType*") IntPointer value,
+                                                   TF_Status status);
+public static native void TF_OperationGetAttrType(TF_Operation oper,
+                                                   @Cast("const char*") BytePointer attr_name,
+                                                   @Cast("TF_DataType*") IntBuffer value,
+                                                   TF_Status status);
+public static native void TF_OperationGetAttrType(TF_Operation oper,
+                                                   String attr_name,
+                                                   @Cast("TF_DataType*") int[] value,
+                                                   TF_Status status);
+
+// Fills in `values` with the value of the attribute `attr_name` of `oper`.
+// `values` must point to an array of length at least `max_values` (ideally set
+// to TF_AttrMetadata.list_size from TF_OperationGetAttrMetadata(oper,
+// attr_name)).
+public static native void TF_OperationGetAttrTypeList(TF_Operation oper,
+                                                       @Cast("const char*") BytePointer attr_name,
+                                                       @Cast("TF_DataType*") IntPointer values,
+                                                       int max_values,
+                                                       TF_Status status);
+public static native void TF_OperationGetAttrTypeList(TF_Operation oper,
+                                                       String attr_name,
+                                                       @Cast("TF_DataType*") IntBuffer values,
+                                                       int max_values,
+                                                       TF_Status status);
+public static native void TF_OperationGetAttrTypeList(TF_Operation oper,
+                                                       @Cast("const char*") BytePointer attr_name,
+                                                       @Cast("TF_DataType*") int[] values,
+                                                       int max_values,
+                                                       TF_Status status);
+public static native void TF_OperationGetAttrTypeList(TF_Operation oper,
+                                                       String attr_name,
+                                                       @Cast("TF_DataType*") IntPointer values,
+                                                       int max_values,
+                                                       TF_Status status);
+public static native void TF_OperationGetAttrTypeList(TF_Operation oper,
+                                                       @Cast("const char*") BytePointer attr_name,
+                                                       @Cast("TF_DataType*") IntBuffer values,
+                                                       int max_values,
+                                                       TF_Status status);
+public static native void TF_OperationGetAttrTypeList(TF_Operation oper,
+                                                       String attr_name,
+                                                       @Cast("TF_DataType*") int[] values,
+                                                       int max_values,
+                                                       TF_Status status);
+
+// Fills in `value` with the value of the attribute `attr_name` of `oper`.
+// `values` must point to an array of length at least `num_dims` (ideally set to
+// TF_Attr_Meta.size from TF_OperationGetAttrMetadata(oper, attr_name)).
+public static native void TF_OperationGetAttrShape(TF_Operation oper,
+                                                    @Cast("const char*") BytePointer attr_name,
+                                                    @Cast("int64_t*") LongPointer value,
+                                                    int num_dims,
+                                                    TF_Status status);
+public static native void TF_OperationGetAttrShape(TF_Operation oper,
+                                                    String attr_name,
+                                                    @Cast("int64_t*") LongBuffer value,
+                                                    int num_dims,
+                                                    TF_Status status);
+public static native void TF_OperationGetAttrShape(TF_Operation oper,
+                                                    @Cast("const char*") BytePointer attr_name,
+                                                    @Cast("int64_t*") long[] value,
+                                                    int num_dims,
+                                                    TF_Status status);
+public static native void TF_OperationGetAttrShape(TF_Operation oper,
+                                                    String attr_name,
+                                                    @Cast("int64_t*") LongPointer value,
+                                                    int num_dims,
+                                                    TF_Status status);
+public static native void TF_OperationGetAttrShape(TF_Operation oper,
+                                                    @Cast("const char*") BytePointer attr_name,
+                                                    @Cast("int64_t*") LongBuffer value,
+                                                    int num_dims,
+                                                    TF_Status status);
+public static native void TF_OperationGetAttrShape(TF_Operation oper,
+                                                    String attr_name,
+                                                    @Cast("int64_t*") long[] value,
+                                                    int num_dims,
+                                                    TF_Status status);
+
+// Fills in `dims` with the list of shapes in the attribute `attr_name` of
+// `oper` and `num_dims` with the corresponding number of dimensions. On return,
+// for every i where `num_dims[i]` > 0, `dims[i]` will be an array of
+// `num_dims[i]` elements. A value of -1 for `num_dims[i]` indicates that the
+// i-th shape in the list is unknown.
+//
+// The elements of `dims` will point to addresses in `storage` which must be
+// large enough to hold at least `storage_size` int64_ts.  Ideally, `num_shapes`
+// would be set to TF_AttrMetadata.list_size and `storage_size` would be set to
+// TF_AttrMetadata.total_size from TF_OperationGetAttrMetadata(oper,
+// attr_name).
+//
+// Fails if storage_size is insufficient to hold the requested shapes.
+public static native void TF_OperationGetAttrShapeList(
+    TF_Operation oper, @Cast("const char*") BytePointer attr_name, @Cast("int64_t**") PointerPointer dims, IntPointer num_dims,
+    int num_shapes, @Cast("int64_t*") LongPointer storage, int storage_size, TF_Status status);
+public static native void TF_OperationGetAttrShapeList(
+    TF_Operation oper, @Cast("const char*") BytePointer attr_name, @Cast("int64_t**") @ByPtrPtr LongPointer dims, IntPointer num_dims,
+    int num_shapes, @Cast("int64_t*") LongPointer storage, int storage_size, TF_Status status);
+public static native void TF_OperationGetAttrShapeList(
+    TF_Operation oper, String attr_name, @Cast("int64_t**") @ByPtrPtr LongBuffer dims, IntBuffer num_dims,
+    int num_shapes, @Cast("int64_t*") LongBuffer storage, int storage_size, TF_Status status);
+public static native void TF_OperationGetAttrShapeList(
+    TF_Operation oper, @Cast("const char*") BytePointer attr_name, @Cast("int64_t**") @ByPtrPtr long[] dims, int[] num_dims,
+    int num_shapes, @Cast("int64_t*") long[] storage, int storage_size, TF_Status status);
+public static native void TF_OperationGetAttrShapeList(
+    TF_Operation oper, String attr_name, @Cast("int64_t**") @ByPtrPtr LongPointer dims, IntPointer num_dims,
+    int num_shapes, @Cast("int64_t*") LongPointer storage, int storage_size, TF_Status status);
+public static native void TF_OperationGetAttrShapeList(
+    TF_Operation oper, @Cast("const char*") BytePointer attr_name, @Cast("int64_t**") @ByPtrPtr LongBuffer dims, IntBuffer num_dims,
+    int num_shapes, @Cast("int64_t*") LongBuffer storage, int storage_size, TF_Status status);
+public static native void TF_OperationGetAttrShapeList(
+    TF_Operation oper, String attr_name, @Cast("int64_t**") @ByPtrPtr long[] dims, int[] num_dims,
+    int num_shapes, @Cast("int64_t*") long[] storage, int storage_size, TF_Status status);
+
+// Sets `value` to the binary-serialized TensorShapeProto of the value of
+// `attr_name` attribute of `oper`'.
+public static native void TF_OperationGetAttrTensorShapeProto(
+    TF_Operation oper, @Cast("const char*") BytePointer attr_name, TF_Buffer value,
+    TF_Status status);
+public static native void TF_OperationGetAttrTensorShapeProto(
+    TF_Operation oper, String attr_name, TF_Buffer value,
+    TF_Status status);
+
+// Fills in `values` with binary-serialized TensorShapeProto values of the
+// attribute `attr_name` of `oper`. `values` must point to an array of length at
+// least `num_values` (ideally set to TF_AttrMetadata.list_size from
+// TF_OperationGetAttrMetadata(oper, attr_name)).
+public static native void TF_OperationGetAttrTensorShapeProtoList(
+    TF_Operation oper, @Cast("const char*") BytePointer attr_name, @Cast("TF_Buffer**") PointerPointer values,
+    int max_values, TF_Status status);
+public static native void TF_OperationGetAttrTensorShapeProtoList(
+    TF_Operation oper, @Cast("const char*") BytePointer attr_name, @ByPtrPtr TF_Buffer values,
+    int max_values, TF_Status status);
+public static native void TF_OperationGetAttrTensorShapeProtoList(
+    TF_Operation oper, String attr_name, @ByPtrPtr TF_Buffer values,
+    int max_values, TF_Status status);
+
+// Gets the TF_Tensor valued attribute of `attr_name` of `oper`.
+//
+// Allocates a new TF_Tensor which the caller is expected to take
+// ownership of (and can deallocate using TF_DeleteTensor).
+public static native void TF_OperationGetAttrTensor(TF_Operation oper,
+                                                     @Cast("const char*") BytePointer attr_name,
+                                                     @Cast("TF_Tensor**") PointerPointer value,
+                                                     TF_Status status);
+public static native void TF_OperationGetAttrTensor(TF_Operation oper,
+                                                     @Cast("const char*") BytePointer attr_name,
+                                                     @ByPtrPtr TF_Tensor value,
+                                                     TF_Status status);
+public static native void TF_OperationGetAttrTensor(TF_Operation oper,
+                                                     String attr_name,
+                                                     @ByPtrPtr TF_Tensor value,
+                                                     TF_Status status);
+
+// Fills in `values` with the TF_Tensor values of the attribute `attr_name` of
+// `oper`. `values` must point to an array of TF_Tensor* of length at least
+// `max_values` (ideally set to TF_AttrMetadata.list_size from
+// TF_OperationGetAttrMetadata(oper, attr_name)).
+//
+// The caller takes ownership of all the non-null TF_Tensor* entries in `values`
+// (which can be deleted using TF_DeleteTensor(values[i])).
+public static native void TF_OperationGetAttrTensorList(TF_Operation oper,
+                                                         @Cast("const char*") BytePointer attr_name,
+                                                         @Cast("TF_Tensor**") PointerPointer values,
+                                                         int max_values,
+                                                         TF_Status status);
+public static native void TF_OperationGetAttrTensorList(TF_Operation oper,
+                                                         @Cast("const char*") BytePointer attr_name,
+                                                         @ByPtrPtr TF_Tensor values,
+                                                         int max_values,
+                                                         TF_Status status);
+public static native void TF_OperationGetAttrTensorList(TF_Operation oper,
+                                                         String attr_name,
+                                                         @ByPtrPtr TF_Tensor values,
+                                                         int max_values,
+                                                         TF_Status status);
+
+// Sets `output_attr_value` to the binary-serialized AttrValue proto
+// representation of the value of the `attr_name` attr of `oper`.
+public static native void TF_OperationGetAttrValueProto(
+    TF_Operation oper, @Cast("const char*") BytePointer attr_name, TF_Buffer output_attr_value,
+    TF_Status status);
+public static native void TF_OperationGetAttrValueProto(
+    TF_Operation oper, String attr_name, TF_Buffer output_attr_value,
+    TF_Status status);
+
+// Returns the operation in the graph with `oper_name`. Returns nullptr if
+// no operation found.
+public static native TF_Operation TF_GraphOperationByName(
+    TF_Graph graph, @Cast("const char*") BytePointer oper_name);
+public static native TF_Operation TF_GraphOperationByName(
+    TF_Graph graph, String oper_name);
+
+// Iterate through the operations of a graph.  To use:
+// size_t pos = 0;
+// TF_Operation* oper;
+// while ((oper = TF_GraphNextOperation(graph, &pos)) != nullptr) {
+//   DoSomethingWithOperation(oper);
+// }
+public static native TF_Operation TF_GraphNextOperation(TF_Graph graph,
+                                                          @Cast("size_t*") SizeTPointer pos);
+
+// Write out a serialized representation of `graph` (as a GraphDef protocol
+// message) to `output_graph_def` (allocated by TF_NewBuffer()).
+// `output_graph_def`'s underlying buffer will be freed when TF_DeleteBuffer()
+// is called.
+//
+// May fail on very large graphs in the future.
+public static native void TF_GraphToGraphDef(TF_Graph graph,
+                                              TF_Buffer output_graph_def,
+                                              TF_Status status);
+
+// Returns the serialized OpDef proto with name `op_name`, or a bad status if no
+// such op exists. This can return OpDefs of functions copied into the graph.
+public static native void TF_GraphGetOpDef(TF_Graph graph,
+                                            @Cast("const char*") BytePointer op_name,
+                                            TF_Buffer output_op_def,
+                                            TF_Status status);
+public static native void TF_GraphGetOpDef(TF_Graph graph,
+                                            String op_name,
+                                            TF_Buffer output_op_def,
+                                            TF_Status status);
+
+// Returns the serialized VersionDef proto for this graph.
+public static native void TF_GraphVersions(TF_Graph graph,
+                                            TF_Buffer output_version_def,
+                                            TF_Status status);
+// Targeting ../TF_ImportGraphDefOptions.java
+
+
+
+public static native TF_ImportGraphDefOptions TF_NewImportGraphDefOptions();
+public static native void TF_DeleteImportGraphDefOptions(
+    TF_ImportGraphDefOptions opts);
+
+// Set the prefix to be prepended to the names of nodes in `graph_def` that will
+// be imported into `graph`. `prefix` is copied and has no lifetime
+// requirements.
+public static native void TF_ImportGraphDefOptionsSetPrefix(
+    TF_ImportGraphDefOptions opts, @Cast("const char*") BytePointer prefix);
+public static native void TF_ImportGraphDefOptionsSetPrefix(
+    TF_ImportGraphDefOptions opts, String prefix);
+
+// Set the execution device for nodes in `graph_def`.
+// Only applies to nodes where a device was not already explicitly specified.
+// `device` is copied and has no lifetime requirements.
+public static native void TF_ImportGraphDefOptionsSetDefaultDevice(
+    TF_ImportGraphDefOptions opts, @Cast("const char*") BytePointer device);
+public static native void TF_ImportGraphDefOptionsSetDefaultDevice(
+    TF_ImportGraphDefOptions opts, String device);
+
+// Set whether to uniquify imported operation names. If true, imported operation
+// names will be modified if their name already exists in the graph. If false,
+// conflicting names will be treated as an error. Note that this option has no
+// effect if a prefix is set, since the prefix will guarantee all names are
+// unique. Defaults to false.
+public static native void TF_ImportGraphDefOptionsSetUniquifyNames(
+    TF_ImportGraphDefOptions opts, @Cast("unsigned char") byte uniquify_names);
+
+// If true, the specified prefix will be modified if it already exists as an
+// operation name or prefix in the graph. If false, a conflicting prefix will be
+// treated as an error. This option has no effect if no prefix is specified.
+public static native void TF_ImportGraphDefOptionsSetUniquifyPrefix(
+    TF_ImportGraphDefOptions opts, @Cast("unsigned char") byte uniquify_prefix);
+
+// Set any imported nodes with input `src_name:src_index` to have that input
+// replaced with `dst`. `src_name` refers to a node in the graph to be imported,
+// `dst` references a node already existing in the graph being imported into.
+// `src_name` is copied and has no lifetime requirements.
+public static native void TF_ImportGraphDefOptionsAddInputMapping(
+    TF_ImportGraphDefOptions opts, @Cast("const char*") BytePointer src_name, int src_index,
+    @ByVal TF_Output dst);
+public static native void TF_ImportGraphDefOptionsAddInputMapping(
+    TF_ImportGraphDefOptions opts, String src_name, int src_index,
+    @ByVal TF_Output dst);
+
+// Set any imported nodes with control input `src_name` to have that input
+// replaced with `dst`. `src_name` refers to a node in the graph to be imported,
+// `dst` references an operation already existing in the graph being imported
+// into. `src_name` is copied and has no lifetime requirements.
+public static native void TF_ImportGraphDefOptionsRemapControlDependency(
+    TF_ImportGraphDefOptions opts, @Cast("const char*") BytePointer src_name, TF_Operation dst);
+public static native void TF_ImportGraphDefOptionsRemapControlDependency(
+    TF_ImportGraphDefOptions opts, String src_name, TF_Operation dst);
+
+// Cause the imported graph to have a control dependency on `oper`. `oper`
+// should exist in the graph being imported into.
+public static native void TF_ImportGraphDefOptionsAddControlDependency(
+    TF_ImportGraphDefOptions opts, TF_Operation oper);
+
+// Add an output in `graph_def` to be returned via the `return_outputs` output
+// parameter of TF_GraphImportGraphDef(). If the output is remapped via an input
+// mapping, the corresponding existing tensor in `graph` will be returned.
+// `oper_name` is copied and has no lifetime requirements.
+public static native void TF_ImportGraphDefOptionsAddReturnOutput(
+    TF_ImportGraphDefOptions opts, @Cast("const char*") BytePointer oper_name, int index);
+public static native void TF_ImportGraphDefOptionsAddReturnOutput(
+    TF_ImportGraphDefOptions opts, String oper_name, int index);
+
+// Returns the number of return outputs added via
+// TF_ImportGraphDefOptionsAddReturnOutput().
+public static native int TF_ImportGraphDefOptionsNumReturnOutputs(
+    @Const TF_ImportGraphDefOptions opts);
+
+// Add an operation in `graph_def` to be returned via the `return_opers` output
+// parameter of TF_GraphImportGraphDef(). `oper_name` is copied and has no
+// lifetime requirements.
+public static native void TF_ImportGraphDefOptionsAddReturnOperation(
+    TF_ImportGraphDefOptions opts, @Cast("const char*") BytePointer oper_name);
+public static native void TF_ImportGraphDefOptionsAddReturnOperation(
+    TF_ImportGraphDefOptions opts, String oper_name);
+
+// Returns the number of return operations added via
+// TF_ImportGraphDefOptionsAddReturnOperation().
+public static native int TF_ImportGraphDefOptionsNumReturnOperations(
+    @Const TF_ImportGraphDefOptions opts);
+// Targeting ../TF_ImportGraphDefResults.java
+
+
+
+// Fetches the return outputs requested via
+// TF_ImportGraphDefOptionsAddReturnOutput(). The number of fetched outputs is
+// returned in `num_outputs`. The array of return outputs is returned in
+// `outputs`. `*outputs` is owned by and has the lifetime of `results`.
+public static native void TF_ImportGraphDefResultsReturnOutputs(
+    TF_ImportGraphDefResults results, IntPointer num_outputs, @Cast("TF_Output**") PointerPointer outputs);
+public static native void TF_ImportGraphDefResultsReturnOutputs(
+    TF_ImportGraphDefResults results, IntPointer num_outputs, @ByPtrPtr TF_Output outputs);
+public static native void TF_ImportGraphDefResultsReturnOutputs(
+    TF_ImportGraphDefResults results, IntBuffer num_outputs, @ByPtrPtr TF_Output outputs);
+public static native void TF_ImportGraphDefResultsReturnOutputs(
+    TF_ImportGraphDefResults results, int[] num_outputs, @ByPtrPtr TF_Output outputs);
+
+// Fetches the return operations requested via
+// TF_ImportGraphDefOptionsAddReturnOperation(). The number of fetched
+// operations is returned in `num_opers`. The array of return operations is
+// returned in `opers`. `*opers` is owned by and has the lifetime of `results`.
+public static native void TF_ImportGraphDefResultsReturnOperations(
+    TF_ImportGraphDefResults results, IntPointer num_opers, @Cast("TF_Operation***") @ByPtrPtr PointerPointer opers);
+public static native void TF_ImportGraphDefResultsReturnOperations(
+    TF_ImportGraphDefResults results, IntBuffer num_opers, @Cast("TF_Operation***") @ByPtrPtr PointerPointer opers);
+public static native void TF_ImportGraphDefResultsReturnOperations(
+    TF_ImportGraphDefResults results, int[] num_opers, @Cast("TF_Operation***") @ByPtrPtr PointerPointer opers);
+
+// Fetches any input mappings requested via
+// TF_ImportGraphDefOptionsAddInputMapping() that didn't appear in the GraphDef
+// and weren't used as input to any node in the imported graph def. The number
+// of fetched mappings is returned in `num_missing_unused_input_mappings`. The
+// array of each mapping's source node name is returned in `src_names`, and the
+// array of each mapping's source index is returned in `src_indexes`.
+//
+// `*src_names`, `*src_indexes`, and the memory backing each string in
+// `src_names` are owned by and have the lifetime of `results`.
+public static native void TF_ImportGraphDefResultsMissingUnusedInputMappings(
+    TF_ImportGraphDefResults results, IntPointer num_missing_unused_input_mappings,
+    @Cast("const char***") @ByPtrPtr PointerPointer src_names, @Cast("int**") PointerPointer src_indexes);
+public static native void TF_ImportGraphDefResultsMissingUnusedInputMappings(
+    TF_ImportGraphDefResults results, IntPointer num_missing_unused_input_mappings,
+    @Cast("const char***") @ByPtrPtr PointerPointer src_names, @ByPtrPtr IntPointer src_indexes);
+public static native void TF_ImportGraphDefResultsMissingUnusedInputMappings(
+    TF_ImportGraphDefResults results, IntBuffer num_missing_unused_input_mappings,
+    @Cast("const char***") @ByPtrPtr PointerPointer src_names, @ByPtrPtr IntBuffer src_indexes);
+public static native void TF_ImportGraphDefResultsMissingUnusedInputMappings(
+    TF_ImportGraphDefResults results, int[] num_missing_unused_input_mappings,
+    @Cast("const char***") @ByPtrPtr PointerPointer src_names, @ByPtrPtr int[] src_indexes);
+
+// Deletes a results object returned by TF_GraphImportGraphDefWithResults().
+public static native void TF_DeleteImportGraphDefResults(
+    TF_ImportGraphDefResults results);
+
+// Import the graph serialized in `graph_def` into `graph`.  Returns nullptr and
+// a bad status on error. Otherwise, returns a populated
+// TF_ImportGraphDefResults instance. The returned instance must be deleted via
+// TF_DeleteImportGraphDefResults().
+public static native TF_ImportGraphDefResults TF_GraphImportGraphDefWithResults(TF_Graph graph, @Const TF_Buffer graph_def,
+                                  @Const TF_ImportGraphDefOptions options,
+                                  TF_Status status);
+
+// Import the graph serialized in `graph_def` into `graph`.
+// Convenience function for when only return outputs are needed.
+//
+// `num_return_outputs` must be the number of return outputs added (i.e. the
+// result of TF_ImportGraphDefOptionsNumReturnOutputs()).  If
+// `num_return_outputs` is non-zero, `return_outputs` must be of length
+// `num_return_outputs`. Otherwise it can be null.
+public static native void TF_GraphImportGraphDefWithReturnOutputs(
+    TF_Graph graph, @Const TF_Buffer graph_def,
+    @Const TF_ImportGraphDefOptions options, TF_Output return_outputs,
+    int num_return_outputs, TF_Status status);
+
+// Import the graph serialized in `graph_def` into `graph`.
+// Convenience function for when no results are needed.
+public static native void TF_GraphImportGraphDef(
+    TF_Graph graph, @Const TF_Buffer graph_def,
+    @Const TF_ImportGraphDefOptions options, TF_Status status);
+
+// Adds a copy of function `func` and optionally its gradient function `grad`
+// to `g`. Once `func`/`grad` is added to `g`, it can be called by creating
+// an operation using the function's name.
+// Any changes to `func`/`grad` (including deleting it) done after this method
+// returns, won't affect the copy of `func`/`grad` in `g`.
+// If `func` or `grad` are already in `g`, TF_GraphCopyFunction has no
+// effect on them, but can establish the function->gradient relationship
+// between them if `func` does not already have a gradient. If `func` already
+// has a gradient different from `grad`, an error is returned.
+//
+// `func` must not be null.
+// If `grad` is null and `func` is not in `g`, `func` is added without a
+// gradient.
+// If `grad` is null and `func` is in `g`, TF_GraphCopyFunction is a noop.
+// `grad` must have appropriate signature as described in the doc of
+// GradientDef in tensorflow/core/framework/function.proto.
+//
+// If successful, status is set to OK and `func` and `grad` are added to `g`.
+// Otherwise, status is set to the encountered error and `g` is unmodified.
+public static native void TF_GraphCopyFunction(TF_Graph g,
+                                                @Const TF_Function func,
+                                                @Const TF_Function grad,
+                                                TF_Status status);
+
+// Returns the number of TF_Functions registered in `g`.
+public static native int TF_GraphNumFunctions(TF_Graph g);
+
+// Fills in `funcs` with the TF_Function* registered in `g`.
+// `funcs` must point to an array of TF_Function* of length at least
+// `max_func`. In usual usage, max_func should be set to the result of
+// TF_GraphNumFunctions(g). In this case, all the functions registered in
+// `g` will be returned. Else, an unspecified subset.
+//
+// If successful, returns the number of TF_Function* successfully set in
+// `funcs` and sets status to OK. The caller takes ownership of
+// all the returned TF_Functions. They must be deleted with TF_DeleteFunction.
+// On error, returns 0, sets status to the encountered error, and the contents
+// of funcs will be undefined.
+public static native int TF_GraphGetFunctions(TF_Graph g, @Cast("TF_Function**") PointerPointer funcs,
+                                               int max_func, TF_Status status);
+public static native int TF_GraphGetFunctions(TF_Graph g, @ByPtrPtr TF_Function funcs,
+                                               int max_func, TF_Status status);
+
+// Note: The following function may fail on very large protos in the future.
+
+public static native void TF_OperationToNodeDef(TF_Operation oper,
+                                                 TF_Buffer output_node_def,
+                                                 TF_Status status);
+// Targeting ../TF_WhileParams.java
+
+
+
+// Creates a TF_WhileParams for creating a while loop in `g`. `inputs` are
+// outputs that already exist in `g` used as initial values for the loop
+// variables.
+//
+// The returned TF_WhileParams will have all fields initialized except
+// `cond_output`, `body_outputs`, and `name`. The `body_outputs` buffer will be
+// allocated to size `ninputs`. The caller should build `cond_graph` and
+// `body_graph` starting from the inputs, and store the final outputs in
+// `cond_output` and `body_outputs`.
+//
+// If `status` is OK, the caller must call either TF_FinishWhile or
+// TF_AbortWhile on the returned TF_WhileParams. If `status` isn't OK, the
+// returned TF_WhileParams is not valid, and the caller should not call
+// TF_FinishWhile() or TF_AbortWhile().
+//
+// Missing functionality (TODO):
+// - Gradients
+// - Reference-type inputs
+// - Directly referencing external tensors from the cond/body graphs (this is
+//   possible in the Python API)
+public static native @ByVal TF_WhileParams TF_NewWhile(TF_Graph g, TF_Output inputs,
+                                                 int ninputs,
+                                                 TF_Status status);
+
+// Builds the while loop specified by `params` and returns the output tensors of
+// the while loop in `outputs`. `outputs` should be allocated to size
+// `params.ninputs`.
+//
+// `params` is no longer valid once this returns.
+//
+// Either this or TF_AbortWhile() must be called after a successful
+// TF_NewWhile() call.
+public static native void TF_FinishWhile(@Const TF_WhileParams params,
+                                          TF_Status status,
+                                          TF_Output outputs);
+
+// Frees `params`s resources without building a while loop. `params` is no
+// longer valid after this returns. Either this or TF_FinishWhile() must be
+// called after a successful TF_NewWhile() call.
+public static native void TF_AbortWhile(@Const TF_WhileParams params);
+
+// Adds operations to compute the partial derivatives of sum of `y`s w.r.t `x`s,
+// i.e., d(y_1 + y_2 + ...)/dx_1, d(y_1 + y_2 + ...)/dx_2...
+//
+// `dx` are used as initial gradients (which represent the symbolic partial
+// derivatives of some loss function `L` w.r.t. `y`).
+// `dx` must be nullptr or have size `ny`.
+// If `dx` is nullptr, the implementation will use dx of `OnesLike` for all
+// shapes in `y`.
+// The partial derivatives are returned in `dy`. `dy` should be allocated to
+// size `nx`.
+//
+// Gradient nodes are automatically named under the "gradients/" prefix. To
+// guarantee name uniqueness, subsequent calls to the same graph will
+// append an incremental tag to the prefix: "gradients_1/", "gradients_2/", ...
+// See TF_AddGradientsWithPrefix, which provides a means to specify a custom
+// name prefix for operations added to a graph to compute the gradients.
+//
+// WARNING: This function does not yet support all the gradients that python
+// supports. See
+// https://www.tensorflow.org/code/tensorflow/cc/gradients/README.md
+// for instructions on how to add C++ more gradients.
+public static native void TF_AddGradients(TF_Graph g, TF_Output y, int ny,
+                                    TF_Output x, int nx, TF_Output dx,
+                                    TF_Status status, TF_Output dy);
+
+// Adds operations to compute the partial derivatives of sum of `y`s w.r.t `x`s,
+// i.e., d(y_1 + y_2 + ...)/dx_1, d(y_1 + y_2 + ...)/dx_2...
+// This is a variant of TF_AddGradients that allows to caller to pass a custom
+// name prefix to the operations added to a graph to compute the gradients.
+//
+// `dx` are used as initial gradients (which represent the symbolic partial
+// derivatives of some loss function `L` w.r.t. `y`).
+// `dx` must be nullptr or have size `ny`.
+// If `dx` is nullptr, the implementation will use dx of `OnesLike` for all
+// shapes in `y`.
+// The partial derivatives are returned in `dy`. `dy` should be allocated to
+// size `nx`.
+// `prefix` names the scope into which all gradients operations are being added.
+// `prefix` must be unique within the provided graph otherwise this operation
+// will fail. If `prefix` is nullptr, the default prefixing behaviour takes
+// place, see TF_AddGradients for more details.
+//
+// WARNING: This function does not yet support all the gradients that python
+// supports. See
+// https://www.tensorflow.org/code/tensorflow/cc/gradients/README.md
+// for instructions on how to add C++ more gradients.
+public static native void TF_AddGradientsWithPrefix(TF_Graph g, @Cast("const char*") BytePointer prefix,
+                                              TF_Output y, int ny,
+                                              TF_Output x, int nx,
+                                              TF_Output dx, TF_Status status,
+                                              TF_Output dy);
+public static native void TF_AddGradientsWithPrefix(TF_Graph g, String prefix,
+                                              TF_Output y, int ny,
+                                              TF_Output x, int nx,
+                                              TF_Output dx, TF_Status status,
+                                              TF_Output dy);
+
+// Create a TF_Function from a TF_Graph
+//
+// Params:
+//  fn_body - the graph whose operations (or subset of whose operations) will be
+//            converted to TF_Function.
+//  fn_name - the name of the new TF_Function. Should match the operation
+//            name (OpDef.name) regexp [A-Z][A-Za-z0-9_.\\-/]*.
+//            If `append_hash_to_fn_name` is false, `fn_name` must be distinct
+//            from other function and operation names (at least those
+//            registered in graphs where this function will be used).
+//  append_hash_to_fn_name - Must be 0 or 1. If set to 1, the actual name
+//                           of the function will be `fn_name` appended with
+//                           '_<hash_of_this_function's_definition>'.
+//                           If set to 0, the function's name will be `fn_name`.
+//  num_opers - `num_opers` contains the number of elements in the `opers` array
+//              or a special value of -1 meaning that no array is given.
+//              The distinction between an empty array of operations and no
+//              array of operations is necessary to distinguish the case of
+//              creating a function with no body (e.g. identity or permutation)
+//              and the case of creating a function whose body contains all
+//              the nodes in the graph (except for the automatic skipping, see
+//              below).
+//  opers - Array of operations to become the body of the function or null.
+//          - If no array is given (`num_opers`  = -1), all the
+//          operations in `fn_body` will become part of the function
+//          except operations referenced in `inputs`. These operations
+//          must have a single output (these operations are typically
+//          placeholders created for the sole purpose of representing
+//          an input. We can relax this constraint if there are
+//          compelling use cases).
+//          - If an array is given (`num_opers` >= 0), all operations
+//          in it will become part of the function. In particular, no
+//          automatic skipping of dummy input operations is performed.
+//  ninputs - number of elements in `inputs` array
+//  inputs - array of TF_Outputs that specify the inputs to the function.
+//           If `ninputs` is zero (the function takes no inputs), `inputs`
+//           can be null. The names used for function inputs are normalized
+//           names of the operations (usually placeholders) pointed to by
+//           `inputs`. These operation names should start with a letter.
+//           Normalization will convert all letters to lowercase and
+//           non-alphanumeric characters to '_' to make resulting names match
+//           the "[a-z][a-z0-9_]*" pattern for operation argument names.
+//           `inputs` cannot contain the same tensor twice.
+//  noutputs - number of elements in `outputs` array
+//  outputs - array of TF_Outputs that specify the outputs of the function.
+//            If `noutputs` is zero (the function returns no outputs), `outputs`
+//            can be null. `outputs` can contain the same tensor more than once.
+//  output_names - The names of the function's outputs. `output_names` array
+//                 must either have the same length as `outputs`
+//                 (i.e. `noutputs`) or be null. In the former case,
+//                 the names should match the regular expression for ArgDef
+//                 names - "[a-z][a-z0-9_]*". In the latter case,
+//                 names for outputs will be generated automatically.
+//  opts - various options for the function, e.g. XLA's inlining control.
+//  description - optional human-readable description of this function.
+//  status - Set to OK on success and an appropriate error on failure.
+//
+// Note that when the same TF_Output is listed as both an input and an output,
+// the corresponding function's output will equal to this input,
+// instead of the original node's output.
+//
+// Callers must also satisfy the following constraints:
+// - `inputs` cannot refer to TF_Outputs within a control flow context. For
+//   example, one cannot use the output of "switch" node as input.
+// - `inputs` and `outputs` cannot have reference types. Reference types are
+//   not exposed through C API and are being replaced with Resources. We support
+//   reference types inside function's body to support legacy code. Do not
+//   use them in new code.
+// - Every node in the function's body must have all of its inputs (including
+//   control inputs). In other words, for every node in the body, each input
+//   must be either listed in `inputs` or must come from another node in
+//   the body. In particular, it is an error to have a control edge going from
+//   a node outside of the body into a node in the body. This applies to control
+//   edges going from nodes referenced in `inputs` to nodes in the body when
+//   the former nodes are not in the body (automatically skipped or not
+//   included in explicitly specified body).
+//
+// Returns:
+//  On success, a newly created TF_Function instance. It must be deleted by
+//  calling TF_DeleteFunction.
+//
+//  On failure, null.
+public static native TF_Function TF_GraphToFunction(
+    @Const TF_Graph fn_body, @Cast("const char*") BytePointer fn_name,
+    @Cast("unsigned char") byte append_hash_to_fn_name, int num_opers,
+    @Cast("const TF_Operation*const*") PointerPointer opers, int ninputs, @Const TF_Output inputs,
+    int noutputs, @Const TF_Output outputs, @Cast("const char*const*") PointerPointer output_names,
+    @Const TF_FunctionOptions opts, @Cast("const char*") BytePointer description, TF_Status status);
+public static native TF_Function TF_GraphToFunction(
+    @Const TF_Graph fn_body, @Cast("const char*") BytePointer fn_name,
+    @Cast("unsigned char") byte append_hash_to_fn_name, int num_opers,
+    @Const @ByPtrPtr TF_Operation opers, int ninputs, @Const TF_Output inputs,
+    int noutputs, @Const TF_Output outputs, @Cast("const char*const*") @ByPtrPtr BytePointer output_names,
+    @Const TF_FunctionOptions opts, @Cast("const char*") BytePointer description, TF_Status status);
+public static native TF_Function TF_GraphToFunction(
+    @Const TF_Graph fn_body, String fn_name,
+    @Cast("unsigned char") byte append_hash_to_fn_name, int num_opers,
+    @Const @ByPtrPtr TF_Operation opers, int ninputs, @Const TF_Output inputs,
+    int noutputs, @Const TF_Output outputs, @Cast("const char*const*") @ByPtrPtr ByteBuffer output_names,
+    @Const TF_FunctionOptions opts, String description, TF_Status status);
+public static native TF_Function TF_GraphToFunction(
+    @Const TF_Graph fn_body, @Cast("const char*") BytePointer fn_name,
+    @Cast("unsigned char") byte append_hash_to_fn_name, int num_opers,
+    @Const @ByPtrPtr TF_Operation opers, int ninputs, @Const TF_Output inputs,
+    int noutputs, @Const TF_Output outputs, @Cast("const char*const*") @ByPtrPtr byte[] output_names,
+    @Const TF_FunctionOptions opts, @Cast("const char*") BytePointer description, TF_Status status);
+public static native TF_Function TF_GraphToFunction(
+    @Const TF_Graph fn_body, String fn_name,
+    @Cast("unsigned char") byte append_hash_to_fn_name, int num_opers,
+    @Const @ByPtrPtr TF_Operation opers, int ninputs, @Const TF_Output inputs,
+    int noutputs, @Const TF_Output outputs, @Cast("const char*const*") @ByPtrPtr BytePointer output_names,
+    @Const TF_FunctionOptions opts, String description, TF_Status status);
+public static native TF_Function TF_GraphToFunction(
+    @Const TF_Graph fn_body, @Cast("const char*") BytePointer fn_name,
+    @Cast("unsigned char") byte append_hash_to_fn_name, int num_opers,
+    @Const @ByPtrPtr TF_Operation opers, int ninputs, @Const TF_Output inputs,
+    int noutputs, @Const TF_Output outputs, @Cast("const char*const*") @ByPtrPtr ByteBuffer output_names,
+    @Const TF_FunctionOptions opts, @Cast("const char*") BytePointer description, TF_Status status);
+public static native TF_Function TF_GraphToFunction(
+    @Const TF_Graph fn_body, String fn_name,
+    @Cast("unsigned char") byte append_hash_to_fn_name, int num_opers,
+    @Const @ByPtrPtr TF_Operation opers, int ninputs, @Const TF_Output inputs,
+    int noutputs, @Const TF_Output outputs, @Cast("const char*const*") @ByPtrPtr byte[] output_names,
+    @Const TF_FunctionOptions opts, String description, TF_Status status);
+
+// Similar to TF_GraphToFunction but allows specifying control outputs of the
+// function.
+//
+//  The arguments of TF_GraphToFunction have the same meaning, but the new
+//  arguments are as follows:
+//
+//    ncontrol_outputs: Number of control outputs of the function.
+//    control_outputs: vector of TF_Operation objects to be marked as control
+//      outputs of the function. Operations marked as control outputs are
+//      guaranteed to execute.
+//    control_output_names: Optional. If not nullptr, vector of strings, one
+//      per control output, with their names to be added to the function's
+//      OpDef.
+public static native TF_Function TF_GraphToFunctionWithControlOutputs(
+    @Const TF_Graph fn_body, @Cast("const char*") BytePointer fn_name,
+    @Cast("unsigned char") byte append_hash_to_fn_name, int num_opers,
+    @Cast("const TF_Operation*const*") PointerPointer opers, int ninputs, @Const TF_Output inputs,
+    int noutputs, @Const TF_Output outputs, @Cast("const char*const*") PointerPointer output_names,
+    int ncontrol_outputs, @Cast("const TF_Operation*const*") PointerPointer control_outputs,
+    @Cast("const char*const*") PointerPointer control_output_names, @Const TF_FunctionOptions opts,
+    @Cast("const char*") BytePointer description, TF_Status status);
+public static native TF_Function TF_GraphToFunctionWithControlOutputs(
+    @Const TF_Graph fn_body, @Cast("const char*") BytePointer fn_name,
+    @Cast("unsigned char") byte append_hash_to_fn_name, int num_opers,
+    @Const @ByPtrPtr TF_Operation opers, int ninputs, @Const TF_Output inputs,
+    int noutputs, @Const TF_Output outputs, @Cast("const char*const*") @ByPtrPtr BytePointer output_names,
+    int ncontrol_outputs, @Const @ByPtrPtr TF_Operation control_outputs,
+    @Cast("const char*const*") @ByPtrPtr BytePointer control_output_names, @Const TF_FunctionOptions opts,
+    @Cast("const char*") BytePointer description, TF_Status status);
+public static native TF_Function TF_GraphToFunctionWithControlOutputs(
+    @Const TF_Graph fn_body, String fn_name,
+    @Cast("unsigned char") byte append_hash_to_fn_name, int num_opers,
+    @Const @ByPtrPtr TF_Operation opers, int ninputs, @Const TF_Output inputs,
+    int noutputs, @Const TF_Output outputs, @Cast("const char*const*") @ByPtrPtr ByteBuffer output_names,
+    int ncontrol_outputs, @Const @ByPtrPtr TF_Operation control_outputs,
+    @Cast("const char*const*") @ByPtrPtr ByteBuffer control_output_names, @Const TF_FunctionOptions opts,
+    String description, TF_Status status);
+public static native TF_Function TF_GraphToFunctionWithControlOutputs(
+    @Const TF_Graph fn_body, @Cast("const char*") BytePointer fn_name,
+    @Cast("unsigned char") byte append_hash_to_fn_name, int num_opers,
+    @Const @ByPtrPtr TF_Operation opers, int ninputs, @Const TF_Output inputs,
+    int noutputs, @Const TF_Output outputs, @Cast("const char*const*") @ByPtrPtr byte[] output_names,
+    int ncontrol_outputs, @Const @ByPtrPtr TF_Operation control_outputs,
+    @Cast("const char*const*") @ByPtrPtr byte[] control_output_names, @Const TF_FunctionOptions opts,
+    @Cast("const char*") BytePointer description, TF_Status status);
+public static native TF_Function TF_GraphToFunctionWithControlOutputs(
+    @Const TF_Graph fn_body, String fn_name,
+    @Cast("unsigned char") byte append_hash_to_fn_name, int num_opers,
+    @Const @ByPtrPtr TF_Operation opers, int ninputs, @Const TF_Output inputs,
+    int noutputs, @Const TF_Output outputs, @Cast("const char*const*") @ByPtrPtr BytePointer output_names,
+    int ncontrol_outputs, @Const @ByPtrPtr TF_Operation control_outputs,
+    @Cast("const char*const*") @ByPtrPtr BytePointer control_output_names, @Const TF_FunctionOptions opts,
+    String description, TF_Status status);
+public static native TF_Function TF_GraphToFunctionWithControlOutputs(
+    @Const TF_Graph fn_body, @Cast("const char*") BytePointer fn_name,
+    @Cast("unsigned char") byte append_hash_to_fn_name, int num_opers,
+    @Const @ByPtrPtr TF_Operation opers, int ninputs, @Const TF_Output inputs,
+    int noutputs, @Const TF_Output outputs, @Cast("const char*const*") @ByPtrPtr ByteBuffer output_names,
+    int ncontrol_outputs, @Const @ByPtrPtr TF_Operation control_outputs,
+    @Cast("const char*const*") @ByPtrPtr ByteBuffer control_output_names, @Const TF_FunctionOptions opts,
+    @Cast("const char*") BytePointer description, TF_Status status);
+public static native TF_Function TF_GraphToFunctionWithControlOutputs(
+    @Const TF_Graph fn_body, String fn_name,
+    @Cast("unsigned char") byte append_hash_to_fn_name, int num_opers,
+    @Const @ByPtrPtr TF_Operation opers, int ninputs, @Const TF_Output inputs,
+    int noutputs, @Const TF_Output outputs, @Cast("const char*const*") @ByPtrPtr byte[] output_names,
+    int ncontrol_outputs, @Const @ByPtrPtr TF_Operation control_outputs,
+    @Cast("const char*const*") @ByPtrPtr byte[] control_output_names, @Const TF_FunctionOptions opts,
+    String description, TF_Status status);
+
+// Returns the name of the graph function.
+// The return value points to memory that is only usable until the next
+// mutation to *func.
+public static native @Cast("const char*") BytePointer TF_FunctionName(TF_Function func);
+
+// Write out a serialized representation of `func` (as a FunctionDef protocol
+// message) to `output_func_def` (allocated by TF_NewBuffer()).
+// `output_func_def`'s underlying buffer will be freed when TF_DeleteBuffer()
+// is called.
+//
+// May fail on very large graphs in the future.
+public static native void TF_FunctionToFunctionDef(TF_Function func,
+                                                    TF_Buffer output_func_def,
+                                                    TF_Status status);
+
+// Construct and return the function whose FunctionDef representation is
+// serialized in `proto`. `proto_len` must equal the number of bytes
+// pointed to by `proto`.
+// Returns:
+//  On success, a newly created TF_Function instance. It must be deleted by
+//  calling TF_DeleteFunction.
+//
+//  On failure, null.
+public static native TF_Function TF_FunctionImportFunctionDef(
+    @Const Pointer proto, @Cast("size_t") long proto_len, TF_Status status);
+
+// Sets function attribute named `attr_name` to value stored in `proto`.
+// If this attribute is already set to another value, it is overridden.
+// `proto` should point to a sequence of bytes of length `proto_len`
+// representing a binary serialization of an AttrValue protocol
+// buffer.
+public static native void TF_FunctionSetAttrValueProto(TF_Function func,
+                                                        @Cast("const char*") BytePointer attr_name,
+                                                        @Const Pointer proto,
+                                                        @Cast("size_t") long proto_len,
+                                                        TF_Status status);
+public static native void TF_FunctionSetAttrValueProto(TF_Function func,
+                                                        String attr_name,
+                                                        @Const Pointer proto,
+                                                        @Cast("size_t") long proto_len,
+                                                        TF_Status status);
+
+// Sets `output_attr_value` to the binary-serialized AttrValue proto
+// representation of the value of the `attr_name` attr of `func`.
+// If `attr_name` attribute is not present, status is set to an error.
+public static native void TF_FunctionGetAttrValueProto(
+    TF_Function func, @Cast("const char*") BytePointer attr_name, TF_Buffer output_attr_value,
+    TF_Status status);
+public static native void TF_FunctionGetAttrValueProto(
+    TF_Function func, String attr_name, TF_Buffer output_attr_value,
+    TF_Status status);
+
+// Frees the memory used by the `func` struct.
+// TF_DeleteFunction is a noop if `func` is null.
+// Deleting a function does not remove it from any graphs it was copied to.
+public static native void TF_DeleteFunction(TF_Function func);
+
+// Attempts to evaluate `output`. This will only be possible if `output` doesn't
+// depend on any graph inputs (this function is safe to call if this isn't the
+// case though).
+//
+// If the evaluation is successful, this function returns true and `output`s
+// value is returned in `result`. Otherwise returns false. An error status is
+// returned if something is wrong with the graph or input. Note that this may
+// return false even if no error status is set.
+public static native @Cast("unsigned char") byte TF_TryEvaluateConstant(TF_Graph graph,
+                                                           @ByVal TF_Output output,
+                                                           @Cast("TF_Tensor**") PointerPointer result,
+                                                           TF_Status status);
+public static native @Cast("unsigned char") byte TF_TryEvaluateConstant(TF_Graph graph,
+                                                           @ByVal TF_Output output,
+                                                           @ByPtrPtr TF_Tensor result,
+                                                           TF_Status status);
+// Targeting ../TF_Session.java
+
+
+
+// Return a new execution session with the associated graph, or NULL on
+// error. Does not take ownership of any input parameters.
+//
+// *`graph` must be a valid graph (not deleted or nullptr). `graph` will be be
+// kept alive for the lifetime of the returned TF_Session. New nodes can still
+// be added to `graph` after this call.
+public static native TF_Session TF_NewSession(TF_Graph graph,
+                                                @Const TF_SessionOptions opts,
+                                                TF_Status status);
+
+// This function creates a new TF_Session (which is created on success) using
+// `session_options`, and then initializes state (restoring tensors and other
+// assets) using `run_options`.
+//
+// Any NULL and non-NULL value combinations for (`run_options, `meta_graph_def`)
+// are valid.
+//
+// - `export_dir` must be set to the path of the exported SavedModel.
+// - `tags` must include the set of tags used to identify one MetaGraphDef in
+//    the SavedModel.
+// - `graph` must be a graph newly allocated with TF_NewGraph().
+//
+// If successful, populates `graph` with the contents of the Graph and
+// `meta_graph_def` with the MetaGraphDef of the loaded model.
+public static native TF_Session TF_LoadSessionFromSavedModel(
+    @Const TF_SessionOptions session_options, @Const TF_Buffer run_options,
+    @Cast("const char*") BytePointer export_dir, @Cast("const char*const*") PointerPointer tags, int tags_len,
+    TF_Graph graph, TF_Buffer meta_graph_def, TF_Status status);
+public static native TF_Session TF_LoadSessionFromSavedModel(
+    @Const TF_SessionOptions session_options, @Const TF_Buffer run_options,
+    @Cast("const char*") BytePointer export_dir, @Cast("const char*const*") @ByPtrPtr BytePointer tags, int tags_len,
+    TF_Graph graph, TF_Buffer meta_graph_def, TF_Status status);
+public static native TF_Session TF_LoadSessionFromSavedModel(
+    @Const TF_SessionOptions session_options, @Const TF_Buffer run_options,
+    String export_dir, @Cast("const char*const*") @ByPtrPtr ByteBuffer tags, int tags_len,
+    TF_Graph graph, TF_Buffer meta_graph_def, TF_Status status);
+public static native TF_Session TF_LoadSessionFromSavedModel(
+    @Const TF_SessionOptions session_options, @Const TF_Buffer run_options,
+    @Cast("const char*") BytePointer export_dir, @Cast("const char*const*") @ByPtrPtr byte[] tags, int tags_len,
+    TF_Graph graph, TF_Buffer meta_graph_def, TF_Status status);
+public static native TF_Session TF_LoadSessionFromSavedModel(
+    @Const TF_SessionOptions session_options, @Const TF_Buffer run_options,
+    String export_dir, @Cast("const char*const*") @ByPtrPtr BytePointer tags, int tags_len,
+    TF_Graph graph, TF_Buffer meta_graph_def, TF_Status status);
+public static native TF_Session TF_LoadSessionFromSavedModel(
+    @Const TF_SessionOptions session_options, @Const TF_Buffer run_options,
+    @Cast("const char*") BytePointer export_dir, @Cast("const char*const*") @ByPtrPtr ByteBuffer tags, int tags_len,
+    TF_Graph graph, TF_Buffer meta_graph_def, TF_Status status);
+public static native TF_Session TF_LoadSessionFromSavedModel(
+    @Const TF_SessionOptions session_options, @Const TF_Buffer run_options,
+    String export_dir, @Cast("const char*const*") @ByPtrPtr byte[] tags, int tags_len,
+    TF_Graph graph, TF_Buffer meta_graph_def, TF_Status status);
+
+// Close a session.
+//
+// Contacts any other processes associated with the session, if applicable.
+// May not be called after TF_DeleteSession().
+public static native void TF_CloseSession(TF_Session arg0, TF_Status status);
+
+// Destroy a session object.
+//
+// Even if error information is recorded in *status, this call discards all
+// local resources associated with the session.  The session may not be used
+// during or after this call (and the session drops its reference to the
+// corresponding graph).
+public static native void TF_DeleteSession(TF_Session arg0, TF_Status status);
+
+// Run the graph associated with the session starting with the supplied inputs
+// (inputs[0,ninputs-1] with corresponding values in input_values[0,ninputs-1]).
+//
+// Any NULL and non-NULL value combinations for (`run_options`,
+// `run_metadata`) are valid.
+//
+//    - `run_options` may be NULL, in which case it will be ignored; or
+//      non-NULL, in which case it must point to a `TF_Buffer` containing the
+//      serialized representation of a `RunOptions` protocol buffer.
+//    - `run_metadata` may be NULL, in which case it will be ignored; or
+//      non-NULL, in which case it must point to an empty, freshly allocated
+//      `TF_Buffer` that may be updated to contain the serialized representation
+//      of a `RunMetadata` protocol buffer.
+//
+// The caller retains ownership of `input_values` (which can be deleted using
+// TF_DeleteTensor). The caller also retains ownership of `run_options` and/or
+// `run_metadata` (when not NULL) and should manually call TF_DeleteBuffer on
+// them.
+//
+// On success, the tensors corresponding to outputs[0,noutputs-1] are placed in
+// output_values[]. Ownership of the elements of output_values[] is transferred
+// to the caller, which must eventually call TF_DeleteTensor on them.
+//
+// On failure, output_values[] contains NULLs.
+public static native void TF_SessionRun(
+    TF_Session session,
+    @Const TF_Buffer run_options,
+    @Const TF_Output inputs, @Cast("TF_Tensor*const*") PointerPointer input_values, int ninputs,
+    @Const TF_Output outputs, @Cast("TF_Tensor**") PointerPointer output_values, int noutputs,
+    @Cast("const TF_Operation*const*") PointerPointer target_opers, int ntargets,
+    TF_Buffer run_metadata,
+    TF_Status arg11);
+public static native void TF_SessionRun(
+    TF_Session session,
+    @Const TF_Buffer run_options,
+    @Const TF_Output inputs, @ByPtrPtr TF_Tensor input_values, int ninputs,
+    @Const TF_Output outputs, @ByPtrPtr TF_Tensor output_values, int noutputs,
+    @Const @ByPtrPtr TF_Operation target_opers, int ntargets,
+    TF_Buffer run_metadata,
+    TF_Status arg11);
+
+// Set up the graph with the intended feeds (inputs) and fetches (outputs) for a
+// sequence of partial run calls.
+//
+// On success, returns a handle that is used for subsequent PRun calls. The
+// handle should be deleted with TF_DeletePRunHandle when it is no longer
+// needed.
+//
+// On failure, out_status contains a tensorflow::Status with an error
+// message. *handle is set to nullptr.
+public static native void TF_SessionPRunSetup(
+    TF_Session arg0,
+    @Const TF_Output inputs, int ninputs,
+    @Const TF_Output outputs, int noutputs,
+    @Cast("const TF_Operation*const*") PointerPointer target_opers, int ntargets,
+    @Cast("const char**") PointerPointer handle,
+    TF_Status arg8);
+public static native void TF_SessionPRunSetup(
+    TF_Session arg0,
+    @Const TF_Output inputs, int ninputs,
+    @Const TF_Output outputs, int noutputs,
+    @Const @ByPtrPtr TF_Operation target_opers, int ntargets,
+    @Cast("const char**") @ByPtrPtr BytePointer handle,
+    TF_Status arg8);
+public static native void TF_SessionPRunSetup(
+    TF_Session arg0,
+    @Const TF_Output inputs, int ninputs,
+    @Const TF_Output outputs, int noutputs,
+    @Const @ByPtrPtr TF_Operation target_opers, int ntargets,
+    @Cast("const char**") @ByPtrPtr ByteBuffer handle,
+    TF_Status arg8);
+public static native void TF_SessionPRunSetup(
+    TF_Session arg0,
+    @Const TF_Output inputs, int ninputs,
+    @Const TF_Output outputs, int noutputs,
+    @Const @ByPtrPtr TF_Operation target_opers, int ntargets,
+    @Cast("const char**") @ByPtrPtr byte[] handle,
+    TF_Status arg8);
+
+// Continue to run the graph with additional feeds and fetches. The
+// execution state is uniquely identified by the handle.
+public static native void TF_SessionPRun(
+    TF_Session arg0, @Cast("const char*") BytePointer handle,
+    @Const TF_Output inputs, @Cast("TF_Tensor*const*") PointerPointer input_values, int ninputs,
+    @Const TF_Output outputs, @Cast("TF_Tensor**") PointerPointer output_values, int noutputs,
+    @Cast("const TF_Operation*const*") PointerPointer target_opers, int ntargets,
+    TF_Status arg10);
+public static native void TF_SessionPRun(
+    TF_Session arg0, @Cast("const char*") BytePointer handle,
+    @Const TF_Output inputs, @ByPtrPtr TF_Tensor input_values, int ninputs,
+    @Const TF_Output outputs, @ByPtrPtr TF_Tensor output_values, int noutputs,
+    @Const @ByPtrPtr TF_Operation target_opers, int ntargets,
+    TF_Status arg10);
+public static native void TF_SessionPRun(
+    TF_Session arg0, String handle,
+    @Const TF_Output inputs, @ByPtrPtr TF_Tensor input_values, int ninputs,
+    @Const TF_Output outputs, @ByPtrPtr TF_Tensor output_values, int noutputs,
+    @Const @ByPtrPtr TF_Operation target_opers, int ntargets,
+    TF_Status arg10);
+
+// Deletes a handle allocated by TF_SessionPRunSetup.
+// Once called, no more calls to TF_SessionPRun should be made.
+public static native void TF_DeletePRunHandle(@Cast("const char*") BytePointer handle);
+public static native void TF_DeletePRunHandle(String handle);
+// Targeting ../TF_DeprecatedSession.java
+
+
+
+public static native TF_DeprecatedSession TF_NewDeprecatedSession(
+    @Const TF_SessionOptions arg0, TF_Status status);
+public static native void TF_CloseDeprecatedSession(TF_DeprecatedSession arg0,
+                                                     TF_Status status);
+public static native void TF_DeleteDeprecatedSession(TF_DeprecatedSession arg0,
+                                                      TF_Status status);
+public static native void TF_Reset(@Const TF_SessionOptions opt,
+                                    @Cast("const char**") PointerPointer containers, int ncontainers,
+                                    TF_Status status);
+public static native void TF_Reset(@Const TF_SessionOptions opt,
+                                    @Cast("const char**") @ByPtrPtr BytePointer containers, int ncontainers,
+                                    TF_Status status);
+public static native void TF_Reset(@Const TF_SessionOptions opt,
+                                    @Cast("const char**") @ByPtrPtr ByteBuffer containers, int ncontainers,
+                                    TF_Status status);
+public static native void TF_Reset(@Const TF_SessionOptions opt,
+                                    @Cast("const char**") @ByPtrPtr byte[] containers, int ncontainers,
+                                    TF_Status status);
+// Treat the bytes proto[0,proto_len-1] as a serialized GraphDef and
+// add the nodes in that GraphDef to the graph for the session.
+//
+// Prefer use of TF_Session and TF_GraphImportGraphDef over this.
+public static native void TF_ExtendGraph(TF_DeprecatedSession arg0,
+                                          @Const Pointer proto, @Cast("size_t") long proto_len,
+                                          TF_Status arg3);
+
+// See TF_SessionRun() above.
+public static native void TF_Run(TF_DeprecatedSession arg0,
+                                  @Const TF_Buffer run_options,
+                                  @Cast("const char**") PointerPointer input_names, @Cast("TF_Tensor**") PointerPointer inputs,
+                                  int ninputs, @Cast("const char**") PointerPointer output_names,
+                                  @Cast("TF_Tensor**") PointerPointer outputs, int noutputs,
+                                  @Cast("const char**") PointerPointer target_oper_names, int ntargets,
+                                  TF_Buffer run_metadata, TF_Status arg11);
+public static native void TF_Run(TF_DeprecatedSession arg0,
+                                  @Const TF_Buffer run_options,
+                                  @Cast("const char**") @ByPtrPtr BytePointer input_names, @ByPtrPtr TF_Tensor inputs,
+                                  int ninputs, @Cast("const char**") @ByPtrPtr BytePointer output_names,
+                                  @ByPtrPtr TF_Tensor outputs, int noutputs,
+                                  @Cast("const char**") @ByPtrPtr BytePointer target_oper_names, int ntargets,
+                                  TF_Buffer run_metadata, TF_Status arg11);
+public static native void TF_Run(TF_DeprecatedSession arg0,
+                                  @Const TF_Buffer run_options,
+                                  @Cast("const char**") @ByPtrPtr ByteBuffer input_names, @ByPtrPtr TF_Tensor inputs,
+                                  int ninputs, @Cast("const char**") @ByPtrPtr ByteBuffer output_names,
+                                  @ByPtrPtr TF_Tensor outputs, int noutputs,
+                                  @Cast("const char**") @ByPtrPtr ByteBuffer target_oper_names, int ntargets,
+                                  TF_Buffer run_metadata, TF_Status arg11);
+public static native void TF_Run(TF_DeprecatedSession arg0,
+                                  @Const TF_Buffer run_options,
+                                  @Cast("const char**") @ByPtrPtr byte[] input_names, @ByPtrPtr TF_Tensor inputs,
+                                  int ninputs, @Cast("const char**") @ByPtrPtr byte[] output_names,
+                                  @ByPtrPtr TF_Tensor outputs, int noutputs,
+                                  @Cast("const char**") @ByPtrPtr byte[] target_oper_names, int ntargets,
+                                  TF_Buffer run_metadata, TF_Status arg11);
+
+// See TF_SessionPRunSetup() above.
+public static native void TF_PRunSetup(TF_DeprecatedSession arg0,
+                                        @Cast("const char**") PointerPointer input_names, int ninputs,
+                                        @Cast("const char**") PointerPointer output_names, int noutputs,
+                                        @Cast("const char**") PointerPointer target_oper_names,
+                                        int ntargets, @Cast("const char**") PointerPointer handle,
+                                        TF_Status arg8);
+public static native void TF_PRunSetup(TF_DeprecatedSession arg0,
+                                        @Cast("const char**") @ByPtrPtr BytePointer input_names, int ninputs,
+                                        @Cast("const char**") @ByPtrPtr BytePointer output_names, int noutputs,
+                                        @Cast("const char**") @ByPtrPtr BytePointer target_oper_names,
+                                        int ntargets, @Cast("const char**") @ByPtrPtr BytePointer handle,
+                                        TF_Status arg8);
+public static native void TF_PRunSetup(TF_DeprecatedSession arg0,
+                                        @Cast("const char**") @ByPtrPtr ByteBuffer input_names, int ninputs,
+                                        @Cast("const char**") @ByPtrPtr ByteBuffer output_names, int noutputs,
+                                        @Cast("const char**") @ByPtrPtr ByteBuffer target_oper_names,
+                                        int ntargets, @Cast("const char**") @ByPtrPtr ByteBuffer handle,
+                                        TF_Status arg8);
+public static native void TF_PRunSetup(TF_DeprecatedSession arg0,
+                                        @Cast("const char**") @ByPtrPtr byte[] input_names, int ninputs,
+                                        @Cast("const char**") @ByPtrPtr byte[] output_names, int noutputs,
+                                        @Cast("const char**") @ByPtrPtr byte[] target_oper_names,
+                                        int ntargets, @Cast("const char**") @ByPtrPtr byte[] handle,
+                                        TF_Status arg8);
+
+// See TF_SessionPRun above.
+public static native void TF_PRun(TF_DeprecatedSession arg0, @Cast("const char*") BytePointer handle,
+                                   @Cast("const char**") PointerPointer input_names, @Cast("TF_Tensor**") PointerPointer inputs,
+                                   int ninputs, @Cast("const char**") PointerPointer output_names,
+                                   @Cast("TF_Tensor**") PointerPointer outputs, int noutputs,
+                                   @Cast("const char**") PointerPointer target_oper_names, int ntargets,
+                                   TF_Status arg10);
+public static native void TF_PRun(TF_DeprecatedSession arg0, @Cast("const char*") BytePointer handle,
+                                   @Cast("const char**") @ByPtrPtr BytePointer input_names, @ByPtrPtr TF_Tensor inputs,
+                                   int ninputs, @Cast("const char**") @ByPtrPtr BytePointer output_names,
+                                   @ByPtrPtr TF_Tensor outputs, int noutputs,
+                                   @Cast("const char**") @ByPtrPtr BytePointer target_oper_names, int ntargets,
+                                   TF_Status arg10);
+public static native void TF_PRun(TF_DeprecatedSession arg0, String handle,
+                                   @Cast("const char**") @ByPtrPtr ByteBuffer input_names, @ByPtrPtr TF_Tensor inputs,
+                                   int ninputs, @Cast("const char**") @ByPtrPtr ByteBuffer output_names,
+                                   @ByPtrPtr TF_Tensor outputs, int noutputs,
+                                   @Cast("const char**") @ByPtrPtr ByteBuffer target_oper_names, int ntargets,
+                                   TF_Status arg10);
+public static native void TF_PRun(TF_DeprecatedSession arg0, @Cast("const char*") BytePointer handle,
+                                   @Cast("const char**") @ByPtrPtr byte[] input_names, @ByPtrPtr TF_Tensor inputs,
+                                   int ninputs, @Cast("const char**") @ByPtrPtr byte[] output_names,
+                                   @ByPtrPtr TF_Tensor outputs, int noutputs,
+                                   @Cast("const char**") @ByPtrPtr byte[] target_oper_names, int ntargets,
+                                   TF_Status arg10);
+public static native void TF_PRun(TF_DeprecatedSession arg0, String handle,
+                                   @Cast("const char**") @ByPtrPtr BytePointer input_names, @ByPtrPtr TF_Tensor inputs,
+                                   int ninputs, @Cast("const char**") @ByPtrPtr BytePointer output_names,
+                                   @ByPtrPtr TF_Tensor outputs, int noutputs,
+                                   @Cast("const char**") @ByPtrPtr BytePointer target_oper_names, int ntargets,
+                                   TF_Status arg10);
+public static native void TF_PRun(TF_DeprecatedSession arg0, @Cast("const char*") BytePointer handle,
+                                   @Cast("const char**") @ByPtrPtr ByteBuffer input_names, @ByPtrPtr TF_Tensor inputs,
+                                   int ninputs, @Cast("const char**") @ByPtrPtr ByteBuffer output_names,
+                                   @ByPtrPtr TF_Tensor outputs, int noutputs,
+                                   @Cast("const char**") @ByPtrPtr ByteBuffer target_oper_names, int ntargets,
+                                   TF_Status arg10);
+public static native void TF_PRun(TF_DeprecatedSession arg0, String handle,
+                                   @Cast("const char**") @ByPtrPtr byte[] input_names, @ByPtrPtr TF_Tensor inputs,
+                                   int ninputs, @Cast("const char**") @ByPtrPtr byte[] output_names,
+                                   @ByPtrPtr TF_Tensor outputs, int noutputs,
+                                   @Cast("const char**") @ByPtrPtr byte[] target_oper_names, int ntargets,
+                                   TF_Status arg10);
+// Targeting ../TF_DeviceList.java
+
+
+
+// Lists all devices in a TF_Session.
+//
+// Caller takes ownership of the returned TF_DeviceList* which must eventually
+// be freed with a call to TF_DeleteDeviceList.
+public static native TF_DeviceList TF_SessionListDevices(TF_Session session,
+                                                           TF_Status status);
+
+// Lists all devices in a TF_Session.
+//
+// Caller takes ownership of the returned TF_DeviceList* which must eventually
+// be freed with a call to TF_DeleteDeviceList.
+public static native TF_DeviceList TF_DeprecatedSessionListDevices(
+    TF_DeprecatedSession session, TF_Status status);
+
+// Deallocates the device list.
+public static native void TF_DeleteDeviceList(TF_DeviceList list);
+
+// Counts the number of elements in the device list.
+public static native int TF_DeviceListCount(@Const TF_DeviceList list);
+
+// Retrieves the full name of the device (e.g. /job:worker/replica:0/...)
+// The return value will be a pointer to a null terminated string. The caller
+// must not modify or delete the string. It will be deallocated upon a call to
+// TF_DeleteDeviceList.
+//
+// If index is out of bounds, an error code will be set in the status object,
+// and a null pointer will be returned.
+public static native @Cast("const char*") BytePointer TF_DeviceListName(@Const TF_DeviceList list,
+                                                    int index,
+                                                    TF_Status status);
+
+// Retrieves the type of the device at the given index.
+//
+// The caller must not modify or delete the string. It will be deallocated upon
+// a call to TF_DeleteDeviceList.
+//
+// If index is out of bounds, an error code will be set in the status object,
+// and a null pointer will be returned.
+public static native @Cast("const char*") BytePointer TF_DeviceListType(@Const TF_DeviceList list,
+                                                    int index,
+                                                    TF_Status status);
+
+// Retrieve the amount of memory associated with a given device.
+//
+// If index is out of bounds, an error code will be set in the status object,
+// and -1 will be returned.
+public static native @Cast("int64_t") long TF_DeviceListMemoryBytes(
+    @Const TF_DeviceList list, int index, TF_Status status);
+
+// Retrieve the incarnation number of a given device.
+//
+// If index is out of bounds, an error code will be set in the status object,
+// and 0 will be returned.
+public static native @Cast("uint64_t") long TF_DeviceListIncarnation(
+    @Const TF_DeviceList list, int index, TF_Status status);
+// Targeting ../TF_Library.java
+
+
+
+// Load the library specified by library_filename and register the ops and
+// kernels present in that library.
+//
+// Pass "library_filename" to a platform-specific mechanism for dynamically
+// loading a library. The rules for determining the exact location of the
+// library are platform-specific and are not documented here.
+//
+// On success, place OK in status and return the newly created library handle.
+// The caller owns the library handle.
+//
+// On failure, place an error status in status and return NULL.
+public static native TF_Library TF_LoadLibrary(@Cast("const char*") BytePointer library_filename,
+                                                 TF_Status status);
+public static native TF_Library TF_LoadLibrary(String library_filename,
+                                                 TF_Status status);
+
+// Get the OpList of OpDefs defined in the library pointed by lib_handle.
+//
+// Returns a TF_Buffer. The memory pointed to by the result is owned by
+// lib_handle. The data in the buffer will be the serialized OpList proto for
+// ops defined in the library.
+public static native @ByVal TF_Buffer TF_GetOpList(TF_Library lib_handle);
+
+// Frees the memory associated with the library handle.
+// Does NOT unload the library.
+public static native void TF_DeleteLibraryHandle(TF_Library lib_handle);
+
+// Get the OpList of all OpDefs defined in this address space.
+// Returns a TF_Buffer, ownership of which is transferred to the caller
+// (and can be freed using TF_DeleteBuffer).
+//
+// The data in the buffer will be the serialized OpList proto for ops registered
+// in this address space.
+public static native TF_Buffer TF_GetAllOpList();
+// Targeting ../TF_ApiDefMap.java
+
+
+
+// Creates a new TF_ApiDefMap instance.
+//
+// Params:
+//  op_list_buffer - TF_Buffer instance containing serialized OpList
+//    protocol buffer. (See
+//    https://www.tensorflow.org/code/tensorflow/core/framework/op_def.proto
+//    for the OpList proto definition).
+//  status - Set to OK on success and an appropriate error on failure.
+public static native TF_ApiDefMap TF_NewApiDefMap(TF_Buffer op_list_buffer,
+                                                    TF_Status status);
+
+// Deallocates a TF_ApiDefMap.
+public static native void TF_DeleteApiDefMap(TF_ApiDefMap apimap);
+
+// Add ApiDefs to the map.
+//
+// `text` corresponds to a text representation of an ApiDefs protocol message.
+// (https://www.tensorflow.org/code/tensorflow/core/framework/api_def.proto).
+//
+// The provided ApiDefs will be merged with existing ones in the map, with
+// precedence given to the newly added version in case of conflicts with
+// previous calls to TF_ApiDefMapPut.
+public static native void TF_ApiDefMapPut(TF_ApiDefMap api_def_map,
+                                           @Cast("const char*") BytePointer text, @Cast("size_t") long text_len,
+                                           TF_Status status);
+public static native void TF_ApiDefMapPut(TF_ApiDefMap api_def_map,
+                                           String text, @Cast("size_t") long text_len,
+                                           TF_Status status);
+
+// Returns a serialized ApiDef protocol buffer for the TensorFlow operation
+// named `name`.
+public static native TF_Buffer TF_ApiDefMapGet(TF_ApiDefMap api_def_map,
+                                                 @Cast("const char*") BytePointer name,
+                                                 @Cast("size_t") long name_len,
+                                                 TF_Status status);
+public static native TF_Buffer TF_ApiDefMapGet(TF_ApiDefMap api_def_map,
+                                                 String name,
+                                                 @Cast("size_t") long name_len,
+                                                 TF_Status status);
+
+// --------------------------------------------------------------------------
+// Kernel definition information.
+
+// Returns a serialized KernelList protocol buffer containing KernelDefs for all
+// registered kernels.
+public static native TF_Buffer TF_GetAllRegisteredKernels(TF_Status status);
+
+// Returns a serialized KernelList protocol buffer containing KernelDefs for all
+// kernels registered for the operation named `name`.
+public static native TF_Buffer TF_GetRegisteredKernelsForOp(
+    @Cast("const char*") BytePointer name, TF_Status status);
+public static native TF_Buffer TF_GetRegisteredKernelsForOp(
+    String name, TF_Status status);
+// Targeting ../TF_Server.java
+
+
+
+// Creates a new in-process TensorFlow server configured using a serialized
+// ServerDef protocol buffer provided via `proto` and `proto_len`.
+//
+// The server will not serve any requests until TF_ServerStart is invoked.
+// The server will stop serving requests once TF_ServerStop or
+// TF_DeleteServer is invoked.
+public static native TF_Server TF_NewServer(@Const Pointer proto,
+                                              @Cast("size_t") long proto_len,
+                                              TF_Status status);
+
+// Starts an in-process TensorFlow server.
+public static native void TF_ServerStart(TF_Server server, TF_Status status);
+
+// Stops an in-process TensorFlow server.
+public static native void TF_ServerStop(TF_Server server, TF_Status status);
+
+// Blocks until the server has been successfully stopped (via TF_ServerStop or
+// TF_ServerClose).
+public static native void TF_ServerJoin(TF_Server server, TF_Status status);
+
+// Returns the target string that can be provided to TF_SetTarget() to connect
+// a TF_Session to `server`.
+//
+// The returned string is valid only until TF_DeleteServer is invoked.
+public static native @Cast("const char*") BytePointer TF_ServerTarget(TF_Server server);
+
+// Destroy an in-process TensorFlow server, frees memory. If server is running
+// it will be stopped and joined.
+public static native void TF_DeleteServer(TF_Server server);
+// Targeting ../Listener_BytePointer.java
+
+
+public static native void TF_RegisterLogListener(
+    Listener_BytePointer listener);
+// Targeting ../Listener_String.java
+
+
+public static native void TF_RegisterLogListener(
+    Listener_String listener);
+
+// #ifdef __cplusplus /* end extern "C" */
+// #endif
+
+// #endif  // TENSORFLOW_C_C_API_H_
+
+
+// Parsed from tensorflow/c/kernels.h
+
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+==============================================================================*/
+
+// #ifndef TENSORFLOW_C_KERNELS_H_
+// #define TENSORFLOW_C_KERNELS_H_
+
+// #include <stdint.h>
+
+// #include "tensorflow/c/tf_datatype.h"
+// #include "tensorflow/c/tf_status.h"
+
+// Macro to control visibility of exported symbols in the shared library (.so,
+// .dylib, .dll).
+// This duplicates the TF_EXPORT macro definition in
+// tensorflow/core/platform/macros.h in order to keep this .h file independent
+// of any other includes.
+// #ifdef SWIG
+// #define TF_CAPI_EXPORT
+// #else
+// #if defined(_WIN32)
+// #ifdef TF_COMPILE_LIBRARY
+// #define TF_CAPI_EXPORT __declspec(dllexport)
+// #else
+// #define TF_CAPI_EXPORT __declspec(dllimport)
+// #endif  // TF_COMPILE_LIBRARY
+// #else
+// #define TF_CAPI_EXPORT __attribute__((visibility("default")))
+// #endif  // _WIN32
+// #endif  // SWIG
+
+// #ifdef __cplusplus
+// #endif
+// Targeting ../TF_KernelBuilder.java
+
+
+// Targeting ../TF_OpKernelConstruction.java
+
+
+// Targeting ../TF_OpKernelContext.java
+
+
+// Targeting ../Create_func_TF_OpKernelConstruction.java
+
+
+// Targeting ../Compute_func_Pointer_TF_OpKernelContext.java
+
+
+// Targeting ../Delete_func_Pointer.java
+
+
+public static native TF_KernelBuilder TF_NewKernelBuilder(
+    @Cast("const char*") BytePointer op_name, @Cast("const char*") BytePointer device_name,
+    Create_func_TF_OpKernelConstruction create_func,
+    Compute_func_Pointer_TF_OpKernelContext compute_func,
+    Delete_func_Pointer delete_func);
+public static native TF_KernelBuilder TF_NewKernelBuilder(
+    String op_name, String device_name,
+    Create_func_TF_OpKernelConstruction create_func,
+    Compute_func_Pointer_TF_OpKernelContext compute_func,
+    Delete_func_Pointer delete_func);
+
+// Specifies that this kernel's attribute only supports the given type.
+public static native void TF_KernelBuilder_TypeConstraint(
+    TF_KernelBuilder kernel_builder, @Cast("const char*") BytePointer attr_name,
+    @Cast("const TF_DataType") int type, TF_Status status);
+public static native void TF_KernelBuilder_TypeConstraint(
+    TF_KernelBuilder kernel_builder, String attr_name,
+    @Cast("const TF_DataType") int type, TF_Status status);
+
+// Specify that this kernel requires/provides an input/output arg
+// in host memory (instead of the default, device memory).
+public static native void TF_KernelBuilder_HostMemory(
+    TF_KernelBuilder kernel_builder, @Cast("const char*") BytePointer arg_name);
+public static native void TF_KernelBuilder_HostMemory(
+    TF_KernelBuilder kernel_builder, String arg_name);
+
+// Register the given kernel builder with the TensorFlow runtime. If
+// registration fails, the given status will be populated.
+//
+// This call takes ownership of the `builder` pointer.
+public static native void TF_RegisterKernelBuilder(@Cast("const char*") BytePointer kernel_name,
+                                                    TF_KernelBuilder builder,
+                                                    TF_Status status);
+public static native void TF_RegisterKernelBuilder(String kernel_name,
+                                                    TF_KernelBuilder builder,
+                                                    TF_Status status);
+
+// Deletes the given TF_KernelBuilder. This should be called only if the kernel
+// builder is not registered with TensorFlow via TF_RegisterKernelBuilder.
+public static native void TF_DeleteKernelBuilder(TF_KernelBuilder builder);
+
+// --------------------------------------------------------------------------
+// OpKernelContext routines
+
+// TF_NumInputs returns the number of inputs available in ctx.
+public static native int TF_NumInputs(TF_OpKernelContext ctx);
+
+// TF_NumOutputs returns the number of outputs to be placed in *ctx by the
+// kernel.
+public static native int TF_NumOutputs(TF_OpKernelContext ctx);
+
+// Retrieves the ith input from ctx. If TF_GetCode(status) is TF_OK, *tensor is
+// populated and its ownership is passed to the caller. In any other case,
+// *tensor is not modified.
+//
+// If i < 0 or i >= TF_NumInputs(ctx), *status is set to TF_OUT_OF_RANGE.
+public static native void TF_GetInput(TF_OpKernelContext ctx, int i,
+                                       @Cast("TF_Tensor**") PointerPointer tensor, TF_Status status);
+public static native void TF_GetInput(TF_OpKernelContext ctx, int i,
+                                       @ByPtrPtr TF_Tensor tensor, TF_Status status);
+
+// Sets the ith output of ctx to tensor. If TF_GetCode(status) is anything but
+// TF_OK, ctx is left unmodified.
+//
+// If i < 0 or i >= TF_NumOutputs(ctx), *status is set to TF_OUT_OF_RANGE.
+public static native void TF_SetOutput(TF_OpKernelContext ctx, int i,
+                                        @Const TF_Tensor tensor,
+                                        TF_Status status);
+
+// Notifies the given OpKernelConstruction that kernel construction has failed.
+public static native void TF_OpKernelConstruction_Failure(
+    TF_OpKernelConstruction ctx, TF_Status status);
+
+// Notifies the given OpKernelContext that the kernel's compute function has
+// failed.
+public static native void TF_OpKernelContext_Failure(TF_OpKernelContext ctx,
+                                                      TF_Status status);
+
+// Returns the expected output data type of the ith output. If i < 0 or
+// i >= TF_NumOutputs(ctx), the program aborts.
+public static native @Cast("TF_DataType") int TF_ExpectedOutputDataType(
+    TF_OpKernelContext ctx, int i);
+
+// Returns the step ID of the given context.
+public static native @Cast("int64_t") long TF_StepId(TF_OpKernelContext ctx);
+
+// Interprets the named kernel construction attribute as a TF_DataType and
+// places it into *val. *status is set to TF_OK.
+//
+// If the attribute could not be found or could not be interpreted as
+// TF_DataType, *status is populated with an error.
+public static native void TF_OpKernelConstruction_GetAttrType(
+    TF_OpKernelConstruction ctx, @Cast("const char*") BytePointer attr_name, @Cast("TF_DataType*") IntPointer val,
+    TF_Status status);
+public static native void TF_OpKernelConstruction_GetAttrType(
+    TF_OpKernelConstruction ctx, String attr_name, @Cast("TF_DataType*") IntBuffer val,
+    TF_Status status);
+public static native void TF_OpKernelConstruction_GetAttrType(
+    TF_OpKernelConstruction ctx, @Cast("const char*") BytePointer attr_name, @Cast("TF_DataType*") int[] val,
+    TF_Status status);
+public static native void TF_OpKernelConstruction_GetAttrType(
+    TF_OpKernelConstruction ctx, String attr_name, @Cast("TF_DataType*") IntPointer val,
+    TF_Status status);
+public static native void TF_OpKernelConstruction_GetAttrType(
+    TF_OpKernelConstruction ctx, @Cast("const char*") BytePointer attr_name, @Cast("TF_DataType*") IntBuffer val,
+    TF_Status status);
+public static native void TF_OpKernelConstruction_GetAttrType(
+    TF_OpKernelConstruction ctx, String attr_name, @Cast("TF_DataType*") int[] val,
+    TF_Status status);
+
+// Interprets the named kernel construction attribute as int32_t and
+// places it into *val. *status is set to TF_OK.
+//
+// If the attribute could not be found or could not be interpreted as
+// int32, *status is populated with an error.
+public static native void TF_OpKernelConstruction_GetAttrInt32(
+    TF_OpKernelConstruction ctx, @Cast("const char*") BytePointer attr_name, IntPointer val,
+    TF_Status status);
+public static native void TF_OpKernelConstruction_GetAttrInt32(
+    TF_OpKernelConstruction ctx, String attr_name, IntBuffer val,
+    TF_Status status);
+public static native void TF_OpKernelConstruction_GetAttrInt32(
+    TF_OpKernelConstruction ctx, @Cast("const char*") BytePointer attr_name, int[] val,
+    TF_Status status);
+public static native void TF_OpKernelConstruction_GetAttrInt32(
+    TF_OpKernelConstruction ctx, String attr_name, IntPointer val,
+    TF_Status status);
+public static native void TF_OpKernelConstruction_GetAttrInt32(
+    TF_OpKernelConstruction ctx, @Cast("const char*") BytePointer attr_name, IntBuffer val,
+    TF_Status status);
+public static native void TF_OpKernelConstruction_GetAttrInt32(
+    TF_OpKernelConstruction ctx, String attr_name, int[] val,
+    TF_Status status);
+
+// Allocates Tensor for output at given index. Caller takes ownership of
+// returned TF_Tensor and should deallocate it using TF_DeleteTensor(tensor).
+//
+// This function should be used to allocate outputs inside kernel
+// compute function.
+public static native TF_Tensor TF_AllocateOutput(TF_OpKernelContext context,
+                                            int index, @Cast("TF_DataType") int dtype,
+                                            @Cast("int64_t*") LongPointer dims, int num_dims,
+                                            @Cast("size_t") long len, TF_Status status);
+public static native TF_Tensor TF_AllocateOutput(TF_OpKernelContext context,
+                                            int index, @Cast("TF_DataType") int dtype,
+                                            @Cast("int64_t*") LongBuffer dims, int num_dims,
+                                            @Cast("size_t") long len, TF_Status status);
+public static native TF_Tensor TF_AllocateOutput(TF_OpKernelContext context,
+                                            int index, @Cast("TF_DataType") int dtype,
+                                            @Cast("int64_t*") long[] dims, int num_dims,
+                                            @Cast("size_t") long len, TF_Status status);
+
+// #ifdef __cplusplus /* end extern "C" */
+// #endif
+
+// #endif  // TENSORFLOW_C_KERNELS_H_
+
+
+// Parsed from tensorflow/c/ops.h
+
+/* Copyright 2019 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+==============================================================================*/
+
+// Routines for registering new ops and for implementing op shape inference
+// functions.
+//
+// This API is alpha software and is subject to change.
+//
+// REGISTRATION
+// ------------
+//
+// In order to register a new op, create a new TF_OpDefinitionBuilder:
+//
+// TF_OpDefinitionBuilder* builder = TF_NewOpDefinitionBuilder("OpName");
+//
+// Inputs, outputs and attributes can be added to the builder with the
+// corresponding functions, e.g.
+//
+// TF_OpDefinitionBuilderAddInput(builder, "input1: int32");
+// TF_OpDefinitionBuilderAddOutput(builder, "output1: int64");
+// TF_OpDefinitionBuilderAddAttr(builder, "attr: int32");
+//
+// The builder may then be registered with TensorFlow using the
+// TF_RegisterOpDefinition function. E.g.
+//
+// TF_Status* status = TF_NewStatus();
+// TF_RegisterOpDefinition(builder, &status);
+// if (TF_GetCode(status) != TF_OK) {
+//   // handle error
+// }
+//
+// SHAPE INFERENCE
+// ---------------
+//
+// You can provide a shape inference function that TensorFlow will call when it
+// wants to understand the shape of outputs that the op will produce. Use the
+// TF_OpDefinitionBuilderSetShapeInferenceFunction function to register a shape
+// inference function pointer with TensorFlow. The following is an example of a
+// very simple shape inference function:
+//
+// void identity_shape_fn(TF_ShapeInferenceContext* ctx, TF_Status* status) {
+//   TF_ShapeHandle* input = TF_NewShapeHandle();
+//   TF_ShapeInferenceContextGetInput(ctx, 0, input, status);
+//   if (TF_GetCode(status) == TF_OK) {
+//     TF_ShapeInferenceContextSetOutput(ctx, 0, input, status);
+//   }
+//   TF_DeleteShapeHandle(input);
+// }
+//
+// The following code registers the inference function with TensorFlow:
+//
+// TF_OpDefinitionBuilderSetShapeInferenceFunction(builder, &identity_shape_fn);
+//
+// For more details about shape inference, see the documentation for
+// TF_OpDefinitionBuilderSetShapeInferenceFunction.
+
+// #ifndef TENSORFLOW_C_OPS_H_
+// #define TENSORFLOW_C_OPS_H_
+
+// #include <stdbool.h>
+// #include <stdint.h>
+// #include <stdlib.h>
+
+// #include "tensorflow/c/tf_datatype.h"
+// #include "tensorflow/c/tf_status.h"
+
+// #ifdef SWIG
+// #define TF_CAPI_EXPORT
+// #else
+// #if defined(_WIN32)
+// #ifdef TF_COMPILE_LIBRARY
+// #define TF_CAPI_EXPORT __declspec(dllexport)
+// #else
+// #define TF_CAPI_EXPORT __declspec(dllimport)
+// #endif  // TF_COMPILE_LIBRARY
+// #else
+// #define TF_CAPI_EXPORT __attribute__((visibility("default")))
+// #endif  // _WIN32
+// #endif  // SWIG
+
+// #ifdef __cplusplus
+// Targeting ../TF_DimensionHandle.java
+
+
+// Targeting ../TF_OpDefinitionBuilder.java
+
+
+// Targeting ../TF_ShapeHandle.java
+
+
+// Targeting ../TF_ShapeInferenceContext.java
+
+
+
+// Returns a newly allocated op definition builder for the given op name. The
+// returned builder may be customized with the `TF_OpDefinitionBuilder...`
+// functions and then registered with TensorFlow with TF_RegisterOpDefinition.
+//
+// The returned pointer is either freed by a call to TF_RegisterOpDefinition, or
+// can be manually deleted by TF_DeleteOpDefinitionBuilder if it is never
+// registered.
+public static native TF_OpDefinitionBuilder TF_NewOpDefinitionBuilder(
+    @Cast("const char*") BytePointer op_name);
+public static native TF_OpDefinitionBuilder TF_NewOpDefinitionBuilder(
+    String op_name);
+
+// Registers the given op builder with TensorFlow. Indicates success or
+// otherwise in the given status.
+//
+// `builder` is freed whether the op was successfully registered or not. You
+// must call either this function or TF_DeleteOpDefinitionBuilder to free the
+// builder, but never both.
+public static native void TF_RegisterOpDefinition(
+    TF_OpDefinitionBuilder builder, TF_Status status);
+
+// Frees the given op definition builder. You must call either this function or
+// TF_RegisterOpDefinition to free the builder, but never both.
+public static native void TF_DeleteOpDefinitionBuilder(
+    TF_OpDefinitionBuilder builder);
+
+//----------------------------------------------------
+// Attribute functions.
+
+// Adds an attr to the given TF_OpDefinitionBuilder. The spec has
+// format "<name>:<type>" or "<name>:<type>=<default>"
+// where <name> matches regexp [a-zA-Z][a-zA-Z0-9_]*.
+// By convention, names containing only capital letters are reserved for
+// attributes whose values can be inferred by the operator implementation if not
+// supplied by the user. If the attribute name contains characters other than
+// capital letters, the operator expects the user to provide the attribute value
+// at operation runtime.
+//
+// <type> can be:
+//   "string", "int", "float", "bool", "type", "shape", or "tensor"
+//   "numbertype", "realnumbertype", "quantizedtype"
+//       (meaning "type" with a restriction on valid values)
+//   "{int32,int64}" or {realnumbertype,quantizedtype,string}"
+//       (meaning "type" with a restriction containing unions of value types)
+//   "{\"foo\", \"bar\n baz\"}", or "{'foo', 'bar\n baz'}"
+//       (meaning "string" with a restriction on valid values)
+//   "list(string)", ..., "list(tensor)", "list(numbertype)", ...
+//       (meaning lists of the above types)
+//   "int >= 2" (meaning "int" with a restriction on valid values)
+//   "list(string) >= 2", "list(int) >= 2"
+//       (meaning "list(string)" / "list(int)" with length at least 2)
+// <default>, if included, should use the Proto text format
+// of <type>.  For lists use [a, b, c] format.
+//
+// Note that any attr specifying the length of an input or output will
+// get a default minimum of 1 unless the >= # syntax is used.
+public static native void TF_OpDefinitionBuilderAddAttr(
+    TF_OpDefinitionBuilder builder, @Cast("const char*") BytePointer attr_spec);
+public static native void TF_OpDefinitionBuilderAddAttr(
+    TF_OpDefinitionBuilder builder, String attr_spec);
+
+// Adds an input to this TF_OpDefinitionBuilder.
+// The spec has form "<name>:<type-expr>" or "<name>:Ref(<type-expr>)"
+// where <name> matches regexp [a-z][a-z0-9_]* and <type-expr> can be:
+// * For a single tensor: <type>
+// * For a sequence of tensors with the same type: <number>*<type>
+// * For a sequence of tensors with different types: <type-list>
+// Where:
+//   <type> is either one of "float", "int32", "string", ...
+//          or the name of an attr (see TF_OpDefinitionBuilderAddAttr)
+//          with type "type".
+//   <number> is the name of an attr with type "int".
+//   <type-list> is the name of an attr with type "list(type)".
+public static native void TF_OpDefinitionBuilderAddInput(
+    TF_OpDefinitionBuilder builder, @Cast("const char*") BytePointer input_spec);
+public static native void TF_OpDefinitionBuilderAddInput(
+    TF_OpDefinitionBuilder builder, String input_spec);
+
+// Adds an output to this TF_OpDefinitionBuilder.
+// The spec has form "<name>:<type-expr>" or "<name>:Ref(<type-expr>)"
+// where <name> matches regexp [a-z][a-z0-9_]* and <type-expr> can be:
+// * For a single tensor: <type>
+// * For a sequence of tensors with the same type: <number>*<type>
+// * For a sequence of tensors with different types: <type-list>
+// Where:
+//   <type> is either one of "float", "int32", "string", ...
+//          or the name of an attr (see TF_OpDefinitionBuilderAddAttr)
+//          with type "type".
+//   <number> is the name of an attr with type "int".
+//   <type-list> is the name of an attr with type "list(type)".
+public static native void TF_OpDefinitionBuilderAddOutput(
+    TF_OpDefinitionBuilder builder, @Cast("const char*") BytePointer output_spec);
+public static native void TF_OpDefinitionBuilderAddOutput(
+    TF_OpDefinitionBuilder builder, String output_spec);
+
+// Sets the commutative property for the op built by the given builder.
+public static native void TF_OpDefinitionBuilderSetIsCommutative(
+    TF_OpDefinitionBuilder builder, @Cast("bool") boolean is_commutative);
+
+// Sets the is_aggregate property of the builder to the given value.
+//
+// If is_aggregate is true, then the operation produced by this builder accepts
+// N >= 2 inputs and produces 1 output all of the same type. Should be
+// associative and commutative, and produce output with the same shape as the
+// input. The optimizer may replace an aggregate op taking input from multiple
+// devices with a tree of aggregate ops that aggregate locally within each
+// device (and possibly within groups of nearby devices) before communicating.
+public static native void TF_OpDefinitionBuilderSetIsAggregate(
+    TF_OpDefinitionBuilder builder, @Cast("bool") boolean is_aggregate);
+
+// Sets the is_stateful property of the builder to the given value.
+//
+// The op built by this builder is stateful if its behavior depends on some
+// state beyond its input tensors (e.g. variable reading op) or if it has a
+// side-effect (e.g. printing or asserting ops). Equivalently, stateless ops
+// must always produce the same output for the same input and have no
+// side-effects.
+//
+// By default Ops may be moved between devices. Stateful ops should either not
+// be moved, or should only be moved if that state can also be moved (e.g. via
+// some sort of save / restore). Stateful ops are guaranteed to never be
+// optimized away by Common Subexpression Elimination (CSE).
+public static native void TF_OpDefinitionBuilderSetIsStateful(
+    TF_OpDefinitionBuilder builder, @Cast("bool") boolean is_stateful);
+
+// Sets the allows_uninitialized_input property of the operation built by this
+// builder.
+//
+// By default, all inputs to an Op must be initialized Tensors. Ops that may
+// initialize tensors for the first time should set this field to true, to allow
+// the Op to take an uninitialized Tensor as input.
+public static native void TF_OpDefinitionBuilderSetAllowsUninitializedInput(
+    TF_OpDefinitionBuilder builder, @Cast("bool") boolean allows_uninitialized_input);
+
+// Adds a deprecation warning for the given op. This indicates to the user that
+// `version` is the first TensorFlow GraphDef version for which the operation is
+// deprecated. `explanation` should contain the reason for the deprecation and
+// what to use instead.
+//
+// This function is only an indicator that the operation may disappear in a
+// version of TensorFlow after `version`. It does not affect op registration.
+public static native void TF_OpDefinitionBuilderDeprecated(
+    TF_OpDefinitionBuilder builder, int version, @Cast("const char*") BytePointer explanation);
+public static native void TF_OpDefinitionBuilderDeprecated(
+    TF_OpDefinitionBuilder builder, int version, String explanation);
+// Targeting ../Shape_inference_func_TF_ShapeInferenceContext_TF_Status.java
+
+
+public static native void TF_OpDefinitionBuilderSetShapeInferenceFunction(
+    TF_OpDefinitionBuilder builder,
+    Shape_inference_func_TF_ShapeInferenceContext_TF_Status shape_inference_func);
+
+//----------------------------------------------------
+// Functions for TF_ShapeInferenceContext.
+//
+// Functions for implementing shape inference functions. TensorFlow uses these
+// functions to determine the shape of tensors produced by an operation without
+// having to actually run the operation. If an operation chooses to provide a
+// shape inference function, it will be invoked by TensorFlow as needed.
+//
+// When invoked by TensorFlow, the shape inference function is provided with a
+// TF_ShapeInferenceContext pointer. The function's implementation will use the
+// accessor and mutator functions with names beginning with
+// TF_ShapeInferenceContext to examine the input state and determine the output
+// shape.
+
+// Returns the number of inputs in the given shape inference context.
+public static native @Cast("int64_t") long TF_ShapeInferenceContextNumInputs(
+    TF_ShapeInferenceContext ctx);
+
+// Returns a newly allocated shape handle. The shapes represented by these
+// handles may be queried or mutated with the corresponding
+// TF_ShapeInferenceContext...  functions.
+public static native TF_ShapeHandle TF_NewShapeHandle();
+
+// Places the ith input of the given shape inference context into the given
+// shape handle, or returns a status other than TF_OK indicating why the input
+// could not be retrieved
+// (for example, if i < 0 || i >= TF_ShapeInferenceContextNumInputs(ctx)).
+public static native void TF_ShapeInferenceContextGetInput(
+    TF_ShapeInferenceContext ctx, int i, TF_ShapeHandle handle,
+    TF_Status status);
+
+// Places the given shape handle into the `i`th output position of the given
+// context. Internally, the shape handle is copied; the caller may subsequently
+// delete `handle`.
+public static native void TF_ShapeInferenceContextSetOutput(TF_ShapeInferenceContext ctx,
+                                              int i, TF_ShapeHandle handle,
+                                              TF_Status status);
+
+// Returns a newly-allocate shape handle representing a vector of the given
+// size. The returned handle should be freed with TF_DeleteShapeHandle.
+public static native TF_ShapeHandle TF_ShapeInferenceContextVectorFromSize(
+    TF_ShapeInferenceContext ctx, @Cast("size_t") long size);
+
+// Returns a newly allocated dimension handle. It must be freed with
+// TF_DeleteDimensionHandle.
+public static native TF_DimensionHandle TF_NewDimensionHandle();
+
+// Interprets the named shape inference context attribute as a TF_DataType and
+// places it into *val. *status is set to TF_OK.
+//
+// If the attribute could not be found or could not be interpreted as
+// TF_DataType, *status is populated with an error.
+public static native void TF_ShapeInferenceContext_GetAttrType(
+    TF_ShapeInferenceContext ctx, @Cast("const char*") BytePointer attr_name, @Cast("TF_DataType*") IntPointer val,
+    TF_Status status);
+public static native void TF_ShapeInferenceContext_GetAttrType(
+    TF_ShapeInferenceContext ctx, String attr_name, @Cast("TF_DataType*") IntBuffer val,
+    TF_Status status);
+public static native void TF_ShapeInferenceContext_GetAttrType(
+    TF_ShapeInferenceContext ctx, @Cast("const char*") BytePointer attr_name, @Cast("TF_DataType*") int[] val,
+    TF_Status status);
+public static native void TF_ShapeInferenceContext_GetAttrType(
+    TF_ShapeInferenceContext ctx, String attr_name, @Cast("TF_DataType*") IntPointer val,
+    TF_Status status);
+public static native void TF_ShapeInferenceContext_GetAttrType(
+    TF_ShapeInferenceContext ctx, @Cast("const char*") BytePointer attr_name, @Cast("TF_DataType*") IntBuffer val,
+    TF_Status status);
+public static native void TF_ShapeInferenceContext_GetAttrType(
+    TF_ShapeInferenceContext ctx, String attr_name, @Cast("TF_DataType*") int[] val,
+    TF_Status status);
+
+// Returns the rank of the shape represented by the given handle.
+public static native @Cast("int64_t") long TF_ShapeInferenceContextRank(
+    TF_ShapeInferenceContext ctx, TF_ShapeHandle handle);
+
+// Returns 1 if `handle` has a known rank, 0 otherwise.
+public static native int TF_ShapeInferenceContextRankKnown(
+    TF_ShapeInferenceContext ctx, TF_ShapeHandle handle);
+
+// If <handle> has rank <rank>, or its rank is unknown, return OK and return the
+// shape with asserted rank in <*result>. Otherwise an error is placed into
+// `status`.
+public static native void TF_ShapeInferenceContextWithRank(
+    TF_ShapeInferenceContext ctx, TF_ShapeHandle handle, @Cast("int64_t") long rank,
+    TF_ShapeHandle result, TF_Status status);
+
+// If <handle> has rank at least <rank>, or its rank is unknown, return OK and
+// return the shape with asserted rank in <*result>. Otherwise an error is
+// placed into `status`.
+public static native void TF_ShapeInferenceContextWithRankAtLeast(
+    TF_ShapeInferenceContext ctx, TF_ShapeHandle handle, @Cast("int64_t") long rank,
+    TF_ShapeHandle result, TF_Status status);
+
+// If <handle> has rank at most <rank>, or its rank is unknown, return OK and
+// return the shape with asserted rank in <*result>. Otherwise an error is
+// placed into `status`.
+public static native void TF_ShapeInferenceContextWithRankAtMost(
+    TF_ShapeInferenceContext ctx, TF_ShapeHandle handle, @Cast("int64_t") long rank,
+    TF_ShapeHandle result, TF_Status status);
+
+// Places a handle to the ith dimension of the given shape into *result.
+public static native void TF_ShapeInferenceContextDim(
+    TF_ShapeInferenceContext ctx, TF_ShapeHandle shape_handle, @Cast("int64_t") long i,
+    TF_DimensionHandle result);
+
+// Returns in <*result> a sub-shape of <shape_handle>, with dimensions
+// [start:end]. <start> and <end> can be negative, to index from the end of the
+// shape. <start> and <end> are set to the rank of <shape_handle> if > rank of
+// <shape_handle>.
+public static native void TF_ShapeInferenceContextSubshape(
+    TF_ShapeInferenceContext ctx, TF_ShapeHandle shape_handle, @Cast("int64_t") long start,
+    @Cast("int64_t") long end, TF_ShapeHandle result, TF_Status status);
+
+// Places an unknown shape in all outputs for the given inference context. Used
+// for shape inference functions with ops whose output shapes are unknown.
+public static native void TF_ShapeInferenceContextSetUnknownShape(
+    TF_ShapeInferenceContext ctx, TF_Status status);
+
+// Returns whether the given handle represents a known dimension.
+public static native int TF_DimensionHandleValueKnown(
+    TF_DimensionHandle dim_handle);
+
+// Returns the value of the given dimension.
+public static native @Cast("int64_t") long TF_DimensionHandleValue(
+    TF_DimensionHandle dim_handle);
+
+// Returns in <*result> the result of appending the dimensions of <second> to
+// those of <first>.
+public static native void TF_ShapeInferenceContextConcatenateShapes(
+    TF_ShapeInferenceContext ctx, TF_ShapeHandle first,
+    TF_ShapeHandle second, TF_ShapeHandle result, TF_Status status);
+
+// Frees the given shape handle.
+public static native void TF_DeleteShapeHandle(TF_ShapeHandle handle);
+
+// Frees the given dimension handle.
+public static native void TF_DeleteDimensionHandle(TF_DimensionHandle handle);
+
+// #ifdef __cplusplus /* end extern "C" */
+// #endif
+
+// #endif  // TENSORFLOW_C_OPS_H_
+
+
+// Parsed from tensorflow/c/eager/c_api.h
+
+/* Copyright 2017 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+==============================================================================*/
+
+// #ifndef TENSORFLOW_C_EAGER_C_API_H_
+// #define TENSORFLOW_C_EAGER_C_API_H_
+
+// C API extensions to experiment with eager execution of kernels.
+// WARNING: Unlike tensorflow/c/c_api.h, the API here is not guaranteed to be
+// stable and can change without notice.
+
+// #include "tensorflow/c/c_api.h"
+
+// Macro to control visibility of exported symbols in the shared library (.so,
+// .dylib, .dll).
+// This duplicates the TF_EXPORT macro definition in
+// tensorflow/core/platform/macros.h in order to keep this .h file independent
+// of any other includes.$a
+// #ifdef SWIG
+// #define TF_CAPI_EXPORT
+// #else
+// #if defined(_WIN32)
+// #ifdef TF_COMPILE_LIBRARY
+// #define TF_CAPI_EXPORT __declspec(dllexport)
+// #else
+// #define TF_CAPI_EXPORT __declspec(dllimport)
+// #endif  // TF_COMPILE_LIBRARY
+// #else
+// #define TF_CAPI_EXPORT __attribute__((visibility("default")))
+// #endif  // _WIN32
+// #endif  // SWIG
+
+// #ifdef __cplusplus
+// Targeting ../TFE_ContextOptions.java
+
+
+
+// Return a new options object.
+public static native TFE_ContextOptions TFE_NewContextOptions();
+
+// Set the config in TF_ContextOptions.options.
+// config should be a serialized tensorflow.ConfigProto proto.
+// If config was not parsed successfully as a ConfigProto, record the
+// error information in *status.
+public static native void TFE_ContextOptionsSetConfig(
+    TFE_ContextOptions options, @Const Pointer proto, @Cast("size_t") long proto_len,
+    TF_Status status);
+
+// Controls how to act when we try to run an operation on a given device but
+// some input tensors are not on that device.
+// LINT.IfChange
+// Note: Keep in sync with internal copy of enum in eager/context.h.
+/** enum TFE_ContextDevicePlacementPolicy */
+public static final int
+  // Running operations with input tensors on the wrong device will fail.
+  TFE_DEVICE_PLACEMENT_EXPLICIT = 0,
+  // Copy the tensor to the right device but log a warning.
+  TFE_DEVICE_PLACEMENT_WARN = 1,
+  // Silently copy the tensor, which has a performance cost since the operation
+  // will be blocked till the copy completes. This is the default placement
+  // policy.
+  TFE_DEVICE_PLACEMENT_SILENT = 2,
+  // Placement policy which silently copies int32 tensors but not other dtypes.
+  TFE_DEVICE_PLACEMENT_SILENT_FOR_INT32 = 3;
+// LINT.ThenChange(//tensorflow/core/common_runtime/eager/context.h)
+
+// Sets the default execution mode (sync/async). Note that this can be
+// overridden per thread using TFE_ContextSetExecutorForThread.
+public static native void TFE_ContextOptionsSetAsync(TFE_ContextOptions arg0,
+                                                      @Cast("unsigned char") byte enable);
+
+public static native void TFE_ContextOptionsSetDevicePlacementPolicy(
+    TFE_ContextOptions arg0, @Cast("TFE_ContextDevicePlacementPolicy") int arg1);
+
+// Destroy an options object.
+public static native void TFE_DeleteContextOptions(TFE_ContextOptions arg0);
+// Targeting ../TFE_Context.java
+
+
+
+public static native TFE_Context TFE_NewContext(
+    @Const TFE_ContextOptions opts, TF_Status status);
+public static native void TFE_DeleteContext(TFE_Context ctx);
+public static native TF_DeviceList TFE_ContextListDevices(TFE_Context ctx,
+                                                            TF_Status status);
+
+// Clears the internal caches in the TFE context. Useful when reseeding random
+// ops.
+public static native void TFE_ContextClearCaches(TFE_Context ctx);
+
+// Sets a thread-local device placement policy. After this call, other calls to
+// TFE_Execute in the same thread will use the device policy specified here
+// instead of the device policy used to construct the context. This has no
+// effect on the device policy used by other program threads.
+public static native void TFE_ContextSetThreadLocalDevicePlacementPolicy(
+    TFE_Context ctx, @Cast("TFE_ContextDevicePlacementPolicy") int policy);
+
+// Returns the device placement policy to be used by this context in the current
+// thread.
+public static native @Cast("TFE_ContextDevicePlacementPolicy") int TFE_ContextGetDevicePlacementPolicy(TFE_Context ctx);
+
+// A tensorflow.ServerDef specifies remote workers (in addition to the current
+// workers name). Operations created on this context can then be executed on
+// any of these remote workers by setting an appropriate device.
+//
+// If the following is set, all servers identified by the
+// ServerDef must be up when the context is created.
+public static native void TFE_ContextSetServerDef(TFE_Context ctx,
+                                                   int keep_alive_secs,
+                                                   @Const Pointer proto,
+                                                   @Cast("size_t") long proto_len,
+                                                   TF_Status status);
+// Targeting ../TFE_TensorHandle.java
+
+
+
+public static native TFE_TensorHandle TFE_NewTensorHandle(@Const TF_Tensor t,
+                                                            TF_Status status);
+// Indicates that the caller will not be using `h` any more.
+public static native void TFE_DeleteTensorHandle(TFE_TensorHandle h);
+public static native @Cast("TF_DataType") int TFE_TensorHandleDataType(TFE_TensorHandle h);
+// This function will block till the operation that produces `h` has completed.
+public static native int TFE_TensorHandleNumDims(TFE_TensorHandle h,
+                                                  TF_Status status);
+public static native @Cast("int64_t") long TFE_TensorHandleNumElements(TFE_TensorHandle h,
+                                                          TF_Status status);
+// This function will block till the operation that produces `h` has completed.
+public static native @Cast("int64_t") long TFE_TensorHandleDim(TFE_TensorHandle h,
+                                                  int dim_index,
+                                                  TF_Status status);
+
+// Returns the device of the operation that produced `h`. If `h` was produced by
+// a copy, returns the destination device of the copy. Note that the returned
+// device name is not always the device holding the tensor handle's memory. If
+// you want the latter, use TFE_TensorHandleBackingDeviceName. This function
+// will block till the operation that produces `h` has completed.
+public static native @Cast("const char*") BytePointer TFE_TensorHandleDeviceName(
+    TFE_TensorHandle h, TF_Status status);
+
+// Returns the name of the device in whose memory `h` resides.
+//
+// This function will block till the operation that produces `h` has completed.
+public static native @Cast("const char*") BytePointer TFE_TensorHandleBackingDeviceName(
+    TFE_TensorHandle h, TF_Status status);
+
+// Return a pointer to a new TFE_TensorHandle that shares the underlying tensor
+// with `h`. On success, `status` is set to OK. On failure, `status` reflects
+// the error and a nullptr is returned.
+public static native TFE_TensorHandle TFE_TensorHandleCopySharingTensor(
+    TFE_TensorHandle h, TF_Status status);
+
+// This function will block till the operation that produces `h` has
+// completed. The memory returned might alias the internal memory used by
+// TensorFlow. Hence, callers should not mutate this memory (for example by
+// modifying the memory region pointed to by TF_TensorData() on the returned
+// TF_Tensor).
+public static native TF_Tensor TFE_TensorHandleResolve(TFE_TensorHandle h,
+                                                         TF_Status status);
+
+// Create a new TFE_TensorHandle with the same contents as 'h' but placed
+// in the memory of the device name 'device_name'.
+// If source and destination are the same device, then this creates a new handle
+// that shares the underlying buffer. Otherwise, it currently requires at least
+// one of the source or destination devices to be CPU (i.e., for the source or
+// destination tensor to be placed in host memory).
+// If async execution is enabled, the copy may be enqueued and the call will
+// return "non-ready" handle. Else, this function returns after the copy has
+// been done.
+public static native TFE_TensorHandle TFE_TensorHandleCopyToDevice(
+    TFE_TensorHandle h, TFE_Context ctx, @Cast("const char*") BytePointer device_name,
+    TF_Status status);
+public static native TFE_TensorHandle TFE_TensorHandleCopyToDevice(
+    TFE_TensorHandle h, TFE_Context ctx, String device_name,
+    TF_Status status);
+// Targeting ../TFE_TensorDebugInfo.java
+
+
+
+// Retrieves TFE_TensorDebugInfo for `handle`.
+// If TFE_TensorHandleTensorDebugInfo succeeds, `status` is set to OK and caller
+// is responsible for deleting returned TFE_TensorDebugInfo.
+// If TFE_TensorHandleTensorDebugInfo fails, `status` is set to appropriate
+// error and nullptr is returned. This function can block till the operation
+// that produces `handle` has completed.
+public static native TFE_TensorDebugInfo TFE_TensorHandleTensorDebugInfo(
+    TFE_TensorHandle h, TF_Status status);
+
+// Deletes `debug_info`.
+public static native void TFE_DeleteTensorDebugInfo(
+    TFE_TensorDebugInfo debug_info);
+
+// Returns the number of dimensions used to represent the tensor on its device.
+// The number of dimensions used to represent the tensor on device can be
+// different from the number returned by TFE_TensorHandleNumDims.
+// The return value was current at the time of TFE_TensorDebugInfo creation.
+public static native int TFE_TensorDebugInfoOnDeviceNumDims(
+    TFE_TensorDebugInfo debug_info);
+
+// Returns the number of elements in dimension `dim_index`.
+// Tensor representation on device can be transposed from its representation
+// on host. The data contained in dimension `dim_index` on device
+// can correspond to the data contained in another dimension in on-host
+// representation. The dimensions are indexed using the standard TensorFlow
+// major-to-minor order (slowest varying dimension first),
+// not the XLA's minor-to-major order.
+// On-device dimensions can be padded. TFE_TensorDebugInfoOnDeviceDim returns
+// the number of elements in a dimension after padding.
+// The return value was current at the time of TFE_TensorDebugInfo creation.
+public static native @Cast("int64_t") long TFE_TensorDebugInfoOnDeviceDim(
+    TFE_TensorDebugInfo debug_info, int dim_index);
+// Targeting ../TFE_Op.java
+
+
+
+public static native TFE_Op TFE_NewOp(TFE_Context ctx,
+                                        @Cast("const char*") BytePointer op_or_function_name,
+                                        TF_Status status);
+public static native TFE_Op TFE_NewOp(TFE_Context ctx,
+                                        String op_or_function_name,
+                                        TF_Status status);
+
+public static native void TFE_DeleteOp(TFE_Op op);
+
+public static native void TFE_OpSetDevice(TFE_Op op, @Cast("const char*") BytePointer device_name,
+                                           TF_Status status);
+public static native void TFE_OpSetDevice(TFE_Op op, String device_name,
+                                           TF_Status status);
+// The returned string remains valid throughout the lifetime of 'op'.
+public static native @Cast("const char*") BytePointer TFE_OpGetDevice(TFE_Op op,
+                                                  TF_Status status);
+
+// When 'enable' is set to 1, and if TensorFlow library is built with XLA
+// support, a subsequent TFE_Execute() call on `op` will run the op via XLA.
+//
+// If the library is not built with XLA support, this call would be a no-op.
+public static native void TFE_OpSetXLACompilation(TFE_Op op,
+                                                   @Cast("unsigned char") byte enable);
+
+public static native void TFE_OpAddInput(TFE_Op op, TFE_TensorHandle input,
+                                          TF_Status status);
+
+public static native void TFE_OpAddInputList(TFE_Op op,
+                                              @Cast("TFE_TensorHandle**") PointerPointer inputs,
+                                              int num_inputs,
+                                              TF_Status status);
+public static native void TFE_OpAddInputList(TFE_Op op,
+                                              @ByPtrPtr TFE_TensorHandle inputs,
+                                              int num_inputs,
+                                              TF_Status status);
+
+public static native @Cast("TF_AttrType") int TFE_OpGetAttrType(TFE_Op op,
+                                                    @Cast("const char*") BytePointer attr_name,
+                                                    @Cast("unsigned char*") BytePointer is_list,
+                                                    TF_Status status);
+public static native @Cast("TF_AttrType") int TFE_OpGetAttrType(TFE_Op op,
+                                                    String attr_name,
+                                                    @Cast("unsigned char*") ByteBuffer is_list,
+                                                    TF_Status status);
+public static native @Cast("TF_AttrType") int TFE_OpGetAttrType(TFE_Op op,
+                                                    @Cast("const char*") BytePointer attr_name,
+                                                    @Cast("unsigned char*") byte[] is_list,
+                                                    TF_Status status);
+public static native @Cast("TF_AttrType") int TFE_OpGetAttrType(TFE_Op op,
+                                                    String attr_name,
+                                                    @Cast("unsigned char*") BytePointer is_list,
+                                                    TF_Status status);
+public static native @Cast("TF_AttrType") int TFE_OpGetAttrType(TFE_Op op,
+                                                    @Cast("const char*") BytePointer attr_name,
+                                                    @Cast("unsigned char*") ByteBuffer is_list,
+                                                    TF_Status status);
+public static native @Cast("TF_AttrType") int TFE_OpGetAttrType(TFE_Op op,
+                                                    String attr_name,
+                                                    @Cast("unsigned char*") byte[] is_list,
+                                                    TF_Status status);
+// Get an attribute type given an op name; a fusion of TFE_NewOp and
+// TFE_OpGetAttrType for use from Python without the overhead of the individual
+// calls and memory management of TFE_Op.
+public static native @Cast("TF_AttrType") int TFE_OpNameGetAttrType(
+    TFE_Context ctx, @Cast("const char*") BytePointer op_or_function_name, @Cast("const char*") BytePointer attr_name,
+    @Cast("unsigned char*") BytePointer is_list, TF_Status status);
+public static native @Cast("TF_AttrType") int TFE_OpNameGetAttrType(
+    TFE_Context ctx, String op_or_function_name, String attr_name,
+    @Cast("unsigned char*") ByteBuffer is_list, TF_Status status);
+public static native @Cast("TF_AttrType") int TFE_OpNameGetAttrType(
+    TFE_Context ctx, @Cast("const char*") BytePointer op_or_function_name, @Cast("const char*") BytePointer attr_name,
+    @Cast("unsigned char*") byte[] is_list, TF_Status status);
+public static native @Cast("TF_AttrType") int TFE_OpNameGetAttrType(
+    TFE_Context ctx, String op_or_function_name, String attr_name,
+    @Cast("unsigned char*") BytePointer is_list, TF_Status status);
+public static native @Cast("TF_AttrType") int TFE_OpNameGetAttrType(
+    TFE_Context ctx, @Cast("const char*") BytePointer op_or_function_name, @Cast("const char*") BytePointer attr_name,
+    @Cast("unsigned char*") ByteBuffer is_list, TF_Status status);
+public static native @Cast("TF_AttrType") int TFE_OpNameGetAttrType(
+    TFE_Context ctx, String op_or_function_name, String attr_name,
+    @Cast("unsigned char*") byte[] is_list, TF_Status status);
+
+public static native void TFE_OpSetAttrString(TFE_Op op,
+                                               @Cast("const char*") BytePointer attr_name,
+                                               @Const Pointer value,
+                                               @Cast("size_t") long length);
+public static native void TFE_OpSetAttrString(TFE_Op op,
+                                               String attr_name,
+                                               @Const Pointer value,
+                                               @Cast("size_t") long length);
+public static native void TFE_OpSetAttrInt(TFE_Op op, @Cast("const char*") BytePointer attr_name,
+                                            @Cast("int64_t") long value);
+public static native void TFE_OpSetAttrInt(TFE_Op op, String attr_name,
+                                            @Cast("int64_t") long value);
+public static native void TFE_OpSetAttrFloat(TFE_Op op, @Cast("const char*") BytePointer attr_name,
+                                              float value);
+public static native void TFE_OpSetAttrFloat(TFE_Op op, String attr_name,
+                                              float value);
+public static native void TFE_OpSetAttrBool(TFE_Op op, @Cast("const char*") BytePointer attr_name,
+                                             @Cast("unsigned char") byte value);
+public static native void TFE_OpSetAttrBool(TFE_Op op, String attr_name,
+                                             @Cast("unsigned char") byte value);
+public static native void TFE_OpSetAttrType(TFE_Op op, @Cast("const char*") BytePointer attr_name,
+                                             @Cast("TF_DataType") int value);
+public static native void TFE_OpSetAttrType(TFE_Op op, String attr_name,
+                                             @Cast("TF_DataType") int value);
+// If the number of dimensions is unknown, `num_dims` must be set to
+// -1 and `dims` can be null.  If a dimension is unknown, the
+// corresponding entry in the `dims` array must be -1.
+public static native void TFE_OpSetAttrShape(TFE_Op op, @Cast("const char*") BytePointer attr_name,
+                                              @Cast("const int64_t*") LongPointer dims,
+                                              int num_dims,
+                                              TF_Status out_status);
+public static native void TFE_OpSetAttrShape(TFE_Op op, String attr_name,
+                                              @Cast("const int64_t*") LongBuffer dims,
+                                              int num_dims,
+                                              TF_Status out_status);
+public static native void TFE_OpSetAttrShape(TFE_Op op, @Cast("const char*") BytePointer attr_name,
+                                              @Cast("const int64_t*") long[] dims,
+                                              int num_dims,
+                                              TF_Status out_status);
+public static native void TFE_OpSetAttrShape(TFE_Op op, String attr_name,
+                                              @Cast("const int64_t*") LongPointer dims,
+                                              int num_dims,
+                                              TF_Status out_status);
+public static native void TFE_OpSetAttrShape(TFE_Op op, @Cast("const char*") BytePointer attr_name,
+                                              @Cast("const int64_t*") LongBuffer dims,
+                                              int num_dims,
+                                              TF_Status out_status);
+public static native void TFE_OpSetAttrShape(TFE_Op op, String attr_name,
+                                              @Cast("const int64_t*") long[] dims,
+                                              int num_dims,
+                                              TF_Status out_status);
+
+// Sets the attribute attr_name to be a function specified by 'function'.
+//
+// TODO(ashankar,iga): Add this functionality to the C API for graph
+// construction. Perhaps we want an AttrValueMap equivalent in the C API?
+public static native void TFE_OpSetAttrFunction(TFE_Op op,
+                                                 @Cast("const char*") BytePointer attr_name,
+                                                 @Const TFE_Op value);
+public static native void TFE_OpSetAttrFunction(TFE_Op op,
+                                                 String attr_name,
+                                                 @Const TFE_Op value);
+
+public static native void TFE_OpSetAttrFunctionName(TFE_Op op, @Cast("const char*") BytePointer attr_name,
+                                              @Cast("const char*") BytePointer data, @Cast("size_t") long length);
+public static native void TFE_OpSetAttrFunctionName(TFE_Op op, String attr_name,
+                                              String data, @Cast("size_t") long length);
+
+public static native void TFE_OpSetAttrTensor(TFE_Op op,
+                                               @Cast("const char*") BytePointer attr_name,
+                                               TF_Tensor tensor,
+                                               TF_Status status);
+public static native void TFE_OpSetAttrTensor(TFE_Op op,
+                                               String attr_name,
+                                               TF_Tensor tensor,
+                                               TF_Status status);
+
+public static native void TFE_OpSetAttrStringList(TFE_Op op,
+                                                   @Cast("const char*") BytePointer attr_name,
+                                                   @Cast("const void*const*") PointerPointer values,
+                                                   @Cast("const size_t*") SizeTPointer lengths,
+                                                   int num_values);
+public static native void TFE_OpSetAttrStringList(TFE_Op op,
+                                                   @Cast("const char*") BytePointer attr_name,
+                                                   @Cast("const void*const*") @ByPtrPtr Pointer values,
+                                                   @Cast("const size_t*") SizeTPointer lengths,
+                                                   int num_values);
+public static native void TFE_OpSetAttrStringList(TFE_Op op,
+                                                   String attr_name,
+                                                   @Cast("const void*const*") @ByPtrPtr Pointer values,
+                                                   @Cast("const size_t*") SizeTPointer lengths,
+                                                   int num_values);
+public static native void TFE_OpSetAttrIntList(TFE_Op op,
+                                                @Cast("const char*") BytePointer attr_name,
+                                                @Cast("const int64_t*") LongPointer values,
+                                                int num_values);
+public static native void TFE_OpSetAttrIntList(TFE_Op op,
+                                                String attr_name,
+                                                @Cast("const int64_t*") LongBuffer values,
+                                                int num_values);
+public static native void TFE_OpSetAttrIntList(TFE_Op op,
+                                                @Cast("const char*") BytePointer attr_name,
+                                                @Cast("const int64_t*") long[] values,
+                                                int num_values);
+public static native void TFE_OpSetAttrIntList(TFE_Op op,
+                                                String attr_name,
+                                                @Cast("const int64_t*") LongPointer values,
+                                                int num_values);
+public static native void TFE_OpSetAttrIntList(TFE_Op op,
+                                                @Cast("const char*") BytePointer attr_name,
+                                                @Cast("const int64_t*") LongBuffer values,
+                                                int num_values);
+public static native void TFE_OpSetAttrIntList(TFE_Op op,
+                                                String attr_name,
+                                                @Cast("const int64_t*") long[] values,
+                                                int num_values);
+public static native void TFE_OpSetAttrFloatList(TFE_Op op,
+                                                  @Cast("const char*") BytePointer attr_name,
+                                                  @Const FloatPointer values,
+                                                  int num_values);
+public static native void TFE_OpSetAttrFloatList(TFE_Op op,
+                                                  String attr_name,
+                                                  @Const FloatBuffer values,
+                                                  int num_values);
+public static native void TFE_OpSetAttrFloatList(TFE_Op op,
+                                                  @Cast("const char*") BytePointer attr_name,
+                                                  @Const float[] values,
+                                                  int num_values);
+public static native void TFE_OpSetAttrFloatList(TFE_Op op,
+                                                  String attr_name,
+                                                  @Const FloatPointer values,
+                                                  int num_values);
+public static native void TFE_OpSetAttrFloatList(TFE_Op op,
+                                                  @Cast("const char*") BytePointer attr_name,
+                                                  @Const FloatBuffer values,
+                                                  int num_values);
+public static native void TFE_OpSetAttrFloatList(TFE_Op op,
+                                                  String attr_name,
+                                                  @Const float[] values,
+                                                  int num_values);
+public static native void TFE_OpSetAttrBoolList(TFE_Op op,
+                                                 @Cast("const char*") BytePointer attr_name,
+                                                 @Cast("const unsigned char*") BytePointer values,
+                                                 int num_values);
+public static native void TFE_OpSetAttrBoolList(TFE_Op op,
+                                                 String attr_name,
+                                                 @Cast("const unsigned char*") ByteBuffer values,
+                                                 int num_values);
+public static native void TFE_OpSetAttrBoolList(TFE_Op op,
+                                                 @Cast("const char*") BytePointer attr_name,
+                                                 @Cast("const unsigned char*") byte[] values,
+                                                 int num_values);
+public static native void TFE_OpSetAttrBoolList(TFE_Op op,
+                                                 String attr_name,
+                                                 @Cast("const unsigned char*") BytePointer values,
+                                                 int num_values);
+public static native void TFE_OpSetAttrBoolList(TFE_Op op,
+                                                 @Cast("const char*") BytePointer attr_name,
+                                                 @Cast("const unsigned char*") ByteBuffer values,
+                                                 int num_values);
+public static native void TFE_OpSetAttrBoolList(TFE_Op op,
+                                                 String attr_name,
+                                                 @Cast("const unsigned char*") byte[] values,
+                                                 int num_values);
+public static native void TFE_OpSetAttrTypeList(TFE_Op op,
+                                                 @Cast("const char*") BytePointer attr_name,
+                                                 @Cast("const TF_DataType*") IntPointer values,
+                                                 int num_values);
+public static native void TFE_OpSetAttrTypeList(TFE_Op op,
+                                                 String attr_name,
+                                                 @Cast("const TF_DataType*") IntBuffer values,
+                                                 int num_values);
+public static native void TFE_OpSetAttrTypeList(TFE_Op op,
+                                                 @Cast("const char*") BytePointer attr_name,
+                                                 @Cast("const TF_DataType*") int[] values,
+                                                 int num_values);
+public static native void TFE_OpSetAttrTypeList(TFE_Op op,
+                                                 String attr_name,
+                                                 @Cast("const TF_DataType*") IntPointer values,
+                                                 int num_values);
+public static native void TFE_OpSetAttrTypeList(TFE_Op op,
+                                                 @Cast("const char*") BytePointer attr_name,
+                                                 @Cast("const TF_DataType*") IntBuffer values,
+                                                 int num_values);
+public static native void TFE_OpSetAttrTypeList(TFE_Op op,
+                                                 String attr_name,
+                                                 @Cast("const TF_DataType*") int[] values,
+                                                 int num_values);
+public static native void TFE_OpSetAttrShapeList(
+    TFE_Op op, @Cast("const char*") BytePointer attr_name, @Cast("const int64_t**") PointerPointer dims,
+    @Const IntPointer num_dims, int num_values, TF_Status out_status);
+public static native void TFE_OpSetAttrShapeList(
+    TFE_Op op, @Cast("const char*") BytePointer attr_name, @Cast("const int64_t**") @ByPtrPtr LongPointer dims,
+    @Const IntPointer num_dims, int num_values, TF_Status out_status);
+public static native void TFE_OpSetAttrShapeList(
+    TFE_Op op, String attr_name, @Cast("const int64_t**") @ByPtrPtr LongBuffer dims,
+    @Const IntBuffer num_dims, int num_values, TF_Status out_status);
+public static native void TFE_OpSetAttrShapeList(
+    TFE_Op op, @Cast("const char*") BytePointer attr_name, @Cast("const int64_t**") @ByPtrPtr long[] dims,
+    @Const int[] num_dims, int num_values, TF_Status out_status);
+public static native void TFE_OpSetAttrShapeList(
+    TFE_Op op, String attr_name, @Cast("const int64_t**") @ByPtrPtr LongPointer dims,
+    @Const IntPointer num_dims, int num_values, TF_Status out_status);
+public static native void TFE_OpSetAttrShapeList(
+    TFE_Op op, @Cast("const char*") BytePointer attr_name, @Cast("const int64_t**") @ByPtrPtr LongBuffer dims,
+    @Const IntBuffer num_dims, int num_values, TF_Status out_status);
+public static native void TFE_OpSetAttrShapeList(
+    TFE_Op op, String attr_name, @Cast("const int64_t**") @ByPtrPtr long[] dims,
+    @Const int[] num_dims, int num_values, TF_Status out_status);
+public static native void TFE_OpSetAttrFunctionList(TFE_Op op,
+                                                     @Cast("const char*") BytePointer attr_name,
+                                                     @Cast("const TFE_Op**") PointerPointer value,
+                                                     int num_values);
+public static native void TFE_OpSetAttrFunctionList(TFE_Op op,
+                                                     @Cast("const char*") BytePointer attr_name,
+                                                     @Const @ByPtrPtr TFE_Op value,
+                                                     int num_values);
+public static native void TFE_OpSetAttrFunctionList(TFE_Op op,
+                                                     String attr_name,
+                                                     @Const @ByPtrPtr TFE_Op value,
+                                                     int num_values);
+
+// Returns the length (number of tensors) of the input argument `input_name`
+// found in the provided `op`.
+public static native int TFE_OpGetInputLength(TFE_Op op,
+                                               @Cast("const char*") BytePointer input_name,
+                                               TF_Status status);
+public static native int TFE_OpGetInputLength(TFE_Op op,
+                                               String input_name,
+                                               TF_Status status);
+
+// Returns the length (number of tensors) of the output argument `output_name`
+// found in the provided `op`.
+public static native int TFE_OpGetOutputLength(TFE_Op op,
+                                                @Cast("const char*") BytePointer output_name,
+                                                TF_Status status);
+public static native int TFE_OpGetOutputLength(TFE_Op op,
+                                                String output_name,
+                                                TF_Status status);
+
+// Execute the operation defined by 'op' and return handles to computed
+// tensors in `retvals`.
+//
+// 'retvals' must point to a pre-allocated array of TFE_TensorHandle* and
+// '*num_retvals' should be set to the size of this array. It is an error if
+// the size of 'retvals' is less than the number of outputs. This call sets
+// *num_retvals to the number of outputs.
+//
+// If async execution is enabled, the call may simply enqueue the execution
+// and return "non-ready" handles in `retvals`. Note that any handles contained
+// in 'op' should not be mutated till the kernel execution actually finishes.
+//
+// For sync execution, if any of the inputs to `op` are not ready, this call
+// will block till they become ready and then return when the kernel execution
+// is done.
+// TODO(agarwal): change num_retvals to int from int*.
+public static native void TFE_Execute(TFE_Op op, @Cast("TFE_TensorHandle**") PointerPointer retvals,
+                                       IntPointer num_retvals, TF_Status status);
+public static native void TFE_Execute(TFE_Op op, @ByPtrPtr TFE_TensorHandle retvals,
+                                       IntPointer num_retvals, TF_Status status);
+public static native void TFE_Execute(TFE_Op op, @ByPtrPtr TFE_TensorHandle retvals,
+                                       IntBuffer num_retvals, TF_Status status);
+public static native void TFE_Execute(TFE_Op op, @ByPtrPtr TFE_TensorHandle retvals,
+                                       int[] num_retvals, TF_Status status);
+
+// Add a function (serialized FunctionDef protocol buffer) to ctx so
+// that it can be invoked using TFE_Execute.
+public static native void TFE_ContextAddFunctionDef(
+    TFE_Context ctx, @Cast("const char*") BytePointer serialized_function_def, @Cast("size_t") long size,
+    TF_Status status);
+public static native void TFE_ContextAddFunctionDef(
+    TFE_Context ctx, String serialized_function_def, @Cast("size_t") long size,
+    TF_Status status);
+
+// Adds a function (created from TF_GraphToFunction or
+// TF_FunctionImportFunctionDef) to the context, allowing it to be executed with
+// TFE_Execute by creating an op with the same name as the function.
+public static native void TFE_ContextAddFunction(TFE_Context ctx,
+                                                  TF_Function function,
+                                                  TF_Status status);
+
+// Removes a function from the context. Once removed, you can no longer
+// TFE_Execute it or TFE_Execute any TFE_Op which has it as an attribute or any
+// other function which calls it as an attribute.
+public static native void TFE_ContextRemoveFunction(TFE_Context ctx,
+                                                     @Cast("const char*") BytePointer name,
+                                                     TF_Status status);
+public static native void TFE_ContextRemoveFunction(TFE_Context ctx,
+                                                     String name,
+                                                     TF_Status status);
+
+// Checks whether a function is registered under `name`.
+public static native @Cast("unsigned char") byte TFE_ContextHasFunction(TFE_Context ctx,
+                                                    @Cast("const char*") BytePointer name);
+public static native @Cast("unsigned char") byte TFE_ContextHasFunction(TFE_Context ctx,
+                                                    String name);
+
+// Enables tracing of RunMetadata on the ops executed from this context.
+public static native void TFE_ContextEnableRunMetadata(TFE_Context ctx);
+
+// Disables tracing of RunMetadata on the ops executed from this context.
+public static native void TFE_ContextDisableRunMetadata(TFE_Context ctx);
+
+// Populates the passed-in buffer with a serialized RunMetadata protocol buffer
+// containing any run metadata information accumulated so far and clears this
+// information.
+// If async mode is enabled, this call blocks till all currently pending ops are
+// done.
+public static native void TFE_ContextExportRunMetadata(TFE_Context ctx,
+                                                        TF_Buffer buf,
+                                                        TF_Status status);
+
+// Some TF ops need a step container to be set to limit the lifetime of some
+// resources (mostly TensorArray and Stack, used in while loop gradients in
+// graph mode). Calling this on a context tells it to start a step.
+public static native void TFE_ContextStartStep(TFE_Context ctx);
+
+// Ends a step. When there is no active step (that is, every started step has
+// been ended) step containers will be cleared. Note: it is not safe to call
+// TFE_ContextEndStep while ops which rely on the step container may be running.
+public static native void TFE_ContextEndStep(TFE_Context ctx);
+
+// #ifdef __cplusplus
+// Targeting ../Tensor.java
+
+
+  // namespace tensorflow
+
+
+// #endif
+
+// #endif  // TENSORFLOW_C_EAGER_C_API_H_
+
+
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/audio/AudioSpectrogram.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/audio/AudioSpectrogram.java
new file mode 100644
index 00000000000..28fdb7791bc
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/audio/AudioSpectrogram.java
@@ -0,0 +1,140 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.audio;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+
+/**
+ * Produces a visualization of audio data over time.
+ * <p>
+ * Spectrograms are a standard way of representing audio information as a series of
+ * slices of frequency information, one slice for each window of time. By joining
+ * these together into a sequence, they form a distinctive fingerprint of the sound
+ * over time.
+ * <p>
+ * This op expects to receive audio data as an input, stored as floats in the range
+ * -1 to 1, together with a window width in samples, and a stride specifying how
+ * far to move the window between slices. From this it generates a three
+ * dimensional output. The first dimension is for the channels in the input, so a
+ * stereo audio input would have two here for example. The second dimension is time,
+ * with successive frequency slices. The third dimension has an amplitude value for
+ * each frequency during that time slice.
+ * <p>
+ * This means the layout when converted and saved as an image is rotated 90 degrees
+ * clockwise from a typical spectrogram. Time is descending down the Y axis, and
+ * the frequency decreases from left to right.
+ * <p>
+ * Each value in the result represents the square root of the sum of the real and
+ * imaginary parts of an FFT on the current window of samples. In this way, the
+ * lowest dimension represents the power of each frequency in the current window,
+ * and adjacent windows are concatenated in the next dimension.
+ * <p>
+ * To get a more intuitive and visual look at what this operation does, you can run
+ * tensorflow/examples/wav_to_spectrogram to read in an audio file and save out the
+ * resulting spectrogram as a PNG image.
+ */
+@Operator(group = "audio")
+public final class AudioSpectrogram extends RawOp implements Operand<TFloat32> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.audio.AudioSpectrogram}
+   */
+  public static class Options {
+    
+    /**
+     * @param magnitudeSquared Whether to return the squared magnitude or just the
+     * magnitude. Using squared magnitude can avoid extra calculations.
+     */
+    public Options magnitudeSquared(Boolean magnitudeSquared) {
+      this.magnitudeSquared = magnitudeSquared;
+      return this;
+    }
+    
+    private Boolean magnitudeSquared;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new AudioSpectrogram operation.
+   * 
+   * @param scope current scope
+   * @param input Float representation of audio data.
+   * @param windowSize How wide the input window is in samples. For the highest efficiency
+   * this should be a power of two, but other values are accepted.
+   * @param stride How widely apart the center of adjacent sample windows should be.
+   * @param options carries optional attributes values
+   * @return a new instance of AudioSpectrogram
+   */
+  @Endpoint(describeByClass = true)
+  public static AudioSpectrogram create(Scope scope, Operand<TFloat32> input, Long windowSize, Long stride, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("AudioSpectrogram", scope.makeOpName("AudioSpectrogram"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("window_size", windowSize);
+    opBuilder.setAttr("stride", stride);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.magnitudeSquared != null) {
+          opBuilder.setAttr("magnitude_squared", opts.magnitudeSquared);
+        }
+      }
+    }
+    return new AudioSpectrogram(opBuilder.build());
+  }
+  
+  /**
+   * @param magnitudeSquared Whether to return the squared magnitude or just the
+   * magnitude. Using squared magnitude can avoid extra calculations.
+   */
+  public static Options magnitudeSquared(Boolean magnitudeSquared) {
+    return new Options().magnitudeSquared(magnitudeSquared);
+  }
+  
+  /**
+   * 3D representation of the audio frequencies as an image.
+   */
+  public Output<TFloat32> spectrogram() {
+    return spectrogram;
+  }
+  
+  @Override
+  public Output<TFloat32> asOutput() {
+    return spectrogram;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "AudioSpectrogram";
+  
+  private Output<TFloat32> spectrogram;
+  
+  private AudioSpectrogram(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    spectrogram = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/audio/DecodeWav.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/audio/DecodeWav.java
new file mode 100644
index 00000000000..ec80ffdaf7c
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/audio/DecodeWav.java
@@ -0,0 +1,147 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.audio;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.TString;
+
+/**
+ * Decode a 16-bit PCM WAV file to a float tensor.
+ * <p>
+ * The -32768 to 32767 signed 16-bit values will be scaled to -1.0 to 1.0 in float.
+ * <p>
+ * When desired_channels is set, if the input contains fewer channels than this
+ * then the last channel will be duplicated to give the requested number, else if
+ * the input has more channels than requested then the additional channels will be
+ * ignored.
+ * <p>
+ * If desired_samples is set, then the audio will be cropped or padded with zeroes
+ * to the requested length.
+ * <p>
+ * The first output contains a Tensor with the content of the audio samples. The
+ * lowest dimension will be the number of channels, and the second will be the
+ * number of samples. For example, a ten-sample-long stereo WAV file should give an
+ * output shape of [10, 2].
+ */
+@Operator(group = "audio")
+public final class DecodeWav extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.audio.DecodeWav}
+   */
+  public static class Options {
+    
+    /**
+     * @param desiredChannels Number of sample channels wanted.
+     */
+    public Options desiredChannels(Long desiredChannels) {
+      this.desiredChannels = desiredChannels;
+      return this;
+    }
+    
+    /**
+     * @param desiredSamples Length of audio requested.
+     */
+    public Options desiredSamples(Long desiredSamples) {
+      this.desiredSamples = desiredSamples;
+      return this;
+    }
+    
+    private Long desiredChannels;
+    private Long desiredSamples;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new DecodeWav operation.
+   * 
+   * @param scope current scope
+   * @param contents The WAV-encoded audio, usually from a file.
+   * @param options carries optional attributes values
+   * @return a new instance of DecodeWav
+   */
+  @Endpoint(describeByClass = true)
+  public static DecodeWav create(Scope scope, Operand<TString> contents, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("DecodeWav", scope.makeOpName("DecodeWav"));
+    opBuilder.addInput(contents.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.desiredChannels != null) {
+          opBuilder.setAttr("desired_channels", opts.desiredChannels);
+        }
+        if (opts.desiredSamples != null) {
+          opBuilder.setAttr("desired_samples", opts.desiredSamples);
+        }
+      }
+    }
+    return new DecodeWav(opBuilder.build());
+  }
+  
+  /**
+   * @param desiredChannels Number of sample channels wanted.
+   */
+  public static Options desiredChannels(Long desiredChannels) {
+    return new Options().desiredChannels(desiredChannels);
+  }
+  
+  /**
+   * @param desiredSamples Length of audio requested.
+   */
+  public static Options desiredSamples(Long desiredSamples) {
+    return new Options().desiredSamples(desiredSamples);
+  }
+  
+  /**
+   * 2-D with shape `[length, channels]`.
+   */
+  public Output<TFloat32> audio() {
+    return audio;
+  }
+  
+  /**
+   * Scalar holding the sample rate found in the WAV header.
+   */
+  public Output<TInt32> sampleRate() {
+    return sampleRate;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "DecodeWav";
+  
+  private Output<TFloat32> audio;
+  private Output<TInt32> sampleRate;
+  
+  private DecodeWav(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    audio = operation.output(outputIdx++);
+    sampleRate = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/audio/EncodeWav.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/audio/EncodeWav.java
new file mode 100644
index 00000000000..b1920318564
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/audio/EncodeWav.java
@@ -0,0 +1,85 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.audio;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.TString;
+
+/**
+ * Encode audio data using the WAV file format.
+ * <p>
+ * This operation will generate a string suitable to be saved out to create a .wav
+ * audio file. It will be encoded in the 16-bit PCM format. It takes in float
+ * values in the range -1.0f to 1.0f, and any outside that value will be clamped to
+ * that range.
+ * <p>
+ * `audio` is a 2-D float Tensor of shape `[length, channels]`.
+ * `sample_rate` is a scalar Tensor holding the rate to use (e.g. 44100).
+ */
+@Operator(group = "audio")
+public final class EncodeWav extends RawOp implements Operand<TString> {
+  
+  /**
+   * Factory method to create a class wrapping a new EncodeWav operation.
+   * 
+   * @param scope current scope
+   * @param audio 2-D with shape `[length, channels]`.
+   * @param sampleRate Scalar containing the sample frequency.
+   * @return a new instance of EncodeWav
+   */
+  @Endpoint(describeByClass = true)
+  public static EncodeWav create(Scope scope, Operand<TFloat32> audio, Operand<TInt32> sampleRate) {
+    OperationBuilder opBuilder = scope.env().opBuilder("EncodeWav", scope.makeOpName("EncodeWav"));
+    opBuilder.addInput(audio.asOutput());
+    opBuilder.addInput(sampleRate.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new EncodeWav(opBuilder.build());
+  }
+  
+  /**
+   * 0-D. WAV-encoded file contents.
+   */
+  public Output<TString> contents() {
+    return contents;
+  }
+  
+  @Override
+  public Output<TString> asOutput() {
+    return contents;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "EncodeWav";
+  
+  private Output<TString> contents;
+  
+  private EncodeWav(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    contents = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/audio/Mfcc.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/audio/Mfcc.java
new file mode 100644
index 00000000000..07ab4a13215
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/audio/Mfcc.java
@@ -0,0 +1,178 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.audio;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.TInt32;
+
+/**
+ * Transforms a spectrogram into a form that's useful for speech recognition.
+ * <p>
+ * Mel Frequency Cepstral Coefficients are a way of representing audio data that's
+ * been effective as an input feature for machine learning. They are created by
+ * taking the spectrum of a spectrogram (a 'cepstrum'), and discarding some of the
+ * higher frequencies that are less significant to the human ear. They have a long
+ * history in the speech recognition world, and https://en.wikipedia.org/wiki/Mel-frequency_cepstrum
+ * is a good resource to learn more.
+ */
+@Operator(group = "audio")
+public final class Mfcc extends RawOp implements Operand<TFloat32> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.audio.Mfcc}
+   */
+  public static class Options {
+    
+    /**
+     * @param upperFrequencyLimit The highest frequency to use when calculating the
+     * ceptstrum.
+     */
+    public Options upperFrequencyLimit(Float upperFrequencyLimit) {
+      this.upperFrequencyLimit = upperFrequencyLimit;
+      return this;
+    }
+    
+    /**
+     * @param lowerFrequencyLimit The lowest frequency to use when calculating the
+     * ceptstrum.
+     */
+    public Options lowerFrequencyLimit(Float lowerFrequencyLimit) {
+      this.lowerFrequencyLimit = lowerFrequencyLimit;
+      return this;
+    }
+    
+    /**
+     * @param filterbankChannelCount Resolution of the Mel bank used internally.
+     */
+    public Options filterbankChannelCount(Long filterbankChannelCount) {
+      this.filterbankChannelCount = filterbankChannelCount;
+      return this;
+    }
+    
+    /**
+     * @param dctCoefficientCount How many output channels to produce per time slice.
+     */
+    public Options dctCoefficientCount(Long dctCoefficientCount) {
+      this.dctCoefficientCount = dctCoefficientCount;
+      return this;
+    }
+    
+    private Float upperFrequencyLimit;
+    private Float lowerFrequencyLimit;
+    private Long filterbankChannelCount;
+    private Long dctCoefficientCount;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new Mfcc operation.
+   * 
+   * @param scope current scope
+   * @param spectrogram Typically produced by the Spectrogram op, with magnitude_squared
+   * set to true.
+   * @param sampleRate How many samples per second the source audio used.
+   * @param options carries optional attributes values
+   * @return a new instance of Mfcc
+   */
+  @Endpoint(describeByClass = true)
+  public static Mfcc create(Scope scope, Operand<TFloat32> spectrogram, Operand<TInt32> sampleRate, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Mfcc", scope.makeOpName("Mfcc"));
+    opBuilder.addInput(spectrogram.asOutput());
+    opBuilder.addInput(sampleRate.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.upperFrequencyLimit != null) {
+          opBuilder.setAttr("upper_frequency_limit", opts.upperFrequencyLimit);
+        }
+        if (opts.lowerFrequencyLimit != null) {
+          opBuilder.setAttr("lower_frequency_limit", opts.lowerFrequencyLimit);
+        }
+        if (opts.filterbankChannelCount != null) {
+          opBuilder.setAttr("filterbank_channel_count", opts.filterbankChannelCount);
+        }
+        if (opts.dctCoefficientCount != null) {
+          opBuilder.setAttr("dct_coefficient_count", opts.dctCoefficientCount);
+        }
+      }
+    }
+    return new Mfcc(opBuilder.build());
+  }
+  
+  /**
+   * @param upperFrequencyLimit The highest frequency to use when calculating the
+   * ceptstrum.
+   */
+  public static Options upperFrequencyLimit(Float upperFrequencyLimit) {
+    return new Options().upperFrequencyLimit(upperFrequencyLimit);
+  }
+  
+  /**
+   * @param lowerFrequencyLimit The lowest frequency to use when calculating the
+   * ceptstrum.
+   */
+  public static Options lowerFrequencyLimit(Float lowerFrequencyLimit) {
+    return new Options().lowerFrequencyLimit(lowerFrequencyLimit);
+  }
+  
+  /**
+   * @param filterbankChannelCount Resolution of the Mel bank used internally.
+   */
+  public static Options filterbankChannelCount(Long filterbankChannelCount) {
+    return new Options().filterbankChannelCount(filterbankChannelCount);
+  }
+  
+  /**
+   * @param dctCoefficientCount How many output channels to produce per time slice.
+   */
+  public static Options dctCoefficientCount(Long dctCoefficientCount) {
+    return new Options().dctCoefficientCount(dctCoefficientCount);
+  }
+  
+  /**
+   */
+  public Output<TFloat32> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<TFloat32> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Mfcc";
+  
+  private Output<TFloat32> output;
+  
+  private Mfcc(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/bitwise/BitwiseAnd.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/bitwise/BitwiseAnd.java
new file mode 100644
index 00000000000..2e3d0cb2949
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/bitwise/BitwiseAnd.java
@@ -0,0 +1,96 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.bitwise;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Elementwise computes the bitwise AND of `x` and `y`.
+ * <p>
+ * The result will have those bits set, that are set in both `x` and `y`. The
+ * computation is performed on the underlying representations of `x` and `y`.
+ * <p>
+ * For example:
+ * <pre>{@code
+ * import tensorflow as tf
+ * from tensorflow.python.ops import bitwise_ops
+ * dtype_list = [tf.int8, tf.int16, tf.int32, tf.int64,
+ *               tf.uint8, tf.uint16, tf.uint32, tf.uint64]
+ * 
+ * for dtype in dtype_list:
+ *   lhs = tf.constant([0, 5, 3, 14], dtype=dtype)
+ *   rhs = tf.constant([5, 0, 7, 11], dtype=dtype)
+ *   exp = tf.constant([0, 0, 3, 10], dtype=tf.float32)
+ * 
+ *   res = bitwise_ops.bitwise_and(lhs, rhs)
+ *   tf.assert_equal(tf.cast(res, tf.float32), exp) # TRUE
+ * }</pre>
+ * 
+ * 
+ * @param <T> data type for {@code z()} output
+ */
+@Operator(group = "bitwise")
+public final class BitwiseAnd<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new BitwiseAnd operation.
+   * 
+   * @param scope current scope
+   * @param x 
+   * @param y 
+   * @return a new instance of BitwiseAnd
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> BitwiseAnd<T> create(Scope scope, Operand<T> x, Operand<T> y) {
+    OperationBuilder opBuilder = scope.env().opBuilder("BitwiseAnd", scope.makeOpName("BitwiseAnd"));
+    opBuilder.addInput(x.asOutput());
+    opBuilder.addInput(y.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new BitwiseAnd<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> z() {
+    return z;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return z;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "BitwiseAnd";
+  
+  private Output<T> z;
+  
+  private BitwiseAnd(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    z = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/bitwise/BitwiseOr.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/bitwise/BitwiseOr.java
new file mode 100644
index 00000000000..abeb509f5c7
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/bitwise/BitwiseOr.java
@@ -0,0 +1,96 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.bitwise;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Elementwise computes the bitwise OR of `x` and `y`.
+ * <p>
+ * The result will have those bits set, that are set in `x`, `y` or both. The
+ * computation is performed on the underlying representations of `x` and `y`.
+ * <p>
+ * For example:
+ * <pre>{@code
+ * import tensorflow as tf
+ * from tensorflow.python.ops import bitwise_ops
+ * dtype_list = [tf.int8, tf.int16, tf.int32, tf.int64,
+ *               tf.uint8, tf.uint16, tf.uint32, tf.uint64]
+ * 
+ * for dtype in dtype_list:
+ *   lhs = tf.constant([0, 5, 3, 14], dtype=dtype)
+ *   rhs = tf.constant([5, 0, 7, 11], dtype=dtype)
+ *   exp = tf.constant([5, 5, 7, 15], dtype=tf.float32)
+ * 
+ *   res = bitwise_ops.bitwise_or(lhs, rhs)
+ *   tf.assert_equal(tf.cast(res,  tf.float32), exp)  # TRUE
+ * }</pre>
+ * 
+ * 
+ * @param <T> data type for {@code z()} output
+ */
+@Operator(group = "bitwise")
+public final class BitwiseOr<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new BitwiseOr operation.
+   * 
+   * @param scope current scope
+   * @param x 
+   * @param y 
+   * @return a new instance of BitwiseOr
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> BitwiseOr<T> create(Scope scope, Operand<T> x, Operand<T> y) {
+    OperationBuilder opBuilder = scope.env().opBuilder("BitwiseOr", scope.makeOpName("BitwiseOr"));
+    opBuilder.addInput(x.asOutput());
+    opBuilder.addInput(y.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new BitwiseOr<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> z() {
+    return z;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return z;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "BitwiseOr";
+  
+  private Output<T> z;
+  
+  private BitwiseOr(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    z = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/bitwise/BitwiseXor.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/bitwise/BitwiseXor.java
new file mode 100644
index 00000000000..4339d2d1df9
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/bitwise/BitwiseXor.java
@@ -0,0 +1,96 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.bitwise;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Elementwise computes the bitwise XOR of `x` and `y`.
+ * <p>
+ * The result will have those bits set, that are different in `x` and `y`. The
+ * computation is performed on the underlying representations of `x` and `y`.
+ * <p>
+ * For example:
+ * <pre>{@code
+ * import tensorflow as tf
+ * from tensorflow.python.ops import bitwise_ops
+ * dtype_list = [tf.int8, tf.int16, tf.int32, tf.int64,
+ *               tf.uint8, tf.uint16, tf.uint32, tf.uint64]
+ * 
+ * for dtype in dtype_list:
+ *   lhs = tf.constant([0, 5, 3, 14], dtype=dtype)
+ *   rhs = tf.constant([5, 0, 7, 11], dtype=dtype)
+ *   exp = tf.constant([5, 5, 4, 5],  dtype=tf.float32)
+ * 
+ *   res = bitwise_ops.bitwise_xor(lhs, rhs)
+ *   tf.assert_equal(tf.cast(res, tf.float32), exp) # TRUE
+ * }</pre>
+ * 
+ * 
+ * @param <T> data type for {@code z()} output
+ */
+@Operator(group = "bitwise")
+public final class BitwiseXor<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new BitwiseXor operation.
+   * 
+   * @param scope current scope
+   * @param x 
+   * @param y 
+   * @return a new instance of BitwiseXor
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> BitwiseXor<T> create(Scope scope, Operand<T> x, Operand<T> y) {
+    OperationBuilder opBuilder = scope.env().opBuilder("BitwiseXor", scope.makeOpName("BitwiseXor"));
+    opBuilder.addInput(x.asOutput());
+    opBuilder.addInput(y.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new BitwiseXor<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> z() {
+    return z;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return z;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "BitwiseXor";
+  
+  private Output<T> z;
+  
+  private BitwiseXor(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    z = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/bitwise/Invert.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/bitwise/Invert.java
new file mode 100644
index 00000000000..589f0047356
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/bitwise/Invert.java
@@ -0,0 +1,115 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.bitwise;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Invert (flip) each bit of supported types; for example, type `uint8` value 01010101 becomes 10101010.
+ * <p>
+ * Flip each bit of supported types.  For example, type `int8` (decimal 2) binary 00000010 becomes (decimal -3) binary 11111101.
+ * This operation is performed on each element of the tensor argument `x`.
+ * <p>
+ * Example:
+ * <pre>{@code
+ * import tensorflow as tf
+ * from tensorflow.python.ops import bitwise_ops
+ * 
+ * # flip 2 (00000010) to -3 (11111101)
+ * tf.assert_equal(-3, bitwise_ops.invert(2))
+ * 
+ * dtype_list = [dtypes.int8, dtypes.int16, dtypes.int32, dtypes.int64,
+ *               dtypes.uint8, dtypes.uint16, dtypes.uint32, dtypes.uint64]
+ * 
+ * inputs = [0, 5, 3, 14]
+ * for dtype in dtype_list:
+ *   # Because of issues with negative numbers, let's test this indirectly.
+ *   # 1. invert(a) and a = 0
+ *   # 2. invert(a) or a = invert(0)
+ *   input_tensor = tf.constant([0, 5, 3, 14], dtype=dtype)
+ *   not_a_and_a, not_a_or_a, not_0 = [bitwise_ops.bitwise_and(
+ *                                       input_tensor, bitwise_ops.invert(input_tensor)),
+ *                                     bitwise_ops.bitwise_or(
+ *                                       input_tensor, bitwise_ops.invert(input_tensor)),
+ *                                     bitwise_ops.invert(
+ *                                       tf.constant(0, dtype=dtype))]
+ * 
+ *   expected = tf.constant([0, 0, 0, 0], dtype=tf.float32)
+ *   tf.assert_equal(tf.cast(not_a_and_a, tf.float32), expected)
+ * 
+ *   expected = tf.cast([not_0] * 4, tf.float32)
+ *   tf.assert_equal(tf.cast(not_a_or_a, tf.float32), expected)
+ * 
+ *   # For unsigned dtypes let's also check the result directly.
+ *   if dtype.is_unsigned:
+ *     inverted = bitwise_ops.invert(input_tensor)
+ *     expected = tf.constant([dtype.max - x for x in inputs], dtype=tf.float32)
+ *     tf.assert_equal(tf.cast(inverted, tf.float32), tf.cast(expected, tf.float32))
+ * }</pre>
+ * 
+ * 
+ * @param <T> data type for {@code y()} output
+ */
+@Operator(group = "bitwise")
+public final class Invert<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new Invert operation.
+   * 
+   * @param scope current scope
+   * @param x 
+   * @return a new instance of Invert
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> Invert<T> create(Scope scope, Operand<T> x) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Invert", scope.makeOpName("Invert"));
+    opBuilder.addInput(x.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new Invert<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> y() {
+    return y;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return y;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Invert";
+  
+  private Output<T> y;
+  
+  private Invert(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    y = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/bitwise/LeftShift.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/bitwise/LeftShift.java
new file mode 100644
index 00000000000..69c818873e9
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/bitwise/LeftShift.java
@@ -0,0 +1,107 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.bitwise;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Elementwise computes the bitwise left-shift of `x` and `y`.
+ * <p>
+ * If `y` is negative, or greater than or equal to the width of `x` in bits the
+ * result is implementation defined.
+ * <p>
+ * Example:
+ * <pre>{@code
+ * import tensorflow as tf
+ * from tensorflow.python.ops import bitwise_ops
+ * import numpy as np
+ * dtype_list = [tf.int8, tf.int16, tf.int32, tf.int64]
+ * 
+ * for dtype in dtype_list:
+ *   lhs = tf.constant([-1, -5, -3, -14], dtype=dtype)
+ *   rhs = tf.constant([5, 0, 7, 11], dtype=dtype)
+ * 
+ *   left_shift_result = bitwise_ops.left_shift(lhs, rhs)
+ * 
+ *   print(left_shift_result)
+ * 
+ * # This will print:
+ * # tf.Tensor([ -32   -5 -128    0], shape=(4,), dtype=int8)
+ * # tf.Tensor([   -32     -5   -384 -28672], shape=(4,), dtype=int16)
+ * # tf.Tensor([   -32     -5   -384 -28672], shape=(4,), dtype=int32)
+ * # tf.Tensor([   -32     -5   -384 -28672], shape=(4,), dtype=int64)
+ * 
+ * lhs = np.array([-2, 64, 101, 32], dtype=np.int8)
+ * rhs = np.array([-1, -5, -3, -14], dtype=np.int8)
+ * bitwise_ops.left_shift(lhs, rhs)
+ * # <tf.Tensor: shape=(4,), dtype=int8, numpy=array([ -2,  64, 101,  32], dtype=int8)>
+ * }</pre>
+ * 
+ * 
+ * @param <T> data type for {@code z()} output
+ */
+@Operator(group = "bitwise")
+public final class LeftShift<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new LeftShift operation.
+   * 
+   * @param scope current scope
+   * @param x 
+   * @param y 
+   * @return a new instance of LeftShift
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> LeftShift<T> create(Scope scope, Operand<T> x, Operand<T> y) {
+    OperationBuilder opBuilder = scope.env().opBuilder("LeftShift", scope.makeOpName("LeftShift"));
+    opBuilder.addInput(x.asOutput());
+    opBuilder.addInput(y.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new LeftShift<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> z() {
+    return z;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return z;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "LeftShift";
+  
+  private Output<T> z;
+  
+  private LeftShift(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    z = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/bitwise/RightShift.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/bitwise/RightShift.java
new file mode 100644
index 00000000000..8106172ee4f
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/bitwise/RightShift.java
@@ -0,0 +1,110 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.bitwise;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Elementwise computes the bitwise right-shift of `x` and `y`.
+ * <p>
+ * Performs a logical shift for unsigned integer types, and an arithmetic shift
+ * for signed integer types.
+ * <p>
+ * If `y` is negative, or greater than or equal to than the width of `x` in bits
+ * the result is implementation defined.
+ * <p>
+ * Example:
+ * <pre>{@code
+ * import tensorflow as tf
+ * from tensorflow.python.ops import bitwise_ops
+ * import numpy as np
+ * dtype_list = [tf.int8, tf.int16, tf.int32, tf.int64]
+ * 
+ * for dtype in dtype_list:
+ *   lhs = tf.constant([-1, -5, -3, -14], dtype=dtype)
+ *   rhs = tf.constant([5, 0, 7, 11], dtype=dtype)
+ * 
+ *   right_shift_result = bitwise_ops.right_shift(lhs, rhs)
+ * 
+ *   print(right_shift_result)
+ * 
+ * # This will print:
+ * # tf.Tensor([-1 -5 -1 -1], shape=(4,), dtype=int8)
+ * # tf.Tensor([-1 -5 -1 -1], shape=(4,), dtype=int16)
+ * # tf.Tensor([-1 -5 -1 -1], shape=(4,), dtype=int32)
+ * # tf.Tensor([-1 -5 -1 -1], shape=(4,), dtype=int64)
+ * 
+ * lhs = np.array([-2, 64, 101, 32], dtype=np.int8)
+ * rhs = np.array([-1, -5, -3, -14], dtype=np.int8)
+ * bitwise_ops.right_shift(lhs, rhs)
+ * # <tf.Tensor: shape=(4,), dtype=int8, numpy=array([ -2,  64, 101,  32], dtype=int8)>
+ * }</pre>
+ * 
+ * 
+ * @param <T> data type for {@code z()} output
+ */
+@Operator(group = "bitwise")
+public final class RightShift<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new RightShift operation.
+   * 
+   * @param scope current scope
+   * @param x 
+   * @param y 
+   * @return a new instance of RightShift
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> RightShift<T> create(Scope scope, Operand<T> x, Operand<T> y) {
+    OperationBuilder opBuilder = scope.env().opBuilder("RightShift", scope.makeOpName("RightShift"));
+    opBuilder.addInput(x.asOutput());
+    opBuilder.addInput(y.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new RightShift<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> z() {
+    return z;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return z;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "RightShift";
+  
+  private Output<T> z;
+  
+  private RightShift(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    z = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/cluster/KMC2ChainInitialization.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/cluster/KMC2ChainInitialization.java
new file mode 100644
index 00000000000..272695dad5c
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/cluster/KMC2ChainInitialization.java
@@ -0,0 +1,81 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.cluster;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.TInt64;
+
+/**
+ * Returns the index of a data point that should be added to the seed set.
+ * <p>
+ * Entries in distances are assumed to be squared distances of candidate points to
+ * the already sampled centers in the seed set. The op constructs one Markov chain
+ * of the k-MC^2 algorithm and returns the index of one candidate point to be added
+ * as an additional cluster center.
+ */
+public final class KMC2ChainInitialization extends RawOp implements Operand<TInt64> {
+  
+  /**
+   * Factory method to create a class wrapping a new KMC2ChainInitialization operation.
+   * 
+   * @param scope current scope
+   * @param distances Vector with squared distances to the closest previously sampled cluster center
+   * for each candidate point.
+   * @param seed Scalar. Seed for initializing the random number generator.
+   * @return a new instance of KMC2ChainInitialization
+   */
+  @Endpoint(describeByClass = true)
+  public static KMC2ChainInitialization create(Scope scope, Operand<TFloat32> distances, Operand<TInt64> seed) {
+    OperationBuilder opBuilder = scope.env().opBuilder("KMC2ChainInitialization", scope.makeOpName("KMC2ChainInitialization"));
+    opBuilder.addInput(distances.asOutput());
+    opBuilder.addInput(seed.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new KMC2ChainInitialization(opBuilder.build());
+  }
+  
+  /**
+   * Scalar with the index of the sampled point.
+   */
+  public Output<TInt64> index() {
+    return index;
+  }
+  
+  @Override
+  public Output<TInt64> asOutput() {
+    return index;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "KMC2ChainInitialization";
+  
+  private Output<TInt64> index;
+  
+  private KMC2ChainInitialization(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    index = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/cluster/KmeansPlusPlusInitialization.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/cluster/KmeansPlusPlusInitialization.java
new file mode 100644
index 00000000000..63d8ef01222
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/cluster/KmeansPlusPlusInitialization.java
@@ -0,0 +1,87 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.cluster;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.TInt64;
+
+/**
+ * Selects num_to_sample rows of input using the KMeans++ criterion.
+ * <p>
+ * Rows of points are assumed to be input points. One row is selected at random.
+ * Subsequent rows are sampled with probability proportional to the squared L2
+ * distance from the nearest row selected thus far till num_to_sample rows have
+ * been sampled.
+ */
+public final class KmeansPlusPlusInitialization extends RawOp implements Operand<TFloat32> {
+  
+  /**
+   * Factory method to create a class wrapping a new KmeansPlusPlusInitialization operation.
+   * 
+   * @param scope current scope
+   * @param points Matrix of shape (n, d). Rows are assumed to be input points.
+   * @param numToSample Scalar. The number of rows to sample. This value must not be larger than n.
+   * @param seed Scalar. Seed for initializing the random number generator.
+   * @param numRetriesPerSample Scalar. For each row that is sampled, this parameter
+   * specifies the number of additional points to draw from the current
+   * distribution before selecting the best. If a negative value is specified, a
+   * heuristic is used to sample O(log(num_to_sample)) additional points.
+   * @return a new instance of KmeansPlusPlusInitialization
+   */
+  @Endpoint(describeByClass = true)
+  public static KmeansPlusPlusInitialization create(Scope scope, Operand<TFloat32> points, Operand<TInt64> numToSample, Operand<TInt64> seed, Operand<TInt64> numRetriesPerSample) {
+    OperationBuilder opBuilder = scope.env().opBuilder("KmeansPlusPlusInitialization", scope.makeOpName("KmeansPlusPlusInitialization"));
+    opBuilder.addInput(points.asOutput());
+    opBuilder.addInput(numToSample.asOutput());
+    opBuilder.addInput(seed.asOutput());
+    opBuilder.addInput(numRetriesPerSample.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new KmeansPlusPlusInitialization(opBuilder.build());
+  }
+  
+  /**
+   * Matrix of shape (num_to_sample, d). The sampled rows.
+   */
+  public Output<TFloat32> samples() {
+    return samples;
+  }
+  
+  @Override
+  public Output<TFloat32> asOutput() {
+    return samples;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "KmeansPlusPlusInitialization";
+  
+  private Output<TFloat32> samples;
+  
+  private KmeansPlusPlusInitialization(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    samples = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/collective/AllReduce.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/collective/AllReduce.java
new file mode 100644
index 00000000000..0ef37a5a9d2
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/collective/AllReduce.java
@@ -0,0 +1,166 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.collective;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Mutually reduces multiple tensors of identical type and shape.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+public final class AllReduce<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.collective.AllReduce}
+   */
+  public static class Options {
+    
+    /**
+     * @param waitFor 
+     */
+    public Options waitFor(List<Long> waitFor) {
+      this.waitFor = waitFor;
+      return this;
+    }
+    
+    /**
+     * @param communicationHint 
+     */
+    public Options communicationHint(String communicationHint) {
+      this.communicationHint = communicationHint;
+      return this;
+    }
+    
+    /**
+     * @param timeoutSeconds 
+     */
+    public Options timeoutSeconds(Float timeoutSeconds) {
+      this.timeoutSeconds = timeoutSeconds;
+      return this;
+    }
+    
+    private List<Long> waitFor;
+    private String communicationHint;
+    private Float timeoutSeconds;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new AllReduce operation.
+   * 
+   * @param scope current scope
+   * @param input 
+   * @param groupSize 
+   * @param groupKey 
+   * @param instanceKey 
+   * @param mergeOp 
+   * @param finalOp 
+   * @param subdivOffsets 
+   * @param options carries optional attributes values
+   * @return a new instance of AllReduce
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> AllReduce<T> create(Scope scope, Operand<T> input, Long groupSize, Long groupKey, Long instanceKey, String mergeOp, String finalOp, List<Long> subdivOffsets, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("CollectiveReduce", scope.makeOpName("AllReduce"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("group_size", groupSize);
+    opBuilder.setAttr("group_key", groupKey);
+    opBuilder.setAttr("instance_key", instanceKey);
+    opBuilder.setAttr("merge_op", mergeOp);
+    opBuilder.setAttr("final_op", finalOp);
+    long[] subdivOffsetsArray = new long[subdivOffsets.size()];
+    for (int i = 0; i < subdivOffsetsArray.length; ++i) {
+      subdivOffsetsArray[i] = subdivOffsets.get(i);
+    }
+    opBuilder.setAttr("subdiv_offsets", subdivOffsetsArray);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.waitFor != null) {
+          long[] waitForArray = new long[opts.waitFor.size()];
+          for (int i = 0; i < waitForArray.length; ++i) {
+            waitForArray[i] = opts.waitFor.get(i);
+          }
+          opBuilder.setAttr("wait_for", waitForArray);
+        }
+        if (opts.communicationHint != null) {
+          opBuilder.setAttr("communication_hint", opts.communicationHint);
+        }
+        if (opts.timeoutSeconds != null) {
+          opBuilder.setAttr("timeout_seconds", opts.timeoutSeconds);
+        }
+      }
+    }
+    return new AllReduce<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param waitFor 
+   */
+  public static Options waitFor(List<Long> waitFor) {
+    return new Options().waitFor(waitFor);
+  }
+  
+  /**
+   * @param communicationHint 
+   */
+  public static Options communicationHint(String communicationHint) {
+    return new Options().communicationHint(communicationHint);
+  }
+  
+  /**
+   * @param timeoutSeconds 
+   */
+  public static Options timeoutSeconds(Float timeoutSeconds) {
+    return new Options().timeoutSeconds(timeoutSeconds);
+  }
+  
+  /**
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "CollectiveReduce";
+  
+  private Output<T> output;
+  
+  private AllReduce(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/collective/BroadcastRecv.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/collective/BroadcastRecv.java
new file mode 100644
index 00000000000..8603acc0e05
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/collective/BroadcastRecv.java
@@ -0,0 +1,135 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.collective;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Receives a tensor value broadcast from another device.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+public final class BroadcastRecv<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.collective.BroadcastRecv}
+   */
+  public static class Options {
+    
+    /**
+     * @param communicationHint 
+     */
+    public Options communicationHint(String communicationHint) {
+      this.communicationHint = communicationHint;
+      return this;
+    }
+    
+    /**
+     * @param timeoutSeconds 
+     */
+    public Options timeoutSeconds(Float timeoutSeconds) {
+      this.timeoutSeconds = timeoutSeconds;
+      return this;
+    }
+    
+    private String communicationHint;
+    private Float timeoutSeconds;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new BroadcastRecv operation.
+   * 
+   * @param scope current scope
+   * @param T 
+   * @param groupSize 
+   * @param groupKey 
+   * @param instanceKey 
+   * @param shape 
+   * @param options carries optional attributes values
+   * @return a new instance of BroadcastRecv
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> BroadcastRecv<T> create(Scope scope, Class<T> T, Long groupSize, Long groupKey, Long instanceKey, Shape shape, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("CollectiveBcastRecv", scope.makeOpName("BroadcastRecv"));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("T", T);
+    opBuilder.setAttr("group_size", groupSize);
+    opBuilder.setAttr("group_key", groupKey);
+    opBuilder.setAttr("instance_key", instanceKey);
+    opBuilder.setAttr("shape", shape);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.communicationHint != null) {
+          opBuilder.setAttr("communication_hint", opts.communicationHint);
+        }
+        if (opts.timeoutSeconds != null) {
+          opBuilder.setAttr("timeout_seconds", opts.timeoutSeconds);
+        }
+      }
+    }
+    return new BroadcastRecv<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param communicationHint 
+   */
+  public static Options communicationHint(String communicationHint) {
+    return new Options().communicationHint(communicationHint);
+  }
+  
+  /**
+   * @param timeoutSeconds 
+   */
+  public static Options timeoutSeconds(Float timeoutSeconds) {
+    return new Options().timeoutSeconds(timeoutSeconds);
+  }
+  
+  /**
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "CollectiveBcastRecv";
+  
+  private Output<T> output;
+  
+  private BroadcastRecv(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/collective/BroadcastSend.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/collective/BroadcastSend.java
new file mode 100644
index 00000000000..f6b3d3ff307
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/collective/BroadcastSend.java
@@ -0,0 +1,135 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.collective;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Broadcasts a tensor value to one or more other devices.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+public final class BroadcastSend<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.collective.BroadcastSend}
+   */
+  public static class Options {
+    
+    /**
+     * @param communicationHint 
+     */
+    public Options communicationHint(String communicationHint) {
+      this.communicationHint = communicationHint;
+      return this;
+    }
+    
+    /**
+     * @param timeoutSeconds 
+     */
+    public Options timeoutSeconds(Float timeoutSeconds) {
+      this.timeoutSeconds = timeoutSeconds;
+      return this;
+    }
+    
+    private String communicationHint;
+    private Float timeoutSeconds;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new BroadcastSend operation.
+   * 
+   * @param scope current scope
+   * @param input 
+   * @param groupSize 
+   * @param groupKey 
+   * @param instanceKey 
+   * @param shape 
+   * @param options carries optional attributes values
+   * @return a new instance of BroadcastSend
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> BroadcastSend<T> create(Scope scope, Operand<T> input, Long groupSize, Long groupKey, Long instanceKey, Shape shape, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("CollectiveBcastSend", scope.makeOpName("BroadcastSend"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("group_size", groupSize);
+    opBuilder.setAttr("group_key", groupKey);
+    opBuilder.setAttr("instance_key", instanceKey);
+    opBuilder.setAttr("shape", shape);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.communicationHint != null) {
+          opBuilder.setAttr("communication_hint", opts.communicationHint);
+        }
+        if (opts.timeoutSeconds != null) {
+          opBuilder.setAttr("timeout_seconds", opts.timeoutSeconds);
+        }
+      }
+    }
+    return new BroadcastSend<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param communicationHint 
+   */
+  public static Options communicationHint(String communicationHint) {
+    return new Options().communicationHint(communicationHint);
+  }
+  
+  /**
+   * @param timeoutSeconds 
+   */
+  public static Options timeoutSeconds(Float timeoutSeconds) {
+    return new Options().timeoutSeconds(timeoutSeconds);
+  }
+  
+  /**
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "CollectiveBcastSend";
+  
+  private Output<T> output;
+  
+  private BroadcastSend(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Abort.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Abort.java
new file mode 100644
index 00000000000..a84f2405b19
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Abort.java
@@ -0,0 +1,110 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+
+/**
+ * Raise a exception to abort the process when called.
+ * <p>
+ * If exit_without_error is true, the process will exit normally,
+ * otherwise it will exit with a SIGABORT signal.
+ * <p>
+ * Returns nothing but an exception.
+ */
+@Operator
+public final class Abort extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.core.Abort}
+   */
+  public static class Options {
+    
+    /**
+     * @param errorMsg A string which is the message associated with the exception.
+     */
+    public Options errorMsg(String errorMsg) {
+      this.errorMsg = errorMsg;
+      return this;
+    }
+    
+    /**
+     * @param exitWithoutError 
+     */
+    public Options exitWithoutError(Boolean exitWithoutError) {
+      this.exitWithoutError = exitWithoutError;
+      return this;
+    }
+    
+    private String errorMsg;
+    private Boolean exitWithoutError;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new Abort operation.
+   * 
+   * @param scope current scope
+   * @param options carries optional attributes values
+   * @return a new instance of Abort
+   */
+  @Endpoint(describeByClass = true)
+  public static Abort create(Scope scope, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Abort", scope.makeOpName("Abort"));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.errorMsg != null) {
+          opBuilder.setAttr("error_msg", opts.errorMsg);
+        }
+        if (opts.exitWithoutError != null) {
+          opBuilder.setAttr("exit_without_error", opts.exitWithoutError);
+        }
+      }
+    }
+    return new Abort(opBuilder.build());
+  }
+  
+  /**
+   * @param errorMsg A string which is the message associated with the exception.
+   */
+  public static Options errorMsg(String errorMsg) {
+    return new Options().errorMsg(errorMsg);
+  }
+  
+  /**
+   * @param exitWithoutError 
+   */
+  public static Options exitWithoutError(Boolean exitWithoutError) {
+    return new Options().exitWithoutError(exitWithoutError);
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Abort";
+  
+  private Abort(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/All.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/All.java
new file mode 100644
index 00000000000..8de82899104
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/All.java
@@ -0,0 +1,116 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TBool;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Computes the "logical and" of elements across dimensions of a tensor.
+ * <p>
+ * Reduces `input` along the dimensions given in `axis`. Unless
+ * `keep_dims` is true, the rank of the tensor is reduced by 1 for each entry in
+ * `axis`. If `keep_dims` is true, the reduced dimensions are
+ * retained with length 1.
+ */
+@Operator
+public final class All extends RawOp implements Operand<TBool> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.core.All}
+   */
+  public static class Options {
+    
+    /**
+     * @param keepDims If true, retain reduced dimensions with length 1.
+     */
+    public Options keepDims(Boolean keepDims) {
+      this.keepDims = keepDims;
+      return this;
+    }
+    
+    private Boolean keepDims;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new All operation.
+   * 
+   * @param scope current scope
+   * @param input The tensor to reduce.
+   * @param axis The dimensions to reduce. Must be in the range
+   * `[-rank(input), rank(input))`.
+   * @param options carries optional attributes values
+   * @return a new instance of All
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> All create(Scope scope, Operand<TBool> input, Operand<T> axis, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("All", scope.makeOpName("All"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(axis.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.keepDims != null) {
+          opBuilder.setAttr("keep_dims", opts.keepDims);
+        }
+      }
+    }
+    return new All(opBuilder.build());
+  }
+  
+  /**
+   * @param keepDims If true, retain reduced dimensions with length 1.
+   */
+  public static Options keepDims(Boolean keepDims) {
+    return new Options().keepDims(keepDims);
+  }
+  
+  /**
+   * The reduced tensor.
+   */
+  public Output<TBool> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<TBool> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "All";
+  
+  private Output<TBool> output;
+  
+  private All(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Any.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Any.java
new file mode 100644
index 00000000000..f5e5f3753e2
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Any.java
@@ -0,0 +1,116 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TBool;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Computes the "logical or" of elements across dimensions of a tensor.
+ * <p>
+ * Reduces `input` along the dimensions given in `axis`. Unless
+ * `keep_dims` is true, the rank of the tensor is reduced by 1 for each entry in
+ * `axis`. If `keep_dims` is true, the reduced dimensions are
+ * retained with length 1.
+ */
+@Operator
+public final class Any extends RawOp implements Operand<TBool> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.core.Any}
+   */
+  public static class Options {
+    
+    /**
+     * @param keepDims If true, retain reduced dimensions with length 1.
+     */
+    public Options keepDims(Boolean keepDims) {
+      this.keepDims = keepDims;
+      return this;
+    }
+    
+    private Boolean keepDims;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new Any operation.
+   * 
+   * @param scope current scope
+   * @param input The tensor to reduce.
+   * @param axis The dimensions to reduce. Must be in the range
+   * `[-rank(input), rank(input))`.
+   * @param options carries optional attributes values
+   * @return a new instance of Any
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> Any create(Scope scope, Operand<TBool> input, Operand<T> axis, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Any", scope.makeOpName("Any"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(axis.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.keepDims != null) {
+          opBuilder.setAttr("keep_dims", opts.keepDims);
+        }
+      }
+    }
+    return new Any(opBuilder.build());
+  }
+  
+  /**
+   * @param keepDims If true, retain reduced dimensions with length 1.
+   */
+  public static Options keepDims(Boolean keepDims) {
+    return new Options().keepDims(keepDims);
+  }
+  
+  /**
+   * The reduced tensor.
+   */
+  public Output<TBool> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<TBool> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Any";
+  
+  private Output<TBool> output;
+  
+  private Any(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/AssertThat.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/AssertThat.java
new file mode 100644
index 00000000000..dce70c04e5a
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/AssertThat.java
@@ -0,0 +1,96 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.Operands;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TBool;
+
+/**
+ * Asserts that the given condition is true.
+ * <p>
+ * If `condition` evaluates to false, print the list of tensors in `data`.
+ * `summarize` determines how many entries of the tensors to print.
+ */
+@Operator
+public final class AssertThat extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.core.AssertThat}
+   */
+  public static class Options {
+    
+    /**
+     * @param summarize Print this many entries of each tensor.
+     */
+    public Options summarize(Long summarize) {
+      this.summarize = summarize;
+      return this;
+    }
+    
+    private Long summarize;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new AssertThat operation.
+   * 
+   * @param scope current scope
+   * @param condition The condition to evaluate.
+   * @param data The tensors to print out when condition is false.
+   * @param options carries optional attributes values
+   * @return a new instance of AssertThat
+   */
+  @Endpoint(describeByClass = true)
+  public static AssertThat create(Scope scope, Operand<TBool> condition, Iterable<Operand<?>> data, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Assert", scope.makeOpName("AssertThat"));
+    opBuilder.addInput(condition.asOutput());
+    opBuilder.addInputList(Operands.asOutputs(data));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.summarize != null) {
+          opBuilder.setAttr("summarize", opts.summarize);
+        }
+      }
+    }
+    return new AssertThat(opBuilder.build());
+  }
+  
+  /**
+   * @param summarize Print this many entries of each tensor.
+   */
+  public static Options summarize(Long summarize) {
+    return new Options().summarize(summarize);
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Assert";
+  
+  private AssertThat(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Assign.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Assign.java
new file mode 100644
index 00000000000..27f66d3da23
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Assign.java
@@ -0,0 +1,140 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Update 'ref' by assigning 'value' to it.
+ * <p>
+ * This operation outputs "ref" after the assignment is done.
+ * This makes it easier to chain operations that need to use the reset value.
+ * 
+ * @param <T> data type for {@code outputRef()} output
+ */
+@Operator
+public final class Assign<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.core.Assign}
+   */
+  public static class Options {
+    
+    /**
+     * @param validateShape If true, the operation will validate that the shape
+     * of 'value' matches the shape of the Tensor being assigned to.  If false,
+     * 'ref' will take on the shape of 'value'.
+     */
+    public Options validateShape(Boolean validateShape) {
+      this.validateShape = validateShape;
+      return this;
+    }
+    
+    /**
+     * @param useLocking If True, the assignment will be protected by a lock;
+     * otherwise the behavior is undefined, but may exhibit less contention.
+     */
+    public Options useLocking(Boolean useLocking) {
+      this.useLocking = useLocking;
+      return this;
+    }
+    
+    private Boolean validateShape;
+    private Boolean useLocking;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new Assign operation.
+   * 
+   * @param scope current scope
+   * @param ref Should be from a `Variable` node. May be uninitialized.
+   * @param value The value to be assigned to the variable.
+   * @param options carries optional attributes values
+   * @return a new instance of Assign
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> Assign<T> create(Scope scope, Operand<T> ref, Operand<T> value, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Assign", scope.makeOpName("Assign"));
+    opBuilder.addInput(ref.asOutput());
+    opBuilder.addInput(value.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.validateShape != null) {
+          opBuilder.setAttr("validate_shape", opts.validateShape);
+        }
+        if (opts.useLocking != null) {
+          opBuilder.setAttr("use_locking", opts.useLocking);
+        }
+      }
+    }
+    return new Assign<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param validateShape If true, the operation will validate that the shape
+   * of 'value' matches the shape of the Tensor being assigned to.  If false,
+   * 'ref' will take on the shape of 'value'.
+   */
+  public static Options validateShape(Boolean validateShape) {
+    return new Options().validateShape(validateShape);
+  }
+  
+  /**
+   * @param useLocking If True, the assignment will be protected by a lock;
+   * otherwise the behavior is undefined, but may exhibit less contention.
+   */
+  public static Options useLocking(Boolean useLocking) {
+    return new Options().useLocking(useLocking);
+  }
+  
+  /**
+   * = Same as "ref".  Returned as a convenience for operations that want
+   * to use the new value after the variable has been reset.
+   */
+  public Output<T> outputRef() {
+    return outputRef;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return outputRef;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Assign";
+  
+  private Output<T> outputRef;
+  
+  private Assign(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    outputRef = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/AssignAdd.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/AssignAdd.java
new file mode 100644
index 00000000000..770842d2764
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/AssignAdd.java
@@ -0,0 +1,117 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Update 'ref' by adding 'value' to it.
+ * <p>
+ * This operation outputs "ref" after the update is done.
+ * This makes it easier to chain operations that need to use the reset value.
+ * 
+ * @param <T> data type for {@code outputRef()} output
+ */
+@Operator
+public final class AssignAdd<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.core.AssignAdd}
+   */
+  public static class Options {
+    
+    /**
+     * @param useLocking If True, the addition will be protected by a lock;
+     * otherwise the behavior is undefined, but may exhibit less contention.
+     */
+    public Options useLocking(Boolean useLocking) {
+      this.useLocking = useLocking;
+      return this;
+    }
+    
+    private Boolean useLocking;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new AssignAdd operation.
+   * 
+   * @param scope current scope
+   * @param ref Should be from a `Variable` node.
+   * @param value The value to be added to the variable.
+   * @param options carries optional attributes values
+   * @return a new instance of AssignAdd
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> AssignAdd<T> create(Scope scope, Operand<T> ref, Operand<T> value, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("AssignAdd", scope.makeOpName("AssignAdd"));
+    opBuilder.addInput(ref.asOutput());
+    opBuilder.addInput(value.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.useLocking != null) {
+          opBuilder.setAttr("use_locking", opts.useLocking);
+        }
+      }
+    }
+    return new AssignAdd<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param useLocking If True, the addition will be protected by a lock;
+   * otherwise the behavior is undefined, but may exhibit less contention.
+   */
+  public static Options useLocking(Boolean useLocking) {
+    return new Options().useLocking(useLocking);
+  }
+  
+  /**
+   * = Same as "ref".  Returned as a convenience for operations that want
+   * to use the new value after the variable has been updated.
+   */
+  public Output<T> outputRef() {
+    return outputRef;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return outputRef;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "AssignAdd";
+  
+  private Output<T> outputRef;
+  
+  private AssignAdd(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    outputRef = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/AssignAddVariableOp.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/AssignAddVariableOp.java
new file mode 100644
index 00000000000..53edc808882
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/AssignAddVariableOp.java
@@ -0,0 +1,61 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Adds a value to the current value of a variable.
+ * <p>
+ * Any ReadVariableOp with a control dependency on this op is guaranteed to
+ * see the incremented value or a subsequent newer one.
+ */
+@Operator
+public final class AssignAddVariableOp extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new AssignAddVariableOp operation.
+   * 
+   * @param scope current scope
+   * @param resource handle to the resource in which to store the variable.
+   * @param value the value by which the variable will be incremented.
+   * @return a new instance of AssignAddVariableOp
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> AssignAddVariableOp create(Scope scope, Operand<?> resource, Operand<T> value) {
+    OperationBuilder opBuilder = scope.env().opBuilder("AssignAddVariableOp", scope.makeOpName("AssignAddVariableOp"));
+    opBuilder.addInput(resource.asOutput());
+    opBuilder.addInput(value.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new AssignAddVariableOp(opBuilder.build());
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "AssignAddVariableOp";
+  
+  private AssignAddVariableOp(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/AssignSub.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/AssignSub.java
new file mode 100644
index 00000000000..37841e0dadc
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/AssignSub.java
@@ -0,0 +1,117 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Update 'ref' by subtracting 'value' from it.
+ * <p>
+ * This operation outputs "ref" after the update is done.
+ * This makes it easier to chain operations that need to use the reset value.
+ * 
+ * @param <T> data type for {@code outputRef()} output
+ */
+@Operator
+public final class AssignSub<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.core.AssignSub}
+   */
+  public static class Options {
+    
+    /**
+     * @param useLocking If True, the subtraction will be protected by a lock;
+     * otherwise the behavior is undefined, but may exhibit less contention.
+     */
+    public Options useLocking(Boolean useLocking) {
+      this.useLocking = useLocking;
+      return this;
+    }
+    
+    private Boolean useLocking;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new AssignSub operation.
+   * 
+   * @param scope current scope
+   * @param ref Should be from a `Variable` node.
+   * @param value The value to be subtracted to the variable.
+   * @param options carries optional attributes values
+   * @return a new instance of AssignSub
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> AssignSub<T> create(Scope scope, Operand<T> ref, Operand<T> value, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("AssignSub", scope.makeOpName("AssignSub"));
+    opBuilder.addInput(ref.asOutput());
+    opBuilder.addInput(value.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.useLocking != null) {
+          opBuilder.setAttr("use_locking", opts.useLocking);
+        }
+      }
+    }
+    return new AssignSub<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param useLocking If True, the subtraction will be protected by a lock;
+   * otherwise the behavior is undefined, but may exhibit less contention.
+   */
+  public static Options useLocking(Boolean useLocking) {
+    return new Options().useLocking(useLocking);
+  }
+  
+  /**
+   * = Same as "ref".  Returned as a convenience for operations that want
+   * to use the new value after the variable has been updated.
+   */
+  public Output<T> outputRef() {
+    return outputRef;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return outputRef;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "AssignSub";
+  
+  private Output<T> outputRef;
+  
+  private AssignSub(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    outputRef = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/AssignSubVariableOp.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/AssignSubVariableOp.java
new file mode 100644
index 00000000000..372a71b2168
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/AssignSubVariableOp.java
@@ -0,0 +1,61 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Subtracts a value from the current value of a variable.
+ * <p>
+ * Any ReadVariableOp with a control dependency on this op is guaranteed to
+ * see the decremented value or a subsequent newer one.
+ */
+@Operator
+public final class AssignSubVariableOp extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new AssignSubVariableOp operation.
+   * 
+   * @param scope current scope
+   * @param resource handle to the resource in which to store the variable.
+   * @param value the value by which the variable will be incremented.
+   * @return a new instance of AssignSubVariableOp
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> AssignSubVariableOp create(Scope scope, Operand<?> resource, Operand<T> value) {
+    OperationBuilder opBuilder = scope.env().opBuilder("AssignSubVariableOp", scope.makeOpName("AssignSubVariableOp"));
+    opBuilder.addInput(resource.asOutput());
+    opBuilder.addInput(value.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new AssignSubVariableOp(opBuilder.build());
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "AssignSubVariableOp";
+  
+  private AssignSubVariableOp(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/AssignVariableOp.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/AssignVariableOp.java
new file mode 100644
index 00000000000..ac08d62f9a8
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/AssignVariableOp.java
@@ -0,0 +1,61 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Assigns a new value to a variable.
+ * <p>
+ * Any ReadVariableOp with a control dependency on this op is guaranteed to return
+ * this value or a subsequent newer value of the variable.
+ */
+@Operator
+public final class AssignVariableOp extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new AssignVariableOp operation.
+   * 
+   * @param scope current scope
+   * @param resource handle to the resource in which to store the variable.
+   * @param value the value to set the new tensor to use.
+   * @return a new instance of AssignVariableOp
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> AssignVariableOp create(Scope scope, Operand<?> resource, Operand<T> value) {
+    OperationBuilder opBuilder = scope.env().opBuilder("AssignVariableOp", scope.makeOpName("AssignVariableOp"));
+    opBuilder.addInput(resource.asOutput());
+    opBuilder.addInput(value.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new AssignVariableOp(opBuilder.build());
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "AssignVariableOp";
+  
+  private AssignVariableOp(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Barrier.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Barrier.java
new file mode 100644
index 00000000000..dc3f8b781bc
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Barrier.java
@@ -0,0 +1,194 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TString;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Defines a barrier that persists across different graph executions.
+ * <p>
+ * A barrier represents a key-value map, where each key is a string, and
+ * each value is a tuple of tensors.
+ * <p>
+ * At runtime, the barrier contains 'complete' and 'incomplete'
+ * elements. A complete element has defined tensors for all components of
+ * its value tuple, and may be accessed using BarrierTakeMany. An
+ * incomplete element has some undefined components in its value tuple,
+ * and may be updated using BarrierInsertMany.
+ */
+@Operator
+public final class Barrier extends RawOp implements Operand<TString> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.core.Barrier}
+   */
+  public static class Options {
+    
+    /**
+     * @param shapes The shape of each component in a value. Each shape must be 1 in the
+     * first dimension. The length of this attr must be the same as the length of
+     * component_types.
+     */
+    public Options shapes(List<Shape> shapes) {
+      this.shapes = shapes;
+      return this;
+    }
+    
+    /**
+     * @param capacity The capacity of the barrier.  The default capacity is MAX_INT32,
+     * which is the largest capacity of the underlying queue.
+     */
+    public Options capacity(Long capacity) {
+      this.capacity = capacity;
+      return this;
+    }
+    
+    /**
+     * @param container If non-empty, this barrier is placed in the given container.
+     * Otherwise, a default container is used.
+     */
+    public Options container(String container) {
+      this.container = container;
+      return this;
+    }
+    
+    /**
+     * @param sharedName If non-empty, this barrier will be shared under the given name
+     * across multiple sessions.
+     */
+    public Options sharedName(String sharedName) {
+      this.sharedName = sharedName;
+      return this;
+    }
+    
+    private List<Shape> shapes;
+    private Long capacity;
+    private String container;
+    private String sharedName;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new Barrier operation.
+   * 
+   * @param scope current scope
+   * @param componentTypes The type of each component in a value.
+   * @param options carries optional attributes values
+   * @return a new instance of Barrier
+   */
+  @Endpoint(describeByClass = true)
+  public static Barrier create(Scope scope, List<Class<? extends TType>> componentTypes, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Barrier", scope.makeOpName("Barrier"));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    Class[] componentTypesArray = new Class[componentTypes.size()];
+    for (int i = 0; i < componentTypesArray.length; ++i) {
+      componentTypesArray[i] = componentTypes.get(i);
+    }
+    opBuilder.setAttr("component_types", componentTypesArray);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.shapes != null) {
+          Shape[] shapesArray = new Shape[opts.shapes.size()];
+          for (int i = 0; i < shapesArray.length; ++i) {
+            shapesArray[i] = opts.shapes.get(i);
+          }
+          opBuilder.setAttr("shapes", shapesArray);
+        }
+        if (opts.capacity != null) {
+          opBuilder.setAttr("capacity", opts.capacity);
+        }
+        if (opts.container != null) {
+          opBuilder.setAttr("container", opts.container);
+        }
+        if (opts.sharedName != null) {
+          opBuilder.setAttr("shared_name", opts.sharedName);
+        }
+      }
+    }
+    return new Barrier(opBuilder.build());
+  }
+  
+  /**
+   * @param shapes The shape of each component in a value. Each shape must be 1 in the
+   * first dimension. The length of this attr must be the same as the length of
+   * component_types.
+   */
+  public static Options shapes(List<Shape> shapes) {
+    return new Options().shapes(shapes);
+  }
+  
+  /**
+   * @param capacity The capacity of the barrier.  The default capacity is MAX_INT32,
+   * which is the largest capacity of the underlying queue.
+   */
+  public static Options capacity(Long capacity) {
+    return new Options().capacity(capacity);
+  }
+  
+  /**
+   * @param container If non-empty, this barrier is placed in the given container.
+   * Otherwise, a default container is used.
+   */
+  public static Options container(String container) {
+    return new Options().container(container);
+  }
+  
+  /**
+   * @param sharedName If non-empty, this barrier will be shared under the given name
+   * across multiple sessions.
+   */
+  public static Options sharedName(String sharedName) {
+    return new Options().sharedName(sharedName);
+  }
+  
+  /**
+   * The handle to the barrier.
+   */
+  public Output<TString> handle() {
+    return handle;
+  }
+  
+  @Override
+  public Output<TString> asOutput() {
+    return handle;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Barrier";
+  
+  private Output<TString> handle;
+  
+  private Barrier(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    handle = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BarrierClose.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BarrierClose.java
new file mode 100644
index 00000000000..514f4f50edf
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BarrierClose.java
@@ -0,0 +1,101 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TString;
+
+/**
+ * Closes the given barrier.
+ * <p>
+ * This operation signals that no more new elements will be inserted in the
+ * given barrier. Subsequent InsertMany that try to introduce a new key will fail.
+ * Subsequent InsertMany operations that just add missing components to already
+ * existing elements will continue to succeed. Subsequent TakeMany operations will
+ * continue to succeed if sufficient completed elements remain in the barrier.
+ * Subsequent TakeMany operations that would block will fail immediately.
+ */
+@Operator
+public final class BarrierClose extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.core.BarrierClose}
+   */
+  public static class Options {
+    
+    /**
+     * @param cancelPendingEnqueues If true, all pending enqueue requests that are
+     * blocked on the barrier's queue will be canceled. InsertMany will fail, even
+     * if no new key is introduced.
+     */
+    public Options cancelPendingEnqueues(Boolean cancelPendingEnqueues) {
+      this.cancelPendingEnqueues = cancelPendingEnqueues;
+      return this;
+    }
+    
+    private Boolean cancelPendingEnqueues;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new BarrierClose operation.
+   * 
+   * @param scope current scope
+   * @param handle The handle to a barrier.
+   * @param options carries optional attributes values
+   * @return a new instance of BarrierClose
+   */
+  @Endpoint(describeByClass = true)
+  public static BarrierClose create(Scope scope, Operand<TString> handle, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("BarrierClose", scope.makeOpName("BarrierClose"));
+    opBuilder.addInput(handle.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.cancelPendingEnqueues != null) {
+          opBuilder.setAttr("cancel_pending_enqueues", opts.cancelPendingEnqueues);
+        }
+      }
+    }
+    return new BarrierClose(opBuilder.build());
+  }
+  
+  /**
+   * @param cancelPendingEnqueues If true, all pending enqueue requests that are
+   * blocked on the barrier's queue will be canceled. InsertMany will fail, even
+   * if no new key is introduced.
+   */
+  public static Options cancelPendingEnqueues(Boolean cancelPendingEnqueues) {
+    return new Options().cancelPendingEnqueues(cancelPendingEnqueues);
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "BarrierClose";
+  
+  private BarrierClose(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BarrierIncompleteSize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BarrierIncompleteSize.java
new file mode 100644
index 00000000000..72dbe1533d6
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BarrierIncompleteSize.java
@@ -0,0 +1,75 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.TString;
+
+/**
+ * Computes the number of incomplete elements in the given barrier.
+ */
+@Operator
+public final class BarrierIncompleteSize extends RawOp implements Operand<TInt32> {
+  
+  /**
+   * Factory method to create a class wrapping a new BarrierIncompleteSize operation.
+   * 
+   * @param scope current scope
+   * @param handle The handle to a barrier.
+   * @return a new instance of BarrierIncompleteSize
+   */
+  @Endpoint(describeByClass = true)
+  public static BarrierIncompleteSize create(Scope scope, Operand<TString> handle) {
+    OperationBuilder opBuilder = scope.env().opBuilder("BarrierIncompleteSize", scope.makeOpName("BarrierIncompleteSize"));
+    opBuilder.addInput(handle.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new BarrierIncompleteSize(opBuilder.build());
+  }
+  
+  /**
+   * The number of incomplete elements (i.e. those with some of their value
+   * components not set) in the barrier.
+   */
+  public Output<TInt32> size() {
+    return size;
+  }
+  
+  @Override
+  public Output<TInt32> asOutput() {
+    return size;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "BarrierIncompleteSize";
+  
+  private Output<TInt32> size;
+  
+  private BarrierIncompleteSize(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    size = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BarrierInsertMany.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BarrierInsertMany.java
new file mode 100644
index 00000000000..b652c11a35c
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BarrierInsertMany.java
@@ -0,0 +1,69 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TString;
+import org.tensorflow.types.family.TType;
+
+/**
+ * For each key, assigns the respective value to the specified component.
+ * <p>
+ * If a key is not found in the barrier, this operation will create a new
+ * incomplete element. If a key is found in the barrier, and the element
+ * already has a value at component_index, this operation will fail with
+ * INVALID_ARGUMENT, and leave the barrier in an undefined state.
+ */
+@Operator
+public final class BarrierInsertMany extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new BarrierInsertMany operation.
+   * 
+   * @param scope current scope
+   * @param handle The handle to a barrier.
+   * @param keys A one-dimensional tensor of keys, with length n.
+   * @param values An any-dimensional tensor of values, which are associated with the
+   * respective keys. The 0th dimension must have length n.
+   * @param componentIndex The component of the barrier elements that is being assigned.
+   * @return a new instance of BarrierInsertMany
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> BarrierInsertMany create(Scope scope, Operand<TString> handle, Operand<TString> keys, Operand<T> values, Long componentIndex) {
+    OperationBuilder opBuilder = scope.env().opBuilder("BarrierInsertMany", scope.makeOpName("BarrierInsertMany"));
+    opBuilder.addInput(handle.asOutput());
+    opBuilder.addInput(keys.asOutput());
+    opBuilder.addInput(values.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("component_index", componentIndex);
+    return new BarrierInsertMany(opBuilder.build());
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "BarrierInsertMany";
+  
+  private BarrierInsertMany(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BarrierReadySize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BarrierReadySize.java
new file mode 100644
index 00000000000..1cb29f97e1d
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BarrierReadySize.java
@@ -0,0 +1,75 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.TString;
+
+/**
+ * Computes the number of complete elements in the given barrier.
+ */
+@Operator
+public final class BarrierReadySize extends RawOp implements Operand<TInt32> {
+  
+  /**
+   * Factory method to create a class wrapping a new BarrierReadySize operation.
+   * 
+   * @param scope current scope
+   * @param handle The handle to a barrier.
+   * @return a new instance of BarrierReadySize
+   */
+  @Endpoint(describeByClass = true)
+  public static BarrierReadySize create(Scope scope, Operand<TString> handle) {
+    OperationBuilder opBuilder = scope.env().opBuilder("BarrierReadySize", scope.makeOpName("BarrierReadySize"));
+    opBuilder.addInput(handle.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new BarrierReadySize(opBuilder.build());
+  }
+  
+  /**
+   * The number of complete elements (i.e. those with all of their value
+   * components set) in the barrier.
+   */
+  public Output<TInt32> size() {
+    return size;
+  }
+  
+  @Override
+  public Output<TInt32> asOutput() {
+    return size;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "BarrierReadySize";
+  
+  private Output<TInt32> size;
+  
+  private BarrierReadySize(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    size = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BarrierTakeMany.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BarrierTakeMany.java
new file mode 100644
index 00000000000..9032635571f
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BarrierTakeMany.java
@@ -0,0 +1,191 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import java.util.Arrays;
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.TString;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Takes the given number of completed elements from a barrier.
+ * <p>
+ * This operation concatenates completed-element component tensors along
+ * the 0th dimension to make a single component tensor.
+ * <p>
+ * Elements come out of the barrier when they are complete, and in the order
+ * in which they were placed into the barrier.  The indices output provides
+ * information about the batch in which each element was originally inserted
+ * into the barrier.
+ */
+@Operator
+public final class BarrierTakeMany extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.core.BarrierTakeMany}
+   */
+  public static class Options {
+    
+    /**
+     * @param allowSmallBatch Allow to return less than num_elements items if barrier is
+     * already closed.
+     */
+    public Options allowSmallBatch(Boolean allowSmallBatch) {
+      this.allowSmallBatch = allowSmallBatch;
+      return this;
+    }
+    
+    /**
+     * @param waitForIncomplete 
+     */
+    public Options waitForIncomplete(Boolean waitForIncomplete) {
+      this.waitForIncomplete = waitForIncomplete;
+      return this;
+    }
+    
+    /**
+     * @param timeoutMs If the queue is empty, this operation will block for up to
+     * timeout_ms milliseconds.
+     * Note: This option is not supported yet.
+     */
+    public Options timeoutMs(Long timeoutMs) {
+      this.timeoutMs = timeoutMs;
+      return this;
+    }
+    
+    private Boolean allowSmallBatch;
+    private Boolean waitForIncomplete;
+    private Long timeoutMs;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new BarrierTakeMany operation.
+   * 
+   * @param scope current scope
+   * @param handle The handle to a barrier.
+   * @param numElements A single-element tensor containing the number of elements to
+   * take.
+   * @param componentTypes The type of each component in a value.
+   * @param options carries optional attributes values
+   * @return a new instance of BarrierTakeMany
+   */
+  @Endpoint(describeByClass = true)
+  public static BarrierTakeMany create(Scope scope, Operand<TString> handle, Operand<TInt32> numElements, List<Class<? extends TType>> componentTypes, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("BarrierTakeMany", scope.makeOpName("BarrierTakeMany"));
+    opBuilder.addInput(handle.asOutput());
+    opBuilder.addInput(numElements.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    Class[] componentTypesArray = new Class[componentTypes.size()];
+    for (int i = 0; i < componentTypesArray.length; ++i) {
+      componentTypesArray[i] = componentTypes.get(i);
+    }
+    opBuilder.setAttr("component_types", componentTypesArray);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.allowSmallBatch != null) {
+          opBuilder.setAttr("allow_small_batch", opts.allowSmallBatch);
+        }
+        if (opts.waitForIncomplete != null) {
+          opBuilder.setAttr("wait_for_incomplete", opts.waitForIncomplete);
+        }
+        if (opts.timeoutMs != null) {
+          opBuilder.setAttr("timeout_ms", opts.timeoutMs);
+        }
+      }
+    }
+    return new BarrierTakeMany(opBuilder.build());
+  }
+  
+  /**
+   * @param allowSmallBatch Allow to return less than num_elements items if barrier is
+   * already closed.
+   */
+  public static Options allowSmallBatch(Boolean allowSmallBatch) {
+    return new Options().allowSmallBatch(allowSmallBatch);
+  }
+  
+  /**
+   * @param waitForIncomplete 
+   */
+  public static Options waitForIncomplete(Boolean waitForIncomplete) {
+    return new Options().waitForIncomplete(waitForIncomplete);
+  }
+  
+  /**
+   * @param timeoutMs If the queue is empty, this operation will block for up to
+   * timeout_ms milliseconds.
+   * Note: This option is not supported yet.
+   */
+  public static Options timeoutMs(Long timeoutMs) {
+    return new Options().timeoutMs(timeoutMs);
+  }
+  
+  /**
+   * A one-dimensional tensor of indices, with length num_elems.
+   * These indices refer to the batch in which the values were placed into the
+   * barrier (starting with MIN_LONG and increasing with each BarrierInsertMany).
+   */
+  public Output<TInt64> indices() {
+    return indices;
+  }
+  
+  /**
+   * A one-dimensional tensor of keys, with length num_elements.
+   */
+  public Output<TString> keys() {
+    return keys;
+  }
+  
+  /**
+   * One any-dimensional tensor per component in a barrier element. All
+   * values have length num_elements in the 0th dimension.
+   */
+  public List<Output<?>> values() {
+    return values;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "BarrierTakeMany";
+  
+  private Output<TInt64> indices;
+  private Output<TString> keys;
+  private List<Output<?>> values;
+  
+  private BarrierTakeMany(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    indices = operation.output(outputIdx++);
+    keys = operation.output(outputIdx++);
+    int valuesLength = operation.outputListLength("values");
+    values = Arrays.asList(operation.outputList(outputIdx, valuesLength));
+    outputIdx += valuesLength;
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Batch.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Batch.java
new file mode 100644
index 00000000000..21d7cf0a6c0
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Batch.java
@@ -0,0 +1,247 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import java.util.Arrays;
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.Operands;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt64;
+
+/**
+ * Batches all input tensors nondeterministically.
+ * <p>
+ * When many instances of this Op are being run concurrently with the same
+ * container/shared_name in the same device, some will output zero-shaped Tensors
+ * and others will output Tensors of size up to max_batch_size.
+ * <p>
+ * All Tensors in in_tensors are batched together (so, for example, labels and
+ * features should be batched with a single instance of this operation.
+ * <p>
+ * Each invocation of batch emits an `id` scalar which will be used to identify
+ * this particular invocation when doing unbatch or its gradient.
+ * <p>
+ * Each op which emits a non-empty batch will also emit a non-empty batch_index
+ * Tensor, which, is a [K, 3] matrix where each row contains the invocation's id,
+ * start, and length of elements of each set of Tensors present in batched_tensors.
+ * <p>
+ * Batched tensors are concatenated along the first dimension, and all tensors in
+ * in_tensors must have the first dimension of the same size.
+ * <p>
+ * in_tensors: The tensors to be batched.
+ * num_batch_threads: Number of scheduling threads for processing batches of work.
+ *  Determines the number of batches processed in parallel.
+ * max_batch_size: Batch sizes will never be bigger than this.
+ * batch_timeout_micros: Maximum number of microseconds to wait before outputting
+ *  an incomplete batch.
+ * allowed_batch_sizes: Optional list of allowed batch sizes. If left empty, does
+ *  nothing. Otherwise, supplies a list of batch sizes, causing the op to pad
+ *  batches up to one of those sizes. The entries must increase monotonically, and
+ *  the final entry must equal max_batch_size.
+ * grad_timeout_micros: The timeout to use for the gradient. See Unbatch.
+ * batched_tensors: Either empty tensors or a batch of concatenated Tensors.
+ * batch_index: If out_tensors is non-empty, has information to invert it.
+ * container: Controls the scope of sharing of this batch.
+ * id: always contains a scalar with a unique ID for this invocation of Batch.
+ * shared_name: Concurrently running instances of batch in the same device with the
+ *  same container and shared_name will batch their elements together. If left
+ *  empty, the op name will be used as the shared name.
+ * T: the types of tensors to be batched.
+ */
+@Operator
+public final class Batch extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.core.Batch}
+   */
+  public static class Options {
+    
+    /**
+     * @param maxEnqueuedBatches 
+     */
+    public Options maxEnqueuedBatches(Long maxEnqueuedBatches) {
+      this.maxEnqueuedBatches = maxEnqueuedBatches;
+      return this;
+    }
+    
+    /**
+     * @param allowedBatchSizes 
+     */
+    public Options allowedBatchSizes(List<Long> allowedBatchSizes) {
+      this.allowedBatchSizes = allowedBatchSizes;
+      return this;
+    }
+    
+    /**
+     * @param container 
+     */
+    public Options container(String container) {
+      this.container = container;
+      return this;
+    }
+    
+    /**
+     * @param sharedName 
+     */
+    public Options sharedName(String sharedName) {
+      this.sharedName = sharedName;
+      return this;
+    }
+    
+    /**
+     * @param batchingQueue 
+     */
+    public Options batchingQueue(String batchingQueue) {
+      this.batchingQueue = batchingQueue;
+      return this;
+    }
+    
+    private Long maxEnqueuedBatches;
+    private List<Long> allowedBatchSizes;
+    private String container;
+    private String sharedName;
+    private String batchingQueue;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new Batch operation.
+   * 
+   * @param scope current scope
+   * @param inTensors 
+   * @param numBatchThreads 
+   * @param maxBatchSize 
+   * @param batchTimeoutMicros 
+   * @param gradTimeoutMicros 
+   * @param options carries optional attributes values
+   * @return a new instance of Batch
+   */
+  @Endpoint(describeByClass = true)
+  public static Batch create(Scope scope, Iterable<Operand<?>> inTensors, Long numBatchThreads, Long maxBatchSize, Long batchTimeoutMicros, Long gradTimeoutMicros, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Batch", scope.makeOpName("Batch"));
+    opBuilder.addInputList(Operands.asOutputs(inTensors));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("num_batch_threads", numBatchThreads);
+    opBuilder.setAttr("max_batch_size", maxBatchSize);
+    opBuilder.setAttr("batch_timeout_micros", batchTimeoutMicros);
+    opBuilder.setAttr("grad_timeout_micros", gradTimeoutMicros);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.maxEnqueuedBatches != null) {
+          opBuilder.setAttr("max_enqueued_batches", opts.maxEnqueuedBatches);
+        }
+        if (opts.allowedBatchSizes != null) {
+          long[] allowedBatchSizesArray = new long[opts.allowedBatchSizes.size()];
+          for (int i = 0; i < allowedBatchSizesArray.length; ++i) {
+            allowedBatchSizesArray[i] = opts.allowedBatchSizes.get(i);
+          }
+          opBuilder.setAttr("allowed_batch_sizes", allowedBatchSizesArray);
+        }
+        if (opts.container != null) {
+          opBuilder.setAttr("container", opts.container);
+        }
+        if (opts.sharedName != null) {
+          opBuilder.setAttr("shared_name", opts.sharedName);
+        }
+        if (opts.batchingQueue != null) {
+          opBuilder.setAttr("batching_queue", opts.batchingQueue);
+        }
+      }
+    }
+    return new Batch(opBuilder.build());
+  }
+  
+  /**
+   * @param maxEnqueuedBatches 
+   */
+  public static Options maxEnqueuedBatches(Long maxEnqueuedBatches) {
+    return new Options().maxEnqueuedBatches(maxEnqueuedBatches);
+  }
+  
+  /**
+   * @param allowedBatchSizes 
+   */
+  public static Options allowedBatchSizes(List<Long> allowedBatchSizes) {
+    return new Options().allowedBatchSizes(allowedBatchSizes);
+  }
+  
+  /**
+   * @param container 
+   */
+  public static Options container(String container) {
+    return new Options().container(container);
+  }
+  
+  /**
+   * @param sharedName 
+   */
+  public static Options sharedName(String sharedName) {
+    return new Options().sharedName(sharedName);
+  }
+  
+  /**
+   * @param batchingQueue 
+   */
+  public static Options batchingQueue(String batchingQueue) {
+    return new Options().batchingQueue(batchingQueue);
+  }
+  
+  /**
+   */
+  public List<Output<?>> batchedTensors() {
+    return batchedTensors;
+  }
+  
+  /**
+   */
+  public Output<TInt64> batchIndex() {
+    return batchIndex;
+  }
+  
+  /**
+   */
+  public Output<TInt64> id() {
+    return id;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Batch";
+  
+  private List<Output<?>> batchedTensors;
+  private Output<TInt64> batchIndex;
+  private Output<TInt64> id;
+  
+  private Batch(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    int batchedTensorsLength = operation.outputListLength("batched_tensors");
+    batchedTensors = Arrays.asList(operation.outputList(outputIdx, batchedTensorsLength));
+    outputIdx += batchedTensorsLength;
+    batchIndex = operation.output(outputIdx++);
+    id = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BatchToSpace.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BatchToSpace.java
new file mode 100644
index 00000000000..b20c60ee4ac
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BatchToSpace.java
@@ -0,0 +1,147 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * BatchToSpace for 4-D tensors of type T.
+ * <p>
+ * This is a legacy version of the more general BatchToSpaceND.
+ * <p>
+ * Rearranges (permutes) data from batch into blocks of spatial data, followed by
+ * cropping. This is the reverse transformation of SpaceToBatch. More specifically,
+ * this op outputs a copy of the input tensor where values from the `batch`
+ * dimension are moved in spatial blocks to the `height` and `width` dimensions,
+ * followed by cropping along the `height` and `width` dimensions.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator
+public final class BatchToSpace<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new BatchToSpace operation.
+   * 
+   * @param scope current scope
+   * @param input 4-D tensor with shape
+   * `[batch<i>block_size</i>block_size, height_pad/block_size, width_pad/block_size,
+   *   depth]`. Note that the batch size of the input tensor must be divisible by
+   * `block_size * block_size`.
+   * @param crops 2-D tensor of non-negative integers with shape `[2, 2]`. It specifies
+   * how many elements to crop from the intermediate result across the spatial
+   * dimensions as follows:
+   * <p>
+   *     crops = [[crop_top, crop_bottom], [crop_left, crop_right]]
+   * @param blockSize 
+   * @return a new instance of BatchToSpace
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType, U extends TNumber> BatchToSpace<T> create(Scope scope, Operand<T> input, Operand<U> crops, Long blockSize) {
+    OperationBuilder opBuilder = scope.env().opBuilder("BatchToSpace", scope.makeOpName("BatchToSpace"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(crops.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("block_size", blockSize);
+    return new BatchToSpace<T>(opBuilder.build());
+  }
+  
+  /**
+   * 4-D with shape `[batch, height, width, depth]`, where:
+   * <p>
+   *       height = height_pad - crop_top - crop_bottom
+   *       width = width_pad - crop_left - crop_right
+   * <p>
+   * The attr `block_size` must be greater than one. It indicates the block size.
+   * <p>
+   * Some examples:
+   * <p>
+   * (1) For the following input of shape `[4, 1, 1, 1]` and block_size of 2:
+   * <pre>{@code
+   * [[[[1]]], [[[2]]], [[[3]]], [[[4]]]]
+   * }</pre>
+   * The output tensor has shape `[1, 2, 2, 1]` and value:
+   * <pre>{@code
+   * x = [[[[1], [2]], [[3], [4]]]]
+   * }</pre>
+   * (2) For the following input of shape `[4, 1, 1, 3]` and block_size of 2:
+   * <pre>{@code
+   * [[[[1, 2, 3]]], [[[4, 5, 6]]], [[[7, 8, 9]]], [[[10, 11, 12]]]]
+   * }</pre>
+   * The output tensor has shape `[1, 2, 2, 3]` and value:
+   * <pre>{@code
+   * x = [[[[1, 2, 3], [4, 5, 6]],
+   *       [[7, 8, 9], [10, 11, 12]]]]
+   * }</pre>
+   * (3) For the following input of shape `[4, 2, 2, 1]` and block_size of 2:
+   * <pre>{@code
+   * x = [[[[1], [3]], [[9], [11]]],
+   *      [[[2], [4]], [[10], [12]]],
+   *      [[[5], [7]], [[13], [15]]],
+   *      [[[6], [8]], [[14], [16]]]]
+   * }</pre>
+   * The output tensor has shape `[1, 4, 4, 1]` and value:
+   * <pre>{@code
+   * x = [[[[1],   [2],  [3],  [4]],
+   *      [[5],   [6],  [7],  [8]],
+   *      [[9],  [10], [11],  [12]],
+   *      [[13], [14], [15],  [16]]]]
+   * }</pre>
+   * (4) For the following input of shape `[8, 1, 2, 1]` and block_size of 2:
+   * <pre>{@code
+   * x = [[[[1], [3]]], [[[9], [11]]], [[[2], [4]]], [[[10], [12]]],
+   *      [[[5], [7]]], [[[13], [15]]], [[[6], [8]]], [[[14], [16]]]]
+   * }</pre>
+   * The output tensor has shape `[2, 2, 4, 1]` and value:
+   * <pre>{@code
+   * x = [[[[1], [3]], [[5], [7]]],
+   *      [[[2], [4]], [[10], [12]]],
+   *      [[[5], [7]], [[13], [15]]],
+   *      [[[6], [8]], [[14], [16]]]]
+   * }</pre>
+   * 
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "BatchToSpace";
+  
+  private Output<T> output;
+  
+  private BatchToSpace(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BatchToSpaceNd.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BatchToSpaceNd.java
new file mode 100644
index 00000000000..ef6f31d0f89
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BatchToSpaceNd.java
@@ -0,0 +1,181 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * BatchToSpace for N-D tensors of type T.
+ * <p>
+ * This operation reshapes the "batch" dimension 0 into `M + 1` dimensions of shape
+ * `block_shape + [batch]`, interleaves these blocks back into the grid defined by
+ * the spatial dimensions `[1, ..., M]`, to obtain a result with the same rank as
+ * the input.  The spatial dimensions of this intermediate result are then
+ * optionally cropped according to `crops` to produce the output.  This is the
+ * reverse of SpaceToBatch.  See below for a precise description.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator
+public final class BatchToSpaceNd<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new BatchToSpaceNd operation.
+   * 
+   * @param scope current scope
+   * @param input N-D with shape `input_shape = [batch] + spatial_shape + remaining_shape`,
+   * where spatial_shape has M dimensions.
+   * @param blockShape 1-D with shape `[M]`, all values must be >= 1.
+   * @param crops 2-D with shape `[M, 2]`, all values must be >= 0.
+   *   `crops[i] = [crop_start, crop_end]` specifies the amount to crop from input
+   *   dimension `i + 1`, which corresponds to spatial dimension `i`.  It is
+   *   required that
+   *   `crop_start[i] + crop_end[i] <= block_shape[i] * input_shape[i + 1]`.
+   * <p>
+   * This operation is equivalent to the following steps:
+   * <p>
+   * 1. Reshape `input` to `reshaped` of shape:
+   *      [block_shape[0], ..., block_shape[M-1],
+   *       batch / prod(block_shape),
+   *       input_shape[1], ..., input_shape[N-1]]
+   * <p>
+   * 2. Permute dimensions of `reshaped` to produce `permuted` of shape
+   *      [batch / prod(block_shape),
+   * <p>
+   *       input_shape[1], block_shape[0],
+   *       ...,
+   *       input_shape[M], block_shape[M-1],
+   * <p>
+   *       input_shape[M+1], ..., input_shape[N-1]]
+   * <p>
+   * 3. Reshape `permuted` to produce `reshaped_permuted` of shape
+   *      [batch / prod(block_shape),
+   * <p>
+   *       input_shape[1] * block_shape[0],
+   *       ...,
+   *       input_shape[M] * block_shape[M-1],
+   * <p>
+   *       input_shape[M+1],
+   *       ...,
+   *       input_shape[N-1]]
+   * <p>
+   * 4. Crop the start and end of dimensions `[1, ..., M]` of
+   *    `reshaped_permuted` according to `crops` to produce the output of shape:
+   *      [batch / prod(block_shape),
+   * <p>
+   *       input_shape[1] * block_shape[0] - crops[0,0] - crops[0,1],
+   *       ...,
+   *       input_shape[M] * block_shape[M-1] - crops[M-1,0] - crops[M-1,1],
+   * <p>
+   *       input_shape[M+1], ..., input_shape[N-1]]
+   * <p>
+   * Some examples:
+   * <p>
+   * (1) For the following input of shape `[4, 1, 1, 1]`, `block_shape = [2, 2]`, and
+   *     `crops = [[0, 0], [0, 0]]`:
+   * <pre>{@code
+   * [[[[1]]], [[[2]]], [[[3]]], [[[4]]]]
+   * }</pre>
+   * The output tensor has shape `[1, 2, 2, 1]` and value:
+   * <pre>{@code
+   * x = [[[[1], [2]], [[3], [4]]]]
+   * }</pre>
+   * (2) For the following input of shape `[4, 1, 1, 3]`, `block_shape = [2, 2]`, and
+   *     `crops = [[0, 0], [0, 0]]`:
+   * <pre>{@code
+   * [[[[1, 2, 3]]], [[[4, 5, 6]]], [[[7, 8, 9]]], [[[10, 11, 12]]]]
+   * }</pre>
+   * The output tensor has shape `[1, 2, 2, 3]` and value:
+   * <pre>{@code
+   * x = [[[[1, 2, 3], [4, 5, 6]],
+   *       [[7, 8, 9], [10, 11, 12]]]]
+   * }</pre>
+   * (3) For the following input of shape `[4, 2, 2, 1]`, `block_shape = [2, 2]`, and
+   *     `crops = [[0, 0], [0, 0]]`:
+   * <pre>{@code
+   * x = [[[[1], [3]], [[9], [11]]],
+   *      [[[2], [4]], [[10], [12]]],
+   *      [[[5], [7]], [[13], [15]]],
+   *      [[[6], [8]], [[14], [16]]]]
+   * }</pre>
+   * The output tensor has shape `[1, 4, 4, 1]` and value:
+   * <pre>{@code
+   * x = [[[[1],   [2],  [3],  [4]],
+   *      [[5],   [6],  [7],  [8]],
+   *      [[9],  [10], [11],  [12]],
+   *      [[13], [14], [15],  [16]]]]
+   * }</pre>
+   * (4) For the following input of shape `[8, 1, 3, 1]`, `block_shape = [2, 2]`, and
+   *     `crops = [[0, 0], [2, 0]]`:
+   * <pre>{@code
+   * x = [[[[0], [1], [3]]], [[[0], [9], [11]]],
+   *      [[[0], [2], [4]]], [[[0], [10], [12]]],
+   *      [[[0], [5], [7]]], [[[0], [13], [15]]],
+   *      [[[0], [6], [8]]], [[[0], [14], [16]]]]
+   * }</pre>
+   * The output tensor has shape `[2, 2, 4, 1]` and value:
+   * <pre>{@code
+   * x = [[[[1],   [2],  [3],  [4]],
+   *       [[5],   [6],  [7],  [8]]],
+   *      [[[9],  [10], [11],  [12]],
+   *       [[13], [14], [15],  [16]]]]
+   * }</pre>
+   * 
+   * @return a new instance of BatchToSpaceNd
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType, U extends TNumber, V extends TNumber> BatchToSpaceNd<T> create(Scope scope, Operand<T> input, Operand<U> blockShape, Operand<V> crops) {
+    OperationBuilder opBuilder = scope.env().opBuilder("BatchToSpaceND", scope.makeOpName("BatchToSpaceNd"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(blockShape.asOutput());
+    opBuilder.addInput(crops.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new BatchToSpaceNd<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "BatchToSpaceND";
+  
+  private Output<T> output;
+  
+  private BatchToSpaceNd(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Bitcast.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Bitcast.java
new file mode 100644
index 00000000000..65d39833a78
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Bitcast.java
@@ -0,0 +1,127 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Bitcasts a tensor from one type to another without copying data.
+ * <p>
+ * Given a tensor `input`, this operation returns a tensor that has the same buffer
+ * data as `input` with datatype `type`.
+ * <p>
+ * If the input datatype `T` is larger than the output datatype `type` then the
+ * shape changes from [...] to [..., sizeof(`T`)/sizeof(`type`)].
+ * <p>
+ * If `T` is smaller than `type`, the operator requires that the rightmost
+ * dimension be equal to sizeof(`type`)/sizeof(`T`). The shape then goes from
+ * [..., sizeof(`type`)/sizeof(`T`)] to [...].
+ * <p>
+ * tf.bitcast() and tf.cast() work differently when real dtype is casted as a complex dtype
+ * (e.g. tf.complex64 or tf.complex128) as tf.cast() make imaginary part 0 while tf.bitcast()
+ * gives module error.
+ * For example,
+ * <p>
+ * Example 1:
+ * <p>
+ * >>> a = [1., 2., 3.]
+ * >>> equality_bitcast = tf.bitcast(a, tf.complex128)
+ * Traceback (most recent call last):
+ * ...
+ * InvalidArgumentError: Cannot bitcast from 1 to 18 [Op:Bitcast]
+ * >>> equality_cast = tf.cast(a, tf.complex128)
+ * >>> print(equality_cast)
+ * tf.Tensor([1.+0.j 2.+0.j 3.+0.j], shape=(3,), dtype=complex128)
+ * <p>
+ * Example 2:
+ * <p>
+ * >>> tf.bitcast(tf.constant(0xffffffff, dtype=tf.uint32), tf.uint8)
+ * <tf.Tensor: shape=(4,), dtype=uint8, numpy=array([255, 255, 255, 255], dtype=uint8)>
+ * <p>
+ * Example 3:
+ * <p>
+ * >>> x = [1., 2., 3.]
+ * >>> y = [0., 2., 3.]
+ * >>> equality= tf.equal(x,y)
+ * >>> equality_cast = tf.cast(equality,tf.float32)
+ * >>> equality_bitcast = tf.bitcast(equality_cast,tf.uint8)
+ * >>> print(equality)
+ * tf.Tensor([False True True], shape=(3,), dtype=bool)
+ * >>> print(equality_cast)
+ * tf.Tensor([0. 1. 1.], shape=(3,), dtype=float32)
+ * >>> print(equality_bitcast)
+ * tf.Tensor(
+ *     [[  0   0   0   0]
+ *      [  0   0 128  63]
+ *      [  0   0 128  63]], shape=(3, 4), dtype=uint8)
+ * <p>
+ * <i>NOTE</i>: Bitcast is implemented as a low-level cast, so machines with different
+ * endian orderings will give different results.
+ * 
+ * @param <U> data type for {@code output()} output
+ */
+@Operator
+public final class Bitcast<U extends TType> extends RawOp implements Operand<U> {
+  
+  /**
+   * Factory method to create a class wrapping a new Bitcast operation.
+   * 
+   * @param scope current scope
+   * @param input 
+   * @param type 
+   * @return a new instance of Bitcast
+   */
+  @Endpoint(describeByClass = true)
+  public static <U extends TType, T extends TType> Bitcast<U> create(Scope scope, Operand<T> input, Class<U> type) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Bitcast", scope.makeOpName("Bitcast"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("type", type);
+    return new Bitcast<U>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<U> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<U> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Bitcast";
+  
+  private Output<U> output;
+  
+  private Bitcast(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BroadcastDynamicShape.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BroadcastDynamicShape.java
new file mode 100644
index 00000000000..df29691cac8
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BroadcastDynamicShape.java
@@ -0,0 +1,79 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Return the shape of s0 op s1 with broadcast.
+ * <p>
+ * Given `s0` and `s1`, tensors that represent shapes, compute `r0`, the
+ * broadcasted shape. `s0`, `s1` and `r0` are all integer vectors.
+ * 
+ * @param <T> data type for {@code r0()} output
+ */
+@Operator
+public final class BroadcastDynamicShape<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new BroadcastDynamicShape operation.
+   * 
+   * @param scope current scope
+   * @param s0 
+   * @param s1 
+   * @return a new instance of BroadcastDynamicShape
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> BroadcastDynamicShape<T> create(Scope scope, Operand<T> s0, Operand<T> s1) {
+    OperationBuilder opBuilder = scope.env().opBuilder("BroadcastArgs", scope.makeOpName("BroadcastDynamicShape"));
+    opBuilder.addInput(s0.asOutput());
+    opBuilder.addInput(s1.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new BroadcastDynamicShape<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> r0() {
+    return r0;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return r0;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "BroadcastArgs";
+  
+  private Output<T> r0;
+  
+  private BroadcastDynamicShape(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    r0 = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BroadcastGradientArgs.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BroadcastGradientArgs.java
new file mode 100644
index 00000000000..dffa39e4bf1
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BroadcastGradientArgs.java
@@ -0,0 +1,80 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Return the reduction indices for computing gradients of s0 op s1 with broadcast.
+ * <p>
+ * This is typically used by gradient computations for a broadcasting operation.
+ * 
+ * @param <T> data type for {@code r0()} output
+ */
+public final class BroadcastGradientArgs<T extends TNumber> extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new BroadcastGradientArgs operation.
+   * 
+   * @param scope current scope
+   * @param s0 
+   * @param s1 
+   * @return a new instance of BroadcastGradientArgs
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> BroadcastGradientArgs<T> create(Scope scope, Operand<T> s0, Operand<T> s1) {
+    OperationBuilder opBuilder = scope.env().opBuilder("BroadcastGradientArgs", scope.makeOpName("BroadcastGradientArgs"));
+    opBuilder.addInput(s0.asOutput());
+    opBuilder.addInput(s1.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new BroadcastGradientArgs<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> r0() {
+    return r0;
+  }
+  
+  /**
+   */
+  public Output<T> r1() {
+    return r1;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "BroadcastGradientArgs";
+  
+  private Output<T> r0;
+  private Output<T> r1;
+  
+  private BroadcastGradientArgs(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    r0 = operation.output(outputIdx++);
+    r1 = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BroadcastTo.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BroadcastTo.java
new file mode 100644
index 00000000000..e5ad5469501
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/BroadcastTo.java
@@ -0,0 +1,106 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Broadcast an array for a compatible shape.
+ * <p>
+ * Broadcasting is the process of making arrays to have compatible shapes
+ * for arithmetic operations. Two shapes are compatible if for each
+ * dimension pair they are either equal or one of them is one. When trying
+ * to broadcast a Tensor to a shape, it starts with the trailing dimensions,
+ * and works its way forward.
+ * <p>
+ * For example,
+ * <p>
+ * >>> x = tf.constant([1, 2, 3])
+ * >>> y = tf.broadcast_to(x, [3, 3])
+ * >>> print(y)
+ * tf.Tensor(
+ *     [[1 2 3]
+ *      [1 2 3]
+ *      [1 2 3]], shape=(3, 3), dtype=int32)
+ * <p>
+ * In the above example, the input Tensor with the shape of `[1, 3]`
+ * is broadcasted to output Tensor with shape of `[3, 3]`.
+ * <p>
+ * When doing broadcasted operations such as multiplying a tensor
+ * by a scalar, broadcasting (usually) confers some time or space
+ * benefit, as the broadcasted tensor is never materialized.
+ * <p>
+ * However, `broadcast_to` does not carry with it any such benefits.
+ * The newly-created tensor takes the full memory of the broadcasted
+ * shape. (In a graph context, `broadcast_to` might be fused to
+ * subsequent operation and then be optimized away, however.)
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator
+public final class BroadcastTo<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new BroadcastTo operation.
+   * 
+   * @param scope current scope
+   * @param input A Tensor to broadcast.
+   * @param shape An 1-D `int` Tensor. The shape of the desired output.
+   * @return a new instance of BroadcastTo
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType, U extends TNumber> BroadcastTo<T> create(Scope scope, Operand<T> input, Operand<U> shape) {
+    OperationBuilder opBuilder = scope.env().opBuilder("BroadcastTo", scope.makeOpName("BroadcastTo"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(shape.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new BroadcastTo<T>(opBuilder.build());
+  }
+  
+  /**
+   * A Tensor.
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "BroadcastTo";
+  
+  private Output<T> output;
+  
+  private BroadcastTo(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Bucketize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Bucketize.java
new file mode 100644
index 00000000000..05b43dc7679
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Bucketize.java
@@ -0,0 +1,96 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Bucketizes 'input' based on 'boundaries'.
+ * <p>
+ * For example, if the inputs are
+ *     boundaries = [0, 10, 100]
+ *     input = [[-5, 10000]
+ *              [150,   10]
+ *              [5,    100]]
+ * <p>
+ * then the output will be
+ *     output = [[0, 3]
+ *               [3, 2]
+ *               [1, 3]]
+ */
+@Operator
+public final class Bucketize extends RawOp implements Operand<TInt32> {
+  
+  /**
+   * Factory method to create a class wrapping a new Bucketize operation.
+   * 
+   * @param scope current scope
+   * @param input Any shape of Tensor contains with int or float type.
+   * @param boundaries A sorted list of floats gives the boundary of the buckets.
+   * @return a new instance of Bucketize
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> Bucketize create(Scope scope, Operand<T> input, List<Float> boundaries) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Bucketize", scope.makeOpName("Bucketize"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    float[] boundariesArray = new float[boundaries.size()];
+    for (int i = 0; i < boundariesArray.length; ++i) {
+      boundariesArray[i] = boundaries.get(i);
+    }
+    opBuilder.setAttr("boundaries", boundariesArray);
+    return new Bucketize(opBuilder.build());
+  }
+  
+  /**
+   * Same shape with 'input', each value of input replaced with bucket index.
+   * <p>
+   * @compatibility(numpy)
+   * Equivalent to np.digitize.
+   * @end_compatibility
+   */
+  public Output<TInt32> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<TInt32> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Bucketize";
+  
+  private Output<TInt32> output;
+  
+  private Bucketize(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ClipByValue.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ClipByValue.java
new file mode 100644
index 00000000000..8a5b99d9389
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ClipByValue.java
@@ -0,0 +1,86 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Clips tensor values to a specified min and max.
+ * <p>
+ * Given a tensor `t`, this operation returns a tensor of the same type and
+ * shape as `t` with its values clipped to `clip_value_min` and `clip_value_max`.
+ * Any values less than `clip_value_min` are set to `clip_value_min`. Any values
+ * greater than `clip_value_max` are set to `clip_value_max`.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator
+public final class ClipByValue<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new ClipByValue operation.
+   * 
+   * @param scope current scope
+   * @param t A `Tensor`.
+   * @param clipValueMin A 0-D (scalar) `Tensor`, or a `Tensor` with the same shape
+   * as `t`. The minimum value to clip by.
+   * @param clipValueMax A 0-D (scalar) `Tensor`, or a `Tensor` with the same shape
+   * as `t`. The maximum value to clip by.
+   * @return a new instance of ClipByValue
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> ClipByValue<T> create(Scope scope, Operand<T> t, Operand<T> clipValueMin, Operand<T> clipValueMax) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ClipByValue", scope.makeOpName("ClipByValue"));
+    opBuilder.addInput(t.asOutput());
+    opBuilder.addInput(clipValueMin.asOutput());
+    opBuilder.addInput(clipValueMax.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new ClipByValue<T>(opBuilder.build());
+  }
+  
+  /**
+   * A clipped `Tensor` with the same shape as input 't'.
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ClipByValue";
+  
+  private Output<T> output;
+  
+  private ClipByValue(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/CollectiveGather.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/CollectiveGather.java
new file mode 100644
index 00000000000..d1f4e84f146
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/CollectiveGather.java
@@ -0,0 +1,135 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Mutually accumulates multiple tensors of identical type and shape.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+public final class CollectiveGather<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.core.CollectiveGather}
+   */
+  public static class Options {
+    
+    /**
+     * @param communicationHint 
+     */
+    public Options communicationHint(String communicationHint) {
+      this.communicationHint = communicationHint;
+      return this;
+    }
+    
+    /**
+     * @param timeoutSeconds 
+     */
+    public Options timeoutSeconds(Float timeoutSeconds) {
+      this.timeoutSeconds = timeoutSeconds;
+      return this;
+    }
+    
+    private String communicationHint;
+    private Float timeoutSeconds;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new CollectiveGather operation.
+   * 
+   * @param scope current scope
+   * @param input 
+   * @param groupSize 
+   * @param groupKey 
+   * @param instanceKey 
+   * @param shape 
+   * @param options carries optional attributes values
+   * @return a new instance of CollectiveGather
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> CollectiveGather<T> create(Scope scope, Operand<T> input, Long groupSize, Long groupKey, Long instanceKey, Shape shape, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("CollectiveGather", scope.makeOpName("CollectiveGather"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("group_size", groupSize);
+    opBuilder.setAttr("group_key", groupKey);
+    opBuilder.setAttr("instance_key", instanceKey);
+    opBuilder.setAttr("shape", shape);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.communicationHint != null) {
+          opBuilder.setAttr("communication_hint", opts.communicationHint);
+        }
+        if (opts.timeoutSeconds != null) {
+          opBuilder.setAttr("timeout_seconds", opts.timeoutSeconds);
+        }
+      }
+    }
+    return new CollectiveGather<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param communicationHint 
+   */
+  public static Options communicationHint(String communicationHint) {
+    return new Options().communicationHint(communicationHint);
+  }
+  
+  /**
+   * @param timeoutSeconds 
+   */
+  public static Options timeoutSeconds(Float timeoutSeconds) {
+    return new Options().timeoutSeconds(timeoutSeconds);
+  }
+  
+  /**
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "CollectiveGather";
+  
+  private Output<T> output;
+  
+  private CollectiveGather(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Concat.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Concat.java
new file mode 100644
index 00000000000..9782b3d6ec9
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Concat.java
@@ -0,0 +1,83 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.Operands;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Concatenates tensors along one dimension.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator
+public final class Concat<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new Concat operation.
+   * 
+   * @param scope current scope
+   * @param values List of `N` Tensors to concatenate. Their ranks and types must match,
+   * and their sizes must match in all dimensions except `concat_dim`.
+   * @param axis 0-D.  The dimension along which to concatenate.  Must be in the
+   * range [-rank(values), rank(values)).
+   * @return a new instance of Concat
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType, U extends TNumber> Concat<T> create(Scope scope, Iterable<Operand<T>> values, Operand<U> axis) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ConcatV2", scope.makeOpName("Concat"));
+    opBuilder.addInputList(Operands.asOutputs(values));
+    opBuilder.addInput(axis.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new Concat<T>(opBuilder.build());
+  }
+  
+  /**
+   * A `Tensor` with the concatenation of values stacked along the
+   * `concat_dim` dimension.  This tensor's shape matches that of `values` except
+   * in `concat_dim` where it has the sum of the sizes.
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ConcatV2";
+  
+  private Output<T> output;
+  
+  private Concat(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ConsumeMutexLock.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ConsumeMutexLock.java
new file mode 100644
index 00000000000..094f6d5e4b0
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ConsumeMutexLock.java
@@ -0,0 +1,63 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+
+/**
+ * This op consumes a lock created by `MutexLock`.
+ * <p>
+ * This op exists to consume a tensor created by `MutexLock` (other than
+ * direct control dependencies).  It should be the only that consumes the tensor,
+ * and will raise an error if it is not.  Its only purpose is to keep the
+ * mutex lock tensor alive until it is consumed by this op.
+ * <p>
+ * <b>NOTE</b>: This operation must run on the same device as its input.  This may
+ * be enforced via the `colocate_with` mechanism.
+ */
+@Operator
+public final class ConsumeMutexLock extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new ConsumeMutexLock operation.
+   * 
+   * @param scope current scope
+   * @param mutexLock A tensor returned by `MutexLock`.
+   * @return a new instance of ConsumeMutexLock
+   */
+  @Endpoint(describeByClass = true)
+  public static ConsumeMutexLock create(Scope scope, Operand<?> mutexLock) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ConsumeMutexLock", scope.makeOpName("ConsumeMutexLock"));
+    opBuilder.addInput(mutexLock.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new ConsumeMutexLock(opBuilder.build());
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ConsumeMutexLock";
+  
+  private ConsumeMutexLock(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ControlTrigger.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ControlTrigger.java
new file mode 100644
index 00000000000..e40715c9f2c
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ControlTrigger.java
@@ -0,0 +1,54 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+
+/**
+ * Does nothing. Serves as a control trigger for scheduling.
+ * <p>
+ * Only useful as a placeholder for control edges.
+ */
+@Operator
+public final class ControlTrigger extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new ControlTrigger operation.
+   * 
+   * @param scope current scope
+   * @return a new instance of ControlTrigger
+   */
+  @Endpoint(describeByClass = true)
+  public static ControlTrigger create(Scope scope) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ControlTrigger", scope.makeOpName("ControlTrigger"));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new ControlTrigger(opBuilder.build());
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ControlTrigger";
+  
+  private ControlTrigger(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Copy.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Copy.java
new file mode 100644
index 00000000000..fb9cbbb122a
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Copy.java
@@ -0,0 +1,148 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Copy a tensor from CPU-to-CPU or GPU-to-GPU.
+ * <p>
+ * Performs CPU-to-CPU or GPU-to-GPU deep-copying of tensor, depending on the
+ * device on which the tensor is allocated.
+ * N.B.: If the all downstream attached debug ops are disabled given the current
+ * gRPC gating status, the output will simply forward the input tensor without
+ * deep-copying. See the documentation of Debug* ops for more details.
+ * <p>
+ * Unlike the CopyHost Op, this op does not have HostMemory constraint on its
+ * input or output.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+public final class Copy<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.core.Copy}
+   */
+  public static class Options {
+    
+    /**
+     * @param tensorName The name of the input tensor.
+     */
+    public Options tensorName(String tensorName) {
+      this.tensorName = tensorName;
+      return this;
+    }
+    
+    /**
+     * @param debugOpsSpec A list of debug op spec (op, url, gated_grpc) for attached debug
+     * ops. Each element of the list has the format
+     * <debug_op>;<grpc_url>;<gated_grpc>, wherein gated_grpc is boolean represented
+     * as 0/1. E.g., "DebugIdentity;grpc://foo:3333;1",
+     * "DebugIdentity;file:///tmp/tfdbg_1;0".
+     */
+    public Options debugOpsSpec(List<String> debugOpsSpec) {
+      this.debugOpsSpec = debugOpsSpec;
+      return this;
+    }
+    
+    private String tensorName;
+    private List<String> debugOpsSpec;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new Copy operation.
+   * 
+   * @param scope current scope
+   * @param input Input tensor.
+   * @param options carries optional attributes values
+   * @return a new instance of Copy
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> Copy<T> create(Scope scope, Operand<T> input, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Copy", scope.makeOpName("Copy"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.tensorName != null) {
+          opBuilder.setAttr("tensor_name", opts.tensorName);
+        }
+        if (opts.debugOpsSpec != null) {
+          String[] debugOpsSpecArray = new String[opts.debugOpsSpec.size()];
+          for (int i = 0; i < debugOpsSpecArray.length; ++i) {
+            debugOpsSpecArray[i] = opts.debugOpsSpec.get(i);
+          }
+          opBuilder.setAttr("debug_ops_spec", debugOpsSpecArray);
+        }
+      }
+    }
+    return new Copy<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param tensorName The name of the input tensor.
+   */
+  public static Options tensorName(String tensorName) {
+    return new Options().tensorName(tensorName);
+  }
+  
+  /**
+   * @param debugOpsSpec A list of debug op spec (op, url, gated_grpc) for attached debug
+   * ops. Each element of the list has the format
+   * <debug_op>;<grpc_url>;<gated_grpc>, wherein gated_grpc is boolean represented
+   * as 0/1. E.g., "DebugIdentity;grpc://foo:3333;1",
+   * "DebugIdentity;file:///tmp/tfdbg_1;0".
+   */
+  public static Options debugOpsSpec(List<String> debugOpsSpec) {
+    return new Options().debugOpsSpec(debugOpsSpec);
+  }
+  
+  /**
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Copy";
+  
+  private Output<T> output;
+  
+  private Copy(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/CopyHost.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/CopyHost.java
new file mode 100644
index 00000000000..566258c0430
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/CopyHost.java
@@ -0,0 +1,146 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Copy a tensor to host.
+ * <p>
+ * Performs CPU-to-CPU deep-copying of tensor.
+ * N.B.: If the all downstream attached debug ops are disabled given the current
+ * gRPC gating status, the output will simply forward the input tensor without
+ * deep-copying. See the documentation of Debug* ops for more details.
+ * <p>
+ * Unlike the Copy Op, this op has HostMemory constraint on its input or output.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+public final class CopyHost<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.core.CopyHost}
+   */
+  public static class Options {
+    
+    /**
+     * @param tensorName The name of the input tensor.
+     */
+    public Options tensorName(String tensorName) {
+      this.tensorName = tensorName;
+      return this;
+    }
+    
+    /**
+     * @param debugOpsSpec A list of debug op spec (op, url, gated_grpc) for attached debug
+     * ops. Each element of the list has the format
+     * <debug_op>;<grpc_url>;<gated_grpc>, wherein gated_grpc is boolean represented
+     * as 0/1. E.g., "DebugIdentity;grpc://foo:3333;1",
+     * "DebugIdentity;file:///tmp/tfdbg_1;0".
+     */
+    public Options debugOpsSpec(List<String> debugOpsSpec) {
+      this.debugOpsSpec = debugOpsSpec;
+      return this;
+    }
+    
+    private String tensorName;
+    private List<String> debugOpsSpec;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new CopyHost operation.
+   * 
+   * @param scope current scope
+   * @param input Input tensor.
+   * @param options carries optional attributes values
+   * @return a new instance of CopyHost
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> CopyHost<T> create(Scope scope, Operand<T> input, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("CopyHost", scope.makeOpName("CopyHost"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.tensorName != null) {
+          opBuilder.setAttr("tensor_name", opts.tensorName);
+        }
+        if (opts.debugOpsSpec != null) {
+          String[] debugOpsSpecArray = new String[opts.debugOpsSpec.size()];
+          for (int i = 0; i < debugOpsSpecArray.length; ++i) {
+            debugOpsSpecArray[i] = opts.debugOpsSpec.get(i);
+          }
+          opBuilder.setAttr("debug_ops_spec", debugOpsSpecArray);
+        }
+      }
+    }
+    return new CopyHost<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param tensorName The name of the input tensor.
+   */
+  public static Options tensorName(String tensorName) {
+    return new Options().tensorName(tensorName);
+  }
+  
+  /**
+   * @param debugOpsSpec A list of debug op spec (op, url, gated_grpc) for attached debug
+   * ops. Each element of the list has the format
+   * <debug_op>;<grpc_url>;<gated_grpc>, wherein gated_grpc is boolean represented
+   * as 0/1. E.g., "DebugIdentity;grpc://foo:3333;1",
+   * "DebugIdentity;file:///tmp/tfdbg_1;0".
+   */
+  public static Options debugOpsSpec(List<String> debugOpsSpec) {
+    return new Options().debugOpsSpec(debugOpsSpec);
+  }
+  
+  /**
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "CopyHost";
+  
+  private Output<T> output;
+  
+  private CopyHost(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/CountUpTo.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/CountUpTo.java
new file mode 100644
index 00000000000..811a955d2fd
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/CountUpTo.java
@@ -0,0 +1,79 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Increments 'ref' until it reaches 'limit'.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator
+public final class CountUpTo<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new CountUpTo operation.
+   * 
+   * @param scope current scope
+   * @param ref Should be from a scalar `Variable` node.
+   * @param limit If incrementing ref would bring it above limit, instead generates an
+   * 'OutOfRange' error.
+   * @return a new instance of CountUpTo
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> CountUpTo<T> create(Scope scope, Operand<T> ref, Long limit) {
+    OperationBuilder opBuilder = scope.env().opBuilder("CountUpTo", scope.makeOpName("CountUpTo"));
+    opBuilder.addInput(ref.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("limit", limit);
+    return new CountUpTo<T>(opBuilder.build());
+  }
+  
+  /**
+   * A copy of the input before increment. If nothing else modifies the
+   * input, the values produced will all be distinct.
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "CountUpTo";
+  
+  private Output<T> output;
+  
+  private CountUpTo(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/DecodeProto.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/DecodeProto.java
new file mode 100644
index 00000000000..40afa08fc97
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/DecodeProto.java
@@ -0,0 +1,223 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import java.util.Arrays;
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.TString;
+import org.tensorflow.types.family.TType;
+
+/**
+ * The op extracts fields from a serialized protocol buffers message into tensors.
+ * <p>
+ * The `decode_proto` op extracts fields from a serialized protocol buffers
+ * message into tensors.  The fields in `field_names` are decoded and converted
+ * to the corresponding `output_types` if possible.
+ * <p>
+ * A `message_type` name must be provided to give context for the field names.
+ * The actual message descriptor can be looked up either in the linked-in
+ * descriptor pool or a filename provided by the caller using the
+ * `descriptor_source` attribute.
+ * <p>
+ * Each output tensor is a dense tensor. This means that it is padded to hold
+ * the largest number of repeated elements seen in the input minibatch. (The
+ * shape is also padded by one to prevent zero-sized dimensions). The actual
+ * repeat counts for each example in the minibatch can be found in the `sizes`
+ * output. In many cases the output of `decode_proto` is fed immediately into
+ * tf.squeeze if missing values are not a concern. When using tf.squeeze, always
+ * pass the squeeze dimension explicitly to avoid surprises.
+ * <p>
+ * For the most part, the mapping between Proto field types and TensorFlow dtypes
+ * is straightforward. However, there are a few special cases:
+ * <p>
+ * - A proto field that contains a submessage or group can only be converted
+ * to `DT_STRING` (the serialized submessage). This is to reduce the complexity
+ * of the API. The resulting string can be used as input to another instance of
+ * the decode_proto op.
+ * <p>
+ * - TensorFlow lacks support for unsigned integers. The ops represent uint64
+ * types as a `DT_INT64` with the same twos-complement bit pattern (the obvious
+ * way). Unsigned int32 values can be represented exactly by specifying type
+ * `DT_INT64`, or using twos-complement if the caller specifies `DT_INT32` in
+ * the `output_types` attribute.
+ * <p>
+ * Both binary and text proto serializations are supported, and can be
+ * chosen using the `format` attribute.
+ * <p>
+ * The `descriptor_source` attribute selects the source of protocol
+ * descriptors to consult when looking up `message_type`. This may be:
+ * <p>
+ * - An empty string  or "local://", in which case protocol descriptors are
+ * created for C++ (not Python) proto definitions linked to the binary.
+ * <p>
+ * - A file, in which case protocol descriptors are created from the file,
+ * which is expected to contain a `FileDescriptorSet` serialized as a string.
+ * NOTE: You can build a `descriptor_source` file using the `--descriptor_set_out`
+ * and `--include_imports` options to the protocol compiler `protoc`.
+ * <p>
+ * - A "bytes://<bytes>", in which protocol descriptors are created from `<bytes>`,
+ * which is expected to be a `FileDescriptorSet` serialized as a string.
+ */
+public final class DecodeProto extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.core.DecodeProto}
+   */
+  public static class Options {
+    
+    /**
+     * @param descriptorSource Either the special value `local://` or a path to a file containing
+     * a serialized `FileDescriptorSet`.
+     */
+    public Options descriptorSource(String descriptorSource) {
+      this.descriptorSource = descriptorSource;
+      return this;
+    }
+    
+    /**
+     * @param messageFormat Either `binary` or `text`.
+     */
+    public Options messageFormat(String messageFormat) {
+      this.messageFormat = messageFormat;
+      return this;
+    }
+    
+    /**
+     * @param sanitize Whether to sanitize the result or not.
+     */
+    public Options sanitize(Boolean sanitize) {
+      this.sanitize = sanitize;
+      return this;
+    }
+    
+    private String descriptorSource;
+    private String messageFormat;
+    private Boolean sanitize;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new DecodeProto operation.
+   * 
+   * @param scope current scope
+   * @param bytes Tensor of serialized protos with shape `batch_shape`.
+   * @param messageType Name of the proto message type to decode.
+   * @param fieldNames List of strings containing proto field names. An extension field can be decoded
+   * by using its full name, e.g. EXT_PACKAGE.EXT_FIELD_NAME.
+   * @param outputTypes List of TF types to use for the respective field in field_names.
+   * @param options carries optional attributes values
+   * @return a new instance of DecodeProto
+   */
+  @Endpoint(describeByClass = true)
+  public static DecodeProto create(Scope scope, Operand<TString> bytes, String messageType, List<String> fieldNames, List<Class<? extends TType>> outputTypes, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("DecodeProtoV2", scope.makeOpName("DecodeProto"));
+    opBuilder.addInput(bytes.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("message_type", messageType);
+    String[] fieldNamesArray = new String[fieldNames.size()];
+    for (int i = 0; i < fieldNamesArray.length; ++i) {
+      fieldNamesArray[i] = fieldNames.get(i);
+    }
+    opBuilder.setAttr("field_names", fieldNamesArray);
+    Class[] outputTypesArray = new Class[outputTypes.size()];
+    for (int i = 0; i < outputTypesArray.length; ++i) {
+      outputTypesArray[i] = outputTypes.get(i);
+    }
+    opBuilder.setAttr("output_types", outputTypesArray);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.descriptorSource != null) {
+          opBuilder.setAttr("descriptor_source", opts.descriptorSource);
+        }
+        if (opts.messageFormat != null) {
+          opBuilder.setAttr("message_format", opts.messageFormat);
+        }
+        if (opts.sanitize != null) {
+          opBuilder.setAttr("sanitize", opts.sanitize);
+        }
+      }
+    }
+    return new DecodeProto(opBuilder.build());
+  }
+  
+  /**
+   * @param descriptorSource Either the special value `local://` or a path to a file containing
+   * a serialized `FileDescriptorSet`.
+   */
+  public static Options descriptorSource(String descriptorSource) {
+    return new Options().descriptorSource(descriptorSource);
+  }
+  
+  /**
+   * @param messageFormat Either `binary` or `text`.
+   */
+  public static Options messageFormat(String messageFormat) {
+    return new Options().messageFormat(messageFormat);
+  }
+  
+  /**
+   * @param sanitize Whether to sanitize the result or not.
+   */
+  public static Options sanitize(Boolean sanitize) {
+    return new Options().sanitize(sanitize);
+  }
+  
+  /**
+   * Tensor of int32 with shape `[batch_shape, len(field_names)]`.
+   * Each entry is the number of values found for the corresponding field.
+   * Optional fields may have 0 or 1 values.
+   */
+  public Output<TInt32> sizes() {
+    return sizes;
+  }
+  
+  /**
+   * List of tensors containing values for the corresponding field.
+   * `values[i]` has datatype `output_types[i]`
+   * and shape `[batch_shape, max(sizes[...,i])]`.
+   */
+  public List<Output<?>> values() {
+    return values;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "DecodeProtoV2";
+  
+  private Output<TInt32> sizes;
+  private List<Output<?>> values;
+  
+  private DecodeProto(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    sizes = operation.output(outputIdx++);
+    int valuesLength = operation.outputListLength("values");
+    values = Arrays.asList(operation.outputList(outputIdx, valuesLength));
+    outputIdx += valuesLength;
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/DeepCopy.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/DeepCopy.java
new file mode 100644
index 00000000000..da22489acc9
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/DeepCopy.java
@@ -0,0 +1,76 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Makes a copy of `x`.
+ * 
+ * @param <T> data type for {@code y()} output
+ */
+@Operator
+public final class DeepCopy<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new DeepCopy operation.
+   * 
+   * @param scope current scope
+   * @param x The source tensor of type `T`.
+   * @return a new instance of DeepCopy
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> DeepCopy<T> create(Scope scope, Operand<T> x) {
+    OperationBuilder opBuilder = scope.env().opBuilder("DeepCopy", scope.makeOpName("DeepCopy"));
+    opBuilder.addInput(x.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new DeepCopy<T>(opBuilder.build());
+  }
+  
+  /**
+   *     y: A `Tensor` of type `T`. A copy of `x`. Guaranteed that `y`
+   *       is not an alias of `x`.
+   */
+  public Output<T> y() {
+    return y;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return y;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "DeepCopy";
+  
+  private Output<T> y;
+  
+  private DeepCopy(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    y = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/DeleteSessionTensor.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/DeleteSessionTensor.java
new file mode 100644
index 00000000000..5f92cc26ca2
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/DeleteSessionTensor.java
@@ -0,0 +1,56 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TString;
+
+/**
+ * Delete the tensor specified by its handle in the session.
+ */
+@Operator
+public final class DeleteSessionTensor extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new DeleteSessionTensor operation.
+   * 
+   * @param scope current scope
+   * @param handle The handle for a tensor stored in the session state.
+   * @return a new instance of DeleteSessionTensor
+   */
+  @Endpoint(describeByClass = true)
+  public static DeleteSessionTensor create(Scope scope, Operand<TString> handle) {
+    OperationBuilder opBuilder = scope.env().opBuilder("DeleteSessionTensor", scope.makeOpName("DeleteSessionTensor"));
+    opBuilder.addInput(handle.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new DeleteSessionTensor(opBuilder.build());
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "DeleteSessionTensor";
+  
+  private DeleteSessionTensor(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/DestroyResourceOp.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/DestroyResourceOp.java
new file mode 100644
index 00000000000..8a427166874
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/DestroyResourceOp.java
@@ -0,0 +1,94 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+
+/**
+ * Deletes the resource specified by the handle.
+ * <p>
+ * All subsequent operations using the resource will result in a NotFound
+ * error status.
+ */
+@Operator
+public final class DestroyResourceOp extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.core.DestroyResourceOp}
+   */
+  public static class Options {
+    
+    /**
+     * @param ignoreLookupError whether to ignore the error when the resource
+     * doesn't exist.
+     */
+    public Options ignoreLookupError(Boolean ignoreLookupError) {
+      this.ignoreLookupError = ignoreLookupError;
+      return this;
+    }
+    
+    private Boolean ignoreLookupError;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new DestroyResourceOp operation.
+   * 
+   * @param scope current scope
+   * @param resource handle to the resource to delete.
+   * @param options carries optional attributes values
+   * @return a new instance of DestroyResourceOp
+   */
+  @Endpoint(describeByClass = true)
+  public static DestroyResourceOp create(Scope scope, Operand<?> resource, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("DestroyResourceOp", scope.makeOpName("DestroyResourceOp"));
+    opBuilder.addInput(resource.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.ignoreLookupError != null) {
+          opBuilder.setAttr("ignore_lookup_error", opts.ignoreLookupError);
+        }
+      }
+    }
+    return new DestroyResourceOp(opBuilder.build());
+  }
+  
+  /**
+   * @param ignoreLookupError whether to ignore the error when the resource
+   * doesn't exist.
+   */
+  public static Options ignoreLookupError(Boolean ignoreLookupError) {
+    return new Options().ignoreLookupError(ignoreLookupError);
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "DestroyResourceOp";
+  
+  private DestroyResourceOp(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/DestroyTemporaryVariable.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/DestroyTemporaryVariable.java
new file mode 100644
index 00000000000..7941a7a5870
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/DestroyTemporaryVariable.java
@@ -0,0 +1,85 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Destroys the temporary variable and returns its final value.
+ * <p>
+ * Sets output to the value of the Tensor pointed to by 'ref', then destroys
+ * the temporary variable called 'var_name'.
+ * All other uses of 'ref' <i>must</i> have executed before this op.
+ * This is typically achieved by chaining the ref through each assign op, or by
+ * using control dependencies.
+ * <p>
+ * Outputs the final value of the tensor pointed to by 'ref'.
+ * 
+ * @param <T> data type for {@code value()} output
+ */
+@Operator
+public final class DestroyTemporaryVariable<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new DestroyTemporaryVariable operation.
+   * 
+   * @param scope current scope
+   * @param ref A reference to the temporary variable tensor.
+   * @param varName Name of the temporary variable, usually the name of the matching
+   * 'TemporaryVariable' op.
+   * @return a new instance of DestroyTemporaryVariable
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> DestroyTemporaryVariable<T> create(Scope scope, Operand<T> ref, String varName) {
+    OperationBuilder opBuilder = scope.env().opBuilder("DestroyTemporaryVariable", scope.makeOpName("DestroyTemporaryVariable"));
+    opBuilder.addInput(ref.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("var_name", varName);
+    return new DestroyTemporaryVariable<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> value() {
+    return value;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return value;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "DestroyTemporaryVariable";
+  
+  private Output<T> value;
+  
+  private DestroyTemporaryVariable(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    value = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/DeviceIndex.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/DeviceIndex.java
new file mode 100644
index 00000000000..f033d3fcc9d
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/DeviceIndex.java
@@ -0,0 +1,81 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+
+/**
+ * Return the index of device the op runs.
+ * <p>
+ * Given a list of device names, this operation returns the index of the device
+ * this op runs. The length of the list is returned in two cases:
+ * (1) Device does not exist in the given device list.
+ * (2) It is in XLA compilation.
+ */
+public final class DeviceIndex extends RawOp implements Operand<TInt32> {
+  
+  /**
+   * Factory method to create a class wrapping a new DeviceIndex operation.
+   * 
+   * @param scope current scope
+   * @param deviceNames 
+   * @return a new instance of DeviceIndex
+   */
+  @Endpoint(describeByClass = true)
+  public static DeviceIndex create(Scope scope, List<String> deviceNames) {
+    OperationBuilder opBuilder = scope.env().opBuilder("DeviceIndex", scope.makeOpName("DeviceIndex"));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    String[] deviceNamesArray = new String[deviceNames.size()];
+    for (int i = 0; i < deviceNamesArray.length; ++i) {
+      deviceNamesArray[i] = deviceNames.get(i);
+    }
+    opBuilder.setAttr("device_names", deviceNamesArray);
+    return new DeviceIndex(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<TInt32> index() {
+    return index;
+  }
+  
+  @Override
+  public Output<TInt32> asOutput() {
+    return index;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "DeviceIndex";
+  
+  private Output<TInt32> index;
+  
+  private DeviceIndex(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    index = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/DummyMemoryCache.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/DummyMemoryCache.java
new file mode 100644
index 00000000000..56f16d48a9b
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/DummyMemoryCache.java
@@ -0,0 +1,69 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ */
+public final class DummyMemoryCache extends RawOp implements Operand<TType> {
+  
+  /**
+   * Factory method to create a class wrapping a new DummyMemoryCache operation.
+   * 
+   * @param scope current scope
+   * @return a new instance of DummyMemoryCache
+   */
+  @Endpoint(describeByClass = true)
+  public static DummyMemoryCache create(Scope scope) {
+    OperationBuilder opBuilder = scope.env().opBuilder("DummyMemoryCache", scope.makeOpName("DummyMemoryCache"));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new DummyMemoryCache(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<?> handle() {
+    return handle;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) handle;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "DummyMemoryCache";
+  
+  private Output<?> handle;
+  
+  private DummyMemoryCache(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    handle = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/DynamicPartition.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/DynamicPartition.java
new file mode 100644
index 00000000000..d9da09b5d34
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/DynamicPartition.java
@@ -0,0 +1,120 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import java.util.Arrays;
+import java.util.Iterator;
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Partitions `data` into `num_partitions` tensors using indices from `partitions`.
+ * <p>
+ * For each index tuple `js` of size `partitions.ndim`, the slice `data[js, ...]`
+ * becomes part of `outputs[partitions[js]]`.  The slices with `partitions[js] = i`
+ * are placed in `outputs[i]` in lexicographic order of `js`, and the first
+ * dimension of `outputs[i]` is the number of entries in `partitions` equal to `i`.
+ * In detail,
+ * <pre>{@code
+ *     outputs[i].shape = [sum(partitions == i)] + data.shape[partitions.ndim:]
+ * 
+ *     outputs[i] = pack([data[js, ...] for js if partitions[js] == i])
+ * }</pre>
+ * `data.shape` must start with `partitions.shape`.
+ * <p>
+ * For example:
+ * <pre>{@code
+ *     # Scalar partitions.
+ *     partitions = 1
+ *     num_partitions = 2
+ *     data = [10, 20]
+ *     outputs[0] = []  # Empty with shape [0, 2]
+ *     outputs[1] = [[10, 20]]
+ * 
+ *     # Vector partitions.
+ *     partitions = [0, 0, 1, 1, 0]
+ *     num_partitions = 2
+ *     data = [10, 20, 30, 40, 50]
+ *     outputs[0] = [10, 20, 50]
+ *     outputs[1] = [30, 40]
+ * }</pre>
+ * See `dynamic_stitch` for an example on how to merge partitions back.
+ * <p>
+ * <div style="width:70%; margin:auto; margin-bottom:10px; margin-top:20px;">
+ * <img style="width:100%" src="https://www.tensorflow.org/images/DynamicPartition.png" alt>
+ * </div>
+ * 
+ * @param <T> data type for {@code outputs()} output
+ */
+@Operator
+public final class DynamicPartition<T extends TType> extends RawOp implements Iterable<Operand<T>> {
+  
+  /**
+   * Factory method to create a class wrapping a new DynamicPartition operation.
+   * 
+   * @param scope current scope
+   * @param data 
+   * @param partitions Any shape.  Indices in the range `[0, num_partitions)`.
+   * @param numPartitions The number of partitions to output.
+   * @return a new instance of DynamicPartition
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> DynamicPartition<T> create(Scope scope, Operand<T> data, Operand<TInt32> partitions, Long numPartitions) {
+    OperationBuilder opBuilder = scope.env().opBuilder("DynamicPartition", scope.makeOpName("DynamicPartition"));
+    opBuilder.addInput(data.asOutput());
+    opBuilder.addInput(partitions.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("num_partitions", numPartitions);
+    return new DynamicPartition<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public List<Output<T>> outputs() {
+    return outputs;
+  }
+  
+  @Override
+  @SuppressWarnings({"rawtypes", "unchecked"})
+  public Iterator<Operand<T>> iterator() {
+    return (Iterator) outputs.iterator();
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "DynamicPartition";
+  
+  private List<Output<T>> outputs;
+  
+  @SuppressWarnings("unchecked")
+  private DynamicPartition(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    int outputsLength = operation.outputListLength("outputs");
+    outputs = Arrays.asList((Output<T>[])operation.outputList(outputIdx, outputsLength));
+    outputIdx += outputsLength;
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/DynamicStitch.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/DynamicStitch.java
new file mode 100644
index 00000000000..b7dd72f8a57
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/DynamicStitch.java
@@ -0,0 +1,133 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.Operands;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Interleave the values from the `data` tensors into a single tensor.
+ * <p>
+ * Builds a merged tensor such that
+ * <pre>{@code
+ *     merged[indices[m][i, ..., j], ...] = data[m][i, ..., j, ...]
+ * }</pre>
+ * For example, if each `indices[m]` is scalar or vector, we have
+ * <pre>{@code
+ *     # Scalar indices:
+ *     merged[indices[m], ...] = data[m][...]
+ * 
+ *     # Vector indices:
+ *     merged[indices[m][i], ...] = data[m][i, ...]
+ * }</pre>
+ * Each `data[i].shape` must start with the corresponding `indices[i].shape`,
+ * and the rest of `data[i].shape` must be constant w.r.t. `i`.  That is, we
+ * must have `data[i].shape = indices[i].shape + constant`.  In terms of this
+ * `constant`, the output shape is
+ * <p>
+ *     merged.shape = [max(indices)] + constant
+ * <p>
+ * Values are merged in order, so if an index appears in both `indices[m][i]` and
+ * `indices[n][j]` for `(m,i) < (n,j)` the slice `data[n][j]` will appear in the
+ * merged result. If you do not need this guarantee, ParallelDynamicStitch might
+ * perform better on some devices.
+ * <p>
+ * For example:
+ * <pre>{@code
+ *     indices[0] = 6
+ *     indices[1] = [4, 1]
+ *     indices[2] = [[5, 2], [0, 3]]
+ *     data[0] = [61, 62]
+ *     data[1] = [[41, 42], [11, 12]]
+ *     data[2] = [[[51, 52], [21, 22]], [[1, 2], [31, 32]]]
+ *     merged = [[1, 2], [11, 12], [21, 22], [31, 32], [41, 42],
+ *               [51, 52], [61, 62]]
+ * }</pre>
+ * This method can be used to merge partitions created by `dynamic_partition`
+ * as illustrated on the following example:
+ * <pre>{@code
+ *     # Apply function (increments x_i) on elements for which a certain condition
+ *     # apply (x_i != -1 in this example).
+ *     x=tf.constant([0.1, -1., 5.2, 4.3, -1., 7.4])
+ *     condition_mask=tf.not_equal(x,tf.constant(-1.))
+ *     partitioned_data = tf.dynamic_partition(
+ *         x, tf.cast(condition_mask, tf.int32) , 2)
+ *     partitioned_data[1] = partitioned_data[1] + 1.0
+ *     condition_indices = tf.dynamic_partition(
+ *         tf.range(tf.shape(x)[0]), tf.cast(condition_mask, tf.int32) , 2)
+ *     x = tf.dynamic_stitch(condition_indices, partitioned_data)
+ *     # Here x=[1.1, -1., 6.2, 5.3, -1, 8.4], the -1. values remain
+ *     # unchanged.
+ * }</pre>
+ * <div style="width:70%; margin:auto; margin-bottom:10px; margin-top:20px;">
+ * <img style="width:100%" src="https://www.tensorflow.org/images/DynamicStitch.png" alt>
+ * </div>
+ * 
+ * @param <T> data type for {@code merged()} output
+ */
+@Operator
+public final class DynamicStitch<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new DynamicStitch operation.
+   * 
+   * @param scope current scope
+   * @param indices 
+   * @param data 
+   * @return a new instance of DynamicStitch
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> DynamicStitch<T> create(Scope scope, Iterable<Operand<TInt32>> indices, Iterable<Operand<T>> data) {
+    OperationBuilder opBuilder = scope.env().opBuilder("DynamicStitch", scope.makeOpName("DynamicStitch"));
+    opBuilder.addInputList(Operands.asOutputs(indices));
+    opBuilder.addInputList(Operands.asOutputs(data));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new DynamicStitch<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> merged() {
+    return merged;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return merged;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "DynamicStitch";
+  
+  private Output<T> merged;
+  
+  private DynamicStitch(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    merged = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/EditDistance.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/EditDistance.java
new file mode 100644
index 00000000000..29423a3c548
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/EditDistance.java
@@ -0,0 +1,164 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Computes the (possibly normalized) Levenshtein Edit Distance.
+ * <p>
+ * The inputs are variable-length sequences provided by SparseTensors
+ *   (hypothesis_indices, hypothesis_values, hypothesis_shape)
+ * and
+ *   (truth_indices, truth_values, truth_shape).
+ * <p>
+ * The inputs are:
+ */
+@Operator
+public final class EditDistance extends RawOp implements Operand<TFloat32> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.core.EditDistance}
+   */
+  public static class Options {
+    
+    /**
+     * @param normalize boolean (if true, edit distances are normalized by length of truth).
+     * <p>
+     * The output is:
+     */
+    public Options normalize(Boolean normalize) {
+      this.normalize = normalize;
+      return this;
+    }
+    
+    private Boolean normalize;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new EditDistance operation.
+   * 
+   * @param scope current scope
+   * @param hypothesisIndices The indices of the hypothesis list SparseTensor.
+   * This is an N x R int64 matrix.
+   * @param hypothesisValues The values of the hypothesis list SparseTensor.
+   * This is an N-length vector.
+   * @param hypothesisShape The shape of the hypothesis list SparseTensor.
+   * This is an R-length vector.
+   * @param truthIndices The indices of the truth list SparseTensor.
+   * This is an M x R int64 matrix.
+   * @param truthValues The values of the truth list SparseTensor.
+   * This is an M-length vector.
+   * @param truthShape truth indices, vector.
+   * @param options carries optional attributes values
+   * @return a new instance of EditDistance
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> EditDistance create(Scope scope, Operand<TInt64> hypothesisIndices, Operand<T> hypothesisValues, Operand<TInt64> hypothesisShape, Operand<TInt64> truthIndices, Operand<T> truthValues, Operand<TInt64> truthShape, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("EditDistance", scope.makeOpName("EditDistance"));
+    opBuilder.addInput(hypothesisIndices.asOutput());
+    opBuilder.addInput(hypothesisValues.asOutput());
+    opBuilder.addInput(hypothesisShape.asOutput());
+    opBuilder.addInput(truthIndices.asOutput());
+    opBuilder.addInput(truthValues.asOutput());
+    opBuilder.addInput(truthShape.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.normalize != null) {
+          opBuilder.setAttr("normalize", opts.normalize);
+        }
+      }
+    }
+    return new EditDistance(opBuilder.build());
+  }
+  
+  /**
+   * @param normalize boolean (if true, edit distances are normalized by length of truth).
+   * <p>
+   * The output is:
+   */
+  public static Options normalize(Boolean normalize) {
+    return new Options().normalize(normalize);
+  }
+  
+  /**
+   * A dense float tensor with rank R - 1.
+   * <p>
+   * For the example input:
+   * <p>
+   *     // hypothesis represents a 2x1 matrix with variable-length values:
+   *     //   (0,0) = ["a"]
+   *     //   (1,0) = ["b"]
+   *     hypothesis_indices = [[0, 0, 0],
+   *                           [1, 0, 0]]
+   *     hypothesis_values = ["a", "b"]
+   *     hypothesis_shape = [2, 1, 1]
+   * <p>
+   *     // truth represents a 2x2 matrix with variable-length values:
+   *     //   (0,0) = []
+   *     //   (0,1) = ["a"]
+   *     //   (1,0) = ["b", "c"]
+   *     //   (1,1) = ["a"]
+   *     truth_indices = [[0, 1, 0],
+   *                      [1, 0, 0],
+   *                      [1, 0, 1],
+   *                      [1, 1, 0]]
+   *     truth_values = ["a", "b", "c", "a"]
+   *     truth_shape = [2, 2, 2]
+   *     normalize = true
+   * <p>
+   * The output will be:
+   * <p>
+   *     // output is a 2x2 matrix with edit distances normalized by truth lengths.
+   *     output = [[inf, 1.0],  // (0,0): no truth, (0,1): no hypothesis
+   *               [0.5, 1.0]]  // (1,0): addition, (1,1): no hypothesis
+   */
+  public Output<TFloat32> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<TFloat32> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "EditDistance";
+  
+  private Output<TFloat32> output;
+  
+  private EditDistance(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Empty.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Empty.java
new file mode 100644
index 00000000000..9330b2dfee3
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Empty.java
@@ -0,0 +1,114 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Creates a tensor with the given shape.
+ * <p>
+ * This operation creates a tensor of `shape` and `dtype`.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator
+public final class Empty<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.core.Empty}
+   */
+  public static class Options {
+    
+    /**
+     * @param init If True, initialize the returned tensor with the default value of dtype.  Otherwise, the implementation is free not to initializethe tensor's content.
+     */
+    public Options init(Boolean init) {
+      this.init = init;
+      return this;
+    }
+    
+    private Boolean init;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new Empty operation.
+   * 
+   * @param scope current scope
+   * @param shape 1-D. Represents the shape of the output tensor.
+   * @param dtype 
+   * @param options carries optional attributes values
+   * @return a new instance of Empty
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> Empty<T> create(Scope scope, Operand<TInt32> shape, Class<T> dtype, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Empty", scope.makeOpName("Empty"));
+    opBuilder.addInput(shape.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("dtype", dtype);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.init != null) {
+          opBuilder.setAttr("init", opts.init);
+        }
+      }
+    }
+    return new Empty<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param init If True, initialize the returned tensor with the default value of dtype.  Otherwise, the implementation is free not to initializethe tensor's content.
+   */
+  public static Options init(Boolean init) {
+    return new Options().init(init);
+  }
+  
+  /**
+   * A `Tensor` of type `T`.
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Empty";
+  
+  private Output<T> output;
+  
+  private Empty(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/EmptyTensorList.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/EmptyTensorList.java
new file mode 100644
index 00000000000..ecd16df5097
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/EmptyTensorList.java
@@ -0,0 +1,86 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Creates and returns an empty tensor list.
+ * <p>
+ * All list elements must be tensors of dtype element_dtype and shape compatible
+ * with element_shape.
+ * <p>
+ * handle: an empty tensor list.
+ * element_dtype: the type of elements in the list.
+ * element_shape: a shape compatible with that of elements in the list.
+ */
+@Operator
+public final class EmptyTensorList extends RawOp implements Operand<TType> {
+  
+  /**
+   * Factory method to create a class wrapping a new EmptyTensorList operation.
+   * 
+   * @param scope current scope
+   * @param elementShape 
+   * @param maxNumElements 
+   * @param elementDtype 
+   * @return a new instance of EmptyTensorList
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber, U extends TType> EmptyTensorList create(Scope scope, Operand<T> elementShape, Operand<TInt32> maxNumElements, Class<U> elementDtype) {
+    OperationBuilder opBuilder = scope.env().opBuilder("EmptyTensorList", scope.makeOpName("EmptyTensorList"));
+    opBuilder.addInput(elementShape.asOutput());
+    opBuilder.addInput(maxNumElements.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("element_dtype", elementDtype);
+    return new EmptyTensorList(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<?> handle() {
+    return handle;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) handle;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "EmptyTensorList";
+  
+  private Output<?> handle;
+  
+  private EmptyTensorList(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    handle = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/EncodeProto.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/EncodeProto.java
new file mode 100644
index 00000000000..4d0396dcb9d
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/EncodeProto.java
@@ -0,0 +1,160 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.Operands;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.TString;
+
+/**
+ * The op serializes protobuf messages provided in the input tensors.
+ * <p>
+ * The types of the tensors in `values` must match the schema for the fields
+ * specified in `field_names`. All the tensors in `values` must have a common
+ * shape prefix, <i>batch_shape</i>.
+ * <p>
+ * The `sizes` tensor specifies repeat counts for each field.  The repeat count
+ * (last dimension) of a each tensor in `values` must be greater than or equal
+ * to corresponding repeat count in `sizes`.
+ * <p>
+ * A `message_type` name must be provided to give context for the field names.
+ * The actual message descriptor can be looked up either in the linked-in
+ * descriptor pool or a filename provided by the caller using the
+ * `descriptor_source` attribute.
+ * <p>
+ * For the most part, the mapping between Proto field types and TensorFlow dtypes
+ * is straightforward. However, there are a few special cases:
+ * <p>
+ * - A proto field that contains a submessage or group can only be converted
+ * to `DT_STRING` (the serialized submessage). This is to reduce the complexity
+ * of the API. The resulting string can be used as input to another instance of
+ * the decode_proto op.
+ * <p>
+ * - TensorFlow lacks support for unsigned integers. The ops represent uint64
+ * types as a `DT_INT64` with the same twos-complement bit pattern (the obvious
+ * way). Unsigned int32 values can be represented exactly by specifying type
+ * `DT_INT64`, or using twos-complement if the caller specifies `DT_INT32` in
+ * the `output_types` attribute.
+ * <p>
+ * The `descriptor_source` attribute selects the source of protocol
+ * descriptors to consult when looking up `message_type`. This may be:
+ * <p>
+ * - An empty string  or "local://", in which case protocol descriptors are
+ * created for C++ (not Python) proto definitions linked to the binary.
+ * <p>
+ * - A file, in which case protocol descriptors are created from the file,
+ * which is expected to contain a `FileDescriptorSet` serialized as a string.
+ * NOTE: You can build a `descriptor_source` file using the `--descriptor_set_out`
+ * and `--include_imports` options to the protocol compiler `protoc`.
+ * <p>
+ * - A "bytes://<bytes>", in which protocol descriptors are created from `<bytes>`,
+ * which is expected to be a `FileDescriptorSet` serialized as a string.
+ */
+public final class EncodeProto extends RawOp implements Operand<TString> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.core.EncodeProto}
+   */
+  public static class Options {
+    
+    /**
+     * @param descriptorSource 
+     */
+    public Options descriptorSource(String descriptorSource) {
+      this.descriptorSource = descriptorSource;
+      return this;
+    }
+    
+    private String descriptorSource;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new EncodeProto operation.
+   * 
+   * @param scope current scope
+   * @param sizes Tensor of int32 with shape `[batch_shape, len(field_names)]`.
+   * @param values List of tensors containing values for the corresponding field.
+   * @param fieldNames List of strings containing proto field names.
+   * @param messageType Name of the proto message type to decode.
+   * @param options carries optional attributes values
+   * @return a new instance of EncodeProto
+   */
+  @Endpoint(describeByClass = true)
+  public static EncodeProto create(Scope scope, Operand<TInt32> sizes, Iterable<Operand<?>> values, List<String> fieldNames, String messageType, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("EncodeProto", scope.makeOpName("EncodeProto"));
+    opBuilder.addInput(sizes.asOutput());
+    opBuilder.addInputList(Operands.asOutputs(values));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    String[] fieldNamesArray = new String[fieldNames.size()];
+    for (int i = 0; i < fieldNamesArray.length; ++i) {
+      fieldNamesArray[i] = fieldNames.get(i);
+    }
+    opBuilder.setAttr("field_names", fieldNamesArray);
+    opBuilder.setAttr("message_type", messageType);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.descriptorSource != null) {
+          opBuilder.setAttr("descriptor_source", opts.descriptorSource);
+        }
+      }
+    }
+    return new EncodeProto(opBuilder.build());
+  }
+  
+  /**
+   * @param descriptorSource 
+   */
+  public static Options descriptorSource(String descriptorSource) {
+    return new Options().descriptorSource(descriptorSource);
+  }
+  
+  /**
+   * Tensor of serialized protos with shape `batch_shape`.
+   */
+  public Output<TString> bytes() {
+    return bytes;
+  }
+  
+  @Override
+  public Output<TString> asOutput() {
+    return bytes;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "EncodeProto";
+  
+  private Output<TString> bytes;
+  
+  private EncodeProto(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    bytes = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/EnsureShape.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/EnsureShape.java
new file mode 100644
index 00000000000..8ea86427b7c
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/EnsureShape.java
@@ -0,0 +1,81 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Ensures that the tensor's shape matches the expected shape.
+ * <p>
+ * Raises an error if the input tensor's shape does not match the specified shape.
+ * Returns the input tensor otherwise.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator
+public final class EnsureShape<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new EnsureShape operation.
+   * 
+   * @param scope current scope
+   * @param input A tensor, whose shape is to be validated.
+   * @param shape The expected (possibly partially specified) shape of the input tensor.
+   * @return a new instance of EnsureShape
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> EnsureShape<T> create(Scope scope, Operand<T> input, Shape shape) {
+    OperationBuilder opBuilder = scope.env().opBuilder("EnsureShape", scope.makeOpName("EnsureShape"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("shape", shape);
+    return new EnsureShape<T>(opBuilder.build());
+  }
+  
+  /**
+   * A tensor with the same shape and contents as the input tensor or value.
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "EnsureShape";
+  
+  private Output<T> output;
+  
+  private EnsureShape(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Enter.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Enter.java
new file mode 100644
index 00000000000..fa19763defb
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Enter.java
@@ -0,0 +1,135 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Creates or finds a child frame, and makes `data` available to the child frame.
+ * <p>
+ * This op is used together with `Exit` to create loops in the graph.
+ * The unique `frame_name` is used by the `Executor` to identify frames. If
+ * `is_constant` is true, `output` is a constant in the child frame; otherwise
+ * it may be changed in the child frame. At most `parallel_iterations` iterations
+ * are run in parallel in the child frame.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+public final class Enter<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.core.Enter}
+   */
+  public static class Options {
+    
+    /**
+     * @param isConstant If true, the output is constant within the child frame.
+     */
+    public Options isConstant(Boolean isConstant) {
+      this.isConstant = isConstant;
+      return this;
+    }
+    
+    /**
+     * @param parallelIterations The number of iterations allowed to run in parallel.
+     */
+    public Options parallelIterations(Long parallelIterations) {
+      this.parallelIterations = parallelIterations;
+      return this;
+    }
+    
+    private Boolean isConstant;
+    private Long parallelIterations;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new Enter operation.
+   * 
+   * @param scope current scope
+   * @param data The tensor to be made available to the child frame.
+   * @param frameName The name of the child frame.
+   * @param options carries optional attributes values
+   * @return a new instance of Enter
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> Enter<T> create(Scope scope, Operand<T> data, String frameName, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Enter", scope.makeOpName("Enter"));
+    opBuilder.addInput(data.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("frame_name", frameName);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.isConstant != null) {
+          opBuilder.setAttr("is_constant", opts.isConstant);
+        }
+        if (opts.parallelIterations != null) {
+          opBuilder.setAttr("parallel_iterations", opts.parallelIterations);
+        }
+      }
+    }
+    return new Enter<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param isConstant If true, the output is constant within the child frame.
+   */
+  public static Options isConstant(Boolean isConstant) {
+    return new Options().isConstant(isConstant);
+  }
+  
+  /**
+   * @param parallelIterations The number of iterations allowed to run in parallel.
+   */
+  public static Options parallelIterations(Long parallelIterations) {
+    return new Options().parallelIterations(parallelIterations);
+  }
+  
+  /**
+   * The same tensor as `data`.
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Enter";
+  
+  private Output<T> output;
+  
+  private Enter(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Exit.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Exit.java
new file mode 100644
index 00000000000..4e42bb0bb0e
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Exit.java
@@ -0,0 +1,76 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Exits the current frame to its parent frame.
+ * <p>
+ * Exit makes its input `data` available to the parent frame.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+public final class Exit<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new Exit operation.
+   * 
+   * @param scope current scope
+   * @param data The tensor to be made available to the parent frame.
+   * @return a new instance of Exit
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> Exit<T> create(Scope scope, Operand<T> data) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Exit", scope.makeOpName("Exit"));
+    opBuilder.addInput(data.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new Exit<T>(opBuilder.build());
+  }
+  
+  /**
+   * The same tensor as `data`.
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Exit";
+  
+  private Output<T> output;
+  
+  private Exit(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ExpandDims.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ExpandDims.java
new file mode 100644
index 00000000000..eb0e977e163
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ExpandDims.java
@@ -0,0 +1,110 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Inserts a dimension of 1 into a tensor's shape.
+ * <p>
+ * Given a tensor `input`, this operation inserts a dimension of 1 at the
+ * dimension index `axis` of `input`'s shape. The dimension index `axis` starts at
+ * zero; if you specify a negative number for `axis` it is counted backward from
+ * the end.
+ * <p>
+ * This operation is useful if you want to add a batch dimension to a single
+ * element. For example, if you have a single image of shape `[height, width,
+ * channels]`, you can make it a batch of 1 image with `expand_dims(image, 0)`,
+ * which will make the shape `[1, height, width, channels]`.
+ * <p>
+ * Other examples:
+ * <pre>{@code
+ * # 't' is a tensor of shape [2]
+ * shape(expand_dims(t, 0)) ==> [1, 2]
+ * shape(expand_dims(t, 1)) ==> [2, 1]
+ * shape(expand_dims(t, -1)) ==> [2, 1]
+ * 
+ * # 't2' is a tensor of shape [2, 3, 5]
+ * shape(expand_dims(t2, 0)) ==> [1, 2, 3, 5]
+ * shape(expand_dims(t2, 2)) ==> [2, 3, 1, 5]
+ * shape(expand_dims(t2, 3)) ==> [2, 3, 5, 1]
+ * }</pre>
+ * This operation requires that:
+ * <p>
+ * `-1-input.dims() <= dim <= input.dims()`
+ * <p>
+ * This operation is related to `squeeze()`, which removes dimensions of
+ * size 1.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator
+public final class ExpandDims<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new ExpandDims operation.
+   * 
+   * @param scope current scope
+   * @param input 
+   * @param axis 0-D (scalar). Specifies the dimension index at which to
+   * expand the shape of `input`. Must be in the range
+   * `[-rank(input) - 1, rank(input)]`.
+   * @return a new instance of ExpandDims
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType, U extends TNumber> ExpandDims<T> create(Scope scope, Operand<T> input, Operand<U> axis) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ExpandDims", scope.makeOpName("ExpandDims"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(axis.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new ExpandDims<T>(opBuilder.build());
+  }
+  
+  /**
+   * Contains the same data as `input`, but its shape has an additional
+   * dimension of size 1 added.
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ExpandDims";
+  
+  private Output<T> output;
+  
+  private ExpandDims(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ExtractVolumePatches.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ExtractVolumePatches.java
new file mode 100644
index 00000000000..487289a8a48
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ExtractVolumePatches.java
@@ -0,0 +1,102 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Extract `patches` from `input` and put them in the "depth" output dimension. 3D extension of `extract_image_patches`.
+ * 
+ * @param <T> data type for {@code patches()} output
+ */
+@Operator
+public final class ExtractVolumePatches<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new ExtractVolumePatches operation.
+   * 
+   * @param scope current scope
+   * @param input 5-D Tensor with shape `[batch, in_planes, in_rows, in_cols, depth]`.
+   * @param ksizes The size of the sliding window for each dimension of `input`.
+   * @param strides 1-D of length 5. How far the centers of two consecutive patches are in
+   * `input`. Must be: `[1, stride_planes, stride_rows, stride_cols, 1]`.
+   * @param padding The type of padding algorithm to use.
+   * <p>
+   * We specify the size-related attributes as:
+   * <pre>{@code
+   *       ksizes = [1, ksize_planes, ksize_rows, ksize_cols, 1]
+   *       strides = [1, stride_planes, strides_rows, strides_cols, 1]
+   * }</pre>
+   * 
+   * @return a new instance of ExtractVolumePatches
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> ExtractVolumePatches<T> create(Scope scope, Operand<T> input, List<Long> ksizes, List<Long> strides, String padding) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ExtractVolumePatches", scope.makeOpName("ExtractVolumePatches"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    long[] ksizesArray = new long[ksizes.size()];
+    for (int i = 0; i < ksizesArray.length; ++i) {
+      ksizesArray[i] = ksizes.get(i);
+    }
+    opBuilder.setAttr("ksizes", ksizesArray);
+    long[] stridesArray = new long[strides.size()];
+    for (int i = 0; i < stridesArray.length; ++i) {
+      stridesArray[i] = strides.get(i);
+    }
+    opBuilder.setAttr("strides", stridesArray);
+    opBuilder.setAttr("padding", padding);
+    return new ExtractVolumePatches<T>(opBuilder.build());
+  }
+  
+  /**
+   * 5-D Tensor with shape `[batch, out_planes, out_rows, out_cols,
+   * ksize_planes * ksize_rows * ksize_cols * depth]` containing patches
+   * with size `ksize_planes x ksize_rows x ksize_cols x depth` vectorized
+   * in the "depth" dimension. Note `out_planes`, `out_rows` and `out_cols`
+   * are the dimensions of the output patches.
+   */
+  public Output<T> patches() {
+    return patches;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return patches;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ExtractVolumePatches";
+  
+  private Output<T> patches;
+  
+  private ExtractVolumePatches(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    patches = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Fill.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Fill.java
new file mode 100644
index 00000000000..79b827ea2aa
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Fill.java
@@ -0,0 +1,104 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Creates a tensor filled with a scalar value.
+ * <p>
+ * This operation creates a tensor of shape `dims` and fills it with `value`.
+ * <p>
+ * For example:
+ * <pre>{@code
+ * # Output tensor has shape [2, 3].
+ * fill([2, 3], 9) ==> [[9, 9, 9]
+ *                      [9, 9, 9]]
+ * }</pre>
+ * `tf.fill` differs from `tf.constant` in a few ways:
+ * <ul>
+ * <li>
+ * `tf.fill` only supports scalar contents, whereas `tf.constant` supports
+ *     Tensor values.
+ * </li>
+ * <li>
+ * `tf.fill` creates an Op in the computation graph that constructs the actual
+ *     Tensor value at runtime. This is in contrast to `tf.constant` which embeds
+ *     the entire Tensor into the graph with a `Const` node.
+ * </li>
+ * <li>
+ * Because `tf.fill` evaluates at graph runtime, it supports dynamic shapes
+ *     based on other runtime Tensors, unlike `tf.constant`.
+ * 
+ * @param <U> data type for {@code output()} output
+ */
+@Operator
+public final class Fill<U extends TType> extends RawOp implements Operand<U> {
+  
+  /**
+   * Factory method to create a class wrapping a new Fill operation.
+   * 
+   * @param scope current scope
+   * @param dims 1-D. Represents the shape of the output tensor.
+   * @param value 0-D (scalar). Value to fill the returned tensor.
+   * <p>
+   * @compatibility(numpy)
+   * Equivalent to np.full
+   * @end_compatibility
+   * @return a new instance of Fill
+   */
+  @Endpoint(describeByClass = true)
+  public static <U extends TType, T extends TNumber> Fill<U> create(Scope scope, Operand<T> dims, Operand<U> value) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Fill", scope.makeOpName("Fill"));
+    opBuilder.addInput(dims.asOutput());
+    opBuilder.addInput(value.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new Fill<U>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<U> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<U> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Fill";
+  
+  private Output<U> output;
+  
+  private Fill(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Fingerprint.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Fingerprint.java
new file mode 100644
index 00000000000..0ea88f2b1f1
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Fingerprint.java
@@ -0,0 +1,109 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TString;
+import org.tensorflow.types.TUint8;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Generates fingerprint values.
+ * <p>
+ * Generates fingerprint values of `data`.
+ * <p>
+ * Fingerprint op considers the first dimension of `data` as the batch dimension,
+ * and `output[i]` contains the fingerprint value generated from contents in
+ * `data[i, ...]` for all `i`.
+ * <p>
+ * Fingerprint op writes fingerprint values as byte arrays. For example, the
+ * default method `farmhash64` generates a 64-bit fingerprint value at a time.
+ * This 8-byte value is written out as an `uint8` array of size 8, in little-endian
+ * order.
+ * <p>
+ * For example, suppose that `data` has data type `DT_INT32` and shape (2, 3, 4),
+ * and that the fingerprint method is `farmhash64`. In this case, the output shape
+ * is (2, 8), where 2 is the batch dimension size of `data`, and 8 is the size of
+ * each fingerprint value in bytes. `output[0, :]` is generated from 12 integers in
+ * `data[0, :, :]` and similarly `output[1, :]` is generated from other 12 integers
+ * in `data[1, :, :]`.
+ * <p>
+ * Note that this op fingerprints the raw underlying buffer, and it does not
+ * fingerprint Tensor's metadata such as data type and/or shape. For example, the
+ * fingerprint values are invariant under reshapes and bitcasts as long as the
+ * batch dimension remain the same:
+ * <pre>{@code
+ * Fingerprint(data) == Fingerprint(Reshape(data, ...))
+ * Fingerprint(data) == Fingerprint(Bitcast(data, ...))
+ * }</pre>
+ * For string data, one should expect `Fingerprint(data) !=
+ * Fingerprint(ReduceJoin(data))` in general.
+ */
+@Operator
+public final class Fingerprint extends RawOp implements Operand<TUint8> {
+  
+  /**
+   * Factory method to create a class wrapping a new Fingerprint operation.
+   * 
+   * @param scope current scope
+   * @param data Must have rank 1 or higher.
+   * @param method Fingerprint method used by this op. Currently available method is
+   * `farmhash::fingerprint64`.
+   * @return a new instance of Fingerprint
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> Fingerprint create(Scope scope, Operand<T> data, Operand<TString> method) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Fingerprint", scope.makeOpName("Fingerprint"));
+    opBuilder.addInput(data.asOutput());
+    opBuilder.addInput(method.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new Fingerprint(opBuilder.build());
+  }
+  
+  /**
+   * A two-dimensional `Tensor` of type `tf.uint8`. The first dimension equals to
+   * `data`'s first dimension, and the second dimension size depends on the
+   * fingerprint algorithm.
+   */
+  public Output<TUint8> fingerprint() {
+    return fingerprint;
+  }
+  
+  @Override
+  public Output<TUint8> asOutput() {
+    return fingerprint;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Fingerprint";
+  
+  private Output<TUint8> fingerprint;
+  
+  private Fingerprint(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    fingerprint = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Gather.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Gather.java
new file mode 100644
index 00000000000..a9bb2f72e69
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Gather.java
@@ -0,0 +1,143 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Gather slices from `params` axis `axis` according to `indices`.
+ * <p>
+ * `indices` must be an integer tensor of any dimension (usually 0-D or 1-D).
+ * Produces an output tensor with shape `params.shape[:axis] +
+ * indices.shape[batch_dims:] + params.shape[axis + 1:]` where:
+ * <pre>{@code
+ *     # Scalar indices (output is rank(params) - 1).
+ *     output[a_0, ..., a_n, b_0, ..., b_n] =
+ *       params[a_0, ..., a_n, indices, b_0, ..., b_n]
+ * 
+ *     # Vector indices (output is rank(params)).
+ *     output[a_0, ..., a_n, i, b_0, ..., b_n] =
+ *       params[a_0, ..., a_n, indices[i], b_0, ..., b_n]
+ * 
+ *     # Higher rank indices (output is rank(params) + rank(indices) - 1).
+ *     output[a_0, ..., a_n, i, ..., j, b_0, ... b_n] =
+ *       params[a_0, ..., a_n, indices[i, ..., j], b_0, ..., b_n]
+ * }</pre>
+ * <div style="width:70%; margin:auto; margin-bottom:10px; margin-top:20px;">
+ * <img style="width:100%" src="https://www.tensorflow.org/images/Gather.png" alt>
+ * </div>
+ * <p>
+ * Note that on CPU, if an out of bound index is found, an error is returned.
+ * On GPU, if an out of bound index is found, a 0 is stored in the
+ * corresponding output value.
+ * <p>
+ * See also `tf.batch_gather` and `tf.gather_nd`.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator
+public final class Gather<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.core.Gather}
+   */
+  public static class Options {
+    
+    /**
+     * @param batchDims 
+     */
+    public Options batchDims(Long batchDims) {
+      this.batchDims = batchDims;
+      return this;
+    }
+    
+    private Long batchDims;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new Gather operation.
+   * 
+   * @param scope current scope
+   * @param params The tensor from which to gather values. Must be at least rank
+   * `axis + 1`.
+   * @param indices Index tensor. Must be in range `[0, params.shape[axis])`.
+   * @param axis The axis in `params` to gather `indices` from. Defaults to the first
+   * dimension. Supports negative indexes.
+   * @param options carries optional attributes values
+   * @return a new instance of Gather
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType, U extends TNumber, V extends TNumber> Gather<T> create(Scope scope, Operand<T> params, Operand<U> indices, Operand<V> axis, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("GatherV2", scope.makeOpName("Gather"));
+    opBuilder.addInput(params.asOutput());
+    opBuilder.addInput(indices.asOutput());
+    opBuilder.addInput(axis.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.batchDims != null) {
+          opBuilder.setAttr("batch_dims", opts.batchDims);
+        }
+      }
+    }
+    return new Gather<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param batchDims 
+   */
+  public static Options batchDims(Long batchDims) {
+    return new Options().batchDims(batchDims);
+  }
+  
+  /**
+   * Values from `params` gathered from indices given by `indices`, with
+   * shape `params.shape[:axis] + indices.shape + params.shape[axis + 1:]`.
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "GatherV2";
+  
+  private Output<T> output;
+  
+  private Gather(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/GatherNd.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/GatherNd.java
new file mode 100644
index 00000000000..e932794001e
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/GatherNd.java
@@ -0,0 +1,172 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Gather slices from `params` into a Tensor with shape specified by `indices`.
+ * <p>
+ * `indices` is a K-dimensional integer tensor, best thought of as a
+ * (K-1)-dimensional tensor of indices into `params`, where each element defines a
+ * slice of `params`:
+ * <p>
+ *     output[\\(i_0, ..., i_{K-2}\\)] = params[indices[\\(i_0, ..., i_{K-2}\\)]]
+ * <p>
+ * Whereas in `tf.gather` `indices` defines slices into the `axis`
+ * dimension of `params`, in `tf.gather_nd`, `indices` defines slices into the
+ * first `N` dimensions of `params`, where `N = indices.shape[-1]`.
+ * <p>
+ * The last dimension of `indices` can be at most the rank of
+ * `params`:
+ * <p>
+ *     indices.shape[-1] <= params.rank
+ * <p>
+ * The last dimension of `indices` corresponds to elements
+ * (if `indices.shape[-1] == params.rank`) or slices
+ * (if `indices.shape[-1] < params.rank`) along dimension `indices.shape[-1]`
+ * of `params`.  The output tensor has shape
+ * <p>
+ *     indices.shape[:-1] + params.shape[indices.shape[-1]:]
+ * <p>
+ * Note that on CPU, if an out of bound index is found, an error is returned.
+ * On GPU, if an out of bound index is found, a 0 is stored in the
+ * corresponding output value.
+ * <p>
+ * Some examples below.
+ * <p>
+ * Simple indexing into a matrix:
+ * <pre>{@code
+ *     indices = [[0, 0], [1, 1]]
+ *     params = [['a', 'b'], ['c', 'd']]
+ *     output = ['a', 'd']
+ * }</pre>
+ * Slice indexing into a matrix:
+ * <pre>{@code
+ *     indices = [[1], [0]]
+ *     params = [['a', 'b'], ['c', 'd']]
+ *     output = [['c', 'd'], ['a', 'b']]
+ * }</pre>
+ * Indexing into a 3-tensor:
+ * <pre>{@code
+ *     indices = [[1]]
+ *     params = [[['a0', 'b0'], ['c0', 'd0']],
+ *               [['a1', 'b1'], ['c1', 'd1']]]
+ *     output = [[['a1', 'b1'], ['c1', 'd1']]]
+ * 
+ * 
+ *     indices = [[0, 1], [1, 0]]
+ *     params = [[['a0', 'b0'], ['c0', 'd0']],
+ *               [['a1', 'b1'], ['c1', 'd1']]]
+ *     output = [['c0', 'd0'], ['a1', 'b1']]
+ * 
+ * 
+ *     indices = [[0, 0, 1], [1, 0, 1]]
+ *     params = [[['a0', 'b0'], ['c0', 'd0']],
+ *               [['a1', 'b1'], ['c1', 'd1']]]
+ *     output = ['b0', 'b1']
+ * }</pre>
+ * Batched indexing into a matrix:
+ * <pre>{@code
+ *     indices = [[[0, 0]], [[0, 1]]]
+ *     params = [['a', 'b'], ['c', 'd']]
+ *     output = [['a'], ['b']]
+ * }</pre>
+ * Batched slice indexing into a matrix:
+ * <pre>{@code
+ *     indices = [[[1]], [[0]]]
+ *     params = [['a', 'b'], ['c', 'd']]
+ *     output = [[['c', 'd']], [['a', 'b']]]
+ * }</pre>
+ * Batched indexing into a 3-tensor:
+ * <pre>{@code
+ *     indices = [[[1]], [[0]]]
+ *     params = [[['a0', 'b0'], ['c0', 'd0']],
+ *               [['a1', 'b1'], ['c1', 'd1']]]
+ *     output = [[[['a1', 'b1'], ['c1', 'd1']]],
+ *               [[['a0', 'b0'], ['c0', 'd0']]]]
+ * 
+ *     indices = [[[0, 1], [1, 0]], [[0, 0], [1, 1]]]
+ *     params = [[['a0', 'b0'], ['c0', 'd0']],
+ *               [['a1', 'b1'], ['c1', 'd1']]]
+ *     output = [[['c0', 'd0'], ['a1', 'b1']],
+ *               [['a0', 'b0'], ['c1', 'd1']]]
+ * 
+ * 
+ *     indices = [[[0, 0, 1], [1, 0, 1]], [[0, 1, 1], [1, 1, 0]]]
+ *     params = [[['a0', 'b0'], ['c0', 'd0']],
+ *               [['a1', 'b1'], ['c1', 'd1']]]
+ *     output = [['b0', 'b1'], ['d0', 'c1']]
+ * }</pre>
+ * See also `tf.gather` and `tf.batch_gather`.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator
+public final class GatherNd<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new GatherNd operation.
+   * 
+   * @param scope current scope
+   * @param params The tensor from which to gather values.
+   * @param indices Index tensor.
+   * @return a new instance of GatherNd
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType, U extends TNumber> GatherNd<T> create(Scope scope, Operand<T> params, Operand<U> indices) {
+    OperationBuilder opBuilder = scope.env().opBuilder("GatherNd", scope.makeOpName("GatherNd"));
+    opBuilder.addInput(params.asOutput());
+    opBuilder.addInput(indices.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new GatherNd<T>(opBuilder.build());
+  }
+  
+  /**
+   * Values from `params` gathered from indices given by `indices`, with
+   * shape `indices.shape[:-1] + params.shape[indices.shape[-1]:]`.
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "GatherNd";
+  
+  private Output<T> output;
+  
+  private GatherNd(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/GetSessionHandle.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/GetSessionHandle.java
new file mode 100644
index 00000000000..93a31f1cba1
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/GetSessionHandle.java
@@ -0,0 +1,75 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Store the input tensor in the state of the current session.
+ */
+@Operator
+public final class GetSessionHandle extends RawOp implements Operand<TType> {
+  
+  /**
+   * Factory method to create a class wrapping a new GetSessionHandle operation.
+   * 
+   * @param scope current scope
+   * @param value The tensor to be stored.
+   * @return a new instance of GetSessionHandle
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> GetSessionHandle create(Scope scope, Operand<T> value) {
+    OperationBuilder opBuilder = scope.env().opBuilder("GetSessionHandleV2", scope.makeOpName("GetSessionHandle"));
+    opBuilder.addInput(value.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new GetSessionHandle(opBuilder.build());
+  }
+  
+  /**
+   * The handle for the tensor stored in the session state, represented
+   * as a ResourceHandle object.
+   */
+  public Output<?> handle() {
+    return handle;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) handle;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "GetSessionHandleV2";
+  
+  private Output<?> handle;
+  
+  private GetSessionHandle(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    handle = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/GetSessionTensor.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/GetSessionTensor.java
new file mode 100644
index 00000000000..d5767515e56
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/GetSessionTensor.java
@@ -0,0 +1,78 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TString;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Get the value of the tensor specified by its handle.
+ * 
+ * @param <T> data type for {@code value()} output
+ */
+@Operator
+public final class GetSessionTensor<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new GetSessionTensor operation.
+   * 
+   * @param scope current scope
+   * @param handle The handle for a tensor stored in the session state.
+   * @param dtype The type of the output value.
+   * @return a new instance of GetSessionTensor
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> GetSessionTensor<T> create(Scope scope, Operand<TString> handle, Class<T> dtype) {
+    OperationBuilder opBuilder = scope.env().opBuilder("GetSessionTensor", scope.makeOpName("GetSessionTensor"));
+    opBuilder.addInput(handle.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("dtype", dtype);
+    return new GetSessionTensor<T>(opBuilder.build());
+  }
+  
+  /**
+   * The tensor for the given handle.
+   */
+  public Output<T> value() {
+    return value;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return value;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "GetSessionTensor";
+  
+  private Output<T> value;
+  
+  private GetSessionTensor(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    value = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/GuaranteeConst.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/GuaranteeConst.java
new file mode 100644
index 00000000000..df0a48ad28f
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/GuaranteeConst.java
@@ -0,0 +1,81 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Gives a guarantee to the TF runtime that the input tensor is a constant.
+ * <p>
+ * The runtime is then free to make optimizations based on this.
+ * <p>
+ * Only accepts value typed tensors as inputs and rejects resource variable handles
+ * as input.
+ * <p>
+ * Returns the input tensor without modification.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator
+public final class GuaranteeConst<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new GuaranteeConst operation.
+   * 
+   * @param scope current scope
+   * @param input 
+   * @return a new instance of GuaranteeConst
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> GuaranteeConst<T> create(Scope scope, Operand<T> input) {
+    OperationBuilder opBuilder = scope.env().opBuilder("GuaranteeConst", scope.makeOpName("GuaranteeConst"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new GuaranteeConst<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "GuaranteeConst";
+  
+  private Output<T> output;
+  
+  private GuaranteeConst(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/HashTable.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/HashTable.java
new file mode 100644
index 00000000000..aea656de1ec
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/HashTable.java
@@ -0,0 +1,158 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Creates a non-initialized hash table.
+ * <p>
+ * This op creates a hash table, specifying the type of its keys and values.
+ * Before using the table you will have to initialize it.  After initialization the
+ * table will be immutable.
+ */
+@Operator
+public final class HashTable extends RawOp implements Operand<TType> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.core.HashTable}
+   */
+  public static class Options {
+    
+    /**
+     * @param container If non-empty, this table is placed in the given container.
+     * Otherwise, a default container is used.
+     */
+    public Options container(String container) {
+      this.container = container;
+      return this;
+    }
+    
+    /**
+     * @param sharedName If non-empty, this table is shared under the given name across
+     * multiple sessions.
+     */
+    public Options sharedName(String sharedName) {
+      this.sharedName = sharedName;
+      return this;
+    }
+    
+    /**
+     * @param useNodeNameSharing If true and shared_name is empty, the table is shared
+     * using the node name.
+     */
+    public Options useNodeNameSharing(Boolean useNodeNameSharing) {
+      this.useNodeNameSharing = useNodeNameSharing;
+      return this;
+    }
+    
+    private String container;
+    private String sharedName;
+    private Boolean useNodeNameSharing;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new HashTable operation.
+   * 
+   * @param scope current scope
+   * @param keyDtype Type of the table keys.
+   * @param valueDtype Type of the table values.
+   * @param options carries optional attributes values
+   * @return a new instance of HashTable
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType, U extends TType> HashTable create(Scope scope, Class<T> keyDtype, Class<U> valueDtype, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("HashTableV2", scope.makeOpName("HashTable"));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("key_dtype", keyDtype);
+    opBuilder.setAttr("value_dtype", valueDtype);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.container != null) {
+          opBuilder.setAttr("container", opts.container);
+        }
+        if (opts.sharedName != null) {
+          opBuilder.setAttr("shared_name", opts.sharedName);
+        }
+        if (opts.useNodeNameSharing != null) {
+          opBuilder.setAttr("use_node_name_sharing", opts.useNodeNameSharing);
+        }
+      }
+    }
+    return new HashTable(opBuilder.build());
+  }
+  
+  /**
+   * @param container If non-empty, this table is placed in the given container.
+   * Otherwise, a default container is used.
+   */
+  public static Options container(String container) {
+    return new Options().container(container);
+  }
+  
+  /**
+   * @param sharedName If non-empty, this table is shared under the given name across
+   * multiple sessions.
+   */
+  public static Options sharedName(String sharedName) {
+    return new Options().sharedName(sharedName);
+  }
+  
+  /**
+   * @param useNodeNameSharing If true and shared_name is empty, the table is shared
+   * using the node name.
+   */
+  public static Options useNodeNameSharing(Boolean useNodeNameSharing) {
+    return new Options().useNodeNameSharing(useNodeNameSharing);
+  }
+  
+  /**
+   * Handle to a table.
+   */
+  public Output<?> tableHandle() {
+    return tableHandle;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) tableHandle;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "HashTableV2";
+  
+  private Output<?> tableHandle;
+  
+  private HashTable(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    tableHandle = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/HistogramFixedWidth.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/HistogramFixedWidth.java
new file mode 100644
index 00000000000..1565dd2a8d0
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/HistogramFixedWidth.java
@@ -0,0 +1,116 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Return histogram of values.
+ * <p>
+ * Given the tensor `values`, this operation returns a rank 1 histogram counting
+ * the number of entries in `values` that fall into every bin.  The bins are
+ * equal width and determined by the arguments `value_range` and `nbins`.
+ * <pre>{@code
+ * # Bins will be:  (-inf, 1), [1, 2), [2, 3), [3, 4), [4, inf)
+ * nbins = 5
+ * value_range = [0.0, 5.0]
+ * new_values = [-1.0, 0.0, 1.5, 2.0, 5.0, 15]
+ * 
+ * with tf.get_default_session() as sess:
+ *   hist = tf.histogram_fixed_width(new_values, value_range, nbins=5)
+ *   variables.global_variables_initializer().run()
+ *   sess.run(hist) => [2, 1, 1, 0, 2]
+ * }</pre>
+ * 
+ * 
+ * @param <U> data type for {@code out()} output
+ */
+@Operator
+public final class HistogramFixedWidth<U extends TNumber> extends RawOp implements Operand<U> {
+  
+  /**
+   * Factory method to create a class wrapping a new HistogramFixedWidth operation.
+   * 
+   * @param scope current scope
+   * @param values Numeric `Tensor`.
+   * @param valueRange Shape [2] `Tensor` of same `dtype` as `values`.
+   * values <= value_range[0] will be mapped to hist[0],
+   * values >= value_range[1] will be mapped to hist[-1].
+   * @param nbins Scalar `int32 Tensor`.  Number of histogram bins.
+   * @param dtype 
+   * @return a new instance of HistogramFixedWidth
+   */
+  @Endpoint(describeByClass = true)
+  public static <U extends TNumber, T extends TNumber> HistogramFixedWidth<U> create(Scope scope, Operand<T> values, Operand<T> valueRange, Operand<TInt32> nbins, Class<U> dtype) {
+    OperationBuilder opBuilder = scope.env().opBuilder("HistogramFixedWidth", scope.makeOpName("HistogramFixedWidth"));
+    opBuilder.addInput(values.asOutput());
+    opBuilder.addInput(valueRange.asOutput());
+    opBuilder.addInput(nbins.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("dtype", dtype);
+    return new HistogramFixedWidth<U>(opBuilder.build());
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new HistogramFixedWidth operation using default output types.
+   * 
+   * @param scope current scope
+   * @param values Numeric `Tensor`.
+   * @param valueRange Shape [2] `Tensor` of same `dtype` as `values`.
+   * values <= value_range[0] will be mapped to hist[0],
+   * values >= value_range[1] will be mapped to hist[-1].
+   * @param nbins Scalar `int32 Tensor`.  Number of histogram bins.
+   * @return a new instance of HistogramFixedWidth
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> HistogramFixedWidth<TInt32> create(Scope scope, Operand<T> values, Operand<T> valueRange, Operand<TInt32> nbins) {
+    return create(scope, values, valueRange, nbins, TInt32.class);
+  }
+  
+  /**
+   * A 1-D `Tensor` holding histogram of values.
+   */
+  public Output<U> out() {
+    return out;
+  }
+  
+  @Override
+  public Output<U> asOutput() {
+    return out;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "HistogramFixedWidth";
+  
+  private Output<U> out;
+  
+  private HistogramFixedWidth(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    out = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Identity.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Identity.java
new file mode 100644
index 00000000000..1cafffdf5c9
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Identity.java
@@ -0,0 +1,74 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Return a tensor with the same shape and contents as the input tensor or value.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator
+public final class Identity<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new Identity operation.
+   * 
+   * @param scope current scope
+   * @param input 
+   * @return a new instance of Identity
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> Identity<T> create(Scope scope, Operand<T> input) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Identity", scope.makeOpName("Identity"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new Identity<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Identity";
+  
+  private Output<T> output;
+  
+  private Identity(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/IdentityN.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/IdentityN.java
new file mode 100644
index 00000000000..e000005eabf
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/IdentityN.java
@@ -0,0 +1,95 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import java.util.Arrays;
+import java.util.Iterator;
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.Operands;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Returns a list of tensors with the same shapes and contents as the input
+ * <p>
+ * tensors.
+ * <p>
+ * This op can be used to override the gradient for complicated functions. For
+ * example, suppose y = f(x) and we wish to apply a custom function g for backprop
+ * such that dx = g(dy). In Python,
+ * <pre>{@code
+ * with tf.get_default_graph().gradient_override_map(
+ *     {'IdentityN': 'OverrideGradientWithG'}):
+ *   y, _ = identity_n([f(x), x])
+ * 
+ * @tf.RegisterGradient('OverrideGradientWithG')
+ * def ApplyG(op, dy, _):
+ *   return [None, g(dy)]  # Do not backprop to f(x).
+ * }</pre>
+ * 
+ */
+@Operator
+public final class IdentityN extends RawOp implements Iterable<Operand<TType>> {
+  
+  /**
+   * Factory method to create a class wrapping a new IdentityN operation.
+   * 
+   * @param scope current scope
+   * @param input 
+   * @return a new instance of IdentityN
+   */
+  @Endpoint(describeByClass = true)
+  public static IdentityN create(Scope scope, Iterable<Operand<?>> input) {
+    OperationBuilder opBuilder = scope.env().opBuilder("IdentityN", scope.makeOpName("IdentityN"));
+    opBuilder.addInputList(Operands.asOutputs(input));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new IdentityN(opBuilder.build());
+  }
+  
+  /**
+   */
+  public List<Output<?>> output() {
+    return output;
+  }
+  
+  @Override
+  @SuppressWarnings({"rawtypes", "unchecked"})
+  public Iterator<Operand<TType>> iterator() {
+    return (Iterator) output.iterator();
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "IdentityN";
+  
+  private List<Output<?>> output;
+  
+  private IdentityN(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    int outputLength = operation.outputListLength("output");
+    output = Arrays.asList(operation.outputList(outputIdx, outputLength));
+    outputIdx += outputLength;
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ImmutableConst.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ImmutableConst.java
new file mode 100644
index 00000000000..3d2ac9625da
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ImmutableConst.java
@@ -0,0 +1,82 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Returns immutable tensor from memory region.
+ * <p>
+ * The current implementation memmaps the tensor from a file.
+ * 
+ * @param <T> data type for {@code tensor()} output
+ */
+@Operator
+public final class ImmutableConst<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new ImmutableConst operation.
+   * 
+   * @param scope current scope
+   * @param dtype Type of the returned tensor.
+   * @param shape Shape of the returned tensor.
+   * @param memoryRegionName Name of readonly memory region used by the tensor, see
+   * NewReadOnlyMemoryRegionFromFile in tensorflow::Env.
+   * @return a new instance of ImmutableConst
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> ImmutableConst<T> create(Scope scope, Class<T> dtype, Shape shape, String memoryRegionName) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ImmutableConst", scope.makeOpName("ImmutableConst"));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("dtype", dtype);
+    opBuilder.setAttr("shape", shape);
+    opBuilder.setAttr("memory_region_name", memoryRegionName);
+    return new ImmutableConst<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> tensor() {
+    return tensor;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return tensor;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ImmutableConst";
+  
+  private Output<T> tensor;
+  
+  private ImmutableConst(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    tensor = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/InitializeTable.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/InitializeTable.java
new file mode 100644
index 00000000000..48662ed420d
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/InitializeTable.java
@@ -0,0 +1,60 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Table initializer that takes two tensors for keys and values respectively.
+ */
+@Operator
+public final class InitializeTable extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new InitializeTable operation.
+   * 
+   * @param scope current scope
+   * @param tableHandle Handle to a table which will be initialized.
+   * @param keys Keys of type Tkey.
+   * @param values Values of type Tval.
+   * @return a new instance of InitializeTable
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType, U extends TType> InitializeTable create(Scope scope, Operand<?> tableHandle, Operand<T> keys, Operand<U> values) {
+    OperationBuilder opBuilder = scope.env().opBuilder("InitializeTableV2", scope.makeOpName("InitializeTable"));
+    opBuilder.addInput(tableHandle.asOutput());
+    opBuilder.addInput(keys.asOutput());
+    opBuilder.addInput(values.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new InitializeTable(opBuilder.build());
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "InitializeTableV2";
+  
+  private InitializeTable(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/InitializeTableFromTextFile.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/InitializeTableFromTextFile.java
new file mode 100644
index 00000000000..2050c4d8628
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/InitializeTableFromTextFile.java
@@ -0,0 +1,127 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TString;
+
+/**
+ * Initializes a table from a text file.
+ * <p>
+ * It inserts one key-value pair into the table for each line of the file.
+ * The key and value is extracted from the whole line content, elements from the
+ * split line based on `delimiter` or the line number (starting from zero).
+ * Where to extract the key and value from a line is specified by `key_index` and
+ * `value_index`.
+ * <p>
+ * - A value of -1 means use the line number(starting from zero), expects `int64`.
+ * - A value of -2 means use the whole line content, expects `string`.
+ * - A value >= 0 means use the index (starting at zero) of the split line based
+ *   on `delimiter`.
+ */
+@Operator
+public final class InitializeTableFromTextFile extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.core.InitializeTableFromTextFile}
+   */
+  public static class Options {
+    
+    /**
+     * @param vocabSize Number of elements of the file, use -1 if unknown.
+     */
+    public Options vocabSize(Long vocabSize) {
+      this.vocabSize = vocabSize;
+      return this;
+    }
+    
+    /**
+     * @param delimiter Delimiter to separate fields in a line.
+     */
+    public Options delimiter(String delimiter) {
+      this.delimiter = delimiter;
+      return this;
+    }
+    
+    private Long vocabSize;
+    private String delimiter;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new InitializeTableFromTextFile operation.
+   * 
+   * @param scope current scope
+   * @param tableHandle Handle to a table which will be initialized.
+   * @param filename Filename of a vocabulary text file.
+   * @param keyIndex Column index in a line to get the table `key` values from.
+   * @param valueIndex Column index that represents information of a line to get the table
+   * `value` values from.
+   * @param options carries optional attributes values
+   * @return a new instance of InitializeTableFromTextFile
+   */
+  @Endpoint(describeByClass = true)
+  public static InitializeTableFromTextFile create(Scope scope, Operand<?> tableHandle, Operand<TString> filename, Long keyIndex, Long valueIndex, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("InitializeTableFromTextFileV2", scope.makeOpName("InitializeTableFromTextFile"));
+    opBuilder.addInput(tableHandle.asOutput());
+    opBuilder.addInput(filename.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("key_index", keyIndex);
+    opBuilder.setAttr("value_index", valueIndex);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.vocabSize != null) {
+          opBuilder.setAttr("vocab_size", opts.vocabSize);
+        }
+        if (opts.delimiter != null) {
+          opBuilder.setAttr("delimiter", opts.delimiter);
+        }
+      }
+    }
+    return new InitializeTableFromTextFile(opBuilder.build());
+  }
+  
+  /**
+   * @param vocabSize Number of elements of the file, use -1 if unknown.
+   */
+  public static Options vocabSize(Long vocabSize) {
+    return new Options().vocabSize(vocabSize);
+  }
+  
+  /**
+   * @param delimiter Delimiter to separate fields in a line.
+   */
+  public static Options delimiter(String delimiter) {
+    return new Options().delimiter(delimiter);
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "InitializeTableFromTextFileV2";
+  
+  private InitializeTableFromTextFile(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/InplaceAdd.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/InplaceAdd.java
new file mode 100644
index 00000000000..8de34bc569f
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/InplaceAdd.java
@@ -0,0 +1,82 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TType;
+
+/**
+ *     Adds v into specified rows of x.
+ * <p>
+ *     Computes y = x; y[i, :] += v; return y.
+ * 
+ * @param <T> data type for {@code y()} output
+ */
+@Operator
+public final class InplaceAdd<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new InplaceAdd operation.
+   * 
+   * @param scope current scope
+   * @param x A `Tensor` of type T.
+   * @param i A vector. Indices into the left-most dimension of `x`.
+   * @param v A `Tensor` of type T. Same dimension sizes as x except the first dimension, which must be the same as i's size.
+   * @return a new instance of InplaceAdd
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> InplaceAdd<T> create(Scope scope, Operand<T> x, Operand<TInt32> i, Operand<T> v) {
+    OperationBuilder opBuilder = scope.env().opBuilder("InplaceAdd", scope.makeOpName("InplaceAdd"));
+    opBuilder.addInput(x.asOutput());
+    opBuilder.addInput(i.asOutput());
+    opBuilder.addInput(v.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new InplaceAdd<T>(opBuilder.build());
+  }
+  
+  /**
+   * A `Tensor` of type T. An alias of `x`. The content of `y` is undefined if there are duplicates in `i`.
+   */
+  public Output<T> y() {
+    return y;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return y;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "InplaceAdd";
+  
+  private Output<T> y;
+  
+  private InplaceAdd(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    y = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/InplaceSub.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/InplaceSub.java
new file mode 100644
index 00000000000..999f793c869
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/InplaceSub.java
@@ -0,0 +1,82 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TType;
+
+/**
+ *     Subtracts `v` into specified rows of `x`.
+ * <p>
+ *     Computes y = x; y[i, :] -= v; return y.
+ * 
+ * @param <T> data type for {@code y()} output
+ */
+@Operator
+public final class InplaceSub<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new InplaceSub operation.
+   * 
+   * @param scope current scope
+   * @param x A `Tensor` of type T.
+   * @param i A vector. Indices into the left-most dimension of `x`.
+   * @param v A `Tensor` of type T. Same dimension sizes as x except the first dimension, which must be the same as i's size.
+   * @return a new instance of InplaceSub
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> InplaceSub<T> create(Scope scope, Operand<T> x, Operand<TInt32> i, Operand<T> v) {
+    OperationBuilder opBuilder = scope.env().opBuilder("InplaceSub", scope.makeOpName("InplaceSub"));
+    opBuilder.addInput(x.asOutput());
+    opBuilder.addInput(i.asOutput());
+    opBuilder.addInput(v.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new InplaceSub<T>(opBuilder.build());
+  }
+  
+  /**
+   * A `Tensor` of type T. An alias of `x`. The content of `y` is undefined if there are duplicates in `i`.
+   */
+  public Output<T> y() {
+    return y;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return y;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "InplaceSub";
+  
+  private Output<T> y;
+  
+  private InplaceSub(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    y = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/InplaceUpdate.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/InplaceUpdate.java
new file mode 100644
index 00000000000..f7d2f79e418
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/InplaceUpdate.java
@@ -0,0 +1,85 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Updates specified rows 'i' with values 'v'.
+ * <p>
+ * Computes `x[i, :] = v; return x`.
+ * <p>
+ * Originally this function is mutative however for compilation we make this
+ * operation create / operate on a copy of `x`.
+ * 
+ * @param <T> data type for {@code y()} output
+ */
+@Operator
+public final class InplaceUpdate<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new InplaceUpdate operation.
+   * 
+   * @param scope current scope
+   * @param x A tensor of type `T`.
+   * @param i A vector. Indices into the left-most dimension of `x`.
+   * @param v A `Tensor` of type T. Same dimension sizes as x except the first dimension, which must be the same as i's size.
+   * @return a new instance of InplaceUpdate
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> InplaceUpdate<T> create(Scope scope, Operand<T> x, Operand<TInt32> i, Operand<T> v) {
+    OperationBuilder opBuilder = scope.env().opBuilder("InplaceUpdate", scope.makeOpName("InplaceUpdate"));
+    opBuilder.addInput(x.asOutput());
+    opBuilder.addInput(i.asOutput());
+    opBuilder.addInput(v.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new InplaceUpdate<T>(opBuilder.build());
+  }
+  
+  /**
+   * A `Tensor` of type T. An alias of `x`. The content of `y` is undefined if there are duplicates in `i`.
+   */
+  public Output<T> y() {
+    return y;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return y;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "InplaceUpdate";
+  
+  private Output<T> y;
+  
+  private InplaceUpdate(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    y = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/IsVariableInitialized.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/IsVariableInitialized.java
new file mode 100644
index 00000000000..e9dc32ec1d7
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/IsVariableInitialized.java
@@ -0,0 +1,75 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TBool;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Checks whether a tensor has been initialized.
+ * <p>
+ * Outputs boolean scalar indicating whether the tensor has been initialized.
+ */
+@Operator
+public final class IsVariableInitialized extends RawOp implements Operand<TBool> {
+  
+  /**
+   * Factory method to create a class wrapping a new IsVariableInitialized operation.
+   * 
+   * @param scope current scope
+   * @param ref Should be from a `Variable` node. May be uninitialized.
+   * @return a new instance of IsVariableInitialized
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> IsVariableInitialized create(Scope scope, Operand<T> ref) {
+    OperationBuilder opBuilder = scope.env().opBuilder("IsVariableInitialized", scope.makeOpName("IsVariableInitialized"));
+    opBuilder.addInput(ref.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new IsVariableInitialized(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<TBool> isInitialized() {
+    return isInitialized;
+  }
+  
+  @Override
+  public Output<TBool> asOutput() {
+    return isInitialized;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "IsVariableInitialized";
+  
+  private Output<TBool> isInitialized;
+  
+  private IsVariableInitialized(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    isInitialized = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LinSpace.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LinSpace.java
new file mode 100644
index 00000000000..fbfa0e69881
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LinSpace.java
@@ -0,0 +1,88 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Generates values in an interval.
+ * <p>
+ * A sequence of `num` evenly-spaced values are generated beginning at `start`.
+ * If `num > 1`, the values in the sequence increase by `stop - start / num - 1`,
+ * so that the last one is exactly `stop`.
+ * <p>
+ * For example:
+ * <pre>{@code
+ * tf.linspace(10.0, 12.0, 3, name="linspace") => [ 10.0  11.0  12.0]
+ * }</pre>
+ * 
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+public final class LinSpace<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new LinSpace operation.
+   * 
+   * @param scope current scope
+   * @param start 0-D tensor. First entry in the range.
+   * @param stop 0-D tensor. Last entry in the range.
+   * @param num 0-D tensor. Number of values to generate.
+   * @return a new instance of LinSpace
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber, U extends TNumber> LinSpace<T> create(Scope scope, Operand<T> start, Operand<T> stop, Operand<U> num) {
+    OperationBuilder opBuilder = scope.env().opBuilder("LinSpace", scope.makeOpName("LinSpace"));
+    opBuilder.addInput(start.asOutput());
+    opBuilder.addInput(stop.asOutput());
+    opBuilder.addInput(num.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new LinSpace<T>(opBuilder.build());
+  }
+  
+  /**
+   * 1-D. The generated values.
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "LinSpace";
+  
+  private Output<T> output;
+  
+  private LinSpace(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LookupTableExport.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LookupTableExport.java
new file mode 100644
index 00000000000..527e70e7fc0
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LookupTableExport.java
@@ -0,0 +1,84 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Outputs all keys and values in the table.
+ * 
+ * @param <T> data type for {@code keys()} output
+ * @param <U> data type for {@code values()} output
+ */
+@Operator
+public final class LookupTableExport<T extends TType, U extends TType> extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new LookupTableExport operation.
+   * 
+   * @param scope current scope
+   * @param tableHandle Handle to the table.
+   * @param Tkeys 
+   * @param Tvalues 
+   * @return a new instance of LookupTableExport
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType, U extends TType> LookupTableExport<T, U> create(Scope scope, Operand<?> tableHandle, Class<T> Tkeys, Class<U> Tvalues) {
+    OperationBuilder opBuilder = scope.env().opBuilder("LookupTableExportV2", scope.makeOpName("LookupTableExport"));
+    opBuilder.addInput(tableHandle.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("Tkeys", Tkeys);
+    opBuilder.setAttr("Tvalues", Tvalues);
+    return new LookupTableExport<T, U>(opBuilder.build());
+  }
+  
+  /**
+   * Vector of all keys present in the table.
+   */
+  public Output<T> keys() {
+    return keys;
+  }
+  
+  /**
+   * Tensor of all values in the table. Indexed in parallel with `keys`.
+   */
+  public Output<U> values() {
+    return values;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "LookupTableExportV2";
+  
+  private Output<T> keys;
+  private Output<U> values;
+  
+  private LookupTableExport(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    keys = operation.output(outputIdx++);
+    values = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LookupTableFind.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LookupTableFind.java
new file mode 100644
index 00000000000..bfc8e92e16f
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LookupTableFind.java
@@ -0,0 +1,86 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Looks up keys in a table, outputs the corresponding values.
+ * <p>
+ * The tensor `keys` must of the same type as the keys of the table.
+ * The output `values` is of the type of the table values.
+ * <p>
+ * The scalar `default_value` is the value output for keys not present in the
+ * table. It must also be of the same type as the table values.
+ * 
+ * @param <U> data type for {@code values()} output
+ */
+@Operator
+public final class LookupTableFind<U extends TType> extends RawOp implements Operand<U> {
+  
+  /**
+   * Factory method to create a class wrapping a new LookupTableFind operation.
+   * 
+   * @param scope current scope
+   * @param tableHandle Handle to the table.
+   * @param keys Any shape.  Keys to look up.
+   * @param defaultValue 
+   * @return a new instance of LookupTableFind
+   */
+  @Endpoint(describeByClass = true)
+  public static <U extends TType, T extends TType> LookupTableFind<U> create(Scope scope, Operand<?> tableHandle, Operand<T> keys, Operand<U> defaultValue) {
+    OperationBuilder opBuilder = scope.env().opBuilder("LookupTableFindV2", scope.makeOpName("LookupTableFind"));
+    opBuilder.addInput(tableHandle.asOutput());
+    opBuilder.addInput(keys.asOutput());
+    opBuilder.addInput(defaultValue.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new LookupTableFind<U>(opBuilder.build());
+  }
+  
+  /**
+   * Same shape as `keys`.  Values found in the table, or `default_values`
+   * for missing keys.
+   */
+  public Output<U> values() {
+    return values;
+  }
+  
+  @Override
+  public Output<U> asOutput() {
+    return values;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "LookupTableFindV2";
+  
+  private Output<U> values;
+  
+  private LookupTableFind(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    values = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LookupTableImport.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LookupTableImport.java
new file mode 100644
index 00000000000..9884a40e3cb
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LookupTableImport.java
@@ -0,0 +1,63 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Replaces the contents of the table with the specified keys and values.
+ * <p>
+ * The tensor `keys` must be of the same type as the keys of the table.
+ * The tensor `values` must be of the type of the table values.
+ */
+@Operator
+public final class LookupTableImport extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new LookupTableImport operation.
+   * 
+   * @param scope current scope
+   * @param tableHandle Handle to the table.
+   * @param keys Any shape.  Keys to look up.
+   * @param values Values to associate with keys.
+   * @return a new instance of LookupTableImport
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType, U extends TType> LookupTableImport create(Scope scope, Operand<?> tableHandle, Operand<T> keys, Operand<U> values) {
+    OperationBuilder opBuilder = scope.env().opBuilder("LookupTableImportV2", scope.makeOpName("LookupTableImport"));
+    opBuilder.addInput(tableHandle.asOutput());
+    opBuilder.addInput(keys.asOutput());
+    opBuilder.addInput(values.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new LookupTableImport(opBuilder.build());
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "LookupTableImportV2";
+  
+  private LookupTableImport(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LookupTableInsert.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LookupTableInsert.java
new file mode 100644
index 00000000000..0f09ae25d1b
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LookupTableInsert.java
@@ -0,0 +1,63 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Updates the table to associates keys with values.
+ * <p>
+ * The tensor `keys` must be of the same type as the keys of the table.
+ * The tensor `values` must be of the type of the table values.
+ */
+@Operator
+public final class LookupTableInsert extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new LookupTableInsert operation.
+   * 
+   * @param scope current scope
+   * @param tableHandle Handle to the table.
+   * @param keys Any shape.  Keys to look up.
+   * @param values Values to associate with keys.
+   * @return a new instance of LookupTableInsert
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType, U extends TType> LookupTableInsert create(Scope scope, Operand<?> tableHandle, Operand<T> keys, Operand<U> values) {
+    OperationBuilder opBuilder = scope.env().opBuilder("LookupTableInsertV2", scope.makeOpName("LookupTableInsert"));
+    opBuilder.addInput(tableHandle.asOutput());
+    opBuilder.addInput(keys.asOutput());
+    opBuilder.addInput(values.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new LookupTableInsert(opBuilder.build());
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "LookupTableInsertV2";
+  
+  private LookupTableInsert(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LookupTableRemove.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LookupTableRemove.java
new file mode 100644
index 00000000000..41463ad7539
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LookupTableRemove.java
@@ -0,0 +1,60 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Removes keys and its associated values from a table.
+ * <p>
+ * The tensor `keys` must of the same type as the keys of the table. Keys not
+ * already in the table are silently ignored.
+ */
+public final class LookupTableRemove extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new LookupTableRemove operation.
+   * 
+   * @param scope current scope
+   * @param tableHandle Handle to the table.
+   * @param keys Any shape.  Keys of the elements to remove.
+   * @return a new instance of LookupTableRemove
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> LookupTableRemove create(Scope scope, Operand<?> tableHandle, Operand<T> keys) {
+    OperationBuilder opBuilder = scope.env().opBuilder("LookupTableRemoveV2", scope.makeOpName("LookupTableRemove"));
+    opBuilder.addInput(tableHandle.asOutput());
+    opBuilder.addInput(keys.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new LookupTableRemove(opBuilder.build());
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "LookupTableRemoveV2";
+  
+  private LookupTableRemove(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LookupTableSize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LookupTableSize.java
new file mode 100644
index 00000000000..10e84953496
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LookupTableSize.java
@@ -0,0 +1,73 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt64;
+
+/**
+ * Computes the number of elements in the given table.
+ */
+@Operator
+public final class LookupTableSize extends RawOp implements Operand<TInt64> {
+  
+  /**
+   * Factory method to create a class wrapping a new LookupTableSize operation.
+   * 
+   * @param scope current scope
+   * @param tableHandle Handle to the table.
+   * @return a new instance of LookupTableSize
+   */
+  @Endpoint(describeByClass = true)
+  public static LookupTableSize create(Scope scope, Operand<?> tableHandle) {
+    OperationBuilder opBuilder = scope.env().opBuilder("LookupTableSizeV2", scope.makeOpName("LookupTableSize"));
+    opBuilder.addInput(tableHandle.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new LookupTableSize(opBuilder.build());
+  }
+  
+  /**
+   * Scalar that contains number of elements in the table.
+   */
+  public Output<TInt64> size() {
+    return size;
+  }
+  
+  @Override
+  public Output<TInt64> asOutput() {
+    return size;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "LookupTableSizeV2";
+  
+  private Output<TInt64> size;
+  
+  private LookupTableSize(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    size = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LoopCond.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LoopCond.java
new file mode 100644
index 00000000000..57b82598f84
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LoopCond.java
@@ -0,0 +1,76 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TBool;
+
+/**
+ * Forwards the input to the output.
+ * <p>
+ * This operator represents the loop termination condition used by the
+ * "pivot" switches of a loop.
+ */
+@Operator
+public final class LoopCond extends RawOp implements Operand<TBool> {
+  
+  /**
+   * Factory method to create a class wrapping a new LoopCond operation.
+   * 
+   * @param scope current scope
+   * @param input A boolean scalar, representing the branch predicate of the Switch op.
+   * @return a new instance of LoopCond
+   */
+  @Endpoint(describeByClass = true)
+  public static LoopCond create(Scope scope, Operand<TBool> input) {
+    OperationBuilder opBuilder = scope.env().opBuilder("LoopCond", scope.makeOpName("LoopCond"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new LoopCond(opBuilder.build());
+  }
+  
+  /**
+   * The same tensor as `input`.
+   */
+  public Output<TBool> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<TBool> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "LoopCond";
+  
+  private Output<TBool> output;
+  
+  private LoopCond(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LowerBound.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LowerBound.java
new file mode 100644
index 00000000000..7ec77e7df0b
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/LowerBound.java
@@ -0,0 +1,115 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Applies lower_bound(sorted_search_values, values) along each row.
+ * <p>
+ * Each set of rows with the same index in (sorted_inputs, values) is treated
+ * independently.  The resulting row is the equivalent of calling
+ * `np.searchsorted(sorted_inputs, values, side='left')`.
+ * <p>
+ * The result is not a global index to the entire
+ * `Tensor`, but rather just the index in the last dimension.
+ * <p>
+ * A 2-D example:
+ *   sorted_sequence = [[0, 3, 9, 9, 10],
+ *                      [1, 2, 3, 4, 5]]
+ *   values = [[2, 4, 9],
+ *             [0, 2, 6]]
+ * <p>
+ *   result = LowerBound(sorted_sequence, values)
+ * <p>
+ *   result == [[1, 2, 2],
+ *              [0, 1, 5]]
+ * 
+ * @param <U> data type for {@code output()} output
+ */
+public final class LowerBound<U extends TNumber> extends RawOp implements Operand<U> {
+  
+  /**
+   * Factory method to create a class wrapping a new LowerBound operation.
+   * 
+   * @param scope current scope
+   * @param sortedInputs 2-D Tensor where each row is ordered.
+   * @param values 2-D Tensor with the same numbers of rows as `sorted_search_values`. Contains
+   * the values that will be searched for in `sorted_search_values`.
+   * @param outType 
+   * @return a new instance of LowerBound
+   */
+  @Endpoint(describeByClass = true)
+  public static <U extends TNumber, T extends TType> LowerBound<U> create(Scope scope, Operand<T> sortedInputs, Operand<T> values, Class<U> outType) {
+    OperationBuilder opBuilder = scope.env().opBuilder("LowerBound", scope.makeOpName("LowerBound"));
+    opBuilder.addInput(sortedInputs.asOutput());
+    opBuilder.addInput(values.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("out_type", outType);
+    return new LowerBound<U>(opBuilder.build());
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new LowerBound operation using default output types.
+   * 
+   * @param scope current scope
+   * @param sortedInputs 2-D Tensor where each row is ordered.
+   * @param values 2-D Tensor with the same numbers of rows as `sorted_search_values`. Contains
+   * the values that will be searched for in `sorted_search_values`.
+   * @return a new instance of LowerBound
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> LowerBound<TInt32> create(Scope scope, Operand<T> sortedInputs, Operand<T> values) {
+    return create(scope, sortedInputs, values, TInt32.class);
+  }
+  
+  /**
+   * A `Tensor` with the same shape as `values`.  It contains the first scalar index
+   * into the last dimension where values can be inserted without changing the
+   * ordered property.
+   */
+  public Output<U> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<U> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "LowerBound";
+  
+  private Output<U> output;
+  
+  private LowerBound(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MapClear.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MapClear.java
new file mode 100644
index 00000000000..3c3dca46cc6
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MapClear.java
@@ -0,0 +1,151 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import java.util.List;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Op removes all elements in the underlying container.
+ */
+@Operator
+public final class MapClear extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.core.MapClear}
+   */
+  public static class Options {
+    
+    /**
+     * @param capacity 
+     */
+    public Options capacity(Long capacity) {
+      this.capacity = capacity;
+      return this;
+    }
+    
+    /**
+     * @param memoryLimit 
+     */
+    public Options memoryLimit(Long memoryLimit) {
+      this.memoryLimit = memoryLimit;
+      return this;
+    }
+    
+    /**
+     * @param container 
+     */
+    public Options container(String container) {
+      this.container = container;
+      return this;
+    }
+    
+    /**
+     * @param sharedName 
+     */
+    public Options sharedName(String sharedName) {
+      this.sharedName = sharedName;
+      return this;
+    }
+    
+    private Long capacity;
+    private Long memoryLimit;
+    private String container;
+    private String sharedName;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new MapClear operation.
+   * 
+   * @param scope current scope
+   * @param dtypes 
+   * @param options carries optional attributes values
+   * @return a new instance of MapClear
+   */
+  @Endpoint(describeByClass = true)
+  public static MapClear create(Scope scope, List<Class<? extends TType>> dtypes, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("MapClear", scope.makeOpName("MapClear"));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    Class[] dtypesArray = new Class[dtypes.size()];
+    for (int i = 0; i < dtypesArray.length; ++i) {
+      dtypesArray[i] = dtypes.get(i);
+    }
+    opBuilder.setAttr("dtypes", dtypesArray);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.capacity != null) {
+          opBuilder.setAttr("capacity", opts.capacity);
+        }
+        if (opts.memoryLimit != null) {
+          opBuilder.setAttr("memory_limit", opts.memoryLimit);
+        }
+        if (opts.container != null) {
+          opBuilder.setAttr("container", opts.container);
+        }
+        if (opts.sharedName != null) {
+          opBuilder.setAttr("shared_name", opts.sharedName);
+        }
+      }
+    }
+    return new MapClear(opBuilder.build());
+  }
+  
+  /**
+   * @param capacity 
+   */
+  public static Options capacity(Long capacity) {
+    return new Options().capacity(capacity);
+  }
+  
+  /**
+   * @param memoryLimit 
+   */
+  public static Options memoryLimit(Long memoryLimit) {
+    return new Options().memoryLimit(memoryLimit);
+  }
+  
+  /**
+   * @param container 
+   */
+  public static Options container(String container) {
+    return new Options().container(container);
+  }
+  
+  /**
+   * @param sharedName 
+   */
+  public static Options sharedName(String sharedName) {
+    return new Options().sharedName(sharedName);
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "MapClear";
+  
+  private MapClear(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MapIncompleteSize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MapIncompleteSize.java
new file mode 100644
index 00000000000..76b1513c6bd
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MapIncompleteSize.java
@@ -0,0 +1,169 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Op returns the number of incomplete elements in the underlying container.
+ */
+@Operator
+public final class MapIncompleteSize extends RawOp implements Operand<TInt32> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.core.MapIncompleteSize}
+   */
+  public static class Options {
+    
+    /**
+     * @param capacity 
+     */
+    public Options capacity(Long capacity) {
+      this.capacity = capacity;
+      return this;
+    }
+    
+    /**
+     * @param memoryLimit 
+     */
+    public Options memoryLimit(Long memoryLimit) {
+      this.memoryLimit = memoryLimit;
+      return this;
+    }
+    
+    /**
+     * @param container 
+     */
+    public Options container(String container) {
+      this.container = container;
+      return this;
+    }
+    
+    /**
+     * @param sharedName 
+     */
+    public Options sharedName(String sharedName) {
+      this.sharedName = sharedName;
+      return this;
+    }
+    
+    private Long capacity;
+    private Long memoryLimit;
+    private String container;
+    private String sharedName;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new MapIncompleteSize operation.
+   * 
+   * @param scope current scope
+   * @param dtypes 
+   * @param options carries optional attributes values
+   * @return a new instance of MapIncompleteSize
+   */
+  @Endpoint(describeByClass = true)
+  public static MapIncompleteSize create(Scope scope, List<Class<? extends TType>> dtypes, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("MapIncompleteSize", scope.makeOpName("MapIncompleteSize"));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    Class[] dtypesArray = new Class[dtypes.size()];
+    for (int i = 0; i < dtypesArray.length; ++i) {
+      dtypesArray[i] = dtypes.get(i);
+    }
+    opBuilder.setAttr("dtypes", dtypesArray);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.capacity != null) {
+          opBuilder.setAttr("capacity", opts.capacity);
+        }
+        if (opts.memoryLimit != null) {
+          opBuilder.setAttr("memory_limit", opts.memoryLimit);
+        }
+        if (opts.container != null) {
+          opBuilder.setAttr("container", opts.container);
+        }
+        if (opts.sharedName != null) {
+          opBuilder.setAttr("shared_name", opts.sharedName);
+        }
+      }
+    }
+    return new MapIncompleteSize(opBuilder.build());
+  }
+  
+  /**
+   * @param capacity 
+   */
+  public static Options capacity(Long capacity) {
+    return new Options().capacity(capacity);
+  }
+  
+  /**
+   * @param memoryLimit 
+   */
+  public static Options memoryLimit(Long memoryLimit) {
+    return new Options().memoryLimit(memoryLimit);
+  }
+  
+  /**
+   * @param container 
+   */
+  public static Options container(String container) {
+    return new Options().container(container);
+  }
+  
+  /**
+   * @param sharedName 
+   */
+  public static Options sharedName(String sharedName) {
+    return new Options().sharedName(sharedName);
+  }
+  
+  /**
+   */
+  public Output<TInt32> size() {
+    return size;
+  }
+  
+  @Override
+  public Output<TInt32> asOutput() {
+    return size;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "MapIncompleteSize";
+  
+  private Output<TInt32> size;
+  
+  private MapIncompleteSize(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    size = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MapPeek.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MapPeek.java
new file mode 100644
index 00000000000..46dccfc6a3f
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MapPeek.java
@@ -0,0 +1,182 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import java.util.Arrays;
+import java.util.Iterator;
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Op peeks at the values at the specified key.  If the
+ * <p>
+ * underlying container does not contain this key
+ * this op will block until it does.
+ */
+@Operator
+public final class MapPeek extends RawOp implements Iterable<Operand<TType>> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.core.MapPeek}
+   */
+  public static class Options {
+    
+    /**
+     * @param capacity 
+     */
+    public Options capacity(Long capacity) {
+      this.capacity = capacity;
+      return this;
+    }
+    
+    /**
+     * @param memoryLimit 
+     */
+    public Options memoryLimit(Long memoryLimit) {
+      this.memoryLimit = memoryLimit;
+      return this;
+    }
+    
+    /**
+     * @param container 
+     */
+    public Options container(String container) {
+      this.container = container;
+      return this;
+    }
+    
+    /**
+     * @param sharedName 
+     */
+    public Options sharedName(String sharedName) {
+      this.sharedName = sharedName;
+      return this;
+    }
+    
+    private Long capacity;
+    private Long memoryLimit;
+    private String container;
+    private String sharedName;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new MapPeek operation.
+   * 
+   * @param scope current scope
+   * @param key 
+   * @param indices 
+   * @param dtypes 
+   * @param options carries optional attributes values
+   * @return a new instance of MapPeek
+   */
+  @Endpoint(describeByClass = true)
+  public static MapPeek create(Scope scope, Operand<TInt64> key, Operand<TInt32> indices, List<Class<? extends TType>> dtypes, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("MapPeek", scope.makeOpName("MapPeek"));
+    opBuilder.addInput(key.asOutput());
+    opBuilder.addInput(indices.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    Class[] dtypesArray = new Class[dtypes.size()];
+    for (int i = 0; i < dtypesArray.length; ++i) {
+      dtypesArray[i] = dtypes.get(i);
+    }
+    opBuilder.setAttr("dtypes", dtypesArray);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.capacity != null) {
+          opBuilder.setAttr("capacity", opts.capacity);
+        }
+        if (opts.memoryLimit != null) {
+          opBuilder.setAttr("memory_limit", opts.memoryLimit);
+        }
+        if (opts.container != null) {
+          opBuilder.setAttr("container", opts.container);
+        }
+        if (opts.sharedName != null) {
+          opBuilder.setAttr("shared_name", opts.sharedName);
+        }
+      }
+    }
+    return new MapPeek(opBuilder.build());
+  }
+  
+  /**
+   * @param capacity 
+   */
+  public static Options capacity(Long capacity) {
+    return new Options().capacity(capacity);
+  }
+  
+  /**
+   * @param memoryLimit 
+   */
+  public static Options memoryLimit(Long memoryLimit) {
+    return new Options().memoryLimit(memoryLimit);
+  }
+  
+  /**
+   * @param container 
+   */
+  public static Options container(String container) {
+    return new Options().container(container);
+  }
+  
+  /**
+   * @param sharedName 
+   */
+  public static Options sharedName(String sharedName) {
+    return new Options().sharedName(sharedName);
+  }
+  
+  /**
+   */
+  public List<Output<?>> values() {
+    return values;
+  }
+  
+  @Override
+  @SuppressWarnings({"rawtypes", "unchecked"})
+  public Iterator<Operand<TType>> iterator() {
+    return (Iterator) values.iterator();
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "MapPeek";
+  
+  private List<Output<?>> values;
+  
+  private MapPeek(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    int valuesLength = operation.outputListLength("values");
+    values = Arrays.asList(operation.outputList(outputIdx, valuesLength));
+    outputIdx += valuesLength;
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MapSize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MapSize.java
new file mode 100644
index 00000000000..600b9473da9
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MapSize.java
@@ -0,0 +1,169 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Op returns the number of elements in the underlying container.
+ */
+@Operator
+public final class MapSize extends RawOp implements Operand<TInt32> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.core.MapSize}
+   */
+  public static class Options {
+    
+    /**
+     * @param capacity 
+     */
+    public Options capacity(Long capacity) {
+      this.capacity = capacity;
+      return this;
+    }
+    
+    /**
+     * @param memoryLimit 
+     */
+    public Options memoryLimit(Long memoryLimit) {
+      this.memoryLimit = memoryLimit;
+      return this;
+    }
+    
+    /**
+     * @param container 
+     */
+    public Options container(String container) {
+      this.container = container;
+      return this;
+    }
+    
+    /**
+     * @param sharedName 
+     */
+    public Options sharedName(String sharedName) {
+      this.sharedName = sharedName;
+      return this;
+    }
+    
+    private Long capacity;
+    private Long memoryLimit;
+    private String container;
+    private String sharedName;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new MapSize operation.
+   * 
+   * @param scope current scope
+   * @param dtypes 
+   * @param options carries optional attributes values
+   * @return a new instance of MapSize
+   */
+  @Endpoint(describeByClass = true)
+  public static MapSize create(Scope scope, List<Class<? extends TType>> dtypes, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("MapSize", scope.makeOpName("MapSize"));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    Class[] dtypesArray = new Class[dtypes.size()];
+    for (int i = 0; i < dtypesArray.length; ++i) {
+      dtypesArray[i] = dtypes.get(i);
+    }
+    opBuilder.setAttr("dtypes", dtypesArray);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.capacity != null) {
+          opBuilder.setAttr("capacity", opts.capacity);
+        }
+        if (opts.memoryLimit != null) {
+          opBuilder.setAttr("memory_limit", opts.memoryLimit);
+        }
+        if (opts.container != null) {
+          opBuilder.setAttr("container", opts.container);
+        }
+        if (opts.sharedName != null) {
+          opBuilder.setAttr("shared_name", opts.sharedName);
+        }
+      }
+    }
+    return new MapSize(opBuilder.build());
+  }
+  
+  /**
+   * @param capacity 
+   */
+  public static Options capacity(Long capacity) {
+    return new Options().capacity(capacity);
+  }
+  
+  /**
+   * @param memoryLimit 
+   */
+  public static Options memoryLimit(Long memoryLimit) {
+    return new Options().memoryLimit(memoryLimit);
+  }
+  
+  /**
+   * @param container 
+   */
+  public static Options container(String container) {
+    return new Options().container(container);
+  }
+  
+  /**
+   * @param sharedName 
+   */
+  public static Options sharedName(String sharedName) {
+    return new Options().sharedName(sharedName);
+  }
+  
+  /**
+   */
+  public Output<TInt32> size() {
+    return size;
+  }
+  
+  @Override
+  public Output<TInt32> asOutput() {
+    return size;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "MapSize";
+  
+  private Output<TInt32> size;
+  
+  private MapSize(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    size = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MapStage.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MapStage.java
new file mode 100644
index 00000000000..eb75d99ba5c
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MapStage.java
@@ -0,0 +1,166 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.Operands;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Stage (key, values) in the underlying container which behaves like a hashtable.
+ */
+@Operator
+public final class MapStage extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.core.MapStage}
+   */
+  public static class Options {
+    
+    /**
+     * @param capacity Maximum number of elements in the Staging Area. If > 0, inserts
+     * on the container will block when the capacity is reached.
+     */
+    public Options capacity(Long capacity) {
+      this.capacity = capacity;
+      return this;
+    }
+    
+    /**
+     * @param memoryLimit 
+     */
+    public Options memoryLimit(Long memoryLimit) {
+      this.memoryLimit = memoryLimit;
+      return this;
+    }
+    
+    /**
+     * @param container If non-empty, this queue is placed in the given container. Otherwise,
+     * a default container is used.
+     */
+    public Options container(String container) {
+      this.container = container;
+      return this;
+    }
+    
+    /**
+     * @param sharedName It is necessary to match this name to the matching Unstage Op.
+     */
+    public Options sharedName(String sharedName) {
+      this.sharedName = sharedName;
+      return this;
+    }
+    
+    private Long capacity;
+    private Long memoryLimit;
+    private String container;
+    private String sharedName;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new MapStage operation.
+   * 
+   * @param scope current scope
+   * @param key int64
+   * @param indices 
+   * @param values a list of tensors
+   * dtypes A list of data types that inserted values should adhere to.
+   * @param dtypes 
+   * @param options carries optional attributes values
+   * @return a new instance of MapStage
+   */
+  @Endpoint(describeByClass = true)
+  public static MapStage create(Scope scope, Operand<TInt64> key, Operand<TInt32> indices, Iterable<Operand<?>> values, List<Class<? extends TType>> dtypes, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("MapStage", scope.makeOpName("MapStage"));
+    opBuilder.addInput(key.asOutput());
+    opBuilder.addInput(indices.asOutput());
+    opBuilder.addInputList(Operands.asOutputs(values));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    Class[] dtypesArray = new Class[dtypes.size()];
+    for (int i = 0; i < dtypesArray.length; ++i) {
+      dtypesArray[i] = dtypes.get(i);
+    }
+    opBuilder.setAttr("dtypes", dtypesArray);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.capacity != null) {
+          opBuilder.setAttr("capacity", opts.capacity);
+        }
+        if (opts.memoryLimit != null) {
+          opBuilder.setAttr("memory_limit", opts.memoryLimit);
+        }
+        if (opts.container != null) {
+          opBuilder.setAttr("container", opts.container);
+        }
+        if (opts.sharedName != null) {
+          opBuilder.setAttr("shared_name", opts.sharedName);
+        }
+      }
+    }
+    return new MapStage(opBuilder.build());
+  }
+  
+  /**
+   * @param capacity Maximum number of elements in the Staging Area. If > 0, inserts
+   * on the container will block when the capacity is reached.
+   */
+  public static Options capacity(Long capacity) {
+    return new Options().capacity(capacity);
+  }
+  
+  /**
+   * @param memoryLimit 
+   */
+  public static Options memoryLimit(Long memoryLimit) {
+    return new Options().memoryLimit(memoryLimit);
+  }
+  
+  /**
+   * @param container If non-empty, this queue is placed in the given container. Otherwise,
+   * a default container is used.
+   */
+  public static Options container(String container) {
+    return new Options().container(container);
+  }
+  
+  /**
+   * @param sharedName It is necessary to match this name to the matching Unstage Op.
+   */
+  public static Options sharedName(String sharedName) {
+    return new Options().sharedName(sharedName);
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "MapStage";
+  
+  private MapStage(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MapUnstage.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MapUnstage.java
new file mode 100644
index 00000000000..ba02210d5c4
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MapUnstage.java
@@ -0,0 +1,182 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import java.util.Arrays;
+import java.util.Iterator;
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Op removes and returns the values associated with the key
+ * <p>
+ * from the underlying container.   If the underlying container
+ * does not contain this key, the op will block until it does.
+ */
+@Operator
+public final class MapUnstage extends RawOp implements Iterable<Operand<TType>> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.core.MapUnstage}
+   */
+  public static class Options {
+    
+    /**
+     * @param capacity 
+     */
+    public Options capacity(Long capacity) {
+      this.capacity = capacity;
+      return this;
+    }
+    
+    /**
+     * @param memoryLimit 
+     */
+    public Options memoryLimit(Long memoryLimit) {
+      this.memoryLimit = memoryLimit;
+      return this;
+    }
+    
+    /**
+     * @param container 
+     */
+    public Options container(String container) {
+      this.container = container;
+      return this;
+    }
+    
+    /**
+     * @param sharedName 
+     */
+    public Options sharedName(String sharedName) {
+      this.sharedName = sharedName;
+      return this;
+    }
+    
+    private Long capacity;
+    private Long memoryLimit;
+    private String container;
+    private String sharedName;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new MapUnstage operation.
+   * 
+   * @param scope current scope
+   * @param key 
+   * @param indices 
+   * @param dtypes 
+   * @param options carries optional attributes values
+   * @return a new instance of MapUnstage
+   */
+  @Endpoint(describeByClass = true)
+  public static MapUnstage create(Scope scope, Operand<TInt64> key, Operand<TInt32> indices, List<Class<? extends TType>> dtypes, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("MapUnstage", scope.makeOpName("MapUnstage"));
+    opBuilder.addInput(key.asOutput());
+    opBuilder.addInput(indices.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    Class[] dtypesArray = new Class[dtypes.size()];
+    for (int i = 0; i < dtypesArray.length; ++i) {
+      dtypesArray[i] = dtypes.get(i);
+    }
+    opBuilder.setAttr("dtypes", dtypesArray);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.capacity != null) {
+          opBuilder.setAttr("capacity", opts.capacity);
+        }
+        if (opts.memoryLimit != null) {
+          opBuilder.setAttr("memory_limit", opts.memoryLimit);
+        }
+        if (opts.container != null) {
+          opBuilder.setAttr("container", opts.container);
+        }
+        if (opts.sharedName != null) {
+          opBuilder.setAttr("shared_name", opts.sharedName);
+        }
+      }
+    }
+    return new MapUnstage(opBuilder.build());
+  }
+  
+  /**
+   * @param capacity 
+   */
+  public static Options capacity(Long capacity) {
+    return new Options().capacity(capacity);
+  }
+  
+  /**
+   * @param memoryLimit 
+   */
+  public static Options memoryLimit(Long memoryLimit) {
+    return new Options().memoryLimit(memoryLimit);
+  }
+  
+  /**
+   * @param container 
+   */
+  public static Options container(String container) {
+    return new Options().container(container);
+  }
+  
+  /**
+   * @param sharedName 
+   */
+  public static Options sharedName(String sharedName) {
+    return new Options().sharedName(sharedName);
+  }
+  
+  /**
+   */
+  public List<Output<?>> values() {
+    return values;
+  }
+  
+  @Override
+  @SuppressWarnings({"rawtypes", "unchecked"})
+  public Iterator<Operand<TType>> iterator() {
+    return (Iterator) values.iterator();
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "MapUnstage";
+  
+  private List<Output<?>> values;
+  
+  private MapUnstage(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    int valuesLength = operation.outputListLength("values");
+    values = Arrays.asList(operation.outputList(outputIdx, valuesLength));
+    outputIdx += valuesLength;
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MapUnstageNoKey.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MapUnstageNoKey.java
new file mode 100644
index 00000000000..f6cfe6c1d73
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MapUnstageNoKey.java
@@ -0,0 +1,181 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import java.util.Arrays;
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Op removes and returns a random (key, value)
+ * <p>
+ * from the underlying container.   If the underlying container
+ * does not contain elements, the op will block until it does.
+ */
+@Operator
+public final class MapUnstageNoKey extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.core.MapUnstageNoKey}
+   */
+  public static class Options {
+    
+    /**
+     * @param capacity 
+     */
+    public Options capacity(Long capacity) {
+      this.capacity = capacity;
+      return this;
+    }
+    
+    /**
+     * @param memoryLimit 
+     */
+    public Options memoryLimit(Long memoryLimit) {
+      this.memoryLimit = memoryLimit;
+      return this;
+    }
+    
+    /**
+     * @param container 
+     */
+    public Options container(String container) {
+      this.container = container;
+      return this;
+    }
+    
+    /**
+     * @param sharedName 
+     */
+    public Options sharedName(String sharedName) {
+      this.sharedName = sharedName;
+      return this;
+    }
+    
+    private Long capacity;
+    private Long memoryLimit;
+    private String container;
+    private String sharedName;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new MapUnstageNoKey operation.
+   * 
+   * @param scope current scope
+   * @param indices 
+   * @param dtypes 
+   * @param options carries optional attributes values
+   * @return a new instance of MapUnstageNoKey
+   */
+  @Endpoint(describeByClass = true)
+  public static MapUnstageNoKey create(Scope scope, Operand<TInt32> indices, List<Class<? extends TType>> dtypes, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("MapUnstageNoKey", scope.makeOpName("MapUnstageNoKey"));
+    opBuilder.addInput(indices.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    Class[] dtypesArray = new Class[dtypes.size()];
+    for (int i = 0; i < dtypesArray.length; ++i) {
+      dtypesArray[i] = dtypes.get(i);
+    }
+    opBuilder.setAttr("dtypes", dtypesArray);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.capacity != null) {
+          opBuilder.setAttr("capacity", opts.capacity);
+        }
+        if (opts.memoryLimit != null) {
+          opBuilder.setAttr("memory_limit", opts.memoryLimit);
+        }
+        if (opts.container != null) {
+          opBuilder.setAttr("container", opts.container);
+        }
+        if (opts.sharedName != null) {
+          opBuilder.setAttr("shared_name", opts.sharedName);
+        }
+      }
+    }
+    return new MapUnstageNoKey(opBuilder.build());
+  }
+  
+  /**
+   * @param capacity 
+   */
+  public static Options capacity(Long capacity) {
+    return new Options().capacity(capacity);
+  }
+  
+  /**
+   * @param memoryLimit 
+   */
+  public static Options memoryLimit(Long memoryLimit) {
+    return new Options().memoryLimit(memoryLimit);
+  }
+  
+  /**
+   * @param container 
+   */
+  public static Options container(String container) {
+    return new Options().container(container);
+  }
+  
+  /**
+   * @param sharedName 
+   */
+  public static Options sharedName(String sharedName) {
+    return new Options().sharedName(sharedName);
+  }
+  
+  /**
+   */
+  public Output<TInt64> key() {
+    return key;
+  }
+  
+  /**
+   */
+  public List<Output<?>> values() {
+    return values;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "MapUnstageNoKey";
+  
+  private Output<TInt64> key;
+  private List<Output<?>> values;
+  
+  private MapUnstageNoKey(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    key = operation.output(outputIdx++);
+    int valuesLength = operation.outputListLength("values");
+    values = Arrays.asList(operation.outputList(outputIdx, valuesLength));
+    outputIdx += valuesLength;
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Max.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Max.java
new file mode 100644
index 00000000000..94bcd0f544c
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Max.java
@@ -0,0 +1,118 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Computes the maximum of elements across dimensions of a tensor.
+ * <p>
+ * Reduces `input` along the dimensions given in `axis`. Unless
+ * `keep_dims` is true, the rank of the tensor is reduced by 1 for each entry in
+ * `axis`. If `keep_dims` is true, the reduced dimensions are
+ * retained with length 1.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator
+public final class Max<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.core.Max}
+   */
+  public static class Options {
+    
+    /**
+     * @param keepDims If true, retain reduced dimensions with length 1.
+     */
+    public Options keepDims(Boolean keepDims) {
+      this.keepDims = keepDims;
+      return this;
+    }
+    
+    private Boolean keepDims;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new Max operation.
+   * 
+   * @param scope current scope
+   * @param input The tensor to reduce.
+   * @param axis The dimensions to reduce. Must be in the range
+   * `[-rank(input), rank(input))`.
+   * @param options carries optional attributes values
+   * @return a new instance of Max
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType, U extends TNumber> Max<T> create(Scope scope, Operand<T> input, Operand<U> axis, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Max", scope.makeOpName("Max"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(axis.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.keepDims != null) {
+          opBuilder.setAttr("keep_dims", opts.keepDims);
+        }
+      }
+    }
+    return new Max<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param keepDims If true, retain reduced dimensions with length 1.
+   */
+  public static Options keepDims(Boolean keepDims) {
+    return new Options().keepDims(keepDims);
+  }
+  
+  /**
+   * The reduced tensor.
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Max";
+  
+  private Output<T> output;
+  
+  private Max(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Merge.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Merge.java
new file mode 100644
index 00000000000..41eca5fc33f
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Merge.java
@@ -0,0 +1,87 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.Operands;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Forwards the value of an available tensor from `inputs` to `output`.
+ * <p>
+ * `Merge` waits for at least one of the tensors in `inputs` to become available.
+ * It is usually combined with `Switch` to implement branching.
+ * <p>
+ * `Merge` forwards the first tensor to become available to `output`, and sets
+ * `value_index` to its index in `inputs`.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator
+public final class Merge<T extends TType> extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new Merge operation.
+   * 
+   * @param scope current scope
+   * @param inputs The input tensors, exactly one of which will become available.
+   * @return a new instance of Merge
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> Merge<T> create(Scope scope, Iterable<Operand<T>> inputs) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Merge", scope.makeOpName("Merge"));
+    opBuilder.addInputList(Operands.asOutputs(inputs));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new Merge<T>(opBuilder.build());
+  }
+  
+  /**
+   * Will be set to the available input tensor.
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  /**
+   * The index of the chosen input tensor in `inputs`.
+   */
+  public Output<TInt32> valueIndex() {
+    return valueIndex;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Merge";
+  
+  private Output<T> output;
+  private Output<TInt32> valueIndex;
+  
+  private Merge(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+    valueIndex = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Min.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Min.java
new file mode 100644
index 00000000000..ed8ea366e7a
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Min.java
@@ -0,0 +1,118 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Computes the minimum of elements across dimensions of a tensor.
+ * <p>
+ * Reduces `input` along the dimensions given in `axis`. Unless
+ * `keep_dims` is true, the rank of the tensor is reduced by 1 for each entry in
+ * `axis`. If `keep_dims` is true, the reduced dimensions are
+ * retained with length 1.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator
+public final class Min<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.core.Min}
+   */
+  public static class Options {
+    
+    /**
+     * @param keepDims If true, retain reduced dimensions with length 1.
+     */
+    public Options keepDims(Boolean keepDims) {
+      this.keepDims = keepDims;
+      return this;
+    }
+    
+    private Boolean keepDims;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new Min operation.
+   * 
+   * @param scope current scope
+   * @param input The tensor to reduce.
+   * @param axis The dimensions to reduce. Must be in the range
+   * `[-rank(input), rank(input))`.
+   * @param options carries optional attributes values
+   * @return a new instance of Min
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType, U extends TNumber> Min<T> create(Scope scope, Operand<T> input, Operand<U> axis, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Min", scope.makeOpName("Min"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(axis.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.keepDims != null) {
+          opBuilder.setAttr("keep_dims", opts.keepDims);
+        }
+      }
+    }
+    return new Min<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param keepDims If true, retain reduced dimensions with length 1.
+   */
+  public static Options keepDims(Boolean keepDims) {
+    return new Options().keepDims(keepDims);
+  }
+  
+  /**
+   * The reduced tensor.
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Min";
+  
+  private Output<T> output;
+  
+  private Min(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MirrorPad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MirrorPad.java
new file mode 100644
index 00000000000..6c8292a671a
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MirrorPad.java
@@ -0,0 +1,111 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Pads a tensor with mirrored values.
+ * <p>
+ * This operation pads a `input` with mirrored values according to the `paddings`
+ * you specify. `paddings` is an integer tensor with shape `[n, 2]`, where n is
+ * the rank of `input`. For each dimension D of `input`, `paddings[D, 0]` indicates
+ * how many values to add before the contents of `input` in that dimension, and
+ * `paddings[D, 1]` indicates how many values to add after the contents of `input`
+ * in that dimension. Both `paddings[D, 0]` and `paddings[D, 1]` must be no greater
+ * than `input.dim_size(D)` (or `input.dim_size(D) - 1`) if `copy_border` is true
+ * (if false, respectively).
+ * <p>
+ * The padded size of each dimension D of the output is:
+ * <p>
+ * `paddings(D, 0) + input.dim_size(D) + paddings(D, 1)`
+ * <p>
+ * For example:
+ * <pre>{@code
+ * # 't' is [[1, 2, 3], [4, 5, 6]].
+ * # 'paddings' is [[1, 1]], [2, 2]].
+ * # 'mode' is SYMMETRIC.
+ * # rank of 't' is 2.
+ * pad(t, paddings) ==> [[2, 1, 1, 2, 3, 3, 2]
+ *                       [2, 1, 1, 2, 3, 3, 2]
+ *                       [5, 4, 4, 5, 6, 6, 5]
+ *                       [5, 4, 4, 5, 6, 6, 5]]
+ * }</pre>
+ * 
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator
+public final class MirrorPad<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new MirrorPad operation.
+   * 
+   * @param scope current scope
+   * @param input The input tensor to be padded.
+   * @param paddings A two-column matrix specifying the padding sizes. The number of
+   * rows must be the same as the rank of `input`.
+   * @param mode Either `REFLECT` or `SYMMETRIC`. In reflect mode the padded regions
+   * do not include the borders, while in symmetric mode the padded regions
+   * do include the borders. For example, if `input` is `[1, 2, 3]` and `paddings`
+   * is `[0, 2]`, then the output is `[1, 2, 3, 2, 1]` in reflect mode, and
+   * it is `[1, 2, 3, 3, 2]` in symmetric mode.
+   * @return a new instance of MirrorPad
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType, U extends TNumber> MirrorPad<T> create(Scope scope, Operand<T> input, Operand<U> paddings, String mode) {
+    OperationBuilder opBuilder = scope.env().opBuilder("MirrorPad", scope.makeOpName("MirrorPad"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(paddings.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("mode", mode);
+    return new MirrorPad<T>(opBuilder.build());
+  }
+  
+  /**
+   * The padded tensor.
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "MirrorPad";
+  
+  private Output<T> output;
+  
+  private MirrorPad(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MirrorPadGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MirrorPadGrad.java
new file mode 100644
index 00000000000..7b1727c3fb5
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MirrorPadGrad.java
@@ -0,0 +1,99 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Gradient op for `MirrorPad` op. This op folds a mirror-padded tensor.
+ * <p>
+ * This operation folds the padded areas of `input` by `MirrorPad` according to the
+ * `paddings` you specify. `paddings` must be the same as `paddings` argument
+ * given to the corresponding `MirrorPad` op.
+ * <p>
+ * The folded size of each dimension D of the output is:
+ * <p>
+ * `input.dim_size(D) - paddings(D, 0) - paddings(D, 1)`
+ * <p>
+ * For example:
+ * <pre>{@code
+ * # 't' is [[1, 2, 3], [4, 5, 6], [7, 8, 9]].
+ * # 'paddings' is [[0, 1]], [0, 1]].
+ * # 'mode' is SYMMETRIC.
+ * # rank of 't' is 2.
+ * pad(t, paddings) ==> [[ 1,  5]
+ *                       [11, 28]]
+ * }</pre>
+ * 
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+public final class MirrorPadGrad<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new MirrorPadGrad operation.
+   * 
+   * @param scope current scope
+   * @param input The input tensor to be folded.
+   * @param paddings A two-column matrix specifying the padding sizes. The number of
+   * rows must be the same as the rank of `input`.
+   * @param mode The mode used in the `MirrorPad` op.
+   * @return a new instance of MirrorPadGrad
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType, U extends TNumber> MirrorPadGrad<T> create(Scope scope, Operand<T> input, Operand<U> paddings, String mode) {
+    OperationBuilder opBuilder = scope.env().opBuilder("MirrorPadGrad", scope.makeOpName("MirrorPadGrad"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(paddings.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("mode", mode);
+    return new MirrorPadGrad<T>(opBuilder.build());
+  }
+  
+  /**
+   * The folded tensor.
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "MirrorPadGrad";
+  
+  private Output<T> output;
+  
+  private MirrorPadGrad(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MlirPassthroughOp.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MlirPassthroughOp.java
new file mode 100644
index 00000000000..027c75bf163
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MlirPassthroughOp.java
@@ -0,0 +1,117 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import java.util.Arrays;
+import java.util.Iterator;
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.Operands;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Wraps an arbitrary MLIR computation expressed as a module with a main() function.
+ * <p>
+ * This operation does not have an associated kernel and is not intended to be
+ * executed in a regular TensorFlow session. Instead it is intended to be used for
+ * testing or for special case where a user intends to pass custom MLIR computation
+ * through a TensorFlow graph with the intent of having custom tooling processing
+ * it downstream (when targeting a different environment, like TensorFlow lite for
+ * example).
+ * The MLIR module is expected to have a main() function that will be used as an
+ * entry point. The inputs to the operations will be passed as argument to the
+ * main() function and the returned values of the main function mapped to the
+ * outputs.
+ * Example usage:
+ * <pre>{@code
+ * import tensorflow as tf
+ * from tensorflow.compiler.mlir.tensorflow.gen_mlir_passthrough_op import mlir_passthrough_op
+ * 
+ * mlir_module = '''python
+ * func @main(%arg0 : tensor<10xf32>, %arg1 : tensor<10xf32>) -> tensor<10x10xf32> {
+ *    %add = "magic.op"(%arg0, %arg1) : (tensor<10xf32>, tensor<10xf32>) -> tensor<10x10xf32>
+ *    return %ret : tensor<10x10xf32>
+ * }
+ * '''
+ * 
+ * @tf.function
+ * def foo(x, y):
+ *   return mlir_passthrough_op([x, y], mlir_module, Toutputs=[tf.float32])
+ * 
+ * graph_def = foo.get_concrete_function(tf.TensorSpec([10], tf.float32), tf.TensorSpec([10], tf.float32)).graph.as_graph_def()
+ * }</pre>
+ * 
+ */
+@Operator
+public final class MlirPassthroughOp extends RawOp implements Iterable<Operand<TType>> {
+  
+  /**
+   * Factory method to create a class wrapping a new MlirPassthroughOp operation.
+   * 
+   * @param scope current scope
+   * @param inputs 
+   * @param mlirModule 
+   * @param Toutputs 
+   * @return a new instance of MlirPassthroughOp
+   */
+  @Endpoint(describeByClass = true)
+  public static MlirPassthroughOp create(Scope scope, Iterable<Operand<?>> inputs, String mlirModule, List<Class<? extends TType>> Toutputs) {
+    OperationBuilder opBuilder = scope.env().opBuilder("MlirPassthroughOp", scope.makeOpName("MlirPassthroughOp"));
+    opBuilder.addInputList(Operands.asOutputs(inputs));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("mlir_module", mlirModule);
+    Class[] ToutputsArray = new Class[Toutputs.size()];
+    for (int i = 0; i < ToutputsArray.length; ++i) {
+      ToutputsArray[i] = Toutputs.get(i);
+    }
+    opBuilder.setAttr("Toutputs", ToutputsArray);
+    return new MlirPassthroughOp(opBuilder.build());
+  }
+  
+  /**
+   */
+  public List<Output<?>> outputs() {
+    return outputs;
+  }
+  
+  @Override
+  @SuppressWarnings({"rawtypes", "unchecked"})
+  public Iterator<Operand<TType>> iterator() {
+    return (Iterator) outputs.iterator();
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "MlirPassthroughOp";
+  
+  private List<Output<?>> outputs;
+  
+  private MlirPassthroughOp(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    int outputsLength = operation.outputListLength("outputs");
+    outputs = Arrays.asList(operation.outputList(outputIdx, outputsLength));
+    outputIdx += outputsLength;
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MutableDenseHashTable.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MutableDenseHashTable.java
new file mode 100644
index 00000000000..0e6b35f502a
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MutableDenseHashTable.java
@@ -0,0 +1,224 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Creates an empty hash table that uses tensors as the backing store.
+ * <p>
+ * It uses "open addressing" with quadratic reprobing to resolve
+ * collisions.
+ * <p>
+ * This op creates a mutable hash table, specifying the type of its keys and
+ * values. Each value must be a scalar. Data can be inserted into the table using
+ * the insert operations. It does not support the initialization operation.
+ */
+@Operator
+public final class MutableDenseHashTable extends RawOp implements Operand<TType> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.core.MutableDenseHashTable}
+   */
+  public static class Options {
+    
+    /**
+     * @param container If non-empty, this table is placed in the given container.
+     * Otherwise, a default container is used.
+     */
+    public Options container(String container) {
+      this.container = container;
+      return this;
+    }
+    
+    /**
+     * @param sharedName If non-empty, this table is shared under the given name across
+     * multiple sessions.
+     */
+    public Options sharedName(String sharedName) {
+      this.sharedName = sharedName;
+      return this;
+    }
+    
+    /**
+     * @param useNodeNameSharing 
+     */
+    public Options useNodeNameSharing(Boolean useNodeNameSharing) {
+      this.useNodeNameSharing = useNodeNameSharing;
+      return this;
+    }
+    
+    /**
+     * @param valueShape The shape of each value.
+     */
+    public Options valueShape(Shape valueShape) {
+      this.valueShape = valueShape;
+      return this;
+    }
+    
+    /**
+     * @param initialNumBuckets The initial number of hash table buckets. Must be a power
+     * to 2.
+     */
+    public Options initialNumBuckets(Long initialNumBuckets) {
+      this.initialNumBuckets = initialNumBuckets;
+      return this;
+    }
+    
+    /**
+     * @param maxLoadFactor The maximum ratio between number of entries and number of
+     * buckets before growing the table. Must be between 0 and 1.
+     */
+    public Options maxLoadFactor(Float maxLoadFactor) {
+      this.maxLoadFactor = maxLoadFactor;
+      return this;
+    }
+    
+    private String container;
+    private String sharedName;
+    private Boolean useNodeNameSharing;
+    private Shape valueShape;
+    private Long initialNumBuckets;
+    private Float maxLoadFactor;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new MutableDenseHashTable operation.
+   * 
+   * @param scope current scope
+   * @param emptyKey The key used to represent empty key buckets internally. Must not
+   * be used in insert or lookup operations.
+   * @param deletedKey 
+   * @param valueDtype Type of the table values.
+   * @param options carries optional attributes values
+   * @return a new instance of MutableDenseHashTable
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType, U extends TType> MutableDenseHashTable create(Scope scope, Operand<T> emptyKey, Operand<T> deletedKey, Class<U> valueDtype, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("MutableDenseHashTableV2", scope.makeOpName("MutableDenseHashTable"));
+    opBuilder.addInput(emptyKey.asOutput());
+    opBuilder.addInput(deletedKey.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("value_dtype", valueDtype);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.container != null) {
+          opBuilder.setAttr("container", opts.container);
+        }
+        if (opts.sharedName != null) {
+          opBuilder.setAttr("shared_name", opts.sharedName);
+        }
+        if (opts.useNodeNameSharing != null) {
+          opBuilder.setAttr("use_node_name_sharing", opts.useNodeNameSharing);
+        }
+        if (opts.valueShape != null) {
+          opBuilder.setAttr("value_shape", opts.valueShape);
+        }
+        if (opts.initialNumBuckets != null) {
+          opBuilder.setAttr("initial_num_buckets", opts.initialNumBuckets);
+        }
+        if (opts.maxLoadFactor != null) {
+          opBuilder.setAttr("max_load_factor", opts.maxLoadFactor);
+        }
+      }
+    }
+    return new MutableDenseHashTable(opBuilder.build());
+  }
+  
+  /**
+   * @param container If non-empty, this table is placed in the given container.
+   * Otherwise, a default container is used.
+   */
+  public static Options container(String container) {
+    return new Options().container(container);
+  }
+  
+  /**
+   * @param sharedName If non-empty, this table is shared under the given name across
+   * multiple sessions.
+   */
+  public static Options sharedName(String sharedName) {
+    return new Options().sharedName(sharedName);
+  }
+  
+  /**
+   * @param useNodeNameSharing 
+   */
+  public static Options useNodeNameSharing(Boolean useNodeNameSharing) {
+    return new Options().useNodeNameSharing(useNodeNameSharing);
+  }
+  
+  /**
+   * @param valueShape The shape of each value.
+   */
+  public static Options valueShape(Shape valueShape) {
+    return new Options().valueShape(valueShape);
+  }
+  
+  /**
+   * @param initialNumBuckets The initial number of hash table buckets. Must be a power
+   * to 2.
+   */
+  public static Options initialNumBuckets(Long initialNumBuckets) {
+    return new Options().initialNumBuckets(initialNumBuckets);
+  }
+  
+  /**
+   * @param maxLoadFactor The maximum ratio between number of entries and number of
+   * buckets before growing the table. Must be between 0 and 1.
+   */
+  public static Options maxLoadFactor(Float maxLoadFactor) {
+    return new Options().maxLoadFactor(maxLoadFactor);
+  }
+  
+  /**
+   * Handle to a table.
+   */
+  public Output<?> tableHandle() {
+    return tableHandle;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) tableHandle;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "MutableDenseHashTableV2";
+  
+  private Output<?> tableHandle;
+  
+  private MutableDenseHashTable(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    tableHandle = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MutableHashTable.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MutableHashTable.java
new file mode 100644
index 00000000000..0e2dd872331
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MutableHashTable.java
@@ -0,0 +1,158 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Creates an empty hash table.
+ * <p>
+ * This op creates a mutable hash table, specifying the type of its keys and
+ * values. Each value must be a scalar. Data can be inserted into the table using
+ * the insert operations. It does not support the initialization operation.
+ */
+@Operator
+public final class MutableHashTable extends RawOp implements Operand<TType> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.core.MutableHashTable}
+   */
+  public static class Options {
+    
+    /**
+     * @param container If non-empty, this table is placed in the given container.
+     * Otherwise, a default container is used.
+     */
+    public Options container(String container) {
+      this.container = container;
+      return this;
+    }
+    
+    /**
+     * @param sharedName If non-empty, this table is shared under the given name across
+     * multiple sessions.
+     */
+    public Options sharedName(String sharedName) {
+      this.sharedName = sharedName;
+      return this;
+    }
+    
+    /**
+     * @param useNodeNameSharing If true and shared_name is empty, the table is shared
+     * using the node name.
+     */
+    public Options useNodeNameSharing(Boolean useNodeNameSharing) {
+      this.useNodeNameSharing = useNodeNameSharing;
+      return this;
+    }
+    
+    private String container;
+    private String sharedName;
+    private Boolean useNodeNameSharing;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new MutableHashTable operation.
+   * 
+   * @param scope current scope
+   * @param keyDtype Type of the table keys.
+   * @param valueDtype Type of the table values.
+   * @param options carries optional attributes values
+   * @return a new instance of MutableHashTable
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType, U extends TType> MutableHashTable create(Scope scope, Class<T> keyDtype, Class<U> valueDtype, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("MutableHashTableV2", scope.makeOpName("MutableHashTable"));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("key_dtype", keyDtype);
+    opBuilder.setAttr("value_dtype", valueDtype);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.container != null) {
+          opBuilder.setAttr("container", opts.container);
+        }
+        if (opts.sharedName != null) {
+          opBuilder.setAttr("shared_name", opts.sharedName);
+        }
+        if (opts.useNodeNameSharing != null) {
+          opBuilder.setAttr("use_node_name_sharing", opts.useNodeNameSharing);
+        }
+      }
+    }
+    return new MutableHashTable(opBuilder.build());
+  }
+  
+  /**
+   * @param container If non-empty, this table is placed in the given container.
+   * Otherwise, a default container is used.
+   */
+  public static Options container(String container) {
+    return new Options().container(container);
+  }
+  
+  /**
+   * @param sharedName If non-empty, this table is shared under the given name across
+   * multiple sessions.
+   */
+  public static Options sharedName(String sharedName) {
+    return new Options().sharedName(sharedName);
+  }
+  
+  /**
+   * @param useNodeNameSharing If true and shared_name is empty, the table is shared
+   * using the node name.
+   */
+  public static Options useNodeNameSharing(Boolean useNodeNameSharing) {
+    return new Options().useNodeNameSharing(useNodeNameSharing);
+  }
+  
+  /**
+   * Handle to a table.
+   */
+  public Output<?> tableHandle() {
+    return tableHandle;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) tableHandle;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "MutableHashTableV2";
+  
+  private Output<?> tableHandle;
+  
+  private MutableHashTable(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    tableHandle = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MutableHashTableOfTensors.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MutableHashTableOfTensors.java
new file mode 100644
index 00000000000..49d2e3c0373
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MutableHashTableOfTensors.java
@@ -0,0 +1,176 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Creates an empty hash table.
+ * <p>
+ * This op creates a mutable hash table, specifying the type of its keys and
+ * values. Each value must be a vector. Data can be inserted into the table using
+ * the insert operations. It does not support the initialization operation.
+ */
+@Operator
+public final class MutableHashTableOfTensors extends RawOp implements Operand<TType> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.core.MutableHashTableOfTensors}
+   */
+  public static class Options {
+    
+    /**
+     * @param container If non-empty, this table is placed in the given container.
+     * Otherwise, a default container is used.
+     */
+    public Options container(String container) {
+      this.container = container;
+      return this;
+    }
+    
+    /**
+     * @param sharedName If non-empty, this table is shared under the given name across
+     * multiple sessions.
+     */
+    public Options sharedName(String sharedName) {
+      this.sharedName = sharedName;
+      return this;
+    }
+    
+    /**
+     * @param useNodeNameSharing 
+     */
+    public Options useNodeNameSharing(Boolean useNodeNameSharing) {
+      this.useNodeNameSharing = useNodeNameSharing;
+      return this;
+    }
+    
+    /**
+     * @param valueShape 
+     */
+    public Options valueShape(Shape valueShape) {
+      this.valueShape = valueShape;
+      return this;
+    }
+    
+    private String container;
+    private String sharedName;
+    private Boolean useNodeNameSharing;
+    private Shape valueShape;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new MutableHashTableOfTensors operation.
+   * 
+   * @param scope current scope
+   * @param keyDtype Type of the table keys.
+   * @param valueDtype Type of the table values.
+   * @param options carries optional attributes values
+   * @return a new instance of MutableHashTableOfTensors
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType, U extends TType> MutableHashTableOfTensors create(Scope scope, Class<T> keyDtype, Class<U> valueDtype, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("MutableHashTableOfTensorsV2", scope.makeOpName("MutableHashTableOfTensors"));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("key_dtype", keyDtype);
+    opBuilder.setAttr("value_dtype", valueDtype);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.container != null) {
+          opBuilder.setAttr("container", opts.container);
+        }
+        if (opts.sharedName != null) {
+          opBuilder.setAttr("shared_name", opts.sharedName);
+        }
+        if (opts.useNodeNameSharing != null) {
+          opBuilder.setAttr("use_node_name_sharing", opts.useNodeNameSharing);
+        }
+        if (opts.valueShape != null) {
+          opBuilder.setAttr("value_shape", opts.valueShape);
+        }
+      }
+    }
+    return new MutableHashTableOfTensors(opBuilder.build());
+  }
+  
+  /**
+   * @param container If non-empty, this table is placed in the given container.
+   * Otherwise, a default container is used.
+   */
+  public static Options container(String container) {
+    return new Options().container(container);
+  }
+  
+  /**
+   * @param sharedName If non-empty, this table is shared under the given name across
+   * multiple sessions.
+   */
+  public static Options sharedName(String sharedName) {
+    return new Options().sharedName(sharedName);
+  }
+  
+  /**
+   * @param useNodeNameSharing 
+   */
+  public static Options useNodeNameSharing(Boolean useNodeNameSharing) {
+    return new Options().useNodeNameSharing(useNodeNameSharing);
+  }
+  
+  /**
+   * @param valueShape 
+   */
+  public static Options valueShape(Shape valueShape) {
+    return new Options().valueShape(valueShape);
+  }
+  
+  /**
+   * Handle to a table.
+   */
+  public Output<?> tableHandle() {
+    return tableHandle;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) tableHandle;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "MutableHashTableOfTensorsV2";
+  
+  private Output<?> tableHandle;
+  
+  private MutableHashTableOfTensors(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    tableHandle = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Mutex.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Mutex.java
new file mode 100644
index 00000000000..1af2b439372
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Mutex.java
@@ -0,0 +1,129 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Creates a Mutex resource that can be locked by `MutexLock`.
+ */
+@Operator
+public final class Mutex extends RawOp implements Operand<TType> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.core.Mutex}
+   */
+  public static class Options {
+    
+    /**
+     * @param container If non-empty, this variable is placed in the given container.
+     * Otherwise, a default container is used.
+     */
+    public Options container(String container) {
+      this.container = container;
+      return this;
+    }
+    
+    /**
+     * @param sharedName If non-empty, this variable is named in the given bucket
+     * with this shared_name. Otherwise, the node name is used instead.
+     */
+    public Options sharedName(String sharedName) {
+      this.sharedName = sharedName;
+      return this;
+    }
+    
+    private String container;
+    private String sharedName;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new Mutex operation.
+   * 
+   * @param scope current scope
+   * @param options carries optional attributes values
+   * @return a new instance of Mutex
+   */
+  @Endpoint(describeByClass = true)
+  public static Mutex create(Scope scope, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("MutexV2", scope.makeOpName("Mutex"));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.container != null) {
+          opBuilder.setAttr("container", opts.container);
+        }
+        if (opts.sharedName != null) {
+          opBuilder.setAttr("shared_name", opts.sharedName);
+        }
+      }
+    }
+    return new Mutex(opBuilder.build());
+  }
+  
+  /**
+   * @param container If non-empty, this variable is placed in the given container.
+   * Otherwise, a default container is used.
+   */
+  public static Options container(String container) {
+    return new Options().container(container);
+  }
+  
+  /**
+   * @param sharedName If non-empty, this variable is named in the given bucket
+   * with this shared_name. Otherwise, the node name is used instead.
+   */
+  public static Options sharedName(String sharedName) {
+    return new Options().sharedName(sharedName);
+  }
+  
+  /**
+   * The mutex resource.
+   */
+  public Output<?> resource() {
+    return resource;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) resource;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "MutexV2";
+  
+  private Output<?> resource;
+  
+  private Mutex(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    resource = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MutexLock.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MutexLock.java
new file mode 100644
index 00000000000..de28b8f5025
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/MutexLock.java
@@ -0,0 +1,112 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Locks a mutex resource.  The output is the lock.  So long as the lock tensor
+ * <p>
+ * is alive, any other request to use `MutexLock` with this mutex will wait.
+ * <p>
+ * This is particularly useful for creating a critical section when used in
+ * conjunction with `MutexLockIdentity`:
+ * <pre>{@code
+ * mutex = mutex_v2(
+ *   shared_name=handle_name, container=container, name=name)
+ * 
+ * def execute_in_critical_section(fn, *args, **kwargs):
+ *   lock = gen_resource_variable_ops.mutex_lock(mutex)
+ * 
+ *   with ops.control_dependencies([lock]):
+ *     r = fn(*args, **kwargs)
+ * 
+ *   with ops.control_dependencies(nest.flatten(r)):
+ *     with ops.colocate_with(mutex):
+ *       ensure_lock_exists = mutex_lock_identity(lock)
+ * 
+ *     # Make sure that if any element of r is accessed, all of
+ *     # them are executed together.
+ *     r = nest.map_structure(tf.identity, r)
+ * 
+ *   with ops.control_dependencies([ensure_lock_exists]):
+ *     return nest.map_structure(tf.identity, r)
+ * }</pre>
+ * While `fn` is running in the critical section, no other functions which wish to
+ * use this critical section may run.
+ * <p>
+ * Often the use case is that two executions of the same graph, in parallel,
+ * wish to run `fn`; and we wish to ensure that only one of them executes
+ * at a time.  This is especially important if `fn` modifies one or more
+ * variables at a time.
+ * <p>
+ * It is also useful if two separate functions must share a resource, but we
+ * wish to ensure the usage is exclusive.
+ */
+@Operator
+public final class MutexLock extends RawOp implements Operand<TType> {
+  
+  /**
+   * Factory method to create a class wrapping a new MutexLock operation.
+   * 
+   * @param scope current scope
+   * @param mutex The mutex resource to lock.
+   * @return a new instance of MutexLock
+   */
+  @Endpoint(describeByClass = true)
+  public static MutexLock create(Scope scope, Operand<?> mutex) {
+    OperationBuilder opBuilder = scope.env().opBuilder("MutexLock", scope.makeOpName("MutexLock"));
+    opBuilder.addInput(mutex.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new MutexLock(opBuilder.build());
+  }
+  
+  /**
+   * A tensor that keeps a shared pointer to a lock on the mutex;
+   * when the Tensor is destroyed, the use count on the shared pointer is decreased
+   * by 1.  When it reaches 0, the lock is released.
+   */
+  public Output<?> mutexLock() {
+    return mutexLock;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) mutexLock;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "MutexLock";
+  
+  private Output<?> mutexLock;
+  
+  private MutexLock(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    mutexLock = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/NcclAllReduce.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/NcclAllReduce.java
new file mode 100644
index 00000000000..171cf01d750
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/NcclAllReduce.java
@@ -0,0 +1,92 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Outputs a tensor containing the reduction across all input tensors.
+ * <p>
+ * Outputs a tensor containing the reduction across all input tensors passed to ops
+ * within the same `shared_name.
+ * <p>
+ * The graph should be constructed so if one op runs with shared_name value `c`,
+ * then `num_devices` ops will run with shared_name value `c`.  Failure to do so
+ * will cause the graph execution to fail to complete.
+ * <p>
+ * input: the input to the reduction
+ * data: the value of the reduction across all `num_devices` devices.
+ * reduction: the reduction operation to perform.
+ * num_devices: The number of devices participating in this reduction.
+ * shared_name: Identifier that shared between ops of the same reduction.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+public final class NcclAllReduce<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new NcclAllReduce operation.
+   * 
+   * @param scope current scope
+   * @param input 
+   * @param reduction 
+   * @param numDevices 
+   * @param sharedName 
+   * @return a new instance of NcclAllReduce
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> NcclAllReduce<T> create(Scope scope, Operand<T> input, String reduction, Long numDevices, String sharedName) {
+    OperationBuilder opBuilder = scope.env().opBuilder("NcclAllReduce", scope.makeOpName("NcclAllReduce"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("reduction", reduction);
+    opBuilder.setAttr("num_devices", numDevices);
+    opBuilder.setAttr("shared_name", sharedName);
+    return new NcclAllReduce<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "NcclAllReduce";
+  
+  private Output<T> output;
+  
+  private NcclAllReduce(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/NcclBroadcast.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/NcclBroadcast.java
new file mode 100644
index 00000000000..17db313a42e
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/NcclBroadcast.java
@@ -0,0 +1,86 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Sends `input` to all devices that are connected to the output.
+ * <p>
+ * Sends `input` to all devices that are connected to the output.
+ * <p>
+ * The graph should be constructed so that all ops connected to the output have a
+ * valid device assignment, and the op itself is assigned one of these devices.
+ * <p>
+ * input: The input to the broadcast.
+ * output: The same as input.
+ * shape: The shape of the input tensor.
+ * 
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+public final class NcclBroadcast<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new NcclBroadcast operation.
+   * 
+   * @param scope current scope
+   * @param input 
+   * @param shape 
+   * @return a new instance of NcclBroadcast
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> NcclBroadcast<T> create(Scope scope, Operand<T> input, Shape shape) {
+    OperationBuilder opBuilder = scope.env().opBuilder("NcclBroadcast", scope.makeOpName("NcclBroadcast"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("shape", shape);
+    return new NcclBroadcast<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "NcclBroadcast";
+  
+  private Output<T> output;
+  
+  private NcclBroadcast(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/NcclReduce.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/NcclReduce.java
new file mode 100644
index 00000000000..6156dbc5551
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/NcclReduce.java
@@ -0,0 +1,85 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.Operands;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Reduces `input` from `num_devices` using `reduction` to a single device.
+ * <p>
+ * Reduces `input` from `num_devices` using `reduction` to a single device.
+ * <p>
+ * The graph should be constructed so that all inputs have a valid device
+ * assignment, and the op itself is assigned one of these devices.
+ * <p>
+ * input: The input to the reduction.
+ * data: the value of the reduction across all `num_devices` devices.
+ * reduction: the reduction operation to perform.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+public final class NcclReduce<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new NcclReduce operation.
+   * 
+   * @param scope current scope
+   * @param input 
+   * @param reduction 
+   * @return a new instance of NcclReduce
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> NcclReduce<T> create(Scope scope, Iterable<Operand<T>> input, String reduction) {
+    OperationBuilder opBuilder = scope.env().opBuilder("NcclReduce", scope.makeOpName("NcclReduce"));
+    opBuilder.addInputList(Operands.asOutputs(input));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("reduction", reduction);
+    return new NcclReduce<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "NcclReduce";
+  
+  private Output<T> output;
+  
+  private NcclReduce(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/NextIteration.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/NextIteration.java
new file mode 100644
index 00000000000..69b6be6a810
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/NextIteration.java
@@ -0,0 +1,75 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Makes its input available to the next iteration.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator
+public final class NextIteration<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new NextIteration operation.
+   * 
+   * @param scope current scope
+   * @param data The tensor to be made available to the next iteration.
+   * @return a new instance of NextIteration
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> NextIteration<T> create(Scope scope, Operand<T> data) {
+    OperationBuilder opBuilder = scope.env().opBuilder("NextIteration", scope.makeOpName("NextIteration"));
+    opBuilder.addInput(data.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new NextIteration<T>(opBuilder.build());
+  }
+  
+  /**
+   * The same tensor as `data`.
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "NextIteration";
+  
+  private Output<T> output;
+  
+  private NextIteration(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/NoOp.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/NoOp.java
new file mode 100644
index 00000000000..922b5d55ce3
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/NoOp.java
@@ -0,0 +1,52 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+
+/**
+ * Does nothing. Only useful as a placeholder for control edges.
+ */
+@Operator
+public final class NoOp extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new NoOp operation.
+   * 
+   * @param scope current scope
+   * @return a new instance of NoOp
+   */
+  @Endpoint(describeByClass = true)
+  public static NoOp create(Scope scope) {
+    OperationBuilder opBuilder = scope.env().opBuilder("NoOp", scope.makeOpName("NoOp"));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new NoOp(opBuilder.build());
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "NoOp";
+  
+  private NoOp(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OneHot.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OneHot.java
new file mode 100644
index 00000000000..442b9ae9108
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OneHot.java
@@ -0,0 +1,198 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Returns a one-hot tensor.
+ * <p>
+ * The locations represented by indices in `indices` take value `on_value`,
+ * while all other locations take value `off_value`.
+ * <p>
+ * If the input `indices` is rank `N`, the output will have rank `N+1`,
+ * The new axis is created at dimension `axis` (default: the new axis is
+ * appended at the end).
+ * <p>
+ * If `indices` is a scalar the output shape will be a vector of length `depth`.
+ * <p>
+ * If `indices` is a vector of length `features`, the output shape will be:
+ * <pre>{@code
+ *   features x depth if axis == -1
+ *   depth x features if axis == 0
+ * }</pre>
+ * If `indices` is a matrix (batch) with shape `[batch, features]`,
+ * the output shape will be:
+ * <pre>{@code
+ *   batch x features x depth if axis == -1
+ *   batch x depth x features if axis == 1
+ *   depth x batch x features if axis == 0
+ * }</pre>
+ * Examples
+ * =========
+ * <p>
+ * Suppose that
+ * <pre>{@code
+ *   indices = [0, 2, -1, 1]
+ *   depth = 3
+ *   on_value = 5.0
+ *   off_value = 0.0
+ *   axis = -1
+ * }</pre>
+ * Then output is `[4 x 3]`:
+ * <pre>{@code
+ * output =
+ *   [5.0 0.0 0.0]  // one_hot(0)
+ *   [0.0 0.0 5.0]  // one_hot(2)
+ *   [0.0 0.0 0.0]  // one_hot(-1)
+ *   [0.0 5.0 0.0]  // one_hot(1)
+ * }</pre>
+ * Suppose that
+ * <pre>{@code
+ *   indices = [0, 2, -1, 1]
+ *   depth = 3
+ *   on_value = 0.0
+ *   off_value = 3.0
+ *   axis = 0
+ * }</pre>
+ * Then output is `[3 x 4]`:
+ * <pre>{@code
+ * output =
+ *   [0.0 3.0 3.0 3.0]
+ *   [3.0 3.0 3.0 0.0]
+ *   [3.0 3.0 3.0 3.0]
+ *   [3.0 0.0 3.0 3.0]
+ * //  ^                one_hot(0)
+ * //      ^            one_hot(2)
+ * //          ^        one_hot(-1)
+ * //              ^    one_hot(1)
+ * }</pre>
+ * Suppose that
+ * <pre>{@code
+ *   indices = [[0, 2], [1, -1]]
+ *   depth = 3
+ *   on_value = 1.0
+ *   off_value = 0.0
+ *   axis = -1
+ * }</pre>
+ * Then output is `[2 x 2 x 3]`:
+ * <pre>{@code
+ * output =
+ *   [
+ *     [1.0, 0.0, 0.0]  // one_hot(0)
+ *     [0.0, 0.0, 1.0]  // one_hot(2)
+ *   ][
+ *     [0.0, 1.0, 0.0]  // one_hot(1)
+ *     [0.0, 0.0, 0.0]  // one_hot(-1)
+ *   ]
+ * }</pre>
+ * 
+ * 
+ * @param <U> data type for {@code output()} output
+ */
+@Operator
+public final class OneHot<U extends TType> extends RawOp implements Operand<U> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.core.OneHot}
+   */
+  public static class Options {
+    
+    /**
+     * @param axis The axis to fill (default: -1, a new inner-most axis).
+     */
+    public Options axis(Long axis) {
+      this.axis = axis;
+      return this;
+    }
+    
+    private Long axis;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new OneHot operation.
+   * 
+   * @param scope current scope
+   * @param indices A tensor of indices.
+   * @param depth A scalar defining the depth of the one hot dimension.
+   * @param onValue A scalar defining the value to fill in output when `indices[j] = i`.
+   * @param offValue A scalar defining the value to fill in output when `indices[j] != i`.
+   * @param options carries optional attributes values
+   * @return a new instance of OneHot
+   */
+  @Endpoint(describeByClass = true)
+  public static <U extends TType, T extends TNumber> OneHot<U> create(Scope scope, Operand<T> indices, Operand<TInt32> depth, Operand<U> onValue, Operand<U> offValue, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("OneHot", scope.makeOpName("OneHot"));
+    opBuilder.addInput(indices.asOutput());
+    opBuilder.addInput(depth.asOutput());
+    opBuilder.addInput(onValue.asOutput());
+    opBuilder.addInput(offValue.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.axis != null) {
+          opBuilder.setAttr("axis", opts.axis);
+        }
+      }
+    }
+    return new OneHot<U>(opBuilder.build());
+  }
+  
+  /**
+   * @param axis The axis to fill (default: -1, a new inner-most axis).
+   */
+  public static Options axis(Long axis) {
+    return new Options().axis(axis);
+  }
+  
+  /**
+   * The one-hot tensor.
+   */
+  public Output<U> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<U> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "OneHot";
+  
+  private Output<U> output;
+  
+  private OneHot(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OnesLike.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OnesLike.java
new file mode 100644
index 00000000000..986d0201f26
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OnesLike.java
@@ -0,0 +1,75 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Returns a tensor of ones with the same shape and type as x.
+ * 
+ * @param <T> data type for {@code y()} output
+ */
+@Operator
+public final class OnesLike<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new OnesLike operation.
+   * 
+   * @param scope current scope
+   * @param x a tensor of type T.
+   * @return a new instance of OnesLike
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> OnesLike<T> create(Scope scope, Operand<T> x) {
+    OperationBuilder opBuilder = scope.env().opBuilder("OnesLike", scope.makeOpName("OnesLike"));
+    opBuilder.addInput(x.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new OnesLike<T>(opBuilder.build());
+  }
+  
+  /**
+   * a tensor of the same shape and type as x but filled with ones.
+   */
+  public Output<T> y() {
+    return y;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return y;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "OnesLike";
+  
+  private Output<T> y;
+  
+  private OnesLike(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    y = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OrderedMapClear.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OrderedMapClear.java
new file mode 100644
index 00000000000..df4b4f26bfc
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OrderedMapClear.java
@@ -0,0 +1,151 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import java.util.List;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Op removes all elements in the underlying container.
+ */
+@Operator
+public final class OrderedMapClear extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.core.OrderedMapClear}
+   */
+  public static class Options {
+    
+    /**
+     * @param capacity 
+     */
+    public Options capacity(Long capacity) {
+      this.capacity = capacity;
+      return this;
+    }
+    
+    /**
+     * @param memoryLimit 
+     */
+    public Options memoryLimit(Long memoryLimit) {
+      this.memoryLimit = memoryLimit;
+      return this;
+    }
+    
+    /**
+     * @param container 
+     */
+    public Options container(String container) {
+      this.container = container;
+      return this;
+    }
+    
+    /**
+     * @param sharedName 
+     */
+    public Options sharedName(String sharedName) {
+      this.sharedName = sharedName;
+      return this;
+    }
+    
+    private Long capacity;
+    private Long memoryLimit;
+    private String container;
+    private String sharedName;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new OrderedMapClear operation.
+   * 
+   * @param scope current scope
+   * @param dtypes 
+   * @param options carries optional attributes values
+   * @return a new instance of OrderedMapClear
+   */
+  @Endpoint(describeByClass = true)
+  public static OrderedMapClear create(Scope scope, List<Class<? extends TType>> dtypes, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("OrderedMapClear", scope.makeOpName("OrderedMapClear"));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    Class[] dtypesArray = new Class[dtypes.size()];
+    for (int i = 0; i < dtypesArray.length; ++i) {
+      dtypesArray[i] = dtypes.get(i);
+    }
+    opBuilder.setAttr("dtypes", dtypesArray);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.capacity != null) {
+          opBuilder.setAttr("capacity", opts.capacity);
+        }
+        if (opts.memoryLimit != null) {
+          opBuilder.setAttr("memory_limit", opts.memoryLimit);
+        }
+        if (opts.container != null) {
+          opBuilder.setAttr("container", opts.container);
+        }
+        if (opts.sharedName != null) {
+          opBuilder.setAttr("shared_name", opts.sharedName);
+        }
+      }
+    }
+    return new OrderedMapClear(opBuilder.build());
+  }
+  
+  /**
+   * @param capacity 
+   */
+  public static Options capacity(Long capacity) {
+    return new Options().capacity(capacity);
+  }
+  
+  /**
+   * @param memoryLimit 
+   */
+  public static Options memoryLimit(Long memoryLimit) {
+    return new Options().memoryLimit(memoryLimit);
+  }
+  
+  /**
+   * @param container 
+   */
+  public static Options container(String container) {
+    return new Options().container(container);
+  }
+  
+  /**
+   * @param sharedName 
+   */
+  public static Options sharedName(String sharedName) {
+    return new Options().sharedName(sharedName);
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "OrderedMapClear";
+  
+  private OrderedMapClear(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OrderedMapIncompleteSize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OrderedMapIncompleteSize.java
new file mode 100644
index 00000000000..1b686b2f14b
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OrderedMapIncompleteSize.java
@@ -0,0 +1,169 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Op returns the number of incomplete elements in the underlying container.
+ */
+@Operator
+public final class OrderedMapIncompleteSize extends RawOp implements Operand<TInt32> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.core.OrderedMapIncompleteSize}
+   */
+  public static class Options {
+    
+    /**
+     * @param capacity 
+     */
+    public Options capacity(Long capacity) {
+      this.capacity = capacity;
+      return this;
+    }
+    
+    /**
+     * @param memoryLimit 
+     */
+    public Options memoryLimit(Long memoryLimit) {
+      this.memoryLimit = memoryLimit;
+      return this;
+    }
+    
+    /**
+     * @param container 
+     */
+    public Options container(String container) {
+      this.container = container;
+      return this;
+    }
+    
+    /**
+     * @param sharedName 
+     */
+    public Options sharedName(String sharedName) {
+      this.sharedName = sharedName;
+      return this;
+    }
+    
+    private Long capacity;
+    private Long memoryLimit;
+    private String container;
+    private String sharedName;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new OrderedMapIncompleteSize operation.
+   * 
+   * @param scope current scope
+   * @param dtypes 
+   * @param options carries optional attributes values
+   * @return a new instance of OrderedMapIncompleteSize
+   */
+  @Endpoint(describeByClass = true)
+  public static OrderedMapIncompleteSize create(Scope scope, List<Class<? extends TType>> dtypes, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("OrderedMapIncompleteSize", scope.makeOpName("OrderedMapIncompleteSize"));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    Class[] dtypesArray = new Class[dtypes.size()];
+    for (int i = 0; i < dtypesArray.length; ++i) {
+      dtypesArray[i] = dtypes.get(i);
+    }
+    opBuilder.setAttr("dtypes", dtypesArray);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.capacity != null) {
+          opBuilder.setAttr("capacity", opts.capacity);
+        }
+        if (opts.memoryLimit != null) {
+          opBuilder.setAttr("memory_limit", opts.memoryLimit);
+        }
+        if (opts.container != null) {
+          opBuilder.setAttr("container", opts.container);
+        }
+        if (opts.sharedName != null) {
+          opBuilder.setAttr("shared_name", opts.sharedName);
+        }
+      }
+    }
+    return new OrderedMapIncompleteSize(opBuilder.build());
+  }
+  
+  /**
+   * @param capacity 
+   */
+  public static Options capacity(Long capacity) {
+    return new Options().capacity(capacity);
+  }
+  
+  /**
+   * @param memoryLimit 
+   */
+  public static Options memoryLimit(Long memoryLimit) {
+    return new Options().memoryLimit(memoryLimit);
+  }
+  
+  /**
+   * @param container 
+   */
+  public static Options container(String container) {
+    return new Options().container(container);
+  }
+  
+  /**
+   * @param sharedName 
+   */
+  public static Options sharedName(String sharedName) {
+    return new Options().sharedName(sharedName);
+  }
+  
+  /**
+   */
+  public Output<TInt32> size() {
+    return size;
+  }
+  
+  @Override
+  public Output<TInt32> asOutput() {
+    return size;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "OrderedMapIncompleteSize";
+  
+  private Output<TInt32> size;
+  
+  private OrderedMapIncompleteSize(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    size = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OrderedMapPeek.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OrderedMapPeek.java
new file mode 100644
index 00000000000..6b052e5f00c
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OrderedMapPeek.java
@@ -0,0 +1,183 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import java.util.Arrays;
+import java.util.Iterator;
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Op peeks at the values at the specified key.  If the
+ * <p>
+ * underlying container does not contain this key
+ * this op will block until it does.   This Op is optimized for
+ * performance.
+ */
+@Operator
+public final class OrderedMapPeek extends RawOp implements Iterable<Operand<TType>> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.core.OrderedMapPeek}
+   */
+  public static class Options {
+    
+    /**
+     * @param capacity 
+     */
+    public Options capacity(Long capacity) {
+      this.capacity = capacity;
+      return this;
+    }
+    
+    /**
+     * @param memoryLimit 
+     */
+    public Options memoryLimit(Long memoryLimit) {
+      this.memoryLimit = memoryLimit;
+      return this;
+    }
+    
+    /**
+     * @param container 
+     */
+    public Options container(String container) {
+      this.container = container;
+      return this;
+    }
+    
+    /**
+     * @param sharedName 
+     */
+    public Options sharedName(String sharedName) {
+      this.sharedName = sharedName;
+      return this;
+    }
+    
+    private Long capacity;
+    private Long memoryLimit;
+    private String container;
+    private String sharedName;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new OrderedMapPeek operation.
+   * 
+   * @param scope current scope
+   * @param key 
+   * @param indices 
+   * @param dtypes 
+   * @param options carries optional attributes values
+   * @return a new instance of OrderedMapPeek
+   */
+  @Endpoint(describeByClass = true)
+  public static OrderedMapPeek create(Scope scope, Operand<TInt64> key, Operand<TInt32> indices, List<Class<? extends TType>> dtypes, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("OrderedMapPeek", scope.makeOpName("OrderedMapPeek"));
+    opBuilder.addInput(key.asOutput());
+    opBuilder.addInput(indices.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    Class[] dtypesArray = new Class[dtypes.size()];
+    for (int i = 0; i < dtypesArray.length; ++i) {
+      dtypesArray[i] = dtypes.get(i);
+    }
+    opBuilder.setAttr("dtypes", dtypesArray);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.capacity != null) {
+          opBuilder.setAttr("capacity", opts.capacity);
+        }
+        if (opts.memoryLimit != null) {
+          opBuilder.setAttr("memory_limit", opts.memoryLimit);
+        }
+        if (opts.container != null) {
+          opBuilder.setAttr("container", opts.container);
+        }
+        if (opts.sharedName != null) {
+          opBuilder.setAttr("shared_name", opts.sharedName);
+        }
+      }
+    }
+    return new OrderedMapPeek(opBuilder.build());
+  }
+  
+  /**
+   * @param capacity 
+   */
+  public static Options capacity(Long capacity) {
+    return new Options().capacity(capacity);
+  }
+  
+  /**
+   * @param memoryLimit 
+   */
+  public static Options memoryLimit(Long memoryLimit) {
+    return new Options().memoryLimit(memoryLimit);
+  }
+  
+  /**
+   * @param container 
+   */
+  public static Options container(String container) {
+    return new Options().container(container);
+  }
+  
+  /**
+   * @param sharedName 
+   */
+  public static Options sharedName(String sharedName) {
+    return new Options().sharedName(sharedName);
+  }
+  
+  /**
+   */
+  public List<Output<?>> values() {
+    return values;
+  }
+  
+  @Override
+  @SuppressWarnings({"rawtypes", "unchecked"})
+  public Iterator<Operand<TType>> iterator() {
+    return (Iterator) values.iterator();
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "OrderedMapPeek";
+  
+  private List<Output<?>> values;
+  
+  private OrderedMapPeek(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    int valuesLength = operation.outputListLength("values");
+    values = Arrays.asList(operation.outputList(outputIdx, valuesLength));
+    outputIdx += valuesLength;
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OrderedMapSize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OrderedMapSize.java
new file mode 100644
index 00000000000..69c7851b13c
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OrderedMapSize.java
@@ -0,0 +1,169 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Op returns the number of elements in the underlying container.
+ */
+@Operator
+public final class OrderedMapSize extends RawOp implements Operand<TInt32> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.core.OrderedMapSize}
+   */
+  public static class Options {
+    
+    /**
+     * @param capacity 
+     */
+    public Options capacity(Long capacity) {
+      this.capacity = capacity;
+      return this;
+    }
+    
+    /**
+     * @param memoryLimit 
+     */
+    public Options memoryLimit(Long memoryLimit) {
+      this.memoryLimit = memoryLimit;
+      return this;
+    }
+    
+    /**
+     * @param container 
+     */
+    public Options container(String container) {
+      this.container = container;
+      return this;
+    }
+    
+    /**
+     * @param sharedName 
+     */
+    public Options sharedName(String sharedName) {
+      this.sharedName = sharedName;
+      return this;
+    }
+    
+    private Long capacity;
+    private Long memoryLimit;
+    private String container;
+    private String sharedName;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new OrderedMapSize operation.
+   * 
+   * @param scope current scope
+   * @param dtypes 
+   * @param options carries optional attributes values
+   * @return a new instance of OrderedMapSize
+   */
+  @Endpoint(describeByClass = true)
+  public static OrderedMapSize create(Scope scope, List<Class<? extends TType>> dtypes, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("OrderedMapSize", scope.makeOpName("OrderedMapSize"));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    Class[] dtypesArray = new Class[dtypes.size()];
+    for (int i = 0; i < dtypesArray.length; ++i) {
+      dtypesArray[i] = dtypes.get(i);
+    }
+    opBuilder.setAttr("dtypes", dtypesArray);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.capacity != null) {
+          opBuilder.setAttr("capacity", opts.capacity);
+        }
+        if (opts.memoryLimit != null) {
+          opBuilder.setAttr("memory_limit", opts.memoryLimit);
+        }
+        if (opts.container != null) {
+          opBuilder.setAttr("container", opts.container);
+        }
+        if (opts.sharedName != null) {
+          opBuilder.setAttr("shared_name", opts.sharedName);
+        }
+      }
+    }
+    return new OrderedMapSize(opBuilder.build());
+  }
+  
+  /**
+   * @param capacity 
+   */
+  public static Options capacity(Long capacity) {
+    return new Options().capacity(capacity);
+  }
+  
+  /**
+   * @param memoryLimit 
+   */
+  public static Options memoryLimit(Long memoryLimit) {
+    return new Options().memoryLimit(memoryLimit);
+  }
+  
+  /**
+   * @param container 
+   */
+  public static Options container(String container) {
+    return new Options().container(container);
+  }
+  
+  /**
+   * @param sharedName 
+   */
+  public static Options sharedName(String sharedName) {
+    return new Options().sharedName(sharedName);
+  }
+  
+  /**
+   */
+  public Output<TInt32> size() {
+    return size;
+  }
+  
+  @Override
+  public Output<TInt32> asOutput() {
+    return size;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "OrderedMapSize";
+  
+  private Output<TInt32> size;
+  
+  private OrderedMapSize(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    size = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OrderedMapStage.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OrderedMapStage.java
new file mode 100644
index 00000000000..12f3155d661
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OrderedMapStage.java
@@ -0,0 +1,168 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.Operands;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Stage (key, values) in the underlying container which behaves like a ordered
+ * <p>
+ * associative container.   Elements are ordered by key.
+ */
+@Operator
+public final class OrderedMapStage extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.core.OrderedMapStage}
+   */
+  public static class Options {
+    
+    /**
+     * @param capacity Maximum number of elements in the Staging Area. If > 0, inserts
+     * on the container will block when the capacity is reached.
+     */
+    public Options capacity(Long capacity) {
+      this.capacity = capacity;
+      return this;
+    }
+    
+    /**
+     * @param memoryLimit 
+     */
+    public Options memoryLimit(Long memoryLimit) {
+      this.memoryLimit = memoryLimit;
+      return this;
+    }
+    
+    /**
+     * @param container If non-empty, this queue is placed in the given container. Otherwise,
+     * a default container is used.
+     */
+    public Options container(String container) {
+      this.container = container;
+      return this;
+    }
+    
+    /**
+     * @param sharedName It is necessary to match this name to the matching Unstage Op.
+     */
+    public Options sharedName(String sharedName) {
+      this.sharedName = sharedName;
+      return this;
+    }
+    
+    private Long capacity;
+    private Long memoryLimit;
+    private String container;
+    private String sharedName;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new OrderedMapStage operation.
+   * 
+   * @param scope current scope
+   * @param key int64
+   * @param indices 
+   * @param values a list of tensors
+   * dtypes A list of data types that inserted values should adhere to.
+   * @param dtypes 
+   * @param options carries optional attributes values
+   * @return a new instance of OrderedMapStage
+   */
+  @Endpoint(describeByClass = true)
+  public static OrderedMapStage create(Scope scope, Operand<TInt64> key, Operand<TInt32> indices, Iterable<Operand<?>> values, List<Class<? extends TType>> dtypes, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("OrderedMapStage", scope.makeOpName("OrderedMapStage"));
+    opBuilder.addInput(key.asOutput());
+    opBuilder.addInput(indices.asOutput());
+    opBuilder.addInputList(Operands.asOutputs(values));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    Class[] dtypesArray = new Class[dtypes.size()];
+    for (int i = 0; i < dtypesArray.length; ++i) {
+      dtypesArray[i] = dtypes.get(i);
+    }
+    opBuilder.setAttr("dtypes", dtypesArray);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.capacity != null) {
+          opBuilder.setAttr("capacity", opts.capacity);
+        }
+        if (opts.memoryLimit != null) {
+          opBuilder.setAttr("memory_limit", opts.memoryLimit);
+        }
+        if (opts.container != null) {
+          opBuilder.setAttr("container", opts.container);
+        }
+        if (opts.sharedName != null) {
+          opBuilder.setAttr("shared_name", opts.sharedName);
+        }
+      }
+    }
+    return new OrderedMapStage(opBuilder.build());
+  }
+  
+  /**
+   * @param capacity Maximum number of elements in the Staging Area. If > 0, inserts
+   * on the container will block when the capacity is reached.
+   */
+  public static Options capacity(Long capacity) {
+    return new Options().capacity(capacity);
+  }
+  
+  /**
+   * @param memoryLimit 
+   */
+  public static Options memoryLimit(Long memoryLimit) {
+    return new Options().memoryLimit(memoryLimit);
+  }
+  
+  /**
+   * @param container If non-empty, this queue is placed in the given container. Otherwise,
+   * a default container is used.
+   */
+  public static Options container(String container) {
+    return new Options().container(container);
+  }
+  
+  /**
+   * @param sharedName It is necessary to match this name to the matching Unstage Op.
+   */
+  public static Options sharedName(String sharedName) {
+    return new Options().sharedName(sharedName);
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "OrderedMapStage";
+  
+  private OrderedMapStage(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OrderedMapUnstage.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OrderedMapUnstage.java
new file mode 100644
index 00000000000..8b08d7449c2
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OrderedMapUnstage.java
@@ -0,0 +1,182 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import java.util.Arrays;
+import java.util.Iterator;
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Op removes and returns the values associated with the key
+ * <p>
+ * from the underlying container.   If the underlying container
+ * does not contain this key, the op will block until it does.
+ */
+@Operator
+public final class OrderedMapUnstage extends RawOp implements Iterable<Operand<TType>> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.core.OrderedMapUnstage}
+   */
+  public static class Options {
+    
+    /**
+     * @param capacity 
+     */
+    public Options capacity(Long capacity) {
+      this.capacity = capacity;
+      return this;
+    }
+    
+    /**
+     * @param memoryLimit 
+     */
+    public Options memoryLimit(Long memoryLimit) {
+      this.memoryLimit = memoryLimit;
+      return this;
+    }
+    
+    /**
+     * @param container 
+     */
+    public Options container(String container) {
+      this.container = container;
+      return this;
+    }
+    
+    /**
+     * @param sharedName 
+     */
+    public Options sharedName(String sharedName) {
+      this.sharedName = sharedName;
+      return this;
+    }
+    
+    private Long capacity;
+    private Long memoryLimit;
+    private String container;
+    private String sharedName;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new OrderedMapUnstage operation.
+   * 
+   * @param scope current scope
+   * @param key 
+   * @param indices 
+   * @param dtypes 
+   * @param options carries optional attributes values
+   * @return a new instance of OrderedMapUnstage
+   */
+  @Endpoint(describeByClass = true)
+  public static OrderedMapUnstage create(Scope scope, Operand<TInt64> key, Operand<TInt32> indices, List<Class<? extends TType>> dtypes, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("OrderedMapUnstage", scope.makeOpName("OrderedMapUnstage"));
+    opBuilder.addInput(key.asOutput());
+    opBuilder.addInput(indices.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    Class[] dtypesArray = new Class[dtypes.size()];
+    for (int i = 0; i < dtypesArray.length; ++i) {
+      dtypesArray[i] = dtypes.get(i);
+    }
+    opBuilder.setAttr("dtypes", dtypesArray);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.capacity != null) {
+          opBuilder.setAttr("capacity", opts.capacity);
+        }
+        if (opts.memoryLimit != null) {
+          opBuilder.setAttr("memory_limit", opts.memoryLimit);
+        }
+        if (opts.container != null) {
+          opBuilder.setAttr("container", opts.container);
+        }
+        if (opts.sharedName != null) {
+          opBuilder.setAttr("shared_name", opts.sharedName);
+        }
+      }
+    }
+    return new OrderedMapUnstage(opBuilder.build());
+  }
+  
+  /**
+   * @param capacity 
+   */
+  public static Options capacity(Long capacity) {
+    return new Options().capacity(capacity);
+  }
+  
+  /**
+   * @param memoryLimit 
+   */
+  public static Options memoryLimit(Long memoryLimit) {
+    return new Options().memoryLimit(memoryLimit);
+  }
+  
+  /**
+   * @param container 
+   */
+  public static Options container(String container) {
+    return new Options().container(container);
+  }
+  
+  /**
+   * @param sharedName 
+   */
+  public static Options sharedName(String sharedName) {
+    return new Options().sharedName(sharedName);
+  }
+  
+  /**
+   */
+  public List<Output<?>> values() {
+    return values;
+  }
+  
+  @Override
+  @SuppressWarnings({"rawtypes", "unchecked"})
+  public Iterator<Operand<TType>> iterator() {
+    return (Iterator) values.iterator();
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "OrderedMapUnstage";
+  
+  private List<Output<?>> values;
+  
+  private OrderedMapUnstage(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    int valuesLength = operation.outputListLength("values");
+    values = Arrays.asList(operation.outputList(outputIdx, valuesLength));
+    outputIdx += valuesLength;
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OrderedMapUnstageNoKey.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OrderedMapUnstageNoKey.java
new file mode 100644
index 00000000000..edb9e82401a
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/OrderedMapUnstageNoKey.java
@@ -0,0 +1,181 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import java.util.Arrays;
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Op removes and returns the (key, value) element with the smallest
+ * <p>
+ * key from the underlying container.   If the underlying container
+ * does not contain elements, the op will block until it does.
+ */
+@Operator
+public final class OrderedMapUnstageNoKey extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.core.OrderedMapUnstageNoKey}
+   */
+  public static class Options {
+    
+    /**
+     * @param capacity 
+     */
+    public Options capacity(Long capacity) {
+      this.capacity = capacity;
+      return this;
+    }
+    
+    /**
+     * @param memoryLimit 
+     */
+    public Options memoryLimit(Long memoryLimit) {
+      this.memoryLimit = memoryLimit;
+      return this;
+    }
+    
+    /**
+     * @param container 
+     */
+    public Options container(String container) {
+      this.container = container;
+      return this;
+    }
+    
+    /**
+     * @param sharedName 
+     */
+    public Options sharedName(String sharedName) {
+      this.sharedName = sharedName;
+      return this;
+    }
+    
+    private Long capacity;
+    private Long memoryLimit;
+    private String container;
+    private String sharedName;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new OrderedMapUnstageNoKey operation.
+   * 
+   * @param scope current scope
+   * @param indices 
+   * @param dtypes 
+   * @param options carries optional attributes values
+   * @return a new instance of OrderedMapUnstageNoKey
+   */
+  @Endpoint(describeByClass = true)
+  public static OrderedMapUnstageNoKey create(Scope scope, Operand<TInt32> indices, List<Class<? extends TType>> dtypes, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("OrderedMapUnstageNoKey", scope.makeOpName("OrderedMapUnstageNoKey"));
+    opBuilder.addInput(indices.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    Class[] dtypesArray = new Class[dtypes.size()];
+    for (int i = 0; i < dtypesArray.length; ++i) {
+      dtypesArray[i] = dtypes.get(i);
+    }
+    opBuilder.setAttr("dtypes", dtypesArray);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.capacity != null) {
+          opBuilder.setAttr("capacity", opts.capacity);
+        }
+        if (opts.memoryLimit != null) {
+          opBuilder.setAttr("memory_limit", opts.memoryLimit);
+        }
+        if (opts.container != null) {
+          opBuilder.setAttr("container", opts.container);
+        }
+        if (opts.sharedName != null) {
+          opBuilder.setAttr("shared_name", opts.sharedName);
+        }
+      }
+    }
+    return new OrderedMapUnstageNoKey(opBuilder.build());
+  }
+  
+  /**
+   * @param capacity 
+   */
+  public static Options capacity(Long capacity) {
+    return new Options().capacity(capacity);
+  }
+  
+  /**
+   * @param memoryLimit 
+   */
+  public static Options memoryLimit(Long memoryLimit) {
+    return new Options().memoryLimit(memoryLimit);
+  }
+  
+  /**
+   * @param container 
+   */
+  public static Options container(String container) {
+    return new Options().container(container);
+  }
+  
+  /**
+   * @param sharedName 
+   */
+  public static Options sharedName(String sharedName) {
+    return new Options().sharedName(sharedName);
+  }
+  
+  /**
+   */
+  public Output<TInt64> key() {
+    return key;
+  }
+  
+  /**
+   */
+  public List<Output<?>> values() {
+    return values;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "OrderedMapUnstageNoKey";
+  
+  private Output<TInt64> key;
+  private List<Output<?>> values;
+  
+  private OrderedMapUnstageNoKey(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    key = operation.output(outputIdx++);
+    int valuesLength = operation.outputListLength("values");
+    values = Arrays.asList(operation.outputList(outputIdx, valuesLength));
+    outputIdx += valuesLength;
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Pad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Pad.java
new file mode 100644
index 00000000000..205e6ab0a62
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Pad.java
@@ -0,0 +1,104 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Pads a tensor.
+ * <p>
+ * This operation pads `input` according to the `paddings` and `constant_values`
+ * you specify. `paddings` is an integer tensor with shape `[Dn, 2]`, where n is
+ * the rank of `input`. For each dimension D of `input`, `paddings[D, 0]` indicates
+ * how many padding values to add before the contents of `input` in that dimension,
+ * and `paddings[D, 1]` indicates how many padding values to add after the contents
+ * of `input` in that dimension. `constant_values` is a scalar tensor of the same
+ * type as `input` that indicates the value to use for padding `input`.
+ * <p>
+ * The padded size of each dimension D of the output is:
+ * <p>
+ * `paddings(D, 0) + input.dim_size(D) + paddings(D, 1)`
+ * <p>
+ * For example:
+ * <pre>{@code
+ * # 't' is [[1, 1], [2, 2]]
+ * # 'paddings' is [[1, 1], [2, 2]]
+ * # 'constant_values' is 0
+ * # rank of 't' is 2
+ * pad(t, paddings) ==> [[0, 0, 0, 0, 0, 0]
+ *                       [0, 0, 1, 1, 0, 0]
+ *                       [0, 0, 2, 2, 0, 0]
+ *                       [0, 0, 0, 0, 0, 0]]
+ * }</pre>
+ * 
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator
+public final class Pad<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new Pad operation.
+   * 
+   * @param scope current scope
+   * @param input 
+   * @param paddings 
+   * @param constantValues 
+   * @return a new instance of Pad
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType, U extends TNumber> Pad<T> create(Scope scope, Operand<T> input, Operand<U> paddings, Operand<T> constantValues) {
+    OperationBuilder opBuilder = scope.env().opBuilder("PadV2", scope.makeOpName("Pad"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(paddings.asOutput());
+    opBuilder.addInput(constantValues.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new Pad<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "PadV2";
+  
+  private Output<T> output;
+  
+  private Pad(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ParallelConcat.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ParallelConcat.java
new file mode 100644
index 00000000000..e373612bb68
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ParallelConcat.java
@@ -0,0 +1,96 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.Operands;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Concatenates a list of `N` tensors along the first dimension.
+ * <p>
+ * The input tensors are all required to have size 1 in the first dimension.
+ * <p>
+ * For example:
+ * <pre>{@code
+ * # 'x' is [[1, 4]]
+ * # 'y' is [[2, 5]]
+ * # 'z' is [[3, 6]]
+ * parallel_concat([x, y, z]) => [[1, 4], [2, 5], [3, 6]]  # Pack along first dim.
+ * }</pre>
+ * The difference between concat and parallel_concat is that concat requires all
+ * of the inputs be computed before the operation will begin but doesn't require
+ * that the input shapes be known during graph construction.  Parallel concat
+ * will copy pieces of the input into the output as they become available, in
+ * some situations this can provide a performance benefit.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator
+public final class ParallelConcat<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new ParallelConcat operation.
+   * 
+   * @param scope current scope
+   * @param values Tensors to be concatenated. All must have size 1 in the first dimension
+   * and same shape.
+   * @param shape the final shape of the result; should be equal to the shapes of any input
+   * but with the number of input values in the first dimension.
+   * @return a new instance of ParallelConcat
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> ParallelConcat<T> create(Scope scope, Iterable<Operand<T>> values, Shape shape) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ParallelConcat", scope.makeOpName("ParallelConcat"));
+    opBuilder.addInputList(Operands.asOutputs(values));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("shape", shape);
+    return new ParallelConcat<T>(opBuilder.build());
+  }
+  
+  /**
+   * The concatenated tensor.
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ParallelConcat";
+  
+  private Output<T> output;
+  
+  private ParallelConcat(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ParallelDynamicStitch.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ParallelDynamicStitch.java
new file mode 100644
index 00000000000..2770fa01591
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ParallelDynamicStitch.java
@@ -0,0 +1,132 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.Operands;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Interleave the values from the `data` tensors into a single tensor.
+ * <p>
+ * Builds a merged tensor such that
+ * <pre>{@code
+ *     merged[indices[m][i, ..., j], ...] = data[m][i, ..., j, ...]
+ * }</pre>
+ * For example, if each `indices[m]` is scalar or vector, we have
+ * <pre>{@code
+ *     # Scalar indices:
+ *     merged[indices[m], ...] = data[m][...]
+ * 
+ *     # Vector indices:
+ *     merged[indices[m][i], ...] = data[m][i, ...]
+ * }</pre>
+ * Each `data[i].shape` must start with the corresponding `indices[i].shape`,
+ * and the rest of `data[i].shape` must be constant w.r.t. `i`.  That is, we
+ * must have `data[i].shape = indices[i].shape + constant`.  In terms of this
+ * `constant`, the output shape is
+ * <p>
+ *     merged.shape = [max(indices)] + constant
+ * <p>
+ * Values may be merged in parallel, so if an index appears in both `indices[m][i]`
+ * and `indices[n][j]`, the result may be invalid. This differs from the normal
+ * DynamicStitch operator that defines the behavior in that case.
+ * <p>
+ * For example:
+ * <pre>{@code
+ *     indices[0] = 6
+ *     indices[1] = [4, 1]
+ *     indices[2] = [[5, 2], [0, 3]]
+ *     data[0] = [61, 62]
+ *     data[1] = [[41, 42], [11, 12]]
+ *     data[2] = [[[51, 52], [21, 22]], [[1, 2], [31, 32]]]
+ *     merged = [[1, 2], [11, 12], [21, 22], [31, 32], [41, 42],
+ *               [51, 52], [61, 62]]
+ * }</pre>
+ * This method can be used to merge partitions created by `dynamic_partition`
+ * as illustrated on the following example:
+ * <pre>{@code
+ *     # Apply function (increments x_i) on elements for which a certain condition
+ *     # apply (x_i != -1 in this example).
+ *     x=tf.constant([0.1, -1., 5.2, 4.3, -1., 7.4])
+ *     condition_mask=tf.not_equal(x,tf.constant(-1.))
+ *     partitioned_data = tf.dynamic_partition(
+ *         x, tf.cast(condition_mask, tf.int32) , 2)
+ *     partitioned_data[1] = partitioned_data[1] + 1.0
+ *     condition_indices = tf.dynamic_partition(
+ *         tf.range(tf.shape(x)[0]), tf.cast(condition_mask, tf.int32) , 2)
+ *     x = tf.dynamic_stitch(condition_indices, partitioned_data)
+ *     # Here x=[1.1, -1., 6.2, 5.3, -1, 8.4], the -1. values remain
+ *     # unchanged.
+ * }</pre>
+ * <div style="width:70%; margin:auto; margin-bottom:10px; margin-top:20px;">
+ * <img style="width:100%" src="https://www.tensorflow.org/images/DynamicStitch.png" alt>
+ * </div>
+ * 
+ * @param <T> data type for {@code merged()} output
+ */
+@Operator
+public final class ParallelDynamicStitch<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new ParallelDynamicStitch operation.
+   * 
+   * @param scope current scope
+   * @param indices 
+   * @param data 
+   * @return a new instance of ParallelDynamicStitch
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> ParallelDynamicStitch<T> create(Scope scope, Iterable<Operand<TInt32>> indices, Iterable<Operand<T>> data) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ParallelDynamicStitch", scope.makeOpName("ParallelDynamicStitch"));
+    opBuilder.addInputList(Operands.asOutputs(indices));
+    opBuilder.addInputList(Operands.asOutputs(data));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new ParallelDynamicStitch<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> merged() {
+    return merged;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return merged;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ParallelDynamicStitch";
+  
+  private Output<T> merged;
+  
+  private ParallelDynamicStitch(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    merged = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Placeholder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Placeholder.java
new file mode 100644
index 00000000000..75d2326d007
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Placeholder.java
@@ -0,0 +1,116 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * A placeholder op for a value that will be fed into the computation.
+ * <p>
+ * N.B. This operation will fail with an error if it is executed. It is
+ * intended as a way to represent a value that will always be fed, and to
+ * provide attrs that enable the fed value to be checked at runtime.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator
+public final class Placeholder<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.core.Placeholder}
+   */
+  public static class Options {
+    
+    /**
+     * @param shape (Optional) The shape of the tensor. If the shape has 0 dimensions, the
+     * shape is unconstrained.
+     */
+    public Options shape(Shape shape) {
+      this.shape = shape;
+      return this;
+    }
+    
+    private Shape shape;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new Placeholder operation.
+   * 
+   * @param scope current scope
+   * @param dtype The type of elements in the tensor.
+   * @param options carries optional attributes values
+   * @return a new instance of Placeholder
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> Placeholder<T> create(Scope scope, Class<T> dtype, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Placeholder", scope.makeOpName("Placeholder"));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("dtype", dtype);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.shape != null) {
+          opBuilder.setAttr("shape", opts.shape);
+        }
+      }
+    }
+    return new Placeholder<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param shape (Optional) The shape of the tensor. If the shape has 0 dimensions, the
+   * shape is unconstrained.
+   */
+  public static Options shape(Shape shape) {
+    return new Options().shape(shape);
+  }
+  
+  /**
+   * A placeholder tensor that must be replaced using the feed mechanism.
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Placeholder";
+  
+  private Output<T> output;
+  
+  private Placeholder(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/PlaceholderWithDefault.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/PlaceholderWithDefault.java
new file mode 100644
index 00000000000..59a9ca223ab
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/PlaceholderWithDefault.java
@@ -0,0 +1,78 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * A placeholder op that passes through `input` when its output is not fed.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator
+public final class PlaceholderWithDefault<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new PlaceholderWithDefault operation.
+   * 
+   * @param scope current scope
+   * @param input The default value to produce when `output` is not fed.
+   * @param shape The (possibly partial) shape of the tensor.
+   * @return a new instance of PlaceholderWithDefault
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> PlaceholderWithDefault<T> create(Scope scope, Operand<T> input, Shape shape) {
+    OperationBuilder opBuilder = scope.env().opBuilder("PlaceholderWithDefault", scope.makeOpName("PlaceholderWithDefault"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("shape", shape);
+    return new PlaceholderWithDefault<T>(opBuilder.build());
+  }
+  
+  /**
+   * A placeholder tensor that defaults to `input` if it is not fed.
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "PlaceholderWithDefault";
+  
+  private Output<T> output;
+  
+  private PlaceholderWithDefault(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Print.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Print.java
new file mode 100644
index 00000000000..52b933329a0
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Print.java
@@ -0,0 +1,111 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TString;
+
+/**
+ * Prints a string scalar.
+ * <p>
+ * Prints a string scalar to the desired output_stream.
+ */
+@Operator
+public final class Print extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.core.Print}
+   */
+  public static class Options {
+    
+    /**
+     * @param outputStream A string specifying the output stream or logging level to print to.
+     */
+    public Options outputStream(String outputStream) {
+      this.outputStream = outputStream;
+      return this;
+    }
+    
+    /**
+     * @param end 
+     */
+    public Options end(String end) {
+      this.end = end;
+      return this;
+    }
+    
+    private String outputStream;
+    private String end;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new Print operation.
+   * 
+   * @param scope current scope
+   * @param input The string scalar to print.
+   * @param options carries optional attributes values
+   * @return a new instance of Print
+   */
+  @Endpoint(describeByClass = true)
+  public static Print create(Scope scope, Operand<TString> input, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("PrintV2", scope.makeOpName("Print"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.outputStream != null) {
+          opBuilder.setAttr("output_stream", opts.outputStream);
+        }
+        if (opts.end != null) {
+          opBuilder.setAttr("end", opts.end);
+        }
+      }
+    }
+    return new Print(opBuilder.build());
+  }
+  
+  /**
+   * @param outputStream A string specifying the output stream or logging level to print to.
+   */
+  public static Options outputStream(String outputStream) {
+    return new Options().outputStream(outputStream);
+  }
+  
+  /**
+   * @param end 
+   */
+  public static Options end(String end) {
+    return new Options().end(end);
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "PrintV2";
+  
+  private Print(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Prod.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Prod.java
new file mode 100644
index 00000000000..6bf1b7cd37d
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Prod.java
@@ -0,0 +1,118 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Computes the product of elements across dimensions of a tensor.
+ * <p>
+ * Reduces `input` along the dimensions given in `axis`. Unless
+ * `keep_dims` is true, the rank of the tensor is reduced by 1 for each entry in
+ * `axis`. If `keep_dims` is true, the reduced dimensions are
+ * retained with length 1.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator
+public final class Prod<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.core.Prod}
+   */
+  public static class Options {
+    
+    /**
+     * @param keepDims If true, retain reduced dimensions with length 1.
+     */
+    public Options keepDims(Boolean keepDims) {
+      this.keepDims = keepDims;
+      return this;
+    }
+    
+    private Boolean keepDims;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new Prod operation.
+   * 
+   * @param scope current scope
+   * @param input The tensor to reduce.
+   * @param axis The dimensions to reduce. Must be in the range
+   * `[-rank(input), rank(input))`.
+   * @param options carries optional attributes values
+   * @return a new instance of Prod
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType, U extends TNumber> Prod<T> create(Scope scope, Operand<T> input, Operand<U> axis, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Prod", scope.makeOpName("Prod"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(axis.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.keepDims != null) {
+          opBuilder.setAttr("keep_dims", opts.keepDims);
+        }
+      }
+    }
+    return new Prod<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param keepDims If true, retain reduced dimensions with length 1.
+   */
+  public static Options keepDims(Boolean keepDims) {
+    return new Options().keepDims(keepDims);
+  }
+  
+  /**
+   * The reduced tensor.
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Prod";
+  
+  private Output<T> output;
+  
+  private Prod(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/QuantizedReshape.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/QuantizedReshape.java
new file mode 100644
index 00000000000..e644855bbc9
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/QuantizedReshape.java
@@ -0,0 +1,97 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Reshapes a quantized tensor as per the Reshape op.
+ * <p>
+ * ```
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator
+public final class QuantizedReshape<T extends TType> extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new QuantizedReshape operation.
+   * 
+   * @param scope current scope
+   * @param tensor 
+   * @param shape Defines the shape of the output tensor.
+   * @param inputMin The minimum value of the input.
+   * @param inputMax The maximum value of the input.
+   * @return a new instance of QuantizedReshape
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType, U extends TNumber> QuantizedReshape<T> create(Scope scope, Operand<T> tensor, Operand<U> shape, Operand<TFloat32> inputMin, Operand<TFloat32> inputMax) {
+    OperationBuilder opBuilder = scope.env().opBuilder("QuantizedReshape", scope.makeOpName("QuantizedReshape"));
+    opBuilder.addInput(tensor.asOutput());
+    opBuilder.addInput(shape.asOutput());
+    opBuilder.addInput(inputMin.asOutput());
+    opBuilder.addInput(inputMax.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new QuantizedReshape<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  /**
+   * This value is copied from input_min.
+   */
+  public Output<TFloat32> outputMin() {
+    return outputMin;
+  }
+  
+  /**
+   * This value is copied from input_max.
+   */
+  public Output<TFloat32> outputMax() {
+    return outputMax;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "QuantizedReshape";
+  
+  private Output<T> output;
+  private Output<TFloat32> outputMin;
+  private Output<TFloat32> outputMax;
+  
+  private QuantizedReshape(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+    outputMin = operation.output(outputIdx++);
+    outputMax = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Range.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Range.java
new file mode 100644
index 00000000000..4f71eafe3c0
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Range.java
@@ -0,0 +1,91 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Creates a sequence of numbers.
+ * <p>
+ * This operation creates a sequence of numbers that begins at `start` and
+ * extends by increments of `delta` up to but not including `limit`.
+ * <p>
+ * For example:
+ * <pre>{@code
+ * # 'start' is 3
+ * # 'limit' is 18
+ * # 'delta' is 3
+ * tf.range(start, limit, delta) ==> [3, 6, 9, 12, 15]
+ * }</pre>
+ * 
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator
+public final class Range<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new Range operation.
+   * 
+   * @param scope current scope
+   * @param start 0-D (scalar). First entry in the sequence.
+   * @param limit 0-D (scalar). Upper limit of sequence, exclusive.
+   * @param delta 0-D (scalar). Optional. Default is 1. Number that increments `start`.
+   * @return a new instance of Range
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> Range<T> create(Scope scope, Operand<T> start, Operand<T> limit, Operand<T> delta) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Range", scope.makeOpName("Range"));
+    opBuilder.addInput(start.asOutput());
+    opBuilder.addInput(limit.asOutput());
+    opBuilder.addInput(delta.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new Range<T>(opBuilder.build());
+  }
+  
+  /**
+   * 1-D.
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Range";
+  
+  private Output<T> output;
+  
+  private Range(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Rank.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Rank.java
new file mode 100644
index 00000000000..071551d6492
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Rank.java
@@ -0,0 +1,85 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Returns the rank of a tensor.
+ * <p>
+ * This operation returns an integer representing the rank of `input`.
+ * <p>
+ * For example:
+ * <pre>{@code
+ * # 't' is [[[1, 1, 1], [2, 2, 2]], [[3, 3, 3], [4, 4, 4]]]
+ * # shape of tensor 't' is [2, 2, 3]
+ * rank(t) ==> 3
+ * }</pre>
+ * <b>Note</b>: The rank of a tensor is not the same as the rank of a matrix. The rank
+ * of a tensor is the number of indices required to uniquely select each element
+ * of the tensor. Rank is also known as "order", "degree", or "ndims."
+ */
+@Operator
+public final class Rank extends RawOp implements Operand<TInt32> {
+  
+  /**
+   * Factory method to create a class wrapping a new Rank operation.
+   * 
+   * @param scope current scope
+   * @param input 
+   * @return a new instance of Rank
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> Rank create(Scope scope, Operand<T> input) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Rank", scope.makeOpName("Rank"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new Rank(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<TInt32> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<TInt32> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Rank";
+  
+  private Output<TInt32> output;
+  
+  private Rank(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ReadVariableOp.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ReadVariableOp.java
new file mode 100644
index 00000000000..6f33217ec56
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ReadVariableOp.java
@@ -0,0 +1,83 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Reads the value of a variable.
+ * <p>
+ * The tensor returned by this operation is immutable.
+ * <p>
+ * The value returned by this operation is guaranteed to be influenced by all the
+ * writes on which this operation depends directly or indirectly, and to not be
+ * influenced by any of the writes which depend directly or indirectly on this
+ * operation.
+ * 
+ * @param <T> data type for {@code value()} output
+ */
+@Operator
+public final class ReadVariableOp<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new ReadVariableOp operation.
+   * 
+   * @param scope current scope
+   * @param resource handle to the resource in which to store the variable.
+   * @param dtype the dtype of the value.
+   * @return a new instance of ReadVariableOp
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> ReadVariableOp<T> create(Scope scope, Operand<?> resource, Class<T> dtype) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ReadVariableOp", scope.makeOpName("ReadVariableOp"));
+    opBuilder.addInput(resource.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("dtype", dtype);
+    return new ReadVariableOp<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> value() {
+    return value;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return value;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ReadVariableOp";
+  
+  private Output<T> value;
+  
+  private ReadVariableOp(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    value = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Recv.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Recv.java
new file mode 100644
index 00000000000..81ebf79a0e8
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Recv.java
@@ -0,0 +1,122 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Receives the named tensor from send_device on recv_device.
+ * 
+ * @param <T> data type for {@code tensor()} output
+ */
+public final class Recv<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.core.Recv}
+   */
+  public static class Options {
+    
+    /**
+     * @param clientTerminated If set to true, this indicates that the node was added
+     * to the graph as a result of a client-side feed or fetch of Tensor data,
+     * in which case the corresponding send or recv is expected to be managed
+     * locally by the caller.
+     */
+    public Options clientTerminated(Boolean clientTerminated) {
+      this.clientTerminated = clientTerminated;
+      return this;
+    }
+    
+    private Boolean clientTerminated;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new Recv operation.
+   * 
+   * @param scope current scope
+   * @param tensorType 
+   * @param tensorName The name of the tensor to receive.
+   * @param sendDevice The name of the device sending the tensor.
+   * @param sendDeviceIncarnation The current incarnation of send_device.
+   * @param recvDevice The name of the device receiving the tensor.
+   * @param options carries optional attributes values
+   * @return a new instance of Recv
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> Recv<T> create(Scope scope, Class<T> tensorType, String tensorName, String sendDevice, Long sendDeviceIncarnation, String recvDevice, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Recv", scope.makeOpName("Recv"));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("tensor_type", tensorType);
+    opBuilder.setAttr("tensor_name", tensorName);
+    opBuilder.setAttr("send_device", sendDevice);
+    opBuilder.setAttr("send_device_incarnation", sendDeviceIncarnation);
+    opBuilder.setAttr("recv_device", recvDevice);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.clientTerminated != null) {
+          opBuilder.setAttr("client_terminated", opts.clientTerminated);
+        }
+      }
+    }
+    return new Recv<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param clientTerminated If set to true, this indicates that the node was added
+   * to the graph as a result of a client-side feed or fetch of Tensor data,
+   * in which case the corresponding send or recv is expected to be managed
+   * locally by the caller.
+   */
+  public static Options clientTerminated(Boolean clientTerminated) {
+    return new Options().clientTerminated(clientTerminated);
+  }
+  
+  /**
+   * The tensor to receive.
+   */
+  public Output<T> tensor() {
+    return tensor;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return tensor;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Recv";
+  
+  private Output<T> tensor;
+  
+  private Recv(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    tensor = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ReduceAll.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ReduceAll.java
new file mode 100644
index 00000000000..1e940c95fb5
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ReduceAll.java
@@ -0,0 +1,116 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TBool;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Computes the "logical and" of elements across dimensions of a tensor.
+ * <p>
+ * Reduces `input` along the dimensions given in `axis`. Unless
+ * `keep_dims` is true, the rank of the tensor is reduced by 1 for each entry in
+ * `axis`. If `keep_dims` is true, the reduced dimensions are
+ * retained with length 1.
+ */
+@Operator
+public final class ReduceAll extends RawOp implements Operand<TBool> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.core.ReduceAll}
+   */
+  public static class Options {
+    
+    /**
+     * @param keepDims If true, retain reduced dimensions with length 1.
+     */
+    public Options keepDims(Boolean keepDims) {
+      this.keepDims = keepDims;
+      return this;
+    }
+    
+    private Boolean keepDims;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new ReduceAll operation.
+   * 
+   * @param scope current scope
+   * @param input The tensor to reduce.
+   * @param axis The dimensions to reduce. Must be in the range
+   * `[-rank(input), rank(input))`.
+   * @param options carries optional attributes values
+   * @return a new instance of ReduceAll
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> ReduceAll create(Scope scope, Operand<TBool> input, Operand<T> axis, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("All", scope.makeOpName("ReduceAll"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(axis.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.keepDims != null) {
+          opBuilder.setAttr("keep_dims", opts.keepDims);
+        }
+      }
+    }
+    return new ReduceAll(opBuilder.build());
+  }
+  
+  /**
+   * @param keepDims If true, retain reduced dimensions with length 1.
+   */
+  public static Options keepDims(Boolean keepDims) {
+    return new Options().keepDims(keepDims);
+  }
+  
+  /**
+   * The reduced tensor.
+   */
+  public Output<TBool> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<TBool> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "All";
+  
+  private Output<TBool> output;
+  
+  private ReduceAll(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ReduceAny.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ReduceAny.java
new file mode 100644
index 00000000000..412e00fd07e
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ReduceAny.java
@@ -0,0 +1,116 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TBool;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Computes the "logical or" of elements across dimensions of a tensor.
+ * <p>
+ * Reduces `input` along the dimensions given in `axis`. Unless
+ * `keep_dims` is true, the rank of the tensor is reduced by 1 for each entry in
+ * `axis`. If `keep_dims` is true, the reduced dimensions are
+ * retained with length 1.
+ */
+@Operator
+public final class ReduceAny extends RawOp implements Operand<TBool> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.core.ReduceAny}
+   */
+  public static class Options {
+    
+    /**
+     * @param keepDims If true, retain reduced dimensions with length 1.
+     */
+    public Options keepDims(Boolean keepDims) {
+      this.keepDims = keepDims;
+      return this;
+    }
+    
+    private Boolean keepDims;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new ReduceAny operation.
+   * 
+   * @param scope current scope
+   * @param input The tensor to reduce.
+   * @param axis The dimensions to reduce. Must be in the range
+   * `[-rank(input), rank(input))`.
+   * @param options carries optional attributes values
+   * @return a new instance of ReduceAny
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> ReduceAny create(Scope scope, Operand<TBool> input, Operand<T> axis, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Any", scope.makeOpName("ReduceAny"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(axis.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.keepDims != null) {
+          opBuilder.setAttr("keep_dims", opts.keepDims);
+        }
+      }
+    }
+    return new ReduceAny(opBuilder.build());
+  }
+  
+  /**
+   * @param keepDims If true, retain reduced dimensions with length 1.
+   */
+  public static Options keepDims(Boolean keepDims) {
+    return new Options().keepDims(keepDims);
+  }
+  
+  /**
+   * The reduced tensor.
+   */
+  public Output<TBool> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<TBool> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Any";
+  
+  private Output<TBool> output;
+  
+  private ReduceAny(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ReduceMax.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ReduceMax.java
new file mode 100644
index 00000000000..997481542b8
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ReduceMax.java
@@ -0,0 +1,118 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Computes the maximum of elements across dimensions of a tensor.
+ * <p>
+ * Reduces `input` along the dimensions given in `axis`. Unless
+ * `keep_dims` is true, the rank of the tensor is reduced by 1 for each entry in
+ * `axis`. If `keep_dims` is true, the reduced dimensions are
+ * retained with length 1.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator
+public final class ReduceMax<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.core.ReduceMax}
+   */
+  public static class Options {
+    
+    /**
+     * @param keepDims If true, retain reduced dimensions with length 1.
+     */
+    public Options keepDims(Boolean keepDims) {
+      this.keepDims = keepDims;
+      return this;
+    }
+    
+    private Boolean keepDims;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new ReduceMax operation.
+   * 
+   * @param scope current scope
+   * @param input The tensor to reduce.
+   * @param axis The dimensions to reduce. Must be in the range
+   * `[-rank(input), rank(input))`.
+   * @param options carries optional attributes values
+   * @return a new instance of ReduceMax
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType, U extends TNumber> ReduceMax<T> create(Scope scope, Operand<T> input, Operand<U> axis, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Max", scope.makeOpName("ReduceMax"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(axis.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.keepDims != null) {
+          opBuilder.setAttr("keep_dims", opts.keepDims);
+        }
+      }
+    }
+    return new ReduceMax<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param keepDims If true, retain reduced dimensions with length 1.
+   */
+  public static Options keepDims(Boolean keepDims) {
+    return new Options().keepDims(keepDims);
+  }
+  
+  /**
+   * The reduced tensor.
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Max";
+  
+  private Output<T> output;
+  
+  private ReduceMax(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ReduceMin.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ReduceMin.java
new file mode 100644
index 00000000000..60d64c3a58c
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ReduceMin.java
@@ -0,0 +1,118 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Computes the minimum of elements across dimensions of a tensor.
+ * <p>
+ * Reduces `input` along the dimensions given in `axis`. Unless
+ * `keep_dims` is true, the rank of the tensor is reduced by 1 for each entry in
+ * `axis`. If `keep_dims` is true, the reduced dimensions are
+ * retained with length 1.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator
+public final class ReduceMin<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.core.ReduceMin}
+   */
+  public static class Options {
+    
+    /**
+     * @param keepDims If true, retain reduced dimensions with length 1.
+     */
+    public Options keepDims(Boolean keepDims) {
+      this.keepDims = keepDims;
+      return this;
+    }
+    
+    private Boolean keepDims;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new ReduceMin operation.
+   * 
+   * @param scope current scope
+   * @param input The tensor to reduce.
+   * @param axis The dimensions to reduce. Must be in the range
+   * `[-rank(input), rank(input))`.
+   * @param options carries optional attributes values
+   * @return a new instance of ReduceMin
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType, U extends TNumber> ReduceMin<T> create(Scope scope, Operand<T> input, Operand<U> axis, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Min", scope.makeOpName("ReduceMin"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(axis.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.keepDims != null) {
+          opBuilder.setAttr("keep_dims", opts.keepDims);
+        }
+      }
+    }
+    return new ReduceMin<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param keepDims If true, retain reduced dimensions with length 1.
+   */
+  public static Options keepDims(Boolean keepDims) {
+    return new Options().keepDims(keepDims);
+  }
+  
+  /**
+   * The reduced tensor.
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Min";
+  
+  private Output<T> output;
+  
+  private ReduceMin(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ReduceProd.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ReduceProd.java
new file mode 100644
index 00000000000..7c9872758b1
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ReduceProd.java
@@ -0,0 +1,118 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Computes the product of elements across dimensions of a tensor.
+ * <p>
+ * Reduces `input` along the dimensions given in `axis`. Unless
+ * `keep_dims` is true, the rank of the tensor is reduced by 1 for each entry in
+ * `axis`. If `keep_dims` is true, the reduced dimensions are
+ * retained with length 1.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator
+public final class ReduceProd<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.core.ReduceProd}
+   */
+  public static class Options {
+    
+    /**
+     * @param keepDims If true, retain reduced dimensions with length 1.
+     */
+    public Options keepDims(Boolean keepDims) {
+      this.keepDims = keepDims;
+      return this;
+    }
+    
+    private Boolean keepDims;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new ReduceProd operation.
+   * 
+   * @param scope current scope
+   * @param input The tensor to reduce.
+   * @param axis The dimensions to reduce. Must be in the range
+   * `[-rank(input), rank(input))`.
+   * @param options carries optional attributes values
+   * @return a new instance of ReduceProd
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType, U extends TNumber> ReduceProd<T> create(Scope scope, Operand<T> input, Operand<U> axis, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Prod", scope.makeOpName("ReduceProd"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(axis.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.keepDims != null) {
+          opBuilder.setAttr("keep_dims", opts.keepDims);
+        }
+      }
+    }
+    return new ReduceProd<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param keepDims If true, retain reduced dimensions with length 1.
+   */
+  public static Options keepDims(Boolean keepDims) {
+    return new Options().keepDims(keepDims);
+  }
+  
+  /**
+   * The reduced tensor.
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Prod";
+  
+  private Output<T> output;
+  
+  private ReduceProd(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ReduceSum.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ReduceSum.java
new file mode 100644
index 00000000000..f87e5b22e15
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ReduceSum.java
@@ -0,0 +1,118 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Computes the sum of elements across dimensions of a tensor.
+ * <p>
+ * Reduces `input` along the dimensions given in `axis`. Unless
+ * `keep_dims` is true, the rank of the tensor is reduced by 1 for each entry in
+ * `axis`. If `keep_dims` is true, the reduced dimensions are
+ * retained with length 1.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator
+public final class ReduceSum<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.core.ReduceSum}
+   */
+  public static class Options {
+    
+    /**
+     * @param keepDims If true, retain reduced dimensions with length 1.
+     */
+    public Options keepDims(Boolean keepDims) {
+      this.keepDims = keepDims;
+      return this;
+    }
+    
+    private Boolean keepDims;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new ReduceSum operation.
+   * 
+   * @param scope current scope
+   * @param input The tensor to reduce.
+   * @param axis The dimensions to reduce. Must be in the range
+   * `[-rank(input), rank(input))`.
+   * @param options carries optional attributes values
+   * @return a new instance of ReduceSum
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType, U extends TNumber> ReduceSum<T> create(Scope scope, Operand<T> input, Operand<U> axis, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Sum", scope.makeOpName("ReduceSum"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(axis.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.keepDims != null) {
+          opBuilder.setAttr("keep_dims", opts.keepDims);
+        }
+      }
+    }
+    return new ReduceSum<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param keepDims If true, retain reduced dimensions with length 1.
+   */
+  public static Options keepDims(Boolean keepDims) {
+    return new Options().keepDims(keepDims);
+  }
+  
+  /**
+   * The reduced tensor.
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Sum";
+  
+  private Output<T> output;
+  
+  private ReduceSum(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/RefEnter.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/RefEnter.java
new file mode 100644
index 00000000000..8292dcbb379
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/RefEnter.java
@@ -0,0 +1,134 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Creates or finds a child frame, and makes `data` available to the child frame.
+ * <p>
+ * The unique `frame_name` is used by the `Executor` to identify frames. If
+ * `is_constant` is true, `output` is a constant in the child frame; otherwise
+ * it may be changed in the child frame. At most `parallel_iterations` iterations
+ * are run in parallel in the child frame.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+public final class RefEnter<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.core.RefEnter}
+   */
+  public static class Options {
+    
+    /**
+     * @param isConstant If true, the output is constant within the child frame.
+     */
+    public Options isConstant(Boolean isConstant) {
+      this.isConstant = isConstant;
+      return this;
+    }
+    
+    /**
+     * @param parallelIterations The number of iterations allowed to run in parallel.
+     */
+    public Options parallelIterations(Long parallelIterations) {
+      this.parallelIterations = parallelIterations;
+      return this;
+    }
+    
+    private Boolean isConstant;
+    private Long parallelIterations;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new RefEnter operation.
+   * 
+   * @param scope current scope
+   * @param data The tensor to be made available to the child frame.
+   * @param frameName The name of the child frame.
+   * @param options carries optional attributes values
+   * @return a new instance of RefEnter
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> RefEnter<T> create(Scope scope, Operand<T> data, String frameName, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("RefEnter", scope.makeOpName("RefEnter"));
+    opBuilder.addInput(data.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("frame_name", frameName);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.isConstant != null) {
+          opBuilder.setAttr("is_constant", opts.isConstant);
+        }
+        if (opts.parallelIterations != null) {
+          opBuilder.setAttr("parallel_iterations", opts.parallelIterations);
+        }
+      }
+    }
+    return new RefEnter<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param isConstant If true, the output is constant within the child frame.
+   */
+  public static Options isConstant(Boolean isConstant) {
+    return new Options().isConstant(isConstant);
+  }
+  
+  /**
+   * @param parallelIterations The number of iterations allowed to run in parallel.
+   */
+  public static Options parallelIterations(Long parallelIterations) {
+    return new Options().parallelIterations(parallelIterations);
+  }
+  
+  /**
+   * The same tensor as `data`.
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "RefEnter";
+  
+  private Output<T> output;
+  
+  private RefEnter(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/RefExit.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/RefExit.java
new file mode 100644
index 00000000000..d8896a59211
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/RefExit.java
@@ -0,0 +1,76 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Exits the current frame to its parent frame.
+ * <p>
+ * Exit makes its input `data` available to the parent frame.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+public final class RefExit<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new RefExit operation.
+   * 
+   * @param scope current scope
+   * @param data The tensor to be made available to the parent frame.
+   * @return a new instance of RefExit
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> RefExit<T> create(Scope scope, Operand<T> data) {
+    OperationBuilder opBuilder = scope.env().opBuilder("RefExit", scope.makeOpName("RefExit"));
+    opBuilder.addInput(data.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new RefExit<T>(opBuilder.build());
+  }
+  
+  /**
+   * The same tensor as `data`.
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "RefExit";
+  
+  private Output<T> output;
+  
+  private RefExit(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/RefIdentity.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/RefIdentity.java
new file mode 100644
index 00000000000..9deff66c3a9
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/RefIdentity.java
@@ -0,0 +1,73 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Return the same ref tensor as the input ref tensor.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+public final class RefIdentity<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new RefIdentity operation.
+   * 
+   * @param scope current scope
+   * @param input 
+   * @return a new instance of RefIdentity
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> RefIdentity<T> create(Scope scope, Operand<T> input) {
+    OperationBuilder opBuilder = scope.env().opBuilder("RefIdentity", scope.makeOpName("RefIdentity"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new RefIdentity<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "RefIdentity";
+  
+  private Output<T> output;
+  
+  private RefIdentity(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/RefMerge.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/RefMerge.java
new file mode 100644
index 00000000000..990cfaaadb6
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/RefMerge.java
@@ -0,0 +1,86 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.Operands;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Forwards the value of an available tensor from `inputs` to `output`.
+ * <p>
+ * `Merge` waits for at least one of the tensors in `inputs` to become available.
+ * It is usually combined with `Switch` to implement branching.
+ * <p>
+ * `Merge` forwards the first tensor for become available to `output`, and sets
+ * `value_index` to its index in `inputs`.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+public final class RefMerge<T extends TType> extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new RefMerge operation.
+   * 
+   * @param scope current scope
+   * @param inputs The input tensors, exactly one of which will become available.
+   * @return a new instance of RefMerge
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> RefMerge<T> create(Scope scope, Iterable<Operand<T>> inputs) {
+    OperationBuilder opBuilder = scope.env().opBuilder("RefMerge", scope.makeOpName("RefMerge"));
+    opBuilder.addInputList(Operands.asOutputs(inputs));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new RefMerge<T>(opBuilder.build());
+  }
+  
+  /**
+   * Will be set to the available input tensor.
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  /**
+   * The index of the chosen input tensor in `inputs`.
+   */
+  public Output<TInt32> valueIndex() {
+    return valueIndex;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "RefMerge";
+  
+  private Output<T> output;
+  private Output<TInt32> valueIndex;
+  
+  private RefMerge(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+    valueIndex = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/RefNextIteration.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/RefNextIteration.java
new file mode 100644
index 00000000000..c151662820e
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/RefNextIteration.java
@@ -0,0 +1,75 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Makes its input available to the next iteration.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator
+public final class RefNextIteration<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new RefNextIteration operation.
+   * 
+   * @param scope current scope
+   * @param data The tensor to be made available to the next iteration.
+   * @return a new instance of RefNextIteration
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> RefNextIteration<T> create(Scope scope, Operand<T> data) {
+    OperationBuilder opBuilder = scope.env().opBuilder("RefNextIteration", scope.makeOpName("RefNextIteration"));
+    opBuilder.addInput(data.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new RefNextIteration<T>(opBuilder.build());
+  }
+  
+  /**
+   * The same tensor as `data`.
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "RefNextIteration";
+  
+  private Output<T> output;
+  
+  private RefNextIteration(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/RefSelect.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/RefSelect.java
new file mode 100644
index 00000000000..8be3ae40e9e
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/RefSelect.java
@@ -0,0 +1,79 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.Operands;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Forwards the `index`th element of `inputs` to `output`.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator
+public final class RefSelect<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new RefSelect operation.
+   * 
+   * @param scope current scope
+   * @param index A scalar that determines the input that gets selected.
+   * @param inputs A list of ref tensors, one of which will be forwarded to `output`.
+   * @return a new instance of RefSelect
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> RefSelect<T> create(Scope scope, Operand<TInt32> index, Iterable<Operand<T>> inputs) {
+    OperationBuilder opBuilder = scope.env().opBuilder("RefSelect", scope.makeOpName("RefSelect"));
+    opBuilder.addInput(index.asOutput());
+    opBuilder.addInputList(Operands.asOutputs(inputs));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new RefSelect<T>(opBuilder.build());
+  }
+  
+  /**
+   * The forwarded tensor.
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "RefSelect";
+  
+  private Output<T> output;
+  
+  private RefSelect(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/RefSwitch.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/RefSwitch.java
new file mode 100644
index 00000000000..f0128f668fc
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/RefSwitch.java
@@ -0,0 +1,87 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TBool;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Forwards the ref tensor `data` to the output port determined by `pred`.
+ * <p>
+ * If `pred` is true, the `data` input is forwarded to `output_true`. Otherwise,
+ * the data goes to `output_false`.
+ * <p>
+ * See also `Switch` and `Merge`.
+ * 
+ * @param <T> data type for {@code outputFalse()} output
+ */
+@Operator
+public final class RefSwitch<T extends TType> extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new RefSwitch operation.
+   * 
+   * @param scope current scope
+   * @param data The ref tensor to be forwarded to the appropriate output.
+   * @param pred A scalar that specifies which output port will receive data.
+   * @return a new instance of RefSwitch
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> RefSwitch<T> create(Scope scope, Operand<T> data, Operand<TBool> pred) {
+    OperationBuilder opBuilder = scope.env().opBuilder("RefSwitch", scope.makeOpName("RefSwitch"));
+    opBuilder.addInput(data.asOutput());
+    opBuilder.addInput(pred.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new RefSwitch<T>(opBuilder.build());
+  }
+  
+  /**
+   * If `pred` is false, data will be forwarded to this output.
+   */
+  public Output<T> outputFalse() {
+    return outputFalse;
+  }
+  
+  /**
+   * If `pred` is true, data will be forwarded to this output.
+   */
+  public Output<T> outputTrue() {
+    return outputTrue;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "RefSwitch";
+  
+  private Output<T> outputFalse;
+  private Output<T> outputTrue;
+  
+  private RefSwitch(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    outputFalse = operation.output(outputIdx++);
+    outputTrue = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/RemoteFusedGraphExecute.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/RemoteFusedGraphExecute.java
new file mode 100644
index 00000000000..f1a70aa35e6
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/RemoteFusedGraphExecute.java
@@ -0,0 +1,97 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import java.util.Arrays;
+import java.util.Iterator;
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.Operands;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Execute a sub graph on a remote processor.
+ * <p>
+ * The graph specifications(such as graph itself, input tensors and output names)
+ * are stored as a serialized protocol buffer of RemoteFusedGraphExecuteInfo
+ * as serialized_remote_fused_graph_execute_info.
+ * The specifications will be passed to a dedicated registered
+ * remote fused graph executor.  The executor will send the graph specifications
+ * to a remote processor and execute that graph.  The execution results
+ * will be passed to consumer nodes as outputs of this node.
+ */
+@Operator
+public final class RemoteFusedGraphExecute extends RawOp implements Iterable<Operand<TType>> {
+  
+  /**
+   * Factory method to create a class wrapping a new RemoteFusedGraphExecute operation.
+   * 
+   * @param scope current scope
+   * @param inputs Arbitrary number of tensors with arbitrary data types
+   * @param Toutputs 
+   * @param serializedRemoteFusedGraphExecuteInfo Serialized protocol buffer
+   * of RemoteFusedGraphExecuteInfo which contains graph specifications.
+   * @return a new instance of RemoteFusedGraphExecute
+   */
+  @Endpoint(describeByClass = true)
+  public static RemoteFusedGraphExecute create(Scope scope, Iterable<Operand<?>> inputs, List<Class<? extends TType>> Toutputs, String serializedRemoteFusedGraphExecuteInfo) {
+    OperationBuilder opBuilder = scope.env().opBuilder("RemoteFusedGraphExecute", scope.makeOpName("RemoteFusedGraphExecute"));
+    opBuilder.addInputList(Operands.asOutputs(inputs));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    Class[] ToutputsArray = new Class[Toutputs.size()];
+    for (int i = 0; i < ToutputsArray.length; ++i) {
+      ToutputsArray[i] = Toutputs.get(i);
+    }
+    opBuilder.setAttr("Toutputs", ToutputsArray);
+    opBuilder.setAttr("serialized_remote_fused_graph_execute_info", serializedRemoteFusedGraphExecuteInfo);
+    return new RemoteFusedGraphExecute(opBuilder.build());
+  }
+  
+  /**
+   * Arbitrary number of tensors with arbitrary data types
+   */
+  public List<Output<?>> outputs() {
+    return outputs;
+  }
+  
+  @Override
+  @SuppressWarnings({"rawtypes", "unchecked"})
+  public Iterator<Operand<TType>> iterator() {
+    return (Iterator) outputs.iterator();
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "RemoteFusedGraphExecute";
+  
+  private List<Output<?>> outputs;
+  
+  private RemoteFusedGraphExecute(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    int outputsLength = operation.outputListLength("outputs");
+    outputs = Arrays.asList(operation.outputList(outputIdx, outputsLength));
+    outputIdx += outputsLength;
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Reshape.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Reshape.java
new file mode 100644
index 00000000000..2829a332b0e
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Reshape.java
@@ -0,0 +1,137 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Reshapes a tensor.
+ * <p>
+ * Given `tensor`, this operation returns a tensor that has the same values
+ * as `tensor` with shape `shape`.
+ * <p>
+ * If one component of 1-D tensor `shape` is the special value -1, the size of that
+ * dimension is computed so that the total size remains constant.  In particular, a
+ * `shape` of `[-1]` flattens into 1-D.  At most one component of `shape` may be
+ * unknown.
+ * <p>
+ * The `shape` must be 1-D and the operation returns a tensor with shape
+ * `shape` filled with the values of `tensor`. In this case, the number of elements
+ * implied by `shape` must be the same as the number of elements in `tensor`.
+ * <p>
+ * It is an error if `shape` is not 1-D.
+ * <p>
+ * For example:
+ * <pre>{@code
+ * # tensor 't' is [1, 2, 3, 4, 5, 6, 7, 8, 9]
+ * # tensor 't' has shape [9]
+ * reshape(t, [3, 3]) ==> [[1, 2, 3],
+ *                         [4, 5, 6],
+ *                         [7, 8, 9]]
+ * 
+ * # tensor 't' is [[[1, 1], [2, 2]],
+ * #                [[3, 3], [4, 4]]]
+ * # tensor 't' has shape [2, 2, 2]
+ * reshape(t, [2, 4]) ==> [[1, 1, 2, 2],
+ *                         [3, 3, 4, 4]]
+ * 
+ * # tensor 't' is [[[1, 1, 1],
+ * #                 [2, 2, 2]],
+ * #                [[3, 3, 3],
+ * #                 [4, 4, 4]],
+ * #                [[5, 5, 5],
+ * #                 [6, 6, 6]]]
+ * # tensor 't' has shape [3, 2, 3]
+ * # pass '[-1]' to flatten 't'
+ * reshape(t, [-1]) ==> [1, 1, 1, 2, 2, 2, 3, 3, 3, 4, 4, 4, 5, 5, 5, 6, 6, 6]
+ * 
+ * # -1 can also be used to infer the shape
+ * 
+ * # -1 is inferred to be 9:
+ * reshape(t, [2, -1]) ==> [[1, 1, 1, 2, 2, 2, 3, 3, 3],
+ *                          [4, 4, 4, 5, 5, 5, 6, 6, 6]]
+ * # -1 is inferred to be 2:
+ * reshape(t, [-1, 9]) ==> [[1, 1, 1, 2, 2, 2, 3, 3, 3],
+ *                          [4, 4, 4, 5, 5, 5, 6, 6, 6]]
+ * # -1 is inferred to be 3:
+ * reshape(t, [ 2, -1, 3]) ==> [[[1, 1, 1],
+ *                               [2, 2, 2],
+ *                               [3, 3, 3]],
+ *                              [[4, 4, 4],
+ *                               [5, 5, 5],
+ *                               [6, 6, 6]]]
+ * 
+ * # tensor 't' is [7]
+ * # shape `[]` reshapes to a scalar
+ * reshape(t, []) ==> 7
+ * }</pre>
+ * 
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator
+public final class Reshape<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new Reshape operation.
+   * 
+   * @param scope current scope
+   * @param tensor 
+   * @param shape Defines the shape of the output tensor.
+   * @return a new instance of Reshape
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType, U extends TNumber> Reshape<T> create(Scope scope, Operand<T> tensor, Operand<U> shape) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Reshape", scope.makeOpName("Reshape"));
+    opBuilder.addInput(tensor.asOutput());
+    opBuilder.addInput(shape.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new Reshape<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Reshape";
+  
+  private Output<T> output;
+  
+  private Reshape(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceCountUpTo.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceCountUpTo.java
new file mode 100644
index 00000000000..072c70a1862
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceCountUpTo.java
@@ -0,0 +1,81 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Increments variable pointed to by 'resource' until it reaches 'limit'.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator
+public final class ResourceCountUpTo<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new ResourceCountUpTo operation.
+   * 
+   * @param scope current scope
+   * @param resource Should be from a scalar `Variable` node.
+   * @param limit If incrementing ref would bring it above limit, instead generates an
+   * 'OutOfRange' error.
+   * @param T 
+   * @return a new instance of ResourceCountUpTo
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> ResourceCountUpTo<T> create(Scope scope, Operand<?> resource, Long limit, Class<T> T) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ResourceCountUpTo", scope.makeOpName("ResourceCountUpTo"));
+    opBuilder.addInput(resource.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("limit", limit);
+    opBuilder.setAttr("T", T);
+    return new ResourceCountUpTo<T>(opBuilder.build());
+  }
+  
+  /**
+   * A copy of the input before increment. If nothing else modifies the
+   * input, the values produced will all be distinct.
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ResourceCountUpTo";
+  
+  private Output<T> output;
+  
+  private ResourceCountUpTo(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceGather.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceGather.java
new file mode 100644
index 00000000000..d470cf09127
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceGather.java
@@ -0,0 +1,146 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Gather slices from the variable pointed to by `resource` according to `indices`.
+ * <p>
+ * `indices` must be an integer tensor of any dimension (usually 0-D or 1-D).
+ * Produces an output tensor with shape `indices.shape + params.shape[1:]` where:
+ * <pre>{@code
+ *     # Scalar indices
+ *     output[:, ..., :] = params[indices, :, ... :]
+ * 
+ *     # Vector indices
+ *     output[i, :, ..., :] = params[indices[i], :, ... :]
+ * 
+ *     # Higher rank indices
+ *     output[i, ..., j, :, ... :] = params[indices[i, ..., j], :, ..., :]
+ * }</pre>
+ * 
+ * 
+ * @param <U> data type for {@code output()} output
+ */
+@Operator
+public final class ResourceGather<U extends TType> extends RawOp implements Operand<U> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.core.ResourceGather}
+   */
+  public static class Options {
+    
+    /**
+     * @param batchDims 
+     */
+    public Options batchDims(Long batchDims) {
+      this.batchDims = batchDims;
+      return this;
+    }
+    
+    /**
+     * @param validateIndices 
+     */
+    public Options validateIndices(Boolean validateIndices) {
+      this.validateIndices = validateIndices;
+      return this;
+    }
+    
+    private Long batchDims;
+    private Boolean validateIndices;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new ResourceGather operation.
+   * 
+   * @param scope current scope
+   * @param resource 
+   * @param indices 
+   * @param dtype 
+   * @param options carries optional attributes values
+   * @return a new instance of ResourceGather
+   */
+  @Endpoint(describeByClass = true)
+  public static <U extends TType, T extends TNumber> ResourceGather<U> create(Scope scope, Operand<?> resource, Operand<T> indices, Class<U> dtype, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ResourceGather", scope.makeOpName("ResourceGather"));
+    opBuilder.addInput(resource.asOutput());
+    opBuilder.addInput(indices.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("dtype", dtype);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.batchDims != null) {
+          opBuilder.setAttr("batch_dims", opts.batchDims);
+        }
+        if (opts.validateIndices != null) {
+          opBuilder.setAttr("validate_indices", opts.validateIndices);
+        }
+      }
+    }
+    return new ResourceGather<U>(opBuilder.build());
+  }
+  
+  /**
+   * @param batchDims 
+   */
+  public static Options batchDims(Long batchDims) {
+    return new Options().batchDims(batchDims);
+  }
+  
+  /**
+   * @param validateIndices 
+   */
+  public static Options validateIndices(Boolean validateIndices) {
+    return new Options().validateIndices(validateIndices);
+  }
+  
+  /**
+   */
+  public Output<U> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<U> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ResourceGather";
+  
+  private Output<U> output;
+  
+  private ResourceGather(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceGatherNd.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceGatherNd.java
new file mode 100644
index 00000000000..3f8237d98e6
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceGatherNd.java
@@ -0,0 +1,77 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * @param <U> data type for {@code output()} output
+ */
+@Operator
+public final class ResourceGatherNd<U extends TType> extends RawOp implements Operand<U> {
+  
+  /**
+   * Factory method to create a class wrapping a new ResourceGatherNd operation.
+   * 
+   * @param scope current scope
+   * @param resource 
+   * @param indices 
+   * @param dtype 
+   * @return a new instance of ResourceGatherNd
+   */
+  @Endpoint(describeByClass = true)
+  public static <U extends TType, T extends TNumber> ResourceGatherNd<U> create(Scope scope, Operand<?> resource, Operand<T> indices, Class<U> dtype) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ResourceGatherNd", scope.makeOpName("ResourceGatherNd"));
+    opBuilder.addInput(resource.asOutput());
+    opBuilder.addInput(indices.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("dtype", dtype);
+    return new ResourceGatherNd<U>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<U> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<U> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ResourceGatherNd";
+  
+  private Output<U> output;
+  
+  private ResourceGatherNd(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterAdd.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterAdd.java
new file mode 100644
index 00000000000..0966dd5fcc4
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterAdd.java
@@ -0,0 +1,81 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Adds sparse updates to the variable referenced by `resource`.
+ * <p>
+ * This operation computes
+ * <p>
+ *     # Scalar indices
+ *     ref[indices, ...] += updates[...]
+ * <p>
+ *     # Vector indices (for each i)
+ *     ref[indices[i], ...] += updates[i, ...]
+ * <p>
+ *     # High rank indices (for each i, ..., j)
+ *     ref[indices[i, ..., j], ...] += updates[i, ..., j, ...]
+ * <p>
+ * Duplicate entries are handled correctly: if multiple `indices` reference
+ * the same location, their contributions add.
+ * <p>
+ * Requires `updates.shape = indices.shape + ref.shape[1:]` or `updates.shape = []`.
+ * <p>
+ * <div style="width:70%; margin:auto; margin-bottom:10px; margin-top:20px;">
+ * <img style="width:100%" src='https://www.tensorflow.org/images/ScatterAdd.png' alt>
+ * </div>
+ */
+@Operator
+public final class ResourceScatterAdd extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new ResourceScatterAdd operation.
+   * 
+   * @param scope current scope
+   * @param resource Should be from a `Variable` node.
+   * @param indices A tensor of indices into the first dimension of `ref`.
+   * @param updates A tensor of updated values to add to `ref`.
+   * @return a new instance of ResourceScatterAdd
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber, U extends TType> ResourceScatterAdd create(Scope scope, Operand<?> resource, Operand<T> indices, Operand<U> updates) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ResourceScatterAdd", scope.makeOpName("ResourceScatterAdd"));
+    opBuilder.addInput(resource.asOutput());
+    opBuilder.addInput(indices.asOutput());
+    opBuilder.addInput(updates.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new ResourceScatterAdd(opBuilder.build());
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ResourceScatterAdd";
+  
+  private ResourceScatterAdd(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterDiv.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterDiv.java
new file mode 100644
index 00000000000..9560bddf284
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterDiv.java
@@ -0,0 +1,81 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Divides sparse updates into the variable referenced by `resource`.
+ * <p>
+ * This operation computes
+ * <p>
+ *     # Scalar indices
+ *     ref[indices, ...] /= updates[...]
+ * <p>
+ *     # Vector indices (for each i)
+ *     ref[indices[i], ...] /= updates[i, ...]
+ * <p>
+ *     # High rank indices (for each i, ..., j)
+ *     ref[indices[i, ..., j], ...] /= updates[i, ..., j, ...]
+ * <p>
+ * Duplicate entries are handled correctly: if multiple `indices` reference
+ * the same location, their contributions multiply.
+ * <p>
+ * Requires `updates.shape = indices.shape + ref.shape[1:]` or `updates.shape = []`.
+ * <p>
+ * <div style="width:70%; margin:auto; margin-bottom:10px; margin-top:20px;">
+ * <img style="width:100%" src='https://www.tensorflow.org/images/ScatterAdd.png' alt>
+ * </div>
+ */
+@Operator
+public final class ResourceScatterDiv extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new ResourceScatterDiv operation.
+   * 
+   * @param scope current scope
+   * @param resource Should be from a `Variable` node.
+   * @param indices A tensor of indices into the first dimension of `ref`.
+   * @param updates A tensor of updated values to add to `ref`.
+   * @return a new instance of ResourceScatterDiv
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber, U extends TType> ResourceScatterDiv create(Scope scope, Operand<?> resource, Operand<T> indices, Operand<U> updates) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ResourceScatterDiv", scope.makeOpName("ResourceScatterDiv"));
+    opBuilder.addInput(resource.asOutput());
+    opBuilder.addInput(indices.asOutput());
+    opBuilder.addInput(updates.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new ResourceScatterDiv(opBuilder.build());
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ResourceScatterDiv";
+  
+  private ResourceScatterDiv(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterMax.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterMax.java
new file mode 100644
index 00000000000..ce952ee19ba
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterMax.java
@@ -0,0 +1,81 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Reduces sparse updates into the variable referenced by `resource` using the `max` operation.
+ * <p>
+ * This operation computes
+ * <p>
+ *     # Scalar indices
+ *     ref[indices, ...] = max(ref[indices, ...], updates[...])
+ * <p>
+ *     # Vector indices (for each i)
+ *     ref[indices[i], ...] = max(ref[indices[i], ...], updates[i, ...])
+ * <p>
+ *     # High rank indices (for each i, ..., j)
+ *     ref[indices[i, ..., j], ...] = max(ref[indices[i, ..., j], ...], updates[i, ..., j, ...])
+ * <p>
+ * Duplicate entries are handled correctly: if multiple `indices` reference
+ * the same location, their contributions are combined.
+ * <p>
+ * Requires `updates.shape = indices.shape + ref.shape[1:]` or `updates.shape = []`.
+ * <p>
+ * <div style="width:70%; margin:auto; margin-bottom:10px; margin-top:20px;">
+ * <img style="width:100%" src='https://www.tensorflow.org/images/ScatterAdd.png' alt>
+ * </div>
+ */
+@Operator
+public final class ResourceScatterMax extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new ResourceScatterMax operation.
+   * 
+   * @param scope current scope
+   * @param resource Should be from a `Variable` node.
+   * @param indices A tensor of indices into the first dimension of `ref`.
+   * @param updates A tensor of updated values to add to `ref`.
+   * @return a new instance of ResourceScatterMax
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber, U extends TType> ResourceScatterMax create(Scope scope, Operand<?> resource, Operand<T> indices, Operand<U> updates) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ResourceScatterMax", scope.makeOpName("ResourceScatterMax"));
+    opBuilder.addInput(resource.asOutput());
+    opBuilder.addInput(indices.asOutput());
+    opBuilder.addInput(updates.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new ResourceScatterMax(opBuilder.build());
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ResourceScatterMax";
+  
+  private ResourceScatterMax(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterMin.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterMin.java
new file mode 100644
index 00000000000..51ec6b7637e
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterMin.java
@@ -0,0 +1,81 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Reduces sparse updates into the variable referenced by `resource` using the `min` operation.
+ * <p>
+ * This operation computes
+ * <p>
+ *     # Scalar indices
+ *     ref[indices, ...] = min(ref[indices, ...], updates[...])
+ * <p>
+ *     # Vector indices (for each i)
+ *     ref[indices[i], ...] = min(ref[indices[i], ...], updates[i, ...])
+ * <p>
+ *     # High rank indices (for each i, ..., j)
+ *     ref[indices[i, ..., j], ...] = min(ref[indices[i, ..., j], ...], updates[i, ..., j, ...])
+ * <p>
+ * Duplicate entries are handled correctly: if multiple `indices` reference
+ * the same location, their contributions are combined.
+ * <p>
+ * Requires `updates.shape = indices.shape + ref.shape[1:]` or `updates.shape = []`.
+ * <p>
+ * <div style="width:70%; margin:auto; margin-bottom:10px; margin-top:20px;">
+ * <img style="width:100%" src='https://www.tensorflow.org/images/ScatterAdd.png' alt>
+ * </div>
+ */
+@Operator
+public final class ResourceScatterMin extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new ResourceScatterMin operation.
+   * 
+   * @param scope current scope
+   * @param resource Should be from a `Variable` node.
+   * @param indices A tensor of indices into the first dimension of `ref`.
+   * @param updates A tensor of updated values to add to `ref`.
+   * @return a new instance of ResourceScatterMin
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber, U extends TType> ResourceScatterMin create(Scope scope, Operand<?> resource, Operand<T> indices, Operand<U> updates) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ResourceScatterMin", scope.makeOpName("ResourceScatterMin"));
+    opBuilder.addInput(resource.asOutput());
+    opBuilder.addInput(indices.asOutput());
+    opBuilder.addInput(updates.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new ResourceScatterMin(opBuilder.build());
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ResourceScatterMin";
+  
+  private ResourceScatterMin(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterMul.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterMul.java
new file mode 100644
index 00000000000..2d5f71e006d
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterMul.java
@@ -0,0 +1,81 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Multiplies sparse updates into the variable referenced by `resource`.
+ * <p>
+ * This operation computes
+ * <p>
+ *     # Scalar indices
+ *     ref[indices, ...] *= updates[...]
+ * <p>
+ *     # Vector indices (for each i)
+ *     ref[indices[i], ...] *= updates[i, ...]
+ * <p>
+ *     # High rank indices (for each i, ..., j)
+ *     ref[indices[i, ..., j], ...] *= updates[i, ..., j, ...]
+ * <p>
+ * Duplicate entries are handled correctly: if multiple `indices` reference
+ * the same location, their contributions multiply.
+ * <p>
+ * Requires `updates.shape = indices.shape + ref.shape[1:]` or `updates.shape = []`.
+ * <p>
+ * <div style="width:70%; margin:auto; margin-bottom:10px; margin-top:20px;">
+ * <img style="width:100%" src='https://www.tensorflow.org/images/ScatterAdd.png' alt>
+ * </div>
+ */
+@Operator
+public final class ResourceScatterMul extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new ResourceScatterMul operation.
+   * 
+   * @param scope current scope
+   * @param resource Should be from a `Variable` node.
+   * @param indices A tensor of indices into the first dimension of `ref`.
+   * @param updates A tensor of updated values to add to `ref`.
+   * @return a new instance of ResourceScatterMul
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber, U extends TType> ResourceScatterMul create(Scope scope, Operand<?> resource, Operand<T> indices, Operand<U> updates) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ResourceScatterMul", scope.makeOpName("ResourceScatterMul"));
+    opBuilder.addInput(resource.asOutput());
+    opBuilder.addInput(indices.asOutput());
+    opBuilder.addInput(updates.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new ResourceScatterMul(opBuilder.build());
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ResourceScatterMul";
+  
+  private ResourceScatterMul(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterNdAdd.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterNdAdd.java
new file mode 100644
index 00000000000..11e45c33098
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterNdAdd.java
@@ -0,0 +1,131 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Applies sparse addition to individual values or slices in a Variable.
+ * <p>
+ * `ref` is a `Tensor` with rank `P` and `indices` is a `Tensor` of rank `Q`.
+ * <p>
+ * `indices` must be integer tensor, containing indices into `ref`.
+ * It must be shape `[d_0, ..., d_{Q-2}, K]` where `0 < K <= P`.
+ * <p>
+ * The innermost dimension of `indices` (with length `K`) corresponds to
+ * indices into elements (if `K = P`) or slices (if `K < P`) along the `K`th
+ * dimension of `ref`.
+ * <p>
+ * `updates` is `Tensor` of rank `Q-1+P-K` with shape:
+ * <pre>{@code
+ * [d_0, ..., d_{Q-2}, ref.shape[K], ..., ref.shape[P-1]]
+ * }</pre>
+ * For example, say we want to add 4 scattered elements to a rank-1 tensor to
+ * 8 elements. In Python, that addition would look like this:
+ * <pre>{@code
+ * ref = tf.Variable([1, 2, 3, 4, 5, 6, 7, 8], use_resource=True)
+ * indices = tf.constant([[4], [3], [1], [7]])
+ * updates = tf.constant([9, 10, 11, 12])
+ * add = tf.scatter_nd_add(ref, indices, updates)
+ * with tf.Session() as sess:
+ *   print sess.run(add)
+ * }</pre>
+ * The resulting update to ref would look like this:
+ * <p>
+ *     [1, 13, 3, 14, 14, 6, 7, 20]
+ * <p>
+ * See `tf.scatter_nd` for more details about how to make updates to
+ * slices.
+ */
+@Operator
+public final class ResourceScatterNdAdd extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.core.ResourceScatterNdAdd}
+   */
+  public static class Options {
+    
+    /**
+     * @param useLocking An optional bool. Defaults to True. If True, the assignment will
+     * be protected by a lock; otherwise the behavior is undefined,
+     * but may exhibit less contention.
+     */
+    public Options useLocking(Boolean useLocking) {
+      this.useLocking = useLocking;
+      return this;
+    }
+    
+    private Boolean useLocking;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new ResourceScatterNdAdd operation.
+   * 
+   * @param scope current scope
+   * @param ref A resource handle. Must be from a VarHandleOp.
+   * @param indices A Tensor. Must be one of the following types: int32, int64.
+   * A tensor of indices into ref.
+   * @param updates A Tensor. Must have the same type as ref. A tensor of
+   * values to add to ref.
+   * @param options carries optional attributes values
+   * @return a new instance of ResourceScatterNdAdd
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber, U extends TType> ResourceScatterNdAdd create(Scope scope, Operand<?> ref, Operand<T> indices, Operand<U> updates, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ResourceScatterNdAdd", scope.makeOpName("ResourceScatterNdAdd"));
+    opBuilder.addInput(ref.asOutput());
+    opBuilder.addInput(indices.asOutput());
+    opBuilder.addInput(updates.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.useLocking != null) {
+          opBuilder.setAttr("use_locking", opts.useLocking);
+        }
+      }
+    }
+    return new ResourceScatterNdAdd(opBuilder.build());
+  }
+  
+  /**
+   * @param useLocking An optional bool. Defaults to True. If True, the assignment will
+   * be protected by a lock; otherwise the behavior is undefined,
+   * but may exhibit less contention.
+   */
+  public static Options useLocking(Boolean useLocking) {
+    return new Options().useLocking(useLocking);
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ResourceScatterNdAdd";
+  
+  private ResourceScatterNdAdd(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterNdMax.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterNdMax.java
new file mode 100644
index 00000000000..82c1f766308
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterNdMax.java
@@ -0,0 +1,100 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ */
+@Operator
+public final class ResourceScatterNdMax extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.core.ResourceScatterNdMax}
+   */
+  public static class Options {
+    
+    /**
+     * @param useLocking An optional bool. Defaults to True. If True, the assignment will
+     * be protected by a lock; otherwise the behavior is undefined,
+     * but may exhibit less contention.
+     */
+    public Options useLocking(Boolean useLocking) {
+      this.useLocking = useLocking;
+      return this;
+    }
+    
+    private Boolean useLocking;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new ResourceScatterNdMax operation.
+   * 
+   * @param scope current scope
+   * @param ref A resource handle. Must be from a VarHandleOp.
+   * @param indices A Tensor. Must be one of the following types: int32, int64.
+   * A tensor of indices into ref.
+   * @param updates A Tensor. Must have the same type as ref. A tensor of
+   * values whose element wise max is taken with ref
+   * @param options carries optional attributes values
+   * @return a new instance of ResourceScatterNdMax
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber, U extends TType> ResourceScatterNdMax create(Scope scope, Operand<?> ref, Operand<T> indices, Operand<U> updates, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ResourceScatterNdMax", scope.makeOpName("ResourceScatterNdMax"));
+    opBuilder.addInput(ref.asOutput());
+    opBuilder.addInput(indices.asOutput());
+    opBuilder.addInput(updates.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.useLocking != null) {
+          opBuilder.setAttr("use_locking", opts.useLocking);
+        }
+      }
+    }
+    return new ResourceScatterNdMax(opBuilder.build());
+  }
+  
+  /**
+   * @param useLocking An optional bool. Defaults to True. If True, the assignment will
+   * be protected by a lock; otherwise the behavior is undefined,
+   * but may exhibit less contention.
+   */
+  public static Options useLocking(Boolean useLocking) {
+    return new Options().useLocking(useLocking);
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ResourceScatterNdMax";
+  
+  private ResourceScatterNdMax(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterNdMin.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterNdMin.java
new file mode 100644
index 00000000000..88e107c65c7
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterNdMin.java
@@ -0,0 +1,100 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ */
+@Operator
+public final class ResourceScatterNdMin extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.core.ResourceScatterNdMin}
+   */
+  public static class Options {
+    
+    /**
+     * @param useLocking An optional bool. Defaults to True. If True, the assignment will
+     * be protected by a lock; otherwise the behavior is undefined,
+     * but may exhibit less contention.
+     */
+    public Options useLocking(Boolean useLocking) {
+      this.useLocking = useLocking;
+      return this;
+    }
+    
+    private Boolean useLocking;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new ResourceScatterNdMin operation.
+   * 
+   * @param scope current scope
+   * @param ref A resource handle. Must be from a VarHandleOp.
+   * @param indices A Tensor. Must be one of the following types: int32, int64.
+   * A tensor of indices into ref.
+   * @param updates A Tensor. Must have the same type as ref. A tensor of
+   * values whose element wise min is taken with ref.
+   * @param options carries optional attributes values
+   * @return a new instance of ResourceScatterNdMin
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber, U extends TType> ResourceScatterNdMin create(Scope scope, Operand<?> ref, Operand<T> indices, Operand<U> updates, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ResourceScatterNdMin", scope.makeOpName("ResourceScatterNdMin"));
+    opBuilder.addInput(ref.asOutput());
+    opBuilder.addInput(indices.asOutput());
+    opBuilder.addInput(updates.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.useLocking != null) {
+          opBuilder.setAttr("use_locking", opts.useLocking);
+        }
+      }
+    }
+    return new ResourceScatterNdMin(opBuilder.build());
+  }
+  
+  /**
+   * @param useLocking An optional bool. Defaults to True. If True, the assignment will
+   * be protected by a lock; otherwise the behavior is undefined,
+   * but may exhibit less contention.
+   */
+  public static Options useLocking(Boolean useLocking) {
+    return new Options().useLocking(useLocking);
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ResourceScatterNdMin";
+  
+  private ResourceScatterNdMin(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterNdSub.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterNdSub.java
new file mode 100644
index 00000000000..267099b7cfc
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterNdSub.java
@@ -0,0 +1,131 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Applies sparse subtraction to individual values or slices in a Variable.
+ * <p>
+ * `ref` is a `Tensor` with rank `P` and `indices` is a `Tensor` of rank `Q`.
+ * <p>
+ * `indices` must be integer tensor, containing indices into `ref`.
+ * It must be shape `[d_0, ..., d_{Q-2}, K]` where `0 < K <= P`.
+ * <p>
+ * The innermost dimension of `indices` (with length `K`) corresponds to
+ * indices into elements (if `K = P`) or slices (if `K < P`) along the `K`th
+ * dimension of `ref`.
+ * <p>
+ * `updates` is `Tensor` of rank `Q-1+P-K` with shape:
+ * <pre>{@code
+ * [d_0, ..., d_{Q-2}, ref.shape[K], ..., ref.shape[P-1]]
+ * }</pre>
+ * For example, say we want to subtract 4 scattered elements from a rank-1 tensor
+ * with 8 elements. In Python, that subtraction would look like this:
+ * <pre>{@code
+ * ref = tf.Variable([1, 2, 3, 4, 5, 6, 7, 8], use_resource=True)
+ * indices = tf.constant([[4], [3], [1], [7]])
+ * updates = tf.constant([9, 10, 11, 12])
+ * sub = tf.scatter_nd_sub(ref, indices, updates)
+ * with tf.Session() as sess:
+ *   print sess.run(sub)
+ * }</pre>
+ * The resulting update to ref would look like this:
+ * <p>
+ *     [1, -9, 3, -6, -4, 6, 7, -4]
+ * <p>
+ * See `tf.scatter_nd` for more details about how to make updates to
+ * slices.
+ */
+@Operator
+public final class ResourceScatterNdSub extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.core.ResourceScatterNdSub}
+   */
+  public static class Options {
+    
+    /**
+     * @param useLocking An optional bool. Defaults to True. If True, the assignment will
+     * be protected by a lock; otherwise the behavior is undefined,
+     * but may exhibit less contention.
+     */
+    public Options useLocking(Boolean useLocking) {
+      this.useLocking = useLocking;
+      return this;
+    }
+    
+    private Boolean useLocking;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new ResourceScatterNdSub operation.
+   * 
+   * @param scope current scope
+   * @param ref A resource handle. Must be from a VarHandleOp.
+   * @param indices A Tensor. Must be one of the following types: int32, int64.
+   * A tensor of indices into ref.
+   * @param updates A Tensor. Must have the same type as ref. A tensor of
+   * values to add to ref.
+   * @param options carries optional attributes values
+   * @return a new instance of ResourceScatterNdSub
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber, U extends TType> ResourceScatterNdSub create(Scope scope, Operand<?> ref, Operand<T> indices, Operand<U> updates, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ResourceScatterNdSub", scope.makeOpName("ResourceScatterNdSub"));
+    opBuilder.addInput(ref.asOutput());
+    opBuilder.addInput(indices.asOutput());
+    opBuilder.addInput(updates.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.useLocking != null) {
+          opBuilder.setAttr("use_locking", opts.useLocking);
+        }
+      }
+    }
+    return new ResourceScatterNdSub(opBuilder.build());
+  }
+  
+  /**
+   * @param useLocking An optional bool. Defaults to True. If True, the assignment will
+   * be protected by a lock; otherwise the behavior is undefined,
+   * but may exhibit less contention.
+   */
+  public static Options useLocking(Boolean useLocking) {
+    return new Options().useLocking(useLocking);
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ResourceScatterNdSub";
+  
+  private ResourceScatterNdSub(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterNdUpdate.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterNdUpdate.java
new file mode 100644
index 00000000000..4a1e875bc97
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterNdUpdate.java
@@ -0,0 +1,133 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Applies sparse `updates` to individual values or slices within a given
+ * <p>
+ * variable according to `indices`.
+ * <p>
+ * `ref` is a `Tensor` with rank `P` and `indices` is a `Tensor` of rank `Q`.
+ * <p>
+ * `indices` must be integer tensor, containing indices into `ref`.
+ * It must be shape `[d_0, ..., d_{Q-2}, K]` where `0 < K <= P`.
+ * <p>
+ * The innermost dimension of `indices` (with length `K`) corresponds to
+ * indices into elements (if `K = P`) or slices (if `K < P`) along the `K`th
+ * dimension of `ref`.
+ * <p>
+ * `updates` is `Tensor` of rank `Q-1+P-K` with shape:
+ * <pre>{@code
+ * [d_0, ..., d_{Q-2}, ref.shape[K], ..., ref.shape[P-1]].
+ * }</pre>
+ * For example, say we want to update 4 scattered elements to a rank-1 tensor to
+ * 8 elements. In Python, that update would look like this:
+ * <pre>{@code
+ *     ref = tf.Variable([1, 2, 3, 4, 5, 6, 7, 8])
+ *     indices = tf.constant([[4], [3], [1] ,[7]])
+ *     updates = tf.constant([9, 10, 11, 12])
+ *     update = tf.scatter_nd_update(ref, indices, updates)
+ *     with tf.Session() as sess:
+ *       print sess.run(update)
+ * }</pre>
+ * The resulting update to ref would look like this:
+ * <p>
+ *     [1, 11, 3, 10, 9, 6, 7, 12]
+ * <p>
+ * See `tf.scatter_nd` for more details about how to make updates to
+ * slices.
+ */
+@Operator
+public final class ResourceScatterNdUpdate extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.core.ResourceScatterNdUpdate}
+   */
+  public static class Options {
+    
+    /**
+     * @param useLocking An optional bool. Defaults to True. If True, the assignment will
+     * be protected by a lock; otherwise the behavior is undefined,
+     * but may exhibit less contention.
+     */
+    public Options useLocking(Boolean useLocking) {
+      this.useLocking = useLocking;
+      return this;
+    }
+    
+    private Boolean useLocking;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new ResourceScatterNdUpdate operation.
+   * 
+   * @param scope current scope
+   * @param ref A resource handle. Must be from a VarHandleOp.
+   * @param indices A Tensor. Must be one of the following types: int32, int64.
+   * A tensor of indices into ref.
+   * @param updates A Tensor. Must have the same type as ref. A tensor of updated
+   * values to add to ref.
+   * @param options carries optional attributes values
+   * @return a new instance of ResourceScatterNdUpdate
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber, U extends TType> ResourceScatterNdUpdate create(Scope scope, Operand<?> ref, Operand<T> indices, Operand<U> updates, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ResourceScatterNdUpdate", scope.makeOpName("ResourceScatterNdUpdate"));
+    opBuilder.addInput(ref.asOutput());
+    opBuilder.addInput(indices.asOutput());
+    opBuilder.addInput(updates.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.useLocking != null) {
+          opBuilder.setAttr("use_locking", opts.useLocking);
+        }
+      }
+    }
+    return new ResourceScatterNdUpdate(opBuilder.build());
+  }
+  
+  /**
+   * @param useLocking An optional bool. Defaults to True. If True, the assignment will
+   * be protected by a lock; otherwise the behavior is undefined,
+   * but may exhibit less contention.
+   */
+  public static Options useLocking(Boolean useLocking) {
+    return new Options().useLocking(useLocking);
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ResourceScatterNdUpdate";
+  
+  private ResourceScatterNdUpdate(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterSub.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterSub.java
new file mode 100644
index 00000000000..7b772fab997
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterSub.java
@@ -0,0 +1,81 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Subtracts sparse updates from the variable referenced by `resource`.
+ * <p>
+ * This operation computes
+ * <p>
+ *     # Scalar indices
+ *     ref[indices, ...] -= updates[...]
+ * <p>
+ *     # Vector indices (for each i)
+ *     ref[indices[i], ...] -= updates[i, ...]
+ * <p>
+ *     # High rank indices (for each i, ..., j)
+ *     ref[indices[i, ..., j], ...] -= updates[i, ..., j, ...]
+ * <p>
+ * Duplicate entries are handled correctly: if multiple `indices` reference
+ * the same location, their contributions add.
+ * <p>
+ * Requires `updates.shape = indices.shape + ref.shape[1:]` or `updates.shape = []`.
+ * <p>
+ * <div style="width:70%; margin:auto; margin-bottom:10px; margin-top:20px;">
+ * <img style="width:100%" src='https://www.tensorflow.org/images/ScatterAdd.png' alt>
+ * </div>
+ */
+@Operator
+public final class ResourceScatterSub extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new ResourceScatterSub operation.
+   * 
+   * @param scope current scope
+   * @param resource Should be from a `Variable` node.
+   * @param indices A tensor of indices into the first dimension of `ref`.
+   * @param updates A tensor of updated values to add to `ref`.
+   * @return a new instance of ResourceScatterSub
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber, U extends TType> ResourceScatterSub create(Scope scope, Operand<?> resource, Operand<T> indices, Operand<U> updates) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ResourceScatterSub", scope.makeOpName("ResourceScatterSub"));
+    opBuilder.addInput(resource.asOutput());
+    opBuilder.addInput(indices.asOutput());
+    opBuilder.addInput(updates.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new ResourceScatterSub(opBuilder.build());
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ResourceScatterSub";
+  
+  private ResourceScatterSub(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterUpdate.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterUpdate.java
new file mode 100644
index 00000000000..067ddf5f205
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceScatterUpdate.java
@@ -0,0 +1,72 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Assigns sparse updates to the variable referenced by `resource`.
+ * <p>
+ * This operation computes
+ * <p>
+ *     # Scalar indices
+ *     ref[indices, ...] = updates[...]
+ * <p>
+ *     # Vector indices (for each i)
+ *     ref[indices[i], ...] = updates[i, ...]
+ * <p>
+ *     # High rank indices (for each i, ..., j)
+ *     ref[indices[i, ..., j], ...] = updates[i, ..., j, ...]
+ */
+@Operator
+public final class ResourceScatterUpdate extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new ResourceScatterUpdate operation.
+   * 
+   * @param scope current scope
+   * @param resource Should be from a `Variable` node.
+   * @param indices A tensor of indices into the first dimension of `ref`.
+   * @param updates A tensor of updated values to add to `ref`.
+   * @return a new instance of ResourceScatterUpdate
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber, U extends TType> ResourceScatterUpdate create(Scope scope, Operand<?> resource, Operand<T> indices, Operand<U> updates) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ResourceScatterUpdate", scope.makeOpName("ResourceScatterUpdate"));
+    opBuilder.addInput(resource.asOutput());
+    opBuilder.addInput(indices.asOutput());
+    opBuilder.addInput(updates.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new ResourceScatterUpdate(opBuilder.build());
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ResourceScatterUpdate";
+  
+  private ResourceScatterUpdate(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceStridedSliceAssign.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceStridedSliceAssign.java
new file mode 100644
index 00000000000..4deb4c55f64
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ResourceStridedSliceAssign.java
@@ -0,0 +1,182 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Assign `value` to the sliced l-value reference of `ref`.
+ * <p>
+ * The values of `value` are assigned to the positions in the variable
+ * `ref` that are selected by the slice parameters. The slice parameters
+ * `begin, `end`, `strides`, etc. work exactly as in `StridedSlice`.
+ * <p>
+ * NOTE this op currently does not support broadcasting and so `value`'s
+ * shape must be exactly the shape produced by the slice of `ref`.
+ */
+@Operator
+public final class ResourceStridedSliceAssign extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.core.ResourceStridedSliceAssign}
+   */
+  public static class Options {
+    
+    /**
+     * @param beginMask 
+     */
+    public Options beginMask(Long beginMask) {
+      this.beginMask = beginMask;
+      return this;
+    }
+    
+    /**
+     * @param endMask 
+     */
+    public Options endMask(Long endMask) {
+      this.endMask = endMask;
+      return this;
+    }
+    
+    /**
+     * @param ellipsisMask 
+     */
+    public Options ellipsisMask(Long ellipsisMask) {
+      this.ellipsisMask = ellipsisMask;
+      return this;
+    }
+    
+    /**
+     * @param newAxisMask 
+     */
+    public Options newAxisMask(Long newAxisMask) {
+      this.newAxisMask = newAxisMask;
+      return this;
+    }
+    
+    /**
+     * @param shrinkAxisMask 
+     */
+    public Options shrinkAxisMask(Long shrinkAxisMask) {
+      this.shrinkAxisMask = shrinkAxisMask;
+      return this;
+    }
+    
+    private Long beginMask;
+    private Long endMask;
+    private Long ellipsisMask;
+    private Long newAxisMask;
+    private Long shrinkAxisMask;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new ResourceStridedSliceAssign operation.
+   * 
+   * @param scope current scope
+   * @param ref 
+   * @param begin 
+   * @param end 
+   * @param strides 
+   * @param value 
+   * @param options carries optional attributes values
+   * @return a new instance of ResourceStridedSliceAssign
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber, U extends TType> ResourceStridedSliceAssign create(Scope scope, Operand<?> ref, Operand<T> begin, Operand<T> end, Operand<T> strides, Operand<U> value, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ResourceStridedSliceAssign", scope.makeOpName("ResourceStridedSliceAssign"));
+    opBuilder.addInput(ref.asOutput());
+    opBuilder.addInput(begin.asOutput());
+    opBuilder.addInput(end.asOutput());
+    opBuilder.addInput(strides.asOutput());
+    opBuilder.addInput(value.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.beginMask != null) {
+          opBuilder.setAttr("begin_mask", opts.beginMask);
+        }
+        if (opts.endMask != null) {
+          opBuilder.setAttr("end_mask", opts.endMask);
+        }
+        if (opts.ellipsisMask != null) {
+          opBuilder.setAttr("ellipsis_mask", opts.ellipsisMask);
+        }
+        if (opts.newAxisMask != null) {
+          opBuilder.setAttr("new_axis_mask", opts.newAxisMask);
+        }
+        if (opts.shrinkAxisMask != null) {
+          opBuilder.setAttr("shrink_axis_mask", opts.shrinkAxisMask);
+        }
+      }
+    }
+    return new ResourceStridedSliceAssign(opBuilder.build());
+  }
+  
+  /**
+   * @param beginMask 
+   */
+  public static Options beginMask(Long beginMask) {
+    return new Options().beginMask(beginMask);
+  }
+  
+  /**
+   * @param endMask 
+   */
+  public static Options endMask(Long endMask) {
+    return new Options().endMask(endMask);
+  }
+  
+  /**
+   * @param ellipsisMask 
+   */
+  public static Options ellipsisMask(Long ellipsisMask) {
+    return new Options().ellipsisMask(ellipsisMask);
+  }
+  
+  /**
+   * @param newAxisMask 
+   */
+  public static Options newAxisMask(Long newAxisMask) {
+    return new Options().newAxisMask(newAxisMask);
+  }
+  
+  /**
+   * @param shrinkAxisMask 
+   */
+  public static Options shrinkAxisMask(Long shrinkAxisMask) {
+    return new Options().shrinkAxisMask(shrinkAxisMask);
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ResourceStridedSliceAssign";
+  
+  private ResourceStridedSliceAssign(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Reverse.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Reverse.java
new file mode 100644
index 00000000000..340336ee9f9
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Reverse.java
@@ -0,0 +1,126 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Reverses specific dimensions of a tensor.
+ * <p>
+ * NOTE `tf.reverse` has now changed behavior in preparation for 1.0.
+ * `tf.reverse_v2` is currently an alias that will be deprecated before TF 1.0.
+ * <p>
+ * Given a `tensor`, and a `int32` tensor `axis` representing the set of
+ * dimensions of `tensor` to reverse. This operation reverses each dimension
+ * `i` for which there exists `j` s.t. `axis[j] == i`.
+ * <p>
+ * `tensor` can have up to 8 dimensions. The number of dimensions specified
+ * in `axis` may be 0 or more entries. If an index is specified more than
+ * once, a InvalidArgument error is raised.
+ * <p>
+ * For example:
+ * <pre>{@code
+ * # tensor 't' is [[[[ 0,  1,  2,  3],
+ * #                  [ 4,  5,  6,  7],
+ * #                  [ 8,  9, 10, 11]],
+ * #                 [[12, 13, 14, 15],
+ * #                  [16, 17, 18, 19],
+ * #                  [20, 21, 22, 23]]]]
+ * # tensor 't' shape is [1, 2, 3, 4]
+ * 
+ * # 'dims' is [3] or 'dims' is [-1]
+ * reverse(t, dims) ==> [[[[ 3,  2,  1,  0],
+ *                         [ 7,  6,  5,  4],
+ *                         [ 11, 10, 9, 8]],
+ *                        [[15, 14, 13, 12],
+ *                         [19, 18, 17, 16],
+ *                         [23, 22, 21, 20]]]]
+ * 
+ * # 'dims' is '[1]' (or 'dims' is '[-3]')
+ * reverse(t, dims) ==> [[[[12, 13, 14, 15],
+ *                         [16, 17, 18, 19],
+ *                         [20, 21, 22, 23]
+ *                        [[ 0,  1,  2,  3],
+ *                         [ 4,  5,  6,  7],
+ *                         [ 8,  9, 10, 11]]]]
+ * 
+ * # 'dims' is '[2]' (or 'dims' is '[-2]')
+ * reverse(t, dims) ==> [[[[8, 9, 10, 11],
+ *                         [4, 5, 6, 7],
+ *                         [0, 1, 2, 3]]
+ *                        [[20, 21, 22, 23],
+ *                         [16, 17, 18, 19],
+ *                         [12, 13, 14, 15]]]]
+ * }</pre>
+ * 
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator
+public final class Reverse<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new Reverse operation.
+   * 
+   * @param scope current scope
+   * @param tensor Up to 8-D.
+   * @param axis 1-D. The indices of the dimensions to reverse. Must be in the range
+   * `[-rank(tensor), rank(tensor))`.
+   * @return a new instance of Reverse
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType, U extends TNumber> Reverse<T> create(Scope scope, Operand<T> tensor, Operand<U> axis) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ReverseV2", scope.makeOpName("Reverse"));
+    opBuilder.addInput(tensor.asOutput());
+    opBuilder.addInput(axis.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new Reverse<T>(opBuilder.build());
+  }
+  
+  /**
+   * The same shape as `tensor`.
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ReverseV2";
+  
+  private Output<T> output;
+  
+  private Reverse(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ReverseSequence.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ReverseSequence.java
new file mode 100644
index 00000000000..46fd2b54aa9
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ReverseSequence.java
@@ -0,0 +1,168 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Reverses variable length slices.
+ * <p>
+ * This op first slices `input` along the dimension `batch_dim`, and for each
+ * slice `i`, reverses the first `seq_lengths[i]` elements along
+ * the dimension `seq_dim`.
+ * <p>
+ * The elements of `seq_lengths` must obey `seq_lengths[i] <= input.dims[seq_dim]`,
+ * and `seq_lengths` must be a vector of length `input.dims[batch_dim]`.
+ * <p>
+ * The output slice `i` along dimension `batch_dim` is then given by input
+ * slice `i`, with the first `seq_lengths[i]` slices along dimension
+ * `seq_dim` reversed.
+ * <p>
+ * For example:
+ * <pre>{@code
+ * # Given this:
+ * batch_dim = 0
+ * seq_dim = 1
+ * input.dims = (4, 8, ...)
+ * seq_lengths = [7, 2, 3, 5]
+ * 
+ * # then slices of input are reversed on seq_dim, but only up to seq_lengths:
+ * output[0, 0:7, :, ...] = input[0, 7:0:-1, :, ...]
+ * output[1, 0:2, :, ...] = input[1, 2:0:-1, :, ...]
+ * output[2, 0:3, :, ...] = input[2, 3:0:-1, :, ...]
+ * output[3, 0:5, :, ...] = input[3, 5:0:-1, :, ...]
+ * 
+ * # while entries past seq_lens are copied through:
+ * output[0, 7:, :, ...] = input[0, 7:, :, ...]
+ * output[1, 2:, :, ...] = input[1, 2:, :, ...]
+ * output[2, 3:, :, ...] = input[2, 3:, :, ...]
+ * output[3, 2:, :, ...] = input[3, 2:, :, ...]
+ * }</pre>
+ * In contrast, if:
+ * <pre>{@code
+ * # Given this:
+ * batch_dim = 2
+ * seq_dim = 0
+ * input.dims = (8, ?, 4, ...)
+ * seq_lengths = [7, 2, 3, 5]
+ * 
+ * # then slices of input are reversed on seq_dim, but only up to seq_lengths:
+ * output[0:7, :, 0, :, ...] = input[7:0:-1, :, 0, :, ...]
+ * output[0:2, :, 1, :, ...] = input[2:0:-1, :, 1, :, ...]
+ * output[0:3, :, 2, :, ...] = input[3:0:-1, :, 2, :, ...]
+ * output[0:5, :, 3, :, ...] = input[5:0:-1, :, 3, :, ...]
+ * 
+ * # while entries past seq_lens are copied through:
+ * output[7:, :, 0, :, ...] = input[7:, :, 0, :, ...]
+ * output[2:, :, 1, :, ...] = input[2:, :, 1, :, ...]
+ * output[3:, :, 2, :, ...] = input[3:, :, 2, :, ...]
+ * output[2:, :, 3, :, ...] = input[2:, :, 3, :, ...]
+ * }</pre>
+ * 
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator
+public final class ReverseSequence<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.core.ReverseSequence}
+   */
+  public static class Options {
+    
+    /**
+     * @param batchDim The dimension along which reversal is performed.
+     */
+    public Options batchDim(Long batchDim) {
+      this.batchDim = batchDim;
+      return this;
+    }
+    
+    private Long batchDim;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new ReverseSequence operation.
+   * 
+   * @param scope current scope
+   * @param input The input to reverse.
+   * @param seqLengths 1-D with length `input.dims(batch_dim)` and
+   * `max(seq_lengths) <= input.dims(seq_dim)`
+   * @param seqDim The dimension which is partially reversed.
+   * @param options carries optional attributes values
+   * @return a new instance of ReverseSequence
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType, U extends TNumber> ReverseSequence<T> create(Scope scope, Operand<T> input, Operand<U> seqLengths, Long seqDim, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ReverseSequence", scope.makeOpName("ReverseSequence"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(seqLengths.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("seq_dim", seqDim);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.batchDim != null) {
+          opBuilder.setAttr("batch_dim", opts.batchDim);
+        }
+      }
+    }
+    return new ReverseSequence<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param batchDim The dimension along which reversal is performed.
+   */
+  public static Options batchDim(Long batchDim) {
+    return new Options().batchDim(batchDim);
+  }
+  
+  /**
+   * The partially reversed input. It has the same shape as `input`.
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ReverseSequence";
+  
+  private Output<T> output;
+  
+  private ReverseSequence(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Roll.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Roll.java
new file mode 100644
index 00000000000..761be92e533
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Roll.java
@@ -0,0 +1,109 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Rolls the elements of a tensor along an axis.
+ * <p>
+ * The elements are shifted positively (towards larger indices) by the offset of
+ * `shift` along the dimension of `axis`. Negative `shift` values will shift
+ * elements in the opposite direction. Elements that roll passed the last position
+ * will wrap around to the first and vice versa. Multiple shifts along multiple
+ * axes may be specified.
+ * <p>
+ * For example:
+ * <pre>{@code
+ * # 't' is [0, 1, 2, 3, 4]
+ * roll(t, shift=2, axis=0) ==> [3, 4, 0, 1, 2]
+ * 
+ * # shifting along multiple dimensions
+ * # 't' is [[0, 1, 2, 3, 4], [5, 6, 7, 8, 9]]
+ * roll(t, shift=[1, -2], axis=[0, 1]) ==> [[7, 8, 9, 5, 6], [2, 3, 4, 0, 1]]
+ * 
+ * # shifting along the same axis multiple times
+ * # 't' is [[0, 1, 2, 3, 4], [5, 6, 7, 8, 9]]
+ * roll(t, shift=[2, -3], axis=[1, 1]) ==> [[1, 2, 3, 4, 0], [6, 7, 8, 9, 5]]
+ * }</pre>
+ * 
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator
+public final class Roll<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new Roll operation.
+   * 
+   * @param scope current scope
+   * @param input 
+   * @param shift Dimension must be 0-D or 1-D. `shift[i]` specifies the number of places by which
+   * elements are shifted positively (towards larger indices) along the dimension
+   * specified by `axis[i]`. Negative shifts will roll the elements in the opposite
+   * direction.
+   * @param axis Dimension must be 0-D or 1-D. `axis[i]` specifies the dimension that the shift
+   * `shift[i]` should occur. If the same axis is referenced more than once, the
+   * total shift for that axis will be the sum of all the shifts that belong to that
+   * axis.
+   * @return a new instance of Roll
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType, U extends TNumber, V extends TNumber> Roll<T> create(Scope scope, Operand<T> input, Operand<U> shift, Operand<V> axis) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Roll", scope.makeOpName("Roll"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(shift.asOutput());
+    opBuilder.addInput(axis.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new Roll<T>(opBuilder.build());
+  }
+  
+  /**
+   * Has the same shape and size as the input. The elements are shifted
+   * positively (towards larger indices) by the offsets of `shift` along the
+   * dimensions of `axis`.
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Roll";
+  
+  private Output<T> output;
+  
+  private Roll(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Rpc.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Rpc.java
new file mode 100644
index 00000000000..ad3195933bc
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Rpc.java
@@ -0,0 +1,209 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TString;
+
+/**
+ * Perform batches of RPC requests.
+ * <p>
+ * This op asynchronously performs either a single RPC request, or a batch
+ * of requests.  RPC requests are defined by three main parameters:
+ * <p>
+ *   - `address` (the host+port or BNS address of the request)
+ *   - `method` (the RPC method name for the request)
+ *   - `request` (the serialized proto string, or vector of strings,
+ *      of the RPC request argument).
+ * <p>
+ * For example, if you have an RPC service running on port localhost:2345,
+ * and its interface is configured with the following proto declaration:
+ * <pre>{@code
+ * service MyService {
+ *   rpc MyMethod(MyRequestProto) returns (MyResponseProto) {
+ *   }
+ * };
+ * }</pre>
+ * then call this op with arguments:
+ * <pre>{@code
+ * address = "localhost:2345"
+ * method = "MyService/MyMethod"
+ * }</pre>
+ * The `request` tensor is a string tensor representing serialized `MyRequestProto`
+ * strings; and the output string tensor `response` will have the same shape
+ * and contain (upon successful completion) corresponding serialized
+ * `MyResponseProto` strings.
+ * <p>
+ * For example, to send a single, empty, `MyRequestProto`, call
+ * this op with `request = ""`.  To send 5 <b>parallel</b> empty requests,
+ * call this op with `request = ["", "", "", "", ""]`.
+ * <p>
+ * More generally, one can create a batch of `MyRequestProto` serialized protos
+ * from regular batched tensors using the `encode_proto` op, and convert
+ * the response `MyResponseProto` serialized protos to batched tensors
+ * using the `decode_proto` op.
+ * <p>
+ * <b>NOTE</b> Working with serialized proto strings is faster than instantiating
+ * actual proto objects in memory, so no performance degradation is expected
+ * compared to writing custom kernels for this workflow.
+ * <p>
+ * If the connection fails or the remote worker returns an error
+ * status, the op reraises this exception locally.
+ * <p>
+ * See the `TryRpc` op if you prefer to handle RPC failures manually in the graph.
+ */
+@Operator
+public final class Rpc extends RawOp implements Operand<TString> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.core.Rpc}
+   */
+  public static class Options {
+    
+    /**
+     * @param protocol RPC protocol to use.  Empty string means use the default protocol.
+     * Options include 'grpc'.
+     */
+    public Options protocol(String protocol) {
+      this.protocol = protocol;
+      return this;
+    }
+    
+    /**
+     * @param failFast `boolean`. If `true` (default), then failures to connect
+     * (i.e., the server does not immediately respond) cause an RPC failure.
+     */
+    public Options failFast(Boolean failFast) {
+      this.failFast = failFast;
+      return this;
+    }
+    
+    /**
+     * @param timeoutInMs `int`. If `0` (default), then the kernel will run the RPC
+     * request and only time out if the RPC deadline passes or the session times out.
+     * If this value is greater than `0`, then the op will raise an exception if
+     * the RPC takes longer than `timeout_in_ms`.
+     */
+    public Options timeoutInMs(Long timeoutInMs) {
+      this.timeoutInMs = timeoutInMs;
+      return this;
+    }
+    
+    private String protocol;
+    private Boolean failFast;
+    private Long timeoutInMs;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new Rpc operation.
+   * 
+   * @param scope current scope
+   * @param address `0-D` or `1-D`.  The address (i.e. host_name:port) of the RPC server.
+   * If this tensor has more than 1 element, then multiple parallel rpc requests
+   * are sent.  This argument broadcasts with `method` and `request`.
+   * @param method `0-D` or `1-D`.  The method address on the RPC server.
+   * If this tensor has more than 1 element, then multiple parallel rpc requests
+   * are sent.  This argument broadcasts with `address` and `request`.
+   * @param request `0-D` or `1-D`.  Serialized proto strings: the rpc request argument.
+   * If this tensor has more than 1 element, then multiple parallel rpc requests
+   * are sent.  This argument broadcasts with `address` and `method`.
+   * @param options carries optional attributes values
+   * @return a new instance of Rpc
+   */
+  @Endpoint(describeByClass = true)
+  public static Rpc create(Scope scope, Operand<TString> address, Operand<TString> method, Operand<TString> request, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Rpc", scope.makeOpName("Rpc"));
+    opBuilder.addInput(address.asOutput());
+    opBuilder.addInput(method.asOutput());
+    opBuilder.addInput(request.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.protocol != null) {
+          opBuilder.setAttr("protocol", opts.protocol);
+        }
+        if (opts.failFast != null) {
+          opBuilder.setAttr("fail_fast", opts.failFast);
+        }
+        if (opts.timeoutInMs != null) {
+          opBuilder.setAttr("timeout_in_ms", opts.timeoutInMs);
+        }
+      }
+    }
+    return new Rpc(opBuilder.build());
+  }
+  
+  /**
+   * @param protocol RPC protocol to use.  Empty string means use the default protocol.
+   * Options include 'grpc'.
+   */
+  public static Options protocol(String protocol) {
+    return new Options().protocol(protocol);
+  }
+  
+  /**
+   * @param failFast `boolean`. If `true` (default), then failures to connect
+   * (i.e., the server does not immediately respond) cause an RPC failure.
+   */
+  public static Options failFast(Boolean failFast) {
+    return new Options().failFast(failFast);
+  }
+  
+  /**
+   * @param timeoutInMs `int`. If `0` (default), then the kernel will run the RPC
+   * request and only time out if the RPC deadline passes or the session times out.
+   * If this value is greater than `0`, then the op will raise an exception if
+   * the RPC takes longer than `timeout_in_ms`.
+   */
+  public static Options timeoutInMs(Long timeoutInMs) {
+    return new Options().timeoutInMs(timeoutInMs);
+  }
+  
+  /**
+   * Same shape as `request`. Serialized proto strings: the rpc responses.
+   */
+  public Output<TString> response() {
+    return response;
+  }
+  
+  @Override
+  public Output<TString> asOutput() {
+    return response;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Rpc";
+  
+  private Output<TString> response;
+  
+  private Rpc(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    response = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterAdd.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterAdd.java
new file mode 100644
index 00000000000..c91e799a643
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterAdd.java
@@ -0,0 +1,140 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Adds sparse updates to a variable reference.
+ * <p>
+ * This operation computes
+ * <p>
+ *     # Scalar indices
+ *     ref[indices, ...] += updates[...]
+ * <p>
+ *     # Vector indices (for each i)
+ *     ref[indices[i], ...] += updates[i, ...]
+ * <p>
+ *     # High rank indices (for each i, ..., j)
+ *     ref[indices[i, ..., j], ...] += updates[i, ..., j, ...]
+ * <p>
+ * This operation outputs `ref` after the update is done.
+ * This makes it easier to chain operations that need to use the reset value.
+ * <p>
+ * Duplicate entries are handled correctly: if multiple `indices` reference
+ * the same location, their contributions add.
+ * <p>
+ * Requires `updates.shape = indices.shape + ref.shape[1:]` or `updates.shape = []`.
+ * <p>
+ * <div style="width:70%; margin:auto; margin-bottom:10px; margin-top:20px;">
+ * <img style="width:100%" src="https://www.tensorflow.org/images/ScatterAdd.png" alt>
+ * </div>
+ * 
+ * @param <T> data type for {@code outputRef()} output
+ */
+@Operator
+public final class ScatterAdd<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.core.ScatterAdd}
+   */
+  public static class Options {
+    
+    /**
+     * @param useLocking If True, the addition will be protected by a lock;
+     * otherwise the behavior is undefined, but may exhibit less contention.
+     */
+    public Options useLocking(Boolean useLocking) {
+      this.useLocking = useLocking;
+      return this;
+    }
+    
+    private Boolean useLocking;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new ScatterAdd operation.
+   * 
+   * @param scope current scope
+   * @param ref Should be from a `Variable` node.
+   * @param indices A tensor of indices into the first dimension of `ref`.
+   * @param updates A tensor of updated values to add to `ref`.
+   * @param options carries optional attributes values
+   * @return a new instance of ScatterAdd
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType, U extends TNumber> ScatterAdd<T> create(Scope scope, Operand<T> ref, Operand<U> indices, Operand<T> updates, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ScatterAdd", scope.makeOpName("ScatterAdd"));
+    opBuilder.addInput(ref.asOutput());
+    opBuilder.addInput(indices.asOutput());
+    opBuilder.addInput(updates.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.useLocking != null) {
+          opBuilder.setAttr("use_locking", opts.useLocking);
+        }
+      }
+    }
+    return new ScatterAdd<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param useLocking If True, the addition will be protected by a lock;
+   * otherwise the behavior is undefined, but may exhibit less contention.
+   */
+  public static Options useLocking(Boolean useLocking) {
+    return new Options().useLocking(useLocking);
+  }
+  
+  /**
+   * = Same as `ref`.  Returned as a convenience for operations that want
+   * to use the updated values after the update is done.
+   */
+  public Output<T> outputRef() {
+    return outputRef;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return outputRef;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ScatterAdd";
+  
+  private Output<T> outputRef;
+  
+  private ScatterAdd(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    outputRef = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterDiv.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterDiv.java
new file mode 100644
index 00000000000..12a27c67960
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterDiv.java
@@ -0,0 +1,136 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Divides a variable reference by sparse updates.
+ * <p>
+ * This operation computes
+ * <pre>{@code
+ *     # Scalar indices
+ *     ref[indices, ...] /= updates[...]
+ * 
+ *     # Vector indices (for each i)
+ *     ref[indices[i], ...] /= updates[i, ...]
+ * 
+ *     # High rank indices (for each i, ..., j)
+ *     ref[indices[i, ..., j], ...] /= updates[i, ..., j, ...]
+ * }</pre>
+ * This operation outputs `ref` after the update is done.
+ * This makes it easier to chain operations that need to use the reset value.
+ * <p>
+ * Duplicate entries are handled correctly: if multiple `indices` reference
+ * the same location, their contributions divide.
+ * <p>
+ * Requires `updates.shape = indices.shape + ref.shape[1:]` or `updates.shape = []`.
+ * 
+ * @param <T> data type for {@code outputRef()} output
+ */
+@Operator
+public final class ScatterDiv<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.core.ScatterDiv}
+   */
+  public static class Options {
+    
+    /**
+     * @param useLocking If True, the operation will be protected by a lock;
+     * otherwise the behavior is undefined, but may exhibit less contention.
+     */
+    public Options useLocking(Boolean useLocking) {
+      this.useLocking = useLocking;
+      return this;
+    }
+    
+    private Boolean useLocking;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new ScatterDiv operation.
+   * 
+   * @param scope current scope
+   * @param ref Should be from a `Variable` node.
+   * @param indices A tensor of indices into the first dimension of `ref`.
+   * @param updates A tensor of values that `ref` is divided by.
+   * @param options carries optional attributes values
+   * @return a new instance of ScatterDiv
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType, U extends TNumber> ScatterDiv<T> create(Scope scope, Operand<T> ref, Operand<U> indices, Operand<T> updates, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ScatterDiv", scope.makeOpName("ScatterDiv"));
+    opBuilder.addInput(ref.asOutput());
+    opBuilder.addInput(indices.asOutput());
+    opBuilder.addInput(updates.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.useLocking != null) {
+          opBuilder.setAttr("use_locking", opts.useLocking);
+        }
+      }
+    }
+    return new ScatterDiv<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param useLocking If True, the operation will be protected by a lock;
+   * otherwise the behavior is undefined, but may exhibit less contention.
+   */
+  public static Options useLocking(Boolean useLocking) {
+    return new Options().useLocking(useLocking);
+  }
+  
+  /**
+   * = Same as `ref`.  Returned as a convenience for operations that want
+   * to use the updated values after the update is done.
+   */
+  public Output<T> outputRef() {
+    return outputRef;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return outputRef;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ScatterDiv";
+  
+  private Output<T> outputRef;
+  
+  private ScatterDiv(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    outputRef = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterMax.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterMax.java
new file mode 100644
index 00000000000..360cb496ee8
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterMax.java
@@ -0,0 +1,139 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Reduces sparse updates into a variable reference using the `max` operation.
+ * <p>
+ * This operation computes
+ * <p>
+ *     # Scalar indices
+ *     ref[indices, ...] = max(ref[indices, ...], updates[...])
+ * <p>
+ *     # Vector indices (for each i)
+ *     ref[indices[i], ...] = max(ref[indices[i], ...], updates[i, ...])
+ * <p>
+ *     # High rank indices (for each i, ..., j)
+ *     ref[indices[i, ..., j], ...] = max(ref[indices[i, ..., j], ...], updates[i, ..., j, ...])
+ * <p>
+ * This operation outputs `ref` after the update is done.
+ * This makes it easier to chain operations that need to use the reset value.
+ * <p>
+ * Duplicate entries are handled correctly: if multiple `indices` reference
+ * the same location, their contributions combine.
+ * <p>
+ * Requires `updates.shape = indices.shape + ref.shape[1:]` or `updates.shape = []`.
+ * <p>
+ * <div style="width:70%; margin:auto; margin-bottom:10px; margin-top:20px;">
+ * <img style="width:100%" src="https://www.tensorflow.org/images/ScatterAdd.png" alt>
+ * </div>
+ * 
+ * @param <T> data type for {@code outputRef()} output
+ */
+@Operator
+public final class ScatterMax<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.core.ScatterMax}
+   */
+  public static class Options {
+    
+    /**
+     * @param useLocking If True, the update will be protected by a lock;
+     * otherwise the behavior is undefined, but may exhibit less contention.
+     */
+    public Options useLocking(Boolean useLocking) {
+      this.useLocking = useLocking;
+      return this;
+    }
+    
+    private Boolean useLocking;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new ScatterMax operation.
+   * 
+   * @param scope current scope
+   * @param ref Should be from a `Variable` node.
+   * @param indices A tensor of indices into the first dimension of `ref`.
+   * @param updates A tensor of updated values to reduce into `ref`.
+   * @param options carries optional attributes values
+   * @return a new instance of ScatterMax
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber, U extends TNumber> ScatterMax<T> create(Scope scope, Operand<T> ref, Operand<U> indices, Operand<T> updates, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ScatterMax", scope.makeOpName("ScatterMax"));
+    opBuilder.addInput(ref.asOutput());
+    opBuilder.addInput(indices.asOutput());
+    opBuilder.addInput(updates.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.useLocking != null) {
+          opBuilder.setAttr("use_locking", opts.useLocking);
+        }
+      }
+    }
+    return new ScatterMax<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param useLocking If True, the update will be protected by a lock;
+   * otherwise the behavior is undefined, but may exhibit less contention.
+   */
+  public static Options useLocking(Boolean useLocking) {
+    return new Options().useLocking(useLocking);
+  }
+  
+  /**
+   * = Same as `ref`.  Returned as a convenience for operations that want
+   * to use the updated values after the update is done.
+   */
+  public Output<T> outputRef() {
+    return outputRef;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return outputRef;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ScatterMax";
+  
+  private Output<T> outputRef;
+  
+  private ScatterMax(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    outputRef = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterMin.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterMin.java
new file mode 100644
index 00000000000..14f83e0756a
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterMin.java
@@ -0,0 +1,139 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Reduces sparse updates into a variable reference using the `min` operation.
+ * <p>
+ * This operation computes
+ * <p>
+ *     # Scalar indices
+ *     ref[indices, ...] = min(ref[indices, ...], updates[...])
+ * <p>
+ *     # Vector indices (for each i)
+ *     ref[indices[i], ...] = min(ref[indices[i], ...], updates[i, ...])
+ * <p>
+ *     # High rank indices (for each i, ..., j)
+ *     ref[indices[i, ..., j], ...] = min(ref[indices[i, ..., j], ...], updates[i, ..., j, ...])
+ * <p>
+ * This operation outputs `ref` after the update is done.
+ * This makes it easier to chain operations that need to use the reset value.
+ * <p>
+ * Duplicate entries are handled correctly: if multiple `indices` reference
+ * the same location, their contributions combine.
+ * <p>
+ * Requires `updates.shape = indices.shape + ref.shape[1:]` or `updates.shape = []`.
+ * <p>
+ * <div style="width:70%; margin:auto; margin-bottom:10px; margin-top:20px;">
+ * <img style="width:100%" src="https://www.tensorflow.org/images/ScatterAdd.png" alt>
+ * </div>
+ * 
+ * @param <T> data type for {@code outputRef()} output
+ */
+@Operator
+public final class ScatterMin<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.core.ScatterMin}
+   */
+  public static class Options {
+    
+    /**
+     * @param useLocking If True, the update will be protected by a lock;
+     * otherwise the behavior is undefined, but may exhibit less contention.
+     */
+    public Options useLocking(Boolean useLocking) {
+      this.useLocking = useLocking;
+      return this;
+    }
+    
+    private Boolean useLocking;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new ScatterMin operation.
+   * 
+   * @param scope current scope
+   * @param ref Should be from a `Variable` node.
+   * @param indices A tensor of indices into the first dimension of `ref`.
+   * @param updates A tensor of updated values to reduce into `ref`.
+   * @param options carries optional attributes values
+   * @return a new instance of ScatterMin
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber, U extends TNumber> ScatterMin<T> create(Scope scope, Operand<T> ref, Operand<U> indices, Operand<T> updates, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ScatterMin", scope.makeOpName("ScatterMin"));
+    opBuilder.addInput(ref.asOutput());
+    opBuilder.addInput(indices.asOutput());
+    opBuilder.addInput(updates.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.useLocking != null) {
+          opBuilder.setAttr("use_locking", opts.useLocking);
+        }
+      }
+    }
+    return new ScatterMin<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param useLocking If True, the update will be protected by a lock;
+   * otherwise the behavior is undefined, but may exhibit less contention.
+   */
+  public static Options useLocking(Boolean useLocking) {
+    return new Options().useLocking(useLocking);
+  }
+  
+  /**
+   * = Same as `ref`.  Returned as a convenience for operations that want
+   * to use the updated values after the update is done.
+   */
+  public Output<T> outputRef() {
+    return outputRef;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return outputRef;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ScatterMin";
+  
+  private Output<T> outputRef;
+  
+  private ScatterMin(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    outputRef = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterMul.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterMul.java
new file mode 100644
index 00000000000..ba0892e1a04
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterMul.java
@@ -0,0 +1,136 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Multiplies sparse updates into a variable reference.
+ * <p>
+ * This operation computes
+ * <pre>{@code
+ *     # Scalar indices
+ *     ref[indices, ...] *= updates[...]
+ * 
+ *     # Vector indices (for each i)
+ *     ref[indices[i], ...] *= updates[i, ...]
+ * 
+ *     # High rank indices (for each i, ..., j)
+ *     ref[indices[i, ..., j], ...] *= updates[i, ..., j, ...]
+ * }</pre>
+ * This operation outputs `ref` after the update is done.
+ * This makes it easier to chain operations that need to use the reset value.
+ * <p>
+ * Duplicate entries are handled correctly: if multiple `indices` reference
+ * the same location, their contributions multiply.
+ * <p>
+ * Requires `updates.shape = indices.shape + ref.shape[1:]` or `updates.shape = []`.
+ * 
+ * @param <T> data type for {@code outputRef()} output
+ */
+@Operator
+public final class ScatterMul<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.core.ScatterMul}
+   */
+  public static class Options {
+    
+    /**
+     * @param useLocking If True, the operation will be protected by a lock;
+     * otherwise the behavior is undefined, but may exhibit less contention.
+     */
+    public Options useLocking(Boolean useLocking) {
+      this.useLocking = useLocking;
+      return this;
+    }
+    
+    private Boolean useLocking;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new ScatterMul operation.
+   * 
+   * @param scope current scope
+   * @param ref Should be from a `Variable` node.
+   * @param indices A tensor of indices into the first dimension of `ref`.
+   * @param updates A tensor of updated values to multiply to `ref`.
+   * @param options carries optional attributes values
+   * @return a new instance of ScatterMul
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType, U extends TNumber> ScatterMul<T> create(Scope scope, Operand<T> ref, Operand<U> indices, Operand<T> updates, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ScatterMul", scope.makeOpName("ScatterMul"));
+    opBuilder.addInput(ref.asOutput());
+    opBuilder.addInput(indices.asOutput());
+    opBuilder.addInput(updates.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.useLocking != null) {
+          opBuilder.setAttr("use_locking", opts.useLocking);
+        }
+      }
+    }
+    return new ScatterMul<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param useLocking If True, the operation will be protected by a lock;
+   * otherwise the behavior is undefined, but may exhibit less contention.
+   */
+  public static Options useLocking(Boolean useLocking) {
+    return new Options().useLocking(useLocking);
+  }
+  
+  /**
+   * = Same as `ref`.  Returned as a convenience for operations that want
+   * to use the updated values after the update is done.
+   */
+  public Output<T> outputRef() {
+    return outputRef;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return outputRef;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ScatterMul";
+  
+  private Output<T> outputRef;
+  
+  private ScatterMul(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    outputRef = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterNd.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterNd.java
new file mode 100644
index 00000000000..86acb2935db
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterNd.java
@@ -0,0 +1,158 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Scatter `updates` into a new tensor according to `indices`.
+ * <p>
+ * Creates a new tensor by applying sparse `updates` to individual values or
+ * slices within a tensor (initially zero for numeric, empty for string) of
+ * the given `shape` according to indices.  This operator is the inverse of the
+ * `tf.gather_nd` operator which extracts values or slices from a given tensor.
+ * <p>
+ * This operation is similar to tensor_scatter_add, except that the tensor is
+ * zero-initialized. Calling `tf.scatter_nd(indices, values, shape)` is identical
+ * to `tensor_scatter_add(tf.zeros(shape, values.dtype), indices, values)`
+ * <p>
+ * If `indices` contains duplicates, then their updates are accumulated (summed).
+ * <p>
+ * <b>WARNING</b>: The order in which updates are applied is nondeterministic, so the
+ * output will be nondeterministic if `indices` contains duplicates -- because
+ * of some numerical approximation issues, numbers summed in different order
+ * may yield different results.
+ * <p>
+ * `indices` is an integer tensor containing indices into a new tensor of shape
+ * `shape`.  The last dimension of `indices` can be at most the rank of `shape`:
+ * <p>
+ *     indices.shape[-1] <= shape.rank
+ * <p>
+ * The last dimension of `indices` corresponds to indices into elements
+ * (if `indices.shape[-1] = shape.rank`) or slices
+ * (if `indices.shape[-1] < shape.rank`) along dimension `indices.shape[-1]` of
+ * `shape`.  `updates` is a tensor with shape
+ * <p>
+ *     indices.shape[:-1] + shape[indices.shape[-1]:]
+ * <p>
+ * The simplest form of scatter is to insert individual elements in a tensor by
+ * index. For example, say we want to insert 4 scattered elements in a rank-1
+ * tensor with 8 elements.
+ * <p>
+ * <div style="width:70%; margin:auto; margin-bottom:10px; margin-top:20px;">
+ * <img style="width:100%" src="https://www.tensorflow.org/images/ScatterNd1.png" alt>
+ * </div>
+ * <p>
+ * In Python, this scatter operation would look like this:
+ * <pre>{@code
+ *     indices = tf.constant([[4], [3], [1], [7]])
+ *     updates = tf.constant([9, 10, 11, 12])
+ *     shape = tf.constant([8])
+ *     scatter = tf.scatter_nd(indices, updates, shape)
+ *     print(scatter)
+ * }</pre>
+ * The resulting tensor would look like this:
+ * <p>
+ *     [0, 11, 0, 10, 9, 0, 0, 12]
+ * <p>
+ * We can also, insert entire slices of a higher rank tensor all at once. For
+ * example, if we wanted to insert two slices in the first dimension of a
+ * rank-3 tensor with two matrices of new values.
+ * <p>
+ * <div style="width:70%; margin:auto; margin-bottom:10px; margin-top:20px;">
+ * <img style="width:100%" src="https://www.tensorflow.org/images/ScatterNd2.png" alt>
+ * </div>
+ * <p>
+ * In Python, this scatter operation would look like this:
+ * <pre>{@code
+ *     indices = tf.constant([[0], [2]])
+ *     updates = tf.constant([[[5, 5, 5, 5], [6, 6, 6, 6],
+ *                             [7, 7, 7, 7], [8, 8, 8, 8]],
+ *                            [[5, 5, 5, 5], [6, 6, 6, 6],
+ *                             [7, 7, 7, 7], [8, 8, 8, 8]]])
+ *     shape = tf.constant([4, 4, 4])
+ *     scatter = tf.scatter_nd(indices, updates, shape)
+ *     print(scatter)
+ * }</pre>
+ * The resulting tensor would look like this:
+ * <p>
+ *     [[[5, 5, 5, 5], [6, 6, 6, 6], [7, 7, 7, 7], [8, 8, 8, 8]],
+ *      [[0, 0, 0, 0], [0, 0, 0, 0], [0, 0, 0, 0], [0, 0, 0, 0]],
+ *      [[5, 5, 5, 5], [6, 6, 6, 6], [7, 7, 7, 7], [8, 8, 8, 8]],
+ *      [[0, 0, 0, 0], [0, 0, 0, 0], [0, 0, 0, 0], [0, 0, 0, 0]]]
+ * <p>
+ * Note that on CPU, if an out of bound index is found, an error is returned.
+ * On GPU, if an out of bound index is found, the index is ignored.
+ * 
+ * @param <U> data type for {@code output()} output
+ */
+@Operator
+public final class ScatterNd<U extends TType> extends RawOp implements Operand<U> {
+  
+  /**
+   * Factory method to create a class wrapping a new ScatterNd operation.
+   * 
+   * @param scope current scope
+   * @param indices Index tensor.
+   * @param updates Updates to scatter into output.
+   * @param shape 1-D. The shape of the resulting tensor.
+   * @return a new instance of ScatterNd
+   */
+  @Endpoint(describeByClass = true)
+  public static <U extends TType, T extends TNumber> ScatterNd<U> create(Scope scope, Operand<T> indices, Operand<U> updates, Operand<T> shape) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ScatterNd", scope.makeOpName("ScatterNd"));
+    opBuilder.addInput(indices.asOutput());
+    opBuilder.addInput(updates.asOutput());
+    opBuilder.addInput(shape.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new ScatterNd<U>(opBuilder.build());
+  }
+  
+  /**
+   * A new tensor with the given shape and updates applied according
+   * to the indices.
+   */
+  public Output<U> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<U> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ScatterNd";
+  
+  private Output<U> output;
+  
+  private ScatterNd(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterNdAdd.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterNdAdd.java
new file mode 100644
index 00000000000..09192e601e9
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterNdAdd.java
@@ -0,0 +1,151 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Applies sparse addition to individual values or slices in a Variable.
+ * <p>
+ * `ref` is a `Tensor` with rank `P` and `indices` is a `Tensor` of rank `Q`.
+ * <p>
+ * `indices` must be integer tensor, containing indices into `ref`.
+ * It must be shape `[d_0, ..., d_{Q-2}, K]` where `0 < K <= P`.
+ * <p>
+ * The innermost dimension of `indices` (with length `K`) corresponds to
+ * indices into elements (if `K = P`) or slices (if `K < P`) along the `K`th
+ * dimension of `ref`.
+ * <p>
+ * `updates` is `Tensor` of rank `Q-1+P-K` with shape:
+ * <pre>{@code
+ * [d_0, ..., d_{Q-2}, ref.shape[K], ..., ref.shape[P-1]]
+ * }</pre>
+ * For example, say we want to add 4 scattered elements to a rank-1 tensor to
+ * 8 elements. In Python, that addition would look like this:
+ * <pre>{@code
+ * ref = tf.Variable([1, 2, 3, 4, 5, 6, 7, 8])
+ * indices = tf.constant([[4], [3], [1], [7]])
+ * updates = tf.constant([9, 10, 11, 12])
+ * add = tf.scatter_nd_add(ref, indices, updates)
+ * with tf.Session() as sess:
+ *   print sess.run(add)
+ * }</pre>
+ * The resulting update to ref would look like this:
+ * <p>
+ *     [1, 13, 3, 14, 14, 6, 7, 20]
+ * <p>
+ * See `tf.scatter_nd` for more details about how to make updates to
+ * slices.
+ * 
+ * @param <T> data type for {@code outputRef()} output
+ */
+@Operator
+public final class ScatterNdAdd<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.core.ScatterNdAdd}
+   */
+  public static class Options {
+    
+    /**
+     * @param useLocking An optional bool. Defaults to True. If True, the assignment will
+     * be protected by a lock; otherwise the behavior is undefined,
+     * but may exhibit less contention.
+     */
+    public Options useLocking(Boolean useLocking) {
+      this.useLocking = useLocking;
+      return this;
+    }
+    
+    private Boolean useLocking;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new ScatterNdAdd operation.
+   * 
+   * @param scope current scope
+   * @param ref A mutable Tensor. Should be from a Variable node.
+   * @param indices A Tensor. Must be one of the following types: int32, int64.
+   * A tensor of indices into ref.
+   * @param updates A Tensor. Must have the same type as ref. A tensor of updated values
+   * to add to ref.
+   * @param options carries optional attributes values
+   * @return a new instance of ScatterNdAdd
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType, U extends TNumber> ScatterNdAdd<T> create(Scope scope, Operand<T> ref, Operand<U> indices, Operand<T> updates, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ScatterNdAdd", scope.makeOpName("ScatterNdAdd"));
+    opBuilder.addInput(ref.asOutput());
+    opBuilder.addInput(indices.asOutput());
+    opBuilder.addInput(updates.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.useLocking != null) {
+          opBuilder.setAttr("use_locking", opts.useLocking);
+        }
+      }
+    }
+    return new ScatterNdAdd<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param useLocking An optional bool. Defaults to True. If True, the assignment will
+   * be protected by a lock; otherwise the behavior is undefined,
+   * but may exhibit less contention.
+   */
+  public static Options useLocking(Boolean useLocking) {
+    return new Options().useLocking(useLocking);
+  }
+  
+  /**
+   * Same as ref. Returned as a convenience for operations that want
+   * to use the updated values after the update is done.
+   */
+  public Output<T> outputRef() {
+    return outputRef;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return outputRef;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ScatterNdAdd";
+  
+  private Output<T> outputRef;
+  
+  private ScatterNdAdd(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    outputRef = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterNdMax.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterNdMax.java
new file mode 100644
index 00000000000..da94c783cae
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterNdMax.java
@@ -0,0 +1,120 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Computes element-wise maximum.
+ * 
+ * @param <T> data type for {@code outputRef()} output
+ */
+public final class ScatterNdMax<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.core.ScatterNdMax}
+   */
+  public static class Options {
+    
+    /**
+     * @param useLocking An optional bool. Defaults to True. If True, the assignment will
+     * be protected by a lock; otherwise the behavior is undefined,
+     * but may exhibit less contention.
+     */
+    public Options useLocking(Boolean useLocking) {
+      this.useLocking = useLocking;
+      return this;
+    }
+    
+    private Boolean useLocking;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new ScatterNdMax operation.
+   * 
+   * @param scope current scope
+   * @param ref A mutable Tensor. Should be from a Variable node.
+   * @param indices A Tensor. Must be one of the following types: int32, int64.
+   * A tensor of indices into ref.
+   * @param updates A Tensor. Must have the same type as ref. A tensor of updated values
+   * to add to ref.
+   * @param options carries optional attributes values
+   * @return a new instance of ScatterNdMax
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType, U extends TNumber> ScatterNdMax<T> create(Scope scope, Operand<T> ref, Operand<U> indices, Operand<T> updates, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ScatterNdMax", scope.makeOpName("ScatterNdMax"));
+    opBuilder.addInput(ref.asOutput());
+    opBuilder.addInput(indices.asOutput());
+    opBuilder.addInput(updates.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.useLocking != null) {
+          opBuilder.setAttr("use_locking", opts.useLocking);
+        }
+      }
+    }
+    return new ScatterNdMax<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param useLocking An optional bool. Defaults to True. If True, the assignment will
+   * be protected by a lock; otherwise the behavior is undefined,
+   * but may exhibit less contention.
+   */
+  public static Options useLocking(Boolean useLocking) {
+    return new Options().useLocking(useLocking);
+  }
+  
+  /**
+   * Same as ref. Returned as a convenience for operations that want
+   * to use the updated values after the update is done.
+   */
+  public Output<T> outputRef() {
+    return outputRef;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return outputRef;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ScatterNdMax";
+  
+  private Output<T> outputRef;
+  
+  private ScatterNdMax(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    outputRef = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterNdMin.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterNdMin.java
new file mode 100644
index 00000000000..5aea70bc929
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterNdMin.java
@@ -0,0 +1,120 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Computes element-wise minimum.
+ * 
+ * @param <T> data type for {@code outputRef()} output
+ */
+public final class ScatterNdMin<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.core.ScatterNdMin}
+   */
+  public static class Options {
+    
+    /**
+     * @param useLocking An optional bool. Defaults to True. If True, the assignment will
+     * be protected by a lock; otherwise the behavior is undefined,
+     * but may exhibit less contention.
+     */
+    public Options useLocking(Boolean useLocking) {
+      this.useLocking = useLocking;
+      return this;
+    }
+    
+    private Boolean useLocking;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new ScatterNdMin operation.
+   * 
+   * @param scope current scope
+   * @param ref A mutable Tensor. Should be from a Variable node.
+   * @param indices A Tensor. Must be one of the following types: int32, int64.
+   * A tensor of indices into ref.
+   * @param updates A Tensor. Must have the same type as ref. A tensor of updated values
+   * to add to ref.
+   * @param options carries optional attributes values
+   * @return a new instance of ScatterNdMin
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType, U extends TNumber> ScatterNdMin<T> create(Scope scope, Operand<T> ref, Operand<U> indices, Operand<T> updates, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ScatterNdMin", scope.makeOpName("ScatterNdMin"));
+    opBuilder.addInput(ref.asOutput());
+    opBuilder.addInput(indices.asOutput());
+    opBuilder.addInput(updates.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.useLocking != null) {
+          opBuilder.setAttr("use_locking", opts.useLocking);
+        }
+      }
+    }
+    return new ScatterNdMin<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param useLocking An optional bool. Defaults to True. If True, the assignment will
+   * be protected by a lock; otherwise the behavior is undefined,
+   * but may exhibit less contention.
+   */
+  public static Options useLocking(Boolean useLocking) {
+    return new Options().useLocking(useLocking);
+  }
+  
+  /**
+   * Same as ref. Returned as a convenience for operations that want
+   * to use the updated values after the update is done.
+   */
+  public Output<T> outputRef() {
+    return outputRef;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return outputRef;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ScatterNdMin";
+  
+  private Output<T> outputRef;
+  
+  private ScatterNdMin(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    outputRef = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterNdNonAliasingAdd.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterNdNonAliasingAdd.java
new file mode 100644
index 00000000000..98ce3cceba4
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterNdNonAliasingAdd.java
@@ -0,0 +1,117 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Applies sparse addition to `input` using individual values or slices
+ * <p>
+ * from `updates` according to indices `indices`.  The updates are non-aliasing:
+ * `input` is only modified in-place if no other operations will use it.
+ * Otherwise, a copy of `input` is made.  This operation has a gradient with
+ * respect to both `input` and `updates`.
+ * <p>
+ * `input` is a `Tensor` with rank `P` and `indices` is a `Tensor` of rank `Q`.
+ * <p>
+ * `indices` must be integer tensor, containing indices into `input`.
+ * It must be shape \\([d_0, ..., d_{Q-2}, K]\\) where `0 < K <= P`.
+ * <p>
+ * The innermost dimension of `indices` (with length `K`) corresponds to
+ * indices into elements (if `K = P`) or `(P-K)`-dimensional slices
+ * (if `K < P`) along the `K`th dimension of `input`.
+ * <p>
+ * `updates` is `Tensor` of rank `Q-1+P-K` with shape:
+ * <p>
+ * $$[d_0, ..., d_{Q-2}, input.shape[K], ..., input.shape[P-1]].$$
+ * <p>
+ * For example, say we want to add 4 scattered elements to a rank-1 tensor to 8
+ * elements. In Python, that addition would look like this:
+ * <p>
+ *     input = tf.constant([1, 2, 3, 4, 5, 6, 7, 8])
+ *     indices = tf.constant([[4], [3], [1], [7]])
+ *     updates = tf.constant([9, 10, 11, 12])
+ *     output = tf.scatter_nd_non_aliasing_add(input, indices, updates)
+ *     with tf.Session() as sess:
+ *       print(sess.run(output))
+ * <p>
+ * The resulting value `output` would look like this:
+ * <p>
+ *     [1, 13, 3, 14, 14, 6, 7, 20]
+ * <p>
+ * See `tf.scatter_nd` for more details about how to make updates to slices.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator
+public final class ScatterNdNonAliasingAdd<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new ScatterNdNonAliasingAdd operation.
+   * 
+   * @param scope current scope
+   * @param input A Tensor.
+   * @param indices A Tensor. Must be one of the following types: `int32`, `int64`.
+   * A tensor of indices into `input`.
+   * @param updates A Tensor. Must have the same type as ref. A tensor of updated values
+   * to add to `input`.
+   * @return a new instance of ScatterNdNonAliasingAdd
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType, U extends TNumber> ScatterNdNonAliasingAdd<T> create(Scope scope, Operand<T> input, Operand<U> indices, Operand<T> updates) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ScatterNdNonAliasingAdd", scope.makeOpName("ScatterNdNonAliasingAdd"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(indices.asOutput());
+    opBuilder.addInput(updates.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new ScatterNdNonAliasingAdd<T>(opBuilder.build());
+  }
+  
+  /**
+   * A `Tensor` with the same shape as `input`, containing values of `input`
+   * updated with `updates`.
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ScatterNdNonAliasingAdd";
+  
+  private Output<T> output;
+  
+  private ScatterNdNonAliasingAdd(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterNdSub.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterNdSub.java
new file mode 100644
index 00000000000..39327cd553d
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterNdSub.java
@@ -0,0 +1,153 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Applies sparse subtraction to individual values or slices in a Variable.
+ * <p>
+ * within a given variable according to `indices`.
+ * <p>
+ * `ref` is a `Tensor` with rank `P` and `indices` is a `Tensor` of rank `Q`.
+ * <p>
+ * `indices` must be integer tensor, containing indices into `ref`.
+ * It must be shape `[d_0, ..., d_{Q-2}, K]` where `0 < K <= P`.
+ * <p>
+ * The innermost dimension of `indices` (with length `K`) corresponds to
+ * indices into elements (if `K = P`) or slices (if `K < P`) along the `K`th
+ * dimension of `ref`.
+ * <p>
+ * `updates` is `Tensor` of rank `Q-1+P-K` with shape:
+ * <pre>{@code
+ * [d_0, ..., d_{Q-2}, ref.shape[K], ..., ref.shape[P-1]]
+ * }</pre>
+ * For example, say we want to subtract 4 scattered elements from a rank-1 tensor
+ * with 8 elements. In Python, that subtraction would look like this:
+ * <pre>{@code
+ * ref = tf.Variable([1, 2, 3, 4, 5, 6, 7, 8])
+ * indices = tf.constant([[4], [3], [1], [7]])
+ * updates = tf.constant([9, 10, 11, 12])
+ * sub = tf.scatter_nd_sub(ref, indices, updates)
+ * with tf.Session() as sess:
+ *   print sess.run(sub)
+ * }</pre>
+ * The resulting update to ref would look like this:
+ * <p>
+ *     [1, -9, 3, -6, -4, 6, 7, -4]
+ * <p>
+ * See `tf.scatter_nd` for more details about how to make updates to
+ * slices.
+ * 
+ * @param <T> data type for {@code outputRef()} output
+ */
+@Operator
+public final class ScatterNdSub<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.core.ScatterNdSub}
+   */
+  public static class Options {
+    
+    /**
+     * @param useLocking An optional bool. Defaults to True. If True, the assignment will
+     * be protected by a lock; otherwise the behavior is undefined,
+     * but may exhibit less contention.
+     */
+    public Options useLocking(Boolean useLocking) {
+      this.useLocking = useLocking;
+      return this;
+    }
+    
+    private Boolean useLocking;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new ScatterNdSub operation.
+   * 
+   * @param scope current scope
+   * @param ref A mutable Tensor. Should be from a Variable node.
+   * @param indices A Tensor. Must be one of the following types: int32, int64.
+   * A tensor of indices into ref.
+   * @param updates A Tensor. Must have the same type as ref. A tensor of updated values
+   * to subtract from ref.
+   * @param options carries optional attributes values
+   * @return a new instance of ScatterNdSub
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType, U extends TNumber> ScatterNdSub<T> create(Scope scope, Operand<T> ref, Operand<U> indices, Operand<T> updates, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ScatterNdSub", scope.makeOpName("ScatterNdSub"));
+    opBuilder.addInput(ref.asOutput());
+    opBuilder.addInput(indices.asOutput());
+    opBuilder.addInput(updates.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.useLocking != null) {
+          opBuilder.setAttr("use_locking", opts.useLocking);
+        }
+      }
+    }
+    return new ScatterNdSub<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param useLocking An optional bool. Defaults to True. If True, the assignment will
+   * be protected by a lock; otherwise the behavior is undefined,
+   * but may exhibit less contention.
+   */
+  public static Options useLocking(Boolean useLocking) {
+    return new Options().useLocking(useLocking);
+  }
+  
+  /**
+   * Same as ref. Returned as a convenience for operations that want
+   * to use the updated values after the update is done.
+   */
+  public Output<T> outputRef() {
+    return outputRef;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return outputRef;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ScatterNdSub";
+  
+  private Output<T> outputRef;
+  
+  private ScatterNdSub(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    outputRef = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterNdUpdate.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterNdUpdate.java
new file mode 100644
index 00000000000..87e5949c07c
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterNdUpdate.java
@@ -0,0 +1,155 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Applies sparse `updates` to individual values or slices within a given
+ * <p>
+ * variable according to `indices`.
+ * <p>
+ * `ref` is a `Tensor` with rank `P` and `indices` is a `Tensor` of rank `Q`.
+ * <p>
+ * `indices` must be integer tensor, containing indices into `ref`.
+ * It must be shape \\([d_0, ..., d_{Q-2}, K]\\) where `0 < K <= P`.
+ * <p>
+ * The innermost dimension of `indices` (with length `K`) corresponds to
+ * indices into elements (if `K = P`) or slices (if `K < P`) along the `K`th
+ * dimension of `ref`.
+ * <p>
+ * `updates` is `Tensor` of rank `Q-1+P-K` with shape:
+ * <p>
+ * $$[d_0, ..., d_{Q-2}, ref.shape[K], ..., ref.shape[P-1]].$$
+ * <p>
+ * For example, say we want to update 4 scattered elements to a rank-1 tensor to
+ * 8 elements. In Python, that update would look like this:
+ * <pre>{@code
+ *     ref = tf.Variable([1, 2, 3, 4, 5, 6, 7, 8])
+ *     indices = tf.constant([[4], [3], [1] ,[7]])
+ *     updates = tf.constant([9, 10, 11, 12])
+ *     update = tf.scatter_nd_update(ref, indices, updates)
+ *     with tf.Session() as sess:
+ *       print sess.run(update)
+ * }</pre>
+ * The resulting update to ref would look like this:
+ * <p>
+ *     [1, 11, 3, 10, 9, 6, 7, 12]
+ * <p>
+ * See `tf.scatter_nd` for more details about how to make updates to
+ * slices.
+ * <p>
+ * See also `tf.scatter_update` and `tf.batch_scatter_update`.
+ * 
+ * @param <T> data type for {@code outputRef()} output
+ */
+@Operator
+public final class ScatterNdUpdate<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.core.ScatterNdUpdate}
+   */
+  public static class Options {
+    
+    /**
+     * @param useLocking An optional bool. Defaults to True. If True, the assignment will
+     * be protected by a lock; otherwise the behavior is undefined,
+     * but may exhibit less contention.
+     */
+    public Options useLocking(Boolean useLocking) {
+      this.useLocking = useLocking;
+      return this;
+    }
+    
+    private Boolean useLocking;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new ScatterNdUpdate operation.
+   * 
+   * @param scope current scope
+   * @param ref A mutable Tensor. Should be from a Variable node.
+   * @param indices A Tensor. Must be one of the following types: int32, int64.
+   * A tensor of indices into ref.
+   * @param updates A Tensor. Must have the same type as ref. A tensor of updated
+   * values to add to ref.
+   * @param options carries optional attributes values
+   * @return a new instance of ScatterNdUpdate
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType, U extends TNumber> ScatterNdUpdate<T> create(Scope scope, Operand<T> ref, Operand<U> indices, Operand<T> updates, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ScatterNdUpdate", scope.makeOpName("ScatterNdUpdate"));
+    opBuilder.addInput(ref.asOutput());
+    opBuilder.addInput(indices.asOutput());
+    opBuilder.addInput(updates.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.useLocking != null) {
+          opBuilder.setAttr("use_locking", opts.useLocking);
+        }
+      }
+    }
+    return new ScatterNdUpdate<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param useLocking An optional bool. Defaults to True. If True, the assignment will
+   * be protected by a lock; otherwise the behavior is undefined,
+   * but may exhibit less contention.
+   */
+  public static Options useLocking(Boolean useLocking) {
+    return new Options().useLocking(useLocking);
+  }
+  
+  /**
+   * Same as ref. Returned as a convenience for operations that want to
+   * use the updated values after the update is done.
+   */
+  public Output<T> outputRef() {
+    return outputRef;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return outputRef;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ScatterNdUpdate";
+  
+  private Output<T> outputRef;
+  
+  private ScatterNdUpdate(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    outputRef = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterSub.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterSub.java
new file mode 100644
index 00000000000..bad245da96e
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterSub.java
@@ -0,0 +1,139 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Subtracts sparse updates to a variable reference.
+ * <p>
+ * <pre>{@code
+ *     # Scalar indices
+ *     ref[indices, ...] -= updates[...]
+ * 
+ *     # Vector indices (for each i)
+ *     ref[indices[i], ...] -= updates[i, ...]
+ * 
+ *     # High rank indices (for each i, ..., j)
+ *     ref[indices[i, ..., j], ...] -= updates[i, ..., j, ...]
+ * }</pre>
+ * This operation outputs `ref` after the update is done.
+ * This makes it easier to chain operations that need to use the reset value.
+ * <p>
+ * Duplicate entries are handled correctly: if multiple `indices` reference
+ * the same location, their (negated) contributions add.
+ * <p>
+ * Requires `updates.shape = indices.shape + ref.shape[1:]` or `updates.shape = []`.
+ * <p>
+ * <div style="width:70%; margin:auto; margin-bottom:10px; margin-top:20px;">
+ * <img style="width:100%" src="https://www.tensorflow.org/images/ScatterSub.png" alt>
+ * </div>
+ * 
+ * @param <T> data type for {@code outputRef()} output
+ */
+@Operator
+public final class ScatterSub<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.core.ScatterSub}
+   */
+  public static class Options {
+    
+    /**
+     * @param useLocking If True, the subtraction will be protected by a lock;
+     * otherwise the behavior is undefined, but may exhibit less contention.
+     */
+    public Options useLocking(Boolean useLocking) {
+      this.useLocking = useLocking;
+      return this;
+    }
+    
+    private Boolean useLocking;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new ScatterSub operation.
+   * 
+   * @param scope current scope
+   * @param ref Should be from a `Variable` node.
+   * @param indices A tensor of indices into the first dimension of `ref`.
+   * @param updates A tensor of updated values to subtract from `ref`.
+   * @param options carries optional attributes values
+   * @return a new instance of ScatterSub
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType, U extends TNumber> ScatterSub<T> create(Scope scope, Operand<T> ref, Operand<U> indices, Operand<T> updates, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ScatterSub", scope.makeOpName("ScatterSub"));
+    opBuilder.addInput(ref.asOutput());
+    opBuilder.addInput(indices.asOutput());
+    opBuilder.addInput(updates.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.useLocking != null) {
+          opBuilder.setAttr("use_locking", opts.useLocking);
+        }
+      }
+    }
+    return new ScatterSub<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param useLocking If True, the subtraction will be protected by a lock;
+   * otherwise the behavior is undefined, but may exhibit less contention.
+   */
+  public static Options useLocking(Boolean useLocking) {
+    return new Options().useLocking(useLocking);
+  }
+  
+  /**
+   * = Same as `ref`.  Returned as a convenience for operations that want
+   * to use the updated values after the update is done.
+   */
+  public Output<T> outputRef() {
+    return outputRef;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return outputRef;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ScatterSub";
+  
+  private Output<T> outputRef;
+  
+  private ScatterSub(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    outputRef = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterUpdate.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterUpdate.java
new file mode 100644
index 00000000000..9790e2fc934
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ScatterUpdate.java
@@ -0,0 +1,143 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Applies sparse updates to a variable reference.
+ * <p>
+ * This operation computes
+ * <pre>{@code
+ *     # Scalar indices
+ *     ref[indices, ...] = updates[...]
+ * 
+ *     # Vector indices (for each i)
+ *     ref[indices[i], ...] = updates[i, ...]
+ * 
+ *     # High rank indices (for each i, ..., j)
+ *     ref[indices[i, ..., j], ...] = updates[i, ..., j, ...]
+ * }</pre>
+ * This operation outputs `ref` after the update is done.
+ * This makes it easier to chain operations that need to use the reset value.
+ * <p>
+ * If values in `ref` is to be updated more than once, because there are
+ * duplicate entries in `indices`, the order at which the updates happen
+ * for each value is undefined.
+ * <p>
+ * Requires `updates.shape = indices.shape + ref.shape[1:]` or `updates.shape = []`.
+ * <p>
+ * <div style="width:70%; margin:auto; margin-bottom:10px; margin-top:20px;">
+ * <img style="width:100%" src="https://www.tensorflow.org/images/ScatterUpdate.png" alt>
+ * </div>
+ * <p>
+ * See also `tf.batch_scatter_update` and `tf.scatter_nd_update`.
+ * 
+ * @param <T> data type for {@code outputRef()} output
+ */
+@Operator
+public final class ScatterUpdate<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.core.ScatterUpdate}
+   */
+  public static class Options {
+    
+    /**
+     * @param useLocking If True, the assignment will be protected by a lock;
+     * otherwise the behavior is undefined, but may exhibit less contention.
+     */
+    public Options useLocking(Boolean useLocking) {
+      this.useLocking = useLocking;
+      return this;
+    }
+    
+    private Boolean useLocking;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new ScatterUpdate operation.
+   * 
+   * @param scope current scope
+   * @param ref Should be from a `Variable` node.
+   * @param indices A tensor of indices into the first dimension of `ref`.
+   * @param updates A tensor of updated values to store in `ref`.
+   * @param options carries optional attributes values
+   * @return a new instance of ScatterUpdate
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType, U extends TNumber> ScatterUpdate<T> create(Scope scope, Operand<T> ref, Operand<U> indices, Operand<T> updates, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ScatterUpdate", scope.makeOpName("ScatterUpdate"));
+    opBuilder.addInput(ref.asOutput());
+    opBuilder.addInput(indices.asOutput());
+    opBuilder.addInput(updates.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.useLocking != null) {
+          opBuilder.setAttr("use_locking", opts.useLocking);
+        }
+      }
+    }
+    return new ScatterUpdate<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param useLocking If True, the assignment will be protected by a lock;
+   * otherwise the behavior is undefined, but may exhibit less contention.
+   */
+  public static Options useLocking(Boolean useLocking) {
+    return new Options().useLocking(useLocking);
+  }
+  
+  /**
+   * = Same as `ref`.  Returned as a convenience for operations that want
+   * to use the updated values after the update is done.
+   */
+  public Output<T> outputRef() {
+    return outputRef;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return outputRef;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ScatterUpdate";
+  
+  private Output<T> outputRef;
+  
+  private ScatterUpdate(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    outputRef = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Select.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Select.java
new file mode 100644
index 00000000000..0aaa338eb1a
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Select.java
@@ -0,0 +1,77 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TBool;
+import org.tensorflow.types.family.TType;
+
+/**
+ * @param <T> data type for {@code output()} output
+ */
+@Operator
+public final class Select<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new Select operation.
+   * 
+   * @param scope current scope
+   * @param condition 
+   * @param t 
+   * @param e 
+   * @return a new instance of Select
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> Select<T> create(Scope scope, Operand<TBool> condition, Operand<T> t, Operand<T> e) {
+    OperationBuilder opBuilder = scope.env().opBuilder("SelectV2", scope.makeOpName("Select"));
+    opBuilder.addInput(condition.asOutput());
+    opBuilder.addInput(t.asOutput());
+    opBuilder.addInput(e.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new Select<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "SelectV2";
+  
+  private Output<T> output;
+  
+  private Select(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Send.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Send.java
new file mode 100644
index 00000000000..d679b85319a
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Send.java
@@ -0,0 +1,103 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Sends the named tensor from send_device to recv_device.
+ */
+public final class Send extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.core.Send}
+   */
+  public static class Options {
+    
+    /**
+     * @param clientTerminated If set to true, this indicates that the node was added
+     * to the graph as a result of a client-side feed or fetch of Tensor data,
+     * in which case the corresponding send or recv is expected to be managed
+     * locally by the caller.
+     */
+    public Options clientTerminated(Boolean clientTerminated) {
+      this.clientTerminated = clientTerminated;
+      return this;
+    }
+    
+    private Boolean clientTerminated;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new Send operation.
+   * 
+   * @param scope current scope
+   * @param tensor The tensor to send.
+   * @param tensorName The name of the tensor to send.
+   * @param sendDevice The name of the device sending the tensor.
+   * @param sendDeviceIncarnation The current incarnation of send_device.
+   * @param recvDevice The name of the device receiving the tensor.
+   * @param options carries optional attributes values
+   * @return a new instance of Send
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> Send create(Scope scope, Operand<T> tensor, String tensorName, String sendDevice, Long sendDeviceIncarnation, String recvDevice, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Send", scope.makeOpName("Send"));
+    opBuilder.addInput(tensor.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("tensor_name", tensorName);
+    opBuilder.setAttr("send_device", sendDevice);
+    opBuilder.setAttr("send_device_incarnation", sendDeviceIncarnation);
+    opBuilder.setAttr("recv_device", recvDevice);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.clientTerminated != null) {
+          opBuilder.setAttr("client_terminated", opts.clientTerminated);
+        }
+      }
+    }
+    return new Send(opBuilder.build());
+  }
+  
+  /**
+   * @param clientTerminated If set to true, this indicates that the node was added
+   * to the graph as a result of a client-side feed or fetch of Tensor data,
+   * in which case the corresponding send or recv is expected to be managed
+   * locally by the caller.
+   */
+  public static Options clientTerminated(Boolean clientTerminated) {
+    return new Options().clientTerminated(clientTerminated);
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Send";
+  
+  private Send(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/SetDiff1d.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/SetDiff1d.java
new file mode 100644
index 00000000000..d1c53319c4f
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/SetDiff1d.java
@@ -0,0 +1,119 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Computes the difference between two lists of numbers or strings.
+ * <p>
+ * Given a list `x` and a list `y`, this operation returns a list `out` that
+ * represents all values that are in `x` but not in `y`. The returned list `out`
+ * is sorted in the same order that the numbers appear in `x` (duplicates are
+ * preserved). This operation also returns a list `idx` that represents the
+ * position of each `out` element in `x`. In other words:
+ * <p>
+ * `out[i] = x[idx[i]] for i in [0, 1, ..., len(out) - 1]`
+ * <p>
+ * For example, given this input:
+ * <pre>{@code
+ * x = [1, 2, 3, 4, 5, 6]
+ * y = [1, 3, 5]
+ * }</pre>
+ * This operation would return:
+ * <pre>{@code
+ * out ==> [2, 4, 6]
+ * idx ==> [1, 3, 5]
+ * }</pre>
+ * 
+ * 
+ * @param <T> data type for {@code out()} output
+ * @param <U> data type for {@code idx()} output
+ */
+@Operator
+public final class SetDiff1d<T extends TType, U extends TNumber> extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new SetDiff1d operation.
+   * 
+   * @param scope current scope
+   * @param x 1-D. Values to keep.
+   * @param y 1-D. Values to remove.
+   * @param outIdx 
+   * @return a new instance of SetDiff1d
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType, U extends TNumber> SetDiff1d<T, U> create(Scope scope, Operand<T> x, Operand<T> y, Class<U> outIdx) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ListDiff", scope.makeOpName("SetDiff1d"));
+    opBuilder.addInput(x.asOutput());
+    opBuilder.addInput(y.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("out_idx", outIdx);
+    return new SetDiff1d<T, U>(opBuilder.build());
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new SetDiff1d operation using default output types.
+   * 
+   * @param scope current scope
+   * @param x 1-D. Values to keep.
+   * @param y 1-D. Values to remove.
+   * @return a new instance of SetDiff1d
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> SetDiff1d<T, TInt32> create(Scope scope, Operand<T> x, Operand<T> y) {
+    return create(scope, x, y, TInt32.class);
+  }
+  
+  /**
+   * 1-D. Values present in `x` but not in `y`.
+   */
+  public Output<T> out() {
+    return out;
+  }
+  
+  /**
+   * 1-D. Positions of `x` values preserved in `out`.
+   */
+  public Output<U> idx() {
+    return idx;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ListDiff";
+  
+  private Output<T> out;
+  private Output<U> idx;
+  
+  private SetDiff1d(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    out = operation.output(outputIdx++);
+    idx = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/SetSize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/SetSize.java
new file mode 100644
index 00000000000..2fcdf728542
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/SetSize.java
@@ -0,0 +1,122 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Number of unique elements along last dimension of input `set`.
+ * <p>
+ * Input `set` is a `SparseTensor` represented by `set_indices`, `set_values`,
+ * and `set_shape`. The last dimension contains values in a set, duplicates are
+ * allowed but ignored.
+ * <p>
+ * If `validate_indices` is `True`, this op validates the order and range of `set`
+ * indices.
+ */
+@Operator
+public final class SetSize extends RawOp implements Operand<TInt32> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.core.SetSize}
+   */
+  public static class Options {
+    
+    /**
+     * @param validateIndices 
+     */
+    public Options validateIndices(Boolean validateIndices) {
+      this.validateIndices = validateIndices;
+      return this;
+    }
+    
+    private Boolean validateIndices;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new SetSize operation.
+   * 
+   * @param scope current scope
+   * @param setIndices 2D `Tensor`, indices of a `SparseTensor`.
+   * @param setValues 1D `Tensor`, values of a `SparseTensor`.
+   * @param setShape 1D `Tensor`, shape of a `SparseTensor`.
+   * @param options carries optional attributes values
+   * @return a new instance of SetSize
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> SetSize create(Scope scope, Operand<TInt64> setIndices, Operand<T> setValues, Operand<TInt64> setShape, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("SetSize", scope.makeOpName("SetSize"));
+    opBuilder.addInput(setIndices.asOutput());
+    opBuilder.addInput(setValues.asOutput());
+    opBuilder.addInput(setShape.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.validateIndices != null) {
+          opBuilder.setAttr("validate_indices", opts.validateIndices);
+        }
+      }
+    }
+    return new SetSize(opBuilder.build());
+  }
+  
+  /**
+   * @param validateIndices 
+   */
+  public static Options validateIndices(Boolean validateIndices) {
+    return new Options().validateIndices(validateIndices);
+  }
+  
+  /**
+   * For `set` ranked `n`, this is a `Tensor` with rank `n-1`, and the same 1st
+   * `n-1` dimensions as `set`. Each value is the number of unique elements in
+   * the corresponding `[0...n-1]` dimension of `set`.
+   */
+  public Output<TInt32> size() {
+    return size;
+  }
+  
+  @Override
+  public Output<TInt32> asOutput() {
+    return size;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "SetSize";
+  
+  private Output<TInt32> size;
+  
+  private SetSize(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    size = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Shape.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Shape.java
new file mode 100644
index 00000000000..3ee92514a35
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Shape.java
@@ -0,0 +1,99 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Returns the shape of a tensor.
+ * <p>
+ * This operation returns a 1-D integer tensor representing the shape of `input`.
+ * <p>
+ * For example:
+ * <pre>{@code
+ * # 't' is [[[1, 1, 1], [2, 2, 2]], [[3, 3, 3], [4, 4, 4]]]
+ * shape(t) ==> [2, 2, 3]
+ * }</pre>
+ * 
+ * 
+ * @param <U> data type for {@code output()} output
+ */
+@Operator
+public final class Shape<U extends TNumber> extends RawOp implements Operand<U> {
+  
+  /**
+   * Factory method to create a class wrapping a new Shape operation.
+   * 
+   * @param scope current scope
+   * @param input 
+   * @param outType 
+   * @return a new instance of Shape
+   */
+  @Endpoint(describeByClass = true)
+  public static <U extends TNumber, T extends TType> Shape<U> create(Scope scope, Operand<T> input, Class<U> outType) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Shape", scope.makeOpName("Shape"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("out_type", outType);
+    return new Shape<U>(opBuilder.build());
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new Shape operation using default output types.
+   * 
+   * @param scope current scope
+   * @param input 
+   * @return a new instance of Shape
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> Shape<TInt32> create(Scope scope, Operand<T> input) {
+    return create(scope, input, TInt32.class);
+  }
+  
+  /**
+   */
+  public Output<U> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<U> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Shape";
+  
+  private Output<U> output;
+  
+  private Shape(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ShapeN.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ShapeN.java
new file mode 100644
index 00000000000..1b1af62e7ac
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ShapeN.java
@@ -0,0 +1,100 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import java.util.Arrays;
+import java.util.Iterator;
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.Operands;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Returns shape of tensors.
+ * <p>
+ * This operation returns N 1-D integer tensors representing shape of `input[i]s`.
+ * 
+ * @param <U> data type for {@code output()} output
+ */
+@Operator
+public final class ShapeN<U extends TNumber> extends RawOp implements Iterable<Operand<U>> {
+  
+  /**
+   * Factory method to create a class wrapping a new ShapeN operation.
+   * 
+   * @param scope current scope
+   * @param input 
+   * @param outType 
+   * @return a new instance of ShapeN
+   */
+  @Endpoint(describeByClass = true)
+  public static <U extends TNumber, T extends TType> ShapeN<U> create(Scope scope, Iterable<Operand<T>> input, Class<U> outType) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ShapeN", scope.makeOpName("ShapeN"));
+    opBuilder.addInputList(Operands.asOutputs(input));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("out_type", outType);
+    return new ShapeN<U>(opBuilder.build());
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new ShapeN operation using default output types.
+   * 
+   * @param scope current scope
+   * @param input 
+   * @return a new instance of ShapeN
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> ShapeN<TInt32> create(Scope scope, Iterable<Operand<T>> input) {
+    return create(scope, input, TInt32.class);
+  }
+  
+  /**
+   */
+  public List<Output<U>> output() {
+    return output;
+  }
+  
+  @Override
+  @SuppressWarnings({"rawtypes", "unchecked"})
+  public Iterator<Operand<U>> iterator() {
+    return (Iterator) output.iterator();
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ShapeN";
+  
+  private List<Output<U>> output;
+  
+  @SuppressWarnings("unchecked")
+  private ShapeN(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    int outputLength = operation.outputListLength("output");
+    output = Arrays.asList((Output<U>[])operation.outputList(outputIdx, outputLength));
+    outputIdx += outputLength;
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Size.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Size.java
new file mode 100644
index 00000000000..ddd912f68f8
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Size.java
@@ -0,0 +1,100 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Returns the size of a tensor.
+ * <p>
+ * This operation returns an integer representing the number of elements in
+ * `input`.
+ * <p>
+ * For example:
+ * <pre>{@code
+ * # 't' is [[[1, 1,, 1], [2, 2, 2]], [[3, 3, 3], [4, 4, 4]]]]
+ * size(t) ==> 12
+ * }</pre>
+ * 
+ * 
+ * @param <U> data type for {@code output()} output
+ */
+@Operator
+public final class Size<U extends TNumber> extends RawOp implements Operand<U> {
+  
+  /**
+   * Factory method to create a class wrapping a new Size operation.
+   * 
+   * @param scope current scope
+   * @param input 
+   * @param outType 
+   * @return a new instance of Size
+   */
+  @Endpoint(describeByClass = true)
+  public static <U extends TNumber, T extends TType> Size<U> create(Scope scope, Operand<T> input, Class<U> outType) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Size", scope.makeOpName("Size"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("out_type", outType);
+    return new Size<U>(opBuilder.build());
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new Size operation using default output types.
+   * 
+   * @param scope current scope
+   * @param input 
+   * @return a new instance of Size
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> Size<TInt32> create(Scope scope, Operand<T> input) {
+    return create(scope, input, TInt32.class);
+  }
+  
+  /**
+   */
+  public Output<U> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<U> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Size";
+  
+  private Output<U> output;
+  
+  private Size(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Skipgram.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Skipgram.java
new file mode 100644
index 00000000000..9c63d317f98
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Skipgram.java
@@ -0,0 +1,201 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.TString;
+
+/**
+ * Parses a text file and creates a batch of examples.
+ */
+@Operator
+public final class Skipgram extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.core.Skipgram}
+   */
+  public static class Options {
+    
+    /**
+     * @param windowSize The number of words to predict to the left and right of the target.
+     */
+    public Options windowSize(Long windowSize) {
+      this.windowSize = windowSize;
+      return this;
+    }
+    
+    /**
+     * @param minCount The minimum number of word occurrences for it to be included in the
+     * vocabulary.
+     */
+    public Options minCount(Long minCount) {
+      this.minCount = minCount;
+      return this;
+    }
+    
+    /**
+     * @param subsample Threshold for word occurrence. Words that appear with higher
+     * frequency will be randomly down-sampled. Set to 0 to disable.
+     */
+    public Options subsample(Float subsample) {
+      this.subsample = subsample;
+      return this;
+    }
+    
+    private Long windowSize;
+    private Long minCount;
+    private Float subsample;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new Skipgram operation.
+   * 
+   * @param scope current scope
+   * @param filename The corpus's text file name.
+   * @param batchSize The size of produced batch.
+   * @param options carries optional attributes values
+   * @return a new instance of Skipgram
+   */
+  @Endpoint(describeByClass = true)
+  public static Skipgram create(Scope scope, String filename, Long batchSize, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Skipgram", scope.makeOpName("Skipgram"));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("filename", filename);
+    opBuilder.setAttr("batch_size", batchSize);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.windowSize != null) {
+          opBuilder.setAttr("window_size", opts.windowSize);
+        }
+        if (opts.minCount != null) {
+          opBuilder.setAttr("min_count", opts.minCount);
+        }
+        if (opts.subsample != null) {
+          opBuilder.setAttr("subsample", opts.subsample);
+        }
+      }
+    }
+    return new Skipgram(opBuilder.build());
+  }
+  
+  /**
+   * @param windowSize The number of words to predict to the left and right of the target.
+   */
+  public static Options windowSize(Long windowSize) {
+    return new Options().windowSize(windowSize);
+  }
+  
+  /**
+   * @param minCount The minimum number of word occurrences for it to be included in the
+   * vocabulary.
+   */
+  public static Options minCount(Long minCount) {
+    return new Options().minCount(minCount);
+  }
+  
+  /**
+   * @param subsample Threshold for word occurrence. Words that appear with higher
+   * frequency will be randomly down-sampled. Set to 0 to disable.
+   */
+  public static Options subsample(Float subsample) {
+    return new Options().subsample(subsample);
+  }
+  
+  /**
+   * A vector of words in the corpus.
+   */
+  public Output<TString> vocabWord() {
+    return vocabWord;
+  }
+  
+  /**
+   * Frequencies of words. Sorted in the non-ascending order.
+   */
+  public Output<TInt32> vocabFreq() {
+    return vocabFreq;
+  }
+  
+  /**
+   * Number of words per epoch in the data file.
+   */
+  public Output<TInt64> wordsPerEpoch() {
+    return wordsPerEpoch;
+  }
+  
+  /**
+   * The current epoch number.
+   */
+  public Output<TInt32> currentEpoch() {
+    return currentEpoch;
+  }
+  
+  /**
+   * The total number of words processed so far.
+   */
+  public Output<TInt64> totalWordsProcessed() {
+    return totalWordsProcessed;
+  }
+  
+  /**
+   * A vector of word ids.
+   */
+  public Output<TInt32> examples() {
+    return examples;
+  }
+  
+  /**
+   * A vector of word ids.
+   */
+  public Output<TInt32> labels() {
+    return labels;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Skipgram";
+  
+  private Output<TString> vocabWord;
+  private Output<TInt32> vocabFreq;
+  private Output<TInt64> wordsPerEpoch;
+  private Output<TInt32> currentEpoch;
+  private Output<TInt64> totalWordsProcessed;
+  private Output<TInt32> examples;
+  private Output<TInt32> labels;
+  
+  private Skipgram(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    vocabWord = operation.output(outputIdx++);
+    vocabFreq = operation.output(outputIdx++);
+    wordsPerEpoch = operation.output(outputIdx++);
+    currentEpoch = operation.output(outputIdx++);
+    totalWordsProcessed = operation.output(outputIdx++);
+    examples = operation.output(outputIdx++);
+    labels = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Slice.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Slice.java
new file mode 100644
index 00000000000..31edde00b81
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Slice.java
@@ -0,0 +1,90 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Return a slice from 'input'.
+ * <p>
+ * The output tensor is a tensor with dimensions described by 'size'
+ * whose values are extracted from 'input' starting at the offsets in
+ * 'begin'.
+ * <p>
+ * <i>Requirements</i>:
+ *   0 <= begin[i] <= begin[i] + size[i] <= Di  for i in [0, n)
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator
+public final class Slice<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new Slice operation.
+   * 
+   * @param scope current scope
+   * @param input 
+   * @param begin begin[i] specifies the offset into the 'i'th dimension of
+   * 'input' to slice from.
+   * @param size size[i] specifies the number of elements of the 'i'th dimension
+   * of 'input' to slice. If size[i] is -1, all remaining elements in dimension
+   * i are included in the slice (i.e. this is equivalent to setting
+   * size[i] = input.dim_size(i) - begin[i]).
+   * @return a new instance of Slice
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType, U extends TNumber> Slice<T> create(Scope scope, Operand<T> input, Operand<U> begin, Operand<U> size) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Slice", scope.makeOpName("Slice"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(begin.asOutput());
+    opBuilder.addInput(size.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new Slice<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Slice";
+  
+  private Output<T> output;
+  
+  private Slice(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Snapshot.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Snapshot.java
new file mode 100644
index 00000000000..7209bccaaee
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Snapshot.java
@@ -0,0 +1,74 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Returns a copy of the input tensor.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator
+public final class Snapshot<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new Snapshot operation.
+   * 
+   * @param scope current scope
+   * @param input 
+   * @return a new instance of Snapshot
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> Snapshot<T> create(Scope scope, Operand<T> input) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Snapshot", scope.makeOpName("Snapshot"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new Snapshot<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Snapshot";
+  
+  private Output<T> output;
+  
+  private Snapshot(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/SpaceToBatchNd.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/SpaceToBatchNd.java
new file mode 100644
index 00000000000..150e18256e5
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/SpaceToBatchNd.java
@@ -0,0 +1,180 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * SpaceToBatch for N-D tensors of type T.
+ * <p>
+ * This operation divides "spatial" dimensions `[1, ..., M]` of the input into a
+ * grid of blocks of shape `block_shape`, and interleaves these blocks with the
+ * "batch" dimension (0) such that in the output, the spatial dimensions
+ * `[1, ..., M]` correspond to the position within the grid, and the batch
+ * dimension combines both the position within a spatial block and the original
+ * batch position.  Prior to division into blocks, the spatial dimensions of the
+ * input are optionally zero padded according to `paddings`.  See below for a
+ * precise description.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator
+public final class SpaceToBatchNd<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new SpaceToBatchNd operation.
+   * 
+   * @param scope current scope
+   * @param input N-D with shape `input_shape = [batch] + spatial_shape + remaining_shape`,
+   * where spatial_shape has `M` dimensions.
+   * @param blockShape 1-D with shape `[M]`, all values must be >= 1.
+   * @param paddings 2-D with shape `[M, 2]`, all values must be >= 0.
+   *   `paddings[i] = [pad_start, pad_end]` specifies the padding for input dimension
+   *   `i + 1`, which corresponds to spatial dimension `i`.  It is required that
+   *   `block_shape[i]` divides `input_shape[i + 1] + pad_start + pad_end`.
+   * <p>
+   * This operation is equivalent to the following steps:
+   * <p>
+   * 1. Zero-pad the start and end of dimensions `[1, ..., M]` of the
+   *    input according to `paddings` to produce `padded` of shape `padded_shape`.
+   * <p>
+   * 2. Reshape `padded` to `reshaped_padded` of shape:
+   * <p>
+   *      [batch] +
+   *      [padded_shape[1] / block_shape[0],
+   *        block_shape[0],
+   *       ...,
+   *       padded_shape[M] / block_shape[M-1],
+   *       block_shape[M-1]] +
+   *      remaining_shape
+   * <p>
+   * 3. Permute dimensions of `reshaped_padded` to produce
+   *    `permuted_reshaped_padded` of shape:
+   * <p>
+   *      block_shape +
+   *      [batch] +
+   *      [padded_shape[1] / block_shape[0],
+   *       ...,
+   *       padded_shape[M] / block_shape[M-1]] +
+   *      remaining_shape
+   * <p>
+   * 4. Reshape `permuted_reshaped_padded` to flatten `block_shape` into the batch
+   *    dimension, producing an output tensor of shape:
+   * <p>
+   *      [batch * prod(block_shape)] +
+   *      [padded_shape[1] / block_shape[0],
+   *       ...,
+   *       padded_shape[M] / block_shape[M-1]] +
+   *      remaining_shape
+   * <p>
+   * Some examples:
+   * <p>
+   * (1) For the following input of shape `[1, 2, 2, 1]`, `block_shape = [2, 2]`, and
+   *     `paddings = [[0, 0], [0, 0]]`:
+   * <pre>{@code
+   * x = [[[[1], [2]], [[3], [4]]]]
+   * }</pre>
+   * The output tensor has shape `[4, 1, 1, 1]` and value:
+   * <pre>{@code
+   * [[[[1]]], [[[2]]], [[[3]]], [[[4]]]]
+   * }</pre>
+   * (2) For the following input of shape `[1, 2, 2, 3]`, `block_shape = [2, 2]`, and
+   *     `paddings = [[0, 0], [0, 0]]`:
+   * <pre>{@code
+   * x = [[[[1, 2, 3], [4, 5, 6]],
+   *       [[7, 8, 9], [10, 11, 12]]]]
+   * }</pre>
+   * The output tensor has shape `[4, 1, 1, 3]` and value:
+   * <pre>{@code
+   * [[[[1, 2, 3]]], [[[4, 5, 6]]], [[[7, 8, 9]]], [[[10, 11, 12]]]]
+   * }</pre>
+   * (3) For the following input of shape `[1, 4, 4, 1]`, `block_shape = [2, 2]`, and
+   *     `paddings = [[0, 0], [0, 0]]`:
+   * <pre>{@code
+   * x = [[[[1],   [2],  [3],  [4]],
+   *       [[5],   [6],  [7],  [8]],
+   *       [[9],  [10], [11],  [12]],
+   *       [[13], [14], [15],  [16]]]]
+   * }</pre>
+   * The output tensor has shape `[4, 2, 2, 1]` and value:
+   * <pre>{@code
+   * x = [[[[1], [3]], [[9], [11]]],
+   *      [[[2], [4]], [[10], [12]]],
+   *      [[[5], [7]], [[13], [15]]],
+   *      [[[6], [8]], [[14], [16]]]]
+   * }</pre>
+   * (4) For the following input of shape `[2, 2, 4, 1]`, block_shape = `[2, 2]`, and
+   *     paddings = `[[0, 0], [2, 0]]`:
+   * <pre>{@code
+   * x = [[[[1],   [2],  [3],  [4]],
+   *       [[5],   [6],  [7],  [8]]],
+   *      [[[9],  [10], [11],  [12]],
+   *       [[13], [14], [15],  [16]]]]
+   * }</pre>
+   * The output tensor has shape `[8, 1, 3, 1]` and value:
+   * <pre>{@code
+   * x = [[[[0], [1], [3]]], [[[0], [9], [11]]],
+   *      [[[0], [2], [4]]], [[[0], [10], [12]]],
+   *      [[[0], [5], [7]]], [[[0], [13], [15]]],
+   *      [[[0], [6], [8]]], [[[0], [14], [16]]]]
+   * }</pre>
+   * Among others, this operation is useful for reducing atrous convolution into
+   * regular convolution.
+   * @return a new instance of SpaceToBatchNd
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType, U extends TNumber, V extends TNumber> SpaceToBatchNd<T> create(Scope scope, Operand<T> input, Operand<U> blockShape, Operand<V> paddings) {
+    OperationBuilder opBuilder = scope.env().opBuilder("SpaceToBatchND", scope.makeOpName("SpaceToBatchNd"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(blockShape.asOutput());
+    opBuilder.addInput(paddings.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new SpaceToBatchNd<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "SpaceToBatchND";
+  
+  private Output<T> output;
+  
+  private SpaceToBatchNd(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Split.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Split.java
new file mode 100644
index 00000000000..fffd73f8ef5
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Split.java
@@ -0,0 +1,91 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import java.util.Arrays;
+import java.util.Iterator;
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Splits a tensor into `num_split` tensors along one dimension.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator
+public final class Split<T extends TType> extends RawOp implements Iterable<Operand<T>> {
+  
+  /**
+   * Factory method to create a class wrapping a new Split operation.
+   * 
+   * @param scope current scope
+   * @param axis 0-D.  The dimension along which to split.  Must be in the range
+   * `[-rank(value), rank(value))`.
+   * @param value The tensor to split.
+   * @param numSplit The number of ways to split.  Must evenly divide
+   * `value.shape[split_dim]`.
+   * @return a new instance of Split
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> Split<T> create(Scope scope, Operand<TInt32> axis, Operand<T> value, Long numSplit) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Split", scope.makeOpName("Split"));
+    opBuilder.addInput(axis.asOutput());
+    opBuilder.addInput(value.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("num_split", numSplit);
+    return new Split<T>(opBuilder.build());
+  }
+  
+  /**
+   * They are identically shaped tensors, whose shape matches that of `value`
+   * except along `axis`, where their sizes are
+   * `values.shape[split_dim] / num_split`.
+   */
+  public List<Output<T>> output() {
+    return output;
+  }
+  
+  @Override
+  @SuppressWarnings({"rawtypes", "unchecked"})
+  public Iterator<Operand<T>> iterator() {
+    return (Iterator) output.iterator();
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Split";
+  
+  private List<Output<T>> output;
+  
+  @SuppressWarnings("unchecked")
+  private Split(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    int outputLength = operation.outputListLength("output");
+    output = Arrays.asList((Output<T>[])operation.outputList(outputIdx, outputLength));
+    outputIdx += outputLength;
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/SplitV.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/SplitV.java
new file mode 100644
index 00000000000..02b740c4d03
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/SplitV.java
@@ -0,0 +1,95 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import java.util.Arrays;
+import java.util.Iterator;
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Splits a tensor into `num_split` tensors along one dimension.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator
+public final class SplitV<T extends TType> extends RawOp implements Iterable<Operand<T>> {
+  
+  /**
+   * Factory method to create a class wrapping a new SplitV operation.
+   * 
+   * @param scope current scope
+   * @param value The tensor to split.
+   * @param sizeSplits list containing the sizes of each output tensor along the split
+   * dimension. Must sum to the dimension of value along split_dim.
+   * Can contain one -1 indicating that dimension is to be inferred.
+   * @param axis 0-D.  The dimension along which to split.  Must be in the range
+   * `[-rank(value), rank(value))`.
+   * @param numSplit 
+   * @return a new instance of SplitV
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType, U extends TNumber> SplitV<T> create(Scope scope, Operand<T> value, Operand<U> sizeSplits, Operand<TInt32> axis, Long numSplit) {
+    OperationBuilder opBuilder = scope.env().opBuilder("SplitV", scope.makeOpName("SplitV"));
+    opBuilder.addInput(value.asOutput());
+    opBuilder.addInput(sizeSplits.asOutput());
+    opBuilder.addInput(axis.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("num_split", numSplit);
+    return new SplitV<T>(opBuilder.build());
+  }
+  
+  /**
+   * Tensors whose shape matches that of `value`
+   * except along `axis`, where their sizes are
+   * `size_splits[i]`.
+   */
+  public List<Output<T>> output() {
+    return output;
+  }
+  
+  @Override
+  @SuppressWarnings({"rawtypes", "unchecked"})
+  public Iterator<Operand<T>> iterator() {
+    return (Iterator) output.iterator();
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "SplitV";
+  
+  private List<Output<T>> output;
+  
+  @SuppressWarnings("unchecked")
+  private SplitV(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    int outputLength = operation.outputListLength("output");
+    output = Arrays.asList((Output<T>[])operation.outputList(outputIdx, outputLength));
+    outputIdx += outputLength;
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Squeeze.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Squeeze.java
new file mode 100644
index 00000000000..3f962adc789
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Squeeze.java
@@ -0,0 +1,136 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Removes dimensions of size 1 from the shape of a tensor.
+ * <p>
+ * Given a tensor `input`, this operation returns a tensor of the same type with
+ * all dimensions of size 1 removed. If you don't want to remove all size 1
+ * dimensions, you can remove specific size 1 dimensions by specifying
+ * `axis`.
+ * <p>
+ * For example:
+ * <pre>{@code
+ * # 't' is a tensor of shape [1, 2, 1, 3, 1, 1]
+ * shape(squeeze(t)) ==> [2, 3]
+ * }</pre>
+ * Or, to remove specific size 1 dimensions:
+ * <pre>{@code
+ * # 't' is a tensor of shape [1, 2, 1, 3, 1, 1]
+ * shape(squeeze(t, [2, 4])) ==> [1, 2, 3, 1]
+ * }</pre>
+ * 
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator
+public final class Squeeze<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.core.Squeeze}
+   */
+  public static class Options {
+    
+    /**
+     * @param axis If specified, only squeezes the dimensions listed. The dimension
+     * index starts at 0. It is an error to squeeze a dimension that is not 1. Must
+     * be in the range `[-rank(input), rank(input))`.
+     */
+    public Options axis(List<Long> axis) {
+      this.axis = axis;
+      return this;
+    }
+    
+    private List<Long> axis;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new Squeeze operation.
+   * 
+   * @param scope current scope
+   * @param input The `input` to squeeze.
+   * @param options carries optional attributes values
+   * @return a new instance of Squeeze
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> Squeeze<T> create(Scope scope, Operand<T> input, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Squeeze", scope.makeOpName("Squeeze"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.axis != null) {
+          long[] axisArray = new long[opts.axis.size()];
+          for (int i = 0; i < axisArray.length; ++i) {
+            axisArray[i] = opts.axis.get(i);
+          }
+          opBuilder.setAttr("squeeze_dims", axisArray);
+        }
+      }
+    }
+    return new Squeeze<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param axis If specified, only squeezes the dimensions listed. The dimension
+   * index starts at 0. It is an error to squeeze a dimension that is not 1. Must
+   * be in the range `[-rank(input), rank(input))`.
+   */
+  public static Options axis(List<Long> axis) {
+    return new Options().axis(axis);
+  }
+  
+  /**
+   * Contains the same data as `input`, but has one or more dimensions of
+   * size 1 removed.
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Squeeze";
+  
+  private Output<T> output;
+  
+  private Squeeze(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Stack.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Stack.java
new file mode 100644
index 00000000000..f70f50b0d2b
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Stack.java
@@ -0,0 +1,130 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.Operands;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Packs a list of `N` rank-`R` tensors into one rank-`(R+1)` tensor.
+ * <p>
+ * Packs the `N` tensors in `values` into a tensor with rank one higher than each
+ * tensor in `values`, by packing them along the `axis` dimension.
+ * Given a list of tensors of shape `(A, B, C)`;
+ * <p>
+ * if `axis == 0` then the `output` tensor will have the shape `(N, A, B, C)`.
+ * if `axis == 1` then the `output` tensor will have the shape `(A, N, B, C)`.
+ * Etc.
+ * <p>
+ * For example:
+ * <pre>{@code
+ * # 'x' is [1, 4]
+ * # 'y' is [2, 5]
+ * # 'z' is [3, 6]
+ * pack([x, y, z]) => [[1, 4], [2, 5], [3, 6]]  # Pack along first dim.
+ * pack([x, y, z], axis=1) => [[1, 2, 3], [4, 5, 6]]
+ * }</pre>
+ * This is the opposite of `unpack`.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator
+public final class Stack<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.core.Stack}
+   */
+  public static class Options {
+    
+    /**
+     * @param axis Dimension along which to pack.  Negative values wrap around, so the
+     * valid range is `[-(R+1), R+1)`.
+     */
+    public Options axis(Long axis) {
+      this.axis = axis;
+      return this;
+    }
+    
+    private Long axis;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new Stack operation.
+   * 
+   * @param scope current scope
+   * @param values Must be of same shape and type.
+   * @param options carries optional attributes values
+   * @return a new instance of Stack
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> Stack<T> create(Scope scope, Iterable<Operand<T>> values, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Pack", scope.makeOpName("Stack"));
+    opBuilder.addInputList(Operands.asOutputs(values));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.axis != null) {
+          opBuilder.setAttr("axis", opts.axis);
+        }
+      }
+    }
+    return new Stack<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param axis Dimension along which to pack.  Negative values wrap around, so the
+   * valid range is `[-(R+1), R+1)`.
+   */
+  public static Options axis(Long axis) {
+    return new Options().axis(axis);
+  }
+  
+  /**
+   * The packed tensor.
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Pack";
+  
+  private Output<T> output;
+  
+  private Stack(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Stage.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Stage.java
new file mode 100644
index 00000000000..526462b02f4
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Stage.java
@@ -0,0 +1,157 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.Operands;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+
+/**
+ * Stage values similar to a lightweight Enqueue.
+ * <p>
+ * The basic functionality of this Op is similar to a queue with many
+ * fewer capabilities and options.  This Op is optimized for performance.
+ */
+@Operator
+public final class Stage extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.core.Stage}
+   */
+  public static class Options {
+    
+    /**
+     * @param capacity Maximum number of elements in the Staging Area. If > 0, inserts
+     * on the container will block when the capacity is reached.
+     */
+    public Options capacity(Long capacity) {
+      this.capacity = capacity;
+      return this;
+    }
+    
+    /**
+     * @param memoryLimit The maximum number of bytes allowed for Tensors in the Staging Area.
+     * If > 0, inserts will block until sufficient space is available.
+     */
+    public Options memoryLimit(Long memoryLimit) {
+      this.memoryLimit = memoryLimit;
+      return this;
+    }
+    
+    /**
+     * @param container If non-empty, this queue is placed in the given container. Otherwise,
+     * a default container is used.
+     */
+    public Options container(String container) {
+      this.container = container;
+      return this;
+    }
+    
+    /**
+     * @param sharedName It is necessary to match this name to the matching Unstage Op.
+     */
+    public Options sharedName(String sharedName) {
+      this.sharedName = sharedName;
+      return this;
+    }
+    
+    private Long capacity;
+    private Long memoryLimit;
+    private String container;
+    private String sharedName;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new Stage operation.
+   * 
+   * @param scope current scope
+   * @param values a list of tensors
+   * dtypes A list of data types that inserted values should adhere to.
+   * @param options carries optional attributes values
+   * @return a new instance of Stage
+   */
+  @Endpoint(describeByClass = true)
+  public static Stage create(Scope scope, Iterable<Operand<?>> values, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Stage", scope.makeOpName("Stage"));
+    opBuilder.addInputList(Operands.asOutputs(values));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.capacity != null) {
+          opBuilder.setAttr("capacity", opts.capacity);
+        }
+        if (opts.memoryLimit != null) {
+          opBuilder.setAttr("memory_limit", opts.memoryLimit);
+        }
+        if (opts.container != null) {
+          opBuilder.setAttr("container", opts.container);
+        }
+        if (opts.sharedName != null) {
+          opBuilder.setAttr("shared_name", opts.sharedName);
+        }
+      }
+    }
+    return new Stage(opBuilder.build());
+  }
+  
+  /**
+   * @param capacity Maximum number of elements in the Staging Area. If > 0, inserts
+   * on the container will block when the capacity is reached.
+   */
+  public static Options capacity(Long capacity) {
+    return new Options().capacity(capacity);
+  }
+  
+  /**
+   * @param memoryLimit The maximum number of bytes allowed for Tensors in the Staging Area.
+   * If > 0, inserts will block until sufficient space is available.
+   */
+  public static Options memoryLimit(Long memoryLimit) {
+    return new Options().memoryLimit(memoryLimit);
+  }
+  
+  /**
+   * @param container If non-empty, this queue is placed in the given container. Otherwise,
+   * a default container is used.
+   */
+  public static Options container(String container) {
+    return new Options().container(container);
+  }
+  
+  /**
+   * @param sharedName It is necessary to match this name to the matching Unstage Op.
+   */
+  public static Options sharedName(String sharedName) {
+    return new Options().sharedName(sharedName);
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Stage";
+  
+  private Stage(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/StageClear.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/StageClear.java
new file mode 100644
index 00000000000..1bb2fde0fdf
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/StageClear.java
@@ -0,0 +1,151 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import java.util.List;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Op removes all elements in the underlying container.
+ */
+@Operator
+public final class StageClear extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.core.StageClear}
+   */
+  public static class Options {
+    
+    /**
+     * @param capacity 
+     */
+    public Options capacity(Long capacity) {
+      this.capacity = capacity;
+      return this;
+    }
+    
+    /**
+     * @param memoryLimit 
+     */
+    public Options memoryLimit(Long memoryLimit) {
+      this.memoryLimit = memoryLimit;
+      return this;
+    }
+    
+    /**
+     * @param container 
+     */
+    public Options container(String container) {
+      this.container = container;
+      return this;
+    }
+    
+    /**
+     * @param sharedName 
+     */
+    public Options sharedName(String sharedName) {
+      this.sharedName = sharedName;
+      return this;
+    }
+    
+    private Long capacity;
+    private Long memoryLimit;
+    private String container;
+    private String sharedName;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new StageClear operation.
+   * 
+   * @param scope current scope
+   * @param dtypes 
+   * @param options carries optional attributes values
+   * @return a new instance of StageClear
+   */
+  @Endpoint(describeByClass = true)
+  public static StageClear create(Scope scope, List<Class<? extends TType>> dtypes, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("StageClear", scope.makeOpName("StageClear"));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    Class[] dtypesArray = new Class[dtypes.size()];
+    for (int i = 0; i < dtypesArray.length; ++i) {
+      dtypesArray[i] = dtypes.get(i);
+    }
+    opBuilder.setAttr("dtypes", dtypesArray);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.capacity != null) {
+          opBuilder.setAttr("capacity", opts.capacity);
+        }
+        if (opts.memoryLimit != null) {
+          opBuilder.setAttr("memory_limit", opts.memoryLimit);
+        }
+        if (opts.container != null) {
+          opBuilder.setAttr("container", opts.container);
+        }
+        if (opts.sharedName != null) {
+          opBuilder.setAttr("shared_name", opts.sharedName);
+        }
+      }
+    }
+    return new StageClear(opBuilder.build());
+  }
+  
+  /**
+   * @param capacity 
+   */
+  public static Options capacity(Long capacity) {
+    return new Options().capacity(capacity);
+  }
+  
+  /**
+   * @param memoryLimit 
+   */
+  public static Options memoryLimit(Long memoryLimit) {
+    return new Options().memoryLimit(memoryLimit);
+  }
+  
+  /**
+   * @param container 
+   */
+  public static Options container(String container) {
+    return new Options().container(container);
+  }
+  
+  /**
+   * @param sharedName 
+   */
+  public static Options sharedName(String sharedName) {
+    return new Options().sharedName(sharedName);
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "StageClear";
+  
+  private StageClear(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/StagePeek.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/StagePeek.java
new file mode 100644
index 00000000000..07ce640c947
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/StagePeek.java
@@ -0,0 +1,180 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import java.util.Arrays;
+import java.util.Iterator;
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Op peeks at the values at the specified index.  If the
+ * <p>
+ * underlying container does not contain sufficient elements
+ * this op will block until it does.   This Op is optimized for
+ * performance.
+ */
+@Operator
+public final class StagePeek extends RawOp implements Iterable<Operand<TType>> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.core.StagePeek}
+   */
+  public static class Options {
+    
+    /**
+     * @param capacity 
+     */
+    public Options capacity(Long capacity) {
+      this.capacity = capacity;
+      return this;
+    }
+    
+    /**
+     * @param memoryLimit 
+     */
+    public Options memoryLimit(Long memoryLimit) {
+      this.memoryLimit = memoryLimit;
+      return this;
+    }
+    
+    /**
+     * @param container 
+     */
+    public Options container(String container) {
+      this.container = container;
+      return this;
+    }
+    
+    /**
+     * @param sharedName 
+     */
+    public Options sharedName(String sharedName) {
+      this.sharedName = sharedName;
+      return this;
+    }
+    
+    private Long capacity;
+    private Long memoryLimit;
+    private String container;
+    private String sharedName;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new StagePeek operation.
+   * 
+   * @param scope current scope
+   * @param index 
+   * @param dtypes 
+   * @param options carries optional attributes values
+   * @return a new instance of StagePeek
+   */
+  @Endpoint(describeByClass = true)
+  public static StagePeek create(Scope scope, Operand<TInt32> index, List<Class<? extends TType>> dtypes, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("StagePeek", scope.makeOpName("StagePeek"));
+    opBuilder.addInput(index.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    Class[] dtypesArray = new Class[dtypes.size()];
+    for (int i = 0; i < dtypesArray.length; ++i) {
+      dtypesArray[i] = dtypes.get(i);
+    }
+    opBuilder.setAttr("dtypes", dtypesArray);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.capacity != null) {
+          opBuilder.setAttr("capacity", opts.capacity);
+        }
+        if (opts.memoryLimit != null) {
+          opBuilder.setAttr("memory_limit", opts.memoryLimit);
+        }
+        if (opts.container != null) {
+          opBuilder.setAttr("container", opts.container);
+        }
+        if (opts.sharedName != null) {
+          opBuilder.setAttr("shared_name", opts.sharedName);
+        }
+      }
+    }
+    return new StagePeek(opBuilder.build());
+  }
+  
+  /**
+   * @param capacity 
+   */
+  public static Options capacity(Long capacity) {
+    return new Options().capacity(capacity);
+  }
+  
+  /**
+   * @param memoryLimit 
+   */
+  public static Options memoryLimit(Long memoryLimit) {
+    return new Options().memoryLimit(memoryLimit);
+  }
+  
+  /**
+   * @param container 
+   */
+  public static Options container(String container) {
+    return new Options().container(container);
+  }
+  
+  /**
+   * @param sharedName 
+   */
+  public static Options sharedName(String sharedName) {
+    return new Options().sharedName(sharedName);
+  }
+  
+  /**
+   */
+  public List<Output<?>> values() {
+    return values;
+  }
+  
+  @Override
+  @SuppressWarnings({"rawtypes", "unchecked"})
+  public Iterator<Operand<TType>> iterator() {
+    return (Iterator) values.iterator();
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "StagePeek";
+  
+  private List<Output<?>> values;
+  
+  private StagePeek(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    int valuesLength = operation.outputListLength("values");
+    values = Arrays.asList(operation.outputList(outputIdx, valuesLength));
+    outputIdx += valuesLength;
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/StageSize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/StageSize.java
new file mode 100644
index 00000000000..f67f23e5bd3
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/StageSize.java
@@ -0,0 +1,169 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Op returns the number of elements in the underlying container.
+ */
+@Operator
+public final class StageSize extends RawOp implements Operand<TInt32> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.core.StageSize}
+   */
+  public static class Options {
+    
+    /**
+     * @param capacity 
+     */
+    public Options capacity(Long capacity) {
+      this.capacity = capacity;
+      return this;
+    }
+    
+    /**
+     * @param memoryLimit 
+     */
+    public Options memoryLimit(Long memoryLimit) {
+      this.memoryLimit = memoryLimit;
+      return this;
+    }
+    
+    /**
+     * @param container 
+     */
+    public Options container(String container) {
+      this.container = container;
+      return this;
+    }
+    
+    /**
+     * @param sharedName 
+     */
+    public Options sharedName(String sharedName) {
+      this.sharedName = sharedName;
+      return this;
+    }
+    
+    private Long capacity;
+    private Long memoryLimit;
+    private String container;
+    private String sharedName;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new StageSize operation.
+   * 
+   * @param scope current scope
+   * @param dtypes 
+   * @param options carries optional attributes values
+   * @return a new instance of StageSize
+   */
+  @Endpoint(describeByClass = true)
+  public static StageSize create(Scope scope, List<Class<? extends TType>> dtypes, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("StageSize", scope.makeOpName("StageSize"));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    Class[] dtypesArray = new Class[dtypes.size()];
+    for (int i = 0; i < dtypesArray.length; ++i) {
+      dtypesArray[i] = dtypes.get(i);
+    }
+    opBuilder.setAttr("dtypes", dtypesArray);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.capacity != null) {
+          opBuilder.setAttr("capacity", opts.capacity);
+        }
+        if (opts.memoryLimit != null) {
+          opBuilder.setAttr("memory_limit", opts.memoryLimit);
+        }
+        if (opts.container != null) {
+          opBuilder.setAttr("container", opts.container);
+        }
+        if (opts.sharedName != null) {
+          opBuilder.setAttr("shared_name", opts.sharedName);
+        }
+      }
+    }
+    return new StageSize(opBuilder.build());
+  }
+  
+  /**
+   * @param capacity 
+   */
+  public static Options capacity(Long capacity) {
+    return new Options().capacity(capacity);
+  }
+  
+  /**
+   * @param memoryLimit 
+   */
+  public static Options memoryLimit(Long memoryLimit) {
+    return new Options().memoryLimit(memoryLimit);
+  }
+  
+  /**
+   * @param container 
+   */
+  public static Options container(String container) {
+    return new Options().container(container);
+  }
+  
+  /**
+   * @param sharedName 
+   */
+  public static Options sharedName(String sharedName) {
+    return new Options().sharedName(sharedName);
+  }
+  
+  /**
+   */
+  public Output<TInt32> size() {
+    return size;
+  }
+  
+  @Override
+  public Output<TInt32> asOutput() {
+    return size;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "StageSize";
+  
+  private Output<TInt32> size;
+  
+  private StageSize(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    size = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/StopGradient.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/StopGradient.java
new file mode 100644
index 00000000000..b3c7d144ca5
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/StopGradient.java
@@ -0,0 +1,99 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Stops gradient computation.
+ * <p>
+ * When executed in a graph, this op outputs its input tensor as-is.
+ * <p>
+ * When building ops to compute gradients, this op prevents the contribution of
+ * its inputs to be taken into account.  Normally, the gradient generator adds ops
+ * to a graph to compute the derivatives of a specified 'loss' by recursively
+ * finding out inputs that contributed to its computation.  If you insert this op
+ * in the graph it inputs are masked from the gradient generator.  They are not
+ * taken into account for computing gradients.
+ * <p>
+ * This is useful any time you want to compute a value with TensorFlow but need
+ * to pretend that the value was a constant. Some examples include:
+ * <ul>
+ * <li>
+ * The <i>EM</i> algorithm where the <i>M-step</i> should not involve backpropagation
+ *    through the output of the <i>E-step</i>.
+ * </li>
+ * <li>
+ * Contrastive divergence training of Boltzmann machines where, when
+ *    differentiating the energy function, the training must not backpropagate
+ *    through the graph that generated the samples from the model.
+ * </li>
+ * <li>
+ * Adversarial training, where no backprop should happen through the adversarial
+ *    example generation process.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator
+public final class StopGradient<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new StopGradient operation.
+   * 
+   * @param scope current scope
+   * @param input 
+   * @return a new instance of StopGradient
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> StopGradient<T> create(Scope scope, Operand<T> input) {
+    OperationBuilder opBuilder = scope.env().opBuilder("StopGradient", scope.makeOpName("StopGradient"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new StopGradient<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "StopGradient";
+  
+  private Output<T> output;
+  
+  private StopGradient(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/StridedSlice.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/StridedSlice.java
new file mode 100644
index 00000000000..8067f96728d
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/StridedSlice.java
@@ -0,0 +1,317 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Return a strided slice from `input`.
+ * <p>
+ * Note, most python users will want to use the Python `Tensor.__getitem__`
+ * or `Variable.__getitem__` rather than this op directly.
+ * <p>
+ * The goal of this op is to produce a new tensor with a subset of
+ * the elements from the `n` dimensional `input` tensor. The subset is chosen using
+ * a sequence of `m` sparse range specifications encoded into the arguments
+ * of this function. Note, in some cases
+ * `m` could be equal to `n`, but this need not be the case. Each
+ * range specification entry can be one of the following:
+ * <p>
+ * - An ellipsis (...). Ellipses are used to imply zero or more
+ *   dimensions of full-dimension selection and are produced using
+ *   `ellipsis_mask`. For example, `foo[...]` is the identity slice.
+ * <p>
+ * - A new axis. This is used to insert a new shape=1 dimension and is
+ *   produced using `new_axis_mask`. For example, `foo[:, ...]` where
+ *   `foo` is shape `(3, 4)` produces a `(1, 3, 4)` tensor.
+ * <p>
+ * - A range `begin:end:stride`. This is used to specify how much to choose from
+ *   a given dimension. `stride` can be any integer but 0.  `begin` is an integer
+ *   which represents the index of the first value to select while `end` represents
+ *   the index of the last value to select. The number of values selected in each
+ *   dimension is `end - begin` if `stride > 0` and `begin - end` if `stride < 0`.
+ *   `begin` and `end` can be negative where `-1` is the last element, `-2` is
+ *   the second to last. `begin_mask` controls whether to replace the explicitly
+ *   given `begin` with an implicit effective value of `0` if `stride > 0` and
+ *   `-1` if `stride < 0`. `end_mask` is analogous but produces the number
+ *   required to create the largest open interval. For example, given a shape
+ *   `(3,)` tensor `foo[:]`, the effective `begin` and `end` are `0` and `3`. Do
+ *   not assume this is equivalent to `foo[0:-1]` which has an effective `begin`
+ *   and `end` of `0` and `2`. Another example is `foo[-2::-1]` which reverses the
+ *   first dimension of a tensor while dropping the last two (in the original
+ *   order elements). For example `foo = [1,2,3,4]; foo[-2::-1]` is `[4,3]`.
+ * <p>
+ * - A single index. This is used to keep only elements that have a given
+ *   index. For example (`foo[2, :]` on a shape `(5,6)` tensor produces a
+ *   shape `(6,)` tensor. This is encoded in `begin` and `end` and
+ *   `shrink_axis_mask`.
+ * <p>
+ * Each conceptual range specification is encoded in the op's argument. This
+ * encoding is best understand by considering a non-trivial example. In
+ * particular,
+ * `foo[1, 2:4, None, ..., :-3:-1, :]` will be encoded as
+ * <pre>{@code
+ * begin = [1, 2, x, x, 0, x] # x denotes don't care (usually 0)
+ * end = [2, 4, x, x, -3, x]
+ * strides = [1, 1, x, x, -1, 1]
+ * begin_mask = 1<<4 | 1<<5 = 48
+ * end_mask = 1<<5 = 32
+ * ellipsis_mask = 1<<3 = 8
+ * new_axis_mask = 1<<2 = 4
+ * shrink_axis_mask = 1<<0 = 1
+ * }</pre>
+ * In this case if `foo.shape` is (5, 5, 5, 5, 5, 5) the final shape of
+ * the slice becomes (2, 1, 5, 5, 2, 5).
+ * Let us walk step by step through each argument specification.
+ * <p>
+ * 1.  The first argument in the example slice is turned into `begin = 1` and
+ * `end = begin + 1 = 2`. To disambiguate from the original spec `2:4` we
+ * also set the appropriate bit in `shrink_axis_mask`.
+ * <p>
+ * 2. `2:4` is contributes 2, 4, 1 to begin, end, and stride. All masks have
+ * zero bits contributed.
+ * <p>
+ * 3. None is a synonym for `tf.newaxis`. This means insert a dimension of size 1
+ * dimension in the final shape. Dummy values are contributed to begin,
+ * end and stride, while the new_axis_mask bit is set.
+ * <p>
+ * 4. `...` grab the full ranges from as many dimensions as needed to
+ * fully specify a slice for every dimension of the input shape.
+ * <p>
+ * 5. `:-3:-1` shows the use of negative indices. A negative index `i` associated
+ * with a dimension that has shape `s` is converted to a positive index
+ * `s + i`. So `-1` becomes `s-1` (i.e. the last element). This conversion
+ * is done internally so begin, end and strides receive x, -3, and -1.
+ * The appropriate begin_mask bit is set to indicate the start range is the
+ * full range (ignoring the x).
+ * <p>
+ * 6. `:` indicates that the entire contents of the corresponding dimension
+ * is selected. This is equivalent to `::` or `0::1`. begin, end, and strides
+ * receive 0, 0, and 1, respectively. The appropriate bits in `begin_mask` and
+ * `end_mask` are also set.
+ * <p>
+ * <i>Requirements</i>:
+ *   `0 != strides[i] for i in [0, m)`
+ *   `ellipsis_mask must be a power of two (only one ellipsis)`
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator
+public final class StridedSlice<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.core.StridedSlice}
+   */
+  public static class Options {
+    
+    /**
+     * @param beginMask a bitmask where a bit i being 1 means to ignore the begin
+     * value and instead use the largest interval possible. At runtime
+     * begin[i] will be replaced with `[0, n-1)` if `stride[i] > 0` or
+     * `[-1, n-1]` if `stride[i] < 0`
+     */
+    public Options beginMask(Long beginMask) {
+      this.beginMask = beginMask;
+      return this;
+    }
+    
+    /**
+     * @param endMask analogous to `begin_mask`
+     */
+    public Options endMask(Long endMask) {
+      this.endMask = endMask;
+      return this;
+    }
+    
+    /**
+     * @param ellipsisMask a bitmask where bit `i` being 1 means the `i`th
+     * position is actually an ellipsis. One bit at most can be 1.
+     * If `ellipsis_mask == 0`, then an implicit ellipsis mask of `1 << (m+1)`
+     * is provided. This means that `foo[3:5] == foo[3:5, ...]`. An ellipsis
+     * implicitly creates as many range specifications as necessary to fully
+     * specify the sliced range for every dimension. For example for a 4-dimensional
+     * tensor `foo` the slice `foo[2, ..., 5:8]` implies `foo[2, :, :, 5:8]`.
+     */
+    public Options ellipsisMask(Long ellipsisMask) {
+      this.ellipsisMask = ellipsisMask;
+      return this;
+    }
+    
+    /**
+     * @param newAxisMask a bitmask where bit `i` being 1 means the `i`th
+     * specification creates a new shape 1 dimension. For example
+     * `foo[:4, tf.newaxis, :2]` would produce a shape `(4, 1, 2)` tensor.
+     */
+    public Options newAxisMask(Long newAxisMask) {
+      this.newAxisMask = newAxisMask;
+      return this;
+    }
+    
+    /**
+     * @param shrinkAxisMask a bitmask where bit `i` implies that the `i`th
+     * specification should shrink the dimensionality. begin and end
+     * must imply a slice of size 1 in the dimension. For example in
+     * python one might do `foo[:, 3, :]` which would result in
+     * `shrink_axis_mask` being 2.
+     */
+    public Options shrinkAxisMask(Long shrinkAxisMask) {
+      this.shrinkAxisMask = shrinkAxisMask;
+      return this;
+    }
+    
+    private Long beginMask;
+    private Long endMask;
+    private Long ellipsisMask;
+    private Long newAxisMask;
+    private Long shrinkAxisMask;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new StridedSlice operation.
+   * 
+   * @param scope current scope
+   * @param input 
+   * @param begin `begin[k]` specifies the offset into the `k`th range specification.
+   * The exact dimension this corresponds to will be determined by context.
+   * Out-of-bounds values will be silently clamped. If the `k`th bit of
+   * `begin_mask` then `begin[k]` is ignored and the full range of the
+   * appropriate dimension is used instead. Negative values causes indexing
+   * to start from the highest element e.g. If `foo==[1,2,3]` then `foo[-1]==3`.
+   * @param end `end[i]` is like `begin` with the exception that `end_mask` is
+   * used to determine full ranges.
+   * @param strides `strides[i]` specifies the increment in the `i`th specification
+   * after extracting a given element. Negative indices will reverse
+   * the original order. Out or range values are
+   * clamped to `[0,dim[i]) if slice[i]>0` or `[-1,dim[i]-1] if slice[i] < 0`
+   * @param options carries optional attributes values
+   * @return a new instance of StridedSlice
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType, U extends TNumber> StridedSlice<T> create(Scope scope, Operand<T> input, Operand<U> begin, Operand<U> end, Operand<U> strides, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("StridedSlice", scope.makeOpName("StridedSlice"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(begin.asOutput());
+    opBuilder.addInput(end.asOutput());
+    opBuilder.addInput(strides.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.beginMask != null) {
+          opBuilder.setAttr("begin_mask", opts.beginMask);
+        }
+        if (opts.endMask != null) {
+          opBuilder.setAttr("end_mask", opts.endMask);
+        }
+        if (opts.ellipsisMask != null) {
+          opBuilder.setAttr("ellipsis_mask", opts.ellipsisMask);
+        }
+        if (opts.newAxisMask != null) {
+          opBuilder.setAttr("new_axis_mask", opts.newAxisMask);
+        }
+        if (opts.shrinkAxisMask != null) {
+          opBuilder.setAttr("shrink_axis_mask", opts.shrinkAxisMask);
+        }
+      }
+    }
+    return new StridedSlice<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param beginMask a bitmask where a bit i being 1 means to ignore the begin
+   * value and instead use the largest interval possible. At runtime
+   * begin[i] will be replaced with `[0, n-1)` if `stride[i] > 0` or
+   * `[-1, n-1]` if `stride[i] < 0`
+   */
+  public static Options beginMask(Long beginMask) {
+    return new Options().beginMask(beginMask);
+  }
+  
+  /**
+   * @param endMask analogous to `begin_mask`
+   */
+  public static Options endMask(Long endMask) {
+    return new Options().endMask(endMask);
+  }
+  
+  /**
+   * @param ellipsisMask a bitmask where bit `i` being 1 means the `i`th
+   * position is actually an ellipsis. One bit at most can be 1.
+   * If `ellipsis_mask == 0`, then an implicit ellipsis mask of `1 << (m+1)`
+   * is provided. This means that `foo[3:5] == foo[3:5, ...]`. An ellipsis
+   * implicitly creates as many range specifications as necessary to fully
+   * specify the sliced range for every dimension. For example for a 4-dimensional
+   * tensor `foo` the slice `foo[2, ..., 5:8]` implies `foo[2, :, :, 5:8]`.
+   */
+  public static Options ellipsisMask(Long ellipsisMask) {
+    return new Options().ellipsisMask(ellipsisMask);
+  }
+  
+  /**
+   * @param newAxisMask a bitmask where bit `i` being 1 means the `i`th
+   * specification creates a new shape 1 dimension. For example
+   * `foo[:4, tf.newaxis, :2]` would produce a shape `(4, 1, 2)` tensor.
+   */
+  public static Options newAxisMask(Long newAxisMask) {
+    return new Options().newAxisMask(newAxisMask);
+  }
+  
+  /**
+   * @param shrinkAxisMask a bitmask where bit `i` implies that the `i`th
+   * specification should shrink the dimensionality. begin and end
+   * must imply a slice of size 1 in the dimension. For example in
+   * python one might do `foo[:, 3, :]` which would result in
+   * `shrink_axis_mask` being 2.
+   */
+  public static Options shrinkAxisMask(Long shrinkAxisMask) {
+    return new Options().shrinkAxisMask(shrinkAxisMask);
+  }
+  
+  /**
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "StridedSlice";
+  
+  private Output<T> output;
+  
+  private StridedSlice(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/StridedSliceAssign.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/StridedSliceAssign.java
new file mode 100644
index 00000000000..b6836c65fca
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/StridedSliceAssign.java
@@ -0,0 +1,200 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Assign `value` to the sliced l-value reference of `ref`.
+ * <p>
+ * The values of `value` are assigned to the positions in the variable
+ * `ref` that are selected by the slice parameters. The slice parameters
+ * `begin`, `end`, `strides`, etc. work exactly as in `StridedSlice`.
+ * <p>
+ * NOTE this op currently does not support broadcasting and so `value`'s
+ * shape must be exactly the shape produced by the slice of `ref`.
+ * 
+ * @param <T> data type for {@code outputRef()} output
+ */
+@Operator
+public final class StridedSliceAssign<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.core.StridedSliceAssign}
+   */
+  public static class Options {
+    
+    /**
+     * @param beginMask 
+     */
+    public Options beginMask(Long beginMask) {
+      this.beginMask = beginMask;
+      return this;
+    }
+    
+    /**
+     * @param endMask 
+     */
+    public Options endMask(Long endMask) {
+      this.endMask = endMask;
+      return this;
+    }
+    
+    /**
+     * @param ellipsisMask 
+     */
+    public Options ellipsisMask(Long ellipsisMask) {
+      this.ellipsisMask = ellipsisMask;
+      return this;
+    }
+    
+    /**
+     * @param newAxisMask 
+     */
+    public Options newAxisMask(Long newAxisMask) {
+      this.newAxisMask = newAxisMask;
+      return this;
+    }
+    
+    /**
+     * @param shrinkAxisMask 
+     */
+    public Options shrinkAxisMask(Long shrinkAxisMask) {
+      this.shrinkAxisMask = shrinkAxisMask;
+      return this;
+    }
+    
+    private Long beginMask;
+    private Long endMask;
+    private Long ellipsisMask;
+    private Long newAxisMask;
+    private Long shrinkAxisMask;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new StridedSliceAssign operation.
+   * 
+   * @param scope current scope
+   * @param ref 
+   * @param begin 
+   * @param end 
+   * @param strides 
+   * @param value 
+   * @param options carries optional attributes values
+   * @return a new instance of StridedSliceAssign
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType, U extends TNumber> StridedSliceAssign<T> create(Scope scope, Operand<T> ref, Operand<U> begin, Operand<U> end, Operand<U> strides, Operand<T> value, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("StridedSliceAssign", scope.makeOpName("StridedSliceAssign"));
+    opBuilder.addInput(ref.asOutput());
+    opBuilder.addInput(begin.asOutput());
+    opBuilder.addInput(end.asOutput());
+    opBuilder.addInput(strides.asOutput());
+    opBuilder.addInput(value.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.beginMask != null) {
+          opBuilder.setAttr("begin_mask", opts.beginMask);
+        }
+        if (opts.endMask != null) {
+          opBuilder.setAttr("end_mask", opts.endMask);
+        }
+        if (opts.ellipsisMask != null) {
+          opBuilder.setAttr("ellipsis_mask", opts.ellipsisMask);
+        }
+        if (opts.newAxisMask != null) {
+          opBuilder.setAttr("new_axis_mask", opts.newAxisMask);
+        }
+        if (opts.shrinkAxisMask != null) {
+          opBuilder.setAttr("shrink_axis_mask", opts.shrinkAxisMask);
+        }
+      }
+    }
+    return new StridedSliceAssign<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param beginMask 
+   */
+  public static Options beginMask(Long beginMask) {
+    return new Options().beginMask(beginMask);
+  }
+  
+  /**
+   * @param endMask 
+   */
+  public static Options endMask(Long endMask) {
+    return new Options().endMask(endMask);
+  }
+  
+  /**
+   * @param ellipsisMask 
+   */
+  public static Options ellipsisMask(Long ellipsisMask) {
+    return new Options().ellipsisMask(ellipsisMask);
+  }
+  
+  /**
+   * @param newAxisMask 
+   */
+  public static Options newAxisMask(Long newAxisMask) {
+    return new Options().newAxisMask(newAxisMask);
+  }
+  
+  /**
+   * @param shrinkAxisMask 
+   */
+  public static Options shrinkAxisMask(Long shrinkAxisMask) {
+    return new Options().shrinkAxisMask(shrinkAxisMask);
+  }
+  
+  /**
+   */
+  public Output<T> outputRef() {
+    return outputRef;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return outputRef;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "StridedSliceAssign";
+  
+  private Output<T> outputRef;
+  
+  private StridedSliceAssign(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    outputRef = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/StridedSliceGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/StridedSliceGrad.java
new file mode 100644
index 00000000000..241d6d567e9
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/StridedSliceGrad.java
@@ -0,0 +1,202 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Returns the gradient of `StridedSlice`.
+ * <p>
+ * Since `StridedSlice` cuts out pieces of its `input` which is size
+ * `shape`, its gradient will have the same shape (which is passed here
+ * as `shape`). The gradient will be zero in any element that the slice
+ * does not select.
+ * <p>
+ * Arguments are the same as StridedSliceGrad with the exception that
+ * `dy` is the input gradient to be propagated and `shape` is the
+ * shape of `StridedSlice`'s `input`.
+ * 
+ * @param <U> data type for {@code output()} output
+ */
+@Operator
+public final class StridedSliceGrad<U extends TType> extends RawOp implements Operand<U> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.core.StridedSliceGrad}
+   */
+  public static class Options {
+    
+    /**
+     * @param beginMask 
+     */
+    public Options beginMask(Long beginMask) {
+      this.beginMask = beginMask;
+      return this;
+    }
+    
+    /**
+     * @param endMask 
+     */
+    public Options endMask(Long endMask) {
+      this.endMask = endMask;
+      return this;
+    }
+    
+    /**
+     * @param ellipsisMask 
+     */
+    public Options ellipsisMask(Long ellipsisMask) {
+      this.ellipsisMask = ellipsisMask;
+      return this;
+    }
+    
+    /**
+     * @param newAxisMask 
+     */
+    public Options newAxisMask(Long newAxisMask) {
+      this.newAxisMask = newAxisMask;
+      return this;
+    }
+    
+    /**
+     * @param shrinkAxisMask 
+     */
+    public Options shrinkAxisMask(Long shrinkAxisMask) {
+      this.shrinkAxisMask = shrinkAxisMask;
+      return this;
+    }
+    
+    private Long beginMask;
+    private Long endMask;
+    private Long ellipsisMask;
+    private Long newAxisMask;
+    private Long shrinkAxisMask;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new StridedSliceGrad operation.
+   * 
+   * @param scope current scope
+   * @param shape 
+   * @param begin 
+   * @param end 
+   * @param strides 
+   * @param dy 
+   * @param options carries optional attributes values
+   * @return a new instance of StridedSliceGrad
+   */
+  @Endpoint(describeByClass = true)
+  public static <U extends TType, T extends TNumber> StridedSliceGrad<U> create(Scope scope, Operand<T> shape, Operand<T> begin, Operand<T> end, Operand<T> strides, Operand<U> dy, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("StridedSliceGrad", scope.makeOpName("StridedSliceGrad"));
+    opBuilder.addInput(shape.asOutput());
+    opBuilder.addInput(begin.asOutput());
+    opBuilder.addInput(end.asOutput());
+    opBuilder.addInput(strides.asOutput());
+    opBuilder.addInput(dy.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.beginMask != null) {
+          opBuilder.setAttr("begin_mask", opts.beginMask);
+        }
+        if (opts.endMask != null) {
+          opBuilder.setAttr("end_mask", opts.endMask);
+        }
+        if (opts.ellipsisMask != null) {
+          opBuilder.setAttr("ellipsis_mask", opts.ellipsisMask);
+        }
+        if (opts.newAxisMask != null) {
+          opBuilder.setAttr("new_axis_mask", opts.newAxisMask);
+        }
+        if (opts.shrinkAxisMask != null) {
+          opBuilder.setAttr("shrink_axis_mask", opts.shrinkAxisMask);
+        }
+      }
+    }
+    return new StridedSliceGrad<U>(opBuilder.build());
+  }
+  
+  /**
+   * @param beginMask 
+   */
+  public static Options beginMask(Long beginMask) {
+    return new Options().beginMask(beginMask);
+  }
+  
+  /**
+   * @param endMask 
+   */
+  public static Options endMask(Long endMask) {
+    return new Options().endMask(endMask);
+  }
+  
+  /**
+   * @param ellipsisMask 
+   */
+  public static Options ellipsisMask(Long ellipsisMask) {
+    return new Options().ellipsisMask(ellipsisMask);
+  }
+  
+  /**
+   * @param newAxisMask 
+   */
+  public static Options newAxisMask(Long newAxisMask) {
+    return new Options().newAxisMask(newAxisMask);
+  }
+  
+  /**
+   * @param shrinkAxisMask 
+   */
+  public static Options shrinkAxisMask(Long shrinkAxisMask) {
+    return new Options().shrinkAxisMask(shrinkAxisMask);
+  }
+  
+  /**
+   */
+  public Output<U> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<U> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "StridedSliceGrad";
+  
+  private Output<U> output;
+  
+  private StridedSliceGrad(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Sum.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Sum.java
new file mode 100644
index 00000000000..c153cdf4760
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Sum.java
@@ -0,0 +1,118 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Computes the sum of elements across dimensions of a tensor.
+ * <p>
+ * Reduces `input` along the dimensions given in `axis`. Unless
+ * `keep_dims` is true, the rank of the tensor is reduced by 1 for each entry in
+ * `axis`. If `keep_dims` is true, the reduced dimensions are
+ * retained with length 1.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator
+public final class Sum<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.core.Sum}
+   */
+  public static class Options {
+    
+    /**
+     * @param keepDims If true, retain reduced dimensions with length 1.
+     */
+    public Options keepDims(Boolean keepDims) {
+      this.keepDims = keepDims;
+      return this;
+    }
+    
+    private Boolean keepDims;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new Sum operation.
+   * 
+   * @param scope current scope
+   * @param input The tensor to reduce.
+   * @param axis The dimensions to reduce. Must be in the range
+   * `[-rank(input), rank(input))`.
+   * @param options carries optional attributes values
+   * @return a new instance of Sum
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType, U extends TNumber> Sum<T> create(Scope scope, Operand<T> input, Operand<U> axis, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Sum", scope.makeOpName("Sum"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(axis.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.keepDims != null) {
+          opBuilder.setAttr("keep_dims", opts.keepDims);
+        }
+      }
+    }
+    return new Sum<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param keepDims If true, retain reduced dimensions with length 1.
+   */
+  public static Options keepDims(Boolean keepDims) {
+    return new Options().keepDims(keepDims);
+  }
+  
+  /**
+   * The reduced tensor.
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Sum";
+  
+  private Output<T> output;
+  
+  private Sum(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/SwitchCond.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/SwitchCond.java
new file mode 100644
index 00000000000..191eb6292d2
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/SwitchCond.java
@@ -0,0 +1,87 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TBool;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Forwards `data` to the output port determined by `pred`.
+ * <p>
+ * If `pred` is true, the `data` input is forwarded to `output_true`. Otherwise,
+ * the data goes to `output_false`.
+ * <p>
+ * See also `RefSwitch` and `Merge`.
+ * 
+ * @param <T> data type for {@code outputFalse()} output
+ */
+@Operator
+public final class SwitchCond<T extends TType> extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new SwitchCond operation.
+   * 
+   * @param scope current scope
+   * @param data The tensor to be forwarded to the appropriate output.
+   * @param pred A scalar that specifies which output port will receive data.
+   * @return a new instance of SwitchCond
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> SwitchCond<T> create(Scope scope, Operand<T> data, Operand<TBool> pred) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Switch", scope.makeOpName("SwitchCond"));
+    opBuilder.addInput(data.asOutput());
+    opBuilder.addInput(pred.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new SwitchCond<T>(opBuilder.build());
+  }
+  
+  /**
+   * If `pred` is false, data will be forwarded to this output.
+   */
+  public Output<T> outputFalse() {
+    return outputFalse;
+  }
+  
+  /**
+   * If `pred` is true, data will be forwarded to this output.
+   */
+  public Output<T> outputTrue() {
+    return outputTrue;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Switch";
+  
+  private Output<T> outputFalse;
+  private Output<T> outputTrue;
+  
+  private SwitchCond(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    outputFalse = operation.output(outputIdx++);
+    outputTrue = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TemporaryVariable.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TemporaryVariable.java
new file mode 100644
index 00000000000..452d380f859
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TemporaryVariable.java
@@ -0,0 +1,129 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Returns a tensor that may be mutated, but only persists within a single step.
+ * <p>
+ * This is an experimental op for internal use only and it is possible to use this
+ * op in unsafe ways.  DO NOT USE unless you fully understand the risks.
+ * <p>
+ * It is the caller's responsibility to ensure that 'ref' is eventually passed to a
+ * matching 'DestroyTemporaryVariable' op after all other uses have completed.
+ * <p>
+ * Outputs a ref to the tensor state so it may be read or modified.
+ * <p>
+ *   E.g.
+ *       var = state_ops._temporary_variable([1, 2], types.float_)
+ *       var_name = var.op.name
+ *       var = state_ops.assign(var, [[4.0, 5.0]])
+ *       var = state_ops.assign_add(var, [[6.0, 7.0]])
+ *       final = state_ops._destroy_temporary_variable(var, var_name=var_name)
+ * 
+ * @param <T> data type for {@code ref()} output
+ */
+@Operator
+public final class TemporaryVariable<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.core.TemporaryVariable}
+   */
+  public static class Options {
+    
+    /**
+     * @param varName Overrides the name used for the temporary variable resource. Default
+     * value is the name of the 'TemporaryVariable' op (which is guaranteed unique).
+     */
+    public Options varName(String varName) {
+      this.varName = varName;
+      return this;
+    }
+    
+    private String varName;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new TemporaryVariable operation.
+   * 
+   * @param scope current scope
+   * @param shape The shape of the variable tensor.
+   * @param dtype The type of elements in the variable tensor.
+   * @param options carries optional attributes values
+   * @return a new instance of TemporaryVariable
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> TemporaryVariable<T> create(Scope scope, Shape shape, Class<T> dtype, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("TemporaryVariable", scope.makeOpName("TemporaryVariable"));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("shape", shape);
+    opBuilder.setAttr("dtype", dtype);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.varName != null) {
+          opBuilder.setAttr("var_name", opts.varName);
+        }
+      }
+    }
+    return new TemporaryVariable<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param varName Overrides the name used for the temporary variable resource. Default
+   * value is the name of the 'TemporaryVariable' op (which is guaranteed unique).
+   */
+  public static Options varName(String varName) {
+    return new Options().varName(varName);
+  }
+  
+  /**
+   * A reference to the variable tensor.
+   */
+  public Output<T> ref() {
+    return ref;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return ref;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "TemporaryVariable";
+  
+  private Output<T> ref;
+  
+  private TemporaryVariable(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    ref = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArray.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArray.java
new file mode 100644
index 00000000000..fb6014a5efd
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArray.java
@@ -0,0 +1,218 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TType;
+
+/**
+ * An array of Tensors of given size.
+ * <p>
+ * Write data via Write and read via Read or Pack.
+ */
+@Operator
+public final class TensorArray extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.core.TensorArray}
+   */
+  public static class Options {
+    
+    /**
+     * @param elementShape The expected shape of an element, if known. Used to
+     * validate the shapes of TensorArray elements. If this shape is not
+     * fully specified, gathering zero-size TensorArrays is an error.
+     */
+    public Options elementShape(Shape elementShape) {
+      this.elementShape = elementShape;
+      return this;
+    }
+    
+    /**
+     * @param dynamicSize A boolean that determines whether writes to the TensorArray
+     * are allowed to grow the size.  By default, this is not allowed.
+     */
+    public Options dynamicSize(Boolean dynamicSize) {
+      this.dynamicSize = dynamicSize;
+      return this;
+    }
+    
+    /**
+     * @param clearAfterRead If true (default), Tensors in the TensorArray are cleared
+     * after being read.  This disables multiple read semantics but allows early
+     * release of memory.
+     */
+    public Options clearAfterRead(Boolean clearAfterRead) {
+      this.clearAfterRead = clearAfterRead;
+      return this;
+    }
+    
+    /**
+     * @param identicalElementShapes If true (default is false), then all
+     * elements in the TensorArray will be expected to have have identical shapes.
+     * This allows certain behaviors, like dynamically checking for
+     * consistent shapes on write, and being able to fill in properly
+     * shaped zero tensors on stack -- even if the element_shape attribute
+     * is not fully defined.
+     */
+    public Options identicalElementShapes(Boolean identicalElementShapes) {
+      this.identicalElementShapes = identicalElementShapes;
+      return this;
+    }
+    
+    /**
+     * @param tensorArrayName Overrides the name used for the temporary tensor_array
+     * resource. Default value is the name of the 'TensorArray' op (which
+     * is guaranteed unique).
+     */
+    public Options tensorArrayName(String tensorArrayName) {
+      this.tensorArrayName = tensorArrayName;
+      return this;
+    }
+    
+    private Shape elementShape;
+    private Boolean dynamicSize;
+    private Boolean clearAfterRead;
+    private Boolean identicalElementShapes;
+    private String tensorArrayName;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new TensorArray operation.
+   * 
+   * @param scope current scope
+   * @param size The size of the array.
+   * @param dtype The type of the elements on the tensor_array.
+   * @param options carries optional attributes values
+   * @return a new instance of TensorArray
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> TensorArray create(Scope scope, Operand<TInt32> size, Class<T> dtype, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("TensorArrayV3", scope.makeOpName("TensorArray"));
+    opBuilder.addInput(size.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("dtype", dtype);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.elementShape != null) {
+          opBuilder.setAttr("element_shape", opts.elementShape);
+        }
+        if (opts.dynamicSize != null) {
+          opBuilder.setAttr("dynamic_size", opts.dynamicSize);
+        }
+        if (opts.clearAfterRead != null) {
+          opBuilder.setAttr("clear_after_read", opts.clearAfterRead);
+        }
+        if (opts.identicalElementShapes != null) {
+          opBuilder.setAttr("identical_element_shapes", opts.identicalElementShapes);
+        }
+        if (opts.tensorArrayName != null) {
+          opBuilder.setAttr("tensor_array_name", opts.tensorArrayName);
+        }
+      }
+    }
+    return new TensorArray(opBuilder.build());
+  }
+  
+  /**
+   * @param elementShape The expected shape of an element, if known. Used to
+   * validate the shapes of TensorArray elements. If this shape is not
+   * fully specified, gathering zero-size TensorArrays is an error.
+   */
+  public static Options elementShape(Shape elementShape) {
+    return new Options().elementShape(elementShape);
+  }
+  
+  /**
+   * @param dynamicSize A boolean that determines whether writes to the TensorArray
+   * are allowed to grow the size.  By default, this is not allowed.
+   */
+  public static Options dynamicSize(Boolean dynamicSize) {
+    return new Options().dynamicSize(dynamicSize);
+  }
+  
+  /**
+   * @param clearAfterRead If true (default), Tensors in the TensorArray are cleared
+   * after being read.  This disables multiple read semantics but allows early
+   * release of memory.
+   */
+  public static Options clearAfterRead(Boolean clearAfterRead) {
+    return new Options().clearAfterRead(clearAfterRead);
+  }
+  
+  /**
+   * @param identicalElementShapes If true (default is false), then all
+   * elements in the TensorArray will be expected to have have identical shapes.
+   * This allows certain behaviors, like dynamically checking for
+   * consistent shapes on write, and being able to fill in properly
+   * shaped zero tensors on stack -- even if the element_shape attribute
+   * is not fully defined.
+   */
+  public static Options identicalElementShapes(Boolean identicalElementShapes) {
+    return new Options().identicalElementShapes(identicalElementShapes);
+  }
+  
+  /**
+   * @param tensorArrayName Overrides the name used for the temporary tensor_array
+   * resource. Default value is the name of the 'TensorArray' op (which
+   * is guaranteed unique).
+   */
+  public static Options tensorArrayName(String tensorArrayName) {
+    return new Options().tensorArrayName(tensorArrayName);
+  }
+  
+  /**
+   * The handle to the TensorArray.
+   */
+  public Output<?> handle() {
+    return handle;
+  }
+  
+  /**
+   * A scalar used to control gradient flow.
+   */
+  public Output<TFloat32> flow() {
+    return flow;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "TensorArrayV3";
+  
+  private Output<?> handle;
+  private Output<TFloat32> flow;
+  
+  private TensorArray(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    handle = operation.output(outputIdx++);
+    flow = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayClose.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayClose.java
new file mode 100644
index 00000000000..62180e8e5ff
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayClose.java
@@ -0,0 +1,58 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+
+/**
+ * Delete the TensorArray from its resource container.
+ * <p>
+ * This enables the user to close and release the resource in the middle
+ * of a step/run.
+ */
+@Operator
+public final class TensorArrayClose extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new TensorArrayClose operation.
+   * 
+   * @param scope current scope
+   * @param handle The handle to a TensorArray (output of TensorArray or TensorArrayGrad).
+   * @return a new instance of TensorArrayClose
+   */
+  @Endpoint(describeByClass = true)
+  public static TensorArrayClose create(Scope scope, Operand<?> handle) {
+    OperationBuilder opBuilder = scope.env().opBuilder("TensorArrayCloseV3", scope.makeOpName("TensorArrayClose"));
+    opBuilder.addInput(handle.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new TensorArrayClose(opBuilder.build());
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "TensorArrayCloseV3";
+  
+  private TensorArrayClose(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayConcat.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayConcat.java
new file mode 100644
index 00000000000..47d78d77f1d
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayConcat.java
@@ -0,0 +1,140 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Concat the elements from the TensorArray into value `value`.
+ * <p>
+ * Takes `T` elements of shapes
+ * <p>
+ *   <pre>{@code
+ *   (n0 x d0 x d1 x ...), (n1 x d0 x d1 x ...), ..., (n(T-1) x d0 x d1 x ...)
+ *   }</pre>
+ * and concatenates them into a Tensor of shape:
+ * <p>
+ *   <pre>{@code
+ * (n0 + n1 + ... + n(T-1) x d0 x d1 x ...)}</pre>
+ * All elements must have the same shape (excepting the first dimension).
+ * 
+ * @param <T> data type for {@code value()} output
+ */
+@Operator
+public final class TensorArrayConcat<T extends TType> extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.core.TensorArrayConcat}
+   */
+  public static class Options {
+    
+    /**
+     * @param elementShapeExcept0 The expected shape of an element, if known,
+     * excluding the first dimension. Used to validate the shapes of
+     * TensorArray elements. If this shape is not fully specified, concatenating
+     * zero-size TensorArrays is an error.
+     */
+    public Options elementShapeExcept0(Shape elementShapeExcept0) {
+      this.elementShapeExcept0 = elementShapeExcept0;
+      return this;
+    }
+    
+    private Shape elementShapeExcept0;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new TensorArrayConcat operation.
+   * 
+   * @param scope current scope
+   * @param handle The handle to a TensorArray.
+   * @param flowIn A float scalar that enforces proper chaining of operations.
+   * @param dtype The type of the elem that is returned.
+   * @param options carries optional attributes values
+   * @return a new instance of TensorArrayConcat
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> TensorArrayConcat<T> create(Scope scope, Operand<?> handle, Operand<TFloat32> flowIn, Class<T> dtype, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("TensorArrayConcatV3", scope.makeOpName("TensorArrayConcat"));
+    opBuilder.addInput(handle.asOutput());
+    opBuilder.addInput(flowIn.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("dtype", dtype);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.elementShapeExcept0 != null) {
+          opBuilder.setAttr("element_shape_except0", opts.elementShapeExcept0);
+        }
+      }
+    }
+    return new TensorArrayConcat<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param elementShapeExcept0 The expected shape of an element, if known,
+   * excluding the first dimension. Used to validate the shapes of
+   * TensorArray elements. If this shape is not fully specified, concatenating
+   * zero-size TensorArrays is an error.
+   */
+  public static Options elementShapeExcept0(Shape elementShapeExcept0) {
+    return new Options().elementShapeExcept0(elementShapeExcept0);
+  }
+  
+  /**
+   * All of the elements in the TensorArray, concatenated along the first
+   * axis.
+   */
+  public Output<T> value() {
+    return value;
+  }
+  
+  /**
+   * A vector of the row sizes of the original T elements in the
+   * value output.  In the example above, this would be the values:
+   * `(n1, n2, ..., n(T-1))`.
+   */
+  public Output<TInt64> lengths() {
+    return lengths;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "TensorArrayConcatV3";
+  
+  private Output<T> value;
+  private Output<TInt64> lengths;
+  
+  private TensorArrayConcat(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    value = operation.output(outputIdx++);
+    lengths = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayGather.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayGather.java
new file mode 100644
index 00000000000..9ba3bca49f7
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayGather.java
@@ -0,0 +1,125 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Gather specific elements from the TensorArray into output `value`.
+ * <p>
+ * All elements selected by `indices` must have the same shape.
+ * 
+ * @param <T> data type for {@code value()} output
+ */
+@Operator
+public final class TensorArrayGather<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.core.TensorArrayGather}
+   */
+  public static class Options {
+    
+    /**
+     * @param elementShape The expected shape of an element, if known. Used to
+     * validate the shapes of TensorArray elements. If this shape is not
+     * fully specified, gathering zero-size TensorArrays is an error.
+     */
+    public Options elementShape(Shape elementShape) {
+      this.elementShape = elementShape;
+      return this;
+    }
+    
+    private Shape elementShape;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new TensorArrayGather operation.
+   * 
+   * @param scope current scope
+   * @param handle The handle to a TensorArray.
+   * @param indices The locations in the TensorArray from which to read tensor elements.
+   * @param flowIn A float scalar that enforces proper chaining of operations.
+   * @param dtype The type of the elem that is returned.
+   * @param options carries optional attributes values
+   * @return a new instance of TensorArrayGather
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> TensorArrayGather<T> create(Scope scope, Operand<?> handle, Operand<TInt32> indices, Operand<TFloat32> flowIn, Class<T> dtype, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("TensorArrayGatherV3", scope.makeOpName("TensorArrayGather"));
+    opBuilder.addInput(handle.asOutput());
+    opBuilder.addInput(indices.asOutput());
+    opBuilder.addInput(flowIn.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("dtype", dtype);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.elementShape != null) {
+          opBuilder.setAttr("element_shape", opts.elementShape);
+        }
+      }
+    }
+    return new TensorArrayGather<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param elementShape The expected shape of an element, if known. Used to
+   * validate the shapes of TensorArray elements. If this shape is not
+   * fully specified, gathering zero-size TensorArrays is an error.
+   */
+  public static Options elementShape(Shape elementShape) {
+    return new Options().elementShape(elementShape);
+  }
+  
+  /**
+   * All of the elements in the TensorArray, concatenated along a new
+   * axis (the new dimension 0).
+   */
+  public Output<T> value() {
+    return value;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return value;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "TensorArrayGatherV3";
+  
+  private Output<T> value;
+  
+  private TensorArrayGather(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    value = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayGrad.java
new file mode 100644
index 00000000000..dca362c6934
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayGrad.java
@@ -0,0 +1,117 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+
+/**
+ * Creates a TensorArray for storing the gradients of values in the given handle.
+ * <p>
+ * If the given TensorArray gradient already exists, returns a reference to it.
+ * <p>
+ * Locks the size of the original TensorArray by disabling its dynamic size flag.
+ * <p>
+ * **A note about the input flow_in:**
+ * <p>
+ * The handle flow_in forces the execution of the gradient lookup to occur
+ * only after certain other operations have occurred.  For example, when
+ * the forward TensorArray is dynamically sized, writes to this TensorArray
+ * may resize the object.  The gradient TensorArray is statically sized based
+ * on the size of the forward TensorArray when this operation executes.
+ * Furthermore, the size of the forward TensorArray is frozen by this call.
+ * As a result, the flow is used to ensure that the call to generate the gradient
+ * TensorArray only happens after all writes are executed.
+ * <p>
+ * In the case of dynamically sized TensorArrays, gradient computation should
+ * only be performed on read operations that have themselves been chained via
+ * flow to occur only after all writes have executed. That way the final size
+ * of the forward TensorArray is known when this operation is called.
+ * <p>
+ * **A note about the source attribute:**
+ * <p>
+ * TensorArray gradient calls use an accumulator TensorArray object.  If
+ * multiple gradients are calculated and run in the same session, the multiple
+ * gradient nodes may accidentally flow through the same accumulator TensorArray.
+ * This double counts and generally breaks the TensorArray gradient flow.
+ * <p>
+ * The solution is to identify which gradient call this particular
+ * TensorArray gradient is being called in.  This is performed by identifying
+ * a unique string (e.g. "gradients", "gradients_1", ...) from the input
+ * gradient Tensor's name.  This string is used as a suffix when creating
+ * the TensorArray gradient object here (the attribute `source`).
+ * <p>
+ * The attribute `source` is added as a suffix to the forward TensorArray's
+ * name when performing the creation / lookup, so that each separate gradient
+ * calculation gets its own TensorArray accumulator.
+ */
+@Operator
+public final class TensorArrayGrad extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new TensorArrayGrad operation.
+   * 
+   * @param scope current scope
+   * @param handle The handle to the forward TensorArray.
+   * @param flowIn A float scalar that enforces proper chaining of operations.
+   * @param source The gradient source string, used to decide which gradient TensorArray
+   * to return.
+   * @return a new instance of TensorArrayGrad
+   */
+  @Endpoint(describeByClass = true)
+  public static TensorArrayGrad create(Scope scope, Operand<?> handle, Operand<TFloat32> flowIn, String source) {
+    OperationBuilder opBuilder = scope.env().opBuilder("TensorArrayGradV3", scope.makeOpName("TensorArrayGrad"));
+    opBuilder.addInput(handle.asOutput());
+    opBuilder.addInput(flowIn.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("source", source);
+    return new TensorArrayGrad(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<?> gradHandle() {
+    return gradHandle;
+  }
+  
+  /**
+   */
+  public Output<TFloat32> flowOut() {
+    return flowOut;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "TensorArrayGradV3";
+  
+  private Output<?> gradHandle;
+  private Output<TFloat32> flowOut;
+  
+  private TensorArrayGrad(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    gradHandle = operation.output(outputIdx++);
+    flowOut = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayGradWithShape.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayGradWithShape.java
new file mode 100644
index 00000000000..6128775dce6
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayGradWithShape.java
@@ -0,0 +1,90 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.TInt32;
+
+/**
+ * Creates a TensorArray for storing multiple gradients of values in the given handle.
+ * <p>
+ * Similar to TensorArrayGradV3. However it creates an accumulator with an
+ * expanded shape compared to the input TensorArray whose gradient is being
+ * computed. This enables multiple gradients for the same TensorArray to be
+ * calculated using the same accumulator.
+ */
+@Operator
+public final class TensorArrayGradWithShape extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new TensorArrayGradWithShape operation.
+   * 
+   * @param scope current scope
+   * @param handle The handle to the forward TensorArray.
+   * @param flowIn A float scalar that enforces proper chaining of operations.
+   * @param shapeToPrepend An int32 vector representing a shape. Elements in the gradient accumulator will
+   * have shape which is this shape_to_prepend value concatenated with shape of the
+   * elements in the TensorArray corresponding to the input handle.
+   * @param source The gradient source string, used to decide which gradient TensorArray
+   * to return.
+   * @return a new instance of TensorArrayGradWithShape
+   */
+  @Endpoint(describeByClass = true)
+  public static TensorArrayGradWithShape create(Scope scope, Operand<?> handle, Operand<TFloat32> flowIn, Operand<TInt32> shapeToPrepend, String source) {
+    OperationBuilder opBuilder = scope.env().opBuilder("TensorArrayGradWithShape", scope.makeOpName("TensorArrayGradWithShape"));
+    opBuilder.addInput(handle.asOutput());
+    opBuilder.addInput(flowIn.asOutput());
+    opBuilder.addInput(shapeToPrepend.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("source", source);
+    return new TensorArrayGradWithShape(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<?> gradHandle() {
+    return gradHandle;
+  }
+  
+  /**
+   */
+  public Output<TFloat32> flowOut() {
+    return flowOut;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "TensorArrayGradWithShape";
+  
+  private Output<?> gradHandle;
+  private Output<TFloat32> flowOut;
+  
+  private TensorArrayGradWithShape(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    gradHandle = operation.output(outputIdx++);
+    flowOut = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayPack.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayPack.java
new file mode 100644
index 00000000000..d622be4d3e7
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayPack.java
@@ -0,0 +1,113 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.TString;
+import org.tensorflow.types.family.TType;
+
+/**
+ * @param <T> data type for {@code value()} output
+ */
+@Operator
+public final class TensorArrayPack<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.core.TensorArrayPack}
+   */
+  public static class Options {
+    
+    /**
+     * @param elementShape 
+     */
+    public Options elementShape(Shape elementShape) {
+      this.elementShape = elementShape;
+      return this;
+    }
+    
+    private Shape elementShape;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new TensorArrayPack operation.
+   * 
+   * @param scope current scope
+   * @param handle 
+   * @param flowIn 
+   * @param dtype 
+   * @param options carries optional attributes values
+   * @return a new instance of TensorArrayPack
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> TensorArrayPack<T> create(Scope scope, Operand<TString> handle, Operand<TFloat32> flowIn, Class<T> dtype, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("TensorArrayPack", scope.makeOpName("TensorArrayPack"));
+    opBuilder.addInput(handle.asOutput());
+    opBuilder.addInput(flowIn.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("dtype", dtype);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.elementShape != null) {
+          opBuilder.setAttr("element_shape", opts.elementShape);
+        }
+      }
+    }
+    return new TensorArrayPack<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param elementShape 
+   */
+  public static Options elementShape(Shape elementShape) {
+    return new Options().elementShape(elementShape);
+  }
+  
+  /**
+   */
+  public Output<T> value() {
+    return value;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return value;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "TensorArrayPack";
+  
+  private Output<T> value;
+  
+  private TensorArrayPack(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    value = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayRead.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayRead.java
new file mode 100644
index 00000000000..b71aa6c58a8
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayRead.java
@@ -0,0 +1,83 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Read an element from the TensorArray into output `value`.
+ * 
+ * @param <T> data type for {@code value()} output
+ */
+@Operator
+public final class TensorArrayRead<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new TensorArrayRead operation.
+   * 
+   * @param scope current scope
+   * @param handle The handle to a TensorArray.
+   * @param index 
+   * @param flowIn A float scalar that enforces proper chaining of operations.
+   * @param dtype The type of the elem that is returned.
+   * @return a new instance of TensorArrayRead
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> TensorArrayRead<T> create(Scope scope, Operand<?> handle, Operand<TInt32> index, Operand<TFloat32> flowIn, Class<T> dtype) {
+    OperationBuilder opBuilder = scope.env().opBuilder("TensorArrayReadV3", scope.makeOpName("TensorArrayRead"));
+    opBuilder.addInput(handle.asOutput());
+    opBuilder.addInput(index.asOutput());
+    opBuilder.addInput(flowIn.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("dtype", dtype);
+    return new TensorArrayRead<T>(opBuilder.build());
+  }
+  
+  /**
+   * The tensor that is read from the TensorArray.
+   */
+  public Output<T> value() {
+    return value;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return value;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "TensorArrayReadV3";
+  
+  private Output<T> value;
+  
+  private TensorArrayRead(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    value = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayScatter.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayScatter.java
new file mode 100644
index 00000000000..7b3eb238026
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayScatter.java
@@ -0,0 +1,83 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Scatter the data from the input value into specific TensorArray elements.
+ * <p>
+ * `indices` must be a vector, its length must match the first dim of `value`.
+ */
+@Operator
+public final class TensorArrayScatter extends RawOp implements Operand<TFloat32> {
+  
+  /**
+   * Factory method to create a class wrapping a new TensorArrayScatter operation.
+   * 
+   * @param scope current scope
+   * @param handle The handle to a TensorArray.
+   * @param indices The locations at which to write the tensor elements.
+   * @param value The concatenated tensor to write to the TensorArray.
+   * @param flowIn A float scalar that enforces proper chaining of operations.
+   * @return a new instance of TensorArrayScatter
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> TensorArrayScatter create(Scope scope, Operand<?> handle, Operand<TInt32> indices, Operand<T> value, Operand<TFloat32> flowIn) {
+    OperationBuilder opBuilder = scope.env().opBuilder("TensorArrayScatterV3", scope.makeOpName("TensorArrayScatter"));
+    opBuilder.addInput(handle.asOutput());
+    opBuilder.addInput(indices.asOutput());
+    opBuilder.addInput(value.asOutput());
+    opBuilder.addInput(flowIn.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new TensorArrayScatter(opBuilder.build());
+  }
+  
+  /**
+   * A float scalar that enforces proper chaining of operations.
+   */
+  public Output<TFloat32> flowOut() {
+    return flowOut;
+  }
+  
+  @Override
+  public Output<TFloat32> asOutput() {
+    return flowOut;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "TensorArrayScatterV3";
+  
+  private Output<TFloat32> flowOut;
+  
+  private TensorArrayScatter(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    flowOut = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArraySize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArraySize.java
new file mode 100644
index 00000000000..3e7987bc388
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArraySize.java
@@ -0,0 +1,76 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.TInt32;
+
+/**
+ * Get the current size of the TensorArray.
+ */
+@Operator
+public final class TensorArraySize extends RawOp implements Operand<TInt32> {
+  
+  /**
+   * Factory method to create a class wrapping a new TensorArraySize operation.
+   * 
+   * @param scope current scope
+   * @param handle The handle to a TensorArray (output of TensorArray or TensorArrayGrad).
+   * @param flowIn A float scalar that enforces proper chaining of operations.
+   * @return a new instance of TensorArraySize
+   */
+  @Endpoint(describeByClass = true)
+  public static TensorArraySize create(Scope scope, Operand<?> handle, Operand<TFloat32> flowIn) {
+    OperationBuilder opBuilder = scope.env().opBuilder("TensorArraySizeV3", scope.makeOpName("TensorArraySize"));
+    opBuilder.addInput(handle.asOutput());
+    opBuilder.addInput(flowIn.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new TensorArraySize(opBuilder.build());
+  }
+  
+  /**
+   * The current size of the TensorArray.
+   */
+  public Output<TInt32> size() {
+    return size;
+  }
+  
+  @Override
+  public Output<TInt32> asOutput() {
+    return size;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "TensorArraySizeV3";
+  
+  private Output<TInt32> size;
+  
+  private TensorArraySize(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    size = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArraySplit.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArraySplit.java
new file mode 100644
index 00000000000..d6aedc34a4a
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArraySplit.java
@@ -0,0 +1,104 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Split the data from the input value into TensorArray elements.
+ * <p>
+ * Assuming that `lengths` takes on values
+ * <p>
+ *   <pre>{@code
+ * (n0, n1, ..., n(T-1))}</pre>
+ * and that `value` has shape
+ * <p>
+ *   <pre>{@code
+ * (n0 + n1 + ... + n(T-1) x d0 x d1 x ...)}</pre>
+ * ,
+ * <p>
+ * this splits values into a TensorArray with T tensors.
+ * <p>
+ * TensorArray index t will be the subtensor of values with starting position
+ * <p>
+ *   <pre>{@code
+ * (n0 + n1 + ... + n(t-1), 0, 0, ...)}</pre>
+ * and having size
+ * <p>
+ *   <pre>{@code
+ * nt x d0 x d1 x ...}</pre>
+ * 
+ */
+@Operator
+public final class TensorArraySplit extends RawOp implements Operand<TFloat32> {
+  
+  /**
+   * Factory method to create a class wrapping a new TensorArraySplit operation.
+   * 
+   * @param scope current scope
+   * @param handle The handle to a TensorArray.
+   * @param value The concatenated tensor to write to the TensorArray.
+   * @param lengths The vector of lengths, how to split the rows of value into the
+   * TensorArray.
+   * @param flowIn A float scalar that enforces proper chaining of operations.
+   * @return a new instance of TensorArraySplit
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> TensorArraySplit create(Scope scope, Operand<?> handle, Operand<T> value, Operand<TInt64> lengths, Operand<TFloat32> flowIn) {
+    OperationBuilder opBuilder = scope.env().opBuilder("TensorArraySplitV3", scope.makeOpName("TensorArraySplit"));
+    opBuilder.addInput(handle.asOutput());
+    opBuilder.addInput(value.asOutput());
+    opBuilder.addInput(lengths.asOutput());
+    opBuilder.addInput(flowIn.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new TensorArraySplit(opBuilder.build());
+  }
+  
+  /**
+   * A float scalar that enforces proper chaining of operations.
+   */
+  public Output<TFloat32> flowOut() {
+    return flowOut;
+  }
+  
+  @Override
+  public Output<TFloat32> asOutput() {
+    return flowOut;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "TensorArraySplitV3";
+  
+  private Output<TFloat32> flowOut;
+  
+  private TensorArraySplit(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    flowOut = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayUnpack.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayUnpack.java
new file mode 100644
index 00000000000..9d2d92c4b85
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayUnpack.java
@@ -0,0 +1,77 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.TString;
+import org.tensorflow.types.family.TType;
+
+/**
+ */
+@Operator
+public final class TensorArrayUnpack extends RawOp implements Operand<TFloat32> {
+  
+  /**
+   * Factory method to create a class wrapping a new TensorArrayUnpack operation.
+   * 
+   * @param scope current scope
+   * @param handle 
+   * @param value 
+   * @param flowIn 
+   * @return a new instance of TensorArrayUnpack
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> TensorArrayUnpack create(Scope scope, Operand<TString> handle, Operand<T> value, Operand<TFloat32> flowIn) {
+    OperationBuilder opBuilder = scope.env().opBuilder("TensorArrayUnpack", scope.makeOpName("TensorArrayUnpack"));
+    opBuilder.addInput(handle.asOutput());
+    opBuilder.addInput(value.asOutput());
+    opBuilder.addInput(flowIn.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new TensorArrayUnpack(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<TFloat32> flowOut() {
+    return flowOut;
+  }
+  
+  @Override
+  public Output<TFloat32> asOutput() {
+    return flowOut;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "TensorArrayUnpack";
+  
+  private Output<TFloat32> flowOut;
+  
+  private TensorArrayUnpack(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    flowOut = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayWrite.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayWrite.java
new file mode 100644
index 00000000000..a83a07a458b
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorArrayWrite.java
@@ -0,0 +1,81 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Push an element onto the tensor_array.
+ */
+@Operator
+public final class TensorArrayWrite extends RawOp implements Operand<TFloat32> {
+  
+  /**
+   * Factory method to create a class wrapping a new TensorArrayWrite operation.
+   * 
+   * @param scope current scope
+   * @param handle The handle to a TensorArray.
+   * @param index The position to write to inside the TensorArray.
+   * @param value The tensor to write to the TensorArray.
+   * @param flowIn A float scalar that enforces proper chaining of operations.
+   * @return a new instance of TensorArrayWrite
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> TensorArrayWrite create(Scope scope, Operand<?> handle, Operand<TInt32> index, Operand<T> value, Operand<TFloat32> flowIn) {
+    OperationBuilder opBuilder = scope.env().opBuilder("TensorArrayWriteV3", scope.makeOpName("TensorArrayWrite"));
+    opBuilder.addInput(handle.asOutput());
+    opBuilder.addInput(index.asOutput());
+    opBuilder.addInput(value.asOutput());
+    opBuilder.addInput(flowIn.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new TensorArrayWrite(opBuilder.build());
+  }
+  
+  /**
+   * A float scalar that enforces proper chaining of operations.
+   */
+  public Output<TFloat32> flowOut() {
+    return flowOut;
+  }
+  
+  @Override
+  public Output<TFloat32> asOutput() {
+    return flowOut;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "TensorArrayWriteV3";
+  
+  private Output<TFloat32> flowOut;
+  
+  private TensorArrayWrite(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    flowOut = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorForestCreateTreeVariable.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorForestCreateTreeVariable.java
new file mode 100644
index 00000000000..5ca6ffa1cd0
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorForestCreateTreeVariable.java
@@ -0,0 +1,57 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TString;
+
+/**
+ * Creates a tree resource and returns a handle to it.
+ */
+public final class TensorForestCreateTreeVariable extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new TensorForestCreateTreeVariable operation.
+   * 
+   * @param scope current scope
+   * @param treeHandle Handle to the tree resource to be created.
+   * @param treeConfig Serialized proto string of the boosted_trees.Tree.
+   * @return a new instance of TensorForestCreateTreeVariable
+   */
+  @Endpoint(describeByClass = true)
+  public static TensorForestCreateTreeVariable create(Scope scope, Operand<?> treeHandle, Operand<TString> treeConfig) {
+    OperationBuilder opBuilder = scope.env().opBuilder("TensorForestCreateTreeVariable", scope.makeOpName("TensorForestCreateTreeVariable"));
+    opBuilder.addInput(treeHandle.asOutput());
+    opBuilder.addInput(treeConfig.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new TensorForestCreateTreeVariable(opBuilder.build());
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "TensorForestCreateTreeVariable";
+  
+  private TensorForestCreateTreeVariable(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorForestTreeDeserialize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorForestTreeDeserialize.java
new file mode 100644
index 00000000000..a5e1638035e
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorForestTreeDeserialize.java
@@ -0,0 +1,57 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TString;
+
+/**
+ * Deserializes a proto into the tree handle
+ */
+public final class TensorForestTreeDeserialize extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new TensorForestTreeDeserialize operation.
+   * 
+   * @param scope current scope
+   * @param treeHandle Handle to the tree resource to be restored.
+   * @param treeConfig Serialied proto string of the boosted_trees.Tree proto.
+   * @return a new instance of TensorForestTreeDeserialize
+   */
+  @Endpoint(describeByClass = true)
+  public static TensorForestTreeDeserialize create(Scope scope, Operand<?> treeHandle, Operand<TString> treeConfig) {
+    OperationBuilder opBuilder = scope.env().opBuilder("TensorForestTreeDeserialize", scope.makeOpName("TensorForestTreeDeserialize"));
+    opBuilder.addInput(treeHandle.asOutput());
+    opBuilder.addInput(treeConfig.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new TensorForestTreeDeserialize(opBuilder.build());
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "TensorForestTreeDeserialize";
+  
+  private TensorForestTreeDeserialize(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorForestTreeIsInitializedOp.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorForestTreeIsInitializedOp.java
new file mode 100644
index 00000000000..4dfe64cdf19
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorForestTreeIsInitializedOp.java
@@ -0,0 +1,72 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TBool;
+
+/**
+ * Checks whether a tree has been initialized.
+ */
+public final class TensorForestTreeIsInitializedOp extends RawOp implements Operand<TBool> {
+  
+  /**
+   * Factory method to create a class wrapping a new TensorForestTreeIsInitializedOp operation.
+   * 
+   * @param scope current scope
+   * @param treeHandle Handle to the tree.
+   * @return a new instance of TensorForestTreeIsInitializedOp
+   */
+  @Endpoint(describeByClass = true)
+  public static TensorForestTreeIsInitializedOp create(Scope scope, Operand<?> treeHandle) {
+    OperationBuilder opBuilder = scope.env().opBuilder("TensorForestTreeIsInitializedOp", scope.makeOpName("TensorForestTreeIsInitializedOp"));
+    opBuilder.addInput(treeHandle.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new TensorForestTreeIsInitializedOp(opBuilder.build());
+  }
+  
+  /**
+   * Whether the tree is initialized.
+   */
+  public Output<TBool> isInitialized() {
+    return isInitialized;
+  }
+  
+  @Override
+  public Output<TBool> asOutput() {
+    return isInitialized;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "TensorForestTreeIsInitializedOp";
+  
+  private Output<TBool> isInitialized;
+  
+  private TensorForestTreeIsInitializedOp(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    isInitialized = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorForestTreePredict.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorForestTreePredict.java
new file mode 100644
index 00000000000..3962212c5b7
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorForestTreePredict.java
@@ -0,0 +1,76 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+
+/**
+ * Output the logits for the given input data
+ */
+public final class TensorForestTreePredict extends RawOp implements Operand<TFloat32> {
+  
+  /**
+   * Factory method to create a class wrapping a new TensorForestTreePredict operation.
+   * 
+   * @param scope current scope
+   * @param treeHandle Handle to the tree resource.
+   * @param denseFeatures Rank 2 dense features tensor.
+   * @param logitsDimension Scalar, dimension of the logits.
+   * @return a new instance of TensorForestTreePredict
+   */
+  @Endpoint(describeByClass = true)
+  public static TensorForestTreePredict create(Scope scope, Operand<?> treeHandle, Operand<TFloat32> denseFeatures, Long logitsDimension) {
+    OperationBuilder opBuilder = scope.env().opBuilder("TensorForestTreePredict", scope.makeOpName("TensorForestTreePredict"));
+    opBuilder.addInput(treeHandle.asOutput());
+    opBuilder.addInput(denseFeatures.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("logits_dimension", logitsDimension);
+    return new TensorForestTreePredict(opBuilder.build());
+  }
+  
+  /**
+   * The logits predictions from the tree for each instance in the batch.
+   */
+  public Output<TFloat32> logits() {
+    return logits;
+  }
+  
+  @Override
+  public Output<TFloat32> asOutput() {
+    return logits;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "TensorForestTreePredict";
+  
+  private Output<TFloat32> logits;
+  
+  private TensorForestTreePredict(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    logits = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorForestTreeResourceHandleOp.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorForestTreeResourceHandleOp.java
new file mode 100644
index 00000000000..c72860e1ea7
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorForestTreeResourceHandleOp.java
@@ -0,0 +1,123 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Creates a handle to a TensorForestTreeResource
+ */
+public final class TensorForestTreeResourceHandleOp extends RawOp implements Operand<TType> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.core.TensorForestTreeResourceHandleOp}
+   */
+  public static class Options {
+    
+    /**
+     * @param container 
+     */
+    public Options container(String container) {
+      this.container = container;
+      return this;
+    }
+    
+    /**
+     * @param sharedName 
+     */
+    public Options sharedName(String sharedName) {
+      this.sharedName = sharedName;
+      return this;
+    }
+    
+    private String container;
+    private String sharedName;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new TensorForestTreeResourceHandleOp operation.
+   * 
+   * @param scope current scope
+   * @param options carries optional attributes values
+   * @return a new instance of TensorForestTreeResourceHandleOp
+   */
+  @Endpoint(describeByClass = true)
+  public static TensorForestTreeResourceHandleOp create(Scope scope, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("TensorForestTreeResourceHandleOp", scope.makeOpName("TensorForestTreeResourceHandleOp"));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.container != null) {
+          opBuilder.setAttr("container", opts.container);
+        }
+        if (opts.sharedName != null) {
+          opBuilder.setAttr("shared_name", opts.sharedName);
+        }
+      }
+    }
+    return new TensorForestTreeResourceHandleOp(opBuilder.build());
+  }
+  
+  /**
+   * @param container 
+   */
+  public static Options container(String container) {
+    return new Options().container(container);
+  }
+  
+  /**
+   * @param sharedName 
+   */
+  public static Options sharedName(String sharedName) {
+    return new Options().sharedName(sharedName);
+  }
+  
+  /**
+   */
+  public Output<?> resource() {
+    return resource;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) resource;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "TensorForestTreeResourceHandleOp";
+  
+  private Output<?> resource;
+  
+  private TensorForestTreeResourceHandleOp(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    resource = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorForestTreeSerialize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorForestTreeSerialize.java
new file mode 100644
index 00000000000..d46e23a08ab
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorForestTreeSerialize.java
@@ -0,0 +1,72 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TString;
+
+/**
+ * Serializes the tree handle to a proto
+ */
+public final class TensorForestTreeSerialize extends RawOp implements Operand<TString> {
+  
+  /**
+   * Factory method to create a class wrapping a new TensorForestTreeSerialize operation.
+   * 
+   * @param scope current scope
+   * @param treeHandle Handle to the tree resource to be serialized.
+   * @return a new instance of TensorForestTreeSerialize
+   */
+  @Endpoint(describeByClass = true)
+  public static TensorForestTreeSerialize create(Scope scope, Operand<?> treeHandle) {
+    OperationBuilder opBuilder = scope.env().opBuilder("TensorForestTreeSerialize", scope.makeOpName("TensorForestTreeSerialize"));
+    opBuilder.addInput(treeHandle.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new TensorForestTreeSerialize(opBuilder.build());
+  }
+  
+  /**
+   * Serialied proto string of the tree resource.
+   */
+  public Output<TString> treeConfig() {
+    return treeConfig;
+  }
+  
+  @Override
+  public Output<TString> asOutput() {
+    return treeConfig;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "TensorForestTreeSerialize";
+  
+  private Output<TString> treeConfig;
+  
+  private TensorForestTreeSerialize(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    treeConfig = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorForestTreeSize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorForestTreeSize.java
new file mode 100644
index 00000000000..17cf008d470
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorForestTreeSize.java
@@ -0,0 +1,72 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+
+/**
+ * Get the number of nodes in a tree
+ */
+public final class TensorForestTreeSize extends RawOp implements Operand<TInt32> {
+  
+  /**
+   * Factory method to create a class wrapping a new TensorForestTreeSize operation.
+   * 
+   * @param scope current scope
+   * @param treeHandle Handle to the tree resource.
+   * @return a new instance of TensorForestTreeSize
+   */
+  @Endpoint(describeByClass = true)
+  public static TensorForestTreeSize create(Scope scope, Operand<?> treeHandle) {
+    OperationBuilder opBuilder = scope.env().opBuilder("TensorForestTreeSize", scope.makeOpName("TensorForestTreeSize"));
+    opBuilder.addInput(treeHandle.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new TensorForestTreeSize(opBuilder.build());
+  }
+  
+  /**
+   * The size of the tree.
+   */
+  public Output<TInt32> treeSize() {
+    return treeSize;
+  }
+  
+  @Override
+  public Output<TInt32> asOutput() {
+    return treeSize;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "TensorForestTreeSize";
+  
+  private Output<TInt32> treeSize;
+  
+  private TensorForestTreeSize(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    treeSize = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListConcat.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListConcat.java
new file mode 100644
index 00000000000..6f83df9232c
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListConcat.java
@@ -0,0 +1,98 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Concats all tensors in the list along the 0th dimension.
+ * <p>
+ * Requires that all tensors have the same shape except the first dimension.
+ * <p>
+ * input_handle: The input list.
+ * element_shape: The shape of the uninitialized elements in the list. If the first
+ *   dimension is not -1, it is assumed that all list elements have the same
+ *   leading dim.
+ * leading_dims: The list of leading dims of uninitialized list elements. Used if
+ *   the leading dim of input_handle.element_shape or the element_shape input arg
+ *   is not already set.
+ * tensor: The concated result.
+ * lengths: Output tensor containing sizes of the 0th dimension of tensors in the list, used for computing the gradient.
+ * 
+ * 
+ * @param <U> data type for {@code tensor()} output
+ */
+@Operator
+public final class TensorListConcat<U extends TType> extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new TensorListConcat operation.
+   * 
+   * @param scope current scope
+   * @param inputHandle 
+   * @param elementShape 
+   * @param leadingDims 
+   * @param elementDtype 
+   * @return a new instance of TensorListConcat
+   */
+  @Endpoint(describeByClass = true)
+  public static <U extends TType, T extends TNumber> TensorListConcat<U> create(Scope scope, Operand<?> inputHandle, Operand<T> elementShape, Operand<TInt64> leadingDims, Class<U> elementDtype) {
+    OperationBuilder opBuilder = scope.env().opBuilder("TensorListConcatV2", scope.makeOpName("TensorListConcat"));
+    opBuilder.addInput(inputHandle.asOutput());
+    opBuilder.addInput(elementShape.asOutput());
+    opBuilder.addInput(leadingDims.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("element_dtype", elementDtype);
+    return new TensorListConcat<U>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<U> tensor() {
+    return tensor;
+  }
+  
+  /**
+   */
+  public Output<TInt64> lengths() {
+    return lengths;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "TensorListConcatV2";
+  
+  private Output<U> tensor;
+  private Output<TInt64> lengths;
+  
+  private TensorListConcat(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    tensor = operation.output(outputIdx++);
+    lengths = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListConcatLists.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListConcatLists.java
new file mode 100644
index 00000000000..285bc6e734a
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListConcatLists.java
@@ -0,0 +1,76 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ */
+@Operator
+public final class TensorListConcatLists extends RawOp implements Operand<TType> {
+  
+  /**
+   * Factory method to create a class wrapping a new TensorListConcatLists operation.
+   * 
+   * @param scope current scope
+   * @param inputA 
+   * @param inputB 
+   * @param elementDtype 
+   * @return a new instance of TensorListConcatLists
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> TensorListConcatLists create(Scope scope, Operand<?> inputA, Operand<?> inputB, Class<T> elementDtype) {
+    OperationBuilder opBuilder = scope.env().opBuilder("TensorListConcatLists", scope.makeOpName("TensorListConcatLists"));
+    opBuilder.addInput(inputA.asOutput());
+    opBuilder.addInput(inputB.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("element_dtype", elementDtype);
+    return new TensorListConcatLists(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<?> output() {
+    return output;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "TensorListConcatLists";
+  
+  private Output<?> output;
+  
+  private TensorListConcatLists(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListElementShape.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListElementShape.java
new file mode 100644
index 00000000000..17dfddcb958
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListElementShape.java
@@ -0,0 +1,79 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * The shape of the elements of the given list, as a tensor.
+ * <p>
+ *   input_handle: the list
+ *   element_shape: the shape of elements of the list
+ * 
+ * @param <T> data type for {@code elementShape()} output
+ */
+@Operator
+public final class TensorListElementShape<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new TensorListElementShape operation.
+   * 
+   * @param scope current scope
+   * @param inputHandle 
+   * @param shapeType 
+   * @return a new instance of TensorListElementShape
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> TensorListElementShape<T> create(Scope scope, Operand<?> inputHandle, Class<T> shapeType) {
+    OperationBuilder opBuilder = scope.env().opBuilder("TensorListElementShape", scope.makeOpName("TensorListElementShape"));
+    opBuilder.addInput(inputHandle.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("shape_type", shapeType);
+    return new TensorListElementShape<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> elementShape() {
+    return elementShape;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return elementShape;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "TensorListElementShape";
+  
+  private Output<T> elementShape;
+  
+  private TensorListElementShape(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    elementShape = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListFromTensor.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListFromTensor.java
new file mode 100644
index 00000000000..97a45046767
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListFromTensor.java
@@ -0,0 +1,81 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Creates a TensorList which, when stacked, has the value of `tensor`.
+ * <p>
+ * Each tensor in the result list corresponds to one row of the input tensor.
+ * <p>
+ * tensor: The input tensor.
+ * output_handle: The list.
+ */
+@Operator
+public final class TensorListFromTensor extends RawOp implements Operand<TType> {
+  
+  /**
+   * Factory method to create a class wrapping a new TensorListFromTensor operation.
+   * 
+   * @param scope current scope
+   * @param tensor 
+   * @param elementShape 
+   * @return a new instance of TensorListFromTensor
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType, U extends TNumber> TensorListFromTensor create(Scope scope, Operand<T> tensor, Operand<U> elementShape) {
+    OperationBuilder opBuilder = scope.env().opBuilder("TensorListFromTensor", scope.makeOpName("TensorListFromTensor"));
+    opBuilder.addInput(tensor.asOutput());
+    opBuilder.addInput(elementShape.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new TensorListFromTensor(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<?> outputHandle() {
+    return outputHandle;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) outputHandle;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "TensorListFromTensor";
+  
+  private Output<?> outputHandle;
+  
+  private TensorListFromTensor(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    outputHandle = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListGather.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListGather.java
new file mode 100644
index 00000000000..d99e658f936
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListGather.java
@@ -0,0 +1,88 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Creates a Tensor by indexing into the TensorList.
+ * <p>
+ * Each row in the produced Tensor corresponds to the element in the TensorList
+ * specified by the given index (see `tf.gather`).
+ * <p>
+ * input_handle: The input tensor list.
+ * indices: The indices used to index into the list.
+ * values: The tensor.
+ * 
+ * @param <T> data type for {@code values()} output
+ */
+@Operator
+public final class TensorListGather<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new TensorListGather operation.
+   * 
+   * @param scope current scope
+   * @param inputHandle 
+   * @param indices 
+   * @param elementShape 
+   * @param elementDtype 
+   * @return a new instance of TensorListGather
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> TensorListGather<T> create(Scope scope, Operand<?> inputHandle, Operand<TInt32> indices, Operand<TInt32> elementShape, Class<T> elementDtype) {
+    OperationBuilder opBuilder = scope.env().opBuilder("TensorListGather", scope.makeOpName("TensorListGather"));
+    opBuilder.addInput(inputHandle.asOutput());
+    opBuilder.addInput(indices.asOutput());
+    opBuilder.addInput(elementShape.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("element_dtype", elementDtype);
+    return new TensorListGather<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> values() {
+    return values;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return values;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "TensorListGather";
+  
+  private Output<T> values;
+  
+  private TensorListGather(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    values = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListGetItem.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListGetItem.java
new file mode 100644
index 00000000000..ed92723ab9f
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListGetItem.java
@@ -0,0 +1,79 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TType;
+
+/**
+ * @param <T> data type for {@code item()} output
+ */
+@Operator
+public final class TensorListGetItem<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new TensorListGetItem operation.
+   * 
+   * @param scope current scope
+   * @param inputHandle 
+   * @param index 
+   * @param elementShape 
+   * @param elementDtype 
+   * @return a new instance of TensorListGetItem
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> TensorListGetItem<T> create(Scope scope, Operand<?> inputHandle, Operand<TInt32> index, Operand<TInt32> elementShape, Class<T> elementDtype) {
+    OperationBuilder opBuilder = scope.env().opBuilder("TensorListGetItem", scope.makeOpName("TensorListGetItem"));
+    opBuilder.addInput(inputHandle.asOutput());
+    opBuilder.addInput(index.asOutput());
+    opBuilder.addInput(elementShape.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("element_dtype", elementDtype);
+    return new TensorListGetItem<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> item() {
+    return item;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return item;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "TensorListGetItem";
+  
+  private Output<T> item;
+  
+  private TensorListGetItem(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    item = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListLength.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListLength.java
new file mode 100644
index 00000000000..d2dc93a6e06
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListLength.java
@@ -0,0 +1,75 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+
+/**
+ * Returns the number of tensors in the input tensor list.
+ * <p>
+ * input_handle: the input list
+ * length: the number of tensors in the list
+ */
+@Operator
+public final class TensorListLength extends RawOp implements Operand<TInt32> {
+  
+  /**
+   * Factory method to create a class wrapping a new TensorListLength operation.
+   * 
+   * @param scope current scope
+   * @param inputHandle 
+   * @return a new instance of TensorListLength
+   */
+  @Endpoint(describeByClass = true)
+  public static TensorListLength create(Scope scope, Operand<?> inputHandle) {
+    OperationBuilder opBuilder = scope.env().opBuilder("TensorListLength", scope.makeOpName("TensorListLength"));
+    opBuilder.addInput(inputHandle.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new TensorListLength(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<TInt32> length() {
+    return length;
+  }
+  
+  @Override
+  public Output<TInt32> asOutput() {
+    return length;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "TensorListLength";
+  
+  private Output<TInt32> length;
+  
+  private TensorListLength(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    length = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListPopBack.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListPopBack.java
new file mode 100644
index 00000000000..1cb362c913c
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListPopBack.java
@@ -0,0 +1,89 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Returns the last element of the input list as well as a list with all but that element.
+ * <p>
+ * Fails if the list is empty.
+ * <p>
+ * input_handle: the input list
+ * tensor: the withdrawn last element of the list
+ * element_dtype: the type of elements in the list
+ * element_shape: the shape of the output tensor
+ * 
+ * @param <T> data type for {@code tensor()} output
+ */
+@Operator
+public final class TensorListPopBack<T extends TType> extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new TensorListPopBack operation.
+   * 
+   * @param scope current scope
+   * @param inputHandle 
+   * @param elementShape 
+   * @param elementDtype 
+   * @return a new instance of TensorListPopBack
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> TensorListPopBack<T> create(Scope scope, Operand<?> inputHandle, Operand<TInt32> elementShape, Class<T> elementDtype) {
+    OperationBuilder opBuilder = scope.env().opBuilder("TensorListPopBack", scope.makeOpName("TensorListPopBack"));
+    opBuilder.addInput(inputHandle.asOutput());
+    opBuilder.addInput(elementShape.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("element_dtype", elementDtype);
+    return new TensorListPopBack<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<?> outputHandle() {
+    return outputHandle;
+  }
+  
+  /**
+   */
+  public Output<T> tensor() {
+    return tensor;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "TensorListPopBack";
+  
+  private Output<?> outputHandle;
+  private Output<T> tensor;
+  
+  private TensorListPopBack(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    outputHandle = operation.output(outputIdx++);
+    tensor = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListPushBack.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListPushBack.java
new file mode 100644
index 00000000000..3bde4981422
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListPushBack.java
@@ -0,0 +1,81 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Returns a list which has the passed-in `Tensor` as last element and the other elements of the given list in `input_handle`.
+ * <p>
+ * tensor: The tensor to put on the list.
+ * input_handle: The old list.
+ * output_handle: A list with the elements of the old list followed by tensor.
+ * element_dtype: the type of elements in the list.
+ * element_shape: a shape compatible with that of elements in the list.
+ */
+@Operator
+public final class TensorListPushBack extends RawOp implements Operand<TType> {
+  
+  /**
+   * Factory method to create a class wrapping a new TensorListPushBack operation.
+   * 
+   * @param scope current scope
+   * @param inputHandle 
+   * @param tensor 
+   * @return a new instance of TensorListPushBack
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> TensorListPushBack create(Scope scope, Operand<?> inputHandle, Operand<T> tensor) {
+    OperationBuilder opBuilder = scope.env().opBuilder("TensorListPushBack", scope.makeOpName("TensorListPushBack"));
+    opBuilder.addInput(inputHandle.asOutput());
+    opBuilder.addInput(tensor.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new TensorListPushBack(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<?> outputHandle() {
+    return outputHandle;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) outputHandle;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "TensorListPushBack";
+  
+  private Output<?> outputHandle;
+  
+  private TensorListPushBack(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    outputHandle = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListPushBackBatch.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListPushBackBatch.java
new file mode 100644
index 00000000000..de772f74891
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListPushBackBatch.java
@@ -0,0 +1,74 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ */
+@Operator
+public final class TensorListPushBackBatch extends RawOp implements Operand<TType> {
+  
+  /**
+   * Factory method to create a class wrapping a new TensorListPushBackBatch operation.
+   * 
+   * @param scope current scope
+   * @param inputHandles 
+   * @param tensor 
+   * @return a new instance of TensorListPushBackBatch
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> TensorListPushBackBatch create(Scope scope, Operand<?> inputHandles, Operand<T> tensor) {
+    OperationBuilder opBuilder = scope.env().opBuilder("TensorListPushBackBatch", scope.makeOpName("TensorListPushBackBatch"));
+    opBuilder.addInput(inputHandles.asOutput());
+    opBuilder.addInput(tensor.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new TensorListPushBackBatch(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<?> outputHandles() {
+    return outputHandles;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) outputHandles;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "TensorListPushBackBatch";
+  
+  private Output<?> outputHandles;
+  
+  private TensorListPushBackBatch(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    outputHandles = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListReserve.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListReserve.java
new file mode 100644
index 00000000000..e7c6abda2c5
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListReserve.java
@@ -0,0 +1,84 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * List of the given size with empty elements.
+ * <p>
+ * element_shape: the shape of the future elements of the list
+ * num_elements: the number of elements to reserve
+ * handle: the output list
+ * element_dtype: the desired type of elements in the list.
+ */
+@Operator
+public final class TensorListReserve extends RawOp implements Operand<TType> {
+  
+  /**
+   * Factory method to create a class wrapping a new TensorListReserve operation.
+   * 
+   * @param scope current scope
+   * @param elementShape 
+   * @param numElements 
+   * @param elementDtype 
+   * @return a new instance of TensorListReserve
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber, U extends TType> TensorListReserve create(Scope scope, Operand<T> elementShape, Operand<TInt32> numElements, Class<U> elementDtype) {
+    OperationBuilder opBuilder = scope.env().opBuilder("TensorListReserve", scope.makeOpName("TensorListReserve"));
+    opBuilder.addInput(elementShape.asOutput());
+    opBuilder.addInput(numElements.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("element_dtype", elementDtype);
+    return new TensorListReserve(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<?> handle() {
+    return handle;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) handle;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "TensorListReserve";
+  
+  private Output<?> handle;
+  
+  private TensorListReserve(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    handle = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListResize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListResize.java
new file mode 100644
index 00000000000..4b70c31f277
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListResize.java
@@ -0,0 +1,81 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Resizes the list.
+ * <p>
+ * 
+ * input_handle: the input list
+ * size: size of the output list
+ * 
+ */
+@Operator
+public final class TensorListResize extends RawOp implements Operand<TType> {
+  
+  /**
+   * Factory method to create a class wrapping a new TensorListResize operation.
+   * 
+   * @param scope current scope
+   * @param inputHandle 
+   * @param size 
+   * @return a new instance of TensorListResize
+   */
+  @Endpoint(describeByClass = true)
+  public static TensorListResize create(Scope scope, Operand<?> inputHandle, Operand<TInt32> size) {
+    OperationBuilder opBuilder = scope.env().opBuilder("TensorListResize", scope.makeOpName("TensorListResize"));
+    opBuilder.addInput(inputHandle.asOutput());
+    opBuilder.addInput(size.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new TensorListResize(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<?> outputHandle() {
+    return outputHandle;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) outputHandle;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "TensorListResize";
+  
+  private Output<?> outputHandle;
+  
+  private TensorListResize(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    outputHandle = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListScatter.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListScatter.java
new file mode 100644
index 00000000000..700e3e715a7
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListScatter.java
@@ -0,0 +1,93 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Creates a TensorList by indexing into a Tensor.
+ * <p>
+ * Each member of the TensorList corresponds to one row of the input tensor,
+ * specified by the given index (see `tf.gather`).
+ * <p>
+ * tensor: The input tensor.
+ * indices: The indices used to index into the list.
+ * element_shape: The shape of the elements in the list (can be less specified than
+ *   the shape of the tensor).
+ * num_elements: The size of the output list. Must be large enough to accommodate
+ *   the largest index in indices. If -1, the list is just large enough to include
+ *   the largest index in indices.
+ * output_handle: The TensorList.
+ */
+@Operator
+public final class TensorListScatter extends RawOp implements Operand<TType> {
+  
+  /**
+   * Factory method to create a class wrapping a new TensorListScatter operation.
+   * 
+   * @param scope current scope
+   * @param tensor 
+   * @param indices 
+   * @param elementShape 
+   * @param numElements 
+   * @return a new instance of TensorListScatter
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType, U extends TNumber> TensorListScatter create(Scope scope, Operand<T> tensor, Operand<TInt32> indices, Operand<U> elementShape, Operand<TInt32> numElements) {
+    OperationBuilder opBuilder = scope.env().opBuilder("TensorListScatterV2", scope.makeOpName("TensorListScatter"));
+    opBuilder.addInput(tensor.asOutput());
+    opBuilder.addInput(indices.asOutput());
+    opBuilder.addInput(elementShape.asOutput());
+    opBuilder.addInput(numElements.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new TensorListScatter(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<?> outputHandle() {
+    return outputHandle;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) outputHandle;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "TensorListScatterV2";
+  
+  private Output<?> outputHandle;
+  
+  private TensorListScatter(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    outputHandle = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListScatterIntoExistingList.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListScatterIntoExistingList.java
new file mode 100644
index 00000000000..4b91b1bc0b3
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListScatterIntoExistingList.java
@@ -0,0 +1,86 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Scatters tensor at indices in an input list.
+ * <p>
+ * Each member of the TensorList corresponds to one row of the input tensor,
+ * specified by the given index (see `tf.gather`).
+ * <p>
+ * input_handle: The list to scatter into.
+ * tensor: The input tensor.
+ * indices: The indices used to index into the list.
+ * output_handle: The TensorList.
+ */
+@Operator
+public final class TensorListScatterIntoExistingList extends RawOp implements Operand<TType> {
+  
+  /**
+   * Factory method to create a class wrapping a new TensorListScatterIntoExistingList operation.
+   * 
+   * @param scope current scope
+   * @param inputHandle 
+   * @param tensor 
+   * @param indices 
+   * @return a new instance of TensorListScatterIntoExistingList
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> TensorListScatterIntoExistingList create(Scope scope, Operand<?> inputHandle, Operand<T> tensor, Operand<TInt32> indices) {
+    OperationBuilder opBuilder = scope.env().opBuilder("TensorListScatterIntoExistingList", scope.makeOpName("TensorListScatterIntoExistingList"));
+    opBuilder.addInput(inputHandle.asOutput());
+    opBuilder.addInput(tensor.asOutput());
+    opBuilder.addInput(indices.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new TensorListScatterIntoExistingList(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<?> outputHandle() {
+    return outputHandle;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) outputHandle;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "TensorListScatterIntoExistingList";
+  
+  private Output<?> outputHandle;
+  
+  private TensorListScatterIntoExistingList(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    outputHandle = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListSetItem.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListSetItem.java
new file mode 100644
index 00000000000..358397a927e
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListSetItem.java
@@ -0,0 +1,77 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TType;
+
+/**
+ */
+@Operator
+public final class TensorListSetItem extends RawOp implements Operand<TType> {
+  
+  /**
+   * Factory method to create a class wrapping a new TensorListSetItem operation.
+   * 
+   * @param scope current scope
+   * @param inputHandle 
+   * @param index 
+   * @param item 
+   * @return a new instance of TensorListSetItem
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> TensorListSetItem create(Scope scope, Operand<?> inputHandle, Operand<TInt32> index, Operand<T> item) {
+    OperationBuilder opBuilder = scope.env().opBuilder("TensorListSetItem", scope.makeOpName("TensorListSetItem"));
+    opBuilder.addInput(inputHandle.asOutput());
+    opBuilder.addInput(index.asOutput());
+    opBuilder.addInput(item.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new TensorListSetItem(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<?> outputHandle() {
+    return outputHandle;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) outputHandle;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "TensorListSetItem";
+  
+  private Output<?> outputHandle;
+  
+  private TensorListSetItem(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    outputHandle = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListSplit.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListSplit.java
new file mode 100644
index 00000000000..e93635d7713
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListSplit.java
@@ -0,0 +1,87 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Splits a tensor into a list.
+ * <p>
+ * list[i] corresponds to lengths[i] tensors from the input tensor.
+ * The tensor must have rank at least 1 and contain exactly sum(lengths) elements.
+ * <p>
+ * tensor: The input tensor.
+ * element_shape: A shape compatible with that of elements in the tensor.
+ * lengths: Vector of sizes of the 0th dimension of tensors in the list.
+ * output_handle: The list.
+ */
+@Operator
+public final class TensorListSplit extends RawOp implements Operand<TType> {
+  
+  /**
+   * Factory method to create a class wrapping a new TensorListSplit operation.
+   * 
+   * @param scope current scope
+   * @param tensor 
+   * @param elementShape 
+   * @param lengths 
+   * @return a new instance of TensorListSplit
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType, U extends TNumber> TensorListSplit create(Scope scope, Operand<T> tensor, Operand<U> elementShape, Operand<TInt64> lengths) {
+    OperationBuilder opBuilder = scope.env().opBuilder("TensorListSplit", scope.makeOpName("TensorListSplit"));
+    opBuilder.addInput(tensor.asOutput());
+    opBuilder.addInput(elementShape.asOutput());
+    opBuilder.addInput(lengths.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new TensorListSplit(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<?> outputHandle() {
+    return outputHandle;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) outputHandle;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "TensorListSplit";
+  
+  private Output<?> outputHandle;
+  
+  private TensorListSplit(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    outputHandle = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListStack.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListStack.java
new file mode 100644
index 00000000000..dfc2357d18c
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorListStack.java
@@ -0,0 +1,120 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Stacks all tensors in the list.
+ * <p>
+ * Requires that all tensors have the same shape.
+ * <p>
+ * input_handle: the input list
+ * tensor: the gathered result
+ * num_elements: optional. If not -1, the number of elements in the list.
+ * 
+ * 
+ * @param <T> data type for {@code tensor()} output
+ */
+@Operator
+public final class TensorListStack<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.core.TensorListStack}
+   */
+  public static class Options {
+    
+    /**
+     * @param numElements 
+     */
+    public Options numElements(Long numElements) {
+      this.numElements = numElements;
+      return this;
+    }
+    
+    private Long numElements;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new TensorListStack operation.
+   * 
+   * @param scope current scope
+   * @param inputHandle 
+   * @param elementShape 
+   * @param elementDtype 
+   * @param options carries optional attributes values
+   * @return a new instance of TensorListStack
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> TensorListStack<T> create(Scope scope, Operand<?> inputHandle, Operand<TInt32> elementShape, Class<T> elementDtype, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("TensorListStack", scope.makeOpName("TensorListStack"));
+    opBuilder.addInput(inputHandle.asOutput());
+    opBuilder.addInput(elementShape.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("element_dtype", elementDtype);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.numElements != null) {
+          opBuilder.setAttr("num_elements", opts.numElements);
+        }
+      }
+    }
+    return new TensorListStack<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param numElements 
+   */
+  public static Options numElements(Long numElements) {
+    return new Options().numElements(numElements);
+  }
+  
+  /**
+   */
+  public Output<T> tensor() {
+    return tensor;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return tensor;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "TensorListStack";
+  
+  private Output<T> tensor;
+  
+  private TensorListStack(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    tensor = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorScatterNdAdd.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorScatterNdAdd.java
new file mode 100644
index 00000000000..b5d32055bd9
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorScatterNdAdd.java
@@ -0,0 +1,140 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Adds sparse `updates` to an existing tensor according to `indices`.
+ * <p>
+ * This operation creates a new tensor by adding sparse `updates` to the passed
+ * in `tensor`.
+ * This operation is very similar to `tf.scatter_nd_add`, except that the updates
+ * are added onto an existing tensor (as opposed to a variable). If the memory
+ * for the existing tensor cannot be re-used, a copy is made and updated.
+ * <p>
+ * `indices` is an integer tensor containing indices into a new tensor of shape
+ * `tensor.shape`.  The last dimension of `indices` can be at most the rank of
+ * `tensor.shape`:
+ * <p>
+ *     indices.shape[-1] <= tensor.shape.rank
+ * <p>
+ * The last dimension of `indices` corresponds to indices into elements
+ * (if `indices.shape[-1] = tensor.shape.rank`) or slices
+ * (if `indices.shape[-1] < tensor.shape.rank`) along dimension
+ * `indices.shape[-1]` of `tensor.shape`.  `updates` is a tensor with shape
+ * <p>
+ *     indices.shape[:-1] + tensor.shape[indices.shape[-1]:]
+ * <p>
+ * The simplest form of tensor_scatter_add is to add individual elements to a
+ * tensor by index. For example, say we want to add 4 elements in a rank-1
+ * tensor with 8 elements.
+ * <p>
+ * In Python, this scatter add operation would look like this:
+ * <pre>{@code
+ *     indices = tf.constant([[4], [3], [1], [7]])
+ *     updates = tf.constant([9, 10, 11, 12])
+ *     tensor = tf.ones([8], dtype=tf.int32)
+ *     updated = tf.tensor_scatter_nd_add(tensor, indices, updates)
+ *     print(updated)
+ * }</pre>
+ * The resulting tensor would look like this:
+ * <p>
+ *     [1, 12, 1, 11, 10, 1, 1, 13]
+ * <p>
+ * We can also, insert entire slices of a higher rank tensor all at once. For
+ * example, if we wanted to insert two slices in the first dimension of a
+ * rank-3 tensor with two matrices of new values.
+ * <p>
+ * In Python, this scatter add operation would look like this:
+ * <pre>{@code
+ *     indices = tf.constant([[0], [2]])
+ *     updates = tf.constant([[[5, 5, 5, 5], [6, 6, 6, 6],
+ *                             [7, 7, 7, 7], [8, 8, 8, 8]],
+ *                            [[5, 5, 5, 5], [6, 6, 6, 6],
+ *                             [7, 7, 7, 7], [8, 8, 8, 8]]])
+ *     tensor = tf.ones([4, 4, 4],dtype=tf.int32)
+ *     updated = tf.tensor_scatter_nd_add(tensor, indices, updates)
+ *     print(updated)
+ * }</pre>
+ * The resulting tensor would look like this:
+ * <p>
+ *     [[[6, 6, 6, 6], [7, 7, 7, 7], [8, 8, 8, 8], [9, 9, 9, 9]],
+ *      [[1, 1, 1, 1], [1, 1, 1, 1], [1, 1, 1, 1], [1, 1, 1, 1]],
+ *      [[6, 6, 6, 6], [7, 7, 7, 7], [8, 8, 8, 8], [9, 9, 9, 9]],
+ *      [[1, 1, 1, 1], [1, 1, 1, 1], [1, 1, 1, 1], [1, 1, 1, 1]]]
+ * <p>
+ * Note that on CPU, if an out of bound index is found, an error is returned.
+ * On GPU, if an out of bound index is found, the index is ignored.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator
+public final class TensorScatterNdAdd<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new TensorScatterNdAdd operation.
+   * 
+   * @param scope current scope
+   * @param tensor Tensor to copy/update.
+   * @param indices Index tensor.
+   * @param updates Updates to scatter into output.
+   * @return a new instance of TensorScatterNdAdd
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType, U extends TNumber> TensorScatterNdAdd<T> create(Scope scope, Operand<T> tensor, Operand<U> indices, Operand<T> updates) {
+    OperationBuilder opBuilder = scope.env().opBuilder("TensorScatterAdd", scope.makeOpName("TensorScatterNdAdd"));
+    opBuilder.addInput(tensor.asOutput());
+    opBuilder.addInput(indices.asOutput());
+    opBuilder.addInput(updates.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new TensorScatterNdAdd<T>(opBuilder.build());
+  }
+  
+  /**
+   * A new tensor copied from tensor and updates added according to the indices.
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "TensorScatterAdd";
+  
+  private Output<T> output;
+  
+  private TensorScatterNdAdd(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorScatterNdMax.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorScatterNdMax.java
new file mode 100644
index 00000000000..a14b20195af
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorScatterNdMax.java
@@ -0,0 +1,78 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * @param <T> data type for {@code output()} output
+ */
+@Operator
+public final class TensorScatterNdMax<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new TensorScatterNdMax operation.
+   * 
+   * @param scope current scope
+   * @param tensor Tensor to update.
+   * @param indices Index tensor.
+   * @param updates Updates to scatter into output.
+   * @return a new instance of TensorScatterNdMax
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType, U extends TNumber> TensorScatterNdMax<T> create(Scope scope, Operand<T> tensor, Operand<U> indices, Operand<T> updates) {
+    OperationBuilder opBuilder = scope.env().opBuilder("TensorScatterMax", scope.makeOpName("TensorScatterNdMax"));
+    opBuilder.addInput(tensor.asOutput());
+    opBuilder.addInput(indices.asOutput());
+    opBuilder.addInput(updates.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new TensorScatterNdMax<T>(opBuilder.build());
+  }
+  
+  /**
+   * A new tensor copied from tensor whose values are element-wise maximum between tensor and updates according to the indices.
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "TensorScatterMax";
+  
+  private Output<T> output;
+  
+  private TensorScatterNdMax(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorScatterNdMin.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorScatterNdMin.java
new file mode 100644
index 00000000000..b202b72eebd
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorScatterNdMin.java
@@ -0,0 +1,78 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * @param <T> data type for {@code output()} output
+ */
+@Operator
+public final class TensorScatterNdMin<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new TensorScatterNdMin operation.
+   * 
+   * @param scope current scope
+   * @param tensor Tensor to update.
+   * @param indices Index tensor.
+   * @param updates Updates to scatter into output.
+   * @return a new instance of TensorScatterNdMin
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType, U extends TNumber> TensorScatterNdMin<T> create(Scope scope, Operand<T> tensor, Operand<U> indices, Operand<T> updates) {
+    OperationBuilder opBuilder = scope.env().opBuilder("TensorScatterMin", scope.makeOpName("TensorScatterNdMin"));
+    opBuilder.addInput(tensor.asOutput());
+    opBuilder.addInput(indices.asOutput());
+    opBuilder.addInput(updates.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new TensorScatterNdMin<T>(opBuilder.build());
+  }
+  
+  /**
+   * A new tensor copied from tensor whose values are element-wise minimum between tensor and updates according to the indices.
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "TensorScatterMin";
+  
+  private Output<T> output;
+  
+  private TensorScatterNdMin(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorScatterNdSub.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorScatterNdSub.java
new file mode 100644
index 00000000000..14737a1fd95
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorScatterNdSub.java
@@ -0,0 +1,139 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Subtracts sparse `updates` from an existing tensor according to `indices`.
+ * <p>
+ * This operation creates a new tensor by subtracting sparse `updates` from the
+ * passed in `tensor`.
+ * This operation is very similar to `tf.scatter_nd_sub`, except that the updates
+ * are subtracted from an existing tensor (as opposed to a variable). If the memory
+ * for the existing tensor cannot be re-used, a copy is made and updated.
+ * <p>
+ * `indices` is an integer tensor containing indices into a new tensor of shape
+ * `shape`.  The last dimension of `indices` can be at most the rank of `shape`:
+ * <p>
+ *     indices.shape[-1] <= shape.rank
+ * <p>
+ * The last dimension of `indices` corresponds to indices into elements
+ * (if `indices.shape[-1] = shape.rank`) or slices
+ * (if `indices.shape[-1] < shape.rank`) along dimension `indices.shape[-1]` of
+ * `shape`.  `updates` is a tensor with shape
+ * <p>
+ *     indices.shape[:-1] + shape[indices.shape[-1]:]
+ * <p>
+ * The simplest form of tensor_scatter_sub is to subtract individual elements
+ * from a tensor by index. For example, say we want to insert 4 scattered elements
+ * in a rank-1 tensor with 8 elements.
+ * <p>
+ * In Python, this scatter subtract operation would look like this:
+ * <pre>{@code
+ *     indices = tf.constant([[4], [3], [1], [7]])
+ *     updates = tf.constant([9, 10, 11, 12])
+ *     tensor = tf.ones([8], dtype=tf.int32)
+ *     updated = tf.tensor_scatter_nd_sub(tensor, indices, updates)
+ *     print(updated)
+ * }</pre>
+ * The resulting tensor would look like this:
+ * <p>
+ *     [1, -10, 1, -9, -8, 1, 1, -11]
+ * <p>
+ * We can also, insert entire slices of a higher rank tensor all at once. For
+ * example, if we wanted to insert two slices in the first dimension of a
+ * rank-3 tensor with two matrices of new values.
+ * <p>
+ * In Python, this scatter add operation would look like this:
+ * <pre>{@code
+ *     indices = tf.constant([[0], [2]])
+ *     updates = tf.constant([[[5, 5, 5, 5], [6, 6, 6, 6],
+ *                             [7, 7, 7, 7], [8, 8, 8, 8]],
+ *                            [[5, 5, 5, 5], [6, 6, 6, 6],
+ *                             [7, 7, 7, 7], [8, 8, 8, 8]]])
+ *     tensor = tf.ones([4, 4, 4],dtype=tf.int32)
+ *     updated = tf.tensor_scatter_nd_sub(tensor, indices, updates)
+ *     print(updated)
+ * }</pre>
+ * The resulting tensor would look like this:
+ * <p>
+ *     [[[-4, -4, -4, -4], [-5, -5, -5, -5], [-6, -6, -6, -6], [-7, -7, -7, -7]],
+ *      [[1, 1, 1, 1], [1, 1, 1, 1], [1, 1, 1, 1], [1, 1, 1, 1]],
+ *      [[-4, -4, -4, -4], [-5, -5, -5, -5], [-6, -6, -6, -6], [-7, -7, -7, -7]],
+ *      [[1, 1, 1, 1], [1, 1, 1, 1], [1, 1, 1, 1], [1, 1, 1, 1]]]
+ * <p>
+ * Note that on CPU, if an out of bound index is found, an error is returned.
+ * On GPU, if an out of bound index is found, the index is ignored.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator
+public final class TensorScatterNdSub<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new TensorScatterNdSub operation.
+   * 
+   * @param scope current scope
+   * @param tensor Tensor to copy/update.
+   * @param indices Index tensor.
+   * @param updates Updates to scatter into output.
+   * @return a new instance of TensorScatterNdSub
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType, U extends TNumber> TensorScatterNdSub<T> create(Scope scope, Operand<T> tensor, Operand<U> indices, Operand<T> updates) {
+    OperationBuilder opBuilder = scope.env().opBuilder("TensorScatterSub", scope.makeOpName("TensorScatterNdSub"));
+    opBuilder.addInput(tensor.asOutput());
+    opBuilder.addInput(indices.asOutput());
+    opBuilder.addInput(updates.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new TensorScatterNdSub<T>(opBuilder.build());
+  }
+  
+  /**
+   * A new tensor copied from tensor and updates subtracted according to the indices.
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "TensorScatterSub";
+  
+  private Output<T> output;
+  
+  private TensorScatterNdSub(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorScatterNdUpdate.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorScatterNdUpdate.java
new file mode 100644
index 00000000000..7dff39f733f
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorScatterNdUpdate.java
@@ -0,0 +1,155 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Scatter `updates` into an existing tensor according to `indices`.
+ * <p>
+ * This operation creates a new tensor by applying sparse `updates` to the passed
+ * in `tensor`.
+ * This operation is very similar to `tf.scatter_nd`, except that the updates are
+ * scattered onto an existing tensor (as opposed to a zero-tensor). If the memory
+ * for the existing tensor cannot be re-used, a copy is made and updated.
+ * <p>
+ * If `indices` contains duplicates, then their updates are accumulated (summed).
+ * <p>
+ * <b>WARNING</b>: The order in which updates are applied is nondeterministic, so the
+ * output will be nondeterministic if `indices` contains duplicates -- because
+ * of some numerical approximation issues, numbers summed in different order
+ * may yield different results.
+ * <p>
+ * `indices` is an integer tensor containing indices into a new tensor of shape
+ * `shape`.  The last dimension of `indices` can be at most the rank of `shape`:
+ * <p>
+ *     indices.shape[-1] <= shape.rank
+ * <p>
+ * The last dimension of `indices` corresponds to indices into elements
+ * (if `indices.shape[-1] = shape.rank`) or slices
+ * (if `indices.shape[-1] < shape.rank`) along dimension `indices.shape[-1]` of
+ * `shape`.  `updates` is a tensor with shape
+ * <p>
+ *     indices.shape[:-1] + shape[indices.shape[-1]:]
+ * <p>
+ * The simplest form of scatter is to insert individual elements in a tensor by
+ * index. For example, say we want to insert 4 scattered elements in a rank-1
+ * tensor with 8 elements.
+ * <p>
+ * <div style="width:70%; margin:auto; margin-bottom:10px; margin-top:20px;">
+ * <img style="width:100%" src="https://www.tensorflow.org/images/ScatterNd1.png" alt>
+ * </div>
+ * <p>
+ * In Python, this scatter operation would look like this:
+ * <p>
+ *     >>> indices = tf.constant([[4], [3], [1], [7]])
+ *     >>> updates = tf.constant([9, 10, 11, 12])
+ *     >>> tensor = tf.ones([8], dtype=tf.int32)
+ *     >>> print(tf.tensor_scatter_nd_update(tensor, indices, updates))
+ *     tf.Tensor([ 1 11  1 10  9  1  1 12], shape=(8,), dtype=int32)
+ * <p>
+ * We can also, insert entire slices of a higher rank tensor all at once. For
+ * example, if we wanted to insert two slices in the first dimension of a
+ * rank-3 tensor with two matrices of new values.
+ * <p>
+ * In Python, this scatter operation would look like this:
+ * <p>
+ *     >>> indices = tf.constant([[0], [2]])
+ *     >>> updates = tf.constant([[[5, 5, 5, 5], [6, 6, 6, 6],
+ *     ...                         [7, 7, 7, 7], [8, 8, 8, 8]],
+ *     ...                        [[5, 5, 5, 5], [6, 6, 6, 6],
+ *     ...                         [7, 7, 7, 7], [8, 8, 8, 8]]])
+ *     >>> tensor = tf.ones([4, 4, 4], dtype=tf.int32)
+ *     >>> print(tf.tensor_scatter_nd_update(tensor, indices, updates).numpy())
+ *     [[[5 5 5 5]
+ *       [6 6 6 6]
+ *       [7 7 7 7]
+ *       [8 8 8 8]]
+ *      [[1 1 1 1]
+ *       [1 1 1 1]
+ *       [1 1 1 1]
+ *       [1 1 1 1]]
+ *      [[5 5 5 5]
+ *       [6 6 6 6]
+ *       [7 7 7 7]
+ *       [8 8 8 8]]
+ *      [[1 1 1 1]
+ *       [1 1 1 1]
+ *       [1 1 1 1]
+ *       [1 1 1 1]]]
+ * <p>
+ * Note that on CPU, if an out of bound index is found, an error is returned.
+ * On GPU, if an out of bound index is found, the index is ignored.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator
+public final class TensorScatterNdUpdate<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new TensorScatterNdUpdate operation.
+   * 
+   * @param scope current scope
+   * @param tensor Tensor to copy/update.
+   * @param indices Index tensor.
+   * @param updates Updates to scatter into output.
+   * @return a new instance of TensorScatterNdUpdate
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType, U extends TNumber> TensorScatterNdUpdate<T> create(Scope scope, Operand<T> tensor, Operand<U> indices, Operand<T> updates) {
+    OperationBuilder opBuilder = scope.env().opBuilder("TensorScatterUpdate", scope.makeOpName("TensorScatterNdUpdate"));
+    opBuilder.addInput(tensor.asOutput());
+    opBuilder.addInput(indices.asOutput());
+    opBuilder.addInput(updates.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new TensorScatterNdUpdate<T>(opBuilder.build());
+  }
+  
+  /**
+   * A new tensor with the given shape and updates applied according
+   * to the indices.
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "TensorScatterUpdate";
+  
+  private Output<T> output;
+  
+  private TensorScatterNdUpdate(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorStridedSliceUpdate.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorStridedSliceUpdate.java
new file mode 100644
index 00000000000..c8c5e36e3c6
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TensorStridedSliceUpdate.java
@@ -0,0 +1,200 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Assign `value` to the sliced l-value reference of `input`.
+ * <p>
+ * The values of `value` are assigned to the positions in the tensor `input` that
+ * are selected by the slice parameters. The slice parameters `begin` `end`
+ * `strides` etc. work exactly as in `StridedSlice`.
+ * <p>
+ * NOTE this op currently does not support broadcasting and so `value`'s shape
+ * must be exactly the shape produced by the slice of `input`.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator
+public final class TensorStridedSliceUpdate<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.core.TensorStridedSliceUpdate}
+   */
+  public static class Options {
+    
+    /**
+     * @param beginMask 
+     */
+    public Options beginMask(Long beginMask) {
+      this.beginMask = beginMask;
+      return this;
+    }
+    
+    /**
+     * @param endMask 
+     */
+    public Options endMask(Long endMask) {
+      this.endMask = endMask;
+      return this;
+    }
+    
+    /**
+     * @param ellipsisMask 
+     */
+    public Options ellipsisMask(Long ellipsisMask) {
+      this.ellipsisMask = ellipsisMask;
+      return this;
+    }
+    
+    /**
+     * @param newAxisMask 
+     */
+    public Options newAxisMask(Long newAxisMask) {
+      this.newAxisMask = newAxisMask;
+      return this;
+    }
+    
+    /**
+     * @param shrinkAxisMask 
+     */
+    public Options shrinkAxisMask(Long shrinkAxisMask) {
+      this.shrinkAxisMask = shrinkAxisMask;
+      return this;
+    }
+    
+    private Long beginMask;
+    private Long endMask;
+    private Long ellipsisMask;
+    private Long newAxisMask;
+    private Long shrinkAxisMask;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new TensorStridedSliceUpdate operation.
+   * 
+   * @param scope current scope
+   * @param input 
+   * @param begin 
+   * @param end 
+   * @param strides 
+   * @param value 
+   * @param options carries optional attributes values
+   * @return a new instance of TensorStridedSliceUpdate
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType, U extends TNumber> TensorStridedSliceUpdate<T> create(Scope scope, Operand<T> input, Operand<U> begin, Operand<U> end, Operand<U> strides, Operand<T> value, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("TensorStridedSliceUpdate", scope.makeOpName("TensorStridedSliceUpdate"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(begin.asOutput());
+    opBuilder.addInput(end.asOutput());
+    opBuilder.addInput(strides.asOutput());
+    opBuilder.addInput(value.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.beginMask != null) {
+          opBuilder.setAttr("begin_mask", opts.beginMask);
+        }
+        if (opts.endMask != null) {
+          opBuilder.setAttr("end_mask", opts.endMask);
+        }
+        if (opts.ellipsisMask != null) {
+          opBuilder.setAttr("ellipsis_mask", opts.ellipsisMask);
+        }
+        if (opts.newAxisMask != null) {
+          opBuilder.setAttr("new_axis_mask", opts.newAxisMask);
+        }
+        if (opts.shrinkAxisMask != null) {
+          opBuilder.setAttr("shrink_axis_mask", opts.shrinkAxisMask);
+        }
+      }
+    }
+    return new TensorStridedSliceUpdate<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param beginMask 
+   */
+  public static Options beginMask(Long beginMask) {
+    return new Options().beginMask(beginMask);
+  }
+  
+  /**
+   * @param endMask 
+   */
+  public static Options endMask(Long endMask) {
+    return new Options().endMask(endMask);
+  }
+  
+  /**
+   * @param ellipsisMask 
+   */
+  public static Options ellipsisMask(Long ellipsisMask) {
+    return new Options().ellipsisMask(ellipsisMask);
+  }
+  
+  /**
+   * @param newAxisMask 
+   */
+  public static Options newAxisMask(Long newAxisMask) {
+    return new Options().newAxisMask(newAxisMask);
+  }
+  
+  /**
+   * @param shrinkAxisMask 
+   */
+  public static Options shrinkAxisMask(Long shrinkAxisMask) {
+    return new Options().shrinkAxisMask(shrinkAxisMask);
+  }
+  
+  /**
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "TensorStridedSliceUpdate";
+  
+  private Output<T> output;
+  
+  private TensorStridedSliceUpdate(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Tile.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Tile.java
new file mode 100644
index 00000000000..d28588cada5
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Tile.java
@@ -0,0 +1,104 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Constructs a tensor by tiling a given tensor.
+ * <p>
+ * This operation creates a new tensor by replicating `input` `multiples` times.
+ * The output tensor's i'th dimension has `input.dims(i) * multiples[i]` elements,
+ * and the values of `input` are replicated `multiples[i]` times along the 'i'th
+ * dimension. For example, tiling `[a b c d]` by `[2]` produces
+ * `[a b c d a b c d]`.
+ * <p>
+ * >>> a = tf.constant([[1,2,3],[4,5,6]], tf.int32)
+ * >>> b = tf.constant([1,2], tf.int32)
+ * >>> tf.tile(a, b)
+ * <tf.Tensor: shape=(2, 6), dtype=int32, numpy=
+ * array([[1, 2, 3, 1, 2, 3],
+ *        [4, 5, 6, 4, 5, 6]], dtype=int32)>
+ * >>> c = tf.constant([2,1], tf.int32)
+ * >>> tf.tile(a, c)
+ * <tf.Tensor: shape=(4, 3), dtype=int32, numpy=
+ * array([[1, 2, 3],
+ *        [4, 5, 6],
+ *        [1, 2, 3],
+ *        [4, 5, 6]], dtype=int32)>
+ * >>> d = tf.constant([2,2], tf.int32)
+ * >>> tf.tile(a, d)
+ * <tf.Tensor: shape=(4, 6), dtype=int32, numpy=
+ * array([[1, 2, 3, 1, 2, 3],
+ *        [4, 5, 6, 4, 5, 6],
+ *        [1, 2, 3, 1, 2, 3],
+ *        [4, 5, 6, 4, 5, 6]], dtype=int32)>
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator
+public final class Tile<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new Tile operation.
+   * 
+   * @param scope current scope
+   * @param input 1-D or higher.
+   * @param multiples 1-D. Length must be the same as the number of dimensions in `input`
+   * @return a new instance of Tile
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType, U extends TNumber> Tile<T> create(Scope scope, Operand<T> input, Operand<U> multiples) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Tile", scope.makeOpName("Tile"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(multiples.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new Tile<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Tile";
+  
+  private Output<T> output;
+  
+  private Tile(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Timestamp.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Timestamp.java
new file mode 100644
index 00000000000..cbbf650a32f
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Timestamp.java
@@ -0,0 +1,75 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat64;
+
+/**
+ * Provides the time since epoch in seconds.
+ * <p>
+ * Returns the timestamp as a `float64` for seconds since the Unix epoch.
+ * <p>
+ * Note: the timestamp is computed when the op is executed, not when it is added
+ * to the graph.
+ */
+@Operator
+public final class Timestamp extends RawOp implements Operand<TFloat64> {
+  
+  /**
+   * Factory method to create a class wrapping a new Timestamp operation.
+   * 
+   * @param scope current scope
+   * @return a new instance of Timestamp
+   */
+  @Endpoint(describeByClass = true)
+  public static Timestamp create(Scope scope) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Timestamp", scope.makeOpName("Timestamp"));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new Timestamp(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<TFloat64> ts() {
+    return ts;
+  }
+  
+  @Override
+  public Output<TFloat64> asOutput() {
+    return ts;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Timestamp";
+  
+  private Output<TFloat64> ts;
+  
+  private Timestamp(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    ts = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TryRpc.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TryRpc.java
new file mode 100644
index 00000000000..3aa6c3a76ad
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/TryRpc.java
@@ -0,0 +1,226 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.TString;
+
+/**
+ * Perform batches of RPC requests.
+ * <p>
+ * This op asynchronously performs either a single RPC request, or a batch
+ * of requests.  RPC requests are defined by three main parameters:
+ * <p>
+ *   - `address` (the host+port or BNS address of the request)
+ *   - `method` (the method name for the request)
+ *   - `request` (the serialized proto string, or vector of strings,
+ *      of the RPC request argument).
+ * <p>
+ * For example, if you have an RPC service running on port localhost:2345,
+ * and its interface is configured with the following proto declaration:
+ * <pre>{@code
+ * service MyService {
+ *   rpc MyMethod(MyRequestProto) returns (MyResponseProto) {
+ *   }
+ * };
+ * }</pre>
+ * then call this op with arguments:
+ * <pre>{@code
+ * address = "localhost:2345"
+ * method = "MyService/MyMethod"
+ * }</pre>
+ * The `request` tensor is a string tensor representing serialized `MyRequestProto`
+ * strings; and the output string tensor `response` will have the same shape
+ * and contain (upon successful completion) corresponding serialized
+ * `MyResponseProto` strings.
+ * <p>
+ * For example, to send a single, empty, `MyRequestProto`, call
+ * this op with `request = ""`.  To send 5 <b>parallel</b> empty requests,
+ * call this op with `request = ["", "", "", "", ""]`.
+ * <p>
+ * More generally, one can create a batch of `MyRequestProto` serialized protos
+ * from regular batched tensors using the `encode_proto` op, and convert
+ * the response `MyResponseProto` serialized protos to batched tensors
+ * using the `decode_proto` op.
+ * <p>
+ * <b>NOTE</b> Working with serialized proto strings is faster than instantiating
+ * actual proto objects in memory, so no performance degradation is expected
+ * compared to writing custom kernels for this workflow.
+ * <p>
+ * Unlike the standard `Rpc` op, if the connection fails or the remote worker
+ * returns an error status, this op does <b>not</b> reraise the exception.
+ * Instead, the `status_code` and `status_message` entry for the corresponding RPC
+ * call is set with the error returned from the RPC call.  The `response` tensor
+ * will contain valid response values for those minibatch entries whose RPCs did
+ * not fail; the rest of the entries will have empty strings.
+ */
+@Operator
+public final class TryRpc extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.core.TryRpc}
+   */
+  public static class Options {
+    
+    /**
+     * @param protocol RPC protocol to use.  Empty string means use the default protocol.
+     * Options include 'grpc'.
+     */
+    public Options protocol(String protocol) {
+      this.protocol = protocol;
+      return this;
+    }
+    
+    /**
+     * @param failFast `boolean`. If `true` (default), then failures to connect
+     * (i.e., the server does not immediately respond) cause an RPC failure.
+     */
+    public Options failFast(Boolean failFast) {
+      this.failFast = failFast;
+      return this;
+    }
+    
+    /**
+     * @param timeoutInMs `int`. If `0` (default), then the kernel will run the RPC
+     * request and only time out if the RPC deadline passes or the session times out.
+     * If this value is greater than `0`, then the op will raise an exception if
+     * the RPC takes longer than `timeout_in_ms`.
+     */
+    public Options timeoutInMs(Long timeoutInMs) {
+      this.timeoutInMs = timeoutInMs;
+      return this;
+    }
+    
+    private String protocol;
+    private Boolean failFast;
+    private Long timeoutInMs;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new TryRpc operation.
+   * 
+   * @param scope current scope
+   * @param address `0-D` or `1-D`.  The address (i.e. host_name:port) of the RPC server.
+   * If this tensor has more than 1 element, then multiple parallel rpc requests
+   * are sent.  This argument broadcasts with `method` and `request`.
+   * @param method `0-D` or `1-D`.  The method address on the RPC server.
+   * If this tensor has more than 1 element, then multiple parallel rpc requests
+   * are sent.  This argument broadcasts with `address` and `request`.
+   * @param request `0-D` or `1-D`.  Serialized proto strings: the rpc request argument.
+   * If this tensor has more than 1 element, then multiple parallel rpc requests
+   * are sent.  This argument broadcasts with `address` and `method`.
+   * @param options carries optional attributes values
+   * @return a new instance of TryRpc
+   */
+  @Endpoint(describeByClass = true)
+  public static TryRpc create(Scope scope, Operand<TString> address, Operand<TString> method, Operand<TString> request, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("TryRpc", scope.makeOpName("TryRpc"));
+    opBuilder.addInput(address.asOutput());
+    opBuilder.addInput(method.asOutput());
+    opBuilder.addInput(request.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.protocol != null) {
+          opBuilder.setAttr("protocol", opts.protocol);
+        }
+        if (opts.failFast != null) {
+          opBuilder.setAttr("fail_fast", opts.failFast);
+        }
+        if (opts.timeoutInMs != null) {
+          opBuilder.setAttr("timeout_in_ms", opts.timeoutInMs);
+        }
+      }
+    }
+    return new TryRpc(opBuilder.build());
+  }
+  
+  /**
+   * @param protocol RPC protocol to use.  Empty string means use the default protocol.
+   * Options include 'grpc'.
+   */
+  public static Options protocol(String protocol) {
+    return new Options().protocol(protocol);
+  }
+  
+  /**
+   * @param failFast `boolean`. If `true` (default), then failures to connect
+   * (i.e., the server does not immediately respond) cause an RPC failure.
+   */
+  public static Options failFast(Boolean failFast) {
+    return new Options().failFast(failFast);
+  }
+  
+  /**
+   * @param timeoutInMs `int`. If `0` (default), then the kernel will run the RPC
+   * request and only time out if the RPC deadline passes or the session times out.
+   * If this value is greater than `0`, then the op will raise an exception if
+   * the RPC takes longer than `timeout_in_ms`.
+   */
+  public static Options timeoutInMs(Long timeoutInMs) {
+    return new Options().timeoutInMs(timeoutInMs);
+  }
+  
+  /**
+   * Same shape as `request`. Serialized proto strings: the rpc responses.
+   */
+  public Output<TString> response() {
+    return response;
+  }
+  
+  /**
+   * Same shape as `request`.  Values correspond to tensorflow Status enum codes.
+   */
+  public Output<TInt32> statusCode() {
+    return statusCode;
+  }
+  
+  /**
+   * Same shape as `request`.  Values correspond to Status messages
+   * returned from the RPC calls.
+   */
+  public Output<TString> statusMessage() {
+    return statusMessage;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "TryRpc";
+  
+  private Output<TString> response;
+  private Output<TInt32> statusCode;
+  private Output<TString> statusMessage;
+  
+  private TryRpc(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    response = operation.output(outputIdx++);
+    statusCode = operation.output(outputIdx++);
+    statusMessage = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Unbatch.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Unbatch.java
new file mode 100644
index 00000000000..84d2063ae8c
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Unbatch.java
@@ -0,0 +1,153 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Reverses the operation of Batch for a single output Tensor.
+ * <p>
+ * An instance of Unbatch either receives an empty batched_tensor, in which case it
+ * asynchronously waits until the values become available from a concurrently
+ * running instance of Unbatch with the same container and shared_name, or receives
+ * a non-empty batched_tensor in which case it finalizes all other concurrently
+ * running instances and outputs its own element from the batch.
+ * <p>
+ * batched_tensor: The possibly transformed output of Batch. The size of the first
+ *  dimension should remain unchanged by the transformations for the operation to
+ *  work.
+ * batch_index: The matching batch_index obtained from Batch.
+ * id: The id scalar emitted by Batch.
+ * unbatched_tensor: The Tensor corresponding to this execution.
+ * timeout_micros: Maximum amount of time (in microseconds) to wait to receive the
+ *  batched input tensor associated with a given invocation of the op.
+ * container: Container to control resource sharing.
+ * shared_name: Instances of Unbatch with the same container and shared_name are
+ *  assumed to possibly belong to the same batch. If left empty, the op name will
+ *  be used as the shared name.
+ * 
+ * @param <T> data type for {@code unbatchedTensor()} output
+ */
+@Operator
+public final class Unbatch<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.core.Unbatch}
+   */
+  public static class Options {
+    
+    /**
+     * @param container 
+     */
+    public Options container(String container) {
+      this.container = container;
+      return this;
+    }
+    
+    /**
+     * @param sharedName 
+     */
+    public Options sharedName(String sharedName) {
+      this.sharedName = sharedName;
+      return this;
+    }
+    
+    private String container;
+    private String sharedName;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new Unbatch operation.
+   * 
+   * @param scope current scope
+   * @param batchedTensor 
+   * @param batchIndex 
+   * @param id 
+   * @param timeoutMicros 
+   * @param options carries optional attributes values
+   * @return a new instance of Unbatch
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> Unbatch<T> create(Scope scope, Operand<T> batchedTensor, Operand<TInt64> batchIndex, Operand<TInt64> id, Long timeoutMicros, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Unbatch", scope.makeOpName("Unbatch"));
+    opBuilder.addInput(batchedTensor.asOutput());
+    opBuilder.addInput(batchIndex.asOutput());
+    opBuilder.addInput(id.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("timeout_micros", timeoutMicros);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.container != null) {
+          opBuilder.setAttr("container", opts.container);
+        }
+        if (opts.sharedName != null) {
+          opBuilder.setAttr("shared_name", opts.sharedName);
+        }
+      }
+    }
+    return new Unbatch<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param container 
+   */
+  public static Options container(String container) {
+    return new Options().container(container);
+  }
+  
+  /**
+   * @param sharedName 
+   */
+  public static Options sharedName(String sharedName) {
+    return new Options().sharedName(sharedName);
+  }
+  
+  /**
+   */
+  public Output<T> unbatchedTensor() {
+    return unbatchedTensor;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return unbatchedTensor;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Unbatch";
+  
+  private Output<T> unbatchedTensor;
+  
+  private Unbatch(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    unbatchedTensor = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/UnbatchGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/UnbatchGrad.java
new file mode 100644
index 00000000000..10062f204fc
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/UnbatchGrad.java
@@ -0,0 +1,149 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Gradient of Unbatch.
+ * <p>
+ * Acts like Batch but using the given batch_index index of batching things as they
+ * become available. This ensures that the gradients are propagated back in the
+ * same session which did the forward pass.
+ * <p>
+ * original_input: The input to the Unbatch operation this is the gradient of.
+ * batch_index: The batch_index given to the Unbatch operation this is the gradient
+ * of.
+ * grad: The downstream gradient.
+ * id: The id scalar emitted by Batch.
+ * batched_grad: The return value, either an empty tensor or the batched gradient.
+ * container: Container to control resource sharing.
+ * shared_name: Instances of UnbatchGrad with the same container and shared_name
+ *  are assumed to possibly belong to the same batch. If left empty, the op name
+ *  will be used as the shared name.
+ * 
+ * @param <T> data type for {@code batchedGrad()} output
+ */
+@Operator
+public final class UnbatchGrad<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.core.UnbatchGrad}
+   */
+  public static class Options {
+    
+    /**
+     * @param container 
+     */
+    public Options container(String container) {
+      this.container = container;
+      return this;
+    }
+    
+    /**
+     * @param sharedName 
+     */
+    public Options sharedName(String sharedName) {
+      this.sharedName = sharedName;
+      return this;
+    }
+    
+    private String container;
+    private String sharedName;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new UnbatchGrad operation.
+   * 
+   * @param scope current scope
+   * @param originalInput 
+   * @param batchIndex 
+   * @param grad 
+   * @param id 
+   * @param options carries optional attributes values
+   * @return a new instance of UnbatchGrad
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> UnbatchGrad<T> create(Scope scope, Operand<T> originalInput, Operand<TInt64> batchIndex, Operand<T> grad, Operand<TInt64> id, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("UnbatchGrad", scope.makeOpName("UnbatchGrad"));
+    opBuilder.addInput(originalInput.asOutput());
+    opBuilder.addInput(batchIndex.asOutput());
+    opBuilder.addInput(grad.asOutput());
+    opBuilder.addInput(id.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.container != null) {
+          opBuilder.setAttr("container", opts.container);
+        }
+        if (opts.sharedName != null) {
+          opBuilder.setAttr("shared_name", opts.sharedName);
+        }
+      }
+    }
+    return new UnbatchGrad<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param container 
+   */
+  public static Options container(String container) {
+    return new Options().container(container);
+  }
+  
+  /**
+   * @param sharedName 
+   */
+  public static Options sharedName(String sharedName) {
+    return new Options().sharedName(sharedName);
+  }
+  
+  /**
+   */
+  public Output<T> batchedGrad() {
+    return batchedGrad;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return batchedGrad;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "UnbatchGrad";
+  
+  private Output<T> batchedGrad;
+  
+  private UnbatchGrad(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    batchedGrad = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Unique.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Unique.java
new file mode 100644
index 00000000000..8d1bcc7a6ef
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Unique.java
@@ -0,0 +1,142 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Finds unique elements along an axis of a tensor.
+ * <p>
+ * This operation either returns a tensor `y` containing unique elements
+ * along the `axis` of a tensor. The returned unique elements is sorted
+ * in the same order as they occur along `axis` in `x`.
+ * This operation also returns a tensor `idx` that is the same size as
+ * the number of the elements in `x` along the `axis` dimension. It
+ * contains the index in the unique output `y`.
+ * In other words, for an `1-D` tensor `x` with `axis = None:
+ * <p>
+ * `y[idx[i]] = x[i] for i in [0, 1,...,rank(x) - 1]`
+ * <p>
+ * For example:
+ * <pre>{@code
+ * # tensor 'x' is [1, 1, 2, 4, 4, 4, 7, 8, 8]
+ * y, idx = unique(x)
+ * y ==> [1, 2, 4, 7, 8]
+ * idx ==> [0, 0, 1, 2, 2, 2, 3, 4, 4]
+ * }</pre>
+ * For an `2-D` tensor `x` with `axis = 0`:
+ * <pre>{@code
+ * # tensor 'x' is [[1, 0, 0],
+ * #                [1, 0, 0],
+ * #                [2, 0, 0]]
+ * y, idx = unique(x, axis=0)
+ * y ==> [[1, 0, 0],
+ *        [2, 0, 0]]
+ * idx ==> [0, 0, 1]
+ * }</pre>
+ * For an `2-D` tensor `x` with `axis = 1`:
+ * <pre>{@code
+ * # tensor 'x' is [[1, 0, 0],
+ * #                [1, 0, 0],
+ * #                [2, 0, 0]]
+ * y, idx = unique(x, axis=1)
+ * y ==> [[1, 0],
+ *        [1, 0],
+ *        [2, 0]]
+ * idx ==> [0, 1, 1]
+ * }</pre>
+ * 
+ * 
+ * @param <T> data type for {@code y()} output
+ * @param <V> data type for {@code idx()} output
+ */
+@Operator
+public final class Unique<T extends TType, V extends TNumber> extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new Unique operation.
+   * 
+   * @param scope current scope
+   * @param x A `Tensor`.
+   * @param axis A `Tensor` of type `int32` (default: None). The axis of the Tensor to
+   * find the unique elements.
+   * @param outIdx 
+   * @return a new instance of Unique
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType, V extends TNumber, U extends TNumber> Unique<T, V> create(Scope scope, Operand<T> x, Operand<U> axis, Class<V> outIdx) {
+    OperationBuilder opBuilder = scope.env().opBuilder("UniqueV2", scope.makeOpName("Unique"));
+    opBuilder.addInput(x.asOutput());
+    opBuilder.addInput(axis.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("out_idx", outIdx);
+    return new Unique<T, V>(opBuilder.build());
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new Unique operation using default output types.
+   * 
+   * @param scope current scope
+   * @param x A `Tensor`.
+   * @param axis A `Tensor` of type `int32` (default: None). The axis of the Tensor to
+   * find the unique elements.
+   * @return a new instance of Unique
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType, U extends TNumber> Unique<T, TInt32> create(Scope scope, Operand<T> x, Operand<U> axis) {
+    return create(scope, x, axis, TInt32.class);
+  }
+  
+  /**
+   * A `Tensor`. Unique elements along the `axis` of `Tensor` x.
+   */
+  public Output<T> y() {
+    return y;
+  }
+  
+  /**
+   * A 1-D Tensor. Has the same type as x that contains the index of each
+   * value of x in the output y.
+   */
+  public Output<V> idx() {
+    return idx;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "UniqueV2";
+  
+  private Output<T> y;
+  private Output<V> idx;
+  
+  private Unique(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    y = operation.output(outputIdx++);
+    idx = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/UniqueWithCounts.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/UniqueWithCounts.java
new file mode 100644
index 00000000000..6316f3659c9
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/UniqueWithCounts.java
@@ -0,0 +1,155 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Finds unique elements along an axis of a tensor.
+ * <p>
+ * This operation either returns a tensor `y` containing unique elements
+ * along the `axis` of a tensor. The returned unique elements is sorted
+ * in the same order as they occur along `axis` in `x`.
+ * This operation also returns a tensor `idx` and a tensor `count`
+ * that are the same size as the number of the elements in `x` along the
+ * `axis` dimension. The `idx` contains the index in the unique output `y`
+ * and the `count` contains the count in the unique output `y`.
+ * In other words, for an `1-D` tensor `x` with `axis = None:
+ * <p>
+ * `y[idx[i]] = x[i] for i in [0, 1,...,rank(x) - 1]`
+ * <p>
+ * For example:
+ * <pre>{@code
+ * # tensor 'x' is [1, 1, 2, 4, 4, 4, 7, 8, 8]
+ * y, idx, count = unique_with_counts(x)
+ * y ==> [1, 2, 4, 7, 8]
+ * idx ==> [0, 0, 1, 2, 2, 2, 3, 4, 4]
+ * count ==> [2, 1, 3, 1, 2]
+ * }</pre>
+ * For an `2-D` tensor `x` with `axis = 0`:
+ * <pre>{@code
+ * # tensor 'x' is [[1, 0, 0],
+ * #                [1, 0, 0],
+ * #                [2, 0, 0]]
+ * y, idx, count = unique_with_counts(x, axis=0)
+ * y ==> [[1, 0, 0],
+ *        [2, 0, 0]]
+ * idx ==> [0, 0, 1]
+ * count ==> [2, 1]
+ * }</pre>
+ * For an `2-D` tensor `x` with `axis = 1`:
+ * <pre>{@code
+ * # tensor 'x' is [[1, 0, 0],
+ * #                [1, 0, 0],
+ * #                [2, 0, 0]]
+ * y, idx, count = unique_with_counts(x, axis=1)
+ * y ==> [[1, 0],
+ *        [1, 0],
+ *        [2, 0]]
+ * idx ==> [0, 1, 1]
+ * count ==> [1, 2]
+ * }</pre>
+ * 
+ * 
+ * @param <T> data type for {@code y()} output
+ * @param <V> data type for {@code idx()} output
+ */
+@Operator
+public final class UniqueWithCounts<T extends TType, V extends TNumber> extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new UniqueWithCounts operation.
+   * 
+   * @param scope current scope
+   * @param x A `Tensor`.
+   * @param axis A `Tensor` of type `int32` (default: None). The axis of the Tensor to
+   * find the unique elements.
+   * @param outIdx 
+   * @return a new instance of UniqueWithCounts
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType, V extends TNumber, U extends TNumber> UniqueWithCounts<T, V> create(Scope scope, Operand<T> x, Operand<U> axis, Class<V> outIdx) {
+    OperationBuilder opBuilder = scope.env().opBuilder("UniqueWithCountsV2", scope.makeOpName("UniqueWithCounts"));
+    opBuilder.addInput(x.asOutput());
+    opBuilder.addInput(axis.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("out_idx", outIdx);
+    return new UniqueWithCounts<T, V>(opBuilder.build());
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new UniqueWithCounts operation using default output types.
+   * 
+   * @param scope current scope
+   * @param x A `Tensor`.
+   * @param axis A `Tensor` of type `int32` (default: None). The axis of the Tensor to
+   * find the unique elements.
+   * @return a new instance of UniqueWithCounts
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType, U extends TNumber> UniqueWithCounts<T, TInt32> create(Scope scope, Operand<T> x, Operand<U> axis) {
+    return create(scope, x, axis, TInt32.class);
+  }
+  
+  /**
+   * A `Tensor`. Unique elements along the `axis` of `Tensor` x.
+   */
+  public Output<T> y() {
+    return y;
+  }
+  
+  /**
+   * A 1-D Tensor. Has the same type as x that contains the index of each
+   * value of x in the output y.
+   */
+  public Output<V> idx() {
+    return idx;
+  }
+  
+  /**
+   * A 1-D Tensor. The count of each value of x in the output y.
+   */
+  public Output<V> count() {
+    return count;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "UniqueWithCountsV2";
+  
+  private Output<T> y;
+  private Output<V> idx;
+  private Output<V> count;
+  
+  private UniqueWithCounts(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    y = operation.output(outputIdx++);
+    idx = operation.output(outputIdx++);
+    count = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/UnravelIndex.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/UnravelIndex.java
new file mode 100644
index 00000000000..37c5f2ded90
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/UnravelIndex.java
@@ -0,0 +1,99 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Converts an array of flat indices into a tuple of coordinate arrays.
+ * <p>
+ * 
+ * Example:
+ * <pre>{@code
+ * y = tf.unravel_index(indices=[2, 5, 7], dims=[3, 3])
+ * # 'dims' represent a hypothetical (3, 3) tensor of indices:
+ * # [[0, 1, *2*],
+ * #  [3, 4, *5*],
+ * #  [6, *7*, 8]]
+ * # For each entry from 'indices', this operation returns
+ * # its coordinates (marked with '*'), such as
+ * # 2 ==> (0, 2)
+ * # 5 ==> (1, 2)
+ * # 7 ==> (2, 1)
+ * y ==> [[0, 1, 2], [2, 2, 1]]
+ * }</pre>
+ * @compatibility(numpy)
+ * Equivalent to np.unravel_index
+ * @end_compatibility
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator
+public final class UnravelIndex<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new UnravelIndex operation.
+   * 
+   * @param scope current scope
+   * @param indices An 0-D or 1-D `int` Tensor whose elements are indices into the
+   * flattened version of an array of dimensions dims.
+   * @param dims An 1-D `int` Tensor. The shape of the array to use for unraveling
+   * indices.
+   * @return a new instance of UnravelIndex
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> UnravelIndex<T> create(Scope scope, Operand<T> indices, Operand<T> dims) {
+    OperationBuilder opBuilder = scope.env().opBuilder("UnravelIndex", scope.makeOpName("UnravelIndex"));
+    opBuilder.addInput(indices.asOutput());
+    opBuilder.addInput(dims.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new UnravelIndex<T>(opBuilder.build());
+  }
+  
+  /**
+   * An 2-D (or 1-D if indices is 0-D) tensor where each row has the
+   * same shape as the indices array.
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "UnravelIndex";
+  
+  private Output<T> output;
+  
+  private UnravelIndex(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Unstack.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Unstack.java
new file mode 100644
index 00000000000..a48f57289ae
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Unstack.java
@@ -0,0 +1,133 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import java.util.Arrays;
+import java.util.Iterator;
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Unpacks a given dimension of a rank-`R` tensor into `num` rank-`(R-1)` tensors.
+ * <p>
+ * Unpacks `num` tensors from `value` by chipping it along the `axis` dimension.
+ * For example, given a tensor of shape `(A, B, C, D)`;
+ * <p>
+ * If `axis == 0` then the i'th tensor in `output` is the slice `value[i, :, :, :]`
+ *   and each tensor in `output` will have shape `(B, C, D)`. (Note that the
+ *   dimension unpacked along is gone, unlike `split`).
+ * <p>
+ * If `axis == 1` then the i'th tensor in `output` is the slice `value[:, i, :, :]`
+ *   and each tensor in `output` will have shape `(A, C, D)`.
+ * Etc.
+ * <p>
+ * This is the opposite of `pack`.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator
+public final class Unstack<T extends TType> extends RawOp implements Iterable<Operand<T>> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.core.Unstack}
+   */
+  public static class Options {
+    
+    /**
+     * @param axis Dimension along which to unpack.  Negative values wrap around, so the
+     * valid range is `[-R, R)`.
+     */
+    public Options axis(Long axis) {
+      this.axis = axis;
+      return this;
+    }
+    
+    private Long axis;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new Unstack operation.
+   * 
+   * @param scope current scope
+   * @param value 1-D or higher, with `axis` dimension size equal to `num`.
+   * @param num 
+   * @param options carries optional attributes values
+   * @return a new instance of Unstack
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> Unstack<T> create(Scope scope, Operand<T> value, Long num, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Unpack", scope.makeOpName("Unstack"));
+    opBuilder.addInput(value.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("num", num);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.axis != null) {
+          opBuilder.setAttr("axis", opts.axis);
+        }
+      }
+    }
+    return new Unstack<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param axis Dimension along which to unpack.  Negative values wrap around, so the
+   * valid range is `[-R, R)`.
+   */
+  public static Options axis(Long axis) {
+    return new Options().axis(axis);
+  }
+  
+  /**
+   * The list of tensors unpacked from `value`.
+   */
+  public List<Output<T>> output() {
+    return output;
+  }
+  
+  @Override
+  @SuppressWarnings({"rawtypes", "unchecked"})
+  public Iterator<Operand<T>> iterator() {
+    return (Iterator) output.iterator();
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Unpack";
+  
+  private List<Output<T>> output;
+  
+  @SuppressWarnings("unchecked")
+  private Unstack(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    int outputLength = operation.outputListLength("output");
+    output = Arrays.asList((Output<T>[])operation.outputList(outputIdx, outputLength));
+    outputIdx += outputLength;
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Unstage.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Unstage.java
new file mode 100644
index 00000000000..f6bd852f898
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Unstage.java
@@ -0,0 +1,176 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import java.util.Arrays;
+import java.util.Iterator;
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Op is similar to a lightweight Dequeue.
+ * <p>
+ * The basic functionality is similar to dequeue with many fewer
+ * capabilities and options.  This Op is optimized for performance.
+ */
+@Operator
+public final class Unstage extends RawOp implements Iterable<Operand<TType>> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.core.Unstage}
+   */
+  public static class Options {
+    
+    /**
+     * @param capacity 
+     */
+    public Options capacity(Long capacity) {
+      this.capacity = capacity;
+      return this;
+    }
+    
+    /**
+     * @param memoryLimit 
+     */
+    public Options memoryLimit(Long memoryLimit) {
+      this.memoryLimit = memoryLimit;
+      return this;
+    }
+    
+    /**
+     * @param container 
+     */
+    public Options container(String container) {
+      this.container = container;
+      return this;
+    }
+    
+    /**
+     * @param sharedName 
+     */
+    public Options sharedName(String sharedName) {
+      this.sharedName = sharedName;
+      return this;
+    }
+    
+    private Long capacity;
+    private Long memoryLimit;
+    private String container;
+    private String sharedName;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new Unstage operation.
+   * 
+   * @param scope current scope
+   * @param dtypes 
+   * @param options carries optional attributes values
+   * @return a new instance of Unstage
+   */
+  @Endpoint(describeByClass = true)
+  public static Unstage create(Scope scope, List<Class<? extends TType>> dtypes, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Unstage", scope.makeOpName("Unstage"));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    Class[] dtypesArray = new Class[dtypes.size()];
+    for (int i = 0; i < dtypesArray.length; ++i) {
+      dtypesArray[i] = dtypes.get(i);
+    }
+    opBuilder.setAttr("dtypes", dtypesArray);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.capacity != null) {
+          opBuilder.setAttr("capacity", opts.capacity);
+        }
+        if (opts.memoryLimit != null) {
+          opBuilder.setAttr("memory_limit", opts.memoryLimit);
+        }
+        if (opts.container != null) {
+          opBuilder.setAttr("container", opts.container);
+        }
+        if (opts.sharedName != null) {
+          opBuilder.setAttr("shared_name", opts.sharedName);
+        }
+      }
+    }
+    return new Unstage(opBuilder.build());
+  }
+  
+  /**
+   * @param capacity 
+   */
+  public static Options capacity(Long capacity) {
+    return new Options().capacity(capacity);
+  }
+  
+  /**
+   * @param memoryLimit 
+   */
+  public static Options memoryLimit(Long memoryLimit) {
+    return new Options().memoryLimit(memoryLimit);
+  }
+  
+  /**
+   * @param container 
+   */
+  public static Options container(String container) {
+    return new Options().container(container);
+  }
+  
+  /**
+   * @param sharedName 
+   */
+  public static Options sharedName(String sharedName) {
+    return new Options().sharedName(sharedName);
+  }
+  
+  /**
+   */
+  public List<Output<?>> values() {
+    return values;
+  }
+  
+  @Override
+  @SuppressWarnings({"rawtypes", "unchecked"})
+  public Iterator<Operand<TType>> iterator() {
+    return (Iterator) values.iterator();
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Unstage";
+  
+  private List<Output<?>> values;
+  
+  private Unstage(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    int valuesLength = operation.outputListLength("values");
+    values = Arrays.asList(operation.outputList(outputIdx, valuesLength));
+    outputIdx += valuesLength;
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/UpperBound.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/UpperBound.java
new file mode 100644
index 00000000000..67d2e9255e4
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/UpperBound.java
@@ -0,0 +1,115 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Applies upper_bound(sorted_search_values, values) along each row.
+ * <p>
+ * Each set of rows with the same index in (sorted_inputs, values) is treated
+ * independently.  The resulting row is the equivalent of calling
+ * `np.searchsorted(sorted_inputs, values, side='right')`.
+ * <p>
+ * The result is not a global index to the entire
+ * `Tensor`, but rather just the index in the last dimension.
+ * <p>
+ * A 2-D example:
+ *   sorted_sequence = [[0, 3, 9, 9, 10],
+ *                      [1, 2, 3, 4, 5]]
+ *   values = [[2, 4, 9],
+ *             [0, 2, 6]]
+ * <p>
+ *   result = UpperBound(sorted_sequence, values)
+ * <p>
+ *   result == [[1, 2, 4],
+ *              [0, 2, 5]]
+ * 
+ * @param <U> data type for {@code output()} output
+ */
+public final class UpperBound<U extends TNumber> extends RawOp implements Operand<U> {
+  
+  /**
+   * Factory method to create a class wrapping a new UpperBound operation.
+   * 
+   * @param scope current scope
+   * @param sortedInputs 2-D Tensor where each row is ordered.
+   * @param values 2-D Tensor with the same numbers of rows as `sorted_search_values`. Contains
+   * the values that will be searched for in `sorted_search_values`.
+   * @param outType 
+   * @return a new instance of UpperBound
+   */
+  @Endpoint(describeByClass = true)
+  public static <U extends TNumber, T extends TType> UpperBound<U> create(Scope scope, Operand<T> sortedInputs, Operand<T> values, Class<U> outType) {
+    OperationBuilder opBuilder = scope.env().opBuilder("UpperBound", scope.makeOpName("UpperBound"));
+    opBuilder.addInput(sortedInputs.asOutput());
+    opBuilder.addInput(values.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("out_type", outType);
+    return new UpperBound<U>(opBuilder.build());
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new UpperBound operation using default output types.
+   * 
+   * @param scope current scope
+   * @param sortedInputs 2-D Tensor where each row is ordered.
+   * @param values 2-D Tensor with the same numbers of rows as `sorted_search_values`. Contains
+   * the values that will be searched for in `sorted_search_values`.
+   * @return a new instance of UpperBound
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> UpperBound<TInt32> create(Scope scope, Operand<T> sortedInputs, Operand<T> values) {
+    return create(scope, sortedInputs, values, TInt32.class);
+  }
+  
+  /**
+   * A `Tensor` with the same shape as `values`.  It contains the last scalar index
+   * into the last dimension where values can be inserted without changing the
+   * ordered property.
+   */
+  public Output<U> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<U> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "UpperBound";
+  
+  private Output<U> output;
+  
+  private UpperBound(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/VarHandleOp.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/VarHandleOp.java
new file mode 100644
index 00000000000..c264519b454
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/VarHandleOp.java
@@ -0,0 +1,156 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Creates a handle to a Variable resource.
+ */
+@Operator
+public final class VarHandleOp extends RawOp implements Operand<TType> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.core.VarHandleOp}
+   */
+  public static class Options {
+    
+    /**
+     * @param container the container this variable is placed in.
+     */
+    public Options container(String container) {
+      this.container = container;
+      return this;
+    }
+    
+    /**
+     * @param sharedName the name by which this variable is referred to.
+     */
+    public Options sharedName(String sharedName) {
+      this.sharedName = sharedName;
+      return this;
+    }
+    
+    /**
+     * @param allowedDevices DEPRECATED. The allowed devices containing the resource variable. Set when the
+     * output ResourceHandle represents a per-replica/partitioned resource variable.
+     */
+    public Options allowedDevices(List<String> allowedDevices) {
+      this.allowedDevices = allowedDevices;
+      return this;
+    }
+    
+    private String container;
+    private String sharedName;
+    private List<String> allowedDevices;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new VarHandleOp operation.
+   * 
+   * @param scope current scope
+   * @param dtype the type of this variable. Must agree with the dtypes
+   * of all ops using this variable.
+   * @param shape The (possibly partially specified) shape of this variable.
+   * @param options carries optional attributes values
+   * @return a new instance of VarHandleOp
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> VarHandleOp create(Scope scope, Class<T> dtype, Shape shape, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("VarHandleOp", scope.makeOpName("VarHandleOp"));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("dtype", dtype);
+    opBuilder.setAttr("shape", shape);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.container != null) {
+          opBuilder.setAttr("container", opts.container);
+        }
+        if (opts.sharedName != null) {
+          opBuilder.setAttr("shared_name", opts.sharedName);
+        }
+        if (opts.allowedDevices != null) {
+          String[] allowedDevicesArray = new String[opts.allowedDevices.size()];
+          for (int i = 0; i < allowedDevicesArray.length; ++i) {
+            allowedDevicesArray[i] = opts.allowedDevices.get(i);
+          }
+          opBuilder.setAttr("allowed_devices", allowedDevicesArray);
+        }
+      }
+    }
+    return new VarHandleOp(opBuilder.build());
+  }
+  
+  /**
+   * @param container the container this variable is placed in.
+   */
+  public static Options container(String container) {
+    return new Options().container(container);
+  }
+  
+  /**
+   * @param sharedName the name by which this variable is referred to.
+   */
+  public static Options sharedName(String sharedName) {
+    return new Options().sharedName(sharedName);
+  }
+  
+  /**
+   * @param allowedDevices DEPRECATED. The allowed devices containing the resource variable. Set when the
+   * output ResourceHandle represents a per-replica/partitioned resource variable.
+   */
+  public static Options allowedDevices(List<String> allowedDevices) {
+    return new Options().allowedDevices(allowedDevices);
+  }
+  
+  /**
+   */
+  public Output<?> resource() {
+    return resource;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) resource;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "VarHandleOp";
+  
+  private Output<?> resource;
+  
+  private VarHandleOp(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    resource = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/VarIsInitializedOp.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/VarIsInitializedOp.java
new file mode 100644
index 00000000000..f2c3df0456d
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/VarIsInitializedOp.java
@@ -0,0 +1,74 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TBool;
+
+/**
+ * Checks whether a resource handle-based variable has been initialized.
+ */
+@Operator
+public final class VarIsInitializedOp extends RawOp implements Operand<TBool> {
+  
+  /**
+   * Factory method to create a class wrapping a new VarIsInitializedOp operation.
+   * 
+   * @param scope current scope
+   * @param resource the input resource handle.
+   * @return a new instance of VarIsInitializedOp
+   */
+  @Endpoint(describeByClass = true)
+  public static VarIsInitializedOp create(Scope scope, Operand<?> resource) {
+    OperationBuilder opBuilder = scope.env().opBuilder("VarIsInitializedOp", scope.makeOpName("VarIsInitializedOp"));
+    opBuilder.addInput(resource.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new VarIsInitializedOp(opBuilder.build());
+  }
+  
+  /**
+   * a scalar boolean which is true if the variable has been
+   * initialized.
+   */
+  public Output<TBool> isInitialized() {
+    return isInitialized;
+  }
+  
+  @Override
+  public Output<TBool> asOutput() {
+    return isInitialized;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "VarIsInitializedOp";
+  
+  private Output<TBool> isInitialized;
+  
+  private VarIsInitializedOp(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    isInitialized = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Variable.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Variable.java
new file mode 100644
index 00000000000..bcd9719c932
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Variable.java
@@ -0,0 +1,139 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Holds state in the form of a tensor that persists across steps.
+ * <p>
+ * Outputs a ref to the tensor state so it may be read or modified.
+ * TODO(zhifengc/mrry): Adds a pointer to a more detail document
+ * about sharing states in tensorflow.
+ * 
+ * @param <T> data type for {@code ref()} output
+ */
+@Operator
+public final class Variable<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.core.Variable}
+   */
+  public static class Options {
+    
+    /**
+     * @param container If non-empty, this variable is placed in the given container.
+     * Otherwise, a default container is used.
+     */
+    public Options container(String container) {
+      this.container = container;
+      return this;
+    }
+    
+    /**
+     * @param sharedName If non-empty, this variable is named in the given bucket
+     * with this shared_name. Otherwise, the node name is used instead.
+     */
+    public Options sharedName(String sharedName) {
+      this.sharedName = sharedName;
+      return this;
+    }
+    
+    private String container;
+    private String sharedName;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new Variable operation.
+   * 
+   * @param scope current scope
+   * @param shape The shape of the variable tensor.
+   * @param dtype The type of elements in the variable tensor.
+   * @param options carries optional attributes values
+   * @return a new instance of Variable
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> Variable<T> create(Scope scope, Shape shape, Class<T> dtype, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("VariableV2", scope.makeOpName("Variable"));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("shape", shape);
+    opBuilder.setAttr("dtype", dtype);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.container != null) {
+          opBuilder.setAttr("container", opts.container);
+        }
+        if (opts.sharedName != null) {
+          opBuilder.setAttr("shared_name", opts.sharedName);
+        }
+      }
+    }
+    return new Variable<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param container If non-empty, this variable is placed in the given container.
+   * Otherwise, a default container is used.
+   */
+  public static Options container(String container) {
+    return new Options().container(container);
+  }
+  
+  /**
+   * @param sharedName If non-empty, this variable is named in the given bucket
+   * with this shared_name. Otherwise, the node name is used instead.
+   */
+  public static Options sharedName(String sharedName) {
+    return new Options().sharedName(sharedName);
+  }
+  
+  /**
+   * A reference to the variable tensor.
+   */
+  public Output<T> ref() {
+    return ref;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return ref;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "VariableV2";
+  
+  private Output<T> ref;
+  
+  private Variable(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    ref = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/VariableShape.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/VariableShape.java
new file mode 100644
index 00000000000..2a4c0ba60ec
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/VariableShape.java
@@ -0,0 +1,98 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Returns the shape of the variable pointed to by `resource`.
+ * <p>
+ * This operation returns a 1-D integer tensor representing the shape of `input`.
+ * <p>
+ * For example:
+ * <pre>{@code
+ * # 't' is [[[1, 1, 1], [2, 2, 2]], [[3, 3, 3], [4, 4, 4]]]
+ * shape(t) ==> [2, 2, 3]
+ * }</pre>
+ * 
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator
+public final class VariableShape<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new VariableShape operation.
+   * 
+   * @param scope current scope
+   * @param input 
+   * @param outType 
+   * @return a new instance of VariableShape
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> VariableShape<T> create(Scope scope, Operand<?> input, Class<T> outType) {
+    OperationBuilder opBuilder = scope.env().opBuilder("VariableShape", scope.makeOpName("VariableShape"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("out_type", outType);
+    return new VariableShape<T>(opBuilder.build());
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new VariableShape operation using default output types.
+   * 
+   * @param scope current scope
+   * @param input 
+   * @return a new instance of VariableShape
+   */
+  @Endpoint(describeByClass = true)
+  public static VariableShape<TInt32> create(Scope scope, Operand<?> input) {
+    return create(scope, input, TInt32.class);
+  }
+  
+  /**
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "VariableShape";
+  
+  private Output<T> output;
+  
+  private VariableShape(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Where.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Where.java
new file mode 100644
index 00000000000..a1af7f9282a
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/Where.java
@@ -0,0 +1,133 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Returns locations of nonzero / true values in a tensor.
+ * <p>
+ * This operation returns the coordinates of true elements in `condition`. The
+ * coordinates are returned in a 2-D tensor where the first dimension (rows)
+ * represents the number of true elements, and the second dimension (columns)
+ * represents the coordinates of the true elements. Keep in mind, the shape of
+ * the output tensor can vary depending on how many true values there are in
+ * `condition`. Indices are output in row-major order.
+ * <p>
+ * For example:
+ * <pre>{@code
+ * # 'input' tensor is [[True, False]
+ * #                    [True, False]]
+ * # 'input' has two true values, so output has two coordinates.
+ * # 'input' has rank of 2, so coordinates have two indices.
+ * where(input) ==> [[0, 0],
+ *                   [1, 0]]
+ * 
+ * # `condition` tensor is [[[True, False]
+ * #                     [True, False]]
+ * #                    [[False, True]
+ * #                     [False, True]]
+ * #                    [[False, False]
+ * #                     [False, True]]]
+ * # 'input' has 5 true values, so output has 5 coordinates.
+ * # 'input' has rank of 3, so coordinates have three indices.
+ * where(input) ==> [[0, 0, 0],
+ *                   [0, 1, 0],
+ *                   [1, 0, 1],
+ *                   [1, 1, 1],
+ *                   [2, 1, 1]]
+ * 
+ * # `condition` tensor is [[[1.5,  0.0]
+ * #                     [-0.5, 0.0]]
+ * #                    [[0.0,  0.25]
+ * #                     [0.0,  0.75]]
+ * #                    [[0.0,  0.0]
+ * #                     [0.0,  0.01]]]
+ * # 'input' has 5 nonzero values, so output has 5 coordinates.
+ * # 'input' has rank of 3, so coordinates have three indices.
+ * where(input) ==> [[0, 0, 0],
+ *                   [0, 1, 0],
+ *                   [1, 0, 1],
+ *                   [1, 1, 1],
+ *                   [2, 1, 1]]
+ * 
+ * # `condition` tensor is [[[1.5 + 0.0j, 0.0  + 0.0j]
+ * #                     [0.0 + 0.5j, 0.0  + 0.0j]]
+ * #                    [[0.0 + 0.0j, 0.25 + 1.5j]
+ * #                     [0.0 + 0.0j, 0.75 + 0.0j]]
+ * #                    [[0.0 + 0.0j, 0.0  + 0.0j]
+ * #                     [0.0 + 0.0j, 0.01 + 0.0j]]]
+ * # 'input' has 5 nonzero magnitude values, so output has 5 coordinates.
+ * # 'input' has rank of 3, so coordinates have three indices.
+ * where(input) ==> [[0, 0, 0],
+ *                   [0, 1, 0],
+ *                   [1, 0, 1],
+ *                   [1, 1, 1],
+ *                   [2, 1, 1]]
+ * }</pre>
+ * 
+ */
+@Operator
+public final class Where extends RawOp implements Operand<TInt64> {
+  
+  /**
+   * Factory method to create a class wrapping a new Where operation.
+   * 
+   * @param scope current scope
+   * @param condition 
+   * @return a new instance of Where
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> Where create(Scope scope, Operand<T> condition) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Where", scope.makeOpName("Where"));
+    opBuilder.addInput(condition.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new Where(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<TInt64> index() {
+    return index;
+  }
+  
+  @Override
+  public Output<TInt64> asOutput() {
+    return index;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Where";
+  
+  private Output<TInt64> index;
+  
+  private Where(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    index = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/XlaSpmdFullToShardShape.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/XlaSpmdFullToShardShape.java
new file mode 100644
index 00000000000..6615d2ef9f6
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/XlaSpmdFullToShardShape.java
@@ -0,0 +1,81 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * An op used by XLA SPMD partitioner to switch from automatic partitioning to
+ * <p>
+ * manual partitioning. It annotates the input (full-shape, to be automatically
+ * partitioned) with the same sharding used by manual partitioning, and outputs a
+ * shard-shaped tensor to be consumed by later manually-partitioned ops. If the
+ * shape is not evenly partitionable, the padding region will be masked with 0s.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator
+public final class XlaSpmdFullToShardShape<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new XlaSpmdFullToShardShape operation.
+   * 
+   * @param scope current scope
+   * @param input 
+   * @param manualSharding 
+   * @return a new instance of XlaSpmdFullToShardShape
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> XlaSpmdFullToShardShape<T> create(Scope scope, Operand<T> input, String manualSharding) {
+    OperationBuilder opBuilder = scope.env().opBuilder("XlaSpmdFullToShardShape", scope.makeOpName("XlaSpmdFullToShardShape"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("manual_sharding", manualSharding);
+    return new XlaSpmdFullToShardShape<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "XlaSpmdFullToShardShape";
+  
+  private Output<T> output;
+  
+  private XlaSpmdFullToShardShape(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/XlaSpmdShardToFullShape.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/XlaSpmdShardToFullShape.java
new file mode 100644
index 00000000000..75e31c7c317
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/XlaSpmdShardToFullShape.java
@@ -0,0 +1,83 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * An op used by XLA SPMD partitioner to switch from manual partitioning to
+ * <p>
+ * automatic partitioning. It converts the shard-shaped, manually partitioned input
+ * into full-shaped tensor to be partitioned automatically with the same sharding
+ * used by manual partitioning.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator
+public final class XlaSpmdShardToFullShape<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new XlaSpmdShardToFullShape operation.
+   * 
+   * @param scope current scope
+   * @param input 
+   * @param manualSharding 
+   * @param fullShape 
+   * @return a new instance of XlaSpmdShardToFullShape
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> XlaSpmdShardToFullShape<T> create(Scope scope, Operand<T> input, String manualSharding, Shape fullShape) {
+    OperationBuilder opBuilder = scope.env().opBuilder("XlaSpmdShardToFullShape", scope.makeOpName("XlaSpmdShardToFullShape"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("manual_sharding", manualSharding);
+    opBuilder.setAttr("full_shape", fullShape);
+    return new XlaSpmdShardToFullShape<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "XlaSpmdShardToFullShape";
+  
+  private Output<T> output;
+  
+  private XlaSpmdShardToFullShape(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ZerosLike.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ZerosLike.java
new file mode 100644
index 00000000000..5b9352ef6bd
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/core/ZerosLike.java
@@ -0,0 +1,75 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.core;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Returns a tensor of zeros with the same shape and type as x.
+ * 
+ * @param <T> data type for {@code y()} output
+ */
+@Operator
+public final class ZerosLike<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new ZerosLike operation.
+   * 
+   * @param scope current scope
+   * @param x a tensor of type T.
+   * @return a new instance of ZerosLike
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> ZerosLike<T> create(Scope scope, Operand<T> x) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ZerosLike", scope.makeOpName("ZerosLike"));
+    opBuilder.addInput(x.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new ZerosLike<T>(opBuilder.build());
+  }
+  
+  /**
+   * a tensor of the same shape and type as x but filled with zeros.
+   */
+  public Output<T> y() {
+    return y;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return y;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ZerosLike";
+  
+  private Output<T> y;
+  
+  private ZerosLike(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    y = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/AnonymousIterator.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/AnonymousIterator.java
new file mode 100644
index 00000000000..8236e750ff5
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/AnonymousIterator.java
@@ -0,0 +1,91 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data;
+
+import java.util.List;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * A container for an iterator resource.
+ */
+@Operator(group = "data")
+public final class AnonymousIterator extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new AnonymousIterator operation.
+   * 
+   * @param scope current scope
+   * @param outputTypes 
+   * @param outputShapes 
+   * @return a new instance of AnonymousIterator
+   */
+  @Endpoint(describeByClass = true)
+  public static AnonymousIterator create(Scope scope, List<Class<? extends TType>> outputTypes, List<Shape> outputShapes) {
+    OperationBuilder opBuilder = scope.env().opBuilder("AnonymousIteratorV2", scope.makeOpName("AnonymousIterator"));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    Class[] outputTypesArray = new Class[outputTypes.size()];
+    for (int i = 0; i < outputTypesArray.length; ++i) {
+      outputTypesArray[i] = outputTypes.get(i);
+    }
+    opBuilder.setAttr("output_types", outputTypesArray);
+    Shape[] outputShapesArray = new Shape[outputShapes.size()];
+    for (int i = 0; i < outputShapesArray.length; ++i) {
+      outputShapesArray[i] = outputShapes.get(i);
+    }
+    opBuilder.setAttr("output_shapes", outputShapesArray);
+    return new AnonymousIterator(opBuilder.build());
+  }
+  
+  /**
+   * A handle to the iterator that can be passed to a "MakeIterator" or
+   * "IteratorGetNext" op. In contrast to Iterator, AnonymousIterator prevents
+   * resource sharing by name, and does not keep a reference to the resource
+   * container.
+   */
+  public Output<?> handle() {
+    return handle;
+  }
+  
+  /**
+   * A variant deleter that should be passed into the op that deletes the iterator.
+   */
+  public Output<?> deleter() {
+    return deleter;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "AnonymousIteratorV2";
+  
+  private Output<?> handle;
+  private Output<?> deleter;
+  
+  private AnonymousIterator(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    handle = operation.output(outputIdx++);
+    deleter = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/AnonymousMemoryCache.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/AnonymousMemoryCache.java
new file mode 100644
index 00000000000..4bc4523c1ea
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/AnonymousMemoryCache.java
@@ -0,0 +1,69 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data;
+
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+
+/**
+ */
+public final class AnonymousMemoryCache extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new AnonymousMemoryCache operation.
+   * 
+   * @param scope current scope
+   * @return a new instance of AnonymousMemoryCache
+   */
+  @Endpoint(describeByClass = true)
+  public static AnonymousMemoryCache create(Scope scope) {
+    OperationBuilder opBuilder = scope.env().opBuilder("AnonymousMemoryCache", scope.makeOpName("AnonymousMemoryCache"));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new AnonymousMemoryCache(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<?> handle() {
+    return handle;
+  }
+  
+  /**
+   */
+  public Output<?> deleter() {
+    return deleter;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "AnonymousMemoryCache";
+  
+  private Output<?> handle;
+  private Output<?> deleter;
+  
+  private AnonymousMemoryCache(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    handle = operation.output(outputIdx++);
+    deleter = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/AnonymousMultiDeviceIterator.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/AnonymousMultiDeviceIterator.java
new file mode 100644
index 00000000000..9b5ba5b7fd9
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/AnonymousMultiDeviceIterator.java
@@ -0,0 +1,96 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data;
+
+import java.util.List;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * A container for a multi device iterator resource.
+ */
+public final class AnonymousMultiDeviceIterator extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new AnonymousMultiDeviceIterator operation.
+   * 
+   * @param scope current scope
+   * @param devices 
+   * @param outputTypes 
+   * @param outputShapes 
+   * @return a new instance of AnonymousMultiDeviceIterator
+   */
+  @Endpoint(describeByClass = true)
+  public static AnonymousMultiDeviceIterator create(Scope scope, List<String> devices, List<Class<? extends TType>> outputTypes, List<Shape> outputShapes) {
+    OperationBuilder opBuilder = scope.env().opBuilder("AnonymousMultiDeviceIterator", scope.makeOpName("AnonymousMultiDeviceIterator"));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    String[] devicesArray = new String[devices.size()];
+    for (int i = 0; i < devicesArray.length; ++i) {
+      devicesArray[i] = devices.get(i);
+    }
+    opBuilder.setAttr("devices", devicesArray);
+    Class[] outputTypesArray = new Class[outputTypes.size()];
+    for (int i = 0; i < outputTypesArray.length; ++i) {
+      outputTypesArray[i] = outputTypes.get(i);
+    }
+    opBuilder.setAttr("output_types", outputTypesArray);
+    Shape[] outputShapesArray = new Shape[outputShapes.size()];
+    for (int i = 0; i < outputShapesArray.length; ++i) {
+      outputShapesArray[i] = outputShapes.get(i);
+    }
+    opBuilder.setAttr("output_shapes", outputShapesArray);
+    return new AnonymousMultiDeviceIterator(opBuilder.build());
+  }
+  
+  /**
+   * A handle to a multi device iterator that can be passed to a
+   * "MultiDeviceIteratorGetNextFromShard" op. In contrast to MultiDeviceIterator,
+   * AnonymousIterator prevents resource sharing by name, and does not keep a
+   * reference to the resource container.
+   */
+  public Output<?> handle() {
+    return handle;
+  }
+  
+  /**
+   * A variant deleter that should be passed into the op that deletes the iterator.
+   */
+  public Output<?> deleter() {
+    return deleter;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "AnonymousMultiDeviceIterator";
+  
+  private Output<?> handle;
+  private Output<?> deleter;
+  
+  private AnonymousMultiDeviceIterator(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    handle = operation.output(outputIdx++);
+    deleter = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/AssertNextDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/AssertNextDataset.java
new file mode 100644
index 00000000000..0fd73311cf3
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/AssertNextDataset.java
@@ -0,0 +1,100 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TString;
+import org.tensorflow.types.family.TType;
+
+/**
+ * A transformation that asserts which transformations happen next.
+ * <p>
+ * This transformation checks whether the camel-case names (i.e. "FlatMap", not
+ * "flat_map") of the transformations following this transformation match the list
+ * of names in the `transformations` argument. If there is a mismatch, the
+ * transformation raises an exception.
+ * <p>
+ * The check occurs when iterating over the contents of the dataset, which
+ * means that the check happens <i>after</i> any static optimizations are applied
+ * to the dataset graph.
+ */
+public final class AssertNextDataset extends RawOp implements Operand<TType> {
+  
+  /**
+   * Factory method to create a class wrapping a new AssertNextDataset operation.
+   * 
+   * @param scope current scope
+   * @param inputDataset A variant tensor representing the input dataset.
+   * `data.AssertNextDataset` passes through the outputs of its input dataset.
+   * @param transformations A `tf.string` vector `tf.Tensor` identifying the transformations that are
+   * expected to happen next.
+   * @param outputTypes 
+   * @param outputShapes 
+   * @return a new instance of AssertNextDataset
+   */
+  @Endpoint(describeByClass = true)
+  public static AssertNextDataset create(Scope scope, Operand<?> inputDataset, Operand<TString> transformations, List<Class<? extends TType>> outputTypes, List<Shape> outputShapes) {
+    OperationBuilder opBuilder = scope.env().opBuilder("AssertNextDataset", scope.makeOpName("AssertNextDataset"));
+    opBuilder.addInput(inputDataset.asOutput());
+    opBuilder.addInput(transformations.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    Class[] outputTypesArray = new Class[outputTypes.size()];
+    for (int i = 0; i < outputTypesArray.length; ++i) {
+      outputTypesArray[i] = outputTypes.get(i);
+    }
+    opBuilder.setAttr("output_types", outputTypesArray);
+    Shape[] outputShapesArray = new Shape[outputShapes.size()];
+    for (int i = 0; i < outputShapesArray.length; ++i) {
+      outputShapesArray[i] = outputShapes.get(i);
+    }
+    opBuilder.setAttr("output_shapes", outputShapesArray);
+    return new AssertNextDataset(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<?> handle() {
+    return handle;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) handle;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "AssertNextDataset";
+  
+  private Output<?> handle;
+  
+  private AssertNextDataset(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    handle = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/AutoShardDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/AutoShardDataset.java
new file mode 100644
index 00000000000..c71a1b96f5e
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/AutoShardDataset.java
@@ -0,0 +1,133 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Creates a dataset that shards the input dataset.
+ * <p>
+ * Creates a dataset that shards the input dataset by num_workers, returning a
+ * sharded dataset for the index-th worker. This attempts to automatically shard
+ * a dataset by examining the Dataset graph and inserting a shard op before the
+ * inputs to a reader Dataset (e.g. CSVDataset, TFRecordDataset).
+ * <p>
+ * This dataset will throw a NotFound error if we cannot shard the dataset
+ * automatically.
+ */
+public final class AutoShardDataset extends RawOp implements Operand<TType> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.data.AutoShardDataset}
+   */
+  public static class Options {
+    
+    /**
+     * @param autoShardPolicy 
+     */
+    public Options autoShardPolicy(Long autoShardPolicy) {
+      this.autoShardPolicy = autoShardPolicy;
+      return this;
+    }
+    
+    private Long autoShardPolicy;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new AutoShardDataset operation.
+   * 
+   * @param scope current scope
+   * @param inputDataset A variant tensor representing the input dataset.
+   * @param numWorkers A scalar representing the number of workers to distribute this dataset across.
+   * @param index A scalar representing the index of the current worker out of num_workers.
+   * @param outputTypes 
+   * @param outputShapes 
+   * @param options carries optional attributes values
+   * @return a new instance of AutoShardDataset
+   */
+  @Endpoint(describeByClass = true)
+  public static AutoShardDataset create(Scope scope, Operand<?> inputDataset, Operand<TInt64> numWorkers, Operand<TInt64> index, List<Class<? extends TType>> outputTypes, List<Shape> outputShapes, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("AutoShardDataset", scope.makeOpName("AutoShardDataset"));
+    opBuilder.addInput(inputDataset.asOutput());
+    opBuilder.addInput(numWorkers.asOutput());
+    opBuilder.addInput(index.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    Class[] outputTypesArray = new Class[outputTypes.size()];
+    for (int i = 0; i < outputTypesArray.length; ++i) {
+      outputTypesArray[i] = outputTypes.get(i);
+    }
+    opBuilder.setAttr("output_types", outputTypesArray);
+    Shape[] outputShapesArray = new Shape[outputShapes.size()];
+    for (int i = 0; i < outputShapesArray.length; ++i) {
+      outputShapesArray[i] = outputShapes.get(i);
+    }
+    opBuilder.setAttr("output_shapes", outputShapesArray);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.autoShardPolicy != null) {
+          opBuilder.setAttr("auto_shard_policy", opts.autoShardPolicy);
+        }
+      }
+    }
+    return new AutoShardDataset(opBuilder.build());
+  }
+  
+  /**
+   * @param autoShardPolicy 
+   */
+  public static Options autoShardPolicy(Long autoShardPolicy) {
+    return new Options().autoShardPolicy(autoShardPolicy);
+  }
+  
+  /**
+   */
+  public Output<?> handle() {
+    return handle;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) handle;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "AutoShardDataset";
+  
+  private Output<?> handle;
+  
+  private AutoShardDataset(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    handle = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/BatchDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/BatchDataset.java
new file mode 100644
index 00000000000..7409bdade59
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/BatchDataset.java
@@ -0,0 +1,128 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TBool;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Creates a dataset that batches `batch_size` elements from `input_dataset`.
+ */
+@Operator(group = "data")
+public final class BatchDataset extends RawOp implements Operand<TType> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.data.BatchDataset}
+   */
+  public static class Options {
+    
+    /**
+     * @param parallelCopy 
+     */
+    public Options parallelCopy(Boolean parallelCopy) {
+      this.parallelCopy = parallelCopy;
+      return this;
+    }
+    
+    private Boolean parallelCopy;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new BatchDataset operation.
+   * 
+   * @param scope current scope
+   * @param inputDataset 
+   * @param batchSize A scalar representing the number of elements to accumulate in a batch.
+   * @param dropRemainder A scalar representing whether the last batch should be dropped in case its size
+   * is smaller than desired.
+   * @param outputTypes 
+   * @param outputShapes 
+   * @param options carries optional attributes values
+   * @return a new instance of BatchDataset
+   */
+  @Endpoint(describeByClass = true)
+  public static BatchDataset create(Scope scope, Operand<?> inputDataset, Operand<TInt64> batchSize, Operand<TBool> dropRemainder, List<Class<? extends TType>> outputTypes, List<Shape> outputShapes, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("BatchDatasetV2", scope.makeOpName("BatchDataset"));
+    opBuilder.addInput(inputDataset.asOutput());
+    opBuilder.addInput(batchSize.asOutput());
+    opBuilder.addInput(dropRemainder.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    Class[] outputTypesArray = new Class[outputTypes.size()];
+    for (int i = 0; i < outputTypesArray.length; ++i) {
+      outputTypesArray[i] = outputTypes.get(i);
+    }
+    opBuilder.setAttr("output_types", outputTypesArray);
+    Shape[] outputShapesArray = new Shape[outputShapes.size()];
+    for (int i = 0; i < outputShapesArray.length; ++i) {
+      outputShapesArray[i] = outputShapes.get(i);
+    }
+    opBuilder.setAttr("output_shapes", outputShapesArray);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.parallelCopy != null) {
+          opBuilder.setAttr("parallel_copy", opts.parallelCopy);
+        }
+      }
+    }
+    return new BatchDataset(opBuilder.build());
+  }
+  
+  /**
+   * @param parallelCopy 
+   */
+  public static Options parallelCopy(Boolean parallelCopy) {
+    return new Options().parallelCopy(parallelCopy);
+  }
+  
+  /**
+   */
+  public Output<?> handle() {
+    return handle;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) handle;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "BatchDatasetV2";
+  
+  private Output<?> handle;
+  
+  private BatchDataset(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    handle = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/BytesProducedStatsDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/BytesProducedStatsDataset.java
new file mode 100644
index 00000000000..4000a238b86
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/BytesProducedStatsDataset.java
@@ -0,0 +1,89 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TString;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Records the bytes size of each element of `input_dataset` in a StatsAggregator.
+ */
+public final class BytesProducedStatsDataset extends RawOp implements Operand<TType> {
+  
+  /**
+   * Factory method to create a class wrapping a new BytesProducedStatsDataset operation.
+   * 
+   * @param scope current scope
+   * @param inputDataset 
+   * @param tag 
+   * @param outputTypes 
+   * @param outputShapes 
+   * @return a new instance of BytesProducedStatsDataset
+   */
+  @Endpoint(describeByClass = true)
+  public static BytesProducedStatsDataset create(Scope scope, Operand<?> inputDataset, Operand<TString> tag, List<Class<? extends TType>> outputTypes, List<Shape> outputShapes) {
+    OperationBuilder opBuilder = scope.env().opBuilder("BytesProducedStatsDataset", scope.makeOpName("BytesProducedStatsDataset"));
+    opBuilder.addInput(inputDataset.asOutput());
+    opBuilder.addInput(tag.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    Class[] outputTypesArray = new Class[outputTypes.size()];
+    for (int i = 0; i < outputTypesArray.length; ++i) {
+      outputTypesArray[i] = outputTypes.get(i);
+    }
+    opBuilder.setAttr("output_types", outputTypesArray);
+    Shape[] outputShapesArray = new Shape[outputShapes.size()];
+    for (int i = 0; i < outputShapesArray.length; ++i) {
+      outputShapesArray[i] = outputShapes.get(i);
+    }
+    opBuilder.setAttr("output_shapes", outputShapesArray);
+    return new BytesProducedStatsDataset(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<?> handle() {
+    return handle;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) handle;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "BytesProducedStatsDataset";
+  
+  private Output<?> handle;
+  
+  private BytesProducedStatsDataset(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    handle = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/CSVDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/CSVDataset.java
new file mode 100644
index 00000000000..f6c8f72232d
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/CSVDataset.java
@@ -0,0 +1,100 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.Operands;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TBool;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.TString;
+import org.tensorflow.types.family.TType;
+
+/**
+ */
+@Operator(group = "data")
+public final class CSVDataset extends RawOp implements Operand<TType> {
+  
+  /**
+   * Factory method to create a class wrapping a new CSVDataset operation.
+   * 
+   * @param scope current scope
+   * @param filenames 
+   * @param compressionType 
+   * @param bufferSize 
+   * @param header 
+   * @param fieldDelim 
+   * @param useQuoteDelim 
+   * @param naValue 
+   * @param selectCols 
+   * @param recordDefaults 
+   * @param outputShapes 
+   * @return a new instance of CSVDataset
+   */
+  @Endpoint(describeByClass = true)
+  public static CSVDataset create(Scope scope, Operand<TString> filenames, Operand<TString> compressionType, Operand<TInt64> bufferSize, Operand<TBool> header, Operand<TString> fieldDelim, Operand<TBool> useQuoteDelim, Operand<TString> naValue, Operand<TInt64> selectCols, Iterable<Operand<?>> recordDefaults, List<Shape> outputShapes) {
+    OperationBuilder opBuilder = scope.env().opBuilder("CSVDataset", scope.makeOpName("CSVDataset"));
+    opBuilder.addInput(filenames.asOutput());
+    opBuilder.addInput(compressionType.asOutput());
+    opBuilder.addInput(bufferSize.asOutput());
+    opBuilder.addInput(header.asOutput());
+    opBuilder.addInput(fieldDelim.asOutput());
+    opBuilder.addInput(useQuoteDelim.asOutput());
+    opBuilder.addInput(naValue.asOutput());
+    opBuilder.addInput(selectCols.asOutput());
+    opBuilder.addInputList(Operands.asOutputs(recordDefaults));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    Shape[] outputShapesArray = new Shape[outputShapes.size()];
+    for (int i = 0; i < outputShapesArray.length; ++i) {
+      outputShapesArray[i] = outputShapes.get(i);
+    }
+    opBuilder.setAttr("output_shapes", outputShapesArray);
+    return new CSVDataset(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<?> handle() {
+    return handle;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) handle;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "CSVDataset";
+  
+  private Output<?> handle;
+  
+  private CSVDataset(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    handle = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/CacheDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/CacheDataset.java
new file mode 100644
index 00000000000..1de01e0eccd
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/CacheDataset.java
@@ -0,0 +1,95 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TString;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Creates a dataset that caches elements from `input_dataset`.
+ * <p>
+ * A CacheDataset will iterate over the input_dataset, and store tensors. If the
+ * cache already exists, the cache will be used. If the cache is inappropriate
+ * (e.g. cannot be opened, contains tensors of the wrong shape / size), an error
+ * will the returned when used.
+ */
+public final class CacheDataset extends RawOp implements Operand<TType> {
+  
+  /**
+   * Factory method to create a class wrapping a new CacheDataset operation.
+   * 
+   * @param scope current scope
+   * @param inputDataset 
+   * @param filename A path on the filesystem where we should cache the dataset. Note: this
+   * will be a directory.
+   * @param outputTypes 
+   * @param outputShapes 
+   * @return a new instance of CacheDataset
+   */
+  @Endpoint(describeByClass = true)
+  public static CacheDataset create(Scope scope, Operand<?> inputDataset, Operand<TString> filename, List<Class<? extends TType>> outputTypes, List<Shape> outputShapes) {
+    OperationBuilder opBuilder = scope.env().opBuilder("CacheDataset", scope.makeOpName("CacheDataset"));
+    opBuilder.addInput(inputDataset.asOutput());
+    opBuilder.addInput(filename.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    Class[] outputTypesArray = new Class[outputTypes.size()];
+    for (int i = 0; i < outputTypesArray.length; ++i) {
+      outputTypesArray[i] = outputTypes.get(i);
+    }
+    opBuilder.setAttr("output_types", outputTypesArray);
+    Shape[] outputShapesArray = new Shape[outputShapes.size()];
+    for (int i = 0; i < outputShapesArray.length; ++i) {
+      outputShapesArray[i] = outputShapes.get(i);
+    }
+    opBuilder.setAttr("output_shapes", outputShapesArray);
+    return new CacheDataset(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<?> handle() {
+    return handle;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) handle;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "CacheDataset";
+  
+  private Output<?> handle;
+  
+  private CacheDataset(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    handle = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/CacheDatasetV2.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/CacheDatasetV2.java
new file mode 100644
index 00000000000..3bcab52b253
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/CacheDatasetV2.java
@@ -0,0 +1,90 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TString;
+import org.tensorflow.types.family.TType;
+
+/**
+ */
+public final class CacheDatasetV2 extends RawOp implements Operand<TType> {
+  
+  /**
+   * Factory method to create a class wrapping a new CacheDatasetV2 operation.
+   * 
+   * @param scope current scope
+   * @param inputDataset 
+   * @param filename 
+   * @param cache 
+   * @param outputTypes 
+   * @param outputShapes 
+   * @return a new instance of CacheDatasetV2
+   */
+  @Endpoint(describeByClass = true)
+  public static CacheDatasetV2 create(Scope scope, Operand<?> inputDataset, Operand<TString> filename, Operand<?> cache, List<Class<? extends TType>> outputTypes, List<Shape> outputShapes) {
+    OperationBuilder opBuilder = scope.env().opBuilder("CacheDatasetV2", scope.makeOpName("CacheDatasetV2"));
+    opBuilder.addInput(inputDataset.asOutput());
+    opBuilder.addInput(filename.asOutput());
+    opBuilder.addInput(cache.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    Class[] outputTypesArray = new Class[outputTypes.size()];
+    for (int i = 0; i < outputTypesArray.length; ++i) {
+      outputTypesArray[i] = outputTypes.get(i);
+    }
+    opBuilder.setAttr("output_types", outputTypesArray);
+    Shape[] outputShapesArray = new Shape[outputShapes.size()];
+    for (int i = 0; i < outputShapesArray.length; ++i) {
+      outputShapesArray[i] = outputShapes.get(i);
+    }
+    opBuilder.setAttr("output_shapes", outputShapesArray);
+    return new CacheDatasetV2(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<?> handle() {
+    return handle;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) handle;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "CacheDatasetV2";
+  
+  private Output<?> handle;
+  
+  private CacheDatasetV2(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    handle = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ChooseFastestDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ChooseFastestDataset.java
new file mode 100644
index 00000000000..74f34bb8023
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ChooseFastestDataset.java
@@ -0,0 +1,88 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.Operands;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ */
+public final class ChooseFastestDataset extends RawOp implements Operand<TType> {
+  
+  /**
+   * Factory method to create a class wrapping a new ChooseFastestDataset operation.
+   * 
+   * @param scope current scope
+   * @param inputDatasets 
+   * @param numExperiments 
+   * @param outputTypes 
+   * @param outputShapes 
+   * @return a new instance of ChooseFastestDataset
+   */
+  @Endpoint(describeByClass = true)
+  public static ChooseFastestDataset create(Scope scope, Iterable<Operand<?>> inputDatasets, Long numExperiments, List<Class<? extends TType>> outputTypes, List<Shape> outputShapes) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ChooseFastestDataset", scope.makeOpName("ChooseFastestDataset"));
+    opBuilder.addInputList(Operands.asOutputs(inputDatasets));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("num_experiments", numExperiments);
+    Class[] outputTypesArray = new Class[outputTypes.size()];
+    for (int i = 0; i < outputTypesArray.length; ++i) {
+      outputTypesArray[i] = outputTypes.get(i);
+    }
+    opBuilder.setAttr("output_types", outputTypesArray);
+    Shape[] outputShapesArray = new Shape[outputShapes.size()];
+    for (int i = 0; i < outputShapesArray.length; ++i) {
+      outputShapesArray[i] = outputShapes.get(i);
+    }
+    opBuilder.setAttr("output_shapes", outputShapesArray);
+    return new ChooseFastestDataset(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<?> handle() {
+    return handle;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) handle;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ChooseFastestDataset";
+  
+  private Output<?> handle;
+  
+  private ChooseFastestDataset(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    handle = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ConcatenateDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ConcatenateDataset.java
new file mode 100644
index 00000000000..34326835089
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ConcatenateDataset.java
@@ -0,0 +1,89 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Creates a dataset that concatenates `input_dataset` with `another_dataset`.
+ */
+@Operator(group = "data")
+public final class ConcatenateDataset extends RawOp implements Operand<TType> {
+  
+  /**
+   * Factory method to create a class wrapping a new ConcatenateDataset operation.
+   * 
+   * @param scope current scope
+   * @param inputDataset 
+   * @param anotherDataset 
+   * @param outputTypes 
+   * @param outputShapes 
+   * @return a new instance of ConcatenateDataset
+   */
+  @Endpoint(describeByClass = true)
+  public static ConcatenateDataset create(Scope scope, Operand<?> inputDataset, Operand<?> anotherDataset, List<Class<? extends TType>> outputTypes, List<Shape> outputShapes) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ConcatenateDataset", scope.makeOpName("ConcatenateDataset"));
+    opBuilder.addInput(inputDataset.asOutput());
+    opBuilder.addInput(anotherDataset.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    Class[] outputTypesArray = new Class[outputTypes.size()];
+    for (int i = 0; i < outputTypesArray.length; ++i) {
+      outputTypesArray[i] = outputTypes.get(i);
+    }
+    opBuilder.setAttr("output_types", outputTypesArray);
+    Shape[] outputShapesArray = new Shape[outputShapes.size()];
+    for (int i = 0; i < outputShapesArray.length; ++i) {
+      outputShapesArray[i] = outputShapes.get(i);
+    }
+    opBuilder.setAttr("output_shapes", outputShapesArray);
+    return new ConcatenateDataset(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<?> handle() {
+    return handle;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) handle;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ConcatenateDataset";
+  
+  private Output<?> handle;
+  
+  private ConcatenateDataset(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    handle = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DatasetCardinality.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DatasetCardinality.java
new file mode 100644
index 00000000000..d851c211420
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DatasetCardinality.java
@@ -0,0 +1,75 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt64;
+
+/**
+ * Returns the cardinality of `input_dataset`.
+ * <p>
+ * Returns the cardinality of `input_dataset`.
+ */
+public final class DatasetCardinality extends RawOp implements Operand<TInt64> {
+  
+  /**
+   * Factory method to create a class wrapping a new DatasetCardinality operation.
+   * 
+   * @param scope current scope
+   * @param inputDataset A variant tensor representing the dataset to return cardinality for.
+   * @return a new instance of DatasetCardinality
+   */
+  @Endpoint(describeByClass = true)
+  public static DatasetCardinality create(Scope scope, Operand<?> inputDataset) {
+    OperationBuilder opBuilder = scope.env().opBuilder("DatasetCardinality", scope.makeOpName("DatasetCardinality"));
+    opBuilder.addInput(inputDataset.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new DatasetCardinality(opBuilder.build());
+  }
+  
+  /**
+   * The cardinality of `input_dataset`. Named constants are used to represent
+   * infinite and unknown cardinality.
+   */
+  public Output<TInt64> cardinality() {
+    return cardinality;
+  }
+  
+  @Override
+  public Output<TInt64> asOutput() {
+    return cardinality;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "DatasetCardinality";
+  
+  private Output<TInt64> cardinality;
+  
+  private DatasetCardinality(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    cardinality = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DatasetFromGraph.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DatasetFromGraph.java
new file mode 100644
index 00000000000..6223baa7251
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DatasetFromGraph.java
@@ -0,0 +1,76 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TString;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Creates a dataset from the given `graph_def`.
+ * <p>
+ * Creates a dataset from the provided `graph_def`.
+ */
+public final class DatasetFromGraph extends RawOp implements Operand<TType> {
+  
+  /**
+   * Factory method to create a class wrapping a new DatasetFromGraph operation.
+   * 
+   * @param scope current scope
+   * @param graphDef The graph representation of the dataset (as serialized GraphDef).
+   * @return a new instance of DatasetFromGraph
+   */
+  @Endpoint(describeByClass = true)
+  public static DatasetFromGraph create(Scope scope, Operand<TString> graphDef) {
+    OperationBuilder opBuilder = scope.env().opBuilder("DatasetFromGraph", scope.makeOpName("DatasetFromGraph"));
+    opBuilder.addInput(graphDef.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new DatasetFromGraph(opBuilder.build());
+  }
+  
+  /**
+   * A variant tensor representing the dataset.
+   */
+  public Output<?> handle() {
+    return handle;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) handle;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "DatasetFromGraph";
+  
+  private Output<?> handle;
+  
+  private DatasetFromGraph(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    handle = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DatasetToGraph.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DatasetToGraph.java
new file mode 100644
index 00000000000..afaab68330a
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DatasetToGraph.java
@@ -0,0 +1,127 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TString;
+
+/**
+ * Returns a serialized GraphDef representing `input_dataset`.
+ * <p>
+ * Returns a graph representation for `input_dataset`.
+ */
+public final class DatasetToGraph extends RawOp implements Operand<TString> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.data.DatasetToGraph}
+   */
+  public static class Options {
+    
+    /**
+     * @param externalStatePolicy 
+     */
+    public Options externalStatePolicy(Long externalStatePolicy) {
+      this.externalStatePolicy = externalStatePolicy;
+      return this;
+    }
+    
+    /**
+     * @param stripDeviceAssignment 
+     */
+    public Options stripDeviceAssignment(Boolean stripDeviceAssignment) {
+      this.stripDeviceAssignment = stripDeviceAssignment;
+      return this;
+    }
+    
+    private Long externalStatePolicy;
+    private Boolean stripDeviceAssignment;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new DatasetToGraph operation.
+   * 
+   * @param scope current scope
+   * @param inputDataset A variant tensor representing the dataset to return the graph representation for.
+   * @param options carries optional attributes values
+   * @return a new instance of DatasetToGraph
+   */
+  @Endpoint(describeByClass = true)
+  public static DatasetToGraph create(Scope scope, Operand<?> inputDataset, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("DatasetToGraphV2", scope.makeOpName("DatasetToGraph"));
+    opBuilder.addInput(inputDataset.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.externalStatePolicy != null) {
+          opBuilder.setAttr("external_state_policy", opts.externalStatePolicy);
+        }
+        if (opts.stripDeviceAssignment != null) {
+          opBuilder.setAttr("strip_device_assignment", opts.stripDeviceAssignment);
+        }
+      }
+    }
+    return new DatasetToGraph(opBuilder.build());
+  }
+  
+  /**
+   * @param externalStatePolicy 
+   */
+  public static Options externalStatePolicy(Long externalStatePolicy) {
+    return new Options().externalStatePolicy(externalStatePolicy);
+  }
+  
+  /**
+   * @param stripDeviceAssignment 
+   */
+  public static Options stripDeviceAssignment(Boolean stripDeviceAssignment) {
+    return new Options().stripDeviceAssignment(stripDeviceAssignment);
+  }
+  
+  /**
+   * The graph representation of the dataset (as serialized GraphDef).
+   */
+  public Output<TString> graph() {
+    return graph;
+  }
+  
+  @Override
+  public Output<TString> asOutput() {
+    return graph;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "DatasetToGraphV2";
+  
+  private Output<TString> graph;
+  
+  private DatasetToGraph(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    graph = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DatasetToSingleElement.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DatasetToSingleElement.java
new file mode 100644
index 00000000000..64a20123b56
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DatasetToSingleElement.java
@@ -0,0 +1,91 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data;
+
+import java.util.Arrays;
+import java.util.Iterator;
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Outputs the single element from the given dataset.
+ */
+public final class DatasetToSingleElement extends RawOp implements Iterable<Operand<TType>> {
+  
+  /**
+   * Factory method to create a class wrapping a new DatasetToSingleElement operation.
+   * 
+   * @param scope current scope
+   * @param dataset A handle to a dataset that contains a single element.
+   * @param outputTypes 
+   * @param outputShapes 
+   * @return a new instance of DatasetToSingleElement
+   */
+  @Endpoint(describeByClass = true)
+  public static DatasetToSingleElement create(Scope scope, Operand<?> dataset, List<Class<? extends TType>> outputTypes, List<Shape> outputShapes) {
+    OperationBuilder opBuilder = scope.env().opBuilder("DatasetToSingleElement", scope.makeOpName("DatasetToSingleElement"));
+    opBuilder.addInput(dataset.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    Class[] outputTypesArray = new Class[outputTypes.size()];
+    for (int i = 0; i < outputTypesArray.length; ++i) {
+      outputTypesArray[i] = outputTypes.get(i);
+    }
+    opBuilder.setAttr("output_types", outputTypesArray);
+    Shape[] outputShapesArray = new Shape[outputShapes.size()];
+    for (int i = 0; i < outputShapesArray.length; ++i) {
+      outputShapesArray[i] = outputShapes.get(i);
+    }
+    opBuilder.setAttr("output_shapes", outputShapesArray);
+    return new DatasetToSingleElement(opBuilder.build());
+  }
+  
+  /**
+   * The components of the single element of `input`.
+   */
+  public List<Output<?>> components() {
+    return components;
+  }
+  
+  @Override
+  @SuppressWarnings({"rawtypes", "unchecked"})
+  public Iterator<Operand<TType>> iterator() {
+    return (Iterator) components.iterator();
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "DatasetToSingleElement";
+  
+  private List<Output<?>> components;
+  
+  private DatasetToSingleElement(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    int componentsLength = operation.outputListLength("components");
+    components = Arrays.asList(operation.outputList(outputIdx, componentsLength));
+    outputIdx += componentsLength;
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DatasetToTfRecord.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DatasetToTfRecord.java
new file mode 100644
index 00000000000..114e11074dc
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DatasetToTfRecord.java
@@ -0,0 +1,60 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TString;
+
+/**
+ * Writes the given dataset to the given file using the TFRecord format.
+ */
+public final class DatasetToTfRecord extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new DatasetToTfRecord operation.
+   * 
+   * @param scope current scope
+   * @param inputDataset A variant tensor representing the dataset to write.
+   * @param filename A scalar string tensor representing the filename to use.
+   * @param compressionType A scalar string tensor containing either (i) the empty string (no
+   * compression), (ii) "ZLIB", or (iii) "GZIP".
+   * @return a new instance of DatasetToTfRecord
+   */
+  @Endpoint(describeByClass = true)
+  public static DatasetToTfRecord create(Scope scope, Operand<?> inputDataset, Operand<TString> filename, Operand<TString> compressionType) {
+    OperationBuilder opBuilder = scope.env().opBuilder("DatasetToTFRecord", scope.makeOpName("DatasetToTfRecord"));
+    opBuilder.addInput(inputDataset.asOutput());
+    opBuilder.addInput(filename.asOutput());
+    opBuilder.addInput(compressionType.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new DatasetToTfRecord(opBuilder.build());
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "DatasetToTFRecord";
+  
+  private DatasetToTfRecord(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DeleteIterator.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DeleteIterator.java
new file mode 100644
index 00000000000..69f3af096bb
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DeleteIterator.java
@@ -0,0 +1,57 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+
+/**
+ * A container for an iterator resource.
+ */
+@Operator(group = "data")
+public final class DeleteIterator extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new DeleteIterator operation.
+   * 
+   * @param scope current scope
+   * @param handle A handle to the iterator to delete.
+   * @param deleter A variant deleter.
+   * @return a new instance of DeleteIterator
+   */
+  @Endpoint(describeByClass = true)
+  public static DeleteIterator create(Scope scope, Operand<?> handle, Operand<?> deleter) {
+    OperationBuilder opBuilder = scope.env().opBuilder("DeleteIterator", scope.makeOpName("DeleteIterator"));
+    opBuilder.addInput(handle.asOutput());
+    opBuilder.addInput(deleter.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new DeleteIterator(opBuilder.build());
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "DeleteIterator";
+  
+  private DeleteIterator(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DeleteMemoryCache.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DeleteMemoryCache.java
new file mode 100644
index 00000000000..21c33030b66
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DeleteMemoryCache.java
@@ -0,0 +1,55 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+
+/**
+ */
+public final class DeleteMemoryCache extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new DeleteMemoryCache operation.
+   * 
+   * @param scope current scope
+   * @param handle 
+   * @param deleter 
+   * @return a new instance of DeleteMemoryCache
+   */
+  @Endpoint(describeByClass = true)
+  public static DeleteMemoryCache create(Scope scope, Operand<?> handle, Operand<?> deleter) {
+    OperationBuilder opBuilder = scope.env().opBuilder("DeleteMemoryCache", scope.makeOpName("DeleteMemoryCache"));
+    opBuilder.addInput(handle.asOutput());
+    opBuilder.addInput(deleter.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new DeleteMemoryCache(opBuilder.build());
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "DeleteMemoryCache";
+  
+  private DeleteMemoryCache(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DeleteMultiDeviceIterator.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DeleteMultiDeviceIterator.java
new file mode 100644
index 00000000000..966d6a7dbf1
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DeleteMultiDeviceIterator.java
@@ -0,0 +1,59 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.Operands;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+
+/**
+ * A container for an iterator resource.
+ */
+public final class DeleteMultiDeviceIterator extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new DeleteMultiDeviceIterator operation.
+   * 
+   * @param scope current scope
+   * @param multiDeviceIterator A handle to the multi device iterator to delete.
+   * @param iterators A list of iterator handles (unused). This is added so that automatic control dependencies get added during function tracing that ensure this op runs after all the dependent iterators are deleted.
+   * @param deleter A variant deleter.
+   * @return a new instance of DeleteMultiDeviceIterator
+   */
+  @Endpoint(describeByClass = true)
+  public static DeleteMultiDeviceIterator create(Scope scope, Operand<?> multiDeviceIterator, Iterable<Operand<?>> iterators, Operand<?> deleter) {
+    OperationBuilder opBuilder = scope.env().opBuilder("DeleteMultiDeviceIterator", scope.makeOpName("DeleteMultiDeviceIterator"));
+    opBuilder.addInput(multiDeviceIterator.asOutput());
+    opBuilder.addInputList(Operands.asOutputs(iterators));
+    opBuilder.addInput(deleter.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new DeleteMultiDeviceIterator(opBuilder.build());
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "DeleteMultiDeviceIterator";
+  
+  private DeleteMultiDeviceIterator(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DenseToSparseBatchDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DenseToSparseBatchDataset.java
new file mode 100644
index 00000000000..1882f3e8beb
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DenseToSparseBatchDataset.java
@@ -0,0 +1,94 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Creates a dataset that batches input elements into a SparseTensor.
+ */
+public final class DenseToSparseBatchDataset extends RawOp implements Operand<TType> {
+  
+  /**
+   * Factory method to create a class wrapping a new DenseToSparseBatchDataset operation.
+   * 
+   * @param scope current scope
+   * @param inputDataset A handle to an input dataset. Must have a single component.
+   * @param batchSize A scalar representing the number of elements to accumulate in a
+   * batch.
+   * @param rowShape A vector representing the dense shape of each row in the produced
+   * SparseTensor. The shape may be partially specified, using `-1` to indicate
+   * that a particular dimension should use the maximum size of all batch elements.
+   * @param outputTypes 
+   * @param outputShapes 
+   * @return a new instance of DenseToSparseBatchDataset
+   */
+  @Endpoint(describeByClass = true)
+  public static DenseToSparseBatchDataset create(Scope scope, Operand<?> inputDataset, Operand<TInt64> batchSize, Operand<TInt64> rowShape, List<Class<? extends TType>> outputTypes, List<Shape> outputShapes) {
+    OperationBuilder opBuilder = scope.env().opBuilder("DenseToSparseBatchDataset", scope.makeOpName("DenseToSparseBatchDataset"));
+    opBuilder.addInput(inputDataset.asOutput());
+    opBuilder.addInput(batchSize.asOutput());
+    opBuilder.addInput(rowShape.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    Class[] outputTypesArray = new Class[outputTypes.size()];
+    for (int i = 0; i < outputTypesArray.length; ++i) {
+      outputTypesArray[i] = outputTypes.get(i);
+    }
+    opBuilder.setAttr("output_types", outputTypesArray);
+    Shape[] outputShapesArray = new Shape[outputShapes.size()];
+    for (int i = 0; i < outputShapesArray.length; ++i) {
+      outputShapesArray[i] = outputShapes.get(i);
+    }
+    opBuilder.setAttr("output_shapes", outputShapesArray);
+    return new DenseToSparseBatchDataset(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<?> handle() {
+    return handle;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) handle;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "DenseToSparseBatchDataset";
+  
+  private Output<?> handle;
+  
+  private DenseToSparseBatchDataset(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    handle = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DeserializeIterator.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DeserializeIterator.java
new file mode 100644
index 00000000000..4f772fd5028
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DeserializeIterator.java
@@ -0,0 +1,58 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+
+/**
+ * Converts the given variant tensor to an iterator and stores it in the given resource.
+ */
+@Operator(group = "data")
+public final class DeserializeIterator extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new DeserializeIterator operation.
+   * 
+   * @param scope current scope
+   * @param resourceHandle A handle to an iterator resource.
+   * @param serialized A variant tensor storing the state of the iterator contained in the
+   * resource.
+   * @return a new instance of DeserializeIterator
+   */
+  @Endpoint(describeByClass = true)
+  public static DeserializeIterator create(Scope scope, Operand<?> resourceHandle, Operand<?> serialized) {
+    OperationBuilder opBuilder = scope.env().opBuilder("DeserializeIterator", scope.makeOpName("DeserializeIterator"));
+    opBuilder.addInput(resourceHandle.asOutput());
+    opBuilder.addInput(serialized.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new DeserializeIterator(opBuilder.build());
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "DeserializeIterator";
+  
+  private DeserializeIterator(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DirectedInterleaveDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DirectedInterleaveDataset.java
new file mode 100644
index 00000000000..4bf4ecdf141
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/DirectedInterleaveDataset.java
@@ -0,0 +1,91 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.Operands;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * A substitute for `InterleaveDataset` on a fixed list of `N` datasets.
+ */
+public final class DirectedInterleaveDataset extends RawOp implements Operand<TType> {
+  
+  /**
+   * Factory method to create a class wrapping a new DirectedInterleaveDataset operation.
+   * 
+   * @param scope current scope
+   * @param selectorInputDataset A dataset of scalar `DT_INT64` elements that determines which of the
+   * `N` data inputs should produce the next output element.
+   * @param dataInputDatasets `N` datasets with the same type that will be interleaved according to
+   * the values of `selector_input_dataset`.
+   * @param outputTypes 
+   * @param outputShapes 
+   * @return a new instance of DirectedInterleaveDataset
+   */
+  @Endpoint(describeByClass = true)
+  public static DirectedInterleaveDataset create(Scope scope, Operand<?> selectorInputDataset, Iterable<Operand<?>> dataInputDatasets, List<Class<? extends TType>> outputTypes, List<Shape> outputShapes) {
+    OperationBuilder opBuilder = scope.env().opBuilder("DirectedInterleaveDataset", scope.makeOpName("DirectedInterleaveDataset"));
+    opBuilder.addInput(selectorInputDataset.asOutput());
+    opBuilder.addInputList(Operands.asOutputs(dataInputDatasets));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    Class[] outputTypesArray = new Class[outputTypes.size()];
+    for (int i = 0; i < outputTypesArray.length; ++i) {
+      outputTypesArray[i] = outputTypes.get(i);
+    }
+    opBuilder.setAttr("output_types", outputTypesArray);
+    Shape[] outputShapesArray = new Shape[outputShapes.size()];
+    for (int i = 0; i < outputShapesArray.length; ++i) {
+      outputShapesArray[i] = outputShapes.get(i);
+    }
+    opBuilder.setAttr("output_shapes", outputShapesArray);
+    return new DirectedInterleaveDataset(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<?> handle() {
+    return handle;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) handle;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "DirectedInterleaveDataset";
+  
+  private Output<?> handle;
+  
+  private DirectedInterleaveDataset(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    handle = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/FilterByLastComponentDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/FilterByLastComponentDataset.java
new file mode 100644
index 00000000000..ea090e46fc6
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/FilterByLastComponentDataset.java
@@ -0,0 +1,86 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Creates a dataset containing elements of first component of `input_dataset` having true in the last component.
+ */
+public final class FilterByLastComponentDataset extends RawOp implements Operand<TType> {
+  
+  /**
+   * Factory method to create a class wrapping a new FilterByLastComponentDataset operation.
+   * 
+   * @param scope current scope
+   * @param inputDataset 
+   * @param outputTypes 
+   * @param outputShapes 
+   * @return a new instance of FilterByLastComponentDataset
+   */
+  @Endpoint(describeByClass = true)
+  public static FilterByLastComponentDataset create(Scope scope, Operand<?> inputDataset, List<Class<? extends TType>> outputTypes, List<Shape> outputShapes) {
+    OperationBuilder opBuilder = scope.env().opBuilder("FilterByLastComponentDataset", scope.makeOpName("FilterByLastComponentDataset"));
+    opBuilder.addInput(inputDataset.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    Class[] outputTypesArray = new Class[outputTypes.size()];
+    for (int i = 0; i < outputTypesArray.length; ++i) {
+      outputTypesArray[i] = outputTypes.get(i);
+    }
+    opBuilder.setAttr("output_types", outputTypesArray);
+    Shape[] outputShapesArray = new Shape[outputShapes.size()];
+    for (int i = 0; i < outputShapesArray.length; ++i) {
+      outputShapesArray[i] = outputShapes.get(i);
+    }
+    opBuilder.setAttr("output_shapes", outputShapesArray);
+    return new FilterByLastComponentDataset(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<?> output() {
+    return output;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "FilterByLastComponentDataset";
+  
+  private Output<?> output;
+  
+  private FilterByLastComponentDataset(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/FixedLengthRecordDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/FixedLengthRecordDataset.java
new file mode 100644
index 00000000000..d1bfa77177b
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/FixedLengthRecordDataset.java
@@ -0,0 +1,83 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.TString;
+import org.tensorflow.types.family.TType;
+
+/**
+ */
+public final class FixedLengthRecordDataset extends RawOp implements Operand<TType> {
+  
+  /**
+   * Factory method to create a class wrapping a new FixedLengthRecordDataset operation.
+   * 
+   * @param scope current scope
+   * @param filenames 
+   * @param headerBytes 
+   * @param recordBytes 
+   * @param footerBytes 
+   * @param bufferSize 
+   * @param compressionType 
+   * @return a new instance of FixedLengthRecordDataset
+   */
+  @Endpoint(describeByClass = true)
+  public static FixedLengthRecordDataset create(Scope scope, Operand<TString> filenames, Operand<TInt64> headerBytes, Operand<TInt64> recordBytes, Operand<TInt64> footerBytes, Operand<TInt64> bufferSize, Operand<TString> compressionType) {
+    OperationBuilder opBuilder = scope.env().opBuilder("FixedLengthRecordDatasetV2", scope.makeOpName("FixedLengthRecordDataset"));
+    opBuilder.addInput(filenames.asOutput());
+    opBuilder.addInput(headerBytes.asOutput());
+    opBuilder.addInput(recordBytes.asOutput());
+    opBuilder.addInput(footerBytes.asOutput());
+    opBuilder.addInput(bufferSize.asOutput());
+    opBuilder.addInput(compressionType.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new FixedLengthRecordDataset(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<?> handle() {
+    return handle;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) handle;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "FixedLengthRecordDatasetV2";
+  
+  private Output<?> handle;
+  
+  private FixedLengthRecordDataset(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    handle = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/IgnoreErrorsDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/IgnoreErrorsDataset.java
new file mode 100644
index 00000000000..08aaf2228a1
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/IgnoreErrorsDataset.java
@@ -0,0 +1,86 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Creates a dataset that contains the elements of `input_dataset` ignoring errors.
+ */
+public final class IgnoreErrorsDataset extends RawOp implements Operand<TType> {
+  
+  /**
+   * Factory method to create a class wrapping a new IgnoreErrorsDataset operation.
+   * 
+   * @param scope current scope
+   * @param inputDataset 
+   * @param outputTypes 
+   * @param outputShapes 
+   * @return a new instance of IgnoreErrorsDataset
+   */
+  @Endpoint(describeByClass = true)
+  public static IgnoreErrorsDataset create(Scope scope, Operand<?> inputDataset, List<Class<? extends TType>> outputTypes, List<Shape> outputShapes) {
+    OperationBuilder opBuilder = scope.env().opBuilder("IgnoreErrorsDataset", scope.makeOpName("IgnoreErrorsDataset"));
+    opBuilder.addInput(inputDataset.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    Class[] outputTypesArray = new Class[outputTypes.size()];
+    for (int i = 0; i < outputTypesArray.length; ++i) {
+      outputTypesArray[i] = outputTypes.get(i);
+    }
+    opBuilder.setAttr("output_types", outputTypesArray);
+    Shape[] outputShapesArray = new Shape[outputShapes.size()];
+    for (int i = 0; i < outputShapesArray.length; ++i) {
+      outputShapesArray[i] = outputShapes.get(i);
+    }
+    opBuilder.setAttr("output_shapes", outputShapesArray);
+    return new IgnoreErrorsDataset(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<?> handle() {
+    return handle;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) handle;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "IgnoreErrorsDataset";
+  
+  private Output<?> handle;
+  
+  private IgnoreErrorsDataset(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    handle = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/InitializeTableFromDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/InitializeTableFromDataset.java
new file mode 100644
index 00000000000..05a263a4ec1
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/InitializeTableFromDataset.java
@@ -0,0 +1,55 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+
+/**
+ */
+public final class InitializeTableFromDataset extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new InitializeTableFromDataset operation.
+   * 
+   * @param scope current scope
+   * @param tableHandle 
+   * @param dataset 
+   * @return a new instance of InitializeTableFromDataset
+   */
+  @Endpoint(describeByClass = true)
+  public static InitializeTableFromDataset create(Scope scope, Operand<?> tableHandle, Operand<?> dataset) {
+    OperationBuilder opBuilder = scope.env().opBuilder("InitializeTableFromDataset", scope.makeOpName("InitializeTableFromDataset"));
+    opBuilder.addInput(tableHandle.asOutput());
+    opBuilder.addInput(dataset.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new InitializeTableFromDataset(opBuilder.build());
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "InitializeTableFromDataset";
+  
+  private InitializeTableFromDataset(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/Iterator.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/Iterator.java
new file mode 100644
index 00000000000..860c089fd40
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/Iterator.java
@@ -0,0 +1,88 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ */
+@Operator(group = "data")
+public final class Iterator extends RawOp implements Operand<TType> {
+  
+  /**
+   * Factory method to create a class wrapping a new Iterator operation.
+   * 
+   * @param scope current scope
+   * @param sharedName 
+   * @param container 
+   * @param outputTypes 
+   * @param outputShapes 
+   * @return a new instance of Iterator
+   */
+  @Endpoint(describeByClass = true)
+  public static Iterator create(Scope scope, String sharedName, String container, List<Class<? extends TType>> outputTypes, List<Shape> outputShapes) {
+    OperationBuilder opBuilder = scope.env().opBuilder("IteratorV2", scope.makeOpName("Iterator"));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("shared_name", sharedName);
+    opBuilder.setAttr("container", container);
+    Class[] outputTypesArray = new Class[outputTypes.size()];
+    for (int i = 0; i < outputTypesArray.length; ++i) {
+      outputTypesArray[i] = outputTypes.get(i);
+    }
+    opBuilder.setAttr("output_types", outputTypesArray);
+    Shape[] outputShapesArray = new Shape[outputShapes.size()];
+    for (int i = 0; i < outputShapesArray.length; ++i) {
+      outputShapesArray[i] = outputShapes.get(i);
+    }
+    opBuilder.setAttr("output_shapes", outputShapesArray);
+    return new Iterator(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<?> handle() {
+    return handle;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) handle;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "IteratorV2";
+  
+  private Output<?> handle;
+  
+  private Iterator(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    handle = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/IteratorFromStringHandle.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/IteratorFromStringHandle.java
new file mode 100644
index 00000000000..db50d961930
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/IteratorFromStringHandle.java
@@ -0,0 +1,118 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TString;
+import org.tensorflow.types.family.TType;
+
+/**
+ */
+public final class IteratorFromStringHandle extends RawOp implements Operand<TType> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.data.IteratorFromStringHandle}
+   */
+  public static class Options {
+    
+    /**
+     * @param outputShapes 
+     */
+    public Options outputShapes(List<Shape> outputShapes) {
+      this.outputShapes = outputShapes;
+      return this;
+    }
+    
+    private List<Shape> outputShapes;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new IteratorFromStringHandle operation.
+   * 
+   * @param scope current scope
+   * @param stringHandle 
+   * @param outputTypes 
+   * @param options carries optional attributes values
+   * @return a new instance of IteratorFromStringHandle
+   */
+  @Endpoint(describeByClass = true)
+  public static IteratorFromStringHandle create(Scope scope, Operand<TString> stringHandle, List<Class<? extends TType>> outputTypes, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("IteratorFromStringHandleV2", scope.makeOpName("IteratorFromStringHandle"));
+    opBuilder.addInput(stringHandle.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    Class[] outputTypesArray = new Class[outputTypes.size()];
+    for (int i = 0; i < outputTypesArray.length; ++i) {
+      outputTypesArray[i] = outputTypes.get(i);
+    }
+    opBuilder.setAttr("output_types", outputTypesArray);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.outputShapes != null) {
+          Shape[] outputShapesArray = new Shape[opts.outputShapes.size()];
+          for (int i = 0; i < outputShapesArray.length; ++i) {
+            outputShapesArray[i] = opts.outputShapes.get(i);
+          }
+          opBuilder.setAttr("output_shapes", outputShapesArray);
+        }
+      }
+    }
+    return new IteratorFromStringHandle(opBuilder.build());
+  }
+  
+  /**
+   * @param outputShapes 
+   */
+  public static Options outputShapes(List<Shape> outputShapes) {
+    return new Options().outputShapes(outputShapes);
+  }
+  
+  /**
+   */
+  public Output<?> resourceHandle() {
+    return resourceHandle;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) resourceHandle;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "IteratorFromStringHandleV2";
+  
+  private Output<?> resourceHandle;
+  
+  private IteratorFromStringHandle(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    resourceHandle = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/IteratorGetDevice.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/IteratorGetDevice.java
new file mode 100644
index 00000000000..d2aee159583
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/IteratorGetDevice.java
@@ -0,0 +1,71 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TString;
+
+/**
+ * Returns the name of the device on which `resource` has been placed.
+ */
+public final class IteratorGetDevice extends RawOp implements Operand<TString> {
+  
+  /**
+   * Factory method to create a class wrapping a new IteratorGetDevice operation.
+   * 
+   * @param scope current scope
+   * @param resource 
+   * @return a new instance of IteratorGetDevice
+   */
+  @Endpoint(describeByClass = true)
+  public static IteratorGetDevice create(Scope scope, Operand<?> resource) {
+    OperationBuilder opBuilder = scope.env().opBuilder("IteratorGetDevice", scope.makeOpName("IteratorGetDevice"));
+    opBuilder.addInput(resource.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new IteratorGetDevice(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<TString> device() {
+    return device;
+  }
+  
+  @Override
+  public Output<TString> asOutput() {
+    return device;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "IteratorGetDevice";
+  
+  private Output<TString> device;
+  
+  private IteratorGetDevice(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    device = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/IteratorGetNext.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/IteratorGetNext.java
new file mode 100644
index 00000000000..2a02a9dc1ea
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/IteratorGetNext.java
@@ -0,0 +1,91 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data;
+
+import java.util.Arrays;
+import java.util.Iterator;
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Gets the next output from the given iterator .
+ */
+@Operator(group = "data")
+public final class IteratorGetNext extends RawOp implements Iterable<Operand<TType>> {
+  
+  /**
+   * Factory method to create a class wrapping a new IteratorGetNext operation.
+   * 
+   * @param scope current scope
+   * @param iterator 
+   * @param outputTypes 
+   * @param outputShapes 
+   * @return a new instance of IteratorGetNext
+   */
+  @Endpoint(describeByClass = true)
+  public static IteratorGetNext create(Scope scope, Operand<?> iterator, List<Class<? extends TType>> outputTypes, List<Shape> outputShapes) {
+    OperationBuilder opBuilder = scope.env().opBuilder("IteratorGetNext", scope.makeOpName("IteratorGetNext"));
+    opBuilder.addInput(iterator.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    Class[] outputTypesArray = new Class[outputTypes.size()];
+    for (int i = 0; i < outputTypesArray.length; ++i) {
+      outputTypesArray[i] = outputTypes.get(i);
+    }
+    opBuilder.setAttr("output_types", outputTypesArray);
+    Shape[] outputShapesArray = new Shape[outputShapes.size()];
+    for (int i = 0; i < outputShapesArray.length; ++i) {
+      outputShapesArray[i] = outputShapes.get(i);
+    }
+    opBuilder.setAttr("output_shapes", outputShapesArray);
+    return new IteratorGetNext(opBuilder.build());
+  }
+  
+  /**
+   */
+  public List<Output<?>> components() {
+    return components;
+  }
+  
+  @Override
+  @SuppressWarnings({"rawtypes", "unchecked"})
+  public Iterator<Operand<TType>> iterator() {
+    return (Iterator) components.iterator();
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "IteratorGetNext";
+  
+  private List<Output<?>> components;
+  
+  private IteratorGetNext(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    int componentsLength = operation.outputListLength("components");
+    components = Arrays.asList(operation.outputList(outputIdx, componentsLength));
+    outputIdx += componentsLength;
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/IteratorGetNextAsOptional.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/IteratorGetNextAsOptional.java
new file mode 100644
index 00000000000..73d16c7b56f
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/IteratorGetNextAsOptional.java
@@ -0,0 +1,87 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Gets the next output from the given iterator as an Optional variant.
+ */
+@Operator(group = "data")
+public final class IteratorGetNextAsOptional extends RawOp implements Operand<TType> {
+  
+  /**
+   * Factory method to create a class wrapping a new IteratorGetNextAsOptional operation.
+   * 
+   * @param scope current scope
+   * @param iterator 
+   * @param outputTypes 
+   * @param outputShapes 
+   * @return a new instance of IteratorGetNextAsOptional
+   */
+  @Endpoint(describeByClass = true)
+  public static IteratorGetNextAsOptional create(Scope scope, Operand<?> iterator, List<Class<? extends TType>> outputTypes, List<Shape> outputShapes) {
+    OperationBuilder opBuilder = scope.env().opBuilder("IteratorGetNextAsOptional", scope.makeOpName("IteratorGetNextAsOptional"));
+    opBuilder.addInput(iterator.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    Class[] outputTypesArray = new Class[outputTypes.size()];
+    for (int i = 0; i < outputTypesArray.length; ++i) {
+      outputTypesArray[i] = outputTypes.get(i);
+    }
+    opBuilder.setAttr("output_types", outputTypesArray);
+    Shape[] outputShapesArray = new Shape[outputShapes.size()];
+    for (int i = 0; i < outputShapesArray.length; ++i) {
+      outputShapesArray[i] = outputShapes.get(i);
+    }
+    opBuilder.setAttr("output_shapes", outputShapesArray);
+    return new IteratorGetNextAsOptional(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<?> optional() {
+    return optional;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) optional;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "IteratorGetNextAsOptional";
+  
+  private Output<?> optional;
+  
+  private IteratorGetNextAsOptional(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    optional = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/IteratorGetNextSync.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/IteratorGetNextSync.java
new file mode 100644
index 00000000000..27a088fdf67
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/IteratorGetNextSync.java
@@ -0,0 +1,96 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data;
+
+import java.util.Arrays;
+import java.util.Iterator;
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Gets the next output from the given iterator.
+ * <p>
+ * This operation is a synchronous version IteratorGetNext. It should only be used
+ * in situations where the iterator does not block the calling thread, or where
+ * the calling thread is not a member of the thread pool used to execute parallel
+ * operations (e.g. in eager mode).
+ */
+@Operator(group = "data")
+public final class IteratorGetNextSync extends RawOp implements Iterable<Operand<TType>> {
+  
+  /**
+   * Factory method to create a class wrapping a new IteratorGetNextSync operation.
+   * 
+   * @param scope current scope
+   * @param iterator 
+   * @param outputTypes 
+   * @param outputShapes 
+   * @return a new instance of IteratorGetNextSync
+   */
+  @Endpoint(describeByClass = true)
+  public static IteratorGetNextSync create(Scope scope, Operand<?> iterator, List<Class<? extends TType>> outputTypes, List<Shape> outputShapes) {
+    OperationBuilder opBuilder = scope.env().opBuilder("IteratorGetNextSync", scope.makeOpName("IteratorGetNextSync"));
+    opBuilder.addInput(iterator.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    Class[] outputTypesArray = new Class[outputTypes.size()];
+    for (int i = 0; i < outputTypesArray.length; ++i) {
+      outputTypesArray[i] = outputTypes.get(i);
+    }
+    opBuilder.setAttr("output_types", outputTypesArray);
+    Shape[] outputShapesArray = new Shape[outputShapes.size()];
+    for (int i = 0; i < outputShapesArray.length; ++i) {
+      outputShapesArray[i] = outputShapes.get(i);
+    }
+    opBuilder.setAttr("output_shapes", outputShapesArray);
+    return new IteratorGetNextSync(opBuilder.build());
+  }
+  
+  /**
+   */
+  public List<Output<?>> components() {
+    return components;
+  }
+  
+  @Override
+  @SuppressWarnings({"rawtypes", "unchecked"})
+  public Iterator<Operand<TType>> iterator() {
+    return (Iterator) components.iterator();
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "IteratorGetNextSync";
+  
+  private List<Output<?>> components;
+  
+  private IteratorGetNextSync(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    int componentsLength = operation.outputListLength("components");
+    components = Arrays.asList(operation.outputList(outputIdx, componentsLength));
+    outputIdx += componentsLength;
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/IteratorToStringHandle.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/IteratorToStringHandle.java
new file mode 100644
index 00000000000..d715d8b8f2d
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/IteratorToStringHandle.java
@@ -0,0 +1,73 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TString;
+
+/**
+ * Converts the given `resource_handle` representing an iterator to a string.
+ */
+@Operator(group = "data")
+public final class IteratorToStringHandle extends RawOp implements Operand<TString> {
+  
+  /**
+   * Factory method to create a class wrapping a new IteratorToStringHandle operation.
+   * 
+   * @param scope current scope
+   * @param resourceHandle A handle to an iterator resource.
+   * @return a new instance of IteratorToStringHandle
+   */
+  @Endpoint(describeByClass = true)
+  public static IteratorToStringHandle create(Scope scope, Operand<?> resourceHandle) {
+    OperationBuilder opBuilder = scope.env().opBuilder("IteratorToStringHandle", scope.makeOpName("IteratorToStringHandle"));
+    opBuilder.addInput(resourceHandle.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new IteratorToStringHandle(opBuilder.build());
+  }
+  
+  /**
+   * A string representation of the given handle.
+   */
+  public Output<TString> stringHandle() {
+    return stringHandle;
+  }
+  
+  @Override
+  public Output<TString> asOutput() {
+    return stringHandle;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "IteratorToStringHandle";
+  
+  private Output<TString> stringHandle;
+  
+  private IteratorToStringHandle(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    stringHandle = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/LMDBDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/LMDBDataset.java
new file mode 100644
index 00000000000..042e866f469
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/LMDBDataset.java
@@ -0,0 +1,99 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TString;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Creates a dataset that emits the key-value pairs in one or more LMDB files.
+ * <p>
+ * The Lightning Memory-Mapped Database Manager, or LMDB, is an embedded binary
+ * key-value database. This dataset can read the contents of LMDB database files,
+ * the names of which generally have the `.mdb` suffix.
+ * <p>
+ * Each output element consists of a key-value pair represented as a pair of
+ * scalar string `Tensor`s, where the first `Tensor` contains the key and the
+ * second `Tensor` contains the value.
+ * <p>
+ * LMDB uses different file formats on big- and little-endian machines.
+ * `data.LMDBDataset` can only read files in the format of the host machine.
+ */
+public final class LMDBDataset extends RawOp implements Operand<TType> {
+  
+  /**
+   * Factory method to create a class wrapping a new LMDBDataset operation.
+   * 
+   * @param scope current scope
+   * @param filenames A scalar or a vector containing the name(s) of the binary file(s) to be
+   * read.
+   * @param outputTypes 
+   * @param outputShapes 
+   * @return a new instance of LMDBDataset
+   */
+  @Endpoint(describeByClass = true)
+  public static LMDBDataset create(Scope scope, Operand<TString> filenames, List<Class<? extends TType>> outputTypes, List<Shape> outputShapes) {
+    OperationBuilder opBuilder = scope.env().opBuilder("LMDBDataset", scope.makeOpName("LMDBDataset"));
+    opBuilder.addInput(filenames.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    Class[] outputTypesArray = new Class[outputTypes.size()];
+    for (int i = 0; i < outputTypesArray.length; ++i) {
+      outputTypesArray[i] = outputTypes.get(i);
+    }
+    opBuilder.setAttr("output_types", outputTypesArray);
+    Shape[] outputShapesArray = new Shape[outputShapes.size()];
+    for (int i = 0; i < outputShapesArray.length; ++i) {
+      outputShapesArray[i] = outputShapes.get(i);
+    }
+    opBuilder.setAttr("output_shapes", outputShapesArray);
+    return new LMDBDataset(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<?> handle() {
+    return handle;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) handle;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "LMDBDataset";
+  
+  private Output<?> handle;
+  
+  private LMDBDataset(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    handle = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/LatencyStatsDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/LatencyStatsDataset.java
new file mode 100644
index 00000000000..29b9ba70579
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/LatencyStatsDataset.java
@@ -0,0 +1,89 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TString;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Records the latency of producing `input_dataset` elements in a StatsAggregator.
+ */
+public final class LatencyStatsDataset extends RawOp implements Operand<TType> {
+  
+  /**
+   * Factory method to create a class wrapping a new LatencyStatsDataset operation.
+   * 
+   * @param scope current scope
+   * @param inputDataset 
+   * @param tag 
+   * @param outputTypes 
+   * @param outputShapes 
+   * @return a new instance of LatencyStatsDataset
+   */
+  @Endpoint(describeByClass = true)
+  public static LatencyStatsDataset create(Scope scope, Operand<?> inputDataset, Operand<TString> tag, List<Class<? extends TType>> outputTypes, List<Shape> outputShapes) {
+    OperationBuilder opBuilder = scope.env().opBuilder("LatencyStatsDataset", scope.makeOpName("LatencyStatsDataset"));
+    opBuilder.addInput(inputDataset.asOutput());
+    opBuilder.addInput(tag.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    Class[] outputTypesArray = new Class[outputTypes.size()];
+    for (int i = 0; i < outputTypesArray.length; ++i) {
+      outputTypesArray[i] = outputTypes.get(i);
+    }
+    opBuilder.setAttr("output_types", outputTypesArray);
+    Shape[] outputShapesArray = new Shape[outputShapes.size()];
+    for (int i = 0; i < outputShapesArray.length; ++i) {
+      outputShapesArray[i] = outputShapes.get(i);
+    }
+    opBuilder.setAttr("output_shapes", outputShapesArray);
+    return new LatencyStatsDataset(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<?> handle() {
+    return handle;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) handle;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "LatencyStatsDataset";
+  
+  private Output<?> handle;
+  
+  private LatencyStatsDataset(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    handle = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/LeakyReluGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/LeakyReluGrad.java
new file mode 100644
index 00000000000..e0c06a9b3b6
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/LeakyReluGrad.java
@@ -0,0 +1,111 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Computes rectified linear gradients for a LeakyRelu operation.
+ * 
+ * @param <T> data type for {@code backprops()} output
+ */
+public final class LeakyReluGrad<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.data.LeakyReluGrad}
+   */
+  public static class Options {
+    
+    /**
+     * @param alpha 
+     */
+    public Options alpha(Float alpha) {
+      this.alpha = alpha;
+      return this;
+    }
+    
+    private Float alpha;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new LeakyReluGrad operation.
+   * 
+   * @param scope current scope
+   * @param gradients The backpropagated gradients to the corresponding LeakyRelu operation.
+   * @param features The features passed as input to the corresponding LeakyRelu operation,
+   * OR the outputs of that operation (both work equivalently).
+   * @param options carries optional attributes values
+   * @return a new instance of LeakyReluGrad
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> LeakyReluGrad<T> create(Scope scope, Operand<T> gradients, Operand<T> features, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("LeakyReluGrad", scope.makeOpName("LeakyReluGrad"));
+    opBuilder.addInput(gradients.asOutput());
+    opBuilder.addInput(features.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.alpha != null) {
+          opBuilder.setAttr("alpha", opts.alpha);
+        }
+      }
+    }
+    return new LeakyReluGrad<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param alpha 
+   */
+  public static Options alpha(Float alpha) {
+    return new Options().alpha(alpha);
+  }
+  
+  /**
+   * `gradients * (features > 0) + alpha * gradients * (features <= 0)`.
+   */
+  public Output<T> backprops() {
+    return backprops;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return backprops;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "LeakyReluGrad";
+  
+  private Output<T> backprops;
+  
+  private LeakyReluGrad(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    backprops = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MakeIterator.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MakeIterator.java
new file mode 100644
index 00000000000..4aace25184e
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MakeIterator.java
@@ -0,0 +1,60 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+
+/**
+ * Makes a new iterator from the given `dataset` and stores it in `iterator`.
+ * <p>
+ * This operation may be executed multiple times. Each execution will reset the
+ * iterator in `iterator` to the first element of `dataset`.
+ */
+@Operator(group = "data")
+public final class MakeIterator extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new MakeIterator operation.
+   * 
+   * @param scope current scope
+   * @param dataset 
+   * @param iterator 
+   * @return a new instance of MakeIterator
+   */
+  @Endpoint(describeByClass = true)
+  public static MakeIterator create(Scope scope, Operand<?> dataset, Operand<?> iterator) {
+    OperationBuilder opBuilder = scope.env().opBuilder("MakeIterator", scope.makeOpName("MakeIterator"));
+    opBuilder.addInput(dataset.asOutput());
+    opBuilder.addInput(iterator.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new MakeIterator(opBuilder.build());
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "MakeIterator";
+  
+  private MakeIterator(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MatchingFilesDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MatchingFilesDataset.java
new file mode 100644
index 00000000000..2c798a8d807
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MatchingFilesDataset.java
@@ -0,0 +1,72 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TString;
+import org.tensorflow.types.family.TType;
+
+/**
+ */
+public final class MatchingFilesDataset extends RawOp implements Operand<TType> {
+  
+  /**
+   * Factory method to create a class wrapping a new MatchingFilesDataset operation.
+   * 
+   * @param scope current scope
+   * @param patterns 
+   * @return a new instance of MatchingFilesDataset
+   */
+  @Endpoint(describeByClass = true)
+  public static MatchingFilesDataset create(Scope scope, Operand<TString> patterns) {
+    OperationBuilder opBuilder = scope.env().opBuilder("MatchingFilesDataset", scope.makeOpName("MatchingFilesDataset"));
+    opBuilder.addInput(patterns.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new MatchingFilesDataset(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<?> handle() {
+    return handle;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) handle;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "MatchingFilesDataset";
+  
+  private Output<?> handle;
+  
+  private MatchingFilesDataset(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    handle = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MaxIntraOpParallelismDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MaxIntraOpParallelismDataset.java
new file mode 100644
index 00000000000..390c6d281ac
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MaxIntraOpParallelismDataset.java
@@ -0,0 +1,89 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Creates a dataset that overrides the maximum intra-op parallelism.
+ */
+public final class MaxIntraOpParallelismDataset extends RawOp implements Operand<TType> {
+  
+  /**
+   * Factory method to create a class wrapping a new MaxIntraOpParallelismDataset operation.
+   * 
+   * @param scope current scope
+   * @param inputDataset 
+   * @param maxIntraOpParallelism Identifies the maximum intra-op parallelism to use.
+   * @param outputTypes 
+   * @param outputShapes 
+   * @return a new instance of MaxIntraOpParallelismDataset
+   */
+  @Endpoint(describeByClass = true)
+  public static MaxIntraOpParallelismDataset create(Scope scope, Operand<?> inputDataset, Operand<TInt64> maxIntraOpParallelism, List<Class<? extends TType>> outputTypes, List<Shape> outputShapes) {
+    OperationBuilder opBuilder = scope.env().opBuilder("MaxIntraOpParallelismDataset", scope.makeOpName("MaxIntraOpParallelismDataset"));
+    opBuilder.addInput(inputDataset.asOutput());
+    opBuilder.addInput(maxIntraOpParallelism.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    Class[] outputTypesArray = new Class[outputTypes.size()];
+    for (int i = 0; i < outputTypesArray.length; ++i) {
+      outputTypesArray[i] = outputTypes.get(i);
+    }
+    opBuilder.setAttr("output_types", outputTypesArray);
+    Shape[] outputShapesArray = new Shape[outputShapes.size()];
+    for (int i = 0; i < outputShapesArray.length; ++i) {
+      outputShapesArray[i] = outputShapes.get(i);
+    }
+    opBuilder.setAttr("output_shapes", outputShapesArray);
+    return new MaxIntraOpParallelismDataset(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<?> handle() {
+    return handle;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) handle;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "MaxIntraOpParallelismDataset";
+  
+  private Output<?> handle;
+  
+  private MaxIntraOpParallelismDataset(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    handle = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ModelDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ModelDataset.java
new file mode 100644
index 00000000000..fa1acc9e5b2
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ModelDataset.java
@@ -0,0 +1,141 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Identity transformation that models performance.
+ * <p>
+ * Identity transformation that models performance.
+ */
+public final class ModelDataset extends RawOp implements Operand<TType> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.data.ModelDataset}
+   */
+  public static class Options {
+    
+    /**
+     * @param algorithm 
+     */
+    public Options algorithm(Long algorithm) {
+      this.algorithm = algorithm;
+      return this;
+    }
+    
+    /**
+     * @param cpuBudget 
+     */
+    public Options cpuBudget(Long cpuBudget) {
+      this.cpuBudget = cpuBudget;
+      return this;
+    }
+    
+    private Long algorithm;
+    private Long cpuBudget;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new ModelDataset operation.
+   * 
+   * @param scope current scope
+   * @param inputDataset A variant tensor representing the input dataset.
+   * @param outputTypes 
+   * @param outputShapes 
+   * @param options carries optional attributes values
+   * @return a new instance of ModelDataset
+   */
+  @Endpoint(describeByClass = true)
+  public static ModelDataset create(Scope scope, Operand<?> inputDataset, List<Class<? extends TType>> outputTypes, List<Shape> outputShapes, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ModelDataset", scope.makeOpName("ModelDataset"));
+    opBuilder.addInput(inputDataset.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    Class[] outputTypesArray = new Class[outputTypes.size()];
+    for (int i = 0; i < outputTypesArray.length; ++i) {
+      outputTypesArray[i] = outputTypes.get(i);
+    }
+    opBuilder.setAttr("output_types", outputTypesArray);
+    Shape[] outputShapesArray = new Shape[outputShapes.size()];
+    for (int i = 0; i < outputShapesArray.length; ++i) {
+      outputShapesArray[i] = outputShapes.get(i);
+    }
+    opBuilder.setAttr("output_shapes", outputShapesArray);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.algorithm != null) {
+          opBuilder.setAttr("algorithm", opts.algorithm);
+        }
+        if (opts.cpuBudget != null) {
+          opBuilder.setAttr("cpu_budget", opts.cpuBudget);
+        }
+      }
+    }
+    return new ModelDataset(opBuilder.build());
+  }
+  
+  /**
+   * @param algorithm 
+   */
+  public static Options algorithm(Long algorithm) {
+    return new Options().algorithm(algorithm);
+  }
+  
+  /**
+   * @param cpuBudget 
+   */
+  public static Options cpuBudget(Long cpuBudget) {
+    return new Options().cpuBudget(cpuBudget);
+  }
+  
+  /**
+   */
+  public Output<?> handle() {
+    return handle;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) handle;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ModelDataset";
+  
+  private Output<?> handle;
+  
+  private ModelDataset(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    handle = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MultiDeviceIterator.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MultiDeviceIterator.java
new file mode 100644
index 00000000000..cb223414fec
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MultiDeviceIterator.java
@@ -0,0 +1,97 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Creates a MultiDeviceIterator resource.
+ */
+public final class MultiDeviceIterator extends RawOp implements Operand<TType> {
+  
+  /**
+   * Factory method to create a class wrapping a new MultiDeviceIterator operation.
+   * 
+   * @param scope current scope
+   * @param devices A list of devices the iterator works across.
+   * @param sharedName If non-empty, this resource will be shared under the given name
+   * across multiple sessions.
+   * @param container If non-empty, this resource is placed in the given container.
+   * Otherwise, a default container is used.
+   * @param outputTypes The type list for the return values.
+   * @param outputShapes The list of shapes being produced.
+   * @return a new instance of MultiDeviceIterator
+   */
+  @Endpoint(describeByClass = true)
+  public static MultiDeviceIterator create(Scope scope, List<String> devices, String sharedName, String container, List<Class<? extends TType>> outputTypes, List<Shape> outputShapes) {
+    OperationBuilder opBuilder = scope.env().opBuilder("MultiDeviceIterator", scope.makeOpName("MultiDeviceIterator"));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    String[] devicesArray = new String[devices.size()];
+    for (int i = 0; i < devicesArray.length; ++i) {
+      devicesArray[i] = devices.get(i);
+    }
+    opBuilder.setAttr("devices", devicesArray);
+    opBuilder.setAttr("shared_name", sharedName);
+    opBuilder.setAttr("container", container);
+    Class[] outputTypesArray = new Class[outputTypes.size()];
+    for (int i = 0; i < outputTypesArray.length; ++i) {
+      outputTypesArray[i] = outputTypes.get(i);
+    }
+    opBuilder.setAttr("output_types", outputTypesArray);
+    Shape[] outputShapesArray = new Shape[outputShapes.size()];
+    for (int i = 0; i < outputShapesArray.length; ++i) {
+      outputShapesArray[i] = outputShapes.get(i);
+    }
+    opBuilder.setAttr("output_shapes", outputShapesArray);
+    return new MultiDeviceIterator(opBuilder.build());
+  }
+  
+  /**
+   * Handle to the resource created.
+   */
+  public Output<?> handle() {
+    return handle;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) handle;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "MultiDeviceIterator";
+  
+  private Output<?> handle;
+  
+  private MultiDeviceIterator(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    handle = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MultiDeviceIteratorFromStringHandle.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MultiDeviceIteratorFromStringHandle.java
new file mode 100644
index 00000000000..af6394bb70d
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MultiDeviceIteratorFromStringHandle.java
@@ -0,0 +1,120 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TString;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Generates a MultiDeviceIterator resource from its provided string handle.
+ */
+public final class MultiDeviceIteratorFromStringHandle extends RawOp implements Operand<TType> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.data.MultiDeviceIteratorFromStringHandle}
+   */
+  public static class Options {
+    
+    /**
+     * @param outputShapes The list of shapes being produced.
+     */
+    public Options outputShapes(List<Shape> outputShapes) {
+      this.outputShapes = outputShapes;
+      return this;
+    }
+    
+    private List<Shape> outputShapes;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new MultiDeviceIteratorFromStringHandle operation.
+   * 
+   * @param scope current scope
+   * @param stringHandle String representing the resource.
+   * @param outputTypes The type list for the return values.
+   * @param options carries optional attributes values
+   * @return a new instance of MultiDeviceIteratorFromStringHandle
+   */
+  @Endpoint(describeByClass = true)
+  public static MultiDeviceIteratorFromStringHandle create(Scope scope, Operand<TString> stringHandle, List<Class<? extends TType>> outputTypes, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("MultiDeviceIteratorFromStringHandle", scope.makeOpName("MultiDeviceIteratorFromStringHandle"));
+    opBuilder.addInput(stringHandle.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    Class[] outputTypesArray = new Class[outputTypes.size()];
+    for (int i = 0; i < outputTypesArray.length; ++i) {
+      outputTypesArray[i] = outputTypes.get(i);
+    }
+    opBuilder.setAttr("output_types", outputTypesArray);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.outputShapes != null) {
+          Shape[] outputShapesArray = new Shape[opts.outputShapes.size()];
+          for (int i = 0; i < outputShapesArray.length; ++i) {
+            outputShapesArray[i] = opts.outputShapes.get(i);
+          }
+          opBuilder.setAttr("output_shapes", outputShapesArray);
+        }
+      }
+    }
+    return new MultiDeviceIteratorFromStringHandle(opBuilder.build());
+  }
+  
+  /**
+   * @param outputShapes The list of shapes being produced.
+   */
+  public static Options outputShapes(List<Shape> outputShapes) {
+    return new Options().outputShapes(outputShapes);
+  }
+  
+  /**
+   * A MultiDeviceIterator resource.
+   */
+  public Output<?> multiDeviceIterator() {
+    return multiDeviceIterator;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) multiDeviceIterator;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "MultiDeviceIteratorFromStringHandle";
+  
+  private Output<?> multiDeviceIterator;
+  
+  private MultiDeviceIteratorFromStringHandle(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    multiDeviceIterator = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MultiDeviceIteratorGetNextFromShard.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MultiDeviceIteratorGetNextFromShard.java
new file mode 100644
index 00000000000..0c0e79f7c31
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MultiDeviceIteratorGetNextFromShard.java
@@ -0,0 +1,97 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data;
+
+import java.util.Arrays;
+import java.util.Iterator;
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Gets next element for the provided shard number.
+ */
+public final class MultiDeviceIteratorGetNextFromShard extends RawOp implements Iterable<Operand<TType>> {
+  
+  /**
+   * Factory method to create a class wrapping a new MultiDeviceIteratorGetNextFromShard operation.
+   * 
+   * @param scope current scope
+   * @param multiDeviceIterator A MultiDeviceIterator resource.
+   * @param shardNum Integer representing which shard to fetch data for.
+   * @param incarnationId Which incarnation of the MultiDeviceIterator is running.
+   * @param outputTypes The type list for the return values.
+   * @param outputShapes The list of shapes being produced.
+   * @return a new instance of MultiDeviceIteratorGetNextFromShard
+   */
+  @Endpoint(describeByClass = true)
+  public static MultiDeviceIteratorGetNextFromShard create(Scope scope, Operand<?> multiDeviceIterator, Operand<TInt32> shardNum, Operand<TInt64> incarnationId, List<Class<? extends TType>> outputTypes, List<Shape> outputShapes) {
+    OperationBuilder opBuilder = scope.env().opBuilder("MultiDeviceIteratorGetNextFromShard", scope.makeOpName("MultiDeviceIteratorGetNextFromShard"));
+    opBuilder.addInput(multiDeviceIterator.asOutput());
+    opBuilder.addInput(shardNum.asOutput());
+    opBuilder.addInput(incarnationId.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    Class[] outputTypesArray = new Class[outputTypes.size()];
+    for (int i = 0; i < outputTypesArray.length; ++i) {
+      outputTypesArray[i] = outputTypes.get(i);
+    }
+    opBuilder.setAttr("output_types", outputTypesArray);
+    Shape[] outputShapesArray = new Shape[outputShapes.size()];
+    for (int i = 0; i < outputShapesArray.length; ++i) {
+      outputShapesArray[i] = outputShapes.get(i);
+    }
+    opBuilder.setAttr("output_shapes", outputShapesArray);
+    return new MultiDeviceIteratorGetNextFromShard(opBuilder.build());
+  }
+  
+  /**
+   * Result of the get_next on the dataset.
+   */
+  public List<Output<?>> components() {
+    return components;
+  }
+  
+  @Override
+  @SuppressWarnings({"rawtypes", "unchecked"})
+  public Iterator<Operand<TType>> iterator() {
+    return (Iterator) components.iterator();
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "MultiDeviceIteratorGetNextFromShard";
+  
+  private List<Output<?>> components;
+  
+  private MultiDeviceIteratorGetNextFromShard(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    int componentsLength = operation.outputListLength("components");
+    components = Arrays.asList(operation.outputList(outputIdx, componentsLength));
+    outputIdx += componentsLength;
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MultiDeviceIteratorInit.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MultiDeviceIteratorInit.java
new file mode 100644
index 00000000000..a8e957c64b5
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MultiDeviceIteratorInit.java
@@ -0,0 +1,77 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt64;
+
+/**
+ * Initializes the multi device iterator with the given dataset.
+ */
+public final class MultiDeviceIteratorInit extends RawOp implements Operand<TInt64> {
+  
+  /**
+   * Factory method to create a class wrapping a new MultiDeviceIteratorInit operation.
+   * 
+   * @param scope current scope
+   * @param dataset Dataset to be iterated upon.
+   * @param multiDeviceIterator A MultiDeviceIteratorResource.
+   * @param maxBufferSize The maximum size of the host side per device buffer to keep.
+   * @return a new instance of MultiDeviceIteratorInit
+   */
+  @Endpoint(describeByClass = true)
+  public static MultiDeviceIteratorInit create(Scope scope, Operand<?> dataset, Operand<?> multiDeviceIterator, Operand<TInt64> maxBufferSize) {
+    OperationBuilder opBuilder = scope.env().opBuilder("MultiDeviceIteratorInit", scope.makeOpName("MultiDeviceIteratorInit"));
+    opBuilder.addInput(dataset.asOutput());
+    opBuilder.addInput(multiDeviceIterator.asOutput());
+    opBuilder.addInput(maxBufferSize.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new MultiDeviceIteratorInit(opBuilder.build());
+  }
+  
+  /**
+   * An int64 indicating which incarnation of the MultiDeviceIterator
+   * is running.
+   */
+  public Output<TInt64> incarnationId() {
+    return incarnationId;
+  }
+  
+  @Override
+  public Output<TInt64> asOutput() {
+    return incarnationId;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "MultiDeviceIteratorInit";
+  
+  private Output<TInt64> incarnationId;
+  
+  private MultiDeviceIteratorInit(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    incarnationId = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MultiDeviceIteratorToStringHandle.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MultiDeviceIteratorToStringHandle.java
new file mode 100644
index 00000000000..5697c7e7699
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/MultiDeviceIteratorToStringHandle.java
@@ -0,0 +1,72 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TString;
+
+/**
+ * Produces a string handle for the given MultiDeviceIterator.
+ */
+public final class MultiDeviceIteratorToStringHandle extends RawOp implements Operand<TString> {
+  
+  /**
+   * Factory method to create a class wrapping a new MultiDeviceIteratorToStringHandle operation.
+   * 
+   * @param scope current scope
+   * @param multiDeviceIterator A MultiDeviceIterator resource.
+   * @return a new instance of MultiDeviceIteratorToStringHandle
+   */
+  @Endpoint(describeByClass = true)
+  public static MultiDeviceIteratorToStringHandle create(Scope scope, Operand<?> multiDeviceIterator) {
+    OperationBuilder opBuilder = scope.env().opBuilder("MultiDeviceIteratorToStringHandle", scope.makeOpName("MultiDeviceIteratorToStringHandle"));
+    opBuilder.addInput(multiDeviceIterator.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new MultiDeviceIteratorToStringHandle(opBuilder.build());
+  }
+  
+  /**
+   * A string representing the resource.
+   */
+  public Output<TString> stringHandle() {
+    return stringHandle;
+  }
+  
+  @Override
+  public Output<TString> asOutput() {
+    return stringHandle;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "MultiDeviceIteratorToStringHandle";
+  
+  private Output<TString> stringHandle;
+  
+  private MultiDeviceIteratorToStringHandle(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    stringHandle = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/NonSerializableDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/NonSerializableDataset.java
new file mode 100644
index 00000000000..52cdcb8ce6f
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/NonSerializableDataset.java
@@ -0,0 +1,85 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ */
+public final class NonSerializableDataset extends RawOp implements Operand<TType> {
+  
+  /**
+   * Factory method to create a class wrapping a new NonSerializableDataset operation.
+   * 
+   * @param scope current scope
+   * @param inputDataset 
+   * @param outputTypes 
+   * @param outputShapes 
+   * @return a new instance of NonSerializableDataset
+   */
+  @Endpoint(describeByClass = true)
+  public static NonSerializableDataset create(Scope scope, Operand<?> inputDataset, List<Class<? extends TType>> outputTypes, List<Shape> outputShapes) {
+    OperationBuilder opBuilder = scope.env().opBuilder("NonSerializableDataset", scope.makeOpName("NonSerializableDataset"));
+    opBuilder.addInput(inputDataset.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    Class[] outputTypesArray = new Class[outputTypes.size()];
+    for (int i = 0; i < outputTypesArray.length; ++i) {
+      outputTypesArray[i] = outputTypes.get(i);
+    }
+    opBuilder.setAttr("output_types", outputTypesArray);
+    Shape[] outputShapesArray = new Shape[outputShapes.size()];
+    for (int i = 0; i < outputShapesArray.length; ++i) {
+      outputShapesArray[i] = outputShapes.get(i);
+    }
+    opBuilder.setAttr("output_shapes", outputShapesArray);
+    return new NonSerializableDataset(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<?> handle() {
+    return handle;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) handle;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "NonSerializableDataset";
+  
+  private Output<?> handle;
+  
+  private NonSerializableDataset(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    handle = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/OptimizeDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/OptimizeDataset.java
new file mode 100644
index 00000000000..7025879a980
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/OptimizeDataset.java
@@ -0,0 +1,129 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TString;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Creates a dataset by applying optimizations to `input_dataset`.
+ * <p>
+ * Creates a dataset by applying optimizations to `input_dataset`.
+ */
+public final class OptimizeDataset extends RawOp implements Operand<TType> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.data.OptimizeDataset}
+   */
+  public static class Options {
+    
+    /**
+     * @param optimizationConfigs 
+     */
+    public Options optimizationConfigs(List<String> optimizationConfigs) {
+      this.optimizationConfigs = optimizationConfigs;
+      return this;
+    }
+    
+    private List<String> optimizationConfigs;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new OptimizeDataset operation.
+   * 
+   * @param scope current scope
+   * @param inputDataset A variant tensor representing the input dataset.
+   * @param optimizations A `tf.string` vector `tf.Tensor` identifying optimizations to use.
+   * @param outputTypes 
+   * @param outputShapes 
+   * @param options carries optional attributes values
+   * @return a new instance of OptimizeDataset
+   */
+  @Endpoint(describeByClass = true)
+  public static OptimizeDataset create(Scope scope, Operand<?> inputDataset, Operand<TString> optimizations, List<Class<? extends TType>> outputTypes, List<Shape> outputShapes, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("OptimizeDataset", scope.makeOpName("OptimizeDataset"));
+    opBuilder.addInput(inputDataset.asOutput());
+    opBuilder.addInput(optimizations.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    Class[] outputTypesArray = new Class[outputTypes.size()];
+    for (int i = 0; i < outputTypesArray.length; ++i) {
+      outputTypesArray[i] = outputTypes.get(i);
+    }
+    opBuilder.setAttr("output_types", outputTypesArray);
+    Shape[] outputShapesArray = new Shape[outputShapes.size()];
+    for (int i = 0; i < outputShapesArray.length; ++i) {
+      outputShapesArray[i] = outputShapes.get(i);
+    }
+    opBuilder.setAttr("output_shapes", outputShapesArray);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.optimizationConfigs != null) {
+          String[] optimizationConfigsArray = new String[opts.optimizationConfigs.size()];
+          for (int i = 0; i < optimizationConfigsArray.length; ++i) {
+            optimizationConfigsArray[i] = opts.optimizationConfigs.get(i);
+          }
+          opBuilder.setAttr("optimization_configs", optimizationConfigsArray);
+        }
+      }
+    }
+    return new OptimizeDataset(opBuilder.build());
+  }
+  
+  /**
+   * @param optimizationConfigs 
+   */
+  public static Options optimizationConfigs(List<String> optimizationConfigs) {
+    return new Options().optimizationConfigs(optimizationConfigs);
+  }
+  
+  /**
+   */
+  public Output<?> handle() {
+    return handle;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) handle;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "OptimizeDataset";
+  
+  private Output<?> handle;
+  
+  private OptimizeDataset(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    handle = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/OptionalFromValue.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/OptionalFromValue.java
new file mode 100644
index 00000000000..a8fc1d4eaea
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/OptionalFromValue.java
@@ -0,0 +1,74 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.Operands;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Constructs an Optional variant from a tuple of tensors.
+ */
+@Operator(group = "data")
+public final class OptionalFromValue extends RawOp implements Operand<TType> {
+  
+  /**
+   * Factory method to create a class wrapping a new OptionalFromValue operation.
+   * 
+   * @param scope current scope
+   * @param components 
+   * @return a new instance of OptionalFromValue
+   */
+  @Endpoint(describeByClass = true)
+  public static OptionalFromValue create(Scope scope, Iterable<Operand<?>> components) {
+    OperationBuilder opBuilder = scope.env().opBuilder("OptionalFromValue", scope.makeOpName("OptionalFromValue"));
+    opBuilder.addInputList(Operands.asOutputs(components));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new OptionalFromValue(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<?> optional() {
+    return optional;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) optional;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "OptionalFromValue";
+  
+  private Output<?> optional;
+  
+  private OptionalFromValue(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    optional = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/OptionalGetValue.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/OptionalGetValue.java
new file mode 100644
index 00000000000..97a6bc8aea9
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/OptionalGetValue.java
@@ -0,0 +1,91 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data;
+
+import java.util.Arrays;
+import java.util.Iterator;
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Returns the value stored in an Optional variant or raises an error if none exists.
+ */
+@Operator(group = "data")
+public final class OptionalGetValue extends RawOp implements Iterable<Operand<TType>> {
+  
+  /**
+   * Factory method to create a class wrapping a new OptionalGetValue operation.
+   * 
+   * @param scope current scope
+   * @param optional 
+   * @param outputTypes 
+   * @param outputShapes 
+   * @return a new instance of OptionalGetValue
+   */
+  @Endpoint(describeByClass = true)
+  public static OptionalGetValue create(Scope scope, Operand<?> optional, List<Class<? extends TType>> outputTypes, List<Shape> outputShapes) {
+    OperationBuilder opBuilder = scope.env().opBuilder("OptionalGetValue", scope.makeOpName("OptionalGetValue"));
+    opBuilder.addInput(optional.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    Class[] outputTypesArray = new Class[outputTypes.size()];
+    for (int i = 0; i < outputTypesArray.length; ++i) {
+      outputTypesArray[i] = outputTypes.get(i);
+    }
+    opBuilder.setAttr("output_types", outputTypesArray);
+    Shape[] outputShapesArray = new Shape[outputShapes.size()];
+    for (int i = 0; i < outputShapesArray.length; ++i) {
+      outputShapesArray[i] = outputShapes.get(i);
+    }
+    opBuilder.setAttr("output_shapes", outputShapesArray);
+    return new OptionalGetValue(opBuilder.build());
+  }
+  
+  /**
+   */
+  public List<Output<?>> components() {
+    return components;
+  }
+  
+  @Override
+  @SuppressWarnings({"rawtypes", "unchecked"})
+  public Iterator<Operand<TType>> iterator() {
+    return (Iterator) components.iterator();
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "OptionalGetValue";
+  
+  private List<Output<?>> components;
+  
+  private OptionalGetValue(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    int componentsLength = operation.outputListLength("components");
+    components = Arrays.asList(operation.outputList(outputIdx, componentsLength));
+    outputIdx += componentsLength;
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/OptionalHasValue.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/OptionalHasValue.java
new file mode 100644
index 00000000000..af5900c38a9
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/OptionalHasValue.java
@@ -0,0 +1,72 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TBool;
+
+/**
+ * Returns true if and only if the given Optional variant has a value.
+ */
+@Operator(group = "data")
+public final class OptionalHasValue extends RawOp implements Operand<TBool> {
+  
+  /**
+   * Factory method to create a class wrapping a new OptionalHasValue operation.
+   * 
+   * @param scope current scope
+   * @param optional 
+   * @return a new instance of OptionalHasValue
+   */
+  @Endpoint(describeByClass = true)
+  public static OptionalHasValue create(Scope scope, Operand<?> optional) {
+    OperationBuilder opBuilder = scope.env().opBuilder("OptionalHasValue", scope.makeOpName("OptionalHasValue"));
+    opBuilder.addInput(optional.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new OptionalHasValue(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<TBool> hasValue() {
+    return hasValue;
+  }
+  
+  @Override
+  public Output<TBool> asOutput() {
+    return hasValue;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "OptionalHasValue";
+  
+  private Output<TBool> hasValue;
+  
+  private OptionalHasValue(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    hasValue = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/OptionalNone.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/OptionalNone.java
new file mode 100644
index 00000000000..fa3deabf28f
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/OptionalNone.java
@@ -0,0 +1,71 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Creates an Optional variant with no value.
+ */
+@Operator(group = "data")
+public final class OptionalNone extends RawOp implements Operand<TType> {
+  
+  /**
+   * Factory method to create a class wrapping a new OptionalNone operation.
+   * 
+   * @param scope current scope
+   * @return a new instance of OptionalNone
+   */
+  @Endpoint(describeByClass = true)
+  public static OptionalNone create(Scope scope) {
+    OperationBuilder opBuilder = scope.env().opBuilder("OptionalNone", scope.makeOpName("OptionalNone"));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new OptionalNone(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<?> optional() {
+    return optional;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) optional;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "OptionalNone";
+  
+  private Output<?> optional;
+  
+  private OptionalNone(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    optional = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/PaddedBatchDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/PaddedBatchDataset.java
new file mode 100644
index 00000000000..0685e5e9e74
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/PaddedBatchDataset.java
@@ -0,0 +1,131 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.Operands;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TBool;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Creates a dataset that batches and pads `batch_size` elements from the input.
+ */
+public final class PaddedBatchDataset extends RawOp implements Operand<TType> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.data.PaddedBatchDataset}
+   */
+  public static class Options {
+    
+    /**
+     * @param parallelCopy 
+     */
+    public Options parallelCopy(Boolean parallelCopy) {
+      this.parallelCopy = parallelCopy;
+      return this;
+    }
+    
+    private Boolean parallelCopy;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new PaddedBatchDataset operation.
+   * 
+   * @param scope current scope
+   * @param inputDataset 
+   * @param batchSize A scalar representing the number of elements to accumulate in a
+   * batch.
+   * @param paddedShapes A list of int64 tensors representing the desired padded shapes
+   * of the corresponding output components. These shapes may be partially
+   * specified, using `-1` to indicate that a particular dimension should be
+   * padded to the maximum size of all batch elements.
+   * @param paddingValues A list of scalars containing the padding value to use for
+   * each of the outputs.
+   * @param dropRemainder A scalar representing whether the last batch should be dropped in case its size
+   * is smaller than desired.
+   * @param outputShapes 
+   * @param options carries optional attributes values
+   * @return a new instance of PaddedBatchDataset
+   */
+  @Endpoint(describeByClass = true)
+  public static PaddedBatchDataset create(Scope scope, Operand<?> inputDataset, Operand<TInt64> batchSize, Iterable<Operand<TInt64>> paddedShapes, Iterable<Operand<?>> paddingValues, Operand<TBool> dropRemainder, List<Shape> outputShapes, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("PaddedBatchDatasetV2", scope.makeOpName("PaddedBatchDataset"));
+    opBuilder.addInput(inputDataset.asOutput());
+    opBuilder.addInput(batchSize.asOutput());
+    opBuilder.addInputList(Operands.asOutputs(paddedShapes));
+    opBuilder.addInputList(Operands.asOutputs(paddingValues));
+    opBuilder.addInput(dropRemainder.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    Shape[] outputShapesArray = new Shape[outputShapes.size()];
+    for (int i = 0; i < outputShapesArray.length; ++i) {
+      outputShapesArray[i] = outputShapes.get(i);
+    }
+    opBuilder.setAttr("output_shapes", outputShapesArray);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.parallelCopy != null) {
+          opBuilder.setAttr("parallel_copy", opts.parallelCopy);
+        }
+      }
+    }
+    return new PaddedBatchDataset(opBuilder.build());
+  }
+  
+  /**
+   * @param parallelCopy 
+   */
+  public static Options parallelCopy(Boolean parallelCopy) {
+    return new Options().parallelCopy(parallelCopy);
+  }
+  
+  /**
+   */
+  public Output<?> handle() {
+    return handle;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) handle;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "PaddedBatchDatasetV2";
+  
+  private Output<?> handle;
+  
+  private PaddedBatchDataset(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    handle = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/PrefetchDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/PrefetchDataset.java
new file mode 100644
index 00000000000..0da54c538a0
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/PrefetchDataset.java
@@ -0,0 +1,143 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Creates a dataset that asynchronously prefetches elements from `input_dataset`.
+ */
+public final class PrefetchDataset extends RawOp implements Operand<TType> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.data.PrefetchDataset}
+   */
+  public static class Options {
+    
+    /**
+     * @param slackPeriod 
+     */
+    public Options slackPeriod(Long slackPeriod) {
+      this.slackPeriod = slackPeriod;
+      return this;
+    }
+    
+    /**
+     * @param legacyAutotune 
+     */
+    public Options legacyAutotune(Boolean legacyAutotune) {
+      this.legacyAutotune = legacyAutotune;
+      return this;
+    }
+    
+    private Long slackPeriod;
+    private Boolean legacyAutotune;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new PrefetchDataset operation.
+   * 
+   * @param scope current scope
+   * @param inputDataset 
+   * @param bufferSize The maximum number of elements to buffer in an iterator over
+   * this dataset.
+   * @param outputTypes 
+   * @param outputShapes 
+   * @param options carries optional attributes values
+   * @return a new instance of PrefetchDataset
+   */
+  @Endpoint(describeByClass = true)
+  public static PrefetchDataset create(Scope scope, Operand<?> inputDataset, Operand<TInt64> bufferSize, List<Class<? extends TType>> outputTypes, List<Shape> outputShapes, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("PrefetchDataset", scope.makeOpName("PrefetchDataset"));
+    opBuilder.addInput(inputDataset.asOutput());
+    opBuilder.addInput(bufferSize.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    Class[] outputTypesArray = new Class[outputTypes.size()];
+    for (int i = 0; i < outputTypesArray.length; ++i) {
+      outputTypesArray[i] = outputTypes.get(i);
+    }
+    opBuilder.setAttr("output_types", outputTypesArray);
+    Shape[] outputShapesArray = new Shape[outputShapes.size()];
+    for (int i = 0; i < outputShapesArray.length; ++i) {
+      outputShapesArray[i] = outputShapes.get(i);
+    }
+    opBuilder.setAttr("output_shapes", outputShapesArray);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.slackPeriod != null) {
+          opBuilder.setAttr("slack_period", opts.slackPeriod);
+        }
+        if (opts.legacyAutotune != null) {
+          opBuilder.setAttr("legacy_autotune", opts.legacyAutotune);
+        }
+      }
+    }
+    return new PrefetchDataset(opBuilder.build());
+  }
+  
+  /**
+   * @param slackPeriod 
+   */
+  public static Options slackPeriod(Long slackPeriod) {
+    return new Options().slackPeriod(slackPeriod);
+  }
+  
+  /**
+   * @param legacyAutotune 
+   */
+  public static Options legacyAutotune(Boolean legacyAutotune) {
+    return new Options().legacyAutotune(legacyAutotune);
+  }
+  
+  /**
+   */
+  public Output<?> handle() {
+    return handle;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) handle;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "PrefetchDataset";
+  
+  private Output<?> handle;
+  
+  private PrefetchDataset(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    handle = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/PrivateThreadPoolDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/PrivateThreadPoolDataset.java
new file mode 100644
index 00000000000..035a6402a74
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/PrivateThreadPoolDataset.java
@@ -0,0 +1,89 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Creates a dataset that uses a custom thread pool to compute `input_dataset`.
+ */
+public final class PrivateThreadPoolDataset extends RawOp implements Operand<TType> {
+  
+  /**
+   * Factory method to create a class wrapping a new PrivateThreadPoolDataset operation.
+   * 
+   * @param scope current scope
+   * @param inputDataset 
+   * @param numThreads Identifies the number of threads to use for the private threadpool.
+   * @param outputTypes 
+   * @param outputShapes 
+   * @return a new instance of PrivateThreadPoolDataset
+   */
+  @Endpoint(describeByClass = true)
+  public static PrivateThreadPoolDataset create(Scope scope, Operand<?> inputDataset, Operand<TInt64> numThreads, List<Class<? extends TType>> outputTypes, List<Shape> outputShapes) {
+    OperationBuilder opBuilder = scope.env().opBuilder("PrivateThreadPoolDataset", scope.makeOpName("PrivateThreadPoolDataset"));
+    opBuilder.addInput(inputDataset.asOutput());
+    opBuilder.addInput(numThreads.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    Class[] outputTypesArray = new Class[outputTypes.size()];
+    for (int i = 0; i < outputTypesArray.length; ++i) {
+      outputTypesArray[i] = outputTypes.get(i);
+    }
+    opBuilder.setAttr("output_types", outputTypesArray);
+    Shape[] outputShapesArray = new Shape[outputShapes.size()];
+    for (int i = 0; i < outputShapesArray.length; ++i) {
+      outputShapesArray[i] = outputShapes.get(i);
+    }
+    opBuilder.setAttr("output_shapes", outputShapesArray);
+    return new PrivateThreadPoolDataset(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<?> handle() {
+    return handle;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) handle;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "PrivateThreadPoolDataset";
+  
+  private Output<?> handle;
+  
+  private PrivateThreadPoolDataset(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    handle = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/RandomDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/RandomDataset.java
new file mode 100644
index 00000000000..577e512c476
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/RandomDataset.java
@@ -0,0 +1,102 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Creates a Dataset that returns pseudorandom numbers.
+ * <p>
+ * Creates a Dataset that returns a stream of uniformly distributed
+ * pseudorandom 64-bit signed integers.
+ * <p>
+ * In the TensorFlow Python API, you can instantiate this dataset via the
+ * class `tf.data.experimental.RandomDataset`.
+ * <p>
+ * Instances of this dataset are also created as a result of the
+ * `hoist_random_uniform` static optimization. Whether this optimization is
+ * performed is determined by the `experimental_optimization.hoist_random_uniform`
+ * option of `tf.data.Options`.
+ */
+public final class RandomDataset extends RawOp implements Operand<TType> {
+  
+  /**
+   * Factory method to create a class wrapping a new RandomDataset operation.
+   * 
+   * @param scope current scope
+   * @param seed A scalar seed for the random number generator. If either seed or
+   * seed2 is set to be non-zero, the random number generator is seeded
+   * by the given seed.  Otherwise, a random seed is used.
+   * @param seed2 A second scalar seed to avoid seed collision.
+   * @param outputTypes 
+   * @param outputShapes 
+   * @return a new instance of RandomDataset
+   */
+  @Endpoint(describeByClass = true)
+  public static RandomDataset create(Scope scope, Operand<TInt64> seed, Operand<TInt64> seed2, List<Class<? extends TType>> outputTypes, List<Shape> outputShapes) {
+    OperationBuilder opBuilder = scope.env().opBuilder("RandomDataset", scope.makeOpName("RandomDataset"));
+    opBuilder.addInput(seed.asOutput());
+    opBuilder.addInput(seed2.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    Class[] outputTypesArray = new Class[outputTypes.size()];
+    for (int i = 0; i < outputTypesArray.length; ++i) {
+      outputTypesArray[i] = outputTypes.get(i);
+    }
+    opBuilder.setAttr("output_types", outputTypesArray);
+    Shape[] outputShapesArray = new Shape[outputShapes.size()];
+    for (int i = 0; i < outputShapesArray.length; ++i) {
+      outputShapesArray[i] = outputShapes.get(i);
+    }
+    opBuilder.setAttr("output_shapes", outputShapesArray);
+    return new RandomDataset(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<?> handle() {
+    return handle;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) handle;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "RandomDataset";
+  
+  private Output<?> handle;
+  
+  private RandomDataset(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    handle = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/RangeDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/RangeDataset.java
new file mode 100644
index 00000000000..1387e2582e6
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/RangeDataset.java
@@ -0,0 +1,92 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Creates a dataset with a range of values. Corresponds to python's xrange.
+ */
+@Operator(group = "data")
+public final class RangeDataset extends RawOp implements Operand<TType> {
+  
+  /**
+   * Factory method to create a class wrapping a new RangeDataset operation.
+   * 
+   * @param scope current scope
+   * @param start corresponds to start in python's xrange().
+   * @param stop corresponds to stop in python's xrange().
+   * @param step corresponds to step in python's xrange().
+   * @param outputTypes 
+   * @param outputShapes 
+   * @return a new instance of RangeDataset
+   */
+  @Endpoint(describeByClass = true)
+  public static RangeDataset create(Scope scope, Operand<TInt64> start, Operand<TInt64> stop, Operand<TInt64> step, List<Class<? extends TType>> outputTypes, List<Shape> outputShapes) {
+    OperationBuilder opBuilder = scope.env().opBuilder("RangeDataset", scope.makeOpName("RangeDataset"));
+    opBuilder.addInput(start.asOutput());
+    opBuilder.addInput(stop.asOutput());
+    opBuilder.addInput(step.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    Class[] outputTypesArray = new Class[outputTypes.size()];
+    for (int i = 0; i < outputTypesArray.length; ++i) {
+      outputTypesArray[i] = outputTypes.get(i);
+    }
+    opBuilder.setAttr("output_types", outputTypesArray);
+    Shape[] outputShapesArray = new Shape[outputShapes.size()];
+    for (int i = 0; i < outputShapesArray.length; ++i) {
+      outputShapesArray[i] = outputShapes.get(i);
+    }
+    opBuilder.setAttr("output_shapes", outputShapesArray);
+    return new RangeDataset(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<?> handle() {
+    return handle;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) handle;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "RangeDataset";
+  
+  private Output<?> handle;
+  
+  private RangeDataset(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    handle = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/RebatchDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/RebatchDataset.java
new file mode 100644
index 00000000000..d29ba3dd541
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/RebatchDataset.java
@@ -0,0 +1,128 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Creates a dataset that changes the batch size.
+ * <p>
+ * Creates a dataset that changes the batch size of the dataset to current batch
+ * size // num_workers.
+ */
+public final class RebatchDataset extends RawOp implements Operand<TType> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.data.RebatchDataset}
+   */
+  public static class Options {
+    
+    /**
+     * @param useFallback 
+     */
+    public Options useFallback(Boolean useFallback) {
+      this.useFallback = useFallback;
+      return this;
+    }
+    
+    private Boolean useFallback;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new RebatchDataset operation.
+   * 
+   * @param scope current scope
+   * @param inputDataset A variant tensor representing the input dataset.
+   * @param numReplicas A scalar representing the number of replicas to distribute this batch across. As
+   * a result of this transformation the current batch size would end up being
+   * divided  by this parameter.
+   * @param outputTypes 
+   * @param outputShapes 
+   * @param options carries optional attributes values
+   * @return a new instance of RebatchDataset
+   */
+  @Endpoint(describeByClass = true)
+  public static RebatchDataset create(Scope scope, Operand<?> inputDataset, Operand<TInt64> numReplicas, List<Class<? extends TType>> outputTypes, List<Shape> outputShapes, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("RebatchDataset", scope.makeOpName("RebatchDataset"));
+    opBuilder.addInput(inputDataset.asOutput());
+    opBuilder.addInput(numReplicas.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    Class[] outputTypesArray = new Class[outputTypes.size()];
+    for (int i = 0; i < outputTypesArray.length; ++i) {
+      outputTypesArray[i] = outputTypes.get(i);
+    }
+    opBuilder.setAttr("output_types", outputTypesArray);
+    Shape[] outputShapesArray = new Shape[outputShapes.size()];
+    for (int i = 0; i < outputShapesArray.length; ++i) {
+      outputShapesArray[i] = outputShapes.get(i);
+    }
+    opBuilder.setAttr("output_shapes", outputShapesArray);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.useFallback != null) {
+          opBuilder.setAttr("use_fallback", opts.useFallback);
+        }
+      }
+    }
+    return new RebatchDataset(opBuilder.build());
+  }
+  
+  /**
+   * @param useFallback 
+   */
+  public static Options useFallback(Boolean useFallback) {
+    return new Options().useFallback(useFallback);
+  }
+  
+  /**
+   */
+  public Output<?> handle() {
+    return handle;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) handle;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "RebatchDataset";
+  
+  private Output<?> handle;
+  
+  private RebatchDataset(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    handle = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/RegisterDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/RegisterDataset.java
new file mode 100644
index 00000000000..5705413165c
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/RegisterDataset.java
@@ -0,0 +1,78 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.TString;
+
+/**
+ * Registers a dataset with the tf.data service.
+ */
+public final class RegisterDataset extends RawOp implements Operand<TInt64> {
+  
+  /**
+   * Factory method to create a class wrapping a new RegisterDataset operation.
+   * 
+   * @param scope current scope
+   * @param dataset 
+   * @param address 
+   * @param protocol 
+   * @param externalStatePolicy 
+   * @return a new instance of RegisterDataset
+   */
+  @Endpoint(describeByClass = true)
+  public static RegisterDataset create(Scope scope, Operand<?> dataset, Operand<TString> address, Operand<TString> protocol, Long externalStatePolicy) {
+    OperationBuilder opBuilder = scope.env().opBuilder("RegisterDataset", scope.makeOpName("RegisterDataset"));
+    opBuilder.addInput(dataset.asOutput());
+    opBuilder.addInput(address.asOutput());
+    opBuilder.addInput(protocol.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("external_state_policy", externalStatePolicy);
+    return new RegisterDataset(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<TInt64> datasetId() {
+    return datasetId;
+  }
+  
+  @Override
+  public Output<TInt64> asOutput() {
+    return datasetId;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "RegisterDataset";
+  
+  private Output<TInt64> datasetId;
+  
+  private RegisterDataset(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    datasetId = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/RepeatDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/RepeatDataset.java
new file mode 100644
index 00000000000..b12001c2439
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/RepeatDataset.java
@@ -0,0 +1,91 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Creates a dataset that emits the outputs of `input_dataset` `count` times.
+ */
+@Operator(group = "data")
+public final class RepeatDataset extends RawOp implements Operand<TType> {
+  
+  /**
+   * Factory method to create a class wrapping a new RepeatDataset operation.
+   * 
+   * @param scope current scope
+   * @param inputDataset 
+   * @param count A scalar representing the number of times that `input_dataset` should
+   * be repeated. A value of `-1` indicates that it should be repeated infinitely.
+   * @param outputTypes 
+   * @param outputShapes 
+   * @return a new instance of RepeatDataset
+   */
+  @Endpoint(describeByClass = true)
+  public static RepeatDataset create(Scope scope, Operand<?> inputDataset, Operand<TInt64> count, List<Class<? extends TType>> outputTypes, List<Shape> outputShapes) {
+    OperationBuilder opBuilder = scope.env().opBuilder("RepeatDataset", scope.makeOpName("RepeatDataset"));
+    opBuilder.addInput(inputDataset.asOutput());
+    opBuilder.addInput(count.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    Class[] outputTypesArray = new Class[outputTypes.size()];
+    for (int i = 0; i < outputTypesArray.length; ++i) {
+      outputTypesArray[i] = outputTypes.get(i);
+    }
+    opBuilder.setAttr("output_types", outputTypesArray);
+    Shape[] outputShapesArray = new Shape[outputShapes.size()];
+    for (int i = 0; i < outputShapesArray.length; ++i) {
+      outputShapesArray[i] = outputShapes.get(i);
+    }
+    opBuilder.setAttr("output_shapes", outputShapesArray);
+    return new RepeatDataset(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<?> handle() {
+    return handle;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) handle;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "RepeatDataset";
+  
+  private Output<?> handle;
+  
+  private RepeatDataset(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    handle = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SamplingDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SamplingDataset.java
new file mode 100644
index 00000000000..716c2c03128
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SamplingDataset.java
@@ -0,0 +1,101 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Creates a dataset that takes a Bernoulli sample of the contents of another dataset.
+ * <p>
+ * There is no transformation in the `tf.data` Python API for creating this dataset.
+ * Instead, it is created as a result of the `filter_with_random_uniform_fusion`
+ * static optimization. Whether this optimization is performed is determined by the
+ * `experimental_optimization.filter_with_random_uniform_fusion` option of
+ * `tf.data.Options`.
+ */
+public final class SamplingDataset extends RawOp implements Operand<TType> {
+  
+  /**
+   * Factory method to create a class wrapping a new SamplingDataset operation.
+   * 
+   * @param scope current scope
+   * @param inputDataset 
+   * @param rate A scalar representing the sample rate. Each element of `input_dataset` is
+   * retained with this probability, independent of all other elements.
+   * @param seed A scalar representing seed of random number generator.
+   * @param seed2 A scalar representing seed2 of random number generator.
+   * @param outputTypes 
+   * @param outputShapes 
+   * @return a new instance of SamplingDataset
+   */
+  @Endpoint(describeByClass = true)
+  public static SamplingDataset create(Scope scope, Operand<?> inputDataset, Operand<TFloat32> rate, Operand<TInt64> seed, Operand<TInt64> seed2, List<Class<? extends TType>> outputTypes, List<Shape> outputShapes) {
+    OperationBuilder opBuilder = scope.env().opBuilder("SamplingDataset", scope.makeOpName("SamplingDataset"));
+    opBuilder.addInput(inputDataset.asOutput());
+    opBuilder.addInput(rate.asOutput());
+    opBuilder.addInput(seed.asOutput());
+    opBuilder.addInput(seed2.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    Class[] outputTypesArray = new Class[outputTypes.size()];
+    for (int i = 0; i < outputTypesArray.length; ++i) {
+      outputTypesArray[i] = outputTypes.get(i);
+    }
+    opBuilder.setAttr("output_types", outputTypesArray);
+    Shape[] outputShapesArray = new Shape[outputShapes.size()];
+    for (int i = 0; i < outputShapesArray.length; ++i) {
+      outputShapesArray[i] = outputShapes.get(i);
+    }
+    opBuilder.setAttr("output_shapes", outputShapesArray);
+    return new SamplingDataset(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<?> handle() {
+    return handle;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) handle;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "SamplingDataset";
+  
+  private Output<?> handle;
+  
+  private SamplingDataset(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    handle = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SerializeIterator.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SerializeIterator.java
new file mode 100644
index 00000000000..f051994e6f5
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SerializeIterator.java
@@ -0,0 +1,109 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Converts the given `resource_handle` representing an iterator to a variant tensor.
+ */
+@Operator(group = "data")
+public final class SerializeIterator extends RawOp implements Operand<TType> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.data.SerializeIterator}
+   */
+  public static class Options {
+    
+    /**
+     * @param externalStatePolicy 
+     */
+    public Options externalStatePolicy(Long externalStatePolicy) {
+      this.externalStatePolicy = externalStatePolicy;
+      return this;
+    }
+    
+    private Long externalStatePolicy;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new SerializeIterator operation.
+   * 
+   * @param scope current scope
+   * @param resourceHandle A handle to an iterator resource.
+   * @param options carries optional attributes values
+   * @return a new instance of SerializeIterator
+   */
+  @Endpoint(describeByClass = true)
+  public static SerializeIterator create(Scope scope, Operand<?> resourceHandle, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("SerializeIterator", scope.makeOpName("SerializeIterator"));
+    opBuilder.addInput(resourceHandle.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.externalStatePolicy != null) {
+          opBuilder.setAttr("external_state_policy", opts.externalStatePolicy);
+        }
+      }
+    }
+    return new SerializeIterator(opBuilder.build());
+  }
+  
+  /**
+   * @param externalStatePolicy 
+   */
+  public static Options externalStatePolicy(Long externalStatePolicy) {
+    return new Options().externalStatePolicy(externalStatePolicy);
+  }
+  
+  /**
+   * A variant tensor storing the state of the iterator contained in the
+   * resource.
+   */
+  public Output<?> serialized() {
+    return serialized;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) serialized;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "SerializeIterator";
+  
+  private Output<?> serialized;
+  
+  private SerializeIterator(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    serialized = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SetStatsAggregatorDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SetStatsAggregatorDataset.java
new file mode 100644
index 00000000000..a107e2198b4
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SetStatsAggregatorDataset.java
@@ -0,0 +1,92 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TString;
+import org.tensorflow.types.family.TType;
+
+/**
+ */
+public final class SetStatsAggregatorDataset extends RawOp implements Operand<TType> {
+  
+  /**
+   * Factory method to create a class wrapping a new SetStatsAggregatorDataset operation.
+   * 
+   * @param scope current scope
+   * @param inputDataset 
+   * @param statsAggregator 
+   * @param tag 
+   * @param counterPrefix 
+   * @param outputTypes 
+   * @param outputShapes 
+   * @return a new instance of SetStatsAggregatorDataset
+   */
+  @Endpoint(describeByClass = true)
+  public static SetStatsAggregatorDataset create(Scope scope, Operand<?> inputDataset, Operand<?> statsAggregator, Operand<TString> tag, Operand<TString> counterPrefix, List<Class<? extends TType>> outputTypes, List<Shape> outputShapes) {
+    OperationBuilder opBuilder = scope.env().opBuilder("SetStatsAggregatorDataset", scope.makeOpName("SetStatsAggregatorDataset"));
+    opBuilder.addInput(inputDataset.asOutput());
+    opBuilder.addInput(statsAggregator.asOutput());
+    opBuilder.addInput(tag.asOutput());
+    opBuilder.addInput(counterPrefix.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    Class[] outputTypesArray = new Class[outputTypes.size()];
+    for (int i = 0; i < outputTypesArray.length; ++i) {
+      outputTypesArray[i] = outputTypes.get(i);
+    }
+    opBuilder.setAttr("output_types", outputTypesArray);
+    Shape[] outputShapesArray = new Shape[outputShapes.size()];
+    for (int i = 0; i < outputShapesArray.length; ++i) {
+      outputShapesArray[i] = outputShapes.get(i);
+    }
+    opBuilder.setAttr("output_shapes", outputShapesArray);
+    return new SetStatsAggregatorDataset(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<?> handle() {
+    return handle;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) handle;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "SetStatsAggregatorDataset";
+  
+  private Output<?> handle;
+  
+  private SetStatsAggregatorDataset(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    handle = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ShardDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ShardDataset.java
new file mode 100644
index 00000000000..98e043f8e5f
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ShardDataset.java
@@ -0,0 +1,125 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Creates a `Dataset` that includes only 1/`num_shards` of this dataset.
+ */
+public final class ShardDataset extends RawOp implements Operand<TType> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.data.ShardDataset}
+   */
+  public static class Options {
+    
+    /**
+     * @param requireNonEmpty 
+     */
+    public Options requireNonEmpty(Boolean requireNonEmpty) {
+      this.requireNonEmpty = requireNonEmpty;
+      return this;
+    }
+    
+    private Boolean requireNonEmpty;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new ShardDataset operation.
+   * 
+   * @param scope current scope
+   * @param inputDataset 
+   * @param numShards An integer representing the number of shards operating in parallel.
+   * @param index An integer representing the current worker index.
+   * @param outputTypes 
+   * @param outputShapes 
+   * @param options carries optional attributes values
+   * @return a new instance of ShardDataset
+   */
+  @Endpoint(describeByClass = true)
+  public static ShardDataset create(Scope scope, Operand<?> inputDataset, Operand<TInt64> numShards, Operand<TInt64> index, List<Class<? extends TType>> outputTypes, List<Shape> outputShapes, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ShardDataset", scope.makeOpName("ShardDataset"));
+    opBuilder.addInput(inputDataset.asOutput());
+    opBuilder.addInput(numShards.asOutput());
+    opBuilder.addInput(index.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    Class[] outputTypesArray = new Class[outputTypes.size()];
+    for (int i = 0; i < outputTypesArray.length; ++i) {
+      outputTypesArray[i] = outputTypes.get(i);
+    }
+    opBuilder.setAttr("output_types", outputTypesArray);
+    Shape[] outputShapesArray = new Shape[outputShapes.size()];
+    for (int i = 0; i < outputShapesArray.length; ++i) {
+      outputShapesArray[i] = outputShapes.get(i);
+    }
+    opBuilder.setAttr("output_shapes", outputShapesArray);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.requireNonEmpty != null) {
+          opBuilder.setAttr("require_non_empty", opts.requireNonEmpty);
+        }
+      }
+    }
+    return new ShardDataset(opBuilder.build());
+  }
+  
+  /**
+   * @param requireNonEmpty 
+   */
+  public static Options requireNonEmpty(Boolean requireNonEmpty) {
+    return new Options().requireNonEmpty(requireNonEmpty);
+  }
+  
+  /**
+   */
+  public Output<?> handle() {
+    return handle;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) handle;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ShardDataset";
+  
+  private Output<?> handle;
+  
+  private ShardDataset(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    handle = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ShuffleAndRepeatDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ShuffleAndRepeatDataset.java
new file mode 100644
index 00000000000..fa5538deff9
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ShuffleAndRepeatDataset.java
@@ -0,0 +1,130 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
+
+/**
+ */
+public final class ShuffleAndRepeatDataset extends RawOp implements Operand<TType> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.data.ShuffleAndRepeatDataset}
+   */
+  public static class Options {
+    
+    /**
+     * @param reshuffleEachIteration 
+     */
+    public Options reshuffleEachIteration(Boolean reshuffleEachIteration) {
+      this.reshuffleEachIteration = reshuffleEachIteration;
+      return this;
+    }
+    
+    private Boolean reshuffleEachIteration;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new ShuffleAndRepeatDataset operation.
+   * 
+   * @param scope current scope
+   * @param inputDataset 
+   * @param bufferSize 
+   * @param seed 
+   * @param seed2 
+   * @param count 
+   * @param seedGenerator 
+   * @param outputTypes 
+   * @param outputShapes 
+   * @param options carries optional attributes values
+   * @return a new instance of ShuffleAndRepeatDataset
+   */
+  @Endpoint(describeByClass = true)
+  public static ShuffleAndRepeatDataset create(Scope scope, Operand<?> inputDataset, Operand<TInt64> bufferSize, Operand<TInt64> seed, Operand<TInt64> seed2, Operand<TInt64> count, Operand<?> seedGenerator, List<Class<? extends TType>> outputTypes, List<Shape> outputShapes, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ShuffleAndRepeatDatasetV2", scope.makeOpName("ShuffleAndRepeatDataset"));
+    opBuilder.addInput(inputDataset.asOutput());
+    opBuilder.addInput(bufferSize.asOutput());
+    opBuilder.addInput(seed.asOutput());
+    opBuilder.addInput(seed2.asOutput());
+    opBuilder.addInput(count.asOutput());
+    opBuilder.addInput(seedGenerator.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    Class[] outputTypesArray = new Class[outputTypes.size()];
+    for (int i = 0; i < outputTypesArray.length; ++i) {
+      outputTypesArray[i] = outputTypes.get(i);
+    }
+    opBuilder.setAttr("output_types", outputTypesArray);
+    Shape[] outputShapesArray = new Shape[outputShapes.size()];
+    for (int i = 0; i < outputShapesArray.length; ++i) {
+      outputShapesArray[i] = outputShapes.get(i);
+    }
+    opBuilder.setAttr("output_shapes", outputShapesArray);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.reshuffleEachIteration != null) {
+          opBuilder.setAttr("reshuffle_each_iteration", opts.reshuffleEachIteration);
+        }
+      }
+    }
+    return new ShuffleAndRepeatDataset(opBuilder.build());
+  }
+  
+  /**
+   * @param reshuffleEachIteration 
+   */
+  public static Options reshuffleEachIteration(Boolean reshuffleEachIteration) {
+    return new Options().reshuffleEachIteration(reshuffleEachIteration);
+  }
+  
+  /**
+   */
+  public Output<?> handle() {
+    return handle;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) handle;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ShuffleAndRepeatDatasetV2";
+  
+  private Output<?> handle;
+  
+  private ShuffleAndRepeatDataset(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    handle = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ShuffleDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ShuffleDataset.java
new file mode 100644
index 00000000000..e8682dec91c
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ShuffleDataset.java
@@ -0,0 +1,128 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
+
+/**
+ */
+public final class ShuffleDataset extends RawOp implements Operand<TType> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.data.ShuffleDataset}
+   */
+  public static class Options {
+    
+    /**
+     * @param reshuffleEachIteration 
+     */
+    public Options reshuffleEachIteration(Boolean reshuffleEachIteration) {
+      this.reshuffleEachIteration = reshuffleEachIteration;
+      return this;
+    }
+    
+    private Boolean reshuffleEachIteration;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new ShuffleDataset operation.
+   * 
+   * @param scope current scope
+   * @param inputDataset 
+   * @param bufferSize 
+   * @param seed 
+   * @param seed2 
+   * @param seedGenerator 
+   * @param outputTypes 
+   * @param outputShapes 
+   * @param options carries optional attributes values
+   * @return a new instance of ShuffleDataset
+   */
+  @Endpoint(describeByClass = true)
+  public static ShuffleDataset create(Scope scope, Operand<?> inputDataset, Operand<TInt64> bufferSize, Operand<TInt64> seed, Operand<TInt64> seed2, Operand<?> seedGenerator, List<Class<? extends TType>> outputTypes, List<Shape> outputShapes, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ShuffleDatasetV3", scope.makeOpName("ShuffleDataset"));
+    opBuilder.addInput(inputDataset.asOutput());
+    opBuilder.addInput(bufferSize.asOutput());
+    opBuilder.addInput(seed.asOutput());
+    opBuilder.addInput(seed2.asOutput());
+    opBuilder.addInput(seedGenerator.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    Class[] outputTypesArray = new Class[outputTypes.size()];
+    for (int i = 0; i < outputTypesArray.length; ++i) {
+      outputTypesArray[i] = outputTypes.get(i);
+    }
+    opBuilder.setAttr("output_types", outputTypesArray);
+    Shape[] outputShapesArray = new Shape[outputShapes.size()];
+    for (int i = 0; i < outputShapesArray.length; ++i) {
+      outputShapesArray[i] = outputShapes.get(i);
+    }
+    opBuilder.setAttr("output_shapes", outputShapesArray);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.reshuffleEachIteration != null) {
+          opBuilder.setAttr("reshuffle_each_iteration", opts.reshuffleEachIteration);
+        }
+      }
+    }
+    return new ShuffleDataset(opBuilder.build());
+  }
+  
+  /**
+   * @param reshuffleEachIteration 
+   */
+  public static Options reshuffleEachIteration(Boolean reshuffleEachIteration) {
+    return new Options().reshuffleEachIteration(reshuffleEachIteration);
+  }
+  
+  /**
+   */
+  public Output<?> handle() {
+    return handle;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) handle;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ShuffleDatasetV3";
+  
+  private Output<?> handle;
+  
+  private ShuffleDataset(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    handle = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SkipDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SkipDataset.java
new file mode 100644
index 00000000000..2d7171854a2
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SkipDataset.java
@@ -0,0 +1,91 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Creates a dataset that skips `count` elements from the `input_dataset`.
+ */
+@Operator(group = "data")
+public final class SkipDataset extends RawOp implements Operand<TType> {
+  
+  /**
+   * Factory method to create a class wrapping a new SkipDataset operation.
+   * 
+   * @param scope current scope
+   * @param inputDataset 
+   * @param count A scalar representing the number of elements from the `input_dataset`
+   * that should be skipped.  If count is -1, skips everything.
+   * @param outputTypes 
+   * @param outputShapes 
+   * @return a new instance of SkipDataset
+   */
+  @Endpoint(describeByClass = true)
+  public static SkipDataset create(Scope scope, Operand<?> inputDataset, Operand<TInt64> count, List<Class<? extends TType>> outputTypes, List<Shape> outputShapes) {
+    OperationBuilder opBuilder = scope.env().opBuilder("SkipDataset", scope.makeOpName("SkipDataset"));
+    opBuilder.addInput(inputDataset.asOutput());
+    opBuilder.addInput(count.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    Class[] outputTypesArray = new Class[outputTypes.size()];
+    for (int i = 0; i < outputTypesArray.length; ++i) {
+      outputTypesArray[i] = outputTypes.get(i);
+    }
+    opBuilder.setAttr("output_types", outputTypesArray);
+    Shape[] outputShapesArray = new Shape[outputShapes.size()];
+    for (int i = 0; i < outputShapesArray.length; ++i) {
+      outputShapesArray[i] = outputShapes.get(i);
+    }
+    opBuilder.setAttr("output_shapes", outputShapesArray);
+    return new SkipDataset(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<?> handle() {
+    return handle;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) handle;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "SkipDataset";
+  
+  private Output<?> handle;
+  
+  private SkipDataset(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    handle = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SleepDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SleepDataset.java
new file mode 100644
index 00000000000..5feb04bb601
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SleepDataset.java
@@ -0,0 +1,88 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
+
+/**
+ */
+public final class SleepDataset extends RawOp implements Operand<TType> {
+  
+  /**
+   * Factory method to create a class wrapping a new SleepDataset operation.
+   * 
+   * @param scope current scope
+   * @param inputDataset 
+   * @param sleepMicroseconds 
+   * @param outputTypes 
+   * @param outputShapes 
+   * @return a new instance of SleepDataset
+   */
+  @Endpoint(describeByClass = true)
+  public static SleepDataset create(Scope scope, Operand<?> inputDataset, Operand<TInt64> sleepMicroseconds, List<Class<? extends TType>> outputTypes, List<Shape> outputShapes) {
+    OperationBuilder opBuilder = scope.env().opBuilder("SleepDataset", scope.makeOpName("SleepDataset"));
+    opBuilder.addInput(inputDataset.asOutput());
+    opBuilder.addInput(sleepMicroseconds.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    Class[] outputTypesArray = new Class[outputTypes.size()];
+    for (int i = 0; i < outputTypesArray.length; ++i) {
+      outputTypesArray[i] = outputTypes.get(i);
+    }
+    opBuilder.setAttr("output_types", outputTypesArray);
+    Shape[] outputShapesArray = new Shape[outputShapes.size()];
+    for (int i = 0; i < outputShapesArray.length; ++i) {
+      outputShapesArray[i] = outputShapes.get(i);
+    }
+    opBuilder.setAttr("output_shapes", outputShapesArray);
+    return new SleepDataset(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<?> handle() {
+    return handle;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) handle;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "SleepDataset";
+  
+  private Output<?> handle;
+  
+  private SleepDataset(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    handle = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SlidingWindowDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SlidingWindowDataset.java
new file mode 100644
index 00000000000..5f4fc56c523
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SlidingWindowDataset.java
@@ -0,0 +1,96 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Creates a dataset that passes a sliding window over `input_dataset`.
+ */
+public final class SlidingWindowDataset extends RawOp implements Operand<TType> {
+  
+  /**
+   * Factory method to create a class wrapping a new SlidingWindowDataset operation.
+   * 
+   * @param scope current scope
+   * @param inputDataset 
+   * @param windowSize A scalar representing the number of elements in the
+   * sliding window.
+   * @param windowShift A scalar representing the steps moving the sliding window
+   * forward in one iteration. It must be positive.
+   * @param windowStride A scalar representing the stride of the input elements of the sliding window.
+   * It must be positive.
+   * @param outputTypes 
+   * @param outputShapes 
+   * @return a new instance of SlidingWindowDataset
+   */
+  @Endpoint(describeByClass = true)
+  public static SlidingWindowDataset create(Scope scope, Operand<?> inputDataset, Operand<TInt64> windowSize, Operand<TInt64> windowShift, Operand<TInt64> windowStride, List<Class<? extends TType>> outputTypes, List<Shape> outputShapes) {
+    OperationBuilder opBuilder = scope.env().opBuilder("SlidingWindowDataset", scope.makeOpName("SlidingWindowDataset"));
+    opBuilder.addInput(inputDataset.asOutput());
+    opBuilder.addInput(windowSize.asOutput());
+    opBuilder.addInput(windowShift.asOutput());
+    opBuilder.addInput(windowStride.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    Class[] outputTypesArray = new Class[outputTypes.size()];
+    for (int i = 0; i < outputTypesArray.length; ++i) {
+      outputTypesArray[i] = outputTypes.get(i);
+    }
+    opBuilder.setAttr("output_types", outputTypesArray);
+    Shape[] outputShapesArray = new Shape[outputShapes.size()];
+    for (int i = 0; i < outputShapesArray.length; ++i) {
+      outputShapesArray[i] = outputShapes.get(i);
+    }
+    opBuilder.setAttr("output_shapes", outputShapesArray);
+    return new SlidingWindowDataset(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<?> handle() {
+    return handle;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) handle;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "SlidingWindowDataset";
+  
+  private Output<?> handle;
+  
+  private SlidingWindowDataset(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    handle = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SparseTensorSliceDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SparseTensorSliceDataset.java
new file mode 100644
index 00000000000..fa5b61137ab
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SparseTensorSliceDataset.java
@@ -0,0 +1,77 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Creates a dataset that splits a SparseTensor into elements row-wise.
+ */
+public final class SparseTensorSliceDataset extends RawOp implements Operand<TType> {
+  
+  /**
+   * Factory method to create a class wrapping a new SparseTensorSliceDataset operation.
+   * 
+   * @param scope current scope
+   * @param indices 
+   * @param values 
+   * @param denseShape 
+   * @return a new instance of SparseTensorSliceDataset
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> SparseTensorSliceDataset create(Scope scope, Operand<TInt64> indices, Operand<T> values, Operand<TInt64> denseShape) {
+    OperationBuilder opBuilder = scope.env().opBuilder("SparseTensorSliceDataset", scope.makeOpName("SparseTensorSliceDataset"));
+    opBuilder.addInput(indices.asOutput());
+    opBuilder.addInput(values.asOutput());
+    opBuilder.addInput(denseShape.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new SparseTensorSliceDataset(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<?> handle() {
+    return handle;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) handle;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "SparseTensorSliceDataset";
+  
+  private Output<?> handle;
+  
+  private SparseTensorSliceDataset(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    handle = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SqlDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SqlDataset.java
new file mode 100644
index 00000000000..cb7140de450
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/SqlDataset.java
@@ -0,0 +1,91 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TString;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Creates a dataset that executes a SQL query and emits rows of the result set.
+ */
+public final class SqlDataset extends RawOp implements Operand<TType> {
+  
+  /**
+   * Factory method to create a class wrapping a new SqlDataset operation.
+   * 
+   * @param scope current scope
+   * @param driverName The database type. Currently, the only supported type is 'sqlite'.
+   * @param dataSourceName A connection string to connect to the database.
+   * @param query A SQL query to execute.
+   * @param outputTypes 
+   * @param outputShapes 
+   * @return a new instance of SqlDataset
+   */
+  @Endpoint(describeByClass = true)
+  public static SqlDataset create(Scope scope, Operand<TString> driverName, Operand<TString> dataSourceName, Operand<TString> query, List<Class<? extends TType>> outputTypes, List<Shape> outputShapes) {
+    OperationBuilder opBuilder = scope.env().opBuilder("SqlDataset", scope.makeOpName("SqlDataset"));
+    opBuilder.addInput(driverName.asOutput());
+    opBuilder.addInput(dataSourceName.asOutput());
+    opBuilder.addInput(query.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    Class[] outputTypesArray = new Class[outputTypes.size()];
+    for (int i = 0; i < outputTypesArray.length; ++i) {
+      outputTypesArray[i] = outputTypes.get(i);
+    }
+    opBuilder.setAttr("output_types", outputTypesArray);
+    Shape[] outputShapesArray = new Shape[outputShapes.size()];
+    for (int i = 0; i < outputShapesArray.length; ++i) {
+      outputShapesArray[i] = outputShapes.get(i);
+    }
+    opBuilder.setAttr("output_shapes", outputShapesArray);
+    return new SqlDataset(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<?> handle() {
+    return handle;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) handle;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "SqlDataset";
+  
+  private Output<?> handle;
+  
+  private SqlDataset(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    handle = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/StatsAggregatorHandle.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/StatsAggregatorHandle.java
new file mode 100644
index 00000000000..f510b437c7d
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/StatsAggregatorHandle.java
@@ -0,0 +1,123 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Creates a statistics manager resource.
+ */
+public final class StatsAggregatorHandle extends RawOp implements Operand<TType> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.data.StatsAggregatorHandle}
+   */
+  public static class Options {
+    
+    /**
+     * @param container 
+     */
+    public Options container(String container) {
+      this.container = container;
+      return this;
+    }
+    
+    /**
+     * @param sharedName 
+     */
+    public Options sharedName(String sharedName) {
+      this.sharedName = sharedName;
+      return this;
+    }
+    
+    private String container;
+    private String sharedName;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new StatsAggregatorHandle operation.
+   * 
+   * @param scope current scope
+   * @param options carries optional attributes values
+   * @return a new instance of StatsAggregatorHandle
+   */
+  @Endpoint(describeByClass = true)
+  public static StatsAggregatorHandle create(Scope scope, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("StatsAggregatorHandle", scope.makeOpName("StatsAggregatorHandle"));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.container != null) {
+          opBuilder.setAttr("container", opts.container);
+        }
+        if (opts.sharedName != null) {
+          opBuilder.setAttr("shared_name", opts.sharedName);
+        }
+      }
+    }
+    return new StatsAggregatorHandle(opBuilder.build());
+  }
+  
+  /**
+   * @param container 
+   */
+  public static Options container(String container) {
+    return new Options().container(container);
+  }
+  
+  /**
+   * @param sharedName 
+   */
+  public static Options sharedName(String sharedName) {
+    return new Options().sharedName(sharedName);
+  }
+  
+  /**
+   */
+  public Output<?> handle() {
+    return handle;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) handle;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "StatsAggregatorHandle";
+  
+  private Output<?> handle;
+  
+  private StatsAggregatorHandle(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    handle = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/TakeDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/TakeDataset.java
new file mode 100644
index 00000000000..8c12d56db8a
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/TakeDataset.java
@@ -0,0 +1,92 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Creates a dataset that contains `count` elements from the `input_dataset`.
+ */
+@Operator(group = "data")
+public final class TakeDataset extends RawOp implements Operand<TType> {
+  
+  /**
+   * Factory method to create a class wrapping a new TakeDataset operation.
+   * 
+   * @param scope current scope
+   * @param inputDataset 
+   * @param count A scalar representing the number of elements from the `input_dataset`
+   * that should be taken. A value of `-1` indicates that all of `input_dataset`
+   * is taken.
+   * @param outputTypes 
+   * @param outputShapes 
+   * @return a new instance of TakeDataset
+   */
+  @Endpoint(describeByClass = true)
+  public static TakeDataset create(Scope scope, Operand<?> inputDataset, Operand<TInt64> count, List<Class<? extends TType>> outputTypes, List<Shape> outputShapes) {
+    OperationBuilder opBuilder = scope.env().opBuilder("TakeDataset", scope.makeOpName("TakeDataset"));
+    opBuilder.addInput(inputDataset.asOutput());
+    opBuilder.addInput(count.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    Class[] outputTypesArray = new Class[outputTypes.size()];
+    for (int i = 0; i < outputTypesArray.length; ++i) {
+      outputTypesArray[i] = outputTypes.get(i);
+    }
+    opBuilder.setAttr("output_types", outputTypesArray);
+    Shape[] outputShapesArray = new Shape[outputShapes.size()];
+    for (int i = 0; i < outputShapesArray.length; ++i) {
+      outputShapesArray[i] = outputShapes.get(i);
+    }
+    opBuilder.setAttr("output_shapes", outputShapesArray);
+    return new TakeDataset(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<?> handle() {
+    return handle;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) handle;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "TakeDataset";
+  
+  private Output<?> handle;
+  
+  private TakeDataset(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    handle = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/TensorDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/TensorDataset.java
new file mode 100644
index 00000000000..974f9562859
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/TensorDataset.java
@@ -0,0 +1,81 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.Operands;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Creates a dataset that emits `components` as a tuple of tensors once.
+ */
+public final class TensorDataset extends RawOp implements Operand<TType> {
+  
+  /**
+   * Factory method to create a class wrapping a new TensorDataset operation.
+   * 
+   * @param scope current scope
+   * @param components 
+   * @param outputShapes 
+   * @return a new instance of TensorDataset
+   */
+  @Endpoint(describeByClass = true)
+  public static TensorDataset create(Scope scope, Iterable<Operand<?>> components, List<Shape> outputShapes) {
+    OperationBuilder opBuilder = scope.env().opBuilder("TensorDataset", scope.makeOpName("TensorDataset"));
+    opBuilder.addInputList(Operands.asOutputs(components));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    Shape[] outputShapesArray = new Shape[outputShapes.size()];
+    for (int i = 0; i < outputShapesArray.length; ++i) {
+      outputShapesArray[i] = outputShapes.get(i);
+    }
+    opBuilder.setAttr("output_shapes", outputShapesArray);
+    return new TensorDataset(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<?> handle() {
+    return handle;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) handle;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "TensorDataset";
+  
+  private Output<?> handle;
+  
+  private TensorDataset(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    handle = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/TensorSliceDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/TensorSliceDataset.java
new file mode 100644
index 00000000000..0d0ca40de32
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/TensorSliceDataset.java
@@ -0,0 +1,82 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.Operands;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Creates a dataset that emits each dim-0 slice of `components` once.
+ */
+@Operator(group = "data")
+public final class TensorSliceDataset extends RawOp implements Operand<TType> {
+  
+  /**
+   * Factory method to create a class wrapping a new TensorSliceDataset operation.
+   * 
+   * @param scope current scope
+   * @param components 
+   * @param outputShapes 
+   * @return a new instance of TensorSliceDataset
+   */
+  @Endpoint(describeByClass = true)
+  public static TensorSliceDataset create(Scope scope, Iterable<Operand<?>> components, List<Shape> outputShapes) {
+    OperationBuilder opBuilder = scope.env().opBuilder("TensorSliceDataset", scope.makeOpName("TensorSliceDataset"));
+    opBuilder.addInputList(Operands.asOutputs(components));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    Shape[] outputShapesArray = new Shape[outputShapes.size()];
+    for (int i = 0; i < outputShapesArray.length; ++i) {
+      outputShapesArray[i] = outputShapes.get(i);
+    }
+    opBuilder.setAttr("output_shapes", outputShapesArray);
+    return new TensorSliceDataset(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<?> handle() {
+    return handle;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) handle;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "TensorSliceDataset";
+  
+  private Output<?> handle;
+  
+  private TensorSliceDataset(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    handle = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/TextLineDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/TextLineDataset.java
new file mode 100644
index 00000000000..a80851bb2ed
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/TextLineDataset.java
@@ -0,0 +1,81 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.TString;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Creates a dataset that emits the lines of one or more text files.
+ */
+@Operator(group = "data")
+public final class TextLineDataset extends RawOp implements Operand<TType> {
+  
+  /**
+   * Factory method to create a class wrapping a new TextLineDataset operation.
+   * 
+   * @param scope current scope
+   * @param filenames A scalar or a vector containing the name(s) of the file(s) to be
+   * read.
+   * @param compressionType A scalar containing either (i) the empty string (no
+   * compression), (ii) "ZLIB", or (iii) "GZIP".
+   * @param bufferSize A scalar containing the number of bytes to buffer.
+   * @return a new instance of TextLineDataset
+   */
+  @Endpoint(describeByClass = true)
+  public static TextLineDataset create(Scope scope, Operand<TString> filenames, Operand<TString> compressionType, Operand<TInt64> bufferSize) {
+    OperationBuilder opBuilder = scope.env().opBuilder("TextLineDataset", scope.makeOpName("TextLineDataset"));
+    opBuilder.addInput(filenames.asOutput());
+    opBuilder.addInput(compressionType.asOutput());
+    opBuilder.addInput(bufferSize.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new TextLineDataset(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<?> handle() {
+    return handle;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) handle;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "TextLineDataset";
+  
+  private Output<?> handle;
+  
+  private TextLineDataset(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    handle = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/TfRecordDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/TfRecordDataset.java
new file mode 100644
index 00000000000..4795e623f82
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/TfRecordDataset.java
@@ -0,0 +1,82 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.TString;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Creates a dataset that emits the records from one or more TFRecord files.
+ */
+@Operator(group = "data")
+public final class TfRecordDataset extends RawOp implements Operand<TType> {
+  
+  /**
+   * Factory method to create a class wrapping a new TfRecordDataset operation.
+   * 
+   * @param scope current scope
+   * @param filenames A scalar or vector containing the name(s) of the file(s) to be
+   * read.
+   * @param compressionType A scalar containing either (i) the empty string (no
+   * compression), (ii) "ZLIB", or (iii) "GZIP".
+   * @param bufferSize A scalar representing the number of bytes to buffer. A value of
+   * 0 means no buffering will be performed.
+   * @return a new instance of TfRecordDataset
+   */
+  @Endpoint(describeByClass = true)
+  public static TfRecordDataset create(Scope scope, Operand<TString> filenames, Operand<TString> compressionType, Operand<TInt64> bufferSize) {
+    OperationBuilder opBuilder = scope.env().opBuilder("TFRecordDataset", scope.makeOpName("TfRecordDataset"));
+    opBuilder.addInput(filenames.asOutput());
+    opBuilder.addInput(compressionType.asOutput());
+    opBuilder.addInput(bufferSize.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new TfRecordDataset(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<?> handle() {
+    return handle;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) handle;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "TFRecordDataset";
+  
+  private Output<?> handle;
+  
+  private TfRecordDataset(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    handle = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ThreadPoolDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ThreadPoolDataset.java
new file mode 100644
index 00000000000..9cc2ad12bba
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ThreadPoolDataset.java
@@ -0,0 +1,88 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Creates a dataset that uses a custom thread pool to compute `input_dataset`.
+ */
+public final class ThreadPoolDataset extends RawOp implements Operand<TType> {
+  
+  /**
+   * Factory method to create a class wrapping a new ThreadPoolDataset operation.
+   * 
+   * @param scope current scope
+   * @param inputDataset 
+   * @param threadPool A resource produced by the ThreadPoolHandle op.
+   * @param outputTypes 
+   * @param outputShapes 
+   * @return a new instance of ThreadPoolDataset
+   */
+  @Endpoint(describeByClass = true)
+  public static ThreadPoolDataset create(Scope scope, Operand<?> inputDataset, Operand<?> threadPool, List<Class<? extends TType>> outputTypes, List<Shape> outputShapes) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ThreadPoolDataset", scope.makeOpName("ThreadPoolDataset"));
+    opBuilder.addInput(inputDataset.asOutput());
+    opBuilder.addInput(threadPool.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    Class[] outputTypesArray = new Class[outputTypes.size()];
+    for (int i = 0; i < outputTypesArray.length; ++i) {
+      outputTypesArray[i] = outputTypes.get(i);
+    }
+    opBuilder.setAttr("output_types", outputTypesArray);
+    Shape[] outputShapesArray = new Shape[outputShapes.size()];
+    for (int i = 0; i < outputShapesArray.length; ++i) {
+      outputShapesArray[i] = outputShapes.get(i);
+    }
+    opBuilder.setAttr("output_shapes", outputShapesArray);
+    return new ThreadPoolDataset(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<?> handle() {
+    return handle;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) handle;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ThreadPoolDataset";
+  
+  private Output<?> handle;
+  
+  private ThreadPoolDataset(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    handle = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ThreadPoolHandle.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ThreadPoolHandle.java
new file mode 100644
index 00000000000..6d9434cdff3
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ThreadPoolHandle.java
@@ -0,0 +1,152 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Creates a dataset that uses a custom thread pool to compute `input_dataset`.
+ */
+public final class ThreadPoolHandle extends RawOp implements Operand<TType> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.data.ThreadPoolHandle}
+   */
+  public static class Options {
+    
+    /**
+     * @param maxIntraOpParallelism The maximum degree of parallelism to use within operations that execute on this
+     * threadpool.
+     */
+    public Options maxIntraOpParallelism(Long maxIntraOpParallelism) {
+      this.maxIntraOpParallelism = maxIntraOpParallelism;
+      return this;
+    }
+    
+    /**
+     * @param container 
+     */
+    public Options container(String container) {
+      this.container = container;
+      return this;
+    }
+    
+    /**
+     * @param sharedName 
+     */
+    public Options sharedName(String sharedName) {
+      this.sharedName = sharedName;
+      return this;
+    }
+    
+    private Long maxIntraOpParallelism;
+    private String container;
+    private String sharedName;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new ThreadPoolHandle operation.
+   * 
+   * @param scope current scope
+   * @param numThreads The number of threads in the thread pool.
+   * @param displayName A human-readable name for the threads that may be visible in some
+   * visualizations.
+   * threadpool.
+   * @param options carries optional attributes values
+   * @return a new instance of ThreadPoolHandle
+   */
+  @Endpoint(describeByClass = true)
+  public static ThreadPoolHandle create(Scope scope, Long numThreads, String displayName, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ThreadPoolHandle", scope.makeOpName("ThreadPoolHandle"));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("num_threads", numThreads);
+    opBuilder.setAttr("display_name", displayName);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.maxIntraOpParallelism != null) {
+          opBuilder.setAttr("max_intra_op_parallelism", opts.maxIntraOpParallelism);
+        }
+        if (opts.container != null) {
+          opBuilder.setAttr("container", opts.container);
+        }
+        if (opts.sharedName != null) {
+          opBuilder.setAttr("shared_name", opts.sharedName);
+        }
+      }
+    }
+    return new ThreadPoolHandle(opBuilder.build());
+  }
+  
+  /**
+   * @param maxIntraOpParallelism The maximum degree of parallelism to use within operations that execute on this
+   * threadpool.
+   */
+  public static Options maxIntraOpParallelism(Long maxIntraOpParallelism) {
+    return new Options().maxIntraOpParallelism(maxIntraOpParallelism);
+  }
+  
+  /**
+   * @param container 
+   */
+  public static Options container(String container) {
+    return new Options().container(container);
+  }
+  
+  /**
+   * @param sharedName 
+   */
+  public static Options sharedName(String sharedName) {
+    return new Options().sharedName(sharedName);
+  }
+  
+  /**
+   * A resource that can be consumed by one or more ExperimentalThreadPoolDataset
+   * ops.
+   */
+  public Output<?> handle() {
+    return handle;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) handle;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ThreadPoolHandle";
+  
+  private Output<?> handle;
+  
+  private ThreadPoolHandle(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    handle = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/UnbatchDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/UnbatchDataset.java
new file mode 100644
index 00000000000..0ead7ff2ef1
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/UnbatchDataset.java
@@ -0,0 +1,86 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * A dataset that splits the elements of its input into multiple elements.
+ */
+public final class UnbatchDataset extends RawOp implements Operand<TType> {
+  
+  /**
+   * Factory method to create a class wrapping a new UnbatchDataset operation.
+   * 
+   * @param scope current scope
+   * @param inputDataset 
+   * @param outputTypes 
+   * @param outputShapes 
+   * @return a new instance of UnbatchDataset
+   */
+  @Endpoint(describeByClass = true)
+  public static UnbatchDataset create(Scope scope, Operand<?> inputDataset, List<Class<? extends TType>> outputTypes, List<Shape> outputShapes) {
+    OperationBuilder opBuilder = scope.env().opBuilder("UnbatchDataset", scope.makeOpName("UnbatchDataset"));
+    opBuilder.addInput(inputDataset.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    Class[] outputTypesArray = new Class[outputTypes.size()];
+    for (int i = 0; i < outputTypesArray.length; ++i) {
+      outputTypesArray[i] = outputTypes.get(i);
+    }
+    opBuilder.setAttr("output_types", outputTypesArray);
+    Shape[] outputShapesArray = new Shape[outputShapes.size()];
+    for (int i = 0; i < outputShapesArray.length; ++i) {
+      outputShapesArray[i] = outputShapes.get(i);
+    }
+    opBuilder.setAttr("output_shapes", outputShapesArray);
+    return new UnbatchDataset(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<?> handle() {
+    return handle;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) handle;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "UnbatchDataset";
+  
+  private Output<?> handle;
+  
+  private UnbatchDataset(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    handle = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/UniqueDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/UniqueDataset.java
new file mode 100644
index 00000000000..f47548f8dad
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/UniqueDataset.java
@@ -0,0 +1,86 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Creates a dataset that contains the unique elements of `input_dataset`.
+ */
+public final class UniqueDataset extends RawOp implements Operand<TType> {
+  
+  /**
+   * Factory method to create a class wrapping a new UniqueDataset operation.
+   * 
+   * @param scope current scope
+   * @param inputDataset 
+   * @param outputTypes 
+   * @param outputShapes 
+   * @return a new instance of UniqueDataset
+   */
+  @Endpoint(describeByClass = true)
+  public static UniqueDataset create(Scope scope, Operand<?> inputDataset, List<Class<? extends TType>> outputTypes, List<Shape> outputShapes) {
+    OperationBuilder opBuilder = scope.env().opBuilder("UniqueDataset", scope.makeOpName("UniqueDataset"));
+    opBuilder.addInput(inputDataset.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    Class[] outputTypesArray = new Class[outputTypes.size()];
+    for (int i = 0; i < outputTypesArray.length; ++i) {
+      outputTypesArray[i] = outputTypes.get(i);
+    }
+    opBuilder.setAttr("output_types", outputTypesArray);
+    Shape[] outputShapesArray = new Shape[outputShapes.size()];
+    for (int i = 0; i < outputShapesArray.length; ++i) {
+      outputShapesArray[i] = outputShapes.get(i);
+    }
+    opBuilder.setAttr("output_shapes", outputShapesArray);
+    return new UniqueDataset(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<?> handle() {
+    return handle;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) handle;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "UniqueDataset";
+  
+  private Output<?> handle;
+  
+  private UniqueDataset(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    handle = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/UnwrapDatasetVariant.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/UnwrapDatasetVariant.java
new file mode 100644
index 00000000000..d5125db92c8
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/UnwrapDatasetVariant.java
@@ -0,0 +1,71 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ */
+public final class UnwrapDatasetVariant extends RawOp implements Operand<TType> {
+  
+  /**
+   * Factory method to create a class wrapping a new UnwrapDatasetVariant operation.
+   * 
+   * @param scope current scope
+   * @param inputHandle 
+   * @return a new instance of UnwrapDatasetVariant
+   */
+  @Endpoint(describeByClass = true)
+  public static UnwrapDatasetVariant create(Scope scope, Operand<?> inputHandle) {
+    OperationBuilder opBuilder = scope.env().opBuilder("UnwrapDatasetVariant", scope.makeOpName("UnwrapDatasetVariant"));
+    opBuilder.addInput(inputHandle.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new UnwrapDatasetVariant(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<?> outputHandle() {
+    return outputHandle;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) outputHandle;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "UnwrapDatasetVariant";
+  
+  private Output<?> outputHandle;
+  
+  private UnwrapDatasetVariant(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    outputHandle = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/WindowDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/WindowDataset.java
new file mode 100644
index 00000000000..332338991f5
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/WindowDataset.java
@@ -0,0 +1,143 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TBool;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
+
+/**
+ *   Combines (nests of) input elements into a dataset of (nests of) windows.
+ * <p>
+ *   A "window" is a finite dataset of flat elements of size `size` (or possibly
+ *   fewer if there are not enough input elements to fill the window and
+ *   `drop_remainder` evaluates to false).
+ * <p>
+ *   The `shift` argument determines the number of input elements by which
+ *   the window moves on each iteration.  The first element in the `k`th window
+ *   will be element
+ * <p>
+ *   <pre>{@code
+ *   1 + (k-1) * shift
+ *   }</pre>
+ * of the input dataset. In particular, the first element of the first window
+ *   will always be the first element of the input dataset.  
+ * <p>
+ *   If the `stride` parameter is greater than 1, then each window will skip
+ *   `(stride - 1)` input elements between each element that appears in the
+ *   window. Output windows will still contain `size` elements regardless of
+ *   the value of `stride`.
+ * <p>
+ *   The `stride` argument determines the stride of the input elements, and the
+ *   `shift` argument determines the shift of the window.
+ * <p>
+ *   For example, letting `{...}` to represent a Dataset:
+ * <p>
+ *   - `tf.data.Dataset.range(7).window(2)` produces
+ *     `{{0, 1}, {2, 3}, {4, 5}, {6}}`
+ *   - `tf.data.Dataset.range(7).window(3, 2, 1, True)` produces
+ *     `{{0, 1, 2}, {2, 3, 4}, {4, 5, 6}}`
+ *   - `tf.data.Dataset.range(7).window(3, 1, 2, True)` produces
+ *     `{{0, 2, 4}, {1, 3, 5}, {2, 4, 6}}`
+ * <p>
+ *   Note that when the `window` transformation is applied to a dataset of
+ *   nested elements, it produces a dataset of nested windows.
+ * <p>
+ *   For example:
+ * <p>
+ *   - `tf.data.Dataset.from_tensor_slices((range(4), range(4))).window(2)`
+ *     produces `{({0, 1}, {0, 1}), ({2, 3}, {2, 3})}`
+ *   - `tf.data.Dataset.from_tensor_slices({"a": range(4)}).window(2)`
+ *     produces `{{"a": {0, 1}}, {"a": {2, 3}}}`
+ */
+public final class WindowDataset extends RawOp implements Operand<TType> {
+  
+  /**
+   * Factory method to create a class wrapping a new WindowDataset operation.
+   * 
+   * @param scope current scope
+   * @param inputDataset 
+   * @param size An integer scalar, representing the number of elements
+   * of the input dataset to combine into a window. Must be positive.
+   * @param shift An integer scalar, representing the number of input elements
+   * by which the window moves in each iteration.  Defaults to `size`.
+   * Must be positive.
+   * @param stride An integer scalar, representing the stride of the input elements
+   * in the sliding window. Must be positive. The default value of 1 means
+   * "retain every input element".
+   * @param dropRemainder A Boolean scalar, representing whether the last window should be
+   * dropped if its size is smaller than `window_size`.
+   * @param outputTypes 
+   * @param outputShapes 
+   * @return a new instance of WindowDataset
+   */
+  @Endpoint(describeByClass = true)
+  public static WindowDataset create(Scope scope, Operand<?> inputDataset, Operand<TInt64> size, Operand<TInt64> shift, Operand<TInt64> stride, Operand<TBool> dropRemainder, List<Class<? extends TType>> outputTypes, List<Shape> outputShapes) {
+    OperationBuilder opBuilder = scope.env().opBuilder("WindowDataset", scope.makeOpName("WindowDataset"));
+    opBuilder.addInput(inputDataset.asOutput());
+    opBuilder.addInput(size.asOutput());
+    opBuilder.addInput(shift.asOutput());
+    opBuilder.addInput(stride.asOutput());
+    opBuilder.addInput(dropRemainder.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    Class[] outputTypesArray = new Class[outputTypes.size()];
+    for (int i = 0; i < outputTypesArray.length; ++i) {
+      outputTypesArray[i] = outputTypes.get(i);
+    }
+    opBuilder.setAttr("output_types", outputTypesArray);
+    Shape[] outputShapesArray = new Shape[outputShapes.size()];
+    for (int i = 0; i < outputShapesArray.length; ++i) {
+      outputShapesArray[i] = outputShapes.get(i);
+    }
+    opBuilder.setAttr("output_shapes", outputShapesArray);
+    return new WindowDataset(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<?> handle() {
+    return handle;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) handle;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "WindowDataset";
+  
+  private Output<?> handle;
+  
+  private WindowDataset(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    handle = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/WrapDatasetVariant.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/WrapDatasetVariant.java
new file mode 100644
index 00000000000..bcea6df7895
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/WrapDatasetVariant.java
@@ -0,0 +1,71 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ */
+public final class WrapDatasetVariant extends RawOp implements Operand<TType> {
+  
+  /**
+   * Factory method to create a class wrapping a new WrapDatasetVariant operation.
+   * 
+   * @param scope current scope
+   * @param inputHandle 
+   * @return a new instance of WrapDatasetVariant
+   */
+  @Endpoint(describeByClass = true)
+  public static WrapDatasetVariant create(Scope scope, Operand<?> inputHandle) {
+    OperationBuilder opBuilder = scope.env().opBuilder("WrapDatasetVariant", scope.makeOpName("WrapDatasetVariant"));
+    opBuilder.addInput(inputHandle.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new WrapDatasetVariant(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<?> outputHandle() {
+    return outputHandle;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) outputHandle;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "WrapDatasetVariant";
+  
+  private Output<?> outputHandle;
+  
+  private WrapDatasetVariant(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    outputHandle = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ZipDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ZipDataset.java
new file mode 100644
index 00000000000..d5414b73ecf
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/ZipDataset.java
@@ -0,0 +1,94 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.Operands;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Creates a dataset that zips together `input_datasets`.
+ * <p>
+ * The elements of the resulting dataset are created by zipping corresponding
+ * elements from each of the input datasets.
+ * <p>
+ * The size of the resulting dataset will match the size of the smallest input
+ * dataset, and no error will be raised if input datasets have different sizes.
+ */
+@Operator(group = "data")
+public final class ZipDataset extends RawOp implements Operand<TType> {
+  
+  /**
+   * Factory method to create a class wrapping a new ZipDataset operation.
+   * 
+   * @param scope current scope
+   * @param inputDatasets List of `N` variant Tensors representing datasets to be zipped together.
+   * @param outputTypes 
+   * @param outputShapes 
+   * @return a new instance of ZipDataset
+   */
+  @Endpoint(describeByClass = true)
+  public static ZipDataset create(Scope scope, Iterable<Operand<?>> inputDatasets, List<Class<? extends TType>> outputTypes, List<Shape> outputShapes) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ZipDataset", scope.makeOpName("ZipDataset"));
+    opBuilder.addInputList(Operands.asOutputs(inputDatasets));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    Class[] outputTypesArray = new Class[outputTypes.size()];
+    for (int i = 0; i < outputTypesArray.length; ++i) {
+      outputTypesArray[i] = outputTypes.get(i);
+    }
+    opBuilder.setAttr("output_types", outputTypesArray);
+    Shape[] outputShapesArray = new Shape[outputShapes.size()];
+    for (int i = 0; i < outputShapesArray.length; ++i) {
+      outputShapesArray[i] = outputShapes.get(i);
+    }
+    opBuilder.setAttr("output_shapes", outputShapesArray);
+    return new ZipDataset(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<?> handle() {
+    return handle;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) handle;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ZipDataset";
+  
+  private Output<?> handle;
+  
+  private ZipDataset(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    handle = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/AssertCardinalityDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/AssertCardinalityDataset.java
new file mode 100644
index 00000000000..6e3694983d4
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/AssertCardinalityDataset.java
@@ -0,0 +1,88 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data.experimental;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
+
+/**
+ */
+public final class AssertCardinalityDataset extends RawOp implements Operand<TType> {
+  
+  /**
+   * Factory method to create a class wrapping a new AssertCardinalityDataset operation.
+   * 
+   * @param scope current scope
+   * @param inputDataset 
+   * @param cardinality 
+   * @param outputTypes 
+   * @param outputShapes 
+   * @return a new instance of AssertCardinalityDataset
+   */
+  @Endpoint(describeByClass = true)
+  public static AssertCardinalityDataset create(Scope scope, Operand<?> inputDataset, Operand<TInt64> cardinality, List<Class<? extends TType>> outputTypes, List<Shape> outputShapes) {
+    OperationBuilder opBuilder = scope.env().opBuilder("AssertCardinalityDataset", scope.makeOpName("AssertCardinalityDataset"));
+    opBuilder.addInput(inputDataset.asOutput());
+    opBuilder.addInput(cardinality.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    Class[] outputTypesArray = new Class[outputTypes.size()];
+    for (int i = 0; i < outputTypesArray.length; ++i) {
+      outputTypesArray[i] = outputTypes.get(i);
+    }
+    opBuilder.setAttr("output_types", outputTypesArray);
+    Shape[] outputShapesArray = new Shape[outputShapes.size()];
+    for (int i = 0; i < outputShapesArray.length; ++i) {
+      outputShapesArray[i] = outputShapes.get(i);
+    }
+    opBuilder.setAttr("output_shapes", outputShapesArray);
+    return new AssertCardinalityDataset(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<?> handle() {
+    return handle;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) handle;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "AssertCardinalityDataset";
+  
+  private Output<?> handle;
+  
+  private AssertCardinalityDataset(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    handle = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/AssertNextDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/AssertNextDataset.java
new file mode 100644
index 00000000000..f4f0b763dc5
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/AssertNextDataset.java
@@ -0,0 +1,88 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data.experimental;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TString;
+import org.tensorflow.types.family.TType;
+
+/**
+ */
+public final class AssertNextDataset extends RawOp implements Operand<TType> {
+  
+  /**
+   * Factory method to create a class wrapping a new AssertNextDataset operation.
+   * 
+   * @param scope current scope
+   * @param inputDataset 
+   * @param transformations 
+   * @param outputTypes 
+   * @param outputShapes 
+   * @return a new instance of AssertNextDataset
+   */
+  @Endpoint(describeByClass = true)
+  public static AssertNextDataset create(Scope scope, Operand<?> inputDataset, Operand<TString> transformations, List<Class<? extends TType>> outputTypes, List<Shape> outputShapes) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ExperimentalAssertNextDataset", scope.makeOpName("AssertNextDataset"));
+    opBuilder.addInput(inputDataset.asOutput());
+    opBuilder.addInput(transformations.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    Class[] outputTypesArray = new Class[outputTypes.size()];
+    for (int i = 0; i < outputTypesArray.length; ++i) {
+      outputTypesArray[i] = outputTypes.get(i);
+    }
+    opBuilder.setAttr("output_types", outputTypesArray);
+    Shape[] outputShapesArray = new Shape[outputShapes.size()];
+    for (int i = 0; i < outputShapesArray.length; ++i) {
+      outputShapesArray[i] = outputShapes.get(i);
+    }
+    opBuilder.setAttr("output_shapes", outputShapesArray);
+    return new AssertNextDataset(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<?> handle() {
+    return handle;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) handle;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ExperimentalAssertNextDataset";
+  
+  private Output<?> handle;
+  
+  private AssertNextDataset(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    handle = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/AutoShardDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/AutoShardDataset.java
new file mode 100644
index 00000000000..d818cdd4175
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/AutoShardDataset.java
@@ -0,0 +1,133 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data.experimental;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Creates a dataset that shards the input dataset.
+ * <p>
+ * Creates a dataset that shards the input dataset by num_workers, returning a
+ * sharded dataset for the index-th worker. This attempts to automatically shard
+ * a dataset by examining the Dataset graph and inserting a shard op before the
+ * inputs to a reader Dataset (e.g. CSVDataset, TFRecordDataset).
+ * <p>
+ * This dataset will throw a NotFound error if we cannot shard the dataset
+ * automatically.
+ */
+public final class AutoShardDataset extends RawOp implements Operand<TType> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.data.experimental.AutoShardDataset}
+   */
+  public static class Options {
+    
+    /**
+     * @param autoShardPolicy 
+     */
+    public Options autoShardPolicy(Long autoShardPolicy) {
+      this.autoShardPolicy = autoShardPolicy;
+      return this;
+    }
+    
+    private Long autoShardPolicy;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new AutoShardDataset operation.
+   * 
+   * @param scope current scope
+   * @param inputDataset A variant tensor representing the input dataset.
+   * @param numWorkers A scalar representing the number of workers to distribute this dataset across.
+   * @param index A scalar representing the index of the current worker out of num_workers.
+   * @param outputTypes 
+   * @param outputShapes 
+   * @param options carries optional attributes values
+   * @return a new instance of AutoShardDataset
+   */
+  @Endpoint(describeByClass = true)
+  public static AutoShardDataset create(Scope scope, Operand<?> inputDataset, Operand<TInt64> numWorkers, Operand<TInt64> index, List<Class<? extends TType>> outputTypes, List<Shape> outputShapes, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ExperimentalAutoShardDataset", scope.makeOpName("AutoShardDataset"));
+    opBuilder.addInput(inputDataset.asOutput());
+    opBuilder.addInput(numWorkers.asOutput());
+    opBuilder.addInput(index.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    Class[] outputTypesArray = new Class[outputTypes.size()];
+    for (int i = 0; i < outputTypesArray.length; ++i) {
+      outputTypesArray[i] = outputTypes.get(i);
+    }
+    opBuilder.setAttr("output_types", outputTypesArray);
+    Shape[] outputShapesArray = new Shape[outputShapes.size()];
+    for (int i = 0; i < outputShapesArray.length; ++i) {
+      outputShapesArray[i] = outputShapes.get(i);
+    }
+    opBuilder.setAttr("output_shapes", outputShapesArray);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.autoShardPolicy != null) {
+          opBuilder.setAttr("auto_shard_policy", opts.autoShardPolicy);
+        }
+      }
+    }
+    return new AutoShardDataset(opBuilder.build());
+  }
+  
+  /**
+   * @param autoShardPolicy 
+   */
+  public static Options autoShardPolicy(Long autoShardPolicy) {
+    return new Options().autoShardPolicy(autoShardPolicy);
+  }
+  
+  /**
+   */
+  public Output<?> handle() {
+    return handle;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) handle;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ExperimentalAutoShardDataset";
+  
+  private Output<?> handle;
+  
+  private AutoShardDataset(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    handle = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/BytesProducedStatsDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/BytesProducedStatsDataset.java
new file mode 100644
index 00000000000..dcf5179b0c9
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/BytesProducedStatsDataset.java
@@ -0,0 +1,89 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data.experimental;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TString;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Records the bytes size of each element of `input_dataset` in a StatsAggregator.
+ */
+public final class BytesProducedStatsDataset extends RawOp implements Operand<TType> {
+  
+  /**
+   * Factory method to create a class wrapping a new BytesProducedStatsDataset operation.
+   * 
+   * @param scope current scope
+   * @param inputDataset 
+   * @param tag 
+   * @param outputTypes 
+   * @param outputShapes 
+   * @return a new instance of BytesProducedStatsDataset
+   */
+  @Endpoint(describeByClass = true)
+  public static BytesProducedStatsDataset create(Scope scope, Operand<?> inputDataset, Operand<TString> tag, List<Class<? extends TType>> outputTypes, List<Shape> outputShapes) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ExperimentalBytesProducedStatsDataset", scope.makeOpName("BytesProducedStatsDataset"));
+    opBuilder.addInput(inputDataset.asOutput());
+    opBuilder.addInput(tag.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    Class[] outputTypesArray = new Class[outputTypes.size()];
+    for (int i = 0; i < outputTypesArray.length; ++i) {
+      outputTypesArray[i] = outputTypes.get(i);
+    }
+    opBuilder.setAttr("output_types", outputTypesArray);
+    Shape[] outputShapesArray = new Shape[outputShapes.size()];
+    for (int i = 0; i < outputShapesArray.length; ++i) {
+      outputShapesArray[i] = outputShapes.get(i);
+    }
+    opBuilder.setAttr("output_shapes", outputShapesArray);
+    return new BytesProducedStatsDataset(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<?> handle() {
+    return handle;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) handle;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ExperimentalBytesProducedStatsDataset";
+  
+  private Output<?> handle;
+  
+  private BytesProducedStatsDataset(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    handle = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/CSVDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/CSVDataset.java
new file mode 100644
index 00000000000..d0498612cde
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/CSVDataset.java
@@ -0,0 +1,99 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data.experimental;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.Operands;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TBool;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.TString;
+import org.tensorflow.types.family.TType;
+
+/**
+ */
+public final class CSVDataset extends RawOp implements Operand<TType> {
+  
+  /**
+   * Factory method to create a class wrapping a new CSVDataset operation.
+   * 
+   * @param scope current scope
+   * @param filenames 
+   * @param compressionType 
+   * @param bufferSize 
+   * @param header 
+   * @param fieldDelim 
+   * @param useQuoteDelim 
+   * @param naValue 
+   * @param selectCols 
+   * @param recordDefaults 
+   * @param outputShapes 
+   * @return a new instance of CSVDataset
+   */
+  @Endpoint(describeByClass = true)
+  public static CSVDataset create(Scope scope, Operand<TString> filenames, Operand<TString> compressionType, Operand<TInt64> bufferSize, Operand<TBool> header, Operand<TString> fieldDelim, Operand<TBool> useQuoteDelim, Operand<TString> naValue, Operand<TInt64> selectCols, Iterable<Operand<?>> recordDefaults, List<Shape> outputShapes) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ExperimentalCSVDataset", scope.makeOpName("CSVDataset"));
+    opBuilder.addInput(filenames.asOutput());
+    opBuilder.addInput(compressionType.asOutput());
+    opBuilder.addInput(bufferSize.asOutput());
+    opBuilder.addInput(header.asOutput());
+    opBuilder.addInput(fieldDelim.asOutput());
+    opBuilder.addInput(useQuoteDelim.asOutput());
+    opBuilder.addInput(naValue.asOutput());
+    opBuilder.addInput(selectCols.asOutput());
+    opBuilder.addInputList(Operands.asOutputs(recordDefaults));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    Shape[] outputShapesArray = new Shape[outputShapes.size()];
+    for (int i = 0; i < outputShapesArray.length; ++i) {
+      outputShapesArray[i] = outputShapes.get(i);
+    }
+    opBuilder.setAttr("output_shapes", outputShapesArray);
+    return new CSVDataset(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<?> handle() {
+    return handle;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) handle;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ExperimentalCSVDataset";
+  
+  private Output<?> handle;
+  
+  private CSVDataset(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    handle = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/ChooseFastestDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/ChooseFastestDataset.java
new file mode 100644
index 00000000000..aee0c946f14
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/ChooseFastestDataset.java
@@ -0,0 +1,88 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data.experimental;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.Operands;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ */
+public final class ChooseFastestDataset extends RawOp implements Operand<TType> {
+  
+  /**
+   * Factory method to create a class wrapping a new ChooseFastestDataset operation.
+   * 
+   * @param scope current scope
+   * @param inputDatasets 
+   * @param numExperiments 
+   * @param outputTypes 
+   * @param outputShapes 
+   * @return a new instance of ChooseFastestDataset
+   */
+  @Endpoint(describeByClass = true)
+  public static ChooseFastestDataset create(Scope scope, Iterable<Operand<?>> inputDatasets, Long numExperiments, List<Class<? extends TType>> outputTypes, List<Shape> outputShapes) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ExperimentalChooseFastestDataset", scope.makeOpName("ChooseFastestDataset"));
+    opBuilder.addInputList(Operands.asOutputs(inputDatasets));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("num_experiments", numExperiments);
+    Class[] outputTypesArray = new Class[outputTypes.size()];
+    for (int i = 0; i < outputTypesArray.length; ++i) {
+      outputTypesArray[i] = outputTypes.get(i);
+    }
+    opBuilder.setAttr("output_types", outputTypesArray);
+    Shape[] outputShapesArray = new Shape[outputShapes.size()];
+    for (int i = 0; i < outputShapesArray.length; ++i) {
+      outputShapesArray[i] = outputShapes.get(i);
+    }
+    opBuilder.setAttr("output_shapes", outputShapesArray);
+    return new ChooseFastestDataset(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<?> handle() {
+    return handle;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) handle;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ExperimentalChooseFastestDataset";
+  
+  private Output<?> handle;
+  
+  private ChooseFastestDataset(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    handle = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/CompressElement.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/CompressElement.java
new file mode 100644
index 00000000000..9e4bfb34a8b
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/CompressElement.java
@@ -0,0 +1,73 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data.experimental;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.Operands;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Compresses a dataset element.
+ */
+public final class CompressElement extends RawOp implements Operand<TType> {
+  
+  /**
+   * Factory method to create a class wrapping a new CompressElement operation.
+   * 
+   * @param scope current scope
+   * @param components 
+   * @return a new instance of CompressElement
+   */
+  @Endpoint(describeByClass = true)
+  public static CompressElement create(Scope scope, Iterable<Operand<?>> components) {
+    OperationBuilder opBuilder = scope.env().opBuilder("CompressElement", scope.makeOpName("CompressElement"));
+    opBuilder.addInputList(Operands.asOutputs(components));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new CompressElement(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<?> compressed() {
+    return compressed;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) compressed;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "CompressElement";
+  
+  private Output<?> compressed;
+  
+  private CompressElement(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    compressed = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/DataServiceDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/DataServiceDataset.java
new file mode 100644
index 00000000000..3330b2163e3
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/DataServiceDataset.java
@@ -0,0 +1,134 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data.experimental;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.TString;
+import org.tensorflow.types.family.TType;
+
+/**
+ */
+@Operator(group = "data.experimental")
+public final class DataServiceDataset extends RawOp implements Operand<TType> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.data.experimental.DataServiceDataset}
+   */
+  public static class Options {
+    
+    /**
+     * @param taskRefreshIntervalHintMs 
+     */
+    public Options taskRefreshIntervalHintMs(Long taskRefreshIntervalHintMs) {
+      this.taskRefreshIntervalHintMs = taskRefreshIntervalHintMs;
+      return this;
+    }
+    
+    private Long taskRefreshIntervalHintMs;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new DataServiceDataset operation.
+   * 
+   * @param scope current scope
+   * @param datasetId 
+   * @param processingMode 
+   * @param address 
+   * @param protocol 
+   * @param jobName 
+   * @param maxOutstandingRequests 
+   * @param iterationCounter 
+   * @param outputTypes 
+   * @param outputShapes 
+   * @param options carries optional attributes values
+   * @return a new instance of DataServiceDataset
+   */
+  @Endpoint(describeByClass = true)
+  public static DataServiceDataset create(Scope scope, Operand<TInt64> datasetId, Operand<TString> processingMode, Operand<TString> address, Operand<TString> protocol, Operand<TString> jobName, Operand<TInt64> maxOutstandingRequests, Operand<?> iterationCounter, List<Class<? extends TType>> outputTypes, List<Shape> outputShapes, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("DataServiceDataset", scope.makeOpName("DataServiceDataset"));
+    opBuilder.addInput(datasetId.asOutput());
+    opBuilder.addInput(processingMode.asOutput());
+    opBuilder.addInput(address.asOutput());
+    opBuilder.addInput(protocol.asOutput());
+    opBuilder.addInput(jobName.asOutput());
+    opBuilder.addInput(maxOutstandingRequests.asOutput());
+    opBuilder.addInput(iterationCounter.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    Class[] outputTypesArray = new Class[outputTypes.size()];
+    for (int i = 0; i < outputTypesArray.length; ++i) {
+      outputTypesArray[i] = outputTypes.get(i);
+    }
+    opBuilder.setAttr("output_types", outputTypesArray);
+    Shape[] outputShapesArray = new Shape[outputShapes.size()];
+    for (int i = 0; i < outputShapesArray.length; ++i) {
+      outputShapesArray[i] = outputShapes.get(i);
+    }
+    opBuilder.setAttr("output_shapes", outputShapesArray);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.taskRefreshIntervalHintMs != null) {
+          opBuilder.setAttr("task_refresh_interval_hint_ms", opts.taskRefreshIntervalHintMs);
+        }
+      }
+    }
+    return new DataServiceDataset(opBuilder.build());
+  }
+  
+  /**
+   * @param taskRefreshIntervalHintMs 
+   */
+  public static Options taskRefreshIntervalHintMs(Long taskRefreshIntervalHintMs) {
+    return new Options().taskRefreshIntervalHintMs(taskRefreshIntervalHintMs);
+  }
+  
+  /**
+   */
+  public Output<?> handle() {
+    return handle;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) handle;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "DataServiceDataset";
+  
+  private Output<?> handle;
+  
+  private DataServiceDataset(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    handle = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/DatasetCardinality.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/DatasetCardinality.java
new file mode 100644
index 00000000000..81ca505ca91
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/DatasetCardinality.java
@@ -0,0 +1,75 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data.experimental;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt64;
+
+/**
+ * Returns the cardinality of `input_dataset`.
+ * <p>
+ * Returns the cardinality of `input_dataset`.
+ */
+public final class DatasetCardinality extends RawOp implements Operand<TInt64> {
+  
+  /**
+   * Factory method to create a class wrapping a new DatasetCardinality operation.
+   * 
+   * @param scope current scope
+   * @param inputDataset A variant tensor representing the dataset to return cardinality for.
+   * @return a new instance of DatasetCardinality
+   */
+  @Endpoint(describeByClass = true)
+  public static DatasetCardinality create(Scope scope, Operand<?> inputDataset) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ExperimentalDatasetCardinality", scope.makeOpName("DatasetCardinality"));
+    opBuilder.addInput(inputDataset.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new DatasetCardinality(opBuilder.build());
+  }
+  
+  /**
+   * The cardinality of `input_dataset`. Named constants are used to represent
+   * infinite and unknown cardinality.
+   */
+  public Output<TInt64> cardinality() {
+    return cardinality;
+  }
+  
+  @Override
+  public Output<TInt64> asOutput() {
+    return cardinality;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ExperimentalDatasetCardinality";
+  
+  private Output<TInt64> cardinality;
+  
+  private DatasetCardinality(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    cardinality = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/DatasetToTFRecord.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/DatasetToTFRecord.java
new file mode 100644
index 00000000000..6e0a0b8f2dc
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/DatasetToTFRecord.java
@@ -0,0 +1,60 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data.experimental;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TString;
+
+/**
+ * Writes the given dataset to the given file using the TFRecord format.
+ */
+public final class DatasetToTFRecord extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new DatasetToTFRecord operation.
+   * 
+   * @param scope current scope
+   * @param inputDataset A variant tensor representing the dataset to write.
+   * @param filename A scalar string tensor representing the filename to use.
+   * @param compressionType A scalar string tensor containing either (i) the empty string (no
+   * compression), (ii) "ZLIB", or (iii) "GZIP".
+   * @return a new instance of DatasetToTFRecord
+   */
+  @Endpoint(describeByClass = true)
+  public static DatasetToTFRecord create(Scope scope, Operand<?> inputDataset, Operand<TString> filename, Operand<TString> compressionType) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ExperimentalDatasetToTFRecord", scope.makeOpName("DatasetToTFRecord"));
+    opBuilder.addInput(inputDataset.asOutput());
+    opBuilder.addInput(filename.asOutput());
+    opBuilder.addInput(compressionType.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new DatasetToTFRecord(opBuilder.build());
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ExperimentalDatasetToTFRecord";
+  
+  private DatasetToTFRecord(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/DenseToSparseBatchDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/DenseToSparseBatchDataset.java
new file mode 100644
index 00000000000..ff2910a2e38
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/DenseToSparseBatchDataset.java
@@ -0,0 +1,94 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data.experimental;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Creates a dataset that batches input elements into a SparseTensor.
+ */
+public final class DenseToSparseBatchDataset extends RawOp implements Operand<TType> {
+  
+  /**
+   * Factory method to create a class wrapping a new DenseToSparseBatchDataset operation.
+   * 
+   * @param scope current scope
+   * @param inputDataset A handle to an input dataset. Must have a single component.
+   * @param batchSize A scalar representing the number of elements to accumulate in a
+   * batch.
+   * @param rowShape A vector representing the dense shape of each row in the produced
+   * SparseTensor. The shape may be partially specified, using `-1` to indicate
+   * that a particular dimension should use the maximum size of all batch elements.
+   * @param outputTypes 
+   * @param outputShapes 
+   * @return a new instance of DenseToSparseBatchDataset
+   */
+  @Endpoint(describeByClass = true)
+  public static DenseToSparseBatchDataset create(Scope scope, Operand<?> inputDataset, Operand<TInt64> batchSize, Operand<TInt64> rowShape, List<Class<? extends TType>> outputTypes, List<Shape> outputShapes) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ExperimentalDenseToSparseBatchDataset", scope.makeOpName("DenseToSparseBatchDataset"));
+    opBuilder.addInput(inputDataset.asOutput());
+    opBuilder.addInput(batchSize.asOutput());
+    opBuilder.addInput(rowShape.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    Class[] outputTypesArray = new Class[outputTypes.size()];
+    for (int i = 0; i < outputTypesArray.length; ++i) {
+      outputTypesArray[i] = outputTypes.get(i);
+    }
+    opBuilder.setAttr("output_types", outputTypesArray);
+    Shape[] outputShapesArray = new Shape[outputShapes.size()];
+    for (int i = 0; i < outputShapesArray.length; ++i) {
+      outputShapesArray[i] = outputShapes.get(i);
+    }
+    opBuilder.setAttr("output_shapes", outputShapesArray);
+    return new DenseToSparseBatchDataset(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<?> handle() {
+    return handle;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) handle;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ExperimentalDenseToSparseBatchDataset";
+  
+  private Output<?> handle;
+  
+  private DenseToSparseBatchDataset(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    handle = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/DirectedInterleaveDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/DirectedInterleaveDataset.java
new file mode 100644
index 00000000000..50ee8a37fcb
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/DirectedInterleaveDataset.java
@@ -0,0 +1,91 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data.experimental;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.Operands;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * A substitute for `InterleaveDataset` on a fixed list of `N` datasets.
+ */
+public final class DirectedInterleaveDataset extends RawOp implements Operand<TType> {
+  
+  /**
+   * Factory method to create a class wrapping a new DirectedInterleaveDataset operation.
+   * 
+   * @param scope current scope
+   * @param selectorInputDataset A dataset of scalar `DT_INT64` elements that determines which of the
+   * `N` data inputs should produce the next output element.
+   * @param dataInputDatasets `N` datasets with the same type that will be interleaved according to
+   * the values of `selector_input_dataset`.
+   * @param outputTypes 
+   * @param outputShapes 
+   * @return a new instance of DirectedInterleaveDataset
+   */
+  @Endpoint(describeByClass = true)
+  public static DirectedInterleaveDataset create(Scope scope, Operand<?> selectorInputDataset, Iterable<Operand<?>> dataInputDatasets, List<Class<? extends TType>> outputTypes, List<Shape> outputShapes) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ExperimentalDirectedInterleaveDataset", scope.makeOpName("DirectedInterleaveDataset"));
+    opBuilder.addInput(selectorInputDataset.asOutput());
+    opBuilder.addInputList(Operands.asOutputs(dataInputDatasets));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    Class[] outputTypesArray = new Class[outputTypes.size()];
+    for (int i = 0; i < outputTypesArray.length; ++i) {
+      outputTypesArray[i] = outputTypes.get(i);
+    }
+    opBuilder.setAttr("output_types", outputTypesArray);
+    Shape[] outputShapesArray = new Shape[outputShapes.size()];
+    for (int i = 0; i < outputShapesArray.length; ++i) {
+      outputShapesArray[i] = outputShapes.get(i);
+    }
+    opBuilder.setAttr("output_shapes", outputShapesArray);
+    return new DirectedInterleaveDataset(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<?> handle() {
+    return handle;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) handle;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ExperimentalDirectedInterleaveDataset";
+  
+  private Output<?> handle;
+  
+  private DirectedInterleaveDataset(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    handle = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/DummyIterationCounter.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/DummyIterationCounter.java
new file mode 100644
index 00000000000..72f83847285
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/DummyIterationCounter.java
@@ -0,0 +1,69 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data.experimental;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ */
+public final class DummyIterationCounter extends RawOp implements Operand<TType> {
+  
+  /**
+   * Factory method to create a class wrapping a new DummyIterationCounter operation.
+   * 
+   * @param scope current scope
+   * @return a new instance of DummyIterationCounter
+   */
+  @Endpoint(describeByClass = true)
+  public static DummyIterationCounter create(Scope scope) {
+    OperationBuilder opBuilder = scope.env().opBuilder("DummyIterationCounter", scope.makeOpName("DummyIterationCounter"));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new DummyIterationCounter(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<?> handle() {
+    return handle;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) handle;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "DummyIterationCounter";
+  
+  private Output<?> handle;
+  
+  private DummyIterationCounter(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    handle = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/IgnoreErrorsDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/IgnoreErrorsDataset.java
new file mode 100644
index 00000000000..524814790f5
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/IgnoreErrorsDataset.java
@@ -0,0 +1,86 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data.experimental;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Creates a dataset that contains the elements of `input_dataset` ignoring errors.
+ */
+public final class IgnoreErrorsDataset extends RawOp implements Operand<TType> {
+  
+  /**
+   * Factory method to create a class wrapping a new IgnoreErrorsDataset operation.
+   * 
+   * @param scope current scope
+   * @param inputDataset 
+   * @param outputTypes 
+   * @param outputShapes 
+   * @return a new instance of IgnoreErrorsDataset
+   */
+  @Endpoint(describeByClass = true)
+  public static IgnoreErrorsDataset create(Scope scope, Operand<?> inputDataset, List<Class<? extends TType>> outputTypes, List<Shape> outputShapes) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ExperimentalIgnoreErrorsDataset", scope.makeOpName("IgnoreErrorsDataset"));
+    opBuilder.addInput(inputDataset.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    Class[] outputTypesArray = new Class[outputTypes.size()];
+    for (int i = 0; i < outputTypesArray.length; ++i) {
+      outputTypesArray[i] = outputTypes.get(i);
+    }
+    opBuilder.setAttr("output_types", outputTypesArray);
+    Shape[] outputShapesArray = new Shape[outputShapes.size()];
+    for (int i = 0; i < outputShapesArray.length; ++i) {
+      outputShapesArray[i] = outputShapes.get(i);
+    }
+    opBuilder.setAttr("output_shapes", outputShapesArray);
+    return new IgnoreErrorsDataset(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<?> handle() {
+    return handle;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) handle;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ExperimentalIgnoreErrorsDataset";
+  
+  private Output<?> handle;
+  
+  private IgnoreErrorsDataset(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    handle = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/IteratorGetDevice.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/IteratorGetDevice.java
new file mode 100644
index 00000000000..7602eca94fc
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/IteratorGetDevice.java
@@ -0,0 +1,71 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data.experimental;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TString;
+
+/**
+ * Returns the name of the device on which `resource` has been placed.
+ */
+public final class IteratorGetDevice extends RawOp implements Operand<TString> {
+  
+  /**
+   * Factory method to create a class wrapping a new IteratorGetDevice operation.
+   * 
+   * @param scope current scope
+   * @param resource 
+   * @return a new instance of IteratorGetDevice
+   */
+  @Endpoint(describeByClass = true)
+  public static IteratorGetDevice create(Scope scope, Operand<?> resource) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ExperimentalIteratorGetDevice", scope.makeOpName("IteratorGetDevice"));
+    opBuilder.addInput(resource.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new IteratorGetDevice(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<TString> device() {
+    return device;
+  }
+  
+  @Override
+  public Output<TString> asOutput() {
+    return device;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ExperimentalIteratorGetDevice";
+  
+  private Output<TString> device;
+  
+  private IteratorGetDevice(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    device = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/LatencyStatsDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/LatencyStatsDataset.java
new file mode 100644
index 00000000000..cd9f0a2c498
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/LatencyStatsDataset.java
@@ -0,0 +1,89 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data.experimental;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TString;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Records the latency of producing `input_dataset` elements in a StatsAggregator.
+ */
+public final class LatencyStatsDataset extends RawOp implements Operand<TType> {
+  
+  /**
+   * Factory method to create a class wrapping a new LatencyStatsDataset operation.
+   * 
+   * @param scope current scope
+   * @param inputDataset 
+   * @param tag 
+   * @param outputTypes 
+   * @param outputShapes 
+   * @return a new instance of LatencyStatsDataset
+   */
+  @Endpoint(describeByClass = true)
+  public static LatencyStatsDataset create(Scope scope, Operand<?> inputDataset, Operand<TString> tag, List<Class<? extends TType>> outputTypes, List<Shape> outputShapes) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ExperimentalLatencyStatsDataset", scope.makeOpName("LatencyStatsDataset"));
+    opBuilder.addInput(inputDataset.asOutput());
+    opBuilder.addInput(tag.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    Class[] outputTypesArray = new Class[outputTypes.size()];
+    for (int i = 0; i < outputTypesArray.length; ++i) {
+      outputTypesArray[i] = outputTypes.get(i);
+    }
+    opBuilder.setAttr("output_types", outputTypesArray);
+    Shape[] outputShapesArray = new Shape[outputShapes.size()];
+    for (int i = 0; i < outputShapesArray.length; ++i) {
+      outputShapesArray[i] = outputShapes.get(i);
+    }
+    opBuilder.setAttr("output_shapes", outputShapesArray);
+    return new LatencyStatsDataset(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<?> handle() {
+    return handle;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) handle;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ExperimentalLatencyStatsDataset";
+  
+  private Output<?> handle;
+  
+  private LatencyStatsDataset(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    handle = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/LmdbDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/LmdbDataset.java
new file mode 100644
index 00000000000..3ec97c34b61
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/LmdbDataset.java
@@ -0,0 +1,86 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data.experimental;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TString;
+import org.tensorflow.types.family.TType;
+
+/**
+ */
+public final class LmdbDataset extends RawOp implements Operand<TType> {
+  
+  /**
+   * Factory method to create a class wrapping a new LmdbDataset operation.
+   * 
+   * @param scope current scope
+   * @param filenames 
+   * @param outputTypes 
+   * @param outputShapes 
+   * @return a new instance of LmdbDataset
+   */
+  @Endpoint(describeByClass = true)
+  public static LmdbDataset create(Scope scope, Operand<TString> filenames, List<Class<? extends TType>> outputTypes, List<Shape> outputShapes) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ExperimentalLMDBDataset", scope.makeOpName("LmdbDataset"));
+    opBuilder.addInput(filenames.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    Class[] outputTypesArray = new Class[outputTypes.size()];
+    for (int i = 0; i < outputTypesArray.length; ++i) {
+      outputTypesArray[i] = outputTypes.get(i);
+    }
+    opBuilder.setAttr("output_types", outputTypesArray);
+    Shape[] outputShapesArray = new Shape[outputShapes.size()];
+    for (int i = 0; i < outputShapesArray.length; ++i) {
+      outputShapesArray[i] = outputShapes.get(i);
+    }
+    opBuilder.setAttr("output_shapes", outputShapesArray);
+    return new LmdbDataset(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<?> handle() {
+    return handle;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) handle;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ExperimentalLMDBDataset";
+  
+  private Output<?> handle;
+  
+  private LmdbDataset(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    handle = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/MatchingFilesDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/MatchingFilesDataset.java
new file mode 100644
index 00000000000..e2e29feeba4
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/MatchingFilesDataset.java
@@ -0,0 +1,72 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data.experimental;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TString;
+import org.tensorflow.types.family.TType;
+
+/**
+ */
+public final class MatchingFilesDataset extends RawOp implements Operand<TType> {
+  
+  /**
+   * Factory method to create a class wrapping a new MatchingFilesDataset operation.
+   * 
+   * @param scope current scope
+   * @param patterns 
+   * @return a new instance of MatchingFilesDataset
+   */
+  @Endpoint(describeByClass = true)
+  public static MatchingFilesDataset create(Scope scope, Operand<TString> patterns) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ExperimentalMatchingFilesDataset", scope.makeOpName("MatchingFilesDataset"));
+    opBuilder.addInput(patterns.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new MatchingFilesDataset(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<?> handle() {
+    return handle;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) handle;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ExperimentalMatchingFilesDataset";
+  
+  private Output<?> handle;
+  
+  private MatchingFilesDataset(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    handle = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/MaxIntraOpParallelismDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/MaxIntraOpParallelismDataset.java
new file mode 100644
index 00000000000..88d924e5e43
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/MaxIntraOpParallelismDataset.java
@@ -0,0 +1,89 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data.experimental;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Creates a dataset that overrides the maximum intra-op parallelism.
+ */
+public final class MaxIntraOpParallelismDataset extends RawOp implements Operand<TType> {
+  
+  /**
+   * Factory method to create a class wrapping a new MaxIntraOpParallelismDataset operation.
+   * 
+   * @param scope current scope
+   * @param inputDataset 
+   * @param maxIntraOpParallelism Identifies the maximum intra-op parallelism to use.
+   * @param outputTypes 
+   * @param outputShapes 
+   * @return a new instance of MaxIntraOpParallelismDataset
+   */
+  @Endpoint(describeByClass = true)
+  public static MaxIntraOpParallelismDataset create(Scope scope, Operand<?> inputDataset, Operand<TInt64> maxIntraOpParallelism, List<Class<? extends TType>> outputTypes, List<Shape> outputShapes) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ExperimentalMaxIntraOpParallelismDataset", scope.makeOpName("MaxIntraOpParallelismDataset"));
+    opBuilder.addInput(inputDataset.asOutput());
+    opBuilder.addInput(maxIntraOpParallelism.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    Class[] outputTypesArray = new Class[outputTypes.size()];
+    for (int i = 0; i < outputTypesArray.length; ++i) {
+      outputTypesArray[i] = outputTypes.get(i);
+    }
+    opBuilder.setAttr("output_types", outputTypesArray);
+    Shape[] outputShapesArray = new Shape[outputShapes.size()];
+    for (int i = 0; i < outputShapesArray.length; ++i) {
+      outputShapesArray[i] = outputShapes.get(i);
+    }
+    opBuilder.setAttr("output_shapes", outputShapesArray);
+    return new MaxIntraOpParallelismDataset(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<?> handle() {
+    return handle;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) handle;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ExperimentalMaxIntraOpParallelismDataset";
+  
+  private Output<?> handle;
+  
+  private MaxIntraOpParallelismDataset(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    handle = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/NonSerializableDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/NonSerializableDataset.java
new file mode 100644
index 00000000000..45866ff08b4
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/NonSerializableDataset.java
@@ -0,0 +1,85 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data.experimental;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ */
+public final class NonSerializableDataset extends RawOp implements Operand<TType> {
+  
+  /**
+   * Factory method to create a class wrapping a new NonSerializableDataset operation.
+   * 
+   * @param scope current scope
+   * @param inputDataset 
+   * @param outputTypes 
+   * @param outputShapes 
+   * @return a new instance of NonSerializableDataset
+   */
+  @Endpoint(describeByClass = true)
+  public static NonSerializableDataset create(Scope scope, Operand<?> inputDataset, List<Class<? extends TType>> outputTypes, List<Shape> outputShapes) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ExperimentalNonSerializableDataset", scope.makeOpName("NonSerializableDataset"));
+    opBuilder.addInput(inputDataset.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    Class[] outputTypesArray = new Class[outputTypes.size()];
+    for (int i = 0; i < outputTypesArray.length; ++i) {
+      outputTypesArray[i] = outputTypes.get(i);
+    }
+    opBuilder.setAttr("output_types", outputTypesArray);
+    Shape[] outputShapesArray = new Shape[outputShapes.size()];
+    for (int i = 0; i < outputShapesArray.length; ++i) {
+      outputShapesArray[i] = outputShapes.get(i);
+    }
+    opBuilder.setAttr("output_shapes", outputShapesArray);
+    return new NonSerializableDataset(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<?> handle() {
+    return handle;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) handle;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ExperimentalNonSerializableDataset";
+  
+  private Output<?> handle;
+  
+  private NonSerializableDataset(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    handle = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/ParseExampleDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/ParseExampleDataset.java
new file mode 100644
index 00000000000..c4602914a42
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/ParseExampleDataset.java
@@ -0,0 +1,207 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data.experimental;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.Operands;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Transforms `input_dataset` containing `Example` protos as vectors of DT_STRING into a dataset of `Tensor` or `SparseTensor` objects representing the parsed features.
+ */
+public final class ParseExampleDataset extends RawOp implements Operand<TType> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.data.experimental.ParseExampleDataset}
+   */
+  public static class Options {
+    
+    /**
+     * @param deterministic A string indicating the op-level determinism to use. Deterministic controls
+     * whether the dataset is allowed to return elements out of order if the next
+     * element to be returned isn't available, but a later element is. Options are
+     * "true", "false", and "default". "default" indicates that determinism should be
+     * decided by the `experimental_deterministic` parameter of `tf.data.Options`.
+     */
+    public Options deterministic(String deterministic) {
+      this.deterministic = deterministic;
+      return this;
+    }
+    
+    /**
+     * @param raggedKeys 
+     */
+    public Options raggedKeys(List<String> raggedKeys) {
+      this.raggedKeys = raggedKeys;
+      return this;
+    }
+    
+    private String deterministic;
+    private List<String> raggedKeys;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new ParseExampleDataset operation.
+   * 
+   * @param scope current scope
+   * @param inputDataset 
+   * @param numParallelCalls 
+   * @param denseDefaults A dict mapping string keys to `Tensor`s.
+   * The keys of the dict must match the dense_keys of the feature.
+   * @param sparseKeys A list of string keys in the examples features.
+   * The results for these keys will be returned as `SparseTensor` objects.
+   * @param denseKeys A list of Ndense string Tensors (scalars).
+   * The keys expected in the Examples features associated with dense values.
+   * @param sparseTypes A list of `DTypes` of the same length as `sparse_keys`.
+   * Only `tf.float32` (`FloatList`), `tf.int64` (`Int64List`),
+   * and `tf.string` (`BytesList`) are supported.
+   * @param denseShapes List of tuples with the same length as `dense_keys`.
+   * The shape of the data for each dense feature referenced by `dense_keys`.
+   * Required for any input tensors identified by `dense_keys`.  Must be
+   * either fully defined, or may contain an unknown first dimension.
+   * An unknown first dimension means the feature is treated as having
+   * a variable number of blocks, and the output shape along this dimension
+   * is considered unknown at graph build time.  Padding is applied for
+   * minibatch elements smaller than the maximum number of blocks for the
+   * given feature along this dimension.
+   * @param outputTypes The type list for the return values.
+   * @param outputShapes The list of shapes being produced.
+   * @param raggedValueTypes 
+   * @param raggedSplitTypes 
+   * @param options carries optional attributes values
+   * @return a new instance of ParseExampleDataset
+   */
+  @Endpoint(describeByClass = true)
+  public static ParseExampleDataset create(Scope scope, Operand<?> inputDataset, Operand<TInt64> numParallelCalls, Iterable<Operand<?>> denseDefaults, List<String> sparseKeys, List<String> denseKeys, List<Class<? extends TType>> sparseTypes, List<Shape> denseShapes, List<Class<? extends TType>> outputTypes, List<Shape> outputShapes, List<Class<? extends TType>> raggedValueTypes, List<Class<? extends TNumber>> raggedSplitTypes, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ParseExampleDatasetV2", scope.makeOpName("ParseExampleDataset"));
+    opBuilder.addInput(inputDataset.asOutput());
+    opBuilder.addInput(numParallelCalls.asOutput());
+    opBuilder.addInputList(Operands.asOutputs(denseDefaults));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    String[] sparseKeysArray = new String[sparseKeys.size()];
+    for (int i = 0; i < sparseKeysArray.length; ++i) {
+      sparseKeysArray[i] = sparseKeys.get(i);
+    }
+    opBuilder.setAttr("sparse_keys", sparseKeysArray);
+    String[] denseKeysArray = new String[denseKeys.size()];
+    for (int i = 0; i < denseKeysArray.length; ++i) {
+      denseKeysArray[i] = denseKeys.get(i);
+    }
+    opBuilder.setAttr("dense_keys", denseKeysArray);
+    Class[] sparseTypesArray = new Class[sparseTypes.size()];
+    for (int i = 0; i < sparseTypesArray.length; ++i) {
+      sparseTypesArray[i] = sparseTypes.get(i);
+    }
+    opBuilder.setAttr("sparse_types", sparseTypesArray);
+    Shape[] denseShapesArray = new Shape[denseShapes.size()];
+    for (int i = 0; i < denseShapesArray.length; ++i) {
+      denseShapesArray[i] = denseShapes.get(i);
+    }
+    opBuilder.setAttr("dense_shapes", denseShapesArray);
+    Class[] outputTypesArray = new Class[outputTypes.size()];
+    for (int i = 0; i < outputTypesArray.length; ++i) {
+      outputTypesArray[i] = outputTypes.get(i);
+    }
+    opBuilder.setAttr("output_types", outputTypesArray);
+    Shape[] outputShapesArray = new Shape[outputShapes.size()];
+    for (int i = 0; i < outputShapesArray.length; ++i) {
+      outputShapesArray[i] = outputShapes.get(i);
+    }
+    opBuilder.setAttr("output_shapes", outputShapesArray);
+    Class[] raggedValueTypesArray = new Class[raggedValueTypes.size()];
+    for (int i = 0; i < raggedValueTypesArray.length; ++i) {
+      raggedValueTypesArray[i] = raggedValueTypes.get(i);
+    }
+    opBuilder.setAttr("ragged_value_types", raggedValueTypesArray);
+    Class[] raggedSplitTypesArray = new Class[raggedSplitTypes.size()];
+    for (int i = 0; i < raggedSplitTypesArray.length; ++i) {
+      raggedSplitTypesArray[i] = raggedSplitTypes.get(i);
+    }
+    opBuilder.setAttr("ragged_split_types", raggedSplitTypesArray);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.deterministic != null) {
+          opBuilder.setAttr("deterministic", opts.deterministic);
+        }
+        if (opts.raggedKeys != null) {
+          String[] raggedKeysArray = new String[opts.raggedKeys.size()];
+          for (int i = 0; i < raggedKeysArray.length; ++i) {
+            raggedKeysArray[i] = opts.raggedKeys.get(i);
+          }
+          opBuilder.setAttr("ragged_keys", raggedKeysArray);
+        }
+      }
+    }
+    return new ParseExampleDataset(opBuilder.build());
+  }
+  
+  /**
+   * @param deterministic A string indicating the op-level determinism to use. Deterministic controls
+   * whether the dataset is allowed to return elements out of order if the next
+   * element to be returned isn't available, but a later element is. Options are
+   * "true", "false", and "default". "default" indicates that determinism should be
+   * decided by the `experimental_deterministic` parameter of `tf.data.Options`.
+   */
+  public static Options deterministic(String deterministic) {
+    return new Options().deterministic(deterministic);
+  }
+  
+  /**
+   * @param raggedKeys 
+   */
+  public static Options raggedKeys(List<String> raggedKeys) {
+    return new Options().raggedKeys(raggedKeys);
+  }
+  
+  /**
+   */
+  public Output<?> handle() {
+    return handle;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) handle;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ParseExampleDatasetV2";
+  
+  private Output<?> handle;
+  
+  private ParseExampleDataset(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    handle = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/PrivateThreadPoolDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/PrivateThreadPoolDataset.java
new file mode 100644
index 00000000000..7d85832bc3e
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/PrivateThreadPoolDataset.java
@@ -0,0 +1,89 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data.experimental;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Creates a dataset that uses a custom thread pool to compute `input_dataset`.
+ */
+public final class PrivateThreadPoolDataset extends RawOp implements Operand<TType> {
+  
+  /**
+   * Factory method to create a class wrapping a new PrivateThreadPoolDataset operation.
+   * 
+   * @param scope current scope
+   * @param inputDataset 
+   * @param numThreads Identifies the number of threads to use for the private threadpool.
+   * @param outputTypes 
+   * @param outputShapes 
+   * @return a new instance of PrivateThreadPoolDataset
+   */
+  @Endpoint(describeByClass = true)
+  public static PrivateThreadPoolDataset create(Scope scope, Operand<?> inputDataset, Operand<TInt64> numThreads, List<Class<? extends TType>> outputTypes, List<Shape> outputShapes) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ExperimentalPrivateThreadPoolDataset", scope.makeOpName("PrivateThreadPoolDataset"));
+    opBuilder.addInput(inputDataset.asOutput());
+    opBuilder.addInput(numThreads.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    Class[] outputTypesArray = new Class[outputTypes.size()];
+    for (int i = 0; i < outputTypesArray.length; ++i) {
+      outputTypesArray[i] = outputTypes.get(i);
+    }
+    opBuilder.setAttr("output_types", outputTypesArray);
+    Shape[] outputShapesArray = new Shape[outputShapes.size()];
+    for (int i = 0; i < outputShapesArray.length; ++i) {
+      outputShapesArray[i] = outputShapes.get(i);
+    }
+    opBuilder.setAttr("output_shapes", outputShapesArray);
+    return new PrivateThreadPoolDataset(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<?> handle() {
+    return handle;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) handle;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ExperimentalPrivateThreadPoolDataset";
+  
+  private Output<?> handle;
+  
+  private PrivateThreadPoolDataset(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    handle = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/RandomDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/RandomDataset.java
new file mode 100644
index 00000000000..f2e06ee1b97
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/RandomDataset.java
@@ -0,0 +1,91 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data.experimental;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Creates a Dataset that returns pseudorandom numbers.
+ */
+public final class RandomDataset extends RawOp implements Operand<TType> {
+  
+  /**
+   * Factory method to create a class wrapping a new RandomDataset operation.
+   * 
+   * @param scope current scope
+   * @param seed A scalar seed for the random number generator. If either seed or
+   * seed2 is set to be non-zero, the random number generator is seeded
+   * by the given seed.  Otherwise, a random seed is used.
+   * @param seed2 A second scalar seed to avoid seed collision.
+   * @param outputTypes 
+   * @param outputShapes 
+   * @return a new instance of RandomDataset
+   */
+  @Endpoint(describeByClass = true)
+  public static RandomDataset create(Scope scope, Operand<TInt64> seed, Operand<TInt64> seed2, List<Class<? extends TType>> outputTypes, List<Shape> outputShapes) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ExperimentalRandomDataset", scope.makeOpName("RandomDataset"));
+    opBuilder.addInput(seed.asOutput());
+    opBuilder.addInput(seed2.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    Class[] outputTypesArray = new Class[outputTypes.size()];
+    for (int i = 0; i < outputTypesArray.length; ++i) {
+      outputTypesArray[i] = outputTypes.get(i);
+    }
+    opBuilder.setAttr("output_types", outputTypesArray);
+    Shape[] outputShapesArray = new Shape[outputShapes.size()];
+    for (int i = 0; i < outputShapesArray.length; ++i) {
+      outputShapesArray[i] = outputShapes.get(i);
+    }
+    opBuilder.setAttr("output_shapes", outputShapesArray);
+    return new RandomDataset(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<?> handle() {
+    return handle;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) handle;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ExperimentalRandomDataset";
+  
+  private Output<?> handle;
+  
+  private RandomDataset(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    handle = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/RebatchDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/RebatchDataset.java
new file mode 100644
index 00000000000..c647ea32f5a
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/RebatchDataset.java
@@ -0,0 +1,128 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data.experimental;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Creates a dataset that changes the batch size.
+ * <p>
+ * Creates a dataset that changes the batch size of the dataset to current batch
+ * size // num_replicas.
+ */
+public final class RebatchDataset extends RawOp implements Operand<TType> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.data.experimental.RebatchDataset}
+   */
+  public static class Options {
+    
+    /**
+     * @param useFallback 
+     */
+    public Options useFallback(Boolean useFallback) {
+      this.useFallback = useFallback;
+      return this;
+    }
+    
+    private Boolean useFallback;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new RebatchDataset operation.
+   * 
+   * @param scope current scope
+   * @param inputDataset A variant tensor representing the input dataset.
+   * @param numReplicas A scalar representing the number of replicas to distribute this batch across. As
+   * a result of this transformation the current batch size would end up being
+   * divided  by this parameter.
+   * @param outputTypes 
+   * @param outputShapes 
+   * @param options carries optional attributes values
+   * @return a new instance of RebatchDataset
+   */
+  @Endpoint(describeByClass = true)
+  public static RebatchDataset create(Scope scope, Operand<?> inputDataset, Operand<TInt64> numReplicas, List<Class<? extends TType>> outputTypes, List<Shape> outputShapes, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ExperimentalRebatchDataset", scope.makeOpName("RebatchDataset"));
+    opBuilder.addInput(inputDataset.asOutput());
+    opBuilder.addInput(numReplicas.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    Class[] outputTypesArray = new Class[outputTypes.size()];
+    for (int i = 0; i < outputTypesArray.length; ++i) {
+      outputTypesArray[i] = outputTypes.get(i);
+    }
+    opBuilder.setAttr("output_types", outputTypesArray);
+    Shape[] outputShapesArray = new Shape[outputShapes.size()];
+    for (int i = 0; i < outputShapesArray.length; ++i) {
+      outputShapesArray[i] = outputShapes.get(i);
+    }
+    opBuilder.setAttr("output_shapes", outputShapesArray);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.useFallback != null) {
+          opBuilder.setAttr("use_fallback", opts.useFallback);
+        }
+      }
+    }
+    return new RebatchDataset(opBuilder.build());
+  }
+  
+  /**
+   * @param useFallback 
+   */
+  public static Options useFallback(Boolean useFallback) {
+    return new Options().useFallback(useFallback);
+  }
+  
+  /**
+   */
+  public Output<?> handle() {
+    return handle;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) handle;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ExperimentalRebatchDataset";
+  
+  private Output<?> handle;
+  
+  private RebatchDataset(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    handle = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/SetStatsAggregatorDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/SetStatsAggregatorDataset.java
new file mode 100644
index 00000000000..842c2fae29b
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/SetStatsAggregatorDataset.java
@@ -0,0 +1,92 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data.experimental;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TString;
+import org.tensorflow.types.family.TType;
+
+/**
+ */
+public final class SetStatsAggregatorDataset extends RawOp implements Operand<TType> {
+  
+  /**
+   * Factory method to create a class wrapping a new SetStatsAggregatorDataset operation.
+   * 
+   * @param scope current scope
+   * @param inputDataset 
+   * @param statsAggregator 
+   * @param tag 
+   * @param counterPrefix 
+   * @param outputTypes 
+   * @param outputShapes 
+   * @return a new instance of SetStatsAggregatorDataset
+   */
+  @Endpoint(describeByClass = true)
+  public static SetStatsAggregatorDataset create(Scope scope, Operand<?> inputDataset, Operand<?> statsAggregator, Operand<TString> tag, Operand<TString> counterPrefix, List<Class<? extends TType>> outputTypes, List<Shape> outputShapes) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ExperimentalSetStatsAggregatorDataset", scope.makeOpName("SetStatsAggregatorDataset"));
+    opBuilder.addInput(inputDataset.asOutput());
+    opBuilder.addInput(statsAggregator.asOutput());
+    opBuilder.addInput(tag.asOutput());
+    opBuilder.addInput(counterPrefix.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    Class[] outputTypesArray = new Class[outputTypes.size()];
+    for (int i = 0; i < outputTypesArray.length; ++i) {
+      outputTypesArray[i] = outputTypes.get(i);
+    }
+    opBuilder.setAttr("output_types", outputTypesArray);
+    Shape[] outputShapesArray = new Shape[outputShapes.size()];
+    for (int i = 0; i < outputShapesArray.length; ++i) {
+      outputShapesArray[i] = outputShapes.get(i);
+    }
+    opBuilder.setAttr("output_shapes", outputShapesArray);
+    return new SetStatsAggregatorDataset(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<?> handle() {
+    return handle;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) handle;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ExperimentalSetStatsAggregatorDataset";
+  
+  private Output<?> handle;
+  
+  private SetStatsAggregatorDataset(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    handle = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/SleepDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/SleepDataset.java
new file mode 100644
index 00000000000..cfe2babce9e
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/SleepDataset.java
@@ -0,0 +1,88 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data.experimental;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
+
+/**
+ */
+public final class SleepDataset extends RawOp implements Operand<TType> {
+  
+  /**
+   * Factory method to create a class wrapping a new SleepDataset operation.
+   * 
+   * @param scope current scope
+   * @param inputDataset 
+   * @param sleepMicroseconds 
+   * @param outputTypes 
+   * @param outputShapes 
+   * @return a new instance of SleepDataset
+   */
+  @Endpoint(describeByClass = true)
+  public static SleepDataset create(Scope scope, Operand<?> inputDataset, Operand<TInt64> sleepMicroseconds, List<Class<? extends TType>> outputTypes, List<Shape> outputShapes) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ExperimentalSleepDataset", scope.makeOpName("SleepDataset"));
+    opBuilder.addInput(inputDataset.asOutput());
+    opBuilder.addInput(sleepMicroseconds.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    Class[] outputTypesArray = new Class[outputTypes.size()];
+    for (int i = 0; i < outputTypesArray.length; ++i) {
+      outputTypesArray[i] = outputTypes.get(i);
+    }
+    opBuilder.setAttr("output_types", outputTypesArray);
+    Shape[] outputShapesArray = new Shape[outputShapes.size()];
+    for (int i = 0; i < outputShapesArray.length; ++i) {
+      outputShapesArray[i] = outputShapes.get(i);
+    }
+    opBuilder.setAttr("output_shapes", outputShapesArray);
+    return new SleepDataset(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<?> handle() {
+    return handle;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) handle;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ExperimentalSleepDataset";
+  
+  private Output<?> handle;
+  
+  private SleepDataset(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    handle = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/SlidingWindowDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/SlidingWindowDataset.java
new file mode 100644
index 00000000000..a7183757b4f
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/SlidingWindowDataset.java
@@ -0,0 +1,96 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data.experimental;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Creates a dataset that passes a sliding window over `input_dataset`.
+ */
+public final class SlidingWindowDataset extends RawOp implements Operand<TType> {
+  
+  /**
+   * Factory method to create a class wrapping a new SlidingWindowDataset operation.
+   * 
+   * @param scope current scope
+   * @param inputDataset 
+   * @param windowSize A scalar representing the number of elements in the
+   * sliding window.
+   * @param windowShift A scalar representing the steps moving the sliding window
+   * forward in one iteration. It must be positive.
+   * @param windowStride A scalar representing the stride of the input elements of the sliding window.
+   * It must be positive.
+   * @param outputTypes 
+   * @param outputShapes 
+   * @return a new instance of SlidingWindowDataset
+   */
+  @Endpoint(describeByClass = true)
+  public static SlidingWindowDataset create(Scope scope, Operand<?> inputDataset, Operand<TInt64> windowSize, Operand<TInt64> windowShift, Operand<TInt64> windowStride, List<Class<? extends TType>> outputTypes, List<Shape> outputShapes) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ExperimentalSlidingWindowDataset", scope.makeOpName("SlidingWindowDataset"));
+    opBuilder.addInput(inputDataset.asOutput());
+    opBuilder.addInput(windowSize.asOutput());
+    opBuilder.addInput(windowShift.asOutput());
+    opBuilder.addInput(windowStride.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    Class[] outputTypesArray = new Class[outputTypes.size()];
+    for (int i = 0; i < outputTypesArray.length; ++i) {
+      outputTypesArray[i] = outputTypes.get(i);
+    }
+    opBuilder.setAttr("output_types", outputTypesArray);
+    Shape[] outputShapesArray = new Shape[outputShapes.size()];
+    for (int i = 0; i < outputShapesArray.length; ++i) {
+      outputShapesArray[i] = outputShapes.get(i);
+    }
+    opBuilder.setAttr("output_shapes", outputShapesArray);
+    return new SlidingWindowDataset(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<?> handle() {
+    return handle;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) handle;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ExperimentalSlidingWindowDataset";
+  
+  private Output<?> handle;
+  
+  private SlidingWindowDataset(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    handle = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/SqlDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/SqlDataset.java
new file mode 100644
index 00000000000..c6bd7b4a433
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/SqlDataset.java
@@ -0,0 +1,91 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data.experimental;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TString;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Creates a dataset that executes a SQL query and emits rows of the result set.
+ */
+public final class SqlDataset extends RawOp implements Operand<TType> {
+  
+  /**
+   * Factory method to create a class wrapping a new SqlDataset operation.
+   * 
+   * @param scope current scope
+   * @param driverName The database type. Currently, the only supported type is 'sqlite'.
+   * @param dataSourceName A connection string to connect to the database.
+   * @param query A SQL query to execute.
+   * @param outputTypes 
+   * @param outputShapes 
+   * @return a new instance of SqlDataset
+   */
+  @Endpoint(describeByClass = true)
+  public static SqlDataset create(Scope scope, Operand<TString> driverName, Operand<TString> dataSourceName, Operand<TString> query, List<Class<? extends TType>> outputTypes, List<Shape> outputShapes) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ExperimentalSqlDataset", scope.makeOpName("SqlDataset"));
+    opBuilder.addInput(driverName.asOutput());
+    opBuilder.addInput(dataSourceName.asOutput());
+    opBuilder.addInput(query.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    Class[] outputTypesArray = new Class[outputTypes.size()];
+    for (int i = 0; i < outputTypesArray.length; ++i) {
+      outputTypesArray[i] = outputTypes.get(i);
+    }
+    opBuilder.setAttr("output_types", outputTypesArray);
+    Shape[] outputShapesArray = new Shape[outputShapes.size()];
+    for (int i = 0; i < outputShapesArray.length; ++i) {
+      outputShapesArray[i] = outputShapes.get(i);
+    }
+    opBuilder.setAttr("output_shapes", outputShapesArray);
+    return new SqlDataset(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<?> handle() {
+    return handle;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) handle;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ExperimentalSqlDataset";
+  
+  private Output<?> handle;
+  
+  private SqlDataset(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    handle = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/StatsAggregatorHandle.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/StatsAggregatorHandle.java
new file mode 100644
index 00000000000..ec2e57ae4e7
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/StatsAggregatorHandle.java
@@ -0,0 +1,122 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data.experimental;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ */
+public final class StatsAggregatorHandle extends RawOp implements Operand<TType> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.data.experimental.StatsAggregatorHandle}
+   */
+  public static class Options {
+    
+    /**
+     * @param container 
+     */
+    public Options container(String container) {
+      this.container = container;
+      return this;
+    }
+    
+    /**
+     * @param sharedName 
+     */
+    public Options sharedName(String sharedName) {
+      this.sharedName = sharedName;
+      return this;
+    }
+    
+    private String container;
+    private String sharedName;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new StatsAggregatorHandle operation.
+   * 
+   * @param scope current scope
+   * @param options carries optional attributes values
+   * @return a new instance of StatsAggregatorHandle
+   */
+  @Endpoint(describeByClass = true)
+  public static StatsAggregatorHandle create(Scope scope, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("StatsAggregatorHandleV2", scope.makeOpName("StatsAggregatorHandle"));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.container != null) {
+          opBuilder.setAttr("container", opts.container);
+        }
+        if (opts.sharedName != null) {
+          opBuilder.setAttr("shared_name", opts.sharedName);
+        }
+      }
+    }
+    return new StatsAggregatorHandle(opBuilder.build());
+  }
+  
+  /**
+   * @param container 
+   */
+  public static Options container(String container) {
+    return new Options().container(container);
+  }
+  
+  /**
+   * @param sharedName 
+   */
+  public static Options sharedName(String sharedName) {
+    return new Options().sharedName(sharedName);
+  }
+  
+  /**
+   */
+  public Output<?> handle() {
+    return handle;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) handle;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "StatsAggregatorHandleV2";
+  
+  private Output<?> handle;
+  
+  private StatsAggregatorHandle(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    handle = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/StatsAggregatorSetSummaryWriter.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/StatsAggregatorSetSummaryWriter.java
new file mode 100644
index 00000000000..1af246d8313
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/StatsAggregatorSetSummaryWriter.java
@@ -0,0 +1,56 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data.experimental;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+
+/**
+ * Set a summary_writer_interface to record statistics using given stats_aggregator.
+ */
+public final class StatsAggregatorSetSummaryWriter extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new StatsAggregatorSetSummaryWriter operation.
+   * 
+   * @param scope current scope
+   * @param statsAggregator 
+   * @param summary 
+   * @return a new instance of StatsAggregatorSetSummaryWriter
+   */
+  @Endpoint(describeByClass = true)
+  public static StatsAggregatorSetSummaryWriter create(Scope scope, Operand<?> statsAggregator, Operand<?> summary) {
+    OperationBuilder opBuilder = scope.env().opBuilder("StatsAggregatorSetSummaryWriter", scope.makeOpName("StatsAggregatorSetSummaryWriter"));
+    opBuilder.addInput(statsAggregator.asOutput());
+    opBuilder.addInput(summary.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new StatsAggregatorSetSummaryWriter(opBuilder.build());
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "StatsAggregatorSetSummaryWriter";
+  
+  private StatsAggregatorSetSummaryWriter(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/StatsAggregatorSummary.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/StatsAggregatorSummary.java
new file mode 100644
index 00000000000..a24bfdbaada
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/StatsAggregatorSummary.java
@@ -0,0 +1,71 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data.experimental;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TString;
+
+/**
+ * Produces a summary of any statistics recorded by the given statistics manager.
+ */
+public final class StatsAggregatorSummary extends RawOp implements Operand<TString> {
+  
+  /**
+   * Factory method to create a class wrapping a new StatsAggregatorSummary operation.
+   * 
+   * @param scope current scope
+   * @param iterator 
+   * @return a new instance of StatsAggregatorSummary
+   */
+  @Endpoint(describeByClass = true)
+  public static StatsAggregatorSummary create(Scope scope, Operand<?> iterator) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ExperimentalStatsAggregatorSummary", scope.makeOpName("StatsAggregatorSummary"));
+    opBuilder.addInput(iterator.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new StatsAggregatorSummary(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<TString> summary() {
+    return summary;
+  }
+  
+  @Override
+  public Output<TString> asOutput() {
+    return summary;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ExperimentalStatsAggregatorSummary";
+  
+  private Output<TString> summary;
+  
+  private StatsAggregatorSummary(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    summary = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/ThreadPoolDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/ThreadPoolDataset.java
new file mode 100644
index 00000000000..2d9a37dba6e
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/ThreadPoolDataset.java
@@ -0,0 +1,88 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data.experimental;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Creates a dataset that uses a custom thread pool to compute `input_dataset`.
+ */
+public final class ThreadPoolDataset extends RawOp implements Operand<TType> {
+  
+  /**
+   * Factory method to create a class wrapping a new ThreadPoolDataset operation.
+   * 
+   * @param scope current scope
+   * @param inputDataset 
+   * @param threadPool A resource produced by the ThreadPoolHandle op.
+   * @param outputTypes 
+   * @param outputShapes 
+   * @return a new instance of ThreadPoolDataset
+   */
+  @Endpoint(describeByClass = true)
+  public static ThreadPoolDataset create(Scope scope, Operand<?> inputDataset, Operand<?> threadPool, List<Class<? extends TType>> outputTypes, List<Shape> outputShapes) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ExperimentalThreadPoolDataset", scope.makeOpName("ThreadPoolDataset"));
+    opBuilder.addInput(inputDataset.asOutput());
+    opBuilder.addInput(threadPool.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    Class[] outputTypesArray = new Class[outputTypes.size()];
+    for (int i = 0; i < outputTypesArray.length; ++i) {
+      outputTypesArray[i] = outputTypes.get(i);
+    }
+    opBuilder.setAttr("output_types", outputTypesArray);
+    Shape[] outputShapesArray = new Shape[outputShapes.size()];
+    for (int i = 0; i < outputShapesArray.length; ++i) {
+      outputShapesArray[i] = outputShapes.get(i);
+    }
+    opBuilder.setAttr("output_shapes", outputShapesArray);
+    return new ThreadPoolDataset(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<?> handle() {
+    return handle;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) handle;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ExperimentalThreadPoolDataset";
+  
+  private Output<?> handle;
+  
+  private ThreadPoolDataset(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    handle = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/ThreadPoolHandle.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/ThreadPoolHandle.java
new file mode 100644
index 00000000000..15adc678211
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/ThreadPoolHandle.java
@@ -0,0 +1,152 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data.experimental;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Creates a dataset that uses a custom thread pool to compute `input_dataset`.
+ */
+public final class ThreadPoolHandle extends RawOp implements Operand<TType> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.data.experimental.ThreadPoolHandle}
+   */
+  public static class Options {
+    
+    /**
+     * @param maxIntraOpParallelism The maximum degree of parallelism to use within operations that execute on this
+     * threadpool.
+     */
+    public Options maxIntraOpParallelism(Long maxIntraOpParallelism) {
+      this.maxIntraOpParallelism = maxIntraOpParallelism;
+      return this;
+    }
+    
+    /**
+     * @param container 
+     */
+    public Options container(String container) {
+      this.container = container;
+      return this;
+    }
+    
+    /**
+     * @param sharedName 
+     */
+    public Options sharedName(String sharedName) {
+      this.sharedName = sharedName;
+      return this;
+    }
+    
+    private Long maxIntraOpParallelism;
+    private String container;
+    private String sharedName;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new ThreadPoolHandle operation.
+   * 
+   * @param scope current scope
+   * @param numThreads The number of threads in the thread pool.
+   * @param displayName A human-readable name for the threads that may be visible in some
+   * visualizations.
+   * threadpool.
+   * @param options carries optional attributes values
+   * @return a new instance of ThreadPoolHandle
+   */
+  @Endpoint(describeByClass = true)
+  public static ThreadPoolHandle create(Scope scope, Long numThreads, String displayName, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ExperimentalThreadPoolHandle", scope.makeOpName("ThreadPoolHandle"));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("num_threads", numThreads);
+    opBuilder.setAttr("display_name", displayName);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.maxIntraOpParallelism != null) {
+          opBuilder.setAttr("max_intra_op_parallelism", opts.maxIntraOpParallelism);
+        }
+        if (opts.container != null) {
+          opBuilder.setAttr("container", opts.container);
+        }
+        if (opts.sharedName != null) {
+          opBuilder.setAttr("shared_name", opts.sharedName);
+        }
+      }
+    }
+    return new ThreadPoolHandle(opBuilder.build());
+  }
+  
+  /**
+   * @param maxIntraOpParallelism The maximum degree of parallelism to use within operations that execute on this
+   * threadpool.
+   */
+  public static Options maxIntraOpParallelism(Long maxIntraOpParallelism) {
+    return new Options().maxIntraOpParallelism(maxIntraOpParallelism);
+  }
+  
+  /**
+   * @param container 
+   */
+  public static Options container(String container) {
+    return new Options().container(container);
+  }
+  
+  /**
+   * @param sharedName 
+   */
+  public static Options sharedName(String sharedName) {
+    return new Options().sharedName(sharedName);
+  }
+  
+  /**
+   * A resource that can be consumed by one or more ExperimentalThreadPoolDataset
+   * ops.
+   */
+  public Output<?> handle() {
+    return handle;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) handle;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ExperimentalThreadPoolHandle";
+  
+  private Output<?> handle;
+  
+  private ThreadPoolHandle(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    handle = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/UnbatchDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/UnbatchDataset.java
new file mode 100644
index 00000000000..f44f8f09fc2
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/UnbatchDataset.java
@@ -0,0 +1,86 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data.experimental;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * A dataset that splits the elements of its input into multiple elements.
+ */
+public final class UnbatchDataset extends RawOp implements Operand<TType> {
+  
+  /**
+   * Factory method to create a class wrapping a new UnbatchDataset operation.
+   * 
+   * @param scope current scope
+   * @param inputDataset 
+   * @param outputTypes 
+   * @param outputShapes 
+   * @return a new instance of UnbatchDataset
+   */
+  @Endpoint(describeByClass = true)
+  public static UnbatchDataset create(Scope scope, Operand<?> inputDataset, List<Class<? extends TType>> outputTypes, List<Shape> outputShapes) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ExperimentalUnbatchDataset", scope.makeOpName("UnbatchDataset"));
+    opBuilder.addInput(inputDataset.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    Class[] outputTypesArray = new Class[outputTypes.size()];
+    for (int i = 0; i < outputTypesArray.length; ++i) {
+      outputTypesArray[i] = outputTypes.get(i);
+    }
+    opBuilder.setAttr("output_types", outputTypesArray);
+    Shape[] outputShapesArray = new Shape[outputShapes.size()];
+    for (int i = 0; i < outputShapesArray.length; ++i) {
+      outputShapesArray[i] = outputShapes.get(i);
+    }
+    opBuilder.setAttr("output_shapes", outputShapesArray);
+    return new UnbatchDataset(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<?> handle() {
+    return handle;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) handle;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ExperimentalUnbatchDataset";
+  
+  private Output<?> handle;
+  
+  private UnbatchDataset(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    handle = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/UncompressElement.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/UncompressElement.java
new file mode 100644
index 00000000000..79d4624779d
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/UncompressElement.java
@@ -0,0 +1,90 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data.experimental;
+
+import java.util.Arrays;
+import java.util.Iterator;
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Uncompresses a compressed dataset element.
+ */
+public final class UncompressElement extends RawOp implements Iterable<Operand<TType>> {
+  
+  /**
+   * Factory method to create a class wrapping a new UncompressElement operation.
+   * 
+   * @param scope current scope
+   * @param compressed 
+   * @param outputTypes 
+   * @param outputShapes 
+   * @return a new instance of UncompressElement
+   */
+  @Endpoint(describeByClass = true)
+  public static UncompressElement create(Scope scope, Operand<?> compressed, List<Class<? extends TType>> outputTypes, List<Shape> outputShapes) {
+    OperationBuilder opBuilder = scope.env().opBuilder("UncompressElement", scope.makeOpName("UncompressElement"));
+    opBuilder.addInput(compressed.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    Class[] outputTypesArray = new Class[outputTypes.size()];
+    for (int i = 0; i < outputTypesArray.length; ++i) {
+      outputTypesArray[i] = outputTypes.get(i);
+    }
+    opBuilder.setAttr("output_types", outputTypesArray);
+    Shape[] outputShapesArray = new Shape[outputShapes.size()];
+    for (int i = 0; i < outputShapesArray.length; ++i) {
+      outputShapesArray[i] = outputShapes.get(i);
+    }
+    opBuilder.setAttr("output_shapes", outputShapesArray);
+    return new UncompressElement(opBuilder.build());
+  }
+  
+  /**
+   */
+  public List<Output<?>> components() {
+    return components;
+  }
+  
+  @Override
+  @SuppressWarnings({"rawtypes", "unchecked"})
+  public Iterator<Operand<TType>> iterator() {
+    return (Iterator) components.iterator();
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "UncompressElement";
+  
+  private List<Output<?>> components;
+  
+  private UncompressElement(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    int componentsLength = operation.outputListLength("components");
+    components = Arrays.asList(operation.outputList(outputIdx, componentsLength));
+    outputIdx += componentsLength;
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/UniqueDataset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/UniqueDataset.java
new file mode 100644
index 00000000000..b05203c4542
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/data/experimental/UniqueDataset.java
@@ -0,0 +1,86 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.data.experimental;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Creates a dataset that contains the unique elements of `input_dataset`.
+ */
+public final class UniqueDataset extends RawOp implements Operand<TType> {
+  
+  /**
+   * Factory method to create a class wrapping a new UniqueDataset operation.
+   * 
+   * @param scope current scope
+   * @param inputDataset 
+   * @param outputTypes 
+   * @param outputShapes 
+   * @return a new instance of UniqueDataset
+   */
+  @Endpoint(describeByClass = true)
+  public static UniqueDataset create(Scope scope, Operand<?> inputDataset, List<Class<? extends TType>> outputTypes, List<Shape> outputShapes) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ExperimentalUniqueDataset", scope.makeOpName("UniqueDataset"));
+    opBuilder.addInput(inputDataset.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    Class[] outputTypesArray = new Class[outputTypes.size()];
+    for (int i = 0; i < outputTypesArray.length; ++i) {
+      outputTypesArray[i] = outputTypes.get(i);
+    }
+    opBuilder.setAttr("output_types", outputTypesArray);
+    Shape[] outputShapesArray = new Shape[outputShapes.size()];
+    for (int i = 0; i < outputShapesArray.length; ++i) {
+      outputShapesArray[i] = outputShapes.get(i);
+    }
+    opBuilder.setAttr("output_shapes", outputShapesArray);
+    return new UniqueDataset(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<?> handle() {
+    return handle;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) handle;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ExperimentalUniqueDataset";
+  
+  private Output<?> handle;
+  
+  private UniqueDataset(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    handle = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/debugging/CheckNumerics.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/debugging/CheckNumerics.java
new file mode 100644
index 00000000000..44ec8ca8aaf
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/debugging/CheckNumerics.java
@@ -0,0 +1,80 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.debugging;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Checks a tensor for NaN, -Inf and +Inf values.
+ * <p>
+ * When run, reports an `InvalidArgument` error if `tensor` has any values
+ * that are not a number (NaN) or infinity (Inf). Otherwise, passes `tensor` as-is.
+ * Unlike CheckNumerics (V1), CheckNumericsV2 distinguishes -Inf and +Inf in the
+ * errors it throws.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+public final class CheckNumerics<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new CheckNumerics operation.
+   * 
+   * @param scope current scope
+   * @param tensor 
+   * @param message Prefix of the error message.
+   * @return a new instance of CheckNumerics
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> CheckNumerics<T> create(Scope scope, Operand<T> tensor, String message) {
+    OperationBuilder opBuilder = scope.env().opBuilder("CheckNumericsV2", scope.makeOpName("CheckNumerics"));
+    opBuilder.addInput(tensor.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("message", message);
+    return new CheckNumerics<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "CheckNumericsV2";
+  
+  private Output<T> output;
+  
+  private CheckNumerics(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/debugging/DebugGradientIdentity.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/debugging/DebugGradientIdentity.java
new file mode 100644
index 00000000000..4214a50ab41
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/debugging/DebugGradientIdentity.java
@@ -0,0 +1,77 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.debugging;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Identity op for gradient debugging.
+ * <p>
+ * This op is hidden from public in Python. It is used by TensorFlow Debugger to
+ * register gradient tensors for gradient debugging.
+ * This op operates on non-reference-type tensors.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+public final class DebugGradientIdentity<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new DebugGradientIdentity operation.
+   * 
+   * @param scope current scope
+   * @param input 
+   * @return a new instance of DebugGradientIdentity
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> DebugGradientIdentity<T> create(Scope scope, Operand<T> input) {
+    OperationBuilder opBuilder = scope.env().opBuilder("DebugGradientIdentity", scope.makeOpName("DebugGradientIdentity"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new DebugGradientIdentity<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "DebugGradientIdentity";
+  
+  private Output<T> output;
+  
+  private DebugGradientIdentity(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/debugging/DebugGradientRefIdentity.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/debugging/DebugGradientRefIdentity.java
new file mode 100644
index 00000000000..fb4a1d4cfef
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/debugging/DebugGradientRefIdentity.java
@@ -0,0 +1,77 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.debugging;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Identity op for gradient debugging.
+ * <p>
+ * This op is hidden from public in Python. It is used by TensorFlow Debugger to
+ * register gradient tensors for gradient debugging.
+ * This op operates on reference-type tensors.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+public final class DebugGradientRefIdentity<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new DebugGradientRefIdentity operation.
+   * 
+   * @param scope current scope
+   * @param input 
+   * @return a new instance of DebugGradientRefIdentity
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> DebugGradientRefIdentity<T> create(Scope scope, Operand<T> input) {
+    OperationBuilder opBuilder = scope.env().opBuilder("DebugGradientRefIdentity", scope.makeOpName("DebugGradientRefIdentity"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new DebugGradientRefIdentity<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "DebugGradientRefIdentity";
+  
+  private Output<T> output;
+  
+  private DebugGradientRefIdentity(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/debugging/DebugIdentity.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/debugging/DebugIdentity.java
new file mode 100644
index 00000000000..14f719c90f8
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/debugging/DebugIdentity.java
@@ -0,0 +1,242 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.debugging;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Debug Identity V2 Op.
+ * <p>
+ * Provides an identity mapping from input to output, while writing the content of
+ * the input tensor by calling DebugEventsWriter.
+ * <p>
+ * The semantics of the input tensor depends on tensor_debug_mode. In typical
+ * usage, the input tensor comes directly from the user computation only when
+ * graph_debug_mode is FULL_TENSOR (see protobuf/debug_event.proto for a
+ * list of all the possible values of graph_debug_mode). For the other debug modes,
+ * the input tensor should be produced by an additional op or subgraph that
+ * computes summary information about one or more tensors.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+public final class DebugIdentity<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.debugging.DebugIdentity}
+   */
+  public static class Options {
+    
+    /**
+     * @param tfdbgContextId A tfdbg-generated ID for the context that the op belongs to,
+     *   e.g., a concrete compiled tf.function.
+     */
+    public Options tfdbgContextId(String tfdbgContextId) {
+      this.tfdbgContextId = tfdbgContextId;
+      return this;
+    }
+    
+    /**
+     * @param opName Optional. Name of the op that the debug op is concerned with.
+     *   Used only for single-tensor trace.
+     */
+    public Options opName(String opName) {
+      this.opName = opName;
+      return this;
+    }
+    
+    /**
+     * @param outputSlot Optional. Output slot index of the tensor that the debug op
+     *   is concerned with. Used only for single-tensor trace.
+     */
+    public Options outputSlot(Long outputSlot) {
+      this.outputSlot = outputSlot;
+      return this;
+    }
+    
+    /**
+     * @param tensorDebugMode TensorDebugMode enum value. See debug_event.proto for details.
+     */
+    public Options tensorDebugMode(Long tensorDebugMode) {
+      this.tensorDebugMode = tensorDebugMode;
+      return this;
+    }
+    
+    /**
+     * @param debugUrls List of URLs to debug targets, e.g., file:///foo/tfdbg_dump.
+     */
+    public Options debugUrls(List<String> debugUrls) {
+      this.debugUrls = debugUrls;
+      return this;
+    }
+    
+    /**
+     * @param circularBufferSize 
+     */
+    public Options circularBufferSize(Long circularBufferSize) {
+      this.circularBufferSize = circularBufferSize;
+      return this;
+    }
+    
+    /**
+     * @param tfdbgRunId 
+     */
+    public Options tfdbgRunId(String tfdbgRunId) {
+      this.tfdbgRunId = tfdbgRunId;
+      return this;
+    }
+    
+    private String tfdbgContextId;
+    private String opName;
+    private Long outputSlot;
+    private Long tensorDebugMode;
+    private List<String> debugUrls;
+    private Long circularBufferSize;
+    private String tfdbgRunId;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new DebugIdentity operation.
+   * 
+   * @param scope current scope
+   * @param input Input tensor, non-Reference type
+   * @param options carries optional attributes values
+   * @return a new instance of DebugIdentity
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> DebugIdentity<T> create(Scope scope, Operand<T> input, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("DebugIdentityV2", scope.makeOpName("DebugIdentity"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.tfdbgContextId != null) {
+          opBuilder.setAttr("tfdbg_context_id", opts.tfdbgContextId);
+        }
+        if (opts.opName != null) {
+          opBuilder.setAttr("op_name", opts.opName);
+        }
+        if (opts.outputSlot != null) {
+          opBuilder.setAttr("output_slot", opts.outputSlot);
+        }
+        if (opts.tensorDebugMode != null) {
+          opBuilder.setAttr("tensor_debug_mode", opts.tensorDebugMode);
+        }
+        if (opts.debugUrls != null) {
+          String[] debugUrlsArray = new String[opts.debugUrls.size()];
+          for (int i = 0; i < debugUrlsArray.length; ++i) {
+            debugUrlsArray[i] = opts.debugUrls.get(i);
+          }
+          opBuilder.setAttr("debug_urls", debugUrlsArray);
+        }
+        if (opts.circularBufferSize != null) {
+          opBuilder.setAttr("circular_buffer_size", opts.circularBufferSize);
+        }
+        if (opts.tfdbgRunId != null) {
+          opBuilder.setAttr("tfdbg_run_id", opts.tfdbgRunId);
+        }
+      }
+    }
+    return new DebugIdentity<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param tfdbgContextId A tfdbg-generated ID for the context that the op belongs to,
+   *   e.g., a concrete compiled tf.function.
+   */
+  public static Options tfdbgContextId(String tfdbgContextId) {
+    return new Options().tfdbgContextId(tfdbgContextId);
+  }
+  
+  /**
+   * @param opName Optional. Name of the op that the debug op is concerned with.
+   *   Used only for single-tensor trace.
+   */
+  public static Options opName(String opName) {
+    return new Options().opName(opName);
+  }
+  
+  /**
+   * @param outputSlot Optional. Output slot index of the tensor that the debug op
+   *   is concerned with. Used only for single-tensor trace.
+   */
+  public static Options outputSlot(Long outputSlot) {
+    return new Options().outputSlot(outputSlot);
+  }
+  
+  /**
+   * @param tensorDebugMode TensorDebugMode enum value. See debug_event.proto for details.
+   */
+  public static Options tensorDebugMode(Long tensorDebugMode) {
+    return new Options().tensorDebugMode(tensorDebugMode);
+  }
+  
+  /**
+   * @param debugUrls List of URLs to debug targets, e.g., file:///foo/tfdbg_dump.
+   */
+  public static Options debugUrls(List<String> debugUrls) {
+    return new Options().debugUrls(debugUrls);
+  }
+  
+  /**
+   * @param circularBufferSize 
+   */
+  public static Options circularBufferSize(Long circularBufferSize) {
+    return new Options().circularBufferSize(circularBufferSize);
+  }
+  
+  /**
+   * @param tfdbgRunId 
+   */
+  public static Options tfdbgRunId(String tfdbgRunId) {
+    return new Options().tfdbgRunId(tfdbgRunId);
+  }
+  
+  /**
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "DebugIdentityV2";
+  
+  private Output<T> output;
+  
+  private DebugIdentity(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/debugging/DebugNanCount.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/debugging/DebugNanCount.java
new file mode 100644
index 00000000000..30bc6931743
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/debugging/DebugNanCount.java
@@ -0,0 +1,182 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.debugging;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Debug NaN Value Counter Op.
+ * <p>
+ * Counts number of NaNs in the input tensor, for debugging.
+ */
+public final class DebugNanCount extends RawOp implements Operand<TInt64> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.debugging.DebugNanCount}
+   */
+  public static class Options {
+    
+    /**
+     * @param deviceName 
+     */
+    public Options deviceName(String deviceName) {
+      this.deviceName = deviceName;
+      return this;
+    }
+    
+    /**
+     * @param tensorName Name of the input tensor.
+     */
+    public Options tensorName(String tensorName) {
+      this.tensorName = tensorName;
+      return this;
+    }
+    
+    /**
+     * @param debugUrls List of URLs to debug targets, e.g.,
+     *   file:///foo/tfdbg_dump, grpc:://localhost:11011.
+     */
+    public Options debugUrls(List<String> debugUrls) {
+      this.debugUrls = debugUrls;
+      return this;
+    }
+    
+    /**
+     * @param gatedGrpc  Whether this op will be gated. If any of the debug_urls of this
+     *   debug node is of the grpc:// scheme, when the value of this attribute is set
+     *   to True, the data will not actually be sent via the grpc stream unless this
+     *   debug op has been enabled at the debug_url. If all of the debug_urls of this
+     *   debug node are of the grpc:// scheme and the debug op is enabled at none of
+     *   them, the output will be an empty Tensor.
+     */
+    public Options gatedGrpc(Boolean gatedGrpc) {
+      this.gatedGrpc = gatedGrpc;
+      return this;
+    }
+    
+    private String deviceName;
+    private String tensorName;
+    private List<String> debugUrls;
+    private Boolean gatedGrpc;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new DebugNanCount operation.
+   * 
+   * @param scope current scope
+   * @param input Input tensor, non-Reference type.
+   * @param options carries optional attributes values
+   * @return a new instance of DebugNanCount
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> DebugNanCount create(Scope scope, Operand<T> input, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("DebugNanCount", scope.makeOpName("DebugNanCount"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.deviceName != null) {
+          opBuilder.setAttr("device_name", opts.deviceName);
+        }
+        if (opts.tensorName != null) {
+          opBuilder.setAttr("tensor_name", opts.tensorName);
+        }
+        if (opts.debugUrls != null) {
+          String[] debugUrlsArray = new String[opts.debugUrls.size()];
+          for (int i = 0; i < debugUrlsArray.length; ++i) {
+            debugUrlsArray[i] = opts.debugUrls.get(i);
+          }
+          opBuilder.setAttr("debug_urls", debugUrlsArray);
+        }
+        if (opts.gatedGrpc != null) {
+          opBuilder.setAttr("gated_grpc", opts.gatedGrpc);
+        }
+      }
+    }
+    return new DebugNanCount(opBuilder.build());
+  }
+  
+  /**
+   * @param deviceName 
+   */
+  public static Options deviceName(String deviceName) {
+    return new Options().deviceName(deviceName);
+  }
+  
+  /**
+   * @param tensorName Name of the input tensor.
+   */
+  public static Options tensorName(String tensorName) {
+    return new Options().tensorName(tensorName);
+  }
+  
+  /**
+   * @param debugUrls List of URLs to debug targets, e.g.,
+   *   file:///foo/tfdbg_dump, grpc:://localhost:11011.
+   */
+  public static Options debugUrls(List<String> debugUrls) {
+    return new Options().debugUrls(debugUrls);
+  }
+  
+  /**
+   * @param gatedGrpc  Whether this op will be gated. If any of the debug_urls of this
+   *   debug node is of the grpc:// scheme, when the value of this attribute is set
+   *   to True, the data will not actually be sent via the grpc stream unless this
+   *   debug op has been enabled at the debug_url. If all of the debug_urls of this
+   *   debug node are of the grpc:// scheme and the debug op is enabled at none of
+   *   them, the output will be an empty Tensor.
+   */
+  public static Options gatedGrpc(Boolean gatedGrpc) {
+    return new Options().gatedGrpc(gatedGrpc);
+  }
+  
+  /**
+   */
+  public Output<TInt64> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<TInt64> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "DebugNanCount";
+  
+  private Output<TInt64> output;
+  
+  private DebugNanCount(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/debugging/DebugNumericsSummary.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/debugging/DebugNumericsSummary.java
new file mode 100644
index 00000000000..3b927a32cf6
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/debugging/DebugNumericsSummary.java
@@ -0,0 +1,253 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.debugging;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Debug Numeric Summary V2 Op.
+ * <p>
+ * Computes a numeric summary of the input tensor. The shape of the output
+ * depends on the tensor_debug_mode attribute.
+ * This op is used internally by TensorFlow Debugger (tfdbg) v2.
+ * 
+ * @param <U> data type for {@code output()} output
+ */
+public final class DebugNumericsSummary<U extends TNumber> extends RawOp implements Operand<U> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.debugging.DebugNumericsSummary}
+   */
+  public static class Options {
+    
+    /**
+     * @param tensorDebugMode Tensor debug mode: the mode in which the input tensor is summarized
+     *   by the op. See the TensorDebugMode enum in
+     *   tensorflow/core/protobuf/debug_event.proto for details.
+     * <p>
+     * Supported values:
+     *   2 (CURT_HEALTH): Output a float32/64 tensor of shape [2]. The 1st
+     *   element is the tensor_id, if provided, and -1 otherwise. The 2nd
+     *   element is a bit which is set to 1 if the input tensor has an
+     *   infinity or nan value, or zero otherwise.
+     * <p>
+     *   3 (CONCISE_HEALTH): Output a float32/64 tensor of shape [5]. The 1st
+     *   element is the tensor_id, if provided, and -1 otherwise. The
+     *   remaining four slots are the total number of elements, -infs,
+     *   +infs, and nans in the input tensor respectively.
+     * <p>
+     *   4 (FULL_HEALTH): Output a float32/64 tensor of shape [11]. The 1st
+     *   element is the tensor_id, if provided, and -1 otherwise. The 2nd
+     *   element is the device_id, if provided, and -1 otherwise. The 3rd
+     *   element holds the datatype value of the input tensor as according
+     *   to the enumerated type in tensorflow/core/framework/types.proto.
+     *   The remaining elements hold the total number of elements, -infs,
+     *   +infs, nans, negative finite numbers, zeros, and positive finite
+     *   numbers in the input tensor respectively.
+     * <p>
+     *   5 (SHAPE): Output a float32/64 tensor of shape [10]. The 1st
+     *   element is the tensor_id, if provided, and -1 otherwise. The 2nd
+     *   element holds the datatype value of the input tensor as according
+     *   to the enumerated type in tensorflow/core/framework/types.proto.
+     *   The 3rd element holds the rank of the tensor. The 4th element holds
+     *   the number of elements within the tensor. Finally the remaining 6
+     *   elements hold the shape of the tensor. If the rank of the tensor
+     *   is lower than 6, the shape is right padded with zeros. If the rank
+     *   is greater than 6, the head of the shape is truncated.
+     * <p>
+     *   6 (FULL_NUMERICS): Output a float32/64 tensor of shape [22]. The 1st
+     *   element is the tensor_id, if provided, and -1 otherwise. The 2nd
+     *   element is the device_id, if provided, and -1 otherwise. The 3rd
+     *   element holds the datatype value of the input tensor as according
+     *   to the enumerated type in tensorflow/core/framework/types.proto.
+     *   The 4th element holds the rank of the tensor. The 5th to 11th
+     *   elements hold the shape of the tensor. If the rank of the tensor
+     *   is lower than 6, the shape is right padded with zeros. If the rank
+     *   is greater than 6, the head of the shape is truncated. The 12th to
+     *   18th elements hold the number of elements, -infs, +infs, nans,
+     *   denormal floats, negative finite numbers, zeros, and positive
+     *   finite numbers in the input tensor respectively. The final four
+     *   elements hold the min value, max value, mean, and variance of the
+     *   input tensor.
+     * <p>
+     *   8 (REDUCE_INF_NAN_THREE_SLOTS): Output a float32/64 tensor of shape
+     *   [3]. The 1st element is -inf if any elements of the input tensor
+     *   is -inf, or zero otherwise. The 2nd element is +inf if any elements
+     *   of the input tensor is +inf, or zero otherwise.  The 3rd element is
+     *   nan if any element of the input tensor is nan, or zero otherwise.
+     */
+    public Options tensorDebugMode(Long tensorDebugMode) {
+      this.tensorDebugMode = tensorDebugMode;
+      return this;
+    }
+    
+    /**
+     * @param tensorId Optional. An integer identifier for the tensor being summarized by this op.
+     */
+    public Options tensorId(Long tensorId) {
+      this.tensorId = tensorId;
+      return this;
+    }
+    
+    private Long tensorDebugMode;
+    private Long tensorId;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new DebugNumericsSummary operation.
+   * 
+   * @param scope current scope
+   * @param input Input tensor, to be summarized by the op.
+   * @param outputDtype Optional. The type of the output. Can be float32 or float64 (default: float32).
+   * @param options carries optional attributes values
+   * @return a new instance of DebugNumericsSummary
+   */
+  @Endpoint(describeByClass = true)
+  public static <U extends TNumber, T extends TType> DebugNumericsSummary<U> create(Scope scope, Operand<T> input, Class<U> outputDtype, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("DebugNumericSummaryV2", scope.makeOpName("DebugNumericsSummary"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("output_dtype", outputDtype);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.tensorDebugMode != null) {
+          opBuilder.setAttr("tensor_debug_mode", opts.tensorDebugMode);
+        }
+        if (opts.tensorId != null) {
+          opBuilder.setAttr("tensor_id", opts.tensorId);
+        }
+      }
+    }
+    return new DebugNumericsSummary<U>(opBuilder.build());
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new DebugNumericsSummary operation using default output types.
+   * 
+   * @param scope current scope
+   * @param input Input tensor, to be summarized by the op.
+   * @param options carries optional attributes values
+   * @return a new instance of DebugNumericsSummary
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> DebugNumericsSummary<TFloat32> create(Scope scope, Operand<T> input, Options... options) {
+    return create(scope, input, TFloat32.class, options);
+  }
+  
+  /**
+   * @param tensorDebugMode Tensor debug mode: the mode in which the input tensor is summarized
+   *   by the op. See the TensorDebugMode enum in
+   *   tensorflow/core/protobuf/debug_event.proto for details.
+   * <p>
+   * Supported values:
+   *   2 (CURT_HEALTH): Output a float32/64 tensor of shape [2]. The 1st
+   *   element is the tensor_id, if provided, and -1 otherwise. The 2nd
+   *   element is a bit which is set to 1 if the input tensor has an
+   *   infinity or nan value, or zero otherwise.
+   * <p>
+   *   3 (CONCISE_HEALTH): Output a float32/64 tensor of shape [5]. The 1st
+   *   element is the tensor_id, if provided, and -1 otherwise. The
+   *   remaining four slots are the total number of elements, -infs,
+   *   +infs, and nans in the input tensor respectively.
+   * <p>
+   *   4 (FULL_HEALTH): Output a float32/64 tensor of shape [11]. The 1st
+   *   element is the tensor_id, if provided, and -1 otherwise. The 2nd
+   *   element is the device_id, if provided, and -1 otherwise. The 3rd
+   *   element holds the datatype value of the input tensor as according
+   *   to the enumerated type in tensorflow/core/framework/types.proto.
+   *   The remaining elements hold the total number of elements, -infs,
+   *   +infs, nans, negative finite numbers, zeros, and positive finite
+   *   numbers in the input tensor respectively.
+   * <p>
+   *   5 (SHAPE): Output a float32/64 tensor of shape [10]. The 1st
+   *   element is the tensor_id, if provided, and -1 otherwise. The 2nd
+   *   element holds the datatype value of the input tensor as according
+   *   to the enumerated type in tensorflow/core/framework/types.proto.
+   *   The 3rd element holds the rank of the tensor. The 4th element holds
+   *   the number of elements within the tensor. Finally the remaining 6
+   *   elements hold the shape of the tensor. If the rank of the tensor
+   *   is lower than 6, the shape is right padded with zeros. If the rank
+   *   is greater than 6, the head of the shape is truncated.
+   * <p>
+   *   6 (FULL_NUMERICS): Output a float32/64 tensor of shape [22]. The 1st
+   *   element is the tensor_id, if provided, and -1 otherwise. The 2nd
+   *   element is the device_id, if provided, and -1 otherwise. The 3rd
+   *   element holds the datatype value of the input tensor as according
+   *   to the enumerated type in tensorflow/core/framework/types.proto.
+   *   The 4th element holds the rank of the tensor. The 5th to 11th
+   *   elements hold the shape of the tensor. If the rank of the tensor
+   *   is lower than 6, the shape is right padded with zeros. If the rank
+   *   is greater than 6, the head of the shape is truncated. The 12th to
+   *   18th elements hold the number of elements, -infs, +infs, nans,
+   *   denormal floats, negative finite numbers, zeros, and positive
+   *   finite numbers in the input tensor respectively. The final four
+   *   elements hold the min value, max value, mean, and variance of the
+   *   input tensor.
+   * <p>
+   *   8 (REDUCE_INF_NAN_THREE_SLOTS): Output a float32/64 tensor of shape
+   *   [3]. The 1st element is -inf if any elements of the input tensor
+   *   is -inf, or zero otherwise. The 2nd element is +inf if any elements
+   *   of the input tensor is +inf, or zero otherwise.  The 3rd element is
+   *   nan if any element of the input tensor is nan, or zero otherwise.
+   */
+  public static Options tensorDebugMode(Long tensorDebugMode) {
+    return new Options().tensorDebugMode(tensorDebugMode);
+  }
+  
+  /**
+   * @param tensorId Optional. An integer identifier for the tensor being summarized by this op.
+   */
+  public static Options tensorId(Long tensorId) {
+    return new Options().tensorId(tensorId);
+  }
+  
+  /**
+   */
+  public Output<U> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<U> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "DebugNumericSummaryV2";
+  
+  private Output<U> output;
+  
+  private DebugNumericsSummary(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/dtypes/AsString.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/dtypes/AsString.java
new file mode 100644
index 00000000000..9d95a0ee6a0
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/dtypes/AsString.java
@@ -0,0 +1,206 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.dtypes;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TString;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Converts each entry in the given tensor to strings.
+ * <p>
+ * Supports many numeric types and boolean.
+ * <p>
+ * For Unicode, see the
+ * [https://www.tensorflow.org/tutorials/representation/unicode](Working with Unicode text)
+ * tutorial.
+ * <p>
+ * Examples:
+ * <p>
+ * >>> tf.strings.as_string([3, 2])
+ * <tf.Tensor: shape=(2,), dtype=string, numpy=array([b'3', b'2'], dtype=object)>
+ * >>> tf.strings.as_string([3.1415926, 2.71828], precision=2).numpy()
+ * array([b'3.14', b'2.72'], dtype=object)
+ */
+@Operator(group = "dtypes")
+public final class AsString extends RawOp implements Operand<TString> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.dtypes.AsString}
+   */
+  public static class Options {
+    
+    /**
+     * @param precision The post-decimal precision to use for floating point numbers.
+     * Only used if precision > -1.
+     */
+    public Options precision(Long precision) {
+      this.precision = precision;
+      return this;
+    }
+    
+    /**
+     * @param scientific Use scientific notation for floating point numbers.
+     */
+    public Options scientific(Boolean scientific) {
+      this.scientific = scientific;
+      return this;
+    }
+    
+    /**
+     * @param shortest Use shortest representation (either scientific or standard) for
+     * floating point numbers.
+     */
+    public Options shortest(Boolean shortest) {
+      this.shortest = shortest;
+      return this;
+    }
+    
+    /**
+     * @param width Pad pre-decimal numbers to this width.
+     * Applies to both floating point and integer numbers.
+     * Only used if width > -1.
+     */
+    public Options width(Long width) {
+      this.width = width;
+      return this;
+    }
+    
+    /**
+     * @param fill The value to pad if width > -1.  If empty, pads with spaces.
+     * Another typical value is '0'.  String cannot be longer than 1 character.
+     */
+    public Options fill(String fill) {
+      this.fill = fill;
+      return this;
+    }
+    
+    private Long precision;
+    private Boolean scientific;
+    private Boolean shortest;
+    private Long width;
+    private String fill;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new AsString operation.
+   * 
+   * @param scope current scope
+   * @param input 
+   * @param options carries optional attributes values
+   * @return a new instance of AsString
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> AsString create(Scope scope, Operand<T> input, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("AsString", scope.makeOpName("AsString"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.precision != null) {
+          opBuilder.setAttr("precision", opts.precision);
+        }
+        if (opts.scientific != null) {
+          opBuilder.setAttr("scientific", opts.scientific);
+        }
+        if (opts.shortest != null) {
+          opBuilder.setAttr("shortest", opts.shortest);
+        }
+        if (opts.width != null) {
+          opBuilder.setAttr("width", opts.width);
+        }
+        if (opts.fill != null) {
+          opBuilder.setAttr("fill", opts.fill);
+        }
+      }
+    }
+    return new AsString(opBuilder.build());
+  }
+  
+  /**
+   * @param precision The post-decimal precision to use for floating point numbers.
+   * Only used if precision > -1.
+   */
+  public static Options precision(Long precision) {
+    return new Options().precision(precision);
+  }
+  
+  /**
+   * @param scientific Use scientific notation for floating point numbers.
+   */
+  public static Options scientific(Boolean scientific) {
+    return new Options().scientific(scientific);
+  }
+  
+  /**
+   * @param shortest Use shortest representation (either scientific or standard) for
+   * floating point numbers.
+   */
+  public static Options shortest(Boolean shortest) {
+    return new Options().shortest(shortest);
+  }
+  
+  /**
+   * @param width Pad pre-decimal numbers to this width.
+   * Applies to both floating point and integer numbers.
+   * Only used if width > -1.
+   */
+  public static Options width(Long width) {
+    return new Options().width(width);
+  }
+  
+  /**
+   * @param fill The value to pad if width > -1.  If empty, pads with spaces.
+   * Another typical value is '0'.  String cannot be longer than 1 character.
+   */
+  public static Options fill(String fill) {
+    return new Options().fill(fill);
+  }
+  
+  /**
+   */
+  public Output<TString> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<TString> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "AsString";
+  
+  private Output<TString> output;
+  
+  private AsString(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/dtypes/Cast.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/dtypes/Cast.java
new file mode 100644
index 00000000000..66c97240952
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/dtypes/Cast.java
@@ -0,0 +1,110 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.dtypes;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Cast x of type SrcT to y of DstT.
+ * 
+ * @param <U> data type for {@code y()} output
+ */
+@Operator(group = "dtypes")
+public final class Cast<U extends TType> extends RawOp implements Operand<U> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.dtypes.Cast}
+   */
+  public static class Options {
+    
+    /**
+     * @param Truncate 
+     */
+    public Options Truncate(Boolean Truncate) {
+      this.Truncate = Truncate;
+      return this;
+    }
+    
+    private Boolean Truncate;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new Cast operation.
+   * 
+   * @param scope current scope
+   * @param x 
+   * @param DstT 
+   * @param options carries optional attributes values
+   * @return a new instance of Cast
+   */
+  @Endpoint(describeByClass = true)
+  public static <U extends TType, T extends TType> Cast<U> create(Scope scope, Operand<T> x, Class<U> DstT, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Cast", scope.makeOpName("Cast"));
+    opBuilder.addInput(x.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("DstT", DstT);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.Truncate != null) {
+          opBuilder.setAttr("Truncate", opts.Truncate);
+        }
+      }
+    }
+    return new Cast<U>(opBuilder.build());
+  }
+  
+  /**
+   * @param Truncate 
+   */
+  public static Options Truncate(Boolean Truncate) {
+    return new Options().Truncate(Truncate);
+  }
+  
+  /**
+   */
+  public Output<U> y() {
+    return y;
+  }
+  
+  @Override
+  public Output<U> asOutput() {
+    return y;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Cast";
+  
+  private Output<U> y;
+  
+  private Cast(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    y = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/dtypes/Complex.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/dtypes/Complex.java
new file mode 100644
index 00000000000..4d245490516
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/dtypes/Complex.java
@@ -0,0 +1,94 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.dtypes;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Converts two real numbers to a complex number.
+ * <p>
+ * Given a tensor `real` representing the real part of a complex number, and a
+ * tensor `imag` representing the imaginary part of a complex number, this
+ * operation returns complex numbers elementwise of the form \\(a + bj\\), where
+ * <i>a</i> represents the `real` part and <i>b</i> represents the `imag` part.
+ * <p>
+ * The input tensors `real` and `imag` must have the same shape.
+ * <p>
+ * For example:
+ * <pre>{@code
+ * # tensor 'real' is [2.25, 3.25]
+ * # tensor `imag` is [4.75, 5.75]
+ * tf.complex(real, imag) ==> [[2.25 + 4.75j], [3.25 + 5.75j]]
+ * }</pre>
+ * 
+ * 
+ * @param <U> data type for {@code out()} output
+ */
+@Operator(group = "dtypes")
+public final class Complex<U extends TType> extends RawOp implements Operand<U> {
+  
+  /**
+   * Factory method to create a class wrapping a new Complex operation.
+   * 
+   * @param scope current scope
+   * @param real 
+   * @param imag 
+   * @param Tout 
+   * @return a new instance of Complex
+   */
+  @Endpoint(describeByClass = true)
+  public static <U extends TType, T extends TNumber> Complex<U> create(Scope scope, Operand<T> real, Operand<T> imag, Class<U> Tout) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Complex", scope.makeOpName("Complex"));
+    opBuilder.addInput(real.asOutput());
+    opBuilder.addInput(imag.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("Tout", Tout);
+    return new Complex<U>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<U> out() {
+    return out;
+  }
+  
+  @Override
+  public Output<U> asOutput() {
+    return out;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Complex";
+  
+  private Output<U> out;
+  
+  private Complex(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    out = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/dtypes/ToBool.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/dtypes/ToBool.java
new file mode 100644
index 00000000000..72b3d46d00c
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/dtypes/ToBool.java
@@ -0,0 +1,85 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.dtypes;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TBool;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Converts a tensor to a scalar predicate.
+ * <p>
+ * Converts a tensor to a scalar predicate with the following rules:
+ * <p>
+ * - For 0D tensors, truthiness is determined by comparing against a "zero"
+ *   value. For numerical types it is the obvious zero. For strings it is the
+ *   empty string.
+ * <p>
+ * - For >0D tensors, truthiness is determined by looking at the number of
+ *   elements. If has zero elements, then the result is false. Otherwise the
+ *   result is true.
+ * <p>
+ * This matches the behavior of If and While for determining if a tensor counts
+ * as true/false for a branch condition.
+ */
+public final class ToBool extends RawOp implements Operand<TBool> {
+  
+  /**
+   * Factory method to create a class wrapping a new ToBool operation.
+   * 
+   * @param scope current scope
+   * @param input 
+   * @return a new instance of ToBool
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> ToBool create(Scope scope, Operand<T> input) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ToBool", scope.makeOpName("ToBool"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new ToBool(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<TBool> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<TBool> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ToBool";
+  
+  private Output<TBool> output;
+  
+  private ToBool(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesAggregateStats.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesAggregateStats.java
new file mode 100644
index 00000000000..ac772ea4ae0
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesAggregateStats.java
@@ -0,0 +1,86 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.estimator;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.TInt32;
+
+/**
+ * Aggregates the summary of accumulated stats for the batch.
+ * <p>
+ * The summary stats contains gradients and hessians accumulated for each node, feature dimension id and bucket.
+ */
+public final class BoostedTreesAggregateStats extends RawOp implements Operand<TFloat32> {
+  
+  /**
+   * Factory method to create a class wrapping a new BoostedTreesAggregateStats operation.
+   * 
+   * @param scope current scope
+   * @param nodeIds int32; Rank 1 Tensor containing node ids for each example, shape [batch_size].
+   * @param gradients float32; Rank 2 Tensor (shape=[batch_size, logits_dimension]) with gradients for each example.
+   * @param hessians float32; Rank 2 Tensor (shape=[batch_size, hessian_dimension]) with hessians for each example.
+   * @param feature int32; Rank 2 feature Tensors (shape=[batch_size, feature_dimension]).
+   * @param maxSplits int; the maximum number of splits possible in the whole tree.
+   * @param numBuckets int; equals to the maximum possible value of bucketized feature.
+   * @return a new instance of BoostedTreesAggregateStats
+   */
+  @Endpoint(describeByClass = true)
+  public static BoostedTreesAggregateStats create(Scope scope, Operand<TInt32> nodeIds, Operand<TFloat32> gradients, Operand<TFloat32> hessians, Operand<TInt32> feature, Long maxSplits, Long numBuckets) {
+    OperationBuilder opBuilder = scope.env().opBuilder("BoostedTreesAggregateStats", scope.makeOpName("BoostedTreesAggregateStats"));
+    opBuilder.addInput(nodeIds.asOutput());
+    opBuilder.addInput(gradients.asOutput());
+    opBuilder.addInput(hessians.asOutput());
+    opBuilder.addInput(feature.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("max_splits", maxSplits);
+    opBuilder.setAttr("num_buckets", numBuckets);
+    return new BoostedTreesAggregateStats(opBuilder.build());
+  }
+  
+  /**
+   * output Rank 4 Tensor (shape=[splits, feature_dimension, buckets, logits_dimension + hessian_dimension])
+   * containing accumulated stats for each node, feature dimension and bucket.
+   */
+  public Output<TFloat32> statsSummary() {
+    return statsSummary;
+  }
+  
+  @Override
+  public Output<TFloat32> asOutput() {
+    return statsSummary;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "BoostedTreesAggregateStats";
+  
+  private Output<TFloat32> statsSummary;
+  
+  private BoostedTreesAggregateStats(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    statsSummary = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesBucketize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesBucketize.java
new file mode 100644
index 00000000000..6d57706c19e
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesBucketize.java
@@ -0,0 +1,87 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.estimator;
+
+import java.util.Arrays;
+import java.util.Iterator;
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.Operands;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.TInt32;
+
+/**
+ * Bucketize each feature based on bucket boundaries.
+ * <p>
+ * An op that returns a list of float tensors, where each tensor represents the
+ * bucketized values for a single feature.
+ */
+public final class BoostedTreesBucketize extends RawOp implements Iterable<Operand<TInt32>> {
+  
+  /**
+   * Factory method to create a class wrapping a new BoostedTreesBucketize operation.
+   * 
+   * @param scope current scope
+   * @param floatValues float; List of Rank 1 Tensor each containing float values for a single feature.
+   * @param bucketBoundaries float; List of Rank 1 Tensors each containing the bucket boundaries for a single
+   * feature.
+   * @return a new instance of BoostedTreesBucketize
+   */
+  @Endpoint(describeByClass = true)
+  public static BoostedTreesBucketize create(Scope scope, Iterable<Operand<TFloat32>> floatValues, Iterable<Operand<TFloat32>> bucketBoundaries) {
+    OperationBuilder opBuilder = scope.env().opBuilder("BoostedTreesBucketize", scope.makeOpName("BoostedTreesBucketize"));
+    opBuilder.addInputList(Operands.asOutputs(floatValues));
+    opBuilder.addInputList(Operands.asOutputs(bucketBoundaries));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new BoostedTreesBucketize(opBuilder.build());
+  }
+  
+  /**
+   * int; List of Rank 1 Tensors each containing the bucketized values for a single feature.
+   */
+  public List<Output<TInt32>> buckets() {
+    return buckets;
+  }
+  
+  @Override
+  @SuppressWarnings({"rawtypes", "unchecked"})
+  public Iterator<Operand<TInt32>> iterator() {
+    return (Iterator) buckets.iterator();
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "BoostedTreesBucketize";
+  
+  private List<Output<TInt32>> buckets;
+  
+  @SuppressWarnings("unchecked")
+  private BoostedTreesBucketize(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    int bucketsLength = operation.outputListLength("buckets");
+    buckets = Arrays.asList((Output<TInt32>[])operation.outputList(outputIdx, bucketsLength));
+    outputIdx += bucketsLength;
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesCalculateBestFeatureSplit.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesCalculateBestFeatureSplit.java
new file mode 100644
index 00000000000..41f1d3a6910
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesCalculateBestFeatureSplit.java
@@ -0,0 +1,179 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.estimator;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.TString;
+
+/**
+ * Calculates gains for each feature and returns the best possible split information for the feature.
+ * <p>
+ * The split information is the best threshold (bucket id), gains and left/right node contributions per node for each feature.
+ * <p>
+ * It is possible that not all nodes can be split on each feature. Hence, the list of possible nodes can differ between the features. Therefore, we return `node_ids_list` for each feature, containing the list of nodes that this feature can be used to split.
+ * <p>
+ * In this manner, the output is the best split per features and per node, so that it needs to be combined later to produce the best split for each node (among all possible features).
+ * <p>
+ * The output shapes are compatible in a way that the first dimension of all tensors are the same and equal to the number of possible split nodes for each feature.
+ */
+public final class BoostedTreesCalculateBestFeatureSplit extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.estimator.BoostedTreesCalculateBestFeatureSplit}
+   */
+  public static class Options {
+    
+    /**
+     * @param splitType A string indicating if this Op should perform inequality split or equality split.
+     */
+    public Options splitType(String splitType) {
+      this.splitType = splitType;
+      return this;
+    }
+    
+    private String splitType;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new BoostedTreesCalculateBestFeatureSplit operation.
+   * 
+   * @param scope current scope
+   * @param nodeIdRange A Rank 1 tensor (shape=[2]) to specify the range [first, last) of node ids to process within `stats_summary_list`. The nodes are iterated between the two nodes specified by the tensor, as like `for node_id in range(node_id_range[0], node_id_range[1])` (Note that the last index node_id_range[1] is exclusive).
+   * @param statsSummary A Rank 4 tensor (#shape=[max_splits, feature_dims, bucket, stats_dims]) for accumulated stats summary (gradient/hessian) per node, per dimension, per buckets for each feature.
+   * The first dimension of the tensor is the maximum number of splits, and thus not all elements of it will be used, but only the indexes specified by node_ids will be used.
+   * @param l1 l1 regularization factor on leaf weights, per instance based.
+   * @param l2 l2 regularization factor on leaf weights, per instance based.
+   * @param treeComplexity adjustment to the gain, per leaf based.
+   * @param minNodeWeight minimum avg of hessians in a node before required for the node to be considered for splitting.
+   * @param logitsDimension The dimension of logit, i.e., number of classes.
+   * @param options carries optional attributes values
+   * @return a new instance of BoostedTreesCalculateBestFeatureSplit
+   */
+  @Endpoint(describeByClass = true)
+  public static BoostedTreesCalculateBestFeatureSplit create(Scope scope, Operand<TInt32> nodeIdRange, Operand<TFloat32> statsSummary, Operand<TFloat32> l1, Operand<TFloat32> l2, Operand<TFloat32> treeComplexity, Operand<TFloat32> minNodeWeight, Long logitsDimension, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("BoostedTreesCalculateBestFeatureSplit", scope.makeOpName("BoostedTreesCalculateBestFeatureSplit"));
+    opBuilder.addInput(nodeIdRange.asOutput());
+    opBuilder.addInput(statsSummary.asOutput());
+    opBuilder.addInput(l1.asOutput());
+    opBuilder.addInput(l2.asOutput());
+    opBuilder.addInput(treeComplexity.asOutput());
+    opBuilder.addInput(minNodeWeight.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("logits_dimension", logitsDimension);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.splitType != null) {
+          opBuilder.setAttr("split_type", opts.splitType);
+        }
+      }
+    }
+    return new BoostedTreesCalculateBestFeatureSplit(opBuilder.build());
+  }
+  
+  /**
+   * @param splitType A string indicating if this Op should perform inequality split or equality split.
+   */
+  public static Options splitType(String splitType) {
+    return new Options().splitType(splitType);
+  }
+  
+  /**
+   * A Rank 1 tensors indicating possible split node ids for each feature. The length of the list is num_features, but each tensor has different size as each feature provides different possible nodes. See above for details like shapes and sizes.
+   */
+  public Output<TInt32> nodeIds() {
+    return nodeIds;
+  }
+  
+  /**
+   * A Rank 1 tensors indicating the best gains for each feature to split for certain nodes. See above for details like shapes and sizes.
+   */
+  public Output<TFloat32> gains() {
+    return gains;
+  }
+  
+  /**
+   * A Rank 1 tensors indicating the best feature dimension for each feature to split for certain nodes if the feature is multi-dimension. See above for details like shapes and sizes.
+   */
+  public Output<TInt32> featureDimensions() {
+    return featureDimensions;
+  }
+  
+  /**
+   * A Rank 1 tensors indicating the bucket id to compare with (as a threshold) for split in each node. See above for details like shapes and sizes.
+   */
+  public Output<TInt32> thresholds() {
+    return thresholds;
+  }
+  
+  /**
+   * A Rank 2 tensors indicating the contribution of the left nodes when branching from parent nodes (given by the tensor element in the output node_ids_list) to the left direction by the given threshold for each feature. This value will be used to make the left node value by adding to the parent node value. Second dimension size is 1 for 1-dimensional logits, but would be larger for multi-class problems. See above for details like shapes and sizes.
+   */
+  public Output<TFloat32> leftNodeContribs() {
+    return leftNodeContribs;
+  }
+  
+  /**
+   * A Rank 2 tensors, with the same shape/conditions as left_node_contribs_list, but just that the value is for the right node.
+   */
+  public Output<TFloat32> rightNodeContribs() {
+    return rightNodeContribs;
+  }
+  
+  /**
+   * A Rank 1 tensors indicating the which direction to go if data is missing. See above for details like shapes and sizes.
+   * Inequality with default left returns 0, inequality with default right returns 1, equality with default right returns 2.
+   */
+  public Output<TString> splitWithDefaultDirections() {
+    return splitWithDefaultDirections;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "BoostedTreesCalculateBestFeatureSplit";
+  
+  private Output<TInt32> nodeIds;
+  private Output<TFloat32> gains;
+  private Output<TInt32> featureDimensions;
+  private Output<TInt32> thresholds;
+  private Output<TFloat32> leftNodeContribs;
+  private Output<TFloat32> rightNodeContribs;
+  private Output<TString> splitWithDefaultDirections;
+  
+  private BoostedTreesCalculateBestFeatureSplit(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    nodeIds = operation.output(outputIdx++);
+    gains = operation.output(outputIdx++);
+    featureDimensions = operation.output(outputIdx++);
+    thresholds = operation.output(outputIdx++);
+    leftNodeContribs = operation.output(outputIdx++);
+    rightNodeContribs = operation.output(outputIdx++);
+    splitWithDefaultDirections = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesCalculateBestFeatureSplitV2.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesCalculateBestFeatureSplitV2.java
new file mode 100644
index 00000000000..94f4afa4386
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesCalculateBestFeatureSplitV2.java
@@ -0,0 +1,159 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.estimator;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.Operands;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.TString;
+
+/**
+ * Calculates gains for each feature and returns the best possible split information for each node. However, if no split is found, then no split information is returned for that node.
+ * <p>
+ * The split information is the best threshold (bucket id), gains and left/right node contributions per node for each feature.
+ * <p>
+ * It is possible that not all nodes can be split on each feature. Hence, the list of possible nodes can differ between the features. Therefore, we return `node_ids_list` for each feature, containing the list of nodes that this feature can be used to split.
+ * <p>
+ * In this manner, the output is the best split per features and per node, so that it needs to be combined later to produce the best split for each node (among all possible features).
+ * <p>
+ * The output shapes are compatible in a way that the first dimension of all tensors are the same and equal to the number of possible split nodes for each feature.
+ */
+public final class BoostedTreesCalculateBestFeatureSplitV2 extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new BoostedTreesCalculateBestFeatureSplitV2 operation.
+   * 
+   * @param scope current scope
+   * @param nodeIdRange A Rank 1 tensor (shape=[2]) to specify the range [first, last) of node ids to process within `stats_summary_list`. The nodes are iterated between the two nodes specified by the tensor, as like `for node_id in range(node_id_range[0], node_id_range[1])` (Note that the last index node_id_range[1] is exclusive).
+   * @param statsSummariesList A list of Rank 4 tensor (#shape=[max_splits, feature_dims, bucket, stats_dims]) for accumulated stats summary (gradient/hessian) per node, per dimension, per buckets for each feature.
+   * The first dimension of the tensor is the maximum number of splits, and thus not all elements of it will be used, but only the indexes specified by node_ids will be used.
+   * @param splitTypes A Rank 1 tensor indicating if this Op should perform inequality split or equality split per feature.
+   * @param candidateFeatureIds Rank 1 tensor with ids for each feature. This is the real id of the feature.
+   * @param l1 l1 regularization factor on leaf weights, per instance based.
+   * @param l2 l2 regularization factor on leaf weights, per instance based.
+   * @param treeComplexity adjustment to the gain, per leaf based.
+   * @param minNodeWeight minimum avg of hessians in a node before required for the node to be considered for splitting.
+   * @param logitsDimension The dimension of logit, i.e., number of classes.
+   * @return a new instance of BoostedTreesCalculateBestFeatureSplitV2
+   */
+  @Endpoint(describeByClass = true)
+  public static BoostedTreesCalculateBestFeatureSplitV2 create(Scope scope, Operand<TInt32> nodeIdRange, Iterable<Operand<TFloat32>> statsSummariesList, Operand<TString> splitTypes, Operand<TInt32> candidateFeatureIds, Operand<TFloat32> l1, Operand<TFloat32> l2, Operand<TFloat32> treeComplexity, Operand<TFloat32> minNodeWeight, Long logitsDimension) {
+    OperationBuilder opBuilder = scope.env().opBuilder("BoostedTreesCalculateBestFeatureSplitV2", scope.makeOpName("BoostedTreesCalculateBestFeatureSplitV2"));
+    opBuilder.addInput(nodeIdRange.asOutput());
+    opBuilder.addInputList(Operands.asOutputs(statsSummariesList));
+    opBuilder.addInput(splitTypes.asOutput());
+    opBuilder.addInput(candidateFeatureIds.asOutput());
+    opBuilder.addInput(l1.asOutput());
+    opBuilder.addInput(l2.asOutput());
+    opBuilder.addInput(treeComplexity.asOutput());
+    opBuilder.addInput(minNodeWeight.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("logits_dimension", logitsDimension);
+    return new BoostedTreesCalculateBestFeatureSplitV2(opBuilder.build());
+  }
+  
+  /**
+   * A Rank 1 tensors indicating possible split node ids for each feature. The length of the list is num_features, but each tensor has different size as each feature provides different possible nodes. See above for details like shapes and sizes.
+   */
+  public Output<TInt32> nodeIds() {
+    return nodeIds;
+  }
+  
+  /**
+   * A Rank 1 tensor indicating the best gains for each feature to split for certain nodes. See above for details like shapes and sizes.
+   */
+  public Output<TFloat32> gains() {
+    return gains;
+  }
+  
+  /**
+   * A Rank 1 tensors indicating the best feature id for each node. See above for details like shapes and sizes.
+   */
+  public Output<TInt32> featureIds() {
+    return featureIds;
+  }
+  
+  /**
+   * A Rank 1 tensors indicating the best feature dimension for each feature to split for certain nodes if the feature is multi-dimension. See above for details like shapes and sizes.
+   */
+  public Output<TInt32> featureDimensions() {
+    return featureDimensions;
+  }
+  
+  /**
+   * A Rank 1 tensors indicating the bucket id to compare with (as a threshold) for split in each node. See above for details like shapes and sizes.
+   */
+  public Output<TInt32> thresholds() {
+    return thresholds;
+  }
+  
+  /**
+   * A Rank 2 tensors indicating the contribution of the left nodes when branching from parent nodes (given by the tensor element in the output node_ids_list) to the left direction by the given threshold for each feature. This value will be used to make the left node value by adding to the parent node value. Second dimension size is 1 for 1-dimensional logits, but would be larger for multi-class problems. See above for details like shapes and sizes.
+   */
+  public Output<TFloat32> leftNodeContribs() {
+    return leftNodeContribs;
+  }
+  
+  /**
+   * A Rank 2 tensors, with the same shape/conditions as left_node_contribs_list, but just that the value is for the right node.
+   */
+  public Output<TFloat32> rightNodeContribs() {
+    return rightNodeContribs;
+  }
+  
+  /**
+   * A Rank 1 tensors indicating the which direction to go if data is missing. See above for details like shapes and sizes.
+   * Inequality with default left returns 0, inequality with default right returns 1, equality with default right returns 2.
+   */
+  public Output<TString> splitWithDefaultDirections() {
+    return splitWithDefaultDirections;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "BoostedTreesCalculateBestFeatureSplitV2";
+  
+  private Output<TInt32> nodeIds;
+  private Output<TFloat32> gains;
+  private Output<TInt32> featureIds;
+  private Output<TInt32> featureDimensions;
+  private Output<TInt32> thresholds;
+  private Output<TFloat32> leftNodeContribs;
+  private Output<TFloat32> rightNodeContribs;
+  private Output<TString> splitWithDefaultDirections;
+  
+  private BoostedTreesCalculateBestFeatureSplitV2(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    nodeIds = operation.output(outputIdx++);
+    gains = operation.output(outputIdx++);
+    featureIds = operation.output(outputIdx++);
+    featureDimensions = operation.output(outputIdx++);
+    thresholds = operation.output(outputIdx++);
+    leftNodeContribs = operation.output(outputIdx++);
+    rightNodeContribs = operation.output(outputIdx++);
+    splitWithDefaultDirections = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesCalculateBestGainsPerFeature.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesCalculateBestGainsPerFeature.java
new file mode 100644
index 00000000000..daade3457ae
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesCalculateBestGainsPerFeature.java
@@ -0,0 +1,139 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.estimator;
+
+import java.util.Arrays;
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.Operands;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.TInt32;
+
+/**
+ * Calculates gains for each feature and returns the best possible split information for the feature.
+ * <p>
+ * The split information is the best threshold (bucket id), gains and left/right node contributions per node for each feature.
+ * <p>
+ * It is possible that not all nodes can be split on each feature. Hence, the list of possible nodes can differ between the features. Therefore, we return `node_ids_list` for each feature, containing the list of nodes that this feature can be used to split.
+ * <p>
+ * In this manner, the output is the best split per features and per node, so that it needs to be combined later to produce the best split for each node (among all possible features).
+ * <p>
+ * The length of output lists are all of the same length, `num_features`.
+ * The output shapes are compatible in a way that the first dimension of all tensors of all lists are the same and equal to the number of possible split nodes for each feature.
+ */
+public final class BoostedTreesCalculateBestGainsPerFeature extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new BoostedTreesCalculateBestGainsPerFeature operation.
+   * 
+   * @param scope current scope
+   * @param nodeIdRange A Rank 1 tensor (shape=[2]) to specify the range [first, last) of node ids to process within `stats_summary_list`. The nodes are iterated between the two nodes specified by the tensor, as like `for node_id in range(node_id_range[0], node_id_range[1])` (Note that the last index node_id_range[1] is exclusive).
+   * @param statsSummaryList A list of Rank 3 tensor (#shape=[max_splits, bucket, 2]) for accumulated stats summary (gradient/hessian) per node per buckets for each feature. The first dimension of the tensor is the maximum number of splits, and thus not all elements of it will be used, but only the indexes specified by node_ids will be used.
+   * @param l1 l1 regularization factor on leaf weights, per instance based.
+   * @param l2 l2 regularization factor on leaf weights, per instance based.
+   * @param treeComplexity adjustment to the gain, per leaf based.
+   * @param minNodeWeight minimum avg of hessians in a node before required for the node to be considered for splitting.
+   * @param maxSplits the number of nodes that can be split in the whole tree. Used as a dimension of output tensors.
+   * @return a new instance of BoostedTreesCalculateBestGainsPerFeature
+   */
+  @Endpoint(describeByClass = true)
+  public static BoostedTreesCalculateBestGainsPerFeature create(Scope scope, Operand<TInt32> nodeIdRange, Iterable<Operand<TFloat32>> statsSummaryList, Operand<TFloat32> l1, Operand<TFloat32> l2, Operand<TFloat32> treeComplexity, Operand<TFloat32> minNodeWeight, Long maxSplits) {
+    OperationBuilder opBuilder = scope.env().opBuilder("BoostedTreesCalculateBestGainsPerFeature", scope.makeOpName("BoostedTreesCalculateBestGainsPerFeature"));
+    opBuilder.addInput(nodeIdRange.asOutput());
+    opBuilder.addInputList(Operands.asOutputs(statsSummaryList));
+    opBuilder.addInput(l1.asOutput());
+    opBuilder.addInput(l2.asOutput());
+    opBuilder.addInput(treeComplexity.asOutput());
+    opBuilder.addInput(minNodeWeight.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("max_splits", maxSplits);
+    return new BoostedTreesCalculateBestGainsPerFeature(opBuilder.build());
+  }
+  
+  /**
+   * An output list of Rank 1 tensors indicating possible split node ids for each feature. The length of the list is num_features, but each tensor has different size as each feature provides different possible nodes. See above for details like shapes and sizes.
+   */
+  public List<Output<TInt32>> nodeIdsList() {
+    return nodeIdsList;
+  }
+  
+  /**
+   * An output list of Rank 1 tensors indicating the best gains for each feature to split for certain nodes. See above for details like shapes and sizes.
+   */
+  public List<Output<TFloat32>> gainsList() {
+    return gainsList;
+  }
+  
+  /**
+   * An output list of Rank 1 tensors indicating the bucket id to compare with (as a threshold) for split in each node. See above for details like shapes and sizes.
+   */
+  public List<Output<TInt32>> thresholdsList() {
+    return thresholdsList;
+  }
+  
+  /**
+   * A list of Rank 2 tensors indicating the contribution of the left nodes when branching from parent nodes (given by the tensor element in the output node_ids_list) to the left direction by the given threshold for each feature. This value will be used to make the left node value by adding to the parent node value. Second dimension size is 1 for 1-dimensional logits, but would be larger for multi-class problems. See above for details like shapes and sizes.
+   */
+  public List<Output<TFloat32>> leftNodeContribsList() {
+    return leftNodeContribsList;
+  }
+  
+  /**
+   * A list of Rank 2 tensors, with the same shape/conditions as left_node_contribs_list, but just that the value is for the right node.
+   */
+  public List<Output<TFloat32>> rightNodeContribsList() {
+    return rightNodeContribsList;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "BoostedTreesCalculateBestGainsPerFeature";
+  
+  private List<Output<TInt32>> nodeIdsList;
+  private List<Output<TFloat32>> gainsList;
+  private List<Output<TInt32>> thresholdsList;
+  private List<Output<TFloat32>> leftNodeContribsList;
+  private List<Output<TFloat32>> rightNodeContribsList;
+  
+  @SuppressWarnings("unchecked")
+  private BoostedTreesCalculateBestGainsPerFeature(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    int nodeIdsListLength = operation.outputListLength("node_ids_list");
+    nodeIdsList = Arrays.asList((Output<TInt32>[])operation.outputList(outputIdx, nodeIdsListLength));
+    outputIdx += nodeIdsListLength;
+    int gainsListLength = operation.outputListLength("gains_list");
+    gainsList = Arrays.asList((Output<TFloat32>[])operation.outputList(outputIdx, gainsListLength));
+    outputIdx += gainsListLength;
+    int thresholdsListLength = operation.outputListLength("thresholds_list");
+    thresholdsList = Arrays.asList((Output<TInt32>[])operation.outputList(outputIdx, thresholdsListLength));
+    outputIdx += thresholdsListLength;
+    int leftNodeContribsListLength = operation.outputListLength("left_node_contribs_list");
+    leftNodeContribsList = Arrays.asList((Output<TFloat32>[])operation.outputList(outputIdx, leftNodeContribsListLength));
+    outputIdx += leftNodeContribsListLength;
+    int rightNodeContribsListLength = operation.outputListLength("right_node_contribs_list");
+    rightNodeContribsList = Arrays.asList((Output<TFloat32>[])operation.outputList(outputIdx, rightNodeContribsListLength));
+    outputIdx += rightNodeContribsListLength;
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesCenterBias.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesCenterBias.java
new file mode 100644
index 00000000000..7359b6039e2
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesCenterBias.java
@@ -0,0 +1,81 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.estimator;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TBool;
+import org.tensorflow.types.TFloat32;
+
+/**
+ * Calculates the prior from the training data (the bias) and fills in the first node with the logits' prior. Returns a boolean indicating whether to continue centering.
+ */
+public final class BoostedTreesCenterBias extends RawOp implements Operand<TBool> {
+  
+  /**
+   * Factory method to create a class wrapping a new BoostedTreesCenterBias operation.
+   * 
+   * @param scope current scope
+   * @param treeEnsembleHandle Handle to the tree ensemble.
+   * @param meanGradients A tensor with shape=[logits_dimension] with mean of gradients for a first node.
+   * @param meanHessians A tensor with shape=[logits_dimension] mean of hessians for a first node.
+   * @param l1 l1 regularization factor on leaf weights, per instance based.
+   * @param l2 l2 regularization factor on leaf weights, per instance based.
+   * @return a new instance of BoostedTreesCenterBias
+   */
+  @Endpoint(describeByClass = true)
+  public static BoostedTreesCenterBias create(Scope scope, Operand<?> treeEnsembleHandle, Operand<TFloat32> meanGradients, Operand<TFloat32> meanHessians, Operand<TFloat32> l1, Operand<TFloat32> l2) {
+    OperationBuilder opBuilder = scope.env().opBuilder("BoostedTreesCenterBias", scope.makeOpName("BoostedTreesCenterBias"));
+    opBuilder.addInput(treeEnsembleHandle.asOutput());
+    opBuilder.addInput(meanGradients.asOutput());
+    opBuilder.addInput(meanHessians.asOutput());
+    opBuilder.addInput(l1.asOutput());
+    opBuilder.addInput(l2.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new BoostedTreesCenterBias(opBuilder.build());
+  }
+  
+  /**
+   * Bool, whether to continue bias centering.
+   */
+  public Output<TBool> continueCentering() {
+    return continueCentering;
+  }
+  
+  @Override
+  public Output<TBool> asOutput() {
+    return continueCentering;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "BoostedTreesCenterBias";
+  
+  private Output<TBool> continueCentering;
+  
+  private BoostedTreesCenterBias(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    continueCentering = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesCreateEnsemble.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesCreateEnsemble.java
new file mode 100644
index 00000000000..8841988b36d
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesCreateEnsemble.java
@@ -0,0 +1,60 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.estimator;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.TString;
+
+/**
+ * Creates a tree ensemble model and returns a handle to it.
+ */
+public final class BoostedTreesCreateEnsemble extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new BoostedTreesCreateEnsemble operation.
+   * 
+   * @param scope current scope
+   * @param treeEnsembleHandle Handle to the tree ensemble resource to be created.
+   * @param stampToken Token to use as the initial value of the resource stamp.
+   * @param treeEnsembleSerialized Serialized proto of the tree ensemble.
+   * @return a new instance of BoostedTreesCreateEnsemble
+   */
+  @Endpoint(describeByClass = true)
+  public static BoostedTreesCreateEnsemble create(Scope scope, Operand<?> treeEnsembleHandle, Operand<TInt64> stampToken, Operand<TString> treeEnsembleSerialized) {
+    OperationBuilder opBuilder = scope.env().opBuilder("BoostedTreesCreateEnsemble", scope.makeOpName("BoostedTreesCreateEnsemble"));
+    opBuilder.addInput(treeEnsembleHandle.asOutput());
+    opBuilder.addInput(stampToken.asOutput());
+    opBuilder.addInput(treeEnsembleSerialized.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new BoostedTreesCreateEnsemble(opBuilder.build());
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "BoostedTreesCreateEnsemble";
+  
+  private BoostedTreesCreateEnsemble(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesCreateQuantileStreamResource.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesCreateQuantileStreamResource.java
new file mode 100644
index 00000000000..802a61ecb2a
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesCreateQuantileStreamResource.java
@@ -0,0 +1,94 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.estimator;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.TInt64;
+
+/**
+ * Create the Resource for Quantile Streams.
+ */
+public final class BoostedTreesCreateQuantileStreamResource extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.estimator.BoostedTreesCreateQuantileStreamResource}
+   */
+  public static class Options {
+    
+    /**
+     * @param maxElements int; The maximum number of data points that can be fed to the stream.
+     */
+    public Options maxElements(Long maxElements) {
+      this.maxElements = maxElements;
+      return this;
+    }
+    
+    private Long maxElements;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new BoostedTreesCreateQuantileStreamResource operation.
+   * 
+   * @param scope current scope
+   * @param quantileStreamResourceHandle resource; Handle to quantile stream resource.
+   * @param epsilon float; The required approximation error of the stream resource.
+   * @param numStreams int; The number of streams managed by the resource that shares the same epsilon.
+   * @param options carries optional attributes values
+   * @return a new instance of BoostedTreesCreateQuantileStreamResource
+   */
+  @Endpoint(describeByClass = true)
+  public static BoostedTreesCreateQuantileStreamResource create(Scope scope, Operand<?> quantileStreamResourceHandle, Operand<TFloat32> epsilon, Operand<TInt64> numStreams, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("BoostedTreesCreateQuantileStreamResource", scope.makeOpName("BoostedTreesCreateQuantileStreamResource"));
+    opBuilder.addInput(quantileStreamResourceHandle.asOutput());
+    opBuilder.addInput(epsilon.asOutput());
+    opBuilder.addInput(numStreams.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.maxElements != null) {
+          opBuilder.setAttr("max_elements", opts.maxElements);
+        }
+      }
+    }
+    return new BoostedTreesCreateQuantileStreamResource(opBuilder.build());
+  }
+  
+  /**
+   * @param maxElements int; The maximum number of data points that can be fed to the stream.
+   */
+  public static Options maxElements(Long maxElements) {
+    return new Options().maxElements(maxElements);
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "BoostedTreesCreateQuantileStreamResource";
+  
+  private BoostedTreesCreateQuantileStreamResource(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesDeserializeEnsemble.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesDeserializeEnsemble.java
new file mode 100644
index 00000000000..15371fb4df9
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesDeserializeEnsemble.java
@@ -0,0 +1,62 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.estimator;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.TString;
+
+/**
+ * Deserializes a serialized tree ensemble config and replaces current tree
+ * <p>
+ * ensemble.
+ */
+public final class BoostedTreesDeserializeEnsemble extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new BoostedTreesDeserializeEnsemble operation.
+   * 
+   * @param scope current scope
+   * @param treeEnsembleHandle Handle to the tree ensemble.
+   * @param stampToken Token to use as the new value of the resource stamp.
+   * @param treeEnsembleSerialized Serialized proto of the ensemble.
+   * @return a new instance of BoostedTreesDeserializeEnsemble
+   */
+  @Endpoint(describeByClass = true)
+  public static BoostedTreesDeserializeEnsemble create(Scope scope, Operand<?> treeEnsembleHandle, Operand<TInt64> stampToken, Operand<TString> treeEnsembleSerialized) {
+    OperationBuilder opBuilder = scope.env().opBuilder("BoostedTreesDeserializeEnsemble", scope.makeOpName("BoostedTreesDeserializeEnsemble"));
+    opBuilder.addInput(treeEnsembleHandle.asOutput());
+    opBuilder.addInput(stampToken.asOutput());
+    opBuilder.addInput(treeEnsembleSerialized.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new BoostedTreesDeserializeEnsemble(opBuilder.build());
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "BoostedTreesDeserializeEnsemble";
+  
+  private BoostedTreesDeserializeEnsemble(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesEnsembleResourceHandleOp.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesEnsembleResourceHandleOp.java
new file mode 100644
index 00000000000..13522dd2fd0
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesEnsembleResourceHandleOp.java
@@ -0,0 +1,123 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.estimator;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Creates a handle to a BoostedTreesEnsembleResource
+ */
+public final class BoostedTreesEnsembleResourceHandleOp extends RawOp implements Operand<TType> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.estimator.BoostedTreesEnsembleResourceHandleOp}
+   */
+  public static class Options {
+    
+    /**
+     * @param container 
+     */
+    public Options container(String container) {
+      this.container = container;
+      return this;
+    }
+    
+    /**
+     * @param sharedName 
+     */
+    public Options sharedName(String sharedName) {
+      this.sharedName = sharedName;
+      return this;
+    }
+    
+    private String container;
+    private String sharedName;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new BoostedTreesEnsembleResourceHandleOp operation.
+   * 
+   * @param scope current scope
+   * @param options carries optional attributes values
+   * @return a new instance of BoostedTreesEnsembleResourceHandleOp
+   */
+  @Endpoint(describeByClass = true)
+  public static BoostedTreesEnsembleResourceHandleOp create(Scope scope, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("BoostedTreesEnsembleResourceHandleOp", scope.makeOpName("BoostedTreesEnsembleResourceHandleOp"));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.container != null) {
+          opBuilder.setAttr("container", opts.container);
+        }
+        if (opts.sharedName != null) {
+          opBuilder.setAttr("shared_name", opts.sharedName);
+        }
+      }
+    }
+    return new BoostedTreesEnsembleResourceHandleOp(opBuilder.build());
+  }
+  
+  /**
+   * @param container 
+   */
+  public static Options container(String container) {
+    return new Options().container(container);
+  }
+  
+  /**
+   * @param sharedName 
+   */
+  public static Options sharedName(String sharedName) {
+    return new Options().sharedName(sharedName);
+  }
+  
+  /**
+   */
+  public Output<?> resource() {
+    return resource;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) resource;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "BoostedTreesEnsembleResourceHandleOp";
+  
+  private Output<?> resource;
+  
+  private BoostedTreesEnsembleResourceHandleOp(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    resource = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesExampleDebugOutputs.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesExampleDebugOutputs.java
new file mode 100644
index 00000000000..8618f401ce4
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesExampleDebugOutputs.java
@@ -0,0 +1,84 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.estimator;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.Operands;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.TString;
+
+/**
+ * Debugging/model interpretability outputs for each example.
+ * <p>
+ * It traverses all the trees and computes debug metrics for individual examples,
+ * such as getting split feature ids and logits after each split along the decision
+ * path used to compute directional feature contributions.
+ */
+public final class BoostedTreesExampleDebugOutputs extends RawOp implements Operand<TString> {
+  
+  /**
+   * Factory method to create a class wrapping a new BoostedTreesExampleDebugOutputs operation.
+   * 
+   * @param scope current scope
+   * @param treeEnsembleHandle 
+   * @param bucketizedFeatures A list of rank 1 Tensors containing bucket id for each
+   * feature.
+   * @param logitsDimension scalar, dimension of the logits, to be used for constructing the protos in
+   * examples_debug_outputs_serialized.
+   * @return a new instance of BoostedTreesExampleDebugOutputs
+   */
+  @Endpoint(describeByClass = true)
+  public static BoostedTreesExampleDebugOutputs create(Scope scope, Operand<?> treeEnsembleHandle, Iterable<Operand<TInt32>> bucketizedFeatures, Long logitsDimension) {
+    OperationBuilder opBuilder = scope.env().opBuilder("BoostedTreesExampleDebugOutputs", scope.makeOpName("BoostedTreesExampleDebugOutputs"));
+    opBuilder.addInput(treeEnsembleHandle.asOutput());
+    opBuilder.addInputList(Operands.asOutputs(bucketizedFeatures));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("logits_dimension", logitsDimension);
+    return new BoostedTreesExampleDebugOutputs(opBuilder.build());
+  }
+  
+  /**
+   * Output rank 1 Tensor containing a proto serialized as a string for each example.
+   */
+  public Output<TString> examplesDebugOutputsSerialized() {
+    return examplesDebugOutputsSerialized;
+  }
+  
+  @Override
+  public Output<TString> asOutput() {
+    return examplesDebugOutputsSerialized;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "BoostedTreesExampleDebugOutputs";
+  
+  private Output<TString> examplesDebugOutputsSerialized;
+  
+  private BoostedTreesExampleDebugOutputs(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    examplesDebugOutputsSerialized = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesFlushQuantileSummaries.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesFlushQuantileSummaries.java
new file mode 100644
index 00000000000..d22acc2ca1b
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesFlushQuantileSummaries.java
@@ -0,0 +1,84 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.estimator;
+
+import java.util.Arrays;
+import java.util.Iterator;
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+
+/**
+ * Flush the quantile summaries from each quantile stream resource.
+ * <p>
+ * An op that outputs a list of quantile summaries of a quantile stream resource.
+ * Each summary Tensor is rank 2, containing summaries (value, weight, min_rank,
+ * max_rank) for a single feature.
+ */
+public final class BoostedTreesFlushQuantileSummaries extends RawOp implements Iterable<Operand<TFloat32>> {
+  
+  /**
+   * Factory method to create a class wrapping a new BoostedTreesFlushQuantileSummaries operation.
+   * 
+   * @param scope current scope
+   * @param quantileStreamResourceHandle resource handle referring to a QuantileStreamResource.
+   * @param numFeatures 
+   * @return a new instance of BoostedTreesFlushQuantileSummaries
+   */
+  @Endpoint(describeByClass = true)
+  public static BoostedTreesFlushQuantileSummaries create(Scope scope, Operand<?> quantileStreamResourceHandle, Long numFeatures) {
+    OperationBuilder opBuilder = scope.env().opBuilder("BoostedTreesFlushQuantileSummaries", scope.makeOpName("BoostedTreesFlushQuantileSummaries"));
+    opBuilder.addInput(quantileStreamResourceHandle.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("num_features", numFeatures);
+    return new BoostedTreesFlushQuantileSummaries(opBuilder.build());
+  }
+  
+  /**
+   */
+  public List<Output<TFloat32>> summaries() {
+    return summaries;
+  }
+  
+  @Override
+  @SuppressWarnings({"rawtypes", "unchecked"})
+  public Iterator<Operand<TFloat32>> iterator() {
+    return (Iterator) summaries.iterator();
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "BoostedTreesFlushQuantileSummaries";
+  
+  private List<Output<TFloat32>> summaries;
+  
+  @SuppressWarnings("unchecked")
+  private BoostedTreesFlushQuantileSummaries(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    int summariesLength = operation.outputListLength("summaries");
+    summaries = Arrays.asList((Output<TFloat32>[])operation.outputList(outputIdx, summariesLength));
+    outputIdx += summariesLength;
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesGetEnsembleStates.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesGetEnsembleStates.java
new file mode 100644
index 00000000000..36e8eb3c4be
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesGetEnsembleStates.java
@@ -0,0 +1,105 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.estimator;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.TInt64;
+
+/**
+ * Retrieves the tree ensemble resource stamp token, number of trees and growing statistics.
+ */
+public final class BoostedTreesGetEnsembleStates extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new BoostedTreesGetEnsembleStates operation.
+   * 
+   * @param scope current scope
+   * @param treeEnsembleHandle Handle to the tree ensemble.
+   * @return a new instance of BoostedTreesGetEnsembleStates
+   */
+  @Endpoint(describeByClass = true)
+  public static BoostedTreesGetEnsembleStates create(Scope scope, Operand<?> treeEnsembleHandle) {
+    OperationBuilder opBuilder = scope.env().opBuilder("BoostedTreesGetEnsembleStates", scope.makeOpName("BoostedTreesGetEnsembleStates"));
+    opBuilder.addInput(treeEnsembleHandle.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new BoostedTreesGetEnsembleStates(opBuilder.build());
+  }
+  
+  /**
+   * Stamp token of the tree ensemble resource.
+   */
+  public Output<TInt64> stampToken() {
+    return stampToken;
+  }
+  
+  /**
+   * The number of trees in the tree ensemble resource.
+   */
+  public Output<TInt32> numTrees() {
+    return numTrees;
+  }
+  
+  /**
+   * The number of trees that were finished successfully.
+   */
+  public Output<TInt32> numFinalizedTrees() {
+    return numFinalizedTrees;
+  }
+  
+  /**
+   * The number of layers we attempted to build (but not necessarily succeeded).
+   */
+  public Output<TInt32> numAttemptedLayers() {
+    return numAttemptedLayers;
+  }
+  
+  /**
+   * Rank size 2 tensor that contains start and end ids of the nodes in the latest
+   * layer.
+   */
+  public Output<TInt32> lastLayerNodesRange() {
+    return lastLayerNodesRange;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "BoostedTreesGetEnsembleStates";
+  
+  private Output<TInt64> stampToken;
+  private Output<TInt32> numTrees;
+  private Output<TInt32> numFinalizedTrees;
+  private Output<TInt32> numAttemptedLayers;
+  private Output<TInt32> lastLayerNodesRange;
+  
+  private BoostedTreesGetEnsembleStates(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    stampToken = operation.output(outputIdx++);
+    numTrees = operation.output(outputIdx++);
+    numFinalizedTrees = operation.output(outputIdx++);
+    numAttemptedLayers = operation.output(outputIdx++);
+    lastLayerNodesRange = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesMakeQuantileSummaries.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesMakeQuantileSummaries.java
new file mode 100644
index 00000000000..cf4156bb2ce
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesMakeQuantileSummaries.java
@@ -0,0 +1,88 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.estimator;
+
+import java.util.Arrays;
+import java.util.Iterator;
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.Operands;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+
+/**
+ * Makes the summary of quantiles for the batch.
+ * <p>
+ * An op that takes a list of tensors (one tensor per feature) and outputs the
+ * quantile summaries for each tensor.
+ */
+public final class BoostedTreesMakeQuantileSummaries extends RawOp implements Iterable<Operand<TFloat32>> {
+  
+  /**
+   * Factory method to create a class wrapping a new BoostedTreesMakeQuantileSummaries operation.
+   * 
+   * @param scope current scope
+   * @param floatValues float; List of Rank 1 Tensors each containing values for a single feature.
+   * @param exampleWeights float; Rank 1 Tensor with weights per instance.
+   * @param epsilon float; The required maximum approximation error.
+   * @return a new instance of BoostedTreesMakeQuantileSummaries
+   */
+  @Endpoint(describeByClass = true)
+  public static BoostedTreesMakeQuantileSummaries create(Scope scope, Iterable<Operand<TFloat32>> floatValues, Operand<TFloat32> exampleWeights, Operand<TFloat32> epsilon) {
+    OperationBuilder opBuilder = scope.env().opBuilder("BoostedTreesMakeQuantileSummaries", scope.makeOpName("BoostedTreesMakeQuantileSummaries"));
+    opBuilder.addInputList(Operands.asOutputs(floatValues));
+    opBuilder.addInput(exampleWeights.asOutput());
+    opBuilder.addInput(epsilon.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new BoostedTreesMakeQuantileSummaries(opBuilder.build());
+  }
+  
+  /**
+   * float; List of Rank 2 Tensors each containing the quantile summary
+   * (value, weight, min_rank, max_rank) of a single feature.
+   */
+  public List<Output<TFloat32>> summaries() {
+    return summaries;
+  }
+  
+  @Override
+  @SuppressWarnings({"rawtypes", "unchecked"})
+  public Iterator<Operand<TFloat32>> iterator() {
+    return (Iterator) summaries.iterator();
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "BoostedTreesMakeQuantileSummaries";
+  
+  private List<Output<TFloat32>> summaries;
+  
+  @SuppressWarnings("unchecked")
+  private BoostedTreesMakeQuantileSummaries(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    int summariesLength = operation.outputListLength("summaries");
+    summaries = Arrays.asList((Output<TFloat32>[])operation.outputList(outputIdx, summariesLength));
+    outputIdx += summariesLength;
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesMakeStatsSummary.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesMakeStatsSummary.java
new file mode 100644
index 00000000000..626e05a6f54
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesMakeStatsSummary.java
@@ -0,0 +1,86 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.estimator;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.Operands;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.TInt32;
+
+/**
+ * Makes the summary of accumulated stats for the batch.
+ * <p>
+ * The summary stats contains gradients and hessians accumulated into the corresponding node and bucket for each example.
+ */
+public final class BoostedTreesMakeStatsSummary extends RawOp implements Operand<TFloat32> {
+  
+  /**
+   * Factory method to create a class wrapping a new BoostedTreesMakeStatsSummary operation.
+   * 
+   * @param scope current scope
+   * @param nodeIds int32 Rank 1 Tensor containing node ids, which each example falls into for the requested layer.
+   * @param gradients float32; Rank 2 Tensor (shape=[#examples, 1]) for gradients.
+   * @param hessians float32; Rank 2 Tensor (shape=[#examples, 1]) for hessians.
+   * @param bucketizedFeaturesList int32 list of Rank 1 Tensors, each containing the bucketized feature (for each feature column).
+   * @param maxSplits int; the maximum number of splits possible in the whole tree.
+   * @param numBuckets int; equals to the maximum possible value of bucketized feature.
+   * @return a new instance of BoostedTreesMakeStatsSummary
+   */
+  @Endpoint(describeByClass = true)
+  public static BoostedTreesMakeStatsSummary create(Scope scope, Operand<TInt32> nodeIds, Operand<TFloat32> gradients, Operand<TFloat32> hessians, Iterable<Operand<TInt32>> bucketizedFeaturesList, Long maxSplits, Long numBuckets) {
+    OperationBuilder opBuilder = scope.env().opBuilder("BoostedTreesMakeStatsSummary", scope.makeOpName("BoostedTreesMakeStatsSummary"));
+    opBuilder.addInput(nodeIds.asOutput());
+    opBuilder.addInput(gradients.asOutput());
+    opBuilder.addInput(hessians.asOutput());
+    opBuilder.addInputList(Operands.asOutputs(bucketizedFeaturesList));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("max_splits", maxSplits);
+    opBuilder.setAttr("num_buckets", numBuckets);
+    return new BoostedTreesMakeStatsSummary(opBuilder.build());
+  }
+  
+  /**
+   * output Rank 4 Tensor (shape=[#features, #splits, #buckets, 2]) containing accumulated stats put into the corresponding node and bucket. The first index of 4th dimension refers to gradients, and the second to hessians.
+   */
+  public Output<TFloat32> statsSummary() {
+    return statsSummary;
+  }
+  
+  @Override
+  public Output<TFloat32> asOutput() {
+    return statsSummary;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "BoostedTreesMakeStatsSummary";
+  
+  private Output<TFloat32> statsSummary;
+  
+  private BoostedTreesMakeStatsSummary(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    statsSummary = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesPredict.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesPredict.java
new file mode 100644
index 00000000000..e9c7d4a2be1
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesPredict.java
@@ -0,0 +1,83 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.estimator;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.Operands;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.TInt32;
+
+/**
+ * Runs multiple additive regression ensemble predictors on input instances and
+ * <p>
+ * computes the logits. It is designed to be used during prediction.
+ * It traverses all the trees and calculates the final score for each instance.
+ */
+public final class BoostedTreesPredict extends RawOp implements Operand<TFloat32> {
+  
+  /**
+   * Factory method to create a class wrapping a new BoostedTreesPredict operation.
+   * 
+   * @param scope current scope
+   * @param treeEnsembleHandle 
+   * @param bucketizedFeatures A list of rank 1 Tensors containing bucket id for each
+   * feature.
+   * @param logitsDimension scalar, dimension of the logits, to be used for partial logits
+   * shape.
+   * @return a new instance of BoostedTreesPredict
+   */
+  @Endpoint(describeByClass = true)
+  public static BoostedTreesPredict create(Scope scope, Operand<?> treeEnsembleHandle, Iterable<Operand<TInt32>> bucketizedFeatures, Long logitsDimension) {
+    OperationBuilder opBuilder = scope.env().opBuilder("BoostedTreesPredict", scope.makeOpName("BoostedTreesPredict"));
+    opBuilder.addInput(treeEnsembleHandle.asOutput());
+    opBuilder.addInputList(Operands.asOutputs(bucketizedFeatures));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("logits_dimension", logitsDimension);
+    return new BoostedTreesPredict(opBuilder.build());
+  }
+  
+  /**
+   * Output rank 2 Tensor containing logits for each example.
+   */
+  public Output<TFloat32> logits() {
+    return logits;
+  }
+  
+  @Override
+  public Output<TFloat32> asOutput() {
+    return logits;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "BoostedTreesPredict";
+  
+  private Output<TFloat32> logits;
+  
+  private BoostedTreesPredict(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    logits = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesQuantileStreamResourceAddSummaries.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesQuantileStreamResourceAddSummaries.java
new file mode 100644
index 00000000000..418ff3b2ff6
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesQuantileStreamResourceAddSummaries.java
@@ -0,0 +1,62 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.estimator;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.Operands;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+
+/**
+ * Add the quantile summaries to each quantile stream resource.
+ * <p>
+ * An op that adds a list of quantile summaries to a quantile stream resource. Each
+ * summary Tensor is rank 2, containing summaries (value, weight, min_rank, max_rank)
+ * for a single feature.
+ */
+public final class BoostedTreesQuantileStreamResourceAddSummaries extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new BoostedTreesQuantileStreamResourceAddSummaries operation.
+   * 
+   * @param scope current scope
+   * @param quantileStreamResourceHandle resource handle referring to a QuantileStreamResource.
+   * @param summaries string; List of Rank 2 Tensor each containing the summaries for a single feature.
+   * @return a new instance of BoostedTreesQuantileStreamResourceAddSummaries
+   */
+  @Endpoint(describeByClass = true)
+  public static BoostedTreesQuantileStreamResourceAddSummaries create(Scope scope, Operand<?> quantileStreamResourceHandle, Iterable<Operand<TFloat32>> summaries) {
+    OperationBuilder opBuilder = scope.env().opBuilder("BoostedTreesQuantileStreamResourceAddSummaries", scope.makeOpName("BoostedTreesQuantileStreamResourceAddSummaries"));
+    opBuilder.addInput(quantileStreamResourceHandle.asOutput());
+    opBuilder.addInputList(Operands.asOutputs(summaries));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new BoostedTreesQuantileStreamResourceAddSummaries(opBuilder.build());
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "BoostedTreesQuantileStreamResourceAddSummaries";
+  
+  private BoostedTreesQuantileStreamResourceAddSummaries(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesQuantileStreamResourceDeserialize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesQuantileStreamResourceDeserialize.java
new file mode 100644
index 00000000000..6efb58ed60c
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesQuantileStreamResourceDeserialize.java
@@ -0,0 +1,60 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.estimator;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.Operands;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+
+/**
+ * Deserialize bucket boundaries and ready flag into current QuantileAccumulator.
+ * <p>
+ * An op that deserializes bucket boundaries and are boundaries ready flag into current QuantileAccumulator.
+ */
+public final class BoostedTreesQuantileStreamResourceDeserialize extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new BoostedTreesQuantileStreamResourceDeserialize operation.
+   * 
+   * @param scope current scope
+   * @param quantileStreamResourceHandle resource handle referring to a QuantileStreamResource.
+   * @param bucketBoundaries float; List of Rank 1 Tensors each containing the bucket boundaries for a feature.
+   * @return a new instance of BoostedTreesQuantileStreamResourceDeserialize
+   */
+  @Endpoint(describeByClass = true)
+  public static BoostedTreesQuantileStreamResourceDeserialize create(Scope scope, Operand<?> quantileStreamResourceHandle, Iterable<Operand<TFloat32>> bucketBoundaries) {
+    OperationBuilder opBuilder = scope.env().opBuilder("BoostedTreesQuantileStreamResourceDeserialize", scope.makeOpName("BoostedTreesQuantileStreamResourceDeserialize"));
+    opBuilder.addInput(quantileStreamResourceHandle.asOutput());
+    opBuilder.addInputList(Operands.asOutputs(bucketBoundaries));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new BoostedTreesQuantileStreamResourceDeserialize(opBuilder.build());
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "BoostedTreesQuantileStreamResourceDeserialize";
+  
+  private BoostedTreesQuantileStreamResourceDeserialize(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesQuantileStreamResourceFlush.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesQuantileStreamResourceFlush.java
new file mode 100644
index 00000000000..cc10434a582
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesQuantileStreamResourceFlush.java
@@ -0,0 +1,103 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.estimator;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt64;
+
+/**
+ * Flush the summaries for a quantile stream resource.
+ * <p>
+ * An op that flushes the summaries for a quantile stream resource.
+ */
+public final class BoostedTreesQuantileStreamResourceFlush extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.estimator.BoostedTreesQuantileStreamResourceFlush}
+   */
+  public static class Options {
+    
+    /**
+     * @param generateQuantiles bool; If True, the output will be the num_quantiles for each stream where the ith
+     * entry is the ith quantile of the input with an approximation error of epsilon.
+     * Duplicate values may be present.
+     * If False, the output will be the points in the histogram that we got which roughly
+     * translates to 1/epsilon boundaries and without any duplicates.
+     * Default to False.
+     */
+    public Options generateQuantiles(Boolean generateQuantiles) {
+      this.generateQuantiles = generateQuantiles;
+      return this;
+    }
+    
+    private Boolean generateQuantiles;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new BoostedTreesQuantileStreamResourceFlush operation.
+   * 
+   * @param scope current scope
+   * @param quantileStreamResourceHandle resource handle referring to a QuantileStreamResource.
+   * @param numBuckets int; approximate number of buckets unless using generate_quantiles.
+   * @param options carries optional attributes values
+   * @return a new instance of BoostedTreesQuantileStreamResourceFlush
+   */
+  @Endpoint(describeByClass = true)
+  public static BoostedTreesQuantileStreamResourceFlush create(Scope scope, Operand<?> quantileStreamResourceHandle, Operand<TInt64> numBuckets, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("BoostedTreesQuantileStreamResourceFlush", scope.makeOpName("BoostedTreesQuantileStreamResourceFlush"));
+    opBuilder.addInput(quantileStreamResourceHandle.asOutput());
+    opBuilder.addInput(numBuckets.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.generateQuantiles != null) {
+          opBuilder.setAttr("generate_quantiles", opts.generateQuantiles);
+        }
+      }
+    }
+    return new BoostedTreesQuantileStreamResourceFlush(opBuilder.build());
+  }
+  
+  /**
+   * @param generateQuantiles bool; If True, the output will be the num_quantiles for each stream where the ith
+   * entry is the ith quantile of the input with an approximation error of epsilon.
+   * Duplicate values may be present.
+   * If False, the output will be the points in the histogram that we got which roughly
+   * translates to 1/epsilon boundaries and without any duplicates.
+   * Default to False.
+   */
+  public static Options generateQuantiles(Boolean generateQuantiles) {
+    return new Options().generateQuantiles(generateQuantiles);
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "BoostedTreesQuantileStreamResourceFlush";
+  
+  private BoostedTreesQuantileStreamResourceFlush(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesQuantileStreamResourceGetBucketBoundaries.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesQuantileStreamResourceGetBucketBoundaries.java
new file mode 100644
index 00000000000..5d421152254
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesQuantileStreamResourceGetBucketBoundaries.java
@@ -0,0 +1,84 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.estimator;
+
+import java.util.Arrays;
+import java.util.Iterator;
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+
+/**
+ * Generate the bucket boundaries for each feature based on accumulated summaries.
+ * <p>
+ * An op that returns a list of float tensors for a quantile stream resource. Each
+ * tensor is Rank 1 containing bucket boundaries for a single feature.
+ */
+public final class BoostedTreesQuantileStreamResourceGetBucketBoundaries extends RawOp implements Iterable<Operand<TFloat32>> {
+  
+  /**
+   * Factory method to create a class wrapping a new BoostedTreesQuantileStreamResourceGetBucketBoundaries operation.
+   * 
+   * @param scope current scope
+   * @param quantileStreamResourceHandle resource handle referring to a QuantileStreamResource.
+   * @param numFeatures inferred int; number of features to get bucket boundaries for.
+   * @return a new instance of BoostedTreesQuantileStreamResourceGetBucketBoundaries
+   */
+  @Endpoint(describeByClass = true)
+  public static BoostedTreesQuantileStreamResourceGetBucketBoundaries create(Scope scope, Operand<?> quantileStreamResourceHandle, Long numFeatures) {
+    OperationBuilder opBuilder = scope.env().opBuilder("BoostedTreesQuantileStreamResourceGetBucketBoundaries", scope.makeOpName("BoostedTreesQuantileStreamResourceGetBucketBoundaries"));
+    opBuilder.addInput(quantileStreamResourceHandle.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("num_features", numFeatures);
+    return new BoostedTreesQuantileStreamResourceGetBucketBoundaries(opBuilder.build());
+  }
+  
+  /**
+   * float; List of Rank 1 Tensors each containing the bucket boundaries for a feature.
+   */
+  public List<Output<TFloat32>> bucketBoundaries() {
+    return bucketBoundaries;
+  }
+  
+  @Override
+  @SuppressWarnings({"rawtypes", "unchecked"})
+  public Iterator<Operand<TFloat32>> iterator() {
+    return (Iterator) bucketBoundaries.iterator();
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "BoostedTreesQuantileStreamResourceGetBucketBoundaries";
+  
+  private List<Output<TFloat32>> bucketBoundaries;
+  
+  @SuppressWarnings("unchecked")
+  private BoostedTreesQuantileStreamResourceGetBucketBoundaries(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    int bucketBoundariesLength = operation.outputListLength("bucket_boundaries");
+    bucketBoundaries = Arrays.asList((Output<TFloat32>[])operation.outputList(outputIdx, bucketBoundariesLength));
+    outputIdx += bucketBoundariesLength;
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesQuantileStreamResourceHandleOp.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesQuantileStreamResourceHandleOp.java
new file mode 100644
index 00000000000..db925d767f9
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesQuantileStreamResourceHandleOp.java
@@ -0,0 +1,123 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.estimator;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Creates a handle to a BoostedTreesQuantileStreamResource.
+ */
+public final class BoostedTreesQuantileStreamResourceHandleOp extends RawOp implements Operand<TType> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.estimator.BoostedTreesQuantileStreamResourceHandleOp}
+   */
+  public static class Options {
+    
+    /**
+     * @param container 
+     */
+    public Options container(String container) {
+      this.container = container;
+      return this;
+    }
+    
+    /**
+     * @param sharedName 
+     */
+    public Options sharedName(String sharedName) {
+      this.sharedName = sharedName;
+      return this;
+    }
+    
+    private String container;
+    private String sharedName;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new BoostedTreesQuantileStreamResourceHandleOp operation.
+   * 
+   * @param scope current scope
+   * @param options carries optional attributes values
+   * @return a new instance of BoostedTreesQuantileStreamResourceHandleOp
+   */
+  @Endpoint(describeByClass = true)
+  public static BoostedTreesQuantileStreamResourceHandleOp create(Scope scope, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("BoostedTreesQuantileStreamResourceHandleOp", scope.makeOpName("BoostedTreesQuantileStreamResourceHandleOp"));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.container != null) {
+          opBuilder.setAttr("container", opts.container);
+        }
+        if (opts.sharedName != null) {
+          opBuilder.setAttr("shared_name", opts.sharedName);
+        }
+      }
+    }
+    return new BoostedTreesQuantileStreamResourceHandleOp(opBuilder.build());
+  }
+  
+  /**
+   * @param container 
+   */
+  public static Options container(String container) {
+    return new Options().container(container);
+  }
+  
+  /**
+   * @param sharedName 
+   */
+  public static Options sharedName(String sharedName) {
+    return new Options().sharedName(sharedName);
+  }
+  
+  /**
+   */
+  public Output<?> resource() {
+    return resource;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) resource;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "BoostedTreesQuantileStreamResourceHandleOp";
+  
+  private Output<?> resource;
+  
+  private BoostedTreesQuantileStreamResourceHandleOp(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    resource = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesSerializeEnsemble.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesSerializeEnsemble.java
new file mode 100644
index 00000000000..287c9e9054a
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesSerializeEnsemble.java
@@ -0,0 +1,77 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.estimator;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.TString;
+
+/**
+ * Serializes the tree ensemble to a proto.
+ */
+public final class BoostedTreesSerializeEnsemble extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new BoostedTreesSerializeEnsemble operation.
+   * 
+   * @param scope current scope
+   * @param treeEnsembleHandle Handle to the tree ensemble.
+   * @return a new instance of BoostedTreesSerializeEnsemble
+   */
+  @Endpoint(describeByClass = true)
+  public static BoostedTreesSerializeEnsemble create(Scope scope, Operand<?> treeEnsembleHandle) {
+    OperationBuilder opBuilder = scope.env().opBuilder("BoostedTreesSerializeEnsemble", scope.makeOpName("BoostedTreesSerializeEnsemble"));
+    opBuilder.addInput(treeEnsembleHandle.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new BoostedTreesSerializeEnsemble(opBuilder.build());
+  }
+  
+  /**
+   * Stamp token of the tree ensemble resource.
+   */
+  public Output<TInt64> stampToken() {
+    return stampToken;
+  }
+  
+  /**
+   * Serialized proto of the ensemble.
+   */
+  public Output<TString> treeEnsembleSerialized() {
+    return treeEnsembleSerialized;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "BoostedTreesSerializeEnsemble";
+  
+  private Output<TInt64> stampToken;
+  private Output<TString> treeEnsembleSerialized;
+  
+  private BoostedTreesSerializeEnsemble(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    stampToken = operation.output(outputIdx++);
+    treeEnsembleSerialized = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesSparseAggregateStats.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesSparseAggregateStats.java
new file mode 100644
index 00000000000..3923ac0e905
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesSparseAggregateStats.java
@@ -0,0 +1,115 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.estimator;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.TInt32;
+
+/**
+ * Aggregates the summary of accumulated stats for the batch.
+ * <p>
+ * The summary stats contains gradients and hessians accumulated for each node, bucket and dimension id.
+ */
+public final class BoostedTreesSparseAggregateStats extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new BoostedTreesSparseAggregateStats operation.
+   * 
+   * @param scope current scope
+   * @param nodeIds int32; Rank 1 Tensor containing node ids for each example, shape [batch_size].
+   * @param gradients float32; Rank 2 Tensor (shape=[batch_size, logits_dimension]) with gradients for each example.
+   * @param hessians float32; Rank 2 Tensor (shape=[batch_size, hessian_dimension]) with hessians for each example.
+   * @param featureIndices int32; Rank 2 indices of feature sparse Tensors (shape=[number of sparse entries, 2]).
+   * Number of sparse entries across all instances from the batch. The first value is
+   * the index of the instance, the second is dimension of the feature. The second axis
+   * can only have 2 values, i.e., the input dense version of Tensor can only be matrix.
+   * @param featureValues int32; Rank 1 values of feature sparse Tensors (shape=[number of sparse entries]).
+   * Number of sparse entries across all instances from the batch. The first value is
+   * the index of the instance, the second is dimension of the feature.
+   * @param featureShape int32; Rank 1 dense shape of feature sparse Tensors (shape=[2]).
+   * The first axis can only have 2 values, [batch_size, feature_dimension].
+   * @param maxSplits int; the maximum number of splits possible in the whole tree.
+   * @param numBuckets int; equals to the maximum possible value of bucketized feature + 1.
+   * @return a new instance of BoostedTreesSparseAggregateStats
+   */
+  @Endpoint(describeByClass = true)
+  public static BoostedTreesSparseAggregateStats create(Scope scope, Operand<TInt32> nodeIds, Operand<TFloat32> gradients, Operand<TFloat32> hessians, Operand<TInt32> featureIndices, Operand<TInt32> featureValues, Operand<TInt32> featureShape, Long maxSplits, Long numBuckets) {
+    OperationBuilder opBuilder = scope.env().opBuilder("BoostedTreesSparseAggregateStats", scope.makeOpName("BoostedTreesSparseAggregateStats"));
+    opBuilder.addInput(nodeIds.asOutput());
+    opBuilder.addInput(gradients.asOutput());
+    opBuilder.addInput(hessians.asOutput());
+    opBuilder.addInput(featureIndices.asOutput());
+    opBuilder.addInput(featureValues.asOutput());
+    opBuilder.addInput(featureShape.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("max_splits", maxSplits);
+    opBuilder.setAttr("num_buckets", numBuckets);
+    return new BoostedTreesSparseAggregateStats(opBuilder.build());
+  }
+  
+  /**
+   * int32; Rank 2 indices of summary sparse Tensors (shape=[number of non zero statistics, 4])
+   * The second axis can only be 4 including node id, feature dimension, bucket id, and statistics_dimension.
+   * statistics_dimension = logits_dimension + hessian_dimension.
+   */
+  public Output<TInt32> statsSummaryIndices() {
+    return statsSummaryIndices;
+  }
+  
+  /**
+   * output Rank 1 Tensor (shape=[number of non zero statistics])
+   */
+  public Output<TFloat32> statsSummaryValues() {
+    return statsSummaryValues;
+  }
+  
+  /**
+   * output Rank 1 Tensor (shape=[4])
+   * The tensor has following 4 values: [max_splits, feature_dimension, num_buckets, statistics_dimension],
+   * where statistics_dimension = gradient_dimension + hessian_dimension. gradient_dimension
+   * is the same as label_dimension, i.e., the output space. hessian_dimension can be the same
+   * as logits dimension when diagonal hessian is used, or label_dimension^2 when full
+   * hessian is used.
+   */
+  public Output<TInt32> statsSummaryShape() {
+    return statsSummaryShape;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "BoostedTreesSparseAggregateStats";
+  
+  private Output<TInt32> statsSummaryIndices;
+  private Output<TFloat32> statsSummaryValues;
+  private Output<TInt32> statsSummaryShape;
+  
+  private BoostedTreesSparseAggregateStats(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    statsSummaryIndices = operation.output(outputIdx++);
+    statsSummaryValues = operation.output(outputIdx++);
+    statsSummaryShape = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesSparseCalculateBestFeatureSplit.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesSparseCalculateBestFeatureSplit.java
new file mode 100644
index 00000000000..b2f2dd1eebf
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesSparseCalculateBestFeatureSplit.java
@@ -0,0 +1,184 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.estimator;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.TString;
+
+/**
+ * Calculates gains for each feature and returns the best possible split information for the feature.
+ * <p>
+ * The split information is the best threshold (bucket id), gains and left/right node contributions per node for each feature.
+ * <p>
+ * It is possible that not all nodes can be split on each feature. Hence, the list of possible nodes can differ between the features. Therefore, we return `node_ids_list` for each feature, containing the list of nodes that this feature can be used to split.
+ * <p>
+ * In this manner, the output is the best split per features and per node, so that it needs to be combined later to produce the best split for each node (among all possible features).
+ * <p>
+ * The output shapes are compatible in a way that the first dimension of all tensors are the same and equal to the number of possible split nodes for each feature.
+ */
+public final class BoostedTreesSparseCalculateBestFeatureSplit extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.estimator.BoostedTreesSparseCalculateBestFeatureSplit}
+   */
+  public static class Options {
+    
+    /**
+     * @param splitType A string indicating if this Op should perform inequality split or equality split.
+     */
+    public Options splitType(String splitType) {
+      this.splitType = splitType;
+      return this;
+    }
+    
+    private String splitType;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new BoostedTreesSparseCalculateBestFeatureSplit operation.
+   * 
+   * @param scope current scope
+   * @param nodeIdRange A Rank 1 tensor (shape=[2]) to specify the range [first, last) of node ids to process within `stats_summary_list`. The nodes are iterated between the two nodes specified by the tensor, as like `for node_id in range(node_id_range[0], node_id_range[1])` (Note that the last index node_id_range[1] is exclusive).
+   * @param statsSummaryIndices A Rank 2 int64 tensor of dense shape [N, 4] (N specifies the number of non-zero values) for accumulated stats summary (gradient/hessian) per node per bucket for each feature. The second dimension contains node id, feature dimension, bucket id, and stats dim.
+   * stats dim is the sum of logits dimension and hessian dimension, hessian dimension can either be logits dimension if diagonal hessian is used, or logits dimension^2 if full hessian is used.
+   * @param statsSummaryValues A Rank 1 float tensor of dense shape [N] (N specifies the number of non-zero values), which supplies the values for each element in summary_indices.
+   * @param statsSummaryShape A Rank 1 float tensor of dense shape [4], which specifies the dense shape of the sparse tensor, which is [num tree nodes, feature dimensions, num buckets, stats dim].
+   * @param l1 l1 regularization factor on leaf weights, per instance based.
+   * @param l2 l2 regularization factor on leaf weights, per instance based.
+   * @param treeComplexity adjustment to the gain, per leaf based.
+   * @param minNodeWeight minimum avg of hessians in a node before required for the node to be considered for splitting.
+   * @param logitsDimension The dimension of logit, i.e., number of classes.
+   * @param options carries optional attributes values
+   * @return a new instance of BoostedTreesSparseCalculateBestFeatureSplit
+   */
+  @Endpoint(describeByClass = true)
+  public static BoostedTreesSparseCalculateBestFeatureSplit create(Scope scope, Operand<TInt32> nodeIdRange, Operand<TInt32> statsSummaryIndices, Operand<TFloat32> statsSummaryValues, Operand<TInt32> statsSummaryShape, Operand<TFloat32> l1, Operand<TFloat32> l2, Operand<TFloat32> treeComplexity, Operand<TFloat32> minNodeWeight, Long logitsDimension, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("BoostedTreesSparseCalculateBestFeatureSplit", scope.makeOpName("BoostedTreesSparseCalculateBestFeatureSplit"));
+    opBuilder.addInput(nodeIdRange.asOutput());
+    opBuilder.addInput(statsSummaryIndices.asOutput());
+    opBuilder.addInput(statsSummaryValues.asOutput());
+    opBuilder.addInput(statsSummaryShape.asOutput());
+    opBuilder.addInput(l1.asOutput());
+    opBuilder.addInput(l2.asOutput());
+    opBuilder.addInput(treeComplexity.asOutput());
+    opBuilder.addInput(minNodeWeight.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("logits_dimension", logitsDimension);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.splitType != null) {
+          opBuilder.setAttr("split_type", opts.splitType);
+        }
+      }
+    }
+    return new BoostedTreesSparseCalculateBestFeatureSplit(opBuilder.build());
+  }
+  
+  /**
+   * @param splitType A string indicating if this Op should perform inequality split or equality split.
+   */
+  public static Options splitType(String splitType) {
+    return new Options().splitType(splitType);
+  }
+  
+  /**
+   * A Rank 1 tensor indicating possible node ids that can be split.
+   */
+  public Output<TInt32> nodeIds() {
+    return nodeIds;
+  }
+  
+  /**
+   * A Rank 1 tensor indicating the best gains to split each node.
+   */
+  public Output<TFloat32> gains() {
+    return gains;
+  }
+  
+  /**
+   * A Rank 1 tensor indicating the best feature dimension for each feature to split for each node.
+   */
+  public Output<TInt32> featureDimensions() {
+    return featureDimensions;
+  }
+  
+  /**
+   * A Rank 1 tensor indicating the bucket id to compare with (as a threshold) for split in each node.
+   */
+  public Output<TInt32> thresholds() {
+    return thresholds;
+  }
+  
+  /**
+   * A Rank 2 tensor indicating the contribution of the left nodes when branching from parent nodes to the left direction by the given threshold for each feature.
+   * This value will be used to make the left node value by adding to the parent node value. Second dimension size is logits dimension.
+   */
+  public Output<TFloat32> leftNodeContribs() {
+    return leftNodeContribs;
+  }
+  
+  /**
+   * A Rank 2 tensor, with the same shape/conditions as left_node_contribs_list, but just that the value is for the right node.
+   */
+  public Output<TFloat32> rightNodeContribs() {
+    return rightNodeContribs;
+  }
+  
+  /**
+   * A Rank 1 tensor indicating which direction to go if data is missing.
+   * Inequality with default left returns 0, inequality with default right returns 1, equality with default right returns 2.
+   */
+  public Output<TString> splitWithDefaultDirections() {
+    return splitWithDefaultDirections;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "BoostedTreesSparseCalculateBestFeatureSplit";
+  
+  private Output<TInt32> nodeIds;
+  private Output<TFloat32> gains;
+  private Output<TInt32> featureDimensions;
+  private Output<TInt32> thresholds;
+  private Output<TFloat32> leftNodeContribs;
+  private Output<TFloat32> rightNodeContribs;
+  private Output<TString> splitWithDefaultDirections;
+  
+  private BoostedTreesSparseCalculateBestFeatureSplit(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    nodeIds = operation.output(outputIdx++);
+    gains = operation.output(outputIdx++);
+    featureDimensions = operation.output(outputIdx++);
+    thresholds = operation.output(outputIdx++);
+    leftNodeContribs = operation.output(outputIdx++);
+    rightNodeContribs = operation.output(outputIdx++);
+    splitWithDefaultDirections = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesTrainingPredict.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesTrainingPredict.java
new file mode 100644
index 00000000000..a81509950b9
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesTrainingPredict.java
@@ -0,0 +1,104 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.estimator;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.Operands;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.TInt32;
+
+/**
+ * Runs multiple additive regression ensemble predictors on input instances and
+ * <p>
+ * computes the update to cached logits. It is designed to be used during training.
+ * It traverses the trees starting from cached tree id and cached node id and
+ * calculates the updates to be pushed to the cache.
+ */
+public final class BoostedTreesTrainingPredict extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new BoostedTreesTrainingPredict operation.
+   * 
+   * @param scope current scope
+   * @param treeEnsembleHandle 
+   * @param cachedTreeIds Rank 1 Tensor containing cached tree ids which is the starting
+   * tree of prediction.
+   * @param cachedNodeIds Rank 1 Tensor containing cached node id which is the starting
+   * node of prediction.
+   * @param bucketizedFeatures A list of rank 1 Tensors containing bucket id for each
+   * feature.
+   * @param logitsDimension scalar, dimension of the logits, to be used for partial logits
+   * shape.
+   * @return a new instance of BoostedTreesTrainingPredict
+   */
+  @Endpoint(describeByClass = true)
+  public static BoostedTreesTrainingPredict create(Scope scope, Operand<?> treeEnsembleHandle, Operand<TInt32> cachedTreeIds, Operand<TInt32> cachedNodeIds, Iterable<Operand<TInt32>> bucketizedFeatures, Long logitsDimension) {
+    OperationBuilder opBuilder = scope.env().opBuilder("BoostedTreesTrainingPredict", scope.makeOpName("BoostedTreesTrainingPredict"));
+    opBuilder.addInput(treeEnsembleHandle.asOutput());
+    opBuilder.addInput(cachedTreeIds.asOutput());
+    opBuilder.addInput(cachedNodeIds.asOutput());
+    opBuilder.addInputList(Operands.asOutputs(bucketizedFeatures));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("logits_dimension", logitsDimension);
+    return new BoostedTreesTrainingPredict(opBuilder.build());
+  }
+  
+  /**
+   * Rank 2 Tensor containing logits update (with respect to cached
+   * values stored) for each example.
+   */
+  public Output<TFloat32> partialLogits() {
+    return partialLogits;
+  }
+  
+  /**
+   * Rank 1 Tensor containing new tree ids for each example.
+   */
+  public Output<TInt32> treeIds() {
+    return treeIds;
+  }
+  
+  /**
+   * Rank 1 Tensor containing new node ids in the new tree_ids.
+   */
+  public Output<TInt32> nodeIds() {
+    return nodeIds;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "BoostedTreesTrainingPredict";
+  
+  private Output<TFloat32> partialLogits;
+  private Output<TInt32> treeIds;
+  private Output<TInt32> nodeIds;
+  
+  private BoostedTreesTrainingPredict(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    partialLogits = operation.output(outputIdx++);
+    treeIds = operation.output(outputIdx++);
+    nodeIds = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesUpdateEnsemble.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesUpdateEnsemble.java
new file mode 100644
index 00000000000..e6ddcf3d2da
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesUpdateEnsemble.java
@@ -0,0 +1,85 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.estimator;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.Operands;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.TInt32;
+
+/**
+ * Updates the tree ensemble by either adding a layer to the last tree being grown
+ * <p>
+ * or by starting a new tree.
+ */
+public final class BoostedTreesUpdateEnsemble extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new BoostedTreesUpdateEnsemble operation.
+   * 
+   * @param scope current scope
+   * @param treeEnsembleHandle Handle to the ensemble variable.
+   * @param featureIds Rank 1 tensor with ids for each feature. This is the real id of
+   * the feature that will be used in the split.
+   * @param nodeIds List of rank 1 tensors representing the nodes for which this feature
+   * has a split.
+   * @param gains List of rank 1 tensors representing the gains for each of the feature's
+   * split.
+   * @param thresholds List of rank 1 tensors representing the thesholds for each of the
+   * feature's split.
+   * @param leftNodeContribs List of rank 2 tensors with left leaf contribs for each of
+   * the feature's splits. Will be added to the previous node values to constitute
+   * the values of the left nodes.
+   * @param rightNodeContribs List of rank 2 tensors with right leaf contribs for each
+   * of the feature's splits. Will be added to the previous node values to constitute
+   * the values of the right nodes.
+   * @param maxDepth Max depth of the tree to build.
+   * @param learningRate shrinkage const for each new tree.
+   * @param pruningMode 0-No pruning, 1-Pre-pruning, 2-Post-pruning.
+   * @return a new instance of BoostedTreesUpdateEnsemble
+   */
+  @Endpoint(describeByClass = true)
+  public static BoostedTreesUpdateEnsemble create(Scope scope, Operand<?> treeEnsembleHandle, Operand<TInt32> featureIds, Iterable<Operand<TInt32>> nodeIds, Iterable<Operand<TFloat32>> gains, Iterable<Operand<TInt32>> thresholds, Iterable<Operand<TFloat32>> leftNodeContribs, Iterable<Operand<TFloat32>> rightNodeContribs, Operand<TInt32> maxDepth, Operand<TFloat32> learningRate, Long pruningMode) {
+    OperationBuilder opBuilder = scope.env().opBuilder("BoostedTreesUpdateEnsemble", scope.makeOpName("BoostedTreesUpdateEnsemble"));
+    opBuilder.addInput(treeEnsembleHandle.asOutput());
+    opBuilder.addInput(featureIds.asOutput());
+    opBuilder.addInputList(Operands.asOutputs(nodeIds));
+    opBuilder.addInputList(Operands.asOutputs(gains));
+    opBuilder.addInputList(Operands.asOutputs(thresholds));
+    opBuilder.addInputList(Operands.asOutputs(leftNodeContribs));
+    opBuilder.addInputList(Operands.asOutputs(rightNodeContribs));
+    opBuilder.addInput(maxDepth.asOutput());
+    opBuilder.addInput(learningRate.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("pruning_mode", pruningMode);
+    return new BoostedTreesUpdateEnsemble(opBuilder.build());
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "BoostedTreesUpdateEnsemble";
+  
+  private BoostedTreesUpdateEnsemble(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesUpdateEnsembleV2.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesUpdateEnsembleV2.java
new file mode 100644
index 00000000000..ceaff116fd1
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/BoostedTreesUpdateEnsembleV2.java
@@ -0,0 +1,124 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.estimator;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.Operands;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.TString;
+
+/**
+ * Updates the tree ensemble by adding a layer to the last tree being grown
+ * <p>
+ * or by starting a new tree.
+ */
+public final class BoostedTreesUpdateEnsembleV2 extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.estimator.BoostedTreesUpdateEnsembleV2}
+   */
+  public static class Options {
+    
+    /**
+     * @param logitsDimension scalar, dimension of the logits
+     */
+    public Options logitsDimension(Long logitsDimension) {
+      this.logitsDimension = logitsDimension;
+      return this;
+    }
+    
+    private Long logitsDimension;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new BoostedTreesUpdateEnsembleV2 operation.
+   * 
+   * @param scope current scope
+   * @param treeEnsembleHandle Handle to the ensemble variable.
+   * @param featureIds Rank 1 tensor with ids for each feature. This is the real id of
+   * the feature that will be used in the split.
+   * @param dimensionIds List of rank 1 tensors representing the dimension in each feature.
+   * @param nodeIds List of rank 1 tensors representing the nodes for which this feature
+   * has a split.
+   * @param gains List of rank 1 tensors representing the gains for each of the feature's
+   * split.
+   * @param thresholds List of rank 1 tensors representing the thesholds for each of the
+   * feature's split.
+   * @param leftNodeContribs List of rank 2 tensors with left leaf contribs for each of
+   * the feature's splits. Will be added to the previous node values to constitute
+   * the values of the left nodes.
+   * @param rightNodeContribs List of rank 2 tensors with right leaf contribs for each
+   * of the feature's splits. Will be added to the previous node values to constitute
+   * the values of the right nodes.
+   * @param splitTypes List of rank 1 tensors representing the split type for each feature.
+   * @param maxDepth Max depth of the tree to build.
+   * @param learningRate shrinkage const for each new tree.
+   * @param pruningMode 0-No pruning, 1-Pre-pruning, 2-Post-pruning.
+   * @param options carries optional attributes values
+   * @return a new instance of BoostedTreesUpdateEnsembleV2
+   */
+  @Endpoint(describeByClass = true)
+  public static BoostedTreesUpdateEnsembleV2 create(Scope scope, Operand<?> treeEnsembleHandle, Iterable<Operand<TInt32>> featureIds, Iterable<Operand<TInt32>> dimensionIds, Iterable<Operand<TInt32>> nodeIds, Iterable<Operand<TFloat32>> gains, Iterable<Operand<TInt32>> thresholds, Iterable<Operand<TFloat32>> leftNodeContribs, Iterable<Operand<TFloat32>> rightNodeContribs, Iterable<Operand<TString>> splitTypes, Operand<TInt32> maxDepth, Operand<TFloat32> learningRate, Operand<TInt32> pruningMode, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("BoostedTreesUpdateEnsembleV2", scope.makeOpName("BoostedTreesUpdateEnsembleV2"));
+    opBuilder.addInput(treeEnsembleHandle.asOutput());
+    opBuilder.addInputList(Operands.asOutputs(featureIds));
+    opBuilder.addInputList(Operands.asOutputs(dimensionIds));
+    opBuilder.addInputList(Operands.asOutputs(nodeIds));
+    opBuilder.addInputList(Operands.asOutputs(gains));
+    opBuilder.addInputList(Operands.asOutputs(thresholds));
+    opBuilder.addInputList(Operands.asOutputs(leftNodeContribs));
+    opBuilder.addInputList(Operands.asOutputs(rightNodeContribs));
+    opBuilder.addInputList(Operands.asOutputs(splitTypes));
+    opBuilder.addInput(maxDepth.asOutput());
+    opBuilder.addInput(learningRate.asOutput());
+    opBuilder.addInput(pruningMode.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.logitsDimension != null) {
+          opBuilder.setAttr("logits_dimension", opts.logitsDimension);
+        }
+      }
+    }
+    return new BoostedTreesUpdateEnsembleV2(opBuilder.build());
+  }
+  
+  /**
+   * @param logitsDimension scalar, dimension of the logits
+   */
+  public static Options logitsDimension(Long logitsDimension) {
+    return new Options().logitsDimension(logitsDimension);
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "BoostedTreesUpdateEnsembleV2";
+  
+  private BoostedTreesUpdateEnsembleV2(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/IsBoostedTreesEnsembleInitialized.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/IsBoostedTreesEnsembleInitialized.java
new file mode 100644
index 00000000000..7bf185408df
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/IsBoostedTreesEnsembleInitialized.java
@@ -0,0 +1,72 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.estimator;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TBool;
+
+/**
+ * Checks whether a tree ensemble has been initialized.
+ */
+public final class IsBoostedTreesEnsembleInitialized extends RawOp implements Operand<TBool> {
+  
+  /**
+   * Factory method to create a class wrapping a new IsBoostedTreesEnsembleInitialized operation.
+   * 
+   * @param scope current scope
+   * @param treeEnsembleHandle Handle to the tree ensemble resource.
+   * @return a new instance of IsBoostedTreesEnsembleInitialized
+   */
+  @Endpoint(describeByClass = true)
+  public static IsBoostedTreesEnsembleInitialized create(Scope scope, Operand<?> treeEnsembleHandle) {
+    OperationBuilder opBuilder = scope.env().opBuilder("IsBoostedTreesEnsembleInitialized", scope.makeOpName("IsBoostedTreesEnsembleInitialized"));
+    opBuilder.addInput(treeEnsembleHandle.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new IsBoostedTreesEnsembleInitialized(opBuilder.build());
+  }
+  
+  /**
+   * output boolean on whether it is initialized or not.
+   */
+  public Output<TBool> isInitialized() {
+    return isInitialized;
+  }
+  
+  @Override
+  public Output<TBool> asOutput() {
+    return isInitialized;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "IsBoostedTreesEnsembleInitialized";
+  
+  private Output<TBool> isInitialized;
+  
+  private IsBoostedTreesEnsembleInitialized(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    isInitialized = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/IsBoostedTreesQuantileStreamResourceInitialized.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/IsBoostedTreesQuantileStreamResourceInitialized.java
new file mode 100644
index 00000000000..66ad4435dfd
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/estimator/IsBoostedTreesQuantileStreamResourceInitialized.java
@@ -0,0 +1,74 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.estimator;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TBool;
+
+/**
+ * Checks whether a quantile stream has been initialized.
+ * <p>
+ * An Op that checks if quantile stream resource is initialized.
+ */
+public final class IsBoostedTreesQuantileStreamResourceInitialized extends RawOp implements Operand<TBool> {
+  
+  /**
+   * Factory method to create a class wrapping a new IsBoostedTreesQuantileStreamResourceInitialized operation.
+   * 
+   * @param scope current scope
+   * @param quantileStreamResourceHandle resource; The reference to quantile stream resource handle.
+   * @return a new instance of IsBoostedTreesQuantileStreamResourceInitialized
+   */
+  @Endpoint(describeByClass = true)
+  public static IsBoostedTreesQuantileStreamResourceInitialized create(Scope scope, Operand<?> quantileStreamResourceHandle) {
+    OperationBuilder opBuilder = scope.env().opBuilder("IsBoostedTreesQuantileStreamResourceInitialized", scope.makeOpName("IsBoostedTreesQuantileStreamResourceInitialized"));
+    opBuilder.addInput(quantileStreamResourceHandle.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new IsBoostedTreesQuantileStreamResourceInitialized(opBuilder.build());
+  }
+  
+  /**
+   * bool; True if the resource is initialized, False otherwise.
+   */
+  public Output<TBool> isInitialized() {
+    return isInitialized;
+  }
+  
+  @Override
+  public Output<TBool> asOutput() {
+    return isInitialized;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "IsBoostedTreesQuantileStreamResourceInitialized";
+  
+  private Output<TBool> isInitialized;
+  
+  private IsBoostedTreesQuantileStreamResourceInitialized(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    isInitialized = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/AdjustContrast.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/AdjustContrast.java
new file mode 100644
index 00000000000..b33b91ed953
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/AdjustContrast.java
@@ -0,0 +1,88 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.image;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Adjust the contrast of one or more images.
+ * <p>
+ * `images` is a tensor of at least 3 dimensions.  The last 3 dimensions are
+ * interpreted as `[height, width, channels]`.  The other dimensions only
+ * represent a collection of images, such as `[batch, height, width, channels].`
+ * <p>
+ * Contrast is adjusted independently for each channel of each image.
+ * <p>
+ * For each channel, the Op first computes the mean of the image pixels in the
+ * channel and then adjusts each component of each pixel to
+ * `(x - mean) * contrast_factor + mean`.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator(group = "image")
+public final class AdjustContrast<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new AdjustContrast operation.
+   * 
+   * @param scope current scope
+   * @param images Images to adjust.  At least 3-D.
+   * @param contrastFactor A float multiplier for adjusting contrast.
+   * @return a new instance of AdjustContrast
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> AdjustContrast<T> create(Scope scope, Operand<T> images, Operand<TFloat32> contrastFactor) {
+    OperationBuilder opBuilder = scope.env().opBuilder("AdjustContrastv2", scope.makeOpName("AdjustContrast"));
+    opBuilder.addInput(images.asOutput());
+    opBuilder.addInput(contrastFactor.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new AdjustContrast<T>(opBuilder.build());
+  }
+  
+  /**
+   * The contrast-adjusted image or images.
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "AdjustContrastv2";
+  
+  private Output<T> output;
+  
+  private AdjustContrast(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/AdjustHue.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/AdjustHue.java
new file mode 100644
index 00000000000..7f1c320f052
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/AdjustHue.java
@@ -0,0 +1,85 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.image;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Adjust the hue of one or more images.
+ * <p>
+ * `images` is a tensor of at least 3 dimensions.  The last dimension is
+ * interpreted as channels, and must be three.
+ * <p>
+ * The input image is considered in the RGB colorspace. Conceptually, the RGB
+ * colors are first mapped into HSV. A delta is then applied all the hue values,
+ * and then remapped back to RGB colorspace.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator(group = "image")
+public final class AdjustHue<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new AdjustHue operation.
+   * 
+   * @param scope current scope
+   * @param images Images to adjust.  At least 3-D.
+   * @param delta A float delta to add to the hue.
+   * @return a new instance of AdjustHue
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> AdjustHue<T> create(Scope scope, Operand<T> images, Operand<TFloat32> delta) {
+    OperationBuilder opBuilder = scope.env().opBuilder("AdjustHue", scope.makeOpName("AdjustHue"));
+    opBuilder.addInput(images.asOutput());
+    opBuilder.addInput(delta.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new AdjustHue<T>(opBuilder.build());
+  }
+  
+  /**
+   * The hue-adjusted image or images.
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "AdjustHue";
+  
+  private Output<T> output;
+  
+  private AdjustHue(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/AdjustSaturation.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/AdjustSaturation.java
new file mode 100644
index 00000000000..49a7f6ef6ae
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/AdjustSaturation.java
@@ -0,0 +1,85 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.image;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Adjust the saturation of one or more images.
+ * <p>
+ * `images` is a tensor of at least 3 dimensions.  The last dimension is
+ * interpreted as channels, and must be three.
+ * <p>
+ * The input image is considered in the RGB colorspace. Conceptually, the RGB
+ * colors are first mapped into HSV. A scale is then applied all the saturation
+ * values, and then remapped back to RGB colorspace.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator(group = "image")
+public final class AdjustSaturation<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new AdjustSaturation operation.
+   * 
+   * @param scope current scope
+   * @param images Images to adjust.  At least 3-D.
+   * @param scale A float scale to add to the saturation.
+   * @return a new instance of AdjustSaturation
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> AdjustSaturation<T> create(Scope scope, Operand<T> images, Operand<TFloat32> scale) {
+    OperationBuilder opBuilder = scope.env().opBuilder("AdjustSaturation", scope.makeOpName("AdjustSaturation"));
+    opBuilder.addInput(images.asOutput());
+    opBuilder.addInput(scale.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new AdjustSaturation<T>(opBuilder.build());
+  }
+  
+  /**
+   * The hue-adjusted image or images.
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "AdjustSaturation";
+  
+  private Output<T> output;
+  
+  private AdjustSaturation(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/CombinedNonMaxSuppression.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/CombinedNonMaxSuppression.java
new file mode 100644
index 00000000000..42bf54e7143
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/CombinedNonMaxSuppression.java
@@ -0,0 +1,197 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.image;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.TInt32;
+
+/**
+ * Greedily selects a subset of bounding boxes in descending order of score,
+ * <p>
+ * This operation performs non_max_suppression on the inputs per batch, across
+ * all classes.
+ * Prunes away boxes that have high intersection-over-union (IOU) overlap
+ * with previously selected boxes.  Bounding boxes are supplied as
+ * [y1, x1, y2, x2], where (y1, x1) and (y2, x2) are the coordinates of any
+ * diagonal pair of box corners and the coordinates can be provided as normalized
+ * (i.e., lying in the interval [0, 1]) or absolute.  Note that this algorithm
+ * is agnostic to where the origin is in the coordinate system. Also note that
+ * this algorithm is invariant to orthogonal transformations and translations
+ * of the coordinate system; thus translating or reflections of the coordinate
+ * system result in the same boxes being selected by the algorithm.
+ * The output of this operation is the final boxes, scores and classes tensor
+ * returned after performing non_max_suppression.
+ */
+@Operator(group = "image")
+public final class CombinedNonMaxSuppression extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.image.CombinedNonMaxSuppression}
+   */
+  public static class Options {
+    
+    /**
+     * @param padPerClass If false, the output nmsed boxes, scores and classes
+     * are padded/clipped to `max_total_size`. If true, the
+     * output nmsed boxes, scores and classes are padded to be of length
+     * `max_size_per_class`*`num_classes`, unless it exceeds `max_total_size` in
+     * which case it is clipped to `max_total_size`. Defaults to false.
+     */
+    public Options padPerClass(Boolean padPerClass) {
+      this.padPerClass = padPerClass;
+      return this;
+    }
+    
+    /**
+     * @param clipBoxes If true, assume the box coordinates are between [0, 1] and clip the output boxes
+     * if they fall beyond [0, 1]. If false, do not do clipping and output the box
+     * coordinates as it is.
+     */
+    public Options clipBoxes(Boolean clipBoxes) {
+      this.clipBoxes = clipBoxes;
+      return this;
+    }
+    
+    private Boolean padPerClass;
+    private Boolean clipBoxes;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new CombinedNonMaxSuppression operation.
+   * 
+   * @param scope current scope
+   * @param boxes A 4-D float tensor of shape `[batch_size, num_boxes, q, 4]`. If `q` is 1 then
+   * same boxes are used for all classes otherwise, if `q` is equal to number of
+   * classes, class-specific boxes are used.
+   * @param scores A 3-D float tensor of shape `[batch_size, num_boxes, num_classes]`
+   * representing a single score corresponding to each box (each row of boxes).
+   * @param maxOutputSizePerClass A scalar integer tensor representing the maximum number of
+   * boxes to be selected by non max suppression per class
+   * @param maxTotalSize A scalar representing maximum number of boxes retained over all classes.
+   * @param iouThreshold A 0-D float tensor representing the threshold for deciding whether
+   * boxes overlap too much with respect to IOU.
+   * @param scoreThreshold A 0-D float tensor representing the threshold for deciding when to remove
+   * boxes based on score.
+   * @param options carries optional attributes values
+   * @return a new instance of CombinedNonMaxSuppression
+   */
+  @Endpoint(describeByClass = true)
+  public static CombinedNonMaxSuppression create(Scope scope, Operand<TFloat32> boxes, Operand<TFloat32> scores, Operand<TInt32> maxOutputSizePerClass, Operand<TInt32> maxTotalSize, Operand<TFloat32> iouThreshold, Operand<TFloat32> scoreThreshold, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("CombinedNonMaxSuppression", scope.makeOpName("CombinedNonMaxSuppression"));
+    opBuilder.addInput(boxes.asOutput());
+    opBuilder.addInput(scores.asOutput());
+    opBuilder.addInput(maxOutputSizePerClass.asOutput());
+    opBuilder.addInput(maxTotalSize.asOutput());
+    opBuilder.addInput(iouThreshold.asOutput());
+    opBuilder.addInput(scoreThreshold.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.padPerClass != null) {
+          opBuilder.setAttr("pad_per_class", opts.padPerClass);
+        }
+        if (opts.clipBoxes != null) {
+          opBuilder.setAttr("clip_boxes", opts.clipBoxes);
+        }
+      }
+    }
+    return new CombinedNonMaxSuppression(opBuilder.build());
+  }
+  
+  /**
+   * @param padPerClass If false, the output nmsed boxes, scores and classes
+   * are padded/clipped to `max_total_size`. If true, the
+   * output nmsed boxes, scores and classes are padded to be of length
+   * `max_size_per_class`*`num_classes`, unless it exceeds `max_total_size` in
+   * which case it is clipped to `max_total_size`. Defaults to false.
+   */
+  public static Options padPerClass(Boolean padPerClass) {
+    return new Options().padPerClass(padPerClass);
+  }
+  
+  /**
+   * @param clipBoxes If true, assume the box coordinates are between [0, 1] and clip the output boxes
+   * if they fall beyond [0, 1]. If false, do not do clipping and output the box
+   * coordinates as it is.
+   */
+  public static Options clipBoxes(Boolean clipBoxes) {
+    return new Options().clipBoxes(clipBoxes);
+  }
+  
+  /**
+   * A [batch_size, max_detections, 4] float32 tensor
+   * containing the non-max suppressed boxes.
+   */
+  public Output<TFloat32> nmsedBoxes() {
+    return nmsedBoxes;
+  }
+  
+  /**
+   * A [batch_size, max_detections] float32 tensor
+   * containing the scores for the boxes.
+   */
+  public Output<TFloat32> nmsedScores() {
+    return nmsedScores;
+  }
+  
+  /**
+   * A [batch_size, max_detections] float32 tensor
+   * containing the classes for the boxes.
+   */
+  public Output<TFloat32> nmsedClasses() {
+    return nmsedClasses;
+  }
+  
+  /**
+   * A [batch_size] int32 tensor indicating the number of
+   * valid detections per batch item. Only the top num_detections[i] entries in
+   * nms_boxes[i], nms_scores[i] and nms_class[i] are valid. The rest of the
+   * entries are zero paddings.
+   */
+  public Output<TInt32> validDetections() {
+    return validDetections;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "CombinedNonMaxSuppression";
+  
+  private Output<TFloat32> nmsedBoxes;
+  private Output<TFloat32> nmsedScores;
+  private Output<TFloat32> nmsedClasses;
+  private Output<TInt32> validDetections;
+  
+  private CombinedNonMaxSuppression(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    nmsedBoxes = operation.output(outputIdx++);
+    nmsedScores = operation.output(outputIdx++);
+    nmsedClasses = operation.output(outputIdx++);
+    validDetections = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/CropAndResize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/CropAndResize.java
new file mode 100644
index 00000000000..58e9ed905d1
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/CropAndResize.java
@@ -0,0 +1,168 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.image;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Extracts crops from the input image tensor and resizes them.
+ * <p>
+ * Extracts crops from the input image tensor and resizes them using bilinear
+ * sampling or nearest neighbor sampling (possibly with aspect ratio change) to a
+ * common output size specified by `crop_size`. This is more general than the
+ * `crop_to_bounding_box` op which extracts a fixed size slice from the input image
+ * and does not allow resizing or aspect ratio change.
+ * <p>
+ * Returns a tensor with `crops` from the input `image` at positions defined at the
+ * bounding box locations in `boxes`. The cropped boxes are all resized (with
+ * bilinear or nearest neighbor interpolation) to a fixed
+ * `size = [crop_height, crop_width]`. The result is a 4-D tensor
+ * `[num_boxes, crop_height, crop_width, depth]`. The resizing is corner aligned.
+ * In particular, if `boxes = [[0, 0, 1, 1]]`, the method will give identical
+ * results to using `tf.image.resize_bilinear()` or
+ * `tf.image.resize_nearest_neighbor()`(depends on the `method` argument) with
+ * `align_corners=True`.
+ */
+@Operator(group = "image")
+public final class CropAndResize extends RawOp implements Operand<TFloat32> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.image.CropAndResize}
+   */
+  public static class Options {
+    
+    /**
+     * @param method A string specifying the sampling method for resizing. It can be either
+     * `"bilinear"` or `"nearest"` and default to `"bilinear"`. Currently two sampling
+     * methods are supported: Bilinear and Nearest Neighbor.
+     */
+    public Options method(String method) {
+      this.method = method;
+      return this;
+    }
+    
+    /**
+     * @param extrapolationValue Value used for extrapolation, when applicable.
+     */
+    public Options extrapolationValue(Float extrapolationValue) {
+      this.extrapolationValue = extrapolationValue;
+      return this;
+    }
+    
+    private String method;
+    private Float extrapolationValue;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new CropAndResize operation.
+   * 
+   * @param scope current scope
+   * @param image A 4-D tensor of shape `[batch, image_height, image_width, depth]`.
+   * Both `image_height` and `image_width` need to be positive.
+   * @param boxes A 2-D tensor of shape `[num_boxes, 4]`. The `i`-th row of the tensor
+   * specifies the coordinates of a box in the `box_ind[i]` image and is specified
+   * in normalized coordinates `[y1, x1, y2, x2]`. A normalized coordinate value of
+   * `y` is mapped to the image coordinate at `y * (image_height - 1)`, so as the
+   * `[0, 1]` interval of normalized image height is mapped to
+   * `[0, image_height - 1]` in image height coordinates. We do allow `y1` > `y2`, in
+   * which case the sampled crop is an up-down flipped version of the original
+   * image. The width dimension is treated similarly. Normalized coordinates
+   * outside the `[0, 1]` range are allowed, in which case we use
+   * `extrapolation_value` to extrapolate the input image values.
+   * @param boxInd A 1-D tensor of shape `[num_boxes]` with int32 values in `[0, batch)`.
+   * The value of `box_ind[i]` specifies the image that the `i`-th box refers to.
+   * @param cropSize A 1-D tensor of 2 elements, `size = [crop_height, crop_width]`. All
+   * cropped image patches are resized to this size. The aspect ratio of the image
+   * content is not preserved. Both `crop_height` and `crop_width` need to be
+   * positive.
+   * @param options carries optional attributes values
+   * @return a new instance of CropAndResize
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> CropAndResize create(Scope scope, Operand<T> image, Operand<TFloat32> boxes, Operand<TInt32> boxInd, Operand<TInt32> cropSize, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("CropAndResize", scope.makeOpName("CropAndResize"));
+    opBuilder.addInput(image.asOutput());
+    opBuilder.addInput(boxes.asOutput());
+    opBuilder.addInput(boxInd.asOutput());
+    opBuilder.addInput(cropSize.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.method != null) {
+          opBuilder.setAttr("method", opts.method);
+        }
+        if (opts.extrapolationValue != null) {
+          opBuilder.setAttr("extrapolation_value", opts.extrapolationValue);
+        }
+      }
+    }
+    return new CropAndResize(opBuilder.build());
+  }
+  
+  /**
+   * @param method A string specifying the sampling method for resizing. It can be either
+   * `"bilinear"` or `"nearest"` and default to `"bilinear"`. Currently two sampling
+   * methods are supported: Bilinear and Nearest Neighbor.
+   */
+  public static Options method(String method) {
+    return new Options().method(method);
+  }
+  
+  /**
+   * @param extrapolationValue Value used for extrapolation, when applicable.
+   */
+  public static Options extrapolationValue(Float extrapolationValue) {
+    return new Options().extrapolationValue(extrapolationValue);
+  }
+  
+  /**
+   * A 4-D tensor of shape `[num_boxes, crop_height, crop_width, depth]`.
+   */
+  public Output<TFloat32> crops() {
+    return crops;
+  }
+  
+  @Override
+  public Output<TFloat32> asOutput() {
+    return crops;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "CropAndResize";
+  
+  private Output<TFloat32> crops;
+  
+  private CropAndResize(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    crops = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/CropAndResizeGradBoxes.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/CropAndResizeGradBoxes.java
new file mode 100644
index 00000000000..4c3c0f678c1
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/CropAndResizeGradBoxes.java
@@ -0,0 +1,128 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.image;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Computes the gradient of the crop_and_resize op wrt the input boxes tensor.
+ */
+@Operator(group = "image")
+public final class CropAndResizeGradBoxes extends RawOp implements Operand<TFloat32> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.image.CropAndResizeGradBoxes}
+   */
+  public static class Options {
+    
+    /**
+     * @param method A string specifying the interpolation method. Only 'bilinear' is
+     * supported for now.
+     */
+    public Options method(String method) {
+      this.method = method;
+      return this;
+    }
+    
+    private String method;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new CropAndResizeGradBoxes operation.
+   * 
+   * @param scope current scope
+   * @param grads A 4-D tensor of shape `[num_boxes, crop_height, crop_width, depth]`.
+   * @param image A 4-D tensor of shape `[batch, image_height, image_width, depth]`.
+   * Both `image_height` and `image_width` need to be positive.
+   * @param boxes A 2-D tensor of shape `[num_boxes, 4]`. The `i`-th row of the tensor
+   * specifies the coordinates of a box in the `box_ind[i]` image and is specified
+   * in normalized coordinates `[y1, x1, y2, x2]`. A normalized coordinate value of
+   * `y` is mapped to the image coordinate at `y * (image_height - 1)`, so as the
+   * `[0, 1]` interval of normalized image height is mapped to
+   * `[0, image_height - 1] in image height coordinates. We do allow y1 > y2, in
+   * which case the sampled crop is an up-down flipped version of the original
+   * image. The width dimension is treated similarly. Normalized coordinates
+   * outside the `[0, 1]` range are allowed, in which case we use
+   * `extrapolation_value` to extrapolate the input image values.
+   * @param boxInd A 1-D tensor of shape `[num_boxes]` with int32 values in `[0, batch)`.
+   * The value of `box_ind[i]` specifies the image that the `i`-th box refers to.
+   * @param options carries optional attributes values
+   * @return a new instance of CropAndResizeGradBoxes
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> CropAndResizeGradBoxes create(Scope scope, Operand<TFloat32> grads, Operand<T> image, Operand<TFloat32> boxes, Operand<TInt32> boxInd, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("CropAndResizeGradBoxes", scope.makeOpName("CropAndResizeGradBoxes"));
+    opBuilder.addInput(grads.asOutput());
+    opBuilder.addInput(image.asOutput());
+    opBuilder.addInput(boxes.asOutput());
+    opBuilder.addInput(boxInd.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.method != null) {
+          opBuilder.setAttr("method", opts.method);
+        }
+      }
+    }
+    return new CropAndResizeGradBoxes(opBuilder.build());
+  }
+  
+  /**
+   * @param method A string specifying the interpolation method. Only 'bilinear' is
+   * supported for now.
+   */
+  public static Options method(String method) {
+    return new Options().method(method);
+  }
+  
+  /**
+   * A 2-D tensor of shape `[num_boxes, 4]`.
+   */
+  public Output<TFloat32> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<TFloat32> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "CropAndResizeGradBoxes";
+  
+  private Output<TFloat32> output;
+  
+  private CropAndResizeGradBoxes(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/CropAndResizeGradImage.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/CropAndResizeGradImage.java
new file mode 100644
index 00000000000..32a7bc6d247
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/CropAndResizeGradImage.java
@@ -0,0 +1,133 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.image;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Computes the gradient of the crop_and_resize op wrt the input image tensor.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator(group = "image")
+public final class CropAndResizeGradImage<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.image.CropAndResizeGradImage}
+   */
+  public static class Options {
+    
+    /**
+     * @param method A string specifying the interpolation method. Only 'bilinear' is
+     * supported for now.
+     */
+    public Options method(String method) {
+      this.method = method;
+      return this;
+    }
+    
+    private String method;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new CropAndResizeGradImage operation.
+   * 
+   * @param scope current scope
+   * @param grads A 4-D tensor of shape `[num_boxes, crop_height, crop_width, depth]`.
+   * @param boxes A 2-D tensor of shape `[num_boxes, 4]`. The `i`-th row of the tensor
+   * specifies the coordinates of a box in the `box_ind[i]` image and is specified
+   * in normalized coordinates `[y1, x1, y2, x2]`. A normalized coordinate value of
+   * `y` is mapped to the image coordinate at `y * (image_height - 1)`, so as the
+   * `[0, 1]` interval of normalized image height is mapped to
+   * `[0, image_height - 1] in image height coordinates. We do allow y1 > y2, in
+   * which case the sampled crop is an up-down flipped version of the original
+   * image. The width dimension is treated similarly. Normalized coordinates
+   * outside the `[0, 1]` range are allowed, in which case we use
+   * `extrapolation_value` to extrapolate the input image values.
+   * @param boxInd A 1-D tensor of shape `[num_boxes]` with int32 values in `[0, batch)`.
+   * The value of `box_ind[i]` specifies the image that the `i`-th box refers to.
+   * @param imageSize A 1-D tensor with value `[batch, image_height, image_width, depth]`
+   * containing the original image size. Both `image_height` and `image_width` need
+   * to be positive.
+   * @param T 
+   * @param options carries optional attributes values
+   * @return a new instance of CropAndResizeGradImage
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> CropAndResizeGradImage<T> create(Scope scope, Operand<TFloat32> grads, Operand<TFloat32> boxes, Operand<TInt32> boxInd, Operand<TInt32> imageSize, Class<T> T, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("CropAndResizeGradImage", scope.makeOpName("CropAndResizeGradImage"));
+    opBuilder.addInput(grads.asOutput());
+    opBuilder.addInput(boxes.asOutput());
+    opBuilder.addInput(boxInd.asOutput());
+    opBuilder.addInput(imageSize.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("T", T);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.method != null) {
+          opBuilder.setAttr("method", opts.method);
+        }
+      }
+    }
+    return new CropAndResizeGradImage<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param method A string specifying the interpolation method. Only 'bilinear' is
+   * supported for now.
+   */
+  public static Options method(String method) {
+    return new Options().method(method);
+  }
+  
+  /**
+   * A 4-D tensor of shape `[batch, image_height, image_width, depth]`.
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "CropAndResizeGradImage";
+  
+  private Output<T> output;
+  
+  private CropAndResizeGradImage(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/DecodeAndCropJpeg.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/DecodeAndCropJpeg.java
new file mode 100644
index 00000000000..0c2ac4b1471
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/DecodeAndCropJpeg.java
@@ -0,0 +1,245 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.image;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.TString;
+import org.tensorflow.types.TUint8;
+
+/**
+ * Decode and Crop a JPEG-encoded image to a uint8 tensor.
+ * <p>
+ * The attr `channels` indicates the desired number of color channels for the
+ * decoded image.
+ * <p>
+ * Accepted values are:
+ * <ul>
+ * <li>
+ * 0: Use the number of channels in the JPEG-encoded image.
+ * </li>
+ * <li>
+ * 1: output a grayscale image.
+ * </li>
+ * <li>
+ * 3: output an RGB image.
+ * </li>
+ * </ul>
+ * If needed, the JPEG-encoded image is transformed to match the requested number
+ * of color channels.
+ * <p>
+ * The attr `ratio` allows downscaling the image by an integer factor during
+ * decoding.  Allowed values are: 1, 2, 4, and 8.  This is much faster than
+ * downscaling the image later.
+ * <p>
+ * It is equivalent to a combination of decode and crop, but much faster by only
+ * decoding partial jpeg image.
+ */
+@Operator(group = "image")
+public final class DecodeAndCropJpeg extends RawOp implements Operand<TUint8> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.image.DecodeAndCropJpeg}
+   */
+  public static class Options {
+    
+    /**
+     * @param channels Number of color channels for the decoded image.
+     */
+    public Options channels(Long channels) {
+      this.channels = channels;
+      return this;
+    }
+    
+    /**
+     * @param ratio Downscaling ratio.
+     */
+    public Options ratio(Long ratio) {
+      this.ratio = ratio;
+      return this;
+    }
+    
+    /**
+     * @param fancyUpscaling If true use a slower but nicer upscaling of the
+     * chroma planes (yuv420/422 only).
+     */
+    public Options fancyUpscaling(Boolean fancyUpscaling) {
+      this.fancyUpscaling = fancyUpscaling;
+      return this;
+    }
+    
+    /**
+     * @param tryRecoverTruncated If true try to recover an image from truncated input.
+     */
+    public Options tryRecoverTruncated(Boolean tryRecoverTruncated) {
+      this.tryRecoverTruncated = tryRecoverTruncated;
+      return this;
+    }
+    
+    /**
+     * @param acceptableFraction The minimum required fraction of lines before a truncated
+     * input is accepted.
+     */
+    public Options acceptableFraction(Float acceptableFraction) {
+      this.acceptableFraction = acceptableFraction;
+      return this;
+    }
+    
+    /**
+     * @param dctMethod string specifying a hint about the algorithm used for
+     * decompression.  Defaults to "" which maps to a system-specific
+     * default.  Currently valid values are ["INTEGER_FAST",
+     * "INTEGER_ACCURATE"].  The hint may be ignored (e.g., the internal
+     * jpeg library changes to a version that does not have that specific
+     * option.)
+     */
+    public Options dctMethod(String dctMethod) {
+      this.dctMethod = dctMethod;
+      return this;
+    }
+    
+    private Long channels;
+    private Long ratio;
+    private Boolean fancyUpscaling;
+    private Boolean tryRecoverTruncated;
+    private Float acceptableFraction;
+    private String dctMethod;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new DecodeAndCropJpeg operation.
+   * 
+   * @param scope current scope
+   * @param contents 0-D.  The JPEG-encoded image.
+   * @param cropWindow 1-D.  The crop window: [crop_y, crop_x, crop_height, crop_width].
+   * @param options carries optional attributes values
+   * @return a new instance of DecodeAndCropJpeg
+   */
+  @Endpoint(describeByClass = true)
+  public static DecodeAndCropJpeg create(Scope scope, Operand<TString> contents, Operand<TInt32> cropWindow, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("DecodeAndCropJpeg", scope.makeOpName("DecodeAndCropJpeg"));
+    opBuilder.addInput(contents.asOutput());
+    opBuilder.addInput(cropWindow.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.channels != null) {
+          opBuilder.setAttr("channels", opts.channels);
+        }
+        if (opts.ratio != null) {
+          opBuilder.setAttr("ratio", opts.ratio);
+        }
+        if (opts.fancyUpscaling != null) {
+          opBuilder.setAttr("fancy_upscaling", opts.fancyUpscaling);
+        }
+        if (opts.tryRecoverTruncated != null) {
+          opBuilder.setAttr("try_recover_truncated", opts.tryRecoverTruncated);
+        }
+        if (opts.acceptableFraction != null) {
+          opBuilder.setAttr("acceptable_fraction", opts.acceptableFraction);
+        }
+        if (opts.dctMethod != null) {
+          opBuilder.setAttr("dct_method", opts.dctMethod);
+        }
+      }
+    }
+    return new DecodeAndCropJpeg(opBuilder.build());
+  }
+  
+  /**
+   * @param channels Number of color channels for the decoded image.
+   */
+  public static Options channels(Long channels) {
+    return new Options().channels(channels);
+  }
+  
+  /**
+   * @param ratio Downscaling ratio.
+   */
+  public static Options ratio(Long ratio) {
+    return new Options().ratio(ratio);
+  }
+  
+  /**
+   * @param fancyUpscaling If true use a slower but nicer upscaling of the
+   * chroma planes (yuv420/422 only).
+   */
+  public static Options fancyUpscaling(Boolean fancyUpscaling) {
+    return new Options().fancyUpscaling(fancyUpscaling);
+  }
+  
+  /**
+   * @param tryRecoverTruncated If true try to recover an image from truncated input.
+   */
+  public static Options tryRecoverTruncated(Boolean tryRecoverTruncated) {
+    return new Options().tryRecoverTruncated(tryRecoverTruncated);
+  }
+  
+  /**
+   * @param acceptableFraction The minimum required fraction of lines before a truncated
+   * input is accepted.
+   */
+  public static Options acceptableFraction(Float acceptableFraction) {
+    return new Options().acceptableFraction(acceptableFraction);
+  }
+  
+  /**
+   * @param dctMethod string specifying a hint about the algorithm used for
+   * decompression.  Defaults to "" which maps to a system-specific
+   * default.  Currently valid values are ["INTEGER_FAST",
+   * "INTEGER_ACCURATE"].  The hint may be ignored (e.g., the internal
+   * jpeg library changes to a version that does not have that specific
+   * option.)
+   */
+  public static Options dctMethod(String dctMethod) {
+    return new Options().dctMethod(dctMethod);
+  }
+  
+  /**
+   * 3-D with shape `[height, width, channels]`..
+   */
+  public Output<TUint8> image() {
+    return image;
+  }
+  
+  @Override
+  public Output<TUint8> asOutput() {
+    return image;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "DecodeAndCropJpeg";
+  
+  private Output<TUint8> image;
+  
+  private DecodeAndCropJpeg(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    image = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/DecodeBmp.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/DecodeBmp.java
new file mode 100644
index 00000000000..3801822db4e
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/DecodeBmp.java
@@ -0,0 +1,122 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.image;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TString;
+import org.tensorflow.types.TUint8;
+
+/**
+ * Decode the first frame of a BMP-encoded image to a uint8 tensor.
+ * <p>
+ * The attr `channels` indicates the desired number of color channels for the
+ * decoded image.
+ * <p>
+ * Accepted values are:
+ * <ul>
+ * <li>
+ * 0: Use the number of channels in the BMP-encoded image.
+ * </li>
+ * <li>
+ * 3: output an RGB image.
+ * </li>
+ * <li>
+ * 4: output an RGBA image.
+ */
+@Operator(group = "image")
+public final class DecodeBmp extends RawOp implements Operand<TUint8> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.image.DecodeBmp}
+   */
+  public static class Options {
+    
+    /**
+     * @param channels 
+     */
+    public Options channels(Long channels) {
+      this.channels = channels;
+      return this;
+    }
+    
+    private Long channels;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new DecodeBmp operation.
+   * 
+   * @param scope current scope
+   * @param contents 0-D.  The BMP-encoded image.
+   * @param options carries optional attributes values
+   * @return a new instance of DecodeBmp
+   */
+  @Endpoint(describeByClass = true)
+  public static DecodeBmp create(Scope scope, Operand<TString> contents, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("DecodeBmp", scope.makeOpName("DecodeBmp"));
+    opBuilder.addInput(contents.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.channels != null) {
+          opBuilder.setAttr("channels", opts.channels);
+        }
+      }
+    }
+    return new DecodeBmp(opBuilder.build());
+  }
+  
+  /**
+   * @param channels 
+   */
+  public static Options channels(Long channels) {
+    return new Options().channels(channels);
+  }
+  
+  /**
+   * 3-D with shape `[height, width, channels]`. RGB order
+   */
+  public Output<TUint8> image() {
+    return image;
+  }
+  
+  @Override
+  public Output<TUint8> asOutput() {
+    return image;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "DecodeBmp";
+  
+  private Output<TUint8> image;
+  
+  private DecodeBmp(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    image = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/DecodeGif.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/DecodeGif.java
new file mode 100644
index 00000000000..7c25b8c5b0c
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/DecodeGif.java
@@ -0,0 +1,83 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.image;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TString;
+import org.tensorflow.types.TUint8;
+
+/**
+ * Decode the frame(s) of a GIF-encoded image to a uint8 tensor.
+ * <p>
+ * GIF images with frame or transparency compression are not supported.
+ * On Linux and MacOS systems, convert animated GIFs from compressed to
+ * uncompressed by running:
+ * <p>
+ *     convert $src.gif -coalesce $dst.gif
+ * <p>
+ * This op also supports decoding JPEGs and PNGs, though it is cleaner to use
+ * `tf.io.decode_image`.
+ */
+@Operator(group = "image")
+public final class DecodeGif extends RawOp implements Operand<TUint8> {
+  
+  /**
+   * Factory method to create a class wrapping a new DecodeGif operation.
+   * 
+   * @param scope current scope
+   * @param contents 0-D.  The GIF-encoded image.
+   * @return a new instance of DecodeGif
+   */
+  @Endpoint(describeByClass = true)
+  public static DecodeGif create(Scope scope, Operand<TString> contents) {
+    OperationBuilder opBuilder = scope.env().opBuilder("DecodeGif", scope.makeOpName("DecodeGif"));
+    opBuilder.addInput(contents.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new DecodeGif(opBuilder.build());
+  }
+  
+  /**
+   * 4-D with shape `[num_frames, height, width, 3]`. RGB channel order.
+   */
+  public Output<TUint8> image() {
+    return image;
+  }
+  
+  @Override
+  public Output<TUint8> asOutput() {
+    return image;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "DecodeGif";
+  
+  private Output<TUint8> image;
+  
+  private DecodeGif(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    image = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/DecodeJpeg.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/DecodeJpeg.java
new file mode 100644
index 00000000000..e2b20460b2e
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/DecodeJpeg.java
@@ -0,0 +1,242 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.image;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TString;
+import org.tensorflow.types.TUint8;
+
+/**
+ * Decode a JPEG-encoded image to a uint8 tensor.
+ * <p>
+ * The attr `channels` indicates the desired number of color channels for the
+ * decoded image.
+ * <p>
+ * Accepted values are:
+ * <ul>
+ * <li>
+ * 0: Use the number of channels in the JPEG-encoded image.
+ * </li>
+ * <li>
+ * 1: output a grayscale image.
+ * </li>
+ * <li>
+ * 3: output an RGB image.
+ * </li>
+ * </ul>
+ * If needed, the JPEG-encoded image is transformed to match the requested number
+ * of color channels.
+ * <p>
+ * The attr `ratio` allows downscaling the image by an integer factor during
+ * decoding.  Allowed values are: 1, 2, 4, and 8.  This is much faster than
+ * downscaling the image later.
+ * <p>
+ * This op also supports decoding PNGs and non-animated GIFs since the interface is
+ * the same, though it is cleaner to use `tf.io.decode_image`.
+ */
+@Operator(group = "image")
+public final class DecodeJpeg extends RawOp implements Operand<TUint8> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.image.DecodeJpeg}
+   */
+  public static class Options {
+    
+    /**
+     * @param channels Number of color channels for the decoded image.
+     */
+    public Options channels(Long channels) {
+      this.channels = channels;
+      return this;
+    }
+    
+    /**
+     * @param ratio Downscaling ratio.
+     */
+    public Options ratio(Long ratio) {
+      this.ratio = ratio;
+      return this;
+    }
+    
+    /**
+     * @param fancyUpscaling If true use a slower but nicer upscaling of the
+     * chroma planes (yuv420/422 only).
+     */
+    public Options fancyUpscaling(Boolean fancyUpscaling) {
+      this.fancyUpscaling = fancyUpscaling;
+      return this;
+    }
+    
+    /**
+     * @param tryRecoverTruncated If true try to recover an image from truncated input.
+     */
+    public Options tryRecoverTruncated(Boolean tryRecoverTruncated) {
+      this.tryRecoverTruncated = tryRecoverTruncated;
+      return this;
+    }
+    
+    /**
+     * @param acceptableFraction The minimum required fraction of lines before a truncated
+     * input is accepted.
+     */
+    public Options acceptableFraction(Float acceptableFraction) {
+      this.acceptableFraction = acceptableFraction;
+      return this;
+    }
+    
+    /**
+     * @param dctMethod string specifying a hint about the algorithm used for
+     * decompression.  Defaults to "" which maps to a system-specific
+     * default.  Currently valid values are ["INTEGER_FAST",
+     * "INTEGER_ACCURATE"].  The hint may be ignored (e.g., the internal
+     * jpeg library changes to a version that does not have that specific
+     * option.)
+     */
+    public Options dctMethod(String dctMethod) {
+      this.dctMethod = dctMethod;
+      return this;
+    }
+    
+    private Long channels;
+    private Long ratio;
+    private Boolean fancyUpscaling;
+    private Boolean tryRecoverTruncated;
+    private Float acceptableFraction;
+    private String dctMethod;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new DecodeJpeg operation.
+   * 
+   * @param scope current scope
+   * @param contents 0-D.  The JPEG-encoded image.
+   * @param options carries optional attributes values
+   * @return a new instance of DecodeJpeg
+   */
+  @Endpoint(describeByClass = true)
+  public static DecodeJpeg create(Scope scope, Operand<TString> contents, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("DecodeJpeg", scope.makeOpName("DecodeJpeg"));
+    opBuilder.addInput(contents.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.channels != null) {
+          opBuilder.setAttr("channels", opts.channels);
+        }
+        if (opts.ratio != null) {
+          opBuilder.setAttr("ratio", opts.ratio);
+        }
+        if (opts.fancyUpscaling != null) {
+          opBuilder.setAttr("fancy_upscaling", opts.fancyUpscaling);
+        }
+        if (opts.tryRecoverTruncated != null) {
+          opBuilder.setAttr("try_recover_truncated", opts.tryRecoverTruncated);
+        }
+        if (opts.acceptableFraction != null) {
+          opBuilder.setAttr("acceptable_fraction", opts.acceptableFraction);
+        }
+        if (opts.dctMethod != null) {
+          opBuilder.setAttr("dct_method", opts.dctMethod);
+        }
+      }
+    }
+    return new DecodeJpeg(opBuilder.build());
+  }
+  
+  /**
+   * @param channels Number of color channels for the decoded image.
+   */
+  public static Options channels(Long channels) {
+    return new Options().channels(channels);
+  }
+  
+  /**
+   * @param ratio Downscaling ratio.
+   */
+  public static Options ratio(Long ratio) {
+    return new Options().ratio(ratio);
+  }
+  
+  /**
+   * @param fancyUpscaling If true use a slower but nicer upscaling of the
+   * chroma planes (yuv420/422 only).
+   */
+  public static Options fancyUpscaling(Boolean fancyUpscaling) {
+    return new Options().fancyUpscaling(fancyUpscaling);
+  }
+  
+  /**
+   * @param tryRecoverTruncated If true try to recover an image from truncated input.
+   */
+  public static Options tryRecoverTruncated(Boolean tryRecoverTruncated) {
+    return new Options().tryRecoverTruncated(tryRecoverTruncated);
+  }
+  
+  /**
+   * @param acceptableFraction The minimum required fraction of lines before a truncated
+   * input is accepted.
+   */
+  public static Options acceptableFraction(Float acceptableFraction) {
+    return new Options().acceptableFraction(acceptableFraction);
+  }
+  
+  /**
+   * @param dctMethod string specifying a hint about the algorithm used for
+   * decompression.  Defaults to "" which maps to a system-specific
+   * default.  Currently valid values are ["INTEGER_FAST",
+   * "INTEGER_ACCURATE"].  The hint may be ignored (e.g., the internal
+   * jpeg library changes to a version that does not have that specific
+   * option.)
+   */
+  public static Options dctMethod(String dctMethod) {
+    return new Options().dctMethod(dctMethod);
+  }
+  
+  /**
+   * 3-D with shape `[height, width, channels]`..
+   */
+  public Output<TUint8> image() {
+    return image;
+  }
+  
+  @Override
+  public Output<TUint8> asOutput() {
+    return image;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "DecodeJpeg";
+  
+  private Output<TUint8> image;
+  
+  private DecodeJpeg(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    image = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/DecodePng.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/DecodePng.java
new file mode 100644
index 00000000000..dc917cb869a
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/DecodePng.java
@@ -0,0 +1,150 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.image;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TString;
+import org.tensorflow.types.TUint8;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Decode a PNG-encoded image to a uint8 or uint16 tensor.
+ * <p>
+ * The attr `channels` indicates the desired number of color channels for the
+ * decoded image.
+ * <p>
+ * Accepted values are:
+ * <ul>
+ * <li>
+ * 0: Use the number of channels in the PNG-encoded image.
+ * </li>
+ * <li>
+ * 1: output a grayscale image.
+ * </li>
+ * <li>
+ * 3: output an RGB image.
+ * </li>
+ * <li>
+ * 4: output an RGBA image.
+ * </li>
+ * </ul>
+ * If needed, the PNG-encoded image is transformed to match the requested number
+ * of color channels.
+ * <p>
+ * This op also supports decoding JPEGs and non-animated GIFs since the interface
+ * is the same, though it is cleaner to use `tf.io.decode_image`.
+ * 
+ * @param <T> data type for {@code image()} output
+ */
+@Operator(group = "image")
+public final class DecodePng<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.image.DecodePng}
+   */
+  public static class Options {
+    
+    /**
+     * @param channels Number of color channels for the decoded image.
+     */
+    public Options channels(Long channels) {
+      this.channels = channels;
+      return this;
+    }
+    
+    private Long channels;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new DecodePng operation.
+   * 
+   * @param scope current scope
+   * @param contents 0-D.  The PNG-encoded image.
+   * @param dtype 
+   * @param options carries optional attributes values
+   * @return a new instance of DecodePng
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> DecodePng<T> create(Scope scope, Operand<TString> contents, Class<T> dtype, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("DecodePng", scope.makeOpName("DecodePng"));
+    opBuilder.addInput(contents.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("dtype", dtype);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.channels != null) {
+          opBuilder.setAttr("channels", opts.channels);
+        }
+      }
+    }
+    return new DecodePng<T>(opBuilder.build());
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new DecodePng operation using default output types.
+   * 
+   * @param scope current scope
+   * @param contents 0-D.  The PNG-encoded image.
+   * @param options carries optional attributes values
+   * @return a new instance of DecodePng
+   */
+  @Endpoint(describeByClass = true)
+  public static DecodePng<TUint8> create(Scope scope, Operand<TString> contents, Options... options) {
+    return create(scope, contents, TUint8.class, options);
+  }
+  
+  /**
+   * @param channels Number of color channels for the decoded image.
+   */
+  public static Options channels(Long channels) {
+    return new Options().channels(channels);
+  }
+  
+  /**
+   * 3-D with shape `[height, width, channels]`.
+   */
+  public Output<T> image() {
+    return image;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return image;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "DecodePng";
+  
+  private Output<T> image;
+  
+  private DecodePng(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    image = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/DrawBoundingBoxes.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/DrawBoundingBoxes.java
new file mode 100644
index 00000000000..edd3f0654d0
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/DrawBoundingBoxes.java
@@ -0,0 +1,94 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.image;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Draw bounding boxes on a batch of images.
+ * <p>
+ * Outputs a copy of `images` but draws on top of the pixels zero or more bounding
+ * boxes specified by the locations in `boxes`. The coordinates of the each
+ * bounding box in `boxes` are encoded as `[y_min, x_min, y_max, x_max]`. The
+ * bounding box coordinates are floats in `[0.0, 1.0]` relative to the width and
+ * height of the underlying image.
+ * <p>
+ * For example, if an image is 100 x 200 pixels (height x width) and the bounding
+ * box is `[0.1, 0.2, 0.5, 0.9]`, the upper-left and bottom-right coordinates of
+ * the bounding box will be `(40, 10)` to `(100, 50)` (in (x,y) coordinates).
+ * <p>
+ * Parts of the bounding box may fall outside the image.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator(group = "image")
+public final class DrawBoundingBoxes<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new DrawBoundingBoxes operation.
+   * 
+   * @param scope current scope
+   * @param images 4-D with shape `[batch, height, width, depth]`. A batch of images.
+   * @param boxes 3-D with shape `[batch, num_bounding_boxes, 4]` containing bounding
+   * boxes.
+   * @param colors 2-D. A list of RGBA colors to cycle through for the boxes.
+   * @return a new instance of DrawBoundingBoxes
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> DrawBoundingBoxes<T> create(Scope scope, Operand<T> images, Operand<TFloat32> boxes, Operand<TFloat32> colors) {
+    OperationBuilder opBuilder = scope.env().opBuilder("DrawBoundingBoxesV2", scope.makeOpName("DrawBoundingBoxes"));
+    opBuilder.addInput(images.asOutput());
+    opBuilder.addInput(boxes.asOutput());
+    opBuilder.addInput(colors.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new DrawBoundingBoxes<T>(opBuilder.build());
+  }
+  
+  /**
+   * 4-D with the same shape as `images`. The batch of input images with
+   * bounding boxes drawn on the images.
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "DrawBoundingBoxesV2";
+  
+  private Output<T> output;
+  
+  private DrawBoundingBoxes(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/EncodeJpeg.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/EncodeJpeg.java
new file mode 100644
index 00000000000..920961e1115
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/EncodeJpeg.java
@@ -0,0 +1,288 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.image;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TString;
+import org.tensorflow.types.TUint8;
+
+/**
+ * JPEG-encode an image.
+ * <p>
+ * `image` is a 3-D uint8 Tensor of shape `[height, width, channels]`.
+ * <p>
+ * The attr `format` can be used to override the color format of the encoded
+ * output.  Values can be:
+ * <ul>
+ * <li>
+ * `''`: Use a default format based on the number of channels in the image.
+ * </li>
+ * <li>
+ * `grayscale`: Output a grayscale JPEG image.  The `channels` dimension
+ *     of `image` must be 1.
+ * </li>
+ * <li>
+ * `rgb`: Output an RGB JPEG image. The `channels` dimension
+ *     of `image` must be 3.
+ * </li>
+ * </ul>
+ * If `format` is not specified or is the empty string, a default format is picked
+ * in function of the number of channels in `image`:
+ * <ul>
+ * <li>
+ * 1: Output a grayscale image.
+ * </li>
+ * <li>
+ * 3: Output an RGB image.
+ */
+@Operator(group = "image")
+public final class EncodeJpeg extends RawOp implements Operand<TString> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.image.EncodeJpeg}
+   */
+  public static class Options {
+    
+    /**
+     * @param format Per pixel image format.
+     */
+    public Options format(String format) {
+      this.format = format;
+      return this;
+    }
+    
+    /**
+     * @param quality Quality of the compression from 0 to 100 (higher is better and slower).
+     */
+    public Options quality(Long quality) {
+      this.quality = quality;
+      return this;
+    }
+    
+    /**
+     * @param progressive If True, create a JPEG that loads progressively (coarse to fine).
+     */
+    public Options progressive(Boolean progressive) {
+      this.progressive = progressive;
+      return this;
+    }
+    
+    /**
+     * @param optimizeSize If True, spend CPU/RAM to reduce size with no quality change.
+     */
+    public Options optimizeSize(Boolean optimizeSize) {
+      this.optimizeSize = optimizeSize;
+      return this;
+    }
+    
+    /**
+     * @param chromaDownsampling See http://en.wikipedia.org/wiki/Chroma_subsampling.
+     */
+    public Options chromaDownsampling(Boolean chromaDownsampling) {
+      this.chromaDownsampling = chromaDownsampling;
+      return this;
+    }
+    
+    /**
+     * @param densityUnit Unit used to specify `x_density` and `y_density`:
+     * pixels per inch (`'in'`) or centimeter (`'cm'`).
+     */
+    public Options densityUnit(String densityUnit) {
+      this.densityUnit = densityUnit;
+      return this;
+    }
+    
+    /**
+     * @param xDensity Horizontal pixels per density unit.
+     */
+    public Options xDensity(Long xDensity) {
+      this.xDensity = xDensity;
+      return this;
+    }
+    
+    /**
+     * @param yDensity Vertical pixels per density unit.
+     */
+    public Options yDensity(Long yDensity) {
+      this.yDensity = yDensity;
+      return this;
+    }
+    
+    /**
+     * @param xmpMetadata If not empty, embed this XMP metadata in the image header.
+     */
+    public Options xmpMetadata(String xmpMetadata) {
+      this.xmpMetadata = xmpMetadata;
+      return this;
+    }
+    
+    private String format;
+    private Long quality;
+    private Boolean progressive;
+    private Boolean optimizeSize;
+    private Boolean chromaDownsampling;
+    private String densityUnit;
+    private Long xDensity;
+    private Long yDensity;
+    private String xmpMetadata;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new EncodeJpeg operation.
+   * 
+   * @param scope current scope
+   * @param image 3-D with shape `[height, width, channels]`.
+   * @param options carries optional attributes values
+   * @return a new instance of EncodeJpeg
+   */
+  @Endpoint(describeByClass = true)
+  public static EncodeJpeg create(Scope scope, Operand<TUint8> image, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("EncodeJpeg", scope.makeOpName("EncodeJpeg"));
+    opBuilder.addInput(image.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.format != null) {
+          opBuilder.setAttr("format", opts.format);
+        }
+        if (opts.quality != null) {
+          opBuilder.setAttr("quality", opts.quality);
+        }
+        if (opts.progressive != null) {
+          opBuilder.setAttr("progressive", opts.progressive);
+        }
+        if (opts.optimizeSize != null) {
+          opBuilder.setAttr("optimize_size", opts.optimizeSize);
+        }
+        if (opts.chromaDownsampling != null) {
+          opBuilder.setAttr("chroma_downsampling", opts.chromaDownsampling);
+        }
+        if (opts.densityUnit != null) {
+          opBuilder.setAttr("density_unit", opts.densityUnit);
+        }
+        if (opts.xDensity != null) {
+          opBuilder.setAttr("x_density", opts.xDensity);
+        }
+        if (opts.yDensity != null) {
+          opBuilder.setAttr("y_density", opts.yDensity);
+        }
+        if (opts.xmpMetadata != null) {
+          opBuilder.setAttr("xmp_metadata", opts.xmpMetadata);
+        }
+      }
+    }
+    return new EncodeJpeg(opBuilder.build());
+  }
+  
+  /**
+   * @param format Per pixel image format.
+   */
+  public static Options format(String format) {
+    return new Options().format(format);
+  }
+  
+  /**
+   * @param quality Quality of the compression from 0 to 100 (higher is better and slower).
+   */
+  public static Options quality(Long quality) {
+    return new Options().quality(quality);
+  }
+  
+  /**
+   * @param progressive If True, create a JPEG that loads progressively (coarse to fine).
+   */
+  public static Options progressive(Boolean progressive) {
+    return new Options().progressive(progressive);
+  }
+  
+  /**
+   * @param optimizeSize If True, spend CPU/RAM to reduce size with no quality change.
+   */
+  public static Options optimizeSize(Boolean optimizeSize) {
+    return new Options().optimizeSize(optimizeSize);
+  }
+  
+  /**
+   * @param chromaDownsampling See http://en.wikipedia.org/wiki/Chroma_subsampling.
+   */
+  public static Options chromaDownsampling(Boolean chromaDownsampling) {
+    return new Options().chromaDownsampling(chromaDownsampling);
+  }
+  
+  /**
+   * @param densityUnit Unit used to specify `x_density` and `y_density`:
+   * pixels per inch (`'in'`) or centimeter (`'cm'`).
+   */
+  public static Options densityUnit(String densityUnit) {
+    return new Options().densityUnit(densityUnit);
+  }
+  
+  /**
+   * @param xDensity Horizontal pixels per density unit.
+   */
+  public static Options xDensity(Long xDensity) {
+    return new Options().xDensity(xDensity);
+  }
+  
+  /**
+   * @param yDensity Vertical pixels per density unit.
+   */
+  public static Options yDensity(Long yDensity) {
+    return new Options().yDensity(yDensity);
+  }
+  
+  /**
+   * @param xmpMetadata If not empty, embed this XMP metadata in the image header.
+   */
+  public static Options xmpMetadata(String xmpMetadata) {
+    return new Options().xmpMetadata(xmpMetadata);
+  }
+  
+  /**
+   * 0-D. JPEG-encoded image.
+   */
+  public Output<TString> contents() {
+    return contents;
+  }
+  
+  @Override
+  public Output<TString> asOutput() {
+    return contents;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "EncodeJpeg";
+  
+  private Output<TString> contents;
+  
+  private EncodeJpeg(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    contents = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/EncodeJpegVariableQuality.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/EncodeJpegVariableQuality.java
new file mode 100644
index 00000000000..c07fc341f00
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/EncodeJpegVariableQuality.java
@@ -0,0 +1,81 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.image;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.TString;
+import org.tensorflow.types.TUint8;
+
+/**
+ * JPEG encode input image with provided compression quality.
+ * <p>
+ * `image` is a 3-D uint8 Tensor of shape `[height, width, channels]`.
+ * `quality` is an int32 jpeg compression quality value between 0 and 100.
+ * 
+ */
+@Operator(group = "image")
+public final class EncodeJpegVariableQuality extends RawOp implements Operand<TString> {
+  
+  /**
+   * Factory method to create a class wrapping a new EncodeJpegVariableQuality operation.
+   * 
+   * @param scope current scope
+   * @param images Images to adjust.  At least 3-D.
+   * @param quality An int quality to encode to.
+   * @return a new instance of EncodeJpegVariableQuality
+   */
+  @Endpoint(describeByClass = true)
+  public static EncodeJpegVariableQuality create(Scope scope, Operand<TUint8> images, Operand<TInt32> quality) {
+    OperationBuilder opBuilder = scope.env().opBuilder("EncodeJpegVariableQuality", scope.makeOpName("EncodeJpegVariableQuality"));
+    opBuilder.addInput(images.asOutput());
+    opBuilder.addInput(quality.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new EncodeJpegVariableQuality(opBuilder.build());
+  }
+  
+  /**
+   * 0-D. JPEG-encoded image.
+   */
+  public Output<TString> contents() {
+    return contents;
+  }
+  
+  @Override
+  public Output<TString> asOutput() {
+    return contents;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "EncodeJpegVariableQuality";
+  
+  private Output<TString> contents;
+  
+  private EncodeJpegVariableQuality(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    contents = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/EncodePng.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/EncodePng.java
new file mode 100644
index 00000000000..5944fbc9f87
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/EncodePng.java
@@ -0,0 +1,128 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.image;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TString;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * PNG-encode an image.
+ * <p>
+ * `image` is a 3-D uint8 or uint16 Tensor of shape `[height, width, channels]`
+ * where `channels` is:
+ * <ul>
+ * <li>
+ * 1: for grayscale.
+ * </li>
+ * <li>
+ * 2: for grayscale + alpha.
+ * </li>
+ * <li>
+ * 3: for RGB.
+ * </li>
+ * <li>
+ * 4: for RGBA.
+ * </li>
+ * </ul>
+ * The ZLIB compression level, `compression`, can be -1 for the PNG-encoder
+ * default or a value from 0 to 9.  9 is the highest compression level, generating
+ * the smallest output, but is slower.
+ */
+@Operator(group = "image")
+public final class EncodePng extends RawOp implements Operand<TString> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.image.EncodePng}
+   */
+  public static class Options {
+    
+    /**
+     * @param compression Compression level.
+     */
+    public Options compression(Long compression) {
+      this.compression = compression;
+      return this;
+    }
+    
+    private Long compression;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new EncodePng operation.
+   * 
+   * @param scope current scope
+   * @param image 3-D with shape `[height, width, channels]`.
+   * @param options carries optional attributes values
+   * @return a new instance of EncodePng
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> EncodePng create(Scope scope, Operand<T> image, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("EncodePng", scope.makeOpName("EncodePng"));
+    opBuilder.addInput(image.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.compression != null) {
+          opBuilder.setAttr("compression", opts.compression);
+        }
+      }
+    }
+    return new EncodePng(opBuilder.build());
+  }
+  
+  /**
+   * @param compression Compression level.
+   */
+  public static Options compression(Long compression) {
+    return new Options().compression(compression);
+  }
+  
+  /**
+   * 0-D. PNG-encoded image.
+   */
+  public Output<TString> contents() {
+    return contents;
+  }
+  
+  @Override
+  public Output<TString> asOutput() {
+    return contents;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "EncodePng";
+  
+  private Output<TString> contents;
+  
+  private EncodePng(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    contents = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ExtractGlimpse.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ExtractGlimpse.java
new file mode 100644
index 00000000000..05bc1d924b3
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ExtractGlimpse.java
@@ -0,0 +1,211 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.image;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.TInt32;
+
+/**
+ * Extracts a glimpse from the input tensor.
+ * <p>
+ * Returns a set of windows called glimpses extracted at location
+ * `offsets` from the input tensor. If the windows only partially
+ * overlaps the inputs, the non overlapping areas will be filled with
+ * random noise.
+ * <p>
+ * The result is a 4-D tensor of shape `[batch_size, glimpse_height,
+ * glimpse_width, channels]`. The channels and batch dimensions are the
+ * same as that of the input tensor. The height and width of the output
+ * windows are specified in the `size` parameter.
+ * <p>
+ * The argument `normalized` and `centered` controls how the windows are built:
+ * <ul>
+ * <li>
+ * If the coordinates are normalized but not centered, 0.0 and 1.0
+ *   correspond to the minimum and maximum of each height and width
+ *   dimension.
+ * </li>
+ * <li>
+ * If the coordinates are both normalized and centered, they range from
+ *   -1.0 to 1.0. The coordinates (-1.0, -1.0) correspond to the upper
+ *   left corner, the lower right corner is located at (1.0, 1.0) and the
+ *   center is at (0, 0).
+ * </li>
+ * <li>
+ * If the coordinates are not normalized they are interpreted as
+ *   numbers of pixels.
+ */
+public final class ExtractGlimpse extends RawOp implements Operand<TFloat32> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.image.ExtractGlimpse}
+   */
+  public static class Options {
+    
+    /**
+     * @param centered indicates if the offset coordinates are centered relative to
+     * the image, in which case the (0, 0) offset is relative to the center
+     * of the input images. If false, the (0,0) offset corresponds to the
+     * upper left corner of the input images.
+     */
+    public Options centered(Boolean centered) {
+      this.centered = centered;
+      return this;
+    }
+    
+    /**
+     * @param normalized indicates if the offset coordinates are normalized.
+     */
+    public Options normalized(Boolean normalized) {
+      this.normalized = normalized;
+      return this;
+    }
+    
+    /**
+     * @param uniformNoise indicates if the noise should be generated using a
+     * uniform distribution or a Gaussian distribution.
+     */
+    public Options uniformNoise(Boolean uniformNoise) {
+      this.uniformNoise = uniformNoise;
+      return this;
+    }
+    
+    /**
+     * @param noise indicates if the noise should `uniform`, `gaussian`, or
+     * `zero`. The default is `uniform` which means the the noise type
+     * will be decided by `uniform_noise`.
+     */
+    public Options noise(String noise) {
+      this.noise = noise;
+      return this;
+    }
+    
+    private Boolean centered;
+    private Boolean normalized;
+    private Boolean uniformNoise;
+    private String noise;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new ExtractGlimpse operation.
+   * 
+   * @param scope current scope
+   * @param input A 4-D float tensor of shape `[batch_size, height, width, channels]`.
+   * @param size A 1-D tensor of 2 elements containing the size of the glimpses
+   * to extract.  The glimpse height must be specified first, following
+   * by the glimpse width.
+   * @param offsets A 2-D integer tensor of shape `[batch_size, 2]` containing
+   * the y, x locations of the center of each window.
+   * @param options carries optional attributes values
+   * @return a new instance of ExtractGlimpse
+   */
+  @Endpoint(describeByClass = true)
+  public static ExtractGlimpse create(Scope scope, Operand<TFloat32> input, Operand<TInt32> size, Operand<TFloat32> offsets, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ExtractGlimpseV2", scope.makeOpName("ExtractGlimpse"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(size.asOutput());
+    opBuilder.addInput(offsets.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.centered != null) {
+          opBuilder.setAttr("centered", opts.centered);
+        }
+        if (opts.normalized != null) {
+          opBuilder.setAttr("normalized", opts.normalized);
+        }
+        if (opts.uniformNoise != null) {
+          opBuilder.setAttr("uniform_noise", opts.uniformNoise);
+        }
+        if (opts.noise != null) {
+          opBuilder.setAttr("noise", opts.noise);
+        }
+      }
+    }
+    return new ExtractGlimpse(opBuilder.build());
+  }
+  
+  /**
+   * @param centered indicates if the offset coordinates are centered relative to
+   * the image, in which case the (0, 0) offset is relative to the center
+   * of the input images. If false, the (0,0) offset corresponds to the
+   * upper left corner of the input images.
+   */
+  public static Options centered(Boolean centered) {
+    return new Options().centered(centered);
+  }
+  
+  /**
+   * @param normalized indicates if the offset coordinates are normalized.
+   */
+  public static Options normalized(Boolean normalized) {
+    return new Options().normalized(normalized);
+  }
+  
+  /**
+   * @param uniformNoise indicates if the noise should be generated using a
+   * uniform distribution or a Gaussian distribution.
+   */
+  public static Options uniformNoise(Boolean uniformNoise) {
+    return new Options().uniformNoise(uniformNoise);
+  }
+  
+  /**
+   * @param noise indicates if the noise should `uniform`, `gaussian`, or
+   * `zero`. The default is `uniform` which means the the noise type
+   * will be decided by `uniform_noise`.
+   */
+  public static Options noise(String noise) {
+    return new Options().noise(noise);
+  }
+  
+  /**
+   * A tensor representing the glimpses `[batch_size,
+   * glimpse_height, glimpse_width, channels]`.
+   */
+  public Output<TFloat32> glimpse() {
+    return glimpse;
+  }
+  
+  @Override
+  public Output<TFloat32> asOutput() {
+    return glimpse;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ExtractGlimpseV2";
+  
+  private Output<TFloat32> glimpse;
+  
+  private ExtractGlimpse(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    glimpse = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ExtractImagePatches.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ExtractImagePatches.java
new file mode 100644
index 00000000000..d02d422a005
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ExtractImagePatches.java
@@ -0,0 +1,105 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.image;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Extract `patches` from `images` and put them in the "depth" output dimension.
+ * 
+ * @param <T> data type for {@code patches()} output
+ */
+@Operator(group = "image")
+public final class ExtractImagePatches<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new ExtractImagePatches operation.
+   * 
+   * @param scope current scope
+   * @param images 4-D Tensor with shape `[batch, in_rows, in_cols, depth]`.
+   * @param ksizes The size of the sliding window for each dimension of `images`.
+   * @param strides How far the centers of two consecutive patches are in
+   * the images. Must be: `[1, stride_rows, stride_cols, 1]`.
+   * @param rates Must be: `[1, rate_rows, rate_cols, 1]`. This is the
+   * input stride, specifying how far two consecutive patch samples are in the
+   * input. Equivalent to extracting patches with
+   * `patch_sizes_eff = patch_sizes + (patch_sizes - 1) * (rates - 1)`, followed by
+   * subsampling them spatially by a factor of `rates`. This is equivalent to
+   * `rate` in dilated (a.k.a. Atrous) convolutions.
+   * @param padding The type of padding algorithm to use.
+   * @return a new instance of ExtractImagePatches
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> ExtractImagePatches<T> create(Scope scope, Operand<T> images, List<Long> ksizes, List<Long> strides, List<Long> rates, String padding) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ExtractImagePatches", scope.makeOpName("ExtractImagePatches"));
+    opBuilder.addInput(images.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    long[] ksizesArray = new long[ksizes.size()];
+    for (int i = 0; i < ksizesArray.length; ++i) {
+      ksizesArray[i] = ksizes.get(i);
+    }
+    opBuilder.setAttr("ksizes", ksizesArray);
+    long[] stridesArray = new long[strides.size()];
+    for (int i = 0; i < stridesArray.length; ++i) {
+      stridesArray[i] = strides.get(i);
+    }
+    opBuilder.setAttr("strides", stridesArray);
+    long[] ratesArray = new long[rates.size()];
+    for (int i = 0; i < ratesArray.length; ++i) {
+      ratesArray[i] = rates.get(i);
+    }
+    opBuilder.setAttr("rates", ratesArray);
+    opBuilder.setAttr("padding", padding);
+    return new ExtractImagePatches<T>(opBuilder.build());
+  }
+  
+  /**
+   * 4-D Tensor with shape `[batch, out_rows, out_cols, ksize_rows *
+   * ksize_cols * depth]` containing image patches with size
+   * `ksize_rows x ksize_cols x depth` vectorized in the "depth" dimension. Note
+   * `out_rows` and `out_cols` are the dimensions of the output patches.
+   */
+  public Output<T> patches() {
+    return patches;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return patches;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ExtractImagePatches";
+  
+  private Output<T> patches;
+  
+  private ExtractImagePatches(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    patches = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ExtractJpegShape.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ExtractJpegShape.java
new file mode 100644
index 00000000000..1f5ff7163fb
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ExtractJpegShape.java
@@ -0,0 +1,94 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.image;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.TString;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Extract the shape information of a JPEG-encoded image.
+ * <p>
+ * This op only parses the image header, so it is much faster than DecodeJpeg.
+ * 
+ * @param <T> data type for {@code imageShape()} output
+ */
+@Operator(group = "image")
+public final class ExtractJpegShape<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new ExtractJpegShape operation.
+   * 
+   * @param scope current scope
+   * @param contents 0-D. The JPEG-encoded image.
+   * @param outputType (Optional) The output type of the operation (int32 or int64).
+   * Defaults to int32.
+   * @return a new instance of ExtractJpegShape
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> ExtractJpegShape<T> create(Scope scope, Operand<TString> contents, Class<T> outputType) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ExtractJpegShape", scope.makeOpName("ExtractJpegShape"));
+    opBuilder.addInput(contents.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("output_type", outputType);
+    return new ExtractJpegShape<T>(opBuilder.build());
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new ExtractJpegShape operation using default output types.
+   * 
+   * @param scope current scope
+   * @param contents 0-D. The JPEG-encoded image.
+   * @return a new instance of ExtractJpegShape
+   */
+  @Endpoint(describeByClass = true)
+  public static ExtractJpegShape<TInt32> create(Scope scope, Operand<TString> contents) {
+    return create(scope, contents, TInt32.class);
+  }
+  
+  /**
+   * 1-D. The image shape with format [height, width, channels].
+   */
+  public Output<T> imageShape() {
+    return imageShape;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return imageShape;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ExtractJpegShape";
+  
+  private Output<T> imageShape;
+  
+  private ExtractJpegShape(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    imageShape = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/GenerateBoundingBoxProposals.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/GenerateBoundingBoxProposals.java
new file mode 100644
index 00000000000..e287a419cdd
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/GenerateBoundingBoxProposals.java
@@ -0,0 +1,138 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.image;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.TInt32;
+
+/**
+ * This op produces Region of Interests from given bounding boxes(bbox_deltas) encoded wrt anchors according to eq.2 in arXiv:1506.01497
+ * <p>
+ *       The op selects top `pre_nms_topn` scoring boxes, decodes them with respect to anchors,
+ *       applies non-maximal suppression on overlapping boxes with higher than
+ *       `nms_threshold` intersection-over-union (iou) value, discarding boxes where shorter
+ *       side is less than `min_size`.
+ *       Inputs:
+ *       `scores`: A 4D tensor of shape [Batch, Height, Width, Num Anchors] containing the scores per anchor at given position
+ *       `bbox_deltas`: is a tensor of shape [Batch, Height, Width, 4 x Num Anchors] boxes encoded to each anchor
+ *       `anchors`: A 1D tensor of shape [4 x Num Anchors], representing the anchors.
+ *       Outputs:
+ *       `rois`: output RoIs, a 3D tensor of shape [Batch, post_nms_topn, 4], padded by 0 if less than post_nms_topn candidates found.
+ *       `roi_probabilities`: probability scores of each roi in 'rois', a 2D tensor of shape [Batch,post_nms_topn], padded with 0 if needed, sorted by scores.
+ */
+public final class GenerateBoundingBoxProposals extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.image.GenerateBoundingBoxProposals}
+   */
+  public static class Options {
+    
+    /**
+     * @param postNmsTopn An integer. Maximum number of rois in the output.
+     */
+    public Options postNmsTopn(Long postNmsTopn) {
+      this.postNmsTopn = postNmsTopn;
+      return this;
+    }
+    
+    private Long postNmsTopn;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new GenerateBoundingBoxProposals operation.
+   * 
+   * @param scope current scope
+   * @param scores A 4-D float tensor of shape `[num_images, height, width, num_achors]` containing scores of the boxes for given anchors, can be unsorted.
+   * @param bboxDeltas A 4-D float tensor of shape `[num_images, height, width, 4 x num_anchors]`. encoding boxes with respec to each anchor.
+   * Coordinates are given in the form [dy, dx, dh, dw].
+   * @param imageInfo A 2-D float tensor of shape `[num_images, 5]` containing image information Height, Width, Scale.
+   * @param anchors A 2-D float tensor of shape `[num_anchors, 4]` describing the anchor boxes. Boxes are formatted in the form [y1, x1, y2, x2].
+   * @param nmsThreshold A scalar float tensor for non-maximal-suppression threshold.
+   * @param preNmsTopn A scalar int tensor for the number of top scoring boxes to be used as input.
+   * @param minSize A scalar float tensor. Any box that has a smaller size than min_size will be discarded.
+   * @param options carries optional attributes values
+   * @return a new instance of GenerateBoundingBoxProposals
+   */
+  @Endpoint(describeByClass = true)
+  public static GenerateBoundingBoxProposals create(Scope scope, Operand<TFloat32> scores, Operand<TFloat32> bboxDeltas, Operand<TFloat32> imageInfo, Operand<TFloat32> anchors, Operand<TFloat32> nmsThreshold, Operand<TInt32> preNmsTopn, Operand<TFloat32> minSize, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("GenerateBoundingBoxProposals", scope.makeOpName("GenerateBoundingBoxProposals"));
+    opBuilder.addInput(scores.asOutput());
+    opBuilder.addInput(bboxDeltas.asOutput());
+    opBuilder.addInput(imageInfo.asOutput());
+    opBuilder.addInput(anchors.asOutput());
+    opBuilder.addInput(nmsThreshold.asOutput());
+    opBuilder.addInput(preNmsTopn.asOutput());
+    opBuilder.addInput(minSize.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.postNmsTopn != null) {
+          opBuilder.setAttr("post_nms_topn", opts.postNmsTopn);
+        }
+      }
+    }
+    return new GenerateBoundingBoxProposals(opBuilder.build());
+  }
+  
+  /**
+   * @param postNmsTopn An integer. Maximum number of rois in the output.
+   */
+  public static Options postNmsTopn(Long postNmsTopn) {
+    return new Options().postNmsTopn(postNmsTopn);
+  }
+  
+  /**
+   * A 3-D float tensor of shape `[num_images,post_nms_topn,4]` representing the selected
+   * region of interest boxes. Sorted in descending order in scores.
+   */
+  public Output<TFloat32> rois() {
+    return rois;
+  }
+  
+  /**
+   * A 2-D float tensor of shape `[num_images, post_nms_topn]` representing the score of the
+   * region of interest box in `rois` tensor at the same index.
+   */
+  public Output<TFloat32> roiProbabilities() {
+    return roiProbabilities;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "GenerateBoundingBoxProposals";
+  
+  private Output<TFloat32> rois;
+  private Output<TFloat32> roiProbabilities;
+  
+  private GenerateBoundingBoxProposals(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    rois = operation.output(outputIdx++);
+    roiProbabilities = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/HsvToRgb.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/HsvToRgb.java
new file mode 100644
index 00000000000..90b9c5ea776
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/HsvToRgb.java
@@ -0,0 +1,81 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.image;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Convert one or more images from HSV to RGB.
+ * <p>
+ * Outputs a tensor of the same shape as the `images` tensor, containing the RGB
+ * value of the pixels. The output is only well defined if the value in `images`
+ * are in `[0,1]`.
+ * <p>
+ * See `rgb_to_hsv` for a description of the HSV encoding.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator(group = "image")
+public final class HsvToRgb<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new HsvToRgb operation.
+   * 
+   * @param scope current scope
+   * @param images 1-D or higher rank. HSV data to convert. Last dimension must be size 3.
+   * @return a new instance of HsvToRgb
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> HsvToRgb<T> create(Scope scope, Operand<T> images) {
+    OperationBuilder opBuilder = scope.env().opBuilder("HSVToRGB", scope.makeOpName("HsvToRgb"));
+    opBuilder.addInput(images.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new HsvToRgb<T>(opBuilder.build());
+  }
+  
+  /**
+   * `images` converted to RGB.
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "HSVToRGB";
+  
+  private Output<T> output;
+  
+  private HsvToRgb(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ImageProjectiveTransformV2.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ImageProjectiveTransformV2.java
new file mode 100644
index 00000000000..a1c786cb619
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ImageProjectiveTransformV2.java
@@ -0,0 +1,125 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.image;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Applies the given transform to each of the images.
+ * <p>
+ * If one row of `transforms` is `[a0, a1, a2, b0, b1, b2, c0, c1]`, then it maps
+ * the <i>output</i> point `(x, y)` to a transformed <i>input</i> point
+ * `(x', y') = ((a0 x + a1 y + a2) / k, (b0 x + b1 y + b2) / k)`, where
+ * `k = c0 x + c1 y + 1`. If the transformed point lays outside of the input
+ * image, the output pixel is set to 0.
+ * 
+ * @param <T> data type for {@code transformedImages()} output
+ */
+public final class ImageProjectiveTransformV2<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.image.ImageProjectiveTransformV2}
+   */
+  public static class Options {
+    
+    /**
+     * @param fillMode Fill mode, "REFLECT", "WRAP", or "CONSTANT".
+     */
+    public Options fillMode(String fillMode) {
+      this.fillMode = fillMode;
+      return this;
+    }
+    
+    private String fillMode;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new ImageProjectiveTransformV2 operation.
+   * 
+   * @param scope current scope
+   * @param images 4-D with shape `[batch, height, width, channels]`.
+   * @param transforms 2-D Tensor, `[batch, 8]` or `[1, 8]` matrix, where each row corresponds to a 3 x 3
+   * projective transformation matrix, with the last entry assumed to be 1. If there
+   * is one row, the same transformation will be applied to all images.
+   * @param outputShape 1-D Tensor [new_height, new_width].
+   * @param interpolation Interpolation method, "NEAREST" or "BILINEAR".
+   * @param options carries optional attributes values
+   * @return a new instance of ImageProjectiveTransformV2
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> ImageProjectiveTransformV2<T> create(Scope scope, Operand<T> images, Operand<TFloat32> transforms, Operand<TInt32> outputShape, String interpolation, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ImageProjectiveTransformV2", scope.makeOpName("ImageProjectiveTransformV2"));
+    opBuilder.addInput(images.asOutput());
+    opBuilder.addInput(transforms.asOutput());
+    opBuilder.addInput(outputShape.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("interpolation", interpolation);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.fillMode != null) {
+          opBuilder.setAttr("fill_mode", opts.fillMode);
+        }
+      }
+    }
+    return new ImageProjectiveTransformV2<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param fillMode Fill mode, "REFLECT", "WRAP", or "CONSTANT".
+   */
+  public static Options fillMode(String fillMode) {
+    return new Options().fillMode(fillMode);
+  }
+  
+  /**
+   * 4-D with shape
+   * `[batch, new_height, new_width, channels]`.
+   */
+  public Output<T> transformedImages() {
+    return transformedImages;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return transformedImages;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ImageProjectiveTransformV2";
+  
+  private Output<T> transformedImages;
+  
+  private ImageProjectiveTransformV2(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    transformedImages = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/NearestNeighbors.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/NearestNeighbors.java
new file mode 100644
index 00000000000..f3ad7cb890b
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/NearestNeighbors.java
@@ -0,0 +1,88 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.image;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.TInt64;
+
+/**
+ * Selects the k nearest centers for each point.
+ * <p>
+ * Rows of points are assumed to be input points. Rows of centers are assumed to be
+ * the list of candidate centers. For each point, the k centers that have least L2
+ * distance to it are computed.
+ */
+public final class NearestNeighbors extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new NearestNeighbors operation.
+   * 
+   * @param scope current scope
+   * @param points Matrix of shape (n, d). Rows are assumed to be input points.
+   * @param centers Matrix of shape (m, d). Rows are assumed to be centers.
+   * @param k Number of nearest centers to return for each point. If k is larger than m, then
+   * only m centers are returned.
+   * @return a new instance of NearestNeighbors
+   */
+  @Endpoint(describeByClass = true)
+  public static NearestNeighbors create(Scope scope, Operand<TFloat32> points, Operand<TFloat32> centers, Operand<TInt64> k) {
+    OperationBuilder opBuilder = scope.env().opBuilder("NearestNeighbors", scope.makeOpName("NearestNeighbors"));
+    opBuilder.addInput(points.asOutput());
+    opBuilder.addInput(centers.asOutput());
+    opBuilder.addInput(k.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new NearestNeighbors(opBuilder.build());
+  }
+  
+  /**
+   * Matrix of shape (n, min(m, k)). Each row contains the indices of the centers
+   * closest to the corresponding point, ordered by increasing distance.
+   */
+  public Output<TInt64> nearestCenterIndices() {
+    return nearestCenterIndices;
+  }
+  
+  /**
+   * Matrix of shape (n, min(m, k)). Each row contains the squared L2 distance to the
+   * corresponding center in nearest_center_indices.
+   */
+  public Output<TFloat32> nearestCenterDistances() {
+    return nearestCenterDistances;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "NearestNeighbors";
+  
+  private Output<TInt64> nearestCenterIndices;
+  private Output<TFloat32> nearestCenterDistances;
+  
+  private NearestNeighbors(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    nearestCenterIndices = operation.output(outputIdx++);
+    nearestCenterDistances = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/NonMaxSuppression.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/NonMaxSuppression.java
new file mode 100644
index 00000000000..919a761f5c4
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/NonMaxSuppression.java
@@ -0,0 +1,169 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.image;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Greedily selects a subset of bounding boxes in descending order of score,
+ * <p>
+ * pruning away boxes that have high intersection-over-union (IOU) overlap
+ * with previously selected boxes.  Bounding boxes with score less than
+ * `score_threshold` are removed.  Bounding boxes are supplied as
+ * [y1, x1, y2, x2], where (y1, x1) and (y2, x2) are the coordinates of any
+ * diagonal pair of box corners and the coordinates can be provided as normalized
+ * (i.e., lying in the interval [0, 1]) or absolute.  Note that this algorithm
+ * is agnostic to where the origin is in the coordinate system and more
+ * generally is invariant to orthogonal transformations and translations
+ * of the coordinate system; thus translating or reflections of the coordinate
+ * system result in the same boxes being selected by the algorithm.
+ * The output of this operation is a set of integers indexing into the input
+ * collection of bounding boxes representing the selected boxes.  The bounding
+ * box coordinates corresponding to the selected indices can then be obtained
+ * using the `tf.gather operation`.  For example:
+ *   selected_indices = tf.image.non_max_suppression_v2(
+ *       boxes, scores, max_output_size, iou_threshold, score_threshold)
+ *   selected_boxes = tf.gather(boxes, selected_indices)
+ * This op also supports a Soft-NMS (with Gaussian weighting) mode (c.f.
+ * Bodla et al, https://arxiv.org/abs/1704.04503) where boxes reduce the score
+ * of other overlapping boxes instead of directly causing them to be pruned.
+ * To enable this Soft-NMS mode, set the `soft_nms_sigma` parameter to be
+ * larger than 0.
+ * 
+ * @param <T> data type for {@code selectedScores()} output
+ */
+@Operator(group = "image")
+public final class NonMaxSuppression<T extends TNumber> extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.image.NonMaxSuppression}
+   */
+  public static class Options {
+    
+    /**
+     * @param padToMaxOutputSize If true, the output `selected_indices` is padded to be of length
+     * `max_output_size`. Defaults to false.
+     */
+    public Options padToMaxOutputSize(Boolean padToMaxOutputSize) {
+      this.padToMaxOutputSize = padToMaxOutputSize;
+      return this;
+    }
+    
+    private Boolean padToMaxOutputSize;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new NonMaxSuppression operation.
+   * 
+   * @param scope current scope
+   * @param boxes A 2-D float tensor of shape `[num_boxes, 4]`.
+   * @param scores A 1-D float tensor of shape `[num_boxes]` representing a single
+   * score corresponding to each box (each row of boxes).
+   * @param maxOutputSize A scalar integer tensor representing the maximum number of
+   * boxes to be selected by non max suppression.
+   * @param iouThreshold A 0-D float tensor representing the threshold for deciding whether
+   * boxes overlap too much with respect to IOU.
+   * @param scoreThreshold A 0-D float tensor representing the threshold for deciding when to remove
+   * boxes based on score.
+   * @param softNmsSigma A 0-D float tensor representing the sigma parameter for Soft NMS; see Bodla et
+   * al (c.f. https://arxiv.org/abs/1704.04503).  When `soft_nms_sigma=0.0` (which
+   * is default), we fall back to standard (hard) NMS.
+   * @param options carries optional attributes values
+   * @return a new instance of NonMaxSuppression
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> NonMaxSuppression<T> create(Scope scope, Operand<T> boxes, Operand<T> scores, Operand<TInt32> maxOutputSize, Operand<T> iouThreshold, Operand<T> scoreThreshold, Operand<T> softNmsSigma, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("NonMaxSuppressionV5", scope.makeOpName("NonMaxSuppression"));
+    opBuilder.addInput(boxes.asOutput());
+    opBuilder.addInput(scores.asOutput());
+    opBuilder.addInput(maxOutputSize.asOutput());
+    opBuilder.addInput(iouThreshold.asOutput());
+    opBuilder.addInput(scoreThreshold.asOutput());
+    opBuilder.addInput(softNmsSigma.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.padToMaxOutputSize != null) {
+          opBuilder.setAttr("pad_to_max_output_size", opts.padToMaxOutputSize);
+        }
+      }
+    }
+    return new NonMaxSuppression<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param padToMaxOutputSize If true, the output `selected_indices` is padded to be of length
+   * `max_output_size`. Defaults to false.
+   */
+  public static Options padToMaxOutputSize(Boolean padToMaxOutputSize) {
+    return new Options().padToMaxOutputSize(padToMaxOutputSize);
+  }
+  
+  /**
+   * A 1-D integer tensor of shape `[M]` representing the selected
+   * indices from the boxes tensor, where `M <= max_output_size`.
+   */
+  public Output<TInt32> selectedIndices() {
+    return selectedIndices;
+  }
+  
+  /**
+   * A 1-D float tensor of shape `[M]` representing the corresponding
+   * scores for each selected box, where `M <= max_output_size`.  Scores only differ
+   * from corresponding input scores when using Soft NMS (i.e. when
+   * `soft_nms_sigma>0`)
+   */
+  public Output<T> selectedScores() {
+    return selectedScores;
+  }
+  
+  /**
+   * A 0-D integer tensor representing the number of valid elements in
+   * `selected_indices`, with the valid elements appearing first.
+   */
+  public Output<TInt32> validOutputs() {
+    return validOutputs;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "NonMaxSuppressionV5";
+  
+  private Output<TInt32> selectedIndices;
+  private Output<T> selectedScores;
+  private Output<TInt32> validOutputs;
+  
+  private NonMaxSuppression(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    selectedIndices = operation.output(outputIdx++);
+    selectedScores = operation.output(outputIdx++);
+    validOutputs = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/NonMaxSuppressionWithOverlaps.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/NonMaxSuppressionWithOverlaps.java
new file mode 100644
index 00000000000..073aee26ff9
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/NonMaxSuppressionWithOverlaps.java
@@ -0,0 +1,103 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.image;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.TInt32;
+
+/**
+ * Greedily selects a subset of bounding boxes in descending order of score,
+ * <p>
+ * pruning away boxes that have high overlaps
+ * with previously selected boxes.  Bounding boxes with score less than
+ * `score_threshold` are removed. N-by-n overlap values are supplied as square matrix,
+ * which allows for defining a custom overlap criterium (eg. intersection over union,
+ * intersection over area, etc.).
+ * <p>
+ * The output of this operation is a set of integers indexing into the input
+ * collection of bounding boxes representing the selected boxes.  The bounding
+ * box coordinates corresponding to the selected indices can then be obtained
+ * using the `tf.gather operation`.  For example:
+ * <p>
+ *   selected_indices = tf.image.non_max_suppression_with_overlaps(
+ *       overlaps, scores, max_output_size, overlap_threshold, score_threshold)
+ *   selected_boxes = tf.gather(boxes, selected_indices)
+ */
+@Operator(group = "image")
+public final class NonMaxSuppressionWithOverlaps extends RawOp implements Operand<TInt32> {
+  
+  /**
+   * Factory method to create a class wrapping a new NonMaxSuppressionWithOverlaps operation.
+   * 
+   * @param scope current scope
+   * @param overlaps A 2-D float tensor of shape `[num_boxes, num_boxes]` representing
+   * the n-by-n box overlap values.
+   * @param scores A 1-D float tensor of shape `[num_boxes]` representing a single
+   * score corresponding to each box (each row of boxes).
+   * @param maxOutputSize A scalar integer tensor representing the maximum number of
+   * boxes to be selected by non max suppression.
+   * @param overlapThreshold A 0-D float tensor representing the threshold for deciding whether
+   * boxes overlap too.
+   * @param scoreThreshold A 0-D float tensor representing the threshold for deciding when to remove
+   * boxes based on score.
+   * @return a new instance of NonMaxSuppressionWithOverlaps
+   */
+  @Endpoint(describeByClass = true)
+  public static NonMaxSuppressionWithOverlaps create(Scope scope, Operand<TFloat32> overlaps, Operand<TFloat32> scores, Operand<TInt32> maxOutputSize, Operand<TFloat32> overlapThreshold, Operand<TFloat32> scoreThreshold) {
+    OperationBuilder opBuilder = scope.env().opBuilder("NonMaxSuppressionWithOverlaps", scope.makeOpName("NonMaxSuppressionWithOverlaps"));
+    opBuilder.addInput(overlaps.asOutput());
+    opBuilder.addInput(scores.asOutput());
+    opBuilder.addInput(maxOutputSize.asOutput());
+    opBuilder.addInput(overlapThreshold.asOutput());
+    opBuilder.addInput(scoreThreshold.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new NonMaxSuppressionWithOverlaps(opBuilder.build());
+  }
+  
+  /**
+   * A 1-D integer tensor of shape `[M]` representing the selected
+   * indices from the boxes tensor, where `M <= max_output_size`.
+   */
+  public Output<TInt32> selectedIndices() {
+    return selectedIndices;
+  }
+  
+  @Override
+  public Output<TInt32> asOutput() {
+    return selectedIndices;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "NonMaxSuppressionWithOverlaps";
+  
+  private Output<TInt32> selectedIndices;
+  
+  private NonMaxSuppressionWithOverlaps(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    selectedIndices = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/QuantizedResizeBilinear.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/QuantizedResizeBilinear.java
new file mode 100644
index 00000000000..7eb6d8a3a92
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/QuantizedResizeBilinear.java
@@ -0,0 +1,153 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.image;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Resize quantized `images` to `size` using quantized bilinear interpolation.
+ * <p>
+ * Input images and output images must be quantized types.
+ * 
+ * @param <T> data type for {@code resizedImages()} output
+ */
+@Operator(group = "image")
+public final class QuantizedResizeBilinear<T extends TType> extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.image.QuantizedResizeBilinear}
+   */
+  public static class Options {
+    
+    /**
+     * @param alignCorners If true, the centers of the 4 corner pixels of the input and output tensors are
+     * aligned, preserving the values at the corner pixels. Defaults to false.
+     */
+    public Options alignCorners(Boolean alignCorners) {
+      this.alignCorners = alignCorners;
+      return this;
+    }
+    
+    /**
+     * @param halfPixelCenters 
+     */
+    public Options halfPixelCenters(Boolean halfPixelCenters) {
+      this.halfPixelCenters = halfPixelCenters;
+      return this;
+    }
+    
+    private Boolean alignCorners;
+    private Boolean halfPixelCenters;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new QuantizedResizeBilinear operation.
+   * 
+   * @param scope current scope
+   * @param images 4-D with shape `[batch, height, width, channels]`.
+   * @param size = A 1-D int32 Tensor of 2 elements: `new_height, new_width`.  The
+   * new size for the images.
+   * @param min 
+   * @param max 
+   * @param options carries optional attributes values
+   * @return a new instance of QuantizedResizeBilinear
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> QuantizedResizeBilinear<T> create(Scope scope, Operand<T> images, Operand<TInt32> size, Operand<TFloat32> min, Operand<TFloat32> max, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("QuantizedResizeBilinear", scope.makeOpName("QuantizedResizeBilinear"));
+    opBuilder.addInput(images.asOutput());
+    opBuilder.addInput(size.asOutput());
+    opBuilder.addInput(min.asOutput());
+    opBuilder.addInput(max.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.alignCorners != null) {
+          opBuilder.setAttr("align_corners", opts.alignCorners);
+        }
+        if (opts.halfPixelCenters != null) {
+          opBuilder.setAttr("half_pixel_centers", opts.halfPixelCenters);
+        }
+      }
+    }
+    return new QuantizedResizeBilinear<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param alignCorners If true, the centers of the 4 corner pixels of the input and output tensors are
+   * aligned, preserving the values at the corner pixels. Defaults to false.
+   */
+  public static Options alignCorners(Boolean alignCorners) {
+    return new Options().alignCorners(alignCorners);
+  }
+  
+  /**
+   * @param halfPixelCenters 
+   */
+  public static Options halfPixelCenters(Boolean halfPixelCenters) {
+    return new Options().halfPixelCenters(halfPixelCenters);
+  }
+  
+  /**
+   * 4-D with shape
+   * `[batch, new_height, new_width, channels]`.
+   */
+  public Output<T> resizedImages() {
+    return resizedImages;
+  }
+  
+  /**
+   */
+  public Output<TFloat32> outMin() {
+    return outMin;
+  }
+  
+  /**
+   */
+  public Output<TFloat32> outMax() {
+    return outMax;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "QuantizedResizeBilinear";
+  
+  private Output<T> resizedImages;
+  private Output<TFloat32> outMin;
+  private Output<TFloat32> outMax;
+  
+  private QuantizedResizeBilinear(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    resizedImages = operation.output(outputIdx++);
+    outMin = operation.output(outputIdx++);
+    outMax = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/RandomCrop.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/RandomCrop.java
new file mode 100644
index 00000000000..a5e2de121b1
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/RandomCrop.java
@@ -0,0 +1,142 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.image;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Randomly crop `image`.
+ * <p>
+ * `size` is a 1-D int64 tensor with 2 elements representing the crop height and
+ * width.  The values must be non negative.
+ * <p>
+ * This Op picks a random location in `image` and crops a `height` by `width`
+ * rectangle from that location.  The random location is picked so the cropped
+ * area will fit inside the original image.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator(group = "image")
+public final class RandomCrop<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.image.RandomCrop}
+   */
+  public static class Options {
+    
+    /**
+     * @param seed If either seed or seed2 are set to be non-zero, the random number
+     * generator is seeded by the given seed.  Otherwise, it is seeded by a
+     * random seed.
+     */
+    public Options seed(Long seed) {
+      this.seed = seed;
+      return this;
+    }
+    
+    /**
+     * @param seed2 An second seed to avoid seed collision.
+     */
+    public Options seed2(Long seed2) {
+      this.seed2 = seed2;
+      return this;
+    }
+    
+    private Long seed;
+    private Long seed2;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new RandomCrop operation.
+   * 
+   * @param scope current scope
+   * @param image 3-D of shape `[height, width, channels]`.
+   * @param size 1-D of length 2 containing: `crop_height`, `crop_width`..
+   * @param options carries optional attributes values
+   * @return a new instance of RandomCrop
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> RandomCrop<T> create(Scope scope, Operand<T> image, Operand<TInt64> size, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("RandomCrop", scope.makeOpName("RandomCrop"));
+    opBuilder.addInput(image.asOutput());
+    opBuilder.addInput(size.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.seed != null) {
+          opBuilder.setAttr("seed", opts.seed);
+        }
+        if (opts.seed2 != null) {
+          opBuilder.setAttr("seed2", opts.seed2);
+        }
+      }
+    }
+    return new RandomCrop<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param seed If either seed or seed2 are set to be non-zero, the random number
+   * generator is seeded by the given seed.  Otherwise, it is seeded by a
+   * random seed.
+   */
+  public static Options seed(Long seed) {
+    return new Options().seed(seed);
+  }
+  
+  /**
+   * @param seed2 An second seed to avoid seed collision.
+   */
+  public static Options seed2(Long seed2) {
+    return new Options().seed2(seed2);
+  }
+  
+  /**
+   * 3-D of shape `[crop_height, crop_width, channels].`
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "RandomCrop";
+  
+  private Output<T> output;
+  
+  private RandomCrop(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ResizeArea.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ResizeArea.java
new file mode 100644
index 00000000000..895bdfac587
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ResizeArea.java
@@ -0,0 +1,127 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.image;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Resize `images` to `size` using area interpolation.
+ * <p>
+ * Input images can be of different types but output images are always float.
+ * <p>
+ * The range of pixel values for the output image might be slightly different
+ * from the range for the input image because of limited numerical precision.
+ * To guarantee an output range, for example `[0.0, 1.0]`, apply
+ * `tf.clip_by_value` to the output.
+ * <p>
+ * Each output pixel is computed by first transforming the pixel's footprint into
+ * the input tensor and then averaging the pixels that intersect the footprint. An
+ * input pixel's contribution to the average is weighted by the fraction of its
+ * area that intersects the footprint.  This is the same as OpenCV's INTER_AREA.
+ */
+@Operator(group = "image")
+public final class ResizeArea extends RawOp implements Operand<TFloat32> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.image.ResizeArea}
+   */
+  public static class Options {
+    
+    /**
+     * @param alignCorners If true, the centers of the 4 corner pixels of the input and output tensors are
+     * aligned, preserving the values at the corner pixels. Defaults to false.
+     */
+    public Options alignCorners(Boolean alignCorners) {
+      this.alignCorners = alignCorners;
+      return this;
+    }
+    
+    private Boolean alignCorners;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new ResizeArea operation.
+   * 
+   * @param scope current scope
+   * @param images 4-D with shape `[batch, height, width, channels]`.
+   * @param size = A 1-D int32 Tensor of 2 elements: `new_height, new_width`.  The
+   * new size for the images.
+   * @param options carries optional attributes values
+   * @return a new instance of ResizeArea
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> ResizeArea create(Scope scope, Operand<T> images, Operand<TInt32> size, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ResizeArea", scope.makeOpName("ResizeArea"));
+    opBuilder.addInput(images.asOutput());
+    opBuilder.addInput(size.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.alignCorners != null) {
+          opBuilder.setAttr("align_corners", opts.alignCorners);
+        }
+      }
+    }
+    return new ResizeArea(opBuilder.build());
+  }
+  
+  /**
+   * @param alignCorners If true, the centers of the 4 corner pixels of the input and output tensors are
+   * aligned, preserving the values at the corner pixels. Defaults to false.
+   */
+  public static Options alignCorners(Boolean alignCorners) {
+    return new Options().alignCorners(alignCorners);
+  }
+  
+  /**
+   * 4-D with shape
+   * `[batch, new_height, new_width, channels]`.
+   */
+  public Output<TFloat32> resizedImages() {
+    return resizedImages;
+  }
+  
+  @Override
+  public Output<TFloat32> asOutput() {
+    return resizedImages;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ResizeArea";
+  
+  private Output<TFloat32> resizedImages;
+  
+  private ResizeArea(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    resizedImages = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ResizeBicubic.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ResizeBicubic.java
new file mode 100644
index 00000000000..7b3b9ffdcdd
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ResizeBicubic.java
@@ -0,0 +1,136 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.image;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Resize `images` to `size` using bicubic interpolation.
+ * <p>
+ * Input images can be of different types but output images are always float.
+ */
+@Operator(group = "image")
+public final class ResizeBicubic extends RawOp implements Operand<TFloat32> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.image.ResizeBicubic}
+   */
+  public static class Options {
+    
+    /**
+     * @param alignCorners If true, the centers of the 4 corner pixels of the input and output tensors are
+     * aligned, preserving the values at the corner pixels. Defaults to false.
+     */
+    public Options alignCorners(Boolean alignCorners) {
+      this.alignCorners = alignCorners;
+      return this;
+    }
+    
+    /**
+     * @param halfPixelCenters 
+     */
+    public Options halfPixelCenters(Boolean halfPixelCenters) {
+      this.halfPixelCenters = halfPixelCenters;
+      return this;
+    }
+    
+    private Boolean alignCorners;
+    private Boolean halfPixelCenters;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new ResizeBicubic operation.
+   * 
+   * @param scope current scope
+   * @param images 4-D with shape `[batch, height, width, channels]`.
+   * @param size = A 1-D int32 Tensor of 2 elements: `new_height, new_width`.  The
+   * new size for the images.
+   * @param options carries optional attributes values
+   * @return a new instance of ResizeBicubic
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> ResizeBicubic create(Scope scope, Operand<T> images, Operand<TInt32> size, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ResizeBicubic", scope.makeOpName("ResizeBicubic"));
+    opBuilder.addInput(images.asOutput());
+    opBuilder.addInput(size.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.alignCorners != null) {
+          opBuilder.setAttr("align_corners", opts.alignCorners);
+        }
+        if (opts.halfPixelCenters != null) {
+          opBuilder.setAttr("half_pixel_centers", opts.halfPixelCenters);
+        }
+      }
+    }
+    return new ResizeBicubic(opBuilder.build());
+  }
+  
+  /**
+   * @param alignCorners If true, the centers of the 4 corner pixels of the input and output tensors are
+   * aligned, preserving the values at the corner pixels. Defaults to false.
+   */
+  public static Options alignCorners(Boolean alignCorners) {
+    return new Options().alignCorners(alignCorners);
+  }
+  
+  /**
+   * @param halfPixelCenters 
+   */
+  public static Options halfPixelCenters(Boolean halfPixelCenters) {
+    return new Options().halfPixelCenters(halfPixelCenters);
+  }
+  
+  /**
+   * 4-D with shape
+   * `[batch, new_height, new_width, channels]`.
+   */
+  public Output<TFloat32> resizedImages() {
+    return resizedImages;
+  }
+  
+  @Override
+  public Output<TFloat32> asOutput() {
+    return resizedImages;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ResizeBicubic";
+  
+  private Output<TFloat32> resizedImages;
+  
+  private ResizeBicubic(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    resizedImages = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ResizeBicubicGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ResizeBicubicGrad.java
new file mode 100644
index 00000000000..e57cd749a72
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ResizeBicubicGrad.java
@@ -0,0 +1,135 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.image;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Computes the gradient of bicubic interpolation.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+public final class ResizeBicubicGrad<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.image.ResizeBicubicGrad}
+   */
+  public static class Options {
+    
+    /**
+     * @param alignCorners If true, the centers of the 4 corner pixels of the input and grad tensors are
+     * aligned. Defaults to false.
+     */
+    public Options alignCorners(Boolean alignCorners) {
+      this.alignCorners = alignCorners;
+      return this;
+    }
+    
+    /**
+     * @param halfPixelCenters 
+     */
+    public Options halfPixelCenters(Boolean halfPixelCenters) {
+      this.halfPixelCenters = halfPixelCenters;
+      return this;
+    }
+    
+    private Boolean alignCorners;
+    private Boolean halfPixelCenters;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new ResizeBicubicGrad operation.
+   * 
+   * @param scope current scope
+   * @param grads 4-D with shape `[batch, height, width, channels]`.
+   * @param originalImage 4-D with shape `[batch, orig_height, orig_width, channels]`,
+   * The image tensor that was resized.
+   * @param options carries optional attributes values
+   * @return a new instance of ResizeBicubicGrad
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> ResizeBicubicGrad<T> create(Scope scope, Operand<TFloat32> grads, Operand<T> originalImage, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ResizeBicubicGrad", scope.makeOpName("ResizeBicubicGrad"));
+    opBuilder.addInput(grads.asOutput());
+    opBuilder.addInput(originalImage.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.alignCorners != null) {
+          opBuilder.setAttr("align_corners", opts.alignCorners);
+        }
+        if (opts.halfPixelCenters != null) {
+          opBuilder.setAttr("half_pixel_centers", opts.halfPixelCenters);
+        }
+      }
+    }
+    return new ResizeBicubicGrad<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param alignCorners If true, the centers of the 4 corner pixels of the input and grad tensors are
+   * aligned. Defaults to false.
+   */
+  public static Options alignCorners(Boolean alignCorners) {
+    return new Options().alignCorners(alignCorners);
+  }
+  
+  /**
+   * @param halfPixelCenters 
+   */
+  public static Options halfPixelCenters(Boolean halfPixelCenters) {
+    return new Options().halfPixelCenters(halfPixelCenters);
+  }
+  
+  /**
+   * 4-D with shape `[batch, orig_height, orig_width, channels]`.
+   * Gradients with respect to the input image. Input image must have been
+   * float or double.
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ResizeBicubicGrad";
+  
+  private Output<T> output;
+  
+  private ResizeBicubicGrad(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ResizeBilinear.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ResizeBilinear.java
new file mode 100644
index 00000000000..8b3bba2c9ae
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ResizeBilinear.java
@@ -0,0 +1,136 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.image;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Resize `images` to `size` using bilinear interpolation.
+ * <p>
+ * Input images can be of different types but output images are always float.
+ */
+@Operator(group = "image")
+public final class ResizeBilinear extends RawOp implements Operand<TFloat32> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.image.ResizeBilinear}
+   */
+  public static class Options {
+    
+    /**
+     * @param alignCorners If true, the centers of the 4 corner pixels of the input and output tensors are
+     * aligned, preserving the values at the corner pixels. Defaults to false.
+     */
+    public Options alignCorners(Boolean alignCorners) {
+      this.alignCorners = alignCorners;
+      return this;
+    }
+    
+    /**
+     * @param halfPixelCenters 
+     */
+    public Options halfPixelCenters(Boolean halfPixelCenters) {
+      this.halfPixelCenters = halfPixelCenters;
+      return this;
+    }
+    
+    private Boolean alignCorners;
+    private Boolean halfPixelCenters;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new ResizeBilinear operation.
+   * 
+   * @param scope current scope
+   * @param images 4-D with shape `[batch, height, width, channels]`.
+   * @param size = A 1-D int32 Tensor of 2 elements: `new_height, new_width`.  The
+   * new size for the images.
+   * @param options carries optional attributes values
+   * @return a new instance of ResizeBilinear
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> ResizeBilinear create(Scope scope, Operand<T> images, Operand<TInt32> size, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ResizeBilinear", scope.makeOpName("ResizeBilinear"));
+    opBuilder.addInput(images.asOutput());
+    opBuilder.addInput(size.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.alignCorners != null) {
+          opBuilder.setAttr("align_corners", opts.alignCorners);
+        }
+        if (opts.halfPixelCenters != null) {
+          opBuilder.setAttr("half_pixel_centers", opts.halfPixelCenters);
+        }
+      }
+    }
+    return new ResizeBilinear(opBuilder.build());
+  }
+  
+  /**
+   * @param alignCorners If true, the centers of the 4 corner pixels of the input and output tensors are
+   * aligned, preserving the values at the corner pixels. Defaults to false.
+   */
+  public static Options alignCorners(Boolean alignCorners) {
+    return new Options().alignCorners(alignCorners);
+  }
+  
+  /**
+   * @param halfPixelCenters 
+   */
+  public static Options halfPixelCenters(Boolean halfPixelCenters) {
+    return new Options().halfPixelCenters(halfPixelCenters);
+  }
+  
+  /**
+   * 4-D with shape
+   * `[batch, new_height, new_width, channels]`.
+   */
+  public Output<TFloat32> resizedImages() {
+    return resizedImages;
+  }
+  
+  @Override
+  public Output<TFloat32> asOutput() {
+    return resizedImages;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ResizeBilinear";
+  
+  private Output<TFloat32> resizedImages;
+  
+  private ResizeBilinear(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    resizedImages = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ResizeBilinearGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ResizeBilinearGrad.java
new file mode 100644
index 00000000000..572f8693d4b
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ResizeBilinearGrad.java
@@ -0,0 +1,135 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.image;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Computes the gradient of bilinear interpolation.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+public final class ResizeBilinearGrad<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.image.ResizeBilinearGrad}
+   */
+  public static class Options {
+    
+    /**
+     * @param alignCorners If true, the centers of the 4 corner pixels of the input and grad tensors are
+     * aligned. Defaults to false.
+     */
+    public Options alignCorners(Boolean alignCorners) {
+      this.alignCorners = alignCorners;
+      return this;
+    }
+    
+    /**
+     * @param halfPixelCenters 
+     */
+    public Options halfPixelCenters(Boolean halfPixelCenters) {
+      this.halfPixelCenters = halfPixelCenters;
+      return this;
+    }
+    
+    private Boolean alignCorners;
+    private Boolean halfPixelCenters;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new ResizeBilinearGrad operation.
+   * 
+   * @param scope current scope
+   * @param grads 4-D with shape `[batch, height, width, channels]`.
+   * @param originalImage 4-D with shape `[batch, orig_height, orig_width, channels]`,
+   * The image tensor that was resized.
+   * @param options carries optional attributes values
+   * @return a new instance of ResizeBilinearGrad
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> ResizeBilinearGrad<T> create(Scope scope, Operand<TFloat32> grads, Operand<T> originalImage, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ResizeBilinearGrad", scope.makeOpName("ResizeBilinearGrad"));
+    opBuilder.addInput(grads.asOutput());
+    opBuilder.addInput(originalImage.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.alignCorners != null) {
+          opBuilder.setAttr("align_corners", opts.alignCorners);
+        }
+        if (opts.halfPixelCenters != null) {
+          opBuilder.setAttr("half_pixel_centers", opts.halfPixelCenters);
+        }
+      }
+    }
+    return new ResizeBilinearGrad<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param alignCorners If true, the centers of the 4 corner pixels of the input and grad tensors are
+   * aligned. Defaults to false.
+   */
+  public static Options alignCorners(Boolean alignCorners) {
+    return new Options().alignCorners(alignCorners);
+  }
+  
+  /**
+   * @param halfPixelCenters 
+   */
+  public static Options halfPixelCenters(Boolean halfPixelCenters) {
+    return new Options().halfPixelCenters(halfPixelCenters);
+  }
+  
+  /**
+   * 4-D with shape `[batch, orig_height, orig_width, channels]`.
+   * Gradients with respect to the input image. Input image must have been
+   * float or double.
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ResizeBilinearGrad";
+  
+  private Output<T> output;
+  
+  private ResizeBilinearGrad(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ResizeNearestNeighbor.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ResizeNearestNeighbor.java
new file mode 100644
index 00000000000..1da719fb569
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ResizeNearestNeighbor.java
@@ -0,0 +1,135 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.image;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Resize `images` to `size` using nearest neighbor interpolation.
+ * 
+ * @param <T> data type for {@code resizedImages()} output
+ */
+@Operator(group = "image")
+public final class ResizeNearestNeighbor<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.image.ResizeNearestNeighbor}
+   */
+  public static class Options {
+    
+    /**
+     * @param alignCorners If true, the centers of the 4 corner pixels of the input and output tensors are
+     * aligned, preserving the values at the corner pixels. Defaults to false.
+     */
+    public Options alignCorners(Boolean alignCorners) {
+      this.alignCorners = alignCorners;
+      return this;
+    }
+    
+    /**
+     * @param halfPixelCenters 
+     */
+    public Options halfPixelCenters(Boolean halfPixelCenters) {
+      this.halfPixelCenters = halfPixelCenters;
+      return this;
+    }
+    
+    private Boolean alignCorners;
+    private Boolean halfPixelCenters;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new ResizeNearestNeighbor operation.
+   * 
+   * @param scope current scope
+   * @param images 4-D with shape `[batch, height, width, channels]`.
+   * @param size = A 1-D int32 Tensor of 2 elements: `new_height, new_width`.  The
+   * new size for the images.
+   * @param options carries optional attributes values
+   * @return a new instance of ResizeNearestNeighbor
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> ResizeNearestNeighbor<T> create(Scope scope, Operand<T> images, Operand<TInt32> size, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ResizeNearestNeighbor", scope.makeOpName("ResizeNearestNeighbor"));
+    opBuilder.addInput(images.asOutput());
+    opBuilder.addInput(size.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.alignCorners != null) {
+          opBuilder.setAttr("align_corners", opts.alignCorners);
+        }
+        if (opts.halfPixelCenters != null) {
+          opBuilder.setAttr("half_pixel_centers", opts.halfPixelCenters);
+        }
+      }
+    }
+    return new ResizeNearestNeighbor<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param alignCorners If true, the centers of the 4 corner pixels of the input and output tensors are
+   * aligned, preserving the values at the corner pixels. Defaults to false.
+   */
+  public static Options alignCorners(Boolean alignCorners) {
+    return new Options().alignCorners(alignCorners);
+  }
+  
+  /**
+   * @param halfPixelCenters 
+   */
+  public static Options halfPixelCenters(Boolean halfPixelCenters) {
+    return new Options().halfPixelCenters(halfPixelCenters);
+  }
+  
+  /**
+   * 4-D with shape
+   * `[batch, new_height, new_width, channels]`.
+   */
+  public Output<T> resizedImages() {
+    return resizedImages;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return resizedImages;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ResizeNearestNeighbor";
+  
+  private Output<T> resizedImages;
+  
+  private ResizeNearestNeighbor(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    resizedImages = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ResizeNearestNeighborGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ResizeNearestNeighborGrad.java
new file mode 100644
index 00000000000..f0139dcca6c
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ResizeNearestNeighborGrad.java
@@ -0,0 +1,134 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.image;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Computes the gradient of nearest neighbor interpolation.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+public final class ResizeNearestNeighborGrad<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.image.ResizeNearestNeighborGrad}
+   */
+  public static class Options {
+    
+    /**
+     * @param alignCorners If true, the centers of the 4 corner pixels of the input and grad tensors are
+     * aligned. Defaults to false.
+     */
+    public Options alignCorners(Boolean alignCorners) {
+      this.alignCorners = alignCorners;
+      return this;
+    }
+    
+    /**
+     * @param halfPixelCenters 
+     */
+    public Options halfPixelCenters(Boolean halfPixelCenters) {
+      this.halfPixelCenters = halfPixelCenters;
+      return this;
+    }
+    
+    private Boolean alignCorners;
+    private Boolean halfPixelCenters;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new ResizeNearestNeighborGrad operation.
+   * 
+   * @param scope current scope
+   * @param grads 4-D with shape `[batch, height, width, channels]`.
+   * @param size = A 1-D int32 Tensor of 2 elements: `orig_height, orig_width`. The
+   * original input size.
+   * @param options carries optional attributes values
+   * @return a new instance of ResizeNearestNeighborGrad
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> ResizeNearestNeighborGrad<T> create(Scope scope, Operand<T> grads, Operand<TInt32> size, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ResizeNearestNeighborGrad", scope.makeOpName("ResizeNearestNeighborGrad"));
+    opBuilder.addInput(grads.asOutput());
+    opBuilder.addInput(size.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.alignCorners != null) {
+          opBuilder.setAttr("align_corners", opts.alignCorners);
+        }
+        if (opts.halfPixelCenters != null) {
+          opBuilder.setAttr("half_pixel_centers", opts.halfPixelCenters);
+        }
+      }
+    }
+    return new ResizeNearestNeighborGrad<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param alignCorners If true, the centers of the 4 corner pixels of the input and grad tensors are
+   * aligned. Defaults to false.
+   */
+  public static Options alignCorners(Boolean alignCorners) {
+    return new Options().alignCorners(alignCorners);
+  }
+  
+  /**
+   * @param halfPixelCenters 
+   */
+  public static Options halfPixelCenters(Boolean halfPixelCenters) {
+    return new Options().halfPixelCenters(halfPixelCenters);
+  }
+  
+  /**
+   * 4-D with shape `[batch, orig_height, orig_width, channels]`. Gradients
+   * with respect to the input image.
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ResizeNearestNeighborGrad";
+  
+  private Output<T> output;
+  
+  private ResizeNearestNeighborGrad(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/RgbToHsv.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/RgbToHsv.java
new file mode 100644
index 00000000000..0ac21df35a5
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/RgbToHsv.java
@@ -0,0 +1,95 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.image;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Converts one or more images from RGB to HSV.
+ * <p>
+ * Outputs a tensor of the same shape as the `images` tensor, containing the HSV
+ * value of the pixels. The output is only well defined if the value in `images`
+ * are in `[0,1]`.
+ * <p>
+ * `output[..., 0]` contains hue, `output[..., 1]` contains saturation, and
+ * `output[..., 2]` contains value. All HSV values are in `[0,1]`. A hue of 0
+ * corresponds to pure red, hue 1/3 is pure green, and 2/3 is pure blue.
+ * <p>
+ * Usage Example:
+ * <p>
+ * >>> blue_image = tf.stack([
+ * ...    tf.zeros([5,5]),
+ * ...    tf.zeros([5,5]),
+ * ...    tf.ones([5,5])],
+ * ...    axis=-1)
+ * >>> blue_hsv_image = tf.image.rgb_to_hsv(blue_image)
+ * >>> blue_hsv_image[0,0].numpy()
+ * array([0.6666667, 1. , 1. ], dtype=float32)
+ * 
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator(group = "image")
+public final class RgbToHsv<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new RgbToHsv operation.
+   * 
+   * @param scope current scope
+   * @param images 1-D or higher rank. RGB data to convert. Last dimension must be size 3.
+   * @return a new instance of RgbToHsv
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> RgbToHsv<T> create(Scope scope, Operand<T> images) {
+    OperationBuilder opBuilder = scope.env().opBuilder("RGBToHSV", scope.makeOpName("RgbToHsv"));
+    opBuilder.addInput(images.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new RgbToHsv<T>(opBuilder.build());
+  }
+  
+  /**
+   * `images` converted to HSV.
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "RGBToHSV";
+  
+  private Output<T> output;
+  
+  private RgbToHsv(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/SampleDistortedBoundingBox.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/SampleDistortedBoundingBox.java
new file mode 100644
index 00000000000..583da52ce0e
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/SampleDistortedBoundingBox.java
@@ -0,0 +1,291 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.image;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Generate a single randomly distorted bounding box for an image.
+ * <p>
+ * Bounding box annotations are often supplied in addition to ground-truth labels
+ * in image recognition or object localization tasks. A common technique for
+ * training such a system is to randomly distort an image while preserving
+ * its content, i.e. <i>data augmentation</i>. This Op outputs a randomly distorted
+ * localization of an object, i.e. bounding box, given an `image_size`,
+ * `bounding_boxes` and a series of constraints.
+ * <p>
+ * The output of this Op is a single bounding box that may be used to crop the
+ * original image. The output is returned as 3 tensors: `begin`, `size` and
+ * `bboxes`. The first 2 tensors can be fed directly into `tf.slice` to crop the
+ * image. The latter may be supplied to `tf.image.draw_bounding_boxes` to visualize
+ * what the bounding box looks like.
+ * <p>
+ * Bounding boxes are supplied and returned as `[y_min, x_min, y_max, x_max]`. The
+ * bounding box coordinates are floats in `[0.0, 1.0]` relative to the width and
+ * height of the underlying image.
+ * <p>
+ * For example,
+ * <pre>{@code
+ *     # Generate a single distorted bounding box.
+ *     begin, size, bbox_for_draw = tf.image.sample_distorted_bounding_box(
+ *         tf.shape(image),
+ *         bounding_boxes=bounding_boxes)
+ * 
+ *     # Draw the bounding box in an image summary.
+ *     image_with_box = tf.image.draw_bounding_boxes(tf.expand_dims(image, 0),
+ *                                                   bbox_for_draw)
+ *     tf.summary.image('images_with_box', image_with_box)
+ * 
+ *     # Employ the bounding box to distort the image.
+ *     distorted_image = tf.slice(image, begin, size)
+ * }</pre>
+ * Note that if no bounding box information is available, setting
+ * `use_image_if_no_bounding_boxes = true` will assume there is a single implicit
+ * bounding box covering the whole image. If `use_image_if_no_bounding_boxes` is
+ * false and no bounding boxes are supplied, an error is raised.
+ * 
+ * @param <T> data type for {@code begin()} output
+ */
+@Operator(group = "image")
+public final class SampleDistortedBoundingBox<T extends TNumber> extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.image.SampleDistortedBoundingBox}
+   */
+  public static class Options {
+    
+    /**
+     * @param seed If either `seed` or `seed2` are set to non-zero, the random number
+     * generator is seeded by the given `seed`.  Otherwise, it is seeded by a random
+     * seed.
+     */
+    public Options seed(Long seed) {
+      this.seed = seed;
+      return this;
+    }
+    
+    /**
+     * @param seed2 A second seed to avoid seed collision.
+     */
+    public Options seed2(Long seed2) {
+      this.seed2 = seed2;
+      return this;
+    }
+    
+    /**
+     * @param aspectRatioRange The cropped area of the image must have an aspect ratio =
+     * width / height within this range.
+     */
+    public Options aspectRatioRange(List<Float> aspectRatioRange) {
+      this.aspectRatioRange = aspectRatioRange;
+      return this;
+    }
+    
+    /**
+     * @param areaRange The cropped area of the image must contain a fraction of the
+     * supplied image within this range.
+     */
+    public Options areaRange(List<Float> areaRange) {
+      this.areaRange = areaRange;
+      return this;
+    }
+    
+    /**
+     * @param maxAttempts Number of attempts at generating a cropped region of the image
+     * of the specified constraints. After `max_attempts` failures, return the entire
+     * image.
+     */
+    public Options maxAttempts(Long maxAttempts) {
+      this.maxAttempts = maxAttempts;
+      return this;
+    }
+    
+    /**
+     * @param useImageIfNoBoundingBoxes Controls behavior if no bounding boxes supplied.
+     * If true, assume an implicit bounding box covering the whole input. If false,
+     * raise an error.
+     */
+    public Options useImageIfNoBoundingBoxes(Boolean useImageIfNoBoundingBoxes) {
+      this.useImageIfNoBoundingBoxes = useImageIfNoBoundingBoxes;
+      return this;
+    }
+    
+    private Long seed;
+    private Long seed2;
+    private List<Float> aspectRatioRange;
+    private List<Float> areaRange;
+    private Long maxAttempts;
+    private Boolean useImageIfNoBoundingBoxes;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new SampleDistortedBoundingBox operation.
+   * 
+   * @param scope current scope
+   * @param imageSize 1-D, containing `[height, width, channels]`.
+   * @param boundingBoxes 3-D with shape `[batch, N, 4]` describing the N bounding boxes
+   * associated with the image.
+   * @param minObjectCovered The cropped area of the image must contain at least this
+   * fraction of any bounding box supplied. The value of this parameter should be
+   * non-negative. In the case of 0, the cropped area does not need to overlap
+   * any of the bounding boxes supplied.
+   * @param options carries optional attributes values
+   * @return a new instance of SampleDistortedBoundingBox
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> SampleDistortedBoundingBox<T> create(Scope scope, Operand<T> imageSize, Operand<TFloat32> boundingBoxes, Operand<TFloat32> minObjectCovered, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("SampleDistortedBoundingBoxV2", scope.makeOpName("SampleDistortedBoundingBox"));
+    opBuilder.addInput(imageSize.asOutput());
+    opBuilder.addInput(boundingBoxes.asOutput());
+    opBuilder.addInput(minObjectCovered.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.seed != null) {
+          opBuilder.setAttr("seed", opts.seed);
+        }
+        if (opts.seed2 != null) {
+          opBuilder.setAttr("seed2", opts.seed2);
+        }
+        if (opts.aspectRatioRange != null) {
+          float[] aspectRatioRangeArray = new float[opts.aspectRatioRange.size()];
+          for (int i = 0; i < aspectRatioRangeArray.length; ++i) {
+            aspectRatioRangeArray[i] = opts.aspectRatioRange.get(i);
+          }
+          opBuilder.setAttr("aspect_ratio_range", aspectRatioRangeArray);
+        }
+        if (opts.areaRange != null) {
+          float[] areaRangeArray = new float[opts.areaRange.size()];
+          for (int i = 0; i < areaRangeArray.length; ++i) {
+            areaRangeArray[i] = opts.areaRange.get(i);
+          }
+          opBuilder.setAttr("area_range", areaRangeArray);
+        }
+        if (opts.maxAttempts != null) {
+          opBuilder.setAttr("max_attempts", opts.maxAttempts);
+        }
+        if (opts.useImageIfNoBoundingBoxes != null) {
+          opBuilder.setAttr("use_image_if_no_bounding_boxes", opts.useImageIfNoBoundingBoxes);
+        }
+      }
+    }
+    return new SampleDistortedBoundingBox<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param seed If either `seed` or `seed2` are set to non-zero, the random number
+   * generator is seeded by the given `seed`.  Otherwise, it is seeded by a random
+   * seed.
+   */
+  public static Options seed(Long seed) {
+    return new Options().seed(seed);
+  }
+  
+  /**
+   * @param seed2 A second seed to avoid seed collision.
+   */
+  public static Options seed2(Long seed2) {
+    return new Options().seed2(seed2);
+  }
+  
+  /**
+   * @param aspectRatioRange The cropped area of the image must have an aspect ratio =
+   * width / height within this range.
+   */
+  public static Options aspectRatioRange(List<Float> aspectRatioRange) {
+    return new Options().aspectRatioRange(aspectRatioRange);
+  }
+  
+  /**
+   * @param areaRange The cropped area of the image must contain a fraction of the
+   * supplied image within this range.
+   */
+  public static Options areaRange(List<Float> areaRange) {
+    return new Options().areaRange(areaRange);
+  }
+  
+  /**
+   * @param maxAttempts Number of attempts at generating a cropped region of the image
+   * of the specified constraints. After `max_attempts` failures, return the entire
+   * image.
+   */
+  public static Options maxAttempts(Long maxAttempts) {
+    return new Options().maxAttempts(maxAttempts);
+  }
+  
+  /**
+   * @param useImageIfNoBoundingBoxes Controls behavior if no bounding boxes supplied.
+   * If true, assume an implicit bounding box covering the whole input. If false,
+   * raise an error.
+   */
+  public static Options useImageIfNoBoundingBoxes(Boolean useImageIfNoBoundingBoxes) {
+    return new Options().useImageIfNoBoundingBoxes(useImageIfNoBoundingBoxes);
+  }
+  
+  /**
+   * 1-D, containing `[offset_height, offset_width, 0]`. Provide as input to
+   * `tf.slice`.
+   */
+  public Output<T> begin() {
+    return begin;
+  }
+  
+  /**
+   * 1-D, containing `[target_height, target_width, -1]`. Provide as input to
+   * `tf.slice`.
+   */
+  public Output<T> size() {
+    return size;
+  }
+  
+  /**
+   * 3-D with shape `[1, 1, 4]` containing the distorted bounding box.
+   * Provide as input to `tf.image.draw_bounding_boxes`.
+   */
+  public Output<TFloat32> bboxes() {
+    return bboxes;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "SampleDistortedBoundingBoxV2";
+  
+  private Output<T> begin;
+  private Output<T> size;
+  private Output<TFloat32> bboxes;
+  
+  private SampleDistortedBoundingBox(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    begin = operation.output(outputIdx++);
+    size = operation.output(outputIdx++);
+    bboxes = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ScaleAndTranslate.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ScaleAndTranslate.java
new file mode 100644
index 00000000000..cf52ddc87d4
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ScaleAndTranslate.java
@@ -0,0 +1,132 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.image;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ */
+@Operator(group = "image")
+public final class ScaleAndTranslate extends RawOp implements Operand<TFloat32> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.image.ScaleAndTranslate}
+   */
+  public static class Options {
+    
+    /**
+     * @param kernelType 
+     */
+    public Options kernelType(String kernelType) {
+      this.kernelType = kernelType;
+      return this;
+    }
+    
+    /**
+     * @param antialias 
+     */
+    public Options antialias(Boolean antialias) {
+      this.antialias = antialias;
+      return this;
+    }
+    
+    private String kernelType;
+    private Boolean antialias;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new ScaleAndTranslate operation.
+   * 
+   * @param scope current scope
+   * @param images 
+   * @param size 
+   * @param scale 
+   * @param translation 
+   * @param options carries optional attributes values
+   * @return a new instance of ScaleAndTranslate
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> ScaleAndTranslate create(Scope scope, Operand<T> images, Operand<TInt32> size, Operand<TFloat32> scale, Operand<TFloat32> translation, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ScaleAndTranslate", scope.makeOpName("ScaleAndTranslate"));
+    opBuilder.addInput(images.asOutput());
+    opBuilder.addInput(size.asOutput());
+    opBuilder.addInput(scale.asOutput());
+    opBuilder.addInput(translation.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.kernelType != null) {
+          opBuilder.setAttr("kernel_type", opts.kernelType);
+        }
+        if (opts.antialias != null) {
+          opBuilder.setAttr("antialias", opts.antialias);
+        }
+      }
+    }
+    return new ScaleAndTranslate(opBuilder.build());
+  }
+  
+  /**
+   * @param kernelType 
+   */
+  public static Options kernelType(String kernelType) {
+    return new Options().kernelType(kernelType);
+  }
+  
+  /**
+   * @param antialias 
+   */
+  public static Options antialias(Boolean antialias) {
+    return new Options().antialias(antialias);
+  }
+  
+  /**
+   */
+  public Output<TFloat32> resizedImages() {
+    return resizedImages;
+  }
+  
+  @Override
+  public Output<TFloat32> asOutput() {
+    return resizedImages;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ScaleAndTranslate";
+  
+  private Output<TFloat32> resizedImages;
+  
+  private ScaleAndTranslate(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    resizedImages = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ScaleAndTranslateGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ScaleAndTranslateGrad.java
new file mode 100644
index 00000000000..79d6c35a46e
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/image/ScaleAndTranslateGrad.java
@@ -0,0 +1,131 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.image;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * @param <T> data type for {@code output()} output
+ */
+public final class ScaleAndTranslateGrad<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.image.ScaleAndTranslateGrad}
+   */
+  public static class Options {
+    
+    /**
+     * @param kernelType 
+     */
+    public Options kernelType(String kernelType) {
+      this.kernelType = kernelType;
+      return this;
+    }
+    
+    /**
+     * @param antialias 
+     */
+    public Options antialias(Boolean antialias) {
+      this.antialias = antialias;
+      return this;
+    }
+    
+    private String kernelType;
+    private Boolean antialias;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new ScaleAndTranslateGrad operation.
+   * 
+   * @param scope current scope
+   * @param grads 
+   * @param originalImage 
+   * @param scale 
+   * @param translation 
+   * @param options carries optional attributes values
+   * @return a new instance of ScaleAndTranslateGrad
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> ScaleAndTranslateGrad<T> create(Scope scope, Operand<T> grads, Operand<T> originalImage, Operand<TFloat32> scale, Operand<TFloat32> translation, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ScaleAndTranslateGrad", scope.makeOpName("ScaleAndTranslateGrad"));
+    opBuilder.addInput(grads.asOutput());
+    opBuilder.addInput(originalImage.asOutput());
+    opBuilder.addInput(scale.asOutput());
+    opBuilder.addInput(translation.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.kernelType != null) {
+          opBuilder.setAttr("kernel_type", opts.kernelType);
+        }
+        if (opts.antialias != null) {
+          opBuilder.setAttr("antialias", opts.antialias);
+        }
+      }
+    }
+    return new ScaleAndTranslateGrad<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param kernelType 
+   */
+  public static Options kernelType(String kernelType) {
+    return new Options().kernelType(kernelType);
+  }
+  
+  /**
+   * @param antialias 
+   */
+  public static Options antialias(Boolean antialias) {
+    return new Options().antialias(antialias);
+  }
+  
+  /**
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ScaleAndTranslateGrad";
+  
+  private Output<T> output;
+  
+  private ScaleAndTranslateGrad(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/DecodeBase64.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/DecodeBase64.java
new file mode 100644
index 00000000000..db026fd820c
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/DecodeBase64.java
@@ -0,0 +1,76 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.io;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TString;
+
+/**
+ * Decode web-safe base64-encoded strings.
+ * <p>
+ * Input may or may not have padding at the end. See EncodeBase64 for padding.
+ * Web-safe means that input must use - and _ instead of + and /.
+ */
+@Operator(group = "io")
+public final class DecodeBase64 extends RawOp implements Operand<TString> {
+  
+  /**
+   * Factory method to create a class wrapping a new DecodeBase64 operation.
+   * 
+   * @param scope current scope
+   * @param input Base64 strings to decode.
+   * @return a new instance of DecodeBase64
+   */
+  @Endpoint(describeByClass = true)
+  public static DecodeBase64 create(Scope scope, Operand<TString> input) {
+    OperationBuilder opBuilder = scope.env().opBuilder("DecodeBase64", scope.makeOpName("DecodeBase64"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new DecodeBase64(opBuilder.build());
+  }
+  
+  /**
+   * Decoded strings.
+   */
+  public Output<TString> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<TString> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "DecodeBase64";
+  
+  private Output<TString> output;
+  
+  private DecodeBase64(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/DecodeCompressed.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/DecodeCompressed.java
new file mode 100644
index 00000000000..745dba6d063
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/DecodeCompressed.java
@@ -0,0 +1,117 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.io;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TString;
+
+/**
+ * Decompress strings.
+ * <p>
+ * This op decompresses each element of the `bytes` input `Tensor`, which
+ * is assumed to be compressed using the given `compression_type`.
+ * <p>
+ * The `output` is a string `Tensor` of the same shape as `bytes`,
+ * each element containing the decompressed data from the corresponding
+ * element in `bytes`.
+ */
+@Operator(group = "io")
+public final class DecodeCompressed extends RawOp implements Operand<TString> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.io.DecodeCompressed}
+   */
+  public static class Options {
+    
+    /**
+     * @param compressionType A scalar containing either (i) the empty string (no
+     * compression), (ii) "ZLIB", or (iii) "GZIP".
+     */
+    public Options compressionType(String compressionType) {
+      this.compressionType = compressionType;
+      return this;
+    }
+    
+    private String compressionType;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new DecodeCompressed operation.
+   * 
+   * @param scope current scope
+   * @param bytes A Tensor of string which is compressed.
+   * @param options carries optional attributes values
+   * @return a new instance of DecodeCompressed
+   */
+  @Endpoint(describeByClass = true)
+  public static DecodeCompressed create(Scope scope, Operand<TString> bytes, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("DecodeCompressed", scope.makeOpName("DecodeCompressed"));
+    opBuilder.addInput(bytes.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.compressionType != null) {
+          opBuilder.setAttr("compression_type", opts.compressionType);
+        }
+      }
+    }
+    return new DecodeCompressed(opBuilder.build());
+  }
+  
+  /**
+   * @param compressionType A scalar containing either (i) the empty string (no
+   * compression), (ii) "ZLIB", or (iii) "GZIP".
+   */
+  public static Options compressionType(String compressionType) {
+    return new Options().compressionType(compressionType);
+  }
+  
+  /**
+   * A Tensor with the same shape as input `bytes`, uncompressed
+   * from bytes.
+   */
+  public Output<TString> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<TString> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "DecodeCompressed";
+  
+  private Output<TString> output;
+  
+  private DecodeCompressed(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/DecodeCsv.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/DecodeCsv.java
new file mode 100644
index 00000000000..6b417cb8717
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/DecodeCsv.java
@@ -0,0 +1,189 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.io;
+
+import java.util.Arrays;
+import java.util.Iterator;
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.Operands;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TString;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Convert CSV records to tensors. Each column maps to one tensor.
+ * <p>
+ * RFC 4180 format is expected for the CSV records.
+ * (https://tools.ietf.org/html/rfc4180)
+ * Note that we allow leading and trailing spaces with int or float field.
+ */
+@Operator(group = "io")
+public final class DecodeCsv extends RawOp implements Iterable<Operand<TType>> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.io.DecodeCsv}
+   */
+  public static class Options {
+    
+    /**
+     * @param fieldDelim char delimiter to separate fields in a record.
+     */
+    public Options fieldDelim(String fieldDelim) {
+      this.fieldDelim = fieldDelim;
+      return this;
+    }
+    
+    /**
+     * @param useQuoteDelim If false, treats double quotation marks as regular
+     * characters inside of the string fields (ignoring RFC 4180, Section 2,
+     * Bullet 5).
+     */
+    public Options useQuoteDelim(Boolean useQuoteDelim) {
+      this.useQuoteDelim = useQuoteDelim;
+      return this;
+    }
+    
+    /**
+     * @param naValue Additional string to recognize as NA/NaN.
+     */
+    public Options naValue(String naValue) {
+      this.naValue = naValue;
+      return this;
+    }
+    
+    /**
+     * @param selectCols 
+     */
+    public Options selectCols(List<Long> selectCols) {
+      this.selectCols = selectCols;
+      return this;
+    }
+    
+    private String fieldDelim;
+    private Boolean useQuoteDelim;
+    private String naValue;
+    private List<Long> selectCols;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new DecodeCsv operation.
+   * 
+   * @param scope current scope
+   * @param records Each string is a record/row in the csv and all records should have
+   * the same format.
+   * @param recordDefaults One tensor per column of the input record, with either a
+   * scalar default value for that column or an empty vector if the column is
+   * required.
+   * @param options carries optional attributes values
+   * @return a new instance of DecodeCsv
+   */
+  @Endpoint(describeByClass = true)
+  public static DecodeCsv create(Scope scope, Operand<TString> records, Iterable<Operand<?>> recordDefaults, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("DecodeCSV", scope.makeOpName("DecodeCsv"));
+    opBuilder.addInput(records.asOutput());
+    opBuilder.addInputList(Operands.asOutputs(recordDefaults));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.fieldDelim != null) {
+          opBuilder.setAttr("field_delim", opts.fieldDelim);
+        }
+        if (opts.useQuoteDelim != null) {
+          opBuilder.setAttr("use_quote_delim", opts.useQuoteDelim);
+        }
+        if (opts.naValue != null) {
+          opBuilder.setAttr("na_value", opts.naValue);
+        }
+        if (opts.selectCols != null) {
+          long[] selectColsArray = new long[opts.selectCols.size()];
+          for (int i = 0; i < selectColsArray.length; ++i) {
+            selectColsArray[i] = opts.selectCols.get(i);
+          }
+          opBuilder.setAttr("select_cols", selectColsArray);
+        }
+      }
+    }
+    return new DecodeCsv(opBuilder.build());
+  }
+  
+  /**
+   * @param fieldDelim char delimiter to separate fields in a record.
+   */
+  public static Options fieldDelim(String fieldDelim) {
+    return new Options().fieldDelim(fieldDelim);
+  }
+  
+  /**
+   * @param useQuoteDelim If false, treats double quotation marks as regular
+   * characters inside of the string fields (ignoring RFC 4180, Section 2,
+   * Bullet 5).
+   */
+  public static Options useQuoteDelim(Boolean useQuoteDelim) {
+    return new Options().useQuoteDelim(useQuoteDelim);
+  }
+  
+  /**
+   * @param naValue Additional string to recognize as NA/NaN.
+   */
+  public static Options naValue(String naValue) {
+    return new Options().naValue(naValue);
+  }
+  
+  /**
+   * @param selectCols 
+   */
+  public static Options selectCols(List<Long> selectCols) {
+    return new Options().selectCols(selectCols);
+  }
+  
+  /**
+   * Each tensor will have the same shape as records.
+   */
+  public List<Output<?>> output() {
+    return output;
+  }
+  
+  @Override
+  @SuppressWarnings({"rawtypes", "unchecked"})
+  public Iterator<Operand<TType>> iterator() {
+    return (Iterator) output.iterator();
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "DecodeCSV";
+  
+  private List<Output<?>> output;
+  
+  private DecodeCsv(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    int outputLength = operation.outputListLength("output");
+    output = Arrays.asList(operation.outputList(outputIdx, outputLength));
+    outputIdx += outputLength;
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/DecodeJsonExample.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/DecodeJsonExample.java
new file mode 100644
index 00000000000..f3da58ca98d
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/DecodeJsonExample.java
@@ -0,0 +1,82 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.io;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TString;
+
+/**
+ * Convert JSON-encoded Example records to binary protocol buffer strings.
+ * <p>
+ * This op translates a tensor containing Example records, encoded using
+ * the [standard JSON
+ * mapping](https://developers.google.com/protocol-buffers/docs/proto3#json),
+ * into a tensor containing the same records encoded as binary protocol
+ * buffers. The resulting tensor can then be fed to any of the other
+ * Example-parsing ops.
+ */
+@Operator(group = "io")
+public final class DecodeJsonExample extends RawOp implements Operand<TString> {
+  
+  /**
+   * Factory method to create a class wrapping a new DecodeJsonExample operation.
+   * 
+   * @param scope current scope
+   * @param jsonExamples Each string is a JSON object serialized according to the JSON
+   * mapping of the Example proto.
+   * @return a new instance of DecodeJsonExample
+   */
+  @Endpoint(describeByClass = true)
+  public static DecodeJsonExample create(Scope scope, Operand<TString> jsonExamples) {
+    OperationBuilder opBuilder = scope.env().opBuilder("DecodeJSONExample", scope.makeOpName("DecodeJsonExample"));
+    opBuilder.addInput(jsonExamples.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new DecodeJsonExample(opBuilder.build());
+  }
+  
+  /**
+   * Each string is a binary Example protocol buffer corresponding
+   * to the respective element of `json_examples`.
+   */
+  public Output<TString> binaryExamples() {
+    return binaryExamples;
+  }
+  
+  @Override
+  public Output<TString> asOutput() {
+    return binaryExamples;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "DecodeJSONExample";
+  
+  private Output<TString> binaryExamples;
+  
+  private DecodeJsonExample(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    binaryExamples = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/DecodePaddedRaw.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/DecodePaddedRaw.java
new file mode 100644
index 00000000000..87a4ed8300c
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/DecodePaddedRaw.java
@@ -0,0 +1,120 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.io;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.TString;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Reinterpret the bytes of a string as a vector of numbers.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator(group = "io")
+public final class DecodePaddedRaw<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.io.DecodePaddedRaw}
+   */
+  public static class Options {
+    
+    /**
+     * @param littleEndian Whether the input `input_bytes` is in little-endian order. Ignored for
+     * `out_type` values that are stored in a single byte, like `uint8`
+     */
+    public Options littleEndian(Boolean littleEndian) {
+      this.littleEndian = littleEndian;
+      return this;
+    }
+    
+    private Boolean littleEndian;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new DecodePaddedRaw operation.
+   * 
+   * @param scope current scope
+   * @param inputBytes Tensor of string to be decoded.
+   * @param fixedLength Length in bytes for each element of the decoded output. Must be a multiple
+   * of the size of the output type.
+   * @param outType 
+   * @param options carries optional attributes values
+   * @return a new instance of DecodePaddedRaw
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> DecodePaddedRaw<T> create(Scope scope, Operand<TString> inputBytes, Operand<TInt32> fixedLength, Class<T> outType, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("DecodePaddedRaw", scope.makeOpName("DecodePaddedRaw"));
+    opBuilder.addInput(inputBytes.asOutput());
+    opBuilder.addInput(fixedLength.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("out_type", outType);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.littleEndian != null) {
+          opBuilder.setAttr("little_endian", opts.littleEndian);
+        }
+      }
+    }
+    return new DecodePaddedRaw<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param littleEndian Whether the input `input_bytes` is in little-endian order. Ignored for
+   * `out_type` values that are stored in a single byte, like `uint8`
+   */
+  public static Options littleEndian(Boolean littleEndian) {
+    return new Options().littleEndian(littleEndian);
+  }
+  
+  /**
+   * A Tensor with one more dimension than the input `bytes`. The added dimension
+   * will have size equal to the length of the elements of `bytes` divided by the
+   * number of bytes to represent `out_type`.
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "DecodePaddedRaw";
+  
+  private Output<T> output;
+  
+  private DecodePaddedRaw(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/DecodeRaw.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/DecodeRaw.java
new file mode 100644
index 00000000000..24b013325f3
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/DecodeRaw.java
@@ -0,0 +1,118 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.io;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TString;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Reinterpret the bytes of a string as a vector of numbers.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator(group = "io")
+public final class DecodeRaw<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.io.DecodeRaw}
+   */
+  public static class Options {
+    
+    /**
+     * @param littleEndian Whether the input `bytes` are in little-endian order.
+     * Ignored for `out_type` values that are stored in a single byte like
+     * `uint8`.
+     */
+    public Options littleEndian(Boolean littleEndian) {
+      this.littleEndian = littleEndian;
+      return this;
+    }
+    
+    private Boolean littleEndian;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new DecodeRaw operation.
+   * 
+   * @param scope current scope
+   * @param bytes All the elements must have the same length.
+   * @param outType 
+   * @param options carries optional attributes values
+   * @return a new instance of DecodeRaw
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> DecodeRaw<T> create(Scope scope, Operand<TString> bytes, Class<T> outType, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("DecodeRaw", scope.makeOpName("DecodeRaw"));
+    opBuilder.addInput(bytes.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("out_type", outType);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.littleEndian != null) {
+          opBuilder.setAttr("little_endian", opts.littleEndian);
+        }
+      }
+    }
+    return new DecodeRaw<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param littleEndian Whether the input `bytes` are in little-endian order.
+   * Ignored for `out_type` values that are stored in a single byte like
+   * `uint8`.
+   */
+  public static Options littleEndian(Boolean littleEndian) {
+    return new Options().littleEndian(littleEndian);
+  }
+  
+  /**
+   * A Tensor with one more dimension than the input `bytes`.  The
+   * added dimension will have size equal to the length of the elements
+   * of `bytes` divided by the number of bytes to represent `out_type`.
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "DecodeRaw";
+  
+  private Output<T> output;
+  
+  private DecodeRaw(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/DeserializeManySparse.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/DeserializeManySparse.java
new file mode 100644
index 00000000000..2a00c03b6a2
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/DeserializeManySparse.java
@@ -0,0 +1,132 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.io;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.TString;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Deserialize and concatenate `SparseTensors` from a serialized minibatch.
+ * <p>
+ * The input `serialized_sparse` must be a string matrix of shape `[N x 3]` where
+ * `N` is the minibatch size and the rows correspond to packed outputs of
+ * `SerializeSparse`.  The ranks of the original `SparseTensor` objects
+ * must all match.  When the final `SparseTensor` is created, it has rank one
+ * higher than the ranks of the incoming `SparseTensor` objects
+ * (they have been concatenated along a new row dimension).
+ * <p>
+ * The output `SparseTensor` object's shape values for all dimensions but the
+ * first are the max across the input `SparseTensor` objects' shape values
+ * for the corresponding dimensions.  Its first shape value is `N`, the minibatch
+ * size.
+ * <p>
+ * The input `SparseTensor` objects' indices are assumed ordered in
+ * standard lexicographic order.  If this is not the case, after this
+ * step run `SparseReorder` to restore index ordering.
+ * <p>
+ * For example, if the serialized input is a `[2 x 3]` matrix representing two
+ * original `SparseTensor` objects:
+ * <p>
+ *     index = [ 0]
+ *             [10]
+ *             [20]
+ *     values = [1, 2, 3]
+ *     shape = [50]
+ * <p>
+ * and
+ * <p>
+ *     index = [ 2]
+ *             [10]
+ *     values = [4, 5]
+ *     shape = [30]
+ * <p>
+ * then the final deserialized `SparseTensor` will be:
+ * <p>
+ *     index = [0  0]
+ *             [0 10]
+ *             [0 20]
+ *             [1  2]
+ *             [1 10]
+ *     values = [1, 2, 3, 4, 5]
+ *     shape = [2 50]
+ * 
+ * @param <T> data type for {@code sparseValues()} output
+ */
+@Operator(group = "io")
+public final class DeserializeManySparse<T extends TType> extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new DeserializeManySparse operation.
+   * 
+   * @param scope current scope
+   * @param serializedSparse 2-D, The `N` serialized `SparseTensor` objects.
+   * Must have 3 columns.
+   * @param dtype The `dtype` of the serialized `SparseTensor` objects.
+   * @return a new instance of DeserializeManySparse
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> DeserializeManySparse<T> create(Scope scope, Operand<TString> serializedSparse, Class<T> dtype) {
+    OperationBuilder opBuilder = scope.env().opBuilder("DeserializeManySparse", scope.makeOpName("DeserializeManySparse"));
+    opBuilder.addInput(serializedSparse.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("dtype", dtype);
+    return new DeserializeManySparse<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<TInt64> sparseIndices() {
+    return sparseIndices;
+  }
+  
+  /**
+   */
+  public Output<T> sparseValues() {
+    return sparseValues;
+  }
+  
+  /**
+   */
+  public Output<TInt64> sparseShape() {
+    return sparseShape;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "DeserializeManySparse";
+  
+  private Output<TInt64> sparseIndices;
+  private Output<T> sparseValues;
+  private Output<TInt64> sparseShape;
+  
+  private DeserializeManySparse(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    sparseIndices = operation.output(outputIdx++);
+    sparseValues = operation.output(outputIdx++);
+    sparseShape = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/EncodeBase64.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/EncodeBase64.java
new file mode 100644
index 00000000000..ddb331b4a48
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/EncodeBase64.java
@@ -0,0 +1,114 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.io;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TString;
+
+/**
+ * Encode strings into web-safe base64 format.
+ * <p>
+ * Refer to the following article for more information on base64 format:
+ * en.wikipedia.org/wiki/Base64. Base64 strings may have padding with '=' at the
+ * end so that the encoded has length multiple of 4. See Padding section of the
+ * link above.
+ * <p>
+ * Web-safe means that the encoder uses - and _ instead of + and /.
+ */
+@Operator(group = "io")
+public final class EncodeBase64 extends RawOp implements Operand<TString> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.io.EncodeBase64}
+   */
+  public static class Options {
+    
+    /**
+     * @param pad Bool whether padding is applied at the ends.
+     */
+    public Options pad(Boolean pad) {
+      this.pad = pad;
+      return this;
+    }
+    
+    private Boolean pad;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new EncodeBase64 operation.
+   * 
+   * @param scope current scope
+   * @param input Strings to be encoded.
+   * @param options carries optional attributes values
+   * @return a new instance of EncodeBase64
+   */
+  @Endpoint(describeByClass = true)
+  public static EncodeBase64 create(Scope scope, Operand<TString> input, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("EncodeBase64", scope.makeOpName("EncodeBase64"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.pad != null) {
+          opBuilder.setAttr("pad", opts.pad);
+        }
+      }
+    }
+    return new EncodeBase64(opBuilder.build());
+  }
+  
+  /**
+   * @param pad Bool whether padding is applied at the ends.
+   */
+  public static Options pad(Boolean pad) {
+    return new Options().pad(pad);
+  }
+  
+  /**
+   * Input strings encoded in base64.
+   */
+  public Output<TString> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<TString> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "EncodeBase64";
+  
+  private Output<TString> output;
+  
+  private EncodeBase64(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/FifoQueue.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/FifoQueue.java
new file mode 100644
index 00000000000..0e507d45e89
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/FifoQueue.java
@@ -0,0 +1,187 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.io;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * A queue that produces elements in first-in first-out order.
+ */
+@Operator(group = "io")
+public final class FifoQueue extends RawOp implements Operand<TType> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.io.FifoQueue}
+   */
+  public static class Options {
+    
+    /**
+     * @param shapes The shape of each component in a value. The length of this attr must
+     * be either 0 or the same as the length of component_types. If the length of
+     * this attr is 0, the shapes of queue elements are not constrained, and
+     * only one element may be dequeued at a time.
+     */
+    public Options shapes(List<Shape> shapes) {
+      this.shapes = shapes;
+      return this;
+    }
+    
+    /**
+     * @param capacity The upper bound on the number of elements in this queue.
+     * Negative numbers mean no limit.
+     */
+    public Options capacity(Long capacity) {
+      this.capacity = capacity;
+      return this;
+    }
+    
+    /**
+     * @param container If non-empty, this queue is placed in the given container.
+     * Otherwise, a default container is used.
+     */
+    public Options container(String container) {
+      this.container = container;
+      return this;
+    }
+    
+    /**
+     * @param sharedName If non-empty, this queue will be shared under the given name
+     * across multiple sessions.
+     */
+    public Options sharedName(String sharedName) {
+      this.sharedName = sharedName;
+      return this;
+    }
+    
+    private List<Shape> shapes;
+    private Long capacity;
+    private String container;
+    private String sharedName;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new FifoQueue operation.
+   * 
+   * @param scope current scope
+   * @param componentTypes The type of each component in a value.
+   * @param options carries optional attributes values
+   * @return a new instance of FifoQueue
+   */
+  @Endpoint(describeByClass = true)
+  public static FifoQueue create(Scope scope, List<Class<? extends TType>> componentTypes, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("FIFOQueueV2", scope.makeOpName("FifoQueue"));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    Class[] componentTypesArray = new Class[componentTypes.size()];
+    for (int i = 0; i < componentTypesArray.length; ++i) {
+      componentTypesArray[i] = componentTypes.get(i);
+    }
+    opBuilder.setAttr("component_types", componentTypesArray);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.shapes != null) {
+          Shape[] shapesArray = new Shape[opts.shapes.size()];
+          for (int i = 0; i < shapesArray.length; ++i) {
+            shapesArray[i] = opts.shapes.get(i);
+          }
+          opBuilder.setAttr("shapes", shapesArray);
+        }
+        if (opts.capacity != null) {
+          opBuilder.setAttr("capacity", opts.capacity);
+        }
+        if (opts.container != null) {
+          opBuilder.setAttr("container", opts.container);
+        }
+        if (opts.sharedName != null) {
+          opBuilder.setAttr("shared_name", opts.sharedName);
+        }
+      }
+    }
+    return new FifoQueue(opBuilder.build());
+  }
+  
+  /**
+   * @param shapes The shape of each component in a value. The length of this attr must
+   * be either 0 or the same as the length of component_types. If the length of
+   * this attr is 0, the shapes of queue elements are not constrained, and
+   * only one element may be dequeued at a time.
+   */
+  public static Options shapes(List<Shape> shapes) {
+    return new Options().shapes(shapes);
+  }
+  
+  /**
+   * @param capacity The upper bound on the number of elements in this queue.
+   * Negative numbers mean no limit.
+   */
+  public static Options capacity(Long capacity) {
+    return new Options().capacity(capacity);
+  }
+  
+  /**
+   * @param container If non-empty, this queue is placed in the given container.
+   * Otherwise, a default container is used.
+   */
+  public static Options container(String container) {
+    return new Options().container(container);
+  }
+  
+  /**
+   * @param sharedName If non-empty, this queue will be shared under the given name
+   * across multiple sessions.
+   */
+  public static Options sharedName(String sharedName) {
+    return new Options().sharedName(sharedName);
+  }
+  
+  /**
+   * The handle to the queue.
+   */
+  public Output<?> handle() {
+    return handle;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) handle;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "FIFOQueueV2";
+  
+  private Output<?> handle;
+  
+  private FifoQueue(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    handle = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/FixedLengthRecordReader.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/FixedLengthRecordReader.java
new file mode 100644
index 00000000000..3678b8190b9
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/FixedLengthRecordReader.java
@@ -0,0 +1,211 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.io;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * A Reader that outputs fixed-length records from a file.
+ */
+@Operator(group = "io")
+public final class FixedLengthRecordReader extends RawOp implements Operand<TType> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.io.FixedLengthRecordReader}
+   */
+  public static class Options {
+    
+    /**
+     * @param headerBytes Number of bytes in the header, defaults to 0.
+     */
+    public Options headerBytes(Long headerBytes) {
+      this.headerBytes = headerBytes;
+      return this;
+    }
+    
+    /**
+     * @param footerBytes Number of bytes in the footer, defaults to 0.
+     */
+    public Options footerBytes(Long footerBytes) {
+      this.footerBytes = footerBytes;
+      return this;
+    }
+    
+    /**
+     * @param hopBytes Number of bytes to hop before each read. Default of 0 means using
+     * record_bytes.
+     */
+    public Options hopBytes(Long hopBytes) {
+      this.hopBytes = hopBytes;
+      return this;
+    }
+    
+    /**
+     * @param container If non-empty, this reader is placed in the given container.
+     * Otherwise, a default container is used.
+     */
+    public Options container(String container) {
+      this.container = container;
+      return this;
+    }
+    
+    /**
+     * @param sharedName If non-empty, this reader is named in the given bucket
+     * with this shared_name. Otherwise, the node name is used instead.
+     */
+    public Options sharedName(String sharedName) {
+      this.sharedName = sharedName;
+      return this;
+    }
+    
+    /**
+     * @param encoding The type of encoding for the file. Currently ZLIB and GZIP
+     * are supported. Defaults to none.
+     */
+    public Options encoding(String encoding) {
+      this.encoding = encoding;
+      return this;
+    }
+    
+    private Long headerBytes;
+    private Long footerBytes;
+    private Long hopBytes;
+    private String container;
+    private String sharedName;
+    private String encoding;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new FixedLengthRecordReader operation.
+   * 
+   * @param scope current scope
+   * @param recordBytes Number of bytes in the record.
+   * @param options carries optional attributes values
+   * @return a new instance of FixedLengthRecordReader
+   */
+  @Endpoint(describeByClass = true)
+  public static FixedLengthRecordReader create(Scope scope, Long recordBytes, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("FixedLengthRecordReaderV2", scope.makeOpName("FixedLengthRecordReader"));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("record_bytes", recordBytes);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.headerBytes != null) {
+          opBuilder.setAttr("header_bytes", opts.headerBytes);
+        }
+        if (opts.footerBytes != null) {
+          opBuilder.setAttr("footer_bytes", opts.footerBytes);
+        }
+        if (opts.hopBytes != null) {
+          opBuilder.setAttr("hop_bytes", opts.hopBytes);
+        }
+        if (opts.container != null) {
+          opBuilder.setAttr("container", opts.container);
+        }
+        if (opts.sharedName != null) {
+          opBuilder.setAttr("shared_name", opts.sharedName);
+        }
+        if (opts.encoding != null) {
+          opBuilder.setAttr("encoding", opts.encoding);
+        }
+      }
+    }
+    return new FixedLengthRecordReader(opBuilder.build());
+  }
+  
+  /**
+   * @param headerBytes Number of bytes in the header, defaults to 0.
+   */
+  public static Options headerBytes(Long headerBytes) {
+    return new Options().headerBytes(headerBytes);
+  }
+  
+  /**
+   * @param footerBytes Number of bytes in the footer, defaults to 0.
+   */
+  public static Options footerBytes(Long footerBytes) {
+    return new Options().footerBytes(footerBytes);
+  }
+  
+  /**
+   * @param hopBytes Number of bytes to hop before each read. Default of 0 means using
+   * record_bytes.
+   */
+  public static Options hopBytes(Long hopBytes) {
+    return new Options().hopBytes(hopBytes);
+  }
+  
+  /**
+   * @param container If non-empty, this reader is placed in the given container.
+   * Otherwise, a default container is used.
+   */
+  public static Options container(String container) {
+    return new Options().container(container);
+  }
+  
+  /**
+   * @param sharedName If non-empty, this reader is named in the given bucket
+   * with this shared_name. Otherwise, the node name is used instead.
+   */
+  public static Options sharedName(String sharedName) {
+    return new Options().sharedName(sharedName);
+  }
+  
+  /**
+   * @param encoding The type of encoding for the file. Currently ZLIB and GZIP
+   * are supported. Defaults to none.
+   */
+  public static Options encoding(String encoding) {
+    return new Options().encoding(encoding);
+  }
+  
+  /**
+   * The handle to reference the Reader.
+   */
+  public Output<?> readerHandle() {
+    return readerHandle;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) readerHandle;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "FixedLengthRecordReaderV2";
+  
+  private Output<?> readerHandle;
+  
+  private FixedLengthRecordReader(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    readerHandle = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/IdentityReader.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/IdentityReader.java
new file mode 100644
index 00000000000..99ae027a82e
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/IdentityReader.java
@@ -0,0 +1,132 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.io;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * A Reader that outputs the queued work as both the key and value.
+ * <p>
+ * To use, enqueue strings in a Queue.  ReaderRead will take the front
+ * work string and output (work, work).
+ */
+@Operator(group = "io")
+public final class IdentityReader extends RawOp implements Operand<TType> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.io.IdentityReader}
+   */
+  public static class Options {
+    
+    /**
+     * @param container If non-empty, this reader is placed in the given container.
+     * Otherwise, a default container is used.
+     */
+    public Options container(String container) {
+      this.container = container;
+      return this;
+    }
+    
+    /**
+     * @param sharedName If non-empty, this reader is named in the given bucket
+     * with this shared_name. Otherwise, the node name is used instead.
+     */
+    public Options sharedName(String sharedName) {
+      this.sharedName = sharedName;
+      return this;
+    }
+    
+    private String container;
+    private String sharedName;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new IdentityReader operation.
+   * 
+   * @param scope current scope
+   * @param options carries optional attributes values
+   * @return a new instance of IdentityReader
+   */
+  @Endpoint(describeByClass = true)
+  public static IdentityReader create(Scope scope, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("IdentityReaderV2", scope.makeOpName("IdentityReader"));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.container != null) {
+          opBuilder.setAttr("container", opts.container);
+        }
+        if (opts.sharedName != null) {
+          opBuilder.setAttr("shared_name", opts.sharedName);
+        }
+      }
+    }
+    return new IdentityReader(opBuilder.build());
+  }
+  
+  /**
+   * @param container If non-empty, this reader is placed in the given container.
+   * Otherwise, a default container is used.
+   */
+  public static Options container(String container) {
+    return new Options().container(container);
+  }
+  
+  /**
+   * @param sharedName If non-empty, this reader is named in the given bucket
+   * with this shared_name. Otherwise, the node name is used instead.
+   */
+  public static Options sharedName(String sharedName) {
+    return new Options().sharedName(sharedName);
+  }
+  
+  /**
+   * The handle to reference the Reader.
+   */
+  public Output<?> readerHandle() {
+    return readerHandle;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) readerHandle;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "IdentityReaderV2";
+  
+  private Output<?> readerHandle;
+  
+  private IdentityReader(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    readerHandle = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/LmdbReader.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/LmdbReader.java
new file mode 100644
index 00000000000..b3b10b2809c
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/LmdbReader.java
@@ -0,0 +1,128 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.io;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TString;
+
+/**
+ * A Reader that outputs the records from a LMDB file.
+ */
+@Operator(group = "io")
+public final class LmdbReader extends RawOp implements Operand<TString> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.io.LmdbReader}
+   */
+  public static class Options {
+    
+    /**
+     * @param container If non-empty, this reader is placed in the given container.
+     * Otherwise, a default container is used.
+     */
+    public Options container(String container) {
+      this.container = container;
+      return this;
+    }
+    
+    /**
+     * @param sharedName If non-empty, this reader is named in the given bucket
+     * with this shared_name. Otherwise, the node name is used instead.
+     */
+    public Options sharedName(String sharedName) {
+      this.sharedName = sharedName;
+      return this;
+    }
+    
+    private String container;
+    private String sharedName;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new LmdbReader operation.
+   * 
+   * @param scope current scope
+   * @param options carries optional attributes values
+   * @return a new instance of LmdbReader
+   */
+  @Endpoint(describeByClass = true)
+  public static LmdbReader create(Scope scope, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("LMDBReader", scope.makeOpName("LmdbReader"));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.container != null) {
+          opBuilder.setAttr("container", opts.container);
+        }
+        if (opts.sharedName != null) {
+          opBuilder.setAttr("shared_name", opts.sharedName);
+        }
+      }
+    }
+    return new LmdbReader(opBuilder.build());
+  }
+  
+  /**
+   * @param container If non-empty, this reader is placed in the given container.
+   * Otherwise, a default container is used.
+   */
+  public static Options container(String container) {
+    return new Options().container(container);
+  }
+  
+  /**
+   * @param sharedName If non-empty, this reader is named in the given bucket
+   * with this shared_name. Otherwise, the node name is used instead.
+   */
+  public static Options sharedName(String sharedName) {
+    return new Options().sharedName(sharedName);
+  }
+  
+  /**
+   * The handle to reference the Reader.
+   */
+  public Output<TString> readerHandle() {
+    return readerHandle;
+  }
+  
+  @Override
+  public Output<TString> asOutput() {
+    return readerHandle;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "LMDBReader";
+  
+  private Output<TString> readerHandle;
+  
+  private LmdbReader(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    readerHandle = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/MatchingFiles.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/MatchingFiles.java
new file mode 100644
index 00000000000..ed2b33adea3
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/MatchingFiles.java
@@ -0,0 +1,77 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.io;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TString;
+
+/**
+ * Returns the set of files matching one or more glob patterns.
+ * <p>
+ * Note that this routine only supports wildcard characters in the
+ * basename portion of the pattern, not in the directory portion.
+ * Note also that the order of filenames returned is deterministic.
+ */
+@Operator(group = "io")
+public final class MatchingFiles extends RawOp implements Operand<TString> {
+  
+  /**
+   * Factory method to create a class wrapping a new MatchingFiles operation.
+   * 
+   * @param scope current scope
+   * @param pattern Shell wildcard pattern(s). Scalar or vector of type string.
+   * @return a new instance of MatchingFiles
+   */
+  @Endpoint(describeByClass = true)
+  public static MatchingFiles create(Scope scope, Operand<TString> pattern) {
+    OperationBuilder opBuilder = scope.env().opBuilder("MatchingFiles", scope.makeOpName("MatchingFiles"));
+    opBuilder.addInput(pattern.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new MatchingFiles(opBuilder.build());
+  }
+  
+  /**
+   * A vector of matching filenames.
+   */
+  public Output<TString> filenames() {
+    return filenames;
+  }
+  
+  @Override
+  public Output<TString> asOutput() {
+    return filenames;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "MatchingFiles";
+  
+  private Output<TString> filenames;
+  
+  private MatchingFiles(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    filenames = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/PaddingFifoQueue.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/PaddingFifoQueue.java
new file mode 100644
index 00000000000..9efc68bbdc7
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/PaddingFifoQueue.java
@@ -0,0 +1,199 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.io;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * A queue that produces elements in first-in first-out order.
+ * <p>
+ * Variable-size shapes are allowed by setting the corresponding shape dimensions
+ * to 0 in the shape attr.  In this case DequeueMany will pad up to the maximum
+ * size of any given element in the minibatch.  See below for details.
+ */
+@Operator(group = "io")
+public final class PaddingFifoQueue extends RawOp implements Operand<TType> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.io.PaddingFifoQueue}
+   */
+  public static class Options {
+    
+    /**
+     * @param shapes The shape of each component in a value. The length of this attr must
+     * be either 0 or the same as the length of component_types.
+     * Shapes of fixed rank but variable size are allowed by setting
+     * any shape dimension to -1.  In this case, the inputs' shape may vary along
+     * the given dimension, and DequeueMany will pad the given dimension with
+     * zeros up to the maximum shape of all elements in the given batch.
+     * If the length of this attr is 0, different queue elements may have
+     * different ranks and shapes, but only one element may be dequeued at a time.
+     */
+    public Options shapes(List<Shape> shapes) {
+      this.shapes = shapes;
+      return this;
+    }
+    
+    /**
+     * @param capacity The upper bound on the number of elements in this queue.
+     * Negative numbers mean no limit.
+     */
+    public Options capacity(Long capacity) {
+      this.capacity = capacity;
+      return this;
+    }
+    
+    /**
+     * @param container If non-empty, this queue is placed in the given container.
+     * Otherwise, a default container is used.
+     */
+    public Options container(String container) {
+      this.container = container;
+      return this;
+    }
+    
+    /**
+     * @param sharedName If non-empty, this queue will be shared under the given name
+     * across multiple sessions.
+     */
+    public Options sharedName(String sharedName) {
+      this.sharedName = sharedName;
+      return this;
+    }
+    
+    private List<Shape> shapes;
+    private Long capacity;
+    private String container;
+    private String sharedName;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new PaddingFifoQueue operation.
+   * 
+   * @param scope current scope
+   * @param componentTypes The type of each component in a value.
+   * @param options carries optional attributes values
+   * @return a new instance of PaddingFifoQueue
+   */
+  @Endpoint(describeByClass = true)
+  public static PaddingFifoQueue create(Scope scope, List<Class<? extends TType>> componentTypes, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("PaddingFIFOQueueV2", scope.makeOpName("PaddingFifoQueue"));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    Class[] componentTypesArray = new Class[componentTypes.size()];
+    for (int i = 0; i < componentTypesArray.length; ++i) {
+      componentTypesArray[i] = componentTypes.get(i);
+    }
+    opBuilder.setAttr("component_types", componentTypesArray);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.shapes != null) {
+          Shape[] shapesArray = new Shape[opts.shapes.size()];
+          for (int i = 0; i < shapesArray.length; ++i) {
+            shapesArray[i] = opts.shapes.get(i);
+          }
+          opBuilder.setAttr("shapes", shapesArray);
+        }
+        if (opts.capacity != null) {
+          opBuilder.setAttr("capacity", opts.capacity);
+        }
+        if (opts.container != null) {
+          opBuilder.setAttr("container", opts.container);
+        }
+        if (opts.sharedName != null) {
+          opBuilder.setAttr("shared_name", opts.sharedName);
+        }
+      }
+    }
+    return new PaddingFifoQueue(opBuilder.build());
+  }
+  
+  /**
+   * @param shapes The shape of each component in a value. The length of this attr must
+   * be either 0 or the same as the length of component_types.
+   * Shapes of fixed rank but variable size are allowed by setting
+   * any shape dimension to -1.  In this case, the inputs' shape may vary along
+   * the given dimension, and DequeueMany will pad the given dimension with
+   * zeros up to the maximum shape of all elements in the given batch.
+   * If the length of this attr is 0, different queue elements may have
+   * different ranks and shapes, but only one element may be dequeued at a time.
+   */
+  public static Options shapes(List<Shape> shapes) {
+    return new Options().shapes(shapes);
+  }
+  
+  /**
+   * @param capacity The upper bound on the number of elements in this queue.
+   * Negative numbers mean no limit.
+   */
+  public static Options capacity(Long capacity) {
+    return new Options().capacity(capacity);
+  }
+  
+  /**
+   * @param container If non-empty, this queue is placed in the given container.
+   * Otherwise, a default container is used.
+   */
+  public static Options container(String container) {
+    return new Options().container(container);
+  }
+  
+  /**
+   * @param sharedName If non-empty, this queue will be shared under the given name
+   * across multiple sessions.
+   */
+  public static Options sharedName(String sharedName) {
+    return new Options().sharedName(sharedName);
+  }
+  
+  /**
+   * The handle to the queue.
+   */
+  public Output<?> handle() {
+    return handle;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) handle;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "PaddingFIFOQueueV2";
+  
+  private Output<?> handle;
+  
+  private PaddingFifoQueue(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    handle = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ParseExample.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ParseExample.java
new file mode 100644
index 00000000000..cbcf7e4deef
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ParseExample.java
@@ -0,0 +1,202 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.io;
+
+import java.util.Arrays;
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.Operands;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.TString;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Transforms a vector of tf.Example protos (as strings) into typed tensors.
+ */
+@Operator(group = "io")
+public final class ParseExample extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new ParseExample operation.
+   * 
+   * @param scope current scope
+   * @param serialized A scalar or vector containing binary serialized Example protos.
+   * @param names A tensor containing the names of the serialized protos.
+   * Corresponds 1:1 with the `serialized` tensor.
+   * May contain, for example, table key (descriptive) names for the
+   * corresponding serialized protos.  These are purely useful for debugging
+   * purposes, and the presence of values here has no effect on the output.
+   * May also be an empty vector if no names are available.
+   * If non-empty, this tensor must have the same shape as "serialized".
+   * @param sparseKeys Vector of strings.
+   * The keys expected in the Examples' features associated with sparse values.
+   * @param denseKeys Vector of strings.
+   * The keys expected in the Examples' features associated with dense values.
+   * @param raggedKeys Vector of strings.
+   * The keys expected in the Examples' features associated with ragged values.
+   * @param denseDefaults A list of Tensors (some may be empty).  Corresponds 1:1 with `dense_keys`.
+   * dense_defaults[j] provides default values
+   * when the example's feature_map lacks dense_key[j].  If an empty Tensor is
+   * provided for dense_defaults[j], then the Feature dense_keys[j] is required.
+   * The input type is inferred from dense_defaults[j], even when it's empty.
+   * If dense_defaults[j] is not empty, and dense_shapes[j] is fully defined,
+   * then the shape of dense_defaults[j] must match that of dense_shapes[j].
+   * If dense_shapes[j] has an undefined major dimension (variable strides dense
+   * feature), dense_defaults[j] must contain a single element:
+   * the padding element.
+   * @param numSparse The number of sparse keys.
+   * @param sparseTypes A list of `num_sparse` types; the data types of data in each Feature
+   * given in sparse_keys.
+   * Currently the ParseExample supports DT_FLOAT (FloatList),
+   * DT_INT64 (Int64List), and DT_STRING (BytesList).
+   * @param raggedValueTypes A list of `num_ragged` types; the data types of data in each Feature
+   * given in ragged_keys (where `num_ragged = sparse_keys.size()`).
+   * Currently the ParseExample supports DT_FLOAT (FloatList),
+   * DT_INT64 (Int64List), and DT_STRING (BytesList).
+   * @param raggedSplitTypes A list of `num_ragged` types; the data types of row_splits in each Feature
+   * given in ragged_keys (where `num_ragged = sparse_keys.size()`).
+   * May be DT_INT32 or DT_INT64.
+   * @param denseShapes A list of `num_dense` shapes; the shapes of data in each Feature
+   * given in dense_keys (where `num_dense = dense_keys.size()`).
+   * The number of elements in the Feature corresponding to dense_key[j]
+   * must always equal dense_shapes[j].NumEntries().
+   * If dense_shapes[j] == (D0, D1, ..., DN) then the shape of output
+   * Tensor dense_values[j] will be (|serialized|, D0, D1, ..., DN):
+   * The dense outputs are just the inputs row-stacked by batch.
+   * This works for dense_shapes[j] = (-1, D1, ..., DN).  In this case
+   * the shape of the output Tensor dense_values[j] will be
+   * (|serialized|, M, D1, .., DN), where M is the maximum number of blocks
+   * of elements of length D1 * .... * DN, across all minibatch entries
+   * in the input.  Any minibatch entry with less than M blocks of elements of
+   * length D1 * ... * DN will be padded with the corresponding default_value
+   * scalar element along the second dimension.
+   * @return a new instance of ParseExample
+   */
+  @Endpoint(describeByClass = true)
+  public static ParseExample create(Scope scope, Operand<TString> serialized, Operand<TString> names, Operand<TString> sparseKeys, Operand<TString> denseKeys, Operand<TString> raggedKeys, Iterable<Operand<?>> denseDefaults, Long numSparse, List<Class<? extends TType>> sparseTypes, List<Class<? extends TType>> raggedValueTypes, List<Class<? extends TNumber>> raggedSplitTypes, List<Shape> denseShapes) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ParseExampleV2", scope.makeOpName("ParseExample"));
+    opBuilder.addInput(serialized.asOutput());
+    opBuilder.addInput(names.asOutput());
+    opBuilder.addInput(sparseKeys.asOutput());
+    opBuilder.addInput(denseKeys.asOutput());
+    opBuilder.addInput(raggedKeys.asOutput());
+    opBuilder.addInputList(Operands.asOutputs(denseDefaults));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("num_sparse", numSparse);
+    Class[] sparseTypesArray = new Class[sparseTypes.size()];
+    for (int i = 0; i < sparseTypesArray.length; ++i) {
+      sparseTypesArray[i] = sparseTypes.get(i);
+    }
+    opBuilder.setAttr("sparse_types", sparseTypesArray);
+    Class[] raggedValueTypesArray = new Class[raggedValueTypes.size()];
+    for (int i = 0; i < raggedValueTypesArray.length; ++i) {
+      raggedValueTypesArray[i] = raggedValueTypes.get(i);
+    }
+    opBuilder.setAttr("ragged_value_types", raggedValueTypesArray);
+    Class[] raggedSplitTypesArray = new Class[raggedSplitTypes.size()];
+    for (int i = 0; i < raggedSplitTypesArray.length; ++i) {
+      raggedSplitTypesArray[i] = raggedSplitTypes.get(i);
+    }
+    opBuilder.setAttr("ragged_split_types", raggedSplitTypesArray);
+    Shape[] denseShapesArray = new Shape[denseShapes.size()];
+    for (int i = 0; i < denseShapesArray.length; ++i) {
+      denseShapesArray[i] = denseShapes.get(i);
+    }
+    opBuilder.setAttr("dense_shapes", denseShapesArray);
+    return new ParseExample(opBuilder.build());
+  }
+  
+  /**
+   */
+  public List<Output<TInt64>> sparseIndices() {
+    return sparseIndices;
+  }
+  
+  /**
+   */
+  public List<Output<?>> sparseValues() {
+    return sparseValues;
+  }
+  
+  /**
+   */
+  public List<Output<TInt64>> sparseShapes() {
+    return sparseShapes;
+  }
+  
+  /**
+   */
+  public List<Output<?>> denseValues() {
+    return denseValues;
+  }
+  
+  /**
+   */
+  public List<Output<?>> raggedValues() {
+    return raggedValues;
+  }
+  
+  /**
+   */
+  public List<Output<?>> raggedRowSplits() {
+    return raggedRowSplits;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ParseExampleV2";
+  
+  private List<Output<TInt64>> sparseIndices;
+  private List<Output<?>> sparseValues;
+  private List<Output<TInt64>> sparseShapes;
+  private List<Output<?>> denseValues;
+  private List<Output<?>> raggedValues;
+  private List<Output<?>> raggedRowSplits;
+  
+  @SuppressWarnings("unchecked")
+  private ParseExample(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    int sparseIndicesLength = operation.outputListLength("sparse_indices");
+    sparseIndices = Arrays.asList((Output<TInt64>[])operation.outputList(outputIdx, sparseIndicesLength));
+    outputIdx += sparseIndicesLength;
+    int sparseValuesLength = operation.outputListLength("sparse_values");
+    sparseValues = Arrays.asList(operation.outputList(outputIdx, sparseValuesLength));
+    outputIdx += sparseValuesLength;
+    int sparseShapesLength = operation.outputListLength("sparse_shapes");
+    sparseShapes = Arrays.asList((Output<TInt64>[])operation.outputList(outputIdx, sparseShapesLength));
+    outputIdx += sparseShapesLength;
+    int denseValuesLength = operation.outputListLength("dense_values");
+    denseValues = Arrays.asList(operation.outputList(outputIdx, denseValuesLength));
+    outputIdx += denseValuesLength;
+    int raggedValuesLength = operation.outputListLength("ragged_values");
+    raggedValues = Arrays.asList(operation.outputList(outputIdx, raggedValuesLength));
+    outputIdx += raggedValuesLength;
+    int raggedRowSplitsLength = operation.outputListLength("ragged_row_splits");
+    raggedRowSplits = Arrays.asList(operation.outputList(outputIdx, raggedRowSplitsLength));
+    outputIdx += raggedRowSplitsLength;
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ParseSequenceExample.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ParseSequenceExample.java
new file mode 100644
index 00000000000..df0db9fabc0
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ParseSequenceExample.java
@@ -0,0 +1,426 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.io;
+
+import java.util.Arrays;
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.Operands;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TBool;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.TString;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Transforms a vector of tf.io.SequenceExample protos (as strings) into
+ * typed tensors.
+ */
+@Operator(group = "io")
+public final class ParseSequenceExample extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.io.ParseSequenceExample}
+   */
+  public static class Options {
+    
+    /**
+     * @param NcontextSparse 
+     */
+    public Options NcontextSparse(Long NcontextSparse) {
+      this.NcontextSparse = NcontextSparse;
+      return this;
+    }
+    
+    /**
+     * @param contextDenseShapes A list of Ncontext_dense shapes; the shapes of data in
+     * each context Feature given in context_dense_keys.
+     * The number of elements in the Feature corresponding to context_dense_key[j]
+     * must always equal context_dense_shapes[j].NumEntries().
+     * The shape of context_dense_values[j] will match context_dense_shapes[j].
+     */
+    public Options contextDenseShapes(List<Shape> contextDenseShapes) {
+      this.contextDenseShapes = contextDenseShapes;
+      return this;
+    }
+    
+    /**
+     * @param NfeatureListSparse 
+     */
+    public Options NfeatureListSparse(Long NfeatureListSparse) {
+      this.NfeatureListSparse = NfeatureListSparse;
+      return this;
+    }
+    
+    /**
+     * @param NfeatureListDense 
+     */
+    public Options NfeatureListDense(Long NfeatureListDense) {
+      this.NfeatureListDense = NfeatureListDense;
+      return this;
+    }
+    
+    /**
+     * @param featureListDenseShapes A list of Nfeature_list_dense shapes; the shapes of
+     * data in each FeatureList given in feature_list_dense_keys.
+     * The shape of each Feature in the FeatureList corresponding to
+     * feature_list_dense_key[j] must always equal
+     * feature_list_dense_shapes[j].NumEntries().
+     */
+    public Options featureListDenseShapes(List<Shape> featureListDenseShapes) {
+      this.featureListDenseShapes = featureListDenseShapes;
+      return this;
+    }
+    
+    private Long NcontextSparse;
+    private List<Shape> contextDenseShapes;
+    private Long NfeatureListSparse;
+    private Long NfeatureListDense;
+    private List<Shape> featureListDenseShapes;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new ParseSequenceExample operation.
+   * 
+   * @param scope current scope
+   * @param serialized A scalar or vector containing binary serialized SequenceExample protos.
+   * @param debugName A scalar or vector containing the names of the serialized protos.
+   * May contain, for example, table key (descriptive) name for the
+   * corresponding serialized proto.  This is purely useful for debugging
+   * purposes, and the presence of values here has no effect on the output.
+   * May also be an empty vector if no name is available.
+   * @param contextSparseKeys The keys expected in the Examples' features associated with context_sparse
+   * values.
+   * @param contextDenseKeys The keys expected in the SequenceExamples' context features associated with
+   * dense values.
+   * @param contextRaggedKeys The keys expected in the Examples' features associated with context_ragged
+   * values.
+   * @param featureListSparseKeys The keys expected in the FeatureLists associated with sparse values.
+   * @param featureListDenseKeys The keys expected in the SequenceExamples' feature_lists associated
+   * with lists of dense values.
+   * @param featureListRaggedKeys The keys expected in the FeatureLists associated with ragged values.
+   * @param featureListDenseMissingAssumedEmpty A vector corresponding 1:1 with feature_list_dense_keys, indicating which
+   * features may be missing from the SequenceExamples.  If the associated
+   * FeatureList is missing, it is treated as empty.
+   * @param contextDenseDefaults A list of Ncontext_dense Tensors (some may be empty).
+   * context_dense_defaults[j] provides default values
+   * when the SequenceExample's context map lacks context_dense_key[j].
+   * If an empty Tensor is provided for context_dense_defaults[j],
+   * then the Feature context_dense_keys[j] is required.
+   * The input type is inferred from context_dense_defaults[j], even when it's
+   * empty.  If context_dense_defaults[j] is not empty, its shape must match
+   * context_dense_shapes[j].
+   * @param contextSparseTypes A list of Ncontext_sparse types; the data types of data in
+   * each context Feature given in context_sparse_keys.
+   * Currently the ParseSingleSequenceExample supports DT_FLOAT (FloatList),
+   * DT_INT64 (Int64List), and DT_STRING (BytesList).
+   * @param contextRaggedValueTypes RaggedTensor.value dtypes for the ragged context features.
+   * @param contextRaggedSplitTypes RaggedTensor.row_split dtypes for the ragged context features.
+   * @param featureListDenseTypes 
+   * @param featureListSparseTypes A list of Nfeature_list_sparse types; the data types
+   * of data in each FeatureList given in feature_list_sparse_keys.
+   * Currently the ParseSingleSequenceExample supports DT_FLOAT (FloatList),
+   * DT_INT64 (Int64List), and DT_STRING (BytesList).
+   * @param featureListRaggedValueTypes RaggedTensor.value dtypes for the ragged FeatureList features.
+   * @param featureListRaggedSplitTypes RaggedTensor.row_split dtypes for the ragged FeatureList features.
+   * @param options carries optional attributes values
+   * @return a new instance of ParseSequenceExample
+   */
+  @Endpoint(describeByClass = true)
+  public static ParseSequenceExample create(Scope scope, Operand<TString> serialized, Operand<TString> debugName, Operand<TString> contextSparseKeys, Operand<TString> contextDenseKeys, Operand<TString> contextRaggedKeys, Operand<TString> featureListSparseKeys, Operand<TString> featureListDenseKeys, Operand<TString> featureListRaggedKeys, Operand<TBool> featureListDenseMissingAssumedEmpty, Iterable<Operand<?>> contextDenseDefaults, List<Class<? extends TType>> contextSparseTypes, List<Class<? extends TType>> contextRaggedValueTypes, List<Class<? extends TNumber>> contextRaggedSplitTypes, List<Class<? extends TType>> featureListDenseTypes, List<Class<? extends TType>> featureListSparseTypes, List<Class<? extends TType>> featureListRaggedValueTypes, List<Class<? extends TNumber>> featureListRaggedSplitTypes, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ParseSequenceExampleV2", scope.makeOpName("ParseSequenceExample"));
+    opBuilder.addInput(serialized.asOutput());
+    opBuilder.addInput(debugName.asOutput());
+    opBuilder.addInput(contextSparseKeys.asOutput());
+    opBuilder.addInput(contextDenseKeys.asOutput());
+    opBuilder.addInput(contextRaggedKeys.asOutput());
+    opBuilder.addInput(featureListSparseKeys.asOutput());
+    opBuilder.addInput(featureListDenseKeys.asOutput());
+    opBuilder.addInput(featureListRaggedKeys.asOutput());
+    opBuilder.addInput(featureListDenseMissingAssumedEmpty.asOutput());
+    opBuilder.addInputList(Operands.asOutputs(contextDenseDefaults));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    Class[] contextSparseTypesArray = new Class[contextSparseTypes.size()];
+    for (int i = 0; i < contextSparseTypesArray.length; ++i) {
+      contextSparseTypesArray[i] = contextSparseTypes.get(i);
+    }
+    opBuilder.setAttr("context_sparse_types", contextSparseTypesArray);
+    Class[] contextRaggedValueTypesArray = new Class[contextRaggedValueTypes.size()];
+    for (int i = 0; i < contextRaggedValueTypesArray.length; ++i) {
+      contextRaggedValueTypesArray[i] = contextRaggedValueTypes.get(i);
+    }
+    opBuilder.setAttr("context_ragged_value_types", contextRaggedValueTypesArray);
+    Class[] contextRaggedSplitTypesArray = new Class[contextRaggedSplitTypes.size()];
+    for (int i = 0; i < contextRaggedSplitTypesArray.length; ++i) {
+      contextRaggedSplitTypesArray[i] = contextRaggedSplitTypes.get(i);
+    }
+    opBuilder.setAttr("context_ragged_split_types", contextRaggedSplitTypesArray);
+    Class[] featureListDenseTypesArray = new Class[featureListDenseTypes.size()];
+    for (int i = 0; i < featureListDenseTypesArray.length; ++i) {
+      featureListDenseTypesArray[i] = featureListDenseTypes.get(i);
+    }
+    opBuilder.setAttr("feature_list_dense_types", featureListDenseTypesArray);
+    Class[] featureListSparseTypesArray = new Class[featureListSparseTypes.size()];
+    for (int i = 0; i < featureListSparseTypesArray.length; ++i) {
+      featureListSparseTypesArray[i] = featureListSparseTypes.get(i);
+    }
+    opBuilder.setAttr("feature_list_sparse_types", featureListSparseTypesArray);
+    Class[] featureListRaggedValueTypesArray = new Class[featureListRaggedValueTypes.size()];
+    for (int i = 0; i < featureListRaggedValueTypesArray.length; ++i) {
+      featureListRaggedValueTypesArray[i] = featureListRaggedValueTypes.get(i);
+    }
+    opBuilder.setAttr("feature_list_ragged_value_types", featureListRaggedValueTypesArray);
+    Class[] featureListRaggedSplitTypesArray = new Class[featureListRaggedSplitTypes.size()];
+    for (int i = 0; i < featureListRaggedSplitTypesArray.length; ++i) {
+      featureListRaggedSplitTypesArray[i] = featureListRaggedSplitTypes.get(i);
+    }
+    opBuilder.setAttr("feature_list_ragged_split_types", featureListRaggedSplitTypesArray);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.NcontextSparse != null) {
+          opBuilder.setAttr("Ncontext_sparse", opts.NcontextSparse);
+        }
+        if (opts.contextDenseShapes != null) {
+          Shape[] contextDenseShapesArray = new Shape[opts.contextDenseShapes.size()];
+          for (int i = 0; i < contextDenseShapesArray.length; ++i) {
+            contextDenseShapesArray[i] = opts.contextDenseShapes.get(i);
+          }
+          opBuilder.setAttr("context_dense_shapes", contextDenseShapesArray);
+        }
+        if (opts.NfeatureListSparse != null) {
+          opBuilder.setAttr("Nfeature_list_sparse", opts.NfeatureListSparse);
+        }
+        if (opts.NfeatureListDense != null) {
+          opBuilder.setAttr("Nfeature_list_dense", opts.NfeatureListDense);
+        }
+        if (opts.featureListDenseShapes != null) {
+          Shape[] featureListDenseShapesArray = new Shape[opts.featureListDenseShapes.size()];
+          for (int i = 0; i < featureListDenseShapesArray.length; ++i) {
+            featureListDenseShapesArray[i] = opts.featureListDenseShapes.get(i);
+          }
+          opBuilder.setAttr("feature_list_dense_shapes", featureListDenseShapesArray);
+        }
+      }
+    }
+    return new ParseSequenceExample(opBuilder.build());
+  }
+  
+  /**
+   * @param NcontextSparse 
+   */
+  public static Options NcontextSparse(Long NcontextSparse) {
+    return new Options().NcontextSparse(NcontextSparse);
+  }
+  
+  /**
+   * @param contextDenseShapes A list of Ncontext_dense shapes; the shapes of data in
+   * each context Feature given in context_dense_keys.
+   * The number of elements in the Feature corresponding to context_dense_key[j]
+   * must always equal context_dense_shapes[j].NumEntries().
+   * The shape of context_dense_values[j] will match context_dense_shapes[j].
+   */
+  public static Options contextDenseShapes(List<Shape> contextDenseShapes) {
+    return new Options().contextDenseShapes(contextDenseShapes);
+  }
+  
+  /**
+   * @param NfeatureListSparse 
+   */
+  public static Options NfeatureListSparse(Long NfeatureListSparse) {
+    return new Options().NfeatureListSparse(NfeatureListSparse);
+  }
+  
+  /**
+   * @param NfeatureListDense 
+   */
+  public static Options NfeatureListDense(Long NfeatureListDense) {
+    return new Options().NfeatureListDense(NfeatureListDense);
+  }
+  
+  /**
+   * @param featureListDenseShapes A list of Nfeature_list_dense shapes; the shapes of
+   * data in each FeatureList given in feature_list_dense_keys.
+   * The shape of each Feature in the FeatureList corresponding to
+   * feature_list_dense_key[j] must always equal
+   * feature_list_dense_shapes[j].NumEntries().
+   */
+  public static Options featureListDenseShapes(List<Shape> featureListDenseShapes) {
+    return new Options().featureListDenseShapes(featureListDenseShapes);
+  }
+  
+  /**
+   */
+  public List<Output<TInt64>> contextSparseIndices() {
+    return contextSparseIndices;
+  }
+  
+  /**
+   */
+  public List<Output<?>> contextSparseValues() {
+    return contextSparseValues;
+  }
+  
+  /**
+   */
+  public List<Output<TInt64>> contextSparseShapes() {
+    return contextSparseShapes;
+  }
+  
+  /**
+   */
+  public List<Output<?>> contextDenseValues() {
+    return contextDenseValues;
+  }
+  
+  /**
+   */
+  public List<Output<?>> contextRaggedValues() {
+    return contextRaggedValues;
+  }
+  
+  /**
+   */
+  public List<Output<?>> contextRaggedRowSplits() {
+    return contextRaggedRowSplits;
+  }
+  
+  /**
+   */
+  public List<Output<TInt64>> featureListSparseIndices() {
+    return featureListSparseIndices;
+  }
+  
+  /**
+   */
+  public List<Output<?>> featureListSparseValues() {
+    return featureListSparseValues;
+  }
+  
+  /**
+   */
+  public List<Output<TInt64>> featureListSparseShapes() {
+    return featureListSparseShapes;
+  }
+  
+  /**
+   */
+  public List<Output<?>> featureListDenseValues() {
+    return featureListDenseValues;
+  }
+  
+  /**
+   */
+  public List<Output<TInt64>> featureListDenseLengths() {
+    return featureListDenseLengths;
+  }
+  
+  /**
+   */
+  public List<Output<?>> featureListRaggedValues() {
+    return featureListRaggedValues;
+  }
+  
+  /**
+   */
+  public List<Output<?>> featureListRaggedOuterSplits() {
+    return featureListRaggedOuterSplits;
+  }
+  
+  /**
+   */
+  public List<Output<?>> featureListRaggedInnerSplits() {
+    return featureListRaggedInnerSplits;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ParseSequenceExampleV2";
+  
+  private List<Output<TInt64>> contextSparseIndices;
+  private List<Output<?>> contextSparseValues;
+  private List<Output<TInt64>> contextSparseShapes;
+  private List<Output<?>> contextDenseValues;
+  private List<Output<?>> contextRaggedValues;
+  private List<Output<?>> contextRaggedRowSplits;
+  private List<Output<TInt64>> featureListSparseIndices;
+  private List<Output<?>> featureListSparseValues;
+  private List<Output<TInt64>> featureListSparseShapes;
+  private List<Output<?>> featureListDenseValues;
+  private List<Output<TInt64>> featureListDenseLengths;
+  private List<Output<?>> featureListRaggedValues;
+  private List<Output<?>> featureListRaggedOuterSplits;
+  private List<Output<?>> featureListRaggedInnerSplits;
+  
+  @SuppressWarnings("unchecked")
+  private ParseSequenceExample(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    int contextSparseIndicesLength = operation.outputListLength("context_sparse_indices");
+    contextSparseIndices = Arrays.asList((Output<TInt64>[])operation.outputList(outputIdx, contextSparseIndicesLength));
+    outputIdx += contextSparseIndicesLength;
+    int contextSparseValuesLength = operation.outputListLength("context_sparse_values");
+    contextSparseValues = Arrays.asList(operation.outputList(outputIdx, contextSparseValuesLength));
+    outputIdx += contextSparseValuesLength;
+    int contextSparseShapesLength = operation.outputListLength("context_sparse_shapes");
+    contextSparseShapes = Arrays.asList((Output<TInt64>[])operation.outputList(outputIdx, contextSparseShapesLength));
+    outputIdx += contextSparseShapesLength;
+    int contextDenseValuesLength = operation.outputListLength("context_dense_values");
+    contextDenseValues = Arrays.asList(operation.outputList(outputIdx, contextDenseValuesLength));
+    outputIdx += contextDenseValuesLength;
+    int contextRaggedValuesLength = operation.outputListLength("context_ragged_values");
+    contextRaggedValues = Arrays.asList(operation.outputList(outputIdx, contextRaggedValuesLength));
+    outputIdx += contextRaggedValuesLength;
+    int contextRaggedRowSplitsLength = operation.outputListLength("context_ragged_row_splits");
+    contextRaggedRowSplits = Arrays.asList(operation.outputList(outputIdx, contextRaggedRowSplitsLength));
+    outputIdx += contextRaggedRowSplitsLength;
+    int featureListSparseIndicesLength = operation.outputListLength("feature_list_sparse_indices");
+    featureListSparseIndices = Arrays.asList((Output<TInt64>[])operation.outputList(outputIdx, featureListSparseIndicesLength));
+    outputIdx += featureListSparseIndicesLength;
+    int featureListSparseValuesLength = operation.outputListLength("feature_list_sparse_values");
+    featureListSparseValues = Arrays.asList(operation.outputList(outputIdx, featureListSparseValuesLength));
+    outputIdx += featureListSparseValuesLength;
+    int featureListSparseShapesLength = operation.outputListLength("feature_list_sparse_shapes");
+    featureListSparseShapes = Arrays.asList((Output<TInt64>[])operation.outputList(outputIdx, featureListSparseShapesLength));
+    outputIdx += featureListSparseShapesLength;
+    int featureListDenseValuesLength = operation.outputListLength("feature_list_dense_values");
+    featureListDenseValues = Arrays.asList(operation.outputList(outputIdx, featureListDenseValuesLength));
+    outputIdx += featureListDenseValuesLength;
+    int featureListDenseLengthsLength = operation.outputListLength("feature_list_dense_lengths");
+    featureListDenseLengths = Arrays.asList((Output<TInt64>[])operation.outputList(outputIdx, featureListDenseLengthsLength));
+    outputIdx += featureListDenseLengthsLength;
+    int featureListRaggedValuesLength = operation.outputListLength("feature_list_ragged_values");
+    featureListRaggedValues = Arrays.asList(operation.outputList(outputIdx, featureListRaggedValuesLength));
+    outputIdx += featureListRaggedValuesLength;
+    int featureListRaggedOuterSplitsLength = operation.outputListLength("feature_list_ragged_outer_splits");
+    featureListRaggedOuterSplits = Arrays.asList(operation.outputList(outputIdx, featureListRaggedOuterSplitsLength));
+    outputIdx += featureListRaggedOuterSplitsLength;
+    int featureListRaggedInnerSplitsLength = operation.outputListLength("feature_list_ragged_inner_splits");
+    featureListRaggedInnerSplits = Arrays.asList(operation.outputList(outputIdx, featureListRaggedInnerSplitsLength));
+    outputIdx += featureListRaggedInnerSplitsLength;
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ParseSingleExample.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ParseSingleExample.java
new file mode 100644
index 00000000000..597e75efa9e
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ParseSingleExample.java
@@ -0,0 +1,157 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.io;
+
+import java.util.Arrays;
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.Operands;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.TString;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Transforms a tf.Example proto (as a string) into typed tensors.
+ */
+@Operator(group = "io")
+public final class ParseSingleExample extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new ParseSingleExample operation.
+   * 
+   * @param scope current scope
+   * @param serialized A vector containing a batch of binary serialized Example protos.
+   * @param denseDefaults A list of Tensors (some may be empty), whose length matches
+   * the length of `dense_keys`. dense_defaults[j] provides default values
+   * when the example's feature_map lacks dense_key[j].  If an empty Tensor is
+   * provided for dense_defaults[j], then the Feature dense_keys[j] is required.
+   * The input type is inferred from dense_defaults[j], even when it's empty.
+   * If dense_defaults[j] is not empty, and dense_shapes[j] is fully defined,
+   * then the shape of dense_defaults[j] must match that of dense_shapes[j].
+   * If dense_shapes[j] has an undefined major dimension (variable strides dense
+   * feature), dense_defaults[j] must contain a single element:
+   * the padding element.
+   * @param numSparse The number of sparse features to be parsed from the example. This
+   * must match the lengths of `sparse_keys` and `sparse_types`.
+   * @param sparseKeys A list of `num_sparse` strings.
+   * The keys expected in the Examples' features associated with sparse values.
+   * @param denseKeys The keys expected in the Examples' features associated with dense
+   * values.
+   * @param sparseTypes A list of `num_sparse` types; the data types of data in each
+   * Feature given in sparse_keys.
+   * Currently the ParseSingleExample op supports DT_FLOAT (FloatList),
+   * DT_INT64 (Int64List), and DT_STRING (BytesList).
+   * @param denseShapes The shapes of data in each Feature given in dense_keys.
+   * The length of this list must match the length of `dense_keys`.  The
+   * number of elements in the Feature corresponding to dense_key[j] must
+   * always equal dense_shapes[j].NumEntries().  If dense_shapes[j] ==
+   * (D0, D1, ..., DN) then the shape of output Tensor dense_values[j]
+   * will be (D0, D1, ..., DN): In the case dense_shapes[j] = (-1, D1,
+   * ..., DN), the shape of the output Tensor dense_values[j] will be (M,
+   * D1, .., DN), where M is the number of blocks of elements of length
+   * D1 * .... * DN, in the input.
+   * @return a new instance of ParseSingleExample
+   */
+  @Endpoint(describeByClass = true)
+  public static ParseSingleExample create(Scope scope, Operand<TString> serialized, Iterable<Operand<?>> denseDefaults, Long numSparse, List<String> sparseKeys, List<String> denseKeys, List<Class<? extends TType>> sparseTypes, List<Shape> denseShapes) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ParseSingleExample", scope.makeOpName("ParseSingleExample"));
+    opBuilder.addInput(serialized.asOutput());
+    opBuilder.addInputList(Operands.asOutputs(denseDefaults));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("num_sparse", numSparse);
+    String[] sparseKeysArray = new String[sparseKeys.size()];
+    for (int i = 0; i < sparseKeysArray.length; ++i) {
+      sparseKeysArray[i] = sparseKeys.get(i);
+    }
+    opBuilder.setAttr("sparse_keys", sparseKeysArray);
+    String[] denseKeysArray = new String[denseKeys.size()];
+    for (int i = 0; i < denseKeysArray.length; ++i) {
+      denseKeysArray[i] = denseKeys.get(i);
+    }
+    opBuilder.setAttr("dense_keys", denseKeysArray);
+    Class[] sparseTypesArray = new Class[sparseTypes.size()];
+    for (int i = 0; i < sparseTypesArray.length; ++i) {
+      sparseTypesArray[i] = sparseTypes.get(i);
+    }
+    opBuilder.setAttr("sparse_types", sparseTypesArray);
+    Shape[] denseShapesArray = new Shape[denseShapes.size()];
+    for (int i = 0; i < denseShapesArray.length; ++i) {
+      denseShapesArray[i] = denseShapes.get(i);
+    }
+    opBuilder.setAttr("dense_shapes", denseShapesArray);
+    return new ParseSingleExample(opBuilder.build());
+  }
+  
+  /**
+   */
+  public List<Output<TInt64>> sparseIndices() {
+    return sparseIndices;
+  }
+  
+  /**
+   */
+  public List<Output<?>> sparseValues() {
+    return sparseValues;
+  }
+  
+  /**
+   */
+  public List<Output<TInt64>> sparseShapes() {
+    return sparseShapes;
+  }
+  
+  /**
+   */
+  public List<Output<?>> denseValues() {
+    return denseValues;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ParseSingleExample";
+  
+  private List<Output<TInt64>> sparseIndices;
+  private List<Output<?>> sparseValues;
+  private List<Output<TInt64>> sparseShapes;
+  private List<Output<?>> denseValues;
+  
+  @SuppressWarnings("unchecked")
+  private ParseSingleExample(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    int sparseIndicesLength = operation.outputListLength("sparse_indices");
+    sparseIndices = Arrays.asList((Output<TInt64>[])operation.outputList(outputIdx, sparseIndicesLength));
+    outputIdx += sparseIndicesLength;
+    int sparseValuesLength = operation.outputListLength("sparse_values");
+    sparseValues = Arrays.asList(operation.outputList(outputIdx, sparseValuesLength));
+    outputIdx += sparseValuesLength;
+    int sparseShapesLength = operation.outputListLength("sparse_shapes");
+    sparseShapes = Arrays.asList((Output<TInt64>[])operation.outputList(outputIdx, sparseShapesLength));
+    outputIdx += sparseShapesLength;
+    int denseValuesLength = operation.outputListLength("dense_values");
+    denseValues = Arrays.asList(operation.outputList(outputIdx, denseValuesLength));
+    outputIdx += denseValuesLength;
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ParseSingleSequenceExample.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ParseSingleSequenceExample.java
new file mode 100644
index 00000000000..89105dc77ab
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ParseSingleSequenceExample.java
@@ -0,0 +1,283 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.io;
+
+import java.util.Arrays;
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.Operands;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.TString;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Transforms a scalar brain.SequenceExample proto (as strings) into typed tensors.
+ */
+@Operator(group = "io")
+public final class ParseSingleSequenceExample extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.io.ParseSingleSequenceExample}
+   */
+  public static class Options {
+    
+    /**
+     * @param contextDenseShapes A list of Ncontext_dense shapes; the shapes of data in
+     * each context Feature given in context_dense_keys.
+     * The number of elements in the Feature corresponding to context_dense_key[j]
+     * must always equal context_dense_shapes[j].NumEntries().
+     * The shape of context_dense_values[j] will match context_dense_shapes[j].
+     */
+    public Options contextDenseShapes(List<Shape> contextDenseShapes) {
+      this.contextDenseShapes = contextDenseShapes;
+      return this;
+    }
+    
+    /**
+     * @param featureListDenseShapes A list of Nfeature_list_dense shapes; the shapes of
+     * data in each FeatureList given in feature_list_dense_keys.
+     * The shape of each Feature in the FeatureList corresponding to
+     * feature_list_dense_key[j] must always equal
+     * feature_list_dense_shapes[j].NumEntries().
+     */
+    public Options featureListDenseShapes(List<Shape> featureListDenseShapes) {
+      this.featureListDenseShapes = featureListDenseShapes;
+      return this;
+    }
+    
+    private List<Shape> contextDenseShapes;
+    private List<Shape> featureListDenseShapes;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new ParseSingleSequenceExample operation.
+   * 
+   * @param scope current scope
+   * @param serialized A scalar containing a binary serialized SequenceExample proto.
+   * @param featureListDenseMissingAssumedEmpty A vector listing the
+   * FeatureList keys which may be missing from the SequenceExample.  If the
+   * associated FeatureList is missing, it is treated as empty.  By default,
+   * any FeatureList not listed in this vector must exist in the SequenceExample.
+   * @param contextSparseKeys A list of Ncontext_sparse string Tensors (scalars).
+   * The keys expected in the Examples' features associated with context_sparse
+   * values.
+   * @param contextDenseKeys A list of Ncontext_dense string Tensors (scalars).
+   * The keys expected in the SequenceExamples' context features associated with
+   * dense values.
+   * @param featureListSparseKeys A list of Nfeature_list_sparse string Tensors
+   * (scalars).  The keys expected in the FeatureLists associated with sparse
+   * values.
+   * @param featureListDenseKeys A list of Nfeature_list_dense string Tensors (scalars).
+   * The keys expected in the SequenceExamples' feature_lists associated
+   * with lists of dense values.
+   * @param contextDenseDefaults A list of Ncontext_dense Tensors (some may be empty).
+   * context_dense_defaults[j] provides default values
+   * when the SequenceExample's context map lacks context_dense_key[j].
+   * If an empty Tensor is provided for context_dense_defaults[j],
+   * then the Feature context_dense_keys[j] is required.
+   * The input type is inferred from context_dense_defaults[j], even when it's
+   * empty.  If context_dense_defaults[j] is not empty, its shape must match
+   * context_dense_shapes[j].
+   * @param debugName A scalar containing the name of the serialized proto.
+   * May contain, for example, table key (descriptive) name for the
+   * corresponding serialized proto.  This is purely useful for debugging
+   * purposes, and the presence of values here has no effect on the output.
+   * May also be an empty scalar if no name is available.
+   * @param contextSparseTypes A list of Ncontext_sparse types; the data types of data in
+   * each context Feature given in context_sparse_keys.
+   * Currently the ParseSingleSequenceExample supports DT_FLOAT (FloatList),
+   * DT_INT64 (Int64List), and DT_STRING (BytesList).
+   * @param featureListDenseTypes 
+   * @param featureListSparseTypes A list of Nfeature_list_sparse types; the data types
+   * of data in each FeatureList given in feature_list_sparse_keys.
+   * Currently the ParseSingleSequenceExample supports DT_FLOAT (FloatList),
+   * DT_INT64 (Int64List), and DT_STRING (BytesList).
+   * @param options carries optional attributes values
+   * @return a new instance of ParseSingleSequenceExample
+   */
+  @Endpoint(describeByClass = true)
+  public static ParseSingleSequenceExample create(Scope scope, Operand<TString> serialized, Operand<TString> featureListDenseMissingAssumedEmpty, Iterable<Operand<TString>> contextSparseKeys, Iterable<Operand<TString>> contextDenseKeys, Iterable<Operand<TString>> featureListSparseKeys, Iterable<Operand<TString>> featureListDenseKeys, Iterable<Operand<?>> contextDenseDefaults, Operand<TString> debugName, List<Class<? extends TType>> contextSparseTypes, List<Class<? extends TType>> featureListDenseTypes, List<Class<? extends TType>> featureListSparseTypes, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ParseSingleSequenceExample", scope.makeOpName("ParseSingleSequenceExample"));
+    opBuilder.addInput(serialized.asOutput());
+    opBuilder.addInput(featureListDenseMissingAssumedEmpty.asOutput());
+    opBuilder.addInputList(Operands.asOutputs(contextSparseKeys));
+    opBuilder.addInputList(Operands.asOutputs(contextDenseKeys));
+    opBuilder.addInputList(Operands.asOutputs(featureListSparseKeys));
+    opBuilder.addInputList(Operands.asOutputs(featureListDenseKeys));
+    opBuilder.addInputList(Operands.asOutputs(contextDenseDefaults));
+    opBuilder.addInput(debugName.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    Class[] contextSparseTypesArray = new Class[contextSparseTypes.size()];
+    for (int i = 0; i < contextSparseTypesArray.length; ++i) {
+      contextSparseTypesArray[i] = contextSparseTypes.get(i);
+    }
+    opBuilder.setAttr("context_sparse_types", contextSparseTypesArray);
+    Class[] featureListDenseTypesArray = new Class[featureListDenseTypes.size()];
+    for (int i = 0; i < featureListDenseTypesArray.length; ++i) {
+      featureListDenseTypesArray[i] = featureListDenseTypes.get(i);
+    }
+    opBuilder.setAttr("feature_list_dense_types", featureListDenseTypesArray);
+    Class[] featureListSparseTypesArray = new Class[featureListSparseTypes.size()];
+    for (int i = 0; i < featureListSparseTypesArray.length; ++i) {
+      featureListSparseTypesArray[i] = featureListSparseTypes.get(i);
+    }
+    opBuilder.setAttr("feature_list_sparse_types", featureListSparseTypesArray);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.contextDenseShapes != null) {
+          Shape[] contextDenseShapesArray = new Shape[opts.contextDenseShapes.size()];
+          for (int i = 0; i < contextDenseShapesArray.length; ++i) {
+            contextDenseShapesArray[i] = opts.contextDenseShapes.get(i);
+          }
+          opBuilder.setAttr("context_dense_shapes", contextDenseShapesArray);
+        }
+        if (opts.featureListDenseShapes != null) {
+          Shape[] featureListDenseShapesArray = new Shape[opts.featureListDenseShapes.size()];
+          for (int i = 0; i < featureListDenseShapesArray.length; ++i) {
+            featureListDenseShapesArray[i] = opts.featureListDenseShapes.get(i);
+          }
+          opBuilder.setAttr("feature_list_dense_shapes", featureListDenseShapesArray);
+        }
+      }
+    }
+    return new ParseSingleSequenceExample(opBuilder.build());
+  }
+  
+  /**
+   * @param contextDenseShapes A list of Ncontext_dense shapes; the shapes of data in
+   * each context Feature given in context_dense_keys.
+   * The number of elements in the Feature corresponding to context_dense_key[j]
+   * must always equal context_dense_shapes[j].NumEntries().
+   * The shape of context_dense_values[j] will match context_dense_shapes[j].
+   */
+  public static Options contextDenseShapes(List<Shape> contextDenseShapes) {
+    return new Options().contextDenseShapes(contextDenseShapes);
+  }
+  
+  /**
+   * @param featureListDenseShapes A list of Nfeature_list_dense shapes; the shapes of
+   * data in each FeatureList given in feature_list_dense_keys.
+   * The shape of each Feature in the FeatureList corresponding to
+   * feature_list_dense_key[j] must always equal
+   * feature_list_dense_shapes[j].NumEntries().
+   */
+  public static Options featureListDenseShapes(List<Shape> featureListDenseShapes) {
+    return new Options().featureListDenseShapes(featureListDenseShapes);
+  }
+  
+  /**
+   */
+  public List<Output<TInt64>> contextSparseIndices() {
+    return contextSparseIndices;
+  }
+  
+  /**
+   */
+  public List<Output<?>> contextSparseValues() {
+    return contextSparseValues;
+  }
+  
+  /**
+   */
+  public List<Output<TInt64>> contextSparseShapes() {
+    return contextSparseShapes;
+  }
+  
+  /**
+   */
+  public List<Output<?>> contextDenseValues() {
+    return contextDenseValues;
+  }
+  
+  /**
+   */
+  public List<Output<TInt64>> featureListSparseIndices() {
+    return featureListSparseIndices;
+  }
+  
+  /**
+   */
+  public List<Output<?>> featureListSparseValues() {
+    return featureListSparseValues;
+  }
+  
+  /**
+   */
+  public List<Output<TInt64>> featureListSparseShapes() {
+    return featureListSparseShapes;
+  }
+  
+  /**
+   */
+  public List<Output<?>> featureListDenseValues() {
+    return featureListDenseValues;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ParseSingleSequenceExample";
+  
+  private List<Output<TInt64>> contextSparseIndices;
+  private List<Output<?>> contextSparseValues;
+  private List<Output<TInt64>> contextSparseShapes;
+  private List<Output<?>> contextDenseValues;
+  private List<Output<TInt64>> featureListSparseIndices;
+  private List<Output<?>> featureListSparseValues;
+  private List<Output<TInt64>> featureListSparseShapes;
+  private List<Output<?>> featureListDenseValues;
+  
+  @SuppressWarnings("unchecked")
+  private ParseSingleSequenceExample(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    int contextSparseIndicesLength = operation.outputListLength("context_sparse_indices");
+    contextSparseIndices = Arrays.asList((Output<TInt64>[])operation.outputList(outputIdx, contextSparseIndicesLength));
+    outputIdx += contextSparseIndicesLength;
+    int contextSparseValuesLength = operation.outputListLength("context_sparse_values");
+    contextSparseValues = Arrays.asList(operation.outputList(outputIdx, contextSparseValuesLength));
+    outputIdx += contextSparseValuesLength;
+    int contextSparseShapesLength = operation.outputListLength("context_sparse_shapes");
+    contextSparseShapes = Arrays.asList((Output<TInt64>[])operation.outputList(outputIdx, contextSparseShapesLength));
+    outputIdx += contextSparseShapesLength;
+    int contextDenseValuesLength = operation.outputListLength("context_dense_values");
+    contextDenseValues = Arrays.asList(operation.outputList(outputIdx, contextDenseValuesLength));
+    outputIdx += contextDenseValuesLength;
+    int featureListSparseIndicesLength = operation.outputListLength("feature_list_sparse_indices");
+    featureListSparseIndices = Arrays.asList((Output<TInt64>[])operation.outputList(outputIdx, featureListSparseIndicesLength));
+    outputIdx += featureListSparseIndicesLength;
+    int featureListSparseValuesLength = operation.outputListLength("feature_list_sparse_values");
+    featureListSparseValues = Arrays.asList(operation.outputList(outputIdx, featureListSparseValuesLength));
+    outputIdx += featureListSparseValuesLength;
+    int featureListSparseShapesLength = operation.outputListLength("feature_list_sparse_shapes");
+    featureListSparseShapes = Arrays.asList((Output<TInt64>[])operation.outputList(outputIdx, featureListSparseShapesLength));
+    outputIdx += featureListSparseShapesLength;
+    int featureListDenseValuesLength = operation.outputListLength("feature_list_dense_values");
+    featureListDenseValues = Arrays.asList(operation.outputList(outputIdx, featureListDenseValuesLength));
+    outputIdx += featureListDenseValuesLength;
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ParseTensor.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ParseTensor.java
new file mode 100644
index 00000000000..a45fa3a542f
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ParseTensor.java
@@ -0,0 +1,79 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.io;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TString;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Transforms a serialized tensorflow.TensorProto proto into a Tensor.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator(group = "io")
+public final class ParseTensor<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new ParseTensor operation.
+   * 
+   * @param scope current scope
+   * @param serialized A scalar string containing a serialized TensorProto proto.
+   * @param outType The type of the serialized tensor.  The provided type must match the
+   * type of the serialized tensor and no implicit conversion will take place.
+   * @return a new instance of ParseTensor
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> ParseTensor<T> create(Scope scope, Operand<TString> serialized, Class<T> outType) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ParseTensor", scope.makeOpName("ParseTensor"));
+    opBuilder.addInput(serialized.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("out_type", outType);
+    return new ParseTensor<T>(opBuilder.build());
+  }
+  
+  /**
+   * A Tensor of type `out_type`.
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ParseTensor";
+  
+  private Output<T> output;
+  
+  private ParseTensor(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/PriorityQueue.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/PriorityQueue.java
new file mode 100644
index 00000000000..b8aba81d587
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/PriorityQueue.java
@@ -0,0 +1,173 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.io;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * A queue that produces elements sorted by the first component value.
+ * <p>
+ * Note that the PriorityQueue requires the first component of any element
+ * to be a scalar int64, in addition to the other elements declared by
+ * component_types.  Therefore calls to Enqueue and EnqueueMany (resp. Dequeue
+ * and DequeueMany) on a PriorityQueue will all require (resp. output) one extra
+ * entry in their input (resp. output) lists.
+ */
+@Operator(group = "io")
+public final class PriorityQueue extends RawOp implements Operand<TType> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.io.PriorityQueue}
+   */
+  public static class Options {
+    
+    /**
+     * @param capacity The upper bound on the number of elements in this queue.
+     * Negative numbers mean no limit.
+     */
+    public Options capacity(Long capacity) {
+      this.capacity = capacity;
+      return this;
+    }
+    
+    /**
+     * @param container If non-empty, this queue is placed in the given container.
+     * Otherwise, a default container is used.
+     */
+    public Options container(String container) {
+      this.container = container;
+      return this;
+    }
+    
+    /**
+     * @param sharedName If non-empty, this queue will be shared under the given name
+     * across multiple sessions.
+     */
+    public Options sharedName(String sharedName) {
+      this.sharedName = sharedName;
+      return this;
+    }
+    
+    private Long capacity;
+    private String container;
+    private String sharedName;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new PriorityQueue operation.
+   * 
+   * @param scope current scope
+   * @param componentTypes The type of each component in a value.
+   * @param shapes The shape of each component in a value. The length of this attr must
+   * be either 0 or the same as the length of component_types. If the length of
+   * this attr is 0, the shapes of queue elements are not constrained, and
+   * only one element may be dequeued at a time.
+   * @param options carries optional attributes values
+   * @return a new instance of PriorityQueue
+   */
+  @Endpoint(describeByClass = true)
+  public static PriorityQueue create(Scope scope, List<Class<? extends TType>> componentTypes, List<Shape> shapes, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("PriorityQueueV2", scope.makeOpName("PriorityQueue"));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    Class[] componentTypesArray = new Class[componentTypes.size()];
+    for (int i = 0; i < componentTypesArray.length; ++i) {
+      componentTypesArray[i] = componentTypes.get(i);
+    }
+    opBuilder.setAttr("component_types", componentTypesArray);
+    Shape[] shapesArray = new Shape[shapes.size()];
+    for (int i = 0; i < shapesArray.length; ++i) {
+      shapesArray[i] = shapes.get(i);
+    }
+    opBuilder.setAttr("shapes", shapesArray);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.capacity != null) {
+          opBuilder.setAttr("capacity", opts.capacity);
+        }
+        if (opts.container != null) {
+          opBuilder.setAttr("container", opts.container);
+        }
+        if (opts.sharedName != null) {
+          opBuilder.setAttr("shared_name", opts.sharedName);
+        }
+      }
+    }
+    return new PriorityQueue(opBuilder.build());
+  }
+  
+  /**
+   * @param capacity The upper bound on the number of elements in this queue.
+   * Negative numbers mean no limit.
+   */
+  public static Options capacity(Long capacity) {
+    return new Options().capacity(capacity);
+  }
+  
+  /**
+   * @param container If non-empty, this queue is placed in the given container.
+   * Otherwise, a default container is used.
+   */
+  public static Options container(String container) {
+    return new Options().container(container);
+  }
+  
+  /**
+   * @param sharedName If non-empty, this queue will be shared under the given name
+   * across multiple sessions.
+   */
+  public static Options sharedName(String sharedName) {
+    return new Options().sharedName(sharedName);
+  }
+  
+  /**
+   * The handle to the queue.
+   */
+  public Output<?> handle() {
+    return handle;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) handle;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "PriorityQueueV2";
+  
+  private Output<?> handle;
+  
+  private PriorityQueue(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    handle = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/QueueClose.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/QueueClose.java
new file mode 100644
index 00000000000..ea7791f143e
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/QueueClose.java
@@ -0,0 +1,97 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.io;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+
+/**
+ * Closes the given queue.
+ * <p>
+ * This operation signals that no more elements will be enqueued in the
+ * given queue. Subsequent Enqueue(Many) operations will fail.
+ * Subsequent Dequeue(Many) operations will continue to succeed if
+ * sufficient elements remain in the queue. Subsequent Dequeue(Many)
+ * operations that would block will fail immediately.
+ */
+@Operator(group = "io")
+public final class QueueClose extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.io.QueueClose}
+   */
+  public static class Options {
+    
+    /**
+     * @param cancelPendingEnqueues If true, all pending enqueue requests that are
+     * blocked on the given queue will be canceled.
+     */
+    public Options cancelPendingEnqueues(Boolean cancelPendingEnqueues) {
+      this.cancelPendingEnqueues = cancelPendingEnqueues;
+      return this;
+    }
+    
+    private Boolean cancelPendingEnqueues;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new QueueClose operation.
+   * 
+   * @param scope current scope
+   * @param handle The handle to a queue.
+   * @param options carries optional attributes values
+   * @return a new instance of QueueClose
+   */
+  @Endpoint(describeByClass = true)
+  public static QueueClose create(Scope scope, Operand<?> handle, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("QueueCloseV2", scope.makeOpName("QueueClose"));
+    opBuilder.addInput(handle.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.cancelPendingEnqueues != null) {
+          opBuilder.setAttr("cancel_pending_enqueues", opts.cancelPendingEnqueues);
+        }
+      }
+    }
+    return new QueueClose(opBuilder.build());
+  }
+  
+  /**
+   * @param cancelPendingEnqueues If true, all pending enqueue requests that are
+   * blocked on the given queue will be canceled.
+   */
+  public static Options cancelPendingEnqueues(Boolean cancelPendingEnqueues) {
+    return new Options().cancelPendingEnqueues(cancelPendingEnqueues);
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "QueueCloseV2";
+  
+  private QueueClose(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/QueueDequeue.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/QueueDequeue.java
new file mode 100644
index 00000000000..bf0cce74ece
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/QueueDequeue.java
@@ -0,0 +1,130 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.io;
+
+import java.util.Arrays;
+import java.util.Iterator;
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Dequeues a tuple of one or more tensors from the given queue.
+ * <p>
+ * This operation has k outputs, where k is the number of components
+ * in the tuples stored in the given queue, and output i is the ith
+ * component of the dequeued tuple.
+ * <p>
+ * N.B. If the queue is empty, this operation will block until an element
+ * has been dequeued (or 'timeout_ms' elapses, if specified).
+ */
+@Operator(group = "io")
+public final class QueueDequeue extends RawOp implements Iterable<Operand<TType>> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.io.QueueDequeue}
+   */
+  public static class Options {
+    
+    /**
+     * @param timeoutMs If the queue is empty, this operation will block for up to
+     * timeout_ms milliseconds.
+     * Note: This option is not supported yet.
+     */
+    public Options timeoutMs(Long timeoutMs) {
+      this.timeoutMs = timeoutMs;
+      return this;
+    }
+    
+    private Long timeoutMs;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new QueueDequeue operation.
+   * 
+   * @param scope current scope
+   * @param handle The handle to a queue.
+   * @param componentTypes The type of each component in a tuple.
+   * @param options carries optional attributes values
+   * @return a new instance of QueueDequeue
+   */
+  @Endpoint(describeByClass = true)
+  public static QueueDequeue create(Scope scope, Operand<?> handle, List<Class<? extends TType>> componentTypes, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("QueueDequeueV2", scope.makeOpName("QueueDequeue"));
+    opBuilder.addInput(handle.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    Class[] componentTypesArray = new Class[componentTypes.size()];
+    for (int i = 0; i < componentTypesArray.length; ++i) {
+      componentTypesArray[i] = componentTypes.get(i);
+    }
+    opBuilder.setAttr("component_types", componentTypesArray);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.timeoutMs != null) {
+          opBuilder.setAttr("timeout_ms", opts.timeoutMs);
+        }
+      }
+    }
+    return new QueueDequeue(opBuilder.build());
+  }
+  
+  /**
+   * @param timeoutMs If the queue is empty, this operation will block for up to
+   * timeout_ms milliseconds.
+   * Note: This option is not supported yet.
+   */
+  public static Options timeoutMs(Long timeoutMs) {
+    return new Options().timeoutMs(timeoutMs);
+  }
+  
+  /**
+   * One or more tensors that were dequeued as a tuple.
+   */
+  public List<Output<?>> components() {
+    return components;
+  }
+  
+  @Override
+  @SuppressWarnings({"rawtypes", "unchecked"})
+  public Iterator<Operand<TType>> iterator() {
+    return (Iterator) components.iterator();
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "QueueDequeueV2";
+  
+  private List<Output<?>> components;
+  
+  private QueueDequeue(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    int componentsLength = operation.outputListLength("components");
+    components = Arrays.asList(operation.outputList(outputIdx, componentsLength));
+    outputIdx += componentsLength;
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/QueueDequeueMany.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/QueueDequeueMany.java
new file mode 100644
index 00000000000..c24cb4251f6
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/QueueDequeueMany.java
@@ -0,0 +1,140 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.io;
+
+import java.util.Arrays;
+import java.util.Iterator;
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Dequeues `n` tuples of one or more tensors from the given queue.
+ * <p>
+ * If the queue is closed and there are fewer than `n` elements, then an
+ * OutOfRange error is returned.
+ * <p>
+ * This operation concatenates queue-element component tensors along the
+ * 0th dimension to make a single component tensor.  All of the components
+ * in the dequeued tuple will have size `n` in the 0th dimension.
+ * <p>
+ * This operation has `k` outputs, where `k` is the number of components in
+ * the tuples stored in the given queue, and output `i` is the ith
+ * component of the dequeued tuple.
+ * <p>
+ * N.B. If the queue is empty, this operation will block until `n` elements
+ * have been dequeued (or 'timeout_ms' elapses, if specified).
+ */
+@Operator(group = "io")
+public final class QueueDequeueMany extends RawOp implements Iterable<Operand<TType>> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.io.QueueDequeueMany}
+   */
+  public static class Options {
+    
+    /**
+     * @param timeoutMs If the queue has fewer than n elements, this operation
+     * will block for up to timeout_ms milliseconds.
+     * Note: This option is not supported yet.
+     */
+    public Options timeoutMs(Long timeoutMs) {
+      this.timeoutMs = timeoutMs;
+      return this;
+    }
+    
+    private Long timeoutMs;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new QueueDequeueMany operation.
+   * 
+   * @param scope current scope
+   * @param handle The handle to a queue.
+   * @param n The number of tuples to dequeue.
+   * @param componentTypes The type of each component in a tuple.
+   * @param options carries optional attributes values
+   * @return a new instance of QueueDequeueMany
+   */
+  @Endpoint(describeByClass = true)
+  public static QueueDequeueMany create(Scope scope, Operand<?> handle, Operand<TInt32> n, List<Class<? extends TType>> componentTypes, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("QueueDequeueManyV2", scope.makeOpName("QueueDequeueMany"));
+    opBuilder.addInput(handle.asOutput());
+    opBuilder.addInput(n.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    Class[] componentTypesArray = new Class[componentTypes.size()];
+    for (int i = 0; i < componentTypesArray.length; ++i) {
+      componentTypesArray[i] = componentTypes.get(i);
+    }
+    opBuilder.setAttr("component_types", componentTypesArray);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.timeoutMs != null) {
+          opBuilder.setAttr("timeout_ms", opts.timeoutMs);
+        }
+      }
+    }
+    return new QueueDequeueMany(opBuilder.build());
+  }
+  
+  /**
+   * @param timeoutMs If the queue has fewer than n elements, this operation
+   * will block for up to timeout_ms milliseconds.
+   * Note: This option is not supported yet.
+   */
+  public static Options timeoutMs(Long timeoutMs) {
+    return new Options().timeoutMs(timeoutMs);
+  }
+  
+  /**
+   * One or more tensors that were dequeued as a tuple.
+   */
+  public List<Output<?>> components() {
+    return components;
+  }
+  
+  @Override
+  @SuppressWarnings({"rawtypes", "unchecked"})
+  public Iterator<Operand<TType>> iterator() {
+    return (Iterator) components.iterator();
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "QueueDequeueManyV2";
+  
+  private List<Output<?>> components;
+  
+  private QueueDequeueMany(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    int componentsLength = operation.outputListLength("components");
+    components = Arrays.asList(operation.outputList(outputIdx, componentsLength));
+    outputIdx += componentsLength;
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/QueueDequeueUpTo.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/QueueDequeueUpTo.java
new file mode 100644
index 00000000000..d2302e77dea
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/QueueDequeueUpTo.java
@@ -0,0 +1,144 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.io;
+
+import java.util.Arrays;
+import java.util.Iterator;
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Dequeues `n` tuples of one or more tensors from the given queue.
+ * <p>
+ * This operation is not supported by all queues.  If a queue does not support
+ * DequeueUpTo, then an Unimplemented error is returned.
+ * <p>
+ * If the queue is closed and there are more than 0 but less than `n`
+ * elements remaining, then instead of returning an OutOfRange error like
+ * QueueDequeueMany, less than `n` elements are returned immediately.  If
+ * the queue is closed and there are 0 elements left in the queue, then
+ * an OutOfRange error is returned just like in QueueDequeueMany.
+ * Otherwise the behavior is identical to QueueDequeueMany:
+ * <p>
+ * This operation concatenates queue-element component tensors along the
+ * 0th dimension to make a single component tensor.  All of the components
+ * in the dequeued tuple will have size n in the 0th dimension.
+ * <p>
+ * This operation has `k` outputs, where `k` is the number of components in
+ * the tuples stored in the given queue, and output `i` is the ith
+ * component of the dequeued tuple.
+ */
+@Operator(group = "io")
+public final class QueueDequeueUpTo extends RawOp implements Iterable<Operand<TType>> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.io.QueueDequeueUpTo}
+   */
+  public static class Options {
+    
+    /**
+     * @param timeoutMs If the queue has fewer than n elements, this operation
+     * will block for up to timeout_ms milliseconds.
+     * Note: This option is not supported yet.
+     */
+    public Options timeoutMs(Long timeoutMs) {
+      this.timeoutMs = timeoutMs;
+      return this;
+    }
+    
+    private Long timeoutMs;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new QueueDequeueUpTo operation.
+   * 
+   * @param scope current scope
+   * @param handle The handle to a queue.
+   * @param n The number of tuples to dequeue.
+   * @param componentTypes The type of each component in a tuple.
+   * @param options carries optional attributes values
+   * @return a new instance of QueueDequeueUpTo
+   */
+  @Endpoint(describeByClass = true)
+  public static QueueDequeueUpTo create(Scope scope, Operand<?> handle, Operand<TInt32> n, List<Class<? extends TType>> componentTypes, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("QueueDequeueUpToV2", scope.makeOpName("QueueDequeueUpTo"));
+    opBuilder.addInput(handle.asOutput());
+    opBuilder.addInput(n.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    Class[] componentTypesArray = new Class[componentTypes.size()];
+    for (int i = 0; i < componentTypesArray.length; ++i) {
+      componentTypesArray[i] = componentTypes.get(i);
+    }
+    opBuilder.setAttr("component_types", componentTypesArray);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.timeoutMs != null) {
+          opBuilder.setAttr("timeout_ms", opts.timeoutMs);
+        }
+      }
+    }
+    return new QueueDequeueUpTo(opBuilder.build());
+  }
+  
+  /**
+   * @param timeoutMs If the queue has fewer than n elements, this operation
+   * will block for up to timeout_ms milliseconds.
+   * Note: This option is not supported yet.
+   */
+  public static Options timeoutMs(Long timeoutMs) {
+    return new Options().timeoutMs(timeoutMs);
+  }
+  
+  /**
+   * One or more tensors that were dequeued as a tuple.
+   */
+  public List<Output<?>> components() {
+    return components;
+  }
+  
+  @Override
+  @SuppressWarnings({"rawtypes", "unchecked"})
+  public Iterator<Operand<TType>> iterator() {
+    return (Iterator) components.iterator();
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "QueueDequeueUpToV2";
+  
+  private List<Output<?>> components;
+  
+  private QueueDequeueUpTo(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    int componentsLength = operation.outputListLength("components");
+    components = Arrays.asList(operation.outputList(outputIdx, componentsLength));
+    outputIdx += componentsLength;
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/QueueEnqueue.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/QueueEnqueue.java
new file mode 100644
index 00000000000..a159b0cd17c
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/QueueEnqueue.java
@@ -0,0 +1,102 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.io;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.Operands;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+
+/**
+ * Enqueues a tuple of one or more tensors in the given queue.
+ * <p>
+ * The components input has k elements, which correspond to the components of
+ * tuples stored in the given queue.
+ * <p>
+ * N.B. If the queue is full, this operation will block until the given
+ * element has been enqueued (or 'timeout_ms' elapses, if specified).
+ */
+@Operator(group = "io")
+public final class QueueEnqueue extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.io.QueueEnqueue}
+   */
+  public static class Options {
+    
+    /**
+     * @param timeoutMs If the queue is full, this operation will block for up to
+     * timeout_ms milliseconds.
+     * Note: This option is not supported yet.
+     */
+    public Options timeoutMs(Long timeoutMs) {
+      this.timeoutMs = timeoutMs;
+      return this;
+    }
+    
+    private Long timeoutMs;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new QueueEnqueue operation.
+   * 
+   * @param scope current scope
+   * @param handle The handle to a queue.
+   * @param components One or more tensors from which the enqueued tensors should be taken.
+   * @param options carries optional attributes values
+   * @return a new instance of QueueEnqueue
+   */
+  @Endpoint(describeByClass = true)
+  public static QueueEnqueue create(Scope scope, Operand<?> handle, Iterable<Operand<?>> components, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("QueueEnqueueV2", scope.makeOpName("QueueEnqueue"));
+    opBuilder.addInput(handle.asOutput());
+    opBuilder.addInputList(Operands.asOutputs(components));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.timeoutMs != null) {
+          opBuilder.setAttr("timeout_ms", opts.timeoutMs);
+        }
+      }
+    }
+    return new QueueEnqueue(opBuilder.build());
+  }
+  
+  /**
+   * @param timeoutMs If the queue is full, this operation will block for up to
+   * timeout_ms milliseconds.
+   * Note: This option is not supported yet.
+   */
+  public static Options timeoutMs(Long timeoutMs) {
+    return new Options().timeoutMs(timeoutMs);
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "QueueEnqueueV2";
+  
+  private QueueEnqueue(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/QueueEnqueueMany.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/QueueEnqueueMany.java
new file mode 100644
index 00000000000..b1f9cbd6807
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/QueueEnqueueMany.java
@@ -0,0 +1,107 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.io;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.Operands;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+
+/**
+ * Enqueues zero or more tuples of one or more tensors in the given queue.
+ * <p>
+ * This operation slices each component tensor along the 0th dimension to
+ * make multiple queue elements. All of the tuple components must have the
+ * same size in the 0th dimension.
+ * <p>
+ * The components input has k elements, which correspond to the components of
+ * tuples stored in the given queue.
+ * <p>
+ * N.B. If the queue is full, this operation will block until the given
+ * elements have been enqueued (or 'timeout_ms' elapses, if specified).
+ */
+@Operator(group = "io")
+public final class QueueEnqueueMany extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.io.QueueEnqueueMany}
+   */
+  public static class Options {
+    
+    /**
+     * @param timeoutMs If the queue is too full, this operation will block for up
+     * to timeout_ms milliseconds.
+     * Note: This option is not supported yet.
+     */
+    public Options timeoutMs(Long timeoutMs) {
+      this.timeoutMs = timeoutMs;
+      return this;
+    }
+    
+    private Long timeoutMs;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new QueueEnqueueMany operation.
+   * 
+   * @param scope current scope
+   * @param handle The handle to a queue.
+   * @param components One or more tensors from which the enqueued tensors should
+   * be taken.
+   * @param options carries optional attributes values
+   * @return a new instance of QueueEnqueueMany
+   */
+  @Endpoint(describeByClass = true)
+  public static QueueEnqueueMany create(Scope scope, Operand<?> handle, Iterable<Operand<?>> components, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("QueueEnqueueManyV2", scope.makeOpName("QueueEnqueueMany"));
+    opBuilder.addInput(handle.asOutput());
+    opBuilder.addInputList(Operands.asOutputs(components));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.timeoutMs != null) {
+          opBuilder.setAttr("timeout_ms", opts.timeoutMs);
+        }
+      }
+    }
+    return new QueueEnqueueMany(opBuilder.build());
+  }
+  
+  /**
+   * @param timeoutMs If the queue is too full, this operation will block for up
+   * to timeout_ms milliseconds.
+   * Note: This option is not supported yet.
+   */
+  public static Options timeoutMs(Long timeoutMs) {
+    return new Options().timeoutMs(timeoutMs);
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "QueueEnqueueManyV2";
+  
+  private QueueEnqueueMany(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/QueueIsClosed.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/QueueIsClosed.java
new file mode 100644
index 00000000000..895965a3c6c
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/QueueIsClosed.java
@@ -0,0 +1,75 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.io;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TBool;
+
+/**
+ * Returns true if queue is closed.
+ * <p>
+ * This operation returns true if the queue is closed and false if the queue
+ * is open.
+ */
+@Operator(group = "io")
+public final class QueueIsClosed extends RawOp implements Operand<TBool> {
+  
+  /**
+   * Factory method to create a class wrapping a new QueueIsClosed operation.
+   * 
+   * @param scope current scope
+   * @param handle The handle to a queue.
+   * @return a new instance of QueueIsClosed
+   */
+  @Endpoint(describeByClass = true)
+  public static QueueIsClosed create(Scope scope, Operand<?> handle) {
+    OperationBuilder opBuilder = scope.env().opBuilder("QueueIsClosedV2", scope.makeOpName("QueueIsClosed"));
+    opBuilder.addInput(handle.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new QueueIsClosed(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<TBool> isClosed() {
+    return isClosed;
+  }
+  
+  @Override
+  public Output<TBool> asOutput() {
+    return isClosed;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "QueueIsClosedV2";
+  
+  private Output<TBool> isClosed;
+  
+  private QueueIsClosed(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    isClosed = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/QueueSize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/QueueSize.java
new file mode 100644
index 00000000000..cefc1b3fd11
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/QueueSize.java
@@ -0,0 +1,73 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.io;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+
+/**
+ * Computes the number of elements in the given queue.
+ */
+@Operator(group = "io")
+public final class QueueSize extends RawOp implements Operand<TInt32> {
+  
+  /**
+   * Factory method to create a class wrapping a new QueueSize operation.
+   * 
+   * @param scope current scope
+   * @param handle The handle to a queue.
+   * @return a new instance of QueueSize
+   */
+  @Endpoint(describeByClass = true)
+  public static QueueSize create(Scope scope, Operand<?> handle) {
+    OperationBuilder opBuilder = scope.env().opBuilder("QueueSizeV2", scope.makeOpName("QueueSize"));
+    opBuilder.addInput(handle.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new QueueSize(opBuilder.build());
+  }
+  
+  /**
+   * The number of elements in the given queue.
+   */
+  public Output<TInt32> size() {
+    return size;
+  }
+  
+  @Override
+  public Output<TInt32> asOutput() {
+    return size;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "QueueSizeV2";
+  
+  private Output<TInt32> size;
+  
+  private QueueSize(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    size = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/RandomShuffleQueue.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/RandomShuffleQueue.java
new file mode 100644
index 00000000000..eb2d6a4ad52
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/RandomShuffleQueue.java
@@ -0,0 +1,250 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.io;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * A queue that randomizes the order of elements.
+ */
+@Operator(group = "io")
+public final class RandomShuffleQueue extends RawOp implements Operand<TType> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.io.RandomShuffleQueue}
+   */
+  public static class Options {
+    
+    /**
+     * @param shapes The shape of each component in a value. The length of this attr must
+     * be either 0 or the same as the length of component_types. If the length of
+     * this attr is 0, the shapes of queue elements are not constrained, and
+     * only one element may be dequeued at a time.
+     */
+    public Options shapes(List<Shape> shapes) {
+      this.shapes = shapes;
+      return this;
+    }
+    
+    /**
+     * @param capacity The upper bound on the number of elements in this queue.
+     * Negative numbers mean no limit.
+     */
+    public Options capacity(Long capacity) {
+      this.capacity = capacity;
+      return this;
+    }
+    
+    /**
+     * @param minAfterDequeue Dequeue will block unless there would be this
+     * many elements after the dequeue or the queue is closed. This
+     * ensures a minimum level of mixing of elements.
+     */
+    public Options minAfterDequeue(Long minAfterDequeue) {
+      this.minAfterDequeue = minAfterDequeue;
+      return this;
+    }
+    
+    /**
+     * @param seed If either seed or seed2 is set to be non-zero, the random number
+     * generator is seeded by the given seed.  Otherwise, a random seed is used.
+     */
+    public Options seed(Long seed) {
+      this.seed = seed;
+      return this;
+    }
+    
+    /**
+     * @param seed2 A second seed to avoid seed collision.
+     */
+    public Options seed2(Long seed2) {
+      this.seed2 = seed2;
+      return this;
+    }
+    
+    /**
+     * @param container If non-empty, this queue is placed in the given container.
+     * Otherwise, a default container is used.
+     */
+    public Options container(String container) {
+      this.container = container;
+      return this;
+    }
+    
+    /**
+     * @param sharedName If non-empty, this queue will be shared under the given name
+     * across multiple sessions.
+     */
+    public Options sharedName(String sharedName) {
+      this.sharedName = sharedName;
+      return this;
+    }
+    
+    private List<Shape> shapes;
+    private Long capacity;
+    private Long minAfterDequeue;
+    private Long seed;
+    private Long seed2;
+    private String container;
+    private String sharedName;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new RandomShuffleQueue operation.
+   * 
+   * @param scope current scope
+   * @param componentTypes The type of each component in a value.
+   * @param options carries optional attributes values
+   * @return a new instance of RandomShuffleQueue
+   */
+  @Endpoint(describeByClass = true)
+  public static RandomShuffleQueue create(Scope scope, List<Class<? extends TType>> componentTypes, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("RandomShuffleQueueV2", scope.makeOpName("RandomShuffleQueue"));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    Class[] componentTypesArray = new Class[componentTypes.size()];
+    for (int i = 0; i < componentTypesArray.length; ++i) {
+      componentTypesArray[i] = componentTypes.get(i);
+    }
+    opBuilder.setAttr("component_types", componentTypesArray);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.shapes != null) {
+          Shape[] shapesArray = new Shape[opts.shapes.size()];
+          for (int i = 0; i < shapesArray.length; ++i) {
+            shapesArray[i] = opts.shapes.get(i);
+          }
+          opBuilder.setAttr("shapes", shapesArray);
+        }
+        if (opts.capacity != null) {
+          opBuilder.setAttr("capacity", opts.capacity);
+        }
+        if (opts.minAfterDequeue != null) {
+          opBuilder.setAttr("min_after_dequeue", opts.minAfterDequeue);
+        }
+        if (opts.seed != null) {
+          opBuilder.setAttr("seed", opts.seed);
+        }
+        if (opts.seed2 != null) {
+          opBuilder.setAttr("seed2", opts.seed2);
+        }
+        if (opts.container != null) {
+          opBuilder.setAttr("container", opts.container);
+        }
+        if (opts.sharedName != null) {
+          opBuilder.setAttr("shared_name", opts.sharedName);
+        }
+      }
+    }
+    return new RandomShuffleQueue(opBuilder.build());
+  }
+  
+  /**
+   * @param shapes The shape of each component in a value. The length of this attr must
+   * be either 0 or the same as the length of component_types. If the length of
+   * this attr is 0, the shapes of queue elements are not constrained, and
+   * only one element may be dequeued at a time.
+   */
+  public static Options shapes(List<Shape> shapes) {
+    return new Options().shapes(shapes);
+  }
+  
+  /**
+   * @param capacity The upper bound on the number of elements in this queue.
+   * Negative numbers mean no limit.
+   */
+  public static Options capacity(Long capacity) {
+    return new Options().capacity(capacity);
+  }
+  
+  /**
+   * @param minAfterDequeue Dequeue will block unless there would be this
+   * many elements after the dequeue or the queue is closed. This
+   * ensures a minimum level of mixing of elements.
+   */
+  public static Options minAfterDequeue(Long minAfterDequeue) {
+    return new Options().minAfterDequeue(minAfterDequeue);
+  }
+  
+  /**
+   * @param seed If either seed or seed2 is set to be non-zero, the random number
+   * generator is seeded by the given seed.  Otherwise, a random seed is used.
+   */
+  public static Options seed(Long seed) {
+    return new Options().seed(seed);
+  }
+  
+  /**
+   * @param seed2 A second seed to avoid seed collision.
+   */
+  public static Options seed2(Long seed2) {
+    return new Options().seed2(seed2);
+  }
+  
+  /**
+   * @param container If non-empty, this queue is placed in the given container.
+   * Otherwise, a default container is used.
+   */
+  public static Options container(String container) {
+    return new Options().container(container);
+  }
+  
+  /**
+   * @param sharedName If non-empty, this queue will be shared under the given name
+   * across multiple sessions.
+   */
+  public static Options sharedName(String sharedName) {
+    return new Options().sharedName(sharedName);
+  }
+  
+  /**
+   * The handle to the queue.
+   */
+  public Output<?> handle() {
+    return handle;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) handle;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "RandomShuffleQueueV2";
+  
+  private Output<?> handle;
+  
+  private RandomShuffleQueue(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    handle = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ReadFile.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ReadFile.java
new file mode 100644
index 00000000000..7109022751a
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ReadFile.java
@@ -0,0 +1,72 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.io;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TString;
+
+/**
+ * Reads and outputs the entire contents of the input filename.
+ */
+@Operator(group = "io")
+public final class ReadFile extends RawOp implements Operand<TString> {
+  
+  /**
+   * Factory method to create a class wrapping a new ReadFile operation.
+   * 
+   * @param scope current scope
+   * @param filename 
+   * @return a new instance of ReadFile
+   */
+  @Endpoint(describeByClass = true)
+  public static ReadFile create(Scope scope, Operand<TString> filename) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ReadFile", scope.makeOpName("ReadFile"));
+    opBuilder.addInput(filename.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new ReadFile(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<TString> contents() {
+    return contents;
+  }
+  
+  @Override
+  public Output<TString> asOutput() {
+    return contents;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ReadFile";
+  
+  private Output<TString> contents;
+  
+  private ReadFile(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    contents = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ReaderNumRecordsProduced.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ReaderNumRecordsProduced.java
new file mode 100644
index 00000000000..8e4cb9c5692
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ReaderNumRecordsProduced.java
@@ -0,0 +1,75 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.io;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt64;
+
+/**
+ * Returns the number of records this Reader has produced.
+ * <p>
+ * This is the same as the number of ReaderRead executions that have
+ * succeeded.
+ */
+@Operator(group = "io")
+public final class ReaderNumRecordsProduced extends RawOp implements Operand<TInt64> {
+  
+  /**
+   * Factory method to create a class wrapping a new ReaderNumRecordsProduced operation.
+   * 
+   * @param scope current scope
+   * @param readerHandle Handle to a Reader.
+   * @return a new instance of ReaderNumRecordsProduced
+   */
+  @Endpoint(describeByClass = true)
+  public static ReaderNumRecordsProduced create(Scope scope, Operand<?> readerHandle) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ReaderNumRecordsProducedV2", scope.makeOpName("ReaderNumRecordsProduced"));
+    opBuilder.addInput(readerHandle.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new ReaderNumRecordsProduced(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<TInt64> recordsProduced() {
+    return recordsProduced;
+  }
+  
+  @Override
+  public Output<TInt64> asOutput() {
+    return recordsProduced;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ReaderNumRecordsProducedV2";
+  
+  private Output<TInt64> recordsProduced;
+  
+  private ReaderNumRecordsProduced(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    recordsProduced = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ReaderNumWorkUnitsCompleted.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ReaderNumWorkUnitsCompleted.java
new file mode 100644
index 00000000000..bbd6fc8075c
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ReaderNumWorkUnitsCompleted.java
@@ -0,0 +1,72 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.io;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt64;
+
+/**
+ * Returns the number of work units this Reader has finished processing.
+ */
+@Operator(group = "io")
+public final class ReaderNumWorkUnitsCompleted extends RawOp implements Operand<TInt64> {
+  
+  /**
+   * Factory method to create a class wrapping a new ReaderNumWorkUnitsCompleted operation.
+   * 
+   * @param scope current scope
+   * @param readerHandle Handle to a Reader.
+   * @return a new instance of ReaderNumWorkUnitsCompleted
+   */
+  @Endpoint(describeByClass = true)
+  public static ReaderNumWorkUnitsCompleted create(Scope scope, Operand<?> readerHandle) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ReaderNumWorkUnitsCompletedV2", scope.makeOpName("ReaderNumWorkUnitsCompleted"));
+    opBuilder.addInput(readerHandle.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new ReaderNumWorkUnitsCompleted(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<TInt64> unitsCompleted() {
+    return unitsCompleted;
+  }
+  
+  @Override
+  public Output<TInt64> asOutput() {
+    return unitsCompleted;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ReaderNumWorkUnitsCompletedV2";
+  
+  private Output<TInt64> unitsCompleted;
+  
+  private ReaderNumWorkUnitsCompleted(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    unitsCompleted = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ReaderRead.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ReaderRead.java
new file mode 100644
index 00000000000..9b7110b10ae
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ReaderRead.java
@@ -0,0 +1,83 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.io;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TString;
+
+/**
+ * Returns the next record (key, value pair) produced by a Reader.
+ * <p>
+ * Will dequeue from the input queue if necessary (e.g. when the
+ * Reader needs to start reading from a new file since it has finished
+ * with the previous file).
+ */
+@Operator(group = "io")
+public final class ReaderRead extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new ReaderRead operation.
+   * 
+   * @param scope current scope
+   * @param readerHandle Handle to a Reader.
+   * @param queueHandle Handle to a Queue, with string work items.
+   * @return a new instance of ReaderRead
+   */
+  @Endpoint(describeByClass = true)
+  public static ReaderRead create(Scope scope, Operand<?> readerHandle, Operand<?> queueHandle) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ReaderReadV2", scope.makeOpName("ReaderRead"));
+    opBuilder.addInput(readerHandle.asOutput());
+    opBuilder.addInput(queueHandle.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new ReaderRead(opBuilder.build());
+  }
+  
+  /**
+   * A scalar.
+   */
+  public Output<TString> key() {
+    return key;
+  }
+  
+  /**
+   * A scalar.
+   */
+  public Output<TString> value() {
+    return value;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ReaderReadV2";
+  
+  private Output<TString> key;
+  private Output<TString> value;
+  
+  private ReaderRead(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    key = operation.output(outputIdx++);
+    value = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ReaderReadUpTo.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ReaderReadUpTo.java
new file mode 100644
index 00000000000..e22d5b95576
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ReaderReadUpTo.java
@@ -0,0 +1,87 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.io;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.TString;
+
+/**
+ * Returns up to `num_records` (key, value) pairs produced by a Reader.
+ * <p>
+ * Will dequeue from the input queue if necessary (e.g. when the
+ * Reader needs to start reading from a new file since it has finished
+ * with the previous file).
+ * It may return less than `num_records` even before the last batch.
+ */
+@Operator(group = "io")
+public final class ReaderReadUpTo extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new ReaderReadUpTo operation.
+   * 
+   * @param scope current scope
+   * @param readerHandle Handle to a `Reader`.
+   * @param queueHandle Handle to a `Queue`, with string work items.
+   * @param numRecords number of records to read from `Reader`.
+   * @return a new instance of ReaderReadUpTo
+   */
+  @Endpoint(describeByClass = true)
+  public static ReaderReadUpTo create(Scope scope, Operand<?> readerHandle, Operand<?> queueHandle, Operand<TInt64> numRecords) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ReaderReadUpToV2", scope.makeOpName("ReaderReadUpTo"));
+    opBuilder.addInput(readerHandle.asOutput());
+    opBuilder.addInput(queueHandle.asOutput());
+    opBuilder.addInput(numRecords.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new ReaderReadUpTo(opBuilder.build());
+  }
+  
+  /**
+   * A 1-D tensor.
+   */
+  public Output<TString> keys() {
+    return keys;
+  }
+  
+  /**
+   * A 1-D tensor.
+   */
+  public Output<TString> values() {
+    return values;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ReaderReadUpToV2";
+  
+  private Output<TString> keys;
+  private Output<TString> values;
+  
+  private ReaderReadUpTo(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    keys = operation.output(outputIdx++);
+    values = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ReaderReset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ReaderReset.java
new file mode 100644
index 00000000000..243d4a72080
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ReaderReset.java
@@ -0,0 +1,55 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.io;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+
+/**
+ * Restore a Reader to its initial clean state.
+ */
+@Operator(group = "io")
+public final class ReaderReset extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new ReaderReset operation.
+   * 
+   * @param scope current scope
+   * @param readerHandle Handle to a Reader.
+   * @return a new instance of ReaderReset
+   */
+  @Endpoint(describeByClass = true)
+  public static ReaderReset create(Scope scope, Operand<?> readerHandle) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ReaderResetV2", scope.makeOpName("ReaderReset"));
+    opBuilder.addInput(readerHandle.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new ReaderReset(opBuilder.build());
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ReaderResetV2";
+  
+  private ReaderReset(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ReaderRestoreState.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ReaderRestoreState.java
new file mode 100644
index 00000000000..431ba079ffc
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ReaderRestoreState.java
@@ -0,0 +1,62 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.io;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TString;
+
+/**
+ * Restore a reader to a previously saved state.
+ * <p>
+ * Not all Readers support being restored, so this can produce an
+ * Unimplemented error.
+ */
+@Operator(group = "io")
+public final class ReaderRestoreState extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new ReaderRestoreState operation.
+   * 
+   * @param scope current scope
+   * @param readerHandle Handle to a Reader.
+   * @param state Result of a ReaderSerializeState of a Reader with type
+   * matching reader_handle.
+   * @return a new instance of ReaderRestoreState
+   */
+  @Endpoint(describeByClass = true)
+  public static ReaderRestoreState create(Scope scope, Operand<?> readerHandle, Operand<TString> state) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ReaderRestoreStateV2", scope.makeOpName("ReaderRestoreState"));
+    opBuilder.addInput(readerHandle.asOutput());
+    opBuilder.addInput(state.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new ReaderRestoreState(opBuilder.build());
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ReaderRestoreStateV2";
+  
+  private ReaderRestoreState(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ReaderSerializeState.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ReaderSerializeState.java
new file mode 100644
index 00000000000..72b927ed171
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ReaderSerializeState.java
@@ -0,0 +1,75 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.io;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TString;
+
+/**
+ * Produce a string tensor that encodes the state of a Reader.
+ * <p>
+ * Not all Readers support being serialized, so this can produce an
+ * Unimplemented error.
+ */
+@Operator(group = "io")
+public final class ReaderSerializeState extends RawOp implements Operand<TString> {
+  
+  /**
+   * Factory method to create a class wrapping a new ReaderSerializeState operation.
+   * 
+   * @param scope current scope
+   * @param readerHandle Handle to a Reader.
+   * @return a new instance of ReaderSerializeState
+   */
+  @Endpoint(describeByClass = true)
+  public static ReaderSerializeState create(Scope scope, Operand<?> readerHandle) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ReaderSerializeStateV2", scope.makeOpName("ReaderSerializeState"));
+    opBuilder.addInput(readerHandle.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new ReaderSerializeState(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<TString> state() {
+    return state;
+  }
+  
+  @Override
+  public Output<TString> asOutput() {
+    return state;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ReaderSerializeStateV2";
+  
+  private Output<TString> state;
+  
+  private ReaderSerializeState(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    state = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/SerializeManySparse.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/SerializeManySparse.java
new file mode 100644
index 00000000000..35b1bbfae09
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/SerializeManySparse.java
@@ -0,0 +1,105 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.io;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.TString;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Serialize an `N`-minibatch `SparseTensor` into an `[N, 3]` `Tensor` object.
+ * <p>
+ * The `SparseTensor` must have rank `R` greater than 1, and the first dimension
+ * is treated as the minibatch dimension.  Elements of the `SparseTensor`
+ * must be sorted in increasing order of this first dimension.  The serialized
+ * `SparseTensor` objects going into each row of `serialized_sparse` will have
+ * rank `R-1`.
+ * <p>
+ * The minibatch size `N` is extracted from `sparse_shape[0]`.
+ * 
+ * @param <U> data type for {@code serializedSparse()} output
+ */
+@Operator(group = "io")
+public final class SerializeManySparse<U extends TType> extends RawOp implements Operand<U> {
+  
+  /**
+   * Factory method to create a class wrapping a new SerializeManySparse operation.
+   * 
+   * @param scope current scope
+   * @param sparseIndices 2-D.  The `indices` of the minibatch `SparseTensor`.
+   * @param sparseValues 1-D.  The `values` of the minibatch `SparseTensor`.
+   * @param sparseShape 1-D.  The `shape` of the minibatch `SparseTensor`.
+   * @param outType The `dtype` to use for serialization; the supported types are `string`
+   * (default) and `variant`.
+   * @return a new instance of SerializeManySparse
+   */
+  @Endpoint(describeByClass = true)
+  public static <U extends TType, T extends TType> SerializeManySparse<U> create(Scope scope, Operand<TInt64> sparseIndices, Operand<T> sparseValues, Operand<TInt64> sparseShape, Class<U> outType) {
+    OperationBuilder opBuilder = scope.env().opBuilder("SerializeManySparse", scope.makeOpName("SerializeManySparse"));
+    opBuilder.addInput(sparseIndices.asOutput());
+    opBuilder.addInput(sparseValues.asOutput());
+    opBuilder.addInput(sparseShape.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("out_type", outType);
+    return new SerializeManySparse<U>(opBuilder.build());
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new SerializeManySparse operation using default output types.
+   * 
+   * @param scope current scope
+   * @param sparseIndices 2-D.  The `indices` of the minibatch `SparseTensor`.
+   * @param sparseValues 1-D.  The `values` of the minibatch `SparseTensor`.
+   * @param sparseShape 1-D.  The `shape` of the minibatch `SparseTensor`.
+   * @return a new instance of SerializeManySparse
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> SerializeManySparse<TString> create(Scope scope, Operand<TInt64> sparseIndices, Operand<T> sparseValues, Operand<TInt64> sparseShape) {
+    return create(scope, sparseIndices, sparseValues, sparseShape, TString.class);
+  }
+  
+  /**
+   */
+  public Output<U> serializedSparse() {
+    return serializedSparse;
+  }
+  
+  @Override
+  public Output<U> asOutput() {
+    return serializedSparse;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "SerializeManySparse";
+  
+  private Output<U> serializedSparse;
+  
+  private SerializeManySparse(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    serializedSparse = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/SerializeSparse.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/SerializeSparse.java
new file mode 100644
index 00000000000..fc4e5c09f21
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/SerializeSparse.java
@@ -0,0 +1,97 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.io;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.TString;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Serialize a `SparseTensor` into a `[3]` `Tensor` object.
+ * 
+ * @param <U> data type for {@code serializedSparse()} output
+ */
+@Operator(group = "io")
+public final class SerializeSparse<U extends TType> extends RawOp implements Operand<U> {
+  
+  /**
+   * Factory method to create a class wrapping a new SerializeSparse operation.
+   * 
+   * @param scope current scope
+   * @param sparseIndices 2-D.  The `indices` of the `SparseTensor`.
+   * @param sparseValues 1-D.  The `values` of the `SparseTensor`.
+   * @param sparseShape 1-D.  The `shape` of the `SparseTensor`.
+   * @param outType The `dtype` to use for serialization; the supported types are `string`
+   * (default) and `variant`.
+   * @return a new instance of SerializeSparse
+   */
+  @Endpoint(describeByClass = true)
+  public static <U extends TType, T extends TType> SerializeSparse<U> create(Scope scope, Operand<TInt64> sparseIndices, Operand<T> sparseValues, Operand<TInt64> sparseShape, Class<U> outType) {
+    OperationBuilder opBuilder = scope.env().opBuilder("SerializeSparse", scope.makeOpName("SerializeSparse"));
+    opBuilder.addInput(sparseIndices.asOutput());
+    opBuilder.addInput(sparseValues.asOutput());
+    opBuilder.addInput(sparseShape.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("out_type", outType);
+    return new SerializeSparse<U>(opBuilder.build());
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new SerializeSparse operation using default output types.
+   * 
+   * @param scope current scope
+   * @param sparseIndices 2-D.  The `indices` of the `SparseTensor`.
+   * @param sparseValues 1-D.  The `values` of the `SparseTensor`.
+   * @param sparseShape 1-D.  The `shape` of the `SparseTensor`.
+   * @return a new instance of SerializeSparse
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> SerializeSparse<TString> create(Scope scope, Operand<TInt64> sparseIndices, Operand<T> sparseValues, Operand<TInt64> sparseShape) {
+    return create(scope, sparseIndices, sparseValues, sparseShape, TString.class);
+  }
+  
+  /**
+   */
+  public Output<U> serializedSparse() {
+    return serializedSparse;
+  }
+  
+  @Override
+  public Output<U> asOutput() {
+    return serializedSparse;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "SerializeSparse";
+  
+  private Output<U> serializedSparse;
+  
+  private SerializeSparse(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    serializedSparse = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/SerializeTensor.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/SerializeTensor.java
new file mode 100644
index 00000000000..d68320b85e3
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/SerializeTensor.java
@@ -0,0 +1,74 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.io;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TString;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Transforms a Tensor into a serialized TensorProto proto.
+ */
+@Operator(group = "io")
+public final class SerializeTensor extends RawOp implements Operand<TString> {
+  
+  /**
+   * Factory method to create a class wrapping a new SerializeTensor operation.
+   * 
+   * @param scope current scope
+   * @param tensor A Tensor of type `T`.
+   * @return a new instance of SerializeTensor
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> SerializeTensor create(Scope scope, Operand<T> tensor) {
+    OperationBuilder opBuilder = scope.env().opBuilder("SerializeTensor", scope.makeOpName("SerializeTensor"));
+    opBuilder.addInput(tensor.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new SerializeTensor(opBuilder.build());
+  }
+  
+  /**
+   * A serialized TensorProto proto of the input tensor.
+   */
+  public Output<TString> serialized() {
+    return serialized;
+  }
+  
+  @Override
+  public Output<TString> asOutput() {
+    return serialized;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "SerializeTensor";
+  
+  private Output<TString> serialized;
+  
+  private SerializeTensor(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    serialized = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ShardedFilename.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ShardedFilename.java
new file mode 100644
index 00000000000..88b422a5f6d
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ShardedFilename.java
@@ -0,0 +1,79 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.io;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.TString;
+
+/**
+ * Generate a sharded filename. The filename is printf formatted as
+ * <p>
+ *    %s-%05d-of-%05d, basename, shard, num_shards.
+ */
+@Operator(group = "io")
+public final class ShardedFilename extends RawOp implements Operand<TString> {
+  
+  /**
+   * Factory method to create a class wrapping a new ShardedFilename operation.
+   * 
+   * @param scope current scope
+   * @param basename 
+   * @param shard 
+   * @param numShards 
+   * @return a new instance of ShardedFilename
+   */
+  @Endpoint(describeByClass = true)
+  public static ShardedFilename create(Scope scope, Operand<TString> basename, Operand<TInt32> shard, Operand<TInt32> numShards) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ShardedFilename", scope.makeOpName("ShardedFilename"));
+    opBuilder.addInput(basename.asOutput());
+    opBuilder.addInput(shard.asOutput());
+    opBuilder.addInput(numShards.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new ShardedFilename(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<TString> filename() {
+    return filename;
+  }
+  
+  @Override
+  public Output<TString> asOutput() {
+    return filename;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ShardedFilename";
+  
+  private Output<TString> filename;
+  
+  private ShardedFilename(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    filename = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ShardedFilespec.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ShardedFilespec.java
new file mode 100644
index 00000000000..f04c0cf1a58
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/ShardedFilespec.java
@@ -0,0 +1,75 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.io;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.TString;
+
+/**
+ * Generate a glob pattern matching all sharded file names.
+ */
+@Operator(group = "io")
+public final class ShardedFilespec extends RawOp implements Operand<TString> {
+  
+  /**
+   * Factory method to create a class wrapping a new ShardedFilespec operation.
+   * 
+   * @param scope current scope
+   * @param basename 
+   * @param numShards 
+   * @return a new instance of ShardedFilespec
+   */
+  @Endpoint(describeByClass = true)
+  public static ShardedFilespec create(Scope scope, Operand<TString> basename, Operand<TInt32> numShards) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ShardedFilespec", scope.makeOpName("ShardedFilespec"));
+    opBuilder.addInput(basename.asOutput());
+    opBuilder.addInput(numShards.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new ShardedFilespec(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<TString> filename() {
+    return filename;
+  }
+  
+  @Override
+  public Output<TString> asOutput() {
+    return filename;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ShardedFilespec";
+  
+  private Output<TString> filename;
+  
+  private ShardedFilespec(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    filename = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/TextLineReader.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/TextLineReader.java
new file mode 100644
index 00000000000..90bf079d2cf
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/TextLineReader.java
@@ -0,0 +1,148 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.io;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * A Reader that outputs the lines of a file delimited by '\n'.
+ */
+@Operator(group = "io")
+public final class TextLineReader extends RawOp implements Operand<TType> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.io.TextLineReader}
+   */
+  public static class Options {
+    
+    /**
+     * @param skipHeaderLines Number of lines to skip from the beginning of every file.
+     */
+    public Options skipHeaderLines(Long skipHeaderLines) {
+      this.skipHeaderLines = skipHeaderLines;
+      return this;
+    }
+    
+    /**
+     * @param container If non-empty, this reader is placed in the given container.
+     * Otherwise, a default container is used.
+     */
+    public Options container(String container) {
+      this.container = container;
+      return this;
+    }
+    
+    /**
+     * @param sharedName If non-empty, this reader is named in the given bucket
+     * with this shared_name. Otherwise, the node name is used instead.
+     */
+    public Options sharedName(String sharedName) {
+      this.sharedName = sharedName;
+      return this;
+    }
+    
+    private Long skipHeaderLines;
+    private String container;
+    private String sharedName;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new TextLineReader operation.
+   * 
+   * @param scope current scope
+   * @param options carries optional attributes values
+   * @return a new instance of TextLineReader
+   */
+  @Endpoint(describeByClass = true)
+  public static TextLineReader create(Scope scope, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("TextLineReaderV2", scope.makeOpName("TextLineReader"));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.skipHeaderLines != null) {
+          opBuilder.setAttr("skip_header_lines", opts.skipHeaderLines);
+        }
+        if (opts.container != null) {
+          opBuilder.setAttr("container", opts.container);
+        }
+        if (opts.sharedName != null) {
+          opBuilder.setAttr("shared_name", opts.sharedName);
+        }
+      }
+    }
+    return new TextLineReader(opBuilder.build());
+  }
+  
+  /**
+   * @param skipHeaderLines Number of lines to skip from the beginning of every file.
+   */
+  public static Options skipHeaderLines(Long skipHeaderLines) {
+    return new Options().skipHeaderLines(skipHeaderLines);
+  }
+  
+  /**
+   * @param container If non-empty, this reader is placed in the given container.
+   * Otherwise, a default container is used.
+   */
+  public static Options container(String container) {
+    return new Options().container(container);
+  }
+  
+  /**
+   * @param sharedName If non-empty, this reader is named in the given bucket
+   * with this shared_name. Otherwise, the node name is used instead.
+   */
+  public static Options sharedName(String sharedName) {
+    return new Options().sharedName(sharedName);
+  }
+  
+  /**
+   * The handle to reference the Reader.
+   */
+  public Output<?> readerHandle() {
+    return readerHandle;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) readerHandle;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "TextLineReaderV2";
+  
+  private Output<?> readerHandle;
+  
+  private TextLineReader(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    readerHandle = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/TfRecordReader.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/TfRecordReader.java
new file mode 100644
index 00000000000..a12145e7972
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/TfRecordReader.java
@@ -0,0 +1,148 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.io;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * A Reader that outputs the records from a TensorFlow Records file.
+ */
+@Operator(group = "io")
+public final class TfRecordReader extends RawOp implements Operand<TType> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.io.TfRecordReader}
+   */
+  public static class Options {
+    
+    /**
+     * @param container If non-empty, this reader is placed in the given container.
+     * Otherwise, a default container is used.
+     */
+    public Options container(String container) {
+      this.container = container;
+      return this;
+    }
+    
+    /**
+     * @param sharedName If non-empty, this reader is named in the given bucket
+     * with this shared_name. Otherwise, the node name is used instead.
+     */
+    public Options sharedName(String sharedName) {
+      this.sharedName = sharedName;
+      return this;
+    }
+    
+    /**
+     * @param compressionType 
+     */
+    public Options compressionType(String compressionType) {
+      this.compressionType = compressionType;
+      return this;
+    }
+    
+    private String container;
+    private String sharedName;
+    private String compressionType;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new TfRecordReader operation.
+   * 
+   * @param scope current scope
+   * @param options carries optional attributes values
+   * @return a new instance of TfRecordReader
+   */
+  @Endpoint(describeByClass = true)
+  public static TfRecordReader create(Scope scope, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("TFRecordReaderV2", scope.makeOpName("TfRecordReader"));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.container != null) {
+          opBuilder.setAttr("container", opts.container);
+        }
+        if (opts.sharedName != null) {
+          opBuilder.setAttr("shared_name", opts.sharedName);
+        }
+        if (opts.compressionType != null) {
+          opBuilder.setAttr("compression_type", opts.compressionType);
+        }
+      }
+    }
+    return new TfRecordReader(opBuilder.build());
+  }
+  
+  /**
+   * @param container If non-empty, this reader is placed in the given container.
+   * Otherwise, a default container is used.
+   */
+  public static Options container(String container) {
+    return new Options().container(container);
+  }
+  
+  /**
+   * @param sharedName If non-empty, this reader is named in the given bucket
+   * with this shared_name. Otherwise, the node name is used instead.
+   */
+  public static Options sharedName(String sharedName) {
+    return new Options().sharedName(sharedName);
+  }
+  
+  /**
+   * @param compressionType 
+   */
+  public static Options compressionType(String compressionType) {
+    return new Options().compressionType(compressionType);
+  }
+  
+  /**
+   * The handle to reference the Reader.
+   */
+  public Output<?> readerHandle() {
+    return readerHandle;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) readerHandle;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "TFRecordReaderV2";
+  
+  private Output<?> readerHandle;
+  
+  private TfRecordReader(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    readerHandle = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/WholeFileReader.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/WholeFileReader.java
new file mode 100644
index 00000000000..8cab4f9b09e
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/WholeFileReader.java
@@ -0,0 +1,132 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.io;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * A Reader that outputs the entire contents of a file as a value.
+ * <p>
+ * To use, enqueue filenames in a Queue.  The output of ReaderRead will
+ * be a filename (key) and the contents of that file (value).
+ */
+@Operator(group = "io")
+public final class WholeFileReader extends RawOp implements Operand<TType> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.io.WholeFileReader}
+   */
+  public static class Options {
+    
+    /**
+     * @param container If non-empty, this reader is placed in the given container.
+     * Otherwise, a default container is used.
+     */
+    public Options container(String container) {
+      this.container = container;
+      return this;
+    }
+    
+    /**
+     * @param sharedName If non-empty, this reader is named in the given bucket
+     * with this shared_name. Otherwise, the node name is used instead.
+     */
+    public Options sharedName(String sharedName) {
+      this.sharedName = sharedName;
+      return this;
+    }
+    
+    private String container;
+    private String sharedName;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new WholeFileReader operation.
+   * 
+   * @param scope current scope
+   * @param options carries optional attributes values
+   * @return a new instance of WholeFileReader
+   */
+  @Endpoint(describeByClass = true)
+  public static WholeFileReader create(Scope scope, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("WholeFileReaderV2", scope.makeOpName("WholeFileReader"));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.container != null) {
+          opBuilder.setAttr("container", opts.container);
+        }
+        if (opts.sharedName != null) {
+          opBuilder.setAttr("shared_name", opts.sharedName);
+        }
+      }
+    }
+    return new WholeFileReader(opBuilder.build());
+  }
+  
+  /**
+   * @param container If non-empty, this reader is placed in the given container.
+   * Otherwise, a default container is used.
+   */
+  public static Options container(String container) {
+    return new Options().container(container);
+  }
+  
+  /**
+   * @param sharedName If non-empty, this reader is named in the given bucket
+   * with this shared_name. Otherwise, the node name is used instead.
+   */
+  public static Options sharedName(String sharedName) {
+    return new Options().sharedName(sharedName);
+  }
+  
+  /**
+   * The handle to reference the Reader.
+   */
+  public Output<?> readerHandle() {
+    return readerHandle;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) readerHandle;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "WholeFileReaderV2";
+  
+  private Output<?> readerHandle;
+  
+  private WholeFileReader(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    readerHandle = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/WriteFile.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/WriteFile.java
new file mode 100644
index 00000000000..d1c9dd9b9c8
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/io/WriteFile.java
@@ -0,0 +1,60 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.io;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TString;
+
+/**
+ * Writes contents to the file at input filename. Creates file and recursively
+ * <p>
+ * creates directory if not existing.
+ */
+@Operator(group = "io")
+public final class WriteFile extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new WriteFile operation.
+   * 
+   * @param scope current scope
+   * @param filename scalar. The name of the file to which we write the contents.
+   * @param contents scalar. The content to be written to the output file.
+   * @return a new instance of WriteFile
+   */
+  @Endpoint(describeByClass = true)
+  public static WriteFile create(Scope scope, Operand<TString> filename, Operand<TString> contents) {
+    OperationBuilder opBuilder = scope.env().opBuilder("WriteFile", scope.makeOpName("WriteFile"));
+    opBuilder.addInput(filename.asOutput());
+    opBuilder.addInput(contents.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new WriteFile(opBuilder.build());
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "WriteFile";
+  
+  private WriteFile(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BandPart.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BandPart.java
new file mode 100644
index 00000000000..11dbb5662a9
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BandPart.java
@@ -0,0 +1,118 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.linalg;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Copy a tensor setting everything outside a central band in each innermost matrix to zero.
+ * <p>
+ * The `band` part is computed as follows:
+ * Assume `input` has `k` dimensions `[I, J, K, ..., M, N]`, then the output is a
+ * tensor with the same shape where
+ * <p>
+ * `band[i, j, k, ..., m, n] = in_band(m, n) * input[i, j, k, ..., m, n]`.
+ * <p>
+ * The indicator function
+ * <p>
+ * `in_band(m, n) = (num_lower < 0 || (m-n) <= num_lower)) &&
+ *                  (num_upper < 0 || (n-m) <= num_upper)`.
+ * <p>
+ * For example:
+ * <pre>{@code
+ * # if 'input' is [[ 0,  1,  2, 3]
+ *                  [-1,  0,  1, 2]
+ *                  [-2, -1,  0, 1]
+ *                  [-3, -2, -1, 0]],
+ * 
+ * tf.matrix_band_part(input, 1, -1) ==> [[ 0,  1,  2, 3]
+ *                                        [-1,  0,  1, 2]
+ *                                        [ 0, -1,  0, 1]
+ *                                        [ 0,  0, -1, 0]],
+ * 
+ * tf.matrix_band_part(input, 2, 1) ==> [[ 0,  1,  0, 0]
+ *                                       [-1,  0,  1, 0]
+ *                                       [-2, -1,  0, 1]
+ *                                       [ 0, -2, -1, 0]]
+ * }</pre>
+ * Useful special cases:
+ * <pre>{@code
+ *  tf.matrix_band_part(input, 0, -1) ==> Upper triangular part.
+ *  tf.matrix_band_part(input, -1, 0) ==> Lower triangular part.
+ *  tf.matrix_band_part(input, 0, 0) ==> Diagonal.
+ * }</pre>
+ * 
+ * 
+ * @param <T> data type for {@code band()} output
+ */
+@Operator(group = "linalg")
+public final class BandPart<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new BandPart operation.
+   * 
+   * @param scope current scope
+   * @param input Rank `k` tensor.
+   * @param numLower 0-D tensor. Number of subdiagonals to keep. If negative, keep entire
+   * lower triangle.
+   * @param numUpper 0-D tensor. Number of superdiagonals to keep. If negative, keep
+   * entire upper triangle.
+   * @return a new instance of BandPart
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType, U extends TNumber> BandPart<T> create(Scope scope, Operand<T> input, Operand<U> numLower, Operand<U> numUpper) {
+    OperationBuilder opBuilder = scope.env().opBuilder("MatrixBandPart", scope.makeOpName("BandPart"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(numLower.asOutput());
+    opBuilder.addInput(numUpper.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new BandPart<T>(opBuilder.build());
+  }
+  
+  /**
+   * Rank `k` tensor of the same shape as input. The extracted banded tensor.
+   */
+  public Output<T> band() {
+    return band;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return band;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "MatrixBandPart";
+  
+  private Output<T> band;
+  
+  private BandPart(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    band = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BandedTriangularSolve.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BandedTriangularSolve.java
new file mode 100644
index 00000000000..8c1e184e6c9
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BandedTriangularSolve.java
@@ -0,0 +1,126 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.linalg;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * @param <T> data type for {@code output()} output
+ */
+public final class BandedTriangularSolve<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.linalg.BandedTriangularSolve}
+   */
+  public static class Options {
+    
+    /**
+     * @param lower 
+     */
+    public Options lower(Boolean lower) {
+      this.lower = lower;
+      return this;
+    }
+    
+    /**
+     * @param adjoint 
+     */
+    public Options adjoint(Boolean adjoint) {
+      this.adjoint = adjoint;
+      return this;
+    }
+    
+    private Boolean lower;
+    private Boolean adjoint;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new BandedTriangularSolve operation.
+   * 
+   * @param scope current scope
+   * @param matrix 
+   * @param rhs 
+   * @param options carries optional attributes values
+   * @return a new instance of BandedTriangularSolve
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> BandedTriangularSolve<T> create(Scope scope, Operand<T> matrix, Operand<T> rhs, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("BandedTriangularSolve", scope.makeOpName("BandedTriangularSolve"));
+    opBuilder.addInput(matrix.asOutput());
+    opBuilder.addInput(rhs.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.lower != null) {
+          opBuilder.setAttr("lower", opts.lower);
+        }
+        if (opts.adjoint != null) {
+          opBuilder.setAttr("adjoint", opts.adjoint);
+        }
+      }
+    }
+    return new BandedTriangularSolve<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param lower 
+   */
+  public static Options lower(Boolean lower) {
+    return new Options().lower(lower);
+  }
+  
+  /**
+   * @param adjoint 
+   */
+  public static Options adjoint(Boolean adjoint) {
+    return new Options().adjoint(adjoint);
+  }
+  
+  /**
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "BandedTriangularSolve";
+  
+  private Output<T> output;
+  
+  private BandedTriangularSolve(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchCholesky.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchCholesky.java
new file mode 100644
index 00000000000..40a66933f4f
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchCholesky.java
@@ -0,0 +1,72 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.linalg;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * @param <T> data type for {@code output()} output
+ */
+@Operator(group = "linalg")
+public final class BatchCholesky<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new BatchCholesky operation.
+   * 
+   * @param scope current scope
+   * @param input 
+   * @return a new instance of BatchCholesky
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> BatchCholesky<T> create(Scope scope, Operand<T> input) {
+    OperationBuilder opBuilder = scope.env().opBuilder("BatchCholesky", scope.makeOpName("BatchCholesky"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new BatchCholesky<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "BatchCholesky";
+  
+  private Output<T> output;
+  
+  private BatchCholesky(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchCholeskyGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchCholeskyGrad.java
new file mode 100644
index 00000000000..caa625ffe79
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchCholeskyGrad.java
@@ -0,0 +1,74 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.linalg;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * @param <T> data type for {@code output()} output
+ */
+@Operator(group = "linalg")
+public final class BatchCholeskyGrad<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new BatchCholeskyGrad operation.
+   * 
+   * @param scope current scope
+   * @param l 
+   * @param grad 
+   * @return a new instance of BatchCholeskyGrad
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> BatchCholeskyGrad<T> create(Scope scope, Operand<T> l, Operand<T> grad) {
+    OperationBuilder opBuilder = scope.env().opBuilder("BatchCholeskyGrad", scope.makeOpName("BatchCholeskyGrad"));
+    opBuilder.addInput(l.asOutput());
+    opBuilder.addInput(grad.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new BatchCholeskyGrad<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "BatchCholeskyGrad";
+  
+  private Output<T> output;
+  
+  private BatchCholeskyGrad(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixBandPart.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixBandPart.java
new file mode 100644
index 00000000000..8c990a8a71c
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixBandPart.java
@@ -0,0 +1,77 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.linalg;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
+
+/**
+ * @param <T> data type for {@code band()} output
+ */
+@Operator(group = "linalg")
+public final class BatchMatrixBandPart<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new BatchMatrixBandPart operation.
+   * 
+   * @param scope current scope
+   * @param input 
+   * @param numLower 
+   * @param numUpper 
+   * @return a new instance of BatchMatrixBandPart
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> BatchMatrixBandPart<T> create(Scope scope, Operand<T> input, Operand<TInt64> numLower, Operand<TInt64> numUpper) {
+    OperationBuilder opBuilder = scope.env().opBuilder("BatchMatrixBandPart", scope.makeOpName("BatchMatrixBandPart"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(numLower.asOutput());
+    opBuilder.addInput(numUpper.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new BatchMatrixBandPart<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> band() {
+    return band;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return band;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "BatchMatrixBandPart";
+  
+  private Output<T> band;
+  
+  private BatchMatrixBandPart(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    band = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixDeterminant.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixDeterminant.java
new file mode 100644
index 00000000000..f513e9f0057
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixDeterminant.java
@@ -0,0 +1,72 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.linalg;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * @param <T> data type for {@code output()} output
+ */
+@Operator(group = "linalg")
+public final class BatchMatrixDeterminant<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new BatchMatrixDeterminant operation.
+   * 
+   * @param scope current scope
+   * @param input 
+   * @return a new instance of BatchMatrixDeterminant
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> BatchMatrixDeterminant<T> create(Scope scope, Operand<T> input) {
+    OperationBuilder opBuilder = scope.env().opBuilder("BatchMatrixDeterminant", scope.makeOpName("BatchMatrixDeterminant"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new BatchMatrixDeterminant<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "BatchMatrixDeterminant";
+  
+  private Output<T> output;
+  
+  private BatchMatrixDeterminant(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixDiag.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixDiag.java
new file mode 100644
index 00000000000..b3f9071f9f6
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixDiag.java
@@ -0,0 +1,72 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.linalg;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * @param <T> data type for {@code output()} output
+ */
+@Operator(group = "linalg")
+public final class BatchMatrixDiag<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new BatchMatrixDiag operation.
+   * 
+   * @param scope current scope
+   * @param diagonal 
+   * @return a new instance of BatchMatrixDiag
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> BatchMatrixDiag<T> create(Scope scope, Operand<T> diagonal) {
+    OperationBuilder opBuilder = scope.env().opBuilder("BatchMatrixDiag", scope.makeOpName("BatchMatrixDiag"));
+    opBuilder.addInput(diagonal.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new BatchMatrixDiag<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "BatchMatrixDiag";
+  
+  private Output<T> output;
+  
+  private BatchMatrixDiag(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixDiagPart.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixDiagPart.java
new file mode 100644
index 00000000000..47f78ed623d
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixDiagPart.java
@@ -0,0 +1,72 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.linalg;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * @param <T> data type for {@code diagonal()} output
+ */
+@Operator(group = "linalg")
+public final class BatchMatrixDiagPart<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new BatchMatrixDiagPart operation.
+   * 
+   * @param scope current scope
+   * @param input 
+   * @return a new instance of BatchMatrixDiagPart
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> BatchMatrixDiagPart<T> create(Scope scope, Operand<T> input) {
+    OperationBuilder opBuilder = scope.env().opBuilder("BatchMatrixDiagPart", scope.makeOpName("BatchMatrixDiagPart"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new BatchMatrixDiagPart<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> diagonal() {
+    return diagonal;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return diagonal;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "BatchMatrixDiagPart";
+  
+  private Output<T> diagonal;
+  
+  private BatchMatrixDiagPart(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    diagonal = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixInverse.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixInverse.java
new file mode 100644
index 00000000000..69017d8dcfc
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixInverse.java
@@ -0,0 +1,106 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.linalg;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * @param <T> data type for {@code output()} output
+ */
+@Operator(group = "linalg")
+public final class BatchMatrixInverse<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.linalg.BatchMatrixInverse}
+   */
+  public static class Options {
+    
+    /**
+     * @param adjoint 
+     */
+    public Options adjoint(Boolean adjoint) {
+      this.adjoint = adjoint;
+      return this;
+    }
+    
+    private Boolean adjoint;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new BatchMatrixInverse operation.
+   * 
+   * @param scope current scope
+   * @param input 
+   * @param options carries optional attributes values
+   * @return a new instance of BatchMatrixInverse
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> BatchMatrixInverse<T> create(Scope scope, Operand<T> input, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("BatchMatrixInverse", scope.makeOpName("BatchMatrixInverse"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.adjoint != null) {
+          opBuilder.setAttr("adjoint", opts.adjoint);
+        }
+      }
+    }
+    return new BatchMatrixInverse<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param adjoint 
+   */
+  public static Options adjoint(Boolean adjoint) {
+    return new Options().adjoint(adjoint);
+  }
+  
+  /**
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "BatchMatrixInverse";
+  
+  private Output<T> output;
+  
+  private BatchMatrixInverse(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixSetDiag.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixSetDiag.java
new file mode 100644
index 00000000000..d83ee0194e7
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixSetDiag.java
@@ -0,0 +1,74 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.linalg;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * @param <T> data type for {@code output()} output
+ */
+@Operator(group = "linalg")
+public final class BatchMatrixSetDiag<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new BatchMatrixSetDiag operation.
+   * 
+   * @param scope current scope
+   * @param input 
+   * @param diagonal 
+   * @return a new instance of BatchMatrixSetDiag
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> BatchMatrixSetDiag<T> create(Scope scope, Operand<T> input, Operand<T> diagonal) {
+    OperationBuilder opBuilder = scope.env().opBuilder("BatchMatrixSetDiag", scope.makeOpName("BatchMatrixSetDiag"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(diagonal.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new BatchMatrixSetDiag<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "BatchMatrixSetDiag";
+  
+  private Output<T> output;
+  
+  private BatchMatrixSetDiag(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixSolve.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixSolve.java
new file mode 100644
index 00000000000..2748e30a02b
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixSolve.java
@@ -0,0 +1,108 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.linalg;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * @param <T> data type for {@code output()} output
+ */
+@Operator(group = "linalg")
+public final class BatchMatrixSolve<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.linalg.BatchMatrixSolve}
+   */
+  public static class Options {
+    
+    /**
+     * @param adjoint 
+     */
+    public Options adjoint(Boolean adjoint) {
+      this.adjoint = adjoint;
+      return this;
+    }
+    
+    private Boolean adjoint;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new BatchMatrixSolve operation.
+   * 
+   * @param scope current scope
+   * @param matrix 
+   * @param rhs 
+   * @param options carries optional attributes values
+   * @return a new instance of BatchMatrixSolve
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> BatchMatrixSolve<T> create(Scope scope, Operand<T> matrix, Operand<T> rhs, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("BatchMatrixSolve", scope.makeOpName("BatchMatrixSolve"));
+    opBuilder.addInput(matrix.asOutput());
+    opBuilder.addInput(rhs.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.adjoint != null) {
+          opBuilder.setAttr("adjoint", opts.adjoint);
+        }
+      }
+    }
+    return new BatchMatrixSolve<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param adjoint 
+   */
+  public static Options adjoint(Boolean adjoint) {
+    return new Options().adjoint(adjoint);
+  }
+  
+  /**
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "BatchMatrixSolve";
+  
+  private Output<T> output;
+  
+  private BatchMatrixSolve(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixSolveLs.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixSolveLs.java
new file mode 100644
index 00000000000..ca30c69f971
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixSolveLs.java
@@ -0,0 +1,111 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.linalg;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat64;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * @param <T> data type for {@code output()} output
+ */
+@Operator(group = "linalg")
+public final class BatchMatrixSolveLs<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.linalg.BatchMatrixSolveLs}
+   */
+  public static class Options {
+    
+    /**
+     * @param fast 
+     */
+    public Options fast(Boolean fast) {
+      this.fast = fast;
+      return this;
+    }
+    
+    private Boolean fast;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new BatchMatrixSolveLs operation.
+   * 
+   * @param scope current scope
+   * @param matrix 
+   * @param rhs 
+   * @param l2Regularizer 
+   * @param options carries optional attributes values
+   * @return a new instance of BatchMatrixSolveLs
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> BatchMatrixSolveLs<T> create(Scope scope, Operand<T> matrix, Operand<T> rhs, Operand<TFloat64> l2Regularizer, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("BatchMatrixSolveLs", scope.makeOpName("BatchMatrixSolveLs"));
+    opBuilder.addInput(matrix.asOutput());
+    opBuilder.addInput(rhs.asOutput());
+    opBuilder.addInput(l2Regularizer.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.fast != null) {
+          opBuilder.setAttr("fast", opts.fast);
+        }
+      }
+    }
+    return new BatchMatrixSolveLs<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param fast 
+   */
+  public static Options fast(Boolean fast) {
+    return new Options().fast(fast);
+  }
+  
+  /**
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "BatchMatrixSolveLs";
+  
+  private Output<T> output;
+  
+  private BatchMatrixSolveLs(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixTriangularSolve.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixTriangularSolve.java
new file mode 100644
index 00000000000..c747840b107
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchMatrixTriangularSolve.java
@@ -0,0 +1,127 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.linalg;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * @param <T> data type for {@code output()} output
+ */
+@Operator(group = "linalg")
+public final class BatchMatrixTriangularSolve<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.linalg.BatchMatrixTriangularSolve}
+   */
+  public static class Options {
+    
+    /**
+     * @param lower 
+     */
+    public Options lower(Boolean lower) {
+      this.lower = lower;
+      return this;
+    }
+    
+    /**
+     * @param adjoint 
+     */
+    public Options adjoint(Boolean adjoint) {
+      this.adjoint = adjoint;
+      return this;
+    }
+    
+    private Boolean lower;
+    private Boolean adjoint;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new BatchMatrixTriangularSolve operation.
+   * 
+   * @param scope current scope
+   * @param matrix 
+   * @param rhs 
+   * @param options carries optional attributes values
+   * @return a new instance of BatchMatrixTriangularSolve
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> BatchMatrixTriangularSolve<T> create(Scope scope, Operand<T> matrix, Operand<T> rhs, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("BatchMatrixTriangularSolve", scope.makeOpName("BatchMatrixTriangularSolve"));
+    opBuilder.addInput(matrix.asOutput());
+    opBuilder.addInput(rhs.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.lower != null) {
+          opBuilder.setAttr("lower", opts.lower);
+        }
+        if (opts.adjoint != null) {
+          opBuilder.setAttr("adjoint", opts.adjoint);
+        }
+      }
+    }
+    return new BatchMatrixTriangularSolve<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param lower 
+   */
+  public static Options lower(Boolean lower) {
+    return new Options().lower(lower);
+  }
+  
+  /**
+   * @param adjoint 
+   */
+  public static Options adjoint(Boolean adjoint) {
+    return new Options().adjoint(adjoint);
+  }
+  
+  /**
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "BatchMatrixTriangularSolve";
+  
+  private Output<T> output;
+  
+  private BatchMatrixTriangularSolve(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchSelfAdjointEig.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchSelfAdjointEig.java
new file mode 100644
index 00000000000..e85fea495fd
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchSelfAdjointEig.java
@@ -0,0 +1,109 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.linalg;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * @param <T> data type for {@code e()} output
+ */
+@Operator(group = "linalg")
+public final class BatchSelfAdjointEig<T extends TNumber> extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.linalg.BatchSelfAdjointEig}
+   */
+  public static class Options {
+    
+    /**
+     * @param computeV 
+     */
+    public Options computeV(Boolean computeV) {
+      this.computeV = computeV;
+      return this;
+    }
+    
+    private Boolean computeV;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new BatchSelfAdjointEig operation.
+   * 
+   * @param scope current scope
+   * @param input 
+   * @param options carries optional attributes values
+   * @return a new instance of BatchSelfAdjointEig
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> BatchSelfAdjointEig<T> create(Scope scope, Operand<T> input, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("BatchSelfAdjointEigV2", scope.makeOpName("BatchSelfAdjointEig"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.computeV != null) {
+          opBuilder.setAttr("compute_v", opts.computeV);
+        }
+      }
+    }
+    return new BatchSelfAdjointEig<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param computeV 
+   */
+  public static Options computeV(Boolean computeV) {
+    return new Options().computeV(computeV);
+  }
+  
+  /**
+   */
+  public Output<T> e() {
+    return e;
+  }
+  
+  /**
+   */
+  public Output<T> v() {
+    return v;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "BatchSelfAdjointEigV2";
+  
+  private Output<T> e;
+  private Output<T> v;
+  
+  private BatchSelfAdjointEig(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    e = operation.output(outputIdx++);
+    v = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchSvd.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchSvd.java
new file mode 100644
index 00000000000..21db016713a
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/BatchSvd.java
@@ -0,0 +1,136 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.linalg;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * @param <T> data type for {@code s()} output
+ */
+@Operator(group = "linalg")
+public final class BatchSvd<T extends TType> extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.linalg.BatchSvd}
+   */
+  public static class Options {
+    
+    /**
+     * @param computeUv 
+     */
+    public Options computeUv(Boolean computeUv) {
+      this.computeUv = computeUv;
+      return this;
+    }
+    
+    /**
+     * @param fullMatrices 
+     */
+    public Options fullMatrices(Boolean fullMatrices) {
+      this.fullMatrices = fullMatrices;
+      return this;
+    }
+    
+    private Boolean computeUv;
+    private Boolean fullMatrices;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new BatchSvd operation.
+   * 
+   * @param scope current scope
+   * @param input 
+   * @param options carries optional attributes values
+   * @return a new instance of BatchSvd
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> BatchSvd<T> create(Scope scope, Operand<T> input, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("BatchSvd", scope.makeOpName("BatchSvd"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.computeUv != null) {
+          opBuilder.setAttr("compute_uv", opts.computeUv);
+        }
+        if (opts.fullMatrices != null) {
+          opBuilder.setAttr("full_matrices", opts.fullMatrices);
+        }
+      }
+    }
+    return new BatchSvd<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param computeUv 
+   */
+  public static Options computeUv(Boolean computeUv) {
+    return new Options().computeUv(computeUv);
+  }
+  
+  /**
+   * @param fullMatrices 
+   */
+  public static Options fullMatrices(Boolean fullMatrices) {
+    return new Options().fullMatrices(fullMatrices);
+  }
+  
+  /**
+   */
+  public Output<T> s() {
+    return s;
+  }
+  
+  /**
+   */
+  public Output<T> u() {
+    return u;
+  }
+  
+  /**
+   */
+  public Output<T> v() {
+    return v;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "BatchSvd";
+  
+  private Output<T> s;
+  private Output<T> u;
+  private Output<T> v;
+  
+  private BatchSvd(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    s = operation.output(outputIdx++);
+    u = operation.output(outputIdx++);
+    v = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Cholesky.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Cholesky.java
new file mode 100644
index 00000000000..3acf951d5a7
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Cholesky.java
@@ -0,0 +1,89 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.linalg;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Computes the Cholesky decomposition of one or more square matrices.
+ * <p>
+ * The input is a tensor of shape `[..., M, M]` whose inner-most 2 dimensions
+ * form square matrices.
+ * <p>
+ * The input has to be symmetric and positive definite. Only the lower-triangular
+ * part of the input will be used for this operation. The upper-triangular part
+ * will not be read.
+ * <p>
+ * The output is a tensor of the same shape as the input
+ * containing the Cholesky decompositions for all input submatrices `[..., :, :]`.
+ * <p>
+ * <b>Note</b>: The gradient computation on GPU is faster for large matrices but
+ * not for large batch dimensions when the submatrices are small. In this
+ * case it might be faster to use the CPU.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator(group = "linalg")
+public final class Cholesky<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new Cholesky operation.
+   * 
+   * @param scope current scope
+   * @param input Shape is `[..., M, M]`.
+   * @return a new instance of Cholesky
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> Cholesky<T> create(Scope scope, Operand<T> input) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Cholesky", scope.makeOpName("Cholesky"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new Cholesky<T>(opBuilder.build());
+  }
+  
+  /**
+   * Shape is `[..., M, M]`.
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Cholesky";
+  
+  private Output<T> output;
+  
+  private Cholesky(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/CholeskyGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/CholeskyGrad.java
new file mode 100644
index 00000000000..e5ac38214dc
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/CholeskyGrad.java
@@ -0,0 +1,84 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.linalg;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Computes the reverse mode backpropagated gradient of the Cholesky algorithm.
+ * <p>
+ * For an explanation see "Differentiation of the Cholesky algorithm" by
+ * Iain Murray http://arxiv.org/abs/1602.07527.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator(group = "linalg")
+public final class CholeskyGrad<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new CholeskyGrad operation.
+   * 
+   * @param scope current scope
+   * @param l Output of batch Cholesky algorithm l = cholesky(A). Shape is `[..., M, M]`.
+   * Algorithm depends only on lower triangular part of the innermost matrices of
+   * this tensor.
+   * @param grad df/dl where f is some scalar function. Shape is `[..., M, M]`.
+   * Algorithm depends only on lower triangular part of the innermost matrices of
+   * this tensor.
+   * @return a new instance of CholeskyGrad
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> CholeskyGrad<T> create(Scope scope, Operand<T> l, Operand<T> grad) {
+    OperationBuilder opBuilder = scope.env().opBuilder("CholeskyGrad", scope.makeOpName("CholeskyGrad"));
+    opBuilder.addInput(l.asOutput());
+    opBuilder.addInput(grad.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new CholeskyGrad<T>(opBuilder.build());
+  }
+  
+  /**
+   * Symmetrized version of df/dA . Shape is `[..., M, M]`
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "CholeskyGrad";
+  
+  private Output<T> output;
+  
+  private CholeskyGrad(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/ConjugateTranspose.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/ConjugateTranspose.java
new file mode 100644
index 00000000000..2da45552f35
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/ConjugateTranspose.java
@@ -0,0 +1,81 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.linalg;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Shuffle dimensions of x according to a permutation and conjugate the result.
+ * <p>
+ * The output `y` has the same rank as `x`. The shapes of `x` and `y` satisfy:
+ *   `y.shape[i] == x.shape[perm[i]] for i in [0, 1, ..., rank(x) - 1]`
+ *   `y[i,j,k,...,s,t,u] == conj(x[perm[i], perm[j], perm[k],...,perm[s], perm[t], perm[u]])`
+ * 
+ * @param <T> data type for {@code y()} output
+ */
+@Operator(group = "linalg")
+public final class ConjugateTranspose<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new ConjugateTranspose operation.
+   * 
+   * @param scope current scope
+   * @param x 
+   * @param perm 
+   * @return a new instance of ConjugateTranspose
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType, U extends TNumber> ConjugateTranspose<T> create(Scope scope, Operand<T> x, Operand<U> perm) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ConjugateTranspose", scope.makeOpName("ConjugateTranspose"));
+    opBuilder.addInput(x.asOutput());
+    opBuilder.addInput(perm.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new ConjugateTranspose<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> y() {
+    return y;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return y;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ConjugateTranspose";
+  
+  private Output<T> y;
+  
+  private ConjugateTranspose(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    y = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Cross.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Cross.java
new file mode 100644
index 00000000000..d223c04922e
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Cross.java
@@ -0,0 +1,81 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.linalg;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Compute the pairwise cross product.
+ * <p>
+ * `a` and `b` must be the same shape; they can either be simple 3-element vectors,
+ * or any shape where the innermost dimension is 3. In the latter case, each pair
+ * of corresponding 3-element vectors is cross-multiplied independently.
+ * 
+ * @param <T> data type for {@code product()} output
+ */
+@Operator(group = "linalg")
+public final class Cross<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new Cross operation.
+   * 
+   * @param scope current scope
+   * @param a A tensor containing 3-element vectors.
+   * @param b Another tensor, of same type and shape as `a`.
+   * @return a new instance of Cross
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> Cross<T> create(Scope scope, Operand<T> a, Operand<T> b) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Cross", scope.makeOpName("Cross"));
+    opBuilder.addInput(a.asOutput());
+    opBuilder.addInput(b.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new Cross<T>(opBuilder.build());
+  }
+  
+  /**
+   * Pairwise cross product of the vectors in `a` and `b`.
+   */
+  public Output<T> product() {
+    return product;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return product;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Cross";
+  
+  private Output<T> product;
+  
+  private Cross(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    product = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Det.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Det.java
new file mode 100644
index 00000000000..7134ef6ea44
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Det.java
@@ -0,0 +1,79 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.linalg;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Computes the determinant of one or more square matrices.
+ * <p>
+ * The input is a tensor of shape `[..., M, M]` whose inner-most 2 dimensions
+ * form square matrices. The output is a tensor containing the determinants
+ * for all input submatrices `[..., :, :]`.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator(group = "linalg")
+public final class Det<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new Det operation.
+   * 
+   * @param scope current scope
+   * @param input Shape is `[..., M, M]`.
+   * @return a new instance of Det
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> Det<T> create(Scope scope, Operand<T> input) {
+    OperationBuilder opBuilder = scope.env().opBuilder("MatrixDeterminant", scope.makeOpName("Det"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new Det<T>(opBuilder.build());
+  }
+  
+  /**
+   * Shape is `[...]`.
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "MatrixDeterminant";
+  
+  private Output<T> output;
+  
+  private Det(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Eig.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Eig.java
new file mode 100644
index 00000000000..e02e54defe6
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Eig.java
@@ -0,0 +1,129 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.linalg;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Computes the eigen decomposition of one or more square matrices.
+ * <p>
+ * Computes the eigenvalues and (optionally) right eigenvectors of each inner matrix in
+ * `input` such that `input[..., :, :] = v[..., :, :] * diag(e[..., :])`. The eigenvalues
+ * are sorted in non-decreasing order.
+ * <pre>{@code
+ * # a is a tensor.
+ * # e is a tensor of eigenvalues.
+ * # v is a tensor of eigenvectors.
+ * e, v = eig(a)
+ * e = eig(a, compute_v=False)
+ * }</pre>
+ * 
+ * 
+ * @param <U> data type for {@code e()} output
+ */
+@Operator(group = "linalg")
+public final class Eig<U extends TType> extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.linalg.Eig}
+   */
+  public static class Options {
+    
+    /**
+     * @param computeV If `True` then eigenvectors will be computed and returned in `v`.
+     * Otherwise, only the eigenvalues will be computed.
+     */
+    public Options computeV(Boolean computeV) {
+      this.computeV = computeV;
+      return this;
+    }
+    
+    private Boolean computeV;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new Eig operation.
+   * 
+   * @param scope current scope
+   * @param input `Tensor` input of shape `[N, N]`.
+   * @param Tout 
+   * @param options carries optional attributes values
+   * @return a new instance of Eig
+   */
+  @Endpoint(describeByClass = true)
+  public static <U extends TType, T extends TType> Eig<U> create(Scope scope, Operand<T> input, Class<U> Tout, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Eig", scope.makeOpName("Eig"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("Tout", Tout);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.computeV != null) {
+          opBuilder.setAttr("compute_v", opts.computeV);
+        }
+      }
+    }
+    return new Eig<U>(opBuilder.build());
+  }
+  
+  /**
+   * @param computeV If `True` then eigenvectors will be computed and returned in `v`.
+   * Otherwise, only the eigenvalues will be computed.
+   */
+  public static Options computeV(Boolean computeV) {
+    return new Options().computeV(computeV);
+  }
+  
+  /**
+   * Eigenvalues. Shape is `[N]`.
+   */
+  public Output<U> e() {
+    return e;
+  }
+  
+  /**
+   * Eigenvectors. Shape is `[N, N]`.
+   */
+  public Output<U> v() {
+    return v;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Eig";
+  
+  private Output<U> e;
+  private Output<U> v;
+  
+  private Eig(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    e = operation.output(outputIdx++);
+    v = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Einsum.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Einsum.java
new file mode 100644
index 00000000000..6817270edd1
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Einsum.java
@@ -0,0 +1,155 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.linalg;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.Operands;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Tensor contraction according to Einstein summation convention.
+ * <p>
+ * Implements generalized Tensor contraction and reduction. Each input Tensor must
+ * have a corresponding input subscript appearing in the comma-separated left-hand
+ * side of the equation. The right-hand side of the equation consists of the
+ * output subscript. The input subscripts and the output subscript should consist
+ * of zero or more named axis labels and at most one ellipsis (`...`).
+ * <p>
+ * The named axis labels may be any single character other than those having
+ * special meaning, namely `,.->`. The behavior of this Op is undefined if it
+ * receives an ill-formatted equation; since the validation is done at
+ * graph-building time, we omit format validation checks at runtime.
+ * <p>
+ * Note: This Op is <i>not</i> intended to be called by the user; instead users should
+ * call `tf.einsum` directly. It is a hidden Op used by `tf.einsum`.
+ * <p>
+ * Operations are applied to the input(s) according to the following rules:
+ * <p>
+ *  (a) Generalized Diagonals: For input dimensions corresponding to axis labels
+ *      appearing more than once in the same input subscript, we take the
+ *      generalized (`k`-dimensional) diagonal.
+ *      For example, in the equation `iii->i` with input shape `[3, 3, 3]`, the
+ *      generalized diagonal would consist of `3` elements at indices `(0, 0, 0)`,
+ *      `(1, 1, 1)` and `(2, 2, 2)` to create a Tensor of shape `[3]`.
+ * <p>
+ *  (b) Reduction: Axes corresponding to labels appearing only in one input
+ *      subscript but not in the output subscript are summed over prior to Tensor
+ *      contraction.
+ *      For example, in the equation `ab,bc->b`, the axis labels `a` and `c` are
+ *      the reduction axis labels.
+ * <p>
+ *  (c) Batch Dimensions: Axes corresponding to labels appearing in each of the
+ *      input subscripts and also in the output subscript make up the batch
+ *      dimensions in Tensor contraction. Unnamed axis labels corresponding to
+ *      ellipsis (`...`) also correspond to batch dimensions.
+ *      For example, for the equation denoting batch matrix multiplication,
+ *      `bij,bjk->bik`, the axis label `b` corresponds to a batch dimension.
+ * <p>
+ *  (d) Contraction: In case of binary einsum, axes corresponding to labels
+ *      appearing in two different inputs (and not in the output) are contracted
+ *      against each other.
+ *      Considering the batch matrix multiplication equation again
+ *      (`bij,bjk->bik`), the contracted axis label is `j`.
+ * <p>
+ *  (e) Expand Diagonal: If the output subscripts contain repeated (explicit) axis
+ *      labels, the opposite operation of (a) is applied. For example, in the
+ *      equation `i->iii`, and input shape `[3]`, the output of shape `[3, 3, 3]`
+ *      are all zeros, except for the (generalized) diagonal which is populated
+ *      with values from the input.
+ *      Note: This operation is not supported by `np.einsum` or `tf.einsum`; it is
+ *      provided to enable computing the symbolic gradient of `tf.einsum`.
+ * <p>
+ * The output subscripts must contain only labels appearing in at least one of the
+ * input subscripts. Furthermore, all dimensions mapping to the same axis label
+ * must be equal.
+ * <p>
+ * Any of the input and output subscripts may contain at most a single ellipsis
+ * (`...`). These ellipsis are mapped against dimensions not corresponding to any
+ * named axis label. If two inputs contain ellipsis, then they are broadcasted
+ * according to standard NumPy broadcasting
+ * [rules](http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html).
+ * <p>
+ * The broadcasted dimensions are placed in the corresponding location of the
+ * ellipsis in the output subscript. If the broadcasted dimensions are non-empty
+ * and the output subscripts do not contain ellipsis, then an InvalidArgument error
+ * is raised.
+ * <p>
+ * @compatibility(numpy)
+ * Similar to [`numpy.einsum`](https://docs.scipy.org/doc/numpy/reference/generated/numpy.einsum.html).
+ * <p>
+ * Comparison with `numpy.einsum`:
+ * <p>
+ *  * This Op only supports unary and binary forms of `numpy.einsum`.
+ *  * This Op does not support implicit form. (i.e. equations without `->`).
+ *  * This Op also supports repeated indices in the output subscript, which is not
+ *    supported by `numpy.einsum`.
+ * @end_compatibility
+ * 
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator(group = "linalg")
+public final class Einsum<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new Einsum operation.
+   * 
+   * @param scope current scope
+   * @param inputs List of 1 or 2 Tensors.
+   * @param equation String describing the Einstein Summation operation; in the format of np.einsum.
+   * @return a new instance of Einsum
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> Einsum<T> create(Scope scope, Iterable<Operand<T>> inputs, String equation) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Einsum", scope.makeOpName("Einsum"));
+    opBuilder.addInputList(Operands.asOutputs(inputs));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("equation", equation);
+    return new Einsum<T>(opBuilder.build());
+  }
+  
+  /**
+   * Output Tensor with shape depending upon `equation`.
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Einsum";
+  
+  private Output<T> output;
+  
+  private Einsum(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/EuclideanNorm.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/EuclideanNorm.java
new file mode 100644
index 00000000000..f8ce40f62bf
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/EuclideanNorm.java
@@ -0,0 +1,118 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.linalg;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Computes the euclidean norm of elements across dimensions of a tensor.
+ * <p>
+ * Reduces `input` along the dimensions given in `axis`. Unless
+ * `keep_dims` is true, the rank of the tensor is reduced by 1 for each entry in
+ * `axis`. If `keep_dims` is true, the reduced dimensions are
+ * retained with length 1.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator(group = "linalg")
+public final class EuclideanNorm<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.linalg.EuclideanNorm}
+   */
+  public static class Options {
+    
+    /**
+     * @param keepDims If true, retain reduced dimensions with length 1.
+     */
+    public Options keepDims(Boolean keepDims) {
+      this.keepDims = keepDims;
+      return this;
+    }
+    
+    private Boolean keepDims;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new EuclideanNorm operation.
+   * 
+   * @param scope current scope
+   * @param input The tensor to reduce.
+   * @param axis The dimensions to reduce. Must be in the range
+   * `[-rank(input), rank(input))`.
+   * @param options carries optional attributes values
+   * @return a new instance of EuclideanNorm
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType, U extends TNumber> EuclideanNorm<T> create(Scope scope, Operand<T> input, Operand<U> axis, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("EuclideanNorm", scope.makeOpName("EuclideanNorm"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(axis.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.keepDims != null) {
+          opBuilder.setAttr("keep_dims", opts.keepDims);
+        }
+      }
+    }
+    return new EuclideanNorm<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param keepDims If true, retain reduced dimensions with length 1.
+   */
+  public static Options keepDims(Boolean keepDims) {
+    return new Options().keepDims(keepDims);
+  }
+  
+  /**
+   * The reduced tensor.
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "EuclideanNorm";
+  
+  private Output<T> output;
+  
+  private EuclideanNorm(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Inv.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Inv.java
new file mode 100644
index 00000000000..e482957bdac
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Inv.java
@@ -0,0 +1,125 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.linalg;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Computes the inverse of one or more square invertible matrices or their
+ * <p>
+ * adjoints (conjugate transposes).
+ * <p>
+ * The input is a tensor of shape `[..., M, M]` whose inner-most 2 dimensions
+ * form square matrices. The output is a tensor of the same shape as the input
+ * containing the inverse for all input submatrices `[..., :, :]`.
+ * <p>
+ * The op uses LU decomposition with partial pivoting to compute the inverses.
+ * <p>
+ * If a matrix is not invertible there is no guarantee what the op does. It
+ * may detect the condition and raise an exception or it may simply return a
+ * garbage result.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator(group = "linalg")
+public final class Inv<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.linalg.Inv}
+   */
+  public static class Options {
+    
+    /**
+     * @param adjoint 
+     */
+    public Options adjoint(Boolean adjoint) {
+      this.adjoint = adjoint;
+      return this;
+    }
+    
+    private Boolean adjoint;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new Inv operation.
+   * 
+   * @param scope current scope
+   * @param input Shape is `[..., M, M]`.
+   * @param options carries optional attributes values
+   * @return a new instance of Inv
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> Inv<T> create(Scope scope, Operand<T> input, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("MatrixInverse", scope.makeOpName("Inv"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.adjoint != null) {
+          opBuilder.setAttr("adjoint", opts.adjoint);
+        }
+      }
+    }
+    return new Inv<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param adjoint 
+   */
+  public static Options adjoint(Boolean adjoint) {
+    return new Options().adjoint(adjoint);
+  }
+  
+  /**
+   * Shape is `[..., M, M]`.
+   * <p>
+   * @compatibility(numpy)
+   * Equivalent to np.linalg.inv
+   * @end_compatibility
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "MatrixInverse";
+  
+  private Output<T> output;
+  
+  private Inv(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/LoadAndRemapMatrix.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/LoadAndRemapMatrix.java
new file mode 100644
index 00000000000..1010c503eba
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/LoadAndRemapMatrix.java
@@ -0,0 +1,175 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.linalg;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.TString;
+
+/**
+ * Loads a 2-D (matrix) `Tensor` with name `old_tensor_name` from the checkpoint
+ * <p>
+ * at `ckpt_path` and potentially reorders its rows and columns using the
+ * specified remappings.
+ * <p>
+ * Most users should use one of the wrapper initializers (such as
+ * `tf.contrib.framework.load_and_remap_matrix_initializer`) instead of this
+ * function directly.
+ * <p>
+ * The remappings are 1-D tensors with the following properties:
+ * <ul>
+ * <li>
+ * `row_remapping` must have exactly `num_rows` entries. Row `i` of the output
+ *   matrix will be initialized from the row corresponding to index
+ *   `row_remapping[i]` in the old `Tensor` from the checkpoint.
+ * </li>
+ * <li>
+ * `col_remapping` must have either 0 entries (indicating that no column
+ *   reordering is needed) or `num_cols` entries. If specified, column `j` of the
+ *   output matrix will be initialized from the column corresponding to index
+ *   `col_remapping[j]` in the old `Tensor` from the checkpoint.
+ * </li>
+ * <li>
+ * A value of -1 in either of the remappings signifies a "missing" entry. In that
+ *   case, values from the `initializing_values` tensor will be used to fill that
+ *   missing row or column. If `row_remapping` has `r` missing entries and
+ *   `col_remapping` has `c` missing entries, then the following condition must be
+ *   true:
+ * </li>
+ * </ul>
+ * `(r * num_cols) + (c * num_rows) - (r * c) == len(initializing_values)`
+ * <p>
+ * The remapping tensors can be generated using the GenerateVocabRemapping op.
+ * <p>
+ * As an example, with row_remapping = [1, 0, -1], col_remapping = [0, 2, -1],
+ * initializing_values = [0.5, -0.5, 0.25, -0.25, 42], and w(i, j) representing
+ * the value from row i, column j of the old tensor in the checkpoint, the output
+ * matrix will look like the following:
+ * <p>
+ * [[w(1, 0),  w(1, 2),  0.5],
+ *  [w(0, 0),  w(0, 2), -0.5],
+ *  [0.25,    -0.25,      42]]
+ */
+@Operator(group = "linalg")
+public final class LoadAndRemapMatrix extends RawOp implements Operand<TFloat32> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.linalg.LoadAndRemapMatrix}
+   */
+  public static class Options {
+    
+    /**
+     * @param maxRowsInMemory The maximum number of rows to load from the checkpoint at
+     * once. If less than or equal to 0, the entire matrix will be loaded into
+     * memory. Setting this arg trades increased disk reads for lower memory usage.
+     */
+    public Options maxRowsInMemory(Long maxRowsInMemory) {
+      this.maxRowsInMemory = maxRowsInMemory;
+      return this;
+    }
+    
+    private Long maxRowsInMemory;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new LoadAndRemapMatrix operation.
+   * 
+   * @param scope current scope
+   * @param ckptPath Path to the TensorFlow checkpoint (version 2, `TensorBundle`) from
+   * which the old matrix `Tensor` will be loaded.
+   * @param oldTensorName Name of the 2-D `Tensor` to load from checkpoint.
+   * @param rowRemapping An int `Tensor` of row remappings (generally created by
+   * `generate_vocab_remapping`).  Even if no row remapping is needed, this must
+   * still be an index-valued Tensor (e.g. [0, 1, 2, ...]), or a shifted
+   * index-valued `Tensor` (e.g. [8, 9, 10, ...], for partitioned `Variables`).
+   * @param colRemapping An int `Tensor` of column remappings (generally created by
+   * `generate_vocab_remapping`).  May be a size-0 `Tensor` if only row remapping
+   * is to be done (e.g. column ordering is the same).
+   * @param initializingValues A float `Tensor` containing  values to fill in for cells
+   * in the output matrix that are not loaded from the checkpoint. Length must be
+   * exactly the same as the number of missing / new cells.
+   * @param numRows Number of rows (length of the 1st dimension) in the output matrix.
+   * @param numCols Number of columns (length of the 2nd dimension) in the output matrix.
+   * @param options carries optional attributes values
+   * @return a new instance of LoadAndRemapMatrix
+   */
+  @Endpoint(describeByClass = true)
+  public static LoadAndRemapMatrix create(Scope scope, Operand<TString> ckptPath, Operand<TString> oldTensorName, Operand<TInt64> rowRemapping, Operand<TInt64> colRemapping, Operand<TFloat32> initializingValues, Long numRows, Long numCols, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("LoadAndRemapMatrix", scope.makeOpName("LoadAndRemapMatrix"));
+    opBuilder.addInput(ckptPath.asOutput());
+    opBuilder.addInput(oldTensorName.asOutput());
+    opBuilder.addInput(rowRemapping.asOutput());
+    opBuilder.addInput(colRemapping.asOutput());
+    opBuilder.addInput(initializingValues.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("num_rows", numRows);
+    opBuilder.setAttr("num_cols", numCols);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.maxRowsInMemory != null) {
+          opBuilder.setAttr("max_rows_in_memory", opts.maxRowsInMemory);
+        }
+      }
+    }
+    return new LoadAndRemapMatrix(opBuilder.build());
+  }
+  
+  /**
+   * @param maxRowsInMemory The maximum number of rows to load from the checkpoint at
+   * once. If less than or equal to 0, the entire matrix will be loaded into
+   * memory. Setting this arg trades increased disk reads for lower memory usage.
+   */
+  public static Options maxRowsInMemory(Long maxRowsInMemory) {
+    return new Options().maxRowsInMemory(maxRowsInMemory);
+  }
+  
+  /**
+   * Output matrix containing existing values loaded from the
+   * checkpoint, and with any missing values filled in from initializing_values.
+   */
+  public Output<TFloat32> outputMatrix() {
+    return outputMatrix;
+  }
+  
+  @Override
+  public Output<TFloat32> asOutput() {
+    return outputMatrix;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "LoadAndRemapMatrix";
+  
+  private Output<TFloat32> outputMatrix;
+  
+  private LoadAndRemapMatrix(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    outputMatrix = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/LogMatrixDeterminant.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/LogMatrixDeterminant.java
new file mode 100644
index 00000000000..33781d561b4
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/LogMatrixDeterminant.java
@@ -0,0 +1,90 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.linalg;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Computes the sign and the log of the absolute value of the determinant of
+ * <p>
+ * one or more square matrices.
+ * <p>
+ * The input is a tensor of shape `[N, M, M]` whose inner-most 2 dimensions
+ * form square matrices. The outputs are two tensors containing the signs and
+ * absolute values of the log determinants for all N input submatrices
+ * `[..., :, :]` such that the determinant = sign*exp(log_abs_determinant).
+ * The log_abs_determinant is computed as det(P)*sum(log(diag(LU))) where LU
+ * is the LU decomposition of the input and P is the corresponding
+ * permutation matrix.
+ * 
+ * @param <T> data type for {@code sign()} output
+ */
+@Operator(group = "linalg")
+public final class LogMatrixDeterminant<T extends TType> extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new LogMatrixDeterminant operation.
+   * 
+   * @param scope current scope
+   * @param input Shape is `[N, M, M]`.
+   * @return a new instance of LogMatrixDeterminant
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> LogMatrixDeterminant<T> create(Scope scope, Operand<T> input) {
+    OperationBuilder opBuilder = scope.env().opBuilder("LogMatrixDeterminant", scope.makeOpName("LogMatrixDeterminant"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new LogMatrixDeterminant<T>(opBuilder.build());
+  }
+  
+  /**
+   * The signs of the log determinants of the inputs. Shape is `[N]`.
+   */
+  public Output<T> sign() {
+    return sign;
+  }
+  
+  /**
+   * The logs of the absolute values of the determinants
+   * of the N input matrices.  Shape is `[N]`.
+   */
+  public Output<T> logAbsDeterminant() {
+    return logAbsDeterminant;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "LogMatrixDeterminant";
+  
+  private Output<T> sign;
+  private Output<T> logAbsDeterminant;
+  
+  private LogMatrixDeterminant(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    sign = operation.output(outputIdx++);
+    logAbsDeterminant = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Lu.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Lu.java
new file mode 100644
index 00000000000..fcbdbd6d80d
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Lu.java
@@ -0,0 +1,125 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.linalg;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Computes the LU decomposition of one or more square matrices.
+ * <p>
+ * The input is a tensor of shape `[..., M, M]` whose inner-most 2 dimensions
+ * form square matrices.
+ * <p>
+ * The input has to be invertible.
+ * <p>
+ * The output consists of two tensors LU and P containing the LU decomposition
+ * of all input submatrices `[..., :, :]`. LU encodes the lower triangular and
+ * upper triangular factors.
+ * <p>
+ * For each input submatrix of shape `[M, M]`, L is a lower triangular matrix of
+ * shape `[M, M]` with unit diagonal whose entries correspond to the strictly lower
+ * triangular part of LU. U is a upper triangular matrix of shape `[M, M]` whose
+ * entries correspond to the upper triangular part, including the diagonal, of LU.
+ * <p>
+ * P represents a permutation matrix encoded as a list of indices each between `0`
+ * and `M-1`, inclusive. If P_mat denotes the permutation matrix corresponding to
+ * P, then the L, U and P satisfies P_mat * input = L * U.
+ * 
+ * @param <T> data type for {@code lu()} output
+ * @param <U> data type for {@code p()} output
+ */
+@Operator(group = "linalg")
+public final class Lu<T extends TType, U extends TNumber> extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new Lu operation.
+   * 
+   * @param scope current scope
+   * @param input A tensor of shape `[..., M, M]` whose inner-most 2 dimensions form matrices of
+   * size `[M, M]`.
+   * @param outputIdxType 
+   * @return a new instance of Lu
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType, U extends TNumber> Lu<T, U> create(Scope scope, Operand<T> input, Class<U> outputIdxType) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Lu", scope.makeOpName("Lu"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("output_idx_type", outputIdxType);
+    return new Lu<T, U>(opBuilder.build());
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new Lu operation using default output types.
+   * 
+   * @param scope current scope
+   * @param input A tensor of shape `[..., M, M]` whose inner-most 2 dimensions form matrices of
+   * size `[M, M]`.
+   * @return a new instance of Lu
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> Lu<T, TInt32> create(Scope scope, Operand<T> input) {
+    return create(scope, input, TInt32.class);
+  }
+  
+  /**
+   * A tensor of shape `[..., M, M]` whose strictly lower triangular part denotes the
+   * lower triangular factor `L` with unit diagonal, and whose upper triangular part
+   * denotes the upper triangular factor `U`.
+   */
+  public Output<T> lu() {
+    return lu;
+  }
+  
+  /**
+   * Permutation of the rows encoded as a list of indices in `0..M-1`. Shape is
+   * `[..., M]`.
+   * @compatibility(scipy)
+   * Similar to `scipy.linalg.lu`, except the triangular factors `L` and `U` are
+   * packed into a single tensor, the permutation is applied to `input` instead of
+   * the right hand side and the permutation `P` is returned as a list of indices
+   * instead of a permutation matrix.
+   * @end_compatibility
+   */
+  public Output<U> p() {
+    return p;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Lu";
+  
+  private Output<T> lu;
+  private Output<U> p;
+  
+  private Lu(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    lu = operation.output(outputIdx++);
+    p = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/MatMul.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/MatMul.java
new file mode 100644
index 00000000000..8b227e3910c
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/MatMul.java
@@ -0,0 +1,137 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.linalg;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Multiply the matrix "a" by the matrix "b".
+ * <p>
+ * The inputs must be two-dimensional matrices and the inner dimension of
+ * "a" (after being transposed if transpose_a is true) must match the
+ * outer dimension of "b" (after being transposed if transposed_b is
+ * true).
+ * <p>
+ * <i>Note</i>: The default kernel implementation for MatMul on GPUs uses
+ * cublas.
+ * 
+ * @param <T> data type for {@code product()} output
+ */
+@Operator(group = "linalg")
+public final class MatMul<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.linalg.MatMul}
+   */
+  public static class Options {
+    
+    /**
+     * @param transposeA If true, "a" is transposed before multiplication.
+     */
+    public Options transposeA(Boolean transposeA) {
+      this.transposeA = transposeA;
+      return this;
+    }
+    
+    /**
+     * @param transposeB If true, "b" is transposed before multiplication.
+     */
+    public Options transposeB(Boolean transposeB) {
+      this.transposeB = transposeB;
+      return this;
+    }
+    
+    private Boolean transposeA;
+    private Boolean transposeB;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new MatMul operation.
+   * 
+   * @param scope current scope
+   * @param a 
+   * @param b 
+   * @param options carries optional attributes values
+   * @return a new instance of MatMul
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> MatMul<T> create(Scope scope, Operand<T> a, Operand<T> b, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("MatMul", scope.makeOpName("MatMul"));
+    opBuilder.addInput(a.asOutput());
+    opBuilder.addInput(b.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.transposeA != null) {
+          opBuilder.setAttr("transpose_a", opts.transposeA);
+        }
+        if (opts.transposeB != null) {
+          opBuilder.setAttr("transpose_b", opts.transposeB);
+        }
+      }
+    }
+    return new MatMul<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param transposeA If true, "a" is transposed before multiplication.
+   */
+  public static Options transposeA(Boolean transposeA) {
+    return new Options().transposeA(transposeA);
+  }
+  
+  /**
+   * @param transposeB If true, "b" is transposed before multiplication.
+   */
+  public static Options transposeB(Boolean transposeB) {
+    return new Options().transposeB(transposeB);
+  }
+  
+  /**
+   */
+  public Output<T> product() {
+    return product;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return product;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "MatMul";
+  
+  private Output<T> product;
+  
+  private MatMul(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    product = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/MatrixDiag.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/MatrixDiag.java
new file mode 100644
index 00000000000..b187ecc0930
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/MatrixDiag.java
@@ -0,0 +1,177 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.linalg;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Returns a batched diagonal tensor with given batched diagonal values.
+ * <p>
+ * Returns a tensor with the contents in `diagonal` as `k[0]`-th to `k[1]`-th
+ * diagonals of a matrix, with everything else padded with `padding`. `num_rows`
+ * and `num_cols` specify the dimension of the innermost matrix of the output. If
+ * both are not specified, the op assumes the innermost matrix is square and infers
+ * its size from `k` and the innermost dimension of `diagonal`. If only one of them
+ * is specified, the op assumes the unspecified value is the smallest possible
+ * based on other criteria.
+ * <p>
+ * Let `diagonal` have `r` dimensions `[I, J, ..., L, M, N]`. The output tensor has
+ * rank `r+1` with shape `[I, J, ..., L, M, num_rows, num_cols]` when only one
+ * diagonal is given (`k` is an integer or `k[0] == k[1]`). Otherwise, it has rank
+ * `r` with shape `[I, J, ..., L, num_rows, num_cols]`.
+ * <p>
+ * The second innermost dimension of `diagonal` has double meaning.
+ * When `k` is scalar or `k[0] == k[1]`, `M` is part of the batch size
+ * [I, J, ..., M], and the output tensor is:
+ * <pre>{@code
+ * output[i, j, ..., l, m, n]
+ *   = diagonal[i, j, ..., l, n-max(d_upper, 0)] ; if n - m == d_upper
+ *     padding_value                             ; otherwise
+ * }</pre>
+ * Otherwise, `M` is treated as the number of diagonals for the matrix in the
+ * same batch (`M = k[1]-k[0]+1`), and the output tensor is:
+ * <pre>{@code
+ * output[i, j, ..., l, m, n]
+ *   = diagonal[i, j, ..., l, diag_index, index_in_diag] ; if k[0] <= d <= k[1]
+ *     padding_value                                     ; otherwise
+ * }</pre>
+ * where `d = n - m`, `diag_index = k[1] - d`, and `index_in_diag = n - max(d, 0)`.
+ * <p>
+ * For example:
+ * <pre>{@code
+ * # The main diagonal.
+ * diagonal = np.array([[1, 2, 3, 4],            # Input shape: (2, 4)
+ *                      [5, 6, 7, 8]])
+ * tf.matrix_diag(diagonal) ==> [[[1, 0, 0, 0],  # Output shape: (2, 4, 4)
+ *                                [0, 2, 0, 0],
+ *                                [0, 0, 3, 0],
+ *                                [0, 0, 0, 4]],
+ *                               [[5, 0, 0, 0],
+ *                                [0, 6, 0, 0],
+ *                                [0, 0, 7, 0],
+ *                                [0, 0, 0, 8]]]
+ * 
+ * # A superdiagonal (per batch).
+ * diagonal = np.array([[1, 2, 3],  # Input shape: (2, 3)
+ *                      [4, 5, 6]])
+ * tf.matrix_diag(diagonal, k = 1)
+ *   ==> [[[0, 1, 0, 0],  # Output shape: (2, 4, 4)
+ *         [0, 0, 2, 0],
+ *         [0, 0, 0, 3],
+ *         [0, 0, 0, 0]],
+ *        [[0, 4, 0, 0],
+ *         [0, 0, 5, 0],
+ *         [0, 0, 0, 6],
+ *         [0, 0, 0, 0]]]
+ * 
+ * # A band of diagonals.
+ * diagonals = np.array([[[1, 2, 3],  # Input shape: (2, 2, 3)
+ *                        [4, 5, 0]],
+ *                       [[6, 7, 9],
+ *                        [9, 1, 0]]])
+ * tf.matrix_diag(diagonals, k = (-1, 0))
+ *   ==> [[[1, 0, 0],  # Output shape: (2, 3, 3)
+ *         [4, 2, 0],
+ *         [0, 5, 3]],
+ *        [[6, 0, 0],
+ *         [9, 7, 0],
+ *         [0, 1, 9]]]
+ * 
+ * # Rectangular matrix.
+ * diagonal = np.array([1, 2])  # Input shape: (2)
+ * tf.matrix_diag(diagonal, k = -1, num_rows = 3, num_cols = 4)
+ *   ==> [[0, 0, 0, 0],  # Output shape: (3, 4)
+ *        [1, 0, 0, 0],
+ *        [0, 2, 0, 0]]
+ * 
+ * # Rectangular matrix with inferred num_cols and padding_value = 9.
+ * tf.matrix_diag(diagonal, k = -1, num_rows = 3, padding_value = 9)
+ *   ==> [[9, 9],  # Output shape: (3, 2)
+ *        [1, 9],
+ *        [9, 2]]
+ * }</pre>
+ * 
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator(group = "linalg")
+public final class MatrixDiag<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new MatrixDiag operation.
+   * 
+   * @param scope current scope
+   * @param diagonal Rank `r`, where `r >= 1`
+   * @param k Diagonal offset(s). Positive value means superdiagonal, 0 refers to the main
+   * diagonal, and negative value means subdiagonals. `k` can be a single integer
+   * (for a single diagonal) or a pair of integers specifying the low and high ends
+   * of a matrix band. `k[0]` must not be larger than `k[1]`.
+   * @param numRows The number of rows of the output matrix. If it is not provided, the op assumes
+   * the output matrix is a square matrix and infers the matrix size from k and the
+   * innermost dimension of `diagonal`.
+   * @param numCols The number of columns of the output matrix. If it is not provided, the op
+   * assumes the output matrix is a square matrix and infers the matrix size from
+   * k and the innermost dimension of `diagonal`.
+   * @param paddingValue The number to fill the area outside the specified diagonal band with.
+   * Default is 0.
+   * @return a new instance of MatrixDiag
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> MatrixDiag<T> create(Scope scope, Operand<T> diagonal, Operand<TInt32> k, Operand<TInt32> numRows, Operand<TInt32> numCols, Operand<T> paddingValue) {
+    OperationBuilder opBuilder = scope.env().opBuilder("MatrixDiagV2", scope.makeOpName("MatrixDiag"));
+    opBuilder.addInput(diagonal.asOutput());
+    opBuilder.addInput(k.asOutput());
+    opBuilder.addInput(numRows.asOutput());
+    opBuilder.addInput(numCols.asOutput());
+    opBuilder.addInput(paddingValue.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new MatrixDiag<T>(opBuilder.build());
+  }
+  
+  /**
+   * Has rank `r+1` when `k` is an integer or `k[0] == k[1]`, rank `r` otherwise.
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "MatrixDiagV2";
+  
+  private Output<T> output;
+  
+  private MatrixDiag(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/MatrixDiagPart.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/MatrixDiagPart.java
new file mode 100644
index 00000000000..c4dee3f7d76
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/MatrixDiagPart.java
@@ -0,0 +1,151 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.linalg;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Returns the batched diagonal part of a batched tensor.
+ * <p>
+ * Returns a tensor with the `k[0]`-th to `k[1]`-th diagonals of the batched
+ * `input`.
+ * <p>
+ * Assume `input` has `r` dimensions `[I, J, ..., L, M, N]`.
+ * Let `max_diag_len` be the maximum length among all diagonals to be extracted,
+ * `max_diag_len = min(M + min(k[1], 0), N + min(-k[0], 0))`
+ * Let `num_diags` be the number of diagonals to extract,
+ * `num_diags = k[1] - k[0] + 1`.
+ * <p>
+ * If `num_diags == 1`, the output tensor is of rank `r - 1` with shape
+ * `[I, J, ..., L, max_diag_len]` and values:
+ * <pre>{@code
+ * diagonal[i, j, ..., l, n]
+ *   = input[i, j, ..., l, n+y, n+x] ; if 0 <= n+y < M and 0 <= n+x < N,
+ *     padding_value                 ; otherwise.
+ * }</pre>
+ * where `y = max(-k[1], 0)`, `x = max(k[1], 0)`.
+ * <p>
+ * Otherwise, the output tensor has rank `r` with dimensions
+ * `[I, J, ..., L, num_diags, max_diag_len]` with values:
+ * <pre>{@code
+ * diagonal[i, j, ..., l, m, n]
+ *   = input[i, j, ..., l, n+y, n+x] ; if 0 <= n+y < M and 0 <= n+x < N,
+ *     padding_value                 ; otherwise.
+ * }</pre>
+ * where `d = k[1] - m`, `y = max(-d, 0)`, and `x = max(d, 0)`.
+ * <p>
+ * The input must be at least a matrix.
+ * <p>
+ * For example:
+ * <pre>{@code
+ * input = np.array([[[1, 2, 3, 4],  # Input shape: (2, 3, 4)
+ *                    [5, 6, 7, 8],
+ *                    [9, 8, 7, 6]],
+ *                   [[5, 4, 3, 2],
+ *                    [1, 2, 3, 4],
+ *                    [5, 6, 7, 8]]])
+ * 
+ * # A main diagonal from each batch.
+ * tf.matrix_diag_part(input) ==> [[1, 6, 7],  # Output shape: (2, 3)
+ *                                 [5, 2, 7]]
+ * 
+ * # A superdiagonal from each batch.
+ * tf.matrix_diag_part(input, k = 1)
+ *   ==> [[2, 7, 6],  # Output shape: (2, 3)
+ *        [4, 3, 8]]
+ * 
+ * # A tridiagonal band from each batch.
+ * tf.matrix_diag_part(input, k = (-1, 1))
+ *   ==> [[[2, 7, 6],  # Output shape: (2, 3, 3)
+ *         [1, 6, 7],
+ *         [5, 8, 0]],
+ *        [[4, 3, 8],
+ *         [5, 2, 7],
+ *         [1, 6, 0]]]
+ * 
+ * # Padding value = 9
+ * tf.matrix_diag_part(input, k = (1, 3), padding_value = 9)
+ *   ==> [[[4, 9, 9],  # Output shape: (2, 3, 3)
+ *         [3, 8, 9],
+ *         [2, 7, 6]],
+ *        [[2, 9, 9],
+ *         [3, 4, 9],
+ *         [4, 3, 8]]]
+ * }</pre>
+ * 
+ * 
+ * @param <T> data type for {@code diagonal()} output
+ */
+@Operator(group = "linalg")
+public final class MatrixDiagPart<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new MatrixDiagPart operation.
+   * 
+   * @param scope current scope
+   * @param input Rank `r` tensor where `r >= 2`.
+   * @param k Diagonal offset(s). Positive value means superdiagonal, 0 refers to the main
+   * diagonal, and negative value means subdiagonals. `k` can be a single integer
+   * (for a single diagonal) or a pair of integers specifying the low and high ends
+   * of a matrix band. `k[0]` must not be larger than `k[1]`.
+   * @param paddingValue The value to fill the area outside the specified diagonal band with.
+   * Default is 0.
+   * @return a new instance of MatrixDiagPart
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> MatrixDiagPart<T> create(Scope scope, Operand<T> input, Operand<TInt32> k, Operand<T> paddingValue) {
+    OperationBuilder opBuilder = scope.env().opBuilder("MatrixDiagPartV2", scope.makeOpName("MatrixDiagPart"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(k.asOutput());
+    opBuilder.addInput(paddingValue.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new MatrixDiagPart<T>(opBuilder.build());
+  }
+  
+  /**
+   * The extracted diagonal(s).
+   */
+  public Output<T> diagonal() {
+    return diagonal;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return diagonal;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "MatrixDiagPartV2";
+  
+  private Output<T> diagonal;
+  
+  private MatrixDiagPart(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    diagonal = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/MatrixDiagPartV3.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/MatrixDiagPartV3.java
new file mode 100644
index 00000000000..5803f9251f5
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/MatrixDiagPartV3.java
@@ -0,0 +1,228 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.linalg;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Returns the batched diagonal part of a batched tensor.
+ * <p>
+ * Returns a tensor with the `k[0]`-th to `k[1]`-th diagonals of the batched
+ * `input`.
+ * <p>
+ * Assume `input` has `r` dimensions `[I, J, ..., L, M, N]`.
+ * Let `max_diag_len` be the maximum length among all diagonals to be extracted,
+ * `max_diag_len = min(M + min(k[1], 0), N + min(-k[0], 0))`
+ * Let `num_diags` be the number of diagonals to extract,
+ * `num_diags = k[1] - k[0] + 1`.
+ * <p>
+ * If `num_diags == 1`, the output tensor is of rank `r - 1` with shape
+ * `[I, J, ..., L, max_diag_len]` and values:
+ * <pre>{@code
+ * diagonal[i, j, ..., l, n]
+ *   = input[i, j, ..., l, n+y, n+x] ; if 0 <= n+y < M and 0 <= n+x < N,
+ *     padding_value                 ; otherwise.
+ * }</pre>
+ * where `y = max(-k[1], 0)`, `x = max(k[1], 0)`.
+ * <p>
+ * Otherwise, the output tensor has rank `r` with dimensions
+ * `[I, J, ..., L, num_diags, max_diag_len]` with values:
+ * <pre>{@code
+ * diagonal[i, j, ..., l, m, n]
+ *   = input[i, j, ..., l, n+y, n+x] ; if 0 <= n+y < M and 0 <= n+x < N,
+ *     padding_value                 ; otherwise.
+ * }</pre>
+ * where `d = k[1] - m`, `y = max(-d, 0) - offset`, and `x = max(d, 0) - offset`.
+ * <p>
+ * `offset` is zero except when the alignment of the diagonal is to the right.
+ * <pre>{@code
+ * offset = max_diag_len - diag_len(d) ; if (`align` in {RIGHT_LEFT, RIGHT_RIGHT}
+ *                                            and `d >= 0`) or
+ *                                          (`align` in {LEFT_RIGHT, RIGHT_RIGHT}
+ *                                            and `d <= 0`)
+ *          0                          ; otherwise
+ * }</pre>
+ * where `diag_len(d) = min(cols - max(d, 0), rows + min(d, 0))`.
+ * <p>
+ * The input must be at least a matrix.
+ * <p>
+ * For example:
+ * <pre>{@code
+ * input = np.array([[[1, 2, 3, 4],  # Input shape: (2, 3, 4)
+ *                    [5, 6, 7, 8],
+ *                    [9, 8, 7, 6]],
+ *                   [[5, 4, 3, 2],
+ *                    [1, 2, 3, 4],
+ *                    [5, 6, 7, 8]]])
+ * 
+ * # A main diagonal from each batch.
+ * tf.matrix_diag_part(input) ==> [[1, 6, 7],  # Output shape: (2, 3)
+ *                                 [5, 2, 7]]
+ * 
+ * # A superdiagonal from each batch.
+ * tf.matrix_diag_part(input, k = 1)
+ *   ==> [[2, 7, 6],  # Output shape: (2, 3)
+ *        [4, 3, 8]]
+ * 
+ * # A band from each batch.
+ * tf.matrix_diag_part(input, k = (-1, 2))
+ *   ==> [[[0, 3, 8],  # Output shape: (2, 4, 3)
+ *         [2, 7, 6],
+ *         [1, 6, 7],
+ *         [5, 8, 0]],
+ *        [[0, 3, 4],
+ *         [4, 3, 8],
+ *         [5, 2, 7],
+ *         [1, 6, 0]]]
+ * 
+ * # LEFT_RIGHT alignment.
+ * tf.matrix_diag_part(input, k = (-1, 2), align="LEFT_RIGHT")
+ *   ==> [[[3, 8, 0],  # Output shape: (2, 4, 3)
+ *         [2, 7, 6],
+ *         [1, 6, 7],
+ *         [0, 5, 8]],
+ *        [[3, 4, 0],
+ *         [4, 3, 8],
+ *         [5, 2, 7],
+ *         [0, 1, 6]]]
+ * 
+ * # max_diag_len can be shorter than the main diagonal.
+ * tf.matrix_diag_part(input, k = (-2, -1))
+ *   ==> [[[5, 8],
+ *         [9, 0]],
+ *        [[1, 6],
+ *         [5, 0]]]
+ * 
+ * # padding_value = 9
+ * tf.matrix_diag_part(input, k = (1, 3), padding_value = 9)
+ *   ==> [[[9, 9, 4],  # Output shape: (2, 3, 3)
+ *         [9, 3, 8],
+ *         [2, 7, 6]],
+ *        [[9, 9, 2],
+ *         [9, 3, 4],
+ *         [4, 3, 8]]]
+ * 
+ * }</pre>
+ * 
+ * 
+ * @param <T> data type for {@code diagonal()} output
+ */
+@Operator(group = "linalg")
+public final class MatrixDiagPartV3<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.linalg.MatrixDiagPartV3}
+   */
+  public static class Options {
+    
+    /**
+     * @param align Some diagonals are shorter than `max_diag_len` and need to be padded. `align` is
+     * a string specifying how superdiagonals and subdiagonals should be aligned,
+     * respectively. There are four possible alignments: "RIGHT_LEFT" (default),
+     * "LEFT_RIGHT", "LEFT_LEFT", and "RIGHT_RIGHT". "RIGHT_LEFT" aligns superdiagonals
+     * to the right (left-pads the row) and subdiagonals to the left (right-pads the
+     * row). It is the packing format LAPACK uses. cuSPARSE uses "LEFT_RIGHT", which is
+     * the opposite alignment.
+     */
+    public Options align(String align) {
+      this.align = align;
+      return this;
+    }
+    
+    private String align;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new MatrixDiagPartV3 operation.
+   * 
+   * @param scope current scope
+   * @param input Rank `r` tensor where `r >= 2`.
+   * @param k Diagonal offset(s). Positive value means superdiagonal, 0 refers to the main
+   * diagonal, and negative value means subdiagonals. `k` can be a single integer
+   * (for a single diagonal) or a pair of integers specifying the low and high ends
+   * of a matrix band. `k[0]` must not be larger than `k[1]`.
+   * @param paddingValue The value to fill the area outside the specified diagonal band with.
+   * Default is 0.
+   * @param options carries optional attributes values
+   * @return a new instance of MatrixDiagPartV3
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> MatrixDiagPartV3<T> create(Scope scope, Operand<T> input, Operand<TInt32> k, Operand<T> paddingValue, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("MatrixDiagPartV3", scope.makeOpName("MatrixDiagPartV3"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(k.asOutput());
+    opBuilder.addInput(paddingValue.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.align != null) {
+          opBuilder.setAttr("align", opts.align);
+        }
+      }
+    }
+    return new MatrixDiagPartV3<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param align Some diagonals are shorter than `max_diag_len` and need to be padded. `align` is
+   * a string specifying how superdiagonals and subdiagonals should be aligned,
+   * respectively. There are four possible alignments: "RIGHT_LEFT" (default),
+   * "LEFT_RIGHT", "LEFT_LEFT", and "RIGHT_RIGHT". "RIGHT_LEFT" aligns superdiagonals
+   * to the right (left-pads the row) and subdiagonals to the left (right-pads the
+   * row). It is the packing format LAPACK uses. cuSPARSE uses "LEFT_RIGHT", which is
+   * the opposite alignment.
+   */
+  public static Options align(String align) {
+    return new Options().align(align);
+  }
+  
+  /**
+   * The extracted diagonal(s).
+   */
+  public Output<T> diagonal() {
+    return diagonal;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return diagonal;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "MatrixDiagPartV3";
+  
+  private Output<T> diagonal;
+  
+  private MatrixDiagPartV3(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    diagonal = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/MatrixDiagV3.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/MatrixDiagV3.java
new file mode 100644
index 00000000000..10b57ecc19a
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/MatrixDiagV3.java
@@ -0,0 +1,252 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.linalg;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Returns a batched diagonal tensor with given batched diagonal values.
+ * <p>
+ * Returns a tensor with the contents in `diagonal` as `k[0]`-th to `k[1]`-th
+ * diagonals of a matrix, with everything else padded with `padding`. `num_rows`
+ * and `num_cols` specify the dimension of the innermost matrix of the output. If
+ * both are not specified, the op assumes the innermost matrix is square and infers
+ * its size from `k` and the innermost dimension of `diagonal`. If only one of them
+ * is specified, the op assumes the unspecified value is the smallest possible
+ * based on other criteria.
+ * <p>
+ * Let `diagonal` have `r` dimensions `[I, J, ..., L, M, N]`. The output tensor has
+ * rank `r+1` with shape `[I, J, ..., L, M, num_rows, num_cols]` when only one
+ * diagonal is given (`k` is an integer or `k[0] == k[1]`). Otherwise, it has rank
+ * `r` with shape `[I, J, ..., L, num_rows, num_cols]`.
+ * <p>
+ * The second innermost dimension of `diagonal` has double meaning.
+ * When `k` is scalar or `k[0] == k[1]`, `M` is part of the batch size
+ * [I, J, ..., M], and the output tensor is:
+ * <pre>{@code
+ * output[i, j, ..., l, m, n]
+ *   = diagonal[i, j, ..., l, n-max(d_upper, 0)] ; if n - m == d_upper
+ *     padding_value                             ; otherwise
+ * }</pre>
+ * Otherwise, `M` is treated as the number of diagonals for the matrix in the
+ * same batch (`M = k[1]-k[0]+1`), and the output tensor is:
+ * <pre>{@code
+ * output[i, j, ..., l, m, n]
+ *   = diagonal[i, j, ..., l, diag_index, index_in_diag] ; if k[0] <= d <= k[1]
+ *     padding_value                                     ; otherwise
+ * }</pre>
+ * where `d = n - m`, `diag_index = [k] - d`, and
+ * `index_in_diag = n - max(d, 0) + offset`.
+ * <p>
+ * `offset` is zero except when the alignment of the diagonal is to the right.
+ * <pre>{@code
+ * offset = max_diag_len - diag_len(d) ; if (`align` in {RIGHT_LEFT, RIGHT_RIGHT}
+ *                                            and `d >= 0`) or
+ *                                          (`align` in {LEFT_RIGHT, RIGHT_RIGHT}
+ *                                            and `d <= 0`)
+ *          0                          ; otherwise
+ * }</pre>
+ * where `diag_len(d) = min(cols - max(d, 0), rows + min(d, 0))`.
+ * <p>
+ * For example:
+ * <pre>{@code
+ * # The main diagonal.
+ * diagonal = np.array([[1, 2, 3, 4],            # Input shape: (2, 4)
+ *                      [5, 6, 7, 8]])
+ * tf.matrix_diag(diagonal) ==> [[[1, 0, 0, 0],  # Output shape: (2, 4, 4)
+ *                                [0, 2, 0, 0],
+ *                                [0, 0, 3, 0],
+ *                                [0, 0, 0, 4]],
+ *                               [[5, 0, 0, 0],
+ *                                [0, 6, 0, 0],
+ *                                [0, 0, 7, 0],
+ *                                [0, 0, 0, 8]]]
+ * 
+ * # A superdiagonal (per batch).
+ * diagonal = np.array([[1, 2, 3],  # Input shape: (2, 3)
+ *                      [4, 5, 6]])
+ * tf.matrix_diag(diagonal, k = 1)
+ *   ==> [[[0, 1, 0, 0],  # Output shape: (2, 4, 4)
+ *         [0, 0, 2, 0],
+ *         [0, 0, 0, 3],
+ *         [0, 0, 0, 0]],
+ *        [[0, 4, 0, 0],
+ *         [0, 0, 5, 0],
+ *         [0, 0, 0, 6],
+ *         [0, 0, 0, 0]]]
+ * 
+ * # A tridiagonal band (per batch).
+ * diagonals = np.array([[[0, 8, 9],  # Input shape: (2, 2, 3)
+ *                        [1, 2, 3],
+ *                        [4, 5, 0]],
+ *                       [[0, 2, 3],
+ *                        [6, 7, 9],
+ *                        [9, 1, 0]]])
+ * tf.matrix_diag(diagonals, k = (-1, 1))
+ *   ==> [[[1, 8, 0],  # Output shape: (2, 3, 3)
+ *         [4, 2, 9],
+ *         [0, 5, 3]],
+ *        [[6, 2, 0],
+ *         [9, 7, 3],
+ *         [0, 1, 9]]]
+ * 
+ * # LEFT_RIGHT alignment.
+ * diagonals = np.array([[[8, 9, 0],  # Input shape: (2, 2, 3)
+ *                        [1, 2, 3],
+ *                        [0, 4, 5]],
+ *                       [[2, 3, 0],
+ *                        [6, 7, 9],
+ *                        [0, 9, 1]]])
+ * tf.matrix_diag(diagonals, k = (-1, 1), align="LEFT_RIGHT")
+ *   ==> [[[1, 8, 0],  # Output shape: (2, 3, 3)
+ *         [4, 2, 9],
+ *         [0, 5, 3]],
+ *        [[6, 2, 0],
+ *         [9, 7, 3],
+ *         [0, 1, 9]]]
+ * 
+ * # Rectangular matrix.
+ * diagonal = np.array([1, 2])  # Input shape: (2)
+ * tf.matrix_diag(diagonal, k = -1, num_rows = 3, num_cols = 4)
+ *   ==> [[0, 0, 0, 0],  # Output shape: (3, 4)
+ *        [1, 0, 0, 0],
+ *        [0, 2, 0, 0]]
+ * 
+ * # Rectangular matrix with inferred num_cols and padding_value = 9.
+ * tf.matrix_diag(diagonal, k = -1, num_rows = 3, padding_value = 9)
+ *   ==> [[9, 9],  # Output shape: (3, 2)
+ *        [1, 9],
+ *        [9, 2]]
+ * 
+ * }</pre>
+ * 
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator(group = "linalg")
+public final class MatrixDiagV3<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.linalg.MatrixDiagV3}
+   */
+  public static class Options {
+    
+    /**
+     * @param align Some diagonals are shorter than `max_diag_len` and need to be padded. `align` is
+     * a string specifying how superdiagonals and subdiagonals should be aligned,
+     * respectively. There are four possible alignments: "RIGHT_LEFT" (default),
+     * "LEFT_RIGHT", "LEFT_LEFT", and "RIGHT_RIGHT". "RIGHT_LEFT" aligns superdiagonals
+     * to the right (left-pads the row) and subdiagonals to the left (right-pads the
+     * row). It is the packing format LAPACK uses. cuSPARSE uses "LEFT_RIGHT", which is
+     * the opposite alignment.
+     */
+    public Options align(String align) {
+      this.align = align;
+      return this;
+    }
+    
+    private String align;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new MatrixDiagV3 operation.
+   * 
+   * @param scope current scope
+   * @param diagonal Rank `r`, where `r >= 1`
+   * @param k Diagonal offset(s). Positive value means superdiagonal, 0 refers to the main
+   * diagonal, and negative value means subdiagonals. `k` can be a single integer
+   * (for a single diagonal) or a pair of integers specifying the low and high ends
+   * of a matrix band. `k[0]` must not be larger than `k[1]`.
+   * @param numRows The number of rows of the output matrix. If it is not provided, the op assumes
+   * the output matrix is a square matrix and infers the matrix size from k and the
+   * innermost dimension of `diagonal`.
+   * @param numCols The number of columns of the output matrix. If it is not provided, the op
+   * assumes the output matrix is a square matrix and infers the matrix size from
+   * k and the innermost dimension of `diagonal`.
+   * @param paddingValue The number to fill the area outside the specified diagonal band with.
+   * Default is 0.
+   * @param options carries optional attributes values
+   * @return a new instance of MatrixDiagV3
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> MatrixDiagV3<T> create(Scope scope, Operand<T> diagonal, Operand<TInt32> k, Operand<TInt32> numRows, Operand<TInt32> numCols, Operand<T> paddingValue, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("MatrixDiagV3", scope.makeOpName("MatrixDiagV3"));
+    opBuilder.addInput(diagonal.asOutput());
+    opBuilder.addInput(k.asOutput());
+    opBuilder.addInput(numRows.asOutput());
+    opBuilder.addInput(numCols.asOutput());
+    opBuilder.addInput(paddingValue.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.align != null) {
+          opBuilder.setAttr("align", opts.align);
+        }
+      }
+    }
+    return new MatrixDiagV3<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param align Some diagonals are shorter than `max_diag_len` and need to be padded. `align` is
+   * a string specifying how superdiagonals and subdiagonals should be aligned,
+   * respectively. There are four possible alignments: "RIGHT_LEFT" (default),
+   * "LEFT_RIGHT", "LEFT_LEFT", and "RIGHT_RIGHT". "RIGHT_LEFT" aligns superdiagonals
+   * to the right (left-pads the row) and subdiagonals to the left (right-pads the
+   * row). It is the packing format LAPACK uses. cuSPARSE uses "LEFT_RIGHT", which is
+   * the opposite alignment.
+   */
+  public static Options align(String align) {
+    return new Options().align(align);
+  }
+  
+  /**
+   * Has rank `r+1` when `k` is an integer or `k[0] == k[1]`, rank `r` otherwise.
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "MatrixDiagV3";
+  
+  private Output<T> output;
+  
+  private MatrixDiagV3(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/MatrixLogarithm.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/MatrixLogarithm.java
new file mode 100644
index 00000000000..71d3a2c9575
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/MatrixLogarithm.java
@@ -0,0 +1,94 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.linalg;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Computes the matrix logarithm of one or more square matrices:
+ * <p>
+ * 
+ * \\(log(exp(A)) = A\\)
+ * <p>
+ * This op is only defined for complex matrices. If A is positive-definite and
+ * real, then casting to a complex matrix, taking the logarithm and casting back
+ * to a real matrix will give the correct result.
+ * <p>
+ * This function computes the matrix logarithm using the Schur-Parlett algorithm.
+ * Details of the algorithm can be found in Section 11.6.2 of:
+ * Nicholas J. Higham, Functions of Matrices: Theory and Computation, SIAM 2008.
+ * ISBN 978-0-898716-46-7.
+ * <p>
+ * The input is a tensor of shape `[..., M, M]` whose inner-most 2 dimensions
+ * form square matrices. The output is a tensor of the same shape as the input
+ * containing the exponential for all input submatrices `[..., :, :]`.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+public final class MatrixLogarithm<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new MatrixLogarithm operation.
+   * 
+   * @param scope current scope
+   * @param input Shape is `[..., M, M]`.
+   * @return a new instance of MatrixLogarithm
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> MatrixLogarithm<T> create(Scope scope, Operand<T> input) {
+    OperationBuilder opBuilder = scope.env().opBuilder("MatrixLogarithm", scope.makeOpName("MatrixLogarithm"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new MatrixLogarithm<T>(opBuilder.build());
+  }
+  
+  /**
+   * Shape is `[..., M, M]`.
+   * <p>
+   * @compatibility(scipy)
+   * Equivalent to scipy.linalg.logm
+   * @end_compatibility
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "MatrixLogarithm";
+  
+  private Output<T> output;
+  
+  private MatrixLogarithm(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/MatrixSetDiag.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/MatrixSetDiag.java
new file mode 100644
index 00000000000..80367a3670b
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/MatrixSetDiag.java
@@ -0,0 +1,232 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.linalg;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Returns a batched matrix tensor with new batched diagonal values.
+ * <p>
+ * Given `input` and `diagonal`, this operation returns a tensor with the
+ * same shape and values as `input`, except for the specified diagonals of the
+ * innermost matrices. These will be overwritten by the values in `diagonal`.
+ * <p>
+ * `input` has `r+1` dimensions `[I, J, ..., L, M, N]`. When `k` is scalar or
+ * `k[0] == k[1]`, `diagonal` has `r` dimensions `[I, J, ..., L, max_diag_len]`.
+ * Otherwise, it has `r+1` dimensions `[I, J, ..., L, num_diags, max_diag_len]`.
+ * `num_diags` is the number of diagonals, `num_diags = k[1] - k[0] + 1`.
+ * `max_diag_len` is the longest diagonal in the range `[k[0], k[1]]`,
+ * `max_diag_len = min(M + min(k[1], 0), N + min(-k[0], 0))`
+ * <p>
+ * The output is a tensor of rank `k+1` with dimensions `[I, J, ..., L, M, N]`.
+ * If `k` is scalar or `k[0] == k[1]`:
+ * <pre>{@code
+ * output[i, j, ..., l, m, n]
+ *   = diagonal[i, j, ..., l, n-max(k[1], 0)] ; if n - m == k[1]
+ *     input[i, j, ..., l, m, n]              ; otherwise
+ * }</pre>
+ * Otherwise,
+ * <pre>{@code
+ * output[i, j, ..., l, m, n]
+ *   = diagonal[i, j, ..., l, diag_index, index_in_diag] ; if k[0] <= d <= k[1]
+ *     input[i, j, ..., l, m, n]                         ; otherwise
+ * }</pre>
+ * where `d = n - m`, `diag_index = k[1] - d`, and
+ * `index_in_diag = n - max(d, 0) + offset`.
+ * <p>
+ * `offset` is zero except when the alignment of the diagonal is to the right.
+ * <pre>{@code
+ * offset = max_diag_len - diag_len(d) ; if (`align` in {RIGHT_LEFT, RIGHT_RIGHT}
+ *                                            and `d >= 0`) or
+ *                                          (`align` in {LEFT_RIGHT, RIGHT_RIGHT}
+ *                                            and `d <= 0`)
+ *          0                          ; otherwise
+ * }</pre>
+ * where `diag_len(d) = min(cols - max(d, 0), rows + min(d, 0))`.
+ * <p>
+ * For example:
+ * <pre>{@code
+ * # The main diagonal.
+ * input = np.array([[[7, 7, 7, 7],              # Input shape: (2, 3, 4)
+ *                    [7, 7, 7, 7],
+ *                    [7, 7, 7, 7]],
+ *                   [[7, 7, 7, 7],
+ *                    [7, 7, 7, 7],
+ *                    [7, 7, 7, 7]]])
+ * diagonal = np.array([[1, 2, 3],               # Diagonal shape: (2, 3)
+ *                      [4, 5, 6]])
+ * tf.matrix_set_diag(input, diagonal)
+ *   ==> [[[1, 7, 7, 7],  # Output shape: (2, 3, 4)
+ *         [7, 2, 7, 7],
+ *         [7, 7, 3, 7]],
+ *        [[4, 7, 7, 7],
+ *         [7, 5, 7, 7],
+ *         [7, 7, 6, 7]]]
+ * 
+ * # A superdiagonal (per batch).
+ * tf.matrix_set_diag(input, diagonal, k = 1)
+ *   ==> [[[7, 1, 7, 7],  # Output shape: (2, 3, 4)
+ *         [7, 7, 2, 7],
+ *         [7, 7, 7, 3]],
+ *        [[7, 4, 7, 7],
+ *         [7, 7, 5, 7],
+ *         [7, 7, 7, 6]]]
+ * 
+ * # A band of diagonals.
+ * diagonals = np.array([[[0, 9, 1],  # Diagonal shape: (2, 4, 3)
+ *                        [6, 5, 8],
+ *                        [1, 2, 3],
+ *                        [4, 5, 0]],
+ *                       [[0, 1, 2],
+ *                        [5, 6, 4],
+ *                        [6, 1, 2],
+ *                        [3, 4, 0]]])
+ * tf.matrix_set_diag(input, diagonals, k = (-1, 2))
+ *   ==> [[[1, 6, 9, 7],  # Output shape: (2, 3, 4)
+ *         [4, 2, 5, 1],
+ *         [7, 5, 3, 8]],
+ *        [[6, 5, 1, 7],
+ *         [3, 1, 6, 2],
+ *         [7, 4, 2, 4]]]
+ * 
+ * # LEFT_RIGHT alignment.
+ * diagonals = np.array([[[9, 1, 0],  # Diagonal shape: (2, 4, 3)
+ *                        [6, 5, 8],
+ *                        [1, 2, 3],
+ *                        [0, 4, 5]],
+ *                       [[1, 2, 0],
+ *                        [5, 6, 4],
+ *                        [6, 1, 2],
+ *                        [0, 3, 4]]])
+ * tf.matrix_set_diag(input, diagonals, k = (-1, 2), align="LEFT_RIGHT")
+ *   ==> [[[1, 6, 9, 7],  # Output shape: (2, 3, 4)
+ *         [4, 2, 5, 1],
+ *         [7, 5, 3, 8]],
+ *        [[6, 5, 1, 7],
+ *         [3, 1, 6, 2],
+ *         [7, 4, 2, 4]]]
+ * 
+ * }</pre>
+ * 
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator(group = "linalg")
+public final class MatrixSetDiag<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.linalg.MatrixSetDiag}
+   */
+  public static class Options {
+    
+    /**
+     * @param align Some diagonals are shorter than `max_diag_len` and need to be padded. `align` is
+     * a string specifying how superdiagonals and subdiagonals should be aligned,
+     * respectively. There are four possible alignments: "RIGHT_LEFT" (default),
+     * "LEFT_RIGHT", "LEFT_LEFT", and "RIGHT_RIGHT". "RIGHT_LEFT" aligns superdiagonals
+     * to the right (left-pads the row) and subdiagonals to the left (right-pads the
+     * row). It is the packing format LAPACK uses. cuSPARSE uses "LEFT_RIGHT", which is
+     * the opposite alignment.
+     */
+    public Options align(String align) {
+      this.align = align;
+      return this;
+    }
+    
+    private String align;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new MatrixSetDiag operation.
+   * 
+   * @param scope current scope
+   * @param input Rank `r+1`, where `r >= 1`.
+   * @param diagonal Rank `r` when `k` is an integer or `k[0] == k[1]`. Otherwise, it has rank `r+1`.
+   * `k >= 1`.
+   * @param k Diagonal offset(s). Positive value means superdiagonal, 0 refers to the main
+   * diagonal, and negative value means subdiagonals. `k` can be a single integer
+   * (for a single diagonal) or a pair of integers specifying the low and high ends
+   * of a matrix band. `k[0]` must not be larger than `k[1]`.
+   * @param options carries optional attributes values
+   * @return a new instance of MatrixSetDiag
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> MatrixSetDiag<T> create(Scope scope, Operand<T> input, Operand<T> diagonal, Operand<TInt32> k, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("MatrixSetDiagV3", scope.makeOpName("MatrixSetDiag"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(diagonal.asOutput());
+    opBuilder.addInput(k.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.align != null) {
+          opBuilder.setAttr("align", opts.align);
+        }
+      }
+    }
+    return new MatrixSetDiag<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param align Some diagonals are shorter than `max_diag_len` and need to be padded. `align` is
+   * a string specifying how superdiagonals and subdiagonals should be aligned,
+   * respectively. There are four possible alignments: "RIGHT_LEFT" (default),
+   * "LEFT_RIGHT", "LEFT_LEFT", and "RIGHT_RIGHT". "RIGHT_LEFT" aligns superdiagonals
+   * to the right (left-pads the row) and subdiagonals to the left (right-pads the
+   * row). It is the packing format LAPACK uses. cuSPARSE uses "LEFT_RIGHT", which is
+   * the opposite alignment.
+   */
+  public static Options align(String align) {
+    return new Options().align(align);
+  }
+  
+  /**
+   * Rank `r+1`, with `output.shape = input.shape`.
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "MatrixSetDiagV3";
+  
+  private Output<T> output;
+  
+  private MatrixSetDiag(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/MatrixSolveLs.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/MatrixSolveLs.java
new file mode 100644
index 00000000000..633145ce24e
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/MatrixSolveLs.java
@@ -0,0 +1,153 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.linalg;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat64;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Solves one or more linear least-squares problems.
+ * <p>
+ * `matrix` is a tensor of shape `[..., M, N]` whose inner-most 2 dimensions
+ * form real or complex matrices of size `[M, N]`. `Rhs` is a tensor of the same
+ * type as `matrix` and shape `[..., M, K]`.
+ * The output is a tensor shape `[..., N, K]` where each output matrix solves
+ * each of the equations
+ * `matrix[..., :, :]` * `output[..., :, :]` = `rhs[..., :, :]`
+ * in the least squares sense.
+ * <p>
+ * We use the following notation for (complex) matrix and right-hand sides
+ * in the batch:
+ * <p>
+ * `matrix`=\\(A \in \mathbb{C}^{m \times n}\\),
+ * `rhs`=\\(B  \in \mathbb{C}^{m \times k}\\),
+ * `output`=\\(X  \in \mathbb{C}^{n \times k}\\),
+ * `l2_regularizer`=\\(\lambda \in \mathbb{R}\\).
+ * <p>
+ * If `fast` is `True`, then the solution is computed by solving the normal
+ * equations using Cholesky decomposition. Specifically, if \\(m \ge n\\) then
+ * \\(X = (A^H A + \lambda I)^{-1} A^H B\\), which solves the least-squares
+ * problem \\(X = \mathrm{argmin}_{Z \in \Re^{n \times k} } ||A Z - B||_F^2 + \lambda ||Z||_F^2\\).
+ * If \\(m \lt n\\) then `output` is computed as
+ * \\(X = A^H (A A^H + \lambda I)^{-1} B\\), which (for \\(\lambda = 0\\)) is the
+ * minimum-norm solution to the under-determined linear system, i.e.
+ * \\(X = \mathrm{argmin}_{Z \in \mathbb{C}^{n \times k} } ||Z||_F^2 \\),
+ * subject to \\(A Z = B\\). Notice that the fast path is only numerically stable
+ * when \\(A\\) is numerically full rank and has a condition number
+ * \\(\mathrm{cond}(A) \lt \frac{1}{\sqrt{\epsilon_{mach} } }\\) or \\(\lambda\\) is
+ * sufficiently large.
+ * <p>
+ * If `fast` is `False` an algorithm based on the numerically robust complete
+ * orthogonal decomposition is used. This computes the minimum-norm
+ * least-squares solution, even when \\(A\\) is rank deficient. This path is
+ * typically 6-7 times slower than the fast path. If `fast` is `False` then
+ * `l2_regularizer` is ignored.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator(group = "linalg")
+public final class MatrixSolveLs<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.linalg.MatrixSolveLs}
+   */
+  public static class Options {
+    
+    /**
+     * @param fast 
+     */
+    public Options fast(Boolean fast) {
+      this.fast = fast;
+      return this;
+    }
+    
+    private Boolean fast;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new MatrixSolveLs operation.
+   * 
+   * @param scope current scope
+   * @param matrix Shape is `[..., M, N]`.
+   * @param rhs Shape is `[..., M, K]`.
+   * @param l2Regularizer Scalar tensor.
+   * <p>
+   * @compatibility(numpy)
+   * Equivalent to np.linalg.lstsq
+   * @end_compatibility
+   * @param options carries optional attributes values
+   * @return a new instance of MatrixSolveLs
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> MatrixSolveLs<T> create(Scope scope, Operand<T> matrix, Operand<T> rhs, Operand<TFloat64> l2Regularizer, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("MatrixSolveLs", scope.makeOpName("MatrixSolveLs"));
+    opBuilder.addInput(matrix.asOutput());
+    opBuilder.addInput(rhs.asOutput());
+    opBuilder.addInput(l2Regularizer.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.fast != null) {
+          opBuilder.setAttr("fast", opts.fast);
+        }
+      }
+    }
+    return new MatrixSolveLs<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param fast 
+   */
+  public static Options fast(Boolean fast) {
+    return new Options().fast(fast);
+  }
+  
+  /**
+   * Shape is `[..., N, K]`.
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "MatrixSolveLs";
+  
+  private Output<T> output;
+  
+  private MatrixSolveLs(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Qr.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Qr.java
new file mode 100644
index 00000000000..cf58f41507f
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Qr.java
@@ -0,0 +1,130 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.linalg;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Computes the QR decompositions of one or more matrices.
+ * <p>
+ * Computes the QR decomposition of each inner matrix in `tensor` such that
+ * `tensor[..., :, :] = q[..., :, :] * r[..., :,:])`
+ * <pre>{@code
+ * # a is a tensor.
+ * # q is a tensor of orthonormal matrices.
+ * # r is a tensor of upper triangular matrices.
+ * q, r = qr(a)
+ * q_full, r_full = qr(a, full_matrices=True)
+ * }</pre>
+ * 
+ * 
+ * @param <T> data type for {@code q()} output
+ */
+@Operator(group = "linalg")
+public final class Qr<T extends TType> extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.linalg.Qr}
+   */
+  public static class Options {
+    
+    /**
+     * @param fullMatrices If true, compute full-sized `q` and `r`. If false
+     * (the default), compute only the leading `P` columns of `q`.
+     */
+    public Options fullMatrices(Boolean fullMatrices) {
+      this.fullMatrices = fullMatrices;
+      return this;
+    }
+    
+    private Boolean fullMatrices;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new Qr operation.
+   * 
+   * @param scope current scope
+   * @param input A tensor of shape `[..., M, N]` whose inner-most 2 dimensions
+   * form matrices of size `[M, N]`. Let `P` be the minimum of `M` and `N`.
+   * @param options carries optional attributes values
+   * @return a new instance of Qr
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> Qr<T> create(Scope scope, Operand<T> input, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Qr", scope.makeOpName("Qr"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.fullMatrices != null) {
+          opBuilder.setAttr("full_matrices", opts.fullMatrices);
+        }
+      }
+    }
+    return new Qr<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param fullMatrices If true, compute full-sized `q` and `r`. If false
+   * (the default), compute only the leading `P` columns of `q`.
+   */
+  public static Options fullMatrices(Boolean fullMatrices) {
+    return new Options().fullMatrices(fullMatrices);
+  }
+  
+  /**
+   * Orthonormal basis for range of `a`. If `full_matrices` is `False` then
+   * shape is `[..., M, P]`; if `full_matrices` is `True` then shape is
+   * `[..., M, M]`.
+   */
+  public Output<T> q() {
+    return q;
+  }
+  
+  /**
+   * Triangular factor. If `full_matrices` is `False` then shape is
+   * `[..., P, N]`. If `full_matrices` is `True` then shape is `[..., M, N]`.
+   */
+  public Output<T> r() {
+    return r;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Qr";
+  
+  private Output<T> q;
+  private Output<T> r;
+  
+  private Qr(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    q = operation.output(outputIdx++);
+    r = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/QuantizedMatMul.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/QuantizedMatMul.java
new file mode 100644
index 00000000000..0fc70a09b7d
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/QuantizedMatMul.java
@@ -0,0 +1,161 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.linalg;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Perform a quantized matrix multiplication of  `a` by the matrix `b`.
+ * <p>
+ * The inputs must be two-dimensional matrices and the inner dimension of
+ * `a` (after being transposed if `transpose_a` is non-zero) must match the
+ * outer dimension of `b` (after being transposed if `transposed_b` is
+ * non-zero).
+ * 
+ * @param <V> data type for {@code out()} output
+ */
+@Operator(group = "linalg")
+public final class QuantizedMatMul<V extends TType> extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.linalg.QuantizedMatMul}
+   */
+  public static class Options {
+    
+    /**
+     * @param transposeA If true, `a` is transposed before multiplication.
+     */
+    public Options transposeA(Boolean transposeA) {
+      this.transposeA = transposeA;
+      return this;
+    }
+    
+    /**
+     * @param transposeB If true, `b` is transposed before multiplication.
+     */
+    public Options transposeB(Boolean transposeB) {
+      this.transposeB = transposeB;
+      return this;
+    }
+    
+    private Boolean transposeA;
+    private Boolean transposeB;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new QuantizedMatMul operation.
+   * 
+   * @param scope current scope
+   * @param a Must be a two-dimensional tensor.
+   * @param b Must be a two-dimensional tensor.
+   * @param minA The float value that the lowest quantized `a` value represents.
+   * @param maxA The float value that the highest quantized `a` value represents.
+   * @param minB The float value that the lowest quantized `b` value represents.
+   * @param maxB The float value that the highest quantized `b` value represents.
+   * @param Toutput 
+   * @param Tactivation The type of output produced by activation function
+   * following this operation.
+   * @param options carries optional attributes values
+   * @return a new instance of QuantizedMatMul
+   */
+  @Endpoint(describeByClass = true)
+  public static <V extends TType, T extends TType, U extends TType, W extends TType> QuantizedMatMul<V> create(Scope scope, Operand<T> a, Operand<U> b, Operand<TFloat32> minA, Operand<TFloat32> maxA, Operand<TFloat32> minB, Operand<TFloat32> maxB, Class<V> Toutput, Class<W> Tactivation, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("QuantizedMatMul", scope.makeOpName("QuantizedMatMul"));
+    opBuilder.addInput(a.asOutput());
+    opBuilder.addInput(b.asOutput());
+    opBuilder.addInput(minA.asOutput());
+    opBuilder.addInput(maxA.asOutput());
+    opBuilder.addInput(minB.asOutput());
+    opBuilder.addInput(maxB.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("Toutput", Toutput);
+    opBuilder.setAttr("Tactivation", Tactivation);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.transposeA != null) {
+          opBuilder.setAttr("transpose_a", opts.transposeA);
+        }
+        if (opts.transposeB != null) {
+          opBuilder.setAttr("transpose_b", opts.transposeB);
+        }
+      }
+    }
+    return new QuantizedMatMul<V>(opBuilder.build());
+  }
+  
+  /**
+   * @param transposeA If true, `a` is transposed before multiplication.
+   */
+  public static Options transposeA(Boolean transposeA) {
+    return new Options().transposeA(transposeA);
+  }
+  
+  /**
+   * @param transposeB If true, `b` is transposed before multiplication.
+   */
+  public static Options transposeB(Boolean transposeB) {
+    return new Options().transposeB(transposeB);
+  }
+  
+  /**
+   */
+  public Output<V> out() {
+    return out;
+  }
+  
+  /**
+   * The float value that the lowest quantized output value represents.
+   */
+  public Output<TFloat32> minOut() {
+    return minOut;
+  }
+  
+  /**
+   * The float value that the highest quantized output value represents.
+   */
+  public Output<TFloat32> maxOut() {
+    return maxOut;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "QuantizedMatMul";
+  
+  private Output<V> out;
+  private Output<TFloat32> minOut;
+  private Output<TFloat32> maxOut;
+  
+  private QuantizedMatMul(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    out = operation.output(outputIdx++);
+    minOut = operation.output(outputIdx++);
+    maxOut = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/QuantizedMatMulWithBias.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/QuantizedMatMulWithBias.java
new file mode 100644
index 00000000000..4f083e85932
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/QuantizedMatMulWithBias.java
@@ -0,0 +1,181 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.linalg;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Performs a quantized matrix multiplication of `a` by the matrix `b` with bias
+ * add.
+ * <p>
+ * The inputs must be two-dimensional matrices and 1D bias vector. And the inner
+ * dimension of `a` (after being transposed if `transpose_a` is non-zero) must
+ * match the outer dimension of `b` (after being transposed if `transposed_b` is
+ * non-zero). Then do broadcast add operation with bias values on the matrix
+ * multiplication result. The bias size must match inner dimension of `b`.
+ * 
+ * @param <W> data type for {@code out()} output
+ */
+public final class QuantizedMatMulWithBias<W extends TType> extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.linalg.QuantizedMatMulWithBias}
+   */
+  public static class Options {
+    
+    /**
+     * @param transposeA If true, `a` is transposed before multiplication.
+     */
+    public Options transposeA(Boolean transposeA) {
+      this.transposeA = transposeA;
+      return this;
+    }
+    
+    /**
+     * @param transposeB If true, `b` is transposed before multiplication.
+     */
+    public Options transposeB(Boolean transposeB) {
+      this.transposeB = transposeB;
+      return this;
+    }
+    
+    /**
+     * @param inputQuantMode Input data quantization mode. Either MIN_FIRST(default) or SCALED.
+     */
+    public Options inputQuantMode(String inputQuantMode) {
+      this.inputQuantMode = inputQuantMode;
+      return this;
+    }
+    
+    private Boolean transposeA;
+    private Boolean transposeB;
+    private String inputQuantMode;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new QuantizedMatMulWithBias operation.
+   * 
+   * @param scope current scope
+   * @param a A matrix to be multiplied. Must be a two-dimensional tensor of type `quint8`.
+   * @param b A matrix to be multiplied and must be a two-dimensional tensor of type `qint8`.
+   * @param bias A 1D bias tensor with size matching inner dimension of `b` (after being
+   * transposed if `transposed_b` is non-zero).
+   * @param minA The float value that the lowest quantized `a` value represents.
+   * @param maxA The float value that the highest quantized `a` value represents.
+   * @param minB The float value that the lowest quantized `b` value represents.
+   * @param maxB The float value that the highest quantized `b` value represents.
+   * @param Toutput 
+   * @param options carries optional attributes values
+   * @return a new instance of QuantizedMatMulWithBias
+   */
+  @Endpoint(describeByClass = true)
+  public static <W extends TType, T extends TType, U extends TType, V extends TType> QuantizedMatMulWithBias<W> create(Scope scope, Operand<T> a, Operand<U> b, Operand<V> bias, Operand<TFloat32> minA, Operand<TFloat32> maxA, Operand<TFloat32> minB, Operand<TFloat32> maxB, Class<W> Toutput, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("QuantizedMatMulWithBias", scope.makeOpName("QuantizedMatMulWithBias"));
+    opBuilder.addInput(a.asOutput());
+    opBuilder.addInput(b.asOutput());
+    opBuilder.addInput(bias.asOutput());
+    opBuilder.addInput(minA.asOutput());
+    opBuilder.addInput(maxA.asOutput());
+    opBuilder.addInput(minB.asOutput());
+    opBuilder.addInput(maxB.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("Toutput", Toutput);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.transposeA != null) {
+          opBuilder.setAttr("transpose_a", opts.transposeA);
+        }
+        if (opts.transposeB != null) {
+          opBuilder.setAttr("transpose_b", opts.transposeB);
+        }
+        if (opts.inputQuantMode != null) {
+          opBuilder.setAttr("input_quant_mode", opts.inputQuantMode);
+        }
+      }
+    }
+    return new QuantizedMatMulWithBias<W>(opBuilder.build());
+  }
+  
+  /**
+   * @param transposeA If true, `a` is transposed before multiplication.
+   */
+  public static Options transposeA(Boolean transposeA) {
+    return new Options().transposeA(transposeA);
+  }
+  
+  /**
+   * @param transposeB If true, `b` is transposed before multiplication.
+   */
+  public static Options transposeB(Boolean transposeB) {
+    return new Options().transposeB(transposeB);
+  }
+  
+  /**
+   * @param inputQuantMode Input data quantization mode. Either MIN_FIRST(default) or SCALED.
+   */
+  public static Options inputQuantMode(String inputQuantMode) {
+    return new Options().inputQuantMode(inputQuantMode);
+  }
+  
+  /**
+   */
+  public Output<W> out() {
+    return out;
+  }
+  
+  /**
+   * The float value that the lowest quantized output value represents.
+   */
+  public Output<TFloat32> minOut() {
+    return minOut;
+  }
+  
+  /**
+   * The float value that the highest quantized output value represents.
+   */
+  public Output<TFloat32> maxOut() {
+    return maxOut;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "QuantizedMatMulWithBias";
+  
+  private Output<W> out;
+  private Output<TFloat32> minOut;
+  private Output<TFloat32> maxOut;
+  
+  private QuantizedMatMulWithBias(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    out = operation.output(outputIdx++);
+    minOut = operation.output(outputIdx++);
+    maxOut = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/QuantizedMatMulWithBiasAndRelu.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/QuantizedMatMulWithBiasAndRelu.java
new file mode 100644
index 00000000000..c5769d336d9
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/QuantizedMatMulWithBiasAndRelu.java
@@ -0,0 +1,182 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.linalg;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Perform a quantized matrix multiplication of  `a` by the matrix `b` with bias
+ * add and relu fusion.
+ * <p>
+ * The inputs must be two-dimensional matrices and 1D bias vector. And the inner
+ * dimension of `a` (after being transposed if `transpose_a` is non-zero) must
+ * match the outer dimension of `b` (after being transposed if `transposed_b` is
+ * non-zero). Then do broadcast add operation with bias values on the matrix
+ * multiplication result. The bias size must match inner dimension of `b`. Then do
+ * relu activation to get non-negative result.
+ * 
+ * @param <V> data type for {@code out()} output
+ */
+public final class QuantizedMatMulWithBiasAndRelu<V extends TType> extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.linalg.QuantizedMatMulWithBiasAndRelu}
+   */
+  public static class Options {
+    
+    /**
+     * @param transposeA If true, `a` is transposed before multiplication.
+     */
+    public Options transposeA(Boolean transposeA) {
+      this.transposeA = transposeA;
+      return this;
+    }
+    
+    /**
+     * @param transposeB If true, `b` is transposed before multiplication.
+     */
+    public Options transposeB(Boolean transposeB) {
+      this.transposeB = transposeB;
+      return this;
+    }
+    
+    /**
+     * @param inputQuantMode Input data quantization mode. Either MIN_FIRST(default) or SCALED.
+     */
+    public Options inputQuantMode(String inputQuantMode) {
+      this.inputQuantMode = inputQuantMode;
+      return this;
+    }
+    
+    private Boolean transposeA;
+    private Boolean transposeB;
+    private String inputQuantMode;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new QuantizedMatMulWithBiasAndRelu operation.
+   * 
+   * @param scope current scope
+   * @param a A matrix to be multiplied. Must be a two-dimensional tensor of type `quint8`.
+   * @param b A matrix to be multiplied and must be a two-dimensional tensor of type `qint8`.
+   * @param bias A 1D bias tensor with size matching with inner dimension of `b` (after being
+   * transposed if `transposed_b` is non-zero).
+   * @param minA The float value that the lowest quantized `a` value represents.
+   * @param maxA The float value that the highest quantized `a` value represents.
+   * @param minB The float value that the lowest quantized `b` value represents.
+   * @param maxB The float value that the highest quantized `b` value represents.
+   * @param Toutput 
+   * @param options carries optional attributes values
+   * @return a new instance of QuantizedMatMulWithBiasAndRelu
+   */
+  @Endpoint(describeByClass = true)
+  public static <V extends TType, T extends TType, U extends TType> QuantizedMatMulWithBiasAndRelu<V> create(Scope scope, Operand<T> a, Operand<U> b, Operand<TFloat32> bias, Operand<TFloat32> minA, Operand<TFloat32> maxA, Operand<TFloat32> minB, Operand<TFloat32> maxB, Class<V> Toutput, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("QuantizedMatMulWithBiasAndRelu", scope.makeOpName("QuantizedMatMulWithBiasAndRelu"));
+    opBuilder.addInput(a.asOutput());
+    opBuilder.addInput(b.asOutput());
+    opBuilder.addInput(bias.asOutput());
+    opBuilder.addInput(minA.asOutput());
+    opBuilder.addInput(maxA.asOutput());
+    opBuilder.addInput(minB.asOutput());
+    opBuilder.addInput(maxB.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("Toutput", Toutput);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.transposeA != null) {
+          opBuilder.setAttr("transpose_a", opts.transposeA);
+        }
+        if (opts.transposeB != null) {
+          opBuilder.setAttr("transpose_b", opts.transposeB);
+        }
+        if (opts.inputQuantMode != null) {
+          opBuilder.setAttr("input_quant_mode", opts.inputQuantMode);
+        }
+      }
+    }
+    return new QuantizedMatMulWithBiasAndRelu<V>(opBuilder.build());
+  }
+  
+  /**
+   * @param transposeA If true, `a` is transposed before multiplication.
+   */
+  public static Options transposeA(Boolean transposeA) {
+    return new Options().transposeA(transposeA);
+  }
+  
+  /**
+   * @param transposeB If true, `b` is transposed before multiplication.
+   */
+  public static Options transposeB(Boolean transposeB) {
+    return new Options().transposeB(transposeB);
+  }
+  
+  /**
+   * @param inputQuantMode Input data quantization mode. Either MIN_FIRST(default) or SCALED.
+   */
+  public static Options inputQuantMode(String inputQuantMode) {
+    return new Options().inputQuantMode(inputQuantMode);
+  }
+  
+  /**
+   */
+  public Output<V> out() {
+    return out;
+  }
+  
+  /**
+   * The float value that the lowest quantized output value represents.
+   */
+  public Output<TFloat32> minOut() {
+    return minOut;
+  }
+  
+  /**
+   * The float value that the highest quantized output value represents.
+   */
+  public Output<TFloat32> maxOut() {
+    return maxOut;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "QuantizedMatMulWithBiasAndRelu";
+  
+  private Output<V> out;
+  private Output<TFloat32> minOut;
+  private Output<TFloat32> maxOut;
+  
+  private QuantizedMatMulWithBiasAndRelu(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    out = operation.output(outputIdx++);
+    minOut = operation.output(outputIdx++);
+    maxOut = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/QuantizedMatMulWithBiasAndReluAndRequantize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/QuantizedMatMulWithBiasAndReluAndRequantize.java
new file mode 100644
index 00000000000..a580c953e74
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/QuantizedMatMulWithBiasAndReluAndRequantize.java
@@ -0,0 +1,187 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.linalg;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Perform a quantized matrix multiplication of  `a` by the matrix `b` with bias
+ * add and relu and requantize fusion.
+ * <p>
+ * The inputs must be two-dimensional matrices and 1D bias vector. And the inner
+ * dimension of `a` (after being transposed if `transpose_a` is non-zero) must
+ * match the outer dimension of `b` (after being transposed if `transposed_b` is
+ * non-zero). Then do broadcast add operation with bias values on the matrix
+ * multiplication result. The bias size must match inner dimension of `b`.  Then do
+ * relu activation to get non-negative result. Then do requantize operation to get
+ * final uint8 result.
+ * 
+ * @param <W> data type for {@code out()} output
+ */
+public final class QuantizedMatMulWithBiasAndReluAndRequantize<W extends TType> extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.linalg.QuantizedMatMulWithBiasAndReluAndRequantize}
+   */
+  public static class Options {
+    
+    /**
+     * @param transposeA If true, `a` is transposed before multiplication.
+     */
+    public Options transposeA(Boolean transposeA) {
+      this.transposeA = transposeA;
+      return this;
+    }
+    
+    /**
+     * @param transposeB If true, `b` is transposed before multiplication.
+     */
+    public Options transposeB(Boolean transposeB) {
+      this.transposeB = transposeB;
+      return this;
+    }
+    
+    /**
+     * @param inputQuantMode Input data quantization mode. Either MIN_FIRST(default) or SCALED.
+     */
+    public Options inputQuantMode(String inputQuantMode) {
+      this.inputQuantMode = inputQuantMode;
+      return this;
+    }
+    
+    private Boolean transposeA;
+    private Boolean transposeB;
+    private String inputQuantMode;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new QuantizedMatMulWithBiasAndReluAndRequantize operation.
+   * 
+   * @param scope current scope
+   * @param a A matrix to be multiplied. Must be a two-dimensional tensor of type `quint8`.
+   * @param b A matrix to be multiplied and must be a two-dimensional tensor of type `qint8`.
+   * @param bias A 1D bias tensor with size matching with inner dimension of `b` (after being
+   * transposed if `transposed_b` is non-zero).
+   * @param minA The float value that the lowest quantized `a` value represents.
+   * @param maxA The float value that the highest quantized `a` value represents.
+   * @param minB The float value that the lowest quantized `b` value represents.
+   * @param maxB The float value that the highest quantized `b` value represents.
+   * @param minFreezedOutput The float value that the highest quantized output value after requantize.
+   * @param maxFreezedOutput 
+   * @param Toutput 
+   * @param options carries optional attributes values
+   * @return a new instance of QuantizedMatMulWithBiasAndReluAndRequantize
+   */
+  @Endpoint(describeByClass = true)
+  public static <W extends TType, T extends TType, U extends TType, V extends TType> QuantizedMatMulWithBiasAndReluAndRequantize<W> create(Scope scope, Operand<T> a, Operand<U> b, Operand<V> bias, Operand<TFloat32> minA, Operand<TFloat32> maxA, Operand<TFloat32> minB, Operand<TFloat32> maxB, Operand<TFloat32> minFreezedOutput, Operand<TFloat32> maxFreezedOutput, Class<W> Toutput, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("QuantizedMatMulWithBiasAndReluAndRequantize", scope.makeOpName("QuantizedMatMulWithBiasAndReluAndRequantize"));
+    opBuilder.addInput(a.asOutput());
+    opBuilder.addInput(b.asOutput());
+    opBuilder.addInput(bias.asOutput());
+    opBuilder.addInput(minA.asOutput());
+    opBuilder.addInput(maxA.asOutput());
+    opBuilder.addInput(minB.asOutput());
+    opBuilder.addInput(maxB.asOutput());
+    opBuilder.addInput(minFreezedOutput.asOutput());
+    opBuilder.addInput(maxFreezedOutput.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("Toutput", Toutput);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.transposeA != null) {
+          opBuilder.setAttr("transpose_a", opts.transposeA);
+        }
+        if (opts.transposeB != null) {
+          opBuilder.setAttr("transpose_b", opts.transposeB);
+        }
+        if (opts.inputQuantMode != null) {
+          opBuilder.setAttr("input_quant_mode", opts.inputQuantMode);
+        }
+      }
+    }
+    return new QuantizedMatMulWithBiasAndReluAndRequantize<W>(opBuilder.build());
+  }
+  
+  /**
+   * @param transposeA If true, `a` is transposed before multiplication.
+   */
+  public static Options transposeA(Boolean transposeA) {
+    return new Options().transposeA(transposeA);
+  }
+  
+  /**
+   * @param transposeB If true, `b` is transposed before multiplication.
+   */
+  public static Options transposeB(Boolean transposeB) {
+    return new Options().transposeB(transposeB);
+  }
+  
+  /**
+   * @param inputQuantMode Input data quantization mode. Either MIN_FIRST(default) or SCALED.
+   */
+  public static Options inputQuantMode(String inputQuantMode) {
+    return new Options().inputQuantMode(inputQuantMode);
+  }
+  
+  /**
+   */
+  public Output<W> out() {
+    return out;
+  }
+  
+  /**
+   * The float value that the lowest quantized output value represents.
+   */
+  public Output<TFloat32> minOut() {
+    return minOut;
+  }
+  
+  /**
+   * The float value that the highest quantized output value represents.
+   */
+  public Output<TFloat32> maxOut() {
+    return maxOut;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "QuantizedMatMulWithBiasAndReluAndRequantize";
+  
+  private Output<W> out;
+  private Output<TFloat32> minOut;
+  private Output<TFloat32> maxOut;
+  
+  private QuantizedMatMulWithBiasAndReluAndRequantize(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    out = operation.output(outputIdx++);
+    minOut = operation.output(outputIdx++);
+    maxOut = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/SelfAdjointEig.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/SelfAdjointEig.java
new file mode 100644
index 00000000000..60bbafe6e70
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/SelfAdjointEig.java
@@ -0,0 +1,127 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.linalg;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Computes the eigen decomposition of one or more square self-adjoint matrices.
+ * <p>
+ * Computes the eigenvalues and (optionally) eigenvectors of each inner matrix in
+ * `input` such that `input[..., :, :] = v[..., :, :] * diag(e[..., :])`. The eigenvalues
+ * are sorted in non-decreasing order.
+ * <pre>{@code
+ * # a is a tensor.
+ * # e is a tensor of eigenvalues.
+ * # v is a tensor of eigenvectors.
+ * e, v = self_adjoint_eig(a)
+ * e = self_adjoint_eig(a, compute_v=False)
+ * }</pre>
+ * 
+ * 
+ * @param <T> data type for {@code e()} output
+ */
+@Operator(group = "linalg")
+public final class SelfAdjointEig<T extends TType> extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.linalg.SelfAdjointEig}
+   */
+  public static class Options {
+    
+    /**
+     * @param computeV If `True` then eigenvectors will be computed and returned in `v`.
+     * Otherwise, only the eigenvalues will be computed.
+     */
+    public Options computeV(Boolean computeV) {
+      this.computeV = computeV;
+      return this;
+    }
+    
+    private Boolean computeV;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new SelfAdjointEig operation.
+   * 
+   * @param scope current scope
+   * @param input `Tensor` input of shape `[N, N]`.
+   * @param options carries optional attributes values
+   * @return a new instance of SelfAdjointEig
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> SelfAdjointEig<T> create(Scope scope, Operand<T> input, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("SelfAdjointEigV2", scope.makeOpName("SelfAdjointEig"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.computeV != null) {
+          opBuilder.setAttr("compute_v", opts.computeV);
+        }
+      }
+    }
+    return new SelfAdjointEig<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param computeV If `True` then eigenvectors will be computed and returned in `v`.
+   * Otherwise, only the eigenvalues will be computed.
+   */
+  public static Options computeV(Boolean computeV) {
+    return new Options().computeV(computeV);
+  }
+  
+  /**
+   * Eigenvalues. Shape is `[N]`.
+   */
+  public Output<T> e() {
+    return e;
+  }
+  
+  /**
+   * Eigenvectors. Shape is `[N, N]`.
+   */
+  public Output<T> v() {
+    return v;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "SelfAdjointEigV2";
+  
+  private Output<T> e;
+  private Output<T> v;
+  
+  private SelfAdjointEig(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    e = operation.output(outputIdx++);
+    v = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Solve.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Solve.java
new file mode 100644
index 00000000000..62f1beb11c8
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Solve.java
@@ -0,0 +1,120 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.linalg;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Solves systems of linear equations.
+ * <p>
+ * `Matrix` is a tensor of shape `[..., M, M]` whose inner-most 2 dimensions
+ * form square matrices. `Rhs` is a tensor of shape `[..., M, K]`. The `output` is
+ * a tensor shape `[..., M, K]`.  If `adjoint` is `False` then each output matrix
+ * satisfies `matrix[..., :, :] * output[..., :, :] = rhs[..., :, :]`.
+ * If `adjoint` is `True` then each output matrix satisfies
+ * `adjoint(matrix[..., :, :]) * output[..., :, :] = rhs[..., :, :]`.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator(group = "linalg")
+public final class Solve<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.linalg.Solve}
+   */
+  public static class Options {
+    
+    /**
+     * @param adjoint Boolean indicating whether to solve with `matrix` or its (block-wise)
+     * adjoint.
+     */
+    public Options adjoint(Boolean adjoint) {
+      this.adjoint = adjoint;
+      return this;
+    }
+    
+    private Boolean adjoint;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new Solve operation.
+   * 
+   * @param scope current scope
+   * @param matrix Shape is `[..., M, M]`.
+   * @param rhs Shape is `[..., M, K]`.
+   * @param options carries optional attributes values
+   * @return a new instance of Solve
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> Solve<T> create(Scope scope, Operand<T> matrix, Operand<T> rhs, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("MatrixSolve", scope.makeOpName("Solve"));
+    opBuilder.addInput(matrix.asOutput());
+    opBuilder.addInput(rhs.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.adjoint != null) {
+          opBuilder.setAttr("adjoint", opts.adjoint);
+        }
+      }
+    }
+    return new Solve<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param adjoint Boolean indicating whether to solve with `matrix` or its (block-wise)
+   * adjoint.
+   */
+  public static Options adjoint(Boolean adjoint) {
+    return new Options().adjoint(adjoint);
+  }
+  
+  /**
+   * Shape is `[..., M, K]`.
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "MatrixSolve";
+  
+  private Output<T> output;
+  
+  private Solve(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Sqrtm.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Sqrtm.java
new file mode 100644
index 00000000000..35c3b88605c
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Sqrtm.java
@@ -0,0 +1,95 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.linalg;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Computes the matrix square root of one or more square matrices:
+ * <p>
+ * matmul(sqrtm(A), sqrtm(A)) = A
+ * <p>
+ * The input matrix should be invertible. If the input matrix is real, it should
+ * have no eigenvalues which are real and negative (pairs of complex conjugate
+ * eigenvalues are allowed).
+ * <p>
+ * The matrix square root is computed by first reducing the matrix to
+ * quasi-triangular form with the real Schur decomposition. The square root
+ * of the quasi-triangular matrix is then computed directly. Details of
+ * the algorithm can be found in: Nicholas J. Higham, "Computing real
+ * square roots of a real matrix", Linear Algebra Appl., 1987.
+ * <p>
+ * The input is a tensor of shape `[..., M, M]` whose inner-most 2 dimensions
+ * form square matrices. The output is a tensor of the same shape as the input
+ * containing the matrix square root for all input submatrices `[..., :, :]`.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator(group = "linalg")
+public final class Sqrtm<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new Sqrtm operation.
+   * 
+   * @param scope current scope
+   * @param input Shape is `[..., M, M]`.
+   * @return a new instance of Sqrtm
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> Sqrtm<T> create(Scope scope, Operand<T> input) {
+    OperationBuilder opBuilder = scope.env().opBuilder("MatrixSquareRoot", scope.makeOpName("Sqrtm"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new Sqrtm<T>(opBuilder.build());
+  }
+  
+  /**
+   * Shape is `[..., M, M]`.
+   * <p>
+   * @compatibility(scipy)
+   * Equivalent to scipy.linalg.sqrtm
+   * @end_compatibility
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "MatrixSquareRoot";
+  
+  private Output<T> output;
+  
+  private Sqrtm(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Svd.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Svd.java
new file mode 100644
index 00000000000..4ff72f0e139
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Svd.java
@@ -0,0 +1,166 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.linalg;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Computes the singular value decompositions of one or more matrices.
+ * <p>
+ * Computes the SVD of each inner matrix in `input` such that
+ * `input[..., :, :] = u[..., :, :] * diag(s[..., :, :]) * transpose(v[..., :, :])`
+ * <pre>{@code
+ * # a is a tensor containing a batch of matrices.
+ * # s is a tensor of singular values for each matrix.
+ * # u is the tensor containing the left singular vectors for each matrix.
+ * # v is the tensor containing the right singular vectors for each matrix.
+ * s, u, v = svd(a)
+ * s, _, _ = svd(a, compute_uv=False)
+ * }</pre>
+ * 
+ * 
+ * @param <T> data type for {@code s()} output
+ */
+@Operator(group = "linalg")
+public final class Svd<T extends TType> extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.linalg.Svd}
+   */
+  public static class Options {
+    
+    /**
+     * @param computeUv If true, left and right singular vectors will be
+     * computed and returned in `u` and `v`, respectively.
+     * If false, `u` and `v` are not set and should never referenced.
+     */
+    public Options computeUv(Boolean computeUv) {
+      this.computeUv = computeUv;
+      return this;
+    }
+    
+    /**
+     * @param fullMatrices If true, compute full-sized `u` and `v`. If false
+     * (the default), compute only the leading `P` singular vectors.
+     * Ignored if `compute_uv` is `False`.
+     */
+    public Options fullMatrices(Boolean fullMatrices) {
+      this.fullMatrices = fullMatrices;
+      return this;
+    }
+    
+    private Boolean computeUv;
+    private Boolean fullMatrices;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new Svd operation.
+   * 
+   * @param scope current scope
+   * @param input A tensor of shape `[..., M, N]` whose inner-most 2 dimensions
+   * form matrices of size `[M, N]`. Let `P` be the minimum of `M` and `N`.
+   * @param options carries optional attributes values
+   * @return a new instance of Svd
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> Svd<T> create(Scope scope, Operand<T> input, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Svd", scope.makeOpName("Svd"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.computeUv != null) {
+          opBuilder.setAttr("compute_uv", opts.computeUv);
+        }
+        if (opts.fullMatrices != null) {
+          opBuilder.setAttr("full_matrices", opts.fullMatrices);
+        }
+      }
+    }
+    return new Svd<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param computeUv If true, left and right singular vectors will be
+   * computed and returned in `u` and `v`, respectively.
+   * If false, `u` and `v` are not set and should never referenced.
+   */
+  public static Options computeUv(Boolean computeUv) {
+    return new Options().computeUv(computeUv);
+  }
+  
+  /**
+   * @param fullMatrices If true, compute full-sized `u` and `v`. If false
+   * (the default), compute only the leading `P` singular vectors.
+   * Ignored if `compute_uv` is `False`.
+   */
+  public static Options fullMatrices(Boolean fullMatrices) {
+    return new Options().fullMatrices(fullMatrices);
+  }
+  
+  /**
+   * Singular values. Shape is `[..., P]`.
+   */
+  public Output<T> s() {
+    return s;
+  }
+  
+  /**
+   * Left singular vectors. If `full_matrices` is `False` then shape is
+   * `[..., M, P]`; if `full_matrices` is `True` then shape is
+   * `[..., M, M]`. Undefined if `compute_uv` is `False`.
+   */
+  public Output<T> u() {
+    return u;
+  }
+  
+  /**
+   * Left singular vectors. If `full_matrices` is `False` then shape is
+   * `[..., N, P]`. If `full_matrices` is `True` then shape is `[..., N, N]`.
+   * Undefined if `compute_uv` is false.
+   */
+  public Output<T> v() {
+    return v;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Svd";
+  
+  private Output<T> s;
+  private Output<T> u;
+  private Output<T> v;
+  
+  private Svd(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    s = operation.output(outputIdx++);
+    u = operation.output(outputIdx++);
+    v = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/TensorDiag.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/TensorDiag.java
new file mode 100644
index 00000000000..126a8c5cde7
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/TensorDiag.java
@@ -0,0 +1,92 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.linalg;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Returns a diagonal tensor with a given diagonal values.
+ * <p>
+ * Given a `diagonal`, this operation returns a tensor with the `diagonal` and
+ * everything else padded with zeros. The diagonal is computed as follows:
+ * <p>
+ * Assume `diagonal` has dimensions [D1,..., Dk], then the output is a tensor of
+ * rank 2k with dimensions [D1,..., Dk, D1,..., Dk] where:
+ * <p>
+ * `output[i1,..., ik, i1,..., ik] = diagonal[i1, ..., ik]` and 0 everywhere else.
+ * <p>
+ * For example:
+ * <pre>{@code
+ * # 'diagonal' is [1, 2, 3, 4]
+ * tf.diag(diagonal) ==> [[1, 0, 0, 0]
+ *                        [0, 2, 0, 0]
+ *                        [0, 0, 3, 0]
+ *                        [0, 0, 0, 4]]
+ * }</pre>
+ * 
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator(group = "linalg")
+public final class TensorDiag<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new TensorDiag operation.
+   * 
+   * @param scope current scope
+   * @param diagonal Rank k tensor where k is at most 1.
+   * @return a new instance of TensorDiag
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> TensorDiag<T> create(Scope scope, Operand<T> diagonal) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Diag", scope.makeOpName("TensorDiag"));
+    opBuilder.addInput(diagonal.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new TensorDiag<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Diag";
+  
+  private Output<T> output;
+  
+  private TensorDiag(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/TensorDiagPart.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/TensorDiagPart.java
new file mode 100644
index 00000000000..e999ca96a4b
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/TensorDiagPart.java
@@ -0,0 +1,94 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.linalg;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Returns the diagonal part of the tensor.
+ * <p>
+ * This operation returns a tensor with the `diagonal` part
+ * of the `input`. The `diagonal` part is computed as follows:
+ * <p>
+ * Assume `input` has dimensions `[D1,..., Dk, D1,..., Dk]`, then the output is a
+ * tensor of rank `k` with dimensions `[D1,..., Dk]` where:
+ * <p>
+ * `diagonal[i1,..., ik] = input[i1, ..., ik, i1,..., ik]`.
+ * <p>
+ * For example:
+ * <pre>{@code
+ * # 'input' is [[1, 0, 0, 0]
+ *               [0, 2, 0, 0]
+ *               [0, 0, 3, 0]
+ *               [0, 0, 0, 4]]
+ * 
+ * tf.diag_part(input) ==> [1, 2, 3, 4]
+ * }</pre>
+ * 
+ * 
+ * @param <T> data type for {@code diagonal()} output
+ */
+@Operator(group = "linalg")
+public final class TensorDiagPart<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new TensorDiagPart operation.
+   * 
+   * @param scope current scope
+   * @param input Rank k tensor where k is even and not zero.
+   * @return a new instance of TensorDiagPart
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> TensorDiagPart<T> create(Scope scope, Operand<T> input) {
+    OperationBuilder opBuilder = scope.env().opBuilder("DiagPart", scope.makeOpName("TensorDiagPart"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new TensorDiagPart<T>(opBuilder.build());
+  }
+  
+  /**
+   * The extracted diagonal.
+   */
+  public Output<T> diagonal() {
+    return diagonal;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return diagonal;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "DiagPart";
+  
+  private Output<T> diagonal;
+  
+  private TensorDiagPart(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    diagonal = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Transpose.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Transpose.java
new file mode 100644
index 00000000000..1df98fabe64
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/Transpose.java
@@ -0,0 +1,80 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.linalg;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Shuffle dimensions of x according to a permutation.
+ * <p>
+ * The output `y` has the same rank as `x`. The shapes of `x` and `y` satisfy:
+ *   `y.shape[i] == x.shape[perm[i]] for i in [0, 1, ..., rank(x) - 1]`
+ * 
+ * @param <T> data type for {@code y()} output
+ */
+@Operator(group = "linalg")
+public final class Transpose<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new Transpose operation.
+   * 
+   * @param scope current scope
+   * @param x 
+   * @param perm 
+   * @return a new instance of Transpose
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType, U extends TNumber> Transpose<T> create(Scope scope, Operand<T> x, Operand<U> perm) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Transpose", scope.makeOpName("Transpose"));
+    opBuilder.addInput(x.asOutput());
+    opBuilder.addInput(perm.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new Transpose<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> y() {
+    return y;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return y;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Transpose";
+  
+  private Output<T> y;
+  
+  private Transpose(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    y = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/TriangularSolve.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/TriangularSolve.java
new file mode 100644
index 00000000000..da86d99b6c6
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/TriangularSolve.java
@@ -0,0 +1,189 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.linalg;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Solves systems of linear equations with upper or lower triangular matrices by backsubstitution.
+ * <p>
+ * 
+ * `matrix` is a tensor of shape `[..., M, M]` whose inner-most 2 dimensions form
+ * square matrices. If `lower` is `True` then the strictly upper triangular part
+ * of each inner-most matrix is assumed to be zero and not accessed.
+ * If `lower` is False then the strictly lower triangular part of each inner-most
+ * matrix is assumed to be zero and not accessed.
+ * `rhs` is a tensor of shape `[..., M, N]`.
+ * <p>
+ * The output is a tensor of shape `[..., M, N]`. If `adjoint` is
+ * `True` then the innermost matrices in `output` satisfy matrix equations
+ * `matrix[..., :, :] * output[..., :, :] = rhs[..., :, :]`.
+ * If `adjoint` is `False` then the strictly then the  innermost matrices in
+ * `output` satisfy matrix equations
+ * `adjoint(matrix[..., i, k]) * output[..., k, j] = rhs[..., i, j]`.
+ * <p>
+ * Note, the batch shapes for the inputs only need to broadcast.
+ * <p>
+ * Example:
+ * <pre>{@code
+ * a = tf.constant([[3,  0,  0,  0],
+ *                  [2,  1,  0,  0],
+ *                  [1,  0,  1,  0],
+ *                  [1,  1,  1,  1]], dtype=tf.float32)
+ * 
+ * b = tf.constant([[4],
+ *                  [2],
+ *                  [4],
+ *                  [2]], dtype=tf.float32)
+ * 
+ * x = tf.linalg.triangular_solve(a, b, lower=True)
+ * x
+ * # <tf.Tensor: shape=(4, 1), dtype=float32, numpy=
+ * # array([[ 1.3333334 ],
+ * #        [-0.66666675],
+ * #        [ 2.6666665 ],
+ * #        [-1.3333331 ]], dtype=float32)>
+ * 
+ * # in python3 one can use `a@x`
+ * tf.matmul(a, x)
+ * # <tf.Tensor: shape=(4, 1), dtype=float32, numpy=
+ * # array([[4.       ],
+ * #        [2.       ],
+ * #        [4.       ],
+ * #        [1.9999999]], dtype=float32)>
+ * }</pre>
+ * 
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator(group = "linalg")
+public final class TriangularSolve<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.linalg.TriangularSolve}
+   */
+  public static class Options {
+    
+    /**
+     * @param lower Boolean indicating whether the innermost matrices in `matrix` are
+     * lower or upper triangular.
+     */
+    public Options lower(Boolean lower) {
+      this.lower = lower;
+      return this;
+    }
+    
+    /**
+     * @param adjoint Boolean indicating whether to solve with `matrix` or its (block-wise)
+     *          adjoint.
+     * <p>
+     * @compatibility(numpy)
+     * Equivalent to scipy.linalg.solve_triangular
+     * @end_compatibility
+     */
+    public Options adjoint(Boolean adjoint) {
+      this.adjoint = adjoint;
+      return this;
+    }
+    
+    private Boolean lower;
+    private Boolean adjoint;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new TriangularSolve operation.
+   * 
+   * @param scope current scope
+   * @param matrix Shape is `[..., M, M]`.
+   * @param rhs Shape is `[..., M, K]`.
+   * @param options carries optional attributes values
+   * @return a new instance of TriangularSolve
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> TriangularSolve<T> create(Scope scope, Operand<T> matrix, Operand<T> rhs, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("MatrixTriangularSolve", scope.makeOpName("TriangularSolve"));
+    opBuilder.addInput(matrix.asOutput());
+    opBuilder.addInput(rhs.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.lower != null) {
+          opBuilder.setAttr("lower", opts.lower);
+        }
+        if (opts.adjoint != null) {
+          opBuilder.setAttr("adjoint", opts.adjoint);
+        }
+      }
+    }
+    return new TriangularSolve<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param lower Boolean indicating whether the innermost matrices in `matrix` are
+   * lower or upper triangular.
+   */
+  public static Options lower(Boolean lower) {
+    return new Options().lower(lower);
+  }
+  
+  /**
+   * @param adjoint Boolean indicating whether to solve with `matrix` or its (block-wise)
+   *          adjoint.
+   * <p>
+   * @compatibility(numpy)
+   * Equivalent to scipy.linalg.solve_triangular
+   * @end_compatibility
+   */
+  public static Options adjoint(Boolean adjoint) {
+    return new Options().adjoint(adjoint);
+  }
+  
+  /**
+   * Shape is `[..., M, K]`.
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "MatrixTriangularSolve";
+  
+  private Output<T> output;
+  
+  private TriangularSolve(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/TridiagonalMatMul.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/TridiagonalMatMul.java
new file mode 100644
index 00000000000..2ff1c18fc60
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/TridiagonalMatMul.java
@@ -0,0 +1,86 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.linalg;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Calculate product with tridiagonal matrix.
+ * <p>
+ * Calculates product of two matrices, where left matrix is a tridiagonal matrix.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+public final class TridiagonalMatMul<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new TridiagonalMatMul operation.
+   * 
+   * @param scope current scope
+   * @param superdiag Tensor of shape `[..., 1, M]`, representing superdiagonals of
+   * tri-diagonal matrices to the left of multiplication. Last element is ignored.
+   * @param maindiag Tensor of shape `[..., 1, M]`, representing main diagonals of tri-diagonal
+   * matrices to the left of multiplication.
+   * @param subdiag Tensor of shape `[..., 1, M]`, representing subdiagonals of tri-diagonal
+   * matrices to the left of multiplication. First element is ignored.
+   * @param rhs Tensor of shape `[..., M, N]`, representing MxN matrices to the right of
+   * multiplication.
+   * @return a new instance of TridiagonalMatMul
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> TridiagonalMatMul<T> create(Scope scope, Operand<T> superdiag, Operand<T> maindiag, Operand<T> subdiag, Operand<T> rhs) {
+    OperationBuilder opBuilder = scope.env().opBuilder("TridiagonalMatMul", scope.makeOpName("TridiagonalMatMul"));
+    opBuilder.addInput(superdiag.asOutput());
+    opBuilder.addInput(maindiag.asOutput());
+    opBuilder.addInput(subdiag.asOutput());
+    opBuilder.addInput(rhs.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new TridiagonalMatMul<T>(opBuilder.build());
+  }
+  
+  /**
+   * Tensor of shape `[..., M, N]` containing the product.
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "TridiagonalMatMul";
+  
+  private Output<T> output;
+  
+  private TridiagonalMatMul(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/TridiagonalSolve.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/TridiagonalSolve.java
new file mode 100644
index 00000000000..84a65b02b55
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/TridiagonalSolve.java
@@ -0,0 +1,124 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.linalg;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Solves tridiagonal systems of equations.
+ * <p>
+ *   Solves tridiagonal systems of equations.
+ *   Supports batch dimensions and multiple right-hand sides per each left-hand
+ *   side.
+ *   On CPU, solution is computed via Gaussian elimination with or without partial
+ *   pivoting, depending on `partial_pivoting` attribute. On GPU, Nvidia's cuSPARSE
+ *   library is used: https://docs.nvidia.com/cuda/cusparse/index.html#gtsv
+ *   Partial pivoting is not yet supported by XLA backends.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+public final class TridiagonalSolve<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.linalg.TridiagonalSolve}
+   */
+  public static class Options {
+    
+    /**
+     * @param partialPivoting Whether to apply partial pivoting. Partial pivoting makes the procedure more
+     * stable, but slower.
+     */
+    public Options partialPivoting(Boolean partialPivoting) {
+      this.partialPivoting = partialPivoting;
+      return this;
+    }
+    
+    private Boolean partialPivoting;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new TridiagonalSolve operation.
+   * 
+   * @param scope current scope
+   * @param diagonals Tensor of shape `[..., 3, M]` whose innermost 2 dimensions represent the
+   * tridiagonal matrices with three rows being the superdiagonal, diagonals, and
+   * subdiagonals, in order. The last element of the superdiagonal and the first
+   * element of the subdiagonal is ignored.
+   * @param rhs Tensor of shape `[..., M, K]`, representing K right-hand sides per each
+   * left-hand side.
+   * @param options carries optional attributes values
+   * @return a new instance of TridiagonalSolve
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> TridiagonalSolve<T> create(Scope scope, Operand<T> diagonals, Operand<T> rhs, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("TridiagonalSolve", scope.makeOpName("TridiagonalSolve"));
+    opBuilder.addInput(diagonals.asOutput());
+    opBuilder.addInput(rhs.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.partialPivoting != null) {
+          opBuilder.setAttr("partial_pivoting", opts.partialPivoting);
+        }
+      }
+    }
+    return new TridiagonalSolve<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param partialPivoting Whether to apply partial pivoting. Partial pivoting makes the procedure more
+   * stable, but slower.
+   */
+  public static Options partialPivoting(Boolean partialPivoting) {
+    return new Options().partialPivoting(partialPivoting);
+  }
+  
+  /**
+   * Tensor of shape `[..., M, K]` containing the solutions
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "TridiagonalSolve";
+  
+  private Output<T> output;
+  
+  private TridiagonalSolve(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/CSRSparseMatrixComponents.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/CSRSparseMatrixComponents.java
new file mode 100644
index 00000000000..3d598e9f450
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/CSRSparseMatrixComponents.java
@@ -0,0 +1,95 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.linalg.sparse;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Reads out the CSR components at batch `index`.
+ * <p>
+ * This op is meant only for debugging / testing, and its interface is not expected
+ * to be stable.
+ * 
+ * @param <T> data type for {@code values()} output
+ */
+public final class CSRSparseMatrixComponents<T extends TType> extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new CSRSparseMatrixComponents operation.
+   * 
+   * @param scope current scope
+   * @param csrSparseMatrix A batched CSRSparseMatrix.
+   * @param index The index in `csr_sparse_matrix`'s batch.
+   * @param type 
+   * @return a new instance of CSRSparseMatrixComponents
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> CSRSparseMatrixComponents<T> create(Scope scope, Operand<?> csrSparseMatrix, Operand<TInt32> index, Class<T> type) {
+    OperationBuilder opBuilder = scope.env().opBuilder("CSRSparseMatrixComponents", scope.makeOpName("CSRSparseMatrixComponents"));
+    opBuilder.addInput(csrSparseMatrix.asOutput());
+    opBuilder.addInput(index.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("type", type);
+    return new CSRSparseMatrixComponents<T>(opBuilder.build());
+  }
+  
+  /**
+   * An array containing CSR matrix row pointers.
+   */
+  public Output<TInt32> rowPtrs() {
+    return rowPtrs;
+  }
+  
+  /**
+   * An array containing CSR matrix column indices.
+   */
+  public Output<TInt32> colInds() {
+    return colInds;
+  }
+  
+  /**
+   * An array containing CSR matrix nonzero values.
+   */
+  public Output<T> values() {
+    return values;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "CSRSparseMatrixComponents";
+  
+  private Output<TInt32> rowPtrs;
+  private Output<TInt32> colInds;
+  private Output<T> values;
+  
+  private CSRSparseMatrixComponents(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    rowPtrs = operation.output(outputIdx++);
+    colInds = operation.output(outputIdx++);
+    values = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/CSRSparseMatrixToDense.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/CSRSparseMatrixToDense.java
new file mode 100644
index 00000000000..6b5fe3c0e82
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/CSRSparseMatrixToDense.java
@@ -0,0 +1,76 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.linalg.sparse;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Convert a (possibly batched) CSRSparseMatrix to dense.
+ * 
+ * @param <T> data type for {@code denseOutput()} output
+ */
+public final class CSRSparseMatrixToDense<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new CSRSparseMatrixToDense operation.
+   * 
+   * @param scope current scope
+   * @param sparseInput A batched CSRSparseMatrix.
+   * @param type 
+   * @return a new instance of CSRSparseMatrixToDense
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> CSRSparseMatrixToDense<T> create(Scope scope, Operand<?> sparseInput, Class<T> type) {
+    OperationBuilder opBuilder = scope.env().opBuilder("CSRSparseMatrixToDense", scope.makeOpName("CSRSparseMatrixToDense"));
+    opBuilder.addInput(sparseInput.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("type", type);
+    return new CSRSparseMatrixToDense<T>(opBuilder.build());
+  }
+  
+  /**
+   * A dense tensor.
+   */
+  public Output<T> denseOutput() {
+    return denseOutput;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return denseOutput;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "CSRSparseMatrixToDense";
+  
+  private Output<T> denseOutput;
+  
+  private CSRSparseMatrixToDense(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    denseOutput = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/CSRSparseMatrixToSparseTensor.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/CSRSparseMatrixToSparseTensor.java
new file mode 100644
index 00000000000..3e203cac761
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/CSRSparseMatrixToSparseTensor.java
@@ -0,0 +1,90 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.linalg.sparse;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Converts a (possibly batched) CSRSparesMatrix to a SparseTensor.
+ * 
+ * @param <T> data type for {@code values()} output
+ */
+public final class CSRSparseMatrixToSparseTensor<T extends TType> extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new CSRSparseMatrixToSparseTensor operation.
+   * 
+   * @param scope current scope
+   * @param sparseMatrix A (possibly batched) CSRSparseMatrix.
+   * @param type 
+   * @return a new instance of CSRSparseMatrixToSparseTensor
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> CSRSparseMatrixToSparseTensor<T> create(Scope scope, Operand<?> sparseMatrix, Class<T> type) {
+    OperationBuilder opBuilder = scope.env().opBuilder("CSRSparseMatrixToSparseTensor", scope.makeOpName("CSRSparseMatrixToSparseTensor"));
+    opBuilder.addInput(sparseMatrix.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("type", type);
+    return new CSRSparseMatrixToSparseTensor<T>(opBuilder.build());
+  }
+  
+  /**
+   * SparseTensor indices.
+   */
+  public Output<TInt64> indices() {
+    return indices;
+  }
+  
+  /**
+   * SparseTensor values.
+   */
+  public Output<T> values() {
+    return values;
+  }
+  
+  /**
+   * SparseTensor dense shape.
+   */
+  public Output<TInt64> denseShape() {
+    return denseShape;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "CSRSparseMatrixToSparseTensor";
+  
+  private Output<TInt64> indices;
+  private Output<T> values;
+  private Output<TInt64> denseShape;
+  
+  private CSRSparseMatrixToSparseTensor(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    indices = operation.output(outputIdx++);
+    values = operation.output(outputIdx++);
+    denseShape = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/DenseToCSRSparseMatrix.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/DenseToCSRSparseMatrix.java
new file mode 100644
index 00000000000..0033c9a9760
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/DenseToCSRSparseMatrix.java
@@ -0,0 +1,76 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.linalg.sparse;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Converts a dense tensor to a (possibly batched) CSRSparseMatrix.
+ */
+public final class DenseToCSRSparseMatrix extends RawOp implements Operand<TType> {
+  
+  /**
+   * Factory method to create a class wrapping a new DenseToCSRSparseMatrix operation.
+   * 
+   * @param scope current scope
+   * @param denseInput A Dense tensor.
+   * @param indices Indices of nonzero elements.
+   * @return a new instance of DenseToCSRSparseMatrix
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> DenseToCSRSparseMatrix create(Scope scope, Operand<T> denseInput, Operand<TInt64> indices) {
+    OperationBuilder opBuilder = scope.env().opBuilder("DenseToCSRSparseMatrix", scope.makeOpName("DenseToCSRSparseMatrix"));
+    opBuilder.addInput(denseInput.asOutput());
+    opBuilder.addInput(indices.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new DenseToCSRSparseMatrix(opBuilder.build());
+  }
+  
+  /**
+   * A (possibly batched) CSRSparseMatrix.
+   */
+  public Output<?> sparseOutput() {
+    return sparseOutput;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) sparseOutput;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "DenseToCSRSparseMatrix";
+  
+  private Output<?> sparseOutput;
+  
+  private DenseToCSRSparseMatrix(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    sparseOutput = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixAdd.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixAdd.java
new file mode 100644
index 00000000000..d8e676743a1
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixAdd.java
@@ -0,0 +1,82 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.linalg.sparse;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Sparse addition of two CSR matrices, C = alpha * A + beta * B.
+ * <p>
+ * The gradients of SparseMatrixAdd outputs with respect to alpha and beta are not
+ * currently defined (TensorFlow will return zeros for these entries).
+ */
+public final class SparseMatrixAdd extends RawOp implements Operand<TType> {
+  
+  /**
+   * Factory method to create a class wrapping a new SparseMatrixAdd operation.
+   * 
+   * @param scope current scope
+   * @param a A CSRSparseMatrix.
+   * @param b A CSRSparseMatrix.
+   * @param alpha A constant scalar.
+   * @param beta A constant scalar.
+   * @return a new instance of SparseMatrixAdd
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> SparseMatrixAdd create(Scope scope, Operand<?> a, Operand<?> b, Operand<T> alpha, Operand<T> beta) {
+    OperationBuilder opBuilder = scope.env().opBuilder("SparseMatrixAdd", scope.makeOpName("SparseMatrixAdd"));
+    opBuilder.addInput(a.asOutput());
+    opBuilder.addInput(b.asOutput());
+    opBuilder.addInput(alpha.asOutput());
+    opBuilder.addInput(beta.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new SparseMatrixAdd(opBuilder.build());
+  }
+  
+  /**
+   * A CSRSparseMatrix.
+   */
+  public Output<?> c() {
+    return c;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) c;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "SparseMatrixAdd";
+  
+  private Output<?> c;
+  
+  private SparseMatrixAdd(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    c = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixMatMul.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixMatMul.java
new file mode 100644
index 00000000000..2ee48d29d79
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixMatMul.java
@@ -0,0 +1,230 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.linalg.sparse;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Matrix-multiplies a sparse matrix with a dense matrix.
+ * <p>
+ * Returns a dense matrix.
+ * For inputs A and B, where A is CSR and B is dense; this op returns a dense C;
+ * <p>
+ * If transpose_output is false, returns:
+ * <pre>{@code
+ *   C = A . B
+ * }</pre>
+ * If transpose_output is `true`, returns:
+ * <pre>{@code
+ *   C = transpose(A . B) = transpose(B) . transpose(A)
+ * }</pre>
+ * where the transposition is performed along the two innermost (matrix)
+ * dimensions.
+ * <p>
+ * If conjugate_output is `true`, returns:
+ * <pre>{@code
+ *   C = conjugate(A . B) = conjugate(A) . conjugate(B)
+ * }</pre>
+ * If both conjugate_output and transpose_output are `true`, returns:
+ * <pre>{@code
+ *   C = conjugate(transpose(A . B)) = conjugate(transpose(B)) .
+ *                                     conjugate(transpose(A))
+ * }</pre>
+ * 
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+public final class SparseMatrixMatMul<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.linalg.sparse.SparseMatrixMatMul}
+   */
+  public static class Options {
+    
+    /**
+     * @param transposeA Indicates whether `a` should be transposed.
+     */
+    public Options transposeA(Boolean transposeA) {
+      this.transposeA = transposeA;
+      return this;
+    }
+    
+    /**
+     * @param transposeB Indicates whether `b` should be transposed.
+     */
+    public Options transposeB(Boolean transposeB) {
+      this.transposeB = transposeB;
+      return this;
+    }
+    
+    /**
+     * @param adjointA Indicates whether `a` should be conjugate-transposed.
+     */
+    public Options adjointA(Boolean adjointA) {
+      this.adjointA = adjointA;
+      return this;
+    }
+    
+    /**
+     * @param adjointB Indicates whether `b` should be conjugate-transposed.
+     */
+    public Options adjointB(Boolean adjointB) {
+      this.adjointB = adjointB;
+      return this;
+    }
+    
+    /**
+     * @param transposeOutput Transposes the product of `a` and `b`.
+     */
+    public Options transposeOutput(Boolean transposeOutput) {
+      this.transposeOutput = transposeOutput;
+      return this;
+    }
+    
+    /**
+     * @param conjugateOutput Conjugates the product of `a` and `b`.
+     */
+    public Options conjugateOutput(Boolean conjugateOutput) {
+      this.conjugateOutput = conjugateOutput;
+      return this;
+    }
+    
+    private Boolean transposeA;
+    private Boolean transposeB;
+    private Boolean adjointA;
+    private Boolean adjointB;
+    private Boolean transposeOutput;
+    private Boolean conjugateOutput;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new SparseMatrixMatMul operation.
+   * 
+   * @param scope current scope
+   * @param a A CSRSparseMatrix.
+   * @param b A dense tensor.
+   * @param options carries optional attributes values
+   * @return a new instance of SparseMatrixMatMul
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> SparseMatrixMatMul<T> create(Scope scope, Operand<?> a, Operand<T> b, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("SparseMatrixMatMul", scope.makeOpName("SparseMatrixMatMul"));
+    opBuilder.addInput(a.asOutput());
+    opBuilder.addInput(b.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.transposeA != null) {
+          opBuilder.setAttr("transpose_a", opts.transposeA);
+        }
+        if (opts.transposeB != null) {
+          opBuilder.setAttr("transpose_b", opts.transposeB);
+        }
+        if (opts.adjointA != null) {
+          opBuilder.setAttr("adjoint_a", opts.adjointA);
+        }
+        if (opts.adjointB != null) {
+          opBuilder.setAttr("adjoint_b", opts.adjointB);
+        }
+        if (opts.transposeOutput != null) {
+          opBuilder.setAttr("transpose_output", opts.transposeOutput);
+        }
+        if (opts.conjugateOutput != null) {
+          opBuilder.setAttr("conjugate_output", opts.conjugateOutput);
+        }
+      }
+    }
+    return new SparseMatrixMatMul<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param transposeA Indicates whether `a` should be transposed.
+   */
+  public static Options transposeA(Boolean transposeA) {
+    return new Options().transposeA(transposeA);
+  }
+  
+  /**
+   * @param transposeB Indicates whether `b` should be transposed.
+   */
+  public static Options transposeB(Boolean transposeB) {
+    return new Options().transposeB(transposeB);
+  }
+  
+  /**
+   * @param adjointA Indicates whether `a` should be conjugate-transposed.
+   */
+  public static Options adjointA(Boolean adjointA) {
+    return new Options().adjointA(adjointA);
+  }
+  
+  /**
+   * @param adjointB Indicates whether `b` should be conjugate-transposed.
+   */
+  public static Options adjointB(Boolean adjointB) {
+    return new Options().adjointB(adjointB);
+  }
+  
+  /**
+   * @param transposeOutput Transposes the product of `a` and `b`.
+   */
+  public static Options transposeOutput(Boolean transposeOutput) {
+    return new Options().transposeOutput(transposeOutput);
+  }
+  
+  /**
+   * @param conjugateOutput Conjugates the product of `a` and `b`.
+   */
+  public static Options conjugateOutput(Boolean conjugateOutput) {
+    return new Options().conjugateOutput(conjugateOutput);
+  }
+  
+  /**
+   * A dense output tensor.
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "SparseMatrixMatMul";
+  
+  private Output<T> output;
+  
+  private SparseMatrixMatMul(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixMul.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixMul.java
new file mode 100644
index 00000000000..ad1571580f5
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixMul.java
@@ -0,0 +1,84 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.linalg.sparse;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Element-wise multiplication of a sparse matrix with a dense tensor.
+ * <p>
+ * Returns a sparse matrix.
+ * <p>
+ * The dense tensor `b` may be either a scalar; otherwise `a` must be a rank-3
+ * `SparseMatrix`; in this case `b` must be shaped `[batch_size, 1, 1]` and the
+ * multiply operation broadcasts.
+ * <p>
+ * <b>NOTE</b> even if `b` is zero, the sparsity structure of the output does not
+ * change.
+ */
+public final class SparseMatrixMul extends RawOp implements Operand<TType> {
+  
+  /**
+   * Factory method to create a class wrapping a new SparseMatrixMul operation.
+   * 
+   * @param scope current scope
+   * @param a A CSRSparseMatrix.
+   * @param b A dense tensor.
+   * @return a new instance of SparseMatrixMul
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> SparseMatrixMul create(Scope scope, Operand<?> a, Operand<T> b) {
+    OperationBuilder opBuilder = scope.env().opBuilder("SparseMatrixMul", scope.makeOpName("SparseMatrixMul"));
+    opBuilder.addInput(a.asOutput());
+    opBuilder.addInput(b.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new SparseMatrixMul(opBuilder.build());
+  }
+  
+  /**
+   * A dense output tensor.
+   */
+  public Output<?> output() {
+    return output;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "SparseMatrixMul";
+  
+  private Output<?> output;
+  
+  private SparseMatrixMul(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixNNZ.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixNNZ.java
new file mode 100644
index 00000000000..e730d181699
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixNNZ.java
@@ -0,0 +1,72 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.linalg.sparse;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+
+/**
+ * Returns the number of nonzeroes of `sparse_matrix`.
+ */
+public final class SparseMatrixNNZ extends RawOp implements Operand<TInt32> {
+  
+  /**
+   * Factory method to create a class wrapping a new SparseMatrixNNZ operation.
+   * 
+   * @param scope current scope
+   * @param sparseMatrix A CSRSparseMatrix.
+   * @return a new instance of SparseMatrixNNZ
+   */
+  @Endpoint(describeByClass = true)
+  public static SparseMatrixNNZ create(Scope scope, Operand<?> sparseMatrix) {
+    OperationBuilder opBuilder = scope.env().opBuilder("SparseMatrixNNZ", scope.makeOpName("SparseMatrixNNZ"));
+    opBuilder.addInput(sparseMatrix.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new SparseMatrixNNZ(opBuilder.build());
+  }
+  
+  /**
+   * The number of nonzeroes of `sparse_matrix`.
+   */
+  public Output<TInt32> nnz() {
+    return nnz;
+  }
+  
+  @Override
+  public Output<TInt32> asOutput() {
+    return nnz;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "SparseMatrixNNZ";
+  
+  private Output<TInt32> nnz;
+  
+  private SparseMatrixNNZ(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    nnz = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixOrderingAMD.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixOrderingAMD.java
new file mode 100644
index 00000000000..b833601b72e
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixOrderingAMD.java
@@ -0,0 +1,120 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.linalg.sparse;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+
+/**
+ * Computes the Approximate Minimum Degree (AMD) ordering of `input`.
+ * <p>
+ * Computes the Approximate Minimum Degree (AMD) ordering for a sparse matrix.
+ * <p>
+ * The returned permutation may be used to permute the rows and columns of the
+ * given sparse matrix. This typically results in permuted sparse matrix's sparse
+ * Cholesky (or other decompositions) in having fewer zero fill-in compared to
+ * decomposition of the original matrix.
+ * <p>
+ * The input sparse matrix may have rank 2 or rank 3. The output Tensor,
+ * representing would then have rank 1 or 2 respectively, with the same batch
+ * shape as the input.
+ * <p>
+ * Each component of the input sparse matrix must represent a square symmetric
+ * matrix; only the lower triangular part of the matrix is read. The values of the
+ * sparse matrix does not affect the returned permutation, only the sparsity
+ * pattern of the sparse matrix is used. Hence, a single AMD ordering may be
+ * reused for the Cholesky decompositions of sparse matrices with the same sparsity
+ * pattern but with possibly different values.
+ * <p>
+ * Each batch component of the output permutation represents a permutation of `N`
+ * elements, where the input sparse matrix components each have `N` rows. That is,
+ * the component contains each of the integers `{0, .. N-1}` exactly once. The
+ * `i`th element represents the row index that the `i`th row maps to.
+ * <p>
+ * Usage example:
+ * <pre>{@code
+ *     from tensorflow.python.ops.linalg.sparse import sparse_csr_matrix_ops
+ * 
+ *     a_indices = np.array([[0, 0], [1, 1], [2, 1], [2, 2], [3, 3]])
+ *     a_values = np.array([1.0, 2.0, 1.0, 3.0, 4.0], np.float32)
+ *     a_dense_shape = [4, 4]
+ * 
+ *     with tf.Session() as sess:
+ *       # Define (COO format) SparseTensor over Numpy array.
+ *       a_st = tf.sparse.SparseTensor(a_indices, a_values, a_dense_shape)
+ * 
+ *       # Convert SparseTensors to CSR SparseMatrix.
+ *       a_sm = sparse_csr_matrix_ops.sparse_tensor_to_csr_sparse_matrix(
+ *           a_st.indices, a_st.values, a_st.dense_shape)
+ * 
+ *       # Obtain the AMD Ordering for the CSR SparseMatrix.
+ *       ordering_amd = sparse_csr_matrix_ops.sparse_matrix_ordering_amd(sparse_matrix)
+ * 
+ *       ordering_amd_value = sess.run(ordering_amd)
+ * }</pre>
+ * `ordering_amd_value` stores the AMD ordering: `[1 2 3 0]`.
+ * <p>
+ * input: A `CSRSparseMatrix`.
+ */
+public final class SparseMatrixOrderingAMD extends RawOp implements Operand<TInt32> {
+  
+  /**
+   * Factory method to create a class wrapping a new SparseMatrixOrderingAMD operation.
+   * 
+   * @param scope current scope
+   * @param input A `CSRSparseMatrix`.
+   * @return a new instance of SparseMatrixOrderingAMD
+   */
+  @Endpoint(describeByClass = true)
+  public static SparseMatrixOrderingAMD create(Scope scope, Operand<?> input) {
+    OperationBuilder opBuilder = scope.env().opBuilder("SparseMatrixOrderingAMD", scope.makeOpName("SparseMatrixOrderingAMD"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new SparseMatrixOrderingAMD(opBuilder.build());
+  }
+  
+  /**
+   * The Approximate Minimum Degree (AMD) ordering of `input`.
+   */
+  public Output<TInt32> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<TInt32> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "SparseMatrixOrderingAMD";
+  
+  private Output<TInt32> output;
+  
+  private SparseMatrixOrderingAMD(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixSoftmax.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixSoftmax.java
new file mode 100644
index 00000000000..e8043662d36
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixSoftmax.java
@@ -0,0 +1,82 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.linalg.sparse;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Calculates the softmax of a CSRSparseMatrix.
+ * <p>
+ * Calculate the softmax of the innermost dimensions of a SparseMatrix.
+ * <p>
+ * Missing values are treated as `-inf` (i.e., logits of zero probability); and
+ * the output has the same sparsity structure as the input (though missing values
+ * in the output may now be treated as having probability zero).
+ */
+public final class SparseMatrixSoftmax extends RawOp implements Operand<TType> {
+  
+  /**
+   * Factory method to create a class wrapping a new SparseMatrixSoftmax operation.
+   * 
+   * @param scope current scope
+   * @param logits A CSRSparseMatrix.
+   * @param type 
+   * @return a new instance of SparseMatrixSoftmax
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> SparseMatrixSoftmax create(Scope scope, Operand<?> logits, Class<T> type) {
+    OperationBuilder opBuilder = scope.env().opBuilder("SparseMatrixSoftmax", scope.makeOpName("SparseMatrixSoftmax"));
+    opBuilder.addInput(logits.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("type", type);
+    return new SparseMatrixSoftmax(opBuilder.build());
+  }
+  
+  /**
+   * A CSRSparseMatrix.
+   */
+  public Output<?> softmax() {
+    return softmax;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) softmax;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "SparseMatrixSoftmax";
+  
+  private Output<?> softmax;
+  
+  private SparseMatrixSoftmax(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    softmax = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixSoftmaxGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixSoftmaxGrad.java
new file mode 100644
index 00000000000..cea37dd4b25
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixSoftmaxGrad.java
@@ -0,0 +1,78 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.linalg.sparse;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Calculates the gradient of the SparseMatrixSoftmax op.
+ */
+public final class SparseMatrixSoftmaxGrad extends RawOp implements Operand<TType> {
+  
+  /**
+   * Factory method to create a class wrapping a new SparseMatrixSoftmaxGrad operation.
+   * 
+   * @param scope current scope
+   * @param softmax A CSRSparseMatrix.
+   * @param gradSoftmax The gradient of `softmax`.
+   * @param type 
+   * @return a new instance of SparseMatrixSoftmaxGrad
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> SparseMatrixSoftmaxGrad create(Scope scope, Operand<?> softmax, Operand<?> gradSoftmax, Class<T> type) {
+    OperationBuilder opBuilder = scope.env().opBuilder("SparseMatrixSoftmaxGrad", scope.makeOpName("SparseMatrixSoftmaxGrad"));
+    opBuilder.addInput(softmax.asOutput());
+    opBuilder.addInput(gradSoftmax.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("type", type);
+    return new SparseMatrixSoftmaxGrad(opBuilder.build());
+  }
+  
+  /**
+   * The output gradient.
+   */
+  public Output<?> gradient() {
+    return gradient;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) gradient;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "SparseMatrixSoftmaxGrad";
+  
+  private Output<?> gradient;
+  
+  private SparseMatrixSoftmaxGrad(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    gradient = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixSparseCholesky.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixSparseCholesky.java
new file mode 100644
index 00000000000..e70ba2b702b
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixSparseCholesky.java
@@ -0,0 +1,149 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.linalg.sparse;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Computes the sparse Cholesky decomposition of `input`.
+ * <p>
+ * Computes the Sparse Cholesky decomposition of a sparse matrix, with the given
+ * fill-in reducing permutation.
+ * <p>
+ * The input sparse matrix and the fill-in reducing permutation `permutation` must
+ * have compatible shapes. If the sparse matrix has rank 3; with the batch
+ * dimension `B`, then the `permutation` must be of rank 2; with the same batch
+ * dimension `B`. There is no support for broadcasting.
+ * <p>
+ * Furthermore, each component vector of `permutation` must be of length `N`,
+ * containing each of the integers {0, 1, ..., N - 1} exactly once, where `N` is
+ * the number of rows of each component of the sparse matrix.
+ * <p>
+ * Each component of the input sparse matrix must represent a symmetric positive
+ * definite (SPD) matrix; although only the lower triangular part of the matrix is
+ * read. If any individual component is not SPD, then an InvalidArgument error is
+ * thrown.
+ * <p>
+ * The returned sparse matrix has the same dense shape as the input sparse matrix.
+ * For each component `A` of the input sparse matrix, the corresponding output
+ * sparse matrix represents `L`, the lower triangular Cholesky factor satisfying
+ * the following identity:
+ * <pre>{@code
+ *   A = L * Lt
+ * }</pre>
+ * where Lt denotes the transpose of L (or its conjugate transpose, if `type` is
+ * `complex64` or `complex128`).
+ * <p>
+ * The `type` parameter denotes the type of the matrix elements. The supported
+ * types are: `float32`, `float64`, `complex64` and `complex128`.
+ * <p>
+ * Usage example:
+ * <pre>{@code
+ *     from tensorflow.python.ops.linalg.sparse import sparse_csr_matrix_ops
+ * 
+ *     a_indices = np.array([[0, 0], [1, 1], [2, 1], [2, 2], [3, 3]])
+ *     a_values = np.array([1.0, 2.0, 1.0, 3.0, 4.0], np.float32)
+ *     a_dense_shape = [4, 4]
+ * 
+ *     with tf.Session() as sess:
+ *       # Define (COO format) SparseTensor over Numpy array.
+ *       a_st = tf.sparse.SparseTensor(a_indices, a_values, a_dense_shape)
+ * 
+ *       # Convert SparseTensors to CSR SparseMatrix.
+ *       a_sm = sparse_csr_matrix_ops.sparse_tensor_to_csr_sparse_matrix(
+ *           a_st.indices, a_st.values, a_st.dense_shape)
+ * 
+ *       # Obtain the Sparse Cholesky factor using AMD Ordering for reducing zero
+ *       # fill-in (number of structural non-zeros in the sparse Cholesky factor).
+ *       ordering_amd = sparse_csr_matrix_ops.sparse_matrix_ordering_amd(sparse_matrix)
+ *       cholesky_sparse_matrices = (
+ *           sparse_csr_matrix_ops.sparse_matrix_sparse_cholesky(
+ *               sparse_matrix, ordering_amd, type=tf.float32))
+ * 
+ *       # Convert the CSRSparseMatrix Cholesky factor to a dense Tensor
+ *       dense_cholesky = sparse_csr_matrix_ops.csr_sparse_matrix_to_dense(
+ *           cholesky_sparse_matrices, tf.float32)
+ * 
+ *       # Evaluate the dense Tensor value.
+ *       dense_cholesky_value = sess.run(dense_cholesky)
+ * }</pre>
+ * `dense_cholesky_value` stores the dense Cholesky factor:
+ * <pre>{@code
+ *     [[  1.  0.    0.    0.]
+ *      [  0.  1.41  0.    0.]
+ *      [  0.  0.70  1.58  0.]
+ *      [  0.  0.    0.    2.]]
+ * }</pre>
+ * input: A `CSRSparseMatrix`.
+ * permutation: A `Tensor`.
+ * type: The type of `input`.
+ */
+public final class SparseMatrixSparseCholesky extends RawOp implements Operand<TType> {
+  
+  /**
+   * Factory method to create a class wrapping a new SparseMatrixSparseCholesky operation.
+   * 
+   * @param scope current scope
+   * @param input A `CSRSparseMatrix`.
+   * @param permutation A fill-in reducing permutation matrix.
+   * @param type 
+   * @return a new instance of SparseMatrixSparseCholesky
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> SparseMatrixSparseCholesky create(Scope scope, Operand<?> input, Operand<TInt32> permutation, Class<T> type) {
+    OperationBuilder opBuilder = scope.env().opBuilder("SparseMatrixSparseCholesky", scope.makeOpName("SparseMatrixSparseCholesky"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(permutation.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("type", type);
+    return new SparseMatrixSparseCholesky(opBuilder.build());
+  }
+  
+  /**
+   * The sparse Cholesky decompsition of `input`.
+   */
+  public Output<?> output() {
+    return output;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "SparseMatrixSparseCholesky";
+  
+  private Output<?> output;
+  
+  private SparseMatrixSparseCholesky(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixSparseMatMul.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixSparseMatMul.java
new file mode 100644
index 00000000000..26ae566f8be
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixSparseMatMul.java
@@ -0,0 +1,241 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.linalg.sparse;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Sparse-matrix-multiplies two CSR matrices `a` and `b`.
+ * <p>
+ * Performs a matrix multiplication of a sparse matrix `a` with a sparse matrix
+ * `b`; returns a sparse matrix `a * b`, unless either `a` or `b` is transposed or
+ * adjointed.
+ * <p>
+ * Each matrix may be transposed or adjointed (conjugated and transposed)
+ * according to the Boolean parameters `transpose_a`, `adjoint_a`, `transpose_b`
+ * and `adjoint_b`. At most one of `transpose_a` or `adjoint_a` may be True.
+ * Similarly, at most one of `transpose_b` or `adjoint_b` may be True.
+ * <p>
+ * The inputs must have compatible shapes. That is, the inner dimension of `a`
+ * must be equal to the outer dimension of `b`. This requirement is adjusted
+ * according to whether either `a` or `b` is transposed or adjointed.
+ * <p>
+ * The `type` parameter denotes the type of the matrix elements. Both `a` and `b`
+ * must have the same type. The supported types are: `float32`, `float64`,
+ * `complex64` and `complex128`.
+ * <p>
+ * Both `a` and `b` must have the same rank. Broadcasting is not supported. If they
+ * have rank 3, each batch of 2D CSRSparseMatrices within `a` and `b` must have the
+ * same dense shape.
+ * <p>
+ * The sparse matrix product may have numeric (non-structural) zeros.
+ * TODO(anudhyan): Consider adding a boolean attribute to control whether to prune
+ * zeros.
+ * <p>
+ * Usage example:
+ * <pre>{@code
+ *     from tensorflow.python.ops.linalg.sparse import sparse_csr_matrix_ops
+ * 
+ *     a_indices = np.array([[0, 0], [2, 3], [2, 4], [3, 0]])
+ *     a_values = np.array([1.0, 5.0, -1.0, -2.0], np.float32)
+ *     a_dense_shape = [4, 5]
+ * 
+ *     b_indices = np.array([[0, 0], [3, 0], [3, 1]])
+ *     b_values = np.array([2.0, 7.0, 8.0], np.float32)
+ *     b_dense_shape = [5, 3]
+ * 
+ *     with tf.Session() as sess:
+ *       # Define (COO format) Sparse Tensors over Numpy arrays
+ *       a_st = tf.sparse.SparseTensor(a_indices, a_values, a_dense_shape)
+ *       b_st = tf.sparse.SparseTensor(b_indices, b_values, b_dense_shape)
+ * 
+ *       # Convert SparseTensors to CSR SparseMatrix
+ *       a_sm = sparse_csr_matrix_ops.sparse_tensor_to_csr_sparse_matrix(
+ *           a_st.indices, a_st.values, a_st.dense_shape)
+ *       b_sm = sparse_csr_matrix_ops.sparse_tensor_to_csr_sparse_matrix(
+ *           b_st.indices, b_st.values, b_st.dense_shape)
+ * 
+ *       # Compute the CSR SparseMatrix matrix multiplication
+ *       c_sm = sparse_csr_matrix_ops.sparse_matrix_sparse_mat_mul(
+ *           a=a_sm, b=b_sm, type=tf.float32)
+ * 
+ *       # Convert the CSR SparseMatrix product to a dense Tensor
+ *       c_sm_dense = sparse_csr_matrix_ops.csr_sparse_matrix_to_dense(
+ *           c_sm, tf.float32)
+ *       # Evaluate the dense Tensor value
+ *       c_sm_dense_value = sess.run(c_sm_dense)
+ * }</pre>
+ * `c_sm_dense_value` stores the dense matrix product:
+ * <pre>{@code
+ *     [[  2.   0.   0.]
+ *      [  0.   0.   0.]
+ *      [ 35.  40.   0.]
+ *      [ -4.   0.   0.]]
+ * }</pre>
+ * a: A `CSRSparseMatrix`.
+ * b: A `CSRSparseMatrix` with the same type and rank as `a`.
+ * type: The type of both `a` and `b`.
+ * transpose_a: If True, `a` transposed before multiplication.
+ * transpose_b: If True, `b` transposed before multiplication.
+ * adjoint_a: If True, `a` adjointed before multiplication.
+ * adjoint_b: If True, `b` adjointed before multiplication.
+ */
+public final class SparseMatrixSparseMatMul extends RawOp implements Operand<TType> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.linalg.sparse.SparseMatrixSparseMatMul}
+   */
+  public static class Options {
+    
+    /**
+     * @param transposeA Indicates whether `a` should be transposed.
+     */
+    public Options transposeA(Boolean transposeA) {
+      this.transposeA = transposeA;
+      return this;
+    }
+    
+    /**
+     * @param transposeB Indicates whether `b` should be transposed.
+     */
+    public Options transposeB(Boolean transposeB) {
+      this.transposeB = transposeB;
+      return this;
+    }
+    
+    /**
+     * @param adjointA Indicates whether `a` should be conjugate-transposed.
+     */
+    public Options adjointA(Boolean adjointA) {
+      this.adjointA = adjointA;
+      return this;
+    }
+    
+    /**
+     * @param adjointB Indicates whether `b` should be conjugate-transposed.
+     */
+    public Options adjointB(Boolean adjointB) {
+      this.adjointB = adjointB;
+      return this;
+    }
+    
+    private Boolean transposeA;
+    private Boolean transposeB;
+    private Boolean adjointA;
+    private Boolean adjointB;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new SparseMatrixSparseMatMul operation.
+   * 
+   * @param scope current scope
+   * @param a A CSRSparseMatrix.
+   * @param b A CSRSparseMatrix.
+   * @param type 
+   * @param options carries optional attributes values
+   * @return a new instance of SparseMatrixSparseMatMul
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> SparseMatrixSparseMatMul create(Scope scope, Operand<?> a, Operand<?> b, Class<T> type, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("SparseMatrixSparseMatMul", scope.makeOpName("SparseMatrixSparseMatMul"));
+    opBuilder.addInput(a.asOutput());
+    opBuilder.addInput(b.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("type", type);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.transposeA != null) {
+          opBuilder.setAttr("transpose_a", opts.transposeA);
+        }
+        if (opts.transposeB != null) {
+          opBuilder.setAttr("transpose_b", opts.transposeB);
+        }
+        if (opts.adjointA != null) {
+          opBuilder.setAttr("adjoint_a", opts.adjointA);
+        }
+        if (opts.adjointB != null) {
+          opBuilder.setAttr("adjoint_b", opts.adjointB);
+        }
+      }
+    }
+    return new SparseMatrixSparseMatMul(opBuilder.build());
+  }
+  
+  /**
+   * @param transposeA Indicates whether `a` should be transposed.
+   */
+  public static Options transposeA(Boolean transposeA) {
+    return new Options().transposeA(transposeA);
+  }
+  
+  /**
+   * @param transposeB Indicates whether `b` should be transposed.
+   */
+  public static Options transposeB(Boolean transposeB) {
+    return new Options().transposeB(transposeB);
+  }
+  
+  /**
+   * @param adjointA Indicates whether `a` should be conjugate-transposed.
+   */
+  public static Options adjointA(Boolean adjointA) {
+    return new Options().adjointA(adjointA);
+  }
+  
+  /**
+   * @param adjointB Indicates whether `b` should be conjugate-transposed.
+   */
+  public static Options adjointB(Boolean adjointB) {
+    return new Options().adjointB(adjointB);
+  }
+  
+  /**
+   * A CSRSparseMatrix.
+   */
+  public Output<?> c() {
+    return c;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) c;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "SparseMatrixSparseMatMul";
+  
+  private Output<?> c;
+  
+  private SparseMatrixSparseMatMul(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    c = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixTranspose.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixTranspose.java
new file mode 100644
index 00000000000..05e54b0e0e6
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixTranspose.java
@@ -0,0 +1,112 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.linalg.sparse;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Transposes the inner (matrix) dimensions of a CSRSparseMatrix.
+ * <p>
+ * Transposes the inner (matrix) dimensions of a SparseMatrix and optionally
+ * conjugates its values.
+ */
+public final class SparseMatrixTranspose extends RawOp implements Operand<TType> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.linalg.sparse.SparseMatrixTranspose}
+   */
+  public static class Options {
+    
+    /**
+     * @param conjugate Indicates whether `input` should be conjugated.
+     */
+    public Options conjugate(Boolean conjugate) {
+      this.conjugate = conjugate;
+      return this;
+    }
+    
+    private Boolean conjugate;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new SparseMatrixTranspose operation.
+   * 
+   * @param scope current scope
+   * @param input A CSRSparseMatrix.
+   * @param type 
+   * @param options carries optional attributes values
+   * @return a new instance of SparseMatrixTranspose
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> SparseMatrixTranspose create(Scope scope, Operand<?> input, Class<T> type, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("SparseMatrixTranspose", scope.makeOpName("SparseMatrixTranspose"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("type", type);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.conjugate != null) {
+          opBuilder.setAttr("conjugate", opts.conjugate);
+        }
+      }
+    }
+    return new SparseMatrixTranspose(opBuilder.build());
+  }
+  
+  /**
+   * @param conjugate Indicates whether `input` should be conjugated.
+   */
+  public static Options conjugate(Boolean conjugate) {
+    return new Options().conjugate(conjugate);
+  }
+  
+  /**
+   * A CSRSparseMatrix.
+   */
+  public Output<?> output() {
+    return output;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "SparseMatrixTranspose";
+  
+  private Output<?> output;
+  
+  private SparseMatrixTranspose(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixZeros.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixZeros.java
new file mode 100644
index 00000000000..439c4ce69b7
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseMatrixZeros.java
@@ -0,0 +1,76 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.linalg.sparse;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Creates an all-zeros CSRSparseMatrix with shape `dense_shape`.
+ */
+public final class SparseMatrixZeros extends RawOp implements Operand<TType> {
+  
+  /**
+   * Factory method to create a class wrapping a new SparseMatrixZeros operation.
+   * 
+   * @param scope current scope
+   * @param denseShape The desired matrix shape.
+   * @param type 
+   * @return a new instance of SparseMatrixZeros
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> SparseMatrixZeros create(Scope scope, Operand<TInt64> denseShape, Class<T> type) {
+    OperationBuilder opBuilder = scope.env().opBuilder("SparseMatrixZeros", scope.makeOpName("SparseMatrixZeros"));
+    opBuilder.addInput(denseShape.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("type", type);
+    return new SparseMatrixZeros(opBuilder.build());
+  }
+  
+  /**
+   * An empty CSR matrix with shape `dense_shape`.
+   */
+  public Output<?> sparseMatrix() {
+    return sparseMatrix;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) sparseMatrix;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "SparseMatrixZeros";
+  
+  private Output<?> sparseMatrix;
+  
+  private SparseMatrixZeros(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    sparseMatrix = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseTensorToCSRSparseMatrix.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseTensorToCSRSparseMatrix.java
new file mode 100644
index 00000000000..afda1e8b5f1
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/linalg/sparse/SparseTensorToCSRSparseMatrix.java
@@ -0,0 +1,78 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.linalg.sparse;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Converts a SparseTensor to a (possibly batched) CSRSparseMatrix.
+ */
+public final class SparseTensorToCSRSparseMatrix extends RawOp implements Operand<TType> {
+  
+  /**
+   * Factory method to create a class wrapping a new SparseTensorToCSRSparseMatrix operation.
+   * 
+   * @param scope current scope
+   * @param indices SparseTensor indices.
+   * @param values SparseTensor values.
+   * @param denseShape SparseTensor dense shape.
+   * @return a new instance of SparseTensorToCSRSparseMatrix
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> SparseTensorToCSRSparseMatrix create(Scope scope, Operand<TInt64> indices, Operand<T> values, Operand<TInt64> denseShape) {
+    OperationBuilder opBuilder = scope.env().opBuilder("SparseTensorToCSRSparseMatrix", scope.makeOpName("SparseTensorToCSRSparseMatrix"));
+    opBuilder.addInput(indices.asOutput());
+    opBuilder.addInput(values.asOutput());
+    opBuilder.addInput(denseShape.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new SparseTensorToCSRSparseMatrix(opBuilder.build());
+  }
+  
+  /**
+   * A (possibly batched) CSRSparseMatrix.
+   */
+  public Output<?> sparseMatrix() {
+    return sparseMatrix;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) sparseMatrix;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "SparseTensorToCSRSparseMatrix";
+  
+  private Output<?> sparseMatrix;
+  
+  private SparseTensorToCSRSparseMatrix(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    sparseMatrix = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Abs.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Abs.java
new file mode 100644
index 00000000000..7fdc1717833
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Abs.java
@@ -0,0 +1,78 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.math;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Computes the absolute value of a tensor.
+ * <p>
+ * Given a tensor `x`, this operation returns a tensor containing the absolute
+ * value of each element in `x`. For example, if x is an input element and y is
+ * an output element, this operation computes \\(y = |x|\\).
+ * 
+ * @param <T> data type for {@code y()} output
+ */
+@Operator(group = "math")
+public final class Abs<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new Abs operation.
+   * 
+   * @param scope current scope
+   * @param x 
+   * @return a new instance of Abs
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> Abs<T> create(Scope scope, Operand<T> x) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Abs", scope.makeOpName("Abs"));
+    opBuilder.addInput(x.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new Abs<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> y() {
+    return y;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return y;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Abs";
+  
+  private Output<T> y;
+  
+  private Abs(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    y = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/AccumulateN.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/AccumulateN.java
new file mode 100644
index 00000000000..dc15d9f9d81
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/AccumulateN.java
@@ -0,0 +1,87 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.math;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.Operands;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Returns the element-wise sum of a list of tensors.
+ * <p>
+ * `tf.accumulate_n_v2` performs the same operation as `tf.add_n`, but does not
+ * wait for all of its inputs to be ready before beginning to sum. This can
+ * save memory if inputs are ready at different times, since minimum temporary
+ * storage is proportional to the output size rather than the inputs size.
+ * <p>
+ * Unlike the original `accumulate_n`, `accumulate_n_v2` is differentiable.
+ * <p>
+ * Returns a `Tensor` of same shape and type as the elements of `inputs`.
+ * 
+ * @param <T> data type for {@code sum()} output
+ */
+@Operator(group = "math")
+public final class AccumulateN<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new AccumulateN operation.
+   * 
+   * @param scope current scope
+   * @param inputs A list of `Tensor` objects, each with same shape and type.
+   * @param shape Shape of elements of `inputs`.
+   * @return a new instance of AccumulateN
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> AccumulateN<T> create(Scope scope, Iterable<Operand<T>> inputs, Shape shape) {
+    OperationBuilder opBuilder = scope.env().opBuilder("AccumulateNV2", scope.makeOpName("AccumulateN"));
+    opBuilder.addInputList(Operands.asOutputs(inputs));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("shape", shape);
+    return new AccumulateN<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> sum() {
+    return sum;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return sum;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "AccumulateNV2";
+  
+  private Output<T> sum;
+  
+  private AccumulateN(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    sum = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Acos.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Acos.java
new file mode 100644
index 00000000000..43ba8e54334
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Acos.java
@@ -0,0 +1,74 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.math;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Computes acos of x element-wise.
+ * 
+ * @param <T> data type for {@code y()} output
+ */
+@Operator(group = "math")
+public final class Acos<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new Acos operation.
+   * 
+   * @param scope current scope
+   * @param x 
+   * @return a new instance of Acos
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> Acos<T> create(Scope scope, Operand<T> x) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Acos", scope.makeOpName("Acos"));
+    opBuilder.addInput(x.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new Acos<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> y() {
+    return y;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return y;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Acos";
+  
+  private Output<T> y;
+  
+  private Acos(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    y = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Acosh.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Acosh.java
new file mode 100644
index 00000000000..0d1aaf95237
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Acosh.java
@@ -0,0 +1,82 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.math;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Computes inverse hyperbolic cosine of x element-wise.
+ * <p>
+ * Given an input tensor, the function computes inverse hyperbolic cosine of every element.
+ * Input range is `[1, inf]`. It returns `nan` if the input lies outside the range.
+ * <pre>{@code
+ * x = tf.constant([-2, -0.5, 1, 1.2, 200, 10000, float("inf")])
+ * tf.math.acosh(x) ==> [nan nan 0. 0.62236255 5.9914584 9.903487 inf]
+ * }</pre>
+ * 
+ * 
+ * @param <T> data type for {@code y()} output
+ */
+@Operator(group = "math")
+public final class Acosh<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new Acosh operation.
+   * 
+   * @param scope current scope
+   * @param x 
+   * @return a new instance of Acosh
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> Acosh<T> create(Scope scope, Operand<T> x) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Acosh", scope.makeOpName("Acosh"));
+    opBuilder.addInput(x.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new Acosh<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> y() {
+    return y;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return y;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Acosh";
+  
+  private Output<T> y;
+  
+  private Acosh(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    y = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Add.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Add.java
new file mode 100644
index 00000000000..86d7c82ae38
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Add.java
@@ -0,0 +1,79 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.math;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Returns x + y element-wise.
+ * <p>
+ * <i>NOTE</i>: `math.Add` supports broadcasting. `AddN` does not. More about broadcasting
+ * [here](http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html)
+ * 
+ * @param <T> data type for {@code z()} output
+ */
+@Operator(group = "math")
+public final class Add<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new Add operation.
+   * 
+   * @param scope current scope
+   * @param x 
+   * @param y 
+   * @return a new instance of Add
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> Add<T> create(Scope scope, Operand<T> x, Operand<T> y) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Add", scope.makeOpName("Add"));
+    opBuilder.addInput(x.asOutput());
+    opBuilder.addInput(y.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new Add<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> z() {
+    return z;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return z;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Add";
+  
+  private Output<T> z;
+  
+  private Add(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    z = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/AddN.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/AddN.java
new file mode 100644
index 00000000000..16a5e22ce16
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/AddN.java
@@ -0,0 +1,83 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.math;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.Operands;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Add all input tensors element wise.
+ * <p>
+ *   Inputs must be of same size and shape.
+ * <p>
+ *   <pre>{@code
+ *   x = [9, 7, 10]
+ *   tf.math.add_n(x) ==> 26
+ *   }</pre>
+ * 
+ * 
+ * @param <T> data type for {@code sum()} output
+ */
+@Operator(group = "math")
+public final class AddN<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new AddN operation.
+   * 
+   * @param scope current scope
+   * @param inputs 
+   * @return a new instance of AddN
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> AddN<T> create(Scope scope, Iterable<Operand<T>> inputs) {
+    OperationBuilder opBuilder = scope.env().opBuilder("AddN", scope.makeOpName("AddN"));
+    opBuilder.addInputList(Operands.asOutputs(inputs));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new AddN<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> sum() {
+    return sum;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return sum;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "AddN";
+  
+  private Output<T> sum;
+  
+  private AddN(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    sum = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Angle.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Angle.java
new file mode 100644
index 00000000000..097a29c35ff
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Angle.java
@@ -0,0 +1,106 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.math;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Returns the argument of a complex number.
+ * <p>
+ * Given a tensor `input` of complex numbers, this operation returns a tensor of
+ * type `float` that is the argument of each element in `input`. All elements in
+ * `input` must be complex numbers of the form \\(a + bj\\), where <i>a</i>
+ * is the real part and <i>b</i> is the imaginary part.
+ * <p>
+ * The argument returned by this operation is of the form \\(atan2(b, a)\\).
+ * <p>
+ * For example:
+ * <pre>{@code
+ * # tensor 'input' is [-2.25 + 4.75j, 3.25 + 5.75j]
+ * tf.angle(input) ==> [2.0132, 1.056]
+ * }</pre>
+ * @compatibility(numpy)
+ * Equivalent to np.angle.
+ * @end_compatibility
+ * 
+ * @param <U> data type for {@code output()} output
+ */
+@Operator(group = "math")
+public final class Angle<U extends TNumber> extends RawOp implements Operand<U> {
+  
+  /**
+   * Factory method to create a class wrapping a new Angle operation.
+   * 
+   * @param scope current scope
+   * @param input 
+   * @param Tout 
+   * @return a new instance of Angle
+   */
+  @Endpoint(describeByClass = true)
+  public static <U extends TNumber, T extends TType> Angle<U> create(Scope scope, Operand<T> input, Class<U> Tout) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Angle", scope.makeOpName("Angle"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("Tout", Tout);
+    return new Angle<U>(opBuilder.build());
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new Angle operation using default output types.
+   * 
+   * @param scope current scope
+   * @param input 
+   * @return a new instance of Angle
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> Angle<TFloat32> create(Scope scope, Operand<T> input) {
+    return create(scope, input, TFloat32.class);
+  }
+  
+  /**
+   */
+  public Output<U> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<U> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Angle";
+  
+  private Output<U> output;
+  
+  private Angle(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/ApproximateEqual.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/ApproximateEqual.java
new file mode 100644
index 00000000000..245c42f269d
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/ApproximateEqual.java
@@ -0,0 +1,109 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.math;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TBool;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Returns the truth value of abs(x-y) < tolerance element-wise.
+ */
+@Operator(group = "math")
+public final class ApproximateEqual extends RawOp implements Operand<TBool> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.math.ApproximateEqual}
+   */
+  public static class Options {
+    
+    /**
+     * @param tolerance 
+     */
+    public Options tolerance(Float tolerance) {
+      this.tolerance = tolerance;
+      return this;
+    }
+    
+    private Float tolerance;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new ApproximateEqual operation.
+   * 
+   * @param scope current scope
+   * @param x 
+   * @param y 
+   * @param options carries optional attributes values
+   * @return a new instance of ApproximateEqual
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> ApproximateEqual create(Scope scope, Operand<T> x, Operand<T> y, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ApproximateEqual", scope.makeOpName("ApproximateEqual"));
+    opBuilder.addInput(x.asOutput());
+    opBuilder.addInput(y.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.tolerance != null) {
+          opBuilder.setAttr("tolerance", opts.tolerance);
+        }
+      }
+    }
+    return new ApproximateEqual(opBuilder.build());
+  }
+  
+  /**
+   * @param tolerance 
+   */
+  public static Options tolerance(Float tolerance) {
+    return new Options().tolerance(tolerance);
+  }
+  
+  /**
+   */
+  public Output<TBool> z() {
+    return z;
+  }
+  
+  @Override
+  public Output<TBool> asOutput() {
+    return z;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ApproximateEqual";
+  
+  private Output<TBool> z;
+  
+  private ApproximateEqual(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    z = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/ArgMax.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/ArgMax.java
new file mode 100644
index 00000000000..42eb6778d24
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/ArgMax.java
@@ -0,0 +1,110 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.math;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Returns the index with the largest value across dimensions of a tensor.
+ * <p>
+ * Note that in case of ties the identity of the return value is not guaranteed.
+ * <p>
+ * Usage:
+ *   <pre>{@code
+ *   import tensorflow as tf
+ *   a = [1, 10, 26.9, 2.8, 166.32, 62.3]
+ *   b = tf.math.argmax(input = a)
+ *   c = tf.keras.backend.eval(b)
+ *   # c = 4
+ *   # here a[4] = 166.32 which is the largest element of a across axis 0
+ *   }</pre>
+ * 
+ * 
+ * @param <V> data type for {@code output()} output
+ */
+@Operator(group = "math")
+public final class ArgMax<V extends TNumber> extends RawOp implements Operand<V> {
+  
+  /**
+   * Factory method to create a class wrapping a new ArgMax operation.
+   * 
+   * @param scope current scope
+   * @param input 
+   * @param dimension int32 or int64, must be in the range `[-rank(input), rank(input))`.
+   * Describes which dimension of the input Tensor to reduce across. For vectors,
+   * use dimension = 0.
+   * @param outputType 
+   * @return a new instance of ArgMax
+   */
+  @Endpoint(describeByClass = true)
+  public static <V extends TNumber, T extends TType, U extends TNumber> ArgMax<V> create(Scope scope, Operand<T> input, Operand<U> dimension, Class<V> outputType) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ArgMax", scope.makeOpName("ArgMax"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(dimension.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("output_type", outputType);
+    return new ArgMax<V>(opBuilder.build());
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new ArgMax operation using default output types.
+   * 
+   * @param scope current scope
+   * @param input 
+   * @param dimension int32 or int64, must be in the range `[-rank(input), rank(input))`.
+   * Describes which dimension of the input Tensor to reduce across. For vectors,
+   * use dimension = 0.
+   * @return a new instance of ArgMax
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType, U extends TNumber> ArgMax<TInt64> create(Scope scope, Operand<T> input, Operand<U> dimension) {
+    return create(scope, input, dimension, TInt64.class);
+  }
+  
+  /**
+   */
+  public Output<V> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<V> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ArgMax";
+  
+  private Output<V> output;
+  
+  private ArgMax(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/ArgMin.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/ArgMin.java
new file mode 100644
index 00000000000..bd6b69d52e1
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/ArgMin.java
@@ -0,0 +1,110 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.math;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Returns the index with the smallest value across dimensions of a tensor.
+ * <p>
+ * Note that in case of ties the identity of the return value is not guaranteed.
+ * <p>
+ * Usage:
+ *   <pre>{@code
+ *   import tensorflow as tf
+ *   a = [1, 10, 26.9, 2.8, 166.32, 62.3]
+ *   b = tf.math.argmin(input = a)
+ *   c = tf.keras.backend.eval(b)
+ *   # c = 0
+ *   # here a[0] = 1 which is the smallest element of a across axis 0
+ *   }</pre>
+ * 
+ * 
+ * @param <V> data type for {@code output()} output
+ */
+@Operator(group = "math")
+public final class ArgMin<V extends TNumber> extends RawOp implements Operand<V> {
+  
+  /**
+   * Factory method to create a class wrapping a new ArgMin operation.
+   * 
+   * @param scope current scope
+   * @param input 
+   * @param dimension int32 or int64, must be in the range `[-rank(input), rank(input))`.
+   * Describes which dimension of the input Tensor to reduce across. For vectors,
+   * use dimension = 0.
+   * @param outputType 
+   * @return a new instance of ArgMin
+   */
+  @Endpoint(describeByClass = true)
+  public static <V extends TNumber, T extends TType, U extends TNumber> ArgMin<V> create(Scope scope, Operand<T> input, Operand<U> dimension, Class<V> outputType) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ArgMin", scope.makeOpName("ArgMin"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(dimension.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("output_type", outputType);
+    return new ArgMin<V>(opBuilder.build());
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new ArgMin operation using default output types.
+   * 
+   * @param scope current scope
+   * @param input 
+   * @param dimension int32 or int64, must be in the range `[-rank(input), rank(input))`.
+   * Describes which dimension of the input Tensor to reduce across. For vectors,
+   * use dimension = 0.
+   * @return a new instance of ArgMin
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType, U extends TNumber> ArgMin<TInt64> create(Scope scope, Operand<T> input, Operand<U> dimension) {
+    return create(scope, input, dimension, TInt64.class);
+  }
+  
+  /**
+   */
+  public Output<V> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<V> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ArgMin";
+  
+  private Output<V> output;
+  
+  private ArgMin(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Asin.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Asin.java
new file mode 100644
index 00000000000..941515b013e
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Asin.java
@@ -0,0 +1,90 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.math;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Computes the trignometric inverse sine of x element-wise.
+ * <p>
+ * The `tf.math.asin` operation returns the inverse of `tf.math.sin`, such that
+ * if `y = tf.math.sin(x)` then, `x = tf.math.asin(y)`.
+ * <p>
+ * <b>Note</b>: The output of `tf.math.asin` will lie within the invertible range
+ * of sine, i.e [-pi/2, pi/2].
+ * <p>
+ * For example:
+ * <pre>{@code
+ * # Note: [1.047, 0.785] ~= [(pi/3), (pi/4)]
+ * x = tf.constant([1.047, 0.785])
+ * y = tf.math.sin(x) # [0.8659266, 0.7068252]
+ * 
+ * tf.math.asin(y) # [1.047, 0.785] = x
+ * }</pre>
+ * 
+ * 
+ * @param <T> data type for {@code y()} output
+ */
+@Operator(group = "math")
+public final class Asin<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new Asin operation.
+   * 
+   * @param scope current scope
+   * @param x 
+   * @return a new instance of Asin
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> Asin<T> create(Scope scope, Operand<T> x) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Asin", scope.makeOpName("Asin"));
+    opBuilder.addInput(x.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new Asin<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> y() {
+    return y;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return y;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Asin";
+  
+  private Output<T> y;
+  
+  private Asin(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    y = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Asinh.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Asinh.java
new file mode 100644
index 00000000000..c2451727573
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Asinh.java
@@ -0,0 +1,84 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.math;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Computes inverse hyperbolic sine of x element-wise.
+ * <p>
+ *   Given an input tensor, this function computes inverse hyperbolic sine
+ *   for every element in the tensor. Both input and output has a range of
+ *   `[-inf, inf]`.
+ * <p>
+ *   <pre>{@code
+ *   x = tf.constant([-float("inf"), -2, -0.5, 1, 1.2, 200, 10000, float("inf")])
+ *   tf.math.asinh(x) ==> [-inf -1.4436355 -0.4812118 0.8813736 1.0159732 5.991471 9.903487 inf]
+ *   }</pre>
+ * 
+ * 
+ * @param <T> data type for {@code y()} output
+ */
+@Operator(group = "math")
+public final class Asinh<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new Asinh operation.
+   * 
+   * @param scope current scope
+   * @param x 
+   * @return a new instance of Asinh
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> Asinh<T> create(Scope scope, Operand<T> x) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Asinh", scope.makeOpName("Asinh"));
+    opBuilder.addInput(x.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new Asinh<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> y() {
+    return y;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return y;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Asinh";
+  
+  private Output<T> y;
+  
+  private Asinh(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    y = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Atan.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Atan.java
new file mode 100644
index 00000000000..a3574d549bc
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Atan.java
@@ -0,0 +1,90 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.math;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Computes the trignometric inverse tangent of x element-wise.
+ * <p>
+ * The `tf.math.atan` operation returns the inverse of `tf.math.tan`, such that
+ * if `y = tf.math.tan(x)` then, `x = tf.math.atan(y)`.
+ * <p>
+ * <b>Note</b>: The output of `tf.math.atan` will lie within the invertible range
+ * of tan, i.e (-pi/2, pi/2).
+ * <p>
+ * For example:
+ * <pre>{@code
+ * # Note: [1.047, 0.785] ~= [(pi/3), (pi/4)]
+ * x = tf.constant([1.047, 0.785])
+ * y = tf.math.tan(x) # [1.731261, 0.99920404]
+ * 
+ * tf.math.atan(y) # [1.047, 0.785] = x
+ * }</pre>
+ * 
+ * 
+ * @param <T> data type for {@code y()} output
+ */
+@Operator(group = "math")
+public final class Atan<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new Atan operation.
+   * 
+   * @param scope current scope
+   * @param x 
+   * @return a new instance of Atan
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> Atan<T> create(Scope scope, Operand<T> x) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Atan", scope.makeOpName("Atan"));
+    opBuilder.addInput(x.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new Atan<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> y() {
+    return y;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return y;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Atan";
+  
+  private Output<T> y;
+  
+  private Atan(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    y = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Atan2.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Atan2.java
new file mode 100644
index 00000000000..a70043da3a3
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Atan2.java
@@ -0,0 +1,82 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.math;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Computes arctangent of `y/x` element-wise, respecting signs of the arguments.
+ * <p>
+ * This is the angle \( \theta \in [-\pi, \pi] \) such that
+ * \[ x = r \cos(\theta) \]
+ * and
+ * \[ y = r \sin(\theta) \]
+ * where \(r = \sqrt(x^2 + y^2) \).
+ * 
+ * @param <T> data type for {@code z()} output
+ */
+@Operator(group = "math")
+public final class Atan2<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new Atan2 operation.
+   * 
+   * @param scope current scope
+   * @param y 
+   * @param x 
+   * @return a new instance of Atan2
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> Atan2<T> create(Scope scope, Operand<T> y, Operand<T> x) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Atan2", scope.makeOpName("Atan2"));
+    opBuilder.addInput(y.asOutput());
+    opBuilder.addInput(x.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new Atan2<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> z() {
+    return z;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return z;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Atan2";
+  
+  private Output<T> z;
+  
+  private Atan2(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    z = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Atanh.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Atanh.java
new file mode 100644
index 00000000000..6c53c8828c0
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Atanh.java
@@ -0,0 +1,86 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.math;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Computes inverse hyperbolic tangent of x element-wise.
+ * <p>
+ *   Given an input tensor, this function computes inverse hyperbolic tangent
+ *   for every element in the tensor. Input range is `[-1,1]` and output range is
+ *   `[-inf, inf]`. If input is `-1`, output will be `-inf` and if the
+ *   input is `1`, output will be `inf`. Values outside the range will have
+ *   `nan` as output.
+ * <p>
+ *   <pre>{@code
+ *   x = tf.constant([-float("inf"), -1, -0.5, 1, 0, 0.5, 10, float("inf")])
+ *   tf.math.atanh(x) ==> [nan -inf -0.54930615 inf  0. 0.54930615 nan nan]
+ *   }</pre>
+ * 
+ * 
+ * @param <T> data type for {@code y()} output
+ */
+@Operator(group = "math")
+public final class Atanh<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new Atanh operation.
+   * 
+   * @param scope current scope
+   * @param x 
+   * @return a new instance of Atanh
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> Atanh<T> create(Scope scope, Operand<T> x) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Atanh", scope.makeOpName("Atanh"));
+    opBuilder.addInput(x.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new Atanh<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> y() {
+    return y;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return y;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Atanh";
+  
+  private Output<T> y;
+  
+  private Atanh(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    y = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/BesselI0.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/BesselI0.java
new file mode 100644
index 00000000000..ecf7aab945c
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/BesselI0.java
@@ -0,0 +1,71 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.math;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * @param <T> data type for {@code y()} output
+ */
+public final class BesselI0<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new BesselI0 operation.
+   * 
+   * @param scope current scope
+   * @param x 
+   * @return a new instance of BesselI0
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> BesselI0<T> create(Scope scope, Operand<T> x) {
+    OperationBuilder opBuilder = scope.env().opBuilder("BesselI0", scope.makeOpName("BesselI0"));
+    opBuilder.addInput(x.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new BesselI0<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> y() {
+    return y;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return y;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "BesselI0";
+  
+  private Output<T> y;
+  
+  private BesselI0(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    y = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/BesselI0e.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/BesselI0e.java
new file mode 100644
index 00000000000..d274265faf0
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/BesselI0e.java
@@ -0,0 +1,71 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.math;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * @param <T> data type for {@code y()} output
+ */
+public final class BesselI0e<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new BesselI0e operation.
+   * 
+   * @param scope current scope
+   * @param x 
+   * @return a new instance of BesselI0e
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> BesselI0e<T> create(Scope scope, Operand<T> x) {
+    OperationBuilder opBuilder = scope.env().opBuilder("BesselI0e", scope.makeOpName("BesselI0e"));
+    opBuilder.addInput(x.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new BesselI0e<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> y() {
+    return y;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return y;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "BesselI0e";
+  
+  private Output<T> y;
+  
+  private BesselI0e(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    y = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/BesselI1.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/BesselI1.java
new file mode 100644
index 00000000000..b836404991e
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/BesselI1.java
@@ -0,0 +1,71 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.math;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * @param <T> data type for {@code y()} output
+ */
+public final class BesselI1<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new BesselI1 operation.
+   * 
+   * @param scope current scope
+   * @param x 
+   * @return a new instance of BesselI1
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> BesselI1<T> create(Scope scope, Operand<T> x) {
+    OperationBuilder opBuilder = scope.env().opBuilder("BesselI1", scope.makeOpName("BesselI1"));
+    opBuilder.addInput(x.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new BesselI1<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> y() {
+    return y;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return y;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "BesselI1";
+  
+  private Output<T> y;
+  
+  private BesselI1(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    y = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/BesselI1e.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/BesselI1e.java
new file mode 100644
index 00000000000..f8d40882585
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/BesselI1e.java
@@ -0,0 +1,71 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.math;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * @param <T> data type for {@code y()} output
+ */
+public final class BesselI1e<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new BesselI1e operation.
+   * 
+   * @param scope current scope
+   * @param x 
+   * @return a new instance of BesselI1e
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> BesselI1e<T> create(Scope scope, Operand<T> x) {
+    OperationBuilder opBuilder = scope.env().opBuilder("BesselI1e", scope.makeOpName("BesselI1e"));
+    opBuilder.addInput(x.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new BesselI1e<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> y() {
+    return y;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return y;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "BesselI1e";
+  
+  private Output<T> y;
+  
+  private BesselI1e(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    y = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Betainc.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Betainc.java
new file mode 100644
index 00000000000..e28da25745f
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Betainc.java
@@ -0,0 +1,89 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.math;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Compute the regularized incomplete beta integral \\(I_x(a, b)\\).
+ * <p>
+ * The regularized incomplete beta integral is defined as:
+ * <p>
+ * \\(I_x(a, b) = \frac{B(x; a, b)}{B(a, b)}\\)
+ * <p>
+ * where
+ * <p>
+ * \\(B(x; a, b) = \int_0^x t^{a-1} (1 - t)^{b-1} dt\\)
+ * <p>
+ * is the incomplete beta function and \\(B(a, b)\\) is the <i>complete</i>
+ * beta function.
+ * 
+ * @param <T> data type for {@code z()} output
+ */
+@Operator(group = "math")
+public final class Betainc<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new Betainc operation.
+   * 
+   * @param scope current scope
+   * @param a 
+   * @param b 
+   * @param x 
+   * @return a new instance of Betainc
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> Betainc<T> create(Scope scope, Operand<T> a, Operand<T> b, Operand<T> x) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Betainc", scope.makeOpName("Betainc"));
+    opBuilder.addInput(a.asOutput());
+    opBuilder.addInput(b.asOutput());
+    opBuilder.addInput(x.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new Betainc<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> z() {
+    return z;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return z;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Betainc";
+  
+  private Output<T> z;
+  
+  private Betainc(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    z = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Bincount.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Bincount.java
new file mode 100644
index 00000000000..565a006c3b4
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Bincount.java
@@ -0,0 +1,91 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.math;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Counts the number of occurrences of each value in an integer array.
+ * <p>
+ * Outputs a vector with length `size` and the same dtype as `weights`. If
+ * `weights` are empty, then index `i` stores the number of times the value `i` is
+ * counted in `arr`. If `weights` are non-empty, then index `i` stores the sum of
+ * the value in `weights` at each index where the corresponding value in `arr` is
+ * `i`.
+ * <p>
+ * Values in `arr` outside of the range [0, size) are ignored.
+ * 
+ * @param <T> data type for {@code bins()} output
+ */
+@Operator(group = "math")
+public final class Bincount<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new Bincount operation.
+   * 
+   * @param scope current scope
+   * @param arr int32 `Tensor`.
+   * @param size non-negative int32 scalar `Tensor`.
+   * @param weights is an int32, int64, float32, or float64 `Tensor` with the same
+   * shape as `arr`, or a length-0 `Tensor`, in which case it acts as all weights
+   * equal to 1.
+   * @return a new instance of Bincount
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> Bincount<T> create(Scope scope, Operand<TInt32> arr, Operand<TInt32> size, Operand<T> weights) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Bincount", scope.makeOpName("Bincount"));
+    opBuilder.addInput(arr.asOutput());
+    opBuilder.addInput(size.asOutput());
+    opBuilder.addInput(weights.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new Bincount<T>(opBuilder.build());
+  }
+  
+  /**
+   * 1D `Tensor` with length equal to `size`. The counts or summed weights for
+   * each value in the range [0, size).
+   */
+  public Output<T> bins() {
+    return bins;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return bins;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Bincount";
+  
+  private Output<T> bins;
+  
+  private Bincount(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    bins = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Ceil.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Ceil.java
new file mode 100644
index 00000000000..bda7247d508
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Ceil.java
@@ -0,0 +1,74 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.math;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Returns element-wise smallest integer not less than x.
+ * 
+ * @param <T> data type for {@code y()} output
+ */
+@Operator(group = "math")
+public final class Ceil<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new Ceil operation.
+   * 
+   * @param scope current scope
+   * @param x 
+   * @return a new instance of Ceil
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> Ceil<T> create(Scope scope, Operand<T> x) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Ceil", scope.makeOpName("Ceil"));
+    opBuilder.addInput(x.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new Ceil<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> y() {
+    return y;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return y;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Ceil";
+  
+  private Output<T> y;
+  
+  private Ceil(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    y = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/CompareAndBitpack.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/CompareAndBitpack.java
new file mode 100644
index 00000000000..2538cc59248
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/CompareAndBitpack.java
@@ -0,0 +1,98 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.math;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TUint8;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Compare values of `input` to `threshold` and pack resulting bits into a `uint8`.
+ * <p>
+ * Each comparison returns a boolean `true` (if `input_value > threshold`)
+ * or and `false` otherwise.
+ * <p>
+ * This operation is useful for Locality-Sensitive-Hashing (LSH) and other
+ * algorithms that use hashing approximations of cosine and `L2` distances;
+ * codes can be generated from an input via:
+ * <pre>{@code
+ * codebook_size = 50
+ * codebook_bits = codebook_size * 32
+ * codebook = tf.get_variable('codebook', [x.shape[-1].value, codebook_bits],
+ *                            dtype=x.dtype,
+ *                            initializer=tf.orthogonal_initializer())
+ * codes = compare_and_threshold(tf.matmul(x, codebook), threshold=0.)
+ * codes = tf.bitcast(codes, tf.int32)  # go from uint8 to int32
+ * # now codes has shape x.shape[:-1] + [codebook_size]
+ * }</pre>
+ * <b>NOTE</b>: Currently, the innermost dimension of the tensor must be divisible
+ * by 8.
+ * <p>
+ * Given an `input` shaped `[s0, s1, ..., s_n]`, the output is
+ * a `uint8` tensor shaped `[s0, s1, ..., s_n / 8]`.
+ */
+@Operator(group = "math")
+public final class CompareAndBitpack extends RawOp implements Operand<TUint8> {
+  
+  /**
+   * Factory method to create a class wrapping a new CompareAndBitpack operation.
+   * 
+   * @param scope current scope
+   * @param input Values to compare against `threshold` and bitpack.
+   * @param threshold Threshold to compare against.
+   * @return a new instance of CompareAndBitpack
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> CompareAndBitpack create(Scope scope, Operand<T> input, Operand<T> threshold) {
+    OperationBuilder opBuilder = scope.env().opBuilder("CompareAndBitpack", scope.makeOpName("CompareAndBitpack"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(threshold.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new CompareAndBitpack(opBuilder.build());
+  }
+  
+  /**
+   * The bitpacked comparisons.
+   */
+  public Output<TUint8> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<TUint8> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "CompareAndBitpack";
+  
+  private Output<TUint8> output;
+  
+  private CompareAndBitpack(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/ComplexAbs.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/ComplexAbs.java
new file mode 100644
index 00000000000..b899c47845a
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/ComplexAbs.java
@@ -0,0 +1,95 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.math;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Computes the complex absolute value of a tensor.
+ * <p>
+ * Given a tensor `x` of complex numbers, this operation returns a tensor of type
+ * `float` or `double` that is the absolute value of each element in `x`. All
+ * elements in `x` must be complex numbers of the form \\(a + bj\\). The absolute
+ * value is computed as \\( \sqrt{a^2 + b^2}\\).
+ * 
+ * @param <U> data type for {@code y()} output
+ */
+@Operator(group = "math")
+public final class ComplexAbs<U extends TNumber> extends RawOp implements Operand<U> {
+  
+  /**
+   * Factory method to create a class wrapping a new ComplexAbs operation.
+   * 
+   * @param scope current scope
+   * @param x 
+   * @param Tout 
+   * @return a new instance of ComplexAbs
+   */
+  @Endpoint(describeByClass = true)
+  public static <U extends TNumber, T extends TType> ComplexAbs<U> create(Scope scope, Operand<T> x, Class<U> Tout) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ComplexAbs", scope.makeOpName("ComplexAbs"));
+    opBuilder.addInput(x.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("Tout", Tout);
+    return new ComplexAbs<U>(opBuilder.build());
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new ComplexAbs operation using default output types.
+   * 
+   * @param scope current scope
+   * @param x 
+   * @return a new instance of ComplexAbs
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> ComplexAbs<TFloat32> create(Scope scope, Operand<T> x) {
+    return create(scope, x, TFloat32.class);
+  }
+  
+  /**
+   */
+  public Output<U> y() {
+    return y;
+  }
+  
+  @Override
+  public Output<U> asOutput() {
+    return y;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ComplexAbs";
+  
+  private Output<U> y;
+  
+  private ComplexAbs(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    y = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Conj.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Conj.java
new file mode 100644
index 00000000000..b565fe9d322
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Conj.java
@@ -0,0 +1,88 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.math;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Returns the complex conjugate of a complex number.
+ * <p>
+ * Given a tensor `input` of complex numbers, this operation returns a tensor of
+ * complex numbers that are the complex conjugate of each element in `input`. The
+ * complex numbers in `input` must be of the form \\(a + bj\\), where <i>a</i> is the
+ * real part and <i>b</i> is the imaginary part.
+ * <p>
+ * The complex conjugate returned by this operation is of the form \\(a - bj\\).
+ * <p>
+ * For example:
+ * <pre>{@code
+ * # tensor 'input' is [-2.25 + 4.75j, 3.25 + 5.75j]
+ * tf.conj(input) ==> [-2.25 - 4.75j, 3.25 - 5.75j]
+ * }</pre>
+ * 
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator(group = "math")
+public final class Conj<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new Conj operation.
+   * 
+   * @param scope current scope
+   * @param input 
+   * @return a new instance of Conj
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> Conj<T> create(Scope scope, Operand<T> input) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Conj", scope.makeOpName("Conj"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new Conj<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Conj";
+  
+  private Output<T> output;
+  
+  private Conj(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Cos.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Cos.java
new file mode 100644
index 00000000000..29f81972b17
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Cos.java
@@ -0,0 +1,85 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.math;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Computes cos of x element-wise.
+ * <p>
+ *   Given an input tensor, this function computes cosine of every
+ *   element in the tensor. Input range is `(-inf, inf)` and
+ *   output range is `[-1,1]`. If input lies outside the boundary, `nan`
+ *   is returned.
+ * <p>
+ *   <pre>{@code
+ *   x = tf.constant([-float("inf"), -9, -0.5, 1, 1.2, 200, 10000, float("inf")])
+ *   tf.math.cos(x) ==> [nan -0.91113025 0.87758255 0.5403023 0.36235774 0.48718765 -0.95215535 nan]
+ *   }</pre>
+ * 
+ * 
+ * @param <T> data type for {@code y()} output
+ */
+@Operator(group = "math")
+public final class Cos<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new Cos operation.
+   * 
+   * @param scope current scope
+   * @param x 
+   * @return a new instance of Cos
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> Cos<T> create(Scope scope, Operand<T> x) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Cos", scope.makeOpName("Cos"));
+    opBuilder.addInput(x.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new Cos<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> y() {
+    return y;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return y;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Cos";
+  
+  private Output<T> y;
+  
+  private Cos(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    y = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Cosh.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Cosh.java
new file mode 100644
index 00000000000..f65e3e952ea
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Cosh.java
@@ -0,0 +1,84 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.math;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Computes hyperbolic cosine of x element-wise.
+ * <p>
+ *   Given an input tensor, this function computes hyperbolic cosine of every
+ *   element in the tensor. Input range is `[-inf, inf]` and output range
+ *   is `[1, inf]`.
+ * <p>
+ *   <pre>{@code
+ *   x = tf.constant([-float("inf"), -9, -0.5, 1, 1.2, 2, 10, float("inf")])
+ *   tf.math.cosh(x) ==> [inf 4.0515420e+03 1.1276259e+00 1.5430807e+00 1.8106556e+00 3.7621956e+00 1.1013233e+04 inf]
+ *   }</pre>
+ * 
+ * 
+ * @param <T> data type for {@code y()} output
+ */
+@Operator(group = "math")
+public final class Cosh<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new Cosh operation.
+   * 
+   * @param scope current scope
+   * @param x 
+   * @return a new instance of Cosh
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> Cosh<T> create(Scope scope, Operand<T> x) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Cosh", scope.makeOpName("Cosh"));
+    opBuilder.addInput(x.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new Cosh<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> y() {
+    return y;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return y;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Cosh";
+  
+  private Output<T> y;
+  
+  private Cosh(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    y = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Cumprod.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Cumprod.java
new file mode 100644
index 00000000000..52154403d72
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Cumprod.java
@@ -0,0 +1,156 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.math;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Compute the cumulative product of the tensor `x` along `axis`.
+ * <p>
+ * By default, this op performs an inclusive cumprod, which means that the first
+ * element of the input is identical to the first element of the output:
+ * <pre>{@code
+ * tf.cumprod([a, b, c])  # => [a, a * b, a * b * c]
+ * }</pre>
+ * By setting the `exclusive` kwarg to `True`, an exclusive cumprod is
+ * performed instead:
+ * <pre>{@code
+ * tf.cumprod([a, b, c], exclusive=True)  # => [1, a, a * b]
+ * }</pre>
+ * By setting the `reverse` kwarg to `True`, the cumprod is performed in the
+ * opposite direction:
+ * <pre>{@code
+ * tf.cumprod([a, b, c], reverse=True)  # => [a * b * c, b * c, c]
+ * }</pre>
+ * This is more efficient than using separate `tf.reverse` ops.
+ * <p>
+ * The `reverse` and `exclusive` kwargs can also be combined:
+ * <pre>{@code
+ * tf.cumprod([a, b, c], exclusive=True, reverse=True)  # => [b * c, c, 1]
+ * }</pre>
+ * 
+ * 
+ * @param <T> data type for {@code out()} output
+ */
+@Operator(group = "math")
+public final class Cumprod<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.math.Cumprod}
+   */
+  public static class Options {
+    
+    /**
+     * @param exclusive If `True`, perform exclusive cumprod.
+     */
+    public Options exclusive(Boolean exclusive) {
+      this.exclusive = exclusive;
+      return this;
+    }
+    
+    /**
+     * @param reverse A `bool` (default: False).
+     */
+    public Options reverse(Boolean reverse) {
+      this.reverse = reverse;
+      return this;
+    }
+    
+    private Boolean exclusive;
+    private Boolean reverse;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new Cumprod operation.
+   * 
+   * @param scope current scope
+   * @param x A `Tensor`. Must be one of the following types: `float32`, `float64`,
+   * `int64`, `int32`, `uint8`, `uint16`, `int16`, `int8`, `complex64`,
+   * `complex128`, `qint8`, `quint8`, `qint32`, `half`.
+   * @param axis A `Tensor` of type `int32` (default: 0). Must be in the range
+   * `[-rank(x), rank(x))`.
+   * @param options carries optional attributes values
+   * @return a new instance of Cumprod
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType, U extends TNumber> Cumprod<T> create(Scope scope, Operand<T> x, Operand<U> axis, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Cumprod", scope.makeOpName("Cumprod"));
+    opBuilder.addInput(x.asOutput());
+    opBuilder.addInput(axis.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.exclusive != null) {
+          opBuilder.setAttr("exclusive", opts.exclusive);
+        }
+        if (opts.reverse != null) {
+          opBuilder.setAttr("reverse", opts.reverse);
+        }
+      }
+    }
+    return new Cumprod<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param exclusive If `True`, perform exclusive cumprod.
+   */
+  public static Options exclusive(Boolean exclusive) {
+    return new Options().exclusive(exclusive);
+  }
+  
+  /**
+   * @param reverse A `bool` (default: False).
+   */
+  public static Options reverse(Boolean reverse) {
+    return new Options().reverse(reverse);
+  }
+  
+  /**
+   */
+  public Output<T> out() {
+    return out;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return out;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Cumprod";
+  
+  private Output<T> out;
+  
+  private Cumprod(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    out = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Cumsum.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Cumsum.java
new file mode 100644
index 00000000000..6895cbed39f
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Cumsum.java
@@ -0,0 +1,156 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.math;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Compute the cumulative sum of the tensor `x` along `axis`.
+ * <p>
+ * By default, this op performs an inclusive cumsum, which means that the first
+ * element of the input is identical to the first element of the output:
+ * <pre>{@code
+ * tf.cumsum([a, b, c])  # => [a, a + b, a + b + c]
+ * }</pre>
+ * By setting the `exclusive` kwarg to `True`, an exclusive cumsum is
+ * performed instead:
+ * <pre>{@code
+ * tf.cumsum([a, b, c], exclusive=True)  # => [0, a, a + b]
+ * }</pre>
+ * By setting the `reverse` kwarg to `True`, the cumsum is performed in the
+ * opposite direction:
+ * <pre>{@code
+ * tf.cumsum([a, b, c], reverse=True)  # => [a + b + c, b + c, c]
+ * }</pre>
+ * This is more efficient than using separate `tf.reverse` ops.
+ * <p>
+ * The `reverse` and `exclusive` kwargs can also be combined:
+ * <pre>{@code
+ * tf.cumsum([a, b, c], exclusive=True, reverse=True)  # => [b + c, c, 0]
+ * }</pre>
+ * 
+ * 
+ * @param <T> data type for {@code out()} output
+ */
+@Operator(group = "math")
+public final class Cumsum<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.math.Cumsum}
+   */
+  public static class Options {
+    
+    /**
+     * @param exclusive If `True`, perform exclusive cumsum.
+     */
+    public Options exclusive(Boolean exclusive) {
+      this.exclusive = exclusive;
+      return this;
+    }
+    
+    /**
+     * @param reverse A `bool` (default: False).
+     */
+    public Options reverse(Boolean reverse) {
+      this.reverse = reverse;
+      return this;
+    }
+    
+    private Boolean exclusive;
+    private Boolean reverse;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new Cumsum operation.
+   * 
+   * @param scope current scope
+   * @param x A `Tensor`. Must be one of the following types: `float32`, `float64`,
+   * `int64`, `int32`, `uint8`, `uint16`, `int16`, `int8`, `complex64`,
+   * `complex128`, `qint8`, `quint8`, `qint32`, `half`.
+   * @param axis A `Tensor` of type `int32` (default: 0). Must be in the range
+   * `[-rank(x), rank(x))`.
+   * @param options carries optional attributes values
+   * @return a new instance of Cumsum
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType, U extends TNumber> Cumsum<T> create(Scope scope, Operand<T> x, Operand<U> axis, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Cumsum", scope.makeOpName("Cumsum"));
+    opBuilder.addInput(x.asOutput());
+    opBuilder.addInput(axis.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.exclusive != null) {
+          opBuilder.setAttr("exclusive", opts.exclusive);
+        }
+        if (opts.reverse != null) {
+          opBuilder.setAttr("reverse", opts.reverse);
+        }
+      }
+    }
+    return new Cumsum<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param exclusive If `True`, perform exclusive cumsum.
+   */
+  public static Options exclusive(Boolean exclusive) {
+    return new Options().exclusive(exclusive);
+  }
+  
+  /**
+   * @param reverse A `bool` (default: False).
+   */
+  public static Options reverse(Boolean reverse) {
+    return new Options().reverse(reverse);
+  }
+  
+  /**
+   */
+  public Output<T> out() {
+    return out;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return out;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Cumsum";
+  
+  private Output<T> out;
+  
+  private Cumsum(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    out = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/CumulativeLogsumexp.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/CumulativeLogsumexp.java
new file mode 100644
index 00000000000..4f199369473
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/CumulativeLogsumexp.java
@@ -0,0 +1,147 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.math;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Compute the cumulative product of the tensor `x` along `axis`.
+ * <p>
+ * By default, this op performs an inclusive cumulative log-sum-exp,
+ * which means that the first
+ * element of the input is identical to the first element of the output:
+ * <pre>{@code
+ * tf.math.cumulative_logsumexp([a, b, c])  # => [a, log(exp(a) + exp(b)), log(exp(a) + exp(b) + exp(c))]
+ * }</pre>
+ * By setting the `exclusive` kwarg to `True`, an exclusive cumulative log-sum-exp is
+ * performed instead:
+ * <pre>{@code
+ * tf.cumulative_logsumexp([a, b, c], exclusive=True)  # => [-inf, a, log(exp(a) * exp(b))]
+ * }</pre>
+ * Note that the neutral element of the log-sum-exp operation is `-inf`,
+ * however, for performance reasons, the minimal value representable by the
+ * floating point type is used instead.
+ * <p>
+ * By setting the `reverse` kwarg to `True`, the cumulative log-sum-exp is performed in the
+ * opposite direction.
+ * 
+ * @param <T> data type for {@code out()} output
+ */
+public final class CumulativeLogsumexp<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.math.CumulativeLogsumexp}
+   */
+  public static class Options {
+    
+    /**
+     * @param exclusive If `True`, perform exclusive cumulative log-sum-exp.
+     */
+    public Options exclusive(Boolean exclusive) {
+      this.exclusive = exclusive;
+      return this;
+    }
+    
+    /**
+     * @param reverse A `bool` (default: False).
+     */
+    public Options reverse(Boolean reverse) {
+      this.reverse = reverse;
+      return this;
+    }
+    
+    private Boolean exclusive;
+    private Boolean reverse;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new CumulativeLogsumexp operation.
+   * 
+   * @param scope current scope
+   * @param x A `Tensor`. Must be one of the following types: `float16`, `float32`, `float64`.
+   * @param axis A `Tensor` of type `int32` (default: 0). Must be in the range
+   * `[-rank(x), rank(x))`.
+   * @param options carries optional attributes values
+   * @return a new instance of CumulativeLogsumexp
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber, U extends TNumber> CumulativeLogsumexp<T> create(Scope scope, Operand<T> x, Operand<U> axis, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("CumulativeLogsumexp", scope.makeOpName("CumulativeLogsumexp"));
+    opBuilder.addInput(x.asOutput());
+    opBuilder.addInput(axis.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.exclusive != null) {
+          opBuilder.setAttr("exclusive", opts.exclusive);
+        }
+        if (opts.reverse != null) {
+          opBuilder.setAttr("reverse", opts.reverse);
+        }
+      }
+    }
+    return new CumulativeLogsumexp<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param exclusive If `True`, perform exclusive cumulative log-sum-exp.
+   */
+  public static Options exclusive(Boolean exclusive) {
+    return new Options().exclusive(exclusive);
+  }
+  
+  /**
+   * @param reverse A `bool` (default: False).
+   */
+  public static Options reverse(Boolean reverse) {
+    return new Options().reverse(reverse);
+  }
+  
+  /**
+   */
+  public Output<T> out() {
+    return out;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return out;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "CumulativeLogsumexp";
+  
+  private Output<T> out;
+  
+  private CumulativeLogsumexp(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    out = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/DenseBincount.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/DenseBincount.java
new file mode 100644
index 00000000000..edbc39b49cb
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/DenseBincount.java
@@ -0,0 +1,124 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.math;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Counts the number of occurrences of each value in an integer array.
+ * <p>
+ * Outputs a vector with length `size` and the same dtype as `weights`. If
+ * `weights` are empty, then index `i` stores the number of times the value `i` is
+ * counted in `arr`. If `weights` are non-empty, then index `i` stores the sum of
+ * the value in `weights` at each index where the corresponding value in `arr` is
+ * `i`.
+ * <p>
+ * Values in `arr` outside of the range [0, size) are ignored.
+ * 
+ * @param <U> data type for {@code output()} output
+ */
+@Operator(group = "math")
+public final class DenseBincount<U extends TNumber> extends RawOp implements Operand<U> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.math.DenseBincount}
+   */
+  public static class Options {
+    
+    /**
+     * @param binaryOutput bool; Whether the kernel should count the appearance or number of occurrences.
+     */
+    public Options binaryOutput(Boolean binaryOutput) {
+      this.binaryOutput = binaryOutput;
+      return this;
+    }
+    
+    private Boolean binaryOutput;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new DenseBincount operation.
+   * 
+   * @param scope current scope
+   * @param input 1D or 2D int `Tensor`.
+   * @param size non-negative int scalar `Tensor`.
+   * @param weights is an int32, int64, float32, or float64 `Tensor` with the same
+   * shape as `arr`, or a length-0 `Tensor`, in which case it acts as all weights
+   * equal to 1.
+   * @param options carries optional attributes values
+   * @return a new instance of DenseBincount
+   */
+  @Endpoint(describeByClass = true)
+  public static <U extends TNumber, T extends TNumber> DenseBincount<U> create(Scope scope, Operand<T> input, Operand<T> size, Operand<U> weights, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("DenseBincount", scope.makeOpName("DenseBincount"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(size.asOutput());
+    opBuilder.addInput(weights.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.binaryOutput != null) {
+          opBuilder.setAttr("binary_output", opts.binaryOutput);
+        }
+      }
+    }
+    return new DenseBincount<U>(opBuilder.build());
+  }
+  
+  /**
+   * @param binaryOutput bool; Whether the kernel should count the appearance or number of occurrences.
+   */
+  public static Options binaryOutput(Boolean binaryOutput) {
+    return new Options().binaryOutput(binaryOutput);
+  }
+  
+  /**
+   * 1D `Tensor` with length equal to `size` or 2D `Tensor` with [batch_size, `size`].
+   * The counts or summed weights for each value in the range [0, size).
+   */
+  public Output<U> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<U> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "DenseBincount";
+  
+  private Output<U> output;
+  
+  private DenseBincount(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Digamma.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Digamma.java
new file mode 100644
index 00000000000..571ade7795b
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Digamma.java
@@ -0,0 +1,76 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.math;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Computes Psi, the derivative of Lgamma (the log of the absolute value of
+ * <p>
+ * `Gamma(x)`), element-wise.
+ * 
+ * @param <T> data type for {@code y()} output
+ */
+@Operator(group = "math")
+public final class Digamma<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new Digamma operation.
+   * 
+   * @param scope current scope
+   * @param x 
+   * @return a new instance of Digamma
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> Digamma<T> create(Scope scope, Operand<T> x) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Digamma", scope.makeOpName("Digamma"));
+    opBuilder.addInput(x.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new Digamma<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> y() {
+    return y;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return y;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Digamma";
+  
+  private Output<T> y;
+  
+  private Digamma(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    y = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Div.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Div.java
new file mode 100644
index 00000000000..0cd80db04df
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Div.java
@@ -0,0 +1,79 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.math;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Returns x / y element-wise.
+ * <p>
+ * <i>NOTE</i>: `math.Div` supports broadcasting. More about broadcasting
+ * [here](http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html)
+ * 
+ * @param <T> data type for {@code z()} output
+ */
+@Operator(group = "math")
+public final class Div<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new Div operation.
+   * 
+   * @param scope current scope
+   * @param x 
+   * @param y 
+   * @return a new instance of Div
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> Div<T> create(Scope scope, Operand<T> x, Operand<T> y) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Div", scope.makeOpName("Div"));
+    opBuilder.addInput(x.asOutput());
+    opBuilder.addInput(y.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new Div<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> z() {
+    return z;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return z;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Div";
+  
+  private Output<T> z;
+  
+  private Div(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    z = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/DivNoNan.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/DivNoNan.java
new file mode 100644
index 00000000000..7a1ea66a50d
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/DivNoNan.java
@@ -0,0 +1,80 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.math;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Returns 0 if the denominator is zero.
+ * <p>
+ * 
+ * <i>NOTE</i>: `math.DivNoNan` supports broadcasting. More about broadcasting
+ * [here](http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html)
+ * 
+ * @param <T> data type for {@code z()} output
+ */
+@Operator(group = "math")
+public final class DivNoNan<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new DivNoNan operation.
+   * 
+   * @param scope current scope
+   * @param x 
+   * @param y 
+   * @return a new instance of DivNoNan
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> DivNoNan<T> create(Scope scope, Operand<T> x, Operand<T> y) {
+    OperationBuilder opBuilder = scope.env().opBuilder("DivNoNan", scope.makeOpName("DivNoNan"));
+    opBuilder.addInput(x.asOutput());
+    opBuilder.addInput(y.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new DivNoNan<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> z() {
+    return z;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return z;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "DivNoNan";
+  
+  private Output<T> z;
+  
+  private DivNoNan(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    z = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Equal.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Equal.java
new file mode 100644
index 00000000000..d67b38e9ca6
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Equal.java
@@ -0,0 +1,122 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.math;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TBool;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Returns the truth value of (x == y) element-wise.
+ * <p>
+ * <i>NOTE</i>: `math.Equal` supports broadcasting. More about broadcasting
+ * [here](http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html)
+ * <pre>{@code
+ * x = tf.constant([2, 4])
+ * y = tf.constant(2)
+ * tf.math.equal(x, y) ==> array([True, False])
+ * 
+ * x = tf.constant([2, 4])
+ * y = tf.constant([2, 4])
+ * tf.math.equal(x, y) ==> array([True,  True])
+ * }</pre>
+ * 
+ */
+@Operator(group = "math")
+public final class Equal extends RawOp implements Operand<TBool> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.math.Equal}
+   */
+  public static class Options {
+    
+    /**
+     * @param incompatibleShapeError 
+     */
+    public Options incompatibleShapeError(Boolean incompatibleShapeError) {
+      this.incompatibleShapeError = incompatibleShapeError;
+      return this;
+    }
+    
+    private Boolean incompatibleShapeError;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new Equal operation.
+   * 
+   * @param scope current scope
+   * @param x 
+   * @param y 
+   * @param options carries optional attributes values
+   * @return a new instance of Equal
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> Equal create(Scope scope, Operand<T> x, Operand<T> y, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Equal", scope.makeOpName("Equal"));
+    opBuilder.addInput(x.asOutput());
+    opBuilder.addInput(y.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.incompatibleShapeError != null) {
+          opBuilder.setAttr("incompatible_shape_error", opts.incompatibleShapeError);
+        }
+      }
+    }
+    return new Equal(opBuilder.build());
+  }
+  
+  /**
+   * @param incompatibleShapeError 
+   */
+  public static Options incompatibleShapeError(Boolean incompatibleShapeError) {
+    return new Options().incompatibleShapeError(incompatibleShapeError);
+  }
+  
+  /**
+   */
+  public Output<TBool> z() {
+    return z;
+  }
+  
+  @Override
+  public Output<TBool> asOutput() {
+    return z;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Equal";
+  
+  private Output<TBool> z;
+  
+  private Equal(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    z = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Erf.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Erf.java
new file mode 100644
index 00000000000..ff1c2d343e8
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Erf.java
@@ -0,0 +1,74 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.math;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Computes the Gauss error function of `x` element-wise.
+ * 
+ * @param <T> data type for {@code y()} output
+ */
+@Operator(group = "math")
+public final class Erf<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new Erf operation.
+   * 
+   * @param scope current scope
+   * @param x 
+   * @return a new instance of Erf
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> Erf<T> create(Scope scope, Operand<T> x) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Erf", scope.makeOpName("Erf"));
+    opBuilder.addInput(x.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new Erf<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> y() {
+    return y;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return y;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Erf";
+  
+  private Output<T> y;
+  
+  private Erf(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    y = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Erfc.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Erfc.java
new file mode 100644
index 00000000000..31826d91508
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Erfc.java
@@ -0,0 +1,74 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.math;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Computes the complementary error function of `x` element-wise.
+ * 
+ * @param <T> data type for {@code y()} output
+ */
+@Operator(group = "math")
+public final class Erfc<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new Erfc operation.
+   * 
+   * @param scope current scope
+   * @param x 
+   * @return a new instance of Erfc
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> Erfc<T> create(Scope scope, Operand<T> x) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Erfc", scope.makeOpName("Erfc"));
+    opBuilder.addInput(x.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new Erfc<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> y() {
+    return y;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return y;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Erfc";
+  
+  private Output<T> y;
+  
+  private Erfc(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    y = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Exp.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Exp.java
new file mode 100644
index 00000000000..cee48319f26
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Exp.java
@@ -0,0 +1,100 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.math;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Computes exponential of x element-wise.  \\(y = e^x\\).
+ * <p>
+ *   This function computes the exponential of every element in the input tensor.
+ *   i.e. `exp(x)` or `e^(x)`, where `x` is the input tensor.
+ *   `e` denotes Euler's number and is approximately equal to 2.718281.
+ *   Output is positive for any real input.
+ * <p>
+ *   <pre>{@code
+ *   x = tf.constant(2.0)
+ *   tf.math.exp(x) ==> 7.389056
+ * 
+ *   x = tf.constant([2.0, 8.0])
+ *   tf.math.exp(x) ==> array([7.389056, 2980.958], dtype=float32)
+ *   }</pre>
+ * For complex numbers, the exponential value is calculated as follows:
+ * <p>
+ *   <pre>{@code
+ *   e^(x+iy) = e^x * e^iy = e^x * (cos y + i sin y)
+ *   }</pre>
+ * Let's consider complex number 1+1j as an example.
+ *   e^1 * (cos 1 + i sin 1) = 2.7182818284590 * (0.54030230586+0.8414709848j)
+ * <p>
+ *   <pre>{@code
+ *   x = tf.constant(1 + 1j)
+ *   tf.math.exp(x) ==> 1.4686939399158851+2.2873552871788423j
+ *   }</pre>
+ * 
+ * 
+ * @param <T> data type for {@code y()} output
+ */
+@Operator(group = "math")
+public final class Exp<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new Exp operation.
+   * 
+   * @param scope current scope
+   * @param x 
+   * @return a new instance of Exp
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> Exp<T> create(Scope scope, Operand<T> x) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Exp", scope.makeOpName("Exp"));
+    opBuilder.addInput(x.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new Exp<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> y() {
+    return y;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return y;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Exp";
+  
+  private Output<T> y;
+  
+  private Exp(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    y = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Expm1.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Expm1.java
new file mode 100644
index 00000000000..9d76f16b774
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Expm1.java
@@ -0,0 +1,89 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.math;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Computes `exp(x) - 1` element-wise.
+ * <p>
+ *   i.e. `exp(x) - 1` or `e^(x) - 1`, where `x` is the input tensor.
+ *   `e` denotes Euler's number and is approximately equal to 2.718281.
+ * <p>
+ *   <pre>{@code
+ *   x = tf.constant(2.0)
+ *   tf.math.expm1(x) ==> 6.389056
+ * 
+ *   x = tf.constant([2.0, 8.0])
+ *   tf.math.expm1(x) ==> array([6.389056, 2979.958], dtype=float32)
+ * 
+ *   x = tf.constant(1 + 1j)
+ *   tf.math.expm1(x) ==> (0.46869393991588515+2.2873552871788423j)
+ *   }</pre>
+ * 
+ * 
+ * @param <T> data type for {@code y()} output
+ */
+@Operator(group = "math")
+public final class Expm1<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new Expm1 operation.
+   * 
+   * @param scope current scope
+   * @param x 
+   * @return a new instance of Expm1
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> Expm1<T> create(Scope scope, Operand<T> x) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Expm1", scope.makeOpName("Expm1"));
+    opBuilder.addInput(x.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new Expm1<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> y() {
+    return y;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return y;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Expm1";
+  
+  private Output<T> y;
+  
+  private Expm1(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    y = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Fact.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Fact.java
new file mode 100644
index 00000000000..31d78966166
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Fact.java
@@ -0,0 +1,70 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.math;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TString;
+
+/**
+ * Output a fact about factorials.
+ */
+@Operator(group = "math")
+public final class Fact extends RawOp implements Operand<TString> {
+  
+  /**
+   * Factory method to create a class wrapping a new Fact operation.
+   * 
+   * @param scope current scope
+   * @return a new instance of Fact
+   */
+  @Endpoint(describeByClass = true)
+  public static Fact create(Scope scope) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Fact", scope.makeOpName("Fact"));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new Fact(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<TString> fact() {
+    return fact;
+  }
+  
+  @Override
+  public Output<TString> asOutput() {
+    return fact;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Fact";
+  
+  private Output<TString> fact;
+  
+  private Fact(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    fact = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Floor.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Floor.java
new file mode 100644
index 00000000000..acd6f3a1d2b
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Floor.java
@@ -0,0 +1,74 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.math;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Returns element-wise largest integer not greater than x.
+ * 
+ * @param <T> data type for {@code y()} output
+ */
+@Operator(group = "math")
+public final class Floor<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new Floor operation.
+   * 
+   * @param scope current scope
+   * @param x 
+   * @return a new instance of Floor
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> Floor<T> create(Scope scope, Operand<T> x) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Floor", scope.makeOpName("Floor"));
+    opBuilder.addInput(x.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new Floor<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> y() {
+    return y;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return y;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Floor";
+  
+  private Output<T> y;
+  
+  private Floor(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    y = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/FloorDiv.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/FloorDiv.java
new file mode 100644
index 00000000000..366b88bc6d9
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/FloorDiv.java
@@ -0,0 +1,79 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.math;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Returns x // y element-wise.
+ * <p>
+ * <i>NOTE</i>: `math.FloorDiv` supports broadcasting. More about broadcasting
+ * [here](http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html)
+ * 
+ * @param <T> data type for {@code z()} output
+ */
+@Operator(group = "math")
+public final class FloorDiv<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new FloorDiv operation.
+   * 
+   * @param scope current scope
+   * @param x 
+   * @param y 
+   * @return a new instance of FloorDiv
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> FloorDiv<T> create(Scope scope, Operand<T> x, Operand<T> y) {
+    OperationBuilder opBuilder = scope.env().opBuilder("FloorDiv", scope.makeOpName("FloorDiv"));
+    opBuilder.addInput(x.asOutput());
+    opBuilder.addInput(y.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new FloorDiv<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> z() {
+    return z;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return z;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "FloorDiv";
+  
+  private Output<T> z;
+  
+  private FloorDiv(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    z = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/FloorMod.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/FloorMod.java
new file mode 100644
index 00000000000..ae4e933ccea
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/FloorMod.java
@@ -0,0 +1,82 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.math;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Returns element-wise remainder of division. When `x < 0` xor `y < 0` is
+ * <p>
+ * true, this follows Python semantics in that the result here is consistent
+ * with a flooring divide. E.g. `floor(x / y) * y + mod(x, y) = x`.
+ * <p>
+ * <i>NOTE</i>: `math.FloorMod` supports broadcasting. More about broadcasting
+ * [here](http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html)
+ * 
+ * @param <T> data type for {@code z()} output
+ */
+@Operator(group = "math")
+public final class FloorMod<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new FloorMod operation.
+   * 
+   * @param scope current scope
+   * @param x 
+   * @param y 
+   * @return a new instance of FloorMod
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> FloorMod<T> create(Scope scope, Operand<T> x, Operand<T> y) {
+    OperationBuilder opBuilder = scope.env().opBuilder("FloorMod", scope.makeOpName("FloorMod"));
+    opBuilder.addInput(x.asOutput());
+    opBuilder.addInput(y.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new FloorMod<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> z() {
+    return z;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return z;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "FloorMod";
+  
+  private Output<T> z;
+  
+  private FloorMod(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    z = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Greater.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Greater.java
new file mode 100644
index 00000000000..a0f100e6b03
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Greater.java
@@ -0,0 +1,90 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.math;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TBool;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Returns the truth value of (x > y) element-wise.
+ * <p>
+ * <i>NOTE</i>: `math.Greater` supports broadcasting. More about broadcasting
+ * [here](http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html)
+ * <p>
+ * Example:
+ * <pre>{@code
+ * x = tf.constant([5, 4, 6])
+ * y = tf.constant([5, 2, 5])
+ * tf.math.greater(x, y) ==> [False, True, True]
+ * 
+ * x = tf.constant([5, 4, 6])
+ * y = tf.constant([5])
+ * tf.math.greater(x, y) ==> [False, False, True]
+ * }</pre>
+ * 
+ */
+@Operator(group = "math")
+public final class Greater extends RawOp implements Operand<TBool> {
+  
+  /**
+   * Factory method to create a class wrapping a new Greater operation.
+   * 
+   * @param scope current scope
+   * @param x 
+   * @param y 
+   * @return a new instance of Greater
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> Greater create(Scope scope, Operand<T> x, Operand<T> y) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Greater", scope.makeOpName("Greater"));
+    opBuilder.addInput(x.asOutput());
+    opBuilder.addInput(y.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new Greater(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<TBool> z() {
+    return z;
+  }
+  
+  @Override
+  public Output<TBool> asOutput() {
+    return z;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Greater";
+  
+  private Output<TBool> z;
+  
+  private Greater(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    z = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/GreaterEqual.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/GreaterEqual.java
new file mode 100644
index 00000000000..fa67314267f
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/GreaterEqual.java
@@ -0,0 +1,90 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.math;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TBool;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Returns the truth value of (x >= y) element-wise.
+ * <p>
+ * <i>NOTE</i>: `math.GreaterEqual` supports broadcasting. More about broadcasting
+ * [here](http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html)
+ * <p>
+ * Example:
+ * <pre>{@code
+ * x = tf.constant([5, 4, 6, 7])
+ * y = tf.constant([5, 2, 5, 10])
+ * tf.math.greater_equal(x, y) ==> [True, True, True, False]
+ * 
+ * x = tf.constant([5, 4, 6, 7])
+ * y = tf.constant([5])
+ * tf.math.greater_equal(x, y) ==> [True, False, True, True]
+ * }</pre>
+ * 
+ */
+@Operator(group = "math")
+public final class GreaterEqual extends RawOp implements Operand<TBool> {
+  
+  /**
+   * Factory method to create a class wrapping a new GreaterEqual operation.
+   * 
+   * @param scope current scope
+   * @param x 
+   * @param y 
+   * @return a new instance of GreaterEqual
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> GreaterEqual create(Scope scope, Operand<T> x, Operand<T> y) {
+    OperationBuilder opBuilder = scope.env().opBuilder("GreaterEqual", scope.makeOpName("GreaterEqual"));
+    opBuilder.addInput(x.asOutput());
+    opBuilder.addInput(y.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new GreaterEqual(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<TBool> z() {
+    return z;
+  }
+  
+  @Override
+  public Output<TBool> asOutput() {
+    return z;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "GreaterEqual";
+  
+  private Output<TBool> z;
+  
+  private GreaterEqual(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    z = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Igamma.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Igamma.java
new file mode 100644
index 00000000000..6d757ea96c6
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Igamma.java
@@ -0,0 +1,89 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.math;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Compute the lower regularized incomplete Gamma function `P(a, x)`.
+ * <p>
+ * The lower regularized incomplete Gamma function is defined as:
+ * <p>
+ * \\(P(a, x) = gamma(a, x) / Gamma(a) = 1 - Q(a, x)\\)
+ * <p>
+ * where
+ * <p>
+ * \\(gamma(a, x) = \\int_{0}^{x} t^{a-1} exp(-t) dt\\)
+ * <p>
+ * is the lower incomplete Gamma function.
+ * <p>
+ * Note, above `Q(a, x)` (`Igammac`) is the upper regularized complete
+ * Gamma function.
+ * 
+ * @param <T> data type for {@code z()} output
+ */
+@Operator(group = "math")
+public final class Igamma<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new Igamma operation.
+   * 
+   * @param scope current scope
+   * @param a 
+   * @param x 
+   * @return a new instance of Igamma
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> Igamma<T> create(Scope scope, Operand<T> a, Operand<T> x) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Igamma", scope.makeOpName("Igamma"));
+    opBuilder.addInput(a.asOutput());
+    opBuilder.addInput(x.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new Igamma<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> z() {
+    return z;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return z;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Igamma";
+  
+  private Output<T> z;
+  
+  private Igamma(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    z = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/IgammaGradA.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/IgammaGradA.java
new file mode 100644
index 00000000000..9e757f8e01d
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/IgammaGradA.java
@@ -0,0 +1,75 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.math;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Computes the gradient of `igamma(a, x)` wrt `a`.
+ * 
+ * @param <T> data type for {@code z()} output
+ */
+public final class IgammaGradA<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new IgammaGradA operation.
+   * 
+   * @param scope current scope
+   * @param a 
+   * @param x 
+   * @return a new instance of IgammaGradA
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> IgammaGradA<T> create(Scope scope, Operand<T> a, Operand<T> x) {
+    OperationBuilder opBuilder = scope.env().opBuilder("IgammaGradA", scope.makeOpName("IgammaGradA"));
+    opBuilder.addInput(a.asOutput());
+    opBuilder.addInput(x.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new IgammaGradA<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> z() {
+    return z;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return z;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "IgammaGradA";
+  
+  private Output<T> z;
+  
+  private IgammaGradA(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    z = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Igammac.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Igammac.java
new file mode 100644
index 00000000000..dec4671fe76
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Igammac.java
@@ -0,0 +1,89 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.math;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Compute the upper regularized incomplete Gamma function `Q(a, x)`.
+ * <p>
+ * The upper regularized incomplete Gamma function is defined as:
+ * <p>
+ * \\(Q(a, x) = Gamma(a, x) / Gamma(a) = 1 - P(a, x)\\)
+ * <p>
+ * where
+ * <p>
+ * \\(Gamma(a, x) = int_{x}^{\infty} t^{a-1} exp(-t) dt\\)
+ * <p>
+ * is the upper incomplete Gama function.
+ * <p>
+ * Note, above `P(a, x)` (`Igamma`) is the lower regularized complete
+ * Gamma function.
+ * 
+ * @param <T> data type for {@code z()} output
+ */
+@Operator(group = "math")
+public final class Igammac<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new Igammac operation.
+   * 
+   * @param scope current scope
+   * @param a 
+   * @param x 
+   * @return a new instance of Igammac
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> Igammac<T> create(Scope scope, Operand<T> a, Operand<T> x) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Igammac", scope.makeOpName("Igammac"));
+    opBuilder.addInput(a.asOutput());
+    opBuilder.addInput(x.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new Igammac<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> z() {
+    return z;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return z;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Igammac";
+  
+  private Output<T> z;
+  
+  private Igammac(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    z = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Imag.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Imag.java
new file mode 100644
index 00000000000..9c4ce01655e
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Imag.java
@@ -0,0 +1,102 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.math;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Returns the imaginary part of a complex number.
+ * <p>
+ * Given a tensor `input` of complex numbers, this operation returns a tensor of
+ * type `float` that is the imaginary part of each element in `input`. All
+ * elements in `input` must be complex numbers of the form \\(a + bj\\), where <i>a</i>
+ * is the real part and <i>b</i> is the imaginary part returned by this operation.
+ * <p>
+ * For example:
+ * <pre>{@code
+ * # tensor 'input' is [-2.25 + 4.75j, 3.25 + 5.75j]
+ * tf.imag(input) ==> [4.75, 5.75]
+ * }</pre>
+ * 
+ * 
+ * @param <U> data type for {@code output()} output
+ */
+@Operator(group = "math")
+public final class Imag<U extends TNumber> extends RawOp implements Operand<U> {
+  
+  /**
+   * Factory method to create a class wrapping a new Imag operation.
+   * 
+   * @param scope current scope
+   * @param input 
+   * @param Tout 
+   * @return a new instance of Imag
+   */
+  @Endpoint(describeByClass = true)
+  public static <U extends TNumber, T extends TType> Imag<U> create(Scope scope, Operand<T> input, Class<U> Tout) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Imag", scope.makeOpName("Imag"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("Tout", Tout);
+    return new Imag<U>(opBuilder.build());
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new Imag operation using default output types.
+   * 
+   * @param scope current scope
+   * @param input 
+   * @return a new instance of Imag
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> Imag<TFloat32> create(Scope scope, Operand<T> input) {
+    return create(scope, input, TFloat32.class);
+  }
+  
+  /**
+   */
+  public Output<U> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<U> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Imag";
+  
+  private Output<U> output;
+  
+  private Imag(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/InvertPermutation.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/InvertPermutation.java
new file mode 100644
index 00000000000..859d648b2c5
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/InvertPermutation.java
@@ -0,0 +1,91 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.math;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Computes the inverse permutation of a tensor.
+ * <p>
+ * This operation computes the inverse of an index permutation. It takes a 1-D
+ * integer tensor `x`, which represents the indices of a zero-based array, and
+ * swaps each value with its index position. In other words, for an output tensor
+ * `y` and an input tensor `x`, this operation computes the following:
+ * <p>
+ * `y[x[i]] = i for i in [0, 1, ..., len(x) - 1]`
+ * <p>
+ * The values must include 0. There can be no duplicate values or negative values.
+ * <p>
+ * For example:
+ * <pre>{@code
+ * # tensor `x` is [3, 4, 0, 2, 1]
+ * invert_permutation(x) ==> [2, 4, 3, 0, 1]
+ * }</pre>
+ * 
+ * 
+ * @param <T> data type for {@code y()} output
+ */
+@Operator(group = "math")
+public final class InvertPermutation<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new InvertPermutation operation.
+   * 
+   * @param scope current scope
+   * @param x 1-D.
+   * @return a new instance of InvertPermutation
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> InvertPermutation<T> create(Scope scope, Operand<T> x) {
+    OperationBuilder opBuilder = scope.env().opBuilder("InvertPermutation", scope.makeOpName("InvertPermutation"));
+    opBuilder.addInput(x.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new InvertPermutation<T>(opBuilder.build());
+  }
+  
+  /**
+   * 1-D.
+   */
+  public Output<T> y() {
+    return y;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return y;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "InvertPermutation";
+  
+  private Output<T> y;
+  
+  private InvertPermutation(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    y = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/IsFinite.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/IsFinite.java
new file mode 100644
index 00000000000..b6536046d53
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/IsFinite.java
@@ -0,0 +1,84 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.math;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TBool;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Returns which elements of x are finite.
+ * <p>
+ * @compatibility(numpy)
+ * Equivalent to np.isfinite
+ * @end_compatibility
+ * <p>
+ * Example:
+ * <pre>{@code
+ * x = tf.constant([5.0, 4.8, 6.8, np.inf, np.nan])
+ * tf.math.is_finite(x) ==> [True, True, True, False, False]
+ * }</pre>
+ * 
+ */
+@Operator(group = "math")
+public final class IsFinite extends RawOp implements Operand<TBool> {
+  
+  /**
+   * Factory method to create a class wrapping a new IsFinite operation.
+   * 
+   * @param scope current scope
+   * @param x 
+   * @return a new instance of IsFinite
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> IsFinite create(Scope scope, Operand<T> x) {
+    OperationBuilder opBuilder = scope.env().opBuilder("IsFinite", scope.makeOpName("IsFinite"));
+    opBuilder.addInput(x.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new IsFinite(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<TBool> y() {
+    return y;
+  }
+  
+  @Override
+  public Output<TBool> asOutput() {
+    return y;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "IsFinite";
+  
+  private Output<TBool> y;
+  
+  private IsFinite(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    y = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/IsInf.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/IsInf.java
new file mode 100644
index 00000000000..cde3b798d16
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/IsInf.java
@@ -0,0 +1,84 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.math;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TBool;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Returns which elements of x are Inf.
+ * <p>
+ * @compatibility(numpy)
+ * Equivalent to np.isinf
+ * @end_compatibility
+ * <p>
+ * Example:
+ * <pre>{@code
+ * x = tf.constant([5.0, np.inf, 6.8, np.inf])
+ * tf.math.is_inf(x) ==> [False, True, False, True]
+ * }</pre>
+ * 
+ */
+@Operator(group = "math")
+public final class IsInf extends RawOp implements Operand<TBool> {
+  
+  /**
+   * Factory method to create a class wrapping a new IsInf operation.
+   * 
+   * @param scope current scope
+   * @param x 
+   * @return a new instance of IsInf
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> IsInf create(Scope scope, Operand<T> x) {
+    OperationBuilder opBuilder = scope.env().opBuilder("IsInf", scope.makeOpName("IsInf"));
+    opBuilder.addInput(x.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new IsInf(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<TBool> y() {
+    return y;
+  }
+  
+  @Override
+  public Output<TBool> asOutput() {
+    return y;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "IsInf";
+  
+  private Output<TBool> y;
+  
+  private IsInf(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    y = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/IsNan.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/IsNan.java
new file mode 100644
index 00000000000..0ecb4836cee
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/IsNan.java
@@ -0,0 +1,84 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.math;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TBool;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Returns which elements of x are NaN.
+ * <p>
+ * @compatibility(numpy)
+ * Equivalent to np.isnan
+ * @end_compatibility
+ * <p>
+ * Example:
+ * <pre>{@code
+ * x = tf.constant([5.0, np.nan, 6.8, np.nan, np.inf])
+ * tf.math.is_nan(x) ==> [False, True, False, True, False]
+ * }</pre>
+ * 
+ */
+@Operator(group = "math")
+public final class IsNan extends RawOp implements Operand<TBool> {
+  
+  /**
+   * Factory method to create a class wrapping a new IsNan operation.
+   * 
+   * @param scope current scope
+   * @param x 
+   * @return a new instance of IsNan
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> IsNan create(Scope scope, Operand<T> x) {
+    OperationBuilder opBuilder = scope.env().opBuilder("IsNan", scope.makeOpName("IsNan"));
+    opBuilder.addInput(x.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new IsNan(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<TBool> y() {
+    return y;
+  }
+  
+  @Override
+  public Output<TBool> asOutput() {
+    return y;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "IsNan";
+  
+  private Output<TBool> y;
+  
+  private IsNan(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    y = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Less.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Less.java
new file mode 100644
index 00000000000..cf696b5a759
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Less.java
@@ -0,0 +1,90 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.math;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TBool;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Returns the truth value of (x < y) element-wise.
+ * <p>
+ * <i>NOTE</i>: `math.Less` supports broadcasting. More about broadcasting
+ * [here](http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html)
+ * <p>
+ * Example:
+ * <pre>{@code
+ * x = tf.constant([5, 4, 6])
+ * y = tf.constant([5])
+ * tf.math.less(x, y) ==> [False, True, False]
+ * 
+ * x = tf.constant([5, 4, 6])
+ * y = tf.constant([5, 6, 7])
+ * tf.math.less(x, y) ==> [False, True, True]
+ * }</pre>
+ * 
+ */
+@Operator(group = "math")
+public final class Less extends RawOp implements Operand<TBool> {
+  
+  /**
+   * Factory method to create a class wrapping a new Less operation.
+   * 
+   * @param scope current scope
+   * @param x 
+   * @param y 
+   * @return a new instance of Less
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> Less create(Scope scope, Operand<T> x, Operand<T> y) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Less", scope.makeOpName("Less"));
+    opBuilder.addInput(x.asOutput());
+    opBuilder.addInput(y.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new Less(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<TBool> z() {
+    return z;
+  }
+  
+  @Override
+  public Output<TBool> asOutput() {
+    return z;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Less";
+  
+  private Output<TBool> z;
+  
+  private Less(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    z = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/LessEqual.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/LessEqual.java
new file mode 100644
index 00000000000..4b4b85a8198
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/LessEqual.java
@@ -0,0 +1,90 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.math;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TBool;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Returns the truth value of (x <= y) element-wise.
+ * <p>
+ * <i>NOTE</i>: `math.LessEqual` supports broadcasting. More about broadcasting
+ * [here](http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html)
+ * <p>
+ * Example:
+ * <pre>{@code
+ * x = tf.constant([5, 4, 6])
+ * y = tf.constant([5])
+ * tf.math.less_equal(x, y) ==> [True, True, False]
+ * 
+ * x = tf.constant([5, 4, 6])
+ * y = tf.constant([5, 6, 6])
+ * tf.math.less_equal(x, y) ==> [True, True, True]
+ * }</pre>
+ * 
+ */
+@Operator(group = "math")
+public final class LessEqual extends RawOp implements Operand<TBool> {
+  
+  /**
+   * Factory method to create a class wrapping a new LessEqual operation.
+   * 
+   * @param scope current scope
+   * @param x 
+   * @param y 
+   * @return a new instance of LessEqual
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> LessEqual create(Scope scope, Operand<T> x, Operand<T> y) {
+    OperationBuilder opBuilder = scope.env().opBuilder("LessEqual", scope.makeOpName("LessEqual"));
+    opBuilder.addInput(x.asOutput());
+    opBuilder.addInput(y.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new LessEqual(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<TBool> z() {
+    return z;
+  }
+  
+  @Override
+  public Output<TBool> asOutput() {
+    return z;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "LessEqual";
+  
+  private Output<TBool> z;
+  
+  private LessEqual(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    z = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Lgamma.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Lgamma.java
new file mode 100644
index 00000000000..8b3070659e1
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Lgamma.java
@@ -0,0 +1,84 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.math;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Computes the log of the absolute value of `Gamma(x)` element-wise.
+ * <p>
+ *   For positive numbers, this function computes log((input - 1)!) for every element in the tensor.
+ *   `lgamma(5) = log((5-1)!) = log(4!) = log(24) = 3.1780539`
+ * <p>
+ * Example:
+ * <pre>{@code
+ * x = tf.constant([0, 0.5, 1, 4.5, -4, -5.6])
+ * tf.math.lgamma(x) ==> [inf, 0.5723649, 0., 2.4537368, inf, -4.6477685]
+ * }</pre>
+ * 
+ * 
+ * @param <T> data type for {@code y()} output
+ */
+@Operator(group = "math")
+public final class Lgamma<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new Lgamma operation.
+   * 
+   * @param scope current scope
+   * @param x 
+   * @return a new instance of Lgamma
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> Lgamma<T> create(Scope scope, Operand<T> x) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Lgamma", scope.makeOpName("Lgamma"));
+    opBuilder.addInput(x.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new Lgamma<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> y() {
+    return y;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return y;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Lgamma";
+  
+  private Output<T> y;
+  
+  private Lgamma(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    y = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Log.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Log.java
new file mode 100644
index 00000000000..9db418d1002
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Log.java
@@ -0,0 +1,83 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.math;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Computes natural logarithm of x element-wise.
+ * <p>
+ * I.e., \\(y = \log_e x\\).
+ * <p>
+ * Example:
+ * <pre>{@code
+ * x = tf.constant([0, 0.5, 1, 5])
+ * tf.math.log(x) ==> [-inf, -0.6931472,  0. ,  1.609438]
+ * }</pre>
+ * 
+ * 
+ * @param <T> data type for {@code y()} output
+ */
+@Operator(group = "math")
+public final class Log<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new Log operation.
+   * 
+   * @param scope current scope
+   * @param x 
+   * @return a new instance of Log
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> Log<T> create(Scope scope, Operand<T> x) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Log", scope.makeOpName("Log"));
+    opBuilder.addInput(x.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new Log<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> y() {
+    return y;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return y;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Log";
+  
+  private Output<T> y;
+  
+  private Log(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    y = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Log1p.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Log1p.java
new file mode 100644
index 00000000000..3cc2656ea9b
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Log1p.java
@@ -0,0 +1,83 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.math;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Computes natural logarithm of (1 + x) element-wise.
+ * <p>
+ * I.e., \\(y = \log_e (1 + x)\\).
+ * <p>
+ * Example:
+ * <pre>{@code
+ * x = tf.constant([0, 0.5, 1, 5])
+ * tf.math.log1p(x) ==> [0., 0.4054651, 0.6931472, 1.7917595]
+ * }</pre>
+ * 
+ * 
+ * @param <T> data type for {@code y()} output
+ */
+@Operator(group = "math")
+public final class Log1p<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new Log1p operation.
+   * 
+   * @param scope current scope
+   * @param x 
+   * @return a new instance of Log1p
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> Log1p<T> create(Scope scope, Operand<T> x) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Log1p", scope.makeOpName("Log1p"));
+    opBuilder.addInput(x.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new Log1p<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> y() {
+    return y;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return y;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Log1p";
+  
+  private Output<T> y;
+  
+  private Log1p(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    y = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/LogicalAnd.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/LogicalAnd.java
new file mode 100644
index 00000000000..89bb2837f46
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/LogicalAnd.java
@@ -0,0 +1,77 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.math;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TBool;
+
+/**
+ * Returns the truth value of x AND y element-wise.
+ * <p>
+ * <i>NOTE</i>: `math.LogicalAnd` supports broadcasting. More about broadcasting
+ * [here](http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html)
+ */
+@Operator(group = "math")
+public final class LogicalAnd extends RawOp implements Operand<TBool> {
+  
+  /**
+   * Factory method to create a class wrapping a new LogicalAnd operation.
+   * 
+   * @param scope current scope
+   * @param x 
+   * @param y 
+   * @return a new instance of LogicalAnd
+   */
+  @Endpoint(describeByClass = true)
+  public static LogicalAnd create(Scope scope, Operand<TBool> x, Operand<TBool> y) {
+    OperationBuilder opBuilder = scope.env().opBuilder("LogicalAnd", scope.makeOpName("LogicalAnd"));
+    opBuilder.addInput(x.asOutput());
+    opBuilder.addInput(y.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new LogicalAnd(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<TBool> z() {
+    return z;
+  }
+  
+  @Override
+  public Output<TBool> asOutput() {
+    return z;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "LogicalAnd";
+  
+  private Output<TBool> z;
+  
+  private LogicalAnd(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    z = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/LogicalNot.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/LogicalNot.java
new file mode 100644
index 00000000000..f59c85fd320
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/LogicalNot.java
@@ -0,0 +1,73 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.math;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TBool;
+
+/**
+ * Returns the truth value of `NOT x` element-wise.
+ */
+@Operator(group = "math")
+public final class LogicalNot extends RawOp implements Operand<TBool> {
+  
+  /**
+   * Factory method to create a class wrapping a new LogicalNot operation.
+   * 
+   * @param scope current scope
+   * @param x A `Tensor` of type `bool`.
+   * @return a new instance of LogicalNot
+   */
+  @Endpoint(describeByClass = true)
+  public static LogicalNot create(Scope scope, Operand<TBool> x) {
+    OperationBuilder opBuilder = scope.env().opBuilder("LogicalNot", scope.makeOpName("LogicalNot"));
+    opBuilder.addInput(x.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new LogicalNot(opBuilder.build());
+  }
+  
+  /**
+   * A `Tensor` of type `bool` with the same shape as `x`. The logical negation of `x`.
+   */
+  public Output<TBool> y() {
+    return y;
+  }
+  
+  @Override
+  public Output<TBool> asOutput() {
+    return y;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "LogicalNot";
+  
+  private Output<TBool> y;
+  
+  private LogicalNot(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    y = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/LogicalOr.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/LogicalOr.java
new file mode 100644
index 00000000000..604e3ea18f5
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/LogicalOr.java
@@ -0,0 +1,77 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.math;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TBool;
+
+/**
+ * Returns the truth value of x OR y element-wise.
+ * <p>
+ * <i>NOTE</i>: `math.LogicalOr` supports broadcasting. More about broadcasting
+ * [here](http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html)
+ */
+@Operator(group = "math")
+public final class LogicalOr extends RawOp implements Operand<TBool> {
+  
+  /**
+   * Factory method to create a class wrapping a new LogicalOr operation.
+   * 
+   * @param scope current scope
+   * @param x 
+   * @param y 
+   * @return a new instance of LogicalOr
+   */
+  @Endpoint(describeByClass = true)
+  public static LogicalOr create(Scope scope, Operand<TBool> x, Operand<TBool> y) {
+    OperationBuilder opBuilder = scope.env().opBuilder("LogicalOr", scope.makeOpName("LogicalOr"));
+    opBuilder.addInput(x.asOutput());
+    opBuilder.addInput(y.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new LogicalOr(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<TBool> z() {
+    return z;
+  }
+  
+  @Override
+  public Output<TBool> asOutput() {
+    return z;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "LogicalOr";
+  
+  private Output<TBool> z;
+  
+  private LogicalOr(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    z = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Maximum.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Maximum.java
new file mode 100644
index 00000000000..b6360e35d91
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Maximum.java
@@ -0,0 +1,79 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.math;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Returns the max of x and y (i.e. x > y ? x : y) element-wise.
+ * <p>
+ * <i>NOTE</i>: `math.Maximum` supports broadcasting. More about broadcasting
+ * [here](http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html)
+ * 
+ * @param <T> data type for {@code z()} output
+ */
+@Operator(group = "math")
+public final class Maximum<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new Maximum operation.
+   * 
+   * @param scope current scope
+   * @param x 
+   * @param y 
+   * @return a new instance of Maximum
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> Maximum<T> create(Scope scope, Operand<T> x, Operand<T> y) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Maximum", scope.makeOpName("Maximum"));
+    opBuilder.addInput(x.asOutput());
+    opBuilder.addInput(y.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new Maximum<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> z() {
+    return z;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return z;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Maximum";
+  
+  private Output<T> z;
+  
+  private Maximum(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    z = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Mean.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Mean.java
new file mode 100644
index 00000000000..7859b7ce10f
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Mean.java
@@ -0,0 +1,118 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.math;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Computes the mean of elements across dimensions of a tensor.
+ * <p>
+ * Reduces `input` along the dimensions given in `axis`. Unless
+ * `keep_dims` is true, the rank of the tensor is reduced by 1 for each entry in
+ * `axis`. If `keep_dims` is true, the reduced dimensions are
+ * retained with length 1.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator(group = "math")
+public final class Mean<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.math.Mean}
+   */
+  public static class Options {
+    
+    /**
+     * @param keepDims If true, retain reduced dimensions with length 1.
+     */
+    public Options keepDims(Boolean keepDims) {
+      this.keepDims = keepDims;
+      return this;
+    }
+    
+    private Boolean keepDims;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new Mean operation.
+   * 
+   * @param scope current scope
+   * @param input The tensor to reduce.
+   * @param axis The dimensions to reduce. Must be in the range
+   * `[-rank(input), rank(input))`.
+   * @param options carries optional attributes values
+   * @return a new instance of Mean
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType, U extends TNumber> Mean<T> create(Scope scope, Operand<T> input, Operand<U> axis, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Mean", scope.makeOpName("Mean"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(axis.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.keepDims != null) {
+          opBuilder.setAttr("keep_dims", opts.keepDims);
+        }
+      }
+    }
+    return new Mean<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param keepDims If true, retain reduced dimensions with length 1.
+   */
+  public static Options keepDims(Boolean keepDims) {
+    return new Options().keepDims(keepDims);
+  }
+  
+  /**
+   * The reduced tensor.
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Mean";
+  
+  private Output<T> output;
+  
+  private Mean(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Minimum.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Minimum.java
new file mode 100644
index 00000000000..7a3e003bfad
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Minimum.java
@@ -0,0 +1,79 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.math;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Returns the min of x and y (i.e. x < y ? x : y) element-wise.
+ * <p>
+ * <i>NOTE</i>: `math.Minimum` supports broadcasting. More about broadcasting
+ * [here](http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html)
+ * 
+ * @param <T> data type for {@code z()} output
+ */
+@Operator(group = "math")
+public final class Minimum<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new Minimum operation.
+   * 
+   * @param scope current scope
+   * @param x 
+   * @param y 
+   * @return a new instance of Minimum
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> Minimum<T> create(Scope scope, Operand<T> x, Operand<T> y) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Minimum", scope.makeOpName("Minimum"));
+    opBuilder.addInput(x.asOutput());
+    opBuilder.addInput(y.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new Minimum<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> z() {
+    return z;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return z;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Minimum";
+  
+  private Output<T> z;
+  
+  private Minimum(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    z = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Mod.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Mod.java
new file mode 100644
index 00000000000..0502e287fc5
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Mod.java
@@ -0,0 +1,82 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.math;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Returns element-wise remainder of division. This emulates C semantics in that
+ * <p>
+ * the result here is consistent with a truncating divide. E.g.
+ * `tf.truncatediv(x, y) * y + truncate_mod(x, y) = x`.
+ * <p>
+ * <i>NOTE</i>: `math.Mod` supports broadcasting. More about broadcasting
+ * [here](http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html)
+ * 
+ * @param <T> data type for {@code z()} output
+ */
+@Operator(group = "math")
+public final class Mod<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new Mod operation.
+   * 
+   * @param scope current scope
+   * @param x 
+   * @param y 
+   * @return a new instance of Mod
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> Mod<T> create(Scope scope, Operand<T> x, Operand<T> y) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Mod", scope.makeOpName("Mod"));
+    opBuilder.addInput(x.asOutput());
+    opBuilder.addInput(y.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new Mod<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> z() {
+    return z;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return z;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Mod";
+  
+  private Output<T> z;
+  
+  private Mod(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    z = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Mul.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Mul.java
new file mode 100644
index 00000000000..20d5f471ba8
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Mul.java
@@ -0,0 +1,79 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.math;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Returns x * y element-wise.
+ * <p>
+ * <i>NOTE</i>: `math.Mul` supports broadcasting. More about broadcasting
+ * [here](http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html)
+ * 
+ * @param <T> data type for {@code z()} output
+ */
+@Operator(group = "math")
+public final class Mul<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new Mul operation.
+   * 
+   * @param scope current scope
+   * @param x 
+   * @param y 
+   * @return a new instance of Mul
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> Mul<T> create(Scope scope, Operand<T> x, Operand<T> y) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Mul", scope.makeOpName("Mul"));
+    opBuilder.addInput(x.asOutput());
+    opBuilder.addInput(y.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new Mul<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> z() {
+    return z;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return z;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Mul";
+  
+  private Output<T> z;
+  
+  private Mul(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    z = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/MulNoNan.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/MulNoNan.java
new file mode 100644
index 00000000000..312a3025870
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/MulNoNan.java
@@ -0,0 +1,79 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.math;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Returns x * y element-wise. Returns zero if y is zero, even if x if infinite or NaN.
+ * <p>
+ * <i>NOTE</i>: `math.MulNoNan` supports broadcasting. More about broadcasting
+ * [here](http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html)
+ * 
+ * @param <T> data type for {@code z()} output
+ */
+@Operator(group = "math")
+public final class MulNoNan<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new MulNoNan operation.
+   * 
+   * @param scope current scope
+   * @param x 
+   * @param y 
+   * @return a new instance of MulNoNan
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> MulNoNan<T> create(Scope scope, Operand<T> x, Operand<T> y) {
+    OperationBuilder opBuilder = scope.env().opBuilder("MulNoNan", scope.makeOpName("MulNoNan"));
+    opBuilder.addInput(x.asOutput());
+    opBuilder.addInput(y.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new MulNoNan<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> z() {
+    return z;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return z;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "MulNoNan";
+  
+  private Output<T> z;
+  
+  private MulNoNan(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    z = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Ndtri.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Ndtri.java
new file mode 100644
index 00000000000..eec0a0d1c9e
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Ndtri.java
@@ -0,0 +1,72 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.math;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * @param <T> data type for {@code y()} output
+ */
+@Operator(group = "math")
+public final class Ndtri<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new Ndtri operation.
+   * 
+   * @param scope current scope
+   * @param x 
+   * @return a new instance of Ndtri
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> Ndtri<T> create(Scope scope, Operand<T> x) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Ndtri", scope.makeOpName("Ndtri"));
+    opBuilder.addInput(x.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new Ndtri<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> y() {
+    return y;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return y;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Ndtri";
+  
+  private Output<T> y;
+  
+  private Ndtri(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    y = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Neg.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Neg.java
new file mode 100644
index 00000000000..53079692e32
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Neg.java
@@ -0,0 +1,76 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.math;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Computes numerical negative value element-wise.
+ * <p>
+ * I.e., \\(y = -x\\).
+ * 
+ * @param <T> data type for {@code y()} output
+ */
+@Operator(group = "math")
+public final class Neg<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new Neg operation.
+   * 
+   * @param scope current scope
+   * @param x 
+   * @return a new instance of Neg
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> Neg<T> create(Scope scope, Operand<T> x) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Neg", scope.makeOpName("Neg"));
+    opBuilder.addInput(x.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new Neg<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> y() {
+    return y;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return y;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Neg";
+  
+  private Output<T> y;
+  
+  private Neg(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    y = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/NextAfter.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/NextAfter.java
new file mode 100644
index 00000000000..7f494a874d5
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/NextAfter.java
@@ -0,0 +1,84 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.math;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Returns the next representable value of `x1` in the direction of `x2`, element-wise.
+ * <p>
+ * This operation returns the same result as the C++ std::nextafter function.
+ * <p>
+ * It can also return a subnormal number.
+ * <p>
+ * @compatibility(cpp)
+ * Equivalent to C++ std::nextafter function.
+ * @end_compatibility
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator(group = "math")
+public final class NextAfter<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new NextAfter operation.
+   * 
+   * @param scope current scope
+   * @param x1 
+   * @param x2 
+   * @return a new instance of NextAfter
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> NextAfter<T> create(Scope scope, Operand<T> x1, Operand<T> x2) {
+    OperationBuilder opBuilder = scope.env().opBuilder("NextAfter", scope.makeOpName("NextAfter"));
+    opBuilder.addInput(x1.asOutput());
+    opBuilder.addInput(x2.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new NextAfter<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "NextAfter";
+  
+  private Output<T> output;
+  
+  private NextAfter(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/NotEqual.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/NotEqual.java
new file mode 100644
index 00000000000..977688031c5
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/NotEqual.java
@@ -0,0 +1,112 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.math;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TBool;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Returns the truth value of (x != y) element-wise.
+ * <p>
+ * <i>NOTE</i>: `math.NotEqual` supports broadcasting. More about broadcasting
+ * [here](http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html)
+ */
+@Operator(group = "math")
+public final class NotEqual extends RawOp implements Operand<TBool> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.math.NotEqual}
+   */
+  public static class Options {
+    
+    /**
+     * @param incompatibleShapeError 
+     */
+    public Options incompatibleShapeError(Boolean incompatibleShapeError) {
+      this.incompatibleShapeError = incompatibleShapeError;
+      return this;
+    }
+    
+    private Boolean incompatibleShapeError;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new NotEqual operation.
+   * 
+   * @param scope current scope
+   * @param x 
+   * @param y 
+   * @param options carries optional attributes values
+   * @return a new instance of NotEqual
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> NotEqual create(Scope scope, Operand<T> x, Operand<T> y, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("NotEqual", scope.makeOpName("NotEqual"));
+    opBuilder.addInput(x.asOutput());
+    opBuilder.addInput(y.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.incompatibleShapeError != null) {
+          opBuilder.setAttr("incompatible_shape_error", opts.incompatibleShapeError);
+        }
+      }
+    }
+    return new NotEqual(opBuilder.build());
+  }
+  
+  /**
+   * @param incompatibleShapeError 
+   */
+  public static Options incompatibleShapeError(Boolean incompatibleShapeError) {
+    return new Options().incompatibleShapeError(incompatibleShapeError);
+  }
+  
+  /**
+   */
+  public Output<TBool> z() {
+    return z;
+  }
+  
+  @Override
+  public Output<TBool> asOutput() {
+    return z;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "NotEqual";
+  
+  private Output<TBool> z;
+  
+  private NotEqual(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    z = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Polygamma.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Polygamma.java
new file mode 100644
index 00000000000..173ce8cd757
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Polygamma.java
@@ -0,0 +1,83 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.math;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Compute the polygamma function \\(\psi^{(n)}(x)\\).
+ * <p>
+ * The polygamma function is defined as:
+ * <p>
+ * \\(\psi^{(a)}(x) = \frac{d^a}{dx^a} \psi(x)\\)
+ * <p>
+ * where \\(\psi(x)\\) is the digamma function.
+ * The polygamma function is defined only for non-negative integer orders \\a\\.
+ * 
+ * @param <T> data type for {@code z()} output
+ */
+@Operator(group = "math")
+public final class Polygamma<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new Polygamma operation.
+   * 
+   * @param scope current scope
+   * @param a 
+   * @param x 
+   * @return a new instance of Polygamma
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> Polygamma<T> create(Scope scope, Operand<T> a, Operand<T> x) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Polygamma", scope.makeOpName("Polygamma"));
+    opBuilder.addInput(a.asOutput());
+    opBuilder.addInput(x.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new Polygamma<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> z() {
+    return z;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return z;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Polygamma";
+  
+  private Output<T> z;
+  
+  private Polygamma(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    z = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/PopulationCount.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/PopulationCount.java
new file mode 100644
index 00000000000..68c9ebf055b
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/PopulationCount.java
@@ -0,0 +1,80 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.math;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TUint8;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Computes element-wise population count (a.k.a. popcount, bitsum, bitcount).
+ * <p>
+ * For each entry in `x`, calculates the number of `1` (on) bits in the binary
+ * representation of that entry.
+ * <p>
+ * <b>NOTE</b>: It is more efficient to first `tf.bitcast` your tensors into
+ * `int32` or `int64` and perform the bitcount on the result, than to feed in
+ * 8- or 16-bit inputs and then aggregate the resulting counts.
+ */
+@Operator(group = "math")
+public final class PopulationCount extends RawOp implements Operand<TUint8> {
+  
+  /**
+   * Factory method to create a class wrapping a new PopulationCount operation.
+   * 
+   * @param scope current scope
+   * @param x 
+   * @return a new instance of PopulationCount
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> PopulationCount create(Scope scope, Operand<T> x) {
+    OperationBuilder opBuilder = scope.env().opBuilder("PopulationCount", scope.makeOpName("PopulationCount"));
+    opBuilder.addInput(x.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new PopulationCount(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<TUint8> y() {
+    return y;
+  }
+  
+  @Override
+  public Output<TUint8> asOutput() {
+    return y;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "PopulationCount";
+  
+  private Output<TUint8> y;
+  
+  private PopulationCount(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    y = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Pow.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Pow.java
new file mode 100644
index 00000000000..27f26e431fe
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Pow.java
@@ -0,0 +1,85 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.math;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Computes the power of one value to another.
+ * <p>
+ * Given a tensor `x` and a tensor `y`, this operation computes \\(x^y\\) for
+ * corresponding elements in `x` and `y`. For example:
+ * <pre>{@code
+ * # tensor 'x' is [[2, 2]], [3, 3]]
+ * # tensor 'y' is [[8, 16], [2, 3]]
+ * tf.pow(x, y) ==> [[256, 65536], [9, 27]]
+ * }</pre>
+ * 
+ * 
+ * @param <T> data type for {@code z()} output
+ */
+@Operator(group = "math")
+public final class Pow<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new Pow operation.
+   * 
+   * @param scope current scope
+   * @param x 
+   * @param y 
+   * @return a new instance of Pow
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> Pow<T> create(Scope scope, Operand<T> x, Operand<T> y) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Pow", scope.makeOpName("Pow"));
+    opBuilder.addInput(x.asOutput());
+    opBuilder.addInput(y.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new Pow<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> z() {
+    return z;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return z;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Pow";
+  
+  private Output<T> z;
+  
+  private Pow(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    z = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/QuantizedAdd.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/QuantizedAdd.java
new file mode 100644
index 00000000000..2c956969e8d
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/QuantizedAdd.java
@@ -0,0 +1,103 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.math;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Returns x + y element-wise, working on quantized buffers.
+ * 
+ * @param <V> data type for {@code z()} output
+ */
+@Operator(group = "math")
+public final class QuantizedAdd<V extends TType> extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new QuantizedAdd operation.
+   * 
+   * @param scope current scope
+   * @param x 
+   * @param y 
+   * @param minX The float value that the lowest quantized `x` value represents.
+   * @param maxX The float value that the highest quantized `x` value represents.
+   * @param minY The float value that the lowest quantized `y` value represents.
+   * @param maxY The float value that the highest quantized `y` value represents.
+   * @param Toutput 
+   * @return a new instance of QuantizedAdd
+   */
+  @Endpoint(describeByClass = true)
+  public static <V extends TType, T extends TType, U extends TType> QuantizedAdd<V> create(Scope scope, Operand<T> x, Operand<U> y, Operand<TFloat32> minX, Operand<TFloat32> maxX, Operand<TFloat32> minY, Operand<TFloat32> maxY, Class<V> Toutput) {
+    OperationBuilder opBuilder = scope.env().opBuilder("QuantizedAdd", scope.makeOpName("QuantizedAdd"));
+    opBuilder.addInput(x.asOutput());
+    opBuilder.addInput(y.asOutput());
+    opBuilder.addInput(minX.asOutput());
+    opBuilder.addInput(maxX.asOutput());
+    opBuilder.addInput(minY.asOutput());
+    opBuilder.addInput(maxY.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("Toutput", Toutput);
+    return new QuantizedAdd<V>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<V> z() {
+    return z;
+  }
+  
+  /**
+   * The float value that the lowest quantized output value represents.
+   */
+  public Output<TFloat32> minZ() {
+    return minZ;
+  }
+  
+  /**
+   * The float value that the highest quantized output value represents.
+   * <p>
+   * <i>NOTE</i>: `math.QuantizedAdd` supports limited forms of broadcasting. More about
+   * broadcasting [here](http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html)
+   */
+  public Output<TFloat32> maxZ() {
+    return maxZ;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "QuantizedAdd";
+  
+  private Output<V> z;
+  private Output<TFloat32> minZ;
+  private Output<TFloat32> maxZ;
+  
+  private QuantizedAdd(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    z = operation.output(outputIdx++);
+    minZ = operation.output(outputIdx++);
+    maxZ = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/QuantizedMul.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/QuantizedMul.java
new file mode 100644
index 00000000000..bd666426bd1
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/QuantizedMul.java
@@ -0,0 +1,103 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.math;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Returns x * y element-wise, working on quantized buffers.
+ * 
+ * @param <V> data type for {@code z()} output
+ */
+@Operator(group = "math")
+public final class QuantizedMul<V extends TType> extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new QuantizedMul operation.
+   * 
+   * @param scope current scope
+   * @param x 
+   * @param y 
+   * @param minX The float value that the lowest quantized `x` value represents.
+   * @param maxX The float value that the highest quantized `x` value represents.
+   * @param minY The float value that the lowest quantized `y` value represents.
+   * @param maxY The float value that the highest quantized `y` value represents.
+   * @param Toutput 
+   * @return a new instance of QuantizedMul
+   */
+  @Endpoint(describeByClass = true)
+  public static <V extends TType, T extends TType, U extends TType> QuantizedMul<V> create(Scope scope, Operand<T> x, Operand<U> y, Operand<TFloat32> minX, Operand<TFloat32> maxX, Operand<TFloat32> minY, Operand<TFloat32> maxY, Class<V> Toutput) {
+    OperationBuilder opBuilder = scope.env().opBuilder("QuantizedMul", scope.makeOpName("QuantizedMul"));
+    opBuilder.addInput(x.asOutput());
+    opBuilder.addInput(y.asOutput());
+    opBuilder.addInput(minX.asOutput());
+    opBuilder.addInput(maxX.asOutput());
+    opBuilder.addInput(minY.asOutput());
+    opBuilder.addInput(maxY.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("Toutput", Toutput);
+    return new QuantizedMul<V>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<V> z() {
+    return z;
+  }
+  
+  /**
+   * The float value that the lowest quantized output value represents.
+   */
+  public Output<TFloat32> minZ() {
+    return minZ;
+  }
+  
+  /**
+   * The float value that the highest quantized output value represents.
+   * <p>
+   * <i>NOTE</i>: `math.QuantizedMul` supports limited forms of broadcasting. More about
+   * broadcasting [here](http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html)
+   */
+  public Output<TFloat32> maxZ() {
+    return maxZ;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "QuantizedMul";
+  
+  private Output<V> z;
+  private Output<TFloat32> minZ;
+  private Output<TFloat32> maxZ;
+  
+  private QuantizedMul(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    z = operation.output(outputIdx++);
+    minZ = operation.output(outputIdx++);
+    maxZ = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Real.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Real.java
new file mode 100644
index 00000000000..bd00e9a4ae6
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Real.java
@@ -0,0 +1,102 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.math;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Returns the real part of a complex number.
+ * <p>
+ * Given a tensor `input` of complex numbers, this operation returns a tensor of
+ * type `float` that is the real part of each element in `input`. All elements in
+ * `input` must be complex numbers of the form \\(a + bj\\), where <i>a</i> is the real
+ *  part returned by this operation and <i>b</i> is the imaginary part.
+ * <p>
+ * For example:
+ * <pre>{@code
+ * # tensor 'input' is [-2.25 + 4.75j, 3.25 + 5.75j]
+ * tf.real(input) ==> [-2.25, 3.25]
+ * }</pre>
+ * 
+ * 
+ * @param <U> data type for {@code output()} output
+ */
+@Operator(group = "math")
+public final class Real<U extends TNumber> extends RawOp implements Operand<U> {
+  
+  /**
+   * Factory method to create a class wrapping a new Real operation.
+   * 
+   * @param scope current scope
+   * @param input 
+   * @param Tout 
+   * @return a new instance of Real
+   */
+  @Endpoint(describeByClass = true)
+  public static <U extends TNumber, T extends TType> Real<U> create(Scope scope, Operand<T> input, Class<U> Tout) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Real", scope.makeOpName("Real"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("Tout", Tout);
+    return new Real<U>(opBuilder.build());
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new Real operation using default output types.
+   * 
+   * @param scope current scope
+   * @param input 
+   * @return a new instance of Real
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> Real<TFloat32> create(Scope scope, Operand<T> input) {
+    return create(scope, input, TFloat32.class);
+  }
+  
+  /**
+   */
+  public Output<U> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<U> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Real";
+  
+  private Output<U> output;
+  
+  private Real(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/RealDiv.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/RealDiv.java
new file mode 100644
index 00000000000..ec322c696db
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/RealDiv.java
@@ -0,0 +1,81 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.math;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Returns x / y element-wise for real types.
+ * <p>
+ * If `x` and `y` are reals, this will return the floating-point division.
+ * <p>
+ * <i>NOTE</i>: `Div` supports broadcasting. More about broadcasting
+ * [here](http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html)
+ * 
+ * @param <T> data type for {@code z()} output
+ */
+@Operator(group = "math")
+public final class RealDiv<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new RealDiv operation.
+   * 
+   * @param scope current scope
+   * @param x 
+   * @param y 
+   * @return a new instance of RealDiv
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> RealDiv<T> create(Scope scope, Operand<T> x, Operand<T> y) {
+    OperationBuilder opBuilder = scope.env().opBuilder("RealDiv", scope.makeOpName("RealDiv"));
+    opBuilder.addInput(x.asOutput());
+    opBuilder.addInput(y.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new RealDiv<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> z() {
+    return z;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return z;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "RealDiv";
+  
+  private Output<T> z;
+  
+  private RealDiv(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    z = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Reciprocal.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Reciprocal.java
new file mode 100644
index 00000000000..a0b53da3b7c
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Reciprocal.java
@@ -0,0 +1,76 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.math;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Computes the reciprocal of x element-wise.
+ * <p>
+ * I.e., \\(y = 1 / x\\).
+ * 
+ * @param <T> data type for {@code y()} output
+ */
+@Operator(group = "math")
+public final class Reciprocal<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new Reciprocal operation.
+   * 
+   * @param scope current scope
+   * @param x 
+   * @return a new instance of Reciprocal
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> Reciprocal<T> create(Scope scope, Operand<T> x) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Reciprocal", scope.makeOpName("Reciprocal"));
+    opBuilder.addInput(x.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new Reciprocal<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> y() {
+    return y;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return y;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Reciprocal";
+  
+  private Output<T> y;
+  
+  private Reciprocal(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    y = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/ReciprocalGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/ReciprocalGrad.java
new file mode 100644
index 00000000000..dee41220ba9
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/ReciprocalGrad.java
@@ -0,0 +1,78 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.math;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Computes the gradient for the inverse of `x` wrt its input.
+ * <p>
+ * Specifically, `grad = -dy <i> y</i>y`, where `y = 1/x`, and `dy`
+ * is the corresponding input gradient.
+ * 
+ * @param <T> data type for {@code z()} output
+ */
+public final class ReciprocalGrad<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new ReciprocalGrad operation.
+   * 
+   * @param scope current scope
+   * @param y 
+   * @param dy 
+   * @return a new instance of ReciprocalGrad
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> ReciprocalGrad<T> create(Scope scope, Operand<T> y, Operand<T> dy) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ReciprocalGrad", scope.makeOpName("ReciprocalGrad"));
+    opBuilder.addInput(y.asOutput());
+    opBuilder.addInput(dy.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new ReciprocalGrad<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> z() {
+    return z;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return z;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ReciprocalGrad";
+  
+  private Output<T> z;
+  
+  private ReciprocalGrad(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    z = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/RequantizationRangePerChannel.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/RequantizationRangePerChannel.java
new file mode 100644
index 00000000000..feda79c70b2
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/RequantizationRangePerChannel.java
@@ -0,0 +1,84 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.math;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Computes requantization range per channel.
+ */
+public final class RequantizationRangePerChannel extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new RequantizationRangePerChannel operation.
+   * 
+   * @param scope current scope
+   * @param input The original input tensor.
+   * @param inputMin The minimum value of the input tensor
+   * @param inputMax The maximum value of the input tensor.
+   * @param clipValueMax The maximum value of the output that needs to be clipped.
+   * Example: set this to 6 for Relu6.
+   * @return a new instance of RequantizationRangePerChannel
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> RequantizationRangePerChannel create(Scope scope, Operand<T> input, Operand<TFloat32> inputMin, Operand<TFloat32> inputMax, Float clipValueMax) {
+    OperationBuilder opBuilder = scope.env().opBuilder("RequantizationRangePerChannel", scope.makeOpName("RequantizationRangePerChannel"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(inputMin.asOutput());
+    opBuilder.addInput(inputMax.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("clip_value_max", clipValueMax);
+    return new RequantizationRangePerChannel(opBuilder.build());
+  }
+  
+  /**
+   * The minimum value of the final output tensor
+   */
+  public Output<TFloat32> outputMin() {
+    return outputMin;
+  }
+  
+  /**
+   * The maximum value of the final output tensor.
+   */
+  public Output<TFloat32> outputMax() {
+    return outputMax;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "RequantizationRangePerChannel";
+  
+  private Output<TFloat32> outputMin;
+  private Output<TFloat32> outputMax;
+  
+  private RequantizationRangePerChannel(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    outputMin = operation.output(outputIdx++);
+    outputMax = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/RequantizePerChannel.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/RequantizePerChannel.java
new file mode 100644
index 00000000000..515da442497
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/RequantizePerChannel.java
@@ -0,0 +1,98 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.math;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Requantizes input with min and max values known per channel.
+ * 
+ * @param <U> data type for {@code output()} output
+ */
+public final class RequantizePerChannel<U extends TType> extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new RequantizePerChannel operation.
+   * 
+   * @param scope current scope
+   * @param input The original input tensor.
+   * @param inputMin The minimum value of the input tensor
+   * @param inputMax The maximum value of the input tensor.
+   * @param requestedOutputMin The minimum value of the output tensor requested.
+   * @param requestedOutputMax The maximum value of the output tensor requested.
+   * @param outType The quantized type of output tensor that needs to be converted.
+   * @return a new instance of RequantizePerChannel
+   */
+  @Endpoint(describeByClass = true)
+  public static <U extends TType, T extends TType> RequantizePerChannel<U> create(Scope scope, Operand<T> input, Operand<TFloat32> inputMin, Operand<TFloat32> inputMax, Operand<TFloat32> requestedOutputMin, Operand<TFloat32> requestedOutputMax, Class<U> outType) {
+    OperationBuilder opBuilder = scope.env().opBuilder("RequantizePerChannel", scope.makeOpName("RequantizePerChannel"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(inputMin.asOutput());
+    opBuilder.addInput(inputMax.asOutput());
+    opBuilder.addInput(requestedOutputMin.asOutput());
+    opBuilder.addInput(requestedOutputMax.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("out_type", outType);
+    return new RequantizePerChannel<U>(opBuilder.build());
+  }
+  
+  /**
+   * Output tensor.
+   */
+  public Output<U> output() {
+    return output;
+  }
+  
+  /**
+   * The minimum value of the final output tensor
+   */
+  public Output<TFloat32> outputMin() {
+    return outputMin;
+  }
+  
+  /**
+   * The maximum value of the final output tensor.
+   */
+  public Output<TFloat32> outputMax() {
+    return outputMax;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "RequantizePerChannel";
+  
+  private Output<U> output;
+  private Output<TFloat32> outputMin;
+  private Output<TFloat32> outputMax;
+  
+  private RequantizePerChannel(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+    outputMin = operation.output(outputIdx++);
+    outputMax = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Rint.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Rint.java
new file mode 100644
index 00000000000..849508eba02
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Rint.java
@@ -0,0 +1,84 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.math;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Returns element-wise integer closest to x.
+ * <p>
+ * If the result is midway between two representable values,
+ * the even representable is chosen.
+ * For example:
+ * <pre>{@code
+ * rint(-1.5) ==> -2.0
+ * rint(0.5000001) ==> 1.0
+ * rint([-1.7, -1.5, -0.2, 0.2, 1.5, 1.7, 2.0]) ==> [-2., -2., -0., 0., 2., 2., 2.]
+ * }</pre>
+ * 
+ * 
+ * @param <T> data type for {@code y()} output
+ */
+@Operator(group = "math")
+public final class Rint<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new Rint operation.
+   * 
+   * @param scope current scope
+   * @param x 
+   * @return a new instance of Rint
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> Rint<T> create(Scope scope, Operand<T> x) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Rint", scope.makeOpName("Rint"));
+    opBuilder.addInput(x.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new Rint<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> y() {
+    return y;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return y;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Rint";
+  
+  private Output<T> y;
+  
+  private Rint(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    y = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Round.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Round.java
new file mode 100644
index 00000000000..7f892d28ed0
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Round.java
@@ -0,0 +1,77 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.math;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Rounds the values of a tensor to the nearest integer, element-wise.
+ * <p>
+ * Rounds half to even.  Also known as bankers rounding. If you want to round
+ * according to the current system rounding mode use std::cint.
+ * 
+ * @param <T> data type for {@code y()} output
+ */
+@Operator(group = "math")
+public final class Round<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new Round operation.
+   * 
+   * @param scope current scope
+   * @param x 
+   * @return a new instance of Round
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> Round<T> create(Scope scope, Operand<T> x) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Round", scope.makeOpName("Round"));
+    opBuilder.addInput(x.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new Round<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> y() {
+    return y;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return y;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Round";
+  
+  private Output<T> y;
+  
+  private Round(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    y = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Rsqrt.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Rsqrt.java
new file mode 100644
index 00000000000..de04c3f1204
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Rsqrt.java
@@ -0,0 +1,76 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.math;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Computes reciprocal of square root of x element-wise.
+ * <p>
+ * I.e., \\(y = 1 / \sqrt{x}\\).
+ * 
+ * @param <T> data type for {@code y()} output
+ */
+@Operator(group = "math")
+public final class Rsqrt<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new Rsqrt operation.
+   * 
+   * @param scope current scope
+   * @param x 
+   * @return a new instance of Rsqrt
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> Rsqrt<T> create(Scope scope, Operand<T> x) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Rsqrt", scope.makeOpName("Rsqrt"));
+    opBuilder.addInput(x.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new Rsqrt<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> y() {
+    return y;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return y;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Rsqrt";
+  
+  private Output<T> y;
+  
+  private Rsqrt(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    y = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/RsqrtGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/RsqrtGrad.java
new file mode 100644
index 00000000000..ee0eec068f3
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/RsqrtGrad.java
@@ -0,0 +1,78 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.math;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Computes the gradient for the rsqrt of `x` wrt its input.
+ * <p>
+ * Specifically, `grad = dy * -0.5 * y^3`, where `y = rsqrt(x)`, and `dy`
+ * is the corresponding input gradient.
+ * 
+ * @param <T> data type for {@code z()} output
+ */
+public final class RsqrtGrad<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new RsqrtGrad operation.
+   * 
+   * @param scope current scope
+   * @param y 
+   * @param dy 
+   * @return a new instance of RsqrtGrad
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> RsqrtGrad<T> create(Scope scope, Operand<T> y, Operand<T> dy) {
+    OperationBuilder opBuilder = scope.env().opBuilder("RsqrtGrad", scope.makeOpName("RsqrtGrad"));
+    opBuilder.addInput(y.asOutput());
+    opBuilder.addInput(dy.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new RsqrtGrad<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> z() {
+    return z;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return z;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "RsqrtGrad";
+  
+  private Output<T> z;
+  
+  private RsqrtGrad(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    z = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SegmentMax.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SegmentMax.java
new file mode 100644
index 00000000000..e044212668b
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SegmentMax.java
@@ -0,0 +1,102 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.math;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Computes the maximum along segments of a tensor.
+ * <p>
+ * Read
+ * [the section on segmentation](https://tensorflow.org/api_docs/python/tf/math#Segmentation)
+ * for an explanation of segments.
+ * <p>
+ * Computes a tensor such that
+ * \\(output_i = \max_j(data_j)\\) where `max` is over `j` such
+ * that `segment_ids[j] == i`.
+ * <p>
+ * If the max is empty for a given segment ID `i`, `output[i] = 0`.
+ * <p>
+ * <div style="width:70%; margin:auto; margin-bottom:10px; margin-top:20px;">
+ * <img style="width:100%" src="https://www.tensorflow.org/images/SegmentMax.png" alt>
+ * </div>
+ * <p>
+ * For example:
+ * <pre>{@code
+ * c = tf.constant([[1,2,3,4], [4, 3, 2, 1], [5,6,7,8]])
+ * tf.segment_max(c, tf.constant([0, 0, 1]))
+ * # ==> [[4, 3, 3, 4],
+ * #      [5, 6, 7, 8]]
+ * }</pre>
+ * 
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator(group = "math")
+public final class SegmentMax<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new SegmentMax operation.
+   * 
+   * @param scope current scope
+   * @param data 
+   * @param segmentIds A 1-D tensor whose size is equal to the size of `data`'s
+   * first dimension.  Values should be sorted and can be repeated.
+   * @return a new instance of SegmentMax
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber, U extends TNumber> SegmentMax<T> create(Scope scope, Operand<T> data, Operand<U> segmentIds) {
+    OperationBuilder opBuilder = scope.env().opBuilder("SegmentMax", scope.makeOpName("SegmentMax"));
+    opBuilder.addInput(data.asOutput());
+    opBuilder.addInput(segmentIds.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new SegmentMax<T>(opBuilder.build());
+  }
+  
+  /**
+   * Has same shape as data, except for dimension 0 which
+   * has size `k`, the number of segments.
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "SegmentMax";
+  
+  private Output<T> output;
+  
+  private SegmentMax(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SegmentMean.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SegmentMean.java
new file mode 100644
index 00000000000..2fe16ce7e12
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SegmentMean.java
@@ -0,0 +1,104 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.math;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Computes the mean along segments of a tensor.
+ * <p>
+ * Read
+ * [the section on segmentation](https://tensorflow.org/api_docs/python/tf/math#Segmentation)
+ * for an explanation of segments.
+ * <p>
+ * Computes a tensor such that
+ * \\(output_i = \frac{\sum_j data_j}{N}\\) where `mean` is
+ * over `j` such that `segment_ids[j] == i` and `N` is the total number of
+ * values summed.
+ * <p>
+ * If the mean is empty for a given segment ID `i`, `output[i] = 0`.
+ * <p>
+ * <div style="width:70%; margin:auto; margin-bottom:10px; margin-top:20px;">
+ * <img style="width:100%" src="https://www.tensorflow.org/images/SegmentMean.png" alt>
+ * </div>
+ * <p>
+ * For example:
+ * <pre>{@code
+ * c = tf.constant([[1.0,2,3,4], [4, 3, 2, 1], [5,6,7,8]])
+ * tf.segment_mean(c, tf.constant([0, 0, 1]))
+ * # ==> [[2.5, 2.5, 2.5, 2.5],
+ * #      [5, 6, 7, 8]]
+ * }</pre>
+ * 
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator(group = "math")
+public final class SegmentMean<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new SegmentMean operation.
+   * 
+   * @param scope current scope
+   * @param data 
+   * @param segmentIds A 1-D tensor whose size is equal to the size of `data`'s
+   * first dimension.  Values should be sorted and can be repeated.
+   * @return a new instance of SegmentMean
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType, U extends TNumber> SegmentMean<T> create(Scope scope, Operand<T> data, Operand<U> segmentIds) {
+    OperationBuilder opBuilder = scope.env().opBuilder("SegmentMean", scope.makeOpName("SegmentMean"));
+    opBuilder.addInput(data.asOutput());
+    opBuilder.addInput(segmentIds.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new SegmentMean<T>(opBuilder.build());
+  }
+  
+  /**
+   * Has same shape as data, except for dimension 0 which
+   * has size `k`, the number of segments.
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "SegmentMean";
+  
+  private Output<T> output;
+  
+  private SegmentMean(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SegmentMin.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SegmentMin.java
new file mode 100644
index 00000000000..b628c6f5eb1
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SegmentMin.java
@@ -0,0 +1,102 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.math;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Computes the minimum along segments of a tensor.
+ * <p>
+ * Read
+ * [the section on segmentation](https://tensorflow.org/api_docs/python/tf/math#Segmentation)
+ * for an explanation of segments.
+ * <p>
+ * Computes a tensor such that
+ * \\(output_i = \min_j(data_j)\\) where `min` is over `j` such
+ * that `segment_ids[j] == i`.
+ * <p>
+ * If the min is empty for a given segment ID `i`, `output[i] = 0`.
+ * <p>
+ * <div style="width:70%; margin:auto; margin-bottom:10px; margin-top:20px;">
+ * <img style="width:100%" src="https://www.tensorflow.org/images/SegmentMin.png" alt>
+ * </div>
+ * <p>
+ * For example:
+ * <pre>{@code
+ * c = tf.constant([[1,2,3,4], [4, 3, 2, 1], [5,6,7,8]])
+ * tf.segment_min(c, tf.constant([0, 0, 1]))
+ * # ==> [[1, 2, 2, 1],
+ * #      [5, 6, 7, 8]]
+ * }</pre>
+ * 
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator(group = "math")
+public final class SegmentMin<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new SegmentMin operation.
+   * 
+   * @param scope current scope
+   * @param data 
+   * @param segmentIds A 1-D tensor whose size is equal to the size of `data`'s
+   * first dimension.  Values should be sorted and can be repeated.
+   * @return a new instance of SegmentMin
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber, U extends TNumber> SegmentMin<T> create(Scope scope, Operand<T> data, Operand<U> segmentIds) {
+    OperationBuilder opBuilder = scope.env().opBuilder("SegmentMin", scope.makeOpName("SegmentMin"));
+    opBuilder.addInput(data.asOutput());
+    opBuilder.addInput(segmentIds.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new SegmentMin<T>(opBuilder.build());
+  }
+  
+  /**
+   * Has same shape as data, except for dimension 0 which
+   * has size `k`, the number of segments.
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "SegmentMin";
+  
+  private Output<T> output;
+  
+  private SegmentMin(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SegmentProd.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SegmentProd.java
new file mode 100644
index 00000000000..0f1179072e5
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SegmentProd.java
@@ -0,0 +1,103 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.math;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Computes the product along segments of a tensor.
+ * <p>
+ * Read
+ * [the section on segmentation](https://tensorflow.org/api_docs/python/tf/math#Segmentation)
+ * for an explanation of segments.
+ * <p>
+ * Computes a tensor such that
+ * \\(output_i = \prod_j data_j\\) where the product is over `j` such
+ * that `segment_ids[j] == i`.
+ * <p>
+ * If the product is empty for a given segment ID `i`, `output[i] = 1`.
+ * <p>
+ * <div style="width:70%; margin:auto; margin-bottom:10px; margin-top:20px;">
+ * <img style="width:100%" src="https://www.tensorflow.org/images/SegmentProd.png" alt>
+ * </div>
+ * <p>
+ * For example:
+ * <pre>{@code
+ * c = tf.constant([[1,2,3,4], [4, 3, 2, 1], [5,6,7,8]])
+ * tf.segment_prod(c, tf.constant([0, 0, 1]))
+ * # ==> [[4, 6, 6, 4],
+ * #      [5, 6, 7, 8]]
+ * }</pre>
+ * 
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator(group = "math")
+public final class SegmentProd<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new SegmentProd operation.
+   * 
+   * @param scope current scope
+   * @param data 
+   * @param segmentIds A 1-D tensor whose size is equal to the size of `data`'s
+   * first dimension.  Values should be sorted and can be repeated.
+   * @return a new instance of SegmentProd
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType, U extends TNumber> SegmentProd<T> create(Scope scope, Operand<T> data, Operand<U> segmentIds) {
+    OperationBuilder opBuilder = scope.env().opBuilder("SegmentProd", scope.makeOpName("SegmentProd"));
+    opBuilder.addInput(data.asOutput());
+    opBuilder.addInput(segmentIds.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new SegmentProd<T>(opBuilder.build());
+  }
+  
+  /**
+   * Has same shape as data, except for dimension 0 which
+   * has size `k`, the number of segments.
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "SegmentProd";
+  
+  private Output<T> output;
+  
+  private SegmentProd(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SegmentSum.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SegmentSum.java
new file mode 100644
index 00000000000..fceec1bfe43
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SegmentSum.java
@@ -0,0 +1,103 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.math;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Computes the sum along segments of a tensor.
+ * <p>
+ * Read
+ * [the section on segmentation](https://tensorflow.org/api_docs/python/tf/math#Segmentation)
+ * for an explanation of segments.
+ * <p>
+ * Computes a tensor such that
+ * \\(output_i = \sum_j data_j\\) where sum is over `j` such
+ * that `segment_ids[j] == i`.
+ * <p>
+ * If the sum is empty for a given segment ID `i`, `output[i] = 0`.
+ * <p>
+ * <div style="width:70%; margin:auto; margin-bottom:10px; margin-top:20px;">
+ * <img style="width:100%" src="https://www.tensorflow.org/images/SegmentSum.png" alt>
+ * </div>
+ * <p>
+ * For example:
+ * <pre>{@code
+ * c = tf.constant([[1,2,3,4], [4, 3, 2, 1], [5,6,7,8]])
+ * tf.segment_sum(c, tf.constant([0, 0, 1]))
+ * # ==> [[5, 5, 5, 5],
+ * #      [5, 6, 7, 8]]
+ * }</pre>
+ * 
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator(group = "math")
+public final class SegmentSum<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new SegmentSum operation.
+   * 
+   * @param scope current scope
+   * @param data 
+   * @param segmentIds A 1-D tensor whose size is equal to the size of `data`'s
+   * first dimension.  Values should be sorted and can be repeated.
+   * @return a new instance of SegmentSum
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType, U extends TNumber> SegmentSum<T> create(Scope scope, Operand<T> data, Operand<U> segmentIds) {
+    OperationBuilder opBuilder = scope.env().opBuilder("SegmentSum", scope.makeOpName("SegmentSum"));
+    opBuilder.addInput(data.asOutput());
+    opBuilder.addInput(segmentIds.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new SegmentSum<T>(opBuilder.build());
+  }
+  
+  /**
+   * Has same shape as data, except for dimension 0 which
+   * has size `k`, the number of segments.
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "SegmentSum";
+  
+  private Output<T> output;
+  
+  private SegmentSum(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Sigmoid.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Sigmoid.java
new file mode 100644
index 00000000000..a54e4ae51ea
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Sigmoid.java
@@ -0,0 +1,76 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.math;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Computes sigmoid of `x` element-wise.
+ * <p>
+ * Specifically, `y = 1 / (1 + exp(-x))`.
+ * 
+ * @param <T> data type for {@code y()} output
+ */
+@Operator(group = "math")
+public final class Sigmoid<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new Sigmoid operation.
+   * 
+   * @param scope current scope
+   * @param x 
+   * @return a new instance of Sigmoid
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> Sigmoid<T> create(Scope scope, Operand<T> x) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Sigmoid", scope.makeOpName("Sigmoid"));
+    opBuilder.addInput(x.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new Sigmoid<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> y() {
+    return y;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return y;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Sigmoid";
+  
+  private Output<T> y;
+  
+  private Sigmoid(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    y = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SigmoidGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SigmoidGrad.java
new file mode 100644
index 00000000000..125397bfd65
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SigmoidGrad.java
@@ -0,0 +1,78 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.math;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Computes the gradient of the sigmoid of `x` wrt its input.
+ * <p>
+ * Specifically, `grad = dy * y * (1 - y)`, where `y = sigmoid(x)`, and
+ * `dy` is the corresponding input gradient.
+ * 
+ * @param <T> data type for {@code z()} output
+ */
+public final class SigmoidGrad<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new SigmoidGrad operation.
+   * 
+   * @param scope current scope
+   * @param y 
+   * @param dy 
+   * @return a new instance of SigmoidGrad
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> SigmoidGrad<T> create(Scope scope, Operand<T> y, Operand<T> dy) {
+    OperationBuilder opBuilder = scope.env().opBuilder("SigmoidGrad", scope.makeOpName("SigmoidGrad"));
+    opBuilder.addInput(y.asOutput());
+    opBuilder.addInput(dy.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new SigmoidGrad<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> z() {
+    return z;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return z;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "SigmoidGrad";
+  
+  private Output<T> z;
+  
+  private SigmoidGrad(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    z = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Sign.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Sign.java
new file mode 100644
index 00000000000..c9d8609872e
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Sign.java
@@ -0,0 +1,82 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.math;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Returns an element-wise indication of the sign of a number.
+ * <p>
+ * `y = sign(x) = -1` if `x < 0`; 0 if `x == 0`; 1 if `x > 0`.
+ * <p>
+ * For complex numbers, `y = sign(x) = x / |x|` if `x != 0`, otherwise `y = 0`.
+ * <p>
+ * Example usage:
+ * >>> tf.math.sign([0., 2., -3.])
+ * <tf.Tensor: shape=(3,), dtype=float32, numpy=array([ 0.,  1., -1.], dtype=float32)>
+ * 
+ * @param <T> data type for {@code y()} output
+ */
+@Operator(group = "math")
+public final class Sign<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new Sign operation.
+   * 
+   * @param scope current scope
+   * @param x 
+   * @return a new instance of Sign
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> Sign<T> create(Scope scope, Operand<T> x) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Sign", scope.makeOpName("Sign"));
+    opBuilder.addInput(x.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new Sign<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> y() {
+    return y;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return y;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Sign";
+  
+  private Output<T> y;
+  
+  private Sign(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    y = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Sin.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Sin.java
new file mode 100644
index 00000000000..a17ea9d288e
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Sin.java
@@ -0,0 +1,84 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.math;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Computes sine of x element-wise.
+ * <p>
+ *   Given an input tensor, this function computes sine of every
+ *   element in the tensor. Input range is `(-inf, inf)` and
+ *   output range is `[-1,1]`.
+ * <p>
+ *   <pre>{@code
+ *   x = tf.constant([-float("inf"), -9, -0.5, 1, 1.2, 200, 10, float("inf")])
+ *   tf.math.sin(x) ==> [nan -0.4121185 -0.47942555 0.84147096 0.9320391 -0.87329733 -0.54402107 nan]
+ *   }</pre>
+ * 
+ * 
+ * @param <T> data type for {@code y()} output
+ */
+@Operator(group = "math")
+public final class Sin<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new Sin operation.
+   * 
+   * @param scope current scope
+   * @param x 
+   * @return a new instance of Sin
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> Sin<T> create(Scope scope, Operand<T> x) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Sin", scope.makeOpName("Sin"));
+    opBuilder.addInput(x.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new Sin<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> y() {
+    return y;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return y;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Sin";
+  
+  private Output<T> y;
+  
+  private Sin(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    y = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Sinh.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Sinh.java
new file mode 100644
index 00000000000..613d983eef5
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Sinh.java
@@ -0,0 +1,84 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.math;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Computes hyperbolic sine of x element-wise.
+ * <p>
+ *   Given an input tensor, this function computes hyperbolic sine of every
+ *   element in the tensor. Input range is `[-inf,inf]` and output range
+ *   is `[-inf,inf]`.
+ * <p>
+ *   <pre>{@code
+ *   x = tf.constant([-float("inf"), -9, -0.5, 1, 1.2, 2, 10, float("inf")])
+ *   tf.math.sinh(x) ==> [-inf -4.0515420e+03 -5.2109528e-01 1.1752012e+00 1.5094614e+00 3.6268604e+00 1.1013232e+04 inf]
+ *   }</pre>
+ * 
+ * 
+ * @param <T> data type for {@code y()} output
+ */
+@Operator(group = "math")
+public final class Sinh<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new Sinh operation.
+   * 
+   * @param scope current scope
+   * @param x 
+   * @return a new instance of Sinh
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> Sinh<T> create(Scope scope, Operand<T> x) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Sinh", scope.makeOpName("Sinh"));
+    opBuilder.addInput(x.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new Sinh<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> y() {
+    return y;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return y;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Sinh";
+  
+  private Output<T> y;
+  
+  private Sinh(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    y = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SobolSample.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SobolSample.java
new file mode 100644
index 00000000000..e7c9c43d20d
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SobolSample.java
@@ -0,0 +1,103 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.math;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Generates points from the Sobol sequence.
+ * <p>
+ * Creates a Sobol sequence with `num_results` samples. Each sample has dimension
+ * `dim`. Skips the first `skip` samples.
+ * 
+ * @param <T> data type for {@code samples()} output
+ */
+public final class SobolSample<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new SobolSample operation.
+   * 
+   * @param scope current scope
+   * @param dim Positive scalar `Tensor` representing each sample's dimension.
+   * @param numResults Positive scalar `Tensor` of dtype int32. The number of Sobol points to return
+   * in the output.
+   * @param skip Positive scalar `Tensor` of dtype int32. The number of initial points of the
+   * Sobol sequence to skip.
+   * @param dtype The type of the sample. One of: `float32` or `float64`.
+   * @return a new instance of SobolSample
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> SobolSample<T> create(Scope scope, Operand<TInt32> dim, Operand<TInt32> numResults, Operand<TInt32> skip, Class<T> dtype) {
+    OperationBuilder opBuilder = scope.env().opBuilder("SobolSample", scope.makeOpName("SobolSample"));
+    opBuilder.addInput(dim.asOutput());
+    opBuilder.addInput(numResults.asOutput());
+    opBuilder.addInput(skip.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("dtype", dtype);
+    return new SobolSample<T>(opBuilder.build());
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new SobolSample operation using default output types.
+   * 
+   * @param scope current scope
+   * @param dim Positive scalar `Tensor` representing each sample's dimension.
+   * @param numResults Positive scalar `Tensor` of dtype int32. The number of Sobol points to return
+   * in the output.
+   * @param skip Positive scalar `Tensor` of dtype int32. The number of initial points of the
+   * Sobol sequence to skip.
+   * @return a new instance of SobolSample
+   */
+  @Endpoint(describeByClass = true)
+  public static SobolSample<TFloat32> create(Scope scope, Operand<TInt32> dim, Operand<TInt32> numResults, Operand<TInt32> skip) {
+    return create(scope, dim, numResults, skip, TFloat32.class);
+  }
+  
+  /**
+   * `Tensor` of samples from Sobol sequence with `shape` [num_results, dim].
+   */
+  public Output<T> samples() {
+    return samples;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return samples;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "SobolSample";
+  
+  private Output<T> samples;
+  
+  private SobolSample(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    samples = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Softplus.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Softplus.java
new file mode 100644
index 00000000000..5065f948cc1
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Softplus.java
@@ -0,0 +1,74 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.math;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Computes softplus: `log(exp(features) + 1)`.
+ * 
+ * @param <T> data type for {@code activations()} output
+ */
+@Operator(group = "math")
+public final class Softplus<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new Softplus operation.
+   * 
+   * @param scope current scope
+   * @param features 
+   * @return a new instance of Softplus
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> Softplus<T> create(Scope scope, Operand<T> features) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Softplus", scope.makeOpName("Softplus"));
+    opBuilder.addInput(features.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new Softplus<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> activations() {
+    return activations;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return activations;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Softplus";
+  
+  private Output<T> activations;
+  
+  private Softplus(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    activations = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SoftplusGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SoftplusGrad.java
new file mode 100644
index 00000000000..100243c60e4
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SoftplusGrad.java
@@ -0,0 +1,76 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.math;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Computes softplus gradients for a softplus operation.
+ * 
+ * @param <T> data type for {@code backprops()} output
+ */
+public final class SoftplusGrad<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new SoftplusGrad operation.
+   * 
+   * @param scope current scope
+   * @param gradients The backpropagated gradients to the corresponding softplus operation.
+   * @param features The features passed as input to the corresponding softplus operation.
+   * @return a new instance of SoftplusGrad
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> SoftplusGrad<T> create(Scope scope, Operand<T> gradients, Operand<T> features) {
+    OperationBuilder opBuilder = scope.env().opBuilder("SoftplusGrad", scope.makeOpName("SoftplusGrad"));
+    opBuilder.addInput(gradients.asOutput());
+    opBuilder.addInput(features.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new SoftplusGrad<T>(opBuilder.build());
+  }
+  
+  /**
+   * The gradients: `gradients / (1 + exp(-features))`.
+   */
+  public Output<T> backprops() {
+    return backprops;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return backprops;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "SoftplusGrad";
+  
+  private Output<T> backprops;
+  
+  private SoftplusGrad(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    backprops = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Sqrt.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Sqrt.java
new file mode 100644
index 00000000000..de62810d7ce
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Sqrt.java
@@ -0,0 +1,76 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.math;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Computes square root of x element-wise.
+ * <p>
+ * I.e., \\(y = \sqrt{x} = x^{1/2}\\).
+ * 
+ * @param <T> data type for {@code y()} output
+ */
+@Operator(group = "math")
+public final class Sqrt<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new Sqrt operation.
+   * 
+   * @param scope current scope
+   * @param x 
+   * @return a new instance of Sqrt
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> Sqrt<T> create(Scope scope, Operand<T> x) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Sqrt", scope.makeOpName("Sqrt"));
+    opBuilder.addInput(x.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new Sqrt<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> y() {
+    return y;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return y;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Sqrt";
+  
+  private Output<T> y;
+  
+  private Sqrt(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    y = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SqrtGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SqrtGrad.java
new file mode 100644
index 00000000000..2c4b301a60b
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SqrtGrad.java
@@ -0,0 +1,78 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.math;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Computes the gradient for the sqrt of `x` wrt its input.
+ * <p>
+ * Specifically, `grad = dy * 0.5 / y`, where `y = sqrt(x)`, and `dy`
+ * is the corresponding input gradient.
+ * 
+ * @param <T> data type for {@code z()} output
+ */
+public final class SqrtGrad<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new SqrtGrad operation.
+   * 
+   * @param scope current scope
+   * @param y 
+   * @param dy 
+   * @return a new instance of SqrtGrad
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> SqrtGrad<T> create(Scope scope, Operand<T> y, Operand<T> dy) {
+    OperationBuilder opBuilder = scope.env().opBuilder("SqrtGrad", scope.makeOpName("SqrtGrad"));
+    opBuilder.addInput(y.asOutput());
+    opBuilder.addInput(dy.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new SqrtGrad<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> z() {
+    return z;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return z;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "SqrtGrad";
+  
+  private Output<T> z;
+  
+  private SqrtGrad(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    z = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Square.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Square.java
new file mode 100644
index 00000000000..01ccd098cc8
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Square.java
@@ -0,0 +1,76 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.math;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Computes square of x element-wise.
+ * <p>
+ * I.e., \\(y = x * x = x^2\\).
+ * 
+ * @param <T> data type for {@code y()} output
+ */
+@Operator(group = "math")
+public final class Square<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new Square operation.
+   * 
+   * @param scope current scope
+   * @param x 
+   * @return a new instance of Square
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> Square<T> create(Scope scope, Operand<T> x) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Square", scope.makeOpName("Square"));
+    opBuilder.addInput(x.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new Square<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> y() {
+    return y;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return y;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Square";
+  
+  private Output<T> y;
+  
+  private Square(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    y = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SquaredDifference.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SquaredDifference.java
new file mode 100644
index 00000000000..bed2fe9bb77
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/SquaredDifference.java
@@ -0,0 +1,79 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.math;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Returns (x - y)(x - y) element-wise.
+ * <p>
+ * <i>NOTE</i>: `math.SquaredDifference` supports broadcasting. More about broadcasting
+ * [here](http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html)
+ * 
+ * @param <T> data type for {@code z()} output
+ */
+@Operator(group = "math")
+public final class SquaredDifference<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new SquaredDifference operation.
+   * 
+   * @param scope current scope
+   * @param x 
+   * @param y 
+   * @return a new instance of SquaredDifference
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> SquaredDifference<T> create(Scope scope, Operand<T> x, Operand<T> y) {
+    OperationBuilder opBuilder = scope.env().opBuilder("SquaredDifference", scope.makeOpName("SquaredDifference"));
+    opBuilder.addInput(x.asOutput());
+    opBuilder.addInput(y.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new SquaredDifference<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> z() {
+    return z;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return z;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "SquaredDifference";
+  
+  private Output<T> z;
+  
+  private SquaredDifference(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    z = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Sub.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Sub.java
new file mode 100644
index 00000000000..3578a962622
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Sub.java
@@ -0,0 +1,79 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.math;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Returns x - y element-wise.
+ * <p>
+ * <i>NOTE</i>: `math.Sub` supports broadcasting. More about broadcasting
+ * [here](http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html)
+ * 
+ * @param <T> data type for {@code z()} output
+ */
+@Operator(group = "math")
+public final class Sub<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new Sub operation.
+   * 
+   * @param scope current scope
+   * @param x 
+   * @param y 
+   * @return a new instance of Sub
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> Sub<T> create(Scope scope, Operand<T> x, Operand<T> y) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Sub", scope.makeOpName("Sub"));
+    opBuilder.addInput(x.asOutput());
+    opBuilder.addInput(y.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new Sub<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> z() {
+    return z;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return z;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Sub";
+  
+  private Output<T> z;
+  
+  private Sub(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    z = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Tan.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Tan.java
new file mode 100644
index 00000000000..f09d909e8cb
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Tan.java
@@ -0,0 +1,85 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.math;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Computes tan of x element-wise.
+ * <p>
+ *   Given an input tensor, this function computes tangent of every
+ *   element in the tensor. Input range is `(-inf, inf)` and
+ *   output range is `(-inf, inf)`. If input lies outside the boundary, `nan`
+ *   is returned.
+ * <p>
+ *   <pre>{@code
+ *   x = tf.constant([-float("inf"), -9, -0.5, 1, 1.2, 200, 10000, float("inf")])
+ *   tf.math.tan(x) ==> [nan 0.45231566 -0.5463025 1.5574077 2.572152 -1.7925274 0.32097113 nan]
+ *   }</pre>
+ * 
+ * 
+ * @param <T> data type for {@code y()} output
+ */
+@Operator(group = "math")
+public final class Tan<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new Tan operation.
+   * 
+   * @param scope current scope
+   * @param x 
+   * @return a new instance of Tan
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> Tan<T> create(Scope scope, Operand<T> x) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Tan", scope.makeOpName("Tan"));
+    opBuilder.addInput(x.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new Tan<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> y() {
+    return y;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return y;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Tan";
+  
+  private Output<T> y;
+  
+  private Tan(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    y = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Tanh.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Tanh.java
new file mode 100644
index 00000000000..04d9d1e092c
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Tanh.java
@@ -0,0 +1,84 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.math;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Computes hyperbolic tangent of `x` element-wise.
+ * <p>
+ *   Given an input tensor, this function computes hyperbolic tangent of every
+ *   element in the tensor. Input range is `[-inf, inf]` and
+ *   output range is `[-1,1]`.
+ * <p>
+ *   <pre>{@code
+ *   x = tf.constant([-float("inf"), -5, -0.5, 1, 1.2, 2, 3, float("inf")])
+ *   tf.math.tanh(x) ==> [-1. -0.99990916 -0.46211717 0.7615942 0.8336547 0.9640276 0.9950547 1.]
+ *   }</pre>
+ * 
+ * 
+ * @param <T> data type for {@code y()} output
+ */
+@Operator(group = "math")
+public final class Tanh<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new Tanh operation.
+   * 
+   * @param scope current scope
+   * @param x 
+   * @return a new instance of Tanh
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> Tanh<T> create(Scope scope, Operand<T> x) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Tanh", scope.makeOpName("Tanh"));
+    opBuilder.addInput(x.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new Tanh<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> y() {
+    return y;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return y;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Tanh";
+  
+  private Output<T> y;
+  
+  private Tanh(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    y = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/TanhGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/TanhGrad.java
new file mode 100644
index 00000000000..273c5a1f76d
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/TanhGrad.java
@@ -0,0 +1,78 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.math;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Computes the gradient for the tanh of `x` wrt its input.
+ * <p>
+ * Specifically, `grad = dy <i> (1 - y</i>y)`, where `y = tanh(x)`, and `dy`
+ * is the corresponding input gradient.
+ * 
+ * @param <T> data type for {@code z()} output
+ */
+public final class TanhGrad<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new TanhGrad operation.
+   * 
+   * @param scope current scope
+   * @param y 
+   * @param dy 
+   * @return a new instance of TanhGrad
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> TanhGrad<T> create(Scope scope, Operand<T> y, Operand<T> dy) {
+    OperationBuilder opBuilder = scope.env().opBuilder("TanhGrad", scope.makeOpName("TanhGrad"));
+    opBuilder.addInput(y.asOutput());
+    opBuilder.addInput(dy.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new TanhGrad<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> z() {
+    return z;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return z;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "TanhGrad";
+  
+  private Output<T> z;
+  
+  private TanhGrad(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    z = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/TruncateDiv.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/TruncateDiv.java
new file mode 100644
index 00000000000..5cd4b52fbed
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/TruncateDiv.java
@@ -0,0 +1,84 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.math;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Returns x / y element-wise for integer types.
+ * <p>
+ * Truncation designates that negative numbers will round fractional quantities
+ * toward zero. I.e. -7 / 5 = -1. This matches C semantics but it is different
+ * than Python semantics. See `FloorDiv` for a division function that matches
+ * Python Semantics.
+ * <p>
+ * <i>NOTE</i>: `math.TruncateDiv` supports broadcasting. More about broadcasting
+ * [here](http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html)
+ * 
+ * @param <T> data type for {@code z()} output
+ */
+@Operator(group = "math")
+public final class TruncateDiv<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new TruncateDiv operation.
+   * 
+   * @param scope current scope
+   * @param x 
+   * @param y 
+   * @return a new instance of TruncateDiv
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> TruncateDiv<T> create(Scope scope, Operand<T> x, Operand<T> y) {
+    OperationBuilder opBuilder = scope.env().opBuilder("TruncateDiv", scope.makeOpName("TruncateDiv"));
+    opBuilder.addInput(x.asOutput());
+    opBuilder.addInput(y.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new TruncateDiv<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> z() {
+    return z;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return z;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "TruncateDiv";
+  
+  private Output<T> z;
+  
+  private TruncateDiv(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    z = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/TruncateMod.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/TruncateMod.java
new file mode 100644
index 00000000000..0fb4397c881
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/TruncateMod.java
@@ -0,0 +1,82 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.math;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Returns element-wise remainder of division. This emulates C semantics in that
+ * <p>
+ * the result here is consistent with a truncating divide. E.g. `truncate(x / y) *
+ * y + truncate_mod(x, y) = x`.
+ * <p>
+ * <i>NOTE</i>: `math.TruncateMod` supports broadcasting. More about broadcasting
+ * [here](http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html)
+ * 
+ * @param <T> data type for {@code z()} output
+ */
+@Operator(group = "math")
+public final class TruncateMod<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new TruncateMod operation.
+   * 
+   * @param scope current scope
+   * @param x 
+   * @param y 
+   * @return a new instance of TruncateMod
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> TruncateMod<T> create(Scope scope, Operand<T> x, Operand<T> y) {
+    OperationBuilder opBuilder = scope.env().opBuilder("TruncateMod", scope.makeOpName("TruncateMod"));
+    opBuilder.addInput(x.asOutput());
+    opBuilder.addInput(y.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new TruncateMod<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> z() {
+    return z;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return z;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "TruncateMod";
+  
+  private Output<T> z;
+  
+  private TruncateMod(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    z = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/UnsortedSegmentMax.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/UnsortedSegmentMax.java
new file mode 100644
index 00000000000..733e3105846
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/UnsortedSegmentMax.java
@@ -0,0 +1,112 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.math;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Computes the maximum along segments of a tensor.
+ * <p>
+ * Read
+ * [the section on segmentation](https://tensorflow.org/api_docs/python/tf/math#Segmentation)
+ * for an explanation of segments.
+ * <p>
+ * This operator is similar to the unsorted segment sum operator found
+ * [(here)](../../../api_docs/python/math_ops.md#UnsortedSegmentSum).
+ * Instead of computing the sum over segments, it computes the maximum such that:
+ * <p>
+ * \\(output_i = \max_{j...} data[j...]\\) where max is over tuples `j...` such
+ * that `segment_ids[j...] == i`.
+ * <p>
+ * If the maximum is empty for a given segment ID `i`, it outputs the smallest
+ * possible value for the specific numeric type,
+ * `output[i] = numeric_limits<T>::lowest()`.
+ * <p>
+ * If the given segment ID `i` is negative, then the corresponding value is
+ * dropped, and will not be included in the result.
+ * <p>
+ * <div style="width:70%; margin:auto; margin-bottom:10px; margin-top:20px;">
+ * <img style="width:100%" src="https://www.tensorflow.org/images/UnsortedSegmentMax.png" alt>
+ * </div>
+ * <p>
+ * For example:
+ * <pre>{@code
+ * c = tf.constant([[1,2,3,4], [5,6,7,8], [4,3,2,1]])
+ * tf.unsorted_segment_max(c, tf.constant([0, 1, 0]), num_segments=2)
+ * # ==> [[ 4,  3, 3, 4],
+ * #       [5,  6, 7, 8]]
+ * }</pre>
+ * 
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator(group = "math")
+public final class UnsortedSegmentMax<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new UnsortedSegmentMax operation.
+   * 
+   * @param scope current scope
+   * @param data 
+   * @param segmentIds A tensor whose shape is a prefix of `data.shape`.
+   * @param numSegments 
+   * @return a new instance of UnsortedSegmentMax
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber, U extends TNumber, V extends TNumber> UnsortedSegmentMax<T> create(Scope scope, Operand<T> data, Operand<U> segmentIds, Operand<V> numSegments) {
+    OperationBuilder opBuilder = scope.env().opBuilder("UnsortedSegmentMax", scope.makeOpName("UnsortedSegmentMax"));
+    opBuilder.addInput(data.asOutput());
+    opBuilder.addInput(segmentIds.asOutput());
+    opBuilder.addInput(numSegments.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new UnsortedSegmentMax<T>(opBuilder.build());
+  }
+  
+  /**
+   * Has same shape as data, except for the first `segment_ids.rank`
+   * dimensions, which are replaced with a single dimension which has size
+   * `num_segments`.
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "UnsortedSegmentMax";
+  
+  private Output<T> output;
+  
+  private UnsortedSegmentMax(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/UnsortedSegmentMin.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/UnsortedSegmentMin.java
new file mode 100644
index 00000000000..408995715a0
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/UnsortedSegmentMin.java
@@ -0,0 +1,106 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.math;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Computes the minimum along segments of a tensor.
+ * <p>
+ * Read
+ * [the section on segmentation](https://tensorflow.org/api_docs/python/tf/math#Segmentation)
+ * for an explanation of segments.
+ * <p>
+ * This operator is similar to the unsorted segment sum operator found
+ * [(here)](../../../api_docs/python/math_ops.md#UnsortedSegmentSum).
+ * Instead of computing the sum over segments, it computes the minimum such that:
+ * <p>
+ * \\(output_i = \min_{j...} data_[j...]\\) where min is over tuples `j...` such
+ * that `segment_ids[j...] == i`.
+ * <p>
+ * If the minimum is empty for a given segment ID `i`, it outputs the largest
+ * possible value for the specific numeric type,
+ * `output[i] = numeric_limits<T>::max()`.
+ * <p>
+ * For example:
+ * <pre>{@code
+ * c = tf.constant([[1,2,3,4], [5,6,7,8], [4,3,2,1]])
+ * tf.unsorted_segment_min(c, tf.constant([0, 1, 0]), num_segments=2)
+ * # ==> [[ 1,  2, 2, 1],
+ * #       [5,  6, 7, 8]]
+ * }</pre>
+ * If the given segment ID `i` is negative, then the corresponding value is
+ * dropped, and will not be included in the result.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator(group = "math")
+public final class UnsortedSegmentMin<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new UnsortedSegmentMin operation.
+   * 
+   * @param scope current scope
+   * @param data 
+   * @param segmentIds A tensor whose shape is a prefix of `data.shape`.
+   * @param numSegments 
+   * @return a new instance of UnsortedSegmentMin
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber, U extends TNumber, V extends TNumber> UnsortedSegmentMin<T> create(Scope scope, Operand<T> data, Operand<U> segmentIds, Operand<V> numSegments) {
+    OperationBuilder opBuilder = scope.env().opBuilder("UnsortedSegmentMin", scope.makeOpName("UnsortedSegmentMin"));
+    opBuilder.addInput(data.asOutput());
+    opBuilder.addInput(segmentIds.asOutput());
+    opBuilder.addInput(numSegments.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new UnsortedSegmentMin<T>(opBuilder.build());
+  }
+  
+  /**
+   * Has same shape as data, except for the first `segment_ids.rank`
+   * dimensions, which are replaced with a single dimension which has size
+   * `num_segments`.
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "UnsortedSegmentMin";
+  
+  private Output<T> output;
+  
+  private UnsortedSegmentMin(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/UnsortedSegmentProd.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/UnsortedSegmentProd.java
new file mode 100644
index 00000000000..c65fc484e4c
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/UnsortedSegmentProd.java
@@ -0,0 +1,106 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.math;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Computes the product along segments of a tensor.
+ * <p>
+ * Read
+ * [the section on segmentation](https://tensorflow.org/api_docs/python/tf/math#Segmentation)
+ * for an explanation of segments.
+ * <p>
+ * This operator is similar to the unsorted segment sum operator found
+ * [(here)](../../../api_docs/python/math_ops.md#UnsortedSegmentSum).
+ * Instead of computing the sum over segments, it computes the product of all
+ * entries belonging to a segment such that:
+ * <p>
+ * \\(output_i = \prod_{j...} data[j...]\\) where the product is over tuples
+ * `j...` such that `segment_ids[j...] == i`.
+ * <p>
+ * For example:
+ * <pre>{@code
+ * c = tf.constant([[1,2,3,4], [5,6,7,8], [4,3,2,1]])
+ * tf.unsorted_segment_prod(c, tf.constant([0, 1, 0]), num_segments=2)
+ * # ==> [[ 4,  6, 6, 4],
+ * #       [5,  6, 7, 8]]
+ * }</pre>
+ * If there is no entry for a given segment ID `i`, it outputs 1.
+ * <p>
+ * If the given segment ID `i` is negative, then the corresponding value is
+ * dropped, and will not be included in the result.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator(group = "math")
+public final class UnsortedSegmentProd<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new UnsortedSegmentProd operation.
+   * 
+   * @param scope current scope
+   * @param data 
+   * @param segmentIds A tensor whose shape is a prefix of `data.shape`.
+   * @param numSegments 
+   * @return a new instance of UnsortedSegmentProd
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType, U extends TNumber, V extends TNumber> UnsortedSegmentProd<T> create(Scope scope, Operand<T> data, Operand<U> segmentIds, Operand<V> numSegments) {
+    OperationBuilder opBuilder = scope.env().opBuilder("UnsortedSegmentProd", scope.makeOpName("UnsortedSegmentProd"));
+    opBuilder.addInput(data.asOutput());
+    opBuilder.addInput(segmentIds.asOutput());
+    opBuilder.addInput(numSegments.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new UnsortedSegmentProd<T>(opBuilder.build());
+  }
+  
+  /**
+   * Has same shape as data, except for the first `segment_ids.rank`
+   * dimensions, which are replaced with a single dimension which has size
+   * `num_segments`.
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "UnsortedSegmentProd";
+  
+  private Output<T> output;
+  
+  private UnsortedSegmentProd(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/UnsortedSegmentSum.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/UnsortedSegmentSum.java
new file mode 100644
index 00000000000..aaaf5b7bd23
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/UnsortedSegmentSum.java
@@ -0,0 +1,109 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.math;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Computes the sum along segments of a tensor.
+ * <p>
+ * Read
+ * [the section on segmentation](https://tensorflow.org/api_docs/python/tf/math#Segmentation)
+ * for an explanation of segments.
+ * <p>
+ * Computes a tensor such that
+ * \\(output[i] = \sum_{j...} data[j...]\\) where the sum is over tuples `j...` such
+ * that `segment_ids[j...] == i`.  Unlike `SegmentSum`, `segment_ids`
+ * need not be sorted and need not cover all values in the full
+ * range of valid values.
+ * <p>
+ * If the sum is empty for a given segment ID `i`, `output[i] = 0`.
+ * If the given segment ID `i` is negative, the value is dropped and will not be
+ * added to the sum of the segment.
+ * <p>
+ * `num_segments` should equal the number of distinct segment IDs.
+ * <p>
+ * <div style="width:70%; margin:auto; margin-bottom:10px; margin-top:20px;">
+ * <img style="width:100%" src="https://www.tensorflow.org/images/UnsortedSegmentSum.png" alt>
+ * </div>
+ * <pre>{@code
+ * c = tf.constant([[1,2,3,4], [5,6,7,8], [4,3,2,1]])
+ * tf.unsorted_segment_sum(c, tf.constant([0, 1, 0]), num_segments=2)
+ * # ==> [[ 5,  5, 5, 5],
+ * #       [5,  6, 7, 8]]
+ * }</pre>
+ * 
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator(group = "math")
+public final class UnsortedSegmentSum<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new UnsortedSegmentSum operation.
+   * 
+   * @param scope current scope
+   * @param data 
+   * @param segmentIds A tensor whose shape is a prefix of `data.shape`.
+   * @param numSegments 
+   * @return a new instance of UnsortedSegmentSum
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType, U extends TNumber, V extends TNumber> UnsortedSegmentSum<T> create(Scope scope, Operand<T> data, Operand<U> segmentIds, Operand<V> numSegments) {
+    OperationBuilder opBuilder = scope.env().opBuilder("UnsortedSegmentSum", scope.makeOpName("UnsortedSegmentSum"));
+    opBuilder.addInput(data.asOutput());
+    opBuilder.addInput(segmentIds.asOutput());
+    opBuilder.addInput(numSegments.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new UnsortedSegmentSum<T>(opBuilder.build());
+  }
+  
+  /**
+   * Has same shape as data, except for the first `segment_ids.rank`
+   * dimensions, which are replaced with a single dimension which has size
+   * `num_segments`.
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "UnsortedSegmentSum";
+  
+  private Output<T> output;
+  
+  private UnsortedSegmentSum(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Xdivy.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Xdivy.java
new file mode 100644
index 00000000000..1a5e57836fc
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Xdivy.java
@@ -0,0 +1,76 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.math;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Returns 0 if x == 0, and x / y otherwise, elementwise.
+ * 
+ * @param <T> data type for {@code z()} output
+ */
+@Operator(group = "math")
+public final class Xdivy<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new Xdivy operation.
+   * 
+   * @param scope current scope
+   * @param x 
+   * @param y 
+   * @return a new instance of Xdivy
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> Xdivy<T> create(Scope scope, Operand<T> x, Operand<T> y) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Xdivy", scope.makeOpName("Xdivy"));
+    opBuilder.addInput(x.asOutput());
+    opBuilder.addInput(y.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new Xdivy<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> z() {
+    return z;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return z;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Xdivy";
+  
+  private Output<T> z;
+  
+  private Xdivy(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    z = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Xlog1py.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Xlog1py.java
new file mode 100644
index 00000000000..a15f29dab04
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Xlog1py.java
@@ -0,0 +1,76 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.math;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Returns 0 if x == 0, and x * log1p(y) otherwise, elementwise.
+ * 
+ * @param <T> data type for {@code z()} output
+ */
+@Operator(group = "math")
+public final class Xlog1py<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new Xlog1py operation.
+   * 
+   * @param scope current scope
+   * @param x 
+   * @param y 
+   * @return a new instance of Xlog1py
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> Xlog1py<T> create(Scope scope, Operand<T> x, Operand<T> y) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Xlog1py", scope.makeOpName("Xlog1py"));
+    opBuilder.addInput(x.asOutput());
+    opBuilder.addInput(y.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new Xlog1py<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> z() {
+    return z;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return z;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Xlog1py";
+  
+  private Output<T> z;
+  
+  private Xlog1py(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    z = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Xlogy.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Xlogy.java
new file mode 100644
index 00000000000..a140c8aa6fe
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Xlogy.java
@@ -0,0 +1,76 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.math;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Returns 0 if x == 0, and x * log(y) otherwise, elementwise.
+ * 
+ * @param <T> data type for {@code z()} output
+ */
+@Operator(group = "math")
+public final class Xlogy<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new Xlogy operation.
+   * 
+   * @param scope current scope
+   * @param x 
+   * @param y 
+   * @return a new instance of Xlogy
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> Xlogy<T> create(Scope scope, Operand<T> x, Operand<T> y) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Xlogy", scope.makeOpName("Xlogy"));
+    opBuilder.addInput(x.asOutput());
+    opBuilder.addInput(y.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new Xlogy<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> z() {
+    return z;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return z;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Xlogy";
+  
+  private Output<T> z;
+  
+  private Xlogy(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    z = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Zeta.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Zeta.java
new file mode 100644
index 00000000000..576af0a1e6e
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/Zeta.java
@@ -0,0 +1,80 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.math;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Compute the Hurwitz zeta function \\(\zeta(x, q)\\).
+ * <p>
+ * The Hurwitz zeta function is defined as:
+ * <p>
+ * \\(\zeta(x, q) = \sum_{n=0}^{\infty} (q + n)^{-x}\\)
+ * 
+ * @param <T> data type for {@code z()} output
+ */
+@Operator(group = "math")
+public final class Zeta<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new Zeta operation.
+   * 
+   * @param scope current scope
+   * @param x 
+   * @param q 
+   * @return a new instance of Zeta
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> Zeta<T> create(Scope scope, Operand<T> x, Operand<T> q) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Zeta", scope.makeOpName("Zeta"));
+    opBuilder.addInput(x.asOutput());
+    opBuilder.addInput(q.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new Zeta<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> z() {
+    return z;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return z;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Zeta";
+  
+  private Output<T> z;
+  
+  private Zeta(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    z = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/erfinv.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/erfinv.java
new file mode 100644
index 00000000000..e74d429e129
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/erfinv.java
@@ -0,0 +1,72 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.math;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * @param <T> data type for {@code y()} output
+ */
+@Operator(group = "math")
+public final class erfinv<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new erfinv operation.
+   * 
+   * @param scope current scope
+   * @param x 
+   * @return a new instance of erfinv
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> erfinv<T> create(Scope scope, Operand<T> x) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Erfinv", scope.makeOpName("erfinv"));
+    opBuilder.addInput(x.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new erfinv<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> y() {
+    return y;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return y;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Erfinv";
+  
+  private Output<T> y;
+  
+  private erfinv(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    y = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/BesselJ0.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/BesselJ0.java
new file mode 100644
index 00000000000..c49c2d69350
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/BesselJ0.java
@@ -0,0 +1,71 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.math.special;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * @param <T> data type for {@code y()} output
+ */
+public final class BesselJ0<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new BesselJ0 operation.
+   * 
+   * @param scope current scope
+   * @param x 
+   * @return a new instance of BesselJ0
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> BesselJ0<T> create(Scope scope, Operand<T> x) {
+    OperationBuilder opBuilder = scope.env().opBuilder("BesselJ0", scope.makeOpName("BesselJ0"));
+    opBuilder.addInput(x.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new BesselJ0<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> y() {
+    return y;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return y;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "BesselJ0";
+  
+  private Output<T> y;
+  
+  private BesselJ0(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    y = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/BesselJ1.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/BesselJ1.java
new file mode 100644
index 00000000000..fb7f29046de
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/BesselJ1.java
@@ -0,0 +1,71 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.math.special;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * @param <T> data type for {@code y()} output
+ */
+public final class BesselJ1<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new BesselJ1 operation.
+   * 
+   * @param scope current scope
+   * @param x 
+   * @return a new instance of BesselJ1
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> BesselJ1<T> create(Scope scope, Operand<T> x) {
+    OperationBuilder opBuilder = scope.env().opBuilder("BesselJ1", scope.makeOpName("BesselJ1"));
+    opBuilder.addInput(x.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new BesselJ1<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> y() {
+    return y;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return y;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "BesselJ1";
+  
+  private Output<T> y;
+  
+  private BesselJ1(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    y = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/BesselK0.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/BesselK0.java
new file mode 100644
index 00000000000..945b617b1dc
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/BesselK0.java
@@ -0,0 +1,71 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.math.special;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * @param <T> data type for {@code y()} output
+ */
+public final class BesselK0<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new BesselK0 operation.
+   * 
+   * @param scope current scope
+   * @param x 
+   * @return a new instance of BesselK0
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> BesselK0<T> create(Scope scope, Operand<T> x) {
+    OperationBuilder opBuilder = scope.env().opBuilder("BesselK0", scope.makeOpName("BesselK0"));
+    opBuilder.addInput(x.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new BesselK0<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> y() {
+    return y;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return y;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "BesselK0";
+  
+  private Output<T> y;
+  
+  private BesselK0(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    y = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/BesselK0e.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/BesselK0e.java
new file mode 100644
index 00000000000..d1a014d4959
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/BesselK0e.java
@@ -0,0 +1,71 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.math.special;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * @param <T> data type for {@code y()} output
+ */
+public final class BesselK0e<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new BesselK0e operation.
+   * 
+   * @param scope current scope
+   * @param x 
+   * @return a new instance of BesselK0e
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> BesselK0e<T> create(Scope scope, Operand<T> x) {
+    OperationBuilder opBuilder = scope.env().opBuilder("BesselK0e", scope.makeOpName("BesselK0e"));
+    opBuilder.addInput(x.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new BesselK0e<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> y() {
+    return y;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return y;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "BesselK0e";
+  
+  private Output<T> y;
+  
+  private BesselK0e(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    y = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/BesselK1.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/BesselK1.java
new file mode 100644
index 00000000000..0cb60fa3d88
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/BesselK1.java
@@ -0,0 +1,71 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.math.special;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * @param <T> data type for {@code y()} output
+ */
+public final class BesselK1<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new BesselK1 operation.
+   * 
+   * @param scope current scope
+   * @param x 
+   * @return a new instance of BesselK1
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> BesselK1<T> create(Scope scope, Operand<T> x) {
+    OperationBuilder opBuilder = scope.env().opBuilder("BesselK1", scope.makeOpName("BesselK1"));
+    opBuilder.addInput(x.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new BesselK1<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> y() {
+    return y;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return y;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "BesselK1";
+  
+  private Output<T> y;
+  
+  private BesselK1(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    y = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/BesselK1e.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/BesselK1e.java
new file mode 100644
index 00000000000..b01a57da175
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/BesselK1e.java
@@ -0,0 +1,71 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.math.special;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * @param <T> data type for {@code y()} output
+ */
+public final class BesselK1e<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new BesselK1e operation.
+   * 
+   * @param scope current scope
+   * @param x 
+   * @return a new instance of BesselK1e
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> BesselK1e<T> create(Scope scope, Operand<T> x) {
+    OperationBuilder opBuilder = scope.env().opBuilder("BesselK1e", scope.makeOpName("BesselK1e"));
+    opBuilder.addInput(x.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new BesselK1e<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> y() {
+    return y;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return y;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "BesselK1e";
+  
+  private Output<T> y;
+  
+  private BesselK1e(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    y = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/BesselY0.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/BesselY0.java
new file mode 100644
index 00000000000..619147cbb6b
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/BesselY0.java
@@ -0,0 +1,71 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.math.special;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * @param <T> data type for {@code y()} output
+ */
+public final class BesselY0<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new BesselY0 operation.
+   * 
+   * @param scope current scope
+   * @param x 
+   * @return a new instance of BesselY0
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> BesselY0<T> create(Scope scope, Operand<T> x) {
+    OperationBuilder opBuilder = scope.env().opBuilder("BesselY0", scope.makeOpName("BesselY0"));
+    opBuilder.addInput(x.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new BesselY0<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> y() {
+    return y;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return y;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "BesselY0";
+  
+  private Output<T> y;
+  
+  private BesselY0(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    y = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/BesselY1.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/BesselY1.java
new file mode 100644
index 00000000000..51eb6e377f2
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/BesselY1.java
@@ -0,0 +1,71 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.math.special;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * @param <T> data type for {@code y()} output
+ */
+public final class BesselY1<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new BesselY1 operation.
+   * 
+   * @param scope current scope
+   * @param x 
+   * @return a new instance of BesselY1
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> BesselY1<T> create(Scope scope, Operand<T> x) {
+    OperationBuilder opBuilder = scope.env().opBuilder("BesselY1", scope.makeOpName("BesselY1"));
+    opBuilder.addInput(x.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new BesselY1<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> y() {
+    return y;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return y;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "BesselY1";
+  
+  private Output<T> y;
+  
+  private BesselY1(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    y = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/Dawsn.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/Dawsn.java
new file mode 100644
index 00000000000..529514c9bf1
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/Dawsn.java
@@ -0,0 +1,71 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.math.special;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * @param <T> data type for {@code y()} output
+ */
+public final class Dawsn<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new Dawsn operation.
+   * 
+   * @param scope current scope
+   * @param x 
+   * @return a new instance of Dawsn
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> Dawsn<T> create(Scope scope, Operand<T> x) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Dawsn", scope.makeOpName("Dawsn"));
+    opBuilder.addInput(x.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new Dawsn<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> y() {
+    return y;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return y;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Dawsn";
+  
+  private Output<T> y;
+  
+  private Dawsn(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    y = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/Expint.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/Expint.java
new file mode 100644
index 00000000000..099cd544ada
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/Expint.java
@@ -0,0 +1,71 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.math.special;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * @param <T> data type for {@code y()} output
+ */
+public final class Expint<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new Expint operation.
+   * 
+   * @param scope current scope
+   * @param x 
+   * @return a new instance of Expint
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> Expint<T> create(Scope scope, Operand<T> x) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Expint", scope.makeOpName("Expint"));
+    opBuilder.addInput(x.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new Expint<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> y() {
+    return y;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return y;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Expint";
+  
+  private Output<T> y;
+  
+  private Expint(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    y = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/FresnelCos.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/FresnelCos.java
new file mode 100644
index 00000000000..5d5ce736e42
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/FresnelCos.java
@@ -0,0 +1,71 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.math.special;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * @param <T> data type for {@code y()} output
+ */
+public final class FresnelCos<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new FresnelCos operation.
+   * 
+   * @param scope current scope
+   * @param x 
+   * @return a new instance of FresnelCos
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> FresnelCos<T> create(Scope scope, Operand<T> x) {
+    OperationBuilder opBuilder = scope.env().opBuilder("FresnelCos", scope.makeOpName("FresnelCos"));
+    opBuilder.addInput(x.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new FresnelCos<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> y() {
+    return y;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return y;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "FresnelCos";
+  
+  private Output<T> y;
+  
+  private FresnelCos(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    y = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/FresnelSin.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/FresnelSin.java
new file mode 100644
index 00000000000..5914fe4da76
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/FresnelSin.java
@@ -0,0 +1,71 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.math.special;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * @param <T> data type for {@code y()} output
+ */
+public final class FresnelSin<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new FresnelSin operation.
+   * 
+   * @param scope current scope
+   * @param x 
+   * @return a new instance of FresnelSin
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> FresnelSin<T> create(Scope scope, Operand<T> x) {
+    OperationBuilder opBuilder = scope.env().opBuilder("FresnelSin", scope.makeOpName("FresnelSin"));
+    opBuilder.addInput(x.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new FresnelSin<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> y() {
+    return y;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return y;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "FresnelSin";
+  
+  private Output<T> y;
+  
+  private FresnelSin(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    y = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/Spence.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/Spence.java
new file mode 100644
index 00000000000..faf21f53510
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/math/special/Spence.java
@@ -0,0 +1,71 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.math.special;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * @param <T> data type for {@code y()} output
+ */
+public final class Spence<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new Spence operation.
+   * 
+   * @param scope current scope
+   * @param x 
+   * @return a new instance of Spence
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> Spence<T> create(Scope scope, Operand<T> x) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Spence", scope.makeOpName("Spence"));
+    opBuilder.addInput(x.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new Spence<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> y() {
+    return y;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return y;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Spence";
+  
+  private Output<T> y;
+  
+  private Spence(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    y = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/AvgPool.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/AvgPool.java
new file mode 100644
index 00000000000..a27f1aa70c7
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/AvgPool.java
@@ -0,0 +1,135 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.nn;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Performs average pooling on the input.
+ * <p>
+ * Each entry in `output` is the mean of the corresponding size `ksize`
+ * window in `value`.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator(group = "nn")
+public final class AvgPool<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.nn.AvgPool}
+   */
+  public static class Options {
+    
+    /**
+     * @param dataFormat Specify the data format of the input and output data. With the
+     * default format "NHWC", the data is stored in the order of:
+     *     [batch, in_height, in_width, in_channels].
+     * Alternatively, the format could be "NCHW", the data storage order of:
+     *     [batch, in_channels, in_height, in_width].
+     */
+    public Options dataFormat(String dataFormat) {
+      this.dataFormat = dataFormat;
+      return this;
+    }
+    
+    private String dataFormat;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new AvgPool operation.
+   * 
+   * @param scope current scope
+   * @param value 4-D with shape `[batch, height, width, channels]`.
+   * @param ksize The size of the sliding window for each dimension of `value`.
+   * @param strides The stride of the sliding window for each dimension of `value`.
+   * @param padding The type of padding algorithm to use.
+   * @param options carries optional attributes values
+   * @return a new instance of AvgPool
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> AvgPool<T> create(Scope scope, Operand<T> value, List<Long> ksize, List<Long> strides, String padding, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("AvgPool", scope.makeOpName("AvgPool"));
+    opBuilder.addInput(value.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    long[] ksizeArray = new long[ksize.size()];
+    for (int i = 0; i < ksizeArray.length; ++i) {
+      ksizeArray[i] = ksize.get(i);
+    }
+    opBuilder.setAttr("ksize", ksizeArray);
+    long[] stridesArray = new long[strides.size()];
+    for (int i = 0; i < stridesArray.length; ++i) {
+      stridesArray[i] = strides.get(i);
+    }
+    opBuilder.setAttr("strides", stridesArray);
+    opBuilder.setAttr("padding", padding);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.dataFormat != null) {
+          opBuilder.setAttr("data_format", opts.dataFormat);
+        }
+      }
+    }
+    return new AvgPool<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param dataFormat Specify the data format of the input and output data. With the
+   * default format "NHWC", the data is stored in the order of:
+   *     [batch, in_height, in_width, in_channels].
+   * Alternatively, the format could be "NCHW", the data storage order of:
+   *     [batch, in_channels, in_height, in_width].
+   */
+  public static Options dataFormat(String dataFormat) {
+    return new Options().dataFormat(dataFormat);
+  }
+  
+  /**
+   * The average pooled output tensor.
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "AvgPool";
+  
+  private Output<T> output;
+  
+  private AvgPool(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/AvgPool3d.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/AvgPool3d.java
new file mode 100644
index 00000000000..177227c0a52
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/AvgPool3d.java
@@ -0,0 +1,134 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.nn;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Performs 3D average pooling on the input.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator(group = "nn")
+public final class AvgPool3d<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.nn.AvgPool3d}
+   */
+  public static class Options {
+    
+    /**
+     * @param dataFormat The data format of the input and output data. With the
+     * default format "NDHWC", the data is stored in the order of:
+     *     [batch, in_depth, in_height, in_width, in_channels].
+     * Alternatively, the format could be "NCDHW", the data storage order is:
+     *     [batch, in_channels, in_depth, in_height, in_width].
+     */
+    public Options dataFormat(String dataFormat) {
+      this.dataFormat = dataFormat;
+      return this;
+    }
+    
+    private String dataFormat;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new AvgPool3d operation.
+   * 
+   * @param scope current scope
+   * @param input Shape `[batch, depth, rows, cols, channels]` tensor to pool over.
+   * @param ksize 1-D tensor of length 5. The size of the window for each dimension of
+   * the input tensor. Must have `ksize[0] = ksize[4] = 1`.
+   * @param strides 1-D tensor of length 5. The stride of the sliding window for each
+   * dimension of `input`. Must have `strides[0] = strides[4] = 1`.
+   * @param padding The type of padding algorithm to use.
+   * @param options carries optional attributes values
+   * @return a new instance of AvgPool3d
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> AvgPool3d<T> create(Scope scope, Operand<T> input, List<Long> ksize, List<Long> strides, String padding, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("AvgPool3D", scope.makeOpName("AvgPool3d"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    long[] ksizeArray = new long[ksize.size()];
+    for (int i = 0; i < ksizeArray.length; ++i) {
+      ksizeArray[i] = ksize.get(i);
+    }
+    opBuilder.setAttr("ksize", ksizeArray);
+    long[] stridesArray = new long[strides.size()];
+    for (int i = 0; i < stridesArray.length; ++i) {
+      stridesArray[i] = strides.get(i);
+    }
+    opBuilder.setAttr("strides", stridesArray);
+    opBuilder.setAttr("padding", padding);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.dataFormat != null) {
+          opBuilder.setAttr("data_format", opts.dataFormat);
+        }
+      }
+    }
+    return new AvgPool3d<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param dataFormat The data format of the input and output data. With the
+   * default format "NDHWC", the data is stored in the order of:
+   *     [batch, in_depth, in_height, in_width, in_channels].
+   * Alternatively, the format could be "NCDHW", the data storage order is:
+   *     [batch, in_channels, in_depth, in_height, in_width].
+   */
+  public static Options dataFormat(String dataFormat) {
+    return new Options().dataFormat(dataFormat);
+  }
+  
+  /**
+   * The average pooled output tensor.
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "AvgPool3D";
+  
+  private Output<T> output;
+  
+  private AvgPool3d(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/AvgPool3dGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/AvgPool3dGrad.java
new file mode 100644
index 00000000000..c23a0c5360c
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/AvgPool3dGrad.java
@@ -0,0 +1,137 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.nn;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Computes gradients of average pooling function.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator(group = "nn")
+public final class AvgPool3dGrad<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.nn.AvgPool3dGrad}
+   */
+  public static class Options {
+    
+    /**
+     * @param dataFormat The data format of the input and output data. With the
+     * default format "NDHWC", the data is stored in the order of:
+     *     [batch, in_depth, in_height, in_width, in_channels].
+     * Alternatively, the format could be "NCDHW", the data storage order is:
+     *     [batch, in_channels, in_depth, in_height, in_width].
+     */
+    public Options dataFormat(String dataFormat) {
+      this.dataFormat = dataFormat;
+      return this;
+    }
+    
+    private String dataFormat;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new AvgPool3dGrad operation.
+   * 
+   * @param scope current scope
+   * @param origInputShape The original input dimensions.
+   * @param grad Output backprop of shape `[batch, depth, rows, cols, channels]`.
+   * @param ksize 1-D tensor of length 5. The size of the window for each dimension of
+   * the input tensor. Must have `ksize[0] = ksize[4] = 1`.
+   * @param strides 1-D tensor of length 5. The stride of the sliding window for each
+   * dimension of `input`. Must have `strides[0] = strides[4] = 1`.
+   * @param padding The type of padding algorithm to use.
+   * @param options carries optional attributes values
+   * @return a new instance of AvgPool3dGrad
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> AvgPool3dGrad<T> create(Scope scope, Operand<TInt32> origInputShape, Operand<T> grad, List<Long> ksize, List<Long> strides, String padding, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("AvgPool3DGrad", scope.makeOpName("AvgPool3dGrad"));
+    opBuilder.addInput(origInputShape.asOutput());
+    opBuilder.addInput(grad.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    long[] ksizeArray = new long[ksize.size()];
+    for (int i = 0; i < ksizeArray.length; ++i) {
+      ksizeArray[i] = ksize.get(i);
+    }
+    opBuilder.setAttr("ksize", ksizeArray);
+    long[] stridesArray = new long[strides.size()];
+    for (int i = 0; i < stridesArray.length; ++i) {
+      stridesArray[i] = strides.get(i);
+    }
+    opBuilder.setAttr("strides", stridesArray);
+    opBuilder.setAttr("padding", padding);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.dataFormat != null) {
+          opBuilder.setAttr("data_format", opts.dataFormat);
+        }
+      }
+    }
+    return new AvgPool3dGrad<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param dataFormat The data format of the input and output data. With the
+   * default format "NDHWC", the data is stored in the order of:
+   *     [batch, in_depth, in_height, in_width, in_channels].
+   * Alternatively, the format could be "NCDHW", the data storage order is:
+   *     [batch, in_channels, in_depth, in_height, in_width].
+   */
+  public static Options dataFormat(String dataFormat) {
+    return new Options().dataFormat(dataFormat);
+  }
+  
+  /**
+   * The backprop for input.
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "AvgPool3DGrad";
+  
+  private Output<T> output;
+  
+  private AvgPool3dGrad(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/AvgPoolGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/AvgPoolGrad.java
new file mode 100644
index 00000000000..ec149807546
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/AvgPoolGrad.java
@@ -0,0 +1,135 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.nn;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Computes gradients of the average pooling function.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+public final class AvgPoolGrad<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.nn.AvgPoolGrad}
+   */
+  public static class Options {
+    
+    /**
+     * @param dataFormat Specify the data format of the input and output data. With the
+     * default format "NHWC", the data is stored in the order of:
+     *     [batch, in_height, in_width, in_channels].
+     * Alternatively, the format could be "NCHW", the data storage order of:
+     *     [batch, in_channels, in_height, in_width].
+     */
+    public Options dataFormat(String dataFormat) {
+      this.dataFormat = dataFormat;
+      return this;
+    }
+    
+    private String dataFormat;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new AvgPoolGrad operation.
+   * 
+   * @param scope current scope
+   * @param origInputShape 1-D.  Shape of the original input to `avg_pool`.
+   * @param grad 4-D with shape `[batch, height, width, channels]`.  Gradients w.r.t.
+   * the output of `avg_pool`.
+   * @param ksize The size of the sliding window for each dimension of the input.
+   * @param strides The stride of the sliding window for each dimension of the input.
+   * @param padding The type of padding algorithm to use.
+   * @param options carries optional attributes values
+   * @return a new instance of AvgPoolGrad
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> AvgPoolGrad<T> create(Scope scope, Operand<TInt32> origInputShape, Operand<T> grad, List<Long> ksize, List<Long> strides, String padding, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("AvgPoolGrad", scope.makeOpName("AvgPoolGrad"));
+    opBuilder.addInput(origInputShape.asOutput());
+    opBuilder.addInput(grad.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    long[] ksizeArray = new long[ksize.size()];
+    for (int i = 0; i < ksizeArray.length; ++i) {
+      ksizeArray[i] = ksize.get(i);
+    }
+    opBuilder.setAttr("ksize", ksizeArray);
+    long[] stridesArray = new long[strides.size()];
+    for (int i = 0; i < stridesArray.length; ++i) {
+      stridesArray[i] = strides.get(i);
+    }
+    opBuilder.setAttr("strides", stridesArray);
+    opBuilder.setAttr("padding", padding);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.dataFormat != null) {
+          opBuilder.setAttr("data_format", opts.dataFormat);
+        }
+      }
+    }
+    return new AvgPoolGrad<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param dataFormat Specify the data format of the input and output data. With the
+   * default format "NHWC", the data is stored in the order of:
+   *     [batch, in_height, in_width, in_channels].
+   * Alternatively, the format could be "NCHW", the data storage order of:
+   *     [batch, in_channels, in_height, in_width].
+   */
+  public static Options dataFormat(String dataFormat) {
+    return new Options().dataFormat(dataFormat);
+  }
+  
+  /**
+   * 4-D.  Gradients w.r.t. the input of `avg_pool`.
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "AvgPoolGrad";
+  
+  private Output<T> output;
+  
+  private AvgPoolGrad(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/BatchNormWithGlobalNormalization.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/BatchNormWithGlobalNormalization.java
new file mode 100644
index 00000000000..6c30a52cf2e
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/BatchNormWithGlobalNormalization.java
@@ -0,0 +1,96 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.nn;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Batch normalization.
+ * <p>
+ * This op is deprecated. Prefer `tf.nn.batch_normalization`.
+ * 
+ * @param <T> data type for {@code result()} output
+ */
+@Operator(group = "nn")
+public final class BatchNormWithGlobalNormalization<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new BatchNormWithGlobalNormalization operation.
+   * 
+   * @param scope current scope
+   * @param t A 4D input Tensor.
+   * @param m A 1D mean Tensor with size matching the last dimension of t.
+   * This is the first output from tf.nn.moments,
+   * or a saved moving average thereof.
+   * @param v A 1D variance Tensor with size matching the last dimension of t.
+   * This is the second output from tf.nn.moments,
+   * or a saved moving average thereof.
+   * @param beta A 1D beta Tensor with size matching the last dimension of t.
+   * An offset to be added to the normalized tensor.
+   * @param gamma A 1D gamma Tensor with size matching the last dimension of t.
+   * If "scale_after_normalization" is true, this tensor will be multiplied
+   * with the normalized tensor.
+   * @param varianceEpsilon A small float number to avoid dividing by 0.
+   * @param scaleAfterNormalization A bool indicating whether the resulted tensor
+   * needs to be multiplied with gamma.
+   * @return a new instance of BatchNormWithGlobalNormalization
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> BatchNormWithGlobalNormalization<T> create(Scope scope, Operand<T> t, Operand<T> m, Operand<T> v, Operand<T> beta, Operand<T> gamma, Float varianceEpsilon, Boolean scaleAfterNormalization) {
+    OperationBuilder opBuilder = scope.env().opBuilder("BatchNormWithGlobalNormalization", scope.makeOpName("BatchNormWithGlobalNormalization"));
+    opBuilder.addInput(t.asOutput());
+    opBuilder.addInput(m.asOutput());
+    opBuilder.addInput(v.asOutput());
+    opBuilder.addInput(beta.asOutput());
+    opBuilder.addInput(gamma.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("variance_epsilon", varianceEpsilon);
+    opBuilder.setAttr("scale_after_normalization", scaleAfterNormalization);
+    return new BatchNormWithGlobalNormalization<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> result() {
+    return result;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return result;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "BatchNormWithGlobalNormalization";
+  
+  private Output<T> result;
+  
+  private BatchNormWithGlobalNormalization(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    result = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/BatchNormWithGlobalNormalizationGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/BatchNormWithGlobalNormalizationGrad.java
new file mode 100644
index 00000000000..be07a5eea25
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/BatchNormWithGlobalNormalizationGrad.java
@@ -0,0 +1,127 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.nn;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Gradients for batch normalization.
+ * <p>
+ * This op is deprecated. See `tf.nn.batch_normalization`.
+ * 
+ * @param <T> data type for {@code dx()} output
+ */
+@Operator(group = "nn")
+public final class BatchNormWithGlobalNormalizationGrad<T extends TType> extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new BatchNormWithGlobalNormalizationGrad operation.
+   * 
+   * @param scope current scope
+   * @param t A 4D input Tensor.
+   * @param m A 1D mean Tensor with size matching the last dimension of t.
+   * This is the first output from tf.nn.moments,
+   * or a saved moving average thereof.
+   * @param v A 1D variance Tensor with size matching the last dimension of t.
+   * This is the second output from tf.nn.moments,
+   * or a saved moving average thereof.
+   * @param gamma A 1D gamma Tensor with size matching the last dimension of t.
+   * If "scale_after_normalization" is true, this Tensor will be multiplied
+   * with the normalized Tensor.
+   * @param backprop 4D backprop Tensor.
+   * @param varianceEpsilon A small float number to avoid dividing by 0.
+   * @param scaleAfterNormalization A bool indicating whether the resulted tensor
+   * needs to be multiplied with gamma.
+   * @return a new instance of BatchNormWithGlobalNormalizationGrad
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> BatchNormWithGlobalNormalizationGrad<T> create(Scope scope, Operand<T> t, Operand<T> m, Operand<T> v, Operand<T> gamma, Operand<T> backprop, Float varianceEpsilon, Boolean scaleAfterNormalization) {
+    OperationBuilder opBuilder = scope.env().opBuilder("BatchNormWithGlobalNormalizationGrad", scope.makeOpName("BatchNormWithGlobalNormalizationGrad"));
+    opBuilder.addInput(t.asOutput());
+    opBuilder.addInput(m.asOutput());
+    opBuilder.addInput(v.asOutput());
+    opBuilder.addInput(gamma.asOutput());
+    opBuilder.addInput(backprop.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("variance_epsilon", varianceEpsilon);
+    opBuilder.setAttr("scale_after_normalization", scaleAfterNormalization);
+    return new BatchNormWithGlobalNormalizationGrad<T>(opBuilder.build());
+  }
+  
+  /**
+   * 4D backprop tensor for input.
+   */
+  public Output<T> dx() {
+    return dx;
+  }
+  
+  /**
+   * 1D backprop tensor for mean.
+   */
+  public Output<T> dm() {
+    return dm;
+  }
+  
+  /**
+   * 1D backprop tensor for variance.
+   */
+  public Output<T> dv() {
+    return dv;
+  }
+  
+  /**
+   * 1D backprop tensor for beta.
+   */
+  public Output<T> db() {
+    return db;
+  }
+  
+  /**
+   * 1D backprop tensor for gamma.
+   */
+  public Output<T> dg() {
+    return dg;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "BatchNormWithGlobalNormalizationGrad";
+  
+  private Output<T> dx;
+  private Output<T> dm;
+  private Output<T> dv;
+  private Output<T> db;
+  private Output<T> dg;
+  
+  private BatchNormWithGlobalNormalizationGrad(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    dx = operation.output(outputIdx++);
+    dm = operation.output(outputIdx++);
+    dv = operation.output(outputIdx++);
+    db = operation.output(outputIdx++);
+    dg = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/BiasAdd.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/BiasAdd.java
new file mode 100644
index 00000000000..462eaa6edc7
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/BiasAdd.java
@@ -0,0 +1,126 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.nn;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Adds `bias` to `value`.
+ * <p>
+ * This is a special case of `tf.add` where `bias` is restricted to be 1-D.
+ * Broadcasting is supported, so `value` may have any number of dimensions.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator(group = "nn")
+public final class BiasAdd<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.nn.BiasAdd}
+   */
+  public static class Options {
+    
+    /**
+     * @param dataFormat Specify the data format of the input and output data. With the
+     * default format "NHWC", the bias tensor will be added to the last dimension
+     * of the value tensor.
+     * Alternatively, the format could be "NCHW", the data storage order of:
+     *     [batch, in_channels, in_height, in_width].
+     * The tensor will be added to "in_channels", the third-to-the-last
+     *     dimension.
+     */
+    public Options dataFormat(String dataFormat) {
+      this.dataFormat = dataFormat;
+      return this;
+    }
+    
+    private String dataFormat;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new BiasAdd operation.
+   * 
+   * @param scope current scope
+   * @param value Any number of dimensions.
+   * @param bias 1-D with size the last dimension of `value`.
+   * @param options carries optional attributes values
+   * @return a new instance of BiasAdd
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> BiasAdd<T> create(Scope scope, Operand<T> value, Operand<T> bias, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("BiasAdd", scope.makeOpName("BiasAdd"));
+    opBuilder.addInput(value.asOutput());
+    opBuilder.addInput(bias.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.dataFormat != null) {
+          opBuilder.setAttr("data_format", opts.dataFormat);
+        }
+      }
+    }
+    return new BiasAdd<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param dataFormat Specify the data format of the input and output data. With the
+   * default format "NHWC", the bias tensor will be added to the last dimension
+   * of the value tensor.
+   * Alternatively, the format could be "NCHW", the data storage order of:
+   *     [batch, in_channels, in_height, in_width].
+   * The tensor will be added to "in_channels", the third-to-the-last
+   *     dimension.
+   */
+  public static Options dataFormat(String dataFormat) {
+    return new Options().dataFormat(dataFormat);
+  }
+  
+  /**
+   * Broadcasted sum of `value` and `bias`.
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "BiasAdd";
+  
+  private Output<T> output;
+  
+  private BiasAdd(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/BiasAddGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/BiasAddGrad.java
new file mode 100644
index 00000000000..df90cf9a2f3
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/BiasAddGrad.java
@@ -0,0 +1,125 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.nn;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * The backward operation for "BiasAdd" on the "bias" tensor.
+ * <p>
+ * It accumulates all the values from out_backprop into the feature dimension.
+ * For NHWC data format, the feature dimension is the last. For NCHW data format,
+ * the feature dimension is the third-to-last.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator(group = "nn")
+public final class BiasAddGrad<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.nn.BiasAddGrad}
+   */
+  public static class Options {
+    
+    /**
+     * @param dataFormat Specify the data format of the input and output data. With the
+     * default format "NHWC", the bias tensor will be added to the last dimension
+     * of the value tensor.
+     * Alternatively, the format could be "NCHW", the data storage order of:
+     *     [batch, in_channels, in_height, in_width].
+     * The tensor will be added to "in_channels", the third-to-the-last
+     *     dimension.
+     */
+    public Options dataFormat(String dataFormat) {
+      this.dataFormat = dataFormat;
+      return this;
+    }
+    
+    private String dataFormat;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new BiasAddGrad operation.
+   * 
+   * @param scope current scope
+   * @param outBackprop Any number of dimensions.
+   * @param options carries optional attributes values
+   * @return a new instance of BiasAddGrad
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> BiasAddGrad<T> create(Scope scope, Operand<T> outBackprop, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("BiasAddGrad", scope.makeOpName("BiasAddGrad"));
+    opBuilder.addInput(outBackprop.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.dataFormat != null) {
+          opBuilder.setAttr("data_format", opts.dataFormat);
+        }
+      }
+    }
+    return new BiasAddGrad<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param dataFormat Specify the data format of the input and output data. With the
+   * default format "NHWC", the bias tensor will be added to the last dimension
+   * of the value tensor.
+   * Alternatively, the format could be "NCHW", the data storage order of:
+   *     [batch, in_channels, in_height, in_width].
+   * The tensor will be added to "in_channels", the third-to-the-last
+   *     dimension.
+   */
+  public static Options dataFormat(String dataFormat) {
+    return new Options().dataFormat(dataFormat);
+  }
+  
+  /**
+   * 1-D with size the feature dimension of `out_backprop`.
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "BiasAddGrad";
+  
+  private Output<T> output;
+  
+  private BiasAddGrad(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/BlockLSTM.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/BlockLSTM.java
new file mode 100644
index 00000000000..c40f2eeea4b
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/BlockLSTM.java
@@ -0,0 +1,216 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.nn;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Computes the LSTM cell forward propagation for all the time steps.
+ * <p>
+ * This is equivalent to applying LSTMBlockCell in a loop, like so:
+ * <pre>{@code
+ * for x1 in unpack(x):
+ *   i1, cs1, f1, o1, ci1, co1, h1 = LSTMBlock(
+ *     x1, cs_prev, h_prev, w, wci, wcf, wco, b)
+ *   cs_prev = cs1
+ *   h_prev = h1
+ *   i.append(i1)
+ *   cs.append(cs1)
+ *   f.append(f1)
+ *   o.append(o1)
+ *   ci.append(ci1)
+ *   co.append(co1)
+ *   h.append(h1)
+ * return pack(i), pack(cs), pack(f), pack(o), pack(ci), pack(ch), pack(h)
+ * 
+ * Note that unlike LSTMBlockCell (and BlockLSTM) which uses ICFO gate layout,
+ * this op uses IFCO. So in order for the following snippet to be equivalent
+ * all gate-related outputs should be reordered.
+ * }</pre>
+ * 
+ * 
+ * @param <T> data type for {@code i()} output
+ */
+public final class BlockLSTM<T extends TNumber> extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.nn.BlockLSTM}
+   */
+  public static class Options {
+    
+    /**
+     * @param cellClip Value to clip the 'cs' value to.
+     */
+    public Options cellClip(Float cellClip) {
+      this.cellClip = cellClip;
+      return this;
+    }
+    
+    /**
+     * @param usePeephole Whether to use peephole weights.
+     */
+    public Options usePeephole(Boolean usePeephole) {
+      this.usePeephole = usePeephole;
+      return this;
+    }
+    
+    private Float cellClip;
+    private Boolean usePeephole;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new BlockLSTM operation.
+   * 
+   * @param scope current scope
+   * @param seqLenMax Maximum time length actually used by this input. Outputs are padded
+   * with zeros beyond this length.
+   * @param x The sequence input to the LSTM, shape (timelen, batch_size, num_inputs).
+   * @param csPrev Value of the initial cell state.
+   * @param hPrev Initial output of cell (to be used for peephole).
+   * @param w The weight matrix.
+   * @param wci The weight matrix for input gate peephole connection.
+   * @param wcf The weight matrix for forget gate peephole connection.
+   * @param wco The weight matrix for output gate peephole connection.
+   * @param b The bias vector.
+   * @param options carries optional attributes values
+   * @return a new instance of BlockLSTM
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> BlockLSTM<T> create(Scope scope, Operand<TInt64> seqLenMax, Operand<T> x, Operand<T> csPrev, Operand<T> hPrev, Operand<T> w, Operand<T> wci, Operand<T> wcf, Operand<T> wco, Operand<T> b, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("BlockLSTMV2", scope.makeOpName("BlockLSTM"));
+    opBuilder.addInput(seqLenMax.asOutput());
+    opBuilder.addInput(x.asOutput());
+    opBuilder.addInput(csPrev.asOutput());
+    opBuilder.addInput(hPrev.asOutput());
+    opBuilder.addInput(w.asOutput());
+    opBuilder.addInput(wci.asOutput());
+    opBuilder.addInput(wcf.asOutput());
+    opBuilder.addInput(wco.asOutput());
+    opBuilder.addInput(b.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.cellClip != null) {
+          opBuilder.setAttr("cell_clip", opts.cellClip);
+        }
+        if (opts.usePeephole != null) {
+          opBuilder.setAttr("use_peephole", opts.usePeephole);
+        }
+      }
+    }
+    return new BlockLSTM<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param cellClip Value to clip the 'cs' value to.
+   */
+  public static Options cellClip(Float cellClip) {
+    return new Options().cellClip(cellClip);
+  }
+  
+  /**
+   * @param usePeephole Whether to use peephole weights.
+   */
+  public static Options usePeephole(Boolean usePeephole) {
+    return new Options().usePeephole(usePeephole);
+  }
+  
+  /**
+   * The input gate over the whole time sequence.
+   */
+  public Output<T> i() {
+    return i;
+  }
+  
+  /**
+   * The cell state before the tanh over the whole time sequence.
+   */
+  public Output<T> cs() {
+    return cs;
+  }
+  
+  /**
+   * The forget gate over the whole time sequence.
+   */
+  public Output<T> f() {
+    return f;
+  }
+  
+  /**
+   * The output gate over the whole time sequence.
+   */
+  public Output<T> o() {
+    return o;
+  }
+  
+  /**
+   * The cell input over the whole time sequence.
+   */
+  public Output<T> ci() {
+    return ci;
+  }
+  
+  /**
+   * The cell after the tanh over the whole time sequence.
+   */
+  public Output<T> co() {
+    return co;
+  }
+  
+  /**
+   * The output h vector over the whole time sequence.
+   */
+  public Output<T> h() {
+    return h;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "BlockLSTMV2";
+  
+  private Output<T> i;
+  private Output<T> cs;
+  private Output<T> f;
+  private Output<T> o;
+  private Output<T> ci;
+  private Output<T> co;
+  private Output<T> h;
+  
+  private BlockLSTM(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    i = operation.output(outputIdx++);
+    cs = operation.output(outputIdx++);
+    f = operation.output(outputIdx++);
+    o = operation.output(outputIdx++);
+    ci = operation.output(outputIdx++);
+    co = operation.output(outputIdx++);
+    h = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/BlockLSTMGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/BlockLSTMGrad.java
new file mode 100644
index 00000000000..f7a17b1bf76
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/BlockLSTMGrad.java
@@ -0,0 +1,172 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.nn;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Computes the LSTM cell backward propagation for the entire time sequence.
+ * <p>
+ * This implementation is to be used in conjunction of BlockLSTMV2.
+ * 
+ * @param <T> data type for {@code xGrad()} output
+ */
+public final class BlockLSTMGrad<T extends TNumber> extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new BlockLSTMGrad operation.
+   * 
+   * @param scope current scope
+   * @param seqLenMax Maximum time length actually used by this input. Outputs are padded
+   * with zeros beyond this length.
+   * @param x The sequence input to the LSTM, shape (timelen, batch_size, num_inputs).
+   * @param csPrev Value of the initial cell state.
+   * @param hPrev Initial output of cell (to be used for peephole).
+   * @param w The weight matrix.
+   * @param wci The weight matrix for input gate peephole connection.
+   * @param wcf The weight matrix for forget gate peephole connection.
+   * @param wco The weight matrix for output gate peephole connection.
+   * @param b The bias vector.
+   * @param i The input gate over the whole time sequence.
+   * @param cs The cell state before the tanh over the whole time sequence.
+   * @param f The forget gate over the whole time sequence.
+   * @param o The output gate over the whole time sequence.
+   * @param ci The cell input over the whole time sequence.
+   * @param co The cell after the tanh over the whole time sequence.
+   * @param h The output h vector over the whole time sequence.
+   * @param csGrad The current gradient of cs.
+   * @param hGrad The gradient of h vector.
+   * @param usePeephole Whether to use peephole weights.
+   * @return a new instance of BlockLSTMGrad
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> BlockLSTMGrad<T> create(Scope scope, Operand<TInt64> seqLenMax, Operand<T> x, Operand<T> csPrev, Operand<T> hPrev, Operand<T> w, Operand<T> wci, Operand<T> wcf, Operand<T> wco, Operand<T> b, Operand<T> i, Operand<T> cs, Operand<T> f, Operand<T> o, Operand<T> ci, Operand<T> co, Operand<T> h, Operand<T> csGrad, Operand<T> hGrad, Boolean usePeephole) {
+    OperationBuilder opBuilder = scope.env().opBuilder("BlockLSTMGradV2", scope.makeOpName("BlockLSTMGrad"));
+    opBuilder.addInput(seqLenMax.asOutput());
+    opBuilder.addInput(x.asOutput());
+    opBuilder.addInput(csPrev.asOutput());
+    opBuilder.addInput(hPrev.asOutput());
+    opBuilder.addInput(w.asOutput());
+    opBuilder.addInput(wci.asOutput());
+    opBuilder.addInput(wcf.asOutput());
+    opBuilder.addInput(wco.asOutput());
+    opBuilder.addInput(b.asOutput());
+    opBuilder.addInput(i.asOutput());
+    opBuilder.addInput(cs.asOutput());
+    opBuilder.addInput(f.asOutput());
+    opBuilder.addInput(o.asOutput());
+    opBuilder.addInput(ci.asOutput());
+    opBuilder.addInput(co.asOutput());
+    opBuilder.addInput(h.asOutput());
+    opBuilder.addInput(csGrad.asOutput());
+    opBuilder.addInput(hGrad.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("use_peephole", usePeephole);
+    return new BlockLSTMGrad<T>(opBuilder.build());
+  }
+  
+  /**
+   * The gradient of x to be back-propped.
+   */
+  public Output<T> xGrad() {
+    return xGrad;
+  }
+  
+  /**
+   * The gradient of cs_prev to be back-propped.
+   */
+  public Output<T> csPrevGrad() {
+    return csPrevGrad;
+  }
+  
+  /**
+   * The gradient of h_prev to be back-propped.
+   */
+  public Output<T> hPrevGrad() {
+    return hPrevGrad;
+  }
+  
+  /**
+   * The gradient for w to be back-propped.
+   */
+  public Output<T> wGrad() {
+    return wGrad;
+  }
+  
+  /**
+   * The gradient for wci to be back-propped.
+   */
+  public Output<T> wciGrad() {
+    return wciGrad;
+  }
+  
+  /**
+   * The gradient for wcf to be back-propped.
+   */
+  public Output<T> wcfGrad() {
+    return wcfGrad;
+  }
+  
+  /**
+   * The gradient for wco to be back-propped.
+   */
+  public Output<T> wcoGrad() {
+    return wcoGrad;
+  }
+  
+  /**
+   * The gradient for w to be back-propped.
+   */
+  public Output<T> bGrad() {
+    return bGrad;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "BlockLSTMGradV2";
+  
+  private Output<T> xGrad;
+  private Output<T> csPrevGrad;
+  private Output<T> hPrevGrad;
+  private Output<T> wGrad;
+  private Output<T> wciGrad;
+  private Output<T> wcfGrad;
+  private Output<T> wcoGrad;
+  private Output<T> bGrad;
+  
+  private BlockLSTMGrad(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    xGrad = operation.output(outputIdx++);
+    csPrevGrad = operation.output(outputIdx++);
+    hPrevGrad = operation.output(outputIdx++);
+    wGrad = operation.output(outputIdx++);
+    wciGrad = operation.output(outputIdx++);
+    wcfGrad = operation.output(outputIdx++);
+    wcoGrad = operation.output(outputIdx++);
+    bGrad = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CTCLossV2.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CTCLossV2.java
new file mode 100644
index 00000000000..c60fef4b2a7
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CTCLossV2.java
@@ -0,0 +1,173 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.nn;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.TInt64;
+
+/**
+ * Calculates the CTC Loss (log probability) for each batch entry.  Also calculates
+ * <p>
+ * the gradient.  This class performs the softmax operation for you, so inputs
+ * should be e.g. linear projections of outputs by an LSTM.
+ */
+public final class CTCLossV2 extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.nn.CTCLossV2}
+   */
+  public static class Options {
+    
+    /**
+     * @param preprocessCollapseRepeated Scalar, if true then repeated labels are
+     * collapsed prior to the CTC calculation.
+     */
+    public Options preprocessCollapseRepeated(Boolean preprocessCollapseRepeated) {
+      this.preprocessCollapseRepeated = preprocessCollapseRepeated;
+      return this;
+    }
+    
+    /**
+     * @param ctcMergeRepeated Scalar.  If set to false, <i>during</i> CTC calculation
+     * repeated non-blank labels will not be merged and are interpreted as
+     * individual labels.  This is a simplified version of CTC.
+     */
+    public Options ctcMergeRepeated(Boolean ctcMergeRepeated) {
+      this.ctcMergeRepeated = ctcMergeRepeated;
+      return this;
+    }
+    
+    /**
+     * @param ignoreLongerOutputsThanInputs Scalar. If set to true, during CTC
+     * calculation, items that have longer output sequences than input sequences
+     * are skipped: they don't contribute to the loss term and have zero-gradient.
+     */
+    public Options ignoreLongerOutputsThanInputs(Boolean ignoreLongerOutputsThanInputs) {
+      this.ignoreLongerOutputsThanInputs = ignoreLongerOutputsThanInputs;
+      return this;
+    }
+    
+    private Boolean preprocessCollapseRepeated;
+    private Boolean ctcMergeRepeated;
+    private Boolean ignoreLongerOutputsThanInputs;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new CTCLossV2 operation.
+   * 
+   * @param scope current scope
+   * @param inputs 3-D, shape: `(max_time x batch_size x num_classes)`, the logits. Default blank
+   * label is 0 rather num_classes - 1.
+   * @param labelsIndices The indices of a `SparseTensor<int32, 2>`.
+   * `labels_indices(i, :) == [b, t]` means `labels_values(i)` stores the id for
+   * `(batch b, time t)`.
+   * @param labelsValues The values (labels) associated with the given batch and time.
+   * @param sequenceLength A vector containing sequence lengths (batch).
+   * @param options carries optional attributes values
+   * @return a new instance of CTCLossV2
+   */
+  @Endpoint(describeByClass = true)
+  public static CTCLossV2 create(Scope scope, Operand<TFloat32> inputs, Operand<TInt64> labelsIndices, Operand<TInt32> labelsValues, Operand<TInt32> sequenceLength, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("CTCLossV2", scope.makeOpName("CTCLossV2"));
+    opBuilder.addInput(inputs.asOutput());
+    opBuilder.addInput(labelsIndices.asOutput());
+    opBuilder.addInput(labelsValues.asOutput());
+    opBuilder.addInput(sequenceLength.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.preprocessCollapseRepeated != null) {
+          opBuilder.setAttr("preprocess_collapse_repeated", opts.preprocessCollapseRepeated);
+        }
+        if (opts.ctcMergeRepeated != null) {
+          opBuilder.setAttr("ctc_merge_repeated", opts.ctcMergeRepeated);
+        }
+        if (opts.ignoreLongerOutputsThanInputs != null) {
+          opBuilder.setAttr("ignore_longer_outputs_than_inputs", opts.ignoreLongerOutputsThanInputs);
+        }
+      }
+    }
+    return new CTCLossV2(opBuilder.build());
+  }
+  
+  /**
+   * @param preprocessCollapseRepeated Scalar, if true then repeated labels are
+   * collapsed prior to the CTC calculation.
+   */
+  public static Options preprocessCollapseRepeated(Boolean preprocessCollapseRepeated) {
+    return new Options().preprocessCollapseRepeated(preprocessCollapseRepeated);
+  }
+  
+  /**
+   * @param ctcMergeRepeated Scalar.  If set to false, <i>during</i> CTC calculation
+   * repeated non-blank labels will not be merged and are interpreted as
+   * individual labels.  This is a simplified version of CTC.
+   */
+  public static Options ctcMergeRepeated(Boolean ctcMergeRepeated) {
+    return new Options().ctcMergeRepeated(ctcMergeRepeated);
+  }
+  
+  /**
+   * @param ignoreLongerOutputsThanInputs Scalar. If set to true, during CTC
+   * calculation, items that have longer output sequences than input sequences
+   * are skipped: they don't contribute to the loss term and have zero-gradient.
+   */
+  public static Options ignoreLongerOutputsThanInputs(Boolean ignoreLongerOutputsThanInputs) {
+    return new Options().ignoreLongerOutputsThanInputs(ignoreLongerOutputsThanInputs);
+  }
+  
+  /**
+   * A vector (batch) containing log-probabilities.
+   */
+  public Output<TFloat32> loss() {
+    return loss;
+  }
+  
+  /**
+   * The gradient of `loss`.  3-D, shape:
+   * `(max_time x batch_size x num_classes)`.
+   */
+  public Output<TFloat32> gradient() {
+    return gradient;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "CTCLossV2";
+  
+  private Output<TFloat32> loss;
+  private Output<TFloat32> gradient;
+  
+  private CTCLossV2(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    loss = operation.output(outputIdx++);
+    gradient = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/ComputeAccidentalHits.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/ComputeAccidentalHits.java
new file mode 100644
index 00000000000..f6daea0df2e
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/ComputeAccidentalHits.java
@@ -0,0 +1,156 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.nn;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.TInt64;
+
+/**
+ * Computes the ids of the positions in sampled_candidates that match true_labels.
+ * <p>
+ * When doing log-odds NCE, the result of this op should be passed through a
+ * SparseToDense op, then added to the logits of the sampled candidates. This has
+ * the effect of 'removing' the sampled labels that match the true labels by
+ * making the classifier sure that they are sampled labels.
+ */
+@Operator(group = "nn")
+public final class ComputeAccidentalHits extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.nn.ComputeAccidentalHits}
+   */
+  public static class Options {
+    
+    /**
+     * @param seed If either seed or seed2 are set to be non-zero, the random number
+     * generator is seeded by the given seed.  Otherwise, it is seeded by a
+     * random seed.
+     */
+    public Options seed(Long seed) {
+      this.seed = seed;
+      return this;
+    }
+    
+    /**
+     * @param seed2 An second seed to avoid seed collision.
+     */
+    public Options seed2(Long seed2) {
+      this.seed2 = seed2;
+      return this;
+    }
+    
+    private Long seed;
+    private Long seed2;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new ComputeAccidentalHits operation.
+   * 
+   * @param scope current scope
+   * @param trueClasses The true_classes output of UnpackSparseLabels.
+   * @param sampledCandidates The sampled_candidates output of CandidateSampler.
+   * @param numTrue Number of true labels per context.
+   * @param options carries optional attributes values
+   * @return a new instance of ComputeAccidentalHits
+   */
+  @Endpoint(describeByClass = true)
+  public static ComputeAccidentalHits create(Scope scope, Operand<TInt64> trueClasses, Operand<TInt64> sampledCandidates, Long numTrue, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ComputeAccidentalHits", scope.makeOpName("ComputeAccidentalHits"));
+    opBuilder.addInput(trueClasses.asOutput());
+    opBuilder.addInput(sampledCandidates.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("num_true", numTrue);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.seed != null) {
+          opBuilder.setAttr("seed", opts.seed);
+        }
+        if (opts.seed2 != null) {
+          opBuilder.setAttr("seed2", opts.seed2);
+        }
+      }
+    }
+    return new ComputeAccidentalHits(opBuilder.build());
+  }
+  
+  /**
+   * @param seed If either seed or seed2 are set to be non-zero, the random number
+   * generator is seeded by the given seed.  Otherwise, it is seeded by a
+   * random seed.
+   */
+  public static Options seed(Long seed) {
+    return new Options().seed(seed);
+  }
+  
+  /**
+   * @param seed2 An second seed to avoid seed collision.
+   */
+  public static Options seed2(Long seed2) {
+    return new Options().seed2(seed2);
+  }
+  
+  /**
+   * A vector of indices corresponding to rows of true_candidates.
+   */
+  public Output<TInt32> indices() {
+    return indices;
+  }
+  
+  /**
+   * A vector of IDs of positions in sampled_candidates that match a true_label
+   * for the row with the corresponding index in indices.
+   */
+  public Output<TInt64> ids() {
+    return ids;
+  }
+  
+  /**
+   * A vector of the same length as indices and ids, in which each element
+   * is -FLOAT_MAX.
+   */
+  public Output<TFloat32> weights() {
+    return weights;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ComputeAccidentalHits";
+  
+  private Output<TInt32> indices;
+  private Output<TInt64> ids;
+  private Output<TFloat32> weights;
+  
+  private ComputeAccidentalHits(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    indices = operation.output(outputIdx++);
+    ids = operation.output(outputIdx++);
+    weights = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Conv2d.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Conv2d.java
new file mode 100644
index 00000000000..70a72bc78da
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Conv2d.java
@@ -0,0 +1,234 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.nn;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Computes a 2-D convolution given 4-D `input` and `filter` tensors.
+ * <p>
+ * Given an input tensor of shape `[batch, in_height, in_width, in_channels]`
+ * and a filter / kernel tensor of shape
+ * `[filter_height, filter_width, in_channels, out_channels]`, this op
+ * performs the following:
+ * <p>
+ * 1. Flattens the filter to a 2-D matrix with shape
+ *    `[filter_height * filter_width * in_channels, output_channels]`.
+ * 2. Extracts image patches from the input tensor to form a <i>virtual</i>
+ *    tensor of shape `[batch, out_height, out_width,
+ *    filter_height * filter_width * in_channels]`.
+ * 3. For each patch, right-multiplies the filter matrix and the image patch
+ *    vector.
+ * <p>
+ * In detail, with the default NHWC format,
+ * <p>
+ *     output[b, i, j, k] =
+ *         sum_{di, dj, q} input[b, strides[1] * i + di, strides[2] * j + dj, q] *
+ *                         filter[di, dj, q, k]
+ * <p>
+ * Must have `strides[0] = strides[3] = 1`.  For the most common case of the same
+ * horizontal and vertices strides, `strides = [1, stride, stride, 1]`.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator(group = "nn")
+public final class Conv2d<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.nn.Conv2d}
+   */
+  public static class Options {
+    
+    /**
+     * @param useCudnnOnGpu 
+     */
+    public Options useCudnnOnGpu(Boolean useCudnnOnGpu) {
+      this.useCudnnOnGpu = useCudnnOnGpu;
+      return this;
+    }
+    
+    /**
+     * @param explicitPaddings If `padding` is `"EXPLICIT"`, the list of explicit padding amounts. For the ith
+     * dimension, the amount of padding inserted before and after the dimension is
+     * `explicit_paddings[2 * i]` and `explicit_paddings[2 * i + 1]`, respectively. If
+     * `padding` is not `"EXPLICIT"`, `explicit_paddings` must be empty.
+     */
+    public Options explicitPaddings(List<Long> explicitPaddings) {
+      this.explicitPaddings = explicitPaddings;
+      return this;
+    }
+    
+    /**
+     * @param dataFormat Specify the data format of the input and output data. With the
+     * default format "NHWC", the data is stored in the order of:
+     *     [batch, height, width, channels].
+     * Alternatively, the format could be "NCHW", the data storage order of:
+     *     [batch, channels, height, width].
+     */
+    public Options dataFormat(String dataFormat) {
+      this.dataFormat = dataFormat;
+      return this;
+    }
+    
+    /**
+     * @param dilations 1-D tensor of length 4.  The dilation factor for each dimension of
+     * `input`. If set to k > 1, there will be k-1 skipped cells between each
+     * filter element on that dimension. The dimension order is determined by the
+     * value of `data_format`, see above for details. Dilations in the batch and
+     * depth dimensions must be 1.
+     */
+    public Options dilations(List<Long> dilations) {
+      this.dilations = dilations;
+      return this;
+    }
+    
+    private Boolean useCudnnOnGpu;
+    private List<Long> explicitPaddings;
+    private String dataFormat;
+    private List<Long> dilations;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new Conv2d operation.
+   * 
+   * @param scope current scope
+   * @param input A 4-D tensor. The dimension order is interpreted according to the value
+   * of `data_format`, see below for details.
+   * @param filter A 4-D tensor of shape
+   * `[filter_height, filter_width, in_channels, out_channels]`
+   * @param strides 1-D tensor of length 4.  The stride of the sliding window for each
+   * dimension of `input`. The dimension order is determined by the value of
+   * `data_format`, see below for details.
+   * @param padding The type of padding algorithm to use.
+   * @param options carries optional attributes values
+   * @return a new instance of Conv2d
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> Conv2d<T> create(Scope scope, Operand<T> input, Operand<T> filter, List<Long> strides, String padding, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Conv2D", scope.makeOpName("Conv2d"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(filter.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    long[] stridesArray = new long[strides.size()];
+    for (int i = 0; i < stridesArray.length; ++i) {
+      stridesArray[i] = strides.get(i);
+    }
+    opBuilder.setAttr("strides", stridesArray);
+    opBuilder.setAttr("padding", padding);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.useCudnnOnGpu != null) {
+          opBuilder.setAttr("use_cudnn_on_gpu", opts.useCudnnOnGpu);
+        }
+        if (opts.explicitPaddings != null) {
+          long[] explicitPaddingsArray = new long[opts.explicitPaddings.size()];
+          for (int i = 0; i < explicitPaddingsArray.length; ++i) {
+            explicitPaddingsArray[i] = opts.explicitPaddings.get(i);
+          }
+          opBuilder.setAttr("explicit_paddings", explicitPaddingsArray);
+        }
+        if (opts.dataFormat != null) {
+          opBuilder.setAttr("data_format", opts.dataFormat);
+        }
+        if (opts.dilations != null) {
+          long[] dilationsArray = new long[opts.dilations.size()];
+          for (int i = 0; i < dilationsArray.length; ++i) {
+            dilationsArray[i] = opts.dilations.get(i);
+          }
+          opBuilder.setAttr("dilations", dilationsArray);
+        }
+      }
+    }
+    return new Conv2d<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param useCudnnOnGpu 
+   */
+  public static Options useCudnnOnGpu(Boolean useCudnnOnGpu) {
+    return new Options().useCudnnOnGpu(useCudnnOnGpu);
+  }
+  
+  /**
+   * @param explicitPaddings If `padding` is `"EXPLICIT"`, the list of explicit padding amounts. For the ith
+   * dimension, the amount of padding inserted before and after the dimension is
+   * `explicit_paddings[2 * i]` and `explicit_paddings[2 * i + 1]`, respectively. If
+   * `padding` is not `"EXPLICIT"`, `explicit_paddings` must be empty.
+   */
+  public static Options explicitPaddings(List<Long> explicitPaddings) {
+    return new Options().explicitPaddings(explicitPaddings);
+  }
+  
+  /**
+   * @param dataFormat Specify the data format of the input and output data. With the
+   * default format "NHWC", the data is stored in the order of:
+   *     [batch, height, width, channels].
+   * Alternatively, the format could be "NCHW", the data storage order of:
+   *     [batch, channels, height, width].
+   */
+  public static Options dataFormat(String dataFormat) {
+    return new Options().dataFormat(dataFormat);
+  }
+  
+  /**
+   * @param dilations 1-D tensor of length 4.  The dilation factor for each dimension of
+   * `input`. If set to k > 1, there will be k-1 skipped cells between each
+   * filter element on that dimension. The dimension order is determined by the
+   * value of `data_format`, see above for details. Dilations in the batch and
+   * depth dimensions must be 1.
+   */
+  public static Options dilations(List<Long> dilations) {
+    return new Options().dilations(dilations);
+  }
+  
+  /**
+   * A 4-D tensor. The dimension order is determined by the value of
+   * `data_format`, see below for details.
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Conv2D";
+  
+  private Output<T> output;
+  
+  private Conv2d(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Conv2dBackpropFilter.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Conv2dBackpropFilter.java
new file mode 100644
index 00000000000..709d59dee28
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Conv2dBackpropFilter.java
@@ -0,0 +1,217 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.nn;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Computes the gradients of convolution with respect to the filter.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator(group = "nn")
+public final class Conv2dBackpropFilter<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.nn.Conv2dBackpropFilter}
+   */
+  public static class Options {
+    
+    /**
+     * @param useCudnnOnGpu 
+     */
+    public Options useCudnnOnGpu(Boolean useCudnnOnGpu) {
+      this.useCudnnOnGpu = useCudnnOnGpu;
+      return this;
+    }
+    
+    /**
+     * @param explicitPaddings If `padding` is `"EXPLICIT"`, the list of explicit padding amounts. For the ith
+     * dimension, the amount of padding inserted before and after the dimension is
+     * `explicit_paddings[2 * i]` and `explicit_paddings[2 * i + 1]`, respectively. If
+     * `padding` is not `"EXPLICIT"`, `explicit_paddings` must be empty.
+     */
+    public Options explicitPaddings(List<Long> explicitPaddings) {
+      this.explicitPaddings = explicitPaddings;
+      return this;
+    }
+    
+    /**
+     * @param dataFormat Specify the data format of the input and output data. With the
+     * default format "NHWC", the data is stored in the order of:
+     *     [batch, in_height, in_width, in_channels].
+     * Alternatively, the format could be "NCHW", the data storage order of:
+     *     [batch, in_channels, in_height, in_width].
+     */
+    public Options dataFormat(String dataFormat) {
+      this.dataFormat = dataFormat;
+      return this;
+    }
+    
+    /**
+     * @param dilations 1-D tensor of length 4.  The dilation factor for each dimension of
+     * `input`. If set to k > 1, there will be k-1 skipped cells between each filter
+     * element on that dimension. The dimension order is determined by the value of
+     * `data_format`, see above for details. Dilations in the batch and depth
+     * dimensions must be 1.
+     */
+    public Options dilations(List<Long> dilations) {
+      this.dilations = dilations;
+      return this;
+    }
+    
+    private Boolean useCudnnOnGpu;
+    private List<Long> explicitPaddings;
+    private String dataFormat;
+    private List<Long> dilations;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new Conv2dBackpropFilter operation.
+   * 
+   * @param scope current scope
+   * @param input 4-D with shape `[batch, in_height, in_width, in_channels]`.
+   * @param filterSizes An integer vector representing the tensor shape of `filter`,
+   * where `filter` is a 4-D
+   * `[filter_height, filter_width, in_channels, out_channels]` tensor.
+   * @param outBackprop 4-D with shape `[batch, out_height, out_width, out_channels]`.
+   * Gradients w.r.t. the output of the convolution.
+   * @param strides The stride of the sliding window for each dimension of the input
+   * of the convolution. Must be in the same order as the dimension specified with
+   * format.
+   * @param padding The type of padding algorithm to use.
+   * @param options carries optional attributes values
+   * @return a new instance of Conv2dBackpropFilter
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> Conv2dBackpropFilter<T> create(Scope scope, Operand<T> input, Operand<TInt32> filterSizes, Operand<T> outBackprop, List<Long> strides, String padding, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Conv2DBackpropFilter", scope.makeOpName("Conv2dBackpropFilter"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(filterSizes.asOutput());
+    opBuilder.addInput(outBackprop.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    long[] stridesArray = new long[strides.size()];
+    for (int i = 0; i < stridesArray.length; ++i) {
+      stridesArray[i] = strides.get(i);
+    }
+    opBuilder.setAttr("strides", stridesArray);
+    opBuilder.setAttr("padding", padding);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.useCudnnOnGpu != null) {
+          opBuilder.setAttr("use_cudnn_on_gpu", opts.useCudnnOnGpu);
+        }
+        if (opts.explicitPaddings != null) {
+          long[] explicitPaddingsArray = new long[opts.explicitPaddings.size()];
+          for (int i = 0; i < explicitPaddingsArray.length; ++i) {
+            explicitPaddingsArray[i] = opts.explicitPaddings.get(i);
+          }
+          opBuilder.setAttr("explicit_paddings", explicitPaddingsArray);
+        }
+        if (opts.dataFormat != null) {
+          opBuilder.setAttr("data_format", opts.dataFormat);
+        }
+        if (opts.dilations != null) {
+          long[] dilationsArray = new long[opts.dilations.size()];
+          for (int i = 0; i < dilationsArray.length; ++i) {
+            dilationsArray[i] = opts.dilations.get(i);
+          }
+          opBuilder.setAttr("dilations", dilationsArray);
+        }
+      }
+    }
+    return new Conv2dBackpropFilter<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param useCudnnOnGpu 
+   */
+  public static Options useCudnnOnGpu(Boolean useCudnnOnGpu) {
+    return new Options().useCudnnOnGpu(useCudnnOnGpu);
+  }
+  
+  /**
+   * @param explicitPaddings If `padding` is `"EXPLICIT"`, the list of explicit padding amounts. For the ith
+   * dimension, the amount of padding inserted before and after the dimension is
+   * `explicit_paddings[2 * i]` and `explicit_paddings[2 * i + 1]`, respectively. If
+   * `padding` is not `"EXPLICIT"`, `explicit_paddings` must be empty.
+   */
+  public static Options explicitPaddings(List<Long> explicitPaddings) {
+    return new Options().explicitPaddings(explicitPaddings);
+  }
+  
+  /**
+   * @param dataFormat Specify the data format of the input and output data. With the
+   * default format "NHWC", the data is stored in the order of:
+   *     [batch, in_height, in_width, in_channels].
+   * Alternatively, the format could be "NCHW", the data storage order of:
+   *     [batch, in_channels, in_height, in_width].
+   */
+  public static Options dataFormat(String dataFormat) {
+    return new Options().dataFormat(dataFormat);
+  }
+  
+  /**
+   * @param dilations 1-D tensor of length 4.  The dilation factor for each dimension of
+   * `input`. If set to k > 1, there will be k-1 skipped cells between each filter
+   * element on that dimension. The dimension order is determined by the value of
+   * `data_format`, see above for details. Dilations in the batch and depth
+   * dimensions must be 1.
+   */
+  public static Options dilations(List<Long> dilations) {
+    return new Options().dilations(dilations);
+  }
+  
+  /**
+   * 4-D with shape
+   * `[filter_height, filter_width, in_channels, out_channels]`.  Gradient w.r.t.
+   * the `filter` input of the convolution.
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Conv2DBackpropFilter";
+  
+  private Output<T> output;
+  
+  private Conv2dBackpropFilter(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Conv2dBackpropInput.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Conv2dBackpropInput.java
new file mode 100644
index 00000000000..b725823e8db
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Conv2dBackpropInput.java
@@ -0,0 +1,216 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.nn;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Computes the gradients of convolution with respect to the input.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator(group = "nn")
+public final class Conv2dBackpropInput<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.nn.Conv2dBackpropInput}
+   */
+  public static class Options {
+    
+    /**
+     * @param useCudnnOnGpu 
+     */
+    public Options useCudnnOnGpu(Boolean useCudnnOnGpu) {
+      this.useCudnnOnGpu = useCudnnOnGpu;
+      return this;
+    }
+    
+    /**
+     * @param explicitPaddings If `padding` is `"EXPLICIT"`, the list of explicit padding amounts. For the ith
+     * dimension, the amount of padding inserted before and after the dimension is
+     * `explicit_paddings[2 * i]` and `explicit_paddings[2 * i + 1]`, respectively. If
+     * `padding` is not `"EXPLICIT"`, `explicit_paddings` must be empty.
+     */
+    public Options explicitPaddings(List<Long> explicitPaddings) {
+      this.explicitPaddings = explicitPaddings;
+      return this;
+    }
+    
+    /**
+     * @param dataFormat Specify the data format of the input and output data. With the
+     * default format "NHWC", the data is stored in the order of:
+     *     [batch, in_height, in_width, in_channels].
+     * Alternatively, the format could be "NCHW", the data storage order of:
+     *     [batch, in_channels, in_height, in_width].
+     */
+    public Options dataFormat(String dataFormat) {
+      this.dataFormat = dataFormat;
+      return this;
+    }
+    
+    /**
+     * @param dilations 1-D tensor of length 4.  The dilation factor for each dimension of
+     * `input`. If set to k > 1, there will be k-1 skipped cells between each filter
+     * element on that dimension. The dimension order is determined by the value of
+     * `data_format`, see above for details. Dilations in the batch and depth
+     * dimensions must be 1.
+     */
+    public Options dilations(List<Long> dilations) {
+      this.dilations = dilations;
+      return this;
+    }
+    
+    private Boolean useCudnnOnGpu;
+    private List<Long> explicitPaddings;
+    private String dataFormat;
+    private List<Long> dilations;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new Conv2dBackpropInput operation.
+   * 
+   * @param scope current scope
+   * @param inputSizes An integer vector representing the shape of `input`,
+   * where `input` is a 4-D `[batch, height, width, channels]` tensor.
+   * @param filter 4-D with shape
+   * `[filter_height, filter_width, in_channels, out_channels]`.
+   * @param outBackprop 4-D with shape `[batch, out_height, out_width, out_channels]`.
+   * Gradients w.r.t. the output of the convolution.
+   * @param strides The stride of the sliding window for each dimension of the input
+   * of the convolution. Must be in the same order as the dimension specified with
+   * format.
+   * @param padding The type of padding algorithm to use.
+   * @param options carries optional attributes values
+   * @return a new instance of Conv2dBackpropInput
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> Conv2dBackpropInput<T> create(Scope scope, Operand<TInt32> inputSizes, Operand<T> filter, Operand<T> outBackprop, List<Long> strides, String padding, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Conv2DBackpropInput", scope.makeOpName("Conv2dBackpropInput"));
+    opBuilder.addInput(inputSizes.asOutput());
+    opBuilder.addInput(filter.asOutput());
+    opBuilder.addInput(outBackprop.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    long[] stridesArray = new long[strides.size()];
+    for (int i = 0; i < stridesArray.length; ++i) {
+      stridesArray[i] = strides.get(i);
+    }
+    opBuilder.setAttr("strides", stridesArray);
+    opBuilder.setAttr("padding", padding);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.useCudnnOnGpu != null) {
+          opBuilder.setAttr("use_cudnn_on_gpu", opts.useCudnnOnGpu);
+        }
+        if (opts.explicitPaddings != null) {
+          long[] explicitPaddingsArray = new long[opts.explicitPaddings.size()];
+          for (int i = 0; i < explicitPaddingsArray.length; ++i) {
+            explicitPaddingsArray[i] = opts.explicitPaddings.get(i);
+          }
+          opBuilder.setAttr("explicit_paddings", explicitPaddingsArray);
+        }
+        if (opts.dataFormat != null) {
+          opBuilder.setAttr("data_format", opts.dataFormat);
+        }
+        if (opts.dilations != null) {
+          long[] dilationsArray = new long[opts.dilations.size()];
+          for (int i = 0; i < dilationsArray.length; ++i) {
+            dilationsArray[i] = opts.dilations.get(i);
+          }
+          opBuilder.setAttr("dilations", dilationsArray);
+        }
+      }
+    }
+    return new Conv2dBackpropInput<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param useCudnnOnGpu 
+   */
+  public static Options useCudnnOnGpu(Boolean useCudnnOnGpu) {
+    return new Options().useCudnnOnGpu(useCudnnOnGpu);
+  }
+  
+  /**
+   * @param explicitPaddings If `padding` is `"EXPLICIT"`, the list of explicit padding amounts. For the ith
+   * dimension, the amount of padding inserted before and after the dimension is
+   * `explicit_paddings[2 * i]` and `explicit_paddings[2 * i + 1]`, respectively. If
+   * `padding` is not `"EXPLICIT"`, `explicit_paddings` must be empty.
+   */
+  public static Options explicitPaddings(List<Long> explicitPaddings) {
+    return new Options().explicitPaddings(explicitPaddings);
+  }
+  
+  /**
+   * @param dataFormat Specify the data format of the input and output data. With the
+   * default format "NHWC", the data is stored in the order of:
+   *     [batch, in_height, in_width, in_channels].
+   * Alternatively, the format could be "NCHW", the data storage order of:
+   *     [batch, in_channels, in_height, in_width].
+   */
+  public static Options dataFormat(String dataFormat) {
+    return new Options().dataFormat(dataFormat);
+  }
+  
+  /**
+   * @param dilations 1-D tensor of length 4.  The dilation factor for each dimension of
+   * `input`. If set to k > 1, there will be k-1 skipped cells between each filter
+   * element on that dimension. The dimension order is determined by the value of
+   * `data_format`, see above for details. Dilations in the batch and depth
+   * dimensions must be 1.
+   */
+  public static Options dilations(List<Long> dilations) {
+    return new Options().dilations(dilations);
+  }
+  
+  /**
+   * 4-D with shape `[batch, in_height, in_width, in_channels]`.  Gradient
+   * w.r.t. the input of the convolution.
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Conv2DBackpropInput";
+  
+  private Output<T> output;
+  
+  private Conv2dBackpropInput(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Conv3d.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Conv3d.java
new file mode 100644
index 00000000000..0720ef0dce5
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Conv3d.java
@@ -0,0 +1,166 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.nn;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Computes a 3-D convolution given 5-D `input` and `filter` tensors.
+ * <p>
+ * In signal processing, cross-correlation is a measure of similarity of
+ * two waveforms as a function of a time-lag applied to one of them. This
+ * is also known as a sliding dot product or sliding inner-product.
+ * <p>
+ * Our Conv3D implements a form of cross-correlation.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator(group = "nn")
+public final class Conv3d<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.nn.Conv3d}
+   */
+  public static class Options {
+    
+    /**
+     * @param dataFormat The data format of the input and output data. With the
+     * default format "NDHWC", the data is stored in the order of:
+     *     [batch, in_depth, in_height, in_width, in_channels].
+     * Alternatively, the format could be "NCDHW", the data storage order is:
+     *     [batch, in_channels, in_depth, in_height, in_width].
+     */
+    public Options dataFormat(String dataFormat) {
+      this.dataFormat = dataFormat;
+      return this;
+    }
+    
+    /**
+     * @param dilations 1-D tensor of length 5.  The dilation factor for each dimension of
+     * `input`. If set to k > 1, there will be k-1 skipped cells between each
+     * filter element on that dimension. The dimension order is determined by the
+     * value of `data_format`, see above for details. Dilations in the batch and
+     * depth dimensions must be 1.
+     */
+    public Options dilations(List<Long> dilations) {
+      this.dilations = dilations;
+      return this;
+    }
+    
+    private String dataFormat;
+    private List<Long> dilations;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new Conv3d operation.
+   * 
+   * @param scope current scope
+   * @param input Shape `[batch, in_depth, in_height, in_width, in_channels]`.
+   * @param filter Shape `[filter_depth, filter_height, filter_width, in_channels,
+   * out_channels]`. `in_channels` must match between `input` and `filter`.
+   * @param strides 1-D tensor of length 5. The stride of the sliding window for each
+   * dimension of `input`. Must have `strides[0] = strides[4] = 1`.
+   * @param padding The type of padding algorithm to use.
+   * @param options carries optional attributes values
+   * @return a new instance of Conv3d
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> Conv3d<T> create(Scope scope, Operand<T> input, Operand<T> filter, List<Long> strides, String padding, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Conv3D", scope.makeOpName("Conv3d"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(filter.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    long[] stridesArray = new long[strides.size()];
+    for (int i = 0; i < stridesArray.length; ++i) {
+      stridesArray[i] = strides.get(i);
+    }
+    opBuilder.setAttr("strides", stridesArray);
+    opBuilder.setAttr("padding", padding);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.dataFormat != null) {
+          opBuilder.setAttr("data_format", opts.dataFormat);
+        }
+        if (opts.dilations != null) {
+          long[] dilationsArray = new long[opts.dilations.size()];
+          for (int i = 0; i < dilationsArray.length; ++i) {
+            dilationsArray[i] = opts.dilations.get(i);
+          }
+          opBuilder.setAttr("dilations", dilationsArray);
+        }
+      }
+    }
+    return new Conv3d<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param dataFormat The data format of the input and output data. With the
+   * default format "NDHWC", the data is stored in the order of:
+   *     [batch, in_depth, in_height, in_width, in_channels].
+   * Alternatively, the format could be "NCDHW", the data storage order is:
+   *     [batch, in_channels, in_depth, in_height, in_width].
+   */
+  public static Options dataFormat(String dataFormat) {
+    return new Options().dataFormat(dataFormat);
+  }
+  
+  /**
+   * @param dilations 1-D tensor of length 5.  The dilation factor for each dimension of
+   * `input`. If set to k > 1, there will be k-1 skipped cells between each
+   * filter element on that dimension. The dimension order is determined by the
+   * value of `data_format`, see above for details. Dilations in the batch and
+   * depth dimensions must be 1.
+   */
+  public static Options dilations(List<Long> dilations) {
+    return new Options().dilations(dilations);
+  }
+  
+  /**
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Conv3D";
+  
+  private Output<T> output;
+  
+  private Conv3d(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Conv3dBackpropFilter.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Conv3dBackpropFilter.java
new file mode 100644
index 00000000000..1de81e9f781
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Conv3dBackpropFilter.java
@@ -0,0 +1,166 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.nn;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Computes the gradients of 3-D convolution with respect to the filter.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator(group = "nn")
+public final class Conv3dBackpropFilter<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.nn.Conv3dBackpropFilter}
+   */
+  public static class Options {
+    
+    /**
+     * @param dataFormat The data format of the input and output data. With the
+     * default format "NDHWC", the data is stored in the order of:
+     *     [batch, in_depth, in_height, in_width, in_channels].
+     * Alternatively, the format could be "NCDHW", the data storage order is:
+     *     [batch, in_channels, in_depth, in_height, in_width].
+     */
+    public Options dataFormat(String dataFormat) {
+      this.dataFormat = dataFormat;
+      return this;
+    }
+    
+    /**
+     * @param dilations 1-D tensor of length 5.  The dilation factor for each dimension of
+     * `input`. If set to k > 1, there will be k-1 skipped cells between each
+     * filter element on that dimension. The dimension order is determined by the
+     * value of `data_format`, see above for details. Dilations in the batch and
+     * depth dimensions must be 1.
+     */
+    public Options dilations(List<Long> dilations) {
+      this.dilations = dilations;
+      return this;
+    }
+    
+    private String dataFormat;
+    private List<Long> dilations;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new Conv3dBackpropFilter operation.
+   * 
+   * @param scope current scope
+   * @param input Shape `[batch, depth, rows, cols, in_channels]`.
+   * @param filterSizes An integer vector representing the tensor shape of `filter`,
+   * where `filter` is a 5-D
+   * `[filter_depth, filter_height, filter_width, in_channels, out_channels]`
+   * tensor.
+   * @param outBackprop Backprop signal of shape `[batch, out_depth, out_rows, out_cols,
+   * out_channels]`.
+   * @param strides 1-D tensor of length 5. The stride of the sliding window for each
+   * dimension of `input`. Must have `strides[0] = strides[4] = 1`.
+   * @param padding The type of padding algorithm to use.
+   * @param options carries optional attributes values
+   * @return a new instance of Conv3dBackpropFilter
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> Conv3dBackpropFilter<T> create(Scope scope, Operand<T> input, Operand<TInt32> filterSizes, Operand<T> outBackprop, List<Long> strides, String padding, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Conv3DBackpropFilterV2", scope.makeOpName("Conv3dBackpropFilter"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(filterSizes.asOutput());
+    opBuilder.addInput(outBackprop.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    long[] stridesArray = new long[strides.size()];
+    for (int i = 0; i < stridesArray.length; ++i) {
+      stridesArray[i] = strides.get(i);
+    }
+    opBuilder.setAttr("strides", stridesArray);
+    opBuilder.setAttr("padding", padding);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.dataFormat != null) {
+          opBuilder.setAttr("data_format", opts.dataFormat);
+        }
+        if (opts.dilations != null) {
+          long[] dilationsArray = new long[opts.dilations.size()];
+          for (int i = 0; i < dilationsArray.length; ++i) {
+            dilationsArray[i] = opts.dilations.get(i);
+          }
+          opBuilder.setAttr("dilations", dilationsArray);
+        }
+      }
+    }
+    return new Conv3dBackpropFilter<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param dataFormat The data format of the input and output data. With the
+   * default format "NDHWC", the data is stored in the order of:
+   *     [batch, in_depth, in_height, in_width, in_channels].
+   * Alternatively, the format could be "NCDHW", the data storage order is:
+   *     [batch, in_channels, in_depth, in_height, in_width].
+   */
+  public static Options dataFormat(String dataFormat) {
+    return new Options().dataFormat(dataFormat);
+  }
+  
+  /**
+   * @param dilations 1-D tensor of length 5.  The dilation factor for each dimension of
+   * `input`. If set to k > 1, there will be k-1 skipped cells between each
+   * filter element on that dimension. The dimension order is determined by the
+   * value of `data_format`, see above for details. Dilations in the batch and
+   * depth dimensions must be 1.
+   */
+  public static Options dilations(List<Long> dilations) {
+    return new Options().dilations(dilations);
+  }
+  
+  /**
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Conv3DBackpropFilterV2";
+  
+  private Output<T> output;
+  
+  private Conv3dBackpropFilter(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Conv3dBackpropInput.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Conv3dBackpropInput.java
new file mode 100644
index 00000000000..1f71f6b64f3
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Conv3dBackpropInput.java
@@ -0,0 +1,165 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.nn;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Computes the gradients of 3-D convolution with respect to the input.
+ * 
+ * @param <U> data type for {@code output()} output
+ */
+@Operator(group = "nn")
+public final class Conv3dBackpropInput<U extends TNumber> extends RawOp implements Operand<U> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.nn.Conv3dBackpropInput}
+   */
+  public static class Options {
+    
+    /**
+     * @param dataFormat The data format of the input and output data. With the
+     * default format "NDHWC", the data is stored in the order of:
+     *     [batch, in_depth, in_height, in_width, in_channels].
+     * Alternatively, the format could be "NCDHW", the data storage order is:
+     *     [batch, in_channels, in_depth, in_height, in_width].
+     */
+    public Options dataFormat(String dataFormat) {
+      this.dataFormat = dataFormat;
+      return this;
+    }
+    
+    /**
+     * @param dilations 1-D tensor of length 5.  The dilation factor for each dimension of
+     * `input`. If set to k > 1, there will be k-1 skipped cells between each
+     * filter element on that dimension. The dimension order is determined by the
+     * value of `data_format`, see above for details. Dilations in the batch and
+     * depth dimensions must be 1.
+     */
+    public Options dilations(List<Long> dilations) {
+      this.dilations = dilations;
+      return this;
+    }
+    
+    private String dataFormat;
+    private List<Long> dilations;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new Conv3dBackpropInput operation.
+   * 
+   * @param scope current scope
+   * @param inputSizes An integer vector representing the tensor shape of `input`,
+   * where `input` is a 5-D
+   * `[batch, depth, rows, cols, in_channels]` tensor.
+   * @param filter Shape `[depth, rows, cols, in_channels, out_channels]`.
+   * `in_channels` must match between `input` and `filter`.
+   * @param outBackprop Backprop signal of shape `[batch, out_depth, out_rows, out_cols,
+   * out_channels]`.
+   * @param strides 1-D tensor of length 5. The stride of the sliding window for each
+   * dimension of `input`. Must have `strides[0] = strides[4] = 1`.
+   * @param padding The type of padding algorithm to use.
+   * @param options carries optional attributes values
+   * @return a new instance of Conv3dBackpropInput
+   */
+  @Endpoint(describeByClass = true)
+  public static <U extends TNumber, T extends TNumber> Conv3dBackpropInput<U> create(Scope scope, Operand<T> inputSizes, Operand<U> filter, Operand<U> outBackprop, List<Long> strides, String padding, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Conv3DBackpropInputV2", scope.makeOpName("Conv3dBackpropInput"));
+    opBuilder.addInput(inputSizes.asOutput());
+    opBuilder.addInput(filter.asOutput());
+    opBuilder.addInput(outBackprop.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    long[] stridesArray = new long[strides.size()];
+    for (int i = 0; i < stridesArray.length; ++i) {
+      stridesArray[i] = strides.get(i);
+    }
+    opBuilder.setAttr("strides", stridesArray);
+    opBuilder.setAttr("padding", padding);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.dataFormat != null) {
+          opBuilder.setAttr("data_format", opts.dataFormat);
+        }
+        if (opts.dilations != null) {
+          long[] dilationsArray = new long[opts.dilations.size()];
+          for (int i = 0; i < dilationsArray.length; ++i) {
+            dilationsArray[i] = opts.dilations.get(i);
+          }
+          opBuilder.setAttr("dilations", dilationsArray);
+        }
+      }
+    }
+    return new Conv3dBackpropInput<U>(opBuilder.build());
+  }
+  
+  /**
+   * @param dataFormat The data format of the input and output data. With the
+   * default format "NDHWC", the data is stored in the order of:
+   *     [batch, in_depth, in_height, in_width, in_channels].
+   * Alternatively, the format could be "NCDHW", the data storage order is:
+   *     [batch, in_channels, in_depth, in_height, in_width].
+   */
+  public static Options dataFormat(String dataFormat) {
+    return new Options().dataFormat(dataFormat);
+  }
+  
+  /**
+   * @param dilations 1-D tensor of length 5.  The dilation factor for each dimension of
+   * `input`. If set to k > 1, there will be k-1 skipped cells between each
+   * filter element on that dimension. The dimension order is determined by the
+   * value of `data_format`, see above for details. Dilations in the batch and
+   * depth dimensions must be 1.
+   */
+  public static Options dilations(List<Long> dilations) {
+    return new Options().dilations(dilations);
+  }
+  
+  /**
+   */
+  public Output<U> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<U> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Conv3DBackpropInputV2";
+  
+  private Output<U> output;
+  
+  private Conv3dBackpropInput(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CtcBeamSearchDecoder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CtcBeamSearchDecoder.java
new file mode 100644
index 00000000000..fadc3985f38
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CtcBeamSearchDecoder.java
@@ -0,0 +1,161 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.nn;
+
+import java.util.Arrays;
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Performs beam search decoding on the logits given in input.
+ * <p>
+ * A note about the attribute merge_repeated: For the beam search decoder,
+ * this means that if consecutive entries in a beam are the same, only
+ * the first of these is emitted.  That is, when the top path is "A B B B B",
+ * "A B" is returned if merge_repeated = True but "A B B B B" is
+ * returned if merge_repeated = False.
+ * 
+ * @param <T> data type for {@code logProbability()} output
+ */
+@Operator(group = "nn")
+public final class CtcBeamSearchDecoder<T extends TNumber> extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.nn.CtcBeamSearchDecoder}
+   */
+  public static class Options {
+    
+    /**
+     * @param mergeRepeated If true, merge repeated classes in output.
+     */
+    public Options mergeRepeated(Boolean mergeRepeated) {
+      this.mergeRepeated = mergeRepeated;
+      return this;
+    }
+    
+    private Boolean mergeRepeated;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new CtcBeamSearchDecoder operation.
+   * 
+   * @param scope current scope
+   * @param inputs 3-D, shape: `(max_time x batch_size x num_classes)`, the logits.
+   * @param sequenceLength A vector containing sequence lengths, size `(batch)`.
+   * @param beamWidth A scalar >= 0 (beam search beam width).
+   * @param topPaths A scalar >= 0, <= beam_width (controls output size).
+   * @param options carries optional attributes values
+   * @return a new instance of CtcBeamSearchDecoder
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> CtcBeamSearchDecoder<T> create(Scope scope, Operand<T> inputs, Operand<TInt32> sequenceLength, Long beamWidth, Long topPaths, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("CTCBeamSearchDecoder", scope.makeOpName("CtcBeamSearchDecoder"));
+    opBuilder.addInput(inputs.asOutput());
+    opBuilder.addInput(sequenceLength.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("beam_width", beamWidth);
+    opBuilder.setAttr("top_paths", topPaths);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.mergeRepeated != null) {
+          opBuilder.setAttr("merge_repeated", opts.mergeRepeated);
+        }
+      }
+    }
+    return new CtcBeamSearchDecoder<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param mergeRepeated If true, merge repeated classes in output.
+   */
+  public static Options mergeRepeated(Boolean mergeRepeated) {
+    return new Options().mergeRepeated(mergeRepeated);
+  }
+  
+  /**
+   * A list (length: top_paths) of indices matrices.  Matrix j,
+   * size `(total_decoded_outputs[j] x 2)`, has indices of a
+   * `SparseTensor<int64, 2>`.  The rows store: [batch, time].
+   */
+  public List<Output<TInt64>> decodedIndices() {
+    return decodedIndices;
+  }
+  
+  /**
+   * A list (length: top_paths) of values vectors.  Vector j,
+   * size `(length total_decoded_outputs[j])`, has the values of a
+   * `SparseTensor<int64, 2>`.  The vector stores the decoded classes for beam j.
+   */
+  public List<Output<TInt64>> decodedValues() {
+    return decodedValues;
+  }
+  
+  /**
+   * A list (length: top_paths) of shape vector.  Vector j,
+   * size `(2)`, stores the shape of the decoded `SparseTensor[j]`.
+   * Its values are: `[batch_size, max_decoded_length[j]]`.
+   */
+  public List<Output<TInt64>> decodedShape() {
+    return decodedShape;
+  }
+  
+  /**
+   * A matrix, shaped: `(batch_size x top_paths)`.  The
+   * sequence log-probabilities.
+   */
+  public Output<T> logProbability() {
+    return logProbability;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "CTCBeamSearchDecoder";
+  
+  private List<Output<TInt64>> decodedIndices;
+  private List<Output<TInt64>> decodedValues;
+  private List<Output<TInt64>> decodedShape;
+  private Output<T> logProbability;
+  
+  @SuppressWarnings("unchecked")
+  private CtcBeamSearchDecoder(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    int decodedIndicesLength = operation.outputListLength("decoded_indices");
+    decodedIndices = Arrays.asList((Output<TInt64>[])operation.outputList(outputIdx, decodedIndicesLength));
+    outputIdx += decodedIndicesLength;
+    int decodedValuesLength = operation.outputListLength("decoded_values");
+    decodedValues = Arrays.asList((Output<TInt64>[])operation.outputList(outputIdx, decodedValuesLength));
+    outputIdx += decodedValuesLength;
+    int decodedShapeLength = operation.outputListLength("decoded_shape");
+    decodedShape = Arrays.asList((Output<TInt64>[])operation.outputList(outputIdx, decodedShapeLength));
+    outputIdx += decodedShapeLength;
+    logProbability = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CtcGreedyDecoder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CtcGreedyDecoder.java
new file mode 100644
index 00000000000..2c985fb4e03
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CtcGreedyDecoder.java
@@ -0,0 +1,149 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.nn;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Performs greedy decoding on the logits given in inputs.
+ * <p>
+ * A note about the attribute merge_repeated: if enabled, when
+ * consecutive logits' maximum indices are the same, only the first of
+ * these is emitted.  Labeling the blank '*', the sequence "A B B * B B"
+ * becomes "A B B" if merge_repeated = True and "A B B B B" if
+ * merge_repeated = False.
+ * <p>
+ * Regardless of the value of merge_repeated, if the maximum index of a given
+ * time and batch corresponds to the blank, index `(num_classes - 1)`, no new
+ * element is emitted.
+ * 
+ * @param <T> data type for {@code logProbability()} output
+ */
+@Operator(group = "nn")
+public final class CtcGreedyDecoder<T extends TNumber> extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.nn.CtcGreedyDecoder}
+   */
+  public static class Options {
+    
+    /**
+     * @param mergeRepeated If True, merge repeated classes in output.
+     */
+    public Options mergeRepeated(Boolean mergeRepeated) {
+      this.mergeRepeated = mergeRepeated;
+      return this;
+    }
+    
+    private Boolean mergeRepeated;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new CtcGreedyDecoder operation.
+   * 
+   * @param scope current scope
+   * @param inputs 3-D, shape: `(max_time x batch_size x num_classes)`, the logits.
+   * @param sequenceLength A vector containing sequence lengths, size `(batch_size)`.
+   * @param options carries optional attributes values
+   * @return a new instance of CtcGreedyDecoder
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> CtcGreedyDecoder<T> create(Scope scope, Operand<T> inputs, Operand<TInt32> sequenceLength, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("CTCGreedyDecoder", scope.makeOpName("CtcGreedyDecoder"));
+    opBuilder.addInput(inputs.asOutput());
+    opBuilder.addInput(sequenceLength.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.mergeRepeated != null) {
+          opBuilder.setAttr("merge_repeated", opts.mergeRepeated);
+        }
+      }
+    }
+    return new CtcGreedyDecoder<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param mergeRepeated If True, merge repeated classes in output.
+   */
+  public static Options mergeRepeated(Boolean mergeRepeated) {
+    return new Options().mergeRepeated(mergeRepeated);
+  }
+  
+  /**
+   * Indices matrix, size `(total_decoded_outputs x 2)`,
+   * of a `SparseTensor<int64, 2>`.  The rows store: [batch, time].
+   */
+  public Output<TInt64> decodedIndices() {
+    return decodedIndices;
+  }
+  
+  /**
+   * Values vector, size: `(total_decoded_outputs)`,
+   * of a `SparseTensor<int64, 2>`.  The vector stores the decoded classes.
+   */
+  public Output<TInt64> decodedValues() {
+    return decodedValues;
+  }
+  
+  /**
+   * Shape vector, size `(2)`, of the decoded SparseTensor.
+   * Values are: `[batch_size, max_decoded_length]`.
+   */
+  public Output<TInt64> decodedShape() {
+    return decodedShape;
+  }
+  
+  /**
+   * Matrix, size `(batch_size x 1)`, containing sequence
+   * log-probabilities.
+   */
+  public Output<T> logProbability() {
+    return logProbability;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "CTCGreedyDecoder";
+  
+  private Output<TInt64> decodedIndices;
+  private Output<TInt64> decodedValues;
+  private Output<TInt64> decodedShape;
+  private Output<T> logProbability;
+  
+  private CtcGreedyDecoder(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    decodedIndices = operation.output(outputIdx++);
+    decodedValues = operation.output(outputIdx++);
+    decodedShape = operation.output(outputIdx++);
+    logProbability = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CtcLoss.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CtcLoss.java
new file mode 100644
index 00000000000..44acdfbad19
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CtcLoss.java
@@ -0,0 +1,175 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.nn;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Calculates the CTC Loss (log probability) for each batch entry.  Also calculates
+ * <p>
+ * the gradient.  This class performs the softmax operation for you, so inputs
+ * should be e.g. linear projections of outputs by an LSTM.
+ * 
+ * @param <T> data type for {@code loss()} output
+ */
+@Operator(group = "nn")
+public final class CtcLoss<T extends TNumber> extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.nn.CtcLoss}
+   */
+  public static class Options {
+    
+    /**
+     * @param preprocessCollapseRepeated Scalar, if true then repeated labels are
+     * collapsed prior to the CTC calculation.
+     */
+    public Options preprocessCollapseRepeated(Boolean preprocessCollapseRepeated) {
+      this.preprocessCollapseRepeated = preprocessCollapseRepeated;
+      return this;
+    }
+    
+    /**
+     * @param ctcMergeRepeated Scalar.  If set to false, <i>during</i> CTC calculation
+     * repeated non-blank labels will not be merged and are interpreted as
+     * individual labels.  This is a simplified version of CTC.
+     */
+    public Options ctcMergeRepeated(Boolean ctcMergeRepeated) {
+      this.ctcMergeRepeated = ctcMergeRepeated;
+      return this;
+    }
+    
+    /**
+     * @param ignoreLongerOutputsThanInputs Scalar. If set to true, during CTC
+     * calculation, items that have longer output sequences than input sequences
+     * are skipped: they don't contribute to the loss term and have zero-gradient.
+     */
+    public Options ignoreLongerOutputsThanInputs(Boolean ignoreLongerOutputsThanInputs) {
+      this.ignoreLongerOutputsThanInputs = ignoreLongerOutputsThanInputs;
+      return this;
+    }
+    
+    private Boolean preprocessCollapseRepeated;
+    private Boolean ctcMergeRepeated;
+    private Boolean ignoreLongerOutputsThanInputs;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new CtcLoss operation.
+   * 
+   * @param scope current scope
+   * @param inputs 3-D, shape: `(max_time x batch_size x num_classes)`, the logits.
+   * @param labelsIndices The indices of a `SparseTensor<int32, 2>`.
+   * `labels_indices(i, :) == [b, t]` means `labels_values(i)` stores the id for
+   * `(batch b, time t)`.
+   * @param labelsValues The values (labels) associated with the given batch and time.
+   * @param sequenceLength A vector containing sequence lengths (batch).
+   * @param options carries optional attributes values
+   * @return a new instance of CtcLoss
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> CtcLoss<T> create(Scope scope, Operand<T> inputs, Operand<TInt64> labelsIndices, Operand<TInt32> labelsValues, Operand<TInt32> sequenceLength, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("CTCLoss", scope.makeOpName("CtcLoss"));
+    opBuilder.addInput(inputs.asOutput());
+    opBuilder.addInput(labelsIndices.asOutput());
+    opBuilder.addInput(labelsValues.asOutput());
+    opBuilder.addInput(sequenceLength.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.preprocessCollapseRepeated != null) {
+          opBuilder.setAttr("preprocess_collapse_repeated", opts.preprocessCollapseRepeated);
+        }
+        if (opts.ctcMergeRepeated != null) {
+          opBuilder.setAttr("ctc_merge_repeated", opts.ctcMergeRepeated);
+        }
+        if (opts.ignoreLongerOutputsThanInputs != null) {
+          opBuilder.setAttr("ignore_longer_outputs_than_inputs", opts.ignoreLongerOutputsThanInputs);
+        }
+      }
+    }
+    return new CtcLoss<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param preprocessCollapseRepeated Scalar, if true then repeated labels are
+   * collapsed prior to the CTC calculation.
+   */
+  public static Options preprocessCollapseRepeated(Boolean preprocessCollapseRepeated) {
+    return new Options().preprocessCollapseRepeated(preprocessCollapseRepeated);
+  }
+  
+  /**
+   * @param ctcMergeRepeated Scalar.  If set to false, <i>during</i> CTC calculation
+   * repeated non-blank labels will not be merged and are interpreted as
+   * individual labels.  This is a simplified version of CTC.
+   */
+  public static Options ctcMergeRepeated(Boolean ctcMergeRepeated) {
+    return new Options().ctcMergeRepeated(ctcMergeRepeated);
+  }
+  
+  /**
+   * @param ignoreLongerOutputsThanInputs Scalar. If set to true, during CTC
+   * calculation, items that have longer output sequences than input sequences
+   * are skipped: they don't contribute to the loss term and have zero-gradient.
+   */
+  public static Options ignoreLongerOutputsThanInputs(Boolean ignoreLongerOutputsThanInputs) {
+    return new Options().ignoreLongerOutputsThanInputs(ignoreLongerOutputsThanInputs);
+  }
+  
+  /**
+   * A vector (batch) containing log-probabilities.
+   */
+  public Output<T> loss() {
+    return loss;
+  }
+  
+  /**
+   * The gradient of `loss`.  3-D, shape:
+   * `(max_time x batch_size x num_classes)`.
+   */
+  public Output<T> gradient() {
+    return gradient;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "CTCLoss";
+  
+  private Output<T> loss;
+  private Output<T> gradient;
+  
+  private CtcLoss(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    loss = operation.output(outputIdx++);
+    gradient = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CudnnRNN.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CudnnRNN.java
new file mode 100644
index 00000000000..44495c8b7a7
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CudnnRNN.java
@@ -0,0 +1,333 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.nn;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * A RNN backed by cuDNN.
+ * <p>
+ * Computes the RNN from the input and initial states, with respect to the params
+ * buffer. Accepts one extra input "sequence_lengths" than CudnnRNN.
+ * <p>
+ * rnn_mode: Indicates the type of the RNN model.
+ * input_mode: Indicates whether there is a linear projection between the input and
+ *   the actual computation before the first layer. 'skip_input' is only allowed
+ *   when input_size == num_units; 'auto_select' implies 'skip_input' when
+ *   input_size == num_units; otherwise, it implies 'linear_input'.
+ * direction: Indicates whether a bidirectional model will be used. Should be
+ *   "unidirectional" or "bidirectional".
+ * dropout: Dropout probability. When set to 0., dropout is disabled.
+ * seed: The 1st part of a seed to initialize dropout.
+ * seed2: The 2nd part of a seed to initialize dropout.
+ * input: If time_major is true, this is a 3-D tensor with the shape of
+ *     [seq_length, batch_size, input_size]. If time_major is false, the shape is
+ *     [batch_size, seq_length, input_size].
+ * input_h: If time_major is true, this is a 3-D tensor with the shape of
+ *     [num_layer * dir, batch_size, num_units]. If time_major is false, the shape
+ *     is [batch_size, num_layer * dir, num_units].
+ * input_c: For LSTM, a 3-D tensor with the shape of
+ *     [num_layer * dir, batch, num_units]. For other models, it is ignored.
+ * params: A 1-D tensor that contains the weights and biases in an opaque layout.
+ *     The size must be created through CudnnRNNParamsSize, and initialized
+ *     separately. Note that they might not be compatible across different
+ *     generations. So it is a good idea to save and restore
+ * sequence_lengths: a vector of lengths of each input sequence.
+ * output: If time_major is true, this is a 3-D tensor with the shape of
+ *     [seq_length, batch_size, dir * num_units]. If time_major is false, the
+ *     shape is [batch_size, seq_length, dir * num_units].
+ * output_h: The same shape has input_h.
+ * output_c: The same shape as input_c for LSTM. An empty tensor for other models.
+ * is_training: Indicates whether this operation is used for inference or
+ *   training.
+ * time_major: Indicates whether the input/output format is time major or batch
+ *     major.
+ * reserve_space: An opaque tensor that can be used in backprop calculation. It
+ *   is only produced if is_training is true.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+public final class CudnnRNN<T extends TNumber> extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.nn.CudnnRNN}
+   */
+  public static class Options {
+    
+    /**
+     * @param rnnMode 
+     */
+    public Options rnnMode(String rnnMode) {
+      this.rnnMode = rnnMode;
+      return this;
+    }
+    
+    /**
+     * @param inputMode 
+     */
+    public Options inputMode(String inputMode) {
+      this.inputMode = inputMode;
+      return this;
+    }
+    
+    /**
+     * @param direction 
+     */
+    public Options direction(String direction) {
+      this.direction = direction;
+      return this;
+    }
+    
+    /**
+     * @param dropout 
+     */
+    public Options dropout(Float dropout) {
+      this.dropout = dropout;
+      return this;
+    }
+    
+    /**
+     * @param seed 
+     */
+    public Options seed(Long seed) {
+      this.seed = seed;
+      return this;
+    }
+    
+    /**
+     * @param seed2 
+     */
+    public Options seed2(Long seed2) {
+      this.seed2 = seed2;
+      return this;
+    }
+    
+    /**
+     * @param numProj 
+     */
+    public Options numProj(Long numProj) {
+      this.numProj = numProj;
+      return this;
+    }
+    
+    /**
+     * @param isTraining 
+     */
+    public Options isTraining(Boolean isTraining) {
+      this.isTraining = isTraining;
+      return this;
+    }
+    
+    /**
+     * @param timeMajor 
+     */
+    public Options timeMajor(Boolean timeMajor) {
+      this.timeMajor = timeMajor;
+      return this;
+    }
+    
+    private String rnnMode;
+    private String inputMode;
+    private String direction;
+    private Float dropout;
+    private Long seed;
+    private Long seed2;
+    private Long numProj;
+    private Boolean isTraining;
+    private Boolean timeMajor;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new CudnnRNN operation.
+   * 
+   * @param scope current scope
+   * @param input 
+   * @param inputH 
+   * @param inputC 
+   * @param params 
+   * @param sequenceLengths 
+   * @param options carries optional attributes values
+   * @return a new instance of CudnnRNN
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> CudnnRNN<T> create(Scope scope, Operand<T> input, Operand<T> inputH, Operand<T> inputC, Operand<T> params, Operand<TInt32> sequenceLengths, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("CudnnRNNV3", scope.makeOpName("CudnnRNN"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(inputH.asOutput());
+    opBuilder.addInput(inputC.asOutput());
+    opBuilder.addInput(params.asOutput());
+    opBuilder.addInput(sequenceLengths.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.rnnMode != null) {
+          opBuilder.setAttr("rnn_mode", opts.rnnMode);
+        }
+        if (opts.inputMode != null) {
+          opBuilder.setAttr("input_mode", opts.inputMode);
+        }
+        if (opts.direction != null) {
+          opBuilder.setAttr("direction", opts.direction);
+        }
+        if (opts.dropout != null) {
+          opBuilder.setAttr("dropout", opts.dropout);
+        }
+        if (opts.seed != null) {
+          opBuilder.setAttr("seed", opts.seed);
+        }
+        if (opts.seed2 != null) {
+          opBuilder.setAttr("seed2", opts.seed2);
+        }
+        if (opts.numProj != null) {
+          opBuilder.setAttr("num_proj", opts.numProj);
+        }
+        if (opts.isTraining != null) {
+          opBuilder.setAttr("is_training", opts.isTraining);
+        }
+        if (opts.timeMajor != null) {
+          opBuilder.setAttr("time_major", opts.timeMajor);
+        }
+      }
+    }
+    return new CudnnRNN<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param rnnMode 
+   */
+  public static Options rnnMode(String rnnMode) {
+    return new Options().rnnMode(rnnMode);
+  }
+  
+  /**
+   * @param inputMode 
+   */
+  public static Options inputMode(String inputMode) {
+    return new Options().inputMode(inputMode);
+  }
+  
+  /**
+   * @param direction 
+   */
+  public static Options direction(String direction) {
+    return new Options().direction(direction);
+  }
+  
+  /**
+   * @param dropout 
+   */
+  public static Options dropout(Float dropout) {
+    return new Options().dropout(dropout);
+  }
+  
+  /**
+   * @param seed 
+   */
+  public static Options seed(Long seed) {
+    return new Options().seed(seed);
+  }
+  
+  /**
+   * @param seed2 
+   */
+  public static Options seed2(Long seed2) {
+    return new Options().seed2(seed2);
+  }
+  
+  /**
+   * @param numProj 
+   */
+  public static Options numProj(Long numProj) {
+    return new Options().numProj(numProj);
+  }
+  
+  /**
+   * @param isTraining 
+   */
+  public static Options isTraining(Boolean isTraining) {
+    return new Options().isTraining(isTraining);
+  }
+  
+  /**
+   * @param timeMajor 
+   */
+  public static Options timeMajor(Boolean timeMajor) {
+    return new Options().timeMajor(timeMajor);
+  }
+  
+  /**
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  /**
+   */
+  public Output<T> outputH() {
+    return outputH;
+  }
+  
+  /**
+   */
+  public Output<T> outputC() {
+    return outputC;
+  }
+  
+  /**
+   */
+  public Output<T> reserveSpace() {
+    return reserveSpace;
+  }
+  
+  /**
+   */
+  public Output<?> hostReserved() {
+    return hostReserved;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "CudnnRNNV3";
+  
+  private Output<T> output;
+  private Output<T> outputH;
+  private Output<T> outputC;
+  private Output<T> reserveSpace;
+  private Output<?> hostReserved;
+  
+  private CudnnRNN(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+    outputH = operation.output(outputIdx++);
+    outputC = operation.output(outputIdx++);
+    reserveSpace = operation.output(outputIdx++);
+    hostReserved = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CudnnRNNBackprop.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CudnnRNNBackprop.java
new file mode 100644
index 00000000000..9e51b12fee7
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CudnnRNNBackprop.java
@@ -0,0 +1,332 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.nn;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Backprop step of CudnnRNNV3.
+ * <p>
+ * Compute the backprop of both data and weights in a RNN. Takes an extra
+ *     "sequence_lengths" input than CudnnRNNBackprop.
+ * <p>
+ * rnn_mode: Indicates the type of the RNN model.
+ * input_mode: Indicates whether there is a linear projection between the input and
+ *     the actual computation before the first layer. 'skip_input' is only allowed
+ *     when input_size == num_units; 'auto_select' implies 'skip_input' when
+ *     input_size == num_units; otherwise, it implies 'linear_input'.
+ * direction: Indicates whether a bidirectional model will be used. Should be
+ *   "unidirectional" or "bidirectional".
+ * dropout: Dropout probability. When set to 0., dropout is disabled.
+ * seed: The 1st part of a seed to initialize dropout.
+ * seed2: The 2nd part of a seed to initialize dropout.
+ * input: If time_major is true, this is a 3-D tensor with the shape of
+ *     [seq_length, batch_size, input_size]. If time_major is false, the shape is
+ *     [batch_size, seq_length, input_size].
+ * input_h: If time_major is true, this is a 3-D tensor with the shape of
+ *     [num_layer * dir, batch_size, num_units]. If time_major is false, the shape
+ *     is [batch_size, num_layer * dir, num_units].
+ * input_c: For LSTM, a 3-D tensor with the shape of
+ *     [num_layer * dir, batch, num_units]. For other models, it is ignored.
+ * params: A 1-D tensor that contains the weights and biases in an opaque layout.
+ *     The size must be created through CudnnRNNParamsSize, and initialized
+ *     separately. Note that they might not be compatible across different
+ *     generations. So it is a good idea to save and restore
+ * sequence_lengths: a vector of lengths of each input sequence.
+ * output: If time_major is true, this is a 3-D tensor with the shape of
+ *     [seq_length, batch_size, dir * num_units]. If time_major is false, the
+ *     shape is [batch_size, seq_length, dir * num_units].
+ * output_h: The same shape has input_h.
+ * output_c: The same shape as input_c for LSTM. An empty tensor for other models.
+ * output_backprop: A 3-D tensor with the same shape as output in the forward pass.
+ * output_h_backprop: A 3-D tensor with the same shape as output_h in the forward
+ *     pass.
+ * output_c_backprop: A 3-D tensor with the same shape as output_c in the forward
+ *     pass.
+ * time_major: Indicates whether the input/output format is time major or batch
+ *     major.
+ * reserve_space: The same reserve_space produced in the forward operation.
+ * input_backprop: The backprop to input in the forward pass. Has the same shape
+ *     as input.
+ * input_h_backprop: The backprop to input_h in the forward pass. Has the same
+ *     shape as input_h.
+ * input_c_backprop: The backprop to input_c in the forward pass. Has the same
+ *     shape as input_c.
+ * params_backprop: The backprop to the params buffer in the forward pass. Has the
+ *     same shape as params.
+ * 
+ * @param <T> data type for {@code inputBackprop()} output
+ */
+public final class CudnnRNNBackprop<T extends TNumber> extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.nn.CudnnRNNBackprop}
+   */
+  public static class Options {
+    
+    /**
+     * @param rnnMode 
+     */
+    public Options rnnMode(String rnnMode) {
+      this.rnnMode = rnnMode;
+      return this;
+    }
+    
+    /**
+     * @param inputMode 
+     */
+    public Options inputMode(String inputMode) {
+      this.inputMode = inputMode;
+      return this;
+    }
+    
+    /**
+     * @param direction 
+     */
+    public Options direction(String direction) {
+      this.direction = direction;
+      return this;
+    }
+    
+    /**
+     * @param dropout 
+     */
+    public Options dropout(Float dropout) {
+      this.dropout = dropout;
+      return this;
+    }
+    
+    /**
+     * @param seed 
+     */
+    public Options seed(Long seed) {
+      this.seed = seed;
+      return this;
+    }
+    
+    /**
+     * @param seed2 
+     */
+    public Options seed2(Long seed2) {
+      this.seed2 = seed2;
+      return this;
+    }
+    
+    /**
+     * @param numProj 
+     */
+    public Options numProj(Long numProj) {
+      this.numProj = numProj;
+      return this;
+    }
+    
+    /**
+     * @param timeMajor 
+     */
+    public Options timeMajor(Boolean timeMajor) {
+      this.timeMajor = timeMajor;
+      return this;
+    }
+    
+    private String rnnMode;
+    private String inputMode;
+    private String direction;
+    private Float dropout;
+    private Long seed;
+    private Long seed2;
+    private Long numProj;
+    private Boolean timeMajor;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new CudnnRNNBackprop operation.
+   * 
+   * @param scope current scope
+   * @param input 
+   * @param inputH 
+   * @param inputC 
+   * @param params 
+   * @param sequenceLengths 
+   * @param output 
+   * @param outputH 
+   * @param outputC 
+   * @param outputBackprop 
+   * @param outputHBackprop 
+   * @param outputCBackprop 
+   * @param reserveSpace 
+   * @param hostReserved 
+   * @param options carries optional attributes values
+   * @return a new instance of CudnnRNNBackprop
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> CudnnRNNBackprop<T> create(Scope scope, Operand<T> input, Operand<T> inputH, Operand<T> inputC, Operand<T> params, Operand<TInt32> sequenceLengths, Operand<T> output, Operand<T> outputH, Operand<T> outputC, Operand<T> outputBackprop, Operand<T> outputHBackprop, Operand<T> outputCBackprop, Operand<T> reserveSpace, Operand<?> hostReserved, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("CudnnRNNBackpropV3", scope.makeOpName("CudnnRNNBackprop"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(inputH.asOutput());
+    opBuilder.addInput(inputC.asOutput());
+    opBuilder.addInput(params.asOutput());
+    opBuilder.addInput(sequenceLengths.asOutput());
+    opBuilder.addInput(output.asOutput());
+    opBuilder.addInput(outputH.asOutput());
+    opBuilder.addInput(outputC.asOutput());
+    opBuilder.addInput(outputBackprop.asOutput());
+    opBuilder.addInput(outputHBackprop.asOutput());
+    opBuilder.addInput(outputCBackprop.asOutput());
+    opBuilder.addInput(reserveSpace.asOutput());
+    opBuilder.addInput(hostReserved.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.rnnMode != null) {
+          opBuilder.setAttr("rnn_mode", opts.rnnMode);
+        }
+        if (opts.inputMode != null) {
+          opBuilder.setAttr("input_mode", opts.inputMode);
+        }
+        if (opts.direction != null) {
+          opBuilder.setAttr("direction", opts.direction);
+        }
+        if (opts.dropout != null) {
+          opBuilder.setAttr("dropout", opts.dropout);
+        }
+        if (opts.seed != null) {
+          opBuilder.setAttr("seed", opts.seed);
+        }
+        if (opts.seed2 != null) {
+          opBuilder.setAttr("seed2", opts.seed2);
+        }
+        if (opts.numProj != null) {
+          opBuilder.setAttr("num_proj", opts.numProj);
+        }
+        if (opts.timeMajor != null) {
+          opBuilder.setAttr("time_major", opts.timeMajor);
+        }
+      }
+    }
+    return new CudnnRNNBackprop<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param rnnMode 
+   */
+  public static Options rnnMode(String rnnMode) {
+    return new Options().rnnMode(rnnMode);
+  }
+  
+  /**
+   * @param inputMode 
+   */
+  public static Options inputMode(String inputMode) {
+    return new Options().inputMode(inputMode);
+  }
+  
+  /**
+   * @param direction 
+   */
+  public static Options direction(String direction) {
+    return new Options().direction(direction);
+  }
+  
+  /**
+   * @param dropout 
+   */
+  public static Options dropout(Float dropout) {
+    return new Options().dropout(dropout);
+  }
+  
+  /**
+   * @param seed 
+   */
+  public static Options seed(Long seed) {
+    return new Options().seed(seed);
+  }
+  
+  /**
+   * @param seed2 
+   */
+  public static Options seed2(Long seed2) {
+    return new Options().seed2(seed2);
+  }
+  
+  /**
+   * @param numProj 
+   */
+  public static Options numProj(Long numProj) {
+    return new Options().numProj(numProj);
+  }
+  
+  /**
+   * @param timeMajor 
+   */
+  public static Options timeMajor(Boolean timeMajor) {
+    return new Options().timeMajor(timeMajor);
+  }
+  
+  /**
+   */
+  public Output<T> inputBackprop() {
+    return inputBackprop;
+  }
+  
+  /**
+   */
+  public Output<T> inputHBackprop() {
+    return inputHBackprop;
+  }
+  
+  /**
+   */
+  public Output<T> inputCBackprop() {
+    return inputCBackprop;
+  }
+  
+  /**
+   */
+  public Output<T> paramsBackprop() {
+    return paramsBackprop;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "CudnnRNNBackpropV3";
+  
+  private Output<T> inputBackprop;
+  private Output<T> inputHBackprop;
+  private Output<T> inputCBackprop;
+  private Output<T> paramsBackprop;
+  
+  private CudnnRNNBackprop(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    inputBackprop = operation.output(outputIdx++);
+    inputHBackprop = operation.output(outputIdx++);
+    inputCBackprop = operation.output(outputIdx++);
+    paramsBackprop = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CudnnRNNCanonicalToParams.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CudnnRNNCanonicalToParams.java
new file mode 100644
index 00000000000..38df41c251a
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CudnnRNNCanonicalToParams.java
@@ -0,0 +1,263 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.nn;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.Operands;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Converts CudnnRNN params from canonical form to usable form. It supports the projection in LSTM.
+ * <p>
+ * Writes a set of weights into the opaque params buffer so they can be used in
+ * upcoming training or inferences.
+ * <p>
+ * Note that the params buffer may not be compatible across different GPUs. So any
+ * save and restoration should be converted to and from the canonical weights and
+ * biases.
+ * <p>
+ * num_layers: Specifies the number of layers in the RNN model.
+ * num_units: Specifies the size of the hidden state.
+ * input_size: Specifies the size of the input state.
+ * weights: the canonical form of weights that can be used for saving
+ *     and restoration. They are more likely to be compatible across different
+ *     generations.
+ * biases: the canonical form of biases that can be used for saving
+ *     and restoration. They are more likely to be compatible across different
+ *     generations.
+ * num_params_weights: number of weight parameter matrix for all layers.
+ * num_params_biases: number of bias parameter vector for all layers.
+ * rnn_mode: Indicates the type of the RNN model.
+ * input_mode: Indicate whether there is a linear projection between the input and
+ *     The actual computation before the first layer. 'skip_input' is only allowed
+ *     when input_size == num_units; 'auto_select' implies 'skip_input' when
+ *     input_size == num_units; otherwise, it implies 'linear_input'.
+ * direction: Indicates whether a bidirectional model will be used.
+ *     dir = (direction == bidirectional) ? 2 : 1
+ * dropout: dropout probability. When set to 0., dropout is disabled.
+ * seed: the 1st part of a seed to initialize dropout.
+ * seed2: the 2nd part of a seed to initialize dropout.
+ * num_proj: The output dimensionality for the projection matrices. If None or 0,
+ *     no projection is performed.
+ * 
+ * @param <T> data type for {@code params()} output
+ */
+@Operator(group = "nn")
+public final class CudnnRNNCanonicalToParams<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.nn.CudnnRNNCanonicalToParams}
+   */
+  public static class Options {
+    
+    /**
+     * @param rnnMode 
+     */
+    public Options rnnMode(String rnnMode) {
+      this.rnnMode = rnnMode;
+      return this;
+    }
+    
+    /**
+     * @param inputMode 
+     */
+    public Options inputMode(String inputMode) {
+      this.inputMode = inputMode;
+      return this;
+    }
+    
+    /**
+     * @param direction 
+     */
+    public Options direction(String direction) {
+      this.direction = direction;
+      return this;
+    }
+    
+    /**
+     * @param dropout 
+     */
+    public Options dropout(Float dropout) {
+      this.dropout = dropout;
+      return this;
+    }
+    
+    /**
+     * @param seed 
+     */
+    public Options seed(Long seed) {
+      this.seed = seed;
+      return this;
+    }
+    
+    /**
+     * @param seed2 
+     */
+    public Options seed2(Long seed2) {
+      this.seed2 = seed2;
+      return this;
+    }
+    
+    /**
+     * @param numProj 
+     */
+    public Options numProj(Long numProj) {
+      this.numProj = numProj;
+      return this;
+    }
+    
+    private String rnnMode;
+    private String inputMode;
+    private String direction;
+    private Float dropout;
+    private Long seed;
+    private Long seed2;
+    private Long numProj;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new CudnnRNNCanonicalToParams operation.
+   * 
+   * @param scope current scope
+   * @param numLayers 
+   * @param numUnits 
+   * @param inputSize 
+   * @param weights 
+   * @param biases 
+   * @param options carries optional attributes values
+   * @return a new instance of CudnnRNNCanonicalToParams
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> CudnnRNNCanonicalToParams<T> create(Scope scope, Operand<TInt32> numLayers, Operand<TInt32> numUnits, Operand<TInt32> inputSize, Iterable<Operand<T>> weights, Iterable<Operand<T>> biases, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("CudnnRNNCanonicalToParamsV2", scope.makeOpName("CudnnRNNCanonicalToParams"));
+    opBuilder.addInput(numLayers.asOutput());
+    opBuilder.addInput(numUnits.asOutput());
+    opBuilder.addInput(inputSize.asOutput());
+    opBuilder.addInputList(Operands.asOutputs(weights));
+    opBuilder.addInputList(Operands.asOutputs(biases));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.rnnMode != null) {
+          opBuilder.setAttr("rnn_mode", opts.rnnMode);
+        }
+        if (opts.inputMode != null) {
+          opBuilder.setAttr("input_mode", opts.inputMode);
+        }
+        if (opts.direction != null) {
+          opBuilder.setAttr("direction", opts.direction);
+        }
+        if (opts.dropout != null) {
+          opBuilder.setAttr("dropout", opts.dropout);
+        }
+        if (opts.seed != null) {
+          opBuilder.setAttr("seed", opts.seed);
+        }
+        if (opts.seed2 != null) {
+          opBuilder.setAttr("seed2", opts.seed2);
+        }
+        if (opts.numProj != null) {
+          opBuilder.setAttr("num_proj", opts.numProj);
+        }
+      }
+    }
+    return new CudnnRNNCanonicalToParams<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param rnnMode 
+   */
+  public static Options rnnMode(String rnnMode) {
+    return new Options().rnnMode(rnnMode);
+  }
+  
+  /**
+   * @param inputMode 
+   */
+  public static Options inputMode(String inputMode) {
+    return new Options().inputMode(inputMode);
+  }
+  
+  /**
+   * @param direction 
+   */
+  public static Options direction(String direction) {
+    return new Options().direction(direction);
+  }
+  
+  /**
+   * @param dropout 
+   */
+  public static Options dropout(Float dropout) {
+    return new Options().dropout(dropout);
+  }
+  
+  /**
+   * @param seed 
+   */
+  public static Options seed(Long seed) {
+    return new Options().seed(seed);
+  }
+  
+  /**
+   * @param seed2 
+   */
+  public static Options seed2(Long seed2) {
+    return new Options().seed2(seed2);
+  }
+  
+  /**
+   * @param numProj 
+   */
+  public static Options numProj(Long numProj) {
+    return new Options().numProj(numProj);
+  }
+  
+  /**
+   */
+  public Output<T> params() {
+    return params;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return params;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "CudnnRNNCanonicalToParamsV2";
+  
+  private Output<T> params;
+  
+  private CudnnRNNCanonicalToParams(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    params = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CudnnRNNParamsToCanonical.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CudnnRNNParamsToCanonical.java
new file mode 100644
index 00000000000..a6fcda3786e
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CudnnRNNParamsToCanonical.java
@@ -0,0 +1,274 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.nn;
+
+import java.util.Arrays;
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Retrieves CudnnRNN params in canonical form. It supports the projection in LSTM.
+ * <p>
+ * Retrieves a set of weights from the opaque params buffer that can be saved and
+ * restored in a way compatible with future runs.
+ * <p>
+ * Note that the params buffer may not be compatible across different GPUs. So any
+ * save and restoration should be converted to and from the canonical weights and
+ * biases.
+ * <p>
+ * num_layers: Specifies the number of layers in the RNN model.
+ * num_units: Specifies the size of the hidden state.
+ * input_size: Specifies the size of the input state.
+ * num_params_weights: number of weight parameter matrix for all layers.
+ * num_params_biases: number of bias parameter vector for all layers.
+ * weights: the canonical form of weights that can be used for saving
+ *     and restoration. They are more likely to be compatible across different
+ *     generations.
+ * biases: the canonical form of biases that can be used for saving
+ *     and restoration. They are more likely to be compatible across different
+ *     generations.
+ * rnn_mode: Indicates the type of the RNN model.
+ * input_mode: Indicate whether there is a linear projection between the input and
+ *     The actual computation before the first layer. 'skip_input' is only allowed
+ *     when input_size == num_units; 'auto_select' implies 'skip_input' when
+ *     input_size == num_units; otherwise, it implies 'linear_input'.
+ * direction: Indicates whether a bidirectional model will be used.
+ *     dir = (direction == bidirectional) ? 2 : 1
+ * dropout: dropout probability. When set to 0., dropout is disabled.
+ * seed: the 1st part of a seed to initialize dropout.
+ * seed2: the 2nd part of a seed to initialize dropout.
+ * num_proj: The output dimensionality for the projection matrices. If None or 0,
+ *     no projection is performed.
+ * 
+ * @param <T> data type for {@code weights()} output
+ */
+@Operator(group = "nn")
+public final class CudnnRNNParamsToCanonical<T extends TNumber> extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.nn.CudnnRNNParamsToCanonical}
+   */
+  public static class Options {
+    
+    /**
+     * @param rnnMode 
+     */
+    public Options rnnMode(String rnnMode) {
+      this.rnnMode = rnnMode;
+      return this;
+    }
+    
+    /**
+     * @param inputMode 
+     */
+    public Options inputMode(String inputMode) {
+      this.inputMode = inputMode;
+      return this;
+    }
+    
+    /**
+     * @param direction 
+     */
+    public Options direction(String direction) {
+      this.direction = direction;
+      return this;
+    }
+    
+    /**
+     * @param dropout 
+     */
+    public Options dropout(Float dropout) {
+      this.dropout = dropout;
+      return this;
+    }
+    
+    /**
+     * @param seed 
+     */
+    public Options seed(Long seed) {
+      this.seed = seed;
+      return this;
+    }
+    
+    /**
+     * @param seed2 
+     */
+    public Options seed2(Long seed2) {
+      this.seed2 = seed2;
+      return this;
+    }
+    
+    /**
+     * @param numProj 
+     */
+    public Options numProj(Long numProj) {
+      this.numProj = numProj;
+      return this;
+    }
+    
+    private String rnnMode;
+    private String inputMode;
+    private String direction;
+    private Float dropout;
+    private Long seed;
+    private Long seed2;
+    private Long numProj;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new CudnnRNNParamsToCanonical operation.
+   * 
+   * @param scope current scope
+   * @param numLayers 
+   * @param numUnits 
+   * @param inputSize 
+   * @param params 
+   * @param numParamsWeights 
+   * @param numParamsBiases 
+   * @param options carries optional attributes values
+   * @return a new instance of CudnnRNNParamsToCanonical
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> CudnnRNNParamsToCanonical<T> create(Scope scope, Operand<TInt32> numLayers, Operand<TInt32> numUnits, Operand<TInt32> inputSize, Operand<T> params, Long numParamsWeights, Long numParamsBiases, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("CudnnRNNParamsToCanonicalV2", scope.makeOpName("CudnnRNNParamsToCanonical"));
+    opBuilder.addInput(numLayers.asOutput());
+    opBuilder.addInput(numUnits.asOutput());
+    opBuilder.addInput(inputSize.asOutput());
+    opBuilder.addInput(params.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("num_params_weights", numParamsWeights);
+    opBuilder.setAttr("num_params_biases", numParamsBiases);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.rnnMode != null) {
+          opBuilder.setAttr("rnn_mode", opts.rnnMode);
+        }
+        if (opts.inputMode != null) {
+          opBuilder.setAttr("input_mode", opts.inputMode);
+        }
+        if (opts.direction != null) {
+          opBuilder.setAttr("direction", opts.direction);
+        }
+        if (opts.dropout != null) {
+          opBuilder.setAttr("dropout", opts.dropout);
+        }
+        if (opts.seed != null) {
+          opBuilder.setAttr("seed", opts.seed);
+        }
+        if (opts.seed2 != null) {
+          opBuilder.setAttr("seed2", opts.seed2);
+        }
+        if (opts.numProj != null) {
+          opBuilder.setAttr("num_proj", opts.numProj);
+        }
+      }
+    }
+    return new CudnnRNNParamsToCanonical<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param rnnMode 
+   */
+  public static Options rnnMode(String rnnMode) {
+    return new Options().rnnMode(rnnMode);
+  }
+  
+  /**
+   * @param inputMode 
+   */
+  public static Options inputMode(String inputMode) {
+    return new Options().inputMode(inputMode);
+  }
+  
+  /**
+   * @param direction 
+   */
+  public static Options direction(String direction) {
+    return new Options().direction(direction);
+  }
+  
+  /**
+   * @param dropout 
+   */
+  public static Options dropout(Float dropout) {
+    return new Options().dropout(dropout);
+  }
+  
+  /**
+   * @param seed 
+   */
+  public static Options seed(Long seed) {
+    return new Options().seed(seed);
+  }
+  
+  /**
+   * @param seed2 
+   */
+  public static Options seed2(Long seed2) {
+    return new Options().seed2(seed2);
+  }
+  
+  /**
+   * @param numProj 
+   */
+  public static Options numProj(Long numProj) {
+    return new Options().numProj(numProj);
+  }
+  
+  /**
+   */
+  public List<Output<T>> weights() {
+    return weights;
+  }
+  
+  /**
+   */
+  public List<Output<T>> biases() {
+    return biases;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "CudnnRNNParamsToCanonicalV2";
+  
+  private List<Output<T>> weights;
+  private List<Output<T>> biases;
+  
+  @SuppressWarnings("unchecked")
+  private CudnnRNNParamsToCanonical(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    int weightsLength = operation.outputListLength("weights");
+    weights = Arrays.asList((Output<T>[])operation.outputList(outputIdx, weightsLength));
+    outputIdx += weightsLength;
+    int biasesLength = operation.outputListLength("biases");
+    biases = Arrays.asList((Output<T>[])operation.outputList(outputIdx, biasesLength));
+    outputIdx += biasesLength;
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CudnnRnnParamsSize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CudnnRnnParamsSize.java
new file mode 100644
index 00000000000..dd3f4ba6392
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/CudnnRnnParamsSize.java
@@ -0,0 +1,253 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.nn;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Computes size of weights that can be used by a Cudnn RNN model.
+ * <p>
+ * Return the params size that can be used by the Cudnn RNN model. Subsequent
+ * weight allocation and initialization should use this size.
+ * <p>
+ * num_layers: Specifies the number of layers in the RNN model.
+ * num_units: Specifies the size of the hidden state.
+ * input_size: Specifies the size of the input state.
+ * rnn_mode: Indicates the type of the RNN model.
+ * input_mode: Indicate whether there is a linear projection between the input and
+ *   The actual computation before the first layer. 'skip_input' is only allowed
+ *   when input_size == num_units; 'auto_select' implies 'skip_input' when
+ *   input_size == num_units; otherwise, it implies 'linear_input'.
+ * direction: Indicates whether a bidirectional model will be used.
+ *   dir = (direction == bidirectional) ? 2 : 1
+ * dropout: dropout probability. When set to 0., dropout is disabled.
+ * seed: the 1st part of a seed to initialize dropout.
+ * seed2: the 2nd part of a seed to initialize dropout.
+ * params_size: The size of the params buffer that should be allocated and
+ *   initialized for this RNN model. Note that this params buffer may not be
+ *   compatible across GPUs. Please use CudnnRNNParamsWeights and
+ *   CudnnRNNParamsBiases to save and restore them in a way that is compatible
+ *   across different runs.
+ * 
+ * @param <U> data type for {@code paramsSize()} output
+ */
+@Operator(group = "nn")
+public final class CudnnRnnParamsSize<U extends TNumber> extends RawOp implements Operand<U> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.nn.CudnnRnnParamsSize}
+   */
+  public static class Options {
+    
+    /**
+     * @param rnnMode 
+     */
+    public Options rnnMode(String rnnMode) {
+      this.rnnMode = rnnMode;
+      return this;
+    }
+    
+    /**
+     * @param inputMode 
+     */
+    public Options inputMode(String inputMode) {
+      this.inputMode = inputMode;
+      return this;
+    }
+    
+    /**
+     * @param direction 
+     */
+    public Options direction(String direction) {
+      this.direction = direction;
+      return this;
+    }
+    
+    /**
+     * @param dropout 
+     */
+    public Options dropout(Float dropout) {
+      this.dropout = dropout;
+      return this;
+    }
+    
+    /**
+     * @param seed 
+     */
+    public Options seed(Long seed) {
+      this.seed = seed;
+      return this;
+    }
+    
+    /**
+     * @param seed2 
+     */
+    public Options seed2(Long seed2) {
+      this.seed2 = seed2;
+      return this;
+    }
+    
+    /**
+     * @param numProj 
+     */
+    public Options numProj(Long numProj) {
+      this.numProj = numProj;
+      return this;
+    }
+    
+    private String rnnMode;
+    private String inputMode;
+    private String direction;
+    private Float dropout;
+    private Long seed;
+    private Long seed2;
+    private Long numProj;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new CudnnRnnParamsSize operation.
+   * 
+   * @param scope current scope
+   * @param numLayers 
+   * @param numUnits 
+   * @param inputSize 
+   * @param T 
+   * @param S 
+   * @param options carries optional attributes values
+   * @return a new instance of CudnnRnnParamsSize
+   */
+  @Endpoint(describeByClass = true)
+  public static <U extends TNumber, T extends TNumber> CudnnRnnParamsSize<U> create(Scope scope, Operand<TInt32> numLayers, Operand<TInt32> numUnits, Operand<TInt32> inputSize, Class<T> T, Class<U> S, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("CudnnRNNParamsSize", scope.makeOpName("CudnnRnnParamsSize"));
+    opBuilder.addInput(numLayers.asOutput());
+    opBuilder.addInput(numUnits.asOutput());
+    opBuilder.addInput(inputSize.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("T", T);
+    opBuilder.setAttr("S", S);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.rnnMode != null) {
+          opBuilder.setAttr("rnn_mode", opts.rnnMode);
+        }
+        if (opts.inputMode != null) {
+          opBuilder.setAttr("input_mode", opts.inputMode);
+        }
+        if (opts.direction != null) {
+          opBuilder.setAttr("direction", opts.direction);
+        }
+        if (opts.dropout != null) {
+          opBuilder.setAttr("dropout", opts.dropout);
+        }
+        if (opts.seed != null) {
+          opBuilder.setAttr("seed", opts.seed);
+        }
+        if (opts.seed2 != null) {
+          opBuilder.setAttr("seed2", opts.seed2);
+        }
+        if (opts.numProj != null) {
+          opBuilder.setAttr("num_proj", opts.numProj);
+        }
+      }
+    }
+    return new CudnnRnnParamsSize<U>(opBuilder.build());
+  }
+  
+  /**
+   * @param rnnMode 
+   */
+  public static Options rnnMode(String rnnMode) {
+    return new Options().rnnMode(rnnMode);
+  }
+  
+  /**
+   * @param inputMode 
+   */
+  public static Options inputMode(String inputMode) {
+    return new Options().inputMode(inputMode);
+  }
+  
+  /**
+   * @param direction 
+   */
+  public static Options direction(String direction) {
+    return new Options().direction(direction);
+  }
+  
+  /**
+   * @param dropout 
+   */
+  public static Options dropout(Float dropout) {
+    return new Options().dropout(dropout);
+  }
+  
+  /**
+   * @param seed 
+   */
+  public static Options seed(Long seed) {
+    return new Options().seed(seed);
+  }
+  
+  /**
+   * @param seed2 
+   */
+  public static Options seed2(Long seed2) {
+    return new Options().seed2(seed2);
+  }
+  
+  /**
+   * @param numProj 
+   */
+  public static Options numProj(Long numProj) {
+    return new Options().numProj(numProj);
+  }
+  
+  /**
+   */
+  public Output<U> paramsSize() {
+    return paramsSize;
+  }
+  
+  @Override
+  public Output<U> asOutput() {
+    return paramsSize;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "CudnnRNNParamsSize";
+  
+  private Output<U> paramsSize;
+  
+  private CudnnRnnParamsSize(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    paramsSize = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/DataFormatDimMap.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/DataFormatDimMap.java
new file mode 100644
index 00000000000..3eb8a46d310
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/DataFormatDimMap.java
@@ -0,0 +1,131 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.nn;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Returns the dimension index in the destination data format given the one in
+ * <p>
+ * the source data format.
+ * 
+ * @param <T> data type for {@code y()} output
+ */
+@Operator(group = "nn")
+public final class DataFormatDimMap<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.nn.DataFormatDimMap}
+   */
+  public static class Options {
+    
+    /**
+     * @param srcFormat source data format.
+     */
+    public Options srcFormat(String srcFormat) {
+      this.srcFormat = srcFormat;
+      return this;
+    }
+    
+    /**
+     * @param dstFormat destination data format.
+     */
+    public Options dstFormat(String dstFormat) {
+      this.dstFormat = dstFormat;
+      return this;
+    }
+    
+    private String srcFormat;
+    private String dstFormat;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new DataFormatDimMap operation.
+   * 
+   * @param scope current scope
+   * @param x A Tensor with each element as a dimension index in source data format.
+   * Must be in the range [-4, 4).
+   * @param options carries optional attributes values
+   * @return a new instance of DataFormatDimMap
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> DataFormatDimMap<T> create(Scope scope, Operand<T> x, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("DataFormatDimMap", scope.makeOpName("DataFormatDimMap"));
+    opBuilder.addInput(x.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.srcFormat != null) {
+          opBuilder.setAttr("src_format", opts.srcFormat);
+        }
+        if (opts.dstFormat != null) {
+          opBuilder.setAttr("dst_format", opts.dstFormat);
+        }
+      }
+    }
+    return new DataFormatDimMap<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param srcFormat source data format.
+   */
+  public static Options srcFormat(String srcFormat) {
+    return new Options().srcFormat(srcFormat);
+  }
+  
+  /**
+   * @param dstFormat destination data format.
+   */
+  public static Options dstFormat(String dstFormat) {
+    return new Options().dstFormat(dstFormat);
+  }
+  
+  /**
+   * A Tensor with each element as a dimension index in destination data format.
+   */
+  public Output<T> y() {
+    return y;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return y;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "DataFormatDimMap";
+  
+  private Output<T> y;
+  
+  private DataFormatDimMap(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    y = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/DataFormatVecPermute.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/DataFormatVecPermute.java
new file mode 100644
index 00000000000..0ba1ddeb4ba
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/DataFormatVecPermute.java
@@ -0,0 +1,130 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.nn;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Returns the permuted vector/tensor in the destination data format given the
+ * <p>
+ * one in the source data format.
+ * 
+ * @param <T> data type for {@code y()} output
+ */
+@Operator(group = "nn")
+public final class DataFormatVecPermute<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.nn.DataFormatVecPermute}
+   */
+  public static class Options {
+    
+    /**
+     * @param srcFormat source data format.
+     */
+    public Options srcFormat(String srcFormat) {
+      this.srcFormat = srcFormat;
+      return this;
+    }
+    
+    /**
+     * @param dstFormat destination data format.
+     */
+    public Options dstFormat(String dstFormat) {
+      this.dstFormat = dstFormat;
+      return this;
+    }
+    
+    private String srcFormat;
+    private String dstFormat;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new DataFormatVecPermute operation.
+   * 
+   * @param scope current scope
+   * @param x Vector of size 4 or Tensor of shape (4, 2) in source data format.
+   * @param options carries optional attributes values
+   * @return a new instance of DataFormatVecPermute
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> DataFormatVecPermute<T> create(Scope scope, Operand<T> x, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("DataFormatVecPermute", scope.makeOpName("DataFormatVecPermute"));
+    opBuilder.addInput(x.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.srcFormat != null) {
+          opBuilder.setAttr("src_format", opts.srcFormat);
+        }
+        if (opts.dstFormat != null) {
+          opBuilder.setAttr("dst_format", opts.dstFormat);
+        }
+      }
+    }
+    return new DataFormatVecPermute<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param srcFormat source data format.
+   */
+  public static Options srcFormat(String srcFormat) {
+    return new Options().srcFormat(srcFormat);
+  }
+  
+  /**
+   * @param dstFormat destination data format.
+   */
+  public static Options dstFormat(String dstFormat) {
+    return new Options().dstFormat(dstFormat);
+  }
+  
+  /**
+   * Vector of size 4 or Tensor of shape (4, 2) in destination data format.
+   */
+  public Output<T> y() {
+    return y;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return y;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "DataFormatVecPermute";
+  
+  private Output<T> y;
+  
+  private DataFormatVecPermute(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    y = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/DepthToSpace.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/DepthToSpace.java
new file mode 100644
index 00000000000..6456eff1f95
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/DepthToSpace.java
@@ -0,0 +1,190 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.nn;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * DepthToSpace for tensors of type T.
+ * <p>
+ * Rearranges data from depth into blocks of spatial data.
+ * This is the reverse transformation of SpaceToDepth. More specifically,
+ * this op outputs a copy of the input tensor where values from the `depth`
+ * dimension are moved in spatial blocks to the `height` and `width` dimensions.
+ * The attr `block_size` indicates the input block size and how the data is moved.
+ * <p>
+ *   * Chunks of data of size `block_size * block_size` from depth are rearranged
+ *     into non-overlapping blocks of size `block_size x block_size`
+ *   * The width the output tensor is `input_depth * block_size`, whereas the
+ *     height is `input_height * block_size`.
+ *   * The Y, X coordinates within each block of the output image are determined
+ *     by the high order component of the input channel index.
+ *   * The depth of the input tensor must be divisible by
+ *     `block_size * block_size`.
+ * <p>
+ * The `data_format` attr specifies the layout of the input and output tensors
+ * with the following options:
+ *   "NHWC": `[ batch, height, width, channels ]`
+ *   "NCHW": `[ batch, channels, height, width ]`
+ *   "NCHW_VECT_C":
+ *       `qint8 [ batch, channels / 4, height, width, 4 ]`
+ * <p>
+ * It is useful to consider the operation as transforming a 6-D Tensor.
+ * e.g. for data_format = NHWC,
+ *      Each element in the input tensor can be specified via 6 coordinates,
+ *      ordered by decreasing memory layout significance as:
+ *      n,iY,iX,bY,bX,oC  (where n=batch index, iX, iY means X or Y coordinates
+ *                         within the input image, bX, bY means coordinates
+ *                         within the output block, oC means output channels).
+ *      The output would be the input transposed to the following layout:
+ *      n,iY,bY,iX,bX,oC
+ * <p>
+ * This operation is useful for resizing the activations between convolutions
+ * (but keeping all data), e.g. instead of pooling. It is also useful for training
+ * purely convolutional models.
+ * <p>
+ * For example, given an input of shape `[1, 1, 1, 4]`, data_format = "NHWC" and
+ * block_size = 2:
+ * <pre>{@code
+ * x = [[[[1, 2, 3, 4]]]]
+ * 
+ * }</pre>
+ * This operation will output a tensor of shape `[1, 2, 2, 1]`:
+ * <pre>{@code
+ *    [[[[1], [2]],
+ *      [[3], [4]]]]
+ * }</pre>
+ * Here, the input has a batch of 1 and each batch element has shape `[1, 1, 4]`,
+ * the corresponding output will have 2x2 elements and will have a depth of
+ * 1 channel (1 = `4 / (block_size * block_size)`).
+ * The output element shape is `[2, 2, 1]`.
+ * <p>
+ * For an input tensor with larger depth, here of shape `[1, 1, 1, 12]`, e.g.
+ * <pre>{@code
+ * x = [[[[1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12]]]]
+ * }</pre>
+ * This operation, for block size of 2, will return the following tensor of shape
+ * `[1, 2, 2, 3]`
+ * <pre>{@code
+ *    [[[[1, 2, 3], [4, 5, 6]],
+ *      [[7, 8, 9], [10, 11, 12]]]]
+ * 
+ * }</pre>
+ * Similarly, for the following input of shape `[1 2 2 4]`, and a block size of 2:
+ * <pre>{@code
+ * x =  [[[[1, 2, 3, 4],
+ *        [5, 6, 7, 8]],
+ *       [[9, 10, 11, 12],
+ *        [13, 14, 15, 16]]]]
+ * }</pre>
+ * the operator will return the following tensor of shape `[1 4 4 1]`:
+ * <pre>{@code
+ * x = [[[ [1],   [2],  [5],  [6]],
+ *       [ [3],   [4],  [7],  [8]],
+ *       [ [9],  [10], [13],  [14]],
+ *       [ [11], [12], [15],  [16]]]]
+ * 
+ * }</pre>
+ * 
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator(group = "nn")
+public final class DepthToSpace<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.nn.DepthToSpace}
+   */
+  public static class Options {
+    
+    /**
+     * @param dataFormat 
+     */
+    public Options dataFormat(String dataFormat) {
+      this.dataFormat = dataFormat;
+      return this;
+    }
+    
+    private String dataFormat;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new DepthToSpace operation.
+   * 
+   * @param scope current scope
+   * @param input 
+   * @param blockSize The size of the spatial block, same as in Space2Depth.
+   * @param options carries optional attributes values
+   * @return a new instance of DepthToSpace
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> DepthToSpace<T> create(Scope scope, Operand<T> input, Long blockSize, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("DepthToSpace", scope.makeOpName("DepthToSpace"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("block_size", blockSize);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.dataFormat != null) {
+          opBuilder.setAttr("data_format", opts.dataFormat);
+        }
+      }
+    }
+    return new DepthToSpace<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param dataFormat 
+   */
+  public static Options dataFormat(String dataFormat) {
+    return new Options().dataFormat(dataFormat);
+  }
+  
+  /**
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "DepthToSpace";
+  
+  private Output<T> output;
+  
+  private DepthToSpace(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/DepthwiseConv2dNative.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/DepthwiseConv2dNative.java
new file mode 100644
index 00000000000..2aff519c733
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/DepthwiseConv2dNative.java
@@ -0,0 +1,199 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.nn;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Computes a 2-D depthwise convolution given 4-D `input` and `filter` tensors.
+ * <p>
+ * Given an input tensor of shape `[batch, in_height, in_width, in_channels]`
+ * and a filter / kernel tensor of shape
+ * `[filter_height, filter_width, in_channels, channel_multiplier]`, containing
+ * `in_channels` convolutional filters of depth 1, `depthwise_conv2d` applies
+ * a different filter to each input channel (expanding from 1 channel to
+ * `channel_multiplier` channels for each), then concatenates the results
+ * together. Thus, the output has `in_channels * channel_multiplier` channels.
+ * <pre>{@code
+ * for k in 0..in_channels-1
+ *   for q in 0..channel_multiplier-1
+ *     output[b, i, j, k * channel_multiplier + q] =
+ *       sum_{di, dj} input[b, strides[1] * i + di, strides[2] * j + dj, k] *
+ *                         filter[di, dj, k, q]
+ * }</pre>
+ * Must have `strides[0] = strides[3] = 1`.  For the most common case of the same
+ * horizontal and vertices strides, `strides = [1, stride, stride, 1]`.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator(group = "nn")
+public final class DepthwiseConv2dNative<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.nn.DepthwiseConv2dNative}
+   */
+  public static class Options {
+    
+    /**
+     * @param explicitPaddings 
+     */
+    public Options explicitPaddings(List<Long> explicitPaddings) {
+      this.explicitPaddings = explicitPaddings;
+      return this;
+    }
+    
+    /**
+     * @param dataFormat Specify the data format of the input and output data. With the
+     * default format "NHWC", the data is stored in the order of:
+     *     [batch, height, width, channels].
+     * Alternatively, the format could be "NCHW", the data storage order of:
+     *     [batch, channels, height, width].
+     */
+    public Options dataFormat(String dataFormat) {
+      this.dataFormat = dataFormat;
+      return this;
+    }
+    
+    /**
+     * @param dilations 1-D tensor of length 4.  The dilation factor for each dimension of
+     * `input`. If set to k > 1, there will be k-1 skipped cells between each filter
+     * element on that dimension. The dimension order is determined by the value of
+     * `data_format`, see above for details. Dilations in the batch and depth
+     * dimensions must be 1.
+     */
+    public Options dilations(List<Long> dilations) {
+      this.dilations = dilations;
+      return this;
+    }
+    
+    private List<Long> explicitPaddings;
+    private String dataFormat;
+    private List<Long> dilations;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new DepthwiseConv2dNative operation.
+   * 
+   * @param scope current scope
+   * @param input 
+   * @param filter 
+   * @param strides 1-D of length 4.  The stride of the sliding window for each dimension
+   * of `input`.
+   * @param padding The type of padding algorithm to use.
+   * @param options carries optional attributes values
+   * @return a new instance of DepthwiseConv2dNative
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> DepthwiseConv2dNative<T> create(Scope scope, Operand<T> input, Operand<T> filter, List<Long> strides, String padding, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("DepthwiseConv2dNative", scope.makeOpName("DepthwiseConv2dNative"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(filter.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    long[] stridesArray = new long[strides.size()];
+    for (int i = 0; i < stridesArray.length; ++i) {
+      stridesArray[i] = strides.get(i);
+    }
+    opBuilder.setAttr("strides", stridesArray);
+    opBuilder.setAttr("padding", padding);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.explicitPaddings != null) {
+          long[] explicitPaddingsArray = new long[opts.explicitPaddings.size()];
+          for (int i = 0; i < explicitPaddingsArray.length; ++i) {
+            explicitPaddingsArray[i] = opts.explicitPaddings.get(i);
+          }
+          opBuilder.setAttr("explicit_paddings", explicitPaddingsArray);
+        }
+        if (opts.dataFormat != null) {
+          opBuilder.setAttr("data_format", opts.dataFormat);
+        }
+        if (opts.dilations != null) {
+          long[] dilationsArray = new long[opts.dilations.size()];
+          for (int i = 0; i < dilationsArray.length; ++i) {
+            dilationsArray[i] = opts.dilations.get(i);
+          }
+          opBuilder.setAttr("dilations", dilationsArray);
+        }
+      }
+    }
+    return new DepthwiseConv2dNative<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param explicitPaddings 
+   */
+  public static Options explicitPaddings(List<Long> explicitPaddings) {
+    return new Options().explicitPaddings(explicitPaddings);
+  }
+  
+  /**
+   * @param dataFormat Specify the data format of the input and output data. With the
+   * default format "NHWC", the data is stored in the order of:
+   *     [batch, height, width, channels].
+   * Alternatively, the format could be "NCHW", the data storage order of:
+   *     [batch, channels, height, width].
+   */
+  public static Options dataFormat(String dataFormat) {
+    return new Options().dataFormat(dataFormat);
+  }
+  
+  /**
+   * @param dilations 1-D tensor of length 4.  The dilation factor for each dimension of
+   * `input`. If set to k > 1, there will be k-1 skipped cells between each filter
+   * element on that dimension. The dimension order is determined by the value of
+   * `data_format`, see above for details. Dilations in the batch and depth
+   * dimensions must be 1.
+   */
+  public static Options dilations(List<Long> dilations) {
+    return new Options().dilations(dilations);
+  }
+  
+  /**
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "DepthwiseConv2dNative";
+  
+  private Output<T> output;
+  
+  private DepthwiseConv2dNative(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/DepthwiseConv2dNativeBackpropFilter.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/DepthwiseConv2dNativeBackpropFilter.java
new file mode 100644
index 00000000000..ac51c06e737
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/DepthwiseConv2dNativeBackpropFilter.java
@@ -0,0 +1,195 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.nn;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Computes the gradients of depthwise convolution with respect to the filter.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator(group = "nn")
+public final class DepthwiseConv2dNativeBackpropFilter<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.nn.DepthwiseConv2dNativeBackpropFilter}
+   */
+  public static class Options {
+    
+    /**
+     * @param explicitPaddings 
+     */
+    public Options explicitPaddings(List<Long> explicitPaddings) {
+      this.explicitPaddings = explicitPaddings;
+      return this;
+    }
+    
+    /**
+     * @param dataFormat Specify the data format of the input and output data. With the
+     * default format "NHWC", the data is stored in the order of:
+     *     [batch, height, width, channels].
+     * Alternatively, the format could be "NCHW", the data storage order of:
+     *     [batch, channels, height, width].
+     */
+    public Options dataFormat(String dataFormat) {
+      this.dataFormat = dataFormat;
+      return this;
+    }
+    
+    /**
+     * @param dilations 1-D tensor of length 4.  The dilation factor for each dimension of
+     * `input`. If set to k > 1, there will be k-1 skipped cells between each filter
+     * element on that dimension. The dimension order is determined by the value of
+     * `data_format`, see above for details. Dilations in the batch and depth
+     * dimensions must be 1.
+     */
+    public Options dilations(List<Long> dilations) {
+      this.dilations = dilations;
+      return this;
+    }
+    
+    private List<Long> explicitPaddings;
+    private String dataFormat;
+    private List<Long> dilations;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new DepthwiseConv2dNativeBackpropFilter operation.
+   * 
+   * @param scope current scope
+   * @param input 4-D with shape based on `data_format`.  For example, if
+   * `data_format` is 'NHWC' then `input` is a 4-D `[batch, in_height,
+   * in_width, in_channels]` tensor.
+   * @param filterSizes An integer vector representing the tensor shape of `filter`,
+   * where `filter` is a 4-D
+   * `[filter_height, filter_width, in_channels, depthwise_multiplier]` tensor.
+   * @param outBackprop 4-D with shape  based on `data_format`.
+   * For example, if `data_format` is 'NHWC' then
+   * out_backprop shape is `[batch, out_height, out_width, out_channels]`.
+   * Gradients w.r.t. the output of the convolution.
+   * @param strides The stride of the sliding window for each dimension of the input
+   * of the convolution.
+   * @param padding The type of padding algorithm to use.
+   * @param options carries optional attributes values
+   * @return a new instance of DepthwiseConv2dNativeBackpropFilter
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> DepthwiseConv2dNativeBackpropFilter<T> create(Scope scope, Operand<T> input, Operand<TInt32> filterSizes, Operand<T> outBackprop, List<Long> strides, String padding, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("DepthwiseConv2dNativeBackpropFilter", scope.makeOpName("DepthwiseConv2dNativeBackpropFilter"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(filterSizes.asOutput());
+    opBuilder.addInput(outBackprop.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    long[] stridesArray = new long[strides.size()];
+    for (int i = 0; i < stridesArray.length; ++i) {
+      stridesArray[i] = strides.get(i);
+    }
+    opBuilder.setAttr("strides", stridesArray);
+    opBuilder.setAttr("padding", padding);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.explicitPaddings != null) {
+          long[] explicitPaddingsArray = new long[opts.explicitPaddings.size()];
+          for (int i = 0; i < explicitPaddingsArray.length; ++i) {
+            explicitPaddingsArray[i] = opts.explicitPaddings.get(i);
+          }
+          opBuilder.setAttr("explicit_paddings", explicitPaddingsArray);
+        }
+        if (opts.dataFormat != null) {
+          opBuilder.setAttr("data_format", opts.dataFormat);
+        }
+        if (opts.dilations != null) {
+          long[] dilationsArray = new long[opts.dilations.size()];
+          for (int i = 0; i < dilationsArray.length; ++i) {
+            dilationsArray[i] = opts.dilations.get(i);
+          }
+          opBuilder.setAttr("dilations", dilationsArray);
+        }
+      }
+    }
+    return new DepthwiseConv2dNativeBackpropFilter<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param explicitPaddings 
+   */
+  public static Options explicitPaddings(List<Long> explicitPaddings) {
+    return new Options().explicitPaddings(explicitPaddings);
+  }
+  
+  /**
+   * @param dataFormat Specify the data format of the input and output data. With the
+   * default format "NHWC", the data is stored in the order of:
+   *     [batch, height, width, channels].
+   * Alternatively, the format could be "NCHW", the data storage order of:
+   *     [batch, channels, height, width].
+   */
+  public static Options dataFormat(String dataFormat) {
+    return new Options().dataFormat(dataFormat);
+  }
+  
+  /**
+   * @param dilations 1-D tensor of length 4.  The dilation factor for each dimension of
+   * `input`. If set to k > 1, there will be k-1 skipped cells between each filter
+   * element on that dimension. The dimension order is determined by the value of
+   * `data_format`, see above for details. Dilations in the batch and depth
+   * dimensions must be 1.
+   */
+  public static Options dilations(List<Long> dilations) {
+    return new Options().dilations(dilations);
+  }
+  
+  /**
+   * 4-D with shape
+   * `[filter_height, filter_width, in_channels, out_channels]`.  Gradient w.r.t.
+   * the `filter` input of the convolution.
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "DepthwiseConv2dNativeBackpropFilter";
+  
+  private Output<T> output;
+  
+  private DepthwiseConv2dNativeBackpropFilter(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/DepthwiseConv2dNativeBackpropInput.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/DepthwiseConv2dNativeBackpropInput.java
new file mode 100644
index 00000000000..26db9841d7d
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/DepthwiseConv2dNativeBackpropInput.java
@@ -0,0 +1,195 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.nn;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Computes the gradients of depthwise convolution with respect to the input.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator(group = "nn")
+public final class DepthwiseConv2dNativeBackpropInput<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.nn.DepthwiseConv2dNativeBackpropInput}
+   */
+  public static class Options {
+    
+    /**
+     * @param explicitPaddings 
+     */
+    public Options explicitPaddings(List<Long> explicitPaddings) {
+      this.explicitPaddings = explicitPaddings;
+      return this;
+    }
+    
+    /**
+     * @param dataFormat Specify the data format of the input and output data. With the
+     * default format "NHWC", the data is stored in the order of:
+     *     [batch, height, width, channels].
+     * Alternatively, the format could be "NCHW", the data storage order of:
+     *     [batch, channels, height, width].
+     */
+    public Options dataFormat(String dataFormat) {
+      this.dataFormat = dataFormat;
+      return this;
+    }
+    
+    /**
+     * @param dilations 1-D tensor of length 4.  The dilation factor for each dimension of
+     * `input`. If set to k > 1, there will be k-1 skipped cells between each filter
+     * element on that dimension. The dimension order is determined by the value of
+     * `data_format`, see above for details. Dilations in the batch and depth
+     * dimensions must be 1.
+     */
+    public Options dilations(List<Long> dilations) {
+      this.dilations = dilations;
+      return this;
+    }
+    
+    private List<Long> explicitPaddings;
+    private String dataFormat;
+    private List<Long> dilations;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new DepthwiseConv2dNativeBackpropInput operation.
+   * 
+   * @param scope current scope
+   * @param inputSizes An integer vector representing the shape of `input`, based
+   * on `data_format`.  For example, if `data_format` is 'NHWC' then
+   *  `input` is a 4-D `[batch, height, width, channels]` tensor.
+   * @param filter 4-D with shape
+   * `[filter_height, filter_width, in_channels, depthwise_multiplier]`.
+   * @param outBackprop 4-D with shape  based on `data_format`.
+   * For example, if `data_format` is 'NHWC' then
+   * out_backprop shape is `[batch, out_height, out_width, out_channels]`.
+   * Gradients w.r.t. the output of the convolution.
+   * @param strides The stride of the sliding window for each dimension of the input
+   * of the convolution.
+   * @param padding The type of padding algorithm to use.
+   * @param options carries optional attributes values
+   * @return a new instance of DepthwiseConv2dNativeBackpropInput
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> DepthwiseConv2dNativeBackpropInput<T> create(Scope scope, Operand<TInt32> inputSizes, Operand<T> filter, Operand<T> outBackprop, List<Long> strides, String padding, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("DepthwiseConv2dNativeBackpropInput", scope.makeOpName("DepthwiseConv2dNativeBackpropInput"));
+    opBuilder.addInput(inputSizes.asOutput());
+    opBuilder.addInput(filter.asOutput());
+    opBuilder.addInput(outBackprop.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    long[] stridesArray = new long[strides.size()];
+    for (int i = 0; i < stridesArray.length; ++i) {
+      stridesArray[i] = strides.get(i);
+    }
+    opBuilder.setAttr("strides", stridesArray);
+    opBuilder.setAttr("padding", padding);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.explicitPaddings != null) {
+          long[] explicitPaddingsArray = new long[opts.explicitPaddings.size()];
+          for (int i = 0; i < explicitPaddingsArray.length; ++i) {
+            explicitPaddingsArray[i] = opts.explicitPaddings.get(i);
+          }
+          opBuilder.setAttr("explicit_paddings", explicitPaddingsArray);
+        }
+        if (opts.dataFormat != null) {
+          opBuilder.setAttr("data_format", opts.dataFormat);
+        }
+        if (opts.dilations != null) {
+          long[] dilationsArray = new long[opts.dilations.size()];
+          for (int i = 0; i < dilationsArray.length; ++i) {
+            dilationsArray[i] = opts.dilations.get(i);
+          }
+          opBuilder.setAttr("dilations", dilationsArray);
+        }
+      }
+    }
+    return new DepthwiseConv2dNativeBackpropInput<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param explicitPaddings 
+   */
+  public static Options explicitPaddings(List<Long> explicitPaddings) {
+    return new Options().explicitPaddings(explicitPaddings);
+  }
+  
+  /**
+   * @param dataFormat Specify the data format of the input and output data. With the
+   * default format "NHWC", the data is stored in the order of:
+   *     [batch, height, width, channels].
+   * Alternatively, the format could be "NCHW", the data storage order of:
+   *     [batch, channels, height, width].
+   */
+  public static Options dataFormat(String dataFormat) {
+    return new Options().dataFormat(dataFormat);
+  }
+  
+  /**
+   * @param dilations 1-D tensor of length 4.  The dilation factor for each dimension of
+   * `input`. If set to k > 1, there will be k-1 skipped cells between each filter
+   * element on that dimension. The dimension order is determined by the value of
+   * `data_format`, see above for details. Dilations in the batch and depth
+   * dimensions must be 1.
+   */
+  public static Options dilations(List<Long> dilations) {
+    return new Options().dilations(dilations);
+  }
+  
+  /**
+   * 4-D with shape according to `data_format`.  For example, if
+   * `data_format` is 'NHWC', output shape is `[batch, in_height,
+   * in_width, in_channels]`.  Gradient w.r.t. the input of the
+   * convolution.
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "DepthwiseConv2dNativeBackpropInput";
+  
+  private Output<T> output;
+  
+  private DepthwiseConv2dNativeBackpropInput(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Dilation2d.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Dilation2d.java
new file mode 100644
index 00000000000..e6bd8cc5b6a
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Dilation2d.java
@@ -0,0 +1,118 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.nn;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Computes the grayscale dilation of 4-D `input` and 3-D `filter` tensors.
+ * <p>
+ * The `input` tensor has shape `[batch, in_height, in_width, depth]` and the
+ * `filter` tensor has shape `[filter_height, filter_width, depth]`, i.e., each
+ * input channel is processed independently of the others with its own structuring
+ * function. The `output` tensor has shape
+ * `[batch, out_height, out_width, depth]`. The spatial dimensions of the output
+ * tensor depend on the `padding` algorithm. We currently only support the default
+ * "NHWC" `data_format`.
+ * <p>
+ * In detail, the grayscale morphological 2-D dilation is the max-sum correlation
+ * (for consistency with `conv2d`, we use unmirrored filters):
+ * <p>
+ *     output[b, y, x, c] =
+ *        max_{dy, dx} input[b,
+ *                           strides[1] * y + rates[1] * dy,
+ *                           strides[2] * x + rates[2] * dx,
+ *                           c] +
+ *                     filter[dy, dx, c]
+ * <p>
+ * Max-pooling is a special case when the filter has size equal to the pooling
+ * kernel size and contains all zeros.
+ * <p>
+ * Note on duality: The dilation of `input` by the `filter` is equal to the
+ * negation of the erosion of `-input` by the reflected `filter`.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator(group = "nn")
+public final class Dilation2d<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new Dilation2d operation.
+   * 
+   * @param scope current scope
+   * @param input 4-D with shape `[batch, in_height, in_width, depth]`.
+   * @param filter 3-D with shape `[filter_height, filter_width, depth]`.
+   * @param strides The stride of the sliding window for each dimension of the input
+   * tensor. Must be: `[1, stride_height, stride_width, 1]`.
+   * @param rates The input stride for atrous morphological dilation. Must be:
+   * `[1, rate_height, rate_width, 1]`.
+   * @param padding The type of padding algorithm to use.
+   * @return a new instance of Dilation2d
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> Dilation2d<T> create(Scope scope, Operand<T> input, Operand<T> filter, List<Long> strides, List<Long> rates, String padding) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Dilation2D", scope.makeOpName("Dilation2d"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(filter.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    long[] stridesArray = new long[strides.size()];
+    for (int i = 0; i < stridesArray.length; ++i) {
+      stridesArray[i] = strides.get(i);
+    }
+    opBuilder.setAttr("strides", stridesArray);
+    long[] ratesArray = new long[rates.size()];
+    for (int i = 0; i < ratesArray.length; ++i) {
+      ratesArray[i] = rates.get(i);
+    }
+    opBuilder.setAttr("rates", ratesArray);
+    opBuilder.setAttr("padding", padding);
+    return new Dilation2d<T>(opBuilder.build());
+  }
+  
+  /**
+   * 4-D with shape `[batch, out_height, out_width, depth]`.
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Dilation2D";
+  
+  private Output<T> output;
+  
+  private Dilation2d(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Dilation2dBackpropFilter.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Dilation2dBackpropFilter.java
new file mode 100644
index 00000000000..04705f53024
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Dilation2dBackpropFilter.java
@@ -0,0 +1,96 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.nn;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Computes the gradient of morphological 2-D dilation with respect to the filter.
+ * 
+ * @param <T> data type for {@code filterBackprop()} output
+ */
+@Operator(group = "nn")
+public final class Dilation2dBackpropFilter<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new Dilation2dBackpropFilter operation.
+   * 
+   * @param scope current scope
+   * @param input 4-D with shape `[batch, in_height, in_width, depth]`.
+   * @param filter 3-D with shape `[filter_height, filter_width, depth]`.
+   * @param outBackprop 4-D with shape `[batch, out_height, out_width, depth]`.
+   * @param strides 1-D of length 4. The stride of the sliding window for each dimension of
+   * the input tensor. Must be: `[1, stride_height, stride_width, 1]`.
+   * @param rates 1-D of length 4. The input stride for atrous morphological dilation.
+   * Must be: `[1, rate_height, rate_width, 1]`.
+   * @param padding The type of padding algorithm to use.
+   * @return a new instance of Dilation2dBackpropFilter
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> Dilation2dBackpropFilter<T> create(Scope scope, Operand<T> input, Operand<T> filter, Operand<T> outBackprop, List<Long> strides, List<Long> rates, String padding) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Dilation2DBackpropFilter", scope.makeOpName("Dilation2dBackpropFilter"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(filter.asOutput());
+    opBuilder.addInput(outBackprop.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    long[] stridesArray = new long[strides.size()];
+    for (int i = 0; i < stridesArray.length; ++i) {
+      stridesArray[i] = strides.get(i);
+    }
+    opBuilder.setAttr("strides", stridesArray);
+    long[] ratesArray = new long[rates.size()];
+    for (int i = 0; i < ratesArray.length; ++i) {
+      ratesArray[i] = rates.get(i);
+    }
+    opBuilder.setAttr("rates", ratesArray);
+    opBuilder.setAttr("padding", padding);
+    return new Dilation2dBackpropFilter<T>(opBuilder.build());
+  }
+  
+  /**
+   * 3-D with shape `[filter_height, filter_width, depth]`.
+   */
+  public Output<T> filterBackprop() {
+    return filterBackprop;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return filterBackprop;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Dilation2DBackpropFilter";
+  
+  private Output<T> filterBackprop;
+  
+  private Dilation2dBackpropFilter(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    filterBackprop = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Dilation2dBackpropInput.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Dilation2dBackpropInput.java
new file mode 100644
index 00000000000..1636545164c
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Dilation2dBackpropInput.java
@@ -0,0 +1,96 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.nn;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Computes the gradient of morphological 2-D dilation with respect to the input.
+ * 
+ * @param <T> data type for {@code inBackprop()} output
+ */
+@Operator(group = "nn")
+public final class Dilation2dBackpropInput<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new Dilation2dBackpropInput operation.
+   * 
+   * @param scope current scope
+   * @param input 4-D with shape `[batch, in_height, in_width, depth]`.
+   * @param filter 3-D with shape `[filter_height, filter_width, depth]`.
+   * @param outBackprop 4-D with shape `[batch, out_height, out_width, depth]`.
+   * @param strides 1-D of length 4. The stride of the sliding window for each dimension of
+   * the input tensor. Must be: `[1, stride_height, stride_width, 1]`.
+   * @param rates 1-D of length 4. The input stride for atrous morphological dilation.
+   * Must be: `[1, rate_height, rate_width, 1]`.
+   * @param padding The type of padding algorithm to use.
+   * @return a new instance of Dilation2dBackpropInput
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> Dilation2dBackpropInput<T> create(Scope scope, Operand<T> input, Operand<T> filter, Operand<T> outBackprop, List<Long> strides, List<Long> rates, String padding) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Dilation2DBackpropInput", scope.makeOpName("Dilation2dBackpropInput"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(filter.asOutput());
+    opBuilder.addInput(outBackprop.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    long[] stridesArray = new long[strides.size()];
+    for (int i = 0; i < stridesArray.length; ++i) {
+      stridesArray[i] = strides.get(i);
+    }
+    opBuilder.setAttr("strides", stridesArray);
+    long[] ratesArray = new long[rates.size()];
+    for (int i = 0; i < ratesArray.length; ++i) {
+      ratesArray[i] = rates.get(i);
+    }
+    opBuilder.setAttr("rates", ratesArray);
+    opBuilder.setAttr("padding", padding);
+    return new Dilation2dBackpropInput<T>(opBuilder.build());
+  }
+  
+  /**
+   * 4-D with shape `[batch, in_height, in_width, depth]`.
+   */
+  public Output<T> inBackprop() {
+    return inBackprop;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return inBackprop;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Dilation2DBackpropInput";
+  
+  private Output<T> inBackprop;
+  
+  private Dilation2dBackpropInput(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    inBackprop = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Elu.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Elu.java
new file mode 100644
index 00000000000..faa70f04b98
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Elu.java
@@ -0,0 +1,77 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.nn;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Computes exponential linear: `exp(features) - 1` if < 0, `features` otherwise.
+ * <p>
+ * See [Fast and Accurate Deep Network Learning by Exponential Linear Units (ELUs)
+ * ](http://arxiv.org/abs/1511.07289)
+ * 
+ * @param <T> data type for {@code activations()} output
+ */
+@Operator(group = "nn")
+public final class Elu<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new Elu operation.
+   * 
+   * @param scope current scope
+   * @param features 
+   * @return a new instance of Elu
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> Elu<T> create(Scope scope, Operand<T> features) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Elu", scope.makeOpName("Elu"));
+    opBuilder.addInput(features.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new Elu<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> activations() {
+    return activations;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return activations;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Elu";
+  
+  private Output<T> activations;
+  
+  private Elu(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    activations = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/EluGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/EluGrad.java
new file mode 100644
index 00000000000..9ead32623ab
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/EluGrad.java
@@ -0,0 +1,77 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.nn;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Computes gradients for the exponential linear (Elu) operation.
+ * 
+ * @param <T> data type for {@code backprops()} output
+ */
+public final class EluGrad<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new EluGrad operation.
+   * 
+   * @param scope current scope
+   * @param gradients The backpropagated gradients to the corresponding Elu operation.
+   * @param outputs The outputs of the corresponding Elu operation.
+   * @return a new instance of EluGrad
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> EluGrad<T> create(Scope scope, Operand<T> gradients, Operand<T> outputs) {
+    OperationBuilder opBuilder = scope.env().opBuilder("EluGrad", scope.makeOpName("EluGrad"));
+    opBuilder.addInput(gradients.asOutput());
+    opBuilder.addInput(outputs.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new EluGrad<T>(opBuilder.build());
+  }
+  
+  /**
+   * The gradients: `gradients * (outputs + 1)` if outputs < 0,
+   * `gradients` otherwise.
+   */
+  public Output<T> backprops() {
+    return backprops;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return backprops;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "EluGrad";
+  
+  private Output<T> backprops;
+  
+  private EluGrad(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    backprops = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/FixedUnigramCandidateSampler.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/FixedUnigramCandidateSampler.java
new file mode 100644
index 00000000000..25fd11071f9
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/FixedUnigramCandidateSampler.java
@@ -0,0 +1,329 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.nn;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.TInt64;
+
+/**
+ * Generates labels for candidate sampling with a learned unigram distribution.
+ * <p>
+ * A unigram sampler could use a fixed unigram distribution read from a
+ * file or passed in as an in-memory array instead of building up the distribution
+ * from data on the fly. There is also an option to skew the distribution by
+ * applying a distortion power to the weights.
+ * <p>
+ * The vocabulary file should be in CSV-like format, with the last field
+ * being the weight associated with the word.
+ * <p>
+ * For each batch, this op picks a single set of sampled candidate labels.
+ * <p>
+ * The advantages of sampling candidates per-batch are simplicity and the
+ * possibility of efficient dense matrix multiplication. The disadvantage is that
+ * the sampled candidates must be chosen independently of the context and of the
+ * true labels.
+ */
+@Operator(group = "nn")
+public final class FixedUnigramCandidateSampler extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.nn.FixedUnigramCandidateSampler}
+   */
+  public static class Options {
+    
+    /**
+     * @param vocabFile Each valid line in this file (which should have a CSV-like format)
+     * corresponds to a valid word ID. IDs are in sequential order, starting from
+     * num_reserved_ids. The last entry in each line is expected to be a value
+     * corresponding to the count or relative probability. Exactly one of vocab_file
+     * and unigrams needs to be passed to this op.
+     */
+    public Options vocabFile(String vocabFile) {
+      this.vocabFile = vocabFile;
+      return this;
+    }
+    
+    /**
+     * @param distortion The distortion is used to skew the unigram probability distribution.
+     * Each weight is first raised to the distortion's power before adding to the
+     * internal unigram distribution. As a result, distortion = 1.0 gives regular
+     * unigram sampling (as defined by the vocab file), and distortion = 0.0 gives
+     * a uniform distribution.
+     */
+    public Options distortion(Float distortion) {
+      this.distortion = distortion;
+      return this;
+    }
+    
+    /**
+     * @param numReservedIds Optionally some reserved IDs can be added in the range [0,
+     * ..., num_reserved_ids) by the users. One use case is that a special unknown
+     * word token is used as ID 0. These IDs will have a sampling probability of 0.
+     */
+    public Options numReservedIds(Long numReservedIds) {
+      this.numReservedIds = numReservedIds;
+      return this;
+    }
+    
+    /**
+     * @param numShards A sampler can be used to sample from a subset of the original range
+     * in order to speed up the whole computation through parallelism. This parameter
+     * (together with 'shard') indicates the number of partitions that are being
+     * used in the overall computation.
+     */
+    public Options numShards(Long numShards) {
+      this.numShards = numShards;
+      return this;
+    }
+    
+    /**
+     * @param shard A sampler can be used to sample from a subset of the original range
+     * in order to speed up the whole computation through parallelism. This parameter
+     * (together with 'num_shards') indicates the particular partition number of a
+     * sampler op, when partitioning is being used.
+     */
+    public Options shard(Long shard) {
+      this.shard = shard;
+      return this;
+    }
+    
+    /**
+     * @param unigrams A list of unigram counts or probabilities, one per ID in sequential
+     * order. Exactly one of vocab_file and unigrams should be passed to this op.
+     */
+    public Options unigrams(List<Float> unigrams) {
+      this.unigrams = unigrams;
+      return this;
+    }
+    
+    /**
+     * @param seed If either seed or seed2 are set to be non-zero, the random number
+     * generator is seeded by the given seed.  Otherwise, it is seeded by a
+     * random seed.
+     */
+    public Options seed(Long seed) {
+      this.seed = seed;
+      return this;
+    }
+    
+    /**
+     * @param seed2 An second seed to avoid seed collision.
+     */
+    public Options seed2(Long seed2) {
+      this.seed2 = seed2;
+      return this;
+    }
+    
+    private String vocabFile;
+    private Float distortion;
+    private Long numReservedIds;
+    private Long numShards;
+    private Long shard;
+    private List<Float> unigrams;
+    private Long seed;
+    private Long seed2;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new FixedUnigramCandidateSampler operation.
+   * 
+   * @param scope current scope
+   * @param trueClasses A batch_size * num_true matrix, in which each row contains the
+   * IDs of the num_true target_classes in the corresponding original label.
+   * @param numTrue Number of true labels per context.
+   * @param numSampled Number of candidates to randomly sample.
+   * @param unique If unique is true, we sample with rejection, so that all sampled
+   * candidates in a batch are unique. This requires some approximation to
+   * estimate the post-rejection sampling probabilities.
+   * @param rangeMax The sampler will sample integers from the interval [0, range_max).
+   * @param options carries optional attributes values
+   * @return a new instance of FixedUnigramCandidateSampler
+   */
+  @Endpoint(describeByClass = true)
+  public static FixedUnigramCandidateSampler create(Scope scope, Operand<TInt64> trueClasses, Long numTrue, Long numSampled, Boolean unique, Long rangeMax, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("FixedUnigramCandidateSampler", scope.makeOpName("FixedUnigramCandidateSampler"));
+    opBuilder.addInput(trueClasses.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("num_true", numTrue);
+    opBuilder.setAttr("num_sampled", numSampled);
+    opBuilder.setAttr("unique", unique);
+    opBuilder.setAttr("range_max", rangeMax);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.vocabFile != null) {
+          opBuilder.setAttr("vocab_file", opts.vocabFile);
+        }
+        if (opts.distortion != null) {
+          opBuilder.setAttr("distortion", opts.distortion);
+        }
+        if (opts.numReservedIds != null) {
+          opBuilder.setAttr("num_reserved_ids", opts.numReservedIds);
+        }
+        if (opts.numShards != null) {
+          opBuilder.setAttr("num_shards", opts.numShards);
+        }
+        if (opts.shard != null) {
+          opBuilder.setAttr("shard", opts.shard);
+        }
+        if (opts.unigrams != null) {
+          float[] unigramsArray = new float[opts.unigrams.size()];
+          for (int i = 0; i < unigramsArray.length; ++i) {
+            unigramsArray[i] = opts.unigrams.get(i);
+          }
+          opBuilder.setAttr("unigrams", unigramsArray);
+        }
+        if (opts.seed != null) {
+          opBuilder.setAttr("seed", opts.seed);
+        }
+        if (opts.seed2 != null) {
+          opBuilder.setAttr("seed2", opts.seed2);
+        }
+      }
+    }
+    return new FixedUnigramCandidateSampler(opBuilder.build());
+  }
+  
+  /**
+   * @param vocabFile Each valid line in this file (which should have a CSV-like format)
+   * corresponds to a valid word ID. IDs are in sequential order, starting from
+   * num_reserved_ids. The last entry in each line is expected to be a value
+   * corresponding to the count or relative probability. Exactly one of vocab_file
+   * and unigrams needs to be passed to this op.
+   */
+  public static Options vocabFile(String vocabFile) {
+    return new Options().vocabFile(vocabFile);
+  }
+  
+  /**
+   * @param distortion The distortion is used to skew the unigram probability distribution.
+   * Each weight is first raised to the distortion's power before adding to the
+   * internal unigram distribution. As a result, distortion = 1.0 gives regular
+   * unigram sampling (as defined by the vocab file), and distortion = 0.0 gives
+   * a uniform distribution.
+   */
+  public static Options distortion(Float distortion) {
+    return new Options().distortion(distortion);
+  }
+  
+  /**
+   * @param numReservedIds Optionally some reserved IDs can be added in the range [0,
+   * ..., num_reserved_ids) by the users. One use case is that a special unknown
+   * word token is used as ID 0. These IDs will have a sampling probability of 0.
+   */
+  public static Options numReservedIds(Long numReservedIds) {
+    return new Options().numReservedIds(numReservedIds);
+  }
+  
+  /**
+   * @param numShards A sampler can be used to sample from a subset of the original range
+   * in order to speed up the whole computation through parallelism. This parameter
+   * (together with 'shard') indicates the number of partitions that are being
+   * used in the overall computation.
+   */
+  public static Options numShards(Long numShards) {
+    return new Options().numShards(numShards);
+  }
+  
+  /**
+   * @param shard A sampler can be used to sample from a subset of the original range
+   * in order to speed up the whole computation through parallelism. This parameter
+   * (together with 'num_shards') indicates the particular partition number of a
+   * sampler op, when partitioning is being used.
+   */
+  public static Options shard(Long shard) {
+    return new Options().shard(shard);
+  }
+  
+  /**
+   * @param unigrams A list of unigram counts or probabilities, one per ID in sequential
+   * order. Exactly one of vocab_file and unigrams should be passed to this op.
+   */
+  public static Options unigrams(List<Float> unigrams) {
+    return new Options().unigrams(unigrams);
+  }
+  
+  /**
+   * @param seed If either seed or seed2 are set to be non-zero, the random number
+   * generator is seeded by the given seed.  Otherwise, it is seeded by a
+   * random seed.
+   */
+  public static Options seed(Long seed) {
+    return new Options().seed(seed);
+  }
+  
+  /**
+   * @param seed2 An second seed to avoid seed collision.
+   */
+  public static Options seed2(Long seed2) {
+    return new Options().seed2(seed2);
+  }
+  
+  /**
+   * A vector of length num_sampled, in which each element is
+   * the ID of a sampled candidate.
+   */
+  public Output<TInt64> sampledCandidates() {
+    return sampledCandidates;
+  }
+  
+  /**
+   * A batch_size * num_true matrix, representing
+   * the number of times each candidate is expected to occur in a batch
+   * of sampled candidates. If unique=true, then this is a probability.
+   */
+  public Output<TFloat32> trueExpectedCount() {
+    return trueExpectedCount;
+  }
+  
+  /**
+   * A vector of length num_sampled, for each sampled
+   * candidate representing the number of times the candidate is expected
+   * to occur in a batch of sampled candidates.  If unique=true, then this is a
+   * probability.
+   */
+  public Output<TFloat32> sampledExpectedCount() {
+    return sampledExpectedCount;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "FixedUnigramCandidateSampler";
+  
+  private Output<TInt64> sampledCandidates;
+  private Output<TFloat32> trueExpectedCount;
+  private Output<TFloat32> sampledExpectedCount;
+  
+  private FixedUnigramCandidateSampler(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    sampledCandidates = operation.output(outputIdx++);
+    trueExpectedCount = operation.output(outputIdx++);
+    sampledExpectedCount = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/FractionalAvgPool.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/FractionalAvgPool.java
new file mode 100644
index 00000000000..93a477ef25e
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/FractionalAvgPool.java
@@ -0,0 +1,246 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.nn;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Performs fractional average pooling on the input.
+ * <p>
+ * Fractional average pooling is similar to Fractional max pooling in the pooling
+ * region generation step. The only difference is that after pooling regions are
+ * generated, a mean operation is performed instead of a max operation in each
+ * pooling region.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator(group = "nn")
+public final class FractionalAvgPool<T extends TNumber> extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.nn.FractionalAvgPool}
+   */
+  public static class Options {
+    
+    /**
+     * @param pseudoRandom When set to True, generates the pooling sequence in a
+     * pseudorandom fashion, otherwise, in a random fashion. Check paper [Benjamin
+     * Graham, Fractional Max-Pooling](http://arxiv.org/abs/1412.6071) for
+     * difference between pseudorandom and random.
+     */
+    public Options pseudoRandom(Boolean pseudoRandom) {
+      this.pseudoRandom = pseudoRandom;
+      return this;
+    }
+    
+    /**
+     * @param overlapping When set to True, it means when pooling, the values at the boundary
+     * of adjacent pooling cells are used by both cells. For example:
+     * <p>
+     * `index  0  1  2  3  4`
+     * <p>
+     * `value  20 5  16 3  7`
+     * <p>
+     * If the pooling sequence is [0, 2, 4], then 16, at index 2 will be used twice.
+     * The result would be [41/3, 26/3] for fractional avg pooling.
+     */
+    public Options overlapping(Boolean overlapping) {
+      this.overlapping = overlapping;
+      return this;
+    }
+    
+    /**
+     * @param deterministic When set to True, a fixed pooling region will be used when
+     * iterating over a FractionalAvgPool node in the computation graph. Mainly used
+     * in unit test to make FractionalAvgPool deterministic.
+     */
+    public Options deterministic(Boolean deterministic) {
+      this.deterministic = deterministic;
+      return this;
+    }
+    
+    /**
+     * @param seed If either seed or seed2 are set to be non-zero, the random number
+     * generator is seeded by the given seed.  Otherwise, it is seeded by a
+     * random seed.
+     */
+    public Options seed(Long seed) {
+      this.seed = seed;
+      return this;
+    }
+    
+    /**
+     * @param seed2 An second seed to avoid seed collision.
+     */
+    public Options seed2(Long seed2) {
+      this.seed2 = seed2;
+      return this;
+    }
+    
+    private Boolean pseudoRandom;
+    private Boolean overlapping;
+    private Boolean deterministic;
+    private Long seed;
+    private Long seed2;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new FractionalAvgPool operation.
+   * 
+   * @param scope current scope
+   * @param value 4-D with shape `[batch, height, width, channels]`.
+   * @param poolingRatio Pooling ratio for each dimension of `value`, currently only
+   * supports row and col dimension and should be >= 1.0. For example, a valid
+   * pooling ratio looks like [1.0, 1.44, 1.73, 1.0]. The first and last elements
+   * must be 1.0 because we don't allow pooling on batch and channels
+   * dimensions. 1.44 and 1.73 are pooling ratio on height and width dimensions
+   * respectively.
+   * @param options carries optional attributes values
+   * @return a new instance of FractionalAvgPool
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> FractionalAvgPool<T> create(Scope scope, Operand<T> value, List<Float> poolingRatio, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("FractionalAvgPool", scope.makeOpName("FractionalAvgPool"));
+    opBuilder.addInput(value.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    float[] poolingRatioArray = new float[poolingRatio.size()];
+    for (int i = 0; i < poolingRatioArray.length; ++i) {
+      poolingRatioArray[i] = poolingRatio.get(i);
+    }
+    opBuilder.setAttr("pooling_ratio", poolingRatioArray);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.pseudoRandom != null) {
+          opBuilder.setAttr("pseudo_random", opts.pseudoRandom);
+        }
+        if (opts.overlapping != null) {
+          opBuilder.setAttr("overlapping", opts.overlapping);
+        }
+        if (opts.deterministic != null) {
+          opBuilder.setAttr("deterministic", opts.deterministic);
+        }
+        if (opts.seed != null) {
+          opBuilder.setAttr("seed", opts.seed);
+        }
+        if (opts.seed2 != null) {
+          opBuilder.setAttr("seed2", opts.seed2);
+        }
+      }
+    }
+    return new FractionalAvgPool<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param pseudoRandom When set to True, generates the pooling sequence in a
+   * pseudorandom fashion, otherwise, in a random fashion. Check paper [Benjamin
+   * Graham, Fractional Max-Pooling](http://arxiv.org/abs/1412.6071) for
+   * difference between pseudorandom and random.
+   */
+  public static Options pseudoRandom(Boolean pseudoRandom) {
+    return new Options().pseudoRandom(pseudoRandom);
+  }
+  
+  /**
+   * @param overlapping When set to True, it means when pooling, the values at the boundary
+   * of adjacent pooling cells are used by both cells. For example:
+   * <p>
+   * `index  0  1  2  3  4`
+   * <p>
+   * `value  20 5  16 3  7`
+   * <p>
+   * If the pooling sequence is [0, 2, 4], then 16, at index 2 will be used twice.
+   * The result would be [41/3, 26/3] for fractional avg pooling.
+   */
+  public static Options overlapping(Boolean overlapping) {
+    return new Options().overlapping(overlapping);
+  }
+  
+  /**
+   * @param deterministic When set to True, a fixed pooling region will be used when
+   * iterating over a FractionalAvgPool node in the computation graph. Mainly used
+   * in unit test to make FractionalAvgPool deterministic.
+   */
+  public static Options deterministic(Boolean deterministic) {
+    return new Options().deterministic(deterministic);
+  }
+  
+  /**
+   * @param seed If either seed or seed2 are set to be non-zero, the random number
+   * generator is seeded by the given seed.  Otherwise, it is seeded by a
+   * random seed.
+   */
+  public static Options seed(Long seed) {
+    return new Options().seed(seed);
+  }
+  
+  /**
+   * @param seed2 An second seed to avoid seed collision.
+   */
+  public static Options seed2(Long seed2) {
+    return new Options().seed2(seed2);
+  }
+  
+  /**
+   * output tensor after fractional avg pooling.
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  /**
+   * row pooling sequence, needed to calculate gradient.
+   */
+  public Output<TInt64> rowPoolingSequence() {
+    return rowPoolingSequence;
+  }
+  
+  /**
+   * column pooling sequence, needed to calculate gradient.
+   */
+  public Output<TInt64> colPoolingSequence() {
+    return colPoolingSequence;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "FractionalAvgPool";
+  
+  private Output<T> output;
+  private Output<TInt64> rowPoolingSequence;
+  private Output<TInt64> colPoolingSequence;
+  
+  private FractionalAvgPool(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+    rowPoolingSequence = operation.output(outputIdx++);
+    colPoolingSequence = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/FractionalAvgPoolGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/FractionalAvgPoolGrad.java
new file mode 100644
index 00000000000..7a4258d18c8
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/FractionalAvgPoolGrad.java
@@ -0,0 +1,140 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.nn;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Computes gradient of the FractionalAvgPool function.
+ * <p>
+ * Unlike FractionalMaxPoolGrad, we don't need to find arg_max for
+ * FractionalAvgPoolGrad, we just need to evenly back-propagate each element of
+ * out_backprop to those indices that form the same pooling cell. Therefore, we
+ * just need to know the shape of original input tensor, instead of the whole
+ * tensor.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+public final class FractionalAvgPoolGrad<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.nn.FractionalAvgPoolGrad}
+   */
+  public static class Options {
+    
+    /**
+     * @param overlapping When set to True, it means when pooling, the values at the boundary
+     * of adjacent pooling cells are used by both cells. For example:
+     * <p>
+     * `index  0  1  2  3  4`
+     * <p>
+     * `value  20 5  16 3  7`
+     * <p>
+     * If the pooling sequence is [0, 2, 4], then 16, at index 2 will be used twice.
+     * The result would be [41/3, 26/3] for fractional avg pooling.
+     */
+    public Options overlapping(Boolean overlapping) {
+      this.overlapping = overlapping;
+      return this;
+    }
+    
+    private Boolean overlapping;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new FractionalAvgPoolGrad operation.
+   * 
+   * @param scope current scope
+   * @param origInputTensorShape Original input tensor shape for `fractional_avg_pool`
+   * @param outBackprop 4-D with shape `[batch, height, width, channels]`.  Gradients
+   * w.r.t. the output of `fractional_avg_pool`.
+   * @param rowPoolingSequence row pooling sequence, form pooling region with
+   * col_pooling_sequence.
+   * @param colPoolingSequence column pooling sequence, form pooling region with
+   * row_pooling sequence.
+   * @param options carries optional attributes values
+   * @return a new instance of FractionalAvgPoolGrad
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> FractionalAvgPoolGrad<T> create(Scope scope, Operand<TInt64> origInputTensorShape, Operand<T> outBackprop, Operand<TInt64> rowPoolingSequence, Operand<TInt64> colPoolingSequence, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("FractionalAvgPoolGrad", scope.makeOpName("FractionalAvgPoolGrad"));
+    opBuilder.addInput(origInputTensorShape.asOutput());
+    opBuilder.addInput(outBackprop.asOutput());
+    opBuilder.addInput(rowPoolingSequence.asOutput());
+    opBuilder.addInput(colPoolingSequence.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.overlapping != null) {
+          opBuilder.setAttr("overlapping", opts.overlapping);
+        }
+      }
+    }
+    return new FractionalAvgPoolGrad<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param overlapping When set to True, it means when pooling, the values at the boundary
+   * of adjacent pooling cells are used by both cells. For example:
+   * <p>
+   * `index  0  1  2  3  4`
+   * <p>
+   * `value  20 5  16 3  7`
+   * <p>
+   * If the pooling sequence is [0, 2, 4], then 16, at index 2 will be used twice.
+   * The result would be [41/3, 26/3] for fractional avg pooling.
+   */
+  public static Options overlapping(Boolean overlapping) {
+    return new Options().overlapping(overlapping);
+  }
+  
+  /**
+   * 4-D.  Gradients w.r.t. the input of `fractional_avg_pool`.
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "FractionalAvgPoolGrad";
+  
+  private Output<T> output;
+  
+  private FractionalAvgPoolGrad(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/FractionalMaxPool.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/FractionalMaxPool.java
new file mode 100644
index 00000000000..9ca640c804b
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/FractionalMaxPool.java
@@ -0,0 +1,270 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.nn;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Performs fractional max pooling on the input.
+ * <p>
+ * Fractional max pooling is slightly different than regular max pooling.  In
+ * regular max pooling, you downsize an input set by taking the maximum value of
+ * smaller N x N subsections of the set (often 2x2), and try to reduce the set by
+ * a factor of N, where N is an integer.  Fractional max pooling, as you might
+ * expect from the word "fractional", means that the overall reduction ratio N
+ * does not have to be an integer.
+ * <p>
+ * The sizes of the pooling regions are generated randomly but are fairly uniform.
+ * For example, let's look at the height dimension, and the constraints on the
+ * list of rows that will be pool boundaries.
+ * <p>
+ * First we define the following:
+ * <p>
+ * 1.  input_row_length : the number of rows from the input set
+ * 2.  output_row_length : which will be smaller than the input
+ * 3.  alpha = input_row_length / output_row_length : our reduction ratio
+ * 4.  K = floor(alpha)
+ * 5.  row_pooling_sequence : this is the result list of pool boundary rows
+ * <p>
+ * Then, row_pooling_sequence should satisfy:
+ * <p>
+ * 1.  a[0] = 0 : the first value of the sequence is 0
+ * 2.  a[end] = input_row_length : the last value of the sequence is the size
+ * 3.  K <= (a[i+1] - a[i]) <= K+1 : all intervals are K or K+1 size
+ * 4.  length(row_pooling_sequence) = output_row_length+1
+ * <p>
+ * For more details on fractional max pooling, see this paper:
+ * [Benjamin Graham, Fractional Max-Pooling](http://arxiv.org/abs/1412.6071)
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator(group = "nn")
+public final class FractionalMaxPool<T extends TNumber> extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.nn.FractionalMaxPool}
+   */
+  public static class Options {
+    
+    /**
+     * @param pseudoRandom When set to True, generates the pooling sequence in a
+     * pseudorandom fashion, otherwise, in a random fashion. Check paper [Benjamin
+     * Graham, Fractional Max-Pooling](http://arxiv.org/abs/1412.6071) for
+     * difference between pseudorandom and random.
+     */
+    public Options pseudoRandom(Boolean pseudoRandom) {
+      this.pseudoRandom = pseudoRandom;
+      return this;
+    }
+    
+    /**
+     * @param overlapping When set to True, it means when pooling, the values at the boundary
+     * of adjacent pooling cells are used by both cells. For example:
+     * <p>
+     * `index  0  1  2  3  4`
+     * <p>
+     * `value  20 5  16 3  7`
+     * <p>
+     * If the pooling sequence is [0, 2, 4], then 16, at index 2 will be used twice.
+     * The result would be [20, 16] for fractional max pooling.
+     */
+    public Options overlapping(Boolean overlapping) {
+      this.overlapping = overlapping;
+      return this;
+    }
+    
+    /**
+     * @param deterministic When set to True, a fixed pooling region will be used when
+     * iterating over a FractionalMaxPool node in the computation graph. Mainly used
+     * in unit test to make FractionalMaxPool deterministic.
+     */
+    public Options deterministic(Boolean deterministic) {
+      this.deterministic = deterministic;
+      return this;
+    }
+    
+    /**
+     * @param seed If either seed or seed2 are set to be non-zero, the random number
+     * generator is seeded by the given seed.  Otherwise, it is seeded by a
+     * random seed.
+     */
+    public Options seed(Long seed) {
+      this.seed = seed;
+      return this;
+    }
+    
+    /**
+     * @param seed2 An second seed to avoid seed collision.
+     */
+    public Options seed2(Long seed2) {
+      this.seed2 = seed2;
+      return this;
+    }
+    
+    private Boolean pseudoRandom;
+    private Boolean overlapping;
+    private Boolean deterministic;
+    private Long seed;
+    private Long seed2;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new FractionalMaxPool operation.
+   * 
+   * @param scope current scope
+   * @param value 4-D with shape `[batch, height, width, channels]`.
+   * @param poolingRatio Pooling ratio for each dimension of `value`, currently only
+   * supports row and col dimension and should be >= 1.0. For example, a valid
+   * pooling ratio looks like [1.0, 1.44, 1.73, 1.0]. The first and last elements
+   * must be 1.0 because we don't allow pooling on batch and channels
+   * dimensions. 1.44 and 1.73 are pooling ratio on height and width dimensions
+   * respectively.
+   * @param options carries optional attributes values
+   * @return a new instance of FractionalMaxPool
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> FractionalMaxPool<T> create(Scope scope, Operand<T> value, List<Float> poolingRatio, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("FractionalMaxPool", scope.makeOpName("FractionalMaxPool"));
+    opBuilder.addInput(value.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    float[] poolingRatioArray = new float[poolingRatio.size()];
+    for (int i = 0; i < poolingRatioArray.length; ++i) {
+      poolingRatioArray[i] = poolingRatio.get(i);
+    }
+    opBuilder.setAttr("pooling_ratio", poolingRatioArray);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.pseudoRandom != null) {
+          opBuilder.setAttr("pseudo_random", opts.pseudoRandom);
+        }
+        if (opts.overlapping != null) {
+          opBuilder.setAttr("overlapping", opts.overlapping);
+        }
+        if (opts.deterministic != null) {
+          opBuilder.setAttr("deterministic", opts.deterministic);
+        }
+        if (opts.seed != null) {
+          opBuilder.setAttr("seed", opts.seed);
+        }
+        if (opts.seed2 != null) {
+          opBuilder.setAttr("seed2", opts.seed2);
+        }
+      }
+    }
+    return new FractionalMaxPool<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param pseudoRandom When set to True, generates the pooling sequence in a
+   * pseudorandom fashion, otherwise, in a random fashion. Check paper [Benjamin
+   * Graham, Fractional Max-Pooling](http://arxiv.org/abs/1412.6071) for
+   * difference between pseudorandom and random.
+   */
+  public static Options pseudoRandom(Boolean pseudoRandom) {
+    return new Options().pseudoRandom(pseudoRandom);
+  }
+  
+  /**
+   * @param overlapping When set to True, it means when pooling, the values at the boundary
+   * of adjacent pooling cells are used by both cells. For example:
+   * <p>
+   * `index  0  1  2  3  4`
+   * <p>
+   * `value  20 5  16 3  7`
+   * <p>
+   * If the pooling sequence is [0, 2, 4], then 16, at index 2 will be used twice.
+   * The result would be [20, 16] for fractional max pooling.
+   */
+  public static Options overlapping(Boolean overlapping) {
+    return new Options().overlapping(overlapping);
+  }
+  
+  /**
+   * @param deterministic When set to True, a fixed pooling region will be used when
+   * iterating over a FractionalMaxPool node in the computation graph. Mainly used
+   * in unit test to make FractionalMaxPool deterministic.
+   */
+  public static Options deterministic(Boolean deterministic) {
+    return new Options().deterministic(deterministic);
+  }
+  
+  /**
+   * @param seed If either seed or seed2 are set to be non-zero, the random number
+   * generator is seeded by the given seed.  Otherwise, it is seeded by a
+   * random seed.
+   */
+  public static Options seed(Long seed) {
+    return new Options().seed(seed);
+  }
+  
+  /**
+   * @param seed2 An second seed to avoid seed collision.
+   */
+  public static Options seed2(Long seed2) {
+    return new Options().seed2(seed2);
+  }
+  
+  /**
+   * output tensor after fractional max pooling.
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  /**
+   * row pooling sequence, needed to calculate gradient.
+   */
+  public Output<TInt64> rowPoolingSequence() {
+    return rowPoolingSequence;
+  }
+  
+  /**
+   * column pooling sequence, needed to calculate gradient.
+   */
+  public Output<TInt64> colPoolingSequence() {
+    return colPoolingSequence;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "FractionalMaxPool";
+  
+  private Output<T> output;
+  private Output<TInt64> rowPoolingSequence;
+  private Output<TInt64> colPoolingSequence;
+  
+  private FractionalMaxPool(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+    rowPoolingSequence = operation.output(outputIdx++);
+    colPoolingSequence = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/FractionalMaxPoolGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/FractionalMaxPoolGrad.java
new file mode 100644
index 00000000000..952c8ce286a
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/FractionalMaxPoolGrad.java
@@ -0,0 +1,136 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.nn;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Computes gradient of the FractionalMaxPool function.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+public final class FractionalMaxPoolGrad<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.nn.FractionalMaxPoolGrad}
+   */
+  public static class Options {
+    
+    /**
+     * @param overlapping When set to True, it means when pooling, the values at the boundary
+     * of adjacent pooling cells are used by both cells. For example:
+     * <p>
+     * `index  0  1  2  3  4`
+     * <p>
+     * `value  20 5  16 3  7`
+     * <p>
+     * If the pooling sequence is [0, 2, 4], then 16, at index 2 will be used twice.
+     * The result would be [20, 16] for fractional max pooling.
+     */
+    public Options overlapping(Boolean overlapping) {
+      this.overlapping = overlapping;
+      return this;
+    }
+    
+    private Boolean overlapping;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new FractionalMaxPoolGrad operation.
+   * 
+   * @param scope current scope
+   * @param origInput Original input for `fractional_max_pool`
+   * @param origOutput Original output for `fractional_max_pool`
+   * @param outBackprop 4-D with shape `[batch, height, width, channels]`.  Gradients
+   * w.r.t. the output of `fractional_max_pool`.
+   * @param rowPoolingSequence row pooling sequence, form pooling region with
+   * col_pooling_sequence.
+   * @param colPoolingSequence column pooling sequence, form pooling region with
+   * row_pooling sequence.
+   * @param options carries optional attributes values
+   * @return a new instance of FractionalMaxPoolGrad
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> FractionalMaxPoolGrad<T> create(Scope scope, Operand<T> origInput, Operand<T> origOutput, Operand<T> outBackprop, Operand<TInt64> rowPoolingSequence, Operand<TInt64> colPoolingSequence, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("FractionalMaxPoolGrad", scope.makeOpName("FractionalMaxPoolGrad"));
+    opBuilder.addInput(origInput.asOutput());
+    opBuilder.addInput(origOutput.asOutput());
+    opBuilder.addInput(outBackprop.asOutput());
+    opBuilder.addInput(rowPoolingSequence.asOutput());
+    opBuilder.addInput(colPoolingSequence.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.overlapping != null) {
+          opBuilder.setAttr("overlapping", opts.overlapping);
+        }
+      }
+    }
+    return new FractionalMaxPoolGrad<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param overlapping When set to True, it means when pooling, the values at the boundary
+   * of adjacent pooling cells are used by both cells. For example:
+   * <p>
+   * `index  0  1  2  3  4`
+   * <p>
+   * `value  20 5  16 3  7`
+   * <p>
+   * If the pooling sequence is [0, 2, 4], then 16, at index 2 will be used twice.
+   * The result would be [20, 16] for fractional max pooling.
+   */
+  public static Options overlapping(Boolean overlapping) {
+    return new Options().overlapping(overlapping);
+  }
+  
+  /**
+   * 4-D.  Gradients w.r.t. the input of `fractional_max_pool`.
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "FractionalMaxPoolGrad";
+  
+  private Output<T> output;
+  
+  private FractionalMaxPoolGrad(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/FusedBatchNorm.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/FusedBatchNorm.java
new file mode 100644
index 00000000000..0d2b3a6dfe9
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/FusedBatchNorm.java
@@ -0,0 +1,227 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.nn;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Batch normalization.
+ * <p>
+ * Note that the size of 4D Tensors are defined by either "NHWC" or "NCHW".
+ * The size of 1D Tensors matches the dimension C of the 4D Tensors.
+ * 
+ * @param <T> data type for {@code y()} output
+ * @param <U> data type for {@code batchMean()} output
+ */
+@Operator(group = "nn")
+public final class FusedBatchNorm<T extends TNumber, U extends TNumber> extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.nn.FusedBatchNorm}
+   */
+  public static class Options {
+    
+    /**
+     * @param epsilon A small float number added to the variance of x.
+     */
+    public Options epsilon(Float epsilon) {
+      this.epsilon = epsilon;
+      return this;
+    }
+    
+    /**
+     * @param exponentialAvgFactor 
+     */
+    public Options exponentialAvgFactor(Float exponentialAvgFactor) {
+      this.exponentialAvgFactor = exponentialAvgFactor;
+      return this;
+    }
+    
+    /**
+     * @param dataFormat The data format for x and y. Either "NHWC" (default) or "NCHW".
+     */
+    public Options dataFormat(String dataFormat) {
+      this.dataFormat = dataFormat;
+      return this;
+    }
+    
+    /**
+     * @param isTraining A bool value to indicate the operation is for training (default)
+     * or inference.
+     */
+    public Options isTraining(Boolean isTraining) {
+      this.isTraining = isTraining;
+      return this;
+    }
+    
+    private Float epsilon;
+    private Float exponentialAvgFactor;
+    private String dataFormat;
+    private Boolean isTraining;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new FusedBatchNorm operation.
+   * 
+   * @param scope current scope
+   * @param x A 4D Tensor for input data.
+   * @param scale A 1D Tensor for scaling factor, to scale the normalized x.
+   * @param offset A 1D Tensor for offset, to shift to the normalized x.
+   * @param mean A 1D Tensor for population mean. Used for inference only;
+   * must be empty for training.
+   * @param variance A 1D Tensor for population variance. Used for inference only;
+   * must be empty for training.
+   * @param options carries optional attributes values
+   * @return a new instance of FusedBatchNorm
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber, U extends TNumber> FusedBatchNorm<T, U> create(Scope scope, Operand<T> x, Operand<U> scale, Operand<U> offset, Operand<U> mean, Operand<U> variance, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("FusedBatchNormV3", scope.makeOpName("FusedBatchNorm"));
+    opBuilder.addInput(x.asOutput());
+    opBuilder.addInput(scale.asOutput());
+    opBuilder.addInput(offset.asOutput());
+    opBuilder.addInput(mean.asOutput());
+    opBuilder.addInput(variance.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.epsilon != null) {
+          opBuilder.setAttr("epsilon", opts.epsilon);
+        }
+        if (opts.exponentialAvgFactor != null) {
+          opBuilder.setAttr("exponential_avg_factor", opts.exponentialAvgFactor);
+        }
+        if (opts.dataFormat != null) {
+          opBuilder.setAttr("data_format", opts.dataFormat);
+        }
+        if (opts.isTraining != null) {
+          opBuilder.setAttr("is_training", opts.isTraining);
+        }
+      }
+    }
+    return new FusedBatchNorm<T, U>(opBuilder.build());
+  }
+  
+  /**
+   * @param epsilon A small float number added to the variance of x.
+   */
+  public static Options epsilon(Float epsilon) {
+    return new Options().epsilon(epsilon);
+  }
+  
+  /**
+   * @param exponentialAvgFactor 
+   */
+  public static Options exponentialAvgFactor(Float exponentialAvgFactor) {
+    return new Options().exponentialAvgFactor(exponentialAvgFactor);
+  }
+  
+  /**
+   * @param dataFormat The data format for x and y. Either "NHWC" (default) or "NCHW".
+   */
+  public static Options dataFormat(String dataFormat) {
+    return new Options().dataFormat(dataFormat);
+  }
+  
+  /**
+   * @param isTraining A bool value to indicate the operation is for training (default)
+   * or inference.
+   */
+  public static Options isTraining(Boolean isTraining) {
+    return new Options().isTraining(isTraining);
+  }
+  
+  /**
+   * A 4D Tensor for output data.
+   */
+  public Output<T> y() {
+    return y;
+  }
+  
+  /**
+   * A 1D Tensor for the computed batch mean, to be used by TensorFlow
+   * to compute the running mean.
+   */
+  public Output<U> batchMean() {
+    return batchMean;
+  }
+  
+  /**
+   * A 1D Tensor for the computed batch variance, to be used by
+   * TensorFlow to compute the running variance.
+   */
+  public Output<U> batchVariance() {
+    return batchVariance;
+  }
+  
+  /**
+   * A 1D Tensor for the computed batch mean, to be reused
+   * in the gradient computation.
+   */
+  public Output<U> reserveSpace1() {
+    return reserveSpace1;
+  }
+  
+  /**
+   * A 1D Tensor for the computed batch variance (inverted variance
+   * in the cuDNN case), to be reused in the gradient computation.
+   */
+  public Output<U> reserveSpace2() {
+    return reserveSpace2;
+  }
+  
+  /**
+   * A 1D Tensor for some intermediate results, to be reused in the gradient
+   * computation for better efficiency.
+   */
+  public Output<U> reserveSpace3() {
+    return reserveSpace3;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "FusedBatchNormV3";
+  
+  private Output<T> y;
+  private Output<U> batchMean;
+  private Output<U> batchVariance;
+  private Output<U> reserveSpace1;
+  private Output<U> reserveSpace2;
+  private Output<U> reserveSpace3;
+  
+  private FusedBatchNorm(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    y = operation.output(outputIdx++);
+    batchMean = operation.output(outputIdx++);
+    batchVariance = operation.output(outputIdx++);
+    reserveSpace1 = operation.output(outputIdx++);
+    reserveSpace2 = operation.output(outputIdx++);
+    reserveSpace3 = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/FusedBatchNormGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/FusedBatchNormGrad.java
new file mode 100644
index 00000000000..8e9f616adcb
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/FusedBatchNormGrad.java
@@ -0,0 +1,207 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.nn;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Gradient for batch normalization.
+ * <p>
+ * Note that the size of 4D Tensors are defined by either "NHWC" or "NCHW".
+ * The size of 1D Tensors matches the dimension C of the 4D Tensors.
+ * 
+ * @param <T> data type for {@code xBackprop()} output
+ * @param <U> data type for {@code scaleBackprop()} output
+ */
+@Operator(group = "nn")
+public final class FusedBatchNormGrad<T extends TNumber, U extends TNumber> extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.nn.FusedBatchNormGrad}
+   */
+  public static class Options {
+    
+    /**
+     * @param epsilon A small float number added to the variance of x.
+     */
+    public Options epsilon(Float epsilon) {
+      this.epsilon = epsilon;
+      return this;
+    }
+    
+    /**
+     * @param dataFormat The data format for y_backprop, x, x_backprop.
+     * Either "NHWC" (default) or "NCHW".
+     */
+    public Options dataFormat(String dataFormat) {
+      this.dataFormat = dataFormat;
+      return this;
+    }
+    
+    /**
+     * @param isTraining A bool value to indicate the operation is for training (default)
+     * or inference.
+     */
+    public Options isTraining(Boolean isTraining) {
+      this.isTraining = isTraining;
+      return this;
+    }
+    
+    private Float epsilon;
+    private String dataFormat;
+    private Boolean isTraining;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new FusedBatchNormGrad operation.
+   * 
+   * @param scope current scope
+   * @param yBackprop A 4D Tensor for the gradient with respect to y.
+   * @param x A 4D Tensor for input data.
+   * @param scale A 1D Tensor for scaling factor, to scale the normalized x.
+   * @param reserveSpace1 When is_training is True, a 1D Tensor for the computed batch
+   * mean to be reused in gradient computation. When is_training is
+   * False, a 1D Tensor for the population mean to be reused in both
+   * 1st and 2nd order gradient computation.
+   * @param reserveSpace2 When is_training is True, a 1D Tensor for the computed batch
+   * variance (inverted variance in the cuDNN case) to be reused in
+   * gradient computation. When is_training is False, a 1D Tensor
+   * for the population variance to be reused in both 1st and 2nd
+   * order gradient computation.
+   * @param reserveSpace3 When is_training is True, a 1D Tensor for some intermediate results to be reused
+   * in gradient computation. When is_training is False, a dummy empty Tensor will be
+   * created.
+   * @param options carries optional attributes values
+   * @return a new instance of FusedBatchNormGrad
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber, U extends TNumber> FusedBatchNormGrad<T, U> create(Scope scope, Operand<T> yBackprop, Operand<T> x, Operand<TFloat32> scale, Operand<U> reserveSpace1, Operand<U> reserveSpace2, Operand<U> reserveSpace3, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("FusedBatchNormGradV3", scope.makeOpName("FusedBatchNormGrad"));
+    opBuilder.addInput(yBackprop.asOutput());
+    opBuilder.addInput(x.asOutput());
+    opBuilder.addInput(scale.asOutput());
+    opBuilder.addInput(reserveSpace1.asOutput());
+    opBuilder.addInput(reserveSpace2.asOutput());
+    opBuilder.addInput(reserveSpace3.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.epsilon != null) {
+          opBuilder.setAttr("epsilon", opts.epsilon);
+        }
+        if (opts.dataFormat != null) {
+          opBuilder.setAttr("data_format", opts.dataFormat);
+        }
+        if (opts.isTraining != null) {
+          opBuilder.setAttr("is_training", opts.isTraining);
+        }
+      }
+    }
+    return new FusedBatchNormGrad<T, U>(opBuilder.build());
+  }
+  
+  /**
+   * @param epsilon A small float number added to the variance of x.
+   */
+  public static Options epsilon(Float epsilon) {
+    return new Options().epsilon(epsilon);
+  }
+  
+  /**
+   * @param dataFormat The data format for y_backprop, x, x_backprop.
+   * Either "NHWC" (default) or "NCHW".
+   */
+  public static Options dataFormat(String dataFormat) {
+    return new Options().dataFormat(dataFormat);
+  }
+  
+  /**
+   * @param isTraining A bool value to indicate the operation is for training (default)
+   * or inference.
+   */
+  public static Options isTraining(Boolean isTraining) {
+    return new Options().isTraining(isTraining);
+  }
+  
+  /**
+   * A 4D Tensor for the gradient with respect to x.
+   */
+  public Output<T> xBackprop() {
+    return xBackprop;
+  }
+  
+  /**
+   * A 1D Tensor for the gradient with respect to scale.
+   */
+  public Output<U> scaleBackprop() {
+    return scaleBackprop;
+  }
+  
+  /**
+   * A 1D Tensor for the gradient with respect to offset.
+   */
+  public Output<U> offsetBackprop() {
+    return offsetBackprop;
+  }
+  
+  /**
+   * Unused placeholder to match the mean input in FusedBatchNorm.
+   */
+  public Output<U> reserveSpace4() {
+    return reserveSpace4;
+  }
+  
+  /**
+   * Unused placeholder to match the variance input
+   * in FusedBatchNorm.
+   */
+  public Output<U> reserveSpace5() {
+    return reserveSpace5;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "FusedBatchNormGradV3";
+  
+  private Output<T> xBackprop;
+  private Output<U> scaleBackprop;
+  private Output<U> offsetBackprop;
+  private Output<U> reserveSpace4;
+  private Output<U> reserveSpace5;
+  
+  private FusedBatchNormGrad(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    xBackprop = operation.output(outputIdx++);
+    scaleBackprop = operation.output(outputIdx++);
+    offsetBackprop = operation.output(outputIdx++);
+    reserveSpace4 = operation.output(outputIdx++);
+    reserveSpace5 = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/FusedPadConv2d.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/FusedPadConv2d.java
new file mode 100644
index 00000000000..89713030f7b
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/FusedPadConv2d.java
@@ -0,0 +1,105 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.nn;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Performs a padding as a preprocess during a convolution.
+ * <p>
+ * Similar to FusedResizeAndPadConv2d, this op allows for an optimized
+ * implementation where the spatial padding transformation stage is fused with the
+ * im2col lookup, but in this case without the bilinear filtering required for
+ * resizing. Fusing the padding prevents the need to write out the intermediate
+ * results as whole tensors, reducing memory pressure, and we can get some latency
+ * gains by merging the transformation calculations.
+ * The data_format attribute for Conv2D isn't supported by this op, and 'NHWC'
+ * order is used instead.
+ * Internally this op uses a single per-graph scratch buffer, which means that it
+ * will block if multiple versions are being run in parallel. This is because this
+ * operator is primarily an optimization to minimize memory usage.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator(group = "nn")
+public final class FusedPadConv2d<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new FusedPadConv2d operation.
+   * 
+   * @param scope current scope
+   * @param input 4-D with shape `[batch, in_height, in_width, in_channels]`.
+   * @param paddings A two-column matrix specifying the padding sizes. The number of
+   * rows must be the same as the rank of `input`.
+   * @param filter 4-D with shape
+   * `[filter_height, filter_width, in_channels, out_channels]`.
+   * @param mode 
+   * @param strides 1-D of length 4.  The stride of the sliding window for each dimension
+   * of `input`. Must be in the same order as the dimension specified with format.
+   * @param padding The type of padding algorithm to use.
+   * @return a new instance of FusedPadConv2d
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> FusedPadConv2d<T> create(Scope scope, Operand<T> input, Operand<TInt32> paddings, Operand<T> filter, String mode, List<Long> strides, String padding) {
+    OperationBuilder opBuilder = scope.env().opBuilder("FusedPadConv2D", scope.makeOpName("FusedPadConv2d"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(paddings.asOutput());
+    opBuilder.addInput(filter.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("mode", mode);
+    long[] stridesArray = new long[strides.size()];
+    for (int i = 0; i < stridesArray.length; ++i) {
+      stridesArray[i] = strides.get(i);
+    }
+    opBuilder.setAttr("strides", stridesArray);
+    opBuilder.setAttr("padding", padding);
+    return new FusedPadConv2d<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "FusedPadConv2D";
+  
+  private Output<T> output;
+  
+  private FusedPadConv2d(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/FusedResizeAndPadConv2d.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/FusedResizeAndPadConv2d.java
new file mode 100644
index 00000000000..b89af628392
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/FusedResizeAndPadConv2d.java
@@ -0,0 +1,143 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.nn;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Performs a resize and padding as a preprocess during a convolution.
+ * <p>
+ * It's often possible to do spatial transformations more efficiently as part of
+ * the packing stage of a convolution, so this op allows for an optimized
+ * implementation where these stages are fused together. This prevents the need to
+ * write out the intermediate results as whole tensors, reducing memory pressure,
+ * and we can get some latency gains by merging the transformation calculations.
+ * The data_format attribute for Conv2D isn't supported by this op, and defaults to
+ * 'NHWC' order.
+ * Internally this op uses a single per-graph scratch buffer, which means that it
+ * will block if multiple versions are being run in parallel. This is because this
+ * operator is primarily an optimization to minimize memory usage.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator(group = "nn")
+public final class FusedResizeAndPadConv2d<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.nn.FusedResizeAndPadConv2d}
+   */
+  public static class Options {
+    
+    /**
+     * @param resizeAlignCorners If true, the centers of the 4 corner pixels of the input and output tensors are
+     * aligned, preserving the values at the corner pixels. Defaults to false.
+     */
+    public Options resizeAlignCorners(Boolean resizeAlignCorners) {
+      this.resizeAlignCorners = resizeAlignCorners;
+      return this;
+    }
+    
+    private Boolean resizeAlignCorners;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new FusedResizeAndPadConv2d operation.
+   * 
+   * @param scope current scope
+   * @param input 4-D with shape `[batch, in_height, in_width, in_channels]`.
+   * @param size A 1-D int32 Tensor of 2 elements: `new_height, new_width`.  The
+   * new size for the images.
+   * @param paddings A two-column matrix specifying the padding sizes. The number of
+   * rows must be the same as the rank of `input`.
+   * @param filter 4-D with shape
+   * `[filter_height, filter_width, in_channels, out_channels]`.
+   * @param mode 
+   * @param strides 1-D of length 4.  The stride of the sliding window for each dimension
+   * of `input`. Must be in the same order as the dimension specified with format.
+   * @param padding The type of padding algorithm to use.
+   * @param options carries optional attributes values
+   * @return a new instance of FusedResizeAndPadConv2d
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> FusedResizeAndPadConv2d<T> create(Scope scope, Operand<T> input, Operand<TInt32> size, Operand<TInt32> paddings, Operand<T> filter, String mode, List<Long> strides, String padding, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("FusedResizeAndPadConv2D", scope.makeOpName("FusedResizeAndPadConv2d"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(size.asOutput());
+    opBuilder.addInput(paddings.asOutput());
+    opBuilder.addInput(filter.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("mode", mode);
+    long[] stridesArray = new long[strides.size()];
+    for (int i = 0; i < stridesArray.length; ++i) {
+      stridesArray[i] = strides.get(i);
+    }
+    opBuilder.setAttr("strides", stridesArray);
+    opBuilder.setAttr("padding", padding);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.resizeAlignCorners != null) {
+          opBuilder.setAttr("resize_align_corners", opts.resizeAlignCorners);
+        }
+      }
+    }
+    return new FusedResizeAndPadConv2d<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param resizeAlignCorners If true, the centers of the 4 corner pixels of the input and output tensors are
+   * aligned, preserving the values at the corner pixels. Defaults to false.
+   */
+  public static Options resizeAlignCorners(Boolean resizeAlignCorners) {
+    return new Options().resizeAlignCorners(resizeAlignCorners);
+  }
+  
+  /**
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "FusedResizeAndPadConv2D";
+  
+  private Output<T> output;
+  
+  private FusedResizeAndPadConv2d(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/GRUBlockCell.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/GRUBlockCell.java
new file mode 100644
index 00000000000..fad66727252
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/GRUBlockCell.java
@@ -0,0 +1,147 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.nn;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Computes the GRU cell forward propagation for 1 time step.
+ * <p>
+ * Args
+ *     x: Input to the GRU cell.
+ *     h_prev: State input from the previous GRU cell.
+ *     w_ru: Weight matrix for the reset and update gate.
+ *     w_c: Weight matrix for the cell connection gate.
+ *     b_ru: Bias vector for the reset and update gate.
+ *     b_c: Bias vector for the cell connection gate.
+ * <p>
+ * Returns
+ *     r: Output of the reset gate.
+ *     u: Output of the update gate.
+ *     c: Output of the cell connection gate.
+ *     h: Current state of the GRU cell.
+ * <p>
+ * Note on notation of the variables:
+ * <p>
+ * Concatenation of a and b is represented by a_b
+ * Element-wise dot product of a and b is represented by ab
+ * Element-wise dot product is represented by \circ
+ * Matrix multiplication is represented by *
+ * <p>
+ * Biases are initialized with :
+ * `b_ru` - constant_initializer(1.0)
+ * `b_c` - constant_initializer(0.0)
+ * <p>
+ * This kernel op implements the following mathematical equations:
+ * <pre>{@code
+ * x_h_prev = [x, h_prev]
+ * 
+ * [r_bar u_bar] = x_h_prev * w_ru + b_ru
+ * 
+ * r = sigmoid(r_bar)
+ * u = sigmoid(u_bar)
+ * 
+ * h_prevr = h_prev \circ r
+ * 
+ * x_h_prevr = [x h_prevr]
+ * 
+ * c_bar = x_h_prevr * w_c + b_c
+ * c = tanh(c_bar)
+ * 
+ * h = (1-u) \circ c + u \circ h_prev
+ * }</pre>
+ * 
+ * 
+ * @param <T> data type for {@code r()} output
+ */
+public final class GRUBlockCell<T extends TNumber> extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new GRUBlockCell operation.
+   * 
+   * @param scope current scope
+   * @param x 
+   * @param hPrev 
+   * @param wRu 
+   * @param wC 
+   * @param bRu 
+   * @param bC 
+   * @return a new instance of GRUBlockCell
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> GRUBlockCell<T> create(Scope scope, Operand<T> x, Operand<T> hPrev, Operand<T> wRu, Operand<T> wC, Operand<T> bRu, Operand<T> bC) {
+    OperationBuilder opBuilder = scope.env().opBuilder("GRUBlockCell", scope.makeOpName("GRUBlockCell"));
+    opBuilder.addInput(x.asOutput());
+    opBuilder.addInput(hPrev.asOutput());
+    opBuilder.addInput(wRu.asOutput());
+    opBuilder.addInput(wC.asOutput());
+    opBuilder.addInput(bRu.asOutput());
+    opBuilder.addInput(bC.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new GRUBlockCell<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> r() {
+    return r;
+  }
+  
+  /**
+   */
+  public Output<T> u() {
+    return u;
+  }
+  
+  /**
+   */
+  public Output<T> c() {
+    return c;
+  }
+  
+  /**
+   */
+  public Output<T> h() {
+    return h;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "GRUBlockCell";
+  
+  private Output<T> r;
+  private Output<T> u;
+  private Output<T> c;
+  private Output<T> h;
+  
+  private GRUBlockCell(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    r = operation.output(outputIdx++);
+    u = operation.output(outputIdx++);
+    c = operation.output(outputIdx++);
+    h = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/GRUBlockCellGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/GRUBlockCellGrad.java
new file mode 100644
index 00000000000..69c4295d5c4
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/GRUBlockCellGrad.java
@@ -0,0 +1,191 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.nn;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Computes the GRU cell back-propagation for 1 time step.
+ * <p>
+ * Args
+ *     x: Input to the GRU cell.
+ *     h_prev: State input from the previous GRU cell.
+ *     w_ru: Weight matrix for the reset and update gate.
+ *     w_c: Weight matrix for the cell connection gate.
+ *     b_ru: Bias vector for the reset and update gate.
+ *     b_c: Bias vector for the cell connection gate.
+ *     r: Output of the reset gate.
+ *     u: Output of the update gate.
+ *     c: Output of the cell connection gate.
+ *     d_h: Gradients of the h_new wrt to objective function.
+ * <p>
+ * Returns
+ *     d_x: Gradients of the x wrt to objective function.
+ *     d_h_prev: Gradients of the h wrt to objective function.
+ *     d_c_bar Gradients of the c_bar wrt to objective function.
+ *     d_r_bar_u_bar Gradients of the r_bar & u_bar wrt to objective function.
+ * <p>
+ * This kernel op implements the following mathematical equations:
+ * <p>
+ * Note on notation of the variables:
+ * <p>
+ * Concatenation of a and b is represented by a_b
+ * Element-wise dot product of a and b is represented by ab
+ * Element-wise dot product is represented by \circ
+ * Matrix multiplication is represented by *
+ * <p>
+ * Additional notes for clarity:
+ * <p>
+ * `w_ru` can be segmented into 4 different matrices.
+ * <pre>{@code
+ * w_ru = [w_r_x w_u_x
+ *         w_r_h_prev w_u_h_prev]
+ * }</pre>
+ * Similarly, `w_c` can be segmented into 2 different matrices.
+ * <pre>{@code
+ * w_c = [w_c_x w_c_h_prevr]
+ * }</pre>
+ * Same goes for biases.
+ * <pre>{@code
+ * b_ru = [b_ru_x b_ru_h]
+ * b_c = [b_c_x b_c_h]
+ * }</pre>
+ * Another note on notation:
+ * <pre>{@code
+ * d_x = d_x_component_1 + d_x_component_2
+ * 
+ * where d_x_component_1 = d_r_bar * w_r_x^T + d_u_bar * w_r_x^T
+ * and d_x_component_2 = d_c_bar * w_c_x^T
+ * 
+ * d_h_prev = d_h_prev_component_1 + d_h_prevr \circ r + d_h \circ u
+ * where d_h_prev_componenet_1 = d_r_bar * w_r_h_prev^T + d_u_bar * w_r_h_prev^T
+ * }</pre>
+ * Mathematics behind the Gradients below:
+ * <pre>{@code
+ * d_c_bar = d_h \circ (1-u) \circ (1-c \circ c)
+ * d_u_bar = d_h \circ (h-c) \circ u \circ (1-u)
+ * 
+ * d_r_bar_u_bar = [d_r_bar d_u_bar]
+ * 
+ * [d_x_component_1 d_h_prev_component_1] = d_r_bar_u_bar * w_ru^T
+ * 
+ * [d_x_component_2 d_h_prevr] = d_c_bar * w_c^T
+ * 
+ * d_x = d_x_component_1 + d_x_component_2
+ * 
+ * d_h_prev = d_h_prev_component_1 + d_h_prevr \circ r + u
+ * }</pre>
+ * Below calculation is performed in the python wrapper for the Gradients
+ * (not in the gradient kernel.)
+ * <pre>{@code
+ * d_w_ru = x_h_prevr^T * d_c_bar
+ * 
+ * d_w_c = x_h_prev^T * d_r_bar_u_bar
+ * 
+ * d_b_ru = sum of d_r_bar_u_bar along axis = 0
+ * 
+ * d_b_c = sum of d_c_bar along axis = 0
+ * }</pre>
+ * 
+ * 
+ * @param <T> data type for {@code dX()} output
+ */
+public final class GRUBlockCellGrad<T extends TNumber> extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new GRUBlockCellGrad operation.
+   * 
+   * @param scope current scope
+   * @param x 
+   * @param hPrev 
+   * @param wRu 
+   * @param wC 
+   * @param bRu 
+   * @param bC 
+   * @param r 
+   * @param u 
+   * @param c 
+   * @param dH 
+   * @return a new instance of GRUBlockCellGrad
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> GRUBlockCellGrad<T> create(Scope scope, Operand<T> x, Operand<T> hPrev, Operand<T> wRu, Operand<T> wC, Operand<T> bRu, Operand<T> bC, Operand<T> r, Operand<T> u, Operand<T> c, Operand<T> dH) {
+    OperationBuilder opBuilder = scope.env().opBuilder("GRUBlockCellGrad", scope.makeOpName("GRUBlockCellGrad"));
+    opBuilder.addInput(x.asOutput());
+    opBuilder.addInput(hPrev.asOutput());
+    opBuilder.addInput(wRu.asOutput());
+    opBuilder.addInput(wC.asOutput());
+    opBuilder.addInput(bRu.asOutput());
+    opBuilder.addInput(bC.asOutput());
+    opBuilder.addInput(r.asOutput());
+    opBuilder.addInput(u.asOutput());
+    opBuilder.addInput(c.asOutput());
+    opBuilder.addInput(dH.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new GRUBlockCellGrad<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> dX() {
+    return dX;
+  }
+  
+  /**
+   */
+  public Output<T> dHPrev() {
+    return dHPrev;
+  }
+  
+  /**
+   */
+  public Output<T> dCBar() {
+    return dCBar;
+  }
+  
+  /**
+   */
+  public Output<T> dRBarUBar() {
+    return dRBarUBar;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "GRUBlockCellGrad";
+  
+  private Output<T> dX;
+  private Output<T> dHPrev;
+  private Output<T> dCBar;
+  private Output<T> dRBarUBar;
+  
+  private GRUBlockCellGrad(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    dX = operation.output(outputIdx++);
+    dHPrev = operation.output(outputIdx++);
+    dCBar = operation.output(outputIdx++);
+    dRBarUBar = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/InTopK.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/InTopK.java
new file mode 100644
index 00000000000..ea2a2652b83
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/InTopK.java
@@ -0,0 +1,94 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.nn;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TBool;
+import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Says whether the targets are in the top `K` predictions.
+ * <p>
+ * This outputs a `batch_size` bool array, an entry `out[i]` is `true` if the
+ * prediction for the target class is among the top `k` predictions among
+ * all predictions for example `i`. Note that the behavior of `InTopK` differs
+ * from the `TopK` op in its handling of ties; if multiple classes have the
+ * same prediction value and straddle the top-`k` boundary, all of those
+ * classes are considered to be in the top `k`.
+ * <p>
+ * More formally, let
+ * <p>
+ *   \\(predictions_i\\) be the predictions for all classes for example `i`,
+ *   \\(targets_i\\) be the target class for example `i`,
+ *   \\(out_i\\) be the output for example `i`,
+ * <p>
+ * $$out_i = predictions_{i, targets_i} \in TopKIncludingTies(predictions_i)$$
+ */
+@Operator(group = "nn")
+public final class InTopK extends RawOp implements Operand<TBool> {
+  
+  /**
+   * Factory method to create a class wrapping a new InTopK operation.
+   * 
+   * @param scope current scope
+   * @param predictions A `batch_size` x `classes` tensor.
+   * @param targets A `batch_size` vector of class ids.
+   * @param k Number of top elements to look at for computing precision.
+   * @return a new instance of InTopK
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> InTopK create(Scope scope, Operand<TFloat32> predictions, Operand<T> targets, Operand<T> k) {
+    OperationBuilder opBuilder = scope.env().opBuilder("InTopKV2", scope.makeOpName("InTopK"));
+    opBuilder.addInput(predictions.asOutput());
+    opBuilder.addInput(targets.asOutput());
+    opBuilder.addInput(k.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new InTopK(opBuilder.build());
+  }
+  
+  /**
+   * Computed precision at `k` as a `bool Tensor`.
+   */
+  public Output<TBool> precision() {
+    return precision;
+  }
+  
+  @Override
+  public Output<TBool> asOutput() {
+    return precision;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "InTopKV2";
+  
+  private Output<TBool> precision;
+  
+  private InTopK(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    precision = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/InvGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/InvGrad.java
new file mode 100644
index 00000000000..161d8771b83
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/InvGrad.java
@@ -0,0 +1,78 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.nn;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Computes the gradient for the inverse of `x` wrt its input.
+ * <p>
+ * Specifically, `grad = -dy <i> y</i>y`, where `y = 1/x`, and `dy`
+ * is the corresponding input gradient.
+ * 
+ * @param <T> data type for {@code z()} output
+ */
+public final class InvGrad<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new InvGrad operation.
+   * 
+   * @param scope current scope
+   * @param y 
+   * @param dy 
+   * @return a new instance of InvGrad
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> InvGrad<T> create(Scope scope, Operand<T> y, Operand<T> dy) {
+    OperationBuilder opBuilder = scope.env().opBuilder("InvGrad", scope.makeOpName("InvGrad"));
+    opBuilder.addInput(y.asOutput());
+    opBuilder.addInput(dy.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new InvGrad<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> z() {
+    return z;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return z;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "InvGrad";
+  
+  private Output<T> z;
+  
+  private InvGrad(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    z = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/L2Loss.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/L2Loss.java
new file mode 100644
index 00000000000..eb38a9baa65
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/L2Loss.java
@@ -0,0 +1,79 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.nn;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * L2 Loss.
+ * <p>
+ * Computes half the L2 norm of a tensor without the `sqrt`:
+ * <p>
+ *     output = sum(t ** 2) / 2
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator(group = "nn")
+public final class L2Loss<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new L2Loss operation.
+   * 
+   * @param scope current scope
+   * @param t Typically 2-D, but may have any dimensions.
+   * @return a new instance of L2Loss
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> L2Loss<T> create(Scope scope, Operand<T> t) {
+    OperationBuilder opBuilder = scope.env().opBuilder("L2Loss", scope.makeOpName("L2Loss"));
+    opBuilder.addInput(t.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new L2Loss<T>(opBuilder.build());
+  }
+  
+  /**
+   * 0-D.
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "L2Loss";
+  
+  private Output<T> output;
+  
+  private L2Loss(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/LSTMBlockCell.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/LSTMBlockCell.java
new file mode 100644
index 00000000000..160d2bc89be
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/LSTMBlockCell.java
@@ -0,0 +1,234 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.nn;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Computes the LSTM cell forward propagation for 1 time step.
+ * <p>
+ * This implementation uses 1 weight matrix and 1 bias vector, and there's an
+ * optional peephole connection.
+ * <p>
+ * This kernel op implements the following mathematical equations:
+ * <pre>{@code
+ * xh = [x, h_prev]
+ * [i, f, ci, o] = xh * w + b
+ * f = f + forget_bias
+ * 
+ * if not use_peephole:
+ *   wci = wcf = wco = 0
+ * 
+ * i = sigmoid(cs_prev * wci + i)
+ * f = sigmoid(cs_prev * wcf + f)
+ * ci = tanh(ci)
+ * 
+ * cs = ci .* i + cs_prev .* f
+ * cs = clip(cs, cell_clip)
+ * 
+ * o = sigmoid(cs * wco + o)
+ * co = tanh(cs)
+ * h = co .* o
+ * }</pre>
+ * 
+ * 
+ * @param <T> data type for {@code i()} output
+ */
+public final class LSTMBlockCell<T extends TNumber> extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.nn.LSTMBlockCell}
+   */
+  public static class Options {
+    
+    /**
+     * @param forgetBias The forget gate bias.
+     */
+    public Options forgetBias(Float forgetBias) {
+      this.forgetBias = forgetBias;
+      return this;
+    }
+    
+    /**
+     * @param cellClip Value to clip the 'cs' value to.
+     */
+    public Options cellClip(Float cellClip) {
+      this.cellClip = cellClip;
+      return this;
+    }
+    
+    /**
+     * @param usePeephole Whether to use peephole weights.
+     */
+    public Options usePeephole(Boolean usePeephole) {
+      this.usePeephole = usePeephole;
+      return this;
+    }
+    
+    private Float forgetBias;
+    private Float cellClip;
+    private Boolean usePeephole;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new LSTMBlockCell operation.
+   * 
+   * @param scope current scope
+   * @param x The input to the LSTM cell, shape (batch_size, num_inputs).
+   * @param csPrev Value of the cell state at previous time step.
+   * @param hPrev Output of the previous cell at previous time step.
+   * @param w The weight matrix.
+   * @param wci The weight matrix for input gate peephole connection.
+   * @param wcf The weight matrix for forget gate peephole connection.
+   * @param wco The weight matrix for output gate peephole connection.
+   * @param b The bias vector.
+   * @param options carries optional attributes values
+   * @return a new instance of LSTMBlockCell
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> LSTMBlockCell<T> create(Scope scope, Operand<T> x, Operand<T> csPrev, Operand<T> hPrev, Operand<T> w, Operand<T> wci, Operand<T> wcf, Operand<T> wco, Operand<T> b, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("LSTMBlockCell", scope.makeOpName("LSTMBlockCell"));
+    opBuilder.addInput(x.asOutput());
+    opBuilder.addInput(csPrev.asOutput());
+    opBuilder.addInput(hPrev.asOutput());
+    opBuilder.addInput(w.asOutput());
+    opBuilder.addInput(wci.asOutput());
+    opBuilder.addInput(wcf.asOutput());
+    opBuilder.addInput(wco.asOutput());
+    opBuilder.addInput(b.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.forgetBias != null) {
+          opBuilder.setAttr("forget_bias", opts.forgetBias);
+        }
+        if (opts.cellClip != null) {
+          opBuilder.setAttr("cell_clip", opts.cellClip);
+        }
+        if (opts.usePeephole != null) {
+          opBuilder.setAttr("use_peephole", opts.usePeephole);
+        }
+      }
+    }
+    return new LSTMBlockCell<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param forgetBias The forget gate bias.
+   */
+  public static Options forgetBias(Float forgetBias) {
+    return new Options().forgetBias(forgetBias);
+  }
+  
+  /**
+   * @param cellClip Value to clip the 'cs' value to.
+   */
+  public static Options cellClip(Float cellClip) {
+    return new Options().cellClip(cellClip);
+  }
+  
+  /**
+   * @param usePeephole Whether to use peephole weights.
+   */
+  public static Options usePeephole(Boolean usePeephole) {
+    return new Options().usePeephole(usePeephole);
+  }
+  
+  /**
+   * The input gate.
+   */
+  public Output<T> i() {
+    return i;
+  }
+  
+  /**
+   * The cell state before the tanh.
+   */
+  public Output<T> cs() {
+    return cs;
+  }
+  
+  /**
+   * The forget gate.
+   */
+  public Output<T> f() {
+    return f;
+  }
+  
+  /**
+   * The output gate.
+   */
+  public Output<T> o() {
+    return o;
+  }
+  
+  /**
+   * The cell input.
+   */
+  public Output<T> ci() {
+    return ci;
+  }
+  
+  /**
+   * The cell after the tanh.
+   */
+  public Output<T> co() {
+    return co;
+  }
+  
+  /**
+   * The output h vector.
+   */
+  public Output<T> h() {
+    return h;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "LSTMBlockCell";
+  
+  private Output<T> i;
+  private Output<T> cs;
+  private Output<T> f;
+  private Output<T> o;
+  private Output<T> ci;
+  private Output<T> co;
+  private Output<T> h;
+  
+  private LSTMBlockCell(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    i = operation.output(outputIdx++);
+    cs = operation.output(outputIdx++);
+    f = operation.output(outputIdx++);
+    o = operation.output(outputIdx++);
+    ci = operation.output(outputIdx++);
+    co = operation.output(outputIdx++);
+    h = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/LSTMBlockCellGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/LSTMBlockCellGrad.java
new file mode 100644
index 00000000000..3a36077d6ca
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/LSTMBlockCellGrad.java
@@ -0,0 +1,139 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.nn;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Computes the LSTM cell backward propagation for 1 timestep.
+ * <p>
+ * This implementation is to be used in conjunction of LSTMBlockCell.
+ * 
+ * @param <T> data type for {@code csPrevGrad()} output
+ */
+public final class LSTMBlockCellGrad<T extends TNumber> extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new LSTMBlockCellGrad operation.
+   * 
+   * @param scope current scope
+   * @param x The input to the LSTM cell, shape (batch_size, num_inputs).
+   * @param csPrev The previous cell state.
+   * @param hPrev The previous h state.
+   * @param w The weight matrix.
+   * @param wci The weight matrix for input gate peephole connection.
+   * @param wcf The weight matrix for forget gate peephole connection.
+   * @param wco The weight matrix for output gate peephole connection.
+   * @param b The bias vector.
+   * @param i The input gate.
+   * @param cs The cell state before the tanh.
+   * @param f The forget gate.
+   * @param o The output gate.
+   * @param ci The cell input.
+   * @param co The cell after the tanh.
+   * @param csGrad The current gradient of cs.
+   * @param hGrad The gradient of h vector.
+   * @param usePeephole Whether the cell uses peephole connections.
+   * @return a new instance of LSTMBlockCellGrad
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> LSTMBlockCellGrad<T> create(Scope scope, Operand<T> x, Operand<T> csPrev, Operand<T> hPrev, Operand<T> w, Operand<T> wci, Operand<T> wcf, Operand<T> wco, Operand<T> b, Operand<T> i, Operand<T> cs, Operand<T> f, Operand<T> o, Operand<T> ci, Operand<T> co, Operand<T> csGrad, Operand<T> hGrad, Boolean usePeephole) {
+    OperationBuilder opBuilder = scope.env().opBuilder("LSTMBlockCellGrad", scope.makeOpName("LSTMBlockCellGrad"));
+    opBuilder.addInput(x.asOutput());
+    opBuilder.addInput(csPrev.asOutput());
+    opBuilder.addInput(hPrev.asOutput());
+    opBuilder.addInput(w.asOutput());
+    opBuilder.addInput(wci.asOutput());
+    opBuilder.addInput(wcf.asOutput());
+    opBuilder.addInput(wco.asOutput());
+    opBuilder.addInput(b.asOutput());
+    opBuilder.addInput(i.asOutput());
+    opBuilder.addInput(cs.asOutput());
+    opBuilder.addInput(f.asOutput());
+    opBuilder.addInput(o.asOutput());
+    opBuilder.addInput(ci.asOutput());
+    opBuilder.addInput(co.asOutput());
+    opBuilder.addInput(csGrad.asOutput());
+    opBuilder.addInput(hGrad.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("use_peephole", usePeephole);
+    return new LSTMBlockCellGrad<T>(opBuilder.build());
+  }
+  
+  /**
+   * The gradient of cs to be back-propped.
+   */
+  public Output<T> csPrevGrad() {
+    return csPrevGrad;
+  }
+  
+  /**
+   * The derivative wrt to [i, cs, f, o].
+   */
+  public Output<T> dicfo() {
+    return dicfo;
+  }
+  
+  /**
+   * The gradient for wci to be back-propped.
+   */
+  public Output<T> wciGrad() {
+    return wciGrad;
+  }
+  
+  /**
+   * The gradient for wcf to be back-propped.
+   */
+  public Output<T> wcfGrad() {
+    return wcfGrad;
+  }
+  
+  /**
+   * The gradient for wco to be back-propped.
+   */
+  public Output<T> wcoGrad() {
+    return wcoGrad;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "LSTMBlockCellGrad";
+  
+  private Output<T> csPrevGrad;
+  private Output<T> dicfo;
+  private Output<T> wciGrad;
+  private Output<T> wcfGrad;
+  private Output<T> wcoGrad;
+  
+  private LSTMBlockCellGrad(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    csPrevGrad = operation.output(outputIdx++);
+    dicfo = operation.output(outputIdx++);
+    wciGrad = operation.output(outputIdx++);
+    wcfGrad = operation.output(outputIdx++);
+    wcoGrad = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/LeakyRelu.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/LeakyRelu.java
index 8db6282d4c9..96f44eacb5a 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/LeakyRelu.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/LeakyRelu.java
@@ -65,7 +65,7 @@ private Options() {
   @Endpoint(describeByClass = true)
   public static <T extends TNumber> LeakyRelu<T> create(Scope scope, Operand<T> features, Options... options) {
     OperationBuilder opBuilder = scope.env().opBuilder("LeakyRelu", scope.makeOpName("LeakyRelu"));
-    opBuilder.addInput(features.asOutput(scope));
+    opBuilder.addInput(features.asOutput());
     opBuilder = scope.applyControlDependencies(opBuilder);
     if (options != null) {
       for (Options opts : options) {
@@ -91,7 +91,7 @@ public Output<T> activations() {
   }
   
   @Override
-  public Output<T> asOutput(Scope scope) {
+  public Output<T> asOutput() {
     return activations;
   }
   
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/LearnedUnigramCandidateSampler.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/LearnedUnigramCandidateSampler.java
new file mode 100644
index 00000000000..3f850ce672c
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/LearnedUnigramCandidateSampler.java
@@ -0,0 +1,171 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.nn;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.TInt64;
+
+/**
+ * Generates labels for candidate sampling with a learned unigram distribution.
+ * <p>
+ * See explanations of candidate sampling and the data formats at
+ * go/candidate-sampling.
+ * <p>
+ * For each batch, this op picks a single set of sampled candidate labels.
+ * <p>
+ * The advantages of sampling candidates per-batch are simplicity and the
+ * possibility of efficient dense matrix multiplication. The disadvantage is that
+ * the sampled candidates must be chosen independently of the context and of the
+ * true labels.
+ */
+@Operator(group = "nn")
+public final class LearnedUnigramCandidateSampler extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.nn.LearnedUnigramCandidateSampler}
+   */
+  public static class Options {
+    
+    /**
+     * @param seed If either seed or seed2 are set to be non-zero, the random number
+     * generator is seeded by the given seed.  Otherwise, it is seeded by a
+     * random seed.
+     */
+    public Options seed(Long seed) {
+      this.seed = seed;
+      return this;
+    }
+    
+    /**
+     * @param seed2 An second seed to avoid seed collision.
+     */
+    public Options seed2(Long seed2) {
+      this.seed2 = seed2;
+      return this;
+    }
+    
+    private Long seed;
+    private Long seed2;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new LearnedUnigramCandidateSampler operation.
+   * 
+   * @param scope current scope
+   * @param trueClasses A batch_size * num_true matrix, in which each row contains the
+   * IDs of the num_true target_classes in the corresponding original label.
+   * @param numTrue Number of true labels per context.
+   * @param numSampled Number of candidates to randomly sample.
+   * @param unique If unique is true, we sample with rejection, so that all sampled
+   * candidates in a batch are unique. This requires some approximation to
+   * estimate the post-rejection sampling probabilities.
+   * @param rangeMax The sampler will sample integers from the interval [0, range_max).
+   * @param options carries optional attributes values
+   * @return a new instance of LearnedUnigramCandidateSampler
+   */
+  @Endpoint(describeByClass = true)
+  public static LearnedUnigramCandidateSampler create(Scope scope, Operand<TInt64> trueClasses, Long numTrue, Long numSampled, Boolean unique, Long rangeMax, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("LearnedUnigramCandidateSampler", scope.makeOpName("LearnedUnigramCandidateSampler"));
+    opBuilder.addInput(trueClasses.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("num_true", numTrue);
+    opBuilder.setAttr("num_sampled", numSampled);
+    opBuilder.setAttr("unique", unique);
+    opBuilder.setAttr("range_max", rangeMax);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.seed != null) {
+          opBuilder.setAttr("seed", opts.seed);
+        }
+        if (opts.seed2 != null) {
+          opBuilder.setAttr("seed2", opts.seed2);
+        }
+      }
+    }
+    return new LearnedUnigramCandidateSampler(opBuilder.build());
+  }
+  
+  /**
+   * @param seed If either seed or seed2 are set to be non-zero, the random number
+   * generator is seeded by the given seed.  Otherwise, it is seeded by a
+   * random seed.
+   */
+  public static Options seed(Long seed) {
+    return new Options().seed(seed);
+  }
+  
+  /**
+   * @param seed2 An second seed to avoid seed collision.
+   */
+  public static Options seed2(Long seed2) {
+    return new Options().seed2(seed2);
+  }
+  
+  /**
+   * A vector of length num_sampled, in which each element is
+   * the ID of a sampled candidate.
+   */
+  public Output<TInt64> sampledCandidates() {
+    return sampledCandidates;
+  }
+  
+  /**
+   * A batch_size * num_true matrix, representing
+   * the number of times each candidate is expected to occur in a batch
+   * of sampled candidates. If unique=true, then this is a probability.
+   */
+  public Output<TFloat32> trueExpectedCount() {
+    return trueExpectedCount;
+  }
+  
+  /**
+   * A vector of length num_sampled, for each sampled
+   * candidate representing the number of times the candidate is expected
+   * to occur in a batch of sampled candidates.  If unique=true, then this is a
+   * probability.
+   */
+  public Output<TFloat32> sampledExpectedCount() {
+    return sampledExpectedCount;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "LearnedUnigramCandidateSampler";
+  
+  private Output<TInt64> sampledCandidates;
+  private Output<TFloat32> trueExpectedCount;
+  private Output<TFloat32> sampledExpectedCount;
+  
+  private LearnedUnigramCandidateSampler(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    sampledCandidates = operation.output(outputIdx++);
+    trueExpectedCount = operation.output(outputIdx++);
+    sampledExpectedCount = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/LocalResponseNormalization.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/LocalResponseNormalization.java
new file mode 100644
index 00000000000..e7d72196b4a
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/LocalResponseNormalization.java
@@ -0,0 +1,177 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.nn;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Local Response Normalization.
+ * <p>
+ * The 4-D `input` tensor is treated as a 3-D array of 1-D vectors (along the last
+ * dimension), and each vector is normalized independently.  Within a given vector,
+ * each component is divided by the weighted, squared sum of inputs within
+ * `depth_radius`.  In detail,
+ * <p>
+ *     sqr_sum[a, b, c, d] =
+ *         sum(input[a, b, c, d - depth_radius : d + depth_radius + 1] ** 2)
+ *     output = input / (bias + alpha * sqr_sum) ** beta
+ * <p>
+ * For details, see [Krizhevsky et al., ImageNet classification with deep
+ * convolutional neural networks (NIPS 2012)](http://papers.nips.cc/paper/4824-imagenet-classification-with-deep-convolutional-neural-networks).
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator(group = "nn")
+public final class LocalResponseNormalization<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.nn.LocalResponseNormalization}
+   */
+  public static class Options {
+    
+    /**
+     * @param depthRadius 0-D.  Half-width of the 1-D normalization window.
+     */
+    public Options depthRadius(Long depthRadius) {
+      this.depthRadius = depthRadius;
+      return this;
+    }
+    
+    /**
+     * @param bias An offset (usually positive to avoid dividing by 0).
+     */
+    public Options bias(Float bias) {
+      this.bias = bias;
+      return this;
+    }
+    
+    /**
+     * @param alpha A scale factor, usually positive.
+     */
+    public Options alpha(Float alpha) {
+      this.alpha = alpha;
+      return this;
+    }
+    
+    /**
+     * @param beta An exponent.
+     */
+    public Options beta(Float beta) {
+      this.beta = beta;
+      return this;
+    }
+    
+    private Long depthRadius;
+    private Float bias;
+    private Float alpha;
+    private Float beta;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new LocalResponseNormalization operation.
+   * 
+   * @param scope current scope
+   * @param input 4-D.
+   * @param options carries optional attributes values
+   * @return a new instance of LocalResponseNormalization
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> LocalResponseNormalization<T> create(Scope scope, Operand<T> input, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("LRN", scope.makeOpName("LocalResponseNormalization"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.depthRadius != null) {
+          opBuilder.setAttr("depth_radius", opts.depthRadius);
+        }
+        if (opts.bias != null) {
+          opBuilder.setAttr("bias", opts.bias);
+        }
+        if (opts.alpha != null) {
+          opBuilder.setAttr("alpha", opts.alpha);
+        }
+        if (opts.beta != null) {
+          opBuilder.setAttr("beta", opts.beta);
+        }
+      }
+    }
+    return new LocalResponseNormalization<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param depthRadius 0-D.  Half-width of the 1-D normalization window.
+   */
+  public static Options depthRadius(Long depthRadius) {
+    return new Options().depthRadius(depthRadius);
+  }
+  
+  /**
+   * @param bias An offset (usually positive to avoid dividing by 0).
+   */
+  public static Options bias(Float bias) {
+    return new Options().bias(bias);
+  }
+  
+  /**
+   * @param alpha A scale factor, usually positive.
+   */
+  public static Options alpha(Float alpha) {
+    return new Options().alpha(alpha);
+  }
+  
+  /**
+   * @param beta An exponent.
+   */
+  public static Options beta(Float beta) {
+    return new Options().beta(beta);
+  }
+  
+  /**
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "LRN";
+  
+  private Output<T> output;
+  
+  private LocalResponseNormalization(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/LocalResponseNormalizationGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/LocalResponseNormalizationGrad.java
new file mode 100644
index 00000000000..d347bec7237
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/LocalResponseNormalizationGrad.java
@@ -0,0 +1,169 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.nn;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Gradients for Local Response Normalization.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+public final class LocalResponseNormalizationGrad<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.nn.LocalResponseNormalizationGrad}
+   */
+  public static class Options {
+    
+    /**
+     * @param depthRadius A depth radius.
+     */
+    public Options depthRadius(Long depthRadius) {
+      this.depthRadius = depthRadius;
+      return this;
+    }
+    
+    /**
+     * @param bias An offset (usually > 0 to avoid dividing by 0).
+     */
+    public Options bias(Float bias) {
+      this.bias = bias;
+      return this;
+    }
+    
+    /**
+     * @param alpha A scale factor, usually positive.
+     */
+    public Options alpha(Float alpha) {
+      this.alpha = alpha;
+      return this;
+    }
+    
+    /**
+     * @param beta An exponent.
+     */
+    public Options beta(Float beta) {
+      this.beta = beta;
+      return this;
+    }
+    
+    private Long depthRadius;
+    private Float bias;
+    private Float alpha;
+    private Float beta;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new LocalResponseNormalizationGrad operation.
+   * 
+   * @param scope current scope
+   * @param inputGrads 4-D with shape `[batch, height, width, channels]`.
+   * @param inputImage 4-D with shape `[batch, height, width, channels]`.
+   * @param outputImage 4-D with shape `[batch, height, width, channels]`.
+   * @param options carries optional attributes values
+   * @return a new instance of LocalResponseNormalizationGrad
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> LocalResponseNormalizationGrad<T> create(Scope scope, Operand<T> inputGrads, Operand<T> inputImage, Operand<T> outputImage, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("LRNGrad", scope.makeOpName("LocalResponseNormalizationGrad"));
+    opBuilder.addInput(inputGrads.asOutput());
+    opBuilder.addInput(inputImage.asOutput());
+    opBuilder.addInput(outputImage.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.depthRadius != null) {
+          opBuilder.setAttr("depth_radius", opts.depthRadius);
+        }
+        if (opts.bias != null) {
+          opBuilder.setAttr("bias", opts.bias);
+        }
+        if (opts.alpha != null) {
+          opBuilder.setAttr("alpha", opts.alpha);
+        }
+        if (opts.beta != null) {
+          opBuilder.setAttr("beta", opts.beta);
+        }
+      }
+    }
+    return new LocalResponseNormalizationGrad<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param depthRadius A depth radius.
+   */
+  public static Options depthRadius(Long depthRadius) {
+    return new Options().depthRadius(depthRadius);
+  }
+  
+  /**
+   * @param bias An offset (usually > 0 to avoid dividing by 0).
+   */
+  public static Options bias(Float bias) {
+    return new Options().bias(bias);
+  }
+  
+  /**
+   * @param alpha A scale factor, usually positive.
+   */
+  public static Options alpha(Float alpha) {
+    return new Options().alpha(alpha);
+  }
+  
+  /**
+   * @param beta An exponent.
+   */
+  public static Options beta(Float beta) {
+    return new Options().beta(beta);
+  }
+  
+  /**
+   * The gradients for LRN.
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "LRNGrad";
+  
+  private Output<T> output;
+  
+  private LocalResponseNormalizationGrad(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/LogSoftmax.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/LogSoftmax.java
new file mode 100644
index 00000000000..e9e123e516d
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/LogSoftmax.java
@@ -0,0 +1,79 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.nn;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Computes log softmax activations.
+ * <p>
+ * For each batch `i` and class `j` we have
+ * <p>
+ *     logsoftmax[i, j] = logits[i, j] - log(sum(exp(logits[i])))
+ * 
+ * @param <T> data type for {@code logsoftmax()} output
+ */
+@Operator(group = "nn")
+public final class LogSoftmax<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new LogSoftmax operation.
+   * 
+   * @param scope current scope
+   * @param logits 2-D with shape `[batch_size, num_classes]`.
+   * @return a new instance of LogSoftmax
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> LogSoftmax<T> create(Scope scope, Operand<T> logits) {
+    OperationBuilder opBuilder = scope.env().opBuilder("LogSoftmax", scope.makeOpName("LogSoftmax"));
+    opBuilder.addInput(logits.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new LogSoftmax<T>(opBuilder.build());
+  }
+  
+  /**
+   * Same shape as `logits`.
+   */
+  public Output<T> logsoftmax() {
+    return logsoftmax;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return logsoftmax;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "LogSoftmax";
+  
+  private Output<T> logsoftmax;
+  
+  private LogSoftmax(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    logsoftmax = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPool.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPool.java
new file mode 100644
index 00000000000..f7b75951e31
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPool.java
@@ -0,0 +1,125 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.nn;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Performs max pooling on the input.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator(group = "nn")
+public final class MaxPool<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.nn.MaxPool}
+   */
+  public static class Options {
+    
+    /**
+     * @param dataFormat Specify the data format of the input and output data. With the
+     * default format "NHWC", the data is stored in the order of:
+     *     [batch, in_height, in_width, in_channels].
+     * Alternatively, the format could be "NCHW", the data storage order of:
+     *     [batch, in_channels, in_height, in_width].
+     */
+    public Options dataFormat(String dataFormat) {
+      this.dataFormat = dataFormat;
+      return this;
+    }
+    
+    private String dataFormat;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new MaxPool operation.
+   * 
+   * @param scope current scope
+   * @param input 4-D input to pool over.
+   * @param ksize The size of the window for each dimension of the input tensor.
+   * @param strides The stride of the sliding window for each dimension of the
+   * input tensor.
+   * @param padding The type of padding algorithm to use.
+   * @param options carries optional attributes values
+   * @return a new instance of MaxPool
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> MaxPool<T> create(Scope scope, Operand<T> input, Operand<TInt32> ksize, Operand<TInt32> strides, String padding, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("MaxPoolV2", scope.makeOpName("MaxPool"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(ksize.asOutput());
+    opBuilder.addInput(strides.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("padding", padding);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.dataFormat != null) {
+          opBuilder.setAttr("data_format", opts.dataFormat);
+        }
+      }
+    }
+    return new MaxPool<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param dataFormat Specify the data format of the input and output data. With the
+   * default format "NHWC", the data is stored in the order of:
+   *     [batch, in_height, in_width, in_channels].
+   * Alternatively, the format could be "NCHW", the data storage order of:
+   *     [batch, in_channels, in_height, in_width].
+   */
+  public static Options dataFormat(String dataFormat) {
+    return new Options().dataFormat(dataFormat);
+  }
+  
+  /**
+   * The max pooled output tensor.
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "MaxPoolV2";
+  
+  private Output<T> output;
+  
+  private MaxPool(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPool3d.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPool3d.java
new file mode 100644
index 00000000000..320f38a084e
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPool3d.java
@@ -0,0 +1,134 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.nn;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Performs 3D max pooling on the input.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator(group = "nn")
+public final class MaxPool3d<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.nn.MaxPool3d}
+   */
+  public static class Options {
+    
+    /**
+     * @param dataFormat The data format of the input and output data. With the
+     * default format "NDHWC", the data is stored in the order of:
+     *     [batch, in_depth, in_height, in_width, in_channels].
+     * Alternatively, the format could be "NCDHW", the data storage order is:
+     *     [batch, in_channels, in_depth, in_height, in_width].
+     */
+    public Options dataFormat(String dataFormat) {
+      this.dataFormat = dataFormat;
+      return this;
+    }
+    
+    private String dataFormat;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new MaxPool3d operation.
+   * 
+   * @param scope current scope
+   * @param input Shape `[batch, depth, rows, cols, channels]` tensor to pool over.
+   * @param ksize 1-D tensor of length 5. The size of the window for each dimension of
+   * the input tensor. Must have `ksize[0] = ksize[4] = 1`.
+   * @param strides 1-D tensor of length 5. The stride of the sliding window for each
+   * dimension of `input`. Must have `strides[0] = strides[4] = 1`.
+   * @param padding The type of padding algorithm to use.
+   * @param options carries optional attributes values
+   * @return a new instance of MaxPool3d
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> MaxPool3d<T> create(Scope scope, Operand<T> input, List<Long> ksize, List<Long> strides, String padding, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("MaxPool3D", scope.makeOpName("MaxPool3d"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    long[] ksizeArray = new long[ksize.size()];
+    for (int i = 0; i < ksizeArray.length; ++i) {
+      ksizeArray[i] = ksize.get(i);
+    }
+    opBuilder.setAttr("ksize", ksizeArray);
+    long[] stridesArray = new long[strides.size()];
+    for (int i = 0; i < stridesArray.length; ++i) {
+      stridesArray[i] = strides.get(i);
+    }
+    opBuilder.setAttr("strides", stridesArray);
+    opBuilder.setAttr("padding", padding);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.dataFormat != null) {
+          opBuilder.setAttr("data_format", opts.dataFormat);
+        }
+      }
+    }
+    return new MaxPool3d<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param dataFormat The data format of the input and output data. With the
+   * default format "NDHWC", the data is stored in the order of:
+   *     [batch, in_depth, in_height, in_width, in_channels].
+   * Alternatively, the format could be "NCDHW", the data storage order is:
+   *     [batch, in_channels, in_depth, in_height, in_width].
+   */
+  public static Options dataFormat(String dataFormat) {
+    return new Options().dataFormat(dataFormat);
+  }
+  
+  /**
+   * The max pooled output tensor.
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "MaxPool3D";
+  
+  private Output<T> output;
+  
+  private MaxPool3d(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPool3dGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPool3dGrad.java
new file mode 100644
index 00000000000..7f2938e8827
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPool3dGrad.java
@@ -0,0 +1,137 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.nn;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Computes gradients of 3D max pooling function.
+ * 
+ * @param <U> data type for {@code output()} output
+ */
+@Operator(group = "nn")
+public final class MaxPool3dGrad<U extends TNumber> extends RawOp implements Operand<U> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.nn.MaxPool3dGrad}
+   */
+  public static class Options {
+    
+    /**
+     * @param dataFormat The data format of the input and output data. With the
+     * default format "NDHWC", the data is stored in the order of:
+     *     [batch, in_depth, in_height, in_width, in_channels].
+     * Alternatively, the format could be "NCDHW", the data storage order is:
+     *     [batch, in_channels, in_depth, in_height, in_width].
+     */
+    public Options dataFormat(String dataFormat) {
+      this.dataFormat = dataFormat;
+      return this;
+    }
+    
+    private String dataFormat;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new MaxPool3dGrad operation.
+   * 
+   * @param scope current scope
+   * @param origInput The original input tensor.
+   * @param origOutput The original output tensor.
+   * @param grad Output backprop of shape `[batch, depth, rows, cols, channels]`.
+   * @param ksize 1-D tensor of length 5. The size of the window for each dimension of
+   * the input tensor. Must have `ksize[0] = ksize[4] = 1`.
+   * @param strides 1-D tensor of length 5. The stride of the sliding window for each
+   * dimension of `input`. Must have `strides[0] = strides[4] = 1`.
+   * @param padding The type of padding algorithm to use.
+   * @param options carries optional attributes values
+   * @return a new instance of MaxPool3dGrad
+   */
+  @Endpoint(describeByClass = true)
+  public static <U extends TNumber, T extends TNumber> MaxPool3dGrad<U> create(Scope scope, Operand<T> origInput, Operand<T> origOutput, Operand<U> grad, List<Long> ksize, List<Long> strides, String padding, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("MaxPool3DGrad", scope.makeOpName("MaxPool3dGrad"));
+    opBuilder.addInput(origInput.asOutput());
+    opBuilder.addInput(origOutput.asOutput());
+    opBuilder.addInput(grad.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    long[] ksizeArray = new long[ksize.size()];
+    for (int i = 0; i < ksizeArray.length; ++i) {
+      ksizeArray[i] = ksize.get(i);
+    }
+    opBuilder.setAttr("ksize", ksizeArray);
+    long[] stridesArray = new long[strides.size()];
+    for (int i = 0; i < stridesArray.length; ++i) {
+      stridesArray[i] = strides.get(i);
+    }
+    opBuilder.setAttr("strides", stridesArray);
+    opBuilder.setAttr("padding", padding);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.dataFormat != null) {
+          opBuilder.setAttr("data_format", opts.dataFormat);
+        }
+      }
+    }
+    return new MaxPool3dGrad<U>(opBuilder.build());
+  }
+  
+  /**
+   * @param dataFormat The data format of the input and output data. With the
+   * default format "NDHWC", the data is stored in the order of:
+   *     [batch, in_depth, in_height, in_width, in_channels].
+   * Alternatively, the format could be "NCDHW", the data storage order is:
+   *     [batch, in_channels, in_depth, in_height, in_width].
+   */
+  public static Options dataFormat(String dataFormat) {
+    return new Options().dataFormat(dataFormat);
+  }
+  
+  /**
+   */
+  public Output<U> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<U> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "MaxPool3DGrad";
+  
+  private Output<U> output;
+  
+  private MaxPool3dGrad(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPool3dGradGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPool3dGradGrad.java
new file mode 100644
index 00000000000..ef4c546c41e
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPool3dGradGrad.java
@@ -0,0 +1,138 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.nn;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Computes second-order gradients of the maxpooling function.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator(group = "nn")
+public final class MaxPool3dGradGrad<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.nn.MaxPool3dGradGrad}
+   */
+  public static class Options {
+    
+    /**
+     * @param dataFormat The data format of the input and output data. With the
+     * default format "NDHWC", the data is stored in the order of:
+     *     [batch, in_depth, in_height, in_width, in_channels].
+     * Alternatively, the format could be "NCDHW", the data storage order is:
+     *     [batch, in_channels, in_depth, in_height, in_width].
+     */
+    public Options dataFormat(String dataFormat) {
+      this.dataFormat = dataFormat;
+      return this;
+    }
+    
+    private String dataFormat;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new MaxPool3dGradGrad operation.
+   * 
+   * @param scope current scope
+   * @param origInput The original input tensor.
+   * @param origOutput The original output tensor.
+   * @param grad Output backprop of shape `[batch, depth, rows, cols, channels]`.
+   * @param ksize 1-D tensor of length 5. The size of the window for each dimension of
+   * the input tensor. Must have `ksize[0] = ksize[4] = 1`.
+   * @param strides 1-D tensor of length 5. The stride of the sliding window for each
+   * dimension of `input`. Must have `strides[0] = strides[4] = 1`.
+   * @param padding The type of padding algorithm to use.
+   * @param options carries optional attributes values
+   * @return a new instance of MaxPool3dGradGrad
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> MaxPool3dGradGrad<T> create(Scope scope, Operand<T> origInput, Operand<T> origOutput, Operand<T> grad, List<Long> ksize, List<Long> strides, String padding, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("MaxPool3DGradGrad", scope.makeOpName("MaxPool3dGradGrad"));
+    opBuilder.addInput(origInput.asOutput());
+    opBuilder.addInput(origOutput.asOutput());
+    opBuilder.addInput(grad.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    long[] ksizeArray = new long[ksize.size()];
+    for (int i = 0; i < ksizeArray.length; ++i) {
+      ksizeArray[i] = ksize.get(i);
+    }
+    opBuilder.setAttr("ksize", ksizeArray);
+    long[] stridesArray = new long[strides.size()];
+    for (int i = 0; i < stridesArray.length; ++i) {
+      stridesArray[i] = strides.get(i);
+    }
+    opBuilder.setAttr("strides", stridesArray);
+    opBuilder.setAttr("padding", padding);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.dataFormat != null) {
+          opBuilder.setAttr("data_format", opts.dataFormat);
+        }
+      }
+    }
+    return new MaxPool3dGradGrad<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param dataFormat The data format of the input and output data. With the
+   * default format "NDHWC", the data is stored in the order of:
+   *     [batch, in_depth, in_height, in_width, in_channels].
+   * Alternatively, the format could be "NCDHW", the data storage order is:
+   *     [batch, in_channels, in_depth, in_height, in_width].
+   */
+  public static Options dataFormat(String dataFormat) {
+    return new Options().dataFormat(dataFormat);
+  }
+  
+  /**
+   * Gradients of gradients w.r.t. the input to `max_pool`.
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "MaxPool3DGradGrad";
+  
+  private Output<T> output;
+  
+  private MaxPool3dGradGrad(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPoolGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPoolGrad.java
new file mode 100644
index 00000000000..e391585b657
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPoolGrad.java
@@ -0,0 +1,129 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.nn;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Computes gradients of the maxpooling function.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator(group = "nn")
+public final class MaxPoolGrad<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.nn.MaxPoolGrad}
+   */
+  public static class Options {
+    
+    /**
+     * @param dataFormat Specify the data format of the input and output data. With the
+     * default format "NHWC", the data is stored in the order of:
+     *     [batch, in_height, in_width, in_channels].
+     * Alternatively, the format could be "NCHW", the data storage order of:
+     *     [batch, in_channels, in_height, in_width].
+     */
+    public Options dataFormat(String dataFormat) {
+      this.dataFormat = dataFormat;
+      return this;
+    }
+    
+    private String dataFormat;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new MaxPoolGrad operation.
+   * 
+   * @param scope current scope
+   * @param origInput The original input tensor.
+   * @param origOutput The original output tensor.
+   * @param grad 4-D.  Gradients w.r.t. the output of `max_pool`.
+   * @param ksize The size of the window for each dimension of the input tensor.
+   * @param strides The stride of the sliding window for each dimension of the
+   * input tensor.
+   * @param padding The type of padding algorithm to use.
+   * @param options carries optional attributes values
+   * @return a new instance of MaxPoolGrad
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> MaxPoolGrad<T> create(Scope scope, Operand<T> origInput, Operand<T> origOutput, Operand<T> grad, Operand<TInt32> ksize, Operand<TInt32> strides, String padding, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("MaxPoolGradV2", scope.makeOpName("MaxPoolGrad"));
+    opBuilder.addInput(origInput.asOutput());
+    opBuilder.addInput(origOutput.asOutput());
+    opBuilder.addInput(grad.asOutput());
+    opBuilder.addInput(ksize.asOutput());
+    opBuilder.addInput(strides.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("padding", padding);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.dataFormat != null) {
+          opBuilder.setAttr("data_format", opts.dataFormat);
+        }
+      }
+    }
+    return new MaxPoolGrad<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param dataFormat Specify the data format of the input and output data. With the
+   * default format "NHWC", the data is stored in the order of:
+   *     [batch, in_height, in_width, in_channels].
+   * Alternatively, the format could be "NCHW", the data storage order of:
+   *     [batch, in_channels, in_height, in_width].
+   */
+  public static Options dataFormat(String dataFormat) {
+    return new Options().dataFormat(dataFormat);
+  }
+  
+  /**
+   * Gradients w.r.t. the input to `max_pool`.
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "MaxPoolGradV2";
+  
+  private Output<T> output;
+  
+  private MaxPoolGrad(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPoolGradGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPoolGradGrad.java
new file mode 100644
index 00000000000..2653f1d51c2
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPoolGradGrad.java
@@ -0,0 +1,129 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.nn;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Computes second-order gradients of the maxpooling function.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator(group = "nn")
+public final class MaxPoolGradGrad<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.nn.MaxPoolGradGrad}
+   */
+  public static class Options {
+    
+    /**
+     * @param dataFormat Specify the data format of the input and output data. With the
+     * default format "NHWC", the data is stored in the order of:
+     *     [batch, in_height, in_width, in_channels].
+     * Alternatively, the format could be "NCHW", the data storage order of:
+     *     [batch, in_channels, in_height, in_width].
+     */
+    public Options dataFormat(String dataFormat) {
+      this.dataFormat = dataFormat;
+      return this;
+    }
+    
+    private String dataFormat;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new MaxPoolGradGrad operation.
+   * 
+   * @param scope current scope
+   * @param origInput The original input tensor.
+   * @param origOutput The original output tensor.
+   * @param grad 4-D.  Gradients of gradients w.r.t. the input of `max_pool`.
+   * @param ksize The size of the window for each dimension of the input tensor.
+   * @param strides The stride of the sliding window for each dimension of the
+   * input tensor.
+   * @param padding The type of padding algorithm to use.
+   * @param options carries optional attributes values
+   * @return a new instance of MaxPoolGradGrad
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> MaxPoolGradGrad<T> create(Scope scope, Operand<T> origInput, Operand<T> origOutput, Operand<T> grad, Operand<TInt32> ksize, Operand<TInt32> strides, String padding, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("MaxPoolGradGradV2", scope.makeOpName("MaxPoolGradGrad"));
+    opBuilder.addInput(origInput.asOutput());
+    opBuilder.addInput(origOutput.asOutput());
+    opBuilder.addInput(grad.asOutput());
+    opBuilder.addInput(ksize.asOutput());
+    opBuilder.addInput(strides.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("padding", padding);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.dataFormat != null) {
+          opBuilder.setAttr("data_format", opts.dataFormat);
+        }
+      }
+    }
+    return new MaxPoolGradGrad<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param dataFormat Specify the data format of the input and output data. With the
+   * default format "NHWC", the data is stored in the order of:
+   *     [batch, in_height, in_width, in_channels].
+   * Alternatively, the format could be "NCHW", the data storage order of:
+   *     [batch, in_channels, in_height, in_width].
+   */
+  public static Options dataFormat(String dataFormat) {
+    return new Options().dataFormat(dataFormat);
+  }
+  
+  /**
+   * Gradients of gradients w.r.t. the input to `max_pool`.
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "MaxPoolGradGradV2";
+  
+  private Output<T> output;
+  
+  private MaxPoolGradGrad(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPoolGradGradWithArgmax.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPoolGradGradWithArgmax.java
new file mode 100644
index 00000000000..a15222e7c4f
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPoolGradGradWithArgmax.java
@@ -0,0 +1,130 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.nn;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Computes second-order gradients of the maxpooling function.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator(group = "nn")
+public final class MaxPoolGradGradWithArgmax<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.nn.MaxPoolGradGradWithArgmax}
+   */
+  public static class Options {
+    
+    /**
+     * @param includeBatchInIndex Whether to include batch dimension in flattened index of `argmax`.
+     */
+    public Options includeBatchInIndex(Boolean includeBatchInIndex) {
+      this.includeBatchInIndex = includeBatchInIndex;
+      return this;
+    }
+    
+    private Boolean includeBatchInIndex;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new MaxPoolGradGradWithArgmax operation.
+   * 
+   * @param scope current scope
+   * @param input The original input.
+   * @param grad 4-D with shape `[batch, height, width, channels]`.  Gradients w.r.t. the
+   * input of `max_pool`.
+   * @param argmax The indices of the maximum values chosen for each output of `max_pool`.
+   * @param ksize The size of the window for each dimension of the input tensor.
+   * @param strides The stride of the sliding window for each dimension of the
+   * input tensor.
+   * @param padding The type of padding algorithm to use.
+   * @param options carries optional attributes values
+   * @return a new instance of MaxPoolGradGradWithArgmax
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber, U extends TNumber> MaxPoolGradGradWithArgmax<T> create(Scope scope, Operand<T> input, Operand<T> grad, Operand<U> argmax, List<Long> ksize, List<Long> strides, String padding, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("MaxPoolGradGradWithArgmax", scope.makeOpName("MaxPoolGradGradWithArgmax"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(grad.asOutput());
+    opBuilder.addInput(argmax.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    long[] ksizeArray = new long[ksize.size()];
+    for (int i = 0; i < ksizeArray.length; ++i) {
+      ksizeArray[i] = ksize.get(i);
+    }
+    opBuilder.setAttr("ksize", ksizeArray);
+    long[] stridesArray = new long[strides.size()];
+    for (int i = 0; i < stridesArray.length; ++i) {
+      stridesArray[i] = strides.get(i);
+    }
+    opBuilder.setAttr("strides", stridesArray);
+    opBuilder.setAttr("padding", padding);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.includeBatchInIndex != null) {
+          opBuilder.setAttr("include_batch_in_index", opts.includeBatchInIndex);
+        }
+      }
+    }
+    return new MaxPoolGradGradWithArgmax<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param includeBatchInIndex Whether to include batch dimension in flattened index of `argmax`.
+   */
+  public static Options includeBatchInIndex(Boolean includeBatchInIndex) {
+    return new Options().includeBatchInIndex(includeBatchInIndex);
+  }
+  
+  /**
+   * Gradients of gradients w.r.t. the input of `max_pool`.
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "MaxPoolGradGradWithArgmax";
+  
+  private Output<T> output;
+  
+  private MaxPoolGradGradWithArgmax(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPoolGradWithArgmax.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPoolGradWithArgmax.java
new file mode 100644
index 00000000000..c51275f68f2
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPoolGradWithArgmax.java
@@ -0,0 +1,129 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.nn;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Computes gradients of the maxpooling function.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+public final class MaxPoolGradWithArgmax<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.nn.MaxPoolGradWithArgmax}
+   */
+  public static class Options {
+    
+    /**
+     * @param includeBatchInIndex Whether to include batch dimension in flattened index of `argmax`.
+     */
+    public Options includeBatchInIndex(Boolean includeBatchInIndex) {
+      this.includeBatchInIndex = includeBatchInIndex;
+      return this;
+    }
+    
+    private Boolean includeBatchInIndex;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new MaxPoolGradWithArgmax operation.
+   * 
+   * @param scope current scope
+   * @param input The original input.
+   * @param grad 4-D with shape `[batch, height, width, channels]`.  Gradients w.r.t. the
+   * output of `max_pool`.
+   * @param argmax The indices of the maximum values chosen for each output of `max_pool`.
+   * @param ksize The size of the window for each dimension of the input tensor.
+   * @param strides The stride of the sliding window for each dimension of the
+   * input tensor.
+   * @param padding The type of padding algorithm to use.
+   * @param options carries optional attributes values
+   * @return a new instance of MaxPoolGradWithArgmax
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber, U extends TNumber> MaxPoolGradWithArgmax<T> create(Scope scope, Operand<T> input, Operand<T> grad, Operand<U> argmax, List<Long> ksize, List<Long> strides, String padding, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("MaxPoolGradWithArgmax", scope.makeOpName("MaxPoolGradWithArgmax"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(grad.asOutput());
+    opBuilder.addInput(argmax.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    long[] ksizeArray = new long[ksize.size()];
+    for (int i = 0; i < ksizeArray.length; ++i) {
+      ksizeArray[i] = ksize.get(i);
+    }
+    opBuilder.setAttr("ksize", ksizeArray);
+    long[] stridesArray = new long[strides.size()];
+    for (int i = 0; i < stridesArray.length; ++i) {
+      stridesArray[i] = strides.get(i);
+    }
+    opBuilder.setAttr("strides", stridesArray);
+    opBuilder.setAttr("padding", padding);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.includeBatchInIndex != null) {
+          opBuilder.setAttr("include_batch_in_index", opts.includeBatchInIndex);
+        }
+      }
+    }
+    return new MaxPoolGradWithArgmax<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param includeBatchInIndex Whether to include batch dimension in flattened index of `argmax`.
+   */
+  public static Options includeBatchInIndex(Boolean includeBatchInIndex) {
+    return new Options().includeBatchInIndex(includeBatchInIndex);
+  }
+  
+  /**
+   * Gradients w.r.t. the input of `max_pool`.
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "MaxPoolGradWithArgmax";
+  
+  private Output<T> output;
+  
+  private MaxPoolGradWithArgmax(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPoolWithArgmax.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPoolWithArgmax.java
new file mode 100644
index 00000000000..841e598ee3a
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/MaxPoolWithArgmax.java
@@ -0,0 +1,160 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.nn;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Performs max pooling on the input and outputs both max values and indices.
+ * <p>
+ * The indices in `argmax` are flattened, so that a maximum value at position
+ * `[b, y, x, c]` becomes flattened index:
+ * `(y * width + x) * channels + c` if `include_batch_in_index` is False;
+ * `((b * height + y) * width + x) * channels + c` if `include_batch_in_index` is True.
+ * <p>
+ * The indices returned are always in `[0, height) x [0, width)` before flattening,
+ * even if padding is involved and the mathematically correct answer is outside
+ * (either negative or too large).  This is a bug, but fixing it is difficult to do
+ * in a safe backwards compatible way, especially due to flattening.
+ * 
+ * @param <T> data type for {@code output()} output
+ * @param <U> data type for {@code argmax()} output
+ */
+@Operator(group = "nn")
+public final class MaxPoolWithArgmax<T extends TNumber, U extends TNumber> extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.nn.MaxPoolWithArgmax}
+   */
+  public static class Options {
+    
+    /**
+     * @param includeBatchInIndex Whether to include batch dimension in flattened index of `argmax`.
+     */
+    public Options includeBatchInIndex(Boolean includeBatchInIndex) {
+      this.includeBatchInIndex = includeBatchInIndex;
+      return this;
+    }
+    
+    private Boolean includeBatchInIndex;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new MaxPoolWithArgmax operation.
+   * 
+   * @param scope current scope
+   * @param input 4-D with shape `[batch, height, width, channels]`.  Input to pool over.
+   * @param ksize The size of the window for each dimension of the input tensor.
+   * @param strides The stride of the sliding window for each dimension of the
+   * input tensor.
+   * @param Targmax 
+   * @param padding The type of padding algorithm to use.
+   * @param options carries optional attributes values
+   * @return a new instance of MaxPoolWithArgmax
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber, U extends TNumber> MaxPoolWithArgmax<T, U> create(Scope scope, Operand<T> input, List<Long> ksize, List<Long> strides, Class<U> Targmax, String padding, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("MaxPoolWithArgmax", scope.makeOpName("MaxPoolWithArgmax"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    long[] ksizeArray = new long[ksize.size()];
+    for (int i = 0; i < ksizeArray.length; ++i) {
+      ksizeArray[i] = ksize.get(i);
+    }
+    opBuilder.setAttr("ksize", ksizeArray);
+    long[] stridesArray = new long[strides.size()];
+    for (int i = 0; i < stridesArray.length; ++i) {
+      stridesArray[i] = strides.get(i);
+    }
+    opBuilder.setAttr("strides", stridesArray);
+    opBuilder.setAttr("Targmax", Targmax);
+    opBuilder.setAttr("padding", padding);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.includeBatchInIndex != null) {
+          opBuilder.setAttr("include_batch_in_index", opts.includeBatchInIndex);
+        }
+      }
+    }
+    return new MaxPoolWithArgmax<T, U>(opBuilder.build());
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new MaxPoolWithArgmax operation using default output types.
+   * 
+   * @param scope current scope
+   * @param input 4-D with shape `[batch, height, width, channels]`.  Input to pool over.
+   * @param ksize The size of the window for each dimension of the input tensor.
+   * @param strides The stride of the sliding window for each dimension of the
+   * input tensor.
+   * @param padding The type of padding algorithm to use.
+   * @param options carries optional attributes values
+   * @return a new instance of MaxPoolWithArgmax
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> MaxPoolWithArgmax<T, TInt64> create(Scope scope, Operand<T> input, List<Long> ksize, List<Long> strides, String padding, Options... options) {
+    return create(scope, input, ksize, strides, TInt64.class, padding, options);
+  }
+  
+  /**
+   * @param includeBatchInIndex Whether to include batch dimension in flattened index of `argmax`.
+   */
+  public static Options includeBatchInIndex(Boolean includeBatchInIndex) {
+    return new Options().includeBatchInIndex(includeBatchInIndex);
+  }
+  
+  /**
+   * The max pooled output tensor.
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  /**
+   * 4-D.  The flattened indices of the max values chosen for each output.
+   */
+  public Output<U> argmax() {
+    return argmax;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "MaxPoolWithArgmax";
+  
+  private Output<T> output;
+  private Output<U> argmax;
+  
+  private MaxPoolWithArgmax(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+    argmax = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/NthElement.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/NthElement.java
new file mode 100644
index 00000000000..35f827f5d01
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/NthElement.java
@@ -0,0 +1,123 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.nn;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Finds values of the `n`-th order statistic for the last dimension.
+ * <p>
+ * If the input is a vector (rank-1), finds the entries which is the nth-smallest
+ * value in the vector and outputs their values as scalar tensor.
+ * <p>
+ * For matrices (resp. higher rank input), computes the entries which is the
+ * nth-smallest value in each row (resp. vector along the last dimension). Thus,
+ * <p>
+ *     values.shape = input.shape[:-1]
+ * 
+ * @param <T> data type for {@code values()} output
+ */
+@Operator(group = "nn")
+public final class NthElement<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.nn.NthElement}
+   */
+  public static class Options {
+    
+    /**
+     * @param reverse When set to True, find the nth-largest value in the vector and vice
+     * versa.
+     */
+    public Options reverse(Boolean reverse) {
+      this.reverse = reverse;
+      return this;
+    }
+    
+    private Boolean reverse;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new NthElement operation.
+   * 
+   * @param scope current scope
+   * @param input 1-D or higher with last dimension at least `n+1`.
+   * @param n 0-D. Position of sorted vector to select along the last dimension (along
+   * each row for matrices). Valid range of n is `[0, input.shape[:-1])`
+   * @param options carries optional attributes values
+   * @return a new instance of NthElement
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> NthElement<T> create(Scope scope, Operand<T> input, Operand<TInt32> n, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("NthElement", scope.makeOpName("NthElement"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(n.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.reverse != null) {
+          opBuilder.setAttr("reverse", opts.reverse);
+        }
+      }
+    }
+    return new NthElement<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param reverse When set to True, find the nth-largest value in the vector and vice
+   * versa.
+   */
+  public static Options reverse(Boolean reverse) {
+    return new Options().reverse(reverse);
+  }
+  
+  /**
+   * The `n`-th order statistic along each last dimensional slice.
+   */
+  public Output<T> values() {
+    return values;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return values;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "NthElement";
+  
+  private Output<T> values;
+  
+  private NthElement(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    values = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedAvgPool.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedAvgPool.java
new file mode 100644
index 00000000000..0da9f10b1ef
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedAvgPool.java
@@ -0,0 +1,109 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.nn;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Produces the average pool of the input tensor for quantized types.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator(group = "nn")
+public final class QuantizedAvgPool<T extends TType> extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new QuantizedAvgPool operation.
+   * 
+   * @param scope current scope
+   * @param input 4-D with shape `[batch, height, width, channels]`.
+   * @param minInput The float value that the lowest quantized input value represents.
+   * @param maxInput The float value that the highest quantized input value represents.
+   * @param ksize The size of the window for each dimension of the input tensor.
+   * The length must be 4 to match the number of dimensions of the input.
+   * @param strides The stride of the sliding window for each dimension of the input
+   * tensor.  The length must be 4 to match the number of dimensions of the input.
+   * @param padding The type of padding algorithm to use.
+   * @return a new instance of QuantizedAvgPool
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> QuantizedAvgPool<T> create(Scope scope, Operand<T> input, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, List<Long> ksize, List<Long> strides, String padding) {
+    OperationBuilder opBuilder = scope.env().opBuilder("QuantizedAvgPool", scope.makeOpName("QuantizedAvgPool"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(minInput.asOutput());
+    opBuilder.addInput(maxInput.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    long[] ksizeArray = new long[ksize.size()];
+    for (int i = 0; i < ksizeArray.length; ++i) {
+      ksizeArray[i] = ksize.get(i);
+    }
+    opBuilder.setAttr("ksize", ksizeArray);
+    long[] stridesArray = new long[strides.size()];
+    for (int i = 0; i < stridesArray.length; ++i) {
+      stridesArray[i] = strides.get(i);
+    }
+    opBuilder.setAttr("strides", stridesArray);
+    opBuilder.setAttr("padding", padding);
+    return new QuantizedAvgPool<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  /**
+   * The float value that the lowest quantized output value represents.
+   */
+  public Output<TFloat32> minOutput() {
+    return minOutput;
+  }
+  
+  /**
+   * The float value that the highest quantized output value represents.
+   */
+  public Output<TFloat32> maxOutput() {
+    return maxOutput;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "QuantizedAvgPool";
+  
+  private Output<T> output;
+  private Output<TFloat32> minOutput;
+  private Output<TFloat32> maxOutput;
+  
+  private QuantizedAvgPool(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+    minOutput = operation.output(outputIdx++);
+    maxOutput = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedBatchNormWithGlobalNormalization.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedBatchNormWithGlobalNormalization.java
new file mode 100644
index 00000000000..57b2af3a66d
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedBatchNormWithGlobalNormalization.java
@@ -0,0 +1,131 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.nn;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Quantized Batch normalization.
+ * <p>
+ * This op is deprecated and will be removed in the future. Prefer
+ * `tf.nn.batch_normalization`.
+ * 
+ * @param <U> data type for {@code result()} output
+ */
+@Operator(group = "nn")
+public final class QuantizedBatchNormWithGlobalNormalization<U extends TType> extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new QuantizedBatchNormWithGlobalNormalization operation.
+   * 
+   * @param scope current scope
+   * @param t A 4D input Tensor.
+   * @param tMin The value represented by the lowest quantized input.
+   * @param tMax The value represented by the highest quantized input.
+   * @param m A 1D mean Tensor with size matching the last dimension of t.
+   * This is the first output from tf.nn.moments,
+   * or a saved moving average thereof.
+   * @param mMin The value represented by the lowest quantized mean.
+   * @param mMax The value represented by the highest quantized mean.
+   * @param v A 1D variance Tensor with size matching the last dimension of t.
+   * This is the second output from tf.nn.moments,
+   * or a saved moving average thereof.
+   * @param vMin The value represented by the lowest quantized variance.
+   * @param vMax The value represented by the highest quantized variance.
+   * @param beta A 1D beta Tensor with size matching the last dimension of t.
+   * An offset to be added to the normalized tensor.
+   * @param betaMin The value represented by the lowest quantized offset.
+   * @param betaMax The value represented by the highest quantized offset.
+   * @param gamma A 1D gamma Tensor with size matching the last dimension of t.
+   * If "scale_after_normalization" is true, this tensor will be multiplied
+   * with the normalized tensor.
+   * @param gammaMin The value represented by the lowest quantized gamma.
+   * @param gammaMax The value represented by the highest quantized gamma.
+   * @param outType 
+   * @param varianceEpsilon A small float number to avoid dividing by 0.
+   * @param scaleAfterNormalization A bool indicating whether the resulted tensor
+   * needs to be multiplied with gamma.
+   * @return a new instance of QuantizedBatchNormWithGlobalNormalization
+   */
+  @Endpoint(describeByClass = true)
+  public static <U extends TType, T extends TType> QuantizedBatchNormWithGlobalNormalization<U> create(Scope scope, Operand<T> t, Operand<TFloat32> tMin, Operand<TFloat32> tMax, Operand<T> m, Operand<TFloat32> mMin, Operand<TFloat32> mMax, Operand<T> v, Operand<TFloat32> vMin, Operand<TFloat32> vMax, Operand<T> beta, Operand<TFloat32> betaMin, Operand<TFloat32> betaMax, Operand<T> gamma, Operand<TFloat32> gammaMin, Operand<TFloat32> gammaMax, Class<U> outType, Float varianceEpsilon, Boolean scaleAfterNormalization) {
+    OperationBuilder opBuilder = scope.env().opBuilder("QuantizedBatchNormWithGlobalNormalization", scope.makeOpName("QuantizedBatchNormWithGlobalNormalization"));
+    opBuilder.addInput(t.asOutput());
+    opBuilder.addInput(tMin.asOutput());
+    opBuilder.addInput(tMax.asOutput());
+    opBuilder.addInput(m.asOutput());
+    opBuilder.addInput(mMin.asOutput());
+    opBuilder.addInput(mMax.asOutput());
+    opBuilder.addInput(v.asOutput());
+    opBuilder.addInput(vMin.asOutput());
+    opBuilder.addInput(vMax.asOutput());
+    opBuilder.addInput(beta.asOutput());
+    opBuilder.addInput(betaMin.asOutput());
+    opBuilder.addInput(betaMax.asOutput());
+    opBuilder.addInput(gamma.asOutput());
+    opBuilder.addInput(gammaMin.asOutput());
+    opBuilder.addInput(gammaMax.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("out_type", outType);
+    opBuilder.setAttr("variance_epsilon", varianceEpsilon);
+    opBuilder.setAttr("scale_after_normalization", scaleAfterNormalization);
+    return new QuantizedBatchNormWithGlobalNormalization<U>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<U> result() {
+    return result;
+  }
+  
+  /**
+   */
+  public Output<TFloat32> resultMin() {
+    return resultMin;
+  }
+  
+  /**
+   */
+  public Output<TFloat32> resultMax() {
+    return resultMax;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "QuantizedBatchNormWithGlobalNormalization";
+  
+  private Output<U> result;
+  private Output<TFloat32> resultMin;
+  private Output<TFloat32> resultMax;
+  
+  private QuantizedBatchNormWithGlobalNormalization(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    result = operation.output(outputIdx++);
+    resultMin = operation.output(outputIdx++);
+    resultMax = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedBiasAdd.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedBiasAdd.java
new file mode 100644
index 00000000000..b39db777a05
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedBiasAdd.java
@@ -0,0 +1,102 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.nn;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Adds Tensor 'bias' to Tensor 'input' for Quantized types.
+ * <p>
+ * Broadcasts the values of bias on dimensions 0..N-2 of 'input'.
+ * 
+ * @param <V> data type for {@code output()} output
+ */
+@Operator(group = "nn")
+public final class QuantizedBiasAdd<V extends TType> extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new QuantizedBiasAdd operation.
+   * 
+   * @param scope current scope
+   * @param input 
+   * @param bias A 1D bias Tensor with size matching the last dimension of 'input'.
+   * @param minInput The float value that the lowest quantized input value represents.
+   * @param maxInput The float value that the highest quantized input value represents.
+   * @param minBias The float value that the lowest quantized bias value represents.
+   * @param maxBias The float value that the highest quantized bias value represents.
+   * @param outType 
+   * @return a new instance of QuantizedBiasAdd
+   */
+  @Endpoint(describeByClass = true)
+  public static <V extends TType, T extends TType, U extends TType> QuantizedBiasAdd<V> create(Scope scope, Operand<T> input, Operand<U> bias, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, Operand<TFloat32> minBias, Operand<TFloat32> maxBias, Class<V> outType) {
+    OperationBuilder opBuilder = scope.env().opBuilder("QuantizedBiasAdd", scope.makeOpName("QuantizedBiasAdd"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(bias.asOutput());
+    opBuilder.addInput(minInput.asOutput());
+    opBuilder.addInput(maxInput.asOutput());
+    opBuilder.addInput(minBias.asOutput());
+    opBuilder.addInput(maxBias.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("out_type", outType);
+    return new QuantizedBiasAdd<V>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<V> output() {
+    return output;
+  }
+  
+  /**
+   * The float value that the lowest quantized output value represents.
+   */
+  public Output<TFloat32> minOut() {
+    return minOut;
+  }
+  
+  /**
+   * The float value that the highest quantized output value represents.
+   */
+  public Output<TFloat32> maxOut() {
+    return maxOut;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "QuantizedBiasAdd";
+  
+  private Output<V> output;
+  private Output<TFloat32> minOut;
+  private Output<TFloat32> maxOut;
+  
+  private QuantizedBiasAdd(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+    minOut = operation.output(outputIdx++);
+    maxOut = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DAndRelu.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DAndRelu.java
new file mode 100644
index 00000000000..362a64d595e
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DAndRelu.java
@@ -0,0 +1,165 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.nn;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.family.TType;
+
+/**
+ * @param <V> data type for {@code output()} output
+ */
+public final class QuantizedConv2DAndRelu<V extends TType> extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.nn.QuantizedConv2DAndRelu}
+   */
+  public static class Options {
+    
+    /**
+     * @param dilations 
+     */
+    public Options dilations(List<Long> dilations) {
+      this.dilations = dilations;
+      return this;
+    }
+    
+    /**
+     * @param paddingList 
+     */
+    public Options paddingList(List<Long> paddingList) {
+      this.paddingList = paddingList;
+      return this;
+    }
+    
+    private List<Long> dilations;
+    private List<Long> paddingList;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new QuantizedConv2DAndRelu operation.
+   * 
+   * @param scope current scope
+   * @param input 
+   * @param filter 
+   * @param minInput 
+   * @param maxInput 
+   * @param minFilter 
+   * @param maxFilter 
+   * @param outType 
+   * @param strides 
+   * @param padding 
+   * @param options carries optional attributes values
+   * @return a new instance of QuantizedConv2DAndRelu
+   */
+  @Endpoint(describeByClass = true)
+  public static <V extends TType, T extends TType, U extends TType> QuantizedConv2DAndRelu<V> create(Scope scope, Operand<T> input, Operand<U> filter, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, Operand<TFloat32> minFilter, Operand<TFloat32> maxFilter, Class<V> outType, List<Long> strides, String padding, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("QuantizedConv2DAndRelu", scope.makeOpName("QuantizedConv2DAndRelu"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(filter.asOutput());
+    opBuilder.addInput(minInput.asOutput());
+    opBuilder.addInput(maxInput.asOutput());
+    opBuilder.addInput(minFilter.asOutput());
+    opBuilder.addInput(maxFilter.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("out_type", outType);
+    long[] stridesArray = new long[strides.size()];
+    for (int i = 0; i < stridesArray.length; ++i) {
+      stridesArray[i] = strides.get(i);
+    }
+    opBuilder.setAttr("strides", stridesArray);
+    opBuilder.setAttr("padding", padding);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.dilations != null) {
+          long[] dilationsArray = new long[opts.dilations.size()];
+          for (int i = 0; i < dilationsArray.length; ++i) {
+            dilationsArray[i] = opts.dilations.get(i);
+          }
+          opBuilder.setAttr("dilations", dilationsArray);
+        }
+        if (opts.paddingList != null) {
+          long[] paddingListArray = new long[opts.paddingList.size()];
+          for (int i = 0; i < paddingListArray.length; ++i) {
+            paddingListArray[i] = opts.paddingList.get(i);
+          }
+          opBuilder.setAttr("padding_list", paddingListArray);
+        }
+      }
+    }
+    return new QuantizedConv2DAndRelu<V>(opBuilder.build());
+  }
+  
+  /**
+   * @param dilations 
+   */
+  public static Options dilations(List<Long> dilations) {
+    return new Options().dilations(dilations);
+  }
+  
+  /**
+   * @param paddingList 
+   */
+  public static Options paddingList(List<Long> paddingList) {
+    return new Options().paddingList(paddingList);
+  }
+  
+  /**
+   */
+  public Output<V> output() {
+    return output;
+  }
+  
+  /**
+   */
+  public Output<TFloat32> minOutput() {
+    return minOutput;
+  }
+  
+  /**
+   */
+  public Output<TFloat32> maxOutput() {
+    return maxOutput;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "QuantizedConv2DAndRelu";
+  
+  private Output<V> output;
+  private Output<TFloat32> minOutput;
+  private Output<TFloat32> maxOutput;
+  
+  private QuantizedConv2DAndRelu(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+    minOutput = operation.output(outputIdx++);
+    maxOutput = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DAndReluAndRequantize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DAndReluAndRequantize.java
new file mode 100644
index 00000000000..f55cae7b48f
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DAndReluAndRequantize.java
@@ -0,0 +1,169 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.nn;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.family.TType;
+
+/**
+ * @param <V> data type for {@code output()} output
+ */
+public final class QuantizedConv2DAndReluAndRequantize<V extends TType> extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.nn.QuantizedConv2DAndReluAndRequantize}
+   */
+  public static class Options {
+    
+    /**
+     * @param dilations 
+     */
+    public Options dilations(List<Long> dilations) {
+      this.dilations = dilations;
+      return this;
+    }
+    
+    /**
+     * @param paddingList 
+     */
+    public Options paddingList(List<Long> paddingList) {
+      this.paddingList = paddingList;
+      return this;
+    }
+    
+    private List<Long> dilations;
+    private List<Long> paddingList;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new QuantizedConv2DAndReluAndRequantize operation.
+   * 
+   * @param scope current scope
+   * @param input 
+   * @param filter 
+   * @param minInput 
+   * @param maxInput 
+   * @param minFilter 
+   * @param maxFilter 
+   * @param minFreezedOutput 
+   * @param maxFreezedOutput 
+   * @param outType 
+   * @param strides 
+   * @param padding 
+   * @param options carries optional attributes values
+   * @return a new instance of QuantizedConv2DAndReluAndRequantize
+   */
+  @Endpoint(describeByClass = true)
+  public static <V extends TType, T extends TType, U extends TType> QuantizedConv2DAndReluAndRequantize<V> create(Scope scope, Operand<T> input, Operand<U> filter, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, Operand<TFloat32> minFilter, Operand<TFloat32> maxFilter, Operand<TFloat32> minFreezedOutput, Operand<TFloat32> maxFreezedOutput, Class<V> outType, List<Long> strides, String padding, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("QuantizedConv2DAndReluAndRequantize", scope.makeOpName("QuantizedConv2DAndReluAndRequantize"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(filter.asOutput());
+    opBuilder.addInput(minInput.asOutput());
+    opBuilder.addInput(maxInput.asOutput());
+    opBuilder.addInput(minFilter.asOutput());
+    opBuilder.addInput(maxFilter.asOutput());
+    opBuilder.addInput(minFreezedOutput.asOutput());
+    opBuilder.addInput(maxFreezedOutput.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("out_type", outType);
+    long[] stridesArray = new long[strides.size()];
+    for (int i = 0; i < stridesArray.length; ++i) {
+      stridesArray[i] = strides.get(i);
+    }
+    opBuilder.setAttr("strides", stridesArray);
+    opBuilder.setAttr("padding", padding);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.dilations != null) {
+          long[] dilationsArray = new long[opts.dilations.size()];
+          for (int i = 0; i < dilationsArray.length; ++i) {
+            dilationsArray[i] = opts.dilations.get(i);
+          }
+          opBuilder.setAttr("dilations", dilationsArray);
+        }
+        if (opts.paddingList != null) {
+          long[] paddingListArray = new long[opts.paddingList.size()];
+          for (int i = 0; i < paddingListArray.length; ++i) {
+            paddingListArray[i] = opts.paddingList.get(i);
+          }
+          opBuilder.setAttr("padding_list", paddingListArray);
+        }
+      }
+    }
+    return new QuantizedConv2DAndReluAndRequantize<V>(opBuilder.build());
+  }
+  
+  /**
+   * @param dilations 
+   */
+  public static Options dilations(List<Long> dilations) {
+    return new Options().dilations(dilations);
+  }
+  
+  /**
+   * @param paddingList 
+   */
+  public static Options paddingList(List<Long> paddingList) {
+    return new Options().paddingList(paddingList);
+  }
+  
+  /**
+   */
+  public Output<V> output() {
+    return output;
+  }
+  
+  /**
+   */
+  public Output<TFloat32> minOutput() {
+    return minOutput;
+  }
+  
+  /**
+   */
+  public Output<TFloat32> maxOutput() {
+    return maxOutput;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "QuantizedConv2DAndReluAndRequantize";
+  
+  private Output<V> output;
+  private Output<TFloat32> minOutput;
+  private Output<TFloat32> maxOutput;
+  
+  private QuantizedConv2DAndReluAndRequantize(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+    minOutput = operation.output(outputIdx++);
+    maxOutput = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DAndRequantize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DAndRequantize.java
new file mode 100644
index 00000000000..30f303ae8f9
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DAndRequantize.java
@@ -0,0 +1,169 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.nn;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.family.TType;
+
+/**
+ * @param <V> data type for {@code output()} output
+ */
+public final class QuantizedConv2DAndRequantize<V extends TType> extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.nn.QuantizedConv2DAndRequantize}
+   */
+  public static class Options {
+    
+    /**
+     * @param dilations 
+     */
+    public Options dilations(List<Long> dilations) {
+      this.dilations = dilations;
+      return this;
+    }
+    
+    /**
+     * @param paddingList 
+     */
+    public Options paddingList(List<Long> paddingList) {
+      this.paddingList = paddingList;
+      return this;
+    }
+    
+    private List<Long> dilations;
+    private List<Long> paddingList;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new QuantizedConv2DAndRequantize operation.
+   * 
+   * @param scope current scope
+   * @param input 
+   * @param filter 
+   * @param minInput 
+   * @param maxInput 
+   * @param minFilter 
+   * @param maxFilter 
+   * @param minFreezedOutput 
+   * @param maxFreezedOutput 
+   * @param outType 
+   * @param strides 
+   * @param padding 
+   * @param options carries optional attributes values
+   * @return a new instance of QuantizedConv2DAndRequantize
+   */
+  @Endpoint(describeByClass = true)
+  public static <V extends TType, T extends TType, U extends TType> QuantizedConv2DAndRequantize<V> create(Scope scope, Operand<T> input, Operand<U> filter, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, Operand<TFloat32> minFilter, Operand<TFloat32> maxFilter, Operand<TFloat32> minFreezedOutput, Operand<TFloat32> maxFreezedOutput, Class<V> outType, List<Long> strides, String padding, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("QuantizedConv2DAndRequantize", scope.makeOpName("QuantizedConv2DAndRequantize"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(filter.asOutput());
+    opBuilder.addInput(minInput.asOutput());
+    opBuilder.addInput(maxInput.asOutput());
+    opBuilder.addInput(minFilter.asOutput());
+    opBuilder.addInput(maxFilter.asOutput());
+    opBuilder.addInput(minFreezedOutput.asOutput());
+    opBuilder.addInput(maxFreezedOutput.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("out_type", outType);
+    long[] stridesArray = new long[strides.size()];
+    for (int i = 0; i < stridesArray.length; ++i) {
+      stridesArray[i] = strides.get(i);
+    }
+    opBuilder.setAttr("strides", stridesArray);
+    opBuilder.setAttr("padding", padding);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.dilations != null) {
+          long[] dilationsArray = new long[opts.dilations.size()];
+          for (int i = 0; i < dilationsArray.length; ++i) {
+            dilationsArray[i] = opts.dilations.get(i);
+          }
+          opBuilder.setAttr("dilations", dilationsArray);
+        }
+        if (opts.paddingList != null) {
+          long[] paddingListArray = new long[opts.paddingList.size()];
+          for (int i = 0; i < paddingListArray.length; ++i) {
+            paddingListArray[i] = opts.paddingList.get(i);
+          }
+          opBuilder.setAttr("padding_list", paddingListArray);
+        }
+      }
+    }
+    return new QuantizedConv2DAndRequantize<V>(opBuilder.build());
+  }
+  
+  /**
+   * @param dilations 
+   */
+  public static Options dilations(List<Long> dilations) {
+    return new Options().dilations(dilations);
+  }
+  
+  /**
+   * @param paddingList 
+   */
+  public static Options paddingList(List<Long> paddingList) {
+    return new Options().paddingList(paddingList);
+  }
+  
+  /**
+   */
+  public Output<V> output() {
+    return output;
+  }
+  
+  /**
+   */
+  public Output<TFloat32> minOutput() {
+    return minOutput;
+  }
+  
+  /**
+   */
+  public Output<TFloat32> maxOutput() {
+    return maxOutput;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "QuantizedConv2DAndRequantize";
+  
+  private Output<V> output;
+  private Output<TFloat32> minOutput;
+  private Output<TFloat32> maxOutput;
+  
+  private QuantizedConv2DAndRequantize(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+    minOutput = operation.output(outputIdx++);
+    maxOutput = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DPerChannel.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DPerChannel.java
new file mode 100644
index 00000000000..23019817b6d
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DPerChannel.java
@@ -0,0 +1,147 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.nn;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Computes QuantizedConv2D per channel.
+ * 
+ * @param <V> data type for {@code output()} output
+ */
+public final class QuantizedConv2DPerChannel<V extends TType> extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.nn.QuantizedConv2DPerChannel}
+   */
+  public static class Options {
+    
+    /**
+     * @param dilations list of dilation values.
+     */
+    public Options dilations(List<Long> dilations) {
+      this.dilations = dilations;
+      return this;
+    }
+    
+    private List<Long> dilations;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new QuantizedConv2DPerChannel operation.
+   * 
+   * @param scope current scope
+   * @param input The original input tensor.
+   * @param filter The original filter tensor.
+   * @param minInput The minimum value of the input tensor
+   * @param maxInput The maximum value of the input tensor.
+   * @param minFilter The minimum value of the filter tensor.
+   * @param maxFilter The maximum value of the filter tensor.
+   * @param outType The quantized type of output tensor that needs to be converted.
+   * @param strides list of stride values.
+   * @param padding 
+   * @param options carries optional attributes values
+   * @return a new instance of QuantizedConv2DPerChannel
+   */
+  @Endpoint(describeByClass = true)
+  public static <V extends TType, T extends TType, U extends TType> QuantizedConv2DPerChannel<V> create(Scope scope, Operand<T> input, Operand<U> filter, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, Operand<TFloat32> minFilter, Operand<TFloat32> maxFilter, Class<V> outType, List<Long> strides, String padding, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("QuantizedConv2DPerChannel", scope.makeOpName("QuantizedConv2DPerChannel"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(filter.asOutput());
+    opBuilder.addInput(minInput.asOutput());
+    opBuilder.addInput(maxInput.asOutput());
+    opBuilder.addInput(minFilter.asOutput());
+    opBuilder.addInput(maxFilter.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("out_type", outType);
+    long[] stridesArray = new long[strides.size()];
+    for (int i = 0; i < stridesArray.length; ++i) {
+      stridesArray[i] = strides.get(i);
+    }
+    opBuilder.setAttr("strides", stridesArray);
+    opBuilder.setAttr("padding", padding);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.dilations != null) {
+          long[] dilationsArray = new long[opts.dilations.size()];
+          for (int i = 0; i < dilationsArray.length; ++i) {
+            dilationsArray[i] = opts.dilations.get(i);
+          }
+          opBuilder.setAttr("dilations", dilationsArray);
+        }
+      }
+    }
+    return new QuantizedConv2DPerChannel<V>(opBuilder.build());
+  }
+  
+  /**
+   * @param dilations list of dilation values.
+   */
+  public static Options dilations(List<Long> dilations) {
+    return new Options().dilations(dilations);
+  }
+  
+  /**
+   * The output tensor.
+   */
+  public Output<V> output() {
+    return output;
+  }
+  
+  /**
+   * The minimum value of the final output tensor.
+   */
+  public Output<TFloat32> minOutput() {
+    return minOutput;
+  }
+  
+  /**
+   * The maximum value of the final output tensor.
+   */
+  public Output<TFloat32> maxOutput() {
+    return maxOutput;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "QuantizedConv2DPerChannel";
+  
+  private Output<V> output;
+  private Output<TFloat32> minOutput;
+  private Output<TFloat32> maxOutput;
+  
+  private QuantizedConv2DPerChannel(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+    minOutput = operation.output(outputIdx++);
+    maxOutput = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DWithBias.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DWithBias.java
new file mode 100644
index 00000000000..a0e9fab70ab
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DWithBias.java
@@ -0,0 +1,167 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.nn;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.family.TType;
+
+/**
+ * @param <V> data type for {@code output()} output
+ */
+public final class QuantizedConv2DWithBias<V extends TType> extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.nn.QuantizedConv2DWithBias}
+   */
+  public static class Options {
+    
+    /**
+     * @param dilations 
+     */
+    public Options dilations(List<Long> dilations) {
+      this.dilations = dilations;
+      return this;
+    }
+    
+    /**
+     * @param paddingList 
+     */
+    public Options paddingList(List<Long> paddingList) {
+      this.paddingList = paddingList;
+      return this;
+    }
+    
+    private List<Long> dilations;
+    private List<Long> paddingList;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new QuantizedConv2DWithBias operation.
+   * 
+   * @param scope current scope
+   * @param input 
+   * @param filter 
+   * @param bias 
+   * @param minInput 
+   * @param maxInput 
+   * @param minFilter 
+   * @param maxFilter 
+   * @param outType 
+   * @param strides 
+   * @param padding 
+   * @param options carries optional attributes values
+   * @return a new instance of QuantizedConv2DWithBias
+   */
+  @Endpoint(describeByClass = true)
+  public static <V extends TType, T extends TType, U extends TType> QuantizedConv2DWithBias<V> create(Scope scope, Operand<T> input, Operand<U> filter, Operand<TFloat32> bias, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, Operand<TFloat32> minFilter, Operand<TFloat32> maxFilter, Class<V> outType, List<Long> strides, String padding, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("QuantizedConv2DWithBias", scope.makeOpName("QuantizedConv2DWithBias"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(filter.asOutput());
+    opBuilder.addInput(bias.asOutput());
+    opBuilder.addInput(minInput.asOutput());
+    opBuilder.addInput(maxInput.asOutput());
+    opBuilder.addInput(minFilter.asOutput());
+    opBuilder.addInput(maxFilter.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("out_type", outType);
+    long[] stridesArray = new long[strides.size()];
+    for (int i = 0; i < stridesArray.length; ++i) {
+      stridesArray[i] = strides.get(i);
+    }
+    opBuilder.setAttr("strides", stridesArray);
+    opBuilder.setAttr("padding", padding);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.dilations != null) {
+          long[] dilationsArray = new long[opts.dilations.size()];
+          for (int i = 0; i < dilationsArray.length; ++i) {
+            dilationsArray[i] = opts.dilations.get(i);
+          }
+          opBuilder.setAttr("dilations", dilationsArray);
+        }
+        if (opts.paddingList != null) {
+          long[] paddingListArray = new long[opts.paddingList.size()];
+          for (int i = 0; i < paddingListArray.length; ++i) {
+            paddingListArray[i] = opts.paddingList.get(i);
+          }
+          opBuilder.setAttr("padding_list", paddingListArray);
+        }
+      }
+    }
+    return new QuantizedConv2DWithBias<V>(opBuilder.build());
+  }
+  
+  /**
+   * @param dilations 
+   */
+  public static Options dilations(List<Long> dilations) {
+    return new Options().dilations(dilations);
+  }
+  
+  /**
+   * @param paddingList 
+   */
+  public static Options paddingList(List<Long> paddingList) {
+    return new Options().paddingList(paddingList);
+  }
+  
+  /**
+   */
+  public Output<V> output() {
+    return output;
+  }
+  
+  /**
+   */
+  public Output<TFloat32> minOutput() {
+    return minOutput;
+  }
+  
+  /**
+   */
+  public Output<TFloat32> maxOutput() {
+    return maxOutput;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "QuantizedConv2DWithBias";
+  
+  private Output<V> output;
+  private Output<TFloat32> minOutput;
+  private Output<TFloat32> maxOutput;
+  
+  private QuantizedConv2DWithBias(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+    minOutput = operation.output(outputIdx++);
+    maxOutput = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DWithBiasAndRelu.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DWithBiasAndRelu.java
new file mode 100644
index 00000000000..b0288a5adf0
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DWithBiasAndRelu.java
@@ -0,0 +1,167 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.nn;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.family.TType;
+
+/**
+ * @param <V> data type for {@code output()} output
+ */
+public final class QuantizedConv2DWithBiasAndRelu<V extends TType> extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.nn.QuantizedConv2DWithBiasAndRelu}
+   */
+  public static class Options {
+    
+    /**
+     * @param dilations 
+     */
+    public Options dilations(List<Long> dilations) {
+      this.dilations = dilations;
+      return this;
+    }
+    
+    /**
+     * @param paddingList 
+     */
+    public Options paddingList(List<Long> paddingList) {
+      this.paddingList = paddingList;
+      return this;
+    }
+    
+    private List<Long> dilations;
+    private List<Long> paddingList;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new QuantizedConv2DWithBiasAndRelu operation.
+   * 
+   * @param scope current scope
+   * @param input 
+   * @param filter 
+   * @param bias 
+   * @param minInput 
+   * @param maxInput 
+   * @param minFilter 
+   * @param maxFilter 
+   * @param outType 
+   * @param strides 
+   * @param padding 
+   * @param options carries optional attributes values
+   * @return a new instance of QuantizedConv2DWithBiasAndRelu
+   */
+  @Endpoint(describeByClass = true)
+  public static <V extends TType, T extends TType, U extends TType> QuantizedConv2DWithBiasAndRelu<V> create(Scope scope, Operand<T> input, Operand<U> filter, Operand<TFloat32> bias, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, Operand<TFloat32> minFilter, Operand<TFloat32> maxFilter, Class<V> outType, List<Long> strides, String padding, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("QuantizedConv2DWithBiasAndRelu", scope.makeOpName("QuantizedConv2DWithBiasAndRelu"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(filter.asOutput());
+    opBuilder.addInput(bias.asOutput());
+    opBuilder.addInput(minInput.asOutput());
+    opBuilder.addInput(maxInput.asOutput());
+    opBuilder.addInput(minFilter.asOutput());
+    opBuilder.addInput(maxFilter.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("out_type", outType);
+    long[] stridesArray = new long[strides.size()];
+    for (int i = 0; i < stridesArray.length; ++i) {
+      stridesArray[i] = strides.get(i);
+    }
+    opBuilder.setAttr("strides", stridesArray);
+    opBuilder.setAttr("padding", padding);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.dilations != null) {
+          long[] dilationsArray = new long[opts.dilations.size()];
+          for (int i = 0; i < dilationsArray.length; ++i) {
+            dilationsArray[i] = opts.dilations.get(i);
+          }
+          opBuilder.setAttr("dilations", dilationsArray);
+        }
+        if (opts.paddingList != null) {
+          long[] paddingListArray = new long[opts.paddingList.size()];
+          for (int i = 0; i < paddingListArray.length; ++i) {
+            paddingListArray[i] = opts.paddingList.get(i);
+          }
+          opBuilder.setAttr("padding_list", paddingListArray);
+        }
+      }
+    }
+    return new QuantizedConv2DWithBiasAndRelu<V>(opBuilder.build());
+  }
+  
+  /**
+   * @param dilations 
+   */
+  public static Options dilations(List<Long> dilations) {
+    return new Options().dilations(dilations);
+  }
+  
+  /**
+   * @param paddingList 
+   */
+  public static Options paddingList(List<Long> paddingList) {
+    return new Options().paddingList(paddingList);
+  }
+  
+  /**
+   */
+  public Output<V> output() {
+    return output;
+  }
+  
+  /**
+   */
+  public Output<TFloat32> minOutput() {
+    return minOutput;
+  }
+  
+  /**
+   */
+  public Output<TFloat32> maxOutput() {
+    return maxOutput;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "QuantizedConv2DWithBiasAndRelu";
+  
+  private Output<V> output;
+  private Output<TFloat32> minOutput;
+  private Output<TFloat32> maxOutput;
+  
+  private QuantizedConv2DWithBiasAndRelu(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+    minOutput = operation.output(outputIdx++);
+    maxOutput = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DWithBiasAndReluAndRequantize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DWithBiasAndReluAndRequantize.java
new file mode 100644
index 00000000000..c06f4ebe73c
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DWithBiasAndReluAndRequantize.java
@@ -0,0 +1,171 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.nn;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.family.TType;
+
+/**
+ * @param <W> data type for {@code output()} output
+ */
+public final class QuantizedConv2DWithBiasAndReluAndRequantize<W extends TType> extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.nn.QuantizedConv2DWithBiasAndReluAndRequantize}
+   */
+  public static class Options {
+    
+    /**
+     * @param dilations 
+     */
+    public Options dilations(List<Long> dilations) {
+      this.dilations = dilations;
+      return this;
+    }
+    
+    /**
+     * @param paddingList 
+     */
+    public Options paddingList(List<Long> paddingList) {
+      this.paddingList = paddingList;
+      return this;
+    }
+    
+    private List<Long> dilations;
+    private List<Long> paddingList;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new QuantizedConv2DWithBiasAndReluAndRequantize operation.
+   * 
+   * @param scope current scope
+   * @param input 
+   * @param filter 
+   * @param bias 
+   * @param minInput 
+   * @param maxInput 
+   * @param minFilter 
+   * @param maxFilter 
+   * @param minFreezedOutput 
+   * @param maxFreezedOutput 
+   * @param outType 
+   * @param strides 
+   * @param padding 
+   * @param options carries optional attributes values
+   * @return a new instance of QuantizedConv2DWithBiasAndReluAndRequantize
+   */
+  @Endpoint(describeByClass = true)
+  public static <W extends TType, T extends TType, U extends TType, V extends TType> QuantizedConv2DWithBiasAndReluAndRequantize<W> create(Scope scope, Operand<T> input, Operand<U> filter, Operand<V> bias, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, Operand<TFloat32> minFilter, Operand<TFloat32> maxFilter, Operand<TFloat32> minFreezedOutput, Operand<TFloat32> maxFreezedOutput, Class<W> outType, List<Long> strides, String padding, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("QuantizedConv2DWithBiasAndReluAndRequantize", scope.makeOpName("QuantizedConv2DWithBiasAndReluAndRequantize"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(filter.asOutput());
+    opBuilder.addInput(bias.asOutput());
+    opBuilder.addInput(minInput.asOutput());
+    opBuilder.addInput(maxInput.asOutput());
+    opBuilder.addInput(minFilter.asOutput());
+    opBuilder.addInput(maxFilter.asOutput());
+    opBuilder.addInput(minFreezedOutput.asOutput());
+    opBuilder.addInput(maxFreezedOutput.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("out_type", outType);
+    long[] stridesArray = new long[strides.size()];
+    for (int i = 0; i < stridesArray.length; ++i) {
+      stridesArray[i] = strides.get(i);
+    }
+    opBuilder.setAttr("strides", stridesArray);
+    opBuilder.setAttr("padding", padding);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.dilations != null) {
+          long[] dilationsArray = new long[opts.dilations.size()];
+          for (int i = 0; i < dilationsArray.length; ++i) {
+            dilationsArray[i] = opts.dilations.get(i);
+          }
+          opBuilder.setAttr("dilations", dilationsArray);
+        }
+        if (opts.paddingList != null) {
+          long[] paddingListArray = new long[opts.paddingList.size()];
+          for (int i = 0; i < paddingListArray.length; ++i) {
+            paddingListArray[i] = opts.paddingList.get(i);
+          }
+          opBuilder.setAttr("padding_list", paddingListArray);
+        }
+      }
+    }
+    return new QuantizedConv2DWithBiasAndReluAndRequantize<W>(opBuilder.build());
+  }
+  
+  /**
+   * @param dilations 
+   */
+  public static Options dilations(List<Long> dilations) {
+    return new Options().dilations(dilations);
+  }
+  
+  /**
+   * @param paddingList 
+   */
+  public static Options paddingList(List<Long> paddingList) {
+    return new Options().paddingList(paddingList);
+  }
+  
+  /**
+   */
+  public Output<W> output() {
+    return output;
+  }
+  
+  /**
+   */
+  public Output<TFloat32> minOutput() {
+    return minOutput;
+  }
+  
+  /**
+   */
+  public Output<TFloat32> maxOutput() {
+    return maxOutput;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "QuantizedConv2DWithBiasAndReluAndRequantize";
+  
+  private Output<W> output;
+  private Output<TFloat32> minOutput;
+  private Output<TFloat32> maxOutput;
+  
+  private QuantizedConv2DWithBiasAndReluAndRequantize(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+    minOutput = operation.output(outputIdx++);
+    maxOutput = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DWithBiasAndRequantize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DWithBiasAndRequantize.java
new file mode 100644
index 00000000000..81d1f4497a0
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DWithBiasAndRequantize.java
@@ -0,0 +1,171 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.nn;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.family.TType;
+
+/**
+ * @param <W> data type for {@code output()} output
+ */
+public final class QuantizedConv2DWithBiasAndRequantize<W extends TType> extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.nn.QuantizedConv2DWithBiasAndRequantize}
+   */
+  public static class Options {
+    
+    /**
+     * @param dilations 
+     */
+    public Options dilations(List<Long> dilations) {
+      this.dilations = dilations;
+      return this;
+    }
+    
+    /**
+     * @param paddingList 
+     */
+    public Options paddingList(List<Long> paddingList) {
+      this.paddingList = paddingList;
+      return this;
+    }
+    
+    private List<Long> dilations;
+    private List<Long> paddingList;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new QuantizedConv2DWithBiasAndRequantize operation.
+   * 
+   * @param scope current scope
+   * @param input 
+   * @param filter 
+   * @param bias 
+   * @param minInput 
+   * @param maxInput 
+   * @param minFilter 
+   * @param maxFilter 
+   * @param minFreezedOutput 
+   * @param maxFreezedOutput 
+   * @param outType 
+   * @param strides 
+   * @param padding 
+   * @param options carries optional attributes values
+   * @return a new instance of QuantizedConv2DWithBiasAndRequantize
+   */
+  @Endpoint(describeByClass = true)
+  public static <W extends TType, T extends TType, U extends TType, V extends TType> QuantizedConv2DWithBiasAndRequantize<W> create(Scope scope, Operand<T> input, Operand<U> filter, Operand<V> bias, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, Operand<TFloat32> minFilter, Operand<TFloat32> maxFilter, Operand<TFloat32> minFreezedOutput, Operand<TFloat32> maxFreezedOutput, Class<W> outType, List<Long> strides, String padding, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("QuantizedConv2DWithBiasAndRequantize", scope.makeOpName("QuantizedConv2DWithBiasAndRequantize"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(filter.asOutput());
+    opBuilder.addInput(bias.asOutput());
+    opBuilder.addInput(minInput.asOutput());
+    opBuilder.addInput(maxInput.asOutput());
+    opBuilder.addInput(minFilter.asOutput());
+    opBuilder.addInput(maxFilter.asOutput());
+    opBuilder.addInput(minFreezedOutput.asOutput());
+    opBuilder.addInput(maxFreezedOutput.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("out_type", outType);
+    long[] stridesArray = new long[strides.size()];
+    for (int i = 0; i < stridesArray.length; ++i) {
+      stridesArray[i] = strides.get(i);
+    }
+    opBuilder.setAttr("strides", stridesArray);
+    opBuilder.setAttr("padding", padding);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.dilations != null) {
+          long[] dilationsArray = new long[opts.dilations.size()];
+          for (int i = 0; i < dilationsArray.length; ++i) {
+            dilationsArray[i] = opts.dilations.get(i);
+          }
+          opBuilder.setAttr("dilations", dilationsArray);
+        }
+        if (opts.paddingList != null) {
+          long[] paddingListArray = new long[opts.paddingList.size()];
+          for (int i = 0; i < paddingListArray.length; ++i) {
+            paddingListArray[i] = opts.paddingList.get(i);
+          }
+          opBuilder.setAttr("padding_list", paddingListArray);
+        }
+      }
+    }
+    return new QuantizedConv2DWithBiasAndRequantize<W>(opBuilder.build());
+  }
+  
+  /**
+   * @param dilations 
+   */
+  public static Options dilations(List<Long> dilations) {
+    return new Options().dilations(dilations);
+  }
+  
+  /**
+   * @param paddingList 
+   */
+  public static Options paddingList(List<Long> paddingList) {
+    return new Options().paddingList(paddingList);
+  }
+  
+  /**
+   */
+  public Output<W> output() {
+    return output;
+  }
+  
+  /**
+   */
+  public Output<TFloat32> minOutput() {
+    return minOutput;
+  }
+  
+  /**
+   */
+  public Output<TFloat32> maxOutput() {
+    return maxOutput;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "QuantizedConv2DWithBiasAndRequantize";
+  
+  private Output<W> output;
+  private Output<TFloat32> minOutput;
+  private Output<TFloat32> maxOutput;
+  
+  private QuantizedConv2DWithBiasAndRequantize(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+    minOutput = operation.output(outputIdx++);
+    maxOutput = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DWithBiasSignedSumAndReluAndRequantize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DWithBiasSignedSumAndReluAndRequantize.java
new file mode 100644
index 00000000000..3c60a3f3a1a
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DWithBiasSignedSumAndReluAndRequantize.java
@@ -0,0 +1,177 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.nn;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.family.TType;
+
+/**
+ * @param <X> data type for {@code output()} output
+ */
+public final class QuantizedConv2DWithBiasSignedSumAndReluAndRequantize<X extends TType> extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.nn.QuantizedConv2DWithBiasSignedSumAndReluAndRequantize}
+   */
+  public static class Options {
+    
+    /**
+     * @param dilations 
+     */
+    public Options dilations(List<Long> dilations) {
+      this.dilations = dilations;
+      return this;
+    }
+    
+    /**
+     * @param paddingList 
+     */
+    public Options paddingList(List<Long> paddingList) {
+      this.paddingList = paddingList;
+      return this;
+    }
+    
+    private List<Long> dilations;
+    private List<Long> paddingList;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new QuantizedConv2DWithBiasSignedSumAndReluAndRequantize operation.
+   * 
+   * @param scope current scope
+   * @param input 
+   * @param filter 
+   * @param bias 
+   * @param minInput 
+   * @param maxInput 
+   * @param minFilter 
+   * @param maxFilter 
+   * @param minFreezedOutput 
+   * @param maxFreezedOutput 
+   * @param summand 
+   * @param minSummand 
+   * @param maxSummand 
+   * @param outType 
+   * @param strides 
+   * @param padding 
+   * @param options carries optional attributes values
+   * @return a new instance of QuantizedConv2DWithBiasSignedSumAndReluAndRequantize
+   */
+  @Endpoint(describeByClass = true)
+  public static <X extends TType, T extends TType, U extends TType, V extends TType, W extends TType> QuantizedConv2DWithBiasSignedSumAndReluAndRequantize<X> create(Scope scope, Operand<T> input, Operand<U> filter, Operand<V> bias, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, Operand<TFloat32> minFilter, Operand<TFloat32> maxFilter, Operand<TFloat32> minFreezedOutput, Operand<TFloat32> maxFreezedOutput, Operand<W> summand, Operand<TFloat32> minSummand, Operand<TFloat32> maxSummand, Class<X> outType, List<Long> strides, String padding, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("QuantizedConv2DWithBiasSignedSumAndReluAndRequantize", scope.makeOpName("QuantizedConv2DWithBiasSignedSumAndReluAndRequantize"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(filter.asOutput());
+    opBuilder.addInput(bias.asOutput());
+    opBuilder.addInput(minInput.asOutput());
+    opBuilder.addInput(maxInput.asOutput());
+    opBuilder.addInput(minFilter.asOutput());
+    opBuilder.addInput(maxFilter.asOutput());
+    opBuilder.addInput(minFreezedOutput.asOutput());
+    opBuilder.addInput(maxFreezedOutput.asOutput());
+    opBuilder.addInput(summand.asOutput());
+    opBuilder.addInput(minSummand.asOutput());
+    opBuilder.addInput(maxSummand.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("out_type", outType);
+    long[] stridesArray = new long[strides.size()];
+    for (int i = 0; i < stridesArray.length; ++i) {
+      stridesArray[i] = strides.get(i);
+    }
+    opBuilder.setAttr("strides", stridesArray);
+    opBuilder.setAttr("padding", padding);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.dilations != null) {
+          long[] dilationsArray = new long[opts.dilations.size()];
+          for (int i = 0; i < dilationsArray.length; ++i) {
+            dilationsArray[i] = opts.dilations.get(i);
+          }
+          opBuilder.setAttr("dilations", dilationsArray);
+        }
+        if (opts.paddingList != null) {
+          long[] paddingListArray = new long[opts.paddingList.size()];
+          for (int i = 0; i < paddingListArray.length; ++i) {
+            paddingListArray[i] = opts.paddingList.get(i);
+          }
+          opBuilder.setAttr("padding_list", paddingListArray);
+        }
+      }
+    }
+    return new QuantizedConv2DWithBiasSignedSumAndReluAndRequantize<X>(opBuilder.build());
+  }
+  
+  /**
+   * @param dilations 
+   */
+  public static Options dilations(List<Long> dilations) {
+    return new Options().dilations(dilations);
+  }
+  
+  /**
+   * @param paddingList 
+   */
+  public static Options paddingList(List<Long> paddingList) {
+    return new Options().paddingList(paddingList);
+  }
+  
+  /**
+   */
+  public Output<X> output() {
+    return output;
+  }
+  
+  /**
+   */
+  public Output<TFloat32> minOutput() {
+    return minOutput;
+  }
+  
+  /**
+   */
+  public Output<TFloat32> maxOutput() {
+    return maxOutput;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "QuantizedConv2DWithBiasSignedSumAndReluAndRequantize";
+  
+  private Output<X> output;
+  private Output<TFloat32> minOutput;
+  private Output<TFloat32> maxOutput;
+  
+  private QuantizedConv2DWithBiasSignedSumAndReluAndRequantize(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+    minOutput = operation.output(outputIdx++);
+    maxOutput = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DWithBiasSumAndRelu.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DWithBiasSumAndRelu.java
new file mode 100644
index 00000000000..0788d608ac1
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DWithBiasSumAndRelu.java
@@ -0,0 +1,169 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.nn;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.family.TType;
+
+/**
+ * @param <V> data type for {@code output()} output
+ */
+public final class QuantizedConv2DWithBiasSumAndRelu<V extends TType> extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.nn.QuantizedConv2DWithBiasSumAndRelu}
+   */
+  public static class Options {
+    
+    /**
+     * @param dilations 
+     */
+    public Options dilations(List<Long> dilations) {
+      this.dilations = dilations;
+      return this;
+    }
+    
+    /**
+     * @param paddingList 
+     */
+    public Options paddingList(List<Long> paddingList) {
+      this.paddingList = paddingList;
+      return this;
+    }
+    
+    private List<Long> dilations;
+    private List<Long> paddingList;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new QuantizedConv2DWithBiasSumAndRelu operation.
+   * 
+   * @param scope current scope
+   * @param input 
+   * @param filter 
+   * @param bias 
+   * @param minInput 
+   * @param maxInput 
+   * @param minFilter 
+   * @param maxFilter 
+   * @param summand 
+   * @param outType 
+   * @param strides 
+   * @param padding 
+   * @param options carries optional attributes values
+   * @return a new instance of QuantizedConv2DWithBiasSumAndRelu
+   */
+  @Endpoint(describeByClass = true)
+  public static <V extends TType, T extends TType, U extends TType> QuantizedConv2DWithBiasSumAndRelu<V> create(Scope scope, Operand<T> input, Operand<U> filter, Operand<TFloat32> bias, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, Operand<TFloat32> minFilter, Operand<TFloat32> maxFilter, Operand<TFloat32> summand, Class<V> outType, List<Long> strides, String padding, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("QuantizedConv2DWithBiasSumAndRelu", scope.makeOpName("QuantizedConv2DWithBiasSumAndRelu"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(filter.asOutput());
+    opBuilder.addInput(bias.asOutput());
+    opBuilder.addInput(minInput.asOutput());
+    opBuilder.addInput(maxInput.asOutput());
+    opBuilder.addInput(minFilter.asOutput());
+    opBuilder.addInput(maxFilter.asOutput());
+    opBuilder.addInput(summand.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("out_type", outType);
+    long[] stridesArray = new long[strides.size()];
+    for (int i = 0; i < stridesArray.length; ++i) {
+      stridesArray[i] = strides.get(i);
+    }
+    opBuilder.setAttr("strides", stridesArray);
+    opBuilder.setAttr("padding", padding);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.dilations != null) {
+          long[] dilationsArray = new long[opts.dilations.size()];
+          for (int i = 0; i < dilationsArray.length; ++i) {
+            dilationsArray[i] = opts.dilations.get(i);
+          }
+          opBuilder.setAttr("dilations", dilationsArray);
+        }
+        if (opts.paddingList != null) {
+          long[] paddingListArray = new long[opts.paddingList.size()];
+          for (int i = 0; i < paddingListArray.length; ++i) {
+            paddingListArray[i] = opts.paddingList.get(i);
+          }
+          opBuilder.setAttr("padding_list", paddingListArray);
+        }
+      }
+    }
+    return new QuantizedConv2DWithBiasSumAndRelu<V>(opBuilder.build());
+  }
+  
+  /**
+   * @param dilations 
+   */
+  public static Options dilations(List<Long> dilations) {
+    return new Options().dilations(dilations);
+  }
+  
+  /**
+   * @param paddingList 
+   */
+  public static Options paddingList(List<Long> paddingList) {
+    return new Options().paddingList(paddingList);
+  }
+  
+  /**
+   */
+  public Output<V> output() {
+    return output;
+  }
+  
+  /**
+   */
+  public Output<TFloat32> minOutput() {
+    return minOutput;
+  }
+  
+  /**
+   */
+  public Output<TFloat32> maxOutput() {
+    return maxOutput;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "QuantizedConv2DWithBiasSumAndRelu";
+  
+  private Output<V> output;
+  private Output<TFloat32> minOutput;
+  private Output<TFloat32> maxOutput;
+  
+  private QuantizedConv2DWithBiasSumAndRelu(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+    minOutput = operation.output(outputIdx++);
+    maxOutput = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DWithBiasSumAndReluAndRequantize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DWithBiasSumAndReluAndRequantize.java
new file mode 100644
index 00000000000..91c903bad91
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2DWithBiasSumAndReluAndRequantize.java
@@ -0,0 +1,177 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.nn;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.family.TType;
+
+/**
+ * @param <X> data type for {@code output()} output
+ */
+public final class QuantizedConv2DWithBiasSumAndReluAndRequantize<X extends TType> extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.nn.QuantizedConv2DWithBiasSumAndReluAndRequantize}
+   */
+  public static class Options {
+    
+    /**
+     * @param dilations 
+     */
+    public Options dilations(List<Long> dilations) {
+      this.dilations = dilations;
+      return this;
+    }
+    
+    /**
+     * @param paddingList 
+     */
+    public Options paddingList(List<Long> paddingList) {
+      this.paddingList = paddingList;
+      return this;
+    }
+    
+    private List<Long> dilations;
+    private List<Long> paddingList;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new QuantizedConv2DWithBiasSumAndReluAndRequantize operation.
+   * 
+   * @param scope current scope
+   * @param input 
+   * @param filter 
+   * @param bias 
+   * @param minInput 
+   * @param maxInput 
+   * @param minFilter 
+   * @param maxFilter 
+   * @param minFreezedOutput 
+   * @param maxFreezedOutput 
+   * @param summand 
+   * @param minSummand 
+   * @param maxSummand 
+   * @param outType 
+   * @param strides 
+   * @param padding 
+   * @param options carries optional attributes values
+   * @return a new instance of QuantizedConv2DWithBiasSumAndReluAndRequantize
+   */
+  @Endpoint(describeByClass = true)
+  public static <X extends TType, T extends TType, U extends TType, V extends TType, W extends TType> QuantizedConv2DWithBiasSumAndReluAndRequantize<X> create(Scope scope, Operand<T> input, Operand<U> filter, Operand<V> bias, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, Operand<TFloat32> minFilter, Operand<TFloat32> maxFilter, Operand<TFloat32> minFreezedOutput, Operand<TFloat32> maxFreezedOutput, Operand<W> summand, Operand<TFloat32> minSummand, Operand<TFloat32> maxSummand, Class<X> outType, List<Long> strides, String padding, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("QuantizedConv2DWithBiasSumAndReluAndRequantize", scope.makeOpName("QuantizedConv2DWithBiasSumAndReluAndRequantize"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(filter.asOutput());
+    opBuilder.addInput(bias.asOutput());
+    opBuilder.addInput(minInput.asOutput());
+    opBuilder.addInput(maxInput.asOutput());
+    opBuilder.addInput(minFilter.asOutput());
+    opBuilder.addInput(maxFilter.asOutput());
+    opBuilder.addInput(minFreezedOutput.asOutput());
+    opBuilder.addInput(maxFreezedOutput.asOutput());
+    opBuilder.addInput(summand.asOutput());
+    opBuilder.addInput(minSummand.asOutput());
+    opBuilder.addInput(maxSummand.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("out_type", outType);
+    long[] stridesArray = new long[strides.size()];
+    for (int i = 0; i < stridesArray.length; ++i) {
+      stridesArray[i] = strides.get(i);
+    }
+    opBuilder.setAttr("strides", stridesArray);
+    opBuilder.setAttr("padding", padding);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.dilations != null) {
+          long[] dilationsArray = new long[opts.dilations.size()];
+          for (int i = 0; i < dilationsArray.length; ++i) {
+            dilationsArray[i] = opts.dilations.get(i);
+          }
+          opBuilder.setAttr("dilations", dilationsArray);
+        }
+        if (opts.paddingList != null) {
+          long[] paddingListArray = new long[opts.paddingList.size()];
+          for (int i = 0; i < paddingListArray.length; ++i) {
+            paddingListArray[i] = opts.paddingList.get(i);
+          }
+          opBuilder.setAttr("padding_list", paddingListArray);
+        }
+      }
+    }
+    return new QuantizedConv2DWithBiasSumAndReluAndRequantize<X>(opBuilder.build());
+  }
+  
+  /**
+   * @param dilations 
+   */
+  public static Options dilations(List<Long> dilations) {
+    return new Options().dilations(dilations);
+  }
+  
+  /**
+   * @param paddingList 
+   */
+  public static Options paddingList(List<Long> paddingList) {
+    return new Options().paddingList(paddingList);
+  }
+  
+  /**
+   */
+  public Output<X> output() {
+    return output;
+  }
+  
+  /**
+   */
+  public Output<TFloat32> minOutput() {
+    return minOutput;
+  }
+  
+  /**
+   */
+  public Output<TFloat32> maxOutput() {
+    return maxOutput;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "QuantizedConv2DWithBiasSumAndReluAndRequantize";
+  
+  private Output<X> output;
+  private Output<TFloat32> minOutput;
+  private Output<TFloat32> maxOutput;
+  
+  private QuantizedConv2DWithBiasSumAndReluAndRequantize(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+    minOutput = operation.output(outputIdx++);
+    maxOutput = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2d.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2d.java
new file mode 100644
index 00000000000..ffd66fc3f23
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedConv2d.java
@@ -0,0 +1,161 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.nn;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Computes a 2D convolution given quantized 4D input and filter tensors.
+ * <p>
+ * The inputs are quantized tensors where the lowest value represents the real
+ * number of the associated minimum, and the highest represents the maximum.
+ * This means that you can only interpret the quantized output in the same way, by
+ * taking the returned minimum and maximum values into account.
+ * 
+ * @param <V> data type for {@code output()} output
+ */
+@Operator(group = "nn")
+public final class QuantizedConv2d<V extends TType> extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.nn.QuantizedConv2d}
+   */
+  public static class Options {
+    
+    /**
+     * @param dilations 1-D tensor of length 4.  The dilation factor for each dimension of
+     * `input`. If set to k > 1, there will be k-1 skipped cells between each
+     * filter element on that dimension. The dimension order is determined by the
+     * value of `data_format`, see above for details. Dilations in the batch and
+     * depth dimensions must be 1.
+     */
+    public Options dilations(List<Long> dilations) {
+      this.dilations = dilations;
+      return this;
+    }
+    
+    private List<Long> dilations;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new QuantizedConv2d operation.
+   * 
+   * @param scope current scope
+   * @param input 
+   * @param filter filter's input_depth dimension must match input's depth dimensions.
+   * @param minInput The float value that the lowest quantized input value represents.
+   * @param maxInput The float value that the highest quantized input value represents.
+   * @param minFilter The float value that the lowest quantized filter value represents.
+   * @param maxFilter The float value that the highest quantized filter value represents.
+   * @param outType 
+   * @param strides The stride of the sliding window for each dimension of the input
+   * tensor.
+   * @param padding The type of padding algorithm to use.
+   * @param options carries optional attributes values
+   * @return a new instance of QuantizedConv2d
+   */
+  @Endpoint(describeByClass = true)
+  public static <V extends TType, T extends TType, U extends TType> QuantizedConv2d<V> create(Scope scope, Operand<T> input, Operand<U> filter, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, Operand<TFloat32> minFilter, Operand<TFloat32> maxFilter, Class<V> outType, List<Long> strides, String padding, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("QuantizedConv2D", scope.makeOpName("QuantizedConv2d"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(filter.asOutput());
+    opBuilder.addInput(minInput.asOutput());
+    opBuilder.addInput(maxInput.asOutput());
+    opBuilder.addInput(minFilter.asOutput());
+    opBuilder.addInput(maxFilter.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("out_type", outType);
+    long[] stridesArray = new long[strides.size()];
+    for (int i = 0; i < stridesArray.length; ++i) {
+      stridesArray[i] = strides.get(i);
+    }
+    opBuilder.setAttr("strides", stridesArray);
+    opBuilder.setAttr("padding", padding);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.dilations != null) {
+          long[] dilationsArray = new long[opts.dilations.size()];
+          for (int i = 0; i < dilationsArray.length; ++i) {
+            dilationsArray[i] = opts.dilations.get(i);
+          }
+          opBuilder.setAttr("dilations", dilationsArray);
+        }
+      }
+    }
+    return new QuantizedConv2d<V>(opBuilder.build());
+  }
+  
+  /**
+   * @param dilations 1-D tensor of length 4.  The dilation factor for each dimension of
+   * `input`. If set to k > 1, there will be k-1 skipped cells between each
+   * filter element on that dimension. The dimension order is determined by the
+   * value of `data_format`, see above for details. Dilations in the batch and
+   * depth dimensions must be 1.
+   */
+  public static Options dilations(List<Long> dilations) {
+    return new Options().dilations(dilations);
+  }
+  
+  /**
+   */
+  public Output<V> output() {
+    return output;
+  }
+  
+  /**
+   * The float value that the lowest quantized output value represents.
+   */
+  public Output<TFloat32> minOutput() {
+    return minOutput;
+  }
+  
+  /**
+   * The float value that the highest quantized output value represents.
+   */
+  public Output<TFloat32> maxOutput() {
+    return maxOutput;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "QuantizedConv2D";
+  
+  private Output<V> output;
+  private Output<TFloat32> minOutput;
+  private Output<TFloat32> maxOutput;
+  
+  private QuantizedConv2d(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+    minOutput = operation.output(outputIdx++);
+    maxOutput = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedDepthwiseConv2D.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedDepthwiseConv2D.java
new file mode 100644
index 00000000000..1e0a8c38e0f
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedDepthwiseConv2D.java
@@ -0,0 +1,147 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.nn;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Computes quantized depthwise Conv2D.
+ * 
+ * @param <V> data type for {@code output()} output
+ */
+public final class QuantizedDepthwiseConv2D<V extends TType> extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.nn.QuantizedDepthwiseConv2D}
+   */
+  public static class Options {
+    
+    /**
+     * @param dilations List of dilation values.
+     */
+    public Options dilations(List<Long> dilations) {
+      this.dilations = dilations;
+      return this;
+    }
+    
+    private List<Long> dilations;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new QuantizedDepthwiseConv2D operation.
+   * 
+   * @param scope current scope
+   * @param input The original input tensor.
+   * @param filter The original filter tensor.
+   * @param minInput The float value that the minimum quantized input value represents.
+   * @param maxInput The float value that the maximum quantized input value represents.
+   * @param minFilter The float value that the minimum quantized filter value represents.
+   * @param maxFilter The float value that the maximum quantized filter value represents.
+   * @param outType The type of the output.
+   * @param strides List of stride values.
+   * @param padding 
+   * @param options carries optional attributes values
+   * @return a new instance of QuantizedDepthwiseConv2D
+   */
+  @Endpoint(describeByClass = true)
+  public static <V extends TType, T extends TType, U extends TType> QuantizedDepthwiseConv2D<V> create(Scope scope, Operand<T> input, Operand<U> filter, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, Operand<TFloat32> minFilter, Operand<TFloat32> maxFilter, Class<V> outType, List<Long> strides, String padding, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("QuantizedDepthwiseConv2D", scope.makeOpName("QuantizedDepthwiseConv2D"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(filter.asOutput());
+    opBuilder.addInput(minInput.asOutput());
+    opBuilder.addInput(maxInput.asOutput());
+    opBuilder.addInput(minFilter.asOutput());
+    opBuilder.addInput(maxFilter.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("out_type", outType);
+    long[] stridesArray = new long[strides.size()];
+    for (int i = 0; i < stridesArray.length; ++i) {
+      stridesArray[i] = strides.get(i);
+    }
+    opBuilder.setAttr("strides", stridesArray);
+    opBuilder.setAttr("padding", padding);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.dilations != null) {
+          long[] dilationsArray = new long[opts.dilations.size()];
+          for (int i = 0; i < dilationsArray.length; ++i) {
+            dilationsArray[i] = opts.dilations.get(i);
+          }
+          opBuilder.setAttr("dilations", dilationsArray);
+        }
+      }
+    }
+    return new QuantizedDepthwiseConv2D<V>(opBuilder.build());
+  }
+  
+  /**
+   * @param dilations List of dilation values.
+   */
+  public static Options dilations(List<Long> dilations) {
+    return new Options().dilations(dilations);
+  }
+  
+  /**
+   * The output tensor.
+   */
+  public Output<V> output() {
+    return output;
+  }
+  
+  /**
+   * The float value that the minimum quantized output value represents.
+   */
+  public Output<TFloat32> minOutput() {
+    return minOutput;
+  }
+  
+  /**
+   * The float value that the maximum quantized output value represents.
+   */
+  public Output<TFloat32> maxOutput() {
+    return maxOutput;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "QuantizedDepthwiseConv2D";
+  
+  private Output<V> output;
+  private Output<TFloat32> minOutput;
+  private Output<TFloat32> maxOutput;
+  
+  private QuantizedDepthwiseConv2D(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+    minOutput = operation.output(outputIdx++);
+    maxOutput = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedDepthwiseConv2DWithBias.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedDepthwiseConv2DWithBias.java
new file mode 100644
index 00000000000..b0d29c64e39
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedDepthwiseConv2DWithBias.java
@@ -0,0 +1,149 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.nn;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Computes quantized depthwise Conv2D with Bias.
+ * 
+ * @param <V> data type for {@code output()} output
+ */
+public final class QuantizedDepthwiseConv2DWithBias<V extends TType> extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.nn.QuantizedDepthwiseConv2DWithBias}
+   */
+  public static class Options {
+    
+    /**
+     * @param dilations List of dilation values.
+     */
+    public Options dilations(List<Long> dilations) {
+      this.dilations = dilations;
+      return this;
+    }
+    
+    private List<Long> dilations;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new QuantizedDepthwiseConv2DWithBias operation.
+   * 
+   * @param scope current scope
+   * @param input The original input tensor.
+   * @param filter The original filter tensor.
+   * @param bias The original bias tensor.
+   * @param minInput The float value that the minimum quantized input value represents.
+   * @param maxInput The float value that the maximum quantized input value represents.
+   * @param minFilter The float value that the minimum quantized filter value represents.
+   * @param maxFilter The float value that the maximum quantized filter value represents.
+   * @param outType The type of the output.
+   * @param strides List of stride values.
+   * @param padding 
+   * @param options carries optional attributes values
+   * @return a new instance of QuantizedDepthwiseConv2DWithBias
+   */
+  @Endpoint(describeByClass = true)
+  public static <V extends TType, T extends TType, U extends TType> QuantizedDepthwiseConv2DWithBias<V> create(Scope scope, Operand<T> input, Operand<U> filter, Operand<TFloat32> bias, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, Operand<TFloat32> minFilter, Operand<TFloat32> maxFilter, Class<V> outType, List<Long> strides, String padding, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("QuantizedDepthwiseConv2DWithBias", scope.makeOpName("QuantizedDepthwiseConv2DWithBias"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(filter.asOutput());
+    opBuilder.addInput(bias.asOutput());
+    opBuilder.addInput(minInput.asOutput());
+    opBuilder.addInput(maxInput.asOutput());
+    opBuilder.addInput(minFilter.asOutput());
+    opBuilder.addInput(maxFilter.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("out_type", outType);
+    long[] stridesArray = new long[strides.size()];
+    for (int i = 0; i < stridesArray.length; ++i) {
+      stridesArray[i] = strides.get(i);
+    }
+    opBuilder.setAttr("strides", stridesArray);
+    opBuilder.setAttr("padding", padding);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.dilations != null) {
+          long[] dilationsArray = new long[opts.dilations.size()];
+          for (int i = 0; i < dilationsArray.length; ++i) {
+            dilationsArray[i] = opts.dilations.get(i);
+          }
+          opBuilder.setAttr("dilations", dilationsArray);
+        }
+      }
+    }
+    return new QuantizedDepthwiseConv2DWithBias<V>(opBuilder.build());
+  }
+  
+  /**
+   * @param dilations List of dilation values.
+   */
+  public static Options dilations(List<Long> dilations) {
+    return new Options().dilations(dilations);
+  }
+  
+  /**
+   * The output tensor.
+   */
+  public Output<V> output() {
+    return output;
+  }
+  
+  /**
+   * The float value that the minimum quantized output value represents.
+   */
+  public Output<TFloat32> minOutput() {
+    return minOutput;
+  }
+  
+  /**
+   * The float value that the maximum quantized output value represents.
+   */
+  public Output<TFloat32> maxOutput() {
+    return maxOutput;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "QuantizedDepthwiseConv2DWithBias";
+  
+  private Output<V> output;
+  private Output<TFloat32> minOutput;
+  private Output<TFloat32> maxOutput;
+  
+  private QuantizedDepthwiseConv2DWithBias(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+    minOutput = operation.output(outputIdx++);
+    maxOutput = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedDepthwiseConv2DWithBiasAndRelu.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedDepthwiseConv2DWithBiasAndRelu.java
new file mode 100644
index 00000000000..e88bddc4f83
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedDepthwiseConv2DWithBiasAndRelu.java
@@ -0,0 +1,172 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.nn;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Computes quantized depthwise Conv2D with Bias and Relu.
+ * 
+ * @param <V> data type for {@code output()} output
+ */
+public final class QuantizedDepthwiseConv2DWithBiasAndRelu<V extends TType> extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.nn.QuantizedDepthwiseConv2DWithBiasAndRelu}
+   */
+  public static class Options {
+    
+    /**
+     * @param dilations List of dilation values.
+     */
+    public Options dilations(List<Long> dilations) {
+      this.dilations = dilations;
+      return this;
+    }
+    
+    /**
+     * @param paddingList 
+     */
+    public Options paddingList(List<Long> paddingList) {
+      this.paddingList = paddingList;
+      return this;
+    }
+    
+    private List<Long> dilations;
+    private List<Long> paddingList;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new QuantizedDepthwiseConv2DWithBiasAndRelu operation.
+   * 
+   * @param scope current scope
+   * @param input The original input tensor.
+   * @param filter The original filter tensor.
+   * @param bias The original bias tensor.
+   * @param minInput The float value that the minimum quantized input value represents.
+   * @param maxInput The float value that the maximum quantized input value represents.
+   * @param minFilter The float value that the minimum quantized filter value represents.
+   * @param maxFilter The float value that the maximum quantized filter value represents.
+   * @param outType The type of the output.
+   * @param strides List of stride values.
+   * @param padding 
+   * @param options carries optional attributes values
+   * @return a new instance of QuantizedDepthwiseConv2DWithBiasAndRelu
+   */
+  @Endpoint(describeByClass = true)
+  public static <V extends TType, T extends TType, U extends TType> QuantizedDepthwiseConv2DWithBiasAndRelu<V> create(Scope scope, Operand<T> input, Operand<U> filter, Operand<TFloat32> bias, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, Operand<TFloat32> minFilter, Operand<TFloat32> maxFilter, Class<V> outType, List<Long> strides, String padding, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("QuantizedDepthwiseConv2DWithBiasAndRelu", scope.makeOpName("QuantizedDepthwiseConv2DWithBiasAndRelu"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(filter.asOutput());
+    opBuilder.addInput(bias.asOutput());
+    opBuilder.addInput(minInput.asOutput());
+    opBuilder.addInput(maxInput.asOutput());
+    opBuilder.addInput(minFilter.asOutput());
+    opBuilder.addInput(maxFilter.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("out_type", outType);
+    long[] stridesArray = new long[strides.size()];
+    for (int i = 0; i < stridesArray.length; ++i) {
+      stridesArray[i] = strides.get(i);
+    }
+    opBuilder.setAttr("strides", stridesArray);
+    opBuilder.setAttr("padding", padding);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.dilations != null) {
+          long[] dilationsArray = new long[opts.dilations.size()];
+          for (int i = 0; i < dilationsArray.length; ++i) {
+            dilationsArray[i] = opts.dilations.get(i);
+          }
+          opBuilder.setAttr("dilations", dilationsArray);
+        }
+        if (opts.paddingList != null) {
+          long[] paddingListArray = new long[opts.paddingList.size()];
+          for (int i = 0; i < paddingListArray.length; ++i) {
+            paddingListArray[i] = opts.paddingList.get(i);
+          }
+          opBuilder.setAttr("padding_list", paddingListArray);
+        }
+      }
+    }
+    return new QuantizedDepthwiseConv2DWithBiasAndRelu<V>(opBuilder.build());
+  }
+  
+  /**
+   * @param dilations List of dilation values.
+   */
+  public static Options dilations(List<Long> dilations) {
+    return new Options().dilations(dilations);
+  }
+  
+  /**
+   * @param paddingList 
+   */
+  public static Options paddingList(List<Long> paddingList) {
+    return new Options().paddingList(paddingList);
+  }
+  
+  /**
+   * The output tensor.
+   */
+  public Output<V> output() {
+    return output;
+  }
+  
+  /**
+   * The float value that the minimum quantized output value represents.
+   */
+  public Output<TFloat32> minOutput() {
+    return minOutput;
+  }
+  
+  /**
+   * The float value that the maximum quantized output value represents.
+   */
+  public Output<TFloat32> maxOutput() {
+    return maxOutput;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "QuantizedDepthwiseConv2DWithBiasAndRelu";
+  
+  private Output<V> output;
+  private Output<TFloat32> minOutput;
+  private Output<TFloat32> maxOutput;
+  
+  private QuantizedDepthwiseConv2DWithBiasAndRelu(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+    minOutput = operation.output(outputIdx++);
+    maxOutput = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedDepthwiseConv2DWithBiasAndReluAndRequantize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedDepthwiseConv2DWithBiasAndReluAndRequantize.java
new file mode 100644
index 00000000000..3430d0a6e2f
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedDepthwiseConv2DWithBiasAndReluAndRequantize.java
@@ -0,0 +1,176 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.nn;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Computes quantized depthwise Conv2D with Bias, Relu and Requantize.
+ * 
+ * @param <W> data type for {@code output()} output
+ */
+public final class QuantizedDepthwiseConv2DWithBiasAndReluAndRequantize<W extends TType> extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.nn.QuantizedDepthwiseConv2DWithBiasAndReluAndRequantize}
+   */
+  public static class Options {
+    
+    /**
+     * @param dilations List of dilation values.
+     */
+    public Options dilations(List<Long> dilations) {
+      this.dilations = dilations;
+      return this;
+    }
+    
+    /**
+     * @param paddingList 
+     */
+    public Options paddingList(List<Long> paddingList) {
+      this.paddingList = paddingList;
+      return this;
+    }
+    
+    private List<Long> dilations;
+    private List<Long> paddingList;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new QuantizedDepthwiseConv2DWithBiasAndReluAndRequantize operation.
+   * 
+   * @param scope current scope
+   * @param input The original input tensor.
+   * @param filter The original filter tensor.
+   * @param bias The original bias tensor.
+   * @param minInput The float value that the minimum quantized input value represents.
+   * @param maxInput The float value that the maximum quantized input value represents.
+   * @param minFilter The float value that the minimum quantized filter value represents.
+   * @param maxFilter The float value that the maximum quantized filter value represents.
+   * @param minFreezedOutput The minimum float value of the output tensor.
+   * @param maxFreezedOutput The maximum float value of the output tensor.
+   * @param outType The type of the output.
+   * @param strides List of stride values.
+   * @param padding 
+   * @param options carries optional attributes values
+   * @return a new instance of QuantizedDepthwiseConv2DWithBiasAndReluAndRequantize
+   */
+  @Endpoint(describeByClass = true)
+  public static <W extends TType, T extends TType, U extends TType, V extends TType> QuantizedDepthwiseConv2DWithBiasAndReluAndRequantize<W> create(Scope scope, Operand<T> input, Operand<U> filter, Operand<V> bias, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, Operand<TFloat32> minFilter, Operand<TFloat32> maxFilter, Operand<TFloat32> minFreezedOutput, Operand<TFloat32> maxFreezedOutput, Class<W> outType, List<Long> strides, String padding, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("QuantizedDepthwiseConv2DWithBiasAndReluAndRequantize", scope.makeOpName("QuantizedDepthwiseConv2DWithBiasAndReluAndRequantize"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(filter.asOutput());
+    opBuilder.addInput(bias.asOutput());
+    opBuilder.addInput(minInput.asOutput());
+    opBuilder.addInput(maxInput.asOutput());
+    opBuilder.addInput(minFilter.asOutput());
+    opBuilder.addInput(maxFilter.asOutput());
+    opBuilder.addInput(minFreezedOutput.asOutput());
+    opBuilder.addInput(maxFreezedOutput.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("out_type", outType);
+    long[] stridesArray = new long[strides.size()];
+    for (int i = 0; i < stridesArray.length; ++i) {
+      stridesArray[i] = strides.get(i);
+    }
+    opBuilder.setAttr("strides", stridesArray);
+    opBuilder.setAttr("padding", padding);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.dilations != null) {
+          long[] dilationsArray = new long[opts.dilations.size()];
+          for (int i = 0; i < dilationsArray.length; ++i) {
+            dilationsArray[i] = opts.dilations.get(i);
+          }
+          opBuilder.setAttr("dilations", dilationsArray);
+        }
+        if (opts.paddingList != null) {
+          long[] paddingListArray = new long[opts.paddingList.size()];
+          for (int i = 0; i < paddingListArray.length; ++i) {
+            paddingListArray[i] = opts.paddingList.get(i);
+          }
+          opBuilder.setAttr("padding_list", paddingListArray);
+        }
+      }
+    }
+    return new QuantizedDepthwiseConv2DWithBiasAndReluAndRequantize<W>(opBuilder.build());
+  }
+  
+  /**
+   * @param dilations List of dilation values.
+   */
+  public static Options dilations(List<Long> dilations) {
+    return new Options().dilations(dilations);
+  }
+  
+  /**
+   * @param paddingList 
+   */
+  public static Options paddingList(List<Long> paddingList) {
+    return new Options().paddingList(paddingList);
+  }
+  
+  /**
+   * The output tensor.
+   */
+  public Output<W> output() {
+    return output;
+  }
+  
+  /**
+   * The float value that the minimum quantized output value represents.
+   */
+  public Output<TFloat32> minOutput() {
+    return minOutput;
+  }
+  
+  /**
+   * The float value that the maximum quantized output value represents.
+   */
+  public Output<TFloat32> maxOutput() {
+    return maxOutput;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "QuantizedDepthwiseConv2DWithBiasAndReluAndRequantize";
+  
+  private Output<W> output;
+  private Output<TFloat32> minOutput;
+  private Output<TFloat32> maxOutput;
+  
+  private QuantizedDepthwiseConv2DWithBiasAndReluAndRequantize(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+    minOutput = operation.output(outputIdx++);
+    maxOutput = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedInstanceNorm.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedInstanceNorm.java
new file mode 100644
index 00000000000..839338a1458
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedInstanceNorm.java
@@ -0,0 +1,207 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.nn;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Quantized Instance normalization.
+ * 
+ * @param <T> data type for {@code y()} output
+ */
+@Operator(group = "nn")
+public final class QuantizedInstanceNorm<T extends TType> extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.nn.QuantizedInstanceNorm}
+   */
+  public static class Options {
+    
+    /**
+     * @param outputRangeGiven If True, `given_y_min` and `given_y_min`
+     * and `given_y_max` are used as the output range. Otherwise,
+     * the implementation computes the output range.
+     */
+    public Options outputRangeGiven(Boolean outputRangeGiven) {
+      this.outputRangeGiven = outputRangeGiven;
+      return this;
+    }
+    
+    /**
+     * @param givenYMin Output in `y_min` if `output_range_given` is True.
+     */
+    public Options givenYMin(Float givenYMin) {
+      this.givenYMin = givenYMin;
+      return this;
+    }
+    
+    /**
+     * @param givenYMax Output in `y_max` if `output_range_given` is True.
+     */
+    public Options givenYMax(Float givenYMax) {
+      this.givenYMax = givenYMax;
+      return this;
+    }
+    
+    /**
+     * @param varianceEpsilon A small float number to avoid dividing by 0.
+     */
+    public Options varianceEpsilon(Float varianceEpsilon) {
+      this.varianceEpsilon = varianceEpsilon;
+      return this;
+    }
+    
+    /**
+     * @param minSeparation Minimum value of `y_max - y_min`
+     */
+    public Options minSeparation(Float minSeparation) {
+      this.minSeparation = minSeparation;
+      return this;
+    }
+    
+    private Boolean outputRangeGiven;
+    private Float givenYMin;
+    private Float givenYMax;
+    private Float varianceEpsilon;
+    private Float minSeparation;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new QuantizedInstanceNorm operation.
+   * 
+   * @param scope current scope
+   * @param x A 4D input Tensor.
+   * @param xMin The value represented by the lowest quantized input.
+   * @param xMax The value represented by the highest quantized input.
+   * @param options carries optional attributes values
+   * @return a new instance of QuantizedInstanceNorm
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> QuantizedInstanceNorm<T> create(Scope scope, Operand<T> x, Operand<TFloat32> xMin, Operand<TFloat32> xMax, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("QuantizedInstanceNorm", scope.makeOpName("QuantizedInstanceNorm"));
+    opBuilder.addInput(x.asOutput());
+    opBuilder.addInput(xMin.asOutput());
+    opBuilder.addInput(xMax.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.outputRangeGiven != null) {
+          opBuilder.setAttr("output_range_given", opts.outputRangeGiven);
+        }
+        if (opts.givenYMin != null) {
+          opBuilder.setAttr("given_y_min", opts.givenYMin);
+        }
+        if (opts.givenYMax != null) {
+          opBuilder.setAttr("given_y_max", opts.givenYMax);
+        }
+        if (opts.varianceEpsilon != null) {
+          opBuilder.setAttr("variance_epsilon", opts.varianceEpsilon);
+        }
+        if (opts.minSeparation != null) {
+          opBuilder.setAttr("min_separation", opts.minSeparation);
+        }
+      }
+    }
+    return new QuantizedInstanceNorm<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param outputRangeGiven If True, `given_y_min` and `given_y_min`
+   * and `given_y_max` are used as the output range. Otherwise,
+   * the implementation computes the output range.
+   */
+  public static Options outputRangeGiven(Boolean outputRangeGiven) {
+    return new Options().outputRangeGiven(outputRangeGiven);
+  }
+  
+  /**
+   * @param givenYMin Output in `y_min` if `output_range_given` is True.
+   */
+  public static Options givenYMin(Float givenYMin) {
+    return new Options().givenYMin(givenYMin);
+  }
+  
+  /**
+   * @param givenYMax Output in `y_max` if `output_range_given` is True.
+   */
+  public static Options givenYMax(Float givenYMax) {
+    return new Options().givenYMax(givenYMax);
+  }
+  
+  /**
+   * @param varianceEpsilon A small float number to avoid dividing by 0.
+   */
+  public static Options varianceEpsilon(Float varianceEpsilon) {
+    return new Options().varianceEpsilon(varianceEpsilon);
+  }
+  
+  /**
+   * @param minSeparation Minimum value of `y_max - y_min`
+   */
+  public static Options minSeparation(Float minSeparation) {
+    return new Options().minSeparation(minSeparation);
+  }
+  
+  /**
+   * A 4D Tensor.
+   */
+  public Output<T> y() {
+    return y;
+  }
+  
+  /**
+   * The value represented by the lowest quantized output.
+   */
+  public Output<TFloat32> yMin() {
+    return yMin;
+  }
+  
+  /**
+   * The value represented by the highest quantized output.
+   */
+  public Output<TFloat32> yMax() {
+    return yMax;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "QuantizedInstanceNorm";
+  
+  private Output<T> y;
+  private Output<TFloat32> yMin;
+  private Output<TFloat32> yMax;
+  
+  private QuantizedInstanceNorm(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    y = operation.output(outputIdx++);
+    yMin = operation.output(outputIdx++);
+    yMax = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedMaxPool.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedMaxPool.java
new file mode 100644
index 00000000000..2cec96e411c
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedMaxPool.java
@@ -0,0 +1,109 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.nn;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Produces the max pool of the input tensor for quantized types.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator(group = "nn")
+public final class QuantizedMaxPool<T extends TType> extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new QuantizedMaxPool operation.
+   * 
+   * @param scope current scope
+   * @param input The 4D (batch x rows x cols x depth) Tensor to MaxReduce over.
+   * @param minInput The float value that the lowest quantized input value represents.
+   * @param maxInput The float value that the highest quantized input value represents.
+   * @param ksize The size of the window for each dimension of the input tensor.
+   * The length must be 4 to match the number of dimensions of the input.
+   * @param strides The stride of the sliding window for each dimension of the input
+   * tensor. The length must be 4 to match the number of dimensions of the input.
+   * @param padding The type of padding algorithm to use.
+   * @return a new instance of QuantizedMaxPool
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> QuantizedMaxPool<T> create(Scope scope, Operand<T> input, Operand<TFloat32> minInput, Operand<TFloat32> maxInput, List<Long> ksize, List<Long> strides, String padding) {
+    OperationBuilder opBuilder = scope.env().opBuilder("QuantizedMaxPool", scope.makeOpName("QuantizedMaxPool"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(minInput.asOutput());
+    opBuilder.addInput(maxInput.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    long[] ksizeArray = new long[ksize.size()];
+    for (int i = 0; i < ksizeArray.length; ++i) {
+      ksizeArray[i] = ksize.get(i);
+    }
+    opBuilder.setAttr("ksize", ksizeArray);
+    long[] stridesArray = new long[strides.size()];
+    for (int i = 0; i < stridesArray.length; ++i) {
+      stridesArray[i] = strides.get(i);
+    }
+    opBuilder.setAttr("strides", stridesArray);
+    opBuilder.setAttr("padding", padding);
+    return new QuantizedMaxPool<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  /**
+   * The float value that the lowest quantized output value represents.
+   */
+  public Output<TFloat32> minOutput() {
+    return minOutput;
+  }
+  
+  /**
+   * The float value that the highest quantized output value represents.
+   */
+  public Output<TFloat32> maxOutput() {
+    return maxOutput;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "QuantizedMaxPool";
+  
+  private Output<T> output;
+  private Output<TFloat32> minOutput;
+  private Output<TFloat32> maxOutput;
+  
+  private QuantizedMaxPool(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+    minOutput = operation.output(outputIdx++);
+    maxOutput = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedRelu.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedRelu.java
new file mode 100644
index 00000000000..8881d2eb824
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedRelu.java
@@ -0,0 +1,95 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.nn;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Computes Quantized Rectified Linear: `max(features, 0)`
+ * 
+ * @param <U> data type for {@code activations()} output
+ */
+@Operator(group = "nn")
+public final class QuantizedRelu<U extends TType> extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new QuantizedRelu operation.
+   * 
+   * @param scope current scope
+   * @param features 
+   * @param minFeatures The float value that the lowest quantized value represents.
+   * @param maxFeatures The float value that the highest quantized value represents.
+   * @param outType 
+   * @return a new instance of QuantizedRelu
+   */
+  @Endpoint(describeByClass = true)
+  public static <U extends TType, T extends TType> QuantizedRelu<U> create(Scope scope, Operand<T> features, Operand<TFloat32> minFeatures, Operand<TFloat32> maxFeatures, Class<U> outType) {
+    OperationBuilder opBuilder = scope.env().opBuilder("QuantizedRelu", scope.makeOpName("QuantizedRelu"));
+    opBuilder.addInput(features.asOutput());
+    opBuilder.addInput(minFeatures.asOutput());
+    opBuilder.addInput(maxFeatures.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("out_type", outType);
+    return new QuantizedRelu<U>(opBuilder.build());
+  }
+  
+  /**
+   * Has the same output shape as "features".
+   */
+  public Output<U> activations() {
+    return activations;
+  }
+  
+  /**
+   * The float value that the lowest quantized value represents.
+   */
+  public Output<TFloat32> minActivations() {
+    return minActivations;
+  }
+  
+  /**
+   * The float value that the highest quantized value represents.
+   */
+  public Output<TFloat32> maxActivations() {
+    return maxActivations;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "QuantizedRelu";
+  
+  private Output<U> activations;
+  private Output<TFloat32> minActivations;
+  private Output<TFloat32> maxActivations;
+  
+  private QuantizedRelu(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    activations = operation.output(outputIdx++);
+    minActivations = operation.output(outputIdx++);
+    maxActivations = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedRelu6.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedRelu6.java
new file mode 100644
index 00000000000..2153c3561df
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedRelu6.java
@@ -0,0 +1,95 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.nn;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Computes Quantized Rectified Linear 6: `min(max(features, 0), 6)`
+ * 
+ * @param <U> data type for {@code activations()} output
+ */
+@Operator(group = "nn")
+public final class QuantizedRelu6<U extends TType> extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new QuantizedRelu6 operation.
+   * 
+   * @param scope current scope
+   * @param features 
+   * @param minFeatures The float value that the lowest quantized value represents.
+   * @param maxFeatures The float value that the highest quantized value represents.
+   * @param outType 
+   * @return a new instance of QuantizedRelu6
+   */
+  @Endpoint(describeByClass = true)
+  public static <U extends TType, T extends TType> QuantizedRelu6<U> create(Scope scope, Operand<T> features, Operand<TFloat32> minFeatures, Operand<TFloat32> maxFeatures, Class<U> outType) {
+    OperationBuilder opBuilder = scope.env().opBuilder("QuantizedRelu6", scope.makeOpName("QuantizedRelu6"));
+    opBuilder.addInput(features.asOutput());
+    opBuilder.addInput(minFeatures.asOutput());
+    opBuilder.addInput(maxFeatures.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("out_type", outType);
+    return new QuantizedRelu6<U>(opBuilder.build());
+  }
+  
+  /**
+   * Has the same output shape as "features".
+   */
+  public Output<U> activations() {
+    return activations;
+  }
+  
+  /**
+   * The float value that the lowest quantized value represents.
+   */
+  public Output<TFloat32> minActivations() {
+    return minActivations;
+  }
+  
+  /**
+   * The float value that the highest quantized value represents.
+   */
+  public Output<TFloat32> maxActivations() {
+    return maxActivations;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "QuantizedRelu6";
+  
+  private Output<U> activations;
+  private Output<TFloat32> minActivations;
+  private Output<TFloat32> maxActivations;
+  
+  private QuantizedRelu6(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    activations = operation.output(outputIdx++);
+    minActivations = operation.output(outputIdx++);
+    maxActivations = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedReluX.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedReluX.java
new file mode 100644
index 00000000000..9db60f2174c
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/QuantizedReluX.java
@@ -0,0 +1,97 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.nn;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Computes Quantized Rectified Linear X: `min(max(features, 0), max_value)`
+ * 
+ * @param <U> data type for {@code activations()} output
+ */
+@Operator(group = "nn")
+public final class QuantizedReluX<U extends TType> extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new QuantizedReluX operation.
+   * 
+   * @param scope current scope
+   * @param features 
+   * @param maxValue 
+   * @param minFeatures The float value that the lowest quantized value represents.
+   * @param maxFeatures The float value that the highest quantized value represents.
+   * @param outType 
+   * @return a new instance of QuantizedReluX
+   */
+  @Endpoint(describeByClass = true)
+  public static <U extends TType, T extends TType> QuantizedReluX<U> create(Scope scope, Operand<T> features, Operand<TFloat32> maxValue, Operand<TFloat32> minFeatures, Operand<TFloat32> maxFeatures, Class<U> outType) {
+    OperationBuilder opBuilder = scope.env().opBuilder("QuantizedReluX", scope.makeOpName("QuantizedReluX"));
+    opBuilder.addInput(features.asOutput());
+    opBuilder.addInput(maxValue.asOutput());
+    opBuilder.addInput(minFeatures.asOutput());
+    opBuilder.addInput(maxFeatures.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("out_type", outType);
+    return new QuantizedReluX<U>(opBuilder.build());
+  }
+  
+  /**
+   * Has the same output shape as "features".
+   */
+  public Output<U> activations() {
+    return activations;
+  }
+  
+  /**
+   * The float value that the lowest quantized value represents.
+   */
+  public Output<TFloat32> minActivations() {
+    return minActivations;
+  }
+  
+  /**
+   * The float value that the highest quantized value represents.
+   */
+  public Output<TFloat32> maxActivations() {
+    return maxActivations;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "QuantizedReluX";
+  
+  private Output<U> activations;
+  private Output<TFloat32> minActivations;
+  private Output<TFloat32> maxActivations;
+  
+  private QuantizedReluX(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    activations = operation.output(outputIdx++);
+    minActivations = operation.output(outputIdx++);
+    maxActivations = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Relu.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Relu.java
new file mode 100644
index 00000000000..c6a515d4c75
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Relu.java
@@ -0,0 +1,79 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.nn;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Computes rectified linear: `max(features, 0)`.
+ * <p>
+ * See: https://en.wikipedia.org/wiki/Rectifier_(neural_networks)
+ * Example usage:
+ * >>> tf.nn.relu([-2., 0., -0., 3.]).numpy()
+ * array([ 0.,  0., -0.,  3.], dtype=float32)
+ * 
+ * @param <T> data type for {@code activations()} output
+ */
+@Operator(group = "nn")
+public final class Relu<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new Relu operation.
+   * 
+   * @param scope current scope
+   * @param features 
+   * @return a new instance of Relu
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> Relu<T> create(Scope scope, Operand<T> features) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Relu", scope.makeOpName("Relu"));
+    opBuilder.addInput(features.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new Relu<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> activations() {
+    return activations;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return activations;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Relu";
+  
+  private Output<T> activations;
+  
+  private Relu(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    activations = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Relu6.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Relu6.java
new file mode 100644
index 00000000000..f227e995b2e
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Relu6.java
@@ -0,0 +1,74 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.nn;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Computes rectified linear 6: `min(max(features, 0), 6)`.
+ * 
+ * @param <T> data type for {@code activations()} output
+ */
+@Operator(group = "nn")
+public final class Relu6<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new Relu6 operation.
+   * 
+   * @param scope current scope
+   * @param features 
+   * @return a new instance of Relu6
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> Relu6<T> create(Scope scope, Operand<T> features) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Relu6", scope.makeOpName("Relu6"));
+    opBuilder.addInput(features.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new Relu6<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> activations() {
+    return activations;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return activations;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Relu6";
+  
+  private Output<T> activations;
+  
+  private Relu6(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    activations = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Relu6Grad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Relu6Grad.java
new file mode 100644
index 00000000000..043f3b5a5d7
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Relu6Grad.java
@@ -0,0 +1,78 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.nn;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Computes rectified linear 6 gradients for a Relu6 operation.
+ * 
+ * @param <T> data type for {@code backprops()} output
+ */
+public final class Relu6Grad<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new Relu6Grad operation.
+   * 
+   * @param scope current scope
+   * @param gradients The backpropagated gradients to the corresponding Relu6 operation.
+   * @param features The features passed as input to the corresponding Relu6 operation, or
+   * its output; using either one produces the same result.
+   * @return a new instance of Relu6Grad
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> Relu6Grad<T> create(Scope scope, Operand<T> gradients, Operand<T> features) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Relu6Grad", scope.makeOpName("Relu6Grad"));
+    opBuilder.addInput(gradients.asOutput());
+    opBuilder.addInput(features.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new Relu6Grad<T>(opBuilder.build());
+  }
+  
+  /**
+   * The gradients:
+   * `gradients * (features > 0) * (features < 6)`.
+   */
+  public Output<T> backprops() {
+    return backprops;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return backprops;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Relu6Grad";
+  
+  private Output<T> backprops;
+  
+  private Relu6Grad(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    backprops = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/ReluGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/ReluGrad.java
new file mode 100644
index 00000000000..b41caedda64
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/ReluGrad.java
@@ -0,0 +1,77 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.nn;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Computes rectified linear gradients for a Relu operation.
+ * 
+ * @param <T> data type for {@code backprops()} output
+ */
+public final class ReluGrad<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new ReluGrad operation.
+   * 
+   * @param scope current scope
+   * @param gradients The backpropagated gradients to the corresponding Relu operation.
+   * @param features The features passed as input to the corresponding Relu operation, OR
+   * the outputs of that operation (both work equivalently).
+   * @return a new instance of ReluGrad
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> ReluGrad<T> create(Scope scope, Operand<T> gradients, Operand<T> features) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ReluGrad", scope.makeOpName("ReluGrad"));
+    opBuilder.addInput(gradients.asOutput());
+    opBuilder.addInput(features.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new ReluGrad<T>(opBuilder.build());
+  }
+  
+  /**
+   * `gradients * (features > 0)`.
+   */
+  public Output<T> backprops() {
+    return backprops;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return backprops;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ReluGrad";
+  
+  private Output<T> backprops;
+  
+  private ReluGrad(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    backprops = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Selu.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Selu.java
new file mode 100644
index 00000000000..355222c25fa
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Selu.java
@@ -0,0 +1,82 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.nn;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Computes scaled exponential linear: `scale * alpha * (exp(features) - 1)`
+ * <p>
+ * if < 0, `scale * features` otherwise.
+ * <p>
+ * To be used together with
+ * `initializer = tf.variance_scaling_initializer(factor=1.0, mode='FAN_IN')`.
+ * For correct dropout, use `tf.contrib.nn.alpha_dropout`.
+ * <p>
+ * See [Self-Normalizing Neural Networks](https://arxiv.org/abs/1706.02515)
+ * 
+ * @param <T> data type for {@code activations()} output
+ */
+@Operator(group = "nn")
+public final class Selu<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new Selu operation.
+   * 
+   * @param scope current scope
+   * @param features 
+   * @return a new instance of Selu
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> Selu<T> create(Scope scope, Operand<T> features) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Selu", scope.makeOpName("Selu"));
+    opBuilder.addInput(features.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new Selu<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> activations() {
+    return activations;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return activations;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Selu";
+  
+  private Output<T> activations;
+  
+  private Selu(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    activations = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/SeluGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/SeluGrad.java
new file mode 100644
index 00000000000..6bf70f4038d
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/SeluGrad.java
@@ -0,0 +1,77 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.nn;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Computes gradients for the scaled exponential linear (Selu) operation.
+ * 
+ * @param <T> data type for {@code backprops()} output
+ */
+public final class SeluGrad<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new SeluGrad operation.
+   * 
+   * @param scope current scope
+   * @param gradients The backpropagated gradients to the corresponding Selu operation.
+   * @param outputs The outputs of the corresponding Selu operation.
+   * @return a new instance of SeluGrad
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> SeluGrad<T> create(Scope scope, Operand<T> gradients, Operand<T> outputs) {
+    OperationBuilder opBuilder = scope.env().opBuilder("SeluGrad", scope.makeOpName("SeluGrad"));
+    opBuilder.addInput(gradients.asOutput());
+    opBuilder.addInput(outputs.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new SeluGrad<T>(opBuilder.build());
+  }
+  
+  /**
+   * The gradients: `gradients * (outputs + scale * alpha)`
+   * if outputs < 0, `scale * gradients` otherwise.
+   */
+  public Output<T> backprops() {
+    return backprops;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return backprops;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "SeluGrad";
+  
+  private Output<T> backprops;
+  
+  private SeluGrad(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    backprops = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Softmax.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Softmax.java
new file mode 100644
index 00000000000..093195aea28
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Softmax.java
@@ -0,0 +1,79 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.nn;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Computes softmax activations.
+ * <p>
+ * For each batch `i` and class `j` we have
+ * <p>
+ *     $$softmax[i, j] = exp(logits[i, j]) / sum_j(exp(logits[i, j]))$$
+ * 
+ * @param <T> data type for {@code softmax()} output
+ */
+@Operator(group = "nn")
+public final class Softmax<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new Softmax operation.
+   * 
+   * @param scope current scope
+   * @param logits 2-D with shape `[batch_size, num_classes]`.
+   * @return a new instance of Softmax
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> Softmax<T> create(Scope scope, Operand<T> logits) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Softmax", scope.makeOpName("Softmax"));
+    opBuilder.addInput(logits.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new Softmax<T>(opBuilder.build());
+  }
+  
+  /**
+   * Same shape as `logits`.
+   */
+  public Output<T> softmax() {
+    return softmax;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return softmax;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Softmax";
+  
+  private Output<T> softmax;
+  
+  private Softmax(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    softmax = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Softsign.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Softsign.java
new file mode 100644
index 00000000000..0b2e3a639e9
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/Softsign.java
@@ -0,0 +1,74 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.nn;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Computes softsign: `features / (abs(features) + 1)`.
+ * 
+ * @param <T> data type for {@code activations()} output
+ */
+@Operator(group = "nn")
+public final class Softsign<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new Softsign operation.
+   * 
+   * @param scope current scope
+   * @param features 
+   * @return a new instance of Softsign
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> Softsign<T> create(Scope scope, Operand<T> features) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Softsign", scope.makeOpName("Softsign"));
+    opBuilder.addInput(features.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new Softsign<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> activations() {
+    return activations;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return activations;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Softsign";
+  
+  private Output<T> activations;
+  
+  private Softsign(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    activations = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/SoftsignGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/SoftsignGrad.java
new file mode 100644
index 00000000000..31367f06df5
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/SoftsignGrad.java
@@ -0,0 +1,76 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.nn;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Computes softsign gradients for a softsign operation.
+ * 
+ * @param <T> data type for {@code backprops()} output
+ */
+public final class SoftsignGrad<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new SoftsignGrad operation.
+   * 
+   * @param scope current scope
+   * @param gradients The backpropagated gradients to the corresponding softsign operation.
+   * @param features The features passed as input to the corresponding softsign operation.
+   * @return a new instance of SoftsignGrad
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> SoftsignGrad<T> create(Scope scope, Operand<T> gradients, Operand<T> features) {
+    OperationBuilder opBuilder = scope.env().opBuilder("SoftsignGrad", scope.makeOpName("SoftsignGrad"));
+    opBuilder.addInput(gradients.asOutput());
+    opBuilder.addInput(features.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new SoftsignGrad<T>(opBuilder.build());
+  }
+  
+  /**
+   * The gradients: `gradients / (1 + abs(features)) ** 2`.
+   */
+  public Output<T> backprops() {
+    return backprops;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return backprops;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "SoftsignGrad";
+  
+  private Output<T> backprops;
+  
+  private SoftsignGrad(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    backprops = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/SpaceToBatch.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/SpaceToBatch.java
new file mode 100644
index 00000000000..9b1796c99aa
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/SpaceToBatch.java
@@ -0,0 +1,155 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.nn;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * SpaceToBatch for 4-D tensors of type T.
+ * <p>
+ * This is a legacy version of the more general SpaceToBatchND.
+ * <p>
+ * Zero-pads and then rearranges (permutes) blocks of spatial data into batch.
+ * More specifically, this op outputs a copy of the input tensor where values from
+ * the `height` and `width` dimensions are moved to the `batch` dimension. After
+ * the zero-padding, both `height` and `width` of the input must be divisible by the
+ * block size.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator(group = "nn")
+public final class SpaceToBatch<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new SpaceToBatch operation.
+   * 
+   * @param scope current scope
+   * @param input 4-D with shape `[batch, height, width, depth]`.
+   * @param paddings 2-D tensor of non-negative integers with shape `[2, 2]`. It specifies
+   *   the padding of the input with zeros across the spatial dimensions as follows:
+   * <p>
+   *       paddings = [[pad_top, pad_bottom], [pad_left, pad_right]]
+   * <p>
+   *   The effective spatial dimensions of the zero-padded input tensor will be:
+   * <p>
+   *       height_pad = pad_top + height + pad_bottom
+   *       width_pad = pad_left + width + pad_right
+   * <p>
+   * The attr `block_size` must be greater than one. It indicates the block size.
+   * <p>
+   *   * Non-overlapping blocks of size `block_size x block size` in the height and
+   *     width dimensions are rearranged into the batch dimension at each location.
+   *   * The batch of the output tensor is `batch * block_size * block_size`.
+   *   * Both height_pad and width_pad must be divisible by block_size.
+   * <p>
+   * The shape of the output will be:
+   * <p>
+   *     [batch<i>block_size</i>block_size, height_pad/block_size, width_pad/block_size,
+   *      depth]
+   * <p>
+   * Some examples:
+   * <p>
+   * (1) For the following input of shape `[1, 2, 2, 1]` and block_size of 2:
+   * <pre>{@code
+   * x = [[[[1], [2]], [[3], [4]]]]
+   * }</pre>
+   * The output tensor has shape `[4, 1, 1, 1]` and value:
+   * <pre>{@code
+   * [[[[1]]], [[[2]]], [[[3]]], [[[4]]]]
+   * }</pre>
+   * (2) For the following input of shape `[1, 2, 2, 3]` and block_size of 2:
+   * <pre>{@code
+   * x = [[[[1, 2, 3], [4, 5, 6]],
+   *       [[7, 8, 9], [10, 11, 12]]]]
+   * }</pre>
+   * The output tensor has shape `[4, 1, 1, 3]` and value:
+   * <pre>{@code
+   * [[[[1, 2, 3]]], [[[4, 5, 6]]], [[[7, 8, 9]]], [[[10, 11, 12]]]]
+   * }</pre>
+   * (3) For the following input of shape `[1, 4, 4, 1]` and block_size of 2:
+   * <pre>{@code
+   * x = [[[[1],   [2],  [3],  [4]],
+   *       [[5],   [6],  [7],  [8]],
+   *       [[9],  [10], [11],  [12]],
+   *       [[13], [14], [15],  [16]]]]
+   * }</pre>
+   * The output tensor has shape `[4, 2, 2, 1]` and value:
+   * <pre>{@code
+   * x = [[[[1], [3]], [[9], [11]]],
+   *      [[[2], [4]], [[10], [12]]],
+   *      [[[5], [7]], [[13], [15]]],
+   *      [[[6], [8]], [[14], [16]]]]
+   * }</pre>
+   * (4) For the following input of shape `[2, 2, 4, 1]` and block_size of 2:
+   * <pre>{@code
+   * x = [[[[1],   [2],  [3],  [4]],
+   *       [[5],   [6],  [7],  [8]]],
+   *      [[[9],  [10], [11],  [12]],
+   *       [[13], [14], [15],  [16]]]]
+   * }</pre>
+   * The output tensor has shape `[8, 1, 2, 1]` and value:
+   * <pre>{@code
+   * x = [[[[1], [3]]], [[[9], [11]]], [[[2], [4]]], [[[10], [12]]],
+   *      [[[5], [7]]], [[[13], [15]]], [[[6], [8]]], [[[14], [16]]]]
+   * }</pre>
+   * Among others, this operation is useful for reducing atrous convolution into
+   * regular convolution.
+   * @param blockSize 
+   * @return a new instance of SpaceToBatch
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType, U extends TNumber> SpaceToBatch<T> create(Scope scope, Operand<T> input, Operand<U> paddings, Long blockSize) {
+    OperationBuilder opBuilder = scope.env().opBuilder("SpaceToBatch", scope.makeOpName("SpaceToBatch"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(paddings.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("block_size", blockSize);
+    return new SpaceToBatch<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "SpaceToBatch";
+  
+  private Output<T> output;
+  
+  private SpaceToBatch(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/SpaceToDepth.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/SpaceToDepth.java
new file mode 100644
index 00000000000..0e9068c9b31
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/SpaceToDepth.java
@@ -0,0 +1,184 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.nn;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * SpaceToDepth for tensors of type T.
+ * <p>
+ * Rearranges blocks of spatial data, into depth. More specifically,
+ * this op outputs a copy of the input tensor where values from the `height`
+ * and `width` dimensions are moved to the `depth` dimension.
+ * The attr `block_size` indicates the input block size.
+ * <p>
+ *   * Non-overlapping blocks of size `block_size x block size` are rearranged
+ *     into depth at each location.
+ *   * The depth of the output tensor is `block_size * block_size * input_depth`.
+ *   * The Y, X coordinates within each block of the input become the high order
+ *     component of the output channel index.
+ *   * The input tensor's height and width must be divisible by block_size.
+ * <p>
+ * The `data_format` attr specifies the layout of the input and output tensors
+ * with the following options:
+ *   "NHWC": `[ batch, height, width, channels ]`
+ *   "NCHW": `[ batch, channels, height, width ]`
+ *   "NCHW_VECT_C":
+ *       `qint8 [ batch, channels / 4, height, width, 4 ]`
+ * <p>
+ * It is useful to consider the operation as transforming a 6-D Tensor.
+ * e.g. for data_format = NHWC,
+ *      Each element in the input tensor can be specified via 6 coordinates,
+ *      ordered by decreasing memory layout significance as:
+ *      n,oY,bY,oX,bX,iC  (where n=batch index, oX, oY means X or Y coordinates
+ *                         within the output image, bX, bY means coordinates
+ *                         within the input block, iC means input channels).
+ *      The output would be a transpose to the following layout:
+ *      n,oY,oX,bY,bX,iC
+ * <p>
+ * This operation is useful for resizing the activations between convolutions
+ * (but keeping all data), e.g. instead of pooling. It is also useful for training
+ * purely convolutional models.
+ * <p>
+ * For example, given an input of shape `[1, 2, 2, 1]`, data_format = "NHWC" and
+ * block_size = 2:
+ * <pre>{@code
+ * x = [[[[1], [2]],
+ *       [[3], [4]]]]
+ * }</pre>
+ * This operation will output a tensor of shape `[1, 1, 1, 4]`:
+ * <pre>{@code
+ * [[[[1, 2, 3, 4]]]]
+ * }</pre>
+ * Here, the input has a batch of 1 and each batch element has shape `[2, 2, 1]`,
+ * the corresponding output will have a single element (i.e. width and height are
+ * both 1) and will have a depth of 4 channels (1 * block_size * block_size).
+ * The output element shape is `[1, 1, 4]`.
+ * <p>
+ * For an input tensor with larger depth, here of shape `[1, 2, 2, 3]`, e.g.
+ * <pre>{@code
+ * x = [[[[1, 2, 3], [4, 5, 6]],
+ *       [[7, 8, 9], [10, 11, 12]]]]
+ * }</pre>
+ * This operation, for block_size of 2, will return the following tensor of shape
+ * `[1, 1, 1, 12]`
+ * <pre>{@code
+ * [[[[1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12]]]]
+ * }</pre>
+ * Similarly, for the following input of shape `[1 4 4 1]`, and a block size of 2:
+ * <pre>{@code
+ * x = [[[[1],   [2],  [5],  [6]],
+ *       [[3],   [4],  [7],  [8]],
+ *       [[9],  [10], [13],  [14]],
+ *       [[11], [12], [15],  [16]]]]
+ * }</pre>
+ * the operator will return the following tensor of shape `[1 2 2 4]`:
+ * <pre>{@code
+ * x = [[[[1, 2, 3, 4],
+ *        [5, 6, 7, 8]],
+ *       [[9, 10, 11, 12],
+ *        [13, 14, 15, 16]]]]
+ * }</pre>
+ * 
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator(group = "nn")
+public final class SpaceToDepth<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.nn.SpaceToDepth}
+   */
+  public static class Options {
+    
+    /**
+     * @param dataFormat 
+     */
+    public Options dataFormat(String dataFormat) {
+      this.dataFormat = dataFormat;
+      return this;
+    }
+    
+    private String dataFormat;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new SpaceToDepth operation.
+   * 
+   * @param scope current scope
+   * @param input 
+   * @param blockSize The size of the spatial block.
+   * @param options carries optional attributes values
+   * @return a new instance of SpaceToDepth
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> SpaceToDepth<T> create(Scope scope, Operand<T> input, Long blockSize, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("SpaceToDepth", scope.makeOpName("SpaceToDepth"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("block_size", blockSize);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.dataFormat != null) {
+          opBuilder.setAttr("data_format", opts.dataFormat);
+        }
+      }
+    }
+    return new SpaceToDepth<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param dataFormat 
+   */
+  public static Options dataFormat(String dataFormat) {
+    return new Options().dataFormat(dataFormat);
+  }
+  
+  /**
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "SpaceToDepth";
+  
+  private Output<T> output;
+  
+  private SpaceToDepth(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/TopK.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/TopK.java
new file mode 100644
index 00000000000..a8bfdbea715
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/TopK.java
@@ -0,0 +1,130 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.nn;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Finds values and indices of the `k` largest elements for the last dimension.
+ * <p>
+ * If the input is a vector (rank-1), finds the `k` largest entries in the vector
+ * and outputs their values and indices as vectors.  Thus `values[j]` is the
+ * `j`-th largest entry in `input`, and its index is `indices[j]`.
+ * <p>
+ * For matrices (resp. higher rank input), computes the top `k` entries in each
+ * row (resp. vector along the last dimension).  Thus,
+ * <p>
+ *     values.shape = indices.shape = input.shape[:-1] + [k]
+ * <p>
+ * If two elements are equal, the lower-index element appears first.
+ * 
+ * @param <T> data type for {@code values()} output
+ */
+@Operator(group = "nn")
+public final class TopK<T extends TNumber> extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.nn.TopK}
+   */
+  public static class Options {
+    
+    /**
+     * @param sorted If true the resulting `k` elements will be sorted by the values in
+     * descending order.
+     */
+    public Options sorted(Boolean sorted) {
+      this.sorted = sorted;
+      return this;
+    }
+    
+    private Boolean sorted;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new TopK operation.
+   * 
+   * @param scope current scope
+   * @param input 1-D or higher with last dimension at least `k`.
+   * @param k 0-D.  Number of top elements to look for along the last dimension (along each
+   * row for matrices).
+   * @param options carries optional attributes values
+   * @return a new instance of TopK
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> TopK<T> create(Scope scope, Operand<T> input, Operand<TInt32> k, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("TopKV2", scope.makeOpName("TopK"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(k.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.sorted != null) {
+          opBuilder.setAttr("sorted", opts.sorted);
+        }
+      }
+    }
+    return new TopK<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param sorted If true the resulting `k` elements will be sorted by the values in
+   * descending order.
+   */
+  public static Options sorted(Boolean sorted) {
+    return new Options().sorted(sorted);
+  }
+  
+  /**
+   * The `k` largest elements along each last dimensional slice.
+   */
+  public Output<T> values() {
+    return values;
+  }
+  
+  /**
+   * The indices of `values` within the last dimension of `input`.
+   */
+  public Output<TInt32> indices() {
+    return indices;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "TopKV2";
+  
+  private Output<T> values;
+  private Output<TInt32> indices;
+  
+  private TopK(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    values = operation.output(outputIdx++);
+    indices = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/raw/SoftmaxCrossEntropyWithLogits.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/raw/SoftmaxCrossEntropyWithLogits.java
new file mode 100644
index 00000000000..1a5011c713a
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/raw/SoftmaxCrossEntropyWithLogits.java
@@ -0,0 +1,85 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.nn.raw;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Computes softmax cross entropy cost and gradients to backpropagate.
+ * <p>
+ * Inputs are the logits, not probabilities.
+ * 
+ * @param <T> data type for {@code loss()} output
+ */
+@Operator(group = "nn.raw")
+public final class SoftmaxCrossEntropyWithLogits<T extends TNumber> extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new SoftmaxCrossEntropyWithLogits operation.
+   * 
+   * @param scope current scope
+   * @param features batch_size x num_classes matrix
+   * @param labels batch_size x num_classes matrix
+   * The caller must ensure that each batch of labels represents a valid
+   * probability distribution.
+   * @return a new instance of SoftmaxCrossEntropyWithLogits
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> SoftmaxCrossEntropyWithLogits<T> create(Scope scope, Operand<T> features, Operand<T> labels) {
+    OperationBuilder opBuilder = scope.env().opBuilder("SoftmaxCrossEntropyWithLogits", scope.makeOpName("SoftmaxCrossEntropyWithLogits"));
+    opBuilder.addInput(features.asOutput());
+    opBuilder.addInput(labels.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new SoftmaxCrossEntropyWithLogits<T>(opBuilder.build());
+  }
+  
+  /**
+   * Per example loss (batch_size vector).
+   */
+  public Output<T> loss() {
+    return loss;
+  }
+  
+  /**
+   * backpropagated gradients (batch_size x num_classes matrix).
+   */
+  public Output<T> backprop() {
+    return backprop;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "SoftmaxCrossEntropyWithLogits";
+  
+  private Output<T> loss;
+  private Output<T> backprop;
+  
+  private SoftmaxCrossEntropyWithLogits(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    loss = operation.output(outputIdx++);
+    backprop = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/raw/SparseSoftmaxCrossEntropyWithLogits.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/raw/SparseSoftmaxCrossEntropyWithLogits.java
new file mode 100644
index 00000000000..c8dcac1a84a
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/nn/raw/SparseSoftmaxCrossEntropyWithLogits.java
@@ -0,0 +1,89 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.nn.raw;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Computes softmax cross entropy cost and gradients to backpropagate.
+ * <p>
+ * Unlike `SoftmaxCrossEntropyWithLogits`, this operation does not accept
+ * a matrix of label probabilities, but rather a single label per row
+ * of features.  This label is considered to have probability 1.0 for the
+ * given row.
+ * <p>
+ * Inputs are the logits, not probabilities.
+ * 
+ * @param <T> data type for {@code loss()} output
+ */
+@Operator(group = "nn.raw")
+public final class SparseSoftmaxCrossEntropyWithLogits<T extends TNumber> extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new SparseSoftmaxCrossEntropyWithLogits operation.
+   * 
+   * @param scope current scope
+   * @param features batch_size x num_classes matrix
+   * @param labels batch_size vector with values in [0, num_classes).
+   * This is the label for the given minibatch entry.
+   * @return a new instance of SparseSoftmaxCrossEntropyWithLogits
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber, U extends TNumber> SparseSoftmaxCrossEntropyWithLogits<T> create(Scope scope, Operand<T> features, Operand<U> labels) {
+    OperationBuilder opBuilder = scope.env().opBuilder("SparseSoftmaxCrossEntropyWithLogits", scope.makeOpName("SparseSoftmaxCrossEntropyWithLogits"));
+    opBuilder.addInput(features.asOutput());
+    opBuilder.addInput(labels.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new SparseSoftmaxCrossEntropyWithLogits<T>(opBuilder.build());
+  }
+  
+  /**
+   * Per example loss (batch_size vector).
+   */
+  public Output<T> loss() {
+    return loss;
+  }
+  
+  /**
+   * backpropagated gradients (batch_size x num_classes matrix).
+   */
+  public Output<T> backprop() {
+    return backprop;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "SparseSoftmaxCrossEntropyWithLogits";
+  
+  private Output<T> loss;
+  private Output<T> backprop;
+  
+  private SparseSoftmaxCrossEntropyWithLogits(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    loss = operation.output(outputIdx++);
+    backprop = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/Dequantize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/Dequantize.java
new file mode 100644
index 00000000000..6c9484cab55
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/Dequantize.java
@@ -0,0 +1,219 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.quantization;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Dequantize the 'input' tensor into a float or bfloat16 Tensor.
+ * <p>
+ * [min_range, max_range] are scalar floats that specify the range for
+ * the output. The 'mode' attribute controls exactly which calculations are
+ * used to convert the float values to their quantized equivalents.
+ * <p>
+ * In 'MIN_COMBINED' mode, each value of the tensor will undergo the following:
+ * <pre>{@code
+ * if T == qint8: in[i] += (range(T) + 1)/ 2.0
+ * out[i] = min_range + (in[i]* (max_range - min_range) / range(T))
+ * }</pre>
+ * here `range(T) = numeric_limits<T>::max() - numeric_limits<T>::min()`
+ * <p>
+ * <i>MIN_COMBINED Mode Example</i>
+ * <p>
+ * If the input comes from a QuantizedRelu6, the output type is
+ * quint8 (range of 0-255) but the possible range of QuantizedRelu6 is
+ * 0-6.  The min_range and max_range values are therefore 0.0 and 6.0.
+ * Dequantize on quint8 will take each value, cast to float, and multiply
+ * by 6 / 255.
+ * Note that if quantizedtype is qint8, the operation will additionally add
+ * each value by 128 prior to casting.
+ * <p>
+ * If the mode is 'MIN_FIRST', then this approach is used:
+ * <pre>{@code
+ * num_discrete_values = 1 << (# of bits in T)
+ * range_adjust = num_discrete_values / (num_discrete_values - 1)
+ * range = (range_max - range_min) * range_adjust
+ * range_scale = range / num_discrete_values
+ * const double offset_input = static_cast<double>(input) - lowest_quantized;
+ * result = range_min + ((input - numeric_limits<T>::min()) * range_scale)
+ * }</pre>
+ * If the mode is `SCALED`, dequantization is performed by multiplying each
+ * input value by a scaling_factor. (Thus an input of 0 always maps to 0.0).
+ * <p>
+ * The scaling_factor is determined from `min_range`, `max_range`, and
+ * `narrow_range` in a way that is compatible with `QuantizeAndDequantize{V2|V3}`
+ * and `QuantizeV2`, using the following algorithm:
+ * <pre>{@code
+ *   const int min_expected_T = std::numeric_limits<T>::min() +
+ *     (narrow_range ? 1 : 0);
+ *   const int max_expected_T = std::numeric_limits<T>::max();
+ *   const float max_expected_T = std::numeric_limits<float>::max();
+ * 
+ *   const float scale_factor =
+ *     (std::numeric_limits<T>::min() == 0) ? (max_range / max_expected_T)
+ *                                          : std::max(min_range / min_expected_T,
+ *                                                     max_range / max_expected_T);
+ * }</pre>
+ * 
+ * 
+ * @param <U> data type for {@code output()} output
+ */
+@Operator(group = "quantization")
+public final class Dequantize<U extends TNumber> extends RawOp implements Operand<U> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.quantization.Dequantize}
+   */
+  public static class Options {
+    
+    /**
+     * @param mode 
+     */
+    public Options mode(String mode) {
+      this.mode = mode;
+      return this;
+    }
+    
+    /**
+     * @param narrowRange 
+     */
+    public Options narrowRange(Boolean narrowRange) {
+      this.narrowRange = narrowRange;
+      return this;
+    }
+    
+    /**
+     * @param axis 
+     */
+    public Options axis(Long axis) {
+      this.axis = axis;
+      return this;
+    }
+    
+    private String mode;
+    private Boolean narrowRange;
+    private Long axis;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new Dequantize operation.
+   * 
+   * @param scope current scope
+   * @param input 
+   * @param minRange The minimum scalar value possibly produced for the input.
+   * @param maxRange The maximum scalar value possibly produced for the input.
+   * @param dtype Type of the output tensor. Currently Dequantize supports float and bfloat16.
+   * If 'dtype' is 'bfloat16', it only supports 'MIN_COMBINED' mode.
+   * @param options carries optional attributes values
+   * @return a new instance of Dequantize
+   */
+  @Endpoint(describeByClass = true)
+  public static <U extends TNumber, T extends TType> Dequantize<U> create(Scope scope, Operand<T> input, Operand<TFloat32> minRange, Operand<TFloat32> maxRange, Class<U> dtype, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Dequantize", scope.makeOpName("Dequantize"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(minRange.asOutput());
+    opBuilder.addInput(maxRange.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("dtype", dtype);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.mode != null) {
+          opBuilder.setAttr("mode", opts.mode);
+        }
+        if (opts.narrowRange != null) {
+          opBuilder.setAttr("narrow_range", opts.narrowRange);
+        }
+        if (opts.axis != null) {
+          opBuilder.setAttr("axis", opts.axis);
+        }
+      }
+    }
+    return new Dequantize<U>(opBuilder.build());
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new Dequantize operation using default output types.
+   * 
+   * @param scope current scope
+   * @param input 
+   * @param minRange The minimum scalar value possibly produced for the input.
+   * @param maxRange The maximum scalar value possibly produced for the input.
+   * @param options carries optional attributes values
+   * @return a new instance of Dequantize
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> Dequantize<TFloat32> create(Scope scope, Operand<T> input, Operand<TFloat32> minRange, Operand<TFloat32> maxRange, Options... options) {
+    return create(scope, input, minRange, maxRange, TFloat32.class, options);
+  }
+  
+  /**
+   * @param mode 
+   */
+  public static Options mode(String mode) {
+    return new Options().mode(mode);
+  }
+  
+  /**
+   * @param narrowRange 
+   */
+  public static Options narrowRange(Boolean narrowRange) {
+    return new Options().narrowRange(narrowRange);
+  }
+  
+  /**
+   * @param axis 
+   */
+  public static Options axis(Long axis) {
+    return new Options().axis(axis);
+  }
+  
+  /**
+   */
+  public Output<U> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<U> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Dequantize";
+  
+  private Output<U> output;
+  
+  private Dequantize(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/FakeQuantWithMinMaxArgs.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/FakeQuantWithMinMaxArgs.java
new file mode 100644
index 00000000000..986097339a9
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/FakeQuantWithMinMaxArgs.java
@@ -0,0 +1,196 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.quantization;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+
+/**
+ * Fake-quantize the 'inputs' tensor, type float to 'outputs' tensor of same type.
+ * <p>
+ * Attributes
+ * <ul>
+ * <li>
+ * `[min; max]` define the clamping range for the `inputs` data.
+ * </li>
+ * <li>
+ * `inputs` values are quantized into the quantization range (
+ * `[0; 2^num_bits - 1]` when `narrow_range` is false and `[1; 2^num_bits - 1]`
+ * when it is true) and then de-quantized and output as floats in `[min; max]`
+ * interval.
+ * </li>
+ * <li>
+ * `num_bits` is the bitwidth of the quantization; between 2 and 16, inclusive.
+ * </li>
+ * </ul>
+ * Before quantization, `min` and `max` values are adjusted with the following
+ * logic.
+ * It is suggested to have `min <= 0 <= max`. If `0` is not in the range of values,
+ * the behavior can be unexpected:
+ * <ul>
+ * <li>
+ * If `0 < min < max`: `min_adj = 0` and `max_adj = max - min`.
+ * </li>
+ * <li>
+ * If `min < max < 0`: `min_adj = min - max` and `max_adj = 0`.
+ * </li>
+ * <li>
+ * If `min <= 0 <= max`: `scale = (max - min) / (2^num_bits - 1) `,
+ * `min_adj = scale * round(min / scale)` and `max_adj = max + min_adj - min`.
+ * </li>
+ * </ul>
+ * Quantization is called fake since the output is still in floating point.
+ */
+@Operator(group = "quantization")
+public final class FakeQuantWithMinMaxArgs extends RawOp implements Operand<TFloat32> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.quantization.FakeQuantWithMinMaxArgs}
+   */
+  public static class Options {
+    
+    /**
+     * @param min 
+     */
+    public Options min(Float min) {
+      this.min = min;
+      return this;
+    }
+    
+    /**
+     * @param max 
+     */
+    public Options max(Float max) {
+      this.max = max;
+      return this;
+    }
+    
+    /**
+     * @param numBits 
+     */
+    public Options numBits(Long numBits) {
+      this.numBits = numBits;
+      return this;
+    }
+    
+    /**
+     * @param narrowRange 
+     */
+    public Options narrowRange(Boolean narrowRange) {
+      this.narrowRange = narrowRange;
+      return this;
+    }
+    
+    private Float min;
+    private Float max;
+    private Long numBits;
+    private Boolean narrowRange;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new FakeQuantWithMinMaxArgs operation.
+   * 
+   * @param scope current scope
+   * @param inputs 
+   * @param options carries optional attributes values
+   * @return a new instance of FakeQuantWithMinMaxArgs
+   */
+  @Endpoint(describeByClass = true)
+  public static FakeQuantWithMinMaxArgs create(Scope scope, Operand<TFloat32> inputs, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("FakeQuantWithMinMaxArgs", scope.makeOpName("FakeQuantWithMinMaxArgs"));
+    opBuilder.addInput(inputs.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.min != null) {
+          opBuilder.setAttr("min", opts.min);
+        }
+        if (opts.max != null) {
+          opBuilder.setAttr("max", opts.max);
+        }
+        if (opts.numBits != null) {
+          opBuilder.setAttr("num_bits", opts.numBits);
+        }
+        if (opts.narrowRange != null) {
+          opBuilder.setAttr("narrow_range", opts.narrowRange);
+        }
+      }
+    }
+    return new FakeQuantWithMinMaxArgs(opBuilder.build());
+  }
+  
+  /**
+   * @param min 
+   */
+  public static Options min(Float min) {
+    return new Options().min(min);
+  }
+  
+  /**
+   * @param max 
+   */
+  public static Options max(Float max) {
+    return new Options().max(max);
+  }
+  
+  /**
+   * @param numBits 
+   */
+  public static Options numBits(Long numBits) {
+    return new Options().numBits(numBits);
+  }
+  
+  /**
+   * @param narrowRange 
+   */
+  public static Options narrowRange(Boolean narrowRange) {
+    return new Options().narrowRange(narrowRange);
+  }
+  
+  /**
+   */
+  public Output<TFloat32> outputs() {
+    return outputs;
+  }
+  
+  @Override
+  public Output<TFloat32> asOutput() {
+    return outputs;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "FakeQuantWithMinMaxArgs";
+  
+  private Output<TFloat32> outputs;
+  
+  private FakeQuantWithMinMaxArgs(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    outputs = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/FakeQuantWithMinMaxArgsGradient.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/FakeQuantWithMinMaxArgsGradient.java
new file mode 100644
index 00000000000..62057eaefed
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/FakeQuantWithMinMaxArgsGradient.java
@@ -0,0 +1,167 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.quantization;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+
+/**
+ * Compute gradients for a FakeQuantWithMinMaxArgs operation.
+ */
+@Operator(group = "quantization")
+public final class FakeQuantWithMinMaxArgsGradient extends RawOp implements Operand<TFloat32> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.quantization.FakeQuantWithMinMaxArgsGradient}
+   */
+  public static class Options {
+    
+    /**
+     * @param min 
+     */
+    public Options min(Float min) {
+      this.min = min;
+      return this;
+    }
+    
+    /**
+     * @param max 
+     */
+    public Options max(Float max) {
+      this.max = max;
+      return this;
+    }
+    
+    /**
+     * @param numBits 
+     */
+    public Options numBits(Long numBits) {
+      this.numBits = numBits;
+      return this;
+    }
+    
+    /**
+     * @param narrowRange 
+     */
+    public Options narrowRange(Boolean narrowRange) {
+      this.narrowRange = narrowRange;
+      return this;
+    }
+    
+    private Float min;
+    private Float max;
+    private Long numBits;
+    private Boolean narrowRange;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new FakeQuantWithMinMaxArgsGradient operation.
+   * 
+   * @param scope current scope
+   * @param gradients Backpropagated gradients above the FakeQuantWithMinMaxArgs operation.
+   * @param inputs Values passed as inputs to the FakeQuantWithMinMaxArgs operation.
+   * @param options carries optional attributes values
+   * @return a new instance of FakeQuantWithMinMaxArgsGradient
+   */
+  @Endpoint(describeByClass = true)
+  public static FakeQuantWithMinMaxArgsGradient create(Scope scope, Operand<TFloat32> gradients, Operand<TFloat32> inputs, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("FakeQuantWithMinMaxArgsGradient", scope.makeOpName("FakeQuantWithMinMaxArgsGradient"));
+    opBuilder.addInput(gradients.asOutput());
+    opBuilder.addInput(inputs.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.min != null) {
+          opBuilder.setAttr("min", opts.min);
+        }
+        if (opts.max != null) {
+          opBuilder.setAttr("max", opts.max);
+        }
+        if (opts.numBits != null) {
+          opBuilder.setAttr("num_bits", opts.numBits);
+        }
+        if (opts.narrowRange != null) {
+          opBuilder.setAttr("narrow_range", opts.narrowRange);
+        }
+      }
+    }
+    return new FakeQuantWithMinMaxArgsGradient(opBuilder.build());
+  }
+  
+  /**
+   * @param min 
+   */
+  public static Options min(Float min) {
+    return new Options().min(min);
+  }
+  
+  /**
+   * @param max 
+   */
+  public static Options max(Float max) {
+    return new Options().max(max);
+  }
+  
+  /**
+   * @param numBits 
+   */
+  public static Options numBits(Long numBits) {
+    return new Options().numBits(numBits);
+  }
+  
+  /**
+   * @param narrowRange 
+   */
+  public static Options narrowRange(Boolean narrowRange) {
+    return new Options().narrowRange(narrowRange);
+  }
+  
+  /**
+   * Backpropagated gradients below the FakeQuantWithMinMaxArgs operation:
+   * `gradients * (inputs >= min && inputs <= max)`.
+   */
+  public Output<TFloat32> backprops() {
+    return backprops;
+  }
+  
+  @Override
+  public Output<TFloat32> asOutput() {
+    return backprops;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "FakeQuantWithMinMaxArgsGradient";
+  
+  private Output<TFloat32> backprops;
+  
+  private FakeQuantWithMinMaxArgsGradient(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    backprops = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/FakeQuantWithMinMaxVars.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/FakeQuantWithMinMaxVars.java
new file mode 100644
index 00000000000..b2c1ae3880f
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/FakeQuantWithMinMaxVars.java
@@ -0,0 +1,166 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.quantization;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+
+/**
+ * Fake-quantize the 'inputs' tensor of type float via global float scalars
+ * <p>
+ * Fake-quantize the `inputs` tensor of type float via global float scalars
+ * `min` and `max` to `outputs` tensor of same shape as `inputs`.
+ * <p>
+ * Attributes
+ * <ul>
+ * <li>
+ * `[min; max]` define the clamping range for the `inputs` data.
+ * </li>
+ * <li>
+ * `inputs` values are quantized into the quantization range (
+ * `[0; 2^num_bits - 1]` when `narrow_range` is false and `[1; 2^num_bits - 1]`
+ * when it is true) and then de-quantized and output as floats in `[min; max]`
+ * interval.
+ * </li>
+ * <li>
+ * `num_bits` is the bitwidth of the quantization; between 2 and 16, inclusive.
+ * </li>
+ * </ul>
+ * Before quantization, `min` and `max` values are adjusted with the following
+ * logic.
+ * It is suggested to have `min <= 0 <= max`. If `0` is not in the range of values,
+ * the behavior can be unexpected:
+ * <ul>
+ * <li>
+ * If `0 < min < max`: `min_adj = 0` and `max_adj = max - min`.
+ * </li>
+ * <li>
+ * If `min < max < 0`: `min_adj = min - max` and `max_adj = 0`.
+ * </li>
+ * <li>
+ * If `min <= 0 <= max`: `scale = (max - min) / (2^num_bits - 1) `,
+ * `min_adj = scale * round(min / scale)` and `max_adj = max + min_adj - min`.
+ * </li>
+ * </ul>
+ * This operation has a gradient and thus allows for training `min` and `max`
+ * values.
+ */
+@Operator(group = "quantization")
+public final class FakeQuantWithMinMaxVars extends RawOp implements Operand<TFloat32> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.quantization.FakeQuantWithMinMaxVars}
+   */
+  public static class Options {
+    
+    /**
+     * @param numBits 
+     */
+    public Options numBits(Long numBits) {
+      this.numBits = numBits;
+      return this;
+    }
+    
+    /**
+     * @param narrowRange 
+     */
+    public Options narrowRange(Boolean narrowRange) {
+      this.narrowRange = narrowRange;
+      return this;
+    }
+    
+    private Long numBits;
+    private Boolean narrowRange;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new FakeQuantWithMinMaxVars operation.
+   * 
+   * @param scope current scope
+   * @param inputs 
+   * @param min 
+   * @param max 
+   * @param options carries optional attributes values
+   * @return a new instance of FakeQuantWithMinMaxVars
+   */
+  @Endpoint(describeByClass = true)
+  public static FakeQuantWithMinMaxVars create(Scope scope, Operand<TFloat32> inputs, Operand<TFloat32> min, Operand<TFloat32> max, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("FakeQuantWithMinMaxVars", scope.makeOpName("FakeQuantWithMinMaxVars"));
+    opBuilder.addInput(inputs.asOutput());
+    opBuilder.addInput(min.asOutput());
+    opBuilder.addInput(max.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.numBits != null) {
+          opBuilder.setAttr("num_bits", opts.numBits);
+        }
+        if (opts.narrowRange != null) {
+          opBuilder.setAttr("narrow_range", opts.narrowRange);
+        }
+      }
+    }
+    return new FakeQuantWithMinMaxVars(opBuilder.build());
+  }
+  
+  /**
+   * @param numBits 
+   */
+  public static Options numBits(Long numBits) {
+    return new Options().numBits(numBits);
+  }
+  
+  /**
+   * @param narrowRange 
+   */
+  public static Options narrowRange(Boolean narrowRange) {
+    return new Options().narrowRange(narrowRange);
+  }
+  
+  /**
+   */
+  public Output<TFloat32> outputs() {
+    return outputs;
+  }
+  
+  @Override
+  public Output<TFloat32> asOutput() {
+    return outputs;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "FakeQuantWithMinMaxVars";
+  
+  private Output<TFloat32> outputs;
+  
+  private FakeQuantWithMinMaxVars(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    outputs = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/FakeQuantWithMinMaxVarsGradient.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/FakeQuantWithMinMaxVarsGradient.java
new file mode 100644
index 00000000000..e9fcd339a75
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/FakeQuantWithMinMaxVarsGradient.java
@@ -0,0 +1,149 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.quantization;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+
+/**
+ * Compute gradients for a FakeQuantWithMinMaxVars operation.
+ */
+@Operator(group = "quantization")
+public final class FakeQuantWithMinMaxVarsGradient extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.quantization.FakeQuantWithMinMaxVarsGradient}
+   */
+  public static class Options {
+    
+    /**
+     * @param numBits The bitwidth of the quantization; between 2 and 8, inclusive.
+     */
+    public Options numBits(Long numBits) {
+      this.numBits = numBits;
+      return this;
+    }
+    
+    /**
+     * @param narrowRange Whether to quantize into 2^num_bits - 1 distinct values.
+     */
+    public Options narrowRange(Boolean narrowRange) {
+      this.narrowRange = narrowRange;
+      return this;
+    }
+    
+    private Long numBits;
+    private Boolean narrowRange;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new FakeQuantWithMinMaxVarsGradient operation.
+   * 
+   * @param scope current scope
+   * @param gradients Backpropagated gradients above the FakeQuantWithMinMaxVars operation.
+   * @param inputs Values passed as inputs to the FakeQuantWithMinMaxVars operation.
+   * min, max: Quantization interval, scalar floats.
+   * @param min 
+   * @param max 
+   * @param options carries optional attributes values
+   * @return a new instance of FakeQuantWithMinMaxVarsGradient
+   */
+  @Endpoint(describeByClass = true)
+  public static FakeQuantWithMinMaxVarsGradient create(Scope scope, Operand<TFloat32> gradients, Operand<TFloat32> inputs, Operand<TFloat32> min, Operand<TFloat32> max, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("FakeQuantWithMinMaxVarsGradient", scope.makeOpName("FakeQuantWithMinMaxVarsGradient"));
+    opBuilder.addInput(gradients.asOutput());
+    opBuilder.addInput(inputs.asOutput());
+    opBuilder.addInput(min.asOutput());
+    opBuilder.addInput(max.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.numBits != null) {
+          opBuilder.setAttr("num_bits", opts.numBits);
+        }
+        if (opts.narrowRange != null) {
+          opBuilder.setAttr("narrow_range", opts.narrowRange);
+        }
+      }
+    }
+    return new FakeQuantWithMinMaxVarsGradient(opBuilder.build());
+  }
+  
+  /**
+   * @param numBits The bitwidth of the quantization; between 2 and 8, inclusive.
+   */
+  public static Options numBits(Long numBits) {
+    return new Options().numBits(numBits);
+  }
+  
+  /**
+   * @param narrowRange Whether to quantize into 2^num_bits - 1 distinct values.
+   */
+  public static Options narrowRange(Boolean narrowRange) {
+    return new Options().narrowRange(narrowRange);
+  }
+  
+  /**
+   * Backpropagated gradients w.r.t. inputs:
+   * `gradients * (inputs >= min && inputs <= max)`.
+   */
+  public Output<TFloat32> backpropsWrtInput() {
+    return backpropsWrtInput;
+  }
+  
+  /**
+   * Backpropagated gradients w.r.t. min parameter:
+   * `sum(gradients * (inputs < min))`.
+   */
+  public Output<TFloat32> backpropWrtMin() {
+    return backpropWrtMin;
+  }
+  
+  /**
+   * Backpropagated gradients w.r.t. max parameter:
+   * `sum(gradients * (inputs > max))`.
+   */
+  public Output<TFloat32> backpropWrtMax() {
+    return backpropWrtMax;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "FakeQuantWithMinMaxVarsGradient";
+  
+  private Output<TFloat32> backpropsWrtInput;
+  private Output<TFloat32> backpropWrtMin;
+  private Output<TFloat32> backpropWrtMax;
+  
+  private FakeQuantWithMinMaxVarsGradient(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    backpropsWrtInput = operation.output(outputIdx++);
+    backpropWrtMin = operation.output(outputIdx++);
+    backpropWrtMax = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/FakeQuantWithMinMaxVarsPerChannel.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/FakeQuantWithMinMaxVarsPerChannel.java
new file mode 100644
index 00000000000..ce5649f6c0b
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/FakeQuantWithMinMaxVarsPerChannel.java
@@ -0,0 +1,167 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.quantization;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+
+/**
+ * Fake-quantize the 'inputs' tensor of type float via per-channel floats
+ * <p>
+ * Fake-quantize the `inputs` tensor of type float per-channel and one of the
+ * shapes: `[d]`, `[b, d]` `[b, h, w, d]` via per-channel floats `min` and `max`
+ * of shape `[d]` to `outputs` tensor of same shape as `inputs`.
+ * <p>
+ * Attributes
+ * <ul>
+ * <li>
+ * `[min; max]` define the clamping range for the `inputs` data.
+ * </li>
+ * <li>
+ * `inputs` values are quantized into the quantization range (
+ * `[0; 2^num_bits - 1]` when `narrow_range` is false and `[1; 2^num_bits - 1]`
+ * when it is true) and then de-quantized and output as floats in `[min; max]`
+ * interval.
+ * </li>
+ * <li>
+ * `num_bits` is the bitwidth of the quantization; between 2 and 16, inclusive.
+ * </li>
+ * </ul>
+ * Before quantization, `min` and `max` values are adjusted with the following
+ * logic.
+ * It is suggested to have `min <= 0 <= max`. If `0` is not in the range of values,
+ * the behavior can be unexpected:
+ * <ul>
+ * <li>
+ * If `0 < min < max`: `min_adj = 0` and `max_adj = max - min`.
+ * </li>
+ * <li>
+ * If `min < max < 0`: `min_adj = min - max` and `max_adj = 0`.
+ * </li>
+ * <li>
+ * If `min <= 0 <= max`: `scale = (max - min) / (2^num_bits - 1) `,
+ * `min_adj = scale * round(min / scale)` and `max_adj = max + min_adj - min`.
+ * </li>
+ * </ul>
+ * This operation has a gradient and thus allows for training `min` and `max`
+ * values.
+ */
+@Operator(group = "quantization")
+public final class FakeQuantWithMinMaxVarsPerChannel extends RawOp implements Operand<TFloat32> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.quantization.FakeQuantWithMinMaxVarsPerChannel}
+   */
+  public static class Options {
+    
+    /**
+     * @param numBits 
+     */
+    public Options numBits(Long numBits) {
+      this.numBits = numBits;
+      return this;
+    }
+    
+    /**
+     * @param narrowRange 
+     */
+    public Options narrowRange(Boolean narrowRange) {
+      this.narrowRange = narrowRange;
+      return this;
+    }
+    
+    private Long numBits;
+    private Boolean narrowRange;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new FakeQuantWithMinMaxVarsPerChannel operation.
+   * 
+   * @param scope current scope
+   * @param inputs 
+   * @param min 
+   * @param max 
+   * @param options carries optional attributes values
+   * @return a new instance of FakeQuantWithMinMaxVarsPerChannel
+   */
+  @Endpoint(describeByClass = true)
+  public static FakeQuantWithMinMaxVarsPerChannel create(Scope scope, Operand<TFloat32> inputs, Operand<TFloat32> min, Operand<TFloat32> max, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("FakeQuantWithMinMaxVarsPerChannel", scope.makeOpName("FakeQuantWithMinMaxVarsPerChannel"));
+    opBuilder.addInput(inputs.asOutput());
+    opBuilder.addInput(min.asOutput());
+    opBuilder.addInput(max.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.numBits != null) {
+          opBuilder.setAttr("num_bits", opts.numBits);
+        }
+        if (opts.narrowRange != null) {
+          opBuilder.setAttr("narrow_range", opts.narrowRange);
+        }
+      }
+    }
+    return new FakeQuantWithMinMaxVarsPerChannel(opBuilder.build());
+  }
+  
+  /**
+   * @param numBits 
+   */
+  public static Options numBits(Long numBits) {
+    return new Options().numBits(numBits);
+  }
+  
+  /**
+   * @param narrowRange 
+   */
+  public static Options narrowRange(Boolean narrowRange) {
+    return new Options().narrowRange(narrowRange);
+  }
+  
+  /**
+   */
+  public Output<TFloat32> outputs() {
+    return outputs;
+  }
+  
+  @Override
+  public Output<TFloat32> asOutput() {
+    return outputs;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "FakeQuantWithMinMaxVarsPerChannel";
+  
+  private Output<TFloat32> outputs;
+  
+  private FakeQuantWithMinMaxVarsPerChannel(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    outputs = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/FakeQuantWithMinMaxVarsPerChannelGradient.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/FakeQuantWithMinMaxVarsPerChannelGradient.java
new file mode 100644
index 00000000000..4fba9b56115
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/FakeQuantWithMinMaxVarsPerChannelGradient.java
@@ -0,0 +1,152 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.quantization;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+
+/**
+ * Compute gradients for a FakeQuantWithMinMaxVarsPerChannel operation.
+ */
+@Operator(group = "quantization")
+public final class FakeQuantWithMinMaxVarsPerChannelGradient extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.quantization.FakeQuantWithMinMaxVarsPerChannelGradient}
+   */
+  public static class Options {
+    
+    /**
+     * @param numBits The bitwidth of the quantization; between 2 and 16, inclusive.
+     */
+    public Options numBits(Long numBits) {
+      this.numBits = numBits;
+      return this;
+    }
+    
+    /**
+     * @param narrowRange Whether to quantize into 2^num_bits - 1 distinct values.
+     */
+    public Options narrowRange(Boolean narrowRange) {
+      this.narrowRange = narrowRange;
+      return this;
+    }
+    
+    private Long numBits;
+    private Boolean narrowRange;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new FakeQuantWithMinMaxVarsPerChannelGradient operation.
+   * 
+   * @param scope current scope
+   * @param gradients Backpropagated gradients above the FakeQuantWithMinMaxVars operation,
+   * shape one of: `[d]`, `[b, d]`,  `[b, h, w, d]`.
+   * @param inputs Values passed as inputs to the FakeQuantWithMinMaxVars operation, shape
+   *   same as `gradients`.
+   * min, max: Quantization interval, floats of shape `[d]`.
+   * @param min 
+   * @param max 
+   * @param options carries optional attributes values
+   * @return a new instance of FakeQuantWithMinMaxVarsPerChannelGradient
+   */
+  @Endpoint(describeByClass = true)
+  public static FakeQuantWithMinMaxVarsPerChannelGradient create(Scope scope, Operand<TFloat32> gradients, Operand<TFloat32> inputs, Operand<TFloat32> min, Operand<TFloat32> max, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("FakeQuantWithMinMaxVarsPerChannelGradient", scope.makeOpName("FakeQuantWithMinMaxVarsPerChannelGradient"));
+    opBuilder.addInput(gradients.asOutput());
+    opBuilder.addInput(inputs.asOutput());
+    opBuilder.addInput(min.asOutput());
+    opBuilder.addInput(max.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.numBits != null) {
+          opBuilder.setAttr("num_bits", opts.numBits);
+        }
+        if (opts.narrowRange != null) {
+          opBuilder.setAttr("narrow_range", opts.narrowRange);
+        }
+      }
+    }
+    return new FakeQuantWithMinMaxVarsPerChannelGradient(opBuilder.build());
+  }
+  
+  /**
+   * @param numBits The bitwidth of the quantization; between 2 and 16, inclusive.
+   */
+  public static Options numBits(Long numBits) {
+    return new Options().numBits(numBits);
+  }
+  
+  /**
+   * @param narrowRange Whether to quantize into 2^num_bits - 1 distinct values.
+   */
+  public static Options narrowRange(Boolean narrowRange) {
+    return new Options().narrowRange(narrowRange);
+  }
+  
+  /**
+   * Backpropagated gradients w.r.t. inputs, shape same as
+   * `inputs`:
+   *   `gradients * (inputs >= min && inputs <= max)`.
+   */
+  public Output<TFloat32> backpropsWrtInput() {
+    return backpropsWrtInput;
+  }
+  
+  /**
+   * Backpropagated gradients w.r.t. min parameter, shape `[d]`:
+   * `sum_per_d(gradients * (inputs < min))`.
+   */
+  public Output<TFloat32> backpropWrtMin() {
+    return backpropWrtMin;
+  }
+  
+  /**
+   * Backpropagated gradients w.r.t. max parameter, shape `[d]`:
+   * `sum_per_d(gradients * (inputs > max))`.
+   */
+  public Output<TFloat32> backpropWrtMax() {
+    return backpropWrtMax;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "FakeQuantWithMinMaxVarsPerChannelGradient";
+  
+  private Output<TFloat32> backpropsWrtInput;
+  private Output<TFloat32> backpropWrtMin;
+  private Output<TFloat32> backpropWrtMax;
+  
+  private FakeQuantWithMinMaxVarsPerChannelGradient(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    backpropsWrtInput = operation.output(outputIdx++);
+    backpropWrtMin = operation.output(outputIdx++);
+    backpropWrtMax = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/Quantize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/Quantize.java
new file mode 100644
index 00000000000..61370659157
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/Quantize.java
@@ -0,0 +1,327 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.quantization;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Quantize the 'input' tensor of type float to 'output' tensor of type 'T'.
+ * <p>
+ * [min_range, max_range] are scalar floats that specify the range for
+ * the 'input' data. The 'mode' attribute controls exactly which calculations are
+ * used to convert the float values to their quantized equivalents.  The
+ * 'round_mode' attribute controls which rounding tie-breaking algorithm is used
+ * when rounding float values to their quantized equivalents.
+ * <p>
+ * In 'MIN_COMBINED' mode, each value of the tensor will undergo the following:
+ * <pre>{@code
+ * out[i] = (in[i] - min_range) * range(T) / (max_range - min_range)
+ * if T == qint8: out[i] -= (range(T) + 1) / 2.0
+ * }</pre>
+ * here `range(T) = numeric_limits<T>::max() - numeric_limits<T>::min()`
+ * <p>
+ * <i>MIN_COMBINED Mode Example</i>
+ * <p>
+ * Assume the input is type float and has a possible range of [0.0, 6.0] and the
+ * output type is quint8 ([0, 255]). The min_range and max_range values should be
+ * specified as 0.0 and 6.0. Quantizing from float to quint8 will multiply each
+ * value of the input by 255/6 and cast to quint8.
+ * <p>
+ * If the output type was qint8 ([-128, 127]), the operation will additionally
+ * subtract each value by 128 prior to casting, so that the range of values aligns
+ * with the range of qint8.
+ * <p>
+ * If the mode is 'MIN_FIRST', then this approach is used:
+ * <pre>{@code
+ * num_discrete_values = 1 << (# of bits in T)
+ * range_adjust = num_discrete_values / (num_discrete_values - 1)
+ * range = (range_max - range_min) * range_adjust
+ * range_scale = num_discrete_values / range
+ * quantized = round(input * range_scale) - round(range_min * range_scale) +
+ *   numeric_limits<T>::min()
+ * quantized = max(quantized, numeric_limits<T>::min())
+ * quantized = min(quantized, numeric_limits<T>::max())
+ * }</pre>
+ * The biggest difference between this and MIN_COMBINED is that the minimum range
+ * is rounded first, before it's subtracted from the rounded value. With
+ * MIN_COMBINED, a small bias is introduced where repeated iterations of quantizing
+ * and dequantizing will introduce a larger and larger error.
+ * <p>
+ * <i>SCALED mode Example</i>
+ * <p>
+ * `SCALED` mode matches the quantization approach used in
+ * `QuantizeAndDequantize{V2|V3}`.
+ * <p>
+ * If the mode is `SCALED`, the quantization is performed by multiplying each
+ * input value by a scaling_factor.
+ * The scaling_factor is determined from `min_range` and `max_range` to be as large
+ * as possible such that the range from `min_range` to `max_range` is representable
+ * within values of type T.
+ * <pre>{@code
+ *   const int min_T = std::numeric_limits<T>::min();
+ *   const int max_T = std::numeric_limits<T>::max();
+ *   const float max_float = std::numeric_limits<float>::max();
+ * 
+ *   const float scale_factor_from_min_side =
+ *       (min_T * min_range > 0) ? min_T / min_range : max_float;
+ *   const float scale_factor_from_max_side =
+ *       (max_T * max_range > 0) ? max_T / max_range : max_float;
+ * 
+ *   const float scale_factor = std::min(scale_factor_from_min_side,
+ *                                       scale_factor_from_max_side);
+ * }</pre>
+ * We next use the scale_factor to adjust min_range and max_range as follows:
+ * <pre>{@code
+ *       min_range = min_T / scale_factor;
+ *       max_range = max_T / scale_factor;
+ * }</pre>
+ * e.g. if T = qint8, and initially min_range = -10, and max_range = 9, we would
+ * compare -128/-10.0 = 12.8 to 127/9.0 = 14.11, and set scaling_factor = 12.8
+ * In this case, min_range would remain -10, but max_range would be adjusted to
+ * 127 / 12.8 = 9.921875
+ * <p>
+ * So we will quantize input values in the range (-10, 9.921875) to (-128, 127).
+ * <p>
+ * The input tensor can now be quantized by clipping values to the range
+ * `min_range` to `max_range`, then multiplying by scale_factor as follows:
+ * <pre>{@code
+ * result = round(min(max_range, max(min_range, input)) * scale_factor)
+ * }</pre>
+ * The adjusted `min_range` and `max_range` are returned as outputs 2 and 3 of
+ * this operation. These outputs should be used as the range for any further
+ * calculations.
+ * <p>
+ * <i>narrow_range (bool) attribute</i>
+ * <p>
+ * If true, we do not use the minimum quantized value.
+ * i.e. for int8 the quantized output, it would be restricted to the range
+ * -127..127 instead of the full -128..127 range.
+ * This is provided for compatibility with certain inference backends.
+ * (Only applies to SCALED mode)
+ * <p>
+ * <i>axis (int) attribute</i>
+ * <p>
+ * An optional `axis` attribute can specify a dimension index of the input tensor,
+ * such that quantization ranges will be calculated and applied separately for each
+ * slice of the tensor along that dimension. This is useful for per-channel
+ * quantization.
+ * <p>
+ * If axis is specified, min_range and max_range
+ * <p>
+ * if `axis`=None, per-tensor quantization is performed as normal.
+ * <p>
+ * <i>ensure_minimum_range (float) attribute</i>
+ * <p>
+ * Ensures the minimum quantization range is at least this value.
+ * The legacy default value for this is 0.01, but it is strongly suggested to
+ * set it to 0 for new uses.
+ * 
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator(group = "quantization")
+public final class Quantize<T extends TType> extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.quantization.Quantize}
+   */
+  public static class Options {
+    
+    /**
+     * @param mode 
+     */
+    public Options mode(String mode) {
+      this.mode = mode;
+      return this;
+    }
+    
+    /**
+     * @param roundMode 
+     */
+    public Options roundMode(String roundMode) {
+      this.roundMode = roundMode;
+      return this;
+    }
+    
+    /**
+     * @param narrowRange 
+     */
+    public Options narrowRange(Boolean narrowRange) {
+      this.narrowRange = narrowRange;
+      return this;
+    }
+    
+    /**
+     * @param axis 
+     */
+    public Options axis(Long axis) {
+      this.axis = axis;
+      return this;
+    }
+    
+    /**
+     * @param ensureMinimumRange 
+     */
+    public Options ensureMinimumRange(Float ensureMinimumRange) {
+      this.ensureMinimumRange = ensureMinimumRange;
+      return this;
+    }
+    
+    private String mode;
+    private String roundMode;
+    private Boolean narrowRange;
+    private Long axis;
+    private Float ensureMinimumRange;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new Quantize operation.
+   * 
+   * @param scope current scope
+   * @param input 
+   * @param minRange The minimum value of the quantization range. This value may be adjusted by the
+   * op depending on other parameters. The adjusted value is written to `output_min`.
+   * If the `axis` attribute is specified, this must be a 1-D tensor whose size
+   * matches the `axis` dimension of the input and output tensors.
+   * @param maxRange The maximum value of the quantization range. This value may be adjusted by the
+   * op depending on other parameters. The adjusted value is written to `output_max`.
+   * If the `axis` attribute is specified, this must be a 1-D tensor whose size
+   * matches the `axis` dimension of the input and output tensors.
+   * @param T 
+   * @param options carries optional attributes values
+   * @return a new instance of Quantize
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> Quantize<T> create(Scope scope, Operand<TFloat32> input, Operand<TFloat32> minRange, Operand<TFloat32> maxRange, Class<T> T, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("QuantizeV2", scope.makeOpName("Quantize"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(minRange.asOutput());
+    opBuilder.addInput(maxRange.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("T", T);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.mode != null) {
+          opBuilder.setAttr("mode", opts.mode);
+        }
+        if (opts.roundMode != null) {
+          opBuilder.setAttr("round_mode", opts.roundMode);
+        }
+        if (opts.narrowRange != null) {
+          opBuilder.setAttr("narrow_range", opts.narrowRange);
+        }
+        if (opts.axis != null) {
+          opBuilder.setAttr("axis", opts.axis);
+        }
+        if (opts.ensureMinimumRange != null) {
+          opBuilder.setAttr("ensure_minimum_range", opts.ensureMinimumRange);
+        }
+      }
+    }
+    return new Quantize<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param mode 
+   */
+  public static Options mode(String mode) {
+    return new Options().mode(mode);
+  }
+  
+  /**
+   * @param roundMode 
+   */
+  public static Options roundMode(String roundMode) {
+    return new Options().roundMode(roundMode);
+  }
+  
+  /**
+   * @param narrowRange 
+   */
+  public static Options narrowRange(Boolean narrowRange) {
+    return new Options().narrowRange(narrowRange);
+  }
+  
+  /**
+   * @param axis 
+   */
+  public static Options axis(Long axis) {
+    return new Options().axis(axis);
+  }
+  
+  /**
+   * @param ensureMinimumRange 
+   */
+  public static Options ensureMinimumRange(Float ensureMinimumRange) {
+    return new Options().ensureMinimumRange(ensureMinimumRange);
+  }
+  
+  /**
+   * The quantized data produced from the float input.
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  /**
+   * The final quantization range minimum, used to clip input values before scaling
+   * and rounding them to quantized values.
+   * If the `axis` attribute is specified, this will be a 1-D tensor whose size
+   * matches the `axis` dimension of the input and output tensors.
+   */
+  public Output<TFloat32> outputMin() {
+    return outputMin;
+  }
+  
+  /**
+   * The final quantization range maximum, used to clip input values before scaling
+   * and rounding them to quantized values.
+   * If the `axis` attribute is specified, this will be a 1-D tensor whose size
+   * matches the `axis` dimension of the input and output tensors.
+   */
+  public Output<TFloat32> outputMax() {
+    return outputMax;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "QuantizeV2";
+  
+  private Output<T> output;
+  private Output<TFloat32> outputMin;
+  private Output<TFloat32> outputMax;
+  
+  private Quantize(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+    outputMin = operation.output(outputIdx++);
+    outputMax = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/QuantizeAndDequantize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/QuantizeAndDequantize.java
new file mode 100644
index 00000000000..c14d9f451d7
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/QuantizeAndDequantize.java
@@ -0,0 +1,175 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.quantization;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Quantizes then dequantizes a tensor.
+ * <p>
+ * This is almost identical to QuantizeAndDequantizeV2, except that num_bits is a
+ * tensor, so its value can change during training.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator(group = "quantization")
+public final class QuantizeAndDequantize<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.quantization.QuantizeAndDequantize}
+   */
+  public static class Options {
+    
+    /**
+     * @param signedInput 
+     */
+    public Options signedInput(Boolean signedInput) {
+      this.signedInput = signedInput;
+      return this;
+    }
+    
+    /**
+     * @param rangeGiven 
+     */
+    public Options rangeGiven(Boolean rangeGiven) {
+      this.rangeGiven = rangeGiven;
+      return this;
+    }
+    
+    /**
+     * @param narrowRange 
+     */
+    public Options narrowRange(Boolean narrowRange) {
+      this.narrowRange = narrowRange;
+      return this;
+    }
+    
+    /**
+     * @param axis 
+     */
+    public Options axis(Long axis) {
+      this.axis = axis;
+      return this;
+    }
+    
+    private Boolean signedInput;
+    private Boolean rangeGiven;
+    private Boolean narrowRange;
+    private Long axis;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new QuantizeAndDequantize operation.
+   * 
+   * @param scope current scope
+   * @param input 
+   * @param inputMin 
+   * @param inputMax 
+   * @param numBits 
+   * @param options carries optional attributes values
+   * @return a new instance of QuantizeAndDequantize
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> QuantizeAndDequantize<T> create(Scope scope, Operand<T> input, Operand<T> inputMin, Operand<T> inputMax, Operand<TInt32> numBits, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("QuantizeAndDequantizeV3", scope.makeOpName("QuantizeAndDequantize"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(inputMin.asOutput());
+    opBuilder.addInput(inputMax.asOutput());
+    opBuilder.addInput(numBits.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.signedInput != null) {
+          opBuilder.setAttr("signed_input", opts.signedInput);
+        }
+        if (opts.rangeGiven != null) {
+          opBuilder.setAttr("range_given", opts.rangeGiven);
+        }
+        if (opts.narrowRange != null) {
+          opBuilder.setAttr("narrow_range", opts.narrowRange);
+        }
+        if (opts.axis != null) {
+          opBuilder.setAttr("axis", opts.axis);
+        }
+      }
+    }
+    return new QuantizeAndDequantize<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param signedInput 
+   */
+  public static Options signedInput(Boolean signedInput) {
+    return new Options().signedInput(signedInput);
+  }
+  
+  /**
+   * @param rangeGiven 
+   */
+  public static Options rangeGiven(Boolean rangeGiven) {
+    return new Options().rangeGiven(rangeGiven);
+  }
+  
+  /**
+   * @param narrowRange 
+   */
+  public static Options narrowRange(Boolean narrowRange) {
+    return new Options().narrowRange(narrowRange);
+  }
+  
+  /**
+   * @param axis 
+   */
+  public static Options axis(Long axis) {
+    return new Options().axis(axis);
+  }
+  
+  /**
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "QuantizeAndDequantizeV3";
+  
+  private Output<T> output;
+  
+  private QuantizeAndDequantize(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/QuantizeDownAndShrinkRange.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/QuantizeDownAndShrinkRange.java
new file mode 100644
index 00000000000..04f13bd833d
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/QuantizeDownAndShrinkRange.java
@@ -0,0 +1,117 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.quantization;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Convert the quantized 'input' tensor into a lower-precision 'output', using the
+ * <p>
+ * actual distribution of the values to maximize the usage of the lower bit depth
+ * and adjusting the output min and max ranges accordingly.
+ * <p>
+ * [input_min, input_max] are scalar floats that specify the range for the float
+ * interpretation of the 'input' data. For example, if input_min is -1.0f and
+ * input_max is 1.0f, and we are dealing with quint16 quantized data, then a 0
+ * value in the 16-bit data should be interpreted as -1.0f, and a 65535 means 1.0f.
+ * <p>
+ * This operator tries to squeeze as much precision as possible into an output with
+ * a lower bit depth by calculating the actual min and max values found in the
+ * data. For example, maybe that quint16 input has no values lower than 16,384 and
+ * none higher than 49,152. That means only half the range is actually needed, all
+ * the float interpretations are between -0.5f and 0.5f, so if we want to compress
+ * the data into a quint8 output, we can use that range rather than the theoretical
+ * -1.0f to 1.0f that is suggested by the input min and max.
+ * <p>
+ * In practice, this is most useful for taking output from operations like
+ * QuantizedMatMul that can produce higher bit-depth outputs than their inputs and
+ * may have large potential output ranges, but in practice have a distribution of
+ * input values that only uses a small fraction of the possible range. By feeding
+ * that output into this operator, we can reduce it from 32 bits down to 8 with
+ * minimal loss of accuracy.
+ * 
+ * @param <U> data type for {@code output()} output
+ */
+@Operator(group = "quantization")
+public final class QuantizeDownAndShrinkRange<U extends TType> extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new QuantizeDownAndShrinkRange operation.
+   * 
+   * @param scope current scope
+   * @param input 
+   * @param inputMin The float value that the minimum quantized input value represents.
+   * @param inputMax The float value that the maximum quantized input value represents.
+   * @param outType The type of the output. Should be a lower bit depth than Tinput.
+   * @return a new instance of QuantizeDownAndShrinkRange
+   */
+  @Endpoint(describeByClass = true)
+  public static <U extends TType, T extends TType> QuantizeDownAndShrinkRange<U> create(Scope scope, Operand<T> input, Operand<TFloat32> inputMin, Operand<TFloat32> inputMax, Class<U> outType) {
+    OperationBuilder opBuilder = scope.env().opBuilder("QuantizeDownAndShrinkRange", scope.makeOpName("QuantizeDownAndShrinkRange"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(inputMin.asOutput());
+    opBuilder.addInput(inputMax.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("out_type", outType);
+    return new QuantizeDownAndShrinkRange<U>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<U> output() {
+    return output;
+  }
+  
+  /**
+   * The float value that the minimum quantized output value represents.
+   */
+  public Output<TFloat32> outputMin() {
+    return outputMin;
+  }
+  
+  /**
+   * The float value that the maximum quantized output value represents.
+   */
+  public Output<TFloat32> outputMax() {
+    return outputMax;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "QuantizeDownAndShrinkRange";
+  
+  private Output<U> output;
+  private Output<TFloat32> outputMin;
+  private Output<TFloat32> outputMax;
+  
+  private QuantizeDownAndShrinkRange(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+    outputMin = operation.output(outputIdx++);
+    outputMax = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/QuantizedConcat.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/QuantizedConcat.java
new file mode 100644
index 00000000000..43839d072de
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/QuantizedConcat.java
@@ -0,0 +1,101 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.quantization;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.Operands;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Concatenates quantized tensors along one dimension.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator(group = "quantization")
+public final class QuantizedConcat<T extends TType> extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new QuantizedConcat operation.
+   * 
+   * @param scope current scope
+   * @param concatDim 0-D.  The dimension along which to concatenate.  Must be in the
+   * range [0, rank(values)).
+   * @param values The `N` Tensors to concatenate. Their ranks and types must match,
+   * and their sizes must match in all dimensions except `concat_dim`.
+   * @param inputMins The minimum scalar values for each of the input tensors.
+   * @param inputMaxes The maximum scalar values for each of the input tensors.
+   * @return a new instance of QuantizedConcat
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> QuantizedConcat<T> create(Scope scope, Operand<TInt32> concatDim, Iterable<Operand<T>> values, Iterable<Operand<TFloat32>> inputMins, Iterable<Operand<TFloat32>> inputMaxes) {
+    OperationBuilder opBuilder = scope.env().opBuilder("QuantizedConcat", scope.makeOpName("QuantizedConcat"));
+    opBuilder.addInput(concatDim.asOutput());
+    opBuilder.addInputList(Operands.asOutputs(values));
+    opBuilder.addInputList(Operands.asOutputs(inputMins));
+    opBuilder.addInputList(Operands.asOutputs(inputMaxes));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new QuantizedConcat<T>(opBuilder.build());
+  }
+  
+  /**
+   * A `Tensor` with the concatenation of values stacked along the
+   * `concat_dim` dimension.  This tensor's shape matches that of `values` except
+   * in `concat_dim` where it has the sum of the sizes.
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  /**
+   * The float value that the minimum quantized output value represents.
+   */
+  public Output<TFloat32> outputMin() {
+    return outputMin;
+  }
+  
+  /**
+   * The float value that the maximum quantized output value represents.
+   */
+  public Output<TFloat32> outputMax() {
+    return outputMax;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "QuantizedConcat";
+  
+  private Output<T> output;
+  private Output<TFloat32> outputMin;
+  private Output<TFloat32> outputMax;
+  
+  private QuantizedConcat(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+    outputMin = operation.output(outputIdx++);
+    outputMax = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/QuantizedMatMulWithBiasAndDequantize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/QuantizedMatMulWithBiasAndDequantize.java
new file mode 100644
index 00000000000..114d40bd058
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/QuantizedMatMulWithBiasAndDequantize.java
@@ -0,0 +1,163 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.quantization;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * @param <W> data type for {@code out()} output
+ */
+public final class QuantizedMatMulWithBiasAndDequantize<W extends TNumber> extends RawOp implements Operand<W> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.quantization.QuantizedMatMulWithBiasAndDequantize}
+   */
+  public static class Options {
+    
+    /**
+     * @param transposeA 
+     */
+    public Options transposeA(Boolean transposeA) {
+      this.transposeA = transposeA;
+      return this;
+    }
+    
+    /**
+     * @param transposeB 
+     */
+    public Options transposeB(Boolean transposeB) {
+      this.transposeB = transposeB;
+      return this;
+    }
+    
+    /**
+     * @param inputQuantMode 
+     */
+    public Options inputQuantMode(String inputQuantMode) {
+      this.inputQuantMode = inputQuantMode;
+      return this;
+    }
+    
+    private Boolean transposeA;
+    private Boolean transposeB;
+    private String inputQuantMode;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new QuantizedMatMulWithBiasAndDequantize operation.
+   * 
+   * @param scope current scope
+   * @param a 
+   * @param b 
+   * @param bias 
+   * @param minA 
+   * @param maxA 
+   * @param minB 
+   * @param maxB 
+   * @param minFreezedOutput 
+   * @param maxFreezedOutput 
+   * @param Toutput 
+   * @param options carries optional attributes values
+   * @return a new instance of QuantizedMatMulWithBiasAndDequantize
+   */
+  @Endpoint(describeByClass = true)
+  public static <W extends TNumber, T extends TType, U extends TType, V extends TType> QuantizedMatMulWithBiasAndDequantize<W> create(Scope scope, Operand<T> a, Operand<U> b, Operand<V> bias, Operand<TFloat32> minA, Operand<TFloat32> maxA, Operand<TFloat32> minB, Operand<TFloat32> maxB, Operand<TFloat32> minFreezedOutput, Operand<TFloat32> maxFreezedOutput, Class<W> Toutput, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("QuantizedMatMulWithBiasAndDequantize", scope.makeOpName("QuantizedMatMulWithBiasAndDequantize"));
+    opBuilder.addInput(a.asOutput());
+    opBuilder.addInput(b.asOutput());
+    opBuilder.addInput(bias.asOutput());
+    opBuilder.addInput(minA.asOutput());
+    opBuilder.addInput(maxA.asOutput());
+    opBuilder.addInput(minB.asOutput());
+    opBuilder.addInput(maxB.asOutput());
+    opBuilder.addInput(minFreezedOutput.asOutput());
+    opBuilder.addInput(maxFreezedOutput.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("Toutput", Toutput);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.transposeA != null) {
+          opBuilder.setAttr("transpose_a", opts.transposeA);
+        }
+        if (opts.transposeB != null) {
+          opBuilder.setAttr("transpose_b", opts.transposeB);
+        }
+        if (opts.inputQuantMode != null) {
+          opBuilder.setAttr("input_quant_mode", opts.inputQuantMode);
+        }
+      }
+    }
+    return new QuantizedMatMulWithBiasAndDequantize<W>(opBuilder.build());
+  }
+  
+  /**
+   * @param transposeA 
+   */
+  public static Options transposeA(Boolean transposeA) {
+    return new Options().transposeA(transposeA);
+  }
+  
+  /**
+   * @param transposeB 
+   */
+  public static Options transposeB(Boolean transposeB) {
+    return new Options().transposeB(transposeB);
+  }
+  
+  /**
+   * @param inputQuantMode 
+   */
+  public static Options inputQuantMode(String inputQuantMode) {
+    return new Options().inputQuantMode(inputQuantMode);
+  }
+  
+  /**
+   */
+  public Output<W> out() {
+    return out;
+  }
+  
+  @Override
+  public Output<W> asOutput() {
+    return out;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "QuantizedMatMulWithBiasAndDequantize";
+  
+  private Output<W> out;
+  
+  private QuantizedMatMulWithBiasAndDequantize(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    out = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/QuantizedMatMulWithBiasAndRequantize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/QuantizedMatMulWithBiasAndRequantize.java
new file mode 100644
index 00000000000..d553cd17e60
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/QuantizedMatMulWithBiasAndRequantize.java
@@ -0,0 +1,173 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.quantization;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.family.TType;
+
+/**
+ * @param <W> data type for {@code out()} output
+ */
+public final class QuantizedMatMulWithBiasAndRequantize<W extends TType> extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.quantization.QuantizedMatMulWithBiasAndRequantize}
+   */
+  public static class Options {
+    
+    /**
+     * @param transposeA 
+     */
+    public Options transposeA(Boolean transposeA) {
+      this.transposeA = transposeA;
+      return this;
+    }
+    
+    /**
+     * @param transposeB 
+     */
+    public Options transposeB(Boolean transposeB) {
+      this.transposeB = transposeB;
+      return this;
+    }
+    
+    /**
+     * @param inputQuantMode 
+     */
+    public Options inputQuantMode(String inputQuantMode) {
+      this.inputQuantMode = inputQuantMode;
+      return this;
+    }
+    
+    private Boolean transposeA;
+    private Boolean transposeB;
+    private String inputQuantMode;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new QuantizedMatMulWithBiasAndRequantize operation.
+   * 
+   * @param scope current scope
+   * @param a 
+   * @param b 
+   * @param bias 
+   * @param minA 
+   * @param maxA 
+   * @param minB 
+   * @param maxB 
+   * @param minFreezedOutput 
+   * @param maxFreezedOutput 
+   * @param Toutput 
+   * @param options carries optional attributes values
+   * @return a new instance of QuantizedMatMulWithBiasAndRequantize
+   */
+  @Endpoint(describeByClass = true)
+  public static <W extends TType, T extends TType, U extends TType, V extends TType> QuantizedMatMulWithBiasAndRequantize<W> create(Scope scope, Operand<T> a, Operand<U> b, Operand<V> bias, Operand<TFloat32> minA, Operand<TFloat32> maxA, Operand<TFloat32> minB, Operand<TFloat32> maxB, Operand<TFloat32> minFreezedOutput, Operand<TFloat32> maxFreezedOutput, Class<W> Toutput, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("QuantizedMatMulWithBiasAndRequantize", scope.makeOpName("QuantizedMatMulWithBiasAndRequantize"));
+    opBuilder.addInput(a.asOutput());
+    opBuilder.addInput(b.asOutput());
+    opBuilder.addInput(bias.asOutput());
+    opBuilder.addInput(minA.asOutput());
+    opBuilder.addInput(maxA.asOutput());
+    opBuilder.addInput(minB.asOutput());
+    opBuilder.addInput(maxB.asOutput());
+    opBuilder.addInput(minFreezedOutput.asOutput());
+    opBuilder.addInput(maxFreezedOutput.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("Toutput", Toutput);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.transposeA != null) {
+          opBuilder.setAttr("transpose_a", opts.transposeA);
+        }
+        if (opts.transposeB != null) {
+          opBuilder.setAttr("transpose_b", opts.transposeB);
+        }
+        if (opts.inputQuantMode != null) {
+          opBuilder.setAttr("input_quant_mode", opts.inputQuantMode);
+        }
+      }
+    }
+    return new QuantizedMatMulWithBiasAndRequantize<W>(opBuilder.build());
+  }
+  
+  /**
+   * @param transposeA 
+   */
+  public static Options transposeA(Boolean transposeA) {
+    return new Options().transposeA(transposeA);
+  }
+  
+  /**
+   * @param transposeB 
+   */
+  public static Options transposeB(Boolean transposeB) {
+    return new Options().transposeB(transposeB);
+  }
+  
+  /**
+   * @param inputQuantMode 
+   */
+  public static Options inputQuantMode(String inputQuantMode) {
+    return new Options().inputQuantMode(inputQuantMode);
+  }
+  
+  /**
+   */
+  public Output<W> out() {
+    return out;
+  }
+  
+  /**
+   */
+  public Output<TFloat32> minOut() {
+    return minOut;
+  }
+  
+  /**
+   */
+  public Output<TFloat32> maxOut() {
+    return maxOut;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "QuantizedMatMulWithBiasAndRequantize";
+  
+  private Output<W> out;
+  private Output<TFloat32> minOut;
+  private Output<TFloat32> maxOut;
+  
+  private QuantizedMatMulWithBiasAndRequantize(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    out = operation.output(outputIdx++);
+    minOut = operation.output(outputIdx++);
+    maxOut = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/RequantizationRange.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/RequantizationRange.java
new file mode 100644
index 00000000000..86ab731bb50
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/RequantizationRange.java
@@ -0,0 +1,87 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.quantization;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Computes a range that covers the actual values present in a quantized tensor.
+ * <p>
+ * Given a quantized tensor described by `(input, input_min, input_max)`, outputs a
+ * range that covers the actual values present in that tensor. This op is typically
+ * used to produce the `requested_output_min` and `requested_output_max` for
+ * `Requantize`.
+ */
+@Operator(group = "quantization")
+public final class RequantizationRange extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new RequantizationRange operation.
+   * 
+   * @param scope current scope
+   * @param input 
+   * @param inputMin The float value that the minimum quantized input value represents.
+   * @param inputMax The float value that the maximum quantized input value represents.
+   * @return a new instance of RequantizationRange
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> RequantizationRange create(Scope scope, Operand<T> input, Operand<TFloat32> inputMin, Operand<TFloat32> inputMax) {
+    OperationBuilder opBuilder = scope.env().opBuilder("RequantizationRange", scope.makeOpName("RequantizationRange"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(inputMin.asOutput());
+    opBuilder.addInput(inputMax.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new RequantizationRange(opBuilder.build());
+  }
+  
+  /**
+   * The computed min output.
+   */
+  public Output<TFloat32> outputMin() {
+    return outputMin;
+  }
+  
+  /**
+   * the computed max output.
+   */
+  public Output<TFloat32> outputMax() {
+    return outputMax;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "RequantizationRange";
+  
+  private Output<TFloat32> outputMin;
+  private Output<TFloat32> outputMax;
+  
+  private RequantizationRange(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    outputMin = operation.output(outputIdx++);
+    outputMax = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/Requantize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/Requantize.java
new file mode 100644
index 00000000000..1ee6077abe1
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/quantization/Requantize.java
@@ -0,0 +1,106 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.quantization;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Converts the quantized `input` tensor into a lower-precision `output`.
+ * <p>
+ * Converts the quantized `input` tensor into a lower-precision `output`, using the
+ * output range specified with `requested_output_min` and `requested_output_max`.
+ * <p>
+ * `[input_min, input_max]` are scalar floats that specify the range for the float
+ * interpretation of the `input` data. For example, if `input_min` is -1.0f and
+ * `input_max` is 1.0f, and we are dealing with `quint16` quantized data, then a 0
+ * value in the 16-bit data should be interpreted as -1.0f, and a 65535 means 1.0f.
+ * 
+ * @param <U> data type for {@code output()} output
+ */
+@Operator(group = "quantization")
+public final class Requantize<U extends TType> extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new Requantize operation.
+   * 
+   * @param scope current scope
+   * @param input 
+   * @param inputMin The float value that the minimum quantized input value represents.
+   * @param inputMax The float value that the maximum quantized input value represents.
+   * @param requestedOutputMin The float value that the minimum quantized output value represents.
+   * @param requestedOutputMax The float value that the maximum quantized output value represents.
+   * @param outType The type of the output. Should be a lower bit depth than Tinput.
+   * @return a new instance of Requantize
+   */
+  @Endpoint(describeByClass = true)
+  public static <U extends TType, T extends TType> Requantize<U> create(Scope scope, Operand<T> input, Operand<TFloat32> inputMin, Operand<TFloat32> inputMax, Operand<TFloat32> requestedOutputMin, Operand<TFloat32> requestedOutputMax, Class<U> outType) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Requantize", scope.makeOpName("Requantize"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(inputMin.asOutput());
+    opBuilder.addInput(inputMax.asOutput());
+    opBuilder.addInput(requestedOutputMin.asOutput());
+    opBuilder.addInput(requestedOutputMax.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("out_type", outType);
+    return new Requantize<U>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<U> output() {
+    return output;
+  }
+  
+  /**
+   * The requested_output_min value is copied into this output.
+   */
+  public Output<TFloat32> outputMin() {
+    return outputMin;
+  }
+  
+  /**
+   * The requested_output_max value is copied into this output.
+   */
+  public Output<TFloat32> outputMax() {
+    return outputMax;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Requantize";
+  
+  private Output<U> output;
+  private Output<TFloat32> outputMin;
+  private Output<TFloat32> outputMax;
+  
+  private Requantize(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+    outputMin = operation.output(outputIdx++);
+    outputMax = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/ragged/RaggedBincount.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/ragged/RaggedBincount.java
new file mode 100644
index 00000000000..b5e54a6a822
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/ragged/RaggedBincount.java
@@ -0,0 +1,127 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.ragged;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Counts the number of occurrences of each value in an integer array.
+ * <p>
+ * Outputs a vector with length `size` and the same dtype as `weights`. If
+ * `weights` are empty, then index `i` stores the number of times the value `i` is
+ * counted in `arr`. If `weights` are non-empty, then index `i` stores the sum of
+ * the value in `weights` at each index where the corresponding value in `arr` is
+ * `i`.
+ * <p>
+ * Values in `arr` outside of the range [0, size) are ignored.
+ * 
+ * @param <U> data type for {@code output()} output
+ */
+@Operator(group = "ragged")
+public final class RaggedBincount<U extends TNumber> extends RawOp implements Operand<U> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.ragged.RaggedBincount}
+   */
+  public static class Options {
+    
+    /**
+     * @param binaryOutput bool; Whether the kernel should count the appearance or number of occurrences.
+     */
+    public Options binaryOutput(Boolean binaryOutput) {
+      this.binaryOutput = binaryOutput;
+      return this;
+    }
+    
+    private Boolean binaryOutput;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new RaggedBincount operation.
+   * 
+   * @param scope current scope
+   * @param splits 1D int64 `Tensor`.
+   * @param values 2D int `Tensor`.
+   * @param size non-negative int scalar `Tensor`.
+   * @param weights is an int32, int64, float32, or float64 `Tensor` with the same
+   * shape as `input`, or a length-0 `Tensor`, in which case it acts as all weights
+   * equal to 1.
+   * @param options carries optional attributes values
+   * @return a new instance of RaggedBincount
+   */
+  @Endpoint(describeByClass = true)
+  public static <U extends TNumber, T extends TNumber> RaggedBincount<U> create(Scope scope, Operand<TInt64> splits, Operand<T> values, Operand<T> size, Operand<U> weights, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("RaggedBincount", scope.makeOpName("RaggedBincount"));
+    opBuilder.addInput(splits.asOutput());
+    opBuilder.addInput(values.asOutput());
+    opBuilder.addInput(size.asOutput());
+    opBuilder.addInput(weights.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.binaryOutput != null) {
+          opBuilder.setAttr("binary_output", opts.binaryOutput);
+        }
+      }
+    }
+    return new RaggedBincount<U>(opBuilder.build());
+  }
+  
+  /**
+   * @param binaryOutput bool; Whether the kernel should count the appearance or number of occurrences.
+   */
+  public static Options binaryOutput(Boolean binaryOutput) {
+    return new Options().binaryOutput(binaryOutput);
+  }
+  
+  /**
+   * 1D `Tensor` with length equal to `size` or 2D `Tensor` with [batch_size, `size`].
+   * The counts or summed weights for each value in the range [0, size).
+   */
+  public Output<U> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<U> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "RaggedBincount";
+  
+  private Output<U> output;
+  
+  private RaggedBincount(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/ragged/RaggedCountSparseOutput.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/ragged/RaggedCountSparseOutput.java
new file mode 100644
index 00000000000..5148b8f1f00
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/ragged/RaggedCountSparseOutput.java
@@ -0,0 +1,156 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.ragged;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Performs sparse-output bin counting for a ragged tensor input.
+ * <p>
+ *   Counts the number of times each value occurs in the input.
+ * 
+ * @param <U> data type for {@code outputValues()} output
+ */
+public final class RaggedCountSparseOutput<U extends TNumber> extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.ragged.RaggedCountSparseOutput}
+   */
+  public static class Options {
+    
+    /**
+     * @param minlength Minimum value to count. Can be set to -1 for no minimum.
+     */
+    public Options minlength(Long minlength) {
+      this.minlength = minlength;
+      return this;
+    }
+    
+    /**
+     * @param maxlength Maximum value to count. Can be set to -1 for no maximum.
+     */
+    public Options maxlength(Long maxlength) {
+      this.maxlength = maxlength;
+      return this;
+    }
+    
+    private Long minlength;
+    private Long maxlength;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new RaggedCountSparseOutput operation.
+   * 
+   * @param scope current scope
+   * @param splits Tensor containing the row splits of the ragged tensor to count.
+   * @param values Tensor containing values of the sparse tensor to count.
+   * @param weights A Tensor of the same shape as indices containing per-index weight values.
+   * May also be the empty tensor if no weights are used.
+   * @param binaryOutput Whether to output the number of occurrences of each value or 1.
+   * @param options carries optional attributes values
+   * @return a new instance of RaggedCountSparseOutput
+   */
+  @Endpoint(describeByClass = true)
+  public static <U extends TNumber, T extends TNumber> RaggedCountSparseOutput<U> create(Scope scope, Operand<TInt64> splits, Operand<T> values, Operand<U> weights, Boolean binaryOutput, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("RaggedCountSparseOutput", scope.makeOpName("RaggedCountSparseOutput"));
+    opBuilder.addInput(splits.asOutput());
+    opBuilder.addInput(values.asOutput());
+    opBuilder.addInput(weights.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("binary_output", binaryOutput);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.minlength != null) {
+          opBuilder.setAttr("minlength", opts.minlength);
+        }
+        if (opts.maxlength != null) {
+          opBuilder.setAttr("maxlength", opts.maxlength);
+        }
+      }
+    }
+    return new RaggedCountSparseOutput<U>(opBuilder.build());
+  }
+  
+  /**
+   * @param minlength Minimum value to count. Can be set to -1 for no minimum.
+   */
+  public static Options minlength(Long minlength) {
+    return new Options().minlength(minlength);
+  }
+  
+  /**
+   * @param maxlength Maximum value to count. Can be set to -1 for no maximum.
+   */
+  public static Options maxlength(Long maxlength) {
+    return new Options().maxlength(maxlength);
+  }
+  
+  /**
+   * Indices tensor for the resulting sparse tensor object.
+   */
+  public Output<TInt64> outputIndices() {
+    return outputIndices;
+  }
+  
+  /**
+   * Values tensor for the resulting sparse tensor object.
+   */
+  public Output<U> outputValues() {
+    return outputValues;
+  }
+  
+  /**
+   * Shape tensor for the resulting sparse tensor object.
+   *   END
+   *   }
+   *   attr {
+   *     name: "T"
+   *     description: <<END
+   * Dtype of the input values tensor.
+   */
+  public Output<TInt64> outputDenseShape() {
+    return outputDenseShape;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "RaggedCountSparseOutput";
+  
+  private Output<TInt64> outputIndices;
+  private Output<U> outputValues;
+  private Output<TInt64> outputDenseShape;
+  
+  private RaggedCountSparseOutput(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    outputIndices = operation.output(outputIdx++);
+    outputValues = operation.output(outputIdx++);
+    outputDenseShape = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/ragged/RaggedCross.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/ragged/RaggedCross.java
new file mode 100644
index 00000000000..4a21e8cdfe4
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/ragged/RaggedCross.java
@@ -0,0 +1,108 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.ragged;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.Operands;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Generates a feature cross from a list of tensors, and returns it as a
+ * RaggedTensor.  See `tf.ragged.cross` for more details.
+ * 
+ * @param <T> data type for {@code outputValues()} output
+ * @param <U> data type for {@code outputRowSplits()} output
+ */
+public final class RaggedCross<T extends TType, U extends TNumber> extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new RaggedCross operation.
+   * 
+   * @param scope current scope
+   * @param raggedValues The values tensor for each RaggedTensor input.
+   * @param raggedRowSplits The row_splits tensor for each RaggedTensor input.
+   * @param sparseIndices The indices tensor for each SparseTensor input.
+   * @param sparseValues The values tensor for each SparseTensor input.
+   * @param sparseShape The dense_shape tensor for each SparseTensor input.
+   * @param denseInputs The tf.Tensor inputs.
+   * @param inputOrder String specifying the tensor type for each input.  The `i`th character in
+   * this string specifies the type of the `i`th input, and is one of: 'R' (ragged),
+   * 'D' (dense), or 'S' (sparse).  This attr is used to ensure that the crossed
+   * values are combined in the order of the inputs from the call to tf.ragged.cross.
+   * @param hashedOutput 
+   * @param numBuckets 
+   * @param hashKey 
+   * @param outValuesType 
+   * @param outRowSplitsType 
+   * @return a new instance of RaggedCross
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType, U extends TNumber> RaggedCross<T, U> create(Scope scope, Iterable<Operand<?>> raggedValues, Iterable<Operand<?>> raggedRowSplits, Iterable<Operand<TInt64>> sparseIndices, Iterable<Operand<?>> sparseValues, Iterable<Operand<TInt64>> sparseShape, Iterable<Operand<?>> denseInputs, String inputOrder, Boolean hashedOutput, Long numBuckets, Long hashKey, Class<T> outValuesType, Class<U> outRowSplitsType) {
+    OperationBuilder opBuilder = scope.env().opBuilder("RaggedCross", scope.makeOpName("RaggedCross"));
+    opBuilder.addInputList(Operands.asOutputs(raggedValues));
+    opBuilder.addInputList(Operands.asOutputs(raggedRowSplits));
+    opBuilder.addInputList(Operands.asOutputs(sparseIndices));
+    opBuilder.addInputList(Operands.asOutputs(sparseValues));
+    opBuilder.addInputList(Operands.asOutputs(sparseShape));
+    opBuilder.addInputList(Operands.asOutputs(denseInputs));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("input_order", inputOrder);
+    opBuilder.setAttr("hashed_output", hashedOutput);
+    opBuilder.setAttr("num_buckets", numBuckets);
+    opBuilder.setAttr("hash_key", hashKey);
+    opBuilder.setAttr("out_values_type", outValuesType);
+    opBuilder.setAttr("out_row_splits_type", outRowSplitsType);
+    return new RaggedCross<T, U>(opBuilder.build());
+  }
+  
+  /**
+   * The `values` for the returned `RaggedTensor`.
+   */
+  public Output<T> outputValues() {
+    return outputValues;
+  }
+  
+  /**
+   * The `row_splits` for the returned `RaggedTensor`.
+   */
+  public Output<U> outputRowSplits() {
+    return outputRowSplits;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "RaggedCross";
+  
+  private Output<T> outputValues;
+  private Output<U> outputRowSplits;
+  
+  private RaggedCross(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    outputValues = operation.output(outputIdx++);
+    outputRowSplits = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/ragged/RaggedGather.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/ragged/RaggedGather.java
new file mode 100644
index 00000000000..7cc7de6d3db
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/ragged/RaggedGather.java
@@ -0,0 +1,127 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.ragged;
+
+import java.util.Arrays;
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.Operands;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Gather ragged slices from `params` axis `0` according to `indices`.
+ * <p>
+ * Outputs a `RaggedTensor` output composed from `output_dense_values` and
+ * `output_nested_splits`, such that:
+ * <pre>{@code
+ * output.shape = indices.shape + params.shape[1:]
+ * output.ragged_rank = indices.shape.ndims + params.ragged_rank
+ * output[i...j, d0...dn] = params[indices[i...j], d0...dn]
+ * }</pre>
+ * where
+ * <ul>
+ * <li>
+ * `params =
+ *    ragged.from_nested_row_splits(params_dense_values, params_nested_splits)`
+ *    provides the values that should be gathered.
+ * </li>
+ * <li>
+ * `indices` ia a dense tensor with dtype `int32` or `int64`, indicating which
+ *    values should be gathered.
+ * </li>
+ * <li>
+ * `output =
+ *    ragged.from_nested_row_splits(output_dense_values, output_nested_splits)`
+ *    is the output tensor.
+ * </li>
+ * </ul>
+ * (Note: This c++ op is used to implement the higher-level python
+ * `tf.ragged.gather` op, which also supports ragged indices.)
+ * 
+ * 
+ * @param <T> data type for {@code outputNestedSplits()} output
+ * @param <U> data type for {@code outputDenseValues()} output
+ */
+public final class RaggedGather<T extends TNumber, U extends TType> extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new RaggedGather operation.
+   * 
+   * @param scope current scope
+   * @param paramsNestedSplits The `nested_row_splits` tensors that define the row-partitioning for the
+   * `params` RaggedTensor input.
+   * @param paramsDenseValues The `flat_values` for the `params` RaggedTensor. There was a terminology change
+   * at the python level from dense_values to flat_values, so dense_values is the
+   * deprecated name.
+   * @param indices Indices in the outermost dimension of `params` of the values that should be
+   * gathered.
+   * @param OUTPUTRAGGEDRANK The ragged rank of the output RaggedTensor. `output_nested_splits` will contain
+   * this number of `row_splits` tensors. This value should equal
+   * `indices.shape.ndims + params.ragged_rank - 1`.
+   * @return a new instance of RaggedGather
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber, U extends TType, V extends TNumber> RaggedGather<T, U> create(Scope scope, Iterable<Operand<T>> paramsNestedSplits, Operand<U> paramsDenseValues, Operand<V> indices, Long OUTPUTRAGGEDRANK) {
+    OperationBuilder opBuilder = scope.env().opBuilder("RaggedGather", scope.makeOpName("RaggedGather"));
+    opBuilder.addInputList(Operands.asOutputs(paramsNestedSplits));
+    opBuilder.addInput(paramsDenseValues.asOutput());
+    opBuilder.addInput(indices.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("OUTPUT_RAGGED_RANK", OUTPUTRAGGEDRANK);
+    return new RaggedGather<T, U>(opBuilder.build());
+  }
+  
+  /**
+   * The `nested_row_splits` tensors that define the row-partitioning for the
+   * returned RaggedTensor.
+   */
+  public List<Output<T>> outputNestedSplits() {
+    return outputNestedSplits;
+  }
+  
+  /**
+   * The `flat_values` for the returned RaggedTensor.
+   */
+  public Output<U> outputDenseValues() {
+    return outputDenseValues;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "RaggedGather";
+  
+  private List<Output<T>> outputNestedSplits;
+  private Output<U> outputDenseValues;
+  
+  @SuppressWarnings("unchecked")
+  private RaggedGather(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    int outputNestedSplitsLength = operation.outputListLength("output_nested_splits");
+    outputNestedSplits = Arrays.asList((Output<T>[])operation.outputList(outputIdx, outputNestedSplitsLength));
+    outputIdx += outputNestedSplitsLength;
+    outputDenseValues = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/ragged/RaggedRange.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/ragged/RaggedRange.java
new file mode 100644
index 00000000000..ea23e612ab2
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/ragged/RaggedRange.java
@@ -0,0 +1,115 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.ragged;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Returns a `RaggedTensor` containing the specified sequences of numbers.
+ * <p>
+ * 
+ * Returns a `RaggedTensor` `result` composed from `rt_dense_values` and
+ * `rt_nested_splits`, such that
+ * `result[i] = range(starts[i], limits[i], deltas[i])`.
+ * <pre>{@code
+ * (rt_nested_splits, rt_dense_values) = ragged_range(
+ *       starts=[2, 5, 8], limits=[3, 5, 12], deltas=1)
+ * result = tf.ragged.from_row_splits(rt_dense_values, rt_nested_splits)
+ * print(result)
+ * <tf.RaggedTensor [[2], [], [8, 9, 10, 11]] >
+ * }</pre>
+ * The input tensors `starts`, `limits`, and `deltas` may be scalars or vectors.
+ * The vector inputs must all have the same size.  Scalar inputs are broadcast
+ * to match the size of the vector inputs.
+ * 
+ * @param <U> data type for {@code rtNestedSplits()} output
+ * @param <T> data type for {@code rtDenseValues()} output
+ */
+public final class RaggedRange<U extends TNumber, T extends TNumber> extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new RaggedRange operation.
+   * 
+   * @param scope current scope
+   * @param starts The starts of each range.
+   * @param limits The limits of each range.
+   * @param deltas The deltas of each range.
+   * @param Tsplits 
+   * @return a new instance of RaggedRange
+   */
+  @Endpoint(describeByClass = true)
+  public static <U extends TNumber, T extends TNumber> RaggedRange<U, T> create(Scope scope, Operand<T> starts, Operand<T> limits, Operand<T> deltas, Class<U> Tsplits) {
+    OperationBuilder opBuilder = scope.env().opBuilder("RaggedRange", scope.makeOpName("RaggedRange"));
+    opBuilder.addInput(starts.asOutput());
+    opBuilder.addInput(limits.asOutput());
+    opBuilder.addInput(deltas.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("Tsplits", Tsplits);
+    return new RaggedRange<U, T>(opBuilder.build());
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new RaggedRange operation using default output types.
+   * 
+   * @param scope current scope
+   * @param starts The starts of each range.
+   * @param limits The limits of each range.
+   * @param deltas The deltas of each range.
+   * @return a new instance of RaggedRange
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> RaggedRange<TInt64, T> create(Scope scope, Operand<T> starts, Operand<T> limits, Operand<T> deltas) {
+    return create(scope, starts, limits, deltas, TInt64.class);
+  }
+  
+  /**
+   * The `row_splits` for the returned `RaggedTensor`.
+   */
+  public Output<U> rtNestedSplits() {
+    return rtNestedSplits;
+  }
+  
+  /**
+   * The `flat_values` for the returned `RaggedTensor`.
+   */
+  public Output<T> rtDenseValues() {
+    return rtDenseValues;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "RaggedRange";
+  
+  private Output<U> rtNestedSplits;
+  private Output<T> rtDenseValues;
+  
+  private RaggedRange(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    rtNestedSplits = operation.output(outputIdx++);
+    rtDenseValues = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/ragged/RaggedTensorFromVariant.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/ragged/RaggedTensorFromVariant.java
new file mode 100644
index 00000000000..cd447134359
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/ragged/RaggedTensorFromVariant.java
@@ -0,0 +1,127 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.ragged;
+
+import java.util.Arrays;
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Decodes a `variant` Tensor into a `RaggedTensor`.
+ * <p>
+ * Decodes the given `variant` Tensor and returns a `RaggedTensor`. The input
+ * could be a scalar, meaning it encodes a single `RaggedTensor` with ragged_rank
+ * `output_ragged_rank`. It could also have an arbitrary rank, in which case each
+ * element is decoded into a `RaggedTensor` with ragged_rank `input_ragged_rank`
+ * and these are then stacked according to the input shape to output a single
+ * `RaggedTensor` with ragged_rank `output_ragged_rank`. Each `variant` element in
+ * the input Tensor is decoded by retrieving from the element a 1-D `variant`
+ * Tensor with `input_ragged_rank + 1` Tensors, corresponding to the splits and
+ * values of the decoded `RaggedTensor`. If `input_ragged_rank` is -1, then it is
+ * inferred as `output_ragged_rank` - `rank(encoded_ragged)`. See
+ * `RaggedTensorToVariant` for the corresponding encoding logic.
+ * 
+ * 
+ * @param <U> data type for {@code outputNestedSplits()} output
+ * @param <T> data type for {@code outputDenseValues()} output
+ */
+public final class RaggedTensorFromVariant<U extends TNumber, T extends TType> extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new RaggedTensorFromVariant operation.
+   * 
+   * @param scope current scope
+   * @param encodedRagged A `variant` Tensor containing encoded `RaggedTensor`s.
+   * @param inputRaggedRank The ragged rank of each encoded `RaggedTensor` component in the input. If set to
+   * -1, this is inferred as `output_ragged_rank` - `rank(encoded_ragged)`
+   * @param outputRaggedRank The expected ragged rank of the output `RaggedTensor`. The following must hold:
+   * `output_ragged_rank = rank(encoded_ragged) + input_ragged_rank`.
+   * @param Tvalues 
+   * @param Tsplits 
+   * @return a new instance of RaggedTensorFromVariant
+   */
+  @Endpoint(describeByClass = true)
+  public static <U extends TNumber, T extends TType> RaggedTensorFromVariant<U, T> create(Scope scope, Operand<?> encodedRagged, Long inputRaggedRank, Long outputRaggedRank, Class<T> Tvalues, Class<U> Tsplits) {
+    OperationBuilder opBuilder = scope.env().opBuilder("RaggedTensorFromVariant", scope.makeOpName("RaggedTensorFromVariant"));
+    opBuilder.addInput(encodedRagged.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("input_ragged_rank", inputRaggedRank);
+    opBuilder.setAttr("output_ragged_rank", outputRaggedRank);
+    opBuilder.setAttr("Tvalues", Tvalues);
+    opBuilder.setAttr("Tsplits", Tsplits);
+    return new RaggedTensorFromVariant<U, T>(opBuilder.build());
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new RaggedTensorFromVariant operation using default output types.
+   * 
+   * @param scope current scope
+   * @param encodedRagged A `variant` Tensor containing encoded `RaggedTensor`s.
+   * @param inputRaggedRank The ragged rank of each encoded `RaggedTensor` component in the input. If set to
+   * -1, this is inferred as `output_ragged_rank` - `rank(encoded_ragged)`
+   * @param outputRaggedRank The expected ragged rank of the output `RaggedTensor`. The following must hold:
+   * `output_ragged_rank = rank(encoded_ragged) + input_ragged_rank`.
+   * @param Tvalues 
+   * @return a new instance of RaggedTensorFromVariant
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> RaggedTensorFromVariant<TInt64, T> create(Scope scope, Operand<?> encodedRagged, Long inputRaggedRank, Long outputRaggedRank, Class<T> Tvalues) {
+    return create(scope, encodedRagged, inputRaggedRank, outputRaggedRank, Tvalues, TInt64.class);
+  }
+  
+  /**
+   * A list of one or more Tensors representing the splits of the output
+   * `RaggedTensor`.
+   */
+  public List<Output<U>> outputNestedSplits() {
+    return outputNestedSplits;
+  }
+  
+  /**
+   * A Tensor representing the values of the output `RaggedTensor`.
+   */
+  public Output<T> outputDenseValues() {
+    return outputDenseValues;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "RaggedTensorFromVariant";
+  
+  private List<Output<U>> outputNestedSplits;
+  private Output<T> outputDenseValues;
+  
+  @SuppressWarnings("unchecked")
+  private RaggedTensorFromVariant(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    int outputNestedSplitsLength = operation.outputListLength("output_nested_splits");
+    outputNestedSplits = Arrays.asList((Output<U>[])operation.outputList(outputIdx, outputNestedSplitsLength));
+    outputIdx += outputNestedSplitsLength;
+    outputDenseValues = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/ragged/RaggedTensorToSparse.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/ragged/RaggedTensorToSparse.java
new file mode 100644
index 00000000000..7906048788d
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/ragged/RaggedTensorToSparse.java
@@ -0,0 +1,96 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.ragged;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.Operands;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Converts a `RaggedTensor` into a `SparseTensor` with the same values.
+ * <p>
+ * input=ragged.from_nested_row_splits(rt_dense_values, rt_nested_splits)
+ * output=SparseTensor(indices=sparse_indices, values=sparse_values,
+ *                     dense_shape=sparse_dense_shape)
+ * 
+ * @param <U> data type for {@code sparseValues()} output
+ */
+public final class RaggedTensorToSparse<U extends TType> extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new RaggedTensorToSparse operation.
+   * 
+   * @param scope current scope
+   * @param rtNestedSplits The `row_splits` for the `RaggedTensor`.
+   * @param rtDenseValues The `flat_values` for the `RaggedTensor`.
+   * @return a new instance of RaggedTensorToSparse
+   */
+  @Endpoint(describeByClass = true)
+  public static <U extends TType, T extends TNumber> RaggedTensorToSparse<U> create(Scope scope, Iterable<Operand<T>> rtNestedSplits, Operand<U> rtDenseValues) {
+    OperationBuilder opBuilder = scope.env().opBuilder("RaggedTensorToSparse", scope.makeOpName("RaggedTensorToSparse"));
+    opBuilder.addInputList(Operands.asOutputs(rtNestedSplits));
+    opBuilder.addInput(rtDenseValues.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new RaggedTensorToSparse<U>(opBuilder.build());
+  }
+  
+  /**
+   * The indices for the `SparseTensor`.
+   */
+  public Output<TInt64> sparseIndices() {
+    return sparseIndices;
+  }
+  
+  /**
+   * The values of the `SparseTensor`.
+   */
+  public Output<U> sparseValues() {
+    return sparseValues;
+  }
+  
+  /**
+   * `sparse_dense_shape` is a tight bounding box of the input `RaggedTensor`.
+   */
+  public Output<TInt64> sparseDenseShape() {
+    return sparseDenseShape;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "RaggedTensorToSparse";
+  
+  private Output<TInt64> sparseIndices;
+  private Output<U> sparseValues;
+  private Output<TInt64> sparseDenseShape;
+  
+  private RaggedTensorToSparse(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    sparseIndices = operation.output(outputIdx++);
+    sparseValues = operation.output(outputIdx++);
+    sparseDenseShape = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/ragged/RaggedTensorToTensor.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/ragged/RaggedTensorToTensor.java
new file mode 100644
index 00000000000..e469e6fc020
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/ragged/RaggedTensorToTensor.java
@@ -0,0 +1,146 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.ragged;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.Operands;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Create a dense tensor from a ragged tensor, possibly altering its shape.
+ * <p>
+ * The `ragged_to_dense` op creates a dense tensor from a list of row partition
+ * tensors, a value vector, and default values. If the shape is unspecified, the
+ * minimal shape required to contain all the elements in the ragged tensor (the
+ * natural shape) will be used. If some dimensions are left unspecified, then the
+ * size of the natural shape is used in that dimension.
+ * <p>
+ * The default_value will be broadcast to the output shape. After that, the values
+ * from the ragged tensor overwrite the default values. Note that the default_value
+ * must have less dimensions than the value.
+ * <p>
+ * The row partition tensors are in the order of the dimensions.
+ * At present, the types can be:
+ * <ul>
+ * <li>
+ * "ROW_SPLITS": the row_splits tensor from the ragged tensor.
+ * </li>
+ * <li>
+ * "VALUE_ROWIDS": the value_rowids tensor from the ragged tensor.
+ * </li>
+ * <li>
+ * "FIRST_DIM_SIZE": if value_rowids is used for the first dimension, then it
+ *   is preceded by "FIRST_DIM_SIZE".
+ * 
+ * @param <U> data type for {@code result()} output
+ */
+public final class RaggedTensorToTensor<U extends TType> extends RawOp implements Operand<U> {
+  
+  /**
+   * Factory method to create a class wrapping a new RaggedTensorToTensor operation.
+   * 
+   * @param scope current scope
+   * @param shape The desired shape of the the output tensor. If left unspecified (empty),
+   * the minimal shape required to contain all the elements in the ragged tensor
+   * (the natural shape) will be used. If some dimensions are left unspecified, then
+   * the size of the natural shape is used in that dimension.
+   * <p>
+   * Note that dense dimensions cannot be modified by the shape argument. Trying to
+   * change the size of a dense dimension will cause the op to fail.
+   * Examples:
+   * natural shape: [4, 5, 6]
+   * shape: -1
+   * output shape: [4, 5, 6]
+   * <p>
+   * natural shape: [4, 5, 6]
+   * shape: [3, -1, 2]
+   * output shape: [3, 5, 2]
+   * <p>
+   * natural shape: [4, 5, 6]
+   * shape: [3, 7, 2]
+   * output shape: [3, 7, 2]
+   * 
+   * @param values A 1D tensor representing the values of the ragged tensor.
+   * @param defaultValue The default_value when the shape is larger than the ragged tensor. The
+   * default_value is broadcast until it is the shape of the output tensor, and
+   * then overwritten by values in the ragged tensor. The default value must be
+   * compatible with this broadcast operation, and must have fewer dimensions than
+   * the value tensor.
+   * @param rowPartitionTensors 
+   * @param rowPartitionTypes The types of the row partition tensors. At present, these can be:
+   * <ul>
+   * <li>
+   * "ROW_SPLITS": the row_splits tensor from the ragged tensor.
+   * </li>
+   * <li>
+   * "VALUE_ROWIDS": the value_rowids tensor from the ragged tensor.
+   * </li>
+   * <li>
+   * "FIRST_DIM_SIZE": if value_rowids is used for the first dimension, then it
+   *   is preceeded by "FIRST_DIM_SIZE".
+   * The tensors are in the order of the dimensions.
+   * @return a new instance of RaggedTensorToTensor
+   */
+  @Endpoint(describeByClass = true)
+  public static <U extends TType, T extends TNumber, V extends TNumber> RaggedTensorToTensor<U> create(Scope scope, Operand<T> shape, Operand<U> values, Operand<U> defaultValue, Iterable<Operand<V>> rowPartitionTensors, List<String> rowPartitionTypes) {
+    OperationBuilder opBuilder = scope.env().opBuilder("RaggedTensorToTensor", scope.makeOpName("RaggedTensorToTensor"));
+    opBuilder.addInput(shape.asOutput());
+    opBuilder.addInput(values.asOutput());
+    opBuilder.addInput(defaultValue.asOutput());
+    opBuilder.addInputList(Operands.asOutputs(rowPartitionTensors));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    String[] rowPartitionTypesArray = new String[rowPartitionTypes.size()];
+    for (int i = 0; i < rowPartitionTypesArray.length; ++i) {
+      rowPartitionTypesArray[i] = rowPartitionTypes.get(i);
+    }
+    opBuilder.setAttr("row_partition_types", rowPartitionTypesArray);
+    return new RaggedTensorToTensor<U>(opBuilder.build());
+  }
+  
+  /**
+   * The resulting dense tensor.
+   */
+  public Output<U> result() {
+    return result;
+  }
+  
+  @Override
+  public Output<U> asOutput() {
+    return result;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "RaggedTensorToTensor";
+  
+  private Output<U> result;
+  
+  private RaggedTensorToTensor(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    result = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/ragged/RaggedTensorToVariant.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/ragged/RaggedTensorToVariant.java
new file mode 100644
index 00000000000..6c22a0c4529
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/ragged/RaggedTensorToVariant.java
@@ -0,0 +1,93 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.ragged;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.Operands;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Encodes a `RaggedTensor` into a `variant` Tensor.
+ * <p>
+ * 
+ * Encodes the given `RaggedTensor` and returns a `variant` Tensor. If
+ * `batched_input` is True, then input `RaggedTensor` is unbatched along the
+ * zero-th dimension, each component `RaggedTensor` is encoded into a scalar
+ * `variant` Tensor, and these are stacked to return a 1-D `variant` Tensor.
+ * If `batched_input` is False, then the input `RaggedTensor` is encoded as is and
+ * a scalar `variant` Tensor is returned. A `RaggedTensor` is encoded by first
+ * creating a 1-D `variant` Tensor with `ragged_rank + 1` elements, containing the
+ * splits and values Tensors of the `RaggedTensor`. Then the 1-D `variant` Tensor
+ * is wrapped in a scalar `variant` Tensor. See `RaggedTensorFromVariant` for the
+ * corresponding decoding logic.
+ * 
+ */
+public final class RaggedTensorToVariant extends RawOp implements Operand<TType> {
+  
+  /**
+   * Factory method to create a class wrapping a new RaggedTensorToVariant operation.
+   * 
+   * @param scope current scope
+   * @param rtNestedSplits A list of one or more Tensors representing the splits of the input
+   * `RaggedTensor`.
+   * @param rtDenseValues A Tensor representing the values of the input `RaggedTensor`.
+   * @param batchedInput A `bool` denoting whether the input is a batched `RaggedTensor`.
+   * @return a new instance of RaggedTensorToVariant
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber, U extends TType> RaggedTensorToVariant create(Scope scope, Iterable<Operand<T>> rtNestedSplits, Operand<U> rtDenseValues, Boolean batchedInput) {
+    OperationBuilder opBuilder = scope.env().opBuilder("RaggedTensorToVariant", scope.makeOpName("RaggedTensorToVariant"));
+    opBuilder.addInputList(Operands.asOutputs(rtNestedSplits));
+    opBuilder.addInput(rtDenseValues.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("batched_input", batchedInput);
+    return new RaggedTensorToVariant(opBuilder.build());
+  }
+  
+  /**
+   * A `variant` Tensor that containing encoded `RaggedTensor`.
+   */
+  public Output<?> encodedRagged() {
+    return encodedRagged;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) encodedRagged;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "RaggedTensorToVariant";
+  
+  private Output<?> encodedRagged;
+  
+  private RaggedTensorToVariant(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    encodedRagged = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/AllCandidateSampler.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/AllCandidateSampler.java
new file mode 100644
index 00000000000..de24d12678f
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/AllCandidateSampler.java
@@ -0,0 +1,169 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.random;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.TInt64;
+
+/**
+ * Generates labels for candidate sampling with a learned unigram distribution.
+ * <p>
+ * See explanations of candidate sampling and the data formats at
+ * go/candidate-sampling.
+ * <p>
+ * For each batch, this op picks a single set of sampled candidate labels.
+ * <p>
+ * The advantages of sampling candidates per-batch are simplicity and the
+ * possibility of efficient dense matrix multiplication. The disadvantage is that
+ * the sampled candidates must be chosen independently of the context and of the
+ * true labels.
+ */
+@Operator(group = "random")
+public final class AllCandidateSampler extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.random.AllCandidateSampler}
+   */
+  public static class Options {
+    
+    /**
+     * @param seed If either seed or seed2 are set to be non-zero, the random number
+     * generator is seeded by the given seed.  Otherwise, it is seeded by a
+     * random seed.
+     */
+    public Options seed(Long seed) {
+      this.seed = seed;
+      return this;
+    }
+    
+    /**
+     * @param seed2 An second seed to avoid seed collision.
+     */
+    public Options seed2(Long seed2) {
+      this.seed2 = seed2;
+      return this;
+    }
+    
+    private Long seed;
+    private Long seed2;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new AllCandidateSampler operation.
+   * 
+   * @param scope current scope
+   * @param trueClasses A batch_size * num_true matrix, in which each row contains the
+   * IDs of the num_true target_classes in the corresponding original label.
+   * @param numTrue Number of true labels per context.
+   * @param numSampled Number of candidates to produce.
+   * @param unique If unique is true, we sample with rejection, so that all sampled
+   * candidates in a batch are unique. This requires some approximation to
+   * estimate the post-rejection sampling probabilities.
+   * @param options carries optional attributes values
+   * @return a new instance of AllCandidateSampler
+   */
+  @Endpoint(describeByClass = true)
+  public static AllCandidateSampler create(Scope scope, Operand<TInt64> trueClasses, Long numTrue, Long numSampled, Boolean unique, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("AllCandidateSampler", scope.makeOpName("AllCandidateSampler"));
+    opBuilder.addInput(trueClasses.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("num_true", numTrue);
+    opBuilder.setAttr("num_sampled", numSampled);
+    opBuilder.setAttr("unique", unique);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.seed != null) {
+          opBuilder.setAttr("seed", opts.seed);
+        }
+        if (opts.seed2 != null) {
+          opBuilder.setAttr("seed2", opts.seed2);
+        }
+      }
+    }
+    return new AllCandidateSampler(opBuilder.build());
+  }
+  
+  /**
+   * @param seed If either seed or seed2 are set to be non-zero, the random number
+   * generator is seeded by the given seed.  Otherwise, it is seeded by a
+   * random seed.
+   */
+  public static Options seed(Long seed) {
+    return new Options().seed(seed);
+  }
+  
+  /**
+   * @param seed2 An second seed to avoid seed collision.
+   */
+  public static Options seed2(Long seed2) {
+    return new Options().seed2(seed2);
+  }
+  
+  /**
+   * A vector of length num_sampled, in which each element is
+   * the ID of a sampled candidate.
+   */
+  public Output<TInt64> sampledCandidates() {
+    return sampledCandidates;
+  }
+  
+  /**
+   * A batch_size * num_true matrix, representing
+   * the number of times each candidate is expected to occur in a batch
+   * of sampled candidates. If unique=true, then this is a probability.
+   */
+  public Output<TFloat32> trueExpectedCount() {
+    return trueExpectedCount;
+  }
+  
+  /**
+   * A vector of length num_sampled, for each sampled
+   * candidate representing the number of times the candidate is expected
+   * to occur in a batch of sampled candidates.  If unique=true, then this is a
+   * probability.
+   */
+  public Output<TFloat32> sampledExpectedCount() {
+    return sampledExpectedCount;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "AllCandidateSampler";
+  
+  private Output<TInt64> sampledCandidates;
+  private Output<TFloat32> trueExpectedCount;
+  private Output<TFloat32> sampledExpectedCount;
+  
+  private AllCandidateSampler(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    sampledCandidates = operation.output(outputIdx++);
+    trueExpectedCount = operation.output(outputIdx++);
+    sampledExpectedCount = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/AnonymousRandomSeedGenerator.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/AnonymousRandomSeedGenerator.java
new file mode 100644
index 00000000000..a87026cd7f1
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/AnonymousRandomSeedGenerator.java
@@ -0,0 +1,75 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.random;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt64;
+
+/**
+ */
+public final class AnonymousRandomSeedGenerator extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new AnonymousRandomSeedGenerator operation.
+   * 
+   * @param scope current scope
+   * @param seed 
+   * @param seed2 
+   * @return a new instance of AnonymousRandomSeedGenerator
+   */
+  @Endpoint(describeByClass = true)
+  public static AnonymousRandomSeedGenerator create(Scope scope, Operand<TInt64> seed, Operand<TInt64> seed2) {
+    OperationBuilder opBuilder = scope.env().opBuilder("AnonymousRandomSeedGenerator", scope.makeOpName("AnonymousRandomSeedGenerator"));
+    opBuilder.addInput(seed.asOutput());
+    opBuilder.addInput(seed2.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new AnonymousRandomSeedGenerator(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<?> handle() {
+    return handle;
+  }
+  
+  /**
+   */
+  public Output<?> deleter() {
+    return deleter;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "AnonymousRandomSeedGenerator";
+  
+  private Output<?> handle;
+  private Output<?> deleter;
+  
+  private AnonymousRandomSeedGenerator(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    handle = operation.output(outputIdx++);
+    deleter = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/AnonymousSeedGenerator.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/AnonymousSeedGenerator.java
new file mode 100644
index 00000000000..c724bb6d110
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/AnonymousSeedGenerator.java
@@ -0,0 +1,78 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.random;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TBool;
+import org.tensorflow.types.TInt64;
+
+/**
+ */
+public final class AnonymousSeedGenerator extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new AnonymousSeedGenerator operation.
+   * 
+   * @param scope current scope
+   * @param seed 
+   * @param seed2 
+   * @param reshuffle 
+   * @return a new instance of AnonymousSeedGenerator
+   */
+  @Endpoint(describeByClass = true)
+  public static AnonymousSeedGenerator create(Scope scope, Operand<TInt64> seed, Operand<TInt64> seed2, Operand<TBool> reshuffle) {
+    OperationBuilder opBuilder = scope.env().opBuilder("AnonymousSeedGenerator", scope.makeOpName("AnonymousSeedGenerator"));
+    opBuilder.addInput(seed.asOutput());
+    opBuilder.addInput(seed2.asOutput());
+    opBuilder.addInput(reshuffle.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new AnonymousSeedGenerator(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<?> handle() {
+    return handle;
+  }
+  
+  /**
+   */
+  public Output<?> deleter() {
+    return deleter;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "AnonymousSeedGenerator";
+  
+  private Output<?> handle;
+  private Output<?> deleter;
+  
+  private AnonymousSeedGenerator(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    handle = operation.output(outputIdx++);
+    deleter = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/DeleteRandomSeedGenerator.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/DeleteRandomSeedGenerator.java
new file mode 100644
index 00000000000..23b154f9d75
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/DeleteRandomSeedGenerator.java
@@ -0,0 +1,55 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.random;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+
+/**
+ */
+public final class DeleteRandomSeedGenerator extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new DeleteRandomSeedGenerator operation.
+   * 
+   * @param scope current scope
+   * @param handle 
+   * @param deleter 
+   * @return a new instance of DeleteRandomSeedGenerator
+   */
+  @Endpoint(describeByClass = true)
+  public static DeleteRandomSeedGenerator create(Scope scope, Operand<?> handle, Operand<?> deleter) {
+    OperationBuilder opBuilder = scope.env().opBuilder("DeleteRandomSeedGenerator", scope.makeOpName("DeleteRandomSeedGenerator"));
+    opBuilder.addInput(handle.asOutput());
+    opBuilder.addInput(deleter.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new DeleteRandomSeedGenerator(opBuilder.build());
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "DeleteRandomSeedGenerator";
+  
+  private DeleteRandomSeedGenerator(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/DeleteSeedGenerator.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/DeleteSeedGenerator.java
new file mode 100644
index 00000000000..16982946d1f
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/DeleteSeedGenerator.java
@@ -0,0 +1,55 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.random;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+
+/**
+ */
+public final class DeleteSeedGenerator extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new DeleteSeedGenerator operation.
+   * 
+   * @param scope current scope
+   * @param handle 
+   * @param deleter 
+   * @return a new instance of DeleteSeedGenerator
+   */
+  @Endpoint(describeByClass = true)
+  public static DeleteSeedGenerator create(Scope scope, Operand<?> handle, Operand<?> deleter) {
+    OperationBuilder opBuilder = scope.env().opBuilder("DeleteSeedGenerator", scope.makeOpName("DeleteSeedGenerator"));
+    opBuilder.addInput(handle.asOutput());
+    opBuilder.addInput(deleter.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new DeleteSeedGenerator(opBuilder.build());
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "DeleteSeedGenerator";
+  
+  private DeleteSeedGenerator(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/LogUniformCandidateSampler.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/LogUniformCandidateSampler.java
new file mode 100644
index 00000000000..761f945de1a
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/LogUniformCandidateSampler.java
@@ -0,0 +1,171 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.random;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.TInt64;
+
+/**
+ * Generates labels for candidate sampling with a log-uniform distribution.
+ * <p>
+ * See explanations of candidate sampling and the data formats at
+ * go/candidate-sampling.
+ * <p>
+ * For each batch, this op picks a single set of sampled candidate labels.
+ * <p>
+ * The advantages of sampling candidates per-batch are simplicity and the
+ * possibility of efficient dense matrix multiplication. The disadvantage is that
+ * the sampled candidates must be chosen independently of the context and of the
+ * true labels.
+ */
+@Operator(group = "random")
+public final class LogUniformCandidateSampler extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.random.LogUniformCandidateSampler}
+   */
+  public static class Options {
+    
+    /**
+     * @param seed If either seed or seed2 are set to be non-zero, the random number
+     * generator is seeded by the given seed.  Otherwise, it is seeded by a
+     * random seed.
+     */
+    public Options seed(Long seed) {
+      this.seed = seed;
+      return this;
+    }
+    
+    /**
+     * @param seed2 An second seed to avoid seed collision.
+     */
+    public Options seed2(Long seed2) {
+      this.seed2 = seed2;
+      return this;
+    }
+    
+    private Long seed;
+    private Long seed2;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new LogUniformCandidateSampler operation.
+   * 
+   * @param scope current scope
+   * @param trueClasses A batch_size * num_true matrix, in which each row contains the
+   * IDs of the num_true target_classes in the corresponding original label.
+   * @param numTrue Number of true labels per context.
+   * @param numSampled Number of candidates to randomly sample.
+   * @param unique If unique is true, we sample with rejection, so that all sampled
+   * candidates in a batch are unique. This requires some approximation to
+   * estimate the post-rejection sampling probabilities.
+   * @param rangeMax The sampler will sample integers from the interval [0, range_max).
+   * @param options carries optional attributes values
+   * @return a new instance of LogUniformCandidateSampler
+   */
+  @Endpoint(describeByClass = true)
+  public static LogUniformCandidateSampler create(Scope scope, Operand<TInt64> trueClasses, Long numTrue, Long numSampled, Boolean unique, Long rangeMax, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("LogUniformCandidateSampler", scope.makeOpName("LogUniformCandidateSampler"));
+    opBuilder.addInput(trueClasses.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("num_true", numTrue);
+    opBuilder.setAttr("num_sampled", numSampled);
+    opBuilder.setAttr("unique", unique);
+    opBuilder.setAttr("range_max", rangeMax);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.seed != null) {
+          opBuilder.setAttr("seed", opts.seed);
+        }
+        if (opts.seed2 != null) {
+          opBuilder.setAttr("seed2", opts.seed2);
+        }
+      }
+    }
+    return new LogUniformCandidateSampler(opBuilder.build());
+  }
+  
+  /**
+   * @param seed If either seed or seed2 are set to be non-zero, the random number
+   * generator is seeded by the given seed.  Otherwise, it is seeded by a
+   * random seed.
+   */
+  public static Options seed(Long seed) {
+    return new Options().seed(seed);
+  }
+  
+  /**
+   * @param seed2 An second seed to avoid seed collision.
+   */
+  public static Options seed2(Long seed2) {
+    return new Options().seed2(seed2);
+  }
+  
+  /**
+   * A vector of length num_sampled, in which each element is
+   * the ID of a sampled candidate.
+   */
+  public Output<TInt64> sampledCandidates() {
+    return sampledCandidates;
+  }
+  
+  /**
+   * A batch_size * num_true matrix, representing
+   * the number of times each candidate is expected to occur in a batch
+   * of sampled candidates. If unique=true, then this is a probability.
+   */
+  public Output<TFloat32> trueExpectedCount() {
+    return trueExpectedCount;
+  }
+  
+  /**
+   * A vector of length num_sampled, for each sampled
+   * candidate representing the number of times the candidate is expected
+   * to occur in a batch of sampled candidates.  If unique=true, then this is a
+   * probability.
+   */
+  public Output<TFloat32> sampledExpectedCount() {
+    return sampledExpectedCount;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "LogUniformCandidateSampler";
+  
+  private Output<TInt64> sampledCandidates;
+  private Output<TFloat32> trueExpectedCount;
+  private Output<TFloat32> sampledExpectedCount;
+  
+  private LogUniformCandidateSampler(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    sampledCandidates = operation.output(outputIdx++);
+    trueExpectedCount = operation.output(outputIdx++);
+    sampledExpectedCount = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/Multinomial.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/Multinomial.java
new file mode 100644
index 00000000000..fbfb90fba47
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/Multinomial.java
@@ -0,0 +1,153 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.random;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Draws samples from a multinomial distribution.
+ * 
+ * @param <U> data type for {@code output()} output
+ */
+@Operator(group = "random")
+public final class Multinomial<U extends TNumber> extends RawOp implements Operand<U> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.random.Multinomial}
+   */
+  public static class Options {
+    
+    /**
+     * @param seed If either seed or seed2 is set to be non-zero, the internal random number
+     * generator is seeded by the given seed.  Otherwise, a random seed is used.
+     */
+    public Options seed(Long seed) {
+      this.seed = seed;
+      return this;
+    }
+    
+    /**
+     * @param seed2 A second seed to avoid seed collision.
+     */
+    public Options seed2(Long seed2) {
+      this.seed2 = seed2;
+      return this;
+    }
+    
+    private Long seed;
+    private Long seed2;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new Multinomial operation.
+   * 
+   * @param scope current scope
+   * @param logits 2-D Tensor with shape `[batch_size, num_classes]`.  Each slice `[i, :]`
+   * represents the unnormalized log probabilities for all classes.
+   * @param numSamples 0-D.  Number of independent samples to draw for each row slice.
+   * @param outputDtype 
+   * @param options carries optional attributes values
+   * @return a new instance of Multinomial
+   */
+  @Endpoint(describeByClass = true)
+  public static <U extends TNumber, T extends TNumber> Multinomial<U> create(Scope scope, Operand<T> logits, Operand<TInt32> numSamples, Class<U> outputDtype, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Multinomial", scope.makeOpName("Multinomial"));
+    opBuilder.addInput(logits.asOutput());
+    opBuilder.addInput(numSamples.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("output_dtype", outputDtype);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.seed != null) {
+          opBuilder.setAttr("seed", opts.seed);
+        }
+        if (opts.seed2 != null) {
+          opBuilder.setAttr("seed2", opts.seed2);
+        }
+      }
+    }
+    return new Multinomial<U>(opBuilder.build());
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new Multinomial operation using default output types.
+   * 
+   * @param scope current scope
+   * @param logits 2-D Tensor with shape `[batch_size, num_classes]`.  Each slice `[i, :]`
+   * represents the unnormalized log probabilities for all classes.
+   * @param numSamples 0-D.  Number of independent samples to draw for each row slice.
+   * @param options carries optional attributes values
+   * @return a new instance of Multinomial
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> Multinomial<TInt64> create(Scope scope, Operand<T> logits, Operand<TInt32> numSamples, Options... options) {
+    return create(scope, logits, numSamples, TInt64.class, options);
+  }
+  
+  /**
+   * @param seed If either seed or seed2 is set to be non-zero, the internal random number
+   * generator is seeded by the given seed.  Otherwise, a random seed is used.
+   */
+  public static Options seed(Long seed) {
+    return new Options().seed(seed);
+  }
+  
+  /**
+   * @param seed2 A second seed to avoid seed collision.
+   */
+  public static Options seed2(Long seed2) {
+    return new Options().seed2(seed2);
+  }
+  
+  /**
+   * 2-D Tensor with shape `[batch_size, num_samples]`.  Each slice `[i, :]`
+   * contains the drawn class labels with range `[0, num_classes)`.
+   */
+  public Output<U> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<U> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Multinomial";
+  
+  private Output<U> output;
+  
+  private Multinomial(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/NonDeterministicInts.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/NonDeterministicInts.java
new file mode 100644
index 00000000000..80ee33ef06b
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/NonDeterministicInts.java
@@ -0,0 +1,91 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.random;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Non-deterministically generates some integers.
+ * <p>
+ * This op may use some OS-provided source of non-determinism (e.g. an RNG), so each execution will give different results.
+ * 
+ * @param <U> data type for {@code output()} output
+ */
+public final class NonDeterministicInts<U extends TType> extends RawOp implements Operand<U> {
+  
+  /**
+   * Factory method to create a class wrapping a new NonDeterministicInts operation.
+   * 
+   * @param scope current scope
+   * @param shape The shape of the output tensor.
+   * @param dtype The type of the output.
+   * @return a new instance of NonDeterministicInts
+   */
+  @Endpoint(describeByClass = true)
+  public static <U extends TType, T extends TType> NonDeterministicInts<U> create(Scope scope, Operand<T> shape, Class<U> dtype) {
+    OperationBuilder opBuilder = scope.env().opBuilder("NonDeterministicInts", scope.makeOpName("NonDeterministicInts"));
+    opBuilder.addInput(shape.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("dtype", dtype);
+    return new NonDeterministicInts<U>(opBuilder.build());
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new NonDeterministicInts operation using default output types.
+   * 
+   * @param scope current scope
+   * @param shape The shape of the output tensor.
+   * @return a new instance of NonDeterministicInts
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> NonDeterministicInts<TInt64> create(Scope scope, Operand<T> shape) {
+    return create(scope, shape, TInt64.class);
+  }
+  
+  /**
+   * Non-deterministic integer values with specified shape.
+   */
+  public Output<U> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<U> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "NonDeterministicInts";
+  
+  private Output<U> output;
+  
+  private NonDeterministicInts(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/ParameterizedTruncatedNormal.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/ParameterizedTruncatedNormal.java
new file mode 100644
index 00000000000..c9933d29a17
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/ParameterizedTruncatedNormal.java
@@ -0,0 +1,145 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.random;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Outputs random values from a normal distribution. The parameters may each be a
+ * <p>
+ * scalar which applies to the entire output, or a vector of length shape[0] which
+ * stores the parameters for each batch.
+ * 
+ * @param <U> data type for {@code output()} output
+ */
+@Operator(group = "random")
+public final class ParameterizedTruncatedNormal<U extends TNumber> extends RawOp implements Operand<U> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.random.ParameterizedTruncatedNormal}
+   */
+  public static class Options {
+    
+    /**
+     * @param seed If either `seed` or `seed2` are set to be non-zero, the random number
+     * generator is seeded by the given seed.  Otherwise, it is seeded by a
+     * random seed.
+     */
+    public Options seed(Long seed) {
+      this.seed = seed;
+      return this;
+    }
+    
+    /**
+     * @param seed2 A second seed to avoid seed collision.
+     */
+    public Options seed2(Long seed2) {
+      this.seed2 = seed2;
+      return this;
+    }
+    
+    private Long seed;
+    private Long seed2;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new ParameterizedTruncatedNormal operation.
+   * 
+   * @param scope current scope
+   * @param shape The shape of the output tensor. Batches are indexed by the 0th dimension.
+   * @param means The mean parameter of each batch.
+   * @param stdevs The standard deviation parameter of each batch. Must be greater than 0.
+   * @param minvals The minimum cutoff. May be -infinity.
+   * @param maxvals The maximum cutoff. May be +infinity, and must be more than the minval
+   * for each batch.
+   * @param options carries optional attributes values
+   * @return a new instance of ParameterizedTruncatedNormal
+   */
+  @Endpoint(describeByClass = true)
+  public static <U extends TNumber, T extends TNumber> ParameterizedTruncatedNormal<U> create(Scope scope, Operand<T> shape, Operand<U> means, Operand<U> stdevs, Operand<U> minvals, Operand<U> maxvals, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ParameterizedTruncatedNormal", scope.makeOpName("ParameterizedTruncatedNormal"));
+    opBuilder.addInput(shape.asOutput());
+    opBuilder.addInput(means.asOutput());
+    opBuilder.addInput(stdevs.asOutput());
+    opBuilder.addInput(minvals.asOutput());
+    opBuilder.addInput(maxvals.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.seed != null) {
+          opBuilder.setAttr("seed", opts.seed);
+        }
+        if (opts.seed2 != null) {
+          opBuilder.setAttr("seed2", opts.seed2);
+        }
+      }
+    }
+    return new ParameterizedTruncatedNormal<U>(opBuilder.build());
+  }
+  
+  /**
+   * @param seed If either `seed` or `seed2` are set to be non-zero, the random number
+   * generator is seeded by the given seed.  Otherwise, it is seeded by a
+   * random seed.
+   */
+  public static Options seed(Long seed) {
+    return new Options().seed(seed);
+  }
+  
+  /**
+   * @param seed2 A second seed to avoid seed collision.
+   */
+  public static Options seed2(Long seed2) {
+    return new Options().seed2(seed2);
+  }
+  
+  /**
+   * A matrix of shape num_batches x samples_per_batch, filled with random
+   * truncated normal values using the parameters for each row.
+   */
+  public Output<U> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<U> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ParameterizedTruncatedNormal";
+  
+  private Output<U> output;
+  
+  private ParameterizedTruncatedNormal(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RandomGamma.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RandomGamma.java
new file mode 100644
index 00000000000..aee60a06932
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RandomGamma.java
@@ -0,0 +1,142 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.random;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Outputs random values from the Gamma distribution(s) described by alpha.
+ * <p>
+ * This op uses the algorithm by Marsaglia et al. to acquire samples via
+ * transformation-rejection from pairs of uniform and normal random variables.
+ * See http://dl.acm.org/citation.cfm?id=358414
+ * 
+ * @param <U> data type for {@code output()} output
+ */
+@Operator(group = "random")
+public final class RandomGamma<U extends TNumber> extends RawOp implements Operand<U> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.random.RandomGamma}
+   */
+  public static class Options {
+    
+    /**
+     * @param seed If either `seed` or `seed2` are set to be non-zero, the random number
+     * generator is seeded by the given seed.  Otherwise, it is seeded by a
+     * random seed.
+     */
+    public Options seed(Long seed) {
+      this.seed = seed;
+      return this;
+    }
+    
+    /**
+     * @param seed2 A second seed to avoid seed collision.
+     */
+    public Options seed2(Long seed2) {
+      this.seed2 = seed2;
+      return this;
+    }
+    
+    private Long seed;
+    private Long seed2;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new RandomGamma operation.
+   * 
+   * @param scope current scope
+   * @param shape 1-D integer tensor. Shape of independent samples to draw from each
+   * distribution described by the shape parameters given in alpha.
+   * @param alpha A tensor in which each scalar is a "shape" parameter describing the
+   * associated gamma distribution.
+   * @param options carries optional attributes values
+   * @return a new instance of RandomGamma
+   */
+  @Endpoint(describeByClass = true)
+  public static <U extends TNumber, T extends TNumber> RandomGamma<U> create(Scope scope, Operand<T> shape, Operand<U> alpha, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("RandomGamma", scope.makeOpName("RandomGamma"));
+    opBuilder.addInput(shape.asOutput());
+    opBuilder.addInput(alpha.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.seed != null) {
+          opBuilder.setAttr("seed", opts.seed);
+        }
+        if (opts.seed2 != null) {
+          opBuilder.setAttr("seed2", opts.seed2);
+        }
+      }
+    }
+    return new RandomGamma<U>(opBuilder.build());
+  }
+  
+  /**
+   * @param seed If either `seed` or `seed2` are set to be non-zero, the random number
+   * generator is seeded by the given seed.  Otherwise, it is seeded by a
+   * random seed.
+   */
+  public static Options seed(Long seed) {
+    return new Options().seed(seed);
+  }
+  
+  /**
+   * @param seed2 A second seed to avoid seed collision.
+   */
+  public static Options seed2(Long seed2) {
+    return new Options().seed2(seed2);
+  }
+  
+  /**
+   * A tensor with shape `shape + shape(alpha)`. Each slice
+   * `[:, ..., :, i0, i1, ...iN]` contains the samples drawn for
+   * `alpha[i0, i1, ...iN]`. The dtype of the output matches the dtype of alpha.
+   */
+  public Output<U> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<U> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "RandomGamma";
+  
+  private Output<U> output;
+  
+  private RandomGamma(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RandomGammaGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RandomGammaGrad.java
new file mode 100644
index 00000000000..29c7c625e17
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RandomGammaGrad.java
@@ -0,0 +1,75 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.random;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Computes the derivative of a Gamma random sample w.r.t. `alpha`.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+public final class RandomGammaGrad<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new RandomGammaGrad operation.
+   * 
+   * @param scope current scope
+   * @param alpha 
+   * @param sample 
+   * @return a new instance of RandomGammaGrad
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> RandomGammaGrad<T> create(Scope scope, Operand<T> alpha, Operand<T> sample) {
+    OperationBuilder opBuilder = scope.env().opBuilder("RandomGammaGrad", scope.makeOpName("RandomGammaGrad"));
+    opBuilder.addInput(alpha.asOutput());
+    opBuilder.addInput(sample.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new RandomGammaGrad<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "RandomGammaGrad";
+  
+  private Output<T> output;
+  
+  private RandomGammaGrad(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RandomPoisson.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RandomPoisson.java
new file mode 100644
index 00000000000..04da9a55cdd
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RandomPoisson.java
@@ -0,0 +1,167 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.random;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Outputs random values from the Poisson distribution(s) described by rate.
+ * <p>
+ * This op uses two algorithms, depending on rate. If rate >= 10, then
+ * the algorithm by Hormann is used to acquire samples via
+ * transformation-rejection.
+ * See http://www.sciencedirect.com/science/article/pii/0167668793909974.
+ * <p>
+ * Otherwise, Knuth's algorithm is used to acquire samples via multiplying uniform
+ * random variables.
+ * See Donald E. Knuth (1969). Seminumerical Algorithms. The Art of Computer
+ * Programming, Volume 2. Addison Wesley
+ * 
+ * @param <V> data type for {@code output()} output
+ */
+@Operator(group = "random")
+public final class RandomPoisson<V extends TNumber> extends RawOp implements Operand<V> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.random.RandomPoisson}
+   */
+  public static class Options {
+    
+    /**
+     * @param seed If either `seed` or `seed2` are set to be non-zero, the random number
+     * generator is seeded by the given seed.  Otherwise, it is seeded by a
+     * random seed.
+     */
+    public Options seed(Long seed) {
+      this.seed = seed;
+      return this;
+    }
+    
+    /**
+     * @param seed2 A second seed to avoid seed collision.
+     */
+    public Options seed2(Long seed2) {
+      this.seed2 = seed2;
+      return this;
+    }
+    
+    private Long seed;
+    private Long seed2;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new RandomPoisson operation.
+   * 
+   * @param scope current scope
+   * @param shape 1-D integer tensor. Shape of independent samples to draw from each
+   * distribution described by the shape parameters given in rate.
+   * @param rate A tensor in which each scalar is a "rate" parameter describing the
+   * associated poisson distribution.
+   * @param dtype 
+   * @param options carries optional attributes values
+   * @return a new instance of RandomPoisson
+   */
+  @Endpoint(describeByClass = true)
+  public static <V extends TNumber, T extends TNumber, U extends TNumber> RandomPoisson<V> create(Scope scope, Operand<T> shape, Operand<U> rate, Class<V> dtype, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("RandomPoissonV2", scope.makeOpName("RandomPoisson"));
+    opBuilder.addInput(shape.asOutput());
+    opBuilder.addInput(rate.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("dtype", dtype);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.seed != null) {
+          opBuilder.setAttr("seed", opts.seed);
+        }
+        if (opts.seed2 != null) {
+          opBuilder.setAttr("seed2", opts.seed2);
+        }
+      }
+    }
+    return new RandomPoisson<V>(opBuilder.build());
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new RandomPoisson operation using default output types.
+   * 
+   * @param scope current scope
+   * @param shape 1-D integer tensor. Shape of independent samples to draw from each
+   * distribution described by the shape parameters given in rate.
+   * @param rate A tensor in which each scalar is a "rate" parameter describing the
+   * associated poisson distribution.
+   * @param options carries optional attributes values
+   * @return a new instance of RandomPoisson
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber, U extends TNumber> RandomPoisson<TInt64> create(Scope scope, Operand<T> shape, Operand<U> rate, Options... options) {
+    return create(scope, shape, rate, TInt64.class, options);
+  }
+  
+  /**
+   * @param seed If either `seed` or `seed2` are set to be non-zero, the random number
+   * generator is seeded by the given seed.  Otherwise, it is seeded by a
+   * random seed.
+   */
+  public static Options seed(Long seed) {
+    return new Options().seed(seed);
+  }
+  
+  /**
+   * @param seed2 A second seed to avoid seed collision.
+   */
+  public static Options seed2(Long seed2) {
+    return new Options().seed2(seed2);
+  }
+  
+  /**
+   * A tensor with shape `shape + shape(rate)`. Each slice
+   * `[:, ..., :, i0, i1, ...iN]` contains the samples drawn for
+   * `rate[i0, i1, ...iN]`.
+   */
+  public Output<V> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<V> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "RandomPoissonV2";
+  
+  private Output<V> output;
+  
+  private RandomPoisson(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RandomShuffle.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RandomShuffle.java
new file mode 100644
index 00000000000..b08c31031ec
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RandomShuffle.java
@@ -0,0 +1,143 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.random;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Randomly shuffles a tensor along its first dimension.
+ * <p>
+ *   The tensor is shuffled along dimension 0, such that each `value[j]` is mapped
+ *   to one and only one `output[i]`. For example, a mapping that might occur for a
+ *   3x2 tensor is:
+ * <pre>{@code
+ * [[1, 2],       [[5, 6],
+ *  [3, 4],  ==>   [1, 2],
+ *  [5, 6]]        [3, 4]]
+ * }</pre>
+ * 
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator(group = "random")
+public final class RandomShuffle<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.random.RandomShuffle}
+   */
+  public static class Options {
+    
+    /**
+     * @param seed If either `seed` or `seed2` are set to be non-zero, the random number
+     * generator is seeded by the given seed.  Otherwise, it is seeded by a
+     * random seed.
+     */
+    public Options seed(Long seed) {
+      this.seed = seed;
+      return this;
+    }
+    
+    /**
+     * @param seed2 A second seed to avoid seed collision.
+     */
+    public Options seed2(Long seed2) {
+      this.seed2 = seed2;
+      return this;
+    }
+    
+    private Long seed;
+    private Long seed2;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new RandomShuffle operation.
+   * 
+   * @param scope current scope
+   * @param value The tensor to be shuffled.
+   * @param options carries optional attributes values
+   * @return a new instance of RandomShuffle
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> RandomShuffle<T> create(Scope scope, Operand<T> value, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("RandomShuffle", scope.makeOpName("RandomShuffle"));
+    opBuilder.addInput(value.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.seed != null) {
+          opBuilder.setAttr("seed", opts.seed);
+        }
+        if (opts.seed2 != null) {
+          opBuilder.setAttr("seed2", opts.seed2);
+        }
+      }
+    }
+    return new RandomShuffle<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param seed If either `seed` or `seed2` are set to be non-zero, the random number
+   * generator is seeded by the given seed.  Otherwise, it is seeded by a
+   * random seed.
+   */
+  public static Options seed(Long seed) {
+    return new Options().seed(seed);
+  }
+  
+  /**
+   * @param seed2 A second seed to avoid seed collision.
+   */
+  public static Options seed2(Long seed2) {
+    return new Options().seed2(seed2);
+  }
+  
+  /**
+   * A tensor of same shape and type as `value`, shuffled along its first
+   * dimension.
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "RandomShuffle";
+  
+  private Output<T> output;
+  
+  private RandomShuffle(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RandomStandardNormal.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RandomStandardNormal.java
new file mode 100644
index 00000000000..c9472874e46
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RandomStandardNormal.java
@@ -0,0 +1,136 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.random;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Outputs random values from a normal distribution.
+ * <p>
+ * The generated values will have mean 0 and standard deviation 1.
+ * 
+ * @param <U> data type for {@code output()} output
+ */
+@Operator(group = "random")
+public final class RandomStandardNormal<U extends TNumber> extends RawOp implements Operand<U> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.random.RandomStandardNormal}
+   */
+  public static class Options {
+    
+    /**
+     * @param seed If either `seed` or `seed2` are set to be non-zero, the random number
+     * generator is seeded by the given seed.  Otherwise, it is seeded by a
+     * random seed.
+     */
+    public Options seed(Long seed) {
+      this.seed = seed;
+      return this;
+    }
+    
+    /**
+     * @param seed2 A second seed to avoid seed collision.
+     */
+    public Options seed2(Long seed2) {
+      this.seed2 = seed2;
+      return this;
+    }
+    
+    private Long seed;
+    private Long seed2;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new RandomStandardNormal operation.
+   * 
+   * @param scope current scope
+   * @param shape The shape of the output tensor.
+   * @param dtype The type of the output.
+   * @param options carries optional attributes values
+   * @return a new instance of RandomStandardNormal
+   */
+  @Endpoint(describeByClass = true)
+  public static <U extends TNumber, T extends TNumber> RandomStandardNormal<U> create(Scope scope, Operand<T> shape, Class<U> dtype, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("RandomStandardNormal", scope.makeOpName("RandomStandardNormal"));
+    opBuilder.addInput(shape.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("dtype", dtype);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.seed != null) {
+          opBuilder.setAttr("seed", opts.seed);
+        }
+        if (opts.seed2 != null) {
+          opBuilder.setAttr("seed2", opts.seed2);
+        }
+      }
+    }
+    return new RandomStandardNormal<U>(opBuilder.build());
+  }
+  
+  /**
+   * @param seed If either `seed` or `seed2` are set to be non-zero, the random number
+   * generator is seeded by the given seed.  Otherwise, it is seeded by a
+   * random seed.
+   */
+  public static Options seed(Long seed) {
+    return new Options().seed(seed);
+  }
+  
+  /**
+   * @param seed2 A second seed to avoid seed collision.
+   */
+  public static Options seed2(Long seed2) {
+    return new Options().seed2(seed2);
+  }
+  
+  /**
+   * A tensor of the specified shape filled with random normal values.
+   */
+  public Output<U> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<U> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "RandomStandardNormal";
+  
+  private Output<U> output;
+  
+  private RandomStandardNormal(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RandomUniform.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RandomUniform.java
new file mode 100644
index 00000000000..43952690b7d
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RandomUniform.java
@@ -0,0 +1,137 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.random;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Outputs random values from a uniform distribution.
+ * <p>
+ * The generated values follow a uniform distribution in the range `[0, 1)`. The
+ * lower bound 0 is included in the range, while the upper bound 1 is excluded.
+ * 
+ * @param <U> data type for {@code output()} output
+ */
+@Operator(group = "random")
+public final class RandomUniform<U extends TNumber> extends RawOp implements Operand<U> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.random.RandomUniform}
+   */
+  public static class Options {
+    
+    /**
+     * @param seed If either `seed` or `seed2` are set to be non-zero, the random number
+     * generator is seeded by the given seed.  Otherwise, it is seeded by a
+     * random seed.
+     */
+    public Options seed(Long seed) {
+      this.seed = seed;
+      return this;
+    }
+    
+    /**
+     * @param seed2 A second seed to avoid seed collision.
+     */
+    public Options seed2(Long seed2) {
+      this.seed2 = seed2;
+      return this;
+    }
+    
+    private Long seed;
+    private Long seed2;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new RandomUniform operation.
+   * 
+   * @param scope current scope
+   * @param shape The shape of the output tensor.
+   * @param dtype The type of the output.
+   * @param options carries optional attributes values
+   * @return a new instance of RandomUniform
+   */
+  @Endpoint(describeByClass = true)
+  public static <U extends TNumber, T extends TNumber> RandomUniform<U> create(Scope scope, Operand<T> shape, Class<U> dtype, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("RandomUniform", scope.makeOpName("RandomUniform"));
+    opBuilder.addInput(shape.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("dtype", dtype);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.seed != null) {
+          opBuilder.setAttr("seed", opts.seed);
+        }
+        if (opts.seed2 != null) {
+          opBuilder.setAttr("seed2", opts.seed2);
+        }
+      }
+    }
+    return new RandomUniform<U>(opBuilder.build());
+  }
+  
+  /**
+   * @param seed If either `seed` or `seed2` are set to be non-zero, the random number
+   * generator is seeded by the given seed.  Otherwise, it is seeded by a
+   * random seed.
+   */
+  public static Options seed(Long seed) {
+    return new Options().seed(seed);
+  }
+  
+  /**
+   * @param seed2 A second seed to avoid seed collision.
+   */
+  public static Options seed2(Long seed2) {
+    return new Options().seed2(seed2);
+  }
+  
+  /**
+   * A tensor of the specified shape filled with uniform random values.
+   */
+  public Output<U> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<U> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "RandomUniform";
+  
+  private Output<U> output;
+  
+  private RandomUniform(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RandomUniformInt.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RandomUniformInt.java
new file mode 100644
index 00000000000..1b59cf964e7
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RandomUniformInt.java
@@ -0,0 +1,144 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.random;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Outputs random integers from a uniform distribution.
+ * <p>
+ * The generated values are uniform integers in the range `[minval, maxval)`.
+ * The lower bound `minval` is included in the range, while the upper bound
+ * `maxval` is excluded.
+ * <p>
+ * The random integers are slightly biased unless `maxval - minval` is an exact
+ * power of two.  The bias is small for values of `maxval - minval` significantly
+ * smaller than the range of the output (either `2^32` or `2^64`).
+ * 
+ * @param <U> data type for {@code output()} output
+ */
+@Operator(group = "random")
+public final class RandomUniformInt<U extends TNumber> extends RawOp implements Operand<U> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.random.RandomUniformInt}
+   */
+  public static class Options {
+    
+    /**
+     * @param seed If either `seed` or `seed2` are set to be non-zero, the random number
+     * generator is seeded by the given seed.  Otherwise, it is seeded by a
+     * random seed.
+     */
+    public Options seed(Long seed) {
+      this.seed = seed;
+      return this;
+    }
+    
+    /**
+     * @param seed2 A second seed to avoid seed collision.
+     */
+    public Options seed2(Long seed2) {
+      this.seed2 = seed2;
+      return this;
+    }
+    
+    private Long seed;
+    private Long seed2;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new RandomUniformInt operation.
+   * 
+   * @param scope current scope
+   * @param shape The shape of the output tensor.
+   * @param minval 0-D.  Inclusive lower bound on the generated integers.
+   * @param maxval 0-D.  Exclusive upper bound on the generated integers.
+   * @param options carries optional attributes values
+   * @return a new instance of RandomUniformInt
+   */
+  @Endpoint(describeByClass = true)
+  public static <U extends TNumber, T extends TNumber> RandomUniformInt<U> create(Scope scope, Operand<T> shape, Operand<U> minval, Operand<U> maxval, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("RandomUniformInt", scope.makeOpName("RandomUniformInt"));
+    opBuilder.addInput(shape.asOutput());
+    opBuilder.addInput(minval.asOutput());
+    opBuilder.addInput(maxval.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.seed != null) {
+          opBuilder.setAttr("seed", opts.seed);
+        }
+        if (opts.seed2 != null) {
+          opBuilder.setAttr("seed2", opts.seed2);
+        }
+      }
+    }
+    return new RandomUniformInt<U>(opBuilder.build());
+  }
+  
+  /**
+   * @param seed If either `seed` or `seed2` are set to be non-zero, the random number
+   * generator is seeded by the given seed.  Otherwise, it is seeded by a
+   * random seed.
+   */
+  public static Options seed(Long seed) {
+    return new Options().seed(seed);
+  }
+  
+  /**
+   * @param seed2 A second seed to avoid seed collision.
+   */
+  public static Options seed2(Long seed2) {
+    return new Options().seed2(seed2);
+  }
+  
+  /**
+   * A tensor of the specified shape filled with uniform random integers.
+   */
+  public Output<U> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<U> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "RandomUniformInt";
+  
+  private Output<U> output;
+  
+  private RandomUniformInt(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RecordInput.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RecordInput.java
new file mode 100644
index 00000000000..7e3550cb71c
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RecordInput.java
@@ -0,0 +1,206 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.random;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TString;
+
+/**
+ * Emits randomized records.
+ */
+@Operator(group = "random")
+public final class RecordInput extends RawOp implements Operand<TString> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.random.RecordInput}
+   */
+  public static class Options {
+    
+    /**
+     * @param fileRandomSeed Random seeds used to produce randomized records.
+     */
+    public Options fileRandomSeed(Long fileRandomSeed) {
+      this.fileRandomSeed = fileRandomSeed;
+      return this;
+    }
+    
+    /**
+     * @param fileShuffleShiftRatio Shifts the list of files after the list is randomly
+     * shuffled.
+     */
+    public Options fileShuffleShiftRatio(Float fileShuffleShiftRatio) {
+      this.fileShuffleShiftRatio = fileShuffleShiftRatio;
+      return this;
+    }
+    
+    /**
+     * @param fileBufferSize The randomization shuffling buffer.
+     */
+    public Options fileBufferSize(Long fileBufferSize) {
+      this.fileBufferSize = fileBufferSize;
+      return this;
+    }
+    
+    /**
+     * @param fileParallelism How many sstables are opened and concurrently iterated over.
+     */
+    public Options fileParallelism(Long fileParallelism) {
+      this.fileParallelism = fileParallelism;
+      return this;
+    }
+    
+    /**
+     * @param batchSize The batch size.
+     */
+    public Options batchSize(Long batchSize) {
+      this.batchSize = batchSize;
+      return this;
+    }
+    
+    /**
+     * @param compressionType The type of compression for the file. Currently ZLIB and
+     * GZIP are supported. Defaults to none.
+     */
+    public Options compressionType(String compressionType) {
+      this.compressionType = compressionType;
+      return this;
+    }
+    
+    private Long fileRandomSeed;
+    private Float fileShuffleShiftRatio;
+    private Long fileBufferSize;
+    private Long fileParallelism;
+    private Long batchSize;
+    private String compressionType;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new RecordInput operation.
+   * 
+   * @param scope current scope
+   * @param filePattern Glob pattern for the data files.
+   * @param options carries optional attributes values
+   * @return a new instance of RecordInput
+   */
+  @Endpoint(describeByClass = true)
+  public static RecordInput create(Scope scope, String filePattern, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("RecordInput", scope.makeOpName("RecordInput"));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("file_pattern", filePattern);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.fileRandomSeed != null) {
+          opBuilder.setAttr("file_random_seed", opts.fileRandomSeed);
+        }
+        if (opts.fileShuffleShiftRatio != null) {
+          opBuilder.setAttr("file_shuffle_shift_ratio", opts.fileShuffleShiftRatio);
+        }
+        if (opts.fileBufferSize != null) {
+          opBuilder.setAttr("file_buffer_size", opts.fileBufferSize);
+        }
+        if (opts.fileParallelism != null) {
+          opBuilder.setAttr("file_parallelism", opts.fileParallelism);
+        }
+        if (opts.batchSize != null) {
+          opBuilder.setAttr("batch_size", opts.batchSize);
+        }
+        if (opts.compressionType != null) {
+          opBuilder.setAttr("compression_type", opts.compressionType);
+        }
+      }
+    }
+    return new RecordInput(opBuilder.build());
+  }
+  
+  /**
+   * @param fileRandomSeed Random seeds used to produce randomized records.
+   */
+  public static Options fileRandomSeed(Long fileRandomSeed) {
+    return new Options().fileRandomSeed(fileRandomSeed);
+  }
+  
+  /**
+   * @param fileShuffleShiftRatio Shifts the list of files after the list is randomly
+   * shuffled.
+   */
+  public static Options fileShuffleShiftRatio(Float fileShuffleShiftRatio) {
+    return new Options().fileShuffleShiftRatio(fileShuffleShiftRatio);
+  }
+  
+  /**
+   * @param fileBufferSize The randomization shuffling buffer.
+   */
+  public static Options fileBufferSize(Long fileBufferSize) {
+    return new Options().fileBufferSize(fileBufferSize);
+  }
+  
+  /**
+   * @param fileParallelism How many sstables are opened and concurrently iterated over.
+   */
+  public static Options fileParallelism(Long fileParallelism) {
+    return new Options().fileParallelism(fileParallelism);
+  }
+  
+  /**
+   * @param batchSize The batch size.
+   */
+  public static Options batchSize(Long batchSize) {
+    return new Options().batchSize(batchSize);
+  }
+  
+  /**
+   * @param compressionType The type of compression for the file. Currently ZLIB and
+   * GZIP are supported. Defaults to none.
+   */
+  public static Options compressionType(String compressionType) {
+    return new Options().compressionType(compressionType);
+  }
+  
+  /**
+   * A tensor of shape [batch_size].
+   */
+  public Output<TString> records() {
+    return records;
+  }
+  
+  @Override
+  public Output<TString> asOutput() {
+    return records;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "RecordInput";
+  
+  private Output<TString> records;
+  
+  private RecordInput(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    records = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RngSkip.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RngSkip.java
new file mode 100644
index 00000000000..f41cff35b04
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/RngSkip.java
@@ -0,0 +1,64 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.random;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt64;
+
+/**
+ * Advance the counter of a counter-based RNG.
+ * <p>
+ * The state of the RNG after
+ * `rng_skip(n)` will be the same as that after `stateful_uniform([n])`
+ * (or any other distribution). The actual increment added to the
+ * counter is an unspecified implementation detail.
+ */
+public final class RngSkip extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new RngSkip operation.
+   * 
+   * @param scope current scope
+   * @param resource The handle of the resource variable that stores the state of the RNG.
+   * @param algorithm The RNG algorithm.
+   * @param delta The amount of advancement.
+   * @return a new instance of RngSkip
+   */
+  @Endpoint(describeByClass = true)
+  public static RngSkip create(Scope scope, Operand<?> resource, Operand<TInt64> algorithm, Operand<TInt64> delta) {
+    OperationBuilder opBuilder = scope.env().opBuilder("RngSkip", scope.makeOpName("RngSkip"));
+    opBuilder.addInput(resource.asOutput());
+    opBuilder.addInput(algorithm.asOutput());
+    opBuilder.addInput(delta.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new RngSkip(opBuilder.build());
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "RngSkip";
+  
+  private RngSkip(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatefulRandomBinomial.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatefulRandomBinomial.java
new file mode 100644
index 00000000000..f0d68949af1
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatefulRandomBinomial.java
@@ -0,0 +1,99 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.random;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * @param <V> data type for {@code output()} output
+ */
+@Operator(group = "random")
+public final class StatefulRandomBinomial<V extends TNumber> extends RawOp implements Operand<V> {
+  
+  /**
+   * Factory method to create a class wrapping a new StatefulRandomBinomial operation.
+   * 
+   * @param scope current scope
+   * @param resource 
+   * @param algorithm 
+   * @param shape 
+   * @param counts 
+   * @param probs 
+   * @param dtype 
+   * @return a new instance of StatefulRandomBinomial
+   */
+  @Endpoint(describeByClass = true)
+  public static <V extends TNumber, T extends TNumber, U extends TNumber> StatefulRandomBinomial<V> create(Scope scope, Operand<?> resource, Operand<TInt64> algorithm, Operand<T> shape, Operand<U> counts, Operand<U> probs, Class<V> dtype) {
+    OperationBuilder opBuilder = scope.env().opBuilder("StatefulRandomBinomial", scope.makeOpName("StatefulRandomBinomial"));
+    opBuilder.addInput(resource.asOutput());
+    opBuilder.addInput(algorithm.asOutput());
+    opBuilder.addInput(shape.asOutput());
+    opBuilder.addInput(counts.asOutput());
+    opBuilder.addInput(probs.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("dtype", dtype);
+    return new StatefulRandomBinomial<V>(opBuilder.build());
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new StatefulRandomBinomial operation using default output types.
+   * 
+   * @param scope current scope
+   * @param resource 
+   * @param algorithm 
+   * @param shape 
+   * @param counts 
+   * @param probs 
+   * @return a new instance of StatefulRandomBinomial
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber, U extends TNumber> StatefulRandomBinomial<TInt64> create(Scope scope, Operand<?> resource, Operand<TInt64> algorithm, Operand<T> shape, Operand<U> counts, Operand<U> probs) {
+    return create(scope, resource, algorithm, shape, counts, probs, TInt64.class);
+  }
+  
+  /**
+   */
+  public Output<V> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<V> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "StatefulRandomBinomial";
+  
+  private Output<V> output;
+  
+  private StatefulRandomBinomial(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatefulStandardNormal.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatefulStandardNormal.java
new file mode 100644
index 00000000000..a01c0f740b8
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatefulStandardNormal.java
@@ -0,0 +1,99 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.random;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Outputs random values from a normal distribution.
+ * <p>
+ * The generated values will have mean 0 and standard deviation 1.
+ * 
+ * @param <U> data type for {@code output()} output
+ */
+@Operator(group = "random")
+public final class StatefulStandardNormal<U extends TType> extends RawOp implements Operand<U> {
+  
+  /**
+   * Factory method to create a class wrapping a new StatefulStandardNormal operation.
+   * 
+   * @param scope current scope
+   * @param resource The handle of the resource variable that stores the state of the RNG.
+   * @param algorithm The RNG algorithm.
+   * @param shape The shape of the output tensor.
+   * @param dtype The type of the output.
+   * @return a new instance of StatefulStandardNormal
+   */
+  @Endpoint(describeByClass = true)
+  public static <U extends TType, T extends TType> StatefulStandardNormal<U> create(Scope scope, Operand<?> resource, Operand<TInt64> algorithm, Operand<T> shape, Class<U> dtype) {
+    OperationBuilder opBuilder = scope.env().opBuilder("StatefulStandardNormalV2", scope.makeOpName("StatefulStandardNormal"));
+    opBuilder.addInput(resource.asOutput());
+    opBuilder.addInput(algorithm.asOutput());
+    opBuilder.addInput(shape.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("dtype", dtype);
+    return new StatefulStandardNormal<U>(opBuilder.build());
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new StatefulStandardNormal operation using default output types.
+   * 
+   * @param scope current scope
+   * @param resource The handle of the resource variable that stores the state of the RNG.
+   * @param algorithm The RNG algorithm.
+   * @param shape The shape of the output tensor.
+   * @return a new instance of StatefulStandardNormal
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> StatefulStandardNormal<TFloat32> create(Scope scope, Operand<?> resource, Operand<TInt64> algorithm, Operand<T> shape) {
+    return create(scope, resource, algorithm, shape, TFloat32.class);
+  }
+  
+  /**
+   * A tensor of the specified shape filled with random normal values.
+   */
+  public Output<U> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<U> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "StatefulStandardNormalV2";
+  
+  private Output<U> output;
+  
+  private StatefulStandardNormal(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatefulTruncatedNormal.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatefulTruncatedNormal.java
new file mode 100644
index 00000000000..461389eaec5
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatefulTruncatedNormal.java
@@ -0,0 +1,100 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.random;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Outputs random values from a truncated normal distribution.
+ * <p>
+ * The generated values follow a normal distribution with mean 0 and standard
+ * deviation 1, except that values whose magnitude is more than 2 standard
+ * deviations from the mean are dropped and re-picked.
+ * 
+ * @param <U> data type for {@code output()} output
+ */
+public final class StatefulTruncatedNormal<U extends TType> extends RawOp implements Operand<U> {
+  
+  /**
+   * Factory method to create a class wrapping a new StatefulTruncatedNormal operation.
+   * 
+   * @param scope current scope
+   * @param resource The handle of the resource variable that stores the state of the RNG.
+   * @param algorithm The RNG algorithm.
+   * @param shape The shape of the output tensor.
+   * @param dtype The type of the output.
+   * @return a new instance of StatefulTruncatedNormal
+   */
+  @Endpoint(describeByClass = true)
+  public static <U extends TType, T extends TType> StatefulTruncatedNormal<U> create(Scope scope, Operand<?> resource, Operand<TInt64> algorithm, Operand<T> shape, Class<U> dtype) {
+    OperationBuilder opBuilder = scope.env().opBuilder("StatefulTruncatedNormal", scope.makeOpName("StatefulTruncatedNormal"));
+    opBuilder.addInput(resource.asOutput());
+    opBuilder.addInput(algorithm.asOutput());
+    opBuilder.addInput(shape.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("dtype", dtype);
+    return new StatefulTruncatedNormal<U>(opBuilder.build());
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new StatefulTruncatedNormal operation using default output types.
+   * 
+   * @param scope current scope
+   * @param resource The handle of the resource variable that stores the state of the RNG.
+   * @param algorithm The RNG algorithm.
+   * @param shape The shape of the output tensor.
+   * @return a new instance of StatefulTruncatedNormal
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> StatefulTruncatedNormal<TFloat32> create(Scope scope, Operand<?> resource, Operand<TInt64> algorithm, Operand<T> shape) {
+    return create(scope, resource, algorithm, shape, TFloat32.class);
+  }
+  
+  /**
+   * Random values with specified shape.
+   */
+  public Output<U> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<U> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "StatefulTruncatedNormal";
+  
+  private Output<U> output;
+  
+  private StatefulTruncatedNormal(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatefulUniform.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatefulUniform.java
new file mode 100644
index 00000000000..89fa8bdd32d
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatefulUniform.java
@@ -0,0 +1,99 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.random;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Outputs random values from a uniform distribution.
+ * <p>
+ * The generated values follow a uniform distribution in the range `[0, 1)`. The
+ * lower bound 0 is included in the range, while the upper bound 1 is excluded.
+ * 
+ * @param <U> data type for {@code output()} output
+ */
+public final class StatefulUniform<U extends TType> extends RawOp implements Operand<U> {
+  
+  /**
+   * Factory method to create a class wrapping a new StatefulUniform operation.
+   * 
+   * @param scope current scope
+   * @param resource The handle of the resource variable that stores the state of the RNG.
+   * @param algorithm The RNG algorithm.
+   * @param shape The shape of the output tensor.
+   * @param dtype The type of the output.
+   * @return a new instance of StatefulUniform
+   */
+  @Endpoint(describeByClass = true)
+  public static <U extends TType, T extends TType> StatefulUniform<U> create(Scope scope, Operand<?> resource, Operand<TInt64> algorithm, Operand<T> shape, Class<U> dtype) {
+    OperationBuilder opBuilder = scope.env().opBuilder("StatefulUniform", scope.makeOpName("StatefulUniform"));
+    opBuilder.addInput(resource.asOutput());
+    opBuilder.addInput(algorithm.asOutput());
+    opBuilder.addInput(shape.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("dtype", dtype);
+    return new StatefulUniform<U>(opBuilder.build());
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new StatefulUniform operation using default output types.
+   * 
+   * @param scope current scope
+   * @param resource The handle of the resource variable that stores the state of the RNG.
+   * @param algorithm The RNG algorithm.
+   * @param shape The shape of the output tensor.
+   * @return a new instance of StatefulUniform
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> StatefulUniform<TFloat32> create(Scope scope, Operand<?> resource, Operand<TInt64> algorithm, Operand<T> shape) {
+    return create(scope, resource, algorithm, shape, TFloat32.class);
+  }
+  
+  /**
+   * Random values with specified shape.
+   */
+  public Output<U> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<U> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "StatefulUniform";
+  
+  private Output<U> output;
+  
+  private StatefulUniform(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatefulUniformFullInt.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatefulUniformFullInt.java
new file mode 100644
index 00000000000..457dcc0b1a8
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatefulUniformFullInt.java
@@ -0,0 +1,83 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.random;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Outputs random integers from a uniform distribution.
+ * <p>
+ * The generated values are uniform integers covering the whole range of `dtype`.
+ * 
+ * @param <U> data type for {@code output()} output
+ */
+public final class StatefulUniformFullInt<U extends TType> extends RawOp implements Operand<U> {
+  
+  /**
+   * Factory method to create a class wrapping a new StatefulUniformFullInt operation.
+   * 
+   * @param scope current scope
+   * @param resource The handle of the resource variable that stores the state of the RNG.
+   * @param algorithm The RNG algorithm.
+   * @param shape The shape of the output tensor.
+   * @param dtype The type of the output.
+   * @return a new instance of StatefulUniformFullInt
+   */
+  @Endpoint(describeByClass = true)
+  public static <U extends TType, T extends TType> StatefulUniformFullInt<U> create(Scope scope, Operand<?> resource, Operand<TInt64> algorithm, Operand<T> shape, Class<U> dtype) {
+    OperationBuilder opBuilder = scope.env().opBuilder("StatefulUniformFullInt", scope.makeOpName("StatefulUniformFullInt"));
+    opBuilder.addInput(resource.asOutput());
+    opBuilder.addInput(algorithm.asOutput());
+    opBuilder.addInput(shape.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("dtype", dtype);
+    return new StatefulUniformFullInt<U>(opBuilder.build());
+  }
+  
+  /**
+   * Random values with specified shape.
+   */
+  public Output<U> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<U> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "StatefulUniformFullInt";
+  
+  private Output<U> output;
+  
+  private StatefulUniformFullInt(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatefulUniformInt.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatefulUniformInt.java
new file mode 100644
index 00000000000..74337edc44d
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatefulUniformInt.java
@@ -0,0 +1,91 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.random;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Outputs random integers from a uniform distribution.
+ * <p>
+ * The generated values are uniform integers in the range `[minval, maxval)`.
+ * The lower bound `minval` is included in the range, while the upper bound
+ * `maxval` is excluded.
+ * <p>
+ * The random integers are slightly biased unless `maxval - minval` is an exact
+ * power of two.  The bias is small for values of `maxval - minval` significantly
+ * smaller than the range of the output (either `2^32` or `2^64`).
+ * 
+ * @param <U> data type for {@code output()} output
+ */
+public final class StatefulUniformInt<U extends TType> extends RawOp implements Operand<U> {
+  
+  /**
+   * Factory method to create a class wrapping a new StatefulUniformInt operation.
+   * 
+   * @param scope current scope
+   * @param resource The handle of the resource variable that stores the state of the RNG.
+   * @param algorithm The RNG algorithm.
+   * @param shape The shape of the output tensor.
+   * @param minval Minimum value (inclusive, scalar).
+   * @param maxval Maximum value (exclusive, scalar).
+   * @return a new instance of StatefulUniformInt
+   */
+  @Endpoint(describeByClass = true)
+  public static <U extends TType, T extends TType> StatefulUniformInt<U> create(Scope scope, Operand<?> resource, Operand<TInt64> algorithm, Operand<T> shape, Operand<U> minval, Operand<U> maxval) {
+    OperationBuilder opBuilder = scope.env().opBuilder("StatefulUniformInt", scope.makeOpName("StatefulUniformInt"));
+    opBuilder.addInput(resource.asOutput());
+    opBuilder.addInput(algorithm.asOutput());
+    opBuilder.addInput(shape.asOutput());
+    opBuilder.addInput(minval.asOutput());
+    opBuilder.addInput(maxval.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new StatefulUniformInt<U>(opBuilder.build());
+  }
+  
+  /**
+   * Random values with specified shape.
+   */
+  public Output<U> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<U> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "StatefulUniformInt";
+  
+  private Output<U> output;
+  
+  private StatefulUniformInt(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessMultinomial.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessMultinomial.java
new file mode 100644
index 00000000000..3bc5ab365a9
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessMultinomial.java
@@ -0,0 +1,100 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.random;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Draws samples from a multinomial distribution.
+ * 
+ * @param <V> data type for {@code output()} output
+ */
+@Operator(group = "random")
+public final class StatelessMultinomial<V extends TNumber> extends RawOp implements Operand<V> {
+  
+  /**
+   * Factory method to create a class wrapping a new StatelessMultinomial operation.
+   * 
+   * @param scope current scope
+   * @param logits 2-D Tensor with shape `[batch_size, num_classes]`.  Each slice `[i, :]`
+   * represents the unnormalized log probabilities for all classes.
+   * @param numSamples 0-D.  Number of independent samples to draw for each row slice.
+   * @param seed 2 seeds (shape [2]).
+   * @param outputDtype 
+   * @return a new instance of StatelessMultinomial
+   */
+  @Endpoint(describeByClass = true)
+  public static <V extends TNumber, T extends TNumber, U extends TNumber> StatelessMultinomial<V> create(Scope scope, Operand<T> logits, Operand<TInt32> numSamples, Operand<U> seed, Class<V> outputDtype) {
+    OperationBuilder opBuilder = scope.env().opBuilder("StatelessMultinomial", scope.makeOpName("StatelessMultinomial"));
+    opBuilder.addInput(logits.asOutput());
+    opBuilder.addInput(numSamples.asOutput());
+    opBuilder.addInput(seed.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("output_dtype", outputDtype);
+    return new StatelessMultinomial<V>(opBuilder.build());
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new StatelessMultinomial operation using default output types.
+   * 
+   * @param scope current scope
+   * @param logits 2-D Tensor with shape `[batch_size, num_classes]`.  Each slice `[i, :]`
+   * represents the unnormalized log probabilities for all classes.
+   * @param numSamples 0-D.  Number of independent samples to draw for each row slice.
+   * @param seed 2 seeds (shape [2]).
+   * @return a new instance of StatelessMultinomial
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber, U extends TNumber> StatelessMultinomial<TInt64> create(Scope scope, Operand<T> logits, Operand<TInt32> numSamples, Operand<U> seed) {
+    return create(scope, logits, numSamples, seed, TInt64.class);
+  }
+  
+  /**
+   * 2-D Tensor with shape `[batch_size, num_samples]`.  Each slice `[i, :]`
+   * contains the drawn class labels with range `[0, num_classes)`.
+   */
+  public Output<V> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<V> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "StatelessMultinomial";
+  
+  private Output<V> output;
+  
+  private StatelessMultinomial(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessParameterizedTruncatedNormal.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessParameterizedTruncatedNormal.java
new file mode 100644
index 00000000000..f422a582a0f
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessParameterizedTruncatedNormal.java
@@ -0,0 +1,84 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.random;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * @param <V> data type for {@code output()} output
+ */
+public final class StatelessParameterizedTruncatedNormal<V extends TNumber> extends RawOp implements Operand<V> {
+  
+  /**
+   * Factory method to create a class wrapping a new StatelessParameterizedTruncatedNormal operation.
+   * 
+   * @param scope current scope
+   * @param shape The shape of the output tensor.
+   * @param seed 2 seeds (shape [2]).
+   * @param means The mean parameter of each batch.
+   * @param stddevs The standard deviation parameter of each batch. Must be greater than 0.
+   * @param minvals The minimum cutoff. May be -infinity.
+   * @param maxvals The maximum cutoff. May be +infinity, and must be more than the minval
+   * for each batch.
+   * @return a new instance of StatelessParameterizedTruncatedNormal
+   */
+  @Endpoint(describeByClass = true)
+  public static <V extends TNumber, T extends TNumber, U extends TNumber> StatelessParameterizedTruncatedNormal<V> create(Scope scope, Operand<T> shape, Operand<U> seed, Operand<V> means, Operand<V> stddevs, Operand<V> minvals, Operand<V> maxvals) {
+    OperationBuilder opBuilder = scope.env().opBuilder("StatelessParameterizedTruncatedNormal", scope.makeOpName("StatelessParameterizedTruncatedNormal"));
+    opBuilder.addInput(shape.asOutput());
+    opBuilder.addInput(seed.asOutput());
+    opBuilder.addInput(means.asOutput());
+    opBuilder.addInput(stddevs.asOutput());
+    opBuilder.addInput(minvals.asOutput());
+    opBuilder.addInput(maxvals.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new StatelessParameterizedTruncatedNormal<V>(opBuilder.build());
+  }
+  
+  /**
+   * The outputs are truncated normal samples and are a deterministic function of
+   * `shape`, `seed`, `minvals`, `maxvals`, `means` and `stddevs`.
+   */
+  public Output<V> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<V> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "StatelessParameterizedTruncatedNormal";
+  
+  private Output<V> output;
+  
+  private StatelessParameterizedTruncatedNormal(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessRandomBinomial.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessRandomBinomial.java
new file mode 100644
index 00000000000..2c2fe32cf12
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessRandomBinomial.java
@@ -0,0 +1,106 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.random;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Outputs deterministic pseudorandom random numbers from a binomial distribution.
+ * <p>
+ * Outputs random values from a binomial distribution.
+ * <p>
+ * The outputs are a deterministic function of `shape`, `seed`, `counts`, and `probs`.
+ * 
+ * @param <W> data type for {@code output()} output
+ */
+public final class StatelessRandomBinomial<W extends TNumber> extends RawOp implements Operand<W> {
+  
+  /**
+   * Factory method to create a class wrapping a new StatelessRandomBinomial operation.
+   * 
+   * @param scope current scope
+   * @param shape The shape of the output tensor.
+   * @param seed 2 seeds (shape [2]).
+   * @param counts The counts of the binomial distribution. Must be broadcastable with `probs`,
+   * and broadcastable with the rightmost dimensions of `shape`.
+   * @param probs The probability of success for the binomial distribution. Must be broadcastable
+   * with `counts` and broadcastable with the rightmost dimensions of `shape`.
+   * @param dtype The type of the output.
+   * @return a new instance of StatelessRandomBinomial
+   */
+  @Endpoint(describeByClass = true)
+  public static <W extends TNumber, T extends TNumber, U extends TNumber, V extends TNumber> StatelessRandomBinomial<W> create(Scope scope, Operand<T> shape, Operand<U> seed, Operand<V> counts, Operand<V> probs, Class<W> dtype) {
+    OperationBuilder opBuilder = scope.env().opBuilder("StatelessRandomBinomial", scope.makeOpName("StatelessRandomBinomial"));
+    opBuilder.addInput(shape.asOutput());
+    opBuilder.addInput(seed.asOutput());
+    opBuilder.addInput(counts.asOutput());
+    opBuilder.addInput(probs.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("dtype", dtype);
+    return new StatelessRandomBinomial<W>(opBuilder.build());
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new StatelessRandomBinomial operation using default output types.
+   * 
+   * @param scope current scope
+   * @param shape The shape of the output tensor.
+   * @param seed 2 seeds (shape [2]).
+   * @param counts The counts of the binomial distribution. Must be broadcastable with `probs`,
+   * and broadcastable with the rightmost dimensions of `shape`.
+   * @param probs The probability of success for the binomial distribution. Must be broadcastable
+   * with `counts` and broadcastable with the rightmost dimensions of `shape`.
+   * @return a new instance of StatelessRandomBinomial
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber, U extends TNumber, V extends TNumber> StatelessRandomBinomial<TInt64> create(Scope scope, Operand<T> shape, Operand<U> seed, Operand<V> counts, Operand<V> probs) {
+    return create(scope, shape, seed, counts, probs, TInt64.class);
+  }
+  
+  /**
+   * Random values with specified shape.
+   */
+  public Output<W> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<W> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "StatelessRandomBinomial";
+  
+  private Output<W> output;
+  
+  private StatelessRandomBinomial(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessRandomGamma.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessRandomGamma.java
new file mode 100644
index 00000000000..7c8aabfa082
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessRandomGamma.java
@@ -0,0 +1,83 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.random;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Outputs deterministic pseudorandom random numbers from a gamma distribution.
+ * <p>
+ * Outputs random values from a gamma distribution.
+ * <p>
+ * The outputs are a deterministic function of `shape`, `seed`, and `alpha`.
+ * 
+ * @param <V> data type for {@code output()} output
+ */
+public final class StatelessRandomGamma<V extends TNumber> extends RawOp implements Operand<V> {
+  
+  /**
+   * Factory method to create a class wrapping a new StatelessRandomGamma operation.
+   * 
+   * @param scope current scope
+   * @param shape The shape of the output tensor.
+   * @param seed 2 seeds (shape [2]).
+   * @param alpha The concentration of the gamma distribution. Shape must match the rightmost
+   * dimensions of `shape`.
+   * @return a new instance of StatelessRandomGamma
+   */
+  @Endpoint(describeByClass = true)
+  public static <V extends TNumber, T extends TNumber, U extends TNumber> StatelessRandomGamma<V> create(Scope scope, Operand<T> shape, Operand<U> seed, Operand<V> alpha) {
+    OperationBuilder opBuilder = scope.env().opBuilder("StatelessRandomGammaV2", scope.makeOpName("StatelessRandomGamma"));
+    opBuilder.addInput(shape.asOutput());
+    opBuilder.addInput(seed.asOutput());
+    opBuilder.addInput(alpha.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new StatelessRandomGamma<V>(opBuilder.build());
+  }
+  
+  /**
+   * Random values with specified shape.
+   */
+  public Output<V> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<V> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "StatelessRandomGammaV2";
+  
+  private Output<V> output;
+  
+  private StatelessRandomGamma(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessRandomNormal.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessRandomNormal.java
new file mode 100644
index 00000000000..2b20afb21b1
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessRandomNormal.java
@@ -0,0 +1,97 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.random;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Outputs deterministic pseudorandom values from a normal distribution.
+ * <p>
+ * The generated values will have mean 0 and standard deviation 1.
+ * <p>
+ * The outputs are a deterministic function of `shape` and `seed`.
+ * 
+ * @param <V> data type for {@code output()} output
+ */
+@Operator(group = "random")
+public final class StatelessRandomNormal<V extends TNumber> extends RawOp implements Operand<V> {
+  
+  /**
+   * Factory method to create a class wrapping a new StatelessRandomNormal operation.
+   * 
+   * @param scope current scope
+   * @param shape The shape of the output tensor.
+   * @param seed 2 seeds (shape [2]).
+   * @param dtype The type of the output.
+   * @return a new instance of StatelessRandomNormal
+   */
+  @Endpoint(describeByClass = true)
+  public static <V extends TNumber, T extends TNumber, U extends TNumber> StatelessRandomNormal<V> create(Scope scope, Operand<T> shape, Operand<U> seed, Class<V> dtype) {
+    OperationBuilder opBuilder = scope.env().opBuilder("StatelessRandomNormal", scope.makeOpName("StatelessRandomNormal"));
+    opBuilder.addInput(shape.asOutput());
+    opBuilder.addInput(seed.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("dtype", dtype);
+    return new StatelessRandomNormal<V>(opBuilder.build());
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new StatelessRandomNormal operation using default output types.
+   * 
+   * @param scope current scope
+   * @param shape The shape of the output tensor.
+   * @param seed 2 seeds (shape [2]).
+   * @return a new instance of StatelessRandomNormal
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber, U extends TNumber> StatelessRandomNormal<TFloat32> create(Scope scope, Operand<T> shape, Operand<U> seed) {
+    return create(scope, shape, seed, TFloat32.class);
+  }
+  
+  /**
+   * Random values with specified shape.
+   */
+  public Output<V> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<V> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "StatelessRandomNormal";
+  
+  private Output<V> output;
+  
+  private StatelessRandomNormal(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessRandomPoisson.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessRandomPoisson.java
new file mode 100644
index 00000000000..cde1f6609b9
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessRandomPoisson.java
@@ -0,0 +1,85 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.random;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Outputs deterministic pseudorandom random numbers from a Poisson distribution.
+ * <p>
+ * Outputs random values from a Poisson distribution.
+ * <p>
+ * The outputs are a deterministic function of `shape`, `seed`, and `lam`.
+ * 
+ * @param <W> data type for {@code output()} output
+ */
+public final class StatelessRandomPoisson<W extends TNumber> extends RawOp implements Operand<W> {
+  
+  /**
+   * Factory method to create a class wrapping a new StatelessRandomPoisson operation.
+   * 
+   * @param scope current scope
+   * @param shape The shape of the output tensor.
+   * @param seed 2 seeds (shape [2]).
+   * @param lam The rate of the Poisson distribution. Shape must match the rightmost dimensions
+   * of `shape`.
+   * @param dtype The type of the output.
+   * @return a new instance of StatelessRandomPoisson
+   */
+  @Endpoint(describeByClass = true)
+  public static <W extends TNumber, T extends TNumber, U extends TNumber, V extends TNumber> StatelessRandomPoisson<W> create(Scope scope, Operand<T> shape, Operand<U> seed, Operand<V> lam, Class<W> dtype) {
+    OperationBuilder opBuilder = scope.env().opBuilder("StatelessRandomPoisson", scope.makeOpName("StatelessRandomPoisson"));
+    opBuilder.addInput(shape.asOutput());
+    opBuilder.addInput(seed.asOutput());
+    opBuilder.addInput(lam.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("dtype", dtype);
+    return new StatelessRandomPoisson<W>(opBuilder.build());
+  }
+  
+  /**
+   * Random values with specified shape.
+   */
+  public Output<W> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<W> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "StatelessRandomPoisson";
+  
+  private Output<W> output;
+  
+  private StatelessRandomPoisson(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessRandomUniform.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessRandomUniform.java
new file mode 100644
index 00000000000..c94282163f6
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessRandomUniform.java
@@ -0,0 +1,98 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.random;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Outputs deterministic pseudorandom random values from a uniform distribution.
+ * <p>
+ * The generated values follow a uniform distribution in the range `[0, 1)`. The
+ * lower bound 0 is included in the range, while the upper bound 1 is excluded.
+ * <p>
+ * The outputs are a deterministic function of `shape` and `seed`.
+ * 
+ * @param <V> data type for {@code output()} output
+ */
+@Operator(group = "random")
+public final class StatelessRandomUniform<V extends TNumber> extends RawOp implements Operand<V> {
+  
+  /**
+   * Factory method to create a class wrapping a new StatelessRandomUniform operation.
+   * 
+   * @param scope current scope
+   * @param shape The shape of the output tensor.
+   * @param seed 2 seeds (shape [2]).
+   * @param dtype The type of the output.
+   * @return a new instance of StatelessRandomUniform
+   */
+  @Endpoint(describeByClass = true)
+  public static <V extends TNumber, T extends TNumber, U extends TNumber> StatelessRandomUniform<V> create(Scope scope, Operand<T> shape, Operand<U> seed, Class<V> dtype) {
+    OperationBuilder opBuilder = scope.env().opBuilder("StatelessRandomUniform", scope.makeOpName("StatelessRandomUniform"));
+    opBuilder.addInput(shape.asOutput());
+    opBuilder.addInput(seed.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("dtype", dtype);
+    return new StatelessRandomUniform<V>(opBuilder.build());
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new StatelessRandomUniform operation using default output types.
+   * 
+   * @param scope current scope
+   * @param shape The shape of the output tensor.
+   * @param seed 2 seeds (shape [2]).
+   * @return a new instance of StatelessRandomUniform
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber, U extends TNumber> StatelessRandomUniform<TFloat32> create(Scope scope, Operand<T> shape, Operand<U> seed) {
+    return create(scope, shape, seed, TFloat32.class);
+  }
+  
+  /**
+   * Random values with specified shape.
+   */
+  public Output<V> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<V> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "StatelessRandomUniform";
+  
+  private Output<V> output;
+  
+  private StatelessRandomUniform(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessRandomUniformFullInt.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessRandomUniformFullInt.java
new file mode 100644
index 00000000000..b774c165ec0
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessRandomUniformFullInt.java
@@ -0,0 +1,82 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.random;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Outputs deterministic pseudorandom random integers from a uniform distribution.
+ * <p>
+ * The generated values are uniform integers covering the whole range of `dtype`.
+ * <p>
+ * The outputs are a deterministic function of `shape` and `seed`.
+ * 
+ * @param <V> data type for {@code output()} output
+ */
+public final class StatelessRandomUniformFullInt<V extends TNumber> extends RawOp implements Operand<V> {
+  
+  /**
+   * Factory method to create a class wrapping a new StatelessRandomUniformFullInt operation.
+   * 
+   * @param scope current scope
+   * @param shape The shape of the output tensor.
+   * @param seed 2 seeds (shape [2]).
+   * @param dtype The type of the output.
+   * @return a new instance of StatelessRandomUniformFullInt
+   */
+  @Endpoint(describeByClass = true)
+  public static <V extends TNumber, T extends TNumber, U extends TNumber> StatelessRandomUniformFullInt<V> create(Scope scope, Operand<T> shape, Operand<U> seed, Class<V> dtype) {
+    OperationBuilder opBuilder = scope.env().opBuilder("StatelessRandomUniformFullInt", scope.makeOpName("StatelessRandomUniformFullInt"));
+    opBuilder.addInput(shape.asOutput());
+    opBuilder.addInput(seed.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("dtype", dtype);
+    return new StatelessRandomUniformFullInt<V>(opBuilder.build());
+  }
+  
+  /**
+   * Random values with specified shape.
+   */
+  public Output<V> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<V> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "StatelessRandomUniformFullInt";
+  
+  private Output<V> output;
+  
+  private StatelessRandomUniformFullInt(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessRandomUniformInt.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessRandomUniformInt.java
new file mode 100644
index 00000000000..e8bbba18ab5
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessRandomUniformInt.java
@@ -0,0 +1,84 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.random;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Outputs deterministic pseudorandom random integers from a uniform distribution.
+ * <p>
+ * The generated values follow a uniform distribution in the range `[minval, maxval)`.
+ * <p>
+ * The outputs are a deterministic function of `shape`, `seed`, `minval`, and `maxval`.
+ * 
+ * @param <V> data type for {@code output()} output
+ */
+public final class StatelessRandomUniformInt<V extends TNumber> extends RawOp implements Operand<V> {
+  
+  /**
+   * Factory method to create a class wrapping a new StatelessRandomUniformInt operation.
+   * 
+   * @param scope current scope
+   * @param shape The shape of the output tensor.
+   * @param seed 2 seeds (shape [2]).
+   * @param minval Minimum value (inclusive, scalar).
+   * @param maxval Maximum value (exclusive, scalar).
+   * @return a new instance of StatelessRandomUniformInt
+   */
+  @Endpoint(describeByClass = true)
+  public static <V extends TNumber, T extends TNumber, U extends TNumber> StatelessRandomUniformInt<V> create(Scope scope, Operand<T> shape, Operand<U> seed, Operand<V> minval, Operand<V> maxval) {
+    OperationBuilder opBuilder = scope.env().opBuilder("StatelessRandomUniformInt", scope.makeOpName("StatelessRandomUniformInt"));
+    opBuilder.addInput(shape.asOutput());
+    opBuilder.addInput(seed.asOutput());
+    opBuilder.addInput(minval.asOutput());
+    opBuilder.addInput(maxval.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new StatelessRandomUniformInt<V>(opBuilder.build());
+  }
+  
+  /**
+   * Random values with specified shape.
+   */
+  public Output<V> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<V> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "StatelessRandomUniformInt";
+  
+  private Output<V> output;
+  
+  private StatelessRandomUniformInt(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessTruncatedNormal.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessTruncatedNormal.java
new file mode 100644
index 00000000000..cf260db14a9
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/StatelessTruncatedNormal.java
@@ -0,0 +1,99 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.random;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Outputs deterministic pseudorandom values from a truncated normal distribution.
+ * <p>
+ * The generated values follow a normal distribution with mean 0 and standard
+ * deviation 1, except that values whose magnitude is more than 2 standard
+ * deviations from the mean are dropped and re-picked.
+ * <p>
+ * The outputs are a deterministic function of `shape` and `seed`.
+ * 
+ * @param <V> data type for {@code output()} output
+ */
+@Operator(group = "random")
+public final class StatelessTruncatedNormal<V extends TNumber> extends RawOp implements Operand<V> {
+  
+  /**
+   * Factory method to create a class wrapping a new StatelessTruncatedNormal operation.
+   * 
+   * @param scope current scope
+   * @param shape The shape of the output tensor.
+   * @param seed 2 seeds (shape [2]).
+   * @param dtype The type of the output.
+   * @return a new instance of StatelessTruncatedNormal
+   */
+  @Endpoint(describeByClass = true)
+  public static <V extends TNumber, T extends TNumber, U extends TNumber> StatelessTruncatedNormal<V> create(Scope scope, Operand<T> shape, Operand<U> seed, Class<V> dtype) {
+    OperationBuilder opBuilder = scope.env().opBuilder("StatelessTruncatedNormal", scope.makeOpName("StatelessTruncatedNormal"));
+    opBuilder.addInput(shape.asOutput());
+    opBuilder.addInput(seed.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("dtype", dtype);
+    return new StatelessTruncatedNormal<V>(opBuilder.build());
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new StatelessTruncatedNormal operation using default output types.
+   * 
+   * @param scope current scope
+   * @param shape The shape of the output tensor.
+   * @param seed 2 seeds (shape [2]).
+   * @return a new instance of StatelessTruncatedNormal
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber, U extends TNumber> StatelessTruncatedNormal<TFloat32> create(Scope scope, Operand<T> shape, Operand<U> seed) {
+    return create(scope, shape, seed, TFloat32.class);
+  }
+  
+  /**
+   * Random values with specified shape.
+   */
+  public Output<V> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<V> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "StatelessTruncatedNormal";
+  
+  private Output<V> output;
+  
+  private StatelessTruncatedNormal(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/TruncatedNormal.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/TruncatedNormal.java
new file mode 100644
index 00000000000..dc7aee67856
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/TruncatedNormal.java
@@ -0,0 +1,139 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.random;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Outputs random values from a truncated normal distribution.
+ * <p>
+ * The generated values follow a normal distribution with mean 0 and standard
+ * deviation 1, except that values whose magnitude is more than 2 standard
+ * deviations from the mean are dropped and re-picked.
+ * 
+ * @param <U> data type for {@code output()} output
+ */
+@Operator(group = "random")
+public final class TruncatedNormal<U extends TNumber> extends RawOp implements Operand<U> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.random.TruncatedNormal}
+   */
+  public static class Options {
+    
+    /**
+     * @param seed If either `seed` or `seed2` are set to be non-zero, the random number
+     * generator is seeded by the given seed.  Otherwise, it is seeded by a
+     * random seed.
+     */
+    public Options seed(Long seed) {
+      this.seed = seed;
+      return this;
+    }
+    
+    /**
+     * @param seed2 A second seed to avoid seed collision.
+     */
+    public Options seed2(Long seed2) {
+      this.seed2 = seed2;
+      return this;
+    }
+    
+    private Long seed;
+    private Long seed2;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new TruncatedNormal operation.
+   * 
+   * @param scope current scope
+   * @param shape The shape of the output tensor.
+   * @param dtype The type of the output.
+   * @param options carries optional attributes values
+   * @return a new instance of TruncatedNormal
+   */
+  @Endpoint(describeByClass = true)
+  public static <U extends TNumber, T extends TNumber> TruncatedNormal<U> create(Scope scope, Operand<T> shape, Class<U> dtype, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("TruncatedNormal", scope.makeOpName("TruncatedNormal"));
+    opBuilder.addInput(shape.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("dtype", dtype);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.seed != null) {
+          opBuilder.setAttr("seed", opts.seed);
+        }
+        if (opts.seed2 != null) {
+          opBuilder.setAttr("seed2", opts.seed2);
+        }
+      }
+    }
+    return new TruncatedNormal<U>(opBuilder.build());
+  }
+  
+  /**
+   * @param seed If either `seed` or `seed2` are set to be non-zero, the random number
+   * generator is seeded by the given seed.  Otherwise, it is seeded by a
+   * random seed.
+   */
+  public static Options seed(Long seed) {
+    return new Options().seed(seed);
+  }
+  
+  /**
+   * @param seed2 A second seed to avoid seed collision.
+   */
+  public static Options seed2(Long seed2) {
+    return new Options().seed2(seed2);
+  }
+  
+  /**
+   * A tensor of the specified shape filled with random truncated normal
+   * values.
+   */
+  public Output<U> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<U> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "TruncatedNormal";
+  
+  private Output<U> output;
+  
+  private TruncatedNormal(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/UniformCandidateSampler.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/UniformCandidateSampler.java
new file mode 100644
index 00000000000..9e1f2b1abbd
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/UniformCandidateSampler.java
@@ -0,0 +1,171 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.random;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.TInt64;
+
+/**
+ * Generates labels for candidate sampling with a uniform distribution.
+ * <p>
+ * See explanations of candidate sampling and the data formats at
+ * go/candidate-sampling.
+ * <p>
+ * For each batch, this op picks a single set of sampled candidate labels.
+ * <p>
+ * The advantages of sampling candidates per-batch are simplicity and the
+ * possibility of efficient dense matrix multiplication. The disadvantage is that
+ * the sampled candidates must be chosen independently of the context and of the
+ * true labels.
+ */
+@Operator(group = "random")
+public final class UniformCandidateSampler extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.random.UniformCandidateSampler}
+   */
+  public static class Options {
+    
+    /**
+     * @param seed If either seed or seed2 are set to be non-zero, the random number
+     * generator is seeded by the given seed.  Otherwise, it is seeded by a
+     * random seed.
+     */
+    public Options seed(Long seed) {
+      this.seed = seed;
+      return this;
+    }
+    
+    /**
+     * @param seed2 An second seed to avoid seed collision.
+     */
+    public Options seed2(Long seed2) {
+      this.seed2 = seed2;
+      return this;
+    }
+    
+    private Long seed;
+    private Long seed2;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new UniformCandidateSampler operation.
+   * 
+   * @param scope current scope
+   * @param trueClasses A batch_size * num_true matrix, in which each row contains the
+   * IDs of the num_true target_classes in the corresponding original label.
+   * @param numTrue Number of true labels per context.
+   * @param numSampled Number of candidates to randomly sample.
+   * @param unique If unique is true, we sample with rejection, so that all sampled
+   * candidates in a batch are unique. This requires some approximation to
+   * estimate the post-rejection sampling probabilities.
+   * @param rangeMax The sampler will sample integers from the interval [0, range_max).
+   * @param options carries optional attributes values
+   * @return a new instance of UniformCandidateSampler
+   */
+  @Endpoint(describeByClass = true)
+  public static UniformCandidateSampler create(Scope scope, Operand<TInt64> trueClasses, Long numTrue, Long numSampled, Boolean unique, Long rangeMax, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("UniformCandidateSampler", scope.makeOpName("UniformCandidateSampler"));
+    opBuilder.addInput(trueClasses.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("num_true", numTrue);
+    opBuilder.setAttr("num_sampled", numSampled);
+    opBuilder.setAttr("unique", unique);
+    opBuilder.setAttr("range_max", rangeMax);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.seed != null) {
+          opBuilder.setAttr("seed", opts.seed);
+        }
+        if (opts.seed2 != null) {
+          opBuilder.setAttr("seed2", opts.seed2);
+        }
+      }
+    }
+    return new UniformCandidateSampler(opBuilder.build());
+  }
+  
+  /**
+   * @param seed If either seed or seed2 are set to be non-zero, the random number
+   * generator is seeded by the given seed.  Otherwise, it is seeded by a
+   * random seed.
+   */
+  public static Options seed(Long seed) {
+    return new Options().seed(seed);
+  }
+  
+  /**
+   * @param seed2 An second seed to avoid seed collision.
+   */
+  public static Options seed2(Long seed2) {
+    return new Options().seed2(seed2);
+  }
+  
+  /**
+   * A vector of length num_sampled, in which each element is
+   * the ID of a sampled candidate.
+   */
+  public Output<TInt64> sampledCandidates() {
+    return sampledCandidates;
+  }
+  
+  /**
+   * A batch_size * num_true matrix, representing
+   * the number of times each candidate is expected to occur in a batch
+   * of sampled candidates. If unique=true, then this is a probability.
+   */
+  public Output<TFloat32> trueExpectedCount() {
+    return trueExpectedCount;
+  }
+  
+  /**
+   * A vector of length num_sampled, for each sampled
+   * candidate representing the number of times the candidate is expected
+   * to occur in a batch of sampled candidates.  If unique=true, then this is a
+   * probability.
+   */
+  public Output<TFloat32> sampledExpectedCount() {
+    return sampledExpectedCount;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "UniformCandidateSampler";
+  
+  private Output<TInt64> sampledCandidates;
+  private Output<TFloat32> trueExpectedCount;
+  private Output<TFloat32> sampledExpectedCount;
+  
+  private UniformCandidateSampler(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    sampledCandidates = operation.output(outputIdx++);
+    trueExpectedCount = operation.output(outputIdx++);
+    sampledExpectedCount = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/experimental/DummySeedGenerator.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/experimental/DummySeedGenerator.java
new file mode 100644
index 00000000000..dd537fa2d68
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/random/experimental/DummySeedGenerator.java
@@ -0,0 +1,69 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.random.experimental;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ */
+public final class DummySeedGenerator extends RawOp implements Operand<TType> {
+  
+  /**
+   * Factory method to create a class wrapping a new DummySeedGenerator operation.
+   * 
+   * @param scope current scope
+   * @return a new instance of DummySeedGenerator
+   */
+  @Endpoint(describeByClass = true)
+  public static DummySeedGenerator create(Scope scope) {
+    OperationBuilder opBuilder = scope.env().opBuilder("DummySeedGenerator", scope.makeOpName("DummySeedGenerator"));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new DummySeedGenerator(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<?> handle() {
+    return handle;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) handle;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "DummySeedGenerator";
+  
+  private Output<?> handle;
+  
+  private DummySeedGenerator(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    handle = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/BatchFft.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/BatchFft.java
new file mode 100644
index 00000000000..9eaacf4b26a
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/BatchFft.java
@@ -0,0 +1,72 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.signal;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ */
+@Operator(group = "signal")
+public final class BatchFft extends RawOp implements Operand<TType> {
+  
+  /**
+   * Factory method to create a class wrapping a new BatchFft operation.
+   * 
+   * @param scope current scope
+   * @param input 
+   * @return a new instance of BatchFft
+   */
+  @Endpoint(describeByClass = true)
+  public static BatchFft create(Scope scope, Operand<?> input) {
+    OperationBuilder opBuilder = scope.env().opBuilder("BatchFFT", scope.makeOpName("BatchFft"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new BatchFft(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<?> output() {
+    return output;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "BatchFFT";
+  
+  private Output<?> output;
+  
+  private BatchFft(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/BatchFft2d.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/BatchFft2d.java
new file mode 100644
index 00000000000..551d7c3f139
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/BatchFft2d.java
@@ -0,0 +1,72 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.signal;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ */
+@Operator(group = "signal")
+public final class BatchFft2d extends RawOp implements Operand<TType> {
+  
+  /**
+   * Factory method to create a class wrapping a new BatchFft2d operation.
+   * 
+   * @param scope current scope
+   * @param input 
+   * @return a new instance of BatchFft2d
+   */
+  @Endpoint(describeByClass = true)
+  public static BatchFft2d create(Scope scope, Operand<?> input) {
+    OperationBuilder opBuilder = scope.env().opBuilder("BatchFFT2D", scope.makeOpName("BatchFft2d"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new BatchFft2d(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<?> output() {
+    return output;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "BatchFFT2D";
+  
+  private Output<?> output;
+  
+  private BatchFft2d(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/BatchFft3d.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/BatchFft3d.java
new file mode 100644
index 00000000000..beeb97ee414
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/BatchFft3d.java
@@ -0,0 +1,72 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.signal;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ */
+@Operator(group = "signal")
+public final class BatchFft3d extends RawOp implements Operand<TType> {
+  
+  /**
+   * Factory method to create a class wrapping a new BatchFft3d operation.
+   * 
+   * @param scope current scope
+   * @param input 
+   * @return a new instance of BatchFft3d
+   */
+  @Endpoint(describeByClass = true)
+  public static BatchFft3d create(Scope scope, Operand<?> input) {
+    OperationBuilder opBuilder = scope.env().opBuilder("BatchFFT3D", scope.makeOpName("BatchFft3d"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new BatchFft3d(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<?> output() {
+    return output;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "BatchFFT3D";
+  
+  private Output<?> output;
+  
+  private BatchFft3d(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/BatchIfft.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/BatchIfft.java
new file mode 100644
index 00000000000..262b02d2bc5
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/BatchIfft.java
@@ -0,0 +1,72 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.signal;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ */
+@Operator(group = "signal")
+public final class BatchIfft extends RawOp implements Operand<TType> {
+  
+  /**
+   * Factory method to create a class wrapping a new BatchIfft operation.
+   * 
+   * @param scope current scope
+   * @param input 
+   * @return a new instance of BatchIfft
+   */
+  @Endpoint(describeByClass = true)
+  public static BatchIfft create(Scope scope, Operand<?> input) {
+    OperationBuilder opBuilder = scope.env().opBuilder("BatchIFFT", scope.makeOpName("BatchIfft"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new BatchIfft(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<?> output() {
+    return output;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "BatchIFFT";
+  
+  private Output<?> output;
+  
+  private BatchIfft(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/BatchIfft2d.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/BatchIfft2d.java
new file mode 100644
index 00000000000..2145f201419
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/BatchIfft2d.java
@@ -0,0 +1,72 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.signal;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ */
+@Operator(group = "signal")
+public final class BatchIfft2d extends RawOp implements Operand<TType> {
+  
+  /**
+   * Factory method to create a class wrapping a new BatchIfft2d operation.
+   * 
+   * @param scope current scope
+   * @param input 
+   * @return a new instance of BatchIfft2d
+   */
+  @Endpoint(describeByClass = true)
+  public static BatchIfft2d create(Scope scope, Operand<?> input) {
+    OperationBuilder opBuilder = scope.env().opBuilder("BatchIFFT2D", scope.makeOpName("BatchIfft2d"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new BatchIfft2d(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<?> output() {
+    return output;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "BatchIFFT2D";
+  
+  private Output<?> output;
+  
+  private BatchIfft2d(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/BatchIfft3d.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/BatchIfft3d.java
new file mode 100644
index 00000000000..6a271795308
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/BatchIfft3d.java
@@ -0,0 +1,72 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.signal;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ */
+@Operator(group = "signal")
+public final class BatchIfft3d extends RawOp implements Operand<TType> {
+  
+  /**
+   * Factory method to create a class wrapping a new BatchIfft3d operation.
+   * 
+   * @param scope current scope
+   * @param input 
+   * @return a new instance of BatchIfft3d
+   */
+  @Endpoint(describeByClass = true)
+  public static BatchIfft3d create(Scope scope, Operand<?> input) {
+    OperationBuilder opBuilder = scope.env().opBuilder("BatchIFFT3D", scope.makeOpName("BatchIfft3d"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new BatchIfft3d(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<?> output() {
+    return output;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "BatchIFFT3D";
+  
+  private Output<?> output;
+  
+  private BatchIfft3d(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Fft.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Fft.java
new file mode 100644
index 00000000000..632500fad5e
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Fft.java
@@ -0,0 +1,83 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.signal;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Fast Fourier transform.
+ * <p>
+ * Computes the 1-dimensional discrete Fourier transform over the inner-most
+ * dimension of `input`.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator(group = "signal")
+public final class Fft<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new Fft operation.
+   * 
+   * @param scope current scope
+   * @param input A complex tensor.
+   * @return a new instance of Fft
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> Fft<T> create(Scope scope, Operand<T> input) {
+    OperationBuilder opBuilder = scope.env().opBuilder("FFT", scope.makeOpName("Fft"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new Fft<T>(opBuilder.build());
+  }
+  
+  /**
+   * A complex tensor of the same shape as `input`. The inner-most
+   *   dimension of `input` is replaced with its 1D Fourier transform.
+   * <p>
+   * @compatibility(numpy)
+   * Equivalent to np.fft.fft
+   * @end_compatibility
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "FFT";
+  
+  private Output<T> output;
+  
+  private Fft(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Fft2d.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Fft2d.java
new file mode 100644
index 00000000000..201b2e3c8af
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Fft2d.java
@@ -0,0 +1,83 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.signal;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * 2D fast Fourier transform.
+ * <p>
+ * Computes the 2-dimensional discrete Fourier transform over the inner-most
+ * 2 dimensions of `input`.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator(group = "signal")
+public final class Fft2d<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new Fft2d operation.
+   * 
+   * @param scope current scope
+   * @param input A complex tensor.
+   * @return a new instance of Fft2d
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> Fft2d<T> create(Scope scope, Operand<T> input) {
+    OperationBuilder opBuilder = scope.env().opBuilder("FFT2D", scope.makeOpName("Fft2d"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new Fft2d<T>(opBuilder.build());
+  }
+  
+  /**
+   * A complex tensor of the same shape as `input`. The inner-most 2
+   *   dimensions of `input` are replaced with their 2D Fourier transform.
+   * <p>
+   * @compatibility(numpy)
+   * Equivalent to np.fft.fft2
+   * @end_compatibility
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "FFT2D";
+  
+  private Output<T> output;
+  
+  private Fft2d(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Fft3d.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Fft3d.java
new file mode 100644
index 00000000000..7e840fbfb62
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Fft3d.java
@@ -0,0 +1,83 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.signal;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * 3D fast Fourier transform.
+ * <p>
+ * Computes the 3-dimensional discrete Fourier transform over the inner-most 3
+ * dimensions of `input`.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator(group = "signal")
+public final class Fft3d<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new Fft3d operation.
+   * 
+   * @param scope current scope
+   * @param input A complex tensor.
+   * @return a new instance of Fft3d
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> Fft3d<T> create(Scope scope, Operand<T> input) {
+    OperationBuilder opBuilder = scope.env().opBuilder("FFT3D", scope.makeOpName("Fft3d"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new Fft3d<T>(opBuilder.build());
+  }
+  
+  /**
+   * A complex tensor of the same shape as `input`. The inner-most 3
+   *   dimensions of `input` are replaced with their 3D Fourier transform.
+   * <p>
+   * @compatibility(numpy)
+   * Equivalent to np.fft.fftn with 3 dimensions.
+   * @end_compatibility
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "FFT3D";
+  
+  private Output<T> output;
+  
+  private Fft3d(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Ifft.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Ifft.java
new file mode 100644
index 00000000000..c020956beec
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Ifft.java
@@ -0,0 +1,83 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.signal;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Inverse fast Fourier transform.
+ * <p>
+ * Computes the inverse 1-dimensional discrete Fourier transform over the
+ * inner-most dimension of `input`.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator(group = "signal")
+public final class Ifft<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new Ifft operation.
+   * 
+   * @param scope current scope
+   * @param input A complex tensor.
+   * @return a new instance of Ifft
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> Ifft<T> create(Scope scope, Operand<T> input) {
+    OperationBuilder opBuilder = scope.env().opBuilder("IFFT", scope.makeOpName("Ifft"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new Ifft<T>(opBuilder.build());
+  }
+  
+  /**
+   * A complex tensor of the same shape as `input`. The inner-most
+   *   dimension of `input` is replaced with its inverse 1D Fourier transform.
+   * <p>
+   * @compatibility(numpy)
+   * Equivalent to np.fft.ifft
+   * @end_compatibility
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "IFFT";
+  
+  private Output<T> output;
+  
+  private Ifft(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Ifft2d.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Ifft2d.java
new file mode 100644
index 00000000000..c8cac107ff4
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Ifft2d.java
@@ -0,0 +1,83 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.signal;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Inverse 2D fast Fourier transform.
+ * <p>
+ * Computes the inverse 2-dimensional discrete Fourier transform over the
+ * inner-most 2 dimensions of `input`.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator(group = "signal")
+public final class Ifft2d<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new Ifft2d operation.
+   * 
+   * @param scope current scope
+   * @param input A complex tensor.
+   * @return a new instance of Ifft2d
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> Ifft2d<T> create(Scope scope, Operand<T> input) {
+    OperationBuilder opBuilder = scope.env().opBuilder("IFFT2D", scope.makeOpName("Ifft2d"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new Ifft2d<T>(opBuilder.build());
+  }
+  
+  /**
+   * A complex tensor of the same shape as `input`. The inner-most 2
+   *   dimensions of `input` are replaced with their inverse 2D Fourier transform.
+   * <p>
+   * @compatibility(numpy)
+   * Equivalent to np.fft.ifft2
+   * @end_compatibility
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "IFFT2D";
+  
+  private Output<T> output;
+  
+  private Ifft2d(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Ifft3d.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Ifft3d.java
new file mode 100644
index 00000000000..ce7f87a9aa2
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Ifft3d.java
@@ -0,0 +1,83 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.signal;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Inverse 3D fast Fourier transform.
+ * <p>
+ * Computes the inverse 3-dimensional discrete Fourier transform over the
+ * inner-most 3 dimensions of `input`.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator(group = "signal")
+public final class Ifft3d<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new Ifft3d operation.
+   * 
+   * @param scope current scope
+   * @param input A complex tensor.
+   * @return a new instance of Ifft3d
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> Ifft3d<T> create(Scope scope, Operand<T> input) {
+    OperationBuilder opBuilder = scope.env().opBuilder("IFFT3D", scope.makeOpName("Ifft3d"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new Ifft3d<T>(opBuilder.build());
+  }
+  
+  /**
+   * A complex tensor of the same shape as `input`. The inner-most 3
+   *   dimensions of `input` are replaced with their inverse 3D Fourier transform.
+   * <p>
+   * @compatibility(numpy)
+   * Equivalent to np.fft.ifftn with 3 dimensions.
+   * @end_compatibility
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "IFFT3D";
+  
+  private Output<T> output;
+  
+  private Ifft3d(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Irfft.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Irfft.java
new file mode 100644
index 00000000000..b0441b47911
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Irfft.java
@@ -0,0 +1,115 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.signal;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Inverse real-valued fast Fourier transform.
+ * <p>
+ * Computes the inverse 1-dimensional discrete Fourier transform of a real-valued
+ * signal over the inner-most dimension of `input`.
+ * <p>
+ * The inner-most dimension of `input` is assumed to be the result of `RFFT`: the
+ * `fft_length / 2 + 1` unique components of the DFT of a real-valued signal. If
+ * `fft_length` is not provided, it is computed from the size of the inner-most
+ * dimension of `input` (`fft_length = 2 * (inner - 1)`). If the FFT length used to
+ * compute `input` is odd, it should be provided since it cannot be inferred
+ * properly.
+ * <p>
+ * Along the axis `signal.Irfft` is computed on, if `fft_length / 2 + 1` is smaller
+ * than the corresponding dimension of `input`, the dimension is cropped. If it is
+ * larger, the dimension is padded with zeros.
+ * 
+ * @param <U> data type for {@code output()} output
+ */
+@Operator(group = "signal")
+public final class Irfft<U extends TNumber> extends RawOp implements Operand<U> {
+  
+  /**
+   * Factory method to create a class wrapping a new Irfft operation.
+   * 
+   * @param scope current scope
+   * @param input A complex tensor.
+   * @param fftLength An int32 tensor of shape [1]. The FFT length.
+   * @param Treal 
+   * @return a new instance of Irfft
+   */
+  @Endpoint(describeByClass = true)
+  public static <U extends TNumber, T extends TType> Irfft<U> create(Scope scope, Operand<T> input, Operand<TInt32> fftLength, Class<U> Treal) {
+    OperationBuilder opBuilder = scope.env().opBuilder("IRFFT", scope.makeOpName("Irfft"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(fftLength.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("Treal", Treal);
+    return new Irfft<U>(opBuilder.build());
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new Irfft operation using default output types.
+   * 
+   * @param scope current scope
+   * @param input A complex tensor.
+   * @param fftLength An int32 tensor of shape [1]. The FFT length.
+   * @return a new instance of Irfft
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> Irfft<TFloat32> create(Scope scope, Operand<T> input, Operand<TInt32> fftLength) {
+    return create(scope, input, fftLength, TFloat32.class);
+  }
+  
+  /**
+   * A float32 tensor of the same rank as `input`. The inner-most
+   *   dimension of `input` is replaced with the `fft_length` samples of its inverse
+   *   1D Fourier transform.
+   * <p>
+   * @compatibility(numpy)
+   * Equivalent to np.fft.irfft
+   * @end_compatibility
+   */
+  public Output<U> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<U> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "IRFFT";
+  
+  private Output<U> output;
+  
+  private Irfft(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Irfft2d.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Irfft2d.java
new file mode 100644
index 00000000000..726e1ea31e6
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Irfft2d.java
@@ -0,0 +1,116 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.signal;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Inverse 2D real-valued fast Fourier transform.
+ * <p>
+ * Computes the inverse 2-dimensional discrete Fourier transform of a real-valued
+ * signal over the inner-most 2 dimensions of `input`.
+ * <p>
+ * The inner-most 2 dimensions of `input` are assumed to be the result of `RFFT2D`:
+ * The inner-most dimension contains the `fft_length / 2 + 1` unique components of
+ * the DFT of a real-valued signal. If `fft_length` is not provided, it is computed
+ * from the size of the inner-most 2 dimensions of `input`. If the FFT length used
+ * to compute `input` is odd, it should be provided since it cannot be inferred
+ * properly.
+ * <p>
+ * Along each axis `signal.Irfft2d` is computed on, if `fft_length` (or
+ * `fft_length / 2 + 1` for the inner-most dimension) is smaller than the
+ * corresponding dimension of `input`, the dimension is cropped. If it is larger,
+ * the dimension is padded with zeros.
+ * 
+ * @param <U> data type for {@code output()} output
+ */
+@Operator(group = "signal")
+public final class Irfft2d<U extends TNumber> extends RawOp implements Operand<U> {
+  
+  /**
+   * Factory method to create a class wrapping a new Irfft2d operation.
+   * 
+   * @param scope current scope
+   * @param input A complex tensor.
+   * @param fftLength An int32 tensor of shape [2]. The FFT length for each dimension.
+   * @param Treal 
+   * @return a new instance of Irfft2d
+   */
+  @Endpoint(describeByClass = true)
+  public static <U extends TNumber, T extends TType> Irfft2d<U> create(Scope scope, Operand<T> input, Operand<TInt32> fftLength, Class<U> Treal) {
+    OperationBuilder opBuilder = scope.env().opBuilder("IRFFT2D", scope.makeOpName("Irfft2d"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(fftLength.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("Treal", Treal);
+    return new Irfft2d<U>(opBuilder.build());
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new Irfft2d operation using default output types.
+   * 
+   * @param scope current scope
+   * @param input A complex tensor.
+   * @param fftLength An int32 tensor of shape [2]. The FFT length for each dimension.
+   * @return a new instance of Irfft2d
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> Irfft2d<TFloat32> create(Scope scope, Operand<T> input, Operand<TInt32> fftLength) {
+    return create(scope, input, fftLength, TFloat32.class);
+  }
+  
+  /**
+   * A float32 tensor of the same rank as `input`. The inner-most 2
+   *   dimensions of `input` are replaced with the `fft_length` samples of their
+   *   inverse 2D Fourier transform.
+   * <p>
+   * @compatibility(numpy)
+   * Equivalent to np.fft.irfft2
+   * @end_compatibility
+   */
+  public Output<U> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<U> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "IRFFT2D";
+  
+  private Output<U> output;
+  
+  private Irfft2d(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Irfft3d.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Irfft3d.java
new file mode 100644
index 00000000000..0d76b5591e7
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Irfft3d.java
@@ -0,0 +1,116 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.signal;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Inverse 3D real-valued fast Fourier transform.
+ * <p>
+ * Computes the inverse 3-dimensional discrete Fourier transform of a real-valued
+ * signal over the inner-most 3 dimensions of `input`.
+ * <p>
+ * The inner-most 3 dimensions of `input` are assumed to be the result of `RFFT3D`:
+ * The inner-most dimension contains the `fft_length / 2 + 1` unique components of
+ * the DFT of a real-valued signal. If `fft_length` is not provided, it is computed
+ * from the size of the inner-most 3 dimensions of `input`. If the FFT length used
+ * to compute `input` is odd, it should be provided since it cannot be inferred
+ * properly.
+ * <p>
+ * Along each axis `signal.Irfft3d` is computed on, if `fft_length` (or
+ * `fft_length / 2 + 1` for the inner-most dimension) is smaller than the
+ * corresponding dimension of `input`, the dimension is cropped. If it is larger,
+ * the dimension is padded with zeros.
+ * 
+ * @param <U> data type for {@code output()} output
+ */
+@Operator(group = "signal")
+public final class Irfft3d<U extends TNumber> extends RawOp implements Operand<U> {
+  
+  /**
+   * Factory method to create a class wrapping a new Irfft3d operation.
+   * 
+   * @param scope current scope
+   * @param input A complex tensor.
+   * @param fftLength An int32 tensor of shape [3]. The FFT length for each dimension.
+   * @param Treal 
+   * @return a new instance of Irfft3d
+   */
+  @Endpoint(describeByClass = true)
+  public static <U extends TNumber, T extends TType> Irfft3d<U> create(Scope scope, Operand<T> input, Operand<TInt32> fftLength, Class<U> Treal) {
+    OperationBuilder opBuilder = scope.env().opBuilder("IRFFT3D", scope.makeOpName("Irfft3d"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(fftLength.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("Treal", Treal);
+    return new Irfft3d<U>(opBuilder.build());
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new Irfft3d operation using default output types.
+   * 
+   * @param scope current scope
+   * @param input A complex tensor.
+   * @param fftLength An int32 tensor of shape [3]. The FFT length for each dimension.
+   * @return a new instance of Irfft3d
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> Irfft3d<TFloat32> create(Scope scope, Operand<T> input, Operand<TInt32> fftLength) {
+    return create(scope, input, fftLength, TFloat32.class);
+  }
+  
+  /**
+   * A float32 tensor of the same rank as `input`. The inner-most 3
+   *   dimensions of `input` are replaced with the `fft_length` samples of their
+   *   inverse 3D real Fourier transform.
+   * <p>
+   * @compatibility(numpy)
+   * Equivalent to np.irfftn with 3 dimensions.
+   * @end_compatibility
+   */
+  public Output<U> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<U> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "IRFFT3D";
+  
+  private Output<U> output;
+  
+  private Irfft3d(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Rfft.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Rfft.java
new file mode 100644
index 00000000000..5d03d58a7b7
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Rfft.java
@@ -0,0 +1,98 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.signal;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Real-valued fast Fourier transform.
+ * <p>
+ * Computes the 1-dimensional discrete Fourier transform of a real-valued signal
+ * over the inner-most dimension of `input`.
+ * <p>
+ * Since the DFT of a real signal is Hermitian-symmetric, `signal.Rfft` only returns the
+ * `fft_length / 2 + 1` unique components of the FFT: the zero-frequency term,
+ * followed by the `fft_length / 2` positive-frequency terms.
+ * <p>
+ * Along the axis `signal.Rfft` is computed on, if `fft_length` is smaller than the
+ * corresponding dimension of `input`, the dimension is cropped. If it is larger,
+ * the dimension is padded with zeros.
+ * 
+ * @param <U> data type for {@code output()} output
+ */
+@Operator(group = "signal")
+public final class Rfft<U extends TType> extends RawOp implements Operand<U> {
+  
+  /**
+   * Factory method to create a class wrapping a new Rfft operation.
+   * 
+   * @param scope current scope
+   * @param input A float32 tensor.
+   * @param fftLength An int32 tensor of shape [1]. The FFT length.
+   * @param Tcomplex 
+   * @return a new instance of Rfft
+   */
+  @Endpoint(describeByClass = true)
+  public static <U extends TType, T extends TNumber> Rfft<U> create(Scope scope, Operand<T> input, Operand<TInt32> fftLength, Class<U> Tcomplex) {
+    OperationBuilder opBuilder = scope.env().opBuilder("RFFT", scope.makeOpName("Rfft"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(fftLength.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("Tcomplex", Tcomplex);
+    return new Rfft<U>(opBuilder.build());
+  }
+  
+  /**
+   * A complex64 tensor of the same rank as `input`. The inner-most
+   *   dimension of `input` is replaced with the `fft_length / 2 + 1` unique
+   *   frequency components of its 1D Fourier transform.
+   * <p>
+   * @compatibility(numpy)
+   * Equivalent to np.fft.rfft
+   * @end_compatibility
+   */
+  public Output<U> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<U> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "RFFT";
+  
+  private Output<U> output;
+  
+  private Rfft(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Rfft2d.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Rfft2d.java
new file mode 100644
index 00000000000..ddecd0f133f
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Rfft2d.java
@@ -0,0 +1,100 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.signal;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * 2D real-valued fast Fourier transform.
+ * <p>
+ * Computes the 2-dimensional discrete Fourier transform of a real-valued signal
+ * over the inner-most 2 dimensions of `input`.
+ * <p>
+ * Since the DFT of a real signal is Hermitian-symmetric, `signal.Rfft2d` only returns the
+ * `fft_length / 2 + 1` unique components of the FFT for the inner-most dimension
+ * of `output`: the zero-frequency term, followed by the `fft_length / 2`
+ * positive-frequency terms.
+ * <p>
+ * Along each axis `signal.Rfft2d` is computed on, if `fft_length` is smaller than the
+ * corresponding dimension of `input`, the dimension is cropped. If it is larger,
+ * the dimension is padded with zeros.
+ * 
+ * @param <U> data type for {@code output()} output
+ */
+@Operator(group = "signal")
+public final class Rfft2d<U extends TType> extends RawOp implements Operand<U> {
+  
+  /**
+   * Factory method to create a class wrapping a new Rfft2d operation.
+   * 
+   * @param scope current scope
+   * @param input A float32 tensor.
+   * @param fftLength An int32 tensor of shape [2]. The FFT length for each dimension.
+   * @param Tcomplex 
+   * @return a new instance of Rfft2d
+   */
+  @Endpoint(describeByClass = true)
+  public static <U extends TType, T extends TNumber> Rfft2d<U> create(Scope scope, Operand<T> input, Operand<TInt32> fftLength, Class<U> Tcomplex) {
+    OperationBuilder opBuilder = scope.env().opBuilder("RFFT2D", scope.makeOpName("Rfft2d"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(fftLength.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("Tcomplex", Tcomplex);
+    return new Rfft2d<U>(opBuilder.build());
+  }
+  
+  /**
+   * A complex64 tensor of the same rank as `input`. The inner-most 2
+   *   dimensions of `input` are replaced with their 2D Fourier transform. The
+   *   inner-most dimension contains `fft_length / 2 + 1` unique frequency
+   *   components.
+   * <p>
+   * @compatibility(numpy)
+   * Equivalent to np.fft.rfft2
+   * @end_compatibility
+   */
+  public Output<U> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<U> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "RFFT2D";
+  
+  private Output<U> output;
+  
+  private Rfft2d(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Rfft3d.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Rfft3d.java
new file mode 100644
index 00000000000..8fd625822e4
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/signal/Rfft3d.java
@@ -0,0 +1,100 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.signal;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * 3D real-valued fast Fourier transform.
+ * <p>
+ * Computes the 3-dimensional discrete Fourier transform of a real-valued signal
+ * over the inner-most 3 dimensions of `input`.
+ * <p>
+ * Since the DFT of a real signal is Hermitian-symmetric, `signal.Rfft3d` only returns the
+ * `fft_length / 2 + 1` unique components of the FFT for the inner-most dimension
+ * of `output`: the zero-frequency term, followed by the `fft_length / 2`
+ * positive-frequency terms.
+ * <p>
+ * Along each axis `signal.Rfft3d` is computed on, if `fft_length` is smaller than the
+ * corresponding dimension of `input`, the dimension is cropped. If it is larger,
+ * the dimension is padded with zeros.
+ * 
+ * @param <U> data type for {@code output()} output
+ */
+@Operator(group = "signal")
+public final class Rfft3d<U extends TType> extends RawOp implements Operand<U> {
+  
+  /**
+   * Factory method to create a class wrapping a new Rfft3d operation.
+   * 
+   * @param scope current scope
+   * @param input A float32 tensor.
+   * @param fftLength An int32 tensor of shape [3]. The FFT length for each dimension.
+   * @param Tcomplex 
+   * @return a new instance of Rfft3d
+   */
+  @Endpoint(describeByClass = true)
+  public static <U extends TType, T extends TNumber> Rfft3d<U> create(Scope scope, Operand<T> input, Operand<TInt32> fftLength, Class<U> Tcomplex) {
+    OperationBuilder opBuilder = scope.env().opBuilder("RFFT3D", scope.makeOpName("Rfft3d"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(fftLength.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("Tcomplex", Tcomplex);
+    return new Rfft3d<U>(opBuilder.build());
+  }
+  
+  /**
+   * A complex64 tensor of the same rank as `input`. The inner-most 3
+   *   dimensions of `input` are replaced with the their 3D Fourier transform. The
+   *   inner-most dimension contains `fft_length / 2 + 1` unique frequency
+   *   components.
+   * <p>
+   * @compatibility(numpy)
+   * Equivalent to np.fft.rfftn with 3 dimensions.
+   * @end_compatibility
+   */
+  public Output<U> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<U> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "RFFT3D";
+  
+  private Output<U> output;
+  
+  private Rfft3d(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/AddManySparseToTensorsMap.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/AddManySparseToTensorsMap.java
new file mode 100644
index 00000000000..ec66167f7e6
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/AddManySparseToTensorsMap.java
@@ -0,0 +1,158 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.sparse;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Add an `N`-minibatch `SparseTensor` to a `SparseTensorsMap`, return `N` handles.
+ * <p>
+ * A `SparseTensor` of rank `R` is represented by three tensors: `sparse_indices`,
+ * `sparse_values`, and `sparse_shape`, where
+ * <pre>{@code
+ * sparse_indices.shape[1] == sparse_shape.shape[0] == R}</pre>
+ * An `N`-minibatch of `SparseTensor` objects is represented as a `SparseTensor`
+ * having a first `sparse_indices` column taking values between `[0, N)`, where
+ * the minibatch size `N == sparse_shape[0]`.
+ * <p>
+ * The input `SparseTensor` must have rank `R` greater than 1, and the first
+ * dimension is treated as the minibatch dimension.  Elements of the `SparseTensor`
+ * must be sorted in increasing order of this first dimension.  The stored
+ * `SparseTensor` objects pointed to by each row of the output `sparse_handles`
+ * will have rank `R-1`.
+ * <p>
+ * The `SparseTensor` values can then be read out as part of a minibatch by passing
+ * the given keys as vector elements to `TakeManySparseFromTensorsMap`.  To ensure
+ * the correct `SparseTensorsMap` is accessed, ensure that the same
+ * `container` and `shared_name` are passed to that Op.  If no `shared_name`
+ * is provided here, instead use the <i>name</i> of the Operation created by calling
+ * `sparse.AddManySparseToTensorsMap` as the `shared_name` passed to
+ * `TakeManySparseFromTensorsMap`.  Ensure the Operations are colocated.
+ */
+@Operator(group = "sparse")
+public final class AddManySparseToTensorsMap extends RawOp implements Operand<TInt64> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.sparse.AddManySparseToTensorsMap}
+   */
+  public static class Options {
+    
+    /**
+     * @param container The container name for the `SparseTensorsMap` created by this op.
+     */
+    public Options container(String container) {
+      this.container = container;
+      return this;
+    }
+    
+    /**
+     * @param sharedName The shared name for the `SparseTensorsMap` created by this op.
+     * If blank, the new Operation's unique name is used.
+     */
+    public Options sharedName(String sharedName) {
+      this.sharedName = sharedName;
+      return this;
+    }
+    
+    private String container;
+    private String sharedName;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new AddManySparseToTensorsMap operation.
+   * 
+   * @param scope current scope
+   * @param sparseIndices 2-D.  The `indices` of the minibatch `SparseTensor`.
+   * `sparse_indices[:, 0]` must be ordered values in `[0, N)`.
+   * @param sparseValues 1-D.  The `values` of the minibatch `SparseTensor`.
+   * @param sparseShape 1-D.  The `shape` of the minibatch `SparseTensor`.
+   * The minibatch size `N == sparse_shape[0]`.
+   * @param options carries optional attributes values
+   * @return a new instance of AddManySparseToTensorsMap
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> AddManySparseToTensorsMap create(Scope scope, Operand<TInt64> sparseIndices, Operand<T> sparseValues, Operand<TInt64> sparseShape, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("AddManySparseToTensorsMap", scope.makeOpName("AddManySparseToTensorsMap"));
+    opBuilder.addInput(sparseIndices.asOutput());
+    opBuilder.addInput(sparseValues.asOutput());
+    opBuilder.addInput(sparseShape.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.container != null) {
+          opBuilder.setAttr("container", opts.container);
+        }
+        if (opts.sharedName != null) {
+          opBuilder.setAttr("shared_name", opts.sharedName);
+        }
+      }
+    }
+    return new AddManySparseToTensorsMap(opBuilder.build());
+  }
+  
+  /**
+   * @param container The container name for the `SparseTensorsMap` created by this op.
+   */
+  public static Options container(String container) {
+    return new Options().container(container);
+  }
+  
+  /**
+   * @param sharedName The shared name for the `SparseTensorsMap` created by this op.
+   * If blank, the new Operation's unique name is used.
+   */
+  public static Options sharedName(String sharedName) {
+    return new Options().sharedName(sharedName);
+  }
+  
+  /**
+   * 1-D.  The handles of the `SparseTensor` now stored in the
+   * `SparseTensorsMap`.  Shape: `[N]`.
+   */
+  public Output<TInt64> sparseHandles() {
+    return sparseHandles;
+  }
+  
+  @Override
+  public Output<TInt64> asOutput() {
+    return sparseHandles;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "AddManySparseToTensorsMap";
+  
+  private Output<TInt64> sparseHandles;
+  
+  private AddManySparseToTensorsMap(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    sparseHandles = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/AddSparseToTensorsMap.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/AddSparseToTensorsMap.java
new file mode 100644
index 00000000000..153e326aa9d
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/AddSparseToTensorsMap.java
@@ -0,0 +1,149 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.sparse;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Add a `SparseTensor` to a `SparseTensorsMap` return its handle.
+ * <p>
+ * A `SparseTensor` is represented by three tensors: `sparse_indices`,
+ * `sparse_values`, and `sparse_shape`.
+ * <p>
+ * This operator takes the given `SparseTensor` and adds it to a container
+ * object (a `SparseTensorsMap`).  A unique key within this container is generated
+ * in the form of an `int64`, and this is the value that is returned.
+ * <p>
+ * The `SparseTensor` can then be read out as part of a minibatch by passing
+ * the key as a vector element to `TakeManySparseFromTensorsMap`.  To ensure
+ * the correct `SparseTensorsMap` is accessed, ensure that the same
+ * `container` and `shared_name` are passed to that Op.  If no `shared_name`
+ * is provided here, instead use the <i>name</i> of the Operation created by calling
+ * `sparse.AddSparseToTensorsMap` as the `shared_name` passed to
+ * `TakeManySparseFromTensorsMap`.  Ensure the Operations are colocated.
+ */
+@Operator(group = "sparse")
+public final class AddSparseToTensorsMap extends RawOp implements Operand<TInt64> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.sparse.AddSparseToTensorsMap}
+   */
+  public static class Options {
+    
+    /**
+     * @param container The container name for the `SparseTensorsMap` created by this op.
+     */
+    public Options container(String container) {
+      this.container = container;
+      return this;
+    }
+    
+    /**
+     * @param sharedName The shared name for the `SparseTensorsMap` created by this op.
+     * If blank, the new Operation's unique name is used.
+     */
+    public Options sharedName(String sharedName) {
+      this.sharedName = sharedName;
+      return this;
+    }
+    
+    private String container;
+    private String sharedName;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new AddSparseToTensorsMap operation.
+   * 
+   * @param scope current scope
+   * @param sparseIndices 2-D.  The `indices` of the `SparseTensor`.
+   * @param sparseValues 1-D.  The `values` of the `SparseTensor`.
+   * @param sparseShape 1-D.  The `shape` of the `SparseTensor`.
+   * @param options carries optional attributes values
+   * @return a new instance of AddSparseToTensorsMap
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> AddSparseToTensorsMap create(Scope scope, Operand<TInt64> sparseIndices, Operand<T> sparseValues, Operand<TInt64> sparseShape, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("AddSparseToTensorsMap", scope.makeOpName("AddSparseToTensorsMap"));
+    opBuilder.addInput(sparseIndices.asOutput());
+    opBuilder.addInput(sparseValues.asOutput());
+    opBuilder.addInput(sparseShape.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.container != null) {
+          opBuilder.setAttr("container", opts.container);
+        }
+        if (opts.sharedName != null) {
+          opBuilder.setAttr("shared_name", opts.sharedName);
+        }
+      }
+    }
+    return new AddSparseToTensorsMap(opBuilder.build());
+  }
+  
+  /**
+   * @param container The container name for the `SparseTensorsMap` created by this op.
+   */
+  public static Options container(String container) {
+    return new Options().container(container);
+  }
+  
+  /**
+   * @param sharedName The shared name for the `SparseTensorsMap` created by this op.
+   * If blank, the new Operation's unique name is used.
+   */
+  public static Options sharedName(String sharedName) {
+    return new Options().sharedName(sharedName);
+  }
+  
+  /**
+   * 0-D.  The handle of the `SparseTensor` now stored in the
+   * `SparseTensorsMap`.
+   */
+  public Output<TInt64> sparseHandle() {
+    return sparseHandle;
+  }
+  
+  @Override
+  public Output<TInt64> asOutput() {
+    return sparseHandle;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "AddSparseToTensorsMap";
+  
+  private Output<TInt64> sparseHandle;
+  
+  private AddSparseToTensorsMap(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    sparseHandle = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/DenseCountSparseOutput.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/DenseCountSparseOutput.java
new file mode 100644
index 00000000000..de6c4ef9eb0
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/DenseCountSparseOutput.java
@@ -0,0 +1,148 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.sparse;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Performs sparse-output bin counting for a tf.tensor input.
+ * <p>
+ *   Counts the number of times each value occurs in the input.
+ * 
+ * @param <U> data type for {@code outputValues()} output
+ */
+public final class DenseCountSparseOutput<U extends TNumber> extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.sparse.DenseCountSparseOutput}
+   */
+  public static class Options {
+    
+    /**
+     * @param minlength Minimum value to count. Can be set to -1 for no minimum.
+     */
+    public Options minlength(Long minlength) {
+      this.minlength = minlength;
+      return this;
+    }
+    
+    /**
+     * @param maxlength Maximum value to count. Can be set to -1 for no maximum.
+     */
+    public Options maxlength(Long maxlength) {
+      this.maxlength = maxlength;
+      return this;
+    }
+    
+    private Long minlength;
+    private Long maxlength;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new DenseCountSparseOutput operation.
+   * 
+   * @param scope current scope
+   * @param values Tensor containing data to count.
+   * @param weights A Tensor of the same shape as indices containing per-index weight values. May
+   * also be the empty tensor if no weights are used.
+   * @param binaryOutput Whether to output the number of occurrences of each value or 1.
+   * @param options carries optional attributes values
+   * @return a new instance of DenseCountSparseOutput
+   */
+  @Endpoint(describeByClass = true)
+  public static <U extends TNumber, T extends TNumber> DenseCountSparseOutput<U> create(Scope scope, Operand<T> values, Operand<U> weights, Boolean binaryOutput, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("DenseCountSparseOutput", scope.makeOpName("DenseCountSparseOutput"));
+    opBuilder.addInput(values.asOutput());
+    opBuilder.addInput(weights.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("binary_output", binaryOutput);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.minlength != null) {
+          opBuilder.setAttr("minlength", opts.minlength);
+        }
+        if (opts.maxlength != null) {
+          opBuilder.setAttr("maxlength", opts.maxlength);
+        }
+      }
+    }
+    return new DenseCountSparseOutput<U>(opBuilder.build());
+  }
+  
+  /**
+   * @param minlength Minimum value to count. Can be set to -1 for no minimum.
+   */
+  public static Options minlength(Long minlength) {
+    return new Options().minlength(minlength);
+  }
+  
+  /**
+   * @param maxlength Maximum value to count. Can be set to -1 for no maximum.
+   */
+  public static Options maxlength(Long maxlength) {
+    return new Options().maxlength(maxlength);
+  }
+  
+  /**
+   * Indices tensor for the resulting sparse tensor object.
+   */
+  public Output<TInt64> outputIndices() {
+    return outputIndices;
+  }
+  
+  /**
+   * Values tensor for the resulting sparse tensor object.
+   */
+  public Output<U> outputValues() {
+    return outputValues;
+  }
+  
+  /**
+   * Shape tensor for the resulting sparse tensor object.
+   */
+  public Output<TInt64> outputDenseShape() {
+    return outputDenseShape;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "DenseCountSparseOutput";
+  
+  private Output<TInt64> outputIndices;
+  private Output<U> outputValues;
+  private Output<TInt64> outputDenseShape;
+  
+  private DenseCountSparseOutput(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    outputIndices = operation.output(outputIdx++);
+    outputValues = operation.output(outputIdx++);
+    outputDenseShape = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/DenseToDenseSetOperation.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/DenseToDenseSetOperation.java
new file mode 100644
index 00000000000..15e62be4fa8
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/DenseToDenseSetOperation.java
@@ -0,0 +1,139 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.sparse;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Applies set operation along last dimension of 2 `Tensor` inputs.
+ * <p>
+ * See SetOperationOp::SetOperationFromContext for values of `set_operation`.
+ * <p>
+ * Output `result` is a `SparseTensor` represented by `result_indices`,
+ * `result_values`, and `result_shape`. For `set1` and `set2` ranked `n`, this
+ * has rank `n` and the same 1st `n-1` dimensions as `set1` and `set2`. The `nth`
+ * dimension contains the result of `set_operation` applied to the corresponding
+ * `[0...n-1]` dimension of `set`.
+ * 
+ * @param <T> data type for {@code resultValues()} output
+ */
+@Operator(group = "sparse")
+public final class DenseToDenseSetOperation<T extends TType> extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.sparse.DenseToDenseSetOperation}
+   */
+  public static class Options {
+    
+    /**
+     * @param validateIndices 
+     */
+    public Options validateIndices(Boolean validateIndices) {
+      this.validateIndices = validateIndices;
+      return this;
+    }
+    
+    private Boolean validateIndices;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new DenseToDenseSetOperation operation.
+   * 
+   * @param scope current scope
+   * @param set1 `Tensor` with rank `n`. 1st `n-1` dimensions must be the same as `set2`.
+   * Dimension `n` contains values in a set, duplicates are allowed but ignored.
+   * @param set2 `Tensor` with rank `n`. 1st `n-1` dimensions must be the same as `set1`.
+   * Dimension `n` contains values in a set, duplicates are allowed but ignored.
+   * @param setOperation 
+   * @param options carries optional attributes values
+   * @return a new instance of DenseToDenseSetOperation
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> DenseToDenseSetOperation<T> create(Scope scope, Operand<T> set1, Operand<T> set2, String setOperation, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("DenseToDenseSetOperation", scope.makeOpName("DenseToDenseSetOperation"));
+    opBuilder.addInput(set1.asOutput());
+    opBuilder.addInput(set2.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("set_operation", setOperation);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.validateIndices != null) {
+          opBuilder.setAttr("validate_indices", opts.validateIndices);
+        }
+      }
+    }
+    return new DenseToDenseSetOperation<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param validateIndices 
+   */
+  public static Options validateIndices(Boolean validateIndices) {
+    return new Options().validateIndices(validateIndices);
+  }
+  
+  /**
+   * 2D indices of a `SparseTensor`.
+   */
+  public Output<TInt64> resultIndices() {
+    return resultIndices;
+  }
+  
+  /**
+   * 1D values of a `SparseTensor`.
+   */
+  public Output<T> resultValues() {
+    return resultValues;
+  }
+  
+  /**
+   * 1D `Tensor` shape of a `SparseTensor`. `result_shape[0...n-1]` is
+   * the same as the 1st `n-1` dimensions of `set1` and `set2`, `result_shape[n]`
+   * is the max result set size across all `0...n-1` dimensions.
+   */
+  public Output<TInt64> resultShape() {
+    return resultShape;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "DenseToDenseSetOperation";
+  
+  private Output<TInt64> resultIndices;
+  private Output<T> resultValues;
+  private Output<TInt64> resultShape;
+  
+  private DenseToDenseSetOperation(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    resultIndices = operation.output(outputIdx++);
+    resultValues = operation.output(outputIdx++);
+    resultShape = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/DenseToSparseSetOperation.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/DenseToSparseSetOperation.java
new file mode 100644
index 00000000000..107f32524ea
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/DenseToSparseSetOperation.java
@@ -0,0 +1,154 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.sparse;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Applies set operation along last dimension of `Tensor` and `SparseTensor`.
+ * <p>
+ * See SetOperationOp::SetOperationFromContext for values of `set_operation`.
+ * <p>
+ * Input `set2` is a `SparseTensor` represented by `set2_indices`, `set2_values`,
+ * and `set2_shape`. For `set2` ranked `n`, 1st `n-1` dimensions must be the same
+ * as `set1`. Dimension `n` contains values in a set, duplicates are allowed but
+ * ignored.
+ * <p>
+ * If `validate_indices` is `True`, this op validates the order and range of `set2`
+ * indices.
+ * <p>
+ * Output `result` is a `SparseTensor` represented by `result_indices`,
+ * `result_values`, and `result_shape`. For `set1` and `set2` ranked `n`, this
+ * has rank `n` and the same 1st `n-1` dimensions as `set1` and `set2`. The `nth`
+ * dimension contains the result of `set_operation` applied to the corresponding
+ * `[0...n-1]` dimension of `set`.
+ * 
+ * @param <T> data type for {@code resultValues()} output
+ */
+@Operator(group = "sparse")
+public final class DenseToSparseSetOperation<T extends TType> extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.sparse.DenseToSparseSetOperation}
+   */
+  public static class Options {
+    
+    /**
+     * @param validateIndices 
+     */
+    public Options validateIndices(Boolean validateIndices) {
+      this.validateIndices = validateIndices;
+      return this;
+    }
+    
+    private Boolean validateIndices;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new DenseToSparseSetOperation operation.
+   * 
+   * @param scope current scope
+   * @param set1 `Tensor` with rank `n`. 1st `n-1` dimensions must be the same as `set2`.
+   * Dimension `n` contains values in a set, duplicates are allowed but ignored.
+   * @param set2Indices 2D `Tensor`, indices of a `SparseTensor`. Must be in row-major
+   * order.
+   * @param set2Values 1D `Tensor`, values of a `SparseTensor`. Must be in row-major
+   * order.
+   * @param set2Shape 1D `Tensor`, shape of a `SparseTensor`. `set2_shape[0...n-1]` must
+   * be the same as the 1st `n-1` dimensions of `set1`, `result_shape[n]` is the
+   * max set size across `n-1` dimensions.
+   * @param setOperation 
+   * @param options carries optional attributes values
+   * @return a new instance of DenseToSparseSetOperation
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> DenseToSparseSetOperation<T> create(Scope scope, Operand<T> set1, Operand<TInt64> set2Indices, Operand<T> set2Values, Operand<TInt64> set2Shape, String setOperation, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("DenseToSparseSetOperation", scope.makeOpName("DenseToSparseSetOperation"));
+    opBuilder.addInput(set1.asOutput());
+    opBuilder.addInput(set2Indices.asOutput());
+    opBuilder.addInput(set2Values.asOutput());
+    opBuilder.addInput(set2Shape.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("set_operation", setOperation);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.validateIndices != null) {
+          opBuilder.setAttr("validate_indices", opts.validateIndices);
+        }
+      }
+    }
+    return new DenseToSparseSetOperation<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param validateIndices 
+   */
+  public static Options validateIndices(Boolean validateIndices) {
+    return new Options().validateIndices(validateIndices);
+  }
+  
+  /**
+   * 2D indices of a `SparseTensor`.
+   */
+  public Output<TInt64> resultIndices() {
+    return resultIndices;
+  }
+  
+  /**
+   * 1D values of a `SparseTensor`.
+   */
+  public Output<T> resultValues() {
+    return resultValues;
+  }
+  
+  /**
+   * 1D `Tensor` shape of a `SparseTensor`. `result_shape[0...n-1]` is
+   * the same as the 1st `n-1` dimensions of `set1` and `set2`, `result_shape[n]`
+   * is the max result set size across all `0...n-1` dimensions.
+   */
+  public Output<TInt64> resultShape() {
+    return resultShape;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "DenseToSparseSetOperation";
+  
+  private Output<TInt64> resultIndices;
+  private Output<T> resultValues;
+  private Output<TInt64> resultShape;
+  
+  private DenseToSparseSetOperation(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    resultIndices = operation.output(outputIdx++);
+    resultValues = operation.output(outputIdx++);
+    resultShape = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/DeserializeSparse.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/DeserializeSparse.java
new file mode 100644
index 00000000000..871910a0874
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/DeserializeSparse.java
@@ -0,0 +1,131 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.sparse;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Deserialize `SparseTensor` objects.
+ * <p>
+ * The input `serialized_sparse` must have the shape `[?, ?, ..., ?, 3]` where
+ * the last dimension stores serialized `SparseTensor` objects and the other N
+ * dimensions (N >= 0) correspond to a batch. The ranks of the original
+ * `SparseTensor` objects must all match. When the final `SparseTensor` is
+ * created, its rank is the rank of the incoming `SparseTensor` objects plus N;
+ * the sparse tensors have been concatenated along new dimensions, one for each
+ * batch.
+ * <p>
+ * The output `SparseTensor` object's shape values for the original dimensions
+ * are the max across the input `SparseTensor` objects' shape values for the
+ * corresponding dimensions. The new dimensions match the size of the batch.
+ * <p>
+ * The input `SparseTensor` objects' indices are assumed ordered in
+ * standard lexicographic order.  If this is not the case, after this
+ * step run `SparseReorder` to restore index ordering.
+ * <p>
+ * For example, if the serialized input is a `[2 x 3]` matrix representing two
+ * original `SparseTensor` objects:
+ * <p>
+ *     index = [ 0]
+ *             [10]
+ *             [20]
+ *     values = [1, 2, 3]
+ *     shape = [50]
+ * <p>
+ * and
+ * <p>
+ *     index = [ 2]
+ *             [10]
+ *     values = [4, 5]
+ *     shape = [30]
+ * <p>
+ * then the final deserialized `SparseTensor` will be:
+ * <p>
+ *     index = [0  0]
+ *             [0 10]
+ *             [0 20]
+ *             [1  2]
+ *             [1 10]
+ *     values = [1, 2, 3, 4, 5]
+ *     shape = [2 50]
+ * 
+ * @param <U> data type for {@code sparseValues()} output
+ */
+@Operator(group = "sparse")
+public final class DeserializeSparse<U extends TType> extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new DeserializeSparse operation.
+   * 
+   * @param scope current scope
+   * @param serializedSparse The serialized `SparseTensor` objects. The last dimension
+   * must have 3 columns.
+   * @param dtype The `dtype` of the serialized `SparseTensor` objects.
+   * @return a new instance of DeserializeSparse
+   */
+  @Endpoint(describeByClass = true)
+  public static <U extends TType, T extends TType> DeserializeSparse<U> create(Scope scope, Operand<T> serializedSparse, Class<U> dtype) {
+    OperationBuilder opBuilder = scope.env().opBuilder("DeserializeSparse", scope.makeOpName("DeserializeSparse"));
+    opBuilder.addInput(serializedSparse.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("dtype", dtype);
+    return new DeserializeSparse<U>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<TInt64> sparseIndices() {
+    return sparseIndices;
+  }
+  
+  /**
+   */
+  public Output<U> sparseValues() {
+    return sparseValues;
+  }
+  
+  /**
+   */
+  public Output<TInt64> sparseShape() {
+    return sparseShape;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "DeserializeSparse";
+  
+  private Output<TInt64> sparseIndices;
+  private Output<U> sparseValues;
+  private Output<TInt64> sparseShape;
+  
+  private DeserializeSparse(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    sparseIndices = operation.output(outputIdx++);
+    sparseValues = operation.output(outputIdx++);
+    sparseShape = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseAccumulatorApplyGradient.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseAccumulatorApplyGradient.java
new file mode 100644
index 00000000000..328fe0c49ea
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseAccumulatorApplyGradient.java
@@ -0,0 +1,75 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.sparse;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.TString;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Applies a sparse gradient to a given accumulator.
+ * <p>
+ * Does not add if local_step is smaller than the accumulator's
+ * global_step.
+ */
+@Operator(group = "sparse")
+public final class SparseAccumulatorApplyGradient extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new SparseAccumulatorApplyGradient operation.
+   * 
+   * @param scope current scope
+   * @param handle The handle to a accumulator.
+   * @param localStep The local_step value at which the sparse gradient was computed.
+   * @param gradientIndices Indices of the sparse gradient to be accumulated. Must be a
+   * vector.
+   * @param gradientValues Values are the non-zero slices of the gradient, and must have
+   * the same first dimension as indices, i.e., the nnz represented by indices and
+   * values must be consistent.
+   * @param gradientShape Shape of the sparse gradient to be accumulated.
+   * @param hasKnownShape Boolean indicating whether gradient_shape is unknown, in which
+   * case the input is ignored during validation.
+   * @return a new instance of SparseAccumulatorApplyGradient
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> SparseAccumulatorApplyGradient create(Scope scope, Operand<TString> handle, Operand<TInt64> localStep, Operand<TInt64> gradientIndices, Operand<T> gradientValues, Operand<TInt64> gradientShape, Boolean hasKnownShape) {
+    OperationBuilder opBuilder = scope.env().opBuilder("SparseAccumulatorApplyGradient", scope.makeOpName("SparseAccumulatorApplyGradient"));
+    opBuilder.addInput(handle.asOutput());
+    opBuilder.addInput(localStep.asOutput());
+    opBuilder.addInput(gradientIndices.asOutput());
+    opBuilder.addInput(gradientValues.asOutput());
+    opBuilder.addInput(gradientShape.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("has_known_shape", hasKnownShape);
+    return new SparseAccumulatorApplyGradient(opBuilder.build());
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "SparseAccumulatorApplyGradient";
+  
+  private SparseAccumulatorApplyGradient(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseAccumulatorTakeGradient.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseAccumulatorTakeGradient.java
new file mode 100644
index 00000000000..ebd3260f019
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseAccumulatorTakeGradient.java
@@ -0,0 +1,103 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.sparse;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.TString;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Extracts the average sparse gradient in a SparseConditionalAccumulator.
+ * <p>
+ * The op will blocks until sufficient (i.e., more than num_required)
+ * gradients have been accumulated. If the accumulator has already
+ * aggregated more than num_required gradients, it will return its
+ * average of the accumulated gradients.  Also automatically increments
+ * the recorded global_step in the accumulator by 1, and resets the
+ * aggregate to 0.
+ * 
+ * @param <T> data type for {@code values()} output
+ */
+@Operator(group = "sparse")
+public final class SparseAccumulatorTakeGradient<T extends TType> extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new SparseAccumulatorTakeGradient operation.
+   * 
+   * @param scope current scope
+   * @param handle The handle to a SparseConditionalAccumulator.
+   * @param numRequired Number of gradients required before we return an aggregate.
+   * @param dtype The data type of accumulated gradients. Needs to correspond to the type
+   * of the accumulator.
+   * @return a new instance of SparseAccumulatorTakeGradient
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> SparseAccumulatorTakeGradient<T> create(Scope scope, Operand<TString> handle, Operand<TInt32> numRequired, Class<T> dtype) {
+    OperationBuilder opBuilder = scope.env().opBuilder("SparseAccumulatorTakeGradient", scope.makeOpName("SparseAccumulatorTakeGradient"));
+    opBuilder.addInput(handle.asOutput());
+    opBuilder.addInput(numRequired.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("dtype", dtype);
+    return new SparseAccumulatorTakeGradient<T>(opBuilder.build());
+  }
+  
+  /**
+   * Indices of the average of the accumulated sparse gradients.
+   */
+  public Output<TInt64> indices() {
+    return indices;
+  }
+  
+  /**
+   * Values of the average of the accumulated sparse gradients.
+   */
+  public Output<T> values() {
+    return values;
+  }
+  
+  /**
+   * Shape of the average of the accumulated sparse gradients.
+   */
+  public Output<TInt64> shape() {
+    return shape;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "SparseAccumulatorTakeGradient";
+  
+  private Output<TInt64> indices;
+  private Output<T> values;
+  private Output<TInt64> shape;
+  
+  private SparseAccumulatorTakeGradient(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    indices = operation.output(outputIdx++);
+    values = operation.output(outputIdx++);
+    shape = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseAdd.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseAdd.java
new file mode 100644
index 00000000000..bc8beb287bb
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseAdd.java
@@ -0,0 +1,114 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.sparse;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Adds two `SparseTensor` objects to produce another `SparseTensor`.
+ * <p>
+ * The input `SparseTensor` objects' indices are assumed ordered in standard
+ * lexicographic order.  If this is not the case, before this step run
+ * `SparseReorder` to restore index ordering.
+ * <p>
+ * By default, if two values sum to zero at some index, the output `SparseTensor`
+ * would still include that particular location in its index, storing a zero in the
+ * corresponding value slot.  To override this, callers can specify `thresh`,
+ * indicating that if the sum has a magnitude strictly smaller than `thresh`, its
+ * corresponding value and index would then not be included.  In particular,
+ * `thresh == 0` (default) means everything is kept and actual thresholding happens
+ * only for a positive value.
+ * <p>
+ * In the following shapes, `nnz` is the count after taking `thresh` into account.
+ * 
+ * @param <T> data type for {@code sumValues()} output
+ */
+@Operator(group = "sparse")
+public final class SparseAdd<T extends TType> extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new SparseAdd operation.
+   * 
+   * @param scope current scope
+   * @param aIndices 2-D.  The `indices` of the first `SparseTensor`, size `[nnz, ndims]` Matrix.
+   * @param aValues 1-D.  The `values` of the first `SparseTensor`, size `[nnz]` Vector.
+   * @param aShape 1-D.  The `shape` of the first `SparseTensor`, size `[ndims]` Vector.
+   * @param bIndices 2-D.  The `indices` of the second `SparseTensor`, size `[nnz, ndims]` Matrix.
+   * @param bValues 1-D.  The `values` of the second `SparseTensor`, size `[nnz]` Vector.
+   * @param bShape 1-D.  The `shape` of the second `SparseTensor`, size `[ndims]` Vector.
+   * @param thresh 0-D.  The magnitude threshold that determines if an output value/index
+   * pair takes space.
+   * @return a new instance of SparseAdd
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType, U extends TNumber> SparseAdd<T> create(Scope scope, Operand<TInt64> aIndices, Operand<T> aValues, Operand<TInt64> aShape, Operand<TInt64> bIndices, Operand<T> bValues, Operand<TInt64> bShape, Operand<U> thresh) {
+    OperationBuilder opBuilder = scope.env().opBuilder("SparseAdd", scope.makeOpName("SparseAdd"));
+    opBuilder.addInput(aIndices.asOutput());
+    opBuilder.addInput(aValues.asOutput());
+    opBuilder.addInput(aShape.asOutput());
+    opBuilder.addInput(bIndices.asOutput());
+    opBuilder.addInput(bValues.asOutput());
+    opBuilder.addInput(bShape.asOutput());
+    opBuilder.addInput(thresh.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new SparseAdd<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<TInt64> sumIndices() {
+    return sumIndices;
+  }
+  
+  /**
+   */
+  public Output<T> sumValues() {
+    return sumValues;
+  }
+  
+  /**
+   */
+  public Output<TInt64> sumShape() {
+    return sumShape;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "SparseAdd";
+  
+  private Output<TInt64> sumIndices;
+  private Output<T> sumValues;
+  private Output<TInt64> sumShape;
+  
+  private SparseAdd(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    sumIndices = operation.output(outputIdx++);
+    sumValues = operation.output(outputIdx++);
+    sumShape = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseAddGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseAddGrad.java
new file mode 100644
index 00000000000..c3fdf2952a9
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseAddGrad.java
@@ -0,0 +1,95 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.sparse;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
+
+/**
+ * The gradient operator for the SparseAdd op.
+ * <p>
+ * The SparseAdd op calculates A + B, where A, B, and the sum are all represented
+ * as `SparseTensor` objects.  This op takes in the upstream gradient w.r.t.
+ * non-empty values of the sum, and outputs the gradients w.r.t. the non-empty
+ * values of A and B.
+ * 
+ * @param <T> data type for {@code aValGrad()} output
+ */
+@Operator(group = "sparse")
+public final class SparseAddGrad<T extends TType> extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new SparseAddGrad operation.
+   * 
+   * @param scope current scope
+   * @param backpropValGrad 1-D with shape `[nnz(sum)]`.  The gradient with respect to
+   * the non-empty values of the sum.
+   * @param aIndices 2-D.  The `indices` of the `SparseTensor` A, size `[nnz(A), ndims]`.
+   * @param bIndices 2-D.  The `indices` of the `SparseTensor` B, size `[nnz(B), ndims]`.
+   * @param sumIndices 2-D.  The `indices` of the sum `SparseTensor`, size
+   * `[nnz(sum), ndims]`.
+   * @return a new instance of SparseAddGrad
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> SparseAddGrad<T> create(Scope scope, Operand<T> backpropValGrad, Operand<TInt64> aIndices, Operand<TInt64> bIndices, Operand<TInt64> sumIndices) {
+    OperationBuilder opBuilder = scope.env().opBuilder("SparseAddGrad", scope.makeOpName("SparseAddGrad"));
+    opBuilder.addInput(backpropValGrad.asOutput());
+    opBuilder.addInput(aIndices.asOutput());
+    opBuilder.addInput(bIndices.asOutput());
+    opBuilder.addInput(sumIndices.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new SparseAddGrad<T>(opBuilder.build());
+  }
+  
+  /**
+   * 1-D with shape `[nnz(A)]`. The gradient with respect to the
+   * non-empty values of A.
+   */
+  public Output<T> aValGrad() {
+    return aValGrad;
+  }
+  
+  /**
+   * 1-D with shape `[nnz(B)]`. The gradient with respect to the
+   * non-empty values of B.
+   */
+  public Output<T> bValGrad() {
+    return bValGrad;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "SparseAddGrad";
+  
+  private Output<T> aValGrad;
+  private Output<T> bValGrad;
+  
+  private SparseAddGrad(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    aValGrad = operation.output(outputIdx++);
+    bValGrad = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseBincount.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseBincount.java
new file mode 100644
index 00000000000..81d72b8a81e
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseBincount.java
@@ -0,0 +1,129 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.sparse;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Counts the number of occurrences of each value in an integer array.
+ * <p>
+ * Outputs a vector with length `size` and the same dtype as `weights`. If
+ * `weights` are empty, then index `i` stores the number of times the value `i` is
+ * counted in `arr`. If `weights` are non-empty, then index `i` stores the sum of
+ * the value in `weights` at each index where the corresponding value in `arr` is
+ * `i`.
+ * <p>
+ * Values in `arr` outside of the range [0, size) are ignored.
+ * 
+ * @param <U> data type for {@code output()} output
+ */
+@Operator(group = "sparse")
+public final class SparseBincount<U extends TNumber> extends RawOp implements Operand<U> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.sparse.SparseBincount}
+   */
+  public static class Options {
+    
+    /**
+     * @param binaryOutput bool; Whether the kernel should count the appearance or number of occurrences.
+     */
+    public Options binaryOutput(Boolean binaryOutput) {
+      this.binaryOutput = binaryOutput;
+      return this;
+    }
+    
+    private Boolean binaryOutput;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new SparseBincount operation.
+   * 
+   * @param scope current scope
+   * @param indices 2D int64 `Tensor`.
+   * @param values 1D int `Tensor`.
+   * @param denseShape 1D int64 `Tensor`.
+   * @param size non-negative int scalar `Tensor`.
+   * @param weights is an int32, int64, float32, or float64 `Tensor` with the same
+   * shape as `input`, or a length-0 `Tensor`, in which case it acts as all weights
+   * equal to 1.
+   * @param options carries optional attributes values
+   * @return a new instance of SparseBincount
+   */
+  @Endpoint(describeByClass = true)
+  public static <U extends TNumber, T extends TNumber> SparseBincount<U> create(Scope scope, Operand<TInt64> indices, Operand<T> values, Operand<TInt64> denseShape, Operand<T> size, Operand<U> weights, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("SparseBincount", scope.makeOpName("SparseBincount"));
+    opBuilder.addInput(indices.asOutput());
+    opBuilder.addInput(values.asOutput());
+    opBuilder.addInput(denseShape.asOutput());
+    opBuilder.addInput(size.asOutput());
+    opBuilder.addInput(weights.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.binaryOutput != null) {
+          opBuilder.setAttr("binary_output", opts.binaryOutput);
+        }
+      }
+    }
+    return new SparseBincount<U>(opBuilder.build());
+  }
+  
+  /**
+   * @param binaryOutput bool; Whether the kernel should count the appearance or number of occurrences.
+   */
+  public static Options binaryOutput(Boolean binaryOutput) {
+    return new Options().binaryOutput(binaryOutput);
+  }
+  
+  /**
+   * 1D `Tensor` with length equal to `size` or 2D `Tensor` with [batch_size, `size`].
+   * The counts or summed weights for each value in the range [0, size).
+   */
+  public Output<U> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<U> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "SparseBincount";
+  
+  private Output<U> output;
+  
+  private SparseBincount(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseConcat.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseConcat.java
new file mode 100644
index 00000000000..c6866997e30
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseConcat.java
@@ -0,0 +1,139 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.sparse;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.Operands;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Concatenates a list of `SparseTensor` along the specified dimension.
+ * <p>
+ * Concatenation is with respect to the dense versions of these sparse tensors.
+ * It is assumed that each input is a `SparseTensor` whose elements are ordered
+ * along increasing dimension number.
+ * <p>
+ * All inputs' shapes must match, except for the concat dimension.  The
+ * `indices`, `values`, and `shapes` lists must have the same length.
+ * <p>
+ * The output shape is identical to the inputs', except along the concat
+ * dimension, where it is the sum of the inputs' sizes along that dimension.
+ * <p>
+ * The output elements will be resorted to preserve the sort order along
+ * increasing dimension number.
+ * <p>
+ * This op runs in `O(M log M)` time, where `M` is the total number of non-empty
+ * values across all inputs. This is due to the need for an internal sort in
+ * order to concatenate efficiently across an arbitrary dimension.
+ * <p>
+ * For example, if `concat_dim = 1` and the inputs are
+ * <p>
+ *     sp_inputs[0]: shape = [2, 3]
+ *     [0, 2]: "a"
+ *     [1, 0]: "b"
+ *     [1, 1]: "c"
+ * <p>
+ *     sp_inputs[1]: shape = [2, 4]
+ *     [0, 1]: "d"
+ *     [0, 2]: "e"
+ * <p>
+ * then the output will be
+ * <p>
+ *     shape = [2, 7]
+ *     [0, 2]: "a"
+ *     [0, 4]: "d"
+ *     [0, 5]: "e"
+ *     [1, 0]: "b"
+ *     [1, 1]: "c"
+ * <p>
+ * Graphically this is equivalent to doing
+ * <p>
+ *     [    a] concat [  d e  ] = [    a   d e  ]
+ *     [b c  ]        [       ]   [b c          ]
+ * 
+ * @param <T> data type for {@code outputValues()} output
+ */
+@Operator(group = "sparse")
+public final class SparseConcat<T extends TType> extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new SparseConcat operation.
+   * 
+   * @param scope current scope
+   * @param indices 2-D.  Indices of each input `SparseTensor`.
+   * @param values 1-D.  Non-empty values of each `SparseTensor`.
+   * @param shapes 1-D.  Shapes of each `SparseTensor`.
+   * @param concatDim Dimension to concatenate along. Must be in range [-rank, rank),
+   * where rank is the number of dimensions in each input `SparseTensor`.
+   * @return a new instance of SparseConcat
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> SparseConcat<T> create(Scope scope, Iterable<Operand<TInt64>> indices, Iterable<Operand<T>> values, Iterable<Operand<TInt64>> shapes, Long concatDim) {
+    OperationBuilder opBuilder = scope.env().opBuilder("SparseConcat", scope.makeOpName("SparseConcat"));
+    opBuilder.addInputList(Operands.asOutputs(indices));
+    opBuilder.addInputList(Operands.asOutputs(values));
+    opBuilder.addInputList(Operands.asOutputs(shapes));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("concat_dim", concatDim);
+    return new SparseConcat<T>(opBuilder.build());
+  }
+  
+  /**
+   * 2-D.  Indices of the concatenated `SparseTensor`.
+   */
+  public Output<TInt64> outputIndices() {
+    return outputIndices;
+  }
+  
+  /**
+   * 1-D.  Non-empty values of the concatenated `SparseTensor`.
+   */
+  public Output<T> outputValues() {
+    return outputValues;
+  }
+  
+  /**
+   * 1-D.  Shape of the concatenated `SparseTensor`.
+   */
+  public Output<TInt64> outputShape() {
+    return outputShape;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "SparseConcat";
+  
+  private Output<TInt64> outputIndices;
+  private Output<T> outputValues;
+  private Output<TInt64> outputShape;
+  
+  private SparseConcat(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    outputIndices = operation.output(outputIdx++);
+    outputValues = operation.output(outputIdx++);
+    outputShape = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseConditionalAccumulator.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseConditionalAccumulator.java
new file mode 100644
index 00000000000..863b2d0b8e3
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseConditionalAccumulator.java
@@ -0,0 +1,160 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.sparse;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TString;
+import org.tensorflow.types.family.TType;
+
+/**
+ * A conditional accumulator for aggregating sparse gradients.
+ * <p>
+ * The accumulator accepts gradients marked with local_step greater or
+ * equal to the most recent global_step known to the accumulator. The
+ * average can be extracted from the accumulator, provided sufficient
+ * gradients have been accumulated. Extracting the average automatically
+ * resets the aggregate to 0, and increments the global_step recorded by
+ * the accumulator.
+ */
+@Operator(group = "sparse")
+public final class SparseConditionalAccumulator extends RawOp implements Operand<TString> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.sparse.SparseConditionalAccumulator}
+   */
+  public static class Options {
+    
+    /**
+     * @param container If non-empty, this accumulator is placed in the given container.
+     * Otherwise, a default container is used.
+     */
+    public Options container(String container) {
+      this.container = container;
+      return this;
+    }
+    
+    /**
+     * @param sharedName If non-empty, this accumulator will be shared under the given name
+     * across multiple sessions.
+     */
+    public Options sharedName(String sharedName) {
+      this.sharedName = sharedName;
+      return this;
+    }
+    
+    /**
+     * @param reductionType 
+     */
+    public Options reductionType(String reductionType) {
+      this.reductionType = reductionType;
+      return this;
+    }
+    
+    private String container;
+    private String sharedName;
+    private String reductionType;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new SparseConditionalAccumulator operation.
+   * 
+   * @param scope current scope
+   * @param dtype The type of the value being accumulated.
+   * @param shape The shape of the values.
+   * @param options carries optional attributes values
+   * @return a new instance of SparseConditionalAccumulator
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> SparseConditionalAccumulator create(Scope scope, Class<T> dtype, Shape shape, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("SparseConditionalAccumulator", scope.makeOpName("SparseConditionalAccumulator"));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("dtype", dtype);
+    opBuilder.setAttr("shape", shape);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.container != null) {
+          opBuilder.setAttr("container", opts.container);
+        }
+        if (opts.sharedName != null) {
+          opBuilder.setAttr("shared_name", opts.sharedName);
+        }
+        if (opts.reductionType != null) {
+          opBuilder.setAttr("reduction_type", opts.reductionType);
+        }
+      }
+    }
+    return new SparseConditionalAccumulator(opBuilder.build());
+  }
+  
+  /**
+   * @param container If non-empty, this accumulator is placed in the given container.
+   * Otherwise, a default container is used.
+   */
+  public static Options container(String container) {
+    return new Options().container(container);
+  }
+  
+  /**
+   * @param sharedName If non-empty, this accumulator will be shared under the given name
+   * across multiple sessions.
+   */
+  public static Options sharedName(String sharedName) {
+    return new Options().sharedName(sharedName);
+  }
+  
+  /**
+   * @param reductionType 
+   */
+  public static Options reductionType(String reductionType) {
+    return new Options().reductionType(reductionType);
+  }
+  
+  /**
+   * The handle to the accumulator.
+   */
+  public Output<TString> handle() {
+    return handle;
+  }
+  
+  @Override
+  public Output<TString> asOutput() {
+    return handle;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "SparseConditionalAccumulator";
+  
+  private Output<TString> handle;
+  
+  private SparseConditionalAccumulator(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    handle = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseCountSparseOutput.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseCountSparseOutput.java
new file mode 100644
index 00000000000..9f68897c075
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseCountSparseOutput.java
@@ -0,0 +1,152 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.sparse;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Performs sparse-output bin counting for a sparse tensor input.
+ * <p>
+ *   Counts the number of times each value occurs in the input.
+ * 
+ * @param <U> data type for {@code outputValues()} output
+ */
+public final class SparseCountSparseOutput<U extends TNumber> extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.sparse.SparseCountSparseOutput}
+   */
+  public static class Options {
+    
+    /**
+     * @param minlength Minimum value to count. Can be set to -1 for no minimum.
+     */
+    public Options minlength(Long minlength) {
+      this.minlength = minlength;
+      return this;
+    }
+    
+    /**
+     * @param maxlength Maximum value to count. Can be set to -1 for no maximum.
+     */
+    public Options maxlength(Long maxlength) {
+      this.maxlength = maxlength;
+      return this;
+    }
+    
+    private Long minlength;
+    private Long maxlength;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new SparseCountSparseOutput operation.
+   * 
+   * @param scope current scope
+   * @param indices Tensor containing the indices of the sparse tensor to count.
+   * @param values Tensor containing values of the sparse tensor to count.
+   * @param denseShape Tensor containing the dense shape of the sparse tensor to count.
+   * @param weights A Tensor of the same shape as indices containing per-index weight values.
+   * May also be the empty tensor if no weights are used.
+   * @param binaryOutput Whether to output the number of occurrences of each value or 1.
+   * @param options carries optional attributes values
+   * @return a new instance of SparseCountSparseOutput
+   */
+  @Endpoint(describeByClass = true)
+  public static <U extends TNumber, T extends TNumber> SparseCountSparseOutput<U> create(Scope scope, Operand<TInt64> indices, Operand<T> values, Operand<TInt64> denseShape, Operand<U> weights, Boolean binaryOutput, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("SparseCountSparseOutput", scope.makeOpName("SparseCountSparseOutput"));
+    opBuilder.addInput(indices.asOutput());
+    opBuilder.addInput(values.asOutput());
+    opBuilder.addInput(denseShape.asOutput());
+    opBuilder.addInput(weights.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("binary_output", binaryOutput);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.minlength != null) {
+          opBuilder.setAttr("minlength", opts.minlength);
+        }
+        if (opts.maxlength != null) {
+          opBuilder.setAttr("maxlength", opts.maxlength);
+        }
+      }
+    }
+    return new SparseCountSparseOutput<U>(opBuilder.build());
+  }
+  
+  /**
+   * @param minlength Minimum value to count. Can be set to -1 for no minimum.
+   */
+  public static Options minlength(Long minlength) {
+    return new Options().minlength(minlength);
+  }
+  
+  /**
+   * @param maxlength Maximum value to count. Can be set to -1 for no maximum.
+   */
+  public static Options maxlength(Long maxlength) {
+    return new Options().maxlength(maxlength);
+  }
+  
+  /**
+   * Indices tensor for the resulting sparse tensor object.
+   */
+  public Output<TInt64> outputIndices() {
+    return outputIndices;
+  }
+  
+  /**
+   * Values tensor for the resulting sparse tensor object.
+   */
+  public Output<U> outputValues() {
+    return outputValues;
+  }
+  
+  /**
+   * Shape tensor for the resulting sparse tensor object.
+   */
+  public Output<TInt64> outputDenseShape() {
+    return outputDenseShape;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "SparseCountSparseOutput";
+  
+  private Output<TInt64> outputIndices;
+  private Output<U> outputValues;
+  private Output<TInt64> outputDenseShape;
+  
+  private SparseCountSparseOutput(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    outputIndices = operation.output(outputIdx++);
+    outputValues = operation.output(outputIdx++);
+    outputDenseShape = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseCross.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseCross.java
new file mode 100644
index 00000000000..1cd471349c2
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseCross.java
@@ -0,0 +1,134 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.sparse;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.Operands;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.TString;
+
+/**
+ * Generates sparse cross from a list of sparse and dense tensors.
+ * <p>
+ * The op takes two lists, one of 2D `SparseTensor` and one of 2D `Tensor`, each
+ * representing features of one feature column. It outputs a 2D `SparseTensor` with
+ * the batchwise crosses of these features.
+ * <p>
+ * For example, if the inputs are
+ * <p>
+ *     inputs[0]: SparseTensor with shape = [2, 2]
+ *     [0, 0]: "a"
+ *     [1, 0]: "b"
+ *     [1, 1]: "c"
+ * <p>
+ *     inputs[1]: SparseTensor with shape = [2, 1]
+ *     [0, 0]: "d"
+ *     [1, 0]: "e"
+ * <p>
+ *     inputs[2]: Tensor [["f"], ["g"]]
+ * <p>
+ * then the output will be
+ * <p>
+ *     shape = [2, 2]
+ *     [0, 0]: "a_X_d_X_f"
+ *     [1, 0]: "b_X_e_X_g"
+ *     [1, 1]: "c_X_e_X_g"
+ * <p>
+ * if hashed_output=true then the output will be
+ * <p>
+ *     shape = [2, 2]
+ *     [0, 0]: FingerprintCat64(
+ *                 Fingerprint64("f"), FingerprintCat64(
+ *                     Fingerprint64("d"), Fingerprint64("a")))
+ *     [1, 0]: FingerprintCat64(
+ *                 Fingerprint64("g"), FingerprintCat64(
+ *                     Fingerprint64("e"), Fingerprint64("b")))
+ *     [1, 1]: FingerprintCat64(
+ *                 Fingerprint64("g"), FingerprintCat64(
+ *                     Fingerprint64("e"), Fingerprint64("c")))
+ */
+@Operator(group = "sparse")
+public final class SparseCross extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new SparseCross operation.
+   * 
+   * @param scope current scope
+   * @param indices 2-D.  Indices of each input `SparseTensor`.
+   * @param values 1-D.   values of each `SparseTensor`.
+   * @param shapes 1-D.   Shapes of each `SparseTensor`.
+   * @param denseInputs 2-D.    Columns represented by dense `Tensor`.
+   * @param sep string used when joining a list of string inputs, can be used as separator later.
+   * @return a new instance of SparseCross
+   */
+  @Endpoint(describeByClass = true)
+  public static SparseCross create(Scope scope, Iterable<Operand<TInt64>> indices, Iterable<Operand<?>> values, Iterable<Operand<TInt64>> shapes, Iterable<Operand<?>> denseInputs, Operand<TString> sep) {
+    OperationBuilder opBuilder = scope.env().opBuilder("SparseCrossV2", scope.makeOpName("SparseCross"));
+    opBuilder.addInputList(Operands.asOutputs(indices));
+    opBuilder.addInputList(Operands.asOutputs(values));
+    opBuilder.addInputList(Operands.asOutputs(shapes));
+    opBuilder.addInputList(Operands.asOutputs(denseInputs));
+    opBuilder.addInput(sep.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new SparseCross(opBuilder.build());
+  }
+  
+  /**
+   * 2-D.  Indices of the concatenated `SparseTensor`.
+   */
+  public Output<TInt64> outputIndices() {
+    return outputIndices;
+  }
+  
+  /**
+   * 1-D.  Non-empty values of the concatenated or hashed
+   * `SparseTensor`.
+   */
+  public Output<TString> outputValues() {
+    return outputValues;
+  }
+  
+  /**
+   * 1-D.  Shape of the concatenated `SparseTensor`.
+   */
+  public Output<TInt64> outputShape() {
+    return outputShape;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "SparseCrossV2";
+  
+  private Output<TInt64> outputIndices;
+  private Output<TString> outputValues;
+  private Output<TInt64> outputShape;
+  
+  private SparseCross(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    outputIndices = operation.output(outputIdx++);
+    outputValues = operation.output(outputIdx++);
+    outputShape = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseCrossHashed.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseCrossHashed.java
new file mode 100644
index 00000000000..2fc6976079e
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseCrossHashed.java
@@ -0,0 +1,139 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.sparse;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.Operands;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TBool;
+import org.tensorflow.types.TInt64;
+
+/**
+ * Generates sparse cross from a list of sparse and dense tensors.
+ * <p>
+ * The op takes two lists, one of 2D `SparseTensor` and one of 2D `Tensor`, each
+ * representing features of one feature column. It outputs a 2D `SparseTensor` with
+ * the batchwise crosses of these features.
+ * <p>
+ * For example, if the inputs are
+ * <p>
+ *     inputs[0]: SparseTensor with shape = [2, 2]
+ *     [0, 0]: "a"
+ *     [1, 0]: "b"
+ *     [1, 1]: "c"
+ * <p>
+ *     inputs[1]: SparseTensor with shape = [2, 1]
+ *     [0, 0]: "d"
+ *     [1, 0]: "e"
+ * <p>
+ *     inputs[2]: Tensor [["f"], ["g"]]
+ * <p>
+ * then the output will be
+ * <p>
+ *     shape = [2, 2]
+ *     [0, 0]: "a_X_d_X_f"
+ *     [1, 0]: "b_X_e_X_g"
+ *     [1, 1]: "c_X_e_X_g"
+ * <p>
+ * if hashed_output=true then the output will be
+ * <p>
+ *     shape = [2, 2]
+ *     [0, 0]: FingerprintCat64(
+ *                 Fingerprint64("f"), FingerprintCat64(
+ *                     Fingerprint64("d"), Fingerprint64("a")))
+ *     [1, 0]: FingerprintCat64(
+ *                 Fingerprint64("g"), FingerprintCat64(
+ *                     Fingerprint64("e"), Fingerprint64("b")))
+ *     [1, 1]: FingerprintCat64(
+ *                 Fingerprint64("g"), FingerprintCat64(
+ *                     Fingerprint64("e"), Fingerprint64("c")))
+ */
+@Operator(group = "sparse")
+public final class SparseCrossHashed extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new SparseCrossHashed operation.
+   * 
+   * @param scope current scope
+   * @param indices 2-D.  Indices of each input `SparseTensor`.
+   * @param values 1-D.   values of each `SparseTensor`.
+   * @param shapes 1-D.   Shapes of each `SparseTensor`.
+   * @param denseInputs 2-D.    Columns represented by dense `Tensor`.
+   * @param numBuckets It is used if hashed_output is true.
+   * output = hashed_value%num_buckets if num_buckets > 0 else hashed_value.
+   * @param strongHash boolean, if true, siphash with salt will be used instead of farmhash.
+   * @param salt Specify the salt that will be used by the siphash function.
+   * @return a new instance of SparseCrossHashed
+   */
+  @Endpoint(describeByClass = true)
+  public static SparseCrossHashed create(Scope scope, Iterable<Operand<TInt64>> indices, Iterable<Operand<?>> values, Iterable<Operand<TInt64>> shapes, Iterable<Operand<?>> denseInputs, Operand<TInt64> numBuckets, Operand<TBool> strongHash, Operand<TInt64> salt) {
+    OperationBuilder opBuilder = scope.env().opBuilder("SparseCrossHashed", scope.makeOpName("SparseCrossHashed"));
+    opBuilder.addInputList(Operands.asOutputs(indices));
+    opBuilder.addInputList(Operands.asOutputs(values));
+    opBuilder.addInputList(Operands.asOutputs(shapes));
+    opBuilder.addInputList(Operands.asOutputs(denseInputs));
+    opBuilder.addInput(numBuckets.asOutput());
+    opBuilder.addInput(strongHash.asOutput());
+    opBuilder.addInput(salt.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new SparseCrossHashed(opBuilder.build());
+  }
+  
+  /**
+   * 2-D.  Indices of the concatenated `SparseTensor`.
+   */
+  public Output<TInt64> outputIndices() {
+    return outputIndices;
+  }
+  
+  /**
+   * 1-D.  Non-empty values of the concatenated or hashed
+   * `SparseTensor`.
+   */
+  public Output<TInt64> outputValues() {
+    return outputValues;
+  }
+  
+  /**
+   * 1-D.  Shape of the concatenated `SparseTensor`.
+   */
+  public Output<TInt64> outputShape() {
+    return outputShape;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "SparseCrossHashed";
+  
+  private Output<TInt64> outputIndices;
+  private Output<TInt64> outputValues;
+  private Output<TInt64> outputShape;
+  
+  private SparseCrossHashed(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    outputIndices = operation.output(outputIdx++);
+    outputValues = operation.output(outputIdx++);
+    outputShape = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseDenseCwiseAdd.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseDenseCwiseAdd.java
new file mode 100644
index 00000000000..40c5fc1446b
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseDenseCwiseAdd.java
@@ -0,0 +1,92 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.sparse;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Adds up a SparseTensor and a dense Tensor, using these special rules:
+ * <p>
+ * (1) Broadcasts the dense side to have the same shape as the sparse side, if
+ *     eligible;
+ * (2) Then, only the dense values pointed to by the indices of the SparseTensor
+ *     participate in the cwise addition.
+ * <p>
+ * By these rules, the result is a logical SparseTensor with exactly the same
+ * indices and shape, but possibly with different non-zero values.  The output of
+ * this Op is the resultant non-zero values.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator(group = "sparse")
+public final class SparseDenseCwiseAdd<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new SparseDenseCwiseAdd operation.
+   * 
+   * @param scope current scope
+   * @param spIndices 2-D.  `N x R` matrix with the indices of non-empty values in a
+   * SparseTensor, possibly not in canonical ordering.
+   * @param spValues 1-D.  `N` non-empty values corresponding to `sp_indices`.
+   * @param spShape 1-D.  Shape of the input SparseTensor.
+   * @param dense `R`-D.  The dense Tensor operand.
+   * @return a new instance of SparseDenseCwiseAdd
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> SparseDenseCwiseAdd<T> create(Scope scope, Operand<TInt64> spIndices, Operand<T> spValues, Operand<TInt64> spShape, Operand<T> dense) {
+    OperationBuilder opBuilder = scope.env().opBuilder("SparseDenseCwiseAdd", scope.makeOpName("SparseDenseCwiseAdd"));
+    opBuilder.addInput(spIndices.asOutput());
+    opBuilder.addInput(spValues.asOutput());
+    opBuilder.addInput(spShape.asOutput());
+    opBuilder.addInput(dense.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new SparseDenseCwiseAdd<T>(opBuilder.build());
+  }
+  
+  /**
+   * 1-D.  The `N` values that are operated on.
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "SparseDenseCwiseAdd";
+  
+  private Output<T> output;
+  
+  private SparseDenseCwiseAdd(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseDenseCwiseDiv.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseDenseCwiseDiv.java
new file mode 100644
index 00000000000..1d88f78e482
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseDenseCwiseDiv.java
@@ -0,0 +1,86 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.sparse;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Component-wise divides a SparseTensor by a dense Tensor.
+ * <p>
+ * <i>Limitation</i>: this Op only broadcasts the dense side to the sparse side, but not
+ * the other direction.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator(group = "sparse")
+public final class SparseDenseCwiseDiv<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new SparseDenseCwiseDiv operation.
+   * 
+   * @param scope current scope
+   * @param spIndices 2-D.  `N x R` matrix with the indices of non-empty values in a
+   * SparseTensor, possibly not in canonical ordering.
+   * @param spValues 1-D.  `N` non-empty values corresponding to `sp_indices`.
+   * @param spShape 1-D.  Shape of the input SparseTensor.
+   * @param dense `R`-D.  The dense Tensor operand.
+   * @return a new instance of SparseDenseCwiseDiv
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> SparseDenseCwiseDiv<T> create(Scope scope, Operand<TInt64> spIndices, Operand<T> spValues, Operand<TInt64> spShape, Operand<T> dense) {
+    OperationBuilder opBuilder = scope.env().opBuilder("SparseDenseCwiseDiv", scope.makeOpName("SparseDenseCwiseDiv"));
+    opBuilder.addInput(spIndices.asOutput());
+    opBuilder.addInput(spValues.asOutput());
+    opBuilder.addInput(spShape.asOutput());
+    opBuilder.addInput(dense.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new SparseDenseCwiseDiv<T>(opBuilder.build());
+  }
+  
+  /**
+   * 1-D.  The `N` values that are operated on.
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "SparseDenseCwiseDiv";
+  
+  private Output<T> output;
+  
+  private SparseDenseCwiseDiv(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseDenseCwiseMul.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseDenseCwiseMul.java
new file mode 100644
index 00000000000..b0144601eb3
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseDenseCwiseMul.java
@@ -0,0 +1,90 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.sparse;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Component-wise multiplies a SparseTensor by a dense Tensor.
+ * <p>
+ * The output locations corresponding to the implicitly zero elements in the sparse
+ * tensor will be zero (i.e., will not take up storage space), regardless of the
+ * contents of the dense tensor (even if it's +/-INF and that INF<i>0 == NaN).
+ * <p>
+ * </i>Limitation*: this Op only broadcasts the dense side to the sparse side, but not
+ * the other direction.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator(group = "sparse")
+public final class SparseDenseCwiseMul<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new SparseDenseCwiseMul operation.
+   * 
+   * @param scope current scope
+   * @param spIndices 2-D.  `N x R` matrix with the indices of non-empty values in a
+   * SparseTensor, possibly not in canonical ordering.
+   * @param spValues 1-D.  `N` non-empty values corresponding to `sp_indices`.
+   * @param spShape 1-D.  Shape of the input SparseTensor.
+   * @param dense `R`-D.  The dense Tensor operand.
+   * @return a new instance of SparseDenseCwiseMul
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> SparseDenseCwiseMul<T> create(Scope scope, Operand<TInt64> spIndices, Operand<T> spValues, Operand<TInt64> spShape, Operand<T> dense) {
+    OperationBuilder opBuilder = scope.env().opBuilder("SparseDenseCwiseMul", scope.makeOpName("SparseDenseCwiseMul"));
+    opBuilder.addInput(spIndices.asOutput());
+    opBuilder.addInput(spValues.asOutput());
+    opBuilder.addInput(spShape.asOutput());
+    opBuilder.addInput(dense.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new SparseDenseCwiseMul<T>(opBuilder.build());
+  }
+  
+  /**
+   * 1-D.  The `N` values that are operated on.
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "SparseDenseCwiseMul";
+  
+  private Output<T> output;
+  
+  private SparseDenseCwiseMul(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseFillEmptyRows.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseFillEmptyRows.java
new file mode 100644
index 00000000000..6c8066061fc
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseFillEmptyRows.java
@@ -0,0 +1,144 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.sparse;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TBool;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Fills empty rows in the input 2-D `SparseTensor` with a default value.
+ * <p>
+ * The input `SparseTensor` is represented via the tuple of inputs
+ * (`indices`, `values`, `dense_shape`).  The output `SparseTensor` has the
+ * same `dense_shape` but with indices `output_indices` and values
+ * `output_values`.
+ * <p>
+ * This op inserts a single entry for every row that doesn't have any values.
+ * The index is created as `[row, 0, ..., 0]` and the inserted value
+ * is `default_value`.
+ * <p>
+ * For example, suppose `sp_input` has shape `[5, 6]` and non-empty values:
+ * <p>
+ *     [0, 1]: a
+ *     [0, 3]: b
+ *     [2, 0]: c
+ *     [3, 1]: d
+ * <p>
+ * Rows 1 and 4 are empty, so the output will be of shape `[5, 6]` with values:
+ * <p>
+ *     [0, 1]: a
+ *     [0, 3]: b
+ *     [1, 0]: default_value
+ *     [2, 0]: c
+ *     [3, 1]: d
+ *     [4, 0]: default_value
+ * <p>
+ * The output `SparseTensor` will be in row-major order and will have the
+ * same shape as the input.
+ * <p>
+ * This op also returns an indicator vector shaped `[dense_shape[0]]` such that
+ * <p>
+ *     empty_row_indicator[i] = True iff row i was an empty row.
+ * <p>
+ * And a reverse index map vector shaped `[indices.shape[0]]` that is used during
+ * backpropagation,
+ * <p>
+ *     reverse_index_map[j] = out_j s.t. indices[j, :] == output_indices[out_j, :]
+ * 
+ * @param <T> data type for {@code outputValues()} output
+ */
+@Operator(group = "sparse")
+public final class SparseFillEmptyRows<T extends TType> extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new SparseFillEmptyRows operation.
+   * 
+   * @param scope current scope
+   * @param indices 2-D. the indices of the sparse tensor.
+   * @param values 1-D. the values of the sparse tensor.
+   * @param denseShape 1-D. the shape of the sparse tensor.
+   * @param defaultValue 0-D. default value to insert into location `[row, 0, ..., 0]`
+   *   for rows missing from the input sparse tensor.
+   * output indices: 2-D. the indices of the filled sparse tensor.
+   * @return a new instance of SparseFillEmptyRows
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> SparseFillEmptyRows<T> create(Scope scope, Operand<TInt64> indices, Operand<T> values, Operand<TInt64> denseShape, Operand<T> defaultValue) {
+    OperationBuilder opBuilder = scope.env().opBuilder("SparseFillEmptyRows", scope.makeOpName("SparseFillEmptyRows"));
+    opBuilder.addInput(indices.asOutput());
+    opBuilder.addInput(values.asOutput());
+    opBuilder.addInput(denseShape.asOutput());
+    opBuilder.addInput(defaultValue.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new SparseFillEmptyRows<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<TInt64> outputIndices() {
+    return outputIndices;
+  }
+  
+  /**
+   * 1-D. the values of the filled sparse tensor.
+   */
+  public Output<T> outputValues() {
+    return outputValues;
+  }
+  
+  /**
+   * 1-D. whether the dense row was missing in the
+   * input sparse tensor.
+   */
+  public Output<TBool> emptyRowIndicator() {
+    return emptyRowIndicator;
+  }
+  
+  /**
+   * 1-D. a map from the input indices to the output indices.
+   */
+  public Output<TInt64> reverseIndexMap() {
+    return reverseIndexMap;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "SparseFillEmptyRows";
+  
+  private Output<TInt64> outputIndices;
+  private Output<T> outputValues;
+  private Output<TBool> emptyRowIndicator;
+  private Output<TInt64> reverseIndexMap;
+  
+  private SparseFillEmptyRows(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    outputIndices = operation.output(outputIdx++);
+    outputValues = operation.output(outputIdx++);
+    emptyRowIndicator = operation.output(outputIdx++);
+    reverseIndexMap = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseFillEmptyRowsGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseFillEmptyRowsGrad.java
new file mode 100644
index 00000000000..3942e8bc8b3
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseFillEmptyRowsGrad.java
@@ -0,0 +1,91 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.sparse;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
+
+/**
+ * The gradient of SparseFillEmptyRows.
+ * <p>
+ * Takes vectors reverse_index_map, shaped `[N]`, and grad_values,
+ * shaped `[N_full]`, where `N_full >= N` and copies data into either
+ * `d_values` or `d_default_value`.  Here `d_values` is shaped `[N]` and
+ * `d_default_value` is a scalar.
+ * <p>
+ *   d_values[j] = grad_values[reverse_index_map[j]]
+ *   d_default_value = sum_{k : 0 .. N_full - 1} (
+ *      grad_values[k] * 1{k not in reverse_index_map})
+ * 
+ * @param <T> data type for {@code dValues()} output
+ */
+@Operator(group = "sparse")
+public final class SparseFillEmptyRowsGrad<T extends TType> extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new SparseFillEmptyRowsGrad operation.
+   * 
+   * @param scope current scope
+   * @param reverseIndexMap 1-D.  The reverse index map from SparseFillEmptyRows.
+   * @param gradValues 1-D.  The gradients from backprop.
+   * @return a new instance of SparseFillEmptyRowsGrad
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> SparseFillEmptyRowsGrad<T> create(Scope scope, Operand<TInt64> reverseIndexMap, Operand<T> gradValues) {
+    OperationBuilder opBuilder = scope.env().opBuilder("SparseFillEmptyRowsGrad", scope.makeOpName("SparseFillEmptyRowsGrad"));
+    opBuilder.addInput(reverseIndexMap.asOutput());
+    opBuilder.addInput(gradValues.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new SparseFillEmptyRowsGrad<T>(opBuilder.build());
+  }
+  
+  /**
+   * 1-D.  The backprop into values.
+   */
+  public Output<T> dValues() {
+    return dValues;
+  }
+  
+  /**
+   * 0-D.  The backprop into default_value.
+   */
+  public Output<T> dDefaultValue() {
+    return dDefaultValue;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "SparseFillEmptyRowsGrad";
+  
+  private Output<T> dValues;
+  private Output<T> dDefaultValue;
+  
+  private SparseFillEmptyRowsGrad(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    dValues = operation.output(outputIdx++);
+    dDefaultValue = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseMatMul.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseMatMul.java
new file mode 100644
index 00000000000..fc66887070f
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseMatMul.java
@@ -0,0 +1,176 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.sparse;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Multiply matrix "a" by matrix "b".
+ * <p>
+ * The inputs must be two-dimensional matrices and the inner dimension of "a" must
+ * match the outer dimension of "b". Both "a" and "b" must be `Tensor`s not
+ * `SparseTensor`s.  This op is optimized for the case where at least one of "a" or
+ * "b" is sparse, in the sense that they have a large proportion of zero values.
+ * The breakeven for using this versus a dense matrix multiply on one platform was
+ * 30% zero values in the sparse matrix.
+ * <p>
+ * The gradient computation of this operation will only take advantage of sparsity
+ * in the input gradient when that gradient comes from a Relu.
+ */
+@Operator(group = "sparse")
+public final class SparseMatMul extends RawOp implements Operand<TFloat32> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.sparse.SparseMatMul}
+   */
+  public static class Options {
+    
+    /**
+     * @param transposeA 
+     */
+    public Options transposeA(Boolean transposeA) {
+      this.transposeA = transposeA;
+      return this;
+    }
+    
+    /**
+     * @param transposeB 
+     */
+    public Options transposeB(Boolean transposeB) {
+      this.transposeB = transposeB;
+      return this;
+    }
+    
+    /**
+     * @param aIsSparse 
+     */
+    public Options aIsSparse(Boolean aIsSparse) {
+      this.aIsSparse = aIsSparse;
+      return this;
+    }
+    
+    /**
+     * @param bIsSparse 
+     */
+    public Options bIsSparse(Boolean bIsSparse) {
+      this.bIsSparse = bIsSparse;
+      return this;
+    }
+    
+    private Boolean transposeA;
+    private Boolean transposeB;
+    private Boolean aIsSparse;
+    private Boolean bIsSparse;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new SparseMatMul operation.
+   * 
+   * @param scope current scope
+   * @param a 
+   * @param b 
+   * @param options carries optional attributes values
+   * @return a new instance of SparseMatMul
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber, U extends TNumber> SparseMatMul create(Scope scope, Operand<T> a, Operand<U> b, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("SparseMatMul", scope.makeOpName("SparseMatMul"));
+    opBuilder.addInput(a.asOutput());
+    opBuilder.addInput(b.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.transposeA != null) {
+          opBuilder.setAttr("transpose_a", opts.transposeA);
+        }
+        if (opts.transposeB != null) {
+          opBuilder.setAttr("transpose_b", opts.transposeB);
+        }
+        if (opts.aIsSparse != null) {
+          opBuilder.setAttr("a_is_sparse", opts.aIsSparse);
+        }
+        if (opts.bIsSparse != null) {
+          opBuilder.setAttr("b_is_sparse", opts.bIsSparse);
+        }
+      }
+    }
+    return new SparseMatMul(opBuilder.build());
+  }
+  
+  /**
+   * @param transposeA 
+   */
+  public static Options transposeA(Boolean transposeA) {
+    return new Options().transposeA(transposeA);
+  }
+  
+  /**
+   * @param transposeB 
+   */
+  public static Options transposeB(Boolean transposeB) {
+    return new Options().transposeB(transposeB);
+  }
+  
+  /**
+   * @param aIsSparse 
+   */
+  public static Options aIsSparse(Boolean aIsSparse) {
+    return new Options().aIsSparse(aIsSparse);
+  }
+  
+  /**
+   * @param bIsSparse 
+   */
+  public static Options bIsSparse(Boolean bIsSparse) {
+    return new Options().bIsSparse(bIsSparse);
+  }
+  
+  /**
+   */
+  public Output<TFloat32> product() {
+    return product;
+  }
+  
+  @Override
+  public Output<TFloat32> asOutput() {
+    return product;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "SparseMatMul";
+  
+  private Output<TFloat32> product;
+  
+  private SparseMatMul(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    product = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseReduceMax.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseReduceMax.java
new file mode 100644
index 00000000000..85c01083363
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseReduceMax.java
@@ -0,0 +1,131 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.sparse;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Computes the max of elements across dimensions of a SparseTensor.
+ * <p>
+ * This Op takes a SparseTensor and is the sparse counterpart to
+ * `tf.reduce_max()`.  In particular, this Op also returns a dense `Tensor`
+ * instead of a sparse one.
+ * <p>
+ * Reduces `sp_input` along the dimensions given in `reduction_axes`.  Unless
+ * `keep_dims` is true, the rank of the tensor is reduced by 1 for each entry in
+ * `reduction_axes`. If `keep_dims` is true, the reduced dimensions are retained
+ * with length 1.
+ * <p>
+ * If `reduction_axes` has no entries, all dimensions are reduced, and a tensor
+ * with a single element is returned.  Additionally, the axes can be negative,
+ * which are interpreted according to the indexing rules in Python.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator(group = "sparse")
+public final class SparseReduceMax<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.sparse.SparseReduceMax}
+   */
+  public static class Options {
+    
+    /**
+     * @param keepDims If true, retain reduced dimensions with length 1.
+     */
+    public Options keepDims(Boolean keepDims) {
+      this.keepDims = keepDims;
+      return this;
+    }
+    
+    private Boolean keepDims;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new SparseReduceMax operation.
+   * 
+   * @param scope current scope
+   * @param inputIndices 2-D.  `N x R` matrix with the indices of non-empty values in a
+   * SparseTensor, possibly not in canonical ordering.
+   * @param inputValues 1-D.  `N` non-empty values corresponding to `input_indices`.
+   * @param inputShape 1-D.  Shape of the input SparseTensor.
+   * @param reductionAxes 1-D.  Length-`K` vector containing the reduction axes.
+   * @param options carries optional attributes values
+   * @return a new instance of SparseReduceMax
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> SparseReduceMax<T> create(Scope scope, Operand<TInt64> inputIndices, Operand<T> inputValues, Operand<TInt64> inputShape, Operand<TInt32> reductionAxes, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("SparseReduceMax", scope.makeOpName("SparseReduceMax"));
+    opBuilder.addInput(inputIndices.asOutput());
+    opBuilder.addInput(inputValues.asOutput());
+    opBuilder.addInput(inputShape.asOutput());
+    opBuilder.addInput(reductionAxes.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.keepDims != null) {
+          opBuilder.setAttr("keep_dims", opts.keepDims);
+        }
+      }
+    }
+    return new SparseReduceMax<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param keepDims If true, retain reduced dimensions with length 1.
+   */
+  public static Options keepDims(Boolean keepDims) {
+    return new Options().keepDims(keepDims);
+  }
+  
+  /**
+   * `R-K`-D.  The reduced Tensor.
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "SparseReduceMax";
+  
+  private Output<T> output;
+  
+  private SparseReduceMax(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseReduceMaxSparse.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseReduceMaxSparse.java
new file mode 100644
index 00000000000..2c2bbfb9348
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseReduceMaxSparse.java
@@ -0,0 +1,141 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.sparse;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Computes the max of elements across dimensions of a SparseTensor.
+ * <p>
+ * This Op takes a SparseTensor and is the sparse counterpart to
+ * `tf.reduce_max()`.  In contrast to SparseReduceMax, this Op returns a
+ * SparseTensor.
+ * <p>
+ * Reduces `sp_input` along the dimensions given in `reduction_axes`.  Unless
+ * `keep_dims` is true, the rank of the tensor is reduced by 1 for each entry in
+ * `reduction_axes`. If `keep_dims` is true, the reduced dimensions are retained
+ * with length 1.
+ * <p>
+ * If `reduction_axes` has no entries, all dimensions are reduced, and a tensor
+ * with a single element is returned.  Additionally, the axes can be negative,
+ * which are interpreted according to the indexing rules in Python.
+ * 
+ * @param <T> data type for {@code outputValues()} output
+ */
+@Operator(group = "sparse")
+public final class SparseReduceMaxSparse<T extends TNumber> extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.sparse.SparseReduceMaxSparse}
+   */
+  public static class Options {
+    
+    /**
+     * @param keepDims If true, retain reduced dimensions with length 1.
+     */
+    public Options keepDims(Boolean keepDims) {
+      this.keepDims = keepDims;
+      return this;
+    }
+    
+    private Boolean keepDims;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new SparseReduceMaxSparse operation.
+   * 
+   * @param scope current scope
+   * @param inputIndices 2-D.  `N x R` matrix with the indices of non-empty values in a
+   * SparseTensor, possibly not in canonical ordering.
+   * @param inputValues 1-D.  `N` non-empty values corresponding to `input_indices`.
+   * @param inputShape 1-D.  Shape of the input SparseTensor.
+   * @param reductionAxes 1-D.  Length-`K` vector containing the reduction axes.
+   * @param options carries optional attributes values
+   * @return a new instance of SparseReduceMaxSparse
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> SparseReduceMaxSparse<T> create(Scope scope, Operand<TInt64> inputIndices, Operand<T> inputValues, Operand<TInt64> inputShape, Operand<TInt32> reductionAxes, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("SparseReduceMaxSparse", scope.makeOpName("SparseReduceMaxSparse"));
+    opBuilder.addInput(inputIndices.asOutput());
+    opBuilder.addInput(inputValues.asOutput());
+    opBuilder.addInput(inputShape.asOutput());
+    opBuilder.addInput(reductionAxes.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.keepDims != null) {
+          opBuilder.setAttr("keep_dims", opts.keepDims);
+        }
+      }
+    }
+    return new SparseReduceMaxSparse<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param keepDims If true, retain reduced dimensions with length 1.
+   */
+  public static Options keepDims(Boolean keepDims) {
+    return new Options().keepDims(keepDims);
+  }
+  
+  /**
+   */
+  public Output<TInt64> outputIndices() {
+    return outputIndices;
+  }
+  
+  /**
+   */
+  public Output<T> outputValues() {
+    return outputValues;
+  }
+  
+  /**
+   */
+  public Output<TInt64> outputShape() {
+    return outputShape;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "SparseReduceMaxSparse";
+  
+  private Output<TInt64> outputIndices;
+  private Output<T> outputValues;
+  private Output<TInt64> outputShape;
+  
+  private SparseReduceMaxSparse(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    outputIndices = operation.output(outputIdx++);
+    outputValues = operation.output(outputIdx++);
+    outputShape = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseReduceSum.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseReduceSum.java
new file mode 100644
index 00000000000..138bb1aab5b
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseReduceSum.java
@@ -0,0 +1,131 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.sparse;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Computes the sum of elements across dimensions of a SparseTensor.
+ * <p>
+ * This Op takes a SparseTensor and is the sparse counterpart to
+ * `tf.reduce_sum()`.  In particular, this Op also returns a dense `Tensor`
+ * instead of a sparse one.
+ * <p>
+ * Reduces `sp_input` along the dimensions given in `reduction_axes`.  Unless
+ * `keep_dims` is true, the rank of the tensor is reduced by 1 for each entry in
+ * `reduction_axes`. If `keep_dims` is true, the reduced dimensions are retained
+ * with length 1.
+ * <p>
+ * If `reduction_axes` has no entries, all dimensions are reduced, and a tensor
+ * with a single element is returned.  Additionally, the axes can be negative,
+ * which are interpreted according to the indexing rules in Python.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator(group = "sparse")
+public final class SparseReduceSum<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.sparse.SparseReduceSum}
+   */
+  public static class Options {
+    
+    /**
+     * @param keepDims If true, retain reduced dimensions with length 1.
+     */
+    public Options keepDims(Boolean keepDims) {
+      this.keepDims = keepDims;
+      return this;
+    }
+    
+    private Boolean keepDims;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new SparseReduceSum operation.
+   * 
+   * @param scope current scope
+   * @param inputIndices 2-D.  `N x R` matrix with the indices of non-empty values in a
+   * SparseTensor, possibly not in canonical ordering.
+   * @param inputValues 1-D.  `N` non-empty values corresponding to `input_indices`.
+   * @param inputShape 1-D.  Shape of the input SparseTensor.
+   * @param reductionAxes 1-D.  Length-`K` vector containing the reduction axes.
+   * @param options carries optional attributes values
+   * @return a new instance of SparseReduceSum
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> SparseReduceSum<T> create(Scope scope, Operand<TInt64> inputIndices, Operand<T> inputValues, Operand<TInt64> inputShape, Operand<TInt32> reductionAxes, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("SparseReduceSum", scope.makeOpName("SparseReduceSum"));
+    opBuilder.addInput(inputIndices.asOutput());
+    opBuilder.addInput(inputValues.asOutput());
+    opBuilder.addInput(inputShape.asOutput());
+    opBuilder.addInput(reductionAxes.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.keepDims != null) {
+          opBuilder.setAttr("keep_dims", opts.keepDims);
+        }
+      }
+    }
+    return new SparseReduceSum<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param keepDims If true, retain reduced dimensions with length 1.
+   */
+  public static Options keepDims(Boolean keepDims) {
+    return new Options().keepDims(keepDims);
+  }
+  
+  /**
+   * `R-K`-D.  The reduced Tensor.
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "SparseReduceSum";
+  
+  private Output<T> output;
+  
+  private SparseReduceSum(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseReduceSumSparse.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseReduceSumSparse.java
new file mode 100644
index 00000000000..e27a1a98861
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseReduceSumSparse.java
@@ -0,0 +1,141 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.sparse;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Computes the sum of elements across dimensions of a SparseTensor.
+ * <p>
+ * This Op takes a SparseTensor and is the sparse counterpart to
+ * `tf.reduce_sum()`.  In contrast to SparseReduceSum, this Op returns a
+ * SparseTensor.
+ * <p>
+ * Reduces `sp_input` along the dimensions given in `reduction_axes`.  Unless
+ * `keep_dims` is true, the rank of the tensor is reduced by 1 for each entry in
+ * `reduction_axes`. If `keep_dims` is true, the reduced dimensions are retained
+ * with length 1.
+ * <p>
+ * If `reduction_axes` has no entries, all dimensions are reduced, and a tensor
+ * with a single element is returned.  Additionally, the axes can be negative,
+ * which are interpreted according to the indexing rules in Python.
+ * 
+ * @param <T> data type for {@code outputValues()} output
+ */
+@Operator(group = "sparse")
+public final class SparseReduceSumSparse<T extends TType> extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.sparse.SparseReduceSumSparse}
+   */
+  public static class Options {
+    
+    /**
+     * @param keepDims If true, retain reduced dimensions with length 1.
+     */
+    public Options keepDims(Boolean keepDims) {
+      this.keepDims = keepDims;
+      return this;
+    }
+    
+    private Boolean keepDims;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new SparseReduceSumSparse operation.
+   * 
+   * @param scope current scope
+   * @param inputIndices 2-D.  `N x R` matrix with the indices of non-empty values in a
+   * SparseTensor, possibly not in canonical ordering.
+   * @param inputValues 1-D.  `N` non-empty values corresponding to `input_indices`.
+   * @param inputShape 1-D.  Shape of the input SparseTensor.
+   * @param reductionAxes 1-D.  Length-`K` vector containing the reduction axes.
+   * @param options carries optional attributes values
+   * @return a new instance of SparseReduceSumSparse
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> SparseReduceSumSparse<T> create(Scope scope, Operand<TInt64> inputIndices, Operand<T> inputValues, Operand<TInt64> inputShape, Operand<TInt32> reductionAxes, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("SparseReduceSumSparse", scope.makeOpName("SparseReduceSumSparse"));
+    opBuilder.addInput(inputIndices.asOutput());
+    opBuilder.addInput(inputValues.asOutput());
+    opBuilder.addInput(inputShape.asOutput());
+    opBuilder.addInput(reductionAxes.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.keepDims != null) {
+          opBuilder.setAttr("keep_dims", opts.keepDims);
+        }
+      }
+    }
+    return new SparseReduceSumSparse<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param keepDims If true, retain reduced dimensions with length 1.
+   */
+  public static Options keepDims(Boolean keepDims) {
+    return new Options().keepDims(keepDims);
+  }
+  
+  /**
+   */
+  public Output<TInt64> outputIndices() {
+    return outputIndices;
+  }
+  
+  /**
+   */
+  public Output<T> outputValues() {
+    return outputValues;
+  }
+  
+  /**
+   */
+  public Output<TInt64> outputShape() {
+    return outputShape;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "SparseReduceSumSparse";
+  
+  private Output<TInt64> outputIndices;
+  private Output<T> outputValues;
+  private Output<TInt64> outputShape;
+  
+  private SparseReduceSumSparse(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    outputIndices = operation.output(outputIdx++);
+    outputValues = operation.output(outputIdx++);
+    outputShape = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseReorder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseReorder.java
new file mode 100644
index 00000000000..16713d1f767
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseReorder.java
@@ -0,0 +1,95 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.sparse;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Reorders a SparseTensor into the canonical, row-major ordering.
+ * <p>
+ * Note that by convention, all sparse ops preserve the canonical ordering along
+ * increasing dimension number. The only time ordering can be violated is during
+ * manual manipulation of the indices and values vectors to add entries.
+ * <p>
+ * Reordering does not affect the shape of the SparseTensor.
+ * <p>
+ * If the tensor has rank `R` and `N` non-empty values, `input_indices` has
+ * shape `[N, R]`, input_values has length `N`, and input_shape has length `R`.
+ * 
+ * @param <T> data type for {@code outputValues()} output
+ */
+@Operator(group = "sparse")
+public final class SparseReorder<T extends TType> extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new SparseReorder operation.
+   * 
+   * @param scope current scope
+   * @param inputIndices 2-D.  `N x R` matrix with the indices of non-empty values in a
+   * SparseTensor, possibly not in canonical ordering.
+   * @param inputValues 1-D.  `N` non-empty values corresponding to `input_indices`.
+   * @param inputShape 1-D.  Shape of the input SparseTensor.
+   * @return a new instance of SparseReorder
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> SparseReorder<T> create(Scope scope, Operand<TInt64> inputIndices, Operand<T> inputValues, Operand<TInt64> inputShape) {
+    OperationBuilder opBuilder = scope.env().opBuilder("SparseReorder", scope.makeOpName("SparseReorder"));
+    opBuilder.addInput(inputIndices.asOutput());
+    opBuilder.addInput(inputValues.asOutput());
+    opBuilder.addInput(inputShape.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new SparseReorder<T>(opBuilder.build());
+  }
+  
+  /**
+   * 2-D.  `N x R` matrix with the same indices as input_indices, but
+   * in canonical row-major ordering.
+   */
+  public Output<TInt64> outputIndices() {
+    return outputIndices;
+  }
+  
+  /**
+   * 1-D.  `N` non-empty values corresponding to `output_indices`.
+   */
+  public Output<T> outputValues() {
+    return outputValues;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "SparseReorder";
+  
+  private Output<TInt64> outputIndices;
+  private Output<T> outputValues;
+  
+  private SparseReorder(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    outputIndices = operation.output(outputIdx++);
+    outputValues = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseReshape.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseReshape.java
new file mode 100644
index 00000000000..772b19db9c5
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseReshape.java
@@ -0,0 +1,101 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.sparse;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt64;
+
+/**
+ * Reshapes a SparseTensor to represent values in a new dense shape.
+ * <p>
+ * This operation has the same semantics as reshape on the represented dense
+ * tensor.  The `input_indices` are recomputed based on the requested `new_shape`.
+ * <p>
+ * If one component of `new_shape` is the special value -1, the size of that
+ * dimension is computed so that the total dense size remains constant.  At
+ * most one component of `new_shape` can be -1.  The number of dense elements
+ * implied by `new_shape` must be the same as the number of dense elements
+ * originally implied by `input_shape`.
+ * <p>
+ * Reshaping does not affect the order of values in the SparseTensor.
+ * <p>
+ * If the input tensor has rank `R_in` and `N` non-empty values, and `new_shape`
+ * has length `R_out`, then `input_indices` has shape `[N, R_in]`,
+ * `input_shape` has length `R_in`, `output_indices` has shape `[N, R_out]`, and
+ * `output_shape` has length `R_out`.
+ */
+@Operator(group = "sparse")
+public final class SparseReshape extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new SparseReshape operation.
+   * 
+   * @param scope current scope
+   * @param inputIndices 2-D.  `N x R_in` matrix with the indices of non-empty values in a
+   * SparseTensor.
+   * @param inputShape 1-D.  `R_in` vector with the input SparseTensor's dense shape.
+   * @param newShape 1-D.  `R_out` vector with the requested new dense shape.
+   * @return a new instance of SparseReshape
+   */
+  @Endpoint(describeByClass = true)
+  public static SparseReshape create(Scope scope, Operand<TInt64> inputIndices, Operand<TInt64> inputShape, Operand<TInt64> newShape) {
+    OperationBuilder opBuilder = scope.env().opBuilder("SparseReshape", scope.makeOpName("SparseReshape"));
+    opBuilder.addInput(inputIndices.asOutput());
+    opBuilder.addInput(inputShape.asOutput());
+    opBuilder.addInput(newShape.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new SparseReshape(opBuilder.build());
+  }
+  
+  /**
+   * 2-D.  `N x R_out` matrix with the updated indices of non-empty
+   * values in the output SparseTensor.
+   */
+  public Output<TInt64> outputIndices() {
+    return outputIndices;
+  }
+  
+  /**
+   * 1-D.  `R_out` vector with the full dense shape of the output
+   * SparseTensor.  This is the same as `new_shape` but with any -1 dimensions
+   * filled in.
+   */
+  public Output<TInt64> outputShape() {
+    return outputShape;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "SparseReshape";
+  
+  private Output<TInt64> outputIndices;
+  private Output<TInt64> outputShape;
+  
+  private SparseReshape(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    outputIndices = operation.output(outputIdx++);
+    outputShape = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentMean.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentMean.java
new file mode 100644
index 00000000000..72e5784e390
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentMean.java
@@ -0,0 +1,85 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.sparse;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Computes the mean along sparse segments of a tensor.
+ * <p>
+ * See `tf.sparse.segment_sum` for usage examples.
+ * <p>
+ * Like `SegmentMean`, but `segment_ids` can have rank less than `data`'s first
+ * dimension, selecting a subset of dimension 0, specified by `indices`.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator(group = "sparse")
+public final class SparseSegmentMean<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new SparseSegmentMean operation.
+   * 
+   * @param scope current scope
+   * @param data 
+   * @param indices A 1-D tensor. Has same rank as `segment_ids`.
+   * @param segmentIds A 1-D tensor. Values should be sorted and can be repeated.
+   * @return a new instance of SparseSegmentMean
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber, U extends TNumber, V extends TNumber> SparseSegmentMean<T> create(Scope scope, Operand<T> data, Operand<U> indices, Operand<V> segmentIds) {
+    OperationBuilder opBuilder = scope.env().opBuilder("SparseSegmentMean", scope.makeOpName("SparseSegmentMean"));
+    opBuilder.addInput(data.asOutput());
+    opBuilder.addInput(indices.asOutput());
+    opBuilder.addInput(segmentIds.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new SparseSegmentMean<T>(opBuilder.build());
+  }
+  
+  /**
+   * Has same shape as data, except for dimension 0 which
+   * has size `k`, the number of segments.
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "SparseSegmentMean";
+  
+  private Output<T> output;
+  
+  private SparseSegmentMean(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentMeanGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentMeanGrad.java
new file mode 100644
index 00000000000..7fc5d505ee6
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentMeanGrad.java
@@ -0,0 +1,84 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.sparse;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Computes gradients for SparseSegmentMean.
+ * <p>
+ * Returns tensor "output" with same shape as grad, except for dimension 0 whose
+ * value is output_dim0.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator(group = "sparse")
+public final class SparseSegmentMeanGrad<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new SparseSegmentMeanGrad operation.
+   * 
+   * @param scope current scope
+   * @param grad gradient propagated to the SparseSegmentMean op.
+   * @param indices indices passed to the corresponding SparseSegmentMean op.
+   * @param segmentIds segment_ids passed to the corresponding SparseSegmentMean op.
+   * @param outputDim0 dimension 0 of "data" passed to SparseSegmentMean op.
+   * @return a new instance of SparseSegmentMeanGrad
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber, U extends TNumber, V extends TNumber> SparseSegmentMeanGrad<T> create(Scope scope, Operand<T> grad, Operand<U> indices, Operand<V> segmentIds, Operand<TInt32> outputDim0) {
+    OperationBuilder opBuilder = scope.env().opBuilder("SparseSegmentMeanGrad", scope.makeOpName("SparseSegmentMeanGrad"));
+    opBuilder.addInput(grad.asOutput());
+    opBuilder.addInput(indices.asOutput());
+    opBuilder.addInput(segmentIds.asOutput());
+    opBuilder.addInput(outputDim0.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new SparseSegmentMeanGrad<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "SparseSegmentMeanGrad";
+  
+  private Output<T> output;
+  
+  private SparseSegmentMeanGrad(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentMeanWithNumSegments.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentMeanWithNumSegments.java
new file mode 100644
index 00000000000..9e2047047f8
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentMeanWithNumSegments.java
@@ -0,0 +1,89 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.sparse;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Computes the mean along sparse segments of a tensor.
+ * <p>
+ * Like `SparseSegmentMean`, but allows missing ids in `segment_ids`. If an id is
+ * missing, the `output` tensor at that position will be zeroed.
+ * <p>
+ * Read
+ * [the section on segmentation](https://tensorflow.org/api_docs/python/tf/math#Segmentation)
+ * for an explanation of segments.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator(group = "sparse")
+public final class SparseSegmentMeanWithNumSegments<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new SparseSegmentMeanWithNumSegments operation.
+   * 
+   * @param scope current scope
+   * @param data 
+   * @param indices A 1-D tensor. Has same rank as `segment_ids`.
+   * @param segmentIds A 1-D tensor. Values should be sorted and can be repeated.
+   * @param numSegments Should equal the number of distinct segment IDs.
+   * @return a new instance of SparseSegmentMeanWithNumSegments
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber, U extends TNumber, V extends TNumber, W extends TNumber> SparseSegmentMeanWithNumSegments<T> create(Scope scope, Operand<T> data, Operand<U> indices, Operand<V> segmentIds, Operand<W> numSegments) {
+    OperationBuilder opBuilder = scope.env().opBuilder("SparseSegmentMeanWithNumSegments", scope.makeOpName("SparseSegmentMeanWithNumSegments"));
+    opBuilder.addInput(data.asOutput());
+    opBuilder.addInput(indices.asOutput());
+    opBuilder.addInput(segmentIds.asOutput());
+    opBuilder.addInput(numSegments.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new SparseSegmentMeanWithNumSegments<T>(opBuilder.build());
+  }
+  
+  /**
+   * Has same shape as data, except for dimension 0 which has size
+   * `num_segments`.
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "SparseSegmentMeanWithNumSegments";
+  
+  private Output<T> output;
+  
+  private SparseSegmentMeanWithNumSegments(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentSqrtN.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentSqrtN.java
new file mode 100644
index 00000000000..dccfcc0d1c2
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentSqrtN.java
@@ -0,0 +1,85 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.sparse;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Computes the sum along sparse segments of a tensor divided by the sqrt of N.
+ * <p>
+ * N is the size of the segment being reduced.
+ * <p>
+ * See `tf.sparse.segment_sum` for usage examples.
+ * 
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator(group = "sparse")
+public final class SparseSegmentSqrtN<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new SparseSegmentSqrtN operation.
+   * 
+   * @param scope current scope
+   * @param data 
+   * @param indices A 1-D tensor. Has same rank as `segment_ids`.
+   * @param segmentIds A 1-D tensor. Values should be sorted and can be repeated.
+   * @return a new instance of SparseSegmentSqrtN
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber, U extends TNumber, V extends TNumber> SparseSegmentSqrtN<T> create(Scope scope, Operand<T> data, Operand<U> indices, Operand<V> segmentIds) {
+    OperationBuilder opBuilder = scope.env().opBuilder("SparseSegmentSqrtN", scope.makeOpName("SparseSegmentSqrtN"));
+    opBuilder.addInput(data.asOutput());
+    opBuilder.addInput(indices.asOutput());
+    opBuilder.addInput(segmentIds.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new SparseSegmentSqrtN<T>(opBuilder.build());
+  }
+  
+  /**
+   * Has same shape as data, except for dimension 0 which
+   * has size `k`, the number of segments.
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "SparseSegmentSqrtN";
+  
+  private Output<T> output;
+  
+  private SparseSegmentSqrtN(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentSqrtNGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentSqrtNGrad.java
new file mode 100644
index 00000000000..a9b7fb3023b
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentSqrtNGrad.java
@@ -0,0 +1,84 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.sparse;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Computes gradients for SparseSegmentSqrtN.
+ * <p>
+ * Returns tensor "output" with same shape as grad, except for dimension 0 whose
+ * value is output_dim0.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator(group = "sparse")
+public final class SparseSegmentSqrtNGrad<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new SparseSegmentSqrtNGrad operation.
+   * 
+   * @param scope current scope
+   * @param grad gradient propagated to the SparseSegmentSqrtN op.
+   * @param indices indices passed to the corresponding SparseSegmentSqrtN op.
+   * @param segmentIds segment_ids passed to the corresponding SparseSegmentSqrtN op.
+   * @param outputDim0 dimension 0 of "data" passed to SparseSegmentSqrtN op.
+   * @return a new instance of SparseSegmentSqrtNGrad
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber, U extends TNumber, V extends TNumber> SparseSegmentSqrtNGrad<T> create(Scope scope, Operand<T> grad, Operand<U> indices, Operand<V> segmentIds, Operand<TInt32> outputDim0) {
+    OperationBuilder opBuilder = scope.env().opBuilder("SparseSegmentSqrtNGrad", scope.makeOpName("SparseSegmentSqrtNGrad"));
+    opBuilder.addInput(grad.asOutput());
+    opBuilder.addInput(indices.asOutput());
+    opBuilder.addInput(segmentIds.asOutput());
+    opBuilder.addInput(outputDim0.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new SparseSegmentSqrtNGrad<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "SparseSegmentSqrtNGrad";
+  
+  private Output<T> output;
+  
+  private SparseSegmentSqrtNGrad(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentSqrtNWithNumSegments.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentSqrtNWithNumSegments.java
new file mode 100644
index 00000000000..127ffa87198
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentSqrtNWithNumSegments.java
@@ -0,0 +1,91 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.sparse;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Computes the sum along sparse segments of a tensor divided by the sqrt of N.
+ * <p>
+ * N is the size of the segment being reduced.
+ * <p>
+ * Like `SparseSegmentSqrtN`, but allows missing ids in `segment_ids`. If an id is
+ * missing, the `output` tensor at that position will be zeroed.
+ * <p>
+ * Read
+ * [the section on segmentation](https://tensorflow.org/api_docs/python/tf/math#Segmentation)
+ * for an explanation of segments.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator(group = "sparse")
+public final class SparseSegmentSqrtNWithNumSegments<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new SparseSegmentSqrtNWithNumSegments operation.
+   * 
+   * @param scope current scope
+   * @param data 
+   * @param indices A 1-D tensor. Has same rank as `segment_ids`.
+   * @param segmentIds A 1-D tensor. Values should be sorted and can be repeated.
+   * @param numSegments Should equal the number of distinct segment IDs.
+   * @return a new instance of SparseSegmentSqrtNWithNumSegments
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber, U extends TNumber, V extends TNumber, W extends TNumber> SparseSegmentSqrtNWithNumSegments<T> create(Scope scope, Operand<T> data, Operand<U> indices, Operand<V> segmentIds, Operand<W> numSegments) {
+    OperationBuilder opBuilder = scope.env().opBuilder("SparseSegmentSqrtNWithNumSegments", scope.makeOpName("SparseSegmentSqrtNWithNumSegments"));
+    opBuilder.addInput(data.asOutput());
+    opBuilder.addInput(indices.asOutput());
+    opBuilder.addInput(segmentIds.asOutput());
+    opBuilder.addInput(numSegments.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new SparseSegmentSqrtNWithNumSegments<T>(opBuilder.build());
+  }
+  
+  /**
+   * Has same shape as data, except for dimension 0 which
+   * has size `k`, the number of segments.
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "SparseSegmentSqrtNWithNumSegments";
+  
+  private Output<T> output;
+  
+  private SparseSegmentSqrtNWithNumSegments(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentSum.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentSum.java
new file mode 100644
index 00000000000..321fbaed9ad
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentSum.java
@@ -0,0 +1,110 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.sparse;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Computes the sum along sparse segments of a tensor.
+ * <p>
+ * Read
+ * [the section on segmentation](https://tensorflow.org/api_docs/python/tf/math#Segmentation)
+ * for an explanation of segments.
+ * <p>
+ * Like `SegmentSum`, but `segment_ids` can have rank less than `data`'s first
+ * dimension, selecting a subset of dimension 0, specified by `indices`.
+ * <p>
+ * For example:
+ * <pre>{@code
+ * c = tf.constant([[1,2,3,4], [-1,-2,-3,-4], [5,6,7,8]])
+ * 
+ * # Select two rows, one segment.
+ * tf.sparse_segment_sum(c, tf.constant([0, 1]), tf.constant([0, 0]))
+ * # => [[0 0 0 0]]
+ * 
+ * # Select two rows, two segment.
+ * tf.sparse_segment_sum(c, tf.constant([0, 1]), tf.constant([0, 1]))
+ * # => [[ 1  2  3  4]
+ * #     [-1 -2 -3 -4]]
+ * 
+ * # Select all rows, two segments.
+ * tf.sparse_segment_sum(c, tf.constant([0, 1, 2]), tf.constant([0, 0, 1]))
+ * # => [[0 0 0 0]
+ * #     [5 6 7 8]]
+ * 
+ * # Which is equivalent to:
+ * tf.segment_sum(c, tf.constant([0, 0, 1]))
+ * }</pre>
+ * 
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator(group = "sparse")
+public final class SparseSegmentSum<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new SparseSegmentSum operation.
+   * 
+   * @param scope current scope
+   * @param data 
+   * @param indices A 1-D tensor. Has same rank as `segment_ids`.
+   * @param segmentIds A 1-D tensor. Values should be sorted and can be repeated.
+   * @return a new instance of SparseSegmentSum
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber, U extends TNumber, V extends TNumber> SparseSegmentSum<T> create(Scope scope, Operand<T> data, Operand<U> indices, Operand<V> segmentIds) {
+    OperationBuilder opBuilder = scope.env().opBuilder("SparseSegmentSum", scope.makeOpName("SparseSegmentSum"));
+    opBuilder.addInput(data.asOutput());
+    opBuilder.addInput(indices.asOutput());
+    opBuilder.addInput(segmentIds.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new SparseSegmentSum<T>(opBuilder.build());
+  }
+  
+  /**
+   * Has same shape as data, except for dimension 0 which
+   * has size `k`, the number of segments.
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "SparseSegmentSum";
+  
+  private Output<T> output;
+  
+  private SparseSegmentSum(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentSumWithNumSegments.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentSumWithNumSegments.java
new file mode 100644
index 00000000000..830ca794891
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSegmentSumWithNumSegments.java
@@ -0,0 +1,110 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.sparse;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Computes the sum along sparse segments of a tensor.
+ * <p>
+ * Like `SparseSegmentSum`, but allows missing ids in `segment_ids`. If an id is
+ * missing, the `output` tensor at that position will be zeroed.
+ * <p>
+ * Read
+ * [the section on segmentation](https://tensorflow.org/api_docs/python/tf/sparse#Segmentation)
+ * for an explanation of segments.
+ * <p>
+ * For example:
+ * <pre>{@code
+ * c = tf.constant([[1,2,3,4], [-1,-2,-3,-4], [5,6,7,8]])
+ * 
+ * tf.sparse_segment_sum_with_num_segments(
+ *     c, tf.constant([0, 1]), tf.constant([0, 0]), num_segments=3)
+ * # => [[0 0 0 0]
+ * #     [0 0 0 0]
+ * #     [0 0 0 0]]
+ * 
+ * tf.sparse_segment_sum_with_num_segments(c,
+ *                                         tf.constant([0, 1]),
+ *                                         tf.constant([0, 2],
+ *                                         num_segments=4))
+ * # => [[ 1  2  3  4]
+ * #     [ 0  0  0  0]
+ * #     [-1 -2 -3 -4]
+ * #     [ 0  0  0  0]]
+ * }</pre>
+ * 
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator(group = "sparse")
+public final class SparseSegmentSumWithNumSegments<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new SparseSegmentSumWithNumSegments operation.
+   * 
+   * @param scope current scope
+   * @param data 
+   * @param indices A 1-D tensor. Has same rank as `segment_ids`.
+   * @param segmentIds A 1-D tensor. Values should be sorted and can be repeated.
+   * @param numSegments Should equal the number of distinct segment IDs.
+   * @return a new instance of SparseSegmentSumWithNumSegments
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber, U extends TNumber, V extends TNumber, W extends TNumber> SparseSegmentSumWithNumSegments<T> create(Scope scope, Operand<T> data, Operand<U> indices, Operand<V> segmentIds, Operand<W> numSegments) {
+    OperationBuilder opBuilder = scope.env().opBuilder("SparseSegmentSumWithNumSegments", scope.makeOpName("SparseSegmentSumWithNumSegments"));
+    opBuilder.addInput(data.asOutput());
+    opBuilder.addInput(indices.asOutput());
+    opBuilder.addInput(segmentIds.asOutput());
+    opBuilder.addInput(numSegments.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new SparseSegmentSumWithNumSegments<T>(opBuilder.build());
+  }
+  
+  /**
+   * Has same shape as data, except for dimension 0 which
+   * has size `num_segments`.
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "SparseSegmentSumWithNumSegments";
+  
+  private Output<T> output;
+  
+  private SparseSegmentSumWithNumSegments(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSlice.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSlice.java
new file mode 100644
index 00000000000..4ed30d8e2eb
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSlice.java
@@ -0,0 +1,116 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.sparse;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Slice a `SparseTensor` based on the `start` and `size`.
+ * <p>
+ * For example, if the input is
+ * <p>
+ *     input_tensor = shape = [2, 7]
+ *     [    a   d e  ]
+ *     [b c          ]
+ * <p>
+ * Graphically the output tensors are:
+ * <p>
+ *     sparse_slice([0, 0], [2, 4]) = shape = [2, 4]
+ *     [    a  ]
+ *     [b c    ]
+ * <p>
+ *     sparse_slice([0, 4], [2, 3]) = shape = [2, 3]
+ *     [ d e  ]
+ *     [      ]
+ * 
+ * @param <T> data type for {@code outputValues()} output
+ */
+@Operator(group = "sparse")
+public final class SparseSlice<T extends TType> extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new SparseSlice operation.
+   * 
+   * @param scope current scope
+   * @param indices 2-D tensor represents the indices of the sparse tensor.
+   * @param values 1-D tensor represents the values of the sparse tensor.
+   * @param shape 1-D. tensor represents the shape of the sparse tensor.
+   * @param start 1-D. tensor represents the start of the slice.
+   * @param size 1-D. tensor represents the size of the slice.
+   * output indices: A list of 1-D tensors represents the indices of the output
+   * sparse tensors.
+   * @return a new instance of SparseSlice
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> SparseSlice<T> create(Scope scope, Operand<TInt64> indices, Operand<T> values, Operand<TInt64> shape, Operand<TInt64> start, Operand<TInt64> size) {
+    OperationBuilder opBuilder = scope.env().opBuilder("SparseSlice", scope.makeOpName("SparseSlice"));
+    opBuilder.addInput(indices.asOutput());
+    opBuilder.addInput(values.asOutput());
+    opBuilder.addInput(shape.asOutput());
+    opBuilder.addInput(start.asOutput());
+    opBuilder.addInput(size.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new SparseSlice<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<TInt64> outputIndices() {
+    return outputIndices;
+  }
+  
+  /**
+   * A list of 1-D tensors represents the values of the output sparse
+   * tensors.
+   */
+  public Output<T> outputValues() {
+    return outputValues;
+  }
+  
+  /**
+   * A list of 1-D tensors represents the shape of the output sparse
+   * tensors.
+   */
+  public Output<TInt64> outputShape() {
+    return outputShape;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "SparseSlice";
+  
+  private Output<TInt64> outputIndices;
+  private Output<T> outputValues;
+  private Output<TInt64> outputShape;
+  
+  private SparseSlice(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    outputIndices = operation.output(outputIdx++);
+    outputValues = operation.output(outputIdx++);
+    outputShape = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSliceGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSliceGrad.java
new file mode 100644
index 00000000000..480631eccbc
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSliceGrad.java
@@ -0,0 +1,87 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.sparse;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
+
+/**
+ * The gradient operator for the SparseSlice op.
+ * <p>
+ * This op takes in the upstream gradient w.r.t. non-empty values of
+ * the sliced `SparseTensor`, and outputs the gradients w.r.t.
+ * the non-empty values of input `SparseTensor`.
+ * 
+ * @param <T> data type for {@code valGrad()} output
+ */
+@Operator(group = "sparse")
+public final class SparseSliceGrad<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new SparseSliceGrad operation.
+   * 
+   * @param scope current scope
+   * @param backpropValGrad 1-D. The gradient with respect to
+   * the non-empty values of the sliced `SparseTensor`.
+   * @param inputIndices 2-D.  The `indices` of the input `SparseTensor`.
+   * @param inputStart 1-D. tensor represents the start of the slice.
+   * @param outputIndices 2-D.  The `indices` of the sliced `SparseTensor`.
+   * @return a new instance of SparseSliceGrad
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> SparseSliceGrad<T> create(Scope scope, Operand<T> backpropValGrad, Operand<TInt64> inputIndices, Operand<TInt64> inputStart, Operand<TInt64> outputIndices) {
+    OperationBuilder opBuilder = scope.env().opBuilder("SparseSliceGrad", scope.makeOpName("SparseSliceGrad"));
+    opBuilder.addInput(backpropValGrad.asOutput());
+    opBuilder.addInput(inputIndices.asOutput());
+    opBuilder.addInput(inputStart.asOutput());
+    opBuilder.addInput(outputIndices.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new SparseSliceGrad<T>(opBuilder.build());
+  }
+  
+  /**
+   * 1-D. The gradient with respect to the non-empty values of input `SparseTensor`.
+   */
+  public Output<T> valGrad() {
+    return valGrad;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return valGrad;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "SparseSliceGrad";
+  
+  private Output<T> valGrad;
+  
+  private SparseSliceGrad(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    valGrad = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSoftmax.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSoftmax.java
new file mode 100644
index 00000000000..81f0cd9bfe4
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSoftmax.java
@@ -0,0 +1,97 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.sparse;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Applies softmax to a batched N-D `SparseTensor`.
+ * <p>
+ * The inputs represent an N-D SparseTensor  with logical shape `[..., B, C]`
+ * (where `N >= 2`), and with indices sorted in the canonical lexicographic order.
+ * <p>
+ * This op is equivalent to applying the normal `tf.nn.softmax()` to each innermost
+ * logical submatrix with shape `[B, C]`, but with the catch that <i>the implicitly
+ * zero elements do not participate</i>.  Specifically, the algorithm is equivalent
+ * to the following:
+ * <p>
+ *   (1) Applies `tf.nn.softmax()` to a densified view of each innermost submatrix
+ *       with shape `[B, C]`, along the size-C dimension;
+ *   (2) Masks out the original implicitly-zero locations;
+ *   (3) Renormalizes the remaining elements.
+ * <p>
+ * Hence, the `SparseTensor` result has exactly the same non-zero indices and
+ * shape.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator(group = "sparse")
+public final class SparseSoftmax<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new SparseSoftmax operation.
+   * 
+   * @param scope current scope
+   * @param spIndices 2-D.  `NNZ x R` matrix with the indices of non-empty values in a
+   * SparseTensor, in canonical ordering.
+   * @param spValues 1-D.  `NNZ` non-empty values corresponding to `sp_indices`.
+   * @param spShape 1-D.  Shape of the input SparseTensor.
+   * @return a new instance of SparseSoftmax
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> SparseSoftmax<T> create(Scope scope, Operand<TInt64> spIndices, Operand<T> spValues, Operand<TInt64> spShape) {
+    OperationBuilder opBuilder = scope.env().opBuilder("SparseSoftmax", scope.makeOpName("SparseSoftmax"));
+    opBuilder.addInput(spIndices.asOutput());
+    opBuilder.addInput(spValues.asOutput());
+    opBuilder.addInput(spShape.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new SparseSoftmax<T>(opBuilder.build());
+  }
+  
+  /**
+   * 1-D.  The `NNZ` values for the result `SparseTensor`.
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "SparseSoftmax";
+  
+  private Output<T> output;
+  
+  private SparseSoftmax(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSparseMaximum.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSparseMaximum.java
new file mode 100644
index 00000000000..976ac5b4633
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSparseMaximum.java
@@ -0,0 +1,93 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.sparse;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Returns the element-wise max of two SparseTensors.
+ * <p>
+ * Assumes the two SparseTensors have the same shape, i.e., no broadcasting.
+ * 
+ * @param <T> data type for {@code outputValues()} output
+ */
+@Operator(group = "sparse")
+public final class SparseSparseMaximum<T extends TNumber> extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new SparseSparseMaximum operation.
+   * 
+   * @param scope current scope
+   * @param aIndices 2-D.  `N x R` matrix with the indices of non-empty values in a
+   * SparseTensor, in the canonical lexicographic ordering.
+   * @param aValues 1-D.  `N` non-empty values corresponding to `a_indices`.
+   * @param aShape 1-D.  Shape of the input SparseTensor.
+   * @param bIndices counterpart to `a_indices` for the other operand.
+   * @param bValues counterpart to `a_values` for the other operand; must be of the same dtype.
+   * @param bShape counterpart to `a_shape` for the other operand; the two shapes must be equal.
+   * @return a new instance of SparseSparseMaximum
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> SparseSparseMaximum<T> create(Scope scope, Operand<TInt64> aIndices, Operand<T> aValues, Operand<TInt64> aShape, Operand<TInt64> bIndices, Operand<T> bValues, Operand<TInt64> bShape) {
+    OperationBuilder opBuilder = scope.env().opBuilder("SparseSparseMaximum", scope.makeOpName("SparseSparseMaximum"));
+    opBuilder.addInput(aIndices.asOutput());
+    opBuilder.addInput(aValues.asOutput());
+    opBuilder.addInput(aShape.asOutput());
+    opBuilder.addInput(bIndices.asOutput());
+    opBuilder.addInput(bValues.asOutput());
+    opBuilder.addInput(bShape.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new SparseSparseMaximum<T>(opBuilder.build());
+  }
+  
+  /**
+   * 2-D.  The indices of the output SparseTensor.
+   */
+  public Output<TInt64> outputIndices() {
+    return outputIndices;
+  }
+  
+  /**
+   * 1-D.  The values of the output SparseTensor.
+   */
+  public Output<T> outputValues() {
+    return outputValues;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "SparseSparseMaximum";
+  
+  private Output<TInt64> outputIndices;
+  private Output<T> outputValues;
+  
+  private SparseSparseMaximum(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    outputIndices = operation.output(outputIdx++);
+    outputValues = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSparseMinimum.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSparseMinimum.java
new file mode 100644
index 00000000000..b2f6aedbccb
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSparseMinimum.java
@@ -0,0 +1,93 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.sparse;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Returns the element-wise min of two SparseTensors.
+ * <p>
+ * Assumes the two SparseTensors have the same shape, i.e., no broadcasting.
+ * 
+ * @param <T> data type for {@code outputValues()} output
+ */
+@Operator(group = "sparse")
+public final class SparseSparseMinimum<T extends TType> extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new SparseSparseMinimum operation.
+   * 
+   * @param scope current scope
+   * @param aIndices 2-D.  `N x R` matrix with the indices of non-empty values in a
+   * SparseTensor, in the canonical lexicographic ordering.
+   * @param aValues 1-D.  `N` non-empty values corresponding to `a_indices`.
+   * @param aShape 1-D.  Shape of the input SparseTensor.
+   * @param bIndices counterpart to `a_indices` for the other operand.
+   * @param bValues counterpart to `a_values` for the other operand; must be of the same dtype.
+   * @param bShape counterpart to `a_shape` for the other operand; the two shapes must be equal.
+   * @return a new instance of SparseSparseMinimum
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> SparseSparseMinimum<T> create(Scope scope, Operand<TInt64> aIndices, Operand<T> aValues, Operand<TInt64> aShape, Operand<TInt64> bIndices, Operand<T> bValues, Operand<TInt64> bShape) {
+    OperationBuilder opBuilder = scope.env().opBuilder("SparseSparseMinimum", scope.makeOpName("SparseSparseMinimum"));
+    opBuilder.addInput(aIndices.asOutput());
+    opBuilder.addInput(aValues.asOutput());
+    opBuilder.addInput(aShape.asOutput());
+    opBuilder.addInput(bIndices.asOutput());
+    opBuilder.addInput(bValues.asOutput());
+    opBuilder.addInput(bShape.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new SparseSparseMinimum<T>(opBuilder.build());
+  }
+  
+  /**
+   * 2-D.  The indices of the output SparseTensor.
+   */
+  public Output<TInt64> outputIndices() {
+    return outputIndices;
+  }
+  
+  /**
+   * 1-D.  The values of the output SparseTensor.
+   */
+  public Output<T> outputValues() {
+    return outputValues;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "SparseSparseMinimum";
+  
+  private Output<TInt64> outputIndices;
+  private Output<T> outputValues;
+  
+  private SparseSparseMinimum(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    outputIndices = operation.output(outputIdx++);
+    outputValues = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSplit.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSplit.java
new file mode 100644
index 00000000000..e93f6ebe40e
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseSplit.java
@@ -0,0 +1,128 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.sparse;
+
+import java.util.Arrays;
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Split a `SparseTensor` into `num_split` tensors along one dimension.
+ * <p>
+ * If the `shape[split_dim]` is not an integer multiple of `num_split`. Slices
+ * `[0 : shape[split_dim] % num_split]` gets one extra dimension.
+ * For example, if `split_dim = 1` and `num_split = 2` and the input is
+ * <p>
+ *     input_tensor = shape = [2, 7]
+ *     [    a   d e  ]
+ *     [b c          ]
+ * <p>
+ * Graphically the output tensors are:
+ * <p>
+ *     output_tensor[0] = shape = [2, 4]
+ *     [    a  ]
+ *     [b c    ]
+ * <p>
+ *     output_tensor[1] = shape = [2, 3]
+ *     [ d e  ]
+ *     [      ]
+ * 
+ * @param <T> data type for {@code outputValues()} output
+ */
+@Operator(group = "sparse")
+public final class SparseSplit<T extends TType> extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new SparseSplit operation.
+   * 
+   * @param scope current scope
+   * @param splitDim 0-D.  The dimension along which to split.  Must be in the range
+   * `[0, rank(shape))`.
+   * @param indices 2-D tensor represents the indices of the sparse tensor.
+   * @param values 1-D tensor represents the values of the sparse tensor.
+   * @param shape 1-D. tensor represents the shape of the sparse tensor.
+   * output indices: A list of 1-D tensors represents the indices of the output
+   * sparse tensors.
+   * @param numSplit The number of ways to split.
+   * @return a new instance of SparseSplit
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> SparseSplit<T> create(Scope scope, Operand<TInt64> splitDim, Operand<TInt64> indices, Operand<T> values, Operand<TInt64> shape, Long numSplit) {
+    OperationBuilder opBuilder = scope.env().opBuilder("SparseSplit", scope.makeOpName("SparseSplit"));
+    opBuilder.addInput(splitDim.asOutput());
+    opBuilder.addInput(indices.asOutput());
+    opBuilder.addInput(values.asOutput());
+    opBuilder.addInput(shape.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("num_split", numSplit);
+    return new SparseSplit<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public List<Output<TInt64>> outputIndices() {
+    return outputIndices;
+  }
+  
+  /**
+   * A list of 1-D tensors represents the values of the output sparse
+   * tensors.
+   */
+  public List<Output<T>> outputValues() {
+    return outputValues;
+  }
+  
+  /**
+   * A list of 1-D tensors represents the shape of the output sparse
+   * tensors.
+   */
+  public List<Output<TInt64>> outputShape() {
+    return outputShape;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "SparseSplit";
+  
+  private List<Output<TInt64>> outputIndices;
+  private List<Output<T>> outputValues;
+  private List<Output<TInt64>> outputShape;
+  
+  @SuppressWarnings("unchecked")
+  private SparseSplit(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    int outputIndicesLength = operation.outputListLength("output_indices");
+    outputIndices = Arrays.asList((Output<TInt64>[])operation.outputList(outputIdx, outputIndicesLength));
+    outputIdx += outputIndicesLength;
+    int outputValuesLength = operation.outputListLength("output_values");
+    outputValues = Arrays.asList((Output<T>[])operation.outputList(outputIdx, outputValuesLength));
+    outputIdx += outputValuesLength;
+    int outputShapeLength = operation.outputListLength("output_shape");
+    outputShape = Arrays.asList((Output<TInt64>[])operation.outputList(outputIdx, outputShapeLength));
+    outputIdx += outputShapeLength;
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseTensorDenseAdd.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseTensorDenseAdd.java
new file mode 100644
index 00000000000..9959d57d16b
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseTensorDenseAdd.java
@@ -0,0 +1,83 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.sparse;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Adds up a `SparseTensor` and a dense `Tensor`, producing a dense `Tensor`.
+ * <p>
+ * This Op does not require `a_indices` be sorted in standard lexicographic order.
+ * 
+ * @param <U> data type for {@code output()} output
+ */
+@Operator(group = "sparse")
+public final class SparseTensorDenseAdd<U extends TType> extends RawOp implements Operand<U> {
+  
+  /**
+   * Factory method to create a class wrapping a new SparseTensorDenseAdd operation.
+   * 
+   * @param scope current scope
+   * @param aIndices 2-D.  The `indices` of the `SparseTensor`, with shape `[nnz, ndims]`.
+   * @param aValues 1-D.  The `values` of the `SparseTensor`, with shape `[nnz]`.
+   * @param aShape 1-D.  The `shape` of the `SparseTensor`, with shape `[ndims]`.
+   * @param b `ndims`-D Tensor.  With shape `a_shape`.
+   * @return a new instance of SparseTensorDenseAdd
+   */
+  @Endpoint(describeByClass = true)
+  public static <U extends TType, T extends TNumber> SparseTensorDenseAdd<U> create(Scope scope, Operand<T> aIndices, Operand<U> aValues, Operand<T> aShape, Operand<U> b) {
+    OperationBuilder opBuilder = scope.env().opBuilder("SparseTensorDenseAdd", scope.makeOpName("SparseTensorDenseAdd"));
+    opBuilder.addInput(aIndices.asOutput());
+    opBuilder.addInput(aValues.asOutput());
+    opBuilder.addInput(aShape.asOutput());
+    opBuilder.addInput(b.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new SparseTensorDenseAdd<U>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<U> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<U> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "SparseTensorDenseAdd";
+  
+  private Output<U> output;
+  
+  private SparseTensorDenseAdd(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseTensorDenseMatMul.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseTensorDenseMatMul.java
new file mode 100644
index 00000000000..63f3d3116c0
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseTensorDenseMatMul.java
@@ -0,0 +1,149 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.sparse;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Multiply SparseTensor (of rank 2) "A" by dense matrix "B".
+ * <p>
+ * No validity checking is performed on the indices of A.  However, the following
+ * input format is recommended for optimal behavior:
+ * <p>
+ * if adjoint_a == false:
+ *   A should be sorted in lexicographically increasing order.  Use SparseReorder
+ *   if you're not sure.
+ * if adjoint_a == true:
+ *   A should be sorted in order of increasing dimension 1 (i.e., "column major"
+ *   order instead of "row major" order).
+ * 
+ * @param <U> data type for {@code product()} output
+ */
+@Operator(group = "sparse")
+public final class SparseTensorDenseMatMul<U extends TType> extends RawOp implements Operand<U> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.sparse.SparseTensorDenseMatMul}
+   */
+  public static class Options {
+    
+    /**
+     * @param adjointA Use the adjoint of A in the matrix multiply.  If A is complex, this
+     * is transpose(conj(A)).  Otherwise it's transpose(A).
+     */
+    public Options adjointA(Boolean adjointA) {
+      this.adjointA = adjointA;
+      return this;
+    }
+    
+    /**
+     * @param adjointB Use the adjoint of B in the matrix multiply.  If B is complex, this
+     * is transpose(conj(B)).  Otherwise it's transpose(B).
+     */
+    public Options adjointB(Boolean adjointB) {
+      this.adjointB = adjointB;
+      return this;
+    }
+    
+    private Boolean adjointA;
+    private Boolean adjointB;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new SparseTensorDenseMatMul operation.
+   * 
+   * @param scope current scope
+   * @param aIndices 2-D.  The `indices` of the `SparseTensor`, size `[nnz, 2]` Matrix.
+   * @param aValues 1-D.  The `values` of the `SparseTensor`, size `[nnz]` Vector.
+   * @param aShape 1-D.  The `shape` of the `SparseTensor`, size `[2]` Vector.
+   * @param b 2-D.  A dense Matrix.
+   * @param options carries optional attributes values
+   * @return a new instance of SparseTensorDenseMatMul
+   */
+  @Endpoint(describeByClass = true)
+  public static <U extends TType, T extends TNumber> SparseTensorDenseMatMul<U> create(Scope scope, Operand<T> aIndices, Operand<U> aValues, Operand<TInt64> aShape, Operand<U> b, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("SparseTensorDenseMatMul", scope.makeOpName("SparseTensorDenseMatMul"));
+    opBuilder.addInput(aIndices.asOutput());
+    opBuilder.addInput(aValues.asOutput());
+    opBuilder.addInput(aShape.asOutput());
+    opBuilder.addInput(b.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.adjointA != null) {
+          opBuilder.setAttr("adjoint_a", opts.adjointA);
+        }
+        if (opts.adjointB != null) {
+          opBuilder.setAttr("adjoint_b", opts.adjointB);
+        }
+      }
+    }
+    return new SparseTensorDenseMatMul<U>(opBuilder.build());
+  }
+  
+  /**
+   * @param adjointA Use the adjoint of A in the matrix multiply.  If A is complex, this
+   * is transpose(conj(A)).  Otherwise it's transpose(A).
+   */
+  public static Options adjointA(Boolean adjointA) {
+    return new Options().adjointA(adjointA);
+  }
+  
+  /**
+   * @param adjointB Use the adjoint of B in the matrix multiply.  If B is complex, this
+   * is transpose(conj(B)).  Otherwise it's transpose(B).
+   */
+  public static Options adjointB(Boolean adjointB) {
+    return new Options().adjointB(adjointB);
+  }
+  
+  /**
+   */
+  public Output<U> product() {
+    return product;
+  }
+  
+  @Override
+  public Output<U> asOutput() {
+    return product;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "SparseTensorDenseMatMul";
+  
+  private Output<U> product;
+  
+  private SparseTensorDenseMatMul(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    product = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseToDense.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseToDense.java
new file mode 100644
index 00000000000..0f86af1327d
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseToDense.java
@@ -0,0 +1,139 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.sparse;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Converts a sparse representation into a dense tensor.
+ * <p>
+ * Builds an array `dense` with shape `output_shape` such that
+ * <pre>{@code
+ * # If sparse_indices is scalar
+ * dense[i] = (i == sparse_indices ? sparse_values : default_value)
+ * 
+ * # If sparse_indices is a vector, then for each i
+ * dense[sparse_indices[i]] = sparse_values[i]
+ * 
+ * # If sparse_indices is an n by d matrix, then for each i in [0, n)
+ * dense[sparse_indices[i][0], ..., sparse_indices[i][d-1]] = sparse_values[i]
+ * }</pre>
+ * All other values in `dense` are set to `default_value`.  If `sparse_values` is a
+ * scalar, all sparse indices are set to this single value.
+ * <p>
+ * Indices should be sorted in lexicographic order, and indices must not
+ * contain any repeats. If `validate_indices` is true, these properties
+ * are checked during execution.
+ * 
+ * @param <U> data type for {@code dense()} output
+ */
+@Operator(group = "sparse")
+public final class SparseToDense<U extends TType> extends RawOp implements Operand<U> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.sparse.SparseToDense}
+   */
+  public static class Options {
+    
+    /**
+     * @param validateIndices If true, indices are checked to make sure they are sorted in
+     * lexicographic order and that there are no repeats.
+     */
+    public Options validateIndices(Boolean validateIndices) {
+      this.validateIndices = validateIndices;
+      return this;
+    }
+    
+    private Boolean validateIndices;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new SparseToDense operation.
+   * 
+   * @param scope current scope
+   * @param sparseIndices 0-D, 1-D, or 2-D.  `sparse_indices[i]` contains the complete
+   * index where `sparse_values[i]` will be placed.
+   * @param outputShape 1-D.  Shape of the dense output tensor.
+   * @param sparseValues 1-D.  Values corresponding to each row of `sparse_indices`,
+   * or a scalar value to be used for all sparse indices.
+   * @param defaultValue Scalar value to set for indices not specified in
+   * `sparse_indices`.
+   * @param options carries optional attributes values
+   * @return a new instance of SparseToDense
+   */
+  @Endpoint(describeByClass = true)
+  public static <U extends TType, T extends TNumber> SparseToDense<U> create(Scope scope, Operand<T> sparseIndices, Operand<T> outputShape, Operand<U> sparseValues, Operand<U> defaultValue, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("SparseToDense", scope.makeOpName("SparseToDense"));
+    opBuilder.addInput(sparseIndices.asOutput());
+    opBuilder.addInput(outputShape.asOutput());
+    opBuilder.addInput(sparseValues.asOutput());
+    opBuilder.addInput(defaultValue.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.validateIndices != null) {
+          opBuilder.setAttr("validate_indices", opts.validateIndices);
+        }
+      }
+    }
+    return new SparseToDense<U>(opBuilder.build());
+  }
+  
+  /**
+   * @param validateIndices If true, indices are checked to make sure they are sorted in
+   * lexicographic order and that there are no repeats.
+   */
+  public static Options validateIndices(Boolean validateIndices) {
+    return new Options().validateIndices(validateIndices);
+  }
+  
+  /**
+   * Dense output tensor of shape `output_shape`.
+   */
+  public Output<U> dense() {
+    return dense;
+  }
+  
+  @Override
+  public Output<U> asOutput() {
+    return dense;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "SparseToDense";
+  
+  private Output<U> dense;
+  
+  private SparseToDense(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    dense = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseToSparseSetOperation.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseToSparseSetOperation.java
new file mode 100644
index 00000000000..31cbe67a914
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/SparseToSparseSetOperation.java
@@ -0,0 +1,169 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.sparse;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Applies set operation along last dimension of 2 `SparseTensor` inputs.
+ * <p>
+ * See SetOperationOp::SetOperationFromContext for values of `set_operation`.
+ * <p>
+ * If `validate_indices` is `True`, `sparse.SparseToSparseSetOperation` validates the
+ * order and range of `set1` and `set2` indices.
+ * <p>
+ * Input `set1` is a `SparseTensor` represented by `set1_indices`, `set1_values`,
+ * and `set1_shape`. For `set1` ranked `n`, 1st `n-1` dimensions must be the same
+ * as `set2`. Dimension `n` contains values in a set, duplicates are allowed but
+ * ignored.
+ * <p>
+ * Input `set2` is a `SparseTensor` represented by `set2_indices`, `set2_values`,
+ * and `set2_shape`. For `set2` ranked `n`, 1st `n-1` dimensions must be the same
+ * as `set1`. Dimension `n` contains values in a set, duplicates are allowed but
+ * ignored.
+ * <p>
+ * If `validate_indices` is `True`, this op validates the order and range of `set1`
+ * and `set2` indices.
+ * <p>
+ * Output `result` is a `SparseTensor` represented by `result_indices`,
+ * `result_values`, and `result_shape`. For `set1` and `set2` ranked `n`, this
+ * has rank `n` and the same 1st `n-1` dimensions as `set1` and `set2`. The `nth`
+ * dimension contains the result of `set_operation` applied to the corresponding
+ * `[0...n-1]` dimension of `set`.
+ * 
+ * @param <T> data type for {@code resultValues()} output
+ */
+@Operator(group = "sparse")
+public final class SparseToSparseSetOperation<T extends TType> extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.sparse.SparseToSparseSetOperation}
+   */
+  public static class Options {
+    
+    /**
+     * @param validateIndices 
+     */
+    public Options validateIndices(Boolean validateIndices) {
+      this.validateIndices = validateIndices;
+      return this;
+    }
+    
+    private Boolean validateIndices;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new SparseToSparseSetOperation operation.
+   * 
+   * @param scope current scope
+   * @param set1Indices 2D `Tensor`, indices of a `SparseTensor`. Must be in row-major
+   * order.
+   * @param set1Values 1D `Tensor`, values of a `SparseTensor`. Must be in row-major
+   * order.
+   * @param set1Shape 1D `Tensor`, shape of a `SparseTensor`. `set1_shape[0...n-1]` must
+   * be the same as `set2_shape[0...n-1]`, `set1_shape[n]` is the
+   * max set size across `0...n-1` dimensions.
+   * @param set2Indices 2D `Tensor`, indices of a `SparseTensor`. Must be in row-major
+   * order.
+   * @param set2Values 1D `Tensor`, values of a `SparseTensor`. Must be in row-major
+   * order.
+   * @param set2Shape 1D `Tensor`, shape of a `SparseTensor`. `set2_shape[0...n-1]` must
+   * be the same as `set1_shape[0...n-1]`, `set2_shape[n]` is the
+   * max set size across `0...n-1` dimensions.
+   * @param setOperation 
+   * @param options carries optional attributes values
+   * @return a new instance of SparseToSparseSetOperation
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> SparseToSparseSetOperation<T> create(Scope scope, Operand<TInt64> set1Indices, Operand<T> set1Values, Operand<TInt64> set1Shape, Operand<TInt64> set2Indices, Operand<T> set2Values, Operand<TInt64> set2Shape, String setOperation, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("SparseToSparseSetOperation", scope.makeOpName("SparseToSparseSetOperation"));
+    opBuilder.addInput(set1Indices.asOutput());
+    opBuilder.addInput(set1Values.asOutput());
+    opBuilder.addInput(set1Shape.asOutput());
+    opBuilder.addInput(set2Indices.asOutput());
+    opBuilder.addInput(set2Values.asOutput());
+    opBuilder.addInput(set2Shape.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("set_operation", setOperation);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.validateIndices != null) {
+          opBuilder.setAttr("validate_indices", opts.validateIndices);
+        }
+      }
+    }
+    return new SparseToSparseSetOperation<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param validateIndices 
+   */
+  public static Options validateIndices(Boolean validateIndices) {
+    return new Options().validateIndices(validateIndices);
+  }
+  
+  /**
+   * 2D indices of a `SparseTensor`.
+   */
+  public Output<TInt64> resultIndices() {
+    return resultIndices;
+  }
+  
+  /**
+   * 1D values of a `SparseTensor`.
+   */
+  public Output<T> resultValues() {
+    return resultValues;
+  }
+  
+  /**
+   * 1D `Tensor` shape of a `SparseTensor`. `result_shape[0...n-1]` is
+   * the same as the 1st `n-1` dimensions of `set1` and `set2`, `result_shape[n]`
+   * is the max result set size across all `0...n-1` dimensions.
+   */
+  public Output<TInt64> resultShape() {
+    return resultShape;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "SparseToSparseSetOperation";
+  
+  private Output<TInt64> resultIndices;
+  private Output<T> resultValues;
+  private Output<TInt64> resultShape;
+  
+  private SparseToSparseSetOperation(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    resultIndices = operation.output(outputIdx++);
+    resultValues = operation.output(outputIdx++);
+    resultShape = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/TakeManySparseFromTensorsMap.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/TakeManySparseFromTensorsMap.java
new file mode 100644
index 00000000000..6e1bbb6ed27
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/sparse/TakeManySparseFromTensorsMap.java
@@ -0,0 +1,195 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.sparse;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Read `SparseTensors` from a `SparseTensorsMap` and concatenate them.
+ * <p>
+ * The input `sparse_handles` must be an `int64` matrix of shape `[N, 1]` where
+ * `N` is the minibatch size and the rows correspond to the output handles of
+ * `AddSparseToTensorsMap` or `AddManySparseToTensorsMap`.  The ranks of the
+ * original `SparseTensor` objects that went into the given input ops must all
+ * match.  When the final `SparseTensor` is created, it has rank one
+ * higher than the ranks of the incoming `SparseTensor` objects
+ * (they have been concatenated along a new row dimension on the left).
+ * <p>
+ * The output `SparseTensor` object's shape values for all dimensions but the
+ * first are the max across the input `SparseTensor` objects' shape values
+ * for the corresponding dimensions.  Its first shape value is `N`, the minibatch
+ * size.
+ * <p>
+ * The input `SparseTensor` objects' indices are assumed ordered in
+ * standard lexicographic order.  If this is not the case, after this
+ * step run `SparseReorder` to restore index ordering.
+ * <p>
+ * For example, if the handles represent an input, which is a `[2, 3]` matrix
+ * representing two original `SparseTensor` objects:
+ * <pre>{@code
+ *     index = [ 0]
+ *             [10]
+ *             [20]
+ *     values = [1, 2, 3]
+ *     shape = [50]
+ * }</pre>
+ * and
+ * <pre>{@code
+ *     index = [ 2]
+ *             [10]
+ *     values = [4, 5]
+ *     shape = [30]
+ * }</pre>
+ * then the final `SparseTensor` will be:
+ * <pre>{@code
+ *     index = [0  0]
+ *             [0 10]
+ *             [0 20]
+ *             [1  2]
+ *             [1 10]
+ *     values = [1, 2, 3, 4, 5]
+ *     shape = [2 50]
+ * }</pre>
+ * 
+ * 
+ * @param <T> data type for {@code sparseValues()} output
+ */
+@Operator(group = "sparse")
+public final class TakeManySparseFromTensorsMap<T extends TType> extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.sparse.TakeManySparseFromTensorsMap}
+   */
+  public static class Options {
+    
+    /**
+     * @param container The container name for the `SparseTensorsMap` read by this op.
+     */
+    public Options container(String container) {
+      this.container = container;
+      return this;
+    }
+    
+    /**
+     * @param sharedName The shared name for the `SparseTensorsMap` read by this op.
+     * It should not be blank; rather the `shared_name` or unique Operation name
+     * of the Op that created the original `SparseTensorsMap` should be used.
+     */
+    public Options sharedName(String sharedName) {
+      this.sharedName = sharedName;
+      return this;
+    }
+    
+    private String container;
+    private String sharedName;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new TakeManySparseFromTensorsMap operation.
+   * 
+   * @param scope current scope
+   * @param sparseHandles 1-D, The `N` serialized `SparseTensor` objects.
+   * Shape: `[N]`.
+   * @param dtype The `dtype` of the `SparseTensor` objects stored in the
+   * `SparseTensorsMap`.
+   * @param options carries optional attributes values
+   * @return a new instance of TakeManySparseFromTensorsMap
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> TakeManySparseFromTensorsMap<T> create(Scope scope, Operand<TInt64> sparseHandles, Class<T> dtype, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("TakeManySparseFromTensorsMap", scope.makeOpName("TakeManySparseFromTensorsMap"));
+    opBuilder.addInput(sparseHandles.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("dtype", dtype);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.container != null) {
+          opBuilder.setAttr("container", opts.container);
+        }
+        if (opts.sharedName != null) {
+          opBuilder.setAttr("shared_name", opts.sharedName);
+        }
+      }
+    }
+    return new TakeManySparseFromTensorsMap<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param container The container name for the `SparseTensorsMap` read by this op.
+   */
+  public static Options container(String container) {
+    return new Options().container(container);
+  }
+  
+  /**
+   * @param sharedName The shared name for the `SparseTensorsMap` read by this op.
+   * It should not be blank; rather the `shared_name` or unique Operation name
+   * of the Op that created the original `SparseTensorsMap` should be used.
+   */
+  public static Options sharedName(String sharedName) {
+    return new Options().sharedName(sharedName);
+  }
+  
+  /**
+   * 2-D.  The `indices` of the minibatch `SparseTensor`.
+   */
+  public Output<TInt64> sparseIndices() {
+    return sparseIndices;
+  }
+  
+  /**
+   * 1-D.  The `values` of the minibatch `SparseTensor`.
+   */
+  public Output<T> sparseValues() {
+    return sparseValues;
+  }
+  
+  /**
+   * 1-D.  The `shape` of the minibatch `SparseTensor`.
+   */
+  public Output<TInt64> sparseShape() {
+    return sparseShape;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "TakeManySparseFromTensorsMap";
+  
+  private Output<TInt64> sparseIndices;
+  private Output<T> sparseValues;
+  private Output<TInt64> sparseShape;
+  
+  private TakeManySparseFromTensorsMap(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    sparseIndices = operation.output(outputIdx++);
+    sparseValues = operation.output(outputIdx++);
+    sparseShape = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/Join.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/Join.java
new file mode 100644
index 00000000000..633457793a3
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/Join.java
@@ -0,0 +1,117 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.strings;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.Operands;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TString;
+
+/**
+ * Joins the strings in the given list of string tensors into one tensor;
+ * <p>
+ * with the given separator (default is an empty separator).
+ * <p>
+ * Examples:
+ * <p>
+ * >>> s = ["hello", "world", "tensorflow"]
+ * >>> tf.strings.join(s, " ")
+ * <tf.Tensor: shape=(), dtype=string, numpy=b'hello world tensorflow'>
+ */
+@Operator(group = "strings")
+public final class Join extends RawOp implements Operand<TString> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.strings.Join}
+   */
+  public static class Options {
+    
+    /**
+     * @param separator string, an optional join separator.
+     */
+    public Options separator(String separator) {
+      this.separator = separator;
+      return this;
+    }
+    
+    private String separator;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new Join operation.
+   * 
+   * @param scope current scope
+   * @param inputs A list of string tensors.  The tensors must all have the same shape,
+   * or be scalars.  Scalars may be mixed in; these will be broadcast to the shape
+   * of non-scalar inputs.
+   * @param options carries optional attributes values
+   * @return a new instance of Join
+   */
+  @Endpoint(describeByClass = true)
+  public static Join create(Scope scope, Iterable<Operand<TString>> inputs, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("StringJoin", scope.makeOpName("Join"));
+    opBuilder.addInputList(Operands.asOutputs(inputs));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.separator != null) {
+          opBuilder.setAttr("separator", opts.separator);
+        }
+      }
+    }
+    return new Join(opBuilder.build());
+  }
+  
+  /**
+   * @param separator string, an optional join separator.
+   */
+  public static Options separator(String separator) {
+    return new Options().separator(separator);
+  }
+  
+  /**
+   */
+  public Output<TString> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<TString> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "StringJoin";
+  
+  private Output<TString> output;
+  
+  private Join(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/Lower.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/Lower.java
new file mode 100644
index 00000000000..0b214f47017
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/Lower.java
@@ -0,0 +1,112 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.strings;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TString;
+
+/**
+ * Converts all uppercase characters into their respective lowercase replacements.
+ * <p>
+ * Example:
+ * <p>
+ * >>> tf.strings.lower("CamelCase string and ALL CAPS")
+ * <tf.Tensor: shape=(), dtype=string, numpy=b'camelcase string and all caps'>
+ * 
+ */
+@Operator(group = "strings")
+public final class Lower extends RawOp implements Operand<TString> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.strings.Lower}
+   */
+  public static class Options {
+    
+    /**
+     * @param encoding 
+     */
+    public Options encoding(String encoding) {
+      this.encoding = encoding;
+      return this;
+    }
+    
+    private String encoding;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new Lower operation.
+   * 
+   * @param scope current scope
+   * @param input 
+   * @param options carries optional attributes values
+   * @return a new instance of Lower
+   */
+  @Endpoint(describeByClass = true)
+  public static Lower create(Scope scope, Operand<TString> input, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("StringLower", scope.makeOpName("Lower"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.encoding != null) {
+          opBuilder.setAttr("encoding", opts.encoding);
+        }
+      }
+    }
+    return new Lower(opBuilder.build());
+  }
+  
+  /**
+   * @param encoding 
+   */
+  public static Options encoding(String encoding) {
+    return new Options().encoding(encoding);
+  }
+  
+  /**
+   */
+  public Output<TString> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<TString> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "StringLower";
+  
+  private Output<TString> output;
+  
+  private Lower(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/ReduceJoin.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/ReduceJoin.java
new file mode 100644
index 00000000000..23d90b3a2b1
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/ReduceJoin.java
@@ -0,0 +1,156 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.strings;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.TString;
+
+/**
+ * Joins a string Tensor across the given dimensions.
+ * <p>
+ * Computes the string join across dimensions in the given string Tensor of shape
+ * `[\\(d_0, d_1, ..., d_{n-1}\\)]`.  Returns a new Tensor created by joining the input
+ * strings with the given separator (default: empty string).  Negative indices are
+ * counted backwards from the end, with `-1` being equivalent to `n - 1`.  If
+ * indices are not specified, joins across all dimensions beginning from `n - 1`
+ * through `0`.
+ * <p>
+ * For example:
+ * <pre>{@code
+ * # tensor `a` is [["a", "b"], ["c", "d"]]
+ * tf.reduce_join(a, 0) ==> ["ac", "bd"]
+ * tf.reduce_join(a, 1) ==> ["ab", "cd"]
+ * tf.reduce_join(a, -2) = tf.reduce_join(a, 0) ==> ["ac", "bd"]
+ * tf.reduce_join(a, -1) = tf.reduce_join(a, 1) ==> ["ab", "cd"]
+ * tf.reduce_join(a, 0, keep_dims=True) ==> [["ac", "bd"]]
+ * tf.reduce_join(a, 1, keep_dims=True) ==> [["ab"], ["cd"]]
+ * tf.reduce_join(a, 0, separator=".") ==> ["a.c", "b.d"]
+ * tf.reduce_join(a, [0, 1]) ==> "acbd"
+ * tf.reduce_join(a, [1, 0]) ==> "abcd"
+ * tf.reduce_join(a, []) ==> [["a", "b"], ["c", "d"]]
+ * tf.reduce_join(a) = tf.reduce_join(a, [1, 0]) ==> "abcd"
+ * }</pre>
+ * 
+ */
+@Operator(group = "strings")
+public final class ReduceJoin extends RawOp implements Operand<TString> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.strings.ReduceJoin}
+   */
+  public static class Options {
+    
+    /**
+     * @param keepDims If `True`, retain reduced dimensions with length `1`.
+     */
+    public Options keepDims(Boolean keepDims) {
+      this.keepDims = keepDims;
+      return this;
+    }
+    
+    /**
+     * @param separator The separator to use when joining.
+     */
+    public Options separator(String separator) {
+      this.separator = separator;
+      return this;
+    }
+    
+    private Boolean keepDims;
+    private String separator;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new ReduceJoin operation.
+   * 
+   * @param scope current scope
+   * @param inputs The input to be joined.  All reduced indices must have non-zero size.
+   * @param reductionIndices The dimensions to reduce over.  Dimensions are reduced in the
+   * order specified.  Omitting `reduction_indices` is equivalent to passing
+   * `[n-1, n-2, ..., 0]`.  Negative indices from `-n` to `-1` are supported.
+   * @param options carries optional attributes values
+   * @return a new instance of ReduceJoin
+   */
+  @Endpoint(describeByClass = true)
+  public static ReduceJoin create(Scope scope, Operand<TString> inputs, Operand<TInt32> reductionIndices, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ReduceJoin", scope.makeOpName("ReduceJoin"));
+    opBuilder.addInput(inputs.asOutput());
+    opBuilder.addInput(reductionIndices.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.keepDims != null) {
+          opBuilder.setAttr("keep_dims", opts.keepDims);
+        }
+        if (opts.separator != null) {
+          opBuilder.setAttr("separator", opts.separator);
+        }
+      }
+    }
+    return new ReduceJoin(opBuilder.build());
+  }
+  
+  /**
+   * @param keepDims If `True`, retain reduced dimensions with length `1`.
+   */
+  public static Options keepDims(Boolean keepDims) {
+    return new Options().keepDims(keepDims);
+  }
+  
+  /**
+   * @param separator The separator to use when joining.
+   */
+  public static Options separator(String separator) {
+    return new Options().separator(separator);
+  }
+  
+  /**
+   * Has shape equal to that of the input with reduced dimensions removed or
+   * set to `1` depending on `keep_dims`.
+   */
+  public Output<TString> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<TString> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ReduceJoin";
+  
+  private Output<TString> output;
+  
+  private ReduceJoin(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/RegexFullMatch.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/RegexFullMatch.java
new file mode 100644
index 00000000000..446166c6e29
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/RegexFullMatch.java
@@ -0,0 +1,90 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.strings;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TBool;
+import org.tensorflow.types.TString;
+
+/**
+ * Check if the input matches the regex pattern.
+ * <p>
+ * The input is a string tensor of any shape. The pattern is a scalar
+ * string tensor which is applied to every element of the input tensor.
+ * The boolean values (True or False) of the output tensor indicate
+ * if the input matches the regex pattern provided.
+ * <p>
+ * The pattern follows the re2 syntax (https://github.com/google/re2/wiki/Syntax)
+ * <p>
+ * Examples:
+ * <p>
+ * >>> tf.strings.regex_full_match(["TF lib", "lib TF"], ".*lib$")
+ * <tf.Tensor: shape=(2,), dtype=bool, numpy=array([ True, False])>
+ * >>> tf.strings.regex_full_match(["TF lib", "lib TF"], ".*TF$")
+ * <tf.Tensor: shape=(2,), dtype=bool, numpy=array([False,  True])>
+ */
+@Operator(group = "strings")
+public final class RegexFullMatch extends RawOp implements Operand<TBool> {
+  
+  /**
+   * Factory method to create a class wrapping a new RegexFullMatch operation.
+   * 
+   * @param scope current scope
+   * @param input A string tensor of the text to be processed.
+   * @param pattern A scalar string tensor containing the regular expression to match the input.
+   * @return a new instance of RegexFullMatch
+   */
+  @Endpoint(describeByClass = true)
+  public static RegexFullMatch create(Scope scope, Operand<TString> input, Operand<TString> pattern) {
+    OperationBuilder opBuilder = scope.env().opBuilder("RegexFullMatch", scope.makeOpName("RegexFullMatch"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(pattern.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new RegexFullMatch(opBuilder.build());
+  }
+  
+  /**
+   * A bool tensor with the same shape as `input`.
+   */
+  public Output<TBool> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<TBool> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "RegexFullMatch";
+  
+  private Output<TBool> output;
+  
+  private RegexFullMatch(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/RegexReplace.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/RegexReplace.java
new file mode 100644
index 00000000000..3eaec903fee
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/RegexReplace.java
@@ -0,0 +1,119 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.strings;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TString;
+
+/**
+ * Replaces matches of the `pattern` regular expression in `input` with the
+ * replacement string provided in `rewrite`.
+ * <p>
+ * It follows the re2 syntax (https://github.com/google/re2/wiki/Syntax)
+ */
+@Operator(group = "strings")
+public final class RegexReplace extends RawOp implements Operand<TString> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.strings.RegexReplace}
+   */
+  public static class Options {
+    
+    /**
+     * @param replaceGlobal If True, the replacement is global (that is, all matches of the `pattern` regular
+     * expression in each input string are rewritten), otherwise the `rewrite`
+     * substitution is only made for the first `pattern` match.
+     */
+    public Options replaceGlobal(Boolean replaceGlobal) {
+      this.replaceGlobal = replaceGlobal;
+      return this;
+    }
+    
+    private Boolean replaceGlobal;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new RegexReplace operation.
+   * 
+   * @param scope current scope
+   * @param input The text to be processed.
+   * @param pattern The regular expression to be matched in the `input` strings.
+   * @param rewrite The rewrite string to be substituted for the `pattern` expression where it is
+   * matched in the `input` strings.
+   * @param options carries optional attributes values
+   * @return a new instance of RegexReplace
+   */
+  @Endpoint(describeByClass = true)
+  public static RegexReplace create(Scope scope, Operand<TString> input, Operand<TString> pattern, Operand<TString> rewrite, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("RegexReplace", scope.makeOpName("RegexReplace"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(pattern.asOutput());
+    opBuilder.addInput(rewrite.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.replaceGlobal != null) {
+          opBuilder.setAttr("replace_global", opts.replaceGlobal);
+        }
+      }
+    }
+    return new RegexReplace(opBuilder.build());
+  }
+  
+  /**
+   * @param replaceGlobal If True, the replacement is global (that is, all matches of the `pattern` regular
+   * expression in each input string are rewritten), otherwise the `rewrite`
+   * substitution is only made for the first `pattern` match.
+   */
+  public static Options replaceGlobal(Boolean replaceGlobal) {
+    return new Options().replaceGlobal(replaceGlobal);
+  }
+  
+  /**
+   * The text after applying pattern match and rewrite substitution.
+   */
+  public Output<TString> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<TString> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "RegexReplace";
+  
+  private Output<TString> output;
+  
+  private RegexReplace(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/StaticRegexFullMatch.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/StaticRegexFullMatch.java
new file mode 100644
index 00000000000..3aa7b0827eb
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/StaticRegexFullMatch.java
@@ -0,0 +1,82 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.strings;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TBool;
+import org.tensorflow.types.TString;
+
+/**
+ * Check if the input matches the regex pattern.
+ * <p>
+ * The input is a string tensor of any shape. The pattern is the
+ * regular expression to be matched with every element of the input tensor.
+ * The boolean values (True or False) of the output tensor indicate
+ * if the input matches the regex pattern provided.
+ * <p>
+ * The pattern follows the re2 syntax (https://github.com/google/re2/wiki/Syntax)
+ */
+public final class StaticRegexFullMatch extends RawOp implements Operand<TBool> {
+  
+  /**
+   * Factory method to create a class wrapping a new StaticRegexFullMatch operation.
+   * 
+   * @param scope current scope
+   * @param input A string tensor of the text to be processed.
+   * @param pattern The regular expression to match the input.
+   * @return a new instance of StaticRegexFullMatch
+   */
+  @Endpoint(describeByClass = true)
+  public static StaticRegexFullMatch create(Scope scope, Operand<TString> input, String pattern) {
+    OperationBuilder opBuilder = scope.env().opBuilder("StaticRegexFullMatch", scope.makeOpName("StaticRegexFullMatch"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("pattern", pattern);
+    return new StaticRegexFullMatch(opBuilder.build());
+  }
+  
+  /**
+   * A bool tensor with the same shape as `input`.
+   */
+  public Output<TBool> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<TBool> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "StaticRegexFullMatch";
+  
+  private Output<TBool> output;
+  
+  private StaticRegexFullMatch(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/StaticRegexReplace.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/StaticRegexReplace.java
new file mode 100644
index 00000000000..c1802f48001
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/StaticRegexReplace.java
@@ -0,0 +1,114 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.strings;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TString;
+
+/**
+ * Replaces the match of pattern in input with rewrite.
+ * <p>
+ * It follows the re2 syntax (https://github.com/google/re2/wiki/Syntax)
+ */
+public final class StaticRegexReplace extends RawOp implements Operand<TString> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.strings.StaticRegexReplace}
+   */
+  public static class Options {
+    
+    /**
+     * @param replaceGlobal If True, the replacement is global, otherwise the replacement
+     * is done only on the first match.
+     */
+    public Options replaceGlobal(Boolean replaceGlobal) {
+      this.replaceGlobal = replaceGlobal;
+      return this;
+    }
+    
+    private Boolean replaceGlobal;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new StaticRegexReplace operation.
+   * 
+   * @param scope current scope
+   * @param input The text to be processed.
+   * @param pattern The regular expression to match the input.
+   * @param rewrite The rewrite to be applied to the matched expression.
+   * @param options carries optional attributes values
+   * @return a new instance of StaticRegexReplace
+   */
+  @Endpoint(describeByClass = true)
+  public static StaticRegexReplace create(Scope scope, Operand<TString> input, String pattern, String rewrite, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("StaticRegexReplace", scope.makeOpName("StaticRegexReplace"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("pattern", pattern);
+    opBuilder.setAttr("rewrite", rewrite);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.replaceGlobal != null) {
+          opBuilder.setAttr("replace_global", opts.replaceGlobal);
+        }
+      }
+    }
+    return new StaticRegexReplace(opBuilder.build());
+  }
+  
+  /**
+   * @param replaceGlobal If True, the replacement is global, otherwise the replacement
+   * is done only on the first match.
+   */
+  public static Options replaceGlobal(Boolean replaceGlobal) {
+    return new Options().replaceGlobal(replaceGlobal);
+  }
+  
+  /**
+   * The text after applying pattern and rewrite.
+   */
+  public Output<TString> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<TString> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "StaticRegexReplace";
+  
+  private Output<TString> output;
+  
+  private StaticRegexReplace(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/StringFormat.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/StringFormat.java
new file mode 100644
index 00000000000..3decf1e5c93
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/StringFormat.java
@@ -0,0 +1,148 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.strings;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.Operands;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TString;
+
+/**
+ * Formats a string template using a list of tensors.
+ * <p>
+ * Formats a string template using a list of tensors, pretty-printing tensor summaries.
+ */
+@Operator(group = "strings")
+public final class StringFormat extends RawOp implements Operand<TString> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.strings.StringFormat}
+   */
+  public static class Options {
+    
+    /**
+     * @param template A string, the template to format tensor summaries into.
+     */
+    public Options template(String template) {
+      this.template = template;
+      return this;
+    }
+    
+    /**
+     * @param placeholder A string, at each placeholder in the template a subsequent tensor summary will be inserted.
+     */
+    public Options placeholder(String placeholder) {
+      this.placeholder = placeholder;
+      return this;
+    }
+    
+    /**
+     * @param summarize When formatting the tensor summaries print the first and last summarize entries of each tensor dimension.
+     */
+    public Options summarize(Long summarize) {
+      this.summarize = summarize;
+      return this;
+    }
+    
+    private String template;
+    private String placeholder;
+    private Long summarize;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new StringFormat operation.
+   * 
+   * @param scope current scope
+   * @param inputs The list of tensors to format into the placeholder string.
+   * @param options carries optional attributes values
+   * @return a new instance of StringFormat
+   */
+  @Endpoint(describeByClass = true)
+  public static StringFormat create(Scope scope, Iterable<Operand<?>> inputs, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("StringFormat", scope.makeOpName("StringFormat"));
+    opBuilder.addInputList(Operands.asOutputs(inputs));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.template != null) {
+          opBuilder.setAttr("template", opts.template);
+        }
+        if (opts.placeholder != null) {
+          opBuilder.setAttr("placeholder", opts.placeholder);
+        }
+        if (opts.summarize != null) {
+          opBuilder.setAttr("summarize", opts.summarize);
+        }
+      }
+    }
+    return new StringFormat(opBuilder.build());
+  }
+  
+  /**
+   * @param template A string, the template to format tensor summaries into.
+   */
+  public static Options template(String template) {
+    return new Options().template(template);
+  }
+  
+  /**
+   * @param placeholder A string, at each placeholder in the template a subsequent tensor summary will be inserted.
+   */
+  public static Options placeholder(String placeholder) {
+    return new Options().placeholder(placeholder);
+  }
+  
+  /**
+   * @param summarize When formatting the tensor summaries print the first and last summarize entries of each tensor dimension.
+   */
+  public static Options summarize(Long summarize) {
+    return new Options().summarize(summarize);
+  }
+  
+  /**
+   * = The resulting string scalar.
+   */
+  public Output<TString> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<TString> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "StringFormat";
+  
+  private Output<TString> output;
+  
+  private StringFormat(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/StringLength.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/StringLength.java
new file mode 100644
index 00000000000..d9fc036ad22
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/StringLength.java
@@ -0,0 +1,126 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.strings;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.TString;
+
+/**
+ * String lengths of `input`.
+ * <p>
+ * Computes the length of each string given in the input tensor.
+ * <p>
+ * >>> strings = tf.constant(['Hello','TensorFlow', '\U0001F642'])
+ * >>> tf.strings.length(strings).numpy() # default counts bytes
+ * array([ 5, 10, 4], dtype=int32)
+ * >>> tf.strings.length(strings, unit="UTF8_CHAR").numpy()
+ * array([ 5, 10, 1], dtype=int32)
+ * 
+ */
+@Operator(group = "strings")
+public final class StringLength extends RawOp implements Operand<TInt32> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.strings.StringLength}
+   */
+  public static class Options {
+    
+    /**
+     * @param unit The unit that is counted to compute string length.  One of: `"BYTE"` (for
+     * the number of bytes in each string) or `"UTF8_CHAR"` (for the number of UTF-8
+     * encoded Unicode code points in each string).  Results are undefined
+     * if `unit=UTF8_CHAR` and the `input` strings do not contain structurally
+     * valid UTF-8.
+     */
+    public Options unit(String unit) {
+      this.unit = unit;
+      return this;
+    }
+    
+    private String unit;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new StringLength operation.
+   * 
+   * @param scope current scope
+   * @param input The strings for which to compute the length for each element.
+   * @param options carries optional attributes values
+   * @return a new instance of StringLength
+   */
+  @Endpoint(describeByClass = true)
+  public static StringLength create(Scope scope, Operand<TString> input, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("StringLength", scope.makeOpName("StringLength"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.unit != null) {
+          opBuilder.setAttr("unit", opts.unit);
+        }
+      }
+    }
+    return new StringLength(opBuilder.build());
+  }
+  
+  /**
+   * @param unit The unit that is counted to compute string length.  One of: `"BYTE"` (for
+   * the number of bytes in each string) or `"UTF8_CHAR"` (for the number of UTF-8
+   * encoded Unicode code points in each string).  Results are undefined
+   * if `unit=UTF8_CHAR` and the `input` strings do not contain structurally
+   * valid UTF-8.
+   */
+  public static Options unit(String unit) {
+    return new Options().unit(unit);
+  }
+  
+  /**
+   * Integer tensor that has the same shape as `input`. The output contains the
+   * element-wise string lengths of `input`.
+   */
+  public Output<TInt32> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<TInt32> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "StringLength";
+  
+  private Output<TInt32> output;
+  
+  private StringLength(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/StringNGrams.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/StringNGrams.java
new file mode 100644
index 00000000000..93d8b60ad87
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/StringNGrams.java
@@ -0,0 +1,109 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.strings;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TString;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Creates ngrams from ragged string data.
+ * <p>
+ * This op accepts a ragged tensor with 1 ragged dimension containing only
+ * strings and outputs a ragged tensor with 1 ragged dimension containing ngrams
+ * of that string, joined along the innermost axis.
+ * 
+ * @param <T> data type for {@code ngramsSplits()} output
+ */
+@Operator(group = "strings")
+public final class StringNGrams<T extends TNumber> extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new StringNGrams operation.
+   * 
+   * @param scope current scope
+   * @param data The values tensor of the ragged string tensor to make ngrams out of. Must be a
+   * 1D string tensor.
+   * @param dataSplits The splits tensor of the ragged string tensor to make ngrams out of.
+   * @param separator The string to append between elements of the token. Use "" for no separator.
+   * @param ngramWidths The sizes of the ngrams to create.
+   * @param leftPad The string to use to pad the left side of the ngram sequence. Only used if
+   * pad_width != 0.
+   * @param rightPad The string to use to pad the right side of the ngram sequence. Only used if
+   * pad_width != 0.
+   * @param padWidth The number of padding elements to add to each side of each
+   * sequence. Note that padding will never be greater than 'ngram_widths'-1
+   * regardless of this value. If `pad_width=-1`, then add `max(ngram_widths)-1`
+   * elements.
+   * @param preserveShortSequences 
+   * @return a new instance of StringNGrams
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> StringNGrams<T> create(Scope scope, Operand<TString> data, Operand<T> dataSplits, String separator, List<Long> ngramWidths, String leftPad, String rightPad, Long padWidth, Boolean preserveShortSequences) {
+    OperationBuilder opBuilder = scope.env().opBuilder("StringNGrams", scope.makeOpName("StringNGrams"));
+    opBuilder.addInput(data.asOutput());
+    opBuilder.addInput(dataSplits.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("separator", separator);
+    long[] ngramWidthsArray = new long[ngramWidths.size()];
+    for (int i = 0; i < ngramWidthsArray.length; ++i) {
+      ngramWidthsArray[i] = ngramWidths.get(i);
+    }
+    opBuilder.setAttr("ngram_widths", ngramWidthsArray);
+    opBuilder.setAttr("left_pad", leftPad);
+    opBuilder.setAttr("right_pad", rightPad);
+    opBuilder.setAttr("pad_width", padWidth);
+    opBuilder.setAttr("preserve_short_sequences", preserveShortSequences);
+    return new StringNGrams<T>(opBuilder.build());
+  }
+  
+  /**
+   * The values tensor of the output ngrams ragged tensor.
+   */
+  public Output<TString> ngrams() {
+    return ngrams;
+  }
+  
+  /**
+   * The splits tensor of the output ngrams ragged tensor.
+   */
+  public Output<T> ngramsSplits() {
+    return ngramsSplits;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "StringNGrams";
+  
+  private Output<TString> ngrams;
+  private Output<T> ngramsSplits;
+  
+  private StringNGrams(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    ngrams = operation.output(outputIdx++);
+    ngramsSplits = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/StringSplit.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/StringSplit.java
new file mode 100644
index 00000000000..87debf546cf
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/StringSplit.java
@@ -0,0 +1,144 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.strings;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.TString;
+
+/**
+ * Split elements of `source` based on `sep` into a `SparseTensor`.
+ * <p>
+ * Let N be the size of source (typically N will be the batch size). Split each
+ * element of `source` based on `sep` and return a `SparseTensor`
+ * containing the split tokens. Empty tokens are ignored.
+ * <p>
+ * For example, N = 2, source[0] is 'hello world' and source[1] is 'a b c',
+ * then the output will be
+ * <pre>{@code
+ * st.indices = [0, 0;
+ *               0, 1;
+ *               1, 0;
+ *               1, 1;
+ *               1, 2]
+ * st.shape = [2, 3]
+ * st.values = ['hello', 'world', 'a', 'b', 'c']
+ * }</pre>
+ * If `sep` is given, consecutive delimiters are not grouped together and are
+ * deemed to delimit empty strings. For example, source of `"1<>2<><>3"` and
+ * sep of `"<>"` returns `["1", "2", "", "3"]`. If `sep` is None or an empty
+ * string, consecutive whitespace are regarded as a single separator, and the
+ * result will contain no empty strings at the startor end if the string has
+ * leading or trailing whitespace.
+ * <p>
+ * Note that the above mentioned behavior matches python's str.split.
+ */
+@Operator(group = "strings")
+public final class StringSplit extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.strings.StringSplit}
+   */
+  public static class Options {
+    
+    /**
+     * @param maxsplit An `int`. If `maxsplit > 0`, limit of the split of the result.
+     */
+    public Options maxsplit(Long maxsplit) {
+      this.maxsplit = maxsplit;
+      return this;
+    }
+    
+    private Long maxsplit;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new StringSplit operation.
+   * 
+   * @param scope current scope
+   * @param input `1-D` string `Tensor`, the strings to split.
+   * @param sep `0-D` string `Tensor`, the delimiter character.
+   * @param options carries optional attributes values
+   * @return a new instance of StringSplit
+   */
+  @Endpoint(describeByClass = true)
+  public static StringSplit create(Scope scope, Operand<TString> input, Operand<TString> sep, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("StringSplitV2", scope.makeOpName("StringSplit"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(sep.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.maxsplit != null) {
+          opBuilder.setAttr("maxsplit", opts.maxsplit);
+        }
+      }
+    }
+    return new StringSplit(opBuilder.build());
+  }
+  
+  /**
+   * @param maxsplit An `int`. If `maxsplit > 0`, limit of the split of the result.
+   */
+  public static Options maxsplit(Long maxsplit) {
+    return new Options().maxsplit(maxsplit);
+  }
+  
+  /**
+   */
+  public Output<TInt64> indices() {
+    return indices;
+  }
+  
+  /**
+   */
+  public Output<TString> values() {
+    return values;
+  }
+  
+  /**
+   */
+  public Output<TInt64> shape() {
+    return shape;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "StringSplitV2";
+  
+  private Output<TInt64> indices;
+  private Output<TString> values;
+  private Output<TInt64> shape;
+  
+  private StringSplit(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    indices = operation.output(outputIdx++);
+    values = operation.output(outputIdx++);
+    shape = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/Strip.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/Strip.java
new file mode 100644
index 00000000000..fc6f37bbabb
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/Strip.java
@@ -0,0 +1,78 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.strings;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TString;
+
+/**
+ * Strip leading and trailing whitespaces from the Tensor.
+ */
+@Operator(group = "strings")
+public final class Strip extends RawOp implements Operand<TString> {
+  
+  /**
+   * Factory method to create a class wrapping a new Strip operation.
+   * 
+   * @param scope current scope
+   * @param input A string `Tensor` of any shape.
+   * @return a new instance of Strip
+   */
+  @Endpoint(describeByClass = true)
+  public static Strip create(Scope scope, Operand<TString> input) {
+    OperationBuilder opBuilder = scope.env().opBuilder("StringStrip", scope.makeOpName("Strip"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new Strip(opBuilder.build());
+  }
+  
+  /**
+   * A string `Tensor` of the same shape as the input.
+   * <p>
+   * Examples:
+   * <p>
+   * >>> tf.strings.strip(["\nTensorFlow", "     The python library    "]).numpy()
+   * array([b'TensorFlow', b'The python library'], dtype=object)
+   */
+  public Output<TString> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<TString> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "StringStrip";
+  
+  private Output<TString> output;
+  
+  private Strip(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/Substr.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/Substr.java
new file mode 100644
index 00000000000..d09e93458ba
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/Substr.java
@@ -0,0 +1,196 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.strings;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TString;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Return substrings from `Tensor` of strings.
+ * <p>
+ * For each string in the input `Tensor`, creates a substring starting at index
+ * `pos` with a total length of `len`.
+ * <p>
+ * If `len` defines a substring that would extend beyond the length of the input
+ * string, or if `len` is negative, then as many characters as possible are used.
+ * <p>
+ * A negative `pos` indicates distance within the string backwards from the end.
+ * <p>
+ * If `pos` specifies an index which is out of range for any of the input strings,
+ * then an `InvalidArgumentError` is thrown.
+ * <p>
+ * `pos` and `len` must have the same shape, otherwise a `ValueError` is thrown on
+ * Op creation.
+ * <p>
+ * <i>NOTE</i>: `strings.Substr` supports broadcasting up to two dimensions. More about
+ * broadcasting
+ * [here](http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html)
+ * <p>
+ * ---
+ * <p>
+ * Examples
+ * <p>
+ * Using scalar `pos` and `len`:
+ * <pre>{@code
+ * input = [b'Hello', b'World']
+ * position = 1
+ * length = 3
+ * 
+ * output = [b'ell', b'orl']
+ * }</pre>
+ * Using `pos` and `len` with same shape as `input`:
+ * <pre>{@code
+ * input = [[b'ten', b'eleven', b'twelve'],
+ *          [b'thirteen', b'fourteen', b'fifteen'],
+ *          [b'sixteen', b'seventeen', b'eighteen']]
+ * position = [[1, 2, 3],
+ *             [1, 2, 3],
+ *             [1, 2, 3]]
+ * length =   [[2, 3, 4],
+ *             [4, 3, 2],
+ *             [5, 5, 5]]
+ * 
+ * output = [[b'en', b'eve', b'lve'],
+ *           [b'hirt', b'urt', b'te'],
+ *           [b'ixtee', b'vente', b'hteen']]
+ * }</pre>
+ * Broadcasting `pos` and `len` onto `input`:
+ * <pre>{@code
+ * input = [[b'ten', b'eleven', b'twelve'],
+ *          [b'thirteen', b'fourteen', b'fifteen'],
+ *          [b'sixteen', b'seventeen', b'eighteen'],
+ *          [b'nineteen', b'twenty', b'twentyone']]
+ * position = [1, 2, 3]
+ * length =   [1, 2, 3]
+ * 
+ * output = [[b'e', b'ev', b'lve'],
+ *           [b'h', b'ur', b'tee'],
+ *           [b'i', b've', b'hte'],
+ *           [b'i', b'en', b'nty']]
+ * }</pre>
+ * Broadcasting `input` onto `pos` and `len`:
+ * <pre>{@code
+ * input = b'thirteen'
+ * position = [1, 5, 7]
+ * length =   [3, 2, 1]
+ * 
+ * output = [b'hir', b'ee', b'n']
+ * }</pre>
+ * Raises:
+ * <p>
+ *   * `ValueError`: If the first argument cannot be converted to a
+ *      Tensor of `dtype string`.
+ *   * `InvalidArgumentError`: If indices are out of range.
+ *   * `ValueError`: If `pos` and `len` are not the same shape.
+ * 
+ */
+@Operator(group = "strings")
+public final class Substr extends RawOp implements Operand<TString> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.strings.Substr}
+   */
+  public static class Options {
+    
+    /**
+     * @param unit The unit that is used to create the substring.  One of: `"BYTE"` (for
+     * defining position and length by bytes) or `"UTF8_CHAR"` (for the UTF-8
+     * encoded Unicode code points).  The default is `"BYTE"`. Results are undefined if
+     * `unit=UTF8_CHAR` and the `input` strings do not contain structurally valid
+     * UTF-8.
+     */
+    public Options unit(String unit) {
+      this.unit = unit;
+      return this;
+    }
+    
+    private String unit;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new Substr operation.
+   * 
+   * @param scope current scope
+   * @param input Tensor of strings
+   * @param pos Scalar defining the position of first character in each substring
+   * @param len Scalar defining the number of characters to include in each substring
+   * @param options carries optional attributes values
+   * @return a new instance of Substr
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> Substr create(Scope scope, Operand<TString> input, Operand<T> pos, Operand<T> len, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Substr", scope.makeOpName("Substr"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(pos.asOutput());
+    opBuilder.addInput(len.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.unit != null) {
+          opBuilder.setAttr("unit", opts.unit);
+        }
+      }
+    }
+    return new Substr(opBuilder.build());
+  }
+  
+  /**
+   * @param unit The unit that is used to create the substring.  One of: `"BYTE"` (for
+   * defining position and length by bytes) or `"UTF8_CHAR"` (for the UTF-8
+   * encoded Unicode code points).  The default is `"BYTE"`. Results are undefined if
+   * `unit=UTF8_CHAR` and the `input` strings do not contain structurally valid
+   * UTF-8.
+   */
+  public static Options unit(String unit) {
+    return new Options().unit(unit);
+  }
+  
+  /**
+   * Tensor of substrings
+   */
+  public Output<TString> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<TString> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Substr";
+  
+  private Output<TString> output;
+  
+  private Substr(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/ToHashBucket.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/ToHashBucket.java
new file mode 100644
index 00000000000..11f025a4055
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/ToHashBucket.java
@@ -0,0 +1,83 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.strings;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.TString;
+
+/**
+ * Converts each string in the input Tensor to its hash mod by a number of buckets.
+ * <p>
+ * The hash function is deterministic on the content of the string within the
+ * process.
+ * <p>
+ * Note that the hash function may change from time to time.
+ * This functionality will be deprecated and it's recommended to use
+ * `tf.string_to_hash_bucket_fast()` or `tf.string_to_hash_bucket_strong()`.
+ */
+@Operator(group = "strings")
+public final class ToHashBucket extends RawOp implements Operand<TInt64> {
+  
+  /**
+   * Factory method to create a class wrapping a new ToHashBucket operation.
+   * 
+   * @param scope current scope
+   * @param stringTensor 
+   * @param numBuckets The number of buckets.
+   * @return a new instance of ToHashBucket
+   */
+  @Endpoint(describeByClass = true)
+  public static ToHashBucket create(Scope scope, Operand<TString> stringTensor, Long numBuckets) {
+    OperationBuilder opBuilder = scope.env().opBuilder("StringToHashBucket", scope.makeOpName("ToHashBucket"));
+    opBuilder.addInput(stringTensor.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("num_buckets", numBuckets);
+    return new ToHashBucket(opBuilder.build());
+  }
+  
+  /**
+   * A Tensor of the same shape as the input `string_tensor`.
+   */
+  public Output<TInt64> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<TInt64> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "StringToHashBucket";
+  
+  private Output<TInt64> output;
+  
+  private ToHashBucket(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/ToHashBucketFast.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/ToHashBucketFast.java
new file mode 100644
index 00000000000..b1025fb9ccd
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/ToHashBucketFast.java
@@ -0,0 +1,88 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.strings;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.TString;
+
+/**
+ * Converts each string in the input Tensor to its hash mod by a number of buckets.
+ * <p>
+ * The hash function is deterministic on the content of the string within the
+ * process and will never change. However, it is not suitable for cryptography.
+ * This function may be used when CPU time is scarce and inputs are trusted or
+ * unimportant. There is a risk of adversaries constructing inputs that all hash
+ * to the same bucket. To prevent this problem, use a strong hash function with
+ * `tf.string_to_hash_bucket_strong`.
+ * <p>
+ * Examples:
+ * <p>
+ * >>> tf.strings.to_hash_bucket_fast(["Hello", "TensorFlow", "2.x"], 3).numpy()
+ * array([0, 2, 2])
+ */
+@Operator(group = "strings")
+public final class ToHashBucketFast extends RawOp implements Operand<TInt64> {
+  
+  /**
+   * Factory method to create a class wrapping a new ToHashBucketFast operation.
+   * 
+   * @param scope current scope
+   * @param input The strings to assign a hash bucket.
+   * @param numBuckets The number of buckets.
+   * @return a new instance of ToHashBucketFast
+   */
+  @Endpoint(describeByClass = true)
+  public static ToHashBucketFast create(Scope scope, Operand<TString> input, Long numBuckets) {
+    OperationBuilder opBuilder = scope.env().opBuilder("StringToHashBucketFast", scope.makeOpName("ToHashBucketFast"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("num_buckets", numBuckets);
+    return new ToHashBucketFast(opBuilder.build());
+  }
+  
+  /**
+   * A Tensor of the same shape as the input `string_tensor`.
+   */
+  public Output<TInt64> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<TInt64> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "StringToHashBucketFast";
+  
+  private Output<TInt64> output;
+  
+  private ToHashBucketFast(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/ToHashBucketStrong.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/ToHashBucketStrong.java
new file mode 100644
index 00000000000..28af44e377f
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/ToHashBucketStrong.java
@@ -0,0 +1,103 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.strings;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.TString;
+
+/**
+ * Converts each string in the input Tensor to its hash mod by a number of buckets.
+ * <p>
+ * The hash function is deterministic on the content of the string within the
+ * process. The hash function is a keyed hash function, where attribute `key`
+ * defines the key of the hash function. `key` is an array of 2 elements.
+ * <p>
+ * A strong hash is important when inputs may be malicious, e.g. URLs with
+ * additional components. Adversaries could try to make their inputs hash to the
+ * same bucket for a denial-of-service attack or to skew the results. A strong
+ * hash can be used to make it difficult to find inputs with a skewed hash value
+ * distribution over buckets. This requires that the hash function is
+ * seeded by a high-entropy (random) "key" unknown to the adversary.
+ * <p>
+ * The additional robustness comes at a cost of roughly 4x higher compute
+ * time than `tf.string_to_hash_bucket_fast`.
+ * <p>
+ * Examples:
+ * <p>
+ * >>> tf.strings.to_hash_bucket_strong(["Hello", "TF"], 3, [1, 2]).numpy()
+ * array([2, 0])
+ */
+@Operator(group = "strings")
+public final class ToHashBucketStrong extends RawOp implements Operand<TInt64> {
+  
+  /**
+   * Factory method to create a class wrapping a new ToHashBucketStrong operation.
+   * 
+   * @param scope current scope
+   * @param input The strings to assign a hash bucket.
+   * @param numBuckets The number of buckets.
+   * @param key The key used to seed the hash function, passed as a list of two uint64
+   * elements.
+   * @return a new instance of ToHashBucketStrong
+   */
+  @Endpoint(describeByClass = true)
+  public static ToHashBucketStrong create(Scope scope, Operand<TString> input, Long numBuckets, List<Long> key) {
+    OperationBuilder opBuilder = scope.env().opBuilder("StringToHashBucketStrong", scope.makeOpName("ToHashBucketStrong"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("num_buckets", numBuckets);
+    long[] keyArray = new long[key.size()];
+    for (int i = 0; i < keyArray.length; ++i) {
+      keyArray[i] = key.get(i);
+    }
+    opBuilder.setAttr("key", keyArray);
+    return new ToHashBucketStrong(opBuilder.build());
+  }
+  
+  /**
+   * A Tensor of the same shape as the input `string_tensor`.
+   */
+  public Output<TInt64> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<TInt64> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "StringToHashBucketStrong";
+  
+  private Output<TInt64> output;
+  
+  private ToHashBucketStrong(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/ToNumber.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/ToNumber.java
new file mode 100644
index 00000000000..47ddcfde480
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/ToNumber.java
@@ -0,0 +1,101 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.strings;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.TString;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Converts each string in the input Tensor to the specified numeric type.
+ * <p>
+ * (Note that int32 overflow results in an error while float overflow
+ * results in a rounded value.)
+ * <p>
+ * Example:
+ * <p>
+ * >>> strings = ["5.0", "3.0", "7.0"]
+ * >>> tf.strings.to_number(strings)
+ * <tf.Tensor: shape=(3,), dtype=float32, numpy=array([5., 3., 7.], dtype=float32)>
+ * 
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator(group = "strings")
+public final class ToNumber<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new ToNumber operation.
+   * 
+   * @param scope current scope
+   * @param stringTensor 
+   * @param outType The numeric type to interpret each string in `string_tensor` as.
+   * @return a new instance of ToNumber
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> ToNumber<T> create(Scope scope, Operand<TString> stringTensor, Class<T> outType) {
+    OperationBuilder opBuilder = scope.env().opBuilder("StringToNumber", scope.makeOpName("ToNumber"));
+    opBuilder.addInput(stringTensor.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("out_type", outType);
+    return new ToNumber<T>(opBuilder.build());
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new ToNumber operation using default output types.
+   * 
+   * @param scope current scope
+   * @param stringTensor 
+   * @return a new instance of ToNumber
+   */
+  @Endpoint(describeByClass = true)
+  public static ToNumber<TFloat32> create(Scope scope, Operand<TString> stringTensor) {
+    return create(scope, stringTensor, TFloat32.class);
+  }
+  
+  /**
+   * A Tensor of the same shape as the input `string_tensor`.
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "StringToNumber";
+  
+  private Output<T> output;
+  
+  private ToNumber(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/UnicodeDecode.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/UnicodeDecode.java
new file mode 100644
index 00000000000..1c73739a144
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/UnicodeDecode.java
@@ -0,0 +1,212 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.strings;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.TString;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Decodes each string in `input` into a sequence of Unicode code points.
+ * <p>
+ * The character codepoints for all strings are returned using a single vector
+ * `char_values`, with strings expanded to characters in row-major order.
+ * <p>
+ * The `row_splits` tensor indicates where the codepoints for
+ * each input string begin and end within the `char_values` tensor.
+ * In particular, the values for the `i`th
+ * string (in row-major order) are stored in the slice
+ * `[row_splits[i]:row_splits[i+1]]`. Thus:
+ * <ul>
+ * <li>
+ * `char_values[row_splits[i]+j]` is the Unicode codepoint for the `j`th
+ *   character in the `i`th string (in row-major order).
+ * </li>
+ * <li>
+ * `row_splits[i+1] - row_splits[i]` is the number of characters in the `i`th
+ *   string (in row-major order).
+ * 
+ * @param <T> data type for {@code rowSplits()} output
+ */
+public final class UnicodeDecode<T extends TNumber> extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.strings.UnicodeDecode}
+   */
+  public static class Options {
+    
+    /**
+     * @param errors Error handling policy when there is invalid formatting found in the input.
+     * The value of 'strict' will cause the operation to produce a InvalidArgument
+     * error on any invalid input formatting. A value of 'replace' (the default) will
+     * cause the operation to replace any invalid formatting in the input with the
+     * `replacement_char` codepoint. A value of 'ignore' will cause the operation to
+     * skip any invalid formatting in the input and produce no corresponding output
+     * character.
+     */
+    public Options errors(String errors) {
+      this.errors = errors;
+      return this;
+    }
+    
+    /**
+     * @param replacementChar The replacement character codepoint to be used in place of any invalid
+     * formatting in the input when `errors='replace'`. Any valid unicode codepoint may
+     * be used. The default value is the default unicode replacement character is
+     * 0xFFFD or U+65533.)
+     */
+    public Options replacementChar(Long replacementChar) {
+      this.replacementChar = replacementChar;
+      return this;
+    }
+    
+    /**
+     * @param replaceControlCharacters Whether to replace the C0 control characters (00-1F) with the
+     * `replacement_char`. Default is false.
+     */
+    public Options replaceControlCharacters(Boolean replaceControlCharacters) {
+      this.replaceControlCharacters = replaceControlCharacters;
+      return this;
+    }
+    
+    private String errors;
+    private Long replacementChar;
+    private Boolean replaceControlCharacters;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new UnicodeDecode operation.
+   * 
+   * @param scope current scope
+   * @param input The text to be decoded. Can have any shape. Note that the output is flattened
+   * to a vector of char values.
+   * @param inputEncoding Text encoding of the input strings. This is any of the encodings supported
+   * by ICU ucnv algorithmic converters. Examples: `"UTF-16", "US ASCII", "UTF-8"`.
+   * @param Tsplits 
+   * @param options carries optional attributes values
+   * @return a new instance of UnicodeDecode
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> UnicodeDecode<T> create(Scope scope, Operand<TString> input, String inputEncoding, Class<T> Tsplits, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("UnicodeDecode", scope.makeOpName("UnicodeDecode"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("input_encoding", inputEncoding);
+    opBuilder.setAttr("Tsplits", Tsplits);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.errors != null) {
+          opBuilder.setAttr("errors", opts.errors);
+        }
+        if (opts.replacementChar != null) {
+          opBuilder.setAttr("replacement_char", opts.replacementChar);
+        }
+        if (opts.replaceControlCharacters != null) {
+          opBuilder.setAttr("replace_control_characters", opts.replaceControlCharacters);
+        }
+      }
+    }
+    return new UnicodeDecode<T>(opBuilder.build());
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new UnicodeDecode operation using default output types.
+   * 
+   * @param scope current scope
+   * @param input The text to be decoded. Can have any shape. Note that the output is flattened
+   * to a vector of char values.
+   * @param inputEncoding Text encoding of the input strings. This is any of the encodings supported
+   * by ICU ucnv algorithmic converters. Examples: `"UTF-16", "US ASCII", "UTF-8"`.
+   * @param options carries optional attributes values
+   * @return a new instance of UnicodeDecode
+   */
+  @Endpoint(describeByClass = true)
+  public static UnicodeDecode<TInt64> create(Scope scope, Operand<TString> input, String inputEncoding, Options... options) {
+    return create(scope, input, inputEncoding, TInt64.class, options);
+  }
+  
+  /**
+   * @param errors Error handling policy when there is invalid formatting found in the input.
+   * The value of 'strict' will cause the operation to produce a InvalidArgument
+   * error on any invalid input formatting. A value of 'replace' (the default) will
+   * cause the operation to replace any invalid formatting in the input with the
+   * `replacement_char` codepoint. A value of 'ignore' will cause the operation to
+   * skip any invalid formatting in the input and produce no corresponding output
+   * character.
+   */
+  public static Options errors(String errors) {
+    return new Options().errors(errors);
+  }
+  
+  /**
+   * @param replacementChar The replacement character codepoint to be used in place of any invalid
+   * formatting in the input when `errors='replace'`. Any valid unicode codepoint may
+   * be used. The default value is the default unicode replacement character is
+   * 0xFFFD or U+65533.)
+   */
+  public static Options replacementChar(Long replacementChar) {
+    return new Options().replacementChar(replacementChar);
+  }
+  
+  /**
+   * @param replaceControlCharacters Whether to replace the C0 control characters (00-1F) with the
+   * `replacement_char`. Default is false.
+   */
+  public static Options replaceControlCharacters(Boolean replaceControlCharacters) {
+    return new Options().replaceControlCharacters(replaceControlCharacters);
+  }
+  
+  /**
+   * A 1D int32 tensor containing the row splits.
+   */
+  public Output<T> rowSplits() {
+    return rowSplits;
+  }
+  
+  /**
+   * A 1D int32 Tensor containing the decoded codepoints.
+   */
+  public Output<TInt32> charValues() {
+    return charValues;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "UnicodeDecode";
+  
+  private Output<T> rowSplits;
+  private Output<TInt32> charValues;
+  
+  private UnicodeDecode(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    rowSplits = operation.output(outputIdx++);
+    charValues = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/UnicodeDecodeWithOffsets.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/UnicodeDecodeWithOffsets.java
new file mode 100644
index 00000000000..f0999df6285
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/UnicodeDecodeWithOffsets.java
@@ -0,0 +1,228 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.strings;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.TString;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Decodes each string in `input` into a sequence of Unicode code points.
+ * <p>
+ * The character codepoints for all strings are returned using a single vector
+ * `char_values`, with strings expanded to characters in row-major order.
+ * Similarly, the character start byte offsets are returned using a single vector
+ * `char_to_byte_starts`, with strings expanded in row-major order.
+ * <p>
+ * The `row_splits` tensor indicates where the codepoints and start offsets for
+ * each input string begin and end within the `char_values` and
+ * `char_to_byte_starts` tensors.  In particular, the values for the `i`th
+ * string (in row-major order) are stored in the slice
+ * `[row_splits[i]:row_splits[i+1]]`. Thus:
+ * <ul>
+ * <li>
+ * `char_values[row_splits[i]+j]` is the Unicode codepoint for the `j`th
+ *   character in the `i`th string (in row-major order).
+ * </li>
+ * <li>
+ * `char_to_bytes_starts[row_splits[i]+j]` is the start byte offset for the `j`th
+ *   character in the `i`th string (in row-major order).
+ * </li>
+ * <li>
+ * `row_splits[i+1] - row_splits[i]` is the number of characters in the `i`th
+ *   string (in row-major order).
+ * 
+ * @param <T> data type for {@code rowSplits()} output
+ */
+public final class UnicodeDecodeWithOffsets<T extends TNumber> extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.strings.UnicodeDecodeWithOffsets}
+   */
+  public static class Options {
+    
+    /**
+     * @param errors Error handling policy when there is invalid formatting found in the input.
+     * The value of 'strict' will cause the operation to produce a InvalidArgument
+     * error on any invalid input formatting. A value of 'replace' (the default) will
+     * cause the operation to replace any invalid formatting in the input with the
+     * `replacement_char` codepoint. A value of 'ignore' will cause the operation to
+     * skip any invalid formatting in the input and produce no corresponding output
+     * character.
+     */
+    public Options errors(String errors) {
+      this.errors = errors;
+      return this;
+    }
+    
+    /**
+     * @param replacementChar The replacement character codepoint to be used in place of any invalid
+     * formatting in the input when `errors='replace'`. Any valid unicode codepoint may
+     * be used. The default value is the default unicode replacement character is
+     * 0xFFFD or U+65533.)
+     */
+    public Options replacementChar(Long replacementChar) {
+      this.replacementChar = replacementChar;
+      return this;
+    }
+    
+    /**
+     * @param replaceControlCharacters Whether to replace the C0 control characters (00-1F) with the
+     * `replacement_char`. Default is false.
+     */
+    public Options replaceControlCharacters(Boolean replaceControlCharacters) {
+      this.replaceControlCharacters = replaceControlCharacters;
+      return this;
+    }
+    
+    private String errors;
+    private Long replacementChar;
+    private Boolean replaceControlCharacters;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new UnicodeDecodeWithOffsets operation.
+   * 
+   * @param scope current scope
+   * @param input The text to be decoded. Can have any shape. Note that the output is flattened
+   * to a vector of char values.
+   * @param inputEncoding Text encoding of the input strings. This is any of the encodings supported
+   * by ICU ucnv algorithmic converters. Examples: `"UTF-16", "US ASCII", "UTF-8"`.
+   * @param Tsplits 
+   * @param options carries optional attributes values
+   * @return a new instance of UnicodeDecodeWithOffsets
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> UnicodeDecodeWithOffsets<T> create(Scope scope, Operand<TString> input, String inputEncoding, Class<T> Tsplits, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("UnicodeDecodeWithOffsets", scope.makeOpName("UnicodeDecodeWithOffsets"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("input_encoding", inputEncoding);
+    opBuilder.setAttr("Tsplits", Tsplits);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.errors != null) {
+          opBuilder.setAttr("errors", opts.errors);
+        }
+        if (opts.replacementChar != null) {
+          opBuilder.setAttr("replacement_char", opts.replacementChar);
+        }
+        if (opts.replaceControlCharacters != null) {
+          opBuilder.setAttr("replace_control_characters", opts.replaceControlCharacters);
+        }
+      }
+    }
+    return new UnicodeDecodeWithOffsets<T>(opBuilder.build());
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new UnicodeDecodeWithOffsets operation using default output types.
+   * 
+   * @param scope current scope
+   * @param input The text to be decoded. Can have any shape. Note that the output is flattened
+   * to a vector of char values.
+   * @param inputEncoding Text encoding of the input strings. This is any of the encodings supported
+   * by ICU ucnv algorithmic converters. Examples: `"UTF-16", "US ASCII", "UTF-8"`.
+   * @param options carries optional attributes values
+   * @return a new instance of UnicodeDecodeWithOffsets
+   */
+  @Endpoint(describeByClass = true)
+  public static UnicodeDecodeWithOffsets<TInt64> create(Scope scope, Operand<TString> input, String inputEncoding, Options... options) {
+    return create(scope, input, inputEncoding, TInt64.class, options);
+  }
+  
+  /**
+   * @param errors Error handling policy when there is invalid formatting found in the input.
+   * The value of 'strict' will cause the operation to produce a InvalidArgument
+   * error on any invalid input formatting. A value of 'replace' (the default) will
+   * cause the operation to replace any invalid formatting in the input with the
+   * `replacement_char` codepoint. A value of 'ignore' will cause the operation to
+   * skip any invalid formatting in the input and produce no corresponding output
+   * character.
+   */
+  public static Options errors(String errors) {
+    return new Options().errors(errors);
+  }
+  
+  /**
+   * @param replacementChar The replacement character codepoint to be used in place of any invalid
+   * formatting in the input when `errors='replace'`. Any valid unicode codepoint may
+   * be used. The default value is the default unicode replacement character is
+   * 0xFFFD or U+65533.)
+   */
+  public static Options replacementChar(Long replacementChar) {
+    return new Options().replacementChar(replacementChar);
+  }
+  
+  /**
+   * @param replaceControlCharacters Whether to replace the C0 control characters (00-1F) with the
+   * `replacement_char`. Default is false.
+   */
+  public static Options replaceControlCharacters(Boolean replaceControlCharacters) {
+    return new Options().replaceControlCharacters(replaceControlCharacters);
+  }
+  
+  /**
+   * A 1D int32 tensor containing the row splits.
+   */
+  public Output<T> rowSplits() {
+    return rowSplits;
+  }
+  
+  /**
+   * A 1D int32 Tensor containing the decoded codepoints.
+   */
+  public Output<TInt32> charValues() {
+    return charValues;
+  }
+  
+  /**
+   * A 1D int32 Tensor containing the byte index in the input string where each
+   * character in `char_values` starts.
+   */
+  public Output<TInt64> charToByteStarts() {
+    return charToByteStarts;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "UnicodeDecodeWithOffsets";
+  
+  private Output<T> rowSplits;
+  private Output<TInt32> charValues;
+  private Output<TInt64> charToByteStarts;
+  
+  private UnicodeDecodeWithOffsets(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    rowSplits = operation.output(outputIdx++);
+    charValues = operation.output(outputIdx++);
+    charToByteStarts = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/UnicodeEncode.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/UnicodeEncode.java
new file mode 100644
index 00000000000..e499dee3157
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/UnicodeEncode.java
@@ -0,0 +1,168 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.strings;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.TString;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Encode a tensor of ints into unicode strings.
+ * <p>
+ * Returns a vector of strings, where `output[i]` is constructed by encoding the
+ * Unicode codepoints in `input_values[input_splits[i]:input_splits[i+1]]`
+ * using `output_encoding`.
+ * <p>
+ * ---
+ * <p>
+ * Example:
+ * <pre>{@code
+ * input_values = [72, 101, 108, 108, 111, 87, 111, 114, 108, 100]
+ * input_splits = [0, 5, 10]
+ * output_encoding = 'UTF-8'
+ * 
+ * output = ['Hello', 'World']
+ * }</pre>
+ * 
+ */
+public final class UnicodeEncode extends RawOp implements Operand<TString> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.strings.UnicodeEncode}
+   */
+  public static class Options {
+    
+    /**
+     * @param errors Error handling policy when there is invalid formatting found in the input.
+     * The value of 'strict' will cause the operation to produce a InvalidArgument
+     * error on any invalid input formatting. A value of 'replace' (the default) will
+     * cause the operation to replace any invalid formatting in the input with the
+     * `replacement_char` codepoint. A value of 'ignore' will cause the operation to
+     * skip any invalid formatting in the input and produce no corresponding output
+     * character.
+     */
+    public Options errors(String errors) {
+      this.errors = errors;
+      return this;
+    }
+    
+    /**
+     * @param replacementChar The replacement character codepoint to be used in place of any invalid
+     * formatting in the input when `errors='replace'`. Any valid unicode codepoint may
+     * be used. The default value is the default unicode replacement character is
+     * 0xFFFD (U+65533).
+     */
+    public Options replacementChar(Long replacementChar) {
+      this.replacementChar = replacementChar;
+      return this;
+    }
+    
+    private String errors;
+    private Long replacementChar;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new UnicodeEncode operation.
+   * 
+   * @param scope current scope
+   * @param inputValues A 1D tensor containing the unicode codepoints that should be encoded.
+   * @param inputSplits A 1D tensor specifying how the unicode codepoints should be split into strings.
+   * In particular, `output[i]` is constructed by encoding the codepoints in the
+   * slice `input_values[input_splits[i]:input_splits[i+1]]`.
+   * @param outputEncoding Unicode encoding of the output strings. Valid encodings are: `"UTF-8",
+   * "UTF-16-BE", and "UTF-32-BE"`.
+   * @param options carries optional attributes values
+   * @return a new instance of UnicodeEncode
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> UnicodeEncode create(Scope scope, Operand<TInt32> inputValues, Operand<T> inputSplits, String outputEncoding, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("UnicodeEncode", scope.makeOpName("UnicodeEncode"));
+    opBuilder.addInput(inputValues.asOutput());
+    opBuilder.addInput(inputSplits.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("output_encoding", outputEncoding);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.errors != null) {
+          opBuilder.setAttr("errors", opts.errors);
+        }
+        if (opts.replacementChar != null) {
+          opBuilder.setAttr("replacement_char", opts.replacementChar);
+        }
+      }
+    }
+    return new UnicodeEncode(opBuilder.build());
+  }
+  
+  /**
+   * @param errors Error handling policy when there is invalid formatting found in the input.
+   * The value of 'strict' will cause the operation to produce a InvalidArgument
+   * error on any invalid input formatting. A value of 'replace' (the default) will
+   * cause the operation to replace any invalid formatting in the input with the
+   * `replacement_char` codepoint. A value of 'ignore' will cause the operation to
+   * skip any invalid formatting in the input and produce no corresponding output
+   * character.
+   */
+  public static Options errors(String errors) {
+    return new Options().errors(errors);
+  }
+  
+  /**
+   * @param replacementChar The replacement character codepoint to be used in place of any invalid
+   * formatting in the input when `errors='replace'`. Any valid unicode codepoint may
+   * be used. The default value is the default unicode replacement character is
+   * 0xFFFD (U+65533).
+   */
+  public static Options replacementChar(Long replacementChar) {
+    return new Options().replacementChar(replacementChar);
+  }
+  
+  /**
+   * The 1-D Tensor of strings encoded from the provided unicode codepoints.
+   */
+  public Output<TString> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<TString> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "UnicodeEncode";
+  
+  private Output<TString> output;
+  
+  private UnicodeEncode(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/UnicodeScript.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/UnicodeScript.java
new file mode 100644
index 00000000000..6d3600a1de3
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/UnicodeScript.java
@@ -0,0 +1,84 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.strings;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+
+/**
+ * Determine the script codes of a given tensor of Unicode integer code points.
+ * <p>
+ * This operation converts Unicode code points to script codes corresponding to
+ * each code point. Script codes correspond to International Components for
+ * Unicode (ICU) UScriptCode values. See http://icu-project.org/apiref/icu4c/uscript_8h.html.
+ * Returns -1 (USCRIPT_INVALID_CODE) for invalid codepoints. Output shape will
+ * match input shape.
+ * <p>
+ * Examples:
+ * <p>
+ * >>> tf.strings.unicode_script([1, 31, 38])
+ * <tf.Tensor: shape=(3,), dtype=int32, numpy=array([0, 0, 0], dtype=int32)>
+ */
+@Operator(group = "strings")
+public final class UnicodeScript extends RawOp implements Operand<TInt32> {
+  
+  /**
+   * Factory method to create a class wrapping a new UnicodeScript operation.
+   * 
+   * @param scope current scope
+   * @param input A Tensor of int32 Unicode code points.
+   * @return a new instance of UnicodeScript
+   */
+  @Endpoint(describeByClass = true)
+  public static UnicodeScript create(Scope scope, Operand<TInt32> input) {
+    OperationBuilder opBuilder = scope.env().opBuilder("UnicodeScript", scope.makeOpName("UnicodeScript"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new UnicodeScript(opBuilder.build());
+  }
+  
+  /**
+   * A Tensor of int32 script codes corresponding to each input code point.
+   */
+  public Output<TInt32> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<TInt32> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "UnicodeScript";
+  
+  private Output<TInt32> output;
+  
+  private UnicodeScript(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/UnicodeTranscode.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/UnicodeTranscode.java
new file mode 100644
index 00000000000..8f62db46152
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/UnicodeTranscode.java
@@ -0,0 +1,216 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.strings;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TString;
+
+/**
+ * Transcode the input text from a source encoding to a destination encoding.
+ * <p>
+ * The input is a string tensor of any shape. The output is a string tensor of
+ * the same shape containing the transcoded strings. Output strings are always
+ * valid unicode. If the input contains invalid encoding positions, the
+ * `errors` attribute sets the policy for how to deal with them. If the default
+ * error-handling policy is used, invalid formatting will be substituted in the
+ * output by the `replacement_char`. If the errors policy is to `ignore`, any
+ * invalid encoding positions in the input are skipped and not included in the
+ * output. If it set to `strict` then any invalid formatting will result in an
+ * InvalidArgument error.
+ * <p>
+ * This operation can be used with `output_encoding = input_encoding` to enforce
+ * correct formatting for inputs even if they are already in the desired encoding.
+ * <p>
+ * If the input is prefixed by a Byte Order Mark needed to determine encoding
+ * (e.g. if the encoding is UTF-16 and the BOM indicates big-endian), then that
+ * BOM will be consumed and not emitted into the output. If the input encoding
+ * is marked with an explicit endianness (e.g. UTF-16-BE), then the BOM is
+ * interpreted as a non-breaking-space and is preserved in the output (including
+ * always for UTF-8).
+ * <p>
+ * The end result is that if the input is marked as an explicit endianness the
+ * transcoding is faithful to all codepoints in the source. If it is not marked
+ * with an explicit endianness, the BOM is not considered part of the string itself
+ * but as metadata, and so is not preserved in the output.
+ * <p>
+ * Examples:
+ * <p>
+ * >>> tf.strings.unicode_transcode(["Hello", "TensorFlow", "2.x"], "UTF-8", "UTF-16-BE")
+ * <tf.Tensor: shape=(3,), dtype=string, numpy=
+ * array([b'\x00H\x00e\x00l\x00l\x00o',
+ *        b'\x00T\x00e\x00n\x00s\x00o\x00r\x00F\x00l\x00o\x00w',
+ *        b'\x002\x00.\x00x'], dtype=object)>
+ * >>> tf.strings.unicode_transcode(["A", "B", "C"], "US ASCII", "UTF-8").numpy()
+ * array([b'A', b'B', b'C'], dtype=object)
+ */
+@Operator(group = "strings")
+public final class UnicodeTranscode extends RawOp implements Operand<TString> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.strings.UnicodeTranscode}
+   */
+  public static class Options {
+    
+    /**
+     * @param errors Error handling policy when there is invalid formatting found in the input.
+     * The value of 'strict' will cause the operation to produce a InvalidArgument
+     * error on any invalid input formatting. A value of 'replace' (the default) will
+     * cause the operation to replace any invalid formatting in the input with the
+     * `replacement_char` codepoint. A value of 'ignore' will cause the operation to
+     * skip any invalid formatting in the input and produce no corresponding output
+     * character.
+     */
+    public Options errors(String errors) {
+      this.errors = errors;
+      return this;
+    }
+    
+    /**
+     * @param replacementChar The replacement character codepoint to be used in place of any invalid
+     * formatting in the input when `errors='replace'`. Any valid unicode codepoint may
+     * be used. The default value is the default unicode replacement character is
+     * 0xFFFD or U+65533.)
+     * <p>
+     * Note that for UTF-8, passing a replacement character expressible in 1 byte, such
+     * as ' ', will preserve string alignment to the source since invalid bytes will be
+     * replaced with a 1-byte replacement. For UTF-16-BE and UTF-16-LE, any 1 or 2 byte
+     * replacement character will preserve byte alignment to the source.
+     */
+    public Options replacementChar(Long replacementChar) {
+      this.replacementChar = replacementChar;
+      return this;
+    }
+    
+    /**
+     * @param replaceControlCharacters Whether to replace the C0 control characters (00-1F) with the
+     * `replacement_char`. Default is false.
+     */
+    public Options replaceControlCharacters(Boolean replaceControlCharacters) {
+      this.replaceControlCharacters = replaceControlCharacters;
+      return this;
+    }
+    
+    private String errors;
+    private Long replacementChar;
+    private Boolean replaceControlCharacters;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new UnicodeTranscode operation.
+   * 
+   * @param scope current scope
+   * @param input The text to be processed. Can have any shape.
+   * @param inputEncoding Text encoding of the input strings. This is any of the encodings supported
+   * by ICU ucnv algorithmic converters. Examples: `"UTF-16", "US ASCII", "UTF-8"`.
+   * @param outputEncoding The unicode encoding to use in the output. Must be one of
+   * `"UTF-8", "UTF-16-BE", "UTF-32-BE"`. Multi-byte encodings will be big-endian.
+   * @param options carries optional attributes values
+   * @return a new instance of UnicodeTranscode
+   */
+  @Endpoint(describeByClass = true)
+  public static UnicodeTranscode create(Scope scope, Operand<TString> input, String inputEncoding, String outputEncoding, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("UnicodeTranscode", scope.makeOpName("UnicodeTranscode"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("input_encoding", inputEncoding);
+    opBuilder.setAttr("output_encoding", outputEncoding);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.errors != null) {
+          opBuilder.setAttr("errors", opts.errors);
+        }
+        if (opts.replacementChar != null) {
+          opBuilder.setAttr("replacement_char", opts.replacementChar);
+        }
+        if (opts.replaceControlCharacters != null) {
+          opBuilder.setAttr("replace_control_characters", opts.replaceControlCharacters);
+        }
+      }
+    }
+    return new UnicodeTranscode(opBuilder.build());
+  }
+  
+  /**
+   * @param errors Error handling policy when there is invalid formatting found in the input.
+   * The value of 'strict' will cause the operation to produce a InvalidArgument
+   * error on any invalid input formatting. A value of 'replace' (the default) will
+   * cause the operation to replace any invalid formatting in the input with the
+   * `replacement_char` codepoint. A value of 'ignore' will cause the operation to
+   * skip any invalid formatting in the input and produce no corresponding output
+   * character.
+   */
+  public static Options errors(String errors) {
+    return new Options().errors(errors);
+  }
+  
+  /**
+   * @param replacementChar The replacement character codepoint to be used in place of any invalid
+   * formatting in the input when `errors='replace'`. Any valid unicode codepoint may
+   * be used. The default value is the default unicode replacement character is
+   * 0xFFFD or U+65533.)
+   * <p>
+   * Note that for UTF-8, passing a replacement character expressible in 1 byte, such
+   * as ' ', will preserve string alignment to the source since invalid bytes will be
+   * replaced with a 1-byte replacement. For UTF-16-BE and UTF-16-LE, any 1 or 2 byte
+   * replacement character will preserve byte alignment to the source.
+   */
+  public static Options replacementChar(Long replacementChar) {
+    return new Options().replacementChar(replacementChar);
+  }
+  
+  /**
+   * @param replaceControlCharacters Whether to replace the C0 control characters (00-1F) with the
+   * `replacement_char`. Default is false.
+   */
+  public static Options replaceControlCharacters(Boolean replaceControlCharacters) {
+    return new Options().replaceControlCharacters(replaceControlCharacters);
+  }
+  
+  /**
+   * A string tensor containing unicode text encoded using `output_encoding`.
+   */
+  public Output<TString> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<TString> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "UnicodeTranscode";
+  
+  private Output<TString> output;
+  
+  private UnicodeTranscode(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/UnsortedSegmentJoin.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/UnsortedSegmentJoin.java
new file mode 100644
index 00000000000..2cd28375bd3
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/UnsortedSegmentJoin.java
@@ -0,0 +1,139 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.strings;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TString;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Joins the elements of `inputs` based on `segment_ids`.
+ * <p>
+ * Computes the string join along segments of a tensor.
+ * Given `segment_ids` with rank `N` and `data` with rank `N+M`:
+ * <p>
+ *     `output[i, k1...kM] = strings.join([data[j1...jN, k1...kM])`
+ * <p>
+ * where the join is over all [j1...jN] such that segment_ids[j1...jN] = i.
+ * Strings are joined in row-major order.
+ * <p>
+ * For example:
+ * <pre>{@code
+ * inputs = [['Y', 'q', 'c'], ['Y', '6', '6'], ['p', 'G', 'a']]
+ * output_array = string_ops.unsorted_segment_join(inputs=inputs,
+ *                                                 segment_ids=[1, 0, 1],
+ *                                                 num_segments=2,
+ *                                                 separator=':'))
+ * # output_array ==> [['Y', '6', '6'], ['Y:p', 'q:G', 'c:a']]
+ * 
+ * 
+ * inputs = ['this', 'is', 'a', 'test']
+ * output_array = string_ops.unsorted_segment_join(inputs=inputs,
+ *                                                 segment_ids=[0, 0, 0, 0],
+ *                                                 num_segments=1,
+ *                                                 separator=':'))
+ * # output_array ==> ['this:is:a:test']
+ * }</pre>
+ * 
+ */
+@Operator(group = "strings")
+public final class UnsortedSegmentJoin extends RawOp implements Operand<TString> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.strings.UnsortedSegmentJoin}
+   */
+  public static class Options {
+    
+    /**
+     * @param separator The separator to use when joining.
+     */
+    public Options separator(String separator) {
+      this.separator = separator;
+      return this;
+    }
+    
+    private String separator;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new UnsortedSegmentJoin operation.
+   * 
+   * @param scope current scope
+   * @param inputs The input to be joined.
+   * @param segmentIds A tensor whose shape is a prefix of data.shape.  Negative segment ids are not
+   * supported.
+   * @param numSegments A scalar.
+   * @param options carries optional attributes values
+   * @return a new instance of UnsortedSegmentJoin
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber, U extends TNumber> UnsortedSegmentJoin create(Scope scope, Operand<TString> inputs, Operand<T> segmentIds, Operand<U> numSegments, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("UnsortedSegmentJoin", scope.makeOpName("UnsortedSegmentJoin"));
+    opBuilder.addInput(inputs.asOutput());
+    opBuilder.addInput(segmentIds.asOutput());
+    opBuilder.addInput(numSegments.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.separator != null) {
+          opBuilder.setAttr("separator", opts.separator);
+        }
+      }
+    }
+    return new UnsortedSegmentJoin(opBuilder.build());
+  }
+  
+  /**
+   * @param separator The separator to use when joining.
+   */
+  public static Options separator(String separator) {
+    return new Options().separator(separator);
+  }
+  
+  /**
+   */
+  public Output<TString> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<TString> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "UnsortedSegmentJoin";
+  
+  private Output<TString> output;
+  
+  private UnsortedSegmentJoin(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/Upper.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/Upper.java
new file mode 100644
index 00000000000..e90dc891dd8
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/strings/Upper.java
@@ -0,0 +1,112 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.strings;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TString;
+
+/**
+ * Converts all lowercase characters into their respective uppercase replacements.
+ * <p>
+ * Example:
+ * <p>
+ * >>> tf.strings.upper("CamelCase string and ALL CAPS")
+ * <tf.Tensor: shape=(), dtype=string, numpy=b'CAMELCASE STRING AND ALL CAPS'>
+ * 
+ */
+@Operator(group = "strings")
+public final class Upper extends RawOp implements Operand<TString> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.strings.Upper}
+   */
+  public static class Options {
+    
+    /**
+     * @param encoding 
+     */
+    public Options encoding(String encoding) {
+      this.encoding = encoding;
+      return this;
+    }
+    
+    private String encoding;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new Upper operation.
+   * 
+   * @param scope current scope
+   * @param input 
+   * @param options carries optional attributes values
+   * @return a new instance of Upper
+   */
+  @Endpoint(describeByClass = true)
+  public static Upper create(Scope scope, Operand<TString> input, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("StringUpper", scope.makeOpName("Upper"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.encoding != null) {
+          opBuilder.setAttr("encoding", opts.encoding);
+        }
+      }
+    }
+    return new Upper(opBuilder.build());
+  }
+  
+  /**
+   * @param encoding 
+   */
+  public static Options encoding(String encoding) {
+    return new Options().encoding(encoding);
+  }
+  
+  /**
+   */
+  public Output<TString> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<TString> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "StringUpper";
+  
+  private Output<TString> output;
+  
+  private Upper(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/AudioSummary.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/AudioSummary.java
new file mode 100644
index 00000000000..6743a4ef4c1
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/AudioSummary.java
@@ -0,0 +1,127 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.summary;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.TString;
+
+/**
+ * Outputs a `Summary` protocol buffer with audio.
+ * <p>
+ * The summary has up to `max_outputs` summary values containing audio. The
+ * audio is built from `tensor` which must be 3-D with shape `[batch_size,
+ * frames, channels]` or 2-D with shape `[batch_size, frames]`. The values are
+ * assumed to be in the range of `[-1.0, 1.0]` with a sample rate of `sample_rate`.
+ * <p>
+ * The `tag` argument is a scalar `Tensor` of type `string`.  It is used to
+ * build the `tag` of the summary values:
+ * <ul>
+ * <li>
+ * If `max_outputs` is 1, the summary value tag is '<i>tag</i>/audio'.
+ * </li>
+ * <li>
+ * If `max_outputs` is greater than 1, the summary value tags are
+ *    generated sequentially as '<i>tag</i>/audio/0', '<i>tag</i>/audio/1', etc.
+ */
+@Operator(group = "summary")
+public final class AudioSummary extends RawOp implements Operand<TString> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.summary.AudioSummary}
+   */
+  public static class Options {
+    
+    /**
+     * @param maxOutputs Max number of batch elements to generate audio for.
+     */
+    public Options maxOutputs(Long maxOutputs) {
+      this.maxOutputs = maxOutputs;
+      return this;
+    }
+    
+    private Long maxOutputs;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new AudioSummary operation.
+   * 
+   * @param scope current scope
+   * @param tag Scalar. Used to build the `tag` attribute of the summary values.
+   * @param tensor 2-D of shape `[batch_size, frames]`.
+   * @param sampleRate The sample rate of the signal in hertz.
+   * @param options carries optional attributes values
+   * @return a new instance of AudioSummary
+   */
+  @Endpoint(describeByClass = true)
+  public static AudioSummary create(Scope scope, Operand<TString> tag, Operand<TFloat32> tensor, Operand<TFloat32> sampleRate, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("AudioSummaryV2", scope.makeOpName("AudioSummary"));
+    opBuilder.addInput(tag.asOutput());
+    opBuilder.addInput(tensor.asOutput());
+    opBuilder.addInput(sampleRate.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.maxOutputs != null) {
+          opBuilder.setAttr("max_outputs", opts.maxOutputs);
+        }
+      }
+    }
+    return new AudioSummary(opBuilder.build());
+  }
+  
+  /**
+   * @param maxOutputs Max number of batch elements to generate audio for.
+   */
+  public static Options maxOutputs(Long maxOutputs) {
+    return new Options().maxOutputs(maxOutputs);
+  }
+  
+  /**
+   * Scalar. Serialized `Summary` protocol buffer.
+   */
+  public Output<TString> summary() {
+    return summary;
+  }
+  
+  @Override
+  public Output<TString> asOutput() {
+    return summary;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "AudioSummaryV2";
+  
+  private Output<TString> summary;
+  
+  private AudioSummary(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    summary = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/CloseSummaryWriter.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/CloseSummaryWriter.java
new file mode 100644
index 00000000000..f5d95d50976
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/CloseSummaryWriter.java
@@ -0,0 +1,53 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.summary;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+
+/**
+ */
+public final class CloseSummaryWriter extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new CloseSummaryWriter operation.
+   * 
+   * @param scope current scope
+   * @param writer 
+   * @return a new instance of CloseSummaryWriter
+   */
+  @Endpoint(describeByClass = true)
+  public static CloseSummaryWriter create(Scope scope, Operand<?> writer) {
+    OperationBuilder opBuilder = scope.env().opBuilder("CloseSummaryWriter", scope.makeOpName("CloseSummaryWriter"));
+    opBuilder.addInput(writer.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new CloseSummaryWriter(opBuilder.build());
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "CloseSummaryWriter";
+  
+  private CloseSummaryWriter(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/CreateSummaryDbWriter.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/CreateSummaryDbWriter.java
new file mode 100644
index 00000000000..8e40aa798d6
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/CreateSummaryDbWriter.java
@@ -0,0 +1,62 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.summary;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TString;
+
+/**
+ */
+public final class CreateSummaryDbWriter extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new CreateSummaryDbWriter operation.
+   * 
+   * @param scope current scope
+   * @param writer 
+   * @param dbUri 
+   * @param experimentName 
+   * @param runName 
+   * @param userName 
+   * @return a new instance of CreateSummaryDbWriter
+   */
+  @Endpoint(describeByClass = true)
+  public static CreateSummaryDbWriter create(Scope scope, Operand<?> writer, Operand<TString> dbUri, Operand<TString> experimentName, Operand<TString> runName, Operand<TString> userName) {
+    OperationBuilder opBuilder = scope.env().opBuilder("CreateSummaryDbWriter", scope.makeOpName("CreateSummaryDbWriter"));
+    opBuilder.addInput(writer.asOutput());
+    opBuilder.addInput(dbUri.asOutput());
+    opBuilder.addInput(experimentName.asOutput());
+    opBuilder.addInput(runName.asOutput());
+    opBuilder.addInput(userName.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new CreateSummaryDbWriter(opBuilder.build());
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "CreateSummaryDbWriter";
+  
+  private CreateSummaryDbWriter(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/CreateSummaryFileWriter.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/CreateSummaryFileWriter.java
new file mode 100644
index 00000000000..e429fab20e2
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/CreateSummaryFileWriter.java
@@ -0,0 +1,63 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.summary;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.TString;
+
+/**
+ */
+public final class CreateSummaryFileWriter extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new CreateSummaryFileWriter operation.
+   * 
+   * @param scope current scope
+   * @param writer 
+   * @param logdir 
+   * @param maxQueue 
+   * @param flushMillis 
+   * @param filenameSuffix 
+   * @return a new instance of CreateSummaryFileWriter
+   */
+  @Endpoint(describeByClass = true)
+  public static CreateSummaryFileWriter create(Scope scope, Operand<?> writer, Operand<TString> logdir, Operand<TInt32> maxQueue, Operand<TInt32> flushMillis, Operand<TString> filenameSuffix) {
+    OperationBuilder opBuilder = scope.env().opBuilder("CreateSummaryFileWriter", scope.makeOpName("CreateSummaryFileWriter"));
+    opBuilder.addInput(writer.asOutput());
+    opBuilder.addInput(logdir.asOutput());
+    opBuilder.addInput(maxQueue.asOutput());
+    opBuilder.addInput(flushMillis.asOutput());
+    opBuilder.addInput(filenameSuffix.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new CreateSummaryFileWriter(opBuilder.build());
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "CreateSummaryFileWriter";
+  
+  private CreateSummaryFileWriter(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/FlushSummaryWriter.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/FlushSummaryWriter.java
new file mode 100644
index 00000000000..e1586542972
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/FlushSummaryWriter.java
@@ -0,0 +1,53 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.summary;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+
+/**
+ */
+public final class FlushSummaryWriter extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new FlushSummaryWriter operation.
+   * 
+   * @param scope current scope
+   * @param writer 
+   * @return a new instance of FlushSummaryWriter
+   */
+  @Endpoint(describeByClass = true)
+  public static FlushSummaryWriter create(Scope scope, Operand<?> writer) {
+    OperationBuilder opBuilder = scope.env().opBuilder("FlushSummaryWriter", scope.makeOpName("FlushSummaryWriter"));
+    opBuilder.addInput(writer.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new FlushSummaryWriter(opBuilder.build());
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "FlushSummaryWriter";
+  
+  private FlushSummaryWriter(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/HistogramSummary.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/HistogramSummary.java
new file mode 100644
index 00000000000..b1bf74ba9d0
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/HistogramSummary.java
@@ -0,0 +1,82 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.summary;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TString;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Outputs a `Summary` protocol buffer with a histogram.
+ * <p>
+ * The generated
+ * [`Summary`](https://www.tensorflow.org/code/tensorflow/core/framework/summary.proto)
+ * has one summary value containing a histogram for `values`.
+ * <p>
+ * This op reports an `InvalidArgument` error if any value is not finite.
+ */
+@Operator(group = "summary")
+public final class HistogramSummary extends RawOp implements Operand<TString> {
+  
+  /**
+   * Factory method to create a class wrapping a new HistogramSummary operation.
+   * 
+   * @param scope current scope
+   * @param tag Scalar.  Tag to use for the `Summary.Value`.
+   * @param values Any shape. Values to use to build the histogram.
+   * @return a new instance of HistogramSummary
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> HistogramSummary create(Scope scope, Operand<TString> tag, Operand<T> values) {
+    OperationBuilder opBuilder = scope.env().opBuilder("HistogramSummary", scope.makeOpName("HistogramSummary"));
+    opBuilder.addInput(tag.asOutput());
+    opBuilder.addInput(values.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new HistogramSummary(opBuilder.build());
+  }
+  
+  /**
+   * Scalar. Serialized `Summary` protocol buffer.
+   */
+  public Output<TString> summary() {
+    return summary;
+  }
+  
+  @Override
+  public Output<TString> asOutput() {
+    return summary;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "HistogramSummary";
+  
+  private Output<TString> summary;
+  
+  private HistogramSummary(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    summary = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/ImageSummary.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/ImageSummary.java
new file mode 100644
index 00000000000..b97fbfcf8da
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/ImageSummary.java
@@ -0,0 +1,178 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.summary;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TString;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Outputs a `Summary` protocol buffer with images.
+ * <p>
+ * The summary has up to `max_images` summary values containing images. The
+ * images are built from `tensor` which must be 4-D with shape `[batch_size,
+ * height, width, channels]` and where `channels` can be:
+ * <ul>
+ * <li>
+ * 1: `tensor` is interpreted as Grayscale.
+ * </li>
+ * <li>
+ * 3: `tensor` is interpreted as RGB.
+ * </li>
+ * <li>
+ * 4: `tensor` is interpreted as RGBA.
+ * </li>
+ * </ul>
+ * The images have the same number of channels as the input tensor. For float
+ * input, the values are normalized one image at a time to fit in the range
+ * `[0, 255]`.  `uint8` values are unchanged.  The op uses two different
+ * normalization algorithms:
+ * <ul>
+ * <li>
+ * If the input values are all positive, they are rescaled so the largest one
+ *    is 255.
+ * </li>
+ * <li>
+ * If any input value is negative, the values are shifted so input value 0.0
+ *    is at 127.  They are then rescaled so that either the smallest value is 0,
+ *    or the largest one is 255.
+ * </li>
+ * </ul>
+ * The `tag` argument is a scalar `Tensor` of type `string`.  It is used to
+ * build the `tag` of the summary values:
+ * <ul>
+ * <li>
+ * If `max_images` is 1, the summary value tag is '<i>tag</i>/image'.
+ * </li>
+ * <li>
+ * If `max_images` is greater than 1, the summary value tags are
+ *    generated sequentially as '<i>tag</i>/image/0', '<i>tag</i>/image/1', etc.
+ * </li>
+ * </ul>
+ * The `bad_color` argument is the color to use in the generated images for
+ * non-finite input values.  It is a `uint8` 1-D tensor of length `channels`.
+ * Each element must be in the range `[0, 255]` (It represents the value of a
+ * pixel in the output image).  Non-finite values in the input tensor are
+ * replaced by this tensor in the output image.  The default value is the color
+ * red.
+ */
+@Operator(group = "summary")
+public final class ImageSummary extends RawOp implements Operand<TString> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.summary.ImageSummary}
+   */
+  public static class Options {
+    
+    /**
+     * @param maxImages Max number of batch elements to generate images for.
+     */
+    public Options maxImages(Long maxImages) {
+      this.maxImages = maxImages;
+      return this;
+    }
+    
+    /**
+     * @param badColor Color to use for pixels with non-finite values.
+     */
+    public Options badColor(TType badColor) {
+      this.badColor = badColor;
+      return this;
+    }
+    
+    private Long maxImages;
+    private TType badColor;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new ImageSummary operation.
+   * 
+   * @param scope current scope
+   * @param tag Scalar. Used to build the `tag` attribute of the summary values.
+   * @param tensor 4-D of shape `[batch_size, height, width, channels]` where
+   * `channels` is 1, 3, or 4.
+   * @param options carries optional attributes values
+   * @return a new instance of ImageSummary
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> ImageSummary create(Scope scope, Operand<TString> tag, Operand<T> tensor, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ImageSummary", scope.makeOpName("ImageSummary"));
+    opBuilder.addInput(tag.asOutput());
+    opBuilder.addInput(tensor.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.maxImages != null) {
+          opBuilder.setAttr("max_images", opts.maxImages);
+        }
+        if (opts.badColor != null) {
+          opBuilder.setAttr("bad_color", opts.badColor);
+        }
+      }
+    }
+    return new ImageSummary(opBuilder.build());
+  }
+  
+  /**
+   * @param maxImages Max number of batch elements to generate images for.
+   */
+  public static Options maxImages(Long maxImages) {
+    return new Options().maxImages(maxImages);
+  }
+  
+  /**
+   * @param badColor Color to use for pixels with non-finite values.
+   */
+  public static Options badColor(TType badColor) {
+    return new Options().badColor(badColor);
+  }
+  
+  /**
+   * Scalar. Serialized `Summary` protocol buffer.
+   */
+  public Output<TString> summary() {
+    return summary;
+  }
+  
+  @Override
+  public Output<TString> asOutput() {
+    return summary;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ImageSummary";
+  
+  private Output<TString> summary;
+  
+  private ImageSummary(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    summary = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/ImportEvent.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/ImportEvent.java
new file mode 100644
index 00000000000..7bd97de571e
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/ImportEvent.java
@@ -0,0 +1,56 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.summary;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TString;
+
+/**
+ */
+public final class ImportEvent extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new ImportEvent operation.
+   * 
+   * @param scope current scope
+   * @param writer 
+   * @param event 
+   * @return a new instance of ImportEvent
+   */
+  @Endpoint(describeByClass = true)
+  public static ImportEvent create(Scope scope, Operand<?> writer, Operand<TString> event) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ImportEvent", scope.makeOpName("ImportEvent"));
+    opBuilder.addInput(writer.asOutput());
+    opBuilder.addInput(event.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new ImportEvent(opBuilder.build());
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ImportEvent";
+  
+  private ImportEvent(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/MergeSummary.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/MergeSummary.java
new file mode 100644
index 00000000000..e098bb8cf62
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/MergeSummary.java
@@ -0,0 +1,83 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.summary;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.Operands;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TString;
+
+/**
+ * Merges summaries.
+ * <p>
+ * This op creates a
+ * [`Summary`](https://www.tensorflow.org/code/tensorflow/core/framework/summary.proto)
+ * protocol buffer that contains the union of all the values in the input
+ * summaries.
+ * <p>
+ * When the Op is run, it reports an `InvalidArgument` error if multiple values
+ * in the summaries to merge use the same tag.
+ */
+@Operator(group = "summary")
+public final class MergeSummary extends RawOp implements Operand<TString> {
+  
+  /**
+   * Factory method to create a class wrapping a new MergeSummary operation.
+   * 
+   * @param scope current scope
+   * @param inputs Can be of any shape.  Each must contain serialized `Summary` protocol
+   * buffers.
+   * @return a new instance of MergeSummary
+   */
+  @Endpoint(describeByClass = true)
+  public static MergeSummary create(Scope scope, Iterable<Operand<TString>> inputs) {
+    OperationBuilder opBuilder = scope.env().opBuilder("MergeSummary", scope.makeOpName("MergeSummary"));
+    opBuilder.addInputList(Operands.asOutputs(inputs));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new MergeSummary(opBuilder.build());
+  }
+  
+  /**
+   * Scalar. Serialized `Summary` protocol buffer.
+   */
+  public Output<TString> summary() {
+    return summary;
+  }
+  
+  @Override
+  public Output<TString> asOutput() {
+    return summary;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "MergeSummary";
+  
+  private Output<TString> summary;
+  
+  private MergeSummary(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    summary = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/ScalarSummary.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/ScalarSummary.java
new file mode 100644
index 00000000000..20baf84097f
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/ScalarSummary.java
@@ -0,0 +1,79 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.summary;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TString;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Outputs a `Summary` protocol buffer with scalar values.
+ * <p>
+ * The input `tags` and `values` must have the same shape.  The generated summary
+ * has a summary value for each tag-value pair in `tags` and `values`.
+ */
+@Operator(group = "summary")
+public final class ScalarSummary extends RawOp implements Operand<TString> {
+  
+  /**
+   * Factory method to create a class wrapping a new ScalarSummary operation.
+   * 
+   * @param scope current scope
+   * @param tags Tags for the summary.
+   * @param values Same shape as `tags.  Values for the summary.
+   * @return a new instance of ScalarSummary
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> ScalarSummary create(Scope scope, Operand<TString> tags, Operand<T> values) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ScalarSummary", scope.makeOpName("ScalarSummary"));
+    opBuilder.addInput(tags.asOutput());
+    opBuilder.addInput(values.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new ScalarSummary(opBuilder.build());
+  }
+  
+  /**
+   * Scalar.  Serialized `Summary` protocol buffer.
+   */
+  public Output<TString> summary() {
+    return summary;
+  }
+  
+  @Override
+  public Output<TString> asOutput() {
+    return summary;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ScalarSummary";
+  
+  private Output<TString> summary;
+  
+  private ScalarSummary(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    summary = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/StatsAggregatorSummary.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/StatsAggregatorSummary.java
new file mode 100644
index 00000000000..9cc5fe6d83b
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/StatsAggregatorSummary.java
@@ -0,0 +1,71 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.summary;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TString;
+
+/**
+ * Produces a summary of any statistics recorded by the given statistics manager.
+ */
+public final class StatsAggregatorSummary extends RawOp implements Operand<TString> {
+  
+  /**
+   * Factory method to create a class wrapping a new StatsAggregatorSummary operation.
+   * 
+   * @param scope current scope
+   * @param iterator 
+   * @return a new instance of StatsAggregatorSummary
+   */
+  @Endpoint(describeByClass = true)
+  public static StatsAggregatorSummary create(Scope scope, Operand<?> iterator) {
+    OperationBuilder opBuilder = scope.env().opBuilder("StatsAggregatorSummary", scope.makeOpName("StatsAggregatorSummary"));
+    opBuilder.addInput(iterator.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new StatsAggregatorSummary(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<TString> summary() {
+    return summary;
+  }
+  
+  @Override
+  public Output<TString> asOutput() {
+    return summary;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "StatsAggregatorSummary";
+  
+  private Output<TString> summary;
+  
+  private StatsAggregatorSummary(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    summary = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/SummaryWriter.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/SummaryWriter.java
new file mode 100644
index 00000000000..131cb6e4a9d
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/SummaryWriter.java
@@ -0,0 +1,122 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.summary;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ */
+public final class SummaryWriter extends RawOp implements Operand<TType> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.summary.SummaryWriter}
+   */
+  public static class Options {
+    
+    /**
+     * @param sharedName 
+     */
+    public Options sharedName(String sharedName) {
+      this.sharedName = sharedName;
+      return this;
+    }
+    
+    /**
+     * @param container 
+     */
+    public Options container(String container) {
+      this.container = container;
+      return this;
+    }
+    
+    private String sharedName;
+    private String container;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new SummaryWriter operation.
+   * 
+   * @param scope current scope
+   * @param options carries optional attributes values
+   * @return a new instance of SummaryWriter
+   */
+  @Endpoint(describeByClass = true)
+  public static SummaryWriter create(Scope scope, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("SummaryWriter", scope.makeOpName("SummaryWriter"));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.sharedName != null) {
+          opBuilder.setAttr("shared_name", opts.sharedName);
+        }
+        if (opts.container != null) {
+          opBuilder.setAttr("container", opts.container);
+        }
+      }
+    }
+    return new SummaryWriter(opBuilder.build());
+  }
+  
+  /**
+   * @param sharedName 
+   */
+  public static Options sharedName(String sharedName) {
+    return new Options().sharedName(sharedName);
+  }
+  
+  /**
+   * @param container 
+   */
+  public static Options container(String container) {
+    return new Options().container(container);
+  }
+  
+  /**
+   */
+  public Output<?> writer() {
+    return writer;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) writer;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "SummaryWriter";
+  
+  private Output<?> writer;
+  
+  private SummaryWriter(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    writer = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/TensorSummary.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/TensorSummary.java
new file mode 100644
index 00000000000..85e0f8d533b
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/TensorSummary.java
@@ -0,0 +1,78 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.summary;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TString;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Outputs a `Summary` protocol buffer with a tensor and per-plugin data.
+ */
+@Operator(group = "summary")
+public final class TensorSummary extends RawOp implements Operand<TString> {
+  
+  /**
+   * Factory method to create a class wrapping a new TensorSummary operation.
+   * 
+   * @param scope current scope
+   * @param tag A string attached to this summary. Used for organization in TensorBoard.
+   * @param tensor A tensor to serialize.
+   * @param serializedSummaryMetadata A serialized SummaryMetadata proto. Contains plugin
+   * data.
+   * @return a new instance of TensorSummary
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> TensorSummary create(Scope scope, Operand<TString> tag, Operand<T> tensor, Operand<TString> serializedSummaryMetadata) {
+    OperationBuilder opBuilder = scope.env().opBuilder("TensorSummaryV2", scope.makeOpName("TensorSummary"));
+    opBuilder.addInput(tag.asOutput());
+    opBuilder.addInput(tensor.asOutput());
+    opBuilder.addInput(serializedSummaryMetadata.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new TensorSummary(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<TString> summary() {
+    return summary;
+  }
+  
+  @Override
+  public Output<TString> asOutput() {
+    return summary;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "TensorSummaryV2";
+  
+  private Output<TString> summary;
+  
+  private TensorSummary(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    summary = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/WriteAudioSummary.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/WriteAudioSummary.java
new file mode 100644
index 00000000000..8cdf23f4982
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/WriteAudioSummary.java
@@ -0,0 +1,98 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.summary;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.TString;
+
+/**
+ */
+public final class WriteAudioSummary extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.summary.WriteAudioSummary}
+   */
+  public static class Options {
+    
+    /**
+     * @param maxOutputs 
+     */
+    public Options maxOutputs(Long maxOutputs) {
+      this.maxOutputs = maxOutputs;
+      return this;
+    }
+    
+    private Long maxOutputs;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new WriteAudioSummary operation.
+   * 
+   * @param scope current scope
+   * @param writer 
+   * @param step 
+   * @param tag 
+   * @param tensor 
+   * @param sampleRate 
+   * @param options carries optional attributes values
+   * @return a new instance of WriteAudioSummary
+   */
+  @Endpoint(describeByClass = true)
+  public static WriteAudioSummary create(Scope scope, Operand<?> writer, Operand<TInt64> step, Operand<TString> tag, Operand<TFloat32> tensor, Operand<TFloat32> sampleRate, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("WriteAudioSummary", scope.makeOpName("WriteAudioSummary"));
+    opBuilder.addInput(writer.asOutput());
+    opBuilder.addInput(step.asOutput());
+    opBuilder.addInput(tag.asOutput());
+    opBuilder.addInput(tensor.asOutput());
+    opBuilder.addInput(sampleRate.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.maxOutputs != null) {
+          opBuilder.setAttr("max_outputs", opts.maxOutputs);
+        }
+      }
+    }
+    return new WriteAudioSummary(opBuilder.build());
+  }
+  
+  /**
+   * @param maxOutputs 
+   */
+  public static Options maxOutputs(Long maxOutputs) {
+    return new Options().maxOutputs(maxOutputs);
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "WriteAudioSummary";
+  
+  private WriteAudioSummary(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/WriteGraphSummary.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/WriteGraphSummary.java
new file mode 100644
index 00000000000..dc0cbe0f222
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/WriteGraphSummary.java
@@ -0,0 +1,59 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.summary;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.TString;
+
+/**
+ */
+public final class WriteGraphSummary extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new WriteGraphSummary operation.
+   * 
+   * @param scope current scope
+   * @param writer 
+   * @param step 
+   * @param tensor 
+   * @return a new instance of WriteGraphSummary
+   */
+  @Endpoint(describeByClass = true)
+  public static WriteGraphSummary create(Scope scope, Operand<?> writer, Operand<TInt64> step, Operand<TString> tensor) {
+    OperationBuilder opBuilder = scope.env().opBuilder("WriteGraphSummary", scope.makeOpName("WriteGraphSummary"));
+    opBuilder.addInput(writer.asOutput());
+    opBuilder.addInput(step.asOutput());
+    opBuilder.addInput(tensor.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new WriteGraphSummary(opBuilder.build());
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "WriteGraphSummary";
+  
+  private WriteGraphSummary(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/WriteHistogramSummary.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/WriteHistogramSummary.java
new file mode 100644
index 00000000000..e460ecc8ca2
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/WriteHistogramSummary.java
@@ -0,0 +1,62 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.summary;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.TString;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ */
+public final class WriteHistogramSummary extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new WriteHistogramSummary operation.
+   * 
+   * @param scope current scope
+   * @param writer 
+   * @param step 
+   * @param tag 
+   * @param values 
+   * @return a new instance of WriteHistogramSummary
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> WriteHistogramSummary create(Scope scope, Operand<?> writer, Operand<TInt64> step, Operand<TString> tag, Operand<T> values) {
+    OperationBuilder opBuilder = scope.env().opBuilder("WriteHistogramSummary", scope.makeOpName("WriteHistogramSummary"));
+    opBuilder.addInput(writer.asOutput());
+    opBuilder.addInput(step.asOutput());
+    opBuilder.addInput(tag.asOutput());
+    opBuilder.addInput(values.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new WriteHistogramSummary(opBuilder.build());
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "WriteHistogramSummary";
+  
+  private WriteHistogramSummary(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/WriteImageSummary.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/WriteImageSummary.java
new file mode 100644
index 00000000000..ff10607477b
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/WriteImageSummary.java
@@ -0,0 +1,99 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.summary;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.TString;
+import org.tensorflow.types.TUint8;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ */
+public final class WriteImageSummary extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.summary.WriteImageSummary}
+   */
+  public static class Options {
+    
+    /**
+     * @param maxImages 
+     */
+    public Options maxImages(Long maxImages) {
+      this.maxImages = maxImages;
+      return this;
+    }
+    
+    private Long maxImages;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new WriteImageSummary operation.
+   * 
+   * @param scope current scope
+   * @param writer 
+   * @param step 
+   * @param tag 
+   * @param tensor 
+   * @param badColor 
+   * @param options carries optional attributes values
+   * @return a new instance of WriteImageSummary
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> WriteImageSummary create(Scope scope, Operand<?> writer, Operand<TInt64> step, Operand<TString> tag, Operand<T> tensor, Operand<TUint8> badColor, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("WriteImageSummary", scope.makeOpName("WriteImageSummary"));
+    opBuilder.addInput(writer.asOutput());
+    opBuilder.addInput(step.asOutput());
+    opBuilder.addInput(tag.asOutput());
+    opBuilder.addInput(tensor.asOutput());
+    opBuilder.addInput(badColor.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.maxImages != null) {
+          opBuilder.setAttr("max_images", opts.maxImages);
+        }
+      }
+    }
+    return new WriteImageSummary(opBuilder.build());
+  }
+  
+  /**
+   * @param maxImages 
+   */
+  public static Options maxImages(Long maxImages) {
+    return new Options().maxImages(maxImages);
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "WriteImageSummary";
+  
+  private WriteImageSummary(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/WriteRawProtoSummary.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/WriteRawProtoSummary.java
new file mode 100644
index 00000000000..75499c1ff69
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/WriteRawProtoSummary.java
@@ -0,0 +1,59 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.summary;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.TString;
+
+/**
+ */
+public final class WriteRawProtoSummary extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new WriteRawProtoSummary operation.
+   * 
+   * @param scope current scope
+   * @param writer 
+   * @param step 
+   * @param tensor 
+   * @return a new instance of WriteRawProtoSummary
+   */
+  @Endpoint(describeByClass = true)
+  public static WriteRawProtoSummary create(Scope scope, Operand<?> writer, Operand<TInt64> step, Operand<TString> tensor) {
+    OperationBuilder opBuilder = scope.env().opBuilder("WriteRawProtoSummary", scope.makeOpName("WriteRawProtoSummary"));
+    opBuilder.addInput(writer.asOutput());
+    opBuilder.addInput(step.asOutput());
+    opBuilder.addInput(tensor.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new WriteRawProtoSummary(opBuilder.build());
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "WriteRawProtoSummary";
+  
+  private WriteRawProtoSummary(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/WriteScalarSummary.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/WriteScalarSummary.java
new file mode 100644
index 00000000000..af1291b964f
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/WriteScalarSummary.java
@@ -0,0 +1,62 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.summary;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.TString;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ */
+public final class WriteScalarSummary extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new WriteScalarSummary operation.
+   * 
+   * @param scope current scope
+   * @param writer 
+   * @param step 
+   * @param tag 
+   * @param value 
+   * @return a new instance of WriteScalarSummary
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> WriteScalarSummary create(Scope scope, Operand<?> writer, Operand<TInt64> step, Operand<TString> tag, Operand<T> value) {
+    OperationBuilder opBuilder = scope.env().opBuilder("WriteScalarSummary", scope.makeOpName("WriteScalarSummary"));
+    opBuilder.addInput(writer.asOutput());
+    opBuilder.addInput(step.asOutput());
+    opBuilder.addInput(tag.asOutput());
+    opBuilder.addInput(value.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new WriteScalarSummary(opBuilder.build());
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "WriteScalarSummary";
+  
+  private WriteScalarSummary(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/WriteSummary.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/WriteSummary.java
new file mode 100644
index 00000000000..5404e593f27
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/summary/WriteSummary.java
@@ -0,0 +1,64 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.summary;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.TString;
+import org.tensorflow.types.family.TType;
+
+/**
+ */
+public final class WriteSummary extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new WriteSummary operation.
+   * 
+   * @param scope current scope
+   * @param writer 
+   * @param step 
+   * @param tensor 
+   * @param tag 
+   * @param summaryMetadata 
+   * @return a new instance of WriteSummary
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> WriteSummary create(Scope scope, Operand<?> writer, Operand<TInt64> step, Operand<T> tensor, Operand<TString> tag, Operand<TString> summaryMetadata) {
+    OperationBuilder opBuilder = scope.env().opBuilder("WriteSummary", scope.makeOpName("WriteSummary"));
+    opBuilder.addInput(writer.asOutput());
+    opBuilder.addInput(step.asOutput());
+    opBuilder.addInput(tensor.asOutput());
+    opBuilder.addInput(tag.asOutput());
+    opBuilder.addInput(summaryMetadata.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new WriteSummary(opBuilder.build());
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "WriteSummary";
+  
+  private WriteSummary(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/AllToAll.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/AllToAll.java
new file mode 100644
index 00000000000..9bbac83ac2c
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/AllToAll.java
@@ -0,0 +1,103 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.tpu;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TType;
+
+/**
+ * An Op to exchange data across TPU replicas.
+ * <p>
+ * On each replica, the input is split into `split_count` blocks along
+ * `split_dimension` and send to the other replicas given group_assignment. After
+ * receiving `split_count` - 1 blocks from other replicas, we concatenate the
+ * blocks along `concat_dimension` as the output.
+ * <p>
+ * For example, suppose there are 2 TPU replicas:
+ * replica 0 receives input: `[[A, B]]`
+ * replica 1 receives input: `[[C, D]]`
+ * <p>
+ * group_assignment=`[[0, 1]]`
+ * concat_dimension=0
+ * split_dimension=1
+ * split_count=2
+ * <p>
+ * replica 0's output: `[[A], [C]]`
+ * replica 1's output: `[[B], [D]]`
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+public final class AllToAll<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new AllToAll operation.
+   * 
+   * @param scope current scope
+   * @param input The local input to the sum.
+   * @param groupAssignment An int32 tensor with shape
+   * [num_groups, num_replicas_per_group]. `group_assignment[i]` represents the
+   * replica ids in the ith subgroup.
+   * @param concatDimension The dimension number to concatenate.
+   * @param splitDimension The dimension number to split.
+   * @param splitCount The number of splits, this number must equal to the sub-group
+   * size(group_assignment.get_shape()[1])
+   * @return a new instance of AllToAll
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> AllToAll<T> create(Scope scope, Operand<T> input, Operand<TInt32> groupAssignment, Long concatDimension, Long splitDimension, Long splitCount) {
+    OperationBuilder opBuilder = scope.env().opBuilder("AllToAll", scope.makeOpName("AllToAll"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(groupAssignment.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("concat_dimension", concatDimension);
+    opBuilder.setAttr("split_dimension", splitDimension);
+    opBuilder.setAttr("split_count", splitCount);
+    return new AllToAll<T>(opBuilder.build());
+  }
+  
+  /**
+   * The exchanged result.
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "AllToAll";
+  
+  private Output<T> output;
+  
+  private AllToAll(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/CollectivePermute.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/CollectivePermute.java
new file mode 100644
index 00000000000..bafa9a893f9
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/CollectivePermute.java
@@ -0,0 +1,84 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.tpu;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TType;
+
+/**
+ * An Op to permute tensors across replicated TPU instances.
+ * <p>
+ * Each instance supplies its own input.
+ * <p>
+ * For example, suppose there are 4 TPU instances: `[A, B, C, D]`. Passing
+ * source_target_pairs=`[[0,1],[1,2],[2,3],[3,0]]` gets the outputs:
+ * `[D, A, B, C]`.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+public final class CollectivePermute<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new CollectivePermute operation.
+   * 
+   * @param scope current scope
+   * @param input The local input to be permuted. Currently only supports float and
+   * bfloat16.
+   * @param sourceTargetPairs A tensor with shape [num_pairs, 2].
+   * @return a new instance of CollectivePermute
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> CollectivePermute<T> create(Scope scope, Operand<T> input, Operand<TInt32> sourceTargetPairs) {
+    OperationBuilder opBuilder = scope.env().opBuilder("CollectivePermute", scope.makeOpName("CollectivePermute"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(sourceTargetPairs.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new CollectivePermute<T>(opBuilder.build());
+  }
+  
+  /**
+   * The permuted input.
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "CollectivePermute";
+  
+  private Output<T> output;
+  
+  private CollectivePermute(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/ConfigureDistributedTPU.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/ConfigureDistributedTPU.java
new file mode 100644
index 00000000000..bfdde5b98e2
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/ConfigureDistributedTPU.java
@@ -0,0 +1,183 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.tpu;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TString;
+
+/**
+ * Sets up the centralized structures for a distributed TPU system.
+ */
+public final class ConfigureDistributedTPU extends RawOp implements Operand<TString> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.tpu.ConfigureDistributedTPU}
+   */
+  public static class Options {
+    
+    /**
+     * @param embeddingConfig Reserved. Do not use.
+     */
+    public Options embeddingConfig(String embeddingConfig) {
+      this.embeddingConfig = embeddingConfig;
+      return this;
+    }
+    
+    /**
+     * @param tpuEmbeddingConfig Serialized tensorflow.tpu.TPUEmbeddingConfiguration that
+     * describes the embedding lookups of the program.
+     */
+    public Options tpuEmbeddingConfig(String tpuEmbeddingConfig) {
+      this.tpuEmbeddingConfig = tpuEmbeddingConfig;
+      return this;
+    }
+    
+    /**
+     * @param isGlobalInit Reserved. Do not use.
+     */
+    public Options isGlobalInit(Boolean isGlobalInit) {
+      this.isGlobalInit = isGlobalInit;
+      return this;
+    }
+    
+    /**
+     * @param enableWholeMeshCompilations 
+     */
+    public Options enableWholeMeshCompilations(Boolean enableWholeMeshCompilations) {
+      this.enableWholeMeshCompilations = enableWholeMeshCompilations;
+      return this;
+    }
+    
+    /**
+     * @param compilationFailureClosesChips 
+     */
+    public Options compilationFailureClosesChips(Boolean compilationFailureClosesChips) {
+      this.compilationFailureClosesChips = compilationFailureClosesChips;
+      return this;
+    }
+    
+    private String embeddingConfig;
+    private String tpuEmbeddingConfig;
+    private Boolean isGlobalInit;
+    private Boolean enableWholeMeshCompilations;
+    private Boolean compilationFailureClosesChips;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new ConfigureDistributedTPU operation.
+   * 
+   * @param scope current scope
+   * @param options carries optional attributes values
+   * @return a new instance of ConfigureDistributedTPU
+   */
+  @Endpoint(describeByClass = true)
+  public static ConfigureDistributedTPU create(Scope scope, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ConfigureDistributedTPU", scope.makeOpName("ConfigureDistributedTPU"));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.embeddingConfig != null) {
+          opBuilder.setAttr("embedding_config", opts.embeddingConfig);
+        }
+        if (opts.tpuEmbeddingConfig != null) {
+          opBuilder.setAttr("tpu_embedding_config", opts.tpuEmbeddingConfig);
+        }
+        if (opts.isGlobalInit != null) {
+          opBuilder.setAttr("is_global_init", opts.isGlobalInit);
+        }
+        if (opts.enableWholeMeshCompilations != null) {
+          opBuilder.setAttr("enable_whole_mesh_compilations", opts.enableWholeMeshCompilations);
+        }
+        if (opts.compilationFailureClosesChips != null) {
+          opBuilder.setAttr("compilation_failure_closes_chips", opts.compilationFailureClosesChips);
+        }
+      }
+    }
+    return new ConfigureDistributedTPU(opBuilder.build());
+  }
+  
+  /**
+   * @param embeddingConfig Reserved. Do not use.
+   */
+  public static Options embeddingConfig(String embeddingConfig) {
+    return new Options().embeddingConfig(embeddingConfig);
+  }
+  
+  /**
+   * @param tpuEmbeddingConfig Serialized tensorflow.tpu.TPUEmbeddingConfiguration that
+   * describes the embedding lookups of the program.
+   */
+  public static Options tpuEmbeddingConfig(String tpuEmbeddingConfig) {
+    return new Options().tpuEmbeddingConfig(tpuEmbeddingConfig);
+  }
+  
+  /**
+   * @param isGlobalInit Reserved. Do not use.
+   */
+  public static Options isGlobalInit(Boolean isGlobalInit) {
+    return new Options().isGlobalInit(isGlobalInit);
+  }
+  
+  /**
+   * @param enableWholeMeshCompilations 
+   */
+  public static Options enableWholeMeshCompilations(Boolean enableWholeMeshCompilations) {
+    return new Options().enableWholeMeshCompilations(enableWholeMeshCompilations);
+  }
+  
+  /**
+   * @param compilationFailureClosesChips 
+   */
+  public static Options compilationFailureClosesChips(Boolean compilationFailureClosesChips) {
+    return new Options().compilationFailureClosesChips(compilationFailureClosesChips);
+  }
+  
+  /**
+   * A serialized tensorflow.tpu.TopologyProto that describes the TPU
+   * topology.
+   */
+  public Output<TString> topology() {
+    return topology;
+  }
+  
+  @Override
+  public Output<TString> asOutput() {
+    return topology;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ConfigureDistributedTPU";
+  
+  private Output<TString> topology;
+  
+  private ConfigureDistributedTPU(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    topology = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/ConfigureTPUEmbedding.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/ConfigureTPUEmbedding.java
new file mode 100644
index 00000000000..76bccd51f83
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/ConfigureTPUEmbedding.java
@@ -0,0 +1,54 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.tpu;
+
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+
+/**
+ * Sets up TPUEmbedding in a distributed TPU system.
+ */
+public final class ConfigureTPUEmbedding extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new ConfigureTPUEmbedding operation.
+   * 
+   * @param scope current scope
+   * @param config Serialized tensorflow.tpu.TPUEmbeddingConfiguration that
+   * describes the embedding lookups of the program.
+   * @return a new instance of ConfigureTPUEmbedding
+   */
+  @Endpoint(describeByClass = true)
+  public static ConfigureTPUEmbedding create(Scope scope, String config) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ConfigureTPUEmbedding", scope.makeOpName("ConfigureTPUEmbedding"));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("config", config);
+    return new ConfigureTPUEmbedding(opBuilder.build());
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ConfigureTPUEmbedding";
+  
+  private ConfigureTPUEmbedding(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/CrossReplicaSum.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/CrossReplicaSum.java
new file mode 100644
index 00000000000..d4f982dce61
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/CrossReplicaSum.java
@@ -0,0 +1,86 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.tpu;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * An Op to sum inputs across replicated TPU instances.
+ * <p>
+ * Each instance supplies its own input.
+ * <p>
+ * For example, suppose there are 8 TPU instances: `[A, B, C, D, E, F, G, H]`.
+ * Passing group_assignment=`[[0,2,4,6],[1,3,5,7]]` sets `A, C, E, G` as group 0,
+ * and `B, D, F, H` as group 1. Thus we get the outputs:
+ * `[A+C+E+G, B+D+F+H, A+C+E+G, B+D+F+H, A+C+E+G, B+D+F+H, A+C+E+G, B+D+F+H]`.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+public final class CrossReplicaSum<T extends TNumber> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new CrossReplicaSum operation.
+   * 
+   * @param scope current scope
+   * @param input The local input to the sum.
+   * @param groupAssignment An int32 tensor with shape
+   * [num_groups, num_replicas_per_group]. `group_assignment[i]` represents the
+   * replica ids in the ith subgroup.
+   * @return a new instance of CrossReplicaSum
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber> CrossReplicaSum<T> create(Scope scope, Operand<T> input, Operand<TInt32> groupAssignment) {
+    OperationBuilder opBuilder = scope.env().opBuilder("CrossReplicaSum", scope.makeOpName("CrossReplicaSum"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(groupAssignment.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new CrossReplicaSum<T>(opBuilder.build());
+  }
+  
+  /**
+   * The sum of all the distributed inputs.
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "CrossReplicaSum";
+  
+  private Output<T> output;
+  
+  private CrossReplicaSum(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/EnqueueTPUEmbeddingIntegerBatch.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/EnqueueTPUEmbeddingIntegerBatch.java
new file mode 100644
index 00000000000..0a1a80c7a0a
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/EnqueueTPUEmbeddingIntegerBatch.java
@@ -0,0 +1,99 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.tpu;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.Operands;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.TString;
+
+/**
+ * An op that enqueues a list of input batch tensors to TPUEmbedding.
+ */
+public final class EnqueueTPUEmbeddingIntegerBatch extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.tpu.EnqueueTPUEmbeddingIntegerBatch}
+   */
+  public static class Options {
+    
+    /**
+     * @param deviceOrdinal The TPU device to use. Should be >= 0 and less than the number
+     * of TPU cores in the task on which the node is placed.
+     */
+    public Options deviceOrdinal(Long deviceOrdinal) {
+      this.deviceOrdinal = deviceOrdinal;
+      return this;
+    }
+    
+    private Long deviceOrdinal;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new EnqueueTPUEmbeddingIntegerBatch operation.
+   * 
+   * @param scope current scope
+   * @param batch A list of 1D tensors, one for each embedding table, containing the
+   * indices into the tables.
+   * @param modeOverride A string input that overrides the mode specified in the
+   * TPUEmbeddingConfiguration. Supported values are {'unspecified', 'inference',
+   * 'training', 'backward_pass_only'}. When set to 'unspecified', the mode set
+   * in TPUEmbeddingConfiguration is used, otherwise mode_override is used.
+   * @param options carries optional attributes values
+   * @return a new instance of EnqueueTPUEmbeddingIntegerBatch
+   */
+  @Endpoint(describeByClass = true)
+  public static EnqueueTPUEmbeddingIntegerBatch create(Scope scope, Iterable<Operand<TInt32>> batch, Operand<TString> modeOverride, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("EnqueueTPUEmbeddingIntegerBatch", scope.makeOpName("EnqueueTPUEmbeddingIntegerBatch"));
+    opBuilder.addInputList(Operands.asOutputs(batch));
+    opBuilder.addInput(modeOverride.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.deviceOrdinal != null) {
+          opBuilder.setAttr("device_ordinal", opts.deviceOrdinal);
+        }
+      }
+    }
+    return new EnqueueTPUEmbeddingIntegerBatch(opBuilder.build());
+  }
+  
+  /**
+   * @param deviceOrdinal The TPU device to use. Should be >= 0 and less than the number
+   * of TPU cores in the task on which the node is placed.
+   */
+  public static Options deviceOrdinal(Long deviceOrdinal) {
+    return new Options().deviceOrdinal(deviceOrdinal);
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "EnqueueTPUEmbeddingIntegerBatch";
+  
+  private EnqueueTPUEmbeddingIntegerBatch(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/EnqueueTPUEmbeddingRaggedTensorBatch.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/EnqueueTPUEmbeddingRaggedTensorBatch.java
new file mode 100644
index 00000000000..ea2961fa2b8
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/EnqueueTPUEmbeddingRaggedTensorBatch.java
@@ -0,0 +1,185 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.tpu;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.Operands;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TString;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Eases the porting of code that uses tf.nn.embedding_lookup().
+ * <p>
+ * sample_splits[i], embedding_indices[i] and aggregation_weights[i] correspond
+ * to the ith feature. table_ids[i] indicates which embedding table to look up ith
+ * feature.
+ * <p>
+ * The tensors at corresponding positions in two of the input lists,
+ * embedding_indices and aggregation_weights, must have the same shape, i.e. rank 1
+ * with dim_size() equal to the total number of lookups into the table described by
+ * the corresponding feature.
+ */
+public final class EnqueueTPUEmbeddingRaggedTensorBatch extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.tpu.EnqueueTPUEmbeddingRaggedTensorBatch}
+   */
+  public static class Options {
+    
+    /**
+     * @param deviceOrdinal The TPU device to use. Should be >= 0 and less than the number
+     * of TPU cores in the task on which the node is placed.
+     */
+    public Options deviceOrdinal(Long deviceOrdinal) {
+      this.deviceOrdinal = deviceOrdinal;
+      return this;
+    }
+    
+    /**
+     * @param combiners A list of string scalars, one for each embedding table that specify
+     * how to normalize the embedding activations after weighted summation.
+     * Supported combiners are 'mean', 'sum', or 'sqrtn'. It is invalid to have
+     * the sum of the weights be 0 for 'mean' or the sum of the squared weights be
+     * 0 for 'sqrtn'. If combiners isn't passed, the default is to use 'sum' for
+     * all tables.
+     */
+    public Options combiners(List<String> combiners) {
+      this.combiners = combiners;
+      return this;
+    }
+    
+    /**
+     * @param maxSequenceLengths 
+     */
+    public Options maxSequenceLengths(List<Long> maxSequenceLengths) {
+      this.maxSequenceLengths = maxSequenceLengths;
+      return this;
+    }
+    
+    private Long deviceOrdinal;
+    private List<String> combiners;
+    private List<Long> maxSequenceLengths;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new EnqueueTPUEmbeddingRaggedTensorBatch operation.
+   * 
+   * @param scope current scope
+   * @param sampleSplits A list of rank 1 Tensors specifying the break points for splitting
+   * embedding_indices and aggregation_weights into rows.
+   * It corresponds to ids.row_splits in embedding_lookup(), when ids is a
+   * RaggedTensor.
+   * @param embeddingIndices A list of rank 1 Tensors, indices into the embedding tables.
+   * It corresponds to ids.values in embedding_lookup(), when ids is a RaggedTensor.
+   * @param aggregationWeights A list of rank 1 Tensors containing per training example
+   * aggregation weights. It corresponds to the values field of a RaggedTensor
+   * with the same row_splits as ids in embedding_lookup(), when ids is a
+   * RaggedTensor.
+   * @param modeOverride A string input that overrides the mode specified in the
+   * TPUEmbeddingConfiguration. Supported values are {'unspecified', 'inference',
+   * 'training', 'backward_pass_only'}. When set to 'unspecified', the mode set
+   * in TPUEmbeddingConfiguration is used, otherwise mode_override is used.
+   * @param tableIds A list of integers specifying the identifier of the embedding table
+   * (offset of TableDescriptor in the TPUEmbeddingConfiguration) to lookup the
+   * corresponding input. The ith input is looked up using table_ids[i]. The size
+   * of the table_ids list must be equal to that of sample_indices,
+   * embedding_indices and aggregation_weights.
+   * @param options carries optional attributes values
+   * @return a new instance of EnqueueTPUEmbeddingRaggedTensorBatch
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber, U extends TNumber, V extends TNumber> EnqueueTPUEmbeddingRaggedTensorBatch create(Scope scope, Iterable<Operand<T>> sampleSplits, Iterable<Operand<U>> embeddingIndices, Iterable<Operand<V>> aggregationWeights, Operand<TString> modeOverride, List<Long> tableIds, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("EnqueueTPUEmbeddingRaggedTensorBatch", scope.makeOpName("EnqueueTPUEmbeddingRaggedTensorBatch"));
+    opBuilder.addInputList(Operands.asOutputs(sampleSplits));
+    opBuilder.addInputList(Operands.asOutputs(embeddingIndices));
+    opBuilder.addInputList(Operands.asOutputs(aggregationWeights));
+    opBuilder.addInput(modeOverride.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    long[] tableIdsArray = new long[tableIds.size()];
+    for (int i = 0; i < tableIdsArray.length; ++i) {
+      tableIdsArray[i] = tableIds.get(i);
+    }
+    opBuilder.setAttr("table_ids", tableIdsArray);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.deviceOrdinal != null) {
+          opBuilder.setAttr("device_ordinal", opts.deviceOrdinal);
+        }
+        if (opts.combiners != null) {
+          String[] combinersArray = new String[opts.combiners.size()];
+          for (int i = 0; i < combinersArray.length; ++i) {
+            combinersArray[i] = opts.combiners.get(i);
+          }
+          opBuilder.setAttr("combiners", combinersArray);
+        }
+        if (opts.maxSequenceLengths != null) {
+          long[] maxSequenceLengthsArray = new long[opts.maxSequenceLengths.size()];
+          for (int i = 0; i < maxSequenceLengthsArray.length; ++i) {
+            maxSequenceLengthsArray[i] = opts.maxSequenceLengths.get(i);
+          }
+          opBuilder.setAttr("max_sequence_lengths", maxSequenceLengthsArray);
+        }
+      }
+    }
+    return new EnqueueTPUEmbeddingRaggedTensorBatch(opBuilder.build());
+  }
+  
+  /**
+   * @param deviceOrdinal The TPU device to use. Should be >= 0 and less than the number
+   * of TPU cores in the task on which the node is placed.
+   */
+  public static Options deviceOrdinal(Long deviceOrdinal) {
+    return new Options().deviceOrdinal(deviceOrdinal);
+  }
+  
+  /**
+   * @param combiners A list of string scalars, one for each embedding table that specify
+   * how to normalize the embedding activations after weighted summation.
+   * Supported combiners are 'mean', 'sum', or 'sqrtn'. It is invalid to have
+   * the sum of the weights be 0 for 'mean' or the sum of the squared weights be
+   * 0 for 'sqrtn'. If combiners isn't passed, the default is to use 'sum' for
+   * all tables.
+   */
+  public static Options combiners(List<String> combiners) {
+    return new Options().combiners(combiners);
+  }
+  
+  /**
+   * @param maxSequenceLengths 
+   */
+  public static Options maxSequenceLengths(List<Long> maxSequenceLengths) {
+    return new Options().maxSequenceLengths(maxSequenceLengths);
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "EnqueueTPUEmbeddingRaggedTensorBatch";
+  
+  private EnqueueTPUEmbeddingRaggedTensorBatch(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/EnqueueTPUEmbeddingSparseBatch.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/EnqueueTPUEmbeddingSparseBatch.java
new file mode 100644
index 00000000000..67c4d41ea03
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/EnqueueTPUEmbeddingSparseBatch.java
@@ -0,0 +1,151 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.tpu;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.Operands;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TString;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * An op that enqueues TPUEmbedding input indices from a SparseTensor.
+ * <p>
+ * This Op eases the porting of code that uses embedding_lookup_sparse(),
+ * although some Python preprocessing of the SparseTensor arguments to
+ * embedding_lookup_sparse() is required to produce the arguments to this Op,
+ * since only a single EnqueueTPUEmbeddingSparseBatch Op is allowed per training
+ * step.
+ * <p>
+ * The tensors at corresponding positions in the three input lists
+ * must have the same shape, i.e. rank 1 with dim_size() equal to the total
+ * number of lookups into the table described by the corresponding table_id.
+ */
+public final class EnqueueTPUEmbeddingSparseBatch extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.tpu.EnqueueTPUEmbeddingSparseBatch}
+   */
+  public static class Options {
+    
+    /**
+     * @param deviceOrdinal The TPU device to use. Should be >= 0 and less than the number
+     * of TPU cores in the task on which the node is placed.
+     */
+    public Options deviceOrdinal(Long deviceOrdinal) {
+      this.deviceOrdinal = deviceOrdinal;
+      return this;
+    }
+    
+    /**
+     * @param combiners A list of string scalars, one for each embedding table that specify
+     * how to normalize the embedding activations after weighted summation.
+     * Supported combiners are 'mean', 'sum', or 'sqrtn'. It is invalid to have
+     * the sum of the weights be 0 for 'mean' or the sum of the squared weights be
+     * 0 for 'sqrtn'. If combiners isn't passed, the default is to use 'sum' for
+     * all tables.
+     */
+    public Options combiners(List<String> combiners) {
+      this.combiners = combiners;
+      return this;
+    }
+    
+    private Long deviceOrdinal;
+    private List<String> combiners;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new EnqueueTPUEmbeddingSparseBatch operation.
+   * 
+   * @param scope current scope
+   * @param sampleIndices A list of rank 1 Tensors specifying the training example and
+   * feature to which the corresponding embedding_indices and aggregation_weights
+   * values belong. sample_indices[i] must equal b * nf + f, where nf is the
+   * number of features from the corresponding table, f is in [0, nf), and
+   * b is in [0, batch size).
+   * @param embeddingIndices A list of rank 1 Tensors, indices into the embedding tables.
+   * @param aggregationWeights A list of rank 1 Tensors containing per sample -- i.e. per
+   * (training example, feature) -- aggregation weights.
+   * @param modeOverride A string input that overrides the mode specified in the
+   * TPUEmbeddingConfiguration. Supported values are {'unspecified', 'inference',
+   * 'training', 'backward_pass_only'}. When set to 'unspecified', the mode set
+   * in TPUEmbeddingConfiguration is used, otherwise mode_override is used.
+   * @param options carries optional attributes values
+   * @return a new instance of EnqueueTPUEmbeddingSparseBatch
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber, U extends TNumber, V extends TNumber> EnqueueTPUEmbeddingSparseBatch create(Scope scope, Iterable<Operand<T>> sampleIndices, Iterable<Operand<U>> embeddingIndices, Iterable<Operand<V>> aggregationWeights, Operand<TString> modeOverride, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("EnqueueTPUEmbeddingSparseBatch", scope.makeOpName("EnqueueTPUEmbeddingSparseBatch"));
+    opBuilder.addInputList(Operands.asOutputs(sampleIndices));
+    opBuilder.addInputList(Operands.asOutputs(embeddingIndices));
+    opBuilder.addInputList(Operands.asOutputs(aggregationWeights));
+    opBuilder.addInput(modeOverride.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.deviceOrdinal != null) {
+          opBuilder.setAttr("device_ordinal", opts.deviceOrdinal);
+        }
+        if (opts.combiners != null) {
+          String[] combinersArray = new String[opts.combiners.size()];
+          for (int i = 0; i < combinersArray.length; ++i) {
+            combinersArray[i] = opts.combiners.get(i);
+          }
+          opBuilder.setAttr("combiners", combinersArray);
+        }
+      }
+    }
+    return new EnqueueTPUEmbeddingSparseBatch(opBuilder.build());
+  }
+  
+  /**
+   * @param deviceOrdinal The TPU device to use. Should be >= 0 and less than the number
+   * of TPU cores in the task on which the node is placed.
+   */
+  public static Options deviceOrdinal(Long deviceOrdinal) {
+    return new Options().deviceOrdinal(deviceOrdinal);
+  }
+  
+  /**
+   * @param combiners A list of string scalars, one for each embedding table that specify
+   * how to normalize the embedding activations after weighted summation.
+   * Supported combiners are 'mean', 'sum', or 'sqrtn'. It is invalid to have
+   * the sum of the weights be 0 for 'mean' or the sum of the squared weights be
+   * 0 for 'sqrtn'. If combiners isn't passed, the default is to use 'sum' for
+   * all tables.
+   */
+  public static Options combiners(List<String> combiners) {
+    return new Options().combiners(combiners);
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "EnqueueTPUEmbeddingSparseBatch";
+  
+  private EnqueueTPUEmbeddingSparseBatch(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/EnqueueTPUEmbeddingSparseTensorBatch.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/EnqueueTPUEmbeddingSparseTensorBatch.java
new file mode 100644
index 00000000000..df9ea11bf37
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/EnqueueTPUEmbeddingSparseTensorBatch.java
@@ -0,0 +1,183 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.tpu;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.Operands;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TString;
+import org.tensorflow.types.family.TNumber;
+
+/**
+ * Eases the porting of code that uses tf.nn.embedding_lookup_sparse().
+ * <p>
+ * sample_indices[i], embedding_indices[i] and aggregation_weights[i] correspond
+ * to the ith feature. table_ids[i] indicates which embedding table to look up ith
+ * feature.
+ * <p>
+ * The tensors at corresponding positions in the three input lists (sample_indices,
+ * embedding_indices and aggregation_weights) must have the same shape, i.e. rank 1
+ * with dim_size() equal to the total number of lookups into the table described by
+ * the corresponding feature.
+ */
+public final class EnqueueTPUEmbeddingSparseTensorBatch extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.tpu.EnqueueTPUEmbeddingSparseTensorBatch}
+   */
+  public static class Options {
+    
+    /**
+     * @param deviceOrdinal The TPU device to use. Should be >= 0 and less than the number
+     * of TPU cores in the task on which the node is placed.
+     */
+    public Options deviceOrdinal(Long deviceOrdinal) {
+      this.deviceOrdinal = deviceOrdinal;
+      return this;
+    }
+    
+    /**
+     * @param combiners A list of string scalars, one for each embedding table that specify
+     * how to normalize the embedding activations after weighted summation.
+     * Supported combiners are 'mean', 'sum', or 'sqrtn'. It is invalid to have
+     * the sum of the weights be 0 for 'mean' or the sum of the squared weights be
+     * 0 for 'sqrtn'. If combiners isn't passed, the default is to use 'sum' for
+     * all tables.
+     */
+    public Options combiners(List<String> combiners) {
+      this.combiners = combiners;
+      return this;
+    }
+    
+    /**
+     * @param maxSequenceLengths 
+     */
+    public Options maxSequenceLengths(List<Long> maxSequenceLengths) {
+      this.maxSequenceLengths = maxSequenceLengths;
+      return this;
+    }
+    
+    private Long deviceOrdinal;
+    private List<String> combiners;
+    private List<Long> maxSequenceLengths;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new EnqueueTPUEmbeddingSparseTensorBatch operation.
+   * 
+   * @param scope current scope
+   * @param sampleIndices A list of rank 1 Tensors specifying the training example to
+   * which the corresponding embedding_indices and aggregation_weights values
+   * belong. It corresponds to sp_ids.indices[:,0] in  embedding_lookup_sparse().
+   * @param embeddingIndices A list of rank 1 Tensors, indices into the embedding tables.
+   * It corresponds to sp_ids.values in embedding_lookup_sparse().
+   * @param aggregationWeights A list of rank 1 Tensors containing per training example
+   * aggregation weights. It corresponds to sp_weights.values in
+   * embedding_lookup_sparse().
+   * @param modeOverride A string input that overrides the mode specified in the
+   * TPUEmbeddingConfiguration. Supported values are {'unspecified', 'inference',
+   * 'training', 'backward_pass_only'}. When set to 'unspecified', the mode set
+   * in TPUEmbeddingConfiguration is used, otherwise mode_override is used.
+   * @param tableIds A list of integers specifying the identifier of the embedding table
+   * (offset of TableDescriptor in the TPUEmbeddingConfiguration) to lookup the
+   * corresponding input. The ith input is looked up using table_ids[i]. The size
+   * of the table_ids list must be equal to that of sample_indices,
+   * embedding_indices and aggregation_weights.
+   * @param options carries optional attributes values
+   * @return a new instance of EnqueueTPUEmbeddingSparseTensorBatch
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber, U extends TNumber, V extends TNumber> EnqueueTPUEmbeddingSparseTensorBatch create(Scope scope, Iterable<Operand<T>> sampleIndices, Iterable<Operand<U>> embeddingIndices, Iterable<Operand<V>> aggregationWeights, Operand<TString> modeOverride, List<Long> tableIds, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("EnqueueTPUEmbeddingSparseTensorBatch", scope.makeOpName("EnqueueTPUEmbeddingSparseTensorBatch"));
+    opBuilder.addInputList(Operands.asOutputs(sampleIndices));
+    opBuilder.addInputList(Operands.asOutputs(embeddingIndices));
+    opBuilder.addInputList(Operands.asOutputs(aggregationWeights));
+    opBuilder.addInput(modeOverride.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    long[] tableIdsArray = new long[tableIds.size()];
+    for (int i = 0; i < tableIdsArray.length; ++i) {
+      tableIdsArray[i] = tableIds.get(i);
+    }
+    opBuilder.setAttr("table_ids", tableIdsArray);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.deviceOrdinal != null) {
+          opBuilder.setAttr("device_ordinal", opts.deviceOrdinal);
+        }
+        if (opts.combiners != null) {
+          String[] combinersArray = new String[opts.combiners.size()];
+          for (int i = 0; i < combinersArray.length; ++i) {
+            combinersArray[i] = opts.combiners.get(i);
+          }
+          opBuilder.setAttr("combiners", combinersArray);
+        }
+        if (opts.maxSequenceLengths != null) {
+          long[] maxSequenceLengthsArray = new long[opts.maxSequenceLengths.size()];
+          for (int i = 0; i < maxSequenceLengthsArray.length; ++i) {
+            maxSequenceLengthsArray[i] = opts.maxSequenceLengths.get(i);
+          }
+          opBuilder.setAttr("max_sequence_lengths", maxSequenceLengthsArray);
+        }
+      }
+    }
+    return new EnqueueTPUEmbeddingSparseTensorBatch(opBuilder.build());
+  }
+  
+  /**
+   * @param deviceOrdinal The TPU device to use. Should be >= 0 and less than the number
+   * of TPU cores in the task on which the node is placed.
+   */
+  public static Options deviceOrdinal(Long deviceOrdinal) {
+    return new Options().deviceOrdinal(deviceOrdinal);
+  }
+  
+  /**
+   * @param combiners A list of string scalars, one for each embedding table that specify
+   * how to normalize the embedding activations after weighted summation.
+   * Supported combiners are 'mean', 'sum', or 'sqrtn'. It is invalid to have
+   * the sum of the weights be 0 for 'mean' or the sum of the squared weights be
+   * 0 for 'sqrtn'. If combiners isn't passed, the default is to use 'sum' for
+   * all tables.
+   */
+  public static Options combiners(List<String> combiners) {
+    return new Options().combiners(combiners);
+  }
+  
+  /**
+   * @param maxSequenceLengths 
+   */
+  public static Options maxSequenceLengths(List<Long> maxSequenceLengths) {
+    return new Options().maxSequenceLengths(maxSequenceLengths);
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "EnqueueTPUEmbeddingSparseTensorBatch";
+  
+  private EnqueueTPUEmbeddingSparseTensorBatch(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/InfeedDequeue.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/InfeedDequeue.java
new file mode 100644
index 00000000000..1713f1f1a4d
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/InfeedDequeue.java
@@ -0,0 +1,77 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.tpu;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * A placeholder op for a value that will be fed into the computation.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+public final class InfeedDequeue<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new InfeedDequeue operation.
+   * 
+   * @param scope current scope
+   * @param dtype The type of elements in the tensor.
+   * @param shape The shape of the tensor.
+   * @return a new instance of InfeedDequeue
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> InfeedDequeue<T> create(Scope scope, Class<T> dtype, Shape shape) {
+    OperationBuilder opBuilder = scope.env().opBuilder("InfeedDequeue", scope.makeOpName("InfeedDequeue"));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("dtype", dtype);
+    opBuilder.setAttr("shape", shape);
+    return new InfeedDequeue<T>(opBuilder.build());
+  }
+  
+  /**
+   * A tensor that will be provided using the infeed mechanism.
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "InfeedDequeue";
+  
+  private Output<T> output;
+  
+  private InfeedDequeue(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/InfeedDequeueTuple.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/InfeedDequeueTuple.java
new file mode 100644
index 00000000000..b90665f21c5
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/InfeedDequeueTuple.java
@@ -0,0 +1,89 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.tpu;
+
+import java.util.Arrays;
+import java.util.Iterator;
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Fetches multiple values from infeed as an XLA tuple.
+ */
+public final class InfeedDequeueTuple extends RawOp implements Iterable<Operand<TType>> {
+  
+  /**
+   * Factory method to create a class wrapping a new InfeedDequeueTuple operation.
+   * 
+   * @param scope current scope
+   * @param dtypes The element types of each element in `outputs`.
+   * @param shapes The shapes of each tensor in `outputs`.
+   * @return a new instance of InfeedDequeueTuple
+   */
+  @Endpoint(describeByClass = true)
+  public static InfeedDequeueTuple create(Scope scope, List<Class<? extends TType>> dtypes, List<Shape> shapes) {
+    OperationBuilder opBuilder = scope.env().opBuilder("InfeedDequeueTuple", scope.makeOpName("InfeedDequeueTuple"));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    Class[] dtypesArray = new Class[dtypes.size()];
+    for (int i = 0; i < dtypesArray.length; ++i) {
+      dtypesArray[i] = dtypes.get(i);
+    }
+    opBuilder.setAttr("dtypes", dtypesArray);
+    Shape[] shapesArray = new Shape[shapes.size()];
+    for (int i = 0; i < shapesArray.length; ++i) {
+      shapesArray[i] = shapes.get(i);
+    }
+    opBuilder.setAttr("shapes", shapesArray);
+    return new InfeedDequeueTuple(opBuilder.build());
+  }
+  
+  /**
+   * A list of tensors that will be provided using the infeed mechanism.
+   */
+  public List<Output<?>> outputs() {
+    return outputs;
+  }
+  
+  @Override
+  @SuppressWarnings({"rawtypes", "unchecked"})
+  public Iterator<Operand<TType>> iterator() {
+    return (Iterator) outputs.iterator();
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "InfeedDequeueTuple";
+  
+  private List<Output<?>> outputs;
+  
+  private InfeedDequeueTuple(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    int outputsLength = operation.outputListLength("outputs");
+    outputs = Arrays.asList(operation.outputList(outputIdx, outputsLength));
+    outputIdx += outputsLength;
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/InfeedEnqueue.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/InfeedEnqueue.java
new file mode 100644
index 00000000000..391d51a9ab0
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/InfeedEnqueue.java
@@ -0,0 +1,141 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.tpu;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * An op which feeds a single Tensor value into the computation.
+ */
+public final class InfeedEnqueue extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.tpu.InfeedEnqueue}
+   */
+  public static class Options {
+    
+    /**
+     * @param shape The shape of the tensor.
+     */
+    public Options shape(Shape shape) {
+      this.shape = shape;
+      return this;
+    }
+    
+    /**
+     * @param layout A vector holding the requested layout in minor-to-major sequence.
+     * If a layout attribute is passed, but its values are all -1, the layout will
+     * be computed by the infeed operation.
+     */
+    public Options layout(List<Long> layout) {
+      this.layout = layout;
+      return this;
+    }
+    
+    /**
+     * @param deviceOrdinal The TPU device to use. This should be -1 when the Op
+     * is running on a TPU device, and >= 0 when the Op is running on the CPU
+     * device.
+     */
+    public Options deviceOrdinal(Long deviceOrdinal) {
+      this.deviceOrdinal = deviceOrdinal;
+      return this;
+    }
+    
+    private Shape shape;
+    private List<Long> layout;
+    private Long deviceOrdinal;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new InfeedEnqueue operation.
+   * 
+   * @param scope current scope
+   * @param input A tensor that will be provided using the infeed mechanism.
+   * @param options carries optional attributes values
+   * @return a new instance of InfeedEnqueue
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> InfeedEnqueue create(Scope scope, Operand<T> input, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("InfeedEnqueue", scope.makeOpName("InfeedEnqueue"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.shape != null) {
+          opBuilder.setAttr("shape", opts.shape);
+        }
+        if (opts.layout != null) {
+          long[] layoutArray = new long[opts.layout.size()];
+          for (int i = 0; i < layoutArray.length; ++i) {
+            layoutArray[i] = opts.layout.get(i);
+          }
+          opBuilder.setAttr("layout", layoutArray);
+        }
+        if (opts.deviceOrdinal != null) {
+          opBuilder.setAttr("device_ordinal", opts.deviceOrdinal);
+        }
+      }
+    }
+    return new InfeedEnqueue(opBuilder.build());
+  }
+  
+  /**
+   * @param shape The shape of the tensor.
+   */
+  public static Options shape(Shape shape) {
+    return new Options().shape(shape);
+  }
+  
+  /**
+   * @param layout A vector holding the requested layout in minor-to-major sequence.
+   * If a layout attribute is passed, but its values are all -1, the layout will
+   * be computed by the infeed operation.
+   */
+  public static Options layout(List<Long> layout) {
+    return new Options().layout(layout);
+  }
+  
+  /**
+   * @param deviceOrdinal The TPU device to use. This should be -1 when the Op
+   * is running on a TPU device, and >= 0 when the Op is running on the CPU
+   * device.
+   */
+  public static Options deviceOrdinal(Long deviceOrdinal) {
+    return new Options().deviceOrdinal(deviceOrdinal);
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "InfeedEnqueue";
+  
+  private InfeedEnqueue(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/InfeedEnqueuePrelinearizedBuffer.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/InfeedEnqueuePrelinearizedBuffer.java
new file mode 100644
index 00000000000..9344352791c
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/InfeedEnqueuePrelinearizedBuffer.java
@@ -0,0 +1,90 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.tpu;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+
+/**
+ * An op which enqueues prelinearized buffer into TPU infeed.
+ */
+public final class InfeedEnqueuePrelinearizedBuffer extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.tpu.InfeedEnqueuePrelinearizedBuffer}
+   */
+  public static class Options {
+    
+    /**
+     * @param deviceOrdinal The TPU device to use. This should be -1 when the Op is running on a TPU device
+     * and = 0 when the Op is running on the CPU device.
+     */
+    public Options deviceOrdinal(Long deviceOrdinal) {
+      this.deviceOrdinal = deviceOrdinal;
+      return this;
+    }
+    
+    private Long deviceOrdinal;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new InfeedEnqueuePrelinearizedBuffer operation.
+   * 
+   * @param scope current scope
+   * @param input A variant tensor representing linearized output.
+   * @param options carries optional attributes values
+   * @return a new instance of InfeedEnqueuePrelinearizedBuffer
+   */
+  @Endpoint(describeByClass = true)
+  public static InfeedEnqueuePrelinearizedBuffer create(Scope scope, Operand<?> input, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("InfeedEnqueuePrelinearizedBuffer", scope.makeOpName("InfeedEnqueuePrelinearizedBuffer"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.deviceOrdinal != null) {
+          opBuilder.setAttr("device_ordinal", opts.deviceOrdinal);
+        }
+      }
+    }
+    return new InfeedEnqueuePrelinearizedBuffer(opBuilder.build());
+  }
+  
+  /**
+   * @param deviceOrdinal The TPU device to use. This should be -1 when the Op is running on a TPU device
+   * and = 0 when the Op is running on the CPU device.
+   */
+  public static Options deviceOrdinal(Long deviceOrdinal) {
+    return new Options().deviceOrdinal(deviceOrdinal);
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "InfeedEnqueuePrelinearizedBuffer";
+  
+  private InfeedEnqueuePrelinearizedBuffer(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/InfeedEnqueueTuple.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/InfeedEnqueueTuple.java
new file mode 100644
index 00000000000..b439df84f71
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/InfeedEnqueueTuple.java
@@ -0,0 +1,130 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.tpu;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.Operands;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+
+/**
+ * Feeds multiple Tensor values into the computation as an XLA tuple.
+ */
+public final class InfeedEnqueueTuple extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.tpu.InfeedEnqueueTuple}
+   */
+  public static class Options {
+    
+    /**
+     * @param layouts A vector holding the requested layout in minor-to-major sequence for
+     * all the tuple shapes, in the order the shapes appear in the "shapes" input.
+     * The layout elements for a sub-shape can be set to -1, in which case the
+     * corresponding layout will be computed by the infeed operation.
+     */
+    public Options layouts(List<Long> layouts) {
+      this.layouts = layouts;
+      return this;
+    }
+    
+    /**
+     * @param deviceOrdinal The TPU device to use. This should be -1 when the Op
+     * is running on a TPU device, and >= 0 when the Op is running on the CPU
+     * device.
+     */
+    public Options deviceOrdinal(Long deviceOrdinal) {
+      this.deviceOrdinal = deviceOrdinal;
+      return this;
+    }
+    
+    private List<Long> layouts;
+    private Long deviceOrdinal;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new InfeedEnqueueTuple operation.
+   * 
+   * @param scope current scope
+   * @param inputs A list of tensors that will be provided using the infeed mechanism.
+   * @param shapes The shapes of each tensor in `inputs`.
+   * @param options carries optional attributes values
+   * @return a new instance of InfeedEnqueueTuple
+   */
+  @Endpoint(describeByClass = true)
+  public static InfeedEnqueueTuple create(Scope scope, Iterable<Operand<?>> inputs, List<Shape> shapes, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("InfeedEnqueueTuple", scope.makeOpName("InfeedEnqueueTuple"));
+    opBuilder.addInputList(Operands.asOutputs(inputs));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    Shape[] shapesArray = new Shape[shapes.size()];
+    for (int i = 0; i < shapesArray.length; ++i) {
+      shapesArray[i] = shapes.get(i);
+    }
+    opBuilder.setAttr("shapes", shapesArray);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.layouts != null) {
+          long[] layoutsArray = new long[opts.layouts.size()];
+          for (int i = 0; i < layoutsArray.length; ++i) {
+            layoutsArray[i] = opts.layouts.get(i);
+          }
+          opBuilder.setAttr("layouts", layoutsArray);
+        }
+        if (opts.deviceOrdinal != null) {
+          opBuilder.setAttr("device_ordinal", opts.deviceOrdinal);
+        }
+      }
+    }
+    return new InfeedEnqueueTuple(opBuilder.build());
+  }
+  
+  /**
+   * @param layouts A vector holding the requested layout in minor-to-major sequence for
+   * all the tuple shapes, in the order the shapes appear in the "shapes" input.
+   * The layout elements for a sub-shape can be set to -1, in which case the
+   * corresponding layout will be computed by the infeed operation.
+   */
+  public static Options layouts(List<Long> layouts) {
+    return new Options().layouts(layouts);
+  }
+  
+  /**
+   * @param deviceOrdinal The TPU device to use. This should be -1 when the Op
+   * is running on a TPU device, and >= 0 when the Op is running on the CPU
+   * device.
+   */
+  public static Options deviceOrdinal(Long deviceOrdinal) {
+    return new Options().deviceOrdinal(deviceOrdinal);
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "InfeedEnqueueTuple";
+  
+  private InfeedEnqueueTuple(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingADAMParameters.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingADAMParameters.java
new file mode 100644
index 00000000000..744688cee23
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingADAMParameters.java
@@ -0,0 +1,141 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.tpu;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+
+/**
+ * Load ADAM embedding parameters.
+ * <p>
+ * An op that loads optimization parameters into HBM for embedding. Must be
+ * preceded by a ConfigureTPUEmbeddingHost op that sets up the correct
+ * embedding table configuration. For example, this op is used to install
+ * parameters that are loaded from a checkpoint before a training loop is
+ * executed.
+ */
+public final class LoadTPUEmbeddingADAMParameters extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.tpu.LoadTPUEmbeddingADAMParameters}
+   */
+  public static class Options {
+    
+    /**
+     * @param tableId 
+     */
+    public Options tableId(Long tableId) {
+      this.tableId = tableId;
+      return this;
+    }
+    
+    /**
+     * @param tableName 
+     */
+    public Options tableName(String tableName) {
+      this.tableName = tableName;
+      return this;
+    }
+    
+    /**
+     * @param config 
+     */
+    public Options config(String config) {
+      this.config = config;
+      return this;
+    }
+    
+    private Long tableId;
+    private String tableName;
+    private String config;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new LoadTPUEmbeddingADAMParameters operation.
+   * 
+   * @param scope current scope
+   * @param parameters Value of parameters used in the ADAM optimization algorithm.
+   * @param momenta Value of momenta used in the ADAM optimization algorithm.
+   * @param velocities Value of velocities used in the ADAM optimization algorithm.
+   * @param numShards 
+   * @param shardId 
+   * @param options carries optional attributes values
+   * @return a new instance of LoadTPUEmbeddingADAMParameters
+   */
+  @Endpoint(describeByClass = true)
+  public static LoadTPUEmbeddingADAMParameters create(Scope scope, Operand<TFloat32> parameters, Operand<TFloat32> momenta, Operand<TFloat32> velocities, Long numShards, Long shardId, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("LoadTPUEmbeddingADAMParameters", scope.makeOpName("LoadTPUEmbeddingADAMParameters"));
+    opBuilder.addInput(parameters.asOutput());
+    opBuilder.addInput(momenta.asOutput());
+    opBuilder.addInput(velocities.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("num_shards", numShards);
+    opBuilder.setAttr("shard_id", shardId);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.tableId != null) {
+          opBuilder.setAttr("table_id", opts.tableId);
+        }
+        if (opts.tableName != null) {
+          opBuilder.setAttr("table_name", opts.tableName);
+        }
+        if (opts.config != null) {
+          opBuilder.setAttr("config", opts.config);
+        }
+      }
+    }
+    return new LoadTPUEmbeddingADAMParameters(opBuilder.build());
+  }
+  
+  /**
+   * @param tableId 
+   */
+  public static Options tableId(Long tableId) {
+    return new Options().tableId(tableId);
+  }
+  
+  /**
+   * @param tableName 
+   */
+  public static Options tableName(String tableName) {
+    return new Options().tableName(tableName);
+  }
+  
+  /**
+   * @param config 
+   */
+  public static Options config(String config) {
+    return new Options().config(config);
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "LoadTPUEmbeddingADAMParameters";
+  
+  private LoadTPUEmbeddingADAMParameters(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingADAMParametersGradAccumDebug.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingADAMParametersGradAccumDebug.java
new file mode 100644
index 00000000000..63df2e6aa79
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingADAMParametersGradAccumDebug.java
@@ -0,0 +1,143 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.tpu;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+
+/**
+ * Load ADAM embedding parameters with debug support.
+ * <p>
+ * An op that loads optimization parameters into HBM for embedding. Must be
+ * preceded by a ConfigureTPUEmbeddingHost op that sets up the correct
+ * embedding table configuration. For example, this op is used to install
+ * parameters that are loaded from a checkpoint before a training loop is
+ * executed.
+ */
+public final class LoadTPUEmbeddingADAMParametersGradAccumDebug extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.tpu.LoadTPUEmbeddingADAMParametersGradAccumDebug}
+   */
+  public static class Options {
+    
+    /**
+     * @param tableId 
+     */
+    public Options tableId(Long tableId) {
+      this.tableId = tableId;
+      return this;
+    }
+    
+    /**
+     * @param tableName 
+     */
+    public Options tableName(String tableName) {
+      this.tableName = tableName;
+      return this;
+    }
+    
+    /**
+     * @param config 
+     */
+    public Options config(String config) {
+      this.config = config;
+      return this;
+    }
+    
+    private Long tableId;
+    private String tableName;
+    private String config;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new LoadTPUEmbeddingADAMParametersGradAccumDebug operation.
+   * 
+   * @param scope current scope
+   * @param parameters Value of parameters used in the ADAM optimization algorithm.
+   * @param momenta Value of momenta used in the ADAM optimization algorithm.
+   * @param velocities Value of velocities used in the ADAM optimization algorithm.
+   * @param gradientAccumulators Value of gradient_accumulators used in the ADAM optimization algorithm.
+   * @param numShards 
+   * @param shardId 
+   * @param options carries optional attributes values
+   * @return a new instance of LoadTPUEmbeddingADAMParametersGradAccumDebug
+   */
+  @Endpoint(describeByClass = true)
+  public static LoadTPUEmbeddingADAMParametersGradAccumDebug create(Scope scope, Operand<TFloat32> parameters, Operand<TFloat32> momenta, Operand<TFloat32> velocities, Operand<TFloat32> gradientAccumulators, Long numShards, Long shardId, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("LoadTPUEmbeddingADAMParametersGradAccumDebug", scope.makeOpName("LoadTPUEmbeddingADAMParametersGradAccumDebug"));
+    opBuilder.addInput(parameters.asOutput());
+    opBuilder.addInput(momenta.asOutput());
+    opBuilder.addInput(velocities.asOutput());
+    opBuilder.addInput(gradientAccumulators.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("num_shards", numShards);
+    opBuilder.setAttr("shard_id", shardId);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.tableId != null) {
+          opBuilder.setAttr("table_id", opts.tableId);
+        }
+        if (opts.tableName != null) {
+          opBuilder.setAttr("table_name", opts.tableName);
+        }
+        if (opts.config != null) {
+          opBuilder.setAttr("config", opts.config);
+        }
+      }
+    }
+    return new LoadTPUEmbeddingADAMParametersGradAccumDebug(opBuilder.build());
+  }
+  
+  /**
+   * @param tableId 
+   */
+  public static Options tableId(Long tableId) {
+    return new Options().tableId(tableId);
+  }
+  
+  /**
+   * @param tableName 
+   */
+  public static Options tableName(String tableName) {
+    return new Options().tableName(tableName);
+  }
+  
+  /**
+   * @param config 
+   */
+  public static Options config(String config) {
+    return new Options().config(config);
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "LoadTPUEmbeddingADAMParametersGradAccumDebug";
+  
+  private LoadTPUEmbeddingADAMParametersGradAccumDebug(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingAdadeltaParameters.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingAdadeltaParameters.java
new file mode 100644
index 00000000000..43535a2aff8
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingAdadeltaParameters.java
@@ -0,0 +1,141 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.tpu;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+
+/**
+ * Load Adadelta embedding parameters.
+ * <p>
+ * An op that loads optimization parameters into HBM for embedding. Must be
+ * preceded by a ConfigureTPUEmbeddingHost op that sets up the correct
+ * embedding table configuration. For example, this op is used to install
+ * parameters that are loaded from a checkpoint before a training loop is
+ * executed.
+ */
+public final class LoadTPUEmbeddingAdadeltaParameters extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.tpu.LoadTPUEmbeddingAdadeltaParameters}
+   */
+  public static class Options {
+    
+    /**
+     * @param tableId 
+     */
+    public Options tableId(Long tableId) {
+      this.tableId = tableId;
+      return this;
+    }
+    
+    /**
+     * @param tableName 
+     */
+    public Options tableName(String tableName) {
+      this.tableName = tableName;
+      return this;
+    }
+    
+    /**
+     * @param config 
+     */
+    public Options config(String config) {
+      this.config = config;
+      return this;
+    }
+    
+    private Long tableId;
+    private String tableName;
+    private String config;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new LoadTPUEmbeddingAdadeltaParameters operation.
+   * 
+   * @param scope current scope
+   * @param parameters Value of parameters used in the Adadelta optimization algorithm.
+   * @param accumulators Value of accumulators used in the Adadelta optimization algorithm.
+   * @param updates Value of updates used in the Adadelta optimization algorithm.
+   * @param numShards 
+   * @param shardId 
+   * @param options carries optional attributes values
+   * @return a new instance of LoadTPUEmbeddingAdadeltaParameters
+   */
+  @Endpoint(describeByClass = true)
+  public static LoadTPUEmbeddingAdadeltaParameters create(Scope scope, Operand<TFloat32> parameters, Operand<TFloat32> accumulators, Operand<TFloat32> updates, Long numShards, Long shardId, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("LoadTPUEmbeddingAdadeltaParameters", scope.makeOpName("LoadTPUEmbeddingAdadeltaParameters"));
+    opBuilder.addInput(parameters.asOutput());
+    opBuilder.addInput(accumulators.asOutput());
+    opBuilder.addInput(updates.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("num_shards", numShards);
+    opBuilder.setAttr("shard_id", shardId);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.tableId != null) {
+          opBuilder.setAttr("table_id", opts.tableId);
+        }
+        if (opts.tableName != null) {
+          opBuilder.setAttr("table_name", opts.tableName);
+        }
+        if (opts.config != null) {
+          opBuilder.setAttr("config", opts.config);
+        }
+      }
+    }
+    return new LoadTPUEmbeddingAdadeltaParameters(opBuilder.build());
+  }
+  
+  /**
+   * @param tableId 
+   */
+  public static Options tableId(Long tableId) {
+    return new Options().tableId(tableId);
+  }
+  
+  /**
+   * @param tableName 
+   */
+  public static Options tableName(String tableName) {
+    return new Options().tableName(tableName);
+  }
+  
+  /**
+   * @param config 
+   */
+  public static Options config(String config) {
+    return new Options().config(config);
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "LoadTPUEmbeddingAdadeltaParameters";
+  
+  private LoadTPUEmbeddingAdadeltaParameters(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingAdadeltaParametersGradAccumDebug.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingAdadeltaParametersGradAccumDebug.java
new file mode 100644
index 00000000000..ce1b759ee60
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingAdadeltaParametersGradAccumDebug.java
@@ -0,0 +1,143 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.tpu;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+
+/**
+ * Load Adadelta parameters with debug support.
+ * <p>
+ * An op that loads optimization parameters into HBM for embedding. Must be
+ * preceded by a ConfigureTPUEmbeddingHost op that sets up the correct
+ * embedding table configuration. For example, this op is used to install
+ * parameters that are loaded from a checkpoint before a training loop is
+ * executed.
+ */
+public final class LoadTPUEmbeddingAdadeltaParametersGradAccumDebug extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.tpu.LoadTPUEmbeddingAdadeltaParametersGradAccumDebug}
+   */
+  public static class Options {
+    
+    /**
+     * @param tableId 
+     */
+    public Options tableId(Long tableId) {
+      this.tableId = tableId;
+      return this;
+    }
+    
+    /**
+     * @param tableName 
+     */
+    public Options tableName(String tableName) {
+      this.tableName = tableName;
+      return this;
+    }
+    
+    /**
+     * @param config 
+     */
+    public Options config(String config) {
+      this.config = config;
+      return this;
+    }
+    
+    private Long tableId;
+    private String tableName;
+    private String config;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new LoadTPUEmbeddingAdadeltaParametersGradAccumDebug operation.
+   * 
+   * @param scope current scope
+   * @param parameters Value of parameters used in the Adadelta optimization algorithm.
+   * @param accumulators Value of accumulators used in the Adadelta optimization algorithm.
+   * @param updates Value of updates used in the Adadelta optimization algorithm.
+   * @param gradientAccumulators Value of gradient_accumulators used in the Adadelta optimization algorithm.
+   * @param numShards 
+   * @param shardId 
+   * @param options carries optional attributes values
+   * @return a new instance of LoadTPUEmbeddingAdadeltaParametersGradAccumDebug
+   */
+  @Endpoint(describeByClass = true)
+  public static LoadTPUEmbeddingAdadeltaParametersGradAccumDebug create(Scope scope, Operand<TFloat32> parameters, Operand<TFloat32> accumulators, Operand<TFloat32> updates, Operand<TFloat32> gradientAccumulators, Long numShards, Long shardId, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("LoadTPUEmbeddingAdadeltaParametersGradAccumDebug", scope.makeOpName("LoadTPUEmbeddingAdadeltaParametersGradAccumDebug"));
+    opBuilder.addInput(parameters.asOutput());
+    opBuilder.addInput(accumulators.asOutput());
+    opBuilder.addInput(updates.asOutput());
+    opBuilder.addInput(gradientAccumulators.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("num_shards", numShards);
+    opBuilder.setAttr("shard_id", shardId);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.tableId != null) {
+          opBuilder.setAttr("table_id", opts.tableId);
+        }
+        if (opts.tableName != null) {
+          opBuilder.setAttr("table_name", opts.tableName);
+        }
+        if (opts.config != null) {
+          opBuilder.setAttr("config", opts.config);
+        }
+      }
+    }
+    return new LoadTPUEmbeddingAdadeltaParametersGradAccumDebug(opBuilder.build());
+  }
+  
+  /**
+   * @param tableId 
+   */
+  public static Options tableId(Long tableId) {
+    return new Options().tableId(tableId);
+  }
+  
+  /**
+   * @param tableName 
+   */
+  public static Options tableName(String tableName) {
+    return new Options().tableName(tableName);
+  }
+  
+  /**
+   * @param config 
+   */
+  public static Options config(String config) {
+    return new Options().config(config);
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "LoadTPUEmbeddingAdadeltaParametersGradAccumDebug";
+  
+  private LoadTPUEmbeddingAdadeltaParametersGradAccumDebug(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingAdagradParameters.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingAdagradParameters.java
new file mode 100644
index 00000000000..f9e16c5b5d6
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingAdagradParameters.java
@@ -0,0 +1,139 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.tpu;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+
+/**
+ * Load Adagrad embedding parameters.
+ * <p>
+ * An op that loads optimization parameters into HBM for embedding. Must be
+ * preceded by a ConfigureTPUEmbeddingHost op that sets up the correct
+ * embedding table configuration. For example, this op is used to install
+ * parameters that are loaded from a checkpoint before a training loop is
+ * executed.
+ */
+public final class LoadTPUEmbeddingAdagradParameters extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.tpu.LoadTPUEmbeddingAdagradParameters}
+   */
+  public static class Options {
+    
+    /**
+     * @param tableId 
+     */
+    public Options tableId(Long tableId) {
+      this.tableId = tableId;
+      return this;
+    }
+    
+    /**
+     * @param tableName 
+     */
+    public Options tableName(String tableName) {
+      this.tableName = tableName;
+      return this;
+    }
+    
+    /**
+     * @param config 
+     */
+    public Options config(String config) {
+      this.config = config;
+      return this;
+    }
+    
+    private Long tableId;
+    private String tableName;
+    private String config;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new LoadTPUEmbeddingAdagradParameters operation.
+   * 
+   * @param scope current scope
+   * @param parameters Value of parameters used in the Adagrad optimization algorithm.
+   * @param accumulators Value of accumulators used in the Adagrad optimization algorithm.
+   * @param numShards 
+   * @param shardId 
+   * @param options carries optional attributes values
+   * @return a new instance of LoadTPUEmbeddingAdagradParameters
+   */
+  @Endpoint(describeByClass = true)
+  public static LoadTPUEmbeddingAdagradParameters create(Scope scope, Operand<TFloat32> parameters, Operand<TFloat32> accumulators, Long numShards, Long shardId, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("LoadTPUEmbeddingAdagradParameters", scope.makeOpName("LoadTPUEmbeddingAdagradParameters"));
+    opBuilder.addInput(parameters.asOutput());
+    opBuilder.addInput(accumulators.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("num_shards", numShards);
+    opBuilder.setAttr("shard_id", shardId);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.tableId != null) {
+          opBuilder.setAttr("table_id", opts.tableId);
+        }
+        if (opts.tableName != null) {
+          opBuilder.setAttr("table_name", opts.tableName);
+        }
+        if (opts.config != null) {
+          opBuilder.setAttr("config", opts.config);
+        }
+      }
+    }
+    return new LoadTPUEmbeddingAdagradParameters(opBuilder.build());
+  }
+  
+  /**
+   * @param tableId 
+   */
+  public static Options tableId(Long tableId) {
+    return new Options().tableId(tableId);
+  }
+  
+  /**
+   * @param tableName 
+   */
+  public static Options tableName(String tableName) {
+    return new Options().tableName(tableName);
+  }
+  
+  /**
+   * @param config 
+   */
+  public static Options config(String config) {
+    return new Options().config(config);
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "LoadTPUEmbeddingAdagradParameters";
+  
+  private LoadTPUEmbeddingAdagradParameters(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingAdagradParametersGradAccumDebug.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingAdagradParametersGradAccumDebug.java
new file mode 100644
index 00000000000..7f8df653745
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingAdagradParametersGradAccumDebug.java
@@ -0,0 +1,141 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.tpu;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+
+/**
+ * Load Adagrad embedding parameters with debug support.
+ * <p>
+ * An op that loads optimization parameters into HBM for embedding. Must be
+ * preceded by a ConfigureTPUEmbeddingHost op that sets up the correct
+ * embedding table configuration. For example, this op is used to install
+ * parameters that are loaded from a checkpoint before a training loop is
+ * executed.
+ */
+public final class LoadTPUEmbeddingAdagradParametersGradAccumDebug extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.tpu.LoadTPUEmbeddingAdagradParametersGradAccumDebug}
+   */
+  public static class Options {
+    
+    /**
+     * @param tableId 
+     */
+    public Options tableId(Long tableId) {
+      this.tableId = tableId;
+      return this;
+    }
+    
+    /**
+     * @param tableName 
+     */
+    public Options tableName(String tableName) {
+      this.tableName = tableName;
+      return this;
+    }
+    
+    /**
+     * @param config 
+     */
+    public Options config(String config) {
+      this.config = config;
+      return this;
+    }
+    
+    private Long tableId;
+    private String tableName;
+    private String config;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new LoadTPUEmbeddingAdagradParametersGradAccumDebug operation.
+   * 
+   * @param scope current scope
+   * @param parameters Value of parameters used in the Adagrad optimization algorithm.
+   * @param accumulators Value of accumulators used in the Adagrad optimization algorithm.
+   * @param gradientAccumulators Value of gradient_accumulators used in the Adagrad optimization algorithm.
+   * @param numShards 
+   * @param shardId 
+   * @param options carries optional attributes values
+   * @return a new instance of LoadTPUEmbeddingAdagradParametersGradAccumDebug
+   */
+  @Endpoint(describeByClass = true)
+  public static LoadTPUEmbeddingAdagradParametersGradAccumDebug create(Scope scope, Operand<TFloat32> parameters, Operand<TFloat32> accumulators, Operand<TFloat32> gradientAccumulators, Long numShards, Long shardId, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("LoadTPUEmbeddingAdagradParametersGradAccumDebug", scope.makeOpName("LoadTPUEmbeddingAdagradParametersGradAccumDebug"));
+    opBuilder.addInput(parameters.asOutput());
+    opBuilder.addInput(accumulators.asOutput());
+    opBuilder.addInput(gradientAccumulators.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("num_shards", numShards);
+    opBuilder.setAttr("shard_id", shardId);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.tableId != null) {
+          opBuilder.setAttr("table_id", opts.tableId);
+        }
+        if (opts.tableName != null) {
+          opBuilder.setAttr("table_name", opts.tableName);
+        }
+        if (opts.config != null) {
+          opBuilder.setAttr("config", opts.config);
+        }
+      }
+    }
+    return new LoadTPUEmbeddingAdagradParametersGradAccumDebug(opBuilder.build());
+  }
+  
+  /**
+   * @param tableId 
+   */
+  public static Options tableId(Long tableId) {
+    return new Options().tableId(tableId);
+  }
+  
+  /**
+   * @param tableName 
+   */
+  public static Options tableName(String tableName) {
+    return new Options().tableName(tableName);
+  }
+  
+  /**
+   * @param config 
+   */
+  public static Options config(String config) {
+    return new Options().config(config);
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "LoadTPUEmbeddingAdagradParametersGradAccumDebug";
+  
+  private LoadTPUEmbeddingAdagradParametersGradAccumDebug(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingCenteredRMSPropParameters.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingCenteredRMSPropParameters.java
new file mode 100644
index 00000000000..f0b704cfaa1
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingCenteredRMSPropParameters.java
@@ -0,0 +1,143 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.tpu;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+
+/**
+ * Load centered RMSProp embedding parameters.
+ * <p>
+ * An op that loads optimization parameters into HBM for embedding. Must be
+ * preceded by a ConfigureTPUEmbeddingHost op that sets up the correct
+ * embedding table configuration. For example, this op is used to install
+ * parameters that are loaded from a checkpoint before a training loop is
+ * executed.
+ */
+public final class LoadTPUEmbeddingCenteredRMSPropParameters extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.tpu.LoadTPUEmbeddingCenteredRMSPropParameters}
+   */
+  public static class Options {
+    
+    /**
+     * @param tableId 
+     */
+    public Options tableId(Long tableId) {
+      this.tableId = tableId;
+      return this;
+    }
+    
+    /**
+     * @param tableName 
+     */
+    public Options tableName(String tableName) {
+      this.tableName = tableName;
+      return this;
+    }
+    
+    /**
+     * @param config 
+     */
+    public Options config(String config) {
+      this.config = config;
+      return this;
+    }
+    
+    private Long tableId;
+    private String tableName;
+    private String config;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new LoadTPUEmbeddingCenteredRMSPropParameters operation.
+   * 
+   * @param scope current scope
+   * @param parameters Value of parameters used in the centered RMSProp optimization algorithm.
+   * @param ms Value of ms used in the centered RMSProp optimization algorithm.
+   * @param mom Value of mom used in the centered RMSProp optimization algorithm.
+   * @param mg Value of mg used in the centered RMSProp optimization algorithm.
+   * @param numShards 
+   * @param shardId 
+   * @param options carries optional attributes values
+   * @return a new instance of LoadTPUEmbeddingCenteredRMSPropParameters
+   */
+  @Endpoint(describeByClass = true)
+  public static LoadTPUEmbeddingCenteredRMSPropParameters create(Scope scope, Operand<TFloat32> parameters, Operand<TFloat32> ms, Operand<TFloat32> mom, Operand<TFloat32> mg, Long numShards, Long shardId, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("LoadTPUEmbeddingCenteredRMSPropParameters", scope.makeOpName("LoadTPUEmbeddingCenteredRMSPropParameters"));
+    opBuilder.addInput(parameters.asOutput());
+    opBuilder.addInput(ms.asOutput());
+    opBuilder.addInput(mom.asOutput());
+    opBuilder.addInput(mg.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("num_shards", numShards);
+    opBuilder.setAttr("shard_id", shardId);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.tableId != null) {
+          opBuilder.setAttr("table_id", opts.tableId);
+        }
+        if (opts.tableName != null) {
+          opBuilder.setAttr("table_name", opts.tableName);
+        }
+        if (opts.config != null) {
+          opBuilder.setAttr("config", opts.config);
+        }
+      }
+    }
+    return new LoadTPUEmbeddingCenteredRMSPropParameters(opBuilder.build());
+  }
+  
+  /**
+   * @param tableId 
+   */
+  public static Options tableId(Long tableId) {
+    return new Options().tableId(tableId);
+  }
+  
+  /**
+   * @param tableName 
+   */
+  public static Options tableName(String tableName) {
+    return new Options().tableName(tableName);
+  }
+  
+  /**
+   * @param config 
+   */
+  public static Options config(String config) {
+    return new Options().config(config);
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "LoadTPUEmbeddingCenteredRMSPropParameters";
+  
+  private LoadTPUEmbeddingCenteredRMSPropParameters(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingFTRLParameters.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingFTRLParameters.java
new file mode 100644
index 00000000000..c96edf58894
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingFTRLParameters.java
@@ -0,0 +1,141 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.tpu;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+
+/**
+ * Load FTRL embedding parameters.
+ * <p>
+ * An op that loads optimization parameters into HBM for embedding. Must be
+ * preceded by a ConfigureTPUEmbeddingHost op that sets up the correct
+ * embedding table configuration. For example, this op is used to install
+ * parameters that are loaded from a checkpoint before a training loop is
+ * executed.
+ */
+public final class LoadTPUEmbeddingFTRLParameters extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.tpu.LoadTPUEmbeddingFTRLParameters}
+   */
+  public static class Options {
+    
+    /**
+     * @param tableId 
+     */
+    public Options tableId(Long tableId) {
+      this.tableId = tableId;
+      return this;
+    }
+    
+    /**
+     * @param tableName 
+     */
+    public Options tableName(String tableName) {
+      this.tableName = tableName;
+      return this;
+    }
+    
+    /**
+     * @param config 
+     */
+    public Options config(String config) {
+      this.config = config;
+      return this;
+    }
+    
+    private Long tableId;
+    private String tableName;
+    private String config;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new LoadTPUEmbeddingFTRLParameters operation.
+   * 
+   * @param scope current scope
+   * @param parameters Value of parameters used in the FTRL optimization algorithm.
+   * @param accumulators Value of accumulators used in the FTRL optimization algorithm.
+   * @param linears Value of linears used in the FTRL optimization algorithm.
+   * @param numShards 
+   * @param shardId 
+   * @param options carries optional attributes values
+   * @return a new instance of LoadTPUEmbeddingFTRLParameters
+   */
+  @Endpoint(describeByClass = true)
+  public static LoadTPUEmbeddingFTRLParameters create(Scope scope, Operand<TFloat32> parameters, Operand<TFloat32> accumulators, Operand<TFloat32> linears, Long numShards, Long shardId, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("LoadTPUEmbeddingFTRLParameters", scope.makeOpName("LoadTPUEmbeddingFTRLParameters"));
+    opBuilder.addInput(parameters.asOutput());
+    opBuilder.addInput(accumulators.asOutput());
+    opBuilder.addInput(linears.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("num_shards", numShards);
+    opBuilder.setAttr("shard_id", shardId);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.tableId != null) {
+          opBuilder.setAttr("table_id", opts.tableId);
+        }
+        if (opts.tableName != null) {
+          opBuilder.setAttr("table_name", opts.tableName);
+        }
+        if (opts.config != null) {
+          opBuilder.setAttr("config", opts.config);
+        }
+      }
+    }
+    return new LoadTPUEmbeddingFTRLParameters(opBuilder.build());
+  }
+  
+  /**
+   * @param tableId 
+   */
+  public static Options tableId(Long tableId) {
+    return new Options().tableId(tableId);
+  }
+  
+  /**
+   * @param tableName 
+   */
+  public static Options tableName(String tableName) {
+    return new Options().tableName(tableName);
+  }
+  
+  /**
+   * @param config 
+   */
+  public static Options config(String config) {
+    return new Options().config(config);
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "LoadTPUEmbeddingFTRLParameters";
+  
+  private LoadTPUEmbeddingFTRLParameters(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingFTRLParametersGradAccumDebug.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingFTRLParametersGradAccumDebug.java
new file mode 100644
index 00000000000..f0a85bd945a
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingFTRLParametersGradAccumDebug.java
@@ -0,0 +1,143 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.tpu;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+
+/**
+ * Load FTRL embedding parameters with debug support.
+ * <p>
+ * An op that loads optimization parameters into HBM for embedding. Must be
+ * preceded by a ConfigureTPUEmbeddingHost op that sets up the correct
+ * embedding table configuration. For example, this op is used to install
+ * parameters that are loaded from a checkpoint before a training loop is
+ * executed.
+ */
+public final class LoadTPUEmbeddingFTRLParametersGradAccumDebug extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.tpu.LoadTPUEmbeddingFTRLParametersGradAccumDebug}
+   */
+  public static class Options {
+    
+    /**
+     * @param tableId 
+     */
+    public Options tableId(Long tableId) {
+      this.tableId = tableId;
+      return this;
+    }
+    
+    /**
+     * @param tableName 
+     */
+    public Options tableName(String tableName) {
+      this.tableName = tableName;
+      return this;
+    }
+    
+    /**
+     * @param config 
+     */
+    public Options config(String config) {
+      this.config = config;
+      return this;
+    }
+    
+    private Long tableId;
+    private String tableName;
+    private String config;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new LoadTPUEmbeddingFTRLParametersGradAccumDebug operation.
+   * 
+   * @param scope current scope
+   * @param parameters Value of parameters used in the FTRL optimization algorithm.
+   * @param accumulators Value of accumulators used in the FTRL optimization algorithm.
+   * @param linears Value of linears used in the FTRL optimization algorithm.
+   * @param gradientAccumulators Value of gradient_accumulators used in the FTRL optimization algorithm.
+   * @param numShards 
+   * @param shardId 
+   * @param options carries optional attributes values
+   * @return a new instance of LoadTPUEmbeddingFTRLParametersGradAccumDebug
+   */
+  @Endpoint(describeByClass = true)
+  public static LoadTPUEmbeddingFTRLParametersGradAccumDebug create(Scope scope, Operand<TFloat32> parameters, Operand<TFloat32> accumulators, Operand<TFloat32> linears, Operand<TFloat32> gradientAccumulators, Long numShards, Long shardId, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("LoadTPUEmbeddingFTRLParametersGradAccumDebug", scope.makeOpName("LoadTPUEmbeddingFTRLParametersGradAccumDebug"));
+    opBuilder.addInput(parameters.asOutput());
+    opBuilder.addInput(accumulators.asOutput());
+    opBuilder.addInput(linears.asOutput());
+    opBuilder.addInput(gradientAccumulators.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("num_shards", numShards);
+    opBuilder.setAttr("shard_id", shardId);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.tableId != null) {
+          opBuilder.setAttr("table_id", opts.tableId);
+        }
+        if (opts.tableName != null) {
+          opBuilder.setAttr("table_name", opts.tableName);
+        }
+        if (opts.config != null) {
+          opBuilder.setAttr("config", opts.config);
+        }
+      }
+    }
+    return new LoadTPUEmbeddingFTRLParametersGradAccumDebug(opBuilder.build());
+  }
+  
+  /**
+   * @param tableId 
+   */
+  public static Options tableId(Long tableId) {
+    return new Options().tableId(tableId);
+  }
+  
+  /**
+   * @param tableName 
+   */
+  public static Options tableName(String tableName) {
+    return new Options().tableName(tableName);
+  }
+  
+  /**
+   * @param config 
+   */
+  public static Options config(String config) {
+    return new Options().config(config);
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "LoadTPUEmbeddingFTRLParametersGradAccumDebug";
+  
+  private LoadTPUEmbeddingFTRLParametersGradAccumDebug(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingMDLAdagradLightParameters.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingMDLAdagradLightParameters.java
new file mode 100644
index 00000000000..f418a70cc8c
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingMDLAdagradLightParameters.java
@@ -0,0 +1,143 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.tpu;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+
+/**
+ * Load MDL Adagrad Light embedding parameters.
+ * <p>
+ * An op that loads optimization parameters into HBM for embedding. Must be
+ * preceded by a ConfigureTPUEmbeddingHost op that sets up the correct
+ * embedding table configuration. For example, this op is used to install
+ * parameters that are loaded from a checkpoint before a training loop is
+ * executed.
+ */
+public final class LoadTPUEmbeddingMDLAdagradLightParameters extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.tpu.LoadTPUEmbeddingMDLAdagradLightParameters}
+   */
+  public static class Options {
+    
+    /**
+     * @param tableId 
+     */
+    public Options tableId(Long tableId) {
+      this.tableId = tableId;
+      return this;
+    }
+    
+    /**
+     * @param tableName 
+     */
+    public Options tableName(String tableName) {
+      this.tableName = tableName;
+      return this;
+    }
+    
+    /**
+     * @param config 
+     */
+    public Options config(String config) {
+      this.config = config;
+      return this;
+    }
+    
+    private Long tableId;
+    private String tableName;
+    private String config;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new LoadTPUEmbeddingMDLAdagradLightParameters operation.
+   * 
+   * @param scope current scope
+   * @param parameters Value of parameters used in the MDL Adagrad Light optimization algorithm.
+   * @param accumulators Value of accumulators used in the MDL Adagrad Light optimization algorithm.
+   * @param weights Value of weights used in the MDL Adagrad Light optimization algorithm.
+   * @param benefits Value of benefits used in the MDL Adagrad Light optimization algorithm.
+   * @param numShards 
+   * @param shardId 
+   * @param options carries optional attributes values
+   * @return a new instance of LoadTPUEmbeddingMDLAdagradLightParameters
+   */
+  @Endpoint(describeByClass = true)
+  public static LoadTPUEmbeddingMDLAdagradLightParameters create(Scope scope, Operand<TFloat32> parameters, Operand<TFloat32> accumulators, Operand<TFloat32> weights, Operand<TFloat32> benefits, Long numShards, Long shardId, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("LoadTPUEmbeddingMDLAdagradLightParameters", scope.makeOpName("LoadTPUEmbeddingMDLAdagradLightParameters"));
+    opBuilder.addInput(parameters.asOutput());
+    opBuilder.addInput(accumulators.asOutput());
+    opBuilder.addInput(weights.asOutput());
+    opBuilder.addInput(benefits.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("num_shards", numShards);
+    opBuilder.setAttr("shard_id", shardId);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.tableId != null) {
+          opBuilder.setAttr("table_id", opts.tableId);
+        }
+        if (opts.tableName != null) {
+          opBuilder.setAttr("table_name", opts.tableName);
+        }
+        if (opts.config != null) {
+          opBuilder.setAttr("config", opts.config);
+        }
+      }
+    }
+    return new LoadTPUEmbeddingMDLAdagradLightParameters(opBuilder.build());
+  }
+  
+  /**
+   * @param tableId 
+   */
+  public static Options tableId(Long tableId) {
+    return new Options().tableId(tableId);
+  }
+  
+  /**
+   * @param tableName 
+   */
+  public static Options tableName(String tableName) {
+    return new Options().tableName(tableName);
+  }
+  
+  /**
+   * @param config 
+   */
+  public static Options config(String config) {
+    return new Options().config(config);
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "LoadTPUEmbeddingMDLAdagradLightParameters";
+  
+  private LoadTPUEmbeddingMDLAdagradLightParameters(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingMomentumParameters.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingMomentumParameters.java
new file mode 100644
index 00000000000..718bdc24f5c
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingMomentumParameters.java
@@ -0,0 +1,139 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.tpu;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+
+/**
+ * Load Momentum embedding parameters.
+ * <p>
+ * An op that loads optimization parameters into HBM for embedding. Must be
+ * preceded by a ConfigureTPUEmbeddingHost op that sets up the correct
+ * embedding table configuration. For example, this op is used to install
+ * parameters that are loaded from a checkpoint before a training loop is
+ * executed.
+ */
+public final class LoadTPUEmbeddingMomentumParameters extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.tpu.LoadTPUEmbeddingMomentumParameters}
+   */
+  public static class Options {
+    
+    /**
+     * @param tableId 
+     */
+    public Options tableId(Long tableId) {
+      this.tableId = tableId;
+      return this;
+    }
+    
+    /**
+     * @param tableName 
+     */
+    public Options tableName(String tableName) {
+      this.tableName = tableName;
+      return this;
+    }
+    
+    /**
+     * @param config 
+     */
+    public Options config(String config) {
+      this.config = config;
+      return this;
+    }
+    
+    private Long tableId;
+    private String tableName;
+    private String config;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new LoadTPUEmbeddingMomentumParameters operation.
+   * 
+   * @param scope current scope
+   * @param parameters Value of parameters used in the Momentum optimization algorithm.
+   * @param momenta Value of momenta used in the Momentum optimization algorithm.
+   * @param numShards 
+   * @param shardId 
+   * @param options carries optional attributes values
+   * @return a new instance of LoadTPUEmbeddingMomentumParameters
+   */
+  @Endpoint(describeByClass = true)
+  public static LoadTPUEmbeddingMomentumParameters create(Scope scope, Operand<TFloat32> parameters, Operand<TFloat32> momenta, Long numShards, Long shardId, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("LoadTPUEmbeddingMomentumParameters", scope.makeOpName("LoadTPUEmbeddingMomentumParameters"));
+    opBuilder.addInput(parameters.asOutput());
+    opBuilder.addInput(momenta.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("num_shards", numShards);
+    opBuilder.setAttr("shard_id", shardId);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.tableId != null) {
+          opBuilder.setAttr("table_id", opts.tableId);
+        }
+        if (opts.tableName != null) {
+          opBuilder.setAttr("table_name", opts.tableName);
+        }
+        if (opts.config != null) {
+          opBuilder.setAttr("config", opts.config);
+        }
+      }
+    }
+    return new LoadTPUEmbeddingMomentumParameters(opBuilder.build());
+  }
+  
+  /**
+   * @param tableId 
+   */
+  public static Options tableId(Long tableId) {
+    return new Options().tableId(tableId);
+  }
+  
+  /**
+   * @param tableName 
+   */
+  public static Options tableName(String tableName) {
+    return new Options().tableName(tableName);
+  }
+  
+  /**
+   * @param config 
+   */
+  public static Options config(String config) {
+    return new Options().config(config);
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "LoadTPUEmbeddingMomentumParameters";
+  
+  private LoadTPUEmbeddingMomentumParameters(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingMomentumParametersGradAccumDebug.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingMomentumParametersGradAccumDebug.java
new file mode 100644
index 00000000000..424c3c846c0
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingMomentumParametersGradAccumDebug.java
@@ -0,0 +1,141 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.tpu;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+
+/**
+ * Load Momentum embedding parameters with debug support.
+ * <p>
+ * An op that loads optimization parameters into HBM for embedding. Must be
+ * preceded by a ConfigureTPUEmbeddingHost op that sets up the correct
+ * embedding table configuration. For example, this op is used to install
+ * parameters that are loaded from a checkpoint before a training loop is
+ * executed.
+ */
+public final class LoadTPUEmbeddingMomentumParametersGradAccumDebug extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.tpu.LoadTPUEmbeddingMomentumParametersGradAccumDebug}
+   */
+  public static class Options {
+    
+    /**
+     * @param tableId 
+     */
+    public Options tableId(Long tableId) {
+      this.tableId = tableId;
+      return this;
+    }
+    
+    /**
+     * @param tableName 
+     */
+    public Options tableName(String tableName) {
+      this.tableName = tableName;
+      return this;
+    }
+    
+    /**
+     * @param config 
+     */
+    public Options config(String config) {
+      this.config = config;
+      return this;
+    }
+    
+    private Long tableId;
+    private String tableName;
+    private String config;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new LoadTPUEmbeddingMomentumParametersGradAccumDebug operation.
+   * 
+   * @param scope current scope
+   * @param parameters Value of parameters used in the Momentum optimization algorithm.
+   * @param momenta Value of momenta used in the Momentum optimization algorithm.
+   * @param gradientAccumulators Value of gradient_accumulators used in the Momentum optimization algorithm.
+   * @param numShards 
+   * @param shardId 
+   * @param options carries optional attributes values
+   * @return a new instance of LoadTPUEmbeddingMomentumParametersGradAccumDebug
+   */
+  @Endpoint(describeByClass = true)
+  public static LoadTPUEmbeddingMomentumParametersGradAccumDebug create(Scope scope, Operand<TFloat32> parameters, Operand<TFloat32> momenta, Operand<TFloat32> gradientAccumulators, Long numShards, Long shardId, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("LoadTPUEmbeddingMomentumParametersGradAccumDebug", scope.makeOpName("LoadTPUEmbeddingMomentumParametersGradAccumDebug"));
+    opBuilder.addInput(parameters.asOutput());
+    opBuilder.addInput(momenta.asOutput());
+    opBuilder.addInput(gradientAccumulators.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("num_shards", numShards);
+    opBuilder.setAttr("shard_id", shardId);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.tableId != null) {
+          opBuilder.setAttr("table_id", opts.tableId);
+        }
+        if (opts.tableName != null) {
+          opBuilder.setAttr("table_name", opts.tableName);
+        }
+        if (opts.config != null) {
+          opBuilder.setAttr("config", opts.config);
+        }
+      }
+    }
+    return new LoadTPUEmbeddingMomentumParametersGradAccumDebug(opBuilder.build());
+  }
+  
+  /**
+   * @param tableId 
+   */
+  public static Options tableId(Long tableId) {
+    return new Options().tableId(tableId);
+  }
+  
+  /**
+   * @param tableName 
+   */
+  public static Options tableName(String tableName) {
+    return new Options().tableName(tableName);
+  }
+  
+  /**
+   * @param config 
+   */
+  public static Options config(String config) {
+    return new Options().config(config);
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "LoadTPUEmbeddingMomentumParametersGradAccumDebug";
+  
+  private LoadTPUEmbeddingMomentumParametersGradAccumDebug(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingProximalAdagradParameters.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingProximalAdagradParameters.java
new file mode 100644
index 00000000000..7b7265e9b82
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingProximalAdagradParameters.java
@@ -0,0 +1,139 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.tpu;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+
+/**
+ * Load proximal Adagrad embedding parameters.
+ * <p>
+ * An op that loads optimization parameters into HBM for embedding. Must be
+ * preceded by a ConfigureTPUEmbeddingHost op that sets up the correct
+ * embedding table configuration. For example, this op is used to install
+ * parameters that are loaded from a checkpoint before a training loop is
+ * executed.
+ */
+public final class LoadTPUEmbeddingProximalAdagradParameters extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.tpu.LoadTPUEmbeddingProximalAdagradParameters}
+   */
+  public static class Options {
+    
+    /**
+     * @param tableId 
+     */
+    public Options tableId(Long tableId) {
+      this.tableId = tableId;
+      return this;
+    }
+    
+    /**
+     * @param tableName 
+     */
+    public Options tableName(String tableName) {
+      this.tableName = tableName;
+      return this;
+    }
+    
+    /**
+     * @param config 
+     */
+    public Options config(String config) {
+      this.config = config;
+      return this;
+    }
+    
+    private Long tableId;
+    private String tableName;
+    private String config;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new LoadTPUEmbeddingProximalAdagradParameters operation.
+   * 
+   * @param scope current scope
+   * @param parameters Value of parameters used in the proximal Adagrad optimization algorithm.
+   * @param accumulators Value of accumulators used in the proximal Adagrad optimization algorithm.
+   * @param numShards 
+   * @param shardId 
+   * @param options carries optional attributes values
+   * @return a new instance of LoadTPUEmbeddingProximalAdagradParameters
+   */
+  @Endpoint(describeByClass = true)
+  public static LoadTPUEmbeddingProximalAdagradParameters create(Scope scope, Operand<TFloat32> parameters, Operand<TFloat32> accumulators, Long numShards, Long shardId, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("LoadTPUEmbeddingProximalAdagradParameters", scope.makeOpName("LoadTPUEmbeddingProximalAdagradParameters"));
+    opBuilder.addInput(parameters.asOutput());
+    opBuilder.addInput(accumulators.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("num_shards", numShards);
+    opBuilder.setAttr("shard_id", shardId);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.tableId != null) {
+          opBuilder.setAttr("table_id", opts.tableId);
+        }
+        if (opts.tableName != null) {
+          opBuilder.setAttr("table_name", opts.tableName);
+        }
+        if (opts.config != null) {
+          opBuilder.setAttr("config", opts.config);
+        }
+      }
+    }
+    return new LoadTPUEmbeddingProximalAdagradParameters(opBuilder.build());
+  }
+  
+  /**
+   * @param tableId 
+   */
+  public static Options tableId(Long tableId) {
+    return new Options().tableId(tableId);
+  }
+  
+  /**
+   * @param tableName 
+   */
+  public static Options tableName(String tableName) {
+    return new Options().tableName(tableName);
+  }
+  
+  /**
+   * @param config 
+   */
+  public static Options config(String config) {
+    return new Options().config(config);
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "LoadTPUEmbeddingProximalAdagradParameters";
+  
+  private LoadTPUEmbeddingProximalAdagradParameters(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingProximalAdagradParametersGradAccumDebug.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingProximalAdagradParametersGradAccumDebug.java
new file mode 100644
index 00000000000..c18d2cf22f5
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingProximalAdagradParametersGradAccumDebug.java
@@ -0,0 +1,141 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.tpu;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+
+/**
+ * Load proximal Adagrad embedding parameters with debug support.
+ * <p>
+ * An op that loads optimization parameters into HBM for embedding. Must be
+ * preceded by a ConfigureTPUEmbeddingHost op that sets up the correct
+ * embedding table configuration. For example, this op is used to install
+ * parameters that are loaded from a checkpoint before a training loop is
+ * executed.
+ */
+public final class LoadTPUEmbeddingProximalAdagradParametersGradAccumDebug extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.tpu.LoadTPUEmbeddingProximalAdagradParametersGradAccumDebug}
+   */
+  public static class Options {
+    
+    /**
+     * @param tableId 
+     */
+    public Options tableId(Long tableId) {
+      this.tableId = tableId;
+      return this;
+    }
+    
+    /**
+     * @param tableName 
+     */
+    public Options tableName(String tableName) {
+      this.tableName = tableName;
+      return this;
+    }
+    
+    /**
+     * @param config 
+     */
+    public Options config(String config) {
+      this.config = config;
+      return this;
+    }
+    
+    private Long tableId;
+    private String tableName;
+    private String config;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new LoadTPUEmbeddingProximalAdagradParametersGradAccumDebug operation.
+   * 
+   * @param scope current scope
+   * @param parameters Value of parameters used in the proximal Adagrad optimization algorithm.
+   * @param accumulators Value of accumulators used in the proximal Adagrad optimization algorithm.
+   * @param gradientAccumulators Value of gradient_accumulators used in the proximal Adagrad optimization algorithm.
+   * @param numShards 
+   * @param shardId 
+   * @param options carries optional attributes values
+   * @return a new instance of LoadTPUEmbeddingProximalAdagradParametersGradAccumDebug
+   */
+  @Endpoint(describeByClass = true)
+  public static LoadTPUEmbeddingProximalAdagradParametersGradAccumDebug create(Scope scope, Operand<TFloat32> parameters, Operand<TFloat32> accumulators, Operand<TFloat32> gradientAccumulators, Long numShards, Long shardId, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("LoadTPUEmbeddingProximalAdagradParametersGradAccumDebug", scope.makeOpName("LoadTPUEmbeddingProximalAdagradParametersGradAccumDebug"));
+    opBuilder.addInput(parameters.asOutput());
+    opBuilder.addInput(accumulators.asOutput());
+    opBuilder.addInput(gradientAccumulators.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("num_shards", numShards);
+    opBuilder.setAttr("shard_id", shardId);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.tableId != null) {
+          opBuilder.setAttr("table_id", opts.tableId);
+        }
+        if (opts.tableName != null) {
+          opBuilder.setAttr("table_name", opts.tableName);
+        }
+        if (opts.config != null) {
+          opBuilder.setAttr("config", opts.config);
+        }
+      }
+    }
+    return new LoadTPUEmbeddingProximalAdagradParametersGradAccumDebug(opBuilder.build());
+  }
+  
+  /**
+   * @param tableId 
+   */
+  public static Options tableId(Long tableId) {
+    return new Options().tableId(tableId);
+  }
+  
+  /**
+   * @param tableName 
+   */
+  public static Options tableName(String tableName) {
+    return new Options().tableName(tableName);
+  }
+  
+  /**
+   * @param config 
+   */
+  public static Options config(String config) {
+    return new Options().config(config);
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "LoadTPUEmbeddingProximalAdagradParametersGradAccumDebug";
+  
+  private LoadTPUEmbeddingProximalAdagradParametersGradAccumDebug(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingProximalYogiParameters.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingProximalYogiParameters.java
new file mode 100644
index 00000000000..2a96916c4f5
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingProximalYogiParameters.java
@@ -0,0 +1,134 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.tpu;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+
+/**
+ */
+public final class LoadTPUEmbeddingProximalYogiParameters extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.tpu.LoadTPUEmbeddingProximalYogiParameters}
+   */
+  public static class Options {
+    
+    /**
+     * @param tableId 
+     */
+    public Options tableId(Long tableId) {
+      this.tableId = tableId;
+      return this;
+    }
+    
+    /**
+     * @param tableName 
+     */
+    public Options tableName(String tableName) {
+      this.tableName = tableName;
+      return this;
+    }
+    
+    /**
+     * @param config 
+     */
+    public Options config(String config) {
+      this.config = config;
+      return this;
+    }
+    
+    private Long tableId;
+    private String tableName;
+    private String config;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new LoadTPUEmbeddingProximalYogiParameters operation.
+   * 
+   * @param scope current scope
+   * @param parameters 
+   * @param v 
+   * @param m 
+   * @param numShards 
+   * @param shardId 
+   * @param options carries optional attributes values
+   * @return a new instance of LoadTPUEmbeddingProximalYogiParameters
+   */
+  @Endpoint(describeByClass = true)
+  public static LoadTPUEmbeddingProximalYogiParameters create(Scope scope, Operand<TFloat32> parameters, Operand<TFloat32> v, Operand<TFloat32> m, Long numShards, Long shardId, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("LoadTPUEmbeddingProximalYogiParameters", scope.makeOpName("LoadTPUEmbeddingProximalYogiParameters"));
+    opBuilder.addInput(parameters.asOutput());
+    opBuilder.addInput(v.asOutput());
+    opBuilder.addInput(m.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("num_shards", numShards);
+    opBuilder.setAttr("shard_id", shardId);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.tableId != null) {
+          opBuilder.setAttr("table_id", opts.tableId);
+        }
+        if (opts.tableName != null) {
+          opBuilder.setAttr("table_name", opts.tableName);
+        }
+        if (opts.config != null) {
+          opBuilder.setAttr("config", opts.config);
+        }
+      }
+    }
+    return new LoadTPUEmbeddingProximalYogiParameters(opBuilder.build());
+  }
+  
+  /**
+   * @param tableId 
+   */
+  public static Options tableId(Long tableId) {
+    return new Options().tableId(tableId);
+  }
+  
+  /**
+   * @param tableName 
+   */
+  public static Options tableName(String tableName) {
+    return new Options().tableName(tableName);
+  }
+  
+  /**
+   * @param config 
+   */
+  public static Options config(String config) {
+    return new Options().config(config);
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "LoadTPUEmbeddingProximalYogiParameters";
+  
+  private LoadTPUEmbeddingProximalYogiParameters(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingProximalYogiParametersGradAccumDebug.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingProximalYogiParametersGradAccumDebug.java
new file mode 100644
index 00000000000..e863dc554d6
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingProximalYogiParametersGradAccumDebug.java
@@ -0,0 +1,136 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.tpu;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+
+/**
+ */
+public final class LoadTPUEmbeddingProximalYogiParametersGradAccumDebug extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.tpu.LoadTPUEmbeddingProximalYogiParametersGradAccumDebug}
+   */
+  public static class Options {
+    
+    /**
+     * @param tableId 
+     */
+    public Options tableId(Long tableId) {
+      this.tableId = tableId;
+      return this;
+    }
+    
+    /**
+     * @param tableName 
+     */
+    public Options tableName(String tableName) {
+      this.tableName = tableName;
+      return this;
+    }
+    
+    /**
+     * @param config 
+     */
+    public Options config(String config) {
+      this.config = config;
+      return this;
+    }
+    
+    private Long tableId;
+    private String tableName;
+    private String config;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new LoadTPUEmbeddingProximalYogiParametersGradAccumDebug operation.
+   * 
+   * @param scope current scope
+   * @param parameters 
+   * @param v 
+   * @param m 
+   * @param gradientAccumulators 
+   * @param numShards 
+   * @param shardId 
+   * @param options carries optional attributes values
+   * @return a new instance of LoadTPUEmbeddingProximalYogiParametersGradAccumDebug
+   */
+  @Endpoint(describeByClass = true)
+  public static LoadTPUEmbeddingProximalYogiParametersGradAccumDebug create(Scope scope, Operand<TFloat32> parameters, Operand<TFloat32> v, Operand<TFloat32> m, Operand<TFloat32> gradientAccumulators, Long numShards, Long shardId, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("LoadTPUEmbeddingProximalYogiParametersGradAccumDebug", scope.makeOpName("LoadTPUEmbeddingProximalYogiParametersGradAccumDebug"));
+    opBuilder.addInput(parameters.asOutput());
+    opBuilder.addInput(v.asOutput());
+    opBuilder.addInput(m.asOutput());
+    opBuilder.addInput(gradientAccumulators.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("num_shards", numShards);
+    opBuilder.setAttr("shard_id", shardId);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.tableId != null) {
+          opBuilder.setAttr("table_id", opts.tableId);
+        }
+        if (opts.tableName != null) {
+          opBuilder.setAttr("table_name", opts.tableName);
+        }
+        if (opts.config != null) {
+          opBuilder.setAttr("config", opts.config);
+        }
+      }
+    }
+    return new LoadTPUEmbeddingProximalYogiParametersGradAccumDebug(opBuilder.build());
+  }
+  
+  /**
+   * @param tableId 
+   */
+  public static Options tableId(Long tableId) {
+    return new Options().tableId(tableId);
+  }
+  
+  /**
+   * @param tableName 
+   */
+  public static Options tableName(String tableName) {
+    return new Options().tableName(tableName);
+  }
+  
+  /**
+   * @param config 
+   */
+  public static Options config(String config) {
+    return new Options().config(config);
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "LoadTPUEmbeddingProximalYogiParametersGradAccumDebug";
+  
+  private LoadTPUEmbeddingProximalYogiParametersGradAccumDebug(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingRMSPropParameters.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingRMSPropParameters.java
new file mode 100644
index 00000000000..1f747282d55
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingRMSPropParameters.java
@@ -0,0 +1,141 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.tpu;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+
+/**
+ * Load RMSProp embedding parameters.
+ * <p>
+ * An op that loads optimization parameters into HBM for embedding. Must be
+ * preceded by a ConfigureTPUEmbeddingHost op that sets up the correct
+ * embedding table configuration. For example, this op is used to install
+ * parameters that are loaded from a checkpoint before a training loop is
+ * executed.
+ */
+public final class LoadTPUEmbeddingRMSPropParameters extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.tpu.LoadTPUEmbeddingRMSPropParameters}
+   */
+  public static class Options {
+    
+    /**
+     * @param tableId 
+     */
+    public Options tableId(Long tableId) {
+      this.tableId = tableId;
+      return this;
+    }
+    
+    /**
+     * @param tableName 
+     */
+    public Options tableName(String tableName) {
+      this.tableName = tableName;
+      return this;
+    }
+    
+    /**
+     * @param config 
+     */
+    public Options config(String config) {
+      this.config = config;
+      return this;
+    }
+    
+    private Long tableId;
+    private String tableName;
+    private String config;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new LoadTPUEmbeddingRMSPropParameters operation.
+   * 
+   * @param scope current scope
+   * @param parameters Value of parameters used in the RMSProp optimization algorithm.
+   * @param ms Value of ms used in the RMSProp optimization algorithm.
+   * @param mom Value of mom used in the RMSProp optimization algorithm.
+   * @param numShards 
+   * @param shardId 
+   * @param options carries optional attributes values
+   * @return a new instance of LoadTPUEmbeddingRMSPropParameters
+   */
+  @Endpoint(describeByClass = true)
+  public static LoadTPUEmbeddingRMSPropParameters create(Scope scope, Operand<TFloat32> parameters, Operand<TFloat32> ms, Operand<TFloat32> mom, Long numShards, Long shardId, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("LoadTPUEmbeddingRMSPropParameters", scope.makeOpName("LoadTPUEmbeddingRMSPropParameters"));
+    opBuilder.addInput(parameters.asOutput());
+    opBuilder.addInput(ms.asOutput());
+    opBuilder.addInput(mom.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("num_shards", numShards);
+    opBuilder.setAttr("shard_id", shardId);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.tableId != null) {
+          opBuilder.setAttr("table_id", opts.tableId);
+        }
+        if (opts.tableName != null) {
+          opBuilder.setAttr("table_name", opts.tableName);
+        }
+        if (opts.config != null) {
+          opBuilder.setAttr("config", opts.config);
+        }
+      }
+    }
+    return new LoadTPUEmbeddingRMSPropParameters(opBuilder.build());
+  }
+  
+  /**
+   * @param tableId 
+   */
+  public static Options tableId(Long tableId) {
+    return new Options().tableId(tableId);
+  }
+  
+  /**
+   * @param tableName 
+   */
+  public static Options tableName(String tableName) {
+    return new Options().tableName(tableName);
+  }
+  
+  /**
+   * @param config 
+   */
+  public static Options config(String config) {
+    return new Options().config(config);
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "LoadTPUEmbeddingRMSPropParameters";
+  
+  private LoadTPUEmbeddingRMSPropParameters(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingRMSPropParametersGradAccumDebug.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingRMSPropParametersGradAccumDebug.java
new file mode 100644
index 00000000000..a7c8ed4812c
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingRMSPropParametersGradAccumDebug.java
@@ -0,0 +1,143 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.tpu;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+
+/**
+ * Load RMSProp embedding parameters with debug support.
+ * <p>
+ * An op that loads optimization parameters into HBM for embedding. Must be
+ * preceded by a ConfigureTPUEmbeddingHost op that sets up the correct
+ * embedding table configuration. For example, this op is used to install
+ * parameters that are loaded from a checkpoint before a training loop is
+ * executed.
+ */
+public final class LoadTPUEmbeddingRMSPropParametersGradAccumDebug extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.tpu.LoadTPUEmbeddingRMSPropParametersGradAccumDebug}
+   */
+  public static class Options {
+    
+    /**
+     * @param tableId 
+     */
+    public Options tableId(Long tableId) {
+      this.tableId = tableId;
+      return this;
+    }
+    
+    /**
+     * @param tableName 
+     */
+    public Options tableName(String tableName) {
+      this.tableName = tableName;
+      return this;
+    }
+    
+    /**
+     * @param config 
+     */
+    public Options config(String config) {
+      this.config = config;
+      return this;
+    }
+    
+    private Long tableId;
+    private String tableName;
+    private String config;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new LoadTPUEmbeddingRMSPropParametersGradAccumDebug operation.
+   * 
+   * @param scope current scope
+   * @param parameters Value of parameters used in the RMSProp optimization algorithm.
+   * @param ms Value of ms used in the RMSProp optimization algorithm.
+   * @param mom Value of mom used in the RMSProp optimization algorithm.
+   * @param gradientAccumulators Value of gradient_accumulators used in the RMSProp optimization algorithm.
+   * @param numShards 
+   * @param shardId 
+   * @param options carries optional attributes values
+   * @return a new instance of LoadTPUEmbeddingRMSPropParametersGradAccumDebug
+   */
+  @Endpoint(describeByClass = true)
+  public static LoadTPUEmbeddingRMSPropParametersGradAccumDebug create(Scope scope, Operand<TFloat32> parameters, Operand<TFloat32> ms, Operand<TFloat32> mom, Operand<TFloat32> gradientAccumulators, Long numShards, Long shardId, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("LoadTPUEmbeddingRMSPropParametersGradAccumDebug", scope.makeOpName("LoadTPUEmbeddingRMSPropParametersGradAccumDebug"));
+    opBuilder.addInput(parameters.asOutput());
+    opBuilder.addInput(ms.asOutput());
+    opBuilder.addInput(mom.asOutput());
+    opBuilder.addInput(gradientAccumulators.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("num_shards", numShards);
+    opBuilder.setAttr("shard_id", shardId);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.tableId != null) {
+          opBuilder.setAttr("table_id", opts.tableId);
+        }
+        if (opts.tableName != null) {
+          opBuilder.setAttr("table_name", opts.tableName);
+        }
+        if (opts.config != null) {
+          opBuilder.setAttr("config", opts.config);
+        }
+      }
+    }
+    return new LoadTPUEmbeddingRMSPropParametersGradAccumDebug(opBuilder.build());
+  }
+  
+  /**
+   * @param tableId 
+   */
+  public static Options tableId(Long tableId) {
+    return new Options().tableId(tableId);
+  }
+  
+  /**
+   * @param tableName 
+   */
+  public static Options tableName(String tableName) {
+    return new Options().tableName(tableName);
+  }
+  
+  /**
+   * @param config 
+   */
+  public static Options config(String config) {
+    return new Options().config(config);
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "LoadTPUEmbeddingRMSPropParametersGradAccumDebug";
+  
+  private LoadTPUEmbeddingRMSPropParametersGradAccumDebug(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingStochasticGradientDescentParameters.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingStochasticGradientDescentParameters.java
new file mode 100644
index 00000000000..769d3436eda
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingStochasticGradientDescentParameters.java
@@ -0,0 +1,137 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.tpu;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+
+/**
+ * Load SGD embedding parameters.
+ * <p>
+ * An op that loads optimization parameters into HBM for embedding. Must be
+ * preceded by a ConfigureTPUEmbeddingHost op that sets up the correct
+ * embedding table configuration. For example, this op is used to install
+ * parameters that are loaded from a checkpoint before a training loop is
+ * executed.
+ */
+public final class LoadTPUEmbeddingStochasticGradientDescentParameters extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.tpu.LoadTPUEmbeddingStochasticGradientDescentParameters}
+   */
+  public static class Options {
+    
+    /**
+     * @param tableId 
+     */
+    public Options tableId(Long tableId) {
+      this.tableId = tableId;
+      return this;
+    }
+    
+    /**
+     * @param tableName 
+     */
+    public Options tableName(String tableName) {
+      this.tableName = tableName;
+      return this;
+    }
+    
+    /**
+     * @param config 
+     */
+    public Options config(String config) {
+      this.config = config;
+      return this;
+    }
+    
+    private Long tableId;
+    private String tableName;
+    private String config;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new LoadTPUEmbeddingStochasticGradientDescentParameters operation.
+   * 
+   * @param scope current scope
+   * @param parameters Value of parameters used in the stochastic gradient descent optimization algorithm.
+   * @param numShards 
+   * @param shardId 
+   * @param options carries optional attributes values
+   * @return a new instance of LoadTPUEmbeddingStochasticGradientDescentParameters
+   */
+  @Endpoint(describeByClass = true)
+  public static LoadTPUEmbeddingStochasticGradientDescentParameters create(Scope scope, Operand<TFloat32> parameters, Long numShards, Long shardId, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("LoadTPUEmbeddingStochasticGradientDescentParameters", scope.makeOpName("LoadTPUEmbeddingStochasticGradientDescentParameters"));
+    opBuilder.addInput(parameters.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("num_shards", numShards);
+    opBuilder.setAttr("shard_id", shardId);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.tableId != null) {
+          opBuilder.setAttr("table_id", opts.tableId);
+        }
+        if (opts.tableName != null) {
+          opBuilder.setAttr("table_name", opts.tableName);
+        }
+        if (opts.config != null) {
+          opBuilder.setAttr("config", opts.config);
+        }
+      }
+    }
+    return new LoadTPUEmbeddingStochasticGradientDescentParameters(opBuilder.build());
+  }
+  
+  /**
+   * @param tableId 
+   */
+  public static Options tableId(Long tableId) {
+    return new Options().tableId(tableId);
+  }
+  
+  /**
+   * @param tableName 
+   */
+  public static Options tableName(String tableName) {
+    return new Options().tableName(tableName);
+  }
+  
+  /**
+   * @param config 
+   */
+  public static Options config(String config) {
+    return new Options().config(config);
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "LoadTPUEmbeddingStochasticGradientDescentParameters";
+  
+  private LoadTPUEmbeddingStochasticGradientDescentParameters(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingStochasticGradientDescentParametersGradAccumDebug.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingStochasticGradientDescentParametersGradAccumDebug.java
new file mode 100644
index 00000000000..e408844e484
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/LoadTPUEmbeddingStochasticGradientDescentParametersGradAccumDebug.java
@@ -0,0 +1,139 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.tpu;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+
+/**
+ * Load SGD embedding parameters.
+ * <p>
+ * An op that loads optimization parameters into HBM for embedding. Must be
+ * preceded by a ConfigureTPUEmbeddingHost op that sets up the correct
+ * embedding table configuration. For example, this op is used to install
+ * parameters that are loaded from a checkpoint before a training loop is
+ * executed.
+ */
+public final class LoadTPUEmbeddingStochasticGradientDescentParametersGradAccumDebug extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.tpu.LoadTPUEmbeddingStochasticGradientDescentParametersGradAccumDebug}
+   */
+  public static class Options {
+    
+    /**
+     * @param tableId 
+     */
+    public Options tableId(Long tableId) {
+      this.tableId = tableId;
+      return this;
+    }
+    
+    /**
+     * @param tableName 
+     */
+    public Options tableName(String tableName) {
+      this.tableName = tableName;
+      return this;
+    }
+    
+    /**
+     * @param config 
+     */
+    public Options config(String config) {
+      this.config = config;
+      return this;
+    }
+    
+    private Long tableId;
+    private String tableName;
+    private String config;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new LoadTPUEmbeddingStochasticGradientDescentParametersGradAccumDebug operation.
+   * 
+   * @param scope current scope
+   * @param parameters Value of parameters used in the stochastic gradient descent optimization algorithm.
+   * @param gradientAccumulators Value of gradient_accumulators used in the Adadelta optimization algorithm.
+   * @param numShards 
+   * @param shardId 
+   * @param options carries optional attributes values
+   * @return a new instance of LoadTPUEmbeddingStochasticGradientDescentParametersGradAccumDebug
+   */
+  @Endpoint(describeByClass = true)
+  public static LoadTPUEmbeddingStochasticGradientDescentParametersGradAccumDebug create(Scope scope, Operand<TFloat32> parameters, Operand<TFloat32> gradientAccumulators, Long numShards, Long shardId, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("LoadTPUEmbeddingStochasticGradientDescentParametersGradAccumDebug", scope.makeOpName("LoadTPUEmbeddingStochasticGradientDescentParametersGradAccumDebug"));
+    opBuilder.addInput(parameters.asOutput());
+    opBuilder.addInput(gradientAccumulators.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("num_shards", numShards);
+    opBuilder.setAttr("shard_id", shardId);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.tableId != null) {
+          opBuilder.setAttr("table_id", opts.tableId);
+        }
+        if (opts.tableName != null) {
+          opBuilder.setAttr("table_name", opts.tableName);
+        }
+        if (opts.config != null) {
+          opBuilder.setAttr("config", opts.config);
+        }
+      }
+    }
+    return new LoadTPUEmbeddingStochasticGradientDescentParametersGradAccumDebug(opBuilder.build());
+  }
+  
+  /**
+   * @param tableId 
+   */
+  public static Options tableId(Long tableId) {
+    return new Options().tableId(tableId);
+  }
+  
+  /**
+   * @param tableName 
+   */
+  public static Options tableName(String tableName) {
+    return new Options().tableName(tableName);
+  }
+  
+  /**
+   * @param config 
+   */
+  public static Options config(String config) {
+    return new Options().config(config);
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "LoadTPUEmbeddingStochasticGradientDescentParametersGradAccumDebug";
+  
+  private LoadTPUEmbeddingStochasticGradientDescentParametersGradAccumDebug(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/OutfeedDequeue.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/OutfeedDequeue.java
new file mode 100644
index 00000000000..c11a81f1e54
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/OutfeedDequeue.java
@@ -0,0 +1,117 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.tpu;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Retrieves a single tensor from the computation outfeed.
+ * <p>
+ * This operation will block indefinitely until data is available.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+public final class OutfeedDequeue<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.tpu.OutfeedDequeue}
+   */
+  public static class Options {
+    
+    /**
+     * @param deviceOrdinal The TPU device to use. This should be -1 when the Op
+     * is running on a TPU device, and >= 0 when the Op is running on the CPU
+     * device.
+     */
+    public Options deviceOrdinal(Long deviceOrdinal) {
+      this.deviceOrdinal = deviceOrdinal;
+      return this;
+    }
+    
+    private Long deviceOrdinal;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new OutfeedDequeue operation.
+   * 
+   * @param scope current scope
+   * @param dtype The type of elements in the tensor.
+   * @param shape The shape of the tensor.
+   * @param options carries optional attributes values
+   * @return a new instance of OutfeedDequeue
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> OutfeedDequeue<T> create(Scope scope, Class<T> dtype, Shape shape, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("OutfeedDequeue", scope.makeOpName("OutfeedDequeue"));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("dtype", dtype);
+    opBuilder.setAttr("shape", shape);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.deviceOrdinal != null) {
+          opBuilder.setAttr("device_ordinal", opts.deviceOrdinal);
+        }
+      }
+    }
+    return new OutfeedDequeue<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param deviceOrdinal The TPU device to use. This should be -1 when the Op
+   * is running on a TPU device, and >= 0 when the Op is running on the CPU
+   * device.
+   */
+  public static Options deviceOrdinal(Long deviceOrdinal) {
+    return new Options().deviceOrdinal(deviceOrdinal);
+  }
+  
+  /**
+   * A tensor that will be read from the device outfeed.
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "OutfeedDequeue";
+  
+  private Output<T> output;
+  
+  private OutfeedDequeue(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/OutfeedDequeueTuple.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/OutfeedDequeueTuple.java
new file mode 100644
index 00000000000..284c2d7a0fb
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/OutfeedDequeueTuple.java
@@ -0,0 +1,130 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.tpu;
+
+import java.util.Arrays;
+import java.util.Iterator;
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Retrieve multiple values from the computation outfeed.
+ * <p>
+ * This operation will block indefinitely until data is available. Output `i`
+ * corresponds to XLA tuple element `i`.
+ */
+public final class OutfeedDequeueTuple extends RawOp implements Iterable<Operand<TType>> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.tpu.OutfeedDequeueTuple}
+   */
+  public static class Options {
+    
+    /**
+     * @param deviceOrdinal The TPU device to use. This should be -1 when the Op
+     * is running on a TPU device, and >= 0 when the Op is running on the CPU
+     * device.
+     */
+    public Options deviceOrdinal(Long deviceOrdinal) {
+      this.deviceOrdinal = deviceOrdinal;
+      return this;
+    }
+    
+    private Long deviceOrdinal;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new OutfeedDequeueTuple operation.
+   * 
+   * @param scope current scope
+   * @param dtypes The element types of each element in `outputs`.
+   * @param shapes The shapes of each tensor in `outputs`.
+   * @param options carries optional attributes values
+   * @return a new instance of OutfeedDequeueTuple
+   */
+  @Endpoint(describeByClass = true)
+  public static OutfeedDequeueTuple create(Scope scope, List<Class<? extends TType>> dtypes, List<Shape> shapes, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("OutfeedDequeueTuple", scope.makeOpName("OutfeedDequeueTuple"));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    Class[] dtypesArray = new Class[dtypes.size()];
+    for (int i = 0; i < dtypesArray.length; ++i) {
+      dtypesArray[i] = dtypes.get(i);
+    }
+    opBuilder.setAttr("dtypes", dtypesArray);
+    Shape[] shapesArray = new Shape[shapes.size()];
+    for (int i = 0; i < shapesArray.length; ++i) {
+      shapesArray[i] = shapes.get(i);
+    }
+    opBuilder.setAttr("shapes", shapesArray);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.deviceOrdinal != null) {
+          opBuilder.setAttr("device_ordinal", opts.deviceOrdinal);
+        }
+      }
+    }
+    return new OutfeedDequeueTuple(opBuilder.build());
+  }
+  
+  /**
+   * @param deviceOrdinal The TPU device to use. This should be -1 when the Op
+   * is running on a TPU device, and >= 0 when the Op is running on the CPU
+   * device.
+   */
+  public static Options deviceOrdinal(Long deviceOrdinal) {
+    return new Options().deviceOrdinal(deviceOrdinal);
+  }
+  
+  /**
+   * A list of tensors that will be read from the outfeed.
+   */
+  public List<Output<?>> outputs() {
+    return outputs;
+  }
+  
+  @Override
+  @SuppressWarnings({"rawtypes", "unchecked"})
+  public Iterator<Operand<TType>> iterator() {
+    return (Iterator) outputs.iterator();
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "OutfeedDequeueTuple";
+  
+  private List<Output<?>> outputs;
+  
+  private OutfeedDequeueTuple(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    int outputsLength = operation.outputListLength("outputs");
+    outputs = Arrays.asList(operation.outputList(outputIdx, outputsLength));
+    outputIdx += outputsLength;
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/OutfeedEnqueue.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/OutfeedEnqueue.java
new file mode 100644
index 00000000000..5b5f059a81e
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/OutfeedEnqueue.java
@@ -0,0 +1,55 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.tpu;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Enqueue a Tensor on the computation outfeed.
+ */
+public final class OutfeedEnqueue extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new OutfeedEnqueue operation.
+   * 
+   * @param scope current scope
+   * @param input A tensor that will be inserted into the outfeed queue.
+   * @return a new instance of OutfeedEnqueue
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> OutfeedEnqueue create(Scope scope, Operand<T> input) {
+    OperationBuilder opBuilder = scope.env().opBuilder("OutfeedEnqueue", scope.makeOpName("OutfeedEnqueue"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new OutfeedEnqueue(opBuilder.build());
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "OutfeedEnqueue";
+  
+  private OutfeedEnqueue(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/OutfeedEnqueueTuple.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/OutfeedEnqueueTuple.java
new file mode 100644
index 00000000000..8bfd04b9a2c
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/OutfeedEnqueueTuple.java
@@ -0,0 +1,56 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.tpu;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.Operands;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+
+/**
+ * Enqueue multiple Tensor values on the computation outfeed.
+ */
+public final class OutfeedEnqueueTuple extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new OutfeedEnqueueTuple operation.
+   * 
+   * @param scope current scope
+   * @param inputs A list of tensors that will be inserted into the outfeed queue as an
+   * XLA tuple.
+   * @return a new instance of OutfeedEnqueueTuple
+   */
+  @Endpoint(describeByClass = true)
+  public static OutfeedEnqueueTuple create(Scope scope, Iterable<Operand<?>> inputs) {
+    OperationBuilder opBuilder = scope.env().opBuilder("OutfeedEnqueueTuple", scope.makeOpName("OutfeedEnqueueTuple"));
+    opBuilder.addInputList(Operands.asOutputs(inputs));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new OutfeedEnqueueTuple(opBuilder.build());
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "OutfeedEnqueueTuple";
+  
+  private OutfeedEnqueueTuple(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/Prelinearize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/Prelinearize.java
new file mode 100644
index 00000000000..c5c9573fb75
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/Prelinearize.java
@@ -0,0 +1,135 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.tpu;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * An op which linearizes one Tensor value to an opaque variant tensor.
+ */
+public final class Prelinearize extends RawOp implements Operand<TType> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.tpu.Prelinearize}
+   */
+  public static class Options {
+    
+    /**
+     * @param shape The shape of the tensor.
+     */
+    public Options shape(Shape shape) {
+      this.shape = shape;
+      return this;
+    }
+    
+    /**
+     * @param layout A vector holding the requested layout in minor-to-major sequence. If a layout
+     * attribute is passed but its values are all -1 the layout will be computed by
+     * the infeed operation.
+     */
+    public Options layout(List<Long> layout) {
+      this.layout = layout;
+      return this;
+    }
+    
+    private Shape shape;
+    private List<Long> layout;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new Prelinearize operation.
+   * 
+   * @param scope current scope
+   * @param input A tensor that will be linearized.
+   * @param options carries optional attributes values
+   * @return a new instance of Prelinearize
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> Prelinearize create(Scope scope, Operand<T> input, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("Prelinearize", scope.makeOpName("Prelinearize"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.shape != null) {
+          opBuilder.setAttr("shape", opts.shape);
+        }
+        if (opts.layout != null) {
+          long[] layoutArray = new long[opts.layout.size()];
+          for (int i = 0; i < layoutArray.length; ++i) {
+            layoutArray[i] = opts.layout.get(i);
+          }
+          opBuilder.setAttr("layout", layoutArray);
+        }
+      }
+    }
+    return new Prelinearize(opBuilder.build());
+  }
+  
+  /**
+   * @param shape The shape of the tensor.
+   */
+  public static Options shape(Shape shape) {
+    return new Options().shape(shape);
+  }
+  
+  /**
+   * @param layout A vector holding the requested layout in minor-to-major sequence. If a layout
+   * attribute is passed but its values are all -1 the layout will be computed by
+   * the infeed operation.
+   */
+  public static Options layout(List<Long> layout) {
+    return new Options().layout(layout);
+  }
+  
+  /**
+   */
+  public Output<?> output() {
+    return output;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "Prelinearize";
+  
+  private Output<?> output;
+  
+  private Prelinearize(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/PrelinearizeTuple.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/PrelinearizeTuple.java
new file mode 100644
index 00000000000..6ff29f7b63a
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/PrelinearizeTuple.java
@@ -0,0 +1,125 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.tpu;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.Operands;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * An op which linearizes multiple Tensor values to an opaque variant tensor.
+ */
+public final class PrelinearizeTuple extends RawOp implements Operand<TType> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.tpu.PrelinearizeTuple}
+   */
+  public static class Options {
+    
+    /**
+     * @param layouts A vector holding the requested layout in minor-to-major sequence for all the
+     * tuple shapes in the order the shapes appear in the "shapes" input. The layout
+     * elements for a sub-shape can be set to -1 in which case the corresponding layout
+     * will be computed by the infeed operation.
+     */
+    public Options layouts(List<Long> layouts) {
+      this.layouts = layouts;
+      return this;
+    }
+    
+    private List<Long> layouts;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new PrelinearizeTuple operation.
+   * 
+   * @param scope current scope
+   * @param inputs A list of tensors that will be provided using the infeed mechanism.
+   * @param shapes The shapes of each tensor in `inputs`.
+   * @param options carries optional attributes values
+   * @return a new instance of PrelinearizeTuple
+   */
+  @Endpoint(describeByClass = true)
+  public static PrelinearizeTuple create(Scope scope, Iterable<Operand<?>> inputs, List<Shape> shapes, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("PrelinearizeTuple", scope.makeOpName("PrelinearizeTuple"));
+    opBuilder.addInputList(Operands.asOutputs(inputs));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    Shape[] shapesArray = new Shape[shapes.size()];
+    for (int i = 0; i < shapesArray.length; ++i) {
+      shapesArray[i] = shapes.get(i);
+    }
+    opBuilder.setAttr("shapes", shapesArray);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.layouts != null) {
+          long[] layoutsArray = new long[opts.layouts.size()];
+          for (int i = 0; i < layoutsArray.length; ++i) {
+            layoutsArray[i] = opts.layouts.get(i);
+          }
+          opBuilder.setAttr("layouts", layoutsArray);
+        }
+      }
+    }
+    return new PrelinearizeTuple(opBuilder.build());
+  }
+  
+  /**
+   * @param layouts A vector holding the requested layout in minor-to-major sequence for all the
+   * tuple shapes in the order the shapes appear in the "shapes" input. The layout
+   * elements for a sub-shape can be set to -1 in which case the corresponding layout
+   * will be computed by the infeed operation.
+   */
+  public static Options layouts(List<Long> layouts) {
+    return new Options().layouts(layouts);
+  }
+  
+  /**
+   */
+  public Output<?> output() {
+    return output;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "PrelinearizeTuple";
+  
+  private Output<?> output;
+  
+  private PrelinearizeTuple(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RecvTPUEmbeddingActivations.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RecvTPUEmbeddingActivations.java
new file mode 100644
index 00000000000..e2bacade157
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RecvTPUEmbeddingActivations.java
@@ -0,0 +1,90 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.tpu;
+
+import java.util.Arrays;
+import java.util.Iterator;
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+
+/**
+ * An op that receives embedding activations on the TPU.
+ * <p>
+ * The TPU system performs the embedding lookups and aggregations specified by
+ * the arguments to TPUEmbeddingEnqueue(Integer/Sparse/SparseTensor)Batch. The
+ * results of these aggregations are visible to the Tensorflow Graph as the
+ * outputs of a RecvTPUEmbeddingActivations op. This op returns a list containing
+ * one Tensor of activations per table specified in the model. There can be at
+ * most one RecvTPUEmbeddingActivations op in the TPU graph.
+ */
+public final class RecvTPUEmbeddingActivations extends RawOp implements Iterable<Operand<TFloat32>> {
+  
+  /**
+   * Factory method to create a class wrapping a new RecvTPUEmbeddingActivations operation.
+   * 
+   * @param scope current scope
+   * @param numOutputs The number of output activation tensors, equal to the number of
+   * embedding tables in the model.
+   * @param config Serialized TPUEmbeddingConfiguration proto.
+   * @return a new instance of RecvTPUEmbeddingActivations
+   */
+  @Endpoint(describeByClass = true)
+  public static RecvTPUEmbeddingActivations create(Scope scope, Long numOutputs, String config) {
+    OperationBuilder opBuilder = scope.env().opBuilder("RecvTPUEmbeddingActivations", scope.makeOpName("RecvTPUEmbeddingActivations"));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("num_outputs", numOutputs);
+    opBuilder.setAttr("config", config);
+    return new RecvTPUEmbeddingActivations(opBuilder.build());
+  }
+  
+  /**
+   * A TensorList of embedding activations containing one Tensor per
+   * embedding table in the model.
+   */
+  public List<Output<TFloat32>> outputs() {
+    return outputs;
+  }
+  
+  @Override
+  @SuppressWarnings({"rawtypes", "unchecked"})
+  public Iterator<Operand<TFloat32>> iterator() {
+    return (Iterator) outputs.iterator();
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "RecvTPUEmbeddingActivations";
+  
+  private List<Output<TFloat32>> outputs;
+  
+  @SuppressWarnings("unchecked")
+  private RecvTPUEmbeddingActivations(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    int outputsLength = operation.outputListLength("outputs");
+    outputs = Arrays.asList((Output<TFloat32>[])operation.outputList(outputIdx, outputsLength));
+    outputIdx += outputsLength;
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingADAMParameters.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingADAMParameters.java
new file mode 100644
index 00000000000..6201ed4b9eb
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingADAMParameters.java
@@ -0,0 +1,163 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.tpu;
+
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+
+/**
+ * Retrieve ADAM embedding parameters.
+ * <p>
+ * An op that retrieves optimization parameters from embedding to host
+ * memory. Must be preceded by a ConfigureTPUEmbeddingHost op that sets up
+ * the correct embedding table configuration. For example, this op is
+ * used to retrieve updated parameters before saving a checkpoint.
+ */
+public final class RetrieveTPUEmbeddingADAMParameters extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.tpu.RetrieveTPUEmbeddingADAMParameters}
+   */
+  public static class Options {
+    
+    /**
+     * @param tableId 
+     */
+    public Options tableId(Long tableId) {
+      this.tableId = tableId;
+      return this;
+    }
+    
+    /**
+     * @param tableName 
+     */
+    public Options tableName(String tableName) {
+      this.tableName = tableName;
+      return this;
+    }
+    
+    /**
+     * @param config 
+     */
+    public Options config(String config) {
+      this.config = config;
+      return this;
+    }
+    
+    private Long tableId;
+    private String tableName;
+    private String config;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new RetrieveTPUEmbeddingADAMParameters operation.
+   * 
+   * @param scope current scope
+   * @param numShards 
+   * @param shardId 
+   * @param options carries optional attributes values
+   * @return a new instance of RetrieveTPUEmbeddingADAMParameters
+   */
+  @Endpoint(describeByClass = true)
+  public static RetrieveTPUEmbeddingADAMParameters create(Scope scope, Long numShards, Long shardId, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("RetrieveTPUEmbeddingADAMParameters", scope.makeOpName("RetrieveTPUEmbeddingADAMParameters"));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("num_shards", numShards);
+    opBuilder.setAttr("shard_id", shardId);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.tableId != null) {
+          opBuilder.setAttr("table_id", opts.tableId);
+        }
+        if (opts.tableName != null) {
+          opBuilder.setAttr("table_name", opts.tableName);
+        }
+        if (opts.config != null) {
+          opBuilder.setAttr("config", opts.config);
+        }
+      }
+    }
+    return new RetrieveTPUEmbeddingADAMParameters(opBuilder.build());
+  }
+  
+  /**
+   * @param tableId 
+   */
+  public static Options tableId(Long tableId) {
+    return new Options().tableId(tableId);
+  }
+  
+  /**
+   * @param tableName 
+   */
+  public static Options tableName(String tableName) {
+    return new Options().tableName(tableName);
+  }
+  
+  /**
+   * @param config 
+   */
+  public static Options config(String config) {
+    return new Options().config(config);
+  }
+  
+  /**
+   * Parameter parameters updated by the ADAM optimization algorithm.
+   */
+  public Output<TFloat32> parameters() {
+    return parameters;
+  }
+  
+  /**
+   * Parameter momenta updated by the ADAM optimization algorithm.
+   */
+  public Output<TFloat32> momenta() {
+    return momenta;
+  }
+  
+  /**
+   * Parameter velocities updated by the ADAM optimization algorithm.
+   */
+  public Output<TFloat32> velocities() {
+    return velocities;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "RetrieveTPUEmbeddingADAMParameters";
+  
+  private Output<TFloat32> parameters;
+  private Output<TFloat32> momenta;
+  private Output<TFloat32> velocities;
+  
+  private RetrieveTPUEmbeddingADAMParameters(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    parameters = operation.output(outputIdx++);
+    momenta = operation.output(outputIdx++);
+    velocities = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingADAMParametersGradAccumDebug.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingADAMParametersGradAccumDebug.java
new file mode 100644
index 00000000000..423fc97541e
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingADAMParametersGradAccumDebug.java
@@ -0,0 +1,172 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.tpu;
+
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+
+/**
+ * Retrieve ADAM embedding parameters with debug support.
+ * <p>
+ * An op that retrieves optimization parameters from embedding to host
+ * memory. Must be preceded by a ConfigureTPUEmbeddingHost op that sets up
+ * the correct embedding table configuration. For example, this op is
+ * used to retrieve updated parameters before saving a checkpoint.
+ */
+public final class RetrieveTPUEmbeddingADAMParametersGradAccumDebug extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.tpu.RetrieveTPUEmbeddingADAMParametersGradAccumDebug}
+   */
+  public static class Options {
+    
+    /**
+     * @param tableId 
+     */
+    public Options tableId(Long tableId) {
+      this.tableId = tableId;
+      return this;
+    }
+    
+    /**
+     * @param tableName 
+     */
+    public Options tableName(String tableName) {
+      this.tableName = tableName;
+      return this;
+    }
+    
+    /**
+     * @param config 
+     */
+    public Options config(String config) {
+      this.config = config;
+      return this;
+    }
+    
+    private Long tableId;
+    private String tableName;
+    private String config;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new RetrieveTPUEmbeddingADAMParametersGradAccumDebug operation.
+   * 
+   * @param scope current scope
+   * @param numShards 
+   * @param shardId 
+   * @param options carries optional attributes values
+   * @return a new instance of RetrieveTPUEmbeddingADAMParametersGradAccumDebug
+   */
+  @Endpoint(describeByClass = true)
+  public static RetrieveTPUEmbeddingADAMParametersGradAccumDebug create(Scope scope, Long numShards, Long shardId, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("RetrieveTPUEmbeddingADAMParametersGradAccumDebug", scope.makeOpName("RetrieveTPUEmbeddingADAMParametersGradAccumDebug"));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("num_shards", numShards);
+    opBuilder.setAttr("shard_id", shardId);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.tableId != null) {
+          opBuilder.setAttr("table_id", opts.tableId);
+        }
+        if (opts.tableName != null) {
+          opBuilder.setAttr("table_name", opts.tableName);
+        }
+        if (opts.config != null) {
+          opBuilder.setAttr("config", opts.config);
+        }
+      }
+    }
+    return new RetrieveTPUEmbeddingADAMParametersGradAccumDebug(opBuilder.build());
+  }
+  
+  /**
+   * @param tableId 
+   */
+  public static Options tableId(Long tableId) {
+    return new Options().tableId(tableId);
+  }
+  
+  /**
+   * @param tableName 
+   */
+  public static Options tableName(String tableName) {
+    return new Options().tableName(tableName);
+  }
+  
+  /**
+   * @param config 
+   */
+  public static Options config(String config) {
+    return new Options().config(config);
+  }
+  
+  /**
+   * Parameter parameters updated by the ADAM optimization algorithm.
+   */
+  public Output<TFloat32> parameters() {
+    return parameters;
+  }
+  
+  /**
+   * Parameter momenta updated by the ADAM optimization algorithm.
+   */
+  public Output<TFloat32> momenta() {
+    return momenta;
+  }
+  
+  /**
+   * Parameter velocities updated by the ADAM optimization algorithm.
+   */
+  public Output<TFloat32> velocities() {
+    return velocities;
+  }
+  
+  /**
+   * Parameter gradient_accumulators updated by the ADAM optimization algorithm.
+   */
+  public Output<TFloat32> gradientAccumulators() {
+    return gradientAccumulators;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "RetrieveTPUEmbeddingADAMParametersGradAccumDebug";
+  
+  private Output<TFloat32> parameters;
+  private Output<TFloat32> momenta;
+  private Output<TFloat32> velocities;
+  private Output<TFloat32> gradientAccumulators;
+  
+  private RetrieveTPUEmbeddingADAMParametersGradAccumDebug(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    parameters = operation.output(outputIdx++);
+    momenta = operation.output(outputIdx++);
+    velocities = operation.output(outputIdx++);
+    gradientAccumulators = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingAdadeltaParameters.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingAdadeltaParameters.java
new file mode 100644
index 00000000000..9a00080e017
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingAdadeltaParameters.java
@@ -0,0 +1,163 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.tpu;
+
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+
+/**
+ * Retrieve Adadelta embedding parameters.
+ * <p>
+ * An op that retrieves optimization parameters from embedding to host
+ * memory. Must be preceded by a ConfigureTPUEmbeddingHost op that sets up
+ * the correct embedding table configuration. For example, this op is
+ * used to retrieve updated parameters before saving a checkpoint.
+ */
+public final class RetrieveTPUEmbeddingAdadeltaParameters extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.tpu.RetrieveTPUEmbeddingAdadeltaParameters}
+   */
+  public static class Options {
+    
+    /**
+     * @param tableId 
+     */
+    public Options tableId(Long tableId) {
+      this.tableId = tableId;
+      return this;
+    }
+    
+    /**
+     * @param tableName 
+     */
+    public Options tableName(String tableName) {
+      this.tableName = tableName;
+      return this;
+    }
+    
+    /**
+     * @param config 
+     */
+    public Options config(String config) {
+      this.config = config;
+      return this;
+    }
+    
+    private Long tableId;
+    private String tableName;
+    private String config;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new RetrieveTPUEmbeddingAdadeltaParameters operation.
+   * 
+   * @param scope current scope
+   * @param numShards 
+   * @param shardId 
+   * @param options carries optional attributes values
+   * @return a new instance of RetrieveTPUEmbeddingAdadeltaParameters
+   */
+  @Endpoint(describeByClass = true)
+  public static RetrieveTPUEmbeddingAdadeltaParameters create(Scope scope, Long numShards, Long shardId, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("RetrieveTPUEmbeddingAdadeltaParameters", scope.makeOpName("RetrieveTPUEmbeddingAdadeltaParameters"));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("num_shards", numShards);
+    opBuilder.setAttr("shard_id", shardId);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.tableId != null) {
+          opBuilder.setAttr("table_id", opts.tableId);
+        }
+        if (opts.tableName != null) {
+          opBuilder.setAttr("table_name", opts.tableName);
+        }
+        if (opts.config != null) {
+          opBuilder.setAttr("config", opts.config);
+        }
+      }
+    }
+    return new RetrieveTPUEmbeddingAdadeltaParameters(opBuilder.build());
+  }
+  
+  /**
+   * @param tableId 
+   */
+  public static Options tableId(Long tableId) {
+    return new Options().tableId(tableId);
+  }
+  
+  /**
+   * @param tableName 
+   */
+  public static Options tableName(String tableName) {
+    return new Options().tableName(tableName);
+  }
+  
+  /**
+   * @param config 
+   */
+  public static Options config(String config) {
+    return new Options().config(config);
+  }
+  
+  /**
+   * Parameter parameters updated by the Adadelta optimization algorithm.
+   */
+  public Output<TFloat32> parameters() {
+    return parameters;
+  }
+  
+  /**
+   * Parameter accumulators updated by the Adadelta optimization algorithm.
+   */
+  public Output<TFloat32> accumulators() {
+    return accumulators;
+  }
+  
+  /**
+   * Parameter updates updated by the Adadelta optimization algorithm.
+   */
+  public Output<TFloat32> updates() {
+    return updates;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "RetrieveTPUEmbeddingAdadeltaParameters";
+  
+  private Output<TFloat32> parameters;
+  private Output<TFloat32> accumulators;
+  private Output<TFloat32> updates;
+  
+  private RetrieveTPUEmbeddingAdadeltaParameters(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    parameters = operation.output(outputIdx++);
+    accumulators = operation.output(outputIdx++);
+    updates = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingAdadeltaParametersGradAccumDebug.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingAdadeltaParametersGradAccumDebug.java
new file mode 100644
index 00000000000..3fec5d6a6f7
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingAdadeltaParametersGradAccumDebug.java
@@ -0,0 +1,172 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.tpu;
+
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+
+/**
+ * Retrieve Adadelta embedding parameters with debug support.
+ * <p>
+ * An op that retrieves optimization parameters from embedding to host
+ * memory. Must be preceded by a ConfigureTPUEmbeddingHost op that sets up
+ * the correct embedding table configuration. For example, this op is
+ * used to retrieve updated parameters before saving a checkpoint.
+ */
+public final class RetrieveTPUEmbeddingAdadeltaParametersGradAccumDebug extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.tpu.RetrieveTPUEmbeddingAdadeltaParametersGradAccumDebug}
+   */
+  public static class Options {
+    
+    /**
+     * @param tableId 
+     */
+    public Options tableId(Long tableId) {
+      this.tableId = tableId;
+      return this;
+    }
+    
+    /**
+     * @param tableName 
+     */
+    public Options tableName(String tableName) {
+      this.tableName = tableName;
+      return this;
+    }
+    
+    /**
+     * @param config 
+     */
+    public Options config(String config) {
+      this.config = config;
+      return this;
+    }
+    
+    private Long tableId;
+    private String tableName;
+    private String config;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new RetrieveTPUEmbeddingAdadeltaParametersGradAccumDebug operation.
+   * 
+   * @param scope current scope
+   * @param numShards 
+   * @param shardId 
+   * @param options carries optional attributes values
+   * @return a new instance of RetrieveTPUEmbeddingAdadeltaParametersGradAccumDebug
+   */
+  @Endpoint(describeByClass = true)
+  public static RetrieveTPUEmbeddingAdadeltaParametersGradAccumDebug create(Scope scope, Long numShards, Long shardId, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("RetrieveTPUEmbeddingAdadeltaParametersGradAccumDebug", scope.makeOpName("RetrieveTPUEmbeddingAdadeltaParametersGradAccumDebug"));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("num_shards", numShards);
+    opBuilder.setAttr("shard_id", shardId);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.tableId != null) {
+          opBuilder.setAttr("table_id", opts.tableId);
+        }
+        if (opts.tableName != null) {
+          opBuilder.setAttr("table_name", opts.tableName);
+        }
+        if (opts.config != null) {
+          opBuilder.setAttr("config", opts.config);
+        }
+      }
+    }
+    return new RetrieveTPUEmbeddingAdadeltaParametersGradAccumDebug(opBuilder.build());
+  }
+  
+  /**
+   * @param tableId 
+   */
+  public static Options tableId(Long tableId) {
+    return new Options().tableId(tableId);
+  }
+  
+  /**
+   * @param tableName 
+   */
+  public static Options tableName(String tableName) {
+    return new Options().tableName(tableName);
+  }
+  
+  /**
+   * @param config 
+   */
+  public static Options config(String config) {
+    return new Options().config(config);
+  }
+  
+  /**
+   * Parameter parameters updated by the Adadelta optimization algorithm.
+   */
+  public Output<TFloat32> parameters() {
+    return parameters;
+  }
+  
+  /**
+   * Parameter accumulators updated by the Adadelta optimization algorithm.
+   */
+  public Output<TFloat32> accumulators() {
+    return accumulators;
+  }
+  
+  /**
+   * Parameter updates updated by the Adadelta optimization algorithm.
+   */
+  public Output<TFloat32> updates() {
+    return updates;
+  }
+  
+  /**
+   * Parameter gradient_accumulators updated by the Adadelta optimization algorithm.
+   */
+  public Output<TFloat32> gradientAccumulators() {
+    return gradientAccumulators;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "RetrieveTPUEmbeddingAdadeltaParametersGradAccumDebug";
+  
+  private Output<TFloat32> parameters;
+  private Output<TFloat32> accumulators;
+  private Output<TFloat32> updates;
+  private Output<TFloat32> gradientAccumulators;
+  
+  private RetrieveTPUEmbeddingAdadeltaParametersGradAccumDebug(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    parameters = operation.output(outputIdx++);
+    accumulators = operation.output(outputIdx++);
+    updates = operation.output(outputIdx++);
+    gradientAccumulators = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingAdagradParameters.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingAdagradParameters.java
new file mode 100644
index 00000000000..07ed32ae1f9
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingAdagradParameters.java
@@ -0,0 +1,154 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.tpu;
+
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+
+/**
+ * Retrieve Adagrad embedding parameters.
+ * <p>
+ * An op that retrieves optimization parameters from embedding to host
+ * memory. Must be preceded by a ConfigureTPUEmbeddingHost op that sets up
+ * the correct embedding table configuration. For example, this op is
+ * used to retrieve updated parameters before saving a checkpoint.
+ */
+public final class RetrieveTPUEmbeddingAdagradParameters extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.tpu.RetrieveTPUEmbeddingAdagradParameters}
+   */
+  public static class Options {
+    
+    /**
+     * @param tableId 
+     */
+    public Options tableId(Long tableId) {
+      this.tableId = tableId;
+      return this;
+    }
+    
+    /**
+     * @param tableName 
+     */
+    public Options tableName(String tableName) {
+      this.tableName = tableName;
+      return this;
+    }
+    
+    /**
+     * @param config 
+     */
+    public Options config(String config) {
+      this.config = config;
+      return this;
+    }
+    
+    private Long tableId;
+    private String tableName;
+    private String config;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new RetrieveTPUEmbeddingAdagradParameters operation.
+   * 
+   * @param scope current scope
+   * @param numShards 
+   * @param shardId 
+   * @param options carries optional attributes values
+   * @return a new instance of RetrieveTPUEmbeddingAdagradParameters
+   */
+  @Endpoint(describeByClass = true)
+  public static RetrieveTPUEmbeddingAdagradParameters create(Scope scope, Long numShards, Long shardId, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("RetrieveTPUEmbeddingAdagradParameters", scope.makeOpName("RetrieveTPUEmbeddingAdagradParameters"));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("num_shards", numShards);
+    opBuilder.setAttr("shard_id", shardId);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.tableId != null) {
+          opBuilder.setAttr("table_id", opts.tableId);
+        }
+        if (opts.tableName != null) {
+          opBuilder.setAttr("table_name", opts.tableName);
+        }
+        if (opts.config != null) {
+          opBuilder.setAttr("config", opts.config);
+        }
+      }
+    }
+    return new RetrieveTPUEmbeddingAdagradParameters(opBuilder.build());
+  }
+  
+  /**
+   * @param tableId 
+   */
+  public static Options tableId(Long tableId) {
+    return new Options().tableId(tableId);
+  }
+  
+  /**
+   * @param tableName 
+   */
+  public static Options tableName(String tableName) {
+    return new Options().tableName(tableName);
+  }
+  
+  /**
+   * @param config 
+   */
+  public static Options config(String config) {
+    return new Options().config(config);
+  }
+  
+  /**
+   * Parameter parameters updated by the Adagrad optimization algorithm.
+   */
+  public Output<TFloat32> parameters() {
+    return parameters;
+  }
+  
+  /**
+   * Parameter accumulators updated by the Adagrad optimization algorithm.
+   */
+  public Output<TFloat32> accumulators() {
+    return accumulators;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "RetrieveTPUEmbeddingAdagradParameters";
+  
+  private Output<TFloat32> parameters;
+  private Output<TFloat32> accumulators;
+  
+  private RetrieveTPUEmbeddingAdagradParameters(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    parameters = operation.output(outputIdx++);
+    accumulators = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingAdagradParametersGradAccumDebug.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingAdagradParametersGradAccumDebug.java
new file mode 100644
index 00000000000..39a8a7ab791
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingAdagradParametersGradAccumDebug.java
@@ -0,0 +1,163 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.tpu;
+
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+
+/**
+ * Retrieve Adagrad embedding parameters with debug support.
+ * <p>
+ * An op that retrieves optimization parameters from embedding to host
+ * memory. Must be preceded by a ConfigureTPUEmbeddingHost op that sets up
+ * the correct embedding table configuration. For example, this op is
+ * used to retrieve updated parameters before saving a checkpoint.
+ */
+public final class RetrieveTPUEmbeddingAdagradParametersGradAccumDebug extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.tpu.RetrieveTPUEmbeddingAdagradParametersGradAccumDebug}
+   */
+  public static class Options {
+    
+    /**
+     * @param tableId 
+     */
+    public Options tableId(Long tableId) {
+      this.tableId = tableId;
+      return this;
+    }
+    
+    /**
+     * @param tableName 
+     */
+    public Options tableName(String tableName) {
+      this.tableName = tableName;
+      return this;
+    }
+    
+    /**
+     * @param config 
+     */
+    public Options config(String config) {
+      this.config = config;
+      return this;
+    }
+    
+    private Long tableId;
+    private String tableName;
+    private String config;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new RetrieveTPUEmbeddingAdagradParametersGradAccumDebug operation.
+   * 
+   * @param scope current scope
+   * @param numShards 
+   * @param shardId 
+   * @param options carries optional attributes values
+   * @return a new instance of RetrieveTPUEmbeddingAdagradParametersGradAccumDebug
+   */
+  @Endpoint(describeByClass = true)
+  public static RetrieveTPUEmbeddingAdagradParametersGradAccumDebug create(Scope scope, Long numShards, Long shardId, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("RetrieveTPUEmbeddingAdagradParametersGradAccumDebug", scope.makeOpName("RetrieveTPUEmbeddingAdagradParametersGradAccumDebug"));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("num_shards", numShards);
+    opBuilder.setAttr("shard_id", shardId);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.tableId != null) {
+          opBuilder.setAttr("table_id", opts.tableId);
+        }
+        if (opts.tableName != null) {
+          opBuilder.setAttr("table_name", opts.tableName);
+        }
+        if (opts.config != null) {
+          opBuilder.setAttr("config", opts.config);
+        }
+      }
+    }
+    return new RetrieveTPUEmbeddingAdagradParametersGradAccumDebug(opBuilder.build());
+  }
+  
+  /**
+   * @param tableId 
+   */
+  public static Options tableId(Long tableId) {
+    return new Options().tableId(tableId);
+  }
+  
+  /**
+   * @param tableName 
+   */
+  public static Options tableName(String tableName) {
+    return new Options().tableName(tableName);
+  }
+  
+  /**
+   * @param config 
+   */
+  public static Options config(String config) {
+    return new Options().config(config);
+  }
+  
+  /**
+   * Parameter parameters updated by the Adagrad optimization algorithm.
+   */
+  public Output<TFloat32> parameters() {
+    return parameters;
+  }
+  
+  /**
+   * Parameter accumulators updated by the Adagrad optimization algorithm.
+   */
+  public Output<TFloat32> accumulators() {
+    return accumulators;
+  }
+  
+  /**
+   * Parameter gradient_accumulators updated by the Adagrad optimization algorithm.
+   */
+  public Output<TFloat32> gradientAccumulators() {
+    return gradientAccumulators;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "RetrieveTPUEmbeddingAdagradParametersGradAccumDebug";
+  
+  private Output<TFloat32> parameters;
+  private Output<TFloat32> accumulators;
+  private Output<TFloat32> gradientAccumulators;
+  
+  private RetrieveTPUEmbeddingAdagradParametersGradAccumDebug(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    parameters = operation.output(outputIdx++);
+    accumulators = operation.output(outputIdx++);
+    gradientAccumulators = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingCenteredRMSPropParameters.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingCenteredRMSPropParameters.java
new file mode 100644
index 00000000000..800cbd10aff
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingCenteredRMSPropParameters.java
@@ -0,0 +1,172 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.tpu;
+
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+
+/**
+ * Retrieve centered RMSProp embedding parameters.
+ * <p>
+ * An op that retrieves optimization parameters from embedding to host
+ * memory. Must be preceded by a ConfigureTPUEmbeddingHost op that sets up
+ * the correct embedding table configuration. For example, this op is
+ * used to retrieve updated parameters before saving a checkpoint.
+ */
+public final class RetrieveTPUEmbeddingCenteredRMSPropParameters extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.tpu.RetrieveTPUEmbeddingCenteredRMSPropParameters}
+   */
+  public static class Options {
+    
+    /**
+     * @param tableId 
+     */
+    public Options tableId(Long tableId) {
+      this.tableId = tableId;
+      return this;
+    }
+    
+    /**
+     * @param tableName 
+     */
+    public Options tableName(String tableName) {
+      this.tableName = tableName;
+      return this;
+    }
+    
+    /**
+     * @param config 
+     */
+    public Options config(String config) {
+      this.config = config;
+      return this;
+    }
+    
+    private Long tableId;
+    private String tableName;
+    private String config;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new RetrieveTPUEmbeddingCenteredRMSPropParameters operation.
+   * 
+   * @param scope current scope
+   * @param numShards 
+   * @param shardId 
+   * @param options carries optional attributes values
+   * @return a new instance of RetrieveTPUEmbeddingCenteredRMSPropParameters
+   */
+  @Endpoint(describeByClass = true)
+  public static RetrieveTPUEmbeddingCenteredRMSPropParameters create(Scope scope, Long numShards, Long shardId, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("RetrieveTPUEmbeddingCenteredRMSPropParameters", scope.makeOpName("RetrieveTPUEmbeddingCenteredRMSPropParameters"));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("num_shards", numShards);
+    opBuilder.setAttr("shard_id", shardId);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.tableId != null) {
+          opBuilder.setAttr("table_id", opts.tableId);
+        }
+        if (opts.tableName != null) {
+          opBuilder.setAttr("table_name", opts.tableName);
+        }
+        if (opts.config != null) {
+          opBuilder.setAttr("config", opts.config);
+        }
+      }
+    }
+    return new RetrieveTPUEmbeddingCenteredRMSPropParameters(opBuilder.build());
+  }
+  
+  /**
+   * @param tableId 
+   */
+  public static Options tableId(Long tableId) {
+    return new Options().tableId(tableId);
+  }
+  
+  /**
+   * @param tableName 
+   */
+  public static Options tableName(String tableName) {
+    return new Options().tableName(tableName);
+  }
+  
+  /**
+   * @param config 
+   */
+  public static Options config(String config) {
+    return new Options().config(config);
+  }
+  
+  /**
+   * Parameter parameters updated by the centered RMSProp optimization algorithm.
+   */
+  public Output<TFloat32> parameters() {
+    return parameters;
+  }
+  
+  /**
+   * Parameter ms updated by the centered RMSProp optimization algorithm.
+   */
+  public Output<TFloat32> ms() {
+    return ms;
+  }
+  
+  /**
+   * Parameter mom updated by the centered RMSProp optimization algorithm.
+   */
+  public Output<TFloat32> mom() {
+    return mom;
+  }
+  
+  /**
+   * Parameter mg updated by the centered RMSProp optimization algorithm.
+   */
+  public Output<TFloat32> mg() {
+    return mg;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "RetrieveTPUEmbeddingCenteredRMSPropParameters";
+  
+  private Output<TFloat32> parameters;
+  private Output<TFloat32> ms;
+  private Output<TFloat32> mom;
+  private Output<TFloat32> mg;
+  
+  private RetrieveTPUEmbeddingCenteredRMSPropParameters(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    parameters = operation.output(outputIdx++);
+    ms = operation.output(outputIdx++);
+    mom = operation.output(outputIdx++);
+    mg = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingFTRLParameters.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingFTRLParameters.java
new file mode 100644
index 00000000000..0e7114d3744
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingFTRLParameters.java
@@ -0,0 +1,163 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.tpu;
+
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+
+/**
+ * Retrieve FTRL embedding parameters.
+ * <p>
+ * An op that retrieves optimization parameters from embedding to host
+ * memory. Must be preceded by a ConfigureTPUEmbeddingHost op that sets up
+ * the correct embedding table configuration. For example, this op is
+ * used to retrieve updated parameters before saving a checkpoint.
+ */
+public final class RetrieveTPUEmbeddingFTRLParameters extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.tpu.RetrieveTPUEmbeddingFTRLParameters}
+   */
+  public static class Options {
+    
+    /**
+     * @param tableId 
+     */
+    public Options tableId(Long tableId) {
+      this.tableId = tableId;
+      return this;
+    }
+    
+    /**
+     * @param tableName 
+     */
+    public Options tableName(String tableName) {
+      this.tableName = tableName;
+      return this;
+    }
+    
+    /**
+     * @param config 
+     */
+    public Options config(String config) {
+      this.config = config;
+      return this;
+    }
+    
+    private Long tableId;
+    private String tableName;
+    private String config;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new RetrieveTPUEmbeddingFTRLParameters operation.
+   * 
+   * @param scope current scope
+   * @param numShards 
+   * @param shardId 
+   * @param options carries optional attributes values
+   * @return a new instance of RetrieveTPUEmbeddingFTRLParameters
+   */
+  @Endpoint(describeByClass = true)
+  public static RetrieveTPUEmbeddingFTRLParameters create(Scope scope, Long numShards, Long shardId, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("RetrieveTPUEmbeddingFTRLParameters", scope.makeOpName("RetrieveTPUEmbeddingFTRLParameters"));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("num_shards", numShards);
+    opBuilder.setAttr("shard_id", shardId);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.tableId != null) {
+          opBuilder.setAttr("table_id", opts.tableId);
+        }
+        if (opts.tableName != null) {
+          opBuilder.setAttr("table_name", opts.tableName);
+        }
+        if (opts.config != null) {
+          opBuilder.setAttr("config", opts.config);
+        }
+      }
+    }
+    return new RetrieveTPUEmbeddingFTRLParameters(opBuilder.build());
+  }
+  
+  /**
+   * @param tableId 
+   */
+  public static Options tableId(Long tableId) {
+    return new Options().tableId(tableId);
+  }
+  
+  /**
+   * @param tableName 
+   */
+  public static Options tableName(String tableName) {
+    return new Options().tableName(tableName);
+  }
+  
+  /**
+   * @param config 
+   */
+  public static Options config(String config) {
+    return new Options().config(config);
+  }
+  
+  /**
+   * Parameter parameters updated by the FTRL optimization algorithm.
+   */
+  public Output<TFloat32> parameters() {
+    return parameters;
+  }
+  
+  /**
+   * Parameter accumulators updated by the FTRL optimization algorithm.
+   */
+  public Output<TFloat32> accumulators() {
+    return accumulators;
+  }
+  
+  /**
+   * Parameter linears updated by the FTRL optimization algorithm.
+   */
+  public Output<TFloat32> linears() {
+    return linears;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "RetrieveTPUEmbeddingFTRLParameters";
+  
+  private Output<TFloat32> parameters;
+  private Output<TFloat32> accumulators;
+  private Output<TFloat32> linears;
+  
+  private RetrieveTPUEmbeddingFTRLParameters(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    parameters = operation.output(outputIdx++);
+    accumulators = operation.output(outputIdx++);
+    linears = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingFTRLParametersGradAccumDebug.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingFTRLParametersGradAccumDebug.java
new file mode 100644
index 00000000000..2dbff3ea109
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingFTRLParametersGradAccumDebug.java
@@ -0,0 +1,172 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.tpu;
+
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+
+/**
+ * Retrieve FTRL embedding parameters with debug support.
+ * <p>
+ * An op that retrieves optimization parameters from embedding to host
+ * memory. Must be preceded by a ConfigureTPUEmbeddingHost op that sets up
+ * the correct embedding table configuration. For example, this op is
+ * used to retrieve updated parameters before saving a checkpoint.
+ */
+public final class RetrieveTPUEmbeddingFTRLParametersGradAccumDebug extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.tpu.RetrieveTPUEmbeddingFTRLParametersGradAccumDebug}
+   */
+  public static class Options {
+    
+    /**
+     * @param tableId 
+     */
+    public Options tableId(Long tableId) {
+      this.tableId = tableId;
+      return this;
+    }
+    
+    /**
+     * @param tableName 
+     */
+    public Options tableName(String tableName) {
+      this.tableName = tableName;
+      return this;
+    }
+    
+    /**
+     * @param config 
+     */
+    public Options config(String config) {
+      this.config = config;
+      return this;
+    }
+    
+    private Long tableId;
+    private String tableName;
+    private String config;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new RetrieveTPUEmbeddingFTRLParametersGradAccumDebug operation.
+   * 
+   * @param scope current scope
+   * @param numShards 
+   * @param shardId 
+   * @param options carries optional attributes values
+   * @return a new instance of RetrieveTPUEmbeddingFTRLParametersGradAccumDebug
+   */
+  @Endpoint(describeByClass = true)
+  public static RetrieveTPUEmbeddingFTRLParametersGradAccumDebug create(Scope scope, Long numShards, Long shardId, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("RetrieveTPUEmbeddingFTRLParametersGradAccumDebug", scope.makeOpName("RetrieveTPUEmbeddingFTRLParametersGradAccumDebug"));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("num_shards", numShards);
+    opBuilder.setAttr("shard_id", shardId);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.tableId != null) {
+          opBuilder.setAttr("table_id", opts.tableId);
+        }
+        if (opts.tableName != null) {
+          opBuilder.setAttr("table_name", opts.tableName);
+        }
+        if (opts.config != null) {
+          opBuilder.setAttr("config", opts.config);
+        }
+      }
+    }
+    return new RetrieveTPUEmbeddingFTRLParametersGradAccumDebug(opBuilder.build());
+  }
+  
+  /**
+   * @param tableId 
+   */
+  public static Options tableId(Long tableId) {
+    return new Options().tableId(tableId);
+  }
+  
+  /**
+   * @param tableName 
+   */
+  public static Options tableName(String tableName) {
+    return new Options().tableName(tableName);
+  }
+  
+  /**
+   * @param config 
+   */
+  public static Options config(String config) {
+    return new Options().config(config);
+  }
+  
+  /**
+   * Parameter parameters updated by the FTRL optimization algorithm.
+   */
+  public Output<TFloat32> parameters() {
+    return parameters;
+  }
+  
+  /**
+   * Parameter accumulators updated by the FTRL optimization algorithm.
+   */
+  public Output<TFloat32> accumulators() {
+    return accumulators;
+  }
+  
+  /**
+   * Parameter linears updated by the FTRL optimization algorithm.
+   */
+  public Output<TFloat32> linears() {
+    return linears;
+  }
+  
+  /**
+   * Parameter gradient_accumulators updated by the FTRL optimization algorithm.
+   */
+  public Output<TFloat32> gradientAccumulators() {
+    return gradientAccumulators;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "RetrieveTPUEmbeddingFTRLParametersGradAccumDebug";
+  
+  private Output<TFloat32> parameters;
+  private Output<TFloat32> accumulators;
+  private Output<TFloat32> linears;
+  private Output<TFloat32> gradientAccumulators;
+  
+  private RetrieveTPUEmbeddingFTRLParametersGradAccumDebug(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    parameters = operation.output(outputIdx++);
+    accumulators = operation.output(outputIdx++);
+    linears = operation.output(outputIdx++);
+    gradientAccumulators = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingMDLAdagradLightParameters.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingMDLAdagradLightParameters.java
new file mode 100644
index 00000000000..aa4be2cb318
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingMDLAdagradLightParameters.java
@@ -0,0 +1,172 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.tpu;
+
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+
+/**
+ * Retrieve MDL Adagrad Light embedding parameters.
+ * <p>
+ * An op that retrieves optimization parameters from embedding to host
+ * memory. Must be preceded by a ConfigureTPUEmbeddingHost op that sets up
+ * the correct embedding table configuration. For example, this op is
+ * used to retrieve updated parameters before saving a checkpoint.
+ */
+public final class RetrieveTPUEmbeddingMDLAdagradLightParameters extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.tpu.RetrieveTPUEmbeddingMDLAdagradLightParameters}
+   */
+  public static class Options {
+    
+    /**
+     * @param tableId 
+     */
+    public Options tableId(Long tableId) {
+      this.tableId = tableId;
+      return this;
+    }
+    
+    /**
+     * @param tableName 
+     */
+    public Options tableName(String tableName) {
+      this.tableName = tableName;
+      return this;
+    }
+    
+    /**
+     * @param config 
+     */
+    public Options config(String config) {
+      this.config = config;
+      return this;
+    }
+    
+    private Long tableId;
+    private String tableName;
+    private String config;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new RetrieveTPUEmbeddingMDLAdagradLightParameters operation.
+   * 
+   * @param scope current scope
+   * @param numShards 
+   * @param shardId 
+   * @param options carries optional attributes values
+   * @return a new instance of RetrieveTPUEmbeddingMDLAdagradLightParameters
+   */
+  @Endpoint(describeByClass = true)
+  public static RetrieveTPUEmbeddingMDLAdagradLightParameters create(Scope scope, Long numShards, Long shardId, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("RetrieveTPUEmbeddingMDLAdagradLightParameters", scope.makeOpName("RetrieveTPUEmbeddingMDLAdagradLightParameters"));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("num_shards", numShards);
+    opBuilder.setAttr("shard_id", shardId);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.tableId != null) {
+          opBuilder.setAttr("table_id", opts.tableId);
+        }
+        if (opts.tableName != null) {
+          opBuilder.setAttr("table_name", opts.tableName);
+        }
+        if (opts.config != null) {
+          opBuilder.setAttr("config", opts.config);
+        }
+      }
+    }
+    return new RetrieveTPUEmbeddingMDLAdagradLightParameters(opBuilder.build());
+  }
+  
+  /**
+   * @param tableId 
+   */
+  public static Options tableId(Long tableId) {
+    return new Options().tableId(tableId);
+  }
+  
+  /**
+   * @param tableName 
+   */
+  public static Options tableName(String tableName) {
+    return new Options().tableName(tableName);
+  }
+  
+  /**
+   * @param config 
+   */
+  public static Options config(String config) {
+    return new Options().config(config);
+  }
+  
+  /**
+   * Parameter parameters updated by the MDL Adagrad Light optimization algorithm.
+   */
+  public Output<TFloat32> parameters() {
+    return parameters;
+  }
+  
+  /**
+   * Parameter accumulators updated by the MDL Adagrad Light optimization algorithm.
+   */
+  public Output<TFloat32> accumulators() {
+    return accumulators;
+  }
+  
+  /**
+   * Parameter weights updated by the MDL Adagrad Light optimization algorithm.
+   */
+  public Output<TFloat32> weights() {
+    return weights;
+  }
+  
+  /**
+   * Parameter benefits updated by the MDL Adagrad Light optimization algorithm.
+   */
+  public Output<TFloat32> benefits() {
+    return benefits;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "RetrieveTPUEmbeddingMDLAdagradLightParameters";
+  
+  private Output<TFloat32> parameters;
+  private Output<TFloat32> accumulators;
+  private Output<TFloat32> weights;
+  private Output<TFloat32> benefits;
+  
+  private RetrieveTPUEmbeddingMDLAdagradLightParameters(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    parameters = operation.output(outputIdx++);
+    accumulators = operation.output(outputIdx++);
+    weights = operation.output(outputIdx++);
+    benefits = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingMomentumParameters.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingMomentumParameters.java
new file mode 100644
index 00000000000..7f397adda1c
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingMomentumParameters.java
@@ -0,0 +1,154 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.tpu;
+
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+
+/**
+ * Retrieve Momentum embedding parameters.
+ * <p>
+ * An op that retrieves optimization parameters from embedding to host
+ * memory. Must be preceded by a ConfigureTPUEmbeddingHost op that sets up
+ * the correct embedding table configuration. For example, this op is
+ * used to retrieve updated parameters before saving a checkpoint.
+ */
+public final class RetrieveTPUEmbeddingMomentumParameters extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.tpu.RetrieveTPUEmbeddingMomentumParameters}
+   */
+  public static class Options {
+    
+    /**
+     * @param tableId 
+     */
+    public Options tableId(Long tableId) {
+      this.tableId = tableId;
+      return this;
+    }
+    
+    /**
+     * @param tableName 
+     */
+    public Options tableName(String tableName) {
+      this.tableName = tableName;
+      return this;
+    }
+    
+    /**
+     * @param config 
+     */
+    public Options config(String config) {
+      this.config = config;
+      return this;
+    }
+    
+    private Long tableId;
+    private String tableName;
+    private String config;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new RetrieveTPUEmbeddingMomentumParameters operation.
+   * 
+   * @param scope current scope
+   * @param numShards 
+   * @param shardId 
+   * @param options carries optional attributes values
+   * @return a new instance of RetrieveTPUEmbeddingMomentumParameters
+   */
+  @Endpoint(describeByClass = true)
+  public static RetrieveTPUEmbeddingMomentumParameters create(Scope scope, Long numShards, Long shardId, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("RetrieveTPUEmbeddingMomentumParameters", scope.makeOpName("RetrieveTPUEmbeddingMomentumParameters"));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("num_shards", numShards);
+    opBuilder.setAttr("shard_id", shardId);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.tableId != null) {
+          opBuilder.setAttr("table_id", opts.tableId);
+        }
+        if (opts.tableName != null) {
+          opBuilder.setAttr("table_name", opts.tableName);
+        }
+        if (opts.config != null) {
+          opBuilder.setAttr("config", opts.config);
+        }
+      }
+    }
+    return new RetrieveTPUEmbeddingMomentumParameters(opBuilder.build());
+  }
+  
+  /**
+   * @param tableId 
+   */
+  public static Options tableId(Long tableId) {
+    return new Options().tableId(tableId);
+  }
+  
+  /**
+   * @param tableName 
+   */
+  public static Options tableName(String tableName) {
+    return new Options().tableName(tableName);
+  }
+  
+  /**
+   * @param config 
+   */
+  public static Options config(String config) {
+    return new Options().config(config);
+  }
+  
+  /**
+   * Parameter parameters updated by the Momentum optimization algorithm.
+   */
+  public Output<TFloat32> parameters() {
+    return parameters;
+  }
+  
+  /**
+   * Parameter momenta updated by the Momentum optimization algorithm.
+   */
+  public Output<TFloat32> momenta() {
+    return momenta;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "RetrieveTPUEmbeddingMomentumParameters";
+  
+  private Output<TFloat32> parameters;
+  private Output<TFloat32> momenta;
+  
+  private RetrieveTPUEmbeddingMomentumParameters(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    parameters = operation.output(outputIdx++);
+    momenta = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingMomentumParametersGradAccumDebug.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingMomentumParametersGradAccumDebug.java
new file mode 100644
index 00000000000..15aceae7f2a
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingMomentumParametersGradAccumDebug.java
@@ -0,0 +1,163 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.tpu;
+
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+
+/**
+ * Retrieve Momentum embedding parameters with debug support.
+ * <p>
+ * An op that retrieves optimization parameters from embedding to host
+ * memory. Must be preceded by a ConfigureTPUEmbeddingHost op that sets up
+ * the correct embedding table configuration. For example, this op is
+ * used to retrieve updated parameters before saving a checkpoint.
+ */
+public final class RetrieveTPUEmbeddingMomentumParametersGradAccumDebug extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.tpu.RetrieveTPUEmbeddingMomentumParametersGradAccumDebug}
+   */
+  public static class Options {
+    
+    /**
+     * @param tableId 
+     */
+    public Options tableId(Long tableId) {
+      this.tableId = tableId;
+      return this;
+    }
+    
+    /**
+     * @param tableName 
+     */
+    public Options tableName(String tableName) {
+      this.tableName = tableName;
+      return this;
+    }
+    
+    /**
+     * @param config 
+     */
+    public Options config(String config) {
+      this.config = config;
+      return this;
+    }
+    
+    private Long tableId;
+    private String tableName;
+    private String config;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new RetrieveTPUEmbeddingMomentumParametersGradAccumDebug operation.
+   * 
+   * @param scope current scope
+   * @param numShards 
+   * @param shardId 
+   * @param options carries optional attributes values
+   * @return a new instance of RetrieveTPUEmbeddingMomentumParametersGradAccumDebug
+   */
+  @Endpoint(describeByClass = true)
+  public static RetrieveTPUEmbeddingMomentumParametersGradAccumDebug create(Scope scope, Long numShards, Long shardId, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("RetrieveTPUEmbeddingMomentumParametersGradAccumDebug", scope.makeOpName("RetrieveTPUEmbeddingMomentumParametersGradAccumDebug"));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("num_shards", numShards);
+    opBuilder.setAttr("shard_id", shardId);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.tableId != null) {
+          opBuilder.setAttr("table_id", opts.tableId);
+        }
+        if (opts.tableName != null) {
+          opBuilder.setAttr("table_name", opts.tableName);
+        }
+        if (opts.config != null) {
+          opBuilder.setAttr("config", opts.config);
+        }
+      }
+    }
+    return new RetrieveTPUEmbeddingMomentumParametersGradAccumDebug(opBuilder.build());
+  }
+  
+  /**
+   * @param tableId 
+   */
+  public static Options tableId(Long tableId) {
+    return new Options().tableId(tableId);
+  }
+  
+  /**
+   * @param tableName 
+   */
+  public static Options tableName(String tableName) {
+    return new Options().tableName(tableName);
+  }
+  
+  /**
+   * @param config 
+   */
+  public static Options config(String config) {
+    return new Options().config(config);
+  }
+  
+  /**
+   * Parameter parameters updated by the Momentum optimization algorithm.
+   */
+  public Output<TFloat32> parameters() {
+    return parameters;
+  }
+  
+  /**
+   * Parameter momenta updated by the Momentum optimization algorithm.
+   */
+  public Output<TFloat32> momenta() {
+    return momenta;
+  }
+  
+  /**
+   * Parameter gradient_accumulators updated by the Momentum optimization algorithm.
+   */
+  public Output<TFloat32> gradientAccumulators() {
+    return gradientAccumulators;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "RetrieveTPUEmbeddingMomentumParametersGradAccumDebug";
+  
+  private Output<TFloat32> parameters;
+  private Output<TFloat32> momenta;
+  private Output<TFloat32> gradientAccumulators;
+  
+  private RetrieveTPUEmbeddingMomentumParametersGradAccumDebug(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    parameters = operation.output(outputIdx++);
+    momenta = operation.output(outputIdx++);
+    gradientAccumulators = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingProximalAdagradParameters.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingProximalAdagradParameters.java
new file mode 100644
index 00000000000..4f20964f27f
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingProximalAdagradParameters.java
@@ -0,0 +1,154 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.tpu;
+
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+
+/**
+ * Retrieve proximal Adagrad embedding parameters.
+ * <p>
+ * An op that retrieves optimization parameters from embedding to host
+ * memory. Must be preceded by a ConfigureTPUEmbeddingHost op that sets up
+ * the correct embedding table configuration. For example, this op is
+ * used to retrieve updated parameters before saving a checkpoint.
+ */
+public final class RetrieveTPUEmbeddingProximalAdagradParameters extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.tpu.RetrieveTPUEmbeddingProximalAdagradParameters}
+   */
+  public static class Options {
+    
+    /**
+     * @param tableId 
+     */
+    public Options tableId(Long tableId) {
+      this.tableId = tableId;
+      return this;
+    }
+    
+    /**
+     * @param tableName 
+     */
+    public Options tableName(String tableName) {
+      this.tableName = tableName;
+      return this;
+    }
+    
+    /**
+     * @param config 
+     */
+    public Options config(String config) {
+      this.config = config;
+      return this;
+    }
+    
+    private Long tableId;
+    private String tableName;
+    private String config;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new RetrieveTPUEmbeddingProximalAdagradParameters operation.
+   * 
+   * @param scope current scope
+   * @param numShards 
+   * @param shardId 
+   * @param options carries optional attributes values
+   * @return a new instance of RetrieveTPUEmbeddingProximalAdagradParameters
+   */
+  @Endpoint(describeByClass = true)
+  public static RetrieveTPUEmbeddingProximalAdagradParameters create(Scope scope, Long numShards, Long shardId, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("RetrieveTPUEmbeddingProximalAdagradParameters", scope.makeOpName("RetrieveTPUEmbeddingProximalAdagradParameters"));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("num_shards", numShards);
+    opBuilder.setAttr("shard_id", shardId);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.tableId != null) {
+          opBuilder.setAttr("table_id", opts.tableId);
+        }
+        if (opts.tableName != null) {
+          opBuilder.setAttr("table_name", opts.tableName);
+        }
+        if (opts.config != null) {
+          opBuilder.setAttr("config", opts.config);
+        }
+      }
+    }
+    return new RetrieveTPUEmbeddingProximalAdagradParameters(opBuilder.build());
+  }
+  
+  /**
+   * @param tableId 
+   */
+  public static Options tableId(Long tableId) {
+    return new Options().tableId(tableId);
+  }
+  
+  /**
+   * @param tableName 
+   */
+  public static Options tableName(String tableName) {
+    return new Options().tableName(tableName);
+  }
+  
+  /**
+   * @param config 
+   */
+  public static Options config(String config) {
+    return new Options().config(config);
+  }
+  
+  /**
+   * Parameter parameters updated by the proximal Adagrad optimization algorithm.
+   */
+  public Output<TFloat32> parameters() {
+    return parameters;
+  }
+  
+  /**
+   * Parameter accumulators updated by the proximal Adagrad optimization algorithm.
+   */
+  public Output<TFloat32> accumulators() {
+    return accumulators;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "RetrieveTPUEmbeddingProximalAdagradParameters";
+  
+  private Output<TFloat32> parameters;
+  private Output<TFloat32> accumulators;
+  
+  private RetrieveTPUEmbeddingProximalAdagradParameters(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    parameters = operation.output(outputIdx++);
+    accumulators = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingProximalAdagradParametersGradAccumDebug.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingProximalAdagradParametersGradAccumDebug.java
new file mode 100644
index 00000000000..abde61ec5a9
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingProximalAdagradParametersGradAccumDebug.java
@@ -0,0 +1,163 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.tpu;
+
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+
+/**
+ * Retrieve proximal Adagrad embedding parameters with debug support.
+ * <p>
+ * An op that retrieves optimization parameters from embedding to host
+ * memory. Must be preceded by a ConfigureTPUEmbeddingHost op that sets up
+ * the correct embedding table configuration. For example, this op is
+ * used to retrieve updated parameters before saving a checkpoint.
+ */
+public final class RetrieveTPUEmbeddingProximalAdagradParametersGradAccumDebug extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.tpu.RetrieveTPUEmbeddingProximalAdagradParametersGradAccumDebug}
+   */
+  public static class Options {
+    
+    /**
+     * @param tableId 
+     */
+    public Options tableId(Long tableId) {
+      this.tableId = tableId;
+      return this;
+    }
+    
+    /**
+     * @param tableName 
+     */
+    public Options tableName(String tableName) {
+      this.tableName = tableName;
+      return this;
+    }
+    
+    /**
+     * @param config 
+     */
+    public Options config(String config) {
+      this.config = config;
+      return this;
+    }
+    
+    private Long tableId;
+    private String tableName;
+    private String config;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new RetrieveTPUEmbeddingProximalAdagradParametersGradAccumDebug operation.
+   * 
+   * @param scope current scope
+   * @param numShards 
+   * @param shardId 
+   * @param options carries optional attributes values
+   * @return a new instance of RetrieveTPUEmbeddingProximalAdagradParametersGradAccumDebug
+   */
+  @Endpoint(describeByClass = true)
+  public static RetrieveTPUEmbeddingProximalAdagradParametersGradAccumDebug create(Scope scope, Long numShards, Long shardId, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("RetrieveTPUEmbeddingProximalAdagradParametersGradAccumDebug", scope.makeOpName("RetrieveTPUEmbeddingProximalAdagradParametersGradAccumDebug"));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("num_shards", numShards);
+    opBuilder.setAttr("shard_id", shardId);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.tableId != null) {
+          opBuilder.setAttr("table_id", opts.tableId);
+        }
+        if (opts.tableName != null) {
+          opBuilder.setAttr("table_name", opts.tableName);
+        }
+        if (opts.config != null) {
+          opBuilder.setAttr("config", opts.config);
+        }
+      }
+    }
+    return new RetrieveTPUEmbeddingProximalAdagradParametersGradAccumDebug(opBuilder.build());
+  }
+  
+  /**
+   * @param tableId 
+   */
+  public static Options tableId(Long tableId) {
+    return new Options().tableId(tableId);
+  }
+  
+  /**
+   * @param tableName 
+   */
+  public static Options tableName(String tableName) {
+    return new Options().tableName(tableName);
+  }
+  
+  /**
+   * @param config 
+   */
+  public static Options config(String config) {
+    return new Options().config(config);
+  }
+  
+  /**
+   * Parameter parameters updated by the proximal Adagrad optimization algorithm.
+   */
+  public Output<TFloat32> parameters() {
+    return parameters;
+  }
+  
+  /**
+   * Parameter accumulators updated by the proximal Adagrad optimization algorithm.
+   */
+  public Output<TFloat32> accumulators() {
+    return accumulators;
+  }
+  
+  /**
+   * Parameter gradient_accumulators updated by the proximal Adagrad optimization algorithm.
+   */
+  public Output<TFloat32> gradientAccumulators() {
+    return gradientAccumulators;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "RetrieveTPUEmbeddingProximalAdagradParametersGradAccumDebug";
+  
+  private Output<TFloat32> parameters;
+  private Output<TFloat32> accumulators;
+  private Output<TFloat32> gradientAccumulators;
+  
+  private RetrieveTPUEmbeddingProximalAdagradParametersGradAccumDebug(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    parameters = operation.output(outputIdx++);
+    accumulators = operation.output(outputIdx++);
+    gradientAccumulators = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingProximalYogiParameters.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingProximalYogiParameters.java
new file mode 100644
index 00000000000..a46cae90359
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingProximalYogiParameters.java
@@ -0,0 +1,154 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.tpu;
+
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+
+/**
+ */
+public final class RetrieveTPUEmbeddingProximalYogiParameters extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.tpu.RetrieveTPUEmbeddingProximalYogiParameters}
+   */
+  public static class Options {
+    
+    /**
+     * @param tableId 
+     */
+    public Options tableId(Long tableId) {
+      this.tableId = tableId;
+      return this;
+    }
+    
+    /**
+     * @param tableName 
+     */
+    public Options tableName(String tableName) {
+      this.tableName = tableName;
+      return this;
+    }
+    
+    /**
+     * @param config 
+     */
+    public Options config(String config) {
+      this.config = config;
+      return this;
+    }
+    
+    private Long tableId;
+    private String tableName;
+    private String config;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new RetrieveTPUEmbeddingProximalYogiParameters operation.
+   * 
+   * @param scope current scope
+   * @param numShards 
+   * @param shardId 
+   * @param options carries optional attributes values
+   * @return a new instance of RetrieveTPUEmbeddingProximalYogiParameters
+   */
+  @Endpoint(describeByClass = true)
+  public static RetrieveTPUEmbeddingProximalYogiParameters create(Scope scope, Long numShards, Long shardId, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("RetrieveTPUEmbeddingProximalYogiParameters", scope.makeOpName("RetrieveTPUEmbeddingProximalYogiParameters"));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("num_shards", numShards);
+    opBuilder.setAttr("shard_id", shardId);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.tableId != null) {
+          opBuilder.setAttr("table_id", opts.tableId);
+        }
+        if (opts.tableName != null) {
+          opBuilder.setAttr("table_name", opts.tableName);
+        }
+        if (opts.config != null) {
+          opBuilder.setAttr("config", opts.config);
+        }
+      }
+    }
+    return new RetrieveTPUEmbeddingProximalYogiParameters(opBuilder.build());
+  }
+  
+  /**
+   * @param tableId 
+   */
+  public static Options tableId(Long tableId) {
+    return new Options().tableId(tableId);
+  }
+  
+  /**
+   * @param tableName 
+   */
+  public static Options tableName(String tableName) {
+    return new Options().tableName(tableName);
+  }
+  
+  /**
+   * @param config 
+   */
+  public static Options config(String config) {
+    return new Options().config(config);
+  }
+  
+  /**
+   */
+  public Output<TFloat32> parameters() {
+    return parameters;
+  }
+  
+  /**
+   */
+  public Output<TFloat32> v() {
+    return v;
+  }
+  
+  /**
+   */
+  public Output<TFloat32> m() {
+    return m;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "RetrieveTPUEmbeddingProximalYogiParameters";
+  
+  private Output<TFloat32> parameters;
+  private Output<TFloat32> v;
+  private Output<TFloat32> m;
+  
+  private RetrieveTPUEmbeddingProximalYogiParameters(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    parameters = operation.output(outputIdx++);
+    v = operation.output(outputIdx++);
+    m = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingProximalYogiParametersGradAccumDebug.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingProximalYogiParametersGradAccumDebug.java
new file mode 100644
index 00000000000..55535a573f6
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingProximalYogiParametersGradAccumDebug.java
@@ -0,0 +1,162 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.tpu;
+
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+
+/**
+ */
+public final class RetrieveTPUEmbeddingProximalYogiParametersGradAccumDebug extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.tpu.RetrieveTPUEmbeddingProximalYogiParametersGradAccumDebug}
+   */
+  public static class Options {
+    
+    /**
+     * @param tableId 
+     */
+    public Options tableId(Long tableId) {
+      this.tableId = tableId;
+      return this;
+    }
+    
+    /**
+     * @param tableName 
+     */
+    public Options tableName(String tableName) {
+      this.tableName = tableName;
+      return this;
+    }
+    
+    /**
+     * @param config 
+     */
+    public Options config(String config) {
+      this.config = config;
+      return this;
+    }
+    
+    private Long tableId;
+    private String tableName;
+    private String config;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new RetrieveTPUEmbeddingProximalYogiParametersGradAccumDebug operation.
+   * 
+   * @param scope current scope
+   * @param numShards 
+   * @param shardId 
+   * @param options carries optional attributes values
+   * @return a new instance of RetrieveTPUEmbeddingProximalYogiParametersGradAccumDebug
+   */
+  @Endpoint(describeByClass = true)
+  public static RetrieveTPUEmbeddingProximalYogiParametersGradAccumDebug create(Scope scope, Long numShards, Long shardId, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("RetrieveTPUEmbeddingProximalYogiParametersGradAccumDebug", scope.makeOpName("RetrieveTPUEmbeddingProximalYogiParametersGradAccumDebug"));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("num_shards", numShards);
+    opBuilder.setAttr("shard_id", shardId);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.tableId != null) {
+          opBuilder.setAttr("table_id", opts.tableId);
+        }
+        if (opts.tableName != null) {
+          opBuilder.setAttr("table_name", opts.tableName);
+        }
+        if (opts.config != null) {
+          opBuilder.setAttr("config", opts.config);
+        }
+      }
+    }
+    return new RetrieveTPUEmbeddingProximalYogiParametersGradAccumDebug(opBuilder.build());
+  }
+  
+  /**
+   * @param tableId 
+   */
+  public static Options tableId(Long tableId) {
+    return new Options().tableId(tableId);
+  }
+  
+  /**
+   * @param tableName 
+   */
+  public static Options tableName(String tableName) {
+    return new Options().tableName(tableName);
+  }
+  
+  /**
+   * @param config 
+   */
+  public static Options config(String config) {
+    return new Options().config(config);
+  }
+  
+  /**
+   */
+  public Output<TFloat32> parameters() {
+    return parameters;
+  }
+  
+  /**
+   */
+  public Output<TFloat32> v() {
+    return v;
+  }
+  
+  /**
+   */
+  public Output<TFloat32> m() {
+    return m;
+  }
+  
+  /**
+   */
+  public Output<TFloat32> gradientAccumulators() {
+    return gradientAccumulators;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "RetrieveTPUEmbeddingProximalYogiParametersGradAccumDebug";
+  
+  private Output<TFloat32> parameters;
+  private Output<TFloat32> v;
+  private Output<TFloat32> m;
+  private Output<TFloat32> gradientAccumulators;
+  
+  private RetrieveTPUEmbeddingProximalYogiParametersGradAccumDebug(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    parameters = operation.output(outputIdx++);
+    v = operation.output(outputIdx++);
+    m = operation.output(outputIdx++);
+    gradientAccumulators = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingRMSPropParameters.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingRMSPropParameters.java
new file mode 100644
index 00000000000..fcc01589e38
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingRMSPropParameters.java
@@ -0,0 +1,163 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.tpu;
+
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+
+/**
+ * Retrieve RMSProp embedding parameters.
+ * <p>
+ * An op that retrieves optimization parameters from embedding to host
+ * memory. Must be preceded by a ConfigureTPUEmbeddingHost op that sets up
+ * the correct embedding table configuration. For example, this op is
+ * used to retrieve updated parameters before saving a checkpoint.
+ */
+public final class RetrieveTPUEmbeddingRMSPropParameters extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.tpu.RetrieveTPUEmbeddingRMSPropParameters}
+   */
+  public static class Options {
+    
+    /**
+     * @param tableId 
+     */
+    public Options tableId(Long tableId) {
+      this.tableId = tableId;
+      return this;
+    }
+    
+    /**
+     * @param tableName 
+     */
+    public Options tableName(String tableName) {
+      this.tableName = tableName;
+      return this;
+    }
+    
+    /**
+     * @param config 
+     */
+    public Options config(String config) {
+      this.config = config;
+      return this;
+    }
+    
+    private Long tableId;
+    private String tableName;
+    private String config;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new RetrieveTPUEmbeddingRMSPropParameters operation.
+   * 
+   * @param scope current scope
+   * @param numShards 
+   * @param shardId 
+   * @param options carries optional attributes values
+   * @return a new instance of RetrieveTPUEmbeddingRMSPropParameters
+   */
+  @Endpoint(describeByClass = true)
+  public static RetrieveTPUEmbeddingRMSPropParameters create(Scope scope, Long numShards, Long shardId, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("RetrieveTPUEmbeddingRMSPropParameters", scope.makeOpName("RetrieveTPUEmbeddingRMSPropParameters"));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("num_shards", numShards);
+    opBuilder.setAttr("shard_id", shardId);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.tableId != null) {
+          opBuilder.setAttr("table_id", opts.tableId);
+        }
+        if (opts.tableName != null) {
+          opBuilder.setAttr("table_name", opts.tableName);
+        }
+        if (opts.config != null) {
+          opBuilder.setAttr("config", opts.config);
+        }
+      }
+    }
+    return new RetrieveTPUEmbeddingRMSPropParameters(opBuilder.build());
+  }
+  
+  /**
+   * @param tableId 
+   */
+  public static Options tableId(Long tableId) {
+    return new Options().tableId(tableId);
+  }
+  
+  /**
+   * @param tableName 
+   */
+  public static Options tableName(String tableName) {
+    return new Options().tableName(tableName);
+  }
+  
+  /**
+   * @param config 
+   */
+  public static Options config(String config) {
+    return new Options().config(config);
+  }
+  
+  /**
+   * Parameter parameters updated by the RMSProp optimization algorithm.
+   */
+  public Output<TFloat32> parameters() {
+    return parameters;
+  }
+  
+  /**
+   * Parameter ms updated by the RMSProp optimization algorithm.
+   */
+  public Output<TFloat32> ms() {
+    return ms;
+  }
+  
+  /**
+   * Parameter mom updated by the RMSProp optimization algorithm.
+   */
+  public Output<TFloat32> mom() {
+    return mom;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "RetrieveTPUEmbeddingRMSPropParameters";
+  
+  private Output<TFloat32> parameters;
+  private Output<TFloat32> ms;
+  private Output<TFloat32> mom;
+  
+  private RetrieveTPUEmbeddingRMSPropParameters(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    parameters = operation.output(outputIdx++);
+    ms = operation.output(outputIdx++);
+    mom = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingRMSPropParametersGradAccumDebug.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingRMSPropParametersGradAccumDebug.java
new file mode 100644
index 00000000000..0c3814b23b4
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingRMSPropParametersGradAccumDebug.java
@@ -0,0 +1,172 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.tpu;
+
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+
+/**
+ * Retrieve RMSProp embedding parameters with debug support.
+ * <p>
+ * An op that retrieves optimization parameters from embedding to host
+ * memory. Must be preceded by a ConfigureTPUEmbeddingHost op that sets up
+ * the correct embedding table configuration. For example, this op is
+ * used to retrieve updated parameters before saving a checkpoint.
+ */
+public final class RetrieveTPUEmbeddingRMSPropParametersGradAccumDebug extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.tpu.RetrieveTPUEmbeddingRMSPropParametersGradAccumDebug}
+   */
+  public static class Options {
+    
+    /**
+     * @param tableId 
+     */
+    public Options tableId(Long tableId) {
+      this.tableId = tableId;
+      return this;
+    }
+    
+    /**
+     * @param tableName 
+     */
+    public Options tableName(String tableName) {
+      this.tableName = tableName;
+      return this;
+    }
+    
+    /**
+     * @param config 
+     */
+    public Options config(String config) {
+      this.config = config;
+      return this;
+    }
+    
+    private Long tableId;
+    private String tableName;
+    private String config;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new RetrieveTPUEmbeddingRMSPropParametersGradAccumDebug operation.
+   * 
+   * @param scope current scope
+   * @param numShards 
+   * @param shardId 
+   * @param options carries optional attributes values
+   * @return a new instance of RetrieveTPUEmbeddingRMSPropParametersGradAccumDebug
+   */
+  @Endpoint(describeByClass = true)
+  public static RetrieveTPUEmbeddingRMSPropParametersGradAccumDebug create(Scope scope, Long numShards, Long shardId, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("RetrieveTPUEmbeddingRMSPropParametersGradAccumDebug", scope.makeOpName("RetrieveTPUEmbeddingRMSPropParametersGradAccumDebug"));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("num_shards", numShards);
+    opBuilder.setAttr("shard_id", shardId);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.tableId != null) {
+          opBuilder.setAttr("table_id", opts.tableId);
+        }
+        if (opts.tableName != null) {
+          opBuilder.setAttr("table_name", opts.tableName);
+        }
+        if (opts.config != null) {
+          opBuilder.setAttr("config", opts.config);
+        }
+      }
+    }
+    return new RetrieveTPUEmbeddingRMSPropParametersGradAccumDebug(opBuilder.build());
+  }
+  
+  /**
+   * @param tableId 
+   */
+  public static Options tableId(Long tableId) {
+    return new Options().tableId(tableId);
+  }
+  
+  /**
+   * @param tableName 
+   */
+  public static Options tableName(String tableName) {
+    return new Options().tableName(tableName);
+  }
+  
+  /**
+   * @param config 
+   */
+  public static Options config(String config) {
+    return new Options().config(config);
+  }
+  
+  /**
+   * Parameter parameters updated by the RMSProp optimization algorithm.
+   */
+  public Output<TFloat32> parameters() {
+    return parameters;
+  }
+  
+  /**
+   * Parameter ms updated by the RMSProp optimization algorithm.
+   */
+  public Output<TFloat32> ms() {
+    return ms;
+  }
+  
+  /**
+   * Parameter mom updated by the RMSProp optimization algorithm.
+   */
+  public Output<TFloat32> mom() {
+    return mom;
+  }
+  
+  /**
+   * Parameter gradient_accumulators updated by the RMSProp optimization algorithm.
+   */
+  public Output<TFloat32> gradientAccumulators() {
+    return gradientAccumulators;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "RetrieveTPUEmbeddingRMSPropParametersGradAccumDebug";
+  
+  private Output<TFloat32> parameters;
+  private Output<TFloat32> ms;
+  private Output<TFloat32> mom;
+  private Output<TFloat32> gradientAccumulators;
+  
+  private RetrieveTPUEmbeddingRMSPropParametersGradAccumDebug(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    parameters = operation.output(outputIdx++);
+    ms = operation.output(outputIdx++);
+    mom = operation.output(outputIdx++);
+    gradientAccumulators = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingStochasticGradientDescentParameters.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingStochasticGradientDescentParameters.java
new file mode 100644
index 00000000000..e69247fa0ce
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingStochasticGradientDescentParameters.java
@@ -0,0 +1,151 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.tpu;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+
+/**
+ * Retrieve SGD embedding parameters.
+ * <p>
+ * An op that retrieves optimization parameters from embedding to host
+ * memory. Must be preceded by a ConfigureTPUEmbeddingHost op that sets up
+ * the correct embedding table configuration. For example, this op is
+ * used to retrieve updated parameters before saving a checkpoint.
+ */
+public final class RetrieveTPUEmbeddingStochasticGradientDescentParameters extends RawOp implements Operand<TFloat32> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.tpu.RetrieveTPUEmbeddingStochasticGradientDescentParameters}
+   */
+  public static class Options {
+    
+    /**
+     * @param tableId 
+     */
+    public Options tableId(Long tableId) {
+      this.tableId = tableId;
+      return this;
+    }
+    
+    /**
+     * @param tableName 
+     */
+    public Options tableName(String tableName) {
+      this.tableName = tableName;
+      return this;
+    }
+    
+    /**
+     * @param config 
+     */
+    public Options config(String config) {
+      this.config = config;
+      return this;
+    }
+    
+    private Long tableId;
+    private String tableName;
+    private String config;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new RetrieveTPUEmbeddingStochasticGradientDescentParameters operation.
+   * 
+   * @param scope current scope
+   * @param numShards 
+   * @param shardId 
+   * @param options carries optional attributes values
+   * @return a new instance of RetrieveTPUEmbeddingStochasticGradientDescentParameters
+   */
+  @Endpoint(describeByClass = true)
+  public static RetrieveTPUEmbeddingStochasticGradientDescentParameters create(Scope scope, Long numShards, Long shardId, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("RetrieveTPUEmbeddingStochasticGradientDescentParameters", scope.makeOpName("RetrieveTPUEmbeddingStochasticGradientDescentParameters"));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("num_shards", numShards);
+    opBuilder.setAttr("shard_id", shardId);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.tableId != null) {
+          opBuilder.setAttr("table_id", opts.tableId);
+        }
+        if (opts.tableName != null) {
+          opBuilder.setAttr("table_name", opts.tableName);
+        }
+        if (opts.config != null) {
+          opBuilder.setAttr("config", opts.config);
+        }
+      }
+    }
+    return new RetrieveTPUEmbeddingStochasticGradientDescentParameters(opBuilder.build());
+  }
+  
+  /**
+   * @param tableId 
+   */
+  public static Options tableId(Long tableId) {
+    return new Options().tableId(tableId);
+  }
+  
+  /**
+   * @param tableName 
+   */
+  public static Options tableName(String tableName) {
+    return new Options().tableName(tableName);
+  }
+  
+  /**
+   * @param config 
+   */
+  public static Options config(String config) {
+    return new Options().config(config);
+  }
+  
+  /**
+   * Parameter parameters updated by the stochastic gradient descent optimization algorithm.
+   */
+  public Output<TFloat32> parameters() {
+    return parameters;
+  }
+  
+  @Override
+  public Output<TFloat32> asOutput() {
+    return parameters;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "RetrieveTPUEmbeddingStochasticGradientDescentParameters";
+  
+  private Output<TFloat32> parameters;
+  
+  private RetrieveTPUEmbeddingStochasticGradientDescentParameters(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    parameters = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingStochasticGradientDescentParametersGradAccumDebug.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingStochasticGradientDescentParametersGradAccumDebug.java
new file mode 100644
index 00000000000..9f35ffcd8e0
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/RetrieveTPUEmbeddingStochasticGradientDescentParametersGradAccumDebug.java
@@ -0,0 +1,154 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.tpu;
+
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+
+/**
+ * Retrieve SGD embedding parameters with debug support.
+ * <p>
+ * An op that retrieves optimization parameters from embedding to host
+ * memory. Must be preceded by a ConfigureTPUEmbeddingHost op that sets up
+ * the correct embedding table configuration. For example, this op is
+ * used to retrieve updated parameters before saving a checkpoint.
+ */
+public final class RetrieveTPUEmbeddingStochasticGradientDescentParametersGradAccumDebug extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.tpu.RetrieveTPUEmbeddingStochasticGradientDescentParametersGradAccumDebug}
+   */
+  public static class Options {
+    
+    /**
+     * @param tableId 
+     */
+    public Options tableId(Long tableId) {
+      this.tableId = tableId;
+      return this;
+    }
+    
+    /**
+     * @param tableName 
+     */
+    public Options tableName(String tableName) {
+      this.tableName = tableName;
+      return this;
+    }
+    
+    /**
+     * @param config 
+     */
+    public Options config(String config) {
+      this.config = config;
+      return this;
+    }
+    
+    private Long tableId;
+    private String tableName;
+    private String config;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new RetrieveTPUEmbeddingStochasticGradientDescentParametersGradAccumDebug operation.
+   * 
+   * @param scope current scope
+   * @param numShards 
+   * @param shardId 
+   * @param options carries optional attributes values
+   * @return a new instance of RetrieveTPUEmbeddingStochasticGradientDescentParametersGradAccumDebug
+   */
+  @Endpoint(describeByClass = true)
+  public static RetrieveTPUEmbeddingStochasticGradientDescentParametersGradAccumDebug create(Scope scope, Long numShards, Long shardId, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("RetrieveTPUEmbeddingStochasticGradientDescentParametersGradAccumDebug", scope.makeOpName("RetrieveTPUEmbeddingStochasticGradientDescentParametersGradAccumDebug"));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("num_shards", numShards);
+    opBuilder.setAttr("shard_id", shardId);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.tableId != null) {
+          opBuilder.setAttr("table_id", opts.tableId);
+        }
+        if (opts.tableName != null) {
+          opBuilder.setAttr("table_name", opts.tableName);
+        }
+        if (opts.config != null) {
+          opBuilder.setAttr("config", opts.config);
+        }
+      }
+    }
+    return new RetrieveTPUEmbeddingStochasticGradientDescentParametersGradAccumDebug(opBuilder.build());
+  }
+  
+  /**
+   * @param tableId 
+   */
+  public static Options tableId(Long tableId) {
+    return new Options().tableId(tableId);
+  }
+  
+  /**
+   * @param tableName 
+   */
+  public static Options tableName(String tableName) {
+    return new Options().tableName(tableName);
+  }
+  
+  /**
+   * @param config 
+   */
+  public static Options config(String config) {
+    return new Options().config(config);
+  }
+  
+  /**
+   * Parameter parameters updated by the stochastic gradient descent optimization algorithm.
+   */
+  public Output<TFloat32> parameters() {
+    return parameters;
+  }
+  
+  /**
+   * Parameter gradient_accumulators updated by the Adadelta optimization algorithm.
+   */
+  public Output<TFloat32> gradientAccumulators() {
+    return gradientAccumulators;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "RetrieveTPUEmbeddingStochasticGradientDescentParametersGradAccumDebug";
+  
+  private Output<TFloat32> parameters;
+  private Output<TFloat32> gradientAccumulators;
+  
+  private RetrieveTPUEmbeddingStochasticGradientDescentParametersGradAccumDebug(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    parameters = operation.output(outputIdx++);
+    gradientAccumulators = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/SendTPUEmbeddingGradients.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/SendTPUEmbeddingGradients.java
new file mode 100644
index 00000000000..482080bde5d
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/SendTPUEmbeddingGradients.java
@@ -0,0 +1,70 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.tpu;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.Operands;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+
+/**
+ * Performs gradient updates of embedding tables.
+ */
+public final class SendTPUEmbeddingGradients extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new SendTPUEmbeddingGradients operation.
+   * 
+   * @param scope current scope
+   * @param inputs A TensorList of gradients with which to update embedding tables.
+   * This argument has the same length and shapes as the return value of
+   * RecvTPUEmbeddingActivations, but contains gradients of the model's loss
+   * with respect to the embedding activations. The embedding tables are updated
+   * from these gradients via the optimizer specified in the TPU embedding
+   * configuration given to tpu.initialize_system.
+   * @param learningRates A TensorList of float32 scalars, one for each dynamic learning
+   * rate tag: see the comments in
+   * //third_party/tensorflow/core/protobuf/tpu/optimization_parameters.proto.
+   * Multiple tables can share the same dynamic learning rate tag as specified
+   * in the configuration. If the learning rates for all tables are constant,
+   * this list should be empty.
+   * @param config Serialized TPUEmbeddingConfiguration proto.
+   * @return a new instance of SendTPUEmbeddingGradients
+   */
+  @Endpoint(describeByClass = true)
+  public static SendTPUEmbeddingGradients create(Scope scope, Iterable<Operand<TFloat32>> inputs, Iterable<Operand<TFloat32>> learningRates, String config) {
+    OperationBuilder opBuilder = scope.env().opBuilder("SendTPUEmbeddingGradients", scope.makeOpName("SendTPUEmbeddingGradients"));
+    opBuilder.addInputList(Operands.asOutputs(inputs));
+    opBuilder.addInputList(Operands.asOutputs(learningRates));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("config", config);
+    return new SendTPUEmbeddingGradients(opBuilder.build());
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "SendTPUEmbeddingGradients";
+  
+  private SendTPUEmbeddingGradients(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/ShutdownDistributedTPU.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/ShutdownDistributedTPU.java
new file mode 100644
index 00000000000..9ce924d153a
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/ShutdownDistributedTPU.java
@@ -0,0 +1,53 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.tpu;
+
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+
+/**
+ * Shuts down a running distributed TPU system.
+ * <p>
+ * The op returns an error if no system is running.
+ */
+public final class ShutdownDistributedTPU extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new ShutdownDistributedTPU operation.
+   * 
+   * @param scope current scope
+   * @return a new instance of ShutdownDistributedTPU
+   */
+  @Endpoint(describeByClass = true)
+  public static ShutdownDistributedTPU create(Scope scope) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ShutdownDistributedTPU", scope.makeOpName("ShutdownDistributedTPU"));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new ShutdownDistributedTPU(opBuilder.build());
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ShutdownDistributedTPU";
+  
+  private ShutdownDistributedTPU(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/TPUCompilationResult.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/TPUCompilationResult.java
new file mode 100644
index 00000000000..45280fbc0aa
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/TPUCompilationResult.java
@@ -0,0 +1,73 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.tpu;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TString;
+
+/**
+ * Returns the result of a TPU compilation.
+ * <p>
+ * This operation returns the result of a TPU compilation as a serialized
+ * CompilationResultProto, which holds a status and an error message if an error
+ * occurred during compilation.
+ */
+public final class TPUCompilationResult extends RawOp implements Operand<TString> {
+  
+  /**
+   * Factory method to create a class wrapping a new TPUCompilationResult operation.
+   * 
+   * @param scope current scope
+   * @return a new instance of TPUCompilationResult
+   */
+  @Endpoint(describeByClass = true)
+  public static TPUCompilationResult create(Scope scope) {
+    OperationBuilder opBuilder = scope.env().opBuilder("TPUCompilationResult", scope.makeOpName("TPUCompilationResult"));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new TPUCompilationResult(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<TString> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<TString> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "TPUCompilationResult";
+  
+  private Output<TString> output;
+  
+  private TPUCompilationResult(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/TPUEmbeddingActivations.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/TPUEmbeddingActivations.java
new file mode 100644
index 00000000000..138a7434106
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/TPUEmbeddingActivations.java
@@ -0,0 +1,85 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.tpu;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+
+/**
+ * An op enabling differentiation of TPU Embeddings.
+ * <p>
+ * This op simply returns its first input, which is assumed to have been sliced
+ * from the Tensors returned by TPUEmbeddingDequeueActivations. The presence of
+ * this op, and its first argument being a trainable Variable, enables automatic
+ * differentiation of graphs containing embeddings via the TPU Embedding Python
+ * libraries.
+ */
+public final class TPUEmbeddingActivations extends RawOp implements Operand<TFloat32> {
+  
+  /**
+   * Factory method to create a class wrapping a new TPUEmbeddingActivations operation.
+   * 
+   * @param scope current scope
+   * @param embeddingVariable A trainable variable, enabling optimizers to find this op.
+   * @param slicedActivations The embedding activations Tensor to return.
+   * @param tableId The id of the table in the embedding layer configuration from which
+   * these activations were computed.
+   * @param lookupId Identifier of the set of embedding indices which produced these
+   * activations.
+   * @return a new instance of TPUEmbeddingActivations
+   */
+  @Endpoint(describeByClass = true)
+  public static TPUEmbeddingActivations create(Scope scope, Operand<TFloat32> embeddingVariable, Operand<TFloat32> slicedActivations, Long tableId, Long lookupId) {
+    OperationBuilder opBuilder = scope.env().opBuilder("TPUEmbeddingActivations", scope.makeOpName("TPUEmbeddingActivations"));
+    opBuilder.addInput(embeddingVariable.asOutput());
+    opBuilder.addInput(slicedActivations.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("table_id", tableId);
+    opBuilder.setAttr("lookup_id", lookupId);
+    return new TPUEmbeddingActivations(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<TFloat32> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<TFloat32> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "TPUEmbeddingActivations";
+  
+  private Output<TFloat32> output;
+  
+  private TPUEmbeddingActivations(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/TPUOrdinalSelector.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/TPUOrdinalSelector.java
new file mode 100644
index 00000000000..2df183cb793
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/TPUOrdinalSelector.java
@@ -0,0 +1,74 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.tpu;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+
+/**
+ * A TPU core selector Op.
+ * <p>
+ * This Op produces a set of TPU cores (for warm-up) or a single TPU core
+ * (for regular inference) to execute the TPU program on. The output is
+ * consumed by TPUPartitionedCall.
+ */
+public final class TPUOrdinalSelector extends RawOp implements Operand<TInt32> {
+  
+  /**
+   * Factory method to create a class wrapping a new TPUOrdinalSelector operation.
+   * 
+   * @param scope current scope
+   * @return a new instance of TPUOrdinalSelector
+   */
+  @Endpoint(describeByClass = true)
+  public static TPUOrdinalSelector create(Scope scope) {
+    OperationBuilder opBuilder = scope.env().opBuilder("TPUOrdinalSelector", scope.makeOpName("TPUOrdinalSelector"));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new TPUOrdinalSelector(opBuilder.build());
+  }
+  
+  /**
+   * A vector 1 or more TPU cores.
+   */
+  public Output<TInt32> deviceOrdinals() {
+    return deviceOrdinals;
+  }
+  
+  @Override
+  public Output<TInt32> asOutput() {
+    return deviceOrdinals;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "TPUOrdinalSelector";
+  
+  private Output<TInt32> deviceOrdinals;
+  
+  private TPUOrdinalSelector(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    deviceOrdinals = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/TPUReplicateMetadata.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/TPUReplicateMetadata.java
new file mode 100644
index 00000000000..3fb9b782f9c
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/TPUReplicateMetadata.java
@@ -0,0 +1,258 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.tpu;
+
+import java.util.List;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+
+/**
+ * Metadata indicating how the TPU computation should be replicated.
+ * <p>
+ * This operation holds the metadata common to operations of a `tpu.replicate()` computation subgraph.
+ */
+public final class TPUReplicateMetadata extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.tpu.TPUReplicateMetadata}
+   */
+  public static class Options {
+    
+    /**
+     * @param numCoresPerReplica Number of cores per replica. Used for model parallelism.
+     */
+    public Options numCoresPerReplica(Long numCoresPerReplica) {
+      this.numCoresPerReplica = numCoresPerReplica;
+      return this;
+    }
+    
+    /**
+     * @param topology TopologyProto indicating the topology of the TPU pod slice.
+     */
+    public Options topology(String topology) {
+      this.topology = topology;
+      return this;
+    }
+    
+    /**
+     * @param useTpu Whether to place the computation on the TPU.
+     */
+    public Options useTpu(Boolean useTpu) {
+      this.useTpu = useTpu;
+      return this;
+    }
+    
+    /**
+     * @param deviceAssignment The assignment of devices for the computation.
+     */
+    public Options deviceAssignment(List<Long> deviceAssignment) {
+      this.deviceAssignment = deviceAssignment;
+      return this;
+    }
+    
+    /**
+     * @param computationShape DEPRECATED. Use num_cores_per_replica instead.
+     */
+    public Options computationShape(List<Long> computationShape) {
+      this.computationShape = computationShape;
+      return this;
+    }
+    
+    /**
+     * @param hostComputeCore 
+     */
+    public Options hostComputeCore(List<String> hostComputeCore) {
+      this.hostComputeCore = hostComputeCore;
+      return this;
+    }
+    
+    /**
+     * @param paddingMap 
+     */
+    public Options paddingMap(List<String> paddingMap) {
+      this.paddingMap = paddingMap;
+      return this;
+    }
+    
+    /**
+     * @param stepMarkerLocation 
+     */
+    public Options stepMarkerLocation(String stepMarkerLocation) {
+      this.stepMarkerLocation = stepMarkerLocation;
+      return this;
+    }
+    
+    /**
+     * @param allowSoftPlacement 
+     */
+    public Options allowSoftPlacement(Boolean allowSoftPlacement) {
+      this.allowSoftPlacement = allowSoftPlacement;
+      return this;
+    }
+    
+    private Long numCoresPerReplica;
+    private String topology;
+    private Boolean useTpu;
+    private List<Long> deviceAssignment;
+    private List<Long> computationShape;
+    private List<String> hostComputeCore;
+    private List<String> paddingMap;
+    private String stepMarkerLocation;
+    private Boolean allowSoftPlacement;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new TPUReplicateMetadata operation.
+   * 
+   * @param scope current scope
+   * @param numReplicas Number of replicas of the computation
+   * @param options carries optional attributes values
+   * @return a new instance of TPUReplicateMetadata
+   */
+  @Endpoint(describeByClass = true)
+  public static TPUReplicateMetadata create(Scope scope, Long numReplicas, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("TPUReplicateMetadata", scope.makeOpName("TPUReplicateMetadata"));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("num_replicas", numReplicas);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.numCoresPerReplica != null) {
+          opBuilder.setAttr("num_cores_per_replica", opts.numCoresPerReplica);
+        }
+        if (opts.topology != null) {
+          opBuilder.setAttr("topology", opts.topology);
+        }
+        if (opts.useTpu != null) {
+          opBuilder.setAttr("use_tpu", opts.useTpu);
+        }
+        if (opts.deviceAssignment != null) {
+          long[] deviceAssignmentArray = new long[opts.deviceAssignment.size()];
+          for (int i = 0; i < deviceAssignmentArray.length; ++i) {
+            deviceAssignmentArray[i] = opts.deviceAssignment.get(i);
+          }
+          opBuilder.setAttr("device_assignment", deviceAssignmentArray);
+        }
+        if (opts.computationShape != null) {
+          long[] computationShapeArray = new long[opts.computationShape.size()];
+          for (int i = 0; i < computationShapeArray.length; ++i) {
+            computationShapeArray[i] = opts.computationShape.get(i);
+          }
+          opBuilder.setAttr("computation_shape", computationShapeArray);
+        }
+        if (opts.hostComputeCore != null) {
+          String[] hostComputeCoreArray = new String[opts.hostComputeCore.size()];
+          for (int i = 0; i < hostComputeCoreArray.length; ++i) {
+            hostComputeCoreArray[i] = opts.hostComputeCore.get(i);
+          }
+          opBuilder.setAttr("host_compute_core", hostComputeCoreArray);
+        }
+        if (opts.paddingMap != null) {
+          String[] paddingMapArray = new String[opts.paddingMap.size()];
+          for (int i = 0; i < paddingMapArray.length; ++i) {
+            paddingMapArray[i] = opts.paddingMap.get(i);
+          }
+          opBuilder.setAttr("padding_map", paddingMapArray);
+        }
+        if (opts.stepMarkerLocation != null) {
+          opBuilder.setAttr("step_marker_location", opts.stepMarkerLocation);
+        }
+        if (opts.allowSoftPlacement != null) {
+          opBuilder.setAttr("allow_soft_placement", opts.allowSoftPlacement);
+        }
+      }
+    }
+    return new TPUReplicateMetadata(opBuilder.build());
+  }
+  
+  /**
+   * @param numCoresPerReplica Number of cores per replica. Used for model parallelism.
+   */
+  public static Options numCoresPerReplica(Long numCoresPerReplica) {
+    return new Options().numCoresPerReplica(numCoresPerReplica);
+  }
+  
+  /**
+   * @param topology TopologyProto indicating the topology of the TPU pod slice.
+   */
+  public static Options topology(String topology) {
+    return new Options().topology(topology);
+  }
+  
+  /**
+   * @param useTpu Whether to place the computation on the TPU.
+   */
+  public static Options useTpu(Boolean useTpu) {
+    return new Options().useTpu(useTpu);
+  }
+  
+  /**
+   * @param deviceAssignment The assignment of devices for the computation.
+   */
+  public static Options deviceAssignment(List<Long> deviceAssignment) {
+    return new Options().deviceAssignment(deviceAssignment);
+  }
+  
+  /**
+   * @param computationShape DEPRECATED. Use num_cores_per_replica instead.
+   */
+  public static Options computationShape(List<Long> computationShape) {
+    return new Options().computationShape(computationShape);
+  }
+  
+  /**
+   * @param hostComputeCore 
+   */
+  public static Options hostComputeCore(List<String> hostComputeCore) {
+    return new Options().hostComputeCore(hostComputeCore);
+  }
+  
+  /**
+   * @param paddingMap 
+   */
+  public static Options paddingMap(List<String> paddingMap) {
+    return new Options().paddingMap(paddingMap);
+  }
+  
+  /**
+   * @param stepMarkerLocation 
+   */
+  public static Options stepMarkerLocation(String stepMarkerLocation) {
+    return new Options().stepMarkerLocation(stepMarkerLocation);
+  }
+  
+  /**
+   * @param allowSoftPlacement 
+   */
+  public static Options allowSoftPlacement(Boolean allowSoftPlacement) {
+    return new Options().allowSoftPlacement(allowSoftPlacement);
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "TPUReplicateMetadata";
+  
+  private TPUReplicateMetadata(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/TPUReplicatedInput.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/TPUReplicatedInput.java
new file mode 100644
index 00000000000..51fe43aea06
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/TPUReplicatedInput.java
@@ -0,0 +1,158 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.tpu;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.Operands;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Connects N inputs to an N-way replicated TPU computation.
+ * <p>
+ * This operation holds a replicated input to a `tpu.replicate()` computation subgraph.
+ * Each replicated input has the same shape and type alongside the output.
+ * <p>
+ * For example:
+ * <pre>{@code
+ * %a = "tf.opA"()
+ * %b = "tf.opB"()
+ * %replicated_input = "tf.TPUReplicatedInput"(%a, %b)
+ * %computation = "tf.Computation"(%replicated_input)
+ * }</pre>
+ * The above computation has a replicated input of two replicas.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+public final class TPUReplicatedInput<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.tpu.TPUReplicatedInput}
+   */
+  public static class Options {
+    
+    /**
+     * @param isMirroredVariable 
+     */
+    public Options isMirroredVariable(Boolean isMirroredVariable) {
+      this.isMirroredVariable = isMirroredVariable;
+      return this;
+    }
+    
+    /**
+     * @param index 
+     */
+    public Options index(Long index) {
+      this.index = index;
+      return this;
+    }
+    
+    /**
+     * @param isPacked 
+     */
+    public Options isPacked(Boolean isPacked) {
+      this.isPacked = isPacked;
+      return this;
+    }
+    
+    private Boolean isMirroredVariable;
+    private Long index;
+    private Boolean isPacked;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new TPUReplicatedInput operation.
+   * 
+   * @param scope current scope
+   * @param inputs 
+   * @param options carries optional attributes values
+   * @return a new instance of TPUReplicatedInput
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> TPUReplicatedInput<T> create(Scope scope, Iterable<Operand<T>> inputs, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("TPUReplicatedInput", scope.makeOpName("TPUReplicatedInput"));
+    opBuilder.addInputList(Operands.asOutputs(inputs));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.isMirroredVariable != null) {
+          opBuilder.setAttr("is_mirrored_variable", opts.isMirroredVariable);
+        }
+        if (opts.index != null) {
+          opBuilder.setAttr("index", opts.index);
+        }
+        if (opts.isPacked != null) {
+          opBuilder.setAttr("is_packed", opts.isPacked);
+        }
+      }
+    }
+    return new TPUReplicatedInput<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param isMirroredVariable 
+   */
+  public static Options isMirroredVariable(Boolean isMirroredVariable) {
+    return new Options().isMirroredVariable(isMirroredVariable);
+  }
+  
+  /**
+   * @param index 
+   */
+  public static Options index(Long index) {
+    return new Options().index(index);
+  }
+  
+  /**
+   * @param isPacked 
+   */
+  public static Options isPacked(Boolean isPacked) {
+    return new Options().isPacked(isPacked);
+  }
+  
+  /**
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "TPUReplicatedInput";
+  
+  private Output<T> output;
+  
+  private TPUReplicatedInput(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/TPUReplicatedOutput.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/TPUReplicatedOutput.java
new file mode 100644
index 00000000000..24c9418b0c3
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/TPUReplicatedOutput.java
@@ -0,0 +1,92 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.tpu;
+
+import java.util.Arrays;
+import java.util.Iterator;
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Connects N outputs from an N-way replicated TPU computation.
+ * <p>
+ * This operation holds a replicated output from a `tpu.replicate()` computation subgraph.
+ * Each replicated output has the same shape and type alongside the input.
+ * <p>
+ * For example:
+ * <pre>{@code
+ * %computation = "tf.Computation"()
+ * %replicated_output:2 = "tf.TPUReplicatedOutput"(%computation)
+ * }</pre>
+ * The above computation has a replicated output of two replicas.
+ * 
+ * @param <T> data type for {@code outputs()} output
+ */
+public final class TPUReplicatedOutput<T extends TType> extends RawOp implements Iterable<Operand<T>> {
+  
+  /**
+   * Factory method to create a class wrapping a new TPUReplicatedOutput operation.
+   * 
+   * @param scope current scope
+   * @param input 
+   * @param numReplicas 
+   * @return a new instance of TPUReplicatedOutput
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> TPUReplicatedOutput<T> create(Scope scope, Operand<T> input, Long numReplicas) {
+    OperationBuilder opBuilder = scope.env().opBuilder("TPUReplicatedOutput", scope.makeOpName("TPUReplicatedOutput"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("num_replicas", numReplicas);
+    return new TPUReplicatedOutput<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public List<Output<T>> outputs() {
+    return outputs;
+  }
+  
+  @Override
+  @SuppressWarnings({"rawtypes", "unchecked"})
+  public Iterator<Operand<T>> iterator() {
+    return (Iterator) outputs.iterator();
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "TPUReplicatedOutput";
+  
+  private List<Output<T>> outputs;
+  
+  @SuppressWarnings("unchecked")
+  private TPUReplicatedOutput(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    int outputsLength = operation.outputListLength("outputs");
+    outputs = Arrays.asList((Output<T>[])operation.outputList(outputIdx, outputsLength));
+    outputIdx += outputsLength;
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/WorkerHeartbeat.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/WorkerHeartbeat.java
new file mode 100644
index 00000000000..49895008bd2
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/tpu/WorkerHeartbeat.java
@@ -0,0 +1,75 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.tpu;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TString;
+
+/**
+ * Worker heartbeat op.
+ * <p>
+ * Heartbeats may be sent periodically to indicate the coordinator is still active,
+ * to retrieve the current worker status and to expedite shutdown when necessary.
+ */
+public final class WorkerHeartbeat extends RawOp implements Operand<TString> {
+  
+  /**
+   * Factory method to create a class wrapping a new WorkerHeartbeat operation.
+   * 
+   * @param scope current scope
+   * @param request A string tensor containing a serialized WorkerHeartbeatRequest
+   * @return a new instance of WorkerHeartbeat
+   */
+  @Endpoint(describeByClass = true)
+  public static WorkerHeartbeat create(Scope scope, Operand<TString> request) {
+    OperationBuilder opBuilder = scope.env().opBuilder("WorkerHeartbeat", scope.makeOpName("WorkerHeartbeat"));
+    opBuilder.addInput(request.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new WorkerHeartbeat(opBuilder.build());
+  }
+  
+  /**
+   * A string tensor containing a serialized WorkerHeartbeatResponse
+   */
+  public Output<TString> response() {
+    return response;
+  }
+  
+  @Override
+  public Output<TString> asOutput() {
+    return response;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "WorkerHeartbeat";
+  
+  private Output<TString> response;
+  
+  private WorkerHeartbeat(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    response = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/AccumulatorApplyGradient.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/AccumulatorApplyGradient.java
new file mode 100644
index 00000000000..504acba4c96
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/AccumulatorApplyGradient.java
@@ -0,0 +1,64 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.train;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.TString;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Applies a gradient to a given accumulator.
+ * <p>
+ * Does not add if local_step is lesser than the accumulator's global_step.
+ */
+@Operator(group = "train")
+public final class AccumulatorApplyGradient extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new AccumulatorApplyGradient operation.
+   * 
+   * @param scope current scope
+   * @param handle The handle to a accumulator.
+   * @param localStep The local_step value at which the gradient was computed.
+   * @param gradient A tensor of the gradient to be accumulated.
+   * @return a new instance of AccumulatorApplyGradient
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> AccumulatorApplyGradient create(Scope scope, Operand<TString> handle, Operand<TInt64> localStep, Operand<T> gradient) {
+    OperationBuilder opBuilder = scope.env().opBuilder("AccumulatorApplyGradient", scope.makeOpName("AccumulatorApplyGradient"));
+    opBuilder.addInput(handle.asOutput());
+    opBuilder.addInput(localStep.asOutput());
+    opBuilder.addInput(gradient.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new AccumulatorApplyGradient(opBuilder.build());
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "AccumulatorApplyGradient";
+  
+  private AccumulatorApplyGradient(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/AccumulatorNumAccumulated.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/AccumulatorNumAccumulated.java
new file mode 100644
index 00000000000..9fca915c550
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/AccumulatorNumAccumulated.java
@@ -0,0 +1,74 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.train;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.TString;
+
+/**
+ * Returns the number of gradients aggregated in the given accumulators.
+ */
+@Operator(group = "train")
+public final class AccumulatorNumAccumulated extends RawOp implements Operand<TInt32> {
+  
+  /**
+   * Factory method to create a class wrapping a new AccumulatorNumAccumulated operation.
+   * 
+   * @param scope current scope
+   * @param handle The handle to an accumulator.
+   * @return a new instance of AccumulatorNumAccumulated
+   */
+  @Endpoint(describeByClass = true)
+  public static AccumulatorNumAccumulated create(Scope scope, Operand<TString> handle) {
+    OperationBuilder opBuilder = scope.env().opBuilder("AccumulatorNumAccumulated", scope.makeOpName("AccumulatorNumAccumulated"));
+    opBuilder.addInput(handle.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new AccumulatorNumAccumulated(opBuilder.build());
+  }
+  
+  /**
+   * The number of gradients aggregated in the given accumulator.
+   */
+  public Output<TInt32> numAccumulated() {
+    return numAccumulated;
+  }
+  
+  @Override
+  public Output<TInt32> asOutput() {
+    return numAccumulated;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "AccumulatorNumAccumulated";
+  
+  private Output<TInt32> numAccumulated;
+  
+  private AccumulatorNumAccumulated(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    numAccumulated = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/AccumulatorSetGlobalStep.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/AccumulatorSetGlobalStep.java
new file mode 100644
index 00000000000..9039d3a654d
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/AccumulatorSetGlobalStep.java
@@ -0,0 +1,62 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.train;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.TString;
+
+/**
+ * Updates the accumulator with a new value for global_step.
+ * <p>
+ * Logs warning if the accumulator's value is already higher than
+ * new_global_step.
+ */
+@Operator(group = "train")
+public final class AccumulatorSetGlobalStep extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new AccumulatorSetGlobalStep operation.
+   * 
+   * @param scope current scope
+   * @param handle The handle to an accumulator.
+   * @param newGlobalStep The new global_step value to set.
+   * @return a new instance of AccumulatorSetGlobalStep
+   */
+  @Endpoint(describeByClass = true)
+  public static AccumulatorSetGlobalStep create(Scope scope, Operand<TString> handle, Operand<TInt64> newGlobalStep) {
+    OperationBuilder opBuilder = scope.env().opBuilder("AccumulatorSetGlobalStep", scope.makeOpName("AccumulatorSetGlobalStep"));
+    opBuilder.addInput(handle.asOutput());
+    opBuilder.addInput(newGlobalStep.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new AccumulatorSetGlobalStep(opBuilder.build());
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "AccumulatorSetGlobalStep";
+  
+  private AccumulatorSetGlobalStep(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/AccumulatorTakeGradient.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/AccumulatorTakeGradient.java
new file mode 100644
index 00000000000..c5371e853c6
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/AccumulatorTakeGradient.java
@@ -0,0 +1,88 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.train;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.TString;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Extracts the average gradient in the given ConditionalAccumulator.
+ * <p>
+ * The op blocks until sufficient (i.e., more than num_required)
+ * gradients have been accumulated.  If the accumulator has already
+ * aggregated more than num_required gradients, it returns the average of
+ * the accumulated gradients.  Also automatically increments the recorded
+ * global_step in the accumulator by 1, and resets the aggregate to 0.
+ * 
+ * @param <T> data type for {@code average()} output
+ */
+@Operator(group = "train")
+public final class AccumulatorTakeGradient<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new AccumulatorTakeGradient operation.
+   * 
+   * @param scope current scope
+   * @param handle The handle to an accumulator.
+   * @param numRequired Number of gradients required before we return an aggregate.
+   * @param dtype The data type of accumulated gradients. Needs to correspond to the type
+   * of the accumulator.
+   * @return a new instance of AccumulatorTakeGradient
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> AccumulatorTakeGradient<T> create(Scope scope, Operand<TString> handle, Operand<TInt32> numRequired, Class<T> dtype) {
+    OperationBuilder opBuilder = scope.env().opBuilder("AccumulatorTakeGradient", scope.makeOpName("AccumulatorTakeGradient"));
+    opBuilder.addInput(handle.asOutput());
+    opBuilder.addInput(numRequired.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("dtype", dtype);
+    return new AccumulatorTakeGradient<T>(opBuilder.build());
+  }
+  
+  /**
+   * The average of the accumulated gradients.
+   */
+  public Output<T> average() {
+    return average;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return average;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "AccumulatorTakeGradient";
+  
+  private Output<T> average;
+  
+  private AccumulatorTakeGradient(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    average = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyAdaMax.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyAdaMax.java
new file mode 100644
index 00000000000..1b24ac1fe26
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyAdaMax.java
@@ -0,0 +1,132 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.train;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Update '*var' according to the AdaMax algorithm.
+ * <p>
+ * m_t <- beta1 * m_{t-1} + (1 - beta1) * g
+ * v_t <- max(beta2 * v_{t-1}, abs(g))
+ * variable <- variable - learning_rate / (1 - beta1^t) * m_t / (v_t + epsilon)
+ * 
+ * @param <T> data type for {@code out()} output
+ */
+public final class ApplyAdaMax<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.train.ApplyAdaMax}
+   */
+  public static class Options {
+    
+    /**
+     * @param useLocking If `True`, updating of the var, m, and v tensors will be protected
+     * by a lock; otherwise the behavior is undefined, but may exhibit less
+     * contention.
+     */
+    public Options useLocking(Boolean useLocking) {
+      this.useLocking = useLocking;
+      return this;
+    }
+    
+    private Boolean useLocking;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new ApplyAdaMax operation.
+   * 
+   * @param scope current scope
+   * @param var Should be from a Variable().
+   * @param m Should be from a Variable().
+   * @param v Should be from a Variable().
+   * @param beta1Power Must be a scalar.
+   * @param lr Scaling factor. Must be a scalar.
+   * @param beta1 Momentum factor. Must be a scalar.
+   * @param beta2 Momentum factor. Must be a scalar.
+   * @param epsilon Ridge term. Must be a scalar.
+   * @param grad The gradient.
+   * @param options carries optional attributes values
+   * @return a new instance of ApplyAdaMax
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> ApplyAdaMax<T> create(Scope scope, Operand<T> var, Operand<T> m, Operand<T> v, Operand<T> beta1Power, Operand<T> lr, Operand<T> beta1, Operand<T> beta2, Operand<T> epsilon, Operand<T> grad, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ApplyAdaMax", scope.makeOpName("ApplyAdaMax"));
+    opBuilder.addInput(var.asOutput());
+    opBuilder.addInput(m.asOutput());
+    opBuilder.addInput(v.asOutput());
+    opBuilder.addInput(beta1Power.asOutput());
+    opBuilder.addInput(lr.asOutput());
+    opBuilder.addInput(beta1.asOutput());
+    opBuilder.addInput(beta2.asOutput());
+    opBuilder.addInput(epsilon.asOutput());
+    opBuilder.addInput(grad.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.useLocking != null) {
+          opBuilder.setAttr("use_locking", opts.useLocking);
+        }
+      }
+    }
+    return new ApplyAdaMax<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param useLocking If `True`, updating of the var, m, and v tensors will be protected
+   * by a lock; otherwise the behavior is undefined, but may exhibit less
+   * contention.
+   */
+  public static Options useLocking(Boolean useLocking) {
+    return new Options().useLocking(useLocking);
+  }
+  
+  /**
+   * Same as "var".
+   */
+  public Output<T> out() {
+    return out;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return out;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ApplyAdaMax";
+  
+  private Output<T> out;
+  
+  private ApplyAdaMax(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    out = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyAdadelta.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyAdadelta.java
new file mode 100644
index 00000000000..7e68618da7d
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyAdadelta.java
@@ -0,0 +1,128 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.train;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Update '*var' according to the adadelta scheme.
+ * <p>
+ * accum = rho() * accum + (1 - rho()) * grad.square();
+ * update = (update_accum + epsilon).sqrt() * (accum + epsilon()).rsqrt() * grad;
+ * update_accum = rho() * update_accum + (1 - rho()) * update.square();
+ * var -= update;
+ * 
+ * @param <T> data type for {@code out()} output
+ */
+@Operator(group = "train")
+public final class ApplyAdadelta<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.train.ApplyAdadelta}
+   */
+  public static class Options {
+    
+    /**
+     * @param useLocking If True, updating of the var, accum and update_accum tensors will be protected by
+     * a lock; otherwise the behavior is undefined, but may exhibit less contention.
+     */
+    public Options useLocking(Boolean useLocking) {
+      this.useLocking = useLocking;
+      return this;
+    }
+    
+    private Boolean useLocking;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new ApplyAdadelta operation.
+   * 
+   * @param scope current scope
+   * @param var Should be from a Variable().
+   * @param accum Should be from a Variable().
+   * @param accumUpdate Should be from a Variable().
+   * @param lr Scaling factor. Must be a scalar.
+   * @param rho Decay factor. Must be a scalar.
+   * @param epsilon Constant factor. Must be a scalar.
+   * @param grad The gradient.
+   * @param options carries optional attributes values
+   * @return a new instance of ApplyAdadelta
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> ApplyAdadelta<T> create(Scope scope, Operand<T> var, Operand<T> accum, Operand<T> accumUpdate, Operand<T> lr, Operand<T> rho, Operand<T> epsilon, Operand<T> grad, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ApplyAdadelta", scope.makeOpName("ApplyAdadelta"));
+    opBuilder.addInput(var.asOutput());
+    opBuilder.addInput(accum.asOutput());
+    opBuilder.addInput(accumUpdate.asOutput());
+    opBuilder.addInput(lr.asOutput());
+    opBuilder.addInput(rho.asOutput());
+    opBuilder.addInput(epsilon.asOutput());
+    opBuilder.addInput(grad.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.useLocking != null) {
+          opBuilder.setAttr("use_locking", opts.useLocking);
+        }
+      }
+    }
+    return new ApplyAdadelta<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param useLocking If True, updating of the var, accum and update_accum tensors will be protected by
+   * a lock; otherwise the behavior is undefined, but may exhibit less contention.
+   */
+  public static Options useLocking(Boolean useLocking) {
+    return new Options().useLocking(useLocking);
+  }
+  
+  /**
+   * Same as "var".
+   */
+  public Output<T> out() {
+    return out;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return out;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ApplyAdadelta";
+  
+  private Output<T> out;
+  
+  private ApplyAdadelta(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    out = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyAdagrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyAdagrad.java
new file mode 100644
index 00000000000..c8311ac541a
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyAdagrad.java
@@ -0,0 +1,141 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.train;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Update '*var' according to the adagrad scheme.
+ * <p>
+ * accum += grad * grad
+ * var -= lr * grad * (1 / sqrt(accum))
+ * 
+ * @param <T> data type for {@code out()} output
+ */
+@Operator(group = "train")
+public final class ApplyAdagrad<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.train.ApplyAdagrad}
+   */
+  public static class Options {
+    
+    /**
+     * @param useLocking If `True`, updating of the var and accum tensors will be protected
+     * by a lock; otherwise the behavior is undefined, but may exhibit less
+     * contention.
+     */
+    public Options useLocking(Boolean useLocking) {
+      this.useLocking = useLocking;
+      return this;
+    }
+    
+    /**
+     * @param updateSlots 
+     */
+    public Options updateSlots(Boolean updateSlots) {
+      this.updateSlots = updateSlots;
+      return this;
+    }
+    
+    private Boolean useLocking;
+    private Boolean updateSlots;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new ApplyAdagrad operation.
+   * 
+   * @param scope current scope
+   * @param var Should be from a Variable().
+   * @param accum Should be from a Variable().
+   * @param lr Scaling factor. Must be a scalar.
+   * @param grad The gradient.
+   * @param options carries optional attributes values
+   * @return a new instance of ApplyAdagrad
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> ApplyAdagrad<T> create(Scope scope, Operand<T> var, Operand<T> accum, Operand<T> lr, Operand<T> grad, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ApplyAdagrad", scope.makeOpName("ApplyAdagrad"));
+    opBuilder.addInput(var.asOutput());
+    opBuilder.addInput(accum.asOutput());
+    opBuilder.addInput(lr.asOutput());
+    opBuilder.addInput(grad.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.useLocking != null) {
+          opBuilder.setAttr("use_locking", opts.useLocking);
+        }
+        if (opts.updateSlots != null) {
+          opBuilder.setAttr("update_slots", opts.updateSlots);
+        }
+      }
+    }
+    return new ApplyAdagrad<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param useLocking If `True`, updating of the var and accum tensors will be protected
+   * by a lock; otherwise the behavior is undefined, but may exhibit less
+   * contention.
+   */
+  public static Options useLocking(Boolean useLocking) {
+    return new Options().useLocking(useLocking);
+  }
+  
+  /**
+   * @param updateSlots 
+   */
+  public static Options updateSlots(Boolean updateSlots) {
+    return new Options().updateSlots(updateSlots);
+  }
+  
+  /**
+   * Same as "var".
+   */
+  public Output<T> out() {
+    return out;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return out;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ApplyAdagrad";
+  
+  private Output<T> out;
+  
+  private ApplyAdagrad(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    out = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyAdagradDa.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyAdagradDa.java
new file mode 100644
index 00000000000..3bb9f9cae4b
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyAdagradDa.java
@@ -0,0 +1,126 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.train;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Update '*var' according to the proximal adagrad scheme.
+ * 
+ * @param <T> data type for {@code out()} output
+ */
+@Operator(group = "train")
+public final class ApplyAdagradDa<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.train.ApplyAdagradDa}
+   */
+  public static class Options {
+    
+    /**
+     * @param useLocking If True, updating of the var and accum tensors will be protected by
+     * a lock; otherwise the behavior is undefined, but may exhibit less contention.
+     */
+    public Options useLocking(Boolean useLocking) {
+      this.useLocking = useLocking;
+      return this;
+    }
+    
+    private Boolean useLocking;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new ApplyAdagradDa operation.
+   * 
+   * @param scope current scope
+   * @param var Should be from a Variable().
+   * @param gradientAccumulator Should be from a Variable().
+   * @param gradientSquaredAccumulator Should be from a Variable().
+   * @param grad The gradient.
+   * @param lr Scaling factor. Must be a scalar.
+   * @param l1 L1 regularization. Must be a scalar.
+   * @param l2 L2 regularization. Must be a scalar.
+   * @param globalStep Training step number. Must be a scalar.
+   * @param options carries optional attributes values
+   * @return a new instance of ApplyAdagradDa
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> ApplyAdagradDa<T> create(Scope scope, Operand<T> var, Operand<T> gradientAccumulator, Operand<T> gradientSquaredAccumulator, Operand<T> grad, Operand<T> lr, Operand<T> l1, Operand<T> l2, Operand<TInt64> globalStep, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ApplyAdagradDA", scope.makeOpName("ApplyAdagradDa"));
+    opBuilder.addInput(var.asOutput());
+    opBuilder.addInput(gradientAccumulator.asOutput());
+    opBuilder.addInput(gradientSquaredAccumulator.asOutput());
+    opBuilder.addInput(grad.asOutput());
+    opBuilder.addInput(lr.asOutput());
+    opBuilder.addInput(l1.asOutput());
+    opBuilder.addInput(l2.asOutput());
+    opBuilder.addInput(globalStep.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.useLocking != null) {
+          opBuilder.setAttr("use_locking", opts.useLocking);
+        }
+      }
+    }
+    return new ApplyAdagradDa<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param useLocking If True, updating of the var and accum tensors will be protected by
+   * a lock; otherwise the behavior is undefined, but may exhibit less contention.
+   */
+  public static Options useLocking(Boolean useLocking) {
+    return new Options().useLocking(useLocking);
+  }
+  
+  /**
+   * Same as "var".
+   */
+  public Output<T> out() {
+    return out;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return out;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ApplyAdagradDA";
+  
+  private Output<T> out;
+  
+  private ApplyAdagradDa(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    out = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyAdagradV2.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyAdagradV2.java
new file mode 100644
index 00000000000..da399e31eda
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyAdagradV2.java
@@ -0,0 +1,142 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.train;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Update '*var' according to the adagrad scheme.
+ * <p>
+ * accum += grad * grad
+ * var -= lr * grad * (1 / sqrt(accum))
+ * 
+ * @param <T> data type for {@code out()} output
+ */
+public final class ApplyAdagradV2<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.train.ApplyAdagradV2}
+   */
+  public static class Options {
+    
+    /**
+     * @param useLocking If `True`, updating of the var and accum tensors will be protected
+     * by a lock; otherwise the behavior is undefined, but may exhibit less
+     * contention.
+     */
+    public Options useLocking(Boolean useLocking) {
+      this.useLocking = useLocking;
+      return this;
+    }
+    
+    /**
+     * @param updateSlots 
+     */
+    public Options updateSlots(Boolean updateSlots) {
+      this.updateSlots = updateSlots;
+      return this;
+    }
+    
+    private Boolean useLocking;
+    private Boolean updateSlots;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new ApplyAdagradV2 operation.
+   * 
+   * @param scope current scope
+   * @param var Should be from a Variable().
+   * @param accum Should be from a Variable().
+   * @param lr Scaling factor. Must be a scalar.
+   * @param epsilon Constant factor. Must be a scalar.
+   * @param grad The gradient.
+   * @param options carries optional attributes values
+   * @return a new instance of ApplyAdagradV2
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> ApplyAdagradV2<T> create(Scope scope, Operand<T> var, Operand<T> accum, Operand<T> lr, Operand<T> epsilon, Operand<T> grad, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ApplyAdagradV2", scope.makeOpName("ApplyAdagradV2"));
+    opBuilder.addInput(var.asOutput());
+    opBuilder.addInput(accum.asOutput());
+    opBuilder.addInput(lr.asOutput());
+    opBuilder.addInput(epsilon.asOutput());
+    opBuilder.addInput(grad.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.useLocking != null) {
+          opBuilder.setAttr("use_locking", opts.useLocking);
+        }
+        if (opts.updateSlots != null) {
+          opBuilder.setAttr("update_slots", opts.updateSlots);
+        }
+      }
+    }
+    return new ApplyAdagradV2<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param useLocking If `True`, updating of the var and accum tensors will be protected
+   * by a lock; otherwise the behavior is undefined, but may exhibit less
+   * contention.
+   */
+  public static Options useLocking(Boolean useLocking) {
+    return new Options().useLocking(useLocking);
+  }
+  
+  /**
+   * @param updateSlots 
+   */
+  public static Options updateSlots(Boolean updateSlots) {
+    return new Options().updateSlots(updateSlots);
+  }
+  
+  /**
+   * Same as "var".
+   */
+  public Output<T> out() {
+    return out;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return out;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ApplyAdagradV2";
+  
+  private Output<T> out;
+  
+  private ApplyAdagradV2(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    out = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyAdam.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyAdam.java
new file mode 100644
index 00000000000..31d98fa299b
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyAdam.java
@@ -0,0 +1,155 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.train;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Update '*var' according to the Adam algorithm.
+ * <p>
+ * $$lr_t := \text{learning\_rate} * \sqrt{1 - beta_2^t} / (1 - beta_1^t)$$
+ * $$m_t := beta_1 * m_{t-1} + (1 - beta_1) * g$$
+ * $$v_t := beta_2 * v_{t-1} + (1 - beta_2) * g * g$$
+ * $$variable := variable - lr_t * m_t / (\sqrt{v_t} + \epsilon)$$
+ * 
+ * @param <T> data type for {@code out()} output
+ */
+@Operator(group = "train")
+public final class ApplyAdam<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.train.ApplyAdam}
+   */
+  public static class Options {
+    
+    /**
+     * @param useLocking If `True`, updating of the var, m, and v tensors will be protected
+     * by a lock; otherwise the behavior is undefined, but may exhibit less
+     * contention.
+     */
+    public Options useLocking(Boolean useLocking) {
+      this.useLocking = useLocking;
+      return this;
+    }
+    
+    /**
+     * @param useNesterov If `True`, uses the nesterov update.
+     */
+    public Options useNesterov(Boolean useNesterov) {
+      this.useNesterov = useNesterov;
+      return this;
+    }
+    
+    private Boolean useLocking;
+    private Boolean useNesterov;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new ApplyAdam operation.
+   * 
+   * @param scope current scope
+   * @param var Should be from a Variable().
+   * @param m Should be from a Variable().
+   * @param v Should be from a Variable().
+   * @param beta1Power Must be a scalar.
+   * @param beta2Power Must be a scalar.
+   * @param lr Scaling factor. Must be a scalar.
+   * @param beta1 Momentum factor. Must be a scalar.
+   * @param beta2 Momentum factor. Must be a scalar.
+   * @param epsilon Ridge term. Must be a scalar.
+   * @param grad The gradient.
+   * @param options carries optional attributes values
+   * @return a new instance of ApplyAdam
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> ApplyAdam<T> create(Scope scope, Operand<T> var, Operand<T> m, Operand<T> v, Operand<T> beta1Power, Operand<T> beta2Power, Operand<T> lr, Operand<T> beta1, Operand<T> beta2, Operand<T> epsilon, Operand<T> grad, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ApplyAdam", scope.makeOpName("ApplyAdam"));
+    opBuilder.addInput(var.asOutput());
+    opBuilder.addInput(m.asOutput());
+    opBuilder.addInput(v.asOutput());
+    opBuilder.addInput(beta1Power.asOutput());
+    opBuilder.addInput(beta2Power.asOutput());
+    opBuilder.addInput(lr.asOutput());
+    opBuilder.addInput(beta1.asOutput());
+    opBuilder.addInput(beta2.asOutput());
+    opBuilder.addInput(epsilon.asOutput());
+    opBuilder.addInput(grad.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.useLocking != null) {
+          opBuilder.setAttr("use_locking", opts.useLocking);
+        }
+        if (opts.useNesterov != null) {
+          opBuilder.setAttr("use_nesterov", opts.useNesterov);
+        }
+      }
+    }
+    return new ApplyAdam<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param useLocking If `True`, updating of the var, m, and v tensors will be protected
+   * by a lock; otherwise the behavior is undefined, but may exhibit less
+   * contention.
+   */
+  public static Options useLocking(Boolean useLocking) {
+    return new Options().useLocking(useLocking);
+  }
+  
+  /**
+   * @param useNesterov If `True`, uses the nesterov update.
+   */
+  public static Options useNesterov(Boolean useNesterov) {
+    return new Options().useNesterov(useNesterov);
+  }
+  
+  /**
+   * Same as "var".
+   */
+  public Output<T> out() {
+    return out;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return out;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ApplyAdam";
+  
+  private Output<T> out;
+  
+  private ApplyAdam(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    out = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyAddSign.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyAddSign.java
new file mode 100644
index 00000000000..1fe2cd756af
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyAddSign.java
@@ -0,0 +1,129 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.train;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Update '*var' according to the AddSign update.
+ * <p>
+ * m_t <- beta1 * m_{t-1} + (1 - beta1) * g
+ * update <- (alpha + sign_decay * sign(g) *sign(m)) * g
+ * variable <- variable - lr_t * update
+ * 
+ * @param <T> data type for {@code out()} output
+ */
+@Operator(group = "train")
+public final class ApplyAddSign<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.train.ApplyAddSign}
+   */
+  public static class Options {
+    
+    /**
+     * @param useLocking If `True`, updating of the var and m tensors is
+     * protected by a lock; otherwise the behavior is undefined, but may exhibit less
+     * contention.
+     */
+    public Options useLocking(Boolean useLocking) {
+      this.useLocking = useLocking;
+      return this;
+    }
+    
+    private Boolean useLocking;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new ApplyAddSign operation.
+   * 
+   * @param scope current scope
+   * @param var Should be from a Variable().
+   * @param m Should be from a Variable().
+   * @param lr Scaling factor. Must be a scalar.
+   * @param alpha Must be a scalar.
+   * @param signDecay Must be a scalar.
+   * @param beta Must be a scalar.
+   * @param grad The gradient.
+   * @param options carries optional attributes values
+   * @return a new instance of ApplyAddSign
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> ApplyAddSign<T> create(Scope scope, Operand<T> var, Operand<T> m, Operand<T> lr, Operand<T> alpha, Operand<T> signDecay, Operand<T> beta, Operand<T> grad, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ApplyAddSign", scope.makeOpName("ApplyAddSign"));
+    opBuilder.addInput(var.asOutput());
+    opBuilder.addInput(m.asOutput());
+    opBuilder.addInput(lr.asOutput());
+    opBuilder.addInput(alpha.asOutput());
+    opBuilder.addInput(signDecay.asOutput());
+    opBuilder.addInput(beta.asOutput());
+    opBuilder.addInput(grad.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.useLocking != null) {
+          opBuilder.setAttr("use_locking", opts.useLocking);
+        }
+      }
+    }
+    return new ApplyAddSign<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param useLocking If `True`, updating of the var and m tensors is
+   * protected by a lock; otherwise the behavior is undefined, but may exhibit less
+   * contention.
+   */
+  public static Options useLocking(Boolean useLocking) {
+    return new Options().useLocking(useLocking);
+  }
+  
+  /**
+   * Same as "var".
+   */
+  public Output<T> out() {
+    return out;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return out;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ApplyAddSign";
+  
+  private Output<T> out;
+  
+  private ApplyAddSign(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    out = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyCenteredRmsProp.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyCenteredRmsProp.java
new file mode 100644
index 00000000000..22aeeed0d91
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyCenteredRmsProp.java
@@ -0,0 +1,148 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.train;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Update '*var' according to the centered RMSProp algorithm.
+ * <p>
+ * The centered RMSProp algorithm uses an estimate of the centered second moment
+ * (i.e., the variance) for normalization, as opposed to regular RMSProp, which
+ * uses the (uncentered) second moment. This often helps with training, but is
+ * slightly more expensive in terms of computation and memory.
+ * <p>
+ * Note that in dense implementation of this algorithm, mg, ms, and mom will
+ * update even if the grad is zero, but in this sparse implementation, mg, ms,
+ * and mom will not update in iterations during which the grad is zero.
+ * <p>
+ * mean_square = decay * mean_square + (1-decay) * gradient ** 2
+ * mean_grad = decay * mean_grad + (1-decay) * gradient
+ * <p>
+ * Delta = learning_rate * gradient / sqrt(mean_square + epsilon - mean_grad ** 2)
+ * <p>
+ * mg <- rho * mg_{t-1} + (1-rho) * grad
+ * ms <- rho * ms_{t-1} + (1-rho) * grad * grad
+ * mom <- momentum * mom_{t-1} + lr * grad / sqrt(ms - mg * mg + epsilon)
+ * var <- var - mom
+ * 
+ * @param <T> data type for {@code out()} output
+ */
+@Operator(group = "train")
+public final class ApplyCenteredRmsProp<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.train.ApplyCenteredRmsProp}
+   */
+  public static class Options {
+    
+    /**
+     * @param useLocking If `True`, updating of the var, mg, ms, and mom tensors is
+     * protected by a lock; otherwise the behavior is undefined, but may exhibit less
+     * contention.
+     */
+    public Options useLocking(Boolean useLocking) {
+      this.useLocking = useLocking;
+      return this;
+    }
+    
+    private Boolean useLocking;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new ApplyCenteredRmsProp operation.
+   * 
+   * @param scope current scope
+   * @param var Should be from a Variable().
+   * @param mg Should be from a Variable().
+   * @param ms Should be from a Variable().
+   * @param mom Should be from a Variable().
+   * @param lr Scaling factor. Must be a scalar.
+   * @param rho Decay rate. Must be a scalar.
+   * @param momentum 
+   * @param epsilon Ridge term. Must be a scalar.
+   * @param grad The gradient.
+   * @param options carries optional attributes values
+   * @return a new instance of ApplyCenteredRmsProp
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> ApplyCenteredRmsProp<T> create(Scope scope, Operand<T> var, Operand<T> mg, Operand<T> ms, Operand<T> mom, Operand<T> lr, Operand<T> rho, Operand<T> momentum, Operand<T> epsilon, Operand<T> grad, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ApplyCenteredRMSProp", scope.makeOpName("ApplyCenteredRmsProp"));
+    opBuilder.addInput(var.asOutput());
+    opBuilder.addInput(mg.asOutput());
+    opBuilder.addInput(ms.asOutput());
+    opBuilder.addInput(mom.asOutput());
+    opBuilder.addInput(lr.asOutput());
+    opBuilder.addInput(rho.asOutput());
+    opBuilder.addInput(momentum.asOutput());
+    opBuilder.addInput(epsilon.asOutput());
+    opBuilder.addInput(grad.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.useLocking != null) {
+          opBuilder.setAttr("use_locking", opts.useLocking);
+        }
+      }
+    }
+    return new ApplyCenteredRmsProp<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param useLocking If `True`, updating of the var, mg, ms, and mom tensors is
+   * protected by a lock; otherwise the behavior is undefined, but may exhibit less
+   * contention.
+   */
+  public static Options useLocking(Boolean useLocking) {
+    return new Options().useLocking(useLocking);
+  }
+  
+  /**
+   * Same as "var".
+   */
+  public Output<T> out() {
+    return out;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return out;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ApplyCenteredRMSProp";
+  
+  private Output<T> out;
+  
+  private ApplyCenteredRmsProp(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    out = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyFtrl.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyFtrl.java
new file mode 100644
index 00000000000..f85cf2769c0
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyFtrl.java
@@ -0,0 +1,156 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.train;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Update '*var' according to the Ftrl-proximal scheme.
+ * <p>
+ * grad_with_shrinkage = grad + 2 * l2_shrinkage * var
+ * accum_new = accum + grad * grad
+ * linear += grad_with_shrinkage -
+ *     (accum_new^(-lr_power) - accum^(-lr_power)) / lr * var
+ * quadratic = 1.0 / (accum_new^(lr_power) * lr) + 2 * l2
+ * var = (sign(linear) * l1 - linear) / quadratic if |linear| > l1 else 0.0
+ * accum = accum_new
+ * 
+ * @param <T> data type for {@code out()} output
+ */
+@Operator(group = "train")
+public final class ApplyFtrl<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.train.ApplyFtrl}
+   */
+  public static class Options {
+    
+    /**
+     * @param useLocking If `True`, updating of the var and accum tensors will be protected
+     * by a lock; otherwise the behavior is undefined, but may exhibit less
+     * contention.
+     */
+    public Options useLocking(Boolean useLocking) {
+      this.useLocking = useLocking;
+      return this;
+    }
+    
+    /**
+     * @param multiplyLinearByLr 
+     */
+    public Options multiplyLinearByLr(Boolean multiplyLinearByLr) {
+      this.multiplyLinearByLr = multiplyLinearByLr;
+      return this;
+    }
+    
+    private Boolean useLocking;
+    private Boolean multiplyLinearByLr;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new ApplyFtrl operation.
+   * 
+   * @param scope current scope
+   * @param var Should be from a Variable().
+   * @param accum Should be from a Variable().
+   * @param linear Should be from a Variable().
+   * @param grad The gradient.
+   * @param lr Scaling factor. Must be a scalar.
+   * @param l1 L1 regularization. Must be a scalar.
+   * @param l2 L2 shrinkage regularization. Must be a scalar.
+   * @param l2Shrinkage 
+   * @param lrPower Scaling factor. Must be a scalar.
+   * @param options carries optional attributes values
+   * @return a new instance of ApplyFtrl
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> ApplyFtrl<T> create(Scope scope, Operand<T> var, Operand<T> accum, Operand<T> linear, Operand<T> grad, Operand<T> lr, Operand<T> l1, Operand<T> l2, Operand<T> l2Shrinkage, Operand<T> lrPower, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ApplyFtrlV2", scope.makeOpName("ApplyFtrl"));
+    opBuilder.addInput(var.asOutput());
+    opBuilder.addInput(accum.asOutput());
+    opBuilder.addInput(linear.asOutput());
+    opBuilder.addInput(grad.asOutput());
+    opBuilder.addInput(lr.asOutput());
+    opBuilder.addInput(l1.asOutput());
+    opBuilder.addInput(l2.asOutput());
+    opBuilder.addInput(l2Shrinkage.asOutput());
+    opBuilder.addInput(lrPower.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.useLocking != null) {
+          opBuilder.setAttr("use_locking", opts.useLocking);
+        }
+        if (opts.multiplyLinearByLr != null) {
+          opBuilder.setAttr("multiply_linear_by_lr", opts.multiplyLinearByLr);
+        }
+      }
+    }
+    return new ApplyFtrl<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param useLocking If `True`, updating of the var and accum tensors will be protected
+   * by a lock; otherwise the behavior is undefined, but may exhibit less
+   * contention.
+   */
+  public static Options useLocking(Boolean useLocking) {
+    return new Options().useLocking(useLocking);
+  }
+  
+  /**
+   * @param multiplyLinearByLr 
+   */
+  public static Options multiplyLinearByLr(Boolean multiplyLinearByLr) {
+    return new Options().multiplyLinearByLr(multiplyLinearByLr);
+  }
+  
+  /**
+   * Same as "var".
+   */
+  public Output<T> out() {
+    return out;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return out;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ApplyFtrlV2";
+  
+  private Output<T> out;
+  
+  private ApplyFtrl(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    out = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyGradientDescent.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyGradientDescent.java
new file mode 100644
index 00000000000..729312860fd
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyGradientDescent.java
@@ -0,0 +1,115 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.train;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Update '*var' by subtracting 'alpha' * 'delta' from it.
+ * 
+ * @param <T> data type for {@code out()} output
+ */
+@Operator(group = "train")
+public final class ApplyGradientDescent<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.train.ApplyGradientDescent}
+   */
+  public static class Options {
+    
+    /**
+     * @param useLocking If `True`, the subtraction will be protected by a lock;
+     * otherwise the behavior is undefined, but may exhibit less contention.
+     */
+    public Options useLocking(Boolean useLocking) {
+      this.useLocking = useLocking;
+      return this;
+    }
+    
+    private Boolean useLocking;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new ApplyGradientDescent operation.
+   * 
+   * @param scope current scope
+   * @param var Should be from a Variable().
+   * @param alpha Scaling factor. Must be a scalar.
+   * @param delta The change.
+   * @param options carries optional attributes values
+   * @return a new instance of ApplyGradientDescent
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> ApplyGradientDescent<T> create(Scope scope, Operand<T> var, Operand<T> alpha, Operand<T> delta, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ApplyGradientDescent", scope.makeOpName("ApplyGradientDescent"));
+    opBuilder.addInput(var.asOutput());
+    opBuilder.addInput(alpha.asOutput());
+    opBuilder.addInput(delta.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.useLocking != null) {
+          opBuilder.setAttr("use_locking", opts.useLocking);
+        }
+      }
+    }
+    return new ApplyGradientDescent<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param useLocking If `True`, the subtraction will be protected by a lock;
+   * otherwise the behavior is undefined, but may exhibit less contention.
+   */
+  public static Options useLocking(Boolean useLocking) {
+    return new Options().useLocking(useLocking);
+  }
+  
+  /**
+   * Same as "var".
+   */
+  public Output<T> out() {
+    return out;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return out;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ApplyGradientDescent";
+  
+  private Output<T> out;
+  
+  private ApplyGradientDescent(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    out = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyMomentum.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyMomentum.java
new file mode 100644
index 00000000000..d7885da2e32
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyMomentum.java
@@ -0,0 +1,149 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.train;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Update '*var' according to the momentum scheme.
+ * <p>
+ * Set use_nesterov = True if you want to use Nesterov momentum.
+ * <p>
+ * accum = accum * momentum + grad
+ * var -= lr * accum
+ * 
+ * @param <T> data type for {@code out()} output
+ */
+@Operator(group = "train")
+public final class ApplyMomentum<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.train.ApplyMomentum}
+   */
+  public static class Options {
+    
+    /**
+     * @param useLocking If `True`, updating of the var and accum tensors will be protected
+     * by a lock; otherwise the behavior is undefined, but may exhibit less
+     * contention.
+     */
+    public Options useLocking(Boolean useLocking) {
+      this.useLocking = useLocking;
+      return this;
+    }
+    
+    /**
+     * @param useNesterov If `True`, the tensor passed to compute grad will be
+     * var - lr * momentum * accum, so in the end, the var you get is actually
+     * var - lr * momentum * accum.
+     */
+    public Options useNesterov(Boolean useNesterov) {
+      this.useNesterov = useNesterov;
+      return this;
+    }
+    
+    private Boolean useLocking;
+    private Boolean useNesterov;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new ApplyMomentum operation.
+   * 
+   * @param scope current scope
+   * @param var Should be from a Variable().
+   * @param accum Should be from a Variable().
+   * @param lr Scaling factor. Must be a scalar.
+   * @param grad The gradient.
+   * @param momentum Momentum. Must be a scalar.
+   * @param options carries optional attributes values
+   * @return a new instance of ApplyMomentum
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> ApplyMomentum<T> create(Scope scope, Operand<T> var, Operand<T> accum, Operand<T> lr, Operand<T> grad, Operand<T> momentum, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ApplyMomentum", scope.makeOpName("ApplyMomentum"));
+    opBuilder.addInput(var.asOutput());
+    opBuilder.addInput(accum.asOutput());
+    opBuilder.addInput(lr.asOutput());
+    opBuilder.addInput(grad.asOutput());
+    opBuilder.addInput(momentum.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.useLocking != null) {
+          opBuilder.setAttr("use_locking", opts.useLocking);
+        }
+        if (opts.useNesterov != null) {
+          opBuilder.setAttr("use_nesterov", opts.useNesterov);
+        }
+      }
+    }
+    return new ApplyMomentum<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param useLocking If `True`, updating of the var and accum tensors will be protected
+   * by a lock; otherwise the behavior is undefined, but may exhibit less
+   * contention.
+   */
+  public static Options useLocking(Boolean useLocking) {
+    return new Options().useLocking(useLocking);
+  }
+  
+  /**
+   * @param useNesterov If `True`, the tensor passed to compute grad will be
+   * var - lr * momentum * accum, so in the end, the var you get is actually
+   * var - lr * momentum * accum.
+   */
+  public static Options useNesterov(Boolean useNesterov) {
+    return new Options().useNesterov(useNesterov);
+  }
+  
+  /**
+   * Same as "var".
+   */
+  public Output<T> out() {
+    return out;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return out;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ApplyMomentum";
+  
+  private Output<T> out;
+  
+  private ApplyMomentum(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    out = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyPowerSign.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyPowerSign.java
new file mode 100644
index 00000000000..d508e939714
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyPowerSign.java
@@ -0,0 +1,129 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.train;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Update '*var' according to the AddSign update.
+ * <p>
+ * m_t <- beta1 * m_{t-1} + (1 - beta1) * g
+ * update <- exp(logbase * sign_decay * sign(g) * sign(m_t)) * g
+ * variable <- variable - lr_t * update
+ * 
+ * @param <T> data type for {@code out()} output
+ */
+@Operator(group = "train")
+public final class ApplyPowerSign<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.train.ApplyPowerSign}
+   */
+  public static class Options {
+    
+    /**
+     * @param useLocking If `True`, updating of the var and m tensors is
+     * protected by a lock; otherwise the behavior is undefined, but may exhibit less
+     * contention.
+     */
+    public Options useLocking(Boolean useLocking) {
+      this.useLocking = useLocking;
+      return this;
+    }
+    
+    private Boolean useLocking;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new ApplyPowerSign operation.
+   * 
+   * @param scope current scope
+   * @param var Should be from a Variable().
+   * @param m Should be from a Variable().
+   * @param lr Scaling factor. Must be a scalar.
+   * @param logbase Must be a scalar.
+   * @param signDecay Must be a scalar.
+   * @param beta Must be a scalar.
+   * @param grad The gradient.
+   * @param options carries optional attributes values
+   * @return a new instance of ApplyPowerSign
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> ApplyPowerSign<T> create(Scope scope, Operand<T> var, Operand<T> m, Operand<T> lr, Operand<T> logbase, Operand<T> signDecay, Operand<T> beta, Operand<T> grad, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ApplyPowerSign", scope.makeOpName("ApplyPowerSign"));
+    opBuilder.addInput(var.asOutput());
+    opBuilder.addInput(m.asOutput());
+    opBuilder.addInput(lr.asOutput());
+    opBuilder.addInput(logbase.asOutput());
+    opBuilder.addInput(signDecay.asOutput());
+    opBuilder.addInput(beta.asOutput());
+    opBuilder.addInput(grad.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.useLocking != null) {
+          opBuilder.setAttr("use_locking", opts.useLocking);
+        }
+      }
+    }
+    return new ApplyPowerSign<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param useLocking If `True`, updating of the var and m tensors is
+   * protected by a lock; otherwise the behavior is undefined, but may exhibit less
+   * contention.
+   */
+  public static Options useLocking(Boolean useLocking) {
+    return new Options().useLocking(useLocking);
+  }
+  
+  /**
+   * Same as "var".
+   */
+  public Output<T> out() {
+    return out;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return out;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ApplyPowerSign";
+  
+  private Output<T> out;
+  
+  private ApplyPowerSign(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    out = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyProximalAdagrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyProximalAdagrad.java
new file mode 100644
index 00000000000..bd710eb08ca
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyProximalAdagrad.java
@@ -0,0 +1,125 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.train;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Update '*var' and '*accum' according to FOBOS with Adagrad learning rate.
+ * <p>
+ * accum += grad <i> grad
+ * prox_v = var - lr </i> grad <i> (1 / sqrt(accum))
+ * var = sign(prox_v)/(1+lr</i>l2) <i> max{|prox_v|-lr</i>l1,0}
+ * 
+ * @param <T> data type for {@code out()} output
+ */
+@Operator(group = "train")
+public final class ApplyProximalAdagrad<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.train.ApplyProximalAdagrad}
+   */
+  public static class Options {
+    
+    /**
+     * @param useLocking If True, updating of the var and accum tensors will be protected by
+     * a lock; otherwise the behavior is undefined, but may exhibit less contention.
+     */
+    public Options useLocking(Boolean useLocking) {
+      this.useLocking = useLocking;
+      return this;
+    }
+    
+    private Boolean useLocking;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new ApplyProximalAdagrad operation.
+   * 
+   * @param scope current scope
+   * @param var Should be from a Variable().
+   * @param accum Should be from a Variable().
+   * @param lr Scaling factor. Must be a scalar.
+   * @param l1 L1 regularization. Must be a scalar.
+   * @param l2 L2 regularization. Must be a scalar.
+   * @param grad The gradient.
+   * @param options carries optional attributes values
+   * @return a new instance of ApplyProximalAdagrad
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> ApplyProximalAdagrad<T> create(Scope scope, Operand<T> var, Operand<T> accum, Operand<T> lr, Operand<T> l1, Operand<T> l2, Operand<T> grad, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ApplyProximalAdagrad", scope.makeOpName("ApplyProximalAdagrad"));
+    opBuilder.addInput(var.asOutput());
+    opBuilder.addInput(accum.asOutput());
+    opBuilder.addInput(lr.asOutput());
+    opBuilder.addInput(l1.asOutput());
+    opBuilder.addInput(l2.asOutput());
+    opBuilder.addInput(grad.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.useLocking != null) {
+          opBuilder.setAttr("use_locking", opts.useLocking);
+        }
+      }
+    }
+    return new ApplyProximalAdagrad<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param useLocking If True, updating of the var and accum tensors will be protected by
+   * a lock; otherwise the behavior is undefined, but may exhibit less contention.
+   */
+  public static Options useLocking(Boolean useLocking) {
+    return new Options().useLocking(useLocking);
+  }
+  
+  /**
+   * Same as "var".
+   */
+  public Output<T> out() {
+    return out;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return out;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ApplyProximalAdagrad";
+  
+  private Output<T> out;
+  
+  private ApplyProximalAdagrad(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    out = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyProximalGradientDescent.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyProximalGradientDescent.java
new file mode 100644
index 00000000000..eb5e9a96dc9
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyProximalGradientDescent.java
@@ -0,0 +1,122 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.train;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Update '*var' as FOBOS algorithm with fixed learning rate.
+ * <p>
+ * prox_v = var - alpha <i> delta
+ * var = sign(prox_v)/(1+alpha</i>l2) <i> max{|prox_v|-alpha</i>l1,0}
+ * 
+ * @param <T> data type for {@code out()} output
+ */
+@Operator(group = "train")
+public final class ApplyProximalGradientDescent<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.train.ApplyProximalGradientDescent}
+   */
+  public static class Options {
+    
+    /**
+     * @param useLocking If True, the subtraction will be protected by a lock;
+     * otherwise the behavior is undefined, but may exhibit less contention.
+     */
+    public Options useLocking(Boolean useLocking) {
+      this.useLocking = useLocking;
+      return this;
+    }
+    
+    private Boolean useLocking;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new ApplyProximalGradientDescent operation.
+   * 
+   * @param scope current scope
+   * @param var Should be from a Variable().
+   * @param alpha Scaling factor. Must be a scalar.
+   * @param l1 L1 regularization. Must be a scalar.
+   * @param l2 L2 regularization. Must be a scalar.
+   * @param delta The change.
+   * @param options carries optional attributes values
+   * @return a new instance of ApplyProximalGradientDescent
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> ApplyProximalGradientDescent<T> create(Scope scope, Operand<T> var, Operand<T> alpha, Operand<T> l1, Operand<T> l2, Operand<T> delta, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ApplyProximalGradientDescent", scope.makeOpName("ApplyProximalGradientDescent"));
+    opBuilder.addInput(var.asOutput());
+    opBuilder.addInput(alpha.asOutput());
+    opBuilder.addInput(l1.asOutput());
+    opBuilder.addInput(l2.asOutput());
+    opBuilder.addInput(delta.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.useLocking != null) {
+          opBuilder.setAttr("use_locking", opts.useLocking);
+        }
+      }
+    }
+    return new ApplyProximalGradientDescent<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param useLocking If True, the subtraction will be protected by a lock;
+   * otherwise the behavior is undefined, but may exhibit less contention.
+   */
+  public static Options useLocking(Boolean useLocking) {
+    return new Options().useLocking(useLocking);
+  }
+  
+  /**
+   * Same as "var".
+   */
+  public Output<T> out() {
+    return out;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return out;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ApplyProximalGradientDescent";
+  
+  private Output<T> out;
+  
+  private ApplyProximalGradientDescent(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    out = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyRmsProp.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyRmsProp.java
new file mode 100644
index 00000000000..388fc062c5d
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ApplyRmsProp.java
@@ -0,0 +1,138 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.train;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Update '*var' according to the RMSProp algorithm.
+ * <p>
+ * Note that in dense implementation of this algorithm, ms and mom will
+ * update even if the grad is zero, but in this sparse implementation, ms
+ * and mom will not update in iterations during which the grad is zero.
+ * <p>
+ * mean_square = decay * mean_square + (1-decay) * gradient ** 2
+ * Delta = learning_rate * gradient / sqrt(mean_square + epsilon)
+ * <p>
+ * ms <- rho * ms_{t-1} + (1-rho) * grad * grad
+ * mom <- momentum * mom_{t-1} + lr * grad / sqrt(ms + epsilon)
+ * var <- var - mom
+ * 
+ * @param <T> data type for {@code out()} output
+ */
+@Operator(group = "train")
+public final class ApplyRmsProp<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.train.ApplyRmsProp}
+   */
+  public static class Options {
+    
+    /**
+     * @param useLocking If `True`, updating of the var, ms, and mom tensors is protected
+     * by a lock; otherwise the behavior is undefined, but may exhibit less
+     * contention.
+     */
+    public Options useLocking(Boolean useLocking) {
+      this.useLocking = useLocking;
+      return this;
+    }
+    
+    private Boolean useLocking;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new ApplyRmsProp operation.
+   * 
+   * @param scope current scope
+   * @param var Should be from a Variable().
+   * @param ms Should be from a Variable().
+   * @param mom Should be from a Variable().
+   * @param lr Scaling factor. Must be a scalar.
+   * @param rho Decay rate. Must be a scalar.
+   * @param momentum 
+   * @param epsilon Ridge term. Must be a scalar.
+   * @param grad The gradient.
+   * @param options carries optional attributes values
+   * @return a new instance of ApplyRmsProp
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> ApplyRmsProp<T> create(Scope scope, Operand<T> var, Operand<T> ms, Operand<T> mom, Operand<T> lr, Operand<T> rho, Operand<T> momentum, Operand<T> epsilon, Operand<T> grad, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ApplyRMSProp", scope.makeOpName("ApplyRmsProp"));
+    opBuilder.addInput(var.asOutput());
+    opBuilder.addInput(ms.asOutput());
+    opBuilder.addInput(mom.asOutput());
+    opBuilder.addInput(lr.asOutput());
+    opBuilder.addInput(rho.asOutput());
+    opBuilder.addInput(momentum.asOutput());
+    opBuilder.addInput(epsilon.asOutput());
+    opBuilder.addInput(grad.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.useLocking != null) {
+          opBuilder.setAttr("use_locking", opts.useLocking);
+        }
+      }
+    }
+    return new ApplyRmsProp<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param useLocking If `True`, updating of the var, ms, and mom tensors is protected
+   * by a lock; otherwise the behavior is undefined, but may exhibit less
+   * contention.
+   */
+  public static Options useLocking(Boolean useLocking) {
+    return new Options().useLocking(useLocking);
+  }
+  
+  /**
+   * Same as "var".
+   */
+  public Output<T> out() {
+    return out;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return out;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ApplyRMSProp";
+  
+  private Output<T> out;
+  
+  private ApplyRmsProp(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    out = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/BatchMatMul.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/BatchMatMul.java
new file mode 100644
index 00000000000..3b6da5b0811
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/BatchMatMul.java
@@ -0,0 +1,154 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.train;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Multiplies slices of two tensors in batches.
+ * <p>
+ * Multiplies all slices of `Tensor` `x` and `y` (each slice can be
+ * viewed as an element of a batch), and arranges the individual results
+ * in a single output tensor of the same batch size. Each of the
+ * individual slices can optionally be adjointed (to adjoint a matrix
+ * means to transpose and conjugate it) before multiplication by setting
+ * the `adj_x` or `adj_y` flag to `True`, which are by default `False`.
+ * <p>
+ * The input tensors `x` and `y` are 2-D or higher with shape `[..., r_x, c_x]`
+ * and `[..., r_y, c_y]`.
+ * <p>
+ * The output tensor is 2-D or higher with shape `[..., r_o, c_o]`, where:
+ * <p>
+ *     r_o = c_x if adj_x else r_x
+ *     c_o = r_y if adj_y else c_y
+ * <p>
+ * It is computed as:
+ * <p>
+ *     output[..., :, :] = matrix(x[..., :, :]) * matrix(y[..., :, :])
+ * <p>
+ * <i>NOTE</i>: `train.BatchMatMul` supports broadcasting in the batch dimensions. More
+ * about broadcasting
+ * [here](http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html).
+ * 
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator(group = "train")
+public final class BatchMatMul<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.train.BatchMatMul}
+   */
+  public static class Options {
+    
+    /**
+     * @param adjX If `True`, adjoint the slices of `x`. Defaults to `False`.
+     */
+    public Options adjX(Boolean adjX) {
+      this.adjX = adjX;
+      return this;
+    }
+    
+    /**
+     * @param adjY If `True`, adjoint the slices of `y`. Defaults to `False`.
+     */
+    public Options adjY(Boolean adjY) {
+      this.adjY = adjY;
+      return this;
+    }
+    
+    private Boolean adjX;
+    private Boolean adjY;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new BatchMatMul operation.
+   * 
+   * @param scope current scope
+   * @param x 2-D or higher with shape `[..., r_x, c_x]`.
+   * @param y 2-D or higher with shape `[..., r_y, c_y]`.
+   * @param options carries optional attributes values
+   * @return a new instance of BatchMatMul
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> BatchMatMul<T> create(Scope scope, Operand<T> x, Operand<T> y, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("BatchMatMulV2", scope.makeOpName("BatchMatMul"));
+    opBuilder.addInput(x.asOutput());
+    opBuilder.addInput(y.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.adjX != null) {
+          opBuilder.setAttr("adj_x", opts.adjX);
+        }
+        if (opts.adjY != null) {
+          opBuilder.setAttr("adj_y", opts.adjY);
+        }
+      }
+    }
+    return new BatchMatMul<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param adjX If `True`, adjoint the slices of `x`. Defaults to `False`.
+   */
+  public static Options adjX(Boolean adjX) {
+    return new Options().adjX(adjX);
+  }
+  
+  /**
+   * @param adjY If `True`, adjoint the slices of `y`. Defaults to `False`.
+   */
+  public static Options adjY(Boolean adjY) {
+    return new Options().adjY(adjY);
+  }
+  
+  /**
+   * 3-D or higher with shape `[..., r_o, c_o]`
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "BatchMatMulV2";
+  
+  private Output<T> output;
+  
+  private BatchMatMul(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ConditionalAccumulator.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ConditionalAccumulator.java
new file mode 100644
index 00000000000..aaf8a6f767a
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ConditionalAccumulator.java
@@ -0,0 +1,160 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.train;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TString;
+import org.tensorflow.types.family.TType;
+
+/**
+ * A conditional accumulator for aggregating gradients.
+ * <p>
+ * The accumulator accepts gradients marked with local_step greater or
+ * equal to the most recent global_step known to the accumulator. The
+ * average can be extracted from the accumulator, provided sufficient
+ * gradients have been accumulated. Extracting the average automatically
+ * resets the aggregate to 0, and increments the global_step recorded by
+ * the accumulator.
+ */
+@Operator(group = "train")
+public final class ConditionalAccumulator extends RawOp implements Operand<TString> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.train.ConditionalAccumulator}
+   */
+  public static class Options {
+    
+    /**
+     * @param container If non-empty, this accumulator is placed in the given container.
+     * Otherwise, a default container is used.
+     */
+    public Options container(String container) {
+      this.container = container;
+      return this;
+    }
+    
+    /**
+     * @param sharedName If non-empty, this accumulator will be shared under the
+     * given name across multiple sessions.
+     */
+    public Options sharedName(String sharedName) {
+      this.sharedName = sharedName;
+      return this;
+    }
+    
+    /**
+     * @param reductionType 
+     */
+    public Options reductionType(String reductionType) {
+      this.reductionType = reductionType;
+      return this;
+    }
+    
+    private String container;
+    private String sharedName;
+    private String reductionType;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new ConditionalAccumulator operation.
+   * 
+   * @param scope current scope
+   * @param dtype The type of the value being accumulated.
+   * @param shape The shape of the values, can be [], in which case shape is unknown.
+   * @param options carries optional attributes values
+   * @return a new instance of ConditionalAccumulator
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> ConditionalAccumulator create(Scope scope, Class<T> dtype, Shape shape, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ConditionalAccumulator", scope.makeOpName("ConditionalAccumulator"));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("dtype", dtype);
+    opBuilder.setAttr("shape", shape);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.container != null) {
+          opBuilder.setAttr("container", opts.container);
+        }
+        if (opts.sharedName != null) {
+          opBuilder.setAttr("shared_name", opts.sharedName);
+        }
+        if (opts.reductionType != null) {
+          opBuilder.setAttr("reduction_type", opts.reductionType);
+        }
+      }
+    }
+    return new ConditionalAccumulator(opBuilder.build());
+  }
+  
+  /**
+   * @param container If non-empty, this accumulator is placed in the given container.
+   * Otherwise, a default container is used.
+   */
+  public static Options container(String container) {
+    return new Options().container(container);
+  }
+  
+  /**
+   * @param sharedName If non-empty, this accumulator will be shared under the
+   * given name across multiple sessions.
+   */
+  public static Options sharedName(String sharedName) {
+    return new Options().sharedName(sharedName);
+  }
+  
+  /**
+   * @param reductionType 
+   */
+  public static Options reductionType(String reductionType) {
+    return new Options().reductionType(reductionType);
+  }
+  
+  /**
+   * The handle to the accumulator.
+   */
+  public Output<TString> handle() {
+    return handle;
+  }
+  
+  @Override
+  public Output<TString> asOutput() {
+    return handle;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ConditionalAccumulator";
+  
+  private Output<TString> handle;
+  
+  private ConditionalAccumulator(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    handle = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/GenerateVocabRemapping.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/GenerateVocabRemapping.java
new file mode 100644
index 00000000000..e91afa44cd8
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/GenerateVocabRemapping.java
@@ -0,0 +1,152 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.train;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.TString;
+
+/**
+ * Given a path to new and old vocabulary files, returns a remapping Tensor of
+ * <p>
+ * length `num_new_vocab`, where `remapping[i]` contains the row number in the old
+ * vocabulary that corresponds to row `i` in the new vocabulary (starting at line
+ * `new_vocab_offset` and up to `num_new_vocab` entities), or `-1` if entry `i`
+ * in the new vocabulary is not in the old vocabulary.  The old vocabulary is
+ * constrained to the first `old_vocab_size` entries if `old_vocab_size` is not the
+ * default value of -1.
+ * <p>
+ * `num_vocab_offset` enables
+ * use in the partitioned variable case, and should generally be set through
+ * examining partitioning info.  The format of the files should be a text file,
+ * with each line containing a single entity within the vocabulary.
+ * <p>
+ * For example, with `new_vocab_file` a text file containing each of the following
+ * elements on a single line: `[f0, f1, f2, f3]`, old_vocab_file = [f1, f0, f3],
+ * `num_new_vocab = 3, new_vocab_offset = 1`, the returned remapping would be
+ * `[0, -1, 2]`.
+ * <p>
+ * The op also returns a count of how many entries in the new vocabulary
+ * were present in the old vocabulary, which is used to calculate the number of
+ * values to initialize in a weight matrix remapping
+ * <p>
+ * This functionality can be used to remap both row vocabularies (typically,
+ * features) and column vocabularies (typically, classes) from TensorFlow
+ * checkpoints.  Note that the partitioning logic relies on contiguous vocabularies
+ * corresponding to div-partitioned variables.  Moreover, the underlying remapping
+ * uses an IndexTable (as opposed to an inexact CuckooTable), so client code should
+ * use the corresponding index_table_from_file() as the FeatureColumn framework
+ * does (as opposed to tf.feature_to_id(), which uses a CuckooTable).
+ */
+@Operator(group = "train")
+public final class GenerateVocabRemapping extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.train.GenerateVocabRemapping}
+   */
+  public static class Options {
+    
+    /**
+     * @param oldVocabSize Number of entries in the old vocab file to consider.  If -1,
+     * use the entire old vocabulary.
+     */
+    public Options oldVocabSize(Long oldVocabSize) {
+      this.oldVocabSize = oldVocabSize;
+      return this;
+    }
+    
+    private Long oldVocabSize;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new GenerateVocabRemapping operation.
+   * 
+   * @param scope current scope
+   * @param newVocabFile Path to the new vocab file.
+   * @param oldVocabFile Path to the old vocab file.
+   * @param newVocabOffset How many entries into the new vocab file to start reading.
+   * @param numNewVocab Number of entries in the new vocab file to remap.
+   * @param options carries optional attributes values
+   * @return a new instance of GenerateVocabRemapping
+   */
+  @Endpoint(describeByClass = true)
+  public static GenerateVocabRemapping create(Scope scope, Operand<TString> newVocabFile, Operand<TString> oldVocabFile, Long newVocabOffset, Long numNewVocab, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("GenerateVocabRemapping", scope.makeOpName("GenerateVocabRemapping"));
+    opBuilder.addInput(newVocabFile.asOutput());
+    opBuilder.addInput(oldVocabFile.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("new_vocab_offset", newVocabOffset);
+    opBuilder.setAttr("num_new_vocab", numNewVocab);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.oldVocabSize != null) {
+          opBuilder.setAttr("old_vocab_size", opts.oldVocabSize);
+        }
+      }
+    }
+    return new GenerateVocabRemapping(opBuilder.build());
+  }
+  
+  /**
+   * @param oldVocabSize Number of entries in the old vocab file to consider.  If -1,
+   * use the entire old vocabulary.
+   */
+  public static Options oldVocabSize(Long oldVocabSize) {
+    return new Options().oldVocabSize(oldVocabSize);
+  }
+  
+  /**
+   * A Tensor of length num_new_vocab where the element at index i
+   * is equal to the old ID that maps to the new ID i.  This element is -1 for any
+   * new ID that is not found in the old vocabulary.
+   */
+  public Output<TInt64> remapping() {
+    return remapping;
+  }
+  
+  /**
+   * Number of new vocab entries found in old vocab.
+   */
+  public Output<TInt32> numPresent() {
+    return numPresent;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "GenerateVocabRemapping";
+  
+  private Output<TInt64> remapping;
+  private Output<TInt32> numPresent;
+  
+  private GenerateVocabRemapping(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    remapping = operation.output(outputIdx++);
+    numPresent = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/MergeV2Checkpoints.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/MergeV2Checkpoints.java
new file mode 100644
index 00000000000..986553a2d8e
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/MergeV2Checkpoints.java
@@ -0,0 +1,102 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.train;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TString;
+
+/**
+ * V2 format specific: merges the metadata files of sharded checkpoints.  The
+ * <p>
+ * result is one logical checkpoint, with one physical metadata file and renamed
+ * data files.
+ * <p>
+ * Intended for "grouping" multiple checkpoints in a sharded checkpoint setup.
+ * <p>
+ * If delete_old_dirs is true, attempts to delete recursively the dirname of each
+ * path in the input checkpoint_prefixes.  This is useful when those paths are non
+ * user-facing temporary locations.
+ */
+@Operator(group = "train")
+public final class MergeV2Checkpoints extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.train.MergeV2Checkpoints}
+   */
+  public static class Options {
+    
+    /**
+     * @param deleteOldDirs see above.
+     */
+    public Options deleteOldDirs(Boolean deleteOldDirs) {
+      this.deleteOldDirs = deleteOldDirs;
+      return this;
+    }
+    
+    private Boolean deleteOldDirs;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new MergeV2Checkpoints operation.
+   * 
+   * @param scope current scope
+   * @param checkpointPrefixes prefixes of V2 checkpoints to merge.
+   * @param destinationPrefix scalar.  The desired final prefix.  Allowed to be the same
+   * as one of the checkpoint_prefixes.
+   * @param options carries optional attributes values
+   * @return a new instance of MergeV2Checkpoints
+   */
+  @Endpoint(describeByClass = true)
+  public static MergeV2Checkpoints create(Scope scope, Operand<TString> checkpointPrefixes, Operand<TString> destinationPrefix, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("MergeV2Checkpoints", scope.makeOpName("MergeV2Checkpoints"));
+    opBuilder.addInput(checkpointPrefixes.asOutput());
+    opBuilder.addInput(destinationPrefix.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.deleteOldDirs != null) {
+          opBuilder.setAttr("delete_old_dirs", opts.deleteOldDirs);
+        }
+      }
+    }
+    return new MergeV2Checkpoints(opBuilder.build());
+  }
+  
+  /**
+   * @param deleteOldDirs see above.
+   */
+  public static Options deleteOldDirs(Boolean deleteOldDirs) {
+    return new Options().deleteOldDirs(deleteOldDirs);
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "MergeV2Checkpoints";
+  
+  private MergeV2Checkpoints(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/NegTrain.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/NegTrain.java
new file mode 100644
index 00000000000..b3e5316ad7e
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/NegTrain.java
@@ -0,0 +1,74 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.train;
+
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.TInt32;
+
+/**
+ * Training via negative sampling.
+ */
+@Operator(group = "train")
+public final class NegTrain extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new NegTrain operation.
+   * 
+   * @param scope current scope
+   * @param wIn input word embedding.
+   * @param wOut output word embedding.
+   * @param examples A vector of word ids.
+   * @param labels A vector of word ids.
+   * @param lr 
+   * @param vocabCount Count of words in the vocabulary.
+   * @param numNegativeSamples Number of negative samples per example.
+   * @return a new instance of NegTrain
+   */
+  @Endpoint(describeByClass = true)
+  public static NegTrain create(Scope scope, Operand<TFloat32> wIn, Operand<TFloat32> wOut, Operand<TInt32> examples, Operand<TInt32> labels, Operand<TFloat32> lr, List<Long> vocabCount, Long numNegativeSamples) {
+    OperationBuilder opBuilder = scope.env().opBuilder("NegTrain", scope.makeOpName("NegTrain"));
+    opBuilder.addInput(wIn.asOutput());
+    opBuilder.addInput(wOut.asOutput());
+    opBuilder.addInput(examples.asOutput());
+    opBuilder.addInput(labels.asOutput());
+    opBuilder.addInput(lr.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    long[] vocabCountArray = new long[vocabCount.size()];
+    for (int i = 0; i < vocabCountArray.length; ++i) {
+      vocabCountArray[i] = vocabCount.get(i);
+    }
+    opBuilder.setAttr("vocab_count", vocabCountArray);
+    opBuilder.setAttr("num_negative_samples", numNegativeSamples);
+    return new NegTrain(opBuilder.build());
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "NegTrain";
+  
+  private NegTrain(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/PreventGradient.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/PreventGradient.java
new file mode 100644
index 00000000000..03d342477a9
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/PreventGradient.java
@@ -0,0 +1,119 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.train;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * An identity op that triggers an error if a gradient is requested.
+ * <p>
+ * When executed in a graph, this op outputs its input tensor as-is.
+ * <p>
+ * When building ops to compute gradients, the TensorFlow gradient system
+ * will return an error when trying to lookup the gradient of this op,
+ * because no gradient must ever be registered for this function.  This
+ * op exists to prevent subtle bugs from silently returning unimplemented
+ * gradients in some corner cases.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator(group = "train")
+public final class PreventGradient<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.train.PreventGradient}
+   */
+  public static class Options {
+    
+    /**
+     * @param message Will be printed in the error when anyone tries to differentiate
+     * this operation.
+     */
+    public Options message(String message) {
+      this.message = message;
+      return this;
+    }
+    
+    private String message;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new PreventGradient operation.
+   * 
+   * @param scope current scope
+   * @param input any tensor.
+   * @param options carries optional attributes values
+   * @return a new instance of PreventGradient
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> PreventGradient<T> create(Scope scope, Operand<T> input, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("PreventGradient", scope.makeOpName("PreventGradient"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.message != null) {
+          opBuilder.setAttr("message", opts.message);
+        }
+      }
+    }
+    return new PreventGradient<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param message Will be printed in the error when anyone tries to differentiate
+   * this operation.
+   */
+  public static Options message(String message) {
+    return new Options().message(message);
+  }
+  
+  /**
+   * the same input tensor.
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "PreventGradient";
+  
+  private Output<T> output;
+  
+  private PreventGradient(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceAccumulatorApplyGradient.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceAccumulatorApplyGradient.java
new file mode 100644
index 00000000000..560c3400c3b
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceAccumulatorApplyGradient.java
@@ -0,0 +1,62 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.train;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Applies a gradient to a given accumulator.
+ * <p>
+ * Does not add if local_step is lesser than the accumulator's global_step.
+ */
+public final class ResourceAccumulatorApplyGradient extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new ResourceAccumulatorApplyGradient operation.
+   * 
+   * @param scope current scope
+   * @param handle The handle to a accumulator.
+   * @param localStep The local_step value at which the gradient was computed.
+   * @param gradient A tensor of the gradient to be accumulated.
+   * @return a new instance of ResourceAccumulatorApplyGradient
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> ResourceAccumulatorApplyGradient create(Scope scope, Operand<?> handle, Operand<TInt64> localStep, Operand<T> gradient) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ResourceAccumulatorApplyGradient", scope.makeOpName("ResourceAccumulatorApplyGradient"));
+    opBuilder.addInput(handle.asOutput());
+    opBuilder.addInput(localStep.asOutput());
+    opBuilder.addInput(gradient.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new ResourceAccumulatorApplyGradient(opBuilder.build());
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ResourceAccumulatorApplyGradient";
+  
+  private ResourceAccumulatorApplyGradient(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceAccumulatorNumAccumulated.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceAccumulatorNumAccumulated.java
new file mode 100644
index 00000000000..628a916f659
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceAccumulatorNumAccumulated.java
@@ -0,0 +1,72 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.train;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+
+/**
+ * Returns the number of gradients aggregated in the given accumulators.
+ */
+public final class ResourceAccumulatorNumAccumulated extends RawOp implements Operand<TInt32> {
+  
+  /**
+   * Factory method to create a class wrapping a new ResourceAccumulatorNumAccumulated operation.
+   * 
+   * @param scope current scope
+   * @param handle The handle to an accumulator.
+   * @return a new instance of ResourceAccumulatorNumAccumulated
+   */
+  @Endpoint(describeByClass = true)
+  public static ResourceAccumulatorNumAccumulated create(Scope scope, Operand<?> handle) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ResourceAccumulatorNumAccumulated", scope.makeOpName("ResourceAccumulatorNumAccumulated"));
+    opBuilder.addInput(handle.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new ResourceAccumulatorNumAccumulated(opBuilder.build());
+  }
+  
+  /**
+   * The number of gradients aggregated in the given accumulator.
+   */
+  public Output<TInt32> numAccumulated() {
+    return numAccumulated;
+  }
+  
+  @Override
+  public Output<TInt32> asOutput() {
+    return numAccumulated;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ResourceAccumulatorNumAccumulated";
+  
+  private Output<TInt32> numAccumulated;
+  
+  private ResourceAccumulatorNumAccumulated(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    numAccumulated = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceAccumulatorSetGlobalStep.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceAccumulatorSetGlobalStep.java
new file mode 100644
index 00000000000..37570909340
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceAccumulatorSetGlobalStep.java
@@ -0,0 +1,60 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.train;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt64;
+
+/**
+ * Updates the accumulator with a new value for global_step.
+ * <p>
+ * Logs warning if the accumulator's value is already higher than
+ * new_global_step.
+ */
+public final class ResourceAccumulatorSetGlobalStep extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new ResourceAccumulatorSetGlobalStep operation.
+   * 
+   * @param scope current scope
+   * @param handle The handle to an accumulator.
+   * @param newGlobalStep The new global_step value to set.
+   * @return a new instance of ResourceAccumulatorSetGlobalStep
+   */
+  @Endpoint(describeByClass = true)
+  public static ResourceAccumulatorSetGlobalStep create(Scope scope, Operand<?> handle, Operand<TInt64> newGlobalStep) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ResourceAccumulatorSetGlobalStep", scope.makeOpName("ResourceAccumulatorSetGlobalStep"));
+    opBuilder.addInput(handle.asOutput());
+    opBuilder.addInput(newGlobalStep.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new ResourceAccumulatorSetGlobalStep(opBuilder.build());
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ResourceAccumulatorSetGlobalStep";
+  
+  private ResourceAccumulatorSetGlobalStep(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceAccumulatorTakeGradient.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceAccumulatorTakeGradient.java
new file mode 100644
index 00000000000..75601de5579
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceAccumulatorTakeGradient.java
@@ -0,0 +1,86 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.train;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Extracts the average gradient in the given ConditionalAccumulator.
+ * <p>
+ * The op blocks until sufficient (i.e., more than num_required)
+ * gradients have been accumulated.  If the accumulator has already
+ * aggregated more than num_required gradients, it returns the average of
+ * the accumulated gradients.  Also automatically increments the recorded
+ * global_step in the accumulator by 1, and resets the aggregate to 0.
+ * 
+ * @param <T> data type for {@code average()} output
+ */
+public final class ResourceAccumulatorTakeGradient<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new ResourceAccumulatorTakeGradient operation.
+   * 
+   * @param scope current scope
+   * @param handle The handle to an accumulator.
+   * @param numRequired Number of gradients required before we return an aggregate.
+   * @param dtype The data type of accumulated gradients. Needs to correspond to the type
+   * of the accumulator.
+   * @return a new instance of ResourceAccumulatorTakeGradient
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> ResourceAccumulatorTakeGradient<T> create(Scope scope, Operand<?> handle, Operand<TInt32> numRequired, Class<T> dtype) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ResourceAccumulatorTakeGradient", scope.makeOpName("ResourceAccumulatorTakeGradient"));
+    opBuilder.addInput(handle.asOutput());
+    opBuilder.addInput(numRequired.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("dtype", dtype);
+    return new ResourceAccumulatorTakeGradient<T>(opBuilder.build());
+  }
+  
+  /**
+   * The average of the accumulated gradients.
+   */
+  public Output<T> average() {
+    return average;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return average;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ResourceAccumulatorTakeGradient";
+  
+  private Output<T> average;
+  
+  private ResourceAccumulatorTakeGradient(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    average = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyAdaMax.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyAdaMax.java
new file mode 100644
index 00000000000..169da75fecd
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyAdaMax.java
@@ -0,0 +1,113 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.train;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Update '*var' according to the AdaMax algorithm.
+ * <p>
+ * m_t <- beta1 * m_{t-1} + (1 - beta1) * g
+ * v_t <- max(beta2 * v_{t-1}, abs(g))
+ * variable <- variable - learning_rate / (1 - beta1^t) * m_t / (v_t + epsilon)
+ */
+public final class ResourceApplyAdaMax extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.train.ResourceApplyAdaMax}
+   */
+  public static class Options {
+    
+    /**
+     * @param useLocking If `True`, updating of the var, m, and v tensors will be protected
+     * by a lock; otherwise the behavior is undefined, but may exhibit less
+     * contention.
+     */
+    public Options useLocking(Boolean useLocking) {
+      this.useLocking = useLocking;
+      return this;
+    }
+    
+    private Boolean useLocking;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new ResourceApplyAdaMax operation.
+   * 
+   * @param scope current scope
+   * @param var Should be from a Variable().
+   * @param m Should be from a Variable().
+   * @param v Should be from a Variable().
+   * @param beta1Power Must be a scalar.
+   * @param lr Scaling factor. Must be a scalar.
+   * @param beta1 Momentum factor. Must be a scalar.
+   * @param beta2 Momentum factor. Must be a scalar.
+   * @param epsilon Ridge term. Must be a scalar.
+   * @param grad The gradient.
+   * @param options carries optional attributes values
+   * @return a new instance of ResourceApplyAdaMax
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> ResourceApplyAdaMax create(Scope scope, Operand<?> var, Operand<?> m, Operand<?> v, Operand<T> beta1Power, Operand<T> lr, Operand<T> beta1, Operand<T> beta2, Operand<T> epsilon, Operand<T> grad, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ResourceApplyAdaMax", scope.makeOpName("ResourceApplyAdaMax"));
+    opBuilder.addInput(var.asOutput());
+    opBuilder.addInput(m.asOutput());
+    opBuilder.addInput(v.asOutput());
+    opBuilder.addInput(beta1Power.asOutput());
+    opBuilder.addInput(lr.asOutput());
+    opBuilder.addInput(beta1.asOutput());
+    opBuilder.addInput(beta2.asOutput());
+    opBuilder.addInput(epsilon.asOutput());
+    opBuilder.addInput(grad.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.useLocking != null) {
+          opBuilder.setAttr("use_locking", opts.useLocking);
+        }
+      }
+    }
+    return new ResourceApplyAdaMax(opBuilder.build());
+  }
+  
+  /**
+   * @param useLocking If `True`, updating of the var, m, and v tensors will be protected
+   * by a lock; otherwise the behavior is undefined, but may exhibit less
+   * contention.
+   */
+  public static Options useLocking(Boolean useLocking) {
+    return new Options().useLocking(useLocking);
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ResourceApplyAdaMax";
+  
+  private ResourceApplyAdaMax(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyAdadelta.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyAdadelta.java
new file mode 100644
index 00000000000..4323a39de45
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyAdadelta.java
@@ -0,0 +1,109 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.train;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Update '*var' according to the adadelta scheme.
+ * <p>
+ * accum = rho() * accum + (1 - rho()) * grad.square();
+ * update = (update_accum + epsilon).sqrt() * (accum + epsilon()).rsqrt() * grad;
+ * update_accum = rho() * update_accum + (1 - rho()) * update.square();
+ * var -= update;
+ */
+@Operator(group = "train")
+public final class ResourceApplyAdadelta extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.train.ResourceApplyAdadelta}
+   */
+  public static class Options {
+    
+    /**
+     * @param useLocking If True, updating of the var, accum and update_accum tensors will be protected by
+     * a lock; otherwise the behavior is undefined, but may exhibit less contention.
+     */
+    public Options useLocking(Boolean useLocking) {
+      this.useLocking = useLocking;
+      return this;
+    }
+    
+    private Boolean useLocking;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new ResourceApplyAdadelta operation.
+   * 
+   * @param scope current scope
+   * @param var Should be from a Variable().
+   * @param accum Should be from a Variable().
+   * @param accumUpdate Should be from a Variable().
+   * @param lr Scaling factor. Must be a scalar.
+   * @param rho Decay factor. Must be a scalar.
+   * @param epsilon Constant factor. Must be a scalar.
+   * @param grad The gradient.
+   * @param options carries optional attributes values
+   * @return a new instance of ResourceApplyAdadelta
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> ResourceApplyAdadelta create(Scope scope, Operand<?> var, Operand<?> accum, Operand<?> accumUpdate, Operand<T> lr, Operand<T> rho, Operand<T> epsilon, Operand<T> grad, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ResourceApplyAdadelta", scope.makeOpName("ResourceApplyAdadelta"));
+    opBuilder.addInput(var.asOutput());
+    opBuilder.addInput(accum.asOutput());
+    opBuilder.addInput(accumUpdate.asOutput());
+    opBuilder.addInput(lr.asOutput());
+    opBuilder.addInput(rho.asOutput());
+    opBuilder.addInput(epsilon.asOutput());
+    opBuilder.addInput(grad.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.useLocking != null) {
+          opBuilder.setAttr("use_locking", opts.useLocking);
+        }
+      }
+    }
+    return new ResourceApplyAdadelta(opBuilder.build());
+  }
+  
+  /**
+   * @param useLocking If True, updating of the var, accum and update_accum tensors will be protected by
+   * a lock; otherwise the behavior is undefined, but may exhibit less contention.
+   */
+  public static Options useLocking(Boolean useLocking) {
+    return new Options().useLocking(useLocking);
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ResourceApplyAdadelta";
+  
+  private ResourceApplyAdadelta(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyAdagrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyAdagrad.java
new file mode 100644
index 00000000000..b60ad1fc0e0
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyAdagrad.java
@@ -0,0 +1,123 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.train;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Update '*var' according to the adagrad scheme.
+ * <p>
+ * accum += grad * grad
+ * var -= lr * grad * (1 / (sqrt(accum) + epsilon))
+ */
+public final class ResourceApplyAdagrad extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.train.ResourceApplyAdagrad}
+   */
+  public static class Options {
+    
+    /**
+     * @param useLocking If `True`, updating of the var and accum tensors will be protected
+     * by a lock; otherwise the behavior is undefined, but may exhibit less
+     * contention.
+     */
+    public Options useLocking(Boolean useLocking) {
+      this.useLocking = useLocking;
+      return this;
+    }
+    
+    /**
+     * @param updateSlots 
+     */
+    public Options updateSlots(Boolean updateSlots) {
+      this.updateSlots = updateSlots;
+      return this;
+    }
+    
+    private Boolean useLocking;
+    private Boolean updateSlots;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new ResourceApplyAdagrad operation.
+   * 
+   * @param scope current scope
+   * @param var Should be from a Variable().
+   * @param accum Should be from a Variable().
+   * @param lr Scaling factor. Must be a scalar.
+   * @param epsilon Constant factor. Must be a scalar.
+   * @param grad The gradient.
+   * @param options carries optional attributes values
+   * @return a new instance of ResourceApplyAdagrad
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> ResourceApplyAdagrad create(Scope scope, Operand<?> var, Operand<?> accum, Operand<T> lr, Operand<T> epsilon, Operand<T> grad, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ResourceApplyAdagradV2", scope.makeOpName("ResourceApplyAdagrad"));
+    opBuilder.addInput(var.asOutput());
+    opBuilder.addInput(accum.asOutput());
+    opBuilder.addInput(lr.asOutput());
+    opBuilder.addInput(epsilon.asOutput());
+    opBuilder.addInput(grad.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.useLocking != null) {
+          opBuilder.setAttr("use_locking", opts.useLocking);
+        }
+        if (opts.updateSlots != null) {
+          opBuilder.setAttr("update_slots", opts.updateSlots);
+        }
+      }
+    }
+    return new ResourceApplyAdagrad(opBuilder.build());
+  }
+  
+  /**
+   * @param useLocking If `True`, updating of the var and accum tensors will be protected
+   * by a lock; otherwise the behavior is undefined, but may exhibit less
+   * contention.
+   */
+  public static Options useLocking(Boolean useLocking) {
+    return new Options().useLocking(useLocking);
+  }
+  
+  /**
+   * @param updateSlots 
+   */
+  public static Options updateSlots(Boolean updateSlots) {
+    return new Options().updateSlots(updateSlots);
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ResourceApplyAdagradV2";
+  
+  private ResourceApplyAdagrad(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyAdagradDa.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyAdagradDa.java
new file mode 100644
index 00000000000..7c6f06634ef
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyAdagradDa.java
@@ -0,0 +1,107 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.train;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Update '*var' according to the proximal adagrad scheme.
+ */
+@Operator(group = "train")
+public final class ResourceApplyAdagradDa extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.train.ResourceApplyAdagradDa}
+   */
+  public static class Options {
+    
+    /**
+     * @param useLocking If True, updating of the var and accum tensors will be protected by
+     * a lock; otherwise the behavior is undefined, but may exhibit less contention.
+     */
+    public Options useLocking(Boolean useLocking) {
+      this.useLocking = useLocking;
+      return this;
+    }
+    
+    private Boolean useLocking;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new ResourceApplyAdagradDa operation.
+   * 
+   * @param scope current scope
+   * @param var Should be from a Variable().
+   * @param gradientAccumulator Should be from a Variable().
+   * @param gradientSquaredAccumulator Should be from a Variable().
+   * @param grad The gradient.
+   * @param lr Scaling factor. Must be a scalar.
+   * @param l1 L1 regularization. Must be a scalar.
+   * @param l2 L2 regularization. Must be a scalar.
+   * @param globalStep Training step number. Must be a scalar.
+   * @param options carries optional attributes values
+   * @return a new instance of ResourceApplyAdagradDa
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> ResourceApplyAdagradDa create(Scope scope, Operand<?> var, Operand<?> gradientAccumulator, Operand<?> gradientSquaredAccumulator, Operand<T> grad, Operand<T> lr, Operand<T> l1, Operand<T> l2, Operand<TInt64> globalStep, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ResourceApplyAdagradDA", scope.makeOpName("ResourceApplyAdagradDa"));
+    opBuilder.addInput(var.asOutput());
+    opBuilder.addInput(gradientAccumulator.asOutput());
+    opBuilder.addInput(gradientSquaredAccumulator.asOutput());
+    opBuilder.addInput(grad.asOutput());
+    opBuilder.addInput(lr.asOutput());
+    opBuilder.addInput(l1.asOutput());
+    opBuilder.addInput(l2.asOutput());
+    opBuilder.addInput(globalStep.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.useLocking != null) {
+          opBuilder.setAttr("use_locking", opts.useLocking);
+        }
+      }
+    }
+    return new ResourceApplyAdagradDa(opBuilder.build());
+  }
+  
+  /**
+   * @param useLocking If True, updating of the var and accum tensors will be protected by
+   * a lock; otherwise the behavior is undefined, but may exhibit less contention.
+   */
+  public static Options useLocking(Boolean useLocking) {
+    return new Options().useLocking(useLocking);
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ResourceApplyAdagradDA";
+  
+  private ResourceApplyAdagradDa(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyAdam.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyAdam.java
new file mode 100644
index 00000000000..4a1aea5d355
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyAdam.java
@@ -0,0 +1,136 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.train;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Update '*var' according to the Adam algorithm.
+ * <p>
+ * $$\text{lr}_t := \mathrm{learning_rate} * \sqrt{1 - \beta_2^t} / (1 - \beta_1^t)$$
+ * $$m_t := \beta_1 * m_{t-1} + (1 - \beta_1) * g$$
+ * $$v_t := \beta_2 * v_{t-1} + (1 - \beta_2) * g * g$$
+ * $$\text{variable} := \text{variable} - \text{lr}_t * m_t / (\sqrt{v_t} + \epsilon)$$
+ */
+@Operator(group = "train")
+public final class ResourceApplyAdam extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.train.ResourceApplyAdam}
+   */
+  public static class Options {
+    
+    /**
+     * @param useLocking If `True`, updating of the var, m, and v tensors will be protected
+     * by a lock; otherwise the behavior is undefined, but may exhibit less
+     * contention.
+     */
+    public Options useLocking(Boolean useLocking) {
+      this.useLocking = useLocking;
+      return this;
+    }
+    
+    /**
+     * @param useNesterov If `True`, uses the nesterov update.
+     */
+    public Options useNesterov(Boolean useNesterov) {
+      this.useNesterov = useNesterov;
+      return this;
+    }
+    
+    private Boolean useLocking;
+    private Boolean useNesterov;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new ResourceApplyAdam operation.
+   * 
+   * @param scope current scope
+   * @param var Should be from a Variable().
+   * @param m Should be from a Variable().
+   * @param v Should be from a Variable().
+   * @param beta1Power Must be a scalar.
+   * @param beta2Power Must be a scalar.
+   * @param lr Scaling factor. Must be a scalar.
+   * @param beta1 Momentum factor. Must be a scalar.
+   * @param beta2 Momentum factor. Must be a scalar.
+   * @param epsilon Ridge term. Must be a scalar.
+   * @param grad The gradient.
+   * @param options carries optional attributes values
+   * @return a new instance of ResourceApplyAdam
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> ResourceApplyAdam create(Scope scope, Operand<?> var, Operand<?> m, Operand<?> v, Operand<T> beta1Power, Operand<T> beta2Power, Operand<T> lr, Operand<T> beta1, Operand<T> beta2, Operand<T> epsilon, Operand<T> grad, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ResourceApplyAdam", scope.makeOpName("ResourceApplyAdam"));
+    opBuilder.addInput(var.asOutput());
+    opBuilder.addInput(m.asOutput());
+    opBuilder.addInput(v.asOutput());
+    opBuilder.addInput(beta1Power.asOutput());
+    opBuilder.addInput(beta2Power.asOutput());
+    opBuilder.addInput(lr.asOutput());
+    opBuilder.addInput(beta1.asOutput());
+    opBuilder.addInput(beta2.asOutput());
+    opBuilder.addInput(epsilon.asOutput());
+    opBuilder.addInput(grad.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.useLocking != null) {
+          opBuilder.setAttr("use_locking", opts.useLocking);
+        }
+        if (opts.useNesterov != null) {
+          opBuilder.setAttr("use_nesterov", opts.useNesterov);
+        }
+      }
+    }
+    return new ResourceApplyAdam(opBuilder.build());
+  }
+  
+  /**
+   * @param useLocking If `True`, updating of the var, m, and v tensors will be protected
+   * by a lock; otherwise the behavior is undefined, but may exhibit less
+   * contention.
+   */
+  public static Options useLocking(Boolean useLocking) {
+    return new Options().useLocking(useLocking);
+  }
+  
+  /**
+   * @param useNesterov If `True`, uses the nesterov update.
+   */
+  public static Options useNesterov(Boolean useNesterov) {
+    return new Options().useNesterov(useNesterov);
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ResourceApplyAdam";
+  
+  private ResourceApplyAdam(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyAdamWithAmsgrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyAdamWithAmsgrad.java
new file mode 100644
index 00000000000..a436bc7fdd2
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyAdamWithAmsgrad.java
@@ -0,0 +1,120 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.train;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Update '*var' according to the Adam algorithm.
+ * <p>
+ * $$\text{lr}_t := \mathrm{learning_rate} * \sqrt{1 - \beta_2^t} / (1 - \beta_1^t)$$
+ * $$m_t := \beta_1 * m_{t-1} + (1 - \beta_1) * g$$
+ * $$v_t := \beta_2 * v_{t-1} + (1 - \beta_2) * g * g$$
+ * $$\hat{v}_t := max{\hat{v}_{t-1}, v_t}$$
+ * $$\text{variable} := \text{variable} - \text{lr}_t * m_t / (\sqrt{\hat{v}_t} + \epsilon)$$
+ */
+@Operator(group = "train")
+public final class ResourceApplyAdamWithAmsgrad extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.train.ResourceApplyAdamWithAmsgrad}
+   */
+  public static class Options {
+    
+    /**
+     * @param useLocking If `True`, updating of the var, m, and v tensors will be protected
+     * by a lock; otherwise the behavior is undefined, but may exhibit less
+     * contention.
+     */
+    public Options useLocking(Boolean useLocking) {
+      this.useLocking = useLocking;
+      return this;
+    }
+    
+    private Boolean useLocking;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new ResourceApplyAdamWithAmsgrad operation.
+   * 
+   * @param scope current scope
+   * @param var Should be from a Variable().
+   * @param m Should be from a Variable().
+   * @param v Should be from a Variable().
+   * @param vhat Should be from a Variable().
+   * @param beta1Power Must be a scalar.
+   * @param beta2Power Must be a scalar.
+   * @param lr Scaling factor. Must be a scalar.
+   * @param beta1 Momentum factor. Must be a scalar.
+   * @param beta2 Momentum factor. Must be a scalar.
+   * @param epsilon Ridge term. Must be a scalar.
+   * @param grad The gradient.
+   * @param options carries optional attributes values
+   * @return a new instance of ResourceApplyAdamWithAmsgrad
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> ResourceApplyAdamWithAmsgrad create(Scope scope, Operand<?> var, Operand<?> m, Operand<?> v, Operand<?> vhat, Operand<T> beta1Power, Operand<T> beta2Power, Operand<T> lr, Operand<T> beta1, Operand<T> beta2, Operand<T> epsilon, Operand<T> grad, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ResourceApplyAdamWithAmsgrad", scope.makeOpName("ResourceApplyAdamWithAmsgrad"));
+    opBuilder.addInput(var.asOutput());
+    opBuilder.addInput(m.asOutput());
+    opBuilder.addInput(v.asOutput());
+    opBuilder.addInput(vhat.asOutput());
+    opBuilder.addInput(beta1Power.asOutput());
+    opBuilder.addInput(beta2Power.asOutput());
+    opBuilder.addInput(lr.asOutput());
+    opBuilder.addInput(beta1.asOutput());
+    opBuilder.addInput(beta2.asOutput());
+    opBuilder.addInput(epsilon.asOutput());
+    opBuilder.addInput(grad.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.useLocking != null) {
+          opBuilder.setAttr("use_locking", opts.useLocking);
+        }
+      }
+    }
+    return new ResourceApplyAdamWithAmsgrad(opBuilder.build());
+  }
+  
+  /**
+   * @param useLocking If `True`, updating of the var, m, and v tensors will be protected
+   * by a lock; otherwise the behavior is undefined, but may exhibit less
+   * contention.
+   */
+  public static Options useLocking(Boolean useLocking) {
+    return new Options().useLocking(useLocking);
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ResourceApplyAdamWithAmsgrad";
+  
+  private ResourceApplyAdamWithAmsgrad(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyAddSign.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyAddSign.java
new file mode 100644
index 00000000000..85c9c587979
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyAddSign.java
@@ -0,0 +1,110 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.train;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Update '*var' according to the AddSign update.
+ * <p>
+ * m_t <- beta1 * m_{t-1} + (1 - beta1) * g
+ * update <- (alpha + sign_decay * sign(g) *sign(m)) * g
+ * variable <- variable - lr_t * update
+ */
+@Operator(group = "train")
+public final class ResourceApplyAddSign extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.train.ResourceApplyAddSign}
+   */
+  public static class Options {
+    
+    /**
+     * @param useLocking If `True`, updating of the var and m tensors is
+     * protected by a lock; otherwise the behavior is undefined, but may exhibit less
+     * contention.
+     */
+    public Options useLocking(Boolean useLocking) {
+      this.useLocking = useLocking;
+      return this;
+    }
+    
+    private Boolean useLocking;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new ResourceApplyAddSign operation.
+   * 
+   * @param scope current scope
+   * @param var Should be from a Variable().
+   * @param m Should be from a Variable().
+   * @param lr Scaling factor. Must be a scalar.
+   * @param alpha Must be a scalar.
+   * @param signDecay Must be a scalar.
+   * @param beta Must be a scalar.
+   * @param grad The gradient.
+   * @param options carries optional attributes values
+   * @return a new instance of ResourceApplyAddSign
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> ResourceApplyAddSign create(Scope scope, Operand<?> var, Operand<?> m, Operand<T> lr, Operand<T> alpha, Operand<T> signDecay, Operand<T> beta, Operand<T> grad, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ResourceApplyAddSign", scope.makeOpName("ResourceApplyAddSign"));
+    opBuilder.addInput(var.asOutput());
+    opBuilder.addInput(m.asOutput());
+    opBuilder.addInput(lr.asOutput());
+    opBuilder.addInput(alpha.asOutput());
+    opBuilder.addInput(signDecay.asOutput());
+    opBuilder.addInput(beta.asOutput());
+    opBuilder.addInput(grad.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.useLocking != null) {
+          opBuilder.setAttr("use_locking", opts.useLocking);
+        }
+      }
+    }
+    return new ResourceApplyAddSign(opBuilder.build());
+  }
+  
+  /**
+   * @param useLocking If `True`, updating of the var and m tensors is
+   * protected by a lock; otherwise the behavior is undefined, but may exhibit less
+   * contention.
+   */
+  public static Options useLocking(Boolean useLocking) {
+    return new Options().useLocking(useLocking);
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ResourceApplyAddSign";
+  
+  private ResourceApplyAddSign(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyCenteredRmsProp.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyCenteredRmsProp.java
new file mode 100644
index 00000000000..6fc3a8a02ff
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyCenteredRmsProp.java
@@ -0,0 +1,129 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.train;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Update '*var' according to the centered RMSProp algorithm.
+ * <p>
+ * The centered RMSProp algorithm uses an estimate of the centered second moment
+ * (i.e., the variance) for normalization, as opposed to regular RMSProp, which
+ * uses the (uncentered) second moment. This often helps with training, but is
+ * slightly more expensive in terms of computation and memory.
+ * <p>
+ * Note that in dense implementation of this algorithm, mg, ms, and mom will
+ * update even if the grad is zero, but in this sparse implementation, mg, ms,
+ * and mom will not update in iterations during which the grad is zero.
+ * <p>
+ * mean_square = decay * mean_square + (1-decay) * gradient ** 2
+ * mean_grad = decay * mean_grad + (1-decay) * gradient
+ * <p>
+ * Delta = learning_rate * gradient / sqrt(mean_square + epsilon - mean_grad ** 2)
+ * <p>
+ * mg <- rho * mg_{t-1} + (1-rho) * grad
+ * ms <- rho * ms_{t-1} + (1-rho) * grad * grad
+ * mom <- momentum * mom_{t-1} + lr * grad / sqrt(ms - mg * mg + epsilon)
+ * var <- var - mom
+ */
+@Operator(group = "train")
+public final class ResourceApplyCenteredRmsProp extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.train.ResourceApplyCenteredRmsProp}
+   */
+  public static class Options {
+    
+    /**
+     * @param useLocking If `True`, updating of the var, mg, ms, and mom tensors is
+     * protected by a lock; otherwise the behavior is undefined, but may exhibit less
+     * contention.
+     */
+    public Options useLocking(Boolean useLocking) {
+      this.useLocking = useLocking;
+      return this;
+    }
+    
+    private Boolean useLocking;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new ResourceApplyCenteredRmsProp operation.
+   * 
+   * @param scope current scope
+   * @param var Should be from a Variable().
+   * @param mg Should be from a Variable().
+   * @param ms Should be from a Variable().
+   * @param mom Should be from a Variable().
+   * @param lr Scaling factor. Must be a scalar.
+   * @param rho Decay rate. Must be a scalar.
+   * @param momentum 
+   * @param epsilon Ridge term. Must be a scalar.
+   * @param grad The gradient.
+   * @param options carries optional attributes values
+   * @return a new instance of ResourceApplyCenteredRmsProp
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> ResourceApplyCenteredRmsProp create(Scope scope, Operand<?> var, Operand<?> mg, Operand<?> ms, Operand<?> mom, Operand<T> lr, Operand<T> rho, Operand<T> momentum, Operand<T> epsilon, Operand<T> grad, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ResourceApplyCenteredRMSProp", scope.makeOpName("ResourceApplyCenteredRmsProp"));
+    opBuilder.addInput(var.asOutput());
+    opBuilder.addInput(mg.asOutput());
+    opBuilder.addInput(ms.asOutput());
+    opBuilder.addInput(mom.asOutput());
+    opBuilder.addInput(lr.asOutput());
+    opBuilder.addInput(rho.asOutput());
+    opBuilder.addInput(momentum.asOutput());
+    opBuilder.addInput(epsilon.asOutput());
+    opBuilder.addInput(grad.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.useLocking != null) {
+          opBuilder.setAttr("use_locking", opts.useLocking);
+        }
+      }
+    }
+    return new ResourceApplyCenteredRmsProp(opBuilder.build());
+  }
+  
+  /**
+   * @param useLocking If `True`, updating of the var, mg, ms, and mom tensors is
+   * protected by a lock; otherwise the behavior is undefined, but may exhibit less
+   * contention.
+   */
+  public static Options useLocking(Boolean useLocking) {
+    return new Options().useLocking(useLocking);
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ResourceApplyCenteredRMSProp";
+  
+  private ResourceApplyCenteredRmsProp(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyFtrl.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyFtrl.java
new file mode 100644
index 00000000000..e69b6b99959
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyFtrl.java
@@ -0,0 +1,137 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.train;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Update '*var' according to the Ftrl-proximal scheme.
+ * <p>
+ * grad_with_shrinkage = grad + 2 * l2_shrinkage * var
+ * accum_new = accum + grad_with_shrinkage * grad_with_shrinkage
+ * linear += grad_with_shrinkage +
+ *     (accum_new^(-lr_power) - accum^(-lr_power)) / lr * var
+ * quadratic = 1.0 / (accum_new^(lr_power) * lr) + 2 * l2
+ * var = (sign(linear) * l1 - linear) / quadratic if |linear| > l1 else 0.0
+ * accum = accum_new
+ */
+@Operator(group = "train")
+public final class ResourceApplyFtrl extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.train.ResourceApplyFtrl}
+   */
+  public static class Options {
+    
+    /**
+     * @param useLocking If `True`, updating of the var and accum tensors will be protected
+     * by a lock; otherwise the behavior is undefined, but may exhibit less
+     * contention.
+     */
+    public Options useLocking(Boolean useLocking) {
+      this.useLocking = useLocking;
+      return this;
+    }
+    
+    /**
+     * @param multiplyLinearByLr 
+     */
+    public Options multiplyLinearByLr(Boolean multiplyLinearByLr) {
+      this.multiplyLinearByLr = multiplyLinearByLr;
+      return this;
+    }
+    
+    private Boolean useLocking;
+    private Boolean multiplyLinearByLr;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new ResourceApplyFtrl operation.
+   * 
+   * @param scope current scope
+   * @param var Should be from a Variable().
+   * @param accum Should be from a Variable().
+   * @param linear Should be from a Variable().
+   * @param grad The gradient.
+   * @param lr Scaling factor. Must be a scalar.
+   * @param l1 L1 regularization. Must be a scalar.
+   * @param l2 L2 shrinkage regularization. Must be a scalar.
+   * @param l2Shrinkage 
+   * @param lrPower Scaling factor. Must be a scalar.
+   * @param options carries optional attributes values
+   * @return a new instance of ResourceApplyFtrl
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> ResourceApplyFtrl create(Scope scope, Operand<?> var, Operand<?> accum, Operand<?> linear, Operand<T> grad, Operand<T> lr, Operand<T> l1, Operand<T> l2, Operand<T> l2Shrinkage, Operand<T> lrPower, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ResourceApplyFtrlV2", scope.makeOpName("ResourceApplyFtrl"));
+    opBuilder.addInput(var.asOutput());
+    opBuilder.addInput(accum.asOutput());
+    opBuilder.addInput(linear.asOutput());
+    opBuilder.addInput(grad.asOutput());
+    opBuilder.addInput(lr.asOutput());
+    opBuilder.addInput(l1.asOutput());
+    opBuilder.addInput(l2.asOutput());
+    opBuilder.addInput(l2Shrinkage.asOutput());
+    opBuilder.addInput(lrPower.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.useLocking != null) {
+          opBuilder.setAttr("use_locking", opts.useLocking);
+        }
+        if (opts.multiplyLinearByLr != null) {
+          opBuilder.setAttr("multiply_linear_by_lr", opts.multiplyLinearByLr);
+        }
+      }
+    }
+    return new ResourceApplyFtrl(opBuilder.build());
+  }
+  
+  /**
+   * @param useLocking If `True`, updating of the var and accum tensors will be protected
+   * by a lock; otherwise the behavior is undefined, but may exhibit less
+   * contention.
+   */
+  public static Options useLocking(Boolean useLocking) {
+    return new Options().useLocking(useLocking);
+  }
+  
+  /**
+   * @param multiplyLinearByLr 
+   */
+  public static Options multiplyLinearByLr(Boolean multiplyLinearByLr) {
+    return new Options().multiplyLinearByLr(multiplyLinearByLr);
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ResourceApplyFtrlV2";
+  
+  private ResourceApplyFtrl(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyGradientDescent.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyGradientDescent.java
new file mode 100644
index 00000000000..f33c6b9ca87
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyGradientDescent.java
@@ -0,0 +1,96 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.train;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Update '*var' by subtracting 'alpha' * 'delta' from it.
+ */
+@Operator(group = "train")
+public final class ResourceApplyGradientDescent extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.train.ResourceApplyGradientDescent}
+   */
+  public static class Options {
+    
+    /**
+     * @param useLocking If `True`, the subtraction will be protected by a lock;
+     * otherwise the behavior is undefined, but may exhibit less contention.
+     */
+    public Options useLocking(Boolean useLocking) {
+      this.useLocking = useLocking;
+      return this;
+    }
+    
+    private Boolean useLocking;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new ResourceApplyGradientDescent operation.
+   * 
+   * @param scope current scope
+   * @param var Should be from a Variable().
+   * @param alpha Scaling factor. Must be a scalar.
+   * @param delta The change.
+   * @param options carries optional attributes values
+   * @return a new instance of ResourceApplyGradientDescent
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> ResourceApplyGradientDescent create(Scope scope, Operand<?> var, Operand<T> alpha, Operand<T> delta, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ResourceApplyGradientDescent", scope.makeOpName("ResourceApplyGradientDescent"));
+    opBuilder.addInput(var.asOutput());
+    opBuilder.addInput(alpha.asOutput());
+    opBuilder.addInput(delta.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.useLocking != null) {
+          opBuilder.setAttr("use_locking", opts.useLocking);
+        }
+      }
+    }
+    return new ResourceApplyGradientDescent(opBuilder.build());
+  }
+  
+  /**
+   * @param useLocking If `True`, the subtraction will be protected by a lock;
+   * otherwise the behavior is undefined, but may exhibit less contention.
+   */
+  public static Options useLocking(Boolean useLocking) {
+    return new Options().useLocking(useLocking);
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ResourceApplyGradientDescent";
+  
+  private ResourceApplyGradientDescent(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyKerasMomentum.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyKerasMomentum.java
new file mode 100644
index 00000000000..3922439dcad
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyKerasMomentum.java
@@ -0,0 +1,130 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.train;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Update '*var' according to the momentum scheme.
+ * <p>
+ * Set use_nesterov = True if you want to use Nesterov momentum.
+ * <p>
+ * accum = accum * momentum - lr * grad
+ * var += accum
+ */
+@Operator(group = "train")
+public final class ResourceApplyKerasMomentum extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.train.ResourceApplyKerasMomentum}
+   */
+  public static class Options {
+    
+    /**
+     * @param useLocking If `True`, updating of the var and accum tensors will be protected
+     * by a lock; otherwise the behavior is undefined, but may exhibit less
+     * contention.
+     */
+    public Options useLocking(Boolean useLocking) {
+      this.useLocking = useLocking;
+      return this;
+    }
+    
+    /**
+     * @param useNesterov If `True`, the tensor passed to compute grad will be
+     * var + momentum * accum, so in the end, the var you get is actually
+     * var + momentum * accum.
+     */
+    public Options useNesterov(Boolean useNesterov) {
+      this.useNesterov = useNesterov;
+      return this;
+    }
+    
+    private Boolean useLocking;
+    private Boolean useNesterov;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new ResourceApplyKerasMomentum operation.
+   * 
+   * @param scope current scope
+   * @param var Should be from a Variable().
+   * @param accum Should be from a Variable().
+   * @param lr Scaling factor. Must be a scalar.
+   * @param grad The gradient.
+   * @param momentum Momentum. Must be a scalar.
+   * @param options carries optional attributes values
+   * @return a new instance of ResourceApplyKerasMomentum
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> ResourceApplyKerasMomentum create(Scope scope, Operand<?> var, Operand<?> accum, Operand<T> lr, Operand<T> grad, Operand<T> momentum, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ResourceApplyKerasMomentum", scope.makeOpName("ResourceApplyKerasMomentum"));
+    opBuilder.addInput(var.asOutput());
+    opBuilder.addInput(accum.asOutput());
+    opBuilder.addInput(lr.asOutput());
+    opBuilder.addInput(grad.asOutput());
+    opBuilder.addInput(momentum.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.useLocking != null) {
+          opBuilder.setAttr("use_locking", opts.useLocking);
+        }
+        if (opts.useNesterov != null) {
+          opBuilder.setAttr("use_nesterov", opts.useNesterov);
+        }
+      }
+    }
+    return new ResourceApplyKerasMomentum(opBuilder.build());
+  }
+  
+  /**
+   * @param useLocking If `True`, updating of the var and accum tensors will be protected
+   * by a lock; otherwise the behavior is undefined, but may exhibit less
+   * contention.
+   */
+  public static Options useLocking(Boolean useLocking) {
+    return new Options().useLocking(useLocking);
+  }
+  
+  /**
+   * @param useNesterov If `True`, the tensor passed to compute grad will be
+   * var + momentum * accum, so in the end, the var you get is actually
+   * var + momentum * accum.
+   */
+  public static Options useNesterov(Boolean useNesterov) {
+    return new Options().useNesterov(useNesterov);
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ResourceApplyKerasMomentum";
+  
+  private ResourceApplyKerasMomentum(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyMomentum.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyMomentum.java
new file mode 100644
index 00000000000..c554c8a939d
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyMomentum.java
@@ -0,0 +1,130 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.train;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Update '*var' according to the momentum scheme.
+ * <p>
+ * Set use_nesterov = True if you want to use Nesterov momentum.
+ * <p>
+ * accum = accum * momentum + grad
+ * var -= lr * accum
+ */
+@Operator(group = "train")
+public final class ResourceApplyMomentum extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.train.ResourceApplyMomentum}
+   */
+  public static class Options {
+    
+    /**
+     * @param useLocking If `True`, updating of the var and accum tensors will be protected
+     * by a lock; otherwise the behavior is undefined, but may exhibit less
+     * contention.
+     */
+    public Options useLocking(Boolean useLocking) {
+      this.useLocking = useLocking;
+      return this;
+    }
+    
+    /**
+     * @param useNesterov If `True`, the tensor passed to compute grad will be
+     * var - lr * momentum * accum, so in the end, the var you get is actually
+     * var - lr * momentum * accum.
+     */
+    public Options useNesterov(Boolean useNesterov) {
+      this.useNesterov = useNesterov;
+      return this;
+    }
+    
+    private Boolean useLocking;
+    private Boolean useNesterov;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new ResourceApplyMomentum operation.
+   * 
+   * @param scope current scope
+   * @param var Should be from a Variable().
+   * @param accum Should be from a Variable().
+   * @param lr Scaling factor. Must be a scalar.
+   * @param grad The gradient.
+   * @param momentum Momentum. Must be a scalar.
+   * @param options carries optional attributes values
+   * @return a new instance of ResourceApplyMomentum
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> ResourceApplyMomentum create(Scope scope, Operand<?> var, Operand<?> accum, Operand<T> lr, Operand<T> grad, Operand<T> momentum, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ResourceApplyMomentum", scope.makeOpName("ResourceApplyMomentum"));
+    opBuilder.addInput(var.asOutput());
+    opBuilder.addInput(accum.asOutput());
+    opBuilder.addInput(lr.asOutput());
+    opBuilder.addInput(grad.asOutput());
+    opBuilder.addInput(momentum.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.useLocking != null) {
+          opBuilder.setAttr("use_locking", opts.useLocking);
+        }
+        if (opts.useNesterov != null) {
+          opBuilder.setAttr("use_nesterov", opts.useNesterov);
+        }
+      }
+    }
+    return new ResourceApplyMomentum(opBuilder.build());
+  }
+  
+  /**
+   * @param useLocking If `True`, updating of the var and accum tensors will be protected
+   * by a lock; otherwise the behavior is undefined, but may exhibit less
+   * contention.
+   */
+  public static Options useLocking(Boolean useLocking) {
+    return new Options().useLocking(useLocking);
+  }
+  
+  /**
+   * @param useNesterov If `True`, the tensor passed to compute grad will be
+   * var - lr * momentum * accum, so in the end, the var you get is actually
+   * var - lr * momentum * accum.
+   */
+  public static Options useNesterov(Boolean useNesterov) {
+    return new Options().useNesterov(useNesterov);
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ResourceApplyMomentum";
+  
+  private ResourceApplyMomentum(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyPowerSign.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyPowerSign.java
new file mode 100644
index 00000000000..662c2253264
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyPowerSign.java
@@ -0,0 +1,110 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.train;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Update '*var' according to the AddSign update.
+ * <p>
+ * m_t <- beta1 * m_{t-1} + (1 - beta1) * g
+ * update <- exp(logbase * sign_decay * sign(g) * sign(m_t)) * g
+ * variable <- variable - lr_t * update
+ */
+@Operator(group = "train")
+public final class ResourceApplyPowerSign extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.train.ResourceApplyPowerSign}
+   */
+  public static class Options {
+    
+    /**
+     * @param useLocking If `True`, updating of the var and m tensors is
+     * protected by a lock; otherwise the behavior is undefined, but may exhibit less
+     * contention.
+     */
+    public Options useLocking(Boolean useLocking) {
+      this.useLocking = useLocking;
+      return this;
+    }
+    
+    private Boolean useLocking;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new ResourceApplyPowerSign operation.
+   * 
+   * @param scope current scope
+   * @param var Should be from a Variable().
+   * @param m Should be from a Variable().
+   * @param lr Scaling factor. Must be a scalar.
+   * @param logbase Must be a scalar.
+   * @param signDecay Must be a scalar.
+   * @param beta Must be a scalar.
+   * @param grad The gradient.
+   * @param options carries optional attributes values
+   * @return a new instance of ResourceApplyPowerSign
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> ResourceApplyPowerSign create(Scope scope, Operand<?> var, Operand<?> m, Operand<T> lr, Operand<T> logbase, Operand<T> signDecay, Operand<T> beta, Operand<T> grad, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ResourceApplyPowerSign", scope.makeOpName("ResourceApplyPowerSign"));
+    opBuilder.addInput(var.asOutput());
+    opBuilder.addInput(m.asOutput());
+    opBuilder.addInput(lr.asOutput());
+    opBuilder.addInput(logbase.asOutput());
+    opBuilder.addInput(signDecay.asOutput());
+    opBuilder.addInput(beta.asOutput());
+    opBuilder.addInput(grad.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.useLocking != null) {
+          opBuilder.setAttr("use_locking", opts.useLocking);
+        }
+      }
+    }
+    return new ResourceApplyPowerSign(opBuilder.build());
+  }
+  
+  /**
+   * @param useLocking If `True`, updating of the var and m tensors is
+   * protected by a lock; otherwise the behavior is undefined, but may exhibit less
+   * contention.
+   */
+  public static Options useLocking(Boolean useLocking) {
+    return new Options().useLocking(useLocking);
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ResourceApplyPowerSign";
+  
+  private ResourceApplyPowerSign(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyProximalAdagrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyProximalAdagrad.java
new file mode 100644
index 00000000000..8036d891e33
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyProximalAdagrad.java
@@ -0,0 +1,106 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.train;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Update '*var' and '*accum' according to FOBOS with Adagrad learning rate.
+ * <p>
+ * accum += grad <i> grad
+ * prox_v = var - lr </i> grad <i> (1 / sqrt(accum))
+ * var = sign(prox_v)/(1+lr</i>l2) <i> max{|prox_v|-lr</i>l1,0}
+ */
+@Operator(group = "train")
+public final class ResourceApplyProximalAdagrad extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.train.ResourceApplyProximalAdagrad}
+   */
+  public static class Options {
+    
+    /**
+     * @param useLocking If True, updating of the var and accum tensors will be protected by
+     * a lock; otherwise the behavior is undefined, but may exhibit less contention.
+     */
+    public Options useLocking(Boolean useLocking) {
+      this.useLocking = useLocking;
+      return this;
+    }
+    
+    private Boolean useLocking;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new ResourceApplyProximalAdagrad operation.
+   * 
+   * @param scope current scope
+   * @param var Should be from a Variable().
+   * @param accum Should be from a Variable().
+   * @param lr Scaling factor. Must be a scalar.
+   * @param l1 L1 regularization. Must be a scalar.
+   * @param l2 L2 regularization. Must be a scalar.
+   * @param grad The gradient.
+   * @param options carries optional attributes values
+   * @return a new instance of ResourceApplyProximalAdagrad
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> ResourceApplyProximalAdagrad create(Scope scope, Operand<?> var, Operand<?> accum, Operand<T> lr, Operand<T> l1, Operand<T> l2, Operand<T> grad, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ResourceApplyProximalAdagrad", scope.makeOpName("ResourceApplyProximalAdagrad"));
+    opBuilder.addInput(var.asOutput());
+    opBuilder.addInput(accum.asOutput());
+    opBuilder.addInput(lr.asOutput());
+    opBuilder.addInput(l1.asOutput());
+    opBuilder.addInput(l2.asOutput());
+    opBuilder.addInput(grad.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.useLocking != null) {
+          opBuilder.setAttr("use_locking", opts.useLocking);
+        }
+      }
+    }
+    return new ResourceApplyProximalAdagrad(opBuilder.build());
+  }
+  
+  /**
+   * @param useLocking If True, updating of the var and accum tensors will be protected by
+   * a lock; otherwise the behavior is undefined, but may exhibit less contention.
+   */
+  public static Options useLocking(Boolean useLocking) {
+    return new Options().useLocking(useLocking);
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ResourceApplyProximalAdagrad";
+  
+  private ResourceApplyProximalAdagrad(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyProximalGradientDescent.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyProximalGradientDescent.java
new file mode 100644
index 00000000000..3b217c88c67
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyProximalGradientDescent.java
@@ -0,0 +1,103 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.train;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Update '*var' as FOBOS algorithm with fixed learning rate.
+ * <p>
+ * prox_v = var - alpha <i> delta
+ * var = sign(prox_v)/(1+alpha</i>l2) <i> max{|prox_v|-alpha</i>l1,0}
+ */
+@Operator(group = "train")
+public final class ResourceApplyProximalGradientDescent extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.train.ResourceApplyProximalGradientDescent}
+   */
+  public static class Options {
+    
+    /**
+     * @param useLocking If True, the subtraction will be protected by a lock;
+     * otherwise the behavior is undefined, but may exhibit less contention.
+     */
+    public Options useLocking(Boolean useLocking) {
+      this.useLocking = useLocking;
+      return this;
+    }
+    
+    private Boolean useLocking;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new ResourceApplyProximalGradientDescent operation.
+   * 
+   * @param scope current scope
+   * @param var Should be from a Variable().
+   * @param alpha Scaling factor. Must be a scalar.
+   * @param l1 L1 regularization. Must be a scalar.
+   * @param l2 L2 regularization. Must be a scalar.
+   * @param delta The change.
+   * @param options carries optional attributes values
+   * @return a new instance of ResourceApplyProximalGradientDescent
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> ResourceApplyProximalGradientDescent create(Scope scope, Operand<?> var, Operand<T> alpha, Operand<T> l1, Operand<T> l2, Operand<T> delta, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ResourceApplyProximalGradientDescent", scope.makeOpName("ResourceApplyProximalGradientDescent"));
+    opBuilder.addInput(var.asOutput());
+    opBuilder.addInput(alpha.asOutput());
+    opBuilder.addInput(l1.asOutput());
+    opBuilder.addInput(l2.asOutput());
+    opBuilder.addInput(delta.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.useLocking != null) {
+          opBuilder.setAttr("use_locking", opts.useLocking);
+        }
+      }
+    }
+    return new ResourceApplyProximalGradientDescent(opBuilder.build());
+  }
+  
+  /**
+   * @param useLocking If True, the subtraction will be protected by a lock;
+   * otherwise the behavior is undefined, but may exhibit less contention.
+   */
+  public static Options useLocking(Boolean useLocking) {
+    return new Options().useLocking(useLocking);
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ResourceApplyProximalGradientDescent";
+  
+  private ResourceApplyProximalGradientDescent(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyRmsProp.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyRmsProp.java
new file mode 100644
index 00000000000..ae42295c1f7
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceApplyRmsProp.java
@@ -0,0 +1,119 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.train;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Update '*var' according to the RMSProp algorithm.
+ * <p>
+ * Note that in dense implementation of this algorithm, ms and mom will
+ * update even if the grad is zero, but in this sparse implementation, ms
+ * and mom will not update in iterations during which the grad is zero.
+ * <p>
+ * mean_square = decay * mean_square + (1-decay) * gradient ** 2
+ * Delta = learning_rate * gradient / sqrt(mean_square + epsilon)
+ * <p>
+ * ms <- rho * ms_{t-1} + (1-rho) * grad * grad
+ * mom <- momentum * mom_{t-1} + lr * grad / sqrt(ms + epsilon)
+ * var <- var - mom
+ */
+@Operator(group = "train")
+public final class ResourceApplyRmsProp extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.train.ResourceApplyRmsProp}
+   */
+  public static class Options {
+    
+    /**
+     * @param useLocking If `True`, updating of the var, ms, and mom tensors is protected
+     * by a lock; otherwise the behavior is undefined, but may exhibit less
+     * contention.
+     */
+    public Options useLocking(Boolean useLocking) {
+      this.useLocking = useLocking;
+      return this;
+    }
+    
+    private Boolean useLocking;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new ResourceApplyRmsProp operation.
+   * 
+   * @param scope current scope
+   * @param var Should be from a Variable().
+   * @param ms Should be from a Variable().
+   * @param mom Should be from a Variable().
+   * @param lr Scaling factor. Must be a scalar.
+   * @param rho Decay rate. Must be a scalar.
+   * @param momentum 
+   * @param epsilon Ridge term. Must be a scalar.
+   * @param grad The gradient.
+   * @param options carries optional attributes values
+   * @return a new instance of ResourceApplyRmsProp
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> ResourceApplyRmsProp create(Scope scope, Operand<?> var, Operand<?> ms, Operand<?> mom, Operand<T> lr, Operand<T> rho, Operand<T> momentum, Operand<T> epsilon, Operand<T> grad, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ResourceApplyRMSProp", scope.makeOpName("ResourceApplyRmsProp"));
+    opBuilder.addInput(var.asOutput());
+    opBuilder.addInput(ms.asOutput());
+    opBuilder.addInput(mom.asOutput());
+    opBuilder.addInput(lr.asOutput());
+    opBuilder.addInput(rho.asOutput());
+    opBuilder.addInput(momentum.asOutput());
+    opBuilder.addInput(epsilon.asOutput());
+    opBuilder.addInput(grad.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.useLocking != null) {
+          opBuilder.setAttr("use_locking", opts.useLocking);
+        }
+      }
+    }
+    return new ResourceApplyRmsProp(opBuilder.build());
+  }
+  
+  /**
+   * @param useLocking If `True`, updating of the var, ms, and mom tensors is protected
+   * by a lock; otherwise the behavior is undefined, but may exhibit less
+   * contention.
+   */
+  public static Options useLocking(Boolean useLocking) {
+    return new Options().useLocking(useLocking);
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ResourceApplyRMSProp";
+  
+  private ResourceApplyRmsProp(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceConditionalAccumulator.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceConditionalAccumulator.java
new file mode 100644
index 00000000000..9bc3453a72a
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceConditionalAccumulator.java
@@ -0,0 +1,161 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.train;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * A conditional accumulator for aggregating gradients.
+ * <p>
+ * The accumulator accepts gradients marked with local_step greater or
+ * equal to the most recent global_step known to the accumulator. The
+ * average can be extracted from the accumulator, provided sufficient
+ * gradients have been accumulated. Extracting the average automatically
+ * resets the aggregate to 0, and increments the global_step recorded by
+ * the accumulator.
+ * This is a resource version of ConditionalAccumulator that will work in TF2.0
+ * with tf.cond version 2.
+ */
+public final class ResourceConditionalAccumulator extends RawOp implements Operand<TType> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.train.ResourceConditionalAccumulator}
+   */
+  public static class Options {
+    
+    /**
+     * @param container If non-empty, this accumulator is placed in the given container.
+     * Otherwise, a default container is used.
+     */
+    public Options container(String container) {
+      this.container = container;
+      return this;
+    }
+    
+    /**
+     * @param sharedName If non-empty, this accumulator will be shared under the
+     * given name across multiple sessions.
+     */
+    public Options sharedName(String sharedName) {
+      this.sharedName = sharedName;
+      return this;
+    }
+    
+    /**
+     * @param reductionType 
+     */
+    public Options reductionType(String reductionType) {
+      this.reductionType = reductionType;
+      return this;
+    }
+    
+    private String container;
+    private String sharedName;
+    private String reductionType;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new ResourceConditionalAccumulator operation.
+   * 
+   * @param scope current scope
+   * @param dtype The type of the value being accumulated.
+   * @param shape The shape of the values, can be [], in which case shape is unknown.
+   * @param options carries optional attributes values
+   * @return a new instance of ResourceConditionalAccumulator
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> ResourceConditionalAccumulator create(Scope scope, Class<T> dtype, Shape shape, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ResourceConditionalAccumulator", scope.makeOpName("ResourceConditionalAccumulator"));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("dtype", dtype);
+    opBuilder.setAttr("shape", shape);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.container != null) {
+          opBuilder.setAttr("container", opts.container);
+        }
+        if (opts.sharedName != null) {
+          opBuilder.setAttr("shared_name", opts.sharedName);
+        }
+        if (opts.reductionType != null) {
+          opBuilder.setAttr("reduction_type", opts.reductionType);
+        }
+      }
+    }
+    return new ResourceConditionalAccumulator(opBuilder.build());
+  }
+  
+  /**
+   * @param container If non-empty, this accumulator is placed in the given container.
+   * Otherwise, a default container is used.
+   */
+  public static Options container(String container) {
+    return new Options().container(container);
+  }
+  
+  /**
+   * @param sharedName If non-empty, this accumulator will be shared under the
+   * given name across multiple sessions.
+   */
+  public static Options sharedName(String sharedName) {
+    return new Options().sharedName(sharedName);
+  }
+  
+  /**
+   * @param reductionType 
+   */
+  public static Options reductionType(String reductionType) {
+    return new Options().reductionType(reductionType);
+  }
+  
+  /**
+   * The handle to the accumulator.
+   */
+  public Output<?> handle() {
+    return handle;
+  }
+  
+  @Override
+  @SuppressWarnings("unchecked")
+  public Output<TType> asOutput() {
+    return (Output<TType>) handle;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ResourceConditionalAccumulator";
+  
+  private Output<?> handle;
+  
+  private ResourceConditionalAccumulator(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    handle = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyAdadelta.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyAdadelta.java
new file mode 100644
index 00000000000..baea98fc1f7
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyAdadelta.java
@@ -0,0 +1,107 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.train;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * var: Should be from a Variable().
+ */
+@Operator(group = "train")
+public final class ResourceSparseApplyAdadelta extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.train.ResourceSparseApplyAdadelta}
+   */
+  public static class Options {
+    
+    /**
+     * @param useLocking If True, updating of the var and accum tensors will be protected by
+     * a lock; otherwise the behavior is undefined, but may exhibit less contention.
+     */
+    public Options useLocking(Boolean useLocking) {
+      this.useLocking = useLocking;
+      return this;
+    }
+    
+    private Boolean useLocking;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new ResourceSparseApplyAdadelta operation.
+   * 
+   * @param scope current scope
+   * @param var 
+   * @param accum Should be from a Variable().
+   * @param accumUpdate : Should be from a Variable().
+   * @param lr Learning rate. Must be a scalar.
+   * @param rho Decay factor. Must be a scalar.
+   * @param epsilon Constant factor. Must be a scalar.
+   * @param grad The gradient.
+   * @param indices A vector of indices into the first dimension of var and accum.
+   * @param options carries optional attributes values
+   * @return a new instance of ResourceSparseApplyAdadelta
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType, U extends TNumber> ResourceSparseApplyAdadelta create(Scope scope, Operand<?> var, Operand<?> accum, Operand<?> accumUpdate, Operand<T> lr, Operand<T> rho, Operand<T> epsilon, Operand<T> grad, Operand<U> indices, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ResourceSparseApplyAdadelta", scope.makeOpName("ResourceSparseApplyAdadelta"));
+    opBuilder.addInput(var.asOutput());
+    opBuilder.addInput(accum.asOutput());
+    opBuilder.addInput(accumUpdate.asOutput());
+    opBuilder.addInput(lr.asOutput());
+    opBuilder.addInput(rho.asOutput());
+    opBuilder.addInput(epsilon.asOutput());
+    opBuilder.addInput(grad.asOutput());
+    opBuilder.addInput(indices.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.useLocking != null) {
+          opBuilder.setAttr("use_locking", opts.useLocking);
+        }
+      }
+    }
+    return new ResourceSparseApplyAdadelta(opBuilder.build());
+  }
+  
+  /**
+   * @param useLocking If True, updating of the var and accum tensors will be protected by
+   * a lock; otherwise the behavior is undefined, but may exhibit less contention.
+   */
+  public static Options useLocking(Boolean useLocking) {
+    return new Options().useLocking(useLocking);
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ResourceSparseApplyAdadelta";
+  
+  private ResourceSparseApplyAdadelta(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyAdagrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyAdagrad.java
new file mode 100644
index 00000000000..f7816e78d0c
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyAdagrad.java
@@ -0,0 +1,126 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.train;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Update relevant entries in '*var' and '*accum' according to the adagrad scheme.
+ * <p>
+ * That is for rows we have grad for, we update var and accum as follows:
+ * accum += grad * grad
+ * var -= lr * grad * (1 / sqrt(accum))
+ */
+@Operator(group = "train")
+public final class ResourceSparseApplyAdagrad extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.train.ResourceSparseApplyAdagrad}
+   */
+  public static class Options {
+    
+    /**
+     * @param useLocking If `True`, updating of the var and accum tensors will be protected
+     * by a lock; otherwise the behavior is undefined, but may exhibit less
+     * contention.
+     */
+    public Options useLocking(Boolean useLocking) {
+      this.useLocking = useLocking;
+      return this;
+    }
+    
+    /**
+     * @param updateSlots 
+     */
+    public Options updateSlots(Boolean updateSlots) {
+      this.updateSlots = updateSlots;
+      return this;
+    }
+    
+    private Boolean useLocking;
+    private Boolean updateSlots;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new ResourceSparseApplyAdagrad operation.
+   * 
+   * @param scope current scope
+   * @param var Should be from a Variable().
+   * @param accum Should be from a Variable().
+   * @param lr Learning rate. Must be a scalar.
+   * @param grad The gradient.
+   * @param indices A vector of indices into the first dimension of var and accum.
+   * @param options carries optional attributes values
+   * @return a new instance of ResourceSparseApplyAdagrad
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType, U extends TNumber> ResourceSparseApplyAdagrad create(Scope scope, Operand<?> var, Operand<?> accum, Operand<T> lr, Operand<T> grad, Operand<U> indices, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ResourceSparseApplyAdagrad", scope.makeOpName("ResourceSparseApplyAdagrad"));
+    opBuilder.addInput(var.asOutput());
+    opBuilder.addInput(accum.asOutput());
+    opBuilder.addInput(lr.asOutput());
+    opBuilder.addInput(grad.asOutput());
+    opBuilder.addInput(indices.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.useLocking != null) {
+          opBuilder.setAttr("use_locking", opts.useLocking);
+        }
+        if (opts.updateSlots != null) {
+          opBuilder.setAttr("update_slots", opts.updateSlots);
+        }
+      }
+    }
+    return new ResourceSparseApplyAdagrad(opBuilder.build());
+  }
+  
+  /**
+   * @param useLocking If `True`, updating of the var and accum tensors will be protected
+   * by a lock; otherwise the behavior is undefined, but may exhibit less
+   * contention.
+   */
+  public static Options useLocking(Boolean useLocking) {
+    return new Options().useLocking(useLocking);
+  }
+  
+  /**
+   * @param updateSlots 
+   */
+  public static Options updateSlots(Boolean updateSlots) {
+    return new Options().updateSlots(updateSlots);
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ResourceSparseApplyAdagrad";
+  
+  private ResourceSparseApplyAdagrad(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyAdagradDa.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyAdagradDa.java
new file mode 100644
index 00000000000..417eca86a80
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyAdagradDa.java
@@ -0,0 +1,110 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.train;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Update entries in '*var' and '*accum' according to the proximal adagrad scheme.
+ */
+@Operator(group = "train")
+public final class ResourceSparseApplyAdagradDa extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.train.ResourceSparseApplyAdagradDa}
+   */
+  public static class Options {
+    
+    /**
+     * @param useLocking If True, updating of the var and accum tensors will be protected by
+     * a lock; otherwise the behavior is undefined, but may exhibit less contention.
+     */
+    public Options useLocking(Boolean useLocking) {
+      this.useLocking = useLocking;
+      return this;
+    }
+    
+    private Boolean useLocking;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new ResourceSparseApplyAdagradDa operation.
+   * 
+   * @param scope current scope
+   * @param var Should be from a Variable().
+   * @param gradientAccumulator Should be from a Variable().
+   * @param gradientSquaredAccumulator Should be from a Variable().
+   * @param grad The gradient.
+   * @param indices A vector of indices into the first dimension of var and accum.
+   * @param lr Learning rate. Must be a scalar.
+   * @param l1 L1 regularization. Must be a scalar.
+   * @param l2 L2 regularization. Must be a scalar.
+   * @param globalStep Training step number. Must be a scalar.
+   * @param options carries optional attributes values
+   * @return a new instance of ResourceSparseApplyAdagradDa
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType, U extends TNumber> ResourceSparseApplyAdagradDa create(Scope scope, Operand<?> var, Operand<?> gradientAccumulator, Operand<?> gradientSquaredAccumulator, Operand<T> grad, Operand<U> indices, Operand<T> lr, Operand<T> l1, Operand<T> l2, Operand<TInt64> globalStep, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ResourceSparseApplyAdagradDA", scope.makeOpName("ResourceSparseApplyAdagradDa"));
+    opBuilder.addInput(var.asOutput());
+    opBuilder.addInput(gradientAccumulator.asOutput());
+    opBuilder.addInput(gradientSquaredAccumulator.asOutput());
+    opBuilder.addInput(grad.asOutput());
+    opBuilder.addInput(indices.asOutput());
+    opBuilder.addInput(lr.asOutput());
+    opBuilder.addInput(l1.asOutput());
+    opBuilder.addInput(l2.asOutput());
+    opBuilder.addInput(globalStep.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.useLocking != null) {
+          opBuilder.setAttr("use_locking", opts.useLocking);
+        }
+      }
+    }
+    return new ResourceSparseApplyAdagradDa(opBuilder.build());
+  }
+  
+  /**
+   * @param useLocking If True, updating of the var and accum tensors will be protected by
+   * a lock; otherwise the behavior is undefined, but may exhibit less contention.
+   */
+  public static Options useLocking(Boolean useLocking) {
+    return new Options().useLocking(useLocking);
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ResourceSparseApplyAdagradDA";
+  
+  private ResourceSparseApplyAdagradDa(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyAdagradV2.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyAdagradV2.java
new file mode 100644
index 00000000000..f60d192c368
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyAdagradV2.java
@@ -0,0 +1,127 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.train;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Update relevant entries in '*var' and '*accum' according to the adagrad scheme.
+ * <p>
+ * That is for rows we have grad for, we update var and accum as follows:
+ * accum += grad * grad
+ * var -= lr * grad * (1 / sqrt(accum))
+ */
+public final class ResourceSparseApplyAdagradV2 extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.train.ResourceSparseApplyAdagradV2}
+   */
+  public static class Options {
+    
+    /**
+     * @param useLocking If `True`, updating of the var and accum tensors will be protected
+     * by a lock; otherwise the behavior is undefined, but may exhibit less
+     * contention.
+     */
+    public Options useLocking(Boolean useLocking) {
+      this.useLocking = useLocking;
+      return this;
+    }
+    
+    /**
+     * @param updateSlots 
+     */
+    public Options updateSlots(Boolean updateSlots) {
+      this.updateSlots = updateSlots;
+      return this;
+    }
+    
+    private Boolean useLocking;
+    private Boolean updateSlots;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new ResourceSparseApplyAdagradV2 operation.
+   * 
+   * @param scope current scope
+   * @param var Should be from a Variable().
+   * @param accum Should be from a Variable().
+   * @param lr Learning rate. Must be a scalar.
+   * @param epsilon Constant factor. Must be a scalar.
+   * @param grad The gradient.
+   * @param indices A vector of indices into the first dimension of var and accum.
+   * @param options carries optional attributes values
+   * @return a new instance of ResourceSparseApplyAdagradV2
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType, U extends TNumber> ResourceSparseApplyAdagradV2 create(Scope scope, Operand<?> var, Operand<?> accum, Operand<T> lr, Operand<T> epsilon, Operand<T> grad, Operand<U> indices, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ResourceSparseApplyAdagradV2", scope.makeOpName("ResourceSparseApplyAdagradV2"));
+    opBuilder.addInput(var.asOutput());
+    opBuilder.addInput(accum.asOutput());
+    opBuilder.addInput(lr.asOutput());
+    opBuilder.addInput(epsilon.asOutput());
+    opBuilder.addInput(grad.asOutput());
+    opBuilder.addInput(indices.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.useLocking != null) {
+          opBuilder.setAttr("use_locking", opts.useLocking);
+        }
+        if (opts.updateSlots != null) {
+          opBuilder.setAttr("update_slots", opts.updateSlots);
+        }
+      }
+    }
+    return new ResourceSparseApplyAdagradV2(opBuilder.build());
+  }
+  
+  /**
+   * @param useLocking If `True`, updating of the var and accum tensors will be protected
+   * by a lock; otherwise the behavior is undefined, but may exhibit less
+   * contention.
+   */
+  public static Options useLocking(Boolean useLocking) {
+    return new Options().useLocking(useLocking);
+  }
+  
+  /**
+   * @param updateSlots 
+   */
+  public static Options updateSlots(Boolean updateSlots) {
+    return new Options().updateSlots(updateSlots);
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ResourceSparseApplyAdagradV2";
+  
+  private ResourceSparseApplyAdagradV2(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyCenteredRmsProp.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyCenteredRmsProp.java
new file mode 100644
index 00000000000..d6806c36abf
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyCenteredRmsProp.java
@@ -0,0 +1,130 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.train;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Update '*var' according to the centered RMSProp algorithm.
+ * <p>
+ * The centered RMSProp algorithm uses an estimate of the centered second moment
+ * (i.e., the variance) for normalization, as opposed to regular RMSProp, which
+ * uses the (uncentered) second moment. This often helps with training, but is
+ * slightly more expensive in terms of computation and memory.
+ * <p>
+ * Note that in dense implementation of this algorithm, mg, ms, and mom will
+ * update even if the grad is zero, but in this sparse implementation, mg, ms,
+ * and mom will not update in iterations during which the grad is zero.
+ * <p>
+ * mean_square = decay * mean_square + (1-decay) * gradient ** 2
+ * mean_grad = decay * mean_grad + (1-decay) * gradient
+ * Delta = learning_rate * gradient / sqrt(mean_square + epsilon - mean_grad ** 2)
+ * <p>
+ * ms <- rho * ms_{t-1} + (1-rho) * grad * grad
+ * mom <- momentum * mom_{t-1} + lr * grad / sqrt(ms + epsilon)
+ * var <- var - mom
+ */
+@Operator(group = "train")
+public final class ResourceSparseApplyCenteredRmsProp extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.train.ResourceSparseApplyCenteredRmsProp}
+   */
+  public static class Options {
+    
+    /**
+     * @param useLocking If `True`, updating of the var, mg, ms, and mom tensors is
+     * protected by a lock; otherwise the behavior is undefined, but may exhibit less
+     * contention.
+     */
+    public Options useLocking(Boolean useLocking) {
+      this.useLocking = useLocking;
+      return this;
+    }
+    
+    private Boolean useLocking;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new ResourceSparseApplyCenteredRmsProp operation.
+   * 
+   * @param scope current scope
+   * @param var Should be from a Variable().
+   * @param mg Should be from a Variable().
+   * @param ms Should be from a Variable().
+   * @param mom Should be from a Variable().
+   * @param lr Scaling factor. Must be a scalar.
+   * @param rho Decay rate. Must be a scalar.
+   * @param momentum 
+   * @param epsilon Ridge term. Must be a scalar.
+   * @param grad The gradient.
+   * @param indices A vector of indices into the first dimension of var, ms and mom.
+   * @param options carries optional attributes values
+   * @return a new instance of ResourceSparseApplyCenteredRmsProp
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType, U extends TNumber> ResourceSparseApplyCenteredRmsProp create(Scope scope, Operand<?> var, Operand<?> mg, Operand<?> ms, Operand<?> mom, Operand<T> lr, Operand<T> rho, Operand<T> momentum, Operand<T> epsilon, Operand<T> grad, Operand<U> indices, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ResourceSparseApplyCenteredRMSProp", scope.makeOpName("ResourceSparseApplyCenteredRmsProp"));
+    opBuilder.addInput(var.asOutput());
+    opBuilder.addInput(mg.asOutput());
+    opBuilder.addInput(ms.asOutput());
+    opBuilder.addInput(mom.asOutput());
+    opBuilder.addInput(lr.asOutput());
+    opBuilder.addInput(rho.asOutput());
+    opBuilder.addInput(momentum.asOutput());
+    opBuilder.addInput(epsilon.asOutput());
+    opBuilder.addInput(grad.asOutput());
+    opBuilder.addInput(indices.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.useLocking != null) {
+          opBuilder.setAttr("use_locking", opts.useLocking);
+        }
+      }
+    }
+    return new ResourceSparseApplyCenteredRmsProp(opBuilder.build());
+  }
+  
+  /**
+   * @param useLocking If `True`, updating of the var, mg, ms, and mom tensors is
+   * protected by a lock; otherwise the behavior is undefined, but may exhibit less
+   * contention.
+   */
+  public static Options useLocking(Boolean useLocking) {
+    return new Options().useLocking(useLocking);
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ResourceSparseApplyCenteredRMSProp";
+  
+  private ResourceSparseApplyCenteredRmsProp(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyFtrl.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyFtrl.java
new file mode 100644
index 00000000000..a13382272c8
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyFtrl.java
@@ -0,0 +1,141 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.train;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Update relevant entries in '*var' according to the Ftrl-proximal scheme.
+ * <p>
+ * That is for rows we have grad for, we update var, accum and linear as follows:
+ * grad_with_shrinkage = grad + 2 * l2_shrinkage * var
+ * accum_new = accum + grad_with_shrinkage * grad_with_shrinkage
+ * linear += grad_with_shrinkage +
+ *     (accum_new^(-lr_power) - accum^(-lr_power)) / lr * var
+ * quadratic = 1.0 / (accum_new^(lr_power) * lr) + 2 * l2
+ * var = (sign(linear) * l1 - linear) / quadratic if |linear| > l1 else 0.0
+ * accum = accum_new
+ */
+@Operator(group = "train")
+public final class ResourceSparseApplyFtrl extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.train.ResourceSparseApplyFtrl}
+   */
+  public static class Options {
+    
+    /**
+     * @param useLocking If `True`, updating of the var and accum tensors will be protected
+     * by a lock; otherwise the behavior is undefined, but may exhibit less
+     * contention.
+     */
+    public Options useLocking(Boolean useLocking) {
+      this.useLocking = useLocking;
+      return this;
+    }
+    
+    /**
+     * @param multiplyLinearByLr 
+     */
+    public Options multiplyLinearByLr(Boolean multiplyLinearByLr) {
+      this.multiplyLinearByLr = multiplyLinearByLr;
+      return this;
+    }
+    
+    private Boolean useLocking;
+    private Boolean multiplyLinearByLr;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new ResourceSparseApplyFtrl operation.
+   * 
+   * @param scope current scope
+   * @param var Should be from a Variable().
+   * @param accum Should be from a Variable().
+   * @param linear Should be from a Variable().
+   * @param grad The gradient.
+   * @param indices A vector of indices into the first dimension of var and accum.
+   * @param lr Scaling factor. Must be a scalar.
+   * @param l1 L1 regularization. Must be a scalar.
+   * @param l2 L2 shrinkage regularization. Must be a scalar.
+   * @param l2Shrinkage 
+   * @param lrPower Scaling factor. Must be a scalar.
+   * @param options carries optional attributes values
+   * @return a new instance of ResourceSparseApplyFtrl
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType, U extends TNumber> ResourceSparseApplyFtrl create(Scope scope, Operand<?> var, Operand<?> accum, Operand<?> linear, Operand<T> grad, Operand<U> indices, Operand<T> lr, Operand<T> l1, Operand<T> l2, Operand<T> l2Shrinkage, Operand<T> lrPower, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ResourceSparseApplyFtrlV2", scope.makeOpName("ResourceSparseApplyFtrl"));
+    opBuilder.addInput(var.asOutput());
+    opBuilder.addInput(accum.asOutput());
+    opBuilder.addInput(linear.asOutput());
+    opBuilder.addInput(grad.asOutput());
+    opBuilder.addInput(indices.asOutput());
+    opBuilder.addInput(lr.asOutput());
+    opBuilder.addInput(l1.asOutput());
+    opBuilder.addInput(l2.asOutput());
+    opBuilder.addInput(l2Shrinkage.asOutput());
+    opBuilder.addInput(lrPower.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.useLocking != null) {
+          opBuilder.setAttr("use_locking", opts.useLocking);
+        }
+        if (opts.multiplyLinearByLr != null) {
+          opBuilder.setAttr("multiply_linear_by_lr", opts.multiplyLinearByLr);
+        }
+      }
+    }
+    return new ResourceSparseApplyFtrl(opBuilder.build());
+  }
+  
+  /**
+   * @param useLocking If `True`, updating of the var and accum tensors will be protected
+   * by a lock; otherwise the behavior is undefined, but may exhibit less
+   * contention.
+   */
+  public static Options useLocking(Boolean useLocking) {
+    return new Options().useLocking(useLocking);
+  }
+  
+  /**
+   * @param multiplyLinearByLr 
+   */
+  public static Options multiplyLinearByLr(Boolean multiplyLinearByLr) {
+    return new Options().multiplyLinearByLr(multiplyLinearByLr);
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ResourceSparseApplyFtrlV2";
+  
+  private ResourceSparseApplyFtrl(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyKerasMomentum.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyKerasMomentum.java
new file mode 100644
index 00000000000..b385403f989
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyKerasMomentum.java
@@ -0,0 +1,135 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.train;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Update relevant entries in '*var' and '*accum' according to the momentum scheme.
+ * <p>
+ * Set use_nesterov = True if you want to use Nesterov momentum.
+ * <p>
+ * That is for rows we have grad for, we update var and accum as follows:
+ * <p>
+ * accum = accum * momentum - lr * grad
+ * var += accum
+ */
+@Operator(group = "train")
+public final class ResourceSparseApplyKerasMomentum extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.train.ResourceSparseApplyKerasMomentum}
+   */
+  public static class Options {
+    
+    /**
+     * @param useLocking If `True`, updating of the var and accum tensors will be protected
+     * by a lock; otherwise the behavior is undefined, but may exhibit less
+     * contention.
+     */
+    public Options useLocking(Boolean useLocking) {
+      this.useLocking = useLocking;
+      return this;
+    }
+    
+    /**
+     * @param useNesterov If `True`, the tensor passed to compute grad will be
+     * var + momentum * accum, so in the end, the var you get is actually
+     * var + momentum * accum.
+     */
+    public Options useNesterov(Boolean useNesterov) {
+      this.useNesterov = useNesterov;
+      return this;
+    }
+    
+    private Boolean useLocking;
+    private Boolean useNesterov;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new ResourceSparseApplyKerasMomentum operation.
+   * 
+   * @param scope current scope
+   * @param var Should be from a Variable().
+   * @param accum Should be from a Variable().
+   * @param lr Learning rate. Must be a scalar.
+   * @param grad The gradient.
+   * @param indices A vector of indices into the first dimension of var and accum.
+   * @param momentum Momentum. Must be a scalar.
+   * @param options carries optional attributes values
+   * @return a new instance of ResourceSparseApplyKerasMomentum
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType, U extends TNumber> ResourceSparseApplyKerasMomentum create(Scope scope, Operand<?> var, Operand<?> accum, Operand<T> lr, Operand<T> grad, Operand<U> indices, Operand<T> momentum, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ResourceSparseApplyKerasMomentum", scope.makeOpName("ResourceSparseApplyKerasMomentum"));
+    opBuilder.addInput(var.asOutput());
+    opBuilder.addInput(accum.asOutput());
+    opBuilder.addInput(lr.asOutput());
+    opBuilder.addInput(grad.asOutput());
+    opBuilder.addInput(indices.asOutput());
+    opBuilder.addInput(momentum.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.useLocking != null) {
+          opBuilder.setAttr("use_locking", opts.useLocking);
+        }
+        if (opts.useNesterov != null) {
+          opBuilder.setAttr("use_nesterov", opts.useNesterov);
+        }
+      }
+    }
+    return new ResourceSparseApplyKerasMomentum(opBuilder.build());
+  }
+  
+  /**
+   * @param useLocking If `True`, updating of the var and accum tensors will be protected
+   * by a lock; otherwise the behavior is undefined, but may exhibit less
+   * contention.
+   */
+  public static Options useLocking(Boolean useLocking) {
+    return new Options().useLocking(useLocking);
+  }
+  
+  /**
+   * @param useNesterov If `True`, the tensor passed to compute grad will be
+   * var + momentum * accum, so in the end, the var you get is actually
+   * var + momentum * accum.
+   */
+  public static Options useNesterov(Boolean useNesterov) {
+    return new Options().useNesterov(useNesterov);
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ResourceSparseApplyKerasMomentum";
+  
+  private ResourceSparseApplyKerasMomentum(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyMomentum.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyMomentum.java
new file mode 100644
index 00000000000..bc303bfbbf0
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyMomentum.java
@@ -0,0 +1,135 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.train;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Update relevant entries in '*var' and '*accum' according to the momentum scheme.
+ * <p>
+ * Set use_nesterov = True if you want to use Nesterov momentum.
+ * <p>
+ * That is for rows we have grad for, we update var and accum as follows:
+ * <p>
+ * accum = accum * momentum + grad
+ * var -= lr * accum
+ */
+@Operator(group = "train")
+public final class ResourceSparseApplyMomentum extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.train.ResourceSparseApplyMomentum}
+   */
+  public static class Options {
+    
+    /**
+     * @param useLocking If `True`, updating of the var and accum tensors will be protected
+     * by a lock; otherwise the behavior is undefined, but may exhibit less
+     * contention.
+     */
+    public Options useLocking(Boolean useLocking) {
+      this.useLocking = useLocking;
+      return this;
+    }
+    
+    /**
+     * @param useNesterov If `True`, the tensor passed to compute grad will be
+     * var - lr * momentum * accum, so in the end, the var you get is actually
+     * var - lr * momentum * accum.
+     */
+    public Options useNesterov(Boolean useNesterov) {
+      this.useNesterov = useNesterov;
+      return this;
+    }
+    
+    private Boolean useLocking;
+    private Boolean useNesterov;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new ResourceSparseApplyMomentum operation.
+   * 
+   * @param scope current scope
+   * @param var Should be from a Variable().
+   * @param accum Should be from a Variable().
+   * @param lr Learning rate. Must be a scalar.
+   * @param grad The gradient.
+   * @param indices A vector of indices into the first dimension of var and accum.
+   * @param momentum Momentum. Must be a scalar.
+   * @param options carries optional attributes values
+   * @return a new instance of ResourceSparseApplyMomentum
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType, U extends TNumber> ResourceSparseApplyMomentum create(Scope scope, Operand<?> var, Operand<?> accum, Operand<T> lr, Operand<T> grad, Operand<U> indices, Operand<T> momentum, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ResourceSparseApplyMomentum", scope.makeOpName("ResourceSparseApplyMomentum"));
+    opBuilder.addInput(var.asOutput());
+    opBuilder.addInput(accum.asOutput());
+    opBuilder.addInput(lr.asOutput());
+    opBuilder.addInput(grad.asOutput());
+    opBuilder.addInput(indices.asOutput());
+    opBuilder.addInput(momentum.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.useLocking != null) {
+          opBuilder.setAttr("use_locking", opts.useLocking);
+        }
+        if (opts.useNesterov != null) {
+          opBuilder.setAttr("use_nesterov", opts.useNesterov);
+        }
+      }
+    }
+    return new ResourceSparseApplyMomentum(opBuilder.build());
+  }
+  
+  /**
+   * @param useLocking If `True`, updating of the var and accum tensors will be protected
+   * by a lock; otherwise the behavior is undefined, but may exhibit less
+   * contention.
+   */
+  public static Options useLocking(Boolean useLocking) {
+    return new Options().useLocking(useLocking);
+  }
+  
+  /**
+   * @param useNesterov If `True`, the tensor passed to compute grad will be
+   * var - lr * momentum * accum, so in the end, the var you get is actually
+   * var - lr * momentum * accum.
+   */
+  public static Options useNesterov(Boolean useNesterov) {
+    return new Options().useNesterov(useNesterov);
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ResourceSparseApplyMomentum";
+  
+  private ResourceSparseApplyMomentum(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyProximalAdagrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyProximalAdagrad.java
new file mode 100644
index 00000000000..678601d6aea
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyProximalAdagrad.java
@@ -0,0 +1,111 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.train;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Sparse update entries in '*var' and '*accum' according to FOBOS algorithm.
+ * <p>
+ * That is for rows we have grad for, we update var and accum as follows:
+ * accum += grad <i> grad
+ * prox_v = var
+ * prox_v -= lr </i> grad <i> (1 / sqrt(accum))
+ * var = sign(prox_v)/(1+lr</i>l2) <i> max{|prox_v|-lr</i>l1,0}
+ */
+@Operator(group = "train")
+public final class ResourceSparseApplyProximalAdagrad extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.train.ResourceSparseApplyProximalAdagrad}
+   */
+  public static class Options {
+    
+    /**
+     * @param useLocking If True, updating of the var and accum tensors will be protected by
+     * a lock; otherwise the behavior is undefined, but may exhibit less contention.
+     */
+    public Options useLocking(Boolean useLocking) {
+      this.useLocking = useLocking;
+      return this;
+    }
+    
+    private Boolean useLocking;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new ResourceSparseApplyProximalAdagrad operation.
+   * 
+   * @param scope current scope
+   * @param var Should be from a Variable().
+   * @param accum Should be from a Variable().
+   * @param lr Learning rate. Must be a scalar.
+   * @param l1 L1 regularization. Must be a scalar.
+   * @param l2 L2 regularization. Must be a scalar.
+   * @param grad The gradient.
+   * @param indices A vector of indices into the first dimension of var and accum.
+   * @param options carries optional attributes values
+   * @return a new instance of ResourceSparseApplyProximalAdagrad
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType, U extends TNumber> ResourceSparseApplyProximalAdagrad create(Scope scope, Operand<?> var, Operand<?> accum, Operand<T> lr, Operand<T> l1, Operand<T> l2, Operand<T> grad, Operand<U> indices, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ResourceSparseApplyProximalAdagrad", scope.makeOpName("ResourceSparseApplyProximalAdagrad"));
+    opBuilder.addInput(var.asOutput());
+    opBuilder.addInput(accum.asOutput());
+    opBuilder.addInput(lr.asOutput());
+    opBuilder.addInput(l1.asOutput());
+    opBuilder.addInput(l2.asOutput());
+    opBuilder.addInput(grad.asOutput());
+    opBuilder.addInput(indices.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.useLocking != null) {
+          opBuilder.setAttr("use_locking", opts.useLocking);
+        }
+      }
+    }
+    return new ResourceSparseApplyProximalAdagrad(opBuilder.build());
+  }
+  
+  /**
+   * @param useLocking If True, updating of the var and accum tensors will be protected by
+   * a lock; otherwise the behavior is undefined, but may exhibit less contention.
+   */
+  public static Options useLocking(Boolean useLocking) {
+    return new Options().useLocking(useLocking);
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ResourceSparseApplyProximalAdagrad";
+  
+  private ResourceSparseApplyProximalAdagrad(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyProximalGradientDescent.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyProximalGradientDescent.java
new file mode 100644
index 00000000000..11ad213524c
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyProximalGradientDescent.java
@@ -0,0 +1,107 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.train;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Sparse update '*var' as FOBOS algorithm with fixed learning rate.
+ * <p>
+ * That is for rows we have grad for, we update var as follows:
+ * prox_v = var - alpha <i> grad
+ * var = sign(prox_v)/(1+alpha</i>l2) <i> max{|prox_v|-alpha</i>l1,0}
+ */
+@Operator(group = "train")
+public final class ResourceSparseApplyProximalGradientDescent extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.train.ResourceSparseApplyProximalGradientDescent}
+   */
+  public static class Options {
+    
+    /**
+     * @param useLocking If True, the subtraction will be protected by a lock;
+     * otherwise the behavior is undefined, but may exhibit less contention.
+     */
+    public Options useLocking(Boolean useLocking) {
+      this.useLocking = useLocking;
+      return this;
+    }
+    
+    private Boolean useLocking;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new ResourceSparseApplyProximalGradientDescent operation.
+   * 
+   * @param scope current scope
+   * @param var Should be from a Variable().
+   * @param alpha Scaling factor. Must be a scalar.
+   * @param l1 L1 regularization. Must be a scalar.
+   * @param l2 L2 regularization. Must be a scalar.
+   * @param grad The gradient.
+   * @param indices A vector of indices into the first dimension of var and accum.
+   * @param options carries optional attributes values
+   * @return a new instance of ResourceSparseApplyProximalGradientDescent
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType, U extends TNumber> ResourceSparseApplyProximalGradientDescent create(Scope scope, Operand<?> var, Operand<T> alpha, Operand<T> l1, Operand<T> l2, Operand<T> grad, Operand<U> indices, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ResourceSparseApplyProximalGradientDescent", scope.makeOpName("ResourceSparseApplyProximalGradientDescent"));
+    opBuilder.addInput(var.asOutput());
+    opBuilder.addInput(alpha.asOutput());
+    opBuilder.addInput(l1.asOutput());
+    opBuilder.addInput(l2.asOutput());
+    opBuilder.addInput(grad.asOutput());
+    opBuilder.addInput(indices.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.useLocking != null) {
+          opBuilder.setAttr("use_locking", opts.useLocking);
+        }
+      }
+    }
+    return new ResourceSparseApplyProximalGradientDescent(opBuilder.build());
+  }
+  
+  /**
+   * @param useLocking If True, the subtraction will be protected by a lock;
+   * otherwise the behavior is undefined, but may exhibit less contention.
+   */
+  public static Options useLocking(Boolean useLocking) {
+    return new Options().useLocking(useLocking);
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ResourceSparseApplyProximalGradientDescent";
+  
+  private ResourceSparseApplyProximalGradientDescent(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyRmsProp.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyRmsProp.java
new file mode 100644
index 00000000000..8c519504f89
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/ResourceSparseApplyRmsProp.java
@@ -0,0 +1,122 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.train;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Update '*var' according to the RMSProp algorithm.
+ * <p>
+ * Note that in dense implementation of this algorithm, ms and mom will
+ * update even if the grad is zero, but in this sparse implementation, ms
+ * and mom will not update in iterations during which the grad is zero.
+ * <p>
+ * mean_square = decay * mean_square + (1-decay) * gradient ** 2
+ * Delta = learning_rate * gradient / sqrt(mean_square + epsilon)
+ * <p>
+ * ms <- rho * ms_{t-1} + (1-rho) * grad * grad
+ * mom <- momentum * mom_{t-1} + lr * grad / sqrt(ms + epsilon)
+ * var <- var - mom
+ */
+@Operator(group = "train")
+public final class ResourceSparseApplyRmsProp extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.train.ResourceSparseApplyRmsProp}
+   */
+  public static class Options {
+    
+    /**
+     * @param useLocking If `True`, updating of the var, ms, and mom tensors is protected
+     * by a lock; otherwise the behavior is undefined, but may exhibit less
+     * contention.
+     */
+    public Options useLocking(Boolean useLocking) {
+      this.useLocking = useLocking;
+      return this;
+    }
+    
+    private Boolean useLocking;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new ResourceSparseApplyRmsProp operation.
+   * 
+   * @param scope current scope
+   * @param var Should be from a Variable().
+   * @param ms Should be from a Variable().
+   * @param mom Should be from a Variable().
+   * @param lr Scaling factor. Must be a scalar.
+   * @param rho Decay rate. Must be a scalar.
+   * @param momentum 
+   * @param epsilon Ridge term. Must be a scalar.
+   * @param grad The gradient.
+   * @param indices A vector of indices into the first dimension of var, ms and mom.
+   * @param options carries optional attributes values
+   * @return a new instance of ResourceSparseApplyRmsProp
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType, U extends TNumber> ResourceSparseApplyRmsProp create(Scope scope, Operand<?> var, Operand<?> ms, Operand<?> mom, Operand<T> lr, Operand<T> rho, Operand<T> momentum, Operand<T> epsilon, Operand<T> grad, Operand<U> indices, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("ResourceSparseApplyRMSProp", scope.makeOpName("ResourceSparseApplyRmsProp"));
+    opBuilder.addInput(var.asOutput());
+    opBuilder.addInput(ms.asOutput());
+    opBuilder.addInput(mom.asOutput());
+    opBuilder.addInput(lr.asOutput());
+    opBuilder.addInput(rho.asOutput());
+    opBuilder.addInput(momentum.asOutput());
+    opBuilder.addInput(epsilon.asOutput());
+    opBuilder.addInput(grad.asOutput());
+    opBuilder.addInput(indices.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.useLocking != null) {
+          opBuilder.setAttr("use_locking", opts.useLocking);
+        }
+      }
+    }
+    return new ResourceSparseApplyRmsProp(opBuilder.build());
+  }
+  
+  /**
+   * @param useLocking If `True`, updating of the var, ms, and mom tensors is protected
+   * by a lock; otherwise the behavior is undefined, but may exhibit less
+   * contention.
+   */
+  public static Options useLocking(Boolean useLocking) {
+    return new Options().useLocking(useLocking);
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "ResourceSparseApplyRMSProp";
+  
+  private ResourceSparseApplyRmsProp(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/Restore.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/Restore.java
new file mode 100644
index 00000000000..721e40e058a
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/Restore.java
@@ -0,0 +1,107 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.train;
+
+import java.util.Arrays;
+import java.util.Iterator;
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TString;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Restores tensors from a V2 checkpoint.
+ * <p>
+ * For backward compatibility with the V1 format, this Op currently allows
+ * restoring from a V1 checkpoint as well:
+ *   - This Op first attempts to find the V2 index file pointed to by "prefix", and
+ *     if found proceed to read it as a V2 checkpoint;
+ *   - Otherwise the V1 read path is invoked.
+ * Relying on this behavior is not recommended, as the ability to fall back to read
+ * V1 might be deprecated and eventually removed.
+ * <p>
+ * By default, restores the named tensors in full.  If the caller wishes to restore
+ * specific slices of stored tensors, "shape_and_slices" should be non-empty
+ * strings and correspondingly well-formed.
+ * <p>
+ * Callers must ensure all the named tensors are indeed stored in the checkpoint.
+ */
+@Operator(group = "train")
+public final class Restore extends RawOp implements Iterable<Operand<TType>> {
+  
+  /**
+   * Factory method to create a class wrapping a new Restore operation.
+   * 
+   * @param scope current scope
+   * @param prefix Must have a single element.  The prefix of a V2 checkpoint.
+   * @param tensorNames shape {N}.  The names of the tensors to be restored.
+   * @param shapeAndSlices shape {N}.  The slice specs of the tensors to be restored.
+   * Empty strings indicate that they are non-partitioned tensors.
+   * @param dtypes shape {N}.  The list of expected dtype for the tensors.  Must match
+   * those stored in the checkpoint.
+   * @return a new instance of Restore
+   */
+  @Endpoint(describeByClass = true)
+  public static Restore create(Scope scope, Operand<TString> prefix, Operand<TString> tensorNames, Operand<TString> shapeAndSlices, List<Class<? extends TType>> dtypes) {
+    OperationBuilder opBuilder = scope.env().opBuilder("RestoreV2", scope.makeOpName("Restore"));
+    opBuilder.addInput(prefix.asOutput());
+    opBuilder.addInput(tensorNames.asOutput());
+    opBuilder.addInput(shapeAndSlices.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    Class[] dtypesArray = new Class[dtypes.size()];
+    for (int i = 0; i < dtypesArray.length; ++i) {
+      dtypesArray[i] = dtypes.get(i);
+    }
+    opBuilder.setAttr("dtypes", dtypesArray);
+    return new Restore(opBuilder.build());
+  }
+  
+  /**
+   * shape {N}.  The restored tensors, whose shapes are read from the
+   * checkpoint directly.
+   */
+  public List<Output<?>> tensors() {
+    return tensors;
+  }
+  
+  @Override
+  @SuppressWarnings({"rawtypes", "unchecked"})
+  public Iterator<Operand<TType>> iterator() {
+    return (Iterator) tensors.iterator();
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "RestoreV2";
+  
+  private List<Output<?>> tensors;
+  
+  private Restore(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    int tensorsLength = operation.outputListLength("tensors");
+    tensors = Arrays.asList(operation.outputList(outputIdx, tensorsLength));
+    outputIdx += tensorsLength;
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/RestoreSlice.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/RestoreSlice.java
new file mode 100644
index 00000000000..24b13ac30af
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/RestoreSlice.java
@@ -0,0 +1,128 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.train;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TString;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Restores a tensor from checkpoint files.
+ * <p>
+ * This is like `Restore` except that restored tensor can be listed as filling
+ * only a slice of a larger tensor.  `shape_and_slice` specifies the shape of the
+ * larger tensor and the slice that the restored tensor covers.
+ * <p>
+ * The `shape_and_slice` input has the same format as the
+ * elements of the `shapes_and_slices` input of the `SaveSlices` op.
+ * 
+ * @param <T> data type for {@code tensor()} output
+ */
+@Operator(group = "train")
+public final class RestoreSlice<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.train.RestoreSlice}
+   */
+  public static class Options {
+    
+    /**
+     * @param preferredShard Index of file to open first if multiple files match
+     * `file_pattern`. See the documentation for `Restore`.
+     */
+    public Options preferredShard(Long preferredShard) {
+      this.preferredShard = preferredShard;
+      return this;
+    }
+    
+    private Long preferredShard;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new RestoreSlice operation.
+   * 
+   * @param scope current scope
+   * @param filePattern Must have a single element. The pattern of the files from
+   * which we read the tensor.
+   * @param tensorName Must have a single element. The name of the tensor to be
+   * restored.
+   * @param shapeAndSlice Scalar. The shapes and slice specifications to use when
+   * restoring a tensors.
+   * @param dt The type of the tensor to be restored.
+   * @param options carries optional attributes values
+   * @return a new instance of RestoreSlice
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> RestoreSlice<T> create(Scope scope, Operand<TString> filePattern, Operand<TString> tensorName, Operand<TString> shapeAndSlice, Class<T> dt, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("RestoreSlice", scope.makeOpName("RestoreSlice"));
+    opBuilder.addInput(filePattern.asOutput());
+    opBuilder.addInput(tensorName.asOutput());
+    opBuilder.addInput(shapeAndSlice.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("dt", dt);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.preferredShard != null) {
+          opBuilder.setAttr("preferred_shard", opts.preferredShard);
+        }
+      }
+    }
+    return new RestoreSlice<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param preferredShard Index of file to open first if multiple files match
+   * `file_pattern`. See the documentation for `Restore`.
+   */
+  public static Options preferredShard(Long preferredShard) {
+    return new Options().preferredShard(preferredShard);
+  }
+  
+  /**
+   * The restored tensor.
+   */
+  public Output<T> tensor() {
+    return tensor;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return tensor;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "RestoreSlice";
+  
+  private Output<T> tensor;
+  
+  private RestoreSlice(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    tensor = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/Save.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/Save.java
new file mode 100644
index 00000000000..c5de40fc91b
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/Save.java
@@ -0,0 +1,69 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.train;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.Operands;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TString;
+
+/**
+ * Saves tensors in V2 checkpoint format.
+ * <p>
+ * By default, saves the named tensors in full.  If the caller wishes to save
+ * specific slices of full tensors, "shape_and_slices" should be non-empty strings
+ * and correspondingly well-formed.
+ */
+@Operator(group = "train")
+public final class Save extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new Save operation.
+   * 
+   * @param scope current scope
+   * @param prefix Must have a single element. The prefix of the V2 checkpoint to which we
+   * write the tensors.
+   * @param tensorNames shape {N}. The names of the tensors to be saved.
+   * @param shapeAndSlices shape {N}.  The slice specs of the tensors to be saved.
+   * Empty strings indicate that they are non-partitioned tensors.
+   * @param tensors `N` tensors to save.
+   * @return a new instance of Save
+   */
+  @Endpoint(describeByClass = true)
+  public static Save create(Scope scope, Operand<TString> prefix, Operand<TString> tensorNames, Operand<TString> shapeAndSlices, Iterable<Operand<?>> tensors) {
+    OperationBuilder opBuilder = scope.env().opBuilder("SaveV2", scope.makeOpName("Save"));
+    opBuilder.addInput(prefix.asOutput());
+    opBuilder.addInput(tensorNames.asOutput());
+    opBuilder.addInput(shapeAndSlices.asOutput());
+    opBuilder.addInputList(Operands.asOutputs(tensors));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new Save(opBuilder.build());
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "SaveV2";
+  
+  private Save(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SaveSlices.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SaveSlices.java
new file mode 100644
index 00000000000..73325d1d1bc
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SaveSlices.java
@@ -0,0 +1,95 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.train;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.Operands;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TString;
+
+/**
+ * Saves input tensors slices to disk.
+ * <p>
+ * This is like `Save` except that tensors can be listed in the saved file as being
+ * a slice of a larger tensor.  `shapes_and_slices` specifies the shape of the
+ * larger tensor and the slice that this tensor covers. `shapes_and_slices` must
+ * have as many elements as `tensor_names`.
+ * <p>
+ * Elements of the `shapes_and_slices` input must either be:
+ * <ul>
+ * <li>
+ * The empty string, in which case the corresponding tensor is
+ *    saved normally.
+ * </li>
+ * <li>
+ * A string of the form `dim0 dim1 ... dimN-1 slice-spec` where the
+ *    `dimI` are the dimensions of the larger tensor and `slice-spec`
+ *    specifies what part is covered by the tensor to save.
+ * </li>
+ * </ul>
+ * `slice-spec` itself is a `:`-separated list: `slice0:slice1:...:sliceN-1`
+ * where each `sliceI` is either:
+ * <ul>
+ * <li>
+ * The string `-` meaning that the slice covers all indices of this dimension
+ * </li>
+ * <li>
+ * `start,length` where `start` and `length` are integers.  In that
+ *    case the slice covers `length` indices starting at `start`.
+ * </li>
+ * </ul>
+ * See also `Save`.
+ */
+@Operator(group = "train")
+public final class SaveSlices extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new SaveSlices operation.
+   * 
+   * @param scope current scope
+   * @param filename Must have a single element. The name of the file to which we write the
+   * tensor.
+   * @param tensorNames Shape `[N]`. The names of the tensors to be saved.
+   * @param shapesAndSlices Shape `[N]`.  The shapes and slice specifications to use when
+   * saving the tensors.
+   * @param data `N` tensors to save.
+   * @return a new instance of SaveSlices
+   */
+  @Endpoint(describeByClass = true)
+  public static SaveSlices create(Scope scope, Operand<TString> filename, Operand<TString> tensorNames, Operand<TString> shapesAndSlices, Iterable<Operand<?>> data) {
+    OperationBuilder opBuilder = scope.env().opBuilder("SaveSlices", scope.makeOpName("SaveSlices"));
+    opBuilder.addInput(filename.asOutput());
+    opBuilder.addInput(tensorNames.asOutput());
+    opBuilder.addInput(shapesAndSlices.asOutput());
+    opBuilder.addInputList(Operands.asOutputs(data));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new SaveSlices(opBuilder.build());
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "SaveSlices";
+  
+  private SaveSlices(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SdcaFprint.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SdcaFprint.java
new file mode 100644
index 00000000000..cdee1af05f5
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SdcaFprint.java
@@ -0,0 +1,75 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.train;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.TString;
+
+/**
+ * Computes fingerprints of the input strings.
+ */
+@Operator(group = "train")
+public final class SdcaFprint extends RawOp implements Operand<TInt64> {
+  
+  /**
+   * Factory method to create a class wrapping a new SdcaFprint operation.
+   * 
+   * @param scope current scope
+   * @param input vector of strings to compute fingerprints on.
+   * @return a new instance of SdcaFprint
+   */
+  @Endpoint(describeByClass = true)
+  public static SdcaFprint create(Scope scope, Operand<TString> input) {
+    OperationBuilder opBuilder = scope.env().opBuilder("SdcaFprint", scope.makeOpName("SdcaFprint"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new SdcaFprint(opBuilder.build());
+  }
+  
+  /**
+   * a (N,2) shaped matrix where N is the number of elements in the input
+   * vector. Each row contains the low and high parts of the fingerprint.
+   */
+  public Output<TInt64> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<TInt64> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "SdcaFprint";
+  
+  private Output<TInt64> output;
+  
+  private SdcaFprint(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SdcaOptimizer.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SdcaOptimizer.java
new file mode 100644
index 00000000000..a38f6773d7f
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SdcaOptimizer.java
@@ -0,0 +1,185 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.train;
+
+import java.util.Arrays;
+import java.util.List;
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.Operands;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+import org.tensorflow.types.TInt64;
+
+/**
+ * Distributed version of Stochastic Dual Coordinate Ascent (SDCA) optimizer for
+ * <p>
+ * linear models with L1 + L2 regularization. As global optimization objective is
+ * strongly-convex, the optimizer optimizes the dual objective at each step. The
+ * optimizer applies each update one example at a time. Examples are sampled
+ * uniformly, and the optimizer is learning rate free and enjoys linear convergence
+ * rate.
+ * <p>
+ * [Proximal Stochastic Dual Coordinate Ascent](http://arxiv.org/pdf/1211.2717v1.pdf).<br>
+ * Shai Shalev-Shwartz, Tong Zhang. 2012
+ * <p>
+ * $$Loss Objective = \sum f_{i} (wx_{i}) + (l2 / 2) * |w|^2 + l1 * |w|$$
+ * <p>
+ * [Adding vs. Averaging in Distributed Primal-Dual Optimization](http://arxiv.org/abs/1502.03508).<br>
+ * Chenxin Ma, Virginia Smith, Martin Jaggi, Michael I. Jordan,
+ * Peter Richtarik, Martin Takac. 2015
+ * <p>
+ * [Stochastic Dual Coordinate Ascent with Adaptive Probabilities](https://arxiv.org/abs/1502.08053).<br>
+ * Dominik Csiba, Zheng Qu, Peter Richtarik. 2015
+ */
+public final class SdcaOptimizer extends RawOp {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.train.SdcaOptimizer}
+   */
+  public static class Options {
+    
+    /**
+     * @param adaptive Whether to use Adaptive SDCA for the inner loop.
+     */
+    public Options adaptive(Boolean adaptive) {
+      this.adaptive = adaptive;
+      return this;
+    }
+    
+    private Boolean adaptive;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new SdcaOptimizer operation.
+   * 
+   * @param scope current scope
+   * @param sparseExampleIndices a list of vectors which contain example indices.
+   * @param sparseFeatureIndices a list of vectors which contain feature indices.
+   * @param sparseFeatureValues a list of vectors which contains feature value
+   * associated with each feature group.
+   * @param denseFeatures a list of matrices which contains the dense feature values.
+   * @param exampleWeights a vector which contains the weight associated with each
+   * example.
+   * @param exampleLabels a vector which contains the label/target associated with each
+   * example.
+   * @param sparseIndices a list of vectors where each value is the indices which has
+   * corresponding weights in sparse_weights. This field maybe omitted for the
+   * dense approach.
+   * @param sparseWeights a list of vectors where each value is the weight associated with
+   * a sparse feature group.
+   * @param denseWeights a list of vectors where the values are the weights associated
+   * with a dense feature group.
+   * @param exampleStateData a list of vectors containing the example state data.
+   * @param lossType Type of the primal loss. Currently SdcaSolver supports logistic,
+   * squared and hinge losses.
+   * @param l1 Symmetric l1 regularization strength.
+   * @param l2 Symmetric l2 regularization strength.
+   * @param numLossPartitions Number of partitions of the global loss function.
+   * @param numInnerIterations Number of iterations per mini-batch.
+   * @param options carries optional attributes values
+   * @return a new instance of SdcaOptimizer
+   */
+  @Endpoint(describeByClass = true)
+  public static SdcaOptimizer create(Scope scope, Iterable<Operand<TInt64>> sparseExampleIndices, Iterable<Operand<TInt64>> sparseFeatureIndices, Iterable<Operand<TFloat32>> sparseFeatureValues, Iterable<Operand<TFloat32>> denseFeatures, Operand<TFloat32> exampleWeights, Operand<TFloat32> exampleLabels, Iterable<Operand<TInt64>> sparseIndices, Iterable<Operand<TFloat32>> sparseWeights, Iterable<Operand<TFloat32>> denseWeights, Operand<TFloat32> exampleStateData, String lossType, Float l1, Float l2, Long numLossPartitions, Long numInnerIterations, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("SdcaOptimizerV2", scope.makeOpName("SdcaOptimizer"));
+    opBuilder.addInputList(Operands.asOutputs(sparseExampleIndices));
+    opBuilder.addInputList(Operands.asOutputs(sparseFeatureIndices));
+    opBuilder.addInputList(Operands.asOutputs(sparseFeatureValues));
+    opBuilder.addInputList(Operands.asOutputs(denseFeatures));
+    opBuilder.addInput(exampleWeights.asOutput());
+    opBuilder.addInput(exampleLabels.asOutput());
+    opBuilder.addInputList(Operands.asOutputs(sparseIndices));
+    opBuilder.addInputList(Operands.asOutputs(sparseWeights));
+    opBuilder.addInputList(Operands.asOutputs(denseWeights));
+    opBuilder.addInput(exampleStateData.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("loss_type", lossType);
+    opBuilder.setAttr("l1", l1);
+    opBuilder.setAttr("l2", l2);
+    opBuilder.setAttr("num_loss_partitions", numLossPartitions);
+    opBuilder.setAttr("num_inner_iterations", numInnerIterations);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.adaptive != null) {
+          opBuilder.setAttr("adaptive", opts.adaptive);
+        }
+      }
+    }
+    return new SdcaOptimizer(opBuilder.build());
+  }
+  
+  /**
+   * @param adaptive Whether to use Adaptive SDCA for the inner loop.
+   */
+  public static Options adaptive(Boolean adaptive) {
+    return new Options().adaptive(adaptive);
+  }
+  
+  /**
+   * a list of vectors containing the updated example state
+   * data.
+   */
+  public Output<TFloat32> outExampleStateData() {
+    return outExampleStateData;
+  }
+  
+  /**
+   * a list of vectors where each value is the delta
+   * weights associated with a sparse feature group.
+   */
+  public List<Output<TFloat32>> outDeltaSparseWeights() {
+    return outDeltaSparseWeights;
+  }
+  
+  /**
+   * a list of vectors where the values are the delta
+   * weights associated with a dense feature group.
+   */
+  public List<Output<TFloat32>> outDeltaDenseWeights() {
+    return outDeltaDenseWeights;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "SdcaOptimizerV2";
+  
+  private Output<TFloat32> outExampleStateData;
+  private List<Output<TFloat32>> outDeltaSparseWeights;
+  private List<Output<TFloat32>> outDeltaDenseWeights;
+  
+  @SuppressWarnings("unchecked")
+  private SdcaOptimizer(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    outExampleStateData = operation.output(outputIdx++);
+    int outDeltaSparseWeightsLength = operation.outputListLength("out_delta_sparse_weights");
+    outDeltaSparseWeights = Arrays.asList((Output<TFloat32>[])operation.outputList(outputIdx, outDeltaSparseWeightsLength));
+    outputIdx += outDeltaSparseWeightsLength;
+    int outDeltaDenseWeightsLength = operation.outputListLength("out_delta_dense_weights");
+    outDeltaDenseWeights = Arrays.asList((Output<TFloat32>[])operation.outputList(outputIdx, outDeltaDenseWeightsLength));
+    outputIdx += outDeltaDenseWeightsLength;
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SdcaShrinkL1.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SdcaShrinkL1.java
new file mode 100644
index 00000000000..748a2eacaec
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SdcaShrinkL1.java
@@ -0,0 +1,62 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.train;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.Operands;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TFloat32;
+
+/**
+ * Applies L1 regularization shrink step on the parameters.
+ */
+@Operator(group = "train")
+public final class SdcaShrinkL1 extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new SdcaShrinkL1 operation.
+   * 
+   * @param scope current scope
+   * @param weights a list of vectors where each value is the weight associated with a
+   * feature group.
+   * @param l1 Symmetric l1 regularization strength.
+   * @param l2 Symmetric l2 regularization strength. Should be a positive float.
+   * @return a new instance of SdcaShrinkL1
+   */
+  @Endpoint(describeByClass = true)
+  public static SdcaShrinkL1 create(Scope scope, Iterable<Operand<TFloat32>> weights, Float l1, Float l2) {
+    OperationBuilder opBuilder = scope.env().opBuilder("SdcaShrinkL1", scope.makeOpName("SdcaShrinkL1"));
+    opBuilder.addInputList(Operands.asOutputs(weights));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("l1", l1);
+    opBuilder.setAttr("l2", l2);
+    return new SdcaShrinkL1(opBuilder.build());
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "SdcaShrinkL1";
+  
+  private SdcaShrinkL1(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyAdadelta.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyAdadelta.java
new file mode 100644
index 00000000000..190449c8ca7
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyAdadelta.java
@@ -0,0 +1,126 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.train;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * var: Should be from a Variable().
+ * 
+ * @param <T> data type for {@code out()} output
+ */
+@Operator(group = "train")
+public final class SparseApplyAdadelta<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.train.SparseApplyAdadelta}
+   */
+  public static class Options {
+    
+    /**
+     * @param useLocking If True, updating of the var and accum tensors will be protected by
+     * a lock; otherwise the behavior is undefined, but may exhibit less contention.
+     */
+    public Options useLocking(Boolean useLocking) {
+      this.useLocking = useLocking;
+      return this;
+    }
+    
+    private Boolean useLocking;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new SparseApplyAdadelta operation.
+   * 
+   * @param scope current scope
+   * @param var 
+   * @param accum Should be from a Variable().
+   * @param accumUpdate : Should be from a Variable().
+   * @param lr Learning rate. Must be a scalar.
+   * @param rho Decay factor. Must be a scalar.
+   * @param epsilon Constant factor. Must be a scalar.
+   * @param grad The gradient.
+   * @param indices A vector of indices into the first dimension of var and accum.
+   * @param options carries optional attributes values
+   * @return a new instance of SparseApplyAdadelta
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType, U extends TNumber> SparseApplyAdadelta<T> create(Scope scope, Operand<T> var, Operand<T> accum, Operand<T> accumUpdate, Operand<T> lr, Operand<T> rho, Operand<T> epsilon, Operand<T> grad, Operand<U> indices, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("SparseApplyAdadelta", scope.makeOpName("SparseApplyAdadelta"));
+    opBuilder.addInput(var.asOutput());
+    opBuilder.addInput(accum.asOutput());
+    opBuilder.addInput(accumUpdate.asOutput());
+    opBuilder.addInput(lr.asOutput());
+    opBuilder.addInput(rho.asOutput());
+    opBuilder.addInput(epsilon.asOutput());
+    opBuilder.addInput(grad.asOutput());
+    opBuilder.addInput(indices.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.useLocking != null) {
+          opBuilder.setAttr("use_locking", opts.useLocking);
+        }
+      }
+    }
+    return new SparseApplyAdadelta<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param useLocking If True, updating of the var and accum tensors will be protected by
+   * a lock; otherwise the behavior is undefined, but may exhibit less contention.
+   */
+  public static Options useLocking(Boolean useLocking) {
+    return new Options().useLocking(useLocking);
+  }
+  
+  /**
+   * Same as "var".
+   */
+  public Output<T> out() {
+    return out;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return out;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "SparseApplyAdadelta";
+  
+  private Output<T> out;
+  
+  private SparseApplyAdadelta(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    out = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyAdagrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyAdagrad.java
new file mode 100644
index 00000000000..71776f4fd26
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyAdagrad.java
@@ -0,0 +1,146 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.train;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Update relevant entries in '*var' and '*accum' according to the adagrad scheme.
+ * <p>
+ * That is for rows we have grad for, we update var and accum as follows:
+ * $$accum += grad * grad$$
+ * $$var -= lr * grad * (1 / sqrt(accum))$$
+ * 
+ * @param <T> data type for {@code out()} output
+ */
+public final class SparseApplyAdagrad<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.train.SparseApplyAdagrad}
+   */
+  public static class Options {
+    
+    /**
+     * @param useLocking If `True`, updating of the var and accum tensors will be protected
+     * by a lock; otherwise the behavior is undefined, but may exhibit less
+     * contention.
+     */
+    public Options useLocking(Boolean useLocking) {
+      this.useLocking = useLocking;
+      return this;
+    }
+    
+    /**
+     * @param updateSlots 
+     */
+    public Options updateSlots(Boolean updateSlots) {
+      this.updateSlots = updateSlots;
+      return this;
+    }
+    
+    private Boolean useLocking;
+    private Boolean updateSlots;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new SparseApplyAdagrad operation.
+   * 
+   * @param scope current scope
+   * @param var Should be from a Variable().
+   * @param accum Should be from a Variable().
+   * @param lr Learning rate. Must be a scalar.
+   * @param epsilon Constant factor. Must be a scalar.
+   * @param grad The gradient.
+   * @param indices A vector of indices into the first dimension of var and accum.
+   * @param options carries optional attributes values
+   * @return a new instance of SparseApplyAdagrad
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType, U extends TNumber> SparseApplyAdagrad<T> create(Scope scope, Operand<T> var, Operand<T> accum, Operand<T> lr, Operand<T> epsilon, Operand<T> grad, Operand<U> indices, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("SparseApplyAdagradV2", scope.makeOpName("SparseApplyAdagrad"));
+    opBuilder.addInput(var.asOutput());
+    opBuilder.addInput(accum.asOutput());
+    opBuilder.addInput(lr.asOutput());
+    opBuilder.addInput(epsilon.asOutput());
+    opBuilder.addInput(grad.asOutput());
+    opBuilder.addInput(indices.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.useLocking != null) {
+          opBuilder.setAttr("use_locking", opts.useLocking);
+        }
+        if (opts.updateSlots != null) {
+          opBuilder.setAttr("update_slots", opts.updateSlots);
+        }
+      }
+    }
+    return new SparseApplyAdagrad<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param useLocking If `True`, updating of the var and accum tensors will be protected
+   * by a lock; otherwise the behavior is undefined, but may exhibit less
+   * contention.
+   */
+  public static Options useLocking(Boolean useLocking) {
+    return new Options().useLocking(useLocking);
+  }
+  
+  /**
+   * @param updateSlots 
+   */
+  public static Options updateSlots(Boolean updateSlots) {
+    return new Options().updateSlots(updateSlots);
+  }
+  
+  /**
+   * Same as "var".
+   */
+  public Output<T> out() {
+    return out;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return out;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "SparseApplyAdagradV2";
+  
+  private Output<T> out;
+  
+  private SparseApplyAdagrad(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    out = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyAdagradDa.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyAdagradDa.java
new file mode 100644
index 00000000000..7f13dd080dd
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyAdagradDa.java
@@ -0,0 +1,129 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.train;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt64;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Update entries in '*var' and '*accum' according to the proximal adagrad scheme.
+ * 
+ * @param <T> data type for {@code out()} output
+ */
+@Operator(group = "train")
+public final class SparseApplyAdagradDa<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.train.SparseApplyAdagradDa}
+   */
+  public static class Options {
+    
+    /**
+     * @param useLocking If True, updating of the var and accum tensors will be protected by
+     * a lock; otherwise the behavior is undefined, but may exhibit less contention.
+     */
+    public Options useLocking(Boolean useLocking) {
+      this.useLocking = useLocking;
+      return this;
+    }
+    
+    private Boolean useLocking;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new SparseApplyAdagradDa operation.
+   * 
+   * @param scope current scope
+   * @param var Should be from a Variable().
+   * @param gradientAccumulator Should be from a Variable().
+   * @param gradientSquaredAccumulator Should be from a Variable().
+   * @param grad The gradient.
+   * @param indices A vector of indices into the first dimension of var and accum.
+   * @param lr Learning rate. Must be a scalar.
+   * @param l1 L1 regularization. Must be a scalar.
+   * @param l2 L2 regularization. Must be a scalar.
+   * @param globalStep Training step number. Must be a scalar.
+   * @param options carries optional attributes values
+   * @return a new instance of SparseApplyAdagradDa
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType, U extends TNumber> SparseApplyAdagradDa<T> create(Scope scope, Operand<T> var, Operand<T> gradientAccumulator, Operand<T> gradientSquaredAccumulator, Operand<T> grad, Operand<U> indices, Operand<T> lr, Operand<T> l1, Operand<T> l2, Operand<TInt64> globalStep, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("SparseApplyAdagradDA", scope.makeOpName("SparseApplyAdagradDa"));
+    opBuilder.addInput(var.asOutput());
+    opBuilder.addInput(gradientAccumulator.asOutput());
+    opBuilder.addInput(gradientSquaredAccumulator.asOutput());
+    opBuilder.addInput(grad.asOutput());
+    opBuilder.addInput(indices.asOutput());
+    opBuilder.addInput(lr.asOutput());
+    opBuilder.addInput(l1.asOutput());
+    opBuilder.addInput(l2.asOutput());
+    opBuilder.addInput(globalStep.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.useLocking != null) {
+          opBuilder.setAttr("use_locking", opts.useLocking);
+        }
+      }
+    }
+    return new SparseApplyAdagradDa<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param useLocking If True, updating of the var and accum tensors will be protected by
+   * a lock; otherwise the behavior is undefined, but may exhibit less contention.
+   */
+  public static Options useLocking(Boolean useLocking) {
+    return new Options().useLocking(useLocking);
+  }
+  
+  /**
+   * Same as "var".
+   */
+  public Output<T> out() {
+    return out;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return out;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "SparseApplyAdagradDA";
+  
+  private Output<T> out;
+  
+  private SparseApplyAdagradDa(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    out = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyCenteredRmsProp.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyCenteredRmsProp.java
new file mode 100644
index 00000000000..b73275139d1
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyCenteredRmsProp.java
@@ -0,0 +1,149 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.train;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Update '*var' according to the centered RMSProp algorithm.
+ * <p>
+ * The centered RMSProp algorithm uses an estimate of the centered second moment
+ * (i.e., the variance) for normalization, as opposed to regular RMSProp, which
+ * uses the (uncentered) second moment. This often helps with training, but is
+ * slightly more expensive in terms of computation and memory.
+ * <p>
+ * Note that in dense implementation of this algorithm, mg, ms, and mom will
+ * update even if the grad is zero, but in this sparse implementation, mg, ms,
+ * and mom will not update in iterations during which the grad is zero.
+ * <p>
+ * mean_square = decay * mean_square + (1-decay) * gradient ** 2
+ * mean_grad = decay * mean_grad + (1-decay) * gradient
+ * Delta = learning_rate * gradient / sqrt(mean_square + epsilon - mean_grad ** 2)
+ * <p>
+ * $$ms <- rho * ms_{t-1} + (1-rho) * grad * grad$$
+ * $$mom <- momentum * mom_{t-1} + lr * grad / sqrt(ms + epsilon)$$
+ * $$var <- var - mom$$
+ * 
+ * @param <T> data type for {@code out()} output
+ */
+@Operator(group = "train")
+public final class SparseApplyCenteredRmsProp<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.train.SparseApplyCenteredRmsProp}
+   */
+  public static class Options {
+    
+    /**
+     * @param useLocking If `True`, updating of the var, mg, ms, and mom tensors is
+     * protected by a lock; otherwise the behavior is undefined, but may exhibit less
+     * contention.
+     */
+    public Options useLocking(Boolean useLocking) {
+      this.useLocking = useLocking;
+      return this;
+    }
+    
+    private Boolean useLocking;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new SparseApplyCenteredRmsProp operation.
+   * 
+   * @param scope current scope
+   * @param var Should be from a Variable().
+   * @param mg Should be from a Variable().
+   * @param ms Should be from a Variable().
+   * @param mom Should be from a Variable().
+   * @param lr Scaling factor. Must be a scalar.
+   * @param rho Decay rate. Must be a scalar.
+   * @param momentum 
+   * @param epsilon Ridge term. Must be a scalar.
+   * @param grad The gradient.
+   * @param indices A vector of indices into the first dimension of var, ms and mom.
+   * @param options carries optional attributes values
+   * @return a new instance of SparseApplyCenteredRmsProp
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType, U extends TNumber> SparseApplyCenteredRmsProp<T> create(Scope scope, Operand<T> var, Operand<T> mg, Operand<T> ms, Operand<T> mom, Operand<T> lr, Operand<T> rho, Operand<T> momentum, Operand<T> epsilon, Operand<T> grad, Operand<U> indices, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("SparseApplyCenteredRMSProp", scope.makeOpName("SparseApplyCenteredRmsProp"));
+    opBuilder.addInput(var.asOutput());
+    opBuilder.addInput(mg.asOutput());
+    opBuilder.addInput(ms.asOutput());
+    opBuilder.addInput(mom.asOutput());
+    opBuilder.addInput(lr.asOutput());
+    opBuilder.addInput(rho.asOutput());
+    opBuilder.addInput(momentum.asOutput());
+    opBuilder.addInput(epsilon.asOutput());
+    opBuilder.addInput(grad.asOutput());
+    opBuilder.addInput(indices.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.useLocking != null) {
+          opBuilder.setAttr("use_locking", opts.useLocking);
+        }
+      }
+    }
+    return new SparseApplyCenteredRmsProp<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param useLocking If `True`, updating of the var, mg, ms, and mom tensors is
+   * protected by a lock; otherwise the behavior is undefined, but may exhibit less
+   * contention.
+   */
+  public static Options useLocking(Boolean useLocking) {
+    return new Options().useLocking(useLocking);
+  }
+  
+  /**
+   * Same as "var".
+   */
+  public Output<T> out() {
+    return out;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return out;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "SparseApplyCenteredRMSProp";
+  
+  private Output<T> out;
+  
+  private SparseApplyCenteredRmsProp(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    out = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyFtrl.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyFtrl.java
new file mode 100644
index 00000000000..2eb27c33f45
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyFtrl.java
@@ -0,0 +1,160 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.train;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Update relevant entries in '*var' according to the Ftrl-proximal scheme.
+ * <p>
+ * That is for rows we have grad for, we update var, accum and linear as follows:
+ * grad_with_shrinkage = grad + 2 * l2_shrinkage * var
+ * accum_new = accum + grad * grad
+ * linear += grad_with_shrinkage -
+ *     (accum_new^(-lr_power) - accum^(-lr_power)) / lr * var
+ * quadratic = 1.0 / (accum_new^(lr_power) * lr) + 2 * l2
+ * var = (sign(linear) * l1 - linear) / quadratic if |linear| > l1 else 0.0
+ * accum = accum_new
+ * 
+ * @param <T> data type for {@code out()} output
+ */
+@Operator(group = "train")
+public final class SparseApplyFtrl<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.train.SparseApplyFtrl}
+   */
+  public static class Options {
+    
+    /**
+     * @param useLocking If `True`, updating of the var and accum tensors will be protected
+     * by a lock; otherwise the behavior is undefined, but may exhibit less
+     * contention.
+     */
+    public Options useLocking(Boolean useLocking) {
+      this.useLocking = useLocking;
+      return this;
+    }
+    
+    /**
+     * @param multiplyLinearByLr 
+     */
+    public Options multiplyLinearByLr(Boolean multiplyLinearByLr) {
+      this.multiplyLinearByLr = multiplyLinearByLr;
+      return this;
+    }
+    
+    private Boolean useLocking;
+    private Boolean multiplyLinearByLr;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new SparseApplyFtrl operation.
+   * 
+   * @param scope current scope
+   * @param var Should be from a Variable().
+   * @param accum Should be from a Variable().
+   * @param linear Should be from a Variable().
+   * @param grad The gradient.
+   * @param indices A vector of indices into the first dimension of var and accum.
+   * @param lr Scaling factor. Must be a scalar.
+   * @param l1 L1 regularization. Must be a scalar.
+   * @param l2 L2 shrinkage regularization. Must be a scalar.
+   * @param l2Shrinkage 
+   * @param lrPower Scaling factor. Must be a scalar.
+   * @param options carries optional attributes values
+   * @return a new instance of SparseApplyFtrl
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType, U extends TNumber> SparseApplyFtrl<T> create(Scope scope, Operand<T> var, Operand<T> accum, Operand<T> linear, Operand<T> grad, Operand<U> indices, Operand<T> lr, Operand<T> l1, Operand<T> l2, Operand<T> l2Shrinkage, Operand<T> lrPower, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("SparseApplyFtrlV2", scope.makeOpName("SparseApplyFtrl"));
+    opBuilder.addInput(var.asOutput());
+    opBuilder.addInput(accum.asOutput());
+    opBuilder.addInput(linear.asOutput());
+    opBuilder.addInput(grad.asOutput());
+    opBuilder.addInput(indices.asOutput());
+    opBuilder.addInput(lr.asOutput());
+    opBuilder.addInput(l1.asOutput());
+    opBuilder.addInput(l2.asOutput());
+    opBuilder.addInput(l2Shrinkage.asOutput());
+    opBuilder.addInput(lrPower.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.useLocking != null) {
+          opBuilder.setAttr("use_locking", opts.useLocking);
+        }
+        if (opts.multiplyLinearByLr != null) {
+          opBuilder.setAttr("multiply_linear_by_lr", opts.multiplyLinearByLr);
+        }
+      }
+    }
+    return new SparseApplyFtrl<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param useLocking If `True`, updating of the var and accum tensors will be protected
+   * by a lock; otherwise the behavior is undefined, but may exhibit less
+   * contention.
+   */
+  public static Options useLocking(Boolean useLocking) {
+    return new Options().useLocking(useLocking);
+  }
+  
+  /**
+   * @param multiplyLinearByLr 
+   */
+  public static Options multiplyLinearByLr(Boolean multiplyLinearByLr) {
+    return new Options().multiplyLinearByLr(multiplyLinearByLr);
+  }
+  
+  /**
+   * Same as "var".
+   */
+  public Output<T> out() {
+    return out;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return out;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "SparseApplyFtrlV2";
+  
+  private Output<T> out;
+  
+  private SparseApplyFtrl(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    out = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyMomentum.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyMomentum.java
new file mode 100644
index 00000000000..5e8bf197092
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyMomentum.java
@@ -0,0 +1,154 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.train;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Update relevant entries in '*var' and '*accum' according to the momentum scheme.
+ * <p>
+ * Set use_nesterov = True if you want to use Nesterov momentum.
+ * <p>
+ * That is for rows we have grad for, we update var and accum as follows:
+ * <p>
+ * $$accum = accum * momentum + grad$$
+ * $$var -= lr * accum$$
+ * 
+ * @param <T> data type for {@code out()} output
+ */
+@Operator(group = "train")
+public final class SparseApplyMomentum<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.train.SparseApplyMomentum}
+   */
+  public static class Options {
+    
+    /**
+     * @param useLocking If `True`, updating of the var and accum tensors will be protected
+     * by a lock; otherwise the behavior is undefined, but may exhibit less
+     * contention.
+     */
+    public Options useLocking(Boolean useLocking) {
+      this.useLocking = useLocking;
+      return this;
+    }
+    
+    /**
+     * @param useNesterov If `True`, the tensor passed to compute grad will be
+     * var - lr * momentum * accum, so in the end, the var you get is actually
+     * var - lr * momentum * accum.
+     */
+    public Options useNesterov(Boolean useNesterov) {
+      this.useNesterov = useNesterov;
+      return this;
+    }
+    
+    private Boolean useLocking;
+    private Boolean useNesterov;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new SparseApplyMomentum operation.
+   * 
+   * @param scope current scope
+   * @param var Should be from a Variable().
+   * @param accum Should be from a Variable().
+   * @param lr Learning rate. Must be a scalar.
+   * @param grad The gradient.
+   * @param indices A vector of indices into the first dimension of var and accum.
+   * @param momentum Momentum. Must be a scalar.
+   * @param options carries optional attributes values
+   * @return a new instance of SparseApplyMomentum
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType, U extends TNumber> SparseApplyMomentum<T> create(Scope scope, Operand<T> var, Operand<T> accum, Operand<T> lr, Operand<T> grad, Operand<U> indices, Operand<T> momentum, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("SparseApplyMomentum", scope.makeOpName("SparseApplyMomentum"));
+    opBuilder.addInput(var.asOutput());
+    opBuilder.addInput(accum.asOutput());
+    opBuilder.addInput(lr.asOutput());
+    opBuilder.addInput(grad.asOutput());
+    opBuilder.addInput(indices.asOutput());
+    opBuilder.addInput(momentum.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.useLocking != null) {
+          opBuilder.setAttr("use_locking", opts.useLocking);
+        }
+        if (opts.useNesterov != null) {
+          opBuilder.setAttr("use_nesterov", opts.useNesterov);
+        }
+      }
+    }
+    return new SparseApplyMomentum<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param useLocking If `True`, updating of the var and accum tensors will be protected
+   * by a lock; otherwise the behavior is undefined, but may exhibit less
+   * contention.
+   */
+  public static Options useLocking(Boolean useLocking) {
+    return new Options().useLocking(useLocking);
+  }
+  
+  /**
+   * @param useNesterov If `True`, the tensor passed to compute grad will be
+   * var - lr * momentum * accum, so in the end, the var you get is actually
+   * var - lr * momentum * accum.
+   */
+  public static Options useNesterov(Boolean useNesterov) {
+    return new Options().useNesterov(useNesterov);
+  }
+  
+  /**
+   * Same as "var".
+   */
+  public Output<T> out() {
+    return out;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return out;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "SparseApplyMomentum";
+  
+  private Output<T> out;
+  
+  private SparseApplyMomentum(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    out = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyProximalAdagrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyProximalAdagrad.java
new file mode 100644
index 00000000000..19ede0d5f71
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyProximalAdagrad.java
@@ -0,0 +1,130 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.train;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Sparse update entries in '*var' and '*accum' according to FOBOS algorithm.
+ * <p>
+ * That is for rows we have grad for, we update var and accum as follows:
+ * $$accum += grad <i> grad$$
+ * $$prox_v = var$$
+ * $$prox_v -= lr </i> grad <i> (1 / sqrt(accum))$$
+ * $$var = sign(prox_v)/(1+lr</i>l2) <i> max{|prox_v|-lr</i>l1,0}$$
+ * 
+ * @param <T> data type for {@code out()} output
+ */
+@Operator(group = "train")
+public final class SparseApplyProximalAdagrad<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.train.SparseApplyProximalAdagrad}
+   */
+  public static class Options {
+    
+    /**
+     * @param useLocking If True, updating of the var and accum tensors will be protected by
+     * a lock; otherwise the behavior is undefined, but may exhibit less contention.
+     */
+    public Options useLocking(Boolean useLocking) {
+      this.useLocking = useLocking;
+      return this;
+    }
+    
+    private Boolean useLocking;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new SparseApplyProximalAdagrad operation.
+   * 
+   * @param scope current scope
+   * @param var Should be from a Variable().
+   * @param accum Should be from a Variable().
+   * @param lr Learning rate. Must be a scalar.
+   * @param l1 L1 regularization. Must be a scalar.
+   * @param l2 L2 regularization. Must be a scalar.
+   * @param grad The gradient.
+   * @param indices A vector of indices into the first dimension of var and accum.
+   * @param options carries optional attributes values
+   * @return a new instance of SparseApplyProximalAdagrad
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType, U extends TNumber> SparseApplyProximalAdagrad<T> create(Scope scope, Operand<T> var, Operand<T> accum, Operand<T> lr, Operand<T> l1, Operand<T> l2, Operand<T> grad, Operand<U> indices, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("SparseApplyProximalAdagrad", scope.makeOpName("SparseApplyProximalAdagrad"));
+    opBuilder.addInput(var.asOutput());
+    opBuilder.addInput(accum.asOutput());
+    opBuilder.addInput(lr.asOutput());
+    opBuilder.addInput(l1.asOutput());
+    opBuilder.addInput(l2.asOutput());
+    opBuilder.addInput(grad.asOutput());
+    opBuilder.addInput(indices.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.useLocking != null) {
+          opBuilder.setAttr("use_locking", opts.useLocking);
+        }
+      }
+    }
+    return new SparseApplyProximalAdagrad<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param useLocking If True, updating of the var and accum tensors will be protected by
+   * a lock; otherwise the behavior is undefined, but may exhibit less contention.
+   */
+  public static Options useLocking(Boolean useLocking) {
+    return new Options().useLocking(useLocking);
+  }
+  
+  /**
+   * Same as "var".
+   */
+  public Output<T> out() {
+    return out;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return out;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "SparseApplyProximalAdagrad";
+  
+  private Output<T> out;
+  
+  private SparseApplyProximalAdagrad(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    out = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyProximalGradientDescent.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyProximalGradientDescent.java
new file mode 100644
index 00000000000..14735403f02
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyProximalGradientDescent.java
@@ -0,0 +1,126 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.train;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Sparse update '*var' as FOBOS algorithm with fixed learning rate.
+ * <p>
+ * That is for rows we have grad for, we update var as follows:
+ * $$prox_v = var - alpha <i> grad$$
+ * $$var = sign(prox_v)/(1+alpha</i>l2) <i> max{|prox_v|-alpha</i>l1,0}$$
+ * 
+ * @param <T> data type for {@code out()} output
+ */
+@Operator(group = "train")
+public final class SparseApplyProximalGradientDescent<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.train.SparseApplyProximalGradientDescent}
+   */
+  public static class Options {
+    
+    /**
+     * @param useLocking If True, the subtraction will be protected by a lock;
+     * otherwise the behavior is undefined, but may exhibit less contention.
+     */
+    public Options useLocking(Boolean useLocking) {
+      this.useLocking = useLocking;
+      return this;
+    }
+    
+    private Boolean useLocking;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new SparseApplyProximalGradientDescent operation.
+   * 
+   * @param scope current scope
+   * @param var Should be from a Variable().
+   * @param alpha Scaling factor. Must be a scalar.
+   * @param l1 L1 regularization. Must be a scalar.
+   * @param l2 L2 regularization. Must be a scalar.
+   * @param grad The gradient.
+   * @param indices A vector of indices into the first dimension of var and accum.
+   * @param options carries optional attributes values
+   * @return a new instance of SparseApplyProximalGradientDescent
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType, U extends TNumber> SparseApplyProximalGradientDescent<T> create(Scope scope, Operand<T> var, Operand<T> alpha, Operand<T> l1, Operand<T> l2, Operand<T> grad, Operand<U> indices, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("SparseApplyProximalGradientDescent", scope.makeOpName("SparseApplyProximalGradientDescent"));
+    opBuilder.addInput(var.asOutput());
+    opBuilder.addInput(alpha.asOutput());
+    opBuilder.addInput(l1.asOutput());
+    opBuilder.addInput(l2.asOutput());
+    opBuilder.addInput(grad.asOutput());
+    opBuilder.addInput(indices.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.useLocking != null) {
+          opBuilder.setAttr("use_locking", opts.useLocking);
+        }
+      }
+    }
+    return new SparseApplyProximalGradientDescent<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param useLocking If True, the subtraction will be protected by a lock;
+   * otherwise the behavior is undefined, but may exhibit less contention.
+   */
+  public static Options useLocking(Boolean useLocking) {
+    return new Options().useLocking(useLocking);
+  }
+  
+  /**
+   * Same as "var".
+   */
+  public Output<T> out() {
+    return out;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return out;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "SparseApplyProximalGradientDescent";
+  
+  private Output<T> out;
+  
+  private SparseApplyProximalGradientDescent(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    out = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyRmsProp.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyRmsProp.java
new file mode 100644
index 00000000000..c0b39b65e5e
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/SparseApplyRmsProp.java
@@ -0,0 +1,141 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.train;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Update '*var' according to the RMSProp algorithm.
+ * <p>
+ * Note that in dense implementation of this algorithm, ms and mom will
+ * update even if the grad is zero, but in this sparse implementation, ms
+ * and mom will not update in iterations during which the grad is zero.
+ * <p>
+ * mean_square = decay * mean_square + (1-decay) * gradient ** 2
+ * Delta = learning_rate * gradient / sqrt(mean_square + epsilon)
+ * <p>
+ * $$ms <- rho * ms_{t-1} + (1-rho) * grad * grad$$
+ * $$mom <- momentum * mom_{t-1} + lr * grad / sqrt(ms + epsilon)$$
+ * $$var <- var - mom$$
+ * 
+ * @param <T> data type for {@code out()} output
+ */
+@Operator(group = "train")
+public final class SparseApplyRmsProp<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Optional attributes for {@link org.tensorflow.op.train.SparseApplyRmsProp}
+   */
+  public static class Options {
+    
+    /**
+     * @param useLocking If `True`, updating of the var, ms, and mom tensors is protected
+     * by a lock; otherwise the behavior is undefined, but may exhibit less
+     * contention.
+     */
+    public Options useLocking(Boolean useLocking) {
+      this.useLocking = useLocking;
+      return this;
+    }
+    
+    private Boolean useLocking;
+    
+    private Options() {
+    }
+  }
+  
+  /**
+   * Factory method to create a class wrapping a new SparseApplyRmsProp operation.
+   * 
+   * @param scope current scope
+   * @param var Should be from a Variable().
+   * @param ms Should be from a Variable().
+   * @param mom Should be from a Variable().
+   * @param lr Scaling factor. Must be a scalar.
+   * @param rho Decay rate. Must be a scalar.
+   * @param momentum 
+   * @param epsilon Ridge term. Must be a scalar.
+   * @param grad The gradient.
+   * @param indices A vector of indices into the first dimension of var, ms and mom.
+   * @param options carries optional attributes values
+   * @return a new instance of SparseApplyRmsProp
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType, U extends TNumber> SparseApplyRmsProp<T> create(Scope scope, Operand<T> var, Operand<T> ms, Operand<T> mom, Operand<T> lr, Operand<T> rho, Operand<T> momentum, Operand<T> epsilon, Operand<T> grad, Operand<U> indices, Options... options) {
+    OperationBuilder opBuilder = scope.env().opBuilder("SparseApplyRMSProp", scope.makeOpName("SparseApplyRmsProp"));
+    opBuilder.addInput(var.asOutput());
+    opBuilder.addInput(ms.asOutput());
+    opBuilder.addInput(mom.asOutput());
+    opBuilder.addInput(lr.asOutput());
+    opBuilder.addInput(rho.asOutput());
+    opBuilder.addInput(momentum.asOutput());
+    opBuilder.addInput(epsilon.asOutput());
+    opBuilder.addInput(grad.asOutput());
+    opBuilder.addInput(indices.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    if (options != null) {
+      for (Options opts : options) {
+        if (opts.useLocking != null) {
+          opBuilder.setAttr("use_locking", opts.useLocking);
+        }
+      }
+    }
+    return new SparseApplyRmsProp<T>(opBuilder.build());
+  }
+  
+  /**
+   * @param useLocking If `True`, updating of the var, ms, and mom tensors is protected
+   * by a lock; otherwise the behavior is undefined, but may exhibit less
+   * contention.
+   */
+  public static Options useLocking(Boolean useLocking) {
+    return new Options().useLocking(useLocking);
+  }
+  
+  /**
+   * Same as "var".
+   */
+  public Output<T> out() {
+    return out;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return out;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "SparseApplyRMSProp";
+  
+  private Output<T> out;
+  
+  private SparseApplyRmsProp(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    out = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/TileGrad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/TileGrad.java
new file mode 100644
index 00000000000..1bcfde7b183
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/train/TileGrad.java
@@ -0,0 +1,81 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.train;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Returns the gradient of `Tile`.
+ * <p>
+ * Since `Tile` takes an input and repeats the input `multiples` times
+ * along each dimension, `train.TileGrad` takes in `multiples` and aggregates
+ * each repeated tile of `input` into `output`.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator(group = "train")
+public final class TileGrad<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new TileGrad operation.
+   * 
+   * @param scope current scope
+   * @param input 
+   * @param multiples 
+   * @return a new instance of TileGrad
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> TileGrad<T> create(Scope scope, Operand<T> input, Operand<TInt32> multiples) {
+    OperationBuilder opBuilder = scope.env().opBuilder("TileGrad", scope.makeOpName("TileGrad"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(multiples.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new TileGrad<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "TileGrad";
+  
+  private Output<T> output;
+  
+  private TileGrad(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/BroadcastHelper.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/BroadcastHelper.java
new file mode 100644
index 00000000000..e06919d04bd
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/BroadcastHelper.java
@@ -0,0 +1,88 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.xla;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Helper operator for performing XLA-style broadcasts
+ * <p>
+ * Broadcasts `lhs` and `rhs` to the same rank, by adding size 1 dimensions to
+ * whichever of `lhs` and `rhs` has the lower rank, using XLA's broadcasting rules
+ * for binary operators.
+ * 
+ * @param <T> data type for {@code lhsOutput()} output
+ */
+@Operator(group = "xla")
+public final class BroadcastHelper<T extends TType> extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new BroadcastHelper operation.
+   * 
+   * @param scope current scope
+   * @param lhs the LHS input tensor
+   * @param rhs the RHS input tensor
+   * @param broadcastDims an XLA-style broadcast dimension specification
+   * @return a new instance of BroadcastHelper
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType, U extends TNumber> BroadcastHelper<T> create(Scope scope, Operand<T> lhs, Operand<T> rhs, Operand<U> broadcastDims) {
+    OperationBuilder opBuilder = scope.env().opBuilder("XlaBroadcastHelper", scope.makeOpName("BroadcastHelper"));
+    opBuilder.addInput(lhs.asOutput());
+    opBuilder.addInput(rhs.asOutput());
+    opBuilder.addInput(broadcastDims.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new BroadcastHelper<T>(opBuilder.build());
+  }
+  
+  /**
+   * the broadcasted LHS tensor
+   */
+  public Output<T> lhsOutput() {
+    return lhsOutput;
+  }
+  
+  /**
+   * the broadcasted RHS tensor
+   */
+  public Output<T> rhsOutput() {
+    return rhsOutput;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "XlaBroadcastHelper";
+  
+  private Output<T> lhsOutput;
+  private Output<T> rhsOutput;
+  
+  private BroadcastHelper(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    lhsOutput = operation.output(outputIdx++);
+    rhsOutput = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/ClusterOutput.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/ClusterOutput.java
new file mode 100644
index 00000000000..717b65c931e
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/ClusterOutput.java
@@ -0,0 +1,74 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.xla;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Operator that connects the output of an XLA computation to other consumer graph nodes.
+ * 
+ * @param <T> data type for {@code outputs()} output
+ */
+@Operator(group = "xla")
+public final class ClusterOutput<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new ClusterOutput operation.
+   * 
+   * @param scope current scope
+   * @param input 
+   * @return a new instance of ClusterOutput
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> ClusterOutput<T> create(Scope scope, Operand<T> input) {
+    OperationBuilder opBuilder = scope.env().opBuilder("XlaClusterOutput", scope.makeOpName("ClusterOutput"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new ClusterOutput<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> outputs() {
+    return outputs;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return outputs;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "XlaClusterOutput";
+  
+  private Output<T> outputs;
+  
+  private ClusterOutput(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    outputs = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Conv.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Conv.java
new file mode 100644
index 00000000000..05cec5b6d51
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Conv.java
@@ -0,0 +1,94 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.xla;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Wraps the XLA ConvGeneralDilated operator, documented at
+ * <p>
+ *  https://www.tensorflow.org/performance/xla/operation_semantics#conv_convolution
+ * .
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator(group = "xla")
+public final class Conv<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new Conv operation.
+   * 
+   * @param scope current scope
+   * @param lhs the input tensor
+   * @param rhs the kernel tensor
+   * @param windowStrides the inter-window strides
+   * @param padding the padding to apply at the start and end of each input dimensions
+   * @param lhsDilation dilation to apply between input elements
+   * @param rhsDilation dilation to apply between kernel elements
+   * @param featureGroupCount number of feature groups for grouped convolution.
+   * @param dimensionNumbers a serialized xla::ConvolutionDimensionNumbers proto.
+   * @param precisionConfig a serialized xla::PrecisionConfig proto.
+   * @return a new instance of Conv
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType, U extends TNumber> Conv<T> create(Scope scope, Operand<T> lhs, Operand<T> rhs, Operand<U> windowStrides, Operand<U> padding, Operand<U> lhsDilation, Operand<U> rhsDilation, Operand<U> featureGroupCount, String dimensionNumbers, String precisionConfig) {
+    OperationBuilder opBuilder = scope.env().opBuilder("XlaConv", scope.makeOpName("Conv"));
+    opBuilder.addInput(lhs.asOutput());
+    opBuilder.addInput(rhs.asOutput());
+    opBuilder.addInput(windowStrides.asOutput());
+    opBuilder.addInput(padding.asOutput());
+    opBuilder.addInput(lhsDilation.asOutput());
+    opBuilder.addInput(rhsDilation.asOutput());
+    opBuilder.addInput(featureGroupCount.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("dimension_numbers", dimensionNumbers);
+    opBuilder.setAttr("precision_config", precisionConfig);
+    return new Conv<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "XlaConv";
+  
+  private Output<T> output;
+  
+  private Conv(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Dequantize.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Dequantize.java
new file mode 100644
index 00000000000..f8cdec66d1f
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Dequantize.java
@@ -0,0 +1,86 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.xla;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TBfloat16;
+
+/**
+ * Takes the packed uint32 input and unpacks the input to uint8 to do
+ * <p>
+ * Dequantization on device.
+ */
+@Operator(group = "xla")
+public final class Dequantize extends RawOp implements Operand<TBfloat16> {
+  
+  /**
+   * Factory method to create a class wrapping a new Dequantize operation.
+   * 
+   * @param scope current scope
+   * @param input Input tensors whose types is uint32, shape is [d0, ..., dn].
+   * @param minRange The minimum scalar value possibly produced for the input.
+   * @param maxRange The maximum scalar value possibly produced for the input.
+   * @param mode String to determine the dequantize mode in {"MIN_COMBINED", "MIN_FIRST", "SCALED"}.
+   * @param transposeOutput Boolean to determine if output is transposed. transpose_output
+   * is faster when input is large and rank of input is higher than 1.
+   * @return a new instance of Dequantize
+   */
+  @Endpoint(describeByClass = true)
+  public static Dequantize create(Scope scope, Operand<?> input, Float minRange, Float maxRange, String mode, Boolean transposeOutput) {
+    OperationBuilder opBuilder = scope.env().opBuilder("XlaDequantize", scope.makeOpName("Dequantize"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("min_range", minRange);
+    opBuilder.setAttr("max_range", maxRange);
+    opBuilder.setAttr("mode", mode);
+    opBuilder.setAttr("transpose_output", transposeOutput);
+    return new Dequantize(opBuilder.build());
+  }
+  
+  /**
+   * Output tensors whose types is bloat16. If transpose_output is true,
+   * output shape is [dn * 4, dn-1, ..., d1, d0]. If transpose_output
+   * is false, output shape is [d0,..., dn * 4].
+   */
+  public Output<TBfloat16> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<TBfloat16> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "XlaDequantize";
+  
+  private Output<TBfloat16> output;
+  
+  private Dequantize(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Dot.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Dot.java
new file mode 100644
index 00000000000..9cb83d061cb
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Dot.java
@@ -0,0 +1,83 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.xla;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Wraps the XLA DotGeneral operator, documented at
+ * <p>
+ *  https://www.tensorflow.org/performance/xla/operation_semantics#dotgeneral
+ * .
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator(group = "xla")
+public final class Dot<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new Dot operation.
+   * 
+   * @param scope current scope
+   * @param lhs the LHS tensor
+   * @param rhs the RHS tensor
+   * @param dimensionNumbers a serialized xla::DotDimensionNumbers proto.
+   * @param precisionConfig a serialized xla::PrecisionConfig proto.
+   * @return a new instance of Dot
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> Dot<T> create(Scope scope, Operand<T> lhs, Operand<T> rhs, String dimensionNumbers, String precisionConfig) {
+    OperationBuilder opBuilder = scope.env().opBuilder("XlaDot", scope.makeOpName("Dot"));
+    opBuilder.addInput(lhs.asOutput());
+    opBuilder.addInput(rhs.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("dimension_numbers", dimensionNumbers);
+    opBuilder.setAttr("precision_config", precisionConfig);
+    return new Dot<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "XlaDot";
+  
+  private Output<T> output;
+  
+  private Dot(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/DynamicSlice.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/DynamicSlice.java
new file mode 100644
index 00000000000..b9e2bebf9b8
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/DynamicSlice.java
@@ -0,0 +1,91 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.xla;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Wraps the XLA DynamicSlice operator, documented at
+ * <p>
+ *  https://www.tensorflow.org/performance/xla/operation_semantics#dynamicslice
+ * .
+ * <p>
+ * DynamicSlice extracts a sub-array from the input array at dynamic
+ * start_indices. The size of the slice in each dimension is passed in
+ * size_indices, which specify the end point of exclusive slice intervals in each
+ * dimension -- [start, start + size). The shape of start_indices must have rank 1,
+ * with dimension size equal to the rank of operand.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator(group = "xla")
+public final class DynamicSlice<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new DynamicSlice operation.
+   * 
+   * @param scope current scope
+   * @param input A `Tensor` of type T.
+   * @param startIndices List of N integers containing the slice size for each
+   * dimension. Each value must be strictly greater than zero, and start + size
+   * must be less than or equal to the size of the dimension to avoid
+   * implementation defined behavior.
+   * @param sizeIndices 
+   * @return a new instance of DynamicSlice
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType, U extends TNumber> DynamicSlice<T> create(Scope scope, Operand<T> input, Operand<U> startIndices, Operand<U> sizeIndices) {
+    OperationBuilder opBuilder = scope.env().opBuilder("XlaDynamicSlice", scope.makeOpName("DynamicSlice"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(startIndices.asOutput());
+    opBuilder.addInput(sizeIndices.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new DynamicSlice<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "XlaDynamicSlice";
+  
+  private Output<T> output;
+  
+  private DynamicSlice(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/DynamicUpdateSlice.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/DynamicUpdateSlice.java
new file mode 100644
index 00000000000..dccbdd10da8
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/DynamicUpdateSlice.java
@@ -0,0 +1,91 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.xla;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Wraps the XLA DynamicUpdateSlice operator, documented at
+ * <p>
+ *  https://www.tensorflow.org/performance/xla/operation_semantics#dynamicupdateslice
+ * .
+ * <p>
+ * XlaDynamicUpdateSlice generates a result which is the value of the `input`
+ * operand, with a slice update overwritten at `indices`. The shape of `update`
+ * determines the shape of the sub-array of the result which is updated. The shape
+ * of indices must be rank == 1, with dimension size equal to the rank of `input`.
+ * <p>
+ * Handling of out-of-bounds slice indices is implementation-defined.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator(group = "xla")
+public final class DynamicUpdateSlice<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new DynamicUpdateSlice operation.
+   * 
+   * @param scope current scope
+   * @param input A `Tensor` of type T.
+   * @param update A `Tensor` of type T. Same rank as `input`.
+   * @param indices A vector of indices into `input`. Must have length equal to the rank of
+   * `input`.
+   * @return a new instance of DynamicUpdateSlice
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType, U extends TNumber> DynamicUpdateSlice<T> create(Scope scope, Operand<T> input, Operand<T> update, Operand<U> indices) {
+    OperationBuilder opBuilder = scope.env().opBuilder("XlaDynamicUpdateSlice", scope.makeOpName("DynamicUpdateSlice"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(update.asOutput());
+    opBuilder.addInput(indices.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new DynamicUpdateSlice<T>(opBuilder.build());
+  }
+  
+  /**
+   * A `Tensor` of type T.
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "XlaDynamicUpdateSlice";
+  
+  private Output<T> output;
+  
+  private DynamicUpdateSlice(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Einsum.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Einsum.java
new file mode 100644
index 00000000000..451fb27bc49
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Einsum.java
@@ -0,0 +1,81 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.xla;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * An op which supports basic einsum op with 2 inputs and 1 output.
+ * <p>
+ * This op has better TPU performance since it doesn't have explicitly reshape and
+ * transpose operations as tf.einsum does.
+ * 
+ * @param <T> data type for {@code product()} output
+ */
+@Operator(group = "xla")
+public final class Einsum<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new Einsum operation.
+   * 
+   * @param scope current scope
+   * @param a 
+   * @param b 
+   * @param equation 
+   * @return a new instance of Einsum
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> Einsum<T> create(Scope scope, Operand<T> a, Operand<T> b, String equation) {
+    OperationBuilder opBuilder = scope.env().opBuilder("XlaEinsum", scope.makeOpName("Einsum"));
+    opBuilder.addInput(a.asOutput());
+    opBuilder.addInput(b.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("equation", equation);
+    return new Einsum<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> product() {
+    return product;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return product;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "XlaEinsum";
+  
+  private Output<T> product;
+  
+  private Einsum(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    product = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Gather.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Gather.java
new file mode 100644
index 00000000000..9daa1808620
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Gather.java
@@ -0,0 +1,85 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.xla;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Wraps the XLA Gather operator documented at
+ * <p>
+ *   https://www.tensorflow.org/xla/operation_semantics#gather
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator(group = "xla")
+public final class Gather<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new Gather operation.
+   * 
+   * @param scope current scope
+   * @param operand The array we're gathering from.
+   * @param startIndices Array containing the starting indices of the slices we gather.
+   * @param sliceSizes slice_sizes[i] is the bounds for the slice on dimension i.
+   * @param dimensionNumbers A serialized xla::GatherDimensionNumbers proto.
+   * @param indicesAreSorted Boolean indicating if the indices are sorted.
+   * @return a new instance of Gather
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType, U extends TNumber> Gather<T> create(Scope scope, Operand<T> operand, Operand<U> startIndices, Operand<U> sliceSizes, String dimensionNumbers, Boolean indicesAreSorted) {
+    OperationBuilder opBuilder = scope.env().opBuilder("XlaGather", scope.makeOpName("Gather"));
+    opBuilder.addInput(operand.asOutput());
+    opBuilder.addInput(startIndices.asOutput());
+    opBuilder.addInput(sliceSizes.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("dimension_numbers", dimensionNumbers);
+    opBuilder.setAttr("indices_are_sorted", indicesAreSorted);
+    return new Gather<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "XlaGather";
+  
+  private Output<T> output;
+  
+  private Gather(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/KeyValueSort.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/KeyValueSort.java
new file mode 100644
index 00000000000..d1cf154bd8a
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/KeyValueSort.java
@@ -0,0 +1,88 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.xla;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Wraps the XLA Sort operator, documented at
+ * <p>
+ *  https://www.tensorflow.org/performance/xla/operation_semantics#sort
+ * .
+ * <p>
+ * Sorts a tensor. Currently only sorts in ascending order are supported.
+ * 
+ * @param <T> data type for {@code sortedKeys()} output
+ * @param <U> data type for {@code sortedValues()} output
+ */
+@Operator(group = "xla")
+public final class KeyValueSort<T extends TNumber, U extends TType> extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new KeyValueSort operation.
+   * 
+   * @param scope current scope
+   * @param keys A `Tensor` of type K.
+   * @param values A `Tensor` of type V.
+   * @return a new instance of KeyValueSort
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TNumber, U extends TType> KeyValueSort<T, U> create(Scope scope, Operand<T> keys, Operand<U> values) {
+    OperationBuilder opBuilder = scope.env().opBuilder("XlaKeyValueSort", scope.makeOpName("KeyValueSort"));
+    opBuilder.addInput(keys.asOutput());
+    opBuilder.addInput(values.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new KeyValueSort<T, U>(opBuilder.build());
+  }
+  
+  /**
+   * A `Tensor` of type K.
+   */
+  public Output<T> sortedKeys() {
+    return sortedKeys;
+  }
+  
+  /**
+   * A `Tensor` of type V.
+   */
+  public Output<U> sortedValues() {
+    return sortedValues;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "XlaKeyValueSort";
+  
+  private Output<T> sortedKeys;
+  private Output<U> sortedValues;
+  
+  private KeyValueSort(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    sortedKeys = operation.output(outputIdx++);
+    sortedValues = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Pad.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Pad.java
new file mode 100644
index 00000000000..c56631c48f2
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Pad.java
@@ -0,0 +1,87 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.xla;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TNumber;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Wraps the XLA Pad operator, documented at
+ * <p>
+ *  https://www.tensorflow.org/performance/xla/operation_semantics#pad
+ * .
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator(group = "xla")
+public final class Pad<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new Pad operation.
+   * 
+   * @param scope current scope
+   * @param input A `Tensor` of type T.
+   * @param paddingValue A scalar `Tensor` of type T.
+   * @param paddingLow the padding to apply at the start of each input dimensions
+   * @param paddingHigh the padding to apply at the end of each input dimension.
+   * @param paddingInterior the padding to apply between each input element.
+   * @return a new instance of Pad
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType, U extends TNumber> Pad<T> create(Scope scope, Operand<T> input, Operand<T> paddingValue, Operand<U> paddingLow, Operand<U> paddingHigh, Operand<U> paddingInterior) {
+    OperationBuilder opBuilder = scope.env().opBuilder("XlaPad", scope.makeOpName("Pad"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder.addInput(paddingValue.asOutput());
+    opBuilder.addInput(paddingLow.asOutput());
+    opBuilder.addInput(paddingHigh.asOutput());
+    opBuilder.addInput(paddingInterior.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new Pad<T>(opBuilder.build());
+  }
+  
+  /**
+   * A `Tensor` of type T.
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "XlaPad";
+  
+  private Output<T> output;
+  
+  private Pad(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Recv.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Recv.java
new file mode 100644
index 00000000000..63cdcc2767b
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Recv.java
@@ -0,0 +1,83 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.xla;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.ndarray.Shape;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Receives the named tensor from another XLA computation. Wraps the XLA Recv
+ * <p>
+ * operator documented at
+ *  https://www.tensorflow.org/performance/xla/operation_semantics#recv .
+ * 
+ * @param <T> data type for {@code tensor()} output
+ */
+@Operator(group = "xla")
+public final class Recv<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new Recv operation.
+   * 
+   * @param scope current scope
+   * @param dtype The type of the tensor.
+   * @param tensorName A string key that identifies the channel.
+   * @param shape The shape of the tensor.
+   * @return a new instance of Recv
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> Recv<T> create(Scope scope, Class<T> dtype, String tensorName, Shape shape) {
+    OperationBuilder opBuilder = scope.env().opBuilder("XlaRecv", scope.makeOpName("Recv"));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("dtype", dtype);
+    opBuilder.setAttr("tensor_name", tensorName);
+    opBuilder.setAttr("shape", shape);
+    return new Recv<T>(opBuilder.build());
+  }
+  
+  /**
+   * The tensor to receive.
+   */
+  public Output<T> tensor() {
+    return tensor;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return tensor;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "XlaRecv";
+  
+  private Output<T> tensor;
+  
+  private Recv(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    tensor = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/ReplicaId.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/ReplicaId.java
new file mode 100644
index 00000000000..886cc40b3ab
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/ReplicaId.java
@@ -0,0 +1,70 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.xla;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.TInt32;
+
+/**
+ * Replica ID.
+ */
+@Operator(group = "xla")
+public final class ReplicaId extends RawOp implements Operand<TInt32> {
+  
+  /**
+   * Factory method to create a class wrapping a new ReplicaId operation.
+   * 
+   * @param scope current scope
+   * @return a new instance of ReplicaId
+   */
+  @Endpoint(describeByClass = true)
+  public static ReplicaId create(Scope scope) {
+    OperationBuilder opBuilder = scope.env().opBuilder("XlaReplicaId", scope.makeOpName("ReplicaId"));
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new ReplicaId(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<TInt32> id() {
+    return id;
+  }
+  
+  @Override
+  public Output<TInt32> asOutput() {
+    return id;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "XlaReplicaId";
+  
+  private Output<TInt32> id;
+  
+  private ReplicaId(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    id = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/SelfAdjointEig.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/SelfAdjointEig.java
new file mode 100644
index 00000000000..9ce4468fdfe
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/SelfAdjointEig.java
@@ -0,0 +1,97 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.xla;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Computes the eigen decomposition of a batch of self-adjoint matrices
+ * <p>
+ * (Note: Only real inputs are supported).
+ * <p>
+ * Computes the eigenvalues and eigenvectors of the innermost N-by-N matrices in
+ * tensor such that tensor[...,:,:] * v[..., :,i] = e[..., i] * v[...,:,i], for
+ * i=0...N-1.
+ * 
+ * @param <T> data type for {@code w()} output
+ */
+@Operator(group = "xla")
+public final class SelfAdjointEig<T extends TType> extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new SelfAdjointEig operation.
+   * 
+   * @param scope current scope
+   * @param a the input tensor.
+   * @param lower a boolean specifies whether the calculation is done with the lower
+   * triangular part or the upper triangular part.
+   * @param maxIter maximum number of sweep update, i.e., the whole lower triangular
+   * part or upper triangular part based on parameter lower. Heuristically, it has
+   * been argued that approximately logN sweeps are needed in practice (Ref: Golub &
+   * van Loan "Matrix Computation").
+   * @param epsilon the tolerance ratio.
+   * @return a new instance of SelfAdjointEig
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> SelfAdjointEig<T> create(Scope scope, Operand<T> a, Boolean lower, Long maxIter, Float epsilon) {
+    OperationBuilder opBuilder = scope.env().opBuilder("XlaSelfAdjointEig", scope.makeOpName("SelfAdjointEig"));
+    opBuilder.addInput(a.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("lower", lower);
+    opBuilder.setAttr("max_iter", maxIter);
+    opBuilder.setAttr("epsilon", epsilon);
+    return new SelfAdjointEig<T>(opBuilder.build());
+  }
+  
+  /**
+   * The eigenvalues in ascending order, each repeated according to its
+   * multiplicity.
+   */
+  public Output<T> w() {
+    return w;
+  }
+  
+  /**
+   * The column v[..., :, i] is the normalized eigenvector corresponding to the
+   * eigenvalue w[..., i].
+   */
+  public Output<T> v() {
+    return v;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "XlaSelfAdjointEig";
+  
+  private Output<T> w;
+  private Output<T> v;
+  
+  private SelfAdjointEig(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    w = operation.output(outputIdx++);
+    v = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Send.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Send.java
new file mode 100644
index 00000000000..b18a86458ca
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Send.java
@@ -0,0 +1,61 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.xla;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Sends the named tensor to another XLA computation. Wraps the XLA Send operator
+ * <p>
+ * documented at
+ *  https://www.tensorflow.org/performance/xla/operation_semantics#send .
+ */
+@Operator(group = "xla")
+public final class Send extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new Send operation.
+   * 
+   * @param scope current scope
+   * @param tensor The tensor to send.
+   * @param tensorName A string key that identifies the channel.
+   * @return a new instance of Send
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> Send create(Scope scope, Operand<T> tensor, String tensorName) {
+    OperationBuilder opBuilder = scope.env().opBuilder("XlaSend", scope.makeOpName("Send"));
+    opBuilder.addInput(tensor.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("tensor_name", tensorName);
+    return new Send(opBuilder.build());
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "XlaSend";
+  
+  private Send(Operation operation) {
+    super(operation);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Sharding.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Sharding.java
new file mode 100644
index 00000000000..dbc255aa607
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Sharding.java
@@ -0,0 +1,74 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.xla;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * An op which shards the input based on the given sharding attribute.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator(group = "xla")
+public final class Sharding<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new Sharding operation.
+   * 
+   * @param scope current scope
+   * @param input 
+   * @return a new instance of Sharding
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> Sharding<T> create(Scope scope, Operand<T> input) {
+    OperationBuilder opBuilder = scope.env().opBuilder("XlaSharding", scope.makeOpName("Sharding"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new Sharding<T>(opBuilder.build());
+  }
+  
+  /**
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "XlaSharding";
+  
+  private Output<T> output;
+  
+  private Sharding(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Sort.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Sort.java
new file mode 100644
index 00000000000..d7f9aae6449
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Sort.java
@@ -0,0 +1,80 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.xla;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Wraps the XLA Sort operator, documented at
+ * <p>
+ *  https://www.tensorflow.org/performance/xla/operation_semantics#sort
+ * .
+ * <p>
+ * Sorts a tensor. Currently only sorts in ascending order are supported.
+ * 
+ * @param <T> data type for {@code output()} output
+ */
+@Operator(group = "xla")
+public final class Sort<T extends TType> extends RawOp implements Operand<T> {
+  
+  /**
+   * Factory method to create a class wrapping a new Sort operation.
+   * 
+   * @param scope current scope
+   * @param input A `Tensor` of type T.
+   * @return a new instance of Sort
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> Sort<T> create(Scope scope, Operand<T> input) {
+    OperationBuilder opBuilder = scope.env().opBuilder("XlaSort", scope.makeOpName("Sort"));
+    opBuilder.addInput(input.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    return new Sort<T>(opBuilder.build());
+  }
+  
+  /**
+   * A `Tensor` of type T.
+   */
+  public Output<T> output() {
+    return output;
+  }
+  
+  @Override
+  public Output<T> asOutput() {
+    return output;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "XlaSort";
+  
+  private Output<T> output;
+  
+  private Sort(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    output = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Svd.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Svd.java
new file mode 100644
index 00000000000..cd0a67f33b4
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/op/xla/Svd.java
@@ -0,0 +1,103 @@
+/* Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+=======================================================================*/
+
+// This class has been generated, DO NOT EDIT!
+
+package org.tensorflow.op.xla;
+
+import org.tensorflow.Operand;
+import org.tensorflow.Operation;
+import org.tensorflow.OperationBuilder;
+import org.tensorflow.Output;
+import org.tensorflow.op.RawOp;
+import org.tensorflow.op.Scope;
+import org.tensorflow.op.annotation.Endpoint;
+import org.tensorflow.op.annotation.Operator;
+import org.tensorflow.types.family.TType;
+
+/**
+ * Computes the eigen decomposition of a batch of self-adjoint matrices
+ * <p>
+ * (Note: Only real inputs are supported).
+ * <p>
+ * Computes the eigenvalues and eigenvectors of the innermost M-by-N matrices in
+ * tensor such that tensor[...,:,:] = u[..., :, :] * Diag(s[..., :]) * Transpose(v[...,:,:]).
+ * 
+ * @param <T> data type for {@code s()} output
+ */
+@Operator(group = "xla")
+public final class Svd<T extends TType> extends RawOp {
+  
+  /**
+   * Factory method to create a class wrapping a new Svd operation.
+   * 
+   * @param scope current scope
+   * @param a the input tensor.
+   * @param maxIter maximum number of sweep update, i.e., the whole lower triangular
+   * part or upper triangular part based on parameter lower. Heuristically, it has
+   * been argued that approximately log(min (M, N)) sweeps are needed in practice
+   * (Ref: Golub & van Loan "Matrix Computation").
+   * @param epsilon the tolerance ratio.
+   * @param precisionConfig a serialized xla::PrecisionConfig proto.
+   * @return a new instance of Svd
+   */
+  @Endpoint(describeByClass = true)
+  public static <T extends TType> Svd<T> create(Scope scope, Operand<T> a, Long maxIter, Float epsilon, String precisionConfig) {
+    OperationBuilder opBuilder = scope.env().opBuilder("XlaSvd", scope.makeOpName("Svd"));
+    opBuilder.addInput(a.asOutput());
+    opBuilder = scope.applyControlDependencies(opBuilder);
+    opBuilder.setAttr("max_iter", maxIter);
+    opBuilder.setAttr("epsilon", epsilon);
+    opBuilder.setAttr("precision_config", precisionConfig);
+    return new Svd<T>(opBuilder.build());
+  }
+  
+  /**
+   * Singular values. The values are sorted in reverse order of magnitude, so
+   * s[..., 0] is the largest value, s[..., 1] is the second largest, etc.
+   */
+  public Output<T> s() {
+    return s;
+  }
+  
+  /**
+   * Left singular vectors.
+   */
+  public Output<T> u() {
+    return u;
+  }
+  
+  /**
+   * Right singular vectors.
+   */
+  public Output<T> v() {
+    return v;
+  }
+  
+  /** The name of this op, as known by TensorFlow core engine */
+  public static final String OP_NAME = "XlaSvd";
+  
+  private Output<T> s;
+  private Output<T> u;
+  private Output<T> v;
+  
+  private Svd(Operation operation) {
+    super(operation);
+    int outputIdx = 0;
+    s = operation.output(outputIdx++);
+    u = operation.output(outputIdx++);
+    v = operation.output(outputIdx++);
+  }
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/distruntime/ClusterDef.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/distruntime/ClusterDef.java
new file mode 100644
index 00000000000..f796e47d02f
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/distruntime/ClusterDef.java
@@ -0,0 +1,865 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/cluster.proto
+
+package org.tensorflow.proto.distruntime;
+
+/**
+ * <pre>
+ * Defines a TensorFlow cluster as a set of jobs.
+ * </pre>
+ *
+ * Protobuf type {@code tensorflow.ClusterDef}
+ */
+public  final class ClusterDef extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.ClusterDef)
+    ClusterDefOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use ClusterDef.newBuilder() to construct.
+  private ClusterDef(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private ClusterDef() {
+    job_ = java.util.Collections.emptyList();
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new ClusterDef();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private ClusterDef(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    int mutable_bitField0_ = 0;
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 10: {
+            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
+              job_ = new java.util.ArrayList<org.tensorflow.proto.distruntime.JobDef>();
+              mutable_bitField0_ |= 0x00000001;
+            }
+            job_.add(
+                input.readMessage(org.tensorflow.proto.distruntime.JobDef.parser(), extensionRegistry));
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      if (((mutable_bitField0_ & 0x00000001) != 0)) {
+        job_ = java.util.Collections.unmodifiableList(job_);
+      }
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.distruntime.ClusterProtos.internal_static_tensorflow_ClusterDef_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.distruntime.ClusterProtos.internal_static_tensorflow_ClusterDef_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.distruntime.ClusterDef.class, org.tensorflow.proto.distruntime.ClusterDef.Builder.class);
+  }
+
+  public static final int JOB_FIELD_NUMBER = 1;
+  private java.util.List<org.tensorflow.proto.distruntime.JobDef> job_;
+  /**
+   * <pre>
+   * The jobs that comprise the cluster.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.JobDef job = 1;</code>
+   */
+  public java.util.List<org.tensorflow.proto.distruntime.JobDef> getJobList() {
+    return job_;
+  }
+  /**
+   * <pre>
+   * The jobs that comprise the cluster.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.JobDef job = 1;</code>
+   */
+  public java.util.List<? extends org.tensorflow.proto.distruntime.JobDefOrBuilder> 
+      getJobOrBuilderList() {
+    return job_;
+  }
+  /**
+   * <pre>
+   * The jobs that comprise the cluster.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.JobDef job = 1;</code>
+   */
+  public int getJobCount() {
+    return job_.size();
+  }
+  /**
+   * <pre>
+   * The jobs that comprise the cluster.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.JobDef job = 1;</code>
+   */
+  public org.tensorflow.proto.distruntime.JobDef getJob(int index) {
+    return job_.get(index);
+  }
+  /**
+   * <pre>
+   * The jobs that comprise the cluster.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.JobDef job = 1;</code>
+   */
+  public org.tensorflow.proto.distruntime.JobDefOrBuilder getJobOrBuilder(
+      int index) {
+    return job_.get(index);
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    for (int i = 0; i < job_.size(); i++) {
+      output.writeMessage(1, job_.get(i));
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    for (int i = 0; i < job_.size(); i++) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(1, job_.get(i));
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.distruntime.ClusterDef)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.distruntime.ClusterDef other = (org.tensorflow.proto.distruntime.ClusterDef) obj;
+
+    if (!getJobList()
+        .equals(other.getJobList())) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    if (getJobCount() > 0) {
+      hash = (37 * hash) + JOB_FIELD_NUMBER;
+      hash = (53 * hash) + getJobList().hashCode();
+    }
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.distruntime.ClusterDef parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.distruntime.ClusterDef parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.distruntime.ClusterDef parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.distruntime.ClusterDef parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.distruntime.ClusterDef parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.distruntime.ClusterDef parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.distruntime.ClusterDef parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.distruntime.ClusterDef parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.distruntime.ClusterDef parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.distruntime.ClusterDef parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.distruntime.ClusterDef parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.distruntime.ClusterDef parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.distruntime.ClusterDef prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * <pre>
+   * Defines a TensorFlow cluster as a set of jobs.
+   * </pre>
+   *
+   * Protobuf type {@code tensorflow.ClusterDef}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.ClusterDef)
+      org.tensorflow.proto.distruntime.ClusterDefOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.distruntime.ClusterProtos.internal_static_tensorflow_ClusterDef_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.distruntime.ClusterProtos.internal_static_tensorflow_ClusterDef_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.distruntime.ClusterDef.class, org.tensorflow.proto.distruntime.ClusterDef.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.distruntime.ClusterDef.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+        getJobFieldBuilder();
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      if (jobBuilder_ == null) {
+        job_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000001);
+      } else {
+        jobBuilder_.clear();
+      }
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.distruntime.ClusterProtos.internal_static_tensorflow_ClusterDef_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.distruntime.ClusterDef getDefaultInstanceForType() {
+      return org.tensorflow.proto.distruntime.ClusterDef.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.distruntime.ClusterDef build() {
+      org.tensorflow.proto.distruntime.ClusterDef result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.distruntime.ClusterDef buildPartial() {
+      org.tensorflow.proto.distruntime.ClusterDef result = new org.tensorflow.proto.distruntime.ClusterDef(this);
+      int from_bitField0_ = bitField0_;
+      if (jobBuilder_ == null) {
+        if (((bitField0_ & 0x00000001) != 0)) {
+          job_ = java.util.Collections.unmodifiableList(job_);
+          bitField0_ = (bitField0_ & ~0x00000001);
+        }
+        result.job_ = job_;
+      } else {
+        result.job_ = jobBuilder_.build();
+      }
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.distruntime.ClusterDef) {
+        return mergeFrom((org.tensorflow.proto.distruntime.ClusterDef)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.distruntime.ClusterDef other) {
+      if (other == org.tensorflow.proto.distruntime.ClusterDef.getDefaultInstance()) return this;
+      if (jobBuilder_ == null) {
+        if (!other.job_.isEmpty()) {
+          if (job_.isEmpty()) {
+            job_ = other.job_;
+            bitField0_ = (bitField0_ & ~0x00000001);
+          } else {
+            ensureJobIsMutable();
+            job_.addAll(other.job_);
+          }
+          onChanged();
+        }
+      } else {
+        if (!other.job_.isEmpty()) {
+          if (jobBuilder_.isEmpty()) {
+            jobBuilder_.dispose();
+            jobBuilder_ = null;
+            job_ = other.job_;
+            bitField0_ = (bitField0_ & ~0x00000001);
+            jobBuilder_ = 
+              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
+                 getJobFieldBuilder() : null;
+          } else {
+            jobBuilder_.addAllMessages(other.job_);
+          }
+        }
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.distruntime.ClusterDef parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.distruntime.ClusterDef) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+    private int bitField0_;
+
+    private java.util.List<org.tensorflow.proto.distruntime.JobDef> job_ =
+      java.util.Collections.emptyList();
+    private void ensureJobIsMutable() {
+      if (!((bitField0_ & 0x00000001) != 0)) {
+        job_ = new java.util.ArrayList<org.tensorflow.proto.distruntime.JobDef>(job_);
+        bitField0_ |= 0x00000001;
+       }
+    }
+
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.distruntime.JobDef, org.tensorflow.proto.distruntime.JobDef.Builder, org.tensorflow.proto.distruntime.JobDefOrBuilder> jobBuilder_;
+
+    /**
+     * <pre>
+     * The jobs that comprise the cluster.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.JobDef job = 1;</code>
+     */
+    public java.util.List<org.tensorflow.proto.distruntime.JobDef> getJobList() {
+      if (jobBuilder_ == null) {
+        return java.util.Collections.unmodifiableList(job_);
+      } else {
+        return jobBuilder_.getMessageList();
+      }
+    }
+    /**
+     * <pre>
+     * The jobs that comprise the cluster.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.JobDef job = 1;</code>
+     */
+    public int getJobCount() {
+      if (jobBuilder_ == null) {
+        return job_.size();
+      } else {
+        return jobBuilder_.getCount();
+      }
+    }
+    /**
+     * <pre>
+     * The jobs that comprise the cluster.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.JobDef job = 1;</code>
+     */
+    public org.tensorflow.proto.distruntime.JobDef getJob(int index) {
+      if (jobBuilder_ == null) {
+        return job_.get(index);
+      } else {
+        return jobBuilder_.getMessage(index);
+      }
+    }
+    /**
+     * <pre>
+     * The jobs that comprise the cluster.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.JobDef job = 1;</code>
+     */
+    public Builder setJob(
+        int index, org.tensorflow.proto.distruntime.JobDef value) {
+      if (jobBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureJobIsMutable();
+        job_.set(index, value);
+        onChanged();
+      } else {
+        jobBuilder_.setMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * The jobs that comprise the cluster.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.JobDef job = 1;</code>
+     */
+    public Builder setJob(
+        int index, org.tensorflow.proto.distruntime.JobDef.Builder builderForValue) {
+      if (jobBuilder_ == null) {
+        ensureJobIsMutable();
+        job_.set(index, builderForValue.build());
+        onChanged();
+      } else {
+        jobBuilder_.setMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * The jobs that comprise the cluster.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.JobDef job = 1;</code>
+     */
+    public Builder addJob(org.tensorflow.proto.distruntime.JobDef value) {
+      if (jobBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureJobIsMutable();
+        job_.add(value);
+        onChanged();
+      } else {
+        jobBuilder_.addMessage(value);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * The jobs that comprise the cluster.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.JobDef job = 1;</code>
+     */
+    public Builder addJob(
+        int index, org.tensorflow.proto.distruntime.JobDef value) {
+      if (jobBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureJobIsMutable();
+        job_.add(index, value);
+        onChanged();
+      } else {
+        jobBuilder_.addMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * The jobs that comprise the cluster.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.JobDef job = 1;</code>
+     */
+    public Builder addJob(
+        org.tensorflow.proto.distruntime.JobDef.Builder builderForValue) {
+      if (jobBuilder_ == null) {
+        ensureJobIsMutable();
+        job_.add(builderForValue.build());
+        onChanged();
+      } else {
+        jobBuilder_.addMessage(builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * The jobs that comprise the cluster.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.JobDef job = 1;</code>
+     */
+    public Builder addJob(
+        int index, org.tensorflow.proto.distruntime.JobDef.Builder builderForValue) {
+      if (jobBuilder_ == null) {
+        ensureJobIsMutable();
+        job_.add(index, builderForValue.build());
+        onChanged();
+      } else {
+        jobBuilder_.addMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * The jobs that comprise the cluster.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.JobDef job = 1;</code>
+     */
+    public Builder addAllJob(
+        java.lang.Iterable<? extends org.tensorflow.proto.distruntime.JobDef> values) {
+      if (jobBuilder_ == null) {
+        ensureJobIsMutable();
+        com.google.protobuf.AbstractMessageLite.Builder.addAll(
+            values, job_);
+        onChanged();
+      } else {
+        jobBuilder_.addAllMessages(values);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * The jobs that comprise the cluster.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.JobDef job = 1;</code>
+     */
+    public Builder clearJob() {
+      if (jobBuilder_ == null) {
+        job_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000001);
+        onChanged();
+      } else {
+        jobBuilder_.clear();
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * The jobs that comprise the cluster.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.JobDef job = 1;</code>
+     */
+    public Builder removeJob(int index) {
+      if (jobBuilder_ == null) {
+        ensureJobIsMutable();
+        job_.remove(index);
+        onChanged();
+      } else {
+        jobBuilder_.remove(index);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * The jobs that comprise the cluster.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.JobDef job = 1;</code>
+     */
+    public org.tensorflow.proto.distruntime.JobDef.Builder getJobBuilder(
+        int index) {
+      return getJobFieldBuilder().getBuilder(index);
+    }
+    /**
+     * <pre>
+     * The jobs that comprise the cluster.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.JobDef job = 1;</code>
+     */
+    public org.tensorflow.proto.distruntime.JobDefOrBuilder getJobOrBuilder(
+        int index) {
+      if (jobBuilder_ == null) {
+        return job_.get(index);  } else {
+        return jobBuilder_.getMessageOrBuilder(index);
+      }
+    }
+    /**
+     * <pre>
+     * The jobs that comprise the cluster.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.JobDef job = 1;</code>
+     */
+    public java.util.List<? extends org.tensorflow.proto.distruntime.JobDefOrBuilder> 
+         getJobOrBuilderList() {
+      if (jobBuilder_ != null) {
+        return jobBuilder_.getMessageOrBuilderList();
+      } else {
+        return java.util.Collections.unmodifiableList(job_);
+      }
+    }
+    /**
+     * <pre>
+     * The jobs that comprise the cluster.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.JobDef job = 1;</code>
+     */
+    public org.tensorflow.proto.distruntime.JobDef.Builder addJobBuilder() {
+      return getJobFieldBuilder().addBuilder(
+          org.tensorflow.proto.distruntime.JobDef.getDefaultInstance());
+    }
+    /**
+     * <pre>
+     * The jobs that comprise the cluster.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.JobDef job = 1;</code>
+     */
+    public org.tensorflow.proto.distruntime.JobDef.Builder addJobBuilder(
+        int index) {
+      return getJobFieldBuilder().addBuilder(
+          index, org.tensorflow.proto.distruntime.JobDef.getDefaultInstance());
+    }
+    /**
+     * <pre>
+     * The jobs that comprise the cluster.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.JobDef job = 1;</code>
+     */
+    public java.util.List<org.tensorflow.proto.distruntime.JobDef.Builder> 
+         getJobBuilderList() {
+      return getJobFieldBuilder().getBuilderList();
+    }
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.distruntime.JobDef, org.tensorflow.proto.distruntime.JobDef.Builder, org.tensorflow.proto.distruntime.JobDefOrBuilder> 
+        getJobFieldBuilder() {
+      if (jobBuilder_ == null) {
+        jobBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
+            org.tensorflow.proto.distruntime.JobDef, org.tensorflow.proto.distruntime.JobDef.Builder, org.tensorflow.proto.distruntime.JobDefOrBuilder>(
+                job_,
+                ((bitField0_ & 0x00000001) != 0),
+                getParentForChildren(),
+                isClean());
+        job_ = null;
+      }
+      return jobBuilder_;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.ClusterDef)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.ClusterDef)
+  private static final org.tensorflow.proto.distruntime.ClusterDef DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.distruntime.ClusterDef();
+  }
+
+  public static org.tensorflow.proto.distruntime.ClusterDef getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<ClusterDef>
+      PARSER = new com.google.protobuf.AbstractParser<ClusterDef>() {
+    @java.lang.Override
+    public ClusterDef parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new ClusterDef(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<ClusterDef> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<ClusterDef> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.distruntime.ClusterDef getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/distruntime/ClusterDefOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/distruntime/ClusterDefOrBuilder.java
new file mode 100644
index 00000000000..8cecd262cc2
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/distruntime/ClusterDefOrBuilder.java
@@ -0,0 +1,53 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/cluster.proto
+
+package org.tensorflow.proto.distruntime;
+
+public interface ClusterDefOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.ClusterDef)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <pre>
+   * The jobs that comprise the cluster.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.JobDef job = 1;</code>
+   */
+  java.util.List<org.tensorflow.proto.distruntime.JobDef> 
+      getJobList();
+  /**
+   * <pre>
+   * The jobs that comprise the cluster.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.JobDef job = 1;</code>
+   */
+  org.tensorflow.proto.distruntime.JobDef getJob(int index);
+  /**
+   * <pre>
+   * The jobs that comprise the cluster.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.JobDef job = 1;</code>
+   */
+  int getJobCount();
+  /**
+   * <pre>
+   * The jobs that comprise the cluster.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.JobDef job = 1;</code>
+   */
+  java.util.List<? extends org.tensorflow.proto.distruntime.JobDefOrBuilder> 
+      getJobOrBuilderList();
+  /**
+   * <pre>
+   * The jobs that comprise the cluster.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.JobDef job = 1;</code>
+   */
+  org.tensorflow.proto.distruntime.JobDefOrBuilder getJobOrBuilder(
+      int index);
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/distruntime/ClusterDeviceFilters.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/distruntime/ClusterDeviceFilters.java
new file mode 100644
index 00000000000..f9684c7e0e8
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/distruntime/ClusterDeviceFilters.java
@@ -0,0 +1,773 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/device_filters.proto
+
+package org.tensorflow.proto.distruntime;
+
+/**
+ * <pre>
+ * Defines the device filters for jobs in a cluster.
+ * </pre>
+ *
+ * Protobuf type {@code tensorflow.ClusterDeviceFilters}
+ */
+public  final class ClusterDeviceFilters extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.ClusterDeviceFilters)
+    ClusterDeviceFiltersOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use ClusterDeviceFilters.newBuilder() to construct.
+  private ClusterDeviceFilters(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private ClusterDeviceFilters() {
+    jobs_ = java.util.Collections.emptyList();
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new ClusterDeviceFilters();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private ClusterDeviceFilters(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    int mutable_bitField0_ = 0;
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 10: {
+            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
+              jobs_ = new java.util.ArrayList<org.tensorflow.proto.distruntime.JobDeviceFilters>();
+              mutable_bitField0_ |= 0x00000001;
+            }
+            jobs_.add(
+                input.readMessage(org.tensorflow.proto.distruntime.JobDeviceFilters.parser(), extensionRegistry));
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      if (((mutable_bitField0_ & 0x00000001) != 0)) {
+        jobs_ = java.util.Collections.unmodifiableList(jobs_);
+      }
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.distruntime.DeviceFiltersProtos.internal_static_tensorflow_ClusterDeviceFilters_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.distruntime.DeviceFiltersProtos.internal_static_tensorflow_ClusterDeviceFilters_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.distruntime.ClusterDeviceFilters.class, org.tensorflow.proto.distruntime.ClusterDeviceFilters.Builder.class);
+  }
+
+  public static final int JOBS_FIELD_NUMBER = 1;
+  private java.util.List<org.tensorflow.proto.distruntime.JobDeviceFilters> jobs_;
+  /**
+   * <code>repeated .tensorflow.JobDeviceFilters jobs = 1;</code>
+   */
+  public java.util.List<org.tensorflow.proto.distruntime.JobDeviceFilters> getJobsList() {
+    return jobs_;
+  }
+  /**
+   * <code>repeated .tensorflow.JobDeviceFilters jobs = 1;</code>
+   */
+  public java.util.List<? extends org.tensorflow.proto.distruntime.JobDeviceFiltersOrBuilder> 
+      getJobsOrBuilderList() {
+    return jobs_;
+  }
+  /**
+   * <code>repeated .tensorflow.JobDeviceFilters jobs = 1;</code>
+   */
+  public int getJobsCount() {
+    return jobs_.size();
+  }
+  /**
+   * <code>repeated .tensorflow.JobDeviceFilters jobs = 1;</code>
+   */
+  public org.tensorflow.proto.distruntime.JobDeviceFilters getJobs(int index) {
+    return jobs_.get(index);
+  }
+  /**
+   * <code>repeated .tensorflow.JobDeviceFilters jobs = 1;</code>
+   */
+  public org.tensorflow.proto.distruntime.JobDeviceFiltersOrBuilder getJobsOrBuilder(
+      int index) {
+    return jobs_.get(index);
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    for (int i = 0; i < jobs_.size(); i++) {
+      output.writeMessage(1, jobs_.get(i));
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    for (int i = 0; i < jobs_.size(); i++) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(1, jobs_.get(i));
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.distruntime.ClusterDeviceFilters)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.distruntime.ClusterDeviceFilters other = (org.tensorflow.proto.distruntime.ClusterDeviceFilters) obj;
+
+    if (!getJobsList()
+        .equals(other.getJobsList())) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    if (getJobsCount() > 0) {
+      hash = (37 * hash) + JOBS_FIELD_NUMBER;
+      hash = (53 * hash) + getJobsList().hashCode();
+    }
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.distruntime.ClusterDeviceFilters parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.distruntime.ClusterDeviceFilters parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.distruntime.ClusterDeviceFilters parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.distruntime.ClusterDeviceFilters parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.distruntime.ClusterDeviceFilters parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.distruntime.ClusterDeviceFilters parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.distruntime.ClusterDeviceFilters parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.distruntime.ClusterDeviceFilters parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.distruntime.ClusterDeviceFilters parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.distruntime.ClusterDeviceFilters parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.distruntime.ClusterDeviceFilters parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.distruntime.ClusterDeviceFilters parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.distruntime.ClusterDeviceFilters prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * <pre>
+   * Defines the device filters for jobs in a cluster.
+   * </pre>
+   *
+   * Protobuf type {@code tensorflow.ClusterDeviceFilters}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.ClusterDeviceFilters)
+      org.tensorflow.proto.distruntime.ClusterDeviceFiltersOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.distruntime.DeviceFiltersProtos.internal_static_tensorflow_ClusterDeviceFilters_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.distruntime.DeviceFiltersProtos.internal_static_tensorflow_ClusterDeviceFilters_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.distruntime.ClusterDeviceFilters.class, org.tensorflow.proto.distruntime.ClusterDeviceFilters.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.distruntime.ClusterDeviceFilters.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+        getJobsFieldBuilder();
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      if (jobsBuilder_ == null) {
+        jobs_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000001);
+      } else {
+        jobsBuilder_.clear();
+      }
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.distruntime.DeviceFiltersProtos.internal_static_tensorflow_ClusterDeviceFilters_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.distruntime.ClusterDeviceFilters getDefaultInstanceForType() {
+      return org.tensorflow.proto.distruntime.ClusterDeviceFilters.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.distruntime.ClusterDeviceFilters build() {
+      org.tensorflow.proto.distruntime.ClusterDeviceFilters result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.distruntime.ClusterDeviceFilters buildPartial() {
+      org.tensorflow.proto.distruntime.ClusterDeviceFilters result = new org.tensorflow.proto.distruntime.ClusterDeviceFilters(this);
+      int from_bitField0_ = bitField0_;
+      if (jobsBuilder_ == null) {
+        if (((bitField0_ & 0x00000001) != 0)) {
+          jobs_ = java.util.Collections.unmodifiableList(jobs_);
+          bitField0_ = (bitField0_ & ~0x00000001);
+        }
+        result.jobs_ = jobs_;
+      } else {
+        result.jobs_ = jobsBuilder_.build();
+      }
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.distruntime.ClusterDeviceFilters) {
+        return mergeFrom((org.tensorflow.proto.distruntime.ClusterDeviceFilters)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.distruntime.ClusterDeviceFilters other) {
+      if (other == org.tensorflow.proto.distruntime.ClusterDeviceFilters.getDefaultInstance()) return this;
+      if (jobsBuilder_ == null) {
+        if (!other.jobs_.isEmpty()) {
+          if (jobs_.isEmpty()) {
+            jobs_ = other.jobs_;
+            bitField0_ = (bitField0_ & ~0x00000001);
+          } else {
+            ensureJobsIsMutable();
+            jobs_.addAll(other.jobs_);
+          }
+          onChanged();
+        }
+      } else {
+        if (!other.jobs_.isEmpty()) {
+          if (jobsBuilder_.isEmpty()) {
+            jobsBuilder_.dispose();
+            jobsBuilder_ = null;
+            jobs_ = other.jobs_;
+            bitField0_ = (bitField0_ & ~0x00000001);
+            jobsBuilder_ = 
+              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
+                 getJobsFieldBuilder() : null;
+          } else {
+            jobsBuilder_.addAllMessages(other.jobs_);
+          }
+        }
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.distruntime.ClusterDeviceFilters parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.distruntime.ClusterDeviceFilters) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+    private int bitField0_;
+
+    private java.util.List<org.tensorflow.proto.distruntime.JobDeviceFilters> jobs_ =
+      java.util.Collections.emptyList();
+    private void ensureJobsIsMutable() {
+      if (!((bitField0_ & 0x00000001) != 0)) {
+        jobs_ = new java.util.ArrayList<org.tensorflow.proto.distruntime.JobDeviceFilters>(jobs_);
+        bitField0_ |= 0x00000001;
+       }
+    }
+
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.distruntime.JobDeviceFilters, org.tensorflow.proto.distruntime.JobDeviceFilters.Builder, org.tensorflow.proto.distruntime.JobDeviceFiltersOrBuilder> jobsBuilder_;
+
+    /**
+     * <code>repeated .tensorflow.JobDeviceFilters jobs = 1;</code>
+     */
+    public java.util.List<org.tensorflow.proto.distruntime.JobDeviceFilters> getJobsList() {
+      if (jobsBuilder_ == null) {
+        return java.util.Collections.unmodifiableList(jobs_);
+      } else {
+        return jobsBuilder_.getMessageList();
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.JobDeviceFilters jobs = 1;</code>
+     */
+    public int getJobsCount() {
+      if (jobsBuilder_ == null) {
+        return jobs_.size();
+      } else {
+        return jobsBuilder_.getCount();
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.JobDeviceFilters jobs = 1;</code>
+     */
+    public org.tensorflow.proto.distruntime.JobDeviceFilters getJobs(int index) {
+      if (jobsBuilder_ == null) {
+        return jobs_.get(index);
+      } else {
+        return jobsBuilder_.getMessage(index);
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.JobDeviceFilters jobs = 1;</code>
+     */
+    public Builder setJobs(
+        int index, org.tensorflow.proto.distruntime.JobDeviceFilters value) {
+      if (jobsBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureJobsIsMutable();
+        jobs_.set(index, value);
+        onChanged();
+      } else {
+        jobsBuilder_.setMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.JobDeviceFilters jobs = 1;</code>
+     */
+    public Builder setJobs(
+        int index, org.tensorflow.proto.distruntime.JobDeviceFilters.Builder builderForValue) {
+      if (jobsBuilder_ == null) {
+        ensureJobsIsMutable();
+        jobs_.set(index, builderForValue.build());
+        onChanged();
+      } else {
+        jobsBuilder_.setMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.JobDeviceFilters jobs = 1;</code>
+     */
+    public Builder addJobs(org.tensorflow.proto.distruntime.JobDeviceFilters value) {
+      if (jobsBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureJobsIsMutable();
+        jobs_.add(value);
+        onChanged();
+      } else {
+        jobsBuilder_.addMessage(value);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.JobDeviceFilters jobs = 1;</code>
+     */
+    public Builder addJobs(
+        int index, org.tensorflow.proto.distruntime.JobDeviceFilters value) {
+      if (jobsBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureJobsIsMutable();
+        jobs_.add(index, value);
+        onChanged();
+      } else {
+        jobsBuilder_.addMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.JobDeviceFilters jobs = 1;</code>
+     */
+    public Builder addJobs(
+        org.tensorflow.proto.distruntime.JobDeviceFilters.Builder builderForValue) {
+      if (jobsBuilder_ == null) {
+        ensureJobsIsMutable();
+        jobs_.add(builderForValue.build());
+        onChanged();
+      } else {
+        jobsBuilder_.addMessage(builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.JobDeviceFilters jobs = 1;</code>
+     */
+    public Builder addJobs(
+        int index, org.tensorflow.proto.distruntime.JobDeviceFilters.Builder builderForValue) {
+      if (jobsBuilder_ == null) {
+        ensureJobsIsMutable();
+        jobs_.add(index, builderForValue.build());
+        onChanged();
+      } else {
+        jobsBuilder_.addMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.JobDeviceFilters jobs = 1;</code>
+     */
+    public Builder addAllJobs(
+        java.lang.Iterable<? extends org.tensorflow.proto.distruntime.JobDeviceFilters> values) {
+      if (jobsBuilder_ == null) {
+        ensureJobsIsMutable();
+        com.google.protobuf.AbstractMessageLite.Builder.addAll(
+            values, jobs_);
+        onChanged();
+      } else {
+        jobsBuilder_.addAllMessages(values);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.JobDeviceFilters jobs = 1;</code>
+     */
+    public Builder clearJobs() {
+      if (jobsBuilder_ == null) {
+        jobs_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000001);
+        onChanged();
+      } else {
+        jobsBuilder_.clear();
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.JobDeviceFilters jobs = 1;</code>
+     */
+    public Builder removeJobs(int index) {
+      if (jobsBuilder_ == null) {
+        ensureJobsIsMutable();
+        jobs_.remove(index);
+        onChanged();
+      } else {
+        jobsBuilder_.remove(index);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.JobDeviceFilters jobs = 1;</code>
+     */
+    public org.tensorflow.proto.distruntime.JobDeviceFilters.Builder getJobsBuilder(
+        int index) {
+      return getJobsFieldBuilder().getBuilder(index);
+    }
+    /**
+     * <code>repeated .tensorflow.JobDeviceFilters jobs = 1;</code>
+     */
+    public org.tensorflow.proto.distruntime.JobDeviceFiltersOrBuilder getJobsOrBuilder(
+        int index) {
+      if (jobsBuilder_ == null) {
+        return jobs_.get(index);  } else {
+        return jobsBuilder_.getMessageOrBuilder(index);
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.JobDeviceFilters jobs = 1;</code>
+     */
+    public java.util.List<? extends org.tensorflow.proto.distruntime.JobDeviceFiltersOrBuilder> 
+         getJobsOrBuilderList() {
+      if (jobsBuilder_ != null) {
+        return jobsBuilder_.getMessageOrBuilderList();
+      } else {
+        return java.util.Collections.unmodifiableList(jobs_);
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.JobDeviceFilters jobs = 1;</code>
+     */
+    public org.tensorflow.proto.distruntime.JobDeviceFilters.Builder addJobsBuilder() {
+      return getJobsFieldBuilder().addBuilder(
+          org.tensorflow.proto.distruntime.JobDeviceFilters.getDefaultInstance());
+    }
+    /**
+     * <code>repeated .tensorflow.JobDeviceFilters jobs = 1;</code>
+     */
+    public org.tensorflow.proto.distruntime.JobDeviceFilters.Builder addJobsBuilder(
+        int index) {
+      return getJobsFieldBuilder().addBuilder(
+          index, org.tensorflow.proto.distruntime.JobDeviceFilters.getDefaultInstance());
+    }
+    /**
+     * <code>repeated .tensorflow.JobDeviceFilters jobs = 1;</code>
+     */
+    public java.util.List<org.tensorflow.proto.distruntime.JobDeviceFilters.Builder> 
+         getJobsBuilderList() {
+      return getJobsFieldBuilder().getBuilderList();
+    }
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.distruntime.JobDeviceFilters, org.tensorflow.proto.distruntime.JobDeviceFilters.Builder, org.tensorflow.proto.distruntime.JobDeviceFiltersOrBuilder> 
+        getJobsFieldBuilder() {
+      if (jobsBuilder_ == null) {
+        jobsBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
+            org.tensorflow.proto.distruntime.JobDeviceFilters, org.tensorflow.proto.distruntime.JobDeviceFilters.Builder, org.tensorflow.proto.distruntime.JobDeviceFiltersOrBuilder>(
+                jobs_,
+                ((bitField0_ & 0x00000001) != 0),
+                getParentForChildren(),
+                isClean());
+        jobs_ = null;
+      }
+      return jobsBuilder_;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.ClusterDeviceFilters)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.ClusterDeviceFilters)
+  private static final org.tensorflow.proto.distruntime.ClusterDeviceFilters DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.distruntime.ClusterDeviceFilters();
+  }
+
+  public static org.tensorflow.proto.distruntime.ClusterDeviceFilters getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<ClusterDeviceFilters>
+      PARSER = new com.google.protobuf.AbstractParser<ClusterDeviceFilters>() {
+    @java.lang.Override
+    public ClusterDeviceFilters parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new ClusterDeviceFilters(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<ClusterDeviceFilters> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<ClusterDeviceFilters> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.distruntime.ClusterDeviceFilters getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/distruntime/ClusterDeviceFiltersOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/distruntime/ClusterDeviceFiltersOrBuilder.java
new file mode 100644
index 00000000000..eaaf1635eec
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/distruntime/ClusterDeviceFiltersOrBuilder.java
@@ -0,0 +1,33 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/device_filters.proto
+
+package org.tensorflow.proto.distruntime;
+
+public interface ClusterDeviceFiltersOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.ClusterDeviceFilters)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <code>repeated .tensorflow.JobDeviceFilters jobs = 1;</code>
+   */
+  java.util.List<org.tensorflow.proto.distruntime.JobDeviceFilters> 
+      getJobsList();
+  /**
+   * <code>repeated .tensorflow.JobDeviceFilters jobs = 1;</code>
+   */
+  org.tensorflow.proto.distruntime.JobDeviceFilters getJobs(int index);
+  /**
+   * <code>repeated .tensorflow.JobDeviceFilters jobs = 1;</code>
+   */
+  int getJobsCount();
+  /**
+   * <code>repeated .tensorflow.JobDeviceFilters jobs = 1;</code>
+   */
+  java.util.List<? extends org.tensorflow.proto.distruntime.JobDeviceFiltersOrBuilder> 
+      getJobsOrBuilderList();
+  /**
+   * <code>repeated .tensorflow.JobDeviceFilters jobs = 1;</code>
+   */
+  org.tensorflow.proto.distruntime.JobDeviceFiltersOrBuilder getJobsOrBuilder(
+      int index);
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/distruntime/ClusterProtos.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/distruntime/ClusterProtos.java
new file mode 100644
index 00000000000..06aa24f0049
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/distruntime/ClusterProtos.java
@@ -0,0 +1,76 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/cluster.proto
+
+package org.tensorflow.proto.distruntime;
+
+public final class ClusterProtos {
+  private ClusterProtos() {}
+  public static void registerAllExtensions(
+      com.google.protobuf.ExtensionRegistryLite registry) {
+  }
+
+  public static void registerAllExtensions(
+      com.google.protobuf.ExtensionRegistry registry) {
+    registerAllExtensions(
+        (com.google.protobuf.ExtensionRegistryLite) registry);
+  }
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_JobDef_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_JobDef_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_JobDef_TasksEntry_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_JobDef_TasksEntry_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_ClusterDef_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_ClusterDef_fieldAccessorTable;
+
+  public static com.google.protobuf.Descriptors.FileDescriptor
+      getDescriptor() {
+    return descriptor;
+  }
+  private static  com.google.protobuf.Descriptors.FileDescriptor
+      descriptor;
+  static {
+    java.lang.String[] descriptorData = {
+      "\n&tensorflow/core/protobuf/cluster.proto" +
+      "\022\ntensorflow\"r\n\006JobDef\022\014\n\004name\030\001 \001(\t\022,\n\005" +
+      "tasks\030\002 \003(\0132\035.tensorflow.JobDef.TasksEnt" +
+      "ry\032,\n\nTasksEntry\022\013\n\003key\030\001 \001(\005\022\r\n\005value\030\002" +
+      " \001(\t:\0028\001\"-\n\nClusterDef\022\037\n\003job\030\001 \003(\0132\022.te" +
+      "nsorflow.JobDefB\200\001\n org.tensorflow.proto" +
+      ".distruntimeB\rClusterProtosP\001ZHgithub.co" +
+      "m/tensorflow/tensorflow/tensorflow/go/co" +
+      "re/core_protos_go_proto\370\001\001b\006proto3"
+    };
+    descriptor = com.google.protobuf.Descriptors.FileDescriptor
+      .internalBuildGeneratedFileFrom(descriptorData,
+        new com.google.protobuf.Descriptors.FileDescriptor[] {
+        });
+    internal_static_tensorflow_JobDef_descriptor =
+      getDescriptor().getMessageTypes().get(0);
+    internal_static_tensorflow_JobDef_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_JobDef_descriptor,
+        new java.lang.String[] { "Name", "Tasks", });
+    internal_static_tensorflow_JobDef_TasksEntry_descriptor =
+      internal_static_tensorflow_JobDef_descriptor.getNestedTypes().get(0);
+    internal_static_tensorflow_JobDef_TasksEntry_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_JobDef_TasksEntry_descriptor,
+        new java.lang.String[] { "Key", "Value", });
+    internal_static_tensorflow_ClusterDef_descriptor =
+      getDescriptor().getMessageTypes().get(1);
+    internal_static_tensorflow_ClusterDef_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_ClusterDef_descriptor,
+        new java.lang.String[] { "Job", });
+  }
+
+  // @@protoc_insertion_point(outer_class_scope)
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/distruntime/DeviceFiltersProtos.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/distruntime/DeviceFiltersProtos.java
new file mode 100644
index 00000000000..254d82a2a2a
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/distruntime/DeviceFiltersProtos.java
@@ -0,0 +1,91 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/device_filters.proto
+
+package org.tensorflow.proto.distruntime;
+
+public final class DeviceFiltersProtos {
+  private DeviceFiltersProtos() {}
+  public static void registerAllExtensions(
+      com.google.protobuf.ExtensionRegistryLite registry) {
+  }
+
+  public static void registerAllExtensions(
+      com.google.protobuf.ExtensionRegistry registry) {
+    registerAllExtensions(
+        (com.google.protobuf.ExtensionRegistryLite) registry);
+  }
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_TaskDeviceFilters_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_TaskDeviceFilters_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_JobDeviceFilters_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_JobDeviceFilters_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_JobDeviceFilters_TasksEntry_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_JobDeviceFilters_TasksEntry_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_ClusterDeviceFilters_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_ClusterDeviceFilters_fieldAccessorTable;
+
+  public static com.google.protobuf.Descriptors.FileDescriptor
+      getDescriptor() {
+    return descriptor;
+  }
+  private static  com.google.protobuf.Descriptors.FileDescriptor
+      descriptor;
+  static {
+    java.lang.String[] descriptorData = {
+      "\n-tensorflow/core/protobuf/device_filter" +
+      "s.proto\022\ntensorflow\"+\n\021TaskDeviceFilters" +
+      "\022\026\n\016device_filters\030\001 \003(\t\"\245\001\n\020JobDeviceFi" +
+      "lters\022\014\n\004name\030\001 \001(\t\0226\n\005tasks\030\002 \003(\0132\'.ten" +
+      "sorflow.JobDeviceFilters.TasksEntry\032K\n\nT" +
+      "asksEntry\022\013\n\003key\030\001 \001(\005\022,\n\005value\030\002 \001(\0132\035." +
+      "tensorflow.TaskDeviceFilters:\0028\001\"B\n\024Clus" +
+      "terDeviceFilters\022*\n\004jobs\030\001 \003(\0132\034.tensorf" +
+      "low.JobDeviceFiltersB\206\001\n org.tensorflow." +
+      "proto.distruntimeB\023DeviceFiltersProtosP\001" +
+      "ZHgithub.com/tensorflow/tensorflow/tenso" +
+      "rflow/go/core/core_protos_go_proto\370\001\001b\006p" +
+      "roto3"
+    };
+    descriptor = com.google.protobuf.Descriptors.FileDescriptor
+      .internalBuildGeneratedFileFrom(descriptorData,
+        new com.google.protobuf.Descriptors.FileDescriptor[] {
+        });
+    internal_static_tensorflow_TaskDeviceFilters_descriptor =
+      getDescriptor().getMessageTypes().get(0);
+    internal_static_tensorflow_TaskDeviceFilters_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_TaskDeviceFilters_descriptor,
+        new java.lang.String[] { "DeviceFilters", });
+    internal_static_tensorflow_JobDeviceFilters_descriptor =
+      getDescriptor().getMessageTypes().get(1);
+    internal_static_tensorflow_JobDeviceFilters_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_JobDeviceFilters_descriptor,
+        new java.lang.String[] { "Name", "Tasks", });
+    internal_static_tensorflow_JobDeviceFilters_TasksEntry_descriptor =
+      internal_static_tensorflow_JobDeviceFilters_descriptor.getNestedTypes().get(0);
+    internal_static_tensorflow_JobDeviceFilters_TasksEntry_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_JobDeviceFilters_TasksEntry_descriptor,
+        new java.lang.String[] { "Key", "Value", });
+    internal_static_tensorflow_ClusterDeviceFilters_descriptor =
+      getDescriptor().getMessageTypes().get(2);
+    internal_static_tensorflow_ClusterDeviceFilters_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_ClusterDeviceFilters_descriptor,
+        new java.lang.String[] { "Jobs", });
+  }
+
+  // @@protoc_insertion_point(outer_class_scope)
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/distruntime/JobDef.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/distruntime/JobDef.java
new file mode 100644
index 00000000000..c6766412507
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/distruntime/JobDef.java
@@ -0,0 +1,935 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/cluster.proto
+
+package org.tensorflow.proto.distruntime;
+
+/**
+ * <pre>
+ * Defines a single job in a TensorFlow cluster.
+ * </pre>
+ *
+ * Protobuf type {@code tensorflow.JobDef}
+ */
+public  final class JobDef extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.JobDef)
+    JobDefOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use JobDef.newBuilder() to construct.
+  private JobDef(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private JobDef() {
+    name_ = "";
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new JobDef();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private JobDef(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    int mutable_bitField0_ = 0;
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 10: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            name_ = s;
+            break;
+          }
+          case 18: {
+            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
+              tasks_ = com.google.protobuf.MapField.newMapField(
+                  TasksDefaultEntryHolder.defaultEntry);
+              mutable_bitField0_ |= 0x00000001;
+            }
+            com.google.protobuf.MapEntry<java.lang.Integer, java.lang.String>
+            tasks__ = input.readMessage(
+                TasksDefaultEntryHolder.defaultEntry.getParserForType(), extensionRegistry);
+            tasks_.getMutableMap().put(
+                tasks__.getKey(), tasks__.getValue());
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.distruntime.ClusterProtos.internal_static_tensorflow_JobDef_descriptor;
+  }
+
+  @SuppressWarnings({"rawtypes"})
+  @java.lang.Override
+  protected com.google.protobuf.MapField internalGetMapField(
+      int number) {
+    switch (number) {
+      case 2:
+        return internalGetTasks();
+      default:
+        throw new RuntimeException(
+            "Invalid map field number: " + number);
+    }
+  }
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.distruntime.ClusterProtos.internal_static_tensorflow_JobDef_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.distruntime.JobDef.class, org.tensorflow.proto.distruntime.JobDef.Builder.class);
+  }
+
+  public static final int NAME_FIELD_NUMBER = 1;
+  private volatile java.lang.Object name_;
+  /**
+   * <pre>
+   * The name of this job.
+   * </pre>
+   *
+   * <code>string name = 1;</code>
+   */
+  public java.lang.String getName() {
+    java.lang.Object ref = name_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      name_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * The name of this job.
+   * </pre>
+   *
+   * <code>string name = 1;</code>
+   */
+  public com.google.protobuf.ByteString
+      getNameBytes() {
+    java.lang.Object ref = name_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      name_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int TASKS_FIELD_NUMBER = 2;
+  private static final class TasksDefaultEntryHolder {
+    static final com.google.protobuf.MapEntry<
+        java.lang.Integer, java.lang.String> defaultEntry =
+            com.google.protobuf.MapEntry
+            .<java.lang.Integer, java.lang.String>newDefaultInstance(
+                org.tensorflow.proto.distruntime.ClusterProtos.internal_static_tensorflow_JobDef_TasksEntry_descriptor, 
+                com.google.protobuf.WireFormat.FieldType.INT32,
+                0,
+                com.google.protobuf.WireFormat.FieldType.STRING,
+                "");
+  }
+  private com.google.protobuf.MapField<
+      java.lang.Integer, java.lang.String> tasks_;
+  private com.google.protobuf.MapField<java.lang.Integer, java.lang.String>
+  internalGetTasks() {
+    if (tasks_ == null) {
+      return com.google.protobuf.MapField.emptyMapField(
+          TasksDefaultEntryHolder.defaultEntry);
+    }
+    return tasks_;
+  }
+
+  public int getTasksCount() {
+    return internalGetTasks().getMap().size();
+  }
+  /**
+   * <pre>
+   * Mapping from task ID to "hostname:port" string.
+   * If the `name` field contains "worker", and the `tasks` map contains a
+   * mapping from 7 to "example.org:2222", then the device prefix
+   * "/job:worker/task:7" will be assigned to "example.org:2222".
+   * </pre>
+   *
+   * <code>map&lt;int32, string&gt; tasks = 2;</code>
+   */
+
+  public boolean containsTasks(
+      int key) {
+    
+    return internalGetTasks().getMap().containsKey(key);
+  }
+  /**
+   * Use {@link #getTasksMap()} instead.
+   */
+  @java.lang.Deprecated
+  public java.util.Map<java.lang.Integer, java.lang.String> getTasks() {
+    return getTasksMap();
+  }
+  /**
+   * <pre>
+   * Mapping from task ID to "hostname:port" string.
+   * If the `name` field contains "worker", and the `tasks` map contains a
+   * mapping from 7 to "example.org:2222", then the device prefix
+   * "/job:worker/task:7" will be assigned to "example.org:2222".
+   * </pre>
+   *
+   * <code>map&lt;int32, string&gt; tasks = 2;</code>
+   */
+
+  public java.util.Map<java.lang.Integer, java.lang.String> getTasksMap() {
+    return internalGetTasks().getMap();
+  }
+  /**
+   * <pre>
+   * Mapping from task ID to "hostname:port" string.
+   * If the `name` field contains "worker", and the `tasks` map contains a
+   * mapping from 7 to "example.org:2222", then the device prefix
+   * "/job:worker/task:7" will be assigned to "example.org:2222".
+   * </pre>
+   *
+   * <code>map&lt;int32, string&gt; tasks = 2;</code>
+   */
+
+  public java.lang.String getTasksOrDefault(
+      int key,
+      java.lang.String defaultValue) {
+    
+    java.util.Map<java.lang.Integer, java.lang.String> map =
+        internalGetTasks().getMap();
+    return map.containsKey(key) ? map.get(key) : defaultValue;
+  }
+  /**
+   * <pre>
+   * Mapping from task ID to "hostname:port" string.
+   * If the `name` field contains "worker", and the `tasks` map contains a
+   * mapping from 7 to "example.org:2222", then the device prefix
+   * "/job:worker/task:7" will be assigned to "example.org:2222".
+   * </pre>
+   *
+   * <code>map&lt;int32, string&gt; tasks = 2;</code>
+   */
+
+  public java.lang.String getTasksOrThrow(
+      int key) {
+    
+    java.util.Map<java.lang.Integer, java.lang.String> map =
+        internalGetTasks().getMap();
+    if (!map.containsKey(key)) {
+      throw new java.lang.IllegalArgumentException();
+    }
+    return map.get(key);
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    if (!getNameBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 1, name_);
+    }
+    com.google.protobuf.GeneratedMessageV3
+      .serializeIntegerMapTo(
+        output,
+        internalGetTasks(),
+        TasksDefaultEntryHolder.defaultEntry,
+        2);
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (!getNameBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, name_);
+    }
+    for (java.util.Map.Entry<java.lang.Integer, java.lang.String> entry
+         : internalGetTasks().getMap().entrySet()) {
+      com.google.protobuf.MapEntry<java.lang.Integer, java.lang.String>
+      tasks__ = TasksDefaultEntryHolder.defaultEntry.newBuilderForType()
+          .setKey(entry.getKey())
+          .setValue(entry.getValue())
+          .build();
+      size += com.google.protobuf.CodedOutputStream
+          .computeMessageSize(2, tasks__);
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.distruntime.JobDef)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.distruntime.JobDef other = (org.tensorflow.proto.distruntime.JobDef) obj;
+
+    if (!getName()
+        .equals(other.getName())) return false;
+    if (!internalGetTasks().equals(
+        other.internalGetTasks())) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    hash = (37 * hash) + NAME_FIELD_NUMBER;
+    hash = (53 * hash) + getName().hashCode();
+    if (!internalGetTasks().getMap().isEmpty()) {
+      hash = (37 * hash) + TASKS_FIELD_NUMBER;
+      hash = (53 * hash) + internalGetTasks().hashCode();
+    }
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.distruntime.JobDef parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.distruntime.JobDef parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.distruntime.JobDef parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.distruntime.JobDef parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.distruntime.JobDef parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.distruntime.JobDef parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.distruntime.JobDef parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.distruntime.JobDef parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.distruntime.JobDef parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.distruntime.JobDef parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.distruntime.JobDef parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.distruntime.JobDef parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.distruntime.JobDef prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * <pre>
+   * Defines a single job in a TensorFlow cluster.
+   * </pre>
+   *
+   * Protobuf type {@code tensorflow.JobDef}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.JobDef)
+      org.tensorflow.proto.distruntime.JobDefOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.distruntime.ClusterProtos.internal_static_tensorflow_JobDef_descriptor;
+    }
+
+    @SuppressWarnings({"rawtypes"})
+    protected com.google.protobuf.MapField internalGetMapField(
+        int number) {
+      switch (number) {
+        case 2:
+          return internalGetTasks();
+        default:
+          throw new RuntimeException(
+              "Invalid map field number: " + number);
+      }
+    }
+    @SuppressWarnings({"rawtypes"})
+    protected com.google.protobuf.MapField internalGetMutableMapField(
+        int number) {
+      switch (number) {
+        case 2:
+          return internalGetMutableTasks();
+        default:
+          throw new RuntimeException(
+              "Invalid map field number: " + number);
+      }
+    }
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.distruntime.ClusterProtos.internal_static_tensorflow_JobDef_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.distruntime.JobDef.class, org.tensorflow.proto.distruntime.JobDef.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.distruntime.JobDef.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      name_ = "";
+
+      internalGetMutableTasks().clear();
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.distruntime.ClusterProtos.internal_static_tensorflow_JobDef_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.distruntime.JobDef getDefaultInstanceForType() {
+      return org.tensorflow.proto.distruntime.JobDef.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.distruntime.JobDef build() {
+      org.tensorflow.proto.distruntime.JobDef result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.distruntime.JobDef buildPartial() {
+      org.tensorflow.proto.distruntime.JobDef result = new org.tensorflow.proto.distruntime.JobDef(this);
+      int from_bitField0_ = bitField0_;
+      result.name_ = name_;
+      result.tasks_ = internalGetTasks();
+      result.tasks_.makeImmutable();
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.distruntime.JobDef) {
+        return mergeFrom((org.tensorflow.proto.distruntime.JobDef)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.distruntime.JobDef other) {
+      if (other == org.tensorflow.proto.distruntime.JobDef.getDefaultInstance()) return this;
+      if (!other.getName().isEmpty()) {
+        name_ = other.name_;
+        onChanged();
+      }
+      internalGetMutableTasks().mergeFrom(
+          other.internalGetTasks());
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.distruntime.JobDef parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.distruntime.JobDef) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+    private int bitField0_;
+
+    private java.lang.Object name_ = "";
+    /**
+     * <pre>
+     * The name of this job.
+     * </pre>
+     *
+     * <code>string name = 1;</code>
+     */
+    public java.lang.String getName() {
+      java.lang.Object ref = name_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        name_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * The name of this job.
+     * </pre>
+     *
+     * <code>string name = 1;</code>
+     */
+    public com.google.protobuf.ByteString
+        getNameBytes() {
+      java.lang.Object ref = name_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        name_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * The name of this job.
+     * </pre>
+     *
+     * <code>string name = 1;</code>
+     */
+    public Builder setName(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      name_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * The name of this job.
+     * </pre>
+     *
+     * <code>string name = 1;</code>
+     */
+    public Builder clearName() {
+      
+      name_ = getDefaultInstance().getName();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * The name of this job.
+     * </pre>
+     *
+     * <code>string name = 1;</code>
+     */
+    public Builder setNameBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      name_ = value;
+      onChanged();
+      return this;
+    }
+
+    private com.google.protobuf.MapField<
+        java.lang.Integer, java.lang.String> tasks_;
+    private com.google.protobuf.MapField<java.lang.Integer, java.lang.String>
+    internalGetTasks() {
+      if (tasks_ == null) {
+        return com.google.protobuf.MapField.emptyMapField(
+            TasksDefaultEntryHolder.defaultEntry);
+      }
+      return tasks_;
+    }
+    private com.google.protobuf.MapField<java.lang.Integer, java.lang.String>
+    internalGetMutableTasks() {
+      onChanged();;
+      if (tasks_ == null) {
+        tasks_ = com.google.protobuf.MapField.newMapField(
+            TasksDefaultEntryHolder.defaultEntry);
+      }
+      if (!tasks_.isMutable()) {
+        tasks_ = tasks_.copy();
+      }
+      return tasks_;
+    }
+
+    public int getTasksCount() {
+      return internalGetTasks().getMap().size();
+    }
+    /**
+     * <pre>
+     * Mapping from task ID to "hostname:port" string.
+     * If the `name` field contains "worker", and the `tasks` map contains a
+     * mapping from 7 to "example.org:2222", then the device prefix
+     * "/job:worker/task:7" will be assigned to "example.org:2222".
+     * </pre>
+     *
+     * <code>map&lt;int32, string&gt; tasks = 2;</code>
+     */
+
+    public boolean containsTasks(
+        int key) {
+      
+      return internalGetTasks().getMap().containsKey(key);
+    }
+    /**
+     * Use {@link #getTasksMap()} instead.
+     */
+    @java.lang.Deprecated
+    public java.util.Map<java.lang.Integer, java.lang.String> getTasks() {
+      return getTasksMap();
+    }
+    /**
+     * <pre>
+     * Mapping from task ID to "hostname:port" string.
+     * If the `name` field contains "worker", and the `tasks` map contains a
+     * mapping from 7 to "example.org:2222", then the device prefix
+     * "/job:worker/task:7" will be assigned to "example.org:2222".
+     * </pre>
+     *
+     * <code>map&lt;int32, string&gt; tasks = 2;</code>
+     */
+
+    public java.util.Map<java.lang.Integer, java.lang.String> getTasksMap() {
+      return internalGetTasks().getMap();
+    }
+    /**
+     * <pre>
+     * Mapping from task ID to "hostname:port" string.
+     * If the `name` field contains "worker", and the `tasks` map contains a
+     * mapping from 7 to "example.org:2222", then the device prefix
+     * "/job:worker/task:7" will be assigned to "example.org:2222".
+     * </pre>
+     *
+     * <code>map&lt;int32, string&gt; tasks = 2;</code>
+     */
+
+    public java.lang.String getTasksOrDefault(
+        int key,
+        java.lang.String defaultValue) {
+      
+      java.util.Map<java.lang.Integer, java.lang.String> map =
+          internalGetTasks().getMap();
+      return map.containsKey(key) ? map.get(key) : defaultValue;
+    }
+    /**
+     * <pre>
+     * Mapping from task ID to "hostname:port" string.
+     * If the `name` field contains "worker", and the `tasks` map contains a
+     * mapping from 7 to "example.org:2222", then the device prefix
+     * "/job:worker/task:7" will be assigned to "example.org:2222".
+     * </pre>
+     *
+     * <code>map&lt;int32, string&gt; tasks = 2;</code>
+     */
+
+    public java.lang.String getTasksOrThrow(
+        int key) {
+      
+      java.util.Map<java.lang.Integer, java.lang.String> map =
+          internalGetTasks().getMap();
+      if (!map.containsKey(key)) {
+        throw new java.lang.IllegalArgumentException();
+      }
+      return map.get(key);
+    }
+
+    public Builder clearTasks() {
+      internalGetMutableTasks().getMutableMap()
+          .clear();
+      return this;
+    }
+    /**
+     * <pre>
+     * Mapping from task ID to "hostname:port" string.
+     * If the `name` field contains "worker", and the `tasks` map contains a
+     * mapping from 7 to "example.org:2222", then the device prefix
+     * "/job:worker/task:7" will be assigned to "example.org:2222".
+     * </pre>
+     *
+     * <code>map&lt;int32, string&gt; tasks = 2;</code>
+     */
+
+    public Builder removeTasks(
+        int key) {
+      
+      internalGetMutableTasks().getMutableMap()
+          .remove(key);
+      return this;
+    }
+    /**
+     * Use alternate mutation accessors instead.
+     */
+    @java.lang.Deprecated
+    public java.util.Map<java.lang.Integer, java.lang.String>
+    getMutableTasks() {
+      return internalGetMutableTasks().getMutableMap();
+    }
+    /**
+     * <pre>
+     * Mapping from task ID to "hostname:port" string.
+     * If the `name` field contains "worker", and the `tasks` map contains a
+     * mapping from 7 to "example.org:2222", then the device prefix
+     * "/job:worker/task:7" will be assigned to "example.org:2222".
+     * </pre>
+     *
+     * <code>map&lt;int32, string&gt; tasks = 2;</code>
+     */
+    public Builder putTasks(
+        int key,
+        java.lang.String value) {
+      
+      if (value == null) { throw new java.lang.NullPointerException(); }
+      internalGetMutableTasks().getMutableMap()
+          .put(key, value);
+      return this;
+    }
+    /**
+     * <pre>
+     * Mapping from task ID to "hostname:port" string.
+     * If the `name` field contains "worker", and the `tasks` map contains a
+     * mapping from 7 to "example.org:2222", then the device prefix
+     * "/job:worker/task:7" will be assigned to "example.org:2222".
+     * </pre>
+     *
+     * <code>map&lt;int32, string&gt; tasks = 2;</code>
+     */
+
+    public Builder putAllTasks(
+        java.util.Map<java.lang.Integer, java.lang.String> values) {
+      internalGetMutableTasks().getMutableMap()
+          .putAll(values);
+      return this;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.JobDef)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.JobDef)
+  private static final org.tensorflow.proto.distruntime.JobDef DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.distruntime.JobDef();
+  }
+
+  public static org.tensorflow.proto.distruntime.JobDef getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<JobDef>
+      PARSER = new com.google.protobuf.AbstractParser<JobDef>() {
+    @java.lang.Override
+    public JobDef parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new JobDef(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<JobDef> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<JobDef> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.distruntime.JobDef getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/distruntime/JobDefOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/distruntime/JobDefOrBuilder.java
new file mode 100644
index 00000000000..a9c24b98a74
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/distruntime/JobDefOrBuilder.java
@@ -0,0 +1,96 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/cluster.proto
+
+package org.tensorflow.proto.distruntime;
+
+public interface JobDefOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.JobDef)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <pre>
+   * The name of this job.
+   * </pre>
+   *
+   * <code>string name = 1;</code>
+   */
+  java.lang.String getName();
+  /**
+   * <pre>
+   * The name of this job.
+   * </pre>
+   *
+   * <code>string name = 1;</code>
+   */
+  com.google.protobuf.ByteString
+      getNameBytes();
+
+  /**
+   * <pre>
+   * Mapping from task ID to "hostname:port" string.
+   * If the `name` field contains "worker", and the `tasks` map contains a
+   * mapping from 7 to "example.org:2222", then the device prefix
+   * "/job:worker/task:7" will be assigned to "example.org:2222".
+   * </pre>
+   *
+   * <code>map&lt;int32, string&gt; tasks = 2;</code>
+   */
+  int getTasksCount();
+  /**
+   * <pre>
+   * Mapping from task ID to "hostname:port" string.
+   * If the `name` field contains "worker", and the `tasks` map contains a
+   * mapping from 7 to "example.org:2222", then the device prefix
+   * "/job:worker/task:7" will be assigned to "example.org:2222".
+   * </pre>
+   *
+   * <code>map&lt;int32, string&gt; tasks = 2;</code>
+   */
+  boolean containsTasks(
+      int key);
+  /**
+   * Use {@link #getTasksMap()} instead.
+   */
+  @java.lang.Deprecated
+  java.util.Map<java.lang.Integer, java.lang.String>
+  getTasks();
+  /**
+   * <pre>
+   * Mapping from task ID to "hostname:port" string.
+   * If the `name` field contains "worker", and the `tasks` map contains a
+   * mapping from 7 to "example.org:2222", then the device prefix
+   * "/job:worker/task:7" will be assigned to "example.org:2222".
+   * </pre>
+   *
+   * <code>map&lt;int32, string&gt; tasks = 2;</code>
+   */
+  java.util.Map<java.lang.Integer, java.lang.String>
+  getTasksMap();
+  /**
+   * <pre>
+   * Mapping from task ID to "hostname:port" string.
+   * If the `name` field contains "worker", and the `tasks` map contains a
+   * mapping from 7 to "example.org:2222", then the device prefix
+   * "/job:worker/task:7" will be assigned to "example.org:2222".
+   * </pre>
+   *
+   * <code>map&lt;int32, string&gt; tasks = 2;</code>
+   */
+
+  java.lang.String getTasksOrDefault(
+      int key,
+      java.lang.String defaultValue);
+  /**
+   * <pre>
+   * Mapping from task ID to "hostname:port" string.
+   * If the `name` field contains "worker", and the `tasks` map contains a
+   * mapping from 7 to "example.org:2222", then the device prefix
+   * "/job:worker/task:7" will be assigned to "example.org:2222".
+   * </pre>
+   *
+   * <code>map&lt;int32, string&gt; tasks = 2;</code>
+   */
+
+  java.lang.String getTasksOrThrow(
+      int key);
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/distruntime/JobDeviceFilters.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/distruntime/JobDeviceFilters.java
new file mode 100644
index 00000000000..bbc7a6542be
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/distruntime/JobDeviceFilters.java
@@ -0,0 +1,902 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/device_filters.proto
+
+package org.tensorflow.proto.distruntime;
+
+/**
+ * <pre>
+ * Defines the device filters for tasks in a job.
+ * </pre>
+ *
+ * Protobuf type {@code tensorflow.JobDeviceFilters}
+ */
+public  final class JobDeviceFilters extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.JobDeviceFilters)
+    JobDeviceFiltersOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use JobDeviceFilters.newBuilder() to construct.
+  private JobDeviceFilters(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private JobDeviceFilters() {
+    name_ = "";
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new JobDeviceFilters();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private JobDeviceFilters(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    int mutable_bitField0_ = 0;
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 10: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            name_ = s;
+            break;
+          }
+          case 18: {
+            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
+              tasks_ = com.google.protobuf.MapField.newMapField(
+                  TasksDefaultEntryHolder.defaultEntry);
+              mutable_bitField0_ |= 0x00000001;
+            }
+            com.google.protobuf.MapEntry<java.lang.Integer, org.tensorflow.proto.distruntime.TaskDeviceFilters>
+            tasks__ = input.readMessage(
+                TasksDefaultEntryHolder.defaultEntry.getParserForType(), extensionRegistry);
+            tasks_.getMutableMap().put(
+                tasks__.getKey(), tasks__.getValue());
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.distruntime.DeviceFiltersProtos.internal_static_tensorflow_JobDeviceFilters_descriptor;
+  }
+
+  @SuppressWarnings({"rawtypes"})
+  @java.lang.Override
+  protected com.google.protobuf.MapField internalGetMapField(
+      int number) {
+    switch (number) {
+      case 2:
+        return internalGetTasks();
+      default:
+        throw new RuntimeException(
+            "Invalid map field number: " + number);
+    }
+  }
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.distruntime.DeviceFiltersProtos.internal_static_tensorflow_JobDeviceFilters_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.distruntime.JobDeviceFilters.class, org.tensorflow.proto.distruntime.JobDeviceFilters.Builder.class);
+  }
+
+  public static final int NAME_FIELD_NUMBER = 1;
+  private volatile java.lang.Object name_;
+  /**
+   * <pre>
+   * The name of this job.
+   * </pre>
+   *
+   * <code>string name = 1;</code>
+   */
+  public java.lang.String getName() {
+    java.lang.Object ref = name_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      name_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * The name of this job.
+   * </pre>
+   *
+   * <code>string name = 1;</code>
+   */
+  public com.google.protobuf.ByteString
+      getNameBytes() {
+    java.lang.Object ref = name_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      name_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int TASKS_FIELD_NUMBER = 2;
+  private static final class TasksDefaultEntryHolder {
+    static final com.google.protobuf.MapEntry<
+        java.lang.Integer, org.tensorflow.proto.distruntime.TaskDeviceFilters> defaultEntry =
+            com.google.protobuf.MapEntry
+            .<java.lang.Integer, org.tensorflow.proto.distruntime.TaskDeviceFilters>newDefaultInstance(
+                org.tensorflow.proto.distruntime.DeviceFiltersProtos.internal_static_tensorflow_JobDeviceFilters_TasksEntry_descriptor, 
+                com.google.protobuf.WireFormat.FieldType.INT32,
+                0,
+                com.google.protobuf.WireFormat.FieldType.MESSAGE,
+                org.tensorflow.proto.distruntime.TaskDeviceFilters.getDefaultInstance());
+  }
+  private com.google.protobuf.MapField<
+      java.lang.Integer, org.tensorflow.proto.distruntime.TaskDeviceFilters> tasks_;
+  private com.google.protobuf.MapField<java.lang.Integer, org.tensorflow.proto.distruntime.TaskDeviceFilters>
+  internalGetTasks() {
+    if (tasks_ == null) {
+      return com.google.protobuf.MapField.emptyMapField(
+          TasksDefaultEntryHolder.defaultEntry);
+    }
+    return tasks_;
+  }
+
+  public int getTasksCount() {
+    return internalGetTasks().getMap().size();
+  }
+  /**
+   * <pre>
+   * Mapping from task ID to task device filters.
+   * </pre>
+   *
+   * <code>map&lt;int32, .tensorflow.TaskDeviceFilters&gt; tasks = 2;</code>
+   */
+
+  public boolean containsTasks(
+      int key) {
+    
+    return internalGetTasks().getMap().containsKey(key);
+  }
+  /**
+   * Use {@link #getTasksMap()} instead.
+   */
+  @java.lang.Deprecated
+  public java.util.Map<java.lang.Integer, org.tensorflow.proto.distruntime.TaskDeviceFilters> getTasks() {
+    return getTasksMap();
+  }
+  /**
+   * <pre>
+   * Mapping from task ID to task device filters.
+   * </pre>
+   *
+   * <code>map&lt;int32, .tensorflow.TaskDeviceFilters&gt; tasks = 2;</code>
+   */
+
+  public java.util.Map<java.lang.Integer, org.tensorflow.proto.distruntime.TaskDeviceFilters> getTasksMap() {
+    return internalGetTasks().getMap();
+  }
+  /**
+   * <pre>
+   * Mapping from task ID to task device filters.
+   * </pre>
+   *
+   * <code>map&lt;int32, .tensorflow.TaskDeviceFilters&gt; tasks = 2;</code>
+   */
+
+  public org.tensorflow.proto.distruntime.TaskDeviceFilters getTasksOrDefault(
+      int key,
+      org.tensorflow.proto.distruntime.TaskDeviceFilters defaultValue) {
+    
+    java.util.Map<java.lang.Integer, org.tensorflow.proto.distruntime.TaskDeviceFilters> map =
+        internalGetTasks().getMap();
+    return map.containsKey(key) ? map.get(key) : defaultValue;
+  }
+  /**
+   * <pre>
+   * Mapping from task ID to task device filters.
+   * </pre>
+   *
+   * <code>map&lt;int32, .tensorflow.TaskDeviceFilters&gt; tasks = 2;</code>
+   */
+
+  public org.tensorflow.proto.distruntime.TaskDeviceFilters getTasksOrThrow(
+      int key) {
+    
+    java.util.Map<java.lang.Integer, org.tensorflow.proto.distruntime.TaskDeviceFilters> map =
+        internalGetTasks().getMap();
+    if (!map.containsKey(key)) {
+      throw new java.lang.IllegalArgumentException();
+    }
+    return map.get(key);
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    if (!getNameBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 1, name_);
+    }
+    com.google.protobuf.GeneratedMessageV3
+      .serializeIntegerMapTo(
+        output,
+        internalGetTasks(),
+        TasksDefaultEntryHolder.defaultEntry,
+        2);
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (!getNameBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, name_);
+    }
+    for (java.util.Map.Entry<java.lang.Integer, org.tensorflow.proto.distruntime.TaskDeviceFilters> entry
+         : internalGetTasks().getMap().entrySet()) {
+      com.google.protobuf.MapEntry<java.lang.Integer, org.tensorflow.proto.distruntime.TaskDeviceFilters>
+      tasks__ = TasksDefaultEntryHolder.defaultEntry.newBuilderForType()
+          .setKey(entry.getKey())
+          .setValue(entry.getValue())
+          .build();
+      size += com.google.protobuf.CodedOutputStream
+          .computeMessageSize(2, tasks__);
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.distruntime.JobDeviceFilters)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.distruntime.JobDeviceFilters other = (org.tensorflow.proto.distruntime.JobDeviceFilters) obj;
+
+    if (!getName()
+        .equals(other.getName())) return false;
+    if (!internalGetTasks().equals(
+        other.internalGetTasks())) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    hash = (37 * hash) + NAME_FIELD_NUMBER;
+    hash = (53 * hash) + getName().hashCode();
+    if (!internalGetTasks().getMap().isEmpty()) {
+      hash = (37 * hash) + TASKS_FIELD_NUMBER;
+      hash = (53 * hash) + internalGetTasks().hashCode();
+    }
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.distruntime.JobDeviceFilters parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.distruntime.JobDeviceFilters parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.distruntime.JobDeviceFilters parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.distruntime.JobDeviceFilters parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.distruntime.JobDeviceFilters parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.distruntime.JobDeviceFilters parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.distruntime.JobDeviceFilters parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.distruntime.JobDeviceFilters parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.distruntime.JobDeviceFilters parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.distruntime.JobDeviceFilters parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.distruntime.JobDeviceFilters parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.distruntime.JobDeviceFilters parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.distruntime.JobDeviceFilters prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * <pre>
+   * Defines the device filters for tasks in a job.
+   * </pre>
+   *
+   * Protobuf type {@code tensorflow.JobDeviceFilters}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.JobDeviceFilters)
+      org.tensorflow.proto.distruntime.JobDeviceFiltersOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.distruntime.DeviceFiltersProtos.internal_static_tensorflow_JobDeviceFilters_descriptor;
+    }
+
+    @SuppressWarnings({"rawtypes"})
+    protected com.google.protobuf.MapField internalGetMapField(
+        int number) {
+      switch (number) {
+        case 2:
+          return internalGetTasks();
+        default:
+          throw new RuntimeException(
+              "Invalid map field number: " + number);
+      }
+    }
+    @SuppressWarnings({"rawtypes"})
+    protected com.google.protobuf.MapField internalGetMutableMapField(
+        int number) {
+      switch (number) {
+        case 2:
+          return internalGetMutableTasks();
+        default:
+          throw new RuntimeException(
+              "Invalid map field number: " + number);
+      }
+    }
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.distruntime.DeviceFiltersProtos.internal_static_tensorflow_JobDeviceFilters_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.distruntime.JobDeviceFilters.class, org.tensorflow.proto.distruntime.JobDeviceFilters.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.distruntime.JobDeviceFilters.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      name_ = "";
+
+      internalGetMutableTasks().clear();
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.distruntime.DeviceFiltersProtos.internal_static_tensorflow_JobDeviceFilters_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.distruntime.JobDeviceFilters getDefaultInstanceForType() {
+      return org.tensorflow.proto.distruntime.JobDeviceFilters.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.distruntime.JobDeviceFilters build() {
+      org.tensorflow.proto.distruntime.JobDeviceFilters result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.distruntime.JobDeviceFilters buildPartial() {
+      org.tensorflow.proto.distruntime.JobDeviceFilters result = new org.tensorflow.proto.distruntime.JobDeviceFilters(this);
+      int from_bitField0_ = bitField0_;
+      result.name_ = name_;
+      result.tasks_ = internalGetTasks();
+      result.tasks_.makeImmutable();
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.distruntime.JobDeviceFilters) {
+        return mergeFrom((org.tensorflow.proto.distruntime.JobDeviceFilters)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.distruntime.JobDeviceFilters other) {
+      if (other == org.tensorflow.proto.distruntime.JobDeviceFilters.getDefaultInstance()) return this;
+      if (!other.getName().isEmpty()) {
+        name_ = other.name_;
+        onChanged();
+      }
+      internalGetMutableTasks().mergeFrom(
+          other.internalGetTasks());
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.distruntime.JobDeviceFilters parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.distruntime.JobDeviceFilters) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+    private int bitField0_;
+
+    private java.lang.Object name_ = "";
+    /**
+     * <pre>
+     * The name of this job.
+     * </pre>
+     *
+     * <code>string name = 1;</code>
+     */
+    public java.lang.String getName() {
+      java.lang.Object ref = name_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        name_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * The name of this job.
+     * </pre>
+     *
+     * <code>string name = 1;</code>
+     */
+    public com.google.protobuf.ByteString
+        getNameBytes() {
+      java.lang.Object ref = name_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        name_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * The name of this job.
+     * </pre>
+     *
+     * <code>string name = 1;</code>
+     */
+    public Builder setName(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      name_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * The name of this job.
+     * </pre>
+     *
+     * <code>string name = 1;</code>
+     */
+    public Builder clearName() {
+      
+      name_ = getDefaultInstance().getName();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * The name of this job.
+     * </pre>
+     *
+     * <code>string name = 1;</code>
+     */
+    public Builder setNameBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      name_ = value;
+      onChanged();
+      return this;
+    }
+
+    private com.google.protobuf.MapField<
+        java.lang.Integer, org.tensorflow.proto.distruntime.TaskDeviceFilters> tasks_;
+    private com.google.protobuf.MapField<java.lang.Integer, org.tensorflow.proto.distruntime.TaskDeviceFilters>
+    internalGetTasks() {
+      if (tasks_ == null) {
+        return com.google.protobuf.MapField.emptyMapField(
+            TasksDefaultEntryHolder.defaultEntry);
+      }
+      return tasks_;
+    }
+    private com.google.protobuf.MapField<java.lang.Integer, org.tensorflow.proto.distruntime.TaskDeviceFilters>
+    internalGetMutableTasks() {
+      onChanged();;
+      if (tasks_ == null) {
+        tasks_ = com.google.protobuf.MapField.newMapField(
+            TasksDefaultEntryHolder.defaultEntry);
+      }
+      if (!tasks_.isMutable()) {
+        tasks_ = tasks_.copy();
+      }
+      return tasks_;
+    }
+
+    public int getTasksCount() {
+      return internalGetTasks().getMap().size();
+    }
+    /**
+     * <pre>
+     * Mapping from task ID to task device filters.
+     * </pre>
+     *
+     * <code>map&lt;int32, .tensorflow.TaskDeviceFilters&gt; tasks = 2;</code>
+     */
+
+    public boolean containsTasks(
+        int key) {
+      
+      return internalGetTasks().getMap().containsKey(key);
+    }
+    /**
+     * Use {@link #getTasksMap()} instead.
+     */
+    @java.lang.Deprecated
+    public java.util.Map<java.lang.Integer, org.tensorflow.proto.distruntime.TaskDeviceFilters> getTasks() {
+      return getTasksMap();
+    }
+    /**
+     * <pre>
+     * Mapping from task ID to task device filters.
+     * </pre>
+     *
+     * <code>map&lt;int32, .tensorflow.TaskDeviceFilters&gt; tasks = 2;</code>
+     */
+
+    public java.util.Map<java.lang.Integer, org.tensorflow.proto.distruntime.TaskDeviceFilters> getTasksMap() {
+      return internalGetTasks().getMap();
+    }
+    /**
+     * <pre>
+     * Mapping from task ID to task device filters.
+     * </pre>
+     *
+     * <code>map&lt;int32, .tensorflow.TaskDeviceFilters&gt; tasks = 2;</code>
+     */
+
+    public org.tensorflow.proto.distruntime.TaskDeviceFilters getTasksOrDefault(
+        int key,
+        org.tensorflow.proto.distruntime.TaskDeviceFilters defaultValue) {
+      
+      java.util.Map<java.lang.Integer, org.tensorflow.proto.distruntime.TaskDeviceFilters> map =
+          internalGetTasks().getMap();
+      return map.containsKey(key) ? map.get(key) : defaultValue;
+    }
+    /**
+     * <pre>
+     * Mapping from task ID to task device filters.
+     * </pre>
+     *
+     * <code>map&lt;int32, .tensorflow.TaskDeviceFilters&gt; tasks = 2;</code>
+     */
+
+    public org.tensorflow.proto.distruntime.TaskDeviceFilters getTasksOrThrow(
+        int key) {
+      
+      java.util.Map<java.lang.Integer, org.tensorflow.proto.distruntime.TaskDeviceFilters> map =
+          internalGetTasks().getMap();
+      if (!map.containsKey(key)) {
+        throw new java.lang.IllegalArgumentException();
+      }
+      return map.get(key);
+    }
+
+    public Builder clearTasks() {
+      internalGetMutableTasks().getMutableMap()
+          .clear();
+      return this;
+    }
+    /**
+     * <pre>
+     * Mapping from task ID to task device filters.
+     * </pre>
+     *
+     * <code>map&lt;int32, .tensorflow.TaskDeviceFilters&gt; tasks = 2;</code>
+     */
+
+    public Builder removeTasks(
+        int key) {
+      
+      internalGetMutableTasks().getMutableMap()
+          .remove(key);
+      return this;
+    }
+    /**
+     * Use alternate mutation accessors instead.
+     */
+    @java.lang.Deprecated
+    public java.util.Map<java.lang.Integer, org.tensorflow.proto.distruntime.TaskDeviceFilters>
+    getMutableTasks() {
+      return internalGetMutableTasks().getMutableMap();
+    }
+    /**
+     * <pre>
+     * Mapping from task ID to task device filters.
+     * </pre>
+     *
+     * <code>map&lt;int32, .tensorflow.TaskDeviceFilters&gt; tasks = 2;</code>
+     */
+    public Builder putTasks(
+        int key,
+        org.tensorflow.proto.distruntime.TaskDeviceFilters value) {
+      
+      if (value == null) { throw new java.lang.NullPointerException(); }
+      internalGetMutableTasks().getMutableMap()
+          .put(key, value);
+      return this;
+    }
+    /**
+     * <pre>
+     * Mapping from task ID to task device filters.
+     * </pre>
+     *
+     * <code>map&lt;int32, .tensorflow.TaskDeviceFilters&gt; tasks = 2;</code>
+     */
+
+    public Builder putAllTasks(
+        java.util.Map<java.lang.Integer, org.tensorflow.proto.distruntime.TaskDeviceFilters> values) {
+      internalGetMutableTasks().getMutableMap()
+          .putAll(values);
+      return this;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.JobDeviceFilters)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.JobDeviceFilters)
+  private static final org.tensorflow.proto.distruntime.JobDeviceFilters DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.distruntime.JobDeviceFilters();
+  }
+
+  public static org.tensorflow.proto.distruntime.JobDeviceFilters getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<JobDeviceFilters>
+      PARSER = new com.google.protobuf.AbstractParser<JobDeviceFilters>() {
+    @java.lang.Override
+    public JobDeviceFilters parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new JobDeviceFilters(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<JobDeviceFilters> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<JobDeviceFilters> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.distruntime.JobDeviceFilters getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/distruntime/JobDeviceFiltersOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/distruntime/JobDeviceFiltersOrBuilder.java
new file mode 100644
index 00000000000..92c3ba82935
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/distruntime/JobDeviceFiltersOrBuilder.java
@@ -0,0 +1,81 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/device_filters.proto
+
+package org.tensorflow.proto.distruntime;
+
+public interface JobDeviceFiltersOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.JobDeviceFilters)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <pre>
+   * The name of this job.
+   * </pre>
+   *
+   * <code>string name = 1;</code>
+   */
+  java.lang.String getName();
+  /**
+   * <pre>
+   * The name of this job.
+   * </pre>
+   *
+   * <code>string name = 1;</code>
+   */
+  com.google.protobuf.ByteString
+      getNameBytes();
+
+  /**
+   * <pre>
+   * Mapping from task ID to task device filters.
+   * </pre>
+   *
+   * <code>map&lt;int32, .tensorflow.TaskDeviceFilters&gt; tasks = 2;</code>
+   */
+  int getTasksCount();
+  /**
+   * <pre>
+   * Mapping from task ID to task device filters.
+   * </pre>
+   *
+   * <code>map&lt;int32, .tensorflow.TaskDeviceFilters&gt; tasks = 2;</code>
+   */
+  boolean containsTasks(
+      int key);
+  /**
+   * Use {@link #getTasksMap()} instead.
+   */
+  @java.lang.Deprecated
+  java.util.Map<java.lang.Integer, org.tensorflow.proto.distruntime.TaskDeviceFilters>
+  getTasks();
+  /**
+   * <pre>
+   * Mapping from task ID to task device filters.
+   * </pre>
+   *
+   * <code>map&lt;int32, .tensorflow.TaskDeviceFilters&gt; tasks = 2;</code>
+   */
+  java.util.Map<java.lang.Integer, org.tensorflow.proto.distruntime.TaskDeviceFilters>
+  getTasksMap();
+  /**
+   * <pre>
+   * Mapping from task ID to task device filters.
+   * </pre>
+   *
+   * <code>map&lt;int32, .tensorflow.TaskDeviceFilters&gt; tasks = 2;</code>
+   */
+
+  org.tensorflow.proto.distruntime.TaskDeviceFilters getTasksOrDefault(
+      int key,
+      org.tensorflow.proto.distruntime.TaskDeviceFilters defaultValue);
+  /**
+   * <pre>
+   * Mapping from task ID to task device filters.
+   * </pre>
+   *
+   * <code>map&lt;int32, .tensorflow.TaskDeviceFilters&gt; tasks = 2;</code>
+   */
+
+  org.tensorflow.proto.distruntime.TaskDeviceFilters getTasksOrThrow(
+      int key);
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/distruntime/ServerDef.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/distruntime/ServerDef.java
new file mode 100644
index 00000000000..068264c171e
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/distruntime/ServerDef.java
@@ -0,0 +1,1611 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/tensorflow_server.proto
+
+package org.tensorflow.proto.distruntime;
+
+/**
+ * <pre>
+ * Defines the configuration of a single TensorFlow server.
+ * </pre>
+ *
+ * Protobuf type {@code tensorflow.ServerDef}
+ */
+public  final class ServerDef extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.ServerDef)
+    ServerDefOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use ServerDef.newBuilder() to construct.
+  private ServerDef(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private ServerDef() {
+    jobName_ = "";
+    protocol_ = "";
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new ServerDef();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private ServerDef(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 10: {
+            org.tensorflow.proto.distruntime.ClusterDef.Builder subBuilder = null;
+            if (cluster_ != null) {
+              subBuilder = cluster_.toBuilder();
+            }
+            cluster_ = input.readMessage(org.tensorflow.proto.distruntime.ClusterDef.parser(), extensionRegistry);
+            if (subBuilder != null) {
+              subBuilder.mergeFrom(cluster_);
+              cluster_ = subBuilder.buildPartial();
+            }
+
+            break;
+          }
+          case 18: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            jobName_ = s;
+            break;
+          }
+          case 24: {
+
+            taskIndex_ = input.readInt32();
+            break;
+          }
+          case 34: {
+            org.tensorflow.proto.framework.ConfigProto.Builder subBuilder = null;
+            if (defaultSessionConfig_ != null) {
+              subBuilder = defaultSessionConfig_.toBuilder();
+            }
+            defaultSessionConfig_ = input.readMessage(org.tensorflow.proto.framework.ConfigProto.parser(), extensionRegistry);
+            if (subBuilder != null) {
+              subBuilder.mergeFrom(defaultSessionConfig_);
+              defaultSessionConfig_ = subBuilder.buildPartial();
+            }
+
+            break;
+          }
+          case 42: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            protocol_ = s;
+            break;
+          }
+          case 48: {
+
+            port_ = input.readInt32();
+            break;
+          }
+          case 58: {
+            org.tensorflow.proto.distruntime.ClusterDeviceFilters.Builder subBuilder = null;
+            if (clusterDeviceFilters_ != null) {
+              subBuilder = clusterDeviceFilters_.toBuilder();
+            }
+            clusterDeviceFilters_ = input.readMessage(org.tensorflow.proto.distruntime.ClusterDeviceFilters.parser(), extensionRegistry);
+            if (subBuilder != null) {
+              subBuilder.mergeFrom(clusterDeviceFilters_);
+              clusterDeviceFilters_ = subBuilder.buildPartial();
+            }
+
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.distruntime.ServerProtos.internal_static_tensorflow_ServerDef_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.distruntime.ServerProtos.internal_static_tensorflow_ServerDef_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.distruntime.ServerDef.class, org.tensorflow.proto.distruntime.ServerDef.Builder.class);
+  }
+
+  public static final int CLUSTER_FIELD_NUMBER = 1;
+  private org.tensorflow.proto.distruntime.ClusterDef cluster_;
+  /**
+   * <pre>
+   * The cluster of which this server is a member.
+   * </pre>
+   *
+   * <code>.tensorflow.ClusterDef cluster = 1;</code>
+   */
+  public boolean hasCluster() {
+    return cluster_ != null;
+  }
+  /**
+   * <pre>
+   * The cluster of which this server is a member.
+   * </pre>
+   *
+   * <code>.tensorflow.ClusterDef cluster = 1;</code>
+   */
+  public org.tensorflow.proto.distruntime.ClusterDef getCluster() {
+    return cluster_ == null ? org.tensorflow.proto.distruntime.ClusterDef.getDefaultInstance() : cluster_;
+  }
+  /**
+   * <pre>
+   * The cluster of which this server is a member.
+   * </pre>
+   *
+   * <code>.tensorflow.ClusterDef cluster = 1;</code>
+   */
+  public org.tensorflow.proto.distruntime.ClusterDefOrBuilder getClusterOrBuilder() {
+    return getCluster();
+  }
+
+  public static final int JOB_NAME_FIELD_NUMBER = 2;
+  private volatile java.lang.Object jobName_;
+  /**
+   * <pre>
+   * The name of the job of which this server is a member.
+   * NOTE(mrry): The `cluster` field must contain a `JobDef` with a `name` field
+   * that matches this name.
+   * </pre>
+   *
+   * <code>string job_name = 2;</code>
+   */
+  public java.lang.String getJobName() {
+    java.lang.Object ref = jobName_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      jobName_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * The name of the job of which this server is a member.
+   * NOTE(mrry): The `cluster` field must contain a `JobDef` with a `name` field
+   * that matches this name.
+   * </pre>
+   *
+   * <code>string job_name = 2;</code>
+   */
+  public com.google.protobuf.ByteString
+      getJobNameBytes() {
+    java.lang.Object ref = jobName_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      jobName_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int TASK_INDEX_FIELD_NUMBER = 3;
+  private int taskIndex_;
+  /**
+   * <pre>
+   * The task index of this server in its job.
+   * NOTE: The `cluster` field must contain a `JobDef` with a matching `name`
+   * and a mapping in its `tasks` field for this index.
+   * </pre>
+   *
+   * <code>int32 task_index = 3;</code>
+   */
+  public int getTaskIndex() {
+    return taskIndex_;
+  }
+
+  public static final int DEFAULT_SESSION_CONFIG_FIELD_NUMBER = 4;
+  private org.tensorflow.proto.framework.ConfigProto defaultSessionConfig_;
+  /**
+   * <pre>
+   * The default configuration for sessions that run on this server.
+   * </pre>
+   *
+   * <code>.tensorflow.ConfigProto default_session_config = 4;</code>
+   */
+  public boolean hasDefaultSessionConfig() {
+    return defaultSessionConfig_ != null;
+  }
+  /**
+   * <pre>
+   * The default configuration for sessions that run on this server.
+   * </pre>
+   *
+   * <code>.tensorflow.ConfigProto default_session_config = 4;</code>
+   */
+  public org.tensorflow.proto.framework.ConfigProto getDefaultSessionConfig() {
+    return defaultSessionConfig_ == null ? org.tensorflow.proto.framework.ConfigProto.getDefaultInstance() : defaultSessionConfig_;
+  }
+  /**
+   * <pre>
+   * The default configuration for sessions that run on this server.
+   * </pre>
+   *
+   * <code>.tensorflow.ConfigProto default_session_config = 4;</code>
+   */
+  public org.tensorflow.proto.framework.ConfigProtoOrBuilder getDefaultSessionConfigOrBuilder() {
+    return getDefaultSessionConfig();
+  }
+
+  public static final int PROTOCOL_FIELD_NUMBER = 5;
+  private volatile java.lang.Object protocol_;
+  /**
+   * <pre>
+   * The protocol to be used by this server.
+   * Acceptable values include: "grpc", "grpc+verbs".
+   * </pre>
+   *
+   * <code>string protocol = 5;</code>
+   */
+  public java.lang.String getProtocol() {
+    java.lang.Object ref = protocol_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      protocol_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * The protocol to be used by this server.
+   * Acceptable values include: "grpc", "grpc+verbs".
+   * </pre>
+   *
+   * <code>string protocol = 5;</code>
+   */
+  public com.google.protobuf.ByteString
+      getProtocolBytes() {
+    java.lang.Object ref = protocol_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      protocol_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int PORT_FIELD_NUMBER = 6;
+  private int port_;
+  /**
+   * <pre>
+   * The server port. If not set, then we identify the port from the job_name.
+   * </pre>
+   *
+   * <code>int32 port = 6;</code>
+   */
+  public int getPort() {
+    return port_;
+  }
+
+  public static final int CLUSTER_DEVICE_FILTERS_FIELD_NUMBER = 7;
+  private org.tensorflow.proto.distruntime.ClusterDeviceFilters clusterDeviceFilters_;
+  /**
+   * <pre>
+   * Device filters for remote tasks in the cluster.
+   * NOTE: This is an experimental feature and only effective in TensorFlow 2.x.
+   * </pre>
+   *
+   * <code>.tensorflow.ClusterDeviceFilters cluster_device_filters = 7;</code>
+   */
+  public boolean hasClusterDeviceFilters() {
+    return clusterDeviceFilters_ != null;
+  }
+  /**
+   * <pre>
+   * Device filters for remote tasks in the cluster.
+   * NOTE: This is an experimental feature and only effective in TensorFlow 2.x.
+   * </pre>
+   *
+   * <code>.tensorflow.ClusterDeviceFilters cluster_device_filters = 7;</code>
+   */
+  public org.tensorflow.proto.distruntime.ClusterDeviceFilters getClusterDeviceFilters() {
+    return clusterDeviceFilters_ == null ? org.tensorflow.proto.distruntime.ClusterDeviceFilters.getDefaultInstance() : clusterDeviceFilters_;
+  }
+  /**
+   * <pre>
+   * Device filters for remote tasks in the cluster.
+   * NOTE: This is an experimental feature and only effective in TensorFlow 2.x.
+   * </pre>
+   *
+   * <code>.tensorflow.ClusterDeviceFilters cluster_device_filters = 7;</code>
+   */
+  public org.tensorflow.proto.distruntime.ClusterDeviceFiltersOrBuilder getClusterDeviceFiltersOrBuilder() {
+    return getClusterDeviceFilters();
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    if (cluster_ != null) {
+      output.writeMessage(1, getCluster());
+    }
+    if (!getJobNameBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 2, jobName_);
+    }
+    if (taskIndex_ != 0) {
+      output.writeInt32(3, taskIndex_);
+    }
+    if (defaultSessionConfig_ != null) {
+      output.writeMessage(4, getDefaultSessionConfig());
+    }
+    if (!getProtocolBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 5, protocol_);
+    }
+    if (port_ != 0) {
+      output.writeInt32(6, port_);
+    }
+    if (clusterDeviceFilters_ != null) {
+      output.writeMessage(7, getClusterDeviceFilters());
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (cluster_ != null) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(1, getCluster());
+    }
+    if (!getJobNameBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(2, jobName_);
+    }
+    if (taskIndex_ != 0) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt32Size(3, taskIndex_);
+    }
+    if (defaultSessionConfig_ != null) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(4, getDefaultSessionConfig());
+    }
+    if (!getProtocolBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(5, protocol_);
+    }
+    if (port_ != 0) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt32Size(6, port_);
+    }
+    if (clusterDeviceFilters_ != null) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(7, getClusterDeviceFilters());
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.distruntime.ServerDef)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.distruntime.ServerDef other = (org.tensorflow.proto.distruntime.ServerDef) obj;
+
+    if (hasCluster() != other.hasCluster()) return false;
+    if (hasCluster()) {
+      if (!getCluster()
+          .equals(other.getCluster())) return false;
+    }
+    if (!getJobName()
+        .equals(other.getJobName())) return false;
+    if (getTaskIndex()
+        != other.getTaskIndex()) return false;
+    if (hasDefaultSessionConfig() != other.hasDefaultSessionConfig()) return false;
+    if (hasDefaultSessionConfig()) {
+      if (!getDefaultSessionConfig()
+          .equals(other.getDefaultSessionConfig())) return false;
+    }
+    if (!getProtocol()
+        .equals(other.getProtocol())) return false;
+    if (getPort()
+        != other.getPort()) return false;
+    if (hasClusterDeviceFilters() != other.hasClusterDeviceFilters()) return false;
+    if (hasClusterDeviceFilters()) {
+      if (!getClusterDeviceFilters()
+          .equals(other.getClusterDeviceFilters())) return false;
+    }
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    if (hasCluster()) {
+      hash = (37 * hash) + CLUSTER_FIELD_NUMBER;
+      hash = (53 * hash) + getCluster().hashCode();
+    }
+    hash = (37 * hash) + JOB_NAME_FIELD_NUMBER;
+    hash = (53 * hash) + getJobName().hashCode();
+    hash = (37 * hash) + TASK_INDEX_FIELD_NUMBER;
+    hash = (53 * hash) + getTaskIndex();
+    if (hasDefaultSessionConfig()) {
+      hash = (37 * hash) + DEFAULT_SESSION_CONFIG_FIELD_NUMBER;
+      hash = (53 * hash) + getDefaultSessionConfig().hashCode();
+    }
+    hash = (37 * hash) + PROTOCOL_FIELD_NUMBER;
+    hash = (53 * hash) + getProtocol().hashCode();
+    hash = (37 * hash) + PORT_FIELD_NUMBER;
+    hash = (53 * hash) + getPort();
+    if (hasClusterDeviceFilters()) {
+      hash = (37 * hash) + CLUSTER_DEVICE_FILTERS_FIELD_NUMBER;
+      hash = (53 * hash) + getClusterDeviceFilters().hashCode();
+    }
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.distruntime.ServerDef parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.distruntime.ServerDef parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.distruntime.ServerDef parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.distruntime.ServerDef parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.distruntime.ServerDef parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.distruntime.ServerDef parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.distruntime.ServerDef parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.distruntime.ServerDef parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.distruntime.ServerDef parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.distruntime.ServerDef parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.distruntime.ServerDef parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.distruntime.ServerDef parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.distruntime.ServerDef prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * <pre>
+   * Defines the configuration of a single TensorFlow server.
+   * </pre>
+   *
+   * Protobuf type {@code tensorflow.ServerDef}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.ServerDef)
+      org.tensorflow.proto.distruntime.ServerDefOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.distruntime.ServerProtos.internal_static_tensorflow_ServerDef_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.distruntime.ServerProtos.internal_static_tensorflow_ServerDef_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.distruntime.ServerDef.class, org.tensorflow.proto.distruntime.ServerDef.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.distruntime.ServerDef.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      if (clusterBuilder_ == null) {
+        cluster_ = null;
+      } else {
+        cluster_ = null;
+        clusterBuilder_ = null;
+      }
+      jobName_ = "";
+
+      taskIndex_ = 0;
+
+      if (defaultSessionConfigBuilder_ == null) {
+        defaultSessionConfig_ = null;
+      } else {
+        defaultSessionConfig_ = null;
+        defaultSessionConfigBuilder_ = null;
+      }
+      protocol_ = "";
+
+      port_ = 0;
+
+      if (clusterDeviceFiltersBuilder_ == null) {
+        clusterDeviceFilters_ = null;
+      } else {
+        clusterDeviceFilters_ = null;
+        clusterDeviceFiltersBuilder_ = null;
+      }
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.distruntime.ServerProtos.internal_static_tensorflow_ServerDef_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.distruntime.ServerDef getDefaultInstanceForType() {
+      return org.tensorflow.proto.distruntime.ServerDef.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.distruntime.ServerDef build() {
+      org.tensorflow.proto.distruntime.ServerDef result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.distruntime.ServerDef buildPartial() {
+      org.tensorflow.proto.distruntime.ServerDef result = new org.tensorflow.proto.distruntime.ServerDef(this);
+      if (clusterBuilder_ == null) {
+        result.cluster_ = cluster_;
+      } else {
+        result.cluster_ = clusterBuilder_.build();
+      }
+      result.jobName_ = jobName_;
+      result.taskIndex_ = taskIndex_;
+      if (defaultSessionConfigBuilder_ == null) {
+        result.defaultSessionConfig_ = defaultSessionConfig_;
+      } else {
+        result.defaultSessionConfig_ = defaultSessionConfigBuilder_.build();
+      }
+      result.protocol_ = protocol_;
+      result.port_ = port_;
+      if (clusterDeviceFiltersBuilder_ == null) {
+        result.clusterDeviceFilters_ = clusterDeviceFilters_;
+      } else {
+        result.clusterDeviceFilters_ = clusterDeviceFiltersBuilder_.build();
+      }
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.distruntime.ServerDef) {
+        return mergeFrom((org.tensorflow.proto.distruntime.ServerDef)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.distruntime.ServerDef other) {
+      if (other == org.tensorflow.proto.distruntime.ServerDef.getDefaultInstance()) return this;
+      if (other.hasCluster()) {
+        mergeCluster(other.getCluster());
+      }
+      if (!other.getJobName().isEmpty()) {
+        jobName_ = other.jobName_;
+        onChanged();
+      }
+      if (other.getTaskIndex() != 0) {
+        setTaskIndex(other.getTaskIndex());
+      }
+      if (other.hasDefaultSessionConfig()) {
+        mergeDefaultSessionConfig(other.getDefaultSessionConfig());
+      }
+      if (!other.getProtocol().isEmpty()) {
+        protocol_ = other.protocol_;
+        onChanged();
+      }
+      if (other.getPort() != 0) {
+        setPort(other.getPort());
+      }
+      if (other.hasClusterDeviceFilters()) {
+        mergeClusterDeviceFilters(other.getClusterDeviceFilters());
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.distruntime.ServerDef parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.distruntime.ServerDef) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+
+    private org.tensorflow.proto.distruntime.ClusterDef cluster_;
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.distruntime.ClusterDef, org.tensorflow.proto.distruntime.ClusterDef.Builder, org.tensorflow.proto.distruntime.ClusterDefOrBuilder> clusterBuilder_;
+    /**
+     * <pre>
+     * The cluster of which this server is a member.
+     * </pre>
+     *
+     * <code>.tensorflow.ClusterDef cluster = 1;</code>
+     */
+    public boolean hasCluster() {
+      return clusterBuilder_ != null || cluster_ != null;
+    }
+    /**
+     * <pre>
+     * The cluster of which this server is a member.
+     * </pre>
+     *
+     * <code>.tensorflow.ClusterDef cluster = 1;</code>
+     */
+    public org.tensorflow.proto.distruntime.ClusterDef getCluster() {
+      if (clusterBuilder_ == null) {
+        return cluster_ == null ? org.tensorflow.proto.distruntime.ClusterDef.getDefaultInstance() : cluster_;
+      } else {
+        return clusterBuilder_.getMessage();
+      }
+    }
+    /**
+     * <pre>
+     * The cluster of which this server is a member.
+     * </pre>
+     *
+     * <code>.tensorflow.ClusterDef cluster = 1;</code>
+     */
+    public Builder setCluster(org.tensorflow.proto.distruntime.ClusterDef value) {
+      if (clusterBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        cluster_ = value;
+        onChanged();
+      } else {
+        clusterBuilder_.setMessage(value);
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * The cluster of which this server is a member.
+     * </pre>
+     *
+     * <code>.tensorflow.ClusterDef cluster = 1;</code>
+     */
+    public Builder setCluster(
+        org.tensorflow.proto.distruntime.ClusterDef.Builder builderForValue) {
+      if (clusterBuilder_ == null) {
+        cluster_ = builderForValue.build();
+        onChanged();
+      } else {
+        clusterBuilder_.setMessage(builderForValue.build());
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * The cluster of which this server is a member.
+     * </pre>
+     *
+     * <code>.tensorflow.ClusterDef cluster = 1;</code>
+     */
+    public Builder mergeCluster(org.tensorflow.proto.distruntime.ClusterDef value) {
+      if (clusterBuilder_ == null) {
+        if (cluster_ != null) {
+          cluster_ =
+            org.tensorflow.proto.distruntime.ClusterDef.newBuilder(cluster_).mergeFrom(value).buildPartial();
+        } else {
+          cluster_ = value;
+        }
+        onChanged();
+      } else {
+        clusterBuilder_.mergeFrom(value);
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * The cluster of which this server is a member.
+     * </pre>
+     *
+     * <code>.tensorflow.ClusterDef cluster = 1;</code>
+     */
+    public Builder clearCluster() {
+      if (clusterBuilder_ == null) {
+        cluster_ = null;
+        onChanged();
+      } else {
+        cluster_ = null;
+        clusterBuilder_ = null;
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * The cluster of which this server is a member.
+     * </pre>
+     *
+     * <code>.tensorflow.ClusterDef cluster = 1;</code>
+     */
+    public org.tensorflow.proto.distruntime.ClusterDef.Builder getClusterBuilder() {
+      
+      onChanged();
+      return getClusterFieldBuilder().getBuilder();
+    }
+    /**
+     * <pre>
+     * The cluster of which this server is a member.
+     * </pre>
+     *
+     * <code>.tensorflow.ClusterDef cluster = 1;</code>
+     */
+    public org.tensorflow.proto.distruntime.ClusterDefOrBuilder getClusterOrBuilder() {
+      if (clusterBuilder_ != null) {
+        return clusterBuilder_.getMessageOrBuilder();
+      } else {
+        return cluster_ == null ?
+            org.tensorflow.proto.distruntime.ClusterDef.getDefaultInstance() : cluster_;
+      }
+    }
+    /**
+     * <pre>
+     * The cluster of which this server is a member.
+     * </pre>
+     *
+     * <code>.tensorflow.ClusterDef cluster = 1;</code>
+     */
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.distruntime.ClusterDef, org.tensorflow.proto.distruntime.ClusterDef.Builder, org.tensorflow.proto.distruntime.ClusterDefOrBuilder> 
+        getClusterFieldBuilder() {
+      if (clusterBuilder_ == null) {
+        clusterBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+            org.tensorflow.proto.distruntime.ClusterDef, org.tensorflow.proto.distruntime.ClusterDef.Builder, org.tensorflow.proto.distruntime.ClusterDefOrBuilder>(
+                getCluster(),
+                getParentForChildren(),
+                isClean());
+        cluster_ = null;
+      }
+      return clusterBuilder_;
+    }
+
+    private java.lang.Object jobName_ = "";
+    /**
+     * <pre>
+     * The name of the job of which this server is a member.
+     * NOTE(mrry): The `cluster` field must contain a `JobDef` with a `name` field
+     * that matches this name.
+     * </pre>
+     *
+     * <code>string job_name = 2;</code>
+     */
+    public java.lang.String getJobName() {
+      java.lang.Object ref = jobName_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        jobName_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * The name of the job of which this server is a member.
+     * NOTE(mrry): The `cluster` field must contain a `JobDef` with a `name` field
+     * that matches this name.
+     * </pre>
+     *
+     * <code>string job_name = 2;</code>
+     */
+    public com.google.protobuf.ByteString
+        getJobNameBytes() {
+      java.lang.Object ref = jobName_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        jobName_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * The name of the job of which this server is a member.
+     * NOTE(mrry): The `cluster` field must contain a `JobDef` with a `name` field
+     * that matches this name.
+     * </pre>
+     *
+     * <code>string job_name = 2;</code>
+     */
+    public Builder setJobName(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      jobName_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * The name of the job of which this server is a member.
+     * NOTE(mrry): The `cluster` field must contain a `JobDef` with a `name` field
+     * that matches this name.
+     * </pre>
+     *
+     * <code>string job_name = 2;</code>
+     */
+    public Builder clearJobName() {
+      
+      jobName_ = getDefaultInstance().getJobName();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * The name of the job of which this server is a member.
+     * NOTE(mrry): The `cluster` field must contain a `JobDef` with a `name` field
+     * that matches this name.
+     * </pre>
+     *
+     * <code>string job_name = 2;</code>
+     */
+    public Builder setJobNameBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      jobName_ = value;
+      onChanged();
+      return this;
+    }
+
+    private int taskIndex_ ;
+    /**
+     * <pre>
+     * The task index of this server in its job.
+     * NOTE: The `cluster` field must contain a `JobDef` with a matching `name`
+     * and a mapping in its `tasks` field for this index.
+     * </pre>
+     *
+     * <code>int32 task_index = 3;</code>
+     */
+    public int getTaskIndex() {
+      return taskIndex_;
+    }
+    /**
+     * <pre>
+     * The task index of this server in its job.
+     * NOTE: The `cluster` field must contain a `JobDef` with a matching `name`
+     * and a mapping in its `tasks` field for this index.
+     * </pre>
+     *
+     * <code>int32 task_index = 3;</code>
+     */
+    public Builder setTaskIndex(int value) {
+      
+      taskIndex_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * The task index of this server in its job.
+     * NOTE: The `cluster` field must contain a `JobDef` with a matching `name`
+     * and a mapping in its `tasks` field for this index.
+     * </pre>
+     *
+     * <code>int32 task_index = 3;</code>
+     */
+    public Builder clearTaskIndex() {
+      
+      taskIndex_ = 0;
+      onChanged();
+      return this;
+    }
+
+    private org.tensorflow.proto.framework.ConfigProto defaultSessionConfig_;
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.ConfigProto, org.tensorflow.proto.framework.ConfigProto.Builder, org.tensorflow.proto.framework.ConfigProtoOrBuilder> defaultSessionConfigBuilder_;
+    /**
+     * <pre>
+     * The default configuration for sessions that run on this server.
+     * </pre>
+     *
+     * <code>.tensorflow.ConfigProto default_session_config = 4;</code>
+     */
+    public boolean hasDefaultSessionConfig() {
+      return defaultSessionConfigBuilder_ != null || defaultSessionConfig_ != null;
+    }
+    /**
+     * <pre>
+     * The default configuration for sessions that run on this server.
+     * </pre>
+     *
+     * <code>.tensorflow.ConfigProto default_session_config = 4;</code>
+     */
+    public org.tensorflow.proto.framework.ConfigProto getDefaultSessionConfig() {
+      if (defaultSessionConfigBuilder_ == null) {
+        return defaultSessionConfig_ == null ? org.tensorflow.proto.framework.ConfigProto.getDefaultInstance() : defaultSessionConfig_;
+      } else {
+        return defaultSessionConfigBuilder_.getMessage();
+      }
+    }
+    /**
+     * <pre>
+     * The default configuration for sessions that run on this server.
+     * </pre>
+     *
+     * <code>.tensorflow.ConfigProto default_session_config = 4;</code>
+     */
+    public Builder setDefaultSessionConfig(org.tensorflow.proto.framework.ConfigProto value) {
+      if (defaultSessionConfigBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        defaultSessionConfig_ = value;
+        onChanged();
+      } else {
+        defaultSessionConfigBuilder_.setMessage(value);
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * The default configuration for sessions that run on this server.
+     * </pre>
+     *
+     * <code>.tensorflow.ConfigProto default_session_config = 4;</code>
+     */
+    public Builder setDefaultSessionConfig(
+        org.tensorflow.proto.framework.ConfigProto.Builder builderForValue) {
+      if (defaultSessionConfigBuilder_ == null) {
+        defaultSessionConfig_ = builderForValue.build();
+        onChanged();
+      } else {
+        defaultSessionConfigBuilder_.setMessage(builderForValue.build());
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * The default configuration for sessions that run on this server.
+     * </pre>
+     *
+     * <code>.tensorflow.ConfigProto default_session_config = 4;</code>
+     */
+    public Builder mergeDefaultSessionConfig(org.tensorflow.proto.framework.ConfigProto value) {
+      if (defaultSessionConfigBuilder_ == null) {
+        if (defaultSessionConfig_ != null) {
+          defaultSessionConfig_ =
+            org.tensorflow.proto.framework.ConfigProto.newBuilder(defaultSessionConfig_).mergeFrom(value).buildPartial();
+        } else {
+          defaultSessionConfig_ = value;
+        }
+        onChanged();
+      } else {
+        defaultSessionConfigBuilder_.mergeFrom(value);
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * The default configuration for sessions that run on this server.
+     * </pre>
+     *
+     * <code>.tensorflow.ConfigProto default_session_config = 4;</code>
+     */
+    public Builder clearDefaultSessionConfig() {
+      if (defaultSessionConfigBuilder_ == null) {
+        defaultSessionConfig_ = null;
+        onChanged();
+      } else {
+        defaultSessionConfig_ = null;
+        defaultSessionConfigBuilder_ = null;
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * The default configuration for sessions that run on this server.
+     * </pre>
+     *
+     * <code>.tensorflow.ConfigProto default_session_config = 4;</code>
+     */
+    public org.tensorflow.proto.framework.ConfigProto.Builder getDefaultSessionConfigBuilder() {
+      
+      onChanged();
+      return getDefaultSessionConfigFieldBuilder().getBuilder();
+    }
+    /**
+     * <pre>
+     * The default configuration for sessions that run on this server.
+     * </pre>
+     *
+     * <code>.tensorflow.ConfigProto default_session_config = 4;</code>
+     */
+    public org.tensorflow.proto.framework.ConfigProtoOrBuilder getDefaultSessionConfigOrBuilder() {
+      if (defaultSessionConfigBuilder_ != null) {
+        return defaultSessionConfigBuilder_.getMessageOrBuilder();
+      } else {
+        return defaultSessionConfig_ == null ?
+            org.tensorflow.proto.framework.ConfigProto.getDefaultInstance() : defaultSessionConfig_;
+      }
+    }
+    /**
+     * <pre>
+     * The default configuration for sessions that run on this server.
+     * </pre>
+     *
+     * <code>.tensorflow.ConfigProto default_session_config = 4;</code>
+     */
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.ConfigProto, org.tensorflow.proto.framework.ConfigProto.Builder, org.tensorflow.proto.framework.ConfigProtoOrBuilder> 
+        getDefaultSessionConfigFieldBuilder() {
+      if (defaultSessionConfigBuilder_ == null) {
+        defaultSessionConfigBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+            org.tensorflow.proto.framework.ConfigProto, org.tensorflow.proto.framework.ConfigProto.Builder, org.tensorflow.proto.framework.ConfigProtoOrBuilder>(
+                getDefaultSessionConfig(),
+                getParentForChildren(),
+                isClean());
+        defaultSessionConfig_ = null;
+      }
+      return defaultSessionConfigBuilder_;
+    }
+
+    private java.lang.Object protocol_ = "";
+    /**
+     * <pre>
+     * The protocol to be used by this server.
+     * Acceptable values include: "grpc", "grpc+verbs".
+     * </pre>
+     *
+     * <code>string protocol = 5;</code>
+     */
+    public java.lang.String getProtocol() {
+      java.lang.Object ref = protocol_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        protocol_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * The protocol to be used by this server.
+     * Acceptable values include: "grpc", "grpc+verbs".
+     * </pre>
+     *
+     * <code>string protocol = 5;</code>
+     */
+    public com.google.protobuf.ByteString
+        getProtocolBytes() {
+      java.lang.Object ref = protocol_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        protocol_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * The protocol to be used by this server.
+     * Acceptable values include: "grpc", "grpc+verbs".
+     * </pre>
+     *
+     * <code>string protocol = 5;</code>
+     */
+    public Builder setProtocol(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      protocol_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * The protocol to be used by this server.
+     * Acceptable values include: "grpc", "grpc+verbs".
+     * </pre>
+     *
+     * <code>string protocol = 5;</code>
+     */
+    public Builder clearProtocol() {
+      
+      protocol_ = getDefaultInstance().getProtocol();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * The protocol to be used by this server.
+     * Acceptable values include: "grpc", "grpc+verbs".
+     * </pre>
+     *
+     * <code>string protocol = 5;</code>
+     */
+    public Builder setProtocolBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      protocol_ = value;
+      onChanged();
+      return this;
+    }
+
+    private int port_ ;
+    /**
+     * <pre>
+     * The server port. If not set, then we identify the port from the job_name.
+     * </pre>
+     *
+     * <code>int32 port = 6;</code>
+     */
+    public int getPort() {
+      return port_;
+    }
+    /**
+     * <pre>
+     * The server port. If not set, then we identify the port from the job_name.
+     * </pre>
+     *
+     * <code>int32 port = 6;</code>
+     */
+    public Builder setPort(int value) {
+      
+      port_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * The server port. If not set, then we identify the port from the job_name.
+     * </pre>
+     *
+     * <code>int32 port = 6;</code>
+     */
+    public Builder clearPort() {
+      
+      port_ = 0;
+      onChanged();
+      return this;
+    }
+
+    private org.tensorflow.proto.distruntime.ClusterDeviceFilters clusterDeviceFilters_;
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.distruntime.ClusterDeviceFilters, org.tensorflow.proto.distruntime.ClusterDeviceFilters.Builder, org.tensorflow.proto.distruntime.ClusterDeviceFiltersOrBuilder> clusterDeviceFiltersBuilder_;
+    /**
+     * <pre>
+     * Device filters for remote tasks in the cluster.
+     * NOTE: This is an experimental feature and only effective in TensorFlow 2.x.
+     * </pre>
+     *
+     * <code>.tensorflow.ClusterDeviceFilters cluster_device_filters = 7;</code>
+     */
+    public boolean hasClusterDeviceFilters() {
+      return clusterDeviceFiltersBuilder_ != null || clusterDeviceFilters_ != null;
+    }
+    /**
+     * <pre>
+     * Device filters for remote tasks in the cluster.
+     * NOTE: This is an experimental feature and only effective in TensorFlow 2.x.
+     * </pre>
+     *
+     * <code>.tensorflow.ClusterDeviceFilters cluster_device_filters = 7;</code>
+     */
+    public org.tensorflow.proto.distruntime.ClusterDeviceFilters getClusterDeviceFilters() {
+      if (clusterDeviceFiltersBuilder_ == null) {
+        return clusterDeviceFilters_ == null ? org.tensorflow.proto.distruntime.ClusterDeviceFilters.getDefaultInstance() : clusterDeviceFilters_;
+      } else {
+        return clusterDeviceFiltersBuilder_.getMessage();
+      }
+    }
+    /**
+     * <pre>
+     * Device filters for remote tasks in the cluster.
+     * NOTE: This is an experimental feature and only effective in TensorFlow 2.x.
+     * </pre>
+     *
+     * <code>.tensorflow.ClusterDeviceFilters cluster_device_filters = 7;</code>
+     */
+    public Builder setClusterDeviceFilters(org.tensorflow.proto.distruntime.ClusterDeviceFilters value) {
+      if (clusterDeviceFiltersBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        clusterDeviceFilters_ = value;
+        onChanged();
+      } else {
+        clusterDeviceFiltersBuilder_.setMessage(value);
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Device filters for remote tasks in the cluster.
+     * NOTE: This is an experimental feature and only effective in TensorFlow 2.x.
+     * </pre>
+     *
+     * <code>.tensorflow.ClusterDeviceFilters cluster_device_filters = 7;</code>
+     */
+    public Builder setClusterDeviceFilters(
+        org.tensorflow.proto.distruntime.ClusterDeviceFilters.Builder builderForValue) {
+      if (clusterDeviceFiltersBuilder_ == null) {
+        clusterDeviceFilters_ = builderForValue.build();
+        onChanged();
+      } else {
+        clusterDeviceFiltersBuilder_.setMessage(builderForValue.build());
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Device filters for remote tasks in the cluster.
+     * NOTE: This is an experimental feature and only effective in TensorFlow 2.x.
+     * </pre>
+     *
+     * <code>.tensorflow.ClusterDeviceFilters cluster_device_filters = 7;</code>
+     */
+    public Builder mergeClusterDeviceFilters(org.tensorflow.proto.distruntime.ClusterDeviceFilters value) {
+      if (clusterDeviceFiltersBuilder_ == null) {
+        if (clusterDeviceFilters_ != null) {
+          clusterDeviceFilters_ =
+            org.tensorflow.proto.distruntime.ClusterDeviceFilters.newBuilder(clusterDeviceFilters_).mergeFrom(value).buildPartial();
+        } else {
+          clusterDeviceFilters_ = value;
+        }
+        onChanged();
+      } else {
+        clusterDeviceFiltersBuilder_.mergeFrom(value);
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Device filters for remote tasks in the cluster.
+     * NOTE: This is an experimental feature and only effective in TensorFlow 2.x.
+     * </pre>
+     *
+     * <code>.tensorflow.ClusterDeviceFilters cluster_device_filters = 7;</code>
+     */
+    public Builder clearClusterDeviceFilters() {
+      if (clusterDeviceFiltersBuilder_ == null) {
+        clusterDeviceFilters_ = null;
+        onChanged();
+      } else {
+        clusterDeviceFilters_ = null;
+        clusterDeviceFiltersBuilder_ = null;
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Device filters for remote tasks in the cluster.
+     * NOTE: This is an experimental feature and only effective in TensorFlow 2.x.
+     * </pre>
+     *
+     * <code>.tensorflow.ClusterDeviceFilters cluster_device_filters = 7;</code>
+     */
+    public org.tensorflow.proto.distruntime.ClusterDeviceFilters.Builder getClusterDeviceFiltersBuilder() {
+      
+      onChanged();
+      return getClusterDeviceFiltersFieldBuilder().getBuilder();
+    }
+    /**
+     * <pre>
+     * Device filters for remote tasks in the cluster.
+     * NOTE: This is an experimental feature and only effective in TensorFlow 2.x.
+     * </pre>
+     *
+     * <code>.tensorflow.ClusterDeviceFilters cluster_device_filters = 7;</code>
+     */
+    public org.tensorflow.proto.distruntime.ClusterDeviceFiltersOrBuilder getClusterDeviceFiltersOrBuilder() {
+      if (clusterDeviceFiltersBuilder_ != null) {
+        return clusterDeviceFiltersBuilder_.getMessageOrBuilder();
+      } else {
+        return clusterDeviceFilters_ == null ?
+            org.tensorflow.proto.distruntime.ClusterDeviceFilters.getDefaultInstance() : clusterDeviceFilters_;
+      }
+    }
+    /**
+     * <pre>
+     * Device filters for remote tasks in the cluster.
+     * NOTE: This is an experimental feature and only effective in TensorFlow 2.x.
+     * </pre>
+     *
+     * <code>.tensorflow.ClusterDeviceFilters cluster_device_filters = 7;</code>
+     */
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.distruntime.ClusterDeviceFilters, org.tensorflow.proto.distruntime.ClusterDeviceFilters.Builder, org.tensorflow.proto.distruntime.ClusterDeviceFiltersOrBuilder> 
+        getClusterDeviceFiltersFieldBuilder() {
+      if (clusterDeviceFiltersBuilder_ == null) {
+        clusterDeviceFiltersBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+            org.tensorflow.proto.distruntime.ClusterDeviceFilters, org.tensorflow.proto.distruntime.ClusterDeviceFilters.Builder, org.tensorflow.proto.distruntime.ClusterDeviceFiltersOrBuilder>(
+                getClusterDeviceFilters(),
+                getParentForChildren(),
+                isClean());
+        clusterDeviceFilters_ = null;
+      }
+      return clusterDeviceFiltersBuilder_;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.ServerDef)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.ServerDef)
+  private static final org.tensorflow.proto.distruntime.ServerDef DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.distruntime.ServerDef();
+  }
+
+  public static org.tensorflow.proto.distruntime.ServerDef getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<ServerDef>
+      PARSER = new com.google.protobuf.AbstractParser<ServerDef>() {
+    @java.lang.Override
+    public ServerDef parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new ServerDef(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<ServerDef> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<ServerDef> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.distruntime.ServerDef getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/distruntime/ServerDefOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/distruntime/ServerDefOrBuilder.java
new file mode 100644
index 00000000000..6c1a12b002d
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/distruntime/ServerDefOrBuilder.java
@@ -0,0 +1,149 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/tensorflow_server.proto
+
+package org.tensorflow.proto.distruntime;
+
+public interface ServerDefOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.ServerDef)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <pre>
+   * The cluster of which this server is a member.
+   * </pre>
+   *
+   * <code>.tensorflow.ClusterDef cluster = 1;</code>
+   */
+  boolean hasCluster();
+  /**
+   * <pre>
+   * The cluster of which this server is a member.
+   * </pre>
+   *
+   * <code>.tensorflow.ClusterDef cluster = 1;</code>
+   */
+  org.tensorflow.proto.distruntime.ClusterDef getCluster();
+  /**
+   * <pre>
+   * The cluster of which this server is a member.
+   * </pre>
+   *
+   * <code>.tensorflow.ClusterDef cluster = 1;</code>
+   */
+  org.tensorflow.proto.distruntime.ClusterDefOrBuilder getClusterOrBuilder();
+
+  /**
+   * <pre>
+   * The name of the job of which this server is a member.
+   * NOTE(mrry): The `cluster` field must contain a `JobDef` with a `name` field
+   * that matches this name.
+   * </pre>
+   *
+   * <code>string job_name = 2;</code>
+   */
+  java.lang.String getJobName();
+  /**
+   * <pre>
+   * The name of the job of which this server is a member.
+   * NOTE(mrry): The `cluster` field must contain a `JobDef` with a `name` field
+   * that matches this name.
+   * </pre>
+   *
+   * <code>string job_name = 2;</code>
+   */
+  com.google.protobuf.ByteString
+      getJobNameBytes();
+
+  /**
+   * <pre>
+   * The task index of this server in its job.
+   * NOTE: The `cluster` field must contain a `JobDef` with a matching `name`
+   * and a mapping in its `tasks` field for this index.
+   * </pre>
+   *
+   * <code>int32 task_index = 3;</code>
+   */
+  int getTaskIndex();
+
+  /**
+   * <pre>
+   * The default configuration for sessions that run on this server.
+   * </pre>
+   *
+   * <code>.tensorflow.ConfigProto default_session_config = 4;</code>
+   */
+  boolean hasDefaultSessionConfig();
+  /**
+   * <pre>
+   * The default configuration for sessions that run on this server.
+   * </pre>
+   *
+   * <code>.tensorflow.ConfigProto default_session_config = 4;</code>
+   */
+  org.tensorflow.proto.framework.ConfigProto getDefaultSessionConfig();
+  /**
+   * <pre>
+   * The default configuration for sessions that run on this server.
+   * </pre>
+   *
+   * <code>.tensorflow.ConfigProto default_session_config = 4;</code>
+   */
+  org.tensorflow.proto.framework.ConfigProtoOrBuilder getDefaultSessionConfigOrBuilder();
+
+  /**
+   * <pre>
+   * The protocol to be used by this server.
+   * Acceptable values include: "grpc", "grpc+verbs".
+   * </pre>
+   *
+   * <code>string protocol = 5;</code>
+   */
+  java.lang.String getProtocol();
+  /**
+   * <pre>
+   * The protocol to be used by this server.
+   * Acceptable values include: "grpc", "grpc+verbs".
+   * </pre>
+   *
+   * <code>string protocol = 5;</code>
+   */
+  com.google.protobuf.ByteString
+      getProtocolBytes();
+
+  /**
+   * <pre>
+   * The server port. If not set, then we identify the port from the job_name.
+   * </pre>
+   *
+   * <code>int32 port = 6;</code>
+   */
+  int getPort();
+
+  /**
+   * <pre>
+   * Device filters for remote tasks in the cluster.
+   * NOTE: This is an experimental feature and only effective in TensorFlow 2.x.
+   * </pre>
+   *
+   * <code>.tensorflow.ClusterDeviceFilters cluster_device_filters = 7;</code>
+   */
+  boolean hasClusterDeviceFilters();
+  /**
+   * <pre>
+   * Device filters for remote tasks in the cluster.
+   * NOTE: This is an experimental feature and only effective in TensorFlow 2.x.
+   * </pre>
+   *
+   * <code>.tensorflow.ClusterDeviceFilters cluster_device_filters = 7;</code>
+   */
+  org.tensorflow.proto.distruntime.ClusterDeviceFilters getClusterDeviceFilters();
+  /**
+   * <pre>
+   * Device filters for remote tasks in the cluster.
+   * NOTE: This is an experimental feature and only effective in TensorFlow 2.x.
+   * </pre>
+   *
+   * <code>.tensorflow.ClusterDeviceFilters cluster_device_filters = 7;</code>
+   */
+  org.tensorflow.proto.distruntime.ClusterDeviceFiltersOrBuilder getClusterDeviceFiltersOrBuilder();
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/distruntime/ServerProtos.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/distruntime/ServerProtos.java
new file mode 100644
index 00000000000..d5b21d1500d
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/distruntime/ServerProtos.java
@@ -0,0 +1,66 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/tensorflow_server.proto
+
+package org.tensorflow.proto.distruntime;
+
+public final class ServerProtos {
+  private ServerProtos() {}
+  public static void registerAllExtensions(
+      com.google.protobuf.ExtensionRegistryLite registry) {
+  }
+
+  public static void registerAllExtensions(
+      com.google.protobuf.ExtensionRegistry registry) {
+    registerAllExtensions(
+        (com.google.protobuf.ExtensionRegistryLite) registry);
+  }
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_ServerDef_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_ServerDef_fieldAccessorTable;
+
+  public static com.google.protobuf.Descriptors.FileDescriptor
+      getDescriptor() {
+    return descriptor;
+  }
+  private static  com.google.protobuf.Descriptors.FileDescriptor
+      descriptor;
+  static {
+    java.lang.String[] descriptorData = {
+      "\n0tensorflow/core/protobuf/tensorflow_se" +
+      "rver.proto\022\ntensorflow\032&tensorflow/core/" +
+      "protobuf/cluster.proto\032%tensorflow/core/" +
+      "protobuf/config.proto\032-tensorflow/core/p" +
+      "rotobuf/device_filters.proto\"\365\001\n\tServerD" +
+      "ef\022\'\n\007cluster\030\001 \001(\0132\026.tensorflow.Cluster" +
+      "Def\022\020\n\010job_name\030\002 \001(\t\022\022\n\ntask_index\030\003 \001(" +
+      "\005\0227\n\026default_session_config\030\004 \001(\0132\027.tens" +
+      "orflow.ConfigProto\022\020\n\010protocol\030\005 \001(\t\022\014\n\004" +
+      "port\030\006 \001(\005\022@\n\026cluster_device_filters\030\007 \001" +
+      "(\0132 .tensorflow.ClusterDeviceFiltersB\177\n " +
+      "org.tensorflow.proto.distruntimeB\014Server" +
+      "ProtosP\001ZHgithub.com/tensorflow/tensorfl" +
+      "ow/tensorflow/go/core/core_protos_go_pro" +
+      "to\370\001\001b\006proto3"
+    };
+    descriptor = com.google.protobuf.Descriptors.FileDescriptor
+      .internalBuildGeneratedFileFrom(descriptorData,
+        new com.google.protobuf.Descriptors.FileDescriptor[] {
+          org.tensorflow.proto.distruntime.ClusterProtos.getDescriptor(),
+          org.tensorflow.proto.framework.ConfigProtos.getDescriptor(),
+          org.tensorflow.proto.distruntime.DeviceFiltersProtos.getDescriptor(),
+        });
+    internal_static_tensorflow_ServerDef_descriptor =
+      getDescriptor().getMessageTypes().get(0);
+    internal_static_tensorflow_ServerDef_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_ServerDef_descriptor,
+        new java.lang.String[] { "Cluster", "JobName", "TaskIndex", "DefaultSessionConfig", "Protocol", "Port", "ClusterDeviceFilters", });
+    org.tensorflow.proto.distruntime.ClusterProtos.getDescriptor();
+    org.tensorflow.proto.framework.ConfigProtos.getDescriptor();
+    org.tensorflow.proto.distruntime.DeviceFiltersProtos.getDescriptor();
+  }
+
+  // @@protoc_insertion_point(outer_class_scope)
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/distruntime/TaskDeviceFilters.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/distruntime/TaskDeviceFilters.java
new file mode 100644
index 00000000000..50bc759c880
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/distruntime/TaskDeviceFilters.java
@@ -0,0 +1,600 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/device_filters.proto
+
+package org.tensorflow.proto.distruntime;
+
+/**
+ * <pre>
+ * Defines the device filters for a remote task.
+ * </pre>
+ *
+ * Protobuf type {@code tensorflow.TaskDeviceFilters}
+ */
+public  final class TaskDeviceFilters extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.TaskDeviceFilters)
+    TaskDeviceFiltersOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use TaskDeviceFilters.newBuilder() to construct.
+  private TaskDeviceFilters(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private TaskDeviceFilters() {
+    deviceFilters_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new TaskDeviceFilters();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private TaskDeviceFilters(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    int mutable_bitField0_ = 0;
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 10: {
+            java.lang.String s = input.readStringRequireUtf8();
+            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
+              deviceFilters_ = new com.google.protobuf.LazyStringArrayList();
+              mutable_bitField0_ |= 0x00000001;
+            }
+            deviceFilters_.add(s);
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      if (((mutable_bitField0_ & 0x00000001) != 0)) {
+        deviceFilters_ = deviceFilters_.getUnmodifiableView();
+      }
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.distruntime.DeviceFiltersProtos.internal_static_tensorflow_TaskDeviceFilters_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.distruntime.DeviceFiltersProtos.internal_static_tensorflow_TaskDeviceFilters_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.distruntime.TaskDeviceFilters.class, org.tensorflow.proto.distruntime.TaskDeviceFilters.Builder.class);
+  }
+
+  public static final int DEVICE_FILTERS_FIELD_NUMBER = 1;
+  private com.google.protobuf.LazyStringList deviceFilters_;
+  /**
+   * <code>repeated string device_filters = 1;</code>
+   */
+  public com.google.protobuf.ProtocolStringList
+      getDeviceFiltersList() {
+    return deviceFilters_;
+  }
+  /**
+   * <code>repeated string device_filters = 1;</code>
+   */
+  public int getDeviceFiltersCount() {
+    return deviceFilters_.size();
+  }
+  /**
+   * <code>repeated string device_filters = 1;</code>
+   */
+  public java.lang.String getDeviceFilters(int index) {
+    return deviceFilters_.get(index);
+  }
+  /**
+   * <code>repeated string device_filters = 1;</code>
+   */
+  public com.google.protobuf.ByteString
+      getDeviceFiltersBytes(int index) {
+    return deviceFilters_.getByteString(index);
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    for (int i = 0; i < deviceFilters_.size(); i++) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 1, deviceFilters_.getRaw(i));
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    {
+      int dataSize = 0;
+      for (int i = 0; i < deviceFilters_.size(); i++) {
+        dataSize += computeStringSizeNoTag(deviceFilters_.getRaw(i));
+      }
+      size += dataSize;
+      size += 1 * getDeviceFiltersList().size();
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.distruntime.TaskDeviceFilters)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.distruntime.TaskDeviceFilters other = (org.tensorflow.proto.distruntime.TaskDeviceFilters) obj;
+
+    if (!getDeviceFiltersList()
+        .equals(other.getDeviceFiltersList())) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    if (getDeviceFiltersCount() > 0) {
+      hash = (37 * hash) + DEVICE_FILTERS_FIELD_NUMBER;
+      hash = (53 * hash) + getDeviceFiltersList().hashCode();
+    }
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.distruntime.TaskDeviceFilters parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.distruntime.TaskDeviceFilters parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.distruntime.TaskDeviceFilters parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.distruntime.TaskDeviceFilters parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.distruntime.TaskDeviceFilters parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.distruntime.TaskDeviceFilters parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.distruntime.TaskDeviceFilters parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.distruntime.TaskDeviceFilters parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.distruntime.TaskDeviceFilters parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.distruntime.TaskDeviceFilters parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.distruntime.TaskDeviceFilters parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.distruntime.TaskDeviceFilters parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.distruntime.TaskDeviceFilters prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * <pre>
+   * Defines the device filters for a remote task.
+   * </pre>
+   *
+   * Protobuf type {@code tensorflow.TaskDeviceFilters}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.TaskDeviceFilters)
+      org.tensorflow.proto.distruntime.TaskDeviceFiltersOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.distruntime.DeviceFiltersProtos.internal_static_tensorflow_TaskDeviceFilters_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.distruntime.DeviceFiltersProtos.internal_static_tensorflow_TaskDeviceFilters_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.distruntime.TaskDeviceFilters.class, org.tensorflow.proto.distruntime.TaskDeviceFilters.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.distruntime.TaskDeviceFilters.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      deviceFilters_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+      bitField0_ = (bitField0_ & ~0x00000001);
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.distruntime.DeviceFiltersProtos.internal_static_tensorflow_TaskDeviceFilters_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.distruntime.TaskDeviceFilters getDefaultInstanceForType() {
+      return org.tensorflow.proto.distruntime.TaskDeviceFilters.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.distruntime.TaskDeviceFilters build() {
+      org.tensorflow.proto.distruntime.TaskDeviceFilters result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.distruntime.TaskDeviceFilters buildPartial() {
+      org.tensorflow.proto.distruntime.TaskDeviceFilters result = new org.tensorflow.proto.distruntime.TaskDeviceFilters(this);
+      int from_bitField0_ = bitField0_;
+      if (((bitField0_ & 0x00000001) != 0)) {
+        deviceFilters_ = deviceFilters_.getUnmodifiableView();
+        bitField0_ = (bitField0_ & ~0x00000001);
+      }
+      result.deviceFilters_ = deviceFilters_;
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.distruntime.TaskDeviceFilters) {
+        return mergeFrom((org.tensorflow.proto.distruntime.TaskDeviceFilters)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.distruntime.TaskDeviceFilters other) {
+      if (other == org.tensorflow.proto.distruntime.TaskDeviceFilters.getDefaultInstance()) return this;
+      if (!other.deviceFilters_.isEmpty()) {
+        if (deviceFilters_.isEmpty()) {
+          deviceFilters_ = other.deviceFilters_;
+          bitField0_ = (bitField0_ & ~0x00000001);
+        } else {
+          ensureDeviceFiltersIsMutable();
+          deviceFilters_.addAll(other.deviceFilters_);
+        }
+        onChanged();
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.distruntime.TaskDeviceFilters parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.distruntime.TaskDeviceFilters) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+    private int bitField0_;
+
+    private com.google.protobuf.LazyStringList deviceFilters_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+    private void ensureDeviceFiltersIsMutable() {
+      if (!((bitField0_ & 0x00000001) != 0)) {
+        deviceFilters_ = new com.google.protobuf.LazyStringArrayList(deviceFilters_);
+        bitField0_ |= 0x00000001;
+       }
+    }
+    /**
+     * <code>repeated string device_filters = 1;</code>
+     */
+    public com.google.protobuf.ProtocolStringList
+        getDeviceFiltersList() {
+      return deviceFilters_.getUnmodifiableView();
+    }
+    /**
+     * <code>repeated string device_filters = 1;</code>
+     */
+    public int getDeviceFiltersCount() {
+      return deviceFilters_.size();
+    }
+    /**
+     * <code>repeated string device_filters = 1;</code>
+     */
+    public java.lang.String getDeviceFilters(int index) {
+      return deviceFilters_.get(index);
+    }
+    /**
+     * <code>repeated string device_filters = 1;</code>
+     */
+    public com.google.protobuf.ByteString
+        getDeviceFiltersBytes(int index) {
+      return deviceFilters_.getByteString(index);
+    }
+    /**
+     * <code>repeated string device_filters = 1;</code>
+     */
+    public Builder setDeviceFilters(
+        int index, java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  ensureDeviceFiltersIsMutable();
+      deviceFilters_.set(index, value);
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>repeated string device_filters = 1;</code>
+     */
+    public Builder addDeviceFilters(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  ensureDeviceFiltersIsMutable();
+      deviceFilters_.add(value);
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>repeated string device_filters = 1;</code>
+     */
+    public Builder addAllDeviceFilters(
+        java.lang.Iterable<java.lang.String> values) {
+      ensureDeviceFiltersIsMutable();
+      com.google.protobuf.AbstractMessageLite.Builder.addAll(
+          values, deviceFilters_);
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>repeated string device_filters = 1;</code>
+     */
+    public Builder clearDeviceFilters() {
+      deviceFilters_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+      bitField0_ = (bitField0_ & ~0x00000001);
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>repeated string device_filters = 1;</code>
+     */
+    public Builder addDeviceFiltersBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      ensureDeviceFiltersIsMutable();
+      deviceFilters_.add(value);
+      onChanged();
+      return this;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.TaskDeviceFilters)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.TaskDeviceFilters)
+  private static final org.tensorflow.proto.distruntime.TaskDeviceFilters DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.distruntime.TaskDeviceFilters();
+  }
+
+  public static org.tensorflow.proto.distruntime.TaskDeviceFilters getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<TaskDeviceFilters>
+      PARSER = new com.google.protobuf.AbstractParser<TaskDeviceFilters>() {
+    @java.lang.Override
+    public TaskDeviceFilters parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new TaskDeviceFilters(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<TaskDeviceFilters> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<TaskDeviceFilters> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.distruntime.TaskDeviceFilters getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/distruntime/TaskDeviceFiltersOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/distruntime/TaskDeviceFiltersOrBuilder.java
new file mode 100644
index 00000000000..2c359e1e049
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/distruntime/TaskDeviceFiltersOrBuilder.java
@@ -0,0 +1,28 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/device_filters.proto
+
+package org.tensorflow.proto.distruntime;
+
+public interface TaskDeviceFiltersOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.TaskDeviceFilters)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <code>repeated string device_filters = 1;</code>
+   */
+  java.util.List<java.lang.String>
+      getDeviceFiltersList();
+  /**
+   * <code>repeated string device_filters = 1;</code>
+   */
+  int getDeviceFiltersCount();
+  /**
+   * <code>repeated string device_filters = 1;</code>
+   */
+  java.lang.String getDeviceFilters(int index);
+  /**
+   * <code>repeated string device_filters = 1;</code>
+   */
+  com.google.protobuf.ByteString
+      getDeviceFiltersBytes(int index);
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/distruntime/TransportOptions.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/distruntime/TransportOptions.java
new file mode 100644
index 00000000000..f7498e56de5
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/distruntime/TransportOptions.java
@@ -0,0 +1,635 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/transport_options.proto
+
+package org.tensorflow.proto.distruntime;
+
+public final class TransportOptions {
+  private TransportOptions() {}
+  public static void registerAllExtensions(
+      com.google.protobuf.ExtensionRegistryLite registry) {
+  }
+
+  public static void registerAllExtensions(
+      com.google.protobuf.ExtensionRegistry registry) {
+    registerAllExtensions(
+        (com.google.protobuf.ExtensionRegistryLite) registry);
+  }
+  public interface RecvBufRespExtraOrBuilder extends
+      // @@protoc_insertion_point(interface_extends:tensorflow.RecvBufRespExtra)
+      com.google.protobuf.MessageOrBuilder {
+
+    /**
+     * <code>repeated bytes tensor_content = 1;</code>
+     */
+    java.util.List<com.google.protobuf.ByteString> getTensorContentList();
+    /**
+     * <code>repeated bytes tensor_content = 1;</code>
+     */
+    int getTensorContentCount();
+    /**
+     * <code>repeated bytes tensor_content = 1;</code>
+     */
+    com.google.protobuf.ByteString getTensorContent(int index);
+  }
+  /**
+   * <pre>
+   * Extra data needed on a non-RDMA RecvBufResponse.
+   * </pre>
+   *
+   * Protobuf type {@code tensorflow.RecvBufRespExtra}
+   */
+  public  static final class RecvBufRespExtra extends
+      com.google.protobuf.GeneratedMessageV3 implements
+      // @@protoc_insertion_point(message_implements:tensorflow.RecvBufRespExtra)
+      RecvBufRespExtraOrBuilder {
+  private static final long serialVersionUID = 0L;
+    // Use RecvBufRespExtra.newBuilder() to construct.
+    private RecvBufRespExtra(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+      super(builder);
+    }
+    private RecvBufRespExtra() {
+      tensorContent_ = java.util.Collections.emptyList();
+    }
+
+    @java.lang.Override
+    @SuppressWarnings({"unused"})
+    protected java.lang.Object newInstance(
+        UnusedPrivateParameter unused) {
+      return new RecvBufRespExtra();
+    }
+
+    @java.lang.Override
+    public final com.google.protobuf.UnknownFieldSet
+    getUnknownFields() {
+      return this.unknownFields;
+    }
+    private RecvBufRespExtra(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      this();
+      if (extensionRegistry == null) {
+        throw new java.lang.NullPointerException();
+      }
+      int mutable_bitField0_ = 0;
+      com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+          com.google.protobuf.UnknownFieldSet.newBuilder();
+      try {
+        boolean done = false;
+        while (!done) {
+          int tag = input.readTag();
+          switch (tag) {
+            case 0:
+              done = true;
+              break;
+            case 10: {
+              if (!((mutable_bitField0_ & 0x00000001) != 0)) {
+                tensorContent_ = new java.util.ArrayList<com.google.protobuf.ByteString>();
+                mutable_bitField0_ |= 0x00000001;
+              }
+              tensorContent_.add(input.readBytes());
+              break;
+            }
+            default: {
+              if (!parseUnknownField(
+                  input, unknownFields, extensionRegistry, tag)) {
+                done = true;
+              }
+              break;
+            }
+          }
+        }
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        throw e.setUnfinishedMessage(this);
+      } catch (java.io.IOException e) {
+        throw new com.google.protobuf.InvalidProtocolBufferException(
+            e).setUnfinishedMessage(this);
+      } finally {
+        if (((mutable_bitField0_ & 0x00000001) != 0)) {
+          tensorContent_ = java.util.Collections.unmodifiableList(tensorContent_); // C
+        }
+        this.unknownFields = unknownFields.build();
+        makeExtensionsImmutable();
+      }
+    }
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.distruntime.TransportOptions.internal_static_tensorflow_RecvBufRespExtra_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.distruntime.TransportOptions.internal_static_tensorflow_RecvBufRespExtra_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.distruntime.TransportOptions.RecvBufRespExtra.class, org.tensorflow.proto.distruntime.TransportOptions.RecvBufRespExtra.Builder.class);
+    }
+
+    public static final int TENSOR_CONTENT_FIELD_NUMBER = 1;
+    private java.util.List<com.google.protobuf.ByteString> tensorContent_;
+    /**
+     * <code>repeated bytes tensor_content = 1;</code>
+     */
+    public java.util.List<com.google.protobuf.ByteString>
+        getTensorContentList() {
+      return tensorContent_;
+    }
+    /**
+     * <code>repeated bytes tensor_content = 1;</code>
+     */
+    public int getTensorContentCount() {
+      return tensorContent_.size();
+    }
+    /**
+     * <code>repeated bytes tensor_content = 1;</code>
+     */
+    public com.google.protobuf.ByteString getTensorContent(int index) {
+      return tensorContent_.get(index);
+    }
+
+    private byte memoizedIsInitialized = -1;
+    @java.lang.Override
+    public final boolean isInitialized() {
+      byte isInitialized = memoizedIsInitialized;
+      if (isInitialized == 1) return true;
+      if (isInitialized == 0) return false;
+
+      memoizedIsInitialized = 1;
+      return true;
+    }
+
+    @java.lang.Override
+    public void writeTo(com.google.protobuf.CodedOutputStream output)
+                        throws java.io.IOException {
+      for (int i = 0; i < tensorContent_.size(); i++) {
+        output.writeBytes(1, tensorContent_.get(i));
+      }
+      unknownFields.writeTo(output);
+    }
+
+    @java.lang.Override
+    public int getSerializedSize() {
+      int size = memoizedSize;
+      if (size != -1) return size;
+
+      size = 0;
+      {
+        int dataSize = 0;
+        for (int i = 0; i < tensorContent_.size(); i++) {
+          dataSize += com.google.protobuf.CodedOutputStream
+            .computeBytesSizeNoTag(tensorContent_.get(i));
+        }
+        size += dataSize;
+        size += 1 * getTensorContentList().size();
+      }
+      size += unknownFields.getSerializedSize();
+      memoizedSize = size;
+      return size;
+    }
+
+    @java.lang.Override
+    public boolean equals(final java.lang.Object obj) {
+      if (obj == this) {
+       return true;
+      }
+      if (!(obj instanceof org.tensorflow.proto.distruntime.TransportOptions.RecvBufRespExtra)) {
+        return super.equals(obj);
+      }
+      org.tensorflow.proto.distruntime.TransportOptions.RecvBufRespExtra other = (org.tensorflow.proto.distruntime.TransportOptions.RecvBufRespExtra) obj;
+
+      if (!getTensorContentList()
+          .equals(other.getTensorContentList())) return false;
+      if (!unknownFields.equals(other.unknownFields)) return false;
+      return true;
+    }
+
+    @java.lang.Override
+    public int hashCode() {
+      if (memoizedHashCode != 0) {
+        return memoizedHashCode;
+      }
+      int hash = 41;
+      hash = (19 * hash) + getDescriptor().hashCode();
+      if (getTensorContentCount() > 0) {
+        hash = (37 * hash) + TENSOR_CONTENT_FIELD_NUMBER;
+        hash = (53 * hash) + getTensorContentList().hashCode();
+      }
+      hash = (29 * hash) + unknownFields.hashCode();
+      memoizedHashCode = hash;
+      return hash;
+    }
+
+    public static org.tensorflow.proto.distruntime.TransportOptions.RecvBufRespExtra parseFrom(
+        java.nio.ByteBuffer data)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data);
+    }
+    public static org.tensorflow.proto.distruntime.TransportOptions.RecvBufRespExtra parseFrom(
+        java.nio.ByteBuffer data,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data, extensionRegistry);
+    }
+    public static org.tensorflow.proto.distruntime.TransportOptions.RecvBufRespExtra parseFrom(
+        com.google.protobuf.ByteString data)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data);
+    }
+    public static org.tensorflow.proto.distruntime.TransportOptions.RecvBufRespExtra parseFrom(
+        com.google.protobuf.ByteString data,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data, extensionRegistry);
+    }
+    public static org.tensorflow.proto.distruntime.TransportOptions.RecvBufRespExtra parseFrom(byte[] data)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data);
+    }
+    public static org.tensorflow.proto.distruntime.TransportOptions.RecvBufRespExtra parseFrom(
+        byte[] data,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data, extensionRegistry);
+    }
+    public static org.tensorflow.proto.distruntime.TransportOptions.RecvBufRespExtra parseFrom(java.io.InputStream input)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input);
+    }
+    public static org.tensorflow.proto.distruntime.TransportOptions.RecvBufRespExtra parseFrom(
+        java.io.InputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input, extensionRegistry);
+    }
+    public static org.tensorflow.proto.distruntime.TransportOptions.RecvBufRespExtra parseDelimitedFrom(java.io.InputStream input)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseDelimitedWithIOException(PARSER, input);
+    }
+    public static org.tensorflow.proto.distruntime.TransportOptions.RecvBufRespExtra parseDelimitedFrom(
+        java.io.InputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+    }
+    public static org.tensorflow.proto.distruntime.TransportOptions.RecvBufRespExtra parseFrom(
+        com.google.protobuf.CodedInputStream input)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input);
+    }
+    public static org.tensorflow.proto.distruntime.TransportOptions.RecvBufRespExtra parseFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input, extensionRegistry);
+    }
+
+    @java.lang.Override
+    public Builder newBuilderForType() { return newBuilder(); }
+    public static Builder newBuilder() {
+      return DEFAULT_INSTANCE.toBuilder();
+    }
+    public static Builder newBuilder(org.tensorflow.proto.distruntime.TransportOptions.RecvBufRespExtra prototype) {
+      return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+    }
+    @java.lang.Override
+    public Builder toBuilder() {
+      return this == DEFAULT_INSTANCE
+          ? new Builder() : new Builder().mergeFrom(this);
+    }
+
+    @java.lang.Override
+    protected Builder newBuilderForType(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      Builder builder = new Builder(parent);
+      return builder;
+    }
+    /**
+     * <pre>
+     * Extra data needed on a non-RDMA RecvBufResponse.
+     * </pre>
+     *
+     * Protobuf type {@code tensorflow.RecvBufRespExtra}
+     */
+    public static final class Builder extends
+        com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+        // @@protoc_insertion_point(builder_implements:tensorflow.RecvBufRespExtra)
+        org.tensorflow.proto.distruntime.TransportOptions.RecvBufRespExtraOrBuilder {
+      public static final com.google.protobuf.Descriptors.Descriptor
+          getDescriptor() {
+        return org.tensorflow.proto.distruntime.TransportOptions.internal_static_tensorflow_RecvBufRespExtra_descriptor;
+      }
+
+      @java.lang.Override
+      protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+          internalGetFieldAccessorTable() {
+        return org.tensorflow.proto.distruntime.TransportOptions.internal_static_tensorflow_RecvBufRespExtra_fieldAccessorTable
+            .ensureFieldAccessorsInitialized(
+                org.tensorflow.proto.distruntime.TransportOptions.RecvBufRespExtra.class, org.tensorflow.proto.distruntime.TransportOptions.RecvBufRespExtra.Builder.class);
+      }
+
+      // Construct using org.tensorflow.proto.distruntime.TransportOptions.RecvBufRespExtra.newBuilder()
+      private Builder() {
+        maybeForceBuilderInitialization();
+      }
+
+      private Builder(
+          com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+        super(parent);
+        maybeForceBuilderInitialization();
+      }
+      private void maybeForceBuilderInitialization() {
+        if (com.google.protobuf.GeneratedMessageV3
+                .alwaysUseFieldBuilders) {
+        }
+      }
+      @java.lang.Override
+      public Builder clear() {
+        super.clear();
+        tensorContent_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000001);
+        return this;
+      }
+
+      @java.lang.Override
+      public com.google.protobuf.Descriptors.Descriptor
+          getDescriptorForType() {
+        return org.tensorflow.proto.distruntime.TransportOptions.internal_static_tensorflow_RecvBufRespExtra_descriptor;
+      }
+
+      @java.lang.Override
+      public org.tensorflow.proto.distruntime.TransportOptions.RecvBufRespExtra getDefaultInstanceForType() {
+        return org.tensorflow.proto.distruntime.TransportOptions.RecvBufRespExtra.getDefaultInstance();
+      }
+
+      @java.lang.Override
+      public org.tensorflow.proto.distruntime.TransportOptions.RecvBufRespExtra build() {
+        org.tensorflow.proto.distruntime.TransportOptions.RecvBufRespExtra result = buildPartial();
+        if (!result.isInitialized()) {
+          throw newUninitializedMessageException(result);
+        }
+        return result;
+      }
+
+      @java.lang.Override
+      public org.tensorflow.proto.distruntime.TransportOptions.RecvBufRespExtra buildPartial() {
+        org.tensorflow.proto.distruntime.TransportOptions.RecvBufRespExtra result = new org.tensorflow.proto.distruntime.TransportOptions.RecvBufRespExtra(this);
+        int from_bitField0_ = bitField0_;
+        if (((bitField0_ & 0x00000001) != 0)) {
+          tensorContent_ = java.util.Collections.unmodifiableList(tensorContent_);
+          bitField0_ = (bitField0_ & ~0x00000001);
+        }
+        result.tensorContent_ = tensorContent_;
+        onBuilt();
+        return result;
+      }
+
+      @java.lang.Override
+      public Builder clone() {
+        return super.clone();
+      }
+      @java.lang.Override
+      public Builder setField(
+          com.google.protobuf.Descriptors.FieldDescriptor field,
+          java.lang.Object value) {
+        return super.setField(field, value);
+      }
+      @java.lang.Override
+      public Builder clearField(
+          com.google.protobuf.Descriptors.FieldDescriptor field) {
+        return super.clearField(field);
+      }
+      @java.lang.Override
+      public Builder clearOneof(
+          com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+        return super.clearOneof(oneof);
+      }
+      @java.lang.Override
+      public Builder setRepeatedField(
+          com.google.protobuf.Descriptors.FieldDescriptor field,
+          int index, java.lang.Object value) {
+        return super.setRepeatedField(field, index, value);
+      }
+      @java.lang.Override
+      public Builder addRepeatedField(
+          com.google.protobuf.Descriptors.FieldDescriptor field,
+          java.lang.Object value) {
+        return super.addRepeatedField(field, value);
+      }
+      @java.lang.Override
+      public Builder mergeFrom(com.google.protobuf.Message other) {
+        if (other instanceof org.tensorflow.proto.distruntime.TransportOptions.RecvBufRespExtra) {
+          return mergeFrom((org.tensorflow.proto.distruntime.TransportOptions.RecvBufRespExtra)other);
+        } else {
+          super.mergeFrom(other);
+          return this;
+        }
+      }
+
+      public Builder mergeFrom(org.tensorflow.proto.distruntime.TransportOptions.RecvBufRespExtra other) {
+        if (other == org.tensorflow.proto.distruntime.TransportOptions.RecvBufRespExtra.getDefaultInstance()) return this;
+        if (!other.tensorContent_.isEmpty()) {
+          if (tensorContent_.isEmpty()) {
+            tensorContent_ = other.tensorContent_;
+            bitField0_ = (bitField0_ & ~0x00000001);
+          } else {
+            ensureTensorContentIsMutable();
+            tensorContent_.addAll(other.tensorContent_);
+          }
+          onChanged();
+        }
+        this.mergeUnknownFields(other.unknownFields);
+        onChanged();
+        return this;
+      }
+
+      @java.lang.Override
+      public final boolean isInitialized() {
+        return true;
+      }
+
+      @java.lang.Override
+      public Builder mergeFrom(
+          com.google.protobuf.CodedInputStream input,
+          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+          throws java.io.IOException {
+        org.tensorflow.proto.distruntime.TransportOptions.RecvBufRespExtra parsedMessage = null;
+        try {
+          parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+        } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+          parsedMessage = (org.tensorflow.proto.distruntime.TransportOptions.RecvBufRespExtra) e.getUnfinishedMessage();
+          throw e.unwrapIOException();
+        } finally {
+          if (parsedMessage != null) {
+            mergeFrom(parsedMessage);
+          }
+        }
+        return this;
+      }
+      private int bitField0_;
+
+      private java.util.List<com.google.protobuf.ByteString> tensorContent_ = java.util.Collections.emptyList();
+      private void ensureTensorContentIsMutable() {
+        if (!((bitField0_ & 0x00000001) != 0)) {
+          tensorContent_ = new java.util.ArrayList<com.google.protobuf.ByteString>(tensorContent_);
+          bitField0_ |= 0x00000001;
+         }
+      }
+      /**
+       * <code>repeated bytes tensor_content = 1;</code>
+       */
+      public java.util.List<com.google.protobuf.ByteString>
+          getTensorContentList() {
+        return ((bitField0_ & 0x00000001) != 0) ?
+                 java.util.Collections.unmodifiableList(tensorContent_) : tensorContent_;
+      }
+      /**
+       * <code>repeated bytes tensor_content = 1;</code>
+       */
+      public int getTensorContentCount() {
+        return tensorContent_.size();
+      }
+      /**
+       * <code>repeated bytes tensor_content = 1;</code>
+       */
+      public com.google.protobuf.ByteString getTensorContent(int index) {
+        return tensorContent_.get(index);
+      }
+      /**
+       * <code>repeated bytes tensor_content = 1;</code>
+       */
+      public Builder setTensorContent(
+          int index, com.google.protobuf.ByteString value) {
+        if (value == null) {
+    throw new NullPointerException();
+  }
+  ensureTensorContentIsMutable();
+        tensorContent_.set(index, value);
+        onChanged();
+        return this;
+      }
+      /**
+       * <code>repeated bytes tensor_content = 1;</code>
+       */
+      public Builder addTensorContent(com.google.protobuf.ByteString value) {
+        if (value == null) {
+    throw new NullPointerException();
+  }
+  ensureTensorContentIsMutable();
+        tensorContent_.add(value);
+        onChanged();
+        return this;
+      }
+      /**
+       * <code>repeated bytes tensor_content = 1;</code>
+       */
+      public Builder addAllTensorContent(
+          java.lang.Iterable<? extends com.google.protobuf.ByteString> values) {
+        ensureTensorContentIsMutable();
+        com.google.protobuf.AbstractMessageLite.Builder.addAll(
+            values, tensorContent_);
+        onChanged();
+        return this;
+      }
+      /**
+       * <code>repeated bytes tensor_content = 1;</code>
+       */
+      public Builder clearTensorContent() {
+        tensorContent_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000001);
+        onChanged();
+        return this;
+      }
+      @java.lang.Override
+      public final Builder setUnknownFields(
+          final com.google.protobuf.UnknownFieldSet unknownFields) {
+        return super.setUnknownFields(unknownFields);
+      }
+
+      @java.lang.Override
+      public final Builder mergeUnknownFields(
+          final com.google.protobuf.UnknownFieldSet unknownFields) {
+        return super.mergeUnknownFields(unknownFields);
+      }
+
+
+      // @@protoc_insertion_point(builder_scope:tensorflow.RecvBufRespExtra)
+    }
+
+    // @@protoc_insertion_point(class_scope:tensorflow.RecvBufRespExtra)
+    private static final org.tensorflow.proto.distruntime.TransportOptions.RecvBufRespExtra DEFAULT_INSTANCE;
+    static {
+      DEFAULT_INSTANCE = new org.tensorflow.proto.distruntime.TransportOptions.RecvBufRespExtra();
+    }
+
+    public static org.tensorflow.proto.distruntime.TransportOptions.RecvBufRespExtra getDefaultInstance() {
+      return DEFAULT_INSTANCE;
+    }
+
+    private static final com.google.protobuf.Parser<RecvBufRespExtra>
+        PARSER = new com.google.protobuf.AbstractParser<RecvBufRespExtra>() {
+      @java.lang.Override
+      public RecvBufRespExtra parsePartialFrom(
+          com.google.protobuf.CodedInputStream input,
+          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+          throws com.google.protobuf.InvalidProtocolBufferException {
+        return new RecvBufRespExtra(input, extensionRegistry);
+      }
+    };
+
+    public static com.google.protobuf.Parser<RecvBufRespExtra> parser() {
+      return PARSER;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Parser<RecvBufRespExtra> getParserForType() {
+      return PARSER;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.distruntime.TransportOptions.RecvBufRespExtra getDefaultInstanceForType() {
+      return DEFAULT_INSTANCE;
+    }
+
+  }
+
+  private static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_RecvBufRespExtra_descriptor;
+  private static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_RecvBufRespExtra_fieldAccessorTable;
+
+  public static com.google.protobuf.Descriptors.FileDescriptor
+      getDescriptor() {
+    return descriptor;
+  }
+  private static  com.google.protobuf.Descriptors.FileDescriptor
+      descriptor;
+  static {
+    java.lang.String[] descriptorData = {
+      "\n0tensorflow/core/protobuf/transport_opt" +
+      "ions.proto\022\ntensorflow\"*\n\020RecvBufRespExt" +
+      "ra\022\026\n\016tensor_content\030\001 \003(\014Bl\n org.tensor" +
+      "flow.proto.distruntimeZHgithub.com/tenso" +
+      "rflow/tensorflow/tensorflow/go/core/core" +
+      "_protos_go_protob\006proto3"
+    };
+    descriptor = com.google.protobuf.Descriptors.FileDescriptor
+      .internalBuildGeneratedFileFrom(descriptorData,
+        new com.google.protobuf.Descriptors.FileDescriptor[] {
+        });
+    internal_static_tensorflow_RecvBufRespExtra_descriptor =
+      getDescriptor().getMessageTypes().get(0);
+    internal_static_tensorflow_RecvBufRespExtra_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_RecvBufRespExtra_descriptor,
+        new java.lang.String[] { "TensorContent", });
+  }
+
+  // @@protoc_insertion_point(outer_class_scope)
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/BytesList.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/BytesList.java
new file mode 100644
index 00000000000..d8158b11f38
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/BytesList.java
@@ -0,0 +1,572 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/example/feature.proto
+
+package org.tensorflow.proto.example;
+
+/**
+ * <pre>
+ * Containers to hold repeated fundamental values.
+ * </pre>
+ *
+ * Protobuf type {@code tensorflow.BytesList}
+ */
+public  final class BytesList extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.BytesList)
+    BytesListOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use BytesList.newBuilder() to construct.
+  private BytesList(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private BytesList() {
+    value_ = java.util.Collections.emptyList();
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new BytesList();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private BytesList(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    int mutable_bitField0_ = 0;
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 10: {
+            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
+              value_ = new java.util.ArrayList<com.google.protobuf.ByteString>();
+              mutable_bitField0_ |= 0x00000001;
+            }
+            value_.add(input.readBytes());
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      if (((mutable_bitField0_ & 0x00000001) != 0)) {
+        value_ = java.util.Collections.unmodifiableList(value_); // C
+      }
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.example.FeatureProtos.internal_static_tensorflow_BytesList_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.example.FeatureProtos.internal_static_tensorflow_BytesList_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.example.BytesList.class, org.tensorflow.proto.example.BytesList.Builder.class);
+  }
+
+  public static final int VALUE_FIELD_NUMBER = 1;
+  private java.util.List<com.google.protobuf.ByteString> value_;
+  /**
+   * <code>repeated bytes value = 1;</code>
+   */
+  public java.util.List<com.google.protobuf.ByteString>
+      getValueList() {
+    return value_;
+  }
+  /**
+   * <code>repeated bytes value = 1;</code>
+   */
+  public int getValueCount() {
+    return value_.size();
+  }
+  /**
+   * <code>repeated bytes value = 1;</code>
+   */
+  public com.google.protobuf.ByteString getValue(int index) {
+    return value_.get(index);
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    for (int i = 0; i < value_.size(); i++) {
+      output.writeBytes(1, value_.get(i));
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    {
+      int dataSize = 0;
+      for (int i = 0; i < value_.size(); i++) {
+        dataSize += com.google.protobuf.CodedOutputStream
+          .computeBytesSizeNoTag(value_.get(i));
+      }
+      size += dataSize;
+      size += 1 * getValueList().size();
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.example.BytesList)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.example.BytesList other = (org.tensorflow.proto.example.BytesList) obj;
+
+    if (!getValueList()
+        .equals(other.getValueList())) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    if (getValueCount() > 0) {
+      hash = (37 * hash) + VALUE_FIELD_NUMBER;
+      hash = (53 * hash) + getValueList().hashCode();
+    }
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.example.BytesList parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.example.BytesList parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.example.BytesList parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.example.BytesList parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.example.BytesList parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.example.BytesList parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.example.BytesList parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.example.BytesList parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.example.BytesList parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.example.BytesList parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.example.BytesList parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.example.BytesList parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.example.BytesList prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * <pre>
+   * Containers to hold repeated fundamental values.
+   * </pre>
+   *
+   * Protobuf type {@code tensorflow.BytesList}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.BytesList)
+      org.tensorflow.proto.example.BytesListOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.example.FeatureProtos.internal_static_tensorflow_BytesList_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.example.FeatureProtos.internal_static_tensorflow_BytesList_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.example.BytesList.class, org.tensorflow.proto.example.BytesList.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.example.BytesList.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      value_ = java.util.Collections.emptyList();
+      bitField0_ = (bitField0_ & ~0x00000001);
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.example.FeatureProtos.internal_static_tensorflow_BytesList_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.example.BytesList getDefaultInstanceForType() {
+      return org.tensorflow.proto.example.BytesList.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.example.BytesList build() {
+      org.tensorflow.proto.example.BytesList result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.example.BytesList buildPartial() {
+      org.tensorflow.proto.example.BytesList result = new org.tensorflow.proto.example.BytesList(this);
+      int from_bitField0_ = bitField0_;
+      if (((bitField0_ & 0x00000001) != 0)) {
+        value_ = java.util.Collections.unmodifiableList(value_);
+        bitField0_ = (bitField0_ & ~0x00000001);
+      }
+      result.value_ = value_;
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.example.BytesList) {
+        return mergeFrom((org.tensorflow.proto.example.BytesList)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.example.BytesList other) {
+      if (other == org.tensorflow.proto.example.BytesList.getDefaultInstance()) return this;
+      if (!other.value_.isEmpty()) {
+        if (value_.isEmpty()) {
+          value_ = other.value_;
+          bitField0_ = (bitField0_ & ~0x00000001);
+        } else {
+          ensureValueIsMutable();
+          value_.addAll(other.value_);
+        }
+        onChanged();
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.example.BytesList parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.example.BytesList) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+    private int bitField0_;
+
+    private java.util.List<com.google.protobuf.ByteString> value_ = java.util.Collections.emptyList();
+    private void ensureValueIsMutable() {
+      if (!((bitField0_ & 0x00000001) != 0)) {
+        value_ = new java.util.ArrayList<com.google.protobuf.ByteString>(value_);
+        bitField0_ |= 0x00000001;
+       }
+    }
+    /**
+     * <code>repeated bytes value = 1;</code>
+     */
+    public java.util.List<com.google.protobuf.ByteString>
+        getValueList() {
+      return ((bitField0_ & 0x00000001) != 0) ?
+               java.util.Collections.unmodifiableList(value_) : value_;
+    }
+    /**
+     * <code>repeated bytes value = 1;</code>
+     */
+    public int getValueCount() {
+      return value_.size();
+    }
+    /**
+     * <code>repeated bytes value = 1;</code>
+     */
+    public com.google.protobuf.ByteString getValue(int index) {
+      return value_.get(index);
+    }
+    /**
+     * <code>repeated bytes value = 1;</code>
+     */
+    public Builder setValue(
+        int index, com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  ensureValueIsMutable();
+      value_.set(index, value);
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>repeated bytes value = 1;</code>
+     */
+    public Builder addValue(com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  ensureValueIsMutable();
+      value_.add(value);
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>repeated bytes value = 1;</code>
+     */
+    public Builder addAllValue(
+        java.lang.Iterable<? extends com.google.protobuf.ByteString> values) {
+      ensureValueIsMutable();
+      com.google.protobuf.AbstractMessageLite.Builder.addAll(
+          values, value_);
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>repeated bytes value = 1;</code>
+     */
+    public Builder clearValue() {
+      value_ = java.util.Collections.emptyList();
+      bitField0_ = (bitField0_ & ~0x00000001);
+      onChanged();
+      return this;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.BytesList)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.BytesList)
+  private static final org.tensorflow.proto.example.BytesList DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.example.BytesList();
+  }
+
+  public static org.tensorflow.proto.example.BytesList getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<BytesList>
+      PARSER = new com.google.protobuf.AbstractParser<BytesList>() {
+    @java.lang.Override
+    public BytesList parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new BytesList(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<BytesList> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<BytesList> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.example.BytesList getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/BytesListOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/BytesListOrBuilder.java
new file mode 100644
index 00000000000..11151e1a947
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/BytesListOrBuilder.java
@@ -0,0 +1,22 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/example/feature.proto
+
+package org.tensorflow.proto.example;
+
+public interface BytesListOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.BytesList)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <code>repeated bytes value = 1;</code>
+   */
+  java.util.List<com.google.protobuf.ByteString> getValueList();
+  /**
+   * <code>repeated bytes value = 1;</code>
+   */
+  int getValueCount();
+  /**
+   * <code>repeated bytes value = 1;</code>
+   */
+  com.google.protobuf.ByteString getValue(int index);
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/Example.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/Example.java
new file mode 100644
index 00000000000..ea9d058e425
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/Example.java
@@ -0,0 +1,600 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/example/example.proto
+
+package org.tensorflow.proto.example;
+
+/**
+ * Protobuf type {@code tensorflow.Example}
+ */
+public  final class Example extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.Example)
+    ExampleOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use Example.newBuilder() to construct.
+  private Example(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private Example() {
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new Example();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private Example(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 10: {
+            org.tensorflow.proto.example.Features.Builder subBuilder = null;
+            if (features_ != null) {
+              subBuilder = features_.toBuilder();
+            }
+            features_ = input.readMessage(org.tensorflow.proto.example.Features.parser(), extensionRegistry);
+            if (subBuilder != null) {
+              subBuilder.mergeFrom(features_);
+              features_ = subBuilder.buildPartial();
+            }
+
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.example.ExampleProtos.internal_static_tensorflow_Example_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.example.ExampleProtos.internal_static_tensorflow_Example_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.example.Example.class, org.tensorflow.proto.example.Example.Builder.class);
+  }
+
+  public static final int FEATURES_FIELD_NUMBER = 1;
+  private org.tensorflow.proto.example.Features features_;
+  /**
+   * <code>.tensorflow.Features features = 1;</code>
+   */
+  public boolean hasFeatures() {
+    return features_ != null;
+  }
+  /**
+   * <code>.tensorflow.Features features = 1;</code>
+   */
+  public org.tensorflow.proto.example.Features getFeatures() {
+    return features_ == null ? org.tensorflow.proto.example.Features.getDefaultInstance() : features_;
+  }
+  /**
+   * <code>.tensorflow.Features features = 1;</code>
+   */
+  public org.tensorflow.proto.example.FeaturesOrBuilder getFeaturesOrBuilder() {
+    return getFeatures();
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    if (features_ != null) {
+      output.writeMessage(1, getFeatures());
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (features_ != null) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(1, getFeatures());
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.example.Example)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.example.Example other = (org.tensorflow.proto.example.Example) obj;
+
+    if (hasFeatures() != other.hasFeatures()) return false;
+    if (hasFeatures()) {
+      if (!getFeatures()
+          .equals(other.getFeatures())) return false;
+    }
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    if (hasFeatures()) {
+      hash = (37 * hash) + FEATURES_FIELD_NUMBER;
+      hash = (53 * hash) + getFeatures().hashCode();
+    }
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.example.Example parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.example.Example parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.example.Example parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.example.Example parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.example.Example parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.example.Example parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.example.Example parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.example.Example parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.example.Example parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.example.Example parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.example.Example parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.example.Example parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.example.Example prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * Protobuf type {@code tensorflow.Example}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.Example)
+      org.tensorflow.proto.example.ExampleOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.example.ExampleProtos.internal_static_tensorflow_Example_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.example.ExampleProtos.internal_static_tensorflow_Example_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.example.Example.class, org.tensorflow.proto.example.Example.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.example.Example.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      if (featuresBuilder_ == null) {
+        features_ = null;
+      } else {
+        features_ = null;
+        featuresBuilder_ = null;
+      }
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.example.ExampleProtos.internal_static_tensorflow_Example_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.example.Example getDefaultInstanceForType() {
+      return org.tensorflow.proto.example.Example.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.example.Example build() {
+      org.tensorflow.proto.example.Example result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.example.Example buildPartial() {
+      org.tensorflow.proto.example.Example result = new org.tensorflow.proto.example.Example(this);
+      if (featuresBuilder_ == null) {
+        result.features_ = features_;
+      } else {
+        result.features_ = featuresBuilder_.build();
+      }
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.example.Example) {
+        return mergeFrom((org.tensorflow.proto.example.Example)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.example.Example other) {
+      if (other == org.tensorflow.proto.example.Example.getDefaultInstance()) return this;
+      if (other.hasFeatures()) {
+        mergeFeatures(other.getFeatures());
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.example.Example parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.example.Example) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+
+    private org.tensorflow.proto.example.Features features_;
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.example.Features, org.tensorflow.proto.example.Features.Builder, org.tensorflow.proto.example.FeaturesOrBuilder> featuresBuilder_;
+    /**
+     * <code>.tensorflow.Features features = 1;</code>
+     */
+    public boolean hasFeatures() {
+      return featuresBuilder_ != null || features_ != null;
+    }
+    /**
+     * <code>.tensorflow.Features features = 1;</code>
+     */
+    public org.tensorflow.proto.example.Features getFeatures() {
+      if (featuresBuilder_ == null) {
+        return features_ == null ? org.tensorflow.proto.example.Features.getDefaultInstance() : features_;
+      } else {
+        return featuresBuilder_.getMessage();
+      }
+    }
+    /**
+     * <code>.tensorflow.Features features = 1;</code>
+     */
+    public Builder setFeatures(org.tensorflow.proto.example.Features value) {
+      if (featuresBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        features_ = value;
+        onChanged();
+      } else {
+        featuresBuilder_.setMessage(value);
+      }
+
+      return this;
+    }
+    /**
+     * <code>.tensorflow.Features features = 1;</code>
+     */
+    public Builder setFeatures(
+        org.tensorflow.proto.example.Features.Builder builderForValue) {
+      if (featuresBuilder_ == null) {
+        features_ = builderForValue.build();
+        onChanged();
+      } else {
+        featuresBuilder_.setMessage(builderForValue.build());
+      }
+
+      return this;
+    }
+    /**
+     * <code>.tensorflow.Features features = 1;</code>
+     */
+    public Builder mergeFeatures(org.tensorflow.proto.example.Features value) {
+      if (featuresBuilder_ == null) {
+        if (features_ != null) {
+          features_ =
+            org.tensorflow.proto.example.Features.newBuilder(features_).mergeFrom(value).buildPartial();
+        } else {
+          features_ = value;
+        }
+        onChanged();
+      } else {
+        featuresBuilder_.mergeFrom(value);
+      }
+
+      return this;
+    }
+    /**
+     * <code>.tensorflow.Features features = 1;</code>
+     */
+    public Builder clearFeatures() {
+      if (featuresBuilder_ == null) {
+        features_ = null;
+        onChanged();
+      } else {
+        features_ = null;
+        featuresBuilder_ = null;
+      }
+
+      return this;
+    }
+    /**
+     * <code>.tensorflow.Features features = 1;</code>
+     */
+    public org.tensorflow.proto.example.Features.Builder getFeaturesBuilder() {
+      
+      onChanged();
+      return getFeaturesFieldBuilder().getBuilder();
+    }
+    /**
+     * <code>.tensorflow.Features features = 1;</code>
+     */
+    public org.tensorflow.proto.example.FeaturesOrBuilder getFeaturesOrBuilder() {
+      if (featuresBuilder_ != null) {
+        return featuresBuilder_.getMessageOrBuilder();
+      } else {
+        return features_ == null ?
+            org.tensorflow.proto.example.Features.getDefaultInstance() : features_;
+      }
+    }
+    /**
+     * <code>.tensorflow.Features features = 1;</code>
+     */
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.example.Features, org.tensorflow.proto.example.Features.Builder, org.tensorflow.proto.example.FeaturesOrBuilder> 
+        getFeaturesFieldBuilder() {
+      if (featuresBuilder_ == null) {
+        featuresBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+            org.tensorflow.proto.example.Features, org.tensorflow.proto.example.Features.Builder, org.tensorflow.proto.example.FeaturesOrBuilder>(
+                getFeatures(),
+                getParentForChildren(),
+                isClean());
+        features_ = null;
+      }
+      return featuresBuilder_;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.Example)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.Example)
+  private static final org.tensorflow.proto.example.Example DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.example.Example();
+  }
+
+  public static org.tensorflow.proto.example.Example getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<Example>
+      PARSER = new com.google.protobuf.AbstractParser<Example>() {
+    @java.lang.Override
+    public Example parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new Example(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<Example> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<Example> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.example.Example getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/ExampleOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/ExampleOrBuilder.java
new file mode 100644
index 00000000000..2026bc24075
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/ExampleOrBuilder.java
@@ -0,0 +1,22 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/example/example.proto
+
+package org.tensorflow.proto.example;
+
+public interface ExampleOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.Example)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <code>.tensorflow.Features features = 1;</code>
+   */
+  boolean hasFeatures();
+  /**
+   * <code>.tensorflow.Features features = 1;</code>
+   */
+  org.tensorflow.proto.example.Features getFeatures();
+  /**
+   * <code>.tensorflow.Features features = 1;</code>
+   */
+  org.tensorflow.proto.example.FeaturesOrBuilder getFeaturesOrBuilder();
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/ExampleParserConfiguration.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/ExampleParserConfiguration.java
new file mode 100644
index 00000000000..4829d13d8ff
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/ExampleParserConfiguration.java
@@ -0,0 +1,695 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/example/example_parser_configuration.proto
+
+package org.tensorflow.proto.example;
+
+/**
+ * Protobuf type {@code tensorflow.ExampleParserConfiguration}
+ */
+public  final class ExampleParserConfiguration extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.ExampleParserConfiguration)
+    ExampleParserConfigurationOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use ExampleParserConfiguration.newBuilder() to construct.
+  private ExampleParserConfiguration(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private ExampleParserConfiguration() {
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new ExampleParserConfiguration();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private ExampleParserConfiguration(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    int mutable_bitField0_ = 0;
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 10: {
+            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
+              featureMap_ = com.google.protobuf.MapField.newMapField(
+                  FeatureMapDefaultEntryHolder.defaultEntry);
+              mutable_bitField0_ |= 0x00000001;
+            }
+            com.google.protobuf.MapEntry<java.lang.String, org.tensorflow.proto.example.FeatureConfiguration>
+            featureMap__ = input.readMessage(
+                FeatureMapDefaultEntryHolder.defaultEntry.getParserForType(), extensionRegistry);
+            featureMap_.getMutableMap().put(
+                featureMap__.getKey(), featureMap__.getValue());
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.example.ExampleParserConfigurationProtos.internal_static_tensorflow_ExampleParserConfiguration_descriptor;
+  }
+
+  @SuppressWarnings({"rawtypes"})
+  @java.lang.Override
+  protected com.google.protobuf.MapField internalGetMapField(
+      int number) {
+    switch (number) {
+      case 1:
+        return internalGetFeatureMap();
+      default:
+        throw new RuntimeException(
+            "Invalid map field number: " + number);
+    }
+  }
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.example.ExampleParserConfigurationProtos.internal_static_tensorflow_ExampleParserConfiguration_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.example.ExampleParserConfiguration.class, org.tensorflow.proto.example.ExampleParserConfiguration.Builder.class);
+  }
+
+  public static final int FEATURE_MAP_FIELD_NUMBER = 1;
+  private static final class FeatureMapDefaultEntryHolder {
+    static final com.google.protobuf.MapEntry<
+        java.lang.String, org.tensorflow.proto.example.FeatureConfiguration> defaultEntry =
+            com.google.protobuf.MapEntry
+            .<java.lang.String, org.tensorflow.proto.example.FeatureConfiguration>newDefaultInstance(
+                org.tensorflow.proto.example.ExampleParserConfigurationProtos.internal_static_tensorflow_ExampleParserConfiguration_FeatureMapEntry_descriptor, 
+                com.google.protobuf.WireFormat.FieldType.STRING,
+                "",
+                com.google.protobuf.WireFormat.FieldType.MESSAGE,
+                org.tensorflow.proto.example.FeatureConfiguration.getDefaultInstance());
+  }
+  private com.google.protobuf.MapField<
+      java.lang.String, org.tensorflow.proto.example.FeatureConfiguration> featureMap_;
+  private com.google.protobuf.MapField<java.lang.String, org.tensorflow.proto.example.FeatureConfiguration>
+  internalGetFeatureMap() {
+    if (featureMap_ == null) {
+      return com.google.protobuf.MapField.emptyMapField(
+          FeatureMapDefaultEntryHolder.defaultEntry);
+    }
+    return featureMap_;
+  }
+
+  public int getFeatureMapCount() {
+    return internalGetFeatureMap().getMap().size();
+  }
+  /**
+   * <code>map&lt;string, .tensorflow.FeatureConfiguration&gt; feature_map = 1;</code>
+   */
+
+  public boolean containsFeatureMap(
+      java.lang.String key) {
+    if (key == null) { throw new java.lang.NullPointerException(); }
+    return internalGetFeatureMap().getMap().containsKey(key);
+  }
+  /**
+   * Use {@link #getFeatureMapMap()} instead.
+   */
+  @java.lang.Deprecated
+  public java.util.Map<java.lang.String, org.tensorflow.proto.example.FeatureConfiguration> getFeatureMap() {
+    return getFeatureMapMap();
+  }
+  /**
+   * <code>map&lt;string, .tensorflow.FeatureConfiguration&gt; feature_map = 1;</code>
+   */
+
+  public java.util.Map<java.lang.String, org.tensorflow.proto.example.FeatureConfiguration> getFeatureMapMap() {
+    return internalGetFeatureMap().getMap();
+  }
+  /**
+   * <code>map&lt;string, .tensorflow.FeatureConfiguration&gt; feature_map = 1;</code>
+   */
+
+  public org.tensorflow.proto.example.FeatureConfiguration getFeatureMapOrDefault(
+      java.lang.String key,
+      org.tensorflow.proto.example.FeatureConfiguration defaultValue) {
+    if (key == null) { throw new java.lang.NullPointerException(); }
+    java.util.Map<java.lang.String, org.tensorflow.proto.example.FeatureConfiguration> map =
+        internalGetFeatureMap().getMap();
+    return map.containsKey(key) ? map.get(key) : defaultValue;
+  }
+  /**
+   * <code>map&lt;string, .tensorflow.FeatureConfiguration&gt; feature_map = 1;</code>
+   */
+
+  public org.tensorflow.proto.example.FeatureConfiguration getFeatureMapOrThrow(
+      java.lang.String key) {
+    if (key == null) { throw new java.lang.NullPointerException(); }
+    java.util.Map<java.lang.String, org.tensorflow.proto.example.FeatureConfiguration> map =
+        internalGetFeatureMap().getMap();
+    if (!map.containsKey(key)) {
+      throw new java.lang.IllegalArgumentException();
+    }
+    return map.get(key);
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    com.google.protobuf.GeneratedMessageV3
+      .serializeStringMapTo(
+        output,
+        internalGetFeatureMap(),
+        FeatureMapDefaultEntryHolder.defaultEntry,
+        1);
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    for (java.util.Map.Entry<java.lang.String, org.tensorflow.proto.example.FeatureConfiguration> entry
+         : internalGetFeatureMap().getMap().entrySet()) {
+      com.google.protobuf.MapEntry<java.lang.String, org.tensorflow.proto.example.FeatureConfiguration>
+      featureMap__ = FeatureMapDefaultEntryHolder.defaultEntry.newBuilderForType()
+          .setKey(entry.getKey())
+          .setValue(entry.getValue())
+          .build();
+      size += com.google.protobuf.CodedOutputStream
+          .computeMessageSize(1, featureMap__);
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.example.ExampleParserConfiguration)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.example.ExampleParserConfiguration other = (org.tensorflow.proto.example.ExampleParserConfiguration) obj;
+
+    if (!internalGetFeatureMap().equals(
+        other.internalGetFeatureMap())) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    if (!internalGetFeatureMap().getMap().isEmpty()) {
+      hash = (37 * hash) + FEATURE_MAP_FIELD_NUMBER;
+      hash = (53 * hash) + internalGetFeatureMap().hashCode();
+    }
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.example.ExampleParserConfiguration parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.example.ExampleParserConfiguration parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.example.ExampleParserConfiguration parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.example.ExampleParserConfiguration parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.example.ExampleParserConfiguration parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.example.ExampleParserConfiguration parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.example.ExampleParserConfiguration parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.example.ExampleParserConfiguration parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.example.ExampleParserConfiguration parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.example.ExampleParserConfiguration parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.example.ExampleParserConfiguration parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.example.ExampleParserConfiguration parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.example.ExampleParserConfiguration prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * Protobuf type {@code tensorflow.ExampleParserConfiguration}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.ExampleParserConfiguration)
+      org.tensorflow.proto.example.ExampleParserConfigurationOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.example.ExampleParserConfigurationProtos.internal_static_tensorflow_ExampleParserConfiguration_descriptor;
+    }
+
+    @SuppressWarnings({"rawtypes"})
+    protected com.google.protobuf.MapField internalGetMapField(
+        int number) {
+      switch (number) {
+        case 1:
+          return internalGetFeatureMap();
+        default:
+          throw new RuntimeException(
+              "Invalid map field number: " + number);
+      }
+    }
+    @SuppressWarnings({"rawtypes"})
+    protected com.google.protobuf.MapField internalGetMutableMapField(
+        int number) {
+      switch (number) {
+        case 1:
+          return internalGetMutableFeatureMap();
+        default:
+          throw new RuntimeException(
+              "Invalid map field number: " + number);
+      }
+    }
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.example.ExampleParserConfigurationProtos.internal_static_tensorflow_ExampleParserConfiguration_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.example.ExampleParserConfiguration.class, org.tensorflow.proto.example.ExampleParserConfiguration.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.example.ExampleParserConfiguration.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      internalGetMutableFeatureMap().clear();
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.example.ExampleParserConfigurationProtos.internal_static_tensorflow_ExampleParserConfiguration_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.example.ExampleParserConfiguration getDefaultInstanceForType() {
+      return org.tensorflow.proto.example.ExampleParserConfiguration.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.example.ExampleParserConfiguration build() {
+      org.tensorflow.proto.example.ExampleParserConfiguration result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.example.ExampleParserConfiguration buildPartial() {
+      org.tensorflow.proto.example.ExampleParserConfiguration result = new org.tensorflow.proto.example.ExampleParserConfiguration(this);
+      int from_bitField0_ = bitField0_;
+      result.featureMap_ = internalGetFeatureMap();
+      result.featureMap_.makeImmutable();
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.example.ExampleParserConfiguration) {
+        return mergeFrom((org.tensorflow.proto.example.ExampleParserConfiguration)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.example.ExampleParserConfiguration other) {
+      if (other == org.tensorflow.proto.example.ExampleParserConfiguration.getDefaultInstance()) return this;
+      internalGetMutableFeatureMap().mergeFrom(
+          other.internalGetFeatureMap());
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.example.ExampleParserConfiguration parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.example.ExampleParserConfiguration) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+    private int bitField0_;
+
+    private com.google.protobuf.MapField<
+        java.lang.String, org.tensorflow.proto.example.FeatureConfiguration> featureMap_;
+    private com.google.protobuf.MapField<java.lang.String, org.tensorflow.proto.example.FeatureConfiguration>
+    internalGetFeatureMap() {
+      if (featureMap_ == null) {
+        return com.google.protobuf.MapField.emptyMapField(
+            FeatureMapDefaultEntryHolder.defaultEntry);
+      }
+      return featureMap_;
+    }
+    private com.google.protobuf.MapField<java.lang.String, org.tensorflow.proto.example.FeatureConfiguration>
+    internalGetMutableFeatureMap() {
+      onChanged();;
+      if (featureMap_ == null) {
+        featureMap_ = com.google.protobuf.MapField.newMapField(
+            FeatureMapDefaultEntryHolder.defaultEntry);
+      }
+      if (!featureMap_.isMutable()) {
+        featureMap_ = featureMap_.copy();
+      }
+      return featureMap_;
+    }
+
+    public int getFeatureMapCount() {
+      return internalGetFeatureMap().getMap().size();
+    }
+    /**
+     * <code>map&lt;string, .tensorflow.FeatureConfiguration&gt; feature_map = 1;</code>
+     */
+
+    public boolean containsFeatureMap(
+        java.lang.String key) {
+      if (key == null) { throw new java.lang.NullPointerException(); }
+      return internalGetFeatureMap().getMap().containsKey(key);
+    }
+    /**
+     * Use {@link #getFeatureMapMap()} instead.
+     */
+    @java.lang.Deprecated
+    public java.util.Map<java.lang.String, org.tensorflow.proto.example.FeatureConfiguration> getFeatureMap() {
+      return getFeatureMapMap();
+    }
+    /**
+     * <code>map&lt;string, .tensorflow.FeatureConfiguration&gt; feature_map = 1;</code>
+     */
+
+    public java.util.Map<java.lang.String, org.tensorflow.proto.example.FeatureConfiguration> getFeatureMapMap() {
+      return internalGetFeatureMap().getMap();
+    }
+    /**
+     * <code>map&lt;string, .tensorflow.FeatureConfiguration&gt; feature_map = 1;</code>
+     */
+
+    public org.tensorflow.proto.example.FeatureConfiguration getFeatureMapOrDefault(
+        java.lang.String key,
+        org.tensorflow.proto.example.FeatureConfiguration defaultValue) {
+      if (key == null) { throw new java.lang.NullPointerException(); }
+      java.util.Map<java.lang.String, org.tensorflow.proto.example.FeatureConfiguration> map =
+          internalGetFeatureMap().getMap();
+      return map.containsKey(key) ? map.get(key) : defaultValue;
+    }
+    /**
+     * <code>map&lt;string, .tensorflow.FeatureConfiguration&gt; feature_map = 1;</code>
+     */
+
+    public org.tensorflow.proto.example.FeatureConfiguration getFeatureMapOrThrow(
+        java.lang.String key) {
+      if (key == null) { throw new java.lang.NullPointerException(); }
+      java.util.Map<java.lang.String, org.tensorflow.proto.example.FeatureConfiguration> map =
+          internalGetFeatureMap().getMap();
+      if (!map.containsKey(key)) {
+        throw new java.lang.IllegalArgumentException();
+      }
+      return map.get(key);
+    }
+
+    public Builder clearFeatureMap() {
+      internalGetMutableFeatureMap().getMutableMap()
+          .clear();
+      return this;
+    }
+    /**
+     * <code>map&lt;string, .tensorflow.FeatureConfiguration&gt; feature_map = 1;</code>
+     */
+
+    public Builder removeFeatureMap(
+        java.lang.String key) {
+      if (key == null) { throw new java.lang.NullPointerException(); }
+      internalGetMutableFeatureMap().getMutableMap()
+          .remove(key);
+      return this;
+    }
+    /**
+     * Use alternate mutation accessors instead.
+     */
+    @java.lang.Deprecated
+    public java.util.Map<java.lang.String, org.tensorflow.proto.example.FeatureConfiguration>
+    getMutableFeatureMap() {
+      return internalGetMutableFeatureMap().getMutableMap();
+    }
+    /**
+     * <code>map&lt;string, .tensorflow.FeatureConfiguration&gt; feature_map = 1;</code>
+     */
+    public Builder putFeatureMap(
+        java.lang.String key,
+        org.tensorflow.proto.example.FeatureConfiguration value) {
+      if (key == null) { throw new java.lang.NullPointerException(); }
+      if (value == null) { throw new java.lang.NullPointerException(); }
+      internalGetMutableFeatureMap().getMutableMap()
+          .put(key, value);
+      return this;
+    }
+    /**
+     * <code>map&lt;string, .tensorflow.FeatureConfiguration&gt; feature_map = 1;</code>
+     */
+
+    public Builder putAllFeatureMap(
+        java.util.Map<java.lang.String, org.tensorflow.proto.example.FeatureConfiguration> values) {
+      internalGetMutableFeatureMap().getMutableMap()
+          .putAll(values);
+      return this;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.ExampleParserConfiguration)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.ExampleParserConfiguration)
+  private static final org.tensorflow.proto.example.ExampleParserConfiguration DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.example.ExampleParserConfiguration();
+  }
+
+  public static org.tensorflow.proto.example.ExampleParserConfiguration getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<ExampleParserConfiguration>
+      PARSER = new com.google.protobuf.AbstractParser<ExampleParserConfiguration>() {
+    @java.lang.Override
+    public ExampleParserConfiguration parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new ExampleParserConfiguration(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<ExampleParserConfiguration> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<ExampleParserConfiguration> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.example.ExampleParserConfiguration getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/ExampleParserConfigurationOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/ExampleParserConfigurationOrBuilder.java
new file mode 100644
index 00000000000..33df2a31a5b
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/ExampleParserConfigurationOrBuilder.java
@@ -0,0 +1,43 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/example/example_parser_configuration.proto
+
+package org.tensorflow.proto.example;
+
+public interface ExampleParserConfigurationOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.ExampleParserConfiguration)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <code>map&lt;string, .tensorflow.FeatureConfiguration&gt; feature_map = 1;</code>
+   */
+  int getFeatureMapCount();
+  /**
+   * <code>map&lt;string, .tensorflow.FeatureConfiguration&gt; feature_map = 1;</code>
+   */
+  boolean containsFeatureMap(
+      java.lang.String key);
+  /**
+   * Use {@link #getFeatureMapMap()} instead.
+   */
+  @java.lang.Deprecated
+  java.util.Map<java.lang.String, org.tensorflow.proto.example.FeatureConfiguration>
+  getFeatureMap();
+  /**
+   * <code>map&lt;string, .tensorflow.FeatureConfiguration&gt; feature_map = 1;</code>
+   */
+  java.util.Map<java.lang.String, org.tensorflow.proto.example.FeatureConfiguration>
+  getFeatureMapMap();
+  /**
+   * <code>map&lt;string, .tensorflow.FeatureConfiguration&gt; feature_map = 1;</code>
+   */
+
+  org.tensorflow.proto.example.FeatureConfiguration getFeatureMapOrDefault(
+      java.lang.String key,
+      org.tensorflow.proto.example.FeatureConfiguration defaultValue);
+  /**
+   * <code>map&lt;string, .tensorflow.FeatureConfiguration&gt; feature_map = 1;</code>
+   */
+
+  org.tensorflow.proto.example.FeatureConfiguration getFeatureMapOrThrow(
+      java.lang.String key);
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/ExampleParserConfigurationProtos.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/ExampleParserConfigurationProtos.java
new file mode 100644
index 00000000000..9bc45ea35c4
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/ExampleParserConfigurationProtos.java
@@ -0,0 +1,123 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/example/example_parser_configuration.proto
+
+package org.tensorflow.proto.example;
+
+public final class ExampleParserConfigurationProtos {
+  private ExampleParserConfigurationProtos() {}
+  public static void registerAllExtensions(
+      com.google.protobuf.ExtensionRegistryLite registry) {
+  }
+
+  public static void registerAllExtensions(
+      com.google.protobuf.ExtensionRegistry registry) {
+    registerAllExtensions(
+        (com.google.protobuf.ExtensionRegistryLite) registry);
+  }
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_VarLenFeatureProto_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_VarLenFeatureProto_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_FixedLenFeatureProto_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_FixedLenFeatureProto_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_FeatureConfiguration_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_FeatureConfiguration_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_ExampleParserConfiguration_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_ExampleParserConfiguration_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_ExampleParserConfiguration_FeatureMapEntry_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_ExampleParserConfiguration_FeatureMapEntry_fieldAccessorTable;
+
+  public static com.google.protobuf.Descriptors.FileDescriptor
+      getDescriptor() {
+    return descriptor;
+  }
+  private static  com.google.protobuf.Descriptors.FileDescriptor
+      descriptor;
+  static {
+    java.lang.String[] descriptorData = {
+      "\n:tensorflow/core/example/example_parser" +
+      "_configuration.proto\022\ntensorflow\032&tensor" +
+      "flow/core/framework/tensor.proto\032,tensor" +
+      "flow/core/framework/tensor_shape.proto\032%" +
+      "tensorflow/core/framework/types.proto\"\243\001" +
+      "\n\022VarLenFeatureProto\022#\n\005dtype\030\001 \001(\0162\024.te" +
+      "nsorflow.DataType\022!\n\031values_output_tenso" +
+      "r_name\030\002 \001(\t\022\"\n\032indices_output_tensor_na" +
+      "me\030\003 \001(\t\022!\n\031shapes_output_tensor_name\030\004 " +
+      "\001(\t\"\273\001\n\024FixedLenFeatureProto\022#\n\005dtype\030\001 " +
+      "\001(\0162\024.tensorflow.DataType\022+\n\005shape\030\002 \001(\013" +
+      "2\034.tensorflow.TensorShapeProto\022.\n\rdefaul" +
+      "t_value\030\003 \001(\0132\027.tensorflow.TensorProto\022!" +
+      "\n\031values_output_tensor_name\030\004 \001(\t\"\232\001\n\024Fe" +
+      "atureConfiguration\022=\n\021fixed_len_feature\030" +
+      "\001 \001(\0132 .tensorflow.FixedLenFeatureProtoH" +
+      "\000\0229\n\017var_len_feature\030\002 \001(\0132\036.tensorflow." +
+      "VarLenFeatureProtoH\000B\010\n\006config\"\276\001\n\032Examp" +
+      "leParserConfiguration\022K\n\013feature_map\030\001 \003" +
+      "(\01326.tensorflow.ExampleParserConfigurati" +
+      "on.FeatureMapEntry\032S\n\017FeatureMapEntry\022\013\n" +
+      "\003key\030\001 \001(\t\022/\n\005value\030\002 \001(\0132 .tensorflow.F" +
+      "eatureConfiguration:\0028\001B\250\001\n\034org.tensorfl" +
+      "ow.proto.exampleB ExampleParserConfigura" +
+      "tionProtosP\001Zagithub.com/tensorflow/tens" +
+      "orflow/tensorflow/go/core/example/exampl" +
+      "e_parser_configuration_go_proto\370\001\001b\006prot" +
+      "o3"
+    };
+    descriptor = com.google.protobuf.Descriptors.FileDescriptor
+      .internalBuildGeneratedFileFrom(descriptorData,
+        new com.google.protobuf.Descriptors.FileDescriptor[] {
+          org.tensorflow.proto.framework.TensorProtos.getDescriptor(),
+          org.tensorflow.proto.framework.TensorShapeProtos.getDescriptor(),
+          org.tensorflow.proto.framework.TypesProtos.getDescriptor(),
+        });
+    internal_static_tensorflow_VarLenFeatureProto_descriptor =
+      getDescriptor().getMessageTypes().get(0);
+    internal_static_tensorflow_VarLenFeatureProto_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_VarLenFeatureProto_descriptor,
+        new java.lang.String[] { "Dtype", "ValuesOutputTensorName", "IndicesOutputTensorName", "ShapesOutputTensorName", });
+    internal_static_tensorflow_FixedLenFeatureProto_descriptor =
+      getDescriptor().getMessageTypes().get(1);
+    internal_static_tensorflow_FixedLenFeatureProto_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_FixedLenFeatureProto_descriptor,
+        new java.lang.String[] { "Dtype", "Shape", "DefaultValue", "ValuesOutputTensorName", });
+    internal_static_tensorflow_FeatureConfiguration_descriptor =
+      getDescriptor().getMessageTypes().get(2);
+    internal_static_tensorflow_FeatureConfiguration_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_FeatureConfiguration_descriptor,
+        new java.lang.String[] { "FixedLenFeature", "VarLenFeature", "Config", });
+    internal_static_tensorflow_ExampleParserConfiguration_descriptor =
+      getDescriptor().getMessageTypes().get(3);
+    internal_static_tensorflow_ExampleParserConfiguration_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_ExampleParserConfiguration_descriptor,
+        new java.lang.String[] { "FeatureMap", });
+    internal_static_tensorflow_ExampleParserConfiguration_FeatureMapEntry_descriptor =
+      internal_static_tensorflow_ExampleParserConfiguration_descriptor.getNestedTypes().get(0);
+    internal_static_tensorflow_ExampleParserConfiguration_FeatureMapEntry_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_ExampleParserConfiguration_FeatureMapEntry_descriptor,
+        new java.lang.String[] { "Key", "Value", });
+    org.tensorflow.proto.framework.TensorProtos.getDescriptor();
+    org.tensorflow.proto.framework.TensorShapeProtos.getDescriptor();
+    org.tensorflow.proto.framework.TypesProtos.getDescriptor();
+  }
+
+  // @@protoc_insertion_point(outer_class_scope)
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/ExampleProtos.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/ExampleProtos.java
new file mode 100644
index 00000000000..d5494a720c5
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/ExampleProtos.java
@@ -0,0 +1,68 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/example/example.proto
+
+package org.tensorflow.proto.example;
+
+public final class ExampleProtos {
+  private ExampleProtos() {}
+  public static void registerAllExtensions(
+      com.google.protobuf.ExtensionRegistryLite registry) {
+  }
+
+  public static void registerAllExtensions(
+      com.google.protobuf.ExtensionRegistry registry) {
+    registerAllExtensions(
+        (com.google.protobuf.ExtensionRegistryLite) registry);
+  }
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_Example_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_Example_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_SequenceExample_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_SequenceExample_fieldAccessorTable;
+
+  public static com.google.protobuf.Descriptors.FileDescriptor
+      getDescriptor() {
+    return descriptor;
+  }
+  private static  com.google.protobuf.Descriptors.FileDescriptor
+      descriptor;
+  static {
+    java.lang.String[] descriptorData = {
+      "\n%tensorflow/core/example/example.proto\022" +
+      "\ntensorflow\032%tensorflow/core/example/fea" +
+      "ture.proto\"1\n\007Example\022&\n\010features\030\001 \001(\0132" +
+      "\024.tensorflow.Features\"i\n\017SequenceExample" +
+      "\022%\n\007context\030\001 \001(\0132\024.tensorflow.Features\022" +
+      "/\n\rfeature_lists\030\002 \001(\0132\030.tensorflow.Feat" +
+      "ureListsB\207\001\n\034org.tensorflow.proto.exampl" +
+      "eB\rExampleProtosP\001ZSgithub.com/tensorflo" +
+      "w/tensorflow/tensorflow/go/core/example/" +
+      "example_protos_go_proto\370\001\001b\006proto3"
+    };
+    descriptor = com.google.protobuf.Descriptors.FileDescriptor
+      .internalBuildGeneratedFileFrom(descriptorData,
+        new com.google.protobuf.Descriptors.FileDescriptor[] {
+          org.tensorflow.proto.example.FeatureProtos.getDescriptor(),
+        });
+    internal_static_tensorflow_Example_descriptor =
+      getDescriptor().getMessageTypes().get(0);
+    internal_static_tensorflow_Example_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_Example_descriptor,
+        new java.lang.String[] { "Features", });
+    internal_static_tensorflow_SequenceExample_descriptor =
+      getDescriptor().getMessageTypes().get(1);
+    internal_static_tensorflow_SequenceExample_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_SequenceExample_descriptor,
+        new java.lang.String[] { "Context", "FeatureLists", });
+    org.tensorflow.proto.example.FeatureProtos.getDescriptor();
+  }
+
+  // @@protoc_insertion_point(outer_class_scope)
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/Feature.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/Feature.java
new file mode 100644
index 00000000000..a4aa0733f54
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/Feature.java
@@ -0,0 +1,1105 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/example/feature.proto
+
+package org.tensorflow.proto.example;
+
+/**
+ * <pre>
+ * Containers for non-sequential data.
+ * </pre>
+ *
+ * Protobuf type {@code tensorflow.Feature}
+ */
+public  final class Feature extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.Feature)
+    FeatureOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use Feature.newBuilder() to construct.
+  private Feature(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private Feature() {
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new Feature();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private Feature(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 10: {
+            org.tensorflow.proto.example.BytesList.Builder subBuilder = null;
+            if (kindCase_ == 1) {
+              subBuilder = ((org.tensorflow.proto.example.BytesList) kind_).toBuilder();
+            }
+            kind_ =
+                input.readMessage(org.tensorflow.proto.example.BytesList.parser(), extensionRegistry);
+            if (subBuilder != null) {
+              subBuilder.mergeFrom((org.tensorflow.proto.example.BytesList) kind_);
+              kind_ = subBuilder.buildPartial();
+            }
+            kindCase_ = 1;
+            break;
+          }
+          case 18: {
+            org.tensorflow.proto.example.FloatList.Builder subBuilder = null;
+            if (kindCase_ == 2) {
+              subBuilder = ((org.tensorflow.proto.example.FloatList) kind_).toBuilder();
+            }
+            kind_ =
+                input.readMessage(org.tensorflow.proto.example.FloatList.parser(), extensionRegistry);
+            if (subBuilder != null) {
+              subBuilder.mergeFrom((org.tensorflow.proto.example.FloatList) kind_);
+              kind_ = subBuilder.buildPartial();
+            }
+            kindCase_ = 2;
+            break;
+          }
+          case 26: {
+            org.tensorflow.proto.example.Int64List.Builder subBuilder = null;
+            if (kindCase_ == 3) {
+              subBuilder = ((org.tensorflow.proto.example.Int64List) kind_).toBuilder();
+            }
+            kind_ =
+                input.readMessage(org.tensorflow.proto.example.Int64List.parser(), extensionRegistry);
+            if (subBuilder != null) {
+              subBuilder.mergeFrom((org.tensorflow.proto.example.Int64List) kind_);
+              kind_ = subBuilder.buildPartial();
+            }
+            kindCase_ = 3;
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.example.FeatureProtos.internal_static_tensorflow_Feature_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.example.FeatureProtos.internal_static_tensorflow_Feature_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.example.Feature.class, org.tensorflow.proto.example.Feature.Builder.class);
+  }
+
+  private int kindCase_ = 0;
+  private java.lang.Object kind_;
+  public enum KindCase
+      implements com.google.protobuf.Internal.EnumLite {
+    BYTES_LIST(1),
+    FLOAT_LIST(2),
+    INT64_LIST(3),
+    KIND_NOT_SET(0);
+    private final int value;
+    private KindCase(int value) {
+      this.value = value;
+    }
+    /**
+     * @deprecated Use {@link #forNumber(int)} instead.
+     */
+    @java.lang.Deprecated
+    public static KindCase valueOf(int value) {
+      return forNumber(value);
+    }
+
+    public static KindCase forNumber(int value) {
+      switch (value) {
+        case 1: return BYTES_LIST;
+        case 2: return FLOAT_LIST;
+        case 3: return INT64_LIST;
+        case 0: return KIND_NOT_SET;
+        default: return null;
+      }
+    }
+    public int getNumber() {
+      return this.value;
+    }
+  };
+
+  public KindCase
+  getKindCase() {
+    return KindCase.forNumber(
+        kindCase_);
+  }
+
+  public static final int BYTES_LIST_FIELD_NUMBER = 1;
+  /**
+   * <code>.tensorflow.BytesList bytes_list = 1;</code>
+   */
+  public boolean hasBytesList() {
+    return kindCase_ == 1;
+  }
+  /**
+   * <code>.tensorflow.BytesList bytes_list = 1;</code>
+   */
+  public org.tensorflow.proto.example.BytesList getBytesList() {
+    if (kindCase_ == 1) {
+       return (org.tensorflow.proto.example.BytesList) kind_;
+    }
+    return org.tensorflow.proto.example.BytesList.getDefaultInstance();
+  }
+  /**
+   * <code>.tensorflow.BytesList bytes_list = 1;</code>
+   */
+  public org.tensorflow.proto.example.BytesListOrBuilder getBytesListOrBuilder() {
+    if (kindCase_ == 1) {
+       return (org.tensorflow.proto.example.BytesList) kind_;
+    }
+    return org.tensorflow.proto.example.BytesList.getDefaultInstance();
+  }
+
+  public static final int FLOAT_LIST_FIELD_NUMBER = 2;
+  /**
+   * <code>.tensorflow.FloatList float_list = 2;</code>
+   */
+  public boolean hasFloatList() {
+    return kindCase_ == 2;
+  }
+  /**
+   * <code>.tensorflow.FloatList float_list = 2;</code>
+   */
+  public org.tensorflow.proto.example.FloatList getFloatList() {
+    if (kindCase_ == 2) {
+       return (org.tensorflow.proto.example.FloatList) kind_;
+    }
+    return org.tensorflow.proto.example.FloatList.getDefaultInstance();
+  }
+  /**
+   * <code>.tensorflow.FloatList float_list = 2;</code>
+   */
+  public org.tensorflow.proto.example.FloatListOrBuilder getFloatListOrBuilder() {
+    if (kindCase_ == 2) {
+       return (org.tensorflow.proto.example.FloatList) kind_;
+    }
+    return org.tensorflow.proto.example.FloatList.getDefaultInstance();
+  }
+
+  public static final int INT64_LIST_FIELD_NUMBER = 3;
+  /**
+   * <code>.tensorflow.Int64List int64_list = 3;</code>
+   */
+  public boolean hasInt64List() {
+    return kindCase_ == 3;
+  }
+  /**
+   * <code>.tensorflow.Int64List int64_list = 3;</code>
+   */
+  public org.tensorflow.proto.example.Int64List getInt64List() {
+    if (kindCase_ == 3) {
+       return (org.tensorflow.proto.example.Int64List) kind_;
+    }
+    return org.tensorflow.proto.example.Int64List.getDefaultInstance();
+  }
+  /**
+   * <code>.tensorflow.Int64List int64_list = 3;</code>
+   */
+  public org.tensorflow.proto.example.Int64ListOrBuilder getInt64ListOrBuilder() {
+    if (kindCase_ == 3) {
+       return (org.tensorflow.proto.example.Int64List) kind_;
+    }
+    return org.tensorflow.proto.example.Int64List.getDefaultInstance();
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    if (kindCase_ == 1) {
+      output.writeMessage(1, (org.tensorflow.proto.example.BytesList) kind_);
+    }
+    if (kindCase_ == 2) {
+      output.writeMessage(2, (org.tensorflow.proto.example.FloatList) kind_);
+    }
+    if (kindCase_ == 3) {
+      output.writeMessage(3, (org.tensorflow.proto.example.Int64List) kind_);
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (kindCase_ == 1) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(1, (org.tensorflow.proto.example.BytesList) kind_);
+    }
+    if (kindCase_ == 2) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(2, (org.tensorflow.proto.example.FloatList) kind_);
+    }
+    if (kindCase_ == 3) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(3, (org.tensorflow.proto.example.Int64List) kind_);
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.example.Feature)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.example.Feature other = (org.tensorflow.proto.example.Feature) obj;
+
+    if (!getKindCase().equals(other.getKindCase())) return false;
+    switch (kindCase_) {
+      case 1:
+        if (!getBytesList()
+            .equals(other.getBytesList())) return false;
+        break;
+      case 2:
+        if (!getFloatList()
+            .equals(other.getFloatList())) return false;
+        break;
+      case 3:
+        if (!getInt64List()
+            .equals(other.getInt64List())) return false;
+        break;
+      case 0:
+      default:
+    }
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    switch (kindCase_) {
+      case 1:
+        hash = (37 * hash) + BYTES_LIST_FIELD_NUMBER;
+        hash = (53 * hash) + getBytesList().hashCode();
+        break;
+      case 2:
+        hash = (37 * hash) + FLOAT_LIST_FIELD_NUMBER;
+        hash = (53 * hash) + getFloatList().hashCode();
+        break;
+      case 3:
+        hash = (37 * hash) + INT64_LIST_FIELD_NUMBER;
+        hash = (53 * hash) + getInt64List().hashCode();
+        break;
+      case 0:
+      default:
+    }
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.example.Feature parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.example.Feature parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.example.Feature parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.example.Feature parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.example.Feature parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.example.Feature parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.example.Feature parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.example.Feature parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.example.Feature parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.example.Feature parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.example.Feature parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.example.Feature parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.example.Feature prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * <pre>
+   * Containers for non-sequential data.
+   * </pre>
+   *
+   * Protobuf type {@code tensorflow.Feature}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.Feature)
+      org.tensorflow.proto.example.FeatureOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.example.FeatureProtos.internal_static_tensorflow_Feature_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.example.FeatureProtos.internal_static_tensorflow_Feature_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.example.Feature.class, org.tensorflow.proto.example.Feature.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.example.Feature.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      kindCase_ = 0;
+      kind_ = null;
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.example.FeatureProtos.internal_static_tensorflow_Feature_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.example.Feature getDefaultInstanceForType() {
+      return org.tensorflow.proto.example.Feature.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.example.Feature build() {
+      org.tensorflow.proto.example.Feature result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.example.Feature buildPartial() {
+      org.tensorflow.proto.example.Feature result = new org.tensorflow.proto.example.Feature(this);
+      if (kindCase_ == 1) {
+        if (bytesListBuilder_ == null) {
+          result.kind_ = kind_;
+        } else {
+          result.kind_ = bytesListBuilder_.build();
+        }
+      }
+      if (kindCase_ == 2) {
+        if (floatListBuilder_ == null) {
+          result.kind_ = kind_;
+        } else {
+          result.kind_ = floatListBuilder_.build();
+        }
+      }
+      if (kindCase_ == 3) {
+        if (int64ListBuilder_ == null) {
+          result.kind_ = kind_;
+        } else {
+          result.kind_ = int64ListBuilder_.build();
+        }
+      }
+      result.kindCase_ = kindCase_;
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.example.Feature) {
+        return mergeFrom((org.tensorflow.proto.example.Feature)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.example.Feature other) {
+      if (other == org.tensorflow.proto.example.Feature.getDefaultInstance()) return this;
+      switch (other.getKindCase()) {
+        case BYTES_LIST: {
+          mergeBytesList(other.getBytesList());
+          break;
+        }
+        case FLOAT_LIST: {
+          mergeFloatList(other.getFloatList());
+          break;
+        }
+        case INT64_LIST: {
+          mergeInt64List(other.getInt64List());
+          break;
+        }
+        case KIND_NOT_SET: {
+          break;
+        }
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.example.Feature parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.example.Feature) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+    private int kindCase_ = 0;
+    private java.lang.Object kind_;
+    public KindCase
+        getKindCase() {
+      return KindCase.forNumber(
+          kindCase_);
+    }
+
+    public Builder clearKind() {
+      kindCase_ = 0;
+      kind_ = null;
+      onChanged();
+      return this;
+    }
+
+
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.example.BytesList, org.tensorflow.proto.example.BytesList.Builder, org.tensorflow.proto.example.BytesListOrBuilder> bytesListBuilder_;
+    /**
+     * <code>.tensorflow.BytesList bytes_list = 1;</code>
+     */
+    public boolean hasBytesList() {
+      return kindCase_ == 1;
+    }
+    /**
+     * <code>.tensorflow.BytesList bytes_list = 1;</code>
+     */
+    public org.tensorflow.proto.example.BytesList getBytesList() {
+      if (bytesListBuilder_ == null) {
+        if (kindCase_ == 1) {
+          return (org.tensorflow.proto.example.BytesList) kind_;
+        }
+        return org.tensorflow.proto.example.BytesList.getDefaultInstance();
+      } else {
+        if (kindCase_ == 1) {
+          return bytesListBuilder_.getMessage();
+        }
+        return org.tensorflow.proto.example.BytesList.getDefaultInstance();
+      }
+    }
+    /**
+     * <code>.tensorflow.BytesList bytes_list = 1;</code>
+     */
+    public Builder setBytesList(org.tensorflow.proto.example.BytesList value) {
+      if (bytesListBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        kind_ = value;
+        onChanged();
+      } else {
+        bytesListBuilder_.setMessage(value);
+      }
+      kindCase_ = 1;
+      return this;
+    }
+    /**
+     * <code>.tensorflow.BytesList bytes_list = 1;</code>
+     */
+    public Builder setBytesList(
+        org.tensorflow.proto.example.BytesList.Builder builderForValue) {
+      if (bytesListBuilder_ == null) {
+        kind_ = builderForValue.build();
+        onChanged();
+      } else {
+        bytesListBuilder_.setMessage(builderForValue.build());
+      }
+      kindCase_ = 1;
+      return this;
+    }
+    /**
+     * <code>.tensorflow.BytesList bytes_list = 1;</code>
+     */
+    public Builder mergeBytesList(org.tensorflow.proto.example.BytesList value) {
+      if (bytesListBuilder_ == null) {
+        if (kindCase_ == 1 &&
+            kind_ != org.tensorflow.proto.example.BytesList.getDefaultInstance()) {
+          kind_ = org.tensorflow.proto.example.BytesList.newBuilder((org.tensorflow.proto.example.BytesList) kind_)
+              .mergeFrom(value).buildPartial();
+        } else {
+          kind_ = value;
+        }
+        onChanged();
+      } else {
+        if (kindCase_ == 1) {
+          bytesListBuilder_.mergeFrom(value);
+        }
+        bytesListBuilder_.setMessage(value);
+      }
+      kindCase_ = 1;
+      return this;
+    }
+    /**
+     * <code>.tensorflow.BytesList bytes_list = 1;</code>
+     */
+    public Builder clearBytesList() {
+      if (bytesListBuilder_ == null) {
+        if (kindCase_ == 1) {
+          kindCase_ = 0;
+          kind_ = null;
+          onChanged();
+        }
+      } else {
+        if (kindCase_ == 1) {
+          kindCase_ = 0;
+          kind_ = null;
+        }
+        bytesListBuilder_.clear();
+      }
+      return this;
+    }
+    /**
+     * <code>.tensorflow.BytesList bytes_list = 1;</code>
+     */
+    public org.tensorflow.proto.example.BytesList.Builder getBytesListBuilder() {
+      return getBytesListFieldBuilder().getBuilder();
+    }
+    /**
+     * <code>.tensorflow.BytesList bytes_list = 1;</code>
+     */
+    public org.tensorflow.proto.example.BytesListOrBuilder getBytesListOrBuilder() {
+      if ((kindCase_ == 1) && (bytesListBuilder_ != null)) {
+        return bytesListBuilder_.getMessageOrBuilder();
+      } else {
+        if (kindCase_ == 1) {
+          return (org.tensorflow.proto.example.BytesList) kind_;
+        }
+        return org.tensorflow.proto.example.BytesList.getDefaultInstance();
+      }
+    }
+    /**
+     * <code>.tensorflow.BytesList bytes_list = 1;</code>
+     */
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.example.BytesList, org.tensorflow.proto.example.BytesList.Builder, org.tensorflow.proto.example.BytesListOrBuilder> 
+        getBytesListFieldBuilder() {
+      if (bytesListBuilder_ == null) {
+        if (!(kindCase_ == 1)) {
+          kind_ = org.tensorflow.proto.example.BytesList.getDefaultInstance();
+        }
+        bytesListBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+            org.tensorflow.proto.example.BytesList, org.tensorflow.proto.example.BytesList.Builder, org.tensorflow.proto.example.BytesListOrBuilder>(
+                (org.tensorflow.proto.example.BytesList) kind_,
+                getParentForChildren(),
+                isClean());
+        kind_ = null;
+      }
+      kindCase_ = 1;
+      onChanged();;
+      return bytesListBuilder_;
+    }
+
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.example.FloatList, org.tensorflow.proto.example.FloatList.Builder, org.tensorflow.proto.example.FloatListOrBuilder> floatListBuilder_;
+    /**
+     * <code>.tensorflow.FloatList float_list = 2;</code>
+     */
+    public boolean hasFloatList() {
+      return kindCase_ == 2;
+    }
+    /**
+     * <code>.tensorflow.FloatList float_list = 2;</code>
+     */
+    public org.tensorflow.proto.example.FloatList getFloatList() {
+      if (floatListBuilder_ == null) {
+        if (kindCase_ == 2) {
+          return (org.tensorflow.proto.example.FloatList) kind_;
+        }
+        return org.tensorflow.proto.example.FloatList.getDefaultInstance();
+      } else {
+        if (kindCase_ == 2) {
+          return floatListBuilder_.getMessage();
+        }
+        return org.tensorflow.proto.example.FloatList.getDefaultInstance();
+      }
+    }
+    /**
+     * <code>.tensorflow.FloatList float_list = 2;</code>
+     */
+    public Builder setFloatList(org.tensorflow.proto.example.FloatList value) {
+      if (floatListBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        kind_ = value;
+        onChanged();
+      } else {
+        floatListBuilder_.setMessage(value);
+      }
+      kindCase_ = 2;
+      return this;
+    }
+    /**
+     * <code>.tensorflow.FloatList float_list = 2;</code>
+     */
+    public Builder setFloatList(
+        org.tensorflow.proto.example.FloatList.Builder builderForValue) {
+      if (floatListBuilder_ == null) {
+        kind_ = builderForValue.build();
+        onChanged();
+      } else {
+        floatListBuilder_.setMessage(builderForValue.build());
+      }
+      kindCase_ = 2;
+      return this;
+    }
+    /**
+     * <code>.tensorflow.FloatList float_list = 2;</code>
+     */
+    public Builder mergeFloatList(org.tensorflow.proto.example.FloatList value) {
+      if (floatListBuilder_ == null) {
+        if (kindCase_ == 2 &&
+            kind_ != org.tensorflow.proto.example.FloatList.getDefaultInstance()) {
+          kind_ = org.tensorflow.proto.example.FloatList.newBuilder((org.tensorflow.proto.example.FloatList) kind_)
+              .mergeFrom(value).buildPartial();
+        } else {
+          kind_ = value;
+        }
+        onChanged();
+      } else {
+        if (kindCase_ == 2) {
+          floatListBuilder_.mergeFrom(value);
+        }
+        floatListBuilder_.setMessage(value);
+      }
+      kindCase_ = 2;
+      return this;
+    }
+    /**
+     * <code>.tensorflow.FloatList float_list = 2;</code>
+     */
+    public Builder clearFloatList() {
+      if (floatListBuilder_ == null) {
+        if (kindCase_ == 2) {
+          kindCase_ = 0;
+          kind_ = null;
+          onChanged();
+        }
+      } else {
+        if (kindCase_ == 2) {
+          kindCase_ = 0;
+          kind_ = null;
+        }
+        floatListBuilder_.clear();
+      }
+      return this;
+    }
+    /**
+     * <code>.tensorflow.FloatList float_list = 2;</code>
+     */
+    public org.tensorflow.proto.example.FloatList.Builder getFloatListBuilder() {
+      return getFloatListFieldBuilder().getBuilder();
+    }
+    /**
+     * <code>.tensorflow.FloatList float_list = 2;</code>
+     */
+    public org.tensorflow.proto.example.FloatListOrBuilder getFloatListOrBuilder() {
+      if ((kindCase_ == 2) && (floatListBuilder_ != null)) {
+        return floatListBuilder_.getMessageOrBuilder();
+      } else {
+        if (kindCase_ == 2) {
+          return (org.tensorflow.proto.example.FloatList) kind_;
+        }
+        return org.tensorflow.proto.example.FloatList.getDefaultInstance();
+      }
+    }
+    /**
+     * <code>.tensorflow.FloatList float_list = 2;</code>
+     */
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.example.FloatList, org.tensorflow.proto.example.FloatList.Builder, org.tensorflow.proto.example.FloatListOrBuilder> 
+        getFloatListFieldBuilder() {
+      if (floatListBuilder_ == null) {
+        if (!(kindCase_ == 2)) {
+          kind_ = org.tensorflow.proto.example.FloatList.getDefaultInstance();
+        }
+        floatListBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+            org.tensorflow.proto.example.FloatList, org.tensorflow.proto.example.FloatList.Builder, org.tensorflow.proto.example.FloatListOrBuilder>(
+                (org.tensorflow.proto.example.FloatList) kind_,
+                getParentForChildren(),
+                isClean());
+        kind_ = null;
+      }
+      kindCase_ = 2;
+      onChanged();;
+      return floatListBuilder_;
+    }
+
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.example.Int64List, org.tensorflow.proto.example.Int64List.Builder, org.tensorflow.proto.example.Int64ListOrBuilder> int64ListBuilder_;
+    /**
+     * <code>.tensorflow.Int64List int64_list = 3;</code>
+     */
+    public boolean hasInt64List() {
+      return kindCase_ == 3;
+    }
+    /**
+     * <code>.tensorflow.Int64List int64_list = 3;</code>
+     */
+    public org.tensorflow.proto.example.Int64List getInt64List() {
+      if (int64ListBuilder_ == null) {
+        if (kindCase_ == 3) {
+          return (org.tensorflow.proto.example.Int64List) kind_;
+        }
+        return org.tensorflow.proto.example.Int64List.getDefaultInstance();
+      } else {
+        if (kindCase_ == 3) {
+          return int64ListBuilder_.getMessage();
+        }
+        return org.tensorflow.proto.example.Int64List.getDefaultInstance();
+      }
+    }
+    /**
+     * <code>.tensorflow.Int64List int64_list = 3;</code>
+     */
+    public Builder setInt64List(org.tensorflow.proto.example.Int64List value) {
+      if (int64ListBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        kind_ = value;
+        onChanged();
+      } else {
+        int64ListBuilder_.setMessage(value);
+      }
+      kindCase_ = 3;
+      return this;
+    }
+    /**
+     * <code>.tensorflow.Int64List int64_list = 3;</code>
+     */
+    public Builder setInt64List(
+        org.tensorflow.proto.example.Int64List.Builder builderForValue) {
+      if (int64ListBuilder_ == null) {
+        kind_ = builderForValue.build();
+        onChanged();
+      } else {
+        int64ListBuilder_.setMessage(builderForValue.build());
+      }
+      kindCase_ = 3;
+      return this;
+    }
+    /**
+     * <code>.tensorflow.Int64List int64_list = 3;</code>
+     */
+    public Builder mergeInt64List(org.tensorflow.proto.example.Int64List value) {
+      if (int64ListBuilder_ == null) {
+        if (kindCase_ == 3 &&
+            kind_ != org.tensorflow.proto.example.Int64List.getDefaultInstance()) {
+          kind_ = org.tensorflow.proto.example.Int64List.newBuilder((org.tensorflow.proto.example.Int64List) kind_)
+              .mergeFrom(value).buildPartial();
+        } else {
+          kind_ = value;
+        }
+        onChanged();
+      } else {
+        if (kindCase_ == 3) {
+          int64ListBuilder_.mergeFrom(value);
+        }
+        int64ListBuilder_.setMessage(value);
+      }
+      kindCase_ = 3;
+      return this;
+    }
+    /**
+     * <code>.tensorflow.Int64List int64_list = 3;</code>
+     */
+    public Builder clearInt64List() {
+      if (int64ListBuilder_ == null) {
+        if (kindCase_ == 3) {
+          kindCase_ = 0;
+          kind_ = null;
+          onChanged();
+        }
+      } else {
+        if (kindCase_ == 3) {
+          kindCase_ = 0;
+          kind_ = null;
+        }
+        int64ListBuilder_.clear();
+      }
+      return this;
+    }
+    /**
+     * <code>.tensorflow.Int64List int64_list = 3;</code>
+     */
+    public org.tensorflow.proto.example.Int64List.Builder getInt64ListBuilder() {
+      return getInt64ListFieldBuilder().getBuilder();
+    }
+    /**
+     * <code>.tensorflow.Int64List int64_list = 3;</code>
+     */
+    public org.tensorflow.proto.example.Int64ListOrBuilder getInt64ListOrBuilder() {
+      if ((kindCase_ == 3) && (int64ListBuilder_ != null)) {
+        return int64ListBuilder_.getMessageOrBuilder();
+      } else {
+        if (kindCase_ == 3) {
+          return (org.tensorflow.proto.example.Int64List) kind_;
+        }
+        return org.tensorflow.proto.example.Int64List.getDefaultInstance();
+      }
+    }
+    /**
+     * <code>.tensorflow.Int64List int64_list = 3;</code>
+     */
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.example.Int64List, org.tensorflow.proto.example.Int64List.Builder, org.tensorflow.proto.example.Int64ListOrBuilder> 
+        getInt64ListFieldBuilder() {
+      if (int64ListBuilder_ == null) {
+        if (!(kindCase_ == 3)) {
+          kind_ = org.tensorflow.proto.example.Int64List.getDefaultInstance();
+        }
+        int64ListBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+            org.tensorflow.proto.example.Int64List, org.tensorflow.proto.example.Int64List.Builder, org.tensorflow.proto.example.Int64ListOrBuilder>(
+                (org.tensorflow.proto.example.Int64List) kind_,
+                getParentForChildren(),
+                isClean());
+        kind_ = null;
+      }
+      kindCase_ = 3;
+      onChanged();;
+      return int64ListBuilder_;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.Feature)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.Feature)
+  private static final org.tensorflow.proto.example.Feature DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.example.Feature();
+  }
+
+  public static org.tensorflow.proto.example.Feature getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<Feature>
+      PARSER = new com.google.protobuf.AbstractParser<Feature>() {
+    @java.lang.Override
+    public Feature parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new Feature(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<Feature> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<Feature> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.example.Feature getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/FeatureConfiguration.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/FeatureConfiguration.java
new file mode 100644
index 00000000000..6034a97c93e
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/FeatureConfiguration.java
@@ -0,0 +1,893 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/example/example_parser_configuration.proto
+
+package org.tensorflow.proto.example;
+
+/**
+ * Protobuf type {@code tensorflow.FeatureConfiguration}
+ */
+public  final class FeatureConfiguration extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.FeatureConfiguration)
+    FeatureConfigurationOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use FeatureConfiguration.newBuilder() to construct.
+  private FeatureConfiguration(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private FeatureConfiguration() {
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new FeatureConfiguration();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private FeatureConfiguration(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 10: {
+            org.tensorflow.proto.example.FixedLenFeatureProto.Builder subBuilder = null;
+            if (configCase_ == 1) {
+              subBuilder = ((org.tensorflow.proto.example.FixedLenFeatureProto) config_).toBuilder();
+            }
+            config_ =
+                input.readMessage(org.tensorflow.proto.example.FixedLenFeatureProto.parser(), extensionRegistry);
+            if (subBuilder != null) {
+              subBuilder.mergeFrom((org.tensorflow.proto.example.FixedLenFeatureProto) config_);
+              config_ = subBuilder.buildPartial();
+            }
+            configCase_ = 1;
+            break;
+          }
+          case 18: {
+            org.tensorflow.proto.example.VarLenFeatureProto.Builder subBuilder = null;
+            if (configCase_ == 2) {
+              subBuilder = ((org.tensorflow.proto.example.VarLenFeatureProto) config_).toBuilder();
+            }
+            config_ =
+                input.readMessage(org.tensorflow.proto.example.VarLenFeatureProto.parser(), extensionRegistry);
+            if (subBuilder != null) {
+              subBuilder.mergeFrom((org.tensorflow.proto.example.VarLenFeatureProto) config_);
+              config_ = subBuilder.buildPartial();
+            }
+            configCase_ = 2;
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.example.ExampleParserConfigurationProtos.internal_static_tensorflow_FeatureConfiguration_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.example.ExampleParserConfigurationProtos.internal_static_tensorflow_FeatureConfiguration_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.example.FeatureConfiguration.class, org.tensorflow.proto.example.FeatureConfiguration.Builder.class);
+  }
+
+  private int configCase_ = 0;
+  private java.lang.Object config_;
+  public enum ConfigCase
+      implements com.google.protobuf.Internal.EnumLite {
+    FIXED_LEN_FEATURE(1),
+    VAR_LEN_FEATURE(2),
+    CONFIG_NOT_SET(0);
+    private final int value;
+    private ConfigCase(int value) {
+      this.value = value;
+    }
+    /**
+     * @deprecated Use {@link #forNumber(int)} instead.
+     */
+    @java.lang.Deprecated
+    public static ConfigCase valueOf(int value) {
+      return forNumber(value);
+    }
+
+    public static ConfigCase forNumber(int value) {
+      switch (value) {
+        case 1: return FIXED_LEN_FEATURE;
+        case 2: return VAR_LEN_FEATURE;
+        case 0: return CONFIG_NOT_SET;
+        default: return null;
+      }
+    }
+    public int getNumber() {
+      return this.value;
+    }
+  };
+
+  public ConfigCase
+  getConfigCase() {
+    return ConfigCase.forNumber(
+        configCase_);
+  }
+
+  public static final int FIXED_LEN_FEATURE_FIELD_NUMBER = 1;
+  /**
+   * <code>.tensorflow.FixedLenFeatureProto fixed_len_feature = 1;</code>
+   */
+  public boolean hasFixedLenFeature() {
+    return configCase_ == 1;
+  }
+  /**
+   * <code>.tensorflow.FixedLenFeatureProto fixed_len_feature = 1;</code>
+   */
+  public org.tensorflow.proto.example.FixedLenFeatureProto getFixedLenFeature() {
+    if (configCase_ == 1) {
+       return (org.tensorflow.proto.example.FixedLenFeatureProto) config_;
+    }
+    return org.tensorflow.proto.example.FixedLenFeatureProto.getDefaultInstance();
+  }
+  /**
+   * <code>.tensorflow.FixedLenFeatureProto fixed_len_feature = 1;</code>
+   */
+  public org.tensorflow.proto.example.FixedLenFeatureProtoOrBuilder getFixedLenFeatureOrBuilder() {
+    if (configCase_ == 1) {
+       return (org.tensorflow.proto.example.FixedLenFeatureProto) config_;
+    }
+    return org.tensorflow.proto.example.FixedLenFeatureProto.getDefaultInstance();
+  }
+
+  public static final int VAR_LEN_FEATURE_FIELD_NUMBER = 2;
+  /**
+   * <code>.tensorflow.VarLenFeatureProto var_len_feature = 2;</code>
+   */
+  public boolean hasVarLenFeature() {
+    return configCase_ == 2;
+  }
+  /**
+   * <code>.tensorflow.VarLenFeatureProto var_len_feature = 2;</code>
+   */
+  public org.tensorflow.proto.example.VarLenFeatureProto getVarLenFeature() {
+    if (configCase_ == 2) {
+       return (org.tensorflow.proto.example.VarLenFeatureProto) config_;
+    }
+    return org.tensorflow.proto.example.VarLenFeatureProto.getDefaultInstance();
+  }
+  /**
+   * <code>.tensorflow.VarLenFeatureProto var_len_feature = 2;</code>
+   */
+  public org.tensorflow.proto.example.VarLenFeatureProtoOrBuilder getVarLenFeatureOrBuilder() {
+    if (configCase_ == 2) {
+       return (org.tensorflow.proto.example.VarLenFeatureProto) config_;
+    }
+    return org.tensorflow.proto.example.VarLenFeatureProto.getDefaultInstance();
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    if (configCase_ == 1) {
+      output.writeMessage(1, (org.tensorflow.proto.example.FixedLenFeatureProto) config_);
+    }
+    if (configCase_ == 2) {
+      output.writeMessage(2, (org.tensorflow.proto.example.VarLenFeatureProto) config_);
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (configCase_ == 1) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(1, (org.tensorflow.proto.example.FixedLenFeatureProto) config_);
+    }
+    if (configCase_ == 2) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(2, (org.tensorflow.proto.example.VarLenFeatureProto) config_);
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.example.FeatureConfiguration)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.example.FeatureConfiguration other = (org.tensorflow.proto.example.FeatureConfiguration) obj;
+
+    if (!getConfigCase().equals(other.getConfigCase())) return false;
+    switch (configCase_) {
+      case 1:
+        if (!getFixedLenFeature()
+            .equals(other.getFixedLenFeature())) return false;
+        break;
+      case 2:
+        if (!getVarLenFeature()
+            .equals(other.getVarLenFeature())) return false;
+        break;
+      case 0:
+      default:
+    }
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    switch (configCase_) {
+      case 1:
+        hash = (37 * hash) + FIXED_LEN_FEATURE_FIELD_NUMBER;
+        hash = (53 * hash) + getFixedLenFeature().hashCode();
+        break;
+      case 2:
+        hash = (37 * hash) + VAR_LEN_FEATURE_FIELD_NUMBER;
+        hash = (53 * hash) + getVarLenFeature().hashCode();
+        break;
+      case 0:
+      default:
+    }
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.example.FeatureConfiguration parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.example.FeatureConfiguration parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.example.FeatureConfiguration parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.example.FeatureConfiguration parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.example.FeatureConfiguration parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.example.FeatureConfiguration parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.example.FeatureConfiguration parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.example.FeatureConfiguration parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.example.FeatureConfiguration parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.example.FeatureConfiguration parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.example.FeatureConfiguration parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.example.FeatureConfiguration parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.example.FeatureConfiguration prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * Protobuf type {@code tensorflow.FeatureConfiguration}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.FeatureConfiguration)
+      org.tensorflow.proto.example.FeatureConfigurationOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.example.ExampleParserConfigurationProtos.internal_static_tensorflow_FeatureConfiguration_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.example.ExampleParserConfigurationProtos.internal_static_tensorflow_FeatureConfiguration_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.example.FeatureConfiguration.class, org.tensorflow.proto.example.FeatureConfiguration.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.example.FeatureConfiguration.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      configCase_ = 0;
+      config_ = null;
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.example.ExampleParserConfigurationProtos.internal_static_tensorflow_FeatureConfiguration_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.example.FeatureConfiguration getDefaultInstanceForType() {
+      return org.tensorflow.proto.example.FeatureConfiguration.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.example.FeatureConfiguration build() {
+      org.tensorflow.proto.example.FeatureConfiguration result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.example.FeatureConfiguration buildPartial() {
+      org.tensorflow.proto.example.FeatureConfiguration result = new org.tensorflow.proto.example.FeatureConfiguration(this);
+      if (configCase_ == 1) {
+        if (fixedLenFeatureBuilder_ == null) {
+          result.config_ = config_;
+        } else {
+          result.config_ = fixedLenFeatureBuilder_.build();
+        }
+      }
+      if (configCase_ == 2) {
+        if (varLenFeatureBuilder_ == null) {
+          result.config_ = config_;
+        } else {
+          result.config_ = varLenFeatureBuilder_.build();
+        }
+      }
+      result.configCase_ = configCase_;
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.example.FeatureConfiguration) {
+        return mergeFrom((org.tensorflow.proto.example.FeatureConfiguration)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.example.FeatureConfiguration other) {
+      if (other == org.tensorflow.proto.example.FeatureConfiguration.getDefaultInstance()) return this;
+      switch (other.getConfigCase()) {
+        case FIXED_LEN_FEATURE: {
+          mergeFixedLenFeature(other.getFixedLenFeature());
+          break;
+        }
+        case VAR_LEN_FEATURE: {
+          mergeVarLenFeature(other.getVarLenFeature());
+          break;
+        }
+        case CONFIG_NOT_SET: {
+          break;
+        }
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.example.FeatureConfiguration parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.example.FeatureConfiguration) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+    private int configCase_ = 0;
+    private java.lang.Object config_;
+    public ConfigCase
+        getConfigCase() {
+      return ConfigCase.forNumber(
+          configCase_);
+    }
+
+    public Builder clearConfig() {
+      configCase_ = 0;
+      config_ = null;
+      onChanged();
+      return this;
+    }
+
+
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.example.FixedLenFeatureProto, org.tensorflow.proto.example.FixedLenFeatureProto.Builder, org.tensorflow.proto.example.FixedLenFeatureProtoOrBuilder> fixedLenFeatureBuilder_;
+    /**
+     * <code>.tensorflow.FixedLenFeatureProto fixed_len_feature = 1;</code>
+     */
+    public boolean hasFixedLenFeature() {
+      return configCase_ == 1;
+    }
+    /**
+     * <code>.tensorflow.FixedLenFeatureProto fixed_len_feature = 1;</code>
+     */
+    public org.tensorflow.proto.example.FixedLenFeatureProto getFixedLenFeature() {
+      if (fixedLenFeatureBuilder_ == null) {
+        if (configCase_ == 1) {
+          return (org.tensorflow.proto.example.FixedLenFeatureProto) config_;
+        }
+        return org.tensorflow.proto.example.FixedLenFeatureProto.getDefaultInstance();
+      } else {
+        if (configCase_ == 1) {
+          return fixedLenFeatureBuilder_.getMessage();
+        }
+        return org.tensorflow.proto.example.FixedLenFeatureProto.getDefaultInstance();
+      }
+    }
+    /**
+     * <code>.tensorflow.FixedLenFeatureProto fixed_len_feature = 1;</code>
+     */
+    public Builder setFixedLenFeature(org.tensorflow.proto.example.FixedLenFeatureProto value) {
+      if (fixedLenFeatureBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        config_ = value;
+        onChanged();
+      } else {
+        fixedLenFeatureBuilder_.setMessage(value);
+      }
+      configCase_ = 1;
+      return this;
+    }
+    /**
+     * <code>.tensorflow.FixedLenFeatureProto fixed_len_feature = 1;</code>
+     */
+    public Builder setFixedLenFeature(
+        org.tensorflow.proto.example.FixedLenFeatureProto.Builder builderForValue) {
+      if (fixedLenFeatureBuilder_ == null) {
+        config_ = builderForValue.build();
+        onChanged();
+      } else {
+        fixedLenFeatureBuilder_.setMessage(builderForValue.build());
+      }
+      configCase_ = 1;
+      return this;
+    }
+    /**
+     * <code>.tensorflow.FixedLenFeatureProto fixed_len_feature = 1;</code>
+     */
+    public Builder mergeFixedLenFeature(org.tensorflow.proto.example.FixedLenFeatureProto value) {
+      if (fixedLenFeatureBuilder_ == null) {
+        if (configCase_ == 1 &&
+            config_ != org.tensorflow.proto.example.FixedLenFeatureProto.getDefaultInstance()) {
+          config_ = org.tensorflow.proto.example.FixedLenFeatureProto.newBuilder((org.tensorflow.proto.example.FixedLenFeatureProto) config_)
+              .mergeFrom(value).buildPartial();
+        } else {
+          config_ = value;
+        }
+        onChanged();
+      } else {
+        if (configCase_ == 1) {
+          fixedLenFeatureBuilder_.mergeFrom(value);
+        }
+        fixedLenFeatureBuilder_.setMessage(value);
+      }
+      configCase_ = 1;
+      return this;
+    }
+    /**
+     * <code>.tensorflow.FixedLenFeatureProto fixed_len_feature = 1;</code>
+     */
+    public Builder clearFixedLenFeature() {
+      if (fixedLenFeatureBuilder_ == null) {
+        if (configCase_ == 1) {
+          configCase_ = 0;
+          config_ = null;
+          onChanged();
+        }
+      } else {
+        if (configCase_ == 1) {
+          configCase_ = 0;
+          config_ = null;
+        }
+        fixedLenFeatureBuilder_.clear();
+      }
+      return this;
+    }
+    /**
+     * <code>.tensorflow.FixedLenFeatureProto fixed_len_feature = 1;</code>
+     */
+    public org.tensorflow.proto.example.FixedLenFeatureProto.Builder getFixedLenFeatureBuilder() {
+      return getFixedLenFeatureFieldBuilder().getBuilder();
+    }
+    /**
+     * <code>.tensorflow.FixedLenFeatureProto fixed_len_feature = 1;</code>
+     */
+    public org.tensorflow.proto.example.FixedLenFeatureProtoOrBuilder getFixedLenFeatureOrBuilder() {
+      if ((configCase_ == 1) && (fixedLenFeatureBuilder_ != null)) {
+        return fixedLenFeatureBuilder_.getMessageOrBuilder();
+      } else {
+        if (configCase_ == 1) {
+          return (org.tensorflow.proto.example.FixedLenFeatureProto) config_;
+        }
+        return org.tensorflow.proto.example.FixedLenFeatureProto.getDefaultInstance();
+      }
+    }
+    /**
+     * <code>.tensorflow.FixedLenFeatureProto fixed_len_feature = 1;</code>
+     */
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.example.FixedLenFeatureProto, org.tensorflow.proto.example.FixedLenFeatureProto.Builder, org.tensorflow.proto.example.FixedLenFeatureProtoOrBuilder> 
+        getFixedLenFeatureFieldBuilder() {
+      if (fixedLenFeatureBuilder_ == null) {
+        if (!(configCase_ == 1)) {
+          config_ = org.tensorflow.proto.example.FixedLenFeatureProto.getDefaultInstance();
+        }
+        fixedLenFeatureBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+            org.tensorflow.proto.example.FixedLenFeatureProto, org.tensorflow.proto.example.FixedLenFeatureProto.Builder, org.tensorflow.proto.example.FixedLenFeatureProtoOrBuilder>(
+                (org.tensorflow.proto.example.FixedLenFeatureProto) config_,
+                getParentForChildren(),
+                isClean());
+        config_ = null;
+      }
+      configCase_ = 1;
+      onChanged();;
+      return fixedLenFeatureBuilder_;
+    }
+
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.example.VarLenFeatureProto, org.tensorflow.proto.example.VarLenFeatureProto.Builder, org.tensorflow.proto.example.VarLenFeatureProtoOrBuilder> varLenFeatureBuilder_;
+    /**
+     * <code>.tensorflow.VarLenFeatureProto var_len_feature = 2;</code>
+     */
+    public boolean hasVarLenFeature() {
+      return configCase_ == 2;
+    }
+    /**
+     * <code>.tensorflow.VarLenFeatureProto var_len_feature = 2;</code>
+     */
+    public org.tensorflow.proto.example.VarLenFeatureProto getVarLenFeature() {
+      if (varLenFeatureBuilder_ == null) {
+        if (configCase_ == 2) {
+          return (org.tensorflow.proto.example.VarLenFeatureProto) config_;
+        }
+        return org.tensorflow.proto.example.VarLenFeatureProto.getDefaultInstance();
+      } else {
+        if (configCase_ == 2) {
+          return varLenFeatureBuilder_.getMessage();
+        }
+        return org.tensorflow.proto.example.VarLenFeatureProto.getDefaultInstance();
+      }
+    }
+    /**
+     * <code>.tensorflow.VarLenFeatureProto var_len_feature = 2;</code>
+     */
+    public Builder setVarLenFeature(org.tensorflow.proto.example.VarLenFeatureProto value) {
+      if (varLenFeatureBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        config_ = value;
+        onChanged();
+      } else {
+        varLenFeatureBuilder_.setMessage(value);
+      }
+      configCase_ = 2;
+      return this;
+    }
+    /**
+     * <code>.tensorflow.VarLenFeatureProto var_len_feature = 2;</code>
+     */
+    public Builder setVarLenFeature(
+        org.tensorflow.proto.example.VarLenFeatureProto.Builder builderForValue) {
+      if (varLenFeatureBuilder_ == null) {
+        config_ = builderForValue.build();
+        onChanged();
+      } else {
+        varLenFeatureBuilder_.setMessage(builderForValue.build());
+      }
+      configCase_ = 2;
+      return this;
+    }
+    /**
+     * <code>.tensorflow.VarLenFeatureProto var_len_feature = 2;</code>
+     */
+    public Builder mergeVarLenFeature(org.tensorflow.proto.example.VarLenFeatureProto value) {
+      if (varLenFeatureBuilder_ == null) {
+        if (configCase_ == 2 &&
+            config_ != org.tensorflow.proto.example.VarLenFeatureProto.getDefaultInstance()) {
+          config_ = org.tensorflow.proto.example.VarLenFeatureProto.newBuilder((org.tensorflow.proto.example.VarLenFeatureProto) config_)
+              .mergeFrom(value).buildPartial();
+        } else {
+          config_ = value;
+        }
+        onChanged();
+      } else {
+        if (configCase_ == 2) {
+          varLenFeatureBuilder_.mergeFrom(value);
+        }
+        varLenFeatureBuilder_.setMessage(value);
+      }
+      configCase_ = 2;
+      return this;
+    }
+    /**
+     * <code>.tensorflow.VarLenFeatureProto var_len_feature = 2;</code>
+     */
+    public Builder clearVarLenFeature() {
+      if (varLenFeatureBuilder_ == null) {
+        if (configCase_ == 2) {
+          configCase_ = 0;
+          config_ = null;
+          onChanged();
+        }
+      } else {
+        if (configCase_ == 2) {
+          configCase_ = 0;
+          config_ = null;
+        }
+        varLenFeatureBuilder_.clear();
+      }
+      return this;
+    }
+    /**
+     * <code>.tensorflow.VarLenFeatureProto var_len_feature = 2;</code>
+     */
+    public org.tensorflow.proto.example.VarLenFeatureProto.Builder getVarLenFeatureBuilder() {
+      return getVarLenFeatureFieldBuilder().getBuilder();
+    }
+    /**
+     * <code>.tensorflow.VarLenFeatureProto var_len_feature = 2;</code>
+     */
+    public org.tensorflow.proto.example.VarLenFeatureProtoOrBuilder getVarLenFeatureOrBuilder() {
+      if ((configCase_ == 2) && (varLenFeatureBuilder_ != null)) {
+        return varLenFeatureBuilder_.getMessageOrBuilder();
+      } else {
+        if (configCase_ == 2) {
+          return (org.tensorflow.proto.example.VarLenFeatureProto) config_;
+        }
+        return org.tensorflow.proto.example.VarLenFeatureProto.getDefaultInstance();
+      }
+    }
+    /**
+     * <code>.tensorflow.VarLenFeatureProto var_len_feature = 2;</code>
+     */
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.example.VarLenFeatureProto, org.tensorflow.proto.example.VarLenFeatureProto.Builder, org.tensorflow.proto.example.VarLenFeatureProtoOrBuilder> 
+        getVarLenFeatureFieldBuilder() {
+      if (varLenFeatureBuilder_ == null) {
+        if (!(configCase_ == 2)) {
+          config_ = org.tensorflow.proto.example.VarLenFeatureProto.getDefaultInstance();
+        }
+        varLenFeatureBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+            org.tensorflow.proto.example.VarLenFeatureProto, org.tensorflow.proto.example.VarLenFeatureProto.Builder, org.tensorflow.proto.example.VarLenFeatureProtoOrBuilder>(
+                (org.tensorflow.proto.example.VarLenFeatureProto) config_,
+                getParentForChildren(),
+                isClean());
+        config_ = null;
+      }
+      configCase_ = 2;
+      onChanged();;
+      return varLenFeatureBuilder_;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.FeatureConfiguration)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.FeatureConfiguration)
+  private static final org.tensorflow.proto.example.FeatureConfiguration DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.example.FeatureConfiguration();
+  }
+
+  public static org.tensorflow.proto.example.FeatureConfiguration getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<FeatureConfiguration>
+      PARSER = new com.google.protobuf.AbstractParser<FeatureConfiguration>() {
+    @java.lang.Override
+    public FeatureConfiguration parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new FeatureConfiguration(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<FeatureConfiguration> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<FeatureConfiguration> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.example.FeatureConfiguration getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/FeatureConfigurationOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/FeatureConfigurationOrBuilder.java
new file mode 100644
index 00000000000..917025caa25
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/FeatureConfigurationOrBuilder.java
@@ -0,0 +1,37 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/example/example_parser_configuration.proto
+
+package org.tensorflow.proto.example;
+
+public interface FeatureConfigurationOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.FeatureConfiguration)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <code>.tensorflow.FixedLenFeatureProto fixed_len_feature = 1;</code>
+   */
+  boolean hasFixedLenFeature();
+  /**
+   * <code>.tensorflow.FixedLenFeatureProto fixed_len_feature = 1;</code>
+   */
+  org.tensorflow.proto.example.FixedLenFeatureProto getFixedLenFeature();
+  /**
+   * <code>.tensorflow.FixedLenFeatureProto fixed_len_feature = 1;</code>
+   */
+  org.tensorflow.proto.example.FixedLenFeatureProtoOrBuilder getFixedLenFeatureOrBuilder();
+
+  /**
+   * <code>.tensorflow.VarLenFeatureProto var_len_feature = 2;</code>
+   */
+  boolean hasVarLenFeature();
+  /**
+   * <code>.tensorflow.VarLenFeatureProto var_len_feature = 2;</code>
+   */
+  org.tensorflow.proto.example.VarLenFeatureProto getVarLenFeature();
+  /**
+   * <code>.tensorflow.VarLenFeatureProto var_len_feature = 2;</code>
+   */
+  org.tensorflow.proto.example.VarLenFeatureProtoOrBuilder getVarLenFeatureOrBuilder();
+
+  public org.tensorflow.proto.example.FeatureConfiguration.ConfigCase getConfigCase();
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/FeatureList.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/FeatureList.java
new file mode 100644
index 00000000000..6aa9f29864a
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/FeatureList.java
@@ -0,0 +1,781 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/example/feature.proto
+
+package org.tensorflow.proto.example;
+
+/**
+ * <pre>
+ * Containers for sequential data.
+ * A FeatureList contains lists of Features.  These may hold zero or more
+ * Feature values.
+ * FeatureLists are organized into categories by name.  The FeatureLists message
+ * contains the mapping from name to FeatureList.
+ * </pre>
+ *
+ * Protobuf type {@code tensorflow.FeatureList}
+ */
+public  final class FeatureList extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.FeatureList)
+    FeatureListOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use FeatureList.newBuilder() to construct.
+  private FeatureList(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private FeatureList() {
+    feature_ = java.util.Collections.emptyList();
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new FeatureList();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private FeatureList(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    int mutable_bitField0_ = 0;
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 10: {
+            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
+              feature_ = new java.util.ArrayList<org.tensorflow.proto.example.Feature>();
+              mutable_bitField0_ |= 0x00000001;
+            }
+            feature_.add(
+                input.readMessage(org.tensorflow.proto.example.Feature.parser(), extensionRegistry));
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      if (((mutable_bitField0_ & 0x00000001) != 0)) {
+        feature_ = java.util.Collections.unmodifiableList(feature_);
+      }
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.example.FeatureProtos.internal_static_tensorflow_FeatureList_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.example.FeatureProtos.internal_static_tensorflow_FeatureList_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.example.FeatureList.class, org.tensorflow.proto.example.FeatureList.Builder.class);
+  }
+
+  public static final int FEATURE_FIELD_NUMBER = 1;
+  private java.util.List<org.tensorflow.proto.example.Feature> feature_;
+  /**
+   * <code>repeated .tensorflow.Feature feature = 1;</code>
+   */
+  public java.util.List<org.tensorflow.proto.example.Feature> getFeatureList() {
+    return feature_;
+  }
+  /**
+   * <code>repeated .tensorflow.Feature feature = 1;</code>
+   */
+  public java.util.List<? extends org.tensorflow.proto.example.FeatureOrBuilder> 
+      getFeatureOrBuilderList() {
+    return feature_;
+  }
+  /**
+   * <code>repeated .tensorflow.Feature feature = 1;</code>
+   */
+  public int getFeatureCount() {
+    return feature_.size();
+  }
+  /**
+   * <code>repeated .tensorflow.Feature feature = 1;</code>
+   */
+  public org.tensorflow.proto.example.Feature getFeature(int index) {
+    return feature_.get(index);
+  }
+  /**
+   * <code>repeated .tensorflow.Feature feature = 1;</code>
+   */
+  public org.tensorflow.proto.example.FeatureOrBuilder getFeatureOrBuilder(
+      int index) {
+    return feature_.get(index);
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    for (int i = 0; i < feature_.size(); i++) {
+      output.writeMessage(1, feature_.get(i));
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    for (int i = 0; i < feature_.size(); i++) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(1, feature_.get(i));
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.example.FeatureList)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.example.FeatureList other = (org.tensorflow.proto.example.FeatureList) obj;
+
+    if (!getFeatureList()
+        .equals(other.getFeatureList())) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    if (getFeatureCount() > 0) {
+      hash = (37 * hash) + FEATURE_FIELD_NUMBER;
+      hash = (53 * hash) + getFeatureList().hashCode();
+    }
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.example.FeatureList parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.example.FeatureList parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.example.FeatureList parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.example.FeatureList parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.example.FeatureList parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.example.FeatureList parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.example.FeatureList parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.example.FeatureList parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.example.FeatureList parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.example.FeatureList parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.example.FeatureList parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.example.FeatureList parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.example.FeatureList prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * <pre>
+   * Containers for sequential data.
+   * A FeatureList contains lists of Features.  These may hold zero or more
+   * Feature values.
+   * FeatureLists are organized into categories by name.  The FeatureLists message
+   * contains the mapping from name to FeatureList.
+   * </pre>
+   *
+   * Protobuf type {@code tensorflow.FeatureList}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.FeatureList)
+      org.tensorflow.proto.example.FeatureListOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.example.FeatureProtos.internal_static_tensorflow_FeatureList_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.example.FeatureProtos.internal_static_tensorflow_FeatureList_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.example.FeatureList.class, org.tensorflow.proto.example.FeatureList.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.example.FeatureList.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+        getFeatureFieldBuilder();
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      if (featureBuilder_ == null) {
+        feature_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000001);
+      } else {
+        featureBuilder_.clear();
+      }
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.example.FeatureProtos.internal_static_tensorflow_FeatureList_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.example.FeatureList getDefaultInstanceForType() {
+      return org.tensorflow.proto.example.FeatureList.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.example.FeatureList build() {
+      org.tensorflow.proto.example.FeatureList result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.example.FeatureList buildPartial() {
+      org.tensorflow.proto.example.FeatureList result = new org.tensorflow.proto.example.FeatureList(this);
+      int from_bitField0_ = bitField0_;
+      if (featureBuilder_ == null) {
+        if (((bitField0_ & 0x00000001) != 0)) {
+          feature_ = java.util.Collections.unmodifiableList(feature_);
+          bitField0_ = (bitField0_ & ~0x00000001);
+        }
+        result.feature_ = feature_;
+      } else {
+        result.feature_ = featureBuilder_.build();
+      }
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.example.FeatureList) {
+        return mergeFrom((org.tensorflow.proto.example.FeatureList)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.example.FeatureList other) {
+      if (other == org.tensorflow.proto.example.FeatureList.getDefaultInstance()) return this;
+      if (featureBuilder_ == null) {
+        if (!other.feature_.isEmpty()) {
+          if (feature_.isEmpty()) {
+            feature_ = other.feature_;
+            bitField0_ = (bitField0_ & ~0x00000001);
+          } else {
+            ensureFeatureIsMutable();
+            feature_.addAll(other.feature_);
+          }
+          onChanged();
+        }
+      } else {
+        if (!other.feature_.isEmpty()) {
+          if (featureBuilder_.isEmpty()) {
+            featureBuilder_.dispose();
+            featureBuilder_ = null;
+            feature_ = other.feature_;
+            bitField0_ = (bitField0_ & ~0x00000001);
+            featureBuilder_ = 
+              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
+                 getFeatureFieldBuilder() : null;
+          } else {
+            featureBuilder_.addAllMessages(other.feature_);
+          }
+        }
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.example.FeatureList parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.example.FeatureList) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+    private int bitField0_;
+
+    private java.util.List<org.tensorflow.proto.example.Feature> feature_ =
+      java.util.Collections.emptyList();
+    private void ensureFeatureIsMutable() {
+      if (!((bitField0_ & 0x00000001) != 0)) {
+        feature_ = new java.util.ArrayList<org.tensorflow.proto.example.Feature>(feature_);
+        bitField0_ |= 0x00000001;
+       }
+    }
+
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.example.Feature, org.tensorflow.proto.example.Feature.Builder, org.tensorflow.proto.example.FeatureOrBuilder> featureBuilder_;
+
+    /**
+     * <code>repeated .tensorflow.Feature feature = 1;</code>
+     */
+    public java.util.List<org.tensorflow.proto.example.Feature> getFeatureList() {
+      if (featureBuilder_ == null) {
+        return java.util.Collections.unmodifiableList(feature_);
+      } else {
+        return featureBuilder_.getMessageList();
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.Feature feature = 1;</code>
+     */
+    public int getFeatureCount() {
+      if (featureBuilder_ == null) {
+        return feature_.size();
+      } else {
+        return featureBuilder_.getCount();
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.Feature feature = 1;</code>
+     */
+    public org.tensorflow.proto.example.Feature getFeature(int index) {
+      if (featureBuilder_ == null) {
+        return feature_.get(index);
+      } else {
+        return featureBuilder_.getMessage(index);
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.Feature feature = 1;</code>
+     */
+    public Builder setFeature(
+        int index, org.tensorflow.proto.example.Feature value) {
+      if (featureBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureFeatureIsMutable();
+        feature_.set(index, value);
+        onChanged();
+      } else {
+        featureBuilder_.setMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.Feature feature = 1;</code>
+     */
+    public Builder setFeature(
+        int index, org.tensorflow.proto.example.Feature.Builder builderForValue) {
+      if (featureBuilder_ == null) {
+        ensureFeatureIsMutable();
+        feature_.set(index, builderForValue.build());
+        onChanged();
+      } else {
+        featureBuilder_.setMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.Feature feature = 1;</code>
+     */
+    public Builder addFeature(org.tensorflow.proto.example.Feature value) {
+      if (featureBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureFeatureIsMutable();
+        feature_.add(value);
+        onChanged();
+      } else {
+        featureBuilder_.addMessage(value);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.Feature feature = 1;</code>
+     */
+    public Builder addFeature(
+        int index, org.tensorflow.proto.example.Feature value) {
+      if (featureBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureFeatureIsMutable();
+        feature_.add(index, value);
+        onChanged();
+      } else {
+        featureBuilder_.addMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.Feature feature = 1;</code>
+     */
+    public Builder addFeature(
+        org.tensorflow.proto.example.Feature.Builder builderForValue) {
+      if (featureBuilder_ == null) {
+        ensureFeatureIsMutable();
+        feature_.add(builderForValue.build());
+        onChanged();
+      } else {
+        featureBuilder_.addMessage(builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.Feature feature = 1;</code>
+     */
+    public Builder addFeature(
+        int index, org.tensorflow.proto.example.Feature.Builder builderForValue) {
+      if (featureBuilder_ == null) {
+        ensureFeatureIsMutable();
+        feature_.add(index, builderForValue.build());
+        onChanged();
+      } else {
+        featureBuilder_.addMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.Feature feature = 1;</code>
+     */
+    public Builder addAllFeature(
+        java.lang.Iterable<? extends org.tensorflow.proto.example.Feature> values) {
+      if (featureBuilder_ == null) {
+        ensureFeatureIsMutable();
+        com.google.protobuf.AbstractMessageLite.Builder.addAll(
+            values, feature_);
+        onChanged();
+      } else {
+        featureBuilder_.addAllMessages(values);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.Feature feature = 1;</code>
+     */
+    public Builder clearFeature() {
+      if (featureBuilder_ == null) {
+        feature_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000001);
+        onChanged();
+      } else {
+        featureBuilder_.clear();
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.Feature feature = 1;</code>
+     */
+    public Builder removeFeature(int index) {
+      if (featureBuilder_ == null) {
+        ensureFeatureIsMutable();
+        feature_.remove(index);
+        onChanged();
+      } else {
+        featureBuilder_.remove(index);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.Feature feature = 1;</code>
+     */
+    public org.tensorflow.proto.example.Feature.Builder getFeatureBuilder(
+        int index) {
+      return getFeatureFieldBuilder().getBuilder(index);
+    }
+    /**
+     * <code>repeated .tensorflow.Feature feature = 1;</code>
+     */
+    public org.tensorflow.proto.example.FeatureOrBuilder getFeatureOrBuilder(
+        int index) {
+      if (featureBuilder_ == null) {
+        return feature_.get(index);  } else {
+        return featureBuilder_.getMessageOrBuilder(index);
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.Feature feature = 1;</code>
+     */
+    public java.util.List<? extends org.tensorflow.proto.example.FeatureOrBuilder> 
+         getFeatureOrBuilderList() {
+      if (featureBuilder_ != null) {
+        return featureBuilder_.getMessageOrBuilderList();
+      } else {
+        return java.util.Collections.unmodifiableList(feature_);
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.Feature feature = 1;</code>
+     */
+    public org.tensorflow.proto.example.Feature.Builder addFeatureBuilder() {
+      return getFeatureFieldBuilder().addBuilder(
+          org.tensorflow.proto.example.Feature.getDefaultInstance());
+    }
+    /**
+     * <code>repeated .tensorflow.Feature feature = 1;</code>
+     */
+    public org.tensorflow.proto.example.Feature.Builder addFeatureBuilder(
+        int index) {
+      return getFeatureFieldBuilder().addBuilder(
+          index, org.tensorflow.proto.example.Feature.getDefaultInstance());
+    }
+    /**
+     * <code>repeated .tensorflow.Feature feature = 1;</code>
+     */
+    public java.util.List<org.tensorflow.proto.example.Feature.Builder> 
+         getFeatureBuilderList() {
+      return getFeatureFieldBuilder().getBuilderList();
+    }
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.example.Feature, org.tensorflow.proto.example.Feature.Builder, org.tensorflow.proto.example.FeatureOrBuilder> 
+        getFeatureFieldBuilder() {
+      if (featureBuilder_ == null) {
+        featureBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
+            org.tensorflow.proto.example.Feature, org.tensorflow.proto.example.Feature.Builder, org.tensorflow.proto.example.FeatureOrBuilder>(
+                feature_,
+                ((bitField0_ & 0x00000001) != 0),
+                getParentForChildren(),
+                isClean());
+        feature_ = null;
+      }
+      return featureBuilder_;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.FeatureList)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.FeatureList)
+  private static final org.tensorflow.proto.example.FeatureList DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.example.FeatureList();
+  }
+
+  public static org.tensorflow.proto.example.FeatureList getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<FeatureList>
+      PARSER = new com.google.protobuf.AbstractParser<FeatureList>() {
+    @java.lang.Override
+    public FeatureList parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new FeatureList(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<FeatureList> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<FeatureList> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.example.FeatureList getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/FeatureListOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/FeatureListOrBuilder.java
new file mode 100644
index 00000000000..9302c3512db
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/FeatureListOrBuilder.java
@@ -0,0 +1,33 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/example/feature.proto
+
+package org.tensorflow.proto.example;
+
+public interface FeatureListOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.FeatureList)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <code>repeated .tensorflow.Feature feature = 1;</code>
+   */
+  java.util.List<org.tensorflow.proto.example.Feature> 
+      getFeatureList();
+  /**
+   * <code>repeated .tensorflow.Feature feature = 1;</code>
+   */
+  org.tensorflow.proto.example.Feature getFeature(int index);
+  /**
+   * <code>repeated .tensorflow.Feature feature = 1;</code>
+   */
+  int getFeatureCount();
+  /**
+   * <code>repeated .tensorflow.Feature feature = 1;</code>
+   */
+  java.util.List<? extends org.tensorflow.proto.example.FeatureOrBuilder> 
+      getFeatureOrBuilderList();
+  /**
+   * <code>repeated .tensorflow.Feature feature = 1;</code>
+   */
+  org.tensorflow.proto.example.FeatureOrBuilder getFeatureOrBuilder(
+      int index);
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/FeatureLists.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/FeatureLists.java
new file mode 100644
index 00000000000..4b10f2b7d90
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/FeatureLists.java
@@ -0,0 +1,739 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/example/feature.proto
+
+package org.tensorflow.proto.example;
+
+/**
+ * Protobuf type {@code tensorflow.FeatureLists}
+ */
+public  final class FeatureLists extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.FeatureLists)
+    FeatureListsOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use FeatureLists.newBuilder() to construct.
+  private FeatureLists(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private FeatureLists() {
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new FeatureLists();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private FeatureLists(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    int mutable_bitField0_ = 0;
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 10: {
+            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
+              featureList_ = com.google.protobuf.MapField.newMapField(
+                  FeatureListDefaultEntryHolder.defaultEntry);
+              mutable_bitField0_ |= 0x00000001;
+            }
+            com.google.protobuf.MapEntry<java.lang.String, org.tensorflow.proto.example.FeatureList>
+            featureList__ = input.readMessage(
+                FeatureListDefaultEntryHolder.defaultEntry.getParserForType(), extensionRegistry);
+            featureList_.getMutableMap().put(
+                featureList__.getKey(), featureList__.getValue());
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.example.FeatureProtos.internal_static_tensorflow_FeatureLists_descriptor;
+  }
+
+  @SuppressWarnings({"rawtypes"})
+  @java.lang.Override
+  protected com.google.protobuf.MapField internalGetMapField(
+      int number) {
+    switch (number) {
+      case 1:
+        return internalGetFeatureList();
+      default:
+        throw new RuntimeException(
+            "Invalid map field number: " + number);
+    }
+  }
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.example.FeatureProtos.internal_static_tensorflow_FeatureLists_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.example.FeatureLists.class, org.tensorflow.proto.example.FeatureLists.Builder.class);
+  }
+
+  public static final int FEATURE_LIST_FIELD_NUMBER = 1;
+  private static final class FeatureListDefaultEntryHolder {
+    static final com.google.protobuf.MapEntry<
+        java.lang.String, org.tensorflow.proto.example.FeatureList> defaultEntry =
+            com.google.protobuf.MapEntry
+            .<java.lang.String, org.tensorflow.proto.example.FeatureList>newDefaultInstance(
+                org.tensorflow.proto.example.FeatureProtos.internal_static_tensorflow_FeatureLists_FeatureListEntry_descriptor, 
+                com.google.protobuf.WireFormat.FieldType.STRING,
+                "",
+                com.google.protobuf.WireFormat.FieldType.MESSAGE,
+                org.tensorflow.proto.example.FeatureList.getDefaultInstance());
+  }
+  private com.google.protobuf.MapField<
+      java.lang.String, org.tensorflow.proto.example.FeatureList> featureList_;
+  private com.google.protobuf.MapField<java.lang.String, org.tensorflow.proto.example.FeatureList>
+  internalGetFeatureList() {
+    if (featureList_ == null) {
+      return com.google.protobuf.MapField.emptyMapField(
+          FeatureListDefaultEntryHolder.defaultEntry);
+    }
+    return featureList_;
+  }
+
+  public int getFeatureListCount() {
+    return internalGetFeatureList().getMap().size();
+  }
+  /**
+   * <pre>
+   * Map from feature name to feature list.
+   * </pre>
+   *
+   * <code>map&lt;string, .tensorflow.FeatureList&gt; feature_list = 1;</code>
+   */
+
+  public boolean containsFeatureList(
+      java.lang.String key) {
+    if (key == null) { throw new java.lang.NullPointerException(); }
+    return internalGetFeatureList().getMap().containsKey(key);
+  }
+  /**
+   * Use {@link #getFeatureListMap()} instead.
+   */
+  @java.lang.Deprecated
+  public java.util.Map<java.lang.String, org.tensorflow.proto.example.FeatureList> getFeatureList() {
+    return getFeatureListMap();
+  }
+  /**
+   * <pre>
+   * Map from feature name to feature list.
+   * </pre>
+   *
+   * <code>map&lt;string, .tensorflow.FeatureList&gt; feature_list = 1;</code>
+   */
+
+  public java.util.Map<java.lang.String, org.tensorflow.proto.example.FeatureList> getFeatureListMap() {
+    return internalGetFeatureList().getMap();
+  }
+  /**
+   * <pre>
+   * Map from feature name to feature list.
+   * </pre>
+   *
+   * <code>map&lt;string, .tensorflow.FeatureList&gt; feature_list = 1;</code>
+   */
+
+  public org.tensorflow.proto.example.FeatureList getFeatureListOrDefault(
+      java.lang.String key,
+      org.tensorflow.proto.example.FeatureList defaultValue) {
+    if (key == null) { throw new java.lang.NullPointerException(); }
+    java.util.Map<java.lang.String, org.tensorflow.proto.example.FeatureList> map =
+        internalGetFeatureList().getMap();
+    return map.containsKey(key) ? map.get(key) : defaultValue;
+  }
+  /**
+   * <pre>
+   * Map from feature name to feature list.
+   * </pre>
+   *
+   * <code>map&lt;string, .tensorflow.FeatureList&gt; feature_list = 1;</code>
+   */
+
+  public org.tensorflow.proto.example.FeatureList getFeatureListOrThrow(
+      java.lang.String key) {
+    if (key == null) { throw new java.lang.NullPointerException(); }
+    java.util.Map<java.lang.String, org.tensorflow.proto.example.FeatureList> map =
+        internalGetFeatureList().getMap();
+    if (!map.containsKey(key)) {
+      throw new java.lang.IllegalArgumentException();
+    }
+    return map.get(key);
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    com.google.protobuf.GeneratedMessageV3
+      .serializeStringMapTo(
+        output,
+        internalGetFeatureList(),
+        FeatureListDefaultEntryHolder.defaultEntry,
+        1);
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    for (java.util.Map.Entry<java.lang.String, org.tensorflow.proto.example.FeatureList> entry
+         : internalGetFeatureList().getMap().entrySet()) {
+      com.google.protobuf.MapEntry<java.lang.String, org.tensorflow.proto.example.FeatureList>
+      featureList__ = FeatureListDefaultEntryHolder.defaultEntry.newBuilderForType()
+          .setKey(entry.getKey())
+          .setValue(entry.getValue())
+          .build();
+      size += com.google.protobuf.CodedOutputStream
+          .computeMessageSize(1, featureList__);
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.example.FeatureLists)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.example.FeatureLists other = (org.tensorflow.proto.example.FeatureLists) obj;
+
+    if (!internalGetFeatureList().equals(
+        other.internalGetFeatureList())) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    if (!internalGetFeatureList().getMap().isEmpty()) {
+      hash = (37 * hash) + FEATURE_LIST_FIELD_NUMBER;
+      hash = (53 * hash) + internalGetFeatureList().hashCode();
+    }
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.example.FeatureLists parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.example.FeatureLists parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.example.FeatureLists parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.example.FeatureLists parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.example.FeatureLists parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.example.FeatureLists parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.example.FeatureLists parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.example.FeatureLists parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.example.FeatureLists parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.example.FeatureLists parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.example.FeatureLists parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.example.FeatureLists parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.example.FeatureLists prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * Protobuf type {@code tensorflow.FeatureLists}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.FeatureLists)
+      org.tensorflow.proto.example.FeatureListsOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.example.FeatureProtos.internal_static_tensorflow_FeatureLists_descriptor;
+    }
+
+    @SuppressWarnings({"rawtypes"})
+    protected com.google.protobuf.MapField internalGetMapField(
+        int number) {
+      switch (number) {
+        case 1:
+          return internalGetFeatureList();
+        default:
+          throw new RuntimeException(
+              "Invalid map field number: " + number);
+      }
+    }
+    @SuppressWarnings({"rawtypes"})
+    protected com.google.protobuf.MapField internalGetMutableMapField(
+        int number) {
+      switch (number) {
+        case 1:
+          return internalGetMutableFeatureList();
+        default:
+          throw new RuntimeException(
+              "Invalid map field number: " + number);
+      }
+    }
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.example.FeatureProtos.internal_static_tensorflow_FeatureLists_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.example.FeatureLists.class, org.tensorflow.proto.example.FeatureLists.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.example.FeatureLists.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      internalGetMutableFeatureList().clear();
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.example.FeatureProtos.internal_static_tensorflow_FeatureLists_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.example.FeatureLists getDefaultInstanceForType() {
+      return org.tensorflow.proto.example.FeatureLists.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.example.FeatureLists build() {
+      org.tensorflow.proto.example.FeatureLists result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.example.FeatureLists buildPartial() {
+      org.tensorflow.proto.example.FeatureLists result = new org.tensorflow.proto.example.FeatureLists(this);
+      int from_bitField0_ = bitField0_;
+      result.featureList_ = internalGetFeatureList();
+      result.featureList_.makeImmutable();
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.example.FeatureLists) {
+        return mergeFrom((org.tensorflow.proto.example.FeatureLists)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.example.FeatureLists other) {
+      if (other == org.tensorflow.proto.example.FeatureLists.getDefaultInstance()) return this;
+      internalGetMutableFeatureList().mergeFrom(
+          other.internalGetFeatureList());
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.example.FeatureLists parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.example.FeatureLists) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+    private int bitField0_;
+
+    private com.google.protobuf.MapField<
+        java.lang.String, org.tensorflow.proto.example.FeatureList> featureList_;
+    private com.google.protobuf.MapField<java.lang.String, org.tensorflow.proto.example.FeatureList>
+    internalGetFeatureList() {
+      if (featureList_ == null) {
+        return com.google.protobuf.MapField.emptyMapField(
+            FeatureListDefaultEntryHolder.defaultEntry);
+      }
+      return featureList_;
+    }
+    private com.google.protobuf.MapField<java.lang.String, org.tensorflow.proto.example.FeatureList>
+    internalGetMutableFeatureList() {
+      onChanged();;
+      if (featureList_ == null) {
+        featureList_ = com.google.protobuf.MapField.newMapField(
+            FeatureListDefaultEntryHolder.defaultEntry);
+      }
+      if (!featureList_.isMutable()) {
+        featureList_ = featureList_.copy();
+      }
+      return featureList_;
+    }
+
+    public int getFeatureListCount() {
+      return internalGetFeatureList().getMap().size();
+    }
+    /**
+     * <pre>
+     * Map from feature name to feature list.
+     * </pre>
+     *
+     * <code>map&lt;string, .tensorflow.FeatureList&gt; feature_list = 1;</code>
+     */
+
+    public boolean containsFeatureList(
+        java.lang.String key) {
+      if (key == null) { throw new java.lang.NullPointerException(); }
+      return internalGetFeatureList().getMap().containsKey(key);
+    }
+    /**
+     * Use {@link #getFeatureListMap()} instead.
+     */
+    @java.lang.Deprecated
+    public java.util.Map<java.lang.String, org.tensorflow.proto.example.FeatureList> getFeatureList() {
+      return getFeatureListMap();
+    }
+    /**
+     * <pre>
+     * Map from feature name to feature list.
+     * </pre>
+     *
+     * <code>map&lt;string, .tensorflow.FeatureList&gt; feature_list = 1;</code>
+     */
+
+    public java.util.Map<java.lang.String, org.tensorflow.proto.example.FeatureList> getFeatureListMap() {
+      return internalGetFeatureList().getMap();
+    }
+    /**
+     * <pre>
+     * Map from feature name to feature list.
+     * </pre>
+     *
+     * <code>map&lt;string, .tensorflow.FeatureList&gt; feature_list = 1;</code>
+     */
+
+    public org.tensorflow.proto.example.FeatureList getFeatureListOrDefault(
+        java.lang.String key,
+        org.tensorflow.proto.example.FeatureList defaultValue) {
+      if (key == null) { throw new java.lang.NullPointerException(); }
+      java.util.Map<java.lang.String, org.tensorflow.proto.example.FeatureList> map =
+          internalGetFeatureList().getMap();
+      return map.containsKey(key) ? map.get(key) : defaultValue;
+    }
+    /**
+     * <pre>
+     * Map from feature name to feature list.
+     * </pre>
+     *
+     * <code>map&lt;string, .tensorflow.FeatureList&gt; feature_list = 1;</code>
+     */
+
+    public org.tensorflow.proto.example.FeatureList getFeatureListOrThrow(
+        java.lang.String key) {
+      if (key == null) { throw new java.lang.NullPointerException(); }
+      java.util.Map<java.lang.String, org.tensorflow.proto.example.FeatureList> map =
+          internalGetFeatureList().getMap();
+      if (!map.containsKey(key)) {
+        throw new java.lang.IllegalArgumentException();
+      }
+      return map.get(key);
+    }
+
+    public Builder clearFeatureList() {
+      internalGetMutableFeatureList().getMutableMap()
+          .clear();
+      return this;
+    }
+    /**
+     * <pre>
+     * Map from feature name to feature list.
+     * </pre>
+     *
+     * <code>map&lt;string, .tensorflow.FeatureList&gt; feature_list = 1;</code>
+     */
+
+    public Builder removeFeatureList(
+        java.lang.String key) {
+      if (key == null) { throw new java.lang.NullPointerException(); }
+      internalGetMutableFeatureList().getMutableMap()
+          .remove(key);
+      return this;
+    }
+    /**
+     * Use alternate mutation accessors instead.
+     */
+    @java.lang.Deprecated
+    public java.util.Map<java.lang.String, org.tensorflow.proto.example.FeatureList>
+    getMutableFeatureList() {
+      return internalGetMutableFeatureList().getMutableMap();
+    }
+    /**
+     * <pre>
+     * Map from feature name to feature list.
+     * </pre>
+     *
+     * <code>map&lt;string, .tensorflow.FeatureList&gt; feature_list = 1;</code>
+     */
+    public Builder putFeatureList(
+        java.lang.String key,
+        org.tensorflow.proto.example.FeatureList value) {
+      if (key == null) { throw new java.lang.NullPointerException(); }
+      if (value == null) { throw new java.lang.NullPointerException(); }
+      internalGetMutableFeatureList().getMutableMap()
+          .put(key, value);
+      return this;
+    }
+    /**
+     * <pre>
+     * Map from feature name to feature list.
+     * </pre>
+     *
+     * <code>map&lt;string, .tensorflow.FeatureList&gt; feature_list = 1;</code>
+     */
+
+    public Builder putAllFeatureList(
+        java.util.Map<java.lang.String, org.tensorflow.proto.example.FeatureList> values) {
+      internalGetMutableFeatureList().getMutableMap()
+          .putAll(values);
+      return this;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.FeatureLists)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.FeatureLists)
+  private static final org.tensorflow.proto.example.FeatureLists DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.example.FeatureLists();
+  }
+
+  public static org.tensorflow.proto.example.FeatureLists getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<FeatureLists>
+      PARSER = new com.google.protobuf.AbstractParser<FeatureLists>() {
+    @java.lang.Override
+    public FeatureLists parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new FeatureLists(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<FeatureLists> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<FeatureLists> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.example.FeatureLists getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/FeatureListsOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/FeatureListsOrBuilder.java
new file mode 100644
index 00000000000..9f582b998ce
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/FeatureListsOrBuilder.java
@@ -0,0 +1,63 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/example/feature.proto
+
+package org.tensorflow.proto.example;
+
+public interface FeatureListsOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.FeatureLists)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <pre>
+   * Map from feature name to feature list.
+   * </pre>
+   *
+   * <code>map&lt;string, .tensorflow.FeatureList&gt; feature_list = 1;</code>
+   */
+  int getFeatureListCount();
+  /**
+   * <pre>
+   * Map from feature name to feature list.
+   * </pre>
+   *
+   * <code>map&lt;string, .tensorflow.FeatureList&gt; feature_list = 1;</code>
+   */
+  boolean containsFeatureList(
+      java.lang.String key);
+  /**
+   * Use {@link #getFeatureListMap()} instead.
+   */
+  @java.lang.Deprecated
+  java.util.Map<java.lang.String, org.tensorflow.proto.example.FeatureList>
+  getFeatureList();
+  /**
+   * <pre>
+   * Map from feature name to feature list.
+   * </pre>
+   *
+   * <code>map&lt;string, .tensorflow.FeatureList&gt; feature_list = 1;</code>
+   */
+  java.util.Map<java.lang.String, org.tensorflow.proto.example.FeatureList>
+  getFeatureListMap();
+  /**
+   * <pre>
+   * Map from feature name to feature list.
+   * </pre>
+   *
+   * <code>map&lt;string, .tensorflow.FeatureList&gt; feature_list = 1;</code>
+   */
+
+  org.tensorflow.proto.example.FeatureList getFeatureListOrDefault(
+      java.lang.String key,
+      org.tensorflow.proto.example.FeatureList defaultValue);
+  /**
+   * <pre>
+   * Map from feature name to feature list.
+   * </pre>
+   *
+   * <code>map&lt;string, .tensorflow.FeatureList&gt; feature_list = 1;</code>
+   */
+
+  org.tensorflow.proto.example.FeatureList getFeatureListOrThrow(
+      java.lang.String key);
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/FeatureOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/FeatureOrBuilder.java
new file mode 100644
index 00000000000..92242bf9615
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/FeatureOrBuilder.java
@@ -0,0 +1,50 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/example/feature.proto
+
+package org.tensorflow.proto.example;
+
+public interface FeatureOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.Feature)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <code>.tensorflow.BytesList bytes_list = 1;</code>
+   */
+  boolean hasBytesList();
+  /**
+   * <code>.tensorflow.BytesList bytes_list = 1;</code>
+   */
+  org.tensorflow.proto.example.BytesList getBytesList();
+  /**
+   * <code>.tensorflow.BytesList bytes_list = 1;</code>
+   */
+  org.tensorflow.proto.example.BytesListOrBuilder getBytesListOrBuilder();
+
+  /**
+   * <code>.tensorflow.FloatList float_list = 2;</code>
+   */
+  boolean hasFloatList();
+  /**
+   * <code>.tensorflow.FloatList float_list = 2;</code>
+   */
+  org.tensorflow.proto.example.FloatList getFloatList();
+  /**
+   * <code>.tensorflow.FloatList float_list = 2;</code>
+   */
+  org.tensorflow.proto.example.FloatListOrBuilder getFloatListOrBuilder();
+
+  /**
+   * <code>.tensorflow.Int64List int64_list = 3;</code>
+   */
+  boolean hasInt64List();
+  /**
+   * <code>.tensorflow.Int64List int64_list = 3;</code>
+   */
+  org.tensorflow.proto.example.Int64List getInt64List();
+  /**
+   * <code>.tensorflow.Int64List int64_list = 3;</code>
+   */
+  org.tensorflow.proto.example.Int64ListOrBuilder getInt64ListOrBuilder();
+
+  public org.tensorflow.proto.example.Feature.KindCase getKindCase();
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/FeatureProtos.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/FeatureProtos.java
new file mode 100644
index 00000000000..a03e3fecc50
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/FeatureProtos.java
@@ -0,0 +1,153 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/example/feature.proto
+
+package org.tensorflow.proto.example;
+
+public final class FeatureProtos {
+  private FeatureProtos() {}
+  public static void registerAllExtensions(
+      com.google.protobuf.ExtensionRegistryLite registry) {
+  }
+
+  public static void registerAllExtensions(
+      com.google.protobuf.ExtensionRegistry registry) {
+    registerAllExtensions(
+        (com.google.protobuf.ExtensionRegistryLite) registry);
+  }
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_BytesList_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_BytesList_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_FloatList_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_FloatList_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_Int64List_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_Int64List_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_Feature_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_Feature_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_Features_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_Features_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_Features_FeatureEntry_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_Features_FeatureEntry_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_FeatureList_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_FeatureList_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_FeatureLists_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_FeatureLists_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_FeatureLists_FeatureListEntry_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_FeatureLists_FeatureListEntry_fieldAccessorTable;
+
+  public static com.google.protobuf.Descriptors.FileDescriptor
+      getDescriptor() {
+    return descriptor;
+  }
+  private static  com.google.protobuf.Descriptors.FileDescriptor
+      descriptor;
+  static {
+    java.lang.String[] descriptorData = {
+      "\n%tensorflow/core/example/feature.proto\022" +
+      "\ntensorflow\"\032\n\tBytesList\022\r\n\005value\030\001 \003(\014\"" +
+      "\036\n\tFloatList\022\021\n\005value\030\001 \003(\002B\002\020\001\"\036\n\tInt64" +
+      "List\022\021\n\005value\030\001 \003(\003B\002\020\001\"\230\001\n\007Feature\022+\n\nb" +
+      "ytes_list\030\001 \001(\0132\025.tensorflow.BytesListH\000" +
+      "\022+\n\nfloat_list\030\002 \001(\0132\025.tensorflow.FloatL" +
+      "istH\000\022+\n\nint64_list\030\003 \001(\0132\025.tensorflow.I" +
+      "nt64ListH\000B\006\n\004kind\"\203\001\n\010Features\0222\n\007featu" +
+      "re\030\001 \003(\0132!.tensorflow.Features.FeatureEn" +
+      "try\032C\n\014FeatureEntry\022\013\n\003key\030\001 \001(\t\022\"\n\005valu" +
+      "e\030\002 \001(\0132\023.tensorflow.Feature:\0028\001\"3\n\013Feat" +
+      "ureList\022$\n\007feature\030\001 \003(\0132\023.tensorflow.Fe" +
+      "ature\"\234\001\n\014FeatureLists\022?\n\014feature_list\030\001" +
+      " \003(\0132).tensorflow.FeatureLists.FeatureLi" +
+      "stEntry\032K\n\020FeatureListEntry\022\013\n\003key\030\001 \001(\t" +
+      "\022&\n\005value\030\002 \001(\0132\027.tensorflow.FeatureList" +
+      ":\0028\001B\207\001\n\034org.tensorflow.proto.exampleB\rF" +
+      "eatureProtosP\001ZSgithub.com/tensorflow/te" +
+      "nsorflow/tensorflow/go/core/example/exam" +
+      "ple_protos_go_proto\370\001\001b\006proto3"
+    };
+    descriptor = com.google.protobuf.Descriptors.FileDescriptor
+      .internalBuildGeneratedFileFrom(descriptorData,
+        new com.google.protobuf.Descriptors.FileDescriptor[] {
+        });
+    internal_static_tensorflow_BytesList_descriptor =
+      getDescriptor().getMessageTypes().get(0);
+    internal_static_tensorflow_BytesList_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_BytesList_descriptor,
+        new java.lang.String[] { "Value", });
+    internal_static_tensorflow_FloatList_descriptor =
+      getDescriptor().getMessageTypes().get(1);
+    internal_static_tensorflow_FloatList_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_FloatList_descriptor,
+        new java.lang.String[] { "Value", });
+    internal_static_tensorflow_Int64List_descriptor =
+      getDescriptor().getMessageTypes().get(2);
+    internal_static_tensorflow_Int64List_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_Int64List_descriptor,
+        new java.lang.String[] { "Value", });
+    internal_static_tensorflow_Feature_descriptor =
+      getDescriptor().getMessageTypes().get(3);
+    internal_static_tensorflow_Feature_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_Feature_descriptor,
+        new java.lang.String[] { "BytesList", "FloatList", "Int64List", "Kind", });
+    internal_static_tensorflow_Features_descriptor =
+      getDescriptor().getMessageTypes().get(4);
+    internal_static_tensorflow_Features_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_Features_descriptor,
+        new java.lang.String[] { "Feature", });
+    internal_static_tensorflow_Features_FeatureEntry_descriptor =
+      internal_static_tensorflow_Features_descriptor.getNestedTypes().get(0);
+    internal_static_tensorflow_Features_FeatureEntry_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_Features_FeatureEntry_descriptor,
+        new java.lang.String[] { "Key", "Value", });
+    internal_static_tensorflow_FeatureList_descriptor =
+      getDescriptor().getMessageTypes().get(5);
+    internal_static_tensorflow_FeatureList_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_FeatureList_descriptor,
+        new java.lang.String[] { "Feature", });
+    internal_static_tensorflow_FeatureLists_descriptor =
+      getDescriptor().getMessageTypes().get(6);
+    internal_static_tensorflow_FeatureLists_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_FeatureLists_descriptor,
+        new java.lang.String[] { "FeatureList", });
+    internal_static_tensorflow_FeatureLists_FeatureListEntry_descriptor =
+      internal_static_tensorflow_FeatureLists_descriptor.getNestedTypes().get(0);
+    internal_static_tensorflow_FeatureLists_FeatureListEntry_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_FeatureLists_FeatureListEntry_descriptor,
+        new java.lang.String[] { "Key", "Value", });
+  }
+
+  // @@protoc_insertion_point(outer_class_scope)
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/Features.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/Features.java
new file mode 100644
index 00000000000..0e47b04941f
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/Features.java
@@ -0,0 +1,739 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/example/feature.proto
+
+package org.tensorflow.proto.example;
+
+/**
+ * Protobuf type {@code tensorflow.Features}
+ */
+public  final class Features extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.Features)
+    FeaturesOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use Features.newBuilder() to construct.
+  private Features(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private Features() {
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new Features();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private Features(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    int mutable_bitField0_ = 0;
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 10: {
+            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
+              feature_ = com.google.protobuf.MapField.newMapField(
+                  FeatureDefaultEntryHolder.defaultEntry);
+              mutable_bitField0_ |= 0x00000001;
+            }
+            com.google.protobuf.MapEntry<java.lang.String, org.tensorflow.proto.example.Feature>
+            feature__ = input.readMessage(
+                FeatureDefaultEntryHolder.defaultEntry.getParserForType(), extensionRegistry);
+            feature_.getMutableMap().put(
+                feature__.getKey(), feature__.getValue());
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.example.FeatureProtos.internal_static_tensorflow_Features_descriptor;
+  }
+
+  @SuppressWarnings({"rawtypes"})
+  @java.lang.Override
+  protected com.google.protobuf.MapField internalGetMapField(
+      int number) {
+    switch (number) {
+      case 1:
+        return internalGetFeature();
+      default:
+        throw new RuntimeException(
+            "Invalid map field number: " + number);
+    }
+  }
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.example.FeatureProtos.internal_static_tensorflow_Features_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.example.Features.class, org.tensorflow.proto.example.Features.Builder.class);
+  }
+
+  public static final int FEATURE_FIELD_NUMBER = 1;
+  private static final class FeatureDefaultEntryHolder {
+    static final com.google.protobuf.MapEntry<
+        java.lang.String, org.tensorflow.proto.example.Feature> defaultEntry =
+            com.google.protobuf.MapEntry
+            .<java.lang.String, org.tensorflow.proto.example.Feature>newDefaultInstance(
+                org.tensorflow.proto.example.FeatureProtos.internal_static_tensorflow_Features_FeatureEntry_descriptor, 
+                com.google.protobuf.WireFormat.FieldType.STRING,
+                "",
+                com.google.protobuf.WireFormat.FieldType.MESSAGE,
+                org.tensorflow.proto.example.Feature.getDefaultInstance());
+  }
+  private com.google.protobuf.MapField<
+      java.lang.String, org.tensorflow.proto.example.Feature> feature_;
+  private com.google.protobuf.MapField<java.lang.String, org.tensorflow.proto.example.Feature>
+  internalGetFeature() {
+    if (feature_ == null) {
+      return com.google.protobuf.MapField.emptyMapField(
+          FeatureDefaultEntryHolder.defaultEntry);
+    }
+    return feature_;
+  }
+
+  public int getFeatureCount() {
+    return internalGetFeature().getMap().size();
+  }
+  /**
+   * <pre>
+   * Map from feature name to feature.
+   * </pre>
+   *
+   * <code>map&lt;string, .tensorflow.Feature&gt; feature = 1;</code>
+   */
+
+  public boolean containsFeature(
+      java.lang.String key) {
+    if (key == null) { throw new java.lang.NullPointerException(); }
+    return internalGetFeature().getMap().containsKey(key);
+  }
+  /**
+   * Use {@link #getFeatureMap()} instead.
+   */
+  @java.lang.Deprecated
+  public java.util.Map<java.lang.String, org.tensorflow.proto.example.Feature> getFeature() {
+    return getFeatureMap();
+  }
+  /**
+   * <pre>
+   * Map from feature name to feature.
+   * </pre>
+   *
+   * <code>map&lt;string, .tensorflow.Feature&gt; feature = 1;</code>
+   */
+
+  public java.util.Map<java.lang.String, org.tensorflow.proto.example.Feature> getFeatureMap() {
+    return internalGetFeature().getMap();
+  }
+  /**
+   * <pre>
+   * Map from feature name to feature.
+   * </pre>
+   *
+   * <code>map&lt;string, .tensorflow.Feature&gt; feature = 1;</code>
+   */
+
+  public org.tensorflow.proto.example.Feature getFeatureOrDefault(
+      java.lang.String key,
+      org.tensorflow.proto.example.Feature defaultValue) {
+    if (key == null) { throw new java.lang.NullPointerException(); }
+    java.util.Map<java.lang.String, org.tensorflow.proto.example.Feature> map =
+        internalGetFeature().getMap();
+    return map.containsKey(key) ? map.get(key) : defaultValue;
+  }
+  /**
+   * <pre>
+   * Map from feature name to feature.
+   * </pre>
+   *
+   * <code>map&lt;string, .tensorflow.Feature&gt; feature = 1;</code>
+   */
+
+  public org.tensorflow.proto.example.Feature getFeatureOrThrow(
+      java.lang.String key) {
+    if (key == null) { throw new java.lang.NullPointerException(); }
+    java.util.Map<java.lang.String, org.tensorflow.proto.example.Feature> map =
+        internalGetFeature().getMap();
+    if (!map.containsKey(key)) {
+      throw new java.lang.IllegalArgumentException();
+    }
+    return map.get(key);
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    com.google.protobuf.GeneratedMessageV3
+      .serializeStringMapTo(
+        output,
+        internalGetFeature(),
+        FeatureDefaultEntryHolder.defaultEntry,
+        1);
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    for (java.util.Map.Entry<java.lang.String, org.tensorflow.proto.example.Feature> entry
+         : internalGetFeature().getMap().entrySet()) {
+      com.google.protobuf.MapEntry<java.lang.String, org.tensorflow.proto.example.Feature>
+      feature__ = FeatureDefaultEntryHolder.defaultEntry.newBuilderForType()
+          .setKey(entry.getKey())
+          .setValue(entry.getValue())
+          .build();
+      size += com.google.protobuf.CodedOutputStream
+          .computeMessageSize(1, feature__);
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.example.Features)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.example.Features other = (org.tensorflow.proto.example.Features) obj;
+
+    if (!internalGetFeature().equals(
+        other.internalGetFeature())) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    if (!internalGetFeature().getMap().isEmpty()) {
+      hash = (37 * hash) + FEATURE_FIELD_NUMBER;
+      hash = (53 * hash) + internalGetFeature().hashCode();
+    }
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.example.Features parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.example.Features parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.example.Features parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.example.Features parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.example.Features parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.example.Features parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.example.Features parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.example.Features parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.example.Features parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.example.Features parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.example.Features parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.example.Features parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.example.Features prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * Protobuf type {@code tensorflow.Features}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.Features)
+      org.tensorflow.proto.example.FeaturesOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.example.FeatureProtos.internal_static_tensorflow_Features_descriptor;
+    }
+
+    @SuppressWarnings({"rawtypes"})
+    protected com.google.protobuf.MapField internalGetMapField(
+        int number) {
+      switch (number) {
+        case 1:
+          return internalGetFeature();
+        default:
+          throw new RuntimeException(
+              "Invalid map field number: " + number);
+      }
+    }
+    @SuppressWarnings({"rawtypes"})
+    protected com.google.protobuf.MapField internalGetMutableMapField(
+        int number) {
+      switch (number) {
+        case 1:
+          return internalGetMutableFeature();
+        default:
+          throw new RuntimeException(
+              "Invalid map field number: " + number);
+      }
+    }
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.example.FeatureProtos.internal_static_tensorflow_Features_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.example.Features.class, org.tensorflow.proto.example.Features.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.example.Features.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      internalGetMutableFeature().clear();
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.example.FeatureProtos.internal_static_tensorflow_Features_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.example.Features getDefaultInstanceForType() {
+      return org.tensorflow.proto.example.Features.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.example.Features build() {
+      org.tensorflow.proto.example.Features result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.example.Features buildPartial() {
+      org.tensorflow.proto.example.Features result = new org.tensorflow.proto.example.Features(this);
+      int from_bitField0_ = bitField0_;
+      result.feature_ = internalGetFeature();
+      result.feature_.makeImmutable();
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.example.Features) {
+        return mergeFrom((org.tensorflow.proto.example.Features)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.example.Features other) {
+      if (other == org.tensorflow.proto.example.Features.getDefaultInstance()) return this;
+      internalGetMutableFeature().mergeFrom(
+          other.internalGetFeature());
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.example.Features parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.example.Features) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+    private int bitField0_;
+
+    private com.google.protobuf.MapField<
+        java.lang.String, org.tensorflow.proto.example.Feature> feature_;
+    private com.google.protobuf.MapField<java.lang.String, org.tensorflow.proto.example.Feature>
+    internalGetFeature() {
+      if (feature_ == null) {
+        return com.google.protobuf.MapField.emptyMapField(
+            FeatureDefaultEntryHolder.defaultEntry);
+      }
+      return feature_;
+    }
+    private com.google.protobuf.MapField<java.lang.String, org.tensorflow.proto.example.Feature>
+    internalGetMutableFeature() {
+      onChanged();;
+      if (feature_ == null) {
+        feature_ = com.google.protobuf.MapField.newMapField(
+            FeatureDefaultEntryHolder.defaultEntry);
+      }
+      if (!feature_.isMutable()) {
+        feature_ = feature_.copy();
+      }
+      return feature_;
+    }
+
+    public int getFeatureCount() {
+      return internalGetFeature().getMap().size();
+    }
+    /**
+     * <pre>
+     * Map from feature name to feature.
+     * </pre>
+     *
+     * <code>map&lt;string, .tensorflow.Feature&gt; feature = 1;</code>
+     */
+
+    public boolean containsFeature(
+        java.lang.String key) {
+      if (key == null) { throw new java.lang.NullPointerException(); }
+      return internalGetFeature().getMap().containsKey(key);
+    }
+    /**
+     * Use {@link #getFeatureMap()} instead.
+     */
+    @java.lang.Deprecated
+    public java.util.Map<java.lang.String, org.tensorflow.proto.example.Feature> getFeature() {
+      return getFeatureMap();
+    }
+    /**
+     * <pre>
+     * Map from feature name to feature.
+     * </pre>
+     *
+     * <code>map&lt;string, .tensorflow.Feature&gt; feature = 1;</code>
+     */
+
+    public java.util.Map<java.lang.String, org.tensorflow.proto.example.Feature> getFeatureMap() {
+      return internalGetFeature().getMap();
+    }
+    /**
+     * <pre>
+     * Map from feature name to feature.
+     * </pre>
+     *
+     * <code>map&lt;string, .tensorflow.Feature&gt; feature = 1;</code>
+     */
+
+    public org.tensorflow.proto.example.Feature getFeatureOrDefault(
+        java.lang.String key,
+        org.tensorflow.proto.example.Feature defaultValue) {
+      if (key == null) { throw new java.lang.NullPointerException(); }
+      java.util.Map<java.lang.String, org.tensorflow.proto.example.Feature> map =
+          internalGetFeature().getMap();
+      return map.containsKey(key) ? map.get(key) : defaultValue;
+    }
+    /**
+     * <pre>
+     * Map from feature name to feature.
+     * </pre>
+     *
+     * <code>map&lt;string, .tensorflow.Feature&gt; feature = 1;</code>
+     */
+
+    public org.tensorflow.proto.example.Feature getFeatureOrThrow(
+        java.lang.String key) {
+      if (key == null) { throw new java.lang.NullPointerException(); }
+      java.util.Map<java.lang.String, org.tensorflow.proto.example.Feature> map =
+          internalGetFeature().getMap();
+      if (!map.containsKey(key)) {
+        throw new java.lang.IllegalArgumentException();
+      }
+      return map.get(key);
+    }
+
+    public Builder clearFeature() {
+      internalGetMutableFeature().getMutableMap()
+          .clear();
+      return this;
+    }
+    /**
+     * <pre>
+     * Map from feature name to feature.
+     * </pre>
+     *
+     * <code>map&lt;string, .tensorflow.Feature&gt; feature = 1;</code>
+     */
+
+    public Builder removeFeature(
+        java.lang.String key) {
+      if (key == null) { throw new java.lang.NullPointerException(); }
+      internalGetMutableFeature().getMutableMap()
+          .remove(key);
+      return this;
+    }
+    /**
+     * Use alternate mutation accessors instead.
+     */
+    @java.lang.Deprecated
+    public java.util.Map<java.lang.String, org.tensorflow.proto.example.Feature>
+    getMutableFeature() {
+      return internalGetMutableFeature().getMutableMap();
+    }
+    /**
+     * <pre>
+     * Map from feature name to feature.
+     * </pre>
+     *
+     * <code>map&lt;string, .tensorflow.Feature&gt; feature = 1;</code>
+     */
+    public Builder putFeature(
+        java.lang.String key,
+        org.tensorflow.proto.example.Feature value) {
+      if (key == null) { throw new java.lang.NullPointerException(); }
+      if (value == null) { throw new java.lang.NullPointerException(); }
+      internalGetMutableFeature().getMutableMap()
+          .put(key, value);
+      return this;
+    }
+    /**
+     * <pre>
+     * Map from feature name to feature.
+     * </pre>
+     *
+     * <code>map&lt;string, .tensorflow.Feature&gt; feature = 1;</code>
+     */
+
+    public Builder putAllFeature(
+        java.util.Map<java.lang.String, org.tensorflow.proto.example.Feature> values) {
+      internalGetMutableFeature().getMutableMap()
+          .putAll(values);
+      return this;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.Features)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.Features)
+  private static final org.tensorflow.proto.example.Features DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.example.Features();
+  }
+
+  public static org.tensorflow.proto.example.Features getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<Features>
+      PARSER = new com.google.protobuf.AbstractParser<Features>() {
+    @java.lang.Override
+    public Features parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new Features(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<Features> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<Features> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.example.Features getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/FeaturesOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/FeaturesOrBuilder.java
new file mode 100644
index 00000000000..2ee80ee66a0
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/FeaturesOrBuilder.java
@@ -0,0 +1,63 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/example/feature.proto
+
+package org.tensorflow.proto.example;
+
+public interface FeaturesOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.Features)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <pre>
+   * Map from feature name to feature.
+   * </pre>
+   *
+   * <code>map&lt;string, .tensorflow.Feature&gt; feature = 1;</code>
+   */
+  int getFeatureCount();
+  /**
+   * <pre>
+   * Map from feature name to feature.
+   * </pre>
+   *
+   * <code>map&lt;string, .tensorflow.Feature&gt; feature = 1;</code>
+   */
+  boolean containsFeature(
+      java.lang.String key);
+  /**
+   * Use {@link #getFeatureMap()} instead.
+   */
+  @java.lang.Deprecated
+  java.util.Map<java.lang.String, org.tensorflow.proto.example.Feature>
+  getFeature();
+  /**
+   * <pre>
+   * Map from feature name to feature.
+   * </pre>
+   *
+   * <code>map&lt;string, .tensorflow.Feature&gt; feature = 1;</code>
+   */
+  java.util.Map<java.lang.String, org.tensorflow.proto.example.Feature>
+  getFeatureMap();
+  /**
+   * <pre>
+   * Map from feature name to feature.
+   * </pre>
+   *
+   * <code>map&lt;string, .tensorflow.Feature&gt; feature = 1;</code>
+   */
+
+  org.tensorflow.proto.example.Feature getFeatureOrDefault(
+      java.lang.String key,
+      org.tensorflow.proto.example.Feature defaultValue);
+  /**
+   * <pre>
+   * Map from feature name to feature.
+   * </pre>
+   *
+   * <code>map&lt;string, .tensorflow.Feature&gt; feature = 1;</code>
+   */
+
+  org.tensorflow.proto.example.Feature getFeatureOrThrow(
+      java.lang.String key);
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/FixedLenFeatureProto.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/FixedLenFeatureProto.java
new file mode 100644
index 00000000000..1f378121573
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/FixedLenFeatureProto.java
@@ -0,0 +1,993 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/example/example_parser_configuration.proto
+
+package org.tensorflow.proto.example;
+
+/**
+ * Protobuf type {@code tensorflow.FixedLenFeatureProto}
+ */
+public  final class FixedLenFeatureProto extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.FixedLenFeatureProto)
+    FixedLenFeatureProtoOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use FixedLenFeatureProto.newBuilder() to construct.
+  private FixedLenFeatureProto(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private FixedLenFeatureProto() {
+    dtype_ = 0;
+    valuesOutputTensorName_ = "";
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new FixedLenFeatureProto();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private FixedLenFeatureProto(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 8: {
+            int rawValue = input.readEnum();
+
+            dtype_ = rawValue;
+            break;
+          }
+          case 18: {
+            org.tensorflow.proto.framework.TensorShapeProto.Builder subBuilder = null;
+            if (shape_ != null) {
+              subBuilder = shape_.toBuilder();
+            }
+            shape_ = input.readMessage(org.tensorflow.proto.framework.TensorShapeProto.parser(), extensionRegistry);
+            if (subBuilder != null) {
+              subBuilder.mergeFrom(shape_);
+              shape_ = subBuilder.buildPartial();
+            }
+
+            break;
+          }
+          case 26: {
+            org.tensorflow.proto.framework.TensorProto.Builder subBuilder = null;
+            if (defaultValue_ != null) {
+              subBuilder = defaultValue_.toBuilder();
+            }
+            defaultValue_ = input.readMessage(org.tensorflow.proto.framework.TensorProto.parser(), extensionRegistry);
+            if (subBuilder != null) {
+              subBuilder.mergeFrom(defaultValue_);
+              defaultValue_ = subBuilder.buildPartial();
+            }
+
+            break;
+          }
+          case 34: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            valuesOutputTensorName_ = s;
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.example.ExampleParserConfigurationProtos.internal_static_tensorflow_FixedLenFeatureProto_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.example.ExampleParserConfigurationProtos.internal_static_tensorflow_FixedLenFeatureProto_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.example.FixedLenFeatureProto.class, org.tensorflow.proto.example.FixedLenFeatureProto.Builder.class);
+  }
+
+  public static final int DTYPE_FIELD_NUMBER = 1;
+  private int dtype_;
+  /**
+   * <code>.tensorflow.DataType dtype = 1;</code>
+   */
+  public int getDtypeValue() {
+    return dtype_;
+  }
+  /**
+   * <code>.tensorflow.DataType dtype = 1;</code>
+   */
+  public org.tensorflow.proto.framework.DataType getDtype() {
+    @SuppressWarnings("deprecation")
+    org.tensorflow.proto.framework.DataType result = org.tensorflow.proto.framework.DataType.valueOf(dtype_);
+    return result == null ? org.tensorflow.proto.framework.DataType.UNRECOGNIZED : result;
+  }
+
+  public static final int SHAPE_FIELD_NUMBER = 2;
+  private org.tensorflow.proto.framework.TensorShapeProto shape_;
+  /**
+   * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+   */
+  public boolean hasShape() {
+    return shape_ != null;
+  }
+  /**
+   * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+   */
+  public org.tensorflow.proto.framework.TensorShapeProto getShape() {
+    return shape_ == null ? org.tensorflow.proto.framework.TensorShapeProto.getDefaultInstance() : shape_;
+  }
+  /**
+   * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+   */
+  public org.tensorflow.proto.framework.TensorShapeProtoOrBuilder getShapeOrBuilder() {
+    return getShape();
+  }
+
+  public static final int DEFAULT_VALUE_FIELD_NUMBER = 3;
+  private org.tensorflow.proto.framework.TensorProto defaultValue_;
+  /**
+   * <code>.tensorflow.TensorProto default_value = 3;</code>
+   */
+  public boolean hasDefaultValue() {
+    return defaultValue_ != null;
+  }
+  /**
+   * <code>.tensorflow.TensorProto default_value = 3;</code>
+   */
+  public org.tensorflow.proto.framework.TensorProto getDefaultValue() {
+    return defaultValue_ == null ? org.tensorflow.proto.framework.TensorProto.getDefaultInstance() : defaultValue_;
+  }
+  /**
+   * <code>.tensorflow.TensorProto default_value = 3;</code>
+   */
+  public org.tensorflow.proto.framework.TensorProtoOrBuilder getDefaultValueOrBuilder() {
+    return getDefaultValue();
+  }
+
+  public static final int VALUES_OUTPUT_TENSOR_NAME_FIELD_NUMBER = 4;
+  private volatile java.lang.Object valuesOutputTensorName_;
+  /**
+   * <code>string values_output_tensor_name = 4;</code>
+   */
+  public java.lang.String getValuesOutputTensorName() {
+    java.lang.Object ref = valuesOutputTensorName_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      valuesOutputTensorName_ = s;
+      return s;
+    }
+  }
+  /**
+   * <code>string values_output_tensor_name = 4;</code>
+   */
+  public com.google.protobuf.ByteString
+      getValuesOutputTensorNameBytes() {
+    java.lang.Object ref = valuesOutputTensorName_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      valuesOutputTensorName_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    if (dtype_ != org.tensorflow.proto.framework.DataType.DT_INVALID.getNumber()) {
+      output.writeEnum(1, dtype_);
+    }
+    if (shape_ != null) {
+      output.writeMessage(2, getShape());
+    }
+    if (defaultValue_ != null) {
+      output.writeMessage(3, getDefaultValue());
+    }
+    if (!getValuesOutputTensorNameBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 4, valuesOutputTensorName_);
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (dtype_ != org.tensorflow.proto.framework.DataType.DT_INVALID.getNumber()) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeEnumSize(1, dtype_);
+    }
+    if (shape_ != null) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(2, getShape());
+    }
+    if (defaultValue_ != null) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(3, getDefaultValue());
+    }
+    if (!getValuesOutputTensorNameBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(4, valuesOutputTensorName_);
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.example.FixedLenFeatureProto)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.example.FixedLenFeatureProto other = (org.tensorflow.proto.example.FixedLenFeatureProto) obj;
+
+    if (dtype_ != other.dtype_) return false;
+    if (hasShape() != other.hasShape()) return false;
+    if (hasShape()) {
+      if (!getShape()
+          .equals(other.getShape())) return false;
+    }
+    if (hasDefaultValue() != other.hasDefaultValue()) return false;
+    if (hasDefaultValue()) {
+      if (!getDefaultValue()
+          .equals(other.getDefaultValue())) return false;
+    }
+    if (!getValuesOutputTensorName()
+        .equals(other.getValuesOutputTensorName())) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    hash = (37 * hash) + DTYPE_FIELD_NUMBER;
+    hash = (53 * hash) + dtype_;
+    if (hasShape()) {
+      hash = (37 * hash) + SHAPE_FIELD_NUMBER;
+      hash = (53 * hash) + getShape().hashCode();
+    }
+    if (hasDefaultValue()) {
+      hash = (37 * hash) + DEFAULT_VALUE_FIELD_NUMBER;
+      hash = (53 * hash) + getDefaultValue().hashCode();
+    }
+    hash = (37 * hash) + VALUES_OUTPUT_TENSOR_NAME_FIELD_NUMBER;
+    hash = (53 * hash) + getValuesOutputTensorName().hashCode();
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.example.FixedLenFeatureProto parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.example.FixedLenFeatureProto parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.example.FixedLenFeatureProto parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.example.FixedLenFeatureProto parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.example.FixedLenFeatureProto parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.example.FixedLenFeatureProto parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.example.FixedLenFeatureProto parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.example.FixedLenFeatureProto parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.example.FixedLenFeatureProto parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.example.FixedLenFeatureProto parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.example.FixedLenFeatureProto parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.example.FixedLenFeatureProto parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.example.FixedLenFeatureProto prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * Protobuf type {@code tensorflow.FixedLenFeatureProto}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.FixedLenFeatureProto)
+      org.tensorflow.proto.example.FixedLenFeatureProtoOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.example.ExampleParserConfigurationProtos.internal_static_tensorflow_FixedLenFeatureProto_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.example.ExampleParserConfigurationProtos.internal_static_tensorflow_FixedLenFeatureProto_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.example.FixedLenFeatureProto.class, org.tensorflow.proto.example.FixedLenFeatureProto.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.example.FixedLenFeatureProto.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      dtype_ = 0;
+
+      if (shapeBuilder_ == null) {
+        shape_ = null;
+      } else {
+        shape_ = null;
+        shapeBuilder_ = null;
+      }
+      if (defaultValueBuilder_ == null) {
+        defaultValue_ = null;
+      } else {
+        defaultValue_ = null;
+        defaultValueBuilder_ = null;
+      }
+      valuesOutputTensorName_ = "";
+
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.example.ExampleParserConfigurationProtos.internal_static_tensorflow_FixedLenFeatureProto_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.example.FixedLenFeatureProto getDefaultInstanceForType() {
+      return org.tensorflow.proto.example.FixedLenFeatureProto.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.example.FixedLenFeatureProto build() {
+      org.tensorflow.proto.example.FixedLenFeatureProto result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.example.FixedLenFeatureProto buildPartial() {
+      org.tensorflow.proto.example.FixedLenFeatureProto result = new org.tensorflow.proto.example.FixedLenFeatureProto(this);
+      result.dtype_ = dtype_;
+      if (shapeBuilder_ == null) {
+        result.shape_ = shape_;
+      } else {
+        result.shape_ = shapeBuilder_.build();
+      }
+      if (defaultValueBuilder_ == null) {
+        result.defaultValue_ = defaultValue_;
+      } else {
+        result.defaultValue_ = defaultValueBuilder_.build();
+      }
+      result.valuesOutputTensorName_ = valuesOutputTensorName_;
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.example.FixedLenFeatureProto) {
+        return mergeFrom((org.tensorflow.proto.example.FixedLenFeatureProto)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.example.FixedLenFeatureProto other) {
+      if (other == org.tensorflow.proto.example.FixedLenFeatureProto.getDefaultInstance()) return this;
+      if (other.dtype_ != 0) {
+        setDtypeValue(other.getDtypeValue());
+      }
+      if (other.hasShape()) {
+        mergeShape(other.getShape());
+      }
+      if (other.hasDefaultValue()) {
+        mergeDefaultValue(other.getDefaultValue());
+      }
+      if (!other.getValuesOutputTensorName().isEmpty()) {
+        valuesOutputTensorName_ = other.valuesOutputTensorName_;
+        onChanged();
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.example.FixedLenFeatureProto parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.example.FixedLenFeatureProto) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+
+    private int dtype_ = 0;
+    /**
+     * <code>.tensorflow.DataType dtype = 1;</code>
+     */
+    public int getDtypeValue() {
+      return dtype_;
+    }
+    /**
+     * <code>.tensorflow.DataType dtype = 1;</code>
+     */
+    public Builder setDtypeValue(int value) {
+      dtype_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>.tensorflow.DataType dtype = 1;</code>
+     */
+    public org.tensorflow.proto.framework.DataType getDtype() {
+      @SuppressWarnings("deprecation")
+      org.tensorflow.proto.framework.DataType result = org.tensorflow.proto.framework.DataType.valueOf(dtype_);
+      return result == null ? org.tensorflow.proto.framework.DataType.UNRECOGNIZED : result;
+    }
+    /**
+     * <code>.tensorflow.DataType dtype = 1;</code>
+     */
+    public Builder setDtype(org.tensorflow.proto.framework.DataType value) {
+      if (value == null) {
+        throw new NullPointerException();
+      }
+      
+      dtype_ = value.getNumber();
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>.tensorflow.DataType dtype = 1;</code>
+     */
+    public Builder clearDtype() {
+      
+      dtype_ = 0;
+      onChanged();
+      return this;
+    }
+
+    private org.tensorflow.proto.framework.TensorShapeProto shape_;
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.TensorShapeProto, org.tensorflow.proto.framework.TensorShapeProto.Builder, org.tensorflow.proto.framework.TensorShapeProtoOrBuilder> shapeBuilder_;
+    /**
+     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+     */
+    public boolean hasShape() {
+      return shapeBuilder_ != null || shape_ != null;
+    }
+    /**
+     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+     */
+    public org.tensorflow.proto.framework.TensorShapeProto getShape() {
+      if (shapeBuilder_ == null) {
+        return shape_ == null ? org.tensorflow.proto.framework.TensorShapeProto.getDefaultInstance() : shape_;
+      } else {
+        return shapeBuilder_.getMessage();
+      }
+    }
+    /**
+     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+     */
+    public Builder setShape(org.tensorflow.proto.framework.TensorShapeProto value) {
+      if (shapeBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        shape_ = value;
+        onChanged();
+      } else {
+        shapeBuilder_.setMessage(value);
+      }
+
+      return this;
+    }
+    /**
+     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+     */
+    public Builder setShape(
+        org.tensorflow.proto.framework.TensorShapeProto.Builder builderForValue) {
+      if (shapeBuilder_ == null) {
+        shape_ = builderForValue.build();
+        onChanged();
+      } else {
+        shapeBuilder_.setMessage(builderForValue.build());
+      }
+
+      return this;
+    }
+    /**
+     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+     */
+    public Builder mergeShape(org.tensorflow.proto.framework.TensorShapeProto value) {
+      if (shapeBuilder_ == null) {
+        if (shape_ != null) {
+          shape_ =
+            org.tensorflow.proto.framework.TensorShapeProto.newBuilder(shape_).mergeFrom(value).buildPartial();
+        } else {
+          shape_ = value;
+        }
+        onChanged();
+      } else {
+        shapeBuilder_.mergeFrom(value);
+      }
+
+      return this;
+    }
+    /**
+     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+     */
+    public Builder clearShape() {
+      if (shapeBuilder_ == null) {
+        shape_ = null;
+        onChanged();
+      } else {
+        shape_ = null;
+        shapeBuilder_ = null;
+      }
+
+      return this;
+    }
+    /**
+     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+     */
+    public org.tensorflow.proto.framework.TensorShapeProto.Builder getShapeBuilder() {
+      
+      onChanged();
+      return getShapeFieldBuilder().getBuilder();
+    }
+    /**
+     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+     */
+    public org.tensorflow.proto.framework.TensorShapeProtoOrBuilder getShapeOrBuilder() {
+      if (shapeBuilder_ != null) {
+        return shapeBuilder_.getMessageOrBuilder();
+      } else {
+        return shape_ == null ?
+            org.tensorflow.proto.framework.TensorShapeProto.getDefaultInstance() : shape_;
+      }
+    }
+    /**
+     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+     */
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.TensorShapeProto, org.tensorflow.proto.framework.TensorShapeProto.Builder, org.tensorflow.proto.framework.TensorShapeProtoOrBuilder> 
+        getShapeFieldBuilder() {
+      if (shapeBuilder_ == null) {
+        shapeBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+            org.tensorflow.proto.framework.TensorShapeProto, org.tensorflow.proto.framework.TensorShapeProto.Builder, org.tensorflow.proto.framework.TensorShapeProtoOrBuilder>(
+                getShape(),
+                getParentForChildren(),
+                isClean());
+        shape_ = null;
+      }
+      return shapeBuilder_;
+    }
+
+    private org.tensorflow.proto.framework.TensorProto defaultValue_;
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.TensorProto, org.tensorflow.proto.framework.TensorProto.Builder, org.tensorflow.proto.framework.TensorProtoOrBuilder> defaultValueBuilder_;
+    /**
+     * <code>.tensorflow.TensorProto default_value = 3;</code>
+     */
+    public boolean hasDefaultValue() {
+      return defaultValueBuilder_ != null || defaultValue_ != null;
+    }
+    /**
+     * <code>.tensorflow.TensorProto default_value = 3;</code>
+     */
+    public org.tensorflow.proto.framework.TensorProto getDefaultValue() {
+      if (defaultValueBuilder_ == null) {
+        return defaultValue_ == null ? org.tensorflow.proto.framework.TensorProto.getDefaultInstance() : defaultValue_;
+      } else {
+        return defaultValueBuilder_.getMessage();
+      }
+    }
+    /**
+     * <code>.tensorflow.TensorProto default_value = 3;</code>
+     */
+    public Builder setDefaultValue(org.tensorflow.proto.framework.TensorProto value) {
+      if (defaultValueBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        defaultValue_ = value;
+        onChanged();
+      } else {
+        defaultValueBuilder_.setMessage(value);
+      }
+
+      return this;
+    }
+    /**
+     * <code>.tensorflow.TensorProto default_value = 3;</code>
+     */
+    public Builder setDefaultValue(
+        org.tensorflow.proto.framework.TensorProto.Builder builderForValue) {
+      if (defaultValueBuilder_ == null) {
+        defaultValue_ = builderForValue.build();
+        onChanged();
+      } else {
+        defaultValueBuilder_.setMessage(builderForValue.build());
+      }
+
+      return this;
+    }
+    /**
+     * <code>.tensorflow.TensorProto default_value = 3;</code>
+     */
+    public Builder mergeDefaultValue(org.tensorflow.proto.framework.TensorProto value) {
+      if (defaultValueBuilder_ == null) {
+        if (defaultValue_ != null) {
+          defaultValue_ =
+            org.tensorflow.proto.framework.TensorProto.newBuilder(defaultValue_).mergeFrom(value).buildPartial();
+        } else {
+          defaultValue_ = value;
+        }
+        onChanged();
+      } else {
+        defaultValueBuilder_.mergeFrom(value);
+      }
+
+      return this;
+    }
+    /**
+     * <code>.tensorflow.TensorProto default_value = 3;</code>
+     */
+    public Builder clearDefaultValue() {
+      if (defaultValueBuilder_ == null) {
+        defaultValue_ = null;
+        onChanged();
+      } else {
+        defaultValue_ = null;
+        defaultValueBuilder_ = null;
+      }
+
+      return this;
+    }
+    /**
+     * <code>.tensorflow.TensorProto default_value = 3;</code>
+     */
+    public org.tensorflow.proto.framework.TensorProto.Builder getDefaultValueBuilder() {
+      
+      onChanged();
+      return getDefaultValueFieldBuilder().getBuilder();
+    }
+    /**
+     * <code>.tensorflow.TensorProto default_value = 3;</code>
+     */
+    public org.tensorflow.proto.framework.TensorProtoOrBuilder getDefaultValueOrBuilder() {
+      if (defaultValueBuilder_ != null) {
+        return defaultValueBuilder_.getMessageOrBuilder();
+      } else {
+        return defaultValue_ == null ?
+            org.tensorflow.proto.framework.TensorProto.getDefaultInstance() : defaultValue_;
+      }
+    }
+    /**
+     * <code>.tensorflow.TensorProto default_value = 3;</code>
+     */
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.TensorProto, org.tensorflow.proto.framework.TensorProto.Builder, org.tensorflow.proto.framework.TensorProtoOrBuilder> 
+        getDefaultValueFieldBuilder() {
+      if (defaultValueBuilder_ == null) {
+        defaultValueBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+            org.tensorflow.proto.framework.TensorProto, org.tensorflow.proto.framework.TensorProto.Builder, org.tensorflow.proto.framework.TensorProtoOrBuilder>(
+                getDefaultValue(),
+                getParentForChildren(),
+                isClean());
+        defaultValue_ = null;
+      }
+      return defaultValueBuilder_;
+    }
+
+    private java.lang.Object valuesOutputTensorName_ = "";
+    /**
+     * <code>string values_output_tensor_name = 4;</code>
+     */
+    public java.lang.String getValuesOutputTensorName() {
+      java.lang.Object ref = valuesOutputTensorName_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        valuesOutputTensorName_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <code>string values_output_tensor_name = 4;</code>
+     */
+    public com.google.protobuf.ByteString
+        getValuesOutputTensorNameBytes() {
+      java.lang.Object ref = valuesOutputTensorName_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        valuesOutputTensorName_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <code>string values_output_tensor_name = 4;</code>
+     */
+    public Builder setValuesOutputTensorName(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      valuesOutputTensorName_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>string values_output_tensor_name = 4;</code>
+     */
+    public Builder clearValuesOutputTensorName() {
+      
+      valuesOutputTensorName_ = getDefaultInstance().getValuesOutputTensorName();
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>string values_output_tensor_name = 4;</code>
+     */
+    public Builder setValuesOutputTensorNameBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      valuesOutputTensorName_ = value;
+      onChanged();
+      return this;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.FixedLenFeatureProto)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.FixedLenFeatureProto)
+  private static final org.tensorflow.proto.example.FixedLenFeatureProto DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.example.FixedLenFeatureProto();
+  }
+
+  public static org.tensorflow.proto.example.FixedLenFeatureProto getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<FixedLenFeatureProto>
+      PARSER = new com.google.protobuf.AbstractParser<FixedLenFeatureProto>() {
+    @java.lang.Override
+    public FixedLenFeatureProto parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new FixedLenFeatureProto(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<FixedLenFeatureProto> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<FixedLenFeatureProto> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.example.FixedLenFeatureProto getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/FixedLenFeatureProtoOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/FixedLenFeatureProtoOrBuilder.java
new file mode 100644
index 00000000000..0008b29583f
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/FixedLenFeatureProtoOrBuilder.java
@@ -0,0 +1,54 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/example/example_parser_configuration.proto
+
+package org.tensorflow.proto.example;
+
+public interface FixedLenFeatureProtoOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.FixedLenFeatureProto)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <code>.tensorflow.DataType dtype = 1;</code>
+   */
+  int getDtypeValue();
+  /**
+   * <code>.tensorflow.DataType dtype = 1;</code>
+   */
+  org.tensorflow.proto.framework.DataType getDtype();
+
+  /**
+   * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+   */
+  boolean hasShape();
+  /**
+   * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+   */
+  org.tensorflow.proto.framework.TensorShapeProto getShape();
+  /**
+   * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+   */
+  org.tensorflow.proto.framework.TensorShapeProtoOrBuilder getShapeOrBuilder();
+
+  /**
+   * <code>.tensorflow.TensorProto default_value = 3;</code>
+   */
+  boolean hasDefaultValue();
+  /**
+   * <code>.tensorflow.TensorProto default_value = 3;</code>
+   */
+  org.tensorflow.proto.framework.TensorProto getDefaultValue();
+  /**
+   * <code>.tensorflow.TensorProto default_value = 3;</code>
+   */
+  org.tensorflow.proto.framework.TensorProtoOrBuilder getDefaultValueOrBuilder();
+
+  /**
+   * <code>string values_output_tensor_name = 4;</code>
+   */
+  java.lang.String getValuesOutputTensorName();
+  /**
+   * <code>string values_output_tensor_name = 4;</code>
+   */
+  com.google.protobuf.ByteString
+      getValuesOutputTensorNameBytes();
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/FloatList.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/FloatList.java
new file mode 100644
index 00000000000..efad9ee2fb6
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/FloatList.java
@@ -0,0 +1,579 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/example/feature.proto
+
+package org.tensorflow.proto.example;
+
+/**
+ * Protobuf type {@code tensorflow.FloatList}
+ */
+public  final class FloatList extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.FloatList)
+    FloatListOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use FloatList.newBuilder() to construct.
+  private FloatList(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private FloatList() {
+    value_ = emptyFloatList();
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new FloatList();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private FloatList(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    int mutable_bitField0_ = 0;
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 13: {
+            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
+              value_ = newFloatList();
+              mutable_bitField0_ |= 0x00000001;
+            }
+            value_.addFloat(input.readFloat());
+            break;
+          }
+          case 10: {
+            int length = input.readRawVarint32();
+            int limit = input.pushLimit(length);
+            if (!((mutable_bitField0_ & 0x00000001) != 0) && input.getBytesUntilLimit() > 0) {
+              value_ = newFloatList();
+              mutable_bitField0_ |= 0x00000001;
+            }
+            while (input.getBytesUntilLimit() > 0) {
+              value_.addFloat(input.readFloat());
+            }
+            input.popLimit(limit);
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      if (((mutable_bitField0_ & 0x00000001) != 0)) {
+        value_.makeImmutable(); // C
+      }
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.example.FeatureProtos.internal_static_tensorflow_FloatList_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.example.FeatureProtos.internal_static_tensorflow_FloatList_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.example.FloatList.class, org.tensorflow.proto.example.FloatList.Builder.class);
+  }
+
+  public static final int VALUE_FIELD_NUMBER = 1;
+  private com.google.protobuf.Internal.FloatList value_;
+  /**
+   * <code>repeated float value = 1 [packed = true];</code>
+   */
+  public java.util.List<java.lang.Float>
+      getValueList() {
+    return value_;
+  }
+  /**
+   * <code>repeated float value = 1 [packed = true];</code>
+   */
+  public int getValueCount() {
+    return value_.size();
+  }
+  /**
+   * <code>repeated float value = 1 [packed = true];</code>
+   */
+  public float getValue(int index) {
+    return value_.getFloat(index);
+  }
+  private int valueMemoizedSerializedSize = -1;
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    getSerializedSize();
+    if (getValueList().size() > 0) {
+      output.writeUInt32NoTag(10);
+      output.writeUInt32NoTag(valueMemoizedSerializedSize);
+    }
+    for (int i = 0; i < value_.size(); i++) {
+      output.writeFloatNoTag(value_.getFloat(i));
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    {
+      int dataSize = 0;
+      dataSize = 4 * getValueList().size();
+      size += dataSize;
+      if (!getValueList().isEmpty()) {
+        size += 1;
+        size += com.google.protobuf.CodedOutputStream
+            .computeInt32SizeNoTag(dataSize);
+      }
+      valueMemoizedSerializedSize = dataSize;
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.example.FloatList)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.example.FloatList other = (org.tensorflow.proto.example.FloatList) obj;
+
+    if (!getValueList()
+        .equals(other.getValueList())) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    if (getValueCount() > 0) {
+      hash = (37 * hash) + VALUE_FIELD_NUMBER;
+      hash = (53 * hash) + getValueList().hashCode();
+    }
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.example.FloatList parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.example.FloatList parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.example.FloatList parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.example.FloatList parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.example.FloatList parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.example.FloatList parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.example.FloatList parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.example.FloatList parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.example.FloatList parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.example.FloatList parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.example.FloatList parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.example.FloatList parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.example.FloatList prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * Protobuf type {@code tensorflow.FloatList}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.FloatList)
+      org.tensorflow.proto.example.FloatListOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.example.FeatureProtos.internal_static_tensorflow_FloatList_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.example.FeatureProtos.internal_static_tensorflow_FloatList_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.example.FloatList.class, org.tensorflow.proto.example.FloatList.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.example.FloatList.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      value_ = emptyFloatList();
+      bitField0_ = (bitField0_ & ~0x00000001);
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.example.FeatureProtos.internal_static_tensorflow_FloatList_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.example.FloatList getDefaultInstanceForType() {
+      return org.tensorflow.proto.example.FloatList.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.example.FloatList build() {
+      org.tensorflow.proto.example.FloatList result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.example.FloatList buildPartial() {
+      org.tensorflow.proto.example.FloatList result = new org.tensorflow.proto.example.FloatList(this);
+      int from_bitField0_ = bitField0_;
+      if (((bitField0_ & 0x00000001) != 0)) {
+        value_.makeImmutable();
+        bitField0_ = (bitField0_ & ~0x00000001);
+      }
+      result.value_ = value_;
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.example.FloatList) {
+        return mergeFrom((org.tensorflow.proto.example.FloatList)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.example.FloatList other) {
+      if (other == org.tensorflow.proto.example.FloatList.getDefaultInstance()) return this;
+      if (!other.value_.isEmpty()) {
+        if (value_.isEmpty()) {
+          value_ = other.value_;
+          bitField0_ = (bitField0_ & ~0x00000001);
+        } else {
+          ensureValueIsMutable();
+          value_.addAll(other.value_);
+        }
+        onChanged();
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.example.FloatList parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.example.FloatList) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+    private int bitField0_;
+
+    private com.google.protobuf.Internal.FloatList value_ = emptyFloatList();
+    private void ensureValueIsMutable() {
+      if (!((bitField0_ & 0x00000001) != 0)) {
+        value_ = mutableCopy(value_);
+        bitField0_ |= 0x00000001;
+       }
+    }
+    /**
+     * <code>repeated float value = 1 [packed = true];</code>
+     */
+    public java.util.List<java.lang.Float>
+        getValueList() {
+      return ((bitField0_ & 0x00000001) != 0) ?
+               java.util.Collections.unmodifiableList(value_) : value_;
+    }
+    /**
+     * <code>repeated float value = 1 [packed = true];</code>
+     */
+    public int getValueCount() {
+      return value_.size();
+    }
+    /**
+     * <code>repeated float value = 1 [packed = true];</code>
+     */
+    public float getValue(int index) {
+      return value_.getFloat(index);
+    }
+    /**
+     * <code>repeated float value = 1 [packed = true];</code>
+     */
+    public Builder setValue(
+        int index, float value) {
+      ensureValueIsMutable();
+      value_.setFloat(index, value);
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>repeated float value = 1 [packed = true];</code>
+     */
+    public Builder addValue(float value) {
+      ensureValueIsMutable();
+      value_.addFloat(value);
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>repeated float value = 1 [packed = true];</code>
+     */
+    public Builder addAllValue(
+        java.lang.Iterable<? extends java.lang.Float> values) {
+      ensureValueIsMutable();
+      com.google.protobuf.AbstractMessageLite.Builder.addAll(
+          values, value_);
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>repeated float value = 1 [packed = true];</code>
+     */
+    public Builder clearValue() {
+      value_ = emptyFloatList();
+      bitField0_ = (bitField0_ & ~0x00000001);
+      onChanged();
+      return this;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.FloatList)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.FloatList)
+  private static final org.tensorflow.proto.example.FloatList DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.example.FloatList();
+  }
+
+  public static org.tensorflow.proto.example.FloatList getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<FloatList>
+      PARSER = new com.google.protobuf.AbstractParser<FloatList>() {
+    @java.lang.Override
+    public FloatList parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new FloatList(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<FloatList> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<FloatList> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.example.FloatList getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/FloatListOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/FloatListOrBuilder.java
new file mode 100644
index 00000000000..65856cd5f4e
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/FloatListOrBuilder.java
@@ -0,0 +1,22 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/example/feature.proto
+
+package org.tensorflow.proto.example;
+
+public interface FloatListOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.FloatList)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <code>repeated float value = 1 [packed = true];</code>
+   */
+  java.util.List<java.lang.Float> getValueList();
+  /**
+   * <code>repeated float value = 1 [packed = true];</code>
+   */
+  int getValueCount();
+  /**
+   * <code>repeated float value = 1 [packed = true];</code>
+   */
+  float getValue(int index);
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/Int64List.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/Int64List.java
new file mode 100644
index 00000000000..1cf4f104182
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/Int64List.java
@@ -0,0 +1,582 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/example/feature.proto
+
+package org.tensorflow.proto.example;
+
+/**
+ * Protobuf type {@code tensorflow.Int64List}
+ */
+public  final class Int64List extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.Int64List)
+    Int64ListOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use Int64List.newBuilder() to construct.
+  private Int64List(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private Int64List() {
+    value_ = emptyLongList();
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new Int64List();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private Int64List(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    int mutable_bitField0_ = 0;
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 8: {
+            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
+              value_ = newLongList();
+              mutable_bitField0_ |= 0x00000001;
+            }
+            value_.addLong(input.readInt64());
+            break;
+          }
+          case 10: {
+            int length = input.readRawVarint32();
+            int limit = input.pushLimit(length);
+            if (!((mutable_bitField0_ & 0x00000001) != 0) && input.getBytesUntilLimit() > 0) {
+              value_ = newLongList();
+              mutable_bitField0_ |= 0x00000001;
+            }
+            while (input.getBytesUntilLimit() > 0) {
+              value_.addLong(input.readInt64());
+            }
+            input.popLimit(limit);
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      if (((mutable_bitField0_ & 0x00000001) != 0)) {
+        value_.makeImmutable(); // C
+      }
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.example.FeatureProtos.internal_static_tensorflow_Int64List_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.example.FeatureProtos.internal_static_tensorflow_Int64List_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.example.Int64List.class, org.tensorflow.proto.example.Int64List.Builder.class);
+  }
+
+  public static final int VALUE_FIELD_NUMBER = 1;
+  private com.google.protobuf.Internal.LongList value_;
+  /**
+   * <code>repeated int64 value = 1 [packed = true];</code>
+   */
+  public java.util.List<java.lang.Long>
+      getValueList() {
+    return value_;
+  }
+  /**
+   * <code>repeated int64 value = 1 [packed = true];</code>
+   */
+  public int getValueCount() {
+    return value_.size();
+  }
+  /**
+   * <code>repeated int64 value = 1 [packed = true];</code>
+   */
+  public long getValue(int index) {
+    return value_.getLong(index);
+  }
+  private int valueMemoizedSerializedSize = -1;
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    getSerializedSize();
+    if (getValueList().size() > 0) {
+      output.writeUInt32NoTag(10);
+      output.writeUInt32NoTag(valueMemoizedSerializedSize);
+    }
+    for (int i = 0; i < value_.size(); i++) {
+      output.writeInt64NoTag(value_.getLong(i));
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    {
+      int dataSize = 0;
+      for (int i = 0; i < value_.size(); i++) {
+        dataSize += com.google.protobuf.CodedOutputStream
+          .computeInt64SizeNoTag(value_.getLong(i));
+      }
+      size += dataSize;
+      if (!getValueList().isEmpty()) {
+        size += 1;
+        size += com.google.protobuf.CodedOutputStream
+            .computeInt32SizeNoTag(dataSize);
+      }
+      valueMemoizedSerializedSize = dataSize;
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.example.Int64List)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.example.Int64List other = (org.tensorflow.proto.example.Int64List) obj;
+
+    if (!getValueList()
+        .equals(other.getValueList())) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    if (getValueCount() > 0) {
+      hash = (37 * hash) + VALUE_FIELD_NUMBER;
+      hash = (53 * hash) + getValueList().hashCode();
+    }
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.example.Int64List parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.example.Int64List parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.example.Int64List parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.example.Int64List parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.example.Int64List parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.example.Int64List parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.example.Int64List parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.example.Int64List parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.example.Int64List parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.example.Int64List parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.example.Int64List parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.example.Int64List parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.example.Int64List prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * Protobuf type {@code tensorflow.Int64List}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.Int64List)
+      org.tensorflow.proto.example.Int64ListOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.example.FeatureProtos.internal_static_tensorflow_Int64List_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.example.FeatureProtos.internal_static_tensorflow_Int64List_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.example.Int64List.class, org.tensorflow.proto.example.Int64List.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.example.Int64List.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      value_ = emptyLongList();
+      bitField0_ = (bitField0_ & ~0x00000001);
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.example.FeatureProtos.internal_static_tensorflow_Int64List_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.example.Int64List getDefaultInstanceForType() {
+      return org.tensorflow.proto.example.Int64List.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.example.Int64List build() {
+      org.tensorflow.proto.example.Int64List result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.example.Int64List buildPartial() {
+      org.tensorflow.proto.example.Int64List result = new org.tensorflow.proto.example.Int64List(this);
+      int from_bitField0_ = bitField0_;
+      if (((bitField0_ & 0x00000001) != 0)) {
+        value_.makeImmutable();
+        bitField0_ = (bitField0_ & ~0x00000001);
+      }
+      result.value_ = value_;
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.example.Int64List) {
+        return mergeFrom((org.tensorflow.proto.example.Int64List)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.example.Int64List other) {
+      if (other == org.tensorflow.proto.example.Int64List.getDefaultInstance()) return this;
+      if (!other.value_.isEmpty()) {
+        if (value_.isEmpty()) {
+          value_ = other.value_;
+          bitField0_ = (bitField0_ & ~0x00000001);
+        } else {
+          ensureValueIsMutable();
+          value_.addAll(other.value_);
+        }
+        onChanged();
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.example.Int64List parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.example.Int64List) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+    private int bitField0_;
+
+    private com.google.protobuf.Internal.LongList value_ = emptyLongList();
+    private void ensureValueIsMutable() {
+      if (!((bitField0_ & 0x00000001) != 0)) {
+        value_ = mutableCopy(value_);
+        bitField0_ |= 0x00000001;
+       }
+    }
+    /**
+     * <code>repeated int64 value = 1 [packed = true];</code>
+     */
+    public java.util.List<java.lang.Long>
+        getValueList() {
+      return ((bitField0_ & 0x00000001) != 0) ?
+               java.util.Collections.unmodifiableList(value_) : value_;
+    }
+    /**
+     * <code>repeated int64 value = 1 [packed = true];</code>
+     */
+    public int getValueCount() {
+      return value_.size();
+    }
+    /**
+     * <code>repeated int64 value = 1 [packed = true];</code>
+     */
+    public long getValue(int index) {
+      return value_.getLong(index);
+    }
+    /**
+     * <code>repeated int64 value = 1 [packed = true];</code>
+     */
+    public Builder setValue(
+        int index, long value) {
+      ensureValueIsMutable();
+      value_.setLong(index, value);
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>repeated int64 value = 1 [packed = true];</code>
+     */
+    public Builder addValue(long value) {
+      ensureValueIsMutable();
+      value_.addLong(value);
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>repeated int64 value = 1 [packed = true];</code>
+     */
+    public Builder addAllValue(
+        java.lang.Iterable<? extends java.lang.Long> values) {
+      ensureValueIsMutable();
+      com.google.protobuf.AbstractMessageLite.Builder.addAll(
+          values, value_);
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>repeated int64 value = 1 [packed = true];</code>
+     */
+    public Builder clearValue() {
+      value_ = emptyLongList();
+      bitField0_ = (bitField0_ & ~0x00000001);
+      onChanged();
+      return this;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.Int64List)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.Int64List)
+  private static final org.tensorflow.proto.example.Int64List DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.example.Int64List();
+  }
+
+  public static org.tensorflow.proto.example.Int64List getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<Int64List>
+      PARSER = new com.google.protobuf.AbstractParser<Int64List>() {
+    @java.lang.Override
+    public Int64List parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new Int64List(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<Int64List> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<Int64List> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.example.Int64List getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/Int64ListOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/Int64ListOrBuilder.java
new file mode 100644
index 00000000000..6ca9f168a7d
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/Int64ListOrBuilder.java
@@ -0,0 +1,22 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/example/feature.proto
+
+package org.tensorflow.proto.example;
+
+public interface Int64ListOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.Int64List)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <code>repeated int64 value = 1 [packed = true];</code>
+   */
+  java.util.List<java.lang.Long> getValueList();
+  /**
+   * <code>repeated int64 value = 1 [packed = true];</code>
+   */
+  int getValueCount();
+  /**
+   * <code>repeated int64 value = 1 [packed = true];</code>
+   */
+  long getValue(int index);
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/SequenceExample.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/SequenceExample.java
new file mode 100644
index 00000000000..b88e6f12e34
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/SequenceExample.java
@@ -0,0 +1,781 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/example/example.proto
+
+package org.tensorflow.proto.example;
+
+/**
+ * Protobuf type {@code tensorflow.SequenceExample}
+ */
+public  final class SequenceExample extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.SequenceExample)
+    SequenceExampleOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use SequenceExample.newBuilder() to construct.
+  private SequenceExample(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private SequenceExample() {
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new SequenceExample();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private SequenceExample(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 10: {
+            org.tensorflow.proto.example.Features.Builder subBuilder = null;
+            if (context_ != null) {
+              subBuilder = context_.toBuilder();
+            }
+            context_ = input.readMessage(org.tensorflow.proto.example.Features.parser(), extensionRegistry);
+            if (subBuilder != null) {
+              subBuilder.mergeFrom(context_);
+              context_ = subBuilder.buildPartial();
+            }
+
+            break;
+          }
+          case 18: {
+            org.tensorflow.proto.example.FeatureLists.Builder subBuilder = null;
+            if (featureLists_ != null) {
+              subBuilder = featureLists_.toBuilder();
+            }
+            featureLists_ = input.readMessage(org.tensorflow.proto.example.FeatureLists.parser(), extensionRegistry);
+            if (subBuilder != null) {
+              subBuilder.mergeFrom(featureLists_);
+              featureLists_ = subBuilder.buildPartial();
+            }
+
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.example.ExampleProtos.internal_static_tensorflow_SequenceExample_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.example.ExampleProtos.internal_static_tensorflow_SequenceExample_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.example.SequenceExample.class, org.tensorflow.proto.example.SequenceExample.Builder.class);
+  }
+
+  public static final int CONTEXT_FIELD_NUMBER = 1;
+  private org.tensorflow.proto.example.Features context_;
+  /**
+   * <code>.tensorflow.Features context = 1;</code>
+   */
+  public boolean hasContext() {
+    return context_ != null;
+  }
+  /**
+   * <code>.tensorflow.Features context = 1;</code>
+   */
+  public org.tensorflow.proto.example.Features getContext() {
+    return context_ == null ? org.tensorflow.proto.example.Features.getDefaultInstance() : context_;
+  }
+  /**
+   * <code>.tensorflow.Features context = 1;</code>
+   */
+  public org.tensorflow.proto.example.FeaturesOrBuilder getContextOrBuilder() {
+    return getContext();
+  }
+
+  public static final int FEATURE_LISTS_FIELD_NUMBER = 2;
+  private org.tensorflow.proto.example.FeatureLists featureLists_;
+  /**
+   * <code>.tensorflow.FeatureLists feature_lists = 2;</code>
+   */
+  public boolean hasFeatureLists() {
+    return featureLists_ != null;
+  }
+  /**
+   * <code>.tensorflow.FeatureLists feature_lists = 2;</code>
+   */
+  public org.tensorflow.proto.example.FeatureLists getFeatureLists() {
+    return featureLists_ == null ? org.tensorflow.proto.example.FeatureLists.getDefaultInstance() : featureLists_;
+  }
+  /**
+   * <code>.tensorflow.FeatureLists feature_lists = 2;</code>
+   */
+  public org.tensorflow.proto.example.FeatureListsOrBuilder getFeatureListsOrBuilder() {
+    return getFeatureLists();
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    if (context_ != null) {
+      output.writeMessage(1, getContext());
+    }
+    if (featureLists_ != null) {
+      output.writeMessage(2, getFeatureLists());
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (context_ != null) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(1, getContext());
+    }
+    if (featureLists_ != null) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(2, getFeatureLists());
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.example.SequenceExample)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.example.SequenceExample other = (org.tensorflow.proto.example.SequenceExample) obj;
+
+    if (hasContext() != other.hasContext()) return false;
+    if (hasContext()) {
+      if (!getContext()
+          .equals(other.getContext())) return false;
+    }
+    if (hasFeatureLists() != other.hasFeatureLists()) return false;
+    if (hasFeatureLists()) {
+      if (!getFeatureLists()
+          .equals(other.getFeatureLists())) return false;
+    }
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    if (hasContext()) {
+      hash = (37 * hash) + CONTEXT_FIELD_NUMBER;
+      hash = (53 * hash) + getContext().hashCode();
+    }
+    if (hasFeatureLists()) {
+      hash = (37 * hash) + FEATURE_LISTS_FIELD_NUMBER;
+      hash = (53 * hash) + getFeatureLists().hashCode();
+    }
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.example.SequenceExample parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.example.SequenceExample parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.example.SequenceExample parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.example.SequenceExample parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.example.SequenceExample parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.example.SequenceExample parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.example.SequenceExample parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.example.SequenceExample parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.example.SequenceExample parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.example.SequenceExample parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.example.SequenceExample parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.example.SequenceExample parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.example.SequenceExample prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * Protobuf type {@code tensorflow.SequenceExample}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.SequenceExample)
+      org.tensorflow.proto.example.SequenceExampleOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.example.ExampleProtos.internal_static_tensorflow_SequenceExample_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.example.ExampleProtos.internal_static_tensorflow_SequenceExample_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.example.SequenceExample.class, org.tensorflow.proto.example.SequenceExample.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.example.SequenceExample.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      if (contextBuilder_ == null) {
+        context_ = null;
+      } else {
+        context_ = null;
+        contextBuilder_ = null;
+      }
+      if (featureListsBuilder_ == null) {
+        featureLists_ = null;
+      } else {
+        featureLists_ = null;
+        featureListsBuilder_ = null;
+      }
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.example.ExampleProtos.internal_static_tensorflow_SequenceExample_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.example.SequenceExample getDefaultInstanceForType() {
+      return org.tensorflow.proto.example.SequenceExample.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.example.SequenceExample build() {
+      org.tensorflow.proto.example.SequenceExample result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.example.SequenceExample buildPartial() {
+      org.tensorflow.proto.example.SequenceExample result = new org.tensorflow.proto.example.SequenceExample(this);
+      if (contextBuilder_ == null) {
+        result.context_ = context_;
+      } else {
+        result.context_ = contextBuilder_.build();
+      }
+      if (featureListsBuilder_ == null) {
+        result.featureLists_ = featureLists_;
+      } else {
+        result.featureLists_ = featureListsBuilder_.build();
+      }
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.example.SequenceExample) {
+        return mergeFrom((org.tensorflow.proto.example.SequenceExample)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.example.SequenceExample other) {
+      if (other == org.tensorflow.proto.example.SequenceExample.getDefaultInstance()) return this;
+      if (other.hasContext()) {
+        mergeContext(other.getContext());
+      }
+      if (other.hasFeatureLists()) {
+        mergeFeatureLists(other.getFeatureLists());
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.example.SequenceExample parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.example.SequenceExample) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+
+    private org.tensorflow.proto.example.Features context_;
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.example.Features, org.tensorflow.proto.example.Features.Builder, org.tensorflow.proto.example.FeaturesOrBuilder> contextBuilder_;
+    /**
+     * <code>.tensorflow.Features context = 1;</code>
+     */
+    public boolean hasContext() {
+      return contextBuilder_ != null || context_ != null;
+    }
+    /**
+     * <code>.tensorflow.Features context = 1;</code>
+     */
+    public org.tensorflow.proto.example.Features getContext() {
+      if (contextBuilder_ == null) {
+        return context_ == null ? org.tensorflow.proto.example.Features.getDefaultInstance() : context_;
+      } else {
+        return contextBuilder_.getMessage();
+      }
+    }
+    /**
+     * <code>.tensorflow.Features context = 1;</code>
+     */
+    public Builder setContext(org.tensorflow.proto.example.Features value) {
+      if (contextBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        context_ = value;
+        onChanged();
+      } else {
+        contextBuilder_.setMessage(value);
+      }
+
+      return this;
+    }
+    /**
+     * <code>.tensorflow.Features context = 1;</code>
+     */
+    public Builder setContext(
+        org.tensorflow.proto.example.Features.Builder builderForValue) {
+      if (contextBuilder_ == null) {
+        context_ = builderForValue.build();
+        onChanged();
+      } else {
+        contextBuilder_.setMessage(builderForValue.build());
+      }
+
+      return this;
+    }
+    /**
+     * <code>.tensorflow.Features context = 1;</code>
+     */
+    public Builder mergeContext(org.tensorflow.proto.example.Features value) {
+      if (contextBuilder_ == null) {
+        if (context_ != null) {
+          context_ =
+            org.tensorflow.proto.example.Features.newBuilder(context_).mergeFrom(value).buildPartial();
+        } else {
+          context_ = value;
+        }
+        onChanged();
+      } else {
+        contextBuilder_.mergeFrom(value);
+      }
+
+      return this;
+    }
+    /**
+     * <code>.tensorflow.Features context = 1;</code>
+     */
+    public Builder clearContext() {
+      if (contextBuilder_ == null) {
+        context_ = null;
+        onChanged();
+      } else {
+        context_ = null;
+        contextBuilder_ = null;
+      }
+
+      return this;
+    }
+    /**
+     * <code>.tensorflow.Features context = 1;</code>
+     */
+    public org.tensorflow.proto.example.Features.Builder getContextBuilder() {
+      
+      onChanged();
+      return getContextFieldBuilder().getBuilder();
+    }
+    /**
+     * <code>.tensorflow.Features context = 1;</code>
+     */
+    public org.tensorflow.proto.example.FeaturesOrBuilder getContextOrBuilder() {
+      if (contextBuilder_ != null) {
+        return contextBuilder_.getMessageOrBuilder();
+      } else {
+        return context_ == null ?
+            org.tensorflow.proto.example.Features.getDefaultInstance() : context_;
+      }
+    }
+    /**
+     * <code>.tensorflow.Features context = 1;</code>
+     */
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.example.Features, org.tensorflow.proto.example.Features.Builder, org.tensorflow.proto.example.FeaturesOrBuilder> 
+        getContextFieldBuilder() {
+      if (contextBuilder_ == null) {
+        contextBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+            org.tensorflow.proto.example.Features, org.tensorflow.proto.example.Features.Builder, org.tensorflow.proto.example.FeaturesOrBuilder>(
+                getContext(),
+                getParentForChildren(),
+                isClean());
+        context_ = null;
+      }
+      return contextBuilder_;
+    }
+
+    private org.tensorflow.proto.example.FeatureLists featureLists_;
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.example.FeatureLists, org.tensorflow.proto.example.FeatureLists.Builder, org.tensorflow.proto.example.FeatureListsOrBuilder> featureListsBuilder_;
+    /**
+     * <code>.tensorflow.FeatureLists feature_lists = 2;</code>
+     */
+    public boolean hasFeatureLists() {
+      return featureListsBuilder_ != null || featureLists_ != null;
+    }
+    /**
+     * <code>.tensorflow.FeatureLists feature_lists = 2;</code>
+     */
+    public org.tensorflow.proto.example.FeatureLists getFeatureLists() {
+      if (featureListsBuilder_ == null) {
+        return featureLists_ == null ? org.tensorflow.proto.example.FeatureLists.getDefaultInstance() : featureLists_;
+      } else {
+        return featureListsBuilder_.getMessage();
+      }
+    }
+    /**
+     * <code>.tensorflow.FeatureLists feature_lists = 2;</code>
+     */
+    public Builder setFeatureLists(org.tensorflow.proto.example.FeatureLists value) {
+      if (featureListsBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        featureLists_ = value;
+        onChanged();
+      } else {
+        featureListsBuilder_.setMessage(value);
+      }
+
+      return this;
+    }
+    /**
+     * <code>.tensorflow.FeatureLists feature_lists = 2;</code>
+     */
+    public Builder setFeatureLists(
+        org.tensorflow.proto.example.FeatureLists.Builder builderForValue) {
+      if (featureListsBuilder_ == null) {
+        featureLists_ = builderForValue.build();
+        onChanged();
+      } else {
+        featureListsBuilder_.setMessage(builderForValue.build());
+      }
+
+      return this;
+    }
+    /**
+     * <code>.tensorflow.FeatureLists feature_lists = 2;</code>
+     */
+    public Builder mergeFeatureLists(org.tensorflow.proto.example.FeatureLists value) {
+      if (featureListsBuilder_ == null) {
+        if (featureLists_ != null) {
+          featureLists_ =
+            org.tensorflow.proto.example.FeatureLists.newBuilder(featureLists_).mergeFrom(value).buildPartial();
+        } else {
+          featureLists_ = value;
+        }
+        onChanged();
+      } else {
+        featureListsBuilder_.mergeFrom(value);
+      }
+
+      return this;
+    }
+    /**
+     * <code>.tensorflow.FeatureLists feature_lists = 2;</code>
+     */
+    public Builder clearFeatureLists() {
+      if (featureListsBuilder_ == null) {
+        featureLists_ = null;
+        onChanged();
+      } else {
+        featureLists_ = null;
+        featureListsBuilder_ = null;
+      }
+
+      return this;
+    }
+    /**
+     * <code>.tensorflow.FeatureLists feature_lists = 2;</code>
+     */
+    public org.tensorflow.proto.example.FeatureLists.Builder getFeatureListsBuilder() {
+      
+      onChanged();
+      return getFeatureListsFieldBuilder().getBuilder();
+    }
+    /**
+     * <code>.tensorflow.FeatureLists feature_lists = 2;</code>
+     */
+    public org.tensorflow.proto.example.FeatureListsOrBuilder getFeatureListsOrBuilder() {
+      if (featureListsBuilder_ != null) {
+        return featureListsBuilder_.getMessageOrBuilder();
+      } else {
+        return featureLists_ == null ?
+            org.tensorflow.proto.example.FeatureLists.getDefaultInstance() : featureLists_;
+      }
+    }
+    /**
+     * <code>.tensorflow.FeatureLists feature_lists = 2;</code>
+     */
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.example.FeatureLists, org.tensorflow.proto.example.FeatureLists.Builder, org.tensorflow.proto.example.FeatureListsOrBuilder> 
+        getFeatureListsFieldBuilder() {
+      if (featureListsBuilder_ == null) {
+        featureListsBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+            org.tensorflow.proto.example.FeatureLists, org.tensorflow.proto.example.FeatureLists.Builder, org.tensorflow.proto.example.FeatureListsOrBuilder>(
+                getFeatureLists(),
+                getParentForChildren(),
+                isClean());
+        featureLists_ = null;
+      }
+      return featureListsBuilder_;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.SequenceExample)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.SequenceExample)
+  private static final org.tensorflow.proto.example.SequenceExample DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.example.SequenceExample();
+  }
+
+  public static org.tensorflow.proto.example.SequenceExample getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<SequenceExample>
+      PARSER = new com.google.protobuf.AbstractParser<SequenceExample>() {
+    @java.lang.Override
+    public SequenceExample parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new SequenceExample(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<SequenceExample> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<SequenceExample> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.example.SequenceExample getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/SequenceExampleOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/SequenceExampleOrBuilder.java
new file mode 100644
index 00000000000..9b849c8c3c2
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/SequenceExampleOrBuilder.java
@@ -0,0 +1,35 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/example/example.proto
+
+package org.tensorflow.proto.example;
+
+public interface SequenceExampleOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.SequenceExample)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <code>.tensorflow.Features context = 1;</code>
+   */
+  boolean hasContext();
+  /**
+   * <code>.tensorflow.Features context = 1;</code>
+   */
+  org.tensorflow.proto.example.Features getContext();
+  /**
+   * <code>.tensorflow.Features context = 1;</code>
+   */
+  org.tensorflow.proto.example.FeaturesOrBuilder getContextOrBuilder();
+
+  /**
+   * <code>.tensorflow.FeatureLists feature_lists = 2;</code>
+   */
+  boolean hasFeatureLists();
+  /**
+   * <code>.tensorflow.FeatureLists feature_lists = 2;</code>
+   */
+  org.tensorflow.proto.example.FeatureLists getFeatureLists();
+  /**
+   * <code>.tensorflow.FeatureLists feature_lists = 2;</code>
+   */
+  org.tensorflow.proto.example.FeatureListsOrBuilder getFeatureListsOrBuilder();
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/VarLenFeatureProto.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/VarLenFeatureProto.java
new file mode 100644
index 00000000000..8edd05b1f24
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/VarLenFeatureProto.java
@@ -0,0 +1,885 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/example/example_parser_configuration.proto
+
+package org.tensorflow.proto.example;
+
+/**
+ * Protobuf type {@code tensorflow.VarLenFeatureProto}
+ */
+public  final class VarLenFeatureProto extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.VarLenFeatureProto)
+    VarLenFeatureProtoOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use VarLenFeatureProto.newBuilder() to construct.
+  private VarLenFeatureProto(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private VarLenFeatureProto() {
+    dtype_ = 0;
+    valuesOutputTensorName_ = "";
+    indicesOutputTensorName_ = "";
+    shapesOutputTensorName_ = "";
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new VarLenFeatureProto();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private VarLenFeatureProto(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 8: {
+            int rawValue = input.readEnum();
+
+            dtype_ = rawValue;
+            break;
+          }
+          case 18: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            valuesOutputTensorName_ = s;
+            break;
+          }
+          case 26: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            indicesOutputTensorName_ = s;
+            break;
+          }
+          case 34: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            shapesOutputTensorName_ = s;
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.example.ExampleParserConfigurationProtos.internal_static_tensorflow_VarLenFeatureProto_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.example.ExampleParserConfigurationProtos.internal_static_tensorflow_VarLenFeatureProto_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.example.VarLenFeatureProto.class, org.tensorflow.proto.example.VarLenFeatureProto.Builder.class);
+  }
+
+  public static final int DTYPE_FIELD_NUMBER = 1;
+  private int dtype_;
+  /**
+   * <code>.tensorflow.DataType dtype = 1;</code>
+   */
+  public int getDtypeValue() {
+    return dtype_;
+  }
+  /**
+   * <code>.tensorflow.DataType dtype = 1;</code>
+   */
+  public org.tensorflow.proto.framework.DataType getDtype() {
+    @SuppressWarnings("deprecation")
+    org.tensorflow.proto.framework.DataType result = org.tensorflow.proto.framework.DataType.valueOf(dtype_);
+    return result == null ? org.tensorflow.proto.framework.DataType.UNRECOGNIZED : result;
+  }
+
+  public static final int VALUES_OUTPUT_TENSOR_NAME_FIELD_NUMBER = 2;
+  private volatile java.lang.Object valuesOutputTensorName_;
+  /**
+   * <code>string values_output_tensor_name = 2;</code>
+   */
+  public java.lang.String getValuesOutputTensorName() {
+    java.lang.Object ref = valuesOutputTensorName_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      valuesOutputTensorName_ = s;
+      return s;
+    }
+  }
+  /**
+   * <code>string values_output_tensor_name = 2;</code>
+   */
+  public com.google.protobuf.ByteString
+      getValuesOutputTensorNameBytes() {
+    java.lang.Object ref = valuesOutputTensorName_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      valuesOutputTensorName_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int INDICES_OUTPUT_TENSOR_NAME_FIELD_NUMBER = 3;
+  private volatile java.lang.Object indicesOutputTensorName_;
+  /**
+   * <code>string indices_output_tensor_name = 3;</code>
+   */
+  public java.lang.String getIndicesOutputTensorName() {
+    java.lang.Object ref = indicesOutputTensorName_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      indicesOutputTensorName_ = s;
+      return s;
+    }
+  }
+  /**
+   * <code>string indices_output_tensor_name = 3;</code>
+   */
+  public com.google.protobuf.ByteString
+      getIndicesOutputTensorNameBytes() {
+    java.lang.Object ref = indicesOutputTensorName_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      indicesOutputTensorName_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int SHAPES_OUTPUT_TENSOR_NAME_FIELD_NUMBER = 4;
+  private volatile java.lang.Object shapesOutputTensorName_;
+  /**
+   * <code>string shapes_output_tensor_name = 4;</code>
+   */
+  public java.lang.String getShapesOutputTensorName() {
+    java.lang.Object ref = shapesOutputTensorName_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      shapesOutputTensorName_ = s;
+      return s;
+    }
+  }
+  /**
+   * <code>string shapes_output_tensor_name = 4;</code>
+   */
+  public com.google.protobuf.ByteString
+      getShapesOutputTensorNameBytes() {
+    java.lang.Object ref = shapesOutputTensorName_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      shapesOutputTensorName_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    if (dtype_ != org.tensorflow.proto.framework.DataType.DT_INVALID.getNumber()) {
+      output.writeEnum(1, dtype_);
+    }
+    if (!getValuesOutputTensorNameBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 2, valuesOutputTensorName_);
+    }
+    if (!getIndicesOutputTensorNameBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 3, indicesOutputTensorName_);
+    }
+    if (!getShapesOutputTensorNameBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 4, shapesOutputTensorName_);
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (dtype_ != org.tensorflow.proto.framework.DataType.DT_INVALID.getNumber()) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeEnumSize(1, dtype_);
+    }
+    if (!getValuesOutputTensorNameBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(2, valuesOutputTensorName_);
+    }
+    if (!getIndicesOutputTensorNameBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(3, indicesOutputTensorName_);
+    }
+    if (!getShapesOutputTensorNameBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(4, shapesOutputTensorName_);
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.example.VarLenFeatureProto)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.example.VarLenFeatureProto other = (org.tensorflow.proto.example.VarLenFeatureProto) obj;
+
+    if (dtype_ != other.dtype_) return false;
+    if (!getValuesOutputTensorName()
+        .equals(other.getValuesOutputTensorName())) return false;
+    if (!getIndicesOutputTensorName()
+        .equals(other.getIndicesOutputTensorName())) return false;
+    if (!getShapesOutputTensorName()
+        .equals(other.getShapesOutputTensorName())) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    hash = (37 * hash) + DTYPE_FIELD_NUMBER;
+    hash = (53 * hash) + dtype_;
+    hash = (37 * hash) + VALUES_OUTPUT_TENSOR_NAME_FIELD_NUMBER;
+    hash = (53 * hash) + getValuesOutputTensorName().hashCode();
+    hash = (37 * hash) + INDICES_OUTPUT_TENSOR_NAME_FIELD_NUMBER;
+    hash = (53 * hash) + getIndicesOutputTensorName().hashCode();
+    hash = (37 * hash) + SHAPES_OUTPUT_TENSOR_NAME_FIELD_NUMBER;
+    hash = (53 * hash) + getShapesOutputTensorName().hashCode();
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.example.VarLenFeatureProto parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.example.VarLenFeatureProto parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.example.VarLenFeatureProto parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.example.VarLenFeatureProto parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.example.VarLenFeatureProto parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.example.VarLenFeatureProto parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.example.VarLenFeatureProto parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.example.VarLenFeatureProto parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.example.VarLenFeatureProto parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.example.VarLenFeatureProto parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.example.VarLenFeatureProto parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.example.VarLenFeatureProto parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.example.VarLenFeatureProto prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * Protobuf type {@code tensorflow.VarLenFeatureProto}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.VarLenFeatureProto)
+      org.tensorflow.proto.example.VarLenFeatureProtoOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.example.ExampleParserConfigurationProtos.internal_static_tensorflow_VarLenFeatureProto_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.example.ExampleParserConfigurationProtos.internal_static_tensorflow_VarLenFeatureProto_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.example.VarLenFeatureProto.class, org.tensorflow.proto.example.VarLenFeatureProto.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.example.VarLenFeatureProto.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      dtype_ = 0;
+
+      valuesOutputTensorName_ = "";
+
+      indicesOutputTensorName_ = "";
+
+      shapesOutputTensorName_ = "";
+
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.example.ExampleParserConfigurationProtos.internal_static_tensorflow_VarLenFeatureProto_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.example.VarLenFeatureProto getDefaultInstanceForType() {
+      return org.tensorflow.proto.example.VarLenFeatureProto.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.example.VarLenFeatureProto build() {
+      org.tensorflow.proto.example.VarLenFeatureProto result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.example.VarLenFeatureProto buildPartial() {
+      org.tensorflow.proto.example.VarLenFeatureProto result = new org.tensorflow.proto.example.VarLenFeatureProto(this);
+      result.dtype_ = dtype_;
+      result.valuesOutputTensorName_ = valuesOutputTensorName_;
+      result.indicesOutputTensorName_ = indicesOutputTensorName_;
+      result.shapesOutputTensorName_ = shapesOutputTensorName_;
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.example.VarLenFeatureProto) {
+        return mergeFrom((org.tensorflow.proto.example.VarLenFeatureProto)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.example.VarLenFeatureProto other) {
+      if (other == org.tensorflow.proto.example.VarLenFeatureProto.getDefaultInstance()) return this;
+      if (other.dtype_ != 0) {
+        setDtypeValue(other.getDtypeValue());
+      }
+      if (!other.getValuesOutputTensorName().isEmpty()) {
+        valuesOutputTensorName_ = other.valuesOutputTensorName_;
+        onChanged();
+      }
+      if (!other.getIndicesOutputTensorName().isEmpty()) {
+        indicesOutputTensorName_ = other.indicesOutputTensorName_;
+        onChanged();
+      }
+      if (!other.getShapesOutputTensorName().isEmpty()) {
+        shapesOutputTensorName_ = other.shapesOutputTensorName_;
+        onChanged();
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.example.VarLenFeatureProto parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.example.VarLenFeatureProto) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+
+    private int dtype_ = 0;
+    /**
+     * <code>.tensorflow.DataType dtype = 1;</code>
+     */
+    public int getDtypeValue() {
+      return dtype_;
+    }
+    /**
+     * <code>.tensorflow.DataType dtype = 1;</code>
+     */
+    public Builder setDtypeValue(int value) {
+      dtype_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>.tensorflow.DataType dtype = 1;</code>
+     */
+    public org.tensorflow.proto.framework.DataType getDtype() {
+      @SuppressWarnings("deprecation")
+      org.tensorflow.proto.framework.DataType result = org.tensorflow.proto.framework.DataType.valueOf(dtype_);
+      return result == null ? org.tensorflow.proto.framework.DataType.UNRECOGNIZED : result;
+    }
+    /**
+     * <code>.tensorflow.DataType dtype = 1;</code>
+     */
+    public Builder setDtype(org.tensorflow.proto.framework.DataType value) {
+      if (value == null) {
+        throw new NullPointerException();
+      }
+      
+      dtype_ = value.getNumber();
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>.tensorflow.DataType dtype = 1;</code>
+     */
+    public Builder clearDtype() {
+      
+      dtype_ = 0;
+      onChanged();
+      return this;
+    }
+
+    private java.lang.Object valuesOutputTensorName_ = "";
+    /**
+     * <code>string values_output_tensor_name = 2;</code>
+     */
+    public java.lang.String getValuesOutputTensorName() {
+      java.lang.Object ref = valuesOutputTensorName_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        valuesOutputTensorName_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <code>string values_output_tensor_name = 2;</code>
+     */
+    public com.google.protobuf.ByteString
+        getValuesOutputTensorNameBytes() {
+      java.lang.Object ref = valuesOutputTensorName_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        valuesOutputTensorName_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <code>string values_output_tensor_name = 2;</code>
+     */
+    public Builder setValuesOutputTensorName(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      valuesOutputTensorName_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>string values_output_tensor_name = 2;</code>
+     */
+    public Builder clearValuesOutputTensorName() {
+      
+      valuesOutputTensorName_ = getDefaultInstance().getValuesOutputTensorName();
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>string values_output_tensor_name = 2;</code>
+     */
+    public Builder setValuesOutputTensorNameBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      valuesOutputTensorName_ = value;
+      onChanged();
+      return this;
+    }
+
+    private java.lang.Object indicesOutputTensorName_ = "";
+    /**
+     * <code>string indices_output_tensor_name = 3;</code>
+     */
+    public java.lang.String getIndicesOutputTensorName() {
+      java.lang.Object ref = indicesOutputTensorName_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        indicesOutputTensorName_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <code>string indices_output_tensor_name = 3;</code>
+     */
+    public com.google.protobuf.ByteString
+        getIndicesOutputTensorNameBytes() {
+      java.lang.Object ref = indicesOutputTensorName_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        indicesOutputTensorName_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <code>string indices_output_tensor_name = 3;</code>
+     */
+    public Builder setIndicesOutputTensorName(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      indicesOutputTensorName_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>string indices_output_tensor_name = 3;</code>
+     */
+    public Builder clearIndicesOutputTensorName() {
+      
+      indicesOutputTensorName_ = getDefaultInstance().getIndicesOutputTensorName();
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>string indices_output_tensor_name = 3;</code>
+     */
+    public Builder setIndicesOutputTensorNameBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      indicesOutputTensorName_ = value;
+      onChanged();
+      return this;
+    }
+
+    private java.lang.Object shapesOutputTensorName_ = "";
+    /**
+     * <code>string shapes_output_tensor_name = 4;</code>
+     */
+    public java.lang.String getShapesOutputTensorName() {
+      java.lang.Object ref = shapesOutputTensorName_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        shapesOutputTensorName_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <code>string shapes_output_tensor_name = 4;</code>
+     */
+    public com.google.protobuf.ByteString
+        getShapesOutputTensorNameBytes() {
+      java.lang.Object ref = shapesOutputTensorName_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        shapesOutputTensorName_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <code>string shapes_output_tensor_name = 4;</code>
+     */
+    public Builder setShapesOutputTensorName(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      shapesOutputTensorName_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>string shapes_output_tensor_name = 4;</code>
+     */
+    public Builder clearShapesOutputTensorName() {
+      
+      shapesOutputTensorName_ = getDefaultInstance().getShapesOutputTensorName();
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>string shapes_output_tensor_name = 4;</code>
+     */
+    public Builder setShapesOutputTensorNameBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      shapesOutputTensorName_ = value;
+      onChanged();
+      return this;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.VarLenFeatureProto)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.VarLenFeatureProto)
+  private static final org.tensorflow.proto.example.VarLenFeatureProto DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.example.VarLenFeatureProto();
+  }
+
+  public static org.tensorflow.proto.example.VarLenFeatureProto getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<VarLenFeatureProto>
+      PARSER = new com.google.protobuf.AbstractParser<VarLenFeatureProto>() {
+    @java.lang.Override
+    public VarLenFeatureProto parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new VarLenFeatureProto(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<VarLenFeatureProto> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<VarLenFeatureProto> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.example.VarLenFeatureProto getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/VarLenFeatureProtoOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/VarLenFeatureProtoOrBuilder.java
new file mode 100644
index 00000000000..820c2faa8f8
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/example/VarLenFeatureProtoOrBuilder.java
@@ -0,0 +1,48 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/example/example_parser_configuration.proto
+
+package org.tensorflow.proto.example;
+
+public interface VarLenFeatureProtoOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.VarLenFeatureProto)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <code>.tensorflow.DataType dtype = 1;</code>
+   */
+  int getDtypeValue();
+  /**
+   * <code>.tensorflow.DataType dtype = 1;</code>
+   */
+  org.tensorflow.proto.framework.DataType getDtype();
+
+  /**
+   * <code>string values_output_tensor_name = 2;</code>
+   */
+  java.lang.String getValuesOutputTensorName();
+  /**
+   * <code>string values_output_tensor_name = 2;</code>
+   */
+  com.google.protobuf.ByteString
+      getValuesOutputTensorNameBytes();
+
+  /**
+   * <code>string indices_output_tensor_name = 3;</code>
+   */
+  java.lang.String getIndicesOutputTensorName();
+  /**
+   * <code>string indices_output_tensor_name = 3;</code>
+   */
+  com.google.protobuf.ByteString
+      getIndicesOutputTensorNameBytes();
+
+  /**
+   * <code>string shapes_output_tensor_name = 4;</code>
+   */
+  java.lang.String getShapesOutputTensorName();
+  /**
+   * <code>string shapes_output_tensor_name = 4;</code>
+   */
+  com.google.protobuf.ByteString
+      getShapesOutputTensorNameBytes();
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/AllocationDescription.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/AllocationDescription.java
new file mode 100644
index 00000000000..2abffe7549a
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/AllocationDescription.java
@@ -0,0 +1,944 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/framework/allocation_description.proto
+
+package org.tensorflow.proto.framework;
+
+/**
+ * Protobuf type {@code tensorflow.AllocationDescription}
+ */
+public  final class AllocationDescription extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.AllocationDescription)
+    AllocationDescriptionOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use AllocationDescription.newBuilder() to construct.
+  private AllocationDescription(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private AllocationDescription() {
+    allocatorName_ = "";
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new AllocationDescription();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private AllocationDescription(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 8: {
+
+            requestedBytes_ = input.readInt64();
+            break;
+          }
+          case 16: {
+
+            allocatedBytes_ = input.readInt64();
+            break;
+          }
+          case 26: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            allocatorName_ = s;
+            break;
+          }
+          case 32: {
+
+            allocationId_ = input.readInt64();
+            break;
+          }
+          case 40: {
+
+            hasSingleReference_ = input.readBool();
+            break;
+          }
+          case 48: {
+
+            ptr_ = input.readUInt64();
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.framework.AllocationDescriptionProtos.internal_static_tensorflow_AllocationDescription_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.framework.AllocationDescriptionProtos.internal_static_tensorflow_AllocationDescription_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.framework.AllocationDescription.class, org.tensorflow.proto.framework.AllocationDescription.Builder.class);
+  }
+
+  public static final int REQUESTED_BYTES_FIELD_NUMBER = 1;
+  private long requestedBytes_;
+  /**
+   * <pre>
+   * Total number of bytes requested
+   * </pre>
+   *
+   * <code>int64 requested_bytes = 1;</code>
+   */
+  public long getRequestedBytes() {
+    return requestedBytes_;
+  }
+
+  public static final int ALLOCATED_BYTES_FIELD_NUMBER = 2;
+  private long allocatedBytes_;
+  /**
+   * <pre>
+   * Total number of bytes allocated if known
+   * </pre>
+   *
+   * <code>int64 allocated_bytes = 2;</code>
+   */
+  public long getAllocatedBytes() {
+    return allocatedBytes_;
+  }
+
+  public static final int ALLOCATOR_NAME_FIELD_NUMBER = 3;
+  private volatile java.lang.Object allocatorName_;
+  /**
+   * <pre>
+   * Name of the allocator used
+   * </pre>
+   *
+   * <code>string allocator_name = 3;</code>
+   */
+  public java.lang.String getAllocatorName() {
+    java.lang.Object ref = allocatorName_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      allocatorName_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * Name of the allocator used
+   * </pre>
+   *
+   * <code>string allocator_name = 3;</code>
+   */
+  public com.google.protobuf.ByteString
+      getAllocatorNameBytes() {
+    java.lang.Object ref = allocatorName_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      allocatorName_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int ALLOCATION_ID_FIELD_NUMBER = 4;
+  private long allocationId_;
+  /**
+   * <pre>
+   * Identifier of the allocated buffer if known
+   * </pre>
+   *
+   * <code>int64 allocation_id = 4;</code>
+   */
+  public long getAllocationId() {
+    return allocationId_;
+  }
+
+  public static final int HAS_SINGLE_REFERENCE_FIELD_NUMBER = 5;
+  private boolean hasSingleReference_;
+  /**
+   * <pre>
+   * Set if this tensor only has one remaining reference
+   * </pre>
+   *
+   * <code>bool has_single_reference = 5;</code>
+   */
+  public boolean getHasSingleReference() {
+    return hasSingleReference_;
+  }
+
+  public static final int PTR_FIELD_NUMBER = 6;
+  private long ptr_;
+  /**
+   * <pre>
+   * Address of the allocation.
+   * </pre>
+   *
+   * <code>uint64 ptr = 6;</code>
+   */
+  public long getPtr() {
+    return ptr_;
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    if (requestedBytes_ != 0L) {
+      output.writeInt64(1, requestedBytes_);
+    }
+    if (allocatedBytes_ != 0L) {
+      output.writeInt64(2, allocatedBytes_);
+    }
+    if (!getAllocatorNameBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 3, allocatorName_);
+    }
+    if (allocationId_ != 0L) {
+      output.writeInt64(4, allocationId_);
+    }
+    if (hasSingleReference_ != false) {
+      output.writeBool(5, hasSingleReference_);
+    }
+    if (ptr_ != 0L) {
+      output.writeUInt64(6, ptr_);
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (requestedBytes_ != 0L) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt64Size(1, requestedBytes_);
+    }
+    if (allocatedBytes_ != 0L) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt64Size(2, allocatedBytes_);
+    }
+    if (!getAllocatorNameBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(3, allocatorName_);
+    }
+    if (allocationId_ != 0L) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt64Size(4, allocationId_);
+    }
+    if (hasSingleReference_ != false) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeBoolSize(5, hasSingleReference_);
+    }
+    if (ptr_ != 0L) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeUInt64Size(6, ptr_);
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.framework.AllocationDescription)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.framework.AllocationDescription other = (org.tensorflow.proto.framework.AllocationDescription) obj;
+
+    if (getRequestedBytes()
+        != other.getRequestedBytes()) return false;
+    if (getAllocatedBytes()
+        != other.getAllocatedBytes()) return false;
+    if (!getAllocatorName()
+        .equals(other.getAllocatorName())) return false;
+    if (getAllocationId()
+        != other.getAllocationId()) return false;
+    if (getHasSingleReference()
+        != other.getHasSingleReference()) return false;
+    if (getPtr()
+        != other.getPtr()) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    hash = (37 * hash) + REQUESTED_BYTES_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+        getRequestedBytes());
+    hash = (37 * hash) + ALLOCATED_BYTES_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+        getAllocatedBytes());
+    hash = (37 * hash) + ALLOCATOR_NAME_FIELD_NUMBER;
+    hash = (53 * hash) + getAllocatorName().hashCode();
+    hash = (37 * hash) + ALLOCATION_ID_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+        getAllocationId());
+    hash = (37 * hash) + HAS_SINGLE_REFERENCE_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashBoolean(
+        getHasSingleReference());
+    hash = (37 * hash) + PTR_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+        getPtr());
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.framework.AllocationDescription parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.AllocationDescription parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.AllocationDescription parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.AllocationDescription parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.AllocationDescription parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.AllocationDescription parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.AllocationDescription parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.AllocationDescription parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.AllocationDescription parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.AllocationDescription parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.AllocationDescription parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.AllocationDescription parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.framework.AllocationDescription prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * Protobuf type {@code tensorflow.AllocationDescription}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.AllocationDescription)
+      org.tensorflow.proto.framework.AllocationDescriptionOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.AllocationDescriptionProtos.internal_static_tensorflow_AllocationDescription_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.AllocationDescriptionProtos.internal_static_tensorflow_AllocationDescription_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.AllocationDescription.class, org.tensorflow.proto.framework.AllocationDescription.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.framework.AllocationDescription.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      requestedBytes_ = 0L;
+
+      allocatedBytes_ = 0L;
+
+      allocatorName_ = "";
+
+      allocationId_ = 0L;
+
+      hasSingleReference_ = false;
+
+      ptr_ = 0L;
+
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.framework.AllocationDescriptionProtos.internal_static_tensorflow_AllocationDescription_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.AllocationDescription getDefaultInstanceForType() {
+      return org.tensorflow.proto.framework.AllocationDescription.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.AllocationDescription build() {
+      org.tensorflow.proto.framework.AllocationDescription result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.AllocationDescription buildPartial() {
+      org.tensorflow.proto.framework.AllocationDescription result = new org.tensorflow.proto.framework.AllocationDescription(this);
+      result.requestedBytes_ = requestedBytes_;
+      result.allocatedBytes_ = allocatedBytes_;
+      result.allocatorName_ = allocatorName_;
+      result.allocationId_ = allocationId_;
+      result.hasSingleReference_ = hasSingleReference_;
+      result.ptr_ = ptr_;
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.framework.AllocationDescription) {
+        return mergeFrom((org.tensorflow.proto.framework.AllocationDescription)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.framework.AllocationDescription other) {
+      if (other == org.tensorflow.proto.framework.AllocationDescription.getDefaultInstance()) return this;
+      if (other.getRequestedBytes() != 0L) {
+        setRequestedBytes(other.getRequestedBytes());
+      }
+      if (other.getAllocatedBytes() != 0L) {
+        setAllocatedBytes(other.getAllocatedBytes());
+      }
+      if (!other.getAllocatorName().isEmpty()) {
+        allocatorName_ = other.allocatorName_;
+        onChanged();
+      }
+      if (other.getAllocationId() != 0L) {
+        setAllocationId(other.getAllocationId());
+      }
+      if (other.getHasSingleReference() != false) {
+        setHasSingleReference(other.getHasSingleReference());
+      }
+      if (other.getPtr() != 0L) {
+        setPtr(other.getPtr());
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.framework.AllocationDescription parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.framework.AllocationDescription) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+
+    private long requestedBytes_ ;
+    /**
+     * <pre>
+     * Total number of bytes requested
+     * </pre>
+     *
+     * <code>int64 requested_bytes = 1;</code>
+     */
+    public long getRequestedBytes() {
+      return requestedBytes_;
+    }
+    /**
+     * <pre>
+     * Total number of bytes requested
+     * </pre>
+     *
+     * <code>int64 requested_bytes = 1;</code>
+     */
+    public Builder setRequestedBytes(long value) {
+      
+      requestedBytes_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Total number of bytes requested
+     * </pre>
+     *
+     * <code>int64 requested_bytes = 1;</code>
+     */
+    public Builder clearRequestedBytes() {
+      
+      requestedBytes_ = 0L;
+      onChanged();
+      return this;
+    }
+
+    private long allocatedBytes_ ;
+    /**
+     * <pre>
+     * Total number of bytes allocated if known
+     * </pre>
+     *
+     * <code>int64 allocated_bytes = 2;</code>
+     */
+    public long getAllocatedBytes() {
+      return allocatedBytes_;
+    }
+    /**
+     * <pre>
+     * Total number of bytes allocated if known
+     * </pre>
+     *
+     * <code>int64 allocated_bytes = 2;</code>
+     */
+    public Builder setAllocatedBytes(long value) {
+      
+      allocatedBytes_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Total number of bytes allocated if known
+     * </pre>
+     *
+     * <code>int64 allocated_bytes = 2;</code>
+     */
+    public Builder clearAllocatedBytes() {
+      
+      allocatedBytes_ = 0L;
+      onChanged();
+      return this;
+    }
+
+    private java.lang.Object allocatorName_ = "";
+    /**
+     * <pre>
+     * Name of the allocator used
+     * </pre>
+     *
+     * <code>string allocator_name = 3;</code>
+     */
+    public java.lang.String getAllocatorName() {
+      java.lang.Object ref = allocatorName_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        allocatorName_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Name of the allocator used
+     * </pre>
+     *
+     * <code>string allocator_name = 3;</code>
+     */
+    public com.google.protobuf.ByteString
+        getAllocatorNameBytes() {
+      java.lang.Object ref = allocatorName_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        allocatorName_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Name of the allocator used
+     * </pre>
+     *
+     * <code>string allocator_name = 3;</code>
+     */
+    public Builder setAllocatorName(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      allocatorName_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Name of the allocator used
+     * </pre>
+     *
+     * <code>string allocator_name = 3;</code>
+     */
+    public Builder clearAllocatorName() {
+      
+      allocatorName_ = getDefaultInstance().getAllocatorName();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Name of the allocator used
+     * </pre>
+     *
+     * <code>string allocator_name = 3;</code>
+     */
+    public Builder setAllocatorNameBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      allocatorName_ = value;
+      onChanged();
+      return this;
+    }
+
+    private long allocationId_ ;
+    /**
+     * <pre>
+     * Identifier of the allocated buffer if known
+     * </pre>
+     *
+     * <code>int64 allocation_id = 4;</code>
+     */
+    public long getAllocationId() {
+      return allocationId_;
+    }
+    /**
+     * <pre>
+     * Identifier of the allocated buffer if known
+     * </pre>
+     *
+     * <code>int64 allocation_id = 4;</code>
+     */
+    public Builder setAllocationId(long value) {
+      
+      allocationId_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Identifier of the allocated buffer if known
+     * </pre>
+     *
+     * <code>int64 allocation_id = 4;</code>
+     */
+    public Builder clearAllocationId() {
+      
+      allocationId_ = 0L;
+      onChanged();
+      return this;
+    }
+
+    private boolean hasSingleReference_ ;
+    /**
+     * <pre>
+     * Set if this tensor only has one remaining reference
+     * </pre>
+     *
+     * <code>bool has_single_reference = 5;</code>
+     */
+    public boolean getHasSingleReference() {
+      return hasSingleReference_;
+    }
+    /**
+     * <pre>
+     * Set if this tensor only has one remaining reference
+     * </pre>
+     *
+     * <code>bool has_single_reference = 5;</code>
+     */
+    public Builder setHasSingleReference(boolean value) {
+      
+      hasSingleReference_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Set if this tensor only has one remaining reference
+     * </pre>
+     *
+     * <code>bool has_single_reference = 5;</code>
+     */
+    public Builder clearHasSingleReference() {
+      
+      hasSingleReference_ = false;
+      onChanged();
+      return this;
+    }
+
+    private long ptr_ ;
+    /**
+     * <pre>
+     * Address of the allocation.
+     * </pre>
+     *
+     * <code>uint64 ptr = 6;</code>
+     */
+    public long getPtr() {
+      return ptr_;
+    }
+    /**
+     * <pre>
+     * Address of the allocation.
+     * </pre>
+     *
+     * <code>uint64 ptr = 6;</code>
+     */
+    public Builder setPtr(long value) {
+      
+      ptr_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Address of the allocation.
+     * </pre>
+     *
+     * <code>uint64 ptr = 6;</code>
+     */
+    public Builder clearPtr() {
+      
+      ptr_ = 0L;
+      onChanged();
+      return this;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.AllocationDescription)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.AllocationDescription)
+  private static final org.tensorflow.proto.framework.AllocationDescription DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.AllocationDescription();
+  }
+
+  public static org.tensorflow.proto.framework.AllocationDescription getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<AllocationDescription>
+      PARSER = new com.google.protobuf.AbstractParser<AllocationDescription>() {
+    @java.lang.Override
+    public AllocationDescription parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new AllocationDescription(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<AllocationDescription> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<AllocationDescription> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.framework.AllocationDescription getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/AllocationDescriptionOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/AllocationDescriptionOrBuilder.java
new file mode 100644
index 00000000000..9fbd5df0695
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/AllocationDescriptionOrBuilder.java
@@ -0,0 +1,72 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/framework/allocation_description.proto
+
+package org.tensorflow.proto.framework;
+
+public interface AllocationDescriptionOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.AllocationDescription)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <pre>
+   * Total number of bytes requested
+   * </pre>
+   *
+   * <code>int64 requested_bytes = 1;</code>
+   */
+  long getRequestedBytes();
+
+  /**
+   * <pre>
+   * Total number of bytes allocated if known
+   * </pre>
+   *
+   * <code>int64 allocated_bytes = 2;</code>
+   */
+  long getAllocatedBytes();
+
+  /**
+   * <pre>
+   * Name of the allocator used
+   * </pre>
+   *
+   * <code>string allocator_name = 3;</code>
+   */
+  java.lang.String getAllocatorName();
+  /**
+   * <pre>
+   * Name of the allocator used
+   * </pre>
+   *
+   * <code>string allocator_name = 3;</code>
+   */
+  com.google.protobuf.ByteString
+      getAllocatorNameBytes();
+
+  /**
+   * <pre>
+   * Identifier of the allocated buffer if known
+   * </pre>
+   *
+   * <code>int64 allocation_id = 4;</code>
+   */
+  long getAllocationId();
+
+  /**
+   * <pre>
+   * Set if this tensor only has one remaining reference
+   * </pre>
+   *
+   * <code>bool has_single_reference = 5;</code>
+   */
+  boolean getHasSingleReference();
+
+  /**
+   * <pre>
+   * Address of the allocation.
+   * </pre>
+   *
+   * <code>uint64 ptr = 6;</code>
+   */
+  long getPtr();
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/AllocationDescriptionProtos.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/AllocationDescriptionProtos.java
new file mode 100644
index 00000000000..f3af7b380fc
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/AllocationDescriptionProtos.java
@@ -0,0 +1,56 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/framework/allocation_description.proto
+
+package org.tensorflow.proto.framework;
+
+public final class AllocationDescriptionProtos {
+  private AllocationDescriptionProtos() {}
+  public static void registerAllExtensions(
+      com.google.protobuf.ExtensionRegistryLite registry) {
+  }
+
+  public static void registerAllExtensions(
+      com.google.protobuf.ExtensionRegistry registry) {
+    registerAllExtensions(
+        (com.google.protobuf.ExtensionRegistryLite) registry);
+  }
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_AllocationDescription_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_AllocationDescription_fieldAccessorTable;
+
+  public static com.google.protobuf.Descriptors.FileDescriptor
+      getDescriptor() {
+    return descriptor;
+  }
+  private static  com.google.protobuf.Descriptors.FileDescriptor
+      descriptor;
+  static {
+    java.lang.String[] descriptorData = {
+      "\n6tensorflow/core/framework/allocation_d" +
+      "escription.proto\022\ntensorflow\"\243\001\n\025Allocat" +
+      "ionDescription\022\027\n\017requested_bytes\030\001 \001(\003\022" +
+      "\027\n\017allocated_bytes\030\002 \001(\003\022\026\n\016allocator_na" +
+      "me\030\003 \001(\t\022\025\n\rallocation_id\030\004 \001(\003\022\034\n\024has_s" +
+      "ingle_reference\030\005 \001(\010\022\013\n\003ptr\030\006 \001(\004B\241\001\n\036o" +
+      "rg.tensorflow.proto.frameworkB\033Allocatio" +
+      "nDescriptionProtosP\001Z]github.com/tensorf" +
+      "low/tensorflow/tensorflow/go/core/framew" +
+      "ork/allocation_description_go_proto\370\001\001b\006" +
+      "proto3"
+    };
+    descriptor = com.google.protobuf.Descriptors.FileDescriptor
+      .internalBuildGeneratedFileFrom(descriptorData,
+        new com.google.protobuf.Descriptors.FileDescriptor[] {
+        });
+    internal_static_tensorflow_AllocationDescription_descriptor =
+      getDescriptor().getMessageTypes().get(0);
+    internal_static_tensorflow_AllocationDescription_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_AllocationDescription_descriptor,
+        new java.lang.String[] { "RequestedBytes", "AllocatedBytes", "AllocatorName", "AllocationId", "HasSingleReference", "Ptr", });
+  }
+
+  // @@protoc_insertion_point(outer_class_scope)
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/AllocationRecord.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/AllocationRecord.java
new file mode 100644
index 00000000000..738f1117fbe
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/AllocationRecord.java
@@ -0,0 +1,575 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/framework/step_stats.proto
+
+package org.tensorflow.proto.framework;
+
+/**
+ * <pre>
+ * An allocation/de-allocation operation performed by the allocator.
+ * </pre>
+ *
+ * Protobuf type {@code tensorflow.AllocationRecord}
+ */
+public  final class AllocationRecord extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.AllocationRecord)
+    AllocationRecordOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use AllocationRecord.newBuilder() to construct.
+  private AllocationRecord(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private AllocationRecord() {
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new AllocationRecord();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private AllocationRecord(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 8: {
+
+            allocMicros_ = input.readInt64();
+            break;
+          }
+          case 16: {
+
+            allocBytes_ = input.readInt64();
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.framework.StepStatsProtos.internal_static_tensorflow_AllocationRecord_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.framework.StepStatsProtos.internal_static_tensorflow_AllocationRecord_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.framework.AllocationRecord.class, org.tensorflow.proto.framework.AllocationRecord.Builder.class);
+  }
+
+  public static final int ALLOC_MICROS_FIELD_NUMBER = 1;
+  private long allocMicros_;
+  /**
+   * <pre>
+   * The timestamp of the operation.
+   * </pre>
+   *
+   * <code>int64 alloc_micros = 1;</code>
+   */
+  public long getAllocMicros() {
+    return allocMicros_;
+  }
+
+  public static final int ALLOC_BYTES_FIELD_NUMBER = 2;
+  private long allocBytes_;
+  /**
+   * <pre>
+   * Number of bytes allocated, or de-allocated if negative.
+   * </pre>
+   *
+   * <code>int64 alloc_bytes = 2;</code>
+   */
+  public long getAllocBytes() {
+    return allocBytes_;
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    if (allocMicros_ != 0L) {
+      output.writeInt64(1, allocMicros_);
+    }
+    if (allocBytes_ != 0L) {
+      output.writeInt64(2, allocBytes_);
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (allocMicros_ != 0L) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt64Size(1, allocMicros_);
+    }
+    if (allocBytes_ != 0L) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt64Size(2, allocBytes_);
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.framework.AllocationRecord)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.framework.AllocationRecord other = (org.tensorflow.proto.framework.AllocationRecord) obj;
+
+    if (getAllocMicros()
+        != other.getAllocMicros()) return false;
+    if (getAllocBytes()
+        != other.getAllocBytes()) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    hash = (37 * hash) + ALLOC_MICROS_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+        getAllocMicros());
+    hash = (37 * hash) + ALLOC_BYTES_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+        getAllocBytes());
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.framework.AllocationRecord parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.AllocationRecord parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.AllocationRecord parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.AllocationRecord parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.AllocationRecord parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.AllocationRecord parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.AllocationRecord parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.AllocationRecord parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.AllocationRecord parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.AllocationRecord parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.AllocationRecord parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.AllocationRecord parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.framework.AllocationRecord prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * <pre>
+   * An allocation/de-allocation operation performed by the allocator.
+   * </pre>
+   *
+   * Protobuf type {@code tensorflow.AllocationRecord}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.AllocationRecord)
+      org.tensorflow.proto.framework.AllocationRecordOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.StepStatsProtos.internal_static_tensorflow_AllocationRecord_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.StepStatsProtos.internal_static_tensorflow_AllocationRecord_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.AllocationRecord.class, org.tensorflow.proto.framework.AllocationRecord.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.framework.AllocationRecord.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      allocMicros_ = 0L;
+
+      allocBytes_ = 0L;
+
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.framework.StepStatsProtos.internal_static_tensorflow_AllocationRecord_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.AllocationRecord getDefaultInstanceForType() {
+      return org.tensorflow.proto.framework.AllocationRecord.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.AllocationRecord build() {
+      org.tensorflow.proto.framework.AllocationRecord result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.AllocationRecord buildPartial() {
+      org.tensorflow.proto.framework.AllocationRecord result = new org.tensorflow.proto.framework.AllocationRecord(this);
+      result.allocMicros_ = allocMicros_;
+      result.allocBytes_ = allocBytes_;
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.framework.AllocationRecord) {
+        return mergeFrom((org.tensorflow.proto.framework.AllocationRecord)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.framework.AllocationRecord other) {
+      if (other == org.tensorflow.proto.framework.AllocationRecord.getDefaultInstance()) return this;
+      if (other.getAllocMicros() != 0L) {
+        setAllocMicros(other.getAllocMicros());
+      }
+      if (other.getAllocBytes() != 0L) {
+        setAllocBytes(other.getAllocBytes());
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.framework.AllocationRecord parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.framework.AllocationRecord) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+
+    private long allocMicros_ ;
+    /**
+     * <pre>
+     * The timestamp of the operation.
+     * </pre>
+     *
+     * <code>int64 alloc_micros = 1;</code>
+     */
+    public long getAllocMicros() {
+      return allocMicros_;
+    }
+    /**
+     * <pre>
+     * The timestamp of the operation.
+     * </pre>
+     *
+     * <code>int64 alloc_micros = 1;</code>
+     */
+    public Builder setAllocMicros(long value) {
+      
+      allocMicros_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * The timestamp of the operation.
+     * </pre>
+     *
+     * <code>int64 alloc_micros = 1;</code>
+     */
+    public Builder clearAllocMicros() {
+      
+      allocMicros_ = 0L;
+      onChanged();
+      return this;
+    }
+
+    private long allocBytes_ ;
+    /**
+     * <pre>
+     * Number of bytes allocated, or de-allocated if negative.
+     * </pre>
+     *
+     * <code>int64 alloc_bytes = 2;</code>
+     */
+    public long getAllocBytes() {
+      return allocBytes_;
+    }
+    /**
+     * <pre>
+     * Number of bytes allocated, or de-allocated if negative.
+     * </pre>
+     *
+     * <code>int64 alloc_bytes = 2;</code>
+     */
+    public Builder setAllocBytes(long value) {
+      
+      allocBytes_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Number of bytes allocated, or de-allocated if negative.
+     * </pre>
+     *
+     * <code>int64 alloc_bytes = 2;</code>
+     */
+    public Builder clearAllocBytes() {
+      
+      allocBytes_ = 0L;
+      onChanged();
+      return this;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.AllocationRecord)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.AllocationRecord)
+  private static final org.tensorflow.proto.framework.AllocationRecord DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.AllocationRecord();
+  }
+
+  public static org.tensorflow.proto.framework.AllocationRecord getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<AllocationRecord>
+      PARSER = new com.google.protobuf.AbstractParser<AllocationRecord>() {
+    @java.lang.Override
+    public AllocationRecord parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new AllocationRecord(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<AllocationRecord> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<AllocationRecord> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.framework.AllocationRecord getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/AllocationRecordOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/AllocationRecordOrBuilder.java
new file mode 100644
index 00000000000..c6c5dd502da
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/AllocationRecordOrBuilder.java
@@ -0,0 +1,27 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/framework/step_stats.proto
+
+package org.tensorflow.proto.framework;
+
+public interface AllocationRecordOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.AllocationRecord)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <pre>
+   * The timestamp of the operation.
+   * </pre>
+   *
+   * <code>int64 alloc_micros = 1;</code>
+   */
+  long getAllocMicros();
+
+  /**
+   * <pre>
+   * Number of bytes allocated, or de-allocated if negative.
+   * </pre>
+   *
+   * <code>int64 alloc_bytes = 2;</code>
+   */
+  long getAllocBytes();
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/AllocatorMemoryUsed.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/AllocatorMemoryUsed.java
new file mode 100644
index 00000000000..3cb33266367
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/AllocatorMemoryUsed.java
@@ -0,0 +1,1268 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/framework/step_stats.proto
+
+package org.tensorflow.proto.framework;
+
+/**
+ * Protobuf type {@code tensorflow.AllocatorMemoryUsed}
+ */
+public  final class AllocatorMemoryUsed extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.AllocatorMemoryUsed)
+    AllocatorMemoryUsedOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use AllocatorMemoryUsed.newBuilder() to construct.
+  private AllocatorMemoryUsed(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private AllocatorMemoryUsed() {
+    allocatorName_ = "";
+    allocationRecords_ = java.util.Collections.emptyList();
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new AllocatorMemoryUsed();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private AllocatorMemoryUsed(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    int mutable_bitField0_ = 0;
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 10: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            allocatorName_ = s;
+            break;
+          }
+          case 16: {
+
+            totalBytes_ = input.readInt64();
+            break;
+          }
+          case 24: {
+
+            peakBytes_ = input.readInt64();
+            break;
+          }
+          case 32: {
+
+            liveBytes_ = input.readInt64();
+            break;
+          }
+          case 40: {
+
+            allocatorBytesInUse_ = input.readInt64();
+            break;
+          }
+          case 50: {
+            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
+              allocationRecords_ = new java.util.ArrayList<org.tensorflow.proto.framework.AllocationRecord>();
+              mutable_bitField0_ |= 0x00000001;
+            }
+            allocationRecords_.add(
+                input.readMessage(org.tensorflow.proto.framework.AllocationRecord.parser(), extensionRegistry));
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      if (((mutable_bitField0_ & 0x00000001) != 0)) {
+        allocationRecords_ = java.util.Collections.unmodifiableList(allocationRecords_);
+      }
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.framework.StepStatsProtos.internal_static_tensorflow_AllocatorMemoryUsed_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.framework.StepStatsProtos.internal_static_tensorflow_AllocatorMemoryUsed_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.framework.AllocatorMemoryUsed.class, org.tensorflow.proto.framework.AllocatorMemoryUsed.Builder.class);
+  }
+
+  public static final int ALLOCATOR_NAME_FIELD_NUMBER = 1;
+  private volatile java.lang.Object allocatorName_;
+  /**
+   * <code>string allocator_name = 1;</code>
+   */
+  public java.lang.String getAllocatorName() {
+    java.lang.Object ref = allocatorName_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      allocatorName_ = s;
+      return s;
+    }
+  }
+  /**
+   * <code>string allocator_name = 1;</code>
+   */
+  public com.google.protobuf.ByteString
+      getAllocatorNameBytes() {
+    java.lang.Object ref = allocatorName_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      allocatorName_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int TOTAL_BYTES_FIELD_NUMBER = 2;
+  private long totalBytes_;
+  /**
+   * <pre>
+   * These are per-node allocator memory stats.
+   * </pre>
+   *
+   * <code>int64 total_bytes = 2;</code>
+   */
+  public long getTotalBytes() {
+    return totalBytes_;
+  }
+
+  public static final int PEAK_BYTES_FIELD_NUMBER = 3;
+  private long peakBytes_;
+  /**
+   * <code>int64 peak_bytes = 3;</code>
+   */
+  public long getPeakBytes() {
+    return peakBytes_;
+  }
+
+  public static final int LIVE_BYTES_FIELD_NUMBER = 4;
+  private long liveBytes_;
+  /**
+   * <pre>
+   * The bytes that are not deallocated.
+   * </pre>
+   *
+   * <code>int64 live_bytes = 4;</code>
+   */
+  public long getLiveBytes() {
+    return liveBytes_;
+  }
+
+  public static final int ALLOCATION_RECORDS_FIELD_NUMBER = 6;
+  private java.util.List<org.tensorflow.proto.framework.AllocationRecord> allocationRecords_;
+  /**
+   * <pre>
+   * The allocation and deallocation timeline.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.AllocationRecord allocation_records = 6;</code>
+   */
+  public java.util.List<org.tensorflow.proto.framework.AllocationRecord> getAllocationRecordsList() {
+    return allocationRecords_;
+  }
+  /**
+   * <pre>
+   * The allocation and deallocation timeline.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.AllocationRecord allocation_records = 6;</code>
+   */
+  public java.util.List<? extends org.tensorflow.proto.framework.AllocationRecordOrBuilder> 
+      getAllocationRecordsOrBuilderList() {
+    return allocationRecords_;
+  }
+  /**
+   * <pre>
+   * The allocation and deallocation timeline.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.AllocationRecord allocation_records = 6;</code>
+   */
+  public int getAllocationRecordsCount() {
+    return allocationRecords_.size();
+  }
+  /**
+   * <pre>
+   * The allocation and deallocation timeline.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.AllocationRecord allocation_records = 6;</code>
+   */
+  public org.tensorflow.proto.framework.AllocationRecord getAllocationRecords(int index) {
+    return allocationRecords_.get(index);
+  }
+  /**
+   * <pre>
+   * The allocation and deallocation timeline.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.AllocationRecord allocation_records = 6;</code>
+   */
+  public org.tensorflow.proto.framework.AllocationRecordOrBuilder getAllocationRecordsOrBuilder(
+      int index) {
+    return allocationRecords_.get(index);
+  }
+
+  public static final int ALLOCATOR_BYTES_IN_USE_FIELD_NUMBER = 5;
+  private long allocatorBytesInUse_;
+  /**
+   * <pre>
+   * These are snapshots of the overall allocator memory stats.
+   * The number of live bytes currently allocated by the allocator.
+   * </pre>
+   *
+   * <code>int64 allocator_bytes_in_use = 5;</code>
+   */
+  public long getAllocatorBytesInUse() {
+    return allocatorBytesInUse_;
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    if (!getAllocatorNameBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 1, allocatorName_);
+    }
+    if (totalBytes_ != 0L) {
+      output.writeInt64(2, totalBytes_);
+    }
+    if (peakBytes_ != 0L) {
+      output.writeInt64(3, peakBytes_);
+    }
+    if (liveBytes_ != 0L) {
+      output.writeInt64(4, liveBytes_);
+    }
+    if (allocatorBytesInUse_ != 0L) {
+      output.writeInt64(5, allocatorBytesInUse_);
+    }
+    for (int i = 0; i < allocationRecords_.size(); i++) {
+      output.writeMessage(6, allocationRecords_.get(i));
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (!getAllocatorNameBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, allocatorName_);
+    }
+    if (totalBytes_ != 0L) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt64Size(2, totalBytes_);
+    }
+    if (peakBytes_ != 0L) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt64Size(3, peakBytes_);
+    }
+    if (liveBytes_ != 0L) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt64Size(4, liveBytes_);
+    }
+    if (allocatorBytesInUse_ != 0L) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt64Size(5, allocatorBytesInUse_);
+    }
+    for (int i = 0; i < allocationRecords_.size(); i++) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(6, allocationRecords_.get(i));
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.framework.AllocatorMemoryUsed)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.framework.AllocatorMemoryUsed other = (org.tensorflow.proto.framework.AllocatorMemoryUsed) obj;
+
+    if (!getAllocatorName()
+        .equals(other.getAllocatorName())) return false;
+    if (getTotalBytes()
+        != other.getTotalBytes()) return false;
+    if (getPeakBytes()
+        != other.getPeakBytes()) return false;
+    if (getLiveBytes()
+        != other.getLiveBytes()) return false;
+    if (!getAllocationRecordsList()
+        .equals(other.getAllocationRecordsList())) return false;
+    if (getAllocatorBytesInUse()
+        != other.getAllocatorBytesInUse()) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    hash = (37 * hash) + ALLOCATOR_NAME_FIELD_NUMBER;
+    hash = (53 * hash) + getAllocatorName().hashCode();
+    hash = (37 * hash) + TOTAL_BYTES_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+        getTotalBytes());
+    hash = (37 * hash) + PEAK_BYTES_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+        getPeakBytes());
+    hash = (37 * hash) + LIVE_BYTES_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+        getLiveBytes());
+    if (getAllocationRecordsCount() > 0) {
+      hash = (37 * hash) + ALLOCATION_RECORDS_FIELD_NUMBER;
+      hash = (53 * hash) + getAllocationRecordsList().hashCode();
+    }
+    hash = (37 * hash) + ALLOCATOR_BYTES_IN_USE_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+        getAllocatorBytesInUse());
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.framework.AllocatorMemoryUsed parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.AllocatorMemoryUsed parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.AllocatorMemoryUsed parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.AllocatorMemoryUsed parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.AllocatorMemoryUsed parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.AllocatorMemoryUsed parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.AllocatorMemoryUsed parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.AllocatorMemoryUsed parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.AllocatorMemoryUsed parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.AllocatorMemoryUsed parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.AllocatorMemoryUsed parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.AllocatorMemoryUsed parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.framework.AllocatorMemoryUsed prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * Protobuf type {@code tensorflow.AllocatorMemoryUsed}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.AllocatorMemoryUsed)
+      org.tensorflow.proto.framework.AllocatorMemoryUsedOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.StepStatsProtos.internal_static_tensorflow_AllocatorMemoryUsed_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.StepStatsProtos.internal_static_tensorflow_AllocatorMemoryUsed_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.AllocatorMemoryUsed.class, org.tensorflow.proto.framework.AllocatorMemoryUsed.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.framework.AllocatorMemoryUsed.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+        getAllocationRecordsFieldBuilder();
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      allocatorName_ = "";
+
+      totalBytes_ = 0L;
+
+      peakBytes_ = 0L;
+
+      liveBytes_ = 0L;
+
+      if (allocationRecordsBuilder_ == null) {
+        allocationRecords_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000001);
+      } else {
+        allocationRecordsBuilder_.clear();
+      }
+      allocatorBytesInUse_ = 0L;
+
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.framework.StepStatsProtos.internal_static_tensorflow_AllocatorMemoryUsed_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.AllocatorMemoryUsed getDefaultInstanceForType() {
+      return org.tensorflow.proto.framework.AllocatorMemoryUsed.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.AllocatorMemoryUsed build() {
+      org.tensorflow.proto.framework.AllocatorMemoryUsed result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.AllocatorMemoryUsed buildPartial() {
+      org.tensorflow.proto.framework.AllocatorMemoryUsed result = new org.tensorflow.proto.framework.AllocatorMemoryUsed(this);
+      int from_bitField0_ = bitField0_;
+      result.allocatorName_ = allocatorName_;
+      result.totalBytes_ = totalBytes_;
+      result.peakBytes_ = peakBytes_;
+      result.liveBytes_ = liveBytes_;
+      if (allocationRecordsBuilder_ == null) {
+        if (((bitField0_ & 0x00000001) != 0)) {
+          allocationRecords_ = java.util.Collections.unmodifiableList(allocationRecords_);
+          bitField0_ = (bitField0_ & ~0x00000001);
+        }
+        result.allocationRecords_ = allocationRecords_;
+      } else {
+        result.allocationRecords_ = allocationRecordsBuilder_.build();
+      }
+      result.allocatorBytesInUse_ = allocatorBytesInUse_;
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.framework.AllocatorMemoryUsed) {
+        return mergeFrom((org.tensorflow.proto.framework.AllocatorMemoryUsed)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.framework.AllocatorMemoryUsed other) {
+      if (other == org.tensorflow.proto.framework.AllocatorMemoryUsed.getDefaultInstance()) return this;
+      if (!other.getAllocatorName().isEmpty()) {
+        allocatorName_ = other.allocatorName_;
+        onChanged();
+      }
+      if (other.getTotalBytes() != 0L) {
+        setTotalBytes(other.getTotalBytes());
+      }
+      if (other.getPeakBytes() != 0L) {
+        setPeakBytes(other.getPeakBytes());
+      }
+      if (other.getLiveBytes() != 0L) {
+        setLiveBytes(other.getLiveBytes());
+      }
+      if (allocationRecordsBuilder_ == null) {
+        if (!other.allocationRecords_.isEmpty()) {
+          if (allocationRecords_.isEmpty()) {
+            allocationRecords_ = other.allocationRecords_;
+            bitField0_ = (bitField0_ & ~0x00000001);
+          } else {
+            ensureAllocationRecordsIsMutable();
+            allocationRecords_.addAll(other.allocationRecords_);
+          }
+          onChanged();
+        }
+      } else {
+        if (!other.allocationRecords_.isEmpty()) {
+          if (allocationRecordsBuilder_.isEmpty()) {
+            allocationRecordsBuilder_.dispose();
+            allocationRecordsBuilder_ = null;
+            allocationRecords_ = other.allocationRecords_;
+            bitField0_ = (bitField0_ & ~0x00000001);
+            allocationRecordsBuilder_ = 
+              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
+                 getAllocationRecordsFieldBuilder() : null;
+          } else {
+            allocationRecordsBuilder_.addAllMessages(other.allocationRecords_);
+          }
+        }
+      }
+      if (other.getAllocatorBytesInUse() != 0L) {
+        setAllocatorBytesInUse(other.getAllocatorBytesInUse());
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.framework.AllocatorMemoryUsed parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.framework.AllocatorMemoryUsed) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+    private int bitField0_;
+
+    private java.lang.Object allocatorName_ = "";
+    /**
+     * <code>string allocator_name = 1;</code>
+     */
+    public java.lang.String getAllocatorName() {
+      java.lang.Object ref = allocatorName_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        allocatorName_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <code>string allocator_name = 1;</code>
+     */
+    public com.google.protobuf.ByteString
+        getAllocatorNameBytes() {
+      java.lang.Object ref = allocatorName_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        allocatorName_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <code>string allocator_name = 1;</code>
+     */
+    public Builder setAllocatorName(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      allocatorName_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>string allocator_name = 1;</code>
+     */
+    public Builder clearAllocatorName() {
+      
+      allocatorName_ = getDefaultInstance().getAllocatorName();
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>string allocator_name = 1;</code>
+     */
+    public Builder setAllocatorNameBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      allocatorName_ = value;
+      onChanged();
+      return this;
+    }
+
+    private long totalBytes_ ;
+    /**
+     * <pre>
+     * These are per-node allocator memory stats.
+     * </pre>
+     *
+     * <code>int64 total_bytes = 2;</code>
+     */
+    public long getTotalBytes() {
+      return totalBytes_;
+    }
+    /**
+     * <pre>
+     * These are per-node allocator memory stats.
+     * </pre>
+     *
+     * <code>int64 total_bytes = 2;</code>
+     */
+    public Builder setTotalBytes(long value) {
+      
+      totalBytes_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * These are per-node allocator memory stats.
+     * </pre>
+     *
+     * <code>int64 total_bytes = 2;</code>
+     */
+    public Builder clearTotalBytes() {
+      
+      totalBytes_ = 0L;
+      onChanged();
+      return this;
+    }
+
+    private long peakBytes_ ;
+    /**
+     * <code>int64 peak_bytes = 3;</code>
+     */
+    public long getPeakBytes() {
+      return peakBytes_;
+    }
+    /**
+     * <code>int64 peak_bytes = 3;</code>
+     */
+    public Builder setPeakBytes(long value) {
+      
+      peakBytes_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>int64 peak_bytes = 3;</code>
+     */
+    public Builder clearPeakBytes() {
+      
+      peakBytes_ = 0L;
+      onChanged();
+      return this;
+    }
+
+    private long liveBytes_ ;
+    /**
+     * <pre>
+     * The bytes that are not deallocated.
+     * </pre>
+     *
+     * <code>int64 live_bytes = 4;</code>
+     */
+    public long getLiveBytes() {
+      return liveBytes_;
+    }
+    /**
+     * <pre>
+     * The bytes that are not deallocated.
+     * </pre>
+     *
+     * <code>int64 live_bytes = 4;</code>
+     */
+    public Builder setLiveBytes(long value) {
+      
+      liveBytes_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * The bytes that are not deallocated.
+     * </pre>
+     *
+     * <code>int64 live_bytes = 4;</code>
+     */
+    public Builder clearLiveBytes() {
+      
+      liveBytes_ = 0L;
+      onChanged();
+      return this;
+    }
+
+    private java.util.List<org.tensorflow.proto.framework.AllocationRecord> allocationRecords_ =
+      java.util.Collections.emptyList();
+    private void ensureAllocationRecordsIsMutable() {
+      if (!((bitField0_ & 0x00000001) != 0)) {
+        allocationRecords_ = new java.util.ArrayList<org.tensorflow.proto.framework.AllocationRecord>(allocationRecords_);
+        bitField0_ |= 0x00000001;
+       }
+    }
+
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.framework.AllocationRecord, org.tensorflow.proto.framework.AllocationRecord.Builder, org.tensorflow.proto.framework.AllocationRecordOrBuilder> allocationRecordsBuilder_;
+
+    /**
+     * <pre>
+     * The allocation and deallocation timeline.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.AllocationRecord allocation_records = 6;</code>
+     */
+    public java.util.List<org.tensorflow.proto.framework.AllocationRecord> getAllocationRecordsList() {
+      if (allocationRecordsBuilder_ == null) {
+        return java.util.Collections.unmodifiableList(allocationRecords_);
+      } else {
+        return allocationRecordsBuilder_.getMessageList();
+      }
+    }
+    /**
+     * <pre>
+     * The allocation and deallocation timeline.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.AllocationRecord allocation_records = 6;</code>
+     */
+    public int getAllocationRecordsCount() {
+      if (allocationRecordsBuilder_ == null) {
+        return allocationRecords_.size();
+      } else {
+        return allocationRecordsBuilder_.getCount();
+      }
+    }
+    /**
+     * <pre>
+     * The allocation and deallocation timeline.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.AllocationRecord allocation_records = 6;</code>
+     */
+    public org.tensorflow.proto.framework.AllocationRecord getAllocationRecords(int index) {
+      if (allocationRecordsBuilder_ == null) {
+        return allocationRecords_.get(index);
+      } else {
+        return allocationRecordsBuilder_.getMessage(index);
+      }
+    }
+    /**
+     * <pre>
+     * The allocation and deallocation timeline.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.AllocationRecord allocation_records = 6;</code>
+     */
+    public Builder setAllocationRecords(
+        int index, org.tensorflow.proto.framework.AllocationRecord value) {
+      if (allocationRecordsBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureAllocationRecordsIsMutable();
+        allocationRecords_.set(index, value);
+        onChanged();
+      } else {
+        allocationRecordsBuilder_.setMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * The allocation and deallocation timeline.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.AllocationRecord allocation_records = 6;</code>
+     */
+    public Builder setAllocationRecords(
+        int index, org.tensorflow.proto.framework.AllocationRecord.Builder builderForValue) {
+      if (allocationRecordsBuilder_ == null) {
+        ensureAllocationRecordsIsMutable();
+        allocationRecords_.set(index, builderForValue.build());
+        onChanged();
+      } else {
+        allocationRecordsBuilder_.setMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * The allocation and deallocation timeline.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.AllocationRecord allocation_records = 6;</code>
+     */
+    public Builder addAllocationRecords(org.tensorflow.proto.framework.AllocationRecord value) {
+      if (allocationRecordsBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureAllocationRecordsIsMutable();
+        allocationRecords_.add(value);
+        onChanged();
+      } else {
+        allocationRecordsBuilder_.addMessage(value);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * The allocation and deallocation timeline.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.AllocationRecord allocation_records = 6;</code>
+     */
+    public Builder addAllocationRecords(
+        int index, org.tensorflow.proto.framework.AllocationRecord value) {
+      if (allocationRecordsBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureAllocationRecordsIsMutable();
+        allocationRecords_.add(index, value);
+        onChanged();
+      } else {
+        allocationRecordsBuilder_.addMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * The allocation and deallocation timeline.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.AllocationRecord allocation_records = 6;</code>
+     */
+    public Builder addAllocationRecords(
+        org.tensorflow.proto.framework.AllocationRecord.Builder builderForValue) {
+      if (allocationRecordsBuilder_ == null) {
+        ensureAllocationRecordsIsMutable();
+        allocationRecords_.add(builderForValue.build());
+        onChanged();
+      } else {
+        allocationRecordsBuilder_.addMessage(builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * The allocation and deallocation timeline.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.AllocationRecord allocation_records = 6;</code>
+     */
+    public Builder addAllocationRecords(
+        int index, org.tensorflow.proto.framework.AllocationRecord.Builder builderForValue) {
+      if (allocationRecordsBuilder_ == null) {
+        ensureAllocationRecordsIsMutable();
+        allocationRecords_.add(index, builderForValue.build());
+        onChanged();
+      } else {
+        allocationRecordsBuilder_.addMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * The allocation and deallocation timeline.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.AllocationRecord allocation_records = 6;</code>
+     */
+    public Builder addAllAllocationRecords(
+        java.lang.Iterable<? extends org.tensorflow.proto.framework.AllocationRecord> values) {
+      if (allocationRecordsBuilder_ == null) {
+        ensureAllocationRecordsIsMutable();
+        com.google.protobuf.AbstractMessageLite.Builder.addAll(
+            values, allocationRecords_);
+        onChanged();
+      } else {
+        allocationRecordsBuilder_.addAllMessages(values);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * The allocation and deallocation timeline.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.AllocationRecord allocation_records = 6;</code>
+     */
+    public Builder clearAllocationRecords() {
+      if (allocationRecordsBuilder_ == null) {
+        allocationRecords_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000001);
+        onChanged();
+      } else {
+        allocationRecordsBuilder_.clear();
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * The allocation and deallocation timeline.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.AllocationRecord allocation_records = 6;</code>
+     */
+    public Builder removeAllocationRecords(int index) {
+      if (allocationRecordsBuilder_ == null) {
+        ensureAllocationRecordsIsMutable();
+        allocationRecords_.remove(index);
+        onChanged();
+      } else {
+        allocationRecordsBuilder_.remove(index);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * The allocation and deallocation timeline.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.AllocationRecord allocation_records = 6;</code>
+     */
+    public org.tensorflow.proto.framework.AllocationRecord.Builder getAllocationRecordsBuilder(
+        int index) {
+      return getAllocationRecordsFieldBuilder().getBuilder(index);
+    }
+    /**
+     * <pre>
+     * The allocation and deallocation timeline.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.AllocationRecord allocation_records = 6;</code>
+     */
+    public org.tensorflow.proto.framework.AllocationRecordOrBuilder getAllocationRecordsOrBuilder(
+        int index) {
+      if (allocationRecordsBuilder_ == null) {
+        return allocationRecords_.get(index);  } else {
+        return allocationRecordsBuilder_.getMessageOrBuilder(index);
+      }
+    }
+    /**
+     * <pre>
+     * The allocation and deallocation timeline.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.AllocationRecord allocation_records = 6;</code>
+     */
+    public java.util.List<? extends org.tensorflow.proto.framework.AllocationRecordOrBuilder> 
+         getAllocationRecordsOrBuilderList() {
+      if (allocationRecordsBuilder_ != null) {
+        return allocationRecordsBuilder_.getMessageOrBuilderList();
+      } else {
+        return java.util.Collections.unmodifiableList(allocationRecords_);
+      }
+    }
+    /**
+     * <pre>
+     * The allocation and deallocation timeline.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.AllocationRecord allocation_records = 6;</code>
+     */
+    public org.tensorflow.proto.framework.AllocationRecord.Builder addAllocationRecordsBuilder() {
+      return getAllocationRecordsFieldBuilder().addBuilder(
+          org.tensorflow.proto.framework.AllocationRecord.getDefaultInstance());
+    }
+    /**
+     * <pre>
+     * The allocation and deallocation timeline.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.AllocationRecord allocation_records = 6;</code>
+     */
+    public org.tensorflow.proto.framework.AllocationRecord.Builder addAllocationRecordsBuilder(
+        int index) {
+      return getAllocationRecordsFieldBuilder().addBuilder(
+          index, org.tensorflow.proto.framework.AllocationRecord.getDefaultInstance());
+    }
+    /**
+     * <pre>
+     * The allocation and deallocation timeline.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.AllocationRecord allocation_records = 6;</code>
+     */
+    public java.util.List<org.tensorflow.proto.framework.AllocationRecord.Builder> 
+         getAllocationRecordsBuilderList() {
+      return getAllocationRecordsFieldBuilder().getBuilderList();
+    }
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.framework.AllocationRecord, org.tensorflow.proto.framework.AllocationRecord.Builder, org.tensorflow.proto.framework.AllocationRecordOrBuilder> 
+        getAllocationRecordsFieldBuilder() {
+      if (allocationRecordsBuilder_ == null) {
+        allocationRecordsBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
+            org.tensorflow.proto.framework.AllocationRecord, org.tensorflow.proto.framework.AllocationRecord.Builder, org.tensorflow.proto.framework.AllocationRecordOrBuilder>(
+                allocationRecords_,
+                ((bitField0_ & 0x00000001) != 0),
+                getParentForChildren(),
+                isClean());
+        allocationRecords_ = null;
+      }
+      return allocationRecordsBuilder_;
+    }
+
+    private long allocatorBytesInUse_ ;
+    /**
+     * <pre>
+     * These are snapshots of the overall allocator memory stats.
+     * The number of live bytes currently allocated by the allocator.
+     * </pre>
+     *
+     * <code>int64 allocator_bytes_in_use = 5;</code>
+     */
+    public long getAllocatorBytesInUse() {
+      return allocatorBytesInUse_;
+    }
+    /**
+     * <pre>
+     * These are snapshots of the overall allocator memory stats.
+     * The number of live bytes currently allocated by the allocator.
+     * </pre>
+     *
+     * <code>int64 allocator_bytes_in_use = 5;</code>
+     */
+    public Builder setAllocatorBytesInUse(long value) {
+      
+      allocatorBytesInUse_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * These are snapshots of the overall allocator memory stats.
+     * The number of live bytes currently allocated by the allocator.
+     * </pre>
+     *
+     * <code>int64 allocator_bytes_in_use = 5;</code>
+     */
+    public Builder clearAllocatorBytesInUse() {
+      
+      allocatorBytesInUse_ = 0L;
+      onChanged();
+      return this;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.AllocatorMemoryUsed)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.AllocatorMemoryUsed)
+  private static final org.tensorflow.proto.framework.AllocatorMemoryUsed DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.AllocatorMemoryUsed();
+  }
+
+  public static org.tensorflow.proto.framework.AllocatorMemoryUsed getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<AllocatorMemoryUsed>
+      PARSER = new com.google.protobuf.AbstractParser<AllocatorMemoryUsed>() {
+    @java.lang.Override
+    public AllocatorMemoryUsed parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new AllocatorMemoryUsed(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<AllocatorMemoryUsed> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<AllocatorMemoryUsed> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.framework.AllocatorMemoryUsed getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/AllocatorMemoryUsedOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/AllocatorMemoryUsedOrBuilder.java
new file mode 100644
index 00000000000..44dc47fa31d
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/AllocatorMemoryUsedOrBuilder.java
@@ -0,0 +1,96 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/framework/step_stats.proto
+
+package org.tensorflow.proto.framework;
+
+public interface AllocatorMemoryUsedOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.AllocatorMemoryUsed)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <code>string allocator_name = 1;</code>
+   */
+  java.lang.String getAllocatorName();
+  /**
+   * <code>string allocator_name = 1;</code>
+   */
+  com.google.protobuf.ByteString
+      getAllocatorNameBytes();
+
+  /**
+   * <pre>
+   * These are per-node allocator memory stats.
+   * </pre>
+   *
+   * <code>int64 total_bytes = 2;</code>
+   */
+  long getTotalBytes();
+
+  /**
+   * <code>int64 peak_bytes = 3;</code>
+   */
+  long getPeakBytes();
+
+  /**
+   * <pre>
+   * The bytes that are not deallocated.
+   * </pre>
+   *
+   * <code>int64 live_bytes = 4;</code>
+   */
+  long getLiveBytes();
+
+  /**
+   * <pre>
+   * The allocation and deallocation timeline.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.AllocationRecord allocation_records = 6;</code>
+   */
+  java.util.List<org.tensorflow.proto.framework.AllocationRecord> 
+      getAllocationRecordsList();
+  /**
+   * <pre>
+   * The allocation and deallocation timeline.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.AllocationRecord allocation_records = 6;</code>
+   */
+  org.tensorflow.proto.framework.AllocationRecord getAllocationRecords(int index);
+  /**
+   * <pre>
+   * The allocation and deallocation timeline.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.AllocationRecord allocation_records = 6;</code>
+   */
+  int getAllocationRecordsCount();
+  /**
+   * <pre>
+   * The allocation and deallocation timeline.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.AllocationRecord allocation_records = 6;</code>
+   */
+  java.util.List<? extends org.tensorflow.proto.framework.AllocationRecordOrBuilder> 
+      getAllocationRecordsOrBuilderList();
+  /**
+   * <pre>
+   * The allocation and deallocation timeline.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.AllocationRecord allocation_records = 6;</code>
+   */
+  org.tensorflow.proto.framework.AllocationRecordOrBuilder getAllocationRecordsOrBuilder(
+      int index);
+
+  /**
+   * <pre>
+   * These are snapshots of the overall allocator memory stats.
+   * The number of live bytes currently allocated by the allocator.
+   * </pre>
+   *
+   * <code>int64 allocator_bytes_in_use = 5;</code>
+   */
+  long getAllocatorBytesInUse();
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ApiDef.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ApiDef.java
new file mode 100644
index 00000000000..99916fcf1f2
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ApiDef.java
@@ -0,0 +1,6303 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/framework/api_def.proto
+
+package org.tensorflow.proto.framework;
+
+/**
+ * <pre>
+ * Used to specify and override the default API &amp; behavior in the
+ * generated code for client languages, from what you would get from
+ * the OpDef alone. There will be a set of ApiDefs that are common
+ * to all client languages, and another set per client language.
+ * The per-client-language ApiDefs will inherit values from the
+ * common ApiDefs which it can either replace or modify.
+ * We separate the API definition from the OpDef so we can evolve the
+ * API while remaining backwards compatible when interpretting old
+ * graphs.  Overrides go in an "api_def.pbtxt" file with a text-format
+ * ApiDefs message.
+ * WARNING: Be *very* careful changing the API for any existing op --
+ * you can change the semantics of existing code.  These changes may
+ * need to wait until a major release of TensorFlow to avoid breaking
+ * our compatibility promises.
+ * </pre>
+ *
+ * Protobuf type {@code tensorflow.ApiDef}
+ */
+public  final class ApiDef extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.ApiDef)
+    ApiDefOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use ApiDef.newBuilder() to construct.
+  private ApiDef(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private ApiDef() {
+    graphOpName_ = "";
+    deprecationMessage_ = "";
+    visibility_ = 0;
+    endpoint_ = java.util.Collections.emptyList();
+    inArg_ = java.util.Collections.emptyList();
+    outArg_ = java.util.Collections.emptyList();
+    argOrder_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+    attr_ = java.util.Collections.emptyList();
+    summary_ = "";
+    description_ = "";
+    descriptionPrefix_ = "";
+    descriptionSuffix_ = "";
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new ApiDef();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private ApiDef(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    int mutable_bitField0_ = 0;
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 10: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            graphOpName_ = s;
+            break;
+          }
+          case 16: {
+            int rawValue = input.readEnum();
+
+            visibility_ = rawValue;
+            break;
+          }
+          case 26: {
+            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
+              endpoint_ = new java.util.ArrayList<org.tensorflow.proto.framework.ApiDef.Endpoint>();
+              mutable_bitField0_ |= 0x00000001;
+            }
+            endpoint_.add(
+                input.readMessage(org.tensorflow.proto.framework.ApiDef.Endpoint.parser(), extensionRegistry));
+            break;
+          }
+          case 34: {
+            if (!((mutable_bitField0_ & 0x00000002) != 0)) {
+              inArg_ = new java.util.ArrayList<org.tensorflow.proto.framework.ApiDef.Arg>();
+              mutable_bitField0_ |= 0x00000002;
+            }
+            inArg_.add(
+                input.readMessage(org.tensorflow.proto.framework.ApiDef.Arg.parser(), extensionRegistry));
+            break;
+          }
+          case 42: {
+            if (!((mutable_bitField0_ & 0x00000004) != 0)) {
+              outArg_ = new java.util.ArrayList<org.tensorflow.proto.framework.ApiDef.Arg>();
+              mutable_bitField0_ |= 0x00000004;
+            }
+            outArg_.add(
+                input.readMessage(org.tensorflow.proto.framework.ApiDef.Arg.parser(), extensionRegistry));
+            break;
+          }
+          case 50: {
+            if (!((mutable_bitField0_ & 0x00000010) != 0)) {
+              attr_ = new java.util.ArrayList<org.tensorflow.proto.framework.ApiDef.Attr>();
+              mutable_bitField0_ |= 0x00000010;
+            }
+            attr_.add(
+                input.readMessage(org.tensorflow.proto.framework.ApiDef.Attr.parser(), extensionRegistry));
+            break;
+          }
+          case 58: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            summary_ = s;
+            break;
+          }
+          case 66: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            description_ = s;
+            break;
+          }
+          case 74: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            descriptionPrefix_ = s;
+            break;
+          }
+          case 82: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            descriptionSuffix_ = s;
+            break;
+          }
+          case 90: {
+            java.lang.String s = input.readStringRequireUtf8();
+            if (!((mutable_bitField0_ & 0x00000008) != 0)) {
+              argOrder_ = new com.google.protobuf.LazyStringArrayList();
+              mutable_bitField0_ |= 0x00000008;
+            }
+            argOrder_.add(s);
+            break;
+          }
+          case 98: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            deprecationMessage_ = s;
+            break;
+          }
+          case 104: {
+
+            deprecationVersion_ = input.readInt32();
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      if (((mutable_bitField0_ & 0x00000001) != 0)) {
+        endpoint_ = java.util.Collections.unmodifiableList(endpoint_);
+      }
+      if (((mutable_bitField0_ & 0x00000002) != 0)) {
+        inArg_ = java.util.Collections.unmodifiableList(inArg_);
+      }
+      if (((mutable_bitField0_ & 0x00000004) != 0)) {
+        outArg_ = java.util.Collections.unmodifiableList(outArg_);
+      }
+      if (((mutable_bitField0_ & 0x00000010) != 0)) {
+        attr_ = java.util.Collections.unmodifiableList(attr_);
+      }
+      if (((mutable_bitField0_ & 0x00000008) != 0)) {
+        argOrder_ = argOrder_.getUnmodifiableView();
+      }
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.framework.ApiDefProtos.internal_static_tensorflow_ApiDef_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.framework.ApiDefProtos.internal_static_tensorflow_ApiDef_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.framework.ApiDef.class, org.tensorflow.proto.framework.ApiDef.Builder.class);
+  }
+
+  /**
+   * Protobuf enum {@code tensorflow.ApiDef.Visibility}
+   */
+  public enum Visibility
+      implements com.google.protobuf.ProtocolMessageEnum {
+    /**
+     * <pre>
+     * Normally this is "VISIBLE" unless you are inheriting a
+     * different value from another ApiDef.
+     * </pre>
+     *
+     * <code>DEFAULT_VISIBILITY = 0;</code>
+     */
+    DEFAULT_VISIBILITY(0),
+    /**
+     * <pre>
+     * Publicly visible in the API.
+     * </pre>
+     *
+     * <code>VISIBLE = 1;</code>
+     */
+    VISIBLE(1),
+    /**
+     * <pre>
+     * Do not include this op in the generated API. If visibility is
+     * set to 'SKIP', other fields are ignored for this op.
+     * </pre>
+     *
+     * <code>SKIP = 2;</code>
+     */
+    SKIP(2),
+    /**
+     * <pre>
+     * Hide this op by putting it into an internal namespace (or whatever
+     * is appropriate in the target language).
+     * </pre>
+     *
+     * <code>HIDDEN = 3;</code>
+     */
+    HIDDEN(3),
+    UNRECOGNIZED(-1),
+    ;
+
+    /**
+     * <pre>
+     * Normally this is "VISIBLE" unless you are inheriting a
+     * different value from another ApiDef.
+     * </pre>
+     *
+     * <code>DEFAULT_VISIBILITY = 0;</code>
+     */
+    public static final int DEFAULT_VISIBILITY_VALUE = 0;
+    /**
+     * <pre>
+     * Publicly visible in the API.
+     * </pre>
+     *
+     * <code>VISIBLE = 1;</code>
+     */
+    public static final int VISIBLE_VALUE = 1;
+    /**
+     * <pre>
+     * Do not include this op in the generated API. If visibility is
+     * set to 'SKIP', other fields are ignored for this op.
+     * </pre>
+     *
+     * <code>SKIP = 2;</code>
+     */
+    public static final int SKIP_VALUE = 2;
+    /**
+     * <pre>
+     * Hide this op by putting it into an internal namespace (or whatever
+     * is appropriate in the target language).
+     * </pre>
+     *
+     * <code>HIDDEN = 3;</code>
+     */
+    public static final int HIDDEN_VALUE = 3;
+
+
+    public final int getNumber() {
+      if (this == UNRECOGNIZED) {
+        throw new java.lang.IllegalArgumentException(
+            "Can't get the number of an unknown enum value.");
+      }
+      return value;
+    }
+
+    /**
+     * @deprecated Use {@link #forNumber(int)} instead.
+     */
+    @java.lang.Deprecated
+    public static Visibility valueOf(int value) {
+      return forNumber(value);
+    }
+
+    public static Visibility forNumber(int value) {
+      switch (value) {
+        case 0: return DEFAULT_VISIBILITY;
+        case 1: return VISIBLE;
+        case 2: return SKIP;
+        case 3: return HIDDEN;
+        default: return null;
+      }
+    }
+
+    public static com.google.protobuf.Internal.EnumLiteMap<Visibility>
+        internalGetValueMap() {
+      return internalValueMap;
+    }
+    private static final com.google.protobuf.Internal.EnumLiteMap<
+        Visibility> internalValueMap =
+          new com.google.protobuf.Internal.EnumLiteMap<Visibility>() {
+            public Visibility findValueByNumber(int number) {
+              return Visibility.forNumber(number);
+            }
+          };
+
+    public final com.google.protobuf.Descriptors.EnumValueDescriptor
+        getValueDescriptor() {
+      return getDescriptor().getValues().get(ordinal());
+    }
+    public final com.google.protobuf.Descriptors.EnumDescriptor
+        getDescriptorForType() {
+      return getDescriptor();
+    }
+    public static final com.google.protobuf.Descriptors.EnumDescriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.ApiDef.getDescriptor().getEnumTypes().get(0);
+    }
+
+    private static final Visibility[] VALUES = values();
+
+    public static Visibility valueOf(
+        com.google.protobuf.Descriptors.EnumValueDescriptor desc) {
+      if (desc.getType() != getDescriptor()) {
+        throw new java.lang.IllegalArgumentException(
+          "EnumValueDescriptor is not for this type.");
+      }
+      if (desc.getIndex() == -1) {
+        return UNRECOGNIZED;
+      }
+      return VALUES[desc.getIndex()];
+    }
+
+    private final int value;
+
+    private Visibility(int value) {
+      this.value = value;
+    }
+
+    // @@protoc_insertion_point(enum_scope:tensorflow.ApiDef.Visibility)
+  }
+
+  public interface EndpointOrBuilder extends
+      // @@protoc_insertion_point(interface_extends:tensorflow.ApiDef.Endpoint)
+      com.google.protobuf.MessageOrBuilder {
+
+    /**
+     * <pre>
+     * Name should be either like "CamelCaseName" or
+     * "Package.CamelCaseName". Client-language-specific ApiDefs may
+     * use a snake_case convention instead of CamelCase.
+     * </pre>
+     *
+     * <code>string name = 1;</code>
+     */
+    java.lang.String getName();
+    /**
+     * <pre>
+     * Name should be either like "CamelCaseName" or
+     * "Package.CamelCaseName". Client-language-specific ApiDefs may
+     * use a snake_case convention instead of CamelCase.
+     * </pre>
+     *
+     * <code>string name = 1;</code>
+     */
+    com.google.protobuf.ByteString
+        getNameBytes();
+
+    /**
+     * <pre>
+     * Set if this endpoint is deprecated. If set to true, a message suggesting
+     * to use a non-deprecated endpoint instead will be printed. If all
+     * endpoints are deprecated, set deprecation_message in ApiDef instead.
+     * </pre>
+     *
+     * <code>bool deprecated = 3;</code>
+     */
+    boolean getDeprecated();
+
+    /**
+     * <pre>
+     * Major version when an endpoint will be deleted. For e.g. set this
+     * value to 2 if endpoint should be removed in TensorFlow 2.0 and
+     * deprecated in versions before that.
+     * </pre>
+     *
+     * <code>int32 deprecation_version = 4;</code>
+     */
+    int getDeprecationVersion();
+  }
+  /**
+   * <pre>
+   * If you specify any endpoint, this will replace all of the
+   * inherited endpoints.  The first endpoint should be the
+   * "canonical" endpoint, and should not be deprecated (unless all
+   * endpoints are deprecated).
+   * </pre>
+   *
+   * Protobuf type {@code tensorflow.ApiDef.Endpoint}
+   */
+  public  static final class Endpoint extends
+      com.google.protobuf.GeneratedMessageV3 implements
+      // @@protoc_insertion_point(message_implements:tensorflow.ApiDef.Endpoint)
+      EndpointOrBuilder {
+  private static final long serialVersionUID = 0L;
+    // Use Endpoint.newBuilder() to construct.
+    private Endpoint(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+      super(builder);
+    }
+    private Endpoint() {
+      name_ = "";
+    }
+
+    @java.lang.Override
+    @SuppressWarnings({"unused"})
+    protected java.lang.Object newInstance(
+        UnusedPrivateParameter unused) {
+      return new Endpoint();
+    }
+
+    @java.lang.Override
+    public final com.google.protobuf.UnknownFieldSet
+    getUnknownFields() {
+      return this.unknownFields;
+    }
+    private Endpoint(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      this();
+      if (extensionRegistry == null) {
+        throw new java.lang.NullPointerException();
+      }
+      com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+          com.google.protobuf.UnknownFieldSet.newBuilder();
+      try {
+        boolean done = false;
+        while (!done) {
+          int tag = input.readTag();
+          switch (tag) {
+            case 0:
+              done = true;
+              break;
+            case 10: {
+              java.lang.String s = input.readStringRequireUtf8();
+
+              name_ = s;
+              break;
+            }
+            case 24: {
+
+              deprecated_ = input.readBool();
+              break;
+            }
+            case 32: {
+
+              deprecationVersion_ = input.readInt32();
+              break;
+            }
+            default: {
+              if (!parseUnknownField(
+                  input, unknownFields, extensionRegistry, tag)) {
+                done = true;
+              }
+              break;
+            }
+          }
+        }
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        throw e.setUnfinishedMessage(this);
+      } catch (java.io.IOException e) {
+        throw new com.google.protobuf.InvalidProtocolBufferException(
+            e).setUnfinishedMessage(this);
+      } finally {
+        this.unknownFields = unknownFields.build();
+        makeExtensionsImmutable();
+      }
+    }
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.ApiDefProtos.internal_static_tensorflow_ApiDef_Endpoint_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.ApiDefProtos.internal_static_tensorflow_ApiDef_Endpoint_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.ApiDef.Endpoint.class, org.tensorflow.proto.framework.ApiDef.Endpoint.Builder.class);
+    }
+
+    public static final int NAME_FIELD_NUMBER = 1;
+    private volatile java.lang.Object name_;
+    /**
+     * <pre>
+     * Name should be either like "CamelCaseName" or
+     * "Package.CamelCaseName". Client-language-specific ApiDefs may
+     * use a snake_case convention instead of CamelCase.
+     * </pre>
+     *
+     * <code>string name = 1;</code>
+     */
+    public java.lang.String getName() {
+      java.lang.Object ref = name_;
+      if (ref instanceof java.lang.String) {
+        return (java.lang.String) ref;
+      } else {
+        com.google.protobuf.ByteString bs = 
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        name_ = s;
+        return s;
+      }
+    }
+    /**
+     * <pre>
+     * Name should be either like "CamelCaseName" or
+     * "Package.CamelCaseName". Client-language-specific ApiDefs may
+     * use a snake_case convention instead of CamelCase.
+     * </pre>
+     *
+     * <code>string name = 1;</code>
+     */
+    public com.google.protobuf.ByteString
+        getNameBytes() {
+      java.lang.Object ref = name_;
+      if (ref instanceof java.lang.String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        name_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+
+    public static final int DEPRECATED_FIELD_NUMBER = 3;
+    private boolean deprecated_;
+    /**
+     * <pre>
+     * Set if this endpoint is deprecated. If set to true, a message suggesting
+     * to use a non-deprecated endpoint instead will be printed. If all
+     * endpoints are deprecated, set deprecation_message in ApiDef instead.
+     * </pre>
+     *
+     * <code>bool deprecated = 3;</code>
+     */
+    public boolean getDeprecated() {
+      return deprecated_;
+    }
+
+    public static final int DEPRECATION_VERSION_FIELD_NUMBER = 4;
+    private int deprecationVersion_;
+    /**
+     * <pre>
+     * Major version when an endpoint will be deleted. For e.g. set this
+     * value to 2 if endpoint should be removed in TensorFlow 2.0 and
+     * deprecated in versions before that.
+     * </pre>
+     *
+     * <code>int32 deprecation_version = 4;</code>
+     */
+    public int getDeprecationVersion() {
+      return deprecationVersion_;
+    }
+
+    private byte memoizedIsInitialized = -1;
+    @java.lang.Override
+    public final boolean isInitialized() {
+      byte isInitialized = memoizedIsInitialized;
+      if (isInitialized == 1) return true;
+      if (isInitialized == 0) return false;
+
+      memoizedIsInitialized = 1;
+      return true;
+    }
+
+    @java.lang.Override
+    public void writeTo(com.google.protobuf.CodedOutputStream output)
+                        throws java.io.IOException {
+      if (!getNameBytes().isEmpty()) {
+        com.google.protobuf.GeneratedMessageV3.writeString(output, 1, name_);
+      }
+      if (deprecated_ != false) {
+        output.writeBool(3, deprecated_);
+      }
+      if (deprecationVersion_ != 0) {
+        output.writeInt32(4, deprecationVersion_);
+      }
+      unknownFields.writeTo(output);
+    }
+
+    @java.lang.Override
+    public int getSerializedSize() {
+      int size = memoizedSize;
+      if (size != -1) return size;
+
+      size = 0;
+      if (!getNameBytes().isEmpty()) {
+        size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, name_);
+      }
+      if (deprecated_ != false) {
+        size += com.google.protobuf.CodedOutputStream
+          .computeBoolSize(3, deprecated_);
+      }
+      if (deprecationVersion_ != 0) {
+        size += com.google.protobuf.CodedOutputStream
+          .computeInt32Size(4, deprecationVersion_);
+      }
+      size += unknownFields.getSerializedSize();
+      memoizedSize = size;
+      return size;
+    }
+
+    @java.lang.Override
+    public boolean equals(final java.lang.Object obj) {
+      if (obj == this) {
+       return true;
+      }
+      if (!(obj instanceof org.tensorflow.proto.framework.ApiDef.Endpoint)) {
+        return super.equals(obj);
+      }
+      org.tensorflow.proto.framework.ApiDef.Endpoint other = (org.tensorflow.proto.framework.ApiDef.Endpoint) obj;
+
+      if (!getName()
+          .equals(other.getName())) return false;
+      if (getDeprecated()
+          != other.getDeprecated()) return false;
+      if (getDeprecationVersion()
+          != other.getDeprecationVersion()) return false;
+      if (!unknownFields.equals(other.unknownFields)) return false;
+      return true;
+    }
+
+    @java.lang.Override
+    public int hashCode() {
+      if (memoizedHashCode != 0) {
+        return memoizedHashCode;
+      }
+      int hash = 41;
+      hash = (19 * hash) + getDescriptor().hashCode();
+      hash = (37 * hash) + NAME_FIELD_NUMBER;
+      hash = (53 * hash) + getName().hashCode();
+      hash = (37 * hash) + DEPRECATED_FIELD_NUMBER;
+      hash = (53 * hash) + com.google.protobuf.Internal.hashBoolean(
+          getDeprecated());
+      hash = (37 * hash) + DEPRECATION_VERSION_FIELD_NUMBER;
+      hash = (53 * hash) + getDeprecationVersion();
+      hash = (29 * hash) + unknownFields.hashCode();
+      memoizedHashCode = hash;
+      return hash;
+    }
+
+    public static org.tensorflow.proto.framework.ApiDef.Endpoint parseFrom(
+        java.nio.ByteBuffer data)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data);
+    }
+    public static org.tensorflow.proto.framework.ApiDef.Endpoint parseFrom(
+        java.nio.ByteBuffer data,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.ApiDef.Endpoint parseFrom(
+        com.google.protobuf.ByteString data)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data);
+    }
+    public static org.tensorflow.proto.framework.ApiDef.Endpoint parseFrom(
+        com.google.protobuf.ByteString data,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.ApiDef.Endpoint parseFrom(byte[] data)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data);
+    }
+    public static org.tensorflow.proto.framework.ApiDef.Endpoint parseFrom(
+        byte[] data,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.ApiDef.Endpoint parseFrom(java.io.InputStream input)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input);
+    }
+    public static org.tensorflow.proto.framework.ApiDef.Endpoint parseFrom(
+        java.io.InputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.ApiDef.Endpoint parseDelimitedFrom(java.io.InputStream input)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseDelimitedWithIOException(PARSER, input);
+    }
+    public static org.tensorflow.proto.framework.ApiDef.Endpoint parseDelimitedFrom(
+        java.io.InputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.ApiDef.Endpoint parseFrom(
+        com.google.protobuf.CodedInputStream input)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input);
+    }
+    public static org.tensorflow.proto.framework.ApiDef.Endpoint parseFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input, extensionRegistry);
+    }
+
+    @java.lang.Override
+    public Builder newBuilderForType() { return newBuilder(); }
+    public static Builder newBuilder() {
+      return DEFAULT_INSTANCE.toBuilder();
+    }
+    public static Builder newBuilder(org.tensorflow.proto.framework.ApiDef.Endpoint prototype) {
+      return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+    }
+    @java.lang.Override
+    public Builder toBuilder() {
+      return this == DEFAULT_INSTANCE
+          ? new Builder() : new Builder().mergeFrom(this);
+    }
+
+    @java.lang.Override
+    protected Builder newBuilderForType(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      Builder builder = new Builder(parent);
+      return builder;
+    }
+    /**
+     * <pre>
+     * If you specify any endpoint, this will replace all of the
+     * inherited endpoints.  The first endpoint should be the
+     * "canonical" endpoint, and should not be deprecated (unless all
+     * endpoints are deprecated).
+     * </pre>
+     *
+     * Protobuf type {@code tensorflow.ApiDef.Endpoint}
+     */
+    public static final class Builder extends
+        com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+        // @@protoc_insertion_point(builder_implements:tensorflow.ApiDef.Endpoint)
+        org.tensorflow.proto.framework.ApiDef.EndpointOrBuilder {
+      public static final com.google.protobuf.Descriptors.Descriptor
+          getDescriptor() {
+        return org.tensorflow.proto.framework.ApiDefProtos.internal_static_tensorflow_ApiDef_Endpoint_descriptor;
+      }
+
+      @java.lang.Override
+      protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+          internalGetFieldAccessorTable() {
+        return org.tensorflow.proto.framework.ApiDefProtos.internal_static_tensorflow_ApiDef_Endpoint_fieldAccessorTable
+            .ensureFieldAccessorsInitialized(
+                org.tensorflow.proto.framework.ApiDef.Endpoint.class, org.tensorflow.proto.framework.ApiDef.Endpoint.Builder.class);
+      }
+
+      // Construct using org.tensorflow.proto.framework.ApiDef.Endpoint.newBuilder()
+      private Builder() {
+        maybeForceBuilderInitialization();
+      }
+
+      private Builder(
+          com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+        super(parent);
+        maybeForceBuilderInitialization();
+      }
+      private void maybeForceBuilderInitialization() {
+        if (com.google.protobuf.GeneratedMessageV3
+                .alwaysUseFieldBuilders) {
+        }
+      }
+      @java.lang.Override
+      public Builder clear() {
+        super.clear();
+        name_ = "";
+
+        deprecated_ = false;
+
+        deprecationVersion_ = 0;
+
+        return this;
+      }
+
+      @java.lang.Override
+      public com.google.protobuf.Descriptors.Descriptor
+          getDescriptorForType() {
+        return org.tensorflow.proto.framework.ApiDefProtos.internal_static_tensorflow_ApiDef_Endpoint_descriptor;
+      }
+
+      @java.lang.Override
+      public org.tensorflow.proto.framework.ApiDef.Endpoint getDefaultInstanceForType() {
+        return org.tensorflow.proto.framework.ApiDef.Endpoint.getDefaultInstance();
+      }
+
+      @java.lang.Override
+      public org.tensorflow.proto.framework.ApiDef.Endpoint build() {
+        org.tensorflow.proto.framework.ApiDef.Endpoint result = buildPartial();
+        if (!result.isInitialized()) {
+          throw newUninitializedMessageException(result);
+        }
+        return result;
+      }
+
+      @java.lang.Override
+      public org.tensorflow.proto.framework.ApiDef.Endpoint buildPartial() {
+        org.tensorflow.proto.framework.ApiDef.Endpoint result = new org.tensorflow.proto.framework.ApiDef.Endpoint(this);
+        result.name_ = name_;
+        result.deprecated_ = deprecated_;
+        result.deprecationVersion_ = deprecationVersion_;
+        onBuilt();
+        return result;
+      }
+
+      @java.lang.Override
+      public Builder clone() {
+        return super.clone();
+      }
+      @java.lang.Override
+      public Builder setField(
+          com.google.protobuf.Descriptors.FieldDescriptor field,
+          java.lang.Object value) {
+        return super.setField(field, value);
+      }
+      @java.lang.Override
+      public Builder clearField(
+          com.google.protobuf.Descriptors.FieldDescriptor field) {
+        return super.clearField(field);
+      }
+      @java.lang.Override
+      public Builder clearOneof(
+          com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+        return super.clearOneof(oneof);
+      }
+      @java.lang.Override
+      public Builder setRepeatedField(
+          com.google.protobuf.Descriptors.FieldDescriptor field,
+          int index, java.lang.Object value) {
+        return super.setRepeatedField(field, index, value);
+      }
+      @java.lang.Override
+      public Builder addRepeatedField(
+          com.google.protobuf.Descriptors.FieldDescriptor field,
+          java.lang.Object value) {
+        return super.addRepeatedField(field, value);
+      }
+      @java.lang.Override
+      public Builder mergeFrom(com.google.protobuf.Message other) {
+        if (other instanceof org.tensorflow.proto.framework.ApiDef.Endpoint) {
+          return mergeFrom((org.tensorflow.proto.framework.ApiDef.Endpoint)other);
+        } else {
+          super.mergeFrom(other);
+          return this;
+        }
+      }
+
+      public Builder mergeFrom(org.tensorflow.proto.framework.ApiDef.Endpoint other) {
+        if (other == org.tensorflow.proto.framework.ApiDef.Endpoint.getDefaultInstance()) return this;
+        if (!other.getName().isEmpty()) {
+          name_ = other.name_;
+          onChanged();
+        }
+        if (other.getDeprecated() != false) {
+          setDeprecated(other.getDeprecated());
+        }
+        if (other.getDeprecationVersion() != 0) {
+          setDeprecationVersion(other.getDeprecationVersion());
+        }
+        this.mergeUnknownFields(other.unknownFields);
+        onChanged();
+        return this;
+      }
+
+      @java.lang.Override
+      public final boolean isInitialized() {
+        return true;
+      }
+
+      @java.lang.Override
+      public Builder mergeFrom(
+          com.google.protobuf.CodedInputStream input,
+          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+          throws java.io.IOException {
+        org.tensorflow.proto.framework.ApiDef.Endpoint parsedMessage = null;
+        try {
+          parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+        } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+          parsedMessage = (org.tensorflow.proto.framework.ApiDef.Endpoint) e.getUnfinishedMessage();
+          throw e.unwrapIOException();
+        } finally {
+          if (parsedMessage != null) {
+            mergeFrom(parsedMessage);
+          }
+        }
+        return this;
+      }
+
+      private java.lang.Object name_ = "";
+      /**
+       * <pre>
+       * Name should be either like "CamelCaseName" or
+       * "Package.CamelCaseName". Client-language-specific ApiDefs may
+       * use a snake_case convention instead of CamelCase.
+       * </pre>
+       *
+       * <code>string name = 1;</code>
+       */
+      public java.lang.String getName() {
+        java.lang.Object ref = name_;
+        if (!(ref instanceof java.lang.String)) {
+          com.google.protobuf.ByteString bs =
+              (com.google.protobuf.ByteString) ref;
+          java.lang.String s = bs.toStringUtf8();
+          name_ = s;
+          return s;
+        } else {
+          return (java.lang.String) ref;
+        }
+      }
+      /**
+       * <pre>
+       * Name should be either like "CamelCaseName" or
+       * "Package.CamelCaseName". Client-language-specific ApiDefs may
+       * use a snake_case convention instead of CamelCase.
+       * </pre>
+       *
+       * <code>string name = 1;</code>
+       */
+      public com.google.protobuf.ByteString
+          getNameBytes() {
+        java.lang.Object ref = name_;
+        if (ref instanceof String) {
+          com.google.protobuf.ByteString b = 
+              com.google.protobuf.ByteString.copyFromUtf8(
+                  (java.lang.String) ref);
+          name_ = b;
+          return b;
+        } else {
+          return (com.google.protobuf.ByteString) ref;
+        }
+      }
+      /**
+       * <pre>
+       * Name should be either like "CamelCaseName" or
+       * "Package.CamelCaseName". Client-language-specific ApiDefs may
+       * use a snake_case convention instead of CamelCase.
+       * </pre>
+       *
+       * <code>string name = 1;</code>
+       */
+      public Builder setName(
+          java.lang.String value) {
+        if (value == null) {
+    throw new NullPointerException();
+  }
+  
+        name_ = value;
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * Name should be either like "CamelCaseName" or
+       * "Package.CamelCaseName". Client-language-specific ApiDefs may
+       * use a snake_case convention instead of CamelCase.
+       * </pre>
+       *
+       * <code>string name = 1;</code>
+       */
+      public Builder clearName() {
+        
+        name_ = getDefaultInstance().getName();
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * Name should be either like "CamelCaseName" or
+       * "Package.CamelCaseName". Client-language-specific ApiDefs may
+       * use a snake_case convention instead of CamelCase.
+       * </pre>
+       *
+       * <code>string name = 1;</code>
+       */
+      public Builder setNameBytes(
+          com.google.protobuf.ByteString value) {
+        if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+        
+        name_ = value;
+        onChanged();
+        return this;
+      }
+
+      private boolean deprecated_ ;
+      /**
+       * <pre>
+       * Set if this endpoint is deprecated. If set to true, a message suggesting
+       * to use a non-deprecated endpoint instead will be printed. If all
+       * endpoints are deprecated, set deprecation_message in ApiDef instead.
+       * </pre>
+       *
+       * <code>bool deprecated = 3;</code>
+       */
+      public boolean getDeprecated() {
+        return deprecated_;
+      }
+      /**
+       * <pre>
+       * Set if this endpoint is deprecated. If set to true, a message suggesting
+       * to use a non-deprecated endpoint instead will be printed. If all
+       * endpoints are deprecated, set deprecation_message in ApiDef instead.
+       * </pre>
+       *
+       * <code>bool deprecated = 3;</code>
+       */
+      public Builder setDeprecated(boolean value) {
+        
+        deprecated_ = value;
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * Set if this endpoint is deprecated. If set to true, a message suggesting
+       * to use a non-deprecated endpoint instead will be printed. If all
+       * endpoints are deprecated, set deprecation_message in ApiDef instead.
+       * </pre>
+       *
+       * <code>bool deprecated = 3;</code>
+       */
+      public Builder clearDeprecated() {
+        
+        deprecated_ = false;
+        onChanged();
+        return this;
+      }
+
+      private int deprecationVersion_ ;
+      /**
+       * <pre>
+       * Major version when an endpoint will be deleted. For e.g. set this
+       * value to 2 if endpoint should be removed in TensorFlow 2.0 and
+       * deprecated in versions before that.
+       * </pre>
+       *
+       * <code>int32 deprecation_version = 4;</code>
+       */
+      public int getDeprecationVersion() {
+        return deprecationVersion_;
+      }
+      /**
+       * <pre>
+       * Major version when an endpoint will be deleted. For e.g. set this
+       * value to 2 if endpoint should be removed in TensorFlow 2.0 and
+       * deprecated in versions before that.
+       * </pre>
+       *
+       * <code>int32 deprecation_version = 4;</code>
+       */
+      public Builder setDeprecationVersion(int value) {
+        
+        deprecationVersion_ = value;
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * Major version when an endpoint will be deleted. For e.g. set this
+       * value to 2 if endpoint should be removed in TensorFlow 2.0 and
+       * deprecated in versions before that.
+       * </pre>
+       *
+       * <code>int32 deprecation_version = 4;</code>
+       */
+      public Builder clearDeprecationVersion() {
+        
+        deprecationVersion_ = 0;
+        onChanged();
+        return this;
+      }
+      @java.lang.Override
+      public final Builder setUnknownFields(
+          final com.google.protobuf.UnknownFieldSet unknownFields) {
+        return super.setUnknownFields(unknownFields);
+      }
+
+      @java.lang.Override
+      public final Builder mergeUnknownFields(
+          final com.google.protobuf.UnknownFieldSet unknownFields) {
+        return super.mergeUnknownFields(unknownFields);
+      }
+
+
+      // @@protoc_insertion_point(builder_scope:tensorflow.ApiDef.Endpoint)
+    }
+
+    // @@protoc_insertion_point(class_scope:tensorflow.ApiDef.Endpoint)
+    private static final org.tensorflow.proto.framework.ApiDef.Endpoint DEFAULT_INSTANCE;
+    static {
+      DEFAULT_INSTANCE = new org.tensorflow.proto.framework.ApiDef.Endpoint();
+    }
+
+    public static org.tensorflow.proto.framework.ApiDef.Endpoint getDefaultInstance() {
+      return DEFAULT_INSTANCE;
+    }
+
+    private static final com.google.protobuf.Parser<Endpoint>
+        PARSER = new com.google.protobuf.AbstractParser<Endpoint>() {
+      @java.lang.Override
+      public Endpoint parsePartialFrom(
+          com.google.protobuf.CodedInputStream input,
+          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+          throws com.google.protobuf.InvalidProtocolBufferException {
+        return new Endpoint(input, extensionRegistry);
+      }
+    };
+
+    public static com.google.protobuf.Parser<Endpoint> parser() {
+      return PARSER;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Parser<Endpoint> getParserForType() {
+      return PARSER;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.ApiDef.Endpoint getDefaultInstanceForType() {
+      return DEFAULT_INSTANCE;
+    }
+
+  }
+
+  public interface ArgOrBuilder extends
+      // @@protoc_insertion_point(interface_extends:tensorflow.ApiDef.Arg)
+      com.google.protobuf.MessageOrBuilder {
+
+    /**
+     * <code>string name = 1;</code>
+     */
+    java.lang.String getName();
+    /**
+     * <code>string name = 1;</code>
+     */
+    com.google.protobuf.ByteString
+        getNameBytes();
+
+    /**
+     * <pre>
+     * Change the name used to access this arg in the API from what
+     * is used in the GraphDef.  Note that these names in `backticks`
+     * will also be replaced in the summary &amp; description fields.
+     * </pre>
+     *
+     * <code>string rename_to = 2;</code>
+     */
+    java.lang.String getRenameTo();
+    /**
+     * <pre>
+     * Change the name used to access this arg in the API from what
+     * is used in the GraphDef.  Note that these names in `backticks`
+     * will also be replaced in the summary &amp; description fields.
+     * </pre>
+     *
+     * <code>string rename_to = 2;</code>
+     */
+    com.google.protobuf.ByteString
+        getRenameToBytes();
+
+    /**
+     * <pre>
+     * Note: this will replace any inherited arg doc. There is no
+     * current way of modifying arg descriptions (other than replacing
+     * them entirely) as can be done with op descriptions.
+     * </pre>
+     *
+     * <code>string description = 3;</code>
+     */
+    java.lang.String getDescription();
+    /**
+     * <pre>
+     * Note: this will replace any inherited arg doc. There is no
+     * current way of modifying arg descriptions (other than replacing
+     * them entirely) as can be done with op descriptions.
+     * </pre>
+     *
+     * <code>string description = 3;</code>
+     */
+    com.google.protobuf.ByteString
+        getDescriptionBytes();
+  }
+  /**
+   * Protobuf type {@code tensorflow.ApiDef.Arg}
+   */
+  public  static final class Arg extends
+      com.google.protobuf.GeneratedMessageV3 implements
+      // @@protoc_insertion_point(message_implements:tensorflow.ApiDef.Arg)
+      ArgOrBuilder {
+  private static final long serialVersionUID = 0L;
+    // Use Arg.newBuilder() to construct.
+    private Arg(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+      super(builder);
+    }
+    private Arg() {
+      name_ = "";
+      renameTo_ = "";
+      description_ = "";
+    }
+
+    @java.lang.Override
+    @SuppressWarnings({"unused"})
+    protected java.lang.Object newInstance(
+        UnusedPrivateParameter unused) {
+      return new Arg();
+    }
+
+    @java.lang.Override
+    public final com.google.protobuf.UnknownFieldSet
+    getUnknownFields() {
+      return this.unknownFields;
+    }
+    private Arg(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      this();
+      if (extensionRegistry == null) {
+        throw new java.lang.NullPointerException();
+      }
+      com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+          com.google.protobuf.UnknownFieldSet.newBuilder();
+      try {
+        boolean done = false;
+        while (!done) {
+          int tag = input.readTag();
+          switch (tag) {
+            case 0:
+              done = true;
+              break;
+            case 10: {
+              java.lang.String s = input.readStringRequireUtf8();
+
+              name_ = s;
+              break;
+            }
+            case 18: {
+              java.lang.String s = input.readStringRequireUtf8();
+
+              renameTo_ = s;
+              break;
+            }
+            case 26: {
+              java.lang.String s = input.readStringRequireUtf8();
+
+              description_ = s;
+              break;
+            }
+            default: {
+              if (!parseUnknownField(
+                  input, unknownFields, extensionRegistry, tag)) {
+                done = true;
+              }
+              break;
+            }
+          }
+        }
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        throw e.setUnfinishedMessage(this);
+      } catch (java.io.IOException e) {
+        throw new com.google.protobuf.InvalidProtocolBufferException(
+            e).setUnfinishedMessage(this);
+      } finally {
+        this.unknownFields = unknownFields.build();
+        makeExtensionsImmutable();
+      }
+    }
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.ApiDefProtos.internal_static_tensorflow_ApiDef_Arg_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.ApiDefProtos.internal_static_tensorflow_ApiDef_Arg_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.ApiDef.Arg.class, org.tensorflow.proto.framework.ApiDef.Arg.Builder.class);
+    }
+
+    public static final int NAME_FIELD_NUMBER = 1;
+    private volatile java.lang.Object name_;
+    /**
+     * <code>string name = 1;</code>
+     */
+    public java.lang.String getName() {
+      java.lang.Object ref = name_;
+      if (ref instanceof java.lang.String) {
+        return (java.lang.String) ref;
+      } else {
+        com.google.protobuf.ByteString bs = 
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        name_ = s;
+        return s;
+      }
+    }
+    /**
+     * <code>string name = 1;</code>
+     */
+    public com.google.protobuf.ByteString
+        getNameBytes() {
+      java.lang.Object ref = name_;
+      if (ref instanceof java.lang.String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        name_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+
+    public static final int RENAME_TO_FIELD_NUMBER = 2;
+    private volatile java.lang.Object renameTo_;
+    /**
+     * <pre>
+     * Change the name used to access this arg in the API from what
+     * is used in the GraphDef.  Note that these names in `backticks`
+     * will also be replaced in the summary &amp; description fields.
+     * </pre>
+     *
+     * <code>string rename_to = 2;</code>
+     */
+    public java.lang.String getRenameTo() {
+      java.lang.Object ref = renameTo_;
+      if (ref instanceof java.lang.String) {
+        return (java.lang.String) ref;
+      } else {
+        com.google.protobuf.ByteString bs = 
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        renameTo_ = s;
+        return s;
+      }
+    }
+    /**
+     * <pre>
+     * Change the name used to access this arg in the API from what
+     * is used in the GraphDef.  Note that these names in `backticks`
+     * will also be replaced in the summary &amp; description fields.
+     * </pre>
+     *
+     * <code>string rename_to = 2;</code>
+     */
+    public com.google.protobuf.ByteString
+        getRenameToBytes() {
+      java.lang.Object ref = renameTo_;
+      if (ref instanceof java.lang.String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        renameTo_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+
+    public static final int DESCRIPTION_FIELD_NUMBER = 3;
+    private volatile java.lang.Object description_;
+    /**
+     * <pre>
+     * Note: this will replace any inherited arg doc. There is no
+     * current way of modifying arg descriptions (other than replacing
+     * them entirely) as can be done with op descriptions.
+     * </pre>
+     *
+     * <code>string description = 3;</code>
+     */
+    public java.lang.String getDescription() {
+      java.lang.Object ref = description_;
+      if (ref instanceof java.lang.String) {
+        return (java.lang.String) ref;
+      } else {
+        com.google.protobuf.ByteString bs = 
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        description_ = s;
+        return s;
+      }
+    }
+    /**
+     * <pre>
+     * Note: this will replace any inherited arg doc. There is no
+     * current way of modifying arg descriptions (other than replacing
+     * them entirely) as can be done with op descriptions.
+     * </pre>
+     *
+     * <code>string description = 3;</code>
+     */
+    public com.google.protobuf.ByteString
+        getDescriptionBytes() {
+      java.lang.Object ref = description_;
+      if (ref instanceof java.lang.String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        description_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+
+    private byte memoizedIsInitialized = -1;
+    @java.lang.Override
+    public final boolean isInitialized() {
+      byte isInitialized = memoizedIsInitialized;
+      if (isInitialized == 1) return true;
+      if (isInitialized == 0) return false;
+
+      memoizedIsInitialized = 1;
+      return true;
+    }
+
+    @java.lang.Override
+    public void writeTo(com.google.protobuf.CodedOutputStream output)
+                        throws java.io.IOException {
+      if (!getNameBytes().isEmpty()) {
+        com.google.protobuf.GeneratedMessageV3.writeString(output, 1, name_);
+      }
+      if (!getRenameToBytes().isEmpty()) {
+        com.google.protobuf.GeneratedMessageV3.writeString(output, 2, renameTo_);
+      }
+      if (!getDescriptionBytes().isEmpty()) {
+        com.google.protobuf.GeneratedMessageV3.writeString(output, 3, description_);
+      }
+      unknownFields.writeTo(output);
+    }
+
+    @java.lang.Override
+    public int getSerializedSize() {
+      int size = memoizedSize;
+      if (size != -1) return size;
+
+      size = 0;
+      if (!getNameBytes().isEmpty()) {
+        size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, name_);
+      }
+      if (!getRenameToBytes().isEmpty()) {
+        size += com.google.protobuf.GeneratedMessageV3.computeStringSize(2, renameTo_);
+      }
+      if (!getDescriptionBytes().isEmpty()) {
+        size += com.google.protobuf.GeneratedMessageV3.computeStringSize(3, description_);
+      }
+      size += unknownFields.getSerializedSize();
+      memoizedSize = size;
+      return size;
+    }
+
+    @java.lang.Override
+    public boolean equals(final java.lang.Object obj) {
+      if (obj == this) {
+       return true;
+      }
+      if (!(obj instanceof org.tensorflow.proto.framework.ApiDef.Arg)) {
+        return super.equals(obj);
+      }
+      org.tensorflow.proto.framework.ApiDef.Arg other = (org.tensorflow.proto.framework.ApiDef.Arg) obj;
+
+      if (!getName()
+          .equals(other.getName())) return false;
+      if (!getRenameTo()
+          .equals(other.getRenameTo())) return false;
+      if (!getDescription()
+          .equals(other.getDescription())) return false;
+      if (!unknownFields.equals(other.unknownFields)) return false;
+      return true;
+    }
+
+    @java.lang.Override
+    public int hashCode() {
+      if (memoizedHashCode != 0) {
+        return memoizedHashCode;
+      }
+      int hash = 41;
+      hash = (19 * hash) + getDescriptor().hashCode();
+      hash = (37 * hash) + NAME_FIELD_NUMBER;
+      hash = (53 * hash) + getName().hashCode();
+      hash = (37 * hash) + RENAME_TO_FIELD_NUMBER;
+      hash = (53 * hash) + getRenameTo().hashCode();
+      hash = (37 * hash) + DESCRIPTION_FIELD_NUMBER;
+      hash = (53 * hash) + getDescription().hashCode();
+      hash = (29 * hash) + unknownFields.hashCode();
+      memoizedHashCode = hash;
+      return hash;
+    }
+
+    public static org.tensorflow.proto.framework.ApiDef.Arg parseFrom(
+        java.nio.ByteBuffer data)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data);
+    }
+    public static org.tensorflow.proto.framework.ApiDef.Arg parseFrom(
+        java.nio.ByteBuffer data,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.ApiDef.Arg parseFrom(
+        com.google.protobuf.ByteString data)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data);
+    }
+    public static org.tensorflow.proto.framework.ApiDef.Arg parseFrom(
+        com.google.protobuf.ByteString data,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.ApiDef.Arg parseFrom(byte[] data)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data);
+    }
+    public static org.tensorflow.proto.framework.ApiDef.Arg parseFrom(
+        byte[] data,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.ApiDef.Arg parseFrom(java.io.InputStream input)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input);
+    }
+    public static org.tensorflow.proto.framework.ApiDef.Arg parseFrom(
+        java.io.InputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.ApiDef.Arg parseDelimitedFrom(java.io.InputStream input)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseDelimitedWithIOException(PARSER, input);
+    }
+    public static org.tensorflow.proto.framework.ApiDef.Arg parseDelimitedFrom(
+        java.io.InputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.ApiDef.Arg parseFrom(
+        com.google.protobuf.CodedInputStream input)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input);
+    }
+    public static org.tensorflow.proto.framework.ApiDef.Arg parseFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input, extensionRegistry);
+    }
+
+    @java.lang.Override
+    public Builder newBuilderForType() { return newBuilder(); }
+    public static Builder newBuilder() {
+      return DEFAULT_INSTANCE.toBuilder();
+    }
+    public static Builder newBuilder(org.tensorflow.proto.framework.ApiDef.Arg prototype) {
+      return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+    }
+    @java.lang.Override
+    public Builder toBuilder() {
+      return this == DEFAULT_INSTANCE
+          ? new Builder() : new Builder().mergeFrom(this);
+    }
+
+    @java.lang.Override
+    protected Builder newBuilderForType(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      Builder builder = new Builder(parent);
+      return builder;
+    }
+    /**
+     * Protobuf type {@code tensorflow.ApiDef.Arg}
+     */
+    public static final class Builder extends
+        com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+        // @@protoc_insertion_point(builder_implements:tensorflow.ApiDef.Arg)
+        org.tensorflow.proto.framework.ApiDef.ArgOrBuilder {
+      public static final com.google.protobuf.Descriptors.Descriptor
+          getDescriptor() {
+        return org.tensorflow.proto.framework.ApiDefProtos.internal_static_tensorflow_ApiDef_Arg_descriptor;
+      }
+
+      @java.lang.Override
+      protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+          internalGetFieldAccessorTable() {
+        return org.tensorflow.proto.framework.ApiDefProtos.internal_static_tensorflow_ApiDef_Arg_fieldAccessorTable
+            .ensureFieldAccessorsInitialized(
+                org.tensorflow.proto.framework.ApiDef.Arg.class, org.tensorflow.proto.framework.ApiDef.Arg.Builder.class);
+      }
+
+      // Construct using org.tensorflow.proto.framework.ApiDef.Arg.newBuilder()
+      private Builder() {
+        maybeForceBuilderInitialization();
+      }
+
+      private Builder(
+          com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+        super(parent);
+        maybeForceBuilderInitialization();
+      }
+      private void maybeForceBuilderInitialization() {
+        if (com.google.protobuf.GeneratedMessageV3
+                .alwaysUseFieldBuilders) {
+        }
+      }
+      @java.lang.Override
+      public Builder clear() {
+        super.clear();
+        name_ = "";
+
+        renameTo_ = "";
+
+        description_ = "";
+
+        return this;
+      }
+
+      @java.lang.Override
+      public com.google.protobuf.Descriptors.Descriptor
+          getDescriptorForType() {
+        return org.tensorflow.proto.framework.ApiDefProtos.internal_static_tensorflow_ApiDef_Arg_descriptor;
+      }
+
+      @java.lang.Override
+      public org.tensorflow.proto.framework.ApiDef.Arg getDefaultInstanceForType() {
+        return org.tensorflow.proto.framework.ApiDef.Arg.getDefaultInstance();
+      }
+
+      @java.lang.Override
+      public org.tensorflow.proto.framework.ApiDef.Arg build() {
+        org.tensorflow.proto.framework.ApiDef.Arg result = buildPartial();
+        if (!result.isInitialized()) {
+          throw newUninitializedMessageException(result);
+        }
+        return result;
+      }
+
+      @java.lang.Override
+      public org.tensorflow.proto.framework.ApiDef.Arg buildPartial() {
+        org.tensorflow.proto.framework.ApiDef.Arg result = new org.tensorflow.proto.framework.ApiDef.Arg(this);
+        result.name_ = name_;
+        result.renameTo_ = renameTo_;
+        result.description_ = description_;
+        onBuilt();
+        return result;
+      }
+
+      @java.lang.Override
+      public Builder clone() {
+        return super.clone();
+      }
+      @java.lang.Override
+      public Builder setField(
+          com.google.protobuf.Descriptors.FieldDescriptor field,
+          java.lang.Object value) {
+        return super.setField(field, value);
+      }
+      @java.lang.Override
+      public Builder clearField(
+          com.google.protobuf.Descriptors.FieldDescriptor field) {
+        return super.clearField(field);
+      }
+      @java.lang.Override
+      public Builder clearOneof(
+          com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+        return super.clearOneof(oneof);
+      }
+      @java.lang.Override
+      public Builder setRepeatedField(
+          com.google.protobuf.Descriptors.FieldDescriptor field,
+          int index, java.lang.Object value) {
+        return super.setRepeatedField(field, index, value);
+      }
+      @java.lang.Override
+      public Builder addRepeatedField(
+          com.google.protobuf.Descriptors.FieldDescriptor field,
+          java.lang.Object value) {
+        return super.addRepeatedField(field, value);
+      }
+      @java.lang.Override
+      public Builder mergeFrom(com.google.protobuf.Message other) {
+        if (other instanceof org.tensorflow.proto.framework.ApiDef.Arg) {
+          return mergeFrom((org.tensorflow.proto.framework.ApiDef.Arg)other);
+        } else {
+          super.mergeFrom(other);
+          return this;
+        }
+      }
+
+      public Builder mergeFrom(org.tensorflow.proto.framework.ApiDef.Arg other) {
+        if (other == org.tensorflow.proto.framework.ApiDef.Arg.getDefaultInstance()) return this;
+        if (!other.getName().isEmpty()) {
+          name_ = other.name_;
+          onChanged();
+        }
+        if (!other.getRenameTo().isEmpty()) {
+          renameTo_ = other.renameTo_;
+          onChanged();
+        }
+        if (!other.getDescription().isEmpty()) {
+          description_ = other.description_;
+          onChanged();
+        }
+        this.mergeUnknownFields(other.unknownFields);
+        onChanged();
+        return this;
+      }
+
+      @java.lang.Override
+      public final boolean isInitialized() {
+        return true;
+      }
+
+      @java.lang.Override
+      public Builder mergeFrom(
+          com.google.protobuf.CodedInputStream input,
+          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+          throws java.io.IOException {
+        org.tensorflow.proto.framework.ApiDef.Arg parsedMessage = null;
+        try {
+          parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+        } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+          parsedMessage = (org.tensorflow.proto.framework.ApiDef.Arg) e.getUnfinishedMessage();
+          throw e.unwrapIOException();
+        } finally {
+          if (parsedMessage != null) {
+            mergeFrom(parsedMessage);
+          }
+        }
+        return this;
+      }
+
+      private java.lang.Object name_ = "";
+      /**
+       * <code>string name = 1;</code>
+       */
+      public java.lang.String getName() {
+        java.lang.Object ref = name_;
+        if (!(ref instanceof java.lang.String)) {
+          com.google.protobuf.ByteString bs =
+              (com.google.protobuf.ByteString) ref;
+          java.lang.String s = bs.toStringUtf8();
+          name_ = s;
+          return s;
+        } else {
+          return (java.lang.String) ref;
+        }
+      }
+      /**
+       * <code>string name = 1;</code>
+       */
+      public com.google.protobuf.ByteString
+          getNameBytes() {
+        java.lang.Object ref = name_;
+        if (ref instanceof String) {
+          com.google.protobuf.ByteString b = 
+              com.google.protobuf.ByteString.copyFromUtf8(
+                  (java.lang.String) ref);
+          name_ = b;
+          return b;
+        } else {
+          return (com.google.protobuf.ByteString) ref;
+        }
+      }
+      /**
+       * <code>string name = 1;</code>
+       */
+      public Builder setName(
+          java.lang.String value) {
+        if (value == null) {
+    throw new NullPointerException();
+  }
+  
+        name_ = value;
+        onChanged();
+        return this;
+      }
+      /**
+       * <code>string name = 1;</code>
+       */
+      public Builder clearName() {
+        
+        name_ = getDefaultInstance().getName();
+        onChanged();
+        return this;
+      }
+      /**
+       * <code>string name = 1;</code>
+       */
+      public Builder setNameBytes(
+          com.google.protobuf.ByteString value) {
+        if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+        
+        name_ = value;
+        onChanged();
+        return this;
+      }
+
+      private java.lang.Object renameTo_ = "";
+      /**
+       * <pre>
+       * Change the name used to access this arg in the API from what
+       * is used in the GraphDef.  Note that these names in `backticks`
+       * will also be replaced in the summary &amp; description fields.
+       * </pre>
+       *
+       * <code>string rename_to = 2;</code>
+       */
+      public java.lang.String getRenameTo() {
+        java.lang.Object ref = renameTo_;
+        if (!(ref instanceof java.lang.String)) {
+          com.google.protobuf.ByteString bs =
+              (com.google.protobuf.ByteString) ref;
+          java.lang.String s = bs.toStringUtf8();
+          renameTo_ = s;
+          return s;
+        } else {
+          return (java.lang.String) ref;
+        }
+      }
+      /**
+       * <pre>
+       * Change the name used to access this arg in the API from what
+       * is used in the GraphDef.  Note that these names in `backticks`
+       * will also be replaced in the summary &amp; description fields.
+       * </pre>
+       *
+       * <code>string rename_to = 2;</code>
+       */
+      public com.google.protobuf.ByteString
+          getRenameToBytes() {
+        java.lang.Object ref = renameTo_;
+        if (ref instanceof String) {
+          com.google.protobuf.ByteString b = 
+              com.google.protobuf.ByteString.copyFromUtf8(
+                  (java.lang.String) ref);
+          renameTo_ = b;
+          return b;
+        } else {
+          return (com.google.protobuf.ByteString) ref;
+        }
+      }
+      /**
+       * <pre>
+       * Change the name used to access this arg in the API from what
+       * is used in the GraphDef.  Note that these names in `backticks`
+       * will also be replaced in the summary &amp; description fields.
+       * </pre>
+       *
+       * <code>string rename_to = 2;</code>
+       */
+      public Builder setRenameTo(
+          java.lang.String value) {
+        if (value == null) {
+    throw new NullPointerException();
+  }
+  
+        renameTo_ = value;
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * Change the name used to access this arg in the API from what
+       * is used in the GraphDef.  Note that these names in `backticks`
+       * will also be replaced in the summary &amp; description fields.
+       * </pre>
+       *
+       * <code>string rename_to = 2;</code>
+       */
+      public Builder clearRenameTo() {
+        
+        renameTo_ = getDefaultInstance().getRenameTo();
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * Change the name used to access this arg in the API from what
+       * is used in the GraphDef.  Note that these names in `backticks`
+       * will also be replaced in the summary &amp; description fields.
+       * </pre>
+       *
+       * <code>string rename_to = 2;</code>
+       */
+      public Builder setRenameToBytes(
+          com.google.protobuf.ByteString value) {
+        if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+        
+        renameTo_ = value;
+        onChanged();
+        return this;
+      }
+
+      private java.lang.Object description_ = "";
+      /**
+       * <pre>
+       * Note: this will replace any inherited arg doc. There is no
+       * current way of modifying arg descriptions (other than replacing
+       * them entirely) as can be done with op descriptions.
+       * </pre>
+       *
+       * <code>string description = 3;</code>
+       */
+      public java.lang.String getDescription() {
+        java.lang.Object ref = description_;
+        if (!(ref instanceof java.lang.String)) {
+          com.google.protobuf.ByteString bs =
+              (com.google.protobuf.ByteString) ref;
+          java.lang.String s = bs.toStringUtf8();
+          description_ = s;
+          return s;
+        } else {
+          return (java.lang.String) ref;
+        }
+      }
+      /**
+       * <pre>
+       * Note: this will replace any inherited arg doc. There is no
+       * current way of modifying arg descriptions (other than replacing
+       * them entirely) as can be done with op descriptions.
+       * </pre>
+       *
+       * <code>string description = 3;</code>
+       */
+      public com.google.protobuf.ByteString
+          getDescriptionBytes() {
+        java.lang.Object ref = description_;
+        if (ref instanceof String) {
+          com.google.protobuf.ByteString b = 
+              com.google.protobuf.ByteString.copyFromUtf8(
+                  (java.lang.String) ref);
+          description_ = b;
+          return b;
+        } else {
+          return (com.google.protobuf.ByteString) ref;
+        }
+      }
+      /**
+       * <pre>
+       * Note: this will replace any inherited arg doc. There is no
+       * current way of modifying arg descriptions (other than replacing
+       * them entirely) as can be done with op descriptions.
+       * </pre>
+       *
+       * <code>string description = 3;</code>
+       */
+      public Builder setDescription(
+          java.lang.String value) {
+        if (value == null) {
+    throw new NullPointerException();
+  }
+  
+        description_ = value;
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * Note: this will replace any inherited arg doc. There is no
+       * current way of modifying arg descriptions (other than replacing
+       * them entirely) as can be done with op descriptions.
+       * </pre>
+       *
+       * <code>string description = 3;</code>
+       */
+      public Builder clearDescription() {
+        
+        description_ = getDefaultInstance().getDescription();
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * Note: this will replace any inherited arg doc. There is no
+       * current way of modifying arg descriptions (other than replacing
+       * them entirely) as can be done with op descriptions.
+       * </pre>
+       *
+       * <code>string description = 3;</code>
+       */
+      public Builder setDescriptionBytes(
+          com.google.protobuf.ByteString value) {
+        if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+        
+        description_ = value;
+        onChanged();
+        return this;
+      }
+      @java.lang.Override
+      public final Builder setUnknownFields(
+          final com.google.protobuf.UnknownFieldSet unknownFields) {
+        return super.setUnknownFields(unknownFields);
+      }
+
+      @java.lang.Override
+      public final Builder mergeUnknownFields(
+          final com.google.protobuf.UnknownFieldSet unknownFields) {
+        return super.mergeUnknownFields(unknownFields);
+      }
+
+
+      // @@protoc_insertion_point(builder_scope:tensorflow.ApiDef.Arg)
+    }
+
+    // @@protoc_insertion_point(class_scope:tensorflow.ApiDef.Arg)
+    private static final org.tensorflow.proto.framework.ApiDef.Arg DEFAULT_INSTANCE;
+    static {
+      DEFAULT_INSTANCE = new org.tensorflow.proto.framework.ApiDef.Arg();
+    }
+
+    public static org.tensorflow.proto.framework.ApiDef.Arg getDefaultInstance() {
+      return DEFAULT_INSTANCE;
+    }
+
+    private static final com.google.protobuf.Parser<Arg>
+        PARSER = new com.google.protobuf.AbstractParser<Arg>() {
+      @java.lang.Override
+      public Arg parsePartialFrom(
+          com.google.protobuf.CodedInputStream input,
+          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+          throws com.google.protobuf.InvalidProtocolBufferException {
+        return new Arg(input, extensionRegistry);
+      }
+    };
+
+    public static com.google.protobuf.Parser<Arg> parser() {
+      return PARSER;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Parser<Arg> getParserForType() {
+      return PARSER;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.ApiDef.Arg getDefaultInstanceForType() {
+      return DEFAULT_INSTANCE;
+    }
+
+  }
+
+  public interface AttrOrBuilder extends
+      // @@protoc_insertion_point(interface_extends:tensorflow.ApiDef.Attr)
+      com.google.protobuf.MessageOrBuilder {
+
+    /**
+     * <code>string name = 1;</code>
+     */
+    java.lang.String getName();
+    /**
+     * <code>string name = 1;</code>
+     */
+    com.google.protobuf.ByteString
+        getNameBytes();
+
+    /**
+     * <pre>
+     * Change the name used to access this attr in the API from what
+     * is used in the GraphDef.  Note that these names in `backticks`
+     * will also be replaced in the summary &amp; description fields.
+     * </pre>
+     *
+     * <code>string rename_to = 2;</code>
+     */
+    java.lang.String getRenameTo();
+    /**
+     * <pre>
+     * Change the name used to access this attr in the API from what
+     * is used in the GraphDef.  Note that these names in `backticks`
+     * will also be replaced in the summary &amp; description fields.
+     * </pre>
+     *
+     * <code>string rename_to = 2;</code>
+     */
+    com.google.protobuf.ByteString
+        getRenameToBytes();
+
+    /**
+     * <pre>
+     * Specify a new default value to use for this attr.  This default
+     * will be used when creating new graphs, as opposed to the
+     * default in the OpDef, which will be used when interpreting old
+     * GraphDefs.
+     * </pre>
+     *
+     * <code>.tensorflow.AttrValue default_value = 3;</code>
+     */
+    boolean hasDefaultValue();
+    /**
+     * <pre>
+     * Specify a new default value to use for this attr.  This default
+     * will be used when creating new graphs, as opposed to the
+     * default in the OpDef, which will be used when interpreting old
+     * GraphDefs.
+     * </pre>
+     *
+     * <code>.tensorflow.AttrValue default_value = 3;</code>
+     */
+    org.tensorflow.proto.framework.AttrValue getDefaultValue();
+    /**
+     * <pre>
+     * Specify a new default value to use for this attr.  This default
+     * will be used when creating new graphs, as opposed to the
+     * default in the OpDef, which will be used when interpreting old
+     * GraphDefs.
+     * </pre>
+     *
+     * <code>.tensorflow.AttrValue default_value = 3;</code>
+     */
+    org.tensorflow.proto.framework.AttrValueOrBuilder getDefaultValueOrBuilder();
+
+    /**
+     * <pre>
+     * Note: this will replace any inherited attr doc, there is no current
+     * way of modifying attr descriptions as can be done with op descriptions.
+     * </pre>
+     *
+     * <code>string description = 4;</code>
+     */
+    java.lang.String getDescription();
+    /**
+     * <pre>
+     * Note: this will replace any inherited attr doc, there is no current
+     * way of modifying attr descriptions as can be done with op descriptions.
+     * </pre>
+     *
+     * <code>string description = 4;</code>
+     */
+    com.google.protobuf.ByteString
+        getDescriptionBytes();
+  }
+  /**
+   * <pre>
+   * Description of the graph-construction-time configuration of this
+   * Op.  That is to say, this describes the attr fields that will
+   * be specified in the NodeDef.
+   * </pre>
+   *
+   * Protobuf type {@code tensorflow.ApiDef.Attr}
+   */
+  public  static final class Attr extends
+      com.google.protobuf.GeneratedMessageV3 implements
+      // @@protoc_insertion_point(message_implements:tensorflow.ApiDef.Attr)
+      AttrOrBuilder {
+  private static final long serialVersionUID = 0L;
+    // Use Attr.newBuilder() to construct.
+    private Attr(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+      super(builder);
+    }
+    private Attr() {
+      name_ = "";
+      renameTo_ = "";
+      description_ = "";
+    }
+
+    @java.lang.Override
+    @SuppressWarnings({"unused"})
+    protected java.lang.Object newInstance(
+        UnusedPrivateParameter unused) {
+      return new Attr();
+    }
+
+    @java.lang.Override
+    public final com.google.protobuf.UnknownFieldSet
+    getUnknownFields() {
+      return this.unknownFields;
+    }
+    private Attr(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      this();
+      if (extensionRegistry == null) {
+        throw new java.lang.NullPointerException();
+      }
+      com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+          com.google.protobuf.UnknownFieldSet.newBuilder();
+      try {
+        boolean done = false;
+        while (!done) {
+          int tag = input.readTag();
+          switch (tag) {
+            case 0:
+              done = true;
+              break;
+            case 10: {
+              java.lang.String s = input.readStringRequireUtf8();
+
+              name_ = s;
+              break;
+            }
+            case 18: {
+              java.lang.String s = input.readStringRequireUtf8();
+
+              renameTo_ = s;
+              break;
+            }
+            case 26: {
+              org.tensorflow.proto.framework.AttrValue.Builder subBuilder = null;
+              if (defaultValue_ != null) {
+                subBuilder = defaultValue_.toBuilder();
+              }
+              defaultValue_ = input.readMessage(org.tensorflow.proto.framework.AttrValue.parser(), extensionRegistry);
+              if (subBuilder != null) {
+                subBuilder.mergeFrom(defaultValue_);
+                defaultValue_ = subBuilder.buildPartial();
+              }
+
+              break;
+            }
+            case 34: {
+              java.lang.String s = input.readStringRequireUtf8();
+
+              description_ = s;
+              break;
+            }
+            default: {
+              if (!parseUnknownField(
+                  input, unknownFields, extensionRegistry, tag)) {
+                done = true;
+              }
+              break;
+            }
+          }
+        }
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        throw e.setUnfinishedMessage(this);
+      } catch (java.io.IOException e) {
+        throw new com.google.protobuf.InvalidProtocolBufferException(
+            e).setUnfinishedMessage(this);
+      } finally {
+        this.unknownFields = unknownFields.build();
+        makeExtensionsImmutable();
+      }
+    }
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.ApiDefProtos.internal_static_tensorflow_ApiDef_Attr_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.ApiDefProtos.internal_static_tensorflow_ApiDef_Attr_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.ApiDef.Attr.class, org.tensorflow.proto.framework.ApiDef.Attr.Builder.class);
+    }
+
+    public static final int NAME_FIELD_NUMBER = 1;
+    private volatile java.lang.Object name_;
+    /**
+     * <code>string name = 1;</code>
+     */
+    public java.lang.String getName() {
+      java.lang.Object ref = name_;
+      if (ref instanceof java.lang.String) {
+        return (java.lang.String) ref;
+      } else {
+        com.google.protobuf.ByteString bs = 
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        name_ = s;
+        return s;
+      }
+    }
+    /**
+     * <code>string name = 1;</code>
+     */
+    public com.google.protobuf.ByteString
+        getNameBytes() {
+      java.lang.Object ref = name_;
+      if (ref instanceof java.lang.String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        name_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+
+    public static final int RENAME_TO_FIELD_NUMBER = 2;
+    private volatile java.lang.Object renameTo_;
+    /**
+     * <pre>
+     * Change the name used to access this attr in the API from what
+     * is used in the GraphDef.  Note that these names in `backticks`
+     * will also be replaced in the summary &amp; description fields.
+     * </pre>
+     *
+     * <code>string rename_to = 2;</code>
+     */
+    public java.lang.String getRenameTo() {
+      java.lang.Object ref = renameTo_;
+      if (ref instanceof java.lang.String) {
+        return (java.lang.String) ref;
+      } else {
+        com.google.protobuf.ByteString bs = 
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        renameTo_ = s;
+        return s;
+      }
+    }
+    /**
+     * <pre>
+     * Change the name used to access this attr in the API from what
+     * is used in the GraphDef.  Note that these names in `backticks`
+     * will also be replaced in the summary &amp; description fields.
+     * </pre>
+     *
+     * <code>string rename_to = 2;</code>
+     */
+    public com.google.protobuf.ByteString
+        getRenameToBytes() {
+      java.lang.Object ref = renameTo_;
+      if (ref instanceof java.lang.String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        renameTo_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+
+    public static final int DEFAULT_VALUE_FIELD_NUMBER = 3;
+    private org.tensorflow.proto.framework.AttrValue defaultValue_;
+    /**
+     * <pre>
+     * Specify a new default value to use for this attr.  This default
+     * will be used when creating new graphs, as opposed to the
+     * default in the OpDef, which will be used when interpreting old
+     * GraphDefs.
+     * </pre>
+     *
+     * <code>.tensorflow.AttrValue default_value = 3;</code>
+     */
+    public boolean hasDefaultValue() {
+      return defaultValue_ != null;
+    }
+    /**
+     * <pre>
+     * Specify a new default value to use for this attr.  This default
+     * will be used when creating new graphs, as opposed to the
+     * default in the OpDef, which will be used when interpreting old
+     * GraphDefs.
+     * </pre>
+     *
+     * <code>.tensorflow.AttrValue default_value = 3;</code>
+     */
+    public org.tensorflow.proto.framework.AttrValue getDefaultValue() {
+      return defaultValue_ == null ? org.tensorflow.proto.framework.AttrValue.getDefaultInstance() : defaultValue_;
+    }
+    /**
+     * <pre>
+     * Specify a new default value to use for this attr.  This default
+     * will be used when creating new graphs, as opposed to the
+     * default in the OpDef, which will be used when interpreting old
+     * GraphDefs.
+     * </pre>
+     *
+     * <code>.tensorflow.AttrValue default_value = 3;</code>
+     */
+    public org.tensorflow.proto.framework.AttrValueOrBuilder getDefaultValueOrBuilder() {
+      return getDefaultValue();
+    }
+
+    public static final int DESCRIPTION_FIELD_NUMBER = 4;
+    private volatile java.lang.Object description_;
+    /**
+     * <pre>
+     * Note: this will replace any inherited attr doc, there is no current
+     * way of modifying attr descriptions as can be done with op descriptions.
+     * </pre>
+     *
+     * <code>string description = 4;</code>
+     */
+    public java.lang.String getDescription() {
+      java.lang.Object ref = description_;
+      if (ref instanceof java.lang.String) {
+        return (java.lang.String) ref;
+      } else {
+        com.google.protobuf.ByteString bs = 
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        description_ = s;
+        return s;
+      }
+    }
+    /**
+     * <pre>
+     * Note: this will replace any inherited attr doc, there is no current
+     * way of modifying attr descriptions as can be done with op descriptions.
+     * </pre>
+     *
+     * <code>string description = 4;</code>
+     */
+    public com.google.protobuf.ByteString
+        getDescriptionBytes() {
+      java.lang.Object ref = description_;
+      if (ref instanceof java.lang.String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        description_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+
+    private byte memoizedIsInitialized = -1;
+    @java.lang.Override
+    public final boolean isInitialized() {
+      byte isInitialized = memoizedIsInitialized;
+      if (isInitialized == 1) return true;
+      if (isInitialized == 0) return false;
+
+      memoizedIsInitialized = 1;
+      return true;
+    }
+
+    @java.lang.Override
+    public void writeTo(com.google.protobuf.CodedOutputStream output)
+                        throws java.io.IOException {
+      if (!getNameBytes().isEmpty()) {
+        com.google.protobuf.GeneratedMessageV3.writeString(output, 1, name_);
+      }
+      if (!getRenameToBytes().isEmpty()) {
+        com.google.protobuf.GeneratedMessageV3.writeString(output, 2, renameTo_);
+      }
+      if (defaultValue_ != null) {
+        output.writeMessage(3, getDefaultValue());
+      }
+      if (!getDescriptionBytes().isEmpty()) {
+        com.google.protobuf.GeneratedMessageV3.writeString(output, 4, description_);
+      }
+      unknownFields.writeTo(output);
+    }
+
+    @java.lang.Override
+    public int getSerializedSize() {
+      int size = memoizedSize;
+      if (size != -1) return size;
+
+      size = 0;
+      if (!getNameBytes().isEmpty()) {
+        size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, name_);
+      }
+      if (!getRenameToBytes().isEmpty()) {
+        size += com.google.protobuf.GeneratedMessageV3.computeStringSize(2, renameTo_);
+      }
+      if (defaultValue_ != null) {
+        size += com.google.protobuf.CodedOutputStream
+          .computeMessageSize(3, getDefaultValue());
+      }
+      if (!getDescriptionBytes().isEmpty()) {
+        size += com.google.protobuf.GeneratedMessageV3.computeStringSize(4, description_);
+      }
+      size += unknownFields.getSerializedSize();
+      memoizedSize = size;
+      return size;
+    }
+
+    @java.lang.Override
+    public boolean equals(final java.lang.Object obj) {
+      if (obj == this) {
+       return true;
+      }
+      if (!(obj instanceof org.tensorflow.proto.framework.ApiDef.Attr)) {
+        return super.equals(obj);
+      }
+      org.tensorflow.proto.framework.ApiDef.Attr other = (org.tensorflow.proto.framework.ApiDef.Attr) obj;
+
+      if (!getName()
+          .equals(other.getName())) return false;
+      if (!getRenameTo()
+          .equals(other.getRenameTo())) return false;
+      if (hasDefaultValue() != other.hasDefaultValue()) return false;
+      if (hasDefaultValue()) {
+        if (!getDefaultValue()
+            .equals(other.getDefaultValue())) return false;
+      }
+      if (!getDescription()
+          .equals(other.getDescription())) return false;
+      if (!unknownFields.equals(other.unknownFields)) return false;
+      return true;
+    }
+
+    @java.lang.Override
+    public int hashCode() {
+      if (memoizedHashCode != 0) {
+        return memoizedHashCode;
+      }
+      int hash = 41;
+      hash = (19 * hash) + getDescriptor().hashCode();
+      hash = (37 * hash) + NAME_FIELD_NUMBER;
+      hash = (53 * hash) + getName().hashCode();
+      hash = (37 * hash) + RENAME_TO_FIELD_NUMBER;
+      hash = (53 * hash) + getRenameTo().hashCode();
+      if (hasDefaultValue()) {
+        hash = (37 * hash) + DEFAULT_VALUE_FIELD_NUMBER;
+        hash = (53 * hash) + getDefaultValue().hashCode();
+      }
+      hash = (37 * hash) + DESCRIPTION_FIELD_NUMBER;
+      hash = (53 * hash) + getDescription().hashCode();
+      hash = (29 * hash) + unknownFields.hashCode();
+      memoizedHashCode = hash;
+      return hash;
+    }
+
+    public static org.tensorflow.proto.framework.ApiDef.Attr parseFrom(
+        java.nio.ByteBuffer data)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data);
+    }
+    public static org.tensorflow.proto.framework.ApiDef.Attr parseFrom(
+        java.nio.ByteBuffer data,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.ApiDef.Attr parseFrom(
+        com.google.protobuf.ByteString data)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data);
+    }
+    public static org.tensorflow.proto.framework.ApiDef.Attr parseFrom(
+        com.google.protobuf.ByteString data,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.ApiDef.Attr parseFrom(byte[] data)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data);
+    }
+    public static org.tensorflow.proto.framework.ApiDef.Attr parseFrom(
+        byte[] data,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.ApiDef.Attr parseFrom(java.io.InputStream input)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input);
+    }
+    public static org.tensorflow.proto.framework.ApiDef.Attr parseFrom(
+        java.io.InputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.ApiDef.Attr parseDelimitedFrom(java.io.InputStream input)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseDelimitedWithIOException(PARSER, input);
+    }
+    public static org.tensorflow.proto.framework.ApiDef.Attr parseDelimitedFrom(
+        java.io.InputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.ApiDef.Attr parseFrom(
+        com.google.protobuf.CodedInputStream input)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input);
+    }
+    public static org.tensorflow.proto.framework.ApiDef.Attr parseFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input, extensionRegistry);
+    }
+
+    @java.lang.Override
+    public Builder newBuilderForType() { return newBuilder(); }
+    public static Builder newBuilder() {
+      return DEFAULT_INSTANCE.toBuilder();
+    }
+    public static Builder newBuilder(org.tensorflow.proto.framework.ApiDef.Attr prototype) {
+      return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+    }
+    @java.lang.Override
+    public Builder toBuilder() {
+      return this == DEFAULT_INSTANCE
+          ? new Builder() : new Builder().mergeFrom(this);
+    }
+
+    @java.lang.Override
+    protected Builder newBuilderForType(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      Builder builder = new Builder(parent);
+      return builder;
+    }
+    /**
+     * <pre>
+     * Description of the graph-construction-time configuration of this
+     * Op.  That is to say, this describes the attr fields that will
+     * be specified in the NodeDef.
+     * </pre>
+     *
+     * Protobuf type {@code tensorflow.ApiDef.Attr}
+     */
+    public static final class Builder extends
+        com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+        // @@protoc_insertion_point(builder_implements:tensorflow.ApiDef.Attr)
+        org.tensorflow.proto.framework.ApiDef.AttrOrBuilder {
+      public static final com.google.protobuf.Descriptors.Descriptor
+          getDescriptor() {
+        return org.tensorflow.proto.framework.ApiDefProtos.internal_static_tensorflow_ApiDef_Attr_descriptor;
+      }
+
+      @java.lang.Override
+      protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+          internalGetFieldAccessorTable() {
+        return org.tensorflow.proto.framework.ApiDefProtos.internal_static_tensorflow_ApiDef_Attr_fieldAccessorTable
+            .ensureFieldAccessorsInitialized(
+                org.tensorflow.proto.framework.ApiDef.Attr.class, org.tensorflow.proto.framework.ApiDef.Attr.Builder.class);
+      }
+
+      // Construct using org.tensorflow.proto.framework.ApiDef.Attr.newBuilder()
+      private Builder() {
+        maybeForceBuilderInitialization();
+      }
+
+      private Builder(
+          com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+        super(parent);
+        maybeForceBuilderInitialization();
+      }
+      private void maybeForceBuilderInitialization() {
+        if (com.google.protobuf.GeneratedMessageV3
+                .alwaysUseFieldBuilders) {
+        }
+      }
+      @java.lang.Override
+      public Builder clear() {
+        super.clear();
+        name_ = "";
+
+        renameTo_ = "";
+
+        if (defaultValueBuilder_ == null) {
+          defaultValue_ = null;
+        } else {
+          defaultValue_ = null;
+          defaultValueBuilder_ = null;
+        }
+        description_ = "";
+
+        return this;
+      }
+
+      @java.lang.Override
+      public com.google.protobuf.Descriptors.Descriptor
+          getDescriptorForType() {
+        return org.tensorflow.proto.framework.ApiDefProtos.internal_static_tensorflow_ApiDef_Attr_descriptor;
+      }
+
+      @java.lang.Override
+      public org.tensorflow.proto.framework.ApiDef.Attr getDefaultInstanceForType() {
+        return org.tensorflow.proto.framework.ApiDef.Attr.getDefaultInstance();
+      }
+
+      @java.lang.Override
+      public org.tensorflow.proto.framework.ApiDef.Attr build() {
+        org.tensorflow.proto.framework.ApiDef.Attr result = buildPartial();
+        if (!result.isInitialized()) {
+          throw newUninitializedMessageException(result);
+        }
+        return result;
+      }
+
+      @java.lang.Override
+      public org.tensorflow.proto.framework.ApiDef.Attr buildPartial() {
+        org.tensorflow.proto.framework.ApiDef.Attr result = new org.tensorflow.proto.framework.ApiDef.Attr(this);
+        result.name_ = name_;
+        result.renameTo_ = renameTo_;
+        if (defaultValueBuilder_ == null) {
+          result.defaultValue_ = defaultValue_;
+        } else {
+          result.defaultValue_ = defaultValueBuilder_.build();
+        }
+        result.description_ = description_;
+        onBuilt();
+        return result;
+      }
+
+      @java.lang.Override
+      public Builder clone() {
+        return super.clone();
+      }
+      @java.lang.Override
+      public Builder setField(
+          com.google.protobuf.Descriptors.FieldDescriptor field,
+          java.lang.Object value) {
+        return super.setField(field, value);
+      }
+      @java.lang.Override
+      public Builder clearField(
+          com.google.protobuf.Descriptors.FieldDescriptor field) {
+        return super.clearField(field);
+      }
+      @java.lang.Override
+      public Builder clearOneof(
+          com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+        return super.clearOneof(oneof);
+      }
+      @java.lang.Override
+      public Builder setRepeatedField(
+          com.google.protobuf.Descriptors.FieldDescriptor field,
+          int index, java.lang.Object value) {
+        return super.setRepeatedField(field, index, value);
+      }
+      @java.lang.Override
+      public Builder addRepeatedField(
+          com.google.protobuf.Descriptors.FieldDescriptor field,
+          java.lang.Object value) {
+        return super.addRepeatedField(field, value);
+      }
+      @java.lang.Override
+      public Builder mergeFrom(com.google.protobuf.Message other) {
+        if (other instanceof org.tensorflow.proto.framework.ApiDef.Attr) {
+          return mergeFrom((org.tensorflow.proto.framework.ApiDef.Attr)other);
+        } else {
+          super.mergeFrom(other);
+          return this;
+        }
+      }
+
+      public Builder mergeFrom(org.tensorflow.proto.framework.ApiDef.Attr other) {
+        if (other == org.tensorflow.proto.framework.ApiDef.Attr.getDefaultInstance()) return this;
+        if (!other.getName().isEmpty()) {
+          name_ = other.name_;
+          onChanged();
+        }
+        if (!other.getRenameTo().isEmpty()) {
+          renameTo_ = other.renameTo_;
+          onChanged();
+        }
+        if (other.hasDefaultValue()) {
+          mergeDefaultValue(other.getDefaultValue());
+        }
+        if (!other.getDescription().isEmpty()) {
+          description_ = other.description_;
+          onChanged();
+        }
+        this.mergeUnknownFields(other.unknownFields);
+        onChanged();
+        return this;
+      }
+
+      @java.lang.Override
+      public final boolean isInitialized() {
+        return true;
+      }
+
+      @java.lang.Override
+      public Builder mergeFrom(
+          com.google.protobuf.CodedInputStream input,
+          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+          throws java.io.IOException {
+        org.tensorflow.proto.framework.ApiDef.Attr parsedMessage = null;
+        try {
+          parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+        } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+          parsedMessage = (org.tensorflow.proto.framework.ApiDef.Attr) e.getUnfinishedMessage();
+          throw e.unwrapIOException();
+        } finally {
+          if (parsedMessage != null) {
+            mergeFrom(parsedMessage);
+          }
+        }
+        return this;
+      }
+
+      private java.lang.Object name_ = "";
+      /**
+       * <code>string name = 1;</code>
+       */
+      public java.lang.String getName() {
+        java.lang.Object ref = name_;
+        if (!(ref instanceof java.lang.String)) {
+          com.google.protobuf.ByteString bs =
+              (com.google.protobuf.ByteString) ref;
+          java.lang.String s = bs.toStringUtf8();
+          name_ = s;
+          return s;
+        } else {
+          return (java.lang.String) ref;
+        }
+      }
+      /**
+       * <code>string name = 1;</code>
+       */
+      public com.google.protobuf.ByteString
+          getNameBytes() {
+        java.lang.Object ref = name_;
+        if (ref instanceof String) {
+          com.google.protobuf.ByteString b = 
+              com.google.protobuf.ByteString.copyFromUtf8(
+                  (java.lang.String) ref);
+          name_ = b;
+          return b;
+        } else {
+          return (com.google.protobuf.ByteString) ref;
+        }
+      }
+      /**
+       * <code>string name = 1;</code>
+       */
+      public Builder setName(
+          java.lang.String value) {
+        if (value == null) {
+    throw new NullPointerException();
+  }
+  
+        name_ = value;
+        onChanged();
+        return this;
+      }
+      /**
+       * <code>string name = 1;</code>
+       */
+      public Builder clearName() {
+        
+        name_ = getDefaultInstance().getName();
+        onChanged();
+        return this;
+      }
+      /**
+       * <code>string name = 1;</code>
+       */
+      public Builder setNameBytes(
+          com.google.protobuf.ByteString value) {
+        if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+        
+        name_ = value;
+        onChanged();
+        return this;
+      }
+
+      private java.lang.Object renameTo_ = "";
+      /**
+       * <pre>
+       * Change the name used to access this attr in the API from what
+       * is used in the GraphDef.  Note that these names in `backticks`
+       * will also be replaced in the summary &amp; description fields.
+       * </pre>
+       *
+       * <code>string rename_to = 2;</code>
+       */
+      public java.lang.String getRenameTo() {
+        java.lang.Object ref = renameTo_;
+        if (!(ref instanceof java.lang.String)) {
+          com.google.protobuf.ByteString bs =
+              (com.google.protobuf.ByteString) ref;
+          java.lang.String s = bs.toStringUtf8();
+          renameTo_ = s;
+          return s;
+        } else {
+          return (java.lang.String) ref;
+        }
+      }
+      /**
+       * <pre>
+       * Change the name used to access this attr in the API from what
+       * is used in the GraphDef.  Note that these names in `backticks`
+       * will also be replaced in the summary &amp; description fields.
+       * </pre>
+       *
+       * <code>string rename_to = 2;</code>
+       */
+      public com.google.protobuf.ByteString
+          getRenameToBytes() {
+        java.lang.Object ref = renameTo_;
+        if (ref instanceof String) {
+          com.google.protobuf.ByteString b = 
+              com.google.protobuf.ByteString.copyFromUtf8(
+                  (java.lang.String) ref);
+          renameTo_ = b;
+          return b;
+        } else {
+          return (com.google.protobuf.ByteString) ref;
+        }
+      }
+      /**
+       * <pre>
+       * Change the name used to access this attr in the API from what
+       * is used in the GraphDef.  Note that these names in `backticks`
+       * will also be replaced in the summary &amp; description fields.
+       * </pre>
+       *
+       * <code>string rename_to = 2;</code>
+       */
+      public Builder setRenameTo(
+          java.lang.String value) {
+        if (value == null) {
+    throw new NullPointerException();
+  }
+  
+        renameTo_ = value;
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * Change the name used to access this attr in the API from what
+       * is used in the GraphDef.  Note that these names in `backticks`
+       * will also be replaced in the summary &amp; description fields.
+       * </pre>
+       *
+       * <code>string rename_to = 2;</code>
+       */
+      public Builder clearRenameTo() {
+        
+        renameTo_ = getDefaultInstance().getRenameTo();
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * Change the name used to access this attr in the API from what
+       * is used in the GraphDef.  Note that these names in `backticks`
+       * will also be replaced in the summary &amp; description fields.
+       * </pre>
+       *
+       * <code>string rename_to = 2;</code>
+       */
+      public Builder setRenameToBytes(
+          com.google.protobuf.ByteString value) {
+        if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+        
+        renameTo_ = value;
+        onChanged();
+        return this;
+      }
+
+      private org.tensorflow.proto.framework.AttrValue defaultValue_;
+      private com.google.protobuf.SingleFieldBuilderV3<
+          org.tensorflow.proto.framework.AttrValue, org.tensorflow.proto.framework.AttrValue.Builder, org.tensorflow.proto.framework.AttrValueOrBuilder> defaultValueBuilder_;
+      /**
+       * <pre>
+       * Specify a new default value to use for this attr.  This default
+       * will be used when creating new graphs, as opposed to the
+       * default in the OpDef, which will be used when interpreting old
+       * GraphDefs.
+       * </pre>
+       *
+       * <code>.tensorflow.AttrValue default_value = 3;</code>
+       */
+      public boolean hasDefaultValue() {
+        return defaultValueBuilder_ != null || defaultValue_ != null;
+      }
+      /**
+       * <pre>
+       * Specify a new default value to use for this attr.  This default
+       * will be used when creating new graphs, as opposed to the
+       * default in the OpDef, which will be used when interpreting old
+       * GraphDefs.
+       * </pre>
+       *
+       * <code>.tensorflow.AttrValue default_value = 3;</code>
+       */
+      public org.tensorflow.proto.framework.AttrValue getDefaultValue() {
+        if (defaultValueBuilder_ == null) {
+          return defaultValue_ == null ? org.tensorflow.proto.framework.AttrValue.getDefaultInstance() : defaultValue_;
+        } else {
+          return defaultValueBuilder_.getMessage();
+        }
+      }
+      /**
+       * <pre>
+       * Specify a new default value to use for this attr.  This default
+       * will be used when creating new graphs, as opposed to the
+       * default in the OpDef, which will be used when interpreting old
+       * GraphDefs.
+       * </pre>
+       *
+       * <code>.tensorflow.AttrValue default_value = 3;</code>
+       */
+      public Builder setDefaultValue(org.tensorflow.proto.framework.AttrValue value) {
+        if (defaultValueBuilder_ == null) {
+          if (value == null) {
+            throw new NullPointerException();
+          }
+          defaultValue_ = value;
+          onChanged();
+        } else {
+          defaultValueBuilder_.setMessage(value);
+        }
+
+        return this;
+      }
+      /**
+       * <pre>
+       * Specify a new default value to use for this attr.  This default
+       * will be used when creating new graphs, as opposed to the
+       * default in the OpDef, which will be used when interpreting old
+       * GraphDefs.
+       * </pre>
+       *
+       * <code>.tensorflow.AttrValue default_value = 3;</code>
+       */
+      public Builder setDefaultValue(
+          org.tensorflow.proto.framework.AttrValue.Builder builderForValue) {
+        if (defaultValueBuilder_ == null) {
+          defaultValue_ = builderForValue.build();
+          onChanged();
+        } else {
+          defaultValueBuilder_.setMessage(builderForValue.build());
+        }
+
+        return this;
+      }
+      /**
+       * <pre>
+       * Specify a new default value to use for this attr.  This default
+       * will be used when creating new graphs, as opposed to the
+       * default in the OpDef, which will be used when interpreting old
+       * GraphDefs.
+       * </pre>
+       *
+       * <code>.tensorflow.AttrValue default_value = 3;</code>
+       */
+      public Builder mergeDefaultValue(org.tensorflow.proto.framework.AttrValue value) {
+        if (defaultValueBuilder_ == null) {
+          if (defaultValue_ != null) {
+            defaultValue_ =
+              org.tensorflow.proto.framework.AttrValue.newBuilder(defaultValue_).mergeFrom(value).buildPartial();
+          } else {
+            defaultValue_ = value;
+          }
+          onChanged();
+        } else {
+          defaultValueBuilder_.mergeFrom(value);
+        }
+
+        return this;
+      }
+      /**
+       * <pre>
+       * Specify a new default value to use for this attr.  This default
+       * will be used when creating new graphs, as opposed to the
+       * default in the OpDef, which will be used when interpreting old
+       * GraphDefs.
+       * </pre>
+       *
+       * <code>.tensorflow.AttrValue default_value = 3;</code>
+       */
+      public Builder clearDefaultValue() {
+        if (defaultValueBuilder_ == null) {
+          defaultValue_ = null;
+          onChanged();
+        } else {
+          defaultValue_ = null;
+          defaultValueBuilder_ = null;
+        }
+
+        return this;
+      }
+      /**
+       * <pre>
+       * Specify a new default value to use for this attr.  This default
+       * will be used when creating new graphs, as opposed to the
+       * default in the OpDef, which will be used when interpreting old
+       * GraphDefs.
+       * </pre>
+       *
+       * <code>.tensorflow.AttrValue default_value = 3;</code>
+       */
+      public org.tensorflow.proto.framework.AttrValue.Builder getDefaultValueBuilder() {
+        
+        onChanged();
+        return getDefaultValueFieldBuilder().getBuilder();
+      }
+      /**
+       * <pre>
+       * Specify a new default value to use for this attr.  This default
+       * will be used when creating new graphs, as opposed to the
+       * default in the OpDef, which will be used when interpreting old
+       * GraphDefs.
+       * </pre>
+       *
+       * <code>.tensorflow.AttrValue default_value = 3;</code>
+       */
+      public org.tensorflow.proto.framework.AttrValueOrBuilder getDefaultValueOrBuilder() {
+        if (defaultValueBuilder_ != null) {
+          return defaultValueBuilder_.getMessageOrBuilder();
+        } else {
+          return defaultValue_ == null ?
+              org.tensorflow.proto.framework.AttrValue.getDefaultInstance() : defaultValue_;
+        }
+      }
+      /**
+       * <pre>
+       * Specify a new default value to use for this attr.  This default
+       * will be used when creating new graphs, as opposed to the
+       * default in the OpDef, which will be used when interpreting old
+       * GraphDefs.
+       * </pre>
+       *
+       * <code>.tensorflow.AttrValue default_value = 3;</code>
+       */
+      private com.google.protobuf.SingleFieldBuilderV3<
+          org.tensorflow.proto.framework.AttrValue, org.tensorflow.proto.framework.AttrValue.Builder, org.tensorflow.proto.framework.AttrValueOrBuilder> 
+          getDefaultValueFieldBuilder() {
+        if (defaultValueBuilder_ == null) {
+          defaultValueBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+              org.tensorflow.proto.framework.AttrValue, org.tensorflow.proto.framework.AttrValue.Builder, org.tensorflow.proto.framework.AttrValueOrBuilder>(
+                  getDefaultValue(),
+                  getParentForChildren(),
+                  isClean());
+          defaultValue_ = null;
+        }
+        return defaultValueBuilder_;
+      }
+
+      private java.lang.Object description_ = "";
+      /**
+       * <pre>
+       * Note: this will replace any inherited attr doc, there is no current
+       * way of modifying attr descriptions as can be done with op descriptions.
+       * </pre>
+       *
+       * <code>string description = 4;</code>
+       */
+      public java.lang.String getDescription() {
+        java.lang.Object ref = description_;
+        if (!(ref instanceof java.lang.String)) {
+          com.google.protobuf.ByteString bs =
+              (com.google.protobuf.ByteString) ref;
+          java.lang.String s = bs.toStringUtf8();
+          description_ = s;
+          return s;
+        } else {
+          return (java.lang.String) ref;
+        }
+      }
+      /**
+       * <pre>
+       * Note: this will replace any inherited attr doc, there is no current
+       * way of modifying attr descriptions as can be done with op descriptions.
+       * </pre>
+       *
+       * <code>string description = 4;</code>
+       */
+      public com.google.protobuf.ByteString
+          getDescriptionBytes() {
+        java.lang.Object ref = description_;
+        if (ref instanceof String) {
+          com.google.protobuf.ByteString b = 
+              com.google.protobuf.ByteString.copyFromUtf8(
+                  (java.lang.String) ref);
+          description_ = b;
+          return b;
+        } else {
+          return (com.google.protobuf.ByteString) ref;
+        }
+      }
+      /**
+       * <pre>
+       * Note: this will replace any inherited attr doc, there is no current
+       * way of modifying attr descriptions as can be done with op descriptions.
+       * </pre>
+       *
+       * <code>string description = 4;</code>
+       */
+      public Builder setDescription(
+          java.lang.String value) {
+        if (value == null) {
+    throw new NullPointerException();
+  }
+  
+        description_ = value;
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * Note: this will replace any inherited attr doc, there is no current
+       * way of modifying attr descriptions as can be done with op descriptions.
+       * </pre>
+       *
+       * <code>string description = 4;</code>
+       */
+      public Builder clearDescription() {
+        
+        description_ = getDefaultInstance().getDescription();
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * Note: this will replace any inherited attr doc, there is no current
+       * way of modifying attr descriptions as can be done with op descriptions.
+       * </pre>
+       *
+       * <code>string description = 4;</code>
+       */
+      public Builder setDescriptionBytes(
+          com.google.protobuf.ByteString value) {
+        if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+        
+        description_ = value;
+        onChanged();
+        return this;
+      }
+      @java.lang.Override
+      public final Builder setUnknownFields(
+          final com.google.protobuf.UnknownFieldSet unknownFields) {
+        return super.setUnknownFields(unknownFields);
+      }
+
+      @java.lang.Override
+      public final Builder mergeUnknownFields(
+          final com.google.protobuf.UnknownFieldSet unknownFields) {
+        return super.mergeUnknownFields(unknownFields);
+      }
+
+
+      // @@protoc_insertion_point(builder_scope:tensorflow.ApiDef.Attr)
+    }
+
+    // @@protoc_insertion_point(class_scope:tensorflow.ApiDef.Attr)
+    private static final org.tensorflow.proto.framework.ApiDef.Attr DEFAULT_INSTANCE;
+    static {
+      DEFAULT_INSTANCE = new org.tensorflow.proto.framework.ApiDef.Attr();
+    }
+
+    public static org.tensorflow.proto.framework.ApiDef.Attr getDefaultInstance() {
+      return DEFAULT_INSTANCE;
+    }
+
+    private static final com.google.protobuf.Parser<Attr>
+        PARSER = new com.google.protobuf.AbstractParser<Attr>() {
+      @java.lang.Override
+      public Attr parsePartialFrom(
+          com.google.protobuf.CodedInputStream input,
+          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+          throws com.google.protobuf.InvalidProtocolBufferException {
+        return new Attr(input, extensionRegistry);
+      }
+    };
+
+    public static com.google.protobuf.Parser<Attr> parser() {
+      return PARSER;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Parser<Attr> getParserForType() {
+      return PARSER;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.ApiDef.Attr getDefaultInstanceForType() {
+      return DEFAULT_INSTANCE;
+    }
+
+  }
+
+  public static final int GRAPH_OP_NAME_FIELD_NUMBER = 1;
+  private volatile java.lang.Object graphOpName_;
+  /**
+   * <pre>
+   * Name of the op (in the OpDef) to specify the API for.
+   * </pre>
+   *
+   * <code>string graph_op_name = 1;</code>
+   */
+  public java.lang.String getGraphOpName() {
+    java.lang.Object ref = graphOpName_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      graphOpName_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * Name of the op (in the OpDef) to specify the API for.
+   * </pre>
+   *
+   * <code>string graph_op_name = 1;</code>
+   */
+  public com.google.protobuf.ByteString
+      getGraphOpNameBytes() {
+    java.lang.Object ref = graphOpName_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      graphOpName_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int DEPRECATION_MESSAGE_FIELD_NUMBER = 12;
+  private volatile java.lang.Object deprecationMessage_;
+  /**
+   * <pre>
+   * If this op is deprecated, set deprecation message to the message
+   * that should be logged when this op is used.
+   * The message should indicate alternative op to use, if any.
+   * </pre>
+   *
+   * <code>string deprecation_message = 12;</code>
+   */
+  public java.lang.String getDeprecationMessage() {
+    java.lang.Object ref = deprecationMessage_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      deprecationMessage_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * If this op is deprecated, set deprecation message to the message
+   * that should be logged when this op is used.
+   * The message should indicate alternative op to use, if any.
+   * </pre>
+   *
+   * <code>string deprecation_message = 12;</code>
+   */
+  public com.google.protobuf.ByteString
+      getDeprecationMessageBytes() {
+    java.lang.Object ref = deprecationMessage_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      deprecationMessage_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int DEPRECATION_VERSION_FIELD_NUMBER = 13;
+  private int deprecationVersion_;
+  /**
+   * <pre>
+   * Major version when the op will be deleted. For e.g. set this
+   * value to 2 if op API should be removed in TensorFlow 2.0 and
+   * deprecated in versions before that.
+   * </pre>
+   *
+   * <code>int32 deprecation_version = 13;</code>
+   */
+  public int getDeprecationVersion() {
+    return deprecationVersion_;
+  }
+
+  public static final int VISIBILITY_FIELD_NUMBER = 2;
+  private int visibility_;
+  /**
+   * <code>.tensorflow.ApiDef.Visibility visibility = 2;</code>
+   */
+  public int getVisibilityValue() {
+    return visibility_;
+  }
+  /**
+   * <code>.tensorflow.ApiDef.Visibility visibility = 2;</code>
+   */
+  public org.tensorflow.proto.framework.ApiDef.Visibility getVisibility() {
+    @SuppressWarnings("deprecation")
+    org.tensorflow.proto.framework.ApiDef.Visibility result = org.tensorflow.proto.framework.ApiDef.Visibility.valueOf(visibility_);
+    return result == null ? org.tensorflow.proto.framework.ApiDef.Visibility.UNRECOGNIZED : result;
+  }
+
+  public static final int ENDPOINT_FIELD_NUMBER = 3;
+  private java.util.List<org.tensorflow.proto.framework.ApiDef.Endpoint> endpoint_;
+  /**
+   * <code>repeated .tensorflow.ApiDef.Endpoint endpoint = 3;</code>
+   */
+  public java.util.List<org.tensorflow.proto.framework.ApiDef.Endpoint> getEndpointList() {
+    return endpoint_;
+  }
+  /**
+   * <code>repeated .tensorflow.ApiDef.Endpoint endpoint = 3;</code>
+   */
+  public java.util.List<? extends org.tensorflow.proto.framework.ApiDef.EndpointOrBuilder> 
+      getEndpointOrBuilderList() {
+    return endpoint_;
+  }
+  /**
+   * <code>repeated .tensorflow.ApiDef.Endpoint endpoint = 3;</code>
+   */
+  public int getEndpointCount() {
+    return endpoint_.size();
+  }
+  /**
+   * <code>repeated .tensorflow.ApiDef.Endpoint endpoint = 3;</code>
+   */
+  public org.tensorflow.proto.framework.ApiDef.Endpoint getEndpoint(int index) {
+    return endpoint_.get(index);
+  }
+  /**
+   * <code>repeated .tensorflow.ApiDef.Endpoint endpoint = 3;</code>
+   */
+  public org.tensorflow.proto.framework.ApiDef.EndpointOrBuilder getEndpointOrBuilder(
+      int index) {
+    return endpoint_.get(index);
+  }
+
+  public static final int IN_ARG_FIELD_NUMBER = 4;
+  private java.util.List<org.tensorflow.proto.framework.ApiDef.Arg> inArg_;
+  /**
+   * <code>repeated .tensorflow.ApiDef.Arg in_arg = 4;</code>
+   */
+  public java.util.List<org.tensorflow.proto.framework.ApiDef.Arg> getInArgList() {
+    return inArg_;
+  }
+  /**
+   * <code>repeated .tensorflow.ApiDef.Arg in_arg = 4;</code>
+   */
+  public java.util.List<? extends org.tensorflow.proto.framework.ApiDef.ArgOrBuilder> 
+      getInArgOrBuilderList() {
+    return inArg_;
+  }
+  /**
+   * <code>repeated .tensorflow.ApiDef.Arg in_arg = 4;</code>
+   */
+  public int getInArgCount() {
+    return inArg_.size();
+  }
+  /**
+   * <code>repeated .tensorflow.ApiDef.Arg in_arg = 4;</code>
+   */
+  public org.tensorflow.proto.framework.ApiDef.Arg getInArg(int index) {
+    return inArg_.get(index);
+  }
+  /**
+   * <code>repeated .tensorflow.ApiDef.Arg in_arg = 4;</code>
+   */
+  public org.tensorflow.proto.framework.ApiDef.ArgOrBuilder getInArgOrBuilder(
+      int index) {
+    return inArg_.get(index);
+  }
+
+  public static final int OUT_ARG_FIELD_NUMBER = 5;
+  private java.util.List<org.tensorflow.proto.framework.ApiDef.Arg> outArg_;
+  /**
+   * <code>repeated .tensorflow.ApiDef.Arg out_arg = 5;</code>
+   */
+  public java.util.List<org.tensorflow.proto.framework.ApiDef.Arg> getOutArgList() {
+    return outArg_;
+  }
+  /**
+   * <code>repeated .tensorflow.ApiDef.Arg out_arg = 5;</code>
+   */
+  public java.util.List<? extends org.tensorflow.proto.framework.ApiDef.ArgOrBuilder> 
+      getOutArgOrBuilderList() {
+    return outArg_;
+  }
+  /**
+   * <code>repeated .tensorflow.ApiDef.Arg out_arg = 5;</code>
+   */
+  public int getOutArgCount() {
+    return outArg_.size();
+  }
+  /**
+   * <code>repeated .tensorflow.ApiDef.Arg out_arg = 5;</code>
+   */
+  public org.tensorflow.proto.framework.ApiDef.Arg getOutArg(int index) {
+    return outArg_.get(index);
+  }
+  /**
+   * <code>repeated .tensorflow.ApiDef.Arg out_arg = 5;</code>
+   */
+  public org.tensorflow.proto.framework.ApiDef.ArgOrBuilder getOutArgOrBuilder(
+      int index) {
+    return outArg_.get(index);
+  }
+
+  public static final int ARG_ORDER_FIELD_NUMBER = 11;
+  private com.google.protobuf.LazyStringList argOrder_;
+  /**
+   * <pre>
+   * List of original in_arg names to specify new argument order.
+   * Length of arg_order should be either empty to keep current order
+   * or match size of in_arg.
+   * </pre>
+   *
+   * <code>repeated string arg_order = 11;</code>
+   */
+  public com.google.protobuf.ProtocolStringList
+      getArgOrderList() {
+    return argOrder_;
+  }
+  /**
+   * <pre>
+   * List of original in_arg names to specify new argument order.
+   * Length of arg_order should be either empty to keep current order
+   * or match size of in_arg.
+   * </pre>
+   *
+   * <code>repeated string arg_order = 11;</code>
+   */
+  public int getArgOrderCount() {
+    return argOrder_.size();
+  }
+  /**
+   * <pre>
+   * List of original in_arg names to specify new argument order.
+   * Length of arg_order should be either empty to keep current order
+   * or match size of in_arg.
+   * </pre>
+   *
+   * <code>repeated string arg_order = 11;</code>
+   */
+  public java.lang.String getArgOrder(int index) {
+    return argOrder_.get(index);
+  }
+  /**
+   * <pre>
+   * List of original in_arg names to specify new argument order.
+   * Length of arg_order should be either empty to keep current order
+   * or match size of in_arg.
+   * </pre>
+   *
+   * <code>repeated string arg_order = 11;</code>
+   */
+  public com.google.protobuf.ByteString
+      getArgOrderBytes(int index) {
+    return argOrder_.getByteString(index);
+  }
+
+  public static final int ATTR_FIELD_NUMBER = 6;
+  private java.util.List<org.tensorflow.proto.framework.ApiDef.Attr> attr_;
+  /**
+   * <code>repeated .tensorflow.ApiDef.Attr attr = 6;</code>
+   */
+  public java.util.List<org.tensorflow.proto.framework.ApiDef.Attr> getAttrList() {
+    return attr_;
+  }
+  /**
+   * <code>repeated .tensorflow.ApiDef.Attr attr = 6;</code>
+   */
+  public java.util.List<? extends org.tensorflow.proto.framework.ApiDef.AttrOrBuilder> 
+      getAttrOrBuilderList() {
+    return attr_;
+  }
+  /**
+   * <code>repeated .tensorflow.ApiDef.Attr attr = 6;</code>
+   */
+  public int getAttrCount() {
+    return attr_.size();
+  }
+  /**
+   * <code>repeated .tensorflow.ApiDef.Attr attr = 6;</code>
+   */
+  public org.tensorflow.proto.framework.ApiDef.Attr getAttr(int index) {
+    return attr_.get(index);
+  }
+  /**
+   * <code>repeated .tensorflow.ApiDef.Attr attr = 6;</code>
+   */
+  public org.tensorflow.proto.framework.ApiDef.AttrOrBuilder getAttrOrBuilder(
+      int index) {
+    return attr_.get(index);
+  }
+
+  public static final int SUMMARY_FIELD_NUMBER = 7;
+  private volatile java.lang.Object summary_;
+  /**
+   * <pre>
+   * One-line human-readable description of what the Op does.
+   * </pre>
+   *
+   * <code>string summary = 7;</code>
+   */
+  public java.lang.String getSummary() {
+    java.lang.Object ref = summary_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      summary_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * One-line human-readable description of what the Op does.
+   * </pre>
+   *
+   * <code>string summary = 7;</code>
+   */
+  public com.google.protobuf.ByteString
+      getSummaryBytes() {
+    java.lang.Object ref = summary_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      summary_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int DESCRIPTION_FIELD_NUMBER = 8;
+  private volatile java.lang.Object description_;
+  /**
+   * <pre>
+   * Additional, longer human-readable description of what the Op does.
+   * </pre>
+   *
+   * <code>string description = 8;</code>
+   */
+  public java.lang.String getDescription() {
+    java.lang.Object ref = description_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      description_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * Additional, longer human-readable description of what the Op does.
+   * </pre>
+   *
+   * <code>string description = 8;</code>
+   */
+  public com.google.protobuf.ByteString
+      getDescriptionBytes() {
+    java.lang.Object ref = description_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      description_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int DESCRIPTION_PREFIX_FIELD_NUMBER = 9;
+  private volatile java.lang.Object descriptionPrefix_;
+  /**
+   * <pre>
+   * Modify an existing/inherited description by adding text to the beginning
+   * or end.
+   * </pre>
+   *
+   * <code>string description_prefix = 9;</code>
+   */
+  public java.lang.String getDescriptionPrefix() {
+    java.lang.Object ref = descriptionPrefix_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      descriptionPrefix_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * Modify an existing/inherited description by adding text to the beginning
+   * or end.
+   * </pre>
+   *
+   * <code>string description_prefix = 9;</code>
+   */
+  public com.google.protobuf.ByteString
+      getDescriptionPrefixBytes() {
+    java.lang.Object ref = descriptionPrefix_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      descriptionPrefix_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int DESCRIPTION_SUFFIX_FIELD_NUMBER = 10;
+  private volatile java.lang.Object descriptionSuffix_;
+  /**
+   * <code>string description_suffix = 10;</code>
+   */
+  public java.lang.String getDescriptionSuffix() {
+    java.lang.Object ref = descriptionSuffix_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      descriptionSuffix_ = s;
+      return s;
+    }
+  }
+  /**
+   * <code>string description_suffix = 10;</code>
+   */
+  public com.google.protobuf.ByteString
+      getDescriptionSuffixBytes() {
+    java.lang.Object ref = descriptionSuffix_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      descriptionSuffix_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    if (!getGraphOpNameBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 1, graphOpName_);
+    }
+    if (visibility_ != org.tensorflow.proto.framework.ApiDef.Visibility.DEFAULT_VISIBILITY.getNumber()) {
+      output.writeEnum(2, visibility_);
+    }
+    for (int i = 0; i < endpoint_.size(); i++) {
+      output.writeMessage(3, endpoint_.get(i));
+    }
+    for (int i = 0; i < inArg_.size(); i++) {
+      output.writeMessage(4, inArg_.get(i));
+    }
+    for (int i = 0; i < outArg_.size(); i++) {
+      output.writeMessage(5, outArg_.get(i));
+    }
+    for (int i = 0; i < attr_.size(); i++) {
+      output.writeMessage(6, attr_.get(i));
+    }
+    if (!getSummaryBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 7, summary_);
+    }
+    if (!getDescriptionBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 8, description_);
+    }
+    if (!getDescriptionPrefixBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 9, descriptionPrefix_);
+    }
+    if (!getDescriptionSuffixBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 10, descriptionSuffix_);
+    }
+    for (int i = 0; i < argOrder_.size(); i++) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 11, argOrder_.getRaw(i));
+    }
+    if (!getDeprecationMessageBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 12, deprecationMessage_);
+    }
+    if (deprecationVersion_ != 0) {
+      output.writeInt32(13, deprecationVersion_);
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (!getGraphOpNameBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, graphOpName_);
+    }
+    if (visibility_ != org.tensorflow.proto.framework.ApiDef.Visibility.DEFAULT_VISIBILITY.getNumber()) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeEnumSize(2, visibility_);
+    }
+    for (int i = 0; i < endpoint_.size(); i++) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(3, endpoint_.get(i));
+    }
+    for (int i = 0; i < inArg_.size(); i++) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(4, inArg_.get(i));
+    }
+    for (int i = 0; i < outArg_.size(); i++) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(5, outArg_.get(i));
+    }
+    for (int i = 0; i < attr_.size(); i++) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(6, attr_.get(i));
+    }
+    if (!getSummaryBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(7, summary_);
+    }
+    if (!getDescriptionBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(8, description_);
+    }
+    if (!getDescriptionPrefixBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(9, descriptionPrefix_);
+    }
+    if (!getDescriptionSuffixBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(10, descriptionSuffix_);
+    }
+    {
+      int dataSize = 0;
+      for (int i = 0; i < argOrder_.size(); i++) {
+        dataSize += computeStringSizeNoTag(argOrder_.getRaw(i));
+      }
+      size += dataSize;
+      size += 1 * getArgOrderList().size();
+    }
+    if (!getDeprecationMessageBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(12, deprecationMessage_);
+    }
+    if (deprecationVersion_ != 0) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt32Size(13, deprecationVersion_);
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.framework.ApiDef)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.framework.ApiDef other = (org.tensorflow.proto.framework.ApiDef) obj;
+
+    if (!getGraphOpName()
+        .equals(other.getGraphOpName())) return false;
+    if (!getDeprecationMessage()
+        .equals(other.getDeprecationMessage())) return false;
+    if (getDeprecationVersion()
+        != other.getDeprecationVersion()) return false;
+    if (visibility_ != other.visibility_) return false;
+    if (!getEndpointList()
+        .equals(other.getEndpointList())) return false;
+    if (!getInArgList()
+        .equals(other.getInArgList())) return false;
+    if (!getOutArgList()
+        .equals(other.getOutArgList())) return false;
+    if (!getArgOrderList()
+        .equals(other.getArgOrderList())) return false;
+    if (!getAttrList()
+        .equals(other.getAttrList())) return false;
+    if (!getSummary()
+        .equals(other.getSummary())) return false;
+    if (!getDescription()
+        .equals(other.getDescription())) return false;
+    if (!getDescriptionPrefix()
+        .equals(other.getDescriptionPrefix())) return false;
+    if (!getDescriptionSuffix()
+        .equals(other.getDescriptionSuffix())) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    hash = (37 * hash) + GRAPH_OP_NAME_FIELD_NUMBER;
+    hash = (53 * hash) + getGraphOpName().hashCode();
+    hash = (37 * hash) + DEPRECATION_MESSAGE_FIELD_NUMBER;
+    hash = (53 * hash) + getDeprecationMessage().hashCode();
+    hash = (37 * hash) + DEPRECATION_VERSION_FIELD_NUMBER;
+    hash = (53 * hash) + getDeprecationVersion();
+    hash = (37 * hash) + VISIBILITY_FIELD_NUMBER;
+    hash = (53 * hash) + visibility_;
+    if (getEndpointCount() > 0) {
+      hash = (37 * hash) + ENDPOINT_FIELD_NUMBER;
+      hash = (53 * hash) + getEndpointList().hashCode();
+    }
+    if (getInArgCount() > 0) {
+      hash = (37 * hash) + IN_ARG_FIELD_NUMBER;
+      hash = (53 * hash) + getInArgList().hashCode();
+    }
+    if (getOutArgCount() > 0) {
+      hash = (37 * hash) + OUT_ARG_FIELD_NUMBER;
+      hash = (53 * hash) + getOutArgList().hashCode();
+    }
+    if (getArgOrderCount() > 0) {
+      hash = (37 * hash) + ARG_ORDER_FIELD_NUMBER;
+      hash = (53 * hash) + getArgOrderList().hashCode();
+    }
+    if (getAttrCount() > 0) {
+      hash = (37 * hash) + ATTR_FIELD_NUMBER;
+      hash = (53 * hash) + getAttrList().hashCode();
+    }
+    hash = (37 * hash) + SUMMARY_FIELD_NUMBER;
+    hash = (53 * hash) + getSummary().hashCode();
+    hash = (37 * hash) + DESCRIPTION_FIELD_NUMBER;
+    hash = (53 * hash) + getDescription().hashCode();
+    hash = (37 * hash) + DESCRIPTION_PREFIX_FIELD_NUMBER;
+    hash = (53 * hash) + getDescriptionPrefix().hashCode();
+    hash = (37 * hash) + DESCRIPTION_SUFFIX_FIELD_NUMBER;
+    hash = (53 * hash) + getDescriptionSuffix().hashCode();
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.framework.ApiDef parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.ApiDef parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.ApiDef parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.ApiDef parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.ApiDef parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.ApiDef parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.ApiDef parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.ApiDef parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.ApiDef parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.ApiDef parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.ApiDef parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.ApiDef parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.framework.ApiDef prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * <pre>
+   * Used to specify and override the default API &amp; behavior in the
+   * generated code for client languages, from what you would get from
+   * the OpDef alone. There will be a set of ApiDefs that are common
+   * to all client languages, and another set per client language.
+   * The per-client-language ApiDefs will inherit values from the
+   * common ApiDefs which it can either replace or modify.
+   * We separate the API definition from the OpDef so we can evolve the
+   * API while remaining backwards compatible when interpretting old
+   * graphs.  Overrides go in an "api_def.pbtxt" file with a text-format
+   * ApiDefs message.
+   * WARNING: Be *very* careful changing the API for any existing op --
+   * you can change the semantics of existing code.  These changes may
+   * need to wait until a major release of TensorFlow to avoid breaking
+   * our compatibility promises.
+   * </pre>
+   *
+   * Protobuf type {@code tensorflow.ApiDef}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.ApiDef)
+      org.tensorflow.proto.framework.ApiDefOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.ApiDefProtos.internal_static_tensorflow_ApiDef_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.ApiDefProtos.internal_static_tensorflow_ApiDef_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.ApiDef.class, org.tensorflow.proto.framework.ApiDef.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.framework.ApiDef.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+        getEndpointFieldBuilder();
+        getInArgFieldBuilder();
+        getOutArgFieldBuilder();
+        getAttrFieldBuilder();
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      graphOpName_ = "";
+
+      deprecationMessage_ = "";
+
+      deprecationVersion_ = 0;
+
+      visibility_ = 0;
+
+      if (endpointBuilder_ == null) {
+        endpoint_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000001);
+      } else {
+        endpointBuilder_.clear();
+      }
+      if (inArgBuilder_ == null) {
+        inArg_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000002);
+      } else {
+        inArgBuilder_.clear();
+      }
+      if (outArgBuilder_ == null) {
+        outArg_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000004);
+      } else {
+        outArgBuilder_.clear();
+      }
+      argOrder_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+      bitField0_ = (bitField0_ & ~0x00000008);
+      if (attrBuilder_ == null) {
+        attr_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000010);
+      } else {
+        attrBuilder_.clear();
+      }
+      summary_ = "";
+
+      description_ = "";
+
+      descriptionPrefix_ = "";
+
+      descriptionSuffix_ = "";
+
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.framework.ApiDefProtos.internal_static_tensorflow_ApiDef_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.ApiDef getDefaultInstanceForType() {
+      return org.tensorflow.proto.framework.ApiDef.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.ApiDef build() {
+      org.tensorflow.proto.framework.ApiDef result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.ApiDef buildPartial() {
+      org.tensorflow.proto.framework.ApiDef result = new org.tensorflow.proto.framework.ApiDef(this);
+      int from_bitField0_ = bitField0_;
+      result.graphOpName_ = graphOpName_;
+      result.deprecationMessage_ = deprecationMessage_;
+      result.deprecationVersion_ = deprecationVersion_;
+      result.visibility_ = visibility_;
+      if (endpointBuilder_ == null) {
+        if (((bitField0_ & 0x00000001) != 0)) {
+          endpoint_ = java.util.Collections.unmodifiableList(endpoint_);
+          bitField0_ = (bitField0_ & ~0x00000001);
+        }
+        result.endpoint_ = endpoint_;
+      } else {
+        result.endpoint_ = endpointBuilder_.build();
+      }
+      if (inArgBuilder_ == null) {
+        if (((bitField0_ & 0x00000002) != 0)) {
+          inArg_ = java.util.Collections.unmodifiableList(inArg_);
+          bitField0_ = (bitField0_ & ~0x00000002);
+        }
+        result.inArg_ = inArg_;
+      } else {
+        result.inArg_ = inArgBuilder_.build();
+      }
+      if (outArgBuilder_ == null) {
+        if (((bitField0_ & 0x00000004) != 0)) {
+          outArg_ = java.util.Collections.unmodifiableList(outArg_);
+          bitField0_ = (bitField0_ & ~0x00000004);
+        }
+        result.outArg_ = outArg_;
+      } else {
+        result.outArg_ = outArgBuilder_.build();
+      }
+      if (((bitField0_ & 0x00000008) != 0)) {
+        argOrder_ = argOrder_.getUnmodifiableView();
+        bitField0_ = (bitField0_ & ~0x00000008);
+      }
+      result.argOrder_ = argOrder_;
+      if (attrBuilder_ == null) {
+        if (((bitField0_ & 0x00000010) != 0)) {
+          attr_ = java.util.Collections.unmodifiableList(attr_);
+          bitField0_ = (bitField0_ & ~0x00000010);
+        }
+        result.attr_ = attr_;
+      } else {
+        result.attr_ = attrBuilder_.build();
+      }
+      result.summary_ = summary_;
+      result.description_ = description_;
+      result.descriptionPrefix_ = descriptionPrefix_;
+      result.descriptionSuffix_ = descriptionSuffix_;
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.framework.ApiDef) {
+        return mergeFrom((org.tensorflow.proto.framework.ApiDef)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.framework.ApiDef other) {
+      if (other == org.tensorflow.proto.framework.ApiDef.getDefaultInstance()) return this;
+      if (!other.getGraphOpName().isEmpty()) {
+        graphOpName_ = other.graphOpName_;
+        onChanged();
+      }
+      if (!other.getDeprecationMessage().isEmpty()) {
+        deprecationMessage_ = other.deprecationMessage_;
+        onChanged();
+      }
+      if (other.getDeprecationVersion() != 0) {
+        setDeprecationVersion(other.getDeprecationVersion());
+      }
+      if (other.visibility_ != 0) {
+        setVisibilityValue(other.getVisibilityValue());
+      }
+      if (endpointBuilder_ == null) {
+        if (!other.endpoint_.isEmpty()) {
+          if (endpoint_.isEmpty()) {
+            endpoint_ = other.endpoint_;
+            bitField0_ = (bitField0_ & ~0x00000001);
+          } else {
+            ensureEndpointIsMutable();
+            endpoint_.addAll(other.endpoint_);
+          }
+          onChanged();
+        }
+      } else {
+        if (!other.endpoint_.isEmpty()) {
+          if (endpointBuilder_.isEmpty()) {
+            endpointBuilder_.dispose();
+            endpointBuilder_ = null;
+            endpoint_ = other.endpoint_;
+            bitField0_ = (bitField0_ & ~0x00000001);
+            endpointBuilder_ = 
+              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
+                 getEndpointFieldBuilder() : null;
+          } else {
+            endpointBuilder_.addAllMessages(other.endpoint_);
+          }
+        }
+      }
+      if (inArgBuilder_ == null) {
+        if (!other.inArg_.isEmpty()) {
+          if (inArg_.isEmpty()) {
+            inArg_ = other.inArg_;
+            bitField0_ = (bitField0_ & ~0x00000002);
+          } else {
+            ensureInArgIsMutable();
+            inArg_.addAll(other.inArg_);
+          }
+          onChanged();
+        }
+      } else {
+        if (!other.inArg_.isEmpty()) {
+          if (inArgBuilder_.isEmpty()) {
+            inArgBuilder_.dispose();
+            inArgBuilder_ = null;
+            inArg_ = other.inArg_;
+            bitField0_ = (bitField0_ & ~0x00000002);
+            inArgBuilder_ = 
+              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
+                 getInArgFieldBuilder() : null;
+          } else {
+            inArgBuilder_.addAllMessages(other.inArg_);
+          }
+        }
+      }
+      if (outArgBuilder_ == null) {
+        if (!other.outArg_.isEmpty()) {
+          if (outArg_.isEmpty()) {
+            outArg_ = other.outArg_;
+            bitField0_ = (bitField0_ & ~0x00000004);
+          } else {
+            ensureOutArgIsMutable();
+            outArg_.addAll(other.outArg_);
+          }
+          onChanged();
+        }
+      } else {
+        if (!other.outArg_.isEmpty()) {
+          if (outArgBuilder_.isEmpty()) {
+            outArgBuilder_.dispose();
+            outArgBuilder_ = null;
+            outArg_ = other.outArg_;
+            bitField0_ = (bitField0_ & ~0x00000004);
+            outArgBuilder_ = 
+              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
+                 getOutArgFieldBuilder() : null;
+          } else {
+            outArgBuilder_.addAllMessages(other.outArg_);
+          }
+        }
+      }
+      if (!other.argOrder_.isEmpty()) {
+        if (argOrder_.isEmpty()) {
+          argOrder_ = other.argOrder_;
+          bitField0_ = (bitField0_ & ~0x00000008);
+        } else {
+          ensureArgOrderIsMutable();
+          argOrder_.addAll(other.argOrder_);
+        }
+        onChanged();
+      }
+      if (attrBuilder_ == null) {
+        if (!other.attr_.isEmpty()) {
+          if (attr_.isEmpty()) {
+            attr_ = other.attr_;
+            bitField0_ = (bitField0_ & ~0x00000010);
+          } else {
+            ensureAttrIsMutable();
+            attr_.addAll(other.attr_);
+          }
+          onChanged();
+        }
+      } else {
+        if (!other.attr_.isEmpty()) {
+          if (attrBuilder_.isEmpty()) {
+            attrBuilder_.dispose();
+            attrBuilder_ = null;
+            attr_ = other.attr_;
+            bitField0_ = (bitField0_ & ~0x00000010);
+            attrBuilder_ = 
+              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
+                 getAttrFieldBuilder() : null;
+          } else {
+            attrBuilder_.addAllMessages(other.attr_);
+          }
+        }
+      }
+      if (!other.getSummary().isEmpty()) {
+        summary_ = other.summary_;
+        onChanged();
+      }
+      if (!other.getDescription().isEmpty()) {
+        description_ = other.description_;
+        onChanged();
+      }
+      if (!other.getDescriptionPrefix().isEmpty()) {
+        descriptionPrefix_ = other.descriptionPrefix_;
+        onChanged();
+      }
+      if (!other.getDescriptionSuffix().isEmpty()) {
+        descriptionSuffix_ = other.descriptionSuffix_;
+        onChanged();
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.framework.ApiDef parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.framework.ApiDef) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+    private int bitField0_;
+
+    private java.lang.Object graphOpName_ = "";
+    /**
+     * <pre>
+     * Name of the op (in the OpDef) to specify the API for.
+     * </pre>
+     *
+     * <code>string graph_op_name = 1;</code>
+     */
+    public java.lang.String getGraphOpName() {
+      java.lang.Object ref = graphOpName_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        graphOpName_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Name of the op (in the OpDef) to specify the API for.
+     * </pre>
+     *
+     * <code>string graph_op_name = 1;</code>
+     */
+    public com.google.protobuf.ByteString
+        getGraphOpNameBytes() {
+      java.lang.Object ref = graphOpName_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        graphOpName_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Name of the op (in the OpDef) to specify the API for.
+     * </pre>
+     *
+     * <code>string graph_op_name = 1;</code>
+     */
+    public Builder setGraphOpName(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      graphOpName_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Name of the op (in the OpDef) to specify the API for.
+     * </pre>
+     *
+     * <code>string graph_op_name = 1;</code>
+     */
+    public Builder clearGraphOpName() {
+      
+      graphOpName_ = getDefaultInstance().getGraphOpName();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Name of the op (in the OpDef) to specify the API for.
+     * </pre>
+     *
+     * <code>string graph_op_name = 1;</code>
+     */
+    public Builder setGraphOpNameBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      graphOpName_ = value;
+      onChanged();
+      return this;
+    }
+
+    private java.lang.Object deprecationMessage_ = "";
+    /**
+     * <pre>
+     * If this op is deprecated, set deprecation message to the message
+     * that should be logged when this op is used.
+     * The message should indicate alternative op to use, if any.
+     * </pre>
+     *
+     * <code>string deprecation_message = 12;</code>
+     */
+    public java.lang.String getDeprecationMessage() {
+      java.lang.Object ref = deprecationMessage_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        deprecationMessage_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * If this op is deprecated, set deprecation message to the message
+     * that should be logged when this op is used.
+     * The message should indicate alternative op to use, if any.
+     * </pre>
+     *
+     * <code>string deprecation_message = 12;</code>
+     */
+    public com.google.protobuf.ByteString
+        getDeprecationMessageBytes() {
+      java.lang.Object ref = deprecationMessage_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        deprecationMessage_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * If this op is deprecated, set deprecation message to the message
+     * that should be logged when this op is used.
+     * The message should indicate alternative op to use, if any.
+     * </pre>
+     *
+     * <code>string deprecation_message = 12;</code>
+     */
+    public Builder setDeprecationMessage(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      deprecationMessage_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * If this op is deprecated, set deprecation message to the message
+     * that should be logged when this op is used.
+     * The message should indicate alternative op to use, if any.
+     * </pre>
+     *
+     * <code>string deprecation_message = 12;</code>
+     */
+    public Builder clearDeprecationMessage() {
+      
+      deprecationMessage_ = getDefaultInstance().getDeprecationMessage();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * If this op is deprecated, set deprecation message to the message
+     * that should be logged when this op is used.
+     * The message should indicate alternative op to use, if any.
+     * </pre>
+     *
+     * <code>string deprecation_message = 12;</code>
+     */
+    public Builder setDeprecationMessageBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      deprecationMessage_ = value;
+      onChanged();
+      return this;
+    }
+
+    private int deprecationVersion_ ;
+    /**
+     * <pre>
+     * Major version when the op will be deleted. For e.g. set this
+     * value to 2 if op API should be removed in TensorFlow 2.0 and
+     * deprecated in versions before that.
+     * </pre>
+     *
+     * <code>int32 deprecation_version = 13;</code>
+     */
+    public int getDeprecationVersion() {
+      return deprecationVersion_;
+    }
+    /**
+     * <pre>
+     * Major version when the op will be deleted. For e.g. set this
+     * value to 2 if op API should be removed in TensorFlow 2.0 and
+     * deprecated in versions before that.
+     * </pre>
+     *
+     * <code>int32 deprecation_version = 13;</code>
+     */
+    public Builder setDeprecationVersion(int value) {
+      
+      deprecationVersion_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Major version when the op will be deleted. For e.g. set this
+     * value to 2 if op API should be removed in TensorFlow 2.0 and
+     * deprecated in versions before that.
+     * </pre>
+     *
+     * <code>int32 deprecation_version = 13;</code>
+     */
+    public Builder clearDeprecationVersion() {
+      
+      deprecationVersion_ = 0;
+      onChanged();
+      return this;
+    }
+
+    private int visibility_ = 0;
+    /**
+     * <code>.tensorflow.ApiDef.Visibility visibility = 2;</code>
+     */
+    public int getVisibilityValue() {
+      return visibility_;
+    }
+    /**
+     * <code>.tensorflow.ApiDef.Visibility visibility = 2;</code>
+     */
+    public Builder setVisibilityValue(int value) {
+      visibility_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>.tensorflow.ApiDef.Visibility visibility = 2;</code>
+     */
+    public org.tensorflow.proto.framework.ApiDef.Visibility getVisibility() {
+      @SuppressWarnings("deprecation")
+      org.tensorflow.proto.framework.ApiDef.Visibility result = org.tensorflow.proto.framework.ApiDef.Visibility.valueOf(visibility_);
+      return result == null ? org.tensorflow.proto.framework.ApiDef.Visibility.UNRECOGNIZED : result;
+    }
+    /**
+     * <code>.tensorflow.ApiDef.Visibility visibility = 2;</code>
+     */
+    public Builder setVisibility(org.tensorflow.proto.framework.ApiDef.Visibility value) {
+      if (value == null) {
+        throw new NullPointerException();
+      }
+      
+      visibility_ = value.getNumber();
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>.tensorflow.ApiDef.Visibility visibility = 2;</code>
+     */
+    public Builder clearVisibility() {
+      
+      visibility_ = 0;
+      onChanged();
+      return this;
+    }
+
+    private java.util.List<org.tensorflow.proto.framework.ApiDef.Endpoint> endpoint_ =
+      java.util.Collections.emptyList();
+    private void ensureEndpointIsMutable() {
+      if (!((bitField0_ & 0x00000001) != 0)) {
+        endpoint_ = new java.util.ArrayList<org.tensorflow.proto.framework.ApiDef.Endpoint>(endpoint_);
+        bitField0_ |= 0x00000001;
+       }
+    }
+
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.framework.ApiDef.Endpoint, org.tensorflow.proto.framework.ApiDef.Endpoint.Builder, org.tensorflow.proto.framework.ApiDef.EndpointOrBuilder> endpointBuilder_;
+
+    /**
+     * <code>repeated .tensorflow.ApiDef.Endpoint endpoint = 3;</code>
+     */
+    public java.util.List<org.tensorflow.proto.framework.ApiDef.Endpoint> getEndpointList() {
+      if (endpointBuilder_ == null) {
+        return java.util.Collections.unmodifiableList(endpoint_);
+      } else {
+        return endpointBuilder_.getMessageList();
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.ApiDef.Endpoint endpoint = 3;</code>
+     */
+    public int getEndpointCount() {
+      if (endpointBuilder_ == null) {
+        return endpoint_.size();
+      } else {
+        return endpointBuilder_.getCount();
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.ApiDef.Endpoint endpoint = 3;</code>
+     */
+    public org.tensorflow.proto.framework.ApiDef.Endpoint getEndpoint(int index) {
+      if (endpointBuilder_ == null) {
+        return endpoint_.get(index);
+      } else {
+        return endpointBuilder_.getMessage(index);
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.ApiDef.Endpoint endpoint = 3;</code>
+     */
+    public Builder setEndpoint(
+        int index, org.tensorflow.proto.framework.ApiDef.Endpoint value) {
+      if (endpointBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureEndpointIsMutable();
+        endpoint_.set(index, value);
+        onChanged();
+      } else {
+        endpointBuilder_.setMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.ApiDef.Endpoint endpoint = 3;</code>
+     */
+    public Builder setEndpoint(
+        int index, org.tensorflow.proto.framework.ApiDef.Endpoint.Builder builderForValue) {
+      if (endpointBuilder_ == null) {
+        ensureEndpointIsMutable();
+        endpoint_.set(index, builderForValue.build());
+        onChanged();
+      } else {
+        endpointBuilder_.setMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.ApiDef.Endpoint endpoint = 3;</code>
+     */
+    public Builder addEndpoint(org.tensorflow.proto.framework.ApiDef.Endpoint value) {
+      if (endpointBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureEndpointIsMutable();
+        endpoint_.add(value);
+        onChanged();
+      } else {
+        endpointBuilder_.addMessage(value);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.ApiDef.Endpoint endpoint = 3;</code>
+     */
+    public Builder addEndpoint(
+        int index, org.tensorflow.proto.framework.ApiDef.Endpoint value) {
+      if (endpointBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureEndpointIsMutable();
+        endpoint_.add(index, value);
+        onChanged();
+      } else {
+        endpointBuilder_.addMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.ApiDef.Endpoint endpoint = 3;</code>
+     */
+    public Builder addEndpoint(
+        org.tensorflow.proto.framework.ApiDef.Endpoint.Builder builderForValue) {
+      if (endpointBuilder_ == null) {
+        ensureEndpointIsMutable();
+        endpoint_.add(builderForValue.build());
+        onChanged();
+      } else {
+        endpointBuilder_.addMessage(builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.ApiDef.Endpoint endpoint = 3;</code>
+     */
+    public Builder addEndpoint(
+        int index, org.tensorflow.proto.framework.ApiDef.Endpoint.Builder builderForValue) {
+      if (endpointBuilder_ == null) {
+        ensureEndpointIsMutable();
+        endpoint_.add(index, builderForValue.build());
+        onChanged();
+      } else {
+        endpointBuilder_.addMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.ApiDef.Endpoint endpoint = 3;</code>
+     */
+    public Builder addAllEndpoint(
+        java.lang.Iterable<? extends org.tensorflow.proto.framework.ApiDef.Endpoint> values) {
+      if (endpointBuilder_ == null) {
+        ensureEndpointIsMutable();
+        com.google.protobuf.AbstractMessageLite.Builder.addAll(
+            values, endpoint_);
+        onChanged();
+      } else {
+        endpointBuilder_.addAllMessages(values);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.ApiDef.Endpoint endpoint = 3;</code>
+     */
+    public Builder clearEndpoint() {
+      if (endpointBuilder_ == null) {
+        endpoint_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000001);
+        onChanged();
+      } else {
+        endpointBuilder_.clear();
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.ApiDef.Endpoint endpoint = 3;</code>
+     */
+    public Builder removeEndpoint(int index) {
+      if (endpointBuilder_ == null) {
+        ensureEndpointIsMutable();
+        endpoint_.remove(index);
+        onChanged();
+      } else {
+        endpointBuilder_.remove(index);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.ApiDef.Endpoint endpoint = 3;</code>
+     */
+    public org.tensorflow.proto.framework.ApiDef.Endpoint.Builder getEndpointBuilder(
+        int index) {
+      return getEndpointFieldBuilder().getBuilder(index);
+    }
+    /**
+     * <code>repeated .tensorflow.ApiDef.Endpoint endpoint = 3;</code>
+     */
+    public org.tensorflow.proto.framework.ApiDef.EndpointOrBuilder getEndpointOrBuilder(
+        int index) {
+      if (endpointBuilder_ == null) {
+        return endpoint_.get(index);  } else {
+        return endpointBuilder_.getMessageOrBuilder(index);
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.ApiDef.Endpoint endpoint = 3;</code>
+     */
+    public java.util.List<? extends org.tensorflow.proto.framework.ApiDef.EndpointOrBuilder> 
+         getEndpointOrBuilderList() {
+      if (endpointBuilder_ != null) {
+        return endpointBuilder_.getMessageOrBuilderList();
+      } else {
+        return java.util.Collections.unmodifiableList(endpoint_);
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.ApiDef.Endpoint endpoint = 3;</code>
+     */
+    public org.tensorflow.proto.framework.ApiDef.Endpoint.Builder addEndpointBuilder() {
+      return getEndpointFieldBuilder().addBuilder(
+          org.tensorflow.proto.framework.ApiDef.Endpoint.getDefaultInstance());
+    }
+    /**
+     * <code>repeated .tensorflow.ApiDef.Endpoint endpoint = 3;</code>
+     */
+    public org.tensorflow.proto.framework.ApiDef.Endpoint.Builder addEndpointBuilder(
+        int index) {
+      return getEndpointFieldBuilder().addBuilder(
+          index, org.tensorflow.proto.framework.ApiDef.Endpoint.getDefaultInstance());
+    }
+    /**
+     * <code>repeated .tensorflow.ApiDef.Endpoint endpoint = 3;</code>
+     */
+    public java.util.List<org.tensorflow.proto.framework.ApiDef.Endpoint.Builder> 
+         getEndpointBuilderList() {
+      return getEndpointFieldBuilder().getBuilderList();
+    }
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.framework.ApiDef.Endpoint, org.tensorflow.proto.framework.ApiDef.Endpoint.Builder, org.tensorflow.proto.framework.ApiDef.EndpointOrBuilder> 
+        getEndpointFieldBuilder() {
+      if (endpointBuilder_ == null) {
+        endpointBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
+            org.tensorflow.proto.framework.ApiDef.Endpoint, org.tensorflow.proto.framework.ApiDef.Endpoint.Builder, org.tensorflow.proto.framework.ApiDef.EndpointOrBuilder>(
+                endpoint_,
+                ((bitField0_ & 0x00000001) != 0),
+                getParentForChildren(),
+                isClean());
+        endpoint_ = null;
+      }
+      return endpointBuilder_;
+    }
+
+    private java.util.List<org.tensorflow.proto.framework.ApiDef.Arg> inArg_ =
+      java.util.Collections.emptyList();
+    private void ensureInArgIsMutable() {
+      if (!((bitField0_ & 0x00000002) != 0)) {
+        inArg_ = new java.util.ArrayList<org.tensorflow.proto.framework.ApiDef.Arg>(inArg_);
+        bitField0_ |= 0x00000002;
+       }
+    }
+
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.framework.ApiDef.Arg, org.tensorflow.proto.framework.ApiDef.Arg.Builder, org.tensorflow.proto.framework.ApiDef.ArgOrBuilder> inArgBuilder_;
+
+    /**
+     * <code>repeated .tensorflow.ApiDef.Arg in_arg = 4;</code>
+     */
+    public java.util.List<org.tensorflow.proto.framework.ApiDef.Arg> getInArgList() {
+      if (inArgBuilder_ == null) {
+        return java.util.Collections.unmodifiableList(inArg_);
+      } else {
+        return inArgBuilder_.getMessageList();
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.ApiDef.Arg in_arg = 4;</code>
+     */
+    public int getInArgCount() {
+      if (inArgBuilder_ == null) {
+        return inArg_.size();
+      } else {
+        return inArgBuilder_.getCount();
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.ApiDef.Arg in_arg = 4;</code>
+     */
+    public org.tensorflow.proto.framework.ApiDef.Arg getInArg(int index) {
+      if (inArgBuilder_ == null) {
+        return inArg_.get(index);
+      } else {
+        return inArgBuilder_.getMessage(index);
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.ApiDef.Arg in_arg = 4;</code>
+     */
+    public Builder setInArg(
+        int index, org.tensorflow.proto.framework.ApiDef.Arg value) {
+      if (inArgBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureInArgIsMutable();
+        inArg_.set(index, value);
+        onChanged();
+      } else {
+        inArgBuilder_.setMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.ApiDef.Arg in_arg = 4;</code>
+     */
+    public Builder setInArg(
+        int index, org.tensorflow.proto.framework.ApiDef.Arg.Builder builderForValue) {
+      if (inArgBuilder_ == null) {
+        ensureInArgIsMutable();
+        inArg_.set(index, builderForValue.build());
+        onChanged();
+      } else {
+        inArgBuilder_.setMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.ApiDef.Arg in_arg = 4;</code>
+     */
+    public Builder addInArg(org.tensorflow.proto.framework.ApiDef.Arg value) {
+      if (inArgBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureInArgIsMutable();
+        inArg_.add(value);
+        onChanged();
+      } else {
+        inArgBuilder_.addMessage(value);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.ApiDef.Arg in_arg = 4;</code>
+     */
+    public Builder addInArg(
+        int index, org.tensorflow.proto.framework.ApiDef.Arg value) {
+      if (inArgBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureInArgIsMutable();
+        inArg_.add(index, value);
+        onChanged();
+      } else {
+        inArgBuilder_.addMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.ApiDef.Arg in_arg = 4;</code>
+     */
+    public Builder addInArg(
+        org.tensorflow.proto.framework.ApiDef.Arg.Builder builderForValue) {
+      if (inArgBuilder_ == null) {
+        ensureInArgIsMutable();
+        inArg_.add(builderForValue.build());
+        onChanged();
+      } else {
+        inArgBuilder_.addMessage(builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.ApiDef.Arg in_arg = 4;</code>
+     */
+    public Builder addInArg(
+        int index, org.tensorflow.proto.framework.ApiDef.Arg.Builder builderForValue) {
+      if (inArgBuilder_ == null) {
+        ensureInArgIsMutable();
+        inArg_.add(index, builderForValue.build());
+        onChanged();
+      } else {
+        inArgBuilder_.addMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.ApiDef.Arg in_arg = 4;</code>
+     */
+    public Builder addAllInArg(
+        java.lang.Iterable<? extends org.tensorflow.proto.framework.ApiDef.Arg> values) {
+      if (inArgBuilder_ == null) {
+        ensureInArgIsMutable();
+        com.google.protobuf.AbstractMessageLite.Builder.addAll(
+            values, inArg_);
+        onChanged();
+      } else {
+        inArgBuilder_.addAllMessages(values);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.ApiDef.Arg in_arg = 4;</code>
+     */
+    public Builder clearInArg() {
+      if (inArgBuilder_ == null) {
+        inArg_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000002);
+        onChanged();
+      } else {
+        inArgBuilder_.clear();
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.ApiDef.Arg in_arg = 4;</code>
+     */
+    public Builder removeInArg(int index) {
+      if (inArgBuilder_ == null) {
+        ensureInArgIsMutable();
+        inArg_.remove(index);
+        onChanged();
+      } else {
+        inArgBuilder_.remove(index);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.ApiDef.Arg in_arg = 4;</code>
+     */
+    public org.tensorflow.proto.framework.ApiDef.Arg.Builder getInArgBuilder(
+        int index) {
+      return getInArgFieldBuilder().getBuilder(index);
+    }
+    /**
+     * <code>repeated .tensorflow.ApiDef.Arg in_arg = 4;</code>
+     */
+    public org.tensorflow.proto.framework.ApiDef.ArgOrBuilder getInArgOrBuilder(
+        int index) {
+      if (inArgBuilder_ == null) {
+        return inArg_.get(index);  } else {
+        return inArgBuilder_.getMessageOrBuilder(index);
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.ApiDef.Arg in_arg = 4;</code>
+     */
+    public java.util.List<? extends org.tensorflow.proto.framework.ApiDef.ArgOrBuilder> 
+         getInArgOrBuilderList() {
+      if (inArgBuilder_ != null) {
+        return inArgBuilder_.getMessageOrBuilderList();
+      } else {
+        return java.util.Collections.unmodifiableList(inArg_);
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.ApiDef.Arg in_arg = 4;</code>
+     */
+    public org.tensorflow.proto.framework.ApiDef.Arg.Builder addInArgBuilder() {
+      return getInArgFieldBuilder().addBuilder(
+          org.tensorflow.proto.framework.ApiDef.Arg.getDefaultInstance());
+    }
+    /**
+     * <code>repeated .tensorflow.ApiDef.Arg in_arg = 4;</code>
+     */
+    public org.tensorflow.proto.framework.ApiDef.Arg.Builder addInArgBuilder(
+        int index) {
+      return getInArgFieldBuilder().addBuilder(
+          index, org.tensorflow.proto.framework.ApiDef.Arg.getDefaultInstance());
+    }
+    /**
+     * <code>repeated .tensorflow.ApiDef.Arg in_arg = 4;</code>
+     */
+    public java.util.List<org.tensorflow.proto.framework.ApiDef.Arg.Builder> 
+         getInArgBuilderList() {
+      return getInArgFieldBuilder().getBuilderList();
+    }
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.framework.ApiDef.Arg, org.tensorflow.proto.framework.ApiDef.Arg.Builder, org.tensorflow.proto.framework.ApiDef.ArgOrBuilder> 
+        getInArgFieldBuilder() {
+      if (inArgBuilder_ == null) {
+        inArgBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
+            org.tensorflow.proto.framework.ApiDef.Arg, org.tensorflow.proto.framework.ApiDef.Arg.Builder, org.tensorflow.proto.framework.ApiDef.ArgOrBuilder>(
+                inArg_,
+                ((bitField0_ & 0x00000002) != 0),
+                getParentForChildren(),
+                isClean());
+        inArg_ = null;
+      }
+      return inArgBuilder_;
+    }
+
+    private java.util.List<org.tensorflow.proto.framework.ApiDef.Arg> outArg_ =
+      java.util.Collections.emptyList();
+    private void ensureOutArgIsMutable() {
+      if (!((bitField0_ & 0x00000004) != 0)) {
+        outArg_ = new java.util.ArrayList<org.tensorflow.proto.framework.ApiDef.Arg>(outArg_);
+        bitField0_ |= 0x00000004;
+       }
+    }
+
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.framework.ApiDef.Arg, org.tensorflow.proto.framework.ApiDef.Arg.Builder, org.tensorflow.proto.framework.ApiDef.ArgOrBuilder> outArgBuilder_;
+
+    /**
+     * <code>repeated .tensorflow.ApiDef.Arg out_arg = 5;</code>
+     */
+    public java.util.List<org.tensorflow.proto.framework.ApiDef.Arg> getOutArgList() {
+      if (outArgBuilder_ == null) {
+        return java.util.Collections.unmodifiableList(outArg_);
+      } else {
+        return outArgBuilder_.getMessageList();
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.ApiDef.Arg out_arg = 5;</code>
+     */
+    public int getOutArgCount() {
+      if (outArgBuilder_ == null) {
+        return outArg_.size();
+      } else {
+        return outArgBuilder_.getCount();
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.ApiDef.Arg out_arg = 5;</code>
+     */
+    public org.tensorflow.proto.framework.ApiDef.Arg getOutArg(int index) {
+      if (outArgBuilder_ == null) {
+        return outArg_.get(index);
+      } else {
+        return outArgBuilder_.getMessage(index);
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.ApiDef.Arg out_arg = 5;</code>
+     */
+    public Builder setOutArg(
+        int index, org.tensorflow.proto.framework.ApiDef.Arg value) {
+      if (outArgBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureOutArgIsMutable();
+        outArg_.set(index, value);
+        onChanged();
+      } else {
+        outArgBuilder_.setMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.ApiDef.Arg out_arg = 5;</code>
+     */
+    public Builder setOutArg(
+        int index, org.tensorflow.proto.framework.ApiDef.Arg.Builder builderForValue) {
+      if (outArgBuilder_ == null) {
+        ensureOutArgIsMutable();
+        outArg_.set(index, builderForValue.build());
+        onChanged();
+      } else {
+        outArgBuilder_.setMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.ApiDef.Arg out_arg = 5;</code>
+     */
+    public Builder addOutArg(org.tensorflow.proto.framework.ApiDef.Arg value) {
+      if (outArgBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureOutArgIsMutable();
+        outArg_.add(value);
+        onChanged();
+      } else {
+        outArgBuilder_.addMessage(value);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.ApiDef.Arg out_arg = 5;</code>
+     */
+    public Builder addOutArg(
+        int index, org.tensorflow.proto.framework.ApiDef.Arg value) {
+      if (outArgBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureOutArgIsMutable();
+        outArg_.add(index, value);
+        onChanged();
+      } else {
+        outArgBuilder_.addMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.ApiDef.Arg out_arg = 5;</code>
+     */
+    public Builder addOutArg(
+        org.tensorflow.proto.framework.ApiDef.Arg.Builder builderForValue) {
+      if (outArgBuilder_ == null) {
+        ensureOutArgIsMutable();
+        outArg_.add(builderForValue.build());
+        onChanged();
+      } else {
+        outArgBuilder_.addMessage(builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.ApiDef.Arg out_arg = 5;</code>
+     */
+    public Builder addOutArg(
+        int index, org.tensorflow.proto.framework.ApiDef.Arg.Builder builderForValue) {
+      if (outArgBuilder_ == null) {
+        ensureOutArgIsMutable();
+        outArg_.add(index, builderForValue.build());
+        onChanged();
+      } else {
+        outArgBuilder_.addMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.ApiDef.Arg out_arg = 5;</code>
+     */
+    public Builder addAllOutArg(
+        java.lang.Iterable<? extends org.tensorflow.proto.framework.ApiDef.Arg> values) {
+      if (outArgBuilder_ == null) {
+        ensureOutArgIsMutable();
+        com.google.protobuf.AbstractMessageLite.Builder.addAll(
+            values, outArg_);
+        onChanged();
+      } else {
+        outArgBuilder_.addAllMessages(values);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.ApiDef.Arg out_arg = 5;</code>
+     */
+    public Builder clearOutArg() {
+      if (outArgBuilder_ == null) {
+        outArg_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000004);
+        onChanged();
+      } else {
+        outArgBuilder_.clear();
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.ApiDef.Arg out_arg = 5;</code>
+     */
+    public Builder removeOutArg(int index) {
+      if (outArgBuilder_ == null) {
+        ensureOutArgIsMutable();
+        outArg_.remove(index);
+        onChanged();
+      } else {
+        outArgBuilder_.remove(index);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.ApiDef.Arg out_arg = 5;</code>
+     */
+    public org.tensorflow.proto.framework.ApiDef.Arg.Builder getOutArgBuilder(
+        int index) {
+      return getOutArgFieldBuilder().getBuilder(index);
+    }
+    /**
+     * <code>repeated .tensorflow.ApiDef.Arg out_arg = 5;</code>
+     */
+    public org.tensorflow.proto.framework.ApiDef.ArgOrBuilder getOutArgOrBuilder(
+        int index) {
+      if (outArgBuilder_ == null) {
+        return outArg_.get(index);  } else {
+        return outArgBuilder_.getMessageOrBuilder(index);
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.ApiDef.Arg out_arg = 5;</code>
+     */
+    public java.util.List<? extends org.tensorflow.proto.framework.ApiDef.ArgOrBuilder> 
+         getOutArgOrBuilderList() {
+      if (outArgBuilder_ != null) {
+        return outArgBuilder_.getMessageOrBuilderList();
+      } else {
+        return java.util.Collections.unmodifiableList(outArg_);
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.ApiDef.Arg out_arg = 5;</code>
+     */
+    public org.tensorflow.proto.framework.ApiDef.Arg.Builder addOutArgBuilder() {
+      return getOutArgFieldBuilder().addBuilder(
+          org.tensorflow.proto.framework.ApiDef.Arg.getDefaultInstance());
+    }
+    /**
+     * <code>repeated .tensorflow.ApiDef.Arg out_arg = 5;</code>
+     */
+    public org.tensorflow.proto.framework.ApiDef.Arg.Builder addOutArgBuilder(
+        int index) {
+      return getOutArgFieldBuilder().addBuilder(
+          index, org.tensorflow.proto.framework.ApiDef.Arg.getDefaultInstance());
+    }
+    /**
+     * <code>repeated .tensorflow.ApiDef.Arg out_arg = 5;</code>
+     */
+    public java.util.List<org.tensorflow.proto.framework.ApiDef.Arg.Builder> 
+         getOutArgBuilderList() {
+      return getOutArgFieldBuilder().getBuilderList();
+    }
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.framework.ApiDef.Arg, org.tensorflow.proto.framework.ApiDef.Arg.Builder, org.tensorflow.proto.framework.ApiDef.ArgOrBuilder> 
+        getOutArgFieldBuilder() {
+      if (outArgBuilder_ == null) {
+        outArgBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
+            org.tensorflow.proto.framework.ApiDef.Arg, org.tensorflow.proto.framework.ApiDef.Arg.Builder, org.tensorflow.proto.framework.ApiDef.ArgOrBuilder>(
+                outArg_,
+                ((bitField0_ & 0x00000004) != 0),
+                getParentForChildren(),
+                isClean());
+        outArg_ = null;
+      }
+      return outArgBuilder_;
+    }
+
+    private com.google.protobuf.LazyStringList argOrder_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+    private void ensureArgOrderIsMutable() {
+      if (!((bitField0_ & 0x00000008) != 0)) {
+        argOrder_ = new com.google.protobuf.LazyStringArrayList(argOrder_);
+        bitField0_ |= 0x00000008;
+       }
+    }
+    /**
+     * <pre>
+     * List of original in_arg names to specify new argument order.
+     * Length of arg_order should be either empty to keep current order
+     * or match size of in_arg.
+     * </pre>
+     *
+     * <code>repeated string arg_order = 11;</code>
+     */
+    public com.google.protobuf.ProtocolStringList
+        getArgOrderList() {
+      return argOrder_.getUnmodifiableView();
+    }
+    /**
+     * <pre>
+     * List of original in_arg names to specify new argument order.
+     * Length of arg_order should be either empty to keep current order
+     * or match size of in_arg.
+     * </pre>
+     *
+     * <code>repeated string arg_order = 11;</code>
+     */
+    public int getArgOrderCount() {
+      return argOrder_.size();
+    }
+    /**
+     * <pre>
+     * List of original in_arg names to specify new argument order.
+     * Length of arg_order should be either empty to keep current order
+     * or match size of in_arg.
+     * </pre>
+     *
+     * <code>repeated string arg_order = 11;</code>
+     */
+    public java.lang.String getArgOrder(int index) {
+      return argOrder_.get(index);
+    }
+    /**
+     * <pre>
+     * List of original in_arg names to specify new argument order.
+     * Length of arg_order should be either empty to keep current order
+     * or match size of in_arg.
+     * </pre>
+     *
+     * <code>repeated string arg_order = 11;</code>
+     */
+    public com.google.protobuf.ByteString
+        getArgOrderBytes(int index) {
+      return argOrder_.getByteString(index);
+    }
+    /**
+     * <pre>
+     * List of original in_arg names to specify new argument order.
+     * Length of arg_order should be either empty to keep current order
+     * or match size of in_arg.
+     * </pre>
+     *
+     * <code>repeated string arg_order = 11;</code>
+     */
+    public Builder setArgOrder(
+        int index, java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  ensureArgOrderIsMutable();
+      argOrder_.set(index, value);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * List of original in_arg names to specify new argument order.
+     * Length of arg_order should be either empty to keep current order
+     * or match size of in_arg.
+     * </pre>
+     *
+     * <code>repeated string arg_order = 11;</code>
+     */
+    public Builder addArgOrder(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  ensureArgOrderIsMutable();
+      argOrder_.add(value);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * List of original in_arg names to specify new argument order.
+     * Length of arg_order should be either empty to keep current order
+     * or match size of in_arg.
+     * </pre>
+     *
+     * <code>repeated string arg_order = 11;</code>
+     */
+    public Builder addAllArgOrder(
+        java.lang.Iterable<java.lang.String> values) {
+      ensureArgOrderIsMutable();
+      com.google.protobuf.AbstractMessageLite.Builder.addAll(
+          values, argOrder_);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * List of original in_arg names to specify new argument order.
+     * Length of arg_order should be either empty to keep current order
+     * or match size of in_arg.
+     * </pre>
+     *
+     * <code>repeated string arg_order = 11;</code>
+     */
+    public Builder clearArgOrder() {
+      argOrder_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+      bitField0_ = (bitField0_ & ~0x00000008);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * List of original in_arg names to specify new argument order.
+     * Length of arg_order should be either empty to keep current order
+     * or match size of in_arg.
+     * </pre>
+     *
+     * <code>repeated string arg_order = 11;</code>
+     */
+    public Builder addArgOrderBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      ensureArgOrderIsMutable();
+      argOrder_.add(value);
+      onChanged();
+      return this;
+    }
+
+    private java.util.List<org.tensorflow.proto.framework.ApiDef.Attr> attr_ =
+      java.util.Collections.emptyList();
+    private void ensureAttrIsMutable() {
+      if (!((bitField0_ & 0x00000010) != 0)) {
+        attr_ = new java.util.ArrayList<org.tensorflow.proto.framework.ApiDef.Attr>(attr_);
+        bitField0_ |= 0x00000010;
+       }
+    }
+
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.framework.ApiDef.Attr, org.tensorflow.proto.framework.ApiDef.Attr.Builder, org.tensorflow.proto.framework.ApiDef.AttrOrBuilder> attrBuilder_;
+
+    /**
+     * <code>repeated .tensorflow.ApiDef.Attr attr = 6;</code>
+     */
+    public java.util.List<org.tensorflow.proto.framework.ApiDef.Attr> getAttrList() {
+      if (attrBuilder_ == null) {
+        return java.util.Collections.unmodifiableList(attr_);
+      } else {
+        return attrBuilder_.getMessageList();
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.ApiDef.Attr attr = 6;</code>
+     */
+    public int getAttrCount() {
+      if (attrBuilder_ == null) {
+        return attr_.size();
+      } else {
+        return attrBuilder_.getCount();
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.ApiDef.Attr attr = 6;</code>
+     */
+    public org.tensorflow.proto.framework.ApiDef.Attr getAttr(int index) {
+      if (attrBuilder_ == null) {
+        return attr_.get(index);
+      } else {
+        return attrBuilder_.getMessage(index);
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.ApiDef.Attr attr = 6;</code>
+     */
+    public Builder setAttr(
+        int index, org.tensorflow.proto.framework.ApiDef.Attr value) {
+      if (attrBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureAttrIsMutable();
+        attr_.set(index, value);
+        onChanged();
+      } else {
+        attrBuilder_.setMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.ApiDef.Attr attr = 6;</code>
+     */
+    public Builder setAttr(
+        int index, org.tensorflow.proto.framework.ApiDef.Attr.Builder builderForValue) {
+      if (attrBuilder_ == null) {
+        ensureAttrIsMutable();
+        attr_.set(index, builderForValue.build());
+        onChanged();
+      } else {
+        attrBuilder_.setMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.ApiDef.Attr attr = 6;</code>
+     */
+    public Builder addAttr(org.tensorflow.proto.framework.ApiDef.Attr value) {
+      if (attrBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureAttrIsMutable();
+        attr_.add(value);
+        onChanged();
+      } else {
+        attrBuilder_.addMessage(value);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.ApiDef.Attr attr = 6;</code>
+     */
+    public Builder addAttr(
+        int index, org.tensorflow.proto.framework.ApiDef.Attr value) {
+      if (attrBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureAttrIsMutable();
+        attr_.add(index, value);
+        onChanged();
+      } else {
+        attrBuilder_.addMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.ApiDef.Attr attr = 6;</code>
+     */
+    public Builder addAttr(
+        org.tensorflow.proto.framework.ApiDef.Attr.Builder builderForValue) {
+      if (attrBuilder_ == null) {
+        ensureAttrIsMutable();
+        attr_.add(builderForValue.build());
+        onChanged();
+      } else {
+        attrBuilder_.addMessage(builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.ApiDef.Attr attr = 6;</code>
+     */
+    public Builder addAttr(
+        int index, org.tensorflow.proto.framework.ApiDef.Attr.Builder builderForValue) {
+      if (attrBuilder_ == null) {
+        ensureAttrIsMutable();
+        attr_.add(index, builderForValue.build());
+        onChanged();
+      } else {
+        attrBuilder_.addMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.ApiDef.Attr attr = 6;</code>
+     */
+    public Builder addAllAttr(
+        java.lang.Iterable<? extends org.tensorflow.proto.framework.ApiDef.Attr> values) {
+      if (attrBuilder_ == null) {
+        ensureAttrIsMutable();
+        com.google.protobuf.AbstractMessageLite.Builder.addAll(
+            values, attr_);
+        onChanged();
+      } else {
+        attrBuilder_.addAllMessages(values);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.ApiDef.Attr attr = 6;</code>
+     */
+    public Builder clearAttr() {
+      if (attrBuilder_ == null) {
+        attr_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000010);
+        onChanged();
+      } else {
+        attrBuilder_.clear();
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.ApiDef.Attr attr = 6;</code>
+     */
+    public Builder removeAttr(int index) {
+      if (attrBuilder_ == null) {
+        ensureAttrIsMutable();
+        attr_.remove(index);
+        onChanged();
+      } else {
+        attrBuilder_.remove(index);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.ApiDef.Attr attr = 6;</code>
+     */
+    public org.tensorflow.proto.framework.ApiDef.Attr.Builder getAttrBuilder(
+        int index) {
+      return getAttrFieldBuilder().getBuilder(index);
+    }
+    /**
+     * <code>repeated .tensorflow.ApiDef.Attr attr = 6;</code>
+     */
+    public org.tensorflow.proto.framework.ApiDef.AttrOrBuilder getAttrOrBuilder(
+        int index) {
+      if (attrBuilder_ == null) {
+        return attr_.get(index);  } else {
+        return attrBuilder_.getMessageOrBuilder(index);
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.ApiDef.Attr attr = 6;</code>
+     */
+    public java.util.List<? extends org.tensorflow.proto.framework.ApiDef.AttrOrBuilder> 
+         getAttrOrBuilderList() {
+      if (attrBuilder_ != null) {
+        return attrBuilder_.getMessageOrBuilderList();
+      } else {
+        return java.util.Collections.unmodifiableList(attr_);
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.ApiDef.Attr attr = 6;</code>
+     */
+    public org.tensorflow.proto.framework.ApiDef.Attr.Builder addAttrBuilder() {
+      return getAttrFieldBuilder().addBuilder(
+          org.tensorflow.proto.framework.ApiDef.Attr.getDefaultInstance());
+    }
+    /**
+     * <code>repeated .tensorflow.ApiDef.Attr attr = 6;</code>
+     */
+    public org.tensorflow.proto.framework.ApiDef.Attr.Builder addAttrBuilder(
+        int index) {
+      return getAttrFieldBuilder().addBuilder(
+          index, org.tensorflow.proto.framework.ApiDef.Attr.getDefaultInstance());
+    }
+    /**
+     * <code>repeated .tensorflow.ApiDef.Attr attr = 6;</code>
+     */
+    public java.util.List<org.tensorflow.proto.framework.ApiDef.Attr.Builder> 
+         getAttrBuilderList() {
+      return getAttrFieldBuilder().getBuilderList();
+    }
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.framework.ApiDef.Attr, org.tensorflow.proto.framework.ApiDef.Attr.Builder, org.tensorflow.proto.framework.ApiDef.AttrOrBuilder> 
+        getAttrFieldBuilder() {
+      if (attrBuilder_ == null) {
+        attrBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
+            org.tensorflow.proto.framework.ApiDef.Attr, org.tensorflow.proto.framework.ApiDef.Attr.Builder, org.tensorflow.proto.framework.ApiDef.AttrOrBuilder>(
+                attr_,
+                ((bitField0_ & 0x00000010) != 0),
+                getParentForChildren(),
+                isClean());
+        attr_ = null;
+      }
+      return attrBuilder_;
+    }
+
+    private java.lang.Object summary_ = "";
+    /**
+     * <pre>
+     * One-line human-readable description of what the Op does.
+     * </pre>
+     *
+     * <code>string summary = 7;</code>
+     */
+    public java.lang.String getSummary() {
+      java.lang.Object ref = summary_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        summary_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * One-line human-readable description of what the Op does.
+     * </pre>
+     *
+     * <code>string summary = 7;</code>
+     */
+    public com.google.protobuf.ByteString
+        getSummaryBytes() {
+      java.lang.Object ref = summary_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        summary_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * One-line human-readable description of what the Op does.
+     * </pre>
+     *
+     * <code>string summary = 7;</code>
+     */
+    public Builder setSummary(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      summary_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * One-line human-readable description of what the Op does.
+     * </pre>
+     *
+     * <code>string summary = 7;</code>
+     */
+    public Builder clearSummary() {
+      
+      summary_ = getDefaultInstance().getSummary();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * One-line human-readable description of what the Op does.
+     * </pre>
+     *
+     * <code>string summary = 7;</code>
+     */
+    public Builder setSummaryBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      summary_ = value;
+      onChanged();
+      return this;
+    }
+
+    private java.lang.Object description_ = "";
+    /**
+     * <pre>
+     * Additional, longer human-readable description of what the Op does.
+     * </pre>
+     *
+     * <code>string description = 8;</code>
+     */
+    public java.lang.String getDescription() {
+      java.lang.Object ref = description_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        description_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Additional, longer human-readable description of what the Op does.
+     * </pre>
+     *
+     * <code>string description = 8;</code>
+     */
+    public com.google.protobuf.ByteString
+        getDescriptionBytes() {
+      java.lang.Object ref = description_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        description_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Additional, longer human-readable description of what the Op does.
+     * </pre>
+     *
+     * <code>string description = 8;</code>
+     */
+    public Builder setDescription(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      description_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Additional, longer human-readable description of what the Op does.
+     * </pre>
+     *
+     * <code>string description = 8;</code>
+     */
+    public Builder clearDescription() {
+      
+      description_ = getDefaultInstance().getDescription();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Additional, longer human-readable description of what the Op does.
+     * </pre>
+     *
+     * <code>string description = 8;</code>
+     */
+    public Builder setDescriptionBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      description_ = value;
+      onChanged();
+      return this;
+    }
+
+    private java.lang.Object descriptionPrefix_ = "";
+    /**
+     * <pre>
+     * Modify an existing/inherited description by adding text to the beginning
+     * or end.
+     * </pre>
+     *
+     * <code>string description_prefix = 9;</code>
+     */
+    public java.lang.String getDescriptionPrefix() {
+      java.lang.Object ref = descriptionPrefix_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        descriptionPrefix_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Modify an existing/inherited description by adding text to the beginning
+     * or end.
+     * </pre>
+     *
+     * <code>string description_prefix = 9;</code>
+     */
+    public com.google.protobuf.ByteString
+        getDescriptionPrefixBytes() {
+      java.lang.Object ref = descriptionPrefix_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        descriptionPrefix_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Modify an existing/inherited description by adding text to the beginning
+     * or end.
+     * </pre>
+     *
+     * <code>string description_prefix = 9;</code>
+     */
+    public Builder setDescriptionPrefix(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      descriptionPrefix_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Modify an existing/inherited description by adding text to the beginning
+     * or end.
+     * </pre>
+     *
+     * <code>string description_prefix = 9;</code>
+     */
+    public Builder clearDescriptionPrefix() {
+      
+      descriptionPrefix_ = getDefaultInstance().getDescriptionPrefix();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Modify an existing/inherited description by adding text to the beginning
+     * or end.
+     * </pre>
+     *
+     * <code>string description_prefix = 9;</code>
+     */
+    public Builder setDescriptionPrefixBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      descriptionPrefix_ = value;
+      onChanged();
+      return this;
+    }
+
+    private java.lang.Object descriptionSuffix_ = "";
+    /**
+     * <code>string description_suffix = 10;</code>
+     */
+    public java.lang.String getDescriptionSuffix() {
+      java.lang.Object ref = descriptionSuffix_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        descriptionSuffix_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <code>string description_suffix = 10;</code>
+     */
+    public com.google.protobuf.ByteString
+        getDescriptionSuffixBytes() {
+      java.lang.Object ref = descriptionSuffix_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        descriptionSuffix_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <code>string description_suffix = 10;</code>
+     */
+    public Builder setDescriptionSuffix(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      descriptionSuffix_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>string description_suffix = 10;</code>
+     */
+    public Builder clearDescriptionSuffix() {
+      
+      descriptionSuffix_ = getDefaultInstance().getDescriptionSuffix();
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>string description_suffix = 10;</code>
+     */
+    public Builder setDescriptionSuffixBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      descriptionSuffix_ = value;
+      onChanged();
+      return this;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.ApiDef)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.ApiDef)
+  private static final org.tensorflow.proto.framework.ApiDef DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.ApiDef();
+  }
+
+  public static org.tensorflow.proto.framework.ApiDef getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<ApiDef>
+      PARSER = new com.google.protobuf.AbstractParser<ApiDef>() {
+    @java.lang.Override
+    public ApiDef parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new ApiDef(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<ApiDef> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<ApiDef> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.framework.ApiDef getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ApiDefOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ApiDefOrBuilder.java
new file mode 100644
index 00000000000..1a435d93fdb
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ApiDefOrBuilder.java
@@ -0,0 +1,274 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/framework/api_def.proto
+
+package org.tensorflow.proto.framework;
+
+public interface ApiDefOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.ApiDef)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <pre>
+   * Name of the op (in the OpDef) to specify the API for.
+   * </pre>
+   *
+   * <code>string graph_op_name = 1;</code>
+   */
+  java.lang.String getGraphOpName();
+  /**
+   * <pre>
+   * Name of the op (in the OpDef) to specify the API for.
+   * </pre>
+   *
+   * <code>string graph_op_name = 1;</code>
+   */
+  com.google.protobuf.ByteString
+      getGraphOpNameBytes();
+
+  /**
+   * <pre>
+   * If this op is deprecated, set deprecation message to the message
+   * that should be logged when this op is used.
+   * The message should indicate alternative op to use, if any.
+   * </pre>
+   *
+   * <code>string deprecation_message = 12;</code>
+   */
+  java.lang.String getDeprecationMessage();
+  /**
+   * <pre>
+   * If this op is deprecated, set deprecation message to the message
+   * that should be logged when this op is used.
+   * The message should indicate alternative op to use, if any.
+   * </pre>
+   *
+   * <code>string deprecation_message = 12;</code>
+   */
+  com.google.protobuf.ByteString
+      getDeprecationMessageBytes();
+
+  /**
+   * <pre>
+   * Major version when the op will be deleted. For e.g. set this
+   * value to 2 if op API should be removed in TensorFlow 2.0 and
+   * deprecated in versions before that.
+   * </pre>
+   *
+   * <code>int32 deprecation_version = 13;</code>
+   */
+  int getDeprecationVersion();
+
+  /**
+   * <code>.tensorflow.ApiDef.Visibility visibility = 2;</code>
+   */
+  int getVisibilityValue();
+  /**
+   * <code>.tensorflow.ApiDef.Visibility visibility = 2;</code>
+   */
+  org.tensorflow.proto.framework.ApiDef.Visibility getVisibility();
+
+  /**
+   * <code>repeated .tensorflow.ApiDef.Endpoint endpoint = 3;</code>
+   */
+  java.util.List<org.tensorflow.proto.framework.ApiDef.Endpoint> 
+      getEndpointList();
+  /**
+   * <code>repeated .tensorflow.ApiDef.Endpoint endpoint = 3;</code>
+   */
+  org.tensorflow.proto.framework.ApiDef.Endpoint getEndpoint(int index);
+  /**
+   * <code>repeated .tensorflow.ApiDef.Endpoint endpoint = 3;</code>
+   */
+  int getEndpointCount();
+  /**
+   * <code>repeated .tensorflow.ApiDef.Endpoint endpoint = 3;</code>
+   */
+  java.util.List<? extends org.tensorflow.proto.framework.ApiDef.EndpointOrBuilder> 
+      getEndpointOrBuilderList();
+  /**
+   * <code>repeated .tensorflow.ApiDef.Endpoint endpoint = 3;</code>
+   */
+  org.tensorflow.proto.framework.ApiDef.EndpointOrBuilder getEndpointOrBuilder(
+      int index);
+
+  /**
+   * <code>repeated .tensorflow.ApiDef.Arg in_arg = 4;</code>
+   */
+  java.util.List<org.tensorflow.proto.framework.ApiDef.Arg> 
+      getInArgList();
+  /**
+   * <code>repeated .tensorflow.ApiDef.Arg in_arg = 4;</code>
+   */
+  org.tensorflow.proto.framework.ApiDef.Arg getInArg(int index);
+  /**
+   * <code>repeated .tensorflow.ApiDef.Arg in_arg = 4;</code>
+   */
+  int getInArgCount();
+  /**
+   * <code>repeated .tensorflow.ApiDef.Arg in_arg = 4;</code>
+   */
+  java.util.List<? extends org.tensorflow.proto.framework.ApiDef.ArgOrBuilder> 
+      getInArgOrBuilderList();
+  /**
+   * <code>repeated .tensorflow.ApiDef.Arg in_arg = 4;</code>
+   */
+  org.tensorflow.proto.framework.ApiDef.ArgOrBuilder getInArgOrBuilder(
+      int index);
+
+  /**
+   * <code>repeated .tensorflow.ApiDef.Arg out_arg = 5;</code>
+   */
+  java.util.List<org.tensorflow.proto.framework.ApiDef.Arg> 
+      getOutArgList();
+  /**
+   * <code>repeated .tensorflow.ApiDef.Arg out_arg = 5;</code>
+   */
+  org.tensorflow.proto.framework.ApiDef.Arg getOutArg(int index);
+  /**
+   * <code>repeated .tensorflow.ApiDef.Arg out_arg = 5;</code>
+   */
+  int getOutArgCount();
+  /**
+   * <code>repeated .tensorflow.ApiDef.Arg out_arg = 5;</code>
+   */
+  java.util.List<? extends org.tensorflow.proto.framework.ApiDef.ArgOrBuilder> 
+      getOutArgOrBuilderList();
+  /**
+   * <code>repeated .tensorflow.ApiDef.Arg out_arg = 5;</code>
+   */
+  org.tensorflow.proto.framework.ApiDef.ArgOrBuilder getOutArgOrBuilder(
+      int index);
+
+  /**
+   * <pre>
+   * List of original in_arg names to specify new argument order.
+   * Length of arg_order should be either empty to keep current order
+   * or match size of in_arg.
+   * </pre>
+   *
+   * <code>repeated string arg_order = 11;</code>
+   */
+  java.util.List<java.lang.String>
+      getArgOrderList();
+  /**
+   * <pre>
+   * List of original in_arg names to specify new argument order.
+   * Length of arg_order should be either empty to keep current order
+   * or match size of in_arg.
+   * </pre>
+   *
+   * <code>repeated string arg_order = 11;</code>
+   */
+  int getArgOrderCount();
+  /**
+   * <pre>
+   * List of original in_arg names to specify new argument order.
+   * Length of arg_order should be either empty to keep current order
+   * or match size of in_arg.
+   * </pre>
+   *
+   * <code>repeated string arg_order = 11;</code>
+   */
+  java.lang.String getArgOrder(int index);
+  /**
+   * <pre>
+   * List of original in_arg names to specify new argument order.
+   * Length of arg_order should be either empty to keep current order
+   * or match size of in_arg.
+   * </pre>
+   *
+   * <code>repeated string arg_order = 11;</code>
+   */
+  com.google.protobuf.ByteString
+      getArgOrderBytes(int index);
+
+  /**
+   * <code>repeated .tensorflow.ApiDef.Attr attr = 6;</code>
+   */
+  java.util.List<org.tensorflow.proto.framework.ApiDef.Attr> 
+      getAttrList();
+  /**
+   * <code>repeated .tensorflow.ApiDef.Attr attr = 6;</code>
+   */
+  org.tensorflow.proto.framework.ApiDef.Attr getAttr(int index);
+  /**
+   * <code>repeated .tensorflow.ApiDef.Attr attr = 6;</code>
+   */
+  int getAttrCount();
+  /**
+   * <code>repeated .tensorflow.ApiDef.Attr attr = 6;</code>
+   */
+  java.util.List<? extends org.tensorflow.proto.framework.ApiDef.AttrOrBuilder> 
+      getAttrOrBuilderList();
+  /**
+   * <code>repeated .tensorflow.ApiDef.Attr attr = 6;</code>
+   */
+  org.tensorflow.proto.framework.ApiDef.AttrOrBuilder getAttrOrBuilder(
+      int index);
+
+  /**
+   * <pre>
+   * One-line human-readable description of what the Op does.
+   * </pre>
+   *
+   * <code>string summary = 7;</code>
+   */
+  java.lang.String getSummary();
+  /**
+   * <pre>
+   * One-line human-readable description of what the Op does.
+   * </pre>
+   *
+   * <code>string summary = 7;</code>
+   */
+  com.google.protobuf.ByteString
+      getSummaryBytes();
+
+  /**
+   * <pre>
+   * Additional, longer human-readable description of what the Op does.
+   * </pre>
+   *
+   * <code>string description = 8;</code>
+   */
+  java.lang.String getDescription();
+  /**
+   * <pre>
+   * Additional, longer human-readable description of what the Op does.
+   * </pre>
+   *
+   * <code>string description = 8;</code>
+   */
+  com.google.protobuf.ByteString
+      getDescriptionBytes();
+
+  /**
+   * <pre>
+   * Modify an existing/inherited description by adding text to the beginning
+   * or end.
+   * </pre>
+   *
+   * <code>string description_prefix = 9;</code>
+   */
+  java.lang.String getDescriptionPrefix();
+  /**
+   * <pre>
+   * Modify an existing/inherited description by adding text to the beginning
+   * or end.
+   * </pre>
+   *
+   * <code>string description_prefix = 9;</code>
+   */
+  com.google.protobuf.ByteString
+      getDescriptionPrefixBytes();
+
+  /**
+   * <code>string description_suffix = 10;</code>
+   */
+  java.lang.String getDescriptionSuffix();
+  /**
+   * <code>string description_suffix = 10;</code>
+   */
+  com.google.protobuf.ByteString
+      getDescriptionSuffixBytes();
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ApiDefProtos.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ApiDefProtos.java
new file mode 100644
index 00000000000..6df5379d411
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ApiDefProtos.java
@@ -0,0 +1,117 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/framework/api_def.proto
+
+package org.tensorflow.proto.framework;
+
+public final class ApiDefProtos {
+  private ApiDefProtos() {}
+  public static void registerAllExtensions(
+      com.google.protobuf.ExtensionRegistryLite registry) {
+  }
+
+  public static void registerAllExtensions(
+      com.google.protobuf.ExtensionRegistry registry) {
+    registerAllExtensions(
+        (com.google.protobuf.ExtensionRegistryLite) registry);
+  }
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_ApiDef_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_ApiDef_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_ApiDef_Endpoint_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_ApiDef_Endpoint_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_ApiDef_Arg_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_ApiDef_Arg_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_ApiDef_Attr_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_ApiDef_Attr_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_ApiDefs_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_ApiDefs_fieldAccessorTable;
+
+  public static com.google.protobuf.Descriptors.FileDescriptor
+      getDescriptor() {
+    return descriptor;
+  }
+  private static  com.google.protobuf.Descriptors.FileDescriptor
+      descriptor;
+  static {
+    java.lang.String[] descriptorData = {
+      "\n\'tensorflow/core/framework/api_def.prot" +
+      "o\022\ntensorflow\032*tensorflow/core/framework" +
+      "/attr_value.proto\"\341\005\n\006ApiDef\022\025\n\rgraph_op" +
+      "_name\030\001 \001(\t\022\033\n\023deprecation_message\030\014 \001(\t" +
+      "\022\033\n\023deprecation_version\030\r \001(\005\0221\n\nvisibil" +
+      "ity\030\002 \001(\0162\035.tensorflow.ApiDef.Visibility" +
+      "\022-\n\010endpoint\030\003 \003(\0132\033.tensorflow.ApiDef.E" +
+      "ndpoint\022&\n\006in_arg\030\004 \003(\0132\026.tensorflow.Api" +
+      "Def.Arg\022\'\n\007out_arg\030\005 \003(\0132\026.tensorflow.Ap" +
+      "iDef.Arg\022\021\n\targ_order\030\013 \003(\t\022%\n\004attr\030\006 \003(" +
+      "\0132\027.tensorflow.ApiDef.Attr\022\017\n\007summary\030\007 " +
+      "\001(\t\022\023\n\013description\030\010 \001(\t\022\032\n\022description_" +
+      "prefix\030\t \001(\t\022\032\n\022description_suffix\030\n \001(\t" +
+      "\032I\n\010Endpoint\022\014\n\004name\030\001 \001(\t\022\022\n\ndeprecated" +
+      "\030\003 \001(\010\022\033\n\023deprecation_version\030\004 \001(\005\032;\n\003A" +
+      "rg\022\014\n\004name\030\001 \001(\t\022\021\n\trename_to\030\002 \001(\t\022\023\n\013d" +
+      "escription\030\003 \001(\t\032j\n\004Attr\022\014\n\004name\030\001 \001(\t\022\021" +
+      "\n\trename_to\030\002 \001(\t\022,\n\rdefault_value\030\003 \001(\013" +
+      "2\025.tensorflow.AttrValue\022\023\n\013description\030\004" +
+      " \001(\t\"G\n\nVisibility\022\026\n\022DEFAULT_VISIBILITY" +
+      "\020\000\022\013\n\007VISIBLE\020\001\022\010\n\004SKIP\020\002\022\n\n\006HIDDEN\020\003\")\n" +
+      "\007ApiDefs\022\036\n\002op\030\001 \003(\0132\022.tensorflow.ApiDef" +
+      "B\203\001\n\036org.tensorflow.proto.frameworkB\014Api" +
+      "DefProtosP\001ZNgithub.com/tensorflow/tenso" +
+      "rflow/tensorflow/go/core/framework/api_d" +
+      "ef_go_proto\370\001\001b\006proto3"
+    };
+    descriptor = com.google.protobuf.Descriptors.FileDescriptor
+      .internalBuildGeneratedFileFrom(descriptorData,
+        new com.google.protobuf.Descriptors.FileDescriptor[] {
+          org.tensorflow.proto.framework.AttrValueProtos.getDescriptor(),
+        });
+    internal_static_tensorflow_ApiDef_descriptor =
+      getDescriptor().getMessageTypes().get(0);
+    internal_static_tensorflow_ApiDef_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_ApiDef_descriptor,
+        new java.lang.String[] { "GraphOpName", "DeprecationMessage", "DeprecationVersion", "Visibility", "Endpoint", "InArg", "OutArg", "ArgOrder", "Attr", "Summary", "Description", "DescriptionPrefix", "DescriptionSuffix", });
+    internal_static_tensorflow_ApiDef_Endpoint_descriptor =
+      internal_static_tensorflow_ApiDef_descriptor.getNestedTypes().get(0);
+    internal_static_tensorflow_ApiDef_Endpoint_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_ApiDef_Endpoint_descriptor,
+        new java.lang.String[] { "Name", "Deprecated", "DeprecationVersion", });
+    internal_static_tensorflow_ApiDef_Arg_descriptor =
+      internal_static_tensorflow_ApiDef_descriptor.getNestedTypes().get(1);
+    internal_static_tensorflow_ApiDef_Arg_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_ApiDef_Arg_descriptor,
+        new java.lang.String[] { "Name", "RenameTo", "Description", });
+    internal_static_tensorflow_ApiDef_Attr_descriptor =
+      internal_static_tensorflow_ApiDef_descriptor.getNestedTypes().get(2);
+    internal_static_tensorflow_ApiDef_Attr_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_ApiDef_Attr_descriptor,
+        new java.lang.String[] { "Name", "RenameTo", "DefaultValue", "Description", });
+    internal_static_tensorflow_ApiDefs_descriptor =
+      getDescriptor().getMessageTypes().get(1);
+    internal_static_tensorflow_ApiDefs_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_ApiDefs_descriptor,
+        new java.lang.String[] { "Op", });
+    org.tensorflow.proto.framework.AttrValueProtos.getDescriptor();
+  }
+
+  // @@protoc_insertion_point(outer_class_scope)
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ApiDefs.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ApiDefs.java
new file mode 100644
index 00000000000..a4076265967
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ApiDefs.java
@@ -0,0 +1,765 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/framework/api_def.proto
+
+package org.tensorflow.proto.framework;
+
+/**
+ * Protobuf type {@code tensorflow.ApiDefs}
+ */
+public  final class ApiDefs extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.ApiDefs)
+    ApiDefsOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use ApiDefs.newBuilder() to construct.
+  private ApiDefs(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private ApiDefs() {
+    op_ = java.util.Collections.emptyList();
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new ApiDefs();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private ApiDefs(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    int mutable_bitField0_ = 0;
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 10: {
+            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
+              op_ = new java.util.ArrayList<org.tensorflow.proto.framework.ApiDef>();
+              mutable_bitField0_ |= 0x00000001;
+            }
+            op_.add(
+                input.readMessage(org.tensorflow.proto.framework.ApiDef.parser(), extensionRegistry));
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      if (((mutable_bitField0_ & 0x00000001) != 0)) {
+        op_ = java.util.Collections.unmodifiableList(op_);
+      }
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.framework.ApiDefProtos.internal_static_tensorflow_ApiDefs_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.framework.ApiDefProtos.internal_static_tensorflow_ApiDefs_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.framework.ApiDefs.class, org.tensorflow.proto.framework.ApiDefs.Builder.class);
+  }
+
+  public static final int OP_FIELD_NUMBER = 1;
+  private java.util.List<org.tensorflow.proto.framework.ApiDef> op_;
+  /**
+   * <code>repeated .tensorflow.ApiDef op = 1;</code>
+   */
+  public java.util.List<org.tensorflow.proto.framework.ApiDef> getOpList() {
+    return op_;
+  }
+  /**
+   * <code>repeated .tensorflow.ApiDef op = 1;</code>
+   */
+  public java.util.List<? extends org.tensorflow.proto.framework.ApiDefOrBuilder> 
+      getOpOrBuilderList() {
+    return op_;
+  }
+  /**
+   * <code>repeated .tensorflow.ApiDef op = 1;</code>
+   */
+  public int getOpCount() {
+    return op_.size();
+  }
+  /**
+   * <code>repeated .tensorflow.ApiDef op = 1;</code>
+   */
+  public org.tensorflow.proto.framework.ApiDef getOp(int index) {
+    return op_.get(index);
+  }
+  /**
+   * <code>repeated .tensorflow.ApiDef op = 1;</code>
+   */
+  public org.tensorflow.proto.framework.ApiDefOrBuilder getOpOrBuilder(
+      int index) {
+    return op_.get(index);
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    for (int i = 0; i < op_.size(); i++) {
+      output.writeMessage(1, op_.get(i));
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    for (int i = 0; i < op_.size(); i++) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(1, op_.get(i));
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.framework.ApiDefs)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.framework.ApiDefs other = (org.tensorflow.proto.framework.ApiDefs) obj;
+
+    if (!getOpList()
+        .equals(other.getOpList())) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    if (getOpCount() > 0) {
+      hash = (37 * hash) + OP_FIELD_NUMBER;
+      hash = (53 * hash) + getOpList().hashCode();
+    }
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.framework.ApiDefs parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.ApiDefs parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.ApiDefs parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.ApiDefs parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.ApiDefs parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.ApiDefs parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.ApiDefs parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.ApiDefs parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.ApiDefs parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.ApiDefs parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.ApiDefs parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.ApiDefs parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.framework.ApiDefs prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * Protobuf type {@code tensorflow.ApiDefs}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.ApiDefs)
+      org.tensorflow.proto.framework.ApiDefsOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.ApiDefProtos.internal_static_tensorflow_ApiDefs_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.ApiDefProtos.internal_static_tensorflow_ApiDefs_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.ApiDefs.class, org.tensorflow.proto.framework.ApiDefs.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.framework.ApiDefs.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+        getOpFieldBuilder();
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      if (opBuilder_ == null) {
+        op_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000001);
+      } else {
+        opBuilder_.clear();
+      }
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.framework.ApiDefProtos.internal_static_tensorflow_ApiDefs_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.ApiDefs getDefaultInstanceForType() {
+      return org.tensorflow.proto.framework.ApiDefs.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.ApiDefs build() {
+      org.tensorflow.proto.framework.ApiDefs result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.ApiDefs buildPartial() {
+      org.tensorflow.proto.framework.ApiDefs result = new org.tensorflow.proto.framework.ApiDefs(this);
+      int from_bitField0_ = bitField0_;
+      if (opBuilder_ == null) {
+        if (((bitField0_ & 0x00000001) != 0)) {
+          op_ = java.util.Collections.unmodifiableList(op_);
+          bitField0_ = (bitField0_ & ~0x00000001);
+        }
+        result.op_ = op_;
+      } else {
+        result.op_ = opBuilder_.build();
+      }
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.framework.ApiDefs) {
+        return mergeFrom((org.tensorflow.proto.framework.ApiDefs)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.framework.ApiDefs other) {
+      if (other == org.tensorflow.proto.framework.ApiDefs.getDefaultInstance()) return this;
+      if (opBuilder_ == null) {
+        if (!other.op_.isEmpty()) {
+          if (op_.isEmpty()) {
+            op_ = other.op_;
+            bitField0_ = (bitField0_ & ~0x00000001);
+          } else {
+            ensureOpIsMutable();
+            op_.addAll(other.op_);
+          }
+          onChanged();
+        }
+      } else {
+        if (!other.op_.isEmpty()) {
+          if (opBuilder_.isEmpty()) {
+            opBuilder_.dispose();
+            opBuilder_ = null;
+            op_ = other.op_;
+            bitField0_ = (bitField0_ & ~0x00000001);
+            opBuilder_ = 
+              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
+                 getOpFieldBuilder() : null;
+          } else {
+            opBuilder_.addAllMessages(other.op_);
+          }
+        }
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.framework.ApiDefs parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.framework.ApiDefs) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+    private int bitField0_;
+
+    private java.util.List<org.tensorflow.proto.framework.ApiDef> op_ =
+      java.util.Collections.emptyList();
+    private void ensureOpIsMutable() {
+      if (!((bitField0_ & 0x00000001) != 0)) {
+        op_ = new java.util.ArrayList<org.tensorflow.proto.framework.ApiDef>(op_);
+        bitField0_ |= 0x00000001;
+       }
+    }
+
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.framework.ApiDef, org.tensorflow.proto.framework.ApiDef.Builder, org.tensorflow.proto.framework.ApiDefOrBuilder> opBuilder_;
+
+    /**
+     * <code>repeated .tensorflow.ApiDef op = 1;</code>
+     */
+    public java.util.List<org.tensorflow.proto.framework.ApiDef> getOpList() {
+      if (opBuilder_ == null) {
+        return java.util.Collections.unmodifiableList(op_);
+      } else {
+        return opBuilder_.getMessageList();
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.ApiDef op = 1;</code>
+     */
+    public int getOpCount() {
+      if (opBuilder_ == null) {
+        return op_.size();
+      } else {
+        return opBuilder_.getCount();
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.ApiDef op = 1;</code>
+     */
+    public org.tensorflow.proto.framework.ApiDef getOp(int index) {
+      if (opBuilder_ == null) {
+        return op_.get(index);
+      } else {
+        return opBuilder_.getMessage(index);
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.ApiDef op = 1;</code>
+     */
+    public Builder setOp(
+        int index, org.tensorflow.proto.framework.ApiDef value) {
+      if (opBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureOpIsMutable();
+        op_.set(index, value);
+        onChanged();
+      } else {
+        opBuilder_.setMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.ApiDef op = 1;</code>
+     */
+    public Builder setOp(
+        int index, org.tensorflow.proto.framework.ApiDef.Builder builderForValue) {
+      if (opBuilder_ == null) {
+        ensureOpIsMutable();
+        op_.set(index, builderForValue.build());
+        onChanged();
+      } else {
+        opBuilder_.setMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.ApiDef op = 1;</code>
+     */
+    public Builder addOp(org.tensorflow.proto.framework.ApiDef value) {
+      if (opBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureOpIsMutable();
+        op_.add(value);
+        onChanged();
+      } else {
+        opBuilder_.addMessage(value);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.ApiDef op = 1;</code>
+     */
+    public Builder addOp(
+        int index, org.tensorflow.proto.framework.ApiDef value) {
+      if (opBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureOpIsMutable();
+        op_.add(index, value);
+        onChanged();
+      } else {
+        opBuilder_.addMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.ApiDef op = 1;</code>
+     */
+    public Builder addOp(
+        org.tensorflow.proto.framework.ApiDef.Builder builderForValue) {
+      if (opBuilder_ == null) {
+        ensureOpIsMutable();
+        op_.add(builderForValue.build());
+        onChanged();
+      } else {
+        opBuilder_.addMessage(builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.ApiDef op = 1;</code>
+     */
+    public Builder addOp(
+        int index, org.tensorflow.proto.framework.ApiDef.Builder builderForValue) {
+      if (opBuilder_ == null) {
+        ensureOpIsMutable();
+        op_.add(index, builderForValue.build());
+        onChanged();
+      } else {
+        opBuilder_.addMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.ApiDef op = 1;</code>
+     */
+    public Builder addAllOp(
+        java.lang.Iterable<? extends org.tensorflow.proto.framework.ApiDef> values) {
+      if (opBuilder_ == null) {
+        ensureOpIsMutable();
+        com.google.protobuf.AbstractMessageLite.Builder.addAll(
+            values, op_);
+        onChanged();
+      } else {
+        opBuilder_.addAllMessages(values);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.ApiDef op = 1;</code>
+     */
+    public Builder clearOp() {
+      if (opBuilder_ == null) {
+        op_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000001);
+        onChanged();
+      } else {
+        opBuilder_.clear();
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.ApiDef op = 1;</code>
+     */
+    public Builder removeOp(int index) {
+      if (opBuilder_ == null) {
+        ensureOpIsMutable();
+        op_.remove(index);
+        onChanged();
+      } else {
+        opBuilder_.remove(index);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.ApiDef op = 1;</code>
+     */
+    public org.tensorflow.proto.framework.ApiDef.Builder getOpBuilder(
+        int index) {
+      return getOpFieldBuilder().getBuilder(index);
+    }
+    /**
+     * <code>repeated .tensorflow.ApiDef op = 1;</code>
+     */
+    public org.tensorflow.proto.framework.ApiDefOrBuilder getOpOrBuilder(
+        int index) {
+      if (opBuilder_ == null) {
+        return op_.get(index);  } else {
+        return opBuilder_.getMessageOrBuilder(index);
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.ApiDef op = 1;</code>
+     */
+    public java.util.List<? extends org.tensorflow.proto.framework.ApiDefOrBuilder> 
+         getOpOrBuilderList() {
+      if (opBuilder_ != null) {
+        return opBuilder_.getMessageOrBuilderList();
+      } else {
+        return java.util.Collections.unmodifiableList(op_);
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.ApiDef op = 1;</code>
+     */
+    public org.tensorflow.proto.framework.ApiDef.Builder addOpBuilder() {
+      return getOpFieldBuilder().addBuilder(
+          org.tensorflow.proto.framework.ApiDef.getDefaultInstance());
+    }
+    /**
+     * <code>repeated .tensorflow.ApiDef op = 1;</code>
+     */
+    public org.tensorflow.proto.framework.ApiDef.Builder addOpBuilder(
+        int index) {
+      return getOpFieldBuilder().addBuilder(
+          index, org.tensorflow.proto.framework.ApiDef.getDefaultInstance());
+    }
+    /**
+     * <code>repeated .tensorflow.ApiDef op = 1;</code>
+     */
+    public java.util.List<org.tensorflow.proto.framework.ApiDef.Builder> 
+         getOpBuilderList() {
+      return getOpFieldBuilder().getBuilderList();
+    }
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.framework.ApiDef, org.tensorflow.proto.framework.ApiDef.Builder, org.tensorflow.proto.framework.ApiDefOrBuilder> 
+        getOpFieldBuilder() {
+      if (opBuilder_ == null) {
+        opBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
+            org.tensorflow.proto.framework.ApiDef, org.tensorflow.proto.framework.ApiDef.Builder, org.tensorflow.proto.framework.ApiDefOrBuilder>(
+                op_,
+                ((bitField0_ & 0x00000001) != 0),
+                getParentForChildren(),
+                isClean());
+        op_ = null;
+      }
+      return opBuilder_;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.ApiDefs)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.ApiDefs)
+  private static final org.tensorflow.proto.framework.ApiDefs DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.ApiDefs();
+  }
+
+  public static org.tensorflow.proto.framework.ApiDefs getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<ApiDefs>
+      PARSER = new com.google.protobuf.AbstractParser<ApiDefs>() {
+    @java.lang.Override
+    public ApiDefs parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new ApiDefs(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<ApiDefs> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<ApiDefs> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.framework.ApiDefs getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ApiDefsOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ApiDefsOrBuilder.java
new file mode 100644
index 00000000000..e3ba64964e7
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ApiDefsOrBuilder.java
@@ -0,0 +1,33 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/framework/api_def.proto
+
+package org.tensorflow.proto.framework;
+
+public interface ApiDefsOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.ApiDefs)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <code>repeated .tensorflow.ApiDef op = 1;</code>
+   */
+  java.util.List<org.tensorflow.proto.framework.ApiDef> 
+      getOpList();
+  /**
+   * <code>repeated .tensorflow.ApiDef op = 1;</code>
+   */
+  org.tensorflow.proto.framework.ApiDef getOp(int index);
+  /**
+   * <code>repeated .tensorflow.ApiDef op = 1;</code>
+   */
+  int getOpCount();
+  /**
+   * <code>repeated .tensorflow.ApiDef op = 1;</code>
+   */
+  java.util.List<? extends org.tensorflow.proto.framework.ApiDefOrBuilder> 
+      getOpOrBuilderList();
+  /**
+   * <code>repeated .tensorflow.ApiDef op = 1;</code>
+   */
+  org.tensorflow.proto.framework.ApiDefOrBuilder getOpOrBuilder(
+      int index);
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/AssetFileDef.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/AssetFileDef.java
new file mode 100644
index 00000000000..2d4a4e4b828
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/AssetFileDef.java
@@ -0,0 +1,827 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/meta_graph.proto
+
+package org.tensorflow.proto.framework;
+
+/**
+ * <pre>
+ * An asset file def for a single file or a set of sharded files with the same
+ * name.
+ * </pre>
+ *
+ * Protobuf type {@code tensorflow.AssetFileDef}
+ */
+public  final class AssetFileDef extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.AssetFileDef)
+    AssetFileDefOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use AssetFileDef.newBuilder() to construct.
+  private AssetFileDef(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private AssetFileDef() {
+    filename_ = "";
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new AssetFileDef();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private AssetFileDef(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 10: {
+            org.tensorflow.proto.framework.TensorInfo.Builder subBuilder = null;
+            if (tensorInfo_ != null) {
+              subBuilder = tensorInfo_.toBuilder();
+            }
+            tensorInfo_ = input.readMessage(org.tensorflow.proto.framework.TensorInfo.parser(), extensionRegistry);
+            if (subBuilder != null) {
+              subBuilder.mergeFrom(tensorInfo_);
+              tensorInfo_ = subBuilder.buildPartial();
+            }
+
+            break;
+          }
+          case 18: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            filename_ = s;
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.framework.MetaGraphProtos.internal_static_tensorflow_AssetFileDef_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.framework.MetaGraphProtos.internal_static_tensorflow_AssetFileDef_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.framework.AssetFileDef.class, org.tensorflow.proto.framework.AssetFileDef.Builder.class);
+  }
+
+  public static final int TENSOR_INFO_FIELD_NUMBER = 1;
+  private org.tensorflow.proto.framework.TensorInfo tensorInfo_;
+  /**
+   * <pre>
+   * The tensor to bind the asset filename to.
+   * </pre>
+   *
+   * <code>.tensorflow.TensorInfo tensor_info = 1;</code>
+   */
+  public boolean hasTensorInfo() {
+    return tensorInfo_ != null;
+  }
+  /**
+   * <pre>
+   * The tensor to bind the asset filename to.
+   * </pre>
+   *
+   * <code>.tensorflow.TensorInfo tensor_info = 1;</code>
+   */
+  public org.tensorflow.proto.framework.TensorInfo getTensorInfo() {
+    return tensorInfo_ == null ? org.tensorflow.proto.framework.TensorInfo.getDefaultInstance() : tensorInfo_;
+  }
+  /**
+   * <pre>
+   * The tensor to bind the asset filename to.
+   * </pre>
+   *
+   * <code>.tensorflow.TensorInfo tensor_info = 1;</code>
+   */
+  public org.tensorflow.proto.framework.TensorInfoOrBuilder getTensorInfoOrBuilder() {
+    return getTensorInfo();
+  }
+
+  public static final int FILENAME_FIELD_NUMBER = 2;
+  private volatile java.lang.Object filename_;
+  /**
+   * <pre>
+   * The filename within an assets directory. Note: does not include the path
+   * prefix, i.e. directories. For an asset at /tmp/path/vocab.txt, the filename
+   * would be "vocab.txt".
+   * </pre>
+   *
+   * <code>string filename = 2;</code>
+   */
+  public java.lang.String getFilename() {
+    java.lang.Object ref = filename_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      filename_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * The filename within an assets directory. Note: does not include the path
+   * prefix, i.e. directories. For an asset at /tmp/path/vocab.txt, the filename
+   * would be "vocab.txt".
+   * </pre>
+   *
+   * <code>string filename = 2;</code>
+   */
+  public com.google.protobuf.ByteString
+      getFilenameBytes() {
+    java.lang.Object ref = filename_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      filename_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    if (tensorInfo_ != null) {
+      output.writeMessage(1, getTensorInfo());
+    }
+    if (!getFilenameBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 2, filename_);
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (tensorInfo_ != null) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(1, getTensorInfo());
+    }
+    if (!getFilenameBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(2, filename_);
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.framework.AssetFileDef)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.framework.AssetFileDef other = (org.tensorflow.proto.framework.AssetFileDef) obj;
+
+    if (hasTensorInfo() != other.hasTensorInfo()) return false;
+    if (hasTensorInfo()) {
+      if (!getTensorInfo()
+          .equals(other.getTensorInfo())) return false;
+    }
+    if (!getFilename()
+        .equals(other.getFilename())) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    if (hasTensorInfo()) {
+      hash = (37 * hash) + TENSOR_INFO_FIELD_NUMBER;
+      hash = (53 * hash) + getTensorInfo().hashCode();
+    }
+    hash = (37 * hash) + FILENAME_FIELD_NUMBER;
+    hash = (53 * hash) + getFilename().hashCode();
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.framework.AssetFileDef parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.AssetFileDef parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.AssetFileDef parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.AssetFileDef parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.AssetFileDef parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.AssetFileDef parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.AssetFileDef parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.AssetFileDef parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.AssetFileDef parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.AssetFileDef parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.AssetFileDef parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.AssetFileDef parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.framework.AssetFileDef prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * <pre>
+   * An asset file def for a single file or a set of sharded files with the same
+   * name.
+   * </pre>
+   *
+   * Protobuf type {@code tensorflow.AssetFileDef}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.AssetFileDef)
+      org.tensorflow.proto.framework.AssetFileDefOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.MetaGraphProtos.internal_static_tensorflow_AssetFileDef_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.MetaGraphProtos.internal_static_tensorflow_AssetFileDef_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.AssetFileDef.class, org.tensorflow.proto.framework.AssetFileDef.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.framework.AssetFileDef.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      if (tensorInfoBuilder_ == null) {
+        tensorInfo_ = null;
+      } else {
+        tensorInfo_ = null;
+        tensorInfoBuilder_ = null;
+      }
+      filename_ = "";
+
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.framework.MetaGraphProtos.internal_static_tensorflow_AssetFileDef_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.AssetFileDef getDefaultInstanceForType() {
+      return org.tensorflow.proto.framework.AssetFileDef.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.AssetFileDef build() {
+      org.tensorflow.proto.framework.AssetFileDef result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.AssetFileDef buildPartial() {
+      org.tensorflow.proto.framework.AssetFileDef result = new org.tensorflow.proto.framework.AssetFileDef(this);
+      if (tensorInfoBuilder_ == null) {
+        result.tensorInfo_ = tensorInfo_;
+      } else {
+        result.tensorInfo_ = tensorInfoBuilder_.build();
+      }
+      result.filename_ = filename_;
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.framework.AssetFileDef) {
+        return mergeFrom((org.tensorflow.proto.framework.AssetFileDef)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.framework.AssetFileDef other) {
+      if (other == org.tensorflow.proto.framework.AssetFileDef.getDefaultInstance()) return this;
+      if (other.hasTensorInfo()) {
+        mergeTensorInfo(other.getTensorInfo());
+      }
+      if (!other.getFilename().isEmpty()) {
+        filename_ = other.filename_;
+        onChanged();
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.framework.AssetFileDef parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.framework.AssetFileDef) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+
+    private org.tensorflow.proto.framework.TensorInfo tensorInfo_;
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.TensorInfo, org.tensorflow.proto.framework.TensorInfo.Builder, org.tensorflow.proto.framework.TensorInfoOrBuilder> tensorInfoBuilder_;
+    /**
+     * <pre>
+     * The tensor to bind the asset filename to.
+     * </pre>
+     *
+     * <code>.tensorflow.TensorInfo tensor_info = 1;</code>
+     */
+    public boolean hasTensorInfo() {
+      return tensorInfoBuilder_ != null || tensorInfo_ != null;
+    }
+    /**
+     * <pre>
+     * The tensor to bind the asset filename to.
+     * </pre>
+     *
+     * <code>.tensorflow.TensorInfo tensor_info = 1;</code>
+     */
+    public org.tensorflow.proto.framework.TensorInfo getTensorInfo() {
+      if (tensorInfoBuilder_ == null) {
+        return tensorInfo_ == null ? org.tensorflow.proto.framework.TensorInfo.getDefaultInstance() : tensorInfo_;
+      } else {
+        return tensorInfoBuilder_.getMessage();
+      }
+    }
+    /**
+     * <pre>
+     * The tensor to bind the asset filename to.
+     * </pre>
+     *
+     * <code>.tensorflow.TensorInfo tensor_info = 1;</code>
+     */
+    public Builder setTensorInfo(org.tensorflow.proto.framework.TensorInfo value) {
+      if (tensorInfoBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        tensorInfo_ = value;
+        onChanged();
+      } else {
+        tensorInfoBuilder_.setMessage(value);
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * The tensor to bind the asset filename to.
+     * </pre>
+     *
+     * <code>.tensorflow.TensorInfo tensor_info = 1;</code>
+     */
+    public Builder setTensorInfo(
+        org.tensorflow.proto.framework.TensorInfo.Builder builderForValue) {
+      if (tensorInfoBuilder_ == null) {
+        tensorInfo_ = builderForValue.build();
+        onChanged();
+      } else {
+        tensorInfoBuilder_.setMessage(builderForValue.build());
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * The tensor to bind the asset filename to.
+     * </pre>
+     *
+     * <code>.tensorflow.TensorInfo tensor_info = 1;</code>
+     */
+    public Builder mergeTensorInfo(org.tensorflow.proto.framework.TensorInfo value) {
+      if (tensorInfoBuilder_ == null) {
+        if (tensorInfo_ != null) {
+          tensorInfo_ =
+            org.tensorflow.proto.framework.TensorInfo.newBuilder(tensorInfo_).mergeFrom(value).buildPartial();
+        } else {
+          tensorInfo_ = value;
+        }
+        onChanged();
+      } else {
+        tensorInfoBuilder_.mergeFrom(value);
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * The tensor to bind the asset filename to.
+     * </pre>
+     *
+     * <code>.tensorflow.TensorInfo tensor_info = 1;</code>
+     */
+    public Builder clearTensorInfo() {
+      if (tensorInfoBuilder_ == null) {
+        tensorInfo_ = null;
+        onChanged();
+      } else {
+        tensorInfo_ = null;
+        tensorInfoBuilder_ = null;
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * The tensor to bind the asset filename to.
+     * </pre>
+     *
+     * <code>.tensorflow.TensorInfo tensor_info = 1;</code>
+     */
+    public org.tensorflow.proto.framework.TensorInfo.Builder getTensorInfoBuilder() {
+      
+      onChanged();
+      return getTensorInfoFieldBuilder().getBuilder();
+    }
+    /**
+     * <pre>
+     * The tensor to bind the asset filename to.
+     * </pre>
+     *
+     * <code>.tensorflow.TensorInfo tensor_info = 1;</code>
+     */
+    public org.tensorflow.proto.framework.TensorInfoOrBuilder getTensorInfoOrBuilder() {
+      if (tensorInfoBuilder_ != null) {
+        return tensorInfoBuilder_.getMessageOrBuilder();
+      } else {
+        return tensorInfo_ == null ?
+            org.tensorflow.proto.framework.TensorInfo.getDefaultInstance() : tensorInfo_;
+      }
+    }
+    /**
+     * <pre>
+     * The tensor to bind the asset filename to.
+     * </pre>
+     *
+     * <code>.tensorflow.TensorInfo tensor_info = 1;</code>
+     */
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.TensorInfo, org.tensorflow.proto.framework.TensorInfo.Builder, org.tensorflow.proto.framework.TensorInfoOrBuilder> 
+        getTensorInfoFieldBuilder() {
+      if (tensorInfoBuilder_ == null) {
+        tensorInfoBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+            org.tensorflow.proto.framework.TensorInfo, org.tensorflow.proto.framework.TensorInfo.Builder, org.tensorflow.proto.framework.TensorInfoOrBuilder>(
+                getTensorInfo(),
+                getParentForChildren(),
+                isClean());
+        tensorInfo_ = null;
+      }
+      return tensorInfoBuilder_;
+    }
+
+    private java.lang.Object filename_ = "";
+    /**
+     * <pre>
+     * The filename within an assets directory. Note: does not include the path
+     * prefix, i.e. directories. For an asset at /tmp/path/vocab.txt, the filename
+     * would be "vocab.txt".
+     * </pre>
+     *
+     * <code>string filename = 2;</code>
+     */
+    public java.lang.String getFilename() {
+      java.lang.Object ref = filename_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        filename_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * The filename within an assets directory. Note: does not include the path
+     * prefix, i.e. directories. For an asset at /tmp/path/vocab.txt, the filename
+     * would be "vocab.txt".
+     * </pre>
+     *
+     * <code>string filename = 2;</code>
+     */
+    public com.google.protobuf.ByteString
+        getFilenameBytes() {
+      java.lang.Object ref = filename_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        filename_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * The filename within an assets directory. Note: does not include the path
+     * prefix, i.e. directories. For an asset at /tmp/path/vocab.txt, the filename
+     * would be "vocab.txt".
+     * </pre>
+     *
+     * <code>string filename = 2;</code>
+     */
+    public Builder setFilename(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      filename_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * The filename within an assets directory. Note: does not include the path
+     * prefix, i.e. directories. For an asset at /tmp/path/vocab.txt, the filename
+     * would be "vocab.txt".
+     * </pre>
+     *
+     * <code>string filename = 2;</code>
+     */
+    public Builder clearFilename() {
+      
+      filename_ = getDefaultInstance().getFilename();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * The filename within an assets directory. Note: does not include the path
+     * prefix, i.e. directories. For an asset at /tmp/path/vocab.txt, the filename
+     * would be "vocab.txt".
+     * </pre>
+     *
+     * <code>string filename = 2;</code>
+     */
+    public Builder setFilenameBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      filename_ = value;
+      onChanged();
+      return this;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.AssetFileDef)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.AssetFileDef)
+  private static final org.tensorflow.proto.framework.AssetFileDef DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.AssetFileDef();
+  }
+
+  public static org.tensorflow.proto.framework.AssetFileDef getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<AssetFileDef>
+      PARSER = new com.google.protobuf.AbstractParser<AssetFileDef>() {
+    @java.lang.Override
+    public AssetFileDef parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new AssetFileDef(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<AssetFileDef> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<AssetFileDef> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.framework.AssetFileDef getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/AssetFileDefOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/AssetFileDefOrBuilder.java
new file mode 100644
index 00000000000..dd8ec09799f
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/AssetFileDefOrBuilder.java
@@ -0,0 +1,56 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/meta_graph.proto
+
+package org.tensorflow.proto.framework;
+
+public interface AssetFileDefOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.AssetFileDef)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <pre>
+   * The tensor to bind the asset filename to.
+   * </pre>
+   *
+   * <code>.tensorflow.TensorInfo tensor_info = 1;</code>
+   */
+  boolean hasTensorInfo();
+  /**
+   * <pre>
+   * The tensor to bind the asset filename to.
+   * </pre>
+   *
+   * <code>.tensorflow.TensorInfo tensor_info = 1;</code>
+   */
+  org.tensorflow.proto.framework.TensorInfo getTensorInfo();
+  /**
+   * <pre>
+   * The tensor to bind the asset filename to.
+   * </pre>
+   *
+   * <code>.tensorflow.TensorInfo tensor_info = 1;</code>
+   */
+  org.tensorflow.proto.framework.TensorInfoOrBuilder getTensorInfoOrBuilder();
+
+  /**
+   * <pre>
+   * The filename within an assets directory. Note: does not include the path
+   * prefix, i.e. directories. For an asset at /tmp/path/vocab.txt, the filename
+   * would be "vocab.txt".
+   * </pre>
+   *
+   * <code>string filename = 2;</code>
+   */
+  java.lang.String getFilename();
+  /**
+   * <pre>
+   * The filename within an assets directory. Note: does not include the path
+   * prefix, i.e. directories. For an asset at /tmp/path/vocab.txt, the filename
+   * would be "vocab.txt".
+   * </pre>
+   *
+   * <code>string filename = 2;</code>
+   */
+  com.google.protobuf.ByteString
+      getFilenameBytes();
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/AttrValue.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/AttrValue.java
new file mode 100644
index 00000000000..f64f1a8b3ab
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/AttrValue.java
@@ -0,0 +1,5341 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/framework/attr_value.proto
+
+package org.tensorflow.proto.framework;
+
+/**
+ * <pre>
+ * Protocol buffer representing the value for an attr used to configure an Op.
+ * Comment indicates the corresponding attr type.  Only the field matching the
+ * attr type may be filled.
+ * </pre>
+ *
+ * Protobuf type {@code tensorflow.AttrValue}
+ */
+public  final class AttrValue extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.AttrValue)
+    AttrValueOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use AttrValue.newBuilder() to construct.
+  private AttrValue(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private AttrValue() {
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new AttrValue();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private AttrValue(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 10: {
+            org.tensorflow.proto.framework.AttrValue.ListValue.Builder subBuilder = null;
+            if (valueCase_ == 1) {
+              subBuilder = ((org.tensorflow.proto.framework.AttrValue.ListValue) value_).toBuilder();
+            }
+            value_ =
+                input.readMessage(org.tensorflow.proto.framework.AttrValue.ListValue.parser(), extensionRegistry);
+            if (subBuilder != null) {
+              subBuilder.mergeFrom((org.tensorflow.proto.framework.AttrValue.ListValue) value_);
+              value_ = subBuilder.buildPartial();
+            }
+            valueCase_ = 1;
+            break;
+          }
+          case 18: {
+            valueCase_ = 2;
+            value_ = input.readBytes();
+            break;
+          }
+          case 24: {
+            valueCase_ = 3;
+            value_ = input.readInt64();
+            break;
+          }
+          case 37: {
+            valueCase_ = 4;
+            value_ = input.readFloat();
+            break;
+          }
+          case 40: {
+            valueCase_ = 5;
+            value_ = input.readBool();
+            break;
+          }
+          case 48: {
+            int rawValue = input.readEnum();
+            valueCase_ = 6;
+            value_ = rawValue;
+            break;
+          }
+          case 58: {
+            org.tensorflow.proto.framework.TensorShapeProto.Builder subBuilder = null;
+            if (valueCase_ == 7) {
+              subBuilder = ((org.tensorflow.proto.framework.TensorShapeProto) value_).toBuilder();
+            }
+            value_ =
+                input.readMessage(org.tensorflow.proto.framework.TensorShapeProto.parser(), extensionRegistry);
+            if (subBuilder != null) {
+              subBuilder.mergeFrom((org.tensorflow.proto.framework.TensorShapeProto) value_);
+              value_ = subBuilder.buildPartial();
+            }
+            valueCase_ = 7;
+            break;
+          }
+          case 66: {
+            org.tensorflow.proto.framework.TensorProto.Builder subBuilder = null;
+            if (valueCase_ == 8) {
+              subBuilder = ((org.tensorflow.proto.framework.TensorProto) value_).toBuilder();
+            }
+            value_ =
+                input.readMessage(org.tensorflow.proto.framework.TensorProto.parser(), extensionRegistry);
+            if (subBuilder != null) {
+              subBuilder.mergeFrom((org.tensorflow.proto.framework.TensorProto) value_);
+              value_ = subBuilder.buildPartial();
+            }
+            valueCase_ = 8;
+            break;
+          }
+          case 74: {
+            java.lang.String s = input.readStringRequireUtf8();
+            valueCase_ = 9;
+            value_ = s;
+            break;
+          }
+          case 82: {
+            org.tensorflow.proto.framework.NameAttrList.Builder subBuilder = null;
+            if (valueCase_ == 10) {
+              subBuilder = ((org.tensorflow.proto.framework.NameAttrList) value_).toBuilder();
+            }
+            value_ =
+                input.readMessage(org.tensorflow.proto.framework.NameAttrList.parser(), extensionRegistry);
+            if (subBuilder != null) {
+              subBuilder.mergeFrom((org.tensorflow.proto.framework.NameAttrList) value_);
+              value_ = subBuilder.buildPartial();
+            }
+            valueCase_ = 10;
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.framework.AttrValueProtos.internal_static_tensorflow_AttrValue_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.framework.AttrValueProtos.internal_static_tensorflow_AttrValue_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.framework.AttrValue.class, org.tensorflow.proto.framework.AttrValue.Builder.class);
+  }
+
+  public interface ListValueOrBuilder extends
+      // @@protoc_insertion_point(interface_extends:tensorflow.AttrValue.ListValue)
+      com.google.protobuf.MessageOrBuilder {
+
+    /**
+     * <pre>
+     * "list(string)"
+     * </pre>
+     *
+     * <code>repeated bytes s = 2;</code>
+     */
+    java.util.List<com.google.protobuf.ByteString> getSList();
+    /**
+     * <pre>
+     * "list(string)"
+     * </pre>
+     *
+     * <code>repeated bytes s = 2;</code>
+     */
+    int getSCount();
+    /**
+     * <pre>
+     * "list(string)"
+     * </pre>
+     *
+     * <code>repeated bytes s = 2;</code>
+     */
+    com.google.protobuf.ByteString getS(int index);
+
+    /**
+     * <pre>
+     * "list(int)"
+     * </pre>
+     *
+     * <code>repeated int64 i = 3 [packed = true];</code>
+     */
+    java.util.List<java.lang.Long> getIList();
+    /**
+     * <pre>
+     * "list(int)"
+     * </pre>
+     *
+     * <code>repeated int64 i = 3 [packed = true];</code>
+     */
+    int getICount();
+    /**
+     * <pre>
+     * "list(int)"
+     * </pre>
+     *
+     * <code>repeated int64 i = 3 [packed = true];</code>
+     */
+    long getI(int index);
+
+    /**
+     * <pre>
+     * "list(float)"
+     * </pre>
+     *
+     * <code>repeated float f = 4 [packed = true];</code>
+     */
+    java.util.List<java.lang.Float> getFList();
+    /**
+     * <pre>
+     * "list(float)"
+     * </pre>
+     *
+     * <code>repeated float f = 4 [packed = true];</code>
+     */
+    int getFCount();
+    /**
+     * <pre>
+     * "list(float)"
+     * </pre>
+     *
+     * <code>repeated float f = 4 [packed = true];</code>
+     */
+    float getF(int index);
+
+    /**
+     * <pre>
+     * "list(bool)"
+     * </pre>
+     *
+     * <code>repeated bool b = 5 [packed = true];</code>
+     */
+    java.util.List<java.lang.Boolean> getBList();
+    /**
+     * <pre>
+     * "list(bool)"
+     * </pre>
+     *
+     * <code>repeated bool b = 5 [packed = true];</code>
+     */
+    int getBCount();
+    /**
+     * <pre>
+     * "list(bool)"
+     * </pre>
+     *
+     * <code>repeated bool b = 5 [packed = true];</code>
+     */
+    boolean getB(int index);
+
+    /**
+     * <pre>
+     * "list(type)"
+     * </pre>
+     *
+     * <code>repeated .tensorflow.DataType type = 6 [packed = true];</code>
+     */
+    java.util.List<org.tensorflow.proto.framework.DataType> getTypeList();
+    /**
+     * <pre>
+     * "list(type)"
+     * </pre>
+     *
+     * <code>repeated .tensorflow.DataType type = 6 [packed = true];</code>
+     */
+    int getTypeCount();
+    /**
+     * <pre>
+     * "list(type)"
+     * </pre>
+     *
+     * <code>repeated .tensorflow.DataType type = 6 [packed = true];</code>
+     */
+    org.tensorflow.proto.framework.DataType getType(int index);
+    /**
+     * <pre>
+     * "list(type)"
+     * </pre>
+     *
+     * <code>repeated .tensorflow.DataType type = 6 [packed = true];</code>
+     */
+    java.util.List<java.lang.Integer>
+    getTypeValueList();
+    /**
+     * <pre>
+     * "list(type)"
+     * </pre>
+     *
+     * <code>repeated .tensorflow.DataType type = 6 [packed = true];</code>
+     */
+    int getTypeValue(int index);
+
+    /**
+     * <pre>
+     * "list(shape)"
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorShapeProto shape = 7;</code>
+     */
+    java.util.List<org.tensorflow.proto.framework.TensorShapeProto> 
+        getShapeList();
+    /**
+     * <pre>
+     * "list(shape)"
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorShapeProto shape = 7;</code>
+     */
+    org.tensorflow.proto.framework.TensorShapeProto getShape(int index);
+    /**
+     * <pre>
+     * "list(shape)"
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorShapeProto shape = 7;</code>
+     */
+    int getShapeCount();
+    /**
+     * <pre>
+     * "list(shape)"
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorShapeProto shape = 7;</code>
+     */
+    java.util.List<? extends org.tensorflow.proto.framework.TensorShapeProtoOrBuilder> 
+        getShapeOrBuilderList();
+    /**
+     * <pre>
+     * "list(shape)"
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorShapeProto shape = 7;</code>
+     */
+    org.tensorflow.proto.framework.TensorShapeProtoOrBuilder getShapeOrBuilder(
+        int index);
+
+    /**
+     * <pre>
+     * "list(tensor)"
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorProto tensor = 8;</code>
+     */
+    java.util.List<org.tensorflow.proto.framework.TensorProto> 
+        getTensorList();
+    /**
+     * <pre>
+     * "list(tensor)"
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorProto tensor = 8;</code>
+     */
+    org.tensorflow.proto.framework.TensorProto getTensor(int index);
+    /**
+     * <pre>
+     * "list(tensor)"
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorProto tensor = 8;</code>
+     */
+    int getTensorCount();
+    /**
+     * <pre>
+     * "list(tensor)"
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorProto tensor = 8;</code>
+     */
+    java.util.List<? extends org.tensorflow.proto.framework.TensorProtoOrBuilder> 
+        getTensorOrBuilderList();
+    /**
+     * <pre>
+     * "list(tensor)"
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorProto tensor = 8;</code>
+     */
+    org.tensorflow.proto.framework.TensorProtoOrBuilder getTensorOrBuilder(
+        int index);
+
+    /**
+     * <pre>
+     * "list(attr)"
+     * </pre>
+     *
+     * <code>repeated .tensorflow.NameAttrList func = 9;</code>
+     */
+    java.util.List<org.tensorflow.proto.framework.NameAttrList> 
+        getFuncList();
+    /**
+     * <pre>
+     * "list(attr)"
+     * </pre>
+     *
+     * <code>repeated .tensorflow.NameAttrList func = 9;</code>
+     */
+    org.tensorflow.proto.framework.NameAttrList getFunc(int index);
+    /**
+     * <pre>
+     * "list(attr)"
+     * </pre>
+     *
+     * <code>repeated .tensorflow.NameAttrList func = 9;</code>
+     */
+    int getFuncCount();
+    /**
+     * <pre>
+     * "list(attr)"
+     * </pre>
+     *
+     * <code>repeated .tensorflow.NameAttrList func = 9;</code>
+     */
+    java.util.List<? extends org.tensorflow.proto.framework.NameAttrListOrBuilder> 
+        getFuncOrBuilderList();
+    /**
+     * <pre>
+     * "list(attr)"
+     * </pre>
+     *
+     * <code>repeated .tensorflow.NameAttrList func = 9;</code>
+     */
+    org.tensorflow.proto.framework.NameAttrListOrBuilder getFuncOrBuilder(
+        int index);
+  }
+  /**
+   * <pre>
+   * LINT.IfChange
+   * </pre>
+   *
+   * Protobuf type {@code tensorflow.AttrValue.ListValue}
+   */
+  public  static final class ListValue extends
+      com.google.protobuf.GeneratedMessageV3 implements
+      // @@protoc_insertion_point(message_implements:tensorflow.AttrValue.ListValue)
+      ListValueOrBuilder {
+  private static final long serialVersionUID = 0L;
+    // Use ListValue.newBuilder() to construct.
+    private ListValue(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+      super(builder);
+    }
+    private ListValue() {
+      s_ = java.util.Collections.emptyList();
+      i_ = emptyLongList();
+      f_ = emptyFloatList();
+      b_ = emptyBooleanList();
+      type_ = java.util.Collections.emptyList();
+      shape_ = java.util.Collections.emptyList();
+      tensor_ = java.util.Collections.emptyList();
+      func_ = java.util.Collections.emptyList();
+    }
+
+    @java.lang.Override
+    @SuppressWarnings({"unused"})
+    protected java.lang.Object newInstance(
+        UnusedPrivateParameter unused) {
+      return new ListValue();
+    }
+
+    @java.lang.Override
+    public final com.google.protobuf.UnknownFieldSet
+    getUnknownFields() {
+      return this.unknownFields;
+    }
+    private ListValue(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      this();
+      if (extensionRegistry == null) {
+        throw new java.lang.NullPointerException();
+      }
+      int mutable_bitField0_ = 0;
+      com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+          com.google.protobuf.UnknownFieldSet.newBuilder();
+      try {
+        boolean done = false;
+        while (!done) {
+          int tag = input.readTag();
+          switch (tag) {
+            case 0:
+              done = true;
+              break;
+            case 18: {
+              if (!((mutable_bitField0_ & 0x00000001) != 0)) {
+                s_ = new java.util.ArrayList<com.google.protobuf.ByteString>();
+                mutable_bitField0_ |= 0x00000001;
+              }
+              s_.add(input.readBytes());
+              break;
+            }
+            case 24: {
+              if (!((mutable_bitField0_ & 0x00000002) != 0)) {
+                i_ = newLongList();
+                mutable_bitField0_ |= 0x00000002;
+              }
+              i_.addLong(input.readInt64());
+              break;
+            }
+            case 26: {
+              int length = input.readRawVarint32();
+              int limit = input.pushLimit(length);
+              if (!((mutable_bitField0_ & 0x00000002) != 0) && input.getBytesUntilLimit() > 0) {
+                i_ = newLongList();
+                mutable_bitField0_ |= 0x00000002;
+              }
+              while (input.getBytesUntilLimit() > 0) {
+                i_.addLong(input.readInt64());
+              }
+              input.popLimit(limit);
+              break;
+            }
+            case 37: {
+              if (!((mutable_bitField0_ & 0x00000004) != 0)) {
+                f_ = newFloatList();
+                mutable_bitField0_ |= 0x00000004;
+              }
+              f_.addFloat(input.readFloat());
+              break;
+            }
+            case 34: {
+              int length = input.readRawVarint32();
+              int limit = input.pushLimit(length);
+              if (!((mutable_bitField0_ & 0x00000004) != 0) && input.getBytesUntilLimit() > 0) {
+                f_ = newFloatList();
+                mutable_bitField0_ |= 0x00000004;
+              }
+              while (input.getBytesUntilLimit() > 0) {
+                f_.addFloat(input.readFloat());
+              }
+              input.popLimit(limit);
+              break;
+            }
+            case 40: {
+              if (!((mutable_bitField0_ & 0x00000008) != 0)) {
+                b_ = newBooleanList();
+                mutable_bitField0_ |= 0x00000008;
+              }
+              b_.addBoolean(input.readBool());
+              break;
+            }
+            case 42: {
+              int length = input.readRawVarint32();
+              int limit = input.pushLimit(length);
+              if (!((mutable_bitField0_ & 0x00000008) != 0) && input.getBytesUntilLimit() > 0) {
+                b_ = newBooleanList();
+                mutable_bitField0_ |= 0x00000008;
+              }
+              while (input.getBytesUntilLimit() > 0) {
+                b_.addBoolean(input.readBool());
+              }
+              input.popLimit(limit);
+              break;
+            }
+            case 48: {
+              int rawValue = input.readEnum();
+              if (!((mutable_bitField0_ & 0x00000010) != 0)) {
+                type_ = new java.util.ArrayList<java.lang.Integer>();
+                mutable_bitField0_ |= 0x00000010;
+              }
+              type_.add(rawValue);
+              break;
+            }
+            case 50: {
+              int length = input.readRawVarint32();
+              int oldLimit = input.pushLimit(length);
+              while(input.getBytesUntilLimit() > 0) {
+                int rawValue = input.readEnum();
+                if (!((mutable_bitField0_ & 0x00000010) != 0)) {
+                  type_ = new java.util.ArrayList<java.lang.Integer>();
+                  mutable_bitField0_ |= 0x00000010;
+                }
+                type_.add(rawValue);
+              }
+              input.popLimit(oldLimit);
+              break;
+            }
+            case 58: {
+              if (!((mutable_bitField0_ & 0x00000020) != 0)) {
+                shape_ = new java.util.ArrayList<org.tensorflow.proto.framework.TensorShapeProto>();
+                mutable_bitField0_ |= 0x00000020;
+              }
+              shape_.add(
+                  input.readMessage(org.tensorflow.proto.framework.TensorShapeProto.parser(), extensionRegistry));
+              break;
+            }
+            case 66: {
+              if (!((mutable_bitField0_ & 0x00000040) != 0)) {
+                tensor_ = new java.util.ArrayList<org.tensorflow.proto.framework.TensorProto>();
+                mutable_bitField0_ |= 0x00000040;
+              }
+              tensor_.add(
+                  input.readMessage(org.tensorflow.proto.framework.TensorProto.parser(), extensionRegistry));
+              break;
+            }
+            case 74: {
+              if (!((mutable_bitField0_ & 0x00000080) != 0)) {
+                func_ = new java.util.ArrayList<org.tensorflow.proto.framework.NameAttrList>();
+                mutable_bitField0_ |= 0x00000080;
+              }
+              func_.add(
+                  input.readMessage(org.tensorflow.proto.framework.NameAttrList.parser(), extensionRegistry));
+              break;
+            }
+            default: {
+              if (!parseUnknownField(
+                  input, unknownFields, extensionRegistry, tag)) {
+                done = true;
+              }
+              break;
+            }
+          }
+        }
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        throw e.setUnfinishedMessage(this);
+      } catch (java.io.IOException e) {
+        throw new com.google.protobuf.InvalidProtocolBufferException(
+            e).setUnfinishedMessage(this);
+      } finally {
+        if (((mutable_bitField0_ & 0x00000001) != 0)) {
+          s_ = java.util.Collections.unmodifiableList(s_); // C
+        }
+        if (((mutable_bitField0_ & 0x00000002) != 0)) {
+          i_.makeImmutable(); // C
+        }
+        if (((mutable_bitField0_ & 0x00000004) != 0)) {
+          f_.makeImmutable(); // C
+        }
+        if (((mutable_bitField0_ & 0x00000008) != 0)) {
+          b_.makeImmutable(); // C
+        }
+        if (((mutable_bitField0_ & 0x00000010) != 0)) {
+          type_ = java.util.Collections.unmodifiableList(type_);
+        }
+        if (((mutable_bitField0_ & 0x00000020) != 0)) {
+          shape_ = java.util.Collections.unmodifiableList(shape_);
+        }
+        if (((mutable_bitField0_ & 0x00000040) != 0)) {
+          tensor_ = java.util.Collections.unmodifiableList(tensor_);
+        }
+        if (((mutable_bitField0_ & 0x00000080) != 0)) {
+          func_ = java.util.Collections.unmodifiableList(func_);
+        }
+        this.unknownFields = unknownFields.build();
+        makeExtensionsImmutable();
+      }
+    }
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.AttrValueProtos.internal_static_tensorflow_AttrValue_ListValue_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.AttrValueProtos.internal_static_tensorflow_AttrValue_ListValue_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.AttrValue.ListValue.class, org.tensorflow.proto.framework.AttrValue.ListValue.Builder.class);
+    }
+
+    public static final int S_FIELD_NUMBER = 2;
+    private java.util.List<com.google.protobuf.ByteString> s_;
+    /**
+     * <pre>
+     * "list(string)"
+     * </pre>
+     *
+     * <code>repeated bytes s = 2;</code>
+     */
+    public java.util.List<com.google.protobuf.ByteString>
+        getSList() {
+      return s_;
+    }
+    /**
+     * <pre>
+     * "list(string)"
+     * </pre>
+     *
+     * <code>repeated bytes s = 2;</code>
+     */
+    public int getSCount() {
+      return s_.size();
+    }
+    /**
+     * <pre>
+     * "list(string)"
+     * </pre>
+     *
+     * <code>repeated bytes s = 2;</code>
+     */
+    public com.google.protobuf.ByteString getS(int index) {
+      return s_.get(index);
+    }
+
+    public static final int I_FIELD_NUMBER = 3;
+    private com.google.protobuf.Internal.LongList i_;
+    /**
+     * <pre>
+     * "list(int)"
+     * </pre>
+     *
+     * <code>repeated int64 i = 3 [packed = true];</code>
+     */
+    public java.util.List<java.lang.Long>
+        getIList() {
+      return i_;
+    }
+    /**
+     * <pre>
+     * "list(int)"
+     * </pre>
+     *
+     * <code>repeated int64 i = 3 [packed = true];</code>
+     */
+    public int getICount() {
+      return i_.size();
+    }
+    /**
+     * <pre>
+     * "list(int)"
+     * </pre>
+     *
+     * <code>repeated int64 i = 3 [packed = true];</code>
+     */
+    public long getI(int index) {
+      return i_.getLong(index);
+    }
+    private int iMemoizedSerializedSize = -1;
+
+    public static final int F_FIELD_NUMBER = 4;
+    private com.google.protobuf.Internal.FloatList f_;
+    /**
+     * <pre>
+     * "list(float)"
+     * </pre>
+     *
+     * <code>repeated float f = 4 [packed = true];</code>
+     */
+    public java.util.List<java.lang.Float>
+        getFList() {
+      return f_;
+    }
+    /**
+     * <pre>
+     * "list(float)"
+     * </pre>
+     *
+     * <code>repeated float f = 4 [packed = true];</code>
+     */
+    public int getFCount() {
+      return f_.size();
+    }
+    /**
+     * <pre>
+     * "list(float)"
+     * </pre>
+     *
+     * <code>repeated float f = 4 [packed = true];</code>
+     */
+    public float getF(int index) {
+      return f_.getFloat(index);
+    }
+    private int fMemoizedSerializedSize = -1;
+
+    public static final int B_FIELD_NUMBER = 5;
+    private com.google.protobuf.Internal.BooleanList b_;
+    /**
+     * <pre>
+     * "list(bool)"
+     * </pre>
+     *
+     * <code>repeated bool b = 5 [packed = true];</code>
+     */
+    public java.util.List<java.lang.Boolean>
+        getBList() {
+      return b_;
+    }
+    /**
+     * <pre>
+     * "list(bool)"
+     * </pre>
+     *
+     * <code>repeated bool b = 5 [packed = true];</code>
+     */
+    public int getBCount() {
+      return b_.size();
+    }
+    /**
+     * <pre>
+     * "list(bool)"
+     * </pre>
+     *
+     * <code>repeated bool b = 5 [packed = true];</code>
+     */
+    public boolean getB(int index) {
+      return b_.getBoolean(index);
+    }
+    private int bMemoizedSerializedSize = -1;
+
+    public static final int TYPE_FIELD_NUMBER = 6;
+    private java.util.List<java.lang.Integer> type_;
+    private static final com.google.protobuf.Internal.ListAdapter.Converter<
+        java.lang.Integer, org.tensorflow.proto.framework.DataType> type_converter_ =
+            new com.google.protobuf.Internal.ListAdapter.Converter<
+                java.lang.Integer, org.tensorflow.proto.framework.DataType>() {
+              public org.tensorflow.proto.framework.DataType convert(java.lang.Integer from) {
+                @SuppressWarnings("deprecation")
+                org.tensorflow.proto.framework.DataType result = org.tensorflow.proto.framework.DataType.valueOf(from);
+                return result == null ? org.tensorflow.proto.framework.DataType.UNRECOGNIZED : result;
+              }
+            };
+    /**
+     * <pre>
+     * "list(type)"
+     * </pre>
+     *
+     * <code>repeated .tensorflow.DataType type = 6 [packed = true];</code>
+     */
+    public java.util.List<org.tensorflow.proto.framework.DataType> getTypeList() {
+      return new com.google.protobuf.Internal.ListAdapter<
+          java.lang.Integer, org.tensorflow.proto.framework.DataType>(type_, type_converter_);
+    }
+    /**
+     * <pre>
+     * "list(type)"
+     * </pre>
+     *
+     * <code>repeated .tensorflow.DataType type = 6 [packed = true];</code>
+     */
+    public int getTypeCount() {
+      return type_.size();
+    }
+    /**
+     * <pre>
+     * "list(type)"
+     * </pre>
+     *
+     * <code>repeated .tensorflow.DataType type = 6 [packed = true];</code>
+     */
+    public org.tensorflow.proto.framework.DataType getType(int index) {
+      return type_converter_.convert(type_.get(index));
+    }
+    /**
+     * <pre>
+     * "list(type)"
+     * </pre>
+     *
+     * <code>repeated .tensorflow.DataType type = 6 [packed = true];</code>
+     */
+    public java.util.List<java.lang.Integer>
+    getTypeValueList() {
+      return type_;
+    }
+    /**
+     * <pre>
+     * "list(type)"
+     * </pre>
+     *
+     * <code>repeated .tensorflow.DataType type = 6 [packed = true];</code>
+     */
+    public int getTypeValue(int index) {
+      return type_.get(index);
+    }
+    private int typeMemoizedSerializedSize;
+
+    public static final int SHAPE_FIELD_NUMBER = 7;
+    private java.util.List<org.tensorflow.proto.framework.TensorShapeProto> shape_;
+    /**
+     * <pre>
+     * "list(shape)"
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorShapeProto shape = 7;</code>
+     */
+    public java.util.List<org.tensorflow.proto.framework.TensorShapeProto> getShapeList() {
+      return shape_;
+    }
+    /**
+     * <pre>
+     * "list(shape)"
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorShapeProto shape = 7;</code>
+     */
+    public java.util.List<? extends org.tensorflow.proto.framework.TensorShapeProtoOrBuilder> 
+        getShapeOrBuilderList() {
+      return shape_;
+    }
+    /**
+     * <pre>
+     * "list(shape)"
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorShapeProto shape = 7;</code>
+     */
+    public int getShapeCount() {
+      return shape_.size();
+    }
+    /**
+     * <pre>
+     * "list(shape)"
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorShapeProto shape = 7;</code>
+     */
+    public org.tensorflow.proto.framework.TensorShapeProto getShape(int index) {
+      return shape_.get(index);
+    }
+    /**
+     * <pre>
+     * "list(shape)"
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorShapeProto shape = 7;</code>
+     */
+    public org.tensorflow.proto.framework.TensorShapeProtoOrBuilder getShapeOrBuilder(
+        int index) {
+      return shape_.get(index);
+    }
+
+    public static final int TENSOR_FIELD_NUMBER = 8;
+    private java.util.List<org.tensorflow.proto.framework.TensorProto> tensor_;
+    /**
+     * <pre>
+     * "list(tensor)"
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorProto tensor = 8;</code>
+     */
+    public java.util.List<org.tensorflow.proto.framework.TensorProto> getTensorList() {
+      return tensor_;
+    }
+    /**
+     * <pre>
+     * "list(tensor)"
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorProto tensor = 8;</code>
+     */
+    public java.util.List<? extends org.tensorflow.proto.framework.TensorProtoOrBuilder> 
+        getTensorOrBuilderList() {
+      return tensor_;
+    }
+    /**
+     * <pre>
+     * "list(tensor)"
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorProto tensor = 8;</code>
+     */
+    public int getTensorCount() {
+      return tensor_.size();
+    }
+    /**
+     * <pre>
+     * "list(tensor)"
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorProto tensor = 8;</code>
+     */
+    public org.tensorflow.proto.framework.TensorProto getTensor(int index) {
+      return tensor_.get(index);
+    }
+    /**
+     * <pre>
+     * "list(tensor)"
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorProto tensor = 8;</code>
+     */
+    public org.tensorflow.proto.framework.TensorProtoOrBuilder getTensorOrBuilder(
+        int index) {
+      return tensor_.get(index);
+    }
+
+    public static final int FUNC_FIELD_NUMBER = 9;
+    private java.util.List<org.tensorflow.proto.framework.NameAttrList> func_;
+    /**
+     * <pre>
+     * "list(attr)"
+     * </pre>
+     *
+     * <code>repeated .tensorflow.NameAttrList func = 9;</code>
+     */
+    public java.util.List<org.tensorflow.proto.framework.NameAttrList> getFuncList() {
+      return func_;
+    }
+    /**
+     * <pre>
+     * "list(attr)"
+     * </pre>
+     *
+     * <code>repeated .tensorflow.NameAttrList func = 9;</code>
+     */
+    public java.util.List<? extends org.tensorflow.proto.framework.NameAttrListOrBuilder> 
+        getFuncOrBuilderList() {
+      return func_;
+    }
+    /**
+     * <pre>
+     * "list(attr)"
+     * </pre>
+     *
+     * <code>repeated .tensorflow.NameAttrList func = 9;</code>
+     */
+    public int getFuncCount() {
+      return func_.size();
+    }
+    /**
+     * <pre>
+     * "list(attr)"
+     * </pre>
+     *
+     * <code>repeated .tensorflow.NameAttrList func = 9;</code>
+     */
+    public org.tensorflow.proto.framework.NameAttrList getFunc(int index) {
+      return func_.get(index);
+    }
+    /**
+     * <pre>
+     * "list(attr)"
+     * </pre>
+     *
+     * <code>repeated .tensorflow.NameAttrList func = 9;</code>
+     */
+    public org.tensorflow.proto.framework.NameAttrListOrBuilder getFuncOrBuilder(
+        int index) {
+      return func_.get(index);
+    }
+
+    private byte memoizedIsInitialized = -1;
+    @java.lang.Override
+    public final boolean isInitialized() {
+      byte isInitialized = memoizedIsInitialized;
+      if (isInitialized == 1) return true;
+      if (isInitialized == 0) return false;
+
+      memoizedIsInitialized = 1;
+      return true;
+    }
+
+    @java.lang.Override
+    public void writeTo(com.google.protobuf.CodedOutputStream output)
+                        throws java.io.IOException {
+      getSerializedSize();
+      for (int i = 0; i < s_.size(); i++) {
+        output.writeBytes(2, s_.get(i));
+      }
+      if (getIList().size() > 0) {
+        output.writeUInt32NoTag(26);
+        output.writeUInt32NoTag(iMemoizedSerializedSize);
+      }
+      for (int i = 0; i < i_.size(); i++) {
+        output.writeInt64NoTag(i_.getLong(i));
+      }
+      if (getFList().size() > 0) {
+        output.writeUInt32NoTag(34);
+        output.writeUInt32NoTag(fMemoizedSerializedSize);
+      }
+      for (int i = 0; i < f_.size(); i++) {
+        output.writeFloatNoTag(f_.getFloat(i));
+      }
+      if (getBList().size() > 0) {
+        output.writeUInt32NoTag(42);
+        output.writeUInt32NoTag(bMemoizedSerializedSize);
+      }
+      for (int i = 0; i < b_.size(); i++) {
+        output.writeBoolNoTag(b_.getBoolean(i));
+      }
+      if (getTypeList().size() > 0) {
+        output.writeUInt32NoTag(50);
+        output.writeUInt32NoTag(typeMemoizedSerializedSize);
+      }
+      for (int i = 0; i < type_.size(); i++) {
+        output.writeEnumNoTag(type_.get(i));
+      }
+      for (int i = 0; i < shape_.size(); i++) {
+        output.writeMessage(7, shape_.get(i));
+      }
+      for (int i = 0; i < tensor_.size(); i++) {
+        output.writeMessage(8, tensor_.get(i));
+      }
+      for (int i = 0; i < func_.size(); i++) {
+        output.writeMessage(9, func_.get(i));
+      }
+      unknownFields.writeTo(output);
+    }
+
+    @java.lang.Override
+    public int getSerializedSize() {
+      int size = memoizedSize;
+      if (size != -1) return size;
+
+      size = 0;
+      {
+        int dataSize = 0;
+        for (int i = 0; i < s_.size(); i++) {
+          dataSize += com.google.protobuf.CodedOutputStream
+            .computeBytesSizeNoTag(s_.get(i));
+        }
+        size += dataSize;
+        size += 1 * getSList().size();
+      }
+      {
+        int dataSize = 0;
+        for (int i = 0; i < i_.size(); i++) {
+          dataSize += com.google.protobuf.CodedOutputStream
+            .computeInt64SizeNoTag(i_.getLong(i));
+        }
+        size += dataSize;
+        if (!getIList().isEmpty()) {
+          size += 1;
+          size += com.google.protobuf.CodedOutputStream
+              .computeInt32SizeNoTag(dataSize);
+        }
+        iMemoizedSerializedSize = dataSize;
+      }
+      {
+        int dataSize = 0;
+        dataSize = 4 * getFList().size();
+        size += dataSize;
+        if (!getFList().isEmpty()) {
+          size += 1;
+          size += com.google.protobuf.CodedOutputStream
+              .computeInt32SizeNoTag(dataSize);
+        }
+        fMemoizedSerializedSize = dataSize;
+      }
+      {
+        int dataSize = 0;
+        dataSize = 1 * getBList().size();
+        size += dataSize;
+        if (!getBList().isEmpty()) {
+          size += 1;
+          size += com.google.protobuf.CodedOutputStream
+              .computeInt32SizeNoTag(dataSize);
+        }
+        bMemoizedSerializedSize = dataSize;
+      }
+      {
+        int dataSize = 0;
+        for (int i = 0; i < type_.size(); i++) {
+          dataSize += com.google.protobuf.CodedOutputStream
+            .computeEnumSizeNoTag(type_.get(i));
+        }
+        size += dataSize;
+        if (!getTypeList().isEmpty()) {  size += 1;
+          size += com.google.protobuf.CodedOutputStream
+            .computeUInt32SizeNoTag(dataSize);
+        }typeMemoizedSerializedSize = dataSize;
+      }
+      for (int i = 0; i < shape_.size(); i++) {
+        size += com.google.protobuf.CodedOutputStream
+          .computeMessageSize(7, shape_.get(i));
+      }
+      for (int i = 0; i < tensor_.size(); i++) {
+        size += com.google.protobuf.CodedOutputStream
+          .computeMessageSize(8, tensor_.get(i));
+      }
+      for (int i = 0; i < func_.size(); i++) {
+        size += com.google.protobuf.CodedOutputStream
+          .computeMessageSize(9, func_.get(i));
+      }
+      size += unknownFields.getSerializedSize();
+      memoizedSize = size;
+      return size;
+    }
+
+    @java.lang.Override
+    public boolean equals(final java.lang.Object obj) {
+      if (obj == this) {
+       return true;
+      }
+      if (!(obj instanceof org.tensorflow.proto.framework.AttrValue.ListValue)) {
+        return super.equals(obj);
+      }
+      org.tensorflow.proto.framework.AttrValue.ListValue other = (org.tensorflow.proto.framework.AttrValue.ListValue) obj;
+
+      if (!getSList()
+          .equals(other.getSList())) return false;
+      if (!getIList()
+          .equals(other.getIList())) return false;
+      if (!getFList()
+          .equals(other.getFList())) return false;
+      if (!getBList()
+          .equals(other.getBList())) return false;
+      if (!type_.equals(other.type_)) return false;
+      if (!getShapeList()
+          .equals(other.getShapeList())) return false;
+      if (!getTensorList()
+          .equals(other.getTensorList())) return false;
+      if (!getFuncList()
+          .equals(other.getFuncList())) return false;
+      if (!unknownFields.equals(other.unknownFields)) return false;
+      return true;
+    }
+
+    @java.lang.Override
+    public int hashCode() {
+      if (memoizedHashCode != 0) {
+        return memoizedHashCode;
+      }
+      int hash = 41;
+      hash = (19 * hash) + getDescriptor().hashCode();
+      if (getSCount() > 0) {
+        hash = (37 * hash) + S_FIELD_NUMBER;
+        hash = (53 * hash) + getSList().hashCode();
+      }
+      if (getICount() > 0) {
+        hash = (37 * hash) + I_FIELD_NUMBER;
+        hash = (53 * hash) + getIList().hashCode();
+      }
+      if (getFCount() > 0) {
+        hash = (37 * hash) + F_FIELD_NUMBER;
+        hash = (53 * hash) + getFList().hashCode();
+      }
+      if (getBCount() > 0) {
+        hash = (37 * hash) + B_FIELD_NUMBER;
+        hash = (53 * hash) + getBList().hashCode();
+      }
+      if (getTypeCount() > 0) {
+        hash = (37 * hash) + TYPE_FIELD_NUMBER;
+        hash = (53 * hash) + type_.hashCode();
+      }
+      if (getShapeCount() > 0) {
+        hash = (37 * hash) + SHAPE_FIELD_NUMBER;
+        hash = (53 * hash) + getShapeList().hashCode();
+      }
+      if (getTensorCount() > 0) {
+        hash = (37 * hash) + TENSOR_FIELD_NUMBER;
+        hash = (53 * hash) + getTensorList().hashCode();
+      }
+      if (getFuncCount() > 0) {
+        hash = (37 * hash) + FUNC_FIELD_NUMBER;
+        hash = (53 * hash) + getFuncList().hashCode();
+      }
+      hash = (29 * hash) + unknownFields.hashCode();
+      memoizedHashCode = hash;
+      return hash;
+    }
+
+    public static org.tensorflow.proto.framework.AttrValue.ListValue parseFrom(
+        java.nio.ByteBuffer data)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data);
+    }
+    public static org.tensorflow.proto.framework.AttrValue.ListValue parseFrom(
+        java.nio.ByteBuffer data,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.AttrValue.ListValue parseFrom(
+        com.google.protobuf.ByteString data)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data);
+    }
+    public static org.tensorflow.proto.framework.AttrValue.ListValue parseFrom(
+        com.google.protobuf.ByteString data,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.AttrValue.ListValue parseFrom(byte[] data)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data);
+    }
+    public static org.tensorflow.proto.framework.AttrValue.ListValue parseFrom(
+        byte[] data,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.AttrValue.ListValue parseFrom(java.io.InputStream input)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input);
+    }
+    public static org.tensorflow.proto.framework.AttrValue.ListValue parseFrom(
+        java.io.InputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.AttrValue.ListValue parseDelimitedFrom(java.io.InputStream input)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseDelimitedWithIOException(PARSER, input);
+    }
+    public static org.tensorflow.proto.framework.AttrValue.ListValue parseDelimitedFrom(
+        java.io.InputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.AttrValue.ListValue parseFrom(
+        com.google.protobuf.CodedInputStream input)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input);
+    }
+    public static org.tensorflow.proto.framework.AttrValue.ListValue parseFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input, extensionRegistry);
+    }
+
+    @java.lang.Override
+    public Builder newBuilderForType() { return newBuilder(); }
+    public static Builder newBuilder() {
+      return DEFAULT_INSTANCE.toBuilder();
+    }
+    public static Builder newBuilder(org.tensorflow.proto.framework.AttrValue.ListValue prototype) {
+      return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+    }
+    @java.lang.Override
+    public Builder toBuilder() {
+      return this == DEFAULT_INSTANCE
+          ? new Builder() : new Builder().mergeFrom(this);
+    }
+
+    @java.lang.Override
+    protected Builder newBuilderForType(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      Builder builder = new Builder(parent);
+      return builder;
+    }
+    /**
+     * <pre>
+     * LINT.IfChange
+     * </pre>
+     *
+     * Protobuf type {@code tensorflow.AttrValue.ListValue}
+     */
+    public static final class Builder extends
+        com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+        // @@protoc_insertion_point(builder_implements:tensorflow.AttrValue.ListValue)
+        org.tensorflow.proto.framework.AttrValue.ListValueOrBuilder {
+      public static final com.google.protobuf.Descriptors.Descriptor
+          getDescriptor() {
+        return org.tensorflow.proto.framework.AttrValueProtos.internal_static_tensorflow_AttrValue_ListValue_descriptor;
+      }
+
+      @java.lang.Override
+      protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+          internalGetFieldAccessorTable() {
+        return org.tensorflow.proto.framework.AttrValueProtos.internal_static_tensorflow_AttrValue_ListValue_fieldAccessorTable
+            .ensureFieldAccessorsInitialized(
+                org.tensorflow.proto.framework.AttrValue.ListValue.class, org.tensorflow.proto.framework.AttrValue.ListValue.Builder.class);
+      }
+
+      // Construct using org.tensorflow.proto.framework.AttrValue.ListValue.newBuilder()
+      private Builder() {
+        maybeForceBuilderInitialization();
+      }
+
+      private Builder(
+          com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+        super(parent);
+        maybeForceBuilderInitialization();
+      }
+      private void maybeForceBuilderInitialization() {
+        if (com.google.protobuf.GeneratedMessageV3
+                .alwaysUseFieldBuilders) {
+          getShapeFieldBuilder();
+          getTensorFieldBuilder();
+          getFuncFieldBuilder();
+        }
+      }
+      @java.lang.Override
+      public Builder clear() {
+        super.clear();
+        s_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000001);
+        i_ = emptyLongList();
+        bitField0_ = (bitField0_ & ~0x00000002);
+        f_ = emptyFloatList();
+        bitField0_ = (bitField0_ & ~0x00000004);
+        b_ = emptyBooleanList();
+        bitField0_ = (bitField0_ & ~0x00000008);
+        type_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000010);
+        if (shapeBuilder_ == null) {
+          shape_ = java.util.Collections.emptyList();
+          bitField0_ = (bitField0_ & ~0x00000020);
+        } else {
+          shapeBuilder_.clear();
+        }
+        if (tensorBuilder_ == null) {
+          tensor_ = java.util.Collections.emptyList();
+          bitField0_ = (bitField0_ & ~0x00000040);
+        } else {
+          tensorBuilder_.clear();
+        }
+        if (funcBuilder_ == null) {
+          func_ = java.util.Collections.emptyList();
+          bitField0_ = (bitField0_ & ~0x00000080);
+        } else {
+          funcBuilder_.clear();
+        }
+        return this;
+      }
+
+      @java.lang.Override
+      public com.google.protobuf.Descriptors.Descriptor
+          getDescriptorForType() {
+        return org.tensorflow.proto.framework.AttrValueProtos.internal_static_tensorflow_AttrValue_ListValue_descriptor;
+      }
+
+      @java.lang.Override
+      public org.tensorflow.proto.framework.AttrValue.ListValue getDefaultInstanceForType() {
+        return org.tensorflow.proto.framework.AttrValue.ListValue.getDefaultInstance();
+      }
+
+      @java.lang.Override
+      public org.tensorflow.proto.framework.AttrValue.ListValue build() {
+        org.tensorflow.proto.framework.AttrValue.ListValue result = buildPartial();
+        if (!result.isInitialized()) {
+          throw newUninitializedMessageException(result);
+        }
+        return result;
+      }
+
+      @java.lang.Override
+      public org.tensorflow.proto.framework.AttrValue.ListValue buildPartial() {
+        org.tensorflow.proto.framework.AttrValue.ListValue result = new org.tensorflow.proto.framework.AttrValue.ListValue(this);
+        int from_bitField0_ = bitField0_;
+        if (((bitField0_ & 0x00000001) != 0)) {
+          s_ = java.util.Collections.unmodifiableList(s_);
+          bitField0_ = (bitField0_ & ~0x00000001);
+        }
+        result.s_ = s_;
+        if (((bitField0_ & 0x00000002) != 0)) {
+          i_.makeImmutable();
+          bitField0_ = (bitField0_ & ~0x00000002);
+        }
+        result.i_ = i_;
+        if (((bitField0_ & 0x00000004) != 0)) {
+          f_.makeImmutable();
+          bitField0_ = (bitField0_ & ~0x00000004);
+        }
+        result.f_ = f_;
+        if (((bitField0_ & 0x00000008) != 0)) {
+          b_.makeImmutable();
+          bitField0_ = (bitField0_ & ~0x00000008);
+        }
+        result.b_ = b_;
+        if (((bitField0_ & 0x00000010) != 0)) {
+          type_ = java.util.Collections.unmodifiableList(type_);
+          bitField0_ = (bitField0_ & ~0x00000010);
+        }
+        result.type_ = type_;
+        if (shapeBuilder_ == null) {
+          if (((bitField0_ & 0x00000020) != 0)) {
+            shape_ = java.util.Collections.unmodifiableList(shape_);
+            bitField0_ = (bitField0_ & ~0x00000020);
+          }
+          result.shape_ = shape_;
+        } else {
+          result.shape_ = shapeBuilder_.build();
+        }
+        if (tensorBuilder_ == null) {
+          if (((bitField0_ & 0x00000040) != 0)) {
+            tensor_ = java.util.Collections.unmodifiableList(tensor_);
+            bitField0_ = (bitField0_ & ~0x00000040);
+          }
+          result.tensor_ = tensor_;
+        } else {
+          result.tensor_ = tensorBuilder_.build();
+        }
+        if (funcBuilder_ == null) {
+          if (((bitField0_ & 0x00000080) != 0)) {
+            func_ = java.util.Collections.unmodifiableList(func_);
+            bitField0_ = (bitField0_ & ~0x00000080);
+          }
+          result.func_ = func_;
+        } else {
+          result.func_ = funcBuilder_.build();
+        }
+        onBuilt();
+        return result;
+      }
+
+      @java.lang.Override
+      public Builder clone() {
+        return super.clone();
+      }
+      @java.lang.Override
+      public Builder setField(
+          com.google.protobuf.Descriptors.FieldDescriptor field,
+          java.lang.Object value) {
+        return super.setField(field, value);
+      }
+      @java.lang.Override
+      public Builder clearField(
+          com.google.protobuf.Descriptors.FieldDescriptor field) {
+        return super.clearField(field);
+      }
+      @java.lang.Override
+      public Builder clearOneof(
+          com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+        return super.clearOneof(oneof);
+      }
+      @java.lang.Override
+      public Builder setRepeatedField(
+          com.google.protobuf.Descriptors.FieldDescriptor field,
+          int index, java.lang.Object value) {
+        return super.setRepeatedField(field, index, value);
+      }
+      @java.lang.Override
+      public Builder addRepeatedField(
+          com.google.protobuf.Descriptors.FieldDescriptor field,
+          java.lang.Object value) {
+        return super.addRepeatedField(field, value);
+      }
+      @java.lang.Override
+      public Builder mergeFrom(com.google.protobuf.Message other) {
+        if (other instanceof org.tensorflow.proto.framework.AttrValue.ListValue) {
+          return mergeFrom((org.tensorflow.proto.framework.AttrValue.ListValue)other);
+        } else {
+          super.mergeFrom(other);
+          return this;
+        }
+      }
+
+      public Builder mergeFrom(org.tensorflow.proto.framework.AttrValue.ListValue other) {
+        if (other == org.tensorflow.proto.framework.AttrValue.ListValue.getDefaultInstance()) return this;
+        if (!other.s_.isEmpty()) {
+          if (s_.isEmpty()) {
+            s_ = other.s_;
+            bitField0_ = (bitField0_ & ~0x00000001);
+          } else {
+            ensureSIsMutable();
+            s_.addAll(other.s_);
+          }
+          onChanged();
+        }
+        if (!other.i_.isEmpty()) {
+          if (i_.isEmpty()) {
+            i_ = other.i_;
+            bitField0_ = (bitField0_ & ~0x00000002);
+          } else {
+            ensureIIsMutable();
+            i_.addAll(other.i_);
+          }
+          onChanged();
+        }
+        if (!other.f_.isEmpty()) {
+          if (f_.isEmpty()) {
+            f_ = other.f_;
+            bitField0_ = (bitField0_ & ~0x00000004);
+          } else {
+            ensureFIsMutable();
+            f_.addAll(other.f_);
+          }
+          onChanged();
+        }
+        if (!other.b_.isEmpty()) {
+          if (b_.isEmpty()) {
+            b_ = other.b_;
+            bitField0_ = (bitField0_ & ~0x00000008);
+          } else {
+            ensureBIsMutable();
+            b_.addAll(other.b_);
+          }
+          onChanged();
+        }
+        if (!other.type_.isEmpty()) {
+          if (type_.isEmpty()) {
+            type_ = other.type_;
+            bitField0_ = (bitField0_ & ~0x00000010);
+          } else {
+            ensureTypeIsMutable();
+            type_.addAll(other.type_);
+          }
+          onChanged();
+        }
+        if (shapeBuilder_ == null) {
+          if (!other.shape_.isEmpty()) {
+            if (shape_.isEmpty()) {
+              shape_ = other.shape_;
+              bitField0_ = (bitField0_ & ~0x00000020);
+            } else {
+              ensureShapeIsMutable();
+              shape_.addAll(other.shape_);
+            }
+            onChanged();
+          }
+        } else {
+          if (!other.shape_.isEmpty()) {
+            if (shapeBuilder_.isEmpty()) {
+              shapeBuilder_.dispose();
+              shapeBuilder_ = null;
+              shape_ = other.shape_;
+              bitField0_ = (bitField0_ & ~0x00000020);
+              shapeBuilder_ = 
+                com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
+                   getShapeFieldBuilder() : null;
+            } else {
+              shapeBuilder_.addAllMessages(other.shape_);
+            }
+          }
+        }
+        if (tensorBuilder_ == null) {
+          if (!other.tensor_.isEmpty()) {
+            if (tensor_.isEmpty()) {
+              tensor_ = other.tensor_;
+              bitField0_ = (bitField0_ & ~0x00000040);
+            } else {
+              ensureTensorIsMutable();
+              tensor_.addAll(other.tensor_);
+            }
+            onChanged();
+          }
+        } else {
+          if (!other.tensor_.isEmpty()) {
+            if (tensorBuilder_.isEmpty()) {
+              tensorBuilder_.dispose();
+              tensorBuilder_ = null;
+              tensor_ = other.tensor_;
+              bitField0_ = (bitField0_ & ~0x00000040);
+              tensorBuilder_ = 
+                com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
+                   getTensorFieldBuilder() : null;
+            } else {
+              tensorBuilder_.addAllMessages(other.tensor_);
+            }
+          }
+        }
+        if (funcBuilder_ == null) {
+          if (!other.func_.isEmpty()) {
+            if (func_.isEmpty()) {
+              func_ = other.func_;
+              bitField0_ = (bitField0_ & ~0x00000080);
+            } else {
+              ensureFuncIsMutable();
+              func_.addAll(other.func_);
+            }
+            onChanged();
+          }
+        } else {
+          if (!other.func_.isEmpty()) {
+            if (funcBuilder_.isEmpty()) {
+              funcBuilder_.dispose();
+              funcBuilder_ = null;
+              func_ = other.func_;
+              bitField0_ = (bitField0_ & ~0x00000080);
+              funcBuilder_ = 
+                com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
+                   getFuncFieldBuilder() : null;
+            } else {
+              funcBuilder_.addAllMessages(other.func_);
+            }
+          }
+        }
+        this.mergeUnknownFields(other.unknownFields);
+        onChanged();
+        return this;
+      }
+
+      @java.lang.Override
+      public final boolean isInitialized() {
+        return true;
+      }
+
+      @java.lang.Override
+      public Builder mergeFrom(
+          com.google.protobuf.CodedInputStream input,
+          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+          throws java.io.IOException {
+        org.tensorflow.proto.framework.AttrValue.ListValue parsedMessage = null;
+        try {
+          parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+        } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+          parsedMessage = (org.tensorflow.proto.framework.AttrValue.ListValue) e.getUnfinishedMessage();
+          throw e.unwrapIOException();
+        } finally {
+          if (parsedMessage != null) {
+            mergeFrom(parsedMessage);
+          }
+        }
+        return this;
+      }
+      private int bitField0_;
+
+      private java.util.List<com.google.protobuf.ByteString> s_ = java.util.Collections.emptyList();
+      private void ensureSIsMutable() {
+        if (!((bitField0_ & 0x00000001) != 0)) {
+          s_ = new java.util.ArrayList<com.google.protobuf.ByteString>(s_);
+          bitField0_ |= 0x00000001;
+         }
+      }
+      /**
+       * <pre>
+       * "list(string)"
+       * </pre>
+       *
+       * <code>repeated bytes s = 2;</code>
+       */
+      public java.util.List<com.google.protobuf.ByteString>
+          getSList() {
+        return ((bitField0_ & 0x00000001) != 0) ?
+                 java.util.Collections.unmodifiableList(s_) : s_;
+      }
+      /**
+       * <pre>
+       * "list(string)"
+       * </pre>
+       *
+       * <code>repeated bytes s = 2;</code>
+       */
+      public int getSCount() {
+        return s_.size();
+      }
+      /**
+       * <pre>
+       * "list(string)"
+       * </pre>
+       *
+       * <code>repeated bytes s = 2;</code>
+       */
+      public com.google.protobuf.ByteString getS(int index) {
+        return s_.get(index);
+      }
+      /**
+       * <pre>
+       * "list(string)"
+       * </pre>
+       *
+       * <code>repeated bytes s = 2;</code>
+       */
+      public Builder setS(
+          int index, com.google.protobuf.ByteString value) {
+        if (value == null) {
+    throw new NullPointerException();
+  }
+  ensureSIsMutable();
+        s_.set(index, value);
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * "list(string)"
+       * </pre>
+       *
+       * <code>repeated bytes s = 2;</code>
+       */
+      public Builder addS(com.google.protobuf.ByteString value) {
+        if (value == null) {
+    throw new NullPointerException();
+  }
+  ensureSIsMutable();
+        s_.add(value);
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * "list(string)"
+       * </pre>
+       *
+       * <code>repeated bytes s = 2;</code>
+       */
+      public Builder addAllS(
+          java.lang.Iterable<? extends com.google.protobuf.ByteString> values) {
+        ensureSIsMutable();
+        com.google.protobuf.AbstractMessageLite.Builder.addAll(
+            values, s_);
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * "list(string)"
+       * </pre>
+       *
+       * <code>repeated bytes s = 2;</code>
+       */
+      public Builder clearS() {
+        s_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000001);
+        onChanged();
+        return this;
+      }
+
+      private com.google.protobuf.Internal.LongList i_ = emptyLongList();
+      private void ensureIIsMutable() {
+        if (!((bitField0_ & 0x00000002) != 0)) {
+          i_ = mutableCopy(i_);
+          bitField0_ |= 0x00000002;
+         }
+      }
+      /**
+       * <pre>
+       * "list(int)"
+       * </pre>
+       *
+       * <code>repeated int64 i = 3 [packed = true];</code>
+       */
+      public java.util.List<java.lang.Long>
+          getIList() {
+        return ((bitField0_ & 0x00000002) != 0) ?
+                 java.util.Collections.unmodifiableList(i_) : i_;
+      }
+      /**
+       * <pre>
+       * "list(int)"
+       * </pre>
+       *
+       * <code>repeated int64 i = 3 [packed = true];</code>
+       */
+      public int getICount() {
+        return i_.size();
+      }
+      /**
+       * <pre>
+       * "list(int)"
+       * </pre>
+       *
+       * <code>repeated int64 i = 3 [packed = true];</code>
+       */
+      public long getI(int index) {
+        return i_.getLong(index);
+      }
+      /**
+       * <pre>
+       * "list(int)"
+       * </pre>
+       *
+       * <code>repeated int64 i = 3 [packed = true];</code>
+       */
+      public Builder setI(
+          int index, long value) {
+        ensureIIsMutable();
+        i_.setLong(index, value);
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * "list(int)"
+       * </pre>
+       *
+       * <code>repeated int64 i = 3 [packed = true];</code>
+       */
+      public Builder addI(long value) {
+        ensureIIsMutable();
+        i_.addLong(value);
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * "list(int)"
+       * </pre>
+       *
+       * <code>repeated int64 i = 3 [packed = true];</code>
+       */
+      public Builder addAllI(
+          java.lang.Iterable<? extends java.lang.Long> values) {
+        ensureIIsMutable();
+        com.google.protobuf.AbstractMessageLite.Builder.addAll(
+            values, i_);
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * "list(int)"
+       * </pre>
+       *
+       * <code>repeated int64 i = 3 [packed = true];</code>
+       */
+      public Builder clearI() {
+        i_ = emptyLongList();
+        bitField0_ = (bitField0_ & ~0x00000002);
+        onChanged();
+        return this;
+      }
+
+      private com.google.protobuf.Internal.FloatList f_ = emptyFloatList();
+      private void ensureFIsMutable() {
+        if (!((bitField0_ & 0x00000004) != 0)) {
+          f_ = mutableCopy(f_);
+          bitField0_ |= 0x00000004;
+         }
+      }
+      /**
+       * <pre>
+       * "list(float)"
+       * </pre>
+       *
+       * <code>repeated float f = 4 [packed = true];</code>
+       */
+      public java.util.List<java.lang.Float>
+          getFList() {
+        return ((bitField0_ & 0x00000004) != 0) ?
+                 java.util.Collections.unmodifiableList(f_) : f_;
+      }
+      /**
+       * <pre>
+       * "list(float)"
+       * </pre>
+       *
+       * <code>repeated float f = 4 [packed = true];</code>
+       */
+      public int getFCount() {
+        return f_.size();
+      }
+      /**
+       * <pre>
+       * "list(float)"
+       * </pre>
+       *
+       * <code>repeated float f = 4 [packed = true];</code>
+       */
+      public float getF(int index) {
+        return f_.getFloat(index);
+      }
+      /**
+       * <pre>
+       * "list(float)"
+       * </pre>
+       *
+       * <code>repeated float f = 4 [packed = true];</code>
+       */
+      public Builder setF(
+          int index, float value) {
+        ensureFIsMutable();
+        f_.setFloat(index, value);
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * "list(float)"
+       * </pre>
+       *
+       * <code>repeated float f = 4 [packed = true];</code>
+       */
+      public Builder addF(float value) {
+        ensureFIsMutable();
+        f_.addFloat(value);
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * "list(float)"
+       * </pre>
+       *
+       * <code>repeated float f = 4 [packed = true];</code>
+       */
+      public Builder addAllF(
+          java.lang.Iterable<? extends java.lang.Float> values) {
+        ensureFIsMutable();
+        com.google.protobuf.AbstractMessageLite.Builder.addAll(
+            values, f_);
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * "list(float)"
+       * </pre>
+       *
+       * <code>repeated float f = 4 [packed = true];</code>
+       */
+      public Builder clearF() {
+        f_ = emptyFloatList();
+        bitField0_ = (bitField0_ & ~0x00000004);
+        onChanged();
+        return this;
+      }
+
+      private com.google.protobuf.Internal.BooleanList b_ = emptyBooleanList();
+      private void ensureBIsMutable() {
+        if (!((bitField0_ & 0x00000008) != 0)) {
+          b_ = mutableCopy(b_);
+          bitField0_ |= 0x00000008;
+         }
+      }
+      /**
+       * <pre>
+       * "list(bool)"
+       * </pre>
+       *
+       * <code>repeated bool b = 5 [packed = true];</code>
+       */
+      public java.util.List<java.lang.Boolean>
+          getBList() {
+        return ((bitField0_ & 0x00000008) != 0) ?
+                 java.util.Collections.unmodifiableList(b_) : b_;
+      }
+      /**
+       * <pre>
+       * "list(bool)"
+       * </pre>
+       *
+       * <code>repeated bool b = 5 [packed = true];</code>
+       */
+      public int getBCount() {
+        return b_.size();
+      }
+      /**
+       * <pre>
+       * "list(bool)"
+       * </pre>
+       *
+       * <code>repeated bool b = 5 [packed = true];</code>
+       */
+      public boolean getB(int index) {
+        return b_.getBoolean(index);
+      }
+      /**
+       * <pre>
+       * "list(bool)"
+       * </pre>
+       *
+       * <code>repeated bool b = 5 [packed = true];</code>
+       */
+      public Builder setB(
+          int index, boolean value) {
+        ensureBIsMutable();
+        b_.setBoolean(index, value);
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * "list(bool)"
+       * </pre>
+       *
+       * <code>repeated bool b = 5 [packed = true];</code>
+       */
+      public Builder addB(boolean value) {
+        ensureBIsMutable();
+        b_.addBoolean(value);
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * "list(bool)"
+       * </pre>
+       *
+       * <code>repeated bool b = 5 [packed = true];</code>
+       */
+      public Builder addAllB(
+          java.lang.Iterable<? extends java.lang.Boolean> values) {
+        ensureBIsMutable();
+        com.google.protobuf.AbstractMessageLite.Builder.addAll(
+            values, b_);
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * "list(bool)"
+       * </pre>
+       *
+       * <code>repeated bool b = 5 [packed = true];</code>
+       */
+      public Builder clearB() {
+        b_ = emptyBooleanList();
+        bitField0_ = (bitField0_ & ~0x00000008);
+        onChanged();
+        return this;
+      }
+
+      private java.util.List<java.lang.Integer> type_ =
+        java.util.Collections.emptyList();
+      private void ensureTypeIsMutable() {
+        if (!((bitField0_ & 0x00000010) != 0)) {
+          type_ = new java.util.ArrayList<java.lang.Integer>(type_);
+          bitField0_ |= 0x00000010;
+        }
+      }
+      /**
+       * <pre>
+       * "list(type)"
+       * </pre>
+       *
+       * <code>repeated .tensorflow.DataType type = 6 [packed = true];</code>
+       */
+      public java.util.List<org.tensorflow.proto.framework.DataType> getTypeList() {
+        return new com.google.protobuf.Internal.ListAdapter<
+            java.lang.Integer, org.tensorflow.proto.framework.DataType>(type_, type_converter_);
+      }
+      /**
+       * <pre>
+       * "list(type)"
+       * </pre>
+       *
+       * <code>repeated .tensorflow.DataType type = 6 [packed = true];</code>
+       */
+      public int getTypeCount() {
+        return type_.size();
+      }
+      /**
+       * <pre>
+       * "list(type)"
+       * </pre>
+       *
+       * <code>repeated .tensorflow.DataType type = 6 [packed = true];</code>
+       */
+      public org.tensorflow.proto.framework.DataType getType(int index) {
+        return type_converter_.convert(type_.get(index));
+      }
+      /**
+       * <pre>
+       * "list(type)"
+       * </pre>
+       *
+       * <code>repeated .tensorflow.DataType type = 6 [packed = true];</code>
+       */
+      public Builder setType(
+          int index, org.tensorflow.proto.framework.DataType value) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureTypeIsMutable();
+        type_.set(index, value.getNumber());
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * "list(type)"
+       * </pre>
+       *
+       * <code>repeated .tensorflow.DataType type = 6 [packed = true];</code>
+       */
+      public Builder addType(org.tensorflow.proto.framework.DataType value) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureTypeIsMutable();
+        type_.add(value.getNumber());
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * "list(type)"
+       * </pre>
+       *
+       * <code>repeated .tensorflow.DataType type = 6 [packed = true];</code>
+       */
+      public Builder addAllType(
+          java.lang.Iterable<? extends org.tensorflow.proto.framework.DataType> values) {
+        ensureTypeIsMutable();
+        for (org.tensorflow.proto.framework.DataType value : values) {
+          type_.add(value.getNumber());
+        }
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * "list(type)"
+       * </pre>
+       *
+       * <code>repeated .tensorflow.DataType type = 6 [packed = true];</code>
+       */
+      public Builder clearType() {
+        type_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000010);
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * "list(type)"
+       * </pre>
+       *
+       * <code>repeated .tensorflow.DataType type = 6 [packed = true];</code>
+       */
+      public java.util.List<java.lang.Integer>
+      getTypeValueList() {
+        return java.util.Collections.unmodifiableList(type_);
+      }
+      /**
+       * <pre>
+       * "list(type)"
+       * </pre>
+       *
+       * <code>repeated .tensorflow.DataType type = 6 [packed = true];</code>
+       */
+      public int getTypeValue(int index) {
+        return type_.get(index);
+      }
+      /**
+       * <pre>
+       * "list(type)"
+       * </pre>
+       *
+       * <code>repeated .tensorflow.DataType type = 6 [packed = true];</code>
+       */
+      public Builder setTypeValue(
+          int index, int value) {
+        ensureTypeIsMutable();
+        type_.set(index, value);
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * "list(type)"
+       * </pre>
+       *
+       * <code>repeated .tensorflow.DataType type = 6 [packed = true];</code>
+       */
+      public Builder addTypeValue(int value) {
+        ensureTypeIsMutable();
+        type_.add(value);
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * "list(type)"
+       * </pre>
+       *
+       * <code>repeated .tensorflow.DataType type = 6 [packed = true];</code>
+       */
+      public Builder addAllTypeValue(
+          java.lang.Iterable<java.lang.Integer> values) {
+        ensureTypeIsMutable();
+        for (int value : values) {
+          type_.add(value);
+        }
+        onChanged();
+        return this;
+      }
+
+      private java.util.List<org.tensorflow.proto.framework.TensorShapeProto> shape_ =
+        java.util.Collections.emptyList();
+      private void ensureShapeIsMutable() {
+        if (!((bitField0_ & 0x00000020) != 0)) {
+          shape_ = new java.util.ArrayList<org.tensorflow.proto.framework.TensorShapeProto>(shape_);
+          bitField0_ |= 0x00000020;
+         }
+      }
+
+      private com.google.protobuf.RepeatedFieldBuilderV3<
+          org.tensorflow.proto.framework.TensorShapeProto, org.tensorflow.proto.framework.TensorShapeProto.Builder, org.tensorflow.proto.framework.TensorShapeProtoOrBuilder> shapeBuilder_;
+
+      /**
+       * <pre>
+       * "list(shape)"
+       * </pre>
+       *
+       * <code>repeated .tensorflow.TensorShapeProto shape = 7;</code>
+       */
+      public java.util.List<org.tensorflow.proto.framework.TensorShapeProto> getShapeList() {
+        if (shapeBuilder_ == null) {
+          return java.util.Collections.unmodifiableList(shape_);
+        } else {
+          return shapeBuilder_.getMessageList();
+        }
+      }
+      /**
+       * <pre>
+       * "list(shape)"
+       * </pre>
+       *
+       * <code>repeated .tensorflow.TensorShapeProto shape = 7;</code>
+       */
+      public int getShapeCount() {
+        if (shapeBuilder_ == null) {
+          return shape_.size();
+        } else {
+          return shapeBuilder_.getCount();
+        }
+      }
+      /**
+       * <pre>
+       * "list(shape)"
+       * </pre>
+       *
+       * <code>repeated .tensorflow.TensorShapeProto shape = 7;</code>
+       */
+      public org.tensorflow.proto.framework.TensorShapeProto getShape(int index) {
+        if (shapeBuilder_ == null) {
+          return shape_.get(index);
+        } else {
+          return shapeBuilder_.getMessage(index);
+        }
+      }
+      /**
+       * <pre>
+       * "list(shape)"
+       * </pre>
+       *
+       * <code>repeated .tensorflow.TensorShapeProto shape = 7;</code>
+       */
+      public Builder setShape(
+          int index, org.tensorflow.proto.framework.TensorShapeProto value) {
+        if (shapeBuilder_ == null) {
+          if (value == null) {
+            throw new NullPointerException();
+          }
+          ensureShapeIsMutable();
+          shape_.set(index, value);
+          onChanged();
+        } else {
+          shapeBuilder_.setMessage(index, value);
+        }
+        return this;
+      }
+      /**
+       * <pre>
+       * "list(shape)"
+       * </pre>
+       *
+       * <code>repeated .tensorflow.TensorShapeProto shape = 7;</code>
+       */
+      public Builder setShape(
+          int index, org.tensorflow.proto.framework.TensorShapeProto.Builder builderForValue) {
+        if (shapeBuilder_ == null) {
+          ensureShapeIsMutable();
+          shape_.set(index, builderForValue.build());
+          onChanged();
+        } else {
+          shapeBuilder_.setMessage(index, builderForValue.build());
+        }
+        return this;
+      }
+      /**
+       * <pre>
+       * "list(shape)"
+       * </pre>
+       *
+       * <code>repeated .tensorflow.TensorShapeProto shape = 7;</code>
+       */
+      public Builder addShape(org.tensorflow.proto.framework.TensorShapeProto value) {
+        if (shapeBuilder_ == null) {
+          if (value == null) {
+            throw new NullPointerException();
+          }
+          ensureShapeIsMutable();
+          shape_.add(value);
+          onChanged();
+        } else {
+          shapeBuilder_.addMessage(value);
+        }
+        return this;
+      }
+      /**
+       * <pre>
+       * "list(shape)"
+       * </pre>
+       *
+       * <code>repeated .tensorflow.TensorShapeProto shape = 7;</code>
+       */
+      public Builder addShape(
+          int index, org.tensorflow.proto.framework.TensorShapeProto value) {
+        if (shapeBuilder_ == null) {
+          if (value == null) {
+            throw new NullPointerException();
+          }
+          ensureShapeIsMutable();
+          shape_.add(index, value);
+          onChanged();
+        } else {
+          shapeBuilder_.addMessage(index, value);
+        }
+        return this;
+      }
+      /**
+       * <pre>
+       * "list(shape)"
+       * </pre>
+       *
+       * <code>repeated .tensorflow.TensorShapeProto shape = 7;</code>
+       */
+      public Builder addShape(
+          org.tensorflow.proto.framework.TensorShapeProto.Builder builderForValue) {
+        if (shapeBuilder_ == null) {
+          ensureShapeIsMutable();
+          shape_.add(builderForValue.build());
+          onChanged();
+        } else {
+          shapeBuilder_.addMessage(builderForValue.build());
+        }
+        return this;
+      }
+      /**
+       * <pre>
+       * "list(shape)"
+       * </pre>
+       *
+       * <code>repeated .tensorflow.TensorShapeProto shape = 7;</code>
+       */
+      public Builder addShape(
+          int index, org.tensorflow.proto.framework.TensorShapeProto.Builder builderForValue) {
+        if (shapeBuilder_ == null) {
+          ensureShapeIsMutable();
+          shape_.add(index, builderForValue.build());
+          onChanged();
+        } else {
+          shapeBuilder_.addMessage(index, builderForValue.build());
+        }
+        return this;
+      }
+      /**
+       * <pre>
+       * "list(shape)"
+       * </pre>
+       *
+       * <code>repeated .tensorflow.TensorShapeProto shape = 7;</code>
+       */
+      public Builder addAllShape(
+          java.lang.Iterable<? extends org.tensorflow.proto.framework.TensorShapeProto> values) {
+        if (shapeBuilder_ == null) {
+          ensureShapeIsMutable();
+          com.google.protobuf.AbstractMessageLite.Builder.addAll(
+              values, shape_);
+          onChanged();
+        } else {
+          shapeBuilder_.addAllMessages(values);
+        }
+        return this;
+      }
+      /**
+       * <pre>
+       * "list(shape)"
+       * </pre>
+       *
+       * <code>repeated .tensorflow.TensorShapeProto shape = 7;</code>
+       */
+      public Builder clearShape() {
+        if (shapeBuilder_ == null) {
+          shape_ = java.util.Collections.emptyList();
+          bitField0_ = (bitField0_ & ~0x00000020);
+          onChanged();
+        } else {
+          shapeBuilder_.clear();
+        }
+        return this;
+      }
+      /**
+       * <pre>
+       * "list(shape)"
+       * </pre>
+       *
+       * <code>repeated .tensorflow.TensorShapeProto shape = 7;</code>
+       */
+      public Builder removeShape(int index) {
+        if (shapeBuilder_ == null) {
+          ensureShapeIsMutable();
+          shape_.remove(index);
+          onChanged();
+        } else {
+          shapeBuilder_.remove(index);
+        }
+        return this;
+      }
+      /**
+       * <pre>
+       * "list(shape)"
+       * </pre>
+       *
+       * <code>repeated .tensorflow.TensorShapeProto shape = 7;</code>
+       */
+      public org.tensorflow.proto.framework.TensorShapeProto.Builder getShapeBuilder(
+          int index) {
+        return getShapeFieldBuilder().getBuilder(index);
+      }
+      /**
+       * <pre>
+       * "list(shape)"
+       * </pre>
+       *
+       * <code>repeated .tensorflow.TensorShapeProto shape = 7;</code>
+       */
+      public org.tensorflow.proto.framework.TensorShapeProtoOrBuilder getShapeOrBuilder(
+          int index) {
+        if (shapeBuilder_ == null) {
+          return shape_.get(index);  } else {
+          return shapeBuilder_.getMessageOrBuilder(index);
+        }
+      }
+      /**
+       * <pre>
+       * "list(shape)"
+       * </pre>
+       *
+       * <code>repeated .tensorflow.TensorShapeProto shape = 7;</code>
+       */
+      public java.util.List<? extends org.tensorflow.proto.framework.TensorShapeProtoOrBuilder> 
+           getShapeOrBuilderList() {
+        if (shapeBuilder_ != null) {
+          return shapeBuilder_.getMessageOrBuilderList();
+        } else {
+          return java.util.Collections.unmodifiableList(shape_);
+        }
+      }
+      /**
+       * <pre>
+       * "list(shape)"
+       * </pre>
+       *
+       * <code>repeated .tensorflow.TensorShapeProto shape = 7;</code>
+       */
+      public org.tensorflow.proto.framework.TensorShapeProto.Builder addShapeBuilder() {
+        return getShapeFieldBuilder().addBuilder(
+            org.tensorflow.proto.framework.TensorShapeProto.getDefaultInstance());
+      }
+      /**
+       * <pre>
+       * "list(shape)"
+       * </pre>
+       *
+       * <code>repeated .tensorflow.TensorShapeProto shape = 7;</code>
+       */
+      public org.tensorflow.proto.framework.TensorShapeProto.Builder addShapeBuilder(
+          int index) {
+        return getShapeFieldBuilder().addBuilder(
+            index, org.tensorflow.proto.framework.TensorShapeProto.getDefaultInstance());
+      }
+      /**
+       * <pre>
+       * "list(shape)"
+       * </pre>
+       *
+       * <code>repeated .tensorflow.TensorShapeProto shape = 7;</code>
+       */
+      public java.util.List<org.tensorflow.proto.framework.TensorShapeProto.Builder> 
+           getShapeBuilderList() {
+        return getShapeFieldBuilder().getBuilderList();
+      }
+      private com.google.protobuf.RepeatedFieldBuilderV3<
+          org.tensorflow.proto.framework.TensorShapeProto, org.tensorflow.proto.framework.TensorShapeProto.Builder, org.tensorflow.proto.framework.TensorShapeProtoOrBuilder> 
+          getShapeFieldBuilder() {
+        if (shapeBuilder_ == null) {
+          shapeBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
+              org.tensorflow.proto.framework.TensorShapeProto, org.tensorflow.proto.framework.TensorShapeProto.Builder, org.tensorflow.proto.framework.TensorShapeProtoOrBuilder>(
+                  shape_,
+                  ((bitField0_ & 0x00000020) != 0),
+                  getParentForChildren(),
+                  isClean());
+          shape_ = null;
+        }
+        return shapeBuilder_;
+      }
+
+      private java.util.List<org.tensorflow.proto.framework.TensorProto> tensor_ =
+        java.util.Collections.emptyList();
+      private void ensureTensorIsMutable() {
+        if (!((bitField0_ & 0x00000040) != 0)) {
+          tensor_ = new java.util.ArrayList<org.tensorflow.proto.framework.TensorProto>(tensor_);
+          bitField0_ |= 0x00000040;
+         }
+      }
+
+      private com.google.protobuf.RepeatedFieldBuilderV3<
+          org.tensorflow.proto.framework.TensorProto, org.tensorflow.proto.framework.TensorProto.Builder, org.tensorflow.proto.framework.TensorProtoOrBuilder> tensorBuilder_;
+
+      /**
+       * <pre>
+       * "list(tensor)"
+       * </pre>
+       *
+       * <code>repeated .tensorflow.TensorProto tensor = 8;</code>
+       */
+      public java.util.List<org.tensorflow.proto.framework.TensorProto> getTensorList() {
+        if (tensorBuilder_ == null) {
+          return java.util.Collections.unmodifiableList(tensor_);
+        } else {
+          return tensorBuilder_.getMessageList();
+        }
+      }
+      /**
+       * <pre>
+       * "list(tensor)"
+       * </pre>
+       *
+       * <code>repeated .tensorflow.TensorProto tensor = 8;</code>
+       */
+      public int getTensorCount() {
+        if (tensorBuilder_ == null) {
+          return tensor_.size();
+        } else {
+          return tensorBuilder_.getCount();
+        }
+      }
+      /**
+       * <pre>
+       * "list(tensor)"
+       * </pre>
+       *
+       * <code>repeated .tensorflow.TensorProto tensor = 8;</code>
+       */
+      public org.tensorflow.proto.framework.TensorProto getTensor(int index) {
+        if (tensorBuilder_ == null) {
+          return tensor_.get(index);
+        } else {
+          return tensorBuilder_.getMessage(index);
+        }
+      }
+      /**
+       * <pre>
+       * "list(tensor)"
+       * </pre>
+       *
+       * <code>repeated .tensorflow.TensorProto tensor = 8;</code>
+       */
+      public Builder setTensor(
+          int index, org.tensorflow.proto.framework.TensorProto value) {
+        if (tensorBuilder_ == null) {
+          if (value == null) {
+            throw new NullPointerException();
+          }
+          ensureTensorIsMutable();
+          tensor_.set(index, value);
+          onChanged();
+        } else {
+          tensorBuilder_.setMessage(index, value);
+        }
+        return this;
+      }
+      /**
+       * <pre>
+       * "list(tensor)"
+       * </pre>
+       *
+       * <code>repeated .tensorflow.TensorProto tensor = 8;</code>
+       */
+      public Builder setTensor(
+          int index, org.tensorflow.proto.framework.TensorProto.Builder builderForValue) {
+        if (tensorBuilder_ == null) {
+          ensureTensorIsMutable();
+          tensor_.set(index, builderForValue.build());
+          onChanged();
+        } else {
+          tensorBuilder_.setMessage(index, builderForValue.build());
+        }
+        return this;
+      }
+      /**
+       * <pre>
+       * "list(tensor)"
+       * </pre>
+       *
+       * <code>repeated .tensorflow.TensorProto tensor = 8;</code>
+       */
+      public Builder addTensor(org.tensorflow.proto.framework.TensorProto value) {
+        if (tensorBuilder_ == null) {
+          if (value == null) {
+            throw new NullPointerException();
+          }
+          ensureTensorIsMutable();
+          tensor_.add(value);
+          onChanged();
+        } else {
+          tensorBuilder_.addMessage(value);
+        }
+        return this;
+      }
+      /**
+       * <pre>
+       * "list(tensor)"
+       * </pre>
+       *
+       * <code>repeated .tensorflow.TensorProto tensor = 8;</code>
+       */
+      public Builder addTensor(
+          int index, org.tensorflow.proto.framework.TensorProto value) {
+        if (tensorBuilder_ == null) {
+          if (value == null) {
+            throw new NullPointerException();
+          }
+          ensureTensorIsMutable();
+          tensor_.add(index, value);
+          onChanged();
+        } else {
+          tensorBuilder_.addMessage(index, value);
+        }
+        return this;
+      }
+      /**
+       * <pre>
+       * "list(tensor)"
+       * </pre>
+       *
+       * <code>repeated .tensorflow.TensorProto tensor = 8;</code>
+       */
+      public Builder addTensor(
+          org.tensorflow.proto.framework.TensorProto.Builder builderForValue) {
+        if (tensorBuilder_ == null) {
+          ensureTensorIsMutable();
+          tensor_.add(builderForValue.build());
+          onChanged();
+        } else {
+          tensorBuilder_.addMessage(builderForValue.build());
+        }
+        return this;
+      }
+      /**
+       * <pre>
+       * "list(tensor)"
+       * </pre>
+       *
+       * <code>repeated .tensorflow.TensorProto tensor = 8;</code>
+       */
+      public Builder addTensor(
+          int index, org.tensorflow.proto.framework.TensorProto.Builder builderForValue) {
+        if (tensorBuilder_ == null) {
+          ensureTensorIsMutable();
+          tensor_.add(index, builderForValue.build());
+          onChanged();
+        } else {
+          tensorBuilder_.addMessage(index, builderForValue.build());
+        }
+        return this;
+      }
+      /**
+       * <pre>
+       * "list(tensor)"
+       * </pre>
+       *
+       * <code>repeated .tensorflow.TensorProto tensor = 8;</code>
+       */
+      public Builder addAllTensor(
+          java.lang.Iterable<? extends org.tensorflow.proto.framework.TensorProto> values) {
+        if (tensorBuilder_ == null) {
+          ensureTensorIsMutable();
+          com.google.protobuf.AbstractMessageLite.Builder.addAll(
+              values, tensor_);
+          onChanged();
+        } else {
+          tensorBuilder_.addAllMessages(values);
+        }
+        return this;
+      }
+      /**
+       * <pre>
+       * "list(tensor)"
+       * </pre>
+       *
+       * <code>repeated .tensorflow.TensorProto tensor = 8;</code>
+       */
+      public Builder clearTensor() {
+        if (tensorBuilder_ == null) {
+          tensor_ = java.util.Collections.emptyList();
+          bitField0_ = (bitField0_ & ~0x00000040);
+          onChanged();
+        } else {
+          tensorBuilder_.clear();
+        }
+        return this;
+      }
+      /**
+       * <pre>
+       * "list(tensor)"
+       * </pre>
+       *
+       * <code>repeated .tensorflow.TensorProto tensor = 8;</code>
+       */
+      public Builder removeTensor(int index) {
+        if (tensorBuilder_ == null) {
+          ensureTensorIsMutable();
+          tensor_.remove(index);
+          onChanged();
+        } else {
+          tensorBuilder_.remove(index);
+        }
+        return this;
+      }
+      /**
+       * <pre>
+       * "list(tensor)"
+       * </pre>
+       *
+       * <code>repeated .tensorflow.TensorProto tensor = 8;</code>
+       */
+      public org.tensorflow.proto.framework.TensorProto.Builder getTensorBuilder(
+          int index) {
+        return getTensorFieldBuilder().getBuilder(index);
+      }
+      /**
+       * <pre>
+       * "list(tensor)"
+       * </pre>
+       *
+       * <code>repeated .tensorflow.TensorProto tensor = 8;</code>
+       */
+      public org.tensorflow.proto.framework.TensorProtoOrBuilder getTensorOrBuilder(
+          int index) {
+        if (tensorBuilder_ == null) {
+          return tensor_.get(index);  } else {
+          return tensorBuilder_.getMessageOrBuilder(index);
+        }
+      }
+      /**
+       * <pre>
+       * "list(tensor)"
+       * </pre>
+       *
+       * <code>repeated .tensorflow.TensorProto tensor = 8;</code>
+       */
+      public java.util.List<? extends org.tensorflow.proto.framework.TensorProtoOrBuilder> 
+           getTensorOrBuilderList() {
+        if (tensorBuilder_ != null) {
+          return tensorBuilder_.getMessageOrBuilderList();
+        } else {
+          return java.util.Collections.unmodifiableList(tensor_);
+        }
+      }
+      /**
+       * <pre>
+       * "list(tensor)"
+       * </pre>
+       *
+       * <code>repeated .tensorflow.TensorProto tensor = 8;</code>
+       */
+      public org.tensorflow.proto.framework.TensorProto.Builder addTensorBuilder() {
+        return getTensorFieldBuilder().addBuilder(
+            org.tensorflow.proto.framework.TensorProto.getDefaultInstance());
+      }
+      /**
+       * <pre>
+       * "list(tensor)"
+       * </pre>
+       *
+       * <code>repeated .tensorflow.TensorProto tensor = 8;</code>
+       */
+      public org.tensorflow.proto.framework.TensorProto.Builder addTensorBuilder(
+          int index) {
+        return getTensorFieldBuilder().addBuilder(
+            index, org.tensorflow.proto.framework.TensorProto.getDefaultInstance());
+      }
+      /**
+       * <pre>
+       * "list(tensor)"
+       * </pre>
+       *
+       * <code>repeated .tensorflow.TensorProto tensor = 8;</code>
+       */
+      public java.util.List<org.tensorflow.proto.framework.TensorProto.Builder> 
+           getTensorBuilderList() {
+        return getTensorFieldBuilder().getBuilderList();
+      }
+      private com.google.protobuf.RepeatedFieldBuilderV3<
+          org.tensorflow.proto.framework.TensorProto, org.tensorflow.proto.framework.TensorProto.Builder, org.tensorflow.proto.framework.TensorProtoOrBuilder> 
+          getTensorFieldBuilder() {
+        if (tensorBuilder_ == null) {
+          tensorBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
+              org.tensorflow.proto.framework.TensorProto, org.tensorflow.proto.framework.TensorProto.Builder, org.tensorflow.proto.framework.TensorProtoOrBuilder>(
+                  tensor_,
+                  ((bitField0_ & 0x00000040) != 0),
+                  getParentForChildren(),
+                  isClean());
+          tensor_ = null;
+        }
+        return tensorBuilder_;
+      }
+
+      private java.util.List<org.tensorflow.proto.framework.NameAttrList> func_ =
+        java.util.Collections.emptyList();
+      private void ensureFuncIsMutable() {
+        if (!((bitField0_ & 0x00000080) != 0)) {
+          func_ = new java.util.ArrayList<org.tensorflow.proto.framework.NameAttrList>(func_);
+          bitField0_ |= 0x00000080;
+         }
+      }
+
+      private com.google.protobuf.RepeatedFieldBuilderV3<
+          org.tensorflow.proto.framework.NameAttrList, org.tensorflow.proto.framework.NameAttrList.Builder, org.tensorflow.proto.framework.NameAttrListOrBuilder> funcBuilder_;
+
+      /**
+       * <pre>
+       * "list(attr)"
+       * </pre>
+       *
+       * <code>repeated .tensorflow.NameAttrList func = 9;</code>
+       */
+      public java.util.List<org.tensorflow.proto.framework.NameAttrList> getFuncList() {
+        if (funcBuilder_ == null) {
+          return java.util.Collections.unmodifiableList(func_);
+        } else {
+          return funcBuilder_.getMessageList();
+        }
+      }
+      /**
+       * <pre>
+       * "list(attr)"
+       * </pre>
+       *
+       * <code>repeated .tensorflow.NameAttrList func = 9;</code>
+       */
+      public int getFuncCount() {
+        if (funcBuilder_ == null) {
+          return func_.size();
+        } else {
+          return funcBuilder_.getCount();
+        }
+      }
+      /**
+       * <pre>
+       * "list(attr)"
+       * </pre>
+       *
+       * <code>repeated .tensorflow.NameAttrList func = 9;</code>
+       */
+      public org.tensorflow.proto.framework.NameAttrList getFunc(int index) {
+        if (funcBuilder_ == null) {
+          return func_.get(index);
+        } else {
+          return funcBuilder_.getMessage(index);
+        }
+      }
+      /**
+       * <pre>
+       * "list(attr)"
+       * </pre>
+       *
+       * <code>repeated .tensorflow.NameAttrList func = 9;</code>
+       */
+      public Builder setFunc(
+          int index, org.tensorflow.proto.framework.NameAttrList value) {
+        if (funcBuilder_ == null) {
+          if (value == null) {
+            throw new NullPointerException();
+          }
+          ensureFuncIsMutable();
+          func_.set(index, value);
+          onChanged();
+        } else {
+          funcBuilder_.setMessage(index, value);
+        }
+        return this;
+      }
+      /**
+       * <pre>
+       * "list(attr)"
+       * </pre>
+       *
+       * <code>repeated .tensorflow.NameAttrList func = 9;</code>
+       */
+      public Builder setFunc(
+          int index, org.tensorflow.proto.framework.NameAttrList.Builder builderForValue) {
+        if (funcBuilder_ == null) {
+          ensureFuncIsMutable();
+          func_.set(index, builderForValue.build());
+          onChanged();
+        } else {
+          funcBuilder_.setMessage(index, builderForValue.build());
+        }
+        return this;
+      }
+      /**
+       * <pre>
+       * "list(attr)"
+       * </pre>
+       *
+       * <code>repeated .tensorflow.NameAttrList func = 9;</code>
+       */
+      public Builder addFunc(org.tensorflow.proto.framework.NameAttrList value) {
+        if (funcBuilder_ == null) {
+          if (value == null) {
+            throw new NullPointerException();
+          }
+          ensureFuncIsMutable();
+          func_.add(value);
+          onChanged();
+        } else {
+          funcBuilder_.addMessage(value);
+        }
+        return this;
+      }
+      /**
+       * <pre>
+       * "list(attr)"
+       * </pre>
+       *
+       * <code>repeated .tensorflow.NameAttrList func = 9;</code>
+       */
+      public Builder addFunc(
+          int index, org.tensorflow.proto.framework.NameAttrList value) {
+        if (funcBuilder_ == null) {
+          if (value == null) {
+            throw new NullPointerException();
+          }
+          ensureFuncIsMutable();
+          func_.add(index, value);
+          onChanged();
+        } else {
+          funcBuilder_.addMessage(index, value);
+        }
+        return this;
+      }
+      /**
+       * <pre>
+       * "list(attr)"
+       * </pre>
+       *
+       * <code>repeated .tensorflow.NameAttrList func = 9;</code>
+       */
+      public Builder addFunc(
+          org.tensorflow.proto.framework.NameAttrList.Builder builderForValue) {
+        if (funcBuilder_ == null) {
+          ensureFuncIsMutable();
+          func_.add(builderForValue.build());
+          onChanged();
+        } else {
+          funcBuilder_.addMessage(builderForValue.build());
+        }
+        return this;
+      }
+      /**
+       * <pre>
+       * "list(attr)"
+       * </pre>
+       *
+       * <code>repeated .tensorflow.NameAttrList func = 9;</code>
+       */
+      public Builder addFunc(
+          int index, org.tensorflow.proto.framework.NameAttrList.Builder builderForValue) {
+        if (funcBuilder_ == null) {
+          ensureFuncIsMutable();
+          func_.add(index, builderForValue.build());
+          onChanged();
+        } else {
+          funcBuilder_.addMessage(index, builderForValue.build());
+        }
+        return this;
+      }
+      /**
+       * <pre>
+       * "list(attr)"
+       * </pre>
+       *
+       * <code>repeated .tensorflow.NameAttrList func = 9;</code>
+       */
+      public Builder addAllFunc(
+          java.lang.Iterable<? extends org.tensorflow.proto.framework.NameAttrList> values) {
+        if (funcBuilder_ == null) {
+          ensureFuncIsMutable();
+          com.google.protobuf.AbstractMessageLite.Builder.addAll(
+              values, func_);
+          onChanged();
+        } else {
+          funcBuilder_.addAllMessages(values);
+        }
+        return this;
+      }
+      /**
+       * <pre>
+       * "list(attr)"
+       * </pre>
+       *
+       * <code>repeated .tensorflow.NameAttrList func = 9;</code>
+       */
+      public Builder clearFunc() {
+        if (funcBuilder_ == null) {
+          func_ = java.util.Collections.emptyList();
+          bitField0_ = (bitField0_ & ~0x00000080);
+          onChanged();
+        } else {
+          funcBuilder_.clear();
+        }
+        return this;
+      }
+      /**
+       * <pre>
+       * "list(attr)"
+       * </pre>
+       *
+       * <code>repeated .tensorflow.NameAttrList func = 9;</code>
+       */
+      public Builder removeFunc(int index) {
+        if (funcBuilder_ == null) {
+          ensureFuncIsMutable();
+          func_.remove(index);
+          onChanged();
+        } else {
+          funcBuilder_.remove(index);
+        }
+        return this;
+      }
+      /**
+       * <pre>
+       * "list(attr)"
+       * </pre>
+       *
+       * <code>repeated .tensorflow.NameAttrList func = 9;</code>
+       */
+      public org.tensorflow.proto.framework.NameAttrList.Builder getFuncBuilder(
+          int index) {
+        return getFuncFieldBuilder().getBuilder(index);
+      }
+      /**
+       * <pre>
+       * "list(attr)"
+       * </pre>
+       *
+       * <code>repeated .tensorflow.NameAttrList func = 9;</code>
+       */
+      public org.tensorflow.proto.framework.NameAttrListOrBuilder getFuncOrBuilder(
+          int index) {
+        if (funcBuilder_ == null) {
+          return func_.get(index);  } else {
+          return funcBuilder_.getMessageOrBuilder(index);
+        }
+      }
+      /**
+       * <pre>
+       * "list(attr)"
+       * </pre>
+       *
+       * <code>repeated .tensorflow.NameAttrList func = 9;</code>
+       */
+      public java.util.List<? extends org.tensorflow.proto.framework.NameAttrListOrBuilder> 
+           getFuncOrBuilderList() {
+        if (funcBuilder_ != null) {
+          return funcBuilder_.getMessageOrBuilderList();
+        } else {
+          return java.util.Collections.unmodifiableList(func_);
+        }
+      }
+      /**
+       * <pre>
+       * "list(attr)"
+       * </pre>
+       *
+       * <code>repeated .tensorflow.NameAttrList func = 9;</code>
+       */
+      public org.tensorflow.proto.framework.NameAttrList.Builder addFuncBuilder() {
+        return getFuncFieldBuilder().addBuilder(
+            org.tensorflow.proto.framework.NameAttrList.getDefaultInstance());
+      }
+      /**
+       * <pre>
+       * "list(attr)"
+       * </pre>
+       *
+       * <code>repeated .tensorflow.NameAttrList func = 9;</code>
+       */
+      public org.tensorflow.proto.framework.NameAttrList.Builder addFuncBuilder(
+          int index) {
+        return getFuncFieldBuilder().addBuilder(
+            index, org.tensorflow.proto.framework.NameAttrList.getDefaultInstance());
+      }
+      /**
+       * <pre>
+       * "list(attr)"
+       * </pre>
+       *
+       * <code>repeated .tensorflow.NameAttrList func = 9;</code>
+       */
+      public java.util.List<org.tensorflow.proto.framework.NameAttrList.Builder> 
+           getFuncBuilderList() {
+        return getFuncFieldBuilder().getBuilderList();
+      }
+      private com.google.protobuf.RepeatedFieldBuilderV3<
+          org.tensorflow.proto.framework.NameAttrList, org.tensorflow.proto.framework.NameAttrList.Builder, org.tensorflow.proto.framework.NameAttrListOrBuilder> 
+          getFuncFieldBuilder() {
+        if (funcBuilder_ == null) {
+          funcBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
+              org.tensorflow.proto.framework.NameAttrList, org.tensorflow.proto.framework.NameAttrList.Builder, org.tensorflow.proto.framework.NameAttrListOrBuilder>(
+                  func_,
+                  ((bitField0_ & 0x00000080) != 0),
+                  getParentForChildren(),
+                  isClean());
+          func_ = null;
+        }
+        return funcBuilder_;
+      }
+      @java.lang.Override
+      public final Builder setUnknownFields(
+          final com.google.protobuf.UnknownFieldSet unknownFields) {
+        return super.setUnknownFields(unknownFields);
+      }
+
+      @java.lang.Override
+      public final Builder mergeUnknownFields(
+          final com.google.protobuf.UnknownFieldSet unknownFields) {
+        return super.mergeUnknownFields(unknownFields);
+      }
+
+
+      // @@protoc_insertion_point(builder_scope:tensorflow.AttrValue.ListValue)
+    }
+
+    // @@protoc_insertion_point(class_scope:tensorflow.AttrValue.ListValue)
+    private static final org.tensorflow.proto.framework.AttrValue.ListValue DEFAULT_INSTANCE;
+    static {
+      DEFAULT_INSTANCE = new org.tensorflow.proto.framework.AttrValue.ListValue();
+    }
+
+    public static org.tensorflow.proto.framework.AttrValue.ListValue getDefaultInstance() {
+      return DEFAULT_INSTANCE;
+    }
+
+    private static final com.google.protobuf.Parser<ListValue>
+        PARSER = new com.google.protobuf.AbstractParser<ListValue>() {
+      @java.lang.Override
+      public ListValue parsePartialFrom(
+          com.google.protobuf.CodedInputStream input,
+          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+          throws com.google.protobuf.InvalidProtocolBufferException {
+        return new ListValue(input, extensionRegistry);
+      }
+    };
+
+    public static com.google.protobuf.Parser<ListValue> parser() {
+      return PARSER;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Parser<ListValue> getParserForType() {
+      return PARSER;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.AttrValue.ListValue getDefaultInstanceForType() {
+      return DEFAULT_INSTANCE;
+    }
+
+  }
+
+  private int valueCase_ = 0;
+  private java.lang.Object value_;
+  public enum ValueCase
+      implements com.google.protobuf.Internal.EnumLite {
+    S(2),
+    I(3),
+    F(4),
+    B(5),
+    TYPE(6),
+    SHAPE(7),
+    TENSOR(8),
+    LIST(1),
+    FUNC(10),
+    PLACEHOLDER(9),
+    VALUE_NOT_SET(0);
+    private final int value;
+    private ValueCase(int value) {
+      this.value = value;
+    }
+    /**
+     * @deprecated Use {@link #forNumber(int)} instead.
+     */
+    @java.lang.Deprecated
+    public static ValueCase valueOf(int value) {
+      return forNumber(value);
+    }
+
+    public static ValueCase forNumber(int value) {
+      switch (value) {
+        case 2: return S;
+        case 3: return I;
+        case 4: return F;
+        case 5: return B;
+        case 6: return TYPE;
+        case 7: return SHAPE;
+        case 8: return TENSOR;
+        case 1: return LIST;
+        case 10: return FUNC;
+        case 9: return PLACEHOLDER;
+        case 0: return VALUE_NOT_SET;
+        default: return null;
+      }
+    }
+    public int getNumber() {
+      return this.value;
+    }
+  };
+
+  public ValueCase
+  getValueCase() {
+    return ValueCase.forNumber(
+        valueCase_);
+  }
+
+  public static final int S_FIELD_NUMBER = 2;
+  /**
+   * <pre>
+   * "string"
+   * </pre>
+   *
+   * <code>bytes s = 2;</code>
+   */
+  public com.google.protobuf.ByteString getS() {
+    if (valueCase_ == 2) {
+      return (com.google.protobuf.ByteString) value_;
+    }
+    return com.google.protobuf.ByteString.EMPTY;
+  }
+
+  public static final int I_FIELD_NUMBER = 3;
+  /**
+   * <pre>
+   * "int"
+   * </pre>
+   *
+   * <code>int64 i = 3;</code>
+   */
+  public long getI() {
+    if (valueCase_ == 3) {
+      return (java.lang.Long) value_;
+    }
+    return 0L;
+  }
+
+  public static final int F_FIELD_NUMBER = 4;
+  /**
+   * <pre>
+   * "float"
+   * </pre>
+   *
+   * <code>float f = 4;</code>
+   */
+  public float getF() {
+    if (valueCase_ == 4) {
+      return (java.lang.Float) value_;
+    }
+    return 0F;
+  }
+
+  public static final int B_FIELD_NUMBER = 5;
+  /**
+   * <pre>
+   * "bool"
+   * </pre>
+   *
+   * <code>bool b = 5;</code>
+   */
+  public boolean getB() {
+    if (valueCase_ == 5) {
+      return (java.lang.Boolean) value_;
+    }
+    return false;
+  }
+
+  public static final int TYPE_FIELD_NUMBER = 6;
+  /**
+   * <pre>
+   * "type"
+   * </pre>
+   *
+   * <code>.tensorflow.DataType type = 6;</code>
+   */
+  public int getTypeValue() {
+    if (valueCase_ == 6) {
+      return (java.lang.Integer) value_;
+    }
+    return 0;
+  }
+  /**
+   * <pre>
+   * "type"
+   * </pre>
+   *
+   * <code>.tensorflow.DataType type = 6;</code>
+   */
+  public org.tensorflow.proto.framework.DataType getType() {
+    if (valueCase_ == 6) {
+      @SuppressWarnings("deprecation")
+      org.tensorflow.proto.framework.DataType result = org.tensorflow.proto.framework.DataType.valueOf(
+          (java.lang.Integer) value_);
+      return result == null ? org.tensorflow.proto.framework.DataType.UNRECOGNIZED : result;
+    }
+    return org.tensorflow.proto.framework.DataType.DT_INVALID;
+  }
+
+  public static final int SHAPE_FIELD_NUMBER = 7;
+  /**
+   * <pre>
+   * "shape"
+   * </pre>
+   *
+   * <code>.tensorflow.TensorShapeProto shape = 7;</code>
+   */
+  public boolean hasShape() {
+    return valueCase_ == 7;
+  }
+  /**
+   * <pre>
+   * "shape"
+   * </pre>
+   *
+   * <code>.tensorflow.TensorShapeProto shape = 7;</code>
+   */
+  public org.tensorflow.proto.framework.TensorShapeProto getShape() {
+    if (valueCase_ == 7) {
+       return (org.tensorflow.proto.framework.TensorShapeProto) value_;
+    }
+    return org.tensorflow.proto.framework.TensorShapeProto.getDefaultInstance();
+  }
+  /**
+   * <pre>
+   * "shape"
+   * </pre>
+   *
+   * <code>.tensorflow.TensorShapeProto shape = 7;</code>
+   */
+  public org.tensorflow.proto.framework.TensorShapeProtoOrBuilder getShapeOrBuilder() {
+    if (valueCase_ == 7) {
+       return (org.tensorflow.proto.framework.TensorShapeProto) value_;
+    }
+    return org.tensorflow.proto.framework.TensorShapeProto.getDefaultInstance();
+  }
+
+  public static final int TENSOR_FIELD_NUMBER = 8;
+  /**
+   * <pre>
+   * "tensor"
+   * </pre>
+   *
+   * <code>.tensorflow.TensorProto tensor = 8;</code>
+   */
+  public boolean hasTensor() {
+    return valueCase_ == 8;
+  }
+  /**
+   * <pre>
+   * "tensor"
+   * </pre>
+   *
+   * <code>.tensorflow.TensorProto tensor = 8;</code>
+   */
+  public org.tensorflow.proto.framework.TensorProto getTensor() {
+    if (valueCase_ == 8) {
+       return (org.tensorflow.proto.framework.TensorProto) value_;
+    }
+    return org.tensorflow.proto.framework.TensorProto.getDefaultInstance();
+  }
+  /**
+   * <pre>
+   * "tensor"
+   * </pre>
+   *
+   * <code>.tensorflow.TensorProto tensor = 8;</code>
+   */
+  public org.tensorflow.proto.framework.TensorProtoOrBuilder getTensorOrBuilder() {
+    if (valueCase_ == 8) {
+       return (org.tensorflow.proto.framework.TensorProto) value_;
+    }
+    return org.tensorflow.proto.framework.TensorProto.getDefaultInstance();
+  }
+
+  public static final int LIST_FIELD_NUMBER = 1;
+  /**
+   * <pre>
+   * any "list(...)"
+   * </pre>
+   *
+   * <code>.tensorflow.AttrValue.ListValue list = 1;</code>
+   */
+  public boolean hasList() {
+    return valueCase_ == 1;
+  }
+  /**
+   * <pre>
+   * any "list(...)"
+   * </pre>
+   *
+   * <code>.tensorflow.AttrValue.ListValue list = 1;</code>
+   */
+  public org.tensorflow.proto.framework.AttrValue.ListValue getList() {
+    if (valueCase_ == 1) {
+       return (org.tensorflow.proto.framework.AttrValue.ListValue) value_;
+    }
+    return org.tensorflow.proto.framework.AttrValue.ListValue.getDefaultInstance();
+  }
+  /**
+   * <pre>
+   * any "list(...)"
+   * </pre>
+   *
+   * <code>.tensorflow.AttrValue.ListValue list = 1;</code>
+   */
+  public org.tensorflow.proto.framework.AttrValue.ListValueOrBuilder getListOrBuilder() {
+    if (valueCase_ == 1) {
+       return (org.tensorflow.proto.framework.AttrValue.ListValue) value_;
+    }
+    return org.tensorflow.proto.framework.AttrValue.ListValue.getDefaultInstance();
+  }
+
+  public static final int FUNC_FIELD_NUMBER = 10;
+  /**
+   * <pre>
+   * "func" represents a function. func.name is a function's name or
+   * a primitive op's name. func.attr.first is the name of an attr
+   * defined for that function. func.attr.second is the value for
+   * that attr in the instantiation.
+   * </pre>
+   *
+   * <code>.tensorflow.NameAttrList func = 10;</code>
+   */
+  public boolean hasFunc() {
+    return valueCase_ == 10;
+  }
+  /**
+   * <pre>
+   * "func" represents a function. func.name is a function's name or
+   * a primitive op's name. func.attr.first is the name of an attr
+   * defined for that function. func.attr.second is the value for
+   * that attr in the instantiation.
+   * </pre>
+   *
+   * <code>.tensorflow.NameAttrList func = 10;</code>
+   */
+  public org.tensorflow.proto.framework.NameAttrList getFunc() {
+    if (valueCase_ == 10) {
+       return (org.tensorflow.proto.framework.NameAttrList) value_;
+    }
+    return org.tensorflow.proto.framework.NameAttrList.getDefaultInstance();
+  }
+  /**
+   * <pre>
+   * "func" represents a function. func.name is a function's name or
+   * a primitive op's name. func.attr.first is the name of an attr
+   * defined for that function. func.attr.second is the value for
+   * that attr in the instantiation.
+   * </pre>
+   *
+   * <code>.tensorflow.NameAttrList func = 10;</code>
+   */
+  public org.tensorflow.proto.framework.NameAttrListOrBuilder getFuncOrBuilder() {
+    if (valueCase_ == 10) {
+       return (org.tensorflow.proto.framework.NameAttrList) value_;
+    }
+    return org.tensorflow.proto.framework.NameAttrList.getDefaultInstance();
+  }
+
+  public static final int PLACEHOLDER_FIELD_NUMBER = 9;
+  /**
+   * <pre>
+   * This is a placeholder only used in nodes defined inside a
+   * function.  It indicates the attr value will be supplied when
+   * the function is instantiated.  For example, let us suppose a
+   * node "N" in function "FN". "N" has an attr "A" with value
+   * placeholder = "foo". When FN is instantiated with attr "foo"
+   * set to "bar", the instantiated node N's attr A will have been
+   * given the value "bar".
+   * </pre>
+   *
+   * <code>string placeholder = 9;</code>
+   */
+  public java.lang.String getPlaceholder() {
+    java.lang.Object ref = "";
+    if (valueCase_ == 9) {
+      ref = value_;
+    }
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      if (valueCase_ == 9) {
+        value_ = s;
+      }
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * This is a placeholder only used in nodes defined inside a
+   * function.  It indicates the attr value will be supplied when
+   * the function is instantiated.  For example, let us suppose a
+   * node "N" in function "FN". "N" has an attr "A" with value
+   * placeholder = "foo". When FN is instantiated with attr "foo"
+   * set to "bar", the instantiated node N's attr A will have been
+   * given the value "bar".
+   * </pre>
+   *
+   * <code>string placeholder = 9;</code>
+   */
+  public com.google.protobuf.ByteString
+      getPlaceholderBytes() {
+    java.lang.Object ref = "";
+    if (valueCase_ == 9) {
+      ref = value_;
+    }
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      if (valueCase_ == 9) {
+        value_ = b;
+      }
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    if (valueCase_ == 1) {
+      output.writeMessage(1, (org.tensorflow.proto.framework.AttrValue.ListValue) value_);
+    }
+    if (valueCase_ == 2) {
+      output.writeBytes(
+          2, (com.google.protobuf.ByteString) value_);
+    }
+    if (valueCase_ == 3) {
+      output.writeInt64(
+          3, (long)((java.lang.Long) value_));
+    }
+    if (valueCase_ == 4) {
+      output.writeFloat(
+          4, (float)((java.lang.Float) value_));
+    }
+    if (valueCase_ == 5) {
+      output.writeBool(
+          5, (boolean)((java.lang.Boolean) value_));
+    }
+    if (valueCase_ == 6) {
+      output.writeEnum(6, ((java.lang.Integer) value_));
+    }
+    if (valueCase_ == 7) {
+      output.writeMessage(7, (org.tensorflow.proto.framework.TensorShapeProto) value_);
+    }
+    if (valueCase_ == 8) {
+      output.writeMessage(8, (org.tensorflow.proto.framework.TensorProto) value_);
+    }
+    if (valueCase_ == 9) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 9, value_);
+    }
+    if (valueCase_ == 10) {
+      output.writeMessage(10, (org.tensorflow.proto.framework.NameAttrList) value_);
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (valueCase_ == 1) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(1, (org.tensorflow.proto.framework.AttrValue.ListValue) value_);
+    }
+    if (valueCase_ == 2) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeBytesSize(
+            2, (com.google.protobuf.ByteString) value_);
+    }
+    if (valueCase_ == 3) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt64Size(
+            3, (long)((java.lang.Long) value_));
+    }
+    if (valueCase_ == 4) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeFloatSize(
+            4, (float)((java.lang.Float) value_));
+    }
+    if (valueCase_ == 5) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeBoolSize(
+            5, (boolean)((java.lang.Boolean) value_));
+    }
+    if (valueCase_ == 6) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeEnumSize(6, ((java.lang.Integer) value_));
+    }
+    if (valueCase_ == 7) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(7, (org.tensorflow.proto.framework.TensorShapeProto) value_);
+    }
+    if (valueCase_ == 8) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(8, (org.tensorflow.proto.framework.TensorProto) value_);
+    }
+    if (valueCase_ == 9) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(9, value_);
+    }
+    if (valueCase_ == 10) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(10, (org.tensorflow.proto.framework.NameAttrList) value_);
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.framework.AttrValue)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.framework.AttrValue other = (org.tensorflow.proto.framework.AttrValue) obj;
+
+    if (!getValueCase().equals(other.getValueCase())) return false;
+    switch (valueCase_) {
+      case 2:
+        if (!getS()
+            .equals(other.getS())) return false;
+        break;
+      case 3:
+        if (getI()
+            != other.getI()) return false;
+        break;
+      case 4:
+        if (java.lang.Float.floatToIntBits(getF())
+            != java.lang.Float.floatToIntBits(
+                other.getF())) return false;
+        break;
+      case 5:
+        if (getB()
+            != other.getB()) return false;
+        break;
+      case 6:
+        if (getTypeValue()
+            != other.getTypeValue()) return false;
+        break;
+      case 7:
+        if (!getShape()
+            .equals(other.getShape())) return false;
+        break;
+      case 8:
+        if (!getTensor()
+            .equals(other.getTensor())) return false;
+        break;
+      case 1:
+        if (!getList()
+            .equals(other.getList())) return false;
+        break;
+      case 10:
+        if (!getFunc()
+            .equals(other.getFunc())) return false;
+        break;
+      case 9:
+        if (!getPlaceholder()
+            .equals(other.getPlaceholder())) return false;
+        break;
+      case 0:
+      default:
+    }
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    switch (valueCase_) {
+      case 2:
+        hash = (37 * hash) + S_FIELD_NUMBER;
+        hash = (53 * hash) + getS().hashCode();
+        break;
+      case 3:
+        hash = (37 * hash) + I_FIELD_NUMBER;
+        hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+            getI());
+        break;
+      case 4:
+        hash = (37 * hash) + F_FIELD_NUMBER;
+        hash = (53 * hash) + java.lang.Float.floatToIntBits(
+            getF());
+        break;
+      case 5:
+        hash = (37 * hash) + B_FIELD_NUMBER;
+        hash = (53 * hash) + com.google.protobuf.Internal.hashBoolean(
+            getB());
+        break;
+      case 6:
+        hash = (37 * hash) + TYPE_FIELD_NUMBER;
+        hash = (53 * hash) + getTypeValue();
+        break;
+      case 7:
+        hash = (37 * hash) + SHAPE_FIELD_NUMBER;
+        hash = (53 * hash) + getShape().hashCode();
+        break;
+      case 8:
+        hash = (37 * hash) + TENSOR_FIELD_NUMBER;
+        hash = (53 * hash) + getTensor().hashCode();
+        break;
+      case 1:
+        hash = (37 * hash) + LIST_FIELD_NUMBER;
+        hash = (53 * hash) + getList().hashCode();
+        break;
+      case 10:
+        hash = (37 * hash) + FUNC_FIELD_NUMBER;
+        hash = (53 * hash) + getFunc().hashCode();
+        break;
+      case 9:
+        hash = (37 * hash) + PLACEHOLDER_FIELD_NUMBER;
+        hash = (53 * hash) + getPlaceholder().hashCode();
+        break;
+      case 0:
+      default:
+    }
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.framework.AttrValue parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.AttrValue parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.AttrValue parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.AttrValue parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.AttrValue parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.AttrValue parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.AttrValue parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.AttrValue parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.AttrValue parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.AttrValue parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.AttrValue parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.AttrValue parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.framework.AttrValue prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * <pre>
+   * Protocol buffer representing the value for an attr used to configure an Op.
+   * Comment indicates the corresponding attr type.  Only the field matching the
+   * attr type may be filled.
+   * </pre>
+   *
+   * Protobuf type {@code tensorflow.AttrValue}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.AttrValue)
+      org.tensorflow.proto.framework.AttrValueOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.AttrValueProtos.internal_static_tensorflow_AttrValue_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.AttrValueProtos.internal_static_tensorflow_AttrValue_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.AttrValue.class, org.tensorflow.proto.framework.AttrValue.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.framework.AttrValue.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      valueCase_ = 0;
+      value_ = null;
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.framework.AttrValueProtos.internal_static_tensorflow_AttrValue_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.AttrValue getDefaultInstanceForType() {
+      return org.tensorflow.proto.framework.AttrValue.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.AttrValue build() {
+      org.tensorflow.proto.framework.AttrValue result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.AttrValue buildPartial() {
+      org.tensorflow.proto.framework.AttrValue result = new org.tensorflow.proto.framework.AttrValue(this);
+      if (valueCase_ == 2) {
+        result.value_ = value_;
+      }
+      if (valueCase_ == 3) {
+        result.value_ = value_;
+      }
+      if (valueCase_ == 4) {
+        result.value_ = value_;
+      }
+      if (valueCase_ == 5) {
+        result.value_ = value_;
+      }
+      if (valueCase_ == 6) {
+        result.value_ = value_;
+      }
+      if (valueCase_ == 7) {
+        if (shapeBuilder_ == null) {
+          result.value_ = value_;
+        } else {
+          result.value_ = shapeBuilder_.build();
+        }
+      }
+      if (valueCase_ == 8) {
+        if (tensorBuilder_ == null) {
+          result.value_ = value_;
+        } else {
+          result.value_ = tensorBuilder_.build();
+        }
+      }
+      if (valueCase_ == 1) {
+        if (listBuilder_ == null) {
+          result.value_ = value_;
+        } else {
+          result.value_ = listBuilder_.build();
+        }
+      }
+      if (valueCase_ == 10) {
+        if (funcBuilder_ == null) {
+          result.value_ = value_;
+        } else {
+          result.value_ = funcBuilder_.build();
+        }
+      }
+      if (valueCase_ == 9) {
+        result.value_ = value_;
+      }
+      result.valueCase_ = valueCase_;
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.framework.AttrValue) {
+        return mergeFrom((org.tensorflow.proto.framework.AttrValue)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.framework.AttrValue other) {
+      if (other == org.tensorflow.proto.framework.AttrValue.getDefaultInstance()) return this;
+      switch (other.getValueCase()) {
+        case S: {
+          setS(other.getS());
+          break;
+        }
+        case I: {
+          setI(other.getI());
+          break;
+        }
+        case F: {
+          setF(other.getF());
+          break;
+        }
+        case B: {
+          setB(other.getB());
+          break;
+        }
+        case TYPE: {
+          setTypeValue(other.getTypeValue());
+          break;
+        }
+        case SHAPE: {
+          mergeShape(other.getShape());
+          break;
+        }
+        case TENSOR: {
+          mergeTensor(other.getTensor());
+          break;
+        }
+        case LIST: {
+          mergeList(other.getList());
+          break;
+        }
+        case FUNC: {
+          mergeFunc(other.getFunc());
+          break;
+        }
+        case PLACEHOLDER: {
+          valueCase_ = 9;
+          value_ = other.value_;
+          onChanged();
+          break;
+        }
+        case VALUE_NOT_SET: {
+          break;
+        }
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.framework.AttrValue parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.framework.AttrValue) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+    private int valueCase_ = 0;
+    private java.lang.Object value_;
+    public ValueCase
+        getValueCase() {
+      return ValueCase.forNumber(
+          valueCase_);
+    }
+
+    public Builder clearValue() {
+      valueCase_ = 0;
+      value_ = null;
+      onChanged();
+      return this;
+    }
+
+
+    /**
+     * <pre>
+     * "string"
+     * </pre>
+     *
+     * <code>bytes s = 2;</code>
+     */
+    public com.google.protobuf.ByteString getS() {
+      if (valueCase_ == 2) {
+        return (com.google.protobuf.ByteString) value_;
+      }
+      return com.google.protobuf.ByteString.EMPTY;
+    }
+    /**
+     * <pre>
+     * "string"
+     * </pre>
+     *
+     * <code>bytes s = 2;</code>
+     */
+    public Builder setS(com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  valueCase_ = 2;
+      value_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * "string"
+     * </pre>
+     *
+     * <code>bytes s = 2;</code>
+     */
+    public Builder clearS() {
+      if (valueCase_ == 2) {
+        valueCase_ = 0;
+        value_ = null;
+        onChanged();
+      }
+      return this;
+    }
+
+    /**
+     * <pre>
+     * "int"
+     * </pre>
+     *
+     * <code>int64 i = 3;</code>
+     */
+    public long getI() {
+      if (valueCase_ == 3) {
+        return (java.lang.Long) value_;
+      }
+      return 0L;
+    }
+    /**
+     * <pre>
+     * "int"
+     * </pre>
+     *
+     * <code>int64 i = 3;</code>
+     */
+    public Builder setI(long value) {
+      valueCase_ = 3;
+      value_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * "int"
+     * </pre>
+     *
+     * <code>int64 i = 3;</code>
+     */
+    public Builder clearI() {
+      if (valueCase_ == 3) {
+        valueCase_ = 0;
+        value_ = null;
+        onChanged();
+      }
+      return this;
+    }
+
+    /**
+     * <pre>
+     * "float"
+     * </pre>
+     *
+     * <code>float f = 4;</code>
+     */
+    public float getF() {
+      if (valueCase_ == 4) {
+        return (java.lang.Float) value_;
+      }
+      return 0F;
+    }
+    /**
+     * <pre>
+     * "float"
+     * </pre>
+     *
+     * <code>float f = 4;</code>
+     */
+    public Builder setF(float value) {
+      valueCase_ = 4;
+      value_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * "float"
+     * </pre>
+     *
+     * <code>float f = 4;</code>
+     */
+    public Builder clearF() {
+      if (valueCase_ == 4) {
+        valueCase_ = 0;
+        value_ = null;
+        onChanged();
+      }
+      return this;
+    }
+
+    /**
+     * <pre>
+     * "bool"
+     * </pre>
+     *
+     * <code>bool b = 5;</code>
+     */
+    public boolean getB() {
+      if (valueCase_ == 5) {
+        return (java.lang.Boolean) value_;
+      }
+      return false;
+    }
+    /**
+     * <pre>
+     * "bool"
+     * </pre>
+     *
+     * <code>bool b = 5;</code>
+     */
+    public Builder setB(boolean value) {
+      valueCase_ = 5;
+      value_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * "bool"
+     * </pre>
+     *
+     * <code>bool b = 5;</code>
+     */
+    public Builder clearB() {
+      if (valueCase_ == 5) {
+        valueCase_ = 0;
+        value_ = null;
+        onChanged();
+      }
+      return this;
+    }
+
+    /**
+     * <pre>
+     * "type"
+     * </pre>
+     *
+     * <code>.tensorflow.DataType type = 6;</code>
+     */
+    public int getTypeValue() {
+      if (valueCase_ == 6) {
+        return ((java.lang.Integer) value_).intValue();
+      }
+      return 0;
+    }
+    /**
+     * <pre>
+     * "type"
+     * </pre>
+     *
+     * <code>.tensorflow.DataType type = 6;</code>
+     */
+    public Builder setTypeValue(int value) {
+      valueCase_ = 6;
+      value_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * "type"
+     * </pre>
+     *
+     * <code>.tensorflow.DataType type = 6;</code>
+     */
+    public org.tensorflow.proto.framework.DataType getType() {
+      if (valueCase_ == 6) {
+        @SuppressWarnings("deprecation")
+        org.tensorflow.proto.framework.DataType result = org.tensorflow.proto.framework.DataType.valueOf(
+            (java.lang.Integer) value_);
+        return result == null ? org.tensorflow.proto.framework.DataType.UNRECOGNIZED : result;
+      }
+      return org.tensorflow.proto.framework.DataType.DT_INVALID;
+    }
+    /**
+     * <pre>
+     * "type"
+     * </pre>
+     *
+     * <code>.tensorflow.DataType type = 6;</code>
+     */
+    public Builder setType(org.tensorflow.proto.framework.DataType value) {
+      if (value == null) {
+        throw new NullPointerException();
+      }
+      valueCase_ = 6;
+      value_ = value.getNumber();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * "type"
+     * </pre>
+     *
+     * <code>.tensorflow.DataType type = 6;</code>
+     */
+    public Builder clearType() {
+      if (valueCase_ == 6) {
+        valueCase_ = 0;
+        value_ = null;
+        onChanged();
+      }
+      return this;
+    }
+
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.TensorShapeProto, org.tensorflow.proto.framework.TensorShapeProto.Builder, org.tensorflow.proto.framework.TensorShapeProtoOrBuilder> shapeBuilder_;
+    /**
+     * <pre>
+     * "shape"
+     * </pre>
+     *
+     * <code>.tensorflow.TensorShapeProto shape = 7;</code>
+     */
+    public boolean hasShape() {
+      return valueCase_ == 7;
+    }
+    /**
+     * <pre>
+     * "shape"
+     * </pre>
+     *
+     * <code>.tensorflow.TensorShapeProto shape = 7;</code>
+     */
+    public org.tensorflow.proto.framework.TensorShapeProto getShape() {
+      if (shapeBuilder_ == null) {
+        if (valueCase_ == 7) {
+          return (org.tensorflow.proto.framework.TensorShapeProto) value_;
+        }
+        return org.tensorflow.proto.framework.TensorShapeProto.getDefaultInstance();
+      } else {
+        if (valueCase_ == 7) {
+          return shapeBuilder_.getMessage();
+        }
+        return org.tensorflow.proto.framework.TensorShapeProto.getDefaultInstance();
+      }
+    }
+    /**
+     * <pre>
+     * "shape"
+     * </pre>
+     *
+     * <code>.tensorflow.TensorShapeProto shape = 7;</code>
+     */
+    public Builder setShape(org.tensorflow.proto.framework.TensorShapeProto value) {
+      if (shapeBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        value_ = value;
+        onChanged();
+      } else {
+        shapeBuilder_.setMessage(value);
+      }
+      valueCase_ = 7;
+      return this;
+    }
+    /**
+     * <pre>
+     * "shape"
+     * </pre>
+     *
+     * <code>.tensorflow.TensorShapeProto shape = 7;</code>
+     */
+    public Builder setShape(
+        org.tensorflow.proto.framework.TensorShapeProto.Builder builderForValue) {
+      if (shapeBuilder_ == null) {
+        value_ = builderForValue.build();
+        onChanged();
+      } else {
+        shapeBuilder_.setMessage(builderForValue.build());
+      }
+      valueCase_ = 7;
+      return this;
+    }
+    /**
+     * <pre>
+     * "shape"
+     * </pre>
+     *
+     * <code>.tensorflow.TensorShapeProto shape = 7;</code>
+     */
+    public Builder mergeShape(org.tensorflow.proto.framework.TensorShapeProto value) {
+      if (shapeBuilder_ == null) {
+        if (valueCase_ == 7 &&
+            value_ != org.tensorflow.proto.framework.TensorShapeProto.getDefaultInstance()) {
+          value_ = org.tensorflow.proto.framework.TensorShapeProto.newBuilder((org.tensorflow.proto.framework.TensorShapeProto) value_)
+              .mergeFrom(value).buildPartial();
+        } else {
+          value_ = value;
+        }
+        onChanged();
+      } else {
+        if (valueCase_ == 7) {
+          shapeBuilder_.mergeFrom(value);
+        }
+        shapeBuilder_.setMessage(value);
+      }
+      valueCase_ = 7;
+      return this;
+    }
+    /**
+     * <pre>
+     * "shape"
+     * </pre>
+     *
+     * <code>.tensorflow.TensorShapeProto shape = 7;</code>
+     */
+    public Builder clearShape() {
+      if (shapeBuilder_ == null) {
+        if (valueCase_ == 7) {
+          valueCase_ = 0;
+          value_ = null;
+          onChanged();
+        }
+      } else {
+        if (valueCase_ == 7) {
+          valueCase_ = 0;
+          value_ = null;
+        }
+        shapeBuilder_.clear();
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * "shape"
+     * </pre>
+     *
+     * <code>.tensorflow.TensorShapeProto shape = 7;</code>
+     */
+    public org.tensorflow.proto.framework.TensorShapeProto.Builder getShapeBuilder() {
+      return getShapeFieldBuilder().getBuilder();
+    }
+    /**
+     * <pre>
+     * "shape"
+     * </pre>
+     *
+     * <code>.tensorflow.TensorShapeProto shape = 7;</code>
+     */
+    public org.tensorflow.proto.framework.TensorShapeProtoOrBuilder getShapeOrBuilder() {
+      if ((valueCase_ == 7) && (shapeBuilder_ != null)) {
+        return shapeBuilder_.getMessageOrBuilder();
+      } else {
+        if (valueCase_ == 7) {
+          return (org.tensorflow.proto.framework.TensorShapeProto) value_;
+        }
+        return org.tensorflow.proto.framework.TensorShapeProto.getDefaultInstance();
+      }
+    }
+    /**
+     * <pre>
+     * "shape"
+     * </pre>
+     *
+     * <code>.tensorflow.TensorShapeProto shape = 7;</code>
+     */
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.TensorShapeProto, org.tensorflow.proto.framework.TensorShapeProto.Builder, org.tensorflow.proto.framework.TensorShapeProtoOrBuilder> 
+        getShapeFieldBuilder() {
+      if (shapeBuilder_ == null) {
+        if (!(valueCase_ == 7)) {
+          value_ = org.tensorflow.proto.framework.TensorShapeProto.getDefaultInstance();
+        }
+        shapeBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+            org.tensorflow.proto.framework.TensorShapeProto, org.tensorflow.proto.framework.TensorShapeProto.Builder, org.tensorflow.proto.framework.TensorShapeProtoOrBuilder>(
+                (org.tensorflow.proto.framework.TensorShapeProto) value_,
+                getParentForChildren(),
+                isClean());
+        value_ = null;
+      }
+      valueCase_ = 7;
+      onChanged();;
+      return shapeBuilder_;
+    }
+
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.TensorProto, org.tensorflow.proto.framework.TensorProto.Builder, org.tensorflow.proto.framework.TensorProtoOrBuilder> tensorBuilder_;
+    /**
+     * <pre>
+     * "tensor"
+     * </pre>
+     *
+     * <code>.tensorflow.TensorProto tensor = 8;</code>
+     */
+    public boolean hasTensor() {
+      return valueCase_ == 8;
+    }
+    /**
+     * <pre>
+     * "tensor"
+     * </pre>
+     *
+     * <code>.tensorflow.TensorProto tensor = 8;</code>
+     */
+    public org.tensorflow.proto.framework.TensorProto getTensor() {
+      if (tensorBuilder_ == null) {
+        if (valueCase_ == 8) {
+          return (org.tensorflow.proto.framework.TensorProto) value_;
+        }
+        return org.tensorflow.proto.framework.TensorProto.getDefaultInstance();
+      } else {
+        if (valueCase_ == 8) {
+          return tensorBuilder_.getMessage();
+        }
+        return org.tensorflow.proto.framework.TensorProto.getDefaultInstance();
+      }
+    }
+    /**
+     * <pre>
+     * "tensor"
+     * </pre>
+     *
+     * <code>.tensorflow.TensorProto tensor = 8;</code>
+     */
+    public Builder setTensor(org.tensorflow.proto.framework.TensorProto value) {
+      if (tensorBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        value_ = value;
+        onChanged();
+      } else {
+        tensorBuilder_.setMessage(value);
+      }
+      valueCase_ = 8;
+      return this;
+    }
+    /**
+     * <pre>
+     * "tensor"
+     * </pre>
+     *
+     * <code>.tensorflow.TensorProto tensor = 8;</code>
+     */
+    public Builder setTensor(
+        org.tensorflow.proto.framework.TensorProto.Builder builderForValue) {
+      if (tensorBuilder_ == null) {
+        value_ = builderForValue.build();
+        onChanged();
+      } else {
+        tensorBuilder_.setMessage(builderForValue.build());
+      }
+      valueCase_ = 8;
+      return this;
+    }
+    /**
+     * <pre>
+     * "tensor"
+     * </pre>
+     *
+     * <code>.tensorflow.TensorProto tensor = 8;</code>
+     */
+    public Builder mergeTensor(org.tensorflow.proto.framework.TensorProto value) {
+      if (tensorBuilder_ == null) {
+        if (valueCase_ == 8 &&
+            value_ != org.tensorflow.proto.framework.TensorProto.getDefaultInstance()) {
+          value_ = org.tensorflow.proto.framework.TensorProto.newBuilder((org.tensorflow.proto.framework.TensorProto) value_)
+              .mergeFrom(value).buildPartial();
+        } else {
+          value_ = value;
+        }
+        onChanged();
+      } else {
+        if (valueCase_ == 8) {
+          tensorBuilder_.mergeFrom(value);
+        }
+        tensorBuilder_.setMessage(value);
+      }
+      valueCase_ = 8;
+      return this;
+    }
+    /**
+     * <pre>
+     * "tensor"
+     * </pre>
+     *
+     * <code>.tensorflow.TensorProto tensor = 8;</code>
+     */
+    public Builder clearTensor() {
+      if (tensorBuilder_ == null) {
+        if (valueCase_ == 8) {
+          valueCase_ = 0;
+          value_ = null;
+          onChanged();
+        }
+      } else {
+        if (valueCase_ == 8) {
+          valueCase_ = 0;
+          value_ = null;
+        }
+        tensorBuilder_.clear();
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * "tensor"
+     * </pre>
+     *
+     * <code>.tensorflow.TensorProto tensor = 8;</code>
+     */
+    public org.tensorflow.proto.framework.TensorProto.Builder getTensorBuilder() {
+      return getTensorFieldBuilder().getBuilder();
+    }
+    /**
+     * <pre>
+     * "tensor"
+     * </pre>
+     *
+     * <code>.tensorflow.TensorProto tensor = 8;</code>
+     */
+    public org.tensorflow.proto.framework.TensorProtoOrBuilder getTensorOrBuilder() {
+      if ((valueCase_ == 8) && (tensorBuilder_ != null)) {
+        return tensorBuilder_.getMessageOrBuilder();
+      } else {
+        if (valueCase_ == 8) {
+          return (org.tensorflow.proto.framework.TensorProto) value_;
+        }
+        return org.tensorflow.proto.framework.TensorProto.getDefaultInstance();
+      }
+    }
+    /**
+     * <pre>
+     * "tensor"
+     * </pre>
+     *
+     * <code>.tensorflow.TensorProto tensor = 8;</code>
+     */
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.TensorProto, org.tensorflow.proto.framework.TensorProto.Builder, org.tensorflow.proto.framework.TensorProtoOrBuilder> 
+        getTensorFieldBuilder() {
+      if (tensorBuilder_ == null) {
+        if (!(valueCase_ == 8)) {
+          value_ = org.tensorflow.proto.framework.TensorProto.getDefaultInstance();
+        }
+        tensorBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+            org.tensorflow.proto.framework.TensorProto, org.tensorflow.proto.framework.TensorProto.Builder, org.tensorflow.proto.framework.TensorProtoOrBuilder>(
+                (org.tensorflow.proto.framework.TensorProto) value_,
+                getParentForChildren(),
+                isClean());
+        value_ = null;
+      }
+      valueCase_ = 8;
+      onChanged();;
+      return tensorBuilder_;
+    }
+
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.AttrValue.ListValue, org.tensorflow.proto.framework.AttrValue.ListValue.Builder, org.tensorflow.proto.framework.AttrValue.ListValueOrBuilder> listBuilder_;
+    /**
+     * <pre>
+     * any "list(...)"
+     * </pre>
+     *
+     * <code>.tensorflow.AttrValue.ListValue list = 1;</code>
+     */
+    public boolean hasList() {
+      return valueCase_ == 1;
+    }
+    /**
+     * <pre>
+     * any "list(...)"
+     * </pre>
+     *
+     * <code>.tensorflow.AttrValue.ListValue list = 1;</code>
+     */
+    public org.tensorflow.proto.framework.AttrValue.ListValue getList() {
+      if (listBuilder_ == null) {
+        if (valueCase_ == 1) {
+          return (org.tensorflow.proto.framework.AttrValue.ListValue) value_;
+        }
+        return org.tensorflow.proto.framework.AttrValue.ListValue.getDefaultInstance();
+      } else {
+        if (valueCase_ == 1) {
+          return listBuilder_.getMessage();
+        }
+        return org.tensorflow.proto.framework.AttrValue.ListValue.getDefaultInstance();
+      }
+    }
+    /**
+     * <pre>
+     * any "list(...)"
+     * </pre>
+     *
+     * <code>.tensorflow.AttrValue.ListValue list = 1;</code>
+     */
+    public Builder setList(org.tensorflow.proto.framework.AttrValue.ListValue value) {
+      if (listBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        value_ = value;
+        onChanged();
+      } else {
+        listBuilder_.setMessage(value);
+      }
+      valueCase_ = 1;
+      return this;
+    }
+    /**
+     * <pre>
+     * any "list(...)"
+     * </pre>
+     *
+     * <code>.tensorflow.AttrValue.ListValue list = 1;</code>
+     */
+    public Builder setList(
+        org.tensorflow.proto.framework.AttrValue.ListValue.Builder builderForValue) {
+      if (listBuilder_ == null) {
+        value_ = builderForValue.build();
+        onChanged();
+      } else {
+        listBuilder_.setMessage(builderForValue.build());
+      }
+      valueCase_ = 1;
+      return this;
+    }
+    /**
+     * <pre>
+     * any "list(...)"
+     * </pre>
+     *
+     * <code>.tensorflow.AttrValue.ListValue list = 1;</code>
+     */
+    public Builder mergeList(org.tensorflow.proto.framework.AttrValue.ListValue value) {
+      if (listBuilder_ == null) {
+        if (valueCase_ == 1 &&
+            value_ != org.tensorflow.proto.framework.AttrValue.ListValue.getDefaultInstance()) {
+          value_ = org.tensorflow.proto.framework.AttrValue.ListValue.newBuilder((org.tensorflow.proto.framework.AttrValue.ListValue) value_)
+              .mergeFrom(value).buildPartial();
+        } else {
+          value_ = value;
+        }
+        onChanged();
+      } else {
+        if (valueCase_ == 1) {
+          listBuilder_.mergeFrom(value);
+        }
+        listBuilder_.setMessage(value);
+      }
+      valueCase_ = 1;
+      return this;
+    }
+    /**
+     * <pre>
+     * any "list(...)"
+     * </pre>
+     *
+     * <code>.tensorflow.AttrValue.ListValue list = 1;</code>
+     */
+    public Builder clearList() {
+      if (listBuilder_ == null) {
+        if (valueCase_ == 1) {
+          valueCase_ = 0;
+          value_ = null;
+          onChanged();
+        }
+      } else {
+        if (valueCase_ == 1) {
+          valueCase_ = 0;
+          value_ = null;
+        }
+        listBuilder_.clear();
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * any "list(...)"
+     * </pre>
+     *
+     * <code>.tensorflow.AttrValue.ListValue list = 1;</code>
+     */
+    public org.tensorflow.proto.framework.AttrValue.ListValue.Builder getListBuilder() {
+      return getListFieldBuilder().getBuilder();
+    }
+    /**
+     * <pre>
+     * any "list(...)"
+     * </pre>
+     *
+     * <code>.tensorflow.AttrValue.ListValue list = 1;</code>
+     */
+    public org.tensorflow.proto.framework.AttrValue.ListValueOrBuilder getListOrBuilder() {
+      if ((valueCase_ == 1) && (listBuilder_ != null)) {
+        return listBuilder_.getMessageOrBuilder();
+      } else {
+        if (valueCase_ == 1) {
+          return (org.tensorflow.proto.framework.AttrValue.ListValue) value_;
+        }
+        return org.tensorflow.proto.framework.AttrValue.ListValue.getDefaultInstance();
+      }
+    }
+    /**
+     * <pre>
+     * any "list(...)"
+     * </pre>
+     *
+     * <code>.tensorflow.AttrValue.ListValue list = 1;</code>
+     */
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.AttrValue.ListValue, org.tensorflow.proto.framework.AttrValue.ListValue.Builder, org.tensorflow.proto.framework.AttrValue.ListValueOrBuilder> 
+        getListFieldBuilder() {
+      if (listBuilder_ == null) {
+        if (!(valueCase_ == 1)) {
+          value_ = org.tensorflow.proto.framework.AttrValue.ListValue.getDefaultInstance();
+        }
+        listBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+            org.tensorflow.proto.framework.AttrValue.ListValue, org.tensorflow.proto.framework.AttrValue.ListValue.Builder, org.tensorflow.proto.framework.AttrValue.ListValueOrBuilder>(
+                (org.tensorflow.proto.framework.AttrValue.ListValue) value_,
+                getParentForChildren(),
+                isClean());
+        value_ = null;
+      }
+      valueCase_ = 1;
+      onChanged();;
+      return listBuilder_;
+    }
+
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.NameAttrList, org.tensorflow.proto.framework.NameAttrList.Builder, org.tensorflow.proto.framework.NameAttrListOrBuilder> funcBuilder_;
+    /**
+     * <pre>
+     * "func" represents a function. func.name is a function's name or
+     * a primitive op's name. func.attr.first is the name of an attr
+     * defined for that function. func.attr.second is the value for
+     * that attr in the instantiation.
+     * </pre>
+     *
+     * <code>.tensorflow.NameAttrList func = 10;</code>
+     */
+    public boolean hasFunc() {
+      return valueCase_ == 10;
+    }
+    /**
+     * <pre>
+     * "func" represents a function. func.name is a function's name or
+     * a primitive op's name. func.attr.first is the name of an attr
+     * defined for that function. func.attr.second is the value for
+     * that attr in the instantiation.
+     * </pre>
+     *
+     * <code>.tensorflow.NameAttrList func = 10;</code>
+     */
+    public org.tensorflow.proto.framework.NameAttrList getFunc() {
+      if (funcBuilder_ == null) {
+        if (valueCase_ == 10) {
+          return (org.tensorflow.proto.framework.NameAttrList) value_;
+        }
+        return org.tensorflow.proto.framework.NameAttrList.getDefaultInstance();
+      } else {
+        if (valueCase_ == 10) {
+          return funcBuilder_.getMessage();
+        }
+        return org.tensorflow.proto.framework.NameAttrList.getDefaultInstance();
+      }
+    }
+    /**
+     * <pre>
+     * "func" represents a function. func.name is a function's name or
+     * a primitive op's name. func.attr.first is the name of an attr
+     * defined for that function. func.attr.second is the value for
+     * that attr in the instantiation.
+     * </pre>
+     *
+     * <code>.tensorflow.NameAttrList func = 10;</code>
+     */
+    public Builder setFunc(org.tensorflow.proto.framework.NameAttrList value) {
+      if (funcBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        value_ = value;
+        onChanged();
+      } else {
+        funcBuilder_.setMessage(value);
+      }
+      valueCase_ = 10;
+      return this;
+    }
+    /**
+     * <pre>
+     * "func" represents a function. func.name is a function's name or
+     * a primitive op's name. func.attr.first is the name of an attr
+     * defined for that function. func.attr.second is the value for
+     * that attr in the instantiation.
+     * </pre>
+     *
+     * <code>.tensorflow.NameAttrList func = 10;</code>
+     */
+    public Builder setFunc(
+        org.tensorflow.proto.framework.NameAttrList.Builder builderForValue) {
+      if (funcBuilder_ == null) {
+        value_ = builderForValue.build();
+        onChanged();
+      } else {
+        funcBuilder_.setMessage(builderForValue.build());
+      }
+      valueCase_ = 10;
+      return this;
+    }
+    /**
+     * <pre>
+     * "func" represents a function. func.name is a function's name or
+     * a primitive op's name. func.attr.first is the name of an attr
+     * defined for that function. func.attr.second is the value for
+     * that attr in the instantiation.
+     * </pre>
+     *
+     * <code>.tensorflow.NameAttrList func = 10;</code>
+     */
+    public Builder mergeFunc(org.tensorflow.proto.framework.NameAttrList value) {
+      if (funcBuilder_ == null) {
+        if (valueCase_ == 10 &&
+            value_ != org.tensorflow.proto.framework.NameAttrList.getDefaultInstance()) {
+          value_ = org.tensorflow.proto.framework.NameAttrList.newBuilder((org.tensorflow.proto.framework.NameAttrList) value_)
+              .mergeFrom(value).buildPartial();
+        } else {
+          value_ = value;
+        }
+        onChanged();
+      } else {
+        if (valueCase_ == 10) {
+          funcBuilder_.mergeFrom(value);
+        }
+        funcBuilder_.setMessage(value);
+      }
+      valueCase_ = 10;
+      return this;
+    }
+    /**
+     * <pre>
+     * "func" represents a function. func.name is a function's name or
+     * a primitive op's name. func.attr.first is the name of an attr
+     * defined for that function. func.attr.second is the value for
+     * that attr in the instantiation.
+     * </pre>
+     *
+     * <code>.tensorflow.NameAttrList func = 10;</code>
+     */
+    public Builder clearFunc() {
+      if (funcBuilder_ == null) {
+        if (valueCase_ == 10) {
+          valueCase_ = 0;
+          value_ = null;
+          onChanged();
+        }
+      } else {
+        if (valueCase_ == 10) {
+          valueCase_ = 0;
+          value_ = null;
+        }
+        funcBuilder_.clear();
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * "func" represents a function. func.name is a function's name or
+     * a primitive op's name. func.attr.first is the name of an attr
+     * defined for that function. func.attr.second is the value for
+     * that attr in the instantiation.
+     * </pre>
+     *
+     * <code>.tensorflow.NameAttrList func = 10;</code>
+     */
+    public org.tensorflow.proto.framework.NameAttrList.Builder getFuncBuilder() {
+      return getFuncFieldBuilder().getBuilder();
+    }
+    /**
+     * <pre>
+     * "func" represents a function. func.name is a function's name or
+     * a primitive op's name. func.attr.first is the name of an attr
+     * defined for that function. func.attr.second is the value for
+     * that attr in the instantiation.
+     * </pre>
+     *
+     * <code>.tensorflow.NameAttrList func = 10;</code>
+     */
+    public org.tensorflow.proto.framework.NameAttrListOrBuilder getFuncOrBuilder() {
+      if ((valueCase_ == 10) && (funcBuilder_ != null)) {
+        return funcBuilder_.getMessageOrBuilder();
+      } else {
+        if (valueCase_ == 10) {
+          return (org.tensorflow.proto.framework.NameAttrList) value_;
+        }
+        return org.tensorflow.proto.framework.NameAttrList.getDefaultInstance();
+      }
+    }
+    /**
+     * <pre>
+     * "func" represents a function. func.name is a function's name or
+     * a primitive op's name. func.attr.first is the name of an attr
+     * defined for that function. func.attr.second is the value for
+     * that attr in the instantiation.
+     * </pre>
+     *
+     * <code>.tensorflow.NameAttrList func = 10;</code>
+     */
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.NameAttrList, org.tensorflow.proto.framework.NameAttrList.Builder, org.tensorflow.proto.framework.NameAttrListOrBuilder> 
+        getFuncFieldBuilder() {
+      if (funcBuilder_ == null) {
+        if (!(valueCase_ == 10)) {
+          value_ = org.tensorflow.proto.framework.NameAttrList.getDefaultInstance();
+        }
+        funcBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+            org.tensorflow.proto.framework.NameAttrList, org.tensorflow.proto.framework.NameAttrList.Builder, org.tensorflow.proto.framework.NameAttrListOrBuilder>(
+                (org.tensorflow.proto.framework.NameAttrList) value_,
+                getParentForChildren(),
+                isClean());
+        value_ = null;
+      }
+      valueCase_ = 10;
+      onChanged();;
+      return funcBuilder_;
+    }
+
+    /**
+     * <pre>
+     * This is a placeholder only used in nodes defined inside a
+     * function.  It indicates the attr value will be supplied when
+     * the function is instantiated.  For example, let us suppose a
+     * node "N" in function "FN". "N" has an attr "A" with value
+     * placeholder = "foo". When FN is instantiated with attr "foo"
+     * set to "bar", the instantiated node N's attr A will have been
+     * given the value "bar".
+     * </pre>
+     *
+     * <code>string placeholder = 9;</code>
+     */
+    public java.lang.String getPlaceholder() {
+      java.lang.Object ref = "";
+      if (valueCase_ == 9) {
+        ref = value_;
+      }
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        if (valueCase_ == 9) {
+          value_ = s;
+        }
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * This is a placeholder only used in nodes defined inside a
+     * function.  It indicates the attr value will be supplied when
+     * the function is instantiated.  For example, let us suppose a
+     * node "N" in function "FN". "N" has an attr "A" with value
+     * placeholder = "foo". When FN is instantiated with attr "foo"
+     * set to "bar", the instantiated node N's attr A will have been
+     * given the value "bar".
+     * </pre>
+     *
+     * <code>string placeholder = 9;</code>
+     */
+    public com.google.protobuf.ByteString
+        getPlaceholderBytes() {
+      java.lang.Object ref = "";
+      if (valueCase_ == 9) {
+        ref = value_;
+      }
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        if (valueCase_ == 9) {
+          value_ = b;
+        }
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * This is a placeholder only used in nodes defined inside a
+     * function.  It indicates the attr value will be supplied when
+     * the function is instantiated.  For example, let us suppose a
+     * node "N" in function "FN". "N" has an attr "A" with value
+     * placeholder = "foo". When FN is instantiated with attr "foo"
+     * set to "bar", the instantiated node N's attr A will have been
+     * given the value "bar".
+     * </pre>
+     *
+     * <code>string placeholder = 9;</code>
+     */
+    public Builder setPlaceholder(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  valueCase_ = 9;
+      value_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * This is a placeholder only used in nodes defined inside a
+     * function.  It indicates the attr value will be supplied when
+     * the function is instantiated.  For example, let us suppose a
+     * node "N" in function "FN". "N" has an attr "A" with value
+     * placeholder = "foo". When FN is instantiated with attr "foo"
+     * set to "bar", the instantiated node N's attr A will have been
+     * given the value "bar".
+     * </pre>
+     *
+     * <code>string placeholder = 9;</code>
+     */
+    public Builder clearPlaceholder() {
+      if (valueCase_ == 9) {
+        valueCase_ = 0;
+        value_ = null;
+        onChanged();
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * This is a placeholder only used in nodes defined inside a
+     * function.  It indicates the attr value will be supplied when
+     * the function is instantiated.  For example, let us suppose a
+     * node "N" in function "FN". "N" has an attr "A" with value
+     * placeholder = "foo". When FN is instantiated with attr "foo"
+     * set to "bar", the instantiated node N's attr A will have been
+     * given the value "bar".
+     * </pre>
+     *
+     * <code>string placeholder = 9;</code>
+     */
+    public Builder setPlaceholderBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      valueCase_ = 9;
+      value_ = value;
+      onChanged();
+      return this;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.AttrValue)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.AttrValue)
+  private static final org.tensorflow.proto.framework.AttrValue DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.AttrValue();
+  }
+
+  public static org.tensorflow.proto.framework.AttrValue getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<AttrValue>
+      PARSER = new com.google.protobuf.AbstractParser<AttrValue>() {
+    @java.lang.Override
+    public AttrValue parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new AttrValue(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<AttrValue> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<AttrValue> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.framework.AttrValue getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/AttrValueOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/AttrValueOrBuilder.java
new file mode 100644
index 00000000000..f2624691edf
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/AttrValueOrBuilder.java
@@ -0,0 +1,203 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/framework/attr_value.proto
+
+package org.tensorflow.proto.framework;
+
+public interface AttrValueOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.AttrValue)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <pre>
+   * "string"
+   * </pre>
+   *
+   * <code>bytes s = 2;</code>
+   */
+  com.google.protobuf.ByteString getS();
+
+  /**
+   * <pre>
+   * "int"
+   * </pre>
+   *
+   * <code>int64 i = 3;</code>
+   */
+  long getI();
+
+  /**
+   * <pre>
+   * "float"
+   * </pre>
+   *
+   * <code>float f = 4;</code>
+   */
+  float getF();
+
+  /**
+   * <pre>
+   * "bool"
+   * </pre>
+   *
+   * <code>bool b = 5;</code>
+   */
+  boolean getB();
+
+  /**
+   * <pre>
+   * "type"
+   * </pre>
+   *
+   * <code>.tensorflow.DataType type = 6;</code>
+   */
+  int getTypeValue();
+  /**
+   * <pre>
+   * "type"
+   * </pre>
+   *
+   * <code>.tensorflow.DataType type = 6;</code>
+   */
+  org.tensorflow.proto.framework.DataType getType();
+
+  /**
+   * <pre>
+   * "shape"
+   * </pre>
+   *
+   * <code>.tensorflow.TensorShapeProto shape = 7;</code>
+   */
+  boolean hasShape();
+  /**
+   * <pre>
+   * "shape"
+   * </pre>
+   *
+   * <code>.tensorflow.TensorShapeProto shape = 7;</code>
+   */
+  org.tensorflow.proto.framework.TensorShapeProto getShape();
+  /**
+   * <pre>
+   * "shape"
+   * </pre>
+   *
+   * <code>.tensorflow.TensorShapeProto shape = 7;</code>
+   */
+  org.tensorflow.proto.framework.TensorShapeProtoOrBuilder getShapeOrBuilder();
+
+  /**
+   * <pre>
+   * "tensor"
+   * </pre>
+   *
+   * <code>.tensorflow.TensorProto tensor = 8;</code>
+   */
+  boolean hasTensor();
+  /**
+   * <pre>
+   * "tensor"
+   * </pre>
+   *
+   * <code>.tensorflow.TensorProto tensor = 8;</code>
+   */
+  org.tensorflow.proto.framework.TensorProto getTensor();
+  /**
+   * <pre>
+   * "tensor"
+   * </pre>
+   *
+   * <code>.tensorflow.TensorProto tensor = 8;</code>
+   */
+  org.tensorflow.proto.framework.TensorProtoOrBuilder getTensorOrBuilder();
+
+  /**
+   * <pre>
+   * any "list(...)"
+   * </pre>
+   *
+   * <code>.tensorflow.AttrValue.ListValue list = 1;</code>
+   */
+  boolean hasList();
+  /**
+   * <pre>
+   * any "list(...)"
+   * </pre>
+   *
+   * <code>.tensorflow.AttrValue.ListValue list = 1;</code>
+   */
+  org.tensorflow.proto.framework.AttrValue.ListValue getList();
+  /**
+   * <pre>
+   * any "list(...)"
+   * </pre>
+   *
+   * <code>.tensorflow.AttrValue.ListValue list = 1;</code>
+   */
+  org.tensorflow.proto.framework.AttrValue.ListValueOrBuilder getListOrBuilder();
+
+  /**
+   * <pre>
+   * "func" represents a function. func.name is a function's name or
+   * a primitive op's name. func.attr.first is the name of an attr
+   * defined for that function. func.attr.second is the value for
+   * that attr in the instantiation.
+   * </pre>
+   *
+   * <code>.tensorflow.NameAttrList func = 10;</code>
+   */
+  boolean hasFunc();
+  /**
+   * <pre>
+   * "func" represents a function. func.name is a function's name or
+   * a primitive op's name. func.attr.first is the name of an attr
+   * defined for that function. func.attr.second is the value for
+   * that attr in the instantiation.
+   * </pre>
+   *
+   * <code>.tensorflow.NameAttrList func = 10;</code>
+   */
+  org.tensorflow.proto.framework.NameAttrList getFunc();
+  /**
+   * <pre>
+   * "func" represents a function. func.name is a function's name or
+   * a primitive op's name. func.attr.first is the name of an attr
+   * defined for that function. func.attr.second is the value for
+   * that attr in the instantiation.
+   * </pre>
+   *
+   * <code>.tensorflow.NameAttrList func = 10;</code>
+   */
+  org.tensorflow.proto.framework.NameAttrListOrBuilder getFuncOrBuilder();
+
+  /**
+   * <pre>
+   * This is a placeholder only used in nodes defined inside a
+   * function.  It indicates the attr value will be supplied when
+   * the function is instantiated.  For example, let us suppose a
+   * node "N" in function "FN". "N" has an attr "A" with value
+   * placeholder = "foo". When FN is instantiated with attr "foo"
+   * set to "bar", the instantiated node N's attr A will have been
+   * given the value "bar".
+   * </pre>
+   *
+   * <code>string placeholder = 9;</code>
+   */
+  java.lang.String getPlaceholder();
+  /**
+   * <pre>
+   * This is a placeholder only used in nodes defined inside a
+   * function.  It indicates the attr value will be supplied when
+   * the function is instantiated.  For example, let us suppose a
+   * node "N" in function "FN". "N" has an attr "A" with value
+   * placeholder = "foo". When FN is instantiated with attr "foo"
+   * set to "bar", the instantiated node N's attr A will have been
+   * given the value "bar".
+   * </pre>
+   *
+   * <code>string placeholder = 9;</code>
+   */
+  com.google.protobuf.ByteString
+      getPlaceholderBytes();
+
+  public org.tensorflow.proto.framework.AttrValue.ValueCase getValueCase();
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/AttrValueProtos.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/AttrValueProtos.java
new file mode 100644
index 00000000000..2df18dbc3d0
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/AttrValueProtos.java
@@ -0,0 +1,110 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/framework/attr_value.proto
+
+package org.tensorflow.proto.framework;
+
+public final class AttrValueProtos {
+  private AttrValueProtos() {}
+  public static void registerAllExtensions(
+      com.google.protobuf.ExtensionRegistryLite registry) {
+  }
+
+  public static void registerAllExtensions(
+      com.google.protobuf.ExtensionRegistry registry) {
+    registerAllExtensions(
+        (com.google.protobuf.ExtensionRegistryLite) registry);
+  }
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_AttrValue_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_AttrValue_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_AttrValue_ListValue_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_AttrValue_ListValue_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_NameAttrList_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_NameAttrList_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_NameAttrList_AttrEntry_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_NameAttrList_AttrEntry_fieldAccessorTable;
+
+  public static com.google.protobuf.Descriptors.FileDescriptor
+      getDescriptor() {
+    return descriptor;
+  }
+  private static  com.google.protobuf.Descriptors.FileDescriptor
+      descriptor;
+  static {
+    java.lang.String[] descriptorData = {
+      "\n*tensorflow/core/framework/attr_value.p" +
+      "roto\022\ntensorflow\032&tensorflow/core/framew" +
+      "ork/tensor.proto\032,tensorflow/core/framew" +
+      "ork/tensor_shape.proto\032%tensorflow/core/" +
+      "framework/types.proto\"\246\004\n\tAttrValue\022\013\n\001s" +
+      "\030\002 \001(\014H\000\022\013\n\001i\030\003 \001(\003H\000\022\013\n\001f\030\004 \001(\002H\000\022\013\n\001b\030" +
+      "\005 \001(\010H\000\022$\n\004type\030\006 \001(\0162\024.tensorflow.DataT" +
+      "ypeH\000\022-\n\005shape\030\007 \001(\0132\034.tensorflow.Tensor" +
+      "ShapeProtoH\000\022)\n\006tensor\030\010 \001(\0132\027.tensorflo" +
+      "w.TensorProtoH\000\022/\n\004list\030\001 \001(\0132\037.tensorfl" +
+      "ow.AttrValue.ListValueH\000\022(\n\004func\030\n \001(\0132\030" +
+      ".tensorflow.NameAttrListH\000\022\025\n\013placeholde" +
+      "r\030\t \001(\tH\000\032\351\001\n\tListValue\022\t\n\001s\030\002 \003(\014\022\r\n\001i\030" +
+      "\003 \003(\003B\002\020\001\022\r\n\001f\030\004 \003(\002B\002\020\001\022\r\n\001b\030\005 \003(\010B\002\020\001\022" +
+      "&\n\004type\030\006 \003(\0162\024.tensorflow.DataTypeB\002\020\001\022" +
+      "+\n\005shape\030\007 \003(\0132\034.tensorflow.TensorShapeP" +
+      "roto\022\'\n\006tensor\030\010 \003(\0132\027.tensorflow.Tensor" +
+      "Proto\022&\n\004func\030\t \003(\0132\030.tensorflow.NameAtt" +
+      "rListB\007\n\005value\"\222\001\n\014NameAttrList\022\014\n\004name\030" +
+      "\001 \001(\t\0220\n\004attr\030\002 \003(\0132\".tensorflow.NameAtt" +
+      "rList.AttrEntry\032B\n\tAttrEntry\022\013\n\003key\030\001 \001(" +
+      "\t\022$\n\005value\030\002 \001(\0132\025.tensorflow.AttrValue:" +
+      "\0028\001B\211\001\n\036org.tensorflow.proto.frameworkB\017" +
+      "AttrValueProtosP\001ZQgithub.com/tensorflow" +
+      "/tensorflow/tensorflow/go/core/framework" +
+      "/attr_value_go_proto\370\001\001b\006proto3"
+    };
+    descriptor = com.google.protobuf.Descriptors.FileDescriptor
+      .internalBuildGeneratedFileFrom(descriptorData,
+        new com.google.protobuf.Descriptors.FileDescriptor[] {
+          org.tensorflow.proto.framework.TensorProtos.getDescriptor(),
+          org.tensorflow.proto.framework.TensorShapeProtos.getDescriptor(),
+          org.tensorflow.proto.framework.TypesProtos.getDescriptor(),
+        });
+    internal_static_tensorflow_AttrValue_descriptor =
+      getDescriptor().getMessageTypes().get(0);
+    internal_static_tensorflow_AttrValue_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_AttrValue_descriptor,
+        new java.lang.String[] { "S", "I", "F", "B", "Type", "Shape", "Tensor", "List", "Func", "Placeholder", "Value", });
+    internal_static_tensorflow_AttrValue_ListValue_descriptor =
+      internal_static_tensorflow_AttrValue_descriptor.getNestedTypes().get(0);
+    internal_static_tensorflow_AttrValue_ListValue_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_AttrValue_ListValue_descriptor,
+        new java.lang.String[] { "S", "I", "F", "B", "Type", "Shape", "Tensor", "Func", });
+    internal_static_tensorflow_NameAttrList_descriptor =
+      getDescriptor().getMessageTypes().get(1);
+    internal_static_tensorflow_NameAttrList_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_NameAttrList_descriptor,
+        new java.lang.String[] { "Name", "Attr", });
+    internal_static_tensorflow_NameAttrList_AttrEntry_descriptor =
+      internal_static_tensorflow_NameAttrList_descriptor.getNestedTypes().get(0);
+    internal_static_tensorflow_NameAttrList_AttrEntry_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_NameAttrList_AttrEntry_descriptor,
+        new java.lang.String[] { "Key", "Value", });
+    org.tensorflow.proto.framework.TensorProtos.getDescriptor();
+    org.tensorflow.proto.framework.TensorShapeProtos.getDescriptor();
+    org.tensorflow.proto.framework.TypesProtos.getDescriptor();
+  }
+
+  // @@protoc_insertion_point(outer_class_scope)
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/AutoParallelOptions.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/AutoParallelOptions.java
new file mode 100644
index 00000000000..cbb6bde48fe
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/AutoParallelOptions.java
@@ -0,0 +1,534 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/rewriter_config.proto
+
+package org.tensorflow.proto.framework;
+
+/**
+ * Protobuf type {@code tensorflow.AutoParallelOptions}
+ */
+public  final class AutoParallelOptions extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.AutoParallelOptions)
+    AutoParallelOptionsOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use AutoParallelOptions.newBuilder() to construct.
+  private AutoParallelOptions(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private AutoParallelOptions() {
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new AutoParallelOptions();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private AutoParallelOptions(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 8: {
+
+            enable_ = input.readBool();
+            break;
+          }
+          case 16: {
+
+            numReplicas_ = input.readInt32();
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.framework.RewriterConfigProtos.internal_static_tensorflow_AutoParallelOptions_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.framework.RewriterConfigProtos.internal_static_tensorflow_AutoParallelOptions_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.framework.AutoParallelOptions.class, org.tensorflow.proto.framework.AutoParallelOptions.Builder.class);
+  }
+
+  public static final int ENABLE_FIELD_NUMBER = 1;
+  private boolean enable_;
+  /**
+   * <code>bool enable = 1;</code>
+   */
+  public boolean getEnable() {
+    return enable_;
+  }
+
+  public static final int NUM_REPLICAS_FIELD_NUMBER = 2;
+  private int numReplicas_;
+  /**
+   * <code>int32 num_replicas = 2;</code>
+   */
+  public int getNumReplicas() {
+    return numReplicas_;
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    if (enable_ != false) {
+      output.writeBool(1, enable_);
+    }
+    if (numReplicas_ != 0) {
+      output.writeInt32(2, numReplicas_);
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (enable_ != false) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeBoolSize(1, enable_);
+    }
+    if (numReplicas_ != 0) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt32Size(2, numReplicas_);
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.framework.AutoParallelOptions)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.framework.AutoParallelOptions other = (org.tensorflow.proto.framework.AutoParallelOptions) obj;
+
+    if (getEnable()
+        != other.getEnable()) return false;
+    if (getNumReplicas()
+        != other.getNumReplicas()) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    hash = (37 * hash) + ENABLE_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashBoolean(
+        getEnable());
+    hash = (37 * hash) + NUM_REPLICAS_FIELD_NUMBER;
+    hash = (53 * hash) + getNumReplicas();
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.framework.AutoParallelOptions parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.AutoParallelOptions parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.AutoParallelOptions parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.AutoParallelOptions parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.AutoParallelOptions parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.AutoParallelOptions parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.AutoParallelOptions parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.AutoParallelOptions parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.AutoParallelOptions parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.AutoParallelOptions parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.AutoParallelOptions parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.AutoParallelOptions parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.framework.AutoParallelOptions prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * Protobuf type {@code tensorflow.AutoParallelOptions}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.AutoParallelOptions)
+      org.tensorflow.proto.framework.AutoParallelOptionsOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.RewriterConfigProtos.internal_static_tensorflow_AutoParallelOptions_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.RewriterConfigProtos.internal_static_tensorflow_AutoParallelOptions_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.AutoParallelOptions.class, org.tensorflow.proto.framework.AutoParallelOptions.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.framework.AutoParallelOptions.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      enable_ = false;
+
+      numReplicas_ = 0;
+
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.framework.RewriterConfigProtos.internal_static_tensorflow_AutoParallelOptions_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.AutoParallelOptions getDefaultInstanceForType() {
+      return org.tensorflow.proto.framework.AutoParallelOptions.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.AutoParallelOptions build() {
+      org.tensorflow.proto.framework.AutoParallelOptions result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.AutoParallelOptions buildPartial() {
+      org.tensorflow.proto.framework.AutoParallelOptions result = new org.tensorflow.proto.framework.AutoParallelOptions(this);
+      result.enable_ = enable_;
+      result.numReplicas_ = numReplicas_;
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.framework.AutoParallelOptions) {
+        return mergeFrom((org.tensorflow.proto.framework.AutoParallelOptions)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.framework.AutoParallelOptions other) {
+      if (other == org.tensorflow.proto.framework.AutoParallelOptions.getDefaultInstance()) return this;
+      if (other.getEnable() != false) {
+        setEnable(other.getEnable());
+      }
+      if (other.getNumReplicas() != 0) {
+        setNumReplicas(other.getNumReplicas());
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.framework.AutoParallelOptions parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.framework.AutoParallelOptions) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+
+    private boolean enable_ ;
+    /**
+     * <code>bool enable = 1;</code>
+     */
+    public boolean getEnable() {
+      return enable_;
+    }
+    /**
+     * <code>bool enable = 1;</code>
+     */
+    public Builder setEnable(boolean value) {
+      
+      enable_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>bool enable = 1;</code>
+     */
+    public Builder clearEnable() {
+      
+      enable_ = false;
+      onChanged();
+      return this;
+    }
+
+    private int numReplicas_ ;
+    /**
+     * <code>int32 num_replicas = 2;</code>
+     */
+    public int getNumReplicas() {
+      return numReplicas_;
+    }
+    /**
+     * <code>int32 num_replicas = 2;</code>
+     */
+    public Builder setNumReplicas(int value) {
+      
+      numReplicas_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>int32 num_replicas = 2;</code>
+     */
+    public Builder clearNumReplicas() {
+      
+      numReplicas_ = 0;
+      onChanged();
+      return this;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.AutoParallelOptions)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.AutoParallelOptions)
+  private static final org.tensorflow.proto.framework.AutoParallelOptions DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.AutoParallelOptions();
+  }
+
+  public static org.tensorflow.proto.framework.AutoParallelOptions getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<AutoParallelOptions>
+      PARSER = new com.google.protobuf.AbstractParser<AutoParallelOptions>() {
+    @java.lang.Override
+    public AutoParallelOptions parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new AutoParallelOptions(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<AutoParallelOptions> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<AutoParallelOptions> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.framework.AutoParallelOptions getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/AutoParallelOptionsOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/AutoParallelOptionsOrBuilder.java
new file mode 100644
index 00000000000..0b954c6477c
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/AutoParallelOptionsOrBuilder.java
@@ -0,0 +1,19 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/rewriter_config.proto
+
+package org.tensorflow.proto.framework;
+
+public interface AutoParallelOptionsOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.AutoParallelOptions)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <code>bool enable = 1;</code>
+   */
+  boolean getEnable();
+
+  /**
+   * <code>int32 num_replicas = 2;</code>
+   */
+  int getNumReplicas();
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/BoundedTensorSpecProto.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/BoundedTensorSpecProto.java
new file mode 100644
index 00000000000..2c21852a18c
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/BoundedTensorSpecProto.java
@@ -0,0 +1,1182 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/struct.proto
+
+package org.tensorflow.proto.framework;
+
+/**
+ * <pre>
+ * A protobuf to represent tf.BoundedTensorSpec.
+ * </pre>
+ *
+ * Protobuf type {@code tensorflow.BoundedTensorSpecProto}
+ */
+public  final class BoundedTensorSpecProto extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.BoundedTensorSpecProto)
+    BoundedTensorSpecProtoOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use BoundedTensorSpecProto.newBuilder() to construct.
+  private BoundedTensorSpecProto(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private BoundedTensorSpecProto() {
+    name_ = "";
+    dtype_ = 0;
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new BoundedTensorSpecProto();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private BoundedTensorSpecProto(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 10: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            name_ = s;
+            break;
+          }
+          case 18: {
+            org.tensorflow.proto.framework.TensorShapeProto.Builder subBuilder = null;
+            if (shape_ != null) {
+              subBuilder = shape_.toBuilder();
+            }
+            shape_ = input.readMessage(org.tensorflow.proto.framework.TensorShapeProto.parser(), extensionRegistry);
+            if (subBuilder != null) {
+              subBuilder.mergeFrom(shape_);
+              shape_ = subBuilder.buildPartial();
+            }
+
+            break;
+          }
+          case 24: {
+            int rawValue = input.readEnum();
+
+            dtype_ = rawValue;
+            break;
+          }
+          case 34: {
+            org.tensorflow.proto.framework.TensorProto.Builder subBuilder = null;
+            if (minimum_ != null) {
+              subBuilder = minimum_.toBuilder();
+            }
+            minimum_ = input.readMessage(org.tensorflow.proto.framework.TensorProto.parser(), extensionRegistry);
+            if (subBuilder != null) {
+              subBuilder.mergeFrom(minimum_);
+              minimum_ = subBuilder.buildPartial();
+            }
+
+            break;
+          }
+          case 42: {
+            org.tensorflow.proto.framework.TensorProto.Builder subBuilder = null;
+            if (maximum_ != null) {
+              subBuilder = maximum_.toBuilder();
+            }
+            maximum_ = input.readMessage(org.tensorflow.proto.framework.TensorProto.parser(), extensionRegistry);
+            if (subBuilder != null) {
+              subBuilder.mergeFrom(maximum_);
+              maximum_ = subBuilder.buildPartial();
+            }
+
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.framework.StructProtos.internal_static_tensorflow_BoundedTensorSpecProto_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.framework.StructProtos.internal_static_tensorflow_BoundedTensorSpecProto_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.framework.BoundedTensorSpecProto.class, org.tensorflow.proto.framework.BoundedTensorSpecProto.Builder.class);
+  }
+
+  public static final int NAME_FIELD_NUMBER = 1;
+  private volatile java.lang.Object name_;
+  /**
+   * <code>string name = 1;</code>
+   */
+  public java.lang.String getName() {
+    java.lang.Object ref = name_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      name_ = s;
+      return s;
+    }
+  }
+  /**
+   * <code>string name = 1;</code>
+   */
+  public com.google.protobuf.ByteString
+      getNameBytes() {
+    java.lang.Object ref = name_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      name_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int SHAPE_FIELD_NUMBER = 2;
+  private org.tensorflow.proto.framework.TensorShapeProto shape_;
+  /**
+   * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+   */
+  public boolean hasShape() {
+    return shape_ != null;
+  }
+  /**
+   * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+   */
+  public org.tensorflow.proto.framework.TensorShapeProto getShape() {
+    return shape_ == null ? org.tensorflow.proto.framework.TensorShapeProto.getDefaultInstance() : shape_;
+  }
+  /**
+   * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+   */
+  public org.tensorflow.proto.framework.TensorShapeProtoOrBuilder getShapeOrBuilder() {
+    return getShape();
+  }
+
+  public static final int DTYPE_FIELD_NUMBER = 3;
+  private int dtype_;
+  /**
+   * <code>.tensorflow.DataType dtype = 3;</code>
+   */
+  public int getDtypeValue() {
+    return dtype_;
+  }
+  /**
+   * <code>.tensorflow.DataType dtype = 3;</code>
+   */
+  public org.tensorflow.proto.framework.DataType getDtype() {
+    @SuppressWarnings("deprecation")
+    org.tensorflow.proto.framework.DataType result = org.tensorflow.proto.framework.DataType.valueOf(dtype_);
+    return result == null ? org.tensorflow.proto.framework.DataType.UNRECOGNIZED : result;
+  }
+
+  public static final int MINIMUM_FIELD_NUMBER = 4;
+  private org.tensorflow.proto.framework.TensorProto minimum_;
+  /**
+   * <code>.tensorflow.TensorProto minimum = 4;</code>
+   */
+  public boolean hasMinimum() {
+    return minimum_ != null;
+  }
+  /**
+   * <code>.tensorflow.TensorProto minimum = 4;</code>
+   */
+  public org.tensorflow.proto.framework.TensorProto getMinimum() {
+    return minimum_ == null ? org.tensorflow.proto.framework.TensorProto.getDefaultInstance() : minimum_;
+  }
+  /**
+   * <code>.tensorflow.TensorProto minimum = 4;</code>
+   */
+  public org.tensorflow.proto.framework.TensorProtoOrBuilder getMinimumOrBuilder() {
+    return getMinimum();
+  }
+
+  public static final int MAXIMUM_FIELD_NUMBER = 5;
+  private org.tensorflow.proto.framework.TensorProto maximum_;
+  /**
+   * <code>.tensorflow.TensorProto maximum = 5;</code>
+   */
+  public boolean hasMaximum() {
+    return maximum_ != null;
+  }
+  /**
+   * <code>.tensorflow.TensorProto maximum = 5;</code>
+   */
+  public org.tensorflow.proto.framework.TensorProto getMaximum() {
+    return maximum_ == null ? org.tensorflow.proto.framework.TensorProto.getDefaultInstance() : maximum_;
+  }
+  /**
+   * <code>.tensorflow.TensorProto maximum = 5;</code>
+   */
+  public org.tensorflow.proto.framework.TensorProtoOrBuilder getMaximumOrBuilder() {
+    return getMaximum();
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    if (!getNameBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 1, name_);
+    }
+    if (shape_ != null) {
+      output.writeMessage(2, getShape());
+    }
+    if (dtype_ != org.tensorflow.proto.framework.DataType.DT_INVALID.getNumber()) {
+      output.writeEnum(3, dtype_);
+    }
+    if (minimum_ != null) {
+      output.writeMessage(4, getMinimum());
+    }
+    if (maximum_ != null) {
+      output.writeMessage(5, getMaximum());
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (!getNameBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, name_);
+    }
+    if (shape_ != null) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(2, getShape());
+    }
+    if (dtype_ != org.tensorflow.proto.framework.DataType.DT_INVALID.getNumber()) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeEnumSize(3, dtype_);
+    }
+    if (minimum_ != null) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(4, getMinimum());
+    }
+    if (maximum_ != null) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(5, getMaximum());
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.framework.BoundedTensorSpecProto)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.framework.BoundedTensorSpecProto other = (org.tensorflow.proto.framework.BoundedTensorSpecProto) obj;
+
+    if (!getName()
+        .equals(other.getName())) return false;
+    if (hasShape() != other.hasShape()) return false;
+    if (hasShape()) {
+      if (!getShape()
+          .equals(other.getShape())) return false;
+    }
+    if (dtype_ != other.dtype_) return false;
+    if (hasMinimum() != other.hasMinimum()) return false;
+    if (hasMinimum()) {
+      if (!getMinimum()
+          .equals(other.getMinimum())) return false;
+    }
+    if (hasMaximum() != other.hasMaximum()) return false;
+    if (hasMaximum()) {
+      if (!getMaximum()
+          .equals(other.getMaximum())) return false;
+    }
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    hash = (37 * hash) + NAME_FIELD_NUMBER;
+    hash = (53 * hash) + getName().hashCode();
+    if (hasShape()) {
+      hash = (37 * hash) + SHAPE_FIELD_NUMBER;
+      hash = (53 * hash) + getShape().hashCode();
+    }
+    hash = (37 * hash) + DTYPE_FIELD_NUMBER;
+    hash = (53 * hash) + dtype_;
+    if (hasMinimum()) {
+      hash = (37 * hash) + MINIMUM_FIELD_NUMBER;
+      hash = (53 * hash) + getMinimum().hashCode();
+    }
+    if (hasMaximum()) {
+      hash = (37 * hash) + MAXIMUM_FIELD_NUMBER;
+      hash = (53 * hash) + getMaximum().hashCode();
+    }
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.framework.BoundedTensorSpecProto parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.BoundedTensorSpecProto parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.BoundedTensorSpecProto parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.BoundedTensorSpecProto parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.BoundedTensorSpecProto parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.BoundedTensorSpecProto parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.BoundedTensorSpecProto parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.BoundedTensorSpecProto parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.BoundedTensorSpecProto parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.BoundedTensorSpecProto parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.BoundedTensorSpecProto parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.BoundedTensorSpecProto parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.framework.BoundedTensorSpecProto prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * <pre>
+   * A protobuf to represent tf.BoundedTensorSpec.
+   * </pre>
+   *
+   * Protobuf type {@code tensorflow.BoundedTensorSpecProto}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.BoundedTensorSpecProto)
+      org.tensorflow.proto.framework.BoundedTensorSpecProtoOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.StructProtos.internal_static_tensorflow_BoundedTensorSpecProto_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.StructProtos.internal_static_tensorflow_BoundedTensorSpecProto_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.BoundedTensorSpecProto.class, org.tensorflow.proto.framework.BoundedTensorSpecProto.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.framework.BoundedTensorSpecProto.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      name_ = "";
+
+      if (shapeBuilder_ == null) {
+        shape_ = null;
+      } else {
+        shape_ = null;
+        shapeBuilder_ = null;
+      }
+      dtype_ = 0;
+
+      if (minimumBuilder_ == null) {
+        minimum_ = null;
+      } else {
+        minimum_ = null;
+        minimumBuilder_ = null;
+      }
+      if (maximumBuilder_ == null) {
+        maximum_ = null;
+      } else {
+        maximum_ = null;
+        maximumBuilder_ = null;
+      }
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.framework.StructProtos.internal_static_tensorflow_BoundedTensorSpecProto_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.BoundedTensorSpecProto getDefaultInstanceForType() {
+      return org.tensorflow.proto.framework.BoundedTensorSpecProto.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.BoundedTensorSpecProto build() {
+      org.tensorflow.proto.framework.BoundedTensorSpecProto result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.BoundedTensorSpecProto buildPartial() {
+      org.tensorflow.proto.framework.BoundedTensorSpecProto result = new org.tensorflow.proto.framework.BoundedTensorSpecProto(this);
+      result.name_ = name_;
+      if (shapeBuilder_ == null) {
+        result.shape_ = shape_;
+      } else {
+        result.shape_ = shapeBuilder_.build();
+      }
+      result.dtype_ = dtype_;
+      if (minimumBuilder_ == null) {
+        result.minimum_ = minimum_;
+      } else {
+        result.minimum_ = minimumBuilder_.build();
+      }
+      if (maximumBuilder_ == null) {
+        result.maximum_ = maximum_;
+      } else {
+        result.maximum_ = maximumBuilder_.build();
+      }
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.framework.BoundedTensorSpecProto) {
+        return mergeFrom((org.tensorflow.proto.framework.BoundedTensorSpecProto)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.framework.BoundedTensorSpecProto other) {
+      if (other == org.tensorflow.proto.framework.BoundedTensorSpecProto.getDefaultInstance()) return this;
+      if (!other.getName().isEmpty()) {
+        name_ = other.name_;
+        onChanged();
+      }
+      if (other.hasShape()) {
+        mergeShape(other.getShape());
+      }
+      if (other.dtype_ != 0) {
+        setDtypeValue(other.getDtypeValue());
+      }
+      if (other.hasMinimum()) {
+        mergeMinimum(other.getMinimum());
+      }
+      if (other.hasMaximum()) {
+        mergeMaximum(other.getMaximum());
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.framework.BoundedTensorSpecProto parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.framework.BoundedTensorSpecProto) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+
+    private java.lang.Object name_ = "";
+    /**
+     * <code>string name = 1;</code>
+     */
+    public java.lang.String getName() {
+      java.lang.Object ref = name_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        name_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <code>string name = 1;</code>
+     */
+    public com.google.protobuf.ByteString
+        getNameBytes() {
+      java.lang.Object ref = name_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        name_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <code>string name = 1;</code>
+     */
+    public Builder setName(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      name_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>string name = 1;</code>
+     */
+    public Builder clearName() {
+      
+      name_ = getDefaultInstance().getName();
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>string name = 1;</code>
+     */
+    public Builder setNameBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      name_ = value;
+      onChanged();
+      return this;
+    }
+
+    private org.tensorflow.proto.framework.TensorShapeProto shape_;
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.TensorShapeProto, org.tensorflow.proto.framework.TensorShapeProto.Builder, org.tensorflow.proto.framework.TensorShapeProtoOrBuilder> shapeBuilder_;
+    /**
+     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+     */
+    public boolean hasShape() {
+      return shapeBuilder_ != null || shape_ != null;
+    }
+    /**
+     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+     */
+    public org.tensorflow.proto.framework.TensorShapeProto getShape() {
+      if (shapeBuilder_ == null) {
+        return shape_ == null ? org.tensorflow.proto.framework.TensorShapeProto.getDefaultInstance() : shape_;
+      } else {
+        return shapeBuilder_.getMessage();
+      }
+    }
+    /**
+     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+     */
+    public Builder setShape(org.tensorflow.proto.framework.TensorShapeProto value) {
+      if (shapeBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        shape_ = value;
+        onChanged();
+      } else {
+        shapeBuilder_.setMessage(value);
+      }
+
+      return this;
+    }
+    /**
+     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+     */
+    public Builder setShape(
+        org.tensorflow.proto.framework.TensorShapeProto.Builder builderForValue) {
+      if (shapeBuilder_ == null) {
+        shape_ = builderForValue.build();
+        onChanged();
+      } else {
+        shapeBuilder_.setMessage(builderForValue.build());
+      }
+
+      return this;
+    }
+    /**
+     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+     */
+    public Builder mergeShape(org.tensorflow.proto.framework.TensorShapeProto value) {
+      if (shapeBuilder_ == null) {
+        if (shape_ != null) {
+          shape_ =
+            org.tensorflow.proto.framework.TensorShapeProto.newBuilder(shape_).mergeFrom(value).buildPartial();
+        } else {
+          shape_ = value;
+        }
+        onChanged();
+      } else {
+        shapeBuilder_.mergeFrom(value);
+      }
+
+      return this;
+    }
+    /**
+     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+     */
+    public Builder clearShape() {
+      if (shapeBuilder_ == null) {
+        shape_ = null;
+        onChanged();
+      } else {
+        shape_ = null;
+        shapeBuilder_ = null;
+      }
+
+      return this;
+    }
+    /**
+     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+     */
+    public org.tensorflow.proto.framework.TensorShapeProto.Builder getShapeBuilder() {
+      
+      onChanged();
+      return getShapeFieldBuilder().getBuilder();
+    }
+    /**
+     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+     */
+    public org.tensorflow.proto.framework.TensorShapeProtoOrBuilder getShapeOrBuilder() {
+      if (shapeBuilder_ != null) {
+        return shapeBuilder_.getMessageOrBuilder();
+      } else {
+        return shape_ == null ?
+            org.tensorflow.proto.framework.TensorShapeProto.getDefaultInstance() : shape_;
+      }
+    }
+    /**
+     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+     */
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.TensorShapeProto, org.tensorflow.proto.framework.TensorShapeProto.Builder, org.tensorflow.proto.framework.TensorShapeProtoOrBuilder> 
+        getShapeFieldBuilder() {
+      if (shapeBuilder_ == null) {
+        shapeBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+            org.tensorflow.proto.framework.TensorShapeProto, org.tensorflow.proto.framework.TensorShapeProto.Builder, org.tensorflow.proto.framework.TensorShapeProtoOrBuilder>(
+                getShape(),
+                getParentForChildren(),
+                isClean());
+        shape_ = null;
+      }
+      return shapeBuilder_;
+    }
+
+    private int dtype_ = 0;
+    /**
+     * <code>.tensorflow.DataType dtype = 3;</code>
+     */
+    public int getDtypeValue() {
+      return dtype_;
+    }
+    /**
+     * <code>.tensorflow.DataType dtype = 3;</code>
+     */
+    public Builder setDtypeValue(int value) {
+      dtype_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>.tensorflow.DataType dtype = 3;</code>
+     */
+    public org.tensorflow.proto.framework.DataType getDtype() {
+      @SuppressWarnings("deprecation")
+      org.tensorflow.proto.framework.DataType result = org.tensorflow.proto.framework.DataType.valueOf(dtype_);
+      return result == null ? org.tensorflow.proto.framework.DataType.UNRECOGNIZED : result;
+    }
+    /**
+     * <code>.tensorflow.DataType dtype = 3;</code>
+     */
+    public Builder setDtype(org.tensorflow.proto.framework.DataType value) {
+      if (value == null) {
+        throw new NullPointerException();
+      }
+      
+      dtype_ = value.getNumber();
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>.tensorflow.DataType dtype = 3;</code>
+     */
+    public Builder clearDtype() {
+      
+      dtype_ = 0;
+      onChanged();
+      return this;
+    }
+
+    private org.tensorflow.proto.framework.TensorProto minimum_;
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.TensorProto, org.tensorflow.proto.framework.TensorProto.Builder, org.tensorflow.proto.framework.TensorProtoOrBuilder> minimumBuilder_;
+    /**
+     * <code>.tensorflow.TensorProto minimum = 4;</code>
+     */
+    public boolean hasMinimum() {
+      return minimumBuilder_ != null || minimum_ != null;
+    }
+    /**
+     * <code>.tensorflow.TensorProto minimum = 4;</code>
+     */
+    public org.tensorflow.proto.framework.TensorProto getMinimum() {
+      if (minimumBuilder_ == null) {
+        return minimum_ == null ? org.tensorflow.proto.framework.TensorProto.getDefaultInstance() : minimum_;
+      } else {
+        return minimumBuilder_.getMessage();
+      }
+    }
+    /**
+     * <code>.tensorflow.TensorProto minimum = 4;</code>
+     */
+    public Builder setMinimum(org.tensorflow.proto.framework.TensorProto value) {
+      if (minimumBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        minimum_ = value;
+        onChanged();
+      } else {
+        minimumBuilder_.setMessage(value);
+      }
+
+      return this;
+    }
+    /**
+     * <code>.tensorflow.TensorProto minimum = 4;</code>
+     */
+    public Builder setMinimum(
+        org.tensorflow.proto.framework.TensorProto.Builder builderForValue) {
+      if (minimumBuilder_ == null) {
+        minimum_ = builderForValue.build();
+        onChanged();
+      } else {
+        minimumBuilder_.setMessage(builderForValue.build());
+      }
+
+      return this;
+    }
+    /**
+     * <code>.tensorflow.TensorProto minimum = 4;</code>
+     */
+    public Builder mergeMinimum(org.tensorflow.proto.framework.TensorProto value) {
+      if (minimumBuilder_ == null) {
+        if (minimum_ != null) {
+          minimum_ =
+            org.tensorflow.proto.framework.TensorProto.newBuilder(minimum_).mergeFrom(value).buildPartial();
+        } else {
+          minimum_ = value;
+        }
+        onChanged();
+      } else {
+        minimumBuilder_.mergeFrom(value);
+      }
+
+      return this;
+    }
+    /**
+     * <code>.tensorflow.TensorProto minimum = 4;</code>
+     */
+    public Builder clearMinimum() {
+      if (minimumBuilder_ == null) {
+        minimum_ = null;
+        onChanged();
+      } else {
+        minimum_ = null;
+        minimumBuilder_ = null;
+      }
+
+      return this;
+    }
+    /**
+     * <code>.tensorflow.TensorProto minimum = 4;</code>
+     */
+    public org.tensorflow.proto.framework.TensorProto.Builder getMinimumBuilder() {
+      
+      onChanged();
+      return getMinimumFieldBuilder().getBuilder();
+    }
+    /**
+     * <code>.tensorflow.TensorProto minimum = 4;</code>
+     */
+    public org.tensorflow.proto.framework.TensorProtoOrBuilder getMinimumOrBuilder() {
+      if (minimumBuilder_ != null) {
+        return minimumBuilder_.getMessageOrBuilder();
+      } else {
+        return minimum_ == null ?
+            org.tensorflow.proto.framework.TensorProto.getDefaultInstance() : minimum_;
+      }
+    }
+    /**
+     * <code>.tensorflow.TensorProto minimum = 4;</code>
+     */
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.TensorProto, org.tensorflow.proto.framework.TensorProto.Builder, org.tensorflow.proto.framework.TensorProtoOrBuilder> 
+        getMinimumFieldBuilder() {
+      if (minimumBuilder_ == null) {
+        minimumBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+            org.tensorflow.proto.framework.TensorProto, org.tensorflow.proto.framework.TensorProto.Builder, org.tensorflow.proto.framework.TensorProtoOrBuilder>(
+                getMinimum(),
+                getParentForChildren(),
+                isClean());
+        minimum_ = null;
+      }
+      return minimumBuilder_;
+    }
+
+    private org.tensorflow.proto.framework.TensorProto maximum_;
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.TensorProto, org.tensorflow.proto.framework.TensorProto.Builder, org.tensorflow.proto.framework.TensorProtoOrBuilder> maximumBuilder_;
+    /**
+     * <code>.tensorflow.TensorProto maximum = 5;</code>
+     */
+    public boolean hasMaximum() {
+      return maximumBuilder_ != null || maximum_ != null;
+    }
+    /**
+     * <code>.tensorflow.TensorProto maximum = 5;</code>
+     */
+    public org.tensorflow.proto.framework.TensorProto getMaximum() {
+      if (maximumBuilder_ == null) {
+        return maximum_ == null ? org.tensorflow.proto.framework.TensorProto.getDefaultInstance() : maximum_;
+      } else {
+        return maximumBuilder_.getMessage();
+      }
+    }
+    /**
+     * <code>.tensorflow.TensorProto maximum = 5;</code>
+     */
+    public Builder setMaximum(org.tensorflow.proto.framework.TensorProto value) {
+      if (maximumBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        maximum_ = value;
+        onChanged();
+      } else {
+        maximumBuilder_.setMessage(value);
+      }
+
+      return this;
+    }
+    /**
+     * <code>.tensorflow.TensorProto maximum = 5;</code>
+     */
+    public Builder setMaximum(
+        org.tensorflow.proto.framework.TensorProto.Builder builderForValue) {
+      if (maximumBuilder_ == null) {
+        maximum_ = builderForValue.build();
+        onChanged();
+      } else {
+        maximumBuilder_.setMessage(builderForValue.build());
+      }
+
+      return this;
+    }
+    /**
+     * <code>.tensorflow.TensorProto maximum = 5;</code>
+     */
+    public Builder mergeMaximum(org.tensorflow.proto.framework.TensorProto value) {
+      if (maximumBuilder_ == null) {
+        if (maximum_ != null) {
+          maximum_ =
+            org.tensorflow.proto.framework.TensorProto.newBuilder(maximum_).mergeFrom(value).buildPartial();
+        } else {
+          maximum_ = value;
+        }
+        onChanged();
+      } else {
+        maximumBuilder_.mergeFrom(value);
+      }
+
+      return this;
+    }
+    /**
+     * <code>.tensorflow.TensorProto maximum = 5;</code>
+     */
+    public Builder clearMaximum() {
+      if (maximumBuilder_ == null) {
+        maximum_ = null;
+        onChanged();
+      } else {
+        maximum_ = null;
+        maximumBuilder_ = null;
+      }
+
+      return this;
+    }
+    /**
+     * <code>.tensorflow.TensorProto maximum = 5;</code>
+     */
+    public org.tensorflow.proto.framework.TensorProto.Builder getMaximumBuilder() {
+      
+      onChanged();
+      return getMaximumFieldBuilder().getBuilder();
+    }
+    /**
+     * <code>.tensorflow.TensorProto maximum = 5;</code>
+     */
+    public org.tensorflow.proto.framework.TensorProtoOrBuilder getMaximumOrBuilder() {
+      if (maximumBuilder_ != null) {
+        return maximumBuilder_.getMessageOrBuilder();
+      } else {
+        return maximum_ == null ?
+            org.tensorflow.proto.framework.TensorProto.getDefaultInstance() : maximum_;
+      }
+    }
+    /**
+     * <code>.tensorflow.TensorProto maximum = 5;</code>
+     */
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.TensorProto, org.tensorflow.proto.framework.TensorProto.Builder, org.tensorflow.proto.framework.TensorProtoOrBuilder> 
+        getMaximumFieldBuilder() {
+      if (maximumBuilder_ == null) {
+        maximumBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+            org.tensorflow.proto.framework.TensorProto, org.tensorflow.proto.framework.TensorProto.Builder, org.tensorflow.proto.framework.TensorProtoOrBuilder>(
+                getMaximum(),
+                getParentForChildren(),
+                isClean());
+        maximum_ = null;
+      }
+      return maximumBuilder_;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.BoundedTensorSpecProto)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.BoundedTensorSpecProto)
+  private static final org.tensorflow.proto.framework.BoundedTensorSpecProto DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.BoundedTensorSpecProto();
+  }
+
+  public static org.tensorflow.proto.framework.BoundedTensorSpecProto getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<BoundedTensorSpecProto>
+      PARSER = new com.google.protobuf.AbstractParser<BoundedTensorSpecProto>() {
+    @java.lang.Override
+    public BoundedTensorSpecProto parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new BoundedTensorSpecProto(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<BoundedTensorSpecProto> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<BoundedTensorSpecProto> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.framework.BoundedTensorSpecProto getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/BoundedTensorSpecProtoOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/BoundedTensorSpecProtoOrBuilder.java
new file mode 100644
index 00000000000..d4b34e04e02
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/BoundedTensorSpecProtoOrBuilder.java
@@ -0,0 +1,67 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/struct.proto
+
+package org.tensorflow.proto.framework;
+
+public interface BoundedTensorSpecProtoOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.BoundedTensorSpecProto)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <code>string name = 1;</code>
+   */
+  java.lang.String getName();
+  /**
+   * <code>string name = 1;</code>
+   */
+  com.google.protobuf.ByteString
+      getNameBytes();
+
+  /**
+   * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+   */
+  boolean hasShape();
+  /**
+   * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+   */
+  org.tensorflow.proto.framework.TensorShapeProto getShape();
+  /**
+   * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+   */
+  org.tensorflow.proto.framework.TensorShapeProtoOrBuilder getShapeOrBuilder();
+
+  /**
+   * <code>.tensorflow.DataType dtype = 3;</code>
+   */
+  int getDtypeValue();
+  /**
+   * <code>.tensorflow.DataType dtype = 3;</code>
+   */
+  org.tensorflow.proto.framework.DataType getDtype();
+
+  /**
+   * <code>.tensorflow.TensorProto minimum = 4;</code>
+   */
+  boolean hasMinimum();
+  /**
+   * <code>.tensorflow.TensorProto minimum = 4;</code>
+   */
+  org.tensorflow.proto.framework.TensorProto getMinimum();
+  /**
+   * <code>.tensorflow.TensorProto minimum = 4;</code>
+   */
+  org.tensorflow.proto.framework.TensorProtoOrBuilder getMinimumOrBuilder();
+
+  /**
+   * <code>.tensorflow.TensorProto maximum = 5;</code>
+   */
+  boolean hasMaximum();
+  /**
+   * <code>.tensorflow.TensorProto maximum = 5;</code>
+   */
+  org.tensorflow.proto.framework.TensorProto getMaximum();
+  /**
+   * <code>.tensorflow.TensorProto maximum = 5;</code>
+   */
+  org.tensorflow.proto.framework.TensorProtoOrBuilder getMaximumOrBuilder();
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/CallableOptions.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/CallableOptions.java
new file mode 100644
index 00000000000..1128fdc60f8
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/CallableOptions.java
@@ -0,0 +1,2902 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/config.proto
+
+package org.tensorflow.proto.framework;
+
+/**
+ * <pre>
+ * Defines a subgraph in another `GraphDef` as a set of feed points and nodes
+ * to be fetched or executed.
+ * Compare with the arguments to `Session::Run()`.
+ * </pre>
+ *
+ * Protobuf type {@code tensorflow.CallableOptions}
+ */
+public  final class CallableOptions extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.CallableOptions)
+    CallableOptionsOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use CallableOptions.newBuilder() to construct.
+  private CallableOptions(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private CallableOptions() {
+    feed_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+    fetch_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+    target_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+    tensorConnection_ = java.util.Collections.emptyList();
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new CallableOptions();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private CallableOptions(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    int mutable_bitField0_ = 0;
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 10: {
+            java.lang.String s = input.readStringRequireUtf8();
+            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
+              feed_ = new com.google.protobuf.LazyStringArrayList();
+              mutable_bitField0_ |= 0x00000001;
+            }
+            feed_.add(s);
+            break;
+          }
+          case 18: {
+            java.lang.String s = input.readStringRequireUtf8();
+            if (!((mutable_bitField0_ & 0x00000002) != 0)) {
+              fetch_ = new com.google.protobuf.LazyStringArrayList();
+              mutable_bitField0_ |= 0x00000002;
+            }
+            fetch_.add(s);
+            break;
+          }
+          case 26: {
+            java.lang.String s = input.readStringRequireUtf8();
+            if (!((mutable_bitField0_ & 0x00000004) != 0)) {
+              target_ = new com.google.protobuf.LazyStringArrayList();
+              mutable_bitField0_ |= 0x00000004;
+            }
+            target_.add(s);
+            break;
+          }
+          case 34: {
+            org.tensorflow.proto.framework.RunOptions.Builder subBuilder = null;
+            if (runOptions_ != null) {
+              subBuilder = runOptions_.toBuilder();
+            }
+            runOptions_ = input.readMessage(org.tensorflow.proto.framework.RunOptions.parser(), extensionRegistry);
+            if (subBuilder != null) {
+              subBuilder.mergeFrom(runOptions_);
+              runOptions_ = subBuilder.buildPartial();
+            }
+
+            break;
+          }
+          case 42: {
+            if (!((mutable_bitField0_ & 0x00000008) != 0)) {
+              tensorConnection_ = new java.util.ArrayList<org.tensorflow.proto.framework.TensorConnection>();
+              mutable_bitField0_ |= 0x00000008;
+            }
+            tensorConnection_.add(
+                input.readMessage(org.tensorflow.proto.framework.TensorConnection.parser(), extensionRegistry));
+            break;
+          }
+          case 50: {
+            if (!((mutable_bitField0_ & 0x00000010) != 0)) {
+              feedDevices_ = com.google.protobuf.MapField.newMapField(
+                  FeedDevicesDefaultEntryHolder.defaultEntry);
+              mutable_bitField0_ |= 0x00000010;
+            }
+            com.google.protobuf.MapEntry<java.lang.String, java.lang.String>
+            feedDevices__ = input.readMessage(
+                FeedDevicesDefaultEntryHolder.defaultEntry.getParserForType(), extensionRegistry);
+            feedDevices_.getMutableMap().put(
+                feedDevices__.getKey(), feedDevices__.getValue());
+            break;
+          }
+          case 58: {
+            if (!((mutable_bitField0_ & 0x00000020) != 0)) {
+              fetchDevices_ = com.google.protobuf.MapField.newMapField(
+                  FetchDevicesDefaultEntryHolder.defaultEntry);
+              mutable_bitField0_ |= 0x00000020;
+            }
+            com.google.protobuf.MapEntry<java.lang.String, java.lang.String>
+            fetchDevices__ = input.readMessage(
+                FetchDevicesDefaultEntryHolder.defaultEntry.getParserForType(), extensionRegistry);
+            fetchDevices_.getMutableMap().put(
+                fetchDevices__.getKey(), fetchDevices__.getValue());
+            break;
+          }
+          case 64: {
+
+            fetchSkipSync_ = input.readBool();
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      if (((mutable_bitField0_ & 0x00000001) != 0)) {
+        feed_ = feed_.getUnmodifiableView();
+      }
+      if (((mutable_bitField0_ & 0x00000002) != 0)) {
+        fetch_ = fetch_.getUnmodifiableView();
+      }
+      if (((mutable_bitField0_ & 0x00000004) != 0)) {
+        target_ = target_.getUnmodifiableView();
+      }
+      if (((mutable_bitField0_ & 0x00000008) != 0)) {
+        tensorConnection_ = java.util.Collections.unmodifiableList(tensorConnection_);
+      }
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_CallableOptions_descriptor;
+  }
+
+  @SuppressWarnings({"rawtypes"})
+  @java.lang.Override
+  protected com.google.protobuf.MapField internalGetMapField(
+      int number) {
+    switch (number) {
+      case 6:
+        return internalGetFeedDevices();
+      case 7:
+        return internalGetFetchDevices();
+      default:
+        throw new RuntimeException(
+            "Invalid map field number: " + number);
+    }
+  }
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_CallableOptions_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.framework.CallableOptions.class, org.tensorflow.proto.framework.CallableOptions.Builder.class);
+  }
+
+  public static final int FEED_FIELD_NUMBER = 1;
+  private com.google.protobuf.LazyStringList feed_;
+  /**
+   * <pre>
+   * Tensors to be fed in the callable. Each feed is the name of a tensor.
+   * </pre>
+   *
+   * <code>repeated string feed = 1;</code>
+   */
+  public com.google.protobuf.ProtocolStringList
+      getFeedList() {
+    return feed_;
+  }
+  /**
+   * <pre>
+   * Tensors to be fed in the callable. Each feed is the name of a tensor.
+   * </pre>
+   *
+   * <code>repeated string feed = 1;</code>
+   */
+  public int getFeedCount() {
+    return feed_.size();
+  }
+  /**
+   * <pre>
+   * Tensors to be fed in the callable. Each feed is the name of a tensor.
+   * </pre>
+   *
+   * <code>repeated string feed = 1;</code>
+   */
+  public java.lang.String getFeed(int index) {
+    return feed_.get(index);
+  }
+  /**
+   * <pre>
+   * Tensors to be fed in the callable. Each feed is the name of a tensor.
+   * </pre>
+   *
+   * <code>repeated string feed = 1;</code>
+   */
+  public com.google.protobuf.ByteString
+      getFeedBytes(int index) {
+    return feed_.getByteString(index);
+  }
+
+  public static final int FETCH_FIELD_NUMBER = 2;
+  private com.google.protobuf.LazyStringList fetch_;
+  /**
+   * <pre>
+   * Fetches. A list of tensor names. The caller of the callable expects a
+   * tensor to be returned for each fetch[i] (see RunStepResponse.tensor). The
+   * order of specified fetches does not change the execution order.
+   * </pre>
+   *
+   * <code>repeated string fetch = 2;</code>
+   */
+  public com.google.protobuf.ProtocolStringList
+      getFetchList() {
+    return fetch_;
+  }
+  /**
+   * <pre>
+   * Fetches. A list of tensor names. The caller of the callable expects a
+   * tensor to be returned for each fetch[i] (see RunStepResponse.tensor). The
+   * order of specified fetches does not change the execution order.
+   * </pre>
+   *
+   * <code>repeated string fetch = 2;</code>
+   */
+  public int getFetchCount() {
+    return fetch_.size();
+  }
+  /**
+   * <pre>
+   * Fetches. A list of tensor names. The caller of the callable expects a
+   * tensor to be returned for each fetch[i] (see RunStepResponse.tensor). The
+   * order of specified fetches does not change the execution order.
+   * </pre>
+   *
+   * <code>repeated string fetch = 2;</code>
+   */
+  public java.lang.String getFetch(int index) {
+    return fetch_.get(index);
+  }
+  /**
+   * <pre>
+   * Fetches. A list of tensor names. The caller of the callable expects a
+   * tensor to be returned for each fetch[i] (see RunStepResponse.tensor). The
+   * order of specified fetches does not change the execution order.
+   * </pre>
+   *
+   * <code>repeated string fetch = 2;</code>
+   */
+  public com.google.protobuf.ByteString
+      getFetchBytes(int index) {
+    return fetch_.getByteString(index);
+  }
+
+  public static final int TARGET_FIELD_NUMBER = 3;
+  private com.google.protobuf.LazyStringList target_;
+  /**
+   * <pre>
+   * Target Nodes. A list of node names. The named nodes will be run by the
+   * callable but their outputs will not be returned.
+   * </pre>
+   *
+   * <code>repeated string target = 3;</code>
+   */
+  public com.google.protobuf.ProtocolStringList
+      getTargetList() {
+    return target_;
+  }
+  /**
+   * <pre>
+   * Target Nodes. A list of node names. The named nodes will be run by the
+   * callable but their outputs will not be returned.
+   * </pre>
+   *
+   * <code>repeated string target = 3;</code>
+   */
+  public int getTargetCount() {
+    return target_.size();
+  }
+  /**
+   * <pre>
+   * Target Nodes. A list of node names. The named nodes will be run by the
+   * callable but their outputs will not be returned.
+   * </pre>
+   *
+   * <code>repeated string target = 3;</code>
+   */
+  public java.lang.String getTarget(int index) {
+    return target_.get(index);
+  }
+  /**
+   * <pre>
+   * Target Nodes. A list of node names. The named nodes will be run by the
+   * callable but their outputs will not be returned.
+   * </pre>
+   *
+   * <code>repeated string target = 3;</code>
+   */
+  public com.google.protobuf.ByteString
+      getTargetBytes(int index) {
+    return target_.getByteString(index);
+  }
+
+  public static final int RUN_OPTIONS_FIELD_NUMBER = 4;
+  private org.tensorflow.proto.framework.RunOptions runOptions_;
+  /**
+   * <pre>
+   * Options that will be applied to each run.
+   * </pre>
+   *
+   * <code>.tensorflow.RunOptions run_options = 4;</code>
+   */
+  public boolean hasRunOptions() {
+    return runOptions_ != null;
+  }
+  /**
+   * <pre>
+   * Options that will be applied to each run.
+   * </pre>
+   *
+   * <code>.tensorflow.RunOptions run_options = 4;</code>
+   */
+  public org.tensorflow.proto.framework.RunOptions getRunOptions() {
+    return runOptions_ == null ? org.tensorflow.proto.framework.RunOptions.getDefaultInstance() : runOptions_;
+  }
+  /**
+   * <pre>
+   * Options that will be applied to each run.
+   * </pre>
+   *
+   * <code>.tensorflow.RunOptions run_options = 4;</code>
+   */
+  public org.tensorflow.proto.framework.RunOptionsOrBuilder getRunOptionsOrBuilder() {
+    return getRunOptions();
+  }
+
+  public static final int TENSOR_CONNECTION_FIELD_NUMBER = 5;
+  private java.util.List<org.tensorflow.proto.framework.TensorConnection> tensorConnection_;
+  /**
+   * <pre>
+   * Tensors to be connected in the callable. Each TensorConnection denotes
+   * a pair of tensors in the graph, between which an edge will be created
+   * in the callable.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.TensorConnection tensor_connection = 5;</code>
+   */
+  public java.util.List<org.tensorflow.proto.framework.TensorConnection> getTensorConnectionList() {
+    return tensorConnection_;
+  }
+  /**
+   * <pre>
+   * Tensors to be connected in the callable. Each TensorConnection denotes
+   * a pair of tensors in the graph, between which an edge will be created
+   * in the callable.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.TensorConnection tensor_connection = 5;</code>
+   */
+  public java.util.List<? extends org.tensorflow.proto.framework.TensorConnectionOrBuilder> 
+      getTensorConnectionOrBuilderList() {
+    return tensorConnection_;
+  }
+  /**
+   * <pre>
+   * Tensors to be connected in the callable. Each TensorConnection denotes
+   * a pair of tensors in the graph, between which an edge will be created
+   * in the callable.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.TensorConnection tensor_connection = 5;</code>
+   */
+  public int getTensorConnectionCount() {
+    return tensorConnection_.size();
+  }
+  /**
+   * <pre>
+   * Tensors to be connected in the callable. Each TensorConnection denotes
+   * a pair of tensors in the graph, between which an edge will be created
+   * in the callable.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.TensorConnection tensor_connection = 5;</code>
+   */
+  public org.tensorflow.proto.framework.TensorConnection getTensorConnection(int index) {
+    return tensorConnection_.get(index);
+  }
+  /**
+   * <pre>
+   * Tensors to be connected in the callable. Each TensorConnection denotes
+   * a pair of tensors in the graph, between which an edge will be created
+   * in the callable.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.TensorConnection tensor_connection = 5;</code>
+   */
+  public org.tensorflow.proto.framework.TensorConnectionOrBuilder getTensorConnectionOrBuilder(
+      int index) {
+    return tensorConnection_.get(index);
+  }
+
+  public static final int FEED_DEVICES_FIELD_NUMBER = 6;
+  private static final class FeedDevicesDefaultEntryHolder {
+    static final com.google.protobuf.MapEntry<
+        java.lang.String, java.lang.String> defaultEntry =
+            com.google.protobuf.MapEntry
+            .<java.lang.String, java.lang.String>newDefaultInstance(
+                org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_CallableOptions_FeedDevicesEntry_descriptor, 
+                com.google.protobuf.WireFormat.FieldType.STRING,
+                "",
+                com.google.protobuf.WireFormat.FieldType.STRING,
+                "");
+  }
+  private com.google.protobuf.MapField<
+      java.lang.String, java.lang.String> feedDevices_;
+  private com.google.protobuf.MapField<java.lang.String, java.lang.String>
+  internalGetFeedDevices() {
+    if (feedDevices_ == null) {
+      return com.google.protobuf.MapField.emptyMapField(
+          FeedDevicesDefaultEntryHolder.defaultEntry);
+    }
+    return feedDevices_;
+  }
+
+  public int getFeedDevicesCount() {
+    return internalGetFeedDevices().getMap().size();
+  }
+  /**
+   * <pre>
+   * The Tensor objects fed in the callable and fetched from the callable
+   * are expected to be backed by host (CPU) memory by default.
+   * The options below allow changing that - feeding tensors backed by
+   * device memory, or returning tensors that are backed by device memory.
+   * The maps below map the name of a feed/fetch tensor (which appears in
+   * 'feed' or 'fetch' fields above), to the fully qualified name of the device
+   * owning the memory backing the contents of the tensor.
+   * For example, creating a callable with the following options:
+   * CallableOptions {
+   *   feed: "a:0"
+   *   feed: "b:0"
+   *   fetch: "x:0"
+   *   fetch: "y:0"
+   *   feed_devices: {
+   *     "a:0": "/job:localhost/replica:0/task:0/device:GPU:0"
+   *   }
+   *   fetch_devices: {
+   *     "y:0": "/job:localhost/replica:0/task:0/device:GPU:0"
+   *  }
+   * }
+   * means that the Callable expects:
+   * - The first argument ("a:0") is a Tensor backed by GPU memory.
+   * - The second argument ("b:0") is a Tensor backed by host memory.
+   * and of its return values:
+   * - The first output ("x:0") will be backed by host memory.
+   * - The second output ("y:0") will be backed by GPU memory.
+   * FEEDS:
+   * It is the responsibility of the caller to ensure that the memory of the fed
+   * tensors will be correctly initialized and synchronized before it is
+   * accessed by operations executed during the call to Session::RunCallable().
+   * This is typically ensured by using the TensorFlow memory allocators
+   * (Device::GetAllocator()) to create the Tensor to be fed.
+   * Alternatively, for CUDA-enabled GPU devices, this typically means that the
+   * operation that produced the contents of the tensor has completed, i.e., the
+   * CUDA stream has been synchronized (e.g., via cuCtxSynchronize() or
+   * cuStreamSynchronize()).
+   * </pre>
+   *
+   * <code>map&lt;string, string&gt; feed_devices = 6;</code>
+   */
+
+  public boolean containsFeedDevices(
+      java.lang.String key) {
+    if (key == null) { throw new java.lang.NullPointerException(); }
+    return internalGetFeedDevices().getMap().containsKey(key);
+  }
+  /**
+   * Use {@link #getFeedDevicesMap()} instead.
+   */
+  @java.lang.Deprecated
+  public java.util.Map<java.lang.String, java.lang.String> getFeedDevices() {
+    return getFeedDevicesMap();
+  }
+  /**
+   * <pre>
+   * The Tensor objects fed in the callable and fetched from the callable
+   * are expected to be backed by host (CPU) memory by default.
+   * The options below allow changing that - feeding tensors backed by
+   * device memory, or returning tensors that are backed by device memory.
+   * The maps below map the name of a feed/fetch tensor (which appears in
+   * 'feed' or 'fetch' fields above), to the fully qualified name of the device
+   * owning the memory backing the contents of the tensor.
+   * For example, creating a callable with the following options:
+   * CallableOptions {
+   *   feed: "a:0"
+   *   feed: "b:0"
+   *   fetch: "x:0"
+   *   fetch: "y:0"
+   *   feed_devices: {
+   *     "a:0": "/job:localhost/replica:0/task:0/device:GPU:0"
+   *   }
+   *   fetch_devices: {
+   *     "y:0": "/job:localhost/replica:0/task:0/device:GPU:0"
+   *  }
+   * }
+   * means that the Callable expects:
+   * - The first argument ("a:0") is a Tensor backed by GPU memory.
+   * - The second argument ("b:0") is a Tensor backed by host memory.
+   * and of its return values:
+   * - The first output ("x:0") will be backed by host memory.
+   * - The second output ("y:0") will be backed by GPU memory.
+   * FEEDS:
+   * It is the responsibility of the caller to ensure that the memory of the fed
+   * tensors will be correctly initialized and synchronized before it is
+   * accessed by operations executed during the call to Session::RunCallable().
+   * This is typically ensured by using the TensorFlow memory allocators
+   * (Device::GetAllocator()) to create the Tensor to be fed.
+   * Alternatively, for CUDA-enabled GPU devices, this typically means that the
+   * operation that produced the contents of the tensor has completed, i.e., the
+   * CUDA stream has been synchronized (e.g., via cuCtxSynchronize() or
+   * cuStreamSynchronize()).
+   * </pre>
+   *
+   * <code>map&lt;string, string&gt; feed_devices = 6;</code>
+   */
+
+  public java.util.Map<java.lang.String, java.lang.String> getFeedDevicesMap() {
+    return internalGetFeedDevices().getMap();
+  }
+  /**
+   * <pre>
+   * The Tensor objects fed in the callable and fetched from the callable
+   * are expected to be backed by host (CPU) memory by default.
+   * The options below allow changing that - feeding tensors backed by
+   * device memory, or returning tensors that are backed by device memory.
+   * The maps below map the name of a feed/fetch tensor (which appears in
+   * 'feed' or 'fetch' fields above), to the fully qualified name of the device
+   * owning the memory backing the contents of the tensor.
+   * For example, creating a callable with the following options:
+   * CallableOptions {
+   *   feed: "a:0"
+   *   feed: "b:0"
+   *   fetch: "x:0"
+   *   fetch: "y:0"
+   *   feed_devices: {
+   *     "a:0": "/job:localhost/replica:0/task:0/device:GPU:0"
+   *   }
+   *   fetch_devices: {
+   *     "y:0": "/job:localhost/replica:0/task:0/device:GPU:0"
+   *  }
+   * }
+   * means that the Callable expects:
+   * - The first argument ("a:0") is a Tensor backed by GPU memory.
+   * - The second argument ("b:0") is a Tensor backed by host memory.
+   * and of its return values:
+   * - The first output ("x:0") will be backed by host memory.
+   * - The second output ("y:0") will be backed by GPU memory.
+   * FEEDS:
+   * It is the responsibility of the caller to ensure that the memory of the fed
+   * tensors will be correctly initialized and synchronized before it is
+   * accessed by operations executed during the call to Session::RunCallable().
+   * This is typically ensured by using the TensorFlow memory allocators
+   * (Device::GetAllocator()) to create the Tensor to be fed.
+   * Alternatively, for CUDA-enabled GPU devices, this typically means that the
+   * operation that produced the contents of the tensor has completed, i.e., the
+   * CUDA stream has been synchronized (e.g., via cuCtxSynchronize() or
+   * cuStreamSynchronize()).
+   * </pre>
+   *
+   * <code>map&lt;string, string&gt; feed_devices = 6;</code>
+   */
+
+  public java.lang.String getFeedDevicesOrDefault(
+      java.lang.String key,
+      java.lang.String defaultValue) {
+    if (key == null) { throw new java.lang.NullPointerException(); }
+    java.util.Map<java.lang.String, java.lang.String> map =
+        internalGetFeedDevices().getMap();
+    return map.containsKey(key) ? map.get(key) : defaultValue;
+  }
+  /**
+   * <pre>
+   * The Tensor objects fed in the callable and fetched from the callable
+   * are expected to be backed by host (CPU) memory by default.
+   * The options below allow changing that - feeding tensors backed by
+   * device memory, or returning tensors that are backed by device memory.
+   * The maps below map the name of a feed/fetch tensor (which appears in
+   * 'feed' or 'fetch' fields above), to the fully qualified name of the device
+   * owning the memory backing the contents of the tensor.
+   * For example, creating a callable with the following options:
+   * CallableOptions {
+   *   feed: "a:0"
+   *   feed: "b:0"
+   *   fetch: "x:0"
+   *   fetch: "y:0"
+   *   feed_devices: {
+   *     "a:0": "/job:localhost/replica:0/task:0/device:GPU:0"
+   *   }
+   *   fetch_devices: {
+   *     "y:0": "/job:localhost/replica:0/task:0/device:GPU:0"
+   *  }
+   * }
+   * means that the Callable expects:
+   * - The first argument ("a:0") is a Tensor backed by GPU memory.
+   * - The second argument ("b:0") is a Tensor backed by host memory.
+   * and of its return values:
+   * - The first output ("x:0") will be backed by host memory.
+   * - The second output ("y:0") will be backed by GPU memory.
+   * FEEDS:
+   * It is the responsibility of the caller to ensure that the memory of the fed
+   * tensors will be correctly initialized and synchronized before it is
+   * accessed by operations executed during the call to Session::RunCallable().
+   * This is typically ensured by using the TensorFlow memory allocators
+   * (Device::GetAllocator()) to create the Tensor to be fed.
+   * Alternatively, for CUDA-enabled GPU devices, this typically means that the
+   * operation that produced the contents of the tensor has completed, i.e., the
+   * CUDA stream has been synchronized (e.g., via cuCtxSynchronize() or
+   * cuStreamSynchronize()).
+   * </pre>
+   *
+   * <code>map&lt;string, string&gt; feed_devices = 6;</code>
+   */
+
+  public java.lang.String getFeedDevicesOrThrow(
+      java.lang.String key) {
+    if (key == null) { throw new java.lang.NullPointerException(); }
+    java.util.Map<java.lang.String, java.lang.String> map =
+        internalGetFeedDevices().getMap();
+    if (!map.containsKey(key)) {
+      throw new java.lang.IllegalArgumentException();
+    }
+    return map.get(key);
+  }
+
+  public static final int FETCH_DEVICES_FIELD_NUMBER = 7;
+  private static final class FetchDevicesDefaultEntryHolder {
+    static final com.google.protobuf.MapEntry<
+        java.lang.String, java.lang.String> defaultEntry =
+            com.google.protobuf.MapEntry
+            .<java.lang.String, java.lang.String>newDefaultInstance(
+                org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_CallableOptions_FetchDevicesEntry_descriptor, 
+                com.google.protobuf.WireFormat.FieldType.STRING,
+                "",
+                com.google.protobuf.WireFormat.FieldType.STRING,
+                "");
+  }
+  private com.google.protobuf.MapField<
+      java.lang.String, java.lang.String> fetchDevices_;
+  private com.google.protobuf.MapField<java.lang.String, java.lang.String>
+  internalGetFetchDevices() {
+    if (fetchDevices_ == null) {
+      return com.google.protobuf.MapField.emptyMapField(
+          FetchDevicesDefaultEntryHolder.defaultEntry);
+    }
+    return fetchDevices_;
+  }
+
+  public int getFetchDevicesCount() {
+    return internalGetFetchDevices().getMap().size();
+  }
+  /**
+   * <code>map&lt;string, string&gt; fetch_devices = 7;</code>
+   */
+
+  public boolean containsFetchDevices(
+      java.lang.String key) {
+    if (key == null) { throw new java.lang.NullPointerException(); }
+    return internalGetFetchDevices().getMap().containsKey(key);
+  }
+  /**
+   * Use {@link #getFetchDevicesMap()} instead.
+   */
+  @java.lang.Deprecated
+  public java.util.Map<java.lang.String, java.lang.String> getFetchDevices() {
+    return getFetchDevicesMap();
+  }
+  /**
+   * <code>map&lt;string, string&gt; fetch_devices = 7;</code>
+   */
+
+  public java.util.Map<java.lang.String, java.lang.String> getFetchDevicesMap() {
+    return internalGetFetchDevices().getMap();
+  }
+  /**
+   * <code>map&lt;string, string&gt; fetch_devices = 7;</code>
+   */
+
+  public java.lang.String getFetchDevicesOrDefault(
+      java.lang.String key,
+      java.lang.String defaultValue) {
+    if (key == null) { throw new java.lang.NullPointerException(); }
+    java.util.Map<java.lang.String, java.lang.String> map =
+        internalGetFetchDevices().getMap();
+    return map.containsKey(key) ? map.get(key) : defaultValue;
+  }
+  /**
+   * <code>map&lt;string, string&gt; fetch_devices = 7;</code>
+   */
+
+  public java.lang.String getFetchDevicesOrThrow(
+      java.lang.String key) {
+    if (key == null) { throw new java.lang.NullPointerException(); }
+    java.util.Map<java.lang.String, java.lang.String> map =
+        internalGetFetchDevices().getMap();
+    if (!map.containsKey(key)) {
+      throw new java.lang.IllegalArgumentException();
+    }
+    return map.get(key);
+  }
+
+  public static final int FETCH_SKIP_SYNC_FIELD_NUMBER = 8;
+  private boolean fetchSkipSync_;
+  /**
+   * <pre>
+   * By default, RunCallable() will synchronize the GPU stream before returning
+   * fetched tensors on a GPU device, to ensure that the values in those tensors
+   * have been produced. This simplifies interacting with the tensors, but
+   * potentially incurs a performance hit.
+   * If this options is set to true, the caller is responsible for ensuring
+   * that the values in the fetched tensors have been produced before they are
+   * used. The caller can do this by invoking `Device::Sync()` on the underlying
+   * device(s), or by feeding the tensors back to the same Session using
+   * `feed_devices` with the same corresponding device name.
+   * </pre>
+   *
+   * <code>bool fetch_skip_sync = 8;</code>
+   */
+  public boolean getFetchSkipSync() {
+    return fetchSkipSync_;
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    for (int i = 0; i < feed_.size(); i++) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 1, feed_.getRaw(i));
+    }
+    for (int i = 0; i < fetch_.size(); i++) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 2, fetch_.getRaw(i));
+    }
+    for (int i = 0; i < target_.size(); i++) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 3, target_.getRaw(i));
+    }
+    if (runOptions_ != null) {
+      output.writeMessage(4, getRunOptions());
+    }
+    for (int i = 0; i < tensorConnection_.size(); i++) {
+      output.writeMessage(5, tensorConnection_.get(i));
+    }
+    com.google.protobuf.GeneratedMessageV3
+      .serializeStringMapTo(
+        output,
+        internalGetFeedDevices(),
+        FeedDevicesDefaultEntryHolder.defaultEntry,
+        6);
+    com.google.protobuf.GeneratedMessageV3
+      .serializeStringMapTo(
+        output,
+        internalGetFetchDevices(),
+        FetchDevicesDefaultEntryHolder.defaultEntry,
+        7);
+    if (fetchSkipSync_ != false) {
+      output.writeBool(8, fetchSkipSync_);
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    {
+      int dataSize = 0;
+      for (int i = 0; i < feed_.size(); i++) {
+        dataSize += computeStringSizeNoTag(feed_.getRaw(i));
+      }
+      size += dataSize;
+      size += 1 * getFeedList().size();
+    }
+    {
+      int dataSize = 0;
+      for (int i = 0; i < fetch_.size(); i++) {
+        dataSize += computeStringSizeNoTag(fetch_.getRaw(i));
+      }
+      size += dataSize;
+      size += 1 * getFetchList().size();
+    }
+    {
+      int dataSize = 0;
+      for (int i = 0; i < target_.size(); i++) {
+        dataSize += computeStringSizeNoTag(target_.getRaw(i));
+      }
+      size += dataSize;
+      size += 1 * getTargetList().size();
+    }
+    if (runOptions_ != null) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(4, getRunOptions());
+    }
+    for (int i = 0; i < tensorConnection_.size(); i++) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(5, tensorConnection_.get(i));
+    }
+    for (java.util.Map.Entry<java.lang.String, java.lang.String> entry
+         : internalGetFeedDevices().getMap().entrySet()) {
+      com.google.protobuf.MapEntry<java.lang.String, java.lang.String>
+      feedDevices__ = FeedDevicesDefaultEntryHolder.defaultEntry.newBuilderForType()
+          .setKey(entry.getKey())
+          .setValue(entry.getValue())
+          .build();
+      size += com.google.protobuf.CodedOutputStream
+          .computeMessageSize(6, feedDevices__);
+    }
+    for (java.util.Map.Entry<java.lang.String, java.lang.String> entry
+         : internalGetFetchDevices().getMap().entrySet()) {
+      com.google.protobuf.MapEntry<java.lang.String, java.lang.String>
+      fetchDevices__ = FetchDevicesDefaultEntryHolder.defaultEntry.newBuilderForType()
+          .setKey(entry.getKey())
+          .setValue(entry.getValue())
+          .build();
+      size += com.google.protobuf.CodedOutputStream
+          .computeMessageSize(7, fetchDevices__);
+    }
+    if (fetchSkipSync_ != false) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeBoolSize(8, fetchSkipSync_);
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.framework.CallableOptions)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.framework.CallableOptions other = (org.tensorflow.proto.framework.CallableOptions) obj;
+
+    if (!getFeedList()
+        .equals(other.getFeedList())) return false;
+    if (!getFetchList()
+        .equals(other.getFetchList())) return false;
+    if (!getTargetList()
+        .equals(other.getTargetList())) return false;
+    if (hasRunOptions() != other.hasRunOptions()) return false;
+    if (hasRunOptions()) {
+      if (!getRunOptions()
+          .equals(other.getRunOptions())) return false;
+    }
+    if (!getTensorConnectionList()
+        .equals(other.getTensorConnectionList())) return false;
+    if (!internalGetFeedDevices().equals(
+        other.internalGetFeedDevices())) return false;
+    if (!internalGetFetchDevices().equals(
+        other.internalGetFetchDevices())) return false;
+    if (getFetchSkipSync()
+        != other.getFetchSkipSync()) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    if (getFeedCount() > 0) {
+      hash = (37 * hash) + FEED_FIELD_NUMBER;
+      hash = (53 * hash) + getFeedList().hashCode();
+    }
+    if (getFetchCount() > 0) {
+      hash = (37 * hash) + FETCH_FIELD_NUMBER;
+      hash = (53 * hash) + getFetchList().hashCode();
+    }
+    if (getTargetCount() > 0) {
+      hash = (37 * hash) + TARGET_FIELD_NUMBER;
+      hash = (53 * hash) + getTargetList().hashCode();
+    }
+    if (hasRunOptions()) {
+      hash = (37 * hash) + RUN_OPTIONS_FIELD_NUMBER;
+      hash = (53 * hash) + getRunOptions().hashCode();
+    }
+    if (getTensorConnectionCount() > 0) {
+      hash = (37 * hash) + TENSOR_CONNECTION_FIELD_NUMBER;
+      hash = (53 * hash) + getTensorConnectionList().hashCode();
+    }
+    if (!internalGetFeedDevices().getMap().isEmpty()) {
+      hash = (37 * hash) + FEED_DEVICES_FIELD_NUMBER;
+      hash = (53 * hash) + internalGetFeedDevices().hashCode();
+    }
+    if (!internalGetFetchDevices().getMap().isEmpty()) {
+      hash = (37 * hash) + FETCH_DEVICES_FIELD_NUMBER;
+      hash = (53 * hash) + internalGetFetchDevices().hashCode();
+    }
+    hash = (37 * hash) + FETCH_SKIP_SYNC_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashBoolean(
+        getFetchSkipSync());
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.framework.CallableOptions parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.CallableOptions parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.CallableOptions parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.CallableOptions parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.CallableOptions parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.CallableOptions parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.CallableOptions parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.CallableOptions parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.CallableOptions parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.CallableOptions parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.CallableOptions parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.CallableOptions parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.framework.CallableOptions prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * <pre>
+   * Defines a subgraph in another `GraphDef` as a set of feed points and nodes
+   * to be fetched or executed.
+   * Compare with the arguments to `Session::Run()`.
+   * </pre>
+   *
+   * Protobuf type {@code tensorflow.CallableOptions}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.CallableOptions)
+      org.tensorflow.proto.framework.CallableOptionsOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_CallableOptions_descriptor;
+    }
+
+    @SuppressWarnings({"rawtypes"})
+    protected com.google.protobuf.MapField internalGetMapField(
+        int number) {
+      switch (number) {
+        case 6:
+          return internalGetFeedDevices();
+        case 7:
+          return internalGetFetchDevices();
+        default:
+          throw new RuntimeException(
+              "Invalid map field number: " + number);
+      }
+    }
+    @SuppressWarnings({"rawtypes"})
+    protected com.google.protobuf.MapField internalGetMutableMapField(
+        int number) {
+      switch (number) {
+        case 6:
+          return internalGetMutableFeedDevices();
+        case 7:
+          return internalGetMutableFetchDevices();
+        default:
+          throw new RuntimeException(
+              "Invalid map field number: " + number);
+      }
+    }
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_CallableOptions_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.CallableOptions.class, org.tensorflow.proto.framework.CallableOptions.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.framework.CallableOptions.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+        getTensorConnectionFieldBuilder();
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      feed_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+      bitField0_ = (bitField0_ & ~0x00000001);
+      fetch_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+      bitField0_ = (bitField0_ & ~0x00000002);
+      target_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+      bitField0_ = (bitField0_ & ~0x00000004);
+      if (runOptionsBuilder_ == null) {
+        runOptions_ = null;
+      } else {
+        runOptions_ = null;
+        runOptionsBuilder_ = null;
+      }
+      if (tensorConnectionBuilder_ == null) {
+        tensorConnection_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000008);
+      } else {
+        tensorConnectionBuilder_.clear();
+      }
+      internalGetMutableFeedDevices().clear();
+      internalGetMutableFetchDevices().clear();
+      fetchSkipSync_ = false;
+
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_CallableOptions_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.CallableOptions getDefaultInstanceForType() {
+      return org.tensorflow.proto.framework.CallableOptions.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.CallableOptions build() {
+      org.tensorflow.proto.framework.CallableOptions result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.CallableOptions buildPartial() {
+      org.tensorflow.proto.framework.CallableOptions result = new org.tensorflow.proto.framework.CallableOptions(this);
+      int from_bitField0_ = bitField0_;
+      if (((bitField0_ & 0x00000001) != 0)) {
+        feed_ = feed_.getUnmodifiableView();
+        bitField0_ = (bitField0_ & ~0x00000001);
+      }
+      result.feed_ = feed_;
+      if (((bitField0_ & 0x00000002) != 0)) {
+        fetch_ = fetch_.getUnmodifiableView();
+        bitField0_ = (bitField0_ & ~0x00000002);
+      }
+      result.fetch_ = fetch_;
+      if (((bitField0_ & 0x00000004) != 0)) {
+        target_ = target_.getUnmodifiableView();
+        bitField0_ = (bitField0_ & ~0x00000004);
+      }
+      result.target_ = target_;
+      if (runOptionsBuilder_ == null) {
+        result.runOptions_ = runOptions_;
+      } else {
+        result.runOptions_ = runOptionsBuilder_.build();
+      }
+      if (tensorConnectionBuilder_ == null) {
+        if (((bitField0_ & 0x00000008) != 0)) {
+          tensorConnection_ = java.util.Collections.unmodifiableList(tensorConnection_);
+          bitField0_ = (bitField0_ & ~0x00000008);
+        }
+        result.tensorConnection_ = tensorConnection_;
+      } else {
+        result.tensorConnection_ = tensorConnectionBuilder_.build();
+      }
+      result.feedDevices_ = internalGetFeedDevices();
+      result.feedDevices_.makeImmutable();
+      result.fetchDevices_ = internalGetFetchDevices();
+      result.fetchDevices_.makeImmutable();
+      result.fetchSkipSync_ = fetchSkipSync_;
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.framework.CallableOptions) {
+        return mergeFrom((org.tensorflow.proto.framework.CallableOptions)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.framework.CallableOptions other) {
+      if (other == org.tensorflow.proto.framework.CallableOptions.getDefaultInstance()) return this;
+      if (!other.feed_.isEmpty()) {
+        if (feed_.isEmpty()) {
+          feed_ = other.feed_;
+          bitField0_ = (bitField0_ & ~0x00000001);
+        } else {
+          ensureFeedIsMutable();
+          feed_.addAll(other.feed_);
+        }
+        onChanged();
+      }
+      if (!other.fetch_.isEmpty()) {
+        if (fetch_.isEmpty()) {
+          fetch_ = other.fetch_;
+          bitField0_ = (bitField0_ & ~0x00000002);
+        } else {
+          ensureFetchIsMutable();
+          fetch_.addAll(other.fetch_);
+        }
+        onChanged();
+      }
+      if (!other.target_.isEmpty()) {
+        if (target_.isEmpty()) {
+          target_ = other.target_;
+          bitField0_ = (bitField0_ & ~0x00000004);
+        } else {
+          ensureTargetIsMutable();
+          target_.addAll(other.target_);
+        }
+        onChanged();
+      }
+      if (other.hasRunOptions()) {
+        mergeRunOptions(other.getRunOptions());
+      }
+      if (tensorConnectionBuilder_ == null) {
+        if (!other.tensorConnection_.isEmpty()) {
+          if (tensorConnection_.isEmpty()) {
+            tensorConnection_ = other.tensorConnection_;
+            bitField0_ = (bitField0_ & ~0x00000008);
+          } else {
+            ensureTensorConnectionIsMutable();
+            tensorConnection_.addAll(other.tensorConnection_);
+          }
+          onChanged();
+        }
+      } else {
+        if (!other.tensorConnection_.isEmpty()) {
+          if (tensorConnectionBuilder_.isEmpty()) {
+            tensorConnectionBuilder_.dispose();
+            tensorConnectionBuilder_ = null;
+            tensorConnection_ = other.tensorConnection_;
+            bitField0_ = (bitField0_ & ~0x00000008);
+            tensorConnectionBuilder_ = 
+              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
+                 getTensorConnectionFieldBuilder() : null;
+          } else {
+            tensorConnectionBuilder_.addAllMessages(other.tensorConnection_);
+          }
+        }
+      }
+      internalGetMutableFeedDevices().mergeFrom(
+          other.internalGetFeedDevices());
+      internalGetMutableFetchDevices().mergeFrom(
+          other.internalGetFetchDevices());
+      if (other.getFetchSkipSync() != false) {
+        setFetchSkipSync(other.getFetchSkipSync());
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.framework.CallableOptions parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.framework.CallableOptions) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+    private int bitField0_;
+
+    private com.google.protobuf.LazyStringList feed_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+    private void ensureFeedIsMutable() {
+      if (!((bitField0_ & 0x00000001) != 0)) {
+        feed_ = new com.google.protobuf.LazyStringArrayList(feed_);
+        bitField0_ |= 0x00000001;
+       }
+    }
+    /**
+     * <pre>
+     * Tensors to be fed in the callable. Each feed is the name of a tensor.
+     * </pre>
+     *
+     * <code>repeated string feed = 1;</code>
+     */
+    public com.google.protobuf.ProtocolStringList
+        getFeedList() {
+      return feed_.getUnmodifiableView();
+    }
+    /**
+     * <pre>
+     * Tensors to be fed in the callable. Each feed is the name of a tensor.
+     * </pre>
+     *
+     * <code>repeated string feed = 1;</code>
+     */
+    public int getFeedCount() {
+      return feed_.size();
+    }
+    /**
+     * <pre>
+     * Tensors to be fed in the callable. Each feed is the name of a tensor.
+     * </pre>
+     *
+     * <code>repeated string feed = 1;</code>
+     */
+    public java.lang.String getFeed(int index) {
+      return feed_.get(index);
+    }
+    /**
+     * <pre>
+     * Tensors to be fed in the callable. Each feed is the name of a tensor.
+     * </pre>
+     *
+     * <code>repeated string feed = 1;</code>
+     */
+    public com.google.protobuf.ByteString
+        getFeedBytes(int index) {
+      return feed_.getByteString(index);
+    }
+    /**
+     * <pre>
+     * Tensors to be fed in the callable. Each feed is the name of a tensor.
+     * </pre>
+     *
+     * <code>repeated string feed = 1;</code>
+     */
+    public Builder setFeed(
+        int index, java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  ensureFeedIsMutable();
+      feed_.set(index, value);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Tensors to be fed in the callable. Each feed is the name of a tensor.
+     * </pre>
+     *
+     * <code>repeated string feed = 1;</code>
+     */
+    public Builder addFeed(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  ensureFeedIsMutable();
+      feed_.add(value);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Tensors to be fed in the callable. Each feed is the name of a tensor.
+     * </pre>
+     *
+     * <code>repeated string feed = 1;</code>
+     */
+    public Builder addAllFeed(
+        java.lang.Iterable<java.lang.String> values) {
+      ensureFeedIsMutable();
+      com.google.protobuf.AbstractMessageLite.Builder.addAll(
+          values, feed_);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Tensors to be fed in the callable. Each feed is the name of a tensor.
+     * </pre>
+     *
+     * <code>repeated string feed = 1;</code>
+     */
+    public Builder clearFeed() {
+      feed_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+      bitField0_ = (bitField0_ & ~0x00000001);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Tensors to be fed in the callable. Each feed is the name of a tensor.
+     * </pre>
+     *
+     * <code>repeated string feed = 1;</code>
+     */
+    public Builder addFeedBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      ensureFeedIsMutable();
+      feed_.add(value);
+      onChanged();
+      return this;
+    }
+
+    private com.google.protobuf.LazyStringList fetch_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+    private void ensureFetchIsMutable() {
+      if (!((bitField0_ & 0x00000002) != 0)) {
+        fetch_ = new com.google.protobuf.LazyStringArrayList(fetch_);
+        bitField0_ |= 0x00000002;
+       }
+    }
+    /**
+     * <pre>
+     * Fetches. A list of tensor names. The caller of the callable expects a
+     * tensor to be returned for each fetch[i] (see RunStepResponse.tensor). The
+     * order of specified fetches does not change the execution order.
+     * </pre>
+     *
+     * <code>repeated string fetch = 2;</code>
+     */
+    public com.google.protobuf.ProtocolStringList
+        getFetchList() {
+      return fetch_.getUnmodifiableView();
+    }
+    /**
+     * <pre>
+     * Fetches. A list of tensor names. The caller of the callable expects a
+     * tensor to be returned for each fetch[i] (see RunStepResponse.tensor). The
+     * order of specified fetches does not change the execution order.
+     * </pre>
+     *
+     * <code>repeated string fetch = 2;</code>
+     */
+    public int getFetchCount() {
+      return fetch_.size();
+    }
+    /**
+     * <pre>
+     * Fetches. A list of tensor names. The caller of the callable expects a
+     * tensor to be returned for each fetch[i] (see RunStepResponse.tensor). The
+     * order of specified fetches does not change the execution order.
+     * </pre>
+     *
+     * <code>repeated string fetch = 2;</code>
+     */
+    public java.lang.String getFetch(int index) {
+      return fetch_.get(index);
+    }
+    /**
+     * <pre>
+     * Fetches. A list of tensor names. The caller of the callable expects a
+     * tensor to be returned for each fetch[i] (see RunStepResponse.tensor). The
+     * order of specified fetches does not change the execution order.
+     * </pre>
+     *
+     * <code>repeated string fetch = 2;</code>
+     */
+    public com.google.protobuf.ByteString
+        getFetchBytes(int index) {
+      return fetch_.getByteString(index);
+    }
+    /**
+     * <pre>
+     * Fetches. A list of tensor names. The caller of the callable expects a
+     * tensor to be returned for each fetch[i] (see RunStepResponse.tensor). The
+     * order of specified fetches does not change the execution order.
+     * </pre>
+     *
+     * <code>repeated string fetch = 2;</code>
+     */
+    public Builder setFetch(
+        int index, java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  ensureFetchIsMutable();
+      fetch_.set(index, value);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Fetches. A list of tensor names. The caller of the callable expects a
+     * tensor to be returned for each fetch[i] (see RunStepResponse.tensor). The
+     * order of specified fetches does not change the execution order.
+     * </pre>
+     *
+     * <code>repeated string fetch = 2;</code>
+     */
+    public Builder addFetch(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  ensureFetchIsMutable();
+      fetch_.add(value);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Fetches. A list of tensor names. The caller of the callable expects a
+     * tensor to be returned for each fetch[i] (see RunStepResponse.tensor). The
+     * order of specified fetches does not change the execution order.
+     * </pre>
+     *
+     * <code>repeated string fetch = 2;</code>
+     */
+    public Builder addAllFetch(
+        java.lang.Iterable<java.lang.String> values) {
+      ensureFetchIsMutable();
+      com.google.protobuf.AbstractMessageLite.Builder.addAll(
+          values, fetch_);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Fetches. A list of tensor names. The caller of the callable expects a
+     * tensor to be returned for each fetch[i] (see RunStepResponse.tensor). The
+     * order of specified fetches does not change the execution order.
+     * </pre>
+     *
+     * <code>repeated string fetch = 2;</code>
+     */
+    public Builder clearFetch() {
+      fetch_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+      bitField0_ = (bitField0_ & ~0x00000002);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Fetches. A list of tensor names. The caller of the callable expects a
+     * tensor to be returned for each fetch[i] (see RunStepResponse.tensor). The
+     * order of specified fetches does not change the execution order.
+     * </pre>
+     *
+     * <code>repeated string fetch = 2;</code>
+     */
+    public Builder addFetchBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      ensureFetchIsMutable();
+      fetch_.add(value);
+      onChanged();
+      return this;
+    }
+
+    private com.google.protobuf.LazyStringList target_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+    private void ensureTargetIsMutable() {
+      if (!((bitField0_ & 0x00000004) != 0)) {
+        target_ = new com.google.protobuf.LazyStringArrayList(target_);
+        bitField0_ |= 0x00000004;
+       }
+    }
+    /**
+     * <pre>
+     * Target Nodes. A list of node names. The named nodes will be run by the
+     * callable but their outputs will not be returned.
+     * </pre>
+     *
+     * <code>repeated string target = 3;</code>
+     */
+    public com.google.protobuf.ProtocolStringList
+        getTargetList() {
+      return target_.getUnmodifiableView();
+    }
+    /**
+     * <pre>
+     * Target Nodes. A list of node names. The named nodes will be run by the
+     * callable but their outputs will not be returned.
+     * </pre>
+     *
+     * <code>repeated string target = 3;</code>
+     */
+    public int getTargetCount() {
+      return target_.size();
+    }
+    /**
+     * <pre>
+     * Target Nodes. A list of node names. The named nodes will be run by the
+     * callable but their outputs will not be returned.
+     * </pre>
+     *
+     * <code>repeated string target = 3;</code>
+     */
+    public java.lang.String getTarget(int index) {
+      return target_.get(index);
+    }
+    /**
+     * <pre>
+     * Target Nodes. A list of node names. The named nodes will be run by the
+     * callable but their outputs will not be returned.
+     * </pre>
+     *
+     * <code>repeated string target = 3;</code>
+     */
+    public com.google.protobuf.ByteString
+        getTargetBytes(int index) {
+      return target_.getByteString(index);
+    }
+    /**
+     * <pre>
+     * Target Nodes. A list of node names. The named nodes will be run by the
+     * callable but their outputs will not be returned.
+     * </pre>
+     *
+     * <code>repeated string target = 3;</code>
+     */
+    public Builder setTarget(
+        int index, java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  ensureTargetIsMutable();
+      target_.set(index, value);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Target Nodes. A list of node names. The named nodes will be run by the
+     * callable but their outputs will not be returned.
+     * </pre>
+     *
+     * <code>repeated string target = 3;</code>
+     */
+    public Builder addTarget(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  ensureTargetIsMutable();
+      target_.add(value);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Target Nodes. A list of node names. The named nodes will be run by the
+     * callable but their outputs will not be returned.
+     * </pre>
+     *
+     * <code>repeated string target = 3;</code>
+     */
+    public Builder addAllTarget(
+        java.lang.Iterable<java.lang.String> values) {
+      ensureTargetIsMutable();
+      com.google.protobuf.AbstractMessageLite.Builder.addAll(
+          values, target_);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Target Nodes. A list of node names. The named nodes will be run by the
+     * callable but their outputs will not be returned.
+     * </pre>
+     *
+     * <code>repeated string target = 3;</code>
+     */
+    public Builder clearTarget() {
+      target_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+      bitField0_ = (bitField0_ & ~0x00000004);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Target Nodes. A list of node names. The named nodes will be run by the
+     * callable but their outputs will not be returned.
+     * </pre>
+     *
+     * <code>repeated string target = 3;</code>
+     */
+    public Builder addTargetBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      ensureTargetIsMutable();
+      target_.add(value);
+      onChanged();
+      return this;
+    }
+
+    private org.tensorflow.proto.framework.RunOptions runOptions_;
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.RunOptions, org.tensorflow.proto.framework.RunOptions.Builder, org.tensorflow.proto.framework.RunOptionsOrBuilder> runOptionsBuilder_;
+    /**
+     * <pre>
+     * Options that will be applied to each run.
+     * </pre>
+     *
+     * <code>.tensorflow.RunOptions run_options = 4;</code>
+     */
+    public boolean hasRunOptions() {
+      return runOptionsBuilder_ != null || runOptions_ != null;
+    }
+    /**
+     * <pre>
+     * Options that will be applied to each run.
+     * </pre>
+     *
+     * <code>.tensorflow.RunOptions run_options = 4;</code>
+     */
+    public org.tensorflow.proto.framework.RunOptions getRunOptions() {
+      if (runOptionsBuilder_ == null) {
+        return runOptions_ == null ? org.tensorflow.proto.framework.RunOptions.getDefaultInstance() : runOptions_;
+      } else {
+        return runOptionsBuilder_.getMessage();
+      }
+    }
+    /**
+     * <pre>
+     * Options that will be applied to each run.
+     * </pre>
+     *
+     * <code>.tensorflow.RunOptions run_options = 4;</code>
+     */
+    public Builder setRunOptions(org.tensorflow.proto.framework.RunOptions value) {
+      if (runOptionsBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        runOptions_ = value;
+        onChanged();
+      } else {
+        runOptionsBuilder_.setMessage(value);
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Options that will be applied to each run.
+     * </pre>
+     *
+     * <code>.tensorflow.RunOptions run_options = 4;</code>
+     */
+    public Builder setRunOptions(
+        org.tensorflow.proto.framework.RunOptions.Builder builderForValue) {
+      if (runOptionsBuilder_ == null) {
+        runOptions_ = builderForValue.build();
+        onChanged();
+      } else {
+        runOptionsBuilder_.setMessage(builderForValue.build());
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Options that will be applied to each run.
+     * </pre>
+     *
+     * <code>.tensorflow.RunOptions run_options = 4;</code>
+     */
+    public Builder mergeRunOptions(org.tensorflow.proto.framework.RunOptions value) {
+      if (runOptionsBuilder_ == null) {
+        if (runOptions_ != null) {
+          runOptions_ =
+            org.tensorflow.proto.framework.RunOptions.newBuilder(runOptions_).mergeFrom(value).buildPartial();
+        } else {
+          runOptions_ = value;
+        }
+        onChanged();
+      } else {
+        runOptionsBuilder_.mergeFrom(value);
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Options that will be applied to each run.
+     * </pre>
+     *
+     * <code>.tensorflow.RunOptions run_options = 4;</code>
+     */
+    public Builder clearRunOptions() {
+      if (runOptionsBuilder_ == null) {
+        runOptions_ = null;
+        onChanged();
+      } else {
+        runOptions_ = null;
+        runOptionsBuilder_ = null;
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Options that will be applied to each run.
+     * </pre>
+     *
+     * <code>.tensorflow.RunOptions run_options = 4;</code>
+     */
+    public org.tensorflow.proto.framework.RunOptions.Builder getRunOptionsBuilder() {
+      
+      onChanged();
+      return getRunOptionsFieldBuilder().getBuilder();
+    }
+    /**
+     * <pre>
+     * Options that will be applied to each run.
+     * </pre>
+     *
+     * <code>.tensorflow.RunOptions run_options = 4;</code>
+     */
+    public org.tensorflow.proto.framework.RunOptionsOrBuilder getRunOptionsOrBuilder() {
+      if (runOptionsBuilder_ != null) {
+        return runOptionsBuilder_.getMessageOrBuilder();
+      } else {
+        return runOptions_ == null ?
+            org.tensorflow.proto.framework.RunOptions.getDefaultInstance() : runOptions_;
+      }
+    }
+    /**
+     * <pre>
+     * Options that will be applied to each run.
+     * </pre>
+     *
+     * <code>.tensorflow.RunOptions run_options = 4;</code>
+     */
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.RunOptions, org.tensorflow.proto.framework.RunOptions.Builder, org.tensorflow.proto.framework.RunOptionsOrBuilder> 
+        getRunOptionsFieldBuilder() {
+      if (runOptionsBuilder_ == null) {
+        runOptionsBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+            org.tensorflow.proto.framework.RunOptions, org.tensorflow.proto.framework.RunOptions.Builder, org.tensorflow.proto.framework.RunOptionsOrBuilder>(
+                getRunOptions(),
+                getParentForChildren(),
+                isClean());
+        runOptions_ = null;
+      }
+      return runOptionsBuilder_;
+    }
+
+    private java.util.List<org.tensorflow.proto.framework.TensorConnection> tensorConnection_ =
+      java.util.Collections.emptyList();
+    private void ensureTensorConnectionIsMutable() {
+      if (!((bitField0_ & 0x00000008) != 0)) {
+        tensorConnection_ = new java.util.ArrayList<org.tensorflow.proto.framework.TensorConnection>(tensorConnection_);
+        bitField0_ |= 0x00000008;
+       }
+    }
+
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.framework.TensorConnection, org.tensorflow.proto.framework.TensorConnection.Builder, org.tensorflow.proto.framework.TensorConnectionOrBuilder> tensorConnectionBuilder_;
+
+    /**
+     * <pre>
+     * Tensors to be connected in the callable. Each TensorConnection denotes
+     * a pair of tensors in the graph, between which an edge will be created
+     * in the callable.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorConnection tensor_connection = 5;</code>
+     */
+    public java.util.List<org.tensorflow.proto.framework.TensorConnection> getTensorConnectionList() {
+      if (tensorConnectionBuilder_ == null) {
+        return java.util.Collections.unmodifiableList(tensorConnection_);
+      } else {
+        return tensorConnectionBuilder_.getMessageList();
+      }
+    }
+    /**
+     * <pre>
+     * Tensors to be connected in the callable. Each TensorConnection denotes
+     * a pair of tensors in the graph, between which an edge will be created
+     * in the callable.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorConnection tensor_connection = 5;</code>
+     */
+    public int getTensorConnectionCount() {
+      if (tensorConnectionBuilder_ == null) {
+        return tensorConnection_.size();
+      } else {
+        return tensorConnectionBuilder_.getCount();
+      }
+    }
+    /**
+     * <pre>
+     * Tensors to be connected in the callable. Each TensorConnection denotes
+     * a pair of tensors in the graph, between which an edge will be created
+     * in the callable.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorConnection tensor_connection = 5;</code>
+     */
+    public org.tensorflow.proto.framework.TensorConnection getTensorConnection(int index) {
+      if (tensorConnectionBuilder_ == null) {
+        return tensorConnection_.get(index);
+      } else {
+        return tensorConnectionBuilder_.getMessage(index);
+      }
+    }
+    /**
+     * <pre>
+     * Tensors to be connected in the callable. Each TensorConnection denotes
+     * a pair of tensors in the graph, between which an edge will be created
+     * in the callable.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorConnection tensor_connection = 5;</code>
+     */
+    public Builder setTensorConnection(
+        int index, org.tensorflow.proto.framework.TensorConnection value) {
+      if (tensorConnectionBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureTensorConnectionIsMutable();
+        tensorConnection_.set(index, value);
+        onChanged();
+      } else {
+        tensorConnectionBuilder_.setMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Tensors to be connected in the callable. Each TensorConnection denotes
+     * a pair of tensors in the graph, between which an edge will be created
+     * in the callable.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorConnection tensor_connection = 5;</code>
+     */
+    public Builder setTensorConnection(
+        int index, org.tensorflow.proto.framework.TensorConnection.Builder builderForValue) {
+      if (tensorConnectionBuilder_ == null) {
+        ensureTensorConnectionIsMutable();
+        tensorConnection_.set(index, builderForValue.build());
+        onChanged();
+      } else {
+        tensorConnectionBuilder_.setMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Tensors to be connected in the callable. Each TensorConnection denotes
+     * a pair of tensors in the graph, between which an edge will be created
+     * in the callable.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorConnection tensor_connection = 5;</code>
+     */
+    public Builder addTensorConnection(org.tensorflow.proto.framework.TensorConnection value) {
+      if (tensorConnectionBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureTensorConnectionIsMutable();
+        tensorConnection_.add(value);
+        onChanged();
+      } else {
+        tensorConnectionBuilder_.addMessage(value);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Tensors to be connected in the callable. Each TensorConnection denotes
+     * a pair of tensors in the graph, between which an edge will be created
+     * in the callable.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorConnection tensor_connection = 5;</code>
+     */
+    public Builder addTensorConnection(
+        int index, org.tensorflow.proto.framework.TensorConnection value) {
+      if (tensorConnectionBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureTensorConnectionIsMutable();
+        tensorConnection_.add(index, value);
+        onChanged();
+      } else {
+        tensorConnectionBuilder_.addMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Tensors to be connected in the callable. Each TensorConnection denotes
+     * a pair of tensors in the graph, between which an edge will be created
+     * in the callable.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorConnection tensor_connection = 5;</code>
+     */
+    public Builder addTensorConnection(
+        org.tensorflow.proto.framework.TensorConnection.Builder builderForValue) {
+      if (tensorConnectionBuilder_ == null) {
+        ensureTensorConnectionIsMutable();
+        tensorConnection_.add(builderForValue.build());
+        onChanged();
+      } else {
+        tensorConnectionBuilder_.addMessage(builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Tensors to be connected in the callable. Each TensorConnection denotes
+     * a pair of tensors in the graph, between which an edge will be created
+     * in the callable.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorConnection tensor_connection = 5;</code>
+     */
+    public Builder addTensorConnection(
+        int index, org.tensorflow.proto.framework.TensorConnection.Builder builderForValue) {
+      if (tensorConnectionBuilder_ == null) {
+        ensureTensorConnectionIsMutable();
+        tensorConnection_.add(index, builderForValue.build());
+        onChanged();
+      } else {
+        tensorConnectionBuilder_.addMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Tensors to be connected in the callable. Each TensorConnection denotes
+     * a pair of tensors in the graph, between which an edge will be created
+     * in the callable.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorConnection tensor_connection = 5;</code>
+     */
+    public Builder addAllTensorConnection(
+        java.lang.Iterable<? extends org.tensorflow.proto.framework.TensorConnection> values) {
+      if (tensorConnectionBuilder_ == null) {
+        ensureTensorConnectionIsMutable();
+        com.google.protobuf.AbstractMessageLite.Builder.addAll(
+            values, tensorConnection_);
+        onChanged();
+      } else {
+        tensorConnectionBuilder_.addAllMessages(values);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Tensors to be connected in the callable. Each TensorConnection denotes
+     * a pair of tensors in the graph, between which an edge will be created
+     * in the callable.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorConnection tensor_connection = 5;</code>
+     */
+    public Builder clearTensorConnection() {
+      if (tensorConnectionBuilder_ == null) {
+        tensorConnection_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000008);
+        onChanged();
+      } else {
+        tensorConnectionBuilder_.clear();
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Tensors to be connected in the callable. Each TensorConnection denotes
+     * a pair of tensors in the graph, between which an edge will be created
+     * in the callable.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorConnection tensor_connection = 5;</code>
+     */
+    public Builder removeTensorConnection(int index) {
+      if (tensorConnectionBuilder_ == null) {
+        ensureTensorConnectionIsMutable();
+        tensorConnection_.remove(index);
+        onChanged();
+      } else {
+        tensorConnectionBuilder_.remove(index);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Tensors to be connected in the callable. Each TensorConnection denotes
+     * a pair of tensors in the graph, between which an edge will be created
+     * in the callable.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorConnection tensor_connection = 5;</code>
+     */
+    public org.tensorflow.proto.framework.TensorConnection.Builder getTensorConnectionBuilder(
+        int index) {
+      return getTensorConnectionFieldBuilder().getBuilder(index);
+    }
+    /**
+     * <pre>
+     * Tensors to be connected in the callable. Each TensorConnection denotes
+     * a pair of tensors in the graph, between which an edge will be created
+     * in the callable.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorConnection tensor_connection = 5;</code>
+     */
+    public org.tensorflow.proto.framework.TensorConnectionOrBuilder getTensorConnectionOrBuilder(
+        int index) {
+      if (tensorConnectionBuilder_ == null) {
+        return tensorConnection_.get(index);  } else {
+        return tensorConnectionBuilder_.getMessageOrBuilder(index);
+      }
+    }
+    /**
+     * <pre>
+     * Tensors to be connected in the callable. Each TensorConnection denotes
+     * a pair of tensors in the graph, between which an edge will be created
+     * in the callable.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorConnection tensor_connection = 5;</code>
+     */
+    public java.util.List<? extends org.tensorflow.proto.framework.TensorConnectionOrBuilder> 
+         getTensorConnectionOrBuilderList() {
+      if (tensorConnectionBuilder_ != null) {
+        return tensorConnectionBuilder_.getMessageOrBuilderList();
+      } else {
+        return java.util.Collections.unmodifiableList(tensorConnection_);
+      }
+    }
+    /**
+     * <pre>
+     * Tensors to be connected in the callable. Each TensorConnection denotes
+     * a pair of tensors in the graph, between which an edge will be created
+     * in the callable.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorConnection tensor_connection = 5;</code>
+     */
+    public org.tensorflow.proto.framework.TensorConnection.Builder addTensorConnectionBuilder() {
+      return getTensorConnectionFieldBuilder().addBuilder(
+          org.tensorflow.proto.framework.TensorConnection.getDefaultInstance());
+    }
+    /**
+     * <pre>
+     * Tensors to be connected in the callable. Each TensorConnection denotes
+     * a pair of tensors in the graph, between which an edge will be created
+     * in the callable.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorConnection tensor_connection = 5;</code>
+     */
+    public org.tensorflow.proto.framework.TensorConnection.Builder addTensorConnectionBuilder(
+        int index) {
+      return getTensorConnectionFieldBuilder().addBuilder(
+          index, org.tensorflow.proto.framework.TensorConnection.getDefaultInstance());
+    }
+    /**
+     * <pre>
+     * Tensors to be connected in the callable. Each TensorConnection denotes
+     * a pair of tensors in the graph, between which an edge will be created
+     * in the callable.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorConnection tensor_connection = 5;</code>
+     */
+    public java.util.List<org.tensorflow.proto.framework.TensorConnection.Builder> 
+         getTensorConnectionBuilderList() {
+      return getTensorConnectionFieldBuilder().getBuilderList();
+    }
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.framework.TensorConnection, org.tensorflow.proto.framework.TensorConnection.Builder, org.tensorflow.proto.framework.TensorConnectionOrBuilder> 
+        getTensorConnectionFieldBuilder() {
+      if (tensorConnectionBuilder_ == null) {
+        tensorConnectionBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
+            org.tensorflow.proto.framework.TensorConnection, org.tensorflow.proto.framework.TensorConnection.Builder, org.tensorflow.proto.framework.TensorConnectionOrBuilder>(
+                tensorConnection_,
+                ((bitField0_ & 0x00000008) != 0),
+                getParentForChildren(),
+                isClean());
+        tensorConnection_ = null;
+      }
+      return tensorConnectionBuilder_;
+    }
+
+    private com.google.protobuf.MapField<
+        java.lang.String, java.lang.String> feedDevices_;
+    private com.google.protobuf.MapField<java.lang.String, java.lang.String>
+    internalGetFeedDevices() {
+      if (feedDevices_ == null) {
+        return com.google.protobuf.MapField.emptyMapField(
+            FeedDevicesDefaultEntryHolder.defaultEntry);
+      }
+      return feedDevices_;
+    }
+    private com.google.protobuf.MapField<java.lang.String, java.lang.String>
+    internalGetMutableFeedDevices() {
+      onChanged();;
+      if (feedDevices_ == null) {
+        feedDevices_ = com.google.protobuf.MapField.newMapField(
+            FeedDevicesDefaultEntryHolder.defaultEntry);
+      }
+      if (!feedDevices_.isMutable()) {
+        feedDevices_ = feedDevices_.copy();
+      }
+      return feedDevices_;
+    }
+
+    public int getFeedDevicesCount() {
+      return internalGetFeedDevices().getMap().size();
+    }
+    /**
+     * <pre>
+     * The Tensor objects fed in the callable and fetched from the callable
+     * are expected to be backed by host (CPU) memory by default.
+     * The options below allow changing that - feeding tensors backed by
+     * device memory, or returning tensors that are backed by device memory.
+     * The maps below map the name of a feed/fetch tensor (which appears in
+     * 'feed' or 'fetch' fields above), to the fully qualified name of the device
+     * owning the memory backing the contents of the tensor.
+     * For example, creating a callable with the following options:
+     * CallableOptions {
+     *   feed: "a:0"
+     *   feed: "b:0"
+     *   fetch: "x:0"
+     *   fetch: "y:0"
+     *   feed_devices: {
+     *     "a:0": "/job:localhost/replica:0/task:0/device:GPU:0"
+     *   }
+     *   fetch_devices: {
+     *     "y:0": "/job:localhost/replica:0/task:0/device:GPU:0"
+     *  }
+     * }
+     * means that the Callable expects:
+     * - The first argument ("a:0") is a Tensor backed by GPU memory.
+     * - The second argument ("b:0") is a Tensor backed by host memory.
+     * and of its return values:
+     * - The first output ("x:0") will be backed by host memory.
+     * - The second output ("y:0") will be backed by GPU memory.
+     * FEEDS:
+     * It is the responsibility of the caller to ensure that the memory of the fed
+     * tensors will be correctly initialized and synchronized before it is
+     * accessed by operations executed during the call to Session::RunCallable().
+     * This is typically ensured by using the TensorFlow memory allocators
+     * (Device::GetAllocator()) to create the Tensor to be fed.
+     * Alternatively, for CUDA-enabled GPU devices, this typically means that the
+     * operation that produced the contents of the tensor has completed, i.e., the
+     * CUDA stream has been synchronized (e.g., via cuCtxSynchronize() or
+     * cuStreamSynchronize()).
+     * </pre>
+     *
+     * <code>map&lt;string, string&gt; feed_devices = 6;</code>
+     */
+
+    public boolean containsFeedDevices(
+        java.lang.String key) {
+      if (key == null) { throw new java.lang.NullPointerException(); }
+      return internalGetFeedDevices().getMap().containsKey(key);
+    }
+    /**
+     * Use {@link #getFeedDevicesMap()} instead.
+     */
+    @java.lang.Deprecated
+    public java.util.Map<java.lang.String, java.lang.String> getFeedDevices() {
+      return getFeedDevicesMap();
+    }
+    /**
+     * <pre>
+     * The Tensor objects fed in the callable and fetched from the callable
+     * are expected to be backed by host (CPU) memory by default.
+     * The options below allow changing that - feeding tensors backed by
+     * device memory, or returning tensors that are backed by device memory.
+     * The maps below map the name of a feed/fetch tensor (which appears in
+     * 'feed' or 'fetch' fields above), to the fully qualified name of the device
+     * owning the memory backing the contents of the tensor.
+     * For example, creating a callable with the following options:
+     * CallableOptions {
+     *   feed: "a:0"
+     *   feed: "b:0"
+     *   fetch: "x:0"
+     *   fetch: "y:0"
+     *   feed_devices: {
+     *     "a:0": "/job:localhost/replica:0/task:0/device:GPU:0"
+     *   }
+     *   fetch_devices: {
+     *     "y:0": "/job:localhost/replica:0/task:0/device:GPU:0"
+     *  }
+     * }
+     * means that the Callable expects:
+     * - The first argument ("a:0") is a Tensor backed by GPU memory.
+     * - The second argument ("b:0") is a Tensor backed by host memory.
+     * and of its return values:
+     * - The first output ("x:0") will be backed by host memory.
+     * - The second output ("y:0") will be backed by GPU memory.
+     * FEEDS:
+     * It is the responsibility of the caller to ensure that the memory of the fed
+     * tensors will be correctly initialized and synchronized before it is
+     * accessed by operations executed during the call to Session::RunCallable().
+     * This is typically ensured by using the TensorFlow memory allocators
+     * (Device::GetAllocator()) to create the Tensor to be fed.
+     * Alternatively, for CUDA-enabled GPU devices, this typically means that the
+     * operation that produced the contents of the tensor has completed, i.e., the
+     * CUDA stream has been synchronized (e.g., via cuCtxSynchronize() or
+     * cuStreamSynchronize()).
+     * </pre>
+     *
+     * <code>map&lt;string, string&gt; feed_devices = 6;</code>
+     */
+
+    public java.util.Map<java.lang.String, java.lang.String> getFeedDevicesMap() {
+      return internalGetFeedDevices().getMap();
+    }
+    /**
+     * <pre>
+     * The Tensor objects fed in the callable and fetched from the callable
+     * are expected to be backed by host (CPU) memory by default.
+     * The options below allow changing that - feeding tensors backed by
+     * device memory, or returning tensors that are backed by device memory.
+     * The maps below map the name of a feed/fetch tensor (which appears in
+     * 'feed' or 'fetch' fields above), to the fully qualified name of the device
+     * owning the memory backing the contents of the tensor.
+     * For example, creating a callable with the following options:
+     * CallableOptions {
+     *   feed: "a:0"
+     *   feed: "b:0"
+     *   fetch: "x:0"
+     *   fetch: "y:0"
+     *   feed_devices: {
+     *     "a:0": "/job:localhost/replica:0/task:0/device:GPU:0"
+     *   }
+     *   fetch_devices: {
+     *     "y:0": "/job:localhost/replica:0/task:0/device:GPU:0"
+     *  }
+     * }
+     * means that the Callable expects:
+     * - The first argument ("a:0") is a Tensor backed by GPU memory.
+     * - The second argument ("b:0") is a Tensor backed by host memory.
+     * and of its return values:
+     * - The first output ("x:0") will be backed by host memory.
+     * - The second output ("y:0") will be backed by GPU memory.
+     * FEEDS:
+     * It is the responsibility of the caller to ensure that the memory of the fed
+     * tensors will be correctly initialized and synchronized before it is
+     * accessed by operations executed during the call to Session::RunCallable().
+     * This is typically ensured by using the TensorFlow memory allocators
+     * (Device::GetAllocator()) to create the Tensor to be fed.
+     * Alternatively, for CUDA-enabled GPU devices, this typically means that the
+     * operation that produced the contents of the tensor has completed, i.e., the
+     * CUDA stream has been synchronized (e.g., via cuCtxSynchronize() or
+     * cuStreamSynchronize()).
+     * </pre>
+     *
+     * <code>map&lt;string, string&gt; feed_devices = 6;</code>
+     */
+
+    public java.lang.String getFeedDevicesOrDefault(
+        java.lang.String key,
+        java.lang.String defaultValue) {
+      if (key == null) { throw new java.lang.NullPointerException(); }
+      java.util.Map<java.lang.String, java.lang.String> map =
+          internalGetFeedDevices().getMap();
+      return map.containsKey(key) ? map.get(key) : defaultValue;
+    }
+    /**
+     * <pre>
+     * The Tensor objects fed in the callable and fetched from the callable
+     * are expected to be backed by host (CPU) memory by default.
+     * The options below allow changing that - feeding tensors backed by
+     * device memory, or returning tensors that are backed by device memory.
+     * The maps below map the name of a feed/fetch tensor (which appears in
+     * 'feed' or 'fetch' fields above), to the fully qualified name of the device
+     * owning the memory backing the contents of the tensor.
+     * For example, creating a callable with the following options:
+     * CallableOptions {
+     *   feed: "a:0"
+     *   feed: "b:0"
+     *   fetch: "x:0"
+     *   fetch: "y:0"
+     *   feed_devices: {
+     *     "a:0": "/job:localhost/replica:0/task:0/device:GPU:0"
+     *   }
+     *   fetch_devices: {
+     *     "y:0": "/job:localhost/replica:0/task:0/device:GPU:0"
+     *  }
+     * }
+     * means that the Callable expects:
+     * - The first argument ("a:0") is a Tensor backed by GPU memory.
+     * - The second argument ("b:0") is a Tensor backed by host memory.
+     * and of its return values:
+     * - The first output ("x:0") will be backed by host memory.
+     * - The second output ("y:0") will be backed by GPU memory.
+     * FEEDS:
+     * It is the responsibility of the caller to ensure that the memory of the fed
+     * tensors will be correctly initialized and synchronized before it is
+     * accessed by operations executed during the call to Session::RunCallable().
+     * This is typically ensured by using the TensorFlow memory allocators
+     * (Device::GetAllocator()) to create the Tensor to be fed.
+     * Alternatively, for CUDA-enabled GPU devices, this typically means that the
+     * operation that produced the contents of the tensor has completed, i.e., the
+     * CUDA stream has been synchronized (e.g., via cuCtxSynchronize() or
+     * cuStreamSynchronize()).
+     * </pre>
+     *
+     * <code>map&lt;string, string&gt; feed_devices = 6;</code>
+     */
+
+    public java.lang.String getFeedDevicesOrThrow(
+        java.lang.String key) {
+      if (key == null) { throw new java.lang.NullPointerException(); }
+      java.util.Map<java.lang.String, java.lang.String> map =
+          internalGetFeedDevices().getMap();
+      if (!map.containsKey(key)) {
+        throw new java.lang.IllegalArgumentException();
+      }
+      return map.get(key);
+    }
+
+    public Builder clearFeedDevices() {
+      internalGetMutableFeedDevices().getMutableMap()
+          .clear();
+      return this;
+    }
+    /**
+     * <pre>
+     * The Tensor objects fed in the callable and fetched from the callable
+     * are expected to be backed by host (CPU) memory by default.
+     * The options below allow changing that - feeding tensors backed by
+     * device memory, or returning tensors that are backed by device memory.
+     * The maps below map the name of a feed/fetch tensor (which appears in
+     * 'feed' or 'fetch' fields above), to the fully qualified name of the device
+     * owning the memory backing the contents of the tensor.
+     * For example, creating a callable with the following options:
+     * CallableOptions {
+     *   feed: "a:0"
+     *   feed: "b:0"
+     *   fetch: "x:0"
+     *   fetch: "y:0"
+     *   feed_devices: {
+     *     "a:0": "/job:localhost/replica:0/task:0/device:GPU:0"
+     *   }
+     *   fetch_devices: {
+     *     "y:0": "/job:localhost/replica:0/task:0/device:GPU:0"
+     *  }
+     * }
+     * means that the Callable expects:
+     * - The first argument ("a:0") is a Tensor backed by GPU memory.
+     * - The second argument ("b:0") is a Tensor backed by host memory.
+     * and of its return values:
+     * - The first output ("x:0") will be backed by host memory.
+     * - The second output ("y:0") will be backed by GPU memory.
+     * FEEDS:
+     * It is the responsibility of the caller to ensure that the memory of the fed
+     * tensors will be correctly initialized and synchronized before it is
+     * accessed by operations executed during the call to Session::RunCallable().
+     * This is typically ensured by using the TensorFlow memory allocators
+     * (Device::GetAllocator()) to create the Tensor to be fed.
+     * Alternatively, for CUDA-enabled GPU devices, this typically means that the
+     * operation that produced the contents of the tensor has completed, i.e., the
+     * CUDA stream has been synchronized (e.g., via cuCtxSynchronize() or
+     * cuStreamSynchronize()).
+     * </pre>
+     *
+     * <code>map&lt;string, string&gt; feed_devices = 6;</code>
+     */
+
+    public Builder removeFeedDevices(
+        java.lang.String key) {
+      if (key == null) { throw new java.lang.NullPointerException(); }
+      internalGetMutableFeedDevices().getMutableMap()
+          .remove(key);
+      return this;
+    }
+    /**
+     * Use alternate mutation accessors instead.
+     */
+    @java.lang.Deprecated
+    public java.util.Map<java.lang.String, java.lang.String>
+    getMutableFeedDevices() {
+      return internalGetMutableFeedDevices().getMutableMap();
+    }
+    /**
+     * <pre>
+     * The Tensor objects fed in the callable and fetched from the callable
+     * are expected to be backed by host (CPU) memory by default.
+     * The options below allow changing that - feeding tensors backed by
+     * device memory, or returning tensors that are backed by device memory.
+     * The maps below map the name of a feed/fetch tensor (which appears in
+     * 'feed' or 'fetch' fields above), to the fully qualified name of the device
+     * owning the memory backing the contents of the tensor.
+     * For example, creating a callable with the following options:
+     * CallableOptions {
+     *   feed: "a:0"
+     *   feed: "b:0"
+     *   fetch: "x:0"
+     *   fetch: "y:0"
+     *   feed_devices: {
+     *     "a:0": "/job:localhost/replica:0/task:0/device:GPU:0"
+     *   }
+     *   fetch_devices: {
+     *     "y:0": "/job:localhost/replica:0/task:0/device:GPU:0"
+     *  }
+     * }
+     * means that the Callable expects:
+     * - The first argument ("a:0") is a Tensor backed by GPU memory.
+     * - The second argument ("b:0") is a Tensor backed by host memory.
+     * and of its return values:
+     * - The first output ("x:0") will be backed by host memory.
+     * - The second output ("y:0") will be backed by GPU memory.
+     * FEEDS:
+     * It is the responsibility of the caller to ensure that the memory of the fed
+     * tensors will be correctly initialized and synchronized before it is
+     * accessed by operations executed during the call to Session::RunCallable().
+     * This is typically ensured by using the TensorFlow memory allocators
+     * (Device::GetAllocator()) to create the Tensor to be fed.
+     * Alternatively, for CUDA-enabled GPU devices, this typically means that the
+     * operation that produced the contents of the tensor has completed, i.e., the
+     * CUDA stream has been synchronized (e.g., via cuCtxSynchronize() or
+     * cuStreamSynchronize()).
+     * </pre>
+     *
+     * <code>map&lt;string, string&gt; feed_devices = 6;</code>
+     */
+    public Builder putFeedDevices(
+        java.lang.String key,
+        java.lang.String value) {
+      if (key == null) { throw new java.lang.NullPointerException(); }
+      if (value == null) { throw new java.lang.NullPointerException(); }
+      internalGetMutableFeedDevices().getMutableMap()
+          .put(key, value);
+      return this;
+    }
+    /**
+     * <pre>
+     * The Tensor objects fed in the callable and fetched from the callable
+     * are expected to be backed by host (CPU) memory by default.
+     * The options below allow changing that - feeding tensors backed by
+     * device memory, or returning tensors that are backed by device memory.
+     * The maps below map the name of a feed/fetch tensor (which appears in
+     * 'feed' or 'fetch' fields above), to the fully qualified name of the device
+     * owning the memory backing the contents of the tensor.
+     * For example, creating a callable with the following options:
+     * CallableOptions {
+     *   feed: "a:0"
+     *   feed: "b:0"
+     *   fetch: "x:0"
+     *   fetch: "y:0"
+     *   feed_devices: {
+     *     "a:0": "/job:localhost/replica:0/task:0/device:GPU:0"
+     *   }
+     *   fetch_devices: {
+     *     "y:0": "/job:localhost/replica:0/task:0/device:GPU:0"
+     *  }
+     * }
+     * means that the Callable expects:
+     * - The first argument ("a:0") is a Tensor backed by GPU memory.
+     * - The second argument ("b:0") is a Tensor backed by host memory.
+     * and of its return values:
+     * - The first output ("x:0") will be backed by host memory.
+     * - The second output ("y:0") will be backed by GPU memory.
+     * FEEDS:
+     * It is the responsibility of the caller to ensure that the memory of the fed
+     * tensors will be correctly initialized and synchronized before it is
+     * accessed by operations executed during the call to Session::RunCallable().
+     * This is typically ensured by using the TensorFlow memory allocators
+     * (Device::GetAllocator()) to create the Tensor to be fed.
+     * Alternatively, for CUDA-enabled GPU devices, this typically means that the
+     * operation that produced the contents of the tensor has completed, i.e., the
+     * CUDA stream has been synchronized (e.g., via cuCtxSynchronize() or
+     * cuStreamSynchronize()).
+     * </pre>
+     *
+     * <code>map&lt;string, string&gt; feed_devices = 6;</code>
+     */
+
+    public Builder putAllFeedDevices(
+        java.util.Map<java.lang.String, java.lang.String> values) {
+      internalGetMutableFeedDevices().getMutableMap()
+          .putAll(values);
+      return this;
+    }
+
+    private com.google.protobuf.MapField<
+        java.lang.String, java.lang.String> fetchDevices_;
+    private com.google.protobuf.MapField<java.lang.String, java.lang.String>
+    internalGetFetchDevices() {
+      if (fetchDevices_ == null) {
+        return com.google.protobuf.MapField.emptyMapField(
+            FetchDevicesDefaultEntryHolder.defaultEntry);
+      }
+      return fetchDevices_;
+    }
+    private com.google.protobuf.MapField<java.lang.String, java.lang.String>
+    internalGetMutableFetchDevices() {
+      onChanged();;
+      if (fetchDevices_ == null) {
+        fetchDevices_ = com.google.protobuf.MapField.newMapField(
+            FetchDevicesDefaultEntryHolder.defaultEntry);
+      }
+      if (!fetchDevices_.isMutable()) {
+        fetchDevices_ = fetchDevices_.copy();
+      }
+      return fetchDevices_;
+    }
+
+    public int getFetchDevicesCount() {
+      return internalGetFetchDevices().getMap().size();
+    }
+    /**
+     * <code>map&lt;string, string&gt; fetch_devices = 7;</code>
+     */
+
+    public boolean containsFetchDevices(
+        java.lang.String key) {
+      if (key == null) { throw new java.lang.NullPointerException(); }
+      return internalGetFetchDevices().getMap().containsKey(key);
+    }
+    /**
+     * Use {@link #getFetchDevicesMap()} instead.
+     */
+    @java.lang.Deprecated
+    public java.util.Map<java.lang.String, java.lang.String> getFetchDevices() {
+      return getFetchDevicesMap();
+    }
+    /**
+     * <code>map&lt;string, string&gt; fetch_devices = 7;</code>
+     */
+
+    public java.util.Map<java.lang.String, java.lang.String> getFetchDevicesMap() {
+      return internalGetFetchDevices().getMap();
+    }
+    /**
+     * <code>map&lt;string, string&gt; fetch_devices = 7;</code>
+     */
+
+    public java.lang.String getFetchDevicesOrDefault(
+        java.lang.String key,
+        java.lang.String defaultValue) {
+      if (key == null) { throw new java.lang.NullPointerException(); }
+      java.util.Map<java.lang.String, java.lang.String> map =
+          internalGetFetchDevices().getMap();
+      return map.containsKey(key) ? map.get(key) : defaultValue;
+    }
+    /**
+     * <code>map&lt;string, string&gt; fetch_devices = 7;</code>
+     */
+
+    public java.lang.String getFetchDevicesOrThrow(
+        java.lang.String key) {
+      if (key == null) { throw new java.lang.NullPointerException(); }
+      java.util.Map<java.lang.String, java.lang.String> map =
+          internalGetFetchDevices().getMap();
+      if (!map.containsKey(key)) {
+        throw new java.lang.IllegalArgumentException();
+      }
+      return map.get(key);
+    }
+
+    public Builder clearFetchDevices() {
+      internalGetMutableFetchDevices().getMutableMap()
+          .clear();
+      return this;
+    }
+    /**
+     * <code>map&lt;string, string&gt; fetch_devices = 7;</code>
+     */
+
+    public Builder removeFetchDevices(
+        java.lang.String key) {
+      if (key == null) { throw new java.lang.NullPointerException(); }
+      internalGetMutableFetchDevices().getMutableMap()
+          .remove(key);
+      return this;
+    }
+    /**
+     * Use alternate mutation accessors instead.
+     */
+    @java.lang.Deprecated
+    public java.util.Map<java.lang.String, java.lang.String>
+    getMutableFetchDevices() {
+      return internalGetMutableFetchDevices().getMutableMap();
+    }
+    /**
+     * <code>map&lt;string, string&gt; fetch_devices = 7;</code>
+     */
+    public Builder putFetchDevices(
+        java.lang.String key,
+        java.lang.String value) {
+      if (key == null) { throw new java.lang.NullPointerException(); }
+      if (value == null) { throw new java.lang.NullPointerException(); }
+      internalGetMutableFetchDevices().getMutableMap()
+          .put(key, value);
+      return this;
+    }
+    /**
+     * <code>map&lt;string, string&gt; fetch_devices = 7;</code>
+     */
+
+    public Builder putAllFetchDevices(
+        java.util.Map<java.lang.String, java.lang.String> values) {
+      internalGetMutableFetchDevices().getMutableMap()
+          .putAll(values);
+      return this;
+    }
+
+    private boolean fetchSkipSync_ ;
+    /**
+     * <pre>
+     * By default, RunCallable() will synchronize the GPU stream before returning
+     * fetched tensors on a GPU device, to ensure that the values in those tensors
+     * have been produced. This simplifies interacting with the tensors, but
+     * potentially incurs a performance hit.
+     * If this options is set to true, the caller is responsible for ensuring
+     * that the values in the fetched tensors have been produced before they are
+     * used. The caller can do this by invoking `Device::Sync()` on the underlying
+     * device(s), or by feeding the tensors back to the same Session using
+     * `feed_devices` with the same corresponding device name.
+     * </pre>
+     *
+     * <code>bool fetch_skip_sync = 8;</code>
+     */
+    public boolean getFetchSkipSync() {
+      return fetchSkipSync_;
+    }
+    /**
+     * <pre>
+     * By default, RunCallable() will synchronize the GPU stream before returning
+     * fetched tensors on a GPU device, to ensure that the values in those tensors
+     * have been produced. This simplifies interacting with the tensors, but
+     * potentially incurs a performance hit.
+     * If this options is set to true, the caller is responsible for ensuring
+     * that the values in the fetched tensors have been produced before they are
+     * used. The caller can do this by invoking `Device::Sync()` on the underlying
+     * device(s), or by feeding the tensors back to the same Session using
+     * `feed_devices` with the same corresponding device name.
+     * </pre>
+     *
+     * <code>bool fetch_skip_sync = 8;</code>
+     */
+    public Builder setFetchSkipSync(boolean value) {
+      
+      fetchSkipSync_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * By default, RunCallable() will synchronize the GPU stream before returning
+     * fetched tensors on a GPU device, to ensure that the values in those tensors
+     * have been produced. This simplifies interacting with the tensors, but
+     * potentially incurs a performance hit.
+     * If this options is set to true, the caller is responsible for ensuring
+     * that the values in the fetched tensors have been produced before they are
+     * used. The caller can do this by invoking `Device::Sync()` on the underlying
+     * device(s), or by feeding the tensors back to the same Session using
+     * `feed_devices` with the same corresponding device name.
+     * </pre>
+     *
+     * <code>bool fetch_skip_sync = 8;</code>
+     */
+    public Builder clearFetchSkipSync() {
+      
+      fetchSkipSync_ = false;
+      onChanged();
+      return this;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.CallableOptions)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.CallableOptions)
+  private static final org.tensorflow.proto.framework.CallableOptions DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.CallableOptions();
+  }
+
+  public static org.tensorflow.proto.framework.CallableOptions getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<CallableOptions>
+      PARSER = new com.google.protobuf.AbstractParser<CallableOptions>() {
+    @java.lang.Override
+    public CallableOptions parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new CallableOptions(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<CallableOptions> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<CallableOptions> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.framework.CallableOptions getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/CallableOptionsOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/CallableOptionsOrBuilder.java
new file mode 100644
index 00000000000..d65852d4843
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/CallableOptionsOrBuilder.java
@@ -0,0 +1,485 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/config.proto
+
+package org.tensorflow.proto.framework;
+
+public interface CallableOptionsOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.CallableOptions)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <pre>
+   * Tensors to be fed in the callable. Each feed is the name of a tensor.
+   * </pre>
+   *
+   * <code>repeated string feed = 1;</code>
+   */
+  java.util.List<java.lang.String>
+      getFeedList();
+  /**
+   * <pre>
+   * Tensors to be fed in the callable. Each feed is the name of a tensor.
+   * </pre>
+   *
+   * <code>repeated string feed = 1;</code>
+   */
+  int getFeedCount();
+  /**
+   * <pre>
+   * Tensors to be fed in the callable. Each feed is the name of a tensor.
+   * </pre>
+   *
+   * <code>repeated string feed = 1;</code>
+   */
+  java.lang.String getFeed(int index);
+  /**
+   * <pre>
+   * Tensors to be fed in the callable. Each feed is the name of a tensor.
+   * </pre>
+   *
+   * <code>repeated string feed = 1;</code>
+   */
+  com.google.protobuf.ByteString
+      getFeedBytes(int index);
+
+  /**
+   * <pre>
+   * Fetches. A list of tensor names. The caller of the callable expects a
+   * tensor to be returned for each fetch[i] (see RunStepResponse.tensor). The
+   * order of specified fetches does not change the execution order.
+   * </pre>
+   *
+   * <code>repeated string fetch = 2;</code>
+   */
+  java.util.List<java.lang.String>
+      getFetchList();
+  /**
+   * <pre>
+   * Fetches. A list of tensor names. The caller of the callable expects a
+   * tensor to be returned for each fetch[i] (see RunStepResponse.tensor). The
+   * order of specified fetches does not change the execution order.
+   * </pre>
+   *
+   * <code>repeated string fetch = 2;</code>
+   */
+  int getFetchCount();
+  /**
+   * <pre>
+   * Fetches. A list of tensor names. The caller of the callable expects a
+   * tensor to be returned for each fetch[i] (see RunStepResponse.tensor). The
+   * order of specified fetches does not change the execution order.
+   * </pre>
+   *
+   * <code>repeated string fetch = 2;</code>
+   */
+  java.lang.String getFetch(int index);
+  /**
+   * <pre>
+   * Fetches. A list of tensor names. The caller of the callable expects a
+   * tensor to be returned for each fetch[i] (see RunStepResponse.tensor). The
+   * order of specified fetches does not change the execution order.
+   * </pre>
+   *
+   * <code>repeated string fetch = 2;</code>
+   */
+  com.google.protobuf.ByteString
+      getFetchBytes(int index);
+
+  /**
+   * <pre>
+   * Target Nodes. A list of node names. The named nodes will be run by the
+   * callable but their outputs will not be returned.
+   * </pre>
+   *
+   * <code>repeated string target = 3;</code>
+   */
+  java.util.List<java.lang.String>
+      getTargetList();
+  /**
+   * <pre>
+   * Target Nodes. A list of node names. The named nodes will be run by the
+   * callable but their outputs will not be returned.
+   * </pre>
+   *
+   * <code>repeated string target = 3;</code>
+   */
+  int getTargetCount();
+  /**
+   * <pre>
+   * Target Nodes. A list of node names. The named nodes will be run by the
+   * callable but their outputs will not be returned.
+   * </pre>
+   *
+   * <code>repeated string target = 3;</code>
+   */
+  java.lang.String getTarget(int index);
+  /**
+   * <pre>
+   * Target Nodes. A list of node names. The named nodes will be run by the
+   * callable but their outputs will not be returned.
+   * </pre>
+   *
+   * <code>repeated string target = 3;</code>
+   */
+  com.google.protobuf.ByteString
+      getTargetBytes(int index);
+
+  /**
+   * <pre>
+   * Options that will be applied to each run.
+   * </pre>
+   *
+   * <code>.tensorflow.RunOptions run_options = 4;</code>
+   */
+  boolean hasRunOptions();
+  /**
+   * <pre>
+   * Options that will be applied to each run.
+   * </pre>
+   *
+   * <code>.tensorflow.RunOptions run_options = 4;</code>
+   */
+  org.tensorflow.proto.framework.RunOptions getRunOptions();
+  /**
+   * <pre>
+   * Options that will be applied to each run.
+   * </pre>
+   *
+   * <code>.tensorflow.RunOptions run_options = 4;</code>
+   */
+  org.tensorflow.proto.framework.RunOptionsOrBuilder getRunOptionsOrBuilder();
+
+  /**
+   * <pre>
+   * Tensors to be connected in the callable. Each TensorConnection denotes
+   * a pair of tensors in the graph, between which an edge will be created
+   * in the callable.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.TensorConnection tensor_connection = 5;</code>
+   */
+  java.util.List<org.tensorflow.proto.framework.TensorConnection> 
+      getTensorConnectionList();
+  /**
+   * <pre>
+   * Tensors to be connected in the callable. Each TensorConnection denotes
+   * a pair of tensors in the graph, between which an edge will be created
+   * in the callable.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.TensorConnection tensor_connection = 5;</code>
+   */
+  org.tensorflow.proto.framework.TensorConnection getTensorConnection(int index);
+  /**
+   * <pre>
+   * Tensors to be connected in the callable. Each TensorConnection denotes
+   * a pair of tensors in the graph, between which an edge will be created
+   * in the callable.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.TensorConnection tensor_connection = 5;</code>
+   */
+  int getTensorConnectionCount();
+  /**
+   * <pre>
+   * Tensors to be connected in the callable. Each TensorConnection denotes
+   * a pair of tensors in the graph, between which an edge will be created
+   * in the callable.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.TensorConnection tensor_connection = 5;</code>
+   */
+  java.util.List<? extends org.tensorflow.proto.framework.TensorConnectionOrBuilder> 
+      getTensorConnectionOrBuilderList();
+  /**
+   * <pre>
+   * Tensors to be connected in the callable. Each TensorConnection denotes
+   * a pair of tensors in the graph, between which an edge will be created
+   * in the callable.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.TensorConnection tensor_connection = 5;</code>
+   */
+  org.tensorflow.proto.framework.TensorConnectionOrBuilder getTensorConnectionOrBuilder(
+      int index);
+
+  /**
+   * <pre>
+   * The Tensor objects fed in the callable and fetched from the callable
+   * are expected to be backed by host (CPU) memory by default.
+   * The options below allow changing that - feeding tensors backed by
+   * device memory, or returning tensors that are backed by device memory.
+   * The maps below map the name of a feed/fetch tensor (which appears in
+   * 'feed' or 'fetch' fields above), to the fully qualified name of the device
+   * owning the memory backing the contents of the tensor.
+   * For example, creating a callable with the following options:
+   * CallableOptions {
+   *   feed: "a:0"
+   *   feed: "b:0"
+   *   fetch: "x:0"
+   *   fetch: "y:0"
+   *   feed_devices: {
+   *     "a:0": "/job:localhost/replica:0/task:0/device:GPU:0"
+   *   }
+   *   fetch_devices: {
+   *     "y:0": "/job:localhost/replica:0/task:0/device:GPU:0"
+   *  }
+   * }
+   * means that the Callable expects:
+   * - The first argument ("a:0") is a Tensor backed by GPU memory.
+   * - The second argument ("b:0") is a Tensor backed by host memory.
+   * and of its return values:
+   * - The first output ("x:0") will be backed by host memory.
+   * - The second output ("y:0") will be backed by GPU memory.
+   * FEEDS:
+   * It is the responsibility of the caller to ensure that the memory of the fed
+   * tensors will be correctly initialized and synchronized before it is
+   * accessed by operations executed during the call to Session::RunCallable().
+   * This is typically ensured by using the TensorFlow memory allocators
+   * (Device::GetAllocator()) to create the Tensor to be fed.
+   * Alternatively, for CUDA-enabled GPU devices, this typically means that the
+   * operation that produced the contents of the tensor has completed, i.e., the
+   * CUDA stream has been synchronized (e.g., via cuCtxSynchronize() or
+   * cuStreamSynchronize()).
+   * </pre>
+   *
+   * <code>map&lt;string, string&gt; feed_devices = 6;</code>
+   */
+  int getFeedDevicesCount();
+  /**
+   * <pre>
+   * The Tensor objects fed in the callable and fetched from the callable
+   * are expected to be backed by host (CPU) memory by default.
+   * The options below allow changing that - feeding tensors backed by
+   * device memory, or returning tensors that are backed by device memory.
+   * The maps below map the name of a feed/fetch tensor (which appears in
+   * 'feed' or 'fetch' fields above), to the fully qualified name of the device
+   * owning the memory backing the contents of the tensor.
+   * For example, creating a callable with the following options:
+   * CallableOptions {
+   *   feed: "a:0"
+   *   feed: "b:0"
+   *   fetch: "x:0"
+   *   fetch: "y:0"
+   *   feed_devices: {
+   *     "a:0": "/job:localhost/replica:0/task:0/device:GPU:0"
+   *   }
+   *   fetch_devices: {
+   *     "y:0": "/job:localhost/replica:0/task:0/device:GPU:0"
+   *  }
+   * }
+   * means that the Callable expects:
+   * - The first argument ("a:0") is a Tensor backed by GPU memory.
+   * - The second argument ("b:0") is a Tensor backed by host memory.
+   * and of its return values:
+   * - The first output ("x:0") will be backed by host memory.
+   * - The second output ("y:0") will be backed by GPU memory.
+   * FEEDS:
+   * It is the responsibility of the caller to ensure that the memory of the fed
+   * tensors will be correctly initialized and synchronized before it is
+   * accessed by operations executed during the call to Session::RunCallable().
+   * This is typically ensured by using the TensorFlow memory allocators
+   * (Device::GetAllocator()) to create the Tensor to be fed.
+   * Alternatively, for CUDA-enabled GPU devices, this typically means that the
+   * operation that produced the contents of the tensor has completed, i.e., the
+   * CUDA stream has been synchronized (e.g., via cuCtxSynchronize() or
+   * cuStreamSynchronize()).
+   * </pre>
+   *
+   * <code>map&lt;string, string&gt; feed_devices = 6;</code>
+   */
+  boolean containsFeedDevices(
+      java.lang.String key);
+  /**
+   * Use {@link #getFeedDevicesMap()} instead.
+   */
+  @java.lang.Deprecated
+  java.util.Map<java.lang.String, java.lang.String>
+  getFeedDevices();
+  /**
+   * <pre>
+   * The Tensor objects fed in the callable and fetched from the callable
+   * are expected to be backed by host (CPU) memory by default.
+   * The options below allow changing that - feeding tensors backed by
+   * device memory, or returning tensors that are backed by device memory.
+   * The maps below map the name of a feed/fetch tensor (which appears in
+   * 'feed' or 'fetch' fields above), to the fully qualified name of the device
+   * owning the memory backing the contents of the tensor.
+   * For example, creating a callable with the following options:
+   * CallableOptions {
+   *   feed: "a:0"
+   *   feed: "b:0"
+   *   fetch: "x:0"
+   *   fetch: "y:0"
+   *   feed_devices: {
+   *     "a:0": "/job:localhost/replica:0/task:0/device:GPU:0"
+   *   }
+   *   fetch_devices: {
+   *     "y:0": "/job:localhost/replica:0/task:0/device:GPU:0"
+   *  }
+   * }
+   * means that the Callable expects:
+   * - The first argument ("a:0") is a Tensor backed by GPU memory.
+   * - The second argument ("b:0") is a Tensor backed by host memory.
+   * and of its return values:
+   * - The first output ("x:0") will be backed by host memory.
+   * - The second output ("y:0") will be backed by GPU memory.
+   * FEEDS:
+   * It is the responsibility of the caller to ensure that the memory of the fed
+   * tensors will be correctly initialized and synchronized before it is
+   * accessed by operations executed during the call to Session::RunCallable().
+   * This is typically ensured by using the TensorFlow memory allocators
+   * (Device::GetAllocator()) to create the Tensor to be fed.
+   * Alternatively, for CUDA-enabled GPU devices, this typically means that the
+   * operation that produced the contents of the tensor has completed, i.e., the
+   * CUDA stream has been synchronized (e.g., via cuCtxSynchronize() or
+   * cuStreamSynchronize()).
+   * </pre>
+   *
+   * <code>map&lt;string, string&gt; feed_devices = 6;</code>
+   */
+  java.util.Map<java.lang.String, java.lang.String>
+  getFeedDevicesMap();
+  /**
+   * <pre>
+   * The Tensor objects fed in the callable and fetched from the callable
+   * are expected to be backed by host (CPU) memory by default.
+   * The options below allow changing that - feeding tensors backed by
+   * device memory, or returning tensors that are backed by device memory.
+   * The maps below map the name of a feed/fetch tensor (which appears in
+   * 'feed' or 'fetch' fields above), to the fully qualified name of the device
+   * owning the memory backing the contents of the tensor.
+   * For example, creating a callable with the following options:
+   * CallableOptions {
+   *   feed: "a:0"
+   *   feed: "b:0"
+   *   fetch: "x:0"
+   *   fetch: "y:0"
+   *   feed_devices: {
+   *     "a:0": "/job:localhost/replica:0/task:0/device:GPU:0"
+   *   }
+   *   fetch_devices: {
+   *     "y:0": "/job:localhost/replica:0/task:0/device:GPU:0"
+   *  }
+   * }
+   * means that the Callable expects:
+   * - The first argument ("a:0") is a Tensor backed by GPU memory.
+   * - The second argument ("b:0") is a Tensor backed by host memory.
+   * and of its return values:
+   * - The first output ("x:0") will be backed by host memory.
+   * - The second output ("y:0") will be backed by GPU memory.
+   * FEEDS:
+   * It is the responsibility of the caller to ensure that the memory of the fed
+   * tensors will be correctly initialized and synchronized before it is
+   * accessed by operations executed during the call to Session::RunCallable().
+   * This is typically ensured by using the TensorFlow memory allocators
+   * (Device::GetAllocator()) to create the Tensor to be fed.
+   * Alternatively, for CUDA-enabled GPU devices, this typically means that the
+   * operation that produced the contents of the tensor has completed, i.e., the
+   * CUDA stream has been synchronized (e.g., via cuCtxSynchronize() or
+   * cuStreamSynchronize()).
+   * </pre>
+   *
+   * <code>map&lt;string, string&gt; feed_devices = 6;</code>
+   */
+
+  java.lang.String getFeedDevicesOrDefault(
+      java.lang.String key,
+      java.lang.String defaultValue);
+  /**
+   * <pre>
+   * The Tensor objects fed in the callable and fetched from the callable
+   * are expected to be backed by host (CPU) memory by default.
+   * The options below allow changing that - feeding tensors backed by
+   * device memory, or returning tensors that are backed by device memory.
+   * The maps below map the name of a feed/fetch tensor (which appears in
+   * 'feed' or 'fetch' fields above), to the fully qualified name of the device
+   * owning the memory backing the contents of the tensor.
+   * For example, creating a callable with the following options:
+   * CallableOptions {
+   *   feed: "a:0"
+   *   feed: "b:0"
+   *   fetch: "x:0"
+   *   fetch: "y:0"
+   *   feed_devices: {
+   *     "a:0": "/job:localhost/replica:0/task:0/device:GPU:0"
+   *   }
+   *   fetch_devices: {
+   *     "y:0": "/job:localhost/replica:0/task:0/device:GPU:0"
+   *  }
+   * }
+   * means that the Callable expects:
+   * - The first argument ("a:0") is a Tensor backed by GPU memory.
+   * - The second argument ("b:0") is a Tensor backed by host memory.
+   * and of its return values:
+   * - The first output ("x:0") will be backed by host memory.
+   * - The second output ("y:0") will be backed by GPU memory.
+   * FEEDS:
+   * It is the responsibility of the caller to ensure that the memory of the fed
+   * tensors will be correctly initialized and synchronized before it is
+   * accessed by operations executed during the call to Session::RunCallable().
+   * This is typically ensured by using the TensorFlow memory allocators
+   * (Device::GetAllocator()) to create the Tensor to be fed.
+   * Alternatively, for CUDA-enabled GPU devices, this typically means that the
+   * operation that produced the contents of the tensor has completed, i.e., the
+   * CUDA stream has been synchronized (e.g., via cuCtxSynchronize() or
+   * cuStreamSynchronize()).
+   * </pre>
+   *
+   * <code>map&lt;string, string&gt; feed_devices = 6;</code>
+   */
+
+  java.lang.String getFeedDevicesOrThrow(
+      java.lang.String key);
+
+  /**
+   * <code>map&lt;string, string&gt; fetch_devices = 7;</code>
+   */
+  int getFetchDevicesCount();
+  /**
+   * <code>map&lt;string, string&gt; fetch_devices = 7;</code>
+   */
+  boolean containsFetchDevices(
+      java.lang.String key);
+  /**
+   * Use {@link #getFetchDevicesMap()} instead.
+   */
+  @java.lang.Deprecated
+  java.util.Map<java.lang.String, java.lang.String>
+  getFetchDevices();
+  /**
+   * <code>map&lt;string, string&gt; fetch_devices = 7;</code>
+   */
+  java.util.Map<java.lang.String, java.lang.String>
+  getFetchDevicesMap();
+  /**
+   * <code>map&lt;string, string&gt; fetch_devices = 7;</code>
+   */
+
+  java.lang.String getFetchDevicesOrDefault(
+      java.lang.String key,
+      java.lang.String defaultValue);
+  /**
+   * <code>map&lt;string, string&gt; fetch_devices = 7;</code>
+   */
+
+  java.lang.String getFetchDevicesOrThrow(
+      java.lang.String key);
+
+  /**
+   * <pre>
+   * By default, RunCallable() will synchronize the GPU stream before returning
+   * fetched tensors on a GPU device, to ensure that the values in those tensors
+   * have been produced. This simplifies interacting with the tensors, but
+   * potentially incurs a performance hit.
+   * If this options is set to true, the caller is responsible for ensuring
+   * that the values in the fetched tensors have been produced before they are
+   * used. The caller can do this by invoking `Device::Sync()` on the underlying
+   * device(s), or by feeding the tensors back to the same Session using
+   * `feed_devices` with the same corresponding device name.
+   * </pre>
+   *
+   * <code>bool fetch_skip_sync = 8;</code>
+   */
+  boolean getFetchSkipSync();
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/Code.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/Code.java
new file mode 100644
index 00000000000..c7eda03f92a
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/Code.java
@@ -0,0 +1,536 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/error_codes.proto
+
+package org.tensorflow.proto.framework;
+
+/**
+ * <pre>
+ * The canonical error codes for TensorFlow APIs.
+ * Warnings:
+ * -   Do not change any numeric assignments.
+ * -   Changes to this list should only be made if there is a compelling
+ *     need that can't be satisfied in another way.  Such changes
+ *     must be approved by at least two OWNERS.
+ * -   These error codes must match gRPC and protobuf error codes (except for
+ *     DO_NOT_USE_RESERVED_FOR_FUTURE_EXPANSION_USE_DEFAULT_IN_SWITCH_INSTEAD_).
+ * Sometimes multiple error codes may apply.  Services should return
+ * the most specific error code that applies.  For example, prefer
+ * OUT_OF_RANGE over FAILED_PRECONDITION if both codes apply.
+ * Similarly prefer NOT_FOUND or ALREADY_EXISTS over FAILED_PRECONDITION.
+ * </pre>
+ *
+ * Protobuf enum {@code tensorflow.error.Code}
+ */
+public enum Code
+    implements com.google.protobuf.ProtocolMessageEnum {
+  /**
+   * <pre>
+   * Not an error; returned on success
+   * </pre>
+   *
+   * <code>OK = 0;</code>
+   */
+  OK(0),
+  /**
+   * <pre>
+   * The operation was cancelled (typically by the caller).
+   * </pre>
+   *
+   * <code>CANCELLED = 1;</code>
+   */
+  CANCELLED(1),
+  /**
+   * <pre>
+   * Unknown error.  An example of where this error may be returned is
+   * if a Status value received from another address space belongs to
+   * an error-space that is not known in this address space.  Also
+   * errors raised by APIs that do not return enough error information
+   * may be converted to this error.
+   * </pre>
+   *
+   * <code>UNKNOWN = 2;</code>
+   */
+  UNKNOWN(2),
+  /**
+   * <pre>
+   * Client specified an invalid argument.  Note that this differs
+   * from FAILED_PRECONDITION.  INVALID_ARGUMENT indicates arguments
+   * that are problematic regardless of the state of the system
+   * (e.g., a malformed file name).
+   * </pre>
+   *
+   * <code>INVALID_ARGUMENT = 3;</code>
+   */
+  INVALID_ARGUMENT(3),
+  /**
+   * <pre>
+   * Deadline expired before operation could complete.  For operations
+   * that change the state of the system, this error may be returned
+   * even if the operation has completed successfully.  For example, a
+   * successful response from a server could have been delayed long
+   * enough for the deadline to expire.
+   * </pre>
+   *
+   * <code>DEADLINE_EXCEEDED = 4;</code>
+   */
+  DEADLINE_EXCEEDED(4),
+  /**
+   * <pre>
+   * Some requested entity (e.g., file or directory) was not found.
+   * For privacy reasons, this code *may* be returned when the client
+   * does not have the access right to the entity.
+   * </pre>
+   *
+   * <code>NOT_FOUND = 5;</code>
+   */
+  NOT_FOUND(5),
+  /**
+   * <pre>
+   * Some entity that we attempted to create (e.g., file or directory)
+   * already exists.
+   * </pre>
+   *
+   * <code>ALREADY_EXISTS = 6;</code>
+   */
+  ALREADY_EXISTS(6),
+  /**
+   * <pre>
+   * The caller does not have permission to execute the specified
+   * operation.  PERMISSION_DENIED must not be used for rejections
+   * caused by exhausting some resource (use RESOURCE_EXHAUSTED
+   * instead for those errors).  PERMISSION_DENIED must not be
+   * used if the caller can not be identified (use UNAUTHENTICATED
+   * instead for those errors).
+   * </pre>
+   *
+   * <code>PERMISSION_DENIED = 7;</code>
+   */
+  PERMISSION_DENIED(7),
+  /**
+   * <pre>
+   * The request does not have valid authentication credentials for the
+   * operation.
+   * </pre>
+   *
+   * <code>UNAUTHENTICATED = 16;</code>
+   */
+  UNAUTHENTICATED(16),
+  /**
+   * <pre>
+   * Some resource has been exhausted, perhaps a per-user quota, or
+   * perhaps the entire file system is out of space.
+   * </pre>
+   *
+   * <code>RESOURCE_EXHAUSTED = 8;</code>
+   */
+  RESOURCE_EXHAUSTED(8),
+  /**
+   * <pre>
+   * Operation was rejected because the system is not in a state
+   * required for the operation's execution.  For example, directory
+   * to be deleted may be non-empty, an rmdir operation is applied to
+   * a non-directory, etc.
+   * A litmus test that may help a service implementor in deciding
+   * between FAILED_PRECONDITION, ABORTED, and UNAVAILABLE:
+   *  (a) Use UNAVAILABLE if the client can retry just the failing call.
+   *  (b) Use ABORTED if the client should retry at a higher-level
+   *      (e.g., restarting a read-modify-write sequence).
+   *  (c) Use FAILED_PRECONDITION if the client should not retry until
+   *      the system state has been explicitly fixed.  E.g., if an "rmdir"
+   *      fails because the directory is non-empty, FAILED_PRECONDITION
+   *      should be returned since the client should not retry unless
+   *      they have first fixed up the directory by deleting files from it.
+   *  (d) Use FAILED_PRECONDITION if the client performs conditional
+   *      REST Get/Update/Delete on a resource and the resource on the
+   *      server does not match the condition. E.g., conflicting
+   *      read-modify-write on the same resource.
+   * </pre>
+   *
+   * <code>FAILED_PRECONDITION = 9;</code>
+   */
+  FAILED_PRECONDITION(9),
+  /**
+   * <pre>
+   * The operation was aborted, typically due to a concurrency issue
+   * like sequencer check failures, transaction aborts, etc.
+   * See litmus test above for deciding between FAILED_PRECONDITION,
+   * ABORTED, and UNAVAILABLE.
+   * </pre>
+   *
+   * <code>ABORTED = 10;</code>
+   */
+  ABORTED(10),
+  /**
+   * <pre>
+   * Operation tried to iterate past the valid input range.  E.g., seeking or
+   * reading past end of file.
+   * Unlike INVALID_ARGUMENT, this error indicates a problem that may
+   * be fixed if the system state changes. For example, a 32-bit file
+   * system will generate INVALID_ARGUMENT if asked to read at an
+   * offset that is not in the range [0,2^32-1], but it will generate
+   * OUT_OF_RANGE if asked to read from an offset past the current
+   * file size.
+   * There is a fair bit of overlap between FAILED_PRECONDITION and
+   * OUT_OF_RANGE.  We recommend using OUT_OF_RANGE (the more specific
+   * error) when it applies so that callers who are iterating through
+   * a space can easily look for an OUT_OF_RANGE error to detect when
+   * they are done.
+   * </pre>
+   *
+   * <code>OUT_OF_RANGE = 11;</code>
+   */
+  OUT_OF_RANGE(11),
+  /**
+   * <pre>
+   * Operation is not implemented or not supported/enabled in this service.
+   * </pre>
+   *
+   * <code>UNIMPLEMENTED = 12;</code>
+   */
+  UNIMPLEMENTED(12),
+  /**
+   * <pre>
+   * Internal errors.  Means some invariant expected by the underlying
+   * system has been broken.  If you see one of these errors,
+   * something is very broken.
+   * </pre>
+   *
+   * <code>INTERNAL = 13;</code>
+   */
+  INTERNAL(13),
+  /**
+   * <pre>
+   * The service is currently unavailable.  This is a most likely a
+   * transient condition and may be corrected by retrying with
+   * a backoff.
+   * See litmus test above for deciding between FAILED_PRECONDITION,
+   * ABORTED, and UNAVAILABLE.
+   * </pre>
+   *
+   * <code>UNAVAILABLE = 14;</code>
+   */
+  UNAVAILABLE(14),
+  /**
+   * <pre>
+   * Unrecoverable data loss or corruption.
+   * </pre>
+   *
+   * <code>DATA_LOSS = 15;</code>
+   */
+  DATA_LOSS(15),
+  /**
+   * <pre>
+   * An extra enum entry to prevent people from writing code that
+   * fails to compile when a new code is added.
+   * Nobody should ever reference this enumeration entry. In particular,
+   * if you write C++ code that switches on this enumeration, add a default:
+   * case instead of a case that mentions this enumeration entry.
+   * Nobody should rely on the value (currently 20) listed here.  It
+   * may change in the future.
+   * </pre>
+   *
+   * <code>DO_NOT_USE_RESERVED_FOR_FUTURE_EXPANSION_USE_DEFAULT_IN_SWITCH_INSTEAD_ = 20;</code>
+   */
+  DO_NOT_USE_RESERVED_FOR_FUTURE_EXPANSION_USE_DEFAULT_IN_SWITCH_INSTEAD_(20),
+  UNRECOGNIZED(-1),
+  ;
+
+  /**
+   * <pre>
+   * Not an error; returned on success
+   * </pre>
+   *
+   * <code>OK = 0;</code>
+   */
+  public static final int OK_VALUE = 0;
+  /**
+   * <pre>
+   * The operation was cancelled (typically by the caller).
+   * </pre>
+   *
+   * <code>CANCELLED = 1;</code>
+   */
+  public static final int CANCELLED_VALUE = 1;
+  /**
+   * <pre>
+   * Unknown error.  An example of where this error may be returned is
+   * if a Status value received from another address space belongs to
+   * an error-space that is not known in this address space.  Also
+   * errors raised by APIs that do not return enough error information
+   * may be converted to this error.
+   * </pre>
+   *
+   * <code>UNKNOWN = 2;</code>
+   */
+  public static final int UNKNOWN_VALUE = 2;
+  /**
+   * <pre>
+   * Client specified an invalid argument.  Note that this differs
+   * from FAILED_PRECONDITION.  INVALID_ARGUMENT indicates arguments
+   * that are problematic regardless of the state of the system
+   * (e.g., a malformed file name).
+   * </pre>
+   *
+   * <code>INVALID_ARGUMENT = 3;</code>
+   */
+  public static final int INVALID_ARGUMENT_VALUE = 3;
+  /**
+   * <pre>
+   * Deadline expired before operation could complete.  For operations
+   * that change the state of the system, this error may be returned
+   * even if the operation has completed successfully.  For example, a
+   * successful response from a server could have been delayed long
+   * enough for the deadline to expire.
+   * </pre>
+   *
+   * <code>DEADLINE_EXCEEDED = 4;</code>
+   */
+  public static final int DEADLINE_EXCEEDED_VALUE = 4;
+  /**
+   * <pre>
+   * Some requested entity (e.g., file or directory) was not found.
+   * For privacy reasons, this code *may* be returned when the client
+   * does not have the access right to the entity.
+   * </pre>
+   *
+   * <code>NOT_FOUND = 5;</code>
+   */
+  public static final int NOT_FOUND_VALUE = 5;
+  /**
+   * <pre>
+   * Some entity that we attempted to create (e.g., file or directory)
+   * already exists.
+   * </pre>
+   *
+   * <code>ALREADY_EXISTS = 6;</code>
+   */
+  public static final int ALREADY_EXISTS_VALUE = 6;
+  /**
+   * <pre>
+   * The caller does not have permission to execute the specified
+   * operation.  PERMISSION_DENIED must not be used for rejections
+   * caused by exhausting some resource (use RESOURCE_EXHAUSTED
+   * instead for those errors).  PERMISSION_DENIED must not be
+   * used if the caller can not be identified (use UNAUTHENTICATED
+   * instead for those errors).
+   * </pre>
+   *
+   * <code>PERMISSION_DENIED = 7;</code>
+   */
+  public static final int PERMISSION_DENIED_VALUE = 7;
+  /**
+   * <pre>
+   * The request does not have valid authentication credentials for the
+   * operation.
+   * </pre>
+   *
+   * <code>UNAUTHENTICATED = 16;</code>
+   */
+  public static final int UNAUTHENTICATED_VALUE = 16;
+  /**
+   * <pre>
+   * Some resource has been exhausted, perhaps a per-user quota, or
+   * perhaps the entire file system is out of space.
+   * </pre>
+   *
+   * <code>RESOURCE_EXHAUSTED = 8;</code>
+   */
+  public static final int RESOURCE_EXHAUSTED_VALUE = 8;
+  /**
+   * <pre>
+   * Operation was rejected because the system is not in a state
+   * required for the operation's execution.  For example, directory
+   * to be deleted may be non-empty, an rmdir operation is applied to
+   * a non-directory, etc.
+   * A litmus test that may help a service implementor in deciding
+   * between FAILED_PRECONDITION, ABORTED, and UNAVAILABLE:
+   *  (a) Use UNAVAILABLE if the client can retry just the failing call.
+   *  (b) Use ABORTED if the client should retry at a higher-level
+   *      (e.g., restarting a read-modify-write sequence).
+   *  (c) Use FAILED_PRECONDITION if the client should not retry until
+   *      the system state has been explicitly fixed.  E.g., if an "rmdir"
+   *      fails because the directory is non-empty, FAILED_PRECONDITION
+   *      should be returned since the client should not retry unless
+   *      they have first fixed up the directory by deleting files from it.
+   *  (d) Use FAILED_PRECONDITION if the client performs conditional
+   *      REST Get/Update/Delete on a resource and the resource on the
+   *      server does not match the condition. E.g., conflicting
+   *      read-modify-write on the same resource.
+   * </pre>
+   *
+   * <code>FAILED_PRECONDITION = 9;</code>
+   */
+  public static final int FAILED_PRECONDITION_VALUE = 9;
+  /**
+   * <pre>
+   * The operation was aborted, typically due to a concurrency issue
+   * like sequencer check failures, transaction aborts, etc.
+   * See litmus test above for deciding between FAILED_PRECONDITION,
+   * ABORTED, and UNAVAILABLE.
+   * </pre>
+   *
+   * <code>ABORTED = 10;</code>
+   */
+  public static final int ABORTED_VALUE = 10;
+  /**
+   * <pre>
+   * Operation tried to iterate past the valid input range.  E.g., seeking or
+   * reading past end of file.
+   * Unlike INVALID_ARGUMENT, this error indicates a problem that may
+   * be fixed if the system state changes. For example, a 32-bit file
+   * system will generate INVALID_ARGUMENT if asked to read at an
+   * offset that is not in the range [0,2^32-1], but it will generate
+   * OUT_OF_RANGE if asked to read from an offset past the current
+   * file size.
+   * There is a fair bit of overlap between FAILED_PRECONDITION and
+   * OUT_OF_RANGE.  We recommend using OUT_OF_RANGE (the more specific
+   * error) when it applies so that callers who are iterating through
+   * a space can easily look for an OUT_OF_RANGE error to detect when
+   * they are done.
+   * </pre>
+   *
+   * <code>OUT_OF_RANGE = 11;</code>
+   */
+  public static final int OUT_OF_RANGE_VALUE = 11;
+  /**
+   * <pre>
+   * Operation is not implemented or not supported/enabled in this service.
+   * </pre>
+   *
+   * <code>UNIMPLEMENTED = 12;</code>
+   */
+  public static final int UNIMPLEMENTED_VALUE = 12;
+  /**
+   * <pre>
+   * Internal errors.  Means some invariant expected by the underlying
+   * system has been broken.  If you see one of these errors,
+   * something is very broken.
+   * </pre>
+   *
+   * <code>INTERNAL = 13;</code>
+   */
+  public static final int INTERNAL_VALUE = 13;
+  /**
+   * <pre>
+   * The service is currently unavailable.  This is a most likely a
+   * transient condition and may be corrected by retrying with
+   * a backoff.
+   * See litmus test above for deciding between FAILED_PRECONDITION,
+   * ABORTED, and UNAVAILABLE.
+   * </pre>
+   *
+   * <code>UNAVAILABLE = 14;</code>
+   */
+  public static final int UNAVAILABLE_VALUE = 14;
+  /**
+   * <pre>
+   * Unrecoverable data loss or corruption.
+   * </pre>
+   *
+   * <code>DATA_LOSS = 15;</code>
+   */
+  public static final int DATA_LOSS_VALUE = 15;
+  /**
+   * <pre>
+   * An extra enum entry to prevent people from writing code that
+   * fails to compile when a new code is added.
+   * Nobody should ever reference this enumeration entry. In particular,
+   * if you write C++ code that switches on this enumeration, add a default:
+   * case instead of a case that mentions this enumeration entry.
+   * Nobody should rely on the value (currently 20) listed here.  It
+   * may change in the future.
+   * </pre>
+   *
+   * <code>DO_NOT_USE_RESERVED_FOR_FUTURE_EXPANSION_USE_DEFAULT_IN_SWITCH_INSTEAD_ = 20;</code>
+   */
+  public static final int DO_NOT_USE_RESERVED_FOR_FUTURE_EXPANSION_USE_DEFAULT_IN_SWITCH_INSTEAD__VALUE = 20;
+
+
+  public final int getNumber() {
+    if (this == UNRECOGNIZED) {
+      throw new java.lang.IllegalArgumentException(
+          "Can't get the number of an unknown enum value.");
+    }
+    return value;
+  }
+
+  /**
+   * @deprecated Use {@link #forNumber(int)} instead.
+   */
+  @java.lang.Deprecated
+  public static Code valueOf(int value) {
+    return forNumber(value);
+  }
+
+  public static Code forNumber(int value) {
+    switch (value) {
+      case 0: return OK;
+      case 1: return CANCELLED;
+      case 2: return UNKNOWN;
+      case 3: return INVALID_ARGUMENT;
+      case 4: return DEADLINE_EXCEEDED;
+      case 5: return NOT_FOUND;
+      case 6: return ALREADY_EXISTS;
+      case 7: return PERMISSION_DENIED;
+      case 16: return UNAUTHENTICATED;
+      case 8: return RESOURCE_EXHAUSTED;
+      case 9: return FAILED_PRECONDITION;
+      case 10: return ABORTED;
+      case 11: return OUT_OF_RANGE;
+      case 12: return UNIMPLEMENTED;
+      case 13: return INTERNAL;
+      case 14: return UNAVAILABLE;
+      case 15: return DATA_LOSS;
+      case 20: return DO_NOT_USE_RESERVED_FOR_FUTURE_EXPANSION_USE_DEFAULT_IN_SWITCH_INSTEAD_;
+      default: return null;
+    }
+  }
+
+  public static com.google.protobuf.Internal.EnumLiteMap<Code>
+      internalGetValueMap() {
+    return internalValueMap;
+  }
+  private static final com.google.protobuf.Internal.EnumLiteMap<
+      Code> internalValueMap =
+        new com.google.protobuf.Internal.EnumLiteMap<Code>() {
+          public Code findValueByNumber(int number) {
+            return Code.forNumber(number);
+          }
+        };
+
+  public final com.google.protobuf.Descriptors.EnumValueDescriptor
+      getValueDescriptor() {
+    return getDescriptor().getValues().get(ordinal());
+  }
+  public final com.google.protobuf.Descriptors.EnumDescriptor
+      getDescriptorForType() {
+    return getDescriptor();
+  }
+  public static final com.google.protobuf.Descriptors.EnumDescriptor
+      getDescriptor() {
+    return org.tensorflow.proto.framework.ErrorCodesProtos.getDescriptor().getEnumTypes().get(0);
+  }
+
+  private static final Code[] VALUES = values();
+
+  public static Code valueOf(
+      com.google.protobuf.Descriptors.EnumValueDescriptor desc) {
+    if (desc.getType() != getDescriptor()) {
+      throw new java.lang.IllegalArgumentException(
+        "EnumValueDescriptor is not for this type.");
+    }
+    if (desc.getIndex() == -1) {
+      return UNRECOGNIZED;
+    }
+    return VALUES[desc.getIndex()];
+  }
+
+  private final int value;
+
+  private Code(int value) {
+    this.value = value;
+  }
+
+  // @@protoc_insertion_point(enum_scope:tensorflow.error.Code)
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/CollectionDef.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/CollectionDef.java
new file mode 100644
index 00000000000..f3094be10f6
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/CollectionDef.java
@@ -0,0 +1,4858 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/meta_graph.proto
+
+package org.tensorflow.proto.framework;
+
+/**
+ * <pre>
+ * CollectionDef should cover most collections.
+ * To add a user-defined collection, do one of the following:
+ * 1. For simple data types, such as string, int, float:
+ *      tf.add_to_collection("your_collection_name", your_simple_value)
+ *    strings will be stored as bytes_list.
+ * 2. For Protobuf types, there are three ways to add them:
+ *    1) tf.add_to_collection("your_collection_name",
+ *         your_proto.SerializeToString())
+ *       collection_def {
+ *         key: "user_defined_bytes_collection"
+ *         value {
+ *           bytes_list {
+ *             value: "queue_name: &#92;"test_queue&#92;"&#92;n"
+ *           }
+ *         }
+ *       }
+ *  or
+ *    2) tf.add_to_collection("your_collection_name", str(your_proto))
+ *       collection_def {
+ *         key: "user_defined_string_collection"
+ *         value {
+ *          bytes_list {
+ *             value: "&#92;n&#92;ntest_queue"
+ *           }
+ *         }
+ *       }
+ *  or
+ *    3) any_buf = any_pb2.Any()
+ *       tf.add_to_collection("your_collection_name",
+ *         any_buf.Pack(your_proto))
+ *       collection_def {
+ *         key: "user_defined_any_collection"
+ *         value {
+ *           any_list {
+ *             value {
+ *               type_url: "type.googleapis.com/tensorflow.QueueRunnerDef"
+ *               value: "&#92;n&#92;ntest_queue"
+ *             }
+ *           }
+ *         }
+ *       }
+ * 3. For Python objects, implement to_proto() and from_proto(), and register
+ *    them in the following manner:
+ *    ops.register_proto_function("your_collection_name",
+ *                                proto_type,
+ *                                to_proto=YourPythonObject.to_proto,
+ *                                from_proto=YourPythonObject.from_proto)
+ *    These functions will be invoked to serialize and de-serialize the
+ *    collection. For example,
+ *    ops.register_proto_function(ops.GraphKeys.GLOBAL_VARIABLES,
+ *                                proto_type=variable_pb2.VariableDef,
+ *                                to_proto=Variable.to_proto,
+ *                                from_proto=Variable.from_proto)
+ * </pre>
+ *
+ * Protobuf type {@code tensorflow.CollectionDef}
+ */
+public  final class CollectionDef extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.CollectionDef)
+    CollectionDefOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use CollectionDef.newBuilder() to construct.
+  private CollectionDef(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private CollectionDef() {
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new CollectionDef();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private CollectionDef(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 10: {
+            org.tensorflow.proto.framework.CollectionDef.NodeList.Builder subBuilder = null;
+            if (kindCase_ == 1) {
+              subBuilder = ((org.tensorflow.proto.framework.CollectionDef.NodeList) kind_).toBuilder();
+            }
+            kind_ =
+                input.readMessage(org.tensorflow.proto.framework.CollectionDef.NodeList.parser(), extensionRegistry);
+            if (subBuilder != null) {
+              subBuilder.mergeFrom((org.tensorflow.proto.framework.CollectionDef.NodeList) kind_);
+              kind_ = subBuilder.buildPartial();
+            }
+            kindCase_ = 1;
+            break;
+          }
+          case 18: {
+            org.tensorflow.proto.framework.CollectionDef.BytesList.Builder subBuilder = null;
+            if (kindCase_ == 2) {
+              subBuilder = ((org.tensorflow.proto.framework.CollectionDef.BytesList) kind_).toBuilder();
+            }
+            kind_ =
+                input.readMessage(org.tensorflow.proto.framework.CollectionDef.BytesList.parser(), extensionRegistry);
+            if (subBuilder != null) {
+              subBuilder.mergeFrom((org.tensorflow.proto.framework.CollectionDef.BytesList) kind_);
+              kind_ = subBuilder.buildPartial();
+            }
+            kindCase_ = 2;
+            break;
+          }
+          case 26: {
+            org.tensorflow.proto.framework.CollectionDef.Int64List.Builder subBuilder = null;
+            if (kindCase_ == 3) {
+              subBuilder = ((org.tensorflow.proto.framework.CollectionDef.Int64List) kind_).toBuilder();
+            }
+            kind_ =
+                input.readMessage(org.tensorflow.proto.framework.CollectionDef.Int64List.parser(), extensionRegistry);
+            if (subBuilder != null) {
+              subBuilder.mergeFrom((org.tensorflow.proto.framework.CollectionDef.Int64List) kind_);
+              kind_ = subBuilder.buildPartial();
+            }
+            kindCase_ = 3;
+            break;
+          }
+          case 34: {
+            org.tensorflow.proto.framework.CollectionDef.FloatList.Builder subBuilder = null;
+            if (kindCase_ == 4) {
+              subBuilder = ((org.tensorflow.proto.framework.CollectionDef.FloatList) kind_).toBuilder();
+            }
+            kind_ =
+                input.readMessage(org.tensorflow.proto.framework.CollectionDef.FloatList.parser(), extensionRegistry);
+            if (subBuilder != null) {
+              subBuilder.mergeFrom((org.tensorflow.proto.framework.CollectionDef.FloatList) kind_);
+              kind_ = subBuilder.buildPartial();
+            }
+            kindCase_ = 4;
+            break;
+          }
+          case 42: {
+            org.tensorflow.proto.framework.CollectionDef.AnyList.Builder subBuilder = null;
+            if (kindCase_ == 5) {
+              subBuilder = ((org.tensorflow.proto.framework.CollectionDef.AnyList) kind_).toBuilder();
+            }
+            kind_ =
+                input.readMessage(org.tensorflow.proto.framework.CollectionDef.AnyList.parser(), extensionRegistry);
+            if (subBuilder != null) {
+              subBuilder.mergeFrom((org.tensorflow.proto.framework.CollectionDef.AnyList) kind_);
+              kind_ = subBuilder.buildPartial();
+            }
+            kindCase_ = 5;
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.framework.MetaGraphProtos.internal_static_tensorflow_CollectionDef_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.framework.MetaGraphProtos.internal_static_tensorflow_CollectionDef_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.framework.CollectionDef.class, org.tensorflow.proto.framework.CollectionDef.Builder.class);
+  }
+
+  public interface NodeListOrBuilder extends
+      // @@protoc_insertion_point(interface_extends:tensorflow.CollectionDef.NodeList)
+      com.google.protobuf.MessageOrBuilder {
+
+    /**
+     * <code>repeated string value = 1;</code>
+     */
+    java.util.List<java.lang.String>
+        getValueList();
+    /**
+     * <code>repeated string value = 1;</code>
+     */
+    int getValueCount();
+    /**
+     * <code>repeated string value = 1;</code>
+     */
+    java.lang.String getValue(int index);
+    /**
+     * <code>repeated string value = 1;</code>
+     */
+    com.google.protobuf.ByteString
+        getValueBytes(int index);
+  }
+  /**
+   * <pre>
+   * NodeList is used for collecting nodes in graph. For example
+   * collection_def {
+   *   key: "summaries"
+   *   value {
+   *     node_list {
+   *       value: "input_producer/ScalarSummary:0"
+   *       value: "shuffle_batch/ScalarSummary:0"
+   *       value: "ImageSummary:0"
+   *     }
+   *   }
+   * </pre>
+   *
+   * Protobuf type {@code tensorflow.CollectionDef.NodeList}
+   */
+  public  static final class NodeList extends
+      com.google.protobuf.GeneratedMessageV3 implements
+      // @@protoc_insertion_point(message_implements:tensorflow.CollectionDef.NodeList)
+      NodeListOrBuilder {
+  private static final long serialVersionUID = 0L;
+    // Use NodeList.newBuilder() to construct.
+    private NodeList(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+      super(builder);
+    }
+    private NodeList() {
+      value_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+    }
+
+    @java.lang.Override
+    @SuppressWarnings({"unused"})
+    protected java.lang.Object newInstance(
+        UnusedPrivateParameter unused) {
+      return new NodeList();
+    }
+
+    @java.lang.Override
+    public final com.google.protobuf.UnknownFieldSet
+    getUnknownFields() {
+      return this.unknownFields;
+    }
+    private NodeList(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      this();
+      if (extensionRegistry == null) {
+        throw new java.lang.NullPointerException();
+      }
+      int mutable_bitField0_ = 0;
+      com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+          com.google.protobuf.UnknownFieldSet.newBuilder();
+      try {
+        boolean done = false;
+        while (!done) {
+          int tag = input.readTag();
+          switch (tag) {
+            case 0:
+              done = true;
+              break;
+            case 10: {
+              java.lang.String s = input.readStringRequireUtf8();
+              if (!((mutable_bitField0_ & 0x00000001) != 0)) {
+                value_ = new com.google.protobuf.LazyStringArrayList();
+                mutable_bitField0_ |= 0x00000001;
+              }
+              value_.add(s);
+              break;
+            }
+            default: {
+              if (!parseUnknownField(
+                  input, unknownFields, extensionRegistry, tag)) {
+                done = true;
+              }
+              break;
+            }
+          }
+        }
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        throw e.setUnfinishedMessage(this);
+      } catch (java.io.IOException e) {
+        throw new com.google.protobuf.InvalidProtocolBufferException(
+            e).setUnfinishedMessage(this);
+      } finally {
+        if (((mutable_bitField0_ & 0x00000001) != 0)) {
+          value_ = value_.getUnmodifiableView();
+        }
+        this.unknownFields = unknownFields.build();
+        makeExtensionsImmutable();
+      }
+    }
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.MetaGraphProtos.internal_static_tensorflow_CollectionDef_NodeList_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.MetaGraphProtos.internal_static_tensorflow_CollectionDef_NodeList_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.CollectionDef.NodeList.class, org.tensorflow.proto.framework.CollectionDef.NodeList.Builder.class);
+    }
+
+    public static final int VALUE_FIELD_NUMBER = 1;
+    private com.google.protobuf.LazyStringList value_;
+    /**
+     * <code>repeated string value = 1;</code>
+     */
+    public com.google.protobuf.ProtocolStringList
+        getValueList() {
+      return value_;
+    }
+    /**
+     * <code>repeated string value = 1;</code>
+     */
+    public int getValueCount() {
+      return value_.size();
+    }
+    /**
+     * <code>repeated string value = 1;</code>
+     */
+    public java.lang.String getValue(int index) {
+      return value_.get(index);
+    }
+    /**
+     * <code>repeated string value = 1;</code>
+     */
+    public com.google.protobuf.ByteString
+        getValueBytes(int index) {
+      return value_.getByteString(index);
+    }
+
+    private byte memoizedIsInitialized = -1;
+    @java.lang.Override
+    public final boolean isInitialized() {
+      byte isInitialized = memoizedIsInitialized;
+      if (isInitialized == 1) return true;
+      if (isInitialized == 0) return false;
+
+      memoizedIsInitialized = 1;
+      return true;
+    }
+
+    @java.lang.Override
+    public void writeTo(com.google.protobuf.CodedOutputStream output)
+                        throws java.io.IOException {
+      for (int i = 0; i < value_.size(); i++) {
+        com.google.protobuf.GeneratedMessageV3.writeString(output, 1, value_.getRaw(i));
+      }
+      unknownFields.writeTo(output);
+    }
+
+    @java.lang.Override
+    public int getSerializedSize() {
+      int size = memoizedSize;
+      if (size != -1) return size;
+
+      size = 0;
+      {
+        int dataSize = 0;
+        for (int i = 0; i < value_.size(); i++) {
+          dataSize += computeStringSizeNoTag(value_.getRaw(i));
+        }
+        size += dataSize;
+        size += 1 * getValueList().size();
+      }
+      size += unknownFields.getSerializedSize();
+      memoizedSize = size;
+      return size;
+    }
+
+    @java.lang.Override
+    public boolean equals(final java.lang.Object obj) {
+      if (obj == this) {
+       return true;
+      }
+      if (!(obj instanceof org.tensorflow.proto.framework.CollectionDef.NodeList)) {
+        return super.equals(obj);
+      }
+      org.tensorflow.proto.framework.CollectionDef.NodeList other = (org.tensorflow.proto.framework.CollectionDef.NodeList) obj;
+
+      if (!getValueList()
+          .equals(other.getValueList())) return false;
+      if (!unknownFields.equals(other.unknownFields)) return false;
+      return true;
+    }
+
+    @java.lang.Override
+    public int hashCode() {
+      if (memoizedHashCode != 0) {
+        return memoizedHashCode;
+      }
+      int hash = 41;
+      hash = (19 * hash) + getDescriptor().hashCode();
+      if (getValueCount() > 0) {
+        hash = (37 * hash) + VALUE_FIELD_NUMBER;
+        hash = (53 * hash) + getValueList().hashCode();
+      }
+      hash = (29 * hash) + unknownFields.hashCode();
+      memoizedHashCode = hash;
+      return hash;
+    }
+
+    public static org.tensorflow.proto.framework.CollectionDef.NodeList parseFrom(
+        java.nio.ByteBuffer data)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data);
+    }
+    public static org.tensorflow.proto.framework.CollectionDef.NodeList parseFrom(
+        java.nio.ByteBuffer data,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.CollectionDef.NodeList parseFrom(
+        com.google.protobuf.ByteString data)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data);
+    }
+    public static org.tensorflow.proto.framework.CollectionDef.NodeList parseFrom(
+        com.google.protobuf.ByteString data,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.CollectionDef.NodeList parseFrom(byte[] data)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data);
+    }
+    public static org.tensorflow.proto.framework.CollectionDef.NodeList parseFrom(
+        byte[] data,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.CollectionDef.NodeList parseFrom(java.io.InputStream input)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input);
+    }
+    public static org.tensorflow.proto.framework.CollectionDef.NodeList parseFrom(
+        java.io.InputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.CollectionDef.NodeList parseDelimitedFrom(java.io.InputStream input)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseDelimitedWithIOException(PARSER, input);
+    }
+    public static org.tensorflow.proto.framework.CollectionDef.NodeList parseDelimitedFrom(
+        java.io.InputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.CollectionDef.NodeList parseFrom(
+        com.google.protobuf.CodedInputStream input)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input);
+    }
+    public static org.tensorflow.proto.framework.CollectionDef.NodeList parseFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input, extensionRegistry);
+    }
+
+    @java.lang.Override
+    public Builder newBuilderForType() { return newBuilder(); }
+    public static Builder newBuilder() {
+      return DEFAULT_INSTANCE.toBuilder();
+    }
+    public static Builder newBuilder(org.tensorflow.proto.framework.CollectionDef.NodeList prototype) {
+      return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+    }
+    @java.lang.Override
+    public Builder toBuilder() {
+      return this == DEFAULT_INSTANCE
+          ? new Builder() : new Builder().mergeFrom(this);
+    }
+
+    @java.lang.Override
+    protected Builder newBuilderForType(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      Builder builder = new Builder(parent);
+      return builder;
+    }
+    /**
+     * <pre>
+     * NodeList is used for collecting nodes in graph. For example
+     * collection_def {
+     *   key: "summaries"
+     *   value {
+     *     node_list {
+     *       value: "input_producer/ScalarSummary:0"
+     *       value: "shuffle_batch/ScalarSummary:0"
+     *       value: "ImageSummary:0"
+     *     }
+     *   }
+     * </pre>
+     *
+     * Protobuf type {@code tensorflow.CollectionDef.NodeList}
+     */
+    public static final class Builder extends
+        com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+        // @@protoc_insertion_point(builder_implements:tensorflow.CollectionDef.NodeList)
+        org.tensorflow.proto.framework.CollectionDef.NodeListOrBuilder {
+      public static final com.google.protobuf.Descriptors.Descriptor
+          getDescriptor() {
+        return org.tensorflow.proto.framework.MetaGraphProtos.internal_static_tensorflow_CollectionDef_NodeList_descriptor;
+      }
+
+      @java.lang.Override
+      protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+          internalGetFieldAccessorTable() {
+        return org.tensorflow.proto.framework.MetaGraphProtos.internal_static_tensorflow_CollectionDef_NodeList_fieldAccessorTable
+            .ensureFieldAccessorsInitialized(
+                org.tensorflow.proto.framework.CollectionDef.NodeList.class, org.tensorflow.proto.framework.CollectionDef.NodeList.Builder.class);
+      }
+
+      // Construct using org.tensorflow.proto.framework.CollectionDef.NodeList.newBuilder()
+      private Builder() {
+        maybeForceBuilderInitialization();
+      }
+
+      private Builder(
+          com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+        super(parent);
+        maybeForceBuilderInitialization();
+      }
+      private void maybeForceBuilderInitialization() {
+        if (com.google.protobuf.GeneratedMessageV3
+                .alwaysUseFieldBuilders) {
+        }
+      }
+      @java.lang.Override
+      public Builder clear() {
+        super.clear();
+        value_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+        bitField0_ = (bitField0_ & ~0x00000001);
+        return this;
+      }
+
+      @java.lang.Override
+      public com.google.protobuf.Descriptors.Descriptor
+          getDescriptorForType() {
+        return org.tensorflow.proto.framework.MetaGraphProtos.internal_static_tensorflow_CollectionDef_NodeList_descriptor;
+      }
+
+      @java.lang.Override
+      public org.tensorflow.proto.framework.CollectionDef.NodeList getDefaultInstanceForType() {
+        return org.tensorflow.proto.framework.CollectionDef.NodeList.getDefaultInstance();
+      }
+
+      @java.lang.Override
+      public org.tensorflow.proto.framework.CollectionDef.NodeList build() {
+        org.tensorflow.proto.framework.CollectionDef.NodeList result = buildPartial();
+        if (!result.isInitialized()) {
+          throw newUninitializedMessageException(result);
+        }
+        return result;
+      }
+
+      @java.lang.Override
+      public org.tensorflow.proto.framework.CollectionDef.NodeList buildPartial() {
+        org.tensorflow.proto.framework.CollectionDef.NodeList result = new org.tensorflow.proto.framework.CollectionDef.NodeList(this);
+        int from_bitField0_ = bitField0_;
+        if (((bitField0_ & 0x00000001) != 0)) {
+          value_ = value_.getUnmodifiableView();
+          bitField0_ = (bitField0_ & ~0x00000001);
+        }
+        result.value_ = value_;
+        onBuilt();
+        return result;
+      }
+
+      @java.lang.Override
+      public Builder clone() {
+        return super.clone();
+      }
+      @java.lang.Override
+      public Builder setField(
+          com.google.protobuf.Descriptors.FieldDescriptor field,
+          java.lang.Object value) {
+        return super.setField(field, value);
+      }
+      @java.lang.Override
+      public Builder clearField(
+          com.google.protobuf.Descriptors.FieldDescriptor field) {
+        return super.clearField(field);
+      }
+      @java.lang.Override
+      public Builder clearOneof(
+          com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+        return super.clearOneof(oneof);
+      }
+      @java.lang.Override
+      public Builder setRepeatedField(
+          com.google.protobuf.Descriptors.FieldDescriptor field,
+          int index, java.lang.Object value) {
+        return super.setRepeatedField(field, index, value);
+      }
+      @java.lang.Override
+      public Builder addRepeatedField(
+          com.google.protobuf.Descriptors.FieldDescriptor field,
+          java.lang.Object value) {
+        return super.addRepeatedField(field, value);
+      }
+      @java.lang.Override
+      public Builder mergeFrom(com.google.protobuf.Message other) {
+        if (other instanceof org.tensorflow.proto.framework.CollectionDef.NodeList) {
+          return mergeFrom((org.tensorflow.proto.framework.CollectionDef.NodeList)other);
+        } else {
+          super.mergeFrom(other);
+          return this;
+        }
+      }
+
+      public Builder mergeFrom(org.tensorflow.proto.framework.CollectionDef.NodeList other) {
+        if (other == org.tensorflow.proto.framework.CollectionDef.NodeList.getDefaultInstance()) return this;
+        if (!other.value_.isEmpty()) {
+          if (value_.isEmpty()) {
+            value_ = other.value_;
+            bitField0_ = (bitField0_ & ~0x00000001);
+          } else {
+            ensureValueIsMutable();
+            value_.addAll(other.value_);
+          }
+          onChanged();
+        }
+        this.mergeUnknownFields(other.unknownFields);
+        onChanged();
+        return this;
+      }
+
+      @java.lang.Override
+      public final boolean isInitialized() {
+        return true;
+      }
+
+      @java.lang.Override
+      public Builder mergeFrom(
+          com.google.protobuf.CodedInputStream input,
+          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+          throws java.io.IOException {
+        org.tensorflow.proto.framework.CollectionDef.NodeList parsedMessage = null;
+        try {
+          parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+        } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+          parsedMessage = (org.tensorflow.proto.framework.CollectionDef.NodeList) e.getUnfinishedMessage();
+          throw e.unwrapIOException();
+        } finally {
+          if (parsedMessage != null) {
+            mergeFrom(parsedMessage);
+          }
+        }
+        return this;
+      }
+      private int bitField0_;
+
+      private com.google.protobuf.LazyStringList value_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+      private void ensureValueIsMutable() {
+        if (!((bitField0_ & 0x00000001) != 0)) {
+          value_ = new com.google.protobuf.LazyStringArrayList(value_);
+          bitField0_ |= 0x00000001;
+         }
+      }
+      /**
+       * <code>repeated string value = 1;</code>
+       */
+      public com.google.protobuf.ProtocolStringList
+          getValueList() {
+        return value_.getUnmodifiableView();
+      }
+      /**
+       * <code>repeated string value = 1;</code>
+       */
+      public int getValueCount() {
+        return value_.size();
+      }
+      /**
+       * <code>repeated string value = 1;</code>
+       */
+      public java.lang.String getValue(int index) {
+        return value_.get(index);
+      }
+      /**
+       * <code>repeated string value = 1;</code>
+       */
+      public com.google.protobuf.ByteString
+          getValueBytes(int index) {
+        return value_.getByteString(index);
+      }
+      /**
+       * <code>repeated string value = 1;</code>
+       */
+      public Builder setValue(
+          int index, java.lang.String value) {
+        if (value == null) {
+    throw new NullPointerException();
+  }
+  ensureValueIsMutable();
+        value_.set(index, value);
+        onChanged();
+        return this;
+      }
+      /**
+       * <code>repeated string value = 1;</code>
+       */
+      public Builder addValue(
+          java.lang.String value) {
+        if (value == null) {
+    throw new NullPointerException();
+  }
+  ensureValueIsMutable();
+        value_.add(value);
+        onChanged();
+        return this;
+      }
+      /**
+       * <code>repeated string value = 1;</code>
+       */
+      public Builder addAllValue(
+          java.lang.Iterable<java.lang.String> values) {
+        ensureValueIsMutable();
+        com.google.protobuf.AbstractMessageLite.Builder.addAll(
+            values, value_);
+        onChanged();
+        return this;
+      }
+      /**
+       * <code>repeated string value = 1;</code>
+       */
+      public Builder clearValue() {
+        value_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+        bitField0_ = (bitField0_ & ~0x00000001);
+        onChanged();
+        return this;
+      }
+      /**
+       * <code>repeated string value = 1;</code>
+       */
+      public Builder addValueBytes(
+          com.google.protobuf.ByteString value) {
+        if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+        ensureValueIsMutable();
+        value_.add(value);
+        onChanged();
+        return this;
+      }
+      @java.lang.Override
+      public final Builder setUnknownFields(
+          final com.google.protobuf.UnknownFieldSet unknownFields) {
+        return super.setUnknownFields(unknownFields);
+      }
+
+      @java.lang.Override
+      public final Builder mergeUnknownFields(
+          final com.google.protobuf.UnknownFieldSet unknownFields) {
+        return super.mergeUnknownFields(unknownFields);
+      }
+
+
+      // @@protoc_insertion_point(builder_scope:tensorflow.CollectionDef.NodeList)
+    }
+
+    // @@protoc_insertion_point(class_scope:tensorflow.CollectionDef.NodeList)
+    private static final org.tensorflow.proto.framework.CollectionDef.NodeList DEFAULT_INSTANCE;
+    static {
+      DEFAULT_INSTANCE = new org.tensorflow.proto.framework.CollectionDef.NodeList();
+    }
+
+    public static org.tensorflow.proto.framework.CollectionDef.NodeList getDefaultInstance() {
+      return DEFAULT_INSTANCE;
+    }
+
+    private static final com.google.protobuf.Parser<NodeList>
+        PARSER = new com.google.protobuf.AbstractParser<NodeList>() {
+      @java.lang.Override
+      public NodeList parsePartialFrom(
+          com.google.protobuf.CodedInputStream input,
+          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+          throws com.google.protobuf.InvalidProtocolBufferException {
+        return new NodeList(input, extensionRegistry);
+      }
+    };
+
+    public static com.google.protobuf.Parser<NodeList> parser() {
+      return PARSER;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Parser<NodeList> getParserForType() {
+      return PARSER;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.CollectionDef.NodeList getDefaultInstanceForType() {
+      return DEFAULT_INSTANCE;
+    }
+
+  }
+
+  public interface BytesListOrBuilder extends
+      // @@protoc_insertion_point(interface_extends:tensorflow.CollectionDef.BytesList)
+      com.google.protobuf.MessageOrBuilder {
+
+    /**
+     * <code>repeated bytes value = 1;</code>
+     */
+    java.util.List<com.google.protobuf.ByteString> getValueList();
+    /**
+     * <code>repeated bytes value = 1;</code>
+     */
+    int getValueCount();
+    /**
+     * <code>repeated bytes value = 1;</code>
+     */
+    com.google.protobuf.ByteString getValue(int index);
+  }
+  /**
+   * <pre>
+   * BytesList is used for collecting strings and serialized protobufs. For
+   * example:
+   * collection_def {
+   *   key: "trainable_variables"
+   *   value {
+   *     bytes_list {
+   *       value: "&#92;n&#92;017conv1/weights:0&#92;022&#92;024conv1/weights/Assign
+   *              &#92;032&#92;024conv1/weights/read:0"
+   *       value: "&#92;n&#92;016conv1/biases:0&#92;022&#92;023conv1/biases/Assign&#92;032
+   *              &#92;023conv1/biases/read:0"
+   *     }
+   *   }
+   * }
+   * </pre>
+   *
+   * Protobuf type {@code tensorflow.CollectionDef.BytesList}
+   */
+  public  static final class BytesList extends
+      com.google.protobuf.GeneratedMessageV3 implements
+      // @@protoc_insertion_point(message_implements:tensorflow.CollectionDef.BytesList)
+      BytesListOrBuilder {
+  private static final long serialVersionUID = 0L;
+    // Use BytesList.newBuilder() to construct.
+    private BytesList(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+      super(builder);
+    }
+    private BytesList() {
+      value_ = java.util.Collections.emptyList();
+    }
+
+    @java.lang.Override
+    @SuppressWarnings({"unused"})
+    protected java.lang.Object newInstance(
+        UnusedPrivateParameter unused) {
+      return new BytesList();
+    }
+
+    @java.lang.Override
+    public final com.google.protobuf.UnknownFieldSet
+    getUnknownFields() {
+      return this.unknownFields;
+    }
+    private BytesList(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      this();
+      if (extensionRegistry == null) {
+        throw new java.lang.NullPointerException();
+      }
+      int mutable_bitField0_ = 0;
+      com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+          com.google.protobuf.UnknownFieldSet.newBuilder();
+      try {
+        boolean done = false;
+        while (!done) {
+          int tag = input.readTag();
+          switch (tag) {
+            case 0:
+              done = true;
+              break;
+            case 10: {
+              if (!((mutable_bitField0_ & 0x00000001) != 0)) {
+                value_ = new java.util.ArrayList<com.google.protobuf.ByteString>();
+                mutable_bitField0_ |= 0x00000001;
+              }
+              value_.add(input.readBytes());
+              break;
+            }
+            default: {
+              if (!parseUnknownField(
+                  input, unknownFields, extensionRegistry, tag)) {
+                done = true;
+              }
+              break;
+            }
+          }
+        }
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        throw e.setUnfinishedMessage(this);
+      } catch (java.io.IOException e) {
+        throw new com.google.protobuf.InvalidProtocolBufferException(
+            e).setUnfinishedMessage(this);
+      } finally {
+        if (((mutable_bitField0_ & 0x00000001) != 0)) {
+          value_ = java.util.Collections.unmodifiableList(value_); // C
+        }
+        this.unknownFields = unknownFields.build();
+        makeExtensionsImmutable();
+      }
+    }
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.MetaGraphProtos.internal_static_tensorflow_CollectionDef_BytesList_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.MetaGraphProtos.internal_static_tensorflow_CollectionDef_BytesList_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.CollectionDef.BytesList.class, org.tensorflow.proto.framework.CollectionDef.BytesList.Builder.class);
+    }
+
+    public static final int VALUE_FIELD_NUMBER = 1;
+    private java.util.List<com.google.protobuf.ByteString> value_;
+    /**
+     * <code>repeated bytes value = 1;</code>
+     */
+    public java.util.List<com.google.protobuf.ByteString>
+        getValueList() {
+      return value_;
+    }
+    /**
+     * <code>repeated bytes value = 1;</code>
+     */
+    public int getValueCount() {
+      return value_.size();
+    }
+    /**
+     * <code>repeated bytes value = 1;</code>
+     */
+    public com.google.protobuf.ByteString getValue(int index) {
+      return value_.get(index);
+    }
+
+    private byte memoizedIsInitialized = -1;
+    @java.lang.Override
+    public final boolean isInitialized() {
+      byte isInitialized = memoizedIsInitialized;
+      if (isInitialized == 1) return true;
+      if (isInitialized == 0) return false;
+
+      memoizedIsInitialized = 1;
+      return true;
+    }
+
+    @java.lang.Override
+    public void writeTo(com.google.protobuf.CodedOutputStream output)
+                        throws java.io.IOException {
+      for (int i = 0; i < value_.size(); i++) {
+        output.writeBytes(1, value_.get(i));
+      }
+      unknownFields.writeTo(output);
+    }
+
+    @java.lang.Override
+    public int getSerializedSize() {
+      int size = memoizedSize;
+      if (size != -1) return size;
+
+      size = 0;
+      {
+        int dataSize = 0;
+        for (int i = 0; i < value_.size(); i++) {
+          dataSize += com.google.protobuf.CodedOutputStream
+            .computeBytesSizeNoTag(value_.get(i));
+        }
+        size += dataSize;
+        size += 1 * getValueList().size();
+      }
+      size += unknownFields.getSerializedSize();
+      memoizedSize = size;
+      return size;
+    }
+
+    @java.lang.Override
+    public boolean equals(final java.lang.Object obj) {
+      if (obj == this) {
+       return true;
+      }
+      if (!(obj instanceof org.tensorflow.proto.framework.CollectionDef.BytesList)) {
+        return super.equals(obj);
+      }
+      org.tensorflow.proto.framework.CollectionDef.BytesList other = (org.tensorflow.proto.framework.CollectionDef.BytesList) obj;
+
+      if (!getValueList()
+          .equals(other.getValueList())) return false;
+      if (!unknownFields.equals(other.unknownFields)) return false;
+      return true;
+    }
+
+    @java.lang.Override
+    public int hashCode() {
+      if (memoizedHashCode != 0) {
+        return memoizedHashCode;
+      }
+      int hash = 41;
+      hash = (19 * hash) + getDescriptor().hashCode();
+      if (getValueCount() > 0) {
+        hash = (37 * hash) + VALUE_FIELD_NUMBER;
+        hash = (53 * hash) + getValueList().hashCode();
+      }
+      hash = (29 * hash) + unknownFields.hashCode();
+      memoizedHashCode = hash;
+      return hash;
+    }
+
+    public static org.tensorflow.proto.framework.CollectionDef.BytesList parseFrom(
+        java.nio.ByteBuffer data)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data);
+    }
+    public static org.tensorflow.proto.framework.CollectionDef.BytesList parseFrom(
+        java.nio.ByteBuffer data,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.CollectionDef.BytesList parseFrom(
+        com.google.protobuf.ByteString data)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data);
+    }
+    public static org.tensorflow.proto.framework.CollectionDef.BytesList parseFrom(
+        com.google.protobuf.ByteString data,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.CollectionDef.BytesList parseFrom(byte[] data)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data);
+    }
+    public static org.tensorflow.proto.framework.CollectionDef.BytesList parseFrom(
+        byte[] data,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.CollectionDef.BytesList parseFrom(java.io.InputStream input)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input);
+    }
+    public static org.tensorflow.proto.framework.CollectionDef.BytesList parseFrom(
+        java.io.InputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.CollectionDef.BytesList parseDelimitedFrom(java.io.InputStream input)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseDelimitedWithIOException(PARSER, input);
+    }
+    public static org.tensorflow.proto.framework.CollectionDef.BytesList parseDelimitedFrom(
+        java.io.InputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.CollectionDef.BytesList parseFrom(
+        com.google.protobuf.CodedInputStream input)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input);
+    }
+    public static org.tensorflow.proto.framework.CollectionDef.BytesList parseFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input, extensionRegistry);
+    }
+
+    @java.lang.Override
+    public Builder newBuilderForType() { return newBuilder(); }
+    public static Builder newBuilder() {
+      return DEFAULT_INSTANCE.toBuilder();
+    }
+    public static Builder newBuilder(org.tensorflow.proto.framework.CollectionDef.BytesList prototype) {
+      return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+    }
+    @java.lang.Override
+    public Builder toBuilder() {
+      return this == DEFAULT_INSTANCE
+          ? new Builder() : new Builder().mergeFrom(this);
+    }
+
+    @java.lang.Override
+    protected Builder newBuilderForType(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      Builder builder = new Builder(parent);
+      return builder;
+    }
+    /**
+     * <pre>
+     * BytesList is used for collecting strings and serialized protobufs. For
+     * example:
+     * collection_def {
+     *   key: "trainable_variables"
+     *   value {
+     *     bytes_list {
+     *       value: "&#92;n&#92;017conv1/weights:0&#92;022&#92;024conv1/weights/Assign
+     *              &#92;032&#92;024conv1/weights/read:0"
+     *       value: "&#92;n&#92;016conv1/biases:0&#92;022&#92;023conv1/biases/Assign&#92;032
+     *              &#92;023conv1/biases/read:0"
+     *     }
+     *   }
+     * }
+     * </pre>
+     *
+     * Protobuf type {@code tensorflow.CollectionDef.BytesList}
+     */
+    public static final class Builder extends
+        com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+        // @@protoc_insertion_point(builder_implements:tensorflow.CollectionDef.BytesList)
+        org.tensorflow.proto.framework.CollectionDef.BytesListOrBuilder {
+      public static final com.google.protobuf.Descriptors.Descriptor
+          getDescriptor() {
+        return org.tensorflow.proto.framework.MetaGraphProtos.internal_static_tensorflow_CollectionDef_BytesList_descriptor;
+      }
+
+      @java.lang.Override
+      protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+          internalGetFieldAccessorTable() {
+        return org.tensorflow.proto.framework.MetaGraphProtos.internal_static_tensorflow_CollectionDef_BytesList_fieldAccessorTable
+            .ensureFieldAccessorsInitialized(
+                org.tensorflow.proto.framework.CollectionDef.BytesList.class, org.tensorflow.proto.framework.CollectionDef.BytesList.Builder.class);
+      }
+
+      // Construct using org.tensorflow.proto.framework.CollectionDef.BytesList.newBuilder()
+      private Builder() {
+        maybeForceBuilderInitialization();
+      }
+
+      private Builder(
+          com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+        super(parent);
+        maybeForceBuilderInitialization();
+      }
+      private void maybeForceBuilderInitialization() {
+        if (com.google.protobuf.GeneratedMessageV3
+                .alwaysUseFieldBuilders) {
+        }
+      }
+      @java.lang.Override
+      public Builder clear() {
+        super.clear();
+        value_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000001);
+        return this;
+      }
+
+      @java.lang.Override
+      public com.google.protobuf.Descriptors.Descriptor
+          getDescriptorForType() {
+        return org.tensorflow.proto.framework.MetaGraphProtos.internal_static_tensorflow_CollectionDef_BytesList_descriptor;
+      }
+
+      @java.lang.Override
+      public org.tensorflow.proto.framework.CollectionDef.BytesList getDefaultInstanceForType() {
+        return org.tensorflow.proto.framework.CollectionDef.BytesList.getDefaultInstance();
+      }
+
+      @java.lang.Override
+      public org.tensorflow.proto.framework.CollectionDef.BytesList build() {
+        org.tensorflow.proto.framework.CollectionDef.BytesList result = buildPartial();
+        if (!result.isInitialized()) {
+          throw newUninitializedMessageException(result);
+        }
+        return result;
+      }
+
+      @java.lang.Override
+      public org.tensorflow.proto.framework.CollectionDef.BytesList buildPartial() {
+        org.tensorflow.proto.framework.CollectionDef.BytesList result = new org.tensorflow.proto.framework.CollectionDef.BytesList(this);
+        int from_bitField0_ = bitField0_;
+        if (((bitField0_ & 0x00000001) != 0)) {
+          value_ = java.util.Collections.unmodifiableList(value_);
+          bitField0_ = (bitField0_ & ~0x00000001);
+        }
+        result.value_ = value_;
+        onBuilt();
+        return result;
+      }
+
+      @java.lang.Override
+      public Builder clone() {
+        return super.clone();
+      }
+      @java.lang.Override
+      public Builder setField(
+          com.google.protobuf.Descriptors.FieldDescriptor field,
+          java.lang.Object value) {
+        return super.setField(field, value);
+      }
+      @java.lang.Override
+      public Builder clearField(
+          com.google.protobuf.Descriptors.FieldDescriptor field) {
+        return super.clearField(field);
+      }
+      @java.lang.Override
+      public Builder clearOneof(
+          com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+        return super.clearOneof(oneof);
+      }
+      @java.lang.Override
+      public Builder setRepeatedField(
+          com.google.protobuf.Descriptors.FieldDescriptor field,
+          int index, java.lang.Object value) {
+        return super.setRepeatedField(field, index, value);
+      }
+      @java.lang.Override
+      public Builder addRepeatedField(
+          com.google.protobuf.Descriptors.FieldDescriptor field,
+          java.lang.Object value) {
+        return super.addRepeatedField(field, value);
+      }
+      @java.lang.Override
+      public Builder mergeFrom(com.google.protobuf.Message other) {
+        if (other instanceof org.tensorflow.proto.framework.CollectionDef.BytesList) {
+          return mergeFrom((org.tensorflow.proto.framework.CollectionDef.BytesList)other);
+        } else {
+          super.mergeFrom(other);
+          return this;
+        }
+      }
+
+      public Builder mergeFrom(org.tensorflow.proto.framework.CollectionDef.BytesList other) {
+        if (other == org.tensorflow.proto.framework.CollectionDef.BytesList.getDefaultInstance()) return this;
+        if (!other.value_.isEmpty()) {
+          if (value_.isEmpty()) {
+            value_ = other.value_;
+            bitField0_ = (bitField0_ & ~0x00000001);
+          } else {
+            ensureValueIsMutable();
+            value_.addAll(other.value_);
+          }
+          onChanged();
+        }
+        this.mergeUnknownFields(other.unknownFields);
+        onChanged();
+        return this;
+      }
+
+      @java.lang.Override
+      public final boolean isInitialized() {
+        return true;
+      }
+
+      @java.lang.Override
+      public Builder mergeFrom(
+          com.google.protobuf.CodedInputStream input,
+          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+          throws java.io.IOException {
+        org.tensorflow.proto.framework.CollectionDef.BytesList parsedMessage = null;
+        try {
+          parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+        } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+          parsedMessage = (org.tensorflow.proto.framework.CollectionDef.BytesList) e.getUnfinishedMessage();
+          throw e.unwrapIOException();
+        } finally {
+          if (parsedMessage != null) {
+            mergeFrom(parsedMessage);
+          }
+        }
+        return this;
+      }
+      private int bitField0_;
+
+      private java.util.List<com.google.protobuf.ByteString> value_ = java.util.Collections.emptyList();
+      private void ensureValueIsMutable() {
+        if (!((bitField0_ & 0x00000001) != 0)) {
+          value_ = new java.util.ArrayList<com.google.protobuf.ByteString>(value_);
+          bitField0_ |= 0x00000001;
+         }
+      }
+      /**
+       * <code>repeated bytes value = 1;</code>
+       */
+      public java.util.List<com.google.protobuf.ByteString>
+          getValueList() {
+        return ((bitField0_ & 0x00000001) != 0) ?
+                 java.util.Collections.unmodifiableList(value_) : value_;
+      }
+      /**
+       * <code>repeated bytes value = 1;</code>
+       */
+      public int getValueCount() {
+        return value_.size();
+      }
+      /**
+       * <code>repeated bytes value = 1;</code>
+       */
+      public com.google.protobuf.ByteString getValue(int index) {
+        return value_.get(index);
+      }
+      /**
+       * <code>repeated bytes value = 1;</code>
+       */
+      public Builder setValue(
+          int index, com.google.protobuf.ByteString value) {
+        if (value == null) {
+    throw new NullPointerException();
+  }
+  ensureValueIsMutable();
+        value_.set(index, value);
+        onChanged();
+        return this;
+      }
+      /**
+       * <code>repeated bytes value = 1;</code>
+       */
+      public Builder addValue(com.google.protobuf.ByteString value) {
+        if (value == null) {
+    throw new NullPointerException();
+  }
+  ensureValueIsMutable();
+        value_.add(value);
+        onChanged();
+        return this;
+      }
+      /**
+       * <code>repeated bytes value = 1;</code>
+       */
+      public Builder addAllValue(
+          java.lang.Iterable<? extends com.google.protobuf.ByteString> values) {
+        ensureValueIsMutable();
+        com.google.protobuf.AbstractMessageLite.Builder.addAll(
+            values, value_);
+        onChanged();
+        return this;
+      }
+      /**
+       * <code>repeated bytes value = 1;</code>
+       */
+      public Builder clearValue() {
+        value_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000001);
+        onChanged();
+        return this;
+      }
+      @java.lang.Override
+      public final Builder setUnknownFields(
+          final com.google.protobuf.UnknownFieldSet unknownFields) {
+        return super.setUnknownFields(unknownFields);
+      }
+
+      @java.lang.Override
+      public final Builder mergeUnknownFields(
+          final com.google.protobuf.UnknownFieldSet unknownFields) {
+        return super.mergeUnknownFields(unknownFields);
+      }
+
+
+      // @@protoc_insertion_point(builder_scope:tensorflow.CollectionDef.BytesList)
+    }
+
+    // @@protoc_insertion_point(class_scope:tensorflow.CollectionDef.BytesList)
+    private static final org.tensorflow.proto.framework.CollectionDef.BytesList DEFAULT_INSTANCE;
+    static {
+      DEFAULT_INSTANCE = new org.tensorflow.proto.framework.CollectionDef.BytesList();
+    }
+
+    public static org.tensorflow.proto.framework.CollectionDef.BytesList getDefaultInstance() {
+      return DEFAULT_INSTANCE;
+    }
+
+    private static final com.google.protobuf.Parser<BytesList>
+        PARSER = new com.google.protobuf.AbstractParser<BytesList>() {
+      @java.lang.Override
+      public BytesList parsePartialFrom(
+          com.google.protobuf.CodedInputStream input,
+          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+          throws com.google.protobuf.InvalidProtocolBufferException {
+        return new BytesList(input, extensionRegistry);
+      }
+    };
+
+    public static com.google.protobuf.Parser<BytesList> parser() {
+      return PARSER;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Parser<BytesList> getParserForType() {
+      return PARSER;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.CollectionDef.BytesList getDefaultInstanceForType() {
+      return DEFAULT_INSTANCE;
+    }
+
+  }
+
+  public interface Int64ListOrBuilder extends
+      // @@protoc_insertion_point(interface_extends:tensorflow.CollectionDef.Int64List)
+      com.google.protobuf.MessageOrBuilder {
+
+    /**
+     * <code>repeated int64 value = 1 [packed = true];</code>
+     */
+    java.util.List<java.lang.Long> getValueList();
+    /**
+     * <code>repeated int64 value = 1 [packed = true];</code>
+     */
+    int getValueCount();
+    /**
+     * <code>repeated int64 value = 1 [packed = true];</code>
+     */
+    long getValue(int index);
+  }
+  /**
+   * <pre>
+   * Int64List is used for collecting int, int64 and long values.
+   * </pre>
+   *
+   * Protobuf type {@code tensorflow.CollectionDef.Int64List}
+   */
+  public  static final class Int64List extends
+      com.google.protobuf.GeneratedMessageV3 implements
+      // @@protoc_insertion_point(message_implements:tensorflow.CollectionDef.Int64List)
+      Int64ListOrBuilder {
+  private static final long serialVersionUID = 0L;
+    // Use Int64List.newBuilder() to construct.
+    private Int64List(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+      super(builder);
+    }
+    private Int64List() {
+      value_ = emptyLongList();
+    }
+
+    @java.lang.Override
+    @SuppressWarnings({"unused"})
+    protected java.lang.Object newInstance(
+        UnusedPrivateParameter unused) {
+      return new Int64List();
+    }
+
+    @java.lang.Override
+    public final com.google.protobuf.UnknownFieldSet
+    getUnknownFields() {
+      return this.unknownFields;
+    }
+    private Int64List(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      this();
+      if (extensionRegistry == null) {
+        throw new java.lang.NullPointerException();
+      }
+      int mutable_bitField0_ = 0;
+      com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+          com.google.protobuf.UnknownFieldSet.newBuilder();
+      try {
+        boolean done = false;
+        while (!done) {
+          int tag = input.readTag();
+          switch (tag) {
+            case 0:
+              done = true;
+              break;
+            case 8: {
+              if (!((mutable_bitField0_ & 0x00000001) != 0)) {
+                value_ = newLongList();
+                mutable_bitField0_ |= 0x00000001;
+              }
+              value_.addLong(input.readInt64());
+              break;
+            }
+            case 10: {
+              int length = input.readRawVarint32();
+              int limit = input.pushLimit(length);
+              if (!((mutable_bitField0_ & 0x00000001) != 0) && input.getBytesUntilLimit() > 0) {
+                value_ = newLongList();
+                mutable_bitField0_ |= 0x00000001;
+              }
+              while (input.getBytesUntilLimit() > 0) {
+                value_.addLong(input.readInt64());
+              }
+              input.popLimit(limit);
+              break;
+            }
+            default: {
+              if (!parseUnknownField(
+                  input, unknownFields, extensionRegistry, tag)) {
+                done = true;
+              }
+              break;
+            }
+          }
+        }
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        throw e.setUnfinishedMessage(this);
+      } catch (java.io.IOException e) {
+        throw new com.google.protobuf.InvalidProtocolBufferException(
+            e).setUnfinishedMessage(this);
+      } finally {
+        if (((mutable_bitField0_ & 0x00000001) != 0)) {
+          value_.makeImmutable(); // C
+        }
+        this.unknownFields = unknownFields.build();
+        makeExtensionsImmutable();
+      }
+    }
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.MetaGraphProtos.internal_static_tensorflow_CollectionDef_Int64List_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.MetaGraphProtos.internal_static_tensorflow_CollectionDef_Int64List_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.CollectionDef.Int64List.class, org.tensorflow.proto.framework.CollectionDef.Int64List.Builder.class);
+    }
+
+    public static final int VALUE_FIELD_NUMBER = 1;
+    private com.google.protobuf.Internal.LongList value_;
+    /**
+     * <code>repeated int64 value = 1 [packed = true];</code>
+     */
+    public java.util.List<java.lang.Long>
+        getValueList() {
+      return value_;
+    }
+    /**
+     * <code>repeated int64 value = 1 [packed = true];</code>
+     */
+    public int getValueCount() {
+      return value_.size();
+    }
+    /**
+     * <code>repeated int64 value = 1 [packed = true];</code>
+     */
+    public long getValue(int index) {
+      return value_.getLong(index);
+    }
+    private int valueMemoizedSerializedSize = -1;
+
+    private byte memoizedIsInitialized = -1;
+    @java.lang.Override
+    public final boolean isInitialized() {
+      byte isInitialized = memoizedIsInitialized;
+      if (isInitialized == 1) return true;
+      if (isInitialized == 0) return false;
+
+      memoizedIsInitialized = 1;
+      return true;
+    }
+
+    @java.lang.Override
+    public void writeTo(com.google.protobuf.CodedOutputStream output)
+                        throws java.io.IOException {
+      getSerializedSize();
+      if (getValueList().size() > 0) {
+        output.writeUInt32NoTag(10);
+        output.writeUInt32NoTag(valueMemoizedSerializedSize);
+      }
+      for (int i = 0; i < value_.size(); i++) {
+        output.writeInt64NoTag(value_.getLong(i));
+      }
+      unknownFields.writeTo(output);
+    }
+
+    @java.lang.Override
+    public int getSerializedSize() {
+      int size = memoizedSize;
+      if (size != -1) return size;
+
+      size = 0;
+      {
+        int dataSize = 0;
+        for (int i = 0; i < value_.size(); i++) {
+          dataSize += com.google.protobuf.CodedOutputStream
+            .computeInt64SizeNoTag(value_.getLong(i));
+        }
+        size += dataSize;
+        if (!getValueList().isEmpty()) {
+          size += 1;
+          size += com.google.protobuf.CodedOutputStream
+              .computeInt32SizeNoTag(dataSize);
+        }
+        valueMemoizedSerializedSize = dataSize;
+      }
+      size += unknownFields.getSerializedSize();
+      memoizedSize = size;
+      return size;
+    }
+
+    @java.lang.Override
+    public boolean equals(final java.lang.Object obj) {
+      if (obj == this) {
+       return true;
+      }
+      if (!(obj instanceof org.tensorflow.proto.framework.CollectionDef.Int64List)) {
+        return super.equals(obj);
+      }
+      org.tensorflow.proto.framework.CollectionDef.Int64List other = (org.tensorflow.proto.framework.CollectionDef.Int64List) obj;
+
+      if (!getValueList()
+          .equals(other.getValueList())) return false;
+      if (!unknownFields.equals(other.unknownFields)) return false;
+      return true;
+    }
+
+    @java.lang.Override
+    public int hashCode() {
+      if (memoizedHashCode != 0) {
+        return memoizedHashCode;
+      }
+      int hash = 41;
+      hash = (19 * hash) + getDescriptor().hashCode();
+      if (getValueCount() > 0) {
+        hash = (37 * hash) + VALUE_FIELD_NUMBER;
+        hash = (53 * hash) + getValueList().hashCode();
+      }
+      hash = (29 * hash) + unknownFields.hashCode();
+      memoizedHashCode = hash;
+      return hash;
+    }
+
+    public static org.tensorflow.proto.framework.CollectionDef.Int64List parseFrom(
+        java.nio.ByteBuffer data)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data);
+    }
+    public static org.tensorflow.proto.framework.CollectionDef.Int64List parseFrom(
+        java.nio.ByteBuffer data,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.CollectionDef.Int64List parseFrom(
+        com.google.protobuf.ByteString data)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data);
+    }
+    public static org.tensorflow.proto.framework.CollectionDef.Int64List parseFrom(
+        com.google.protobuf.ByteString data,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.CollectionDef.Int64List parseFrom(byte[] data)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data);
+    }
+    public static org.tensorflow.proto.framework.CollectionDef.Int64List parseFrom(
+        byte[] data,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.CollectionDef.Int64List parseFrom(java.io.InputStream input)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input);
+    }
+    public static org.tensorflow.proto.framework.CollectionDef.Int64List parseFrom(
+        java.io.InputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.CollectionDef.Int64List parseDelimitedFrom(java.io.InputStream input)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseDelimitedWithIOException(PARSER, input);
+    }
+    public static org.tensorflow.proto.framework.CollectionDef.Int64List parseDelimitedFrom(
+        java.io.InputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.CollectionDef.Int64List parseFrom(
+        com.google.protobuf.CodedInputStream input)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input);
+    }
+    public static org.tensorflow.proto.framework.CollectionDef.Int64List parseFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input, extensionRegistry);
+    }
+
+    @java.lang.Override
+    public Builder newBuilderForType() { return newBuilder(); }
+    public static Builder newBuilder() {
+      return DEFAULT_INSTANCE.toBuilder();
+    }
+    public static Builder newBuilder(org.tensorflow.proto.framework.CollectionDef.Int64List prototype) {
+      return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+    }
+    @java.lang.Override
+    public Builder toBuilder() {
+      return this == DEFAULT_INSTANCE
+          ? new Builder() : new Builder().mergeFrom(this);
+    }
+
+    @java.lang.Override
+    protected Builder newBuilderForType(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      Builder builder = new Builder(parent);
+      return builder;
+    }
+    /**
+     * <pre>
+     * Int64List is used for collecting int, int64 and long values.
+     * </pre>
+     *
+     * Protobuf type {@code tensorflow.CollectionDef.Int64List}
+     */
+    public static final class Builder extends
+        com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+        // @@protoc_insertion_point(builder_implements:tensorflow.CollectionDef.Int64List)
+        org.tensorflow.proto.framework.CollectionDef.Int64ListOrBuilder {
+      public static final com.google.protobuf.Descriptors.Descriptor
+          getDescriptor() {
+        return org.tensorflow.proto.framework.MetaGraphProtos.internal_static_tensorflow_CollectionDef_Int64List_descriptor;
+      }
+
+      @java.lang.Override
+      protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+          internalGetFieldAccessorTable() {
+        return org.tensorflow.proto.framework.MetaGraphProtos.internal_static_tensorflow_CollectionDef_Int64List_fieldAccessorTable
+            .ensureFieldAccessorsInitialized(
+                org.tensorflow.proto.framework.CollectionDef.Int64List.class, org.tensorflow.proto.framework.CollectionDef.Int64List.Builder.class);
+      }
+
+      // Construct using org.tensorflow.proto.framework.CollectionDef.Int64List.newBuilder()
+      private Builder() {
+        maybeForceBuilderInitialization();
+      }
+
+      private Builder(
+          com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+        super(parent);
+        maybeForceBuilderInitialization();
+      }
+      private void maybeForceBuilderInitialization() {
+        if (com.google.protobuf.GeneratedMessageV3
+                .alwaysUseFieldBuilders) {
+        }
+      }
+      @java.lang.Override
+      public Builder clear() {
+        super.clear();
+        value_ = emptyLongList();
+        bitField0_ = (bitField0_ & ~0x00000001);
+        return this;
+      }
+
+      @java.lang.Override
+      public com.google.protobuf.Descriptors.Descriptor
+          getDescriptorForType() {
+        return org.tensorflow.proto.framework.MetaGraphProtos.internal_static_tensorflow_CollectionDef_Int64List_descriptor;
+      }
+
+      @java.lang.Override
+      public org.tensorflow.proto.framework.CollectionDef.Int64List getDefaultInstanceForType() {
+        return org.tensorflow.proto.framework.CollectionDef.Int64List.getDefaultInstance();
+      }
+
+      @java.lang.Override
+      public org.tensorflow.proto.framework.CollectionDef.Int64List build() {
+        org.tensorflow.proto.framework.CollectionDef.Int64List result = buildPartial();
+        if (!result.isInitialized()) {
+          throw newUninitializedMessageException(result);
+        }
+        return result;
+      }
+
+      @java.lang.Override
+      public org.tensorflow.proto.framework.CollectionDef.Int64List buildPartial() {
+        org.tensorflow.proto.framework.CollectionDef.Int64List result = new org.tensorflow.proto.framework.CollectionDef.Int64List(this);
+        int from_bitField0_ = bitField0_;
+        if (((bitField0_ & 0x00000001) != 0)) {
+          value_.makeImmutable();
+          bitField0_ = (bitField0_ & ~0x00000001);
+        }
+        result.value_ = value_;
+        onBuilt();
+        return result;
+      }
+
+      @java.lang.Override
+      public Builder clone() {
+        return super.clone();
+      }
+      @java.lang.Override
+      public Builder setField(
+          com.google.protobuf.Descriptors.FieldDescriptor field,
+          java.lang.Object value) {
+        return super.setField(field, value);
+      }
+      @java.lang.Override
+      public Builder clearField(
+          com.google.protobuf.Descriptors.FieldDescriptor field) {
+        return super.clearField(field);
+      }
+      @java.lang.Override
+      public Builder clearOneof(
+          com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+        return super.clearOneof(oneof);
+      }
+      @java.lang.Override
+      public Builder setRepeatedField(
+          com.google.protobuf.Descriptors.FieldDescriptor field,
+          int index, java.lang.Object value) {
+        return super.setRepeatedField(field, index, value);
+      }
+      @java.lang.Override
+      public Builder addRepeatedField(
+          com.google.protobuf.Descriptors.FieldDescriptor field,
+          java.lang.Object value) {
+        return super.addRepeatedField(field, value);
+      }
+      @java.lang.Override
+      public Builder mergeFrom(com.google.protobuf.Message other) {
+        if (other instanceof org.tensorflow.proto.framework.CollectionDef.Int64List) {
+          return mergeFrom((org.tensorflow.proto.framework.CollectionDef.Int64List)other);
+        } else {
+          super.mergeFrom(other);
+          return this;
+        }
+      }
+
+      public Builder mergeFrom(org.tensorflow.proto.framework.CollectionDef.Int64List other) {
+        if (other == org.tensorflow.proto.framework.CollectionDef.Int64List.getDefaultInstance()) return this;
+        if (!other.value_.isEmpty()) {
+          if (value_.isEmpty()) {
+            value_ = other.value_;
+            bitField0_ = (bitField0_ & ~0x00000001);
+          } else {
+            ensureValueIsMutable();
+            value_.addAll(other.value_);
+          }
+          onChanged();
+        }
+        this.mergeUnknownFields(other.unknownFields);
+        onChanged();
+        return this;
+      }
+
+      @java.lang.Override
+      public final boolean isInitialized() {
+        return true;
+      }
+
+      @java.lang.Override
+      public Builder mergeFrom(
+          com.google.protobuf.CodedInputStream input,
+          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+          throws java.io.IOException {
+        org.tensorflow.proto.framework.CollectionDef.Int64List parsedMessage = null;
+        try {
+          parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+        } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+          parsedMessage = (org.tensorflow.proto.framework.CollectionDef.Int64List) e.getUnfinishedMessage();
+          throw e.unwrapIOException();
+        } finally {
+          if (parsedMessage != null) {
+            mergeFrom(parsedMessage);
+          }
+        }
+        return this;
+      }
+      private int bitField0_;
+
+      private com.google.protobuf.Internal.LongList value_ = emptyLongList();
+      private void ensureValueIsMutable() {
+        if (!((bitField0_ & 0x00000001) != 0)) {
+          value_ = mutableCopy(value_);
+          bitField0_ |= 0x00000001;
+         }
+      }
+      /**
+       * <code>repeated int64 value = 1 [packed = true];</code>
+       */
+      public java.util.List<java.lang.Long>
+          getValueList() {
+        return ((bitField0_ & 0x00000001) != 0) ?
+                 java.util.Collections.unmodifiableList(value_) : value_;
+      }
+      /**
+       * <code>repeated int64 value = 1 [packed = true];</code>
+       */
+      public int getValueCount() {
+        return value_.size();
+      }
+      /**
+       * <code>repeated int64 value = 1 [packed = true];</code>
+       */
+      public long getValue(int index) {
+        return value_.getLong(index);
+      }
+      /**
+       * <code>repeated int64 value = 1 [packed = true];</code>
+       */
+      public Builder setValue(
+          int index, long value) {
+        ensureValueIsMutable();
+        value_.setLong(index, value);
+        onChanged();
+        return this;
+      }
+      /**
+       * <code>repeated int64 value = 1 [packed = true];</code>
+       */
+      public Builder addValue(long value) {
+        ensureValueIsMutable();
+        value_.addLong(value);
+        onChanged();
+        return this;
+      }
+      /**
+       * <code>repeated int64 value = 1 [packed = true];</code>
+       */
+      public Builder addAllValue(
+          java.lang.Iterable<? extends java.lang.Long> values) {
+        ensureValueIsMutable();
+        com.google.protobuf.AbstractMessageLite.Builder.addAll(
+            values, value_);
+        onChanged();
+        return this;
+      }
+      /**
+       * <code>repeated int64 value = 1 [packed = true];</code>
+       */
+      public Builder clearValue() {
+        value_ = emptyLongList();
+        bitField0_ = (bitField0_ & ~0x00000001);
+        onChanged();
+        return this;
+      }
+      @java.lang.Override
+      public final Builder setUnknownFields(
+          final com.google.protobuf.UnknownFieldSet unknownFields) {
+        return super.setUnknownFields(unknownFields);
+      }
+
+      @java.lang.Override
+      public final Builder mergeUnknownFields(
+          final com.google.protobuf.UnknownFieldSet unknownFields) {
+        return super.mergeUnknownFields(unknownFields);
+      }
+
+
+      // @@protoc_insertion_point(builder_scope:tensorflow.CollectionDef.Int64List)
+    }
+
+    // @@protoc_insertion_point(class_scope:tensorflow.CollectionDef.Int64List)
+    private static final org.tensorflow.proto.framework.CollectionDef.Int64List DEFAULT_INSTANCE;
+    static {
+      DEFAULT_INSTANCE = new org.tensorflow.proto.framework.CollectionDef.Int64List();
+    }
+
+    public static org.tensorflow.proto.framework.CollectionDef.Int64List getDefaultInstance() {
+      return DEFAULT_INSTANCE;
+    }
+
+    private static final com.google.protobuf.Parser<Int64List>
+        PARSER = new com.google.protobuf.AbstractParser<Int64List>() {
+      @java.lang.Override
+      public Int64List parsePartialFrom(
+          com.google.protobuf.CodedInputStream input,
+          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+          throws com.google.protobuf.InvalidProtocolBufferException {
+        return new Int64List(input, extensionRegistry);
+      }
+    };
+
+    public static com.google.protobuf.Parser<Int64List> parser() {
+      return PARSER;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Parser<Int64List> getParserForType() {
+      return PARSER;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.CollectionDef.Int64List getDefaultInstanceForType() {
+      return DEFAULT_INSTANCE;
+    }
+
+  }
+
+  public interface FloatListOrBuilder extends
+      // @@protoc_insertion_point(interface_extends:tensorflow.CollectionDef.FloatList)
+      com.google.protobuf.MessageOrBuilder {
+
+    /**
+     * <code>repeated float value = 1 [packed = true];</code>
+     */
+    java.util.List<java.lang.Float> getValueList();
+    /**
+     * <code>repeated float value = 1 [packed = true];</code>
+     */
+    int getValueCount();
+    /**
+     * <code>repeated float value = 1 [packed = true];</code>
+     */
+    float getValue(int index);
+  }
+  /**
+   * <pre>
+   * FloatList is used for collecting float values.
+   * </pre>
+   *
+   * Protobuf type {@code tensorflow.CollectionDef.FloatList}
+   */
+  public  static final class FloatList extends
+      com.google.protobuf.GeneratedMessageV3 implements
+      // @@protoc_insertion_point(message_implements:tensorflow.CollectionDef.FloatList)
+      FloatListOrBuilder {
+  private static final long serialVersionUID = 0L;
+    // Use FloatList.newBuilder() to construct.
+    private FloatList(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+      super(builder);
+    }
+    private FloatList() {
+      value_ = emptyFloatList();
+    }
+
+    @java.lang.Override
+    @SuppressWarnings({"unused"})
+    protected java.lang.Object newInstance(
+        UnusedPrivateParameter unused) {
+      return new FloatList();
+    }
+
+    @java.lang.Override
+    public final com.google.protobuf.UnknownFieldSet
+    getUnknownFields() {
+      return this.unknownFields;
+    }
+    private FloatList(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      this();
+      if (extensionRegistry == null) {
+        throw new java.lang.NullPointerException();
+      }
+      int mutable_bitField0_ = 0;
+      com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+          com.google.protobuf.UnknownFieldSet.newBuilder();
+      try {
+        boolean done = false;
+        while (!done) {
+          int tag = input.readTag();
+          switch (tag) {
+            case 0:
+              done = true;
+              break;
+            case 13: {
+              if (!((mutable_bitField0_ & 0x00000001) != 0)) {
+                value_ = newFloatList();
+                mutable_bitField0_ |= 0x00000001;
+              }
+              value_.addFloat(input.readFloat());
+              break;
+            }
+            case 10: {
+              int length = input.readRawVarint32();
+              int limit = input.pushLimit(length);
+              if (!((mutable_bitField0_ & 0x00000001) != 0) && input.getBytesUntilLimit() > 0) {
+                value_ = newFloatList();
+                mutable_bitField0_ |= 0x00000001;
+              }
+              while (input.getBytesUntilLimit() > 0) {
+                value_.addFloat(input.readFloat());
+              }
+              input.popLimit(limit);
+              break;
+            }
+            default: {
+              if (!parseUnknownField(
+                  input, unknownFields, extensionRegistry, tag)) {
+                done = true;
+              }
+              break;
+            }
+          }
+        }
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        throw e.setUnfinishedMessage(this);
+      } catch (java.io.IOException e) {
+        throw new com.google.protobuf.InvalidProtocolBufferException(
+            e).setUnfinishedMessage(this);
+      } finally {
+        if (((mutable_bitField0_ & 0x00000001) != 0)) {
+          value_.makeImmutable(); // C
+        }
+        this.unknownFields = unknownFields.build();
+        makeExtensionsImmutable();
+      }
+    }
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.MetaGraphProtos.internal_static_tensorflow_CollectionDef_FloatList_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.MetaGraphProtos.internal_static_tensorflow_CollectionDef_FloatList_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.CollectionDef.FloatList.class, org.tensorflow.proto.framework.CollectionDef.FloatList.Builder.class);
+    }
+
+    public static final int VALUE_FIELD_NUMBER = 1;
+    private com.google.protobuf.Internal.FloatList value_;
+    /**
+     * <code>repeated float value = 1 [packed = true];</code>
+     */
+    public java.util.List<java.lang.Float>
+        getValueList() {
+      return value_;
+    }
+    /**
+     * <code>repeated float value = 1 [packed = true];</code>
+     */
+    public int getValueCount() {
+      return value_.size();
+    }
+    /**
+     * <code>repeated float value = 1 [packed = true];</code>
+     */
+    public float getValue(int index) {
+      return value_.getFloat(index);
+    }
+    private int valueMemoizedSerializedSize = -1;
+
+    private byte memoizedIsInitialized = -1;
+    @java.lang.Override
+    public final boolean isInitialized() {
+      byte isInitialized = memoizedIsInitialized;
+      if (isInitialized == 1) return true;
+      if (isInitialized == 0) return false;
+
+      memoizedIsInitialized = 1;
+      return true;
+    }
+
+    @java.lang.Override
+    public void writeTo(com.google.protobuf.CodedOutputStream output)
+                        throws java.io.IOException {
+      getSerializedSize();
+      if (getValueList().size() > 0) {
+        output.writeUInt32NoTag(10);
+        output.writeUInt32NoTag(valueMemoizedSerializedSize);
+      }
+      for (int i = 0; i < value_.size(); i++) {
+        output.writeFloatNoTag(value_.getFloat(i));
+      }
+      unknownFields.writeTo(output);
+    }
+
+    @java.lang.Override
+    public int getSerializedSize() {
+      int size = memoizedSize;
+      if (size != -1) return size;
+
+      size = 0;
+      {
+        int dataSize = 0;
+        dataSize = 4 * getValueList().size();
+        size += dataSize;
+        if (!getValueList().isEmpty()) {
+          size += 1;
+          size += com.google.protobuf.CodedOutputStream
+              .computeInt32SizeNoTag(dataSize);
+        }
+        valueMemoizedSerializedSize = dataSize;
+      }
+      size += unknownFields.getSerializedSize();
+      memoizedSize = size;
+      return size;
+    }
+
+    @java.lang.Override
+    public boolean equals(final java.lang.Object obj) {
+      if (obj == this) {
+       return true;
+      }
+      if (!(obj instanceof org.tensorflow.proto.framework.CollectionDef.FloatList)) {
+        return super.equals(obj);
+      }
+      org.tensorflow.proto.framework.CollectionDef.FloatList other = (org.tensorflow.proto.framework.CollectionDef.FloatList) obj;
+
+      if (!getValueList()
+          .equals(other.getValueList())) return false;
+      if (!unknownFields.equals(other.unknownFields)) return false;
+      return true;
+    }
+
+    @java.lang.Override
+    public int hashCode() {
+      if (memoizedHashCode != 0) {
+        return memoizedHashCode;
+      }
+      int hash = 41;
+      hash = (19 * hash) + getDescriptor().hashCode();
+      if (getValueCount() > 0) {
+        hash = (37 * hash) + VALUE_FIELD_NUMBER;
+        hash = (53 * hash) + getValueList().hashCode();
+      }
+      hash = (29 * hash) + unknownFields.hashCode();
+      memoizedHashCode = hash;
+      return hash;
+    }
+
+    public static org.tensorflow.proto.framework.CollectionDef.FloatList parseFrom(
+        java.nio.ByteBuffer data)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data);
+    }
+    public static org.tensorflow.proto.framework.CollectionDef.FloatList parseFrom(
+        java.nio.ByteBuffer data,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.CollectionDef.FloatList parseFrom(
+        com.google.protobuf.ByteString data)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data);
+    }
+    public static org.tensorflow.proto.framework.CollectionDef.FloatList parseFrom(
+        com.google.protobuf.ByteString data,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.CollectionDef.FloatList parseFrom(byte[] data)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data);
+    }
+    public static org.tensorflow.proto.framework.CollectionDef.FloatList parseFrom(
+        byte[] data,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.CollectionDef.FloatList parseFrom(java.io.InputStream input)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input);
+    }
+    public static org.tensorflow.proto.framework.CollectionDef.FloatList parseFrom(
+        java.io.InputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.CollectionDef.FloatList parseDelimitedFrom(java.io.InputStream input)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseDelimitedWithIOException(PARSER, input);
+    }
+    public static org.tensorflow.proto.framework.CollectionDef.FloatList parseDelimitedFrom(
+        java.io.InputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.CollectionDef.FloatList parseFrom(
+        com.google.protobuf.CodedInputStream input)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input);
+    }
+    public static org.tensorflow.proto.framework.CollectionDef.FloatList parseFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input, extensionRegistry);
+    }
+
+    @java.lang.Override
+    public Builder newBuilderForType() { return newBuilder(); }
+    public static Builder newBuilder() {
+      return DEFAULT_INSTANCE.toBuilder();
+    }
+    public static Builder newBuilder(org.tensorflow.proto.framework.CollectionDef.FloatList prototype) {
+      return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+    }
+    @java.lang.Override
+    public Builder toBuilder() {
+      return this == DEFAULT_INSTANCE
+          ? new Builder() : new Builder().mergeFrom(this);
+    }
+
+    @java.lang.Override
+    protected Builder newBuilderForType(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      Builder builder = new Builder(parent);
+      return builder;
+    }
+    /**
+     * <pre>
+     * FloatList is used for collecting float values.
+     * </pre>
+     *
+     * Protobuf type {@code tensorflow.CollectionDef.FloatList}
+     */
+    public static final class Builder extends
+        com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+        // @@protoc_insertion_point(builder_implements:tensorflow.CollectionDef.FloatList)
+        org.tensorflow.proto.framework.CollectionDef.FloatListOrBuilder {
+      public static final com.google.protobuf.Descriptors.Descriptor
+          getDescriptor() {
+        return org.tensorflow.proto.framework.MetaGraphProtos.internal_static_tensorflow_CollectionDef_FloatList_descriptor;
+      }
+
+      @java.lang.Override
+      protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+          internalGetFieldAccessorTable() {
+        return org.tensorflow.proto.framework.MetaGraphProtos.internal_static_tensorflow_CollectionDef_FloatList_fieldAccessorTable
+            .ensureFieldAccessorsInitialized(
+                org.tensorflow.proto.framework.CollectionDef.FloatList.class, org.tensorflow.proto.framework.CollectionDef.FloatList.Builder.class);
+      }
+
+      // Construct using org.tensorflow.proto.framework.CollectionDef.FloatList.newBuilder()
+      private Builder() {
+        maybeForceBuilderInitialization();
+      }
+
+      private Builder(
+          com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+        super(parent);
+        maybeForceBuilderInitialization();
+      }
+      private void maybeForceBuilderInitialization() {
+        if (com.google.protobuf.GeneratedMessageV3
+                .alwaysUseFieldBuilders) {
+        }
+      }
+      @java.lang.Override
+      public Builder clear() {
+        super.clear();
+        value_ = emptyFloatList();
+        bitField0_ = (bitField0_ & ~0x00000001);
+        return this;
+      }
+
+      @java.lang.Override
+      public com.google.protobuf.Descriptors.Descriptor
+          getDescriptorForType() {
+        return org.tensorflow.proto.framework.MetaGraphProtos.internal_static_tensorflow_CollectionDef_FloatList_descriptor;
+      }
+
+      @java.lang.Override
+      public org.tensorflow.proto.framework.CollectionDef.FloatList getDefaultInstanceForType() {
+        return org.tensorflow.proto.framework.CollectionDef.FloatList.getDefaultInstance();
+      }
+
+      @java.lang.Override
+      public org.tensorflow.proto.framework.CollectionDef.FloatList build() {
+        org.tensorflow.proto.framework.CollectionDef.FloatList result = buildPartial();
+        if (!result.isInitialized()) {
+          throw newUninitializedMessageException(result);
+        }
+        return result;
+      }
+
+      @java.lang.Override
+      public org.tensorflow.proto.framework.CollectionDef.FloatList buildPartial() {
+        org.tensorflow.proto.framework.CollectionDef.FloatList result = new org.tensorflow.proto.framework.CollectionDef.FloatList(this);
+        int from_bitField0_ = bitField0_;
+        if (((bitField0_ & 0x00000001) != 0)) {
+          value_.makeImmutable();
+          bitField0_ = (bitField0_ & ~0x00000001);
+        }
+        result.value_ = value_;
+        onBuilt();
+        return result;
+      }
+
+      @java.lang.Override
+      public Builder clone() {
+        return super.clone();
+      }
+      @java.lang.Override
+      public Builder setField(
+          com.google.protobuf.Descriptors.FieldDescriptor field,
+          java.lang.Object value) {
+        return super.setField(field, value);
+      }
+      @java.lang.Override
+      public Builder clearField(
+          com.google.protobuf.Descriptors.FieldDescriptor field) {
+        return super.clearField(field);
+      }
+      @java.lang.Override
+      public Builder clearOneof(
+          com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+        return super.clearOneof(oneof);
+      }
+      @java.lang.Override
+      public Builder setRepeatedField(
+          com.google.protobuf.Descriptors.FieldDescriptor field,
+          int index, java.lang.Object value) {
+        return super.setRepeatedField(field, index, value);
+      }
+      @java.lang.Override
+      public Builder addRepeatedField(
+          com.google.protobuf.Descriptors.FieldDescriptor field,
+          java.lang.Object value) {
+        return super.addRepeatedField(field, value);
+      }
+      @java.lang.Override
+      public Builder mergeFrom(com.google.protobuf.Message other) {
+        if (other instanceof org.tensorflow.proto.framework.CollectionDef.FloatList) {
+          return mergeFrom((org.tensorflow.proto.framework.CollectionDef.FloatList)other);
+        } else {
+          super.mergeFrom(other);
+          return this;
+        }
+      }
+
+      public Builder mergeFrom(org.tensorflow.proto.framework.CollectionDef.FloatList other) {
+        if (other == org.tensorflow.proto.framework.CollectionDef.FloatList.getDefaultInstance()) return this;
+        if (!other.value_.isEmpty()) {
+          if (value_.isEmpty()) {
+            value_ = other.value_;
+            bitField0_ = (bitField0_ & ~0x00000001);
+          } else {
+            ensureValueIsMutable();
+            value_.addAll(other.value_);
+          }
+          onChanged();
+        }
+        this.mergeUnknownFields(other.unknownFields);
+        onChanged();
+        return this;
+      }
+
+      @java.lang.Override
+      public final boolean isInitialized() {
+        return true;
+      }
+
+      @java.lang.Override
+      public Builder mergeFrom(
+          com.google.protobuf.CodedInputStream input,
+          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+          throws java.io.IOException {
+        org.tensorflow.proto.framework.CollectionDef.FloatList parsedMessage = null;
+        try {
+          parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+        } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+          parsedMessage = (org.tensorflow.proto.framework.CollectionDef.FloatList) e.getUnfinishedMessage();
+          throw e.unwrapIOException();
+        } finally {
+          if (parsedMessage != null) {
+            mergeFrom(parsedMessage);
+          }
+        }
+        return this;
+      }
+      private int bitField0_;
+
+      private com.google.protobuf.Internal.FloatList value_ = emptyFloatList();
+      private void ensureValueIsMutable() {
+        if (!((bitField0_ & 0x00000001) != 0)) {
+          value_ = mutableCopy(value_);
+          bitField0_ |= 0x00000001;
+         }
+      }
+      /**
+       * <code>repeated float value = 1 [packed = true];</code>
+       */
+      public java.util.List<java.lang.Float>
+          getValueList() {
+        return ((bitField0_ & 0x00000001) != 0) ?
+                 java.util.Collections.unmodifiableList(value_) : value_;
+      }
+      /**
+       * <code>repeated float value = 1 [packed = true];</code>
+       */
+      public int getValueCount() {
+        return value_.size();
+      }
+      /**
+       * <code>repeated float value = 1 [packed = true];</code>
+       */
+      public float getValue(int index) {
+        return value_.getFloat(index);
+      }
+      /**
+       * <code>repeated float value = 1 [packed = true];</code>
+       */
+      public Builder setValue(
+          int index, float value) {
+        ensureValueIsMutable();
+        value_.setFloat(index, value);
+        onChanged();
+        return this;
+      }
+      /**
+       * <code>repeated float value = 1 [packed = true];</code>
+       */
+      public Builder addValue(float value) {
+        ensureValueIsMutable();
+        value_.addFloat(value);
+        onChanged();
+        return this;
+      }
+      /**
+       * <code>repeated float value = 1 [packed = true];</code>
+       */
+      public Builder addAllValue(
+          java.lang.Iterable<? extends java.lang.Float> values) {
+        ensureValueIsMutable();
+        com.google.protobuf.AbstractMessageLite.Builder.addAll(
+            values, value_);
+        onChanged();
+        return this;
+      }
+      /**
+       * <code>repeated float value = 1 [packed = true];</code>
+       */
+      public Builder clearValue() {
+        value_ = emptyFloatList();
+        bitField0_ = (bitField0_ & ~0x00000001);
+        onChanged();
+        return this;
+      }
+      @java.lang.Override
+      public final Builder setUnknownFields(
+          final com.google.protobuf.UnknownFieldSet unknownFields) {
+        return super.setUnknownFields(unknownFields);
+      }
+
+      @java.lang.Override
+      public final Builder mergeUnknownFields(
+          final com.google.protobuf.UnknownFieldSet unknownFields) {
+        return super.mergeUnknownFields(unknownFields);
+      }
+
+
+      // @@protoc_insertion_point(builder_scope:tensorflow.CollectionDef.FloatList)
+    }
+
+    // @@protoc_insertion_point(class_scope:tensorflow.CollectionDef.FloatList)
+    private static final org.tensorflow.proto.framework.CollectionDef.FloatList DEFAULT_INSTANCE;
+    static {
+      DEFAULT_INSTANCE = new org.tensorflow.proto.framework.CollectionDef.FloatList();
+    }
+
+    public static org.tensorflow.proto.framework.CollectionDef.FloatList getDefaultInstance() {
+      return DEFAULT_INSTANCE;
+    }
+
+    private static final com.google.protobuf.Parser<FloatList>
+        PARSER = new com.google.protobuf.AbstractParser<FloatList>() {
+      @java.lang.Override
+      public FloatList parsePartialFrom(
+          com.google.protobuf.CodedInputStream input,
+          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+          throws com.google.protobuf.InvalidProtocolBufferException {
+        return new FloatList(input, extensionRegistry);
+      }
+    };
+
+    public static com.google.protobuf.Parser<FloatList> parser() {
+      return PARSER;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Parser<FloatList> getParserForType() {
+      return PARSER;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.CollectionDef.FloatList getDefaultInstanceForType() {
+      return DEFAULT_INSTANCE;
+    }
+
+  }
+
+  public interface AnyListOrBuilder extends
+      // @@protoc_insertion_point(interface_extends:tensorflow.CollectionDef.AnyList)
+      com.google.protobuf.MessageOrBuilder {
+
+    /**
+     * <code>repeated .google.protobuf.Any value = 1;</code>
+     */
+    java.util.List<com.google.protobuf.Any> 
+        getValueList();
+    /**
+     * <code>repeated .google.protobuf.Any value = 1;</code>
+     */
+    com.google.protobuf.Any getValue(int index);
+    /**
+     * <code>repeated .google.protobuf.Any value = 1;</code>
+     */
+    int getValueCount();
+    /**
+     * <code>repeated .google.protobuf.Any value = 1;</code>
+     */
+    java.util.List<? extends com.google.protobuf.AnyOrBuilder> 
+        getValueOrBuilderList();
+    /**
+     * <code>repeated .google.protobuf.Any value = 1;</code>
+     */
+    com.google.protobuf.AnyOrBuilder getValueOrBuilder(
+        int index);
+  }
+  /**
+   * <pre>
+   * AnyList is used for collecting Any protos.
+   * </pre>
+   *
+   * Protobuf type {@code tensorflow.CollectionDef.AnyList}
+   */
+  public  static final class AnyList extends
+      com.google.protobuf.GeneratedMessageV3 implements
+      // @@protoc_insertion_point(message_implements:tensorflow.CollectionDef.AnyList)
+      AnyListOrBuilder {
+  private static final long serialVersionUID = 0L;
+    // Use AnyList.newBuilder() to construct.
+    private AnyList(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+      super(builder);
+    }
+    private AnyList() {
+      value_ = java.util.Collections.emptyList();
+    }
+
+    @java.lang.Override
+    @SuppressWarnings({"unused"})
+    protected java.lang.Object newInstance(
+        UnusedPrivateParameter unused) {
+      return new AnyList();
+    }
+
+    @java.lang.Override
+    public final com.google.protobuf.UnknownFieldSet
+    getUnknownFields() {
+      return this.unknownFields;
+    }
+    private AnyList(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      this();
+      if (extensionRegistry == null) {
+        throw new java.lang.NullPointerException();
+      }
+      int mutable_bitField0_ = 0;
+      com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+          com.google.protobuf.UnknownFieldSet.newBuilder();
+      try {
+        boolean done = false;
+        while (!done) {
+          int tag = input.readTag();
+          switch (tag) {
+            case 0:
+              done = true;
+              break;
+            case 10: {
+              if (!((mutable_bitField0_ & 0x00000001) != 0)) {
+                value_ = new java.util.ArrayList<com.google.protobuf.Any>();
+                mutable_bitField0_ |= 0x00000001;
+              }
+              value_.add(
+                  input.readMessage(com.google.protobuf.Any.parser(), extensionRegistry));
+              break;
+            }
+            default: {
+              if (!parseUnknownField(
+                  input, unknownFields, extensionRegistry, tag)) {
+                done = true;
+              }
+              break;
+            }
+          }
+        }
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        throw e.setUnfinishedMessage(this);
+      } catch (java.io.IOException e) {
+        throw new com.google.protobuf.InvalidProtocolBufferException(
+            e).setUnfinishedMessage(this);
+      } finally {
+        if (((mutable_bitField0_ & 0x00000001) != 0)) {
+          value_ = java.util.Collections.unmodifiableList(value_);
+        }
+        this.unknownFields = unknownFields.build();
+        makeExtensionsImmutable();
+      }
+    }
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.MetaGraphProtos.internal_static_tensorflow_CollectionDef_AnyList_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.MetaGraphProtos.internal_static_tensorflow_CollectionDef_AnyList_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.CollectionDef.AnyList.class, org.tensorflow.proto.framework.CollectionDef.AnyList.Builder.class);
+    }
+
+    public static final int VALUE_FIELD_NUMBER = 1;
+    private java.util.List<com.google.protobuf.Any> value_;
+    /**
+     * <code>repeated .google.protobuf.Any value = 1;</code>
+     */
+    public java.util.List<com.google.protobuf.Any> getValueList() {
+      return value_;
+    }
+    /**
+     * <code>repeated .google.protobuf.Any value = 1;</code>
+     */
+    public java.util.List<? extends com.google.protobuf.AnyOrBuilder> 
+        getValueOrBuilderList() {
+      return value_;
+    }
+    /**
+     * <code>repeated .google.protobuf.Any value = 1;</code>
+     */
+    public int getValueCount() {
+      return value_.size();
+    }
+    /**
+     * <code>repeated .google.protobuf.Any value = 1;</code>
+     */
+    public com.google.protobuf.Any getValue(int index) {
+      return value_.get(index);
+    }
+    /**
+     * <code>repeated .google.protobuf.Any value = 1;</code>
+     */
+    public com.google.protobuf.AnyOrBuilder getValueOrBuilder(
+        int index) {
+      return value_.get(index);
+    }
+
+    private byte memoizedIsInitialized = -1;
+    @java.lang.Override
+    public final boolean isInitialized() {
+      byte isInitialized = memoizedIsInitialized;
+      if (isInitialized == 1) return true;
+      if (isInitialized == 0) return false;
+
+      memoizedIsInitialized = 1;
+      return true;
+    }
+
+    @java.lang.Override
+    public void writeTo(com.google.protobuf.CodedOutputStream output)
+                        throws java.io.IOException {
+      for (int i = 0; i < value_.size(); i++) {
+        output.writeMessage(1, value_.get(i));
+      }
+      unknownFields.writeTo(output);
+    }
+
+    @java.lang.Override
+    public int getSerializedSize() {
+      int size = memoizedSize;
+      if (size != -1) return size;
+
+      size = 0;
+      for (int i = 0; i < value_.size(); i++) {
+        size += com.google.protobuf.CodedOutputStream
+          .computeMessageSize(1, value_.get(i));
+      }
+      size += unknownFields.getSerializedSize();
+      memoizedSize = size;
+      return size;
+    }
+
+    @java.lang.Override
+    public boolean equals(final java.lang.Object obj) {
+      if (obj == this) {
+       return true;
+      }
+      if (!(obj instanceof org.tensorflow.proto.framework.CollectionDef.AnyList)) {
+        return super.equals(obj);
+      }
+      org.tensorflow.proto.framework.CollectionDef.AnyList other = (org.tensorflow.proto.framework.CollectionDef.AnyList) obj;
+
+      if (!getValueList()
+          .equals(other.getValueList())) return false;
+      if (!unknownFields.equals(other.unknownFields)) return false;
+      return true;
+    }
+
+    @java.lang.Override
+    public int hashCode() {
+      if (memoizedHashCode != 0) {
+        return memoizedHashCode;
+      }
+      int hash = 41;
+      hash = (19 * hash) + getDescriptor().hashCode();
+      if (getValueCount() > 0) {
+        hash = (37 * hash) + VALUE_FIELD_NUMBER;
+        hash = (53 * hash) + getValueList().hashCode();
+      }
+      hash = (29 * hash) + unknownFields.hashCode();
+      memoizedHashCode = hash;
+      return hash;
+    }
+
+    public static org.tensorflow.proto.framework.CollectionDef.AnyList parseFrom(
+        java.nio.ByteBuffer data)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data);
+    }
+    public static org.tensorflow.proto.framework.CollectionDef.AnyList parseFrom(
+        java.nio.ByteBuffer data,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.CollectionDef.AnyList parseFrom(
+        com.google.protobuf.ByteString data)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data);
+    }
+    public static org.tensorflow.proto.framework.CollectionDef.AnyList parseFrom(
+        com.google.protobuf.ByteString data,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.CollectionDef.AnyList parseFrom(byte[] data)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data);
+    }
+    public static org.tensorflow.proto.framework.CollectionDef.AnyList parseFrom(
+        byte[] data,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.CollectionDef.AnyList parseFrom(java.io.InputStream input)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input);
+    }
+    public static org.tensorflow.proto.framework.CollectionDef.AnyList parseFrom(
+        java.io.InputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.CollectionDef.AnyList parseDelimitedFrom(java.io.InputStream input)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseDelimitedWithIOException(PARSER, input);
+    }
+    public static org.tensorflow.proto.framework.CollectionDef.AnyList parseDelimitedFrom(
+        java.io.InputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.CollectionDef.AnyList parseFrom(
+        com.google.protobuf.CodedInputStream input)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input);
+    }
+    public static org.tensorflow.proto.framework.CollectionDef.AnyList parseFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input, extensionRegistry);
+    }
+
+    @java.lang.Override
+    public Builder newBuilderForType() { return newBuilder(); }
+    public static Builder newBuilder() {
+      return DEFAULT_INSTANCE.toBuilder();
+    }
+    public static Builder newBuilder(org.tensorflow.proto.framework.CollectionDef.AnyList prototype) {
+      return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+    }
+    @java.lang.Override
+    public Builder toBuilder() {
+      return this == DEFAULT_INSTANCE
+          ? new Builder() : new Builder().mergeFrom(this);
+    }
+
+    @java.lang.Override
+    protected Builder newBuilderForType(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      Builder builder = new Builder(parent);
+      return builder;
+    }
+    /**
+     * <pre>
+     * AnyList is used for collecting Any protos.
+     * </pre>
+     *
+     * Protobuf type {@code tensorflow.CollectionDef.AnyList}
+     */
+    public static final class Builder extends
+        com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+        // @@protoc_insertion_point(builder_implements:tensorflow.CollectionDef.AnyList)
+        org.tensorflow.proto.framework.CollectionDef.AnyListOrBuilder {
+      public static final com.google.protobuf.Descriptors.Descriptor
+          getDescriptor() {
+        return org.tensorflow.proto.framework.MetaGraphProtos.internal_static_tensorflow_CollectionDef_AnyList_descriptor;
+      }
+
+      @java.lang.Override
+      protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+          internalGetFieldAccessorTable() {
+        return org.tensorflow.proto.framework.MetaGraphProtos.internal_static_tensorflow_CollectionDef_AnyList_fieldAccessorTable
+            .ensureFieldAccessorsInitialized(
+                org.tensorflow.proto.framework.CollectionDef.AnyList.class, org.tensorflow.proto.framework.CollectionDef.AnyList.Builder.class);
+      }
+
+      // Construct using org.tensorflow.proto.framework.CollectionDef.AnyList.newBuilder()
+      private Builder() {
+        maybeForceBuilderInitialization();
+      }
+
+      private Builder(
+          com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+        super(parent);
+        maybeForceBuilderInitialization();
+      }
+      private void maybeForceBuilderInitialization() {
+        if (com.google.protobuf.GeneratedMessageV3
+                .alwaysUseFieldBuilders) {
+          getValueFieldBuilder();
+        }
+      }
+      @java.lang.Override
+      public Builder clear() {
+        super.clear();
+        if (valueBuilder_ == null) {
+          value_ = java.util.Collections.emptyList();
+          bitField0_ = (bitField0_ & ~0x00000001);
+        } else {
+          valueBuilder_.clear();
+        }
+        return this;
+      }
+
+      @java.lang.Override
+      public com.google.protobuf.Descriptors.Descriptor
+          getDescriptorForType() {
+        return org.tensorflow.proto.framework.MetaGraphProtos.internal_static_tensorflow_CollectionDef_AnyList_descriptor;
+      }
+
+      @java.lang.Override
+      public org.tensorflow.proto.framework.CollectionDef.AnyList getDefaultInstanceForType() {
+        return org.tensorflow.proto.framework.CollectionDef.AnyList.getDefaultInstance();
+      }
+
+      @java.lang.Override
+      public org.tensorflow.proto.framework.CollectionDef.AnyList build() {
+        org.tensorflow.proto.framework.CollectionDef.AnyList result = buildPartial();
+        if (!result.isInitialized()) {
+          throw newUninitializedMessageException(result);
+        }
+        return result;
+      }
+
+      @java.lang.Override
+      public org.tensorflow.proto.framework.CollectionDef.AnyList buildPartial() {
+        org.tensorflow.proto.framework.CollectionDef.AnyList result = new org.tensorflow.proto.framework.CollectionDef.AnyList(this);
+        int from_bitField0_ = bitField0_;
+        if (valueBuilder_ == null) {
+          if (((bitField0_ & 0x00000001) != 0)) {
+            value_ = java.util.Collections.unmodifiableList(value_);
+            bitField0_ = (bitField0_ & ~0x00000001);
+          }
+          result.value_ = value_;
+        } else {
+          result.value_ = valueBuilder_.build();
+        }
+        onBuilt();
+        return result;
+      }
+
+      @java.lang.Override
+      public Builder clone() {
+        return super.clone();
+      }
+      @java.lang.Override
+      public Builder setField(
+          com.google.protobuf.Descriptors.FieldDescriptor field,
+          java.lang.Object value) {
+        return super.setField(field, value);
+      }
+      @java.lang.Override
+      public Builder clearField(
+          com.google.protobuf.Descriptors.FieldDescriptor field) {
+        return super.clearField(field);
+      }
+      @java.lang.Override
+      public Builder clearOneof(
+          com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+        return super.clearOneof(oneof);
+      }
+      @java.lang.Override
+      public Builder setRepeatedField(
+          com.google.protobuf.Descriptors.FieldDescriptor field,
+          int index, java.lang.Object value) {
+        return super.setRepeatedField(field, index, value);
+      }
+      @java.lang.Override
+      public Builder addRepeatedField(
+          com.google.protobuf.Descriptors.FieldDescriptor field,
+          java.lang.Object value) {
+        return super.addRepeatedField(field, value);
+      }
+      @java.lang.Override
+      public Builder mergeFrom(com.google.protobuf.Message other) {
+        if (other instanceof org.tensorflow.proto.framework.CollectionDef.AnyList) {
+          return mergeFrom((org.tensorflow.proto.framework.CollectionDef.AnyList)other);
+        } else {
+          super.mergeFrom(other);
+          return this;
+        }
+      }
+
+      public Builder mergeFrom(org.tensorflow.proto.framework.CollectionDef.AnyList other) {
+        if (other == org.tensorflow.proto.framework.CollectionDef.AnyList.getDefaultInstance()) return this;
+        if (valueBuilder_ == null) {
+          if (!other.value_.isEmpty()) {
+            if (value_.isEmpty()) {
+              value_ = other.value_;
+              bitField0_ = (bitField0_ & ~0x00000001);
+            } else {
+              ensureValueIsMutable();
+              value_.addAll(other.value_);
+            }
+            onChanged();
+          }
+        } else {
+          if (!other.value_.isEmpty()) {
+            if (valueBuilder_.isEmpty()) {
+              valueBuilder_.dispose();
+              valueBuilder_ = null;
+              value_ = other.value_;
+              bitField0_ = (bitField0_ & ~0x00000001);
+              valueBuilder_ = 
+                com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
+                   getValueFieldBuilder() : null;
+            } else {
+              valueBuilder_.addAllMessages(other.value_);
+            }
+          }
+        }
+        this.mergeUnknownFields(other.unknownFields);
+        onChanged();
+        return this;
+      }
+
+      @java.lang.Override
+      public final boolean isInitialized() {
+        return true;
+      }
+
+      @java.lang.Override
+      public Builder mergeFrom(
+          com.google.protobuf.CodedInputStream input,
+          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+          throws java.io.IOException {
+        org.tensorflow.proto.framework.CollectionDef.AnyList parsedMessage = null;
+        try {
+          parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+        } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+          parsedMessage = (org.tensorflow.proto.framework.CollectionDef.AnyList) e.getUnfinishedMessage();
+          throw e.unwrapIOException();
+        } finally {
+          if (parsedMessage != null) {
+            mergeFrom(parsedMessage);
+          }
+        }
+        return this;
+      }
+      private int bitField0_;
+
+      private java.util.List<com.google.protobuf.Any> value_ =
+        java.util.Collections.emptyList();
+      private void ensureValueIsMutable() {
+        if (!((bitField0_ & 0x00000001) != 0)) {
+          value_ = new java.util.ArrayList<com.google.protobuf.Any>(value_);
+          bitField0_ |= 0x00000001;
+         }
+      }
+
+      private com.google.protobuf.RepeatedFieldBuilderV3<
+          com.google.protobuf.Any, com.google.protobuf.Any.Builder, com.google.protobuf.AnyOrBuilder> valueBuilder_;
+
+      /**
+       * <code>repeated .google.protobuf.Any value = 1;</code>
+       */
+      public java.util.List<com.google.protobuf.Any> getValueList() {
+        if (valueBuilder_ == null) {
+          return java.util.Collections.unmodifiableList(value_);
+        } else {
+          return valueBuilder_.getMessageList();
+        }
+      }
+      /**
+       * <code>repeated .google.protobuf.Any value = 1;</code>
+       */
+      public int getValueCount() {
+        if (valueBuilder_ == null) {
+          return value_.size();
+        } else {
+          return valueBuilder_.getCount();
+        }
+      }
+      /**
+       * <code>repeated .google.protobuf.Any value = 1;</code>
+       */
+      public com.google.protobuf.Any getValue(int index) {
+        if (valueBuilder_ == null) {
+          return value_.get(index);
+        } else {
+          return valueBuilder_.getMessage(index);
+        }
+      }
+      /**
+       * <code>repeated .google.protobuf.Any value = 1;</code>
+       */
+      public Builder setValue(
+          int index, com.google.protobuf.Any value) {
+        if (valueBuilder_ == null) {
+          if (value == null) {
+            throw new NullPointerException();
+          }
+          ensureValueIsMutable();
+          value_.set(index, value);
+          onChanged();
+        } else {
+          valueBuilder_.setMessage(index, value);
+        }
+        return this;
+      }
+      /**
+       * <code>repeated .google.protobuf.Any value = 1;</code>
+       */
+      public Builder setValue(
+          int index, com.google.protobuf.Any.Builder builderForValue) {
+        if (valueBuilder_ == null) {
+          ensureValueIsMutable();
+          value_.set(index, builderForValue.build());
+          onChanged();
+        } else {
+          valueBuilder_.setMessage(index, builderForValue.build());
+        }
+        return this;
+      }
+      /**
+       * <code>repeated .google.protobuf.Any value = 1;</code>
+       */
+      public Builder addValue(com.google.protobuf.Any value) {
+        if (valueBuilder_ == null) {
+          if (value == null) {
+            throw new NullPointerException();
+          }
+          ensureValueIsMutable();
+          value_.add(value);
+          onChanged();
+        } else {
+          valueBuilder_.addMessage(value);
+        }
+        return this;
+      }
+      /**
+       * <code>repeated .google.protobuf.Any value = 1;</code>
+       */
+      public Builder addValue(
+          int index, com.google.protobuf.Any value) {
+        if (valueBuilder_ == null) {
+          if (value == null) {
+            throw new NullPointerException();
+          }
+          ensureValueIsMutable();
+          value_.add(index, value);
+          onChanged();
+        } else {
+          valueBuilder_.addMessage(index, value);
+        }
+        return this;
+      }
+      /**
+       * <code>repeated .google.protobuf.Any value = 1;</code>
+       */
+      public Builder addValue(
+          com.google.protobuf.Any.Builder builderForValue) {
+        if (valueBuilder_ == null) {
+          ensureValueIsMutable();
+          value_.add(builderForValue.build());
+          onChanged();
+        } else {
+          valueBuilder_.addMessage(builderForValue.build());
+        }
+        return this;
+      }
+      /**
+       * <code>repeated .google.protobuf.Any value = 1;</code>
+       */
+      public Builder addValue(
+          int index, com.google.protobuf.Any.Builder builderForValue) {
+        if (valueBuilder_ == null) {
+          ensureValueIsMutable();
+          value_.add(index, builderForValue.build());
+          onChanged();
+        } else {
+          valueBuilder_.addMessage(index, builderForValue.build());
+        }
+        return this;
+      }
+      /**
+       * <code>repeated .google.protobuf.Any value = 1;</code>
+       */
+      public Builder addAllValue(
+          java.lang.Iterable<? extends com.google.protobuf.Any> values) {
+        if (valueBuilder_ == null) {
+          ensureValueIsMutable();
+          com.google.protobuf.AbstractMessageLite.Builder.addAll(
+              values, value_);
+          onChanged();
+        } else {
+          valueBuilder_.addAllMessages(values);
+        }
+        return this;
+      }
+      /**
+       * <code>repeated .google.protobuf.Any value = 1;</code>
+       */
+      public Builder clearValue() {
+        if (valueBuilder_ == null) {
+          value_ = java.util.Collections.emptyList();
+          bitField0_ = (bitField0_ & ~0x00000001);
+          onChanged();
+        } else {
+          valueBuilder_.clear();
+        }
+        return this;
+      }
+      /**
+       * <code>repeated .google.protobuf.Any value = 1;</code>
+       */
+      public Builder removeValue(int index) {
+        if (valueBuilder_ == null) {
+          ensureValueIsMutable();
+          value_.remove(index);
+          onChanged();
+        } else {
+          valueBuilder_.remove(index);
+        }
+        return this;
+      }
+      /**
+       * <code>repeated .google.protobuf.Any value = 1;</code>
+       */
+      public com.google.protobuf.Any.Builder getValueBuilder(
+          int index) {
+        return getValueFieldBuilder().getBuilder(index);
+      }
+      /**
+       * <code>repeated .google.protobuf.Any value = 1;</code>
+       */
+      public com.google.protobuf.AnyOrBuilder getValueOrBuilder(
+          int index) {
+        if (valueBuilder_ == null) {
+          return value_.get(index);  } else {
+          return valueBuilder_.getMessageOrBuilder(index);
+        }
+      }
+      /**
+       * <code>repeated .google.protobuf.Any value = 1;</code>
+       */
+      public java.util.List<? extends com.google.protobuf.AnyOrBuilder> 
+           getValueOrBuilderList() {
+        if (valueBuilder_ != null) {
+          return valueBuilder_.getMessageOrBuilderList();
+        } else {
+          return java.util.Collections.unmodifiableList(value_);
+        }
+      }
+      /**
+       * <code>repeated .google.protobuf.Any value = 1;</code>
+       */
+      public com.google.protobuf.Any.Builder addValueBuilder() {
+        return getValueFieldBuilder().addBuilder(
+            com.google.protobuf.Any.getDefaultInstance());
+      }
+      /**
+       * <code>repeated .google.protobuf.Any value = 1;</code>
+       */
+      public com.google.protobuf.Any.Builder addValueBuilder(
+          int index) {
+        return getValueFieldBuilder().addBuilder(
+            index, com.google.protobuf.Any.getDefaultInstance());
+      }
+      /**
+       * <code>repeated .google.protobuf.Any value = 1;</code>
+       */
+      public java.util.List<com.google.protobuf.Any.Builder> 
+           getValueBuilderList() {
+        return getValueFieldBuilder().getBuilderList();
+      }
+      private com.google.protobuf.RepeatedFieldBuilderV3<
+          com.google.protobuf.Any, com.google.protobuf.Any.Builder, com.google.protobuf.AnyOrBuilder> 
+          getValueFieldBuilder() {
+        if (valueBuilder_ == null) {
+          valueBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
+              com.google.protobuf.Any, com.google.protobuf.Any.Builder, com.google.protobuf.AnyOrBuilder>(
+                  value_,
+                  ((bitField0_ & 0x00000001) != 0),
+                  getParentForChildren(),
+                  isClean());
+          value_ = null;
+        }
+        return valueBuilder_;
+      }
+      @java.lang.Override
+      public final Builder setUnknownFields(
+          final com.google.protobuf.UnknownFieldSet unknownFields) {
+        return super.setUnknownFields(unknownFields);
+      }
+
+      @java.lang.Override
+      public final Builder mergeUnknownFields(
+          final com.google.protobuf.UnknownFieldSet unknownFields) {
+        return super.mergeUnknownFields(unknownFields);
+      }
+
+
+      // @@protoc_insertion_point(builder_scope:tensorflow.CollectionDef.AnyList)
+    }
+
+    // @@protoc_insertion_point(class_scope:tensorflow.CollectionDef.AnyList)
+    private static final org.tensorflow.proto.framework.CollectionDef.AnyList DEFAULT_INSTANCE;
+    static {
+      DEFAULT_INSTANCE = new org.tensorflow.proto.framework.CollectionDef.AnyList();
+    }
+
+    public static org.tensorflow.proto.framework.CollectionDef.AnyList getDefaultInstance() {
+      return DEFAULT_INSTANCE;
+    }
+
+    private static final com.google.protobuf.Parser<AnyList>
+        PARSER = new com.google.protobuf.AbstractParser<AnyList>() {
+      @java.lang.Override
+      public AnyList parsePartialFrom(
+          com.google.protobuf.CodedInputStream input,
+          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+          throws com.google.protobuf.InvalidProtocolBufferException {
+        return new AnyList(input, extensionRegistry);
+      }
+    };
+
+    public static com.google.protobuf.Parser<AnyList> parser() {
+      return PARSER;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Parser<AnyList> getParserForType() {
+      return PARSER;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.CollectionDef.AnyList getDefaultInstanceForType() {
+      return DEFAULT_INSTANCE;
+    }
+
+  }
+
+  private int kindCase_ = 0;
+  private java.lang.Object kind_;
+  public enum KindCase
+      implements com.google.protobuf.Internal.EnumLite {
+    NODE_LIST(1),
+    BYTES_LIST(2),
+    INT64_LIST(3),
+    FLOAT_LIST(4),
+    ANY_LIST(5),
+    KIND_NOT_SET(0);
+    private final int value;
+    private KindCase(int value) {
+      this.value = value;
+    }
+    /**
+     * @deprecated Use {@link #forNumber(int)} instead.
+     */
+    @java.lang.Deprecated
+    public static KindCase valueOf(int value) {
+      return forNumber(value);
+    }
+
+    public static KindCase forNumber(int value) {
+      switch (value) {
+        case 1: return NODE_LIST;
+        case 2: return BYTES_LIST;
+        case 3: return INT64_LIST;
+        case 4: return FLOAT_LIST;
+        case 5: return ANY_LIST;
+        case 0: return KIND_NOT_SET;
+        default: return null;
+      }
+    }
+    public int getNumber() {
+      return this.value;
+    }
+  };
+
+  public KindCase
+  getKindCase() {
+    return KindCase.forNumber(
+        kindCase_);
+  }
+
+  public static final int NODE_LIST_FIELD_NUMBER = 1;
+  /**
+   * <code>.tensorflow.CollectionDef.NodeList node_list = 1;</code>
+   */
+  public boolean hasNodeList() {
+    return kindCase_ == 1;
+  }
+  /**
+   * <code>.tensorflow.CollectionDef.NodeList node_list = 1;</code>
+   */
+  public org.tensorflow.proto.framework.CollectionDef.NodeList getNodeList() {
+    if (kindCase_ == 1) {
+       return (org.tensorflow.proto.framework.CollectionDef.NodeList) kind_;
+    }
+    return org.tensorflow.proto.framework.CollectionDef.NodeList.getDefaultInstance();
+  }
+  /**
+   * <code>.tensorflow.CollectionDef.NodeList node_list = 1;</code>
+   */
+  public org.tensorflow.proto.framework.CollectionDef.NodeListOrBuilder getNodeListOrBuilder() {
+    if (kindCase_ == 1) {
+       return (org.tensorflow.proto.framework.CollectionDef.NodeList) kind_;
+    }
+    return org.tensorflow.proto.framework.CollectionDef.NodeList.getDefaultInstance();
+  }
+
+  public static final int BYTES_LIST_FIELD_NUMBER = 2;
+  /**
+   * <code>.tensorflow.CollectionDef.BytesList bytes_list = 2;</code>
+   */
+  public boolean hasBytesList() {
+    return kindCase_ == 2;
+  }
+  /**
+   * <code>.tensorflow.CollectionDef.BytesList bytes_list = 2;</code>
+   */
+  public org.tensorflow.proto.framework.CollectionDef.BytesList getBytesList() {
+    if (kindCase_ == 2) {
+       return (org.tensorflow.proto.framework.CollectionDef.BytesList) kind_;
+    }
+    return org.tensorflow.proto.framework.CollectionDef.BytesList.getDefaultInstance();
+  }
+  /**
+   * <code>.tensorflow.CollectionDef.BytesList bytes_list = 2;</code>
+   */
+  public org.tensorflow.proto.framework.CollectionDef.BytesListOrBuilder getBytesListOrBuilder() {
+    if (kindCase_ == 2) {
+       return (org.tensorflow.proto.framework.CollectionDef.BytesList) kind_;
+    }
+    return org.tensorflow.proto.framework.CollectionDef.BytesList.getDefaultInstance();
+  }
+
+  public static final int INT64_LIST_FIELD_NUMBER = 3;
+  /**
+   * <code>.tensorflow.CollectionDef.Int64List int64_list = 3;</code>
+   */
+  public boolean hasInt64List() {
+    return kindCase_ == 3;
+  }
+  /**
+   * <code>.tensorflow.CollectionDef.Int64List int64_list = 3;</code>
+   */
+  public org.tensorflow.proto.framework.CollectionDef.Int64List getInt64List() {
+    if (kindCase_ == 3) {
+       return (org.tensorflow.proto.framework.CollectionDef.Int64List) kind_;
+    }
+    return org.tensorflow.proto.framework.CollectionDef.Int64List.getDefaultInstance();
+  }
+  /**
+   * <code>.tensorflow.CollectionDef.Int64List int64_list = 3;</code>
+   */
+  public org.tensorflow.proto.framework.CollectionDef.Int64ListOrBuilder getInt64ListOrBuilder() {
+    if (kindCase_ == 3) {
+       return (org.tensorflow.proto.framework.CollectionDef.Int64List) kind_;
+    }
+    return org.tensorflow.proto.framework.CollectionDef.Int64List.getDefaultInstance();
+  }
+
+  public static final int FLOAT_LIST_FIELD_NUMBER = 4;
+  /**
+   * <code>.tensorflow.CollectionDef.FloatList float_list = 4;</code>
+   */
+  public boolean hasFloatList() {
+    return kindCase_ == 4;
+  }
+  /**
+   * <code>.tensorflow.CollectionDef.FloatList float_list = 4;</code>
+   */
+  public org.tensorflow.proto.framework.CollectionDef.FloatList getFloatList() {
+    if (kindCase_ == 4) {
+       return (org.tensorflow.proto.framework.CollectionDef.FloatList) kind_;
+    }
+    return org.tensorflow.proto.framework.CollectionDef.FloatList.getDefaultInstance();
+  }
+  /**
+   * <code>.tensorflow.CollectionDef.FloatList float_list = 4;</code>
+   */
+  public org.tensorflow.proto.framework.CollectionDef.FloatListOrBuilder getFloatListOrBuilder() {
+    if (kindCase_ == 4) {
+       return (org.tensorflow.proto.framework.CollectionDef.FloatList) kind_;
+    }
+    return org.tensorflow.proto.framework.CollectionDef.FloatList.getDefaultInstance();
+  }
+
+  public static final int ANY_LIST_FIELD_NUMBER = 5;
+  /**
+   * <code>.tensorflow.CollectionDef.AnyList any_list = 5;</code>
+   */
+  public boolean hasAnyList() {
+    return kindCase_ == 5;
+  }
+  /**
+   * <code>.tensorflow.CollectionDef.AnyList any_list = 5;</code>
+   */
+  public org.tensorflow.proto.framework.CollectionDef.AnyList getAnyList() {
+    if (kindCase_ == 5) {
+       return (org.tensorflow.proto.framework.CollectionDef.AnyList) kind_;
+    }
+    return org.tensorflow.proto.framework.CollectionDef.AnyList.getDefaultInstance();
+  }
+  /**
+   * <code>.tensorflow.CollectionDef.AnyList any_list = 5;</code>
+   */
+  public org.tensorflow.proto.framework.CollectionDef.AnyListOrBuilder getAnyListOrBuilder() {
+    if (kindCase_ == 5) {
+       return (org.tensorflow.proto.framework.CollectionDef.AnyList) kind_;
+    }
+    return org.tensorflow.proto.framework.CollectionDef.AnyList.getDefaultInstance();
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    if (kindCase_ == 1) {
+      output.writeMessage(1, (org.tensorflow.proto.framework.CollectionDef.NodeList) kind_);
+    }
+    if (kindCase_ == 2) {
+      output.writeMessage(2, (org.tensorflow.proto.framework.CollectionDef.BytesList) kind_);
+    }
+    if (kindCase_ == 3) {
+      output.writeMessage(3, (org.tensorflow.proto.framework.CollectionDef.Int64List) kind_);
+    }
+    if (kindCase_ == 4) {
+      output.writeMessage(4, (org.tensorflow.proto.framework.CollectionDef.FloatList) kind_);
+    }
+    if (kindCase_ == 5) {
+      output.writeMessage(5, (org.tensorflow.proto.framework.CollectionDef.AnyList) kind_);
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (kindCase_ == 1) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(1, (org.tensorflow.proto.framework.CollectionDef.NodeList) kind_);
+    }
+    if (kindCase_ == 2) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(2, (org.tensorflow.proto.framework.CollectionDef.BytesList) kind_);
+    }
+    if (kindCase_ == 3) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(3, (org.tensorflow.proto.framework.CollectionDef.Int64List) kind_);
+    }
+    if (kindCase_ == 4) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(4, (org.tensorflow.proto.framework.CollectionDef.FloatList) kind_);
+    }
+    if (kindCase_ == 5) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(5, (org.tensorflow.proto.framework.CollectionDef.AnyList) kind_);
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.framework.CollectionDef)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.framework.CollectionDef other = (org.tensorflow.proto.framework.CollectionDef) obj;
+
+    if (!getKindCase().equals(other.getKindCase())) return false;
+    switch (kindCase_) {
+      case 1:
+        if (!getNodeList()
+            .equals(other.getNodeList())) return false;
+        break;
+      case 2:
+        if (!getBytesList()
+            .equals(other.getBytesList())) return false;
+        break;
+      case 3:
+        if (!getInt64List()
+            .equals(other.getInt64List())) return false;
+        break;
+      case 4:
+        if (!getFloatList()
+            .equals(other.getFloatList())) return false;
+        break;
+      case 5:
+        if (!getAnyList()
+            .equals(other.getAnyList())) return false;
+        break;
+      case 0:
+      default:
+    }
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    switch (kindCase_) {
+      case 1:
+        hash = (37 * hash) + NODE_LIST_FIELD_NUMBER;
+        hash = (53 * hash) + getNodeList().hashCode();
+        break;
+      case 2:
+        hash = (37 * hash) + BYTES_LIST_FIELD_NUMBER;
+        hash = (53 * hash) + getBytesList().hashCode();
+        break;
+      case 3:
+        hash = (37 * hash) + INT64_LIST_FIELD_NUMBER;
+        hash = (53 * hash) + getInt64List().hashCode();
+        break;
+      case 4:
+        hash = (37 * hash) + FLOAT_LIST_FIELD_NUMBER;
+        hash = (53 * hash) + getFloatList().hashCode();
+        break;
+      case 5:
+        hash = (37 * hash) + ANY_LIST_FIELD_NUMBER;
+        hash = (53 * hash) + getAnyList().hashCode();
+        break;
+      case 0:
+      default:
+    }
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.framework.CollectionDef parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.CollectionDef parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.CollectionDef parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.CollectionDef parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.CollectionDef parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.CollectionDef parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.CollectionDef parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.CollectionDef parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.CollectionDef parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.CollectionDef parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.CollectionDef parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.CollectionDef parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.framework.CollectionDef prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * <pre>
+   * CollectionDef should cover most collections.
+   * To add a user-defined collection, do one of the following:
+   * 1. For simple data types, such as string, int, float:
+   *      tf.add_to_collection("your_collection_name", your_simple_value)
+   *    strings will be stored as bytes_list.
+   * 2. For Protobuf types, there are three ways to add them:
+   *    1) tf.add_to_collection("your_collection_name",
+   *         your_proto.SerializeToString())
+   *       collection_def {
+   *         key: "user_defined_bytes_collection"
+   *         value {
+   *           bytes_list {
+   *             value: "queue_name: &#92;"test_queue&#92;"&#92;n"
+   *           }
+   *         }
+   *       }
+   *  or
+   *    2) tf.add_to_collection("your_collection_name", str(your_proto))
+   *       collection_def {
+   *         key: "user_defined_string_collection"
+   *         value {
+   *          bytes_list {
+   *             value: "&#92;n&#92;ntest_queue"
+   *           }
+   *         }
+   *       }
+   *  or
+   *    3) any_buf = any_pb2.Any()
+   *       tf.add_to_collection("your_collection_name",
+   *         any_buf.Pack(your_proto))
+   *       collection_def {
+   *         key: "user_defined_any_collection"
+   *         value {
+   *           any_list {
+   *             value {
+   *               type_url: "type.googleapis.com/tensorflow.QueueRunnerDef"
+   *               value: "&#92;n&#92;ntest_queue"
+   *             }
+   *           }
+   *         }
+   *       }
+   * 3. For Python objects, implement to_proto() and from_proto(), and register
+   *    them in the following manner:
+   *    ops.register_proto_function("your_collection_name",
+   *                                proto_type,
+   *                                to_proto=YourPythonObject.to_proto,
+   *                                from_proto=YourPythonObject.from_proto)
+   *    These functions will be invoked to serialize and de-serialize the
+   *    collection. For example,
+   *    ops.register_proto_function(ops.GraphKeys.GLOBAL_VARIABLES,
+   *                                proto_type=variable_pb2.VariableDef,
+   *                                to_proto=Variable.to_proto,
+   *                                from_proto=Variable.from_proto)
+   * </pre>
+   *
+   * Protobuf type {@code tensorflow.CollectionDef}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.CollectionDef)
+      org.tensorflow.proto.framework.CollectionDefOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.MetaGraphProtos.internal_static_tensorflow_CollectionDef_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.MetaGraphProtos.internal_static_tensorflow_CollectionDef_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.CollectionDef.class, org.tensorflow.proto.framework.CollectionDef.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.framework.CollectionDef.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      kindCase_ = 0;
+      kind_ = null;
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.framework.MetaGraphProtos.internal_static_tensorflow_CollectionDef_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.CollectionDef getDefaultInstanceForType() {
+      return org.tensorflow.proto.framework.CollectionDef.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.CollectionDef build() {
+      org.tensorflow.proto.framework.CollectionDef result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.CollectionDef buildPartial() {
+      org.tensorflow.proto.framework.CollectionDef result = new org.tensorflow.proto.framework.CollectionDef(this);
+      if (kindCase_ == 1) {
+        if (nodeListBuilder_ == null) {
+          result.kind_ = kind_;
+        } else {
+          result.kind_ = nodeListBuilder_.build();
+        }
+      }
+      if (kindCase_ == 2) {
+        if (bytesListBuilder_ == null) {
+          result.kind_ = kind_;
+        } else {
+          result.kind_ = bytesListBuilder_.build();
+        }
+      }
+      if (kindCase_ == 3) {
+        if (int64ListBuilder_ == null) {
+          result.kind_ = kind_;
+        } else {
+          result.kind_ = int64ListBuilder_.build();
+        }
+      }
+      if (kindCase_ == 4) {
+        if (floatListBuilder_ == null) {
+          result.kind_ = kind_;
+        } else {
+          result.kind_ = floatListBuilder_.build();
+        }
+      }
+      if (kindCase_ == 5) {
+        if (anyListBuilder_ == null) {
+          result.kind_ = kind_;
+        } else {
+          result.kind_ = anyListBuilder_.build();
+        }
+      }
+      result.kindCase_ = kindCase_;
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.framework.CollectionDef) {
+        return mergeFrom((org.tensorflow.proto.framework.CollectionDef)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.framework.CollectionDef other) {
+      if (other == org.tensorflow.proto.framework.CollectionDef.getDefaultInstance()) return this;
+      switch (other.getKindCase()) {
+        case NODE_LIST: {
+          mergeNodeList(other.getNodeList());
+          break;
+        }
+        case BYTES_LIST: {
+          mergeBytesList(other.getBytesList());
+          break;
+        }
+        case INT64_LIST: {
+          mergeInt64List(other.getInt64List());
+          break;
+        }
+        case FLOAT_LIST: {
+          mergeFloatList(other.getFloatList());
+          break;
+        }
+        case ANY_LIST: {
+          mergeAnyList(other.getAnyList());
+          break;
+        }
+        case KIND_NOT_SET: {
+          break;
+        }
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.framework.CollectionDef parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.framework.CollectionDef) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+    private int kindCase_ = 0;
+    private java.lang.Object kind_;
+    public KindCase
+        getKindCase() {
+      return KindCase.forNumber(
+          kindCase_);
+    }
+
+    public Builder clearKind() {
+      kindCase_ = 0;
+      kind_ = null;
+      onChanged();
+      return this;
+    }
+
+
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.CollectionDef.NodeList, org.tensorflow.proto.framework.CollectionDef.NodeList.Builder, org.tensorflow.proto.framework.CollectionDef.NodeListOrBuilder> nodeListBuilder_;
+    /**
+     * <code>.tensorflow.CollectionDef.NodeList node_list = 1;</code>
+     */
+    public boolean hasNodeList() {
+      return kindCase_ == 1;
+    }
+    /**
+     * <code>.tensorflow.CollectionDef.NodeList node_list = 1;</code>
+     */
+    public org.tensorflow.proto.framework.CollectionDef.NodeList getNodeList() {
+      if (nodeListBuilder_ == null) {
+        if (kindCase_ == 1) {
+          return (org.tensorflow.proto.framework.CollectionDef.NodeList) kind_;
+        }
+        return org.tensorflow.proto.framework.CollectionDef.NodeList.getDefaultInstance();
+      } else {
+        if (kindCase_ == 1) {
+          return nodeListBuilder_.getMessage();
+        }
+        return org.tensorflow.proto.framework.CollectionDef.NodeList.getDefaultInstance();
+      }
+    }
+    /**
+     * <code>.tensorflow.CollectionDef.NodeList node_list = 1;</code>
+     */
+    public Builder setNodeList(org.tensorflow.proto.framework.CollectionDef.NodeList value) {
+      if (nodeListBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        kind_ = value;
+        onChanged();
+      } else {
+        nodeListBuilder_.setMessage(value);
+      }
+      kindCase_ = 1;
+      return this;
+    }
+    /**
+     * <code>.tensorflow.CollectionDef.NodeList node_list = 1;</code>
+     */
+    public Builder setNodeList(
+        org.tensorflow.proto.framework.CollectionDef.NodeList.Builder builderForValue) {
+      if (nodeListBuilder_ == null) {
+        kind_ = builderForValue.build();
+        onChanged();
+      } else {
+        nodeListBuilder_.setMessage(builderForValue.build());
+      }
+      kindCase_ = 1;
+      return this;
+    }
+    /**
+     * <code>.tensorflow.CollectionDef.NodeList node_list = 1;</code>
+     */
+    public Builder mergeNodeList(org.tensorflow.proto.framework.CollectionDef.NodeList value) {
+      if (nodeListBuilder_ == null) {
+        if (kindCase_ == 1 &&
+            kind_ != org.tensorflow.proto.framework.CollectionDef.NodeList.getDefaultInstance()) {
+          kind_ = org.tensorflow.proto.framework.CollectionDef.NodeList.newBuilder((org.tensorflow.proto.framework.CollectionDef.NodeList) kind_)
+              .mergeFrom(value).buildPartial();
+        } else {
+          kind_ = value;
+        }
+        onChanged();
+      } else {
+        if (kindCase_ == 1) {
+          nodeListBuilder_.mergeFrom(value);
+        }
+        nodeListBuilder_.setMessage(value);
+      }
+      kindCase_ = 1;
+      return this;
+    }
+    /**
+     * <code>.tensorflow.CollectionDef.NodeList node_list = 1;</code>
+     */
+    public Builder clearNodeList() {
+      if (nodeListBuilder_ == null) {
+        if (kindCase_ == 1) {
+          kindCase_ = 0;
+          kind_ = null;
+          onChanged();
+        }
+      } else {
+        if (kindCase_ == 1) {
+          kindCase_ = 0;
+          kind_ = null;
+        }
+        nodeListBuilder_.clear();
+      }
+      return this;
+    }
+    /**
+     * <code>.tensorflow.CollectionDef.NodeList node_list = 1;</code>
+     */
+    public org.tensorflow.proto.framework.CollectionDef.NodeList.Builder getNodeListBuilder() {
+      return getNodeListFieldBuilder().getBuilder();
+    }
+    /**
+     * <code>.tensorflow.CollectionDef.NodeList node_list = 1;</code>
+     */
+    public org.tensorflow.proto.framework.CollectionDef.NodeListOrBuilder getNodeListOrBuilder() {
+      if ((kindCase_ == 1) && (nodeListBuilder_ != null)) {
+        return nodeListBuilder_.getMessageOrBuilder();
+      } else {
+        if (kindCase_ == 1) {
+          return (org.tensorflow.proto.framework.CollectionDef.NodeList) kind_;
+        }
+        return org.tensorflow.proto.framework.CollectionDef.NodeList.getDefaultInstance();
+      }
+    }
+    /**
+     * <code>.tensorflow.CollectionDef.NodeList node_list = 1;</code>
+     */
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.CollectionDef.NodeList, org.tensorflow.proto.framework.CollectionDef.NodeList.Builder, org.tensorflow.proto.framework.CollectionDef.NodeListOrBuilder> 
+        getNodeListFieldBuilder() {
+      if (nodeListBuilder_ == null) {
+        if (!(kindCase_ == 1)) {
+          kind_ = org.tensorflow.proto.framework.CollectionDef.NodeList.getDefaultInstance();
+        }
+        nodeListBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+            org.tensorflow.proto.framework.CollectionDef.NodeList, org.tensorflow.proto.framework.CollectionDef.NodeList.Builder, org.tensorflow.proto.framework.CollectionDef.NodeListOrBuilder>(
+                (org.tensorflow.proto.framework.CollectionDef.NodeList) kind_,
+                getParentForChildren(),
+                isClean());
+        kind_ = null;
+      }
+      kindCase_ = 1;
+      onChanged();;
+      return nodeListBuilder_;
+    }
+
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.CollectionDef.BytesList, org.tensorflow.proto.framework.CollectionDef.BytesList.Builder, org.tensorflow.proto.framework.CollectionDef.BytesListOrBuilder> bytesListBuilder_;
+    /**
+     * <code>.tensorflow.CollectionDef.BytesList bytes_list = 2;</code>
+     */
+    public boolean hasBytesList() {
+      return kindCase_ == 2;
+    }
+    /**
+     * <code>.tensorflow.CollectionDef.BytesList bytes_list = 2;</code>
+     */
+    public org.tensorflow.proto.framework.CollectionDef.BytesList getBytesList() {
+      if (bytesListBuilder_ == null) {
+        if (kindCase_ == 2) {
+          return (org.tensorflow.proto.framework.CollectionDef.BytesList) kind_;
+        }
+        return org.tensorflow.proto.framework.CollectionDef.BytesList.getDefaultInstance();
+      } else {
+        if (kindCase_ == 2) {
+          return bytesListBuilder_.getMessage();
+        }
+        return org.tensorflow.proto.framework.CollectionDef.BytesList.getDefaultInstance();
+      }
+    }
+    /**
+     * <code>.tensorflow.CollectionDef.BytesList bytes_list = 2;</code>
+     */
+    public Builder setBytesList(org.tensorflow.proto.framework.CollectionDef.BytesList value) {
+      if (bytesListBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        kind_ = value;
+        onChanged();
+      } else {
+        bytesListBuilder_.setMessage(value);
+      }
+      kindCase_ = 2;
+      return this;
+    }
+    /**
+     * <code>.tensorflow.CollectionDef.BytesList bytes_list = 2;</code>
+     */
+    public Builder setBytesList(
+        org.tensorflow.proto.framework.CollectionDef.BytesList.Builder builderForValue) {
+      if (bytesListBuilder_ == null) {
+        kind_ = builderForValue.build();
+        onChanged();
+      } else {
+        bytesListBuilder_.setMessage(builderForValue.build());
+      }
+      kindCase_ = 2;
+      return this;
+    }
+    /**
+     * <code>.tensorflow.CollectionDef.BytesList bytes_list = 2;</code>
+     */
+    public Builder mergeBytesList(org.tensorflow.proto.framework.CollectionDef.BytesList value) {
+      if (bytesListBuilder_ == null) {
+        if (kindCase_ == 2 &&
+            kind_ != org.tensorflow.proto.framework.CollectionDef.BytesList.getDefaultInstance()) {
+          kind_ = org.tensorflow.proto.framework.CollectionDef.BytesList.newBuilder((org.tensorflow.proto.framework.CollectionDef.BytesList) kind_)
+              .mergeFrom(value).buildPartial();
+        } else {
+          kind_ = value;
+        }
+        onChanged();
+      } else {
+        if (kindCase_ == 2) {
+          bytesListBuilder_.mergeFrom(value);
+        }
+        bytesListBuilder_.setMessage(value);
+      }
+      kindCase_ = 2;
+      return this;
+    }
+    /**
+     * <code>.tensorflow.CollectionDef.BytesList bytes_list = 2;</code>
+     */
+    public Builder clearBytesList() {
+      if (bytesListBuilder_ == null) {
+        if (kindCase_ == 2) {
+          kindCase_ = 0;
+          kind_ = null;
+          onChanged();
+        }
+      } else {
+        if (kindCase_ == 2) {
+          kindCase_ = 0;
+          kind_ = null;
+        }
+        bytesListBuilder_.clear();
+      }
+      return this;
+    }
+    /**
+     * <code>.tensorflow.CollectionDef.BytesList bytes_list = 2;</code>
+     */
+    public org.tensorflow.proto.framework.CollectionDef.BytesList.Builder getBytesListBuilder() {
+      return getBytesListFieldBuilder().getBuilder();
+    }
+    /**
+     * <code>.tensorflow.CollectionDef.BytesList bytes_list = 2;</code>
+     */
+    public org.tensorflow.proto.framework.CollectionDef.BytesListOrBuilder getBytesListOrBuilder() {
+      if ((kindCase_ == 2) && (bytesListBuilder_ != null)) {
+        return bytesListBuilder_.getMessageOrBuilder();
+      } else {
+        if (kindCase_ == 2) {
+          return (org.tensorflow.proto.framework.CollectionDef.BytesList) kind_;
+        }
+        return org.tensorflow.proto.framework.CollectionDef.BytesList.getDefaultInstance();
+      }
+    }
+    /**
+     * <code>.tensorflow.CollectionDef.BytesList bytes_list = 2;</code>
+     */
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.CollectionDef.BytesList, org.tensorflow.proto.framework.CollectionDef.BytesList.Builder, org.tensorflow.proto.framework.CollectionDef.BytesListOrBuilder> 
+        getBytesListFieldBuilder() {
+      if (bytesListBuilder_ == null) {
+        if (!(kindCase_ == 2)) {
+          kind_ = org.tensorflow.proto.framework.CollectionDef.BytesList.getDefaultInstance();
+        }
+        bytesListBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+            org.tensorflow.proto.framework.CollectionDef.BytesList, org.tensorflow.proto.framework.CollectionDef.BytesList.Builder, org.tensorflow.proto.framework.CollectionDef.BytesListOrBuilder>(
+                (org.tensorflow.proto.framework.CollectionDef.BytesList) kind_,
+                getParentForChildren(),
+                isClean());
+        kind_ = null;
+      }
+      kindCase_ = 2;
+      onChanged();;
+      return bytesListBuilder_;
+    }
+
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.CollectionDef.Int64List, org.tensorflow.proto.framework.CollectionDef.Int64List.Builder, org.tensorflow.proto.framework.CollectionDef.Int64ListOrBuilder> int64ListBuilder_;
+    /**
+     * <code>.tensorflow.CollectionDef.Int64List int64_list = 3;</code>
+     */
+    public boolean hasInt64List() {
+      return kindCase_ == 3;
+    }
+    /**
+     * <code>.tensorflow.CollectionDef.Int64List int64_list = 3;</code>
+     */
+    public org.tensorflow.proto.framework.CollectionDef.Int64List getInt64List() {
+      if (int64ListBuilder_ == null) {
+        if (kindCase_ == 3) {
+          return (org.tensorflow.proto.framework.CollectionDef.Int64List) kind_;
+        }
+        return org.tensorflow.proto.framework.CollectionDef.Int64List.getDefaultInstance();
+      } else {
+        if (kindCase_ == 3) {
+          return int64ListBuilder_.getMessage();
+        }
+        return org.tensorflow.proto.framework.CollectionDef.Int64List.getDefaultInstance();
+      }
+    }
+    /**
+     * <code>.tensorflow.CollectionDef.Int64List int64_list = 3;</code>
+     */
+    public Builder setInt64List(org.tensorflow.proto.framework.CollectionDef.Int64List value) {
+      if (int64ListBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        kind_ = value;
+        onChanged();
+      } else {
+        int64ListBuilder_.setMessage(value);
+      }
+      kindCase_ = 3;
+      return this;
+    }
+    /**
+     * <code>.tensorflow.CollectionDef.Int64List int64_list = 3;</code>
+     */
+    public Builder setInt64List(
+        org.tensorflow.proto.framework.CollectionDef.Int64List.Builder builderForValue) {
+      if (int64ListBuilder_ == null) {
+        kind_ = builderForValue.build();
+        onChanged();
+      } else {
+        int64ListBuilder_.setMessage(builderForValue.build());
+      }
+      kindCase_ = 3;
+      return this;
+    }
+    /**
+     * <code>.tensorflow.CollectionDef.Int64List int64_list = 3;</code>
+     */
+    public Builder mergeInt64List(org.tensorflow.proto.framework.CollectionDef.Int64List value) {
+      if (int64ListBuilder_ == null) {
+        if (kindCase_ == 3 &&
+            kind_ != org.tensorflow.proto.framework.CollectionDef.Int64List.getDefaultInstance()) {
+          kind_ = org.tensorflow.proto.framework.CollectionDef.Int64List.newBuilder((org.tensorflow.proto.framework.CollectionDef.Int64List) kind_)
+              .mergeFrom(value).buildPartial();
+        } else {
+          kind_ = value;
+        }
+        onChanged();
+      } else {
+        if (kindCase_ == 3) {
+          int64ListBuilder_.mergeFrom(value);
+        }
+        int64ListBuilder_.setMessage(value);
+      }
+      kindCase_ = 3;
+      return this;
+    }
+    /**
+     * <code>.tensorflow.CollectionDef.Int64List int64_list = 3;</code>
+     */
+    public Builder clearInt64List() {
+      if (int64ListBuilder_ == null) {
+        if (kindCase_ == 3) {
+          kindCase_ = 0;
+          kind_ = null;
+          onChanged();
+        }
+      } else {
+        if (kindCase_ == 3) {
+          kindCase_ = 0;
+          kind_ = null;
+        }
+        int64ListBuilder_.clear();
+      }
+      return this;
+    }
+    /**
+     * <code>.tensorflow.CollectionDef.Int64List int64_list = 3;</code>
+     */
+    public org.tensorflow.proto.framework.CollectionDef.Int64List.Builder getInt64ListBuilder() {
+      return getInt64ListFieldBuilder().getBuilder();
+    }
+    /**
+     * <code>.tensorflow.CollectionDef.Int64List int64_list = 3;</code>
+     */
+    public org.tensorflow.proto.framework.CollectionDef.Int64ListOrBuilder getInt64ListOrBuilder() {
+      if ((kindCase_ == 3) && (int64ListBuilder_ != null)) {
+        return int64ListBuilder_.getMessageOrBuilder();
+      } else {
+        if (kindCase_ == 3) {
+          return (org.tensorflow.proto.framework.CollectionDef.Int64List) kind_;
+        }
+        return org.tensorflow.proto.framework.CollectionDef.Int64List.getDefaultInstance();
+      }
+    }
+    /**
+     * <code>.tensorflow.CollectionDef.Int64List int64_list = 3;</code>
+     */
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.CollectionDef.Int64List, org.tensorflow.proto.framework.CollectionDef.Int64List.Builder, org.tensorflow.proto.framework.CollectionDef.Int64ListOrBuilder> 
+        getInt64ListFieldBuilder() {
+      if (int64ListBuilder_ == null) {
+        if (!(kindCase_ == 3)) {
+          kind_ = org.tensorflow.proto.framework.CollectionDef.Int64List.getDefaultInstance();
+        }
+        int64ListBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+            org.tensorflow.proto.framework.CollectionDef.Int64List, org.tensorflow.proto.framework.CollectionDef.Int64List.Builder, org.tensorflow.proto.framework.CollectionDef.Int64ListOrBuilder>(
+                (org.tensorflow.proto.framework.CollectionDef.Int64List) kind_,
+                getParentForChildren(),
+                isClean());
+        kind_ = null;
+      }
+      kindCase_ = 3;
+      onChanged();;
+      return int64ListBuilder_;
+    }
+
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.CollectionDef.FloatList, org.tensorflow.proto.framework.CollectionDef.FloatList.Builder, org.tensorflow.proto.framework.CollectionDef.FloatListOrBuilder> floatListBuilder_;
+    /**
+     * <code>.tensorflow.CollectionDef.FloatList float_list = 4;</code>
+     */
+    public boolean hasFloatList() {
+      return kindCase_ == 4;
+    }
+    /**
+     * <code>.tensorflow.CollectionDef.FloatList float_list = 4;</code>
+     */
+    public org.tensorflow.proto.framework.CollectionDef.FloatList getFloatList() {
+      if (floatListBuilder_ == null) {
+        if (kindCase_ == 4) {
+          return (org.tensorflow.proto.framework.CollectionDef.FloatList) kind_;
+        }
+        return org.tensorflow.proto.framework.CollectionDef.FloatList.getDefaultInstance();
+      } else {
+        if (kindCase_ == 4) {
+          return floatListBuilder_.getMessage();
+        }
+        return org.tensorflow.proto.framework.CollectionDef.FloatList.getDefaultInstance();
+      }
+    }
+    /**
+     * <code>.tensorflow.CollectionDef.FloatList float_list = 4;</code>
+     */
+    public Builder setFloatList(org.tensorflow.proto.framework.CollectionDef.FloatList value) {
+      if (floatListBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        kind_ = value;
+        onChanged();
+      } else {
+        floatListBuilder_.setMessage(value);
+      }
+      kindCase_ = 4;
+      return this;
+    }
+    /**
+     * <code>.tensorflow.CollectionDef.FloatList float_list = 4;</code>
+     */
+    public Builder setFloatList(
+        org.tensorflow.proto.framework.CollectionDef.FloatList.Builder builderForValue) {
+      if (floatListBuilder_ == null) {
+        kind_ = builderForValue.build();
+        onChanged();
+      } else {
+        floatListBuilder_.setMessage(builderForValue.build());
+      }
+      kindCase_ = 4;
+      return this;
+    }
+    /**
+     * <code>.tensorflow.CollectionDef.FloatList float_list = 4;</code>
+     */
+    public Builder mergeFloatList(org.tensorflow.proto.framework.CollectionDef.FloatList value) {
+      if (floatListBuilder_ == null) {
+        if (kindCase_ == 4 &&
+            kind_ != org.tensorflow.proto.framework.CollectionDef.FloatList.getDefaultInstance()) {
+          kind_ = org.tensorflow.proto.framework.CollectionDef.FloatList.newBuilder((org.tensorflow.proto.framework.CollectionDef.FloatList) kind_)
+              .mergeFrom(value).buildPartial();
+        } else {
+          kind_ = value;
+        }
+        onChanged();
+      } else {
+        if (kindCase_ == 4) {
+          floatListBuilder_.mergeFrom(value);
+        }
+        floatListBuilder_.setMessage(value);
+      }
+      kindCase_ = 4;
+      return this;
+    }
+    /**
+     * <code>.tensorflow.CollectionDef.FloatList float_list = 4;</code>
+     */
+    public Builder clearFloatList() {
+      if (floatListBuilder_ == null) {
+        if (kindCase_ == 4) {
+          kindCase_ = 0;
+          kind_ = null;
+          onChanged();
+        }
+      } else {
+        if (kindCase_ == 4) {
+          kindCase_ = 0;
+          kind_ = null;
+        }
+        floatListBuilder_.clear();
+      }
+      return this;
+    }
+    /**
+     * <code>.tensorflow.CollectionDef.FloatList float_list = 4;</code>
+     */
+    public org.tensorflow.proto.framework.CollectionDef.FloatList.Builder getFloatListBuilder() {
+      return getFloatListFieldBuilder().getBuilder();
+    }
+    /**
+     * <code>.tensorflow.CollectionDef.FloatList float_list = 4;</code>
+     */
+    public org.tensorflow.proto.framework.CollectionDef.FloatListOrBuilder getFloatListOrBuilder() {
+      if ((kindCase_ == 4) && (floatListBuilder_ != null)) {
+        return floatListBuilder_.getMessageOrBuilder();
+      } else {
+        if (kindCase_ == 4) {
+          return (org.tensorflow.proto.framework.CollectionDef.FloatList) kind_;
+        }
+        return org.tensorflow.proto.framework.CollectionDef.FloatList.getDefaultInstance();
+      }
+    }
+    /**
+     * <code>.tensorflow.CollectionDef.FloatList float_list = 4;</code>
+     */
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.CollectionDef.FloatList, org.tensorflow.proto.framework.CollectionDef.FloatList.Builder, org.tensorflow.proto.framework.CollectionDef.FloatListOrBuilder> 
+        getFloatListFieldBuilder() {
+      if (floatListBuilder_ == null) {
+        if (!(kindCase_ == 4)) {
+          kind_ = org.tensorflow.proto.framework.CollectionDef.FloatList.getDefaultInstance();
+        }
+        floatListBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+            org.tensorflow.proto.framework.CollectionDef.FloatList, org.tensorflow.proto.framework.CollectionDef.FloatList.Builder, org.tensorflow.proto.framework.CollectionDef.FloatListOrBuilder>(
+                (org.tensorflow.proto.framework.CollectionDef.FloatList) kind_,
+                getParentForChildren(),
+                isClean());
+        kind_ = null;
+      }
+      kindCase_ = 4;
+      onChanged();;
+      return floatListBuilder_;
+    }
+
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.CollectionDef.AnyList, org.tensorflow.proto.framework.CollectionDef.AnyList.Builder, org.tensorflow.proto.framework.CollectionDef.AnyListOrBuilder> anyListBuilder_;
+    /**
+     * <code>.tensorflow.CollectionDef.AnyList any_list = 5;</code>
+     */
+    public boolean hasAnyList() {
+      return kindCase_ == 5;
+    }
+    /**
+     * <code>.tensorflow.CollectionDef.AnyList any_list = 5;</code>
+     */
+    public org.tensorflow.proto.framework.CollectionDef.AnyList getAnyList() {
+      if (anyListBuilder_ == null) {
+        if (kindCase_ == 5) {
+          return (org.tensorflow.proto.framework.CollectionDef.AnyList) kind_;
+        }
+        return org.tensorflow.proto.framework.CollectionDef.AnyList.getDefaultInstance();
+      } else {
+        if (kindCase_ == 5) {
+          return anyListBuilder_.getMessage();
+        }
+        return org.tensorflow.proto.framework.CollectionDef.AnyList.getDefaultInstance();
+      }
+    }
+    /**
+     * <code>.tensorflow.CollectionDef.AnyList any_list = 5;</code>
+     */
+    public Builder setAnyList(org.tensorflow.proto.framework.CollectionDef.AnyList value) {
+      if (anyListBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        kind_ = value;
+        onChanged();
+      } else {
+        anyListBuilder_.setMessage(value);
+      }
+      kindCase_ = 5;
+      return this;
+    }
+    /**
+     * <code>.tensorflow.CollectionDef.AnyList any_list = 5;</code>
+     */
+    public Builder setAnyList(
+        org.tensorflow.proto.framework.CollectionDef.AnyList.Builder builderForValue) {
+      if (anyListBuilder_ == null) {
+        kind_ = builderForValue.build();
+        onChanged();
+      } else {
+        anyListBuilder_.setMessage(builderForValue.build());
+      }
+      kindCase_ = 5;
+      return this;
+    }
+    /**
+     * <code>.tensorflow.CollectionDef.AnyList any_list = 5;</code>
+     */
+    public Builder mergeAnyList(org.tensorflow.proto.framework.CollectionDef.AnyList value) {
+      if (anyListBuilder_ == null) {
+        if (kindCase_ == 5 &&
+            kind_ != org.tensorflow.proto.framework.CollectionDef.AnyList.getDefaultInstance()) {
+          kind_ = org.tensorflow.proto.framework.CollectionDef.AnyList.newBuilder((org.tensorflow.proto.framework.CollectionDef.AnyList) kind_)
+              .mergeFrom(value).buildPartial();
+        } else {
+          kind_ = value;
+        }
+        onChanged();
+      } else {
+        if (kindCase_ == 5) {
+          anyListBuilder_.mergeFrom(value);
+        }
+        anyListBuilder_.setMessage(value);
+      }
+      kindCase_ = 5;
+      return this;
+    }
+    /**
+     * <code>.tensorflow.CollectionDef.AnyList any_list = 5;</code>
+     */
+    public Builder clearAnyList() {
+      if (anyListBuilder_ == null) {
+        if (kindCase_ == 5) {
+          kindCase_ = 0;
+          kind_ = null;
+          onChanged();
+        }
+      } else {
+        if (kindCase_ == 5) {
+          kindCase_ = 0;
+          kind_ = null;
+        }
+        anyListBuilder_.clear();
+      }
+      return this;
+    }
+    /**
+     * <code>.tensorflow.CollectionDef.AnyList any_list = 5;</code>
+     */
+    public org.tensorflow.proto.framework.CollectionDef.AnyList.Builder getAnyListBuilder() {
+      return getAnyListFieldBuilder().getBuilder();
+    }
+    /**
+     * <code>.tensorflow.CollectionDef.AnyList any_list = 5;</code>
+     */
+    public org.tensorflow.proto.framework.CollectionDef.AnyListOrBuilder getAnyListOrBuilder() {
+      if ((kindCase_ == 5) && (anyListBuilder_ != null)) {
+        return anyListBuilder_.getMessageOrBuilder();
+      } else {
+        if (kindCase_ == 5) {
+          return (org.tensorflow.proto.framework.CollectionDef.AnyList) kind_;
+        }
+        return org.tensorflow.proto.framework.CollectionDef.AnyList.getDefaultInstance();
+      }
+    }
+    /**
+     * <code>.tensorflow.CollectionDef.AnyList any_list = 5;</code>
+     */
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.CollectionDef.AnyList, org.tensorflow.proto.framework.CollectionDef.AnyList.Builder, org.tensorflow.proto.framework.CollectionDef.AnyListOrBuilder> 
+        getAnyListFieldBuilder() {
+      if (anyListBuilder_ == null) {
+        if (!(kindCase_ == 5)) {
+          kind_ = org.tensorflow.proto.framework.CollectionDef.AnyList.getDefaultInstance();
+        }
+        anyListBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+            org.tensorflow.proto.framework.CollectionDef.AnyList, org.tensorflow.proto.framework.CollectionDef.AnyList.Builder, org.tensorflow.proto.framework.CollectionDef.AnyListOrBuilder>(
+                (org.tensorflow.proto.framework.CollectionDef.AnyList) kind_,
+                getParentForChildren(),
+                isClean());
+        kind_ = null;
+      }
+      kindCase_ = 5;
+      onChanged();;
+      return anyListBuilder_;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.CollectionDef)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.CollectionDef)
+  private static final org.tensorflow.proto.framework.CollectionDef DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.CollectionDef();
+  }
+
+  public static org.tensorflow.proto.framework.CollectionDef getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<CollectionDef>
+      PARSER = new com.google.protobuf.AbstractParser<CollectionDef>() {
+    @java.lang.Override
+    public CollectionDef parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new CollectionDef(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<CollectionDef> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<CollectionDef> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.framework.CollectionDef getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/CollectionDefOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/CollectionDefOrBuilder.java
new file mode 100644
index 00000000000..f3d5fc7ce61
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/CollectionDefOrBuilder.java
@@ -0,0 +1,76 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/meta_graph.proto
+
+package org.tensorflow.proto.framework;
+
+public interface CollectionDefOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.CollectionDef)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <code>.tensorflow.CollectionDef.NodeList node_list = 1;</code>
+   */
+  boolean hasNodeList();
+  /**
+   * <code>.tensorflow.CollectionDef.NodeList node_list = 1;</code>
+   */
+  org.tensorflow.proto.framework.CollectionDef.NodeList getNodeList();
+  /**
+   * <code>.tensorflow.CollectionDef.NodeList node_list = 1;</code>
+   */
+  org.tensorflow.proto.framework.CollectionDef.NodeListOrBuilder getNodeListOrBuilder();
+
+  /**
+   * <code>.tensorflow.CollectionDef.BytesList bytes_list = 2;</code>
+   */
+  boolean hasBytesList();
+  /**
+   * <code>.tensorflow.CollectionDef.BytesList bytes_list = 2;</code>
+   */
+  org.tensorflow.proto.framework.CollectionDef.BytesList getBytesList();
+  /**
+   * <code>.tensorflow.CollectionDef.BytesList bytes_list = 2;</code>
+   */
+  org.tensorflow.proto.framework.CollectionDef.BytesListOrBuilder getBytesListOrBuilder();
+
+  /**
+   * <code>.tensorflow.CollectionDef.Int64List int64_list = 3;</code>
+   */
+  boolean hasInt64List();
+  /**
+   * <code>.tensorflow.CollectionDef.Int64List int64_list = 3;</code>
+   */
+  org.tensorflow.proto.framework.CollectionDef.Int64List getInt64List();
+  /**
+   * <code>.tensorflow.CollectionDef.Int64List int64_list = 3;</code>
+   */
+  org.tensorflow.proto.framework.CollectionDef.Int64ListOrBuilder getInt64ListOrBuilder();
+
+  /**
+   * <code>.tensorflow.CollectionDef.FloatList float_list = 4;</code>
+   */
+  boolean hasFloatList();
+  /**
+   * <code>.tensorflow.CollectionDef.FloatList float_list = 4;</code>
+   */
+  org.tensorflow.proto.framework.CollectionDef.FloatList getFloatList();
+  /**
+   * <code>.tensorflow.CollectionDef.FloatList float_list = 4;</code>
+   */
+  org.tensorflow.proto.framework.CollectionDef.FloatListOrBuilder getFloatListOrBuilder();
+
+  /**
+   * <code>.tensorflow.CollectionDef.AnyList any_list = 5;</code>
+   */
+  boolean hasAnyList();
+  /**
+   * <code>.tensorflow.CollectionDef.AnyList any_list = 5;</code>
+   */
+  org.tensorflow.proto.framework.CollectionDef.AnyList getAnyList();
+  /**
+   * <code>.tensorflow.CollectionDef.AnyList any_list = 5;</code>
+   */
+  org.tensorflow.proto.framework.CollectionDef.AnyListOrBuilder getAnyListOrBuilder();
+
+  public org.tensorflow.proto.framework.CollectionDef.KindCase getKindCase();
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/CondContextDef.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/CondContextDef.java
new file mode 100644
index 00000000000..3ab6d340e10
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/CondContextDef.java
@@ -0,0 +1,1632 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/control_flow.proto
+
+package org.tensorflow.proto.framework;
+
+/**
+ * <pre>
+ * Protocol buffer representing a CondContext object.
+ * </pre>
+ *
+ * Protobuf type {@code tensorflow.CondContextDef}
+ */
+public  final class CondContextDef extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.CondContextDef)
+    CondContextDefOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use CondContextDef.newBuilder() to construct.
+  private CondContextDef(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private CondContextDef() {
+    contextName_ = "";
+    predName_ = "";
+    pivotName_ = "";
+    nestedContexts_ = java.util.Collections.emptyList();
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new CondContextDef();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private CondContextDef(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    int mutable_bitField0_ = 0;
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 10: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            contextName_ = s;
+            break;
+          }
+          case 18: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            predName_ = s;
+            break;
+          }
+          case 26: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            pivotName_ = s;
+            break;
+          }
+          case 32: {
+
+            branch_ = input.readInt32();
+            break;
+          }
+          case 42: {
+            org.tensorflow.proto.framework.ValuesDef.Builder subBuilder = null;
+            if (valuesDef_ != null) {
+              subBuilder = valuesDef_.toBuilder();
+            }
+            valuesDef_ = input.readMessage(org.tensorflow.proto.framework.ValuesDef.parser(), extensionRegistry);
+            if (subBuilder != null) {
+              subBuilder.mergeFrom(valuesDef_);
+              valuesDef_ = subBuilder.buildPartial();
+            }
+
+            break;
+          }
+          case 50: {
+            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
+              nestedContexts_ = new java.util.ArrayList<org.tensorflow.proto.framework.ControlFlowContextDef>();
+              mutable_bitField0_ |= 0x00000001;
+            }
+            nestedContexts_.add(
+                input.readMessage(org.tensorflow.proto.framework.ControlFlowContextDef.parser(), extensionRegistry));
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      if (((mutable_bitField0_ & 0x00000001) != 0)) {
+        nestedContexts_ = java.util.Collections.unmodifiableList(nestedContexts_);
+      }
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.framework.ControlFlowProtos.internal_static_tensorflow_CondContextDef_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.framework.ControlFlowProtos.internal_static_tensorflow_CondContextDef_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.framework.CondContextDef.class, org.tensorflow.proto.framework.CondContextDef.Builder.class);
+  }
+
+  public static final int CONTEXT_NAME_FIELD_NUMBER = 1;
+  private volatile java.lang.Object contextName_;
+  /**
+   * <pre>
+   * Name of the context.
+   * </pre>
+   *
+   * <code>string context_name = 1;</code>
+   */
+  public java.lang.String getContextName() {
+    java.lang.Object ref = contextName_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      contextName_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * Name of the context.
+   * </pre>
+   *
+   * <code>string context_name = 1;</code>
+   */
+  public com.google.protobuf.ByteString
+      getContextNameBytes() {
+    java.lang.Object ref = contextName_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      contextName_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int PRED_NAME_FIELD_NUMBER = 2;
+  private volatile java.lang.Object predName_;
+  /**
+   * <pre>
+   * Name of the pred tensor.
+   * </pre>
+   *
+   * <code>string pred_name = 2;</code>
+   */
+  public java.lang.String getPredName() {
+    java.lang.Object ref = predName_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      predName_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * Name of the pred tensor.
+   * </pre>
+   *
+   * <code>string pred_name = 2;</code>
+   */
+  public com.google.protobuf.ByteString
+      getPredNameBytes() {
+    java.lang.Object ref = predName_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      predName_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int PIVOT_NAME_FIELD_NUMBER = 3;
+  private volatile java.lang.Object pivotName_;
+  /**
+   * <pre>
+   * Name of the pivot tensor.
+   * </pre>
+   *
+   * <code>string pivot_name = 3;</code>
+   */
+  public java.lang.String getPivotName() {
+    java.lang.Object ref = pivotName_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      pivotName_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * Name of the pivot tensor.
+   * </pre>
+   *
+   * <code>string pivot_name = 3;</code>
+   */
+  public com.google.protobuf.ByteString
+      getPivotNameBytes() {
+    java.lang.Object ref = pivotName_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      pivotName_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int BRANCH_FIELD_NUMBER = 4;
+  private int branch_;
+  /**
+   * <pre>
+   * Branch prediction. 0 or 1.
+   * </pre>
+   *
+   * <code>int32 branch = 4;</code>
+   */
+  public int getBranch() {
+    return branch_;
+  }
+
+  public static final int VALUES_DEF_FIELD_NUMBER = 5;
+  private org.tensorflow.proto.framework.ValuesDef valuesDef_;
+  /**
+   * <pre>
+   * Values and external values in control flow context.
+   * </pre>
+   *
+   * <code>.tensorflow.ValuesDef values_def = 5;</code>
+   */
+  public boolean hasValuesDef() {
+    return valuesDef_ != null;
+  }
+  /**
+   * <pre>
+   * Values and external values in control flow context.
+   * </pre>
+   *
+   * <code>.tensorflow.ValuesDef values_def = 5;</code>
+   */
+  public org.tensorflow.proto.framework.ValuesDef getValuesDef() {
+    return valuesDef_ == null ? org.tensorflow.proto.framework.ValuesDef.getDefaultInstance() : valuesDef_;
+  }
+  /**
+   * <pre>
+   * Values and external values in control flow context.
+   * </pre>
+   *
+   * <code>.tensorflow.ValuesDef values_def = 5;</code>
+   */
+  public org.tensorflow.proto.framework.ValuesDefOrBuilder getValuesDefOrBuilder() {
+    return getValuesDef();
+  }
+
+  public static final int NESTED_CONTEXTS_FIELD_NUMBER = 6;
+  private java.util.List<org.tensorflow.proto.framework.ControlFlowContextDef> nestedContexts_;
+  /**
+   * <pre>
+   * Contexts contained inside this context (e.g. nested conds).
+   * </pre>
+   *
+   * <code>repeated .tensorflow.ControlFlowContextDef nested_contexts = 6;</code>
+   */
+  public java.util.List<org.tensorflow.proto.framework.ControlFlowContextDef> getNestedContextsList() {
+    return nestedContexts_;
+  }
+  /**
+   * <pre>
+   * Contexts contained inside this context (e.g. nested conds).
+   * </pre>
+   *
+   * <code>repeated .tensorflow.ControlFlowContextDef nested_contexts = 6;</code>
+   */
+  public java.util.List<? extends org.tensorflow.proto.framework.ControlFlowContextDefOrBuilder> 
+      getNestedContextsOrBuilderList() {
+    return nestedContexts_;
+  }
+  /**
+   * <pre>
+   * Contexts contained inside this context (e.g. nested conds).
+   * </pre>
+   *
+   * <code>repeated .tensorflow.ControlFlowContextDef nested_contexts = 6;</code>
+   */
+  public int getNestedContextsCount() {
+    return nestedContexts_.size();
+  }
+  /**
+   * <pre>
+   * Contexts contained inside this context (e.g. nested conds).
+   * </pre>
+   *
+   * <code>repeated .tensorflow.ControlFlowContextDef nested_contexts = 6;</code>
+   */
+  public org.tensorflow.proto.framework.ControlFlowContextDef getNestedContexts(int index) {
+    return nestedContexts_.get(index);
+  }
+  /**
+   * <pre>
+   * Contexts contained inside this context (e.g. nested conds).
+   * </pre>
+   *
+   * <code>repeated .tensorflow.ControlFlowContextDef nested_contexts = 6;</code>
+   */
+  public org.tensorflow.proto.framework.ControlFlowContextDefOrBuilder getNestedContextsOrBuilder(
+      int index) {
+    return nestedContexts_.get(index);
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    if (!getContextNameBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 1, contextName_);
+    }
+    if (!getPredNameBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 2, predName_);
+    }
+    if (!getPivotNameBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 3, pivotName_);
+    }
+    if (branch_ != 0) {
+      output.writeInt32(4, branch_);
+    }
+    if (valuesDef_ != null) {
+      output.writeMessage(5, getValuesDef());
+    }
+    for (int i = 0; i < nestedContexts_.size(); i++) {
+      output.writeMessage(6, nestedContexts_.get(i));
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (!getContextNameBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, contextName_);
+    }
+    if (!getPredNameBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(2, predName_);
+    }
+    if (!getPivotNameBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(3, pivotName_);
+    }
+    if (branch_ != 0) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt32Size(4, branch_);
+    }
+    if (valuesDef_ != null) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(5, getValuesDef());
+    }
+    for (int i = 0; i < nestedContexts_.size(); i++) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(6, nestedContexts_.get(i));
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.framework.CondContextDef)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.framework.CondContextDef other = (org.tensorflow.proto.framework.CondContextDef) obj;
+
+    if (!getContextName()
+        .equals(other.getContextName())) return false;
+    if (!getPredName()
+        .equals(other.getPredName())) return false;
+    if (!getPivotName()
+        .equals(other.getPivotName())) return false;
+    if (getBranch()
+        != other.getBranch()) return false;
+    if (hasValuesDef() != other.hasValuesDef()) return false;
+    if (hasValuesDef()) {
+      if (!getValuesDef()
+          .equals(other.getValuesDef())) return false;
+    }
+    if (!getNestedContextsList()
+        .equals(other.getNestedContextsList())) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    hash = (37 * hash) + CONTEXT_NAME_FIELD_NUMBER;
+    hash = (53 * hash) + getContextName().hashCode();
+    hash = (37 * hash) + PRED_NAME_FIELD_NUMBER;
+    hash = (53 * hash) + getPredName().hashCode();
+    hash = (37 * hash) + PIVOT_NAME_FIELD_NUMBER;
+    hash = (53 * hash) + getPivotName().hashCode();
+    hash = (37 * hash) + BRANCH_FIELD_NUMBER;
+    hash = (53 * hash) + getBranch();
+    if (hasValuesDef()) {
+      hash = (37 * hash) + VALUES_DEF_FIELD_NUMBER;
+      hash = (53 * hash) + getValuesDef().hashCode();
+    }
+    if (getNestedContextsCount() > 0) {
+      hash = (37 * hash) + NESTED_CONTEXTS_FIELD_NUMBER;
+      hash = (53 * hash) + getNestedContextsList().hashCode();
+    }
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.framework.CondContextDef parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.CondContextDef parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.CondContextDef parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.CondContextDef parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.CondContextDef parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.CondContextDef parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.CondContextDef parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.CondContextDef parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.CondContextDef parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.CondContextDef parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.CondContextDef parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.CondContextDef parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.framework.CondContextDef prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * <pre>
+   * Protocol buffer representing a CondContext object.
+   * </pre>
+   *
+   * Protobuf type {@code tensorflow.CondContextDef}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.CondContextDef)
+      org.tensorflow.proto.framework.CondContextDefOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.ControlFlowProtos.internal_static_tensorflow_CondContextDef_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.ControlFlowProtos.internal_static_tensorflow_CondContextDef_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.CondContextDef.class, org.tensorflow.proto.framework.CondContextDef.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.framework.CondContextDef.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+        getNestedContextsFieldBuilder();
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      contextName_ = "";
+
+      predName_ = "";
+
+      pivotName_ = "";
+
+      branch_ = 0;
+
+      if (valuesDefBuilder_ == null) {
+        valuesDef_ = null;
+      } else {
+        valuesDef_ = null;
+        valuesDefBuilder_ = null;
+      }
+      if (nestedContextsBuilder_ == null) {
+        nestedContexts_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000001);
+      } else {
+        nestedContextsBuilder_.clear();
+      }
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.framework.ControlFlowProtos.internal_static_tensorflow_CondContextDef_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.CondContextDef getDefaultInstanceForType() {
+      return org.tensorflow.proto.framework.CondContextDef.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.CondContextDef build() {
+      org.tensorflow.proto.framework.CondContextDef result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.CondContextDef buildPartial() {
+      org.tensorflow.proto.framework.CondContextDef result = new org.tensorflow.proto.framework.CondContextDef(this);
+      int from_bitField0_ = bitField0_;
+      result.contextName_ = contextName_;
+      result.predName_ = predName_;
+      result.pivotName_ = pivotName_;
+      result.branch_ = branch_;
+      if (valuesDefBuilder_ == null) {
+        result.valuesDef_ = valuesDef_;
+      } else {
+        result.valuesDef_ = valuesDefBuilder_.build();
+      }
+      if (nestedContextsBuilder_ == null) {
+        if (((bitField0_ & 0x00000001) != 0)) {
+          nestedContexts_ = java.util.Collections.unmodifiableList(nestedContexts_);
+          bitField0_ = (bitField0_ & ~0x00000001);
+        }
+        result.nestedContexts_ = nestedContexts_;
+      } else {
+        result.nestedContexts_ = nestedContextsBuilder_.build();
+      }
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.framework.CondContextDef) {
+        return mergeFrom((org.tensorflow.proto.framework.CondContextDef)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.framework.CondContextDef other) {
+      if (other == org.tensorflow.proto.framework.CondContextDef.getDefaultInstance()) return this;
+      if (!other.getContextName().isEmpty()) {
+        contextName_ = other.contextName_;
+        onChanged();
+      }
+      if (!other.getPredName().isEmpty()) {
+        predName_ = other.predName_;
+        onChanged();
+      }
+      if (!other.getPivotName().isEmpty()) {
+        pivotName_ = other.pivotName_;
+        onChanged();
+      }
+      if (other.getBranch() != 0) {
+        setBranch(other.getBranch());
+      }
+      if (other.hasValuesDef()) {
+        mergeValuesDef(other.getValuesDef());
+      }
+      if (nestedContextsBuilder_ == null) {
+        if (!other.nestedContexts_.isEmpty()) {
+          if (nestedContexts_.isEmpty()) {
+            nestedContexts_ = other.nestedContexts_;
+            bitField0_ = (bitField0_ & ~0x00000001);
+          } else {
+            ensureNestedContextsIsMutable();
+            nestedContexts_.addAll(other.nestedContexts_);
+          }
+          onChanged();
+        }
+      } else {
+        if (!other.nestedContexts_.isEmpty()) {
+          if (nestedContextsBuilder_.isEmpty()) {
+            nestedContextsBuilder_.dispose();
+            nestedContextsBuilder_ = null;
+            nestedContexts_ = other.nestedContexts_;
+            bitField0_ = (bitField0_ & ~0x00000001);
+            nestedContextsBuilder_ = 
+              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
+                 getNestedContextsFieldBuilder() : null;
+          } else {
+            nestedContextsBuilder_.addAllMessages(other.nestedContexts_);
+          }
+        }
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.framework.CondContextDef parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.framework.CondContextDef) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+    private int bitField0_;
+
+    private java.lang.Object contextName_ = "";
+    /**
+     * <pre>
+     * Name of the context.
+     * </pre>
+     *
+     * <code>string context_name = 1;</code>
+     */
+    public java.lang.String getContextName() {
+      java.lang.Object ref = contextName_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        contextName_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Name of the context.
+     * </pre>
+     *
+     * <code>string context_name = 1;</code>
+     */
+    public com.google.protobuf.ByteString
+        getContextNameBytes() {
+      java.lang.Object ref = contextName_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        contextName_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Name of the context.
+     * </pre>
+     *
+     * <code>string context_name = 1;</code>
+     */
+    public Builder setContextName(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      contextName_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Name of the context.
+     * </pre>
+     *
+     * <code>string context_name = 1;</code>
+     */
+    public Builder clearContextName() {
+      
+      contextName_ = getDefaultInstance().getContextName();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Name of the context.
+     * </pre>
+     *
+     * <code>string context_name = 1;</code>
+     */
+    public Builder setContextNameBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      contextName_ = value;
+      onChanged();
+      return this;
+    }
+
+    private java.lang.Object predName_ = "";
+    /**
+     * <pre>
+     * Name of the pred tensor.
+     * </pre>
+     *
+     * <code>string pred_name = 2;</code>
+     */
+    public java.lang.String getPredName() {
+      java.lang.Object ref = predName_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        predName_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Name of the pred tensor.
+     * </pre>
+     *
+     * <code>string pred_name = 2;</code>
+     */
+    public com.google.protobuf.ByteString
+        getPredNameBytes() {
+      java.lang.Object ref = predName_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        predName_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Name of the pred tensor.
+     * </pre>
+     *
+     * <code>string pred_name = 2;</code>
+     */
+    public Builder setPredName(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      predName_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Name of the pred tensor.
+     * </pre>
+     *
+     * <code>string pred_name = 2;</code>
+     */
+    public Builder clearPredName() {
+      
+      predName_ = getDefaultInstance().getPredName();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Name of the pred tensor.
+     * </pre>
+     *
+     * <code>string pred_name = 2;</code>
+     */
+    public Builder setPredNameBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      predName_ = value;
+      onChanged();
+      return this;
+    }
+
+    private java.lang.Object pivotName_ = "";
+    /**
+     * <pre>
+     * Name of the pivot tensor.
+     * </pre>
+     *
+     * <code>string pivot_name = 3;</code>
+     */
+    public java.lang.String getPivotName() {
+      java.lang.Object ref = pivotName_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        pivotName_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Name of the pivot tensor.
+     * </pre>
+     *
+     * <code>string pivot_name = 3;</code>
+     */
+    public com.google.protobuf.ByteString
+        getPivotNameBytes() {
+      java.lang.Object ref = pivotName_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        pivotName_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Name of the pivot tensor.
+     * </pre>
+     *
+     * <code>string pivot_name = 3;</code>
+     */
+    public Builder setPivotName(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      pivotName_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Name of the pivot tensor.
+     * </pre>
+     *
+     * <code>string pivot_name = 3;</code>
+     */
+    public Builder clearPivotName() {
+      
+      pivotName_ = getDefaultInstance().getPivotName();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Name of the pivot tensor.
+     * </pre>
+     *
+     * <code>string pivot_name = 3;</code>
+     */
+    public Builder setPivotNameBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      pivotName_ = value;
+      onChanged();
+      return this;
+    }
+
+    private int branch_ ;
+    /**
+     * <pre>
+     * Branch prediction. 0 or 1.
+     * </pre>
+     *
+     * <code>int32 branch = 4;</code>
+     */
+    public int getBranch() {
+      return branch_;
+    }
+    /**
+     * <pre>
+     * Branch prediction. 0 or 1.
+     * </pre>
+     *
+     * <code>int32 branch = 4;</code>
+     */
+    public Builder setBranch(int value) {
+      
+      branch_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Branch prediction. 0 or 1.
+     * </pre>
+     *
+     * <code>int32 branch = 4;</code>
+     */
+    public Builder clearBranch() {
+      
+      branch_ = 0;
+      onChanged();
+      return this;
+    }
+
+    private org.tensorflow.proto.framework.ValuesDef valuesDef_;
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.ValuesDef, org.tensorflow.proto.framework.ValuesDef.Builder, org.tensorflow.proto.framework.ValuesDefOrBuilder> valuesDefBuilder_;
+    /**
+     * <pre>
+     * Values and external values in control flow context.
+     * </pre>
+     *
+     * <code>.tensorflow.ValuesDef values_def = 5;</code>
+     */
+    public boolean hasValuesDef() {
+      return valuesDefBuilder_ != null || valuesDef_ != null;
+    }
+    /**
+     * <pre>
+     * Values and external values in control flow context.
+     * </pre>
+     *
+     * <code>.tensorflow.ValuesDef values_def = 5;</code>
+     */
+    public org.tensorflow.proto.framework.ValuesDef getValuesDef() {
+      if (valuesDefBuilder_ == null) {
+        return valuesDef_ == null ? org.tensorflow.proto.framework.ValuesDef.getDefaultInstance() : valuesDef_;
+      } else {
+        return valuesDefBuilder_.getMessage();
+      }
+    }
+    /**
+     * <pre>
+     * Values and external values in control flow context.
+     * </pre>
+     *
+     * <code>.tensorflow.ValuesDef values_def = 5;</code>
+     */
+    public Builder setValuesDef(org.tensorflow.proto.framework.ValuesDef value) {
+      if (valuesDefBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        valuesDef_ = value;
+        onChanged();
+      } else {
+        valuesDefBuilder_.setMessage(value);
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Values and external values in control flow context.
+     * </pre>
+     *
+     * <code>.tensorflow.ValuesDef values_def = 5;</code>
+     */
+    public Builder setValuesDef(
+        org.tensorflow.proto.framework.ValuesDef.Builder builderForValue) {
+      if (valuesDefBuilder_ == null) {
+        valuesDef_ = builderForValue.build();
+        onChanged();
+      } else {
+        valuesDefBuilder_.setMessage(builderForValue.build());
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Values and external values in control flow context.
+     * </pre>
+     *
+     * <code>.tensorflow.ValuesDef values_def = 5;</code>
+     */
+    public Builder mergeValuesDef(org.tensorflow.proto.framework.ValuesDef value) {
+      if (valuesDefBuilder_ == null) {
+        if (valuesDef_ != null) {
+          valuesDef_ =
+            org.tensorflow.proto.framework.ValuesDef.newBuilder(valuesDef_).mergeFrom(value).buildPartial();
+        } else {
+          valuesDef_ = value;
+        }
+        onChanged();
+      } else {
+        valuesDefBuilder_.mergeFrom(value);
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Values and external values in control flow context.
+     * </pre>
+     *
+     * <code>.tensorflow.ValuesDef values_def = 5;</code>
+     */
+    public Builder clearValuesDef() {
+      if (valuesDefBuilder_ == null) {
+        valuesDef_ = null;
+        onChanged();
+      } else {
+        valuesDef_ = null;
+        valuesDefBuilder_ = null;
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Values and external values in control flow context.
+     * </pre>
+     *
+     * <code>.tensorflow.ValuesDef values_def = 5;</code>
+     */
+    public org.tensorflow.proto.framework.ValuesDef.Builder getValuesDefBuilder() {
+      
+      onChanged();
+      return getValuesDefFieldBuilder().getBuilder();
+    }
+    /**
+     * <pre>
+     * Values and external values in control flow context.
+     * </pre>
+     *
+     * <code>.tensorflow.ValuesDef values_def = 5;</code>
+     */
+    public org.tensorflow.proto.framework.ValuesDefOrBuilder getValuesDefOrBuilder() {
+      if (valuesDefBuilder_ != null) {
+        return valuesDefBuilder_.getMessageOrBuilder();
+      } else {
+        return valuesDef_ == null ?
+            org.tensorflow.proto.framework.ValuesDef.getDefaultInstance() : valuesDef_;
+      }
+    }
+    /**
+     * <pre>
+     * Values and external values in control flow context.
+     * </pre>
+     *
+     * <code>.tensorflow.ValuesDef values_def = 5;</code>
+     */
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.ValuesDef, org.tensorflow.proto.framework.ValuesDef.Builder, org.tensorflow.proto.framework.ValuesDefOrBuilder> 
+        getValuesDefFieldBuilder() {
+      if (valuesDefBuilder_ == null) {
+        valuesDefBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+            org.tensorflow.proto.framework.ValuesDef, org.tensorflow.proto.framework.ValuesDef.Builder, org.tensorflow.proto.framework.ValuesDefOrBuilder>(
+                getValuesDef(),
+                getParentForChildren(),
+                isClean());
+        valuesDef_ = null;
+      }
+      return valuesDefBuilder_;
+    }
+
+    private java.util.List<org.tensorflow.proto.framework.ControlFlowContextDef> nestedContexts_ =
+      java.util.Collections.emptyList();
+    private void ensureNestedContextsIsMutable() {
+      if (!((bitField0_ & 0x00000001) != 0)) {
+        nestedContexts_ = new java.util.ArrayList<org.tensorflow.proto.framework.ControlFlowContextDef>(nestedContexts_);
+        bitField0_ |= 0x00000001;
+       }
+    }
+
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.framework.ControlFlowContextDef, org.tensorflow.proto.framework.ControlFlowContextDef.Builder, org.tensorflow.proto.framework.ControlFlowContextDefOrBuilder> nestedContextsBuilder_;
+
+    /**
+     * <pre>
+     * Contexts contained inside this context (e.g. nested conds).
+     * </pre>
+     *
+     * <code>repeated .tensorflow.ControlFlowContextDef nested_contexts = 6;</code>
+     */
+    public java.util.List<org.tensorflow.proto.framework.ControlFlowContextDef> getNestedContextsList() {
+      if (nestedContextsBuilder_ == null) {
+        return java.util.Collections.unmodifiableList(nestedContexts_);
+      } else {
+        return nestedContextsBuilder_.getMessageList();
+      }
+    }
+    /**
+     * <pre>
+     * Contexts contained inside this context (e.g. nested conds).
+     * </pre>
+     *
+     * <code>repeated .tensorflow.ControlFlowContextDef nested_contexts = 6;</code>
+     */
+    public int getNestedContextsCount() {
+      if (nestedContextsBuilder_ == null) {
+        return nestedContexts_.size();
+      } else {
+        return nestedContextsBuilder_.getCount();
+      }
+    }
+    /**
+     * <pre>
+     * Contexts contained inside this context (e.g. nested conds).
+     * </pre>
+     *
+     * <code>repeated .tensorflow.ControlFlowContextDef nested_contexts = 6;</code>
+     */
+    public org.tensorflow.proto.framework.ControlFlowContextDef getNestedContexts(int index) {
+      if (nestedContextsBuilder_ == null) {
+        return nestedContexts_.get(index);
+      } else {
+        return nestedContextsBuilder_.getMessage(index);
+      }
+    }
+    /**
+     * <pre>
+     * Contexts contained inside this context (e.g. nested conds).
+     * </pre>
+     *
+     * <code>repeated .tensorflow.ControlFlowContextDef nested_contexts = 6;</code>
+     */
+    public Builder setNestedContexts(
+        int index, org.tensorflow.proto.framework.ControlFlowContextDef value) {
+      if (nestedContextsBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureNestedContextsIsMutable();
+        nestedContexts_.set(index, value);
+        onChanged();
+      } else {
+        nestedContextsBuilder_.setMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Contexts contained inside this context (e.g. nested conds).
+     * </pre>
+     *
+     * <code>repeated .tensorflow.ControlFlowContextDef nested_contexts = 6;</code>
+     */
+    public Builder setNestedContexts(
+        int index, org.tensorflow.proto.framework.ControlFlowContextDef.Builder builderForValue) {
+      if (nestedContextsBuilder_ == null) {
+        ensureNestedContextsIsMutable();
+        nestedContexts_.set(index, builderForValue.build());
+        onChanged();
+      } else {
+        nestedContextsBuilder_.setMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Contexts contained inside this context (e.g. nested conds).
+     * </pre>
+     *
+     * <code>repeated .tensorflow.ControlFlowContextDef nested_contexts = 6;</code>
+     */
+    public Builder addNestedContexts(org.tensorflow.proto.framework.ControlFlowContextDef value) {
+      if (nestedContextsBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureNestedContextsIsMutable();
+        nestedContexts_.add(value);
+        onChanged();
+      } else {
+        nestedContextsBuilder_.addMessage(value);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Contexts contained inside this context (e.g. nested conds).
+     * </pre>
+     *
+     * <code>repeated .tensorflow.ControlFlowContextDef nested_contexts = 6;</code>
+     */
+    public Builder addNestedContexts(
+        int index, org.tensorflow.proto.framework.ControlFlowContextDef value) {
+      if (nestedContextsBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureNestedContextsIsMutable();
+        nestedContexts_.add(index, value);
+        onChanged();
+      } else {
+        nestedContextsBuilder_.addMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Contexts contained inside this context (e.g. nested conds).
+     * </pre>
+     *
+     * <code>repeated .tensorflow.ControlFlowContextDef nested_contexts = 6;</code>
+     */
+    public Builder addNestedContexts(
+        org.tensorflow.proto.framework.ControlFlowContextDef.Builder builderForValue) {
+      if (nestedContextsBuilder_ == null) {
+        ensureNestedContextsIsMutable();
+        nestedContexts_.add(builderForValue.build());
+        onChanged();
+      } else {
+        nestedContextsBuilder_.addMessage(builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Contexts contained inside this context (e.g. nested conds).
+     * </pre>
+     *
+     * <code>repeated .tensorflow.ControlFlowContextDef nested_contexts = 6;</code>
+     */
+    public Builder addNestedContexts(
+        int index, org.tensorflow.proto.framework.ControlFlowContextDef.Builder builderForValue) {
+      if (nestedContextsBuilder_ == null) {
+        ensureNestedContextsIsMutable();
+        nestedContexts_.add(index, builderForValue.build());
+        onChanged();
+      } else {
+        nestedContextsBuilder_.addMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Contexts contained inside this context (e.g. nested conds).
+     * </pre>
+     *
+     * <code>repeated .tensorflow.ControlFlowContextDef nested_contexts = 6;</code>
+     */
+    public Builder addAllNestedContexts(
+        java.lang.Iterable<? extends org.tensorflow.proto.framework.ControlFlowContextDef> values) {
+      if (nestedContextsBuilder_ == null) {
+        ensureNestedContextsIsMutable();
+        com.google.protobuf.AbstractMessageLite.Builder.addAll(
+            values, nestedContexts_);
+        onChanged();
+      } else {
+        nestedContextsBuilder_.addAllMessages(values);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Contexts contained inside this context (e.g. nested conds).
+     * </pre>
+     *
+     * <code>repeated .tensorflow.ControlFlowContextDef nested_contexts = 6;</code>
+     */
+    public Builder clearNestedContexts() {
+      if (nestedContextsBuilder_ == null) {
+        nestedContexts_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000001);
+        onChanged();
+      } else {
+        nestedContextsBuilder_.clear();
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Contexts contained inside this context (e.g. nested conds).
+     * </pre>
+     *
+     * <code>repeated .tensorflow.ControlFlowContextDef nested_contexts = 6;</code>
+     */
+    public Builder removeNestedContexts(int index) {
+      if (nestedContextsBuilder_ == null) {
+        ensureNestedContextsIsMutable();
+        nestedContexts_.remove(index);
+        onChanged();
+      } else {
+        nestedContextsBuilder_.remove(index);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Contexts contained inside this context (e.g. nested conds).
+     * </pre>
+     *
+     * <code>repeated .tensorflow.ControlFlowContextDef nested_contexts = 6;</code>
+     */
+    public org.tensorflow.proto.framework.ControlFlowContextDef.Builder getNestedContextsBuilder(
+        int index) {
+      return getNestedContextsFieldBuilder().getBuilder(index);
+    }
+    /**
+     * <pre>
+     * Contexts contained inside this context (e.g. nested conds).
+     * </pre>
+     *
+     * <code>repeated .tensorflow.ControlFlowContextDef nested_contexts = 6;</code>
+     */
+    public org.tensorflow.proto.framework.ControlFlowContextDefOrBuilder getNestedContextsOrBuilder(
+        int index) {
+      if (nestedContextsBuilder_ == null) {
+        return nestedContexts_.get(index);  } else {
+        return nestedContextsBuilder_.getMessageOrBuilder(index);
+      }
+    }
+    /**
+     * <pre>
+     * Contexts contained inside this context (e.g. nested conds).
+     * </pre>
+     *
+     * <code>repeated .tensorflow.ControlFlowContextDef nested_contexts = 6;</code>
+     */
+    public java.util.List<? extends org.tensorflow.proto.framework.ControlFlowContextDefOrBuilder> 
+         getNestedContextsOrBuilderList() {
+      if (nestedContextsBuilder_ != null) {
+        return nestedContextsBuilder_.getMessageOrBuilderList();
+      } else {
+        return java.util.Collections.unmodifiableList(nestedContexts_);
+      }
+    }
+    /**
+     * <pre>
+     * Contexts contained inside this context (e.g. nested conds).
+     * </pre>
+     *
+     * <code>repeated .tensorflow.ControlFlowContextDef nested_contexts = 6;</code>
+     */
+    public org.tensorflow.proto.framework.ControlFlowContextDef.Builder addNestedContextsBuilder() {
+      return getNestedContextsFieldBuilder().addBuilder(
+          org.tensorflow.proto.framework.ControlFlowContextDef.getDefaultInstance());
+    }
+    /**
+     * <pre>
+     * Contexts contained inside this context (e.g. nested conds).
+     * </pre>
+     *
+     * <code>repeated .tensorflow.ControlFlowContextDef nested_contexts = 6;</code>
+     */
+    public org.tensorflow.proto.framework.ControlFlowContextDef.Builder addNestedContextsBuilder(
+        int index) {
+      return getNestedContextsFieldBuilder().addBuilder(
+          index, org.tensorflow.proto.framework.ControlFlowContextDef.getDefaultInstance());
+    }
+    /**
+     * <pre>
+     * Contexts contained inside this context (e.g. nested conds).
+     * </pre>
+     *
+     * <code>repeated .tensorflow.ControlFlowContextDef nested_contexts = 6;</code>
+     */
+    public java.util.List<org.tensorflow.proto.framework.ControlFlowContextDef.Builder> 
+         getNestedContextsBuilderList() {
+      return getNestedContextsFieldBuilder().getBuilderList();
+    }
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.framework.ControlFlowContextDef, org.tensorflow.proto.framework.ControlFlowContextDef.Builder, org.tensorflow.proto.framework.ControlFlowContextDefOrBuilder> 
+        getNestedContextsFieldBuilder() {
+      if (nestedContextsBuilder_ == null) {
+        nestedContextsBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
+            org.tensorflow.proto.framework.ControlFlowContextDef, org.tensorflow.proto.framework.ControlFlowContextDef.Builder, org.tensorflow.proto.framework.ControlFlowContextDefOrBuilder>(
+                nestedContexts_,
+                ((bitField0_ & 0x00000001) != 0),
+                getParentForChildren(),
+                isClean());
+        nestedContexts_ = null;
+      }
+      return nestedContextsBuilder_;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.CondContextDef)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.CondContextDef)
+  private static final org.tensorflow.proto.framework.CondContextDef DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.CondContextDef();
+  }
+
+  public static org.tensorflow.proto.framework.CondContextDef getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<CondContextDef>
+      PARSER = new com.google.protobuf.AbstractParser<CondContextDef>() {
+    @java.lang.Override
+    public CondContextDef parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new CondContextDef(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<CondContextDef> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<CondContextDef> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.framework.CondContextDef getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/CondContextDefOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/CondContextDefOrBuilder.java
new file mode 100644
index 00000000000..1dcdef5fd71
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/CondContextDefOrBuilder.java
@@ -0,0 +1,141 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/control_flow.proto
+
+package org.tensorflow.proto.framework;
+
+public interface CondContextDefOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.CondContextDef)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <pre>
+   * Name of the context.
+   * </pre>
+   *
+   * <code>string context_name = 1;</code>
+   */
+  java.lang.String getContextName();
+  /**
+   * <pre>
+   * Name of the context.
+   * </pre>
+   *
+   * <code>string context_name = 1;</code>
+   */
+  com.google.protobuf.ByteString
+      getContextNameBytes();
+
+  /**
+   * <pre>
+   * Name of the pred tensor.
+   * </pre>
+   *
+   * <code>string pred_name = 2;</code>
+   */
+  java.lang.String getPredName();
+  /**
+   * <pre>
+   * Name of the pred tensor.
+   * </pre>
+   *
+   * <code>string pred_name = 2;</code>
+   */
+  com.google.protobuf.ByteString
+      getPredNameBytes();
+
+  /**
+   * <pre>
+   * Name of the pivot tensor.
+   * </pre>
+   *
+   * <code>string pivot_name = 3;</code>
+   */
+  java.lang.String getPivotName();
+  /**
+   * <pre>
+   * Name of the pivot tensor.
+   * </pre>
+   *
+   * <code>string pivot_name = 3;</code>
+   */
+  com.google.protobuf.ByteString
+      getPivotNameBytes();
+
+  /**
+   * <pre>
+   * Branch prediction. 0 or 1.
+   * </pre>
+   *
+   * <code>int32 branch = 4;</code>
+   */
+  int getBranch();
+
+  /**
+   * <pre>
+   * Values and external values in control flow context.
+   * </pre>
+   *
+   * <code>.tensorflow.ValuesDef values_def = 5;</code>
+   */
+  boolean hasValuesDef();
+  /**
+   * <pre>
+   * Values and external values in control flow context.
+   * </pre>
+   *
+   * <code>.tensorflow.ValuesDef values_def = 5;</code>
+   */
+  org.tensorflow.proto.framework.ValuesDef getValuesDef();
+  /**
+   * <pre>
+   * Values and external values in control flow context.
+   * </pre>
+   *
+   * <code>.tensorflow.ValuesDef values_def = 5;</code>
+   */
+  org.tensorflow.proto.framework.ValuesDefOrBuilder getValuesDefOrBuilder();
+
+  /**
+   * <pre>
+   * Contexts contained inside this context (e.g. nested conds).
+   * </pre>
+   *
+   * <code>repeated .tensorflow.ControlFlowContextDef nested_contexts = 6;</code>
+   */
+  java.util.List<org.tensorflow.proto.framework.ControlFlowContextDef> 
+      getNestedContextsList();
+  /**
+   * <pre>
+   * Contexts contained inside this context (e.g. nested conds).
+   * </pre>
+   *
+   * <code>repeated .tensorflow.ControlFlowContextDef nested_contexts = 6;</code>
+   */
+  org.tensorflow.proto.framework.ControlFlowContextDef getNestedContexts(int index);
+  /**
+   * <pre>
+   * Contexts contained inside this context (e.g. nested conds).
+   * </pre>
+   *
+   * <code>repeated .tensorflow.ControlFlowContextDef nested_contexts = 6;</code>
+   */
+  int getNestedContextsCount();
+  /**
+   * <pre>
+   * Contexts contained inside this context (e.g. nested conds).
+   * </pre>
+   *
+   * <code>repeated .tensorflow.ControlFlowContextDef nested_contexts = 6;</code>
+   */
+  java.util.List<? extends org.tensorflow.proto.framework.ControlFlowContextDefOrBuilder> 
+      getNestedContextsOrBuilderList();
+  /**
+   * <pre>
+   * Contexts contained inside this context (e.g. nested conds).
+   * </pre>
+   *
+   * <code>repeated .tensorflow.ControlFlowContextDef nested_contexts = 6;</code>
+   */
+  org.tensorflow.proto.framework.ControlFlowContextDefOrBuilder getNestedContextsOrBuilder(
+      int index);
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ConfigProto.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ConfigProto.java
new file mode 100644
index 00000000000..a1c1e579953
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ConfigProto.java
@@ -0,0 +1,6064 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/config.proto
+
+package org.tensorflow.proto.framework;
+
+/**
+ * <pre>
+ * Session configuration parameters.
+ * The system picks appropriate values for fields that are not set.
+ * </pre>
+ *
+ * Protobuf type {@code tensorflow.ConfigProto}
+ */
+public  final class ConfigProto extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.ConfigProto)
+    ConfigProtoOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use ConfigProto.newBuilder() to construct.
+  private ConfigProto(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private ConfigProto() {
+    sessionInterOpThreadPool_ = java.util.Collections.emptyList();
+    deviceFilters_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new ConfigProto();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private ConfigProto(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    int mutable_bitField0_ = 0;
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 10: {
+            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
+              deviceCount_ = com.google.protobuf.MapField.newMapField(
+                  DeviceCountDefaultEntryHolder.defaultEntry);
+              mutable_bitField0_ |= 0x00000001;
+            }
+            com.google.protobuf.MapEntry<java.lang.String, java.lang.Integer>
+            deviceCount__ = input.readMessage(
+                DeviceCountDefaultEntryHolder.defaultEntry.getParserForType(), extensionRegistry);
+            deviceCount_.getMutableMap().put(
+                deviceCount__.getKey(), deviceCount__.getValue());
+            break;
+          }
+          case 16: {
+
+            intraOpParallelismThreads_ = input.readInt32();
+            break;
+          }
+          case 24: {
+
+            placementPeriod_ = input.readInt32();
+            break;
+          }
+          case 34: {
+            java.lang.String s = input.readStringRequireUtf8();
+            if (!((mutable_bitField0_ & 0x00000004) != 0)) {
+              deviceFilters_ = new com.google.protobuf.LazyStringArrayList();
+              mutable_bitField0_ |= 0x00000004;
+            }
+            deviceFilters_.add(s);
+            break;
+          }
+          case 40: {
+
+            interOpParallelismThreads_ = input.readInt32();
+            break;
+          }
+          case 50: {
+            org.tensorflow.proto.framework.GPUOptions.Builder subBuilder = null;
+            if (gpuOptions_ != null) {
+              subBuilder = gpuOptions_.toBuilder();
+            }
+            gpuOptions_ = input.readMessage(org.tensorflow.proto.framework.GPUOptions.parser(), extensionRegistry);
+            if (subBuilder != null) {
+              subBuilder.mergeFrom(gpuOptions_);
+              gpuOptions_ = subBuilder.buildPartial();
+            }
+
+            break;
+          }
+          case 56: {
+
+            allowSoftPlacement_ = input.readBool();
+            break;
+          }
+          case 64: {
+
+            logDevicePlacement_ = input.readBool();
+            break;
+          }
+          case 72: {
+
+            usePerSessionThreads_ = input.readBool();
+            break;
+          }
+          case 82: {
+            org.tensorflow.proto.framework.GraphOptions.Builder subBuilder = null;
+            if (graphOptions_ != null) {
+              subBuilder = graphOptions_.toBuilder();
+            }
+            graphOptions_ = input.readMessage(org.tensorflow.proto.framework.GraphOptions.parser(), extensionRegistry);
+            if (subBuilder != null) {
+              subBuilder.mergeFrom(graphOptions_);
+              graphOptions_ = subBuilder.buildPartial();
+            }
+
+            break;
+          }
+          case 88: {
+
+            operationTimeoutInMs_ = input.readInt64();
+            break;
+          }
+          case 98: {
+            if (!((mutable_bitField0_ & 0x00000002) != 0)) {
+              sessionInterOpThreadPool_ = new java.util.ArrayList<org.tensorflow.proto.framework.ThreadPoolOptionProto>();
+              mutable_bitField0_ |= 0x00000002;
+            }
+            sessionInterOpThreadPool_.add(
+                input.readMessage(org.tensorflow.proto.framework.ThreadPoolOptionProto.parser(), extensionRegistry));
+            break;
+          }
+          case 106: {
+            org.tensorflow.proto.framework.RPCOptions.Builder subBuilder = null;
+            if (rpcOptions_ != null) {
+              subBuilder = rpcOptions_.toBuilder();
+            }
+            rpcOptions_ = input.readMessage(org.tensorflow.proto.framework.RPCOptions.parser(), extensionRegistry);
+            if (subBuilder != null) {
+              subBuilder.mergeFrom(rpcOptions_);
+              rpcOptions_ = subBuilder.buildPartial();
+            }
+
+            break;
+          }
+          case 114: {
+            org.tensorflow.proto.distruntime.ClusterDef.Builder subBuilder = null;
+            if (clusterDef_ != null) {
+              subBuilder = clusterDef_.toBuilder();
+            }
+            clusterDef_ = input.readMessage(org.tensorflow.proto.distruntime.ClusterDef.parser(), extensionRegistry);
+            if (subBuilder != null) {
+              subBuilder.mergeFrom(clusterDef_);
+              clusterDef_ = subBuilder.buildPartial();
+            }
+
+            break;
+          }
+          case 120: {
+
+            isolateSessionState_ = input.readBool();
+            break;
+          }
+          case 130: {
+            org.tensorflow.proto.framework.ConfigProto.Experimental.Builder subBuilder = null;
+            if (experimental_ != null) {
+              subBuilder = experimental_.toBuilder();
+            }
+            experimental_ = input.readMessage(org.tensorflow.proto.framework.ConfigProto.Experimental.parser(), extensionRegistry);
+            if (subBuilder != null) {
+              subBuilder.mergeFrom(experimental_);
+              experimental_ = subBuilder.buildPartial();
+            }
+
+            break;
+          }
+          case 136: {
+
+            shareClusterDevicesInSession_ = input.readBool();
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      if (((mutable_bitField0_ & 0x00000004) != 0)) {
+        deviceFilters_ = deviceFilters_.getUnmodifiableView();
+      }
+      if (((mutable_bitField0_ & 0x00000002) != 0)) {
+        sessionInterOpThreadPool_ = java.util.Collections.unmodifiableList(sessionInterOpThreadPool_);
+      }
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_ConfigProto_descriptor;
+  }
+
+  @SuppressWarnings({"rawtypes"})
+  @java.lang.Override
+  protected com.google.protobuf.MapField internalGetMapField(
+      int number) {
+    switch (number) {
+      case 1:
+        return internalGetDeviceCount();
+      default:
+        throw new RuntimeException(
+            "Invalid map field number: " + number);
+    }
+  }
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_ConfigProto_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.framework.ConfigProto.class, org.tensorflow.proto.framework.ConfigProto.Builder.class);
+  }
+
+  public interface ExperimentalOrBuilder extends
+      // @@protoc_insertion_point(interface_extends:tensorflow.ConfigProto.Experimental)
+      com.google.protobuf.MessageOrBuilder {
+
+    /**
+     * <pre>
+     * Task name for group resolution.
+     * </pre>
+     *
+     * <code>string collective_group_leader = 1;</code>
+     */
+    java.lang.String getCollectiveGroupLeader();
+    /**
+     * <pre>
+     * Task name for group resolution.
+     * </pre>
+     *
+     * <code>string collective_group_leader = 1;</code>
+     */
+    com.google.protobuf.ByteString
+        getCollectiveGroupLeaderBytes();
+
+    /**
+     * <pre>
+     * Which executor to use, the default executor will be used
+     * if it is an empty string or "DEFAULT"
+     * </pre>
+     *
+     * <code>string executor_type = 3;</code>
+     */
+    java.lang.String getExecutorType();
+    /**
+     * <pre>
+     * Which executor to use, the default executor will be used
+     * if it is an empty string or "DEFAULT"
+     * </pre>
+     *
+     * <code>string executor_type = 3;</code>
+     */
+    com.google.protobuf.ByteString
+        getExecutorTypeBytes();
+
+    /**
+     * <pre>
+     * Guidance to formatting of large RecvBuf fields for transfer.
+     * Any positive value sets the max chunk size.  0 defaults to 4096.
+     * Any negative value indicates no max, i.e. one chunk only.
+     * </pre>
+     *
+     * <code>int32 recv_buf_max_chunk = 4;</code>
+     */
+    int getRecvBufMaxChunk();
+
+    /**
+     * <pre>
+     * If true, and supported by the platform, the runtime will attempt to
+     * use NUMA affinity where applicable.  One consequence will be the
+     * existence of as many CPU devices as there are available NUMA nodes.
+     * </pre>
+     *
+     * <code>bool use_numa_affinity = 5;</code>
+     */
+    boolean getUseNumaAffinity();
+
+    /**
+     * <pre>
+     * If true, make collective op execution order sequential and deterministic
+     * for potentially concurrent collective instances.
+     * </pre>
+     *
+     * <code>bool collective_deterministic_sequential_execution = 6;</code>
+     */
+    boolean getCollectiveDeterministicSequentialExecution();
+
+    /**
+     * <pre>
+     * If true, use NCCL for CollectiveOps.  This feature is highly
+     * experimental.
+     * </pre>
+     *
+     * <code>bool collective_nccl = 7;</code>
+     */
+    boolean getCollectiveNccl();
+
+    /**
+     * <pre>
+     * In the following, session state means the value of a variable, elements
+     * in a hash table, or any other resource, accessible by worker sessions
+     * held by a TF server.
+     * When ClusterSpec propagation is enabled, the value of
+     * isolate_session_state is ignored when deciding whether to share session
+     * states in a TF server (for backwards compatibility reasons).
+     * - If share_session_state_in_clusterspec_propagation is true, the session
+     * states are shared.
+     * - If share_session_state_in_clusterspec_propagation is false, session
+     * states are isolated.
+     * When clusterspec propagation is not used, the value of
+     * share_session_state_in_clusterspec_propagation is ignored when deciding
+     * whether to share session states in a TF server.
+     * - If isolate_session_state is true, session states are isolated.
+     * - If isolate_session_state is false, session states are shared.
+     * TODO(b/129330037): Add a single API that consistently treats
+     * isolate_session_state and ClusterSpec propagation.
+     * </pre>
+     *
+     * <code>bool share_session_state_in_clusterspec_propagation = 8;</code>
+     */
+    boolean getShareSessionStateInClusterspecPropagation();
+
+    /**
+     * <pre>
+     * If using a direct session, disable spinning while waiting for work in
+     * the thread pool. This may result in higher latency for completing ops,
+     * but in the case where there is a lot of spinning may result in lower
+     * CPU usage.
+     * </pre>
+     *
+     * <code>bool disable_thread_spinning = 9;</code>
+     */
+    boolean getDisableThreadSpinning();
+
+    /**
+     * <pre>
+     * This was promoted to a non-experimental API. Please use
+     * ConfigProto.share_cluster_devices_in_session instead.
+     * </pre>
+     *
+     * <code>bool share_cluster_devices_in_session = 10;</code>
+     */
+    boolean getShareClusterDevicesInSession();
+
+    /**
+     * <pre>
+     * Metadata about the session.
+     * If set, this can be used by the runtime and the Ops for debugging,
+     * monitoring, etc.
+     * NOTE: This is currently used and propagated only by the direct session.
+     * </pre>
+     *
+     * <code>.tensorflow.SessionMetadata session_metadata = 11;</code>
+     */
+    boolean hasSessionMetadata();
+    /**
+     * <pre>
+     * Metadata about the session.
+     * If set, this can be used by the runtime and the Ops for debugging,
+     * monitoring, etc.
+     * NOTE: This is currently used and propagated only by the direct session.
+     * </pre>
+     *
+     * <code>.tensorflow.SessionMetadata session_metadata = 11;</code>
+     */
+    org.tensorflow.proto.framework.SessionMetadata getSessionMetadata();
+    /**
+     * <pre>
+     * Metadata about the session.
+     * If set, this can be used by the runtime and the Ops for debugging,
+     * monitoring, etc.
+     * NOTE: This is currently used and propagated only by the direct session.
+     * </pre>
+     *
+     * <code>.tensorflow.SessionMetadata session_metadata = 11;</code>
+     */
+    org.tensorflow.proto.framework.SessionMetadataOrBuilder getSessionMetadataOrBuilder();
+
+    /**
+     * <pre>
+     * If true, the session may treat the graph as being static for optimization
+     * purposes.
+     * If this option is set to true when a session is created, the full
+     * GraphDef must be passed in a single call to Session::Create(), and
+     * Session::Extend() may not be supported.
+     * </pre>
+     *
+     * <code>bool optimize_for_static_graph = 12;</code>
+     */
+    boolean getOptimizeForStaticGraph();
+
+    /**
+     * <pre>
+     * Whether to enable the MLIR-based TF-&gt;XLA bridge.
+     * This is a replacement to the existing bridge, and not ready for
+     * production usage yet.
+     * If this option is set to true when a session is created, MLIR is used to
+     * perform the set of graph transformations to put the graph in a form that
+     * can be executed with delegation of some computations to an accelerator.
+     * This builds on the model of XLA where a subset of the graph is
+     * encapsulated and attached to a "compile" operation, whose result is fed
+     * to an "execute" operation. The kernel for these operations is responsible
+     * to lower the encapsulated graph to a particular device.
+     * </pre>
+     *
+     * <code>bool enable_mlir_bridge = 13;</code>
+     */
+    boolean getEnableMlirBridge();
+
+    /**
+     * <pre>
+     * Whether to enable the MLIR-based Graph optimizations.
+     * This will become a part of standard Tensorflow graph optimization
+     * pipeline, currently this is only used for gradual migration and testing
+     * new passes that are replacing existing optimizations in Grappler.
+     * </pre>
+     *
+     * <code>bool enable_mlir_graph_optimization = 16;</code>
+     */
+    boolean getEnableMlirGraphOptimization();
+
+    /**
+     * <pre>
+     * If true, the session will not store an additional copy of the graph for
+     * each subgraph.
+     * If this option is set to true when a session is created, the
+     * `RunOptions.output_partition_graphs` options must not be set.
+     * </pre>
+     *
+     * <code>bool disable_output_partition_graphs = 14;</code>
+     */
+    boolean getDisableOutputPartitionGraphs();
+
+    /**
+     * <pre>
+     * Minimum number of batches run through the XLA graph before XLA fusion
+     * autotuner is enabled. Default value of zero disables the autotuner.
+     * The XLA fusion autotuner can improve performance by executing a heuristic
+     * search on the compiler parameters.
+     * </pre>
+     *
+     * <code>int64 xla_fusion_autotuner_thresh = 15;</code>
+     */
+    long getXlaFusionAutotunerThresh();
+  }
+  /**
+   * <pre>
+   * Everything inside Experimental is subject to change and is not subject
+   * to API stability guarantees in
+   * https://www.tensorflow.org/guide/version_compat.
+   * </pre>
+   *
+   * Protobuf type {@code tensorflow.ConfigProto.Experimental}
+   */
+  public  static final class Experimental extends
+      com.google.protobuf.GeneratedMessageV3 implements
+      // @@protoc_insertion_point(message_implements:tensorflow.ConfigProto.Experimental)
+      ExperimentalOrBuilder {
+  private static final long serialVersionUID = 0L;
+    // Use Experimental.newBuilder() to construct.
+    private Experimental(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+      super(builder);
+    }
+    private Experimental() {
+      collectiveGroupLeader_ = "";
+      executorType_ = "";
+    }
+
+    @java.lang.Override
+    @SuppressWarnings({"unused"})
+    protected java.lang.Object newInstance(
+        UnusedPrivateParameter unused) {
+      return new Experimental();
+    }
+
+    @java.lang.Override
+    public final com.google.protobuf.UnknownFieldSet
+    getUnknownFields() {
+      return this.unknownFields;
+    }
+    private Experimental(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      this();
+      if (extensionRegistry == null) {
+        throw new java.lang.NullPointerException();
+      }
+      com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+          com.google.protobuf.UnknownFieldSet.newBuilder();
+      try {
+        boolean done = false;
+        while (!done) {
+          int tag = input.readTag();
+          switch (tag) {
+            case 0:
+              done = true;
+              break;
+            case 10: {
+              java.lang.String s = input.readStringRequireUtf8();
+
+              collectiveGroupLeader_ = s;
+              break;
+            }
+            case 26: {
+              java.lang.String s = input.readStringRequireUtf8();
+
+              executorType_ = s;
+              break;
+            }
+            case 32: {
+
+              recvBufMaxChunk_ = input.readInt32();
+              break;
+            }
+            case 40: {
+
+              useNumaAffinity_ = input.readBool();
+              break;
+            }
+            case 48: {
+
+              collectiveDeterministicSequentialExecution_ = input.readBool();
+              break;
+            }
+            case 56: {
+
+              collectiveNccl_ = input.readBool();
+              break;
+            }
+            case 64: {
+
+              shareSessionStateInClusterspecPropagation_ = input.readBool();
+              break;
+            }
+            case 72: {
+
+              disableThreadSpinning_ = input.readBool();
+              break;
+            }
+            case 80: {
+
+              shareClusterDevicesInSession_ = input.readBool();
+              break;
+            }
+            case 90: {
+              org.tensorflow.proto.framework.SessionMetadata.Builder subBuilder = null;
+              if (sessionMetadata_ != null) {
+                subBuilder = sessionMetadata_.toBuilder();
+              }
+              sessionMetadata_ = input.readMessage(org.tensorflow.proto.framework.SessionMetadata.parser(), extensionRegistry);
+              if (subBuilder != null) {
+                subBuilder.mergeFrom(sessionMetadata_);
+                sessionMetadata_ = subBuilder.buildPartial();
+              }
+
+              break;
+            }
+            case 96: {
+
+              optimizeForStaticGraph_ = input.readBool();
+              break;
+            }
+            case 104: {
+
+              enableMlirBridge_ = input.readBool();
+              break;
+            }
+            case 112: {
+
+              disableOutputPartitionGraphs_ = input.readBool();
+              break;
+            }
+            case 120: {
+
+              xlaFusionAutotunerThresh_ = input.readInt64();
+              break;
+            }
+            case 128: {
+
+              enableMlirGraphOptimization_ = input.readBool();
+              break;
+            }
+            default: {
+              if (!parseUnknownField(
+                  input, unknownFields, extensionRegistry, tag)) {
+                done = true;
+              }
+              break;
+            }
+          }
+        }
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        throw e.setUnfinishedMessage(this);
+      } catch (java.io.IOException e) {
+        throw new com.google.protobuf.InvalidProtocolBufferException(
+            e).setUnfinishedMessage(this);
+      } finally {
+        this.unknownFields = unknownFields.build();
+        makeExtensionsImmutable();
+      }
+    }
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_ConfigProto_Experimental_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_ConfigProto_Experimental_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.ConfigProto.Experimental.class, org.tensorflow.proto.framework.ConfigProto.Experimental.Builder.class);
+    }
+
+    public static final int COLLECTIVE_GROUP_LEADER_FIELD_NUMBER = 1;
+    private volatile java.lang.Object collectiveGroupLeader_;
+    /**
+     * <pre>
+     * Task name for group resolution.
+     * </pre>
+     *
+     * <code>string collective_group_leader = 1;</code>
+     */
+    public java.lang.String getCollectiveGroupLeader() {
+      java.lang.Object ref = collectiveGroupLeader_;
+      if (ref instanceof java.lang.String) {
+        return (java.lang.String) ref;
+      } else {
+        com.google.protobuf.ByteString bs = 
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        collectiveGroupLeader_ = s;
+        return s;
+      }
+    }
+    /**
+     * <pre>
+     * Task name for group resolution.
+     * </pre>
+     *
+     * <code>string collective_group_leader = 1;</code>
+     */
+    public com.google.protobuf.ByteString
+        getCollectiveGroupLeaderBytes() {
+      java.lang.Object ref = collectiveGroupLeader_;
+      if (ref instanceof java.lang.String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        collectiveGroupLeader_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+
+    public static final int EXECUTOR_TYPE_FIELD_NUMBER = 3;
+    private volatile java.lang.Object executorType_;
+    /**
+     * <pre>
+     * Which executor to use, the default executor will be used
+     * if it is an empty string or "DEFAULT"
+     * </pre>
+     *
+     * <code>string executor_type = 3;</code>
+     */
+    public java.lang.String getExecutorType() {
+      java.lang.Object ref = executorType_;
+      if (ref instanceof java.lang.String) {
+        return (java.lang.String) ref;
+      } else {
+        com.google.protobuf.ByteString bs = 
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        executorType_ = s;
+        return s;
+      }
+    }
+    /**
+     * <pre>
+     * Which executor to use, the default executor will be used
+     * if it is an empty string or "DEFAULT"
+     * </pre>
+     *
+     * <code>string executor_type = 3;</code>
+     */
+    public com.google.protobuf.ByteString
+        getExecutorTypeBytes() {
+      java.lang.Object ref = executorType_;
+      if (ref instanceof java.lang.String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        executorType_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+
+    public static final int RECV_BUF_MAX_CHUNK_FIELD_NUMBER = 4;
+    private int recvBufMaxChunk_;
+    /**
+     * <pre>
+     * Guidance to formatting of large RecvBuf fields for transfer.
+     * Any positive value sets the max chunk size.  0 defaults to 4096.
+     * Any negative value indicates no max, i.e. one chunk only.
+     * </pre>
+     *
+     * <code>int32 recv_buf_max_chunk = 4;</code>
+     */
+    public int getRecvBufMaxChunk() {
+      return recvBufMaxChunk_;
+    }
+
+    public static final int USE_NUMA_AFFINITY_FIELD_NUMBER = 5;
+    private boolean useNumaAffinity_;
+    /**
+     * <pre>
+     * If true, and supported by the platform, the runtime will attempt to
+     * use NUMA affinity where applicable.  One consequence will be the
+     * existence of as many CPU devices as there are available NUMA nodes.
+     * </pre>
+     *
+     * <code>bool use_numa_affinity = 5;</code>
+     */
+    public boolean getUseNumaAffinity() {
+      return useNumaAffinity_;
+    }
+
+    public static final int COLLECTIVE_DETERMINISTIC_SEQUENTIAL_EXECUTION_FIELD_NUMBER = 6;
+    private boolean collectiveDeterministicSequentialExecution_;
+    /**
+     * <pre>
+     * If true, make collective op execution order sequential and deterministic
+     * for potentially concurrent collective instances.
+     * </pre>
+     *
+     * <code>bool collective_deterministic_sequential_execution = 6;</code>
+     */
+    public boolean getCollectiveDeterministicSequentialExecution() {
+      return collectiveDeterministicSequentialExecution_;
+    }
+
+    public static final int COLLECTIVE_NCCL_FIELD_NUMBER = 7;
+    private boolean collectiveNccl_;
+    /**
+     * <pre>
+     * If true, use NCCL for CollectiveOps.  This feature is highly
+     * experimental.
+     * </pre>
+     *
+     * <code>bool collective_nccl = 7;</code>
+     */
+    public boolean getCollectiveNccl() {
+      return collectiveNccl_;
+    }
+
+    public static final int SHARE_SESSION_STATE_IN_CLUSTERSPEC_PROPAGATION_FIELD_NUMBER = 8;
+    private boolean shareSessionStateInClusterspecPropagation_;
+    /**
+     * <pre>
+     * In the following, session state means the value of a variable, elements
+     * in a hash table, or any other resource, accessible by worker sessions
+     * held by a TF server.
+     * When ClusterSpec propagation is enabled, the value of
+     * isolate_session_state is ignored when deciding whether to share session
+     * states in a TF server (for backwards compatibility reasons).
+     * - If share_session_state_in_clusterspec_propagation is true, the session
+     * states are shared.
+     * - If share_session_state_in_clusterspec_propagation is false, session
+     * states are isolated.
+     * When clusterspec propagation is not used, the value of
+     * share_session_state_in_clusterspec_propagation is ignored when deciding
+     * whether to share session states in a TF server.
+     * - If isolate_session_state is true, session states are isolated.
+     * - If isolate_session_state is false, session states are shared.
+     * TODO(b/129330037): Add a single API that consistently treats
+     * isolate_session_state and ClusterSpec propagation.
+     * </pre>
+     *
+     * <code>bool share_session_state_in_clusterspec_propagation = 8;</code>
+     */
+    public boolean getShareSessionStateInClusterspecPropagation() {
+      return shareSessionStateInClusterspecPropagation_;
+    }
+
+    public static final int DISABLE_THREAD_SPINNING_FIELD_NUMBER = 9;
+    private boolean disableThreadSpinning_;
+    /**
+     * <pre>
+     * If using a direct session, disable spinning while waiting for work in
+     * the thread pool. This may result in higher latency for completing ops,
+     * but in the case where there is a lot of spinning may result in lower
+     * CPU usage.
+     * </pre>
+     *
+     * <code>bool disable_thread_spinning = 9;</code>
+     */
+    public boolean getDisableThreadSpinning() {
+      return disableThreadSpinning_;
+    }
+
+    public static final int SHARE_CLUSTER_DEVICES_IN_SESSION_FIELD_NUMBER = 10;
+    private boolean shareClusterDevicesInSession_;
+    /**
+     * <pre>
+     * This was promoted to a non-experimental API. Please use
+     * ConfigProto.share_cluster_devices_in_session instead.
+     * </pre>
+     *
+     * <code>bool share_cluster_devices_in_session = 10;</code>
+     */
+    public boolean getShareClusterDevicesInSession() {
+      return shareClusterDevicesInSession_;
+    }
+
+    public static final int SESSION_METADATA_FIELD_NUMBER = 11;
+    private org.tensorflow.proto.framework.SessionMetadata sessionMetadata_;
+    /**
+     * <pre>
+     * Metadata about the session.
+     * If set, this can be used by the runtime and the Ops for debugging,
+     * monitoring, etc.
+     * NOTE: This is currently used and propagated only by the direct session.
+     * </pre>
+     *
+     * <code>.tensorflow.SessionMetadata session_metadata = 11;</code>
+     */
+    public boolean hasSessionMetadata() {
+      return sessionMetadata_ != null;
+    }
+    /**
+     * <pre>
+     * Metadata about the session.
+     * If set, this can be used by the runtime and the Ops for debugging,
+     * monitoring, etc.
+     * NOTE: This is currently used and propagated only by the direct session.
+     * </pre>
+     *
+     * <code>.tensorflow.SessionMetadata session_metadata = 11;</code>
+     */
+    public org.tensorflow.proto.framework.SessionMetadata getSessionMetadata() {
+      return sessionMetadata_ == null ? org.tensorflow.proto.framework.SessionMetadata.getDefaultInstance() : sessionMetadata_;
+    }
+    /**
+     * <pre>
+     * Metadata about the session.
+     * If set, this can be used by the runtime and the Ops for debugging,
+     * monitoring, etc.
+     * NOTE: This is currently used and propagated only by the direct session.
+     * </pre>
+     *
+     * <code>.tensorflow.SessionMetadata session_metadata = 11;</code>
+     */
+    public org.tensorflow.proto.framework.SessionMetadataOrBuilder getSessionMetadataOrBuilder() {
+      return getSessionMetadata();
+    }
+
+    public static final int OPTIMIZE_FOR_STATIC_GRAPH_FIELD_NUMBER = 12;
+    private boolean optimizeForStaticGraph_;
+    /**
+     * <pre>
+     * If true, the session may treat the graph as being static for optimization
+     * purposes.
+     * If this option is set to true when a session is created, the full
+     * GraphDef must be passed in a single call to Session::Create(), and
+     * Session::Extend() may not be supported.
+     * </pre>
+     *
+     * <code>bool optimize_for_static_graph = 12;</code>
+     */
+    public boolean getOptimizeForStaticGraph() {
+      return optimizeForStaticGraph_;
+    }
+
+    public static final int ENABLE_MLIR_BRIDGE_FIELD_NUMBER = 13;
+    private boolean enableMlirBridge_;
+    /**
+     * <pre>
+     * Whether to enable the MLIR-based TF-&gt;XLA bridge.
+     * This is a replacement to the existing bridge, and not ready for
+     * production usage yet.
+     * If this option is set to true when a session is created, MLIR is used to
+     * perform the set of graph transformations to put the graph in a form that
+     * can be executed with delegation of some computations to an accelerator.
+     * This builds on the model of XLA where a subset of the graph is
+     * encapsulated and attached to a "compile" operation, whose result is fed
+     * to an "execute" operation. The kernel for these operations is responsible
+     * to lower the encapsulated graph to a particular device.
+     * </pre>
+     *
+     * <code>bool enable_mlir_bridge = 13;</code>
+     */
+    public boolean getEnableMlirBridge() {
+      return enableMlirBridge_;
+    }
+
+    public static final int ENABLE_MLIR_GRAPH_OPTIMIZATION_FIELD_NUMBER = 16;
+    private boolean enableMlirGraphOptimization_;
+    /**
+     * <pre>
+     * Whether to enable the MLIR-based Graph optimizations.
+     * This will become a part of standard Tensorflow graph optimization
+     * pipeline, currently this is only used for gradual migration and testing
+     * new passes that are replacing existing optimizations in Grappler.
+     * </pre>
+     *
+     * <code>bool enable_mlir_graph_optimization = 16;</code>
+     */
+    public boolean getEnableMlirGraphOptimization() {
+      return enableMlirGraphOptimization_;
+    }
+
+    public static final int DISABLE_OUTPUT_PARTITION_GRAPHS_FIELD_NUMBER = 14;
+    private boolean disableOutputPartitionGraphs_;
+    /**
+     * <pre>
+     * If true, the session will not store an additional copy of the graph for
+     * each subgraph.
+     * If this option is set to true when a session is created, the
+     * `RunOptions.output_partition_graphs` options must not be set.
+     * </pre>
+     *
+     * <code>bool disable_output_partition_graphs = 14;</code>
+     */
+    public boolean getDisableOutputPartitionGraphs() {
+      return disableOutputPartitionGraphs_;
+    }
+
+    public static final int XLA_FUSION_AUTOTUNER_THRESH_FIELD_NUMBER = 15;
+    private long xlaFusionAutotunerThresh_;
+    /**
+     * <pre>
+     * Minimum number of batches run through the XLA graph before XLA fusion
+     * autotuner is enabled. Default value of zero disables the autotuner.
+     * The XLA fusion autotuner can improve performance by executing a heuristic
+     * search on the compiler parameters.
+     * </pre>
+     *
+     * <code>int64 xla_fusion_autotuner_thresh = 15;</code>
+     */
+    public long getXlaFusionAutotunerThresh() {
+      return xlaFusionAutotunerThresh_;
+    }
+
+    private byte memoizedIsInitialized = -1;
+    @java.lang.Override
+    public final boolean isInitialized() {
+      byte isInitialized = memoizedIsInitialized;
+      if (isInitialized == 1) return true;
+      if (isInitialized == 0) return false;
+
+      memoizedIsInitialized = 1;
+      return true;
+    }
+
+    @java.lang.Override
+    public void writeTo(com.google.protobuf.CodedOutputStream output)
+                        throws java.io.IOException {
+      if (!getCollectiveGroupLeaderBytes().isEmpty()) {
+        com.google.protobuf.GeneratedMessageV3.writeString(output, 1, collectiveGroupLeader_);
+      }
+      if (!getExecutorTypeBytes().isEmpty()) {
+        com.google.protobuf.GeneratedMessageV3.writeString(output, 3, executorType_);
+      }
+      if (recvBufMaxChunk_ != 0) {
+        output.writeInt32(4, recvBufMaxChunk_);
+      }
+      if (useNumaAffinity_ != false) {
+        output.writeBool(5, useNumaAffinity_);
+      }
+      if (collectiveDeterministicSequentialExecution_ != false) {
+        output.writeBool(6, collectiveDeterministicSequentialExecution_);
+      }
+      if (collectiveNccl_ != false) {
+        output.writeBool(7, collectiveNccl_);
+      }
+      if (shareSessionStateInClusterspecPropagation_ != false) {
+        output.writeBool(8, shareSessionStateInClusterspecPropagation_);
+      }
+      if (disableThreadSpinning_ != false) {
+        output.writeBool(9, disableThreadSpinning_);
+      }
+      if (shareClusterDevicesInSession_ != false) {
+        output.writeBool(10, shareClusterDevicesInSession_);
+      }
+      if (sessionMetadata_ != null) {
+        output.writeMessage(11, getSessionMetadata());
+      }
+      if (optimizeForStaticGraph_ != false) {
+        output.writeBool(12, optimizeForStaticGraph_);
+      }
+      if (enableMlirBridge_ != false) {
+        output.writeBool(13, enableMlirBridge_);
+      }
+      if (disableOutputPartitionGraphs_ != false) {
+        output.writeBool(14, disableOutputPartitionGraphs_);
+      }
+      if (xlaFusionAutotunerThresh_ != 0L) {
+        output.writeInt64(15, xlaFusionAutotunerThresh_);
+      }
+      if (enableMlirGraphOptimization_ != false) {
+        output.writeBool(16, enableMlirGraphOptimization_);
+      }
+      unknownFields.writeTo(output);
+    }
+
+    @java.lang.Override
+    public int getSerializedSize() {
+      int size = memoizedSize;
+      if (size != -1) return size;
+
+      size = 0;
+      if (!getCollectiveGroupLeaderBytes().isEmpty()) {
+        size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, collectiveGroupLeader_);
+      }
+      if (!getExecutorTypeBytes().isEmpty()) {
+        size += com.google.protobuf.GeneratedMessageV3.computeStringSize(3, executorType_);
+      }
+      if (recvBufMaxChunk_ != 0) {
+        size += com.google.protobuf.CodedOutputStream
+          .computeInt32Size(4, recvBufMaxChunk_);
+      }
+      if (useNumaAffinity_ != false) {
+        size += com.google.protobuf.CodedOutputStream
+          .computeBoolSize(5, useNumaAffinity_);
+      }
+      if (collectiveDeterministicSequentialExecution_ != false) {
+        size += com.google.protobuf.CodedOutputStream
+          .computeBoolSize(6, collectiveDeterministicSequentialExecution_);
+      }
+      if (collectiveNccl_ != false) {
+        size += com.google.protobuf.CodedOutputStream
+          .computeBoolSize(7, collectiveNccl_);
+      }
+      if (shareSessionStateInClusterspecPropagation_ != false) {
+        size += com.google.protobuf.CodedOutputStream
+          .computeBoolSize(8, shareSessionStateInClusterspecPropagation_);
+      }
+      if (disableThreadSpinning_ != false) {
+        size += com.google.protobuf.CodedOutputStream
+          .computeBoolSize(9, disableThreadSpinning_);
+      }
+      if (shareClusterDevicesInSession_ != false) {
+        size += com.google.protobuf.CodedOutputStream
+          .computeBoolSize(10, shareClusterDevicesInSession_);
+      }
+      if (sessionMetadata_ != null) {
+        size += com.google.protobuf.CodedOutputStream
+          .computeMessageSize(11, getSessionMetadata());
+      }
+      if (optimizeForStaticGraph_ != false) {
+        size += com.google.protobuf.CodedOutputStream
+          .computeBoolSize(12, optimizeForStaticGraph_);
+      }
+      if (enableMlirBridge_ != false) {
+        size += com.google.protobuf.CodedOutputStream
+          .computeBoolSize(13, enableMlirBridge_);
+      }
+      if (disableOutputPartitionGraphs_ != false) {
+        size += com.google.protobuf.CodedOutputStream
+          .computeBoolSize(14, disableOutputPartitionGraphs_);
+      }
+      if (xlaFusionAutotunerThresh_ != 0L) {
+        size += com.google.protobuf.CodedOutputStream
+          .computeInt64Size(15, xlaFusionAutotunerThresh_);
+      }
+      if (enableMlirGraphOptimization_ != false) {
+        size += com.google.protobuf.CodedOutputStream
+          .computeBoolSize(16, enableMlirGraphOptimization_);
+      }
+      size += unknownFields.getSerializedSize();
+      memoizedSize = size;
+      return size;
+    }
+
+    @java.lang.Override
+    public boolean equals(final java.lang.Object obj) {
+      if (obj == this) {
+       return true;
+      }
+      if (!(obj instanceof org.tensorflow.proto.framework.ConfigProto.Experimental)) {
+        return super.equals(obj);
+      }
+      org.tensorflow.proto.framework.ConfigProto.Experimental other = (org.tensorflow.proto.framework.ConfigProto.Experimental) obj;
+
+      if (!getCollectiveGroupLeader()
+          .equals(other.getCollectiveGroupLeader())) return false;
+      if (!getExecutorType()
+          .equals(other.getExecutorType())) return false;
+      if (getRecvBufMaxChunk()
+          != other.getRecvBufMaxChunk()) return false;
+      if (getUseNumaAffinity()
+          != other.getUseNumaAffinity()) return false;
+      if (getCollectiveDeterministicSequentialExecution()
+          != other.getCollectiveDeterministicSequentialExecution()) return false;
+      if (getCollectiveNccl()
+          != other.getCollectiveNccl()) return false;
+      if (getShareSessionStateInClusterspecPropagation()
+          != other.getShareSessionStateInClusterspecPropagation()) return false;
+      if (getDisableThreadSpinning()
+          != other.getDisableThreadSpinning()) return false;
+      if (getShareClusterDevicesInSession()
+          != other.getShareClusterDevicesInSession()) return false;
+      if (hasSessionMetadata() != other.hasSessionMetadata()) return false;
+      if (hasSessionMetadata()) {
+        if (!getSessionMetadata()
+            .equals(other.getSessionMetadata())) return false;
+      }
+      if (getOptimizeForStaticGraph()
+          != other.getOptimizeForStaticGraph()) return false;
+      if (getEnableMlirBridge()
+          != other.getEnableMlirBridge()) return false;
+      if (getEnableMlirGraphOptimization()
+          != other.getEnableMlirGraphOptimization()) return false;
+      if (getDisableOutputPartitionGraphs()
+          != other.getDisableOutputPartitionGraphs()) return false;
+      if (getXlaFusionAutotunerThresh()
+          != other.getXlaFusionAutotunerThresh()) return false;
+      if (!unknownFields.equals(other.unknownFields)) return false;
+      return true;
+    }
+
+    @java.lang.Override
+    public int hashCode() {
+      if (memoizedHashCode != 0) {
+        return memoizedHashCode;
+      }
+      int hash = 41;
+      hash = (19 * hash) + getDescriptor().hashCode();
+      hash = (37 * hash) + COLLECTIVE_GROUP_LEADER_FIELD_NUMBER;
+      hash = (53 * hash) + getCollectiveGroupLeader().hashCode();
+      hash = (37 * hash) + EXECUTOR_TYPE_FIELD_NUMBER;
+      hash = (53 * hash) + getExecutorType().hashCode();
+      hash = (37 * hash) + RECV_BUF_MAX_CHUNK_FIELD_NUMBER;
+      hash = (53 * hash) + getRecvBufMaxChunk();
+      hash = (37 * hash) + USE_NUMA_AFFINITY_FIELD_NUMBER;
+      hash = (53 * hash) + com.google.protobuf.Internal.hashBoolean(
+          getUseNumaAffinity());
+      hash = (37 * hash) + COLLECTIVE_DETERMINISTIC_SEQUENTIAL_EXECUTION_FIELD_NUMBER;
+      hash = (53 * hash) + com.google.protobuf.Internal.hashBoolean(
+          getCollectiveDeterministicSequentialExecution());
+      hash = (37 * hash) + COLLECTIVE_NCCL_FIELD_NUMBER;
+      hash = (53 * hash) + com.google.protobuf.Internal.hashBoolean(
+          getCollectiveNccl());
+      hash = (37 * hash) + SHARE_SESSION_STATE_IN_CLUSTERSPEC_PROPAGATION_FIELD_NUMBER;
+      hash = (53 * hash) + com.google.protobuf.Internal.hashBoolean(
+          getShareSessionStateInClusterspecPropagation());
+      hash = (37 * hash) + DISABLE_THREAD_SPINNING_FIELD_NUMBER;
+      hash = (53 * hash) + com.google.protobuf.Internal.hashBoolean(
+          getDisableThreadSpinning());
+      hash = (37 * hash) + SHARE_CLUSTER_DEVICES_IN_SESSION_FIELD_NUMBER;
+      hash = (53 * hash) + com.google.protobuf.Internal.hashBoolean(
+          getShareClusterDevicesInSession());
+      if (hasSessionMetadata()) {
+        hash = (37 * hash) + SESSION_METADATA_FIELD_NUMBER;
+        hash = (53 * hash) + getSessionMetadata().hashCode();
+      }
+      hash = (37 * hash) + OPTIMIZE_FOR_STATIC_GRAPH_FIELD_NUMBER;
+      hash = (53 * hash) + com.google.protobuf.Internal.hashBoolean(
+          getOptimizeForStaticGraph());
+      hash = (37 * hash) + ENABLE_MLIR_BRIDGE_FIELD_NUMBER;
+      hash = (53 * hash) + com.google.protobuf.Internal.hashBoolean(
+          getEnableMlirBridge());
+      hash = (37 * hash) + ENABLE_MLIR_GRAPH_OPTIMIZATION_FIELD_NUMBER;
+      hash = (53 * hash) + com.google.protobuf.Internal.hashBoolean(
+          getEnableMlirGraphOptimization());
+      hash = (37 * hash) + DISABLE_OUTPUT_PARTITION_GRAPHS_FIELD_NUMBER;
+      hash = (53 * hash) + com.google.protobuf.Internal.hashBoolean(
+          getDisableOutputPartitionGraphs());
+      hash = (37 * hash) + XLA_FUSION_AUTOTUNER_THRESH_FIELD_NUMBER;
+      hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+          getXlaFusionAutotunerThresh());
+      hash = (29 * hash) + unknownFields.hashCode();
+      memoizedHashCode = hash;
+      return hash;
+    }
+
+    public static org.tensorflow.proto.framework.ConfigProto.Experimental parseFrom(
+        java.nio.ByteBuffer data)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data);
+    }
+    public static org.tensorflow.proto.framework.ConfigProto.Experimental parseFrom(
+        java.nio.ByteBuffer data,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.ConfigProto.Experimental parseFrom(
+        com.google.protobuf.ByteString data)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data);
+    }
+    public static org.tensorflow.proto.framework.ConfigProto.Experimental parseFrom(
+        com.google.protobuf.ByteString data,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.ConfigProto.Experimental parseFrom(byte[] data)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data);
+    }
+    public static org.tensorflow.proto.framework.ConfigProto.Experimental parseFrom(
+        byte[] data,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.ConfigProto.Experimental parseFrom(java.io.InputStream input)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input);
+    }
+    public static org.tensorflow.proto.framework.ConfigProto.Experimental parseFrom(
+        java.io.InputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.ConfigProto.Experimental parseDelimitedFrom(java.io.InputStream input)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseDelimitedWithIOException(PARSER, input);
+    }
+    public static org.tensorflow.proto.framework.ConfigProto.Experimental parseDelimitedFrom(
+        java.io.InputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.ConfigProto.Experimental parseFrom(
+        com.google.protobuf.CodedInputStream input)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input);
+    }
+    public static org.tensorflow.proto.framework.ConfigProto.Experimental parseFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input, extensionRegistry);
+    }
+
+    @java.lang.Override
+    public Builder newBuilderForType() { return newBuilder(); }
+    public static Builder newBuilder() {
+      return DEFAULT_INSTANCE.toBuilder();
+    }
+    public static Builder newBuilder(org.tensorflow.proto.framework.ConfigProto.Experimental prototype) {
+      return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+    }
+    @java.lang.Override
+    public Builder toBuilder() {
+      return this == DEFAULT_INSTANCE
+          ? new Builder() : new Builder().mergeFrom(this);
+    }
+
+    @java.lang.Override
+    protected Builder newBuilderForType(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      Builder builder = new Builder(parent);
+      return builder;
+    }
+    /**
+     * <pre>
+     * Everything inside Experimental is subject to change and is not subject
+     * to API stability guarantees in
+     * https://www.tensorflow.org/guide/version_compat.
+     * </pre>
+     *
+     * Protobuf type {@code tensorflow.ConfigProto.Experimental}
+     */
+    public static final class Builder extends
+        com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+        // @@protoc_insertion_point(builder_implements:tensorflow.ConfigProto.Experimental)
+        org.tensorflow.proto.framework.ConfigProto.ExperimentalOrBuilder {
+      public static final com.google.protobuf.Descriptors.Descriptor
+          getDescriptor() {
+        return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_ConfigProto_Experimental_descriptor;
+      }
+
+      @java.lang.Override
+      protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+          internalGetFieldAccessorTable() {
+        return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_ConfigProto_Experimental_fieldAccessorTable
+            .ensureFieldAccessorsInitialized(
+                org.tensorflow.proto.framework.ConfigProto.Experimental.class, org.tensorflow.proto.framework.ConfigProto.Experimental.Builder.class);
+      }
+
+      // Construct using org.tensorflow.proto.framework.ConfigProto.Experimental.newBuilder()
+      private Builder() {
+        maybeForceBuilderInitialization();
+      }
+
+      private Builder(
+          com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+        super(parent);
+        maybeForceBuilderInitialization();
+      }
+      private void maybeForceBuilderInitialization() {
+        if (com.google.protobuf.GeneratedMessageV3
+                .alwaysUseFieldBuilders) {
+        }
+      }
+      @java.lang.Override
+      public Builder clear() {
+        super.clear();
+        collectiveGroupLeader_ = "";
+
+        executorType_ = "";
+
+        recvBufMaxChunk_ = 0;
+
+        useNumaAffinity_ = false;
+
+        collectiveDeterministicSequentialExecution_ = false;
+
+        collectiveNccl_ = false;
+
+        shareSessionStateInClusterspecPropagation_ = false;
+
+        disableThreadSpinning_ = false;
+
+        shareClusterDevicesInSession_ = false;
+
+        if (sessionMetadataBuilder_ == null) {
+          sessionMetadata_ = null;
+        } else {
+          sessionMetadata_ = null;
+          sessionMetadataBuilder_ = null;
+        }
+        optimizeForStaticGraph_ = false;
+
+        enableMlirBridge_ = false;
+
+        enableMlirGraphOptimization_ = false;
+
+        disableOutputPartitionGraphs_ = false;
+
+        xlaFusionAutotunerThresh_ = 0L;
+
+        return this;
+      }
+
+      @java.lang.Override
+      public com.google.protobuf.Descriptors.Descriptor
+          getDescriptorForType() {
+        return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_ConfigProto_Experimental_descriptor;
+      }
+
+      @java.lang.Override
+      public org.tensorflow.proto.framework.ConfigProto.Experimental getDefaultInstanceForType() {
+        return org.tensorflow.proto.framework.ConfigProto.Experimental.getDefaultInstance();
+      }
+
+      @java.lang.Override
+      public org.tensorflow.proto.framework.ConfigProto.Experimental build() {
+        org.tensorflow.proto.framework.ConfigProto.Experimental result = buildPartial();
+        if (!result.isInitialized()) {
+          throw newUninitializedMessageException(result);
+        }
+        return result;
+      }
+
+      @java.lang.Override
+      public org.tensorflow.proto.framework.ConfigProto.Experimental buildPartial() {
+        org.tensorflow.proto.framework.ConfigProto.Experimental result = new org.tensorflow.proto.framework.ConfigProto.Experimental(this);
+        result.collectiveGroupLeader_ = collectiveGroupLeader_;
+        result.executorType_ = executorType_;
+        result.recvBufMaxChunk_ = recvBufMaxChunk_;
+        result.useNumaAffinity_ = useNumaAffinity_;
+        result.collectiveDeterministicSequentialExecution_ = collectiveDeterministicSequentialExecution_;
+        result.collectiveNccl_ = collectiveNccl_;
+        result.shareSessionStateInClusterspecPropagation_ = shareSessionStateInClusterspecPropagation_;
+        result.disableThreadSpinning_ = disableThreadSpinning_;
+        result.shareClusterDevicesInSession_ = shareClusterDevicesInSession_;
+        if (sessionMetadataBuilder_ == null) {
+          result.sessionMetadata_ = sessionMetadata_;
+        } else {
+          result.sessionMetadata_ = sessionMetadataBuilder_.build();
+        }
+        result.optimizeForStaticGraph_ = optimizeForStaticGraph_;
+        result.enableMlirBridge_ = enableMlirBridge_;
+        result.enableMlirGraphOptimization_ = enableMlirGraphOptimization_;
+        result.disableOutputPartitionGraphs_ = disableOutputPartitionGraphs_;
+        result.xlaFusionAutotunerThresh_ = xlaFusionAutotunerThresh_;
+        onBuilt();
+        return result;
+      }
+
+      @java.lang.Override
+      public Builder clone() {
+        return super.clone();
+      }
+      @java.lang.Override
+      public Builder setField(
+          com.google.protobuf.Descriptors.FieldDescriptor field,
+          java.lang.Object value) {
+        return super.setField(field, value);
+      }
+      @java.lang.Override
+      public Builder clearField(
+          com.google.protobuf.Descriptors.FieldDescriptor field) {
+        return super.clearField(field);
+      }
+      @java.lang.Override
+      public Builder clearOneof(
+          com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+        return super.clearOneof(oneof);
+      }
+      @java.lang.Override
+      public Builder setRepeatedField(
+          com.google.protobuf.Descriptors.FieldDescriptor field,
+          int index, java.lang.Object value) {
+        return super.setRepeatedField(field, index, value);
+      }
+      @java.lang.Override
+      public Builder addRepeatedField(
+          com.google.protobuf.Descriptors.FieldDescriptor field,
+          java.lang.Object value) {
+        return super.addRepeatedField(field, value);
+      }
+      @java.lang.Override
+      public Builder mergeFrom(com.google.protobuf.Message other) {
+        if (other instanceof org.tensorflow.proto.framework.ConfigProto.Experimental) {
+          return mergeFrom((org.tensorflow.proto.framework.ConfigProto.Experimental)other);
+        } else {
+          super.mergeFrom(other);
+          return this;
+        }
+      }
+
+      public Builder mergeFrom(org.tensorflow.proto.framework.ConfigProto.Experimental other) {
+        if (other == org.tensorflow.proto.framework.ConfigProto.Experimental.getDefaultInstance()) return this;
+        if (!other.getCollectiveGroupLeader().isEmpty()) {
+          collectiveGroupLeader_ = other.collectiveGroupLeader_;
+          onChanged();
+        }
+        if (!other.getExecutorType().isEmpty()) {
+          executorType_ = other.executorType_;
+          onChanged();
+        }
+        if (other.getRecvBufMaxChunk() != 0) {
+          setRecvBufMaxChunk(other.getRecvBufMaxChunk());
+        }
+        if (other.getUseNumaAffinity() != false) {
+          setUseNumaAffinity(other.getUseNumaAffinity());
+        }
+        if (other.getCollectiveDeterministicSequentialExecution() != false) {
+          setCollectiveDeterministicSequentialExecution(other.getCollectiveDeterministicSequentialExecution());
+        }
+        if (other.getCollectiveNccl() != false) {
+          setCollectiveNccl(other.getCollectiveNccl());
+        }
+        if (other.getShareSessionStateInClusterspecPropagation() != false) {
+          setShareSessionStateInClusterspecPropagation(other.getShareSessionStateInClusterspecPropagation());
+        }
+        if (other.getDisableThreadSpinning() != false) {
+          setDisableThreadSpinning(other.getDisableThreadSpinning());
+        }
+        if (other.getShareClusterDevicesInSession() != false) {
+          setShareClusterDevicesInSession(other.getShareClusterDevicesInSession());
+        }
+        if (other.hasSessionMetadata()) {
+          mergeSessionMetadata(other.getSessionMetadata());
+        }
+        if (other.getOptimizeForStaticGraph() != false) {
+          setOptimizeForStaticGraph(other.getOptimizeForStaticGraph());
+        }
+        if (other.getEnableMlirBridge() != false) {
+          setEnableMlirBridge(other.getEnableMlirBridge());
+        }
+        if (other.getEnableMlirGraphOptimization() != false) {
+          setEnableMlirGraphOptimization(other.getEnableMlirGraphOptimization());
+        }
+        if (other.getDisableOutputPartitionGraphs() != false) {
+          setDisableOutputPartitionGraphs(other.getDisableOutputPartitionGraphs());
+        }
+        if (other.getXlaFusionAutotunerThresh() != 0L) {
+          setXlaFusionAutotunerThresh(other.getXlaFusionAutotunerThresh());
+        }
+        this.mergeUnknownFields(other.unknownFields);
+        onChanged();
+        return this;
+      }
+
+      @java.lang.Override
+      public final boolean isInitialized() {
+        return true;
+      }
+
+      @java.lang.Override
+      public Builder mergeFrom(
+          com.google.protobuf.CodedInputStream input,
+          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+          throws java.io.IOException {
+        org.tensorflow.proto.framework.ConfigProto.Experimental parsedMessage = null;
+        try {
+          parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+        } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+          parsedMessage = (org.tensorflow.proto.framework.ConfigProto.Experimental) e.getUnfinishedMessage();
+          throw e.unwrapIOException();
+        } finally {
+          if (parsedMessage != null) {
+            mergeFrom(parsedMessage);
+          }
+        }
+        return this;
+      }
+
+      private java.lang.Object collectiveGroupLeader_ = "";
+      /**
+       * <pre>
+       * Task name for group resolution.
+       * </pre>
+       *
+       * <code>string collective_group_leader = 1;</code>
+       */
+      public java.lang.String getCollectiveGroupLeader() {
+        java.lang.Object ref = collectiveGroupLeader_;
+        if (!(ref instanceof java.lang.String)) {
+          com.google.protobuf.ByteString bs =
+              (com.google.protobuf.ByteString) ref;
+          java.lang.String s = bs.toStringUtf8();
+          collectiveGroupLeader_ = s;
+          return s;
+        } else {
+          return (java.lang.String) ref;
+        }
+      }
+      /**
+       * <pre>
+       * Task name for group resolution.
+       * </pre>
+       *
+       * <code>string collective_group_leader = 1;</code>
+       */
+      public com.google.protobuf.ByteString
+          getCollectiveGroupLeaderBytes() {
+        java.lang.Object ref = collectiveGroupLeader_;
+        if (ref instanceof String) {
+          com.google.protobuf.ByteString b = 
+              com.google.protobuf.ByteString.copyFromUtf8(
+                  (java.lang.String) ref);
+          collectiveGroupLeader_ = b;
+          return b;
+        } else {
+          return (com.google.protobuf.ByteString) ref;
+        }
+      }
+      /**
+       * <pre>
+       * Task name for group resolution.
+       * </pre>
+       *
+       * <code>string collective_group_leader = 1;</code>
+       */
+      public Builder setCollectiveGroupLeader(
+          java.lang.String value) {
+        if (value == null) {
+    throw new NullPointerException();
+  }
+  
+        collectiveGroupLeader_ = value;
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * Task name for group resolution.
+       * </pre>
+       *
+       * <code>string collective_group_leader = 1;</code>
+       */
+      public Builder clearCollectiveGroupLeader() {
+        
+        collectiveGroupLeader_ = getDefaultInstance().getCollectiveGroupLeader();
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * Task name for group resolution.
+       * </pre>
+       *
+       * <code>string collective_group_leader = 1;</code>
+       */
+      public Builder setCollectiveGroupLeaderBytes(
+          com.google.protobuf.ByteString value) {
+        if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+        
+        collectiveGroupLeader_ = value;
+        onChanged();
+        return this;
+      }
+
+      private java.lang.Object executorType_ = "";
+      /**
+       * <pre>
+       * Which executor to use, the default executor will be used
+       * if it is an empty string or "DEFAULT"
+       * </pre>
+       *
+       * <code>string executor_type = 3;</code>
+       */
+      public java.lang.String getExecutorType() {
+        java.lang.Object ref = executorType_;
+        if (!(ref instanceof java.lang.String)) {
+          com.google.protobuf.ByteString bs =
+              (com.google.protobuf.ByteString) ref;
+          java.lang.String s = bs.toStringUtf8();
+          executorType_ = s;
+          return s;
+        } else {
+          return (java.lang.String) ref;
+        }
+      }
+      /**
+       * <pre>
+       * Which executor to use, the default executor will be used
+       * if it is an empty string or "DEFAULT"
+       * </pre>
+       *
+       * <code>string executor_type = 3;</code>
+       */
+      public com.google.protobuf.ByteString
+          getExecutorTypeBytes() {
+        java.lang.Object ref = executorType_;
+        if (ref instanceof String) {
+          com.google.protobuf.ByteString b = 
+              com.google.protobuf.ByteString.copyFromUtf8(
+                  (java.lang.String) ref);
+          executorType_ = b;
+          return b;
+        } else {
+          return (com.google.protobuf.ByteString) ref;
+        }
+      }
+      /**
+       * <pre>
+       * Which executor to use, the default executor will be used
+       * if it is an empty string or "DEFAULT"
+       * </pre>
+       *
+       * <code>string executor_type = 3;</code>
+       */
+      public Builder setExecutorType(
+          java.lang.String value) {
+        if (value == null) {
+    throw new NullPointerException();
+  }
+  
+        executorType_ = value;
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * Which executor to use, the default executor will be used
+       * if it is an empty string or "DEFAULT"
+       * </pre>
+       *
+       * <code>string executor_type = 3;</code>
+       */
+      public Builder clearExecutorType() {
+        
+        executorType_ = getDefaultInstance().getExecutorType();
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * Which executor to use, the default executor will be used
+       * if it is an empty string or "DEFAULT"
+       * </pre>
+       *
+       * <code>string executor_type = 3;</code>
+       */
+      public Builder setExecutorTypeBytes(
+          com.google.protobuf.ByteString value) {
+        if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+        
+        executorType_ = value;
+        onChanged();
+        return this;
+      }
+
+      private int recvBufMaxChunk_ ;
+      /**
+       * <pre>
+       * Guidance to formatting of large RecvBuf fields for transfer.
+       * Any positive value sets the max chunk size.  0 defaults to 4096.
+       * Any negative value indicates no max, i.e. one chunk only.
+       * </pre>
+       *
+       * <code>int32 recv_buf_max_chunk = 4;</code>
+       */
+      public int getRecvBufMaxChunk() {
+        return recvBufMaxChunk_;
+      }
+      /**
+       * <pre>
+       * Guidance to formatting of large RecvBuf fields for transfer.
+       * Any positive value sets the max chunk size.  0 defaults to 4096.
+       * Any negative value indicates no max, i.e. one chunk only.
+       * </pre>
+       *
+       * <code>int32 recv_buf_max_chunk = 4;</code>
+       */
+      public Builder setRecvBufMaxChunk(int value) {
+        
+        recvBufMaxChunk_ = value;
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * Guidance to formatting of large RecvBuf fields for transfer.
+       * Any positive value sets the max chunk size.  0 defaults to 4096.
+       * Any negative value indicates no max, i.e. one chunk only.
+       * </pre>
+       *
+       * <code>int32 recv_buf_max_chunk = 4;</code>
+       */
+      public Builder clearRecvBufMaxChunk() {
+        
+        recvBufMaxChunk_ = 0;
+        onChanged();
+        return this;
+      }
+
+      private boolean useNumaAffinity_ ;
+      /**
+       * <pre>
+       * If true, and supported by the platform, the runtime will attempt to
+       * use NUMA affinity where applicable.  One consequence will be the
+       * existence of as many CPU devices as there are available NUMA nodes.
+       * </pre>
+       *
+       * <code>bool use_numa_affinity = 5;</code>
+       */
+      public boolean getUseNumaAffinity() {
+        return useNumaAffinity_;
+      }
+      /**
+       * <pre>
+       * If true, and supported by the platform, the runtime will attempt to
+       * use NUMA affinity where applicable.  One consequence will be the
+       * existence of as many CPU devices as there are available NUMA nodes.
+       * </pre>
+       *
+       * <code>bool use_numa_affinity = 5;</code>
+       */
+      public Builder setUseNumaAffinity(boolean value) {
+        
+        useNumaAffinity_ = value;
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * If true, and supported by the platform, the runtime will attempt to
+       * use NUMA affinity where applicable.  One consequence will be the
+       * existence of as many CPU devices as there are available NUMA nodes.
+       * </pre>
+       *
+       * <code>bool use_numa_affinity = 5;</code>
+       */
+      public Builder clearUseNumaAffinity() {
+        
+        useNumaAffinity_ = false;
+        onChanged();
+        return this;
+      }
+
+      private boolean collectiveDeterministicSequentialExecution_ ;
+      /**
+       * <pre>
+       * If true, make collective op execution order sequential and deterministic
+       * for potentially concurrent collective instances.
+       * </pre>
+       *
+       * <code>bool collective_deterministic_sequential_execution = 6;</code>
+       */
+      public boolean getCollectiveDeterministicSequentialExecution() {
+        return collectiveDeterministicSequentialExecution_;
+      }
+      /**
+       * <pre>
+       * If true, make collective op execution order sequential and deterministic
+       * for potentially concurrent collective instances.
+       * </pre>
+       *
+       * <code>bool collective_deterministic_sequential_execution = 6;</code>
+       */
+      public Builder setCollectiveDeterministicSequentialExecution(boolean value) {
+        
+        collectiveDeterministicSequentialExecution_ = value;
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * If true, make collective op execution order sequential and deterministic
+       * for potentially concurrent collective instances.
+       * </pre>
+       *
+       * <code>bool collective_deterministic_sequential_execution = 6;</code>
+       */
+      public Builder clearCollectiveDeterministicSequentialExecution() {
+        
+        collectiveDeterministicSequentialExecution_ = false;
+        onChanged();
+        return this;
+      }
+
+      private boolean collectiveNccl_ ;
+      /**
+       * <pre>
+       * If true, use NCCL for CollectiveOps.  This feature is highly
+       * experimental.
+       * </pre>
+       *
+       * <code>bool collective_nccl = 7;</code>
+       */
+      public boolean getCollectiveNccl() {
+        return collectiveNccl_;
+      }
+      /**
+       * <pre>
+       * If true, use NCCL for CollectiveOps.  This feature is highly
+       * experimental.
+       * </pre>
+       *
+       * <code>bool collective_nccl = 7;</code>
+       */
+      public Builder setCollectiveNccl(boolean value) {
+        
+        collectiveNccl_ = value;
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * If true, use NCCL for CollectiveOps.  This feature is highly
+       * experimental.
+       * </pre>
+       *
+       * <code>bool collective_nccl = 7;</code>
+       */
+      public Builder clearCollectiveNccl() {
+        
+        collectiveNccl_ = false;
+        onChanged();
+        return this;
+      }
+
+      private boolean shareSessionStateInClusterspecPropagation_ ;
+      /**
+       * <pre>
+       * In the following, session state means the value of a variable, elements
+       * in a hash table, or any other resource, accessible by worker sessions
+       * held by a TF server.
+       * When ClusterSpec propagation is enabled, the value of
+       * isolate_session_state is ignored when deciding whether to share session
+       * states in a TF server (for backwards compatibility reasons).
+       * - If share_session_state_in_clusterspec_propagation is true, the session
+       * states are shared.
+       * - If share_session_state_in_clusterspec_propagation is false, session
+       * states are isolated.
+       * When clusterspec propagation is not used, the value of
+       * share_session_state_in_clusterspec_propagation is ignored when deciding
+       * whether to share session states in a TF server.
+       * - If isolate_session_state is true, session states are isolated.
+       * - If isolate_session_state is false, session states are shared.
+       * TODO(b/129330037): Add a single API that consistently treats
+       * isolate_session_state and ClusterSpec propagation.
+       * </pre>
+       *
+       * <code>bool share_session_state_in_clusterspec_propagation = 8;</code>
+       */
+      public boolean getShareSessionStateInClusterspecPropagation() {
+        return shareSessionStateInClusterspecPropagation_;
+      }
+      /**
+       * <pre>
+       * In the following, session state means the value of a variable, elements
+       * in a hash table, or any other resource, accessible by worker sessions
+       * held by a TF server.
+       * When ClusterSpec propagation is enabled, the value of
+       * isolate_session_state is ignored when deciding whether to share session
+       * states in a TF server (for backwards compatibility reasons).
+       * - If share_session_state_in_clusterspec_propagation is true, the session
+       * states are shared.
+       * - If share_session_state_in_clusterspec_propagation is false, session
+       * states are isolated.
+       * When clusterspec propagation is not used, the value of
+       * share_session_state_in_clusterspec_propagation is ignored when deciding
+       * whether to share session states in a TF server.
+       * - If isolate_session_state is true, session states are isolated.
+       * - If isolate_session_state is false, session states are shared.
+       * TODO(b/129330037): Add a single API that consistently treats
+       * isolate_session_state and ClusterSpec propagation.
+       * </pre>
+       *
+       * <code>bool share_session_state_in_clusterspec_propagation = 8;</code>
+       */
+      public Builder setShareSessionStateInClusterspecPropagation(boolean value) {
+        
+        shareSessionStateInClusterspecPropagation_ = value;
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * In the following, session state means the value of a variable, elements
+       * in a hash table, or any other resource, accessible by worker sessions
+       * held by a TF server.
+       * When ClusterSpec propagation is enabled, the value of
+       * isolate_session_state is ignored when deciding whether to share session
+       * states in a TF server (for backwards compatibility reasons).
+       * - If share_session_state_in_clusterspec_propagation is true, the session
+       * states are shared.
+       * - If share_session_state_in_clusterspec_propagation is false, session
+       * states are isolated.
+       * When clusterspec propagation is not used, the value of
+       * share_session_state_in_clusterspec_propagation is ignored when deciding
+       * whether to share session states in a TF server.
+       * - If isolate_session_state is true, session states are isolated.
+       * - If isolate_session_state is false, session states are shared.
+       * TODO(b/129330037): Add a single API that consistently treats
+       * isolate_session_state and ClusterSpec propagation.
+       * </pre>
+       *
+       * <code>bool share_session_state_in_clusterspec_propagation = 8;</code>
+       */
+      public Builder clearShareSessionStateInClusterspecPropagation() {
+        
+        shareSessionStateInClusterspecPropagation_ = false;
+        onChanged();
+        return this;
+      }
+
+      private boolean disableThreadSpinning_ ;
+      /**
+       * <pre>
+       * If using a direct session, disable spinning while waiting for work in
+       * the thread pool. This may result in higher latency for completing ops,
+       * but in the case where there is a lot of spinning may result in lower
+       * CPU usage.
+       * </pre>
+       *
+       * <code>bool disable_thread_spinning = 9;</code>
+       */
+      public boolean getDisableThreadSpinning() {
+        return disableThreadSpinning_;
+      }
+      /**
+       * <pre>
+       * If using a direct session, disable spinning while waiting for work in
+       * the thread pool. This may result in higher latency for completing ops,
+       * but in the case where there is a lot of spinning may result in lower
+       * CPU usage.
+       * </pre>
+       *
+       * <code>bool disable_thread_spinning = 9;</code>
+       */
+      public Builder setDisableThreadSpinning(boolean value) {
+        
+        disableThreadSpinning_ = value;
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * If using a direct session, disable spinning while waiting for work in
+       * the thread pool. This may result in higher latency for completing ops,
+       * but in the case where there is a lot of spinning may result in lower
+       * CPU usage.
+       * </pre>
+       *
+       * <code>bool disable_thread_spinning = 9;</code>
+       */
+      public Builder clearDisableThreadSpinning() {
+        
+        disableThreadSpinning_ = false;
+        onChanged();
+        return this;
+      }
+
+      private boolean shareClusterDevicesInSession_ ;
+      /**
+       * <pre>
+       * This was promoted to a non-experimental API. Please use
+       * ConfigProto.share_cluster_devices_in_session instead.
+       * </pre>
+       *
+       * <code>bool share_cluster_devices_in_session = 10;</code>
+       */
+      public boolean getShareClusterDevicesInSession() {
+        return shareClusterDevicesInSession_;
+      }
+      /**
+       * <pre>
+       * This was promoted to a non-experimental API. Please use
+       * ConfigProto.share_cluster_devices_in_session instead.
+       * </pre>
+       *
+       * <code>bool share_cluster_devices_in_session = 10;</code>
+       */
+      public Builder setShareClusterDevicesInSession(boolean value) {
+        
+        shareClusterDevicesInSession_ = value;
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * This was promoted to a non-experimental API. Please use
+       * ConfigProto.share_cluster_devices_in_session instead.
+       * </pre>
+       *
+       * <code>bool share_cluster_devices_in_session = 10;</code>
+       */
+      public Builder clearShareClusterDevicesInSession() {
+        
+        shareClusterDevicesInSession_ = false;
+        onChanged();
+        return this;
+      }
+
+      private org.tensorflow.proto.framework.SessionMetadata sessionMetadata_;
+      private com.google.protobuf.SingleFieldBuilderV3<
+          org.tensorflow.proto.framework.SessionMetadata, org.tensorflow.proto.framework.SessionMetadata.Builder, org.tensorflow.proto.framework.SessionMetadataOrBuilder> sessionMetadataBuilder_;
+      /**
+       * <pre>
+       * Metadata about the session.
+       * If set, this can be used by the runtime and the Ops for debugging,
+       * monitoring, etc.
+       * NOTE: This is currently used and propagated only by the direct session.
+       * </pre>
+       *
+       * <code>.tensorflow.SessionMetadata session_metadata = 11;</code>
+       */
+      public boolean hasSessionMetadata() {
+        return sessionMetadataBuilder_ != null || sessionMetadata_ != null;
+      }
+      /**
+       * <pre>
+       * Metadata about the session.
+       * If set, this can be used by the runtime and the Ops for debugging,
+       * monitoring, etc.
+       * NOTE: This is currently used and propagated only by the direct session.
+       * </pre>
+       *
+       * <code>.tensorflow.SessionMetadata session_metadata = 11;</code>
+       */
+      public org.tensorflow.proto.framework.SessionMetadata getSessionMetadata() {
+        if (sessionMetadataBuilder_ == null) {
+          return sessionMetadata_ == null ? org.tensorflow.proto.framework.SessionMetadata.getDefaultInstance() : sessionMetadata_;
+        } else {
+          return sessionMetadataBuilder_.getMessage();
+        }
+      }
+      /**
+       * <pre>
+       * Metadata about the session.
+       * If set, this can be used by the runtime and the Ops for debugging,
+       * monitoring, etc.
+       * NOTE: This is currently used and propagated only by the direct session.
+       * </pre>
+       *
+       * <code>.tensorflow.SessionMetadata session_metadata = 11;</code>
+       */
+      public Builder setSessionMetadata(org.tensorflow.proto.framework.SessionMetadata value) {
+        if (sessionMetadataBuilder_ == null) {
+          if (value == null) {
+            throw new NullPointerException();
+          }
+          sessionMetadata_ = value;
+          onChanged();
+        } else {
+          sessionMetadataBuilder_.setMessage(value);
+        }
+
+        return this;
+      }
+      /**
+       * <pre>
+       * Metadata about the session.
+       * If set, this can be used by the runtime and the Ops for debugging,
+       * monitoring, etc.
+       * NOTE: This is currently used and propagated only by the direct session.
+       * </pre>
+       *
+       * <code>.tensorflow.SessionMetadata session_metadata = 11;</code>
+       */
+      public Builder setSessionMetadata(
+          org.tensorflow.proto.framework.SessionMetadata.Builder builderForValue) {
+        if (sessionMetadataBuilder_ == null) {
+          sessionMetadata_ = builderForValue.build();
+          onChanged();
+        } else {
+          sessionMetadataBuilder_.setMessage(builderForValue.build());
+        }
+
+        return this;
+      }
+      /**
+       * <pre>
+       * Metadata about the session.
+       * If set, this can be used by the runtime and the Ops for debugging,
+       * monitoring, etc.
+       * NOTE: This is currently used and propagated only by the direct session.
+       * </pre>
+       *
+       * <code>.tensorflow.SessionMetadata session_metadata = 11;</code>
+       */
+      public Builder mergeSessionMetadata(org.tensorflow.proto.framework.SessionMetadata value) {
+        if (sessionMetadataBuilder_ == null) {
+          if (sessionMetadata_ != null) {
+            sessionMetadata_ =
+              org.tensorflow.proto.framework.SessionMetadata.newBuilder(sessionMetadata_).mergeFrom(value).buildPartial();
+          } else {
+            sessionMetadata_ = value;
+          }
+          onChanged();
+        } else {
+          sessionMetadataBuilder_.mergeFrom(value);
+        }
+
+        return this;
+      }
+      /**
+       * <pre>
+       * Metadata about the session.
+       * If set, this can be used by the runtime and the Ops for debugging,
+       * monitoring, etc.
+       * NOTE: This is currently used and propagated only by the direct session.
+       * </pre>
+       *
+       * <code>.tensorflow.SessionMetadata session_metadata = 11;</code>
+       */
+      public Builder clearSessionMetadata() {
+        if (sessionMetadataBuilder_ == null) {
+          sessionMetadata_ = null;
+          onChanged();
+        } else {
+          sessionMetadata_ = null;
+          sessionMetadataBuilder_ = null;
+        }
+
+        return this;
+      }
+      /**
+       * <pre>
+       * Metadata about the session.
+       * If set, this can be used by the runtime and the Ops for debugging,
+       * monitoring, etc.
+       * NOTE: This is currently used and propagated only by the direct session.
+       * </pre>
+       *
+       * <code>.tensorflow.SessionMetadata session_metadata = 11;</code>
+       */
+      public org.tensorflow.proto.framework.SessionMetadata.Builder getSessionMetadataBuilder() {
+        
+        onChanged();
+        return getSessionMetadataFieldBuilder().getBuilder();
+      }
+      /**
+       * <pre>
+       * Metadata about the session.
+       * If set, this can be used by the runtime and the Ops for debugging,
+       * monitoring, etc.
+       * NOTE: This is currently used and propagated only by the direct session.
+       * </pre>
+       *
+       * <code>.tensorflow.SessionMetadata session_metadata = 11;</code>
+       */
+      public org.tensorflow.proto.framework.SessionMetadataOrBuilder getSessionMetadataOrBuilder() {
+        if (sessionMetadataBuilder_ != null) {
+          return sessionMetadataBuilder_.getMessageOrBuilder();
+        } else {
+          return sessionMetadata_ == null ?
+              org.tensorflow.proto.framework.SessionMetadata.getDefaultInstance() : sessionMetadata_;
+        }
+      }
+      /**
+       * <pre>
+       * Metadata about the session.
+       * If set, this can be used by the runtime and the Ops for debugging,
+       * monitoring, etc.
+       * NOTE: This is currently used and propagated only by the direct session.
+       * </pre>
+       *
+       * <code>.tensorflow.SessionMetadata session_metadata = 11;</code>
+       */
+      private com.google.protobuf.SingleFieldBuilderV3<
+          org.tensorflow.proto.framework.SessionMetadata, org.tensorflow.proto.framework.SessionMetadata.Builder, org.tensorflow.proto.framework.SessionMetadataOrBuilder> 
+          getSessionMetadataFieldBuilder() {
+        if (sessionMetadataBuilder_ == null) {
+          sessionMetadataBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+              org.tensorflow.proto.framework.SessionMetadata, org.tensorflow.proto.framework.SessionMetadata.Builder, org.tensorflow.proto.framework.SessionMetadataOrBuilder>(
+                  getSessionMetadata(),
+                  getParentForChildren(),
+                  isClean());
+          sessionMetadata_ = null;
+        }
+        return sessionMetadataBuilder_;
+      }
+
+      private boolean optimizeForStaticGraph_ ;
+      /**
+       * <pre>
+       * If true, the session may treat the graph as being static for optimization
+       * purposes.
+       * If this option is set to true when a session is created, the full
+       * GraphDef must be passed in a single call to Session::Create(), and
+       * Session::Extend() may not be supported.
+       * </pre>
+       *
+       * <code>bool optimize_for_static_graph = 12;</code>
+       */
+      public boolean getOptimizeForStaticGraph() {
+        return optimizeForStaticGraph_;
+      }
+      /**
+       * <pre>
+       * If true, the session may treat the graph as being static for optimization
+       * purposes.
+       * If this option is set to true when a session is created, the full
+       * GraphDef must be passed in a single call to Session::Create(), and
+       * Session::Extend() may not be supported.
+       * </pre>
+       *
+       * <code>bool optimize_for_static_graph = 12;</code>
+       */
+      public Builder setOptimizeForStaticGraph(boolean value) {
+        
+        optimizeForStaticGraph_ = value;
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * If true, the session may treat the graph as being static for optimization
+       * purposes.
+       * If this option is set to true when a session is created, the full
+       * GraphDef must be passed in a single call to Session::Create(), and
+       * Session::Extend() may not be supported.
+       * </pre>
+       *
+       * <code>bool optimize_for_static_graph = 12;</code>
+       */
+      public Builder clearOptimizeForStaticGraph() {
+        
+        optimizeForStaticGraph_ = false;
+        onChanged();
+        return this;
+      }
+
+      private boolean enableMlirBridge_ ;
+      /**
+       * <pre>
+       * Whether to enable the MLIR-based TF-&gt;XLA bridge.
+       * This is a replacement to the existing bridge, and not ready for
+       * production usage yet.
+       * If this option is set to true when a session is created, MLIR is used to
+       * perform the set of graph transformations to put the graph in a form that
+       * can be executed with delegation of some computations to an accelerator.
+       * This builds on the model of XLA where a subset of the graph is
+       * encapsulated and attached to a "compile" operation, whose result is fed
+       * to an "execute" operation. The kernel for these operations is responsible
+       * to lower the encapsulated graph to a particular device.
+       * </pre>
+       *
+       * <code>bool enable_mlir_bridge = 13;</code>
+       */
+      public boolean getEnableMlirBridge() {
+        return enableMlirBridge_;
+      }
+      /**
+       * <pre>
+       * Whether to enable the MLIR-based TF-&gt;XLA bridge.
+       * This is a replacement to the existing bridge, and not ready for
+       * production usage yet.
+       * If this option is set to true when a session is created, MLIR is used to
+       * perform the set of graph transformations to put the graph in a form that
+       * can be executed with delegation of some computations to an accelerator.
+       * This builds on the model of XLA where a subset of the graph is
+       * encapsulated and attached to a "compile" operation, whose result is fed
+       * to an "execute" operation. The kernel for these operations is responsible
+       * to lower the encapsulated graph to a particular device.
+       * </pre>
+       *
+       * <code>bool enable_mlir_bridge = 13;</code>
+       */
+      public Builder setEnableMlirBridge(boolean value) {
+        
+        enableMlirBridge_ = value;
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * Whether to enable the MLIR-based TF-&gt;XLA bridge.
+       * This is a replacement to the existing bridge, and not ready for
+       * production usage yet.
+       * If this option is set to true when a session is created, MLIR is used to
+       * perform the set of graph transformations to put the graph in a form that
+       * can be executed with delegation of some computations to an accelerator.
+       * This builds on the model of XLA where a subset of the graph is
+       * encapsulated and attached to a "compile" operation, whose result is fed
+       * to an "execute" operation. The kernel for these operations is responsible
+       * to lower the encapsulated graph to a particular device.
+       * </pre>
+       *
+       * <code>bool enable_mlir_bridge = 13;</code>
+       */
+      public Builder clearEnableMlirBridge() {
+        
+        enableMlirBridge_ = false;
+        onChanged();
+        return this;
+      }
+
+      private boolean enableMlirGraphOptimization_ ;
+      /**
+       * <pre>
+       * Whether to enable the MLIR-based Graph optimizations.
+       * This will become a part of standard Tensorflow graph optimization
+       * pipeline, currently this is only used for gradual migration and testing
+       * new passes that are replacing existing optimizations in Grappler.
+       * </pre>
+       *
+       * <code>bool enable_mlir_graph_optimization = 16;</code>
+       */
+      public boolean getEnableMlirGraphOptimization() {
+        return enableMlirGraphOptimization_;
+      }
+      /**
+       * <pre>
+       * Whether to enable the MLIR-based Graph optimizations.
+       * This will become a part of standard Tensorflow graph optimization
+       * pipeline, currently this is only used for gradual migration and testing
+       * new passes that are replacing existing optimizations in Grappler.
+       * </pre>
+       *
+       * <code>bool enable_mlir_graph_optimization = 16;</code>
+       */
+      public Builder setEnableMlirGraphOptimization(boolean value) {
+        
+        enableMlirGraphOptimization_ = value;
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * Whether to enable the MLIR-based Graph optimizations.
+       * This will become a part of standard Tensorflow graph optimization
+       * pipeline, currently this is only used for gradual migration and testing
+       * new passes that are replacing existing optimizations in Grappler.
+       * </pre>
+       *
+       * <code>bool enable_mlir_graph_optimization = 16;</code>
+       */
+      public Builder clearEnableMlirGraphOptimization() {
+        
+        enableMlirGraphOptimization_ = false;
+        onChanged();
+        return this;
+      }
+
+      private boolean disableOutputPartitionGraphs_ ;
+      /**
+       * <pre>
+       * If true, the session will not store an additional copy of the graph for
+       * each subgraph.
+       * If this option is set to true when a session is created, the
+       * `RunOptions.output_partition_graphs` options must not be set.
+       * </pre>
+       *
+       * <code>bool disable_output_partition_graphs = 14;</code>
+       */
+      public boolean getDisableOutputPartitionGraphs() {
+        return disableOutputPartitionGraphs_;
+      }
+      /**
+       * <pre>
+       * If true, the session will not store an additional copy of the graph for
+       * each subgraph.
+       * If this option is set to true when a session is created, the
+       * `RunOptions.output_partition_graphs` options must not be set.
+       * </pre>
+       *
+       * <code>bool disable_output_partition_graphs = 14;</code>
+       */
+      public Builder setDisableOutputPartitionGraphs(boolean value) {
+        
+        disableOutputPartitionGraphs_ = value;
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * If true, the session will not store an additional copy of the graph for
+       * each subgraph.
+       * If this option is set to true when a session is created, the
+       * `RunOptions.output_partition_graphs` options must not be set.
+       * </pre>
+       *
+       * <code>bool disable_output_partition_graphs = 14;</code>
+       */
+      public Builder clearDisableOutputPartitionGraphs() {
+        
+        disableOutputPartitionGraphs_ = false;
+        onChanged();
+        return this;
+      }
+
+      private long xlaFusionAutotunerThresh_ ;
+      /**
+       * <pre>
+       * Minimum number of batches run through the XLA graph before XLA fusion
+       * autotuner is enabled. Default value of zero disables the autotuner.
+       * The XLA fusion autotuner can improve performance by executing a heuristic
+       * search on the compiler parameters.
+       * </pre>
+       *
+       * <code>int64 xla_fusion_autotuner_thresh = 15;</code>
+       */
+      public long getXlaFusionAutotunerThresh() {
+        return xlaFusionAutotunerThresh_;
+      }
+      /**
+       * <pre>
+       * Minimum number of batches run through the XLA graph before XLA fusion
+       * autotuner is enabled. Default value of zero disables the autotuner.
+       * The XLA fusion autotuner can improve performance by executing a heuristic
+       * search on the compiler parameters.
+       * </pre>
+       *
+       * <code>int64 xla_fusion_autotuner_thresh = 15;</code>
+       */
+      public Builder setXlaFusionAutotunerThresh(long value) {
+        
+        xlaFusionAutotunerThresh_ = value;
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * Minimum number of batches run through the XLA graph before XLA fusion
+       * autotuner is enabled. Default value of zero disables the autotuner.
+       * The XLA fusion autotuner can improve performance by executing a heuristic
+       * search on the compiler parameters.
+       * </pre>
+       *
+       * <code>int64 xla_fusion_autotuner_thresh = 15;</code>
+       */
+      public Builder clearXlaFusionAutotunerThresh() {
+        
+        xlaFusionAutotunerThresh_ = 0L;
+        onChanged();
+        return this;
+      }
+      @java.lang.Override
+      public final Builder setUnknownFields(
+          final com.google.protobuf.UnknownFieldSet unknownFields) {
+        return super.setUnknownFields(unknownFields);
+      }
+
+      @java.lang.Override
+      public final Builder mergeUnknownFields(
+          final com.google.protobuf.UnknownFieldSet unknownFields) {
+        return super.mergeUnknownFields(unknownFields);
+      }
+
+
+      // @@protoc_insertion_point(builder_scope:tensorflow.ConfigProto.Experimental)
+    }
+
+    // @@protoc_insertion_point(class_scope:tensorflow.ConfigProto.Experimental)
+    private static final org.tensorflow.proto.framework.ConfigProto.Experimental DEFAULT_INSTANCE;
+    static {
+      DEFAULT_INSTANCE = new org.tensorflow.proto.framework.ConfigProto.Experimental();
+    }
+
+    public static org.tensorflow.proto.framework.ConfigProto.Experimental getDefaultInstance() {
+      return DEFAULT_INSTANCE;
+    }
+
+    private static final com.google.protobuf.Parser<Experimental>
+        PARSER = new com.google.protobuf.AbstractParser<Experimental>() {
+      @java.lang.Override
+      public Experimental parsePartialFrom(
+          com.google.protobuf.CodedInputStream input,
+          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+          throws com.google.protobuf.InvalidProtocolBufferException {
+        return new Experimental(input, extensionRegistry);
+      }
+    };
+
+    public static com.google.protobuf.Parser<Experimental> parser() {
+      return PARSER;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Parser<Experimental> getParserForType() {
+      return PARSER;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.ConfigProto.Experimental getDefaultInstanceForType() {
+      return DEFAULT_INSTANCE;
+    }
+
+  }
+
+  public static final int DEVICE_COUNT_FIELD_NUMBER = 1;
+  private static final class DeviceCountDefaultEntryHolder {
+    static final com.google.protobuf.MapEntry<
+        java.lang.String, java.lang.Integer> defaultEntry =
+            com.google.protobuf.MapEntry
+            .<java.lang.String, java.lang.Integer>newDefaultInstance(
+                org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_ConfigProto_DeviceCountEntry_descriptor, 
+                com.google.protobuf.WireFormat.FieldType.STRING,
+                "",
+                com.google.protobuf.WireFormat.FieldType.INT32,
+                0);
+  }
+  private com.google.protobuf.MapField<
+      java.lang.String, java.lang.Integer> deviceCount_;
+  private com.google.protobuf.MapField<java.lang.String, java.lang.Integer>
+  internalGetDeviceCount() {
+    if (deviceCount_ == null) {
+      return com.google.protobuf.MapField.emptyMapField(
+          DeviceCountDefaultEntryHolder.defaultEntry);
+    }
+    return deviceCount_;
+  }
+
+  public int getDeviceCountCount() {
+    return internalGetDeviceCount().getMap().size();
+  }
+  /**
+   * <pre>
+   * Map from device type name (e.g., "CPU" or "GPU" ) to maximum
+   * number of devices of that type to use.  If a particular device
+   * type is not found in the map, the system picks an appropriate
+   * number.
+   * </pre>
+   *
+   * <code>map&lt;string, int32&gt; device_count = 1;</code>
+   */
+
+  public boolean containsDeviceCount(
+      java.lang.String key) {
+    if (key == null) { throw new java.lang.NullPointerException(); }
+    return internalGetDeviceCount().getMap().containsKey(key);
+  }
+  /**
+   * Use {@link #getDeviceCountMap()} instead.
+   */
+  @java.lang.Deprecated
+  public java.util.Map<java.lang.String, java.lang.Integer> getDeviceCount() {
+    return getDeviceCountMap();
+  }
+  /**
+   * <pre>
+   * Map from device type name (e.g., "CPU" or "GPU" ) to maximum
+   * number of devices of that type to use.  If a particular device
+   * type is not found in the map, the system picks an appropriate
+   * number.
+   * </pre>
+   *
+   * <code>map&lt;string, int32&gt; device_count = 1;</code>
+   */
+
+  public java.util.Map<java.lang.String, java.lang.Integer> getDeviceCountMap() {
+    return internalGetDeviceCount().getMap();
+  }
+  /**
+   * <pre>
+   * Map from device type name (e.g., "CPU" or "GPU" ) to maximum
+   * number of devices of that type to use.  If a particular device
+   * type is not found in the map, the system picks an appropriate
+   * number.
+   * </pre>
+   *
+   * <code>map&lt;string, int32&gt; device_count = 1;</code>
+   */
+
+  public int getDeviceCountOrDefault(
+      java.lang.String key,
+      int defaultValue) {
+    if (key == null) { throw new java.lang.NullPointerException(); }
+    java.util.Map<java.lang.String, java.lang.Integer> map =
+        internalGetDeviceCount().getMap();
+    return map.containsKey(key) ? map.get(key) : defaultValue;
+  }
+  /**
+   * <pre>
+   * Map from device type name (e.g., "CPU" or "GPU" ) to maximum
+   * number of devices of that type to use.  If a particular device
+   * type is not found in the map, the system picks an appropriate
+   * number.
+   * </pre>
+   *
+   * <code>map&lt;string, int32&gt; device_count = 1;</code>
+   */
+
+  public int getDeviceCountOrThrow(
+      java.lang.String key) {
+    if (key == null) { throw new java.lang.NullPointerException(); }
+    java.util.Map<java.lang.String, java.lang.Integer> map =
+        internalGetDeviceCount().getMap();
+    if (!map.containsKey(key)) {
+      throw new java.lang.IllegalArgumentException();
+    }
+    return map.get(key);
+  }
+
+  public static final int INTRA_OP_PARALLELISM_THREADS_FIELD_NUMBER = 2;
+  private int intraOpParallelismThreads_;
+  /**
+   * <pre>
+   * The execution of an individual op (for some op types) can be
+   * parallelized on a pool of intra_op_parallelism_threads.
+   * 0 means the system picks an appropriate number.
+   * If you create an ordinary session, e.g., from Python or C++,
+   * then there is exactly one intra op thread pool per process.
+   * The first session created determines the number of threads in this pool.
+   * All subsequent sessions reuse/share this one global pool.
+   * There are notable exceptions to the default behavior describe above:
+   * 1. There is an environment variable  for overriding this thread pool,
+   *    named TF_OVERRIDE_GLOBAL_THREADPOOL.
+   * 2. When connecting to a server, such as a remote `tf.train.Server`
+   *    instance, then this option will be ignored altogether.
+   * </pre>
+   *
+   * <code>int32 intra_op_parallelism_threads = 2;</code>
+   */
+  public int getIntraOpParallelismThreads() {
+    return intraOpParallelismThreads_;
+  }
+
+  public static final int INTER_OP_PARALLELISM_THREADS_FIELD_NUMBER = 5;
+  private int interOpParallelismThreads_;
+  /**
+   * <pre>
+   * Nodes that perform blocking operations are enqueued on a pool of
+   * inter_op_parallelism_threads available in each process.
+   * 0 means the system picks an appropriate number.
+   * Negative means all operations are performed in caller's thread.
+   * Note that the first Session created in the process sets the
+   * number of threads for all future sessions unless use_per_session_threads is
+   * true or session_inter_op_thread_pool is configured.
+   * </pre>
+   *
+   * <code>int32 inter_op_parallelism_threads = 5;</code>
+   */
+  public int getInterOpParallelismThreads() {
+    return interOpParallelismThreads_;
+  }
+
+  public static final int USE_PER_SESSION_THREADS_FIELD_NUMBER = 9;
+  private boolean usePerSessionThreads_;
+  /**
+   * <pre>
+   * If true, use a new set of threads for this session rather than the global
+   * pool of threads. Only supported by direct sessions.
+   * If false, use the global threads created by the first session, or the
+   * per-session thread pools configured by session_inter_op_thread_pool.
+   * This option is deprecated. The same effect can be achieved by setting
+   * session_inter_op_thread_pool to have one element, whose num_threads equals
+   * inter_op_parallelism_threads.
+   * </pre>
+   *
+   * <code>bool use_per_session_threads = 9;</code>
+   */
+  public boolean getUsePerSessionThreads() {
+    return usePerSessionThreads_;
+  }
+
+  public static final int SESSION_INTER_OP_THREAD_POOL_FIELD_NUMBER = 12;
+  private java.util.List<org.tensorflow.proto.framework.ThreadPoolOptionProto> sessionInterOpThreadPool_;
+  /**
+   * <pre>
+   * This option is experimental - it may be replaced with a different mechanism
+   * in the future.
+   * Configures session thread pools. If this is configured, then RunOptions for
+   * a Run call can select the thread pool to use.
+   * The intended use is for when some session invocations need to run in a
+   * background pool limited to a small number of threads:
+   * - For example, a session may be configured to have one large pool (for
+   * regular compute) and one small pool (for periodic, low priority work);
+   * using the small pool is currently the mechanism for limiting the inter-op
+   * parallelism of the low priority work.  Note that it does not limit the
+   * parallelism of work spawned by a single op kernel implementation.
+   * - Using this setting is normally not needed in training, but may help some
+   * serving use cases.
+   * - It is also generally recommended to set the global_name field of this
+   * proto, to avoid creating multiple large pools. It is typically better to
+   * run the non-low-priority work, even across sessions, in a single large
+   * pool.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.ThreadPoolOptionProto session_inter_op_thread_pool = 12;</code>
+   */
+  public java.util.List<org.tensorflow.proto.framework.ThreadPoolOptionProto> getSessionInterOpThreadPoolList() {
+    return sessionInterOpThreadPool_;
+  }
+  /**
+   * <pre>
+   * This option is experimental - it may be replaced with a different mechanism
+   * in the future.
+   * Configures session thread pools. If this is configured, then RunOptions for
+   * a Run call can select the thread pool to use.
+   * The intended use is for when some session invocations need to run in a
+   * background pool limited to a small number of threads:
+   * - For example, a session may be configured to have one large pool (for
+   * regular compute) and one small pool (for periodic, low priority work);
+   * using the small pool is currently the mechanism for limiting the inter-op
+   * parallelism of the low priority work.  Note that it does not limit the
+   * parallelism of work spawned by a single op kernel implementation.
+   * - Using this setting is normally not needed in training, but may help some
+   * serving use cases.
+   * - It is also generally recommended to set the global_name field of this
+   * proto, to avoid creating multiple large pools. It is typically better to
+   * run the non-low-priority work, even across sessions, in a single large
+   * pool.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.ThreadPoolOptionProto session_inter_op_thread_pool = 12;</code>
+   */
+  public java.util.List<? extends org.tensorflow.proto.framework.ThreadPoolOptionProtoOrBuilder> 
+      getSessionInterOpThreadPoolOrBuilderList() {
+    return sessionInterOpThreadPool_;
+  }
+  /**
+   * <pre>
+   * This option is experimental - it may be replaced with a different mechanism
+   * in the future.
+   * Configures session thread pools. If this is configured, then RunOptions for
+   * a Run call can select the thread pool to use.
+   * The intended use is for when some session invocations need to run in a
+   * background pool limited to a small number of threads:
+   * - For example, a session may be configured to have one large pool (for
+   * regular compute) and one small pool (for periodic, low priority work);
+   * using the small pool is currently the mechanism for limiting the inter-op
+   * parallelism of the low priority work.  Note that it does not limit the
+   * parallelism of work spawned by a single op kernel implementation.
+   * - Using this setting is normally not needed in training, but may help some
+   * serving use cases.
+   * - It is also generally recommended to set the global_name field of this
+   * proto, to avoid creating multiple large pools. It is typically better to
+   * run the non-low-priority work, even across sessions, in a single large
+   * pool.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.ThreadPoolOptionProto session_inter_op_thread_pool = 12;</code>
+   */
+  public int getSessionInterOpThreadPoolCount() {
+    return sessionInterOpThreadPool_.size();
+  }
+  /**
+   * <pre>
+   * This option is experimental - it may be replaced with a different mechanism
+   * in the future.
+   * Configures session thread pools. If this is configured, then RunOptions for
+   * a Run call can select the thread pool to use.
+   * The intended use is for when some session invocations need to run in a
+   * background pool limited to a small number of threads:
+   * - For example, a session may be configured to have one large pool (for
+   * regular compute) and one small pool (for periodic, low priority work);
+   * using the small pool is currently the mechanism for limiting the inter-op
+   * parallelism of the low priority work.  Note that it does not limit the
+   * parallelism of work spawned by a single op kernel implementation.
+   * - Using this setting is normally not needed in training, but may help some
+   * serving use cases.
+   * - It is also generally recommended to set the global_name field of this
+   * proto, to avoid creating multiple large pools. It is typically better to
+   * run the non-low-priority work, even across sessions, in a single large
+   * pool.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.ThreadPoolOptionProto session_inter_op_thread_pool = 12;</code>
+   */
+  public org.tensorflow.proto.framework.ThreadPoolOptionProto getSessionInterOpThreadPool(int index) {
+    return sessionInterOpThreadPool_.get(index);
+  }
+  /**
+   * <pre>
+   * This option is experimental - it may be replaced with a different mechanism
+   * in the future.
+   * Configures session thread pools. If this is configured, then RunOptions for
+   * a Run call can select the thread pool to use.
+   * The intended use is for when some session invocations need to run in a
+   * background pool limited to a small number of threads:
+   * - For example, a session may be configured to have one large pool (for
+   * regular compute) and one small pool (for periodic, low priority work);
+   * using the small pool is currently the mechanism for limiting the inter-op
+   * parallelism of the low priority work.  Note that it does not limit the
+   * parallelism of work spawned by a single op kernel implementation.
+   * - Using this setting is normally not needed in training, but may help some
+   * serving use cases.
+   * - It is also generally recommended to set the global_name field of this
+   * proto, to avoid creating multiple large pools. It is typically better to
+   * run the non-low-priority work, even across sessions, in a single large
+   * pool.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.ThreadPoolOptionProto session_inter_op_thread_pool = 12;</code>
+   */
+  public org.tensorflow.proto.framework.ThreadPoolOptionProtoOrBuilder getSessionInterOpThreadPoolOrBuilder(
+      int index) {
+    return sessionInterOpThreadPool_.get(index);
+  }
+
+  public static final int PLACEMENT_PERIOD_FIELD_NUMBER = 3;
+  private int placementPeriod_;
+  /**
+   * <pre>
+   * Assignment of Nodes to Devices is recomputed every placement_period
+   * steps until the system warms up (at which point the recomputation
+   * typically slows down automatically).
+   * </pre>
+   *
+   * <code>int32 placement_period = 3;</code>
+   */
+  public int getPlacementPeriod() {
+    return placementPeriod_;
+  }
+
+  public static final int DEVICE_FILTERS_FIELD_NUMBER = 4;
+  private com.google.protobuf.LazyStringList deviceFilters_;
+  /**
+   * <pre>
+   * When any filters are present sessions will ignore all devices which do not
+   * match the filters. Each filter can be partially specified, e.g. "/job:ps"
+   * "/job:worker/replica:3", etc.
+   * </pre>
+   *
+   * <code>repeated string device_filters = 4;</code>
+   */
+  public com.google.protobuf.ProtocolStringList
+      getDeviceFiltersList() {
+    return deviceFilters_;
+  }
+  /**
+   * <pre>
+   * When any filters are present sessions will ignore all devices which do not
+   * match the filters. Each filter can be partially specified, e.g. "/job:ps"
+   * "/job:worker/replica:3", etc.
+   * </pre>
+   *
+   * <code>repeated string device_filters = 4;</code>
+   */
+  public int getDeviceFiltersCount() {
+    return deviceFilters_.size();
+  }
+  /**
+   * <pre>
+   * When any filters are present sessions will ignore all devices which do not
+   * match the filters. Each filter can be partially specified, e.g. "/job:ps"
+   * "/job:worker/replica:3", etc.
+   * </pre>
+   *
+   * <code>repeated string device_filters = 4;</code>
+   */
+  public java.lang.String getDeviceFilters(int index) {
+    return deviceFilters_.get(index);
+  }
+  /**
+   * <pre>
+   * When any filters are present sessions will ignore all devices which do not
+   * match the filters. Each filter can be partially specified, e.g. "/job:ps"
+   * "/job:worker/replica:3", etc.
+   * </pre>
+   *
+   * <code>repeated string device_filters = 4;</code>
+   */
+  public com.google.protobuf.ByteString
+      getDeviceFiltersBytes(int index) {
+    return deviceFilters_.getByteString(index);
+  }
+
+  public static final int GPU_OPTIONS_FIELD_NUMBER = 6;
+  private org.tensorflow.proto.framework.GPUOptions gpuOptions_;
+  /**
+   * <pre>
+   * Options that apply to all GPUs.
+   * </pre>
+   *
+   * <code>.tensorflow.GPUOptions gpu_options = 6;</code>
+   */
+  public boolean hasGpuOptions() {
+    return gpuOptions_ != null;
+  }
+  /**
+   * <pre>
+   * Options that apply to all GPUs.
+   * </pre>
+   *
+   * <code>.tensorflow.GPUOptions gpu_options = 6;</code>
+   */
+  public org.tensorflow.proto.framework.GPUOptions getGpuOptions() {
+    return gpuOptions_ == null ? org.tensorflow.proto.framework.GPUOptions.getDefaultInstance() : gpuOptions_;
+  }
+  /**
+   * <pre>
+   * Options that apply to all GPUs.
+   * </pre>
+   *
+   * <code>.tensorflow.GPUOptions gpu_options = 6;</code>
+   */
+  public org.tensorflow.proto.framework.GPUOptionsOrBuilder getGpuOptionsOrBuilder() {
+    return getGpuOptions();
+  }
+
+  public static final int ALLOW_SOFT_PLACEMENT_FIELD_NUMBER = 7;
+  private boolean allowSoftPlacement_;
+  /**
+   * <pre>
+   * Whether soft placement is allowed. If allow_soft_placement is true,
+   * an op will be placed on CPU if
+   *   1. there's no GPU implementation for the OP
+   * or
+   *   2. no GPU devices are known or registered
+   * or
+   *   3. need to co-locate with reftype input(s) which are from CPU.
+   * </pre>
+   *
+   * <code>bool allow_soft_placement = 7;</code>
+   */
+  public boolean getAllowSoftPlacement() {
+    return allowSoftPlacement_;
+  }
+
+  public static final int LOG_DEVICE_PLACEMENT_FIELD_NUMBER = 8;
+  private boolean logDevicePlacement_;
+  /**
+   * <pre>
+   * Whether device placements should be logged.
+   * </pre>
+   *
+   * <code>bool log_device_placement = 8;</code>
+   */
+  public boolean getLogDevicePlacement() {
+    return logDevicePlacement_;
+  }
+
+  public static final int GRAPH_OPTIONS_FIELD_NUMBER = 10;
+  private org.tensorflow.proto.framework.GraphOptions graphOptions_;
+  /**
+   * <pre>
+   * Options that apply to all graphs.
+   * </pre>
+   *
+   * <code>.tensorflow.GraphOptions graph_options = 10;</code>
+   */
+  public boolean hasGraphOptions() {
+    return graphOptions_ != null;
+  }
+  /**
+   * <pre>
+   * Options that apply to all graphs.
+   * </pre>
+   *
+   * <code>.tensorflow.GraphOptions graph_options = 10;</code>
+   */
+  public org.tensorflow.proto.framework.GraphOptions getGraphOptions() {
+    return graphOptions_ == null ? org.tensorflow.proto.framework.GraphOptions.getDefaultInstance() : graphOptions_;
+  }
+  /**
+   * <pre>
+   * Options that apply to all graphs.
+   * </pre>
+   *
+   * <code>.tensorflow.GraphOptions graph_options = 10;</code>
+   */
+  public org.tensorflow.proto.framework.GraphOptionsOrBuilder getGraphOptionsOrBuilder() {
+    return getGraphOptions();
+  }
+
+  public static final int OPERATION_TIMEOUT_IN_MS_FIELD_NUMBER = 11;
+  private long operationTimeoutInMs_;
+  /**
+   * <pre>
+   * Global timeout for all blocking operations in this session.  If non-zero,
+   * and not overridden on a per-operation basis, this value will be used as the
+   * deadline for all blocking operations.
+   * </pre>
+   *
+   * <code>int64 operation_timeout_in_ms = 11;</code>
+   */
+  public long getOperationTimeoutInMs() {
+    return operationTimeoutInMs_;
+  }
+
+  public static final int RPC_OPTIONS_FIELD_NUMBER = 13;
+  private org.tensorflow.proto.framework.RPCOptions rpcOptions_;
+  /**
+   * <pre>
+   * Options that apply when this session uses the distributed runtime.
+   * </pre>
+   *
+   * <code>.tensorflow.RPCOptions rpc_options = 13;</code>
+   */
+  public boolean hasRpcOptions() {
+    return rpcOptions_ != null;
+  }
+  /**
+   * <pre>
+   * Options that apply when this session uses the distributed runtime.
+   * </pre>
+   *
+   * <code>.tensorflow.RPCOptions rpc_options = 13;</code>
+   */
+  public org.tensorflow.proto.framework.RPCOptions getRpcOptions() {
+    return rpcOptions_ == null ? org.tensorflow.proto.framework.RPCOptions.getDefaultInstance() : rpcOptions_;
+  }
+  /**
+   * <pre>
+   * Options that apply when this session uses the distributed runtime.
+   * </pre>
+   *
+   * <code>.tensorflow.RPCOptions rpc_options = 13;</code>
+   */
+  public org.tensorflow.proto.framework.RPCOptionsOrBuilder getRpcOptionsOrBuilder() {
+    return getRpcOptions();
+  }
+
+  public static final int CLUSTER_DEF_FIELD_NUMBER = 14;
+  private org.tensorflow.proto.distruntime.ClusterDef clusterDef_;
+  /**
+   * <pre>
+   * Optional list of all workers to use in this session.
+   * </pre>
+   *
+   * <code>.tensorflow.ClusterDef cluster_def = 14;</code>
+   */
+  public boolean hasClusterDef() {
+    return clusterDef_ != null;
+  }
+  /**
+   * <pre>
+   * Optional list of all workers to use in this session.
+   * </pre>
+   *
+   * <code>.tensorflow.ClusterDef cluster_def = 14;</code>
+   */
+  public org.tensorflow.proto.distruntime.ClusterDef getClusterDef() {
+    return clusterDef_ == null ? org.tensorflow.proto.distruntime.ClusterDef.getDefaultInstance() : clusterDef_;
+  }
+  /**
+   * <pre>
+   * Optional list of all workers to use in this session.
+   * </pre>
+   *
+   * <code>.tensorflow.ClusterDef cluster_def = 14;</code>
+   */
+  public org.tensorflow.proto.distruntime.ClusterDefOrBuilder getClusterDefOrBuilder() {
+    return getClusterDef();
+  }
+
+  public static final int ISOLATE_SESSION_STATE_FIELD_NUMBER = 15;
+  private boolean isolateSessionState_;
+  /**
+   * <pre>
+   * If true, any resources such as Variables used in the session will not be
+   * shared with other sessions. However, when clusterspec propagation is
+   * enabled, this field is ignored and sessions are always isolated.
+   * </pre>
+   *
+   * <code>bool isolate_session_state = 15;</code>
+   */
+  public boolean getIsolateSessionState() {
+    return isolateSessionState_;
+  }
+
+  public static final int SHARE_CLUSTER_DEVICES_IN_SESSION_FIELD_NUMBER = 17;
+  private boolean shareClusterDevicesInSession_;
+  /**
+   * <pre>
+   * When true, WorkerSessions are created with device attributes from the
+   * full cluster.
+   * This is helpful when a worker wants to partition a graph
+   * (for example during a PartitionedCallOp).
+   * </pre>
+   *
+   * <code>bool share_cluster_devices_in_session = 17;</code>
+   */
+  public boolean getShareClusterDevicesInSession() {
+    return shareClusterDevicesInSession_;
+  }
+
+  public static final int EXPERIMENTAL_FIELD_NUMBER = 16;
+  private org.tensorflow.proto.framework.ConfigProto.Experimental experimental_;
+  /**
+   * <code>.tensorflow.ConfigProto.Experimental experimental = 16;</code>
+   */
+  public boolean hasExperimental() {
+    return experimental_ != null;
+  }
+  /**
+   * <code>.tensorflow.ConfigProto.Experimental experimental = 16;</code>
+   */
+  public org.tensorflow.proto.framework.ConfigProto.Experimental getExperimental() {
+    return experimental_ == null ? org.tensorflow.proto.framework.ConfigProto.Experimental.getDefaultInstance() : experimental_;
+  }
+  /**
+   * <code>.tensorflow.ConfigProto.Experimental experimental = 16;</code>
+   */
+  public org.tensorflow.proto.framework.ConfigProto.ExperimentalOrBuilder getExperimentalOrBuilder() {
+    return getExperimental();
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    com.google.protobuf.GeneratedMessageV3
+      .serializeStringMapTo(
+        output,
+        internalGetDeviceCount(),
+        DeviceCountDefaultEntryHolder.defaultEntry,
+        1);
+    if (intraOpParallelismThreads_ != 0) {
+      output.writeInt32(2, intraOpParallelismThreads_);
+    }
+    if (placementPeriod_ != 0) {
+      output.writeInt32(3, placementPeriod_);
+    }
+    for (int i = 0; i < deviceFilters_.size(); i++) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 4, deviceFilters_.getRaw(i));
+    }
+    if (interOpParallelismThreads_ != 0) {
+      output.writeInt32(5, interOpParallelismThreads_);
+    }
+    if (gpuOptions_ != null) {
+      output.writeMessage(6, getGpuOptions());
+    }
+    if (allowSoftPlacement_ != false) {
+      output.writeBool(7, allowSoftPlacement_);
+    }
+    if (logDevicePlacement_ != false) {
+      output.writeBool(8, logDevicePlacement_);
+    }
+    if (usePerSessionThreads_ != false) {
+      output.writeBool(9, usePerSessionThreads_);
+    }
+    if (graphOptions_ != null) {
+      output.writeMessage(10, getGraphOptions());
+    }
+    if (operationTimeoutInMs_ != 0L) {
+      output.writeInt64(11, operationTimeoutInMs_);
+    }
+    for (int i = 0; i < sessionInterOpThreadPool_.size(); i++) {
+      output.writeMessage(12, sessionInterOpThreadPool_.get(i));
+    }
+    if (rpcOptions_ != null) {
+      output.writeMessage(13, getRpcOptions());
+    }
+    if (clusterDef_ != null) {
+      output.writeMessage(14, getClusterDef());
+    }
+    if (isolateSessionState_ != false) {
+      output.writeBool(15, isolateSessionState_);
+    }
+    if (experimental_ != null) {
+      output.writeMessage(16, getExperimental());
+    }
+    if (shareClusterDevicesInSession_ != false) {
+      output.writeBool(17, shareClusterDevicesInSession_);
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    for (java.util.Map.Entry<java.lang.String, java.lang.Integer> entry
+         : internalGetDeviceCount().getMap().entrySet()) {
+      com.google.protobuf.MapEntry<java.lang.String, java.lang.Integer>
+      deviceCount__ = DeviceCountDefaultEntryHolder.defaultEntry.newBuilderForType()
+          .setKey(entry.getKey())
+          .setValue(entry.getValue())
+          .build();
+      size += com.google.protobuf.CodedOutputStream
+          .computeMessageSize(1, deviceCount__);
+    }
+    if (intraOpParallelismThreads_ != 0) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt32Size(2, intraOpParallelismThreads_);
+    }
+    if (placementPeriod_ != 0) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt32Size(3, placementPeriod_);
+    }
+    {
+      int dataSize = 0;
+      for (int i = 0; i < deviceFilters_.size(); i++) {
+        dataSize += computeStringSizeNoTag(deviceFilters_.getRaw(i));
+      }
+      size += dataSize;
+      size += 1 * getDeviceFiltersList().size();
+    }
+    if (interOpParallelismThreads_ != 0) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt32Size(5, interOpParallelismThreads_);
+    }
+    if (gpuOptions_ != null) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(6, getGpuOptions());
+    }
+    if (allowSoftPlacement_ != false) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeBoolSize(7, allowSoftPlacement_);
+    }
+    if (logDevicePlacement_ != false) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeBoolSize(8, logDevicePlacement_);
+    }
+    if (usePerSessionThreads_ != false) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeBoolSize(9, usePerSessionThreads_);
+    }
+    if (graphOptions_ != null) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(10, getGraphOptions());
+    }
+    if (operationTimeoutInMs_ != 0L) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt64Size(11, operationTimeoutInMs_);
+    }
+    for (int i = 0; i < sessionInterOpThreadPool_.size(); i++) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(12, sessionInterOpThreadPool_.get(i));
+    }
+    if (rpcOptions_ != null) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(13, getRpcOptions());
+    }
+    if (clusterDef_ != null) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(14, getClusterDef());
+    }
+    if (isolateSessionState_ != false) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeBoolSize(15, isolateSessionState_);
+    }
+    if (experimental_ != null) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(16, getExperimental());
+    }
+    if (shareClusterDevicesInSession_ != false) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeBoolSize(17, shareClusterDevicesInSession_);
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.framework.ConfigProto)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.framework.ConfigProto other = (org.tensorflow.proto.framework.ConfigProto) obj;
+
+    if (!internalGetDeviceCount().equals(
+        other.internalGetDeviceCount())) return false;
+    if (getIntraOpParallelismThreads()
+        != other.getIntraOpParallelismThreads()) return false;
+    if (getInterOpParallelismThreads()
+        != other.getInterOpParallelismThreads()) return false;
+    if (getUsePerSessionThreads()
+        != other.getUsePerSessionThreads()) return false;
+    if (!getSessionInterOpThreadPoolList()
+        .equals(other.getSessionInterOpThreadPoolList())) return false;
+    if (getPlacementPeriod()
+        != other.getPlacementPeriod()) return false;
+    if (!getDeviceFiltersList()
+        .equals(other.getDeviceFiltersList())) return false;
+    if (hasGpuOptions() != other.hasGpuOptions()) return false;
+    if (hasGpuOptions()) {
+      if (!getGpuOptions()
+          .equals(other.getGpuOptions())) return false;
+    }
+    if (getAllowSoftPlacement()
+        != other.getAllowSoftPlacement()) return false;
+    if (getLogDevicePlacement()
+        != other.getLogDevicePlacement()) return false;
+    if (hasGraphOptions() != other.hasGraphOptions()) return false;
+    if (hasGraphOptions()) {
+      if (!getGraphOptions()
+          .equals(other.getGraphOptions())) return false;
+    }
+    if (getOperationTimeoutInMs()
+        != other.getOperationTimeoutInMs()) return false;
+    if (hasRpcOptions() != other.hasRpcOptions()) return false;
+    if (hasRpcOptions()) {
+      if (!getRpcOptions()
+          .equals(other.getRpcOptions())) return false;
+    }
+    if (hasClusterDef() != other.hasClusterDef()) return false;
+    if (hasClusterDef()) {
+      if (!getClusterDef()
+          .equals(other.getClusterDef())) return false;
+    }
+    if (getIsolateSessionState()
+        != other.getIsolateSessionState()) return false;
+    if (getShareClusterDevicesInSession()
+        != other.getShareClusterDevicesInSession()) return false;
+    if (hasExperimental() != other.hasExperimental()) return false;
+    if (hasExperimental()) {
+      if (!getExperimental()
+          .equals(other.getExperimental())) return false;
+    }
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    if (!internalGetDeviceCount().getMap().isEmpty()) {
+      hash = (37 * hash) + DEVICE_COUNT_FIELD_NUMBER;
+      hash = (53 * hash) + internalGetDeviceCount().hashCode();
+    }
+    hash = (37 * hash) + INTRA_OP_PARALLELISM_THREADS_FIELD_NUMBER;
+    hash = (53 * hash) + getIntraOpParallelismThreads();
+    hash = (37 * hash) + INTER_OP_PARALLELISM_THREADS_FIELD_NUMBER;
+    hash = (53 * hash) + getInterOpParallelismThreads();
+    hash = (37 * hash) + USE_PER_SESSION_THREADS_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashBoolean(
+        getUsePerSessionThreads());
+    if (getSessionInterOpThreadPoolCount() > 0) {
+      hash = (37 * hash) + SESSION_INTER_OP_THREAD_POOL_FIELD_NUMBER;
+      hash = (53 * hash) + getSessionInterOpThreadPoolList().hashCode();
+    }
+    hash = (37 * hash) + PLACEMENT_PERIOD_FIELD_NUMBER;
+    hash = (53 * hash) + getPlacementPeriod();
+    if (getDeviceFiltersCount() > 0) {
+      hash = (37 * hash) + DEVICE_FILTERS_FIELD_NUMBER;
+      hash = (53 * hash) + getDeviceFiltersList().hashCode();
+    }
+    if (hasGpuOptions()) {
+      hash = (37 * hash) + GPU_OPTIONS_FIELD_NUMBER;
+      hash = (53 * hash) + getGpuOptions().hashCode();
+    }
+    hash = (37 * hash) + ALLOW_SOFT_PLACEMENT_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashBoolean(
+        getAllowSoftPlacement());
+    hash = (37 * hash) + LOG_DEVICE_PLACEMENT_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashBoolean(
+        getLogDevicePlacement());
+    if (hasGraphOptions()) {
+      hash = (37 * hash) + GRAPH_OPTIONS_FIELD_NUMBER;
+      hash = (53 * hash) + getGraphOptions().hashCode();
+    }
+    hash = (37 * hash) + OPERATION_TIMEOUT_IN_MS_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+        getOperationTimeoutInMs());
+    if (hasRpcOptions()) {
+      hash = (37 * hash) + RPC_OPTIONS_FIELD_NUMBER;
+      hash = (53 * hash) + getRpcOptions().hashCode();
+    }
+    if (hasClusterDef()) {
+      hash = (37 * hash) + CLUSTER_DEF_FIELD_NUMBER;
+      hash = (53 * hash) + getClusterDef().hashCode();
+    }
+    hash = (37 * hash) + ISOLATE_SESSION_STATE_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashBoolean(
+        getIsolateSessionState());
+    hash = (37 * hash) + SHARE_CLUSTER_DEVICES_IN_SESSION_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashBoolean(
+        getShareClusterDevicesInSession());
+    if (hasExperimental()) {
+      hash = (37 * hash) + EXPERIMENTAL_FIELD_NUMBER;
+      hash = (53 * hash) + getExperimental().hashCode();
+    }
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.framework.ConfigProto parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.ConfigProto parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.ConfigProto parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.ConfigProto parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.ConfigProto parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.ConfigProto parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.ConfigProto parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.ConfigProto parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.ConfigProto parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.ConfigProto parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.ConfigProto parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.ConfigProto parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.framework.ConfigProto prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * <pre>
+   * Session configuration parameters.
+   * The system picks appropriate values for fields that are not set.
+   * </pre>
+   *
+   * Protobuf type {@code tensorflow.ConfigProto}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.ConfigProto)
+      org.tensorflow.proto.framework.ConfigProtoOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_ConfigProto_descriptor;
+    }
+
+    @SuppressWarnings({"rawtypes"})
+    protected com.google.protobuf.MapField internalGetMapField(
+        int number) {
+      switch (number) {
+        case 1:
+          return internalGetDeviceCount();
+        default:
+          throw new RuntimeException(
+              "Invalid map field number: " + number);
+      }
+    }
+    @SuppressWarnings({"rawtypes"})
+    protected com.google.protobuf.MapField internalGetMutableMapField(
+        int number) {
+      switch (number) {
+        case 1:
+          return internalGetMutableDeviceCount();
+        default:
+          throw new RuntimeException(
+              "Invalid map field number: " + number);
+      }
+    }
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_ConfigProto_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.ConfigProto.class, org.tensorflow.proto.framework.ConfigProto.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.framework.ConfigProto.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+        getSessionInterOpThreadPoolFieldBuilder();
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      internalGetMutableDeviceCount().clear();
+      intraOpParallelismThreads_ = 0;
+
+      interOpParallelismThreads_ = 0;
+
+      usePerSessionThreads_ = false;
+
+      if (sessionInterOpThreadPoolBuilder_ == null) {
+        sessionInterOpThreadPool_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000002);
+      } else {
+        sessionInterOpThreadPoolBuilder_.clear();
+      }
+      placementPeriod_ = 0;
+
+      deviceFilters_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+      bitField0_ = (bitField0_ & ~0x00000004);
+      if (gpuOptionsBuilder_ == null) {
+        gpuOptions_ = null;
+      } else {
+        gpuOptions_ = null;
+        gpuOptionsBuilder_ = null;
+      }
+      allowSoftPlacement_ = false;
+
+      logDevicePlacement_ = false;
+
+      if (graphOptionsBuilder_ == null) {
+        graphOptions_ = null;
+      } else {
+        graphOptions_ = null;
+        graphOptionsBuilder_ = null;
+      }
+      operationTimeoutInMs_ = 0L;
+
+      if (rpcOptionsBuilder_ == null) {
+        rpcOptions_ = null;
+      } else {
+        rpcOptions_ = null;
+        rpcOptionsBuilder_ = null;
+      }
+      if (clusterDefBuilder_ == null) {
+        clusterDef_ = null;
+      } else {
+        clusterDef_ = null;
+        clusterDefBuilder_ = null;
+      }
+      isolateSessionState_ = false;
+
+      shareClusterDevicesInSession_ = false;
+
+      if (experimentalBuilder_ == null) {
+        experimental_ = null;
+      } else {
+        experimental_ = null;
+        experimentalBuilder_ = null;
+      }
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_ConfigProto_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.ConfigProto getDefaultInstanceForType() {
+      return org.tensorflow.proto.framework.ConfigProto.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.ConfigProto build() {
+      org.tensorflow.proto.framework.ConfigProto result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.ConfigProto buildPartial() {
+      org.tensorflow.proto.framework.ConfigProto result = new org.tensorflow.proto.framework.ConfigProto(this);
+      int from_bitField0_ = bitField0_;
+      result.deviceCount_ = internalGetDeviceCount();
+      result.deviceCount_.makeImmutable();
+      result.intraOpParallelismThreads_ = intraOpParallelismThreads_;
+      result.interOpParallelismThreads_ = interOpParallelismThreads_;
+      result.usePerSessionThreads_ = usePerSessionThreads_;
+      if (sessionInterOpThreadPoolBuilder_ == null) {
+        if (((bitField0_ & 0x00000002) != 0)) {
+          sessionInterOpThreadPool_ = java.util.Collections.unmodifiableList(sessionInterOpThreadPool_);
+          bitField0_ = (bitField0_ & ~0x00000002);
+        }
+        result.sessionInterOpThreadPool_ = sessionInterOpThreadPool_;
+      } else {
+        result.sessionInterOpThreadPool_ = sessionInterOpThreadPoolBuilder_.build();
+      }
+      result.placementPeriod_ = placementPeriod_;
+      if (((bitField0_ & 0x00000004) != 0)) {
+        deviceFilters_ = deviceFilters_.getUnmodifiableView();
+        bitField0_ = (bitField0_ & ~0x00000004);
+      }
+      result.deviceFilters_ = deviceFilters_;
+      if (gpuOptionsBuilder_ == null) {
+        result.gpuOptions_ = gpuOptions_;
+      } else {
+        result.gpuOptions_ = gpuOptionsBuilder_.build();
+      }
+      result.allowSoftPlacement_ = allowSoftPlacement_;
+      result.logDevicePlacement_ = logDevicePlacement_;
+      if (graphOptionsBuilder_ == null) {
+        result.graphOptions_ = graphOptions_;
+      } else {
+        result.graphOptions_ = graphOptionsBuilder_.build();
+      }
+      result.operationTimeoutInMs_ = operationTimeoutInMs_;
+      if (rpcOptionsBuilder_ == null) {
+        result.rpcOptions_ = rpcOptions_;
+      } else {
+        result.rpcOptions_ = rpcOptionsBuilder_.build();
+      }
+      if (clusterDefBuilder_ == null) {
+        result.clusterDef_ = clusterDef_;
+      } else {
+        result.clusterDef_ = clusterDefBuilder_.build();
+      }
+      result.isolateSessionState_ = isolateSessionState_;
+      result.shareClusterDevicesInSession_ = shareClusterDevicesInSession_;
+      if (experimentalBuilder_ == null) {
+        result.experimental_ = experimental_;
+      } else {
+        result.experimental_ = experimentalBuilder_.build();
+      }
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.framework.ConfigProto) {
+        return mergeFrom((org.tensorflow.proto.framework.ConfigProto)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.framework.ConfigProto other) {
+      if (other == org.tensorflow.proto.framework.ConfigProto.getDefaultInstance()) return this;
+      internalGetMutableDeviceCount().mergeFrom(
+          other.internalGetDeviceCount());
+      if (other.getIntraOpParallelismThreads() != 0) {
+        setIntraOpParallelismThreads(other.getIntraOpParallelismThreads());
+      }
+      if (other.getInterOpParallelismThreads() != 0) {
+        setInterOpParallelismThreads(other.getInterOpParallelismThreads());
+      }
+      if (other.getUsePerSessionThreads() != false) {
+        setUsePerSessionThreads(other.getUsePerSessionThreads());
+      }
+      if (sessionInterOpThreadPoolBuilder_ == null) {
+        if (!other.sessionInterOpThreadPool_.isEmpty()) {
+          if (sessionInterOpThreadPool_.isEmpty()) {
+            sessionInterOpThreadPool_ = other.sessionInterOpThreadPool_;
+            bitField0_ = (bitField0_ & ~0x00000002);
+          } else {
+            ensureSessionInterOpThreadPoolIsMutable();
+            sessionInterOpThreadPool_.addAll(other.sessionInterOpThreadPool_);
+          }
+          onChanged();
+        }
+      } else {
+        if (!other.sessionInterOpThreadPool_.isEmpty()) {
+          if (sessionInterOpThreadPoolBuilder_.isEmpty()) {
+            sessionInterOpThreadPoolBuilder_.dispose();
+            sessionInterOpThreadPoolBuilder_ = null;
+            sessionInterOpThreadPool_ = other.sessionInterOpThreadPool_;
+            bitField0_ = (bitField0_ & ~0x00000002);
+            sessionInterOpThreadPoolBuilder_ = 
+              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
+                 getSessionInterOpThreadPoolFieldBuilder() : null;
+          } else {
+            sessionInterOpThreadPoolBuilder_.addAllMessages(other.sessionInterOpThreadPool_);
+          }
+        }
+      }
+      if (other.getPlacementPeriod() != 0) {
+        setPlacementPeriod(other.getPlacementPeriod());
+      }
+      if (!other.deviceFilters_.isEmpty()) {
+        if (deviceFilters_.isEmpty()) {
+          deviceFilters_ = other.deviceFilters_;
+          bitField0_ = (bitField0_ & ~0x00000004);
+        } else {
+          ensureDeviceFiltersIsMutable();
+          deviceFilters_.addAll(other.deviceFilters_);
+        }
+        onChanged();
+      }
+      if (other.hasGpuOptions()) {
+        mergeGpuOptions(other.getGpuOptions());
+      }
+      if (other.getAllowSoftPlacement() != false) {
+        setAllowSoftPlacement(other.getAllowSoftPlacement());
+      }
+      if (other.getLogDevicePlacement() != false) {
+        setLogDevicePlacement(other.getLogDevicePlacement());
+      }
+      if (other.hasGraphOptions()) {
+        mergeGraphOptions(other.getGraphOptions());
+      }
+      if (other.getOperationTimeoutInMs() != 0L) {
+        setOperationTimeoutInMs(other.getOperationTimeoutInMs());
+      }
+      if (other.hasRpcOptions()) {
+        mergeRpcOptions(other.getRpcOptions());
+      }
+      if (other.hasClusterDef()) {
+        mergeClusterDef(other.getClusterDef());
+      }
+      if (other.getIsolateSessionState() != false) {
+        setIsolateSessionState(other.getIsolateSessionState());
+      }
+      if (other.getShareClusterDevicesInSession() != false) {
+        setShareClusterDevicesInSession(other.getShareClusterDevicesInSession());
+      }
+      if (other.hasExperimental()) {
+        mergeExperimental(other.getExperimental());
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.framework.ConfigProto parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.framework.ConfigProto) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+    private int bitField0_;
+
+    private com.google.protobuf.MapField<
+        java.lang.String, java.lang.Integer> deviceCount_;
+    private com.google.protobuf.MapField<java.lang.String, java.lang.Integer>
+    internalGetDeviceCount() {
+      if (deviceCount_ == null) {
+        return com.google.protobuf.MapField.emptyMapField(
+            DeviceCountDefaultEntryHolder.defaultEntry);
+      }
+      return deviceCount_;
+    }
+    private com.google.protobuf.MapField<java.lang.String, java.lang.Integer>
+    internalGetMutableDeviceCount() {
+      onChanged();;
+      if (deviceCount_ == null) {
+        deviceCount_ = com.google.protobuf.MapField.newMapField(
+            DeviceCountDefaultEntryHolder.defaultEntry);
+      }
+      if (!deviceCount_.isMutable()) {
+        deviceCount_ = deviceCount_.copy();
+      }
+      return deviceCount_;
+    }
+
+    public int getDeviceCountCount() {
+      return internalGetDeviceCount().getMap().size();
+    }
+    /**
+     * <pre>
+     * Map from device type name (e.g., "CPU" or "GPU" ) to maximum
+     * number of devices of that type to use.  If a particular device
+     * type is not found in the map, the system picks an appropriate
+     * number.
+     * </pre>
+     *
+     * <code>map&lt;string, int32&gt; device_count = 1;</code>
+     */
+
+    public boolean containsDeviceCount(
+        java.lang.String key) {
+      if (key == null) { throw new java.lang.NullPointerException(); }
+      return internalGetDeviceCount().getMap().containsKey(key);
+    }
+    /**
+     * Use {@link #getDeviceCountMap()} instead.
+     */
+    @java.lang.Deprecated
+    public java.util.Map<java.lang.String, java.lang.Integer> getDeviceCount() {
+      return getDeviceCountMap();
+    }
+    /**
+     * <pre>
+     * Map from device type name (e.g., "CPU" or "GPU" ) to maximum
+     * number of devices of that type to use.  If a particular device
+     * type is not found in the map, the system picks an appropriate
+     * number.
+     * </pre>
+     *
+     * <code>map&lt;string, int32&gt; device_count = 1;</code>
+     */
+
+    public java.util.Map<java.lang.String, java.lang.Integer> getDeviceCountMap() {
+      return internalGetDeviceCount().getMap();
+    }
+    /**
+     * <pre>
+     * Map from device type name (e.g., "CPU" or "GPU" ) to maximum
+     * number of devices of that type to use.  If a particular device
+     * type is not found in the map, the system picks an appropriate
+     * number.
+     * </pre>
+     *
+     * <code>map&lt;string, int32&gt; device_count = 1;</code>
+     */
+
+    public int getDeviceCountOrDefault(
+        java.lang.String key,
+        int defaultValue) {
+      if (key == null) { throw new java.lang.NullPointerException(); }
+      java.util.Map<java.lang.String, java.lang.Integer> map =
+          internalGetDeviceCount().getMap();
+      return map.containsKey(key) ? map.get(key) : defaultValue;
+    }
+    /**
+     * <pre>
+     * Map from device type name (e.g., "CPU" or "GPU" ) to maximum
+     * number of devices of that type to use.  If a particular device
+     * type is not found in the map, the system picks an appropriate
+     * number.
+     * </pre>
+     *
+     * <code>map&lt;string, int32&gt; device_count = 1;</code>
+     */
+
+    public int getDeviceCountOrThrow(
+        java.lang.String key) {
+      if (key == null) { throw new java.lang.NullPointerException(); }
+      java.util.Map<java.lang.String, java.lang.Integer> map =
+          internalGetDeviceCount().getMap();
+      if (!map.containsKey(key)) {
+        throw new java.lang.IllegalArgumentException();
+      }
+      return map.get(key);
+    }
+
+    public Builder clearDeviceCount() {
+      internalGetMutableDeviceCount().getMutableMap()
+          .clear();
+      return this;
+    }
+    /**
+     * <pre>
+     * Map from device type name (e.g., "CPU" or "GPU" ) to maximum
+     * number of devices of that type to use.  If a particular device
+     * type is not found in the map, the system picks an appropriate
+     * number.
+     * </pre>
+     *
+     * <code>map&lt;string, int32&gt; device_count = 1;</code>
+     */
+
+    public Builder removeDeviceCount(
+        java.lang.String key) {
+      if (key == null) { throw new java.lang.NullPointerException(); }
+      internalGetMutableDeviceCount().getMutableMap()
+          .remove(key);
+      return this;
+    }
+    /**
+     * Use alternate mutation accessors instead.
+     */
+    @java.lang.Deprecated
+    public java.util.Map<java.lang.String, java.lang.Integer>
+    getMutableDeviceCount() {
+      return internalGetMutableDeviceCount().getMutableMap();
+    }
+    /**
+     * <pre>
+     * Map from device type name (e.g., "CPU" or "GPU" ) to maximum
+     * number of devices of that type to use.  If a particular device
+     * type is not found in the map, the system picks an appropriate
+     * number.
+     * </pre>
+     *
+     * <code>map&lt;string, int32&gt; device_count = 1;</code>
+     */
+    public Builder putDeviceCount(
+        java.lang.String key,
+        int value) {
+      if (key == null) { throw new java.lang.NullPointerException(); }
+      
+      internalGetMutableDeviceCount().getMutableMap()
+          .put(key, value);
+      return this;
+    }
+    /**
+     * <pre>
+     * Map from device type name (e.g., "CPU" or "GPU" ) to maximum
+     * number of devices of that type to use.  If a particular device
+     * type is not found in the map, the system picks an appropriate
+     * number.
+     * </pre>
+     *
+     * <code>map&lt;string, int32&gt; device_count = 1;</code>
+     */
+
+    public Builder putAllDeviceCount(
+        java.util.Map<java.lang.String, java.lang.Integer> values) {
+      internalGetMutableDeviceCount().getMutableMap()
+          .putAll(values);
+      return this;
+    }
+
+    private int intraOpParallelismThreads_ ;
+    /**
+     * <pre>
+     * The execution of an individual op (for some op types) can be
+     * parallelized on a pool of intra_op_parallelism_threads.
+     * 0 means the system picks an appropriate number.
+     * If you create an ordinary session, e.g., from Python or C++,
+     * then there is exactly one intra op thread pool per process.
+     * The first session created determines the number of threads in this pool.
+     * All subsequent sessions reuse/share this one global pool.
+     * There are notable exceptions to the default behavior describe above:
+     * 1. There is an environment variable  for overriding this thread pool,
+     *    named TF_OVERRIDE_GLOBAL_THREADPOOL.
+     * 2. When connecting to a server, such as a remote `tf.train.Server`
+     *    instance, then this option will be ignored altogether.
+     * </pre>
+     *
+     * <code>int32 intra_op_parallelism_threads = 2;</code>
+     */
+    public int getIntraOpParallelismThreads() {
+      return intraOpParallelismThreads_;
+    }
+    /**
+     * <pre>
+     * The execution of an individual op (for some op types) can be
+     * parallelized on a pool of intra_op_parallelism_threads.
+     * 0 means the system picks an appropriate number.
+     * If you create an ordinary session, e.g., from Python or C++,
+     * then there is exactly one intra op thread pool per process.
+     * The first session created determines the number of threads in this pool.
+     * All subsequent sessions reuse/share this one global pool.
+     * There are notable exceptions to the default behavior describe above:
+     * 1. There is an environment variable  for overriding this thread pool,
+     *    named TF_OVERRIDE_GLOBAL_THREADPOOL.
+     * 2. When connecting to a server, such as a remote `tf.train.Server`
+     *    instance, then this option will be ignored altogether.
+     * </pre>
+     *
+     * <code>int32 intra_op_parallelism_threads = 2;</code>
+     */
+    public Builder setIntraOpParallelismThreads(int value) {
+      
+      intraOpParallelismThreads_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * The execution of an individual op (for some op types) can be
+     * parallelized on a pool of intra_op_parallelism_threads.
+     * 0 means the system picks an appropriate number.
+     * If you create an ordinary session, e.g., from Python or C++,
+     * then there is exactly one intra op thread pool per process.
+     * The first session created determines the number of threads in this pool.
+     * All subsequent sessions reuse/share this one global pool.
+     * There are notable exceptions to the default behavior describe above:
+     * 1. There is an environment variable  for overriding this thread pool,
+     *    named TF_OVERRIDE_GLOBAL_THREADPOOL.
+     * 2. When connecting to a server, such as a remote `tf.train.Server`
+     *    instance, then this option will be ignored altogether.
+     * </pre>
+     *
+     * <code>int32 intra_op_parallelism_threads = 2;</code>
+     */
+    public Builder clearIntraOpParallelismThreads() {
+      
+      intraOpParallelismThreads_ = 0;
+      onChanged();
+      return this;
+    }
+
+    private int interOpParallelismThreads_ ;
+    /**
+     * <pre>
+     * Nodes that perform blocking operations are enqueued on a pool of
+     * inter_op_parallelism_threads available in each process.
+     * 0 means the system picks an appropriate number.
+     * Negative means all operations are performed in caller's thread.
+     * Note that the first Session created in the process sets the
+     * number of threads for all future sessions unless use_per_session_threads is
+     * true or session_inter_op_thread_pool is configured.
+     * </pre>
+     *
+     * <code>int32 inter_op_parallelism_threads = 5;</code>
+     */
+    public int getInterOpParallelismThreads() {
+      return interOpParallelismThreads_;
+    }
+    /**
+     * <pre>
+     * Nodes that perform blocking operations are enqueued on a pool of
+     * inter_op_parallelism_threads available in each process.
+     * 0 means the system picks an appropriate number.
+     * Negative means all operations are performed in caller's thread.
+     * Note that the first Session created in the process sets the
+     * number of threads for all future sessions unless use_per_session_threads is
+     * true or session_inter_op_thread_pool is configured.
+     * </pre>
+     *
+     * <code>int32 inter_op_parallelism_threads = 5;</code>
+     */
+    public Builder setInterOpParallelismThreads(int value) {
+      
+      interOpParallelismThreads_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Nodes that perform blocking operations are enqueued on a pool of
+     * inter_op_parallelism_threads available in each process.
+     * 0 means the system picks an appropriate number.
+     * Negative means all operations are performed in caller's thread.
+     * Note that the first Session created in the process sets the
+     * number of threads for all future sessions unless use_per_session_threads is
+     * true or session_inter_op_thread_pool is configured.
+     * </pre>
+     *
+     * <code>int32 inter_op_parallelism_threads = 5;</code>
+     */
+    public Builder clearInterOpParallelismThreads() {
+      
+      interOpParallelismThreads_ = 0;
+      onChanged();
+      return this;
+    }
+
+    private boolean usePerSessionThreads_ ;
+    /**
+     * <pre>
+     * If true, use a new set of threads for this session rather than the global
+     * pool of threads. Only supported by direct sessions.
+     * If false, use the global threads created by the first session, or the
+     * per-session thread pools configured by session_inter_op_thread_pool.
+     * This option is deprecated. The same effect can be achieved by setting
+     * session_inter_op_thread_pool to have one element, whose num_threads equals
+     * inter_op_parallelism_threads.
+     * </pre>
+     *
+     * <code>bool use_per_session_threads = 9;</code>
+     */
+    public boolean getUsePerSessionThreads() {
+      return usePerSessionThreads_;
+    }
+    /**
+     * <pre>
+     * If true, use a new set of threads for this session rather than the global
+     * pool of threads. Only supported by direct sessions.
+     * If false, use the global threads created by the first session, or the
+     * per-session thread pools configured by session_inter_op_thread_pool.
+     * This option is deprecated. The same effect can be achieved by setting
+     * session_inter_op_thread_pool to have one element, whose num_threads equals
+     * inter_op_parallelism_threads.
+     * </pre>
+     *
+     * <code>bool use_per_session_threads = 9;</code>
+     */
+    public Builder setUsePerSessionThreads(boolean value) {
+      
+      usePerSessionThreads_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * If true, use a new set of threads for this session rather than the global
+     * pool of threads. Only supported by direct sessions.
+     * If false, use the global threads created by the first session, or the
+     * per-session thread pools configured by session_inter_op_thread_pool.
+     * This option is deprecated. The same effect can be achieved by setting
+     * session_inter_op_thread_pool to have one element, whose num_threads equals
+     * inter_op_parallelism_threads.
+     * </pre>
+     *
+     * <code>bool use_per_session_threads = 9;</code>
+     */
+    public Builder clearUsePerSessionThreads() {
+      
+      usePerSessionThreads_ = false;
+      onChanged();
+      return this;
+    }
+
+    private java.util.List<org.tensorflow.proto.framework.ThreadPoolOptionProto> sessionInterOpThreadPool_ =
+      java.util.Collections.emptyList();
+    private void ensureSessionInterOpThreadPoolIsMutable() {
+      if (!((bitField0_ & 0x00000002) != 0)) {
+        sessionInterOpThreadPool_ = new java.util.ArrayList<org.tensorflow.proto.framework.ThreadPoolOptionProto>(sessionInterOpThreadPool_);
+        bitField0_ |= 0x00000002;
+       }
+    }
+
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.framework.ThreadPoolOptionProto, org.tensorflow.proto.framework.ThreadPoolOptionProto.Builder, org.tensorflow.proto.framework.ThreadPoolOptionProtoOrBuilder> sessionInterOpThreadPoolBuilder_;
+
+    /**
+     * <pre>
+     * This option is experimental - it may be replaced with a different mechanism
+     * in the future.
+     * Configures session thread pools. If this is configured, then RunOptions for
+     * a Run call can select the thread pool to use.
+     * The intended use is for when some session invocations need to run in a
+     * background pool limited to a small number of threads:
+     * - For example, a session may be configured to have one large pool (for
+     * regular compute) and one small pool (for periodic, low priority work);
+     * using the small pool is currently the mechanism for limiting the inter-op
+     * parallelism of the low priority work.  Note that it does not limit the
+     * parallelism of work spawned by a single op kernel implementation.
+     * - Using this setting is normally not needed in training, but may help some
+     * serving use cases.
+     * - It is also generally recommended to set the global_name field of this
+     * proto, to avoid creating multiple large pools. It is typically better to
+     * run the non-low-priority work, even across sessions, in a single large
+     * pool.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.ThreadPoolOptionProto session_inter_op_thread_pool = 12;</code>
+     */
+    public java.util.List<org.tensorflow.proto.framework.ThreadPoolOptionProto> getSessionInterOpThreadPoolList() {
+      if (sessionInterOpThreadPoolBuilder_ == null) {
+        return java.util.Collections.unmodifiableList(sessionInterOpThreadPool_);
+      } else {
+        return sessionInterOpThreadPoolBuilder_.getMessageList();
+      }
+    }
+    /**
+     * <pre>
+     * This option is experimental - it may be replaced with a different mechanism
+     * in the future.
+     * Configures session thread pools. If this is configured, then RunOptions for
+     * a Run call can select the thread pool to use.
+     * The intended use is for when some session invocations need to run in a
+     * background pool limited to a small number of threads:
+     * - For example, a session may be configured to have one large pool (for
+     * regular compute) and one small pool (for periodic, low priority work);
+     * using the small pool is currently the mechanism for limiting the inter-op
+     * parallelism of the low priority work.  Note that it does not limit the
+     * parallelism of work spawned by a single op kernel implementation.
+     * - Using this setting is normally not needed in training, but may help some
+     * serving use cases.
+     * - It is also generally recommended to set the global_name field of this
+     * proto, to avoid creating multiple large pools. It is typically better to
+     * run the non-low-priority work, even across sessions, in a single large
+     * pool.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.ThreadPoolOptionProto session_inter_op_thread_pool = 12;</code>
+     */
+    public int getSessionInterOpThreadPoolCount() {
+      if (sessionInterOpThreadPoolBuilder_ == null) {
+        return sessionInterOpThreadPool_.size();
+      } else {
+        return sessionInterOpThreadPoolBuilder_.getCount();
+      }
+    }
+    /**
+     * <pre>
+     * This option is experimental - it may be replaced with a different mechanism
+     * in the future.
+     * Configures session thread pools. If this is configured, then RunOptions for
+     * a Run call can select the thread pool to use.
+     * The intended use is for when some session invocations need to run in a
+     * background pool limited to a small number of threads:
+     * - For example, a session may be configured to have one large pool (for
+     * regular compute) and one small pool (for periodic, low priority work);
+     * using the small pool is currently the mechanism for limiting the inter-op
+     * parallelism of the low priority work.  Note that it does not limit the
+     * parallelism of work spawned by a single op kernel implementation.
+     * - Using this setting is normally not needed in training, but may help some
+     * serving use cases.
+     * - It is also generally recommended to set the global_name field of this
+     * proto, to avoid creating multiple large pools. It is typically better to
+     * run the non-low-priority work, even across sessions, in a single large
+     * pool.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.ThreadPoolOptionProto session_inter_op_thread_pool = 12;</code>
+     */
+    public org.tensorflow.proto.framework.ThreadPoolOptionProto getSessionInterOpThreadPool(int index) {
+      if (sessionInterOpThreadPoolBuilder_ == null) {
+        return sessionInterOpThreadPool_.get(index);
+      } else {
+        return sessionInterOpThreadPoolBuilder_.getMessage(index);
+      }
+    }
+    /**
+     * <pre>
+     * This option is experimental - it may be replaced with a different mechanism
+     * in the future.
+     * Configures session thread pools. If this is configured, then RunOptions for
+     * a Run call can select the thread pool to use.
+     * The intended use is for when some session invocations need to run in a
+     * background pool limited to a small number of threads:
+     * - For example, a session may be configured to have one large pool (for
+     * regular compute) and one small pool (for periodic, low priority work);
+     * using the small pool is currently the mechanism for limiting the inter-op
+     * parallelism of the low priority work.  Note that it does not limit the
+     * parallelism of work spawned by a single op kernel implementation.
+     * - Using this setting is normally not needed in training, but may help some
+     * serving use cases.
+     * - It is also generally recommended to set the global_name field of this
+     * proto, to avoid creating multiple large pools. It is typically better to
+     * run the non-low-priority work, even across sessions, in a single large
+     * pool.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.ThreadPoolOptionProto session_inter_op_thread_pool = 12;</code>
+     */
+    public Builder setSessionInterOpThreadPool(
+        int index, org.tensorflow.proto.framework.ThreadPoolOptionProto value) {
+      if (sessionInterOpThreadPoolBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureSessionInterOpThreadPoolIsMutable();
+        sessionInterOpThreadPool_.set(index, value);
+        onChanged();
+      } else {
+        sessionInterOpThreadPoolBuilder_.setMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * This option is experimental - it may be replaced with a different mechanism
+     * in the future.
+     * Configures session thread pools. If this is configured, then RunOptions for
+     * a Run call can select the thread pool to use.
+     * The intended use is for when some session invocations need to run in a
+     * background pool limited to a small number of threads:
+     * - For example, a session may be configured to have one large pool (for
+     * regular compute) and one small pool (for periodic, low priority work);
+     * using the small pool is currently the mechanism for limiting the inter-op
+     * parallelism of the low priority work.  Note that it does not limit the
+     * parallelism of work spawned by a single op kernel implementation.
+     * - Using this setting is normally not needed in training, but may help some
+     * serving use cases.
+     * - It is also generally recommended to set the global_name field of this
+     * proto, to avoid creating multiple large pools. It is typically better to
+     * run the non-low-priority work, even across sessions, in a single large
+     * pool.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.ThreadPoolOptionProto session_inter_op_thread_pool = 12;</code>
+     */
+    public Builder setSessionInterOpThreadPool(
+        int index, org.tensorflow.proto.framework.ThreadPoolOptionProto.Builder builderForValue) {
+      if (sessionInterOpThreadPoolBuilder_ == null) {
+        ensureSessionInterOpThreadPoolIsMutable();
+        sessionInterOpThreadPool_.set(index, builderForValue.build());
+        onChanged();
+      } else {
+        sessionInterOpThreadPoolBuilder_.setMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * This option is experimental - it may be replaced with a different mechanism
+     * in the future.
+     * Configures session thread pools. If this is configured, then RunOptions for
+     * a Run call can select the thread pool to use.
+     * The intended use is for when some session invocations need to run in a
+     * background pool limited to a small number of threads:
+     * - For example, a session may be configured to have one large pool (for
+     * regular compute) and one small pool (for periodic, low priority work);
+     * using the small pool is currently the mechanism for limiting the inter-op
+     * parallelism of the low priority work.  Note that it does not limit the
+     * parallelism of work spawned by a single op kernel implementation.
+     * - Using this setting is normally not needed in training, but may help some
+     * serving use cases.
+     * - It is also generally recommended to set the global_name field of this
+     * proto, to avoid creating multiple large pools. It is typically better to
+     * run the non-low-priority work, even across sessions, in a single large
+     * pool.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.ThreadPoolOptionProto session_inter_op_thread_pool = 12;</code>
+     */
+    public Builder addSessionInterOpThreadPool(org.tensorflow.proto.framework.ThreadPoolOptionProto value) {
+      if (sessionInterOpThreadPoolBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureSessionInterOpThreadPoolIsMutable();
+        sessionInterOpThreadPool_.add(value);
+        onChanged();
+      } else {
+        sessionInterOpThreadPoolBuilder_.addMessage(value);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * This option is experimental - it may be replaced with a different mechanism
+     * in the future.
+     * Configures session thread pools. If this is configured, then RunOptions for
+     * a Run call can select the thread pool to use.
+     * The intended use is for when some session invocations need to run in a
+     * background pool limited to a small number of threads:
+     * - For example, a session may be configured to have one large pool (for
+     * regular compute) and one small pool (for periodic, low priority work);
+     * using the small pool is currently the mechanism for limiting the inter-op
+     * parallelism of the low priority work.  Note that it does not limit the
+     * parallelism of work spawned by a single op kernel implementation.
+     * - Using this setting is normally not needed in training, but may help some
+     * serving use cases.
+     * - It is also generally recommended to set the global_name field of this
+     * proto, to avoid creating multiple large pools. It is typically better to
+     * run the non-low-priority work, even across sessions, in a single large
+     * pool.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.ThreadPoolOptionProto session_inter_op_thread_pool = 12;</code>
+     */
+    public Builder addSessionInterOpThreadPool(
+        int index, org.tensorflow.proto.framework.ThreadPoolOptionProto value) {
+      if (sessionInterOpThreadPoolBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureSessionInterOpThreadPoolIsMutable();
+        sessionInterOpThreadPool_.add(index, value);
+        onChanged();
+      } else {
+        sessionInterOpThreadPoolBuilder_.addMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * This option is experimental - it may be replaced with a different mechanism
+     * in the future.
+     * Configures session thread pools. If this is configured, then RunOptions for
+     * a Run call can select the thread pool to use.
+     * The intended use is for when some session invocations need to run in a
+     * background pool limited to a small number of threads:
+     * - For example, a session may be configured to have one large pool (for
+     * regular compute) and one small pool (for periodic, low priority work);
+     * using the small pool is currently the mechanism for limiting the inter-op
+     * parallelism of the low priority work.  Note that it does not limit the
+     * parallelism of work spawned by a single op kernel implementation.
+     * - Using this setting is normally not needed in training, but may help some
+     * serving use cases.
+     * - It is also generally recommended to set the global_name field of this
+     * proto, to avoid creating multiple large pools. It is typically better to
+     * run the non-low-priority work, even across sessions, in a single large
+     * pool.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.ThreadPoolOptionProto session_inter_op_thread_pool = 12;</code>
+     */
+    public Builder addSessionInterOpThreadPool(
+        org.tensorflow.proto.framework.ThreadPoolOptionProto.Builder builderForValue) {
+      if (sessionInterOpThreadPoolBuilder_ == null) {
+        ensureSessionInterOpThreadPoolIsMutable();
+        sessionInterOpThreadPool_.add(builderForValue.build());
+        onChanged();
+      } else {
+        sessionInterOpThreadPoolBuilder_.addMessage(builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * This option is experimental - it may be replaced with a different mechanism
+     * in the future.
+     * Configures session thread pools. If this is configured, then RunOptions for
+     * a Run call can select the thread pool to use.
+     * The intended use is for when some session invocations need to run in a
+     * background pool limited to a small number of threads:
+     * - For example, a session may be configured to have one large pool (for
+     * regular compute) and one small pool (for periodic, low priority work);
+     * using the small pool is currently the mechanism for limiting the inter-op
+     * parallelism of the low priority work.  Note that it does not limit the
+     * parallelism of work spawned by a single op kernel implementation.
+     * - Using this setting is normally not needed in training, but may help some
+     * serving use cases.
+     * - It is also generally recommended to set the global_name field of this
+     * proto, to avoid creating multiple large pools. It is typically better to
+     * run the non-low-priority work, even across sessions, in a single large
+     * pool.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.ThreadPoolOptionProto session_inter_op_thread_pool = 12;</code>
+     */
+    public Builder addSessionInterOpThreadPool(
+        int index, org.tensorflow.proto.framework.ThreadPoolOptionProto.Builder builderForValue) {
+      if (sessionInterOpThreadPoolBuilder_ == null) {
+        ensureSessionInterOpThreadPoolIsMutable();
+        sessionInterOpThreadPool_.add(index, builderForValue.build());
+        onChanged();
+      } else {
+        sessionInterOpThreadPoolBuilder_.addMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * This option is experimental - it may be replaced with a different mechanism
+     * in the future.
+     * Configures session thread pools. If this is configured, then RunOptions for
+     * a Run call can select the thread pool to use.
+     * The intended use is for when some session invocations need to run in a
+     * background pool limited to a small number of threads:
+     * - For example, a session may be configured to have one large pool (for
+     * regular compute) and one small pool (for periodic, low priority work);
+     * using the small pool is currently the mechanism for limiting the inter-op
+     * parallelism of the low priority work.  Note that it does not limit the
+     * parallelism of work spawned by a single op kernel implementation.
+     * - Using this setting is normally not needed in training, but may help some
+     * serving use cases.
+     * - It is also generally recommended to set the global_name field of this
+     * proto, to avoid creating multiple large pools. It is typically better to
+     * run the non-low-priority work, even across sessions, in a single large
+     * pool.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.ThreadPoolOptionProto session_inter_op_thread_pool = 12;</code>
+     */
+    public Builder addAllSessionInterOpThreadPool(
+        java.lang.Iterable<? extends org.tensorflow.proto.framework.ThreadPoolOptionProto> values) {
+      if (sessionInterOpThreadPoolBuilder_ == null) {
+        ensureSessionInterOpThreadPoolIsMutable();
+        com.google.protobuf.AbstractMessageLite.Builder.addAll(
+            values, sessionInterOpThreadPool_);
+        onChanged();
+      } else {
+        sessionInterOpThreadPoolBuilder_.addAllMessages(values);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * This option is experimental - it may be replaced with a different mechanism
+     * in the future.
+     * Configures session thread pools. If this is configured, then RunOptions for
+     * a Run call can select the thread pool to use.
+     * The intended use is for when some session invocations need to run in a
+     * background pool limited to a small number of threads:
+     * - For example, a session may be configured to have one large pool (for
+     * regular compute) and one small pool (for periodic, low priority work);
+     * using the small pool is currently the mechanism for limiting the inter-op
+     * parallelism of the low priority work.  Note that it does not limit the
+     * parallelism of work spawned by a single op kernel implementation.
+     * - Using this setting is normally not needed in training, but may help some
+     * serving use cases.
+     * - It is also generally recommended to set the global_name field of this
+     * proto, to avoid creating multiple large pools. It is typically better to
+     * run the non-low-priority work, even across sessions, in a single large
+     * pool.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.ThreadPoolOptionProto session_inter_op_thread_pool = 12;</code>
+     */
+    public Builder clearSessionInterOpThreadPool() {
+      if (sessionInterOpThreadPoolBuilder_ == null) {
+        sessionInterOpThreadPool_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000002);
+        onChanged();
+      } else {
+        sessionInterOpThreadPoolBuilder_.clear();
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * This option is experimental - it may be replaced with a different mechanism
+     * in the future.
+     * Configures session thread pools. If this is configured, then RunOptions for
+     * a Run call can select the thread pool to use.
+     * The intended use is for when some session invocations need to run in a
+     * background pool limited to a small number of threads:
+     * - For example, a session may be configured to have one large pool (for
+     * regular compute) and one small pool (for periodic, low priority work);
+     * using the small pool is currently the mechanism for limiting the inter-op
+     * parallelism of the low priority work.  Note that it does not limit the
+     * parallelism of work spawned by a single op kernel implementation.
+     * - Using this setting is normally not needed in training, but may help some
+     * serving use cases.
+     * - It is also generally recommended to set the global_name field of this
+     * proto, to avoid creating multiple large pools. It is typically better to
+     * run the non-low-priority work, even across sessions, in a single large
+     * pool.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.ThreadPoolOptionProto session_inter_op_thread_pool = 12;</code>
+     */
+    public Builder removeSessionInterOpThreadPool(int index) {
+      if (sessionInterOpThreadPoolBuilder_ == null) {
+        ensureSessionInterOpThreadPoolIsMutable();
+        sessionInterOpThreadPool_.remove(index);
+        onChanged();
+      } else {
+        sessionInterOpThreadPoolBuilder_.remove(index);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * This option is experimental - it may be replaced with a different mechanism
+     * in the future.
+     * Configures session thread pools. If this is configured, then RunOptions for
+     * a Run call can select the thread pool to use.
+     * The intended use is for when some session invocations need to run in a
+     * background pool limited to a small number of threads:
+     * - For example, a session may be configured to have one large pool (for
+     * regular compute) and one small pool (for periodic, low priority work);
+     * using the small pool is currently the mechanism for limiting the inter-op
+     * parallelism of the low priority work.  Note that it does not limit the
+     * parallelism of work spawned by a single op kernel implementation.
+     * - Using this setting is normally not needed in training, but may help some
+     * serving use cases.
+     * - It is also generally recommended to set the global_name field of this
+     * proto, to avoid creating multiple large pools. It is typically better to
+     * run the non-low-priority work, even across sessions, in a single large
+     * pool.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.ThreadPoolOptionProto session_inter_op_thread_pool = 12;</code>
+     */
+    public org.tensorflow.proto.framework.ThreadPoolOptionProto.Builder getSessionInterOpThreadPoolBuilder(
+        int index) {
+      return getSessionInterOpThreadPoolFieldBuilder().getBuilder(index);
+    }
+    /**
+     * <pre>
+     * This option is experimental - it may be replaced with a different mechanism
+     * in the future.
+     * Configures session thread pools. If this is configured, then RunOptions for
+     * a Run call can select the thread pool to use.
+     * The intended use is for when some session invocations need to run in a
+     * background pool limited to a small number of threads:
+     * - For example, a session may be configured to have one large pool (for
+     * regular compute) and one small pool (for periodic, low priority work);
+     * using the small pool is currently the mechanism for limiting the inter-op
+     * parallelism of the low priority work.  Note that it does not limit the
+     * parallelism of work spawned by a single op kernel implementation.
+     * - Using this setting is normally not needed in training, but may help some
+     * serving use cases.
+     * - It is also generally recommended to set the global_name field of this
+     * proto, to avoid creating multiple large pools. It is typically better to
+     * run the non-low-priority work, even across sessions, in a single large
+     * pool.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.ThreadPoolOptionProto session_inter_op_thread_pool = 12;</code>
+     */
+    public org.tensorflow.proto.framework.ThreadPoolOptionProtoOrBuilder getSessionInterOpThreadPoolOrBuilder(
+        int index) {
+      if (sessionInterOpThreadPoolBuilder_ == null) {
+        return sessionInterOpThreadPool_.get(index);  } else {
+        return sessionInterOpThreadPoolBuilder_.getMessageOrBuilder(index);
+      }
+    }
+    /**
+     * <pre>
+     * This option is experimental - it may be replaced with a different mechanism
+     * in the future.
+     * Configures session thread pools. If this is configured, then RunOptions for
+     * a Run call can select the thread pool to use.
+     * The intended use is for when some session invocations need to run in a
+     * background pool limited to a small number of threads:
+     * - For example, a session may be configured to have one large pool (for
+     * regular compute) and one small pool (for periodic, low priority work);
+     * using the small pool is currently the mechanism for limiting the inter-op
+     * parallelism of the low priority work.  Note that it does not limit the
+     * parallelism of work spawned by a single op kernel implementation.
+     * - Using this setting is normally not needed in training, but may help some
+     * serving use cases.
+     * - It is also generally recommended to set the global_name field of this
+     * proto, to avoid creating multiple large pools. It is typically better to
+     * run the non-low-priority work, even across sessions, in a single large
+     * pool.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.ThreadPoolOptionProto session_inter_op_thread_pool = 12;</code>
+     */
+    public java.util.List<? extends org.tensorflow.proto.framework.ThreadPoolOptionProtoOrBuilder> 
+         getSessionInterOpThreadPoolOrBuilderList() {
+      if (sessionInterOpThreadPoolBuilder_ != null) {
+        return sessionInterOpThreadPoolBuilder_.getMessageOrBuilderList();
+      } else {
+        return java.util.Collections.unmodifiableList(sessionInterOpThreadPool_);
+      }
+    }
+    /**
+     * <pre>
+     * This option is experimental - it may be replaced with a different mechanism
+     * in the future.
+     * Configures session thread pools. If this is configured, then RunOptions for
+     * a Run call can select the thread pool to use.
+     * The intended use is for when some session invocations need to run in a
+     * background pool limited to a small number of threads:
+     * - For example, a session may be configured to have one large pool (for
+     * regular compute) and one small pool (for periodic, low priority work);
+     * using the small pool is currently the mechanism for limiting the inter-op
+     * parallelism of the low priority work.  Note that it does not limit the
+     * parallelism of work spawned by a single op kernel implementation.
+     * - Using this setting is normally not needed in training, but may help some
+     * serving use cases.
+     * - It is also generally recommended to set the global_name field of this
+     * proto, to avoid creating multiple large pools. It is typically better to
+     * run the non-low-priority work, even across sessions, in a single large
+     * pool.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.ThreadPoolOptionProto session_inter_op_thread_pool = 12;</code>
+     */
+    public org.tensorflow.proto.framework.ThreadPoolOptionProto.Builder addSessionInterOpThreadPoolBuilder() {
+      return getSessionInterOpThreadPoolFieldBuilder().addBuilder(
+          org.tensorflow.proto.framework.ThreadPoolOptionProto.getDefaultInstance());
+    }
+    /**
+     * <pre>
+     * This option is experimental - it may be replaced with a different mechanism
+     * in the future.
+     * Configures session thread pools. If this is configured, then RunOptions for
+     * a Run call can select the thread pool to use.
+     * The intended use is for when some session invocations need to run in a
+     * background pool limited to a small number of threads:
+     * - For example, a session may be configured to have one large pool (for
+     * regular compute) and one small pool (for periodic, low priority work);
+     * using the small pool is currently the mechanism for limiting the inter-op
+     * parallelism of the low priority work.  Note that it does not limit the
+     * parallelism of work spawned by a single op kernel implementation.
+     * - Using this setting is normally not needed in training, but may help some
+     * serving use cases.
+     * - It is also generally recommended to set the global_name field of this
+     * proto, to avoid creating multiple large pools. It is typically better to
+     * run the non-low-priority work, even across sessions, in a single large
+     * pool.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.ThreadPoolOptionProto session_inter_op_thread_pool = 12;</code>
+     */
+    public org.tensorflow.proto.framework.ThreadPoolOptionProto.Builder addSessionInterOpThreadPoolBuilder(
+        int index) {
+      return getSessionInterOpThreadPoolFieldBuilder().addBuilder(
+          index, org.tensorflow.proto.framework.ThreadPoolOptionProto.getDefaultInstance());
+    }
+    /**
+     * <pre>
+     * This option is experimental - it may be replaced with a different mechanism
+     * in the future.
+     * Configures session thread pools. If this is configured, then RunOptions for
+     * a Run call can select the thread pool to use.
+     * The intended use is for when some session invocations need to run in a
+     * background pool limited to a small number of threads:
+     * - For example, a session may be configured to have one large pool (for
+     * regular compute) and one small pool (for periodic, low priority work);
+     * using the small pool is currently the mechanism for limiting the inter-op
+     * parallelism of the low priority work.  Note that it does not limit the
+     * parallelism of work spawned by a single op kernel implementation.
+     * - Using this setting is normally not needed in training, but may help some
+     * serving use cases.
+     * - It is also generally recommended to set the global_name field of this
+     * proto, to avoid creating multiple large pools. It is typically better to
+     * run the non-low-priority work, even across sessions, in a single large
+     * pool.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.ThreadPoolOptionProto session_inter_op_thread_pool = 12;</code>
+     */
+    public java.util.List<org.tensorflow.proto.framework.ThreadPoolOptionProto.Builder> 
+         getSessionInterOpThreadPoolBuilderList() {
+      return getSessionInterOpThreadPoolFieldBuilder().getBuilderList();
+    }
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.framework.ThreadPoolOptionProto, org.tensorflow.proto.framework.ThreadPoolOptionProto.Builder, org.tensorflow.proto.framework.ThreadPoolOptionProtoOrBuilder> 
+        getSessionInterOpThreadPoolFieldBuilder() {
+      if (sessionInterOpThreadPoolBuilder_ == null) {
+        sessionInterOpThreadPoolBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
+            org.tensorflow.proto.framework.ThreadPoolOptionProto, org.tensorflow.proto.framework.ThreadPoolOptionProto.Builder, org.tensorflow.proto.framework.ThreadPoolOptionProtoOrBuilder>(
+                sessionInterOpThreadPool_,
+                ((bitField0_ & 0x00000002) != 0),
+                getParentForChildren(),
+                isClean());
+        sessionInterOpThreadPool_ = null;
+      }
+      return sessionInterOpThreadPoolBuilder_;
+    }
+
+    private int placementPeriod_ ;
+    /**
+     * <pre>
+     * Assignment of Nodes to Devices is recomputed every placement_period
+     * steps until the system warms up (at which point the recomputation
+     * typically slows down automatically).
+     * </pre>
+     *
+     * <code>int32 placement_period = 3;</code>
+     */
+    public int getPlacementPeriod() {
+      return placementPeriod_;
+    }
+    /**
+     * <pre>
+     * Assignment of Nodes to Devices is recomputed every placement_period
+     * steps until the system warms up (at which point the recomputation
+     * typically slows down automatically).
+     * </pre>
+     *
+     * <code>int32 placement_period = 3;</code>
+     */
+    public Builder setPlacementPeriod(int value) {
+      
+      placementPeriod_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Assignment of Nodes to Devices is recomputed every placement_period
+     * steps until the system warms up (at which point the recomputation
+     * typically slows down automatically).
+     * </pre>
+     *
+     * <code>int32 placement_period = 3;</code>
+     */
+    public Builder clearPlacementPeriod() {
+      
+      placementPeriod_ = 0;
+      onChanged();
+      return this;
+    }
+
+    private com.google.protobuf.LazyStringList deviceFilters_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+    private void ensureDeviceFiltersIsMutable() {
+      if (!((bitField0_ & 0x00000004) != 0)) {
+        deviceFilters_ = new com.google.protobuf.LazyStringArrayList(deviceFilters_);
+        bitField0_ |= 0x00000004;
+       }
+    }
+    /**
+     * <pre>
+     * When any filters are present sessions will ignore all devices which do not
+     * match the filters. Each filter can be partially specified, e.g. "/job:ps"
+     * "/job:worker/replica:3", etc.
+     * </pre>
+     *
+     * <code>repeated string device_filters = 4;</code>
+     */
+    public com.google.protobuf.ProtocolStringList
+        getDeviceFiltersList() {
+      return deviceFilters_.getUnmodifiableView();
+    }
+    /**
+     * <pre>
+     * When any filters are present sessions will ignore all devices which do not
+     * match the filters. Each filter can be partially specified, e.g. "/job:ps"
+     * "/job:worker/replica:3", etc.
+     * </pre>
+     *
+     * <code>repeated string device_filters = 4;</code>
+     */
+    public int getDeviceFiltersCount() {
+      return deviceFilters_.size();
+    }
+    /**
+     * <pre>
+     * When any filters are present sessions will ignore all devices which do not
+     * match the filters. Each filter can be partially specified, e.g. "/job:ps"
+     * "/job:worker/replica:3", etc.
+     * </pre>
+     *
+     * <code>repeated string device_filters = 4;</code>
+     */
+    public java.lang.String getDeviceFilters(int index) {
+      return deviceFilters_.get(index);
+    }
+    /**
+     * <pre>
+     * When any filters are present sessions will ignore all devices which do not
+     * match the filters. Each filter can be partially specified, e.g. "/job:ps"
+     * "/job:worker/replica:3", etc.
+     * </pre>
+     *
+     * <code>repeated string device_filters = 4;</code>
+     */
+    public com.google.protobuf.ByteString
+        getDeviceFiltersBytes(int index) {
+      return deviceFilters_.getByteString(index);
+    }
+    /**
+     * <pre>
+     * When any filters are present sessions will ignore all devices which do not
+     * match the filters. Each filter can be partially specified, e.g. "/job:ps"
+     * "/job:worker/replica:3", etc.
+     * </pre>
+     *
+     * <code>repeated string device_filters = 4;</code>
+     */
+    public Builder setDeviceFilters(
+        int index, java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  ensureDeviceFiltersIsMutable();
+      deviceFilters_.set(index, value);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * When any filters are present sessions will ignore all devices which do not
+     * match the filters. Each filter can be partially specified, e.g. "/job:ps"
+     * "/job:worker/replica:3", etc.
+     * </pre>
+     *
+     * <code>repeated string device_filters = 4;</code>
+     */
+    public Builder addDeviceFilters(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  ensureDeviceFiltersIsMutable();
+      deviceFilters_.add(value);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * When any filters are present sessions will ignore all devices which do not
+     * match the filters. Each filter can be partially specified, e.g. "/job:ps"
+     * "/job:worker/replica:3", etc.
+     * </pre>
+     *
+     * <code>repeated string device_filters = 4;</code>
+     */
+    public Builder addAllDeviceFilters(
+        java.lang.Iterable<java.lang.String> values) {
+      ensureDeviceFiltersIsMutable();
+      com.google.protobuf.AbstractMessageLite.Builder.addAll(
+          values, deviceFilters_);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * When any filters are present sessions will ignore all devices which do not
+     * match the filters. Each filter can be partially specified, e.g. "/job:ps"
+     * "/job:worker/replica:3", etc.
+     * </pre>
+     *
+     * <code>repeated string device_filters = 4;</code>
+     */
+    public Builder clearDeviceFilters() {
+      deviceFilters_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+      bitField0_ = (bitField0_ & ~0x00000004);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * When any filters are present sessions will ignore all devices which do not
+     * match the filters. Each filter can be partially specified, e.g. "/job:ps"
+     * "/job:worker/replica:3", etc.
+     * </pre>
+     *
+     * <code>repeated string device_filters = 4;</code>
+     */
+    public Builder addDeviceFiltersBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      ensureDeviceFiltersIsMutable();
+      deviceFilters_.add(value);
+      onChanged();
+      return this;
+    }
+
+    private org.tensorflow.proto.framework.GPUOptions gpuOptions_;
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.GPUOptions, org.tensorflow.proto.framework.GPUOptions.Builder, org.tensorflow.proto.framework.GPUOptionsOrBuilder> gpuOptionsBuilder_;
+    /**
+     * <pre>
+     * Options that apply to all GPUs.
+     * </pre>
+     *
+     * <code>.tensorflow.GPUOptions gpu_options = 6;</code>
+     */
+    public boolean hasGpuOptions() {
+      return gpuOptionsBuilder_ != null || gpuOptions_ != null;
+    }
+    /**
+     * <pre>
+     * Options that apply to all GPUs.
+     * </pre>
+     *
+     * <code>.tensorflow.GPUOptions gpu_options = 6;</code>
+     */
+    public org.tensorflow.proto.framework.GPUOptions getGpuOptions() {
+      if (gpuOptionsBuilder_ == null) {
+        return gpuOptions_ == null ? org.tensorflow.proto.framework.GPUOptions.getDefaultInstance() : gpuOptions_;
+      } else {
+        return gpuOptionsBuilder_.getMessage();
+      }
+    }
+    /**
+     * <pre>
+     * Options that apply to all GPUs.
+     * </pre>
+     *
+     * <code>.tensorflow.GPUOptions gpu_options = 6;</code>
+     */
+    public Builder setGpuOptions(org.tensorflow.proto.framework.GPUOptions value) {
+      if (gpuOptionsBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        gpuOptions_ = value;
+        onChanged();
+      } else {
+        gpuOptionsBuilder_.setMessage(value);
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Options that apply to all GPUs.
+     * </pre>
+     *
+     * <code>.tensorflow.GPUOptions gpu_options = 6;</code>
+     */
+    public Builder setGpuOptions(
+        org.tensorflow.proto.framework.GPUOptions.Builder builderForValue) {
+      if (gpuOptionsBuilder_ == null) {
+        gpuOptions_ = builderForValue.build();
+        onChanged();
+      } else {
+        gpuOptionsBuilder_.setMessage(builderForValue.build());
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Options that apply to all GPUs.
+     * </pre>
+     *
+     * <code>.tensorflow.GPUOptions gpu_options = 6;</code>
+     */
+    public Builder mergeGpuOptions(org.tensorflow.proto.framework.GPUOptions value) {
+      if (gpuOptionsBuilder_ == null) {
+        if (gpuOptions_ != null) {
+          gpuOptions_ =
+            org.tensorflow.proto.framework.GPUOptions.newBuilder(gpuOptions_).mergeFrom(value).buildPartial();
+        } else {
+          gpuOptions_ = value;
+        }
+        onChanged();
+      } else {
+        gpuOptionsBuilder_.mergeFrom(value);
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Options that apply to all GPUs.
+     * </pre>
+     *
+     * <code>.tensorflow.GPUOptions gpu_options = 6;</code>
+     */
+    public Builder clearGpuOptions() {
+      if (gpuOptionsBuilder_ == null) {
+        gpuOptions_ = null;
+        onChanged();
+      } else {
+        gpuOptions_ = null;
+        gpuOptionsBuilder_ = null;
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Options that apply to all GPUs.
+     * </pre>
+     *
+     * <code>.tensorflow.GPUOptions gpu_options = 6;</code>
+     */
+    public org.tensorflow.proto.framework.GPUOptions.Builder getGpuOptionsBuilder() {
+      
+      onChanged();
+      return getGpuOptionsFieldBuilder().getBuilder();
+    }
+    /**
+     * <pre>
+     * Options that apply to all GPUs.
+     * </pre>
+     *
+     * <code>.tensorflow.GPUOptions gpu_options = 6;</code>
+     */
+    public org.tensorflow.proto.framework.GPUOptionsOrBuilder getGpuOptionsOrBuilder() {
+      if (gpuOptionsBuilder_ != null) {
+        return gpuOptionsBuilder_.getMessageOrBuilder();
+      } else {
+        return gpuOptions_ == null ?
+            org.tensorflow.proto.framework.GPUOptions.getDefaultInstance() : gpuOptions_;
+      }
+    }
+    /**
+     * <pre>
+     * Options that apply to all GPUs.
+     * </pre>
+     *
+     * <code>.tensorflow.GPUOptions gpu_options = 6;</code>
+     */
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.GPUOptions, org.tensorflow.proto.framework.GPUOptions.Builder, org.tensorflow.proto.framework.GPUOptionsOrBuilder> 
+        getGpuOptionsFieldBuilder() {
+      if (gpuOptionsBuilder_ == null) {
+        gpuOptionsBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+            org.tensorflow.proto.framework.GPUOptions, org.tensorflow.proto.framework.GPUOptions.Builder, org.tensorflow.proto.framework.GPUOptionsOrBuilder>(
+                getGpuOptions(),
+                getParentForChildren(),
+                isClean());
+        gpuOptions_ = null;
+      }
+      return gpuOptionsBuilder_;
+    }
+
+    private boolean allowSoftPlacement_ ;
+    /**
+     * <pre>
+     * Whether soft placement is allowed. If allow_soft_placement is true,
+     * an op will be placed on CPU if
+     *   1. there's no GPU implementation for the OP
+     * or
+     *   2. no GPU devices are known or registered
+     * or
+     *   3. need to co-locate with reftype input(s) which are from CPU.
+     * </pre>
+     *
+     * <code>bool allow_soft_placement = 7;</code>
+     */
+    public boolean getAllowSoftPlacement() {
+      return allowSoftPlacement_;
+    }
+    /**
+     * <pre>
+     * Whether soft placement is allowed. If allow_soft_placement is true,
+     * an op will be placed on CPU if
+     *   1. there's no GPU implementation for the OP
+     * or
+     *   2. no GPU devices are known or registered
+     * or
+     *   3. need to co-locate with reftype input(s) which are from CPU.
+     * </pre>
+     *
+     * <code>bool allow_soft_placement = 7;</code>
+     */
+    public Builder setAllowSoftPlacement(boolean value) {
+      
+      allowSoftPlacement_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Whether soft placement is allowed. If allow_soft_placement is true,
+     * an op will be placed on CPU if
+     *   1. there's no GPU implementation for the OP
+     * or
+     *   2. no GPU devices are known or registered
+     * or
+     *   3. need to co-locate with reftype input(s) which are from CPU.
+     * </pre>
+     *
+     * <code>bool allow_soft_placement = 7;</code>
+     */
+    public Builder clearAllowSoftPlacement() {
+      
+      allowSoftPlacement_ = false;
+      onChanged();
+      return this;
+    }
+
+    private boolean logDevicePlacement_ ;
+    /**
+     * <pre>
+     * Whether device placements should be logged.
+     * </pre>
+     *
+     * <code>bool log_device_placement = 8;</code>
+     */
+    public boolean getLogDevicePlacement() {
+      return logDevicePlacement_;
+    }
+    /**
+     * <pre>
+     * Whether device placements should be logged.
+     * </pre>
+     *
+     * <code>bool log_device_placement = 8;</code>
+     */
+    public Builder setLogDevicePlacement(boolean value) {
+      
+      logDevicePlacement_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Whether device placements should be logged.
+     * </pre>
+     *
+     * <code>bool log_device_placement = 8;</code>
+     */
+    public Builder clearLogDevicePlacement() {
+      
+      logDevicePlacement_ = false;
+      onChanged();
+      return this;
+    }
+
+    private org.tensorflow.proto.framework.GraphOptions graphOptions_;
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.GraphOptions, org.tensorflow.proto.framework.GraphOptions.Builder, org.tensorflow.proto.framework.GraphOptionsOrBuilder> graphOptionsBuilder_;
+    /**
+     * <pre>
+     * Options that apply to all graphs.
+     * </pre>
+     *
+     * <code>.tensorflow.GraphOptions graph_options = 10;</code>
+     */
+    public boolean hasGraphOptions() {
+      return graphOptionsBuilder_ != null || graphOptions_ != null;
+    }
+    /**
+     * <pre>
+     * Options that apply to all graphs.
+     * </pre>
+     *
+     * <code>.tensorflow.GraphOptions graph_options = 10;</code>
+     */
+    public org.tensorflow.proto.framework.GraphOptions getGraphOptions() {
+      if (graphOptionsBuilder_ == null) {
+        return graphOptions_ == null ? org.tensorflow.proto.framework.GraphOptions.getDefaultInstance() : graphOptions_;
+      } else {
+        return graphOptionsBuilder_.getMessage();
+      }
+    }
+    /**
+     * <pre>
+     * Options that apply to all graphs.
+     * </pre>
+     *
+     * <code>.tensorflow.GraphOptions graph_options = 10;</code>
+     */
+    public Builder setGraphOptions(org.tensorflow.proto.framework.GraphOptions value) {
+      if (graphOptionsBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        graphOptions_ = value;
+        onChanged();
+      } else {
+        graphOptionsBuilder_.setMessage(value);
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Options that apply to all graphs.
+     * </pre>
+     *
+     * <code>.tensorflow.GraphOptions graph_options = 10;</code>
+     */
+    public Builder setGraphOptions(
+        org.tensorflow.proto.framework.GraphOptions.Builder builderForValue) {
+      if (graphOptionsBuilder_ == null) {
+        graphOptions_ = builderForValue.build();
+        onChanged();
+      } else {
+        graphOptionsBuilder_.setMessage(builderForValue.build());
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Options that apply to all graphs.
+     * </pre>
+     *
+     * <code>.tensorflow.GraphOptions graph_options = 10;</code>
+     */
+    public Builder mergeGraphOptions(org.tensorflow.proto.framework.GraphOptions value) {
+      if (graphOptionsBuilder_ == null) {
+        if (graphOptions_ != null) {
+          graphOptions_ =
+            org.tensorflow.proto.framework.GraphOptions.newBuilder(graphOptions_).mergeFrom(value).buildPartial();
+        } else {
+          graphOptions_ = value;
+        }
+        onChanged();
+      } else {
+        graphOptionsBuilder_.mergeFrom(value);
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Options that apply to all graphs.
+     * </pre>
+     *
+     * <code>.tensorflow.GraphOptions graph_options = 10;</code>
+     */
+    public Builder clearGraphOptions() {
+      if (graphOptionsBuilder_ == null) {
+        graphOptions_ = null;
+        onChanged();
+      } else {
+        graphOptions_ = null;
+        graphOptionsBuilder_ = null;
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Options that apply to all graphs.
+     * </pre>
+     *
+     * <code>.tensorflow.GraphOptions graph_options = 10;</code>
+     */
+    public org.tensorflow.proto.framework.GraphOptions.Builder getGraphOptionsBuilder() {
+      
+      onChanged();
+      return getGraphOptionsFieldBuilder().getBuilder();
+    }
+    /**
+     * <pre>
+     * Options that apply to all graphs.
+     * </pre>
+     *
+     * <code>.tensorflow.GraphOptions graph_options = 10;</code>
+     */
+    public org.tensorflow.proto.framework.GraphOptionsOrBuilder getGraphOptionsOrBuilder() {
+      if (graphOptionsBuilder_ != null) {
+        return graphOptionsBuilder_.getMessageOrBuilder();
+      } else {
+        return graphOptions_ == null ?
+            org.tensorflow.proto.framework.GraphOptions.getDefaultInstance() : graphOptions_;
+      }
+    }
+    /**
+     * <pre>
+     * Options that apply to all graphs.
+     * </pre>
+     *
+     * <code>.tensorflow.GraphOptions graph_options = 10;</code>
+     */
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.GraphOptions, org.tensorflow.proto.framework.GraphOptions.Builder, org.tensorflow.proto.framework.GraphOptionsOrBuilder> 
+        getGraphOptionsFieldBuilder() {
+      if (graphOptionsBuilder_ == null) {
+        graphOptionsBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+            org.tensorflow.proto.framework.GraphOptions, org.tensorflow.proto.framework.GraphOptions.Builder, org.tensorflow.proto.framework.GraphOptionsOrBuilder>(
+                getGraphOptions(),
+                getParentForChildren(),
+                isClean());
+        graphOptions_ = null;
+      }
+      return graphOptionsBuilder_;
+    }
+
+    private long operationTimeoutInMs_ ;
+    /**
+     * <pre>
+     * Global timeout for all blocking operations in this session.  If non-zero,
+     * and not overridden on a per-operation basis, this value will be used as the
+     * deadline for all blocking operations.
+     * </pre>
+     *
+     * <code>int64 operation_timeout_in_ms = 11;</code>
+     */
+    public long getOperationTimeoutInMs() {
+      return operationTimeoutInMs_;
+    }
+    /**
+     * <pre>
+     * Global timeout for all blocking operations in this session.  If non-zero,
+     * and not overridden on a per-operation basis, this value will be used as the
+     * deadline for all blocking operations.
+     * </pre>
+     *
+     * <code>int64 operation_timeout_in_ms = 11;</code>
+     */
+    public Builder setOperationTimeoutInMs(long value) {
+      
+      operationTimeoutInMs_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Global timeout for all blocking operations in this session.  If non-zero,
+     * and not overridden on a per-operation basis, this value will be used as the
+     * deadline for all blocking operations.
+     * </pre>
+     *
+     * <code>int64 operation_timeout_in_ms = 11;</code>
+     */
+    public Builder clearOperationTimeoutInMs() {
+      
+      operationTimeoutInMs_ = 0L;
+      onChanged();
+      return this;
+    }
+
+    private org.tensorflow.proto.framework.RPCOptions rpcOptions_;
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.RPCOptions, org.tensorflow.proto.framework.RPCOptions.Builder, org.tensorflow.proto.framework.RPCOptionsOrBuilder> rpcOptionsBuilder_;
+    /**
+     * <pre>
+     * Options that apply when this session uses the distributed runtime.
+     * </pre>
+     *
+     * <code>.tensorflow.RPCOptions rpc_options = 13;</code>
+     */
+    public boolean hasRpcOptions() {
+      return rpcOptionsBuilder_ != null || rpcOptions_ != null;
+    }
+    /**
+     * <pre>
+     * Options that apply when this session uses the distributed runtime.
+     * </pre>
+     *
+     * <code>.tensorflow.RPCOptions rpc_options = 13;</code>
+     */
+    public org.tensorflow.proto.framework.RPCOptions getRpcOptions() {
+      if (rpcOptionsBuilder_ == null) {
+        return rpcOptions_ == null ? org.tensorflow.proto.framework.RPCOptions.getDefaultInstance() : rpcOptions_;
+      } else {
+        return rpcOptionsBuilder_.getMessage();
+      }
+    }
+    /**
+     * <pre>
+     * Options that apply when this session uses the distributed runtime.
+     * </pre>
+     *
+     * <code>.tensorflow.RPCOptions rpc_options = 13;</code>
+     */
+    public Builder setRpcOptions(org.tensorflow.proto.framework.RPCOptions value) {
+      if (rpcOptionsBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        rpcOptions_ = value;
+        onChanged();
+      } else {
+        rpcOptionsBuilder_.setMessage(value);
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Options that apply when this session uses the distributed runtime.
+     * </pre>
+     *
+     * <code>.tensorflow.RPCOptions rpc_options = 13;</code>
+     */
+    public Builder setRpcOptions(
+        org.tensorflow.proto.framework.RPCOptions.Builder builderForValue) {
+      if (rpcOptionsBuilder_ == null) {
+        rpcOptions_ = builderForValue.build();
+        onChanged();
+      } else {
+        rpcOptionsBuilder_.setMessage(builderForValue.build());
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Options that apply when this session uses the distributed runtime.
+     * </pre>
+     *
+     * <code>.tensorflow.RPCOptions rpc_options = 13;</code>
+     */
+    public Builder mergeRpcOptions(org.tensorflow.proto.framework.RPCOptions value) {
+      if (rpcOptionsBuilder_ == null) {
+        if (rpcOptions_ != null) {
+          rpcOptions_ =
+            org.tensorflow.proto.framework.RPCOptions.newBuilder(rpcOptions_).mergeFrom(value).buildPartial();
+        } else {
+          rpcOptions_ = value;
+        }
+        onChanged();
+      } else {
+        rpcOptionsBuilder_.mergeFrom(value);
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Options that apply when this session uses the distributed runtime.
+     * </pre>
+     *
+     * <code>.tensorflow.RPCOptions rpc_options = 13;</code>
+     */
+    public Builder clearRpcOptions() {
+      if (rpcOptionsBuilder_ == null) {
+        rpcOptions_ = null;
+        onChanged();
+      } else {
+        rpcOptions_ = null;
+        rpcOptionsBuilder_ = null;
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Options that apply when this session uses the distributed runtime.
+     * </pre>
+     *
+     * <code>.tensorflow.RPCOptions rpc_options = 13;</code>
+     */
+    public org.tensorflow.proto.framework.RPCOptions.Builder getRpcOptionsBuilder() {
+      
+      onChanged();
+      return getRpcOptionsFieldBuilder().getBuilder();
+    }
+    /**
+     * <pre>
+     * Options that apply when this session uses the distributed runtime.
+     * </pre>
+     *
+     * <code>.tensorflow.RPCOptions rpc_options = 13;</code>
+     */
+    public org.tensorflow.proto.framework.RPCOptionsOrBuilder getRpcOptionsOrBuilder() {
+      if (rpcOptionsBuilder_ != null) {
+        return rpcOptionsBuilder_.getMessageOrBuilder();
+      } else {
+        return rpcOptions_ == null ?
+            org.tensorflow.proto.framework.RPCOptions.getDefaultInstance() : rpcOptions_;
+      }
+    }
+    /**
+     * <pre>
+     * Options that apply when this session uses the distributed runtime.
+     * </pre>
+     *
+     * <code>.tensorflow.RPCOptions rpc_options = 13;</code>
+     */
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.RPCOptions, org.tensorflow.proto.framework.RPCOptions.Builder, org.tensorflow.proto.framework.RPCOptionsOrBuilder> 
+        getRpcOptionsFieldBuilder() {
+      if (rpcOptionsBuilder_ == null) {
+        rpcOptionsBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+            org.tensorflow.proto.framework.RPCOptions, org.tensorflow.proto.framework.RPCOptions.Builder, org.tensorflow.proto.framework.RPCOptionsOrBuilder>(
+                getRpcOptions(),
+                getParentForChildren(),
+                isClean());
+        rpcOptions_ = null;
+      }
+      return rpcOptionsBuilder_;
+    }
+
+    private org.tensorflow.proto.distruntime.ClusterDef clusterDef_;
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.distruntime.ClusterDef, org.tensorflow.proto.distruntime.ClusterDef.Builder, org.tensorflow.proto.distruntime.ClusterDefOrBuilder> clusterDefBuilder_;
+    /**
+     * <pre>
+     * Optional list of all workers to use in this session.
+     * </pre>
+     *
+     * <code>.tensorflow.ClusterDef cluster_def = 14;</code>
+     */
+    public boolean hasClusterDef() {
+      return clusterDefBuilder_ != null || clusterDef_ != null;
+    }
+    /**
+     * <pre>
+     * Optional list of all workers to use in this session.
+     * </pre>
+     *
+     * <code>.tensorflow.ClusterDef cluster_def = 14;</code>
+     */
+    public org.tensorflow.proto.distruntime.ClusterDef getClusterDef() {
+      if (clusterDefBuilder_ == null) {
+        return clusterDef_ == null ? org.tensorflow.proto.distruntime.ClusterDef.getDefaultInstance() : clusterDef_;
+      } else {
+        return clusterDefBuilder_.getMessage();
+      }
+    }
+    /**
+     * <pre>
+     * Optional list of all workers to use in this session.
+     * </pre>
+     *
+     * <code>.tensorflow.ClusterDef cluster_def = 14;</code>
+     */
+    public Builder setClusterDef(org.tensorflow.proto.distruntime.ClusterDef value) {
+      if (clusterDefBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        clusterDef_ = value;
+        onChanged();
+      } else {
+        clusterDefBuilder_.setMessage(value);
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Optional list of all workers to use in this session.
+     * </pre>
+     *
+     * <code>.tensorflow.ClusterDef cluster_def = 14;</code>
+     */
+    public Builder setClusterDef(
+        org.tensorflow.proto.distruntime.ClusterDef.Builder builderForValue) {
+      if (clusterDefBuilder_ == null) {
+        clusterDef_ = builderForValue.build();
+        onChanged();
+      } else {
+        clusterDefBuilder_.setMessage(builderForValue.build());
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Optional list of all workers to use in this session.
+     * </pre>
+     *
+     * <code>.tensorflow.ClusterDef cluster_def = 14;</code>
+     */
+    public Builder mergeClusterDef(org.tensorflow.proto.distruntime.ClusterDef value) {
+      if (clusterDefBuilder_ == null) {
+        if (clusterDef_ != null) {
+          clusterDef_ =
+            org.tensorflow.proto.distruntime.ClusterDef.newBuilder(clusterDef_).mergeFrom(value).buildPartial();
+        } else {
+          clusterDef_ = value;
+        }
+        onChanged();
+      } else {
+        clusterDefBuilder_.mergeFrom(value);
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Optional list of all workers to use in this session.
+     * </pre>
+     *
+     * <code>.tensorflow.ClusterDef cluster_def = 14;</code>
+     */
+    public Builder clearClusterDef() {
+      if (clusterDefBuilder_ == null) {
+        clusterDef_ = null;
+        onChanged();
+      } else {
+        clusterDef_ = null;
+        clusterDefBuilder_ = null;
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Optional list of all workers to use in this session.
+     * </pre>
+     *
+     * <code>.tensorflow.ClusterDef cluster_def = 14;</code>
+     */
+    public org.tensorflow.proto.distruntime.ClusterDef.Builder getClusterDefBuilder() {
+      
+      onChanged();
+      return getClusterDefFieldBuilder().getBuilder();
+    }
+    /**
+     * <pre>
+     * Optional list of all workers to use in this session.
+     * </pre>
+     *
+     * <code>.tensorflow.ClusterDef cluster_def = 14;</code>
+     */
+    public org.tensorflow.proto.distruntime.ClusterDefOrBuilder getClusterDefOrBuilder() {
+      if (clusterDefBuilder_ != null) {
+        return clusterDefBuilder_.getMessageOrBuilder();
+      } else {
+        return clusterDef_ == null ?
+            org.tensorflow.proto.distruntime.ClusterDef.getDefaultInstance() : clusterDef_;
+      }
+    }
+    /**
+     * <pre>
+     * Optional list of all workers to use in this session.
+     * </pre>
+     *
+     * <code>.tensorflow.ClusterDef cluster_def = 14;</code>
+     */
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.distruntime.ClusterDef, org.tensorflow.proto.distruntime.ClusterDef.Builder, org.tensorflow.proto.distruntime.ClusterDefOrBuilder> 
+        getClusterDefFieldBuilder() {
+      if (clusterDefBuilder_ == null) {
+        clusterDefBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+            org.tensorflow.proto.distruntime.ClusterDef, org.tensorflow.proto.distruntime.ClusterDef.Builder, org.tensorflow.proto.distruntime.ClusterDefOrBuilder>(
+                getClusterDef(),
+                getParentForChildren(),
+                isClean());
+        clusterDef_ = null;
+      }
+      return clusterDefBuilder_;
+    }
+
+    private boolean isolateSessionState_ ;
+    /**
+     * <pre>
+     * If true, any resources such as Variables used in the session will not be
+     * shared with other sessions. However, when clusterspec propagation is
+     * enabled, this field is ignored and sessions are always isolated.
+     * </pre>
+     *
+     * <code>bool isolate_session_state = 15;</code>
+     */
+    public boolean getIsolateSessionState() {
+      return isolateSessionState_;
+    }
+    /**
+     * <pre>
+     * If true, any resources such as Variables used in the session will not be
+     * shared with other sessions. However, when clusterspec propagation is
+     * enabled, this field is ignored and sessions are always isolated.
+     * </pre>
+     *
+     * <code>bool isolate_session_state = 15;</code>
+     */
+    public Builder setIsolateSessionState(boolean value) {
+      
+      isolateSessionState_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * If true, any resources such as Variables used in the session will not be
+     * shared with other sessions. However, when clusterspec propagation is
+     * enabled, this field is ignored and sessions are always isolated.
+     * </pre>
+     *
+     * <code>bool isolate_session_state = 15;</code>
+     */
+    public Builder clearIsolateSessionState() {
+      
+      isolateSessionState_ = false;
+      onChanged();
+      return this;
+    }
+
+    private boolean shareClusterDevicesInSession_ ;
+    /**
+     * <pre>
+     * When true, WorkerSessions are created with device attributes from the
+     * full cluster.
+     * This is helpful when a worker wants to partition a graph
+     * (for example during a PartitionedCallOp).
+     * </pre>
+     *
+     * <code>bool share_cluster_devices_in_session = 17;</code>
+     */
+    public boolean getShareClusterDevicesInSession() {
+      return shareClusterDevicesInSession_;
+    }
+    /**
+     * <pre>
+     * When true, WorkerSessions are created with device attributes from the
+     * full cluster.
+     * This is helpful when a worker wants to partition a graph
+     * (for example during a PartitionedCallOp).
+     * </pre>
+     *
+     * <code>bool share_cluster_devices_in_session = 17;</code>
+     */
+    public Builder setShareClusterDevicesInSession(boolean value) {
+      
+      shareClusterDevicesInSession_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * When true, WorkerSessions are created with device attributes from the
+     * full cluster.
+     * This is helpful when a worker wants to partition a graph
+     * (for example during a PartitionedCallOp).
+     * </pre>
+     *
+     * <code>bool share_cluster_devices_in_session = 17;</code>
+     */
+    public Builder clearShareClusterDevicesInSession() {
+      
+      shareClusterDevicesInSession_ = false;
+      onChanged();
+      return this;
+    }
+
+    private org.tensorflow.proto.framework.ConfigProto.Experimental experimental_;
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.ConfigProto.Experimental, org.tensorflow.proto.framework.ConfigProto.Experimental.Builder, org.tensorflow.proto.framework.ConfigProto.ExperimentalOrBuilder> experimentalBuilder_;
+    /**
+     * <code>.tensorflow.ConfigProto.Experimental experimental = 16;</code>
+     */
+    public boolean hasExperimental() {
+      return experimentalBuilder_ != null || experimental_ != null;
+    }
+    /**
+     * <code>.tensorflow.ConfigProto.Experimental experimental = 16;</code>
+     */
+    public org.tensorflow.proto.framework.ConfigProto.Experimental getExperimental() {
+      if (experimentalBuilder_ == null) {
+        return experimental_ == null ? org.tensorflow.proto.framework.ConfigProto.Experimental.getDefaultInstance() : experimental_;
+      } else {
+        return experimentalBuilder_.getMessage();
+      }
+    }
+    /**
+     * <code>.tensorflow.ConfigProto.Experimental experimental = 16;</code>
+     */
+    public Builder setExperimental(org.tensorflow.proto.framework.ConfigProto.Experimental value) {
+      if (experimentalBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        experimental_ = value;
+        onChanged();
+      } else {
+        experimentalBuilder_.setMessage(value);
+      }
+
+      return this;
+    }
+    /**
+     * <code>.tensorflow.ConfigProto.Experimental experimental = 16;</code>
+     */
+    public Builder setExperimental(
+        org.tensorflow.proto.framework.ConfigProto.Experimental.Builder builderForValue) {
+      if (experimentalBuilder_ == null) {
+        experimental_ = builderForValue.build();
+        onChanged();
+      } else {
+        experimentalBuilder_.setMessage(builderForValue.build());
+      }
+
+      return this;
+    }
+    /**
+     * <code>.tensorflow.ConfigProto.Experimental experimental = 16;</code>
+     */
+    public Builder mergeExperimental(org.tensorflow.proto.framework.ConfigProto.Experimental value) {
+      if (experimentalBuilder_ == null) {
+        if (experimental_ != null) {
+          experimental_ =
+            org.tensorflow.proto.framework.ConfigProto.Experimental.newBuilder(experimental_).mergeFrom(value).buildPartial();
+        } else {
+          experimental_ = value;
+        }
+        onChanged();
+      } else {
+        experimentalBuilder_.mergeFrom(value);
+      }
+
+      return this;
+    }
+    /**
+     * <code>.tensorflow.ConfigProto.Experimental experimental = 16;</code>
+     */
+    public Builder clearExperimental() {
+      if (experimentalBuilder_ == null) {
+        experimental_ = null;
+        onChanged();
+      } else {
+        experimental_ = null;
+        experimentalBuilder_ = null;
+      }
+
+      return this;
+    }
+    /**
+     * <code>.tensorflow.ConfigProto.Experimental experimental = 16;</code>
+     */
+    public org.tensorflow.proto.framework.ConfigProto.Experimental.Builder getExperimentalBuilder() {
+      
+      onChanged();
+      return getExperimentalFieldBuilder().getBuilder();
+    }
+    /**
+     * <code>.tensorflow.ConfigProto.Experimental experimental = 16;</code>
+     */
+    public org.tensorflow.proto.framework.ConfigProto.ExperimentalOrBuilder getExperimentalOrBuilder() {
+      if (experimentalBuilder_ != null) {
+        return experimentalBuilder_.getMessageOrBuilder();
+      } else {
+        return experimental_ == null ?
+            org.tensorflow.proto.framework.ConfigProto.Experimental.getDefaultInstance() : experimental_;
+      }
+    }
+    /**
+     * <code>.tensorflow.ConfigProto.Experimental experimental = 16;</code>
+     */
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.ConfigProto.Experimental, org.tensorflow.proto.framework.ConfigProto.Experimental.Builder, org.tensorflow.proto.framework.ConfigProto.ExperimentalOrBuilder> 
+        getExperimentalFieldBuilder() {
+      if (experimentalBuilder_ == null) {
+        experimentalBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+            org.tensorflow.proto.framework.ConfigProto.Experimental, org.tensorflow.proto.framework.ConfigProto.Experimental.Builder, org.tensorflow.proto.framework.ConfigProto.ExperimentalOrBuilder>(
+                getExperimental(),
+                getParentForChildren(),
+                isClean());
+        experimental_ = null;
+      }
+      return experimentalBuilder_;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.ConfigProto)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.ConfigProto)
+  private static final org.tensorflow.proto.framework.ConfigProto DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.ConfigProto();
+  }
+
+  public static org.tensorflow.proto.framework.ConfigProto getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<ConfigProto>
+      PARSER = new com.google.protobuf.AbstractParser<ConfigProto>() {
+    @java.lang.Override
+    public ConfigProto parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new ConfigProto(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<ConfigProto> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<ConfigProto> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.framework.ConfigProto getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ConfigProtoOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ConfigProtoOrBuilder.java
new file mode 100644
index 00000000000..a13298b2bb5
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ConfigProtoOrBuilder.java
@@ -0,0 +1,477 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/config.proto
+
+package org.tensorflow.proto.framework;
+
+public interface ConfigProtoOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.ConfigProto)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <pre>
+   * Map from device type name (e.g., "CPU" or "GPU" ) to maximum
+   * number of devices of that type to use.  If a particular device
+   * type is not found in the map, the system picks an appropriate
+   * number.
+   * </pre>
+   *
+   * <code>map&lt;string, int32&gt; device_count = 1;</code>
+   */
+  int getDeviceCountCount();
+  /**
+   * <pre>
+   * Map from device type name (e.g., "CPU" or "GPU" ) to maximum
+   * number of devices of that type to use.  If a particular device
+   * type is not found in the map, the system picks an appropriate
+   * number.
+   * </pre>
+   *
+   * <code>map&lt;string, int32&gt; device_count = 1;</code>
+   */
+  boolean containsDeviceCount(
+      java.lang.String key);
+  /**
+   * Use {@link #getDeviceCountMap()} instead.
+   */
+  @java.lang.Deprecated
+  java.util.Map<java.lang.String, java.lang.Integer>
+  getDeviceCount();
+  /**
+   * <pre>
+   * Map from device type name (e.g., "CPU" or "GPU" ) to maximum
+   * number of devices of that type to use.  If a particular device
+   * type is not found in the map, the system picks an appropriate
+   * number.
+   * </pre>
+   *
+   * <code>map&lt;string, int32&gt; device_count = 1;</code>
+   */
+  java.util.Map<java.lang.String, java.lang.Integer>
+  getDeviceCountMap();
+  /**
+   * <pre>
+   * Map from device type name (e.g., "CPU" or "GPU" ) to maximum
+   * number of devices of that type to use.  If a particular device
+   * type is not found in the map, the system picks an appropriate
+   * number.
+   * </pre>
+   *
+   * <code>map&lt;string, int32&gt; device_count = 1;</code>
+   */
+
+  int getDeviceCountOrDefault(
+      java.lang.String key,
+      int defaultValue);
+  /**
+   * <pre>
+   * Map from device type name (e.g., "CPU" or "GPU" ) to maximum
+   * number of devices of that type to use.  If a particular device
+   * type is not found in the map, the system picks an appropriate
+   * number.
+   * </pre>
+   *
+   * <code>map&lt;string, int32&gt; device_count = 1;</code>
+   */
+
+  int getDeviceCountOrThrow(
+      java.lang.String key);
+
+  /**
+   * <pre>
+   * The execution of an individual op (for some op types) can be
+   * parallelized on a pool of intra_op_parallelism_threads.
+   * 0 means the system picks an appropriate number.
+   * If you create an ordinary session, e.g., from Python or C++,
+   * then there is exactly one intra op thread pool per process.
+   * The first session created determines the number of threads in this pool.
+   * All subsequent sessions reuse/share this one global pool.
+   * There are notable exceptions to the default behavior describe above:
+   * 1. There is an environment variable  for overriding this thread pool,
+   *    named TF_OVERRIDE_GLOBAL_THREADPOOL.
+   * 2. When connecting to a server, such as a remote `tf.train.Server`
+   *    instance, then this option will be ignored altogether.
+   * </pre>
+   *
+   * <code>int32 intra_op_parallelism_threads = 2;</code>
+   */
+  int getIntraOpParallelismThreads();
+
+  /**
+   * <pre>
+   * Nodes that perform blocking operations are enqueued on a pool of
+   * inter_op_parallelism_threads available in each process.
+   * 0 means the system picks an appropriate number.
+   * Negative means all operations are performed in caller's thread.
+   * Note that the first Session created in the process sets the
+   * number of threads for all future sessions unless use_per_session_threads is
+   * true or session_inter_op_thread_pool is configured.
+   * </pre>
+   *
+   * <code>int32 inter_op_parallelism_threads = 5;</code>
+   */
+  int getInterOpParallelismThreads();
+
+  /**
+   * <pre>
+   * If true, use a new set of threads for this session rather than the global
+   * pool of threads. Only supported by direct sessions.
+   * If false, use the global threads created by the first session, or the
+   * per-session thread pools configured by session_inter_op_thread_pool.
+   * This option is deprecated. The same effect can be achieved by setting
+   * session_inter_op_thread_pool to have one element, whose num_threads equals
+   * inter_op_parallelism_threads.
+   * </pre>
+   *
+   * <code>bool use_per_session_threads = 9;</code>
+   */
+  boolean getUsePerSessionThreads();
+
+  /**
+   * <pre>
+   * This option is experimental - it may be replaced with a different mechanism
+   * in the future.
+   * Configures session thread pools. If this is configured, then RunOptions for
+   * a Run call can select the thread pool to use.
+   * The intended use is for when some session invocations need to run in a
+   * background pool limited to a small number of threads:
+   * - For example, a session may be configured to have one large pool (for
+   * regular compute) and one small pool (for periodic, low priority work);
+   * using the small pool is currently the mechanism for limiting the inter-op
+   * parallelism of the low priority work.  Note that it does not limit the
+   * parallelism of work spawned by a single op kernel implementation.
+   * - Using this setting is normally not needed in training, but may help some
+   * serving use cases.
+   * - It is also generally recommended to set the global_name field of this
+   * proto, to avoid creating multiple large pools. It is typically better to
+   * run the non-low-priority work, even across sessions, in a single large
+   * pool.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.ThreadPoolOptionProto session_inter_op_thread_pool = 12;</code>
+   */
+  java.util.List<org.tensorflow.proto.framework.ThreadPoolOptionProto> 
+      getSessionInterOpThreadPoolList();
+  /**
+   * <pre>
+   * This option is experimental - it may be replaced with a different mechanism
+   * in the future.
+   * Configures session thread pools. If this is configured, then RunOptions for
+   * a Run call can select the thread pool to use.
+   * The intended use is for when some session invocations need to run in a
+   * background pool limited to a small number of threads:
+   * - For example, a session may be configured to have one large pool (for
+   * regular compute) and one small pool (for periodic, low priority work);
+   * using the small pool is currently the mechanism for limiting the inter-op
+   * parallelism of the low priority work.  Note that it does not limit the
+   * parallelism of work spawned by a single op kernel implementation.
+   * - Using this setting is normally not needed in training, but may help some
+   * serving use cases.
+   * - It is also generally recommended to set the global_name field of this
+   * proto, to avoid creating multiple large pools. It is typically better to
+   * run the non-low-priority work, even across sessions, in a single large
+   * pool.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.ThreadPoolOptionProto session_inter_op_thread_pool = 12;</code>
+   */
+  org.tensorflow.proto.framework.ThreadPoolOptionProto getSessionInterOpThreadPool(int index);
+  /**
+   * <pre>
+   * This option is experimental - it may be replaced with a different mechanism
+   * in the future.
+   * Configures session thread pools. If this is configured, then RunOptions for
+   * a Run call can select the thread pool to use.
+   * The intended use is for when some session invocations need to run in a
+   * background pool limited to a small number of threads:
+   * - For example, a session may be configured to have one large pool (for
+   * regular compute) and one small pool (for periodic, low priority work);
+   * using the small pool is currently the mechanism for limiting the inter-op
+   * parallelism of the low priority work.  Note that it does not limit the
+   * parallelism of work spawned by a single op kernel implementation.
+   * - Using this setting is normally not needed in training, but may help some
+   * serving use cases.
+   * - It is also generally recommended to set the global_name field of this
+   * proto, to avoid creating multiple large pools. It is typically better to
+   * run the non-low-priority work, even across sessions, in a single large
+   * pool.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.ThreadPoolOptionProto session_inter_op_thread_pool = 12;</code>
+   */
+  int getSessionInterOpThreadPoolCount();
+  /**
+   * <pre>
+   * This option is experimental - it may be replaced with a different mechanism
+   * in the future.
+   * Configures session thread pools. If this is configured, then RunOptions for
+   * a Run call can select the thread pool to use.
+   * The intended use is for when some session invocations need to run in a
+   * background pool limited to a small number of threads:
+   * - For example, a session may be configured to have one large pool (for
+   * regular compute) and one small pool (for periodic, low priority work);
+   * using the small pool is currently the mechanism for limiting the inter-op
+   * parallelism of the low priority work.  Note that it does not limit the
+   * parallelism of work spawned by a single op kernel implementation.
+   * - Using this setting is normally not needed in training, but may help some
+   * serving use cases.
+   * - It is also generally recommended to set the global_name field of this
+   * proto, to avoid creating multiple large pools. It is typically better to
+   * run the non-low-priority work, even across sessions, in a single large
+   * pool.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.ThreadPoolOptionProto session_inter_op_thread_pool = 12;</code>
+   */
+  java.util.List<? extends org.tensorflow.proto.framework.ThreadPoolOptionProtoOrBuilder> 
+      getSessionInterOpThreadPoolOrBuilderList();
+  /**
+   * <pre>
+   * This option is experimental - it may be replaced with a different mechanism
+   * in the future.
+   * Configures session thread pools. If this is configured, then RunOptions for
+   * a Run call can select the thread pool to use.
+   * The intended use is for when some session invocations need to run in a
+   * background pool limited to a small number of threads:
+   * - For example, a session may be configured to have one large pool (for
+   * regular compute) and one small pool (for periodic, low priority work);
+   * using the small pool is currently the mechanism for limiting the inter-op
+   * parallelism of the low priority work.  Note that it does not limit the
+   * parallelism of work spawned by a single op kernel implementation.
+   * - Using this setting is normally not needed in training, but may help some
+   * serving use cases.
+   * - It is also generally recommended to set the global_name field of this
+   * proto, to avoid creating multiple large pools. It is typically better to
+   * run the non-low-priority work, even across sessions, in a single large
+   * pool.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.ThreadPoolOptionProto session_inter_op_thread_pool = 12;</code>
+   */
+  org.tensorflow.proto.framework.ThreadPoolOptionProtoOrBuilder getSessionInterOpThreadPoolOrBuilder(
+      int index);
+
+  /**
+   * <pre>
+   * Assignment of Nodes to Devices is recomputed every placement_period
+   * steps until the system warms up (at which point the recomputation
+   * typically slows down automatically).
+   * </pre>
+   *
+   * <code>int32 placement_period = 3;</code>
+   */
+  int getPlacementPeriod();
+
+  /**
+   * <pre>
+   * When any filters are present sessions will ignore all devices which do not
+   * match the filters. Each filter can be partially specified, e.g. "/job:ps"
+   * "/job:worker/replica:3", etc.
+   * </pre>
+   *
+   * <code>repeated string device_filters = 4;</code>
+   */
+  java.util.List<java.lang.String>
+      getDeviceFiltersList();
+  /**
+   * <pre>
+   * When any filters are present sessions will ignore all devices which do not
+   * match the filters. Each filter can be partially specified, e.g. "/job:ps"
+   * "/job:worker/replica:3", etc.
+   * </pre>
+   *
+   * <code>repeated string device_filters = 4;</code>
+   */
+  int getDeviceFiltersCount();
+  /**
+   * <pre>
+   * When any filters are present sessions will ignore all devices which do not
+   * match the filters. Each filter can be partially specified, e.g. "/job:ps"
+   * "/job:worker/replica:3", etc.
+   * </pre>
+   *
+   * <code>repeated string device_filters = 4;</code>
+   */
+  java.lang.String getDeviceFilters(int index);
+  /**
+   * <pre>
+   * When any filters are present sessions will ignore all devices which do not
+   * match the filters. Each filter can be partially specified, e.g. "/job:ps"
+   * "/job:worker/replica:3", etc.
+   * </pre>
+   *
+   * <code>repeated string device_filters = 4;</code>
+   */
+  com.google.protobuf.ByteString
+      getDeviceFiltersBytes(int index);
+
+  /**
+   * <pre>
+   * Options that apply to all GPUs.
+   * </pre>
+   *
+   * <code>.tensorflow.GPUOptions gpu_options = 6;</code>
+   */
+  boolean hasGpuOptions();
+  /**
+   * <pre>
+   * Options that apply to all GPUs.
+   * </pre>
+   *
+   * <code>.tensorflow.GPUOptions gpu_options = 6;</code>
+   */
+  org.tensorflow.proto.framework.GPUOptions getGpuOptions();
+  /**
+   * <pre>
+   * Options that apply to all GPUs.
+   * </pre>
+   *
+   * <code>.tensorflow.GPUOptions gpu_options = 6;</code>
+   */
+  org.tensorflow.proto.framework.GPUOptionsOrBuilder getGpuOptionsOrBuilder();
+
+  /**
+   * <pre>
+   * Whether soft placement is allowed. If allow_soft_placement is true,
+   * an op will be placed on CPU if
+   *   1. there's no GPU implementation for the OP
+   * or
+   *   2. no GPU devices are known or registered
+   * or
+   *   3. need to co-locate with reftype input(s) which are from CPU.
+   * </pre>
+   *
+   * <code>bool allow_soft_placement = 7;</code>
+   */
+  boolean getAllowSoftPlacement();
+
+  /**
+   * <pre>
+   * Whether device placements should be logged.
+   * </pre>
+   *
+   * <code>bool log_device_placement = 8;</code>
+   */
+  boolean getLogDevicePlacement();
+
+  /**
+   * <pre>
+   * Options that apply to all graphs.
+   * </pre>
+   *
+   * <code>.tensorflow.GraphOptions graph_options = 10;</code>
+   */
+  boolean hasGraphOptions();
+  /**
+   * <pre>
+   * Options that apply to all graphs.
+   * </pre>
+   *
+   * <code>.tensorflow.GraphOptions graph_options = 10;</code>
+   */
+  org.tensorflow.proto.framework.GraphOptions getGraphOptions();
+  /**
+   * <pre>
+   * Options that apply to all graphs.
+   * </pre>
+   *
+   * <code>.tensorflow.GraphOptions graph_options = 10;</code>
+   */
+  org.tensorflow.proto.framework.GraphOptionsOrBuilder getGraphOptionsOrBuilder();
+
+  /**
+   * <pre>
+   * Global timeout for all blocking operations in this session.  If non-zero,
+   * and not overridden on a per-operation basis, this value will be used as the
+   * deadline for all blocking operations.
+   * </pre>
+   *
+   * <code>int64 operation_timeout_in_ms = 11;</code>
+   */
+  long getOperationTimeoutInMs();
+
+  /**
+   * <pre>
+   * Options that apply when this session uses the distributed runtime.
+   * </pre>
+   *
+   * <code>.tensorflow.RPCOptions rpc_options = 13;</code>
+   */
+  boolean hasRpcOptions();
+  /**
+   * <pre>
+   * Options that apply when this session uses the distributed runtime.
+   * </pre>
+   *
+   * <code>.tensorflow.RPCOptions rpc_options = 13;</code>
+   */
+  org.tensorflow.proto.framework.RPCOptions getRpcOptions();
+  /**
+   * <pre>
+   * Options that apply when this session uses the distributed runtime.
+   * </pre>
+   *
+   * <code>.tensorflow.RPCOptions rpc_options = 13;</code>
+   */
+  org.tensorflow.proto.framework.RPCOptionsOrBuilder getRpcOptionsOrBuilder();
+
+  /**
+   * <pre>
+   * Optional list of all workers to use in this session.
+   * </pre>
+   *
+   * <code>.tensorflow.ClusterDef cluster_def = 14;</code>
+   */
+  boolean hasClusterDef();
+  /**
+   * <pre>
+   * Optional list of all workers to use in this session.
+   * </pre>
+   *
+   * <code>.tensorflow.ClusterDef cluster_def = 14;</code>
+   */
+  org.tensorflow.proto.distruntime.ClusterDef getClusterDef();
+  /**
+   * <pre>
+   * Optional list of all workers to use in this session.
+   * </pre>
+   *
+   * <code>.tensorflow.ClusterDef cluster_def = 14;</code>
+   */
+  org.tensorflow.proto.distruntime.ClusterDefOrBuilder getClusterDefOrBuilder();
+
+  /**
+   * <pre>
+   * If true, any resources such as Variables used in the session will not be
+   * shared with other sessions. However, when clusterspec propagation is
+   * enabled, this field is ignored and sessions are always isolated.
+   * </pre>
+   *
+   * <code>bool isolate_session_state = 15;</code>
+   */
+  boolean getIsolateSessionState();
+
+  /**
+   * <pre>
+   * When true, WorkerSessions are created with device attributes from the
+   * full cluster.
+   * This is helpful when a worker wants to partition a graph
+   * (for example during a PartitionedCallOp).
+   * </pre>
+   *
+   * <code>bool share_cluster_devices_in_session = 17;</code>
+   */
+  boolean getShareClusterDevicesInSession();
+
+  /**
+   * <code>.tensorflow.ConfigProto.Experimental experimental = 16;</code>
+   */
+  boolean hasExperimental();
+  /**
+   * <code>.tensorflow.ConfigProto.Experimental experimental = 16;</code>
+   */
+  org.tensorflow.proto.framework.ConfigProto.Experimental getExperimental();
+  /**
+   * <code>.tensorflow.ConfigProto.Experimental experimental = 16;</code>
+   */
+  org.tensorflow.proto.framework.ConfigProto.ExperimentalOrBuilder getExperimentalOrBuilder();
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ConfigProtos.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ConfigProtos.java
new file mode 100644
index 00000000000..9a2c79552d6
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ConfigProtos.java
@@ -0,0 +1,393 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/config.proto
+
+package org.tensorflow.proto.framework;
+
+public final class ConfigProtos {
+  private ConfigProtos() {}
+  public static void registerAllExtensions(
+      com.google.protobuf.ExtensionRegistryLite registry) {
+  }
+
+  public static void registerAllExtensions(
+      com.google.protobuf.ExtensionRegistry registry) {
+    registerAllExtensions(
+        (com.google.protobuf.ExtensionRegistryLite) registry);
+  }
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_GPUOptions_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_GPUOptions_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_GPUOptions_Experimental_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_GPUOptions_Experimental_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_GPUOptions_Experimental_VirtualDevices_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_GPUOptions_Experimental_VirtualDevices_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_OptimizerOptions_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_OptimizerOptions_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_GraphOptions_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_GraphOptions_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_ThreadPoolOptionProto_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_ThreadPoolOptionProto_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_RPCOptions_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_RPCOptions_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_SessionMetadata_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_SessionMetadata_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_ConfigProto_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_ConfigProto_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_ConfigProto_DeviceCountEntry_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_ConfigProto_DeviceCountEntry_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_ConfigProto_Experimental_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_ConfigProto_Experimental_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_RunOptions_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_RunOptions_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_RunOptions_Experimental_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_RunOptions_Experimental_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_RunOptions_Experimental_RunHandlerPoolOptions_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_RunOptions_Experimental_RunHandlerPoolOptions_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_RunMetadata_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_RunMetadata_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_RunMetadata_FunctionGraphs_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_RunMetadata_FunctionGraphs_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_TensorConnection_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_TensorConnection_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_CallableOptions_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_CallableOptions_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_CallableOptions_FeedDevicesEntry_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_CallableOptions_FeedDevicesEntry_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_CallableOptions_FetchDevicesEntry_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_CallableOptions_FetchDevicesEntry_fieldAccessorTable;
+
+  public static com.google.protobuf.Descriptors.FileDescriptor
+      getDescriptor() {
+    return descriptor;
+  }
+  private static  com.google.protobuf.Descriptors.FileDescriptor
+      descriptor;
+  static {
+    java.lang.String[] descriptorData = {
+      "\n%tensorflow/core/protobuf/config.proto\022" +
+      "\ntensorflow\032*tensorflow/core/framework/c" +
+      "ost_graph.proto\032%tensorflow/core/framewo" +
+      "rk/graph.proto\032*tensorflow/core/framewor" +
+      "k/step_stats.proto\032&tensorflow/core/prot" +
+      "obuf/cluster.proto\032$tensorflow/core/prot" +
+      "obuf/debug.proto\032.tensorflow/core/protob" +
+      "uf/rewriter_config.proto\"\311\005\n\nGPUOptions\022" +
+      "\'\n\037per_process_gpu_memory_fraction\030\001 \001(\001" +
+      "\022\024\n\014allow_growth\030\004 \001(\010\022\026\n\016allocator_type" +
+      "\030\002 \001(\t\022\037\n\027deferred_deletion_bytes\030\003 \001(\003\022" +
+      "\033\n\023visible_device_list\030\005 \001(\t\022\"\n\032polling_" +
+      "active_delay_usecs\030\006 \001(\005\022$\n\034polling_inac" +
+      "tive_delay_msecs\030\007 \001(\005\022\034\n\024force_gpu_comp" +
+      "atible\030\010 \001(\010\0229\n\014experimental\030\t \001(\0132#.ten" +
+      "sorflow.GPUOptions.Experimental\032\202\003\n\014Expe" +
+      "rimental\022K\n\017virtual_devices\030\001 \003(\01322.tens" +
+      "orflow.GPUOptions.Experimental.VirtualDe" +
+      "vices\022\032\n\022use_unified_memory\030\002 \001(\010\022#\n\033num" +
+      "_dev_to_dev_copy_streams\030\003 \001(\005\022\035\n\025collec" +
+      "tive_ring_order\030\004 \001(\t\022\035\n\025timestamped_all" +
+      "ocator\030\005 \001(\010\022#\n\033kernel_tracker_max_inter" +
+      "val\030\007 \001(\005\022 \n\030kernel_tracker_max_bytes\030\010 " +
+      "\001(\005\022\"\n\032kernel_tracker_max_pending\030\t \001(\005\032" +
+      ";\n\016VirtualDevices\022\027\n\017memory_limit_mb\030\001 \003" +
+      "(\002\022\020\n\010priority\030\002 \003(\005\"\205\003\n\020OptimizerOption" +
+      "s\022+\n#do_common_subexpression_elimination" +
+      "\030\001 \001(\010\022\033\n\023do_constant_folding\030\002 \001(\010\022$\n\034m" +
+      "ax_folded_constant_in_bytes\030\006 \001(\003\022\034\n\024do_" +
+      "function_inlining\030\004 \001(\010\0225\n\topt_level\030\003 \001" +
+      "(\0162\".tensorflow.OptimizerOptions.Level\022E" +
+      "\n\020global_jit_level\030\005 \001(\0162+.tensorflow.Op" +
+      "timizerOptions.GlobalJitLevel\" \n\005Level\022\006" +
+      "\n\002L1\020\000\022\017\n\002L0\020\377\377\377\377\377\377\377\377\377\001\"C\n\016GlobalJitLeve" +
+      "l\022\013\n\007DEFAULT\020\000\022\020\n\003OFF\020\377\377\377\377\377\377\377\377\377\001\022\010\n\004ON_1" +
+      "\020\001\022\010\n\004ON_2\020\002\"\356\002\n\014GraphOptions\022\036\n\026enable_" +
+      "recv_scheduling\030\002 \001(\010\0227\n\021optimizer_optio" +
+      "ns\030\003 \001(\0132\034.tensorflow.OptimizerOptions\022\030" +
+      "\n\020build_cost_model\030\004 \001(\003\022\036\n\026build_cost_m" +
+      "odel_after\030\t \001(\003\022\024\n\014infer_shapes\030\005 \001(\010\022\032" +
+      "\n\022place_pruned_graph\030\006 \001(\010\022 \n\030enable_bfl" +
+      "oat16_sendrecv\030\007 \001(\010\022\025\n\rtimeline_step\030\010 " +
+      "\001(\005\0223\n\017rewrite_options\030\n \001(\0132\032.tensorflo" +
+      "w.RewriterConfigJ\004\010\001\020\002R%skip_common_sube" +
+      "xpression_elimination\"A\n\025ThreadPoolOptio" +
+      "nProto\022\023\n\013num_threads\030\001 \001(\005\022\023\n\013global_na" +
+      "me\030\002 \001(\t\"\264\001\n\nRPCOptions\022$\n\034use_rpc_for_i" +
+      "nprocess_master\030\001 \001(\010\022\035\n\025compression_alg" +
+      "orithm\030\002 \001(\t\022\031\n\021compression_level\030\003 \001(\005\022" +
+      "\032\n\022cache_rpc_response\030\004 \001(\010\022*\n\"disable_s" +
+      "ession_connection_sharing\030\005 \001(\010\"0\n\017Sessi" +
+      "onMetadata\022\014\n\004name\030\001 \001(\t\022\017\n\007version\030\002 \001(" +
+      "\003\"\310\n\n\013ConfigProto\022>\n\014device_count\030\001 \003(\0132" +
+      "(.tensorflow.ConfigProto.DeviceCountEntr" +
+      "y\022$\n\034intra_op_parallelism_threads\030\002 \001(\005\022" +
+      "$\n\034inter_op_parallelism_threads\030\005 \001(\005\022\037\n" +
+      "\027use_per_session_threads\030\t \001(\010\022G\n\034sessio" +
+      "n_inter_op_thread_pool\030\014 \003(\0132!.tensorflo" +
+      "w.ThreadPoolOptionProto\022\030\n\020placement_per" +
+      "iod\030\003 \001(\005\022\026\n\016device_filters\030\004 \003(\t\022+\n\013gpu" +
+      "_options\030\006 \001(\0132\026.tensorflow.GPUOptions\022\034" +
+      "\n\024allow_soft_placement\030\007 \001(\010\022\034\n\024log_devi" +
+      "ce_placement\030\010 \001(\010\022/\n\rgraph_options\030\n \001(" +
+      "\0132\030.tensorflow.GraphOptions\022\037\n\027operation" +
+      "_timeout_in_ms\030\013 \001(\003\022+\n\013rpc_options\030\r \001(" +
+      "\0132\026.tensorflow.RPCOptions\022+\n\013cluster_def" +
+      "\030\016 \001(\0132\026.tensorflow.ClusterDef\022\035\n\025isolat" +
+      "e_session_state\030\017 \001(\010\022(\n share_cluster_d" +
+      "evices_in_session\030\021 \001(\010\022:\n\014experimental\030" +
+      "\020 \001(\0132$.tensorflow.ConfigProto.Experimen" +
+      "tal\0322\n\020DeviceCountEntry\022\013\n\003key\030\001 \001(\t\022\r\n\005" +
+      "value\030\002 \001(\005:\0028\001\032\302\004\n\014Experimental\022\037\n\027coll" +
+      "ective_group_leader\030\001 \001(\t\022\025\n\rexecutor_ty" +
+      "pe\030\003 \001(\t\022\032\n\022recv_buf_max_chunk\030\004 \001(\005\022\031\n\021" +
+      "use_numa_affinity\030\005 \001(\010\0225\n-collective_de" +
+      "terministic_sequential_execution\030\006 \001(\010\022\027" +
+      "\n\017collective_nccl\030\007 \001(\010\0226\n.share_session" +
+      "_state_in_clusterspec_propagation\030\010 \001(\010\022" +
+      "\037\n\027disable_thread_spinning\030\t \001(\010\022(\n shar" +
+      "e_cluster_devices_in_session\030\n \001(\010\0225\n\020se" +
+      "ssion_metadata\030\013 \001(\0132\033.tensorflow.Sessio" +
+      "nMetadata\022!\n\031optimize_for_static_graph\030\014" +
+      " \001(\010\022\032\n\022enable_mlir_bridge\030\r \001(\010\022&\n\036enab" +
+      "le_mlir_graph_optimization\030\020 \001(\010\022\'\n\037disa" +
+      "ble_output_partition_graphs\030\016 \001(\010\022#\n\033xla" +
+      "_fusion_autotuner_thresh\030\017 \001(\003J\004\010\002\020\003\"\341\004\n" +
+      "\nRunOptions\0226\n\013trace_level\030\001 \001(\0162!.tenso" +
+      "rflow.RunOptions.TraceLevel\022\025\n\rtimeout_i" +
+      "n_ms\030\002 \001(\003\022\034\n\024inter_op_thread_pool\030\003 \001(\005" +
+      "\022\037\n\027output_partition_graphs\030\005 \001(\010\022/\n\rdeb" +
+      "ug_options\030\006 \001(\0132\030.tensorflow.DebugOptio" +
+      "ns\022*\n\"report_tensor_allocations_upon_oom" +
+      "\030\007 \001(\010\0229\n\014experimental\030\010 \001(\0132#.tensorflo" +
+      "w.RunOptions.Experimental\032\322\001\n\014Experiment" +
+      "al\022\034\n\024collective_graph_key\030\001 \001(\003\022\034\n\024use_" +
+      "run_handler_pool\030\002 \001(\010\022[\n\030run_handler_po" +
+      "ol_options\030\003 \001(\01329.tensorflow.RunOptions" +
+      ".Experimental.RunHandlerPoolOptions\032)\n\025R" +
+      "unHandlerPoolOptions\022\020\n\010priority\030\001 \001(\003\"R" +
+      "\n\nTraceLevel\022\014\n\010NO_TRACE\020\000\022\022\n\016SOFTWARE_T" +
+      "RACE\020\001\022\022\n\016HARDWARE_TRACE\020\002\022\016\n\nFULL_TRACE" +
+      "\020\003J\004\010\004\020\005\"\207\003\n\013RunMetadata\022)\n\nstep_stats\030\001" +
+      " \001(\0132\025.tensorflow.StepStats\022,\n\ncost_grap" +
+      "h\030\002 \001(\0132\030.tensorflow.CostGraphDef\022.\n\020par" +
+      "tition_graphs\030\003 \003(\0132\024.tensorflow.GraphDe" +
+      "f\022?\n\017function_graphs\030\004 \003(\0132&.tensorflow." +
+      "RunMetadata.FunctionGraphs\032\255\001\n\016FunctionG" +
+      "raphs\022.\n\020partition_graphs\030\001 \003(\0132\024.tensor" +
+      "flow.GraphDef\0224\n\026pre_optimization_graph\030" +
+      "\002 \001(\0132\024.tensorflow.GraphDef\0225\n\027post_opti" +
+      "mization_graph\030\003 \001(\0132\024.tensorflow.GraphD" +
+      "ef\":\n\020TensorConnection\022\023\n\013from_tensor\030\001 " +
+      "\001(\t\022\021\n\tto_tensor\030\002 \001(\t\"\260\003\n\017CallableOptio" +
+      "ns\022\014\n\004feed\030\001 \003(\t\022\r\n\005fetch\030\002 \003(\t\022\016\n\006targe" +
+      "t\030\003 \003(\t\022+\n\013run_options\030\004 \001(\0132\026.tensorflo" +
+      "w.RunOptions\0227\n\021tensor_connection\030\005 \003(\0132" +
+      "\034.tensorflow.TensorConnection\022B\n\014feed_de" +
+      "vices\030\006 \003(\0132,.tensorflow.CallableOptions" +
+      ".FeedDevicesEntry\022D\n\rfetch_devices\030\007 \003(\013" +
+      "2-.tensorflow.CallableOptions.FetchDevic" +
+      "esEntry\022\027\n\017fetch_skip_sync\030\010 \001(\010\0322\n\020Feed" +
+      "DevicesEntry\022\013\n\003key\030\001 \001(\t\022\r\n\005value\030\002 \001(\t" +
+      ":\0028\001\0323\n\021FetchDevicesEntry\022\013\n\003key\030\001 \001(\t\022\r" +
+      "\n\005value\030\002 \001(\t:\0028\001B}\n\036org.tensorflow.prot" +
+      "o.frameworkB\014ConfigProtosP\001ZHgithub.com/" +
+      "tensorflow/tensorflow/tensorflow/go/core" +
+      "/core_protos_go_proto\370\001\001b\006proto3"
+    };
+    descriptor = com.google.protobuf.Descriptors.FileDescriptor
+      .internalBuildGeneratedFileFrom(descriptorData,
+        new com.google.protobuf.Descriptors.FileDescriptor[] {
+          org.tensorflow.proto.framework.CostGraphProtos.getDescriptor(),
+          org.tensorflow.proto.framework.GraphProtos.getDescriptor(),
+          org.tensorflow.proto.framework.StepStatsProtos.getDescriptor(),
+          org.tensorflow.proto.distruntime.ClusterProtos.getDescriptor(),
+          org.tensorflow.proto.framework.DebugProtos.getDescriptor(),
+          org.tensorflow.proto.framework.RewriterConfigProtos.getDescriptor(),
+        });
+    internal_static_tensorflow_GPUOptions_descriptor =
+      getDescriptor().getMessageTypes().get(0);
+    internal_static_tensorflow_GPUOptions_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_GPUOptions_descriptor,
+        new java.lang.String[] { "PerProcessGpuMemoryFraction", "AllowGrowth", "AllocatorType", "DeferredDeletionBytes", "VisibleDeviceList", "PollingActiveDelayUsecs", "PollingInactiveDelayMsecs", "ForceGpuCompatible", "Experimental", });
+    internal_static_tensorflow_GPUOptions_Experimental_descriptor =
+      internal_static_tensorflow_GPUOptions_descriptor.getNestedTypes().get(0);
+    internal_static_tensorflow_GPUOptions_Experimental_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_GPUOptions_Experimental_descriptor,
+        new java.lang.String[] { "VirtualDevices", "UseUnifiedMemory", "NumDevToDevCopyStreams", "CollectiveRingOrder", "TimestampedAllocator", "KernelTrackerMaxInterval", "KernelTrackerMaxBytes", "KernelTrackerMaxPending", });
+    internal_static_tensorflow_GPUOptions_Experimental_VirtualDevices_descriptor =
+      internal_static_tensorflow_GPUOptions_Experimental_descriptor.getNestedTypes().get(0);
+    internal_static_tensorflow_GPUOptions_Experimental_VirtualDevices_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_GPUOptions_Experimental_VirtualDevices_descriptor,
+        new java.lang.String[] { "MemoryLimitMb", "Priority", });
+    internal_static_tensorflow_OptimizerOptions_descriptor =
+      getDescriptor().getMessageTypes().get(1);
+    internal_static_tensorflow_OptimizerOptions_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_OptimizerOptions_descriptor,
+        new java.lang.String[] { "DoCommonSubexpressionElimination", "DoConstantFolding", "MaxFoldedConstantInBytes", "DoFunctionInlining", "OptLevel", "GlobalJitLevel", });
+    internal_static_tensorflow_GraphOptions_descriptor =
+      getDescriptor().getMessageTypes().get(2);
+    internal_static_tensorflow_GraphOptions_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_GraphOptions_descriptor,
+        new java.lang.String[] { "EnableRecvScheduling", "OptimizerOptions", "BuildCostModel", "BuildCostModelAfter", "InferShapes", "PlacePrunedGraph", "EnableBfloat16Sendrecv", "TimelineStep", "RewriteOptions", });
+    internal_static_tensorflow_ThreadPoolOptionProto_descriptor =
+      getDescriptor().getMessageTypes().get(3);
+    internal_static_tensorflow_ThreadPoolOptionProto_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_ThreadPoolOptionProto_descriptor,
+        new java.lang.String[] { "NumThreads", "GlobalName", });
+    internal_static_tensorflow_RPCOptions_descriptor =
+      getDescriptor().getMessageTypes().get(4);
+    internal_static_tensorflow_RPCOptions_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_RPCOptions_descriptor,
+        new java.lang.String[] { "UseRpcForInprocessMaster", "CompressionAlgorithm", "CompressionLevel", "CacheRpcResponse", "DisableSessionConnectionSharing", });
+    internal_static_tensorflow_SessionMetadata_descriptor =
+      getDescriptor().getMessageTypes().get(5);
+    internal_static_tensorflow_SessionMetadata_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_SessionMetadata_descriptor,
+        new java.lang.String[] { "Name", "Version", });
+    internal_static_tensorflow_ConfigProto_descriptor =
+      getDescriptor().getMessageTypes().get(6);
+    internal_static_tensorflow_ConfigProto_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_ConfigProto_descriptor,
+        new java.lang.String[] { "DeviceCount", "IntraOpParallelismThreads", "InterOpParallelismThreads", "UsePerSessionThreads", "SessionInterOpThreadPool", "PlacementPeriod", "DeviceFilters", "GpuOptions", "AllowSoftPlacement", "LogDevicePlacement", "GraphOptions", "OperationTimeoutInMs", "RpcOptions", "ClusterDef", "IsolateSessionState", "ShareClusterDevicesInSession", "Experimental", });
+    internal_static_tensorflow_ConfigProto_DeviceCountEntry_descriptor =
+      internal_static_tensorflow_ConfigProto_descriptor.getNestedTypes().get(0);
+    internal_static_tensorflow_ConfigProto_DeviceCountEntry_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_ConfigProto_DeviceCountEntry_descriptor,
+        new java.lang.String[] { "Key", "Value", });
+    internal_static_tensorflow_ConfigProto_Experimental_descriptor =
+      internal_static_tensorflow_ConfigProto_descriptor.getNestedTypes().get(1);
+    internal_static_tensorflow_ConfigProto_Experimental_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_ConfigProto_Experimental_descriptor,
+        new java.lang.String[] { "CollectiveGroupLeader", "ExecutorType", "RecvBufMaxChunk", "UseNumaAffinity", "CollectiveDeterministicSequentialExecution", "CollectiveNccl", "ShareSessionStateInClusterspecPropagation", "DisableThreadSpinning", "ShareClusterDevicesInSession", "SessionMetadata", "OptimizeForStaticGraph", "EnableMlirBridge", "EnableMlirGraphOptimization", "DisableOutputPartitionGraphs", "XlaFusionAutotunerThresh", });
+    internal_static_tensorflow_RunOptions_descriptor =
+      getDescriptor().getMessageTypes().get(7);
+    internal_static_tensorflow_RunOptions_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_RunOptions_descriptor,
+        new java.lang.String[] { "TraceLevel", "TimeoutInMs", "InterOpThreadPool", "OutputPartitionGraphs", "DebugOptions", "ReportTensorAllocationsUponOom", "Experimental", });
+    internal_static_tensorflow_RunOptions_Experimental_descriptor =
+      internal_static_tensorflow_RunOptions_descriptor.getNestedTypes().get(0);
+    internal_static_tensorflow_RunOptions_Experimental_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_RunOptions_Experimental_descriptor,
+        new java.lang.String[] { "CollectiveGraphKey", "UseRunHandlerPool", "RunHandlerPoolOptions", });
+    internal_static_tensorflow_RunOptions_Experimental_RunHandlerPoolOptions_descriptor =
+      internal_static_tensorflow_RunOptions_Experimental_descriptor.getNestedTypes().get(0);
+    internal_static_tensorflow_RunOptions_Experimental_RunHandlerPoolOptions_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_RunOptions_Experimental_RunHandlerPoolOptions_descriptor,
+        new java.lang.String[] { "Priority", });
+    internal_static_tensorflow_RunMetadata_descriptor =
+      getDescriptor().getMessageTypes().get(8);
+    internal_static_tensorflow_RunMetadata_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_RunMetadata_descriptor,
+        new java.lang.String[] { "StepStats", "CostGraph", "PartitionGraphs", "FunctionGraphs", });
+    internal_static_tensorflow_RunMetadata_FunctionGraphs_descriptor =
+      internal_static_tensorflow_RunMetadata_descriptor.getNestedTypes().get(0);
+    internal_static_tensorflow_RunMetadata_FunctionGraphs_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_RunMetadata_FunctionGraphs_descriptor,
+        new java.lang.String[] { "PartitionGraphs", "PreOptimizationGraph", "PostOptimizationGraph", });
+    internal_static_tensorflow_TensorConnection_descriptor =
+      getDescriptor().getMessageTypes().get(9);
+    internal_static_tensorflow_TensorConnection_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_TensorConnection_descriptor,
+        new java.lang.String[] { "FromTensor", "ToTensor", });
+    internal_static_tensorflow_CallableOptions_descriptor =
+      getDescriptor().getMessageTypes().get(10);
+    internal_static_tensorflow_CallableOptions_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_CallableOptions_descriptor,
+        new java.lang.String[] { "Feed", "Fetch", "Target", "RunOptions", "TensorConnection", "FeedDevices", "FetchDevices", "FetchSkipSync", });
+    internal_static_tensorflow_CallableOptions_FeedDevicesEntry_descriptor =
+      internal_static_tensorflow_CallableOptions_descriptor.getNestedTypes().get(0);
+    internal_static_tensorflow_CallableOptions_FeedDevicesEntry_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_CallableOptions_FeedDevicesEntry_descriptor,
+        new java.lang.String[] { "Key", "Value", });
+    internal_static_tensorflow_CallableOptions_FetchDevicesEntry_descriptor =
+      internal_static_tensorflow_CallableOptions_descriptor.getNestedTypes().get(1);
+    internal_static_tensorflow_CallableOptions_FetchDevicesEntry_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_CallableOptions_FetchDevicesEntry_descriptor,
+        new java.lang.String[] { "Key", "Value", });
+    org.tensorflow.proto.framework.CostGraphProtos.getDescriptor();
+    org.tensorflow.proto.framework.GraphProtos.getDescriptor();
+    org.tensorflow.proto.framework.StepStatsProtos.getDescriptor();
+    org.tensorflow.proto.distruntime.ClusterProtos.getDescriptor();
+    org.tensorflow.proto.framework.DebugProtos.getDescriptor();
+    org.tensorflow.proto.framework.RewriterConfigProtos.getDescriptor();
+  }
+
+  // @@protoc_insertion_point(outer_class_scope)
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ControlFlowContextDef.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ControlFlowContextDef.java
new file mode 100644
index 00000000000..48b3b2cd638
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ControlFlowContextDef.java
@@ -0,0 +1,903 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/control_flow.proto
+
+package org.tensorflow.proto.framework;
+
+/**
+ * <pre>
+ * Container for any kind of control flow context. Any other control flow
+ * contexts that are added below should also be added here.
+ * </pre>
+ *
+ * Protobuf type {@code tensorflow.ControlFlowContextDef}
+ */
+public  final class ControlFlowContextDef extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.ControlFlowContextDef)
+    ControlFlowContextDefOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use ControlFlowContextDef.newBuilder() to construct.
+  private ControlFlowContextDef(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private ControlFlowContextDef() {
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new ControlFlowContextDef();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private ControlFlowContextDef(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 10: {
+            org.tensorflow.proto.framework.CondContextDef.Builder subBuilder = null;
+            if (ctxtCase_ == 1) {
+              subBuilder = ((org.tensorflow.proto.framework.CondContextDef) ctxt_).toBuilder();
+            }
+            ctxt_ =
+                input.readMessage(org.tensorflow.proto.framework.CondContextDef.parser(), extensionRegistry);
+            if (subBuilder != null) {
+              subBuilder.mergeFrom((org.tensorflow.proto.framework.CondContextDef) ctxt_);
+              ctxt_ = subBuilder.buildPartial();
+            }
+            ctxtCase_ = 1;
+            break;
+          }
+          case 18: {
+            org.tensorflow.proto.framework.WhileContextDef.Builder subBuilder = null;
+            if (ctxtCase_ == 2) {
+              subBuilder = ((org.tensorflow.proto.framework.WhileContextDef) ctxt_).toBuilder();
+            }
+            ctxt_ =
+                input.readMessage(org.tensorflow.proto.framework.WhileContextDef.parser(), extensionRegistry);
+            if (subBuilder != null) {
+              subBuilder.mergeFrom((org.tensorflow.proto.framework.WhileContextDef) ctxt_);
+              ctxt_ = subBuilder.buildPartial();
+            }
+            ctxtCase_ = 2;
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.framework.ControlFlowProtos.internal_static_tensorflow_ControlFlowContextDef_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.framework.ControlFlowProtos.internal_static_tensorflow_ControlFlowContextDef_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.framework.ControlFlowContextDef.class, org.tensorflow.proto.framework.ControlFlowContextDef.Builder.class);
+  }
+
+  private int ctxtCase_ = 0;
+  private java.lang.Object ctxt_;
+  public enum CtxtCase
+      implements com.google.protobuf.Internal.EnumLite {
+    COND_CTXT(1),
+    WHILE_CTXT(2),
+    CTXT_NOT_SET(0);
+    private final int value;
+    private CtxtCase(int value) {
+      this.value = value;
+    }
+    /**
+     * @deprecated Use {@link #forNumber(int)} instead.
+     */
+    @java.lang.Deprecated
+    public static CtxtCase valueOf(int value) {
+      return forNumber(value);
+    }
+
+    public static CtxtCase forNumber(int value) {
+      switch (value) {
+        case 1: return COND_CTXT;
+        case 2: return WHILE_CTXT;
+        case 0: return CTXT_NOT_SET;
+        default: return null;
+      }
+    }
+    public int getNumber() {
+      return this.value;
+    }
+  };
+
+  public CtxtCase
+  getCtxtCase() {
+    return CtxtCase.forNumber(
+        ctxtCase_);
+  }
+
+  public static final int COND_CTXT_FIELD_NUMBER = 1;
+  /**
+   * <code>.tensorflow.CondContextDef cond_ctxt = 1;</code>
+   */
+  public boolean hasCondCtxt() {
+    return ctxtCase_ == 1;
+  }
+  /**
+   * <code>.tensorflow.CondContextDef cond_ctxt = 1;</code>
+   */
+  public org.tensorflow.proto.framework.CondContextDef getCondCtxt() {
+    if (ctxtCase_ == 1) {
+       return (org.tensorflow.proto.framework.CondContextDef) ctxt_;
+    }
+    return org.tensorflow.proto.framework.CondContextDef.getDefaultInstance();
+  }
+  /**
+   * <code>.tensorflow.CondContextDef cond_ctxt = 1;</code>
+   */
+  public org.tensorflow.proto.framework.CondContextDefOrBuilder getCondCtxtOrBuilder() {
+    if (ctxtCase_ == 1) {
+       return (org.tensorflow.proto.framework.CondContextDef) ctxt_;
+    }
+    return org.tensorflow.proto.framework.CondContextDef.getDefaultInstance();
+  }
+
+  public static final int WHILE_CTXT_FIELD_NUMBER = 2;
+  /**
+   * <code>.tensorflow.WhileContextDef while_ctxt = 2;</code>
+   */
+  public boolean hasWhileCtxt() {
+    return ctxtCase_ == 2;
+  }
+  /**
+   * <code>.tensorflow.WhileContextDef while_ctxt = 2;</code>
+   */
+  public org.tensorflow.proto.framework.WhileContextDef getWhileCtxt() {
+    if (ctxtCase_ == 2) {
+       return (org.tensorflow.proto.framework.WhileContextDef) ctxt_;
+    }
+    return org.tensorflow.proto.framework.WhileContextDef.getDefaultInstance();
+  }
+  /**
+   * <code>.tensorflow.WhileContextDef while_ctxt = 2;</code>
+   */
+  public org.tensorflow.proto.framework.WhileContextDefOrBuilder getWhileCtxtOrBuilder() {
+    if (ctxtCase_ == 2) {
+       return (org.tensorflow.proto.framework.WhileContextDef) ctxt_;
+    }
+    return org.tensorflow.proto.framework.WhileContextDef.getDefaultInstance();
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    if (ctxtCase_ == 1) {
+      output.writeMessage(1, (org.tensorflow.proto.framework.CondContextDef) ctxt_);
+    }
+    if (ctxtCase_ == 2) {
+      output.writeMessage(2, (org.tensorflow.proto.framework.WhileContextDef) ctxt_);
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (ctxtCase_ == 1) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(1, (org.tensorflow.proto.framework.CondContextDef) ctxt_);
+    }
+    if (ctxtCase_ == 2) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(2, (org.tensorflow.proto.framework.WhileContextDef) ctxt_);
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.framework.ControlFlowContextDef)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.framework.ControlFlowContextDef other = (org.tensorflow.proto.framework.ControlFlowContextDef) obj;
+
+    if (!getCtxtCase().equals(other.getCtxtCase())) return false;
+    switch (ctxtCase_) {
+      case 1:
+        if (!getCondCtxt()
+            .equals(other.getCondCtxt())) return false;
+        break;
+      case 2:
+        if (!getWhileCtxt()
+            .equals(other.getWhileCtxt())) return false;
+        break;
+      case 0:
+      default:
+    }
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    switch (ctxtCase_) {
+      case 1:
+        hash = (37 * hash) + COND_CTXT_FIELD_NUMBER;
+        hash = (53 * hash) + getCondCtxt().hashCode();
+        break;
+      case 2:
+        hash = (37 * hash) + WHILE_CTXT_FIELD_NUMBER;
+        hash = (53 * hash) + getWhileCtxt().hashCode();
+        break;
+      case 0:
+      default:
+    }
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.framework.ControlFlowContextDef parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.ControlFlowContextDef parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.ControlFlowContextDef parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.ControlFlowContextDef parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.ControlFlowContextDef parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.ControlFlowContextDef parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.ControlFlowContextDef parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.ControlFlowContextDef parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.ControlFlowContextDef parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.ControlFlowContextDef parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.ControlFlowContextDef parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.ControlFlowContextDef parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.framework.ControlFlowContextDef prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * <pre>
+   * Container for any kind of control flow context. Any other control flow
+   * contexts that are added below should also be added here.
+   * </pre>
+   *
+   * Protobuf type {@code tensorflow.ControlFlowContextDef}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.ControlFlowContextDef)
+      org.tensorflow.proto.framework.ControlFlowContextDefOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.ControlFlowProtos.internal_static_tensorflow_ControlFlowContextDef_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.ControlFlowProtos.internal_static_tensorflow_ControlFlowContextDef_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.ControlFlowContextDef.class, org.tensorflow.proto.framework.ControlFlowContextDef.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.framework.ControlFlowContextDef.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      ctxtCase_ = 0;
+      ctxt_ = null;
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.framework.ControlFlowProtos.internal_static_tensorflow_ControlFlowContextDef_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.ControlFlowContextDef getDefaultInstanceForType() {
+      return org.tensorflow.proto.framework.ControlFlowContextDef.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.ControlFlowContextDef build() {
+      org.tensorflow.proto.framework.ControlFlowContextDef result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.ControlFlowContextDef buildPartial() {
+      org.tensorflow.proto.framework.ControlFlowContextDef result = new org.tensorflow.proto.framework.ControlFlowContextDef(this);
+      if (ctxtCase_ == 1) {
+        if (condCtxtBuilder_ == null) {
+          result.ctxt_ = ctxt_;
+        } else {
+          result.ctxt_ = condCtxtBuilder_.build();
+        }
+      }
+      if (ctxtCase_ == 2) {
+        if (whileCtxtBuilder_ == null) {
+          result.ctxt_ = ctxt_;
+        } else {
+          result.ctxt_ = whileCtxtBuilder_.build();
+        }
+      }
+      result.ctxtCase_ = ctxtCase_;
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.framework.ControlFlowContextDef) {
+        return mergeFrom((org.tensorflow.proto.framework.ControlFlowContextDef)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.framework.ControlFlowContextDef other) {
+      if (other == org.tensorflow.proto.framework.ControlFlowContextDef.getDefaultInstance()) return this;
+      switch (other.getCtxtCase()) {
+        case COND_CTXT: {
+          mergeCondCtxt(other.getCondCtxt());
+          break;
+        }
+        case WHILE_CTXT: {
+          mergeWhileCtxt(other.getWhileCtxt());
+          break;
+        }
+        case CTXT_NOT_SET: {
+          break;
+        }
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.framework.ControlFlowContextDef parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.framework.ControlFlowContextDef) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+    private int ctxtCase_ = 0;
+    private java.lang.Object ctxt_;
+    public CtxtCase
+        getCtxtCase() {
+      return CtxtCase.forNumber(
+          ctxtCase_);
+    }
+
+    public Builder clearCtxt() {
+      ctxtCase_ = 0;
+      ctxt_ = null;
+      onChanged();
+      return this;
+    }
+
+
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.CondContextDef, org.tensorflow.proto.framework.CondContextDef.Builder, org.tensorflow.proto.framework.CondContextDefOrBuilder> condCtxtBuilder_;
+    /**
+     * <code>.tensorflow.CondContextDef cond_ctxt = 1;</code>
+     */
+    public boolean hasCondCtxt() {
+      return ctxtCase_ == 1;
+    }
+    /**
+     * <code>.tensorflow.CondContextDef cond_ctxt = 1;</code>
+     */
+    public org.tensorflow.proto.framework.CondContextDef getCondCtxt() {
+      if (condCtxtBuilder_ == null) {
+        if (ctxtCase_ == 1) {
+          return (org.tensorflow.proto.framework.CondContextDef) ctxt_;
+        }
+        return org.tensorflow.proto.framework.CondContextDef.getDefaultInstance();
+      } else {
+        if (ctxtCase_ == 1) {
+          return condCtxtBuilder_.getMessage();
+        }
+        return org.tensorflow.proto.framework.CondContextDef.getDefaultInstance();
+      }
+    }
+    /**
+     * <code>.tensorflow.CondContextDef cond_ctxt = 1;</code>
+     */
+    public Builder setCondCtxt(org.tensorflow.proto.framework.CondContextDef value) {
+      if (condCtxtBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ctxt_ = value;
+        onChanged();
+      } else {
+        condCtxtBuilder_.setMessage(value);
+      }
+      ctxtCase_ = 1;
+      return this;
+    }
+    /**
+     * <code>.tensorflow.CondContextDef cond_ctxt = 1;</code>
+     */
+    public Builder setCondCtxt(
+        org.tensorflow.proto.framework.CondContextDef.Builder builderForValue) {
+      if (condCtxtBuilder_ == null) {
+        ctxt_ = builderForValue.build();
+        onChanged();
+      } else {
+        condCtxtBuilder_.setMessage(builderForValue.build());
+      }
+      ctxtCase_ = 1;
+      return this;
+    }
+    /**
+     * <code>.tensorflow.CondContextDef cond_ctxt = 1;</code>
+     */
+    public Builder mergeCondCtxt(org.tensorflow.proto.framework.CondContextDef value) {
+      if (condCtxtBuilder_ == null) {
+        if (ctxtCase_ == 1 &&
+            ctxt_ != org.tensorflow.proto.framework.CondContextDef.getDefaultInstance()) {
+          ctxt_ = org.tensorflow.proto.framework.CondContextDef.newBuilder((org.tensorflow.proto.framework.CondContextDef) ctxt_)
+              .mergeFrom(value).buildPartial();
+        } else {
+          ctxt_ = value;
+        }
+        onChanged();
+      } else {
+        if (ctxtCase_ == 1) {
+          condCtxtBuilder_.mergeFrom(value);
+        }
+        condCtxtBuilder_.setMessage(value);
+      }
+      ctxtCase_ = 1;
+      return this;
+    }
+    /**
+     * <code>.tensorflow.CondContextDef cond_ctxt = 1;</code>
+     */
+    public Builder clearCondCtxt() {
+      if (condCtxtBuilder_ == null) {
+        if (ctxtCase_ == 1) {
+          ctxtCase_ = 0;
+          ctxt_ = null;
+          onChanged();
+        }
+      } else {
+        if (ctxtCase_ == 1) {
+          ctxtCase_ = 0;
+          ctxt_ = null;
+        }
+        condCtxtBuilder_.clear();
+      }
+      return this;
+    }
+    /**
+     * <code>.tensorflow.CondContextDef cond_ctxt = 1;</code>
+     */
+    public org.tensorflow.proto.framework.CondContextDef.Builder getCondCtxtBuilder() {
+      return getCondCtxtFieldBuilder().getBuilder();
+    }
+    /**
+     * <code>.tensorflow.CondContextDef cond_ctxt = 1;</code>
+     */
+    public org.tensorflow.proto.framework.CondContextDefOrBuilder getCondCtxtOrBuilder() {
+      if ((ctxtCase_ == 1) && (condCtxtBuilder_ != null)) {
+        return condCtxtBuilder_.getMessageOrBuilder();
+      } else {
+        if (ctxtCase_ == 1) {
+          return (org.tensorflow.proto.framework.CondContextDef) ctxt_;
+        }
+        return org.tensorflow.proto.framework.CondContextDef.getDefaultInstance();
+      }
+    }
+    /**
+     * <code>.tensorflow.CondContextDef cond_ctxt = 1;</code>
+     */
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.CondContextDef, org.tensorflow.proto.framework.CondContextDef.Builder, org.tensorflow.proto.framework.CondContextDefOrBuilder> 
+        getCondCtxtFieldBuilder() {
+      if (condCtxtBuilder_ == null) {
+        if (!(ctxtCase_ == 1)) {
+          ctxt_ = org.tensorflow.proto.framework.CondContextDef.getDefaultInstance();
+        }
+        condCtxtBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+            org.tensorflow.proto.framework.CondContextDef, org.tensorflow.proto.framework.CondContextDef.Builder, org.tensorflow.proto.framework.CondContextDefOrBuilder>(
+                (org.tensorflow.proto.framework.CondContextDef) ctxt_,
+                getParentForChildren(),
+                isClean());
+        ctxt_ = null;
+      }
+      ctxtCase_ = 1;
+      onChanged();;
+      return condCtxtBuilder_;
+    }
+
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.WhileContextDef, org.tensorflow.proto.framework.WhileContextDef.Builder, org.tensorflow.proto.framework.WhileContextDefOrBuilder> whileCtxtBuilder_;
+    /**
+     * <code>.tensorflow.WhileContextDef while_ctxt = 2;</code>
+     */
+    public boolean hasWhileCtxt() {
+      return ctxtCase_ == 2;
+    }
+    /**
+     * <code>.tensorflow.WhileContextDef while_ctxt = 2;</code>
+     */
+    public org.tensorflow.proto.framework.WhileContextDef getWhileCtxt() {
+      if (whileCtxtBuilder_ == null) {
+        if (ctxtCase_ == 2) {
+          return (org.tensorflow.proto.framework.WhileContextDef) ctxt_;
+        }
+        return org.tensorflow.proto.framework.WhileContextDef.getDefaultInstance();
+      } else {
+        if (ctxtCase_ == 2) {
+          return whileCtxtBuilder_.getMessage();
+        }
+        return org.tensorflow.proto.framework.WhileContextDef.getDefaultInstance();
+      }
+    }
+    /**
+     * <code>.tensorflow.WhileContextDef while_ctxt = 2;</code>
+     */
+    public Builder setWhileCtxt(org.tensorflow.proto.framework.WhileContextDef value) {
+      if (whileCtxtBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ctxt_ = value;
+        onChanged();
+      } else {
+        whileCtxtBuilder_.setMessage(value);
+      }
+      ctxtCase_ = 2;
+      return this;
+    }
+    /**
+     * <code>.tensorflow.WhileContextDef while_ctxt = 2;</code>
+     */
+    public Builder setWhileCtxt(
+        org.tensorflow.proto.framework.WhileContextDef.Builder builderForValue) {
+      if (whileCtxtBuilder_ == null) {
+        ctxt_ = builderForValue.build();
+        onChanged();
+      } else {
+        whileCtxtBuilder_.setMessage(builderForValue.build());
+      }
+      ctxtCase_ = 2;
+      return this;
+    }
+    /**
+     * <code>.tensorflow.WhileContextDef while_ctxt = 2;</code>
+     */
+    public Builder mergeWhileCtxt(org.tensorflow.proto.framework.WhileContextDef value) {
+      if (whileCtxtBuilder_ == null) {
+        if (ctxtCase_ == 2 &&
+            ctxt_ != org.tensorflow.proto.framework.WhileContextDef.getDefaultInstance()) {
+          ctxt_ = org.tensorflow.proto.framework.WhileContextDef.newBuilder((org.tensorflow.proto.framework.WhileContextDef) ctxt_)
+              .mergeFrom(value).buildPartial();
+        } else {
+          ctxt_ = value;
+        }
+        onChanged();
+      } else {
+        if (ctxtCase_ == 2) {
+          whileCtxtBuilder_.mergeFrom(value);
+        }
+        whileCtxtBuilder_.setMessage(value);
+      }
+      ctxtCase_ = 2;
+      return this;
+    }
+    /**
+     * <code>.tensorflow.WhileContextDef while_ctxt = 2;</code>
+     */
+    public Builder clearWhileCtxt() {
+      if (whileCtxtBuilder_ == null) {
+        if (ctxtCase_ == 2) {
+          ctxtCase_ = 0;
+          ctxt_ = null;
+          onChanged();
+        }
+      } else {
+        if (ctxtCase_ == 2) {
+          ctxtCase_ = 0;
+          ctxt_ = null;
+        }
+        whileCtxtBuilder_.clear();
+      }
+      return this;
+    }
+    /**
+     * <code>.tensorflow.WhileContextDef while_ctxt = 2;</code>
+     */
+    public org.tensorflow.proto.framework.WhileContextDef.Builder getWhileCtxtBuilder() {
+      return getWhileCtxtFieldBuilder().getBuilder();
+    }
+    /**
+     * <code>.tensorflow.WhileContextDef while_ctxt = 2;</code>
+     */
+    public org.tensorflow.proto.framework.WhileContextDefOrBuilder getWhileCtxtOrBuilder() {
+      if ((ctxtCase_ == 2) && (whileCtxtBuilder_ != null)) {
+        return whileCtxtBuilder_.getMessageOrBuilder();
+      } else {
+        if (ctxtCase_ == 2) {
+          return (org.tensorflow.proto.framework.WhileContextDef) ctxt_;
+        }
+        return org.tensorflow.proto.framework.WhileContextDef.getDefaultInstance();
+      }
+    }
+    /**
+     * <code>.tensorflow.WhileContextDef while_ctxt = 2;</code>
+     */
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.WhileContextDef, org.tensorflow.proto.framework.WhileContextDef.Builder, org.tensorflow.proto.framework.WhileContextDefOrBuilder> 
+        getWhileCtxtFieldBuilder() {
+      if (whileCtxtBuilder_ == null) {
+        if (!(ctxtCase_ == 2)) {
+          ctxt_ = org.tensorflow.proto.framework.WhileContextDef.getDefaultInstance();
+        }
+        whileCtxtBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+            org.tensorflow.proto.framework.WhileContextDef, org.tensorflow.proto.framework.WhileContextDef.Builder, org.tensorflow.proto.framework.WhileContextDefOrBuilder>(
+                (org.tensorflow.proto.framework.WhileContextDef) ctxt_,
+                getParentForChildren(),
+                isClean());
+        ctxt_ = null;
+      }
+      ctxtCase_ = 2;
+      onChanged();;
+      return whileCtxtBuilder_;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.ControlFlowContextDef)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.ControlFlowContextDef)
+  private static final org.tensorflow.proto.framework.ControlFlowContextDef DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.ControlFlowContextDef();
+  }
+
+  public static org.tensorflow.proto.framework.ControlFlowContextDef getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<ControlFlowContextDef>
+      PARSER = new com.google.protobuf.AbstractParser<ControlFlowContextDef>() {
+    @java.lang.Override
+    public ControlFlowContextDef parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new ControlFlowContextDef(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<ControlFlowContextDef> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<ControlFlowContextDef> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.framework.ControlFlowContextDef getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ControlFlowContextDefOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ControlFlowContextDefOrBuilder.java
new file mode 100644
index 00000000000..b4482047fd9
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ControlFlowContextDefOrBuilder.java
@@ -0,0 +1,37 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/control_flow.proto
+
+package org.tensorflow.proto.framework;
+
+public interface ControlFlowContextDefOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.ControlFlowContextDef)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <code>.tensorflow.CondContextDef cond_ctxt = 1;</code>
+   */
+  boolean hasCondCtxt();
+  /**
+   * <code>.tensorflow.CondContextDef cond_ctxt = 1;</code>
+   */
+  org.tensorflow.proto.framework.CondContextDef getCondCtxt();
+  /**
+   * <code>.tensorflow.CondContextDef cond_ctxt = 1;</code>
+   */
+  org.tensorflow.proto.framework.CondContextDefOrBuilder getCondCtxtOrBuilder();
+
+  /**
+   * <code>.tensorflow.WhileContextDef while_ctxt = 2;</code>
+   */
+  boolean hasWhileCtxt();
+  /**
+   * <code>.tensorflow.WhileContextDef while_ctxt = 2;</code>
+   */
+  org.tensorflow.proto.framework.WhileContextDef getWhileCtxt();
+  /**
+   * <code>.tensorflow.WhileContextDef while_ctxt = 2;</code>
+   */
+  org.tensorflow.proto.framework.WhileContextDefOrBuilder getWhileCtxtOrBuilder();
+
+  public org.tensorflow.proto.framework.ControlFlowContextDef.CtxtCase getCtxtCase();
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ControlFlowProtos.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ControlFlowProtos.java
new file mode 100644
index 00000000000..97a565aa465
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ControlFlowProtos.java
@@ -0,0 +1,116 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/control_flow.proto
+
+package org.tensorflow.proto.framework;
+
+public final class ControlFlowProtos {
+  private ControlFlowProtos() {}
+  public static void registerAllExtensions(
+      com.google.protobuf.ExtensionRegistryLite registry) {
+  }
+
+  public static void registerAllExtensions(
+      com.google.protobuf.ExtensionRegistry registry) {
+    registerAllExtensions(
+        (com.google.protobuf.ExtensionRegistryLite) registry);
+  }
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_ValuesDef_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_ValuesDef_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_ValuesDef_ExternalValuesEntry_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_ValuesDef_ExternalValuesEntry_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_ControlFlowContextDef_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_ControlFlowContextDef_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_CondContextDef_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_CondContextDef_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_WhileContextDef_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_WhileContextDef_fieldAccessorTable;
+
+  public static com.google.protobuf.Descriptors.FileDescriptor
+      getDescriptor() {
+    return descriptor;
+  }
+  private static  com.google.protobuf.Descriptors.FileDescriptor
+      descriptor;
+  static {
+    java.lang.String[] descriptorData = {
+      "\n+tensorflow/core/protobuf/control_flow." +
+      "proto\022\ntensorflow\"\226\001\n\tValuesDef\022\016\n\006value" +
+      "s\030\001 \003(\t\022B\n\017external_values\030\002 \003(\0132).tenso" +
+      "rflow.ValuesDef.ExternalValuesEntry\0325\n\023E" +
+      "xternalValuesEntry\022\013\n\003key\030\001 \001(\t\022\r\n\005value" +
+      "\030\002 \001(\t:\0028\001\"\203\001\n\025ControlFlowContextDef\022/\n\t" +
+      "cond_ctxt\030\001 \001(\0132\032.tensorflow.CondContext" +
+      "DefH\000\0221\n\nwhile_ctxt\030\002 \001(\0132\033.tensorflow.W" +
+      "hileContextDefH\000B\006\n\004ctxt\"\304\001\n\016CondContext" +
+      "Def\022\024\n\014context_name\030\001 \001(\t\022\021\n\tpred_name\030\002" +
+      " \001(\t\022\022\n\npivot_name\030\003 \001(\t\022\016\n\006branch\030\004 \001(\005" +
+      "\022)\n\nvalues_def\030\005 \001(\0132\025.tensorflow.Values" +
+      "Def\022:\n\017nested_contexts\030\006 \003(\0132!.tensorflo" +
+      "w.ControlFlowContextDef\"\365\002\n\017WhileContext" +
+      "Def\022\024\n\014context_name\030\001 \001(\t\022\033\n\023parallel_it" +
+      "erations\030\002 \001(\005\022\021\n\tback_prop\030\003 \001(\010\022\023\n\013swa" +
+      "p_memory\030\004 \001(\010\022\022\n\npivot_name\030\005 \001(\t\022\033\n\023pi" +
+      "vot_for_pred_name\030\006 \001(\t\022\033\n\023pivot_for_bod" +
+      "y_name\030\007 \001(\t\022\027\n\017loop_exit_names\030\010 \003(\t\022\030\n" +
+      "\020loop_enter_names\030\n \003(\t\022)\n\nvalues_def\030\t " +
+      "\001(\0132\025.tensorflow.ValuesDef\022\037\n\027maximum_it" +
+      "erations_name\030\013 \001(\t\022:\n\017nested_contexts\030\014" +
+      " \003(\0132!.tensorflow.ControlFlowContextDefB" +
+      "\202\001\n\036org.tensorflow.proto.frameworkB\021Cont" +
+      "rolFlowProtosP\001ZHgithub.com/tensorflow/t" +
+      "ensorflow/tensorflow/go/core/core_protos" +
+      "_go_proto\370\001\001b\006proto3"
+    };
+    descriptor = com.google.protobuf.Descriptors.FileDescriptor
+      .internalBuildGeneratedFileFrom(descriptorData,
+        new com.google.protobuf.Descriptors.FileDescriptor[] {
+        });
+    internal_static_tensorflow_ValuesDef_descriptor =
+      getDescriptor().getMessageTypes().get(0);
+    internal_static_tensorflow_ValuesDef_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_ValuesDef_descriptor,
+        new java.lang.String[] { "Values", "ExternalValues", });
+    internal_static_tensorflow_ValuesDef_ExternalValuesEntry_descriptor =
+      internal_static_tensorflow_ValuesDef_descriptor.getNestedTypes().get(0);
+    internal_static_tensorflow_ValuesDef_ExternalValuesEntry_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_ValuesDef_ExternalValuesEntry_descriptor,
+        new java.lang.String[] { "Key", "Value", });
+    internal_static_tensorflow_ControlFlowContextDef_descriptor =
+      getDescriptor().getMessageTypes().get(1);
+    internal_static_tensorflow_ControlFlowContextDef_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_ControlFlowContextDef_descriptor,
+        new java.lang.String[] { "CondCtxt", "WhileCtxt", "Ctxt", });
+    internal_static_tensorflow_CondContextDef_descriptor =
+      getDescriptor().getMessageTypes().get(2);
+    internal_static_tensorflow_CondContextDef_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_CondContextDef_descriptor,
+        new java.lang.String[] { "ContextName", "PredName", "PivotName", "Branch", "ValuesDef", "NestedContexts", });
+    internal_static_tensorflow_WhileContextDef_descriptor =
+      getDescriptor().getMessageTypes().get(3);
+    internal_static_tensorflow_WhileContextDef_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_WhileContextDef_descriptor,
+        new java.lang.String[] { "ContextName", "ParallelIterations", "BackProp", "SwapMemory", "PivotName", "PivotForPredName", "PivotForBodyName", "LoopExitNames", "LoopEnterNames", "ValuesDef", "MaximumIterationsName", "NestedContexts", });
+  }
+
+  // @@protoc_insertion_point(outer_class_scope)
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/CostGraphDef.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/CostGraphDef.java
new file mode 100644
index 00000000000..c06cd1761fa
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/CostGraphDef.java
@@ -0,0 +1,5817 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/framework/cost_graph.proto
+
+package org.tensorflow.proto.framework;
+
+/**
+ * Protobuf type {@code tensorflow.CostGraphDef}
+ */
+public  final class CostGraphDef extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.CostGraphDef)
+    CostGraphDefOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use CostGraphDef.newBuilder() to construct.
+  private CostGraphDef(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private CostGraphDef() {
+    node_ = java.util.Collections.emptyList();
+    cost_ = java.util.Collections.emptyList();
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new CostGraphDef();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private CostGraphDef(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    int mutable_bitField0_ = 0;
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 10: {
+            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
+              node_ = new java.util.ArrayList<org.tensorflow.proto.framework.CostGraphDef.Node>();
+              mutable_bitField0_ |= 0x00000001;
+            }
+            node_.add(
+                input.readMessage(org.tensorflow.proto.framework.CostGraphDef.Node.parser(), extensionRegistry));
+            break;
+          }
+          case 18: {
+            if (!((mutable_bitField0_ & 0x00000002) != 0)) {
+              cost_ = new java.util.ArrayList<org.tensorflow.proto.framework.CostGraphDef.AggregatedCost>();
+              mutable_bitField0_ |= 0x00000002;
+            }
+            cost_.add(
+                input.readMessage(org.tensorflow.proto.framework.CostGraphDef.AggregatedCost.parser(), extensionRegistry));
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      if (((mutable_bitField0_ & 0x00000001) != 0)) {
+        node_ = java.util.Collections.unmodifiableList(node_);
+      }
+      if (((mutable_bitField0_ & 0x00000002) != 0)) {
+        cost_ = java.util.Collections.unmodifiableList(cost_);
+      }
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.framework.CostGraphProtos.internal_static_tensorflow_CostGraphDef_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.framework.CostGraphProtos.internal_static_tensorflow_CostGraphDef_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.framework.CostGraphDef.class, org.tensorflow.proto.framework.CostGraphDef.Builder.class);
+  }
+
+  public interface NodeOrBuilder extends
+      // @@protoc_insertion_point(interface_extends:tensorflow.CostGraphDef.Node)
+      com.google.protobuf.MessageOrBuilder {
+
+    /**
+     * <pre>
+     * The name of the node. Names are globally unique.
+     * </pre>
+     *
+     * <code>string name = 1;</code>
+     */
+    java.lang.String getName();
+    /**
+     * <pre>
+     * The name of the node. Names are globally unique.
+     * </pre>
+     *
+     * <code>string name = 1;</code>
+     */
+    com.google.protobuf.ByteString
+        getNameBytes();
+
+    /**
+     * <pre>
+     * The device of the node. Can be empty if the node is mapped to the
+     * default partition or partitioning hasn't been run yet.
+     * </pre>
+     *
+     * <code>string device = 2;</code>
+     */
+    java.lang.String getDevice();
+    /**
+     * <pre>
+     * The device of the node. Can be empty if the node is mapped to the
+     * default partition or partitioning hasn't been run yet.
+     * </pre>
+     *
+     * <code>string device = 2;</code>
+     */
+    com.google.protobuf.ByteString
+        getDeviceBytes();
+
+    /**
+     * <pre>
+     * The id of the node. Node ids are only unique inside a partition.
+     * </pre>
+     *
+     * <code>int32 id = 3;</code>
+     */
+    int getId();
+
+    /**
+     * <code>repeated .tensorflow.CostGraphDef.Node.InputInfo input_info = 4;</code>
+     */
+    java.util.List<org.tensorflow.proto.framework.CostGraphDef.Node.InputInfo> 
+        getInputInfoList();
+    /**
+     * <code>repeated .tensorflow.CostGraphDef.Node.InputInfo input_info = 4;</code>
+     */
+    org.tensorflow.proto.framework.CostGraphDef.Node.InputInfo getInputInfo(int index);
+    /**
+     * <code>repeated .tensorflow.CostGraphDef.Node.InputInfo input_info = 4;</code>
+     */
+    int getInputInfoCount();
+    /**
+     * <code>repeated .tensorflow.CostGraphDef.Node.InputInfo input_info = 4;</code>
+     */
+    java.util.List<? extends org.tensorflow.proto.framework.CostGraphDef.Node.InputInfoOrBuilder> 
+        getInputInfoOrBuilderList();
+    /**
+     * <code>repeated .tensorflow.CostGraphDef.Node.InputInfo input_info = 4;</code>
+     */
+    org.tensorflow.proto.framework.CostGraphDef.Node.InputInfoOrBuilder getInputInfoOrBuilder(
+        int index);
+
+    /**
+     * <code>repeated .tensorflow.CostGraphDef.Node.OutputInfo output_info = 5;</code>
+     */
+    java.util.List<org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfo> 
+        getOutputInfoList();
+    /**
+     * <code>repeated .tensorflow.CostGraphDef.Node.OutputInfo output_info = 5;</code>
+     */
+    org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfo getOutputInfo(int index);
+    /**
+     * <code>repeated .tensorflow.CostGraphDef.Node.OutputInfo output_info = 5;</code>
+     */
+    int getOutputInfoCount();
+    /**
+     * <code>repeated .tensorflow.CostGraphDef.Node.OutputInfo output_info = 5;</code>
+     */
+    java.util.List<? extends org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfoOrBuilder> 
+        getOutputInfoOrBuilderList();
+    /**
+     * <code>repeated .tensorflow.CostGraphDef.Node.OutputInfo output_info = 5;</code>
+     */
+    org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfoOrBuilder getOutputInfoOrBuilder(
+        int index);
+
+    /**
+     * <pre>
+     * Temporary memory used by this node.
+     * </pre>
+     *
+     * <code>int64 temporary_memory_size = 6;</code>
+     */
+    long getTemporaryMemorySize();
+
+    /**
+     * <pre>
+     * Persistent memory used by this node.
+     * </pre>
+     *
+     * <code>int64 persistent_memory_size = 12;</code>
+     */
+    long getPersistentMemorySize();
+
+    /**
+     * <code>int64 host_temp_memory_size = 10 [deprecated = true];</code>
+     */
+    @java.lang.Deprecated long getHostTempMemorySize();
+
+    /**
+     * <code>int64 device_temp_memory_size = 11 [deprecated = true];</code>
+     */
+    @java.lang.Deprecated long getDeviceTempMemorySize();
+
+    /**
+     * <code>int64 device_persistent_memory_size = 16 [deprecated = true];</code>
+     */
+    @java.lang.Deprecated long getDevicePersistentMemorySize();
+
+    /**
+     * <pre>
+     * Estimate of the computational cost of this node, in microseconds.
+     * </pre>
+     *
+     * <code>int64 compute_cost = 9;</code>
+     */
+    long getComputeCost();
+
+    /**
+     * <pre>
+     * Analytical estimate of the computational cost of this node, in
+     * microseconds.
+     * </pre>
+     *
+     * <code>int64 compute_time = 14;</code>
+     */
+    long getComputeTime();
+
+    /**
+     * <pre>
+     * Analytical estimate of the memory access cost of this node, in
+     * microseconds.
+     * </pre>
+     *
+     * <code>int64 memory_time = 15;</code>
+     */
+    long getMemoryTime();
+
+    /**
+     * <pre>
+     * If true, the output is permanent: it can't be discarded, because this
+     * node is part of the "final output". Nodes may depend on final nodes.
+     * </pre>
+     *
+     * <code>bool is_final = 7;</code>
+     */
+    boolean getIsFinal();
+
+    /**
+     * <pre>
+     * Ids of the control inputs for this node.
+     * </pre>
+     *
+     * <code>repeated int32 control_input = 8;</code>
+     */
+    java.util.List<java.lang.Integer> getControlInputList();
+    /**
+     * <pre>
+     * Ids of the control inputs for this node.
+     * </pre>
+     *
+     * <code>repeated int32 control_input = 8;</code>
+     */
+    int getControlInputCount();
+    /**
+     * <pre>
+     * Ids of the control inputs for this node.
+     * </pre>
+     *
+     * <code>repeated int32 control_input = 8;</code>
+     */
+    int getControlInput(int index);
+
+    /**
+     * <pre>
+     * Are the costs inaccurate?
+     * </pre>
+     *
+     * <code>bool inaccurate = 17;</code>
+     */
+    boolean getInaccurate();
+  }
+  /**
+   * Protobuf type {@code tensorflow.CostGraphDef.Node}
+   */
+  public  static final class Node extends
+      com.google.protobuf.GeneratedMessageV3 implements
+      // @@protoc_insertion_point(message_implements:tensorflow.CostGraphDef.Node)
+      NodeOrBuilder {
+  private static final long serialVersionUID = 0L;
+    // Use Node.newBuilder() to construct.
+    private Node(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+      super(builder);
+    }
+    private Node() {
+      name_ = "";
+      device_ = "";
+      inputInfo_ = java.util.Collections.emptyList();
+      outputInfo_ = java.util.Collections.emptyList();
+      controlInput_ = emptyIntList();
+    }
+
+    @java.lang.Override
+    @SuppressWarnings({"unused"})
+    protected java.lang.Object newInstance(
+        UnusedPrivateParameter unused) {
+      return new Node();
+    }
+
+    @java.lang.Override
+    public final com.google.protobuf.UnknownFieldSet
+    getUnknownFields() {
+      return this.unknownFields;
+    }
+    private Node(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      this();
+      if (extensionRegistry == null) {
+        throw new java.lang.NullPointerException();
+      }
+      int mutable_bitField0_ = 0;
+      com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+          com.google.protobuf.UnknownFieldSet.newBuilder();
+      try {
+        boolean done = false;
+        while (!done) {
+          int tag = input.readTag();
+          switch (tag) {
+            case 0:
+              done = true;
+              break;
+            case 10: {
+              java.lang.String s = input.readStringRequireUtf8();
+
+              name_ = s;
+              break;
+            }
+            case 18: {
+              java.lang.String s = input.readStringRequireUtf8();
+
+              device_ = s;
+              break;
+            }
+            case 24: {
+
+              id_ = input.readInt32();
+              break;
+            }
+            case 34: {
+              if (!((mutable_bitField0_ & 0x00000001) != 0)) {
+                inputInfo_ = new java.util.ArrayList<org.tensorflow.proto.framework.CostGraphDef.Node.InputInfo>();
+                mutable_bitField0_ |= 0x00000001;
+              }
+              inputInfo_.add(
+                  input.readMessage(org.tensorflow.proto.framework.CostGraphDef.Node.InputInfo.parser(), extensionRegistry));
+              break;
+            }
+            case 42: {
+              if (!((mutable_bitField0_ & 0x00000002) != 0)) {
+                outputInfo_ = new java.util.ArrayList<org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfo>();
+                mutable_bitField0_ |= 0x00000002;
+              }
+              outputInfo_.add(
+                  input.readMessage(org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfo.parser(), extensionRegistry));
+              break;
+            }
+            case 48: {
+
+              temporaryMemorySize_ = input.readInt64();
+              break;
+            }
+            case 56: {
+
+              isFinal_ = input.readBool();
+              break;
+            }
+            case 64: {
+              if (!((mutable_bitField0_ & 0x00000004) != 0)) {
+                controlInput_ = newIntList();
+                mutable_bitField0_ |= 0x00000004;
+              }
+              controlInput_.addInt(input.readInt32());
+              break;
+            }
+            case 66: {
+              int length = input.readRawVarint32();
+              int limit = input.pushLimit(length);
+              if (!((mutable_bitField0_ & 0x00000004) != 0) && input.getBytesUntilLimit() > 0) {
+                controlInput_ = newIntList();
+                mutable_bitField0_ |= 0x00000004;
+              }
+              while (input.getBytesUntilLimit() > 0) {
+                controlInput_.addInt(input.readInt32());
+              }
+              input.popLimit(limit);
+              break;
+            }
+            case 72: {
+
+              computeCost_ = input.readInt64();
+              break;
+            }
+            case 80: {
+
+              hostTempMemorySize_ = input.readInt64();
+              break;
+            }
+            case 88: {
+
+              deviceTempMemorySize_ = input.readInt64();
+              break;
+            }
+            case 96: {
+
+              persistentMemorySize_ = input.readInt64();
+              break;
+            }
+            case 112: {
+
+              computeTime_ = input.readInt64();
+              break;
+            }
+            case 120: {
+
+              memoryTime_ = input.readInt64();
+              break;
+            }
+            case 128: {
+
+              devicePersistentMemorySize_ = input.readInt64();
+              break;
+            }
+            case 136: {
+
+              inaccurate_ = input.readBool();
+              break;
+            }
+            default: {
+              if (!parseUnknownField(
+                  input, unknownFields, extensionRegistry, tag)) {
+                done = true;
+              }
+              break;
+            }
+          }
+        }
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        throw e.setUnfinishedMessage(this);
+      } catch (java.io.IOException e) {
+        throw new com.google.protobuf.InvalidProtocolBufferException(
+            e).setUnfinishedMessage(this);
+      } finally {
+        if (((mutable_bitField0_ & 0x00000001) != 0)) {
+          inputInfo_ = java.util.Collections.unmodifiableList(inputInfo_);
+        }
+        if (((mutable_bitField0_ & 0x00000002) != 0)) {
+          outputInfo_ = java.util.Collections.unmodifiableList(outputInfo_);
+        }
+        if (((mutable_bitField0_ & 0x00000004) != 0)) {
+          controlInput_.makeImmutable(); // C
+        }
+        this.unknownFields = unknownFields.build();
+        makeExtensionsImmutable();
+      }
+    }
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.CostGraphProtos.internal_static_tensorflow_CostGraphDef_Node_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.CostGraphProtos.internal_static_tensorflow_CostGraphDef_Node_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.CostGraphDef.Node.class, org.tensorflow.proto.framework.CostGraphDef.Node.Builder.class);
+    }
+
+    public interface InputInfoOrBuilder extends
+        // @@protoc_insertion_point(interface_extends:tensorflow.CostGraphDef.Node.InputInfo)
+        com.google.protobuf.MessageOrBuilder {
+
+      /**
+       * <code>int32 preceding_node = 1;</code>
+       */
+      int getPrecedingNode();
+
+      /**
+       * <code>int32 preceding_port = 2;</code>
+       */
+      int getPrecedingPort();
+    }
+    /**
+     * <pre>
+     * Inputs of this node. They must be executed before this node can be
+     * executed. An input is a particular output of another node, specified
+     * by the node id and the output index.
+     * </pre>
+     *
+     * Protobuf type {@code tensorflow.CostGraphDef.Node.InputInfo}
+     */
+    public  static final class InputInfo extends
+        com.google.protobuf.GeneratedMessageV3 implements
+        // @@protoc_insertion_point(message_implements:tensorflow.CostGraphDef.Node.InputInfo)
+        InputInfoOrBuilder {
+    private static final long serialVersionUID = 0L;
+      // Use InputInfo.newBuilder() to construct.
+      private InputInfo(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+        super(builder);
+      }
+      private InputInfo() {
+      }
+
+      @java.lang.Override
+      @SuppressWarnings({"unused"})
+      protected java.lang.Object newInstance(
+          UnusedPrivateParameter unused) {
+        return new InputInfo();
+      }
+
+      @java.lang.Override
+      public final com.google.protobuf.UnknownFieldSet
+      getUnknownFields() {
+        return this.unknownFields;
+      }
+      private InputInfo(
+          com.google.protobuf.CodedInputStream input,
+          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+          throws com.google.protobuf.InvalidProtocolBufferException {
+        this();
+        if (extensionRegistry == null) {
+          throw new java.lang.NullPointerException();
+        }
+        com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+            com.google.protobuf.UnknownFieldSet.newBuilder();
+        try {
+          boolean done = false;
+          while (!done) {
+            int tag = input.readTag();
+            switch (tag) {
+              case 0:
+                done = true;
+                break;
+              case 8: {
+
+                precedingNode_ = input.readInt32();
+                break;
+              }
+              case 16: {
+
+                precedingPort_ = input.readInt32();
+                break;
+              }
+              default: {
+                if (!parseUnknownField(
+                    input, unknownFields, extensionRegistry, tag)) {
+                  done = true;
+                }
+                break;
+              }
+            }
+          }
+        } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+          throw e.setUnfinishedMessage(this);
+        } catch (java.io.IOException e) {
+          throw new com.google.protobuf.InvalidProtocolBufferException(
+              e).setUnfinishedMessage(this);
+        } finally {
+          this.unknownFields = unknownFields.build();
+          makeExtensionsImmutable();
+        }
+      }
+      public static final com.google.protobuf.Descriptors.Descriptor
+          getDescriptor() {
+        return org.tensorflow.proto.framework.CostGraphProtos.internal_static_tensorflow_CostGraphDef_Node_InputInfo_descriptor;
+      }
+
+      @java.lang.Override
+      protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+          internalGetFieldAccessorTable() {
+        return org.tensorflow.proto.framework.CostGraphProtos.internal_static_tensorflow_CostGraphDef_Node_InputInfo_fieldAccessorTable
+            .ensureFieldAccessorsInitialized(
+                org.tensorflow.proto.framework.CostGraphDef.Node.InputInfo.class, org.tensorflow.proto.framework.CostGraphDef.Node.InputInfo.Builder.class);
+      }
+
+      public static final int PRECEDING_NODE_FIELD_NUMBER = 1;
+      private int precedingNode_;
+      /**
+       * <code>int32 preceding_node = 1;</code>
+       */
+      public int getPrecedingNode() {
+        return precedingNode_;
+      }
+
+      public static final int PRECEDING_PORT_FIELD_NUMBER = 2;
+      private int precedingPort_;
+      /**
+       * <code>int32 preceding_port = 2;</code>
+       */
+      public int getPrecedingPort() {
+        return precedingPort_;
+      }
+
+      private byte memoizedIsInitialized = -1;
+      @java.lang.Override
+      public final boolean isInitialized() {
+        byte isInitialized = memoizedIsInitialized;
+        if (isInitialized == 1) return true;
+        if (isInitialized == 0) return false;
+
+        memoizedIsInitialized = 1;
+        return true;
+      }
+
+      @java.lang.Override
+      public void writeTo(com.google.protobuf.CodedOutputStream output)
+                          throws java.io.IOException {
+        if (precedingNode_ != 0) {
+          output.writeInt32(1, precedingNode_);
+        }
+        if (precedingPort_ != 0) {
+          output.writeInt32(2, precedingPort_);
+        }
+        unknownFields.writeTo(output);
+      }
+
+      @java.lang.Override
+      public int getSerializedSize() {
+        int size = memoizedSize;
+        if (size != -1) return size;
+
+        size = 0;
+        if (precedingNode_ != 0) {
+          size += com.google.protobuf.CodedOutputStream
+            .computeInt32Size(1, precedingNode_);
+        }
+        if (precedingPort_ != 0) {
+          size += com.google.protobuf.CodedOutputStream
+            .computeInt32Size(2, precedingPort_);
+        }
+        size += unknownFields.getSerializedSize();
+        memoizedSize = size;
+        return size;
+      }
+
+      @java.lang.Override
+      public boolean equals(final java.lang.Object obj) {
+        if (obj == this) {
+         return true;
+        }
+        if (!(obj instanceof org.tensorflow.proto.framework.CostGraphDef.Node.InputInfo)) {
+          return super.equals(obj);
+        }
+        org.tensorflow.proto.framework.CostGraphDef.Node.InputInfo other = (org.tensorflow.proto.framework.CostGraphDef.Node.InputInfo) obj;
+
+        if (getPrecedingNode()
+            != other.getPrecedingNode()) return false;
+        if (getPrecedingPort()
+            != other.getPrecedingPort()) return false;
+        if (!unknownFields.equals(other.unknownFields)) return false;
+        return true;
+      }
+
+      @java.lang.Override
+      public int hashCode() {
+        if (memoizedHashCode != 0) {
+          return memoizedHashCode;
+        }
+        int hash = 41;
+        hash = (19 * hash) + getDescriptor().hashCode();
+        hash = (37 * hash) + PRECEDING_NODE_FIELD_NUMBER;
+        hash = (53 * hash) + getPrecedingNode();
+        hash = (37 * hash) + PRECEDING_PORT_FIELD_NUMBER;
+        hash = (53 * hash) + getPrecedingPort();
+        hash = (29 * hash) + unknownFields.hashCode();
+        memoizedHashCode = hash;
+        return hash;
+      }
+
+      public static org.tensorflow.proto.framework.CostGraphDef.Node.InputInfo parseFrom(
+          java.nio.ByteBuffer data)
+          throws com.google.protobuf.InvalidProtocolBufferException {
+        return PARSER.parseFrom(data);
+      }
+      public static org.tensorflow.proto.framework.CostGraphDef.Node.InputInfo parseFrom(
+          java.nio.ByteBuffer data,
+          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+          throws com.google.protobuf.InvalidProtocolBufferException {
+        return PARSER.parseFrom(data, extensionRegistry);
+      }
+      public static org.tensorflow.proto.framework.CostGraphDef.Node.InputInfo parseFrom(
+          com.google.protobuf.ByteString data)
+          throws com.google.protobuf.InvalidProtocolBufferException {
+        return PARSER.parseFrom(data);
+      }
+      public static org.tensorflow.proto.framework.CostGraphDef.Node.InputInfo parseFrom(
+          com.google.protobuf.ByteString data,
+          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+          throws com.google.protobuf.InvalidProtocolBufferException {
+        return PARSER.parseFrom(data, extensionRegistry);
+      }
+      public static org.tensorflow.proto.framework.CostGraphDef.Node.InputInfo parseFrom(byte[] data)
+          throws com.google.protobuf.InvalidProtocolBufferException {
+        return PARSER.parseFrom(data);
+      }
+      public static org.tensorflow.proto.framework.CostGraphDef.Node.InputInfo parseFrom(
+          byte[] data,
+          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+          throws com.google.protobuf.InvalidProtocolBufferException {
+        return PARSER.parseFrom(data, extensionRegistry);
+      }
+      public static org.tensorflow.proto.framework.CostGraphDef.Node.InputInfo parseFrom(java.io.InputStream input)
+          throws java.io.IOException {
+        return com.google.protobuf.GeneratedMessageV3
+            .parseWithIOException(PARSER, input);
+      }
+      public static org.tensorflow.proto.framework.CostGraphDef.Node.InputInfo parseFrom(
+          java.io.InputStream input,
+          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+          throws java.io.IOException {
+        return com.google.protobuf.GeneratedMessageV3
+            .parseWithIOException(PARSER, input, extensionRegistry);
+      }
+      public static org.tensorflow.proto.framework.CostGraphDef.Node.InputInfo parseDelimitedFrom(java.io.InputStream input)
+          throws java.io.IOException {
+        return com.google.protobuf.GeneratedMessageV3
+            .parseDelimitedWithIOException(PARSER, input);
+      }
+      public static org.tensorflow.proto.framework.CostGraphDef.Node.InputInfo parseDelimitedFrom(
+          java.io.InputStream input,
+          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+          throws java.io.IOException {
+        return com.google.protobuf.GeneratedMessageV3
+            .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+      }
+      public static org.tensorflow.proto.framework.CostGraphDef.Node.InputInfo parseFrom(
+          com.google.protobuf.CodedInputStream input)
+          throws java.io.IOException {
+        return com.google.protobuf.GeneratedMessageV3
+            .parseWithIOException(PARSER, input);
+      }
+      public static org.tensorflow.proto.framework.CostGraphDef.Node.InputInfo parseFrom(
+          com.google.protobuf.CodedInputStream input,
+          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+          throws java.io.IOException {
+        return com.google.protobuf.GeneratedMessageV3
+            .parseWithIOException(PARSER, input, extensionRegistry);
+      }
+
+      @java.lang.Override
+      public Builder newBuilderForType() { return newBuilder(); }
+      public static Builder newBuilder() {
+        return DEFAULT_INSTANCE.toBuilder();
+      }
+      public static Builder newBuilder(org.tensorflow.proto.framework.CostGraphDef.Node.InputInfo prototype) {
+        return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+      }
+      @java.lang.Override
+      public Builder toBuilder() {
+        return this == DEFAULT_INSTANCE
+            ? new Builder() : new Builder().mergeFrom(this);
+      }
+
+      @java.lang.Override
+      protected Builder newBuilderForType(
+          com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+        Builder builder = new Builder(parent);
+        return builder;
+      }
+      /**
+       * <pre>
+       * Inputs of this node. They must be executed before this node can be
+       * executed. An input is a particular output of another node, specified
+       * by the node id and the output index.
+       * </pre>
+       *
+       * Protobuf type {@code tensorflow.CostGraphDef.Node.InputInfo}
+       */
+      public static final class Builder extends
+          com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+          // @@protoc_insertion_point(builder_implements:tensorflow.CostGraphDef.Node.InputInfo)
+          org.tensorflow.proto.framework.CostGraphDef.Node.InputInfoOrBuilder {
+        public static final com.google.protobuf.Descriptors.Descriptor
+            getDescriptor() {
+          return org.tensorflow.proto.framework.CostGraphProtos.internal_static_tensorflow_CostGraphDef_Node_InputInfo_descriptor;
+        }
+
+        @java.lang.Override
+        protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+            internalGetFieldAccessorTable() {
+          return org.tensorflow.proto.framework.CostGraphProtos.internal_static_tensorflow_CostGraphDef_Node_InputInfo_fieldAccessorTable
+              .ensureFieldAccessorsInitialized(
+                  org.tensorflow.proto.framework.CostGraphDef.Node.InputInfo.class, org.tensorflow.proto.framework.CostGraphDef.Node.InputInfo.Builder.class);
+        }
+
+        // Construct using org.tensorflow.proto.framework.CostGraphDef.Node.InputInfo.newBuilder()
+        private Builder() {
+          maybeForceBuilderInitialization();
+        }
+
+        private Builder(
+            com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+          super(parent);
+          maybeForceBuilderInitialization();
+        }
+        private void maybeForceBuilderInitialization() {
+          if (com.google.protobuf.GeneratedMessageV3
+                  .alwaysUseFieldBuilders) {
+          }
+        }
+        @java.lang.Override
+        public Builder clear() {
+          super.clear();
+          precedingNode_ = 0;
+
+          precedingPort_ = 0;
+
+          return this;
+        }
+
+        @java.lang.Override
+        public com.google.protobuf.Descriptors.Descriptor
+            getDescriptorForType() {
+          return org.tensorflow.proto.framework.CostGraphProtos.internal_static_tensorflow_CostGraphDef_Node_InputInfo_descriptor;
+        }
+
+        @java.lang.Override
+        public org.tensorflow.proto.framework.CostGraphDef.Node.InputInfo getDefaultInstanceForType() {
+          return org.tensorflow.proto.framework.CostGraphDef.Node.InputInfo.getDefaultInstance();
+        }
+
+        @java.lang.Override
+        public org.tensorflow.proto.framework.CostGraphDef.Node.InputInfo build() {
+          org.tensorflow.proto.framework.CostGraphDef.Node.InputInfo result = buildPartial();
+          if (!result.isInitialized()) {
+            throw newUninitializedMessageException(result);
+          }
+          return result;
+        }
+
+        @java.lang.Override
+        public org.tensorflow.proto.framework.CostGraphDef.Node.InputInfo buildPartial() {
+          org.tensorflow.proto.framework.CostGraphDef.Node.InputInfo result = new org.tensorflow.proto.framework.CostGraphDef.Node.InputInfo(this);
+          result.precedingNode_ = precedingNode_;
+          result.precedingPort_ = precedingPort_;
+          onBuilt();
+          return result;
+        }
+
+        @java.lang.Override
+        public Builder clone() {
+          return super.clone();
+        }
+        @java.lang.Override
+        public Builder setField(
+            com.google.protobuf.Descriptors.FieldDescriptor field,
+            java.lang.Object value) {
+          return super.setField(field, value);
+        }
+        @java.lang.Override
+        public Builder clearField(
+            com.google.protobuf.Descriptors.FieldDescriptor field) {
+          return super.clearField(field);
+        }
+        @java.lang.Override
+        public Builder clearOneof(
+            com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+          return super.clearOneof(oneof);
+        }
+        @java.lang.Override
+        public Builder setRepeatedField(
+            com.google.protobuf.Descriptors.FieldDescriptor field,
+            int index, java.lang.Object value) {
+          return super.setRepeatedField(field, index, value);
+        }
+        @java.lang.Override
+        public Builder addRepeatedField(
+            com.google.protobuf.Descriptors.FieldDescriptor field,
+            java.lang.Object value) {
+          return super.addRepeatedField(field, value);
+        }
+        @java.lang.Override
+        public Builder mergeFrom(com.google.protobuf.Message other) {
+          if (other instanceof org.tensorflow.proto.framework.CostGraphDef.Node.InputInfo) {
+            return mergeFrom((org.tensorflow.proto.framework.CostGraphDef.Node.InputInfo)other);
+          } else {
+            super.mergeFrom(other);
+            return this;
+          }
+        }
+
+        public Builder mergeFrom(org.tensorflow.proto.framework.CostGraphDef.Node.InputInfo other) {
+          if (other == org.tensorflow.proto.framework.CostGraphDef.Node.InputInfo.getDefaultInstance()) return this;
+          if (other.getPrecedingNode() != 0) {
+            setPrecedingNode(other.getPrecedingNode());
+          }
+          if (other.getPrecedingPort() != 0) {
+            setPrecedingPort(other.getPrecedingPort());
+          }
+          this.mergeUnknownFields(other.unknownFields);
+          onChanged();
+          return this;
+        }
+
+        @java.lang.Override
+        public final boolean isInitialized() {
+          return true;
+        }
+
+        @java.lang.Override
+        public Builder mergeFrom(
+            com.google.protobuf.CodedInputStream input,
+            com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+            throws java.io.IOException {
+          org.tensorflow.proto.framework.CostGraphDef.Node.InputInfo parsedMessage = null;
+          try {
+            parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+          } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+            parsedMessage = (org.tensorflow.proto.framework.CostGraphDef.Node.InputInfo) e.getUnfinishedMessage();
+            throw e.unwrapIOException();
+          } finally {
+            if (parsedMessage != null) {
+              mergeFrom(parsedMessage);
+            }
+          }
+          return this;
+        }
+
+        private int precedingNode_ ;
+        /**
+         * <code>int32 preceding_node = 1;</code>
+         */
+        public int getPrecedingNode() {
+          return precedingNode_;
+        }
+        /**
+         * <code>int32 preceding_node = 1;</code>
+         */
+        public Builder setPrecedingNode(int value) {
+          
+          precedingNode_ = value;
+          onChanged();
+          return this;
+        }
+        /**
+         * <code>int32 preceding_node = 1;</code>
+         */
+        public Builder clearPrecedingNode() {
+          
+          precedingNode_ = 0;
+          onChanged();
+          return this;
+        }
+
+        private int precedingPort_ ;
+        /**
+         * <code>int32 preceding_port = 2;</code>
+         */
+        public int getPrecedingPort() {
+          return precedingPort_;
+        }
+        /**
+         * <code>int32 preceding_port = 2;</code>
+         */
+        public Builder setPrecedingPort(int value) {
+          
+          precedingPort_ = value;
+          onChanged();
+          return this;
+        }
+        /**
+         * <code>int32 preceding_port = 2;</code>
+         */
+        public Builder clearPrecedingPort() {
+          
+          precedingPort_ = 0;
+          onChanged();
+          return this;
+        }
+        @java.lang.Override
+        public final Builder setUnknownFields(
+            final com.google.protobuf.UnknownFieldSet unknownFields) {
+          return super.setUnknownFields(unknownFields);
+        }
+
+        @java.lang.Override
+        public final Builder mergeUnknownFields(
+            final com.google.protobuf.UnknownFieldSet unknownFields) {
+          return super.mergeUnknownFields(unknownFields);
+        }
+
+
+        // @@protoc_insertion_point(builder_scope:tensorflow.CostGraphDef.Node.InputInfo)
+      }
+
+      // @@protoc_insertion_point(class_scope:tensorflow.CostGraphDef.Node.InputInfo)
+      private static final org.tensorflow.proto.framework.CostGraphDef.Node.InputInfo DEFAULT_INSTANCE;
+      static {
+        DEFAULT_INSTANCE = new org.tensorflow.proto.framework.CostGraphDef.Node.InputInfo();
+      }
+
+      public static org.tensorflow.proto.framework.CostGraphDef.Node.InputInfo getDefaultInstance() {
+        return DEFAULT_INSTANCE;
+      }
+
+      private static final com.google.protobuf.Parser<InputInfo>
+          PARSER = new com.google.protobuf.AbstractParser<InputInfo>() {
+        @java.lang.Override
+        public InputInfo parsePartialFrom(
+            com.google.protobuf.CodedInputStream input,
+            com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+            throws com.google.protobuf.InvalidProtocolBufferException {
+          return new InputInfo(input, extensionRegistry);
+        }
+      };
+
+      public static com.google.protobuf.Parser<InputInfo> parser() {
+        return PARSER;
+      }
+
+      @java.lang.Override
+      public com.google.protobuf.Parser<InputInfo> getParserForType() {
+        return PARSER;
+      }
+
+      @java.lang.Override
+      public org.tensorflow.proto.framework.CostGraphDef.Node.InputInfo getDefaultInstanceForType() {
+        return DEFAULT_INSTANCE;
+      }
+
+    }
+
+    public interface OutputInfoOrBuilder extends
+        // @@protoc_insertion_point(interface_extends:tensorflow.CostGraphDef.Node.OutputInfo)
+        com.google.protobuf.MessageOrBuilder {
+
+      /**
+       * <code>int64 size = 1;</code>
+       */
+      long getSize();
+
+      /**
+       * <pre>
+       * If &gt;= 0, the output is an alias of an input. Note that an alias input
+       * may itself be an alias. The algorithm will therefore need to follow
+       * those pointers.
+       * </pre>
+       *
+       * <code>int64 alias_input_port = 2;</code>
+       */
+      long getAliasInputPort();
+
+      /**
+       * <code>.tensorflow.TensorShapeProto shape = 3;</code>
+       */
+      boolean hasShape();
+      /**
+       * <code>.tensorflow.TensorShapeProto shape = 3;</code>
+       */
+      org.tensorflow.proto.framework.TensorShapeProto getShape();
+      /**
+       * <code>.tensorflow.TensorShapeProto shape = 3;</code>
+       */
+      org.tensorflow.proto.framework.TensorShapeProtoOrBuilder getShapeOrBuilder();
+
+      /**
+       * <code>.tensorflow.DataType dtype = 4;</code>
+       */
+      int getDtypeValue();
+      /**
+       * <code>.tensorflow.DataType dtype = 4;</code>
+       */
+      org.tensorflow.proto.framework.DataType getDtype();
+    }
+    /**
+     * <pre>
+     * Outputs of this node.
+     * </pre>
+     *
+     * Protobuf type {@code tensorflow.CostGraphDef.Node.OutputInfo}
+     */
+    public  static final class OutputInfo extends
+        com.google.protobuf.GeneratedMessageV3 implements
+        // @@protoc_insertion_point(message_implements:tensorflow.CostGraphDef.Node.OutputInfo)
+        OutputInfoOrBuilder {
+    private static final long serialVersionUID = 0L;
+      // Use OutputInfo.newBuilder() to construct.
+      private OutputInfo(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+        super(builder);
+      }
+      private OutputInfo() {
+        dtype_ = 0;
+      }
+
+      @java.lang.Override
+      @SuppressWarnings({"unused"})
+      protected java.lang.Object newInstance(
+          UnusedPrivateParameter unused) {
+        return new OutputInfo();
+      }
+
+      @java.lang.Override
+      public final com.google.protobuf.UnknownFieldSet
+      getUnknownFields() {
+        return this.unknownFields;
+      }
+      private OutputInfo(
+          com.google.protobuf.CodedInputStream input,
+          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+          throws com.google.protobuf.InvalidProtocolBufferException {
+        this();
+        if (extensionRegistry == null) {
+          throw new java.lang.NullPointerException();
+        }
+        com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+            com.google.protobuf.UnknownFieldSet.newBuilder();
+        try {
+          boolean done = false;
+          while (!done) {
+            int tag = input.readTag();
+            switch (tag) {
+              case 0:
+                done = true;
+                break;
+              case 8: {
+
+                size_ = input.readInt64();
+                break;
+              }
+              case 16: {
+
+                aliasInputPort_ = input.readInt64();
+                break;
+              }
+              case 26: {
+                org.tensorflow.proto.framework.TensorShapeProto.Builder subBuilder = null;
+                if (shape_ != null) {
+                  subBuilder = shape_.toBuilder();
+                }
+                shape_ = input.readMessage(org.tensorflow.proto.framework.TensorShapeProto.parser(), extensionRegistry);
+                if (subBuilder != null) {
+                  subBuilder.mergeFrom(shape_);
+                  shape_ = subBuilder.buildPartial();
+                }
+
+                break;
+              }
+              case 32: {
+                int rawValue = input.readEnum();
+
+                dtype_ = rawValue;
+                break;
+              }
+              default: {
+                if (!parseUnknownField(
+                    input, unknownFields, extensionRegistry, tag)) {
+                  done = true;
+                }
+                break;
+              }
+            }
+          }
+        } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+          throw e.setUnfinishedMessage(this);
+        } catch (java.io.IOException e) {
+          throw new com.google.protobuf.InvalidProtocolBufferException(
+              e).setUnfinishedMessage(this);
+        } finally {
+          this.unknownFields = unknownFields.build();
+          makeExtensionsImmutable();
+        }
+      }
+      public static final com.google.protobuf.Descriptors.Descriptor
+          getDescriptor() {
+        return org.tensorflow.proto.framework.CostGraphProtos.internal_static_tensorflow_CostGraphDef_Node_OutputInfo_descriptor;
+      }
+
+      @java.lang.Override
+      protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+          internalGetFieldAccessorTable() {
+        return org.tensorflow.proto.framework.CostGraphProtos.internal_static_tensorflow_CostGraphDef_Node_OutputInfo_fieldAccessorTable
+            .ensureFieldAccessorsInitialized(
+                org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfo.class, org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfo.Builder.class);
+      }
+
+      public static final int SIZE_FIELD_NUMBER = 1;
+      private long size_;
+      /**
+       * <code>int64 size = 1;</code>
+       */
+      public long getSize() {
+        return size_;
+      }
+
+      public static final int ALIAS_INPUT_PORT_FIELD_NUMBER = 2;
+      private long aliasInputPort_;
+      /**
+       * <pre>
+       * If &gt;= 0, the output is an alias of an input. Note that an alias input
+       * may itself be an alias. The algorithm will therefore need to follow
+       * those pointers.
+       * </pre>
+       *
+       * <code>int64 alias_input_port = 2;</code>
+       */
+      public long getAliasInputPort() {
+        return aliasInputPort_;
+      }
+
+      public static final int SHAPE_FIELD_NUMBER = 3;
+      private org.tensorflow.proto.framework.TensorShapeProto shape_;
+      /**
+       * <code>.tensorflow.TensorShapeProto shape = 3;</code>
+       */
+      public boolean hasShape() {
+        return shape_ != null;
+      }
+      /**
+       * <code>.tensorflow.TensorShapeProto shape = 3;</code>
+       */
+      public org.tensorflow.proto.framework.TensorShapeProto getShape() {
+        return shape_ == null ? org.tensorflow.proto.framework.TensorShapeProto.getDefaultInstance() : shape_;
+      }
+      /**
+       * <code>.tensorflow.TensorShapeProto shape = 3;</code>
+       */
+      public org.tensorflow.proto.framework.TensorShapeProtoOrBuilder getShapeOrBuilder() {
+        return getShape();
+      }
+
+      public static final int DTYPE_FIELD_NUMBER = 4;
+      private int dtype_;
+      /**
+       * <code>.tensorflow.DataType dtype = 4;</code>
+       */
+      public int getDtypeValue() {
+        return dtype_;
+      }
+      /**
+       * <code>.tensorflow.DataType dtype = 4;</code>
+       */
+      public org.tensorflow.proto.framework.DataType getDtype() {
+        @SuppressWarnings("deprecation")
+        org.tensorflow.proto.framework.DataType result = org.tensorflow.proto.framework.DataType.valueOf(dtype_);
+        return result == null ? org.tensorflow.proto.framework.DataType.UNRECOGNIZED : result;
+      }
+
+      private byte memoizedIsInitialized = -1;
+      @java.lang.Override
+      public final boolean isInitialized() {
+        byte isInitialized = memoizedIsInitialized;
+        if (isInitialized == 1) return true;
+        if (isInitialized == 0) return false;
+
+        memoizedIsInitialized = 1;
+        return true;
+      }
+
+      @java.lang.Override
+      public void writeTo(com.google.protobuf.CodedOutputStream output)
+                          throws java.io.IOException {
+        if (size_ != 0L) {
+          output.writeInt64(1, size_);
+        }
+        if (aliasInputPort_ != 0L) {
+          output.writeInt64(2, aliasInputPort_);
+        }
+        if (shape_ != null) {
+          output.writeMessage(3, getShape());
+        }
+        if (dtype_ != org.tensorflow.proto.framework.DataType.DT_INVALID.getNumber()) {
+          output.writeEnum(4, dtype_);
+        }
+        unknownFields.writeTo(output);
+      }
+
+      @java.lang.Override
+      public int getSerializedSize() {
+        int size = memoizedSize;
+        if (size != -1) return size;
+
+        size = 0;
+        if (size_ != 0L) {
+          size += com.google.protobuf.CodedOutputStream
+            .computeInt64Size(1, size_);
+        }
+        if (aliasInputPort_ != 0L) {
+          size += com.google.protobuf.CodedOutputStream
+            .computeInt64Size(2, aliasInputPort_);
+        }
+        if (shape_ != null) {
+          size += com.google.protobuf.CodedOutputStream
+            .computeMessageSize(3, getShape());
+        }
+        if (dtype_ != org.tensorflow.proto.framework.DataType.DT_INVALID.getNumber()) {
+          size += com.google.protobuf.CodedOutputStream
+            .computeEnumSize(4, dtype_);
+        }
+        size += unknownFields.getSerializedSize();
+        memoizedSize = size;
+        return size;
+      }
+
+      @java.lang.Override
+      public boolean equals(final java.lang.Object obj) {
+        if (obj == this) {
+         return true;
+        }
+        if (!(obj instanceof org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfo)) {
+          return super.equals(obj);
+        }
+        org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfo other = (org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfo) obj;
+
+        if (getSize()
+            != other.getSize()) return false;
+        if (getAliasInputPort()
+            != other.getAliasInputPort()) return false;
+        if (hasShape() != other.hasShape()) return false;
+        if (hasShape()) {
+          if (!getShape()
+              .equals(other.getShape())) return false;
+        }
+        if (dtype_ != other.dtype_) return false;
+        if (!unknownFields.equals(other.unknownFields)) return false;
+        return true;
+      }
+
+      @java.lang.Override
+      public int hashCode() {
+        if (memoizedHashCode != 0) {
+          return memoizedHashCode;
+        }
+        int hash = 41;
+        hash = (19 * hash) + getDescriptor().hashCode();
+        hash = (37 * hash) + SIZE_FIELD_NUMBER;
+        hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+            getSize());
+        hash = (37 * hash) + ALIAS_INPUT_PORT_FIELD_NUMBER;
+        hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+            getAliasInputPort());
+        if (hasShape()) {
+          hash = (37 * hash) + SHAPE_FIELD_NUMBER;
+          hash = (53 * hash) + getShape().hashCode();
+        }
+        hash = (37 * hash) + DTYPE_FIELD_NUMBER;
+        hash = (53 * hash) + dtype_;
+        hash = (29 * hash) + unknownFields.hashCode();
+        memoizedHashCode = hash;
+        return hash;
+      }
+
+      public static org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfo parseFrom(
+          java.nio.ByteBuffer data)
+          throws com.google.protobuf.InvalidProtocolBufferException {
+        return PARSER.parseFrom(data);
+      }
+      public static org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfo parseFrom(
+          java.nio.ByteBuffer data,
+          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+          throws com.google.protobuf.InvalidProtocolBufferException {
+        return PARSER.parseFrom(data, extensionRegistry);
+      }
+      public static org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfo parseFrom(
+          com.google.protobuf.ByteString data)
+          throws com.google.protobuf.InvalidProtocolBufferException {
+        return PARSER.parseFrom(data);
+      }
+      public static org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfo parseFrom(
+          com.google.protobuf.ByteString data,
+          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+          throws com.google.protobuf.InvalidProtocolBufferException {
+        return PARSER.parseFrom(data, extensionRegistry);
+      }
+      public static org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfo parseFrom(byte[] data)
+          throws com.google.protobuf.InvalidProtocolBufferException {
+        return PARSER.parseFrom(data);
+      }
+      public static org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfo parseFrom(
+          byte[] data,
+          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+          throws com.google.protobuf.InvalidProtocolBufferException {
+        return PARSER.parseFrom(data, extensionRegistry);
+      }
+      public static org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfo parseFrom(java.io.InputStream input)
+          throws java.io.IOException {
+        return com.google.protobuf.GeneratedMessageV3
+            .parseWithIOException(PARSER, input);
+      }
+      public static org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfo parseFrom(
+          java.io.InputStream input,
+          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+          throws java.io.IOException {
+        return com.google.protobuf.GeneratedMessageV3
+            .parseWithIOException(PARSER, input, extensionRegistry);
+      }
+      public static org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfo parseDelimitedFrom(java.io.InputStream input)
+          throws java.io.IOException {
+        return com.google.protobuf.GeneratedMessageV3
+            .parseDelimitedWithIOException(PARSER, input);
+      }
+      public static org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfo parseDelimitedFrom(
+          java.io.InputStream input,
+          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+          throws java.io.IOException {
+        return com.google.protobuf.GeneratedMessageV3
+            .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+      }
+      public static org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfo parseFrom(
+          com.google.protobuf.CodedInputStream input)
+          throws java.io.IOException {
+        return com.google.protobuf.GeneratedMessageV3
+            .parseWithIOException(PARSER, input);
+      }
+      public static org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfo parseFrom(
+          com.google.protobuf.CodedInputStream input,
+          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+          throws java.io.IOException {
+        return com.google.protobuf.GeneratedMessageV3
+            .parseWithIOException(PARSER, input, extensionRegistry);
+      }
+
+      @java.lang.Override
+      public Builder newBuilderForType() { return newBuilder(); }
+      public static Builder newBuilder() {
+        return DEFAULT_INSTANCE.toBuilder();
+      }
+      public static Builder newBuilder(org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfo prototype) {
+        return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+      }
+      @java.lang.Override
+      public Builder toBuilder() {
+        return this == DEFAULT_INSTANCE
+            ? new Builder() : new Builder().mergeFrom(this);
+      }
+
+      @java.lang.Override
+      protected Builder newBuilderForType(
+          com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+        Builder builder = new Builder(parent);
+        return builder;
+      }
+      /**
+       * <pre>
+       * Outputs of this node.
+       * </pre>
+       *
+       * Protobuf type {@code tensorflow.CostGraphDef.Node.OutputInfo}
+       */
+      public static final class Builder extends
+          com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+          // @@protoc_insertion_point(builder_implements:tensorflow.CostGraphDef.Node.OutputInfo)
+          org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfoOrBuilder {
+        public static final com.google.protobuf.Descriptors.Descriptor
+            getDescriptor() {
+          return org.tensorflow.proto.framework.CostGraphProtos.internal_static_tensorflow_CostGraphDef_Node_OutputInfo_descriptor;
+        }
+
+        @java.lang.Override
+        protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+            internalGetFieldAccessorTable() {
+          return org.tensorflow.proto.framework.CostGraphProtos.internal_static_tensorflow_CostGraphDef_Node_OutputInfo_fieldAccessorTable
+              .ensureFieldAccessorsInitialized(
+                  org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfo.class, org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfo.Builder.class);
+        }
+
+        // Construct using org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfo.newBuilder()
+        private Builder() {
+          maybeForceBuilderInitialization();
+        }
+
+        private Builder(
+            com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+          super(parent);
+          maybeForceBuilderInitialization();
+        }
+        private void maybeForceBuilderInitialization() {
+          if (com.google.protobuf.GeneratedMessageV3
+                  .alwaysUseFieldBuilders) {
+          }
+        }
+        @java.lang.Override
+        public Builder clear() {
+          super.clear();
+          size_ = 0L;
+
+          aliasInputPort_ = 0L;
+
+          if (shapeBuilder_ == null) {
+            shape_ = null;
+          } else {
+            shape_ = null;
+            shapeBuilder_ = null;
+          }
+          dtype_ = 0;
+
+          return this;
+        }
+
+        @java.lang.Override
+        public com.google.protobuf.Descriptors.Descriptor
+            getDescriptorForType() {
+          return org.tensorflow.proto.framework.CostGraphProtos.internal_static_tensorflow_CostGraphDef_Node_OutputInfo_descriptor;
+        }
+
+        @java.lang.Override
+        public org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfo getDefaultInstanceForType() {
+          return org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfo.getDefaultInstance();
+        }
+
+        @java.lang.Override
+        public org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfo build() {
+          org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfo result = buildPartial();
+          if (!result.isInitialized()) {
+            throw newUninitializedMessageException(result);
+          }
+          return result;
+        }
+
+        @java.lang.Override
+        public org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfo buildPartial() {
+          org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfo result = new org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfo(this);
+          result.size_ = size_;
+          result.aliasInputPort_ = aliasInputPort_;
+          if (shapeBuilder_ == null) {
+            result.shape_ = shape_;
+          } else {
+            result.shape_ = shapeBuilder_.build();
+          }
+          result.dtype_ = dtype_;
+          onBuilt();
+          return result;
+        }
+
+        @java.lang.Override
+        public Builder clone() {
+          return super.clone();
+        }
+        @java.lang.Override
+        public Builder setField(
+            com.google.protobuf.Descriptors.FieldDescriptor field,
+            java.lang.Object value) {
+          return super.setField(field, value);
+        }
+        @java.lang.Override
+        public Builder clearField(
+            com.google.protobuf.Descriptors.FieldDescriptor field) {
+          return super.clearField(field);
+        }
+        @java.lang.Override
+        public Builder clearOneof(
+            com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+          return super.clearOneof(oneof);
+        }
+        @java.lang.Override
+        public Builder setRepeatedField(
+            com.google.protobuf.Descriptors.FieldDescriptor field,
+            int index, java.lang.Object value) {
+          return super.setRepeatedField(field, index, value);
+        }
+        @java.lang.Override
+        public Builder addRepeatedField(
+            com.google.protobuf.Descriptors.FieldDescriptor field,
+            java.lang.Object value) {
+          return super.addRepeatedField(field, value);
+        }
+        @java.lang.Override
+        public Builder mergeFrom(com.google.protobuf.Message other) {
+          if (other instanceof org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfo) {
+            return mergeFrom((org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfo)other);
+          } else {
+            super.mergeFrom(other);
+            return this;
+          }
+        }
+
+        public Builder mergeFrom(org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfo other) {
+          if (other == org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfo.getDefaultInstance()) return this;
+          if (other.getSize() != 0L) {
+            setSize(other.getSize());
+          }
+          if (other.getAliasInputPort() != 0L) {
+            setAliasInputPort(other.getAliasInputPort());
+          }
+          if (other.hasShape()) {
+            mergeShape(other.getShape());
+          }
+          if (other.dtype_ != 0) {
+            setDtypeValue(other.getDtypeValue());
+          }
+          this.mergeUnknownFields(other.unknownFields);
+          onChanged();
+          return this;
+        }
+
+        @java.lang.Override
+        public final boolean isInitialized() {
+          return true;
+        }
+
+        @java.lang.Override
+        public Builder mergeFrom(
+            com.google.protobuf.CodedInputStream input,
+            com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+            throws java.io.IOException {
+          org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfo parsedMessage = null;
+          try {
+            parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+          } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+            parsedMessage = (org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfo) e.getUnfinishedMessage();
+            throw e.unwrapIOException();
+          } finally {
+            if (parsedMessage != null) {
+              mergeFrom(parsedMessage);
+            }
+          }
+          return this;
+        }
+
+        private long size_ ;
+        /**
+         * <code>int64 size = 1;</code>
+         */
+        public long getSize() {
+          return size_;
+        }
+        /**
+         * <code>int64 size = 1;</code>
+         */
+        public Builder setSize(long value) {
+          
+          size_ = value;
+          onChanged();
+          return this;
+        }
+        /**
+         * <code>int64 size = 1;</code>
+         */
+        public Builder clearSize() {
+          
+          size_ = 0L;
+          onChanged();
+          return this;
+        }
+
+        private long aliasInputPort_ ;
+        /**
+         * <pre>
+         * If &gt;= 0, the output is an alias of an input. Note that an alias input
+         * may itself be an alias. The algorithm will therefore need to follow
+         * those pointers.
+         * </pre>
+         *
+         * <code>int64 alias_input_port = 2;</code>
+         */
+        public long getAliasInputPort() {
+          return aliasInputPort_;
+        }
+        /**
+         * <pre>
+         * If &gt;= 0, the output is an alias of an input. Note that an alias input
+         * may itself be an alias. The algorithm will therefore need to follow
+         * those pointers.
+         * </pre>
+         *
+         * <code>int64 alias_input_port = 2;</code>
+         */
+        public Builder setAliasInputPort(long value) {
+          
+          aliasInputPort_ = value;
+          onChanged();
+          return this;
+        }
+        /**
+         * <pre>
+         * If &gt;= 0, the output is an alias of an input. Note that an alias input
+         * may itself be an alias. The algorithm will therefore need to follow
+         * those pointers.
+         * </pre>
+         *
+         * <code>int64 alias_input_port = 2;</code>
+         */
+        public Builder clearAliasInputPort() {
+          
+          aliasInputPort_ = 0L;
+          onChanged();
+          return this;
+        }
+
+        private org.tensorflow.proto.framework.TensorShapeProto shape_;
+        private com.google.protobuf.SingleFieldBuilderV3<
+            org.tensorflow.proto.framework.TensorShapeProto, org.tensorflow.proto.framework.TensorShapeProto.Builder, org.tensorflow.proto.framework.TensorShapeProtoOrBuilder> shapeBuilder_;
+        /**
+         * <code>.tensorflow.TensorShapeProto shape = 3;</code>
+         */
+        public boolean hasShape() {
+          return shapeBuilder_ != null || shape_ != null;
+        }
+        /**
+         * <code>.tensorflow.TensorShapeProto shape = 3;</code>
+         */
+        public org.tensorflow.proto.framework.TensorShapeProto getShape() {
+          if (shapeBuilder_ == null) {
+            return shape_ == null ? org.tensorflow.proto.framework.TensorShapeProto.getDefaultInstance() : shape_;
+          } else {
+            return shapeBuilder_.getMessage();
+          }
+        }
+        /**
+         * <code>.tensorflow.TensorShapeProto shape = 3;</code>
+         */
+        public Builder setShape(org.tensorflow.proto.framework.TensorShapeProto value) {
+          if (shapeBuilder_ == null) {
+            if (value == null) {
+              throw new NullPointerException();
+            }
+            shape_ = value;
+            onChanged();
+          } else {
+            shapeBuilder_.setMessage(value);
+          }
+
+          return this;
+        }
+        /**
+         * <code>.tensorflow.TensorShapeProto shape = 3;</code>
+         */
+        public Builder setShape(
+            org.tensorflow.proto.framework.TensorShapeProto.Builder builderForValue) {
+          if (shapeBuilder_ == null) {
+            shape_ = builderForValue.build();
+            onChanged();
+          } else {
+            shapeBuilder_.setMessage(builderForValue.build());
+          }
+
+          return this;
+        }
+        /**
+         * <code>.tensorflow.TensorShapeProto shape = 3;</code>
+         */
+        public Builder mergeShape(org.tensorflow.proto.framework.TensorShapeProto value) {
+          if (shapeBuilder_ == null) {
+            if (shape_ != null) {
+              shape_ =
+                org.tensorflow.proto.framework.TensorShapeProto.newBuilder(shape_).mergeFrom(value).buildPartial();
+            } else {
+              shape_ = value;
+            }
+            onChanged();
+          } else {
+            shapeBuilder_.mergeFrom(value);
+          }
+
+          return this;
+        }
+        /**
+         * <code>.tensorflow.TensorShapeProto shape = 3;</code>
+         */
+        public Builder clearShape() {
+          if (shapeBuilder_ == null) {
+            shape_ = null;
+            onChanged();
+          } else {
+            shape_ = null;
+            shapeBuilder_ = null;
+          }
+
+          return this;
+        }
+        /**
+         * <code>.tensorflow.TensorShapeProto shape = 3;</code>
+         */
+        public org.tensorflow.proto.framework.TensorShapeProto.Builder getShapeBuilder() {
+          
+          onChanged();
+          return getShapeFieldBuilder().getBuilder();
+        }
+        /**
+         * <code>.tensorflow.TensorShapeProto shape = 3;</code>
+         */
+        public org.tensorflow.proto.framework.TensorShapeProtoOrBuilder getShapeOrBuilder() {
+          if (shapeBuilder_ != null) {
+            return shapeBuilder_.getMessageOrBuilder();
+          } else {
+            return shape_ == null ?
+                org.tensorflow.proto.framework.TensorShapeProto.getDefaultInstance() : shape_;
+          }
+        }
+        /**
+         * <code>.tensorflow.TensorShapeProto shape = 3;</code>
+         */
+        private com.google.protobuf.SingleFieldBuilderV3<
+            org.tensorflow.proto.framework.TensorShapeProto, org.tensorflow.proto.framework.TensorShapeProto.Builder, org.tensorflow.proto.framework.TensorShapeProtoOrBuilder> 
+            getShapeFieldBuilder() {
+          if (shapeBuilder_ == null) {
+            shapeBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+                org.tensorflow.proto.framework.TensorShapeProto, org.tensorflow.proto.framework.TensorShapeProto.Builder, org.tensorflow.proto.framework.TensorShapeProtoOrBuilder>(
+                    getShape(),
+                    getParentForChildren(),
+                    isClean());
+            shape_ = null;
+          }
+          return shapeBuilder_;
+        }
+
+        private int dtype_ = 0;
+        /**
+         * <code>.tensorflow.DataType dtype = 4;</code>
+         */
+        public int getDtypeValue() {
+          return dtype_;
+        }
+        /**
+         * <code>.tensorflow.DataType dtype = 4;</code>
+         */
+        public Builder setDtypeValue(int value) {
+          dtype_ = value;
+          onChanged();
+          return this;
+        }
+        /**
+         * <code>.tensorflow.DataType dtype = 4;</code>
+         */
+        public org.tensorflow.proto.framework.DataType getDtype() {
+          @SuppressWarnings("deprecation")
+          org.tensorflow.proto.framework.DataType result = org.tensorflow.proto.framework.DataType.valueOf(dtype_);
+          return result == null ? org.tensorflow.proto.framework.DataType.UNRECOGNIZED : result;
+        }
+        /**
+         * <code>.tensorflow.DataType dtype = 4;</code>
+         */
+        public Builder setDtype(org.tensorflow.proto.framework.DataType value) {
+          if (value == null) {
+            throw new NullPointerException();
+          }
+          
+          dtype_ = value.getNumber();
+          onChanged();
+          return this;
+        }
+        /**
+         * <code>.tensorflow.DataType dtype = 4;</code>
+         */
+        public Builder clearDtype() {
+          
+          dtype_ = 0;
+          onChanged();
+          return this;
+        }
+        @java.lang.Override
+        public final Builder setUnknownFields(
+            final com.google.protobuf.UnknownFieldSet unknownFields) {
+          return super.setUnknownFields(unknownFields);
+        }
+
+        @java.lang.Override
+        public final Builder mergeUnknownFields(
+            final com.google.protobuf.UnknownFieldSet unknownFields) {
+          return super.mergeUnknownFields(unknownFields);
+        }
+
+
+        // @@protoc_insertion_point(builder_scope:tensorflow.CostGraphDef.Node.OutputInfo)
+      }
+
+      // @@protoc_insertion_point(class_scope:tensorflow.CostGraphDef.Node.OutputInfo)
+      private static final org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfo DEFAULT_INSTANCE;
+      static {
+        DEFAULT_INSTANCE = new org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfo();
+      }
+
+      public static org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfo getDefaultInstance() {
+        return DEFAULT_INSTANCE;
+      }
+
+      private static final com.google.protobuf.Parser<OutputInfo>
+          PARSER = new com.google.protobuf.AbstractParser<OutputInfo>() {
+        @java.lang.Override
+        public OutputInfo parsePartialFrom(
+            com.google.protobuf.CodedInputStream input,
+            com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+            throws com.google.protobuf.InvalidProtocolBufferException {
+          return new OutputInfo(input, extensionRegistry);
+        }
+      };
+
+      public static com.google.protobuf.Parser<OutputInfo> parser() {
+        return PARSER;
+      }
+
+      @java.lang.Override
+      public com.google.protobuf.Parser<OutputInfo> getParserForType() {
+        return PARSER;
+      }
+
+      @java.lang.Override
+      public org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfo getDefaultInstanceForType() {
+        return DEFAULT_INSTANCE;
+      }
+
+    }
+
+    public static final int NAME_FIELD_NUMBER = 1;
+    private volatile java.lang.Object name_;
+    /**
+     * <pre>
+     * The name of the node. Names are globally unique.
+     * </pre>
+     *
+     * <code>string name = 1;</code>
+     */
+    public java.lang.String getName() {
+      java.lang.Object ref = name_;
+      if (ref instanceof java.lang.String) {
+        return (java.lang.String) ref;
+      } else {
+        com.google.protobuf.ByteString bs = 
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        name_ = s;
+        return s;
+      }
+    }
+    /**
+     * <pre>
+     * The name of the node. Names are globally unique.
+     * </pre>
+     *
+     * <code>string name = 1;</code>
+     */
+    public com.google.protobuf.ByteString
+        getNameBytes() {
+      java.lang.Object ref = name_;
+      if (ref instanceof java.lang.String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        name_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+
+    public static final int DEVICE_FIELD_NUMBER = 2;
+    private volatile java.lang.Object device_;
+    /**
+     * <pre>
+     * The device of the node. Can be empty if the node is mapped to the
+     * default partition or partitioning hasn't been run yet.
+     * </pre>
+     *
+     * <code>string device = 2;</code>
+     */
+    public java.lang.String getDevice() {
+      java.lang.Object ref = device_;
+      if (ref instanceof java.lang.String) {
+        return (java.lang.String) ref;
+      } else {
+        com.google.protobuf.ByteString bs = 
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        device_ = s;
+        return s;
+      }
+    }
+    /**
+     * <pre>
+     * The device of the node. Can be empty if the node is mapped to the
+     * default partition or partitioning hasn't been run yet.
+     * </pre>
+     *
+     * <code>string device = 2;</code>
+     */
+    public com.google.protobuf.ByteString
+        getDeviceBytes() {
+      java.lang.Object ref = device_;
+      if (ref instanceof java.lang.String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        device_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+
+    public static final int ID_FIELD_NUMBER = 3;
+    private int id_;
+    /**
+     * <pre>
+     * The id of the node. Node ids are only unique inside a partition.
+     * </pre>
+     *
+     * <code>int32 id = 3;</code>
+     */
+    public int getId() {
+      return id_;
+    }
+
+    public static final int INPUT_INFO_FIELD_NUMBER = 4;
+    private java.util.List<org.tensorflow.proto.framework.CostGraphDef.Node.InputInfo> inputInfo_;
+    /**
+     * <code>repeated .tensorflow.CostGraphDef.Node.InputInfo input_info = 4;</code>
+     */
+    public java.util.List<org.tensorflow.proto.framework.CostGraphDef.Node.InputInfo> getInputInfoList() {
+      return inputInfo_;
+    }
+    /**
+     * <code>repeated .tensorflow.CostGraphDef.Node.InputInfo input_info = 4;</code>
+     */
+    public java.util.List<? extends org.tensorflow.proto.framework.CostGraphDef.Node.InputInfoOrBuilder> 
+        getInputInfoOrBuilderList() {
+      return inputInfo_;
+    }
+    /**
+     * <code>repeated .tensorflow.CostGraphDef.Node.InputInfo input_info = 4;</code>
+     */
+    public int getInputInfoCount() {
+      return inputInfo_.size();
+    }
+    /**
+     * <code>repeated .tensorflow.CostGraphDef.Node.InputInfo input_info = 4;</code>
+     */
+    public org.tensorflow.proto.framework.CostGraphDef.Node.InputInfo getInputInfo(int index) {
+      return inputInfo_.get(index);
+    }
+    /**
+     * <code>repeated .tensorflow.CostGraphDef.Node.InputInfo input_info = 4;</code>
+     */
+    public org.tensorflow.proto.framework.CostGraphDef.Node.InputInfoOrBuilder getInputInfoOrBuilder(
+        int index) {
+      return inputInfo_.get(index);
+    }
+
+    public static final int OUTPUT_INFO_FIELD_NUMBER = 5;
+    private java.util.List<org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfo> outputInfo_;
+    /**
+     * <code>repeated .tensorflow.CostGraphDef.Node.OutputInfo output_info = 5;</code>
+     */
+    public java.util.List<org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfo> getOutputInfoList() {
+      return outputInfo_;
+    }
+    /**
+     * <code>repeated .tensorflow.CostGraphDef.Node.OutputInfo output_info = 5;</code>
+     */
+    public java.util.List<? extends org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfoOrBuilder> 
+        getOutputInfoOrBuilderList() {
+      return outputInfo_;
+    }
+    /**
+     * <code>repeated .tensorflow.CostGraphDef.Node.OutputInfo output_info = 5;</code>
+     */
+    public int getOutputInfoCount() {
+      return outputInfo_.size();
+    }
+    /**
+     * <code>repeated .tensorflow.CostGraphDef.Node.OutputInfo output_info = 5;</code>
+     */
+    public org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfo getOutputInfo(int index) {
+      return outputInfo_.get(index);
+    }
+    /**
+     * <code>repeated .tensorflow.CostGraphDef.Node.OutputInfo output_info = 5;</code>
+     */
+    public org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfoOrBuilder getOutputInfoOrBuilder(
+        int index) {
+      return outputInfo_.get(index);
+    }
+
+    public static final int TEMPORARY_MEMORY_SIZE_FIELD_NUMBER = 6;
+    private long temporaryMemorySize_;
+    /**
+     * <pre>
+     * Temporary memory used by this node.
+     * </pre>
+     *
+     * <code>int64 temporary_memory_size = 6;</code>
+     */
+    public long getTemporaryMemorySize() {
+      return temporaryMemorySize_;
+    }
+
+    public static final int PERSISTENT_MEMORY_SIZE_FIELD_NUMBER = 12;
+    private long persistentMemorySize_;
+    /**
+     * <pre>
+     * Persistent memory used by this node.
+     * </pre>
+     *
+     * <code>int64 persistent_memory_size = 12;</code>
+     */
+    public long getPersistentMemorySize() {
+      return persistentMemorySize_;
+    }
+
+    public static final int HOST_TEMP_MEMORY_SIZE_FIELD_NUMBER = 10;
+    private long hostTempMemorySize_;
+    /**
+     * <code>int64 host_temp_memory_size = 10 [deprecated = true];</code>
+     */
+    @java.lang.Deprecated public long getHostTempMemorySize() {
+      return hostTempMemorySize_;
+    }
+
+    public static final int DEVICE_TEMP_MEMORY_SIZE_FIELD_NUMBER = 11;
+    private long deviceTempMemorySize_;
+    /**
+     * <code>int64 device_temp_memory_size = 11 [deprecated = true];</code>
+     */
+    @java.lang.Deprecated public long getDeviceTempMemorySize() {
+      return deviceTempMemorySize_;
+    }
+
+    public static final int DEVICE_PERSISTENT_MEMORY_SIZE_FIELD_NUMBER = 16;
+    private long devicePersistentMemorySize_;
+    /**
+     * <code>int64 device_persistent_memory_size = 16 [deprecated = true];</code>
+     */
+    @java.lang.Deprecated public long getDevicePersistentMemorySize() {
+      return devicePersistentMemorySize_;
+    }
+
+    public static final int COMPUTE_COST_FIELD_NUMBER = 9;
+    private long computeCost_;
+    /**
+     * <pre>
+     * Estimate of the computational cost of this node, in microseconds.
+     * </pre>
+     *
+     * <code>int64 compute_cost = 9;</code>
+     */
+    public long getComputeCost() {
+      return computeCost_;
+    }
+
+    public static final int COMPUTE_TIME_FIELD_NUMBER = 14;
+    private long computeTime_;
+    /**
+     * <pre>
+     * Analytical estimate of the computational cost of this node, in
+     * microseconds.
+     * </pre>
+     *
+     * <code>int64 compute_time = 14;</code>
+     */
+    public long getComputeTime() {
+      return computeTime_;
+    }
+
+    public static final int MEMORY_TIME_FIELD_NUMBER = 15;
+    private long memoryTime_;
+    /**
+     * <pre>
+     * Analytical estimate of the memory access cost of this node, in
+     * microseconds.
+     * </pre>
+     *
+     * <code>int64 memory_time = 15;</code>
+     */
+    public long getMemoryTime() {
+      return memoryTime_;
+    }
+
+    public static final int IS_FINAL_FIELD_NUMBER = 7;
+    private boolean isFinal_;
+    /**
+     * <pre>
+     * If true, the output is permanent: it can't be discarded, because this
+     * node is part of the "final output". Nodes may depend on final nodes.
+     * </pre>
+     *
+     * <code>bool is_final = 7;</code>
+     */
+    public boolean getIsFinal() {
+      return isFinal_;
+    }
+
+    public static final int CONTROL_INPUT_FIELD_NUMBER = 8;
+    private com.google.protobuf.Internal.IntList controlInput_;
+    /**
+     * <pre>
+     * Ids of the control inputs for this node.
+     * </pre>
+     *
+     * <code>repeated int32 control_input = 8;</code>
+     */
+    public java.util.List<java.lang.Integer>
+        getControlInputList() {
+      return controlInput_;
+    }
+    /**
+     * <pre>
+     * Ids of the control inputs for this node.
+     * </pre>
+     *
+     * <code>repeated int32 control_input = 8;</code>
+     */
+    public int getControlInputCount() {
+      return controlInput_.size();
+    }
+    /**
+     * <pre>
+     * Ids of the control inputs for this node.
+     * </pre>
+     *
+     * <code>repeated int32 control_input = 8;</code>
+     */
+    public int getControlInput(int index) {
+      return controlInput_.getInt(index);
+    }
+    private int controlInputMemoizedSerializedSize = -1;
+
+    public static final int INACCURATE_FIELD_NUMBER = 17;
+    private boolean inaccurate_;
+    /**
+     * <pre>
+     * Are the costs inaccurate?
+     * </pre>
+     *
+     * <code>bool inaccurate = 17;</code>
+     */
+    public boolean getInaccurate() {
+      return inaccurate_;
+    }
+
+    private byte memoizedIsInitialized = -1;
+    @java.lang.Override
+    public final boolean isInitialized() {
+      byte isInitialized = memoizedIsInitialized;
+      if (isInitialized == 1) return true;
+      if (isInitialized == 0) return false;
+
+      memoizedIsInitialized = 1;
+      return true;
+    }
+
+    @java.lang.Override
+    public void writeTo(com.google.protobuf.CodedOutputStream output)
+                        throws java.io.IOException {
+      getSerializedSize();
+      if (!getNameBytes().isEmpty()) {
+        com.google.protobuf.GeneratedMessageV3.writeString(output, 1, name_);
+      }
+      if (!getDeviceBytes().isEmpty()) {
+        com.google.protobuf.GeneratedMessageV3.writeString(output, 2, device_);
+      }
+      if (id_ != 0) {
+        output.writeInt32(3, id_);
+      }
+      for (int i = 0; i < inputInfo_.size(); i++) {
+        output.writeMessage(4, inputInfo_.get(i));
+      }
+      for (int i = 0; i < outputInfo_.size(); i++) {
+        output.writeMessage(5, outputInfo_.get(i));
+      }
+      if (temporaryMemorySize_ != 0L) {
+        output.writeInt64(6, temporaryMemorySize_);
+      }
+      if (isFinal_ != false) {
+        output.writeBool(7, isFinal_);
+      }
+      if (getControlInputList().size() > 0) {
+        output.writeUInt32NoTag(66);
+        output.writeUInt32NoTag(controlInputMemoizedSerializedSize);
+      }
+      for (int i = 0; i < controlInput_.size(); i++) {
+        output.writeInt32NoTag(controlInput_.getInt(i));
+      }
+      if (computeCost_ != 0L) {
+        output.writeInt64(9, computeCost_);
+      }
+      if (hostTempMemorySize_ != 0L) {
+        output.writeInt64(10, hostTempMemorySize_);
+      }
+      if (deviceTempMemorySize_ != 0L) {
+        output.writeInt64(11, deviceTempMemorySize_);
+      }
+      if (persistentMemorySize_ != 0L) {
+        output.writeInt64(12, persistentMemorySize_);
+      }
+      if (computeTime_ != 0L) {
+        output.writeInt64(14, computeTime_);
+      }
+      if (memoryTime_ != 0L) {
+        output.writeInt64(15, memoryTime_);
+      }
+      if (devicePersistentMemorySize_ != 0L) {
+        output.writeInt64(16, devicePersistentMemorySize_);
+      }
+      if (inaccurate_ != false) {
+        output.writeBool(17, inaccurate_);
+      }
+      unknownFields.writeTo(output);
+    }
+
+    @java.lang.Override
+    public int getSerializedSize() {
+      int size = memoizedSize;
+      if (size != -1) return size;
+
+      size = 0;
+      if (!getNameBytes().isEmpty()) {
+        size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, name_);
+      }
+      if (!getDeviceBytes().isEmpty()) {
+        size += com.google.protobuf.GeneratedMessageV3.computeStringSize(2, device_);
+      }
+      if (id_ != 0) {
+        size += com.google.protobuf.CodedOutputStream
+          .computeInt32Size(3, id_);
+      }
+      for (int i = 0; i < inputInfo_.size(); i++) {
+        size += com.google.protobuf.CodedOutputStream
+          .computeMessageSize(4, inputInfo_.get(i));
+      }
+      for (int i = 0; i < outputInfo_.size(); i++) {
+        size += com.google.protobuf.CodedOutputStream
+          .computeMessageSize(5, outputInfo_.get(i));
+      }
+      if (temporaryMemorySize_ != 0L) {
+        size += com.google.protobuf.CodedOutputStream
+          .computeInt64Size(6, temporaryMemorySize_);
+      }
+      if (isFinal_ != false) {
+        size += com.google.protobuf.CodedOutputStream
+          .computeBoolSize(7, isFinal_);
+      }
+      {
+        int dataSize = 0;
+        for (int i = 0; i < controlInput_.size(); i++) {
+          dataSize += com.google.protobuf.CodedOutputStream
+            .computeInt32SizeNoTag(controlInput_.getInt(i));
+        }
+        size += dataSize;
+        if (!getControlInputList().isEmpty()) {
+          size += 1;
+          size += com.google.protobuf.CodedOutputStream
+              .computeInt32SizeNoTag(dataSize);
+        }
+        controlInputMemoizedSerializedSize = dataSize;
+      }
+      if (computeCost_ != 0L) {
+        size += com.google.protobuf.CodedOutputStream
+          .computeInt64Size(9, computeCost_);
+      }
+      if (hostTempMemorySize_ != 0L) {
+        size += com.google.protobuf.CodedOutputStream
+          .computeInt64Size(10, hostTempMemorySize_);
+      }
+      if (deviceTempMemorySize_ != 0L) {
+        size += com.google.protobuf.CodedOutputStream
+          .computeInt64Size(11, deviceTempMemorySize_);
+      }
+      if (persistentMemorySize_ != 0L) {
+        size += com.google.protobuf.CodedOutputStream
+          .computeInt64Size(12, persistentMemorySize_);
+      }
+      if (computeTime_ != 0L) {
+        size += com.google.protobuf.CodedOutputStream
+          .computeInt64Size(14, computeTime_);
+      }
+      if (memoryTime_ != 0L) {
+        size += com.google.protobuf.CodedOutputStream
+          .computeInt64Size(15, memoryTime_);
+      }
+      if (devicePersistentMemorySize_ != 0L) {
+        size += com.google.protobuf.CodedOutputStream
+          .computeInt64Size(16, devicePersistentMemorySize_);
+      }
+      if (inaccurate_ != false) {
+        size += com.google.protobuf.CodedOutputStream
+          .computeBoolSize(17, inaccurate_);
+      }
+      size += unknownFields.getSerializedSize();
+      memoizedSize = size;
+      return size;
+    }
+
+    @java.lang.Override
+    public boolean equals(final java.lang.Object obj) {
+      if (obj == this) {
+       return true;
+      }
+      if (!(obj instanceof org.tensorflow.proto.framework.CostGraphDef.Node)) {
+        return super.equals(obj);
+      }
+      org.tensorflow.proto.framework.CostGraphDef.Node other = (org.tensorflow.proto.framework.CostGraphDef.Node) obj;
+
+      if (!getName()
+          .equals(other.getName())) return false;
+      if (!getDevice()
+          .equals(other.getDevice())) return false;
+      if (getId()
+          != other.getId()) return false;
+      if (!getInputInfoList()
+          .equals(other.getInputInfoList())) return false;
+      if (!getOutputInfoList()
+          .equals(other.getOutputInfoList())) return false;
+      if (getTemporaryMemorySize()
+          != other.getTemporaryMemorySize()) return false;
+      if (getPersistentMemorySize()
+          != other.getPersistentMemorySize()) return false;
+      if (getHostTempMemorySize()
+          != other.getHostTempMemorySize()) return false;
+      if (getDeviceTempMemorySize()
+          != other.getDeviceTempMemorySize()) return false;
+      if (getDevicePersistentMemorySize()
+          != other.getDevicePersistentMemorySize()) return false;
+      if (getComputeCost()
+          != other.getComputeCost()) return false;
+      if (getComputeTime()
+          != other.getComputeTime()) return false;
+      if (getMemoryTime()
+          != other.getMemoryTime()) return false;
+      if (getIsFinal()
+          != other.getIsFinal()) return false;
+      if (!getControlInputList()
+          .equals(other.getControlInputList())) return false;
+      if (getInaccurate()
+          != other.getInaccurate()) return false;
+      if (!unknownFields.equals(other.unknownFields)) return false;
+      return true;
+    }
+
+    @java.lang.Override
+    public int hashCode() {
+      if (memoizedHashCode != 0) {
+        return memoizedHashCode;
+      }
+      int hash = 41;
+      hash = (19 * hash) + getDescriptor().hashCode();
+      hash = (37 * hash) + NAME_FIELD_NUMBER;
+      hash = (53 * hash) + getName().hashCode();
+      hash = (37 * hash) + DEVICE_FIELD_NUMBER;
+      hash = (53 * hash) + getDevice().hashCode();
+      hash = (37 * hash) + ID_FIELD_NUMBER;
+      hash = (53 * hash) + getId();
+      if (getInputInfoCount() > 0) {
+        hash = (37 * hash) + INPUT_INFO_FIELD_NUMBER;
+        hash = (53 * hash) + getInputInfoList().hashCode();
+      }
+      if (getOutputInfoCount() > 0) {
+        hash = (37 * hash) + OUTPUT_INFO_FIELD_NUMBER;
+        hash = (53 * hash) + getOutputInfoList().hashCode();
+      }
+      hash = (37 * hash) + TEMPORARY_MEMORY_SIZE_FIELD_NUMBER;
+      hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+          getTemporaryMemorySize());
+      hash = (37 * hash) + PERSISTENT_MEMORY_SIZE_FIELD_NUMBER;
+      hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+          getPersistentMemorySize());
+      hash = (37 * hash) + HOST_TEMP_MEMORY_SIZE_FIELD_NUMBER;
+      hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+          getHostTempMemorySize());
+      hash = (37 * hash) + DEVICE_TEMP_MEMORY_SIZE_FIELD_NUMBER;
+      hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+          getDeviceTempMemorySize());
+      hash = (37 * hash) + DEVICE_PERSISTENT_MEMORY_SIZE_FIELD_NUMBER;
+      hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+          getDevicePersistentMemorySize());
+      hash = (37 * hash) + COMPUTE_COST_FIELD_NUMBER;
+      hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+          getComputeCost());
+      hash = (37 * hash) + COMPUTE_TIME_FIELD_NUMBER;
+      hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+          getComputeTime());
+      hash = (37 * hash) + MEMORY_TIME_FIELD_NUMBER;
+      hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+          getMemoryTime());
+      hash = (37 * hash) + IS_FINAL_FIELD_NUMBER;
+      hash = (53 * hash) + com.google.protobuf.Internal.hashBoolean(
+          getIsFinal());
+      if (getControlInputCount() > 0) {
+        hash = (37 * hash) + CONTROL_INPUT_FIELD_NUMBER;
+        hash = (53 * hash) + getControlInputList().hashCode();
+      }
+      hash = (37 * hash) + INACCURATE_FIELD_NUMBER;
+      hash = (53 * hash) + com.google.protobuf.Internal.hashBoolean(
+          getInaccurate());
+      hash = (29 * hash) + unknownFields.hashCode();
+      memoizedHashCode = hash;
+      return hash;
+    }
+
+    public static org.tensorflow.proto.framework.CostGraphDef.Node parseFrom(
+        java.nio.ByteBuffer data)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data);
+    }
+    public static org.tensorflow.proto.framework.CostGraphDef.Node parseFrom(
+        java.nio.ByteBuffer data,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.CostGraphDef.Node parseFrom(
+        com.google.protobuf.ByteString data)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data);
+    }
+    public static org.tensorflow.proto.framework.CostGraphDef.Node parseFrom(
+        com.google.protobuf.ByteString data,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.CostGraphDef.Node parseFrom(byte[] data)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data);
+    }
+    public static org.tensorflow.proto.framework.CostGraphDef.Node parseFrom(
+        byte[] data,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.CostGraphDef.Node parseFrom(java.io.InputStream input)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input);
+    }
+    public static org.tensorflow.proto.framework.CostGraphDef.Node parseFrom(
+        java.io.InputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.CostGraphDef.Node parseDelimitedFrom(java.io.InputStream input)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseDelimitedWithIOException(PARSER, input);
+    }
+    public static org.tensorflow.proto.framework.CostGraphDef.Node parseDelimitedFrom(
+        java.io.InputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.CostGraphDef.Node parseFrom(
+        com.google.protobuf.CodedInputStream input)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input);
+    }
+    public static org.tensorflow.proto.framework.CostGraphDef.Node parseFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input, extensionRegistry);
+    }
+
+    @java.lang.Override
+    public Builder newBuilderForType() { return newBuilder(); }
+    public static Builder newBuilder() {
+      return DEFAULT_INSTANCE.toBuilder();
+    }
+    public static Builder newBuilder(org.tensorflow.proto.framework.CostGraphDef.Node prototype) {
+      return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+    }
+    @java.lang.Override
+    public Builder toBuilder() {
+      return this == DEFAULT_INSTANCE
+          ? new Builder() : new Builder().mergeFrom(this);
+    }
+
+    @java.lang.Override
+    protected Builder newBuilderForType(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      Builder builder = new Builder(parent);
+      return builder;
+    }
+    /**
+     * Protobuf type {@code tensorflow.CostGraphDef.Node}
+     */
+    public static final class Builder extends
+        com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+        // @@protoc_insertion_point(builder_implements:tensorflow.CostGraphDef.Node)
+        org.tensorflow.proto.framework.CostGraphDef.NodeOrBuilder {
+      public static final com.google.protobuf.Descriptors.Descriptor
+          getDescriptor() {
+        return org.tensorflow.proto.framework.CostGraphProtos.internal_static_tensorflow_CostGraphDef_Node_descriptor;
+      }
+
+      @java.lang.Override
+      protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+          internalGetFieldAccessorTable() {
+        return org.tensorflow.proto.framework.CostGraphProtos.internal_static_tensorflow_CostGraphDef_Node_fieldAccessorTable
+            .ensureFieldAccessorsInitialized(
+                org.tensorflow.proto.framework.CostGraphDef.Node.class, org.tensorflow.proto.framework.CostGraphDef.Node.Builder.class);
+      }
+
+      // Construct using org.tensorflow.proto.framework.CostGraphDef.Node.newBuilder()
+      private Builder() {
+        maybeForceBuilderInitialization();
+      }
+
+      private Builder(
+          com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+        super(parent);
+        maybeForceBuilderInitialization();
+      }
+      private void maybeForceBuilderInitialization() {
+        if (com.google.protobuf.GeneratedMessageV3
+                .alwaysUseFieldBuilders) {
+          getInputInfoFieldBuilder();
+          getOutputInfoFieldBuilder();
+        }
+      }
+      @java.lang.Override
+      public Builder clear() {
+        super.clear();
+        name_ = "";
+
+        device_ = "";
+
+        id_ = 0;
+
+        if (inputInfoBuilder_ == null) {
+          inputInfo_ = java.util.Collections.emptyList();
+          bitField0_ = (bitField0_ & ~0x00000001);
+        } else {
+          inputInfoBuilder_.clear();
+        }
+        if (outputInfoBuilder_ == null) {
+          outputInfo_ = java.util.Collections.emptyList();
+          bitField0_ = (bitField0_ & ~0x00000002);
+        } else {
+          outputInfoBuilder_.clear();
+        }
+        temporaryMemorySize_ = 0L;
+
+        persistentMemorySize_ = 0L;
+
+        hostTempMemorySize_ = 0L;
+
+        deviceTempMemorySize_ = 0L;
+
+        devicePersistentMemorySize_ = 0L;
+
+        computeCost_ = 0L;
+
+        computeTime_ = 0L;
+
+        memoryTime_ = 0L;
+
+        isFinal_ = false;
+
+        controlInput_ = emptyIntList();
+        bitField0_ = (bitField0_ & ~0x00000004);
+        inaccurate_ = false;
+
+        return this;
+      }
+
+      @java.lang.Override
+      public com.google.protobuf.Descriptors.Descriptor
+          getDescriptorForType() {
+        return org.tensorflow.proto.framework.CostGraphProtos.internal_static_tensorflow_CostGraphDef_Node_descriptor;
+      }
+
+      @java.lang.Override
+      public org.tensorflow.proto.framework.CostGraphDef.Node getDefaultInstanceForType() {
+        return org.tensorflow.proto.framework.CostGraphDef.Node.getDefaultInstance();
+      }
+
+      @java.lang.Override
+      public org.tensorflow.proto.framework.CostGraphDef.Node build() {
+        org.tensorflow.proto.framework.CostGraphDef.Node result = buildPartial();
+        if (!result.isInitialized()) {
+          throw newUninitializedMessageException(result);
+        }
+        return result;
+      }
+
+      @java.lang.Override
+      public org.tensorflow.proto.framework.CostGraphDef.Node buildPartial() {
+        org.tensorflow.proto.framework.CostGraphDef.Node result = new org.tensorflow.proto.framework.CostGraphDef.Node(this);
+        int from_bitField0_ = bitField0_;
+        result.name_ = name_;
+        result.device_ = device_;
+        result.id_ = id_;
+        if (inputInfoBuilder_ == null) {
+          if (((bitField0_ & 0x00000001) != 0)) {
+            inputInfo_ = java.util.Collections.unmodifiableList(inputInfo_);
+            bitField0_ = (bitField0_ & ~0x00000001);
+          }
+          result.inputInfo_ = inputInfo_;
+        } else {
+          result.inputInfo_ = inputInfoBuilder_.build();
+        }
+        if (outputInfoBuilder_ == null) {
+          if (((bitField0_ & 0x00000002) != 0)) {
+            outputInfo_ = java.util.Collections.unmodifiableList(outputInfo_);
+            bitField0_ = (bitField0_ & ~0x00000002);
+          }
+          result.outputInfo_ = outputInfo_;
+        } else {
+          result.outputInfo_ = outputInfoBuilder_.build();
+        }
+        result.temporaryMemorySize_ = temporaryMemorySize_;
+        result.persistentMemorySize_ = persistentMemorySize_;
+        result.hostTempMemorySize_ = hostTempMemorySize_;
+        result.deviceTempMemorySize_ = deviceTempMemorySize_;
+        result.devicePersistentMemorySize_ = devicePersistentMemorySize_;
+        result.computeCost_ = computeCost_;
+        result.computeTime_ = computeTime_;
+        result.memoryTime_ = memoryTime_;
+        result.isFinal_ = isFinal_;
+        if (((bitField0_ & 0x00000004) != 0)) {
+          controlInput_.makeImmutable();
+          bitField0_ = (bitField0_ & ~0x00000004);
+        }
+        result.controlInput_ = controlInput_;
+        result.inaccurate_ = inaccurate_;
+        onBuilt();
+        return result;
+      }
+
+      @java.lang.Override
+      public Builder clone() {
+        return super.clone();
+      }
+      @java.lang.Override
+      public Builder setField(
+          com.google.protobuf.Descriptors.FieldDescriptor field,
+          java.lang.Object value) {
+        return super.setField(field, value);
+      }
+      @java.lang.Override
+      public Builder clearField(
+          com.google.protobuf.Descriptors.FieldDescriptor field) {
+        return super.clearField(field);
+      }
+      @java.lang.Override
+      public Builder clearOneof(
+          com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+        return super.clearOneof(oneof);
+      }
+      @java.lang.Override
+      public Builder setRepeatedField(
+          com.google.protobuf.Descriptors.FieldDescriptor field,
+          int index, java.lang.Object value) {
+        return super.setRepeatedField(field, index, value);
+      }
+      @java.lang.Override
+      public Builder addRepeatedField(
+          com.google.protobuf.Descriptors.FieldDescriptor field,
+          java.lang.Object value) {
+        return super.addRepeatedField(field, value);
+      }
+      @java.lang.Override
+      public Builder mergeFrom(com.google.protobuf.Message other) {
+        if (other instanceof org.tensorflow.proto.framework.CostGraphDef.Node) {
+          return mergeFrom((org.tensorflow.proto.framework.CostGraphDef.Node)other);
+        } else {
+          super.mergeFrom(other);
+          return this;
+        }
+      }
+
+      public Builder mergeFrom(org.tensorflow.proto.framework.CostGraphDef.Node other) {
+        if (other == org.tensorflow.proto.framework.CostGraphDef.Node.getDefaultInstance()) return this;
+        if (!other.getName().isEmpty()) {
+          name_ = other.name_;
+          onChanged();
+        }
+        if (!other.getDevice().isEmpty()) {
+          device_ = other.device_;
+          onChanged();
+        }
+        if (other.getId() != 0) {
+          setId(other.getId());
+        }
+        if (inputInfoBuilder_ == null) {
+          if (!other.inputInfo_.isEmpty()) {
+            if (inputInfo_.isEmpty()) {
+              inputInfo_ = other.inputInfo_;
+              bitField0_ = (bitField0_ & ~0x00000001);
+            } else {
+              ensureInputInfoIsMutable();
+              inputInfo_.addAll(other.inputInfo_);
+            }
+            onChanged();
+          }
+        } else {
+          if (!other.inputInfo_.isEmpty()) {
+            if (inputInfoBuilder_.isEmpty()) {
+              inputInfoBuilder_.dispose();
+              inputInfoBuilder_ = null;
+              inputInfo_ = other.inputInfo_;
+              bitField0_ = (bitField0_ & ~0x00000001);
+              inputInfoBuilder_ = 
+                com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
+                   getInputInfoFieldBuilder() : null;
+            } else {
+              inputInfoBuilder_.addAllMessages(other.inputInfo_);
+            }
+          }
+        }
+        if (outputInfoBuilder_ == null) {
+          if (!other.outputInfo_.isEmpty()) {
+            if (outputInfo_.isEmpty()) {
+              outputInfo_ = other.outputInfo_;
+              bitField0_ = (bitField0_ & ~0x00000002);
+            } else {
+              ensureOutputInfoIsMutable();
+              outputInfo_.addAll(other.outputInfo_);
+            }
+            onChanged();
+          }
+        } else {
+          if (!other.outputInfo_.isEmpty()) {
+            if (outputInfoBuilder_.isEmpty()) {
+              outputInfoBuilder_.dispose();
+              outputInfoBuilder_ = null;
+              outputInfo_ = other.outputInfo_;
+              bitField0_ = (bitField0_ & ~0x00000002);
+              outputInfoBuilder_ = 
+                com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
+                   getOutputInfoFieldBuilder() : null;
+            } else {
+              outputInfoBuilder_.addAllMessages(other.outputInfo_);
+            }
+          }
+        }
+        if (other.getTemporaryMemorySize() != 0L) {
+          setTemporaryMemorySize(other.getTemporaryMemorySize());
+        }
+        if (other.getPersistentMemorySize() != 0L) {
+          setPersistentMemorySize(other.getPersistentMemorySize());
+        }
+        if (other.getHostTempMemorySize() != 0L) {
+          setHostTempMemorySize(other.getHostTempMemorySize());
+        }
+        if (other.getDeviceTempMemorySize() != 0L) {
+          setDeviceTempMemorySize(other.getDeviceTempMemorySize());
+        }
+        if (other.getDevicePersistentMemorySize() != 0L) {
+          setDevicePersistentMemorySize(other.getDevicePersistentMemorySize());
+        }
+        if (other.getComputeCost() != 0L) {
+          setComputeCost(other.getComputeCost());
+        }
+        if (other.getComputeTime() != 0L) {
+          setComputeTime(other.getComputeTime());
+        }
+        if (other.getMemoryTime() != 0L) {
+          setMemoryTime(other.getMemoryTime());
+        }
+        if (other.getIsFinal() != false) {
+          setIsFinal(other.getIsFinal());
+        }
+        if (!other.controlInput_.isEmpty()) {
+          if (controlInput_.isEmpty()) {
+            controlInput_ = other.controlInput_;
+            bitField0_ = (bitField0_ & ~0x00000004);
+          } else {
+            ensureControlInputIsMutable();
+            controlInput_.addAll(other.controlInput_);
+          }
+          onChanged();
+        }
+        if (other.getInaccurate() != false) {
+          setInaccurate(other.getInaccurate());
+        }
+        this.mergeUnknownFields(other.unknownFields);
+        onChanged();
+        return this;
+      }
+
+      @java.lang.Override
+      public final boolean isInitialized() {
+        return true;
+      }
+
+      @java.lang.Override
+      public Builder mergeFrom(
+          com.google.protobuf.CodedInputStream input,
+          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+          throws java.io.IOException {
+        org.tensorflow.proto.framework.CostGraphDef.Node parsedMessage = null;
+        try {
+          parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+        } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+          parsedMessage = (org.tensorflow.proto.framework.CostGraphDef.Node) e.getUnfinishedMessage();
+          throw e.unwrapIOException();
+        } finally {
+          if (parsedMessage != null) {
+            mergeFrom(parsedMessage);
+          }
+        }
+        return this;
+      }
+      private int bitField0_;
+
+      private java.lang.Object name_ = "";
+      /**
+       * <pre>
+       * The name of the node. Names are globally unique.
+       * </pre>
+       *
+       * <code>string name = 1;</code>
+       */
+      public java.lang.String getName() {
+        java.lang.Object ref = name_;
+        if (!(ref instanceof java.lang.String)) {
+          com.google.protobuf.ByteString bs =
+              (com.google.protobuf.ByteString) ref;
+          java.lang.String s = bs.toStringUtf8();
+          name_ = s;
+          return s;
+        } else {
+          return (java.lang.String) ref;
+        }
+      }
+      /**
+       * <pre>
+       * The name of the node. Names are globally unique.
+       * </pre>
+       *
+       * <code>string name = 1;</code>
+       */
+      public com.google.protobuf.ByteString
+          getNameBytes() {
+        java.lang.Object ref = name_;
+        if (ref instanceof String) {
+          com.google.protobuf.ByteString b = 
+              com.google.protobuf.ByteString.copyFromUtf8(
+                  (java.lang.String) ref);
+          name_ = b;
+          return b;
+        } else {
+          return (com.google.protobuf.ByteString) ref;
+        }
+      }
+      /**
+       * <pre>
+       * The name of the node. Names are globally unique.
+       * </pre>
+       *
+       * <code>string name = 1;</code>
+       */
+      public Builder setName(
+          java.lang.String value) {
+        if (value == null) {
+    throw new NullPointerException();
+  }
+  
+        name_ = value;
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * The name of the node. Names are globally unique.
+       * </pre>
+       *
+       * <code>string name = 1;</code>
+       */
+      public Builder clearName() {
+        
+        name_ = getDefaultInstance().getName();
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * The name of the node. Names are globally unique.
+       * </pre>
+       *
+       * <code>string name = 1;</code>
+       */
+      public Builder setNameBytes(
+          com.google.protobuf.ByteString value) {
+        if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+        
+        name_ = value;
+        onChanged();
+        return this;
+      }
+
+      private java.lang.Object device_ = "";
+      /**
+       * <pre>
+       * The device of the node. Can be empty if the node is mapped to the
+       * default partition or partitioning hasn't been run yet.
+       * </pre>
+       *
+       * <code>string device = 2;</code>
+       */
+      public java.lang.String getDevice() {
+        java.lang.Object ref = device_;
+        if (!(ref instanceof java.lang.String)) {
+          com.google.protobuf.ByteString bs =
+              (com.google.protobuf.ByteString) ref;
+          java.lang.String s = bs.toStringUtf8();
+          device_ = s;
+          return s;
+        } else {
+          return (java.lang.String) ref;
+        }
+      }
+      /**
+       * <pre>
+       * The device of the node. Can be empty if the node is mapped to the
+       * default partition or partitioning hasn't been run yet.
+       * </pre>
+       *
+       * <code>string device = 2;</code>
+       */
+      public com.google.protobuf.ByteString
+          getDeviceBytes() {
+        java.lang.Object ref = device_;
+        if (ref instanceof String) {
+          com.google.protobuf.ByteString b = 
+              com.google.protobuf.ByteString.copyFromUtf8(
+                  (java.lang.String) ref);
+          device_ = b;
+          return b;
+        } else {
+          return (com.google.protobuf.ByteString) ref;
+        }
+      }
+      /**
+       * <pre>
+       * The device of the node. Can be empty if the node is mapped to the
+       * default partition or partitioning hasn't been run yet.
+       * </pre>
+       *
+       * <code>string device = 2;</code>
+       */
+      public Builder setDevice(
+          java.lang.String value) {
+        if (value == null) {
+    throw new NullPointerException();
+  }
+  
+        device_ = value;
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * The device of the node. Can be empty if the node is mapped to the
+       * default partition or partitioning hasn't been run yet.
+       * </pre>
+       *
+       * <code>string device = 2;</code>
+       */
+      public Builder clearDevice() {
+        
+        device_ = getDefaultInstance().getDevice();
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * The device of the node. Can be empty if the node is mapped to the
+       * default partition or partitioning hasn't been run yet.
+       * </pre>
+       *
+       * <code>string device = 2;</code>
+       */
+      public Builder setDeviceBytes(
+          com.google.protobuf.ByteString value) {
+        if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+        
+        device_ = value;
+        onChanged();
+        return this;
+      }
+
+      private int id_ ;
+      /**
+       * <pre>
+       * The id of the node. Node ids are only unique inside a partition.
+       * </pre>
+       *
+       * <code>int32 id = 3;</code>
+       */
+      public int getId() {
+        return id_;
+      }
+      /**
+       * <pre>
+       * The id of the node. Node ids are only unique inside a partition.
+       * </pre>
+       *
+       * <code>int32 id = 3;</code>
+       */
+      public Builder setId(int value) {
+        
+        id_ = value;
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * The id of the node. Node ids are only unique inside a partition.
+       * </pre>
+       *
+       * <code>int32 id = 3;</code>
+       */
+      public Builder clearId() {
+        
+        id_ = 0;
+        onChanged();
+        return this;
+      }
+
+      private java.util.List<org.tensorflow.proto.framework.CostGraphDef.Node.InputInfo> inputInfo_ =
+        java.util.Collections.emptyList();
+      private void ensureInputInfoIsMutable() {
+        if (!((bitField0_ & 0x00000001) != 0)) {
+          inputInfo_ = new java.util.ArrayList<org.tensorflow.proto.framework.CostGraphDef.Node.InputInfo>(inputInfo_);
+          bitField0_ |= 0x00000001;
+         }
+      }
+
+      private com.google.protobuf.RepeatedFieldBuilderV3<
+          org.tensorflow.proto.framework.CostGraphDef.Node.InputInfo, org.tensorflow.proto.framework.CostGraphDef.Node.InputInfo.Builder, org.tensorflow.proto.framework.CostGraphDef.Node.InputInfoOrBuilder> inputInfoBuilder_;
+
+      /**
+       * <code>repeated .tensorflow.CostGraphDef.Node.InputInfo input_info = 4;</code>
+       */
+      public java.util.List<org.tensorflow.proto.framework.CostGraphDef.Node.InputInfo> getInputInfoList() {
+        if (inputInfoBuilder_ == null) {
+          return java.util.Collections.unmodifiableList(inputInfo_);
+        } else {
+          return inputInfoBuilder_.getMessageList();
+        }
+      }
+      /**
+       * <code>repeated .tensorflow.CostGraphDef.Node.InputInfo input_info = 4;</code>
+       */
+      public int getInputInfoCount() {
+        if (inputInfoBuilder_ == null) {
+          return inputInfo_.size();
+        } else {
+          return inputInfoBuilder_.getCount();
+        }
+      }
+      /**
+       * <code>repeated .tensorflow.CostGraphDef.Node.InputInfo input_info = 4;</code>
+       */
+      public org.tensorflow.proto.framework.CostGraphDef.Node.InputInfo getInputInfo(int index) {
+        if (inputInfoBuilder_ == null) {
+          return inputInfo_.get(index);
+        } else {
+          return inputInfoBuilder_.getMessage(index);
+        }
+      }
+      /**
+       * <code>repeated .tensorflow.CostGraphDef.Node.InputInfo input_info = 4;</code>
+       */
+      public Builder setInputInfo(
+          int index, org.tensorflow.proto.framework.CostGraphDef.Node.InputInfo value) {
+        if (inputInfoBuilder_ == null) {
+          if (value == null) {
+            throw new NullPointerException();
+          }
+          ensureInputInfoIsMutable();
+          inputInfo_.set(index, value);
+          onChanged();
+        } else {
+          inputInfoBuilder_.setMessage(index, value);
+        }
+        return this;
+      }
+      /**
+       * <code>repeated .tensorflow.CostGraphDef.Node.InputInfo input_info = 4;</code>
+       */
+      public Builder setInputInfo(
+          int index, org.tensorflow.proto.framework.CostGraphDef.Node.InputInfo.Builder builderForValue) {
+        if (inputInfoBuilder_ == null) {
+          ensureInputInfoIsMutable();
+          inputInfo_.set(index, builderForValue.build());
+          onChanged();
+        } else {
+          inputInfoBuilder_.setMessage(index, builderForValue.build());
+        }
+        return this;
+      }
+      /**
+       * <code>repeated .tensorflow.CostGraphDef.Node.InputInfo input_info = 4;</code>
+       */
+      public Builder addInputInfo(org.tensorflow.proto.framework.CostGraphDef.Node.InputInfo value) {
+        if (inputInfoBuilder_ == null) {
+          if (value == null) {
+            throw new NullPointerException();
+          }
+          ensureInputInfoIsMutable();
+          inputInfo_.add(value);
+          onChanged();
+        } else {
+          inputInfoBuilder_.addMessage(value);
+        }
+        return this;
+      }
+      /**
+       * <code>repeated .tensorflow.CostGraphDef.Node.InputInfo input_info = 4;</code>
+       */
+      public Builder addInputInfo(
+          int index, org.tensorflow.proto.framework.CostGraphDef.Node.InputInfo value) {
+        if (inputInfoBuilder_ == null) {
+          if (value == null) {
+            throw new NullPointerException();
+          }
+          ensureInputInfoIsMutable();
+          inputInfo_.add(index, value);
+          onChanged();
+        } else {
+          inputInfoBuilder_.addMessage(index, value);
+        }
+        return this;
+      }
+      /**
+       * <code>repeated .tensorflow.CostGraphDef.Node.InputInfo input_info = 4;</code>
+       */
+      public Builder addInputInfo(
+          org.tensorflow.proto.framework.CostGraphDef.Node.InputInfo.Builder builderForValue) {
+        if (inputInfoBuilder_ == null) {
+          ensureInputInfoIsMutable();
+          inputInfo_.add(builderForValue.build());
+          onChanged();
+        } else {
+          inputInfoBuilder_.addMessage(builderForValue.build());
+        }
+        return this;
+      }
+      /**
+       * <code>repeated .tensorflow.CostGraphDef.Node.InputInfo input_info = 4;</code>
+       */
+      public Builder addInputInfo(
+          int index, org.tensorflow.proto.framework.CostGraphDef.Node.InputInfo.Builder builderForValue) {
+        if (inputInfoBuilder_ == null) {
+          ensureInputInfoIsMutable();
+          inputInfo_.add(index, builderForValue.build());
+          onChanged();
+        } else {
+          inputInfoBuilder_.addMessage(index, builderForValue.build());
+        }
+        return this;
+      }
+      /**
+       * <code>repeated .tensorflow.CostGraphDef.Node.InputInfo input_info = 4;</code>
+       */
+      public Builder addAllInputInfo(
+          java.lang.Iterable<? extends org.tensorflow.proto.framework.CostGraphDef.Node.InputInfo> values) {
+        if (inputInfoBuilder_ == null) {
+          ensureInputInfoIsMutable();
+          com.google.protobuf.AbstractMessageLite.Builder.addAll(
+              values, inputInfo_);
+          onChanged();
+        } else {
+          inputInfoBuilder_.addAllMessages(values);
+        }
+        return this;
+      }
+      /**
+       * <code>repeated .tensorflow.CostGraphDef.Node.InputInfo input_info = 4;</code>
+       */
+      public Builder clearInputInfo() {
+        if (inputInfoBuilder_ == null) {
+          inputInfo_ = java.util.Collections.emptyList();
+          bitField0_ = (bitField0_ & ~0x00000001);
+          onChanged();
+        } else {
+          inputInfoBuilder_.clear();
+        }
+        return this;
+      }
+      /**
+       * <code>repeated .tensorflow.CostGraphDef.Node.InputInfo input_info = 4;</code>
+       */
+      public Builder removeInputInfo(int index) {
+        if (inputInfoBuilder_ == null) {
+          ensureInputInfoIsMutable();
+          inputInfo_.remove(index);
+          onChanged();
+        } else {
+          inputInfoBuilder_.remove(index);
+        }
+        return this;
+      }
+      /**
+       * <code>repeated .tensorflow.CostGraphDef.Node.InputInfo input_info = 4;</code>
+       */
+      public org.tensorflow.proto.framework.CostGraphDef.Node.InputInfo.Builder getInputInfoBuilder(
+          int index) {
+        return getInputInfoFieldBuilder().getBuilder(index);
+      }
+      /**
+       * <code>repeated .tensorflow.CostGraphDef.Node.InputInfo input_info = 4;</code>
+       */
+      public org.tensorflow.proto.framework.CostGraphDef.Node.InputInfoOrBuilder getInputInfoOrBuilder(
+          int index) {
+        if (inputInfoBuilder_ == null) {
+          return inputInfo_.get(index);  } else {
+          return inputInfoBuilder_.getMessageOrBuilder(index);
+        }
+      }
+      /**
+       * <code>repeated .tensorflow.CostGraphDef.Node.InputInfo input_info = 4;</code>
+       */
+      public java.util.List<? extends org.tensorflow.proto.framework.CostGraphDef.Node.InputInfoOrBuilder> 
+           getInputInfoOrBuilderList() {
+        if (inputInfoBuilder_ != null) {
+          return inputInfoBuilder_.getMessageOrBuilderList();
+        } else {
+          return java.util.Collections.unmodifiableList(inputInfo_);
+        }
+      }
+      /**
+       * <code>repeated .tensorflow.CostGraphDef.Node.InputInfo input_info = 4;</code>
+       */
+      public org.tensorflow.proto.framework.CostGraphDef.Node.InputInfo.Builder addInputInfoBuilder() {
+        return getInputInfoFieldBuilder().addBuilder(
+            org.tensorflow.proto.framework.CostGraphDef.Node.InputInfo.getDefaultInstance());
+      }
+      /**
+       * <code>repeated .tensorflow.CostGraphDef.Node.InputInfo input_info = 4;</code>
+       */
+      public org.tensorflow.proto.framework.CostGraphDef.Node.InputInfo.Builder addInputInfoBuilder(
+          int index) {
+        return getInputInfoFieldBuilder().addBuilder(
+            index, org.tensorflow.proto.framework.CostGraphDef.Node.InputInfo.getDefaultInstance());
+      }
+      /**
+       * <code>repeated .tensorflow.CostGraphDef.Node.InputInfo input_info = 4;</code>
+       */
+      public java.util.List<org.tensorflow.proto.framework.CostGraphDef.Node.InputInfo.Builder> 
+           getInputInfoBuilderList() {
+        return getInputInfoFieldBuilder().getBuilderList();
+      }
+      private com.google.protobuf.RepeatedFieldBuilderV3<
+          org.tensorflow.proto.framework.CostGraphDef.Node.InputInfo, org.tensorflow.proto.framework.CostGraphDef.Node.InputInfo.Builder, org.tensorflow.proto.framework.CostGraphDef.Node.InputInfoOrBuilder> 
+          getInputInfoFieldBuilder() {
+        if (inputInfoBuilder_ == null) {
+          inputInfoBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
+              org.tensorflow.proto.framework.CostGraphDef.Node.InputInfo, org.tensorflow.proto.framework.CostGraphDef.Node.InputInfo.Builder, org.tensorflow.proto.framework.CostGraphDef.Node.InputInfoOrBuilder>(
+                  inputInfo_,
+                  ((bitField0_ & 0x00000001) != 0),
+                  getParentForChildren(),
+                  isClean());
+          inputInfo_ = null;
+        }
+        return inputInfoBuilder_;
+      }
+
+      private java.util.List<org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfo> outputInfo_ =
+        java.util.Collections.emptyList();
+      private void ensureOutputInfoIsMutable() {
+        if (!((bitField0_ & 0x00000002) != 0)) {
+          outputInfo_ = new java.util.ArrayList<org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfo>(outputInfo_);
+          bitField0_ |= 0x00000002;
+         }
+      }
+
+      private com.google.protobuf.RepeatedFieldBuilderV3<
+          org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfo, org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfo.Builder, org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfoOrBuilder> outputInfoBuilder_;
+
+      /**
+       * <code>repeated .tensorflow.CostGraphDef.Node.OutputInfo output_info = 5;</code>
+       */
+      public java.util.List<org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfo> getOutputInfoList() {
+        if (outputInfoBuilder_ == null) {
+          return java.util.Collections.unmodifiableList(outputInfo_);
+        } else {
+          return outputInfoBuilder_.getMessageList();
+        }
+      }
+      /**
+       * <code>repeated .tensorflow.CostGraphDef.Node.OutputInfo output_info = 5;</code>
+       */
+      public int getOutputInfoCount() {
+        if (outputInfoBuilder_ == null) {
+          return outputInfo_.size();
+        } else {
+          return outputInfoBuilder_.getCount();
+        }
+      }
+      /**
+       * <code>repeated .tensorflow.CostGraphDef.Node.OutputInfo output_info = 5;</code>
+       */
+      public org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfo getOutputInfo(int index) {
+        if (outputInfoBuilder_ == null) {
+          return outputInfo_.get(index);
+        } else {
+          return outputInfoBuilder_.getMessage(index);
+        }
+      }
+      /**
+       * <code>repeated .tensorflow.CostGraphDef.Node.OutputInfo output_info = 5;</code>
+       */
+      public Builder setOutputInfo(
+          int index, org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfo value) {
+        if (outputInfoBuilder_ == null) {
+          if (value == null) {
+            throw new NullPointerException();
+          }
+          ensureOutputInfoIsMutable();
+          outputInfo_.set(index, value);
+          onChanged();
+        } else {
+          outputInfoBuilder_.setMessage(index, value);
+        }
+        return this;
+      }
+      /**
+       * <code>repeated .tensorflow.CostGraphDef.Node.OutputInfo output_info = 5;</code>
+       */
+      public Builder setOutputInfo(
+          int index, org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfo.Builder builderForValue) {
+        if (outputInfoBuilder_ == null) {
+          ensureOutputInfoIsMutable();
+          outputInfo_.set(index, builderForValue.build());
+          onChanged();
+        } else {
+          outputInfoBuilder_.setMessage(index, builderForValue.build());
+        }
+        return this;
+      }
+      /**
+       * <code>repeated .tensorflow.CostGraphDef.Node.OutputInfo output_info = 5;</code>
+       */
+      public Builder addOutputInfo(org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfo value) {
+        if (outputInfoBuilder_ == null) {
+          if (value == null) {
+            throw new NullPointerException();
+          }
+          ensureOutputInfoIsMutable();
+          outputInfo_.add(value);
+          onChanged();
+        } else {
+          outputInfoBuilder_.addMessage(value);
+        }
+        return this;
+      }
+      /**
+       * <code>repeated .tensorflow.CostGraphDef.Node.OutputInfo output_info = 5;</code>
+       */
+      public Builder addOutputInfo(
+          int index, org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfo value) {
+        if (outputInfoBuilder_ == null) {
+          if (value == null) {
+            throw new NullPointerException();
+          }
+          ensureOutputInfoIsMutable();
+          outputInfo_.add(index, value);
+          onChanged();
+        } else {
+          outputInfoBuilder_.addMessage(index, value);
+        }
+        return this;
+      }
+      /**
+       * <code>repeated .tensorflow.CostGraphDef.Node.OutputInfo output_info = 5;</code>
+       */
+      public Builder addOutputInfo(
+          org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfo.Builder builderForValue) {
+        if (outputInfoBuilder_ == null) {
+          ensureOutputInfoIsMutable();
+          outputInfo_.add(builderForValue.build());
+          onChanged();
+        } else {
+          outputInfoBuilder_.addMessage(builderForValue.build());
+        }
+        return this;
+      }
+      /**
+       * <code>repeated .tensorflow.CostGraphDef.Node.OutputInfo output_info = 5;</code>
+       */
+      public Builder addOutputInfo(
+          int index, org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfo.Builder builderForValue) {
+        if (outputInfoBuilder_ == null) {
+          ensureOutputInfoIsMutable();
+          outputInfo_.add(index, builderForValue.build());
+          onChanged();
+        } else {
+          outputInfoBuilder_.addMessage(index, builderForValue.build());
+        }
+        return this;
+      }
+      /**
+       * <code>repeated .tensorflow.CostGraphDef.Node.OutputInfo output_info = 5;</code>
+       */
+      public Builder addAllOutputInfo(
+          java.lang.Iterable<? extends org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfo> values) {
+        if (outputInfoBuilder_ == null) {
+          ensureOutputInfoIsMutable();
+          com.google.protobuf.AbstractMessageLite.Builder.addAll(
+              values, outputInfo_);
+          onChanged();
+        } else {
+          outputInfoBuilder_.addAllMessages(values);
+        }
+        return this;
+      }
+      /**
+       * <code>repeated .tensorflow.CostGraphDef.Node.OutputInfo output_info = 5;</code>
+       */
+      public Builder clearOutputInfo() {
+        if (outputInfoBuilder_ == null) {
+          outputInfo_ = java.util.Collections.emptyList();
+          bitField0_ = (bitField0_ & ~0x00000002);
+          onChanged();
+        } else {
+          outputInfoBuilder_.clear();
+        }
+        return this;
+      }
+      /**
+       * <code>repeated .tensorflow.CostGraphDef.Node.OutputInfo output_info = 5;</code>
+       */
+      public Builder removeOutputInfo(int index) {
+        if (outputInfoBuilder_ == null) {
+          ensureOutputInfoIsMutable();
+          outputInfo_.remove(index);
+          onChanged();
+        } else {
+          outputInfoBuilder_.remove(index);
+        }
+        return this;
+      }
+      /**
+       * <code>repeated .tensorflow.CostGraphDef.Node.OutputInfo output_info = 5;</code>
+       */
+      public org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfo.Builder getOutputInfoBuilder(
+          int index) {
+        return getOutputInfoFieldBuilder().getBuilder(index);
+      }
+      /**
+       * <code>repeated .tensorflow.CostGraphDef.Node.OutputInfo output_info = 5;</code>
+       */
+      public org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfoOrBuilder getOutputInfoOrBuilder(
+          int index) {
+        if (outputInfoBuilder_ == null) {
+          return outputInfo_.get(index);  } else {
+          return outputInfoBuilder_.getMessageOrBuilder(index);
+        }
+      }
+      /**
+       * <code>repeated .tensorflow.CostGraphDef.Node.OutputInfo output_info = 5;</code>
+       */
+      public java.util.List<? extends org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfoOrBuilder> 
+           getOutputInfoOrBuilderList() {
+        if (outputInfoBuilder_ != null) {
+          return outputInfoBuilder_.getMessageOrBuilderList();
+        } else {
+          return java.util.Collections.unmodifiableList(outputInfo_);
+        }
+      }
+      /**
+       * <code>repeated .tensorflow.CostGraphDef.Node.OutputInfo output_info = 5;</code>
+       */
+      public org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfo.Builder addOutputInfoBuilder() {
+        return getOutputInfoFieldBuilder().addBuilder(
+            org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfo.getDefaultInstance());
+      }
+      /**
+       * <code>repeated .tensorflow.CostGraphDef.Node.OutputInfo output_info = 5;</code>
+       */
+      public org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfo.Builder addOutputInfoBuilder(
+          int index) {
+        return getOutputInfoFieldBuilder().addBuilder(
+            index, org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfo.getDefaultInstance());
+      }
+      /**
+       * <code>repeated .tensorflow.CostGraphDef.Node.OutputInfo output_info = 5;</code>
+       */
+      public java.util.List<org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfo.Builder> 
+           getOutputInfoBuilderList() {
+        return getOutputInfoFieldBuilder().getBuilderList();
+      }
+      private com.google.protobuf.RepeatedFieldBuilderV3<
+          org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfo, org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfo.Builder, org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfoOrBuilder> 
+          getOutputInfoFieldBuilder() {
+        if (outputInfoBuilder_ == null) {
+          outputInfoBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
+              org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfo, org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfo.Builder, org.tensorflow.proto.framework.CostGraphDef.Node.OutputInfoOrBuilder>(
+                  outputInfo_,
+                  ((bitField0_ & 0x00000002) != 0),
+                  getParentForChildren(),
+                  isClean());
+          outputInfo_ = null;
+        }
+        return outputInfoBuilder_;
+      }
+
+      private long temporaryMemorySize_ ;
+      /**
+       * <pre>
+       * Temporary memory used by this node.
+       * </pre>
+       *
+       * <code>int64 temporary_memory_size = 6;</code>
+       */
+      public long getTemporaryMemorySize() {
+        return temporaryMemorySize_;
+      }
+      /**
+       * <pre>
+       * Temporary memory used by this node.
+       * </pre>
+       *
+       * <code>int64 temporary_memory_size = 6;</code>
+       */
+      public Builder setTemporaryMemorySize(long value) {
+        
+        temporaryMemorySize_ = value;
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * Temporary memory used by this node.
+       * </pre>
+       *
+       * <code>int64 temporary_memory_size = 6;</code>
+       */
+      public Builder clearTemporaryMemorySize() {
+        
+        temporaryMemorySize_ = 0L;
+        onChanged();
+        return this;
+      }
+
+      private long persistentMemorySize_ ;
+      /**
+       * <pre>
+       * Persistent memory used by this node.
+       * </pre>
+       *
+       * <code>int64 persistent_memory_size = 12;</code>
+       */
+      public long getPersistentMemorySize() {
+        return persistentMemorySize_;
+      }
+      /**
+       * <pre>
+       * Persistent memory used by this node.
+       * </pre>
+       *
+       * <code>int64 persistent_memory_size = 12;</code>
+       */
+      public Builder setPersistentMemorySize(long value) {
+        
+        persistentMemorySize_ = value;
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * Persistent memory used by this node.
+       * </pre>
+       *
+       * <code>int64 persistent_memory_size = 12;</code>
+       */
+      public Builder clearPersistentMemorySize() {
+        
+        persistentMemorySize_ = 0L;
+        onChanged();
+        return this;
+      }
+
+      private long hostTempMemorySize_ ;
+      /**
+       * <code>int64 host_temp_memory_size = 10 [deprecated = true];</code>
+       */
+      @java.lang.Deprecated public long getHostTempMemorySize() {
+        return hostTempMemorySize_;
+      }
+      /**
+       * <code>int64 host_temp_memory_size = 10 [deprecated = true];</code>
+       */
+      @java.lang.Deprecated public Builder setHostTempMemorySize(long value) {
+        
+        hostTempMemorySize_ = value;
+        onChanged();
+        return this;
+      }
+      /**
+       * <code>int64 host_temp_memory_size = 10 [deprecated = true];</code>
+       */
+      @java.lang.Deprecated public Builder clearHostTempMemorySize() {
+        
+        hostTempMemorySize_ = 0L;
+        onChanged();
+        return this;
+      }
+
+      private long deviceTempMemorySize_ ;
+      /**
+       * <code>int64 device_temp_memory_size = 11 [deprecated = true];</code>
+       */
+      @java.lang.Deprecated public long getDeviceTempMemorySize() {
+        return deviceTempMemorySize_;
+      }
+      /**
+       * <code>int64 device_temp_memory_size = 11 [deprecated = true];</code>
+       */
+      @java.lang.Deprecated public Builder setDeviceTempMemorySize(long value) {
+        
+        deviceTempMemorySize_ = value;
+        onChanged();
+        return this;
+      }
+      /**
+       * <code>int64 device_temp_memory_size = 11 [deprecated = true];</code>
+       */
+      @java.lang.Deprecated public Builder clearDeviceTempMemorySize() {
+        
+        deviceTempMemorySize_ = 0L;
+        onChanged();
+        return this;
+      }
+
+      private long devicePersistentMemorySize_ ;
+      /**
+       * <code>int64 device_persistent_memory_size = 16 [deprecated = true];</code>
+       */
+      @java.lang.Deprecated public long getDevicePersistentMemorySize() {
+        return devicePersistentMemorySize_;
+      }
+      /**
+       * <code>int64 device_persistent_memory_size = 16 [deprecated = true];</code>
+       */
+      @java.lang.Deprecated public Builder setDevicePersistentMemorySize(long value) {
+        
+        devicePersistentMemorySize_ = value;
+        onChanged();
+        return this;
+      }
+      /**
+       * <code>int64 device_persistent_memory_size = 16 [deprecated = true];</code>
+       */
+      @java.lang.Deprecated public Builder clearDevicePersistentMemorySize() {
+        
+        devicePersistentMemorySize_ = 0L;
+        onChanged();
+        return this;
+      }
+
+      private long computeCost_ ;
+      /**
+       * <pre>
+       * Estimate of the computational cost of this node, in microseconds.
+       * </pre>
+       *
+       * <code>int64 compute_cost = 9;</code>
+       */
+      public long getComputeCost() {
+        return computeCost_;
+      }
+      /**
+       * <pre>
+       * Estimate of the computational cost of this node, in microseconds.
+       * </pre>
+       *
+       * <code>int64 compute_cost = 9;</code>
+       */
+      public Builder setComputeCost(long value) {
+        
+        computeCost_ = value;
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * Estimate of the computational cost of this node, in microseconds.
+       * </pre>
+       *
+       * <code>int64 compute_cost = 9;</code>
+       */
+      public Builder clearComputeCost() {
+        
+        computeCost_ = 0L;
+        onChanged();
+        return this;
+      }
+
+      private long computeTime_ ;
+      /**
+       * <pre>
+       * Analytical estimate of the computational cost of this node, in
+       * microseconds.
+       * </pre>
+       *
+       * <code>int64 compute_time = 14;</code>
+       */
+      public long getComputeTime() {
+        return computeTime_;
+      }
+      /**
+       * <pre>
+       * Analytical estimate of the computational cost of this node, in
+       * microseconds.
+       * </pre>
+       *
+       * <code>int64 compute_time = 14;</code>
+       */
+      public Builder setComputeTime(long value) {
+        
+        computeTime_ = value;
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * Analytical estimate of the computational cost of this node, in
+       * microseconds.
+       * </pre>
+       *
+       * <code>int64 compute_time = 14;</code>
+       */
+      public Builder clearComputeTime() {
+        
+        computeTime_ = 0L;
+        onChanged();
+        return this;
+      }
+
+      private long memoryTime_ ;
+      /**
+       * <pre>
+       * Analytical estimate of the memory access cost of this node, in
+       * microseconds.
+       * </pre>
+       *
+       * <code>int64 memory_time = 15;</code>
+       */
+      public long getMemoryTime() {
+        return memoryTime_;
+      }
+      /**
+       * <pre>
+       * Analytical estimate of the memory access cost of this node, in
+       * microseconds.
+       * </pre>
+       *
+       * <code>int64 memory_time = 15;</code>
+       */
+      public Builder setMemoryTime(long value) {
+        
+        memoryTime_ = value;
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * Analytical estimate of the memory access cost of this node, in
+       * microseconds.
+       * </pre>
+       *
+       * <code>int64 memory_time = 15;</code>
+       */
+      public Builder clearMemoryTime() {
+        
+        memoryTime_ = 0L;
+        onChanged();
+        return this;
+      }
+
+      private boolean isFinal_ ;
+      /**
+       * <pre>
+       * If true, the output is permanent: it can't be discarded, because this
+       * node is part of the "final output". Nodes may depend on final nodes.
+       * </pre>
+       *
+       * <code>bool is_final = 7;</code>
+       */
+      public boolean getIsFinal() {
+        return isFinal_;
+      }
+      /**
+       * <pre>
+       * If true, the output is permanent: it can't be discarded, because this
+       * node is part of the "final output". Nodes may depend on final nodes.
+       * </pre>
+       *
+       * <code>bool is_final = 7;</code>
+       */
+      public Builder setIsFinal(boolean value) {
+        
+        isFinal_ = value;
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * If true, the output is permanent: it can't be discarded, because this
+       * node is part of the "final output". Nodes may depend on final nodes.
+       * </pre>
+       *
+       * <code>bool is_final = 7;</code>
+       */
+      public Builder clearIsFinal() {
+        
+        isFinal_ = false;
+        onChanged();
+        return this;
+      }
+
+      private com.google.protobuf.Internal.IntList controlInput_ = emptyIntList();
+      private void ensureControlInputIsMutable() {
+        if (!((bitField0_ & 0x00000004) != 0)) {
+          controlInput_ = mutableCopy(controlInput_);
+          bitField0_ |= 0x00000004;
+         }
+      }
+      /**
+       * <pre>
+       * Ids of the control inputs for this node.
+       * </pre>
+       *
+       * <code>repeated int32 control_input = 8;</code>
+       */
+      public java.util.List<java.lang.Integer>
+          getControlInputList() {
+        return ((bitField0_ & 0x00000004) != 0) ?
+                 java.util.Collections.unmodifiableList(controlInput_) : controlInput_;
+      }
+      /**
+       * <pre>
+       * Ids of the control inputs for this node.
+       * </pre>
+       *
+       * <code>repeated int32 control_input = 8;</code>
+       */
+      public int getControlInputCount() {
+        return controlInput_.size();
+      }
+      /**
+       * <pre>
+       * Ids of the control inputs for this node.
+       * </pre>
+       *
+       * <code>repeated int32 control_input = 8;</code>
+       */
+      public int getControlInput(int index) {
+        return controlInput_.getInt(index);
+      }
+      /**
+       * <pre>
+       * Ids of the control inputs for this node.
+       * </pre>
+       *
+       * <code>repeated int32 control_input = 8;</code>
+       */
+      public Builder setControlInput(
+          int index, int value) {
+        ensureControlInputIsMutable();
+        controlInput_.setInt(index, value);
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * Ids of the control inputs for this node.
+       * </pre>
+       *
+       * <code>repeated int32 control_input = 8;</code>
+       */
+      public Builder addControlInput(int value) {
+        ensureControlInputIsMutable();
+        controlInput_.addInt(value);
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * Ids of the control inputs for this node.
+       * </pre>
+       *
+       * <code>repeated int32 control_input = 8;</code>
+       */
+      public Builder addAllControlInput(
+          java.lang.Iterable<? extends java.lang.Integer> values) {
+        ensureControlInputIsMutable();
+        com.google.protobuf.AbstractMessageLite.Builder.addAll(
+            values, controlInput_);
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * Ids of the control inputs for this node.
+       * </pre>
+       *
+       * <code>repeated int32 control_input = 8;</code>
+       */
+      public Builder clearControlInput() {
+        controlInput_ = emptyIntList();
+        bitField0_ = (bitField0_ & ~0x00000004);
+        onChanged();
+        return this;
+      }
+
+      private boolean inaccurate_ ;
+      /**
+       * <pre>
+       * Are the costs inaccurate?
+       * </pre>
+       *
+       * <code>bool inaccurate = 17;</code>
+       */
+      public boolean getInaccurate() {
+        return inaccurate_;
+      }
+      /**
+       * <pre>
+       * Are the costs inaccurate?
+       * </pre>
+       *
+       * <code>bool inaccurate = 17;</code>
+       */
+      public Builder setInaccurate(boolean value) {
+        
+        inaccurate_ = value;
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * Are the costs inaccurate?
+       * </pre>
+       *
+       * <code>bool inaccurate = 17;</code>
+       */
+      public Builder clearInaccurate() {
+        
+        inaccurate_ = false;
+        onChanged();
+        return this;
+      }
+      @java.lang.Override
+      public final Builder setUnknownFields(
+          final com.google.protobuf.UnknownFieldSet unknownFields) {
+        return super.setUnknownFields(unknownFields);
+      }
+
+      @java.lang.Override
+      public final Builder mergeUnknownFields(
+          final com.google.protobuf.UnknownFieldSet unknownFields) {
+        return super.mergeUnknownFields(unknownFields);
+      }
+
+
+      // @@protoc_insertion_point(builder_scope:tensorflow.CostGraphDef.Node)
+    }
+
+    // @@protoc_insertion_point(class_scope:tensorflow.CostGraphDef.Node)
+    private static final org.tensorflow.proto.framework.CostGraphDef.Node DEFAULT_INSTANCE;
+    static {
+      DEFAULT_INSTANCE = new org.tensorflow.proto.framework.CostGraphDef.Node();
+    }
+
+    public static org.tensorflow.proto.framework.CostGraphDef.Node getDefaultInstance() {
+      return DEFAULT_INSTANCE;
+    }
+
+    private static final com.google.protobuf.Parser<Node>
+        PARSER = new com.google.protobuf.AbstractParser<Node>() {
+      @java.lang.Override
+      public Node parsePartialFrom(
+          com.google.protobuf.CodedInputStream input,
+          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+          throws com.google.protobuf.InvalidProtocolBufferException {
+        return new Node(input, extensionRegistry);
+      }
+    };
+
+    public static com.google.protobuf.Parser<Node> parser() {
+      return PARSER;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Parser<Node> getParserForType() {
+      return PARSER;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.CostGraphDef.Node getDefaultInstanceForType() {
+      return DEFAULT_INSTANCE;
+    }
+
+  }
+
+  public interface AggregatedCostOrBuilder extends
+      // @@protoc_insertion_point(interface_extends:tensorflow.CostGraphDef.AggregatedCost)
+      com.google.protobuf.MessageOrBuilder {
+
+    /**
+     * <pre>
+     * Aggregated cost value.
+     * </pre>
+     *
+     * <code>float cost = 1;</code>
+     */
+    float getCost();
+
+    /**
+     * <pre>
+     * Aggregated cost dimension (e.g. 'memory', 'compute', 'network').
+     * </pre>
+     *
+     * <code>string dimension = 2;</code>
+     */
+    java.lang.String getDimension();
+    /**
+     * <pre>
+     * Aggregated cost dimension (e.g. 'memory', 'compute', 'network').
+     * </pre>
+     *
+     * <code>string dimension = 2;</code>
+     */
+    com.google.protobuf.ByteString
+        getDimensionBytes();
+  }
+  /**
+   * <pre>
+   * Total cost of this graph, typically used for balancing decisions.
+   * </pre>
+   *
+   * Protobuf type {@code tensorflow.CostGraphDef.AggregatedCost}
+   */
+  public  static final class AggregatedCost extends
+      com.google.protobuf.GeneratedMessageV3 implements
+      // @@protoc_insertion_point(message_implements:tensorflow.CostGraphDef.AggregatedCost)
+      AggregatedCostOrBuilder {
+  private static final long serialVersionUID = 0L;
+    // Use AggregatedCost.newBuilder() to construct.
+    private AggregatedCost(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+      super(builder);
+    }
+    private AggregatedCost() {
+      dimension_ = "";
+    }
+
+    @java.lang.Override
+    @SuppressWarnings({"unused"})
+    protected java.lang.Object newInstance(
+        UnusedPrivateParameter unused) {
+      return new AggregatedCost();
+    }
+
+    @java.lang.Override
+    public final com.google.protobuf.UnknownFieldSet
+    getUnknownFields() {
+      return this.unknownFields;
+    }
+    private AggregatedCost(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      this();
+      if (extensionRegistry == null) {
+        throw new java.lang.NullPointerException();
+      }
+      com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+          com.google.protobuf.UnknownFieldSet.newBuilder();
+      try {
+        boolean done = false;
+        while (!done) {
+          int tag = input.readTag();
+          switch (tag) {
+            case 0:
+              done = true;
+              break;
+            case 13: {
+
+              cost_ = input.readFloat();
+              break;
+            }
+            case 18: {
+              java.lang.String s = input.readStringRequireUtf8();
+
+              dimension_ = s;
+              break;
+            }
+            default: {
+              if (!parseUnknownField(
+                  input, unknownFields, extensionRegistry, tag)) {
+                done = true;
+              }
+              break;
+            }
+          }
+        }
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        throw e.setUnfinishedMessage(this);
+      } catch (java.io.IOException e) {
+        throw new com.google.protobuf.InvalidProtocolBufferException(
+            e).setUnfinishedMessage(this);
+      } finally {
+        this.unknownFields = unknownFields.build();
+        makeExtensionsImmutable();
+      }
+    }
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.CostGraphProtos.internal_static_tensorflow_CostGraphDef_AggregatedCost_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.CostGraphProtos.internal_static_tensorflow_CostGraphDef_AggregatedCost_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.CostGraphDef.AggregatedCost.class, org.tensorflow.proto.framework.CostGraphDef.AggregatedCost.Builder.class);
+    }
+
+    public static final int COST_FIELD_NUMBER = 1;
+    private float cost_;
+    /**
+     * <pre>
+     * Aggregated cost value.
+     * </pre>
+     *
+     * <code>float cost = 1;</code>
+     */
+    public float getCost() {
+      return cost_;
+    }
+
+    public static final int DIMENSION_FIELD_NUMBER = 2;
+    private volatile java.lang.Object dimension_;
+    /**
+     * <pre>
+     * Aggregated cost dimension (e.g. 'memory', 'compute', 'network').
+     * </pre>
+     *
+     * <code>string dimension = 2;</code>
+     */
+    public java.lang.String getDimension() {
+      java.lang.Object ref = dimension_;
+      if (ref instanceof java.lang.String) {
+        return (java.lang.String) ref;
+      } else {
+        com.google.protobuf.ByteString bs = 
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        dimension_ = s;
+        return s;
+      }
+    }
+    /**
+     * <pre>
+     * Aggregated cost dimension (e.g. 'memory', 'compute', 'network').
+     * </pre>
+     *
+     * <code>string dimension = 2;</code>
+     */
+    public com.google.protobuf.ByteString
+        getDimensionBytes() {
+      java.lang.Object ref = dimension_;
+      if (ref instanceof java.lang.String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        dimension_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+
+    private byte memoizedIsInitialized = -1;
+    @java.lang.Override
+    public final boolean isInitialized() {
+      byte isInitialized = memoizedIsInitialized;
+      if (isInitialized == 1) return true;
+      if (isInitialized == 0) return false;
+
+      memoizedIsInitialized = 1;
+      return true;
+    }
+
+    @java.lang.Override
+    public void writeTo(com.google.protobuf.CodedOutputStream output)
+                        throws java.io.IOException {
+      if (cost_ != 0F) {
+        output.writeFloat(1, cost_);
+      }
+      if (!getDimensionBytes().isEmpty()) {
+        com.google.protobuf.GeneratedMessageV3.writeString(output, 2, dimension_);
+      }
+      unknownFields.writeTo(output);
+    }
+
+    @java.lang.Override
+    public int getSerializedSize() {
+      int size = memoizedSize;
+      if (size != -1) return size;
+
+      size = 0;
+      if (cost_ != 0F) {
+        size += com.google.protobuf.CodedOutputStream
+          .computeFloatSize(1, cost_);
+      }
+      if (!getDimensionBytes().isEmpty()) {
+        size += com.google.protobuf.GeneratedMessageV3.computeStringSize(2, dimension_);
+      }
+      size += unknownFields.getSerializedSize();
+      memoizedSize = size;
+      return size;
+    }
+
+    @java.lang.Override
+    public boolean equals(final java.lang.Object obj) {
+      if (obj == this) {
+       return true;
+      }
+      if (!(obj instanceof org.tensorflow.proto.framework.CostGraphDef.AggregatedCost)) {
+        return super.equals(obj);
+      }
+      org.tensorflow.proto.framework.CostGraphDef.AggregatedCost other = (org.tensorflow.proto.framework.CostGraphDef.AggregatedCost) obj;
+
+      if (java.lang.Float.floatToIntBits(getCost())
+          != java.lang.Float.floatToIntBits(
+              other.getCost())) return false;
+      if (!getDimension()
+          .equals(other.getDimension())) return false;
+      if (!unknownFields.equals(other.unknownFields)) return false;
+      return true;
+    }
+
+    @java.lang.Override
+    public int hashCode() {
+      if (memoizedHashCode != 0) {
+        return memoizedHashCode;
+      }
+      int hash = 41;
+      hash = (19 * hash) + getDescriptor().hashCode();
+      hash = (37 * hash) + COST_FIELD_NUMBER;
+      hash = (53 * hash) + java.lang.Float.floatToIntBits(
+          getCost());
+      hash = (37 * hash) + DIMENSION_FIELD_NUMBER;
+      hash = (53 * hash) + getDimension().hashCode();
+      hash = (29 * hash) + unknownFields.hashCode();
+      memoizedHashCode = hash;
+      return hash;
+    }
+
+    public static org.tensorflow.proto.framework.CostGraphDef.AggregatedCost parseFrom(
+        java.nio.ByteBuffer data)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data);
+    }
+    public static org.tensorflow.proto.framework.CostGraphDef.AggregatedCost parseFrom(
+        java.nio.ByteBuffer data,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.CostGraphDef.AggregatedCost parseFrom(
+        com.google.protobuf.ByteString data)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data);
+    }
+    public static org.tensorflow.proto.framework.CostGraphDef.AggregatedCost parseFrom(
+        com.google.protobuf.ByteString data,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.CostGraphDef.AggregatedCost parseFrom(byte[] data)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data);
+    }
+    public static org.tensorflow.proto.framework.CostGraphDef.AggregatedCost parseFrom(
+        byte[] data,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.CostGraphDef.AggregatedCost parseFrom(java.io.InputStream input)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input);
+    }
+    public static org.tensorflow.proto.framework.CostGraphDef.AggregatedCost parseFrom(
+        java.io.InputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.CostGraphDef.AggregatedCost parseDelimitedFrom(java.io.InputStream input)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseDelimitedWithIOException(PARSER, input);
+    }
+    public static org.tensorflow.proto.framework.CostGraphDef.AggregatedCost parseDelimitedFrom(
+        java.io.InputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.CostGraphDef.AggregatedCost parseFrom(
+        com.google.protobuf.CodedInputStream input)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input);
+    }
+    public static org.tensorflow.proto.framework.CostGraphDef.AggregatedCost parseFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input, extensionRegistry);
+    }
+
+    @java.lang.Override
+    public Builder newBuilderForType() { return newBuilder(); }
+    public static Builder newBuilder() {
+      return DEFAULT_INSTANCE.toBuilder();
+    }
+    public static Builder newBuilder(org.tensorflow.proto.framework.CostGraphDef.AggregatedCost prototype) {
+      return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+    }
+    @java.lang.Override
+    public Builder toBuilder() {
+      return this == DEFAULT_INSTANCE
+          ? new Builder() : new Builder().mergeFrom(this);
+    }
+
+    @java.lang.Override
+    protected Builder newBuilderForType(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      Builder builder = new Builder(parent);
+      return builder;
+    }
+    /**
+     * <pre>
+     * Total cost of this graph, typically used for balancing decisions.
+     * </pre>
+     *
+     * Protobuf type {@code tensorflow.CostGraphDef.AggregatedCost}
+     */
+    public static final class Builder extends
+        com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+        // @@protoc_insertion_point(builder_implements:tensorflow.CostGraphDef.AggregatedCost)
+        org.tensorflow.proto.framework.CostGraphDef.AggregatedCostOrBuilder {
+      public static final com.google.protobuf.Descriptors.Descriptor
+          getDescriptor() {
+        return org.tensorflow.proto.framework.CostGraphProtos.internal_static_tensorflow_CostGraphDef_AggregatedCost_descriptor;
+      }
+
+      @java.lang.Override
+      protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+          internalGetFieldAccessorTable() {
+        return org.tensorflow.proto.framework.CostGraphProtos.internal_static_tensorflow_CostGraphDef_AggregatedCost_fieldAccessorTable
+            .ensureFieldAccessorsInitialized(
+                org.tensorflow.proto.framework.CostGraphDef.AggregatedCost.class, org.tensorflow.proto.framework.CostGraphDef.AggregatedCost.Builder.class);
+      }
+
+      // Construct using org.tensorflow.proto.framework.CostGraphDef.AggregatedCost.newBuilder()
+      private Builder() {
+        maybeForceBuilderInitialization();
+      }
+
+      private Builder(
+          com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+        super(parent);
+        maybeForceBuilderInitialization();
+      }
+      private void maybeForceBuilderInitialization() {
+        if (com.google.protobuf.GeneratedMessageV3
+                .alwaysUseFieldBuilders) {
+        }
+      }
+      @java.lang.Override
+      public Builder clear() {
+        super.clear();
+        cost_ = 0F;
+
+        dimension_ = "";
+
+        return this;
+      }
+
+      @java.lang.Override
+      public com.google.protobuf.Descriptors.Descriptor
+          getDescriptorForType() {
+        return org.tensorflow.proto.framework.CostGraphProtos.internal_static_tensorflow_CostGraphDef_AggregatedCost_descriptor;
+      }
+
+      @java.lang.Override
+      public org.tensorflow.proto.framework.CostGraphDef.AggregatedCost getDefaultInstanceForType() {
+        return org.tensorflow.proto.framework.CostGraphDef.AggregatedCost.getDefaultInstance();
+      }
+
+      @java.lang.Override
+      public org.tensorflow.proto.framework.CostGraphDef.AggregatedCost build() {
+        org.tensorflow.proto.framework.CostGraphDef.AggregatedCost result = buildPartial();
+        if (!result.isInitialized()) {
+          throw newUninitializedMessageException(result);
+        }
+        return result;
+      }
+
+      @java.lang.Override
+      public org.tensorflow.proto.framework.CostGraphDef.AggregatedCost buildPartial() {
+        org.tensorflow.proto.framework.CostGraphDef.AggregatedCost result = new org.tensorflow.proto.framework.CostGraphDef.AggregatedCost(this);
+        result.cost_ = cost_;
+        result.dimension_ = dimension_;
+        onBuilt();
+        return result;
+      }
+
+      @java.lang.Override
+      public Builder clone() {
+        return super.clone();
+      }
+      @java.lang.Override
+      public Builder setField(
+          com.google.protobuf.Descriptors.FieldDescriptor field,
+          java.lang.Object value) {
+        return super.setField(field, value);
+      }
+      @java.lang.Override
+      public Builder clearField(
+          com.google.protobuf.Descriptors.FieldDescriptor field) {
+        return super.clearField(field);
+      }
+      @java.lang.Override
+      public Builder clearOneof(
+          com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+        return super.clearOneof(oneof);
+      }
+      @java.lang.Override
+      public Builder setRepeatedField(
+          com.google.protobuf.Descriptors.FieldDescriptor field,
+          int index, java.lang.Object value) {
+        return super.setRepeatedField(field, index, value);
+      }
+      @java.lang.Override
+      public Builder addRepeatedField(
+          com.google.protobuf.Descriptors.FieldDescriptor field,
+          java.lang.Object value) {
+        return super.addRepeatedField(field, value);
+      }
+      @java.lang.Override
+      public Builder mergeFrom(com.google.protobuf.Message other) {
+        if (other instanceof org.tensorflow.proto.framework.CostGraphDef.AggregatedCost) {
+          return mergeFrom((org.tensorflow.proto.framework.CostGraphDef.AggregatedCost)other);
+        } else {
+          super.mergeFrom(other);
+          return this;
+        }
+      }
+
+      public Builder mergeFrom(org.tensorflow.proto.framework.CostGraphDef.AggregatedCost other) {
+        if (other == org.tensorflow.proto.framework.CostGraphDef.AggregatedCost.getDefaultInstance()) return this;
+        if (other.getCost() != 0F) {
+          setCost(other.getCost());
+        }
+        if (!other.getDimension().isEmpty()) {
+          dimension_ = other.dimension_;
+          onChanged();
+        }
+        this.mergeUnknownFields(other.unknownFields);
+        onChanged();
+        return this;
+      }
+
+      @java.lang.Override
+      public final boolean isInitialized() {
+        return true;
+      }
+
+      @java.lang.Override
+      public Builder mergeFrom(
+          com.google.protobuf.CodedInputStream input,
+          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+          throws java.io.IOException {
+        org.tensorflow.proto.framework.CostGraphDef.AggregatedCost parsedMessage = null;
+        try {
+          parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+        } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+          parsedMessage = (org.tensorflow.proto.framework.CostGraphDef.AggregatedCost) e.getUnfinishedMessage();
+          throw e.unwrapIOException();
+        } finally {
+          if (parsedMessage != null) {
+            mergeFrom(parsedMessage);
+          }
+        }
+        return this;
+      }
+
+      private float cost_ ;
+      /**
+       * <pre>
+       * Aggregated cost value.
+       * </pre>
+       *
+       * <code>float cost = 1;</code>
+       */
+      public float getCost() {
+        return cost_;
+      }
+      /**
+       * <pre>
+       * Aggregated cost value.
+       * </pre>
+       *
+       * <code>float cost = 1;</code>
+       */
+      public Builder setCost(float value) {
+        
+        cost_ = value;
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * Aggregated cost value.
+       * </pre>
+       *
+       * <code>float cost = 1;</code>
+       */
+      public Builder clearCost() {
+        
+        cost_ = 0F;
+        onChanged();
+        return this;
+      }
+
+      private java.lang.Object dimension_ = "";
+      /**
+       * <pre>
+       * Aggregated cost dimension (e.g. 'memory', 'compute', 'network').
+       * </pre>
+       *
+       * <code>string dimension = 2;</code>
+       */
+      public java.lang.String getDimension() {
+        java.lang.Object ref = dimension_;
+        if (!(ref instanceof java.lang.String)) {
+          com.google.protobuf.ByteString bs =
+              (com.google.protobuf.ByteString) ref;
+          java.lang.String s = bs.toStringUtf8();
+          dimension_ = s;
+          return s;
+        } else {
+          return (java.lang.String) ref;
+        }
+      }
+      /**
+       * <pre>
+       * Aggregated cost dimension (e.g. 'memory', 'compute', 'network').
+       * </pre>
+       *
+       * <code>string dimension = 2;</code>
+       */
+      public com.google.protobuf.ByteString
+          getDimensionBytes() {
+        java.lang.Object ref = dimension_;
+        if (ref instanceof String) {
+          com.google.protobuf.ByteString b = 
+              com.google.protobuf.ByteString.copyFromUtf8(
+                  (java.lang.String) ref);
+          dimension_ = b;
+          return b;
+        } else {
+          return (com.google.protobuf.ByteString) ref;
+        }
+      }
+      /**
+       * <pre>
+       * Aggregated cost dimension (e.g. 'memory', 'compute', 'network').
+       * </pre>
+       *
+       * <code>string dimension = 2;</code>
+       */
+      public Builder setDimension(
+          java.lang.String value) {
+        if (value == null) {
+    throw new NullPointerException();
+  }
+  
+        dimension_ = value;
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * Aggregated cost dimension (e.g. 'memory', 'compute', 'network').
+       * </pre>
+       *
+       * <code>string dimension = 2;</code>
+       */
+      public Builder clearDimension() {
+        
+        dimension_ = getDefaultInstance().getDimension();
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * Aggregated cost dimension (e.g. 'memory', 'compute', 'network').
+       * </pre>
+       *
+       * <code>string dimension = 2;</code>
+       */
+      public Builder setDimensionBytes(
+          com.google.protobuf.ByteString value) {
+        if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+        
+        dimension_ = value;
+        onChanged();
+        return this;
+      }
+      @java.lang.Override
+      public final Builder setUnknownFields(
+          final com.google.protobuf.UnknownFieldSet unknownFields) {
+        return super.setUnknownFields(unknownFields);
+      }
+
+      @java.lang.Override
+      public final Builder mergeUnknownFields(
+          final com.google.protobuf.UnknownFieldSet unknownFields) {
+        return super.mergeUnknownFields(unknownFields);
+      }
+
+
+      // @@protoc_insertion_point(builder_scope:tensorflow.CostGraphDef.AggregatedCost)
+    }
+
+    // @@protoc_insertion_point(class_scope:tensorflow.CostGraphDef.AggregatedCost)
+    private static final org.tensorflow.proto.framework.CostGraphDef.AggregatedCost DEFAULT_INSTANCE;
+    static {
+      DEFAULT_INSTANCE = new org.tensorflow.proto.framework.CostGraphDef.AggregatedCost();
+    }
+
+    public static org.tensorflow.proto.framework.CostGraphDef.AggregatedCost getDefaultInstance() {
+      return DEFAULT_INSTANCE;
+    }
+
+    private static final com.google.protobuf.Parser<AggregatedCost>
+        PARSER = new com.google.protobuf.AbstractParser<AggregatedCost>() {
+      @java.lang.Override
+      public AggregatedCost parsePartialFrom(
+          com.google.protobuf.CodedInputStream input,
+          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+          throws com.google.protobuf.InvalidProtocolBufferException {
+        return new AggregatedCost(input, extensionRegistry);
+      }
+    };
+
+    public static com.google.protobuf.Parser<AggregatedCost> parser() {
+      return PARSER;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Parser<AggregatedCost> getParserForType() {
+      return PARSER;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.CostGraphDef.AggregatedCost getDefaultInstanceForType() {
+      return DEFAULT_INSTANCE;
+    }
+
+  }
+
+  public static final int NODE_FIELD_NUMBER = 1;
+  private java.util.List<org.tensorflow.proto.framework.CostGraphDef.Node> node_;
+  /**
+   * <code>repeated .tensorflow.CostGraphDef.Node node = 1;</code>
+   */
+  public java.util.List<org.tensorflow.proto.framework.CostGraphDef.Node> getNodeList() {
+    return node_;
+  }
+  /**
+   * <code>repeated .tensorflow.CostGraphDef.Node node = 1;</code>
+   */
+  public java.util.List<? extends org.tensorflow.proto.framework.CostGraphDef.NodeOrBuilder> 
+      getNodeOrBuilderList() {
+    return node_;
+  }
+  /**
+   * <code>repeated .tensorflow.CostGraphDef.Node node = 1;</code>
+   */
+  public int getNodeCount() {
+    return node_.size();
+  }
+  /**
+   * <code>repeated .tensorflow.CostGraphDef.Node node = 1;</code>
+   */
+  public org.tensorflow.proto.framework.CostGraphDef.Node getNode(int index) {
+    return node_.get(index);
+  }
+  /**
+   * <code>repeated .tensorflow.CostGraphDef.Node node = 1;</code>
+   */
+  public org.tensorflow.proto.framework.CostGraphDef.NodeOrBuilder getNodeOrBuilder(
+      int index) {
+    return node_.get(index);
+  }
+
+  public static final int COST_FIELD_NUMBER = 2;
+  private java.util.List<org.tensorflow.proto.framework.CostGraphDef.AggregatedCost> cost_;
+  /**
+   * <code>repeated .tensorflow.CostGraphDef.AggregatedCost cost = 2;</code>
+   */
+  public java.util.List<org.tensorflow.proto.framework.CostGraphDef.AggregatedCost> getCostList() {
+    return cost_;
+  }
+  /**
+   * <code>repeated .tensorflow.CostGraphDef.AggregatedCost cost = 2;</code>
+   */
+  public java.util.List<? extends org.tensorflow.proto.framework.CostGraphDef.AggregatedCostOrBuilder> 
+      getCostOrBuilderList() {
+    return cost_;
+  }
+  /**
+   * <code>repeated .tensorflow.CostGraphDef.AggregatedCost cost = 2;</code>
+   */
+  public int getCostCount() {
+    return cost_.size();
+  }
+  /**
+   * <code>repeated .tensorflow.CostGraphDef.AggregatedCost cost = 2;</code>
+   */
+  public org.tensorflow.proto.framework.CostGraphDef.AggregatedCost getCost(int index) {
+    return cost_.get(index);
+  }
+  /**
+   * <code>repeated .tensorflow.CostGraphDef.AggregatedCost cost = 2;</code>
+   */
+  public org.tensorflow.proto.framework.CostGraphDef.AggregatedCostOrBuilder getCostOrBuilder(
+      int index) {
+    return cost_.get(index);
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    for (int i = 0; i < node_.size(); i++) {
+      output.writeMessage(1, node_.get(i));
+    }
+    for (int i = 0; i < cost_.size(); i++) {
+      output.writeMessage(2, cost_.get(i));
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    for (int i = 0; i < node_.size(); i++) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(1, node_.get(i));
+    }
+    for (int i = 0; i < cost_.size(); i++) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(2, cost_.get(i));
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.framework.CostGraphDef)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.framework.CostGraphDef other = (org.tensorflow.proto.framework.CostGraphDef) obj;
+
+    if (!getNodeList()
+        .equals(other.getNodeList())) return false;
+    if (!getCostList()
+        .equals(other.getCostList())) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    if (getNodeCount() > 0) {
+      hash = (37 * hash) + NODE_FIELD_NUMBER;
+      hash = (53 * hash) + getNodeList().hashCode();
+    }
+    if (getCostCount() > 0) {
+      hash = (37 * hash) + COST_FIELD_NUMBER;
+      hash = (53 * hash) + getCostList().hashCode();
+    }
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.framework.CostGraphDef parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.CostGraphDef parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.CostGraphDef parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.CostGraphDef parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.CostGraphDef parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.CostGraphDef parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.CostGraphDef parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.CostGraphDef parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.CostGraphDef parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.CostGraphDef parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.CostGraphDef parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.CostGraphDef parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.framework.CostGraphDef prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * Protobuf type {@code tensorflow.CostGraphDef}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.CostGraphDef)
+      org.tensorflow.proto.framework.CostGraphDefOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.CostGraphProtos.internal_static_tensorflow_CostGraphDef_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.CostGraphProtos.internal_static_tensorflow_CostGraphDef_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.CostGraphDef.class, org.tensorflow.proto.framework.CostGraphDef.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.framework.CostGraphDef.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+        getNodeFieldBuilder();
+        getCostFieldBuilder();
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      if (nodeBuilder_ == null) {
+        node_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000001);
+      } else {
+        nodeBuilder_.clear();
+      }
+      if (costBuilder_ == null) {
+        cost_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000002);
+      } else {
+        costBuilder_.clear();
+      }
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.framework.CostGraphProtos.internal_static_tensorflow_CostGraphDef_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.CostGraphDef getDefaultInstanceForType() {
+      return org.tensorflow.proto.framework.CostGraphDef.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.CostGraphDef build() {
+      org.tensorflow.proto.framework.CostGraphDef result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.CostGraphDef buildPartial() {
+      org.tensorflow.proto.framework.CostGraphDef result = new org.tensorflow.proto.framework.CostGraphDef(this);
+      int from_bitField0_ = bitField0_;
+      if (nodeBuilder_ == null) {
+        if (((bitField0_ & 0x00000001) != 0)) {
+          node_ = java.util.Collections.unmodifiableList(node_);
+          bitField0_ = (bitField0_ & ~0x00000001);
+        }
+        result.node_ = node_;
+      } else {
+        result.node_ = nodeBuilder_.build();
+      }
+      if (costBuilder_ == null) {
+        if (((bitField0_ & 0x00000002) != 0)) {
+          cost_ = java.util.Collections.unmodifiableList(cost_);
+          bitField0_ = (bitField0_ & ~0x00000002);
+        }
+        result.cost_ = cost_;
+      } else {
+        result.cost_ = costBuilder_.build();
+      }
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.framework.CostGraphDef) {
+        return mergeFrom((org.tensorflow.proto.framework.CostGraphDef)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.framework.CostGraphDef other) {
+      if (other == org.tensorflow.proto.framework.CostGraphDef.getDefaultInstance()) return this;
+      if (nodeBuilder_ == null) {
+        if (!other.node_.isEmpty()) {
+          if (node_.isEmpty()) {
+            node_ = other.node_;
+            bitField0_ = (bitField0_ & ~0x00000001);
+          } else {
+            ensureNodeIsMutable();
+            node_.addAll(other.node_);
+          }
+          onChanged();
+        }
+      } else {
+        if (!other.node_.isEmpty()) {
+          if (nodeBuilder_.isEmpty()) {
+            nodeBuilder_.dispose();
+            nodeBuilder_ = null;
+            node_ = other.node_;
+            bitField0_ = (bitField0_ & ~0x00000001);
+            nodeBuilder_ = 
+              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
+                 getNodeFieldBuilder() : null;
+          } else {
+            nodeBuilder_.addAllMessages(other.node_);
+          }
+        }
+      }
+      if (costBuilder_ == null) {
+        if (!other.cost_.isEmpty()) {
+          if (cost_.isEmpty()) {
+            cost_ = other.cost_;
+            bitField0_ = (bitField0_ & ~0x00000002);
+          } else {
+            ensureCostIsMutable();
+            cost_.addAll(other.cost_);
+          }
+          onChanged();
+        }
+      } else {
+        if (!other.cost_.isEmpty()) {
+          if (costBuilder_.isEmpty()) {
+            costBuilder_.dispose();
+            costBuilder_ = null;
+            cost_ = other.cost_;
+            bitField0_ = (bitField0_ & ~0x00000002);
+            costBuilder_ = 
+              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
+                 getCostFieldBuilder() : null;
+          } else {
+            costBuilder_.addAllMessages(other.cost_);
+          }
+        }
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.framework.CostGraphDef parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.framework.CostGraphDef) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+    private int bitField0_;
+
+    private java.util.List<org.tensorflow.proto.framework.CostGraphDef.Node> node_ =
+      java.util.Collections.emptyList();
+    private void ensureNodeIsMutable() {
+      if (!((bitField0_ & 0x00000001) != 0)) {
+        node_ = new java.util.ArrayList<org.tensorflow.proto.framework.CostGraphDef.Node>(node_);
+        bitField0_ |= 0x00000001;
+       }
+    }
+
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.framework.CostGraphDef.Node, org.tensorflow.proto.framework.CostGraphDef.Node.Builder, org.tensorflow.proto.framework.CostGraphDef.NodeOrBuilder> nodeBuilder_;
+
+    /**
+     * <code>repeated .tensorflow.CostGraphDef.Node node = 1;</code>
+     */
+    public java.util.List<org.tensorflow.proto.framework.CostGraphDef.Node> getNodeList() {
+      if (nodeBuilder_ == null) {
+        return java.util.Collections.unmodifiableList(node_);
+      } else {
+        return nodeBuilder_.getMessageList();
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.CostGraphDef.Node node = 1;</code>
+     */
+    public int getNodeCount() {
+      if (nodeBuilder_ == null) {
+        return node_.size();
+      } else {
+        return nodeBuilder_.getCount();
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.CostGraphDef.Node node = 1;</code>
+     */
+    public org.tensorflow.proto.framework.CostGraphDef.Node getNode(int index) {
+      if (nodeBuilder_ == null) {
+        return node_.get(index);
+      } else {
+        return nodeBuilder_.getMessage(index);
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.CostGraphDef.Node node = 1;</code>
+     */
+    public Builder setNode(
+        int index, org.tensorflow.proto.framework.CostGraphDef.Node value) {
+      if (nodeBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureNodeIsMutable();
+        node_.set(index, value);
+        onChanged();
+      } else {
+        nodeBuilder_.setMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.CostGraphDef.Node node = 1;</code>
+     */
+    public Builder setNode(
+        int index, org.tensorflow.proto.framework.CostGraphDef.Node.Builder builderForValue) {
+      if (nodeBuilder_ == null) {
+        ensureNodeIsMutable();
+        node_.set(index, builderForValue.build());
+        onChanged();
+      } else {
+        nodeBuilder_.setMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.CostGraphDef.Node node = 1;</code>
+     */
+    public Builder addNode(org.tensorflow.proto.framework.CostGraphDef.Node value) {
+      if (nodeBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureNodeIsMutable();
+        node_.add(value);
+        onChanged();
+      } else {
+        nodeBuilder_.addMessage(value);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.CostGraphDef.Node node = 1;</code>
+     */
+    public Builder addNode(
+        int index, org.tensorflow.proto.framework.CostGraphDef.Node value) {
+      if (nodeBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureNodeIsMutable();
+        node_.add(index, value);
+        onChanged();
+      } else {
+        nodeBuilder_.addMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.CostGraphDef.Node node = 1;</code>
+     */
+    public Builder addNode(
+        org.tensorflow.proto.framework.CostGraphDef.Node.Builder builderForValue) {
+      if (nodeBuilder_ == null) {
+        ensureNodeIsMutable();
+        node_.add(builderForValue.build());
+        onChanged();
+      } else {
+        nodeBuilder_.addMessage(builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.CostGraphDef.Node node = 1;</code>
+     */
+    public Builder addNode(
+        int index, org.tensorflow.proto.framework.CostGraphDef.Node.Builder builderForValue) {
+      if (nodeBuilder_ == null) {
+        ensureNodeIsMutable();
+        node_.add(index, builderForValue.build());
+        onChanged();
+      } else {
+        nodeBuilder_.addMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.CostGraphDef.Node node = 1;</code>
+     */
+    public Builder addAllNode(
+        java.lang.Iterable<? extends org.tensorflow.proto.framework.CostGraphDef.Node> values) {
+      if (nodeBuilder_ == null) {
+        ensureNodeIsMutable();
+        com.google.protobuf.AbstractMessageLite.Builder.addAll(
+            values, node_);
+        onChanged();
+      } else {
+        nodeBuilder_.addAllMessages(values);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.CostGraphDef.Node node = 1;</code>
+     */
+    public Builder clearNode() {
+      if (nodeBuilder_ == null) {
+        node_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000001);
+        onChanged();
+      } else {
+        nodeBuilder_.clear();
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.CostGraphDef.Node node = 1;</code>
+     */
+    public Builder removeNode(int index) {
+      if (nodeBuilder_ == null) {
+        ensureNodeIsMutable();
+        node_.remove(index);
+        onChanged();
+      } else {
+        nodeBuilder_.remove(index);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.CostGraphDef.Node node = 1;</code>
+     */
+    public org.tensorflow.proto.framework.CostGraphDef.Node.Builder getNodeBuilder(
+        int index) {
+      return getNodeFieldBuilder().getBuilder(index);
+    }
+    /**
+     * <code>repeated .tensorflow.CostGraphDef.Node node = 1;</code>
+     */
+    public org.tensorflow.proto.framework.CostGraphDef.NodeOrBuilder getNodeOrBuilder(
+        int index) {
+      if (nodeBuilder_ == null) {
+        return node_.get(index);  } else {
+        return nodeBuilder_.getMessageOrBuilder(index);
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.CostGraphDef.Node node = 1;</code>
+     */
+    public java.util.List<? extends org.tensorflow.proto.framework.CostGraphDef.NodeOrBuilder> 
+         getNodeOrBuilderList() {
+      if (nodeBuilder_ != null) {
+        return nodeBuilder_.getMessageOrBuilderList();
+      } else {
+        return java.util.Collections.unmodifiableList(node_);
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.CostGraphDef.Node node = 1;</code>
+     */
+    public org.tensorflow.proto.framework.CostGraphDef.Node.Builder addNodeBuilder() {
+      return getNodeFieldBuilder().addBuilder(
+          org.tensorflow.proto.framework.CostGraphDef.Node.getDefaultInstance());
+    }
+    /**
+     * <code>repeated .tensorflow.CostGraphDef.Node node = 1;</code>
+     */
+    public org.tensorflow.proto.framework.CostGraphDef.Node.Builder addNodeBuilder(
+        int index) {
+      return getNodeFieldBuilder().addBuilder(
+          index, org.tensorflow.proto.framework.CostGraphDef.Node.getDefaultInstance());
+    }
+    /**
+     * <code>repeated .tensorflow.CostGraphDef.Node node = 1;</code>
+     */
+    public java.util.List<org.tensorflow.proto.framework.CostGraphDef.Node.Builder> 
+         getNodeBuilderList() {
+      return getNodeFieldBuilder().getBuilderList();
+    }
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.framework.CostGraphDef.Node, org.tensorflow.proto.framework.CostGraphDef.Node.Builder, org.tensorflow.proto.framework.CostGraphDef.NodeOrBuilder> 
+        getNodeFieldBuilder() {
+      if (nodeBuilder_ == null) {
+        nodeBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
+            org.tensorflow.proto.framework.CostGraphDef.Node, org.tensorflow.proto.framework.CostGraphDef.Node.Builder, org.tensorflow.proto.framework.CostGraphDef.NodeOrBuilder>(
+                node_,
+                ((bitField0_ & 0x00000001) != 0),
+                getParentForChildren(),
+                isClean());
+        node_ = null;
+      }
+      return nodeBuilder_;
+    }
+
+    private java.util.List<org.tensorflow.proto.framework.CostGraphDef.AggregatedCost> cost_ =
+      java.util.Collections.emptyList();
+    private void ensureCostIsMutable() {
+      if (!((bitField0_ & 0x00000002) != 0)) {
+        cost_ = new java.util.ArrayList<org.tensorflow.proto.framework.CostGraphDef.AggregatedCost>(cost_);
+        bitField0_ |= 0x00000002;
+       }
+    }
+
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.framework.CostGraphDef.AggregatedCost, org.tensorflow.proto.framework.CostGraphDef.AggregatedCost.Builder, org.tensorflow.proto.framework.CostGraphDef.AggregatedCostOrBuilder> costBuilder_;
+
+    /**
+     * <code>repeated .tensorflow.CostGraphDef.AggregatedCost cost = 2;</code>
+     */
+    public java.util.List<org.tensorflow.proto.framework.CostGraphDef.AggregatedCost> getCostList() {
+      if (costBuilder_ == null) {
+        return java.util.Collections.unmodifiableList(cost_);
+      } else {
+        return costBuilder_.getMessageList();
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.CostGraphDef.AggregatedCost cost = 2;</code>
+     */
+    public int getCostCount() {
+      if (costBuilder_ == null) {
+        return cost_.size();
+      } else {
+        return costBuilder_.getCount();
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.CostGraphDef.AggregatedCost cost = 2;</code>
+     */
+    public org.tensorflow.proto.framework.CostGraphDef.AggregatedCost getCost(int index) {
+      if (costBuilder_ == null) {
+        return cost_.get(index);
+      } else {
+        return costBuilder_.getMessage(index);
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.CostGraphDef.AggregatedCost cost = 2;</code>
+     */
+    public Builder setCost(
+        int index, org.tensorflow.proto.framework.CostGraphDef.AggregatedCost value) {
+      if (costBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureCostIsMutable();
+        cost_.set(index, value);
+        onChanged();
+      } else {
+        costBuilder_.setMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.CostGraphDef.AggregatedCost cost = 2;</code>
+     */
+    public Builder setCost(
+        int index, org.tensorflow.proto.framework.CostGraphDef.AggregatedCost.Builder builderForValue) {
+      if (costBuilder_ == null) {
+        ensureCostIsMutable();
+        cost_.set(index, builderForValue.build());
+        onChanged();
+      } else {
+        costBuilder_.setMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.CostGraphDef.AggregatedCost cost = 2;</code>
+     */
+    public Builder addCost(org.tensorflow.proto.framework.CostGraphDef.AggregatedCost value) {
+      if (costBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureCostIsMutable();
+        cost_.add(value);
+        onChanged();
+      } else {
+        costBuilder_.addMessage(value);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.CostGraphDef.AggregatedCost cost = 2;</code>
+     */
+    public Builder addCost(
+        int index, org.tensorflow.proto.framework.CostGraphDef.AggregatedCost value) {
+      if (costBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureCostIsMutable();
+        cost_.add(index, value);
+        onChanged();
+      } else {
+        costBuilder_.addMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.CostGraphDef.AggregatedCost cost = 2;</code>
+     */
+    public Builder addCost(
+        org.tensorflow.proto.framework.CostGraphDef.AggregatedCost.Builder builderForValue) {
+      if (costBuilder_ == null) {
+        ensureCostIsMutable();
+        cost_.add(builderForValue.build());
+        onChanged();
+      } else {
+        costBuilder_.addMessage(builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.CostGraphDef.AggregatedCost cost = 2;</code>
+     */
+    public Builder addCost(
+        int index, org.tensorflow.proto.framework.CostGraphDef.AggregatedCost.Builder builderForValue) {
+      if (costBuilder_ == null) {
+        ensureCostIsMutable();
+        cost_.add(index, builderForValue.build());
+        onChanged();
+      } else {
+        costBuilder_.addMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.CostGraphDef.AggregatedCost cost = 2;</code>
+     */
+    public Builder addAllCost(
+        java.lang.Iterable<? extends org.tensorflow.proto.framework.CostGraphDef.AggregatedCost> values) {
+      if (costBuilder_ == null) {
+        ensureCostIsMutable();
+        com.google.protobuf.AbstractMessageLite.Builder.addAll(
+            values, cost_);
+        onChanged();
+      } else {
+        costBuilder_.addAllMessages(values);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.CostGraphDef.AggregatedCost cost = 2;</code>
+     */
+    public Builder clearCost() {
+      if (costBuilder_ == null) {
+        cost_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000002);
+        onChanged();
+      } else {
+        costBuilder_.clear();
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.CostGraphDef.AggregatedCost cost = 2;</code>
+     */
+    public Builder removeCost(int index) {
+      if (costBuilder_ == null) {
+        ensureCostIsMutable();
+        cost_.remove(index);
+        onChanged();
+      } else {
+        costBuilder_.remove(index);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.CostGraphDef.AggregatedCost cost = 2;</code>
+     */
+    public org.tensorflow.proto.framework.CostGraphDef.AggregatedCost.Builder getCostBuilder(
+        int index) {
+      return getCostFieldBuilder().getBuilder(index);
+    }
+    /**
+     * <code>repeated .tensorflow.CostGraphDef.AggregatedCost cost = 2;</code>
+     */
+    public org.tensorflow.proto.framework.CostGraphDef.AggregatedCostOrBuilder getCostOrBuilder(
+        int index) {
+      if (costBuilder_ == null) {
+        return cost_.get(index);  } else {
+        return costBuilder_.getMessageOrBuilder(index);
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.CostGraphDef.AggregatedCost cost = 2;</code>
+     */
+    public java.util.List<? extends org.tensorflow.proto.framework.CostGraphDef.AggregatedCostOrBuilder> 
+         getCostOrBuilderList() {
+      if (costBuilder_ != null) {
+        return costBuilder_.getMessageOrBuilderList();
+      } else {
+        return java.util.Collections.unmodifiableList(cost_);
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.CostGraphDef.AggregatedCost cost = 2;</code>
+     */
+    public org.tensorflow.proto.framework.CostGraphDef.AggregatedCost.Builder addCostBuilder() {
+      return getCostFieldBuilder().addBuilder(
+          org.tensorflow.proto.framework.CostGraphDef.AggregatedCost.getDefaultInstance());
+    }
+    /**
+     * <code>repeated .tensorflow.CostGraphDef.AggregatedCost cost = 2;</code>
+     */
+    public org.tensorflow.proto.framework.CostGraphDef.AggregatedCost.Builder addCostBuilder(
+        int index) {
+      return getCostFieldBuilder().addBuilder(
+          index, org.tensorflow.proto.framework.CostGraphDef.AggregatedCost.getDefaultInstance());
+    }
+    /**
+     * <code>repeated .tensorflow.CostGraphDef.AggregatedCost cost = 2;</code>
+     */
+    public java.util.List<org.tensorflow.proto.framework.CostGraphDef.AggregatedCost.Builder> 
+         getCostBuilderList() {
+      return getCostFieldBuilder().getBuilderList();
+    }
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.framework.CostGraphDef.AggregatedCost, org.tensorflow.proto.framework.CostGraphDef.AggregatedCost.Builder, org.tensorflow.proto.framework.CostGraphDef.AggregatedCostOrBuilder> 
+        getCostFieldBuilder() {
+      if (costBuilder_ == null) {
+        costBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
+            org.tensorflow.proto.framework.CostGraphDef.AggregatedCost, org.tensorflow.proto.framework.CostGraphDef.AggregatedCost.Builder, org.tensorflow.proto.framework.CostGraphDef.AggregatedCostOrBuilder>(
+                cost_,
+                ((bitField0_ & 0x00000002) != 0),
+                getParentForChildren(),
+                isClean());
+        cost_ = null;
+      }
+      return costBuilder_;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.CostGraphDef)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.CostGraphDef)
+  private static final org.tensorflow.proto.framework.CostGraphDef DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.CostGraphDef();
+  }
+
+  public static org.tensorflow.proto.framework.CostGraphDef getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<CostGraphDef>
+      PARSER = new com.google.protobuf.AbstractParser<CostGraphDef>() {
+    @java.lang.Override
+    public CostGraphDef parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new CostGraphDef(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<CostGraphDef> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<CostGraphDef> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.framework.CostGraphDef getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/CostGraphDefOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/CostGraphDefOrBuilder.java
new file mode 100644
index 00000000000..7d8046a5e27
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/CostGraphDefOrBuilder.java
@@ -0,0 +1,57 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/framework/cost_graph.proto
+
+package org.tensorflow.proto.framework;
+
+public interface CostGraphDefOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.CostGraphDef)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <code>repeated .tensorflow.CostGraphDef.Node node = 1;</code>
+   */
+  java.util.List<org.tensorflow.proto.framework.CostGraphDef.Node> 
+      getNodeList();
+  /**
+   * <code>repeated .tensorflow.CostGraphDef.Node node = 1;</code>
+   */
+  org.tensorflow.proto.framework.CostGraphDef.Node getNode(int index);
+  /**
+   * <code>repeated .tensorflow.CostGraphDef.Node node = 1;</code>
+   */
+  int getNodeCount();
+  /**
+   * <code>repeated .tensorflow.CostGraphDef.Node node = 1;</code>
+   */
+  java.util.List<? extends org.tensorflow.proto.framework.CostGraphDef.NodeOrBuilder> 
+      getNodeOrBuilderList();
+  /**
+   * <code>repeated .tensorflow.CostGraphDef.Node node = 1;</code>
+   */
+  org.tensorflow.proto.framework.CostGraphDef.NodeOrBuilder getNodeOrBuilder(
+      int index);
+
+  /**
+   * <code>repeated .tensorflow.CostGraphDef.AggregatedCost cost = 2;</code>
+   */
+  java.util.List<org.tensorflow.proto.framework.CostGraphDef.AggregatedCost> 
+      getCostList();
+  /**
+   * <code>repeated .tensorflow.CostGraphDef.AggregatedCost cost = 2;</code>
+   */
+  org.tensorflow.proto.framework.CostGraphDef.AggregatedCost getCost(int index);
+  /**
+   * <code>repeated .tensorflow.CostGraphDef.AggregatedCost cost = 2;</code>
+   */
+  int getCostCount();
+  /**
+   * <code>repeated .tensorflow.CostGraphDef.AggregatedCost cost = 2;</code>
+   */
+  java.util.List<? extends org.tensorflow.proto.framework.CostGraphDef.AggregatedCostOrBuilder> 
+      getCostOrBuilderList();
+  /**
+   * <code>repeated .tensorflow.CostGraphDef.AggregatedCost cost = 2;</code>
+   */
+  org.tensorflow.proto.framework.CostGraphDef.AggregatedCostOrBuilder getCostOrBuilder(
+      int index);
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/CostGraphProtos.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/CostGraphProtos.java
new file mode 100644
index 00000000000..b115a24c302
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/CostGraphProtos.java
@@ -0,0 +1,122 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/framework/cost_graph.proto
+
+package org.tensorflow.proto.framework;
+
+public final class CostGraphProtos {
+  private CostGraphProtos() {}
+  public static void registerAllExtensions(
+      com.google.protobuf.ExtensionRegistryLite registry) {
+  }
+
+  public static void registerAllExtensions(
+      com.google.protobuf.ExtensionRegistry registry) {
+    registerAllExtensions(
+        (com.google.protobuf.ExtensionRegistryLite) registry);
+  }
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_CostGraphDef_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_CostGraphDef_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_CostGraphDef_Node_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_CostGraphDef_Node_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_CostGraphDef_Node_InputInfo_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_CostGraphDef_Node_InputInfo_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_CostGraphDef_Node_OutputInfo_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_CostGraphDef_Node_OutputInfo_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_CostGraphDef_AggregatedCost_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_CostGraphDef_AggregatedCost_fieldAccessorTable;
+
+  public static com.google.protobuf.Descriptors.FileDescriptor
+      getDescriptor() {
+    return descriptor;
+  }
+  private static  com.google.protobuf.Descriptors.FileDescriptor
+      descriptor;
+  static {
+    java.lang.String[] descriptorData = {
+      "\n*tensorflow/core/framework/cost_graph.p" +
+      "roto\022\ntensorflow\032,tensorflow/core/framew" +
+      "ork/tensor_shape.proto\032%tensorflow/core/" +
+      "framework/types.proto\"\312\006\n\014CostGraphDef\022+" +
+      "\n\004node\030\001 \003(\0132\035.tensorflow.CostGraphDef.N" +
+      "ode\0225\n\004cost\030\002 \003(\0132\'.tensorflow.CostGraph" +
+      "Def.AggregatedCost\032\242\005\n\004Node\022\014\n\004name\030\001 \001(" +
+      "\t\022\016\n\006device\030\002 \001(\t\022\n\n\002id\030\003 \001(\005\022;\n\ninput_i" +
+      "nfo\030\004 \003(\0132\'.tensorflow.CostGraphDef.Node" +
+      ".InputInfo\022=\n\013output_info\030\005 \003(\0132(.tensor" +
+      "flow.CostGraphDef.Node.OutputInfo\022\035\n\025tem" +
+      "porary_memory_size\030\006 \001(\003\022\036\n\026persistent_m" +
+      "emory_size\030\014 \001(\003\022!\n\025host_temp_memory_siz" +
+      "e\030\n \001(\003B\002\030\001\022#\n\027device_temp_memory_size\030\013" +
+      " \001(\003B\002\030\001\022)\n\035device_persistent_memory_siz" +
+      "e\030\020 \001(\003B\002\030\001\022\024\n\014compute_cost\030\t \001(\003\022\024\n\014com" +
+      "pute_time\030\016 \001(\003\022\023\n\013memory_time\030\017 \001(\003\022\020\n\010" +
+      "is_final\030\007 \001(\010\022\025\n\rcontrol_input\030\010 \003(\005\022\022\n" +
+      "\ninaccurate\030\021 \001(\010\032;\n\tInputInfo\022\026\n\016preced" +
+      "ing_node\030\001 \001(\005\022\026\n\016preceding_port\030\002 \001(\005\032\206" +
+      "\001\n\nOutputInfo\022\014\n\004size\030\001 \001(\003\022\030\n\020alias_inp" +
+      "ut_port\030\002 \001(\003\022+\n\005shape\030\003 \001(\0132\034.tensorflo" +
+      "w.TensorShapeProto\022#\n\005dtype\030\004 \001(\0162\024.tens" +
+      "orflow.DataType\0321\n\016AggregatedCost\022\014\n\004cos" +
+      "t\030\001 \001(\002\022\021\n\tdimension\030\002 \001(\tB\211\001\n\036org.tenso" +
+      "rflow.proto.frameworkB\017CostGraphProtosP\001" +
+      "ZQgithub.com/tensorflow/tensorflow/tenso" +
+      "rflow/go/core/framework/cost_graph_go_pr" +
+      "oto\370\001\001b\006proto3"
+    };
+    descriptor = com.google.protobuf.Descriptors.FileDescriptor
+      .internalBuildGeneratedFileFrom(descriptorData,
+        new com.google.protobuf.Descriptors.FileDescriptor[] {
+          org.tensorflow.proto.framework.TensorShapeProtos.getDescriptor(),
+          org.tensorflow.proto.framework.TypesProtos.getDescriptor(),
+        });
+    internal_static_tensorflow_CostGraphDef_descriptor =
+      getDescriptor().getMessageTypes().get(0);
+    internal_static_tensorflow_CostGraphDef_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_CostGraphDef_descriptor,
+        new java.lang.String[] { "Node", "Cost", });
+    internal_static_tensorflow_CostGraphDef_Node_descriptor =
+      internal_static_tensorflow_CostGraphDef_descriptor.getNestedTypes().get(0);
+    internal_static_tensorflow_CostGraphDef_Node_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_CostGraphDef_Node_descriptor,
+        new java.lang.String[] { "Name", "Device", "Id", "InputInfo", "OutputInfo", "TemporaryMemorySize", "PersistentMemorySize", "HostTempMemorySize", "DeviceTempMemorySize", "DevicePersistentMemorySize", "ComputeCost", "ComputeTime", "MemoryTime", "IsFinal", "ControlInput", "Inaccurate", });
+    internal_static_tensorflow_CostGraphDef_Node_InputInfo_descriptor =
+      internal_static_tensorflow_CostGraphDef_Node_descriptor.getNestedTypes().get(0);
+    internal_static_tensorflow_CostGraphDef_Node_InputInfo_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_CostGraphDef_Node_InputInfo_descriptor,
+        new java.lang.String[] { "PrecedingNode", "PrecedingPort", });
+    internal_static_tensorflow_CostGraphDef_Node_OutputInfo_descriptor =
+      internal_static_tensorflow_CostGraphDef_Node_descriptor.getNestedTypes().get(1);
+    internal_static_tensorflow_CostGraphDef_Node_OutputInfo_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_CostGraphDef_Node_OutputInfo_descriptor,
+        new java.lang.String[] { "Size", "AliasInputPort", "Shape", "Dtype", });
+    internal_static_tensorflow_CostGraphDef_AggregatedCost_descriptor =
+      internal_static_tensorflow_CostGraphDef_descriptor.getNestedTypes().get(1);
+    internal_static_tensorflow_CostGraphDef_AggregatedCost_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_CostGraphDef_AggregatedCost_descriptor,
+        new java.lang.String[] { "Cost", "Dimension", });
+    org.tensorflow.proto.framework.TensorShapeProtos.getDescriptor();
+    org.tensorflow.proto.framework.TypesProtos.getDescriptor();
+  }
+
+  // @@protoc_insertion_point(outer_class_scope)
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DataClass.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DataClass.java
new file mode 100644
index 00000000000..895366e817c
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DataClass.java
@@ -0,0 +1,167 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/framework/summary.proto
+
+package org.tensorflow.proto.framework;
+
+/**
+ * Protobuf enum {@code tensorflow.DataClass}
+ */
+public enum DataClass
+    implements com.google.protobuf.ProtocolMessageEnum {
+  /**
+   * <pre>
+   * Unknown data class, used (implicitly) for legacy data. Will not be
+   * processed by data ingestion pipelines.
+   * </pre>
+   *
+   * <code>DATA_CLASS_UNKNOWN = 0;</code>
+   */
+  DATA_CLASS_UNKNOWN(0),
+  /**
+   * <pre>
+   * Scalar time series. Each `Value` for the corresponding tag must have
+   * `tensor` set to a rank-0 tensor of floating-point dtype, which will be
+   * converted to float64.
+   * </pre>
+   *
+   * <code>DATA_CLASS_SCALAR = 1;</code>
+   */
+  DATA_CLASS_SCALAR(1),
+  /**
+   * <pre>
+   * Tensor time series. Each `Value` for the corresponding tag must have
+   * `tensor` set. The tensor value is arbitrary, but should be small to
+   * accommodate direct storage in database backends: an upper bound of a few
+   * kilobytes is a reasonable rule of thumb.
+   * </pre>
+   *
+   * <code>DATA_CLASS_TENSOR = 2;</code>
+   */
+  DATA_CLASS_TENSOR(2),
+  /**
+   * <pre>
+   * Blob sequence time series. Each `Value` for the corresponding tag must
+   * have `tensor` set to a rank-1 tensor of bytestring dtype.
+   * </pre>
+   *
+   * <code>DATA_CLASS_BLOB_SEQUENCE = 3;</code>
+   */
+  DATA_CLASS_BLOB_SEQUENCE(3),
+  UNRECOGNIZED(-1),
+  ;
+
+  /**
+   * <pre>
+   * Unknown data class, used (implicitly) for legacy data. Will not be
+   * processed by data ingestion pipelines.
+   * </pre>
+   *
+   * <code>DATA_CLASS_UNKNOWN = 0;</code>
+   */
+  public static final int DATA_CLASS_UNKNOWN_VALUE = 0;
+  /**
+   * <pre>
+   * Scalar time series. Each `Value` for the corresponding tag must have
+   * `tensor` set to a rank-0 tensor of floating-point dtype, which will be
+   * converted to float64.
+   * </pre>
+   *
+   * <code>DATA_CLASS_SCALAR = 1;</code>
+   */
+  public static final int DATA_CLASS_SCALAR_VALUE = 1;
+  /**
+   * <pre>
+   * Tensor time series. Each `Value` for the corresponding tag must have
+   * `tensor` set. The tensor value is arbitrary, but should be small to
+   * accommodate direct storage in database backends: an upper bound of a few
+   * kilobytes is a reasonable rule of thumb.
+   * </pre>
+   *
+   * <code>DATA_CLASS_TENSOR = 2;</code>
+   */
+  public static final int DATA_CLASS_TENSOR_VALUE = 2;
+  /**
+   * <pre>
+   * Blob sequence time series. Each `Value` for the corresponding tag must
+   * have `tensor` set to a rank-1 tensor of bytestring dtype.
+   * </pre>
+   *
+   * <code>DATA_CLASS_BLOB_SEQUENCE = 3;</code>
+   */
+  public static final int DATA_CLASS_BLOB_SEQUENCE_VALUE = 3;
+
+
+  public final int getNumber() {
+    if (this == UNRECOGNIZED) {
+      throw new java.lang.IllegalArgumentException(
+          "Can't get the number of an unknown enum value.");
+    }
+    return value;
+  }
+
+  /**
+   * @deprecated Use {@link #forNumber(int)} instead.
+   */
+  @java.lang.Deprecated
+  public static DataClass valueOf(int value) {
+    return forNumber(value);
+  }
+
+  public static DataClass forNumber(int value) {
+    switch (value) {
+      case 0: return DATA_CLASS_UNKNOWN;
+      case 1: return DATA_CLASS_SCALAR;
+      case 2: return DATA_CLASS_TENSOR;
+      case 3: return DATA_CLASS_BLOB_SEQUENCE;
+      default: return null;
+    }
+  }
+
+  public static com.google.protobuf.Internal.EnumLiteMap<DataClass>
+      internalGetValueMap() {
+    return internalValueMap;
+  }
+  private static final com.google.protobuf.Internal.EnumLiteMap<
+      DataClass> internalValueMap =
+        new com.google.protobuf.Internal.EnumLiteMap<DataClass>() {
+          public DataClass findValueByNumber(int number) {
+            return DataClass.forNumber(number);
+          }
+        };
+
+  public final com.google.protobuf.Descriptors.EnumValueDescriptor
+      getValueDescriptor() {
+    return getDescriptor().getValues().get(ordinal());
+  }
+  public final com.google.protobuf.Descriptors.EnumDescriptor
+      getDescriptorForType() {
+    return getDescriptor();
+  }
+  public static final com.google.protobuf.Descriptors.EnumDescriptor
+      getDescriptor() {
+    return org.tensorflow.proto.framework.SummaryProtos.getDescriptor().getEnumTypes().get(0);
+  }
+
+  private static final DataClass[] VALUES = values();
+
+  public static DataClass valueOf(
+      com.google.protobuf.Descriptors.EnumValueDescriptor desc) {
+    if (desc.getType() != getDescriptor()) {
+      throw new java.lang.IllegalArgumentException(
+        "EnumValueDescriptor is not for this type.");
+    }
+    if (desc.getIndex() == -1) {
+      return UNRECOGNIZED;
+    }
+    return VALUES[desc.getIndex()];
+  }
+
+  private final int value;
+
+  private DataClass(int value) {
+    this.value = value;
+  }
+
+  // @@protoc_insertion_point(enum_scope:tensorflow.DataClass)
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DataType.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DataType.java
new file mode 100644
index 00000000000..6b836c8353a
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DataType.java
@@ -0,0 +1,615 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/framework/types.proto
+
+package org.tensorflow.proto.framework;
+
+/**
+ * <pre>
+ * (== suppress_warning documentation-presence ==)
+ * LINT.IfChange
+ * </pre>
+ *
+ * Protobuf enum {@code tensorflow.DataType}
+ */
+public enum DataType
+    implements com.google.protobuf.ProtocolMessageEnum {
+  /**
+   * <pre>
+   * Not a legal value for DataType.  Used to indicate a DataType field
+   * has not been set.
+   * </pre>
+   *
+   * <code>DT_INVALID = 0;</code>
+   */
+  DT_INVALID(0),
+  /**
+   * <pre>
+   * Data types that all computation devices are expected to be
+   * capable to support.
+   * </pre>
+   *
+   * <code>DT_FLOAT = 1;</code>
+   */
+  DT_FLOAT(1),
+  /**
+   * <code>DT_DOUBLE = 2;</code>
+   */
+  DT_DOUBLE(2),
+  /**
+   * <code>DT_INT32 = 3;</code>
+   */
+  DT_INT32(3),
+  /**
+   * <code>DT_UINT8 = 4;</code>
+   */
+  DT_UINT8(4),
+  /**
+   * <code>DT_INT16 = 5;</code>
+   */
+  DT_INT16(5),
+  /**
+   * <code>DT_INT8 = 6;</code>
+   */
+  DT_INT8(6),
+  /**
+   * <code>DT_STRING = 7;</code>
+   */
+  DT_STRING(7),
+  /**
+   * <pre>
+   * Single-precision complex
+   * </pre>
+   *
+   * <code>DT_COMPLEX64 = 8;</code>
+   */
+  DT_COMPLEX64(8),
+  /**
+   * <code>DT_INT64 = 9;</code>
+   */
+  DT_INT64(9),
+  /**
+   * <code>DT_BOOL = 10;</code>
+   */
+  DT_BOOL(10),
+  /**
+   * <pre>
+   * Quantized int8
+   * </pre>
+   *
+   * <code>DT_QINT8 = 11;</code>
+   */
+  DT_QINT8(11),
+  /**
+   * <pre>
+   * Quantized uint8
+   * </pre>
+   *
+   * <code>DT_QUINT8 = 12;</code>
+   */
+  DT_QUINT8(12),
+  /**
+   * <pre>
+   * Quantized int32
+   * </pre>
+   *
+   * <code>DT_QINT32 = 13;</code>
+   */
+  DT_QINT32(13),
+  /**
+   * <pre>
+   * Float32 truncated to 16 bits.  Only for cast ops.
+   * </pre>
+   *
+   * <code>DT_BFLOAT16 = 14;</code>
+   */
+  DT_BFLOAT16(14),
+  /**
+   * <pre>
+   * Quantized int16
+   * </pre>
+   *
+   * <code>DT_QINT16 = 15;</code>
+   */
+  DT_QINT16(15),
+  /**
+   * <pre>
+   * Quantized uint16
+   * </pre>
+   *
+   * <code>DT_QUINT16 = 16;</code>
+   */
+  DT_QUINT16(16),
+  /**
+   * <code>DT_UINT16 = 17;</code>
+   */
+  DT_UINT16(17),
+  /**
+   * <pre>
+   * Double-precision complex
+   * </pre>
+   *
+   * <code>DT_COMPLEX128 = 18;</code>
+   */
+  DT_COMPLEX128(18),
+  /**
+   * <code>DT_HALF = 19;</code>
+   */
+  DT_HALF(19),
+  /**
+   * <code>DT_RESOURCE = 20;</code>
+   */
+  DT_RESOURCE(20),
+  /**
+   * <pre>
+   * Arbitrary C++ data types
+   * </pre>
+   *
+   * <code>DT_VARIANT = 21;</code>
+   */
+  DT_VARIANT(21),
+  /**
+   * <code>DT_UINT32 = 22;</code>
+   */
+  DT_UINT32(22),
+  /**
+   * <code>DT_UINT64 = 23;</code>
+   */
+  DT_UINT64(23),
+  /**
+   * <pre>
+   * Do not use!  These are only for parameters.  Every enum above
+   * should have a corresponding value below (verified by types_test).
+   * </pre>
+   *
+   * <code>DT_FLOAT_REF = 101;</code>
+   */
+  DT_FLOAT_REF(101),
+  /**
+   * <code>DT_DOUBLE_REF = 102;</code>
+   */
+  DT_DOUBLE_REF(102),
+  /**
+   * <code>DT_INT32_REF = 103;</code>
+   */
+  DT_INT32_REF(103),
+  /**
+   * <code>DT_UINT8_REF = 104;</code>
+   */
+  DT_UINT8_REF(104),
+  /**
+   * <code>DT_INT16_REF = 105;</code>
+   */
+  DT_INT16_REF(105),
+  /**
+   * <code>DT_INT8_REF = 106;</code>
+   */
+  DT_INT8_REF(106),
+  /**
+   * <code>DT_STRING_REF = 107;</code>
+   */
+  DT_STRING_REF(107),
+  /**
+   * <code>DT_COMPLEX64_REF = 108;</code>
+   */
+  DT_COMPLEX64_REF(108),
+  /**
+   * <code>DT_INT64_REF = 109;</code>
+   */
+  DT_INT64_REF(109),
+  /**
+   * <code>DT_BOOL_REF = 110;</code>
+   */
+  DT_BOOL_REF(110),
+  /**
+   * <code>DT_QINT8_REF = 111;</code>
+   */
+  DT_QINT8_REF(111),
+  /**
+   * <code>DT_QUINT8_REF = 112;</code>
+   */
+  DT_QUINT8_REF(112),
+  /**
+   * <code>DT_QINT32_REF = 113;</code>
+   */
+  DT_QINT32_REF(113),
+  /**
+   * <code>DT_BFLOAT16_REF = 114;</code>
+   */
+  DT_BFLOAT16_REF(114),
+  /**
+   * <code>DT_QINT16_REF = 115;</code>
+   */
+  DT_QINT16_REF(115),
+  /**
+   * <code>DT_QUINT16_REF = 116;</code>
+   */
+  DT_QUINT16_REF(116),
+  /**
+   * <code>DT_UINT16_REF = 117;</code>
+   */
+  DT_UINT16_REF(117),
+  /**
+   * <code>DT_COMPLEX128_REF = 118;</code>
+   */
+  DT_COMPLEX128_REF(118),
+  /**
+   * <code>DT_HALF_REF = 119;</code>
+   */
+  DT_HALF_REF(119),
+  /**
+   * <code>DT_RESOURCE_REF = 120;</code>
+   */
+  DT_RESOURCE_REF(120),
+  /**
+   * <code>DT_VARIANT_REF = 121;</code>
+   */
+  DT_VARIANT_REF(121),
+  /**
+   * <code>DT_UINT32_REF = 122;</code>
+   */
+  DT_UINT32_REF(122),
+  /**
+   * <code>DT_UINT64_REF = 123;</code>
+   */
+  DT_UINT64_REF(123),
+  UNRECOGNIZED(-1),
+  ;
+
+  /**
+   * <pre>
+   * Not a legal value for DataType.  Used to indicate a DataType field
+   * has not been set.
+   * </pre>
+   *
+   * <code>DT_INVALID = 0;</code>
+   */
+  public static final int DT_INVALID_VALUE = 0;
+  /**
+   * <pre>
+   * Data types that all computation devices are expected to be
+   * capable to support.
+   * </pre>
+   *
+   * <code>DT_FLOAT = 1;</code>
+   */
+  public static final int DT_FLOAT_VALUE = 1;
+  /**
+   * <code>DT_DOUBLE = 2;</code>
+   */
+  public static final int DT_DOUBLE_VALUE = 2;
+  /**
+   * <code>DT_INT32 = 3;</code>
+   */
+  public static final int DT_INT32_VALUE = 3;
+  /**
+   * <code>DT_UINT8 = 4;</code>
+   */
+  public static final int DT_UINT8_VALUE = 4;
+  /**
+   * <code>DT_INT16 = 5;</code>
+   */
+  public static final int DT_INT16_VALUE = 5;
+  /**
+   * <code>DT_INT8 = 6;</code>
+   */
+  public static final int DT_INT8_VALUE = 6;
+  /**
+   * <code>DT_STRING = 7;</code>
+   */
+  public static final int DT_STRING_VALUE = 7;
+  /**
+   * <pre>
+   * Single-precision complex
+   * </pre>
+   *
+   * <code>DT_COMPLEX64 = 8;</code>
+   */
+  public static final int DT_COMPLEX64_VALUE = 8;
+  /**
+   * <code>DT_INT64 = 9;</code>
+   */
+  public static final int DT_INT64_VALUE = 9;
+  /**
+   * <code>DT_BOOL = 10;</code>
+   */
+  public static final int DT_BOOL_VALUE = 10;
+  /**
+   * <pre>
+   * Quantized int8
+   * </pre>
+   *
+   * <code>DT_QINT8 = 11;</code>
+   */
+  public static final int DT_QINT8_VALUE = 11;
+  /**
+   * <pre>
+   * Quantized uint8
+   * </pre>
+   *
+   * <code>DT_QUINT8 = 12;</code>
+   */
+  public static final int DT_QUINT8_VALUE = 12;
+  /**
+   * <pre>
+   * Quantized int32
+   * </pre>
+   *
+   * <code>DT_QINT32 = 13;</code>
+   */
+  public static final int DT_QINT32_VALUE = 13;
+  /**
+   * <pre>
+   * Float32 truncated to 16 bits.  Only for cast ops.
+   * </pre>
+   *
+   * <code>DT_BFLOAT16 = 14;</code>
+   */
+  public static final int DT_BFLOAT16_VALUE = 14;
+  /**
+   * <pre>
+   * Quantized int16
+   * </pre>
+   *
+   * <code>DT_QINT16 = 15;</code>
+   */
+  public static final int DT_QINT16_VALUE = 15;
+  /**
+   * <pre>
+   * Quantized uint16
+   * </pre>
+   *
+   * <code>DT_QUINT16 = 16;</code>
+   */
+  public static final int DT_QUINT16_VALUE = 16;
+  /**
+   * <code>DT_UINT16 = 17;</code>
+   */
+  public static final int DT_UINT16_VALUE = 17;
+  /**
+   * <pre>
+   * Double-precision complex
+   * </pre>
+   *
+   * <code>DT_COMPLEX128 = 18;</code>
+   */
+  public static final int DT_COMPLEX128_VALUE = 18;
+  /**
+   * <code>DT_HALF = 19;</code>
+   */
+  public static final int DT_HALF_VALUE = 19;
+  /**
+   * <code>DT_RESOURCE = 20;</code>
+   */
+  public static final int DT_RESOURCE_VALUE = 20;
+  /**
+   * <pre>
+   * Arbitrary C++ data types
+   * </pre>
+   *
+   * <code>DT_VARIANT = 21;</code>
+   */
+  public static final int DT_VARIANT_VALUE = 21;
+  /**
+   * <code>DT_UINT32 = 22;</code>
+   */
+  public static final int DT_UINT32_VALUE = 22;
+  /**
+   * <code>DT_UINT64 = 23;</code>
+   */
+  public static final int DT_UINT64_VALUE = 23;
+  /**
+   * <pre>
+   * Do not use!  These are only for parameters.  Every enum above
+   * should have a corresponding value below (verified by types_test).
+   * </pre>
+   *
+   * <code>DT_FLOAT_REF = 101;</code>
+   */
+  public static final int DT_FLOAT_REF_VALUE = 101;
+  /**
+   * <code>DT_DOUBLE_REF = 102;</code>
+   */
+  public static final int DT_DOUBLE_REF_VALUE = 102;
+  /**
+   * <code>DT_INT32_REF = 103;</code>
+   */
+  public static final int DT_INT32_REF_VALUE = 103;
+  /**
+   * <code>DT_UINT8_REF = 104;</code>
+   */
+  public static final int DT_UINT8_REF_VALUE = 104;
+  /**
+   * <code>DT_INT16_REF = 105;</code>
+   */
+  public static final int DT_INT16_REF_VALUE = 105;
+  /**
+   * <code>DT_INT8_REF = 106;</code>
+   */
+  public static final int DT_INT8_REF_VALUE = 106;
+  /**
+   * <code>DT_STRING_REF = 107;</code>
+   */
+  public static final int DT_STRING_REF_VALUE = 107;
+  /**
+   * <code>DT_COMPLEX64_REF = 108;</code>
+   */
+  public static final int DT_COMPLEX64_REF_VALUE = 108;
+  /**
+   * <code>DT_INT64_REF = 109;</code>
+   */
+  public static final int DT_INT64_REF_VALUE = 109;
+  /**
+   * <code>DT_BOOL_REF = 110;</code>
+   */
+  public static final int DT_BOOL_REF_VALUE = 110;
+  /**
+   * <code>DT_QINT8_REF = 111;</code>
+   */
+  public static final int DT_QINT8_REF_VALUE = 111;
+  /**
+   * <code>DT_QUINT8_REF = 112;</code>
+   */
+  public static final int DT_QUINT8_REF_VALUE = 112;
+  /**
+   * <code>DT_QINT32_REF = 113;</code>
+   */
+  public static final int DT_QINT32_REF_VALUE = 113;
+  /**
+   * <code>DT_BFLOAT16_REF = 114;</code>
+   */
+  public static final int DT_BFLOAT16_REF_VALUE = 114;
+  /**
+   * <code>DT_QINT16_REF = 115;</code>
+   */
+  public static final int DT_QINT16_REF_VALUE = 115;
+  /**
+   * <code>DT_QUINT16_REF = 116;</code>
+   */
+  public static final int DT_QUINT16_REF_VALUE = 116;
+  /**
+   * <code>DT_UINT16_REF = 117;</code>
+   */
+  public static final int DT_UINT16_REF_VALUE = 117;
+  /**
+   * <code>DT_COMPLEX128_REF = 118;</code>
+   */
+  public static final int DT_COMPLEX128_REF_VALUE = 118;
+  /**
+   * <code>DT_HALF_REF = 119;</code>
+   */
+  public static final int DT_HALF_REF_VALUE = 119;
+  /**
+   * <code>DT_RESOURCE_REF = 120;</code>
+   */
+  public static final int DT_RESOURCE_REF_VALUE = 120;
+  /**
+   * <code>DT_VARIANT_REF = 121;</code>
+   */
+  public static final int DT_VARIANT_REF_VALUE = 121;
+  /**
+   * <code>DT_UINT32_REF = 122;</code>
+   */
+  public static final int DT_UINT32_REF_VALUE = 122;
+  /**
+   * <code>DT_UINT64_REF = 123;</code>
+   */
+  public static final int DT_UINT64_REF_VALUE = 123;
+
+
+  public final int getNumber() {
+    if (this == UNRECOGNIZED) {
+      throw new java.lang.IllegalArgumentException(
+          "Can't get the number of an unknown enum value.");
+    }
+    return value;
+  }
+
+  /**
+   * @deprecated Use {@link #forNumber(int)} instead.
+   */
+  @java.lang.Deprecated
+  public static DataType valueOf(int value) {
+    return forNumber(value);
+  }
+
+  public static DataType forNumber(int value) {
+    switch (value) {
+      case 0: return DT_INVALID;
+      case 1: return DT_FLOAT;
+      case 2: return DT_DOUBLE;
+      case 3: return DT_INT32;
+      case 4: return DT_UINT8;
+      case 5: return DT_INT16;
+      case 6: return DT_INT8;
+      case 7: return DT_STRING;
+      case 8: return DT_COMPLEX64;
+      case 9: return DT_INT64;
+      case 10: return DT_BOOL;
+      case 11: return DT_QINT8;
+      case 12: return DT_QUINT8;
+      case 13: return DT_QINT32;
+      case 14: return DT_BFLOAT16;
+      case 15: return DT_QINT16;
+      case 16: return DT_QUINT16;
+      case 17: return DT_UINT16;
+      case 18: return DT_COMPLEX128;
+      case 19: return DT_HALF;
+      case 20: return DT_RESOURCE;
+      case 21: return DT_VARIANT;
+      case 22: return DT_UINT32;
+      case 23: return DT_UINT64;
+      case 101: return DT_FLOAT_REF;
+      case 102: return DT_DOUBLE_REF;
+      case 103: return DT_INT32_REF;
+      case 104: return DT_UINT8_REF;
+      case 105: return DT_INT16_REF;
+      case 106: return DT_INT8_REF;
+      case 107: return DT_STRING_REF;
+      case 108: return DT_COMPLEX64_REF;
+      case 109: return DT_INT64_REF;
+      case 110: return DT_BOOL_REF;
+      case 111: return DT_QINT8_REF;
+      case 112: return DT_QUINT8_REF;
+      case 113: return DT_QINT32_REF;
+      case 114: return DT_BFLOAT16_REF;
+      case 115: return DT_QINT16_REF;
+      case 116: return DT_QUINT16_REF;
+      case 117: return DT_UINT16_REF;
+      case 118: return DT_COMPLEX128_REF;
+      case 119: return DT_HALF_REF;
+      case 120: return DT_RESOURCE_REF;
+      case 121: return DT_VARIANT_REF;
+      case 122: return DT_UINT32_REF;
+      case 123: return DT_UINT64_REF;
+      default: return null;
+    }
+  }
+
+  public static com.google.protobuf.Internal.EnumLiteMap<DataType>
+      internalGetValueMap() {
+    return internalValueMap;
+  }
+  private static final com.google.protobuf.Internal.EnumLiteMap<
+      DataType> internalValueMap =
+        new com.google.protobuf.Internal.EnumLiteMap<DataType>() {
+          public DataType findValueByNumber(int number) {
+            return DataType.forNumber(number);
+          }
+        };
+
+  public final com.google.protobuf.Descriptors.EnumValueDescriptor
+      getValueDescriptor() {
+    return getDescriptor().getValues().get(ordinal());
+  }
+  public final com.google.protobuf.Descriptors.EnumDescriptor
+      getDescriptorForType() {
+    return getDescriptor();
+  }
+  public static final com.google.protobuf.Descriptors.EnumDescriptor
+      getDescriptor() {
+    return org.tensorflow.proto.framework.TypesProtos.getDescriptor().getEnumTypes().get(0);
+  }
+
+  private static final DataType[] VALUES = values();
+
+  public static DataType valueOf(
+      com.google.protobuf.Descriptors.EnumValueDescriptor desc) {
+    if (desc.getType() != getDescriptor()) {
+      throw new java.lang.IllegalArgumentException(
+        "EnumValueDescriptor is not for this type.");
+    }
+    if (desc.getIndex() == -1) {
+      return UNRECOGNIZED;
+    }
+    return VALUES[desc.getIndex()];
+  }
+
+  private final int value;
+
+  private DataType(int value) {
+    this.value = value;
+  }
+
+  // @@protoc_insertion_point(enum_scope:tensorflow.DataType)
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DebugOptions.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DebugOptions.java
new file mode 100644
index 00000000000..73b4e10af4e
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DebugOptions.java
@@ -0,0 +1,1033 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/debug.proto
+
+package org.tensorflow.proto.framework;
+
+/**
+ * <pre>
+ * Options for initializing DebuggerState in TensorFlow Debugger (tfdbg).
+ * </pre>
+ *
+ * Protobuf type {@code tensorflow.DebugOptions}
+ */
+public  final class DebugOptions extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.DebugOptions)
+    DebugOptionsOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use DebugOptions.newBuilder() to construct.
+  private DebugOptions(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private DebugOptions() {
+    debugTensorWatchOpts_ = java.util.Collections.emptyList();
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new DebugOptions();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private DebugOptions(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    int mutable_bitField0_ = 0;
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 34: {
+            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
+              debugTensorWatchOpts_ = new java.util.ArrayList<org.tensorflow.proto.framework.DebugTensorWatch>();
+              mutable_bitField0_ |= 0x00000001;
+            }
+            debugTensorWatchOpts_.add(
+                input.readMessage(org.tensorflow.proto.framework.DebugTensorWatch.parser(), extensionRegistry));
+            break;
+          }
+          case 80: {
+
+            globalStep_ = input.readInt64();
+            break;
+          }
+          case 88: {
+
+            resetDiskByteUsage_ = input.readBool();
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      if (((mutable_bitField0_ & 0x00000001) != 0)) {
+        debugTensorWatchOpts_ = java.util.Collections.unmodifiableList(debugTensorWatchOpts_);
+      }
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.framework.DebugProtos.internal_static_tensorflow_DebugOptions_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.framework.DebugProtos.internal_static_tensorflow_DebugOptions_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.framework.DebugOptions.class, org.tensorflow.proto.framework.DebugOptions.Builder.class);
+  }
+
+  public static final int DEBUG_TENSOR_WATCH_OPTS_FIELD_NUMBER = 4;
+  private java.util.List<org.tensorflow.proto.framework.DebugTensorWatch> debugTensorWatchOpts_;
+  /**
+   * <pre>
+   * Debugging options
+   * </pre>
+   *
+   * <code>repeated .tensorflow.DebugTensorWatch debug_tensor_watch_opts = 4;</code>
+   */
+  public java.util.List<org.tensorflow.proto.framework.DebugTensorWatch> getDebugTensorWatchOptsList() {
+    return debugTensorWatchOpts_;
+  }
+  /**
+   * <pre>
+   * Debugging options
+   * </pre>
+   *
+   * <code>repeated .tensorflow.DebugTensorWatch debug_tensor_watch_opts = 4;</code>
+   */
+  public java.util.List<? extends org.tensorflow.proto.framework.DebugTensorWatchOrBuilder> 
+      getDebugTensorWatchOptsOrBuilderList() {
+    return debugTensorWatchOpts_;
+  }
+  /**
+   * <pre>
+   * Debugging options
+   * </pre>
+   *
+   * <code>repeated .tensorflow.DebugTensorWatch debug_tensor_watch_opts = 4;</code>
+   */
+  public int getDebugTensorWatchOptsCount() {
+    return debugTensorWatchOpts_.size();
+  }
+  /**
+   * <pre>
+   * Debugging options
+   * </pre>
+   *
+   * <code>repeated .tensorflow.DebugTensorWatch debug_tensor_watch_opts = 4;</code>
+   */
+  public org.tensorflow.proto.framework.DebugTensorWatch getDebugTensorWatchOpts(int index) {
+    return debugTensorWatchOpts_.get(index);
+  }
+  /**
+   * <pre>
+   * Debugging options
+   * </pre>
+   *
+   * <code>repeated .tensorflow.DebugTensorWatch debug_tensor_watch_opts = 4;</code>
+   */
+  public org.tensorflow.proto.framework.DebugTensorWatchOrBuilder getDebugTensorWatchOptsOrBuilder(
+      int index) {
+    return debugTensorWatchOpts_.get(index);
+  }
+
+  public static final int GLOBAL_STEP_FIELD_NUMBER = 10;
+  private long globalStep_;
+  /**
+   * <pre>
+   * Caller-specified global step count.
+   * Note that this is distinct from the session run count and the executor
+   * step count.
+   * </pre>
+   *
+   * <code>int64 global_step = 10;</code>
+   */
+  public long getGlobalStep() {
+    return globalStep_;
+  }
+
+  public static final int RESET_DISK_BYTE_USAGE_FIELD_NUMBER = 11;
+  private boolean resetDiskByteUsage_;
+  /**
+   * <pre>
+   * Whether the total disk usage of tfdbg is to be reset to zero
+   * in this Session.run call. This is used by wrappers and hooks
+   * such as the local CLI ones to indicate that the dumped tensors
+   * are cleaned up from the disk after each Session.run.
+   * </pre>
+   *
+   * <code>bool reset_disk_byte_usage = 11;</code>
+   */
+  public boolean getResetDiskByteUsage() {
+    return resetDiskByteUsage_;
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    for (int i = 0; i < debugTensorWatchOpts_.size(); i++) {
+      output.writeMessage(4, debugTensorWatchOpts_.get(i));
+    }
+    if (globalStep_ != 0L) {
+      output.writeInt64(10, globalStep_);
+    }
+    if (resetDiskByteUsage_ != false) {
+      output.writeBool(11, resetDiskByteUsage_);
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    for (int i = 0; i < debugTensorWatchOpts_.size(); i++) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(4, debugTensorWatchOpts_.get(i));
+    }
+    if (globalStep_ != 0L) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt64Size(10, globalStep_);
+    }
+    if (resetDiskByteUsage_ != false) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeBoolSize(11, resetDiskByteUsage_);
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.framework.DebugOptions)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.framework.DebugOptions other = (org.tensorflow.proto.framework.DebugOptions) obj;
+
+    if (!getDebugTensorWatchOptsList()
+        .equals(other.getDebugTensorWatchOptsList())) return false;
+    if (getGlobalStep()
+        != other.getGlobalStep()) return false;
+    if (getResetDiskByteUsage()
+        != other.getResetDiskByteUsage()) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    if (getDebugTensorWatchOptsCount() > 0) {
+      hash = (37 * hash) + DEBUG_TENSOR_WATCH_OPTS_FIELD_NUMBER;
+      hash = (53 * hash) + getDebugTensorWatchOptsList().hashCode();
+    }
+    hash = (37 * hash) + GLOBAL_STEP_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+        getGlobalStep());
+    hash = (37 * hash) + RESET_DISK_BYTE_USAGE_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashBoolean(
+        getResetDiskByteUsage());
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.framework.DebugOptions parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.DebugOptions parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.DebugOptions parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.DebugOptions parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.DebugOptions parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.DebugOptions parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.DebugOptions parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.DebugOptions parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.DebugOptions parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.DebugOptions parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.DebugOptions parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.DebugOptions parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.framework.DebugOptions prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * <pre>
+   * Options for initializing DebuggerState in TensorFlow Debugger (tfdbg).
+   * </pre>
+   *
+   * Protobuf type {@code tensorflow.DebugOptions}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.DebugOptions)
+      org.tensorflow.proto.framework.DebugOptionsOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.DebugProtos.internal_static_tensorflow_DebugOptions_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.DebugProtos.internal_static_tensorflow_DebugOptions_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.DebugOptions.class, org.tensorflow.proto.framework.DebugOptions.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.framework.DebugOptions.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+        getDebugTensorWatchOptsFieldBuilder();
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      if (debugTensorWatchOptsBuilder_ == null) {
+        debugTensorWatchOpts_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000001);
+      } else {
+        debugTensorWatchOptsBuilder_.clear();
+      }
+      globalStep_ = 0L;
+
+      resetDiskByteUsage_ = false;
+
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.framework.DebugProtos.internal_static_tensorflow_DebugOptions_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.DebugOptions getDefaultInstanceForType() {
+      return org.tensorflow.proto.framework.DebugOptions.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.DebugOptions build() {
+      org.tensorflow.proto.framework.DebugOptions result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.DebugOptions buildPartial() {
+      org.tensorflow.proto.framework.DebugOptions result = new org.tensorflow.proto.framework.DebugOptions(this);
+      int from_bitField0_ = bitField0_;
+      if (debugTensorWatchOptsBuilder_ == null) {
+        if (((bitField0_ & 0x00000001) != 0)) {
+          debugTensorWatchOpts_ = java.util.Collections.unmodifiableList(debugTensorWatchOpts_);
+          bitField0_ = (bitField0_ & ~0x00000001);
+        }
+        result.debugTensorWatchOpts_ = debugTensorWatchOpts_;
+      } else {
+        result.debugTensorWatchOpts_ = debugTensorWatchOptsBuilder_.build();
+      }
+      result.globalStep_ = globalStep_;
+      result.resetDiskByteUsage_ = resetDiskByteUsage_;
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.framework.DebugOptions) {
+        return mergeFrom((org.tensorflow.proto.framework.DebugOptions)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.framework.DebugOptions other) {
+      if (other == org.tensorflow.proto.framework.DebugOptions.getDefaultInstance()) return this;
+      if (debugTensorWatchOptsBuilder_ == null) {
+        if (!other.debugTensorWatchOpts_.isEmpty()) {
+          if (debugTensorWatchOpts_.isEmpty()) {
+            debugTensorWatchOpts_ = other.debugTensorWatchOpts_;
+            bitField0_ = (bitField0_ & ~0x00000001);
+          } else {
+            ensureDebugTensorWatchOptsIsMutable();
+            debugTensorWatchOpts_.addAll(other.debugTensorWatchOpts_);
+          }
+          onChanged();
+        }
+      } else {
+        if (!other.debugTensorWatchOpts_.isEmpty()) {
+          if (debugTensorWatchOptsBuilder_.isEmpty()) {
+            debugTensorWatchOptsBuilder_.dispose();
+            debugTensorWatchOptsBuilder_ = null;
+            debugTensorWatchOpts_ = other.debugTensorWatchOpts_;
+            bitField0_ = (bitField0_ & ~0x00000001);
+            debugTensorWatchOptsBuilder_ = 
+              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
+                 getDebugTensorWatchOptsFieldBuilder() : null;
+          } else {
+            debugTensorWatchOptsBuilder_.addAllMessages(other.debugTensorWatchOpts_);
+          }
+        }
+      }
+      if (other.getGlobalStep() != 0L) {
+        setGlobalStep(other.getGlobalStep());
+      }
+      if (other.getResetDiskByteUsage() != false) {
+        setResetDiskByteUsage(other.getResetDiskByteUsage());
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.framework.DebugOptions parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.framework.DebugOptions) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+    private int bitField0_;
+
+    private java.util.List<org.tensorflow.proto.framework.DebugTensorWatch> debugTensorWatchOpts_ =
+      java.util.Collections.emptyList();
+    private void ensureDebugTensorWatchOptsIsMutable() {
+      if (!((bitField0_ & 0x00000001) != 0)) {
+        debugTensorWatchOpts_ = new java.util.ArrayList<org.tensorflow.proto.framework.DebugTensorWatch>(debugTensorWatchOpts_);
+        bitField0_ |= 0x00000001;
+       }
+    }
+
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.framework.DebugTensorWatch, org.tensorflow.proto.framework.DebugTensorWatch.Builder, org.tensorflow.proto.framework.DebugTensorWatchOrBuilder> debugTensorWatchOptsBuilder_;
+
+    /**
+     * <pre>
+     * Debugging options
+     * </pre>
+     *
+     * <code>repeated .tensorflow.DebugTensorWatch debug_tensor_watch_opts = 4;</code>
+     */
+    public java.util.List<org.tensorflow.proto.framework.DebugTensorWatch> getDebugTensorWatchOptsList() {
+      if (debugTensorWatchOptsBuilder_ == null) {
+        return java.util.Collections.unmodifiableList(debugTensorWatchOpts_);
+      } else {
+        return debugTensorWatchOptsBuilder_.getMessageList();
+      }
+    }
+    /**
+     * <pre>
+     * Debugging options
+     * </pre>
+     *
+     * <code>repeated .tensorflow.DebugTensorWatch debug_tensor_watch_opts = 4;</code>
+     */
+    public int getDebugTensorWatchOptsCount() {
+      if (debugTensorWatchOptsBuilder_ == null) {
+        return debugTensorWatchOpts_.size();
+      } else {
+        return debugTensorWatchOptsBuilder_.getCount();
+      }
+    }
+    /**
+     * <pre>
+     * Debugging options
+     * </pre>
+     *
+     * <code>repeated .tensorflow.DebugTensorWatch debug_tensor_watch_opts = 4;</code>
+     */
+    public org.tensorflow.proto.framework.DebugTensorWatch getDebugTensorWatchOpts(int index) {
+      if (debugTensorWatchOptsBuilder_ == null) {
+        return debugTensorWatchOpts_.get(index);
+      } else {
+        return debugTensorWatchOptsBuilder_.getMessage(index);
+      }
+    }
+    /**
+     * <pre>
+     * Debugging options
+     * </pre>
+     *
+     * <code>repeated .tensorflow.DebugTensorWatch debug_tensor_watch_opts = 4;</code>
+     */
+    public Builder setDebugTensorWatchOpts(
+        int index, org.tensorflow.proto.framework.DebugTensorWatch value) {
+      if (debugTensorWatchOptsBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureDebugTensorWatchOptsIsMutable();
+        debugTensorWatchOpts_.set(index, value);
+        onChanged();
+      } else {
+        debugTensorWatchOptsBuilder_.setMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Debugging options
+     * </pre>
+     *
+     * <code>repeated .tensorflow.DebugTensorWatch debug_tensor_watch_opts = 4;</code>
+     */
+    public Builder setDebugTensorWatchOpts(
+        int index, org.tensorflow.proto.framework.DebugTensorWatch.Builder builderForValue) {
+      if (debugTensorWatchOptsBuilder_ == null) {
+        ensureDebugTensorWatchOptsIsMutable();
+        debugTensorWatchOpts_.set(index, builderForValue.build());
+        onChanged();
+      } else {
+        debugTensorWatchOptsBuilder_.setMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Debugging options
+     * </pre>
+     *
+     * <code>repeated .tensorflow.DebugTensorWatch debug_tensor_watch_opts = 4;</code>
+     */
+    public Builder addDebugTensorWatchOpts(org.tensorflow.proto.framework.DebugTensorWatch value) {
+      if (debugTensorWatchOptsBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureDebugTensorWatchOptsIsMutable();
+        debugTensorWatchOpts_.add(value);
+        onChanged();
+      } else {
+        debugTensorWatchOptsBuilder_.addMessage(value);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Debugging options
+     * </pre>
+     *
+     * <code>repeated .tensorflow.DebugTensorWatch debug_tensor_watch_opts = 4;</code>
+     */
+    public Builder addDebugTensorWatchOpts(
+        int index, org.tensorflow.proto.framework.DebugTensorWatch value) {
+      if (debugTensorWatchOptsBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureDebugTensorWatchOptsIsMutable();
+        debugTensorWatchOpts_.add(index, value);
+        onChanged();
+      } else {
+        debugTensorWatchOptsBuilder_.addMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Debugging options
+     * </pre>
+     *
+     * <code>repeated .tensorflow.DebugTensorWatch debug_tensor_watch_opts = 4;</code>
+     */
+    public Builder addDebugTensorWatchOpts(
+        org.tensorflow.proto.framework.DebugTensorWatch.Builder builderForValue) {
+      if (debugTensorWatchOptsBuilder_ == null) {
+        ensureDebugTensorWatchOptsIsMutable();
+        debugTensorWatchOpts_.add(builderForValue.build());
+        onChanged();
+      } else {
+        debugTensorWatchOptsBuilder_.addMessage(builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Debugging options
+     * </pre>
+     *
+     * <code>repeated .tensorflow.DebugTensorWatch debug_tensor_watch_opts = 4;</code>
+     */
+    public Builder addDebugTensorWatchOpts(
+        int index, org.tensorflow.proto.framework.DebugTensorWatch.Builder builderForValue) {
+      if (debugTensorWatchOptsBuilder_ == null) {
+        ensureDebugTensorWatchOptsIsMutable();
+        debugTensorWatchOpts_.add(index, builderForValue.build());
+        onChanged();
+      } else {
+        debugTensorWatchOptsBuilder_.addMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Debugging options
+     * </pre>
+     *
+     * <code>repeated .tensorflow.DebugTensorWatch debug_tensor_watch_opts = 4;</code>
+     */
+    public Builder addAllDebugTensorWatchOpts(
+        java.lang.Iterable<? extends org.tensorflow.proto.framework.DebugTensorWatch> values) {
+      if (debugTensorWatchOptsBuilder_ == null) {
+        ensureDebugTensorWatchOptsIsMutable();
+        com.google.protobuf.AbstractMessageLite.Builder.addAll(
+            values, debugTensorWatchOpts_);
+        onChanged();
+      } else {
+        debugTensorWatchOptsBuilder_.addAllMessages(values);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Debugging options
+     * </pre>
+     *
+     * <code>repeated .tensorflow.DebugTensorWatch debug_tensor_watch_opts = 4;</code>
+     */
+    public Builder clearDebugTensorWatchOpts() {
+      if (debugTensorWatchOptsBuilder_ == null) {
+        debugTensorWatchOpts_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000001);
+        onChanged();
+      } else {
+        debugTensorWatchOptsBuilder_.clear();
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Debugging options
+     * </pre>
+     *
+     * <code>repeated .tensorflow.DebugTensorWatch debug_tensor_watch_opts = 4;</code>
+     */
+    public Builder removeDebugTensorWatchOpts(int index) {
+      if (debugTensorWatchOptsBuilder_ == null) {
+        ensureDebugTensorWatchOptsIsMutable();
+        debugTensorWatchOpts_.remove(index);
+        onChanged();
+      } else {
+        debugTensorWatchOptsBuilder_.remove(index);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Debugging options
+     * </pre>
+     *
+     * <code>repeated .tensorflow.DebugTensorWatch debug_tensor_watch_opts = 4;</code>
+     */
+    public org.tensorflow.proto.framework.DebugTensorWatch.Builder getDebugTensorWatchOptsBuilder(
+        int index) {
+      return getDebugTensorWatchOptsFieldBuilder().getBuilder(index);
+    }
+    /**
+     * <pre>
+     * Debugging options
+     * </pre>
+     *
+     * <code>repeated .tensorflow.DebugTensorWatch debug_tensor_watch_opts = 4;</code>
+     */
+    public org.tensorflow.proto.framework.DebugTensorWatchOrBuilder getDebugTensorWatchOptsOrBuilder(
+        int index) {
+      if (debugTensorWatchOptsBuilder_ == null) {
+        return debugTensorWatchOpts_.get(index);  } else {
+        return debugTensorWatchOptsBuilder_.getMessageOrBuilder(index);
+      }
+    }
+    /**
+     * <pre>
+     * Debugging options
+     * </pre>
+     *
+     * <code>repeated .tensorflow.DebugTensorWatch debug_tensor_watch_opts = 4;</code>
+     */
+    public java.util.List<? extends org.tensorflow.proto.framework.DebugTensorWatchOrBuilder> 
+         getDebugTensorWatchOptsOrBuilderList() {
+      if (debugTensorWatchOptsBuilder_ != null) {
+        return debugTensorWatchOptsBuilder_.getMessageOrBuilderList();
+      } else {
+        return java.util.Collections.unmodifiableList(debugTensorWatchOpts_);
+      }
+    }
+    /**
+     * <pre>
+     * Debugging options
+     * </pre>
+     *
+     * <code>repeated .tensorflow.DebugTensorWatch debug_tensor_watch_opts = 4;</code>
+     */
+    public org.tensorflow.proto.framework.DebugTensorWatch.Builder addDebugTensorWatchOptsBuilder() {
+      return getDebugTensorWatchOptsFieldBuilder().addBuilder(
+          org.tensorflow.proto.framework.DebugTensorWatch.getDefaultInstance());
+    }
+    /**
+     * <pre>
+     * Debugging options
+     * </pre>
+     *
+     * <code>repeated .tensorflow.DebugTensorWatch debug_tensor_watch_opts = 4;</code>
+     */
+    public org.tensorflow.proto.framework.DebugTensorWatch.Builder addDebugTensorWatchOptsBuilder(
+        int index) {
+      return getDebugTensorWatchOptsFieldBuilder().addBuilder(
+          index, org.tensorflow.proto.framework.DebugTensorWatch.getDefaultInstance());
+    }
+    /**
+     * <pre>
+     * Debugging options
+     * </pre>
+     *
+     * <code>repeated .tensorflow.DebugTensorWatch debug_tensor_watch_opts = 4;</code>
+     */
+    public java.util.List<org.tensorflow.proto.framework.DebugTensorWatch.Builder> 
+         getDebugTensorWatchOptsBuilderList() {
+      return getDebugTensorWatchOptsFieldBuilder().getBuilderList();
+    }
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.framework.DebugTensorWatch, org.tensorflow.proto.framework.DebugTensorWatch.Builder, org.tensorflow.proto.framework.DebugTensorWatchOrBuilder> 
+        getDebugTensorWatchOptsFieldBuilder() {
+      if (debugTensorWatchOptsBuilder_ == null) {
+        debugTensorWatchOptsBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
+            org.tensorflow.proto.framework.DebugTensorWatch, org.tensorflow.proto.framework.DebugTensorWatch.Builder, org.tensorflow.proto.framework.DebugTensorWatchOrBuilder>(
+                debugTensorWatchOpts_,
+                ((bitField0_ & 0x00000001) != 0),
+                getParentForChildren(),
+                isClean());
+        debugTensorWatchOpts_ = null;
+      }
+      return debugTensorWatchOptsBuilder_;
+    }
+
+    private long globalStep_ ;
+    /**
+     * <pre>
+     * Caller-specified global step count.
+     * Note that this is distinct from the session run count and the executor
+     * step count.
+     * </pre>
+     *
+     * <code>int64 global_step = 10;</code>
+     */
+    public long getGlobalStep() {
+      return globalStep_;
+    }
+    /**
+     * <pre>
+     * Caller-specified global step count.
+     * Note that this is distinct from the session run count and the executor
+     * step count.
+     * </pre>
+     *
+     * <code>int64 global_step = 10;</code>
+     */
+    public Builder setGlobalStep(long value) {
+      
+      globalStep_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Caller-specified global step count.
+     * Note that this is distinct from the session run count and the executor
+     * step count.
+     * </pre>
+     *
+     * <code>int64 global_step = 10;</code>
+     */
+    public Builder clearGlobalStep() {
+      
+      globalStep_ = 0L;
+      onChanged();
+      return this;
+    }
+
+    private boolean resetDiskByteUsage_ ;
+    /**
+     * <pre>
+     * Whether the total disk usage of tfdbg is to be reset to zero
+     * in this Session.run call. This is used by wrappers and hooks
+     * such as the local CLI ones to indicate that the dumped tensors
+     * are cleaned up from the disk after each Session.run.
+     * </pre>
+     *
+     * <code>bool reset_disk_byte_usage = 11;</code>
+     */
+    public boolean getResetDiskByteUsage() {
+      return resetDiskByteUsage_;
+    }
+    /**
+     * <pre>
+     * Whether the total disk usage of tfdbg is to be reset to zero
+     * in this Session.run call. This is used by wrappers and hooks
+     * such as the local CLI ones to indicate that the dumped tensors
+     * are cleaned up from the disk after each Session.run.
+     * </pre>
+     *
+     * <code>bool reset_disk_byte_usage = 11;</code>
+     */
+    public Builder setResetDiskByteUsage(boolean value) {
+      
+      resetDiskByteUsage_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Whether the total disk usage of tfdbg is to be reset to zero
+     * in this Session.run call. This is used by wrappers and hooks
+     * such as the local CLI ones to indicate that the dumped tensors
+     * are cleaned up from the disk after each Session.run.
+     * </pre>
+     *
+     * <code>bool reset_disk_byte_usage = 11;</code>
+     */
+    public Builder clearResetDiskByteUsage() {
+      
+      resetDiskByteUsage_ = false;
+      onChanged();
+      return this;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.DebugOptions)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.DebugOptions)
+  private static final org.tensorflow.proto.framework.DebugOptions DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.DebugOptions();
+  }
+
+  public static org.tensorflow.proto.framework.DebugOptions getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<DebugOptions>
+      PARSER = new com.google.protobuf.AbstractParser<DebugOptions>() {
+    @java.lang.Override
+    public DebugOptions parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new DebugOptions(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<DebugOptions> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<DebugOptions> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.framework.DebugOptions getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DebugOptionsOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DebugOptionsOrBuilder.java
new file mode 100644
index 00000000000..65f0407737a
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DebugOptionsOrBuilder.java
@@ -0,0 +1,76 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/debug.proto
+
+package org.tensorflow.proto.framework;
+
+public interface DebugOptionsOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.DebugOptions)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <pre>
+   * Debugging options
+   * </pre>
+   *
+   * <code>repeated .tensorflow.DebugTensorWatch debug_tensor_watch_opts = 4;</code>
+   */
+  java.util.List<org.tensorflow.proto.framework.DebugTensorWatch> 
+      getDebugTensorWatchOptsList();
+  /**
+   * <pre>
+   * Debugging options
+   * </pre>
+   *
+   * <code>repeated .tensorflow.DebugTensorWatch debug_tensor_watch_opts = 4;</code>
+   */
+  org.tensorflow.proto.framework.DebugTensorWatch getDebugTensorWatchOpts(int index);
+  /**
+   * <pre>
+   * Debugging options
+   * </pre>
+   *
+   * <code>repeated .tensorflow.DebugTensorWatch debug_tensor_watch_opts = 4;</code>
+   */
+  int getDebugTensorWatchOptsCount();
+  /**
+   * <pre>
+   * Debugging options
+   * </pre>
+   *
+   * <code>repeated .tensorflow.DebugTensorWatch debug_tensor_watch_opts = 4;</code>
+   */
+  java.util.List<? extends org.tensorflow.proto.framework.DebugTensorWatchOrBuilder> 
+      getDebugTensorWatchOptsOrBuilderList();
+  /**
+   * <pre>
+   * Debugging options
+   * </pre>
+   *
+   * <code>repeated .tensorflow.DebugTensorWatch debug_tensor_watch_opts = 4;</code>
+   */
+  org.tensorflow.proto.framework.DebugTensorWatchOrBuilder getDebugTensorWatchOptsOrBuilder(
+      int index);
+
+  /**
+   * <pre>
+   * Caller-specified global step count.
+   * Note that this is distinct from the session run count and the executor
+   * step count.
+   * </pre>
+   *
+   * <code>int64 global_step = 10;</code>
+   */
+  long getGlobalStep();
+
+  /**
+   * <pre>
+   * Whether the total disk usage of tfdbg is to be reset to zero
+   * in this Session.run call. This is used by wrappers and hooks
+   * such as the local CLI ones to indicate that the dumped tensors
+   * are cleaned up from the disk after each Session.run.
+   * </pre>
+   *
+   * <code>bool reset_disk_byte_usage = 11;</code>
+   */
+  boolean getResetDiskByteUsage();
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DebugProtos.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DebugProtos.java
new file mode 100644
index 00000000000..edc853e9376
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DebugProtos.java
@@ -0,0 +1,95 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/debug.proto
+
+package org.tensorflow.proto.framework;
+
+public final class DebugProtos {
+  private DebugProtos() {}
+  public static void registerAllExtensions(
+      com.google.protobuf.ExtensionRegistryLite registry) {
+  }
+
+  public static void registerAllExtensions(
+      com.google.protobuf.ExtensionRegistry registry) {
+    registerAllExtensions(
+        (com.google.protobuf.ExtensionRegistryLite) registry);
+  }
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_DebugTensorWatch_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_DebugTensorWatch_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_DebugOptions_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_DebugOptions_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_DebuggedSourceFile_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_DebuggedSourceFile_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_DebuggedSourceFiles_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_DebuggedSourceFiles_fieldAccessorTable;
+
+  public static com.google.protobuf.Descriptors.FileDescriptor
+      getDescriptor() {
+    return descriptor;
+  }
+  private static  com.google.protobuf.Descriptors.FileDescriptor
+      descriptor;
+  static {
+    java.lang.String[] descriptorData = {
+      "\n$tensorflow/core/protobuf/debug.proto\022\n" +
+      "tensorflow\"\216\001\n\020DebugTensorWatch\022\021\n\tnode_" +
+      "name\030\001 \001(\t\022\023\n\013output_slot\030\002 \001(\005\022\021\n\tdebug" +
+      "_ops\030\003 \003(\t\022\022\n\ndebug_urls\030\004 \003(\t\022+\n#tolera" +
+      "te_debug_op_creation_failures\030\005 \001(\010\"\201\001\n\014" +
+      "DebugOptions\022=\n\027debug_tensor_watch_opts\030" +
+      "\004 \003(\0132\034.tensorflow.DebugTensorWatch\022\023\n\013g" +
+      "lobal_step\030\n \001(\003\022\035\n\025reset_disk_byte_usag" +
+      "e\030\013 \001(\010\"j\n\022DebuggedSourceFile\022\014\n\004host\030\001 " +
+      "\001(\t\022\021\n\tfile_path\030\002 \001(\t\022\025\n\rlast_modified\030" +
+      "\003 \001(\003\022\r\n\005bytes\030\004 \001(\003\022\r\n\005lines\030\005 \003(\t\"K\n\023D" +
+      "ebuggedSourceFiles\0224\n\014source_files\030\001 \003(\013" +
+      "2\036.tensorflow.DebuggedSourceFileB|\n\036org." +
+      "tensorflow.proto.frameworkB\013DebugProtosP" +
+      "\001ZHgithub.com/tensorflow/tensorflow/tens" +
+      "orflow/go/core/core_protos_go_proto\370\001\001b\006" +
+      "proto3"
+    };
+    descriptor = com.google.protobuf.Descriptors.FileDescriptor
+      .internalBuildGeneratedFileFrom(descriptorData,
+        new com.google.protobuf.Descriptors.FileDescriptor[] {
+        });
+    internal_static_tensorflow_DebugTensorWatch_descriptor =
+      getDescriptor().getMessageTypes().get(0);
+    internal_static_tensorflow_DebugTensorWatch_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_DebugTensorWatch_descriptor,
+        new java.lang.String[] { "NodeName", "OutputSlot", "DebugOps", "DebugUrls", "TolerateDebugOpCreationFailures", });
+    internal_static_tensorflow_DebugOptions_descriptor =
+      getDescriptor().getMessageTypes().get(1);
+    internal_static_tensorflow_DebugOptions_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_DebugOptions_descriptor,
+        new java.lang.String[] { "DebugTensorWatchOpts", "GlobalStep", "ResetDiskByteUsage", });
+    internal_static_tensorflow_DebuggedSourceFile_descriptor =
+      getDescriptor().getMessageTypes().get(2);
+    internal_static_tensorflow_DebuggedSourceFile_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_DebuggedSourceFile_descriptor,
+        new java.lang.String[] { "Host", "FilePath", "LastModified", "Bytes", "Lines", });
+    internal_static_tensorflow_DebuggedSourceFiles_descriptor =
+      getDescriptor().getMessageTypes().get(3);
+    internal_static_tensorflow_DebuggedSourceFiles_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_DebuggedSourceFiles_descriptor,
+        new java.lang.String[] { "SourceFiles", });
+  }
+
+  // @@protoc_insertion_point(outer_class_scope)
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DebugTensorWatch.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DebugTensorWatch.java
new file mode 100644
index 00000000000..6536f3402fc
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DebugTensorWatch.java
@@ -0,0 +1,1444 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/debug.proto
+
+package org.tensorflow.proto.framework;
+
+/**
+ * <pre>
+ * Option for watching a node in TensorFlow Debugger (tfdbg).
+ * </pre>
+ *
+ * Protobuf type {@code tensorflow.DebugTensorWatch}
+ */
+public  final class DebugTensorWatch extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.DebugTensorWatch)
+    DebugTensorWatchOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use DebugTensorWatch.newBuilder() to construct.
+  private DebugTensorWatch(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private DebugTensorWatch() {
+    nodeName_ = "";
+    debugOps_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+    debugUrls_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new DebugTensorWatch();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private DebugTensorWatch(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    int mutable_bitField0_ = 0;
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 10: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            nodeName_ = s;
+            break;
+          }
+          case 16: {
+
+            outputSlot_ = input.readInt32();
+            break;
+          }
+          case 26: {
+            java.lang.String s = input.readStringRequireUtf8();
+            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
+              debugOps_ = new com.google.protobuf.LazyStringArrayList();
+              mutable_bitField0_ |= 0x00000001;
+            }
+            debugOps_.add(s);
+            break;
+          }
+          case 34: {
+            java.lang.String s = input.readStringRequireUtf8();
+            if (!((mutable_bitField0_ & 0x00000002) != 0)) {
+              debugUrls_ = new com.google.protobuf.LazyStringArrayList();
+              mutable_bitField0_ |= 0x00000002;
+            }
+            debugUrls_.add(s);
+            break;
+          }
+          case 40: {
+
+            tolerateDebugOpCreationFailures_ = input.readBool();
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      if (((mutable_bitField0_ & 0x00000001) != 0)) {
+        debugOps_ = debugOps_.getUnmodifiableView();
+      }
+      if (((mutable_bitField0_ & 0x00000002) != 0)) {
+        debugUrls_ = debugUrls_.getUnmodifiableView();
+      }
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.framework.DebugProtos.internal_static_tensorflow_DebugTensorWatch_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.framework.DebugProtos.internal_static_tensorflow_DebugTensorWatch_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.framework.DebugTensorWatch.class, org.tensorflow.proto.framework.DebugTensorWatch.Builder.class);
+  }
+
+  public static final int NODE_NAME_FIELD_NUMBER = 1;
+  private volatile java.lang.Object nodeName_;
+  /**
+   * <pre>
+   * Name of the node to watch.
+   * Use "*" for wildcard. But note: currently, regex is not supported in
+   * general.
+   * </pre>
+   *
+   * <code>string node_name = 1;</code>
+   */
+  public java.lang.String getNodeName() {
+    java.lang.Object ref = nodeName_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      nodeName_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * Name of the node to watch.
+   * Use "*" for wildcard. But note: currently, regex is not supported in
+   * general.
+   * </pre>
+   *
+   * <code>string node_name = 1;</code>
+   */
+  public com.google.protobuf.ByteString
+      getNodeNameBytes() {
+    java.lang.Object ref = nodeName_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      nodeName_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int OUTPUT_SLOT_FIELD_NUMBER = 2;
+  private int outputSlot_;
+  /**
+   * <pre>
+   * Output slot to watch.
+   * The semantics of output_slot == -1 is that all outputs of the node
+   * will be watched (i.e., a wildcard).
+   * Other negative values of output_slot are invalid and will lead to
+   * errors currently.
+   * </pre>
+   *
+   * <code>int32 output_slot = 2;</code>
+   */
+  public int getOutputSlot() {
+    return outputSlot_;
+  }
+
+  public static final int DEBUG_OPS_FIELD_NUMBER = 3;
+  private com.google.protobuf.LazyStringList debugOps_;
+  /**
+   * <pre>
+   * Name(s) of the debugging op(s).
+   * One or more than one probes on a tensor.
+   * e.g., {"DebugIdentity", "DebugNanCount"}
+   * </pre>
+   *
+   * <code>repeated string debug_ops = 3;</code>
+   */
+  public com.google.protobuf.ProtocolStringList
+      getDebugOpsList() {
+    return debugOps_;
+  }
+  /**
+   * <pre>
+   * Name(s) of the debugging op(s).
+   * One or more than one probes on a tensor.
+   * e.g., {"DebugIdentity", "DebugNanCount"}
+   * </pre>
+   *
+   * <code>repeated string debug_ops = 3;</code>
+   */
+  public int getDebugOpsCount() {
+    return debugOps_.size();
+  }
+  /**
+   * <pre>
+   * Name(s) of the debugging op(s).
+   * One or more than one probes on a tensor.
+   * e.g., {"DebugIdentity", "DebugNanCount"}
+   * </pre>
+   *
+   * <code>repeated string debug_ops = 3;</code>
+   */
+  public java.lang.String getDebugOps(int index) {
+    return debugOps_.get(index);
+  }
+  /**
+   * <pre>
+   * Name(s) of the debugging op(s).
+   * One or more than one probes on a tensor.
+   * e.g., {"DebugIdentity", "DebugNanCount"}
+   * </pre>
+   *
+   * <code>repeated string debug_ops = 3;</code>
+   */
+  public com.google.protobuf.ByteString
+      getDebugOpsBytes(int index) {
+    return debugOps_.getByteString(index);
+  }
+
+  public static final int DEBUG_URLS_FIELD_NUMBER = 4;
+  private com.google.protobuf.LazyStringList debugUrls_;
+  /**
+   * <pre>
+   * URL(s) for debug targets(s).
+   * Supported URL formats are:
+   *   - file:///foo/tfdbg_dump: Writes out Event content to file
+   *     /foo/tfdbg_dump.  Assumes all directories can be created if they don't
+   *     already exist.
+   *   - grpc://localhost:11011: Sends an RPC request to an EventListener
+   *     service running at localhost:11011 with the event.
+   *   - memcbk:///event_key: Routes tensors to clients using the
+   *     callback registered with the DebugCallbackRegistry for event_key.
+   * Each debug op listed in debug_ops will publish its output tensor (debug
+   * signal) to all URLs in debug_urls.
+   * N.B. Session::Run() supports concurrent invocations of the same inputs
+   * (feed keys), outputs and target nodes. If such concurrent invocations
+   * are to be debugged, the callers of Session::Run() must use distinct
+   * debug_urls to make sure that the streamed or dumped events do not overlap
+   * among the invocations.
+   * TODO(cais): More visible documentation of this in g3docs.
+   * </pre>
+   *
+   * <code>repeated string debug_urls = 4;</code>
+   */
+  public com.google.protobuf.ProtocolStringList
+      getDebugUrlsList() {
+    return debugUrls_;
+  }
+  /**
+   * <pre>
+   * URL(s) for debug targets(s).
+   * Supported URL formats are:
+   *   - file:///foo/tfdbg_dump: Writes out Event content to file
+   *     /foo/tfdbg_dump.  Assumes all directories can be created if they don't
+   *     already exist.
+   *   - grpc://localhost:11011: Sends an RPC request to an EventListener
+   *     service running at localhost:11011 with the event.
+   *   - memcbk:///event_key: Routes tensors to clients using the
+   *     callback registered with the DebugCallbackRegistry for event_key.
+   * Each debug op listed in debug_ops will publish its output tensor (debug
+   * signal) to all URLs in debug_urls.
+   * N.B. Session::Run() supports concurrent invocations of the same inputs
+   * (feed keys), outputs and target nodes. If such concurrent invocations
+   * are to be debugged, the callers of Session::Run() must use distinct
+   * debug_urls to make sure that the streamed or dumped events do not overlap
+   * among the invocations.
+   * TODO(cais): More visible documentation of this in g3docs.
+   * </pre>
+   *
+   * <code>repeated string debug_urls = 4;</code>
+   */
+  public int getDebugUrlsCount() {
+    return debugUrls_.size();
+  }
+  /**
+   * <pre>
+   * URL(s) for debug targets(s).
+   * Supported URL formats are:
+   *   - file:///foo/tfdbg_dump: Writes out Event content to file
+   *     /foo/tfdbg_dump.  Assumes all directories can be created if they don't
+   *     already exist.
+   *   - grpc://localhost:11011: Sends an RPC request to an EventListener
+   *     service running at localhost:11011 with the event.
+   *   - memcbk:///event_key: Routes tensors to clients using the
+   *     callback registered with the DebugCallbackRegistry for event_key.
+   * Each debug op listed in debug_ops will publish its output tensor (debug
+   * signal) to all URLs in debug_urls.
+   * N.B. Session::Run() supports concurrent invocations of the same inputs
+   * (feed keys), outputs and target nodes. If such concurrent invocations
+   * are to be debugged, the callers of Session::Run() must use distinct
+   * debug_urls to make sure that the streamed or dumped events do not overlap
+   * among the invocations.
+   * TODO(cais): More visible documentation of this in g3docs.
+   * </pre>
+   *
+   * <code>repeated string debug_urls = 4;</code>
+   */
+  public java.lang.String getDebugUrls(int index) {
+    return debugUrls_.get(index);
+  }
+  /**
+   * <pre>
+   * URL(s) for debug targets(s).
+   * Supported URL formats are:
+   *   - file:///foo/tfdbg_dump: Writes out Event content to file
+   *     /foo/tfdbg_dump.  Assumes all directories can be created if they don't
+   *     already exist.
+   *   - grpc://localhost:11011: Sends an RPC request to an EventListener
+   *     service running at localhost:11011 with the event.
+   *   - memcbk:///event_key: Routes tensors to clients using the
+   *     callback registered with the DebugCallbackRegistry for event_key.
+   * Each debug op listed in debug_ops will publish its output tensor (debug
+   * signal) to all URLs in debug_urls.
+   * N.B. Session::Run() supports concurrent invocations of the same inputs
+   * (feed keys), outputs and target nodes. If such concurrent invocations
+   * are to be debugged, the callers of Session::Run() must use distinct
+   * debug_urls to make sure that the streamed or dumped events do not overlap
+   * among the invocations.
+   * TODO(cais): More visible documentation of this in g3docs.
+   * </pre>
+   *
+   * <code>repeated string debug_urls = 4;</code>
+   */
+  public com.google.protobuf.ByteString
+      getDebugUrlsBytes(int index) {
+    return debugUrls_.getByteString(index);
+  }
+
+  public static final int TOLERATE_DEBUG_OP_CREATION_FAILURES_FIELD_NUMBER = 5;
+  private boolean tolerateDebugOpCreationFailures_;
+  /**
+   * <pre>
+   * Do not error out if debug op creation fails (e.g., due to dtype
+   * incompatibility). Instead, just log the failure.
+   * </pre>
+   *
+   * <code>bool tolerate_debug_op_creation_failures = 5;</code>
+   */
+  public boolean getTolerateDebugOpCreationFailures() {
+    return tolerateDebugOpCreationFailures_;
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    if (!getNodeNameBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 1, nodeName_);
+    }
+    if (outputSlot_ != 0) {
+      output.writeInt32(2, outputSlot_);
+    }
+    for (int i = 0; i < debugOps_.size(); i++) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 3, debugOps_.getRaw(i));
+    }
+    for (int i = 0; i < debugUrls_.size(); i++) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 4, debugUrls_.getRaw(i));
+    }
+    if (tolerateDebugOpCreationFailures_ != false) {
+      output.writeBool(5, tolerateDebugOpCreationFailures_);
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (!getNodeNameBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, nodeName_);
+    }
+    if (outputSlot_ != 0) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt32Size(2, outputSlot_);
+    }
+    {
+      int dataSize = 0;
+      for (int i = 0; i < debugOps_.size(); i++) {
+        dataSize += computeStringSizeNoTag(debugOps_.getRaw(i));
+      }
+      size += dataSize;
+      size += 1 * getDebugOpsList().size();
+    }
+    {
+      int dataSize = 0;
+      for (int i = 0; i < debugUrls_.size(); i++) {
+        dataSize += computeStringSizeNoTag(debugUrls_.getRaw(i));
+      }
+      size += dataSize;
+      size += 1 * getDebugUrlsList().size();
+    }
+    if (tolerateDebugOpCreationFailures_ != false) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeBoolSize(5, tolerateDebugOpCreationFailures_);
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.framework.DebugTensorWatch)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.framework.DebugTensorWatch other = (org.tensorflow.proto.framework.DebugTensorWatch) obj;
+
+    if (!getNodeName()
+        .equals(other.getNodeName())) return false;
+    if (getOutputSlot()
+        != other.getOutputSlot()) return false;
+    if (!getDebugOpsList()
+        .equals(other.getDebugOpsList())) return false;
+    if (!getDebugUrlsList()
+        .equals(other.getDebugUrlsList())) return false;
+    if (getTolerateDebugOpCreationFailures()
+        != other.getTolerateDebugOpCreationFailures()) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    hash = (37 * hash) + NODE_NAME_FIELD_NUMBER;
+    hash = (53 * hash) + getNodeName().hashCode();
+    hash = (37 * hash) + OUTPUT_SLOT_FIELD_NUMBER;
+    hash = (53 * hash) + getOutputSlot();
+    if (getDebugOpsCount() > 0) {
+      hash = (37 * hash) + DEBUG_OPS_FIELD_NUMBER;
+      hash = (53 * hash) + getDebugOpsList().hashCode();
+    }
+    if (getDebugUrlsCount() > 0) {
+      hash = (37 * hash) + DEBUG_URLS_FIELD_NUMBER;
+      hash = (53 * hash) + getDebugUrlsList().hashCode();
+    }
+    hash = (37 * hash) + TOLERATE_DEBUG_OP_CREATION_FAILURES_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashBoolean(
+        getTolerateDebugOpCreationFailures());
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.framework.DebugTensorWatch parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.DebugTensorWatch parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.DebugTensorWatch parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.DebugTensorWatch parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.DebugTensorWatch parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.DebugTensorWatch parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.DebugTensorWatch parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.DebugTensorWatch parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.DebugTensorWatch parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.DebugTensorWatch parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.DebugTensorWatch parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.DebugTensorWatch parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.framework.DebugTensorWatch prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * <pre>
+   * Option for watching a node in TensorFlow Debugger (tfdbg).
+   * </pre>
+   *
+   * Protobuf type {@code tensorflow.DebugTensorWatch}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.DebugTensorWatch)
+      org.tensorflow.proto.framework.DebugTensorWatchOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.DebugProtos.internal_static_tensorflow_DebugTensorWatch_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.DebugProtos.internal_static_tensorflow_DebugTensorWatch_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.DebugTensorWatch.class, org.tensorflow.proto.framework.DebugTensorWatch.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.framework.DebugTensorWatch.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      nodeName_ = "";
+
+      outputSlot_ = 0;
+
+      debugOps_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+      bitField0_ = (bitField0_ & ~0x00000001);
+      debugUrls_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+      bitField0_ = (bitField0_ & ~0x00000002);
+      tolerateDebugOpCreationFailures_ = false;
+
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.framework.DebugProtos.internal_static_tensorflow_DebugTensorWatch_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.DebugTensorWatch getDefaultInstanceForType() {
+      return org.tensorflow.proto.framework.DebugTensorWatch.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.DebugTensorWatch build() {
+      org.tensorflow.proto.framework.DebugTensorWatch result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.DebugTensorWatch buildPartial() {
+      org.tensorflow.proto.framework.DebugTensorWatch result = new org.tensorflow.proto.framework.DebugTensorWatch(this);
+      int from_bitField0_ = bitField0_;
+      result.nodeName_ = nodeName_;
+      result.outputSlot_ = outputSlot_;
+      if (((bitField0_ & 0x00000001) != 0)) {
+        debugOps_ = debugOps_.getUnmodifiableView();
+        bitField0_ = (bitField0_ & ~0x00000001);
+      }
+      result.debugOps_ = debugOps_;
+      if (((bitField0_ & 0x00000002) != 0)) {
+        debugUrls_ = debugUrls_.getUnmodifiableView();
+        bitField0_ = (bitField0_ & ~0x00000002);
+      }
+      result.debugUrls_ = debugUrls_;
+      result.tolerateDebugOpCreationFailures_ = tolerateDebugOpCreationFailures_;
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.framework.DebugTensorWatch) {
+        return mergeFrom((org.tensorflow.proto.framework.DebugTensorWatch)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.framework.DebugTensorWatch other) {
+      if (other == org.tensorflow.proto.framework.DebugTensorWatch.getDefaultInstance()) return this;
+      if (!other.getNodeName().isEmpty()) {
+        nodeName_ = other.nodeName_;
+        onChanged();
+      }
+      if (other.getOutputSlot() != 0) {
+        setOutputSlot(other.getOutputSlot());
+      }
+      if (!other.debugOps_.isEmpty()) {
+        if (debugOps_.isEmpty()) {
+          debugOps_ = other.debugOps_;
+          bitField0_ = (bitField0_ & ~0x00000001);
+        } else {
+          ensureDebugOpsIsMutable();
+          debugOps_.addAll(other.debugOps_);
+        }
+        onChanged();
+      }
+      if (!other.debugUrls_.isEmpty()) {
+        if (debugUrls_.isEmpty()) {
+          debugUrls_ = other.debugUrls_;
+          bitField0_ = (bitField0_ & ~0x00000002);
+        } else {
+          ensureDebugUrlsIsMutable();
+          debugUrls_.addAll(other.debugUrls_);
+        }
+        onChanged();
+      }
+      if (other.getTolerateDebugOpCreationFailures() != false) {
+        setTolerateDebugOpCreationFailures(other.getTolerateDebugOpCreationFailures());
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.framework.DebugTensorWatch parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.framework.DebugTensorWatch) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+    private int bitField0_;
+
+    private java.lang.Object nodeName_ = "";
+    /**
+     * <pre>
+     * Name of the node to watch.
+     * Use "*" for wildcard. But note: currently, regex is not supported in
+     * general.
+     * </pre>
+     *
+     * <code>string node_name = 1;</code>
+     */
+    public java.lang.String getNodeName() {
+      java.lang.Object ref = nodeName_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        nodeName_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Name of the node to watch.
+     * Use "*" for wildcard. But note: currently, regex is not supported in
+     * general.
+     * </pre>
+     *
+     * <code>string node_name = 1;</code>
+     */
+    public com.google.protobuf.ByteString
+        getNodeNameBytes() {
+      java.lang.Object ref = nodeName_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        nodeName_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Name of the node to watch.
+     * Use "*" for wildcard. But note: currently, regex is not supported in
+     * general.
+     * </pre>
+     *
+     * <code>string node_name = 1;</code>
+     */
+    public Builder setNodeName(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      nodeName_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Name of the node to watch.
+     * Use "*" for wildcard. But note: currently, regex is not supported in
+     * general.
+     * </pre>
+     *
+     * <code>string node_name = 1;</code>
+     */
+    public Builder clearNodeName() {
+      
+      nodeName_ = getDefaultInstance().getNodeName();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Name of the node to watch.
+     * Use "*" for wildcard. But note: currently, regex is not supported in
+     * general.
+     * </pre>
+     *
+     * <code>string node_name = 1;</code>
+     */
+    public Builder setNodeNameBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      nodeName_ = value;
+      onChanged();
+      return this;
+    }
+
+    private int outputSlot_ ;
+    /**
+     * <pre>
+     * Output slot to watch.
+     * The semantics of output_slot == -1 is that all outputs of the node
+     * will be watched (i.e., a wildcard).
+     * Other negative values of output_slot are invalid and will lead to
+     * errors currently.
+     * </pre>
+     *
+     * <code>int32 output_slot = 2;</code>
+     */
+    public int getOutputSlot() {
+      return outputSlot_;
+    }
+    /**
+     * <pre>
+     * Output slot to watch.
+     * The semantics of output_slot == -1 is that all outputs of the node
+     * will be watched (i.e., a wildcard).
+     * Other negative values of output_slot are invalid and will lead to
+     * errors currently.
+     * </pre>
+     *
+     * <code>int32 output_slot = 2;</code>
+     */
+    public Builder setOutputSlot(int value) {
+      
+      outputSlot_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Output slot to watch.
+     * The semantics of output_slot == -1 is that all outputs of the node
+     * will be watched (i.e., a wildcard).
+     * Other negative values of output_slot are invalid and will lead to
+     * errors currently.
+     * </pre>
+     *
+     * <code>int32 output_slot = 2;</code>
+     */
+    public Builder clearOutputSlot() {
+      
+      outputSlot_ = 0;
+      onChanged();
+      return this;
+    }
+
+    private com.google.protobuf.LazyStringList debugOps_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+    private void ensureDebugOpsIsMutable() {
+      if (!((bitField0_ & 0x00000001) != 0)) {
+        debugOps_ = new com.google.protobuf.LazyStringArrayList(debugOps_);
+        bitField0_ |= 0x00000001;
+       }
+    }
+    /**
+     * <pre>
+     * Name(s) of the debugging op(s).
+     * One or more than one probes on a tensor.
+     * e.g., {"DebugIdentity", "DebugNanCount"}
+     * </pre>
+     *
+     * <code>repeated string debug_ops = 3;</code>
+     */
+    public com.google.protobuf.ProtocolStringList
+        getDebugOpsList() {
+      return debugOps_.getUnmodifiableView();
+    }
+    /**
+     * <pre>
+     * Name(s) of the debugging op(s).
+     * One or more than one probes on a tensor.
+     * e.g., {"DebugIdentity", "DebugNanCount"}
+     * </pre>
+     *
+     * <code>repeated string debug_ops = 3;</code>
+     */
+    public int getDebugOpsCount() {
+      return debugOps_.size();
+    }
+    /**
+     * <pre>
+     * Name(s) of the debugging op(s).
+     * One or more than one probes on a tensor.
+     * e.g., {"DebugIdentity", "DebugNanCount"}
+     * </pre>
+     *
+     * <code>repeated string debug_ops = 3;</code>
+     */
+    public java.lang.String getDebugOps(int index) {
+      return debugOps_.get(index);
+    }
+    /**
+     * <pre>
+     * Name(s) of the debugging op(s).
+     * One or more than one probes on a tensor.
+     * e.g., {"DebugIdentity", "DebugNanCount"}
+     * </pre>
+     *
+     * <code>repeated string debug_ops = 3;</code>
+     */
+    public com.google.protobuf.ByteString
+        getDebugOpsBytes(int index) {
+      return debugOps_.getByteString(index);
+    }
+    /**
+     * <pre>
+     * Name(s) of the debugging op(s).
+     * One or more than one probes on a tensor.
+     * e.g., {"DebugIdentity", "DebugNanCount"}
+     * </pre>
+     *
+     * <code>repeated string debug_ops = 3;</code>
+     */
+    public Builder setDebugOps(
+        int index, java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  ensureDebugOpsIsMutable();
+      debugOps_.set(index, value);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Name(s) of the debugging op(s).
+     * One or more than one probes on a tensor.
+     * e.g., {"DebugIdentity", "DebugNanCount"}
+     * </pre>
+     *
+     * <code>repeated string debug_ops = 3;</code>
+     */
+    public Builder addDebugOps(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  ensureDebugOpsIsMutable();
+      debugOps_.add(value);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Name(s) of the debugging op(s).
+     * One or more than one probes on a tensor.
+     * e.g., {"DebugIdentity", "DebugNanCount"}
+     * </pre>
+     *
+     * <code>repeated string debug_ops = 3;</code>
+     */
+    public Builder addAllDebugOps(
+        java.lang.Iterable<java.lang.String> values) {
+      ensureDebugOpsIsMutable();
+      com.google.protobuf.AbstractMessageLite.Builder.addAll(
+          values, debugOps_);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Name(s) of the debugging op(s).
+     * One or more than one probes on a tensor.
+     * e.g., {"DebugIdentity", "DebugNanCount"}
+     * </pre>
+     *
+     * <code>repeated string debug_ops = 3;</code>
+     */
+    public Builder clearDebugOps() {
+      debugOps_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+      bitField0_ = (bitField0_ & ~0x00000001);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Name(s) of the debugging op(s).
+     * One or more than one probes on a tensor.
+     * e.g., {"DebugIdentity", "DebugNanCount"}
+     * </pre>
+     *
+     * <code>repeated string debug_ops = 3;</code>
+     */
+    public Builder addDebugOpsBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      ensureDebugOpsIsMutable();
+      debugOps_.add(value);
+      onChanged();
+      return this;
+    }
+
+    private com.google.protobuf.LazyStringList debugUrls_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+    private void ensureDebugUrlsIsMutable() {
+      if (!((bitField0_ & 0x00000002) != 0)) {
+        debugUrls_ = new com.google.protobuf.LazyStringArrayList(debugUrls_);
+        bitField0_ |= 0x00000002;
+       }
+    }
+    /**
+     * <pre>
+     * URL(s) for debug targets(s).
+     * Supported URL formats are:
+     *   - file:///foo/tfdbg_dump: Writes out Event content to file
+     *     /foo/tfdbg_dump.  Assumes all directories can be created if they don't
+     *     already exist.
+     *   - grpc://localhost:11011: Sends an RPC request to an EventListener
+     *     service running at localhost:11011 with the event.
+     *   - memcbk:///event_key: Routes tensors to clients using the
+     *     callback registered with the DebugCallbackRegistry for event_key.
+     * Each debug op listed in debug_ops will publish its output tensor (debug
+     * signal) to all URLs in debug_urls.
+     * N.B. Session::Run() supports concurrent invocations of the same inputs
+     * (feed keys), outputs and target nodes. If such concurrent invocations
+     * are to be debugged, the callers of Session::Run() must use distinct
+     * debug_urls to make sure that the streamed or dumped events do not overlap
+     * among the invocations.
+     * TODO(cais): More visible documentation of this in g3docs.
+     * </pre>
+     *
+     * <code>repeated string debug_urls = 4;</code>
+     */
+    public com.google.protobuf.ProtocolStringList
+        getDebugUrlsList() {
+      return debugUrls_.getUnmodifiableView();
+    }
+    /**
+     * <pre>
+     * URL(s) for debug targets(s).
+     * Supported URL formats are:
+     *   - file:///foo/tfdbg_dump: Writes out Event content to file
+     *     /foo/tfdbg_dump.  Assumes all directories can be created if they don't
+     *     already exist.
+     *   - grpc://localhost:11011: Sends an RPC request to an EventListener
+     *     service running at localhost:11011 with the event.
+     *   - memcbk:///event_key: Routes tensors to clients using the
+     *     callback registered with the DebugCallbackRegistry for event_key.
+     * Each debug op listed in debug_ops will publish its output tensor (debug
+     * signal) to all URLs in debug_urls.
+     * N.B. Session::Run() supports concurrent invocations of the same inputs
+     * (feed keys), outputs and target nodes. If such concurrent invocations
+     * are to be debugged, the callers of Session::Run() must use distinct
+     * debug_urls to make sure that the streamed or dumped events do not overlap
+     * among the invocations.
+     * TODO(cais): More visible documentation of this in g3docs.
+     * </pre>
+     *
+     * <code>repeated string debug_urls = 4;</code>
+     */
+    public int getDebugUrlsCount() {
+      return debugUrls_.size();
+    }
+    /**
+     * <pre>
+     * URL(s) for debug targets(s).
+     * Supported URL formats are:
+     *   - file:///foo/tfdbg_dump: Writes out Event content to file
+     *     /foo/tfdbg_dump.  Assumes all directories can be created if they don't
+     *     already exist.
+     *   - grpc://localhost:11011: Sends an RPC request to an EventListener
+     *     service running at localhost:11011 with the event.
+     *   - memcbk:///event_key: Routes tensors to clients using the
+     *     callback registered with the DebugCallbackRegistry for event_key.
+     * Each debug op listed in debug_ops will publish its output tensor (debug
+     * signal) to all URLs in debug_urls.
+     * N.B. Session::Run() supports concurrent invocations of the same inputs
+     * (feed keys), outputs and target nodes. If such concurrent invocations
+     * are to be debugged, the callers of Session::Run() must use distinct
+     * debug_urls to make sure that the streamed or dumped events do not overlap
+     * among the invocations.
+     * TODO(cais): More visible documentation of this in g3docs.
+     * </pre>
+     *
+     * <code>repeated string debug_urls = 4;</code>
+     */
+    public java.lang.String getDebugUrls(int index) {
+      return debugUrls_.get(index);
+    }
+    /**
+     * <pre>
+     * URL(s) for debug targets(s).
+     * Supported URL formats are:
+     *   - file:///foo/tfdbg_dump: Writes out Event content to file
+     *     /foo/tfdbg_dump.  Assumes all directories can be created if they don't
+     *     already exist.
+     *   - grpc://localhost:11011: Sends an RPC request to an EventListener
+     *     service running at localhost:11011 with the event.
+     *   - memcbk:///event_key: Routes tensors to clients using the
+     *     callback registered with the DebugCallbackRegistry for event_key.
+     * Each debug op listed in debug_ops will publish its output tensor (debug
+     * signal) to all URLs in debug_urls.
+     * N.B. Session::Run() supports concurrent invocations of the same inputs
+     * (feed keys), outputs and target nodes. If such concurrent invocations
+     * are to be debugged, the callers of Session::Run() must use distinct
+     * debug_urls to make sure that the streamed or dumped events do not overlap
+     * among the invocations.
+     * TODO(cais): More visible documentation of this in g3docs.
+     * </pre>
+     *
+     * <code>repeated string debug_urls = 4;</code>
+     */
+    public com.google.protobuf.ByteString
+        getDebugUrlsBytes(int index) {
+      return debugUrls_.getByteString(index);
+    }
+    /**
+     * <pre>
+     * URL(s) for debug targets(s).
+     * Supported URL formats are:
+     *   - file:///foo/tfdbg_dump: Writes out Event content to file
+     *     /foo/tfdbg_dump.  Assumes all directories can be created if they don't
+     *     already exist.
+     *   - grpc://localhost:11011: Sends an RPC request to an EventListener
+     *     service running at localhost:11011 with the event.
+     *   - memcbk:///event_key: Routes tensors to clients using the
+     *     callback registered with the DebugCallbackRegistry for event_key.
+     * Each debug op listed in debug_ops will publish its output tensor (debug
+     * signal) to all URLs in debug_urls.
+     * N.B. Session::Run() supports concurrent invocations of the same inputs
+     * (feed keys), outputs and target nodes. If such concurrent invocations
+     * are to be debugged, the callers of Session::Run() must use distinct
+     * debug_urls to make sure that the streamed or dumped events do not overlap
+     * among the invocations.
+     * TODO(cais): More visible documentation of this in g3docs.
+     * </pre>
+     *
+     * <code>repeated string debug_urls = 4;</code>
+     */
+    public Builder setDebugUrls(
+        int index, java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  ensureDebugUrlsIsMutable();
+      debugUrls_.set(index, value);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * URL(s) for debug targets(s).
+     * Supported URL formats are:
+     *   - file:///foo/tfdbg_dump: Writes out Event content to file
+     *     /foo/tfdbg_dump.  Assumes all directories can be created if they don't
+     *     already exist.
+     *   - grpc://localhost:11011: Sends an RPC request to an EventListener
+     *     service running at localhost:11011 with the event.
+     *   - memcbk:///event_key: Routes tensors to clients using the
+     *     callback registered with the DebugCallbackRegistry for event_key.
+     * Each debug op listed in debug_ops will publish its output tensor (debug
+     * signal) to all URLs in debug_urls.
+     * N.B. Session::Run() supports concurrent invocations of the same inputs
+     * (feed keys), outputs and target nodes. If such concurrent invocations
+     * are to be debugged, the callers of Session::Run() must use distinct
+     * debug_urls to make sure that the streamed or dumped events do not overlap
+     * among the invocations.
+     * TODO(cais): More visible documentation of this in g3docs.
+     * </pre>
+     *
+     * <code>repeated string debug_urls = 4;</code>
+     */
+    public Builder addDebugUrls(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  ensureDebugUrlsIsMutable();
+      debugUrls_.add(value);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * URL(s) for debug targets(s).
+     * Supported URL formats are:
+     *   - file:///foo/tfdbg_dump: Writes out Event content to file
+     *     /foo/tfdbg_dump.  Assumes all directories can be created if they don't
+     *     already exist.
+     *   - grpc://localhost:11011: Sends an RPC request to an EventListener
+     *     service running at localhost:11011 with the event.
+     *   - memcbk:///event_key: Routes tensors to clients using the
+     *     callback registered with the DebugCallbackRegistry for event_key.
+     * Each debug op listed in debug_ops will publish its output tensor (debug
+     * signal) to all URLs in debug_urls.
+     * N.B. Session::Run() supports concurrent invocations of the same inputs
+     * (feed keys), outputs and target nodes. If such concurrent invocations
+     * are to be debugged, the callers of Session::Run() must use distinct
+     * debug_urls to make sure that the streamed or dumped events do not overlap
+     * among the invocations.
+     * TODO(cais): More visible documentation of this in g3docs.
+     * </pre>
+     *
+     * <code>repeated string debug_urls = 4;</code>
+     */
+    public Builder addAllDebugUrls(
+        java.lang.Iterable<java.lang.String> values) {
+      ensureDebugUrlsIsMutable();
+      com.google.protobuf.AbstractMessageLite.Builder.addAll(
+          values, debugUrls_);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * URL(s) for debug targets(s).
+     * Supported URL formats are:
+     *   - file:///foo/tfdbg_dump: Writes out Event content to file
+     *     /foo/tfdbg_dump.  Assumes all directories can be created if they don't
+     *     already exist.
+     *   - grpc://localhost:11011: Sends an RPC request to an EventListener
+     *     service running at localhost:11011 with the event.
+     *   - memcbk:///event_key: Routes tensors to clients using the
+     *     callback registered with the DebugCallbackRegistry for event_key.
+     * Each debug op listed in debug_ops will publish its output tensor (debug
+     * signal) to all URLs in debug_urls.
+     * N.B. Session::Run() supports concurrent invocations of the same inputs
+     * (feed keys), outputs and target nodes. If such concurrent invocations
+     * are to be debugged, the callers of Session::Run() must use distinct
+     * debug_urls to make sure that the streamed or dumped events do not overlap
+     * among the invocations.
+     * TODO(cais): More visible documentation of this in g3docs.
+     * </pre>
+     *
+     * <code>repeated string debug_urls = 4;</code>
+     */
+    public Builder clearDebugUrls() {
+      debugUrls_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+      bitField0_ = (bitField0_ & ~0x00000002);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * URL(s) for debug targets(s).
+     * Supported URL formats are:
+     *   - file:///foo/tfdbg_dump: Writes out Event content to file
+     *     /foo/tfdbg_dump.  Assumes all directories can be created if they don't
+     *     already exist.
+     *   - grpc://localhost:11011: Sends an RPC request to an EventListener
+     *     service running at localhost:11011 with the event.
+     *   - memcbk:///event_key: Routes tensors to clients using the
+     *     callback registered with the DebugCallbackRegistry for event_key.
+     * Each debug op listed in debug_ops will publish its output tensor (debug
+     * signal) to all URLs in debug_urls.
+     * N.B. Session::Run() supports concurrent invocations of the same inputs
+     * (feed keys), outputs and target nodes. If such concurrent invocations
+     * are to be debugged, the callers of Session::Run() must use distinct
+     * debug_urls to make sure that the streamed or dumped events do not overlap
+     * among the invocations.
+     * TODO(cais): More visible documentation of this in g3docs.
+     * </pre>
+     *
+     * <code>repeated string debug_urls = 4;</code>
+     */
+    public Builder addDebugUrlsBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      ensureDebugUrlsIsMutable();
+      debugUrls_.add(value);
+      onChanged();
+      return this;
+    }
+
+    private boolean tolerateDebugOpCreationFailures_ ;
+    /**
+     * <pre>
+     * Do not error out if debug op creation fails (e.g., due to dtype
+     * incompatibility). Instead, just log the failure.
+     * </pre>
+     *
+     * <code>bool tolerate_debug_op_creation_failures = 5;</code>
+     */
+    public boolean getTolerateDebugOpCreationFailures() {
+      return tolerateDebugOpCreationFailures_;
+    }
+    /**
+     * <pre>
+     * Do not error out if debug op creation fails (e.g., due to dtype
+     * incompatibility). Instead, just log the failure.
+     * </pre>
+     *
+     * <code>bool tolerate_debug_op_creation_failures = 5;</code>
+     */
+    public Builder setTolerateDebugOpCreationFailures(boolean value) {
+      
+      tolerateDebugOpCreationFailures_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Do not error out if debug op creation fails (e.g., due to dtype
+     * incompatibility). Instead, just log the failure.
+     * </pre>
+     *
+     * <code>bool tolerate_debug_op_creation_failures = 5;</code>
+     */
+    public Builder clearTolerateDebugOpCreationFailures() {
+      
+      tolerateDebugOpCreationFailures_ = false;
+      onChanged();
+      return this;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.DebugTensorWatch)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.DebugTensorWatch)
+  private static final org.tensorflow.proto.framework.DebugTensorWatch DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.DebugTensorWatch();
+  }
+
+  public static org.tensorflow.proto.framework.DebugTensorWatch getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<DebugTensorWatch>
+      PARSER = new com.google.protobuf.AbstractParser<DebugTensorWatch>() {
+    @java.lang.Override
+    public DebugTensorWatch parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new DebugTensorWatch(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<DebugTensorWatch> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<DebugTensorWatch> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.framework.DebugTensorWatch getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DebugTensorWatchOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DebugTensorWatchOrBuilder.java
new file mode 100644
index 00000000000..cce3536c5d1
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DebugTensorWatchOrBuilder.java
@@ -0,0 +1,196 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/debug.proto
+
+package org.tensorflow.proto.framework;
+
+public interface DebugTensorWatchOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.DebugTensorWatch)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <pre>
+   * Name of the node to watch.
+   * Use "*" for wildcard. But note: currently, regex is not supported in
+   * general.
+   * </pre>
+   *
+   * <code>string node_name = 1;</code>
+   */
+  java.lang.String getNodeName();
+  /**
+   * <pre>
+   * Name of the node to watch.
+   * Use "*" for wildcard. But note: currently, regex is not supported in
+   * general.
+   * </pre>
+   *
+   * <code>string node_name = 1;</code>
+   */
+  com.google.protobuf.ByteString
+      getNodeNameBytes();
+
+  /**
+   * <pre>
+   * Output slot to watch.
+   * The semantics of output_slot == -1 is that all outputs of the node
+   * will be watched (i.e., a wildcard).
+   * Other negative values of output_slot are invalid and will lead to
+   * errors currently.
+   * </pre>
+   *
+   * <code>int32 output_slot = 2;</code>
+   */
+  int getOutputSlot();
+
+  /**
+   * <pre>
+   * Name(s) of the debugging op(s).
+   * One or more than one probes on a tensor.
+   * e.g., {"DebugIdentity", "DebugNanCount"}
+   * </pre>
+   *
+   * <code>repeated string debug_ops = 3;</code>
+   */
+  java.util.List<java.lang.String>
+      getDebugOpsList();
+  /**
+   * <pre>
+   * Name(s) of the debugging op(s).
+   * One or more than one probes on a tensor.
+   * e.g., {"DebugIdentity", "DebugNanCount"}
+   * </pre>
+   *
+   * <code>repeated string debug_ops = 3;</code>
+   */
+  int getDebugOpsCount();
+  /**
+   * <pre>
+   * Name(s) of the debugging op(s).
+   * One or more than one probes on a tensor.
+   * e.g., {"DebugIdentity", "DebugNanCount"}
+   * </pre>
+   *
+   * <code>repeated string debug_ops = 3;</code>
+   */
+  java.lang.String getDebugOps(int index);
+  /**
+   * <pre>
+   * Name(s) of the debugging op(s).
+   * One or more than one probes on a tensor.
+   * e.g., {"DebugIdentity", "DebugNanCount"}
+   * </pre>
+   *
+   * <code>repeated string debug_ops = 3;</code>
+   */
+  com.google.protobuf.ByteString
+      getDebugOpsBytes(int index);
+
+  /**
+   * <pre>
+   * URL(s) for debug targets(s).
+   * Supported URL formats are:
+   *   - file:///foo/tfdbg_dump: Writes out Event content to file
+   *     /foo/tfdbg_dump.  Assumes all directories can be created if they don't
+   *     already exist.
+   *   - grpc://localhost:11011: Sends an RPC request to an EventListener
+   *     service running at localhost:11011 with the event.
+   *   - memcbk:///event_key: Routes tensors to clients using the
+   *     callback registered with the DebugCallbackRegistry for event_key.
+   * Each debug op listed in debug_ops will publish its output tensor (debug
+   * signal) to all URLs in debug_urls.
+   * N.B. Session::Run() supports concurrent invocations of the same inputs
+   * (feed keys), outputs and target nodes. If such concurrent invocations
+   * are to be debugged, the callers of Session::Run() must use distinct
+   * debug_urls to make sure that the streamed or dumped events do not overlap
+   * among the invocations.
+   * TODO(cais): More visible documentation of this in g3docs.
+   * </pre>
+   *
+   * <code>repeated string debug_urls = 4;</code>
+   */
+  java.util.List<java.lang.String>
+      getDebugUrlsList();
+  /**
+   * <pre>
+   * URL(s) for debug targets(s).
+   * Supported URL formats are:
+   *   - file:///foo/tfdbg_dump: Writes out Event content to file
+   *     /foo/tfdbg_dump.  Assumes all directories can be created if they don't
+   *     already exist.
+   *   - grpc://localhost:11011: Sends an RPC request to an EventListener
+   *     service running at localhost:11011 with the event.
+   *   - memcbk:///event_key: Routes tensors to clients using the
+   *     callback registered with the DebugCallbackRegistry for event_key.
+   * Each debug op listed in debug_ops will publish its output tensor (debug
+   * signal) to all URLs in debug_urls.
+   * N.B. Session::Run() supports concurrent invocations of the same inputs
+   * (feed keys), outputs and target nodes. If such concurrent invocations
+   * are to be debugged, the callers of Session::Run() must use distinct
+   * debug_urls to make sure that the streamed or dumped events do not overlap
+   * among the invocations.
+   * TODO(cais): More visible documentation of this in g3docs.
+   * </pre>
+   *
+   * <code>repeated string debug_urls = 4;</code>
+   */
+  int getDebugUrlsCount();
+  /**
+   * <pre>
+   * URL(s) for debug targets(s).
+   * Supported URL formats are:
+   *   - file:///foo/tfdbg_dump: Writes out Event content to file
+   *     /foo/tfdbg_dump.  Assumes all directories can be created if they don't
+   *     already exist.
+   *   - grpc://localhost:11011: Sends an RPC request to an EventListener
+   *     service running at localhost:11011 with the event.
+   *   - memcbk:///event_key: Routes tensors to clients using the
+   *     callback registered with the DebugCallbackRegistry for event_key.
+   * Each debug op listed in debug_ops will publish its output tensor (debug
+   * signal) to all URLs in debug_urls.
+   * N.B. Session::Run() supports concurrent invocations of the same inputs
+   * (feed keys), outputs and target nodes. If such concurrent invocations
+   * are to be debugged, the callers of Session::Run() must use distinct
+   * debug_urls to make sure that the streamed or dumped events do not overlap
+   * among the invocations.
+   * TODO(cais): More visible documentation of this in g3docs.
+   * </pre>
+   *
+   * <code>repeated string debug_urls = 4;</code>
+   */
+  java.lang.String getDebugUrls(int index);
+  /**
+   * <pre>
+   * URL(s) for debug targets(s).
+   * Supported URL formats are:
+   *   - file:///foo/tfdbg_dump: Writes out Event content to file
+   *     /foo/tfdbg_dump.  Assumes all directories can be created if they don't
+   *     already exist.
+   *   - grpc://localhost:11011: Sends an RPC request to an EventListener
+   *     service running at localhost:11011 with the event.
+   *   - memcbk:///event_key: Routes tensors to clients using the
+   *     callback registered with the DebugCallbackRegistry for event_key.
+   * Each debug op listed in debug_ops will publish its output tensor (debug
+   * signal) to all URLs in debug_urls.
+   * N.B. Session::Run() supports concurrent invocations of the same inputs
+   * (feed keys), outputs and target nodes. If such concurrent invocations
+   * are to be debugged, the callers of Session::Run() must use distinct
+   * debug_urls to make sure that the streamed or dumped events do not overlap
+   * among the invocations.
+   * TODO(cais): More visible documentation of this in g3docs.
+   * </pre>
+   *
+   * <code>repeated string debug_urls = 4;</code>
+   */
+  com.google.protobuf.ByteString
+      getDebugUrlsBytes(int index);
+
+  /**
+   * <pre>
+   * Do not error out if debug op creation fails (e.g., due to dtype
+   * incompatibility). Instead, just log the failure.
+   * </pre>
+   *
+   * <code>bool tolerate_debug_op_creation_failures = 5;</code>
+   */
+  boolean getTolerateDebugOpCreationFailures();
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DebuggedSourceFile.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DebuggedSourceFile.java
new file mode 100644
index 00000000000..d0752d9022b
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DebuggedSourceFile.java
@@ -0,0 +1,1102 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/debug.proto
+
+package org.tensorflow.proto.framework;
+
+/**
+ * Protobuf type {@code tensorflow.DebuggedSourceFile}
+ */
+public  final class DebuggedSourceFile extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.DebuggedSourceFile)
+    DebuggedSourceFileOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use DebuggedSourceFile.newBuilder() to construct.
+  private DebuggedSourceFile(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private DebuggedSourceFile() {
+    host_ = "";
+    filePath_ = "";
+    lines_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new DebuggedSourceFile();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private DebuggedSourceFile(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    int mutable_bitField0_ = 0;
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 10: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            host_ = s;
+            break;
+          }
+          case 18: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            filePath_ = s;
+            break;
+          }
+          case 24: {
+
+            lastModified_ = input.readInt64();
+            break;
+          }
+          case 32: {
+
+            bytes_ = input.readInt64();
+            break;
+          }
+          case 42: {
+            java.lang.String s = input.readStringRequireUtf8();
+            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
+              lines_ = new com.google.protobuf.LazyStringArrayList();
+              mutable_bitField0_ |= 0x00000001;
+            }
+            lines_.add(s);
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      if (((mutable_bitField0_ & 0x00000001) != 0)) {
+        lines_ = lines_.getUnmodifiableView();
+      }
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.framework.DebugProtos.internal_static_tensorflow_DebuggedSourceFile_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.framework.DebugProtos.internal_static_tensorflow_DebuggedSourceFile_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.framework.DebuggedSourceFile.class, org.tensorflow.proto.framework.DebuggedSourceFile.Builder.class);
+  }
+
+  public static final int HOST_FIELD_NUMBER = 1;
+  private volatile java.lang.Object host_;
+  /**
+   * <pre>
+   * The host name on which a source code file is located.
+   * </pre>
+   *
+   * <code>string host = 1;</code>
+   */
+  public java.lang.String getHost() {
+    java.lang.Object ref = host_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      host_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * The host name on which a source code file is located.
+   * </pre>
+   *
+   * <code>string host = 1;</code>
+   */
+  public com.google.protobuf.ByteString
+      getHostBytes() {
+    java.lang.Object ref = host_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      host_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int FILE_PATH_FIELD_NUMBER = 2;
+  private volatile java.lang.Object filePath_;
+  /**
+   * <pre>
+   * Path to the source code file.
+   * </pre>
+   *
+   * <code>string file_path = 2;</code>
+   */
+  public java.lang.String getFilePath() {
+    java.lang.Object ref = filePath_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      filePath_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * Path to the source code file.
+   * </pre>
+   *
+   * <code>string file_path = 2;</code>
+   */
+  public com.google.protobuf.ByteString
+      getFilePathBytes() {
+    java.lang.Object ref = filePath_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      filePath_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int LAST_MODIFIED_FIELD_NUMBER = 3;
+  private long lastModified_;
+  /**
+   * <pre>
+   * The timestamp at which the source code file is last modified.
+   * </pre>
+   *
+   * <code>int64 last_modified = 3;</code>
+   */
+  public long getLastModified() {
+    return lastModified_;
+  }
+
+  public static final int BYTES_FIELD_NUMBER = 4;
+  private long bytes_;
+  /**
+   * <pre>
+   * Byte size of the file.
+   * </pre>
+   *
+   * <code>int64 bytes = 4;</code>
+   */
+  public long getBytes() {
+    return bytes_;
+  }
+
+  public static final int LINES_FIELD_NUMBER = 5;
+  private com.google.protobuf.LazyStringList lines_;
+  /**
+   * <pre>
+   * Line-by-line content of the source code file.
+   * </pre>
+   *
+   * <code>repeated string lines = 5;</code>
+   */
+  public com.google.protobuf.ProtocolStringList
+      getLinesList() {
+    return lines_;
+  }
+  /**
+   * <pre>
+   * Line-by-line content of the source code file.
+   * </pre>
+   *
+   * <code>repeated string lines = 5;</code>
+   */
+  public int getLinesCount() {
+    return lines_.size();
+  }
+  /**
+   * <pre>
+   * Line-by-line content of the source code file.
+   * </pre>
+   *
+   * <code>repeated string lines = 5;</code>
+   */
+  public java.lang.String getLines(int index) {
+    return lines_.get(index);
+  }
+  /**
+   * <pre>
+   * Line-by-line content of the source code file.
+   * </pre>
+   *
+   * <code>repeated string lines = 5;</code>
+   */
+  public com.google.protobuf.ByteString
+      getLinesBytes(int index) {
+    return lines_.getByteString(index);
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    if (!getHostBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 1, host_);
+    }
+    if (!getFilePathBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 2, filePath_);
+    }
+    if (lastModified_ != 0L) {
+      output.writeInt64(3, lastModified_);
+    }
+    if (bytes_ != 0L) {
+      output.writeInt64(4, bytes_);
+    }
+    for (int i = 0; i < lines_.size(); i++) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 5, lines_.getRaw(i));
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (!getHostBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, host_);
+    }
+    if (!getFilePathBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(2, filePath_);
+    }
+    if (lastModified_ != 0L) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt64Size(3, lastModified_);
+    }
+    if (bytes_ != 0L) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt64Size(4, bytes_);
+    }
+    {
+      int dataSize = 0;
+      for (int i = 0; i < lines_.size(); i++) {
+        dataSize += computeStringSizeNoTag(lines_.getRaw(i));
+      }
+      size += dataSize;
+      size += 1 * getLinesList().size();
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.framework.DebuggedSourceFile)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.framework.DebuggedSourceFile other = (org.tensorflow.proto.framework.DebuggedSourceFile) obj;
+
+    if (!getHost()
+        .equals(other.getHost())) return false;
+    if (!getFilePath()
+        .equals(other.getFilePath())) return false;
+    if (getLastModified()
+        != other.getLastModified()) return false;
+    if (getBytes()
+        != other.getBytes()) return false;
+    if (!getLinesList()
+        .equals(other.getLinesList())) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    hash = (37 * hash) + HOST_FIELD_NUMBER;
+    hash = (53 * hash) + getHost().hashCode();
+    hash = (37 * hash) + FILE_PATH_FIELD_NUMBER;
+    hash = (53 * hash) + getFilePath().hashCode();
+    hash = (37 * hash) + LAST_MODIFIED_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+        getLastModified());
+    hash = (37 * hash) + BYTES_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+        getBytes());
+    if (getLinesCount() > 0) {
+      hash = (37 * hash) + LINES_FIELD_NUMBER;
+      hash = (53 * hash) + getLinesList().hashCode();
+    }
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.framework.DebuggedSourceFile parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.DebuggedSourceFile parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.DebuggedSourceFile parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.DebuggedSourceFile parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.DebuggedSourceFile parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.DebuggedSourceFile parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.DebuggedSourceFile parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.DebuggedSourceFile parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.DebuggedSourceFile parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.DebuggedSourceFile parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.DebuggedSourceFile parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.DebuggedSourceFile parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.framework.DebuggedSourceFile prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * Protobuf type {@code tensorflow.DebuggedSourceFile}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.DebuggedSourceFile)
+      org.tensorflow.proto.framework.DebuggedSourceFileOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.DebugProtos.internal_static_tensorflow_DebuggedSourceFile_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.DebugProtos.internal_static_tensorflow_DebuggedSourceFile_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.DebuggedSourceFile.class, org.tensorflow.proto.framework.DebuggedSourceFile.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.framework.DebuggedSourceFile.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      host_ = "";
+
+      filePath_ = "";
+
+      lastModified_ = 0L;
+
+      bytes_ = 0L;
+
+      lines_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+      bitField0_ = (bitField0_ & ~0x00000001);
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.framework.DebugProtos.internal_static_tensorflow_DebuggedSourceFile_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.DebuggedSourceFile getDefaultInstanceForType() {
+      return org.tensorflow.proto.framework.DebuggedSourceFile.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.DebuggedSourceFile build() {
+      org.tensorflow.proto.framework.DebuggedSourceFile result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.DebuggedSourceFile buildPartial() {
+      org.tensorflow.proto.framework.DebuggedSourceFile result = new org.tensorflow.proto.framework.DebuggedSourceFile(this);
+      int from_bitField0_ = bitField0_;
+      result.host_ = host_;
+      result.filePath_ = filePath_;
+      result.lastModified_ = lastModified_;
+      result.bytes_ = bytes_;
+      if (((bitField0_ & 0x00000001) != 0)) {
+        lines_ = lines_.getUnmodifiableView();
+        bitField0_ = (bitField0_ & ~0x00000001);
+      }
+      result.lines_ = lines_;
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.framework.DebuggedSourceFile) {
+        return mergeFrom((org.tensorflow.proto.framework.DebuggedSourceFile)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.framework.DebuggedSourceFile other) {
+      if (other == org.tensorflow.proto.framework.DebuggedSourceFile.getDefaultInstance()) return this;
+      if (!other.getHost().isEmpty()) {
+        host_ = other.host_;
+        onChanged();
+      }
+      if (!other.getFilePath().isEmpty()) {
+        filePath_ = other.filePath_;
+        onChanged();
+      }
+      if (other.getLastModified() != 0L) {
+        setLastModified(other.getLastModified());
+      }
+      if (other.getBytes() != 0L) {
+        setBytes(other.getBytes());
+      }
+      if (!other.lines_.isEmpty()) {
+        if (lines_.isEmpty()) {
+          lines_ = other.lines_;
+          bitField0_ = (bitField0_ & ~0x00000001);
+        } else {
+          ensureLinesIsMutable();
+          lines_.addAll(other.lines_);
+        }
+        onChanged();
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.framework.DebuggedSourceFile parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.framework.DebuggedSourceFile) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+    private int bitField0_;
+
+    private java.lang.Object host_ = "";
+    /**
+     * <pre>
+     * The host name on which a source code file is located.
+     * </pre>
+     *
+     * <code>string host = 1;</code>
+     */
+    public java.lang.String getHost() {
+      java.lang.Object ref = host_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        host_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * The host name on which a source code file is located.
+     * </pre>
+     *
+     * <code>string host = 1;</code>
+     */
+    public com.google.protobuf.ByteString
+        getHostBytes() {
+      java.lang.Object ref = host_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        host_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * The host name on which a source code file is located.
+     * </pre>
+     *
+     * <code>string host = 1;</code>
+     */
+    public Builder setHost(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      host_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * The host name on which a source code file is located.
+     * </pre>
+     *
+     * <code>string host = 1;</code>
+     */
+    public Builder clearHost() {
+      
+      host_ = getDefaultInstance().getHost();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * The host name on which a source code file is located.
+     * </pre>
+     *
+     * <code>string host = 1;</code>
+     */
+    public Builder setHostBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      host_ = value;
+      onChanged();
+      return this;
+    }
+
+    private java.lang.Object filePath_ = "";
+    /**
+     * <pre>
+     * Path to the source code file.
+     * </pre>
+     *
+     * <code>string file_path = 2;</code>
+     */
+    public java.lang.String getFilePath() {
+      java.lang.Object ref = filePath_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        filePath_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Path to the source code file.
+     * </pre>
+     *
+     * <code>string file_path = 2;</code>
+     */
+    public com.google.protobuf.ByteString
+        getFilePathBytes() {
+      java.lang.Object ref = filePath_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        filePath_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Path to the source code file.
+     * </pre>
+     *
+     * <code>string file_path = 2;</code>
+     */
+    public Builder setFilePath(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      filePath_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Path to the source code file.
+     * </pre>
+     *
+     * <code>string file_path = 2;</code>
+     */
+    public Builder clearFilePath() {
+      
+      filePath_ = getDefaultInstance().getFilePath();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Path to the source code file.
+     * </pre>
+     *
+     * <code>string file_path = 2;</code>
+     */
+    public Builder setFilePathBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      filePath_ = value;
+      onChanged();
+      return this;
+    }
+
+    private long lastModified_ ;
+    /**
+     * <pre>
+     * The timestamp at which the source code file is last modified.
+     * </pre>
+     *
+     * <code>int64 last_modified = 3;</code>
+     */
+    public long getLastModified() {
+      return lastModified_;
+    }
+    /**
+     * <pre>
+     * The timestamp at which the source code file is last modified.
+     * </pre>
+     *
+     * <code>int64 last_modified = 3;</code>
+     */
+    public Builder setLastModified(long value) {
+      
+      lastModified_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * The timestamp at which the source code file is last modified.
+     * </pre>
+     *
+     * <code>int64 last_modified = 3;</code>
+     */
+    public Builder clearLastModified() {
+      
+      lastModified_ = 0L;
+      onChanged();
+      return this;
+    }
+
+    private long bytes_ ;
+    /**
+     * <pre>
+     * Byte size of the file.
+     * </pre>
+     *
+     * <code>int64 bytes = 4;</code>
+     */
+    public long getBytes() {
+      return bytes_;
+    }
+    /**
+     * <pre>
+     * Byte size of the file.
+     * </pre>
+     *
+     * <code>int64 bytes = 4;</code>
+     */
+    public Builder setBytes(long value) {
+      
+      bytes_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Byte size of the file.
+     * </pre>
+     *
+     * <code>int64 bytes = 4;</code>
+     */
+    public Builder clearBytes() {
+      
+      bytes_ = 0L;
+      onChanged();
+      return this;
+    }
+
+    private com.google.protobuf.LazyStringList lines_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+    private void ensureLinesIsMutable() {
+      if (!((bitField0_ & 0x00000001) != 0)) {
+        lines_ = new com.google.protobuf.LazyStringArrayList(lines_);
+        bitField0_ |= 0x00000001;
+       }
+    }
+    /**
+     * <pre>
+     * Line-by-line content of the source code file.
+     * </pre>
+     *
+     * <code>repeated string lines = 5;</code>
+     */
+    public com.google.protobuf.ProtocolStringList
+        getLinesList() {
+      return lines_.getUnmodifiableView();
+    }
+    /**
+     * <pre>
+     * Line-by-line content of the source code file.
+     * </pre>
+     *
+     * <code>repeated string lines = 5;</code>
+     */
+    public int getLinesCount() {
+      return lines_.size();
+    }
+    /**
+     * <pre>
+     * Line-by-line content of the source code file.
+     * </pre>
+     *
+     * <code>repeated string lines = 5;</code>
+     */
+    public java.lang.String getLines(int index) {
+      return lines_.get(index);
+    }
+    /**
+     * <pre>
+     * Line-by-line content of the source code file.
+     * </pre>
+     *
+     * <code>repeated string lines = 5;</code>
+     */
+    public com.google.protobuf.ByteString
+        getLinesBytes(int index) {
+      return lines_.getByteString(index);
+    }
+    /**
+     * <pre>
+     * Line-by-line content of the source code file.
+     * </pre>
+     *
+     * <code>repeated string lines = 5;</code>
+     */
+    public Builder setLines(
+        int index, java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  ensureLinesIsMutable();
+      lines_.set(index, value);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Line-by-line content of the source code file.
+     * </pre>
+     *
+     * <code>repeated string lines = 5;</code>
+     */
+    public Builder addLines(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  ensureLinesIsMutable();
+      lines_.add(value);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Line-by-line content of the source code file.
+     * </pre>
+     *
+     * <code>repeated string lines = 5;</code>
+     */
+    public Builder addAllLines(
+        java.lang.Iterable<java.lang.String> values) {
+      ensureLinesIsMutable();
+      com.google.protobuf.AbstractMessageLite.Builder.addAll(
+          values, lines_);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Line-by-line content of the source code file.
+     * </pre>
+     *
+     * <code>repeated string lines = 5;</code>
+     */
+    public Builder clearLines() {
+      lines_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+      bitField0_ = (bitField0_ & ~0x00000001);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Line-by-line content of the source code file.
+     * </pre>
+     *
+     * <code>repeated string lines = 5;</code>
+     */
+    public Builder addLinesBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      ensureLinesIsMutable();
+      lines_.add(value);
+      onChanged();
+      return this;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.DebuggedSourceFile)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.DebuggedSourceFile)
+  private static final org.tensorflow.proto.framework.DebuggedSourceFile DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.DebuggedSourceFile();
+  }
+
+  public static org.tensorflow.proto.framework.DebuggedSourceFile getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<DebuggedSourceFile>
+      PARSER = new com.google.protobuf.AbstractParser<DebuggedSourceFile>() {
+    @java.lang.Override
+    public DebuggedSourceFile parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new DebuggedSourceFile(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<DebuggedSourceFile> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<DebuggedSourceFile> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.framework.DebuggedSourceFile getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DebuggedSourceFileOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DebuggedSourceFileOrBuilder.java
new file mode 100644
index 00000000000..1a855dc8a07
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DebuggedSourceFileOrBuilder.java
@@ -0,0 +1,98 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/debug.proto
+
+package org.tensorflow.proto.framework;
+
+public interface DebuggedSourceFileOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.DebuggedSourceFile)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <pre>
+   * The host name on which a source code file is located.
+   * </pre>
+   *
+   * <code>string host = 1;</code>
+   */
+  java.lang.String getHost();
+  /**
+   * <pre>
+   * The host name on which a source code file is located.
+   * </pre>
+   *
+   * <code>string host = 1;</code>
+   */
+  com.google.protobuf.ByteString
+      getHostBytes();
+
+  /**
+   * <pre>
+   * Path to the source code file.
+   * </pre>
+   *
+   * <code>string file_path = 2;</code>
+   */
+  java.lang.String getFilePath();
+  /**
+   * <pre>
+   * Path to the source code file.
+   * </pre>
+   *
+   * <code>string file_path = 2;</code>
+   */
+  com.google.protobuf.ByteString
+      getFilePathBytes();
+
+  /**
+   * <pre>
+   * The timestamp at which the source code file is last modified.
+   * </pre>
+   *
+   * <code>int64 last_modified = 3;</code>
+   */
+  long getLastModified();
+
+  /**
+   * <pre>
+   * Byte size of the file.
+   * </pre>
+   *
+   * <code>int64 bytes = 4;</code>
+   */
+  long getBytes();
+
+  /**
+   * <pre>
+   * Line-by-line content of the source code file.
+   * </pre>
+   *
+   * <code>repeated string lines = 5;</code>
+   */
+  java.util.List<java.lang.String>
+      getLinesList();
+  /**
+   * <pre>
+   * Line-by-line content of the source code file.
+   * </pre>
+   *
+   * <code>repeated string lines = 5;</code>
+   */
+  int getLinesCount();
+  /**
+   * <pre>
+   * Line-by-line content of the source code file.
+   * </pre>
+   *
+   * <code>repeated string lines = 5;</code>
+   */
+  java.lang.String getLines(int index);
+  /**
+   * <pre>
+   * Line-by-line content of the source code file.
+   * </pre>
+   *
+   * <code>repeated string lines = 5;</code>
+   */
+  com.google.protobuf.ByteString
+      getLinesBytes(int index);
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DebuggedSourceFiles.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DebuggedSourceFiles.java
new file mode 100644
index 00000000000..c5d014b4bb6
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DebuggedSourceFiles.java
@@ -0,0 +1,857 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/debug.proto
+
+package org.tensorflow.proto.framework;
+
+/**
+ * Protobuf type {@code tensorflow.DebuggedSourceFiles}
+ */
+public  final class DebuggedSourceFiles extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.DebuggedSourceFiles)
+    DebuggedSourceFilesOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use DebuggedSourceFiles.newBuilder() to construct.
+  private DebuggedSourceFiles(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private DebuggedSourceFiles() {
+    sourceFiles_ = java.util.Collections.emptyList();
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new DebuggedSourceFiles();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private DebuggedSourceFiles(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    int mutable_bitField0_ = 0;
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 10: {
+            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
+              sourceFiles_ = new java.util.ArrayList<org.tensorflow.proto.framework.DebuggedSourceFile>();
+              mutable_bitField0_ |= 0x00000001;
+            }
+            sourceFiles_.add(
+                input.readMessage(org.tensorflow.proto.framework.DebuggedSourceFile.parser(), extensionRegistry));
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      if (((mutable_bitField0_ & 0x00000001) != 0)) {
+        sourceFiles_ = java.util.Collections.unmodifiableList(sourceFiles_);
+      }
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.framework.DebugProtos.internal_static_tensorflow_DebuggedSourceFiles_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.framework.DebugProtos.internal_static_tensorflow_DebuggedSourceFiles_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.framework.DebuggedSourceFiles.class, org.tensorflow.proto.framework.DebuggedSourceFiles.Builder.class);
+  }
+
+  public static final int SOURCE_FILES_FIELD_NUMBER = 1;
+  private java.util.List<org.tensorflow.proto.framework.DebuggedSourceFile> sourceFiles_;
+  /**
+   * <pre>
+   * A collection of source code files.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.DebuggedSourceFile source_files = 1;</code>
+   */
+  public java.util.List<org.tensorflow.proto.framework.DebuggedSourceFile> getSourceFilesList() {
+    return sourceFiles_;
+  }
+  /**
+   * <pre>
+   * A collection of source code files.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.DebuggedSourceFile source_files = 1;</code>
+   */
+  public java.util.List<? extends org.tensorflow.proto.framework.DebuggedSourceFileOrBuilder> 
+      getSourceFilesOrBuilderList() {
+    return sourceFiles_;
+  }
+  /**
+   * <pre>
+   * A collection of source code files.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.DebuggedSourceFile source_files = 1;</code>
+   */
+  public int getSourceFilesCount() {
+    return sourceFiles_.size();
+  }
+  /**
+   * <pre>
+   * A collection of source code files.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.DebuggedSourceFile source_files = 1;</code>
+   */
+  public org.tensorflow.proto.framework.DebuggedSourceFile getSourceFiles(int index) {
+    return sourceFiles_.get(index);
+  }
+  /**
+   * <pre>
+   * A collection of source code files.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.DebuggedSourceFile source_files = 1;</code>
+   */
+  public org.tensorflow.proto.framework.DebuggedSourceFileOrBuilder getSourceFilesOrBuilder(
+      int index) {
+    return sourceFiles_.get(index);
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    for (int i = 0; i < sourceFiles_.size(); i++) {
+      output.writeMessage(1, sourceFiles_.get(i));
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    for (int i = 0; i < sourceFiles_.size(); i++) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(1, sourceFiles_.get(i));
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.framework.DebuggedSourceFiles)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.framework.DebuggedSourceFiles other = (org.tensorflow.proto.framework.DebuggedSourceFiles) obj;
+
+    if (!getSourceFilesList()
+        .equals(other.getSourceFilesList())) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    if (getSourceFilesCount() > 0) {
+      hash = (37 * hash) + SOURCE_FILES_FIELD_NUMBER;
+      hash = (53 * hash) + getSourceFilesList().hashCode();
+    }
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.framework.DebuggedSourceFiles parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.DebuggedSourceFiles parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.DebuggedSourceFiles parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.DebuggedSourceFiles parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.DebuggedSourceFiles parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.DebuggedSourceFiles parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.DebuggedSourceFiles parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.DebuggedSourceFiles parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.DebuggedSourceFiles parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.DebuggedSourceFiles parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.DebuggedSourceFiles parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.DebuggedSourceFiles parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.framework.DebuggedSourceFiles prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * Protobuf type {@code tensorflow.DebuggedSourceFiles}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.DebuggedSourceFiles)
+      org.tensorflow.proto.framework.DebuggedSourceFilesOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.DebugProtos.internal_static_tensorflow_DebuggedSourceFiles_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.DebugProtos.internal_static_tensorflow_DebuggedSourceFiles_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.DebuggedSourceFiles.class, org.tensorflow.proto.framework.DebuggedSourceFiles.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.framework.DebuggedSourceFiles.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+        getSourceFilesFieldBuilder();
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      if (sourceFilesBuilder_ == null) {
+        sourceFiles_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000001);
+      } else {
+        sourceFilesBuilder_.clear();
+      }
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.framework.DebugProtos.internal_static_tensorflow_DebuggedSourceFiles_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.DebuggedSourceFiles getDefaultInstanceForType() {
+      return org.tensorflow.proto.framework.DebuggedSourceFiles.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.DebuggedSourceFiles build() {
+      org.tensorflow.proto.framework.DebuggedSourceFiles result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.DebuggedSourceFiles buildPartial() {
+      org.tensorflow.proto.framework.DebuggedSourceFiles result = new org.tensorflow.proto.framework.DebuggedSourceFiles(this);
+      int from_bitField0_ = bitField0_;
+      if (sourceFilesBuilder_ == null) {
+        if (((bitField0_ & 0x00000001) != 0)) {
+          sourceFiles_ = java.util.Collections.unmodifiableList(sourceFiles_);
+          bitField0_ = (bitField0_ & ~0x00000001);
+        }
+        result.sourceFiles_ = sourceFiles_;
+      } else {
+        result.sourceFiles_ = sourceFilesBuilder_.build();
+      }
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.framework.DebuggedSourceFiles) {
+        return mergeFrom((org.tensorflow.proto.framework.DebuggedSourceFiles)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.framework.DebuggedSourceFiles other) {
+      if (other == org.tensorflow.proto.framework.DebuggedSourceFiles.getDefaultInstance()) return this;
+      if (sourceFilesBuilder_ == null) {
+        if (!other.sourceFiles_.isEmpty()) {
+          if (sourceFiles_.isEmpty()) {
+            sourceFiles_ = other.sourceFiles_;
+            bitField0_ = (bitField0_ & ~0x00000001);
+          } else {
+            ensureSourceFilesIsMutable();
+            sourceFiles_.addAll(other.sourceFiles_);
+          }
+          onChanged();
+        }
+      } else {
+        if (!other.sourceFiles_.isEmpty()) {
+          if (sourceFilesBuilder_.isEmpty()) {
+            sourceFilesBuilder_.dispose();
+            sourceFilesBuilder_ = null;
+            sourceFiles_ = other.sourceFiles_;
+            bitField0_ = (bitField0_ & ~0x00000001);
+            sourceFilesBuilder_ = 
+              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
+                 getSourceFilesFieldBuilder() : null;
+          } else {
+            sourceFilesBuilder_.addAllMessages(other.sourceFiles_);
+          }
+        }
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.framework.DebuggedSourceFiles parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.framework.DebuggedSourceFiles) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+    private int bitField0_;
+
+    private java.util.List<org.tensorflow.proto.framework.DebuggedSourceFile> sourceFiles_ =
+      java.util.Collections.emptyList();
+    private void ensureSourceFilesIsMutable() {
+      if (!((bitField0_ & 0x00000001) != 0)) {
+        sourceFiles_ = new java.util.ArrayList<org.tensorflow.proto.framework.DebuggedSourceFile>(sourceFiles_);
+        bitField0_ |= 0x00000001;
+       }
+    }
+
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.framework.DebuggedSourceFile, org.tensorflow.proto.framework.DebuggedSourceFile.Builder, org.tensorflow.proto.framework.DebuggedSourceFileOrBuilder> sourceFilesBuilder_;
+
+    /**
+     * <pre>
+     * A collection of source code files.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.DebuggedSourceFile source_files = 1;</code>
+     */
+    public java.util.List<org.tensorflow.proto.framework.DebuggedSourceFile> getSourceFilesList() {
+      if (sourceFilesBuilder_ == null) {
+        return java.util.Collections.unmodifiableList(sourceFiles_);
+      } else {
+        return sourceFilesBuilder_.getMessageList();
+      }
+    }
+    /**
+     * <pre>
+     * A collection of source code files.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.DebuggedSourceFile source_files = 1;</code>
+     */
+    public int getSourceFilesCount() {
+      if (sourceFilesBuilder_ == null) {
+        return sourceFiles_.size();
+      } else {
+        return sourceFilesBuilder_.getCount();
+      }
+    }
+    /**
+     * <pre>
+     * A collection of source code files.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.DebuggedSourceFile source_files = 1;</code>
+     */
+    public org.tensorflow.proto.framework.DebuggedSourceFile getSourceFiles(int index) {
+      if (sourceFilesBuilder_ == null) {
+        return sourceFiles_.get(index);
+      } else {
+        return sourceFilesBuilder_.getMessage(index);
+      }
+    }
+    /**
+     * <pre>
+     * A collection of source code files.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.DebuggedSourceFile source_files = 1;</code>
+     */
+    public Builder setSourceFiles(
+        int index, org.tensorflow.proto.framework.DebuggedSourceFile value) {
+      if (sourceFilesBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureSourceFilesIsMutable();
+        sourceFiles_.set(index, value);
+        onChanged();
+      } else {
+        sourceFilesBuilder_.setMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * A collection of source code files.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.DebuggedSourceFile source_files = 1;</code>
+     */
+    public Builder setSourceFiles(
+        int index, org.tensorflow.proto.framework.DebuggedSourceFile.Builder builderForValue) {
+      if (sourceFilesBuilder_ == null) {
+        ensureSourceFilesIsMutable();
+        sourceFiles_.set(index, builderForValue.build());
+        onChanged();
+      } else {
+        sourceFilesBuilder_.setMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * A collection of source code files.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.DebuggedSourceFile source_files = 1;</code>
+     */
+    public Builder addSourceFiles(org.tensorflow.proto.framework.DebuggedSourceFile value) {
+      if (sourceFilesBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureSourceFilesIsMutable();
+        sourceFiles_.add(value);
+        onChanged();
+      } else {
+        sourceFilesBuilder_.addMessage(value);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * A collection of source code files.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.DebuggedSourceFile source_files = 1;</code>
+     */
+    public Builder addSourceFiles(
+        int index, org.tensorflow.proto.framework.DebuggedSourceFile value) {
+      if (sourceFilesBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureSourceFilesIsMutable();
+        sourceFiles_.add(index, value);
+        onChanged();
+      } else {
+        sourceFilesBuilder_.addMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * A collection of source code files.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.DebuggedSourceFile source_files = 1;</code>
+     */
+    public Builder addSourceFiles(
+        org.tensorflow.proto.framework.DebuggedSourceFile.Builder builderForValue) {
+      if (sourceFilesBuilder_ == null) {
+        ensureSourceFilesIsMutable();
+        sourceFiles_.add(builderForValue.build());
+        onChanged();
+      } else {
+        sourceFilesBuilder_.addMessage(builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * A collection of source code files.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.DebuggedSourceFile source_files = 1;</code>
+     */
+    public Builder addSourceFiles(
+        int index, org.tensorflow.proto.framework.DebuggedSourceFile.Builder builderForValue) {
+      if (sourceFilesBuilder_ == null) {
+        ensureSourceFilesIsMutable();
+        sourceFiles_.add(index, builderForValue.build());
+        onChanged();
+      } else {
+        sourceFilesBuilder_.addMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * A collection of source code files.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.DebuggedSourceFile source_files = 1;</code>
+     */
+    public Builder addAllSourceFiles(
+        java.lang.Iterable<? extends org.tensorflow.proto.framework.DebuggedSourceFile> values) {
+      if (sourceFilesBuilder_ == null) {
+        ensureSourceFilesIsMutable();
+        com.google.protobuf.AbstractMessageLite.Builder.addAll(
+            values, sourceFiles_);
+        onChanged();
+      } else {
+        sourceFilesBuilder_.addAllMessages(values);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * A collection of source code files.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.DebuggedSourceFile source_files = 1;</code>
+     */
+    public Builder clearSourceFiles() {
+      if (sourceFilesBuilder_ == null) {
+        sourceFiles_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000001);
+        onChanged();
+      } else {
+        sourceFilesBuilder_.clear();
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * A collection of source code files.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.DebuggedSourceFile source_files = 1;</code>
+     */
+    public Builder removeSourceFiles(int index) {
+      if (sourceFilesBuilder_ == null) {
+        ensureSourceFilesIsMutable();
+        sourceFiles_.remove(index);
+        onChanged();
+      } else {
+        sourceFilesBuilder_.remove(index);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * A collection of source code files.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.DebuggedSourceFile source_files = 1;</code>
+     */
+    public org.tensorflow.proto.framework.DebuggedSourceFile.Builder getSourceFilesBuilder(
+        int index) {
+      return getSourceFilesFieldBuilder().getBuilder(index);
+    }
+    /**
+     * <pre>
+     * A collection of source code files.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.DebuggedSourceFile source_files = 1;</code>
+     */
+    public org.tensorflow.proto.framework.DebuggedSourceFileOrBuilder getSourceFilesOrBuilder(
+        int index) {
+      if (sourceFilesBuilder_ == null) {
+        return sourceFiles_.get(index);  } else {
+        return sourceFilesBuilder_.getMessageOrBuilder(index);
+      }
+    }
+    /**
+     * <pre>
+     * A collection of source code files.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.DebuggedSourceFile source_files = 1;</code>
+     */
+    public java.util.List<? extends org.tensorflow.proto.framework.DebuggedSourceFileOrBuilder> 
+         getSourceFilesOrBuilderList() {
+      if (sourceFilesBuilder_ != null) {
+        return sourceFilesBuilder_.getMessageOrBuilderList();
+      } else {
+        return java.util.Collections.unmodifiableList(sourceFiles_);
+      }
+    }
+    /**
+     * <pre>
+     * A collection of source code files.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.DebuggedSourceFile source_files = 1;</code>
+     */
+    public org.tensorflow.proto.framework.DebuggedSourceFile.Builder addSourceFilesBuilder() {
+      return getSourceFilesFieldBuilder().addBuilder(
+          org.tensorflow.proto.framework.DebuggedSourceFile.getDefaultInstance());
+    }
+    /**
+     * <pre>
+     * A collection of source code files.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.DebuggedSourceFile source_files = 1;</code>
+     */
+    public org.tensorflow.proto.framework.DebuggedSourceFile.Builder addSourceFilesBuilder(
+        int index) {
+      return getSourceFilesFieldBuilder().addBuilder(
+          index, org.tensorflow.proto.framework.DebuggedSourceFile.getDefaultInstance());
+    }
+    /**
+     * <pre>
+     * A collection of source code files.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.DebuggedSourceFile source_files = 1;</code>
+     */
+    public java.util.List<org.tensorflow.proto.framework.DebuggedSourceFile.Builder> 
+         getSourceFilesBuilderList() {
+      return getSourceFilesFieldBuilder().getBuilderList();
+    }
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.framework.DebuggedSourceFile, org.tensorflow.proto.framework.DebuggedSourceFile.Builder, org.tensorflow.proto.framework.DebuggedSourceFileOrBuilder> 
+        getSourceFilesFieldBuilder() {
+      if (sourceFilesBuilder_ == null) {
+        sourceFilesBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
+            org.tensorflow.proto.framework.DebuggedSourceFile, org.tensorflow.proto.framework.DebuggedSourceFile.Builder, org.tensorflow.proto.framework.DebuggedSourceFileOrBuilder>(
+                sourceFiles_,
+                ((bitField0_ & 0x00000001) != 0),
+                getParentForChildren(),
+                isClean());
+        sourceFiles_ = null;
+      }
+      return sourceFilesBuilder_;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.DebuggedSourceFiles)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.DebuggedSourceFiles)
+  private static final org.tensorflow.proto.framework.DebuggedSourceFiles DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.DebuggedSourceFiles();
+  }
+
+  public static org.tensorflow.proto.framework.DebuggedSourceFiles getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<DebuggedSourceFiles>
+      PARSER = new com.google.protobuf.AbstractParser<DebuggedSourceFiles>() {
+    @java.lang.Override
+    public DebuggedSourceFiles parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new DebuggedSourceFiles(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<DebuggedSourceFiles> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<DebuggedSourceFiles> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.framework.DebuggedSourceFiles getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DebuggedSourceFilesOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DebuggedSourceFilesOrBuilder.java
new file mode 100644
index 00000000000..3afc8e4f78f
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DebuggedSourceFilesOrBuilder.java
@@ -0,0 +1,53 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/debug.proto
+
+package org.tensorflow.proto.framework;
+
+public interface DebuggedSourceFilesOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.DebuggedSourceFiles)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <pre>
+   * A collection of source code files.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.DebuggedSourceFile source_files = 1;</code>
+   */
+  java.util.List<org.tensorflow.proto.framework.DebuggedSourceFile> 
+      getSourceFilesList();
+  /**
+   * <pre>
+   * A collection of source code files.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.DebuggedSourceFile source_files = 1;</code>
+   */
+  org.tensorflow.proto.framework.DebuggedSourceFile getSourceFiles(int index);
+  /**
+   * <pre>
+   * A collection of source code files.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.DebuggedSourceFile source_files = 1;</code>
+   */
+  int getSourceFilesCount();
+  /**
+   * <pre>
+   * A collection of source code files.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.DebuggedSourceFile source_files = 1;</code>
+   */
+  java.util.List<? extends org.tensorflow.proto.framework.DebuggedSourceFileOrBuilder> 
+      getSourceFilesOrBuilderList();
+  /**
+   * <pre>
+   * A collection of source code files.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.DebuggedSourceFile source_files = 1;</code>
+   */
+  org.tensorflow.proto.framework.DebuggedSourceFileOrBuilder getSourceFilesOrBuilder(
+      int index);
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DeviceAttributes.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DeviceAttributes.java
new file mode 100644
index 00000000000..3da166dfbfd
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DeviceAttributes.java
@@ -0,0 +1,1277 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/framework/device_attributes.proto
+
+package org.tensorflow.proto.framework;
+
+/**
+ * Protobuf type {@code tensorflow.DeviceAttributes}
+ */
+public  final class DeviceAttributes extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.DeviceAttributes)
+    DeviceAttributesOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use DeviceAttributes.newBuilder() to construct.
+  private DeviceAttributes(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private DeviceAttributes() {
+    name_ = "";
+    deviceType_ = "";
+    physicalDeviceDesc_ = "";
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new DeviceAttributes();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private DeviceAttributes(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 10: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            name_ = s;
+            break;
+          }
+          case 18: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            deviceType_ = s;
+            break;
+          }
+          case 32: {
+
+            memoryLimit_ = input.readInt64();
+            break;
+          }
+          case 42: {
+            org.tensorflow.proto.framework.DeviceLocality.Builder subBuilder = null;
+            if (locality_ != null) {
+              subBuilder = locality_.toBuilder();
+            }
+            locality_ = input.readMessage(org.tensorflow.proto.framework.DeviceLocality.parser(), extensionRegistry);
+            if (subBuilder != null) {
+              subBuilder.mergeFrom(locality_);
+              locality_ = subBuilder.buildPartial();
+            }
+
+            break;
+          }
+          case 49: {
+
+            incarnation_ = input.readFixed64();
+            break;
+          }
+          case 58: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            physicalDeviceDesc_ = s;
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.framework.DeviceAttributesProtos.internal_static_tensorflow_DeviceAttributes_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.framework.DeviceAttributesProtos.internal_static_tensorflow_DeviceAttributes_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.framework.DeviceAttributes.class, org.tensorflow.proto.framework.DeviceAttributes.Builder.class);
+  }
+
+  public static final int NAME_FIELD_NUMBER = 1;
+  private volatile java.lang.Object name_;
+  /**
+   * <pre>
+   * Fully specified name of the device within a cluster.
+   * </pre>
+   *
+   * <code>string name = 1;</code>
+   */
+  public java.lang.String getName() {
+    java.lang.Object ref = name_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      name_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * Fully specified name of the device within a cluster.
+   * </pre>
+   *
+   * <code>string name = 1;</code>
+   */
+  public com.google.protobuf.ByteString
+      getNameBytes() {
+    java.lang.Object ref = name_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      name_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int DEVICE_TYPE_FIELD_NUMBER = 2;
+  private volatile java.lang.Object deviceType_;
+  /**
+   * <pre>
+   * String representation of device_type.
+   * </pre>
+   *
+   * <code>string device_type = 2;</code>
+   */
+  public java.lang.String getDeviceType() {
+    java.lang.Object ref = deviceType_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      deviceType_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * String representation of device_type.
+   * </pre>
+   *
+   * <code>string device_type = 2;</code>
+   */
+  public com.google.protobuf.ByteString
+      getDeviceTypeBytes() {
+    java.lang.Object ref = deviceType_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      deviceType_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int MEMORY_LIMIT_FIELD_NUMBER = 4;
+  private long memoryLimit_;
+  /**
+   * <pre>
+   * Memory capacity of device in bytes.
+   * </pre>
+   *
+   * <code>int64 memory_limit = 4;</code>
+   */
+  public long getMemoryLimit() {
+    return memoryLimit_;
+  }
+
+  public static final int LOCALITY_FIELD_NUMBER = 5;
+  private org.tensorflow.proto.framework.DeviceLocality locality_;
+  /**
+   * <pre>
+   * Platform-specific data about device that may be useful
+   * for supporting efficient data transfers.
+   * </pre>
+   *
+   * <code>.tensorflow.DeviceLocality locality = 5;</code>
+   */
+  public boolean hasLocality() {
+    return locality_ != null;
+  }
+  /**
+   * <pre>
+   * Platform-specific data about device that may be useful
+   * for supporting efficient data transfers.
+   * </pre>
+   *
+   * <code>.tensorflow.DeviceLocality locality = 5;</code>
+   */
+  public org.tensorflow.proto.framework.DeviceLocality getLocality() {
+    return locality_ == null ? org.tensorflow.proto.framework.DeviceLocality.getDefaultInstance() : locality_;
+  }
+  /**
+   * <pre>
+   * Platform-specific data about device that may be useful
+   * for supporting efficient data transfers.
+   * </pre>
+   *
+   * <code>.tensorflow.DeviceLocality locality = 5;</code>
+   */
+  public org.tensorflow.proto.framework.DeviceLocalityOrBuilder getLocalityOrBuilder() {
+    return getLocality();
+  }
+
+  public static final int INCARNATION_FIELD_NUMBER = 6;
+  private long incarnation_;
+  /**
+   * <pre>
+   * A device is assigned a global unique number each time it is
+   * initialized. "incarnation" should never be 0.
+   * </pre>
+   *
+   * <code>fixed64 incarnation = 6;</code>
+   */
+  public long getIncarnation() {
+    return incarnation_;
+  }
+
+  public static final int PHYSICAL_DEVICE_DESC_FIELD_NUMBER = 7;
+  private volatile java.lang.Object physicalDeviceDesc_;
+  /**
+   * <pre>
+   * String representation of the physical device that this device maps to.
+   * </pre>
+   *
+   * <code>string physical_device_desc = 7;</code>
+   */
+  public java.lang.String getPhysicalDeviceDesc() {
+    java.lang.Object ref = physicalDeviceDesc_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      physicalDeviceDesc_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * String representation of the physical device that this device maps to.
+   * </pre>
+   *
+   * <code>string physical_device_desc = 7;</code>
+   */
+  public com.google.protobuf.ByteString
+      getPhysicalDeviceDescBytes() {
+    java.lang.Object ref = physicalDeviceDesc_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      physicalDeviceDesc_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    if (!getNameBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 1, name_);
+    }
+    if (!getDeviceTypeBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 2, deviceType_);
+    }
+    if (memoryLimit_ != 0L) {
+      output.writeInt64(4, memoryLimit_);
+    }
+    if (locality_ != null) {
+      output.writeMessage(5, getLocality());
+    }
+    if (incarnation_ != 0L) {
+      output.writeFixed64(6, incarnation_);
+    }
+    if (!getPhysicalDeviceDescBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 7, physicalDeviceDesc_);
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (!getNameBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, name_);
+    }
+    if (!getDeviceTypeBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(2, deviceType_);
+    }
+    if (memoryLimit_ != 0L) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt64Size(4, memoryLimit_);
+    }
+    if (locality_ != null) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(5, getLocality());
+    }
+    if (incarnation_ != 0L) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeFixed64Size(6, incarnation_);
+    }
+    if (!getPhysicalDeviceDescBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(7, physicalDeviceDesc_);
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.framework.DeviceAttributes)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.framework.DeviceAttributes other = (org.tensorflow.proto.framework.DeviceAttributes) obj;
+
+    if (!getName()
+        .equals(other.getName())) return false;
+    if (!getDeviceType()
+        .equals(other.getDeviceType())) return false;
+    if (getMemoryLimit()
+        != other.getMemoryLimit()) return false;
+    if (hasLocality() != other.hasLocality()) return false;
+    if (hasLocality()) {
+      if (!getLocality()
+          .equals(other.getLocality())) return false;
+    }
+    if (getIncarnation()
+        != other.getIncarnation()) return false;
+    if (!getPhysicalDeviceDesc()
+        .equals(other.getPhysicalDeviceDesc())) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    hash = (37 * hash) + NAME_FIELD_NUMBER;
+    hash = (53 * hash) + getName().hashCode();
+    hash = (37 * hash) + DEVICE_TYPE_FIELD_NUMBER;
+    hash = (53 * hash) + getDeviceType().hashCode();
+    hash = (37 * hash) + MEMORY_LIMIT_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+        getMemoryLimit());
+    if (hasLocality()) {
+      hash = (37 * hash) + LOCALITY_FIELD_NUMBER;
+      hash = (53 * hash) + getLocality().hashCode();
+    }
+    hash = (37 * hash) + INCARNATION_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+        getIncarnation());
+    hash = (37 * hash) + PHYSICAL_DEVICE_DESC_FIELD_NUMBER;
+    hash = (53 * hash) + getPhysicalDeviceDesc().hashCode();
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.framework.DeviceAttributes parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.DeviceAttributes parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.DeviceAttributes parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.DeviceAttributes parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.DeviceAttributes parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.DeviceAttributes parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.DeviceAttributes parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.DeviceAttributes parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.DeviceAttributes parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.DeviceAttributes parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.DeviceAttributes parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.DeviceAttributes parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.framework.DeviceAttributes prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * Protobuf type {@code tensorflow.DeviceAttributes}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.DeviceAttributes)
+      org.tensorflow.proto.framework.DeviceAttributesOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.DeviceAttributesProtos.internal_static_tensorflow_DeviceAttributes_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.DeviceAttributesProtos.internal_static_tensorflow_DeviceAttributes_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.DeviceAttributes.class, org.tensorflow.proto.framework.DeviceAttributes.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.framework.DeviceAttributes.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      name_ = "";
+
+      deviceType_ = "";
+
+      memoryLimit_ = 0L;
+
+      if (localityBuilder_ == null) {
+        locality_ = null;
+      } else {
+        locality_ = null;
+        localityBuilder_ = null;
+      }
+      incarnation_ = 0L;
+
+      physicalDeviceDesc_ = "";
+
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.framework.DeviceAttributesProtos.internal_static_tensorflow_DeviceAttributes_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.DeviceAttributes getDefaultInstanceForType() {
+      return org.tensorflow.proto.framework.DeviceAttributes.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.DeviceAttributes build() {
+      org.tensorflow.proto.framework.DeviceAttributes result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.DeviceAttributes buildPartial() {
+      org.tensorflow.proto.framework.DeviceAttributes result = new org.tensorflow.proto.framework.DeviceAttributes(this);
+      result.name_ = name_;
+      result.deviceType_ = deviceType_;
+      result.memoryLimit_ = memoryLimit_;
+      if (localityBuilder_ == null) {
+        result.locality_ = locality_;
+      } else {
+        result.locality_ = localityBuilder_.build();
+      }
+      result.incarnation_ = incarnation_;
+      result.physicalDeviceDesc_ = physicalDeviceDesc_;
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.framework.DeviceAttributes) {
+        return mergeFrom((org.tensorflow.proto.framework.DeviceAttributes)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.framework.DeviceAttributes other) {
+      if (other == org.tensorflow.proto.framework.DeviceAttributes.getDefaultInstance()) return this;
+      if (!other.getName().isEmpty()) {
+        name_ = other.name_;
+        onChanged();
+      }
+      if (!other.getDeviceType().isEmpty()) {
+        deviceType_ = other.deviceType_;
+        onChanged();
+      }
+      if (other.getMemoryLimit() != 0L) {
+        setMemoryLimit(other.getMemoryLimit());
+      }
+      if (other.hasLocality()) {
+        mergeLocality(other.getLocality());
+      }
+      if (other.getIncarnation() != 0L) {
+        setIncarnation(other.getIncarnation());
+      }
+      if (!other.getPhysicalDeviceDesc().isEmpty()) {
+        physicalDeviceDesc_ = other.physicalDeviceDesc_;
+        onChanged();
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.framework.DeviceAttributes parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.framework.DeviceAttributes) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+
+    private java.lang.Object name_ = "";
+    /**
+     * <pre>
+     * Fully specified name of the device within a cluster.
+     * </pre>
+     *
+     * <code>string name = 1;</code>
+     */
+    public java.lang.String getName() {
+      java.lang.Object ref = name_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        name_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Fully specified name of the device within a cluster.
+     * </pre>
+     *
+     * <code>string name = 1;</code>
+     */
+    public com.google.protobuf.ByteString
+        getNameBytes() {
+      java.lang.Object ref = name_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        name_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Fully specified name of the device within a cluster.
+     * </pre>
+     *
+     * <code>string name = 1;</code>
+     */
+    public Builder setName(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      name_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Fully specified name of the device within a cluster.
+     * </pre>
+     *
+     * <code>string name = 1;</code>
+     */
+    public Builder clearName() {
+      
+      name_ = getDefaultInstance().getName();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Fully specified name of the device within a cluster.
+     * </pre>
+     *
+     * <code>string name = 1;</code>
+     */
+    public Builder setNameBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      name_ = value;
+      onChanged();
+      return this;
+    }
+
+    private java.lang.Object deviceType_ = "";
+    /**
+     * <pre>
+     * String representation of device_type.
+     * </pre>
+     *
+     * <code>string device_type = 2;</code>
+     */
+    public java.lang.String getDeviceType() {
+      java.lang.Object ref = deviceType_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        deviceType_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * String representation of device_type.
+     * </pre>
+     *
+     * <code>string device_type = 2;</code>
+     */
+    public com.google.protobuf.ByteString
+        getDeviceTypeBytes() {
+      java.lang.Object ref = deviceType_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        deviceType_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * String representation of device_type.
+     * </pre>
+     *
+     * <code>string device_type = 2;</code>
+     */
+    public Builder setDeviceType(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      deviceType_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * String representation of device_type.
+     * </pre>
+     *
+     * <code>string device_type = 2;</code>
+     */
+    public Builder clearDeviceType() {
+      
+      deviceType_ = getDefaultInstance().getDeviceType();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * String representation of device_type.
+     * </pre>
+     *
+     * <code>string device_type = 2;</code>
+     */
+    public Builder setDeviceTypeBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      deviceType_ = value;
+      onChanged();
+      return this;
+    }
+
+    private long memoryLimit_ ;
+    /**
+     * <pre>
+     * Memory capacity of device in bytes.
+     * </pre>
+     *
+     * <code>int64 memory_limit = 4;</code>
+     */
+    public long getMemoryLimit() {
+      return memoryLimit_;
+    }
+    /**
+     * <pre>
+     * Memory capacity of device in bytes.
+     * </pre>
+     *
+     * <code>int64 memory_limit = 4;</code>
+     */
+    public Builder setMemoryLimit(long value) {
+      
+      memoryLimit_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Memory capacity of device in bytes.
+     * </pre>
+     *
+     * <code>int64 memory_limit = 4;</code>
+     */
+    public Builder clearMemoryLimit() {
+      
+      memoryLimit_ = 0L;
+      onChanged();
+      return this;
+    }
+
+    private org.tensorflow.proto.framework.DeviceLocality locality_;
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.DeviceLocality, org.tensorflow.proto.framework.DeviceLocality.Builder, org.tensorflow.proto.framework.DeviceLocalityOrBuilder> localityBuilder_;
+    /**
+     * <pre>
+     * Platform-specific data about device that may be useful
+     * for supporting efficient data transfers.
+     * </pre>
+     *
+     * <code>.tensorflow.DeviceLocality locality = 5;</code>
+     */
+    public boolean hasLocality() {
+      return localityBuilder_ != null || locality_ != null;
+    }
+    /**
+     * <pre>
+     * Platform-specific data about device that may be useful
+     * for supporting efficient data transfers.
+     * </pre>
+     *
+     * <code>.tensorflow.DeviceLocality locality = 5;</code>
+     */
+    public org.tensorflow.proto.framework.DeviceLocality getLocality() {
+      if (localityBuilder_ == null) {
+        return locality_ == null ? org.tensorflow.proto.framework.DeviceLocality.getDefaultInstance() : locality_;
+      } else {
+        return localityBuilder_.getMessage();
+      }
+    }
+    /**
+     * <pre>
+     * Platform-specific data about device that may be useful
+     * for supporting efficient data transfers.
+     * </pre>
+     *
+     * <code>.tensorflow.DeviceLocality locality = 5;</code>
+     */
+    public Builder setLocality(org.tensorflow.proto.framework.DeviceLocality value) {
+      if (localityBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        locality_ = value;
+        onChanged();
+      } else {
+        localityBuilder_.setMessage(value);
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Platform-specific data about device that may be useful
+     * for supporting efficient data transfers.
+     * </pre>
+     *
+     * <code>.tensorflow.DeviceLocality locality = 5;</code>
+     */
+    public Builder setLocality(
+        org.tensorflow.proto.framework.DeviceLocality.Builder builderForValue) {
+      if (localityBuilder_ == null) {
+        locality_ = builderForValue.build();
+        onChanged();
+      } else {
+        localityBuilder_.setMessage(builderForValue.build());
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Platform-specific data about device that may be useful
+     * for supporting efficient data transfers.
+     * </pre>
+     *
+     * <code>.tensorflow.DeviceLocality locality = 5;</code>
+     */
+    public Builder mergeLocality(org.tensorflow.proto.framework.DeviceLocality value) {
+      if (localityBuilder_ == null) {
+        if (locality_ != null) {
+          locality_ =
+            org.tensorflow.proto.framework.DeviceLocality.newBuilder(locality_).mergeFrom(value).buildPartial();
+        } else {
+          locality_ = value;
+        }
+        onChanged();
+      } else {
+        localityBuilder_.mergeFrom(value);
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Platform-specific data about device that may be useful
+     * for supporting efficient data transfers.
+     * </pre>
+     *
+     * <code>.tensorflow.DeviceLocality locality = 5;</code>
+     */
+    public Builder clearLocality() {
+      if (localityBuilder_ == null) {
+        locality_ = null;
+        onChanged();
+      } else {
+        locality_ = null;
+        localityBuilder_ = null;
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Platform-specific data about device that may be useful
+     * for supporting efficient data transfers.
+     * </pre>
+     *
+     * <code>.tensorflow.DeviceLocality locality = 5;</code>
+     */
+    public org.tensorflow.proto.framework.DeviceLocality.Builder getLocalityBuilder() {
+      
+      onChanged();
+      return getLocalityFieldBuilder().getBuilder();
+    }
+    /**
+     * <pre>
+     * Platform-specific data about device that may be useful
+     * for supporting efficient data transfers.
+     * </pre>
+     *
+     * <code>.tensorflow.DeviceLocality locality = 5;</code>
+     */
+    public org.tensorflow.proto.framework.DeviceLocalityOrBuilder getLocalityOrBuilder() {
+      if (localityBuilder_ != null) {
+        return localityBuilder_.getMessageOrBuilder();
+      } else {
+        return locality_ == null ?
+            org.tensorflow.proto.framework.DeviceLocality.getDefaultInstance() : locality_;
+      }
+    }
+    /**
+     * <pre>
+     * Platform-specific data about device that may be useful
+     * for supporting efficient data transfers.
+     * </pre>
+     *
+     * <code>.tensorflow.DeviceLocality locality = 5;</code>
+     */
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.DeviceLocality, org.tensorflow.proto.framework.DeviceLocality.Builder, org.tensorflow.proto.framework.DeviceLocalityOrBuilder> 
+        getLocalityFieldBuilder() {
+      if (localityBuilder_ == null) {
+        localityBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+            org.tensorflow.proto.framework.DeviceLocality, org.tensorflow.proto.framework.DeviceLocality.Builder, org.tensorflow.proto.framework.DeviceLocalityOrBuilder>(
+                getLocality(),
+                getParentForChildren(),
+                isClean());
+        locality_ = null;
+      }
+      return localityBuilder_;
+    }
+
+    private long incarnation_ ;
+    /**
+     * <pre>
+     * A device is assigned a global unique number each time it is
+     * initialized. "incarnation" should never be 0.
+     * </pre>
+     *
+     * <code>fixed64 incarnation = 6;</code>
+     */
+    public long getIncarnation() {
+      return incarnation_;
+    }
+    /**
+     * <pre>
+     * A device is assigned a global unique number each time it is
+     * initialized. "incarnation" should never be 0.
+     * </pre>
+     *
+     * <code>fixed64 incarnation = 6;</code>
+     */
+    public Builder setIncarnation(long value) {
+      
+      incarnation_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * A device is assigned a global unique number each time it is
+     * initialized. "incarnation" should never be 0.
+     * </pre>
+     *
+     * <code>fixed64 incarnation = 6;</code>
+     */
+    public Builder clearIncarnation() {
+      
+      incarnation_ = 0L;
+      onChanged();
+      return this;
+    }
+
+    private java.lang.Object physicalDeviceDesc_ = "";
+    /**
+     * <pre>
+     * String representation of the physical device that this device maps to.
+     * </pre>
+     *
+     * <code>string physical_device_desc = 7;</code>
+     */
+    public java.lang.String getPhysicalDeviceDesc() {
+      java.lang.Object ref = physicalDeviceDesc_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        physicalDeviceDesc_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * String representation of the physical device that this device maps to.
+     * </pre>
+     *
+     * <code>string physical_device_desc = 7;</code>
+     */
+    public com.google.protobuf.ByteString
+        getPhysicalDeviceDescBytes() {
+      java.lang.Object ref = physicalDeviceDesc_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        physicalDeviceDesc_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * String representation of the physical device that this device maps to.
+     * </pre>
+     *
+     * <code>string physical_device_desc = 7;</code>
+     */
+    public Builder setPhysicalDeviceDesc(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      physicalDeviceDesc_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * String representation of the physical device that this device maps to.
+     * </pre>
+     *
+     * <code>string physical_device_desc = 7;</code>
+     */
+    public Builder clearPhysicalDeviceDesc() {
+      
+      physicalDeviceDesc_ = getDefaultInstance().getPhysicalDeviceDesc();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * String representation of the physical device that this device maps to.
+     * </pre>
+     *
+     * <code>string physical_device_desc = 7;</code>
+     */
+    public Builder setPhysicalDeviceDescBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      physicalDeviceDesc_ = value;
+      onChanged();
+      return this;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.DeviceAttributes)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.DeviceAttributes)
+  private static final org.tensorflow.proto.framework.DeviceAttributes DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.DeviceAttributes();
+  }
+
+  public static org.tensorflow.proto.framework.DeviceAttributes getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<DeviceAttributes>
+      PARSER = new com.google.protobuf.AbstractParser<DeviceAttributes>() {
+    @java.lang.Override
+    public DeviceAttributes parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new DeviceAttributes(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<DeviceAttributes> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<DeviceAttributes> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.framework.DeviceAttributes getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DeviceAttributesOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DeviceAttributesOrBuilder.java
new file mode 100644
index 00000000000..bb8c806b325
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DeviceAttributesOrBuilder.java
@@ -0,0 +1,110 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/framework/device_attributes.proto
+
+package org.tensorflow.proto.framework;
+
+public interface DeviceAttributesOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.DeviceAttributes)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <pre>
+   * Fully specified name of the device within a cluster.
+   * </pre>
+   *
+   * <code>string name = 1;</code>
+   */
+  java.lang.String getName();
+  /**
+   * <pre>
+   * Fully specified name of the device within a cluster.
+   * </pre>
+   *
+   * <code>string name = 1;</code>
+   */
+  com.google.protobuf.ByteString
+      getNameBytes();
+
+  /**
+   * <pre>
+   * String representation of device_type.
+   * </pre>
+   *
+   * <code>string device_type = 2;</code>
+   */
+  java.lang.String getDeviceType();
+  /**
+   * <pre>
+   * String representation of device_type.
+   * </pre>
+   *
+   * <code>string device_type = 2;</code>
+   */
+  com.google.protobuf.ByteString
+      getDeviceTypeBytes();
+
+  /**
+   * <pre>
+   * Memory capacity of device in bytes.
+   * </pre>
+   *
+   * <code>int64 memory_limit = 4;</code>
+   */
+  long getMemoryLimit();
+
+  /**
+   * <pre>
+   * Platform-specific data about device that may be useful
+   * for supporting efficient data transfers.
+   * </pre>
+   *
+   * <code>.tensorflow.DeviceLocality locality = 5;</code>
+   */
+  boolean hasLocality();
+  /**
+   * <pre>
+   * Platform-specific data about device that may be useful
+   * for supporting efficient data transfers.
+   * </pre>
+   *
+   * <code>.tensorflow.DeviceLocality locality = 5;</code>
+   */
+  org.tensorflow.proto.framework.DeviceLocality getLocality();
+  /**
+   * <pre>
+   * Platform-specific data about device that may be useful
+   * for supporting efficient data transfers.
+   * </pre>
+   *
+   * <code>.tensorflow.DeviceLocality locality = 5;</code>
+   */
+  org.tensorflow.proto.framework.DeviceLocalityOrBuilder getLocalityOrBuilder();
+
+  /**
+   * <pre>
+   * A device is assigned a global unique number each time it is
+   * initialized. "incarnation" should never be 0.
+   * </pre>
+   *
+   * <code>fixed64 incarnation = 6;</code>
+   */
+  long getIncarnation();
+
+  /**
+   * <pre>
+   * String representation of the physical device that this device maps to.
+   * </pre>
+   *
+   * <code>string physical_device_desc = 7;</code>
+   */
+  java.lang.String getPhysicalDeviceDesc();
+  /**
+   * <pre>
+   * String representation of the physical device that this device maps to.
+   * </pre>
+   *
+   * <code>string physical_device_desc = 7;</code>
+   */
+  com.google.protobuf.ByteString
+      getPhysicalDeviceDescBytes();
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DeviceAttributesProtos.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DeviceAttributesProtos.java
new file mode 100644
index 00000000000..f9e2dac8a67
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DeviceAttributesProtos.java
@@ -0,0 +1,94 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/framework/device_attributes.proto
+
+package org.tensorflow.proto.framework;
+
+public final class DeviceAttributesProtos {
+  private DeviceAttributesProtos() {}
+  public static void registerAllExtensions(
+      com.google.protobuf.ExtensionRegistryLite registry) {
+  }
+
+  public static void registerAllExtensions(
+      com.google.protobuf.ExtensionRegistry registry) {
+    registerAllExtensions(
+        (com.google.protobuf.ExtensionRegistryLite) registry);
+  }
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_InterconnectLink_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_InterconnectLink_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_LocalLinks_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_LocalLinks_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_DeviceLocality_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_DeviceLocality_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_DeviceAttributes_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_DeviceAttributes_fieldAccessorTable;
+
+  public static com.google.protobuf.Descriptors.FileDescriptor
+      getDescriptor() {
+    return descriptor;
+  }
+  private static  com.google.protobuf.Descriptors.FileDescriptor
+      descriptor;
+  static {
+    java.lang.String[] descriptorData = {
+      "\n1tensorflow/core/framework/device_attri" +
+      "butes.proto\022\ntensorflow\"E\n\020InterconnectL" +
+      "ink\022\021\n\tdevice_id\030\001 \001(\005\022\014\n\004type\030\002 \001(\t\022\020\n\010" +
+      "strength\030\003 \001(\005\"8\n\nLocalLinks\022*\n\004link\030\001 \003" +
+      "(\0132\034.tensorflow.InterconnectLink\"Z\n\016Devi" +
+      "ceLocality\022\016\n\006bus_id\030\001 \001(\005\022\021\n\tnuma_node\030" +
+      "\002 \001(\005\022%\n\005links\030\003 \001(\0132\026.tensorflow.LocalL" +
+      "inks\"\254\001\n\020DeviceAttributes\022\014\n\004name\030\001 \001(\t\022" +
+      "\023\n\013device_type\030\002 \001(\t\022\024\n\014memory_limit\030\004 \001" +
+      "(\003\022,\n\010locality\030\005 \001(\0132\032.tensorflow.Device" +
+      "Locality\022\023\n\013incarnation\030\006 \001(\006\022\034\n\024physica" +
+      "l_device_desc\030\007 \001(\tB\227\001\n\036org.tensorflow.p" +
+      "roto.frameworkB\026DeviceAttributesProtosP\001" +
+      "ZXgithub.com/tensorflow/tensorflow/tenso" +
+      "rflow/go/core/framework/device_attribute" +
+      "s_go_proto\370\001\001b\006proto3"
+    };
+    descriptor = com.google.protobuf.Descriptors.FileDescriptor
+      .internalBuildGeneratedFileFrom(descriptorData,
+        new com.google.protobuf.Descriptors.FileDescriptor[] {
+        });
+    internal_static_tensorflow_InterconnectLink_descriptor =
+      getDescriptor().getMessageTypes().get(0);
+    internal_static_tensorflow_InterconnectLink_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_InterconnectLink_descriptor,
+        new java.lang.String[] { "DeviceId", "Type", "Strength", });
+    internal_static_tensorflow_LocalLinks_descriptor =
+      getDescriptor().getMessageTypes().get(1);
+    internal_static_tensorflow_LocalLinks_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_LocalLinks_descriptor,
+        new java.lang.String[] { "Link", });
+    internal_static_tensorflow_DeviceLocality_descriptor =
+      getDescriptor().getMessageTypes().get(2);
+    internal_static_tensorflow_DeviceLocality_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_DeviceLocality_descriptor,
+        new java.lang.String[] { "BusId", "NumaNode", "Links", });
+    internal_static_tensorflow_DeviceAttributes_descriptor =
+      getDescriptor().getMessageTypes().get(3);
+    internal_static_tensorflow_DeviceAttributes_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_DeviceAttributes_descriptor,
+        new java.lang.String[] { "Name", "DeviceType", "MemoryLimit", "Locality", "Incarnation", "PhysicalDeviceDesc", });
+  }
+
+  // @@protoc_insertion_point(outer_class_scope)
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DeviceLocality.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DeviceLocality.java
new file mode 100644
index 00000000000..25d5784a00d
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DeviceLocality.java
@@ -0,0 +1,798 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/framework/device_attributes.proto
+
+package org.tensorflow.proto.framework;
+
+/**
+ * Protobuf type {@code tensorflow.DeviceLocality}
+ */
+public  final class DeviceLocality extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.DeviceLocality)
+    DeviceLocalityOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use DeviceLocality.newBuilder() to construct.
+  private DeviceLocality(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private DeviceLocality() {
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new DeviceLocality();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private DeviceLocality(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 8: {
+
+            busId_ = input.readInt32();
+            break;
+          }
+          case 16: {
+
+            numaNode_ = input.readInt32();
+            break;
+          }
+          case 26: {
+            org.tensorflow.proto.framework.LocalLinks.Builder subBuilder = null;
+            if (links_ != null) {
+              subBuilder = links_.toBuilder();
+            }
+            links_ = input.readMessage(org.tensorflow.proto.framework.LocalLinks.parser(), extensionRegistry);
+            if (subBuilder != null) {
+              subBuilder.mergeFrom(links_);
+              links_ = subBuilder.buildPartial();
+            }
+
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.framework.DeviceAttributesProtos.internal_static_tensorflow_DeviceLocality_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.framework.DeviceAttributesProtos.internal_static_tensorflow_DeviceLocality_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.framework.DeviceLocality.class, org.tensorflow.proto.framework.DeviceLocality.Builder.class);
+  }
+
+  public static final int BUS_ID_FIELD_NUMBER = 1;
+  private int busId_;
+  /**
+   * <pre>
+   * Optional bus locality of device.  Default value of 0 means
+   * no specific locality.  Specific localities are indexed from 1.
+   * </pre>
+   *
+   * <code>int32 bus_id = 1;</code>
+   */
+  public int getBusId() {
+    return busId_;
+  }
+
+  public static final int NUMA_NODE_FIELD_NUMBER = 2;
+  private int numaNode_;
+  /**
+   * <pre>
+   * Optional NUMA locality of device.
+   * </pre>
+   *
+   * <code>int32 numa_node = 2;</code>
+   */
+  public int getNumaNode() {
+    return numaNode_;
+  }
+
+  public static final int LINKS_FIELD_NUMBER = 3;
+  private org.tensorflow.proto.framework.LocalLinks links_;
+  /**
+   * <pre>
+   * Optional local interconnect links to other devices.
+   * </pre>
+   *
+   * <code>.tensorflow.LocalLinks links = 3;</code>
+   */
+  public boolean hasLinks() {
+    return links_ != null;
+  }
+  /**
+   * <pre>
+   * Optional local interconnect links to other devices.
+   * </pre>
+   *
+   * <code>.tensorflow.LocalLinks links = 3;</code>
+   */
+  public org.tensorflow.proto.framework.LocalLinks getLinks() {
+    return links_ == null ? org.tensorflow.proto.framework.LocalLinks.getDefaultInstance() : links_;
+  }
+  /**
+   * <pre>
+   * Optional local interconnect links to other devices.
+   * </pre>
+   *
+   * <code>.tensorflow.LocalLinks links = 3;</code>
+   */
+  public org.tensorflow.proto.framework.LocalLinksOrBuilder getLinksOrBuilder() {
+    return getLinks();
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    if (busId_ != 0) {
+      output.writeInt32(1, busId_);
+    }
+    if (numaNode_ != 0) {
+      output.writeInt32(2, numaNode_);
+    }
+    if (links_ != null) {
+      output.writeMessage(3, getLinks());
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (busId_ != 0) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt32Size(1, busId_);
+    }
+    if (numaNode_ != 0) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt32Size(2, numaNode_);
+    }
+    if (links_ != null) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(3, getLinks());
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.framework.DeviceLocality)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.framework.DeviceLocality other = (org.tensorflow.proto.framework.DeviceLocality) obj;
+
+    if (getBusId()
+        != other.getBusId()) return false;
+    if (getNumaNode()
+        != other.getNumaNode()) return false;
+    if (hasLinks() != other.hasLinks()) return false;
+    if (hasLinks()) {
+      if (!getLinks()
+          .equals(other.getLinks())) return false;
+    }
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    hash = (37 * hash) + BUS_ID_FIELD_NUMBER;
+    hash = (53 * hash) + getBusId();
+    hash = (37 * hash) + NUMA_NODE_FIELD_NUMBER;
+    hash = (53 * hash) + getNumaNode();
+    if (hasLinks()) {
+      hash = (37 * hash) + LINKS_FIELD_NUMBER;
+      hash = (53 * hash) + getLinks().hashCode();
+    }
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.framework.DeviceLocality parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.DeviceLocality parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.DeviceLocality parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.DeviceLocality parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.DeviceLocality parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.DeviceLocality parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.DeviceLocality parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.DeviceLocality parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.DeviceLocality parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.DeviceLocality parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.DeviceLocality parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.DeviceLocality parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.framework.DeviceLocality prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * Protobuf type {@code tensorflow.DeviceLocality}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.DeviceLocality)
+      org.tensorflow.proto.framework.DeviceLocalityOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.DeviceAttributesProtos.internal_static_tensorflow_DeviceLocality_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.DeviceAttributesProtos.internal_static_tensorflow_DeviceLocality_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.DeviceLocality.class, org.tensorflow.proto.framework.DeviceLocality.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.framework.DeviceLocality.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      busId_ = 0;
+
+      numaNode_ = 0;
+
+      if (linksBuilder_ == null) {
+        links_ = null;
+      } else {
+        links_ = null;
+        linksBuilder_ = null;
+      }
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.framework.DeviceAttributesProtos.internal_static_tensorflow_DeviceLocality_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.DeviceLocality getDefaultInstanceForType() {
+      return org.tensorflow.proto.framework.DeviceLocality.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.DeviceLocality build() {
+      org.tensorflow.proto.framework.DeviceLocality result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.DeviceLocality buildPartial() {
+      org.tensorflow.proto.framework.DeviceLocality result = new org.tensorflow.proto.framework.DeviceLocality(this);
+      result.busId_ = busId_;
+      result.numaNode_ = numaNode_;
+      if (linksBuilder_ == null) {
+        result.links_ = links_;
+      } else {
+        result.links_ = linksBuilder_.build();
+      }
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.framework.DeviceLocality) {
+        return mergeFrom((org.tensorflow.proto.framework.DeviceLocality)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.framework.DeviceLocality other) {
+      if (other == org.tensorflow.proto.framework.DeviceLocality.getDefaultInstance()) return this;
+      if (other.getBusId() != 0) {
+        setBusId(other.getBusId());
+      }
+      if (other.getNumaNode() != 0) {
+        setNumaNode(other.getNumaNode());
+      }
+      if (other.hasLinks()) {
+        mergeLinks(other.getLinks());
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.framework.DeviceLocality parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.framework.DeviceLocality) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+
+    private int busId_ ;
+    /**
+     * <pre>
+     * Optional bus locality of device.  Default value of 0 means
+     * no specific locality.  Specific localities are indexed from 1.
+     * </pre>
+     *
+     * <code>int32 bus_id = 1;</code>
+     */
+    public int getBusId() {
+      return busId_;
+    }
+    /**
+     * <pre>
+     * Optional bus locality of device.  Default value of 0 means
+     * no specific locality.  Specific localities are indexed from 1.
+     * </pre>
+     *
+     * <code>int32 bus_id = 1;</code>
+     */
+    public Builder setBusId(int value) {
+      
+      busId_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Optional bus locality of device.  Default value of 0 means
+     * no specific locality.  Specific localities are indexed from 1.
+     * </pre>
+     *
+     * <code>int32 bus_id = 1;</code>
+     */
+    public Builder clearBusId() {
+      
+      busId_ = 0;
+      onChanged();
+      return this;
+    }
+
+    private int numaNode_ ;
+    /**
+     * <pre>
+     * Optional NUMA locality of device.
+     * </pre>
+     *
+     * <code>int32 numa_node = 2;</code>
+     */
+    public int getNumaNode() {
+      return numaNode_;
+    }
+    /**
+     * <pre>
+     * Optional NUMA locality of device.
+     * </pre>
+     *
+     * <code>int32 numa_node = 2;</code>
+     */
+    public Builder setNumaNode(int value) {
+      
+      numaNode_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Optional NUMA locality of device.
+     * </pre>
+     *
+     * <code>int32 numa_node = 2;</code>
+     */
+    public Builder clearNumaNode() {
+      
+      numaNode_ = 0;
+      onChanged();
+      return this;
+    }
+
+    private org.tensorflow.proto.framework.LocalLinks links_;
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.LocalLinks, org.tensorflow.proto.framework.LocalLinks.Builder, org.tensorflow.proto.framework.LocalLinksOrBuilder> linksBuilder_;
+    /**
+     * <pre>
+     * Optional local interconnect links to other devices.
+     * </pre>
+     *
+     * <code>.tensorflow.LocalLinks links = 3;</code>
+     */
+    public boolean hasLinks() {
+      return linksBuilder_ != null || links_ != null;
+    }
+    /**
+     * <pre>
+     * Optional local interconnect links to other devices.
+     * </pre>
+     *
+     * <code>.tensorflow.LocalLinks links = 3;</code>
+     */
+    public org.tensorflow.proto.framework.LocalLinks getLinks() {
+      if (linksBuilder_ == null) {
+        return links_ == null ? org.tensorflow.proto.framework.LocalLinks.getDefaultInstance() : links_;
+      } else {
+        return linksBuilder_.getMessage();
+      }
+    }
+    /**
+     * <pre>
+     * Optional local interconnect links to other devices.
+     * </pre>
+     *
+     * <code>.tensorflow.LocalLinks links = 3;</code>
+     */
+    public Builder setLinks(org.tensorflow.proto.framework.LocalLinks value) {
+      if (linksBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        links_ = value;
+        onChanged();
+      } else {
+        linksBuilder_.setMessage(value);
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Optional local interconnect links to other devices.
+     * </pre>
+     *
+     * <code>.tensorflow.LocalLinks links = 3;</code>
+     */
+    public Builder setLinks(
+        org.tensorflow.proto.framework.LocalLinks.Builder builderForValue) {
+      if (linksBuilder_ == null) {
+        links_ = builderForValue.build();
+        onChanged();
+      } else {
+        linksBuilder_.setMessage(builderForValue.build());
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Optional local interconnect links to other devices.
+     * </pre>
+     *
+     * <code>.tensorflow.LocalLinks links = 3;</code>
+     */
+    public Builder mergeLinks(org.tensorflow.proto.framework.LocalLinks value) {
+      if (linksBuilder_ == null) {
+        if (links_ != null) {
+          links_ =
+            org.tensorflow.proto.framework.LocalLinks.newBuilder(links_).mergeFrom(value).buildPartial();
+        } else {
+          links_ = value;
+        }
+        onChanged();
+      } else {
+        linksBuilder_.mergeFrom(value);
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Optional local interconnect links to other devices.
+     * </pre>
+     *
+     * <code>.tensorflow.LocalLinks links = 3;</code>
+     */
+    public Builder clearLinks() {
+      if (linksBuilder_ == null) {
+        links_ = null;
+        onChanged();
+      } else {
+        links_ = null;
+        linksBuilder_ = null;
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Optional local interconnect links to other devices.
+     * </pre>
+     *
+     * <code>.tensorflow.LocalLinks links = 3;</code>
+     */
+    public org.tensorflow.proto.framework.LocalLinks.Builder getLinksBuilder() {
+      
+      onChanged();
+      return getLinksFieldBuilder().getBuilder();
+    }
+    /**
+     * <pre>
+     * Optional local interconnect links to other devices.
+     * </pre>
+     *
+     * <code>.tensorflow.LocalLinks links = 3;</code>
+     */
+    public org.tensorflow.proto.framework.LocalLinksOrBuilder getLinksOrBuilder() {
+      if (linksBuilder_ != null) {
+        return linksBuilder_.getMessageOrBuilder();
+      } else {
+        return links_ == null ?
+            org.tensorflow.proto.framework.LocalLinks.getDefaultInstance() : links_;
+      }
+    }
+    /**
+     * <pre>
+     * Optional local interconnect links to other devices.
+     * </pre>
+     *
+     * <code>.tensorflow.LocalLinks links = 3;</code>
+     */
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.LocalLinks, org.tensorflow.proto.framework.LocalLinks.Builder, org.tensorflow.proto.framework.LocalLinksOrBuilder> 
+        getLinksFieldBuilder() {
+      if (linksBuilder_ == null) {
+        linksBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+            org.tensorflow.proto.framework.LocalLinks, org.tensorflow.proto.framework.LocalLinks.Builder, org.tensorflow.proto.framework.LocalLinksOrBuilder>(
+                getLinks(),
+                getParentForChildren(),
+                isClean());
+        links_ = null;
+      }
+      return linksBuilder_;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.DeviceLocality)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.DeviceLocality)
+  private static final org.tensorflow.proto.framework.DeviceLocality DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.DeviceLocality();
+  }
+
+  public static org.tensorflow.proto.framework.DeviceLocality getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<DeviceLocality>
+      PARSER = new com.google.protobuf.AbstractParser<DeviceLocality>() {
+    @java.lang.Override
+    public DeviceLocality parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new DeviceLocality(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<DeviceLocality> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<DeviceLocality> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.framework.DeviceLocality getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DeviceLocalityOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DeviceLocalityOrBuilder.java
new file mode 100644
index 00000000000..cfb7b2c5287
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DeviceLocalityOrBuilder.java
@@ -0,0 +1,53 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/framework/device_attributes.proto
+
+package org.tensorflow.proto.framework;
+
+public interface DeviceLocalityOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.DeviceLocality)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <pre>
+   * Optional bus locality of device.  Default value of 0 means
+   * no specific locality.  Specific localities are indexed from 1.
+   * </pre>
+   *
+   * <code>int32 bus_id = 1;</code>
+   */
+  int getBusId();
+
+  /**
+   * <pre>
+   * Optional NUMA locality of device.
+   * </pre>
+   *
+   * <code>int32 numa_node = 2;</code>
+   */
+  int getNumaNode();
+
+  /**
+   * <pre>
+   * Optional local interconnect links to other devices.
+   * </pre>
+   *
+   * <code>.tensorflow.LocalLinks links = 3;</code>
+   */
+  boolean hasLinks();
+  /**
+   * <pre>
+   * Optional local interconnect links to other devices.
+   * </pre>
+   *
+   * <code>.tensorflow.LocalLinks links = 3;</code>
+   */
+  org.tensorflow.proto.framework.LocalLinks getLinks();
+  /**
+   * <pre>
+   * Optional local interconnect links to other devices.
+   * </pre>
+   *
+   * <code>.tensorflow.LocalLinks links = 3;</code>
+   */
+  org.tensorflow.proto.framework.LocalLinksOrBuilder getLinksOrBuilder();
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DeviceProperties.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DeviceProperties.java
new file mode 100644
index 00000000000..a6e16b28d1f
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DeviceProperties.java
@@ -0,0 +1,1885 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/device_properties.proto
+
+package org.tensorflow.proto.framework;
+
+/**
+ * Protobuf type {@code tensorflow.DeviceProperties}
+ */
+public  final class DeviceProperties extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.DeviceProperties)
+    DevicePropertiesOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use DeviceProperties.newBuilder() to construct.
+  private DeviceProperties(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private DeviceProperties() {
+    type_ = "";
+    vendor_ = "";
+    model_ = "";
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new DeviceProperties();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private DeviceProperties(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    int mutable_bitField0_ = 0;
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 10: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            type_ = s;
+            break;
+          }
+          case 18: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            vendor_ = s;
+            break;
+          }
+          case 26: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            model_ = s;
+            break;
+          }
+          case 32: {
+
+            frequency_ = input.readInt64();
+            break;
+          }
+          case 40: {
+
+            numCores_ = input.readInt64();
+            break;
+          }
+          case 50: {
+            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
+              environment_ = com.google.protobuf.MapField.newMapField(
+                  EnvironmentDefaultEntryHolder.defaultEntry);
+              mutable_bitField0_ |= 0x00000001;
+            }
+            com.google.protobuf.MapEntry<java.lang.String, java.lang.String>
+            environment__ = input.readMessage(
+                EnvironmentDefaultEntryHolder.defaultEntry.getParserForType(), extensionRegistry);
+            environment_.getMutableMap().put(
+                environment__.getKey(), environment__.getValue());
+            break;
+          }
+          case 56: {
+
+            numRegisters_ = input.readInt64();
+            break;
+          }
+          case 64: {
+
+            l1CacheSize_ = input.readInt64();
+            break;
+          }
+          case 72: {
+
+            l2CacheSize_ = input.readInt64();
+            break;
+          }
+          case 80: {
+
+            l3CacheSize_ = input.readInt64();
+            break;
+          }
+          case 88: {
+
+            sharedMemorySizePerMultiprocessor_ = input.readInt64();
+            break;
+          }
+          case 96: {
+
+            memorySize_ = input.readInt64();
+            break;
+          }
+          case 104: {
+
+            bandwidth_ = input.readInt64();
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.framework.DevicePropertiesProtos.internal_static_tensorflow_DeviceProperties_descriptor;
+  }
+
+  @SuppressWarnings({"rawtypes"})
+  @java.lang.Override
+  protected com.google.protobuf.MapField internalGetMapField(
+      int number) {
+    switch (number) {
+      case 6:
+        return internalGetEnvironment();
+      default:
+        throw new RuntimeException(
+            "Invalid map field number: " + number);
+    }
+  }
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.framework.DevicePropertiesProtos.internal_static_tensorflow_DeviceProperties_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.framework.DeviceProperties.class, org.tensorflow.proto.framework.DeviceProperties.Builder.class);
+  }
+
+  public static final int TYPE_FIELD_NUMBER = 1;
+  private volatile java.lang.Object type_;
+  /**
+   * <pre>
+   * Device type (CPU, GPU, ...)
+   * </pre>
+   *
+   * <code>string type = 1;</code>
+   */
+  public java.lang.String getType() {
+    java.lang.Object ref = type_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      type_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * Device type (CPU, GPU, ...)
+   * </pre>
+   *
+   * <code>string type = 1;</code>
+   */
+  public com.google.protobuf.ByteString
+      getTypeBytes() {
+    java.lang.Object ref = type_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      type_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int VENDOR_FIELD_NUMBER = 2;
+  private volatile java.lang.Object vendor_;
+  /**
+   * <pre>
+   * Vendor (Intel, nvidia, ...)
+   * </pre>
+   *
+   * <code>string vendor = 2;</code>
+   */
+  public java.lang.String getVendor() {
+    java.lang.Object ref = vendor_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      vendor_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * Vendor (Intel, nvidia, ...)
+   * </pre>
+   *
+   * <code>string vendor = 2;</code>
+   */
+  public com.google.protobuf.ByteString
+      getVendorBytes() {
+    java.lang.Object ref = vendor_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      vendor_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int MODEL_FIELD_NUMBER = 3;
+  private volatile java.lang.Object model_;
+  /**
+   * <pre>
+   * Model (Haswell, K40, ...)
+   * </pre>
+   *
+   * <code>string model = 3;</code>
+   */
+  public java.lang.String getModel() {
+    java.lang.Object ref = model_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      model_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * Model (Haswell, K40, ...)
+   * </pre>
+   *
+   * <code>string model = 3;</code>
+   */
+  public com.google.protobuf.ByteString
+      getModelBytes() {
+    java.lang.Object ref = model_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      model_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int FREQUENCY_FIELD_NUMBER = 4;
+  private long frequency_;
+  /**
+   * <pre>
+   * Core Frequency in Mhz
+   * </pre>
+   *
+   * <code>int64 frequency = 4;</code>
+   */
+  public long getFrequency() {
+    return frequency_;
+  }
+
+  public static final int NUM_CORES_FIELD_NUMBER = 5;
+  private long numCores_;
+  /**
+   * <pre>
+   * Number of cores
+   * </pre>
+   *
+   * <code>int64 num_cores = 5;</code>
+   */
+  public long getNumCores() {
+    return numCores_;
+  }
+
+  public static final int ENVIRONMENT_FIELD_NUMBER = 6;
+  private static final class EnvironmentDefaultEntryHolder {
+    static final com.google.protobuf.MapEntry<
+        java.lang.String, java.lang.String> defaultEntry =
+            com.google.protobuf.MapEntry
+            .<java.lang.String, java.lang.String>newDefaultInstance(
+                org.tensorflow.proto.framework.DevicePropertiesProtos.internal_static_tensorflow_DeviceProperties_EnvironmentEntry_descriptor, 
+                com.google.protobuf.WireFormat.FieldType.STRING,
+                "",
+                com.google.protobuf.WireFormat.FieldType.STRING,
+                "");
+  }
+  private com.google.protobuf.MapField<
+      java.lang.String, java.lang.String> environment_;
+  private com.google.protobuf.MapField<java.lang.String, java.lang.String>
+  internalGetEnvironment() {
+    if (environment_ == null) {
+      return com.google.protobuf.MapField.emptyMapField(
+          EnvironmentDefaultEntryHolder.defaultEntry);
+    }
+    return environment_;
+  }
+
+  public int getEnvironmentCount() {
+    return internalGetEnvironment().getMap().size();
+  }
+  /**
+   * <pre>
+   * Version of the tools and libraries used with this device (e.g. gcc 4.9,
+   * cudnn 5.1)
+   * </pre>
+   *
+   * <code>map&lt;string, string&gt; environment = 6;</code>
+   */
+
+  public boolean containsEnvironment(
+      java.lang.String key) {
+    if (key == null) { throw new java.lang.NullPointerException(); }
+    return internalGetEnvironment().getMap().containsKey(key);
+  }
+  /**
+   * Use {@link #getEnvironmentMap()} instead.
+   */
+  @java.lang.Deprecated
+  public java.util.Map<java.lang.String, java.lang.String> getEnvironment() {
+    return getEnvironmentMap();
+  }
+  /**
+   * <pre>
+   * Version of the tools and libraries used with this device (e.g. gcc 4.9,
+   * cudnn 5.1)
+   * </pre>
+   *
+   * <code>map&lt;string, string&gt; environment = 6;</code>
+   */
+
+  public java.util.Map<java.lang.String, java.lang.String> getEnvironmentMap() {
+    return internalGetEnvironment().getMap();
+  }
+  /**
+   * <pre>
+   * Version of the tools and libraries used with this device (e.g. gcc 4.9,
+   * cudnn 5.1)
+   * </pre>
+   *
+   * <code>map&lt;string, string&gt; environment = 6;</code>
+   */
+
+  public java.lang.String getEnvironmentOrDefault(
+      java.lang.String key,
+      java.lang.String defaultValue) {
+    if (key == null) { throw new java.lang.NullPointerException(); }
+    java.util.Map<java.lang.String, java.lang.String> map =
+        internalGetEnvironment().getMap();
+    return map.containsKey(key) ? map.get(key) : defaultValue;
+  }
+  /**
+   * <pre>
+   * Version of the tools and libraries used with this device (e.g. gcc 4.9,
+   * cudnn 5.1)
+   * </pre>
+   *
+   * <code>map&lt;string, string&gt; environment = 6;</code>
+   */
+
+  public java.lang.String getEnvironmentOrThrow(
+      java.lang.String key) {
+    if (key == null) { throw new java.lang.NullPointerException(); }
+    java.util.Map<java.lang.String, java.lang.String> map =
+        internalGetEnvironment().getMap();
+    if (!map.containsKey(key)) {
+      throw new java.lang.IllegalArgumentException();
+    }
+    return map.get(key);
+  }
+
+  public static final int NUM_REGISTERS_FIELD_NUMBER = 7;
+  private long numRegisters_;
+  /**
+   * <pre>
+   * Number of registers per core.
+   * </pre>
+   *
+   * <code>int64 num_registers = 7;</code>
+   */
+  public long getNumRegisters() {
+    return numRegisters_;
+  }
+
+  public static final int L1_CACHE_SIZE_FIELD_NUMBER = 8;
+  private long l1CacheSize_;
+  /**
+   * <pre>
+   * L1 cache size in bytes
+   * </pre>
+   *
+   * <code>int64 l1_cache_size = 8;</code>
+   */
+  public long getL1CacheSize() {
+    return l1CacheSize_;
+  }
+
+  public static final int L2_CACHE_SIZE_FIELD_NUMBER = 9;
+  private long l2CacheSize_;
+  /**
+   * <pre>
+   * L2 cache size in bytes
+   * </pre>
+   *
+   * <code>int64 l2_cache_size = 9;</code>
+   */
+  public long getL2CacheSize() {
+    return l2CacheSize_;
+  }
+
+  public static final int L3_CACHE_SIZE_FIELD_NUMBER = 10;
+  private long l3CacheSize_;
+  /**
+   * <pre>
+   * L3 cache size in bytes
+   * </pre>
+   *
+   * <code>int64 l3_cache_size = 10;</code>
+   */
+  public long getL3CacheSize() {
+    return l3CacheSize_;
+  }
+
+  public static final int SHARED_MEMORY_SIZE_PER_MULTIPROCESSOR_FIELD_NUMBER = 11;
+  private long sharedMemorySizePerMultiprocessor_;
+  /**
+   * <pre>
+   * Shared memory size per multiprocessor in bytes. This field is
+   * applicable to GPUs only.
+   * </pre>
+   *
+   * <code>int64 shared_memory_size_per_multiprocessor = 11;</code>
+   */
+  public long getSharedMemorySizePerMultiprocessor() {
+    return sharedMemorySizePerMultiprocessor_;
+  }
+
+  public static final int MEMORY_SIZE_FIELD_NUMBER = 12;
+  private long memorySize_;
+  /**
+   * <pre>
+   * Memory size in bytes
+   * </pre>
+   *
+   * <code>int64 memory_size = 12;</code>
+   */
+  public long getMemorySize() {
+    return memorySize_;
+  }
+
+  public static final int BANDWIDTH_FIELD_NUMBER = 13;
+  private long bandwidth_;
+  /**
+   * <pre>
+   * Memory bandwidth in KB/s
+   * </pre>
+   *
+   * <code>int64 bandwidth = 13;</code>
+   */
+  public long getBandwidth() {
+    return bandwidth_;
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    if (!getTypeBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 1, type_);
+    }
+    if (!getVendorBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 2, vendor_);
+    }
+    if (!getModelBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 3, model_);
+    }
+    if (frequency_ != 0L) {
+      output.writeInt64(4, frequency_);
+    }
+    if (numCores_ != 0L) {
+      output.writeInt64(5, numCores_);
+    }
+    com.google.protobuf.GeneratedMessageV3
+      .serializeStringMapTo(
+        output,
+        internalGetEnvironment(),
+        EnvironmentDefaultEntryHolder.defaultEntry,
+        6);
+    if (numRegisters_ != 0L) {
+      output.writeInt64(7, numRegisters_);
+    }
+    if (l1CacheSize_ != 0L) {
+      output.writeInt64(8, l1CacheSize_);
+    }
+    if (l2CacheSize_ != 0L) {
+      output.writeInt64(9, l2CacheSize_);
+    }
+    if (l3CacheSize_ != 0L) {
+      output.writeInt64(10, l3CacheSize_);
+    }
+    if (sharedMemorySizePerMultiprocessor_ != 0L) {
+      output.writeInt64(11, sharedMemorySizePerMultiprocessor_);
+    }
+    if (memorySize_ != 0L) {
+      output.writeInt64(12, memorySize_);
+    }
+    if (bandwidth_ != 0L) {
+      output.writeInt64(13, bandwidth_);
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (!getTypeBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, type_);
+    }
+    if (!getVendorBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(2, vendor_);
+    }
+    if (!getModelBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(3, model_);
+    }
+    if (frequency_ != 0L) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt64Size(4, frequency_);
+    }
+    if (numCores_ != 0L) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt64Size(5, numCores_);
+    }
+    for (java.util.Map.Entry<java.lang.String, java.lang.String> entry
+         : internalGetEnvironment().getMap().entrySet()) {
+      com.google.protobuf.MapEntry<java.lang.String, java.lang.String>
+      environment__ = EnvironmentDefaultEntryHolder.defaultEntry.newBuilderForType()
+          .setKey(entry.getKey())
+          .setValue(entry.getValue())
+          .build();
+      size += com.google.protobuf.CodedOutputStream
+          .computeMessageSize(6, environment__);
+    }
+    if (numRegisters_ != 0L) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt64Size(7, numRegisters_);
+    }
+    if (l1CacheSize_ != 0L) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt64Size(8, l1CacheSize_);
+    }
+    if (l2CacheSize_ != 0L) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt64Size(9, l2CacheSize_);
+    }
+    if (l3CacheSize_ != 0L) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt64Size(10, l3CacheSize_);
+    }
+    if (sharedMemorySizePerMultiprocessor_ != 0L) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt64Size(11, sharedMemorySizePerMultiprocessor_);
+    }
+    if (memorySize_ != 0L) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt64Size(12, memorySize_);
+    }
+    if (bandwidth_ != 0L) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt64Size(13, bandwidth_);
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.framework.DeviceProperties)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.framework.DeviceProperties other = (org.tensorflow.proto.framework.DeviceProperties) obj;
+
+    if (!getType()
+        .equals(other.getType())) return false;
+    if (!getVendor()
+        .equals(other.getVendor())) return false;
+    if (!getModel()
+        .equals(other.getModel())) return false;
+    if (getFrequency()
+        != other.getFrequency()) return false;
+    if (getNumCores()
+        != other.getNumCores()) return false;
+    if (!internalGetEnvironment().equals(
+        other.internalGetEnvironment())) return false;
+    if (getNumRegisters()
+        != other.getNumRegisters()) return false;
+    if (getL1CacheSize()
+        != other.getL1CacheSize()) return false;
+    if (getL2CacheSize()
+        != other.getL2CacheSize()) return false;
+    if (getL3CacheSize()
+        != other.getL3CacheSize()) return false;
+    if (getSharedMemorySizePerMultiprocessor()
+        != other.getSharedMemorySizePerMultiprocessor()) return false;
+    if (getMemorySize()
+        != other.getMemorySize()) return false;
+    if (getBandwidth()
+        != other.getBandwidth()) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    hash = (37 * hash) + TYPE_FIELD_NUMBER;
+    hash = (53 * hash) + getType().hashCode();
+    hash = (37 * hash) + VENDOR_FIELD_NUMBER;
+    hash = (53 * hash) + getVendor().hashCode();
+    hash = (37 * hash) + MODEL_FIELD_NUMBER;
+    hash = (53 * hash) + getModel().hashCode();
+    hash = (37 * hash) + FREQUENCY_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+        getFrequency());
+    hash = (37 * hash) + NUM_CORES_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+        getNumCores());
+    if (!internalGetEnvironment().getMap().isEmpty()) {
+      hash = (37 * hash) + ENVIRONMENT_FIELD_NUMBER;
+      hash = (53 * hash) + internalGetEnvironment().hashCode();
+    }
+    hash = (37 * hash) + NUM_REGISTERS_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+        getNumRegisters());
+    hash = (37 * hash) + L1_CACHE_SIZE_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+        getL1CacheSize());
+    hash = (37 * hash) + L2_CACHE_SIZE_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+        getL2CacheSize());
+    hash = (37 * hash) + L3_CACHE_SIZE_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+        getL3CacheSize());
+    hash = (37 * hash) + SHARED_MEMORY_SIZE_PER_MULTIPROCESSOR_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+        getSharedMemorySizePerMultiprocessor());
+    hash = (37 * hash) + MEMORY_SIZE_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+        getMemorySize());
+    hash = (37 * hash) + BANDWIDTH_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+        getBandwidth());
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.framework.DeviceProperties parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.DeviceProperties parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.DeviceProperties parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.DeviceProperties parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.DeviceProperties parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.DeviceProperties parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.DeviceProperties parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.DeviceProperties parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.DeviceProperties parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.DeviceProperties parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.DeviceProperties parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.DeviceProperties parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.framework.DeviceProperties prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * Protobuf type {@code tensorflow.DeviceProperties}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.DeviceProperties)
+      org.tensorflow.proto.framework.DevicePropertiesOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.DevicePropertiesProtos.internal_static_tensorflow_DeviceProperties_descriptor;
+    }
+
+    @SuppressWarnings({"rawtypes"})
+    protected com.google.protobuf.MapField internalGetMapField(
+        int number) {
+      switch (number) {
+        case 6:
+          return internalGetEnvironment();
+        default:
+          throw new RuntimeException(
+              "Invalid map field number: " + number);
+      }
+    }
+    @SuppressWarnings({"rawtypes"})
+    protected com.google.protobuf.MapField internalGetMutableMapField(
+        int number) {
+      switch (number) {
+        case 6:
+          return internalGetMutableEnvironment();
+        default:
+          throw new RuntimeException(
+              "Invalid map field number: " + number);
+      }
+    }
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.DevicePropertiesProtos.internal_static_tensorflow_DeviceProperties_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.DeviceProperties.class, org.tensorflow.proto.framework.DeviceProperties.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.framework.DeviceProperties.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      type_ = "";
+
+      vendor_ = "";
+
+      model_ = "";
+
+      frequency_ = 0L;
+
+      numCores_ = 0L;
+
+      internalGetMutableEnvironment().clear();
+      numRegisters_ = 0L;
+
+      l1CacheSize_ = 0L;
+
+      l2CacheSize_ = 0L;
+
+      l3CacheSize_ = 0L;
+
+      sharedMemorySizePerMultiprocessor_ = 0L;
+
+      memorySize_ = 0L;
+
+      bandwidth_ = 0L;
+
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.framework.DevicePropertiesProtos.internal_static_tensorflow_DeviceProperties_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.DeviceProperties getDefaultInstanceForType() {
+      return org.tensorflow.proto.framework.DeviceProperties.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.DeviceProperties build() {
+      org.tensorflow.proto.framework.DeviceProperties result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.DeviceProperties buildPartial() {
+      org.tensorflow.proto.framework.DeviceProperties result = new org.tensorflow.proto.framework.DeviceProperties(this);
+      int from_bitField0_ = bitField0_;
+      result.type_ = type_;
+      result.vendor_ = vendor_;
+      result.model_ = model_;
+      result.frequency_ = frequency_;
+      result.numCores_ = numCores_;
+      result.environment_ = internalGetEnvironment();
+      result.environment_.makeImmutable();
+      result.numRegisters_ = numRegisters_;
+      result.l1CacheSize_ = l1CacheSize_;
+      result.l2CacheSize_ = l2CacheSize_;
+      result.l3CacheSize_ = l3CacheSize_;
+      result.sharedMemorySizePerMultiprocessor_ = sharedMemorySizePerMultiprocessor_;
+      result.memorySize_ = memorySize_;
+      result.bandwidth_ = bandwidth_;
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.framework.DeviceProperties) {
+        return mergeFrom((org.tensorflow.proto.framework.DeviceProperties)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.framework.DeviceProperties other) {
+      if (other == org.tensorflow.proto.framework.DeviceProperties.getDefaultInstance()) return this;
+      if (!other.getType().isEmpty()) {
+        type_ = other.type_;
+        onChanged();
+      }
+      if (!other.getVendor().isEmpty()) {
+        vendor_ = other.vendor_;
+        onChanged();
+      }
+      if (!other.getModel().isEmpty()) {
+        model_ = other.model_;
+        onChanged();
+      }
+      if (other.getFrequency() != 0L) {
+        setFrequency(other.getFrequency());
+      }
+      if (other.getNumCores() != 0L) {
+        setNumCores(other.getNumCores());
+      }
+      internalGetMutableEnvironment().mergeFrom(
+          other.internalGetEnvironment());
+      if (other.getNumRegisters() != 0L) {
+        setNumRegisters(other.getNumRegisters());
+      }
+      if (other.getL1CacheSize() != 0L) {
+        setL1CacheSize(other.getL1CacheSize());
+      }
+      if (other.getL2CacheSize() != 0L) {
+        setL2CacheSize(other.getL2CacheSize());
+      }
+      if (other.getL3CacheSize() != 0L) {
+        setL3CacheSize(other.getL3CacheSize());
+      }
+      if (other.getSharedMemorySizePerMultiprocessor() != 0L) {
+        setSharedMemorySizePerMultiprocessor(other.getSharedMemorySizePerMultiprocessor());
+      }
+      if (other.getMemorySize() != 0L) {
+        setMemorySize(other.getMemorySize());
+      }
+      if (other.getBandwidth() != 0L) {
+        setBandwidth(other.getBandwidth());
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.framework.DeviceProperties parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.framework.DeviceProperties) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+    private int bitField0_;
+
+    private java.lang.Object type_ = "";
+    /**
+     * <pre>
+     * Device type (CPU, GPU, ...)
+     * </pre>
+     *
+     * <code>string type = 1;</code>
+     */
+    public java.lang.String getType() {
+      java.lang.Object ref = type_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        type_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Device type (CPU, GPU, ...)
+     * </pre>
+     *
+     * <code>string type = 1;</code>
+     */
+    public com.google.protobuf.ByteString
+        getTypeBytes() {
+      java.lang.Object ref = type_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        type_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Device type (CPU, GPU, ...)
+     * </pre>
+     *
+     * <code>string type = 1;</code>
+     */
+    public Builder setType(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      type_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Device type (CPU, GPU, ...)
+     * </pre>
+     *
+     * <code>string type = 1;</code>
+     */
+    public Builder clearType() {
+      
+      type_ = getDefaultInstance().getType();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Device type (CPU, GPU, ...)
+     * </pre>
+     *
+     * <code>string type = 1;</code>
+     */
+    public Builder setTypeBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      type_ = value;
+      onChanged();
+      return this;
+    }
+
+    private java.lang.Object vendor_ = "";
+    /**
+     * <pre>
+     * Vendor (Intel, nvidia, ...)
+     * </pre>
+     *
+     * <code>string vendor = 2;</code>
+     */
+    public java.lang.String getVendor() {
+      java.lang.Object ref = vendor_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        vendor_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Vendor (Intel, nvidia, ...)
+     * </pre>
+     *
+     * <code>string vendor = 2;</code>
+     */
+    public com.google.protobuf.ByteString
+        getVendorBytes() {
+      java.lang.Object ref = vendor_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        vendor_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Vendor (Intel, nvidia, ...)
+     * </pre>
+     *
+     * <code>string vendor = 2;</code>
+     */
+    public Builder setVendor(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      vendor_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Vendor (Intel, nvidia, ...)
+     * </pre>
+     *
+     * <code>string vendor = 2;</code>
+     */
+    public Builder clearVendor() {
+      
+      vendor_ = getDefaultInstance().getVendor();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Vendor (Intel, nvidia, ...)
+     * </pre>
+     *
+     * <code>string vendor = 2;</code>
+     */
+    public Builder setVendorBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      vendor_ = value;
+      onChanged();
+      return this;
+    }
+
+    private java.lang.Object model_ = "";
+    /**
+     * <pre>
+     * Model (Haswell, K40, ...)
+     * </pre>
+     *
+     * <code>string model = 3;</code>
+     */
+    public java.lang.String getModel() {
+      java.lang.Object ref = model_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        model_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Model (Haswell, K40, ...)
+     * </pre>
+     *
+     * <code>string model = 3;</code>
+     */
+    public com.google.protobuf.ByteString
+        getModelBytes() {
+      java.lang.Object ref = model_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        model_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Model (Haswell, K40, ...)
+     * </pre>
+     *
+     * <code>string model = 3;</code>
+     */
+    public Builder setModel(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      model_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Model (Haswell, K40, ...)
+     * </pre>
+     *
+     * <code>string model = 3;</code>
+     */
+    public Builder clearModel() {
+      
+      model_ = getDefaultInstance().getModel();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Model (Haswell, K40, ...)
+     * </pre>
+     *
+     * <code>string model = 3;</code>
+     */
+    public Builder setModelBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      model_ = value;
+      onChanged();
+      return this;
+    }
+
+    private long frequency_ ;
+    /**
+     * <pre>
+     * Core Frequency in Mhz
+     * </pre>
+     *
+     * <code>int64 frequency = 4;</code>
+     */
+    public long getFrequency() {
+      return frequency_;
+    }
+    /**
+     * <pre>
+     * Core Frequency in Mhz
+     * </pre>
+     *
+     * <code>int64 frequency = 4;</code>
+     */
+    public Builder setFrequency(long value) {
+      
+      frequency_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Core Frequency in Mhz
+     * </pre>
+     *
+     * <code>int64 frequency = 4;</code>
+     */
+    public Builder clearFrequency() {
+      
+      frequency_ = 0L;
+      onChanged();
+      return this;
+    }
+
+    private long numCores_ ;
+    /**
+     * <pre>
+     * Number of cores
+     * </pre>
+     *
+     * <code>int64 num_cores = 5;</code>
+     */
+    public long getNumCores() {
+      return numCores_;
+    }
+    /**
+     * <pre>
+     * Number of cores
+     * </pre>
+     *
+     * <code>int64 num_cores = 5;</code>
+     */
+    public Builder setNumCores(long value) {
+      
+      numCores_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Number of cores
+     * </pre>
+     *
+     * <code>int64 num_cores = 5;</code>
+     */
+    public Builder clearNumCores() {
+      
+      numCores_ = 0L;
+      onChanged();
+      return this;
+    }
+
+    private com.google.protobuf.MapField<
+        java.lang.String, java.lang.String> environment_;
+    private com.google.protobuf.MapField<java.lang.String, java.lang.String>
+    internalGetEnvironment() {
+      if (environment_ == null) {
+        return com.google.protobuf.MapField.emptyMapField(
+            EnvironmentDefaultEntryHolder.defaultEntry);
+      }
+      return environment_;
+    }
+    private com.google.protobuf.MapField<java.lang.String, java.lang.String>
+    internalGetMutableEnvironment() {
+      onChanged();;
+      if (environment_ == null) {
+        environment_ = com.google.protobuf.MapField.newMapField(
+            EnvironmentDefaultEntryHolder.defaultEntry);
+      }
+      if (!environment_.isMutable()) {
+        environment_ = environment_.copy();
+      }
+      return environment_;
+    }
+
+    public int getEnvironmentCount() {
+      return internalGetEnvironment().getMap().size();
+    }
+    /**
+     * <pre>
+     * Version of the tools and libraries used with this device (e.g. gcc 4.9,
+     * cudnn 5.1)
+     * </pre>
+     *
+     * <code>map&lt;string, string&gt; environment = 6;</code>
+     */
+
+    public boolean containsEnvironment(
+        java.lang.String key) {
+      if (key == null) { throw new java.lang.NullPointerException(); }
+      return internalGetEnvironment().getMap().containsKey(key);
+    }
+    /**
+     * Use {@link #getEnvironmentMap()} instead.
+     */
+    @java.lang.Deprecated
+    public java.util.Map<java.lang.String, java.lang.String> getEnvironment() {
+      return getEnvironmentMap();
+    }
+    /**
+     * <pre>
+     * Version of the tools and libraries used with this device (e.g. gcc 4.9,
+     * cudnn 5.1)
+     * </pre>
+     *
+     * <code>map&lt;string, string&gt; environment = 6;</code>
+     */
+
+    public java.util.Map<java.lang.String, java.lang.String> getEnvironmentMap() {
+      return internalGetEnvironment().getMap();
+    }
+    /**
+     * <pre>
+     * Version of the tools and libraries used with this device (e.g. gcc 4.9,
+     * cudnn 5.1)
+     * </pre>
+     *
+     * <code>map&lt;string, string&gt; environment = 6;</code>
+     */
+
+    public java.lang.String getEnvironmentOrDefault(
+        java.lang.String key,
+        java.lang.String defaultValue) {
+      if (key == null) { throw new java.lang.NullPointerException(); }
+      java.util.Map<java.lang.String, java.lang.String> map =
+          internalGetEnvironment().getMap();
+      return map.containsKey(key) ? map.get(key) : defaultValue;
+    }
+    /**
+     * <pre>
+     * Version of the tools and libraries used with this device (e.g. gcc 4.9,
+     * cudnn 5.1)
+     * </pre>
+     *
+     * <code>map&lt;string, string&gt; environment = 6;</code>
+     */
+
+    public java.lang.String getEnvironmentOrThrow(
+        java.lang.String key) {
+      if (key == null) { throw new java.lang.NullPointerException(); }
+      java.util.Map<java.lang.String, java.lang.String> map =
+          internalGetEnvironment().getMap();
+      if (!map.containsKey(key)) {
+        throw new java.lang.IllegalArgumentException();
+      }
+      return map.get(key);
+    }
+
+    public Builder clearEnvironment() {
+      internalGetMutableEnvironment().getMutableMap()
+          .clear();
+      return this;
+    }
+    /**
+     * <pre>
+     * Version of the tools and libraries used with this device (e.g. gcc 4.9,
+     * cudnn 5.1)
+     * </pre>
+     *
+     * <code>map&lt;string, string&gt; environment = 6;</code>
+     */
+
+    public Builder removeEnvironment(
+        java.lang.String key) {
+      if (key == null) { throw new java.lang.NullPointerException(); }
+      internalGetMutableEnvironment().getMutableMap()
+          .remove(key);
+      return this;
+    }
+    /**
+     * Use alternate mutation accessors instead.
+     */
+    @java.lang.Deprecated
+    public java.util.Map<java.lang.String, java.lang.String>
+    getMutableEnvironment() {
+      return internalGetMutableEnvironment().getMutableMap();
+    }
+    /**
+     * <pre>
+     * Version of the tools and libraries used with this device (e.g. gcc 4.9,
+     * cudnn 5.1)
+     * </pre>
+     *
+     * <code>map&lt;string, string&gt; environment = 6;</code>
+     */
+    public Builder putEnvironment(
+        java.lang.String key,
+        java.lang.String value) {
+      if (key == null) { throw new java.lang.NullPointerException(); }
+      if (value == null) { throw new java.lang.NullPointerException(); }
+      internalGetMutableEnvironment().getMutableMap()
+          .put(key, value);
+      return this;
+    }
+    /**
+     * <pre>
+     * Version of the tools and libraries used with this device (e.g. gcc 4.9,
+     * cudnn 5.1)
+     * </pre>
+     *
+     * <code>map&lt;string, string&gt; environment = 6;</code>
+     */
+
+    public Builder putAllEnvironment(
+        java.util.Map<java.lang.String, java.lang.String> values) {
+      internalGetMutableEnvironment().getMutableMap()
+          .putAll(values);
+      return this;
+    }
+
+    private long numRegisters_ ;
+    /**
+     * <pre>
+     * Number of registers per core.
+     * </pre>
+     *
+     * <code>int64 num_registers = 7;</code>
+     */
+    public long getNumRegisters() {
+      return numRegisters_;
+    }
+    /**
+     * <pre>
+     * Number of registers per core.
+     * </pre>
+     *
+     * <code>int64 num_registers = 7;</code>
+     */
+    public Builder setNumRegisters(long value) {
+      
+      numRegisters_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Number of registers per core.
+     * </pre>
+     *
+     * <code>int64 num_registers = 7;</code>
+     */
+    public Builder clearNumRegisters() {
+      
+      numRegisters_ = 0L;
+      onChanged();
+      return this;
+    }
+
+    private long l1CacheSize_ ;
+    /**
+     * <pre>
+     * L1 cache size in bytes
+     * </pre>
+     *
+     * <code>int64 l1_cache_size = 8;</code>
+     */
+    public long getL1CacheSize() {
+      return l1CacheSize_;
+    }
+    /**
+     * <pre>
+     * L1 cache size in bytes
+     * </pre>
+     *
+     * <code>int64 l1_cache_size = 8;</code>
+     */
+    public Builder setL1CacheSize(long value) {
+      
+      l1CacheSize_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * L1 cache size in bytes
+     * </pre>
+     *
+     * <code>int64 l1_cache_size = 8;</code>
+     */
+    public Builder clearL1CacheSize() {
+      
+      l1CacheSize_ = 0L;
+      onChanged();
+      return this;
+    }
+
+    private long l2CacheSize_ ;
+    /**
+     * <pre>
+     * L2 cache size in bytes
+     * </pre>
+     *
+     * <code>int64 l2_cache_size = 9;</code>
+     */
+    public long getL2CacheSize() {
+      return l2CacheSize_;
+    }
+    /**
+     * <pre>
+     * L2 cache size in bytes
+     * </pre>
+     *
+     * <code>int64 l2_cache_size = 9;</code>
+     */
+    public Builder setL2CacheSize(long value) {
+      
+      l2CacheSize_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * L2 cache size in bytes
+     * </pre>
+     *
+     * <code>int64 l2_cache_size = 9;</code>
+     */
+    public Builder clearL2CacheSize() {
+      
+      l2CacheSize_ = 0L;
+      onChanged();
+      return this;
+    }
+
+    private long l3CacheSize_ ;
+    /**
+     * <pre>
+     * L3 cache size in bytes
+     * </pre>
+     *
+     * <code>int64 l3_cache_size = 10;</code>
+     */
+    public long getL3CacheSize() {
+      return l3CacheSize_;
+    }
+    /**
+     * <pre>
+     * L3 cache size in bytes
+     * </pre>
+     *
+     * <code>int64 l3_cache_size = 10;</code>
+     */
+    public Builder setL3CacheSize(long value) {
+      
+      l3CacheSize_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * L3 cache size in bytes
+     * </pre>
+     *
+     * <code>int64 l3_cache_size = 10;</code>
+     */
+    public Builder clearL3CacheSize() {
+      
+      l3CacheSize_ = 0L;
+      onChanged();
+      return this;
+    }
+
+    private long sharedMemorySizePerMultiprocessor_ ;
+    /**
+     * <pre>
+     * Shared memory size per multiprocessor in bytes. This field is
+     * applicable to GPUs only.
+     * </pre>
+     *
+     * <code>int64 shared_memory_size_per_multiprocessor = 11;</code>
+     */
+    public long getSharedMemorySizePerMultiprocessor() {
+      return sharedMemorySizePerMultiprocessor_;
+    }
+    /**
+     * <pre>
+     * Shared memory size per multiprocessor in bytes. This field is
+     * applicable to GPUs only.
+     * </pre>
+     *
+     * <code>int64 shared_memory_size_per_multiprocessor = 11;</code>
+     */
+    public Builder setSharedMemorySizePerMultiprocessor(long value) {
+      
+      sharedMemorySizePerMultiprocessor_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Shared memory size per multiprocessor in bytes. This field is
+     * applicable to GPUs only.
+     * </pre>
+     *
+     * <code>int64 shared_memory_size_per_multiprocessor = 11;</code>
+     */
+    public Builder clearSharedMemorySizePerMultiprocessor() {
+      
+      sharedMemorySizePerMultiprocessor_ = 0L;
+      onChanged();
+      return this;
+    }
+
+    private long memorySize_ ;
+    /**
+     * <pre>
+     * Memory size in bytes
+     * </pre>
+     *
+     * <code>int64 memory_size = 12;</code>
+     */
+    public long getMemorySize() {
+      return memorySize_;
+    }
+    /**
+     * <pre>
+     * Memory size in bytes
+     * </pre>
+     *
+     * <code>int64 memory_size = 12;</code>
+     */
+    public Builder setMemorySize(long value) {
+      
+      memorySize_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Memory size in bytes
+     * </pre>
+     *
+     * <code>int64 memory_size = 12;</code>
+     */
+    public Builder clearMemorySize() {
+      
+      memorySize_ = 0L;
+      onChanged();
+      return this;
+    }
+
+    private long bandwidth_ ;
+    /**
+     * <pre>
+     * Memory bandwidth in KB/s
+     * </pre>
+     *
+     * <code>int64 bandwidth = 13;</code>
+     */
+    public long getBandwidth() {
+      return bandwidth_;
+    }
+    /**
+     * <pre>
+     * Memory bandwidth in KB/s
+     * </pre>
+     *
+     * <code>int64 bandwidth = 13;</code>
+     */
+    public Builder setBandwidth(long value) {
+      
+      bandwidth_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Memory bandwidth in KB/s
+     * </pre>
+     *
+     * <code>int64 bandwidth = 13;</code>
+     */
+    public Builder clearBandwidth() {
+      
+      bandwidth_ = 0L;
+      onChanged();
+      return this;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.DeviceProperties)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.DeviceProperties)
+  private static final org.tensorflow.proto.framework.DeviceProperties DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.DeviceProperties();
+  }
+
+  public static org.tensorflow.proto.framework.DeviceProperties getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<DeviceProperties>
+      PARSER = new com.google.protobuf.AbstractParser<DeviceProperties>() {
+    @java.lang.Override
+    public DeviceProperties parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new DeviceProperties(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<DeviceProperties> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<DeviceProperties> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.framework.DeviceProperties getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DevicePropertiesOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DevicePropertiesOrBuilder.java
new file mode 100644
index 00000000000..675627a60fa
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DevicePropertiesOrBuilder.java
@@ -0,0 +1,204 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/device_properties.proto
+
+package org.tensorflow.proto.framework;
+
+public interface DevicePropertiesOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.DeviceProperties)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <pre>
+   * Device type (CPU, GPU, ...)
+   * </pre>
+   *
+   * <code>string type = 1;</code>
+   */
+  java.lang.String getType();
+  /**
+   * <pre>
+   * Device type (CPU, GPU, ...)
+   * </pre>
+   *
+   * <code>string type = 1;</code>
+   */
+  com.google.protobuf.ByteString
+      getTypeBytes();
+
+  /**
+   * <pre>
+   * Vendor (Intel, nvidia, ...)
+   * </pre>
+   *
+   * <code>string vendor = 2;</code>
+   */
+  java.lang.String getVendor();
+  /**
+   * <pre>
+   * Vendor (Intel, nvidia, ...)
+   * </pre>
+   *
+   * <code>string vendor = 2;</code>
+   */
+  com.google.protobuf.ByteString
+      getVendorBytes();
+
+  /**
+   * <pre>
+   * Model (Haswell, K40, ...)
+   * </pre>
+   *
+   * <code>string model = 3;</code>
+   */
+  java.lang.String getModel();
+  /**
+   * <pre>
+   * Model (Haswell, K40, ...)
+   * </pre>
+   *
+   * <code>string model = 3;</code>
+   */
+  com.google.protobuf.ByteString
+      getModelBytes();
+
+  /**
+   * <pre>
+   * Core Frequency in Mhz
+   * </pre>
+   *
+   * <code>int64 frequency = 4;</code>
+   */
+  long getFrequency();
+
+  /**
+   * <pre>
+   * Number of cores
+   * </pre>
+   *
+   * <code>int64 num_cores = 5;</code>
+   */
+  long getNumCores();
+
+  /**
+   * <pre>
+   * Version of the tools and libraries used with this device (e.g. gcc 4.9,
+   * cudnn 5.1)
+   * </pre>
+   *
+   * <code>map&lt;string, string&gt; environment = 6;</code>
+   */
+  int getEnvironmentCount();
+  /**
+   * <pre>
+   * Version of the tools and libraries used with this device (e.g. gcc 4.9,
+   * cudnn 5.1)
+   * </pre>
+   *
+   * <code>map&lt;string, string&gt; environment = 6;</code>
+   */
+  boolean containsEnvironment(
+      java.lang.String key);
+  /**
+   * Use {@link #getEnvironmentMap()} instead.
+   */
+  @java.lang.Deprecated
+  java.util.Map<java.lang.String, java.lang.String>
+  getEnvironment();
+  /**
+   * <pre>
+   * Version of the tools and libraries used with this device (e.g. gcc 4.9,
+   * cudnn 5.1)
+   * </pre>
+   *
+   * <code>map&lt;string, string&gt; environment = 6;</code>
+   */
+  java.util.Map<java.lang.String, java.lang.String>
+  getEnvironmentMap();
+  /**
+   * <pre>
+   * Version of the tools and libraries used with this device (e.g. gcc 4.9,
+   * cudnn 5.1)
+   * </pre>
+   *
+   * <code>map&lt;string, string&gt; environment = 6;</code>
+   */
+
+  java.lang.String getEnvironmentOrDefault(
+      java.lang.String key,
+      java.lang.String defaultValue);
+  /**
+   * <pre>
+   * Version of the tools and libraries used with this device (e.g. gcc 4.9,
+   * cudnn 5.1)
+   * </pre>
+   *
+   * <code>map&lt;string, string&gt; environment = 6;</code>
+   */
+
+  java.lang.String getEnvironmentOrThrow(
+      java.lang.String key);
+
+  /**
+   * <pre>
+   * Number of registers per core.
+   * </pre>
+   *
+   * <code>int64 num_registers = 7;</code>
+   */
+  long getNumRegisters();
+
+  /**
+   * <pre>
+   * L1 cache size in bytes
+   * </pre>
+   *
+   * <code>int64 l1_cache_size = 8;</code>
+   */
+  long getL1CacheSize();
+
+  /**
+   * <pre>
+   * L2 cache size in bytes
+   * </pre>
+   *
+   * <code>int64 l2_cache_size = 9;</code>
+   */
+  long getL2CacheSize();
+
+  /**
+   * <pre>
+   * L3 cache size in bytes
+   * </pre>
+   *
+   * <code>int64 l3_cache_size = 10;</code>
+   */
+  long getL3CacheSize();
+
+  /**
+   * <pre>
+   * Shared memory size per multiprocessor in bytes. This field is
+   * applicable to GPUs only.
+   * </pre>
+   *
+   * <code>int64 shared_memory_size_per_multiprocessor = 11;</code>
+   */
+  long getSharedMemorySizePerMultiprocessor();
+
+  /**
+   * <pre>
+   * Memory size in bytes
+   * </pre>
+   *
+   * <code>int64 memory_size = 12;</code>
+   */
+  long getMemorySize();
+
+  /**
+   * <pre>
+   * Memory bandwidth in KB/s
+   * </pre>
+   *
+   * <code>int64 bandwidth = 13;</code>
+   */
+  long getBandwidth();
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DevicePropertiesProtos.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DevicePropertiesProtos.java
new file mode 100644
index 00000000000..7f6f979fca2
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DevicePropertiesProtos.java
@@ -0,0 +1,85 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/device_properties.proto
+
+package org.tensorflow.proto.framework;
+
+public final class DevicePropertiesProtos {
+  private DevicePropertiesProtos() {}
+  public static void registerAllExtensions(
+      com.google.protobuf.ExtensionRegistryLite registry) {
+  }
+
+  public static void registerAllExtensions(
+      com.google.protobuf.ExtensionRegistry registry) {
+    registerAllExtensions(
+        (com.google.protobuf.ExtensionRegistryLite) registry);
+  }
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_DeviceProperties_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_DeviceProperties_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_DeviceProperties_EnvironmentEntry_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_DeviceProperties_EnvironmentEntry_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_NamedDevice_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_NamedDevice_fieldAccessorTable;
+
+  public static com.google.protobuf.Descriptors.FileDescriptor
+      getDescriptor() {
+    return descriptor;
+  }
+  private static  com.google.protobuf.Descriptors.FileDescriptor
+      descriptor;
+  static {
+    java.lang.String[] descriptorData = {
+      "\n0tensorflow/core/protobuf/device_proper" +
+      "ties.proto\022\ntensorflow\"\220\003\n\020DevicePropert" +
+      "ies\022\014\n\004type\030\001 \001(\t\022\016\n\006vendor\030\002 \001(\t\022\r\n\005mod" +
+      "el\030\003 \001(\t\022\021\n\tfrequency\030\004 \001(\003\022\021\n\tnum_cores" +
+      "\030\005 \001(\003\022B\n\013environment\030\006 \003(\0132-.tensorflow" +
+      ".DeviceProperties.EnvironmentEntry\022\025\n\rnu" +
+      "m_registers\030\007 \001(\003\022\025\n\rl1_cache_size\030\010 \001(\003" +
+      "\022\025\n\rl2_cache_size\030\t \001(\003\022\025\n\rl3_cache_size" +
+      "\030\n \001(\003\022-\n%shared_memory_size_per_multipr" +
+      "ocessor\030\013 \001(\003\022\023\n\013memory_size\030\014 \001(\003\022\021\n\tba" +
+      "ndwidth\030\r \001(\003\0322\n\020EnvironmentEntry\022\013\n\003key" +
+      "\030\001 \001(\t\022\r\n\005value\030\002 \001(\t:\0028\001\"M\n\013NamedDevice" +
+      "\022\014\n\004name\030\001 \001(\t\0220\n\nproperties\030\002 \001(\0132\034.ten" +
+      "sorflow.DevicePropertiesB\207\001\n\036org.tensorf" +
+      "low.proto.frameworkB\026DevicePropertiesPro" +
+      "tosP\001ZHgithub.com/tensorflow/tensorflow/" +
+      "tensorflow/go/core/core_protos_go_proto\370" +
+      "\001\001b\006proto3"
+    };
+    descriptor = com.google.protobuf.Descriptors.FileDescriptor
+      .internalBuildGeneratedFileFrom(descriptorData,
+        new com.google.protobuf.Descriptors.FileDescriptor[] {
+        });
+    internal_static_tensorflow_DeviceProperties_descriptor =
+      getDescriptor().getMessageTypes().get(0);
+    internal_static_tensorflow_DeviceProperties_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_DeviceProperties_descriptor,
+        new java.lang.String[] { "Type", "Vendor", "Model", "Frequency", "NumCores", "Environment", "NumRegisters", "L1CacheSize", "L2CacheSize", "L3CacheSize", "SharedMemorySizePerMultiprocessor", "MemorySize", "Bandwidth", });
+    internal_static_tensorflow_DeviceProperties_EnvironmentEntry_descriptor =
+      internal_static_tensorflow_DeviceProperties_descriptor.getNestedTypes().get(0);
+    internal_static_tensorflow_DeviceProperties_EnvironmentEntry_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_DeviceProperties_EnvironmentEntry_descriptor,
+        new java.lang.String[] { "Key", "Value", });
+    internal_static_tensorflow_NamedDevice_descriptor =
+      getDescriptor().getMessageTypes().get(1);
+    internal_static_tensorflow_NamedDevice_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_NamedDevice_descriptor,
+        new java.lang.String[] { "Name", "Properties", });
+  }
+
+  // @@protoc_insertion_point(outer_class_scope)
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DeviceStepStats.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DeviceStepStats.java
new file mode 100644
index 00000000000..8da65e2b3d5
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DeviceStepStats.java
@@ -0,0 +1,1209 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/framework/step_stats.proto
+
+package org.tensorflow.proto.framework;
+
+/**
+ * Protobuf type {@code tensorflow.DeviceStepStats}
+ */
+public  final class DeviceStepStats extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.DeviceStepStats)
+    DeviceStepStatsOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use DeviceStepStats.newBuilder() to construct.
+  private DeviceStepStats(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private DeviceStepStats() {
+    device_ = "";
+    nodeStats_ = java.util.Collections.emptyList();
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new DeviceStepStats();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private DeviceStepStats(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    int mutable_bitField0_ = 0;
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 10: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            device_ = s;
+            break;
+          }
+          case 18: {
+            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
+              nodeStats_ = new java.util.ArrayList<org.tensorflow.proto.framework.NodeExecStats>();
+              mutable_bitField0_ |= 0x00000001;
+            }
+            nodeStats_.add(
+                input.readMessage(org.tensorflow.proto.framework.NodeExecStats.parser(), extensionRegistry));
+            break;
+          }
+          case 26: {
+            if (!((mutable_bitField0_ & 0x00000002) != 0)) {
+              threadNames_ = com.google.protobuf.MapField.newMapField(
+                  ThreadNamesDefaultEntryHolder.defaultEntry);
+              mutable_bitField0_ |= 0x00000002;
+            }
+            com.google.protobuf.MapEntry<java.lang.Integer, java.lang.String>
+            threadNames__ = input.readMessage(
+                ThreadNamesDefaultEntryHolder.defaultEntry.getParserForType(), extensionRegistry);
+            threadNames_.getMutableMap().put(
+                threadNames__.getKey(), threadNames__.getValue());
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      if (((mutable_bitField0_ & 0x00000001) != 0)) {
+        nodeStats_ = java.util.Collections.unmodifiableList(nodeStats_);
+      }
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.framework.StepStatsProtos.internal_static_tensorflow_DeviceStepStats_descriptor;
+  }
+
+  @SuppressWarnings({"rawtypes"})
+  @java.lang.Override
+  protected com.google.protobuf.MapField internalGetMapField(
+      int number) {
+    switch (number) {
+      case 3:
+        return internalGetThreadNames();
+      default:
+        throw new RuntimeException(
+            "Invalid map field number: " + number);
+    }
+  }
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.framework.StepStatsProtos.internal_static_tensorflow_DeviceStepStats_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.framework.DeviceStepStats.class, org.tensorflow.proto.framework.DeviceStepStats.Builder.class);
+  }
+
+  public static final int DEVICE_FIELD_NUMBER = 1;
+  private volatile java.lang.Object device_;
+  /**
+   * <code>string device = 1;</code>
+   */
+  public java.lang.String getDevice() {
+    java.lang.Object ref = device_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      device_ = s;
+      return s;
+    }
+  }
+  /**
+   * <code>string device = 1;</code>
+   */
+  public com.google.protobuf.ByteString
+      getDeviceBytes() {
+    java.lang.Object ref = device_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      device_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int NODE_STATS_FIELD_NUMBER = 2;
+  private java.util.List<org.tensorflow.proto.framework.NodeExecStats> nodeStats_;
+  /**
+   * <code>repeated .tensorflow.NodeExecStats node_stats = 2;</code>
+   */
+  public java.util.List<org.tensorflow.proto.framework.NodeExecStats> getNodeStatsList() {
+    return nodeStats_;
+  }
+  /**
+   * <code>repeated .tensorflow.NodeExecStats node_stats = 2;</code>
+   */
+  public java.util.List<? extends org.tensorflow.proto.framework.NodeExecStatsOrBuilder> 
+      getNodeStatsOrBuilderList() {
+    return nodeStats_;
+  }
+  /**
+   * <code>repeated .tensorflow.NodeExecStats node_stats = 2;</code>
+   */
+  public int getNodeStatsCount() {
+    return nodeStats_.size();
+  }
+  /**
+   * <code>repeated .tensorflow.NodeExecStats node_stats = 2;</code>
+   */
+  public org.tensorflow.proto.framework.NodeExecStats getNodeStats(int index) {
+    return nodeStats_.get(index);
+  }
+  /**
+   * <code>repeated .tensorflow.NodeExecStats node_stats = 2;</code>
+   */
+  public org.tensorflow.proto.framework.NodeExecStatsOrBuilder getNodeStatsOrBuilder(
+      int index) {
+    return nodeStats_.get(index);
+  }
+
+  public static final int THREAD_NAMES_FIELD_NUMBER = 3;
+  private static final class ThreadNamesDefaultEntryHolder {
+    static final com.google.protobuf.MapEntry<
+        java.lang.Integer, java.lang.String> defaultEntry =
+            com.google.protobuf.MapEntry
+            .<java.lang.Integer, java.lang.String>newDefaultInstance(
+                org.tensorflow.proto.framework.StepStatsProtos.internal_static_tensorflow_DeviceStepStats_ThreadNamesEntry_descriptor, 
+                com.google.protobuf.WireFormat.FieldType.UINT32,
+                0,
+                com.google.protobuf.WireFormat.FieldType.STRING,
+                "");
+  }
+  private com.google.protobuf.MapField<
+      java.lang.Integer, java.lang.String> threadNames_;
+  private com.google.protobuf.MapField<java.lang.Integer, java.lang.String>
+  internalGetThreadNames() {
+    if (threadNames_ == null) {
+      return com.google.protobuf.MapField.emptyMapField(
+          ThreadNamesDefaultEntryHolder.defaultEntry);
+    }
+    return threadNames_;
+  }
+
+  public int getThreadNamesCount() {
+    return internalGetThreadNames().getMap().size();
+  }
+  /**
+   * <pre>
+   * Its key is thread id.
+   * </pre>
+   *
+   * <code>map&lt;uint32, string&gt; thread_names = 3;</code>
+   */
+
+  public boolean containsThreadNames(
+      int key) {
+    
+    return internalGetThreadNames().getMap().containsKey(key);
+  }
+  /**
+   * Use {@link #getThreadNamesMap()} instead.
+   */
+  @java.lang.Deprecated
+  public java.util.Map<java.lang.Integer, java.lang.String> getThreadNames() {
+    return getThreadNamesMap();
+  }
+  /**
+   * <pre>
+   * Its key is thread id.
+   * </pre>
+   *
+   * <code>map&lt;uint32, string&gt; thread_names = 3;</code>
+   */
+
+  public java.util.Map<java.lang.Integer, java.lang.String> getThreadNamesMap() {
+    return internalGetThreadNames().getMap();
+  }
+  /**
+   * <pre>
+   * Its key is thread id.
+   * </pre>
+   *
+   * <code>map&lt;uint32, string&gt; thread_names = 3;</code>
+   */
+
+  public java.lang.String getThreadNamesOrDefault(
+      int key,
+      java.lang.String defaultValue) {
+    
+    java.util.Map<java.lang.Integer, java.lang.String> map =
+        internalGetThreadNames().getMap();
+    return map.containsKey(key) ? map.get(key) : defaultValue;
+  }
+  /**
+   * <pre>
+   * Its key is thread id.
+   * </pre>
+   *
+   * <code>map&lt;uint32, string&gt; thread_names = 3;</code>
+   */
+
+  public java.lang.String getThreadNamesOrThrow(
+      int key) {
+    
+    java.util.Map<java.lang.Integer, java.lang.String> map =
+        internalGetThreadNames().getMap();
+    if (!map.containsKey(key)) {
+      throw new java.lang.IllegalArgumentException();
+    }
+    return map.get(key);
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    if (!getDeviceBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 1, device_);
+    }
+    for (int i = 0; i < nodeStats_.size(); i++) {
+      output.writeMessage(2, nodeStats_.get(i));
+    }
+    com.google.protobuf.GeneratedMessageV3
+      .serializeIntegerMapTo(
+        output,
+        internalGetThreadNames(),
+        ThreadNamesDefaultEntryHolder.defaultEntry,
+        3);
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (!getDeviceBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, device_);
+    }
+    for (int i = 0; i < nodeStats_.size(); i++) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(2, nodeStats_.get(i));
+    }
+    for (java.util.Map.Entry<java.lang.Integer, java.lang.String> entry
+         : internalGetThreadNames().getMap().entrySet()) {
+      com.google.protobuf.MapEntry<java.lang.Integer, java.lang.String>
+      threadNames__ = ThreadNamesDefaultEntryHolder.defaultEntry.newBuilderForType()
+          .setKey(entry.getKey())
+          .setValue(entry.getValue())
+          .build();
+      size += com.google.protobuf.CodedOutputStream
+          .computeMessageSize(3, threadNames__);
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.framework.DeviceStepStats)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.framework.DeviceStepStats other = (org.tensorflow.proto.framework.DeviceStepStats) obj;
+
+    if (!getDevice()
+        .equals(other.getDevice())) return false;
+    if (!getNodeStatsList()
+        .equals(other.getNodeStatsList())) return false;
+    if (!internalGetThreadNames().equals(
+        other.internalGetThreadNames())) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    hash = (37 * hash) + DEVICE_FIELD_NUMBER;
+    hash = (53 * hash) + getDevice().hashCode();
+    if (getNodeStatsCount() > 0) {
+      hash = (37 * hash) + NODE_STATS_FIELD_NUMBER;
+      hash = (53 * hash) + getNodeStatsList().hashCode();
+    }
+    if (!internalGetThreadNames().getMap().isEmpty()) {
+      hash = (37 * hash) + THREAD_NAMES_FIELD_NUMBER;
+      hash = (53 * hash) + internalGetThreadNames().hashCode();
+    }
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.framework.DeviceStepStats parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.DeviceStepStats parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.DeviceStepStats parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.DeviceStepStats parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.DeviceStepStats parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.DeviceStepStats parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.DeviceStepStats parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.DeviceStepStats parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.DeviceStepStats parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.DeviceStepStats parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.DeviceStepStats parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.DeviceStepStats parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.framework.DeviceStepStats prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * Protobuf type {@code tensorflow.DeviceStepStats}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.DeviceStepStats)
+      org.tensorflow.proto.framework.DeviceStepStatsOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.StepStatsProtos.internal_static_tensorflow_DeviceStepStats_descriptor;
+    }
+
+    @SuppressWarnings({"rawtypes"})
+    protected com.google.protobuf.MapField internalGetMapField(
+        int number) {
+      switch (number) {
+        case 3:
+          return internalGetThreadNames();
+        default:
+          throw new RuntimeException(
+              "Invalid map field number: " + number);
+      }
+    }
+    @SuppressWarnings({"rawtypes"})
+    protected com.google.protobuf.MapField internalGetMutableMapField(
+        int number) {
+      switch (number) {
+        case 3:
+          return internalGetMutableThreadNames();
+        default:
+          throw new RuntimeException(
+              "Invalid map field number: " + number);
+      }
+    }
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.StepStatsProtos.internal_static_tensorflow_DeviceStepStats_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.DeviceStepStats.class, org.tensorflow.proto.framework.DeviceStepStats.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.framework.DeviceStepStats.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+        getNodeStatsFieldBuilder();
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      device_ = "";
+
+      if (nodeStatsBuilder_ == null) {
+        nodeStats_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000001);
+      } else {
+        nodeStatsBuilder_.clear();
+      }
+      internalGetMutableThreadNames().clear();
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.framework.StepStatsProtos.internal_static_tensorflow_DeviceStepStats_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.DeviceStepStats getDefaultInstanceForType() {
+      return org.tensorflow.proto.framework.DeviceStepStats.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.DeviceStepStats build() {
+      org.tensorflow.proto.framework.DeviceStepStats result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.DeviceStepStats buildPartial() {
+      org.tensorflow.proto.framework.DeviceStepStats result = new org.tensorflow.proto.framework.DeviceStepStats(this);
+      int from_bitField0_ = bitField0_;
+      result.device_ = device_;
+      if (nodeStatsBuilder_ == null) {
+        if (((bitField0_ & 0x00000001) != 0)) {
+          nodeStats_ = java.util.Collections.unmodifiableList(nodeStats_);
+          bitField0_ = (bitField0_ & ~0x00000001);
+        }
+        result.nodeStats_ = nodeStats_;
+      } else {
+        result.nodeStats_ = nodeStatsBuilder_.build();
+      }
+      result.threadNames_ = internalGetThreadNames();
+      result.threadNames_.makeImmutable();
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.framework.DeviceStepStats) {
+        return mergeFrom((org.tensorflow.proto.framework.DeviceStepStats)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.framework.DeviceStepStats other) {
+      if (other == org.tensorflow.proto.framework.DeviceStepStats.getDefaultInstance()) return this;
+      if (!other.getDevice().isEmpty()) {
+        device_ = other.device_;
+        onChanged();
+      }
+      if (nodeStatsBuilder_ == null) {
+        if (!other.nodeStats_.isEmpty()) {
+          if (nodeStats_.isEmpty()) {
+            nodeStats_ = other.nodeStats_;
+            bitField0_ = (bitField0_ & ~0x00000001);
+          } else {
+            ensureNodeStatsIsMutable();
+            nodeStats_.addAll(other.nodeStats_);
+          }
+          onChanged();
+        }
+      } else {
+        if (!other.nodeStats_.isEmpty()) {
+          if (nodeStatsBuilder_.isEmpty()) {
+            nodeStatsBuilder_.dispose();
+            nodeStatsBuilder_ = null;
+            nodeStats_ = other.nodeStats_;
+            bitField0_ = (bitField0_ & ~0x00000001);
+            nodeStatsBuilder_ = 
+              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
+                 getNodeStatsFieldBuilder() : null;
+          } else {
+            nodeStatsBuilder_.addAllMessages(other.nodeStats_);
+          }
+        }
+      }
+      internalGetMutableThreadNames().mergeFrom(
+          other.internalGetThreadNames());
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.framework.DeviceStepStats parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.framework.DeviceStepStats) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+    private int bitField0_;
+
+    private java.lang.Object device_ = "";
+    /**
+     * <code>string device = 1;</code>
+     */
+    public java.lang.String getDevice() {
+      java.lang.Object ref = device_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        device_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <code>string device = 1;</code>
+     */
+    public com.google.protobuf.ByteString
+        getDeviceBytes() {
+      java.lang.Object ref = device_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        device_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <code>string device = 1;</code>
+     */
+    public Builder setDevice(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      device_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>string device = 1;</code>
+     */
+    public Builder clearDevice() {
+      
+      device_ = getDefaultInstance().getDevice();
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>string device = 1;</code>
+     */
+    public Builder setDeviceBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      device_ = value;
+      onChanged();
+      return this;
+    }
+
+    private java.util.List<org.tensorflow.proto.framework.NodeExecStats> nodeStats_ =
+      java.util.Collections.emptyList();
+    private void ensureNodeStatsIsMutable() {
+      if (!((bitField0_ & 0x00000001) != 0)) {
+        nodeStats_ = new java.util.ArrayList<org.tensorflow.proto.framework.NodeExecStats>(nodeStats_);
+        bitField0_ |= 0x00000001;
+       }
+    }
+
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.framework.NodeExecStats, org.tensorflow.proto.framework.NodeExecStats.Builder, org.tensorflow.proto.framework.NodeExecStatsOrBuilder> nodeStatsBuilder_;
+
+    /**
+     * <code>repeated .tensorflow.NodeExecStats node_stats = 2;</code>
+     */
+    public java.util.List<org.tensorflow.proto.framework.NodeExecStats> getNodeStatsList() {
+      if (nodeStatsBuilder_ == null) {
+        return java.util.Collections.unmodifiableList(nodeStats_);
+      } else {
+        return nodeStatsBuilder_.getMessageList();
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.NodeExecStats node_stats = 2;</code>
+     */
+    public int getNodeStatsCount() {
+      if (nodeStatsBuilder_ == null) {
+        return nodeStats_.size();
+      } else {
+        return nodeStatsBuilder_.getCount();
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.NodeExecStats node_stats = 2;</code>
+     */
+    public org.tensorflow.proto.framework.NodeExecStats getNodeStats(int index) {
+      if (nodeStatsBuilder_ == null) {
+        return nodeStats_.get(index);
+      } else {
+        return nodeStatsBuilder_.getMessage(index);
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.NodeExecStats node_stats = 2;</code>
+     */
+    public Builder setNodeStats(
+        int index, org.tensorflow.proto.framework.NodeExecStats value) {
+      if (nodeStatsBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureNodeStatsIsMutable();
+        nodeStats_.set(index, value);
+        onChanged();
+      } else {
+        nodeStatsBuilder_.setMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.NodeExecStats node_stats = 2;</code>
+     */
+    public Builder setNodeStats(
+        int index, org.tensorflow.proto.framework.NodeExecStats.Builder builderForValue) {
+      if (nodeStatsBuilder_ == null) {
+        ensureNodeStatsIsMutable();
+        nodeStats_.set(index, builderForValue.build());
+        onChanged();
+      } else {
+        nodeStatsBuilder_.setMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.NodeExecStats node_stats = 2;</code>
+     */
+    public Builder addNodeStats(org.tensorflow.proto.framework.NodeExecStats value) {
+      if (nodeStatsBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureNodeStatsIsMutable();
+        nodeStats_.add(value);
+        onChanged();
+      } else {
+        nodeStatsBuilder_.addMessage(value);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.NodeExecStats node_stats = 2;</code>
+     */
+    public Builder addNodeStats(
+        int index, org.tensorflow.proto.framework.NodeExecStats value) {
+      if (nodeStatsBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureNodeStatsIsMutable();
+        nodeStats_.add(index, value);
+        onChanged();
+      } else {
+        nodeStatsBuilder_.addMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.NodeExecStats node_stats = 2;</code>
+     */
+    public Builder addNodeStats(
+        org.tensorflow.proto.framework.NodeExecStats.Builder builderForValue) {
+      if (nodeStatsBuilder_ == null) {
+        ensureNodeStatsIsMutable();
+        nodeStats_.add(builderForValue.build());
+        onChanged();
+      } else {
+        nodeStatsBuilder_.addMessage(builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.NodeExecStats node_stats = 2;</code>
+     */
+    public Builder addNodeStats(
+        int index, org.tensorflow.proto.framework.NodeExecStats.Builder builderForValue) {
+      if (nodeStatsBuilder_ == null) {
+        ensureNodeStatsIsMutable();
+        nodeStats_.add(index, builderForValue.build());
+        onChanged();
+      } else {
+        nodeStatsBuilder_.addMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.NodeExecStats node_stats = 2;</code>
+     */
+    public Builder addAllNodeStats(
+        java.lang.Iterable<? extends org.tensorflow.proto.framework.NodeExecStats> values) {
+      if (nodeStatsBuilder_ == null) {
+        ensureNodeStatsIsMutable();
+        com.google.protobuf.AbstractMessageLite.Builder.addAll(
+            values, nodeStats_);
+        onChanged();
+      } else {
+        nodeStatsBuilder_.addAllMessages(values);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.NodeExecStats node_stats = 2;</code>
+     */
+    public Builder clearNodeStats() {
+      if (nodeStatsBuilder_ == null) {
+        nodeStats_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000001);
+        onChanged();
+      } else {
+        nodeStatsBuilder_.clear();
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.NodeExecStats node_stats = 2;</code>
+     */
+    public Builder removeNodeStats(int index) {
+      if (nodeStatsBuilder_ == null) {
+        ensureNodeStatsIsMutable();
+        nodeStats_.remove(index);
+        onChanged();
+      } else {
+        nodeStatsBuilder_.remove(index);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.NodeExecStats node_stats = 2;</code>
+     */
+    public org.tensorflow.proto.framework.NodeExecStats.Builder getNodeStatsBuilder(
+        int index) {
+      return getNodeStatsFieldBuilder().getBuilder(index);
+    }
+    /**
+     * <code>repeated .tensorflow.NodeExecStats node_stats = 2;</code>
+     */
+    public org.tensorflow.proto.framework.NodeExecStatsOrBuilder getNodeStatsOrBuilder(
+        int index) {
+      if (nodeStatsBuilder_ == null) {
+        return nodeStats_.get(index);  } else {
+        return nodeStatsBuilder_.getMessageOrBuilder(index);
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.NodeExecStats node_stats = 2;</code>
+     */
+    public java.util.List<? extends org.tensorflow.proto.framework.NodeExecStatsOrBuilder> 
+         getNodeStatsOrBuilderList() {
+      if (nodeStatsBuilder_ != null) {
+        return nodeStatsBuilder_.getMessageOrBuilderList();
+      } else {
+        return java.util.Collections.unmodifiableList(nodeStats_);
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.NodeExecStats node_stats = 2;</code>
+     */
+    public org.tensorflow.proto.framework.NodeExecStats.Builder addNodeStatsBuilder() {
+      return getNodeStatsFieldBuilder().addBuilder(
+          org.tensorflow.proto.framework.NodeExecStats.getDefaultInstance());
+    }
+    /**
+     * <code>repeated .tensorflow.NodeExecStats node_stats = 2;</code>
+     */
+    public org.tensorflow.proto.framework.NodeExecStats.Builder addNodeStatsBuilder(
+        int index) {
+      return getNodeStatsFieldBuilder().addBuilder(
+          index, org.tensorflow.proto.framework.NodeExecStats.getDefaultInstance());
+    }
+    /**
+     * <code>repeated .tensorflow.NodeExecStats node_stats = 2;</code>
+     */
+    public java.util.List<org.tensorflow.proto.framework.NodeExecStats.Builder> 
+         getNodeStatsBuilderList() {
+      return getNodeStatsFieldBuilder().getBuilderList();
+    }
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.framework.NodeExecStats, org.tensorflow.proto.framework.NodeExecStats.Builder, org.tensorflow.proto.framework.NodeExecStatsOrBuilder> 
+        getNodeStatsFieldBuilder() {
+      if (nodeStatsBuilder_ == null) {
+        nodeStatsBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
+            org.tensorflow.proto.framework.NodeExecStats, org.tensorflow.proto.framework.NodeExecStats.Builder, org.tensorflow.proto.framework.NodeExecStatsOrBuilder>(
+                nodeStats_,
+                ((bitField0_ & 0x00000001) != 0),
+                getParentForChildren(),
+                isClean());
+        nodeStats_ = null;
+      }
+      return nodeStatsBuilder_;
+    }
+
+    private com.google.protobuf.MapField<
+        java.lang.Integer, java.lang.String> threadNames_;
+    private com.google.protobuf.MapField<java.lang.Integer, java.lang.String>
+    internalGetThreadNames() {
+      if (threadNames_ == null) {
+        return com.google.protobuf.MapField.emptyMapField(
+            ThreadNamesDefaultEntryHolder.defaultEntry);
+      }
+      return threadNames_;
+    }
+    private com.google.protobuf.MapField<java.lang.Integer, java.lang.String>
+    internalGetMutableThreadNames() {
+      onChanged();;
+      if (threadNames_ == null) {
+        threadNames_ = com.google.protobuf.MapField.newMapField(
+            ThreadNamesDefaultEntryHolder.defaultEntry);
+      }
+      if (!threadNames_.isMutable()) {
+        threadNames_ = threadNames_.copy();
+      }
+      return threadNames_;
+    }
+
+    public int getThreadNamesCount() {
+      return internalGetThreadNames().getMap().size();
+    }
+    /**
+     * <pre>
+     * Its key is thread id.
+     * </pre>
+     *
+     * <code>map&lt;uint32, string&gt; thread_names = 3;</code>
+     */
+
+    public boolean containsThreadNames(
+        int key) {
+      
+      return internalGetThreadNames().getMap().containsKey(key);
+    }
+    /**
+     * Use {@link #getThreadNamesMap()} instead.
+     */
+    @java.lang.Deprecated
+    public java.util.Map<java.lang.Integer, java.lang.String> getThreadNames() {
+      return getThreadNamesMap();
+    }
+    /**
+     * <pre>
+     * Its key is thread id.
+     * </pre>
+     *
+     * <code>map&lt;uint32, string&gt; thread_names = 3;</code>
+     */
+
+    public java.util.Map<java.lang.Integer, java.lang.String> getThreadNamesMap() {
+      return internalGetThreadNames().getMap();
+    }
+    /**
+     * <pre>
+     * Its key is thread id.
+     * </pre>
+     *
+     * <code>map&lt;uint32, string&gt; thread_names = 3;</code>
+     */
+
+    public java.lang.String getThreadNamesOrDefault(
+        int key,
+        java.lang.String defaultValue) {
+      
+      java.util.Map<java.lang.Integer, java.lang.String> map =
+          internalGetThreadNames().getMap();
+      return map.containsKey(key) ? map.get(key) : defaultValue;
+    }
+    /**
+     * <pre>
+     * Its key is thread id.
+     * </pre>
+     *
+     * <code>map&lt;uint32, string&gt; thread_names = 3;</code>
+     */
+
+    public java.lang.String getThreadNamesOrThrow(
+        int key) {
+      
+      java.util.Map<java.lang.Integer, java.lang.String> map =
+          internalGetThreadNames().getMap();
+      if (!map.containsKey(key)) {
+        throw new java.lang.IllegalArgumentException();
+      }
+      return map.get(key);
+    }
+
+    public Builder clearThreadNames() {
+      internalGetMutableThreadNames().getMutableMap()
+          .clear();
+      return this;
+    }
+    /**
+     * <pre>
+     * Its key is thread id.
+     * </pre>
+     *
+     * <code>map&lt;uint32, string&gt; thread_names = 3;</code>
+     */
+
+    public Builder removeThreadNames(
+        int key) {
+      
+      internalGetMutableThreadNames().getMutableMap()
+          .remove(key);
+      return this;
+    }
+    /**
+     * Use alternate mutation accessors instead.
+     */
+    @java.lang.Deprecated
+    public java.util.Map<java.lang.Integer, java.lang.String>
+    getMutableThreadNames() {
+      return internalGetMutableThreadNames().getMutableMap();
+    }
+    /**
+     * <pre>
+     * Its key is thread id.
+     * </pre>
+     *
+     * <code>map&lt;uint32, string&gt; thread_names = 3;</code>
+     */
+    public Builder putThreadNames(
+        int key,
+        java.lang.String value) {
+      
+      if (value == null) { throw new java.lang.NullPointerException(); }
+      internalGetMutableThreadNames().getMutableMap()
+          .put(key, value);
+      return this;
+    }
+    /**
+     * <pre>
+     * Its key is thread id.
+     * </pre>
+     *
+     * <code>map&lt;uint32, string&gt; thread_names = 3;</code>
+     */
+
+    public Builder putAllThreadNames(
+        java.util.Map<java.lang.Integer, java.lang.String> values) {
+      internalGetMutableThreadNames().getMutableMap()
+          .putAll(values);
+      return this;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.DeviceStepStats)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.DeviceStepStats)
+  private static final org.tensorflow.proto.framework.DeviceStepStats DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.DeviceStepStats();
+  }
+
+  public static org.tensorflow.proto.framework.DeviceStepStats getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<DeviceStepStats>
+      PARSER = new com.google.protobuf.AbstractParser<DeviceStepStats>() {
+    @java.lang.Override
+    public DeviceStepStats parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new DeviceStepStats(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<DeviceStepStats> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<DeviceStepStats> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.framework.DeviceStepStats getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DeviceStepStatsOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DeviceStepStatsOrBuilder.java
new file mode 100644
index 00000000000..d53ab25f2e0
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DeviceStepStatsOrBuilder.java
@@ -0,0 +1,97 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/framework/step_stats.proto
+
+package org.tensorflow.proto.framework;
+
+public interface DeviceStepStatsOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.DeviceStepStats)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <code>string device = 1;</code>
+   */
+  java.lang.String getDevice();
+  /**
+   * <code>string device = 1;</code>
+   */
+  com.google.protobuf.ByteString
+      getDeviceBytes();
+
+  /**
+   * <code>repeated .tensorflow.NodeExecStats node_stats = 2;</code>
+   */
+  java.util.List<org.tensorflow.proto.framework.NodeExecStats> 
+      getNodeStatsList();
+  /**
+   * <code>repeated .tensorflow.NodeExecStats node_stats = 2;</code>
+   */
+  org.tensorflow.proto.framework.NodeExecStats getNodeStats(int index);
+  /**
+   * <code>repeated .tensorflow.NodeExecStats node_stats = 2;</code>
+   */
+  int getNodeStatsCount();
+  /**
+   * <code>repeated .tensorflow.NodeExecStats node_stats = 2;</code>
+   */
+  java.util.List<? extends org.tensorflow.proto.framework.NodeExecStatsOrBuilder> 
+      getNodeStatsOrBuilderList();
+  /**
+   * <code>repeated .tensorflow.NodeExecStats node_stats = 2;</code>
+   */
+  org.tensorflow.proto.framework.NodeExecStatsOrBuilder getNodeStatsOrBuilder(
+      int index);
+
+  /**
+   * <pre>
+   * Its key is thread id.
+   * </pre>
+   *
+   * <code>map&lt;uint32, string&gt; thread_names = 3;</code>
+   */
+  int getThreadNamesCount();
+  /**
+   * <pre>
+   * Its key is thread id.
+   * </pre>
+   *
+   * <code>map&lt;uint32, string&gt; thread_names = 3;</code>
+   */
+  boolean containsThreadNames(
+      int key);
+  /**
+   * Use {@link #getThreadNamesMap()} instead.
+   */
+  @java.lang.Deprecated
+  java.util.Map<java.lang.Integer, java.lang.String>
+  getThreadNames();
+  /**
+   * <pre>
+   * Its key is thread id.
+   * </pre>
+   *
+   * <code>map&lt;uint32, string&gt; thread_names = 3;</code>
+   */
+  java.util.Map<java.lang.Integer, java.lang.String>
+  getThreadNamesMap();
+  /**
+   * <pre>
+   * Its key is thread id.
+   * </pre>
+   *
+   * <code>map&lt;uint32, string&gt; thread_names = 3;</code>
+   */
+
+  java.lang.String getThreadNamesOrDefault(
+      int key,
+      java.lang.String defaultValue);
+  /**
+   * <pre>
+   * Its key is thread id.
+   * </pre>
+   *
+   * <code>map&lt;uint32, string&gt; thread_names = 3;</code>
+   */
+
+  java.lang.String getThreadNamesOrThrow(
+      int key);
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DictValue.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DictValue.java
new file mode 100644
index 00000000000..e8e6321372f
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DictValue.java
@@ -0,0 +1,705 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/struct.proto
+
+package org.tensorflow.proto.framework;
+
+/**
+ * <pre>
+ * Represents a Python dict keyed by `str`.
+ * The comment on Unicode from Value.string_value applies analogously.
+ * </pre>
+ *
+ * Protobuf type {@code tensorflow.DictValue}
+ */
+public  final class DictValue extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.DictValue)
+    DictValueOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use DictValue.newBuilder() to construct.
+  private DictValue(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private DictValue() {
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new DictValue();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private DictValue(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    int mutable_bitField0_ = 0;
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 10: {
+            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
+              fields_ = com.google.protobuf.MapField.newMapField(
+                  FieldsDefaultEntryHolder.defaultEntry);
+              mutable_bitField0_ |= 0x00000001;
+            }
+            com.google.protobuf.MapEntry<java.lang.String, org.tensorflow.proto.framework.StructuredValue>
+            fields__ = input.readMessage(
+                FieldsDefaultEntryHolder.defaultEntry.getParserForType(), extensionRegistry);
+            fields_.getMutableMap().put(
+                fields__.getKey(), fields__.getValue());
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.framework.StructProtos.internal_static_tensorflow_DictValue_descriptor;
+  }
+
+  @SuppressWarnings({"rawtypes"})
+  @java.lang.Override
+  protected com.google.protobuf.MapField internalGetMapField(
+      int number) {
+    switch (number) {
+      case 1:
+        return internalGetFields();
+      default:
+        throw new RuntimeException(
+            "Invalid map field number: " + number);
+    }
+  }
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.framework.StructProtos.internal_static_tensorflow_DictValue_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.framework.DictValue.class, org.tensorflow.proto.framework.DictValue.Builder.class);
+  }
+
+  public static final int FIELDS_FIELD_NUMBER = 1;
+  private static final class FieldsDefaultEntryHolder {
+    static final com.google.protobuf.MapEntry<
+        java.lang.String, org.tensorflow.proto.framework.StructuredValue> defaultEntry =
+            com.google.protobuf.MapEntry
+            .<java.lang.String, org.tensorflow.proto.framework.StructuredValue>newDefaultInstance(
+                org.tensorflow.proto.framework.StructProtos.internal_static_tensorflow_DictValue_FieldsEntry_descriptor, 
+                com.google.protobuf.WireFormat.FieldType.STRING,
+                "",
+                com.google.protobuf.WireFormat.FieldType.MESSAGE,
+                org.tensorflow.proto.framework.StructuredValue.getDefaultInstance());
+  }
+  private com.google.protobuf.MapField<
+      java.lang.String, org.tensorflow.proto.framework.StructuredValue> fields_;
+  private com.google.protobuf.MapField<java.lang.String, org.tensorflow.proto.framework.StructuredValue>
+  internalGetFields() {
+    if (fields_ == null) {
+      return com.google.protobuf.MapField.emptyMapField(
+          FieldsDefaultEntryHolder.defaultEntry);
+    }
+    return fields_;
+  }
+
+  public int getFieldsCount() {
+    return internalGetFields().getMap().size();
+  }
+  /**
+   * <code>map&lt;string, .tensorflow.StructuredValue&gt; fields = 1;</code>
+   */
+
+  public boolean containsFields(
+      java.lang.String key) {
+    if (key == null) { throw new java.lang.NullPointerException(); }
+    return internalGetFields().getMap().containsKey(key);
+  }
+  /**
+   * Use {@link #getFieldsMap()} instead.
+   */
+  @java.lang.Deprecated
+  public java.util.Map<java.lang.String, org.tensorflow.proto.framework.StructuredValue> getFields() {
+    return getFieldsMap();
+  }
+  /**
+   * <code>map&lt;string, .tensorflow.StructuredValue&gt; fields = 1;</code>
+   */
+
+  public java.util.Map<java.lang.String, org.tensorflow.proto.framework.StructuredValue> getFieldsMap() {
+    return internalGetFields().getMap();
+  }
+  /**
+   * <code>map&lt;string, .tensorflow.StructuredValue&gt; fields = 1;</code>
+   */
+
+  public org.tensorflow.proto.framework.StructuredValue getFieldsOrDefault(
+      java.lang.String key,
+      org.tensorflow.proto.framework.StructuredValue defaultValue) {
+    if (key == null) { throw new java.lang.NullPointerException(); }
+    java.util.Map<java.lang.String, org.tensorflow.proto.framework.StructuredValue> map =
+        internalGetFields().getMap();
+    return map.containsKey(key) ? map.get(key) : defaultValue;
+  }
+  /**
+   * <code>map&lt;string, .tensorflow.StructuredValue&gt; fields = 1;</code>
+   */
+
+  public org.tensorflow.proto.framework.StructuredValue getFieldsOrThrow(
+      java.lang.String key) {
+    if (key == null) { throw new java.lang.NullPointerException(); }
+    java.util.Map<java.lang.String, org.tensorflow.proto.framework.StructuredValue> map =
+        internalGetFields().getMap();
+    if (!map.containsKey(key)) {
+      throw new java.lang.IllegalArgumentException();
+    }
+    return map.get(key);
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    com.google.protobuf.GeneratedMessageV3
+      .serializeStringMapTo(
+        output,
+        internalGetFields(),
+        FieldsDefaultEntryHolder.defaultEntry,
+        1);
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    for (java.util.Map.Entry<java.lang.String, org.tensorflow.proto.framework.StructuredValue> entry
+         : internalGetFields().getMap().entrySet()) {
+      com.google.protobuf.MapEntry<java.lang.String, org.tensorflow.proto.framework.StructuredValue>
+      fields__ = FieldsDefaultEntryHolder.defaultEntry.newBuilderForType()
+          .setKey(entry.getKey())
+          .setValue(entry.getValue())
+          .build();
+      size += com.google.protobuf.CodedOutputStream
+          .computeMessageSize(1, fields__);
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.framework.DictValue)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.framework.DictValue other = (org.tensorflow.proto.framework.DictValue) obj;
+
+    if (!internalGetFields().equals(
+        other.internalGetFields())) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    if (!internalGetFields().getMap().isEmpty()) {
+      hash = (37 * hash) + FIELDS_FIELD_NUMBER;
+      hash = (53 * hash) + internalGetFields().hashCode();
+    }
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.framework.DictValue parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.DictValue parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.DictValue parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.DictValue parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.DictValue parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.DictValue parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.DictValue parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.DictValue parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.DictValue parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.DictValue parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.DictValue parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.DictValue parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.framework.DictValue prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * <pre>
+   * Represents a Python dict keyed by `str`.
+   * The comment on Unicode from Value.string_value applies analogously.
+   * </pre>
+   *
+   * Protobuf type {@code tensorflow.DictValue}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.DictValue)
+      org.tensorflow.proto.framework.DictValueOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.StructProtos.internal_static_tensorflow_DictValue_descriptor;
+    }
+
+    @SuppressWarnings({"rawtypes"})
+    protected com.google.protobuf.MapField internalGetMapField(
+        int number) {
+      switch (number) {
+        case 1:
+          return internalGetFields();
+        default:
+          throw new RuntimeException(
+              "Invalid map field number: " + number);
+      }
+    }
+    @SuppressWarnings({"rawtypes"})
+    protected com.google.protobuf.MapField internalGetMutableMapField(
+        int number) {
+      switch (number) {
+        case 1:
+          return internalGetMutableFields();
+        default:
+          throw new RuntimeException(
+              "Invalid map field number: " + number);
+      }
+    }
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.StructProtos.internal_static_tensorflow_DictValue_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.DictValue.class, org.tensorflow.proto.framework.DictValue.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.framework.DictValue.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      internalGetMutableFields().clear();
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.framework.StructProtos.internal_static_tensorflow_DictValue_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.DictValue getDefaultInstanceForType() {
+      return org.tensorflow.proto.framework.DictValue.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.DictValue build() {
+      org.tensorflow.proto.framework.DictValue result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.DictValue buildPartial() {
+      org.tensorflow.proto.framework.DictValue result = new org.tensorflow.proto.framework.DictValue(this);
+      int from_bitField0_ = bitField0_;
+      result.fields_ = internalGetFields();
+      result.fields_.makeImmutable();
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.framework.DictValue) {
+        return mergeFrom((org.tensorflow.proto.framework.DictValue)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.framework.DictValue other) {
+      if (other == org.tensorflow.proto.framework.DictValue.getDefaultInstance()) return this;
+      internalGetMutableFields().mergeFrom(
+          other.internalGetFields());
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.framework.DictValue parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.framework.DictValue) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+    private int bitField0_;
+
+    private com.google.protobuf.MapField<
+        java.lang.String, org.tensorflow.proto.framework.StructuredValue> fields_;
+    private com.google.protobuf.MapField<java.lang.String, org.tensorflow.proto.framework.StructuredValue>
+    internalGetFields() {
+      if (fields_ == null) {
+        return com.google.protobuf.MapField.emptyMapField(
+            FieldsDefaultEntryHolder.defaultEntry);
+      }
+      return fields_;
+    }
+    private com.google.protobuf.MapField<java.lang.String, org.tensorflow.proto.framework.StructuredValue>
+    internalGetMutableFields() {
+      onChanged();;
+      if (fields_ == null) {
+        fields_ = com.google.protobuf.MapField.newMapField(
+            FieldsDefaultEntryHolder.defaultEntry);
+      }
+      if (!fields_.isMutable()) {
+        fields_ = fields_.copy();
+      }
+      return fields_;
+    }
+
+    public int getFieldsCount() {
+      return internalGetFields().getMap().size();
+    }
+    /**
+     * <code>map&lt;string, .tensorflow.StructuredValue&gt; fields = 1;</code>
+     */
+
+    public boolean containsFields(
+        java.lang.String key) {
+      if (key == null) { throw new java.lang.NullPointerException(); }
+      return internalGetFields().getMap().containsKey(key);
+    }
+    /**
+     * Use {@link #getFieldsMap()} instead.
+     */
+    @java.lang.Deprecated
+    public java.util.Map<java.lang.String, org.tensorflow.proto.framework.StructuredValue> getFields() {
+      return getFieldsMap();
+    }
+    /**
+     * <code>map&lt;string, .tensorflow.StructuredValue&gt; fields = 1;</code>
+     */
+
+    public java.util.Map<java.lang.String, org.tensorflow.proto.framework.StructuredValue> getFieldsMap() {
+      return internalGetFields().getMap();
+    }
+    /**
+     * <code>map&lt;string, .tensorflow.StructuredValue&gt; fields = 1;</code>
+     */
+
+    public org.tensorflow.proto.framework.StructuredValue getFieldsOrDefault(
+        java.lang.String key,
+        org.tensorflow.proto.framework.StructuredValue defaultValue) {
+      if (key == null) { throw new java.lang.NullPointerException(); }
+      java.util.Map<java.lang.String, org.tensorflow.proto.framework.StructuredValue> map =
+          internalGetFields().getMap();
+      return map.containsKey(key) ? map.get(key) : defaultValue;
+    }
+    /**
+     * <code>map&lt;string, .tensorflow.StructuredValue&gt; fields = 1;</code>
+     */
+
+    public org.tensorflow.proto.framework.StructuredValue getFieldsOrThrow(
+        java.lang.String key) {
+      if (key == null) { throw new java.lang.NullPointerException(); }
+      java.util.Map<java.lang.String, org.tensorflow.proto.framework.StructuredValue> map =
+          internalGetFields().getMap();
+      if (!map.containsKey(key)) {
+        throw new java.lang.IllegalArgumentException();
+      }
+      return map.get(key);
+    }
+
+    public Builder clearFields() {
+      internalGetMutableFields().getMutableMap()
+          .clear();
+      return this;
+    }
+    /**
+     * <code>map&lt;string, .tensorflow.StructuredValue&gt; fields = 1;</code>
+     */
+
+    public Builder removeFields(
+        java.lang.String key) {
+      if (key == null) { throw new java.lang.NullPointerException(); }
+      internalGetMutableFields().getMutableMap()
+          .remove(key);
+      return this;
+    }
+    /**
+     * Use alternate mutation accessors instead.
+     */
+    @java.lang.Deprecated
+    public java.util.Map<java.lang.String, org.tensorflow.proto.framework.StructuredValue>
+    getMutableFields() {
+      return internalGetMutableFields().getMutableMap();
+    }
+    /**
+     * <code>map&lt;string, .tensorflow.StructuredValue&gt; fields = 1;</code>
+     */
+    public Builder putFields(
+        java.lang.String key,
+        org.tensorflow.proto.framework.StructuredValue value) {
+      if (key == null) { throw new java.lang.NullPointerException(); }
+      if (value == null) { throw new java.lang.NullPointerException(); }
+      internalGetMutableFields().getMutableMap()
+          .put(key, value);
+      return this;
+    }
+    /**
+     * <code>map&lt;string, .tensorflow.StructuredValue&gt; fields = 1;</code>
+     */
+
+    public Builder putAllFields(
+        java.util.Map<java.lang.String, org.tensorflow.proto.framework.StructuredValue> values) {
+      internalGetMutableFields().getMutableMap()
+          .putAll(values);
+      return this;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.DictValue)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.DictValue)
+  private static final org.tensorflow.proto.framework.DictValue DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.DictValue();
+  }
+
+  public static org.tensorflow.proto.framework.DictValue getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<DictValue>
+      PARSER = new com.google.protobuf.AbstractParser<DictValue>() {
+    @java.lang.Override
+    public DictValue parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new DictValue(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<DictValue> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<DictValue> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.framework.DictValue getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DictValueOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DictValueOrBuilder.java
new file mode 100644
index 00000000000..527fdffc935
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/DictValueOrBuilder.java
@@ -0,0 +1,43 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/struct.proto
+
+package org.tensorflow.proto.framework;
+
+public interface DictValueOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.DictValue)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <code>map&lt;string, .tensorflow.StructuredValue&gt; fields = 1;</code>
+   */
+  int getFieldsCount();
+  /**
+   * <code>map&lt;string, .tensorflow.StructuredValue&gt; fields = 1;</code>
+   */
+  boolean containsFields(
+      java.lang.String key);
+  /**
+   * Use {@link #getFieldsMap()} instead.
+   */
+  @java.lang.Deprecated
+  java.util.Map<java.lang.String, org.tensorflow.proto.framework.StructuredValue>
+  getFields();
+  /**
+   * <code>map&lt;string, .tensorflow.StructuredValue&gt; fields = 1;</code>
+   */
+  java.util.Map<java.lang.String, org.tensorflow.proto.framework.StructuredValue>
+  getFieldsMap();
+  /**
+   * <code>map&lt;string, .tensorflow.StructuredValue&gt; fields = 1;</code>
+   */
+
+  org.tensorflow.proto.framework.StructuredValue getFieldsOrDefault(
+      java.lang.String key,
+      org.tensorflow.proto.framework.StructuredValue defaultValue);
+  /**
+   * <code>map&lt;string, .tensorflow.StructuredValue&gt; fields = 1;</code>
+   */
+
+  org.tensorflow.proto.framework.StructuredValue getFieldsOrThrow(
+      java.lang.String key);
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ErrorCodes.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ErrorCodes.java
new file mode 100644
index 00000000000..113b8465373
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ErrorCodes.java
@@ -0,0 +1,40 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/lib/core/error_codes.proto
+
+package org.tensorflow.proto.framework;
+
+public final class ErrorCodes {
+  private ErrorCodes() {}
+  public static void registerAllExtensions(
+      com.google.protobuf.ExtensionRegistryLite registry) {
+  }
+
+  public static void registerAllExtensions(
+      com.google.protobuf.ExtensionRegistry registry) {
+    registerAllExtensions(
+        (com.google.protobuf.ExtensionRegistryLite) registry);
+  }
+
+  public static com.google.protobuf.Descriptors.FileDescriptor
+      getDescriptor() {
+    return descriptor;
+  }
+  private static  com.google.protobuf.Descriptors.FileDescriptor
+      descriptor;
+  static {
+    java.lang.String[] descriptorData = {
+      "\n*tensorflow/core/lib/core/error_codes.p" +
+      "roto\032*tensorflow/core/protobuf/error_cod" +
+      "es.protoB \n\036org.tensorflow.proto.framewo" +
+      "rkP\000b\006proto3"
+    };
+    descriptor = com.google.protobuf.Descriptors.FileDescriptor
+      .internalBuildGeneratedFileFrom(descriptorData,
+        new com.google.protobuf.Descriptors.FileDescriptor[] {
+          org.tensorflow.proto.framework.ErrorCodesProtos.getDescriptor(),
+        });
+    org.tensorflow.proto.framework.ErrorCodesProtos.getDescriptor();
+  }
+
+  // @@protoc_insertion_point(outer_class_scope)
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ErrorCodesProtos.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ErrorCodesProtos.java
new file mode 100644
index 00000000000..9e424017f74
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ErrorCodesProtos.java
@@ -0,0 +1,49 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/error_codes.proto
+
+package org.tensorflow.proto.framework;
+
+public final class ErrorCodesProtos {
+  private ErrorCodesProtos() {}
+  public static void registerAllExtensions(
+      com.google.protobuf.ExtensionRegistryLite registry) {
+  }
+
+  public static void registerAllExtensions(
+      com.google.protobuf.ExtensionRegistry registry) {
+    registerAllExtensions(
+        (com.google.protobuf.ExtensionRegistryLite) registry);
+  }
+
+  public static com.google.protobuf.Descriptors.FileDescriptor
+      getDescriptor() {
+    return descriptor;
+  }
+  private static  com.google.protobuf.Descriptors.FileDescriptor
+      descriptor;
+  static {
+    java.lang.String[] descriptorData = {
+      "\n*tensorflow/core/protobuf/error_codes.p" +
+      "roto\022\020tensorflow.error*\204\003\n\004Code\022\006\n\002OK\020\000\022" +
+      "\r\n\tCANCELLED\020\001\022\013\n\007UNKNOWN\020\002\022\024\n\020INVALID_A" +
+      "RGUMENT\020\003\022\025\n\021DEADLINE_EXCEEDED\020\004\022\r\n\tNOT_" +
+      "FOUND\020\005\022\022\n\016ALREADY_EXISTS\020\006\022\025\n\021PERMISSIO" +
+      "N_DENIED\020\007\022\023\n\017UNAUTHENTICATED\020\020\022\026\n\022RESOU" +
+      "RCE_EXHAUSTED\020\010\022\027\n\023FAILED_PRECONDITION\020\t" +
+      "\022\013\n\007ABORTED\020\n\022\020\n\014OUT_OF_RANGE\020\013\022\021\n\rUNIMP" +
+      "LEMENTED\020\014\022\014\n\010INTERNAL\020\r\022\017\n\013UNAVAILABLE\020" +
+      "\016\022\r\n\tDATA_LOSS\020\017\022K\nGDO_NOT_USE_RESERVED_" +
+      "FOR_FUTURE_EXPANSION_USE_DEFAULT_IN_SWIT" +
+      "CH_INSTEAD_\020\024B\201\001\n\036org.tensorflow.proto.f" +
+      "rameworkB\020ErrorCodesProtosP\001ZHgithub.com" +
+      "/tensorflow/tensorflow/tensorflow/go/cor" +
+      "e/core_protos_go_proto\370\001\001b\006proto3"
+    };
+    descriptor = com.google.protobuf.Descriptors.FileDescriptor
+      .internalBuildGeneratedFileFrom(descriptorData,
+        new com.google.protobuf.Descriptors.FileDescriptor[] {
+        });
+  }
+
+  // @@protoc_insertion_point(outer_class_scope)
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/FunctionDef.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/FunctionDef.java
new file mode 100644
index 00000000000..3437857d610
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/FunctionDef.java
@@ -0,0 +1,3415 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/framework/function.proto
+
+package org.tensorflow.proto.framework;
+
+/**
+ * <pre>
+ * A function can be instantiated when the runtime can bind every attr
+ * with a value. When a GraphDef has a call to a function, it must
+ * have binding for every attr defined in the signature.
+ * TODO(zhifengc):
+ *   * device spec, etc.
+ * </pre>
+ *
+ * Protobuf type {@code tensorflow.FunctionDef}
+ */
+public  final class FunctionDef extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.FunctionDef)
+    FunctionDefOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use FunctionDef.newBuilder() to construct.
+  private FunctionDef(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private FunctionDef() {
+    nodeDef_ = java.util.Collections.emptyList();
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new FunctionDef();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private FunctionDef(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    int mutable_bitField0_ = 0;
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 10: {
+            org.tensorflow.proto.framework.OpDef.Builder subBuilder = null;
+            if (signature_ != null) {
+              subBuilder = signature_.toBuilder();
+            }
+            signature_ = input.readMessage(org.tensorflow.proto.framework.OpDef.parser(), extensionRegistry);
+            if (subBuilder != null) {
+              subBuilder.mergeFrom(signature_);
+              signature_ = subBuilder.buildPartial();
+            }
+
+            break;
+          }
+          case 26: {
+            if (!((mutable_bitField0_ & 0x00000008) != 0)) {
+              nodeDef_ = new java.util.ArrayList<org.tensorflow.proto.framework.NodeDef>();
+              mutable_bitField0_ |= 0x00000008;
+            }
+            nodeDef_.add(
+                input.readMessage(org.tensorflow.proto.framework.NodeDef.parser(), extensionRegistry));
+            break;
+          }
+          case 34: {
+            if (!((mutable_bitField0_ & 0x00000010) != 0)) {
+              ret_ = com.google.protobuf.MapField.newMapField(
+                  RetDefaultEntryHolder.defaultEntry);
+              mutable_bitField0_ |= 0x00000010;
+            }
+            com.google.protobuf.MapEntry<java.lang.String, java.lang.String>
+            ret__ = input.readMessage(
+                RetDefaultEntryHolder.defaultEntry.getParserForType(), extensionRegistry);
+            ret_.getMutableMap().put(
+                ret__.getKey(), ret__.getValue());
+            break;
+          }
+          case 42: {
+            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
+              attr_ = com.google.protobuf.MapField.newMapField(
+                  AttrDefaultEntryHolder.defaultEntry);
+              mutable_bitField0_ |= 0x00000001;
+            }
+            com.google.protobuf.MapEntry<java.lang.String, org.tensorflow.proto.framework.AttrValue>
+            attr__ = input.readMessage(
+                AttrDefaultEntryHolder.defaultEntry.getParserForType(), extensionRegistry);
+            attr_.getMutableMap().put(
+                attr__.getKey(), attr__.getValue());
+            break;
+          }
+          case 50: {
+            if (!((mutable_bitField0_ & 0x00000020) != 0)) {
+              controlRet_ = com.google.protobuf.MapField.newMapField(
+                  ControlRetDefaultEntryHolder.defaultEntry);
+              mutable_bitField0_ |= 0x00000020;
+            }
+            com.google.protobuf.MapEntry<java.lang.String, java.lang.String>
+            controlRet__ = input.readMessage(
+                ControlRetDefaultEntryHolder.defaultEntry.getParserForType(), extensionRegistry);
+            controlRet_.getMutableMap().put(
+                controlRet__.getKey(), controlRet__.getValue());
+            break;
+          }
+          case 58: {
+            if (!((mutable_bitField0_ & 0x00000002) != 0)) {
+              argAttr_ = com.google.protobuf.MapField.newMapField(
+                  ArgAttrDefaultEntryHolder.defaultEntry);
+              mutable_bitField0_ |= 0x00000002;
+            }
+            com.google.protobuf.MapEntry<java.lang.Integer, org.tensorflow.proto.framework.FunctionDef.ArgAttrs>
+            argAttr__ = input.readMessage(
+                ArgAttrDefaultEntryHolder.defaultEntry.getParserForType(), extensionRegistry);
+            argAttr_.getMutableMap().put(
+                argAttr__.getKey(), argAttr__.getValue());
+            break;
+          }
+          case 66: {
+            if (!((mutable_bitField0_ & 0x00000004) != 0)) {
+              resourceArgUniqueId_ = com.google.protobuf.MapField.newMapField(
+                  ResourceArgUniqueIdDefaultEntryHolder.defaultEntry);
+              mutable_bitField0_ |= 0x00000004;
+            }
+            com.google.protobuf.MapEntry<java.lang.Integer, java.lang.Integer>
+            resourceArgUniqueId__ = input.readMessage(
+                ResourceArgUniqueIdDefaultEntryHolder.defaultEntry.getParserForType(), extensionRegistry);
+            resourceArgUniqueId_.getMutableMap().put(
+                resourceArgUniqueId__.getKey(), resourceArgUniqueId__.getValue());
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      if (((mutable_bitField0_ & 0x00000008) != 0)) {
+        nodeDef_ = java.util.Collections.unmodifiableList(nodeDef_);
+      }
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.framework.FunctionProtos.internal_static_tensorflow_FunctionDef_descriptor;
+  }
+
+  @SuppressWarnings({"rawtypes"})
+  @java.lang.Override
+  protected com.google.protobuf.MapField internalGetMapField(
+      int number) {
+    switch (number) {
+      case 5:
+        return internalGetAttr();
+      case 7:
+        return internalGetArgAttr();
+      case 8:
+        return internalGetResourceArgUniqueId();
+      case 4:
+        return internalGetRet();
+      case 6:
+        return internalGetControlRet();
+      default:
+        throw new RuntimeException(
+            "Invalid map field number: " + number);
+    }
+  }
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.framework.FunctionProtos.internal_static_tensorflow_FunctionDef_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.framework.FunctionDef.class, org.tensorflow.proto.framework.FunctionDef.Builder.class);
+  }
+
+  public interface ArgAttrsOrBuilder extends
+      // @@protoc_insertion_point(interface_extends:tensorflow.FunctionDef.ArgAttrs)
+      com.google.protobuf.MessageOrBuilder {
+
+    /**
+     * <code>map&lt;string, .tensorflow.AttrValue&gt; attr = 1;</code>
+     */
+    int getAttrCount();
+    /**
+     * <code>map&lt;string, .tensorflow.AttrValue&gt; attr = 1;</code>
+     */
+    boolean containsAttr(
+        java.lang.String key);
+    /**
+     * Use {@link #getAttrMap()} instead.
+     */
+    @java.lang.Deprecated
+    java.util.Map<java.lang.String, org.tensorflow.proto.framework.AttrValue>
+    getAttr();
+    /**
+     * <code>map&lt;string, .tensorflow.AttrValue&gt; attr = 1;</code>
+     */
+    java.util.Map<java.lang.String, org.tensorflow.proto.framework.AttrValue>
+    getAttrMap();
+    /**
+     * <code>map&lt;string, .tensorflow.AttrValue&gt; attr = 1;</code>
+     */
+
+    org.tensorflow.proto.framework.AttrValue getAttrOrDefault(
+        java.lang.String key,
+        org.tensorflow.proto.framework.AttrValue defaultValue);
+    /**
+     * <code>map&lt;string, .tensorflow.AttrValue&gt; attr = 1;</code>
+     */
+
+    org.tensorflow.proto.framework.AttrValue getAttrOrThrow(
+        java.lang.String key);
+  }
+  /**
+   * <pre>
+   * Attributes for function arguments. These attributes are the same set of
+   * valid attributes as to _Arg nodes.
+   * </pre>
+   *
+   * Protobuf type {@code tensorflow.FunctionDef.ArgAttrs}
+   */
+  public  static final class ArgAttrs extends
+      com.google.protobuf.GeneratedMessageV3 implements
+      // @@protoc_insertion_point(message_implements:tensorflow.FunctionDef.ArgAttrs)
+      ArgAttrsOrBuilder {
+  private static final long serialVersionUID = 0L;
+    // Use ArgAttrs.newBuilder() to construct.
+    private ArgAttrs(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+      super(builder);
+    }
+    private ArgAttrs() {
+    }
+
+    @java.lang.Override
+    @SuppressWarnings({"unused"})
+    protected java.lang.Object newInstance(
+        UnusedPrivateParameter unused) {
+      return new ArgAttrs();
+    }
+
+    @java.lang.Override
+    public final com.google.protobuf.UnknownFieldSet
+    getUnknownFields() {
+      return this.unknownFields;
+    }
+    private ArgAttrs(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      this();
+      if (extensionRegistry == null) {
+        throw new java.lang.NullPointerException();
+      }
+      int mutable_bitField0_ = 0;
+      com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+          com.google.protobuf.UnknownFieldSet.newBuilder();
+      try {
+        boolean done = false;
+        while (!done) {
+          int tag = input.readTag();
+          switch (tag) {
+            case 0:
+              done = true;
+              break;
+            case 10: {
+              if (!((mutable_bitField0_ & 0x00000001) != 0)) {
+                attr_ = com.google.protobuf.MapField.newMapField(
+                    AttrDefaultEntryHolder.defaultEntry);
+                mutable_bitField0_ |= 0x00000001;
+              }
+              com.google.protobuf.MapEntry<java.lang.String, org.tensorflow.proto.framework.AttrValue>
+              attr__ = input.readMessage(
+                  AttrDefaultEntryHolder.defaultEntry.getParserForType(), extensionRegistry);
+              attr_.getMutableMap().put(
+                  attr__.getKey(), attr__.getValue());
+              break;
+            }
+            default: {
+              if (!parseUnknownField(
+                  input, unknownFields, extensionRegistry, tag)) {
+                done = true;
+              }
+              break;
+            }
+          }
+        }
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        throw e.setUnfinishedMessage(this);
+      } catch (java.io.IOException e) {
+        throw new com.google.protobuf.InvalidProtocolBufferException(
+            e).setUnfinishedMessage(this);
+      } finally {
+        this.unknownFields = unknownFields.build();
+        makeExtensionsImmutable();
+      }
+    }
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.FunctionProtos.internal_static_tensorflow_FunctionDef_ArgAttrs_descriptor;
+    }
+
+    @SuppressWarnings({"rawtypes"})
+    @java.lang.Override
+    protected com.google.protobuf.MapField internalGetMapField(
+        int number) {
+      switch (number) {
+        case 1:
+          return internalGetAttr();
+        default:
+          throw new RuntimeException(
+              "Invalid map field number: " + number);
+      }
+    }
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.FunctionProtos.internal_static_tensorflow_FunctionDef_ArgAttrs_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.FunctionDef.ArgAttrs.class, org.tensorflow.proto.framework.FunctionDef.ArgAttrs.Builder.class);
+    }
+
+    public static final int ATTR_FIELD_NUMBER = 1;
+    private static final class AttrDefaultEntryHolder {
+      static final com.google.protobuf.MapEntry<
+          java.lang.String, org.tensorflow.proto.framework.AttrValue> defaultEntry =
+              com.google.protobuf.MapEntry
+              .<java.lang.String, org.tensorflow.proto.framework.AttrValue>newDefaultInstance(
+                  org.tensorflow.proto.framework.FunctionProtos.internal_static_tensorflow_FunctionDef_ArgAttrs_AttrEntry_descriptor, 
+                  com.google.protobuf.WireFormat.FieldType.STRING,
+                  "",
+                  com.google.protobuf.WireFormat.FieldType.MESSAGE,
+                  org.tensorflow.proto.framework.AttrValue.getDefaultInstance());
+    }
+    private com.google.protobuf.MapField<
+        java.lang.String, org.tensorflow.proto.framework.AttrValue> attr_;
+    private com.google.protobuf.MapField<java.lang.String, org.tensorflow.proto.framework.AttrValue>
+    internalGetAttr() {
+      if (attr_ == null) {
+        return com.google.protobuf.MapField.emptyMapField(
+            AttrDefaultEntryHolder.defaultEntry);
+      }
+      return attr_;
+    }
+
+    public int getAttrCount() {
+      return internalGetAttr().getMap().size();
+    }
+    /**
+     * <code>map&lt;string, .tensorflow.AttrValue&gt; attr = 1;</code>
+     */
+
+    public boolean containsAttr(
+        java.lang.String key) {
+      if (key == null) { throw new java.lang.NullPointerException(); }
+      return internalGetAttr().getMap().containsKey(key);
+    }
+    /**
+     * Use {@link #getAttrMap()} instead.
+     */
+    @java.lang.Deprecated
+    public java.util.Map<java.lang.String, org.tensorflow.proto.framework.AttrValue> getAttr() {
+      return getAttrMap();
+    }
+    /**
+     * <code>map&lt;string, .tensorflow.AttrValue&gt; attr = 1;</code>
+     */
+
+    public java.util.Map<java.lang.String, org.tensorflow.proto.framework.AttrValue> getAttrMap() {
+      return internalGetAttr().getMap();
+    }
+    /**
+     * <code>map&lt;string, .tensorflow.AttrValue&gt; attr = 1;</code>
+     */
+
+    public org.tensorflow.proto.framework.AttrValue getAttrOrDefault(
+        java.lang.String key,
+        org.tensorflow.proto.framework.AttrValue defaultValue) {
+      if (key == null) { throw new java.lang.NullPointerException(); }
+      java.util.Map<java.lang.String, org.tensorflow.proto.framework.AttrValue> map =
+          internalGetAttr().getMap();
+      return map.containsKey(key) ? map.get(key) : defaultValue;
+    }
+    /**
+     * <code>map&lt;string, .tensorflow.AttrValue&gt; attr = 1;</code>
+     */
+
+    public org.tensorflow.proto.framework.AttrValue getAttrOrThrow(
+        java.lang.String key) {
+      if (key == null) { throw new java.lang.NullPointerException(); }
+      java.util.Map<java.lang.String, org.tensorflow.proto.framework.AttrValue> map =
+          internalGetAttr().getMap();
+      if (!map.containsKey(key)) {
+        throw new java.lang.IllegalArgumentException();
+      }
+      return map.get(key);
+    }
+
+    private byte memoizedIsInitialized = -1;
+    @java.lang.Override
+    public final boolean isInitialized() {
+      byte isInitialized = memoizedIsInitialized;
+      if (isInitialized == 1) return true;
+      if (isInitialized == 0) return false;
+
+      memoizedIsInitialized = 1;
+      return true;
+    }
+
+    @java.lang.Override
+    public void writeTo(com.google.protobuf.CodedOutputStream output)
+                        throws java.io.IOException {
+      com.google.protobuf.GeneratedMessageV3
+        .serializeStringMapTo(
+          output,
+          internalGetAttr(),
+          AttrDefaultEntryHolder.defaultEntry,
+          1);
+      unknownFields.writeTo(output);
+    }
+
+    @java.lang.Override
+    public int getSerializedSize() {
+      int size = memoizedSize;
+      if (size != -1) return size;
+
+      size = 0;
+      for (java.util.Map.Entry<java.lang.String, org.tensorflow.proto.framework.AttrValue> entry
+           : internalGetAttr().getMap().entrySet()) {
+        com.google.protobuf.MapEntry<java.lang.String, org.tensorflow.proto.framework.AttrValue>
+        attr__ = AttrDefaultEntryHolder.defaultEntry.newBuilderForType()
+            .setKey(entry.getKey())
+            .setValue(entry.getValue())
+            .build();
+        size += com.google.protobuf.CodedOutputStream
+            .computeMessageSize(1, attr__);
+      }
+      size += unknownFields.getSerializedSize();
+      memoizedSize = size;
+      return size;
+    }
+
+    @java.lang.Override
+    public boolean equals(final java.lang.Object obj) {
+      if (obj == this) {
+       return true;
+      }
+      if (!(obj instanceof org.tensorflow.proto.framework.FunctionDef.ArgAttrs)) {
+        return super.equals(obj);
+      }
+      org.tensorflow.proto.framework.FunctionDef.ArgAttrs other = (org.tensorflow.proto.framework.FunctionDef.ArgAttrs) obj;
+
+      if (!internalGetAttr().equals(
+          other.internalGetAttr())) return false;
+      if (!unknownFields.equals(other.unknownFields)) return false;
+      return true;
+    }
+
+    @java.lang.Override
+    public int hashCode() {
+      if (memoizedHashCode != 0) {
+        return memoizedHashCode;
+      }
+      int hash = 41;
+      hash = (19 * hash) + getDescriptor().hashCode();
+      if (!internalGetAttr().getMap().isEmpty()) {
+        hash = (37 * hash) + ATTR_FIELD_NUMBER;
+        hash = (53 * hash) + internalGetAttr().hashCode();
+      }
+      hash = (29 * hash) + unknownFields.hashCode();
+      memoizedHashCode = hash;
+      return hash;
+    }
+
+    public static org.tensorflow.proto.framework.FunctionDef.ArgAttrs parseFrom(
+        java.nio.ByteBuffer data)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data);
+    }
+    public static org.tensorflow.proto.framework.FunctionDef.ArgAttrs parseFrom(
+        java.nio.ByteBuffer data,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.FunctionDef.ArgAttrs parseFrom(
+        com.google.protobuf.ByteString data)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data);
+    }
+    public static org.tensorflow.proto.framework.FunctionDef.ArgAttrs parseFrom(
+        com.google.protobuf.ByteString data,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.FunctionDef.ArgAttrs parseFrom(byte[] data)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data);
+    }
+    public static org.tensorflow.proto.framework.FunctionDef.ArgAttrs parseFrom(
+        byte[] data,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.FunctionDef.ArgAttrs parseFrom(java.io.InputStream input)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input);
+    }
+    public static org.tensorflow.proto.framework.FunctionDef.ArgAttrs parseFrom(
+        java.io.InputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.FunctionDef.ArgAttrs parseDelimitedFrom(java.io.InputStream input)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseDelimitedWithIOException(PARSER, input);
+    }
+    public static org.tensorflow.proto.framework.FunctionDef.ArgAttrs parseDelimitedFrom(
+        java.io.InputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.FunctionDef.ArgAttrs parseFrom(
+        com.google.protobuf.CodedInputStream input)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input);
+    }
+    public static org.tensorflow.proto.framework.FunctionDef.ArgAttrs parseFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input, extensionRegistry);
+    }
+
+    @java.lang.Override
+    public Builder newBuilderForType() { return newBuilder(); }
+    public static Builder newBuilder() {
+      return DEFAULT_INSTANCE.toBuilder();
+    }
+    public static Builder newBuilder(org.tensorflow.proto.framework.FunctionDef.ArgAttrs prototype) {
+      return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+    }
+    @java.lang.Override
+    public Builder toBuilder() {
+      return this == DEFAULT_INSTANCE
+          ? new Builder() : new Builder().mergeFrom(this);
+    }
+
+    @java.lang.Override
+    protected Builder newBuilderForType(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      Builder builder = new Builder(parent);
+      return builder;
+    }
+    /**
+     * <pre>
+     * Attributes for function arguments. These attributes are the same set of
+     * valid attributes as to _Arg nodes.
+     * </pre>
+     *
+     * Protobuf type {@code tensorflow.FunctionDef.ArgAttrs}
+     */
+    public static final class Builder extends
+        com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+        // @@protoc_insertion_point(builder_implements:tensorflow.FunctionDef.ArgAttrs)
+        org.tensorflow.proto.framework.FunctionDef.ArgAttrsOrBuilder {
+      public static final com.google.protobuf.Descriptors.Descriptor
+          getDescriptor() {
+        return org.tensorflow.proto.framework.FunctionProtos.internal_static_tensorflow_FunctionDef_ArgAttrs_descriptor;
+      }
+
+      @SuppressWarnings({"rawtypes"})
+      protected com.google.protobuf.MapField internalGetMapField(
+          int number) {
+        switch (number) {
+          case 1:
+            return internalGetAttr();
+          default:
+            throw new RuntimeException(
+                "Invalid map field number: " + number);
+        }
+      }
+      @SuppressWarnings({"rawtypes"})
+      protected com.google.protobuf.MapField internalGetMutableMapField(
+          int number) {
+        switch (number) {
+          case 1:
+            return internalGetMutableAttr();
+          default:
+            throw new RuntimeException(
+                "Invalid map field number: " + number);
+        }
+      }
+      @java.lang.Override
+      protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+          internalGetFieldAccessorTable() {
+        return org.tensorflow.proto.framework.FunctionProtos.internal_static_tensorflow_FunctionDef_ArgAttrs_fieldAccessorTable
+            .ensureFieldAccessorsInitialized(
+                org.tensorflow.proto.framework.FunctionDef.ArgAttrs.class, org.tensorflow.proto.framework.FunctionDef.ArgAttrs.Builder.class);
+      }
+
+      // Construct using org.tensorflow.proto.framework.FunctionDef.ArgAttrs.newBuilder()
+      private Builder() {
+        maybeForceBuilderInitialization();
+      }
+
+      private Builder(
+          com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+        super(parent);
+        maybeForceBuilderInitialization();
+      }
+      private void maybeForceBuilderInitialization() {
+        if (com.google.protobuf.GeneratedMessageV3
+                .alwaysUseFieldBuilders) {
+        }
+      }
+      @java.lang.Override
+      public Builder clear() {
+        super.clear();
+        internalGetMutableAttr().clear();
+        return this;
+      }
+
+      @java.lang.Override
+      public com.google.protobuf.Descriptors.Descriptor
+          getDescriptorForType() {
+        return org.tensorflow.proto.framework.FunctionProtos.internal_static_tensorflow_FunctionDef_ArgAttrs_descriptor;
+      }
+
+      @java.lang.Override
+      public org.tensorflow.proto.framework.FunctionDef.ArgAttrs getDefaultInstanceForType() {
+        return org.tensorflow.proto.framework.FunctionDef.ArgAttrs.getDefaultInstance();
+      }
+
+      @java.lang.Override
+      public org.tensorflow.proto.framework.FunctionDef.ArgAttrs build() {
+        org.tensorflow.proto.framework.FunctionDef.ArgAttrs result = buildPartial();
+        if (!result.isInitialized()) {
+          throw newUninitializedMessageException(result);
+        }
+        return result;
+      }
+
+      @java.lang.Override
+      public org.tensorflow.proto.framework.FunctionDef.ArgAttrs buildPartial() {
+        org.tensorflow.proto.framework.FunctionDef.ArgAttrs result = new org.tensorflow.proto.framework.FunctionDef.ArgAttrs(this);
+        int from_bitField0_ = bitField0_;
+        result.attr_ = internalGetAttr();
+        result.attr_.makeImmutable();
+        onBuilt();
+        return result;
+      }
+
+      @java.lang.Override
+      public Builder clone() {
+        return super.clone();
+      }
+      @java.lang.Override
+      public Builder setField(
+          com.google.protobuf.Descriptors.FieldDescriptor field,
+          java.lang.Object value) {
+        return super.setField(field, value);
+      }
+      @java.lang.Override
+      public Builder clearField(
+          com.google.protobuf.Descriptors.FieldDescriptor field) {
+        return super.clearField(field);
+      }
+      @java.lang.Override
+      public Builder clearOneof(
+          com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+        return super.clearOneof(oneof);
+      }
+      @java.lang.Override
+      public Builder setRepeatedField(
+          com.google.protobuf.Descriptors.FieldDescriptor field,
+          int index, java.lang.Object value) {
+        return super.setRepeatedField(field, index, value);
+      }
+      @java.lang.Override
+      public Builder addRepeatedField(
+          com.google.protobuf.Descriptors.FieldDescriptor field,
+          java.lang.Object value) {
+        return super.addRepeatedField(field, value);
+      }
+      @java.lang.Override
+      public Builder mergeFrom(com.google.protobuf.Message other) {
+        if (other instanceof org.tensorflow.proto.framework.FunctionDef.ArgAttrs) {
+          return mergeFrom((org.tensorflow.proto.framework.FunctionDef.ArgAttrs)other);
+        } else {
+          super.mergeFrom(other);
+          return this;
+        }
+      }
+
+      public Builder mergeFrom(org.tensorflow.proto.framework.FunctionDef.ArgAttrs other) {
+        if (other == org.tensorflow.proto.framework.FunctionDef.ArgAttrs.getDefaultInstance()) return this;
+        internalGetMutableAttr().mergeFrom(
+            other.internalGetAttr());
+        this.mergeUnknownFields(other.unknownFields);
+        onChanged();
+        return this;
+      }
+
+      @java.lang.Override
+      public final boolean isInitialized() {
+        return true;
+      }
+
+      @java.lang.Override
+      public Builder mergeFrom(
+          com.google.protobuf.CodedInputStream input,
+          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+          throws java.io.IOException {
+        org.tensorflow.proto.framework.FunctionDef.ArgAttrs parsedMessage = null;
+        try {
+          parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+        } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+          parsedMessage = (org.tensorflow.proto.framework.FunctionDef.ArgAttrs) e.getUnfinishedMessage();
+          throw e.unwrapIOException();
+        } finally {
+          if (parsedMessage != null) {
+            mergeFrom(parsedMessage);
+          }
+        }
+        return this;
+      }
+      private int bitField0_;
+
+      private com.google.protobuf.MapField<
+          java.lang.String, org.tensorflow.proto.framework.AttrValue> attr_;
+      private com.google.protobuf.MapField<java.lang.String, org.tensorflow.proto.framework.AttrValue>
+      internalGetAttr() {
+        if (attr_ == null) {
+          return com.google.protobuf.MapField.emptyMapField(
+              AttrDefaultEntryHolder.defaultEntry);
+        }
+        return attr_;
+      }
+      private com.google.protobuf.MapField<java.lang.String, org.tensorflow.proto.framework.AttrValue>
+      internalGetMutableAttr() {
+        onChanged();;
+        if (attr_ == null) {
+          attr_ = com.google.protobuf.MapField.newMapField(
+              AttrDefaultEntryHolder.defaultEntry);
+        }
+        if (!attr_.isMutable()) {
+          attr_ = attr_.copy();
+        }
+        return attr_;
+      }
+
+      public int getAttrCount() {
+        return internalGetAttr().getMap().size();
+      }
+      /**
+       * <code>map&lt;string, .tensorflow.AttrValue&gt; attr = 1;</code>
+       */
+
+      public boolean containsAttr(
+          java.lang.String key) {
+        if (key == null) { throw new java.lang.NullPointerException(); }
+        return internalGetAttr().getMap().containsKey(key);
+      }
+      /**
+       * Use {@link #getAttrMap()} instead.
+       */
+      @java.lang.Deprecated
+      public java.util.Map<java.lang.String, org.tensorflow.proto.framework.AttrValue> getAttr() {
+        return getAttrMap();
+      }
+      /**
+       * <code>map&lt;string, .tensorflow.AttrValue&gt; attr = 1;</code>
+       */
+
+      public java.util.Map<java.lang.String, org.tensorflow.proto.framework.AttrValue> getAttrMap() {
+        return internalGetAttr().getMap();
+      }
+      /**
+       * <code>map&lt;string, .tensorflow.AttrValue&gt; attr = 1;</code>
+       */
+
+      public org.tensorflow.proto.framework.AttrValue getAttrOrDefault(
+          java.lang.String key,
+          org.tensorflow.proto.framework.AttrValue defaultValue) {
+        if (key == null) { throw new java.lang.NullPointerException(); }
+        java.util.Map<java.lang.String, org.tensorflow.proto.framework.AttrValue> map =
+            internalGetAttr().getMap();
+        return map.containsKey(key) ? map.get(key) : defaultValue;
+      }
+      /**
+       * <code>map&lt;string, .tensorflow.AttrValue&gt; attr = 1;</code>
+       */
+
+      public org.tensorflow.proto.framework.AttrValue getAttrOrThrow(
+          java.lang.String key) {
+        if (key == null) { throw new java.lang.NullPointerException(); }
+        java.util.Map<java.lang.String, org.tensorflow.proto.framework.AttrValue> map =
+            internalGetAttr().getMap();
+        if (!map.containsKey(key)) {
+          throw new java.lang.IllegalArgumentException();
+        }
+        return map.get(key);
+      }
+
+      public Builder clearAttr() {
+        internalGetMutableAttr().getMutableMap()
+            .clear();
+        return this;
+      }
+      /**
+       * <code>map&lt;string, .tensorflow.AttrValue&gt; attr = 1;</code>
+       */
+
+      public Builder removeAttr(
+          java.lang.String key) {
+        if (key == null) { throw new java.lang.NullPointerException(); }
+        internalGetMutableAttr().getMutableMap()
+            .remove(key);
+        return this;
+      }
+      /**
+       * Use alternate mutation accessors instead.
+       */
+      @java.lang.Deprecated
+      public java.util.Map<java.lang.String, org.tensorflow.proto.framework.AttrValue>
+      getMutableAttr() {
+        return internalGetMutableAttr().getMutableMap();
+      }
+      /**
+       * <code>map&lt;string, .tensorflow.AttrValue&gt; attr = 1;</code>
+       */
+      public Builder putAttr(
+          java.lang.String key,
+          org.tensorflow.proto.framework.AttrValue value) {
+        if (key == null) { throw new java.lang.NullPointerException(); }
+        if (value == null) { throw new java.lang.NullPointerException(); }
+        internalGetMutableAttr().getMutableMap()
+            .put(key, value);
+        return this;
+      }
+      /**
+       * <code>map&lt;string, .tensorflow.AttrValue&gt; attr = 1;</code>
+       */
+
+      public Builder putAllAttr(
+          java.util.Map<java.lang.String, org.tensorflow.proto.framework.AttrValue> values) {
+        internalGetMutableAttr().getMutableMap()
+            .putAll(values);
+        return this;
+      }
+      @java.lang.Override
+      public final Builder setUnknownFields(
+          final com.google.protobuf.UnknownFieldSet unknownFields) {
+        return super.setUnknownFields(unknownFields);
+      }
+
+      @java.lang.Override
+      public final Builder mergeUnknownFields(
+          final com.google.protobuf.UnknownFieldSet unknownFields) {
+        return super.mergeUnknownFields(unknownFields);
+      }
+
+
+      // @@protoc_insertion_point(builder_scope:tensorflow.FunctionDef.ArgAttrs)
+    }
+
+    // @@protoc_insertion_point(class_scope:tensorflow.FunctionDef.ArgAttrs)
+    private static final org.tensorflow.proto.framework.FunctionDef.ArgAttrs DEFAULT_INSTANCE;
+    static {
+      DEFAULT_INSTANCE = new org.tensorflow.proto.framework.FunctionDef.ArgAttrs();
+    }
+
+    public static org.tensorflow.proto.framework.FunctionDef.ArgAttrs getDefaultInstance() {
+      return DEFAULT_INSTANCE;
+    }
+
+    private static final com.google.protobuf.Parser<ArgAttrs>
+        PARSER = new com.google.protobuf.AbstractParser<ArgAttrs>() {
+      @java.lang.Override
+      public ArgAttrs parsePartialFrom(
+          com.google.protobuf.CodedInputStream input,
+          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+          throws com.google.protobuf.InvalidProtocolBufferException {
+        return new ArgAttrs(input, extensionRegistry);
+      }
+    };
+
+    public static com.google.protobuf.Parser<ArgAttrs> parser() {
+      return PARSER;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Parser<ArgAttrs> getParserForType() {
+      return PARSER;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.FunctionDef.ArgAttrs getDefaultInstanceForType() {
+      return DEFAULT_INSTANCE;
+    }
+
+  }
+
+  public static final int SIGNATURE_FIELD_NUMBER = 1;
+  private org.tensorflow.proto.framework.OpDef signature_;
+  /**
+   * <pre>
+   * The definition of the function's name, arguments, return values,
+   * attrs etc.
+   * </pre>
+   *
+   * <code>.tensorflow.OpDef signature = 1;</code>
+   */
+  public boolean hasSignature() {
+    return signature_ != null;
+  }
+  /**
+   * <pre>
+   * The definition of the function's name, arguments, return values,
+   * attrs etc.
+   * </pre>
+   *
+   * <code>.tensorflow.OpDef signature = 1;</code>
+   */
+  public org.tensorflow.proto.framework.OpDef getSignature() {
+    return signature_ == null ? org.tensorflow.proto.framework.OpDef.getDefaultInstance() : signature_;
+  }
+  /**
+   * <pre>
+   * The definition of the function's name, arguments, return values,
+   * attrs etc.
+   * </pre>
+   *
+   * <code>.tensorflow.OpDef signature = 1;</code>
+   */
+  public org.tensorflow.proto.framework.OpDefOrBuilder getSignatureOrBuilder() {
+    return getSignature();
+  }
+
+  public static final int ATTR_FIELD_NUMBER = 5;
+  private static final class AttrDefaultEntryHolder {
+    static final com.google.protobuf.MapEntry<
+        java.lang.String, org.tensorflow.proto.framework.AttrValue> defaultEntry =
+            com.google.protobuf.MapEntry
+            .<java.lang.String, org.tensorflow.proto.framework.AttrValue>newDefaultInstance(
+                org.tensorflow.proto.framework.FunctionProtos.internal_static_tensorflow_FunctionDef_AttrEntry_descriptor, 
+                com.google.protobuf.WireFormat.FieldType.STRING,
+                "",
+                com.google.protobuf.WireFormat.FieldType.MESSAGE,
+                org.tensorflow.proto.framework.AttrValue.getDefaultInstance());
+  }
+  private com.google.protobuf.MapField<
+      java.lang.String, org.tensorflow.proto.framework.AttrValue> attr_;
+  private com.google.protobuf.MapField<java.lang.String, org.tensorflow.proto.framework.AttrValue>
+  internalGetAttr() {
+    if (attr_ == null) {
+      return com.google.protobuf.MapField.emptyMapField(
+          AttrDefaultEntryHolder.defaultEntry);
+    }
+    return attr_;
+  }
+
+  public int getAttrCount() {
+    return internalGetAttr().getMap().size();
+  }
+  /**
+   * <pre>
+   * Attributes specific to this function definition.
+   * </pre>
+   *
+   * <code>map&lt;string, .tensorflow.AttrValue&gt; attr = 5;</code>
+   */
+
+  public boolean containsAttr(
+      java.lang.String key) {
+    if (key == null) { throw new java.lang.NullPointerException(); }
+    return internalGetAttr().getMap().containsKey(key);
+  }
+  /**
+   * Use {@link #getAttrMap()} instead.
+   */
+  @java.lang.Deprecated
+  public java.util.Map<java.lang.String, org.tensorflow.proto.framework.AttrValue> getAttr() {
+    return getAttrMap();
+  }
+  /**
+   * <pre>
+   * Attributes specific to this function definition.
+   * </pre>
+   *
+   * <code>map&lt;string, .tensorflow.AttrValue&gt; attr = 5;</code>
+   */
+
+  public java.util.Map<java.lang.String, org.tensorflow.proto.framework.AttrValue> getAttrMap() {
+    return internalGetAttr().getMap();
+  }
+  /**
+   * <pre>
+   * Attributes specific to this function definition.
+   * </pre>
+   *
+   * <code>map&lt;string, .tensorflow.AttrValue&gt; attr = 5;</code>
+   */
+
+  public org.tensorflow.proto.framework.AttrValue getAttrOrDefault(
+      java.lang.String key,
+      org.tensorflow.proto.framework.AttrValue defaultValue) {
+    if (key == null) { throw new java.lang.NullPointerException(); }
+    java.util.Map<java.lang.String, org.tensorflow.proto.framework.AttrValue> map =
+        internalGetAttr().getMap();
+    return map.containsKey(key) ? map.get(key) : defaultValue;
+  }
+  /**
+   * <pre>
+   * Attributes specific to this function definition.
+   * </pre>
+   *
+   * <code>map&lt;string, .tensorflow.AttrValue&gt; attr = 5;</code>
+   */
+
+  public org.tensorflow.proto.framework.AttrValue getAttrOrThrow(
+      java.lang.String key) {
+    if (key == null) { throw new java.lang.NullPointerException(); }
+    java.util.Map<java.lang.String, org.tensorflow.proto.framework.AttrValue> map =
+        internalGetAttr().getMap();
+    if (!map.containsKey(key)) {
+      throw new java.lang.IllegalArgumentException();
+    }
+    return map.get(key);
+  }
+
+  public static final int ARG_ATTR_FIELD_NUMBER = 7;
+  private static final class ArgAttrDefaultEntryHolder {
+    static final com.google.protobuf.MapEntry<
+        java.lang.Integer, org.tensorflow.proto.framework.FunctionDef.ArgAttrs> defaultEntry =
+            com.google.protobuf.MapEntry
+            .<java.lang.Integer, org.tensorflow.proto.framework.FunctionDef.ArgAttrs>newDefaultInstance(
+                org.tensorflow.proto.framework.FunctionProtos.internal_static_tensorflow_FunctionDef_ArgAttrEntry_descriptor, 
+                com.google.protobuf.WireFormat.FieldType.UINT32,
+                0,
+                com.google.protobuf.WireFormat.FieldType.MESSAGE,
+                org.tensorflow.proto.framework.FunctionDef.ArgAttrs.getDefaultInstance());
+  }
+  private com.google.protobuf.MapField<
+      java.lang.Integer, org.tensorflow.proto.framework.FunctionDef.ArgAttrs> argAttr_;
+  private com.google.protobuf.MapField<java.lang.Integer, org.tensorflow.proto.framework.FunctionDef.ArgAttrs>
+  internalGetArgAttr() {
+    if (argAttr_ == null) {
+      return com.google.protobuf.MapField.emptyMapField(
+          ArgAttrDefaultEntryHolder.defaultEntry);
+    }
+    return argAttr_;
+  }
+
+  public int getArgAttrCount() {
+    return internalGetArgAttr().getMap().size();
+  }
+  /**
+   * <code>map&lt;uint32, .tensorflow.FunctionDef.ArgAttrs&gt; arg_attr = 7;</code>
+   */
+
+  public boolean containsArgAttr(
+      int key) {
+    
+    return internalGetArgAttr().getMap().containsKey(key);
+  }
+  /**
+   * Use {@link #getArgAttrMap()} instead.
+   */
+  @java.lang.Deprecated
+  public java.util.Map<java.lang.Integer, org.tensorflow.proto.framework.FunctionDef.ArgAttrs> getArgAttr() {
+    return getArgAttrMap();
+  }
+  /**
+   * <code>map&lt;uint32, .tensorflow.FunctionDef.ArgAttrs&gt; arg_attr = 7;</code>
+   */
+
+  public java.util.Map<java.lang.Integer, org.tensorflow.proto.framework.FunctionDef.ArgAttrs> getArgAttrMap() {
+    return internalGetArgAttr().getMap();
+  }
+  /**
+   * <code>map&lt;uint32, .tensorflow.FunctionDef.ArgAttrs&gt; arg_attr = 7;</code>
+   */
+
+  public org.tensorflow.proto.framework.FunctionDef.ArgAttrs getArgAttrOrDefault(
+      int key,
+      org.tensorflow.proto.framework.FunctionDef.ArgAttrs defaultValue) {
+    
+    java.util.Map<java.lang.Integer, org.tensorflow.proto.framework.FunctionDef.ArgAttrs> map =
+        internalGetArgAttr().getMap();
+    return map.containsKey(key) ? map.get(key) : defaultValue;
+  }
+  /**
+   * <code>map&lt;uint32, .tensorflow.FunctionDef.ArgAttrs&gt; arg_attr = 7;</code>
+   */
+
+  public org.tensorflow.proto.framework.FunctionDef.ArgAttrs getArgAttrOrThrow(
+      int key) {
+    
+    java.util.Map<java.lang.Integer, org.tensorflow.proto.framework.FunctionDef.ArgAttrs> map =
+        internalGetArgAttr().getMap();
+    if (!map.containsKey(key)) {
+      throw new java.lang.IllegalArgumentException();
+    }
+    return map.get(key);
+  }
+
+  public static final int RESOURCE_ARG_UNIQUE_ID_FIELD_NUMBER = 8;
+  private static final class ResourceArgUniqueIdDefaultEntryHolder {
+    static final com.google.protobuf.MapEntry<
+        java.lang.Integer, java.lang.Integer> defaultEntry =
+            com.google.protobuf.MapEntry
+            .<java.lang.Integer, java.lang.Integer>newDefaultInstance(
+                org.tensorflow.proto.framework.FunctionProtos.internal_static_tensorflow_FunctionDef_ResourceArgUniqueIdEntry_descriptor, 
+                com.google.protobuf.WireFormat.FieldType.UINT32,
+                0,
+                com.google.protobuf.WireFormat.FieldType.UINT32,
+                0);
+  }
+  private com.google.protobuf.MapField<
+      java.lang.Integer, java.lang.Integer> resourceArgUniqueId_;
+  private com.google.protobuf.MapField<java.lang.Integer, java.lang.Integer>
+  internalGetResourceArgUniqueId() {
+    if (resourceArgUniqueId_ == null) {
+      return com.google.protobuf.MapField.emptyMapField(
+          ResourceArgUniqueIdDefaultEntryHolder.defaultEntry);
+    }
+    return resourceArgUniqueId_;
+  }
+
+  public int getResourceArgUniqueIdCount() {
+    return internalGetResourceArgUniqueId().getMap().size();
+  }
+  /**
+   * <pre>
+   * Unique IDs for each resource argument, used to track aliasing resources. If
+   * Argument A and Argument B alias each other, then
+   * resource_arg_unique_ids[A.index] == resource_arg_unique_ids[B.index].
+   * If this field is empty, none of the arguments could alias; otherwise, every
+   * resource argument should have an entry in this field.
+   * When instantiated, the unique IDs will be attached to the _Arg nodes'
+   * "_resource_arg_unique_id" attribute.
+   * </pre>
+   *
+   * <code>map&lt;uint32, uint32&gt; resource_arg_unique_id = 8;</code>
+   */
+
+  public boolean containsResourceArgUniqueId(
+      int key) {
+    
+    return internalGetResourceArgUniqueId().getMap().containsKey(key);
+  }
+  /**
+   * Use {@link #getResourceArgUniqueIdMap()} instead.
+   */
+  @java.lang.Deprecated
+  public java.util.Map<java.lang.Integer, java.lang.Integer> getResourceArgUniqueId() {
+    return getResourceArgUniqueIdMap();
+  }
+  /**
+   * <pre>
+   * Unique IDs for each resource argument, used to track aliasing resources. If
+   * Argument A and Argument B alias each other, then
+   * resource_arg_unique_ids[A.index] == resource_arg_unique_ids[B.index].
+   * If this field is empty, none of the arguments could alias; otherwise, every
+   * resource argument should have an entry in this field.
+   * When instantiated, the unique IDs will be attached to the _Arg nodes'
+   * "_resource_arg_unique_id" attribute.
+   * </pre>
+   *
+   * <code>map&lt;uint32, uint32&gt; resource_arg_unique_id = 8;</code>
+   */
+
+  public java.util.Map<java.lang.Integer, java.lang.Integer> getResourceArgUniqueIdMap() {
+    return internalGetResourceArgUniqueId().getMap();
+  }
+  /**
+   * <pre>
+   * Unique IDs for each resource argument, used to track aliasing resources. If
+   * Argument A and Argument B alias each other, then
+   * resource_arg_unique_ids[A.index] == resource_arg_unique_ids[B.index].
+   * If this field is empty, none of the arguments could alias; otherwise, every
+   * resource argument should have an entry in this field.
+   * When instantiated, the unique IDs will be attached to the _Arg nodes'
+   * "_resource_arg_unique_id" attribute.
+   * </pre>
+   *
+   * <code>map&lt;uint32, uint32&gt; resource_arg_unique_id = 8;</code>
+   */
+
+  public int getResourceArgUniqueIdOrDefault(
+      int key,
+      int defaultValue) {
+    
+    java.util.Map<java.lang.Integer, java.lang.Integer> map =
+        internalGetResourceArgUniqueId().getMap();
+    return map.containsKey(key) ? map.get(key) : defaultValue;
+  }
+  /**
+   * <pre>
+   * Unique IDs for each resource argument, used to track aliasing resources. If
+   * Argument A and Argument B alias each other, then
+   * resource_arg_unique_ids[A.index] == resource_arg_unique_ids[B.index].
+   * If this field is empty, none of the arguments could alias; otherwise, every
+   * resource argument should have an entry in this field.
+   * When instantiated, the unique IDs will be attached to the _Arg nodes'
+   * "_resource_arg_unique_id" attribute.
+   * </pre>
+   *
+   * <code>map&lt;uint32, uint32&gt; resource_arg_unique_id = 8;</code>
+   */
+
+  public int getResourceArgUniqueIdOrThrow(
+      int key) {
+    
+    java.util.Map<java.lang.Integer, java.lang.Integer> map =
+        internalGetResourceArgUniqueId().getMap();
+    if (!map.containsKey(key)) {
+      throw new java.lang.IllegalArgumentException();
+    }
+    return map.get(key);
+  }
+
+  public static final int NODE_DEF_FIELD_NUMBER = 3;
+  private java.util.List<org.tensorflow.proto.framework.NodeDef> nodeDef_;
+  /**
+   * <pre>
+   * By convention, "op" in node_def is resolved by consulting with a
+   * user-defined library first. If not resolved, "func" is assumed to
+   * be a builtin op.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.NodeDef node_def = 3;</code>
+   */
+  public java.util.List<org.tensorflow.proto.framework.NodeDef> getNodeDefList() {
+    return nodeDef_;
+  }
+  /**
+   * <pre>
+   * By convention, "op" in node_def is resolved by consulting with a
+   * user-defined library first. If not resolved, "func" is assumed to
+   * be a builtin op.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.NodeDef node_def = 3;</code>
+   */
+  public java.util.List<? extends org.tensorflow.proto.framework.NodeDefOrBuilder> 
+      getNodeDefOrBuilderList() {
+    return nodeDef_;
+  }
+  /**
+   * <pre>
+   * By convention, "op" in node_def is resolved by consulting with a
+   * user-defined library first. If not resolved, "func" is assumed to
+   * be a builtin op.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.NodeDef node_def = 3;</code>
+   */
+  public int getNodeDefCount() {
+    return nodeDef_.size();
+  }
+  /**
+   * <pre>
+   * By convention, "op" in node_def is resolved by consulting with a
+   * user-defined library first. If not resolved, "func" is assumed to
+   * be a builtin op.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.NodeDef node_def = 3;</code>
+   */
+  public org.tensorflow.proto.framework.NodeDef getNodeDef(int index) {
+    return nodeDef_.get(index);
+  }
+  /**
+   * <pre>
+   * By convention, "op" in node_def is resolved by consulting with a
+   * user-defined library first. If not resolved, "func" is assumed to
+   * be a builtin op.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.NodeDef node_def = 3;</code>
+   */
+  public org.tensorflow.proto.framework.NodeDefOrBuilder getNodeDefOrBuilder(
+      int index) {
+    return nodeDef_.get(index);
+  }
+
+  public static final int RET_FIELD_NUMBER = 4;
+  private static final class RetDefaultEntryHolder {
+    static final com.google.protobuf.MapEntry<
+        java.lang.String, java.lang.String> defaultEntry =
+            com.google.protobuf.MapEntry
+            .<java.lang.String, java.lang.String>newDefaultInstance(
+                org.tensorflow.proto.framework.FunctionProtos.internal_static_tensorflow_FunctionDef_RetEntry_descriptor, 
+                com.google.protobuf.WireFormat.FieldType.STRING,
+                "",
+                com.google.protobuf.WireFormat.FieldType.STRING,
+                "");
+  }
+  private com.google.protobuf.MapField<
+      java.lang.String, java.lang.String> ret_;
+  private com.google.protobuf.MapField<java.lang.String, java.lang.String>
+  internalGetRet() {
+    if (ret_ == null) {
+      return com.google.protobuf.MapField.emptyMapField(
+          RetDefaultEntryHolder.defaultEntry);
+    }
+    return ret_;
+  }
+
+  public int getRetCount() {
+    return internalGetRet().getMap().size();
+  }
+  /**
+   * <pre>
+   * A mapping from the output arg names from `signature` to the
+   * outputs from `node_def` that should be returned by the function.
+   * </pre>
+   *
+   * <code>map&lt;string, string&gt; ret = 4;</code>
+   */
+
+  public boolean containsRet(
+      java.lang.String key) {
+    if (key == null) { throw new java.lang.NullPointerException(); }
+    return internalGetRet().getMap().containsKey(key);
+  }
+  /**
+   * Use {@link #getRetMap()} instead.
+   */
+  @java.lang.Deprecated
+  public java.util.Map<java.lang.String, java.lang.String> getRet() {
+    return getRetMap();
+  }
+  /**
+   * <pre>
+   * A mapping from the output arg names from `signature` to the
+   * outputs from `node_def` that should be returned by the function.
+   * </pre>
+   *
+   * <code>map&lt;string, string&gt; ret = 4;</code>
+   */
+
+  public java.util.Map<java.lang.String, java.lang.String> getRetMap() {
+    return internalGetRet().getMap();
+  }
+  /**
+   * <pre>
+   * A mapping from the output arg names from `signature` to the
+   * outputs from `node_def` that should be returned by the function.
+   * </pre>
+   *
+   * <code>map&lt;string, string&gt; ret = 4;</code>
+   */
+
+  public java.lang.String getRetOrDefault(
+      java.lang.String key,
+      java.lang.String defaultValue) {
+    if (key == null) { throw new java.lang.NullPointerException(); }
+    java.util.Map<java.lang.String, java.lang.String> map =
+        internalGetRet().getMap();
+    return map.containsKey(key) ? map.get(key) : defaultValue;
+  }
+  /**
+   * <pre>
+   * A mapping from the output arg names from `signature` to the
+   * outputs from `node_def` that should be returned by the function.
+   * </pre>
+   *
+   * <code>map&lt;string, string&gt; ret = 4;</code>
+   */
+
+  public java.lang.String getRetOrThrow(
+      java.lang.String key) {
+    if (key == null) { throw new java.lang.NullPointerException(); }
+    java.util.Map<java.lang.String, java.lang.String> map =
+        internalGetRet().getMap();
+    if (!map.containsKey(key)) {
+      throw new java.lang.IllegalArgumentException();
+    }
+    return map.get(key);
+  }
+
+  public static final int CONTROL_RET_FIELD_NUMBER = 6;
+  private static final class ControlRetDefaultEntryHolder {
+    static final com.google.protobuf.MapEntry<
+        java.lang.String, java.lang.String> defaultEntry =
+            com.google.protobuf.MapEntry
+            .<java.lang.String, java.lang.String>newDefaultInstance(
+                org.tensorflow.proto.framework.FunctionProtos.internal_static_tensorflow_FunctionDef_ControlRetEntry_descriptor, 
+                com.google.protobuf.WireFormat.FieldType.STRING,
+                "",
+                com.google.protobuf.WireFormat.FieldType.STRING,
+                "");
+  }
+  private com.google.protobuf.MapField<
+      java.lang.String, java.lang.String> controlRet_;
+  private com.google.protobuf.MapField<java.lang.String, java.lang.String>
+  internalGetControlRet() {
+    if (controlRet_ == null) {
+      return com.google.protobuf.MapField.emptyMapField(
+          ControlRetDefaultEntryHolder.defaultEntry);
+    }
+    return controlRet_;
+  }
+
+  public int getControlRetCount() {
+    return internalGetControlRet().getMap().size();
+  }
+  /**
+   * <pre>
+   * A mapping from control output names from `signature` to node names in
+   * `node_def` which should be control outputs of this function.
+   * </pre>
+   *
+   * <code>map&lt;string, string&gt; control_ret = 6;</code>
+   */
+
+  public boolean containsControlRet(
+      java.lang.String key) {
+    if (key == null) { throw new java.lang.NullPointerException(); }
+    return internalGetControlRet().getMap().containsKey(key);
+  }
+  /**
+   * Use {@link #getControlRetMap()} instead.
+   */
+  @java.lang.Deprecated
+  public java.util.Map<java.lang.String, java.lang.String> getControlRet() {
+    return getControlRetMap();
+  }
+  /**
+   * <pre>
+   * A mapping from control output names from `signature` to node names in
+   * `node_def` which should be control outputs of this function.
+   * </pre>
+   *
+   * <code>map&lt;string, string&gt; control_ret = 6;</code>
+   */
+
+  public java.util.Map<java.lang.String, java.lang.String> getControlRetMap() {
+    return internalGetControlRet().getMap();
+  }
+  /**
+   * <pre>
+   * A mapping from control output names from `signature` to node names in
+   * `node_def` which should be control outputs of this function.
+   * </pre>
+   *
+   * <code>map&lt;string, string&gt; control_ret = 6;</code>
+   */
+
+  public java.lang.String getControlRetOrDefault(
+      java.lang.String key,
+      java.lang.String defaultValue) {
+    if (key == null) { throw new java.lang.NullPointerException(); }
+    java.util.Map<java.lang.String, java.lang.String> map =
+        internalGetControlRet().getMap();
+    return map.containsKey(key) ? map.get(key) : defaultValue;
+  }
+  /**
+   * <pre>
+   * A mapping from control output names from `signature` to node names in
+   * `node_def` which should be control outputs of this function.
+   * </pre>
+   *
+   * <code>map&lt;string, string&gt; control_ret = 6;</code>
+   */
+
+  public java.lang.String getControlRetOrThrow(
+      java.lang.String key) {
+    if (key == null) { throw new java.lang.NullPointerException(); }
+    java.util.Map<java.lang.String, java.lang.String> map =
+        internalGetControlRet().getMap();
+    if (!map.containsKey(key)) {
+      throw new java.lang.IllegalArgumentException();
+    }
+    return map.get(key);
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    if (signature_ != null) {
+      output.writeMessage(1, getSignature());
+    }
+    for (int i = 0; i < nodeDef_.size(); i++) {
+      output.writeMessage(3, nodeDef_.get(i));
+    }
+    com.google.protobuf.GeneratedMessageV3
+      .serializeStringMapTo(
+        output,
+        internalGetRet(),
+        RetDefaultEntryHolder.defaultEntry,
+        4);
+    com.google.protobuf.GeneratedMessageV3
+      .serializeStringMapTo(
+        output,
+        internalGetAttr(),
+        AttrDefaultEntryHolder.defaultEntry,
+        5);
+    com.google.protobuf.GeneratedMessageV3
+      .serializeStringMapTo(
+        output,
+        internalGetControlRet(),
+        ControlRetDefaultEntryHolder.defaultEntry,
+        6);
+    com.google.protobuf.GeneratedMessageV3
+      .serializeIntegerMapTo(
+        output,
+        internalGetArgAttr(),
+        ArgAttrDefaultEntryHolder.defaultEntry,
+        7);
+    com.google.protobuf.GeneratedMessageV3
+      .serializeIntegerMapTo(
+        output,
+        internalGetResourceArgUniqueId(),
+        ResourceArgUniqueIdDefaultEntryHolder.defaultEntry,
+        8);
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (signature_ != null) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(1, getSignature());
+    }
+    for (int i = 0; i < nodeDef_.size(); i++) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(3, nodeDef_.get(i));
+    }
+    for (java.util.Map.Entry<java.lang.String, java.lang.String> entry
+         : internalGetRet().getMap().entrySet()) {
+      com.google.protobuf.MapEntry<java.lang.String, java.lang.String>
+      ret__ = RetDefaultEntryHolder.defaultEntry.newBuilderForType()
+          .setKey(entry.getKey())
+          .setValue(entry.getValue())
+          .build();
+      size += com.google.protobuf.CodedOutputStream
+          .computeMessageSize(4, ret__);
+    }
+    for (java.util.Map.Entry<java.lang.String, org.tensorflow.proto.framework.AttrValue> entry
+         : internalGetAttr().getMap().entrySet()) {
+      com.google.protobuf.MapEntry<java.lang.String, org.tensorflow.proto.framework.AttrValue>
+      attr__ = AttrDefaultEntryHolder.defaultEntry.newBuilderForType()
+          .setKey(entry.getKey())
+          .setValue(entry.getValue())
+          .build();
+      size += com.google.protobuf.CodedOutputStream
+          .computeMessageSize(5, attr__);
+    }
+    for (java.util.Map.Entry<java.lang.String, java.lang.String> entry
+         : internalGetControlRet().getMap().entrySet()) {
+      com.google.protobuf.MapEntry<java.lang.String, java.lang.String>
+      controlRet__ = ControlRetDefaultEntryHolder.defaultEntry.newBuilderForType()
+          .setKey(entry.getKey())
+          .setValue(entry.getValue())
+          .build();
+      size += com.google.protobuf.CodedOutputStream
+          .computeMessageSize(6, controlRet__);
+    }
+    for (java.util.Map.Entry<java.lang.Integer, org.tensorflow.proto.framework.FunctionDef.ArgAttrs> entry
+         : internalGetArgAttr().getMap().entrySet()) {
+      com.google.protobuf.MapEntry<java.lang.Integer, org.tensorflow.proto.framework.FunctionDef.ArgAttrs>
+      argAttr__ = ArgAttrDefaultEntryHolder.defaultEntry.newBuilderForType()
+          .setKey(entry.getKey())
+          .setValue(entry.getValue())
+          .build();
+      size += com.google.protobuf.CodedOutputStream
+          .computeMessageSize(7, argAttr__);
+    }
+    for (java.util.Map.Entry<java.lang.Integer, java.lang.Integer> entry
+         : internalGetResourceArgUniqueId().getMap().entrySet()) {
+      com.google.protobuf.MapEntry<java.lang.Integer, java.lang.Integer>
+      resourceArgUniqueId__ = ResourceArgUniqueIdDefaultEntryHolder.defaultEntry.newBuilderForType()
+          .setKey(entry.getKey())
+          .setValue(entry.getValue())
+          .build();
+      size += com.google.protobuf.CodedOutputStream
+          .computeMessageSize(8, resourceArgUniqueId__);
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.framework.FunctionDef)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.framework.FunctionDef other = (org.tensorflow.proto.framework.FunctionDef) obj;
+
+    if (hasSignature() != other.hasSignature()) return false;
+    if (hasSignature()) {
+      if (!getSignature()
+          .equals(other.getSignature())) return false;
+    }
+    if (!internalGetAttr().equals(
+        other.internalGetAttr())) return false;
+    if (!internalGetArgAttr().equals(
+        other.internalGetArgAttr())) return false;
+    if (!internalGetResourceArgUniqueId().equals(
+        other.internalGetResourceArgUniqueId())) return false;
+    if (!getNodeDefList()
+        .equals(other.getNodeDefList())) return false;
+    if (!internalGetRet().equals(
+        other.internalGetRet())) return false;
+    if (!internalGetControlRet().equals(
+        other.internalGetControlRet())) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    if (hasSignature()) {
+      hash = (37 * hash) + SIGNATURE_FIELD_NUMBER;
+      hash = (53 * hash) + getSignature().hashCode();
+    }
+    if (!internalGetAttr().getMap().isEmpty()) {
+      hash = (37 * hash) + ATTR_FIELD_NUMBER;
+      hash = (53 * hash) + internalGetAttr().hashCode();
+    }
+    if (!internalGetArgAttr().getMap().isEmpty()) {
+      hash = (37 * hash) + ARG_ATTR_FIELD_NUMBER;
+      hash = (53 * hash) + internalGetArgAttr().hashCode();
+    }
+    if (!internalGetResourceArgUniqueId().getMap().isEmpty()) {
+      hash = (37 * hash) + RESOURCE_ARG_UNIQUE_ID_FIELD_NUMBER;
+      hash = (53 * hash) + internalGetResourceArgUniqueId().hashCode();
+    }
+    if (getNodeDefCount() > 0) {
+      hash = (37 * hash) + NODE_DEF_FIELD_NUMBER;
+      hash = (53 * hash) + getNodeDefList().hashCode();
+    }
+    if (!internalGetRet().getMap().isEmpty()) {
+      hash = (37 * hash) + RET_FIELD_NUMBER;
+      hash = (53 * hash) + internalGetRet().hashCode();
+    }
+    if (!internalGetControlRet().getMap().isEmpty()) {
+      hash = (37 * hash) + CONTROL_RET_FIELD_NUMBER;
+      hash = (53 * hash) + internalGetControlRet().hashCode();
+    }
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.framework.FunctionDef parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.FunctionDef parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.FunctionDef parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.FunctionDef parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.FunctionDef parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.FunctionDef parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.FunctionDef parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.FunctionDef parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.FunctionDef parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.FunctionDef parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.FunctionDef parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.FunctionDef parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.framework.FunctionDef prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * <pre>
+   * A function can be instantiated when the runtime can bind every attr
+   * with a value. When a GraphDef has a call to a function, it must
+   * have binding for every attr defined in the signature.
+   * TODO(zhifengc):
+   *   * device spec, etc.
+   * </pre>
+   *
+   * Protobuf type {@code tensorflow.FunctionDef}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.FunctionDef)
+      org.tensorflow.proto.framework.FunctionDefOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.FunctionProtos.internal_static_tensorflow_FunctionDef_descriptor;
+    }
+
+    @SuppressWarnings({"rawtypes"})
+    protected com.google.protobuf.MapField internalGetMapField(
+        int number) {
+      switch (number) {
+        case 5:
+          return internalGetAttr();
+        case 7:
+          return internalGetArgAttr();
+        case 8:
+          return internalGetResourceArgUniqueId();
+        case 4:
+          return internalGetRet();
+        case 6:
+          return internalGetControlRet();
+        default:
+          throw new RuntimeException(
+              "Invalid map field number: " + number);
+      }
+    }
+    @SuppressWarnings({"rawtypes"})
+    protected com.google.protobuf.MapField internalGetMutableMapField(
+        int number) {
+      switch (number) {
+        case 5:
+          return internalGetMutableAttr();
+        case 7:
+          return internalGetMutableArgAttr();
+        case 8:
+          return internalGetMutableResourceArgUniqueId();
+        case 4:
+          return internalGetMutableRet();
+        case 6:
+          return internalGetMutableControlRet();
+        default:
+          throw new RuntimeException(
+              "Invalid map field number: " + number);
+      }
+    }
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.FunctionProtos.internal_static_tensorflow_FunctionDef_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.FunctionDef.class, org.tensorflow.proto.framework.FunctionDef.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.framework.FunctionDef.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+        getNodeDefFieldBuilder();
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      if (signatureBuilder_ == null) {
+        signature_ = null;
+      } else {
+        signature_ = null;
+        signatureBuilder_ = null;
+      }
+      internalGetMutableAttr().clear();
+      internalGetMutableArgAttr().clear();
+      internalGetMutableResourceArgUniqueId().clear();
+      if (nodeDefBuilder_ == null) {
+        nodeDef_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000008);
+      } else {
+        nodeDefBuilder_.clear();
+      }
+      internalGetMutableRet().clear();
+      internalGetMutableControlRet().clear();
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.framework.FunctionProtos.internal_static_tensorflow_FunctionDef_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.FunctionDef getDefaultInstanceForType() {
+      return org.tensorflow.proto.framework.FunctionDef.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.FunctionDef build() {
+      org.tensorflow.proto.framework.FunctionDef result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.FunctionDef buildPartial() {
+      org.tensorflow.proto.framework.FunctionDef result = new org.tensorflow.proto.framework.FunctionDef(this);
+      int from_bitField0_ = bitField0_;
+      if (signatureBuilder_ == null) {
+        result.signature_ = signature_;
+      } else {
+        result.signature_ = signatureBuilder_.build();
+      }
+      result.attr_ = internalGetAttr();
+      result.attr_.makeImmutable();
+      result.argAttr_ = internalGetArgAttr();
+      result.argAttr_.makeImmutable();
+      result.resourceArgUniqueId_ = internalGetResourceArgUniqueId();
+      result.resourceArgUniqueId_.makeImmutable();
+      if (nodeDefBuilder_ == null) {
+        if (((bitField0_ & 0x00000008) != 0)) {
+          nodeDef_ = java.util.Collections.unmodifiableList(nodeDef_);
+          bitField0_ = (bitField0_ & ~0x00000008);
+        }
+        result.nodeDef_ = nodeDef_;
+      } else {
+        result.nodeDef_ = nodeDefBuilder_.build();
+      }
+      result.ret_ = internalGetRet();
+      result.ret_.makeImmutable();
+      result.controlRet_ = internalGetControlRet();
+      result.controlRet_.makeImmutable();
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.framework.FunctionDef) {
+        return mergeFrom((org.tensorflow.proto.framework.FunctionDef)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.framework.FunctionDef other) {
+      if (other == org.tensorflow.proto.framework.FunctionDef.getDefaultInstance()) return this;
+      if (other.hasSignature()) {
+        mergeSignature(other.getSignature());
+      }
+      internalGetMutableAttr().mergeFrom(
+          other.internalGetAttr());
+      internalGetMutableArgAttr().mergeFrom(
+          other.internalGetArgAttr());
+      internalGetMutableResourceArgUniqueId().mergeFrom(
+          other.internalGetResourceArgUniqueId());
+      if (nodeDefBuilder_ == null) {
+        if (!other.nodeDef_.isEmpty()) {
+          if (nodeDef_.isEmpty()) {
+            nodeDef_ = other.nodeDef_;
+            bitField0_ = (bitField0_ & ~0x00000008);
+          } else {
+            ensureNodeDefIsMutable();
+            nodeDef_.addAll(other.nodeDef_);
+          }
+          onChanged();
+        }
+      } else {
+        if (!other.nodeDef_.isEmpty()) {
+          if (nodeDefBuilder_.isEmpty()) {
+            nodeDefBuilder_.dispose();
+            nodeDefBuilder_ = null;
+            nodeDef_ = other.nodeDef_;
+            bitField0_ = (bitField0_ & ~0x00000008);
+            nodeDefBuilder_ = 
+              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
+                 getNodeDefFieldBuilder() : null;
+          } else {
+            nodeDefBuilder_.addAllMessages(other.nodeDef_);
+          }
+        }
+      }
+      internalGetMutableRet().mergeFrom(
+          other.internalGetRet());
+      internalGetMutableControlRet().mergeFrom(
+          other.internalGetControlRet());
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.framework.FunctionDef parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.framework.FunctionDef) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+    private int bitField0_;
+
+    private org.tensorflow.proto.framework.OpDef signature_;
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.OpDef, org.tensorflow.proto.framework.OpDef.Builder, org.tensorflow.proto.framework.OpDefOrBuilder> signatureBuilder_;
+    /**
+     * <pre>
+     * The definition of the function's name, arguments, return values,
+     * attrs etc.
+     * </pre>
+     *
+     * <code>.tensorflow.OpDef signature = 1;</code>
+     */
+    public boolean hasSignature() {
+      return signatureBuilder_ != null || signature_ != null;
+    }
+    /**
+     * <pre>
+     * The definition of the function's name, arguments, return values,
+     * attrs etc.
+     * </pre>
+     *
+     * <code>.tensorflow.OpDef signature = 1;</code>
+     */
+    public org.tensorflow.proto.framework.OpDef getSignature() {
+      if (signatureBuilder_ == null) {
+        return signature_ == null ? org.tensorflow.proto.framework.OpDef.getDefaultInstance() : signature_;
+      } else {
+        return signatureBuilder_.getMessage();
+      }
+    }
+    /**
+     * <pre>
+     * The definition of the function's name, arguments, return values,
+     * attrs etc.
+     * </pre>
+     *
+     * <code>.tensorflow.OpDef signature = 1;</code>
+     */
+    public Builder setSignature(org.tensorflow.proto.framework.OpDef value) {
+      if (signatureBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        signature_ = value;
+        onChanged();
+      } else {
+        signatureBuilder_.setMessage(value);
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * The definition of the function's name, arguments, return values,
+     * attrs etc.
+     * </pre>
+     *
+     * <code>.tensorflow.OpDef signature = 1;</code>
+     */
+    public Builder setSignature(
+        org.tensorflow.proto.framework.OpDef.Builder builderForValue) {
+      if (signatureBuilder_ == null) {
+        signature_ = builderForValue.build();
+        onChanged();
+      } else {
+        signatureBuilder_.setMessage(builderForValue.build());
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * The definition of the function's name, arguments, return values,
+     * attrs etc.
+     * </pre>
+     *
+     * <code>.tensorflow.OpDef signature = 1;</code>
+     */
+    public Builder mergeSignature(org.tensorflow.proto.framework.OpDef value) {
+      if (signatureBuilder_ == null) {
+        if (signature_ != null) {
+          signature_ =
+            org.tensorflow.proto.framework.OpDef.newBuilder(signature_).mergeFrom(value).buildPartial();
+        } else {
+          signature_ = value;
+        }
+        onChanged();
+      } else {
+        signatureBuilder_.mergeFrom(value);
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * The definition of the function's name, arguments, return values,
+     * attrs etc.
+     * </pre>
+     *
+     * <code>.tensorflow.OpDef signature = 1;</code>
+     */
+    public Builder clearSignature() {
+      if (signatureBuilder_ == null) {
+        signature_ = null;
+        onChanged();
+      } else {
+        signature_ = null;
+        signatureBuilder_ = null;
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * The definition of the function's name, arguments, return values,
+     * attrs etc.
+     * </pre>
+     *
+     * <code>.tensorflow.OpDef signature = 1;</code>
+     */
+    public org.tensorflow.proto.framework.OpDef.Builder getSignatureBuilder() {
+      
+      onChanged();
+      return getSignatureFieldBuilder().getBuilder();
+    }
+    /**
+     * <pre>
+     * The definition of the function's name, arguments, return values,
+     * attrs etc.
+     * </pre>
+     *
+     * <code>.tensorflow.OpDef signature = 1;</code>
+     */
+    public org.tensorflow.proto.framework.OpDefOrBuilder getSignatureOrBuilder() {
+      if (signatureBuilder_ != null) {
+        return signatureBuilder_.getMessageOrBuilder();
+      } else {
+        return signature_ == null ?
+            org.tensorflow.proto.framework.OpDef.getDefaultInstance() : signature_;
+      }
+    }
+    /**
+     * <pre>
+     * The definition of the function's name, arguments, return values,
+     * attrs etc.
+     * </pre>
+     *
+     * <code>.tensorflow.OpDef signature = 1;</code>
+     */
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.OpDef, org.tensorflow.proto.framework.OpDef.Builder, org.tensorflow.proto.framework.OpDefOrBuilder> 
+        getSignatureFieldBuilder() {
+      if (signatureBuilder_ == null) {
+        signatureBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+            org.tensorflow.proto.framework.OpDef, org.tensorflow.proto.framework.OpDef.Builder, org.tensorflow.proto.framework.OpDefOrBuilder>(
+                getSignature(),
+                getParentForChildren(),
+                isClean());
+        signature_ = null;
+      }
+      return signatureBuilder_;
+    }
+
+    private com.google.protobuf.MapField<
+        java.lang.String, org.tensorflow.proto.framework.AttrValue> attr_;
+    private com.google.protobuf.MapField<java.lang.String, org.tensorflow.proto.framework.AttrValue>
+    internalGetAttr() {
+      if (attr_ == null) {
+        return com.google.protobuf.MapField.emptyMapField(
+            AttrDefaultEntryHolder.defaultEntry);
+      }
+      return attr_;
+    }
+    private com.google.protobuf.MapField<java.lang.String, org.tensorflow.proto.framework.AttrValue>
+    internalGetMutableAttr() {
+      onChanged();;
+      if (attr_ == null) {
+        attr_ = com.google.protobuf.MapField.newMapField(
+            AttrDefaultEntryHolder.defaultEntry);
+      }
+      if (!attr_.isMutable()) {
+        attr_ = attr_.copy();
+      }
+      return attr_;
+    }
+
+    public int getAttrCount() {
+      return internalGetAttr().getMap().size();
+    }
+    /**
+     * <pre>
+     * Attributes specific to this function definition.
+     * </pre>
+     *
+     * <code>map&lt;string, .tensorflow.AttrValue&gt; attr = 5;</code>
+     */
+
+    public boolean containsAttr(
+        java.lang.String key) {
+      if (key == null) { throw new java.lang.NullPointerException(); }
+      return internalGetAttr().getMap().containsKey(key);
+    }
+    /**
+     * Use {@link #getAttrMap()} instead.
+     */
+    @java.lang.Deprecated
+    public java.util.Map<java.lang.String, org.tensorflow.proto.framework.AttrValue> getAttr() {
+      return getAttrMap();
+    }
+    /**
+     * <pre>
+     * Attributes specific to this function definition.
+     * </pre>
+     *
+     * <code>map&lt;string, .tensorflow.AttrValue&gt; attr = 5;</code>
+     */
+
+    public java.util.Map<java.lang.String, org.tensorflow.proto.framework.AttrValue> getAttrMap() {
+      return internalGetAttr().getMap();
+    }
+    /**
+     * <pre>
+     * Attributes specific to this function definition.
+     * </pre>
+     *
+     * <code>map&lt;string, .tensorflow.AttrValue&gt; attr = 5;</code>
+     */
+
+    public org.tensorflow.proto.framework.AttrValue getAttrOrDefault(
+        java.lang.String key,
+        org.tensorflow.proto.framework.AttrValue defaultValue) {
+      if (key == null) { throw new java.lang.NullPointerException(); }
+      java.util.Map<java.lang.String, org.tensorflow.proto.framework.AttrValue> map =
+          internalGetAttr().getMap();
+      return map.containsKey(key) ? map.get(key) : defaultValue;
+    }
+    /**
+     * <pre>
+     * Attributes specific to this function definition.
+     * </pre>
+     *
+     * <code>map&lt;string, .tensorflow.AttrValue&gt; attr = 5;</code>
+     */
+
+    public org.tensorflow.proto.framework.AttrValue getAttrOrThrow(
+        java.lang.String key) {
+      if (key == null) { throw new java.lang.NullPointerException(); }
+      java.util.Map<java.lang.String, org.tensorflow.proto.framework.AttrValue> map =
+          internalGetAttr().getMap();
+      if (!map.containsKey(key)) {
+        throw new java.lang.IllegalArgumentException();
+      }
+      return map.get(key);
+    }
+
+    public Builder clearAttr() {
+      internalGetMutableAttr().getMutableMap()
+          .clear();
+      return this;
+    }
+    /**
+     * <pre>
+     * Attributes specific to this function definition.
+     * </pre>
+     *
+     * <code>map&lt;string, .tensorflow.AttrValue&gt; attr = 5;</code>
+     */
+
+    public Builder removeAttr(
+        java.lang.String key) {
+      if (key == null) { throw new java.lang.NullPointerException(); }
+      internalGetMutableAttr().getMutableMap()
+          .remove(key);
+      return this;
+    }
+    /**
+     * Use alternate mutation accessors instead.
+     */
+    @java.lang.Deprecated
+    public java.util.Map<java.lang.String, org.tensorflow.proto.framework.AttrValue>
+    getMutableAttr() {
+      return internalGetMutableAttr().getMutableMap();
+    }
+    /**
+     * <pre>
+     * Attributes specific to this function definition.
+     * </pre>
+     *
+     * <code>map&lt;string, .tensorflow.AttrValue&gt; attr = 5;</code>
+     */
+    public Builder putAttr(
+        java.lang.String key,
+        org.tensorflow.proto.framework.AttrValue value) {
+      if (key == null) { throw new java.lang.NullPointerException(); }
+      if (value == null) { throw new java.lang.NullPointerException(); }
+      internalGetMutableAttr().getMutableMap()
+          .put(key, value);
+      return this;
+    }
+    /**
+     * <pre>
+     * Attributes specific to this function definition.
+     * </pre>
+     *
+     * <code>map&lt;string, .tensorflow.AttrValue&gt; attr = 5;</code>
+     */
+
+    public Builder putAllAttr(
+        java.util.Map<java.lang.String, org.tensorflow.proto.framework.AttrValue> values) {
+      internalGetMutableAttr().getMutableMap()
+          .putAll(values);
+      return this;
+    }
+
+    private com.google.protobuf.MapField<
+        java.lang.Integer, org.tensorflow.proto.framework.FunctionDef.ArgAttrs> argAttr_;
+    private com.google.protobuf.MapField<java.lang.Integer, org.tensorflow.proto.framework.FunctionDef.ArgAttrs>
+    internalGetArgAttr() {
+      if (argAttr_ == null) {
+        return com.google.protobuf.MapField.emptyMapField(
+            ArgAttrDefaultEntryHolder.defaultEntry);
+      }
+      return argAttr_;
+    }
+    private com.google.protobuf.MapField<java.lang.Integer, org.tensorflow.proto.framework.FunctionDef.ArgAttrs>
+    internalGetMutableArgAttr() {
+      onChanged();;
+      if (argAttr_ == null) {
+        argAttr_ = com.google.protobuf.MapField.newMapField(
+            ArgAttrDefaultEntryHolder.defaultEntry);
+      }
+      if (!argAttr_.isMutable()) {
+        argAttr_ = argAttr_.copy();
+      }
+      return argAttr_;
+    }
+
+    public int getArgAttrCount() {
+      return internalGetArgAttr().getMap().size();
+    }
+    /**
+     * <code>map&lt;uint32, .tensorflow.FunctionDef.ArgAttrs&gt; arg_attr = 7;</code>
+     */
+
+    public boolean containsArgAttr(
+        int key) {
+      
+      return internalGetArgAttr().getMap().containsKey(key);
+    }
+    /**
+     * Use {@link #getArgAttrMap()} instead.
+     */
+    @java.lang.Deprecated
+    public java.util.Map<java.lang.Integer, org.tensorflow.proto.framework.FunctionDef.ArgAttrs> getArgAttr() {
+      return getArgAttrMap();
+    }
+    /**
+     * <code>map&lt;uint32, .tensorflow.FunctionDef.ArgAttrs&gt; arg_attr = 7;</code>
+     */
+
+    public java.util.Map<java.lang.Integer, org.tensorflow.proto.framework.FunctionDef.ArgAttrs> getArgAttrMap() {
+      return internalGetArgAttr().getMap();
+    }
+    /**
+     * <code>map&lt;uint32, .tensorflow.FunctionDef.ArgAttrs&gt; arg_attr = 7;</code>
+     */
+
+    public org.tensorflow.proto.framework.FunctionDef.ArgAttrs getArgAttrOrDefault(
+        int key,
+        org.tensorflow.proto.framework.FunctionDef.ArgAttrs defaultValue) {
+      
+      java.util.Map<java.lang.Integer, org.tensorflow.proto.framework.FunctionDef.ArgAttrs> map =
+          internalGetArgAttr().getMap();
+      return map.containsKey(key) ? map.get(key) : defaultValue;
+    }
+    /**
+     * <code>map&lt;uint32, .tensorflow.FunctionDef.ArgAttrs&gt; arg_attr = 7;</code>
+     */
+
+    public org.tensorflow.proto.framework.FunctionDef.ArgAttrs getArgAttrOrThrow(
+        int key) {
+      
+      java.util.Map<java.lang.Integer, org.tensorflow.proto.framework.FunctionDef.ArgAttrs> map =
+          internalGetArgAttr().getMap();
+      if (!map.containsKey(key)) {
+        throw new java.lang.IllegalArgumentException();
+      }
+      return map.get(key);
+    }
+
+    public Builder clearArgAttr() {
+      internalGetMutableArgAttr().getMutableMap()
+          .clear();
+      return this;
+    }
+    /**
+     * <code>map&lt;uint32, .tensorflow.FunctionDef.ArgAttrs&gt; arg_attr = 7;</code>
+     */
+
+    public Builder removeArgAttr(
+        int key) {
+      
+      internalGetMutableArgAttr().getMutableMap()
+          .remove(key);
+      return this;
+    }
+    /**
+     * Use alternate mutation accessors instead.
+     */
+    @java.lang.Deprecated
+    public java.util.Map<java.lang.Integer, org.tensorflow.proto.framework.FunctionDef.ArgAttrs>
+    getMutableArgAttr() {
+      return internalGetMutableArgAttr().getMutableMap();
+    }
+    /**
+     * <code>map&lt;uint32, .tensorflow.FunctionDef.ArgAttrs&gt; arg_attr = 7;</code>
+     */
+    public Builder putArgAttr(
+        int key,
+        org.tensorflow.proto.framework.FunctionDef.ArgAttrs value) {
+      
+      if (value == null) { throw new java.lang.NullPointerException(); }
+      internalGetMutableArgAttr().getMutableMap()
+          .put(key, value);
+      return this;
+    }
+    /**
+     * <code>map&lt;uint32, .tensorflow.FunctionDef.ArgAttrs&gt; arg_attr = 7;</code>
+     */
+
+    public Builder putAllArgAttr(
+        java.util.Map<java.lang.Integer, org.tensorflow.proto.framework.FunctionDef.ArgAttrs> values) {
+      internalGetMutableArgAttr().getMutableMap()
+          .putAll(values);
+      return this;
+    }
+
+    private com.google.protobuf.MapField<
+        java.lang.Integer, java.lang.Integer> resourceArgUniqueId_;
+    private com.google.protobuf.MapField<java.lang.Integer, java.lang.Integer>
+    internalGetResourceArgUniqueId() {
+      if (resourceArgUniqueId_ == null) {
+        return com.google.protobuf.MapField.emptyMapField(
+            ResourceArgUniqueIdDefaultEntryHolder.defaultEntry);
+      }
+      return resourceArgUniqueId_;
+    }
+    private com.google.protobuf.MapField<java.lang.Integer, java.lang.Integer>
+    internalGetMutableResourceArgUniqueId() {
+      onChanged();;
+      if (resourceArgUniqueId_ == null) {
+        resourceArgUniqueId_ = com.google.protobuf.MapField.newMapField(
+            ResourceArgUniqueIdDefaultEntryHolder.defaultEntry);
+      }
+      if (!resourceArgUniqueId_.isMutable()) {
+        resourceArgUniqueId_ = resourceArgUniqueId_.copy();
+      }
+      return resourceArgUniqueId_;
+    }
+
+    public int getResourceArgUniqueIdCount() {
+      return internalGetResourceArgUniqueId().getMap().size();
+    }
+    /**
+     * <pre>
+     * Unique IDs for each resource argument, used to track aliasing resources. If
+     * Argument A and Argument B alias each other, then
+     * resource_arg_unique_ids[A.index] == resource_arg_unique_ids[B.index].
+     * If this field is empty, none of the arguments could alias; otherwise, every
+     * resource argument should have an entry in this field.
+     * When instantiated, the unique IDs will be attached to the _Arg nodes'
+     * "_resource_arg_unique_id" attribute.
+     * </pre>
+     *
+     * <code>map&lt;uint32, uint32&gt; resource_arg_unique_id = 8;</code>
+     */
+
+    public boolean containsResourceArgUniqueId(
+        int key) {
+      
+      return internalGetResourceArgUniqueId().getMap().containsKey(key);
+    }
+    /**
+     * Use {@link #getResourceArgUniqueIdMap()} instead.
+     */
+    @java.lang.Deprecated
+    public java.util.Map<java.lang.Integer, java.lang.Integer> getResourceArgUniqueId() {
+      return getResourceArgUniqueIdMap();
+    }
+    /**
+     * <pre>
+     * Unique IDs for each resource argument, used to track aliasing resources. If
+     * Argument A and Argument B alias each other, then
+     * resource_arg_unique_ids[A.index] == resource_arg_unique_ids[B.index].
+     * If this field is empty, none of the arguments could alias; otherwise, every
+     * resource argument should have an entry in this field.
+     * When instantiated, the unique IDs will be attached to the _Arg nodes'
+     * "_resource_arg_unique_id" attribute.
+     * </pre>
+     *
+     * <code>map&lt;uint32, uint32&gt; resource_arg_unique_id = 8;</code>
+     */
+
+    public java.util.Map<java.lang.Integer, java.lang.Integer> getResourceArgUniqueIdMap() {
+      return internalGetResourceArgUniqueId().getMap();
+    }
+    /**
+     * <pre>
+     * Unique IDs for each resource argument, used to track aliasing resources. If
+     * Argument A and Argument B alias each other, then
+     * resource_arg_unique_ids[A.index] == resource_arg_unique_ids[B.index].
+     * If this field is empty, none of the arguments could alias; otherwise, every
+     * resource argument should have an entry in this field.
+     * When instantiated, the unique IDs will be attached to the _Arg nodes'
+     * "_resource_arg_unique_id" attribute.
+     * </pre>
+     *
+     * <code>map&lt;uint32, uint32&gt; resource_arg_unique_id = 8;</code>
+     */
+
+    public int getResourceArgUniqueIdOrDefault(
+        int key,
+        int defaultValue) {
+      
+      java.util.Map<java.lang.Integer, java.lang.Integer> map =
+          internalGetResourceArgUniqueId().getMap();
+      return map.containsKey(key) ? map.get(key) : defaultValue;
+    }
+    /**
+     * <pre>
+     * Unique IDs for each resource argument, used to track aliasing resources. If
+     * Argument A and Argument B alias each other, then
+     * resource_arg_unique_ids[A.index] == resource_arg_unique_ids[B.index].
+     * If this field is empty, none of the arguments could alias; otherwise, every
+     * resource argument should have an entry in this field.
+     * When instantiated, the unique IDs will be attached to the _Arg nodes'
+     * "_resource_arg_unique_id" attribute.
+     * </pre>
+     *
+     * <code>map&lt;uint32, uint32&gt; resource_arg_unique_id = 8;</code>
+     */
+
+    public int getResourceArgUniqueIdOrThrow(
+        int key) {
+      
+      java.util.Map<java.lang.Integer, java.lang.Integer> map =
+          internalGetResourceArgUniqueId().getMap();
+      if (!map.containsKey(key)) {
+        throw new java.lang.IllegalArgumentException();
+      }
+      return map.get(key);
+    }
+
+    public Builder clearResourceArgUniqueId() {
+      internalGetMutableResourceArgUniqueId().getMutableMap()
+          .clear();
+      return this;
+    }
+    /**
+     * <pre>
+     * Unique IDs for each resource argument, used to track aliasing resources. If
+     * Argument A and Argument B alias each other, then
+     * resource_arg_unique_ids[A.index] == resource_arg_unique_ids[B.index].
+     * If this field is empty, none of the arguments could alias; otherwise, every
+     * resource argument should have an entry in this field.
+     * When instantiated, the unique IDs will be attached to the _Arg nodes'
+     * "_resource_arg_unique_id" attribute.
+     * </pre>
+     *
+     * <code>map&lt;uint32, uint32&gt; resource_arg_unique_id = 8;</code>
+     */
+
+    public Builder removeResourceArgUniqueId(
+        int key) {
+      
+      internalGetMutableResourceArgUniqueId().getMutableMap()
+          .remove(key);
+      return this;
+    }
+    /**
+     * Use alternate mutation accessors instead.
+     */
+    @java.lang.Deprecated
+    public java.util.Map<java.lang.Integer, java.lang.Integer>
+    getMutableResourceArgUniqueId() {
+      return internalGetMutableResourceArgUniqueId().getMutableMap();
+    }
+    /**
+     * <pre>
+     * Unique IDs for each resource argument, used to track aliasing resources. If
+     * Argument A and Argument B alias each other, then
+     * resource_arg_unique_ids[A.index] == resource_arg_unique_ids[B.index].
+     * If this field is empty, none of the arguments could alias; otherwise, every
+     * resource argument should have an entry in this field.
+     * When instantiated, the unique IDs will be attached to the _Arg nodes'
+     * "_resource_arg_unique_id" attribute.
+     * </pre>
+     *
+     * <code>map&lt;uint32, uint32&gt; resource_arg_unique_id = 8;</code>
+     */
+    public Builder putResourceArgUniqueId(
+        int key,
+        int value) {
+      
+      
+      internalGetMutableResourceArgUniqueId().getMutableMap()
+          .put(key, value);
+      return this;
+    }
+    /**
+     * <pre>
+     * Unique IDs for each resource argument, used to track aliasing resources. If
+     * Argument A and Argument B alias each other, then
+     * resource_arg_unique_ids[A.index] == resource_arg_unique_ids[B.index].
+     * If this field is empty, none of the arguments could alias; otherwise, every
+     * resource argument should have an entry in this field.
+     * When instantiated, the unique IDs will be attached to the _Arg nodes'
+     * "_resource_arg_unique_id" attribute.
+     * </pre>
+     *
+     * <code>map&lt;uint32, uint32&gt; resource_arg_unique_id = 8;</code>
+     */
+
+    public Builder putAllResourceArgUniqueId(
+        java.util.Map<java.lang.Integer, java.lang.Integer> values) {
+      internalGetMutableResourceArgUniqueId().getMutableMap()
+          .putAll(values);
+      return this;
+    }
+
+    private java.util.List<org.tensorflow.proto.framework.NodeDef> nodeDef_ =
+      java.util.Collections.emptyList();
+    private void ensureNodeDefIsMutable() {
+      if (!((bitField0_ & 0x00000008) != 0)) {
+        nodeDef_ = new java.util.ArrayList<org.tensorflow.proto.framework.NodeDef>(nodeDef_);
+        bitField0_ |= 0x00000008;
+       }
+    }
+
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.framework.NodeDef, org.tensorflow.proto.framework.NodeDef.Builder, org.tensorflow.proto.framework.NodeDefOrBuilder> nodeDefBuilder_;
+
+    /**
+     * <pre>
+     * By convention, "op" in node_def is resolved by consulting with a
+     * user-defined library first. If not resolved, "func" is assumed to
+     * be a builtin op.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.NodeDef node_def = 3;</code>
+     */
+    public java.util.List<org.tensorflow.proto.framework.NodeDef> getNodeDefList() {
+      if (nodeDefBuilder_ == null) {
+        return java.util.Collections.unmodifiableList(nodeDef_);
+      } else {
+        return nodeDefBuilder_.getMessageList();
+      }
+    }
+    /**
+     * <pre>
+     * By convention, "op" in node_def is resolved by consulting with a
+     * user-defined library first. If not resolved, "func" is assumed to
+     * be a builtin op.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.NodeDef node_def = 3;</code>
+     */
+    public int getNodeDefCount() {
+      if (nodeDefBuilder_ == null) {
+        return nodeDef_.size();
+      } else {
+        return nodeDefBuilder_.getCount();
+      }
+    }
+    /**
+     * <pre>
+     * By convention, "op" in node_def is resolved by consulting with a
+     * user-defined library first. If not resolved, "func" is assumed to
+     * be a builtin op.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.NodeDef node_def = 3;</code>
+     */
+    public org.tensorflow.proto.framework.NodeDef getNodeDef(int index) {
+      if (nodeDefBuilder_ == null) {
+        return nodeDef_.get(index);
+      } else {
+        return nodeDefBuilder_.getMessage(index);
+      }
+    }
+    /**
+     * <pre>
+     * By convention, "op" in node_def is resolved by consulting with a
+     * user-defined library first. If not resolved, "func" is assumed to
+     * be a builtin op.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.NodeDef node_def = 3;</code>
+     */
+    public Builder setNodeDef(
+        int index, org.tensorflow.proto.framework.NodeDef value) {
+      if (nodeDefBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureNodeDefIsMutable();
+        nodeDef_.set(index, value);
+        onChanged();
+      } else {
+        nodeDefBuilder_.setMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * By convention, "op" in node_def is resolved by consulting with a
+     * user-defined library first. If not resolved, "func" is assumed to
+     * be a builtin op.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.NodeDef node_def = 3;</code>
+     */
+    public Builder setNodeDef(
+        int index, org.tensorflow.proto.framework.NodeDef.Builder builderForValue) {
+      if (nodeDefBuilder_ == null) {
+        ensureNodeDefIsMutable();
+        nodeDef_.set(index, builderForValue.build());
+        onChanged();
+      } else {
+        nodeDefBuilder_.setMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * By convention, "op" in node_def is resolved by consulting with a
+     * user-defined library first. If not resolved, "func" is assumed to
+     * be a builtin op.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.NodeDef node_def = 3;</code>
+     */
+    public Builder addNodeDef(org.tensorflow.proto.framework.NodeDef value) {
+      if (nodeDefBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureNodeDefIsMutable();
+        nodeDef_.add(value);
+        onChanged();
+      } else {
+        nodeDefBuilder_.addMessage(value);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * By convention, "op" in node_def is resolved by consulting with a
+     * user-defined library first. If not resolved, "func" is assumed to
+     * be a builtin op.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.NodeDef node_def = 3;</code>
+     */
+    public Builder addNodeDef(
+        int index, org.tensorflow.proto.framework.NodeDef value) {
+      if (nodeDefBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureNodeDefIsMutable();
+        nodeDef_.add(index, value);
+        onChanged();
+      } else {
+        nodeDefBuilder_.addMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * By convention, "op" in node_def is resolved by consulting with a
+     * user-defined library first. If not resolved, "func" is assumed to
+     * be a builtin op.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.NodeDef node_def = 3;</code>
+     */
+    public Builder addNodeDef(
+        org.tensorflow.proto.framework.NodeDef.Builder builderForValue) {
+      if (nodeDefBuilder_ == null) {
+        ensureNodeDefIsMutable();
+        nodeDef_.add(builderForValue.build());
+        onChanged();
+      } else {
+        nodeDefBuilder_.addMessage(builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * By convention, "op" in node_def is resolved by consulting with a
+     * user-defined library first. If not resolved, "func" is assumed to
+     * be a builtin op.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.NodeDef node_def = 3;</code>
+     */
+    public Builder addNodeDef(
+        int index, org.tensorflow.proto.framework.NodeDef.Builder builderForValue) {
+      if (nodeDefBuilder_ == null) {
+        ensureNodeDefIsMutable();
+        nodeDef_.add(index, builderForValue.build());
+        onChanged();
+      } else {
+        nodeDefBuilder_.addMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * By convention, "op" in node_def is resolved by consulting with a
+     * user-defined library first. If not resolved, "func" is assumed to
+     * be a builtin op.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.NodeDef node_def = 3;</code>
+     */
+    public Builder addAllNodeDef(
+        java.lang.Iterable<? extends org.tensorflow.proto.framework.NodeDef> values) {
+      if (nodeDefBuilder_ == null) {
+        ensureNodeDefIsMutable();
+        com.google.protobuf.AbstractMessageLite.Builder.addAll(
+            values, nodeDef_);
+        onChanged();
+      } else {
+        nodeDefBuilder_.addAllMessages(values);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * By convention, "op" in node_def is resolved by consulting with a
+     * user-defined library first. If not resolved, "func" is assumed to
+     * be a builtin op.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.NodeDef node_def = 3;</code>
+     */
+    public Builder clearNodeDef() {
+      if (nodeDefBuilder_ == null) {
+        nodeDef_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000008);
+        onChanged();
+      } else {
+        nodeDefBuilder_.clear();
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * By convention, "op" in node_def is resolved by consulting with a
+     * user-defined library first. If not resolved, "func" is assumed to
+     * be a builtin op.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.NodeDef node_def = 3;</code>
+     */
+    public Builder removeNodeDef(int index) {
+      if (nodeDefBuilder_ == null) {
+        ensureNodeDefIsMutable();
+        nodeDef_.remove(index);
+        onChanged();
+      } else {
+        nodeDefBuilder_.remove(index);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * By convention, "op" in node_def is resolved by consulting with a
+     * user-defined library first. If not resolved, "func" is assumed to
+     * be a builtin op.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.NodeDef node_def = 3;</code>
+     */
+    public org.tensorflow.proto.framework.NodeDef.Builder getNodeDefBuilder(
+        int index) {
+      return getNodeDefFieldBuilder().getBuilder(index);
+    }
+    /**
+     * <pre>
+     * By convention, "op" in node_def is resolved by consulting with a
+     * user-defined library first. If not resolved, "func" is assumed to
+     * be a builtin op.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.NodeDef node_def = 3;</code>
+     */
+    public org.tensorflow.proto.framework.NodeDefOrBuilder getNodeDefOrBuilder(
+        int index) {
+      if (nodeDefBuilder_ == null) {
+        return nodeDef_.get(index);  } else {
+        return nodeDefBuilder_.getMessageOrBuilder(index);
+      }
+    }
+    /**
+     * <pre>
+     * By convention, "op" in node_def is resolved by consulting with a
+     * user-defined library first. If not resolved, "func" is assumed to
+     * be a builtin op.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.NodeDef node_def = 3;</code>
+     */
+    public java.util.List<? extends org.tensorflow.proto.framework.NodeDefOrBuilder> 
+         getNodeDefOrBuilderList() {
+      if (nodeDefBuilder_ != null) {
+        return nodeDefBuilder_.getMessageOrBuilderList();
+      } else {
+        return java.util.Collections.unmodifiableList(nodeDef_);
+      }
+    }
+    /**
+     * <pre>
+     * By convention, "op" in node_def is resolved by consulting with a
+     * user-defined library first. If not resolved, "func" is assumed to
+     * be a builtin op.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.NodeDef node_def = 3;</code>
+     */
+    public org.tensorflow.proto.framework.NodeDef.Builder addNodeDefBuilder() {
+      return getNodeDefFieldBuilder().addBuilder(
+          org.tensorflow.proto.framework.NodeDef.getDefaultInstance());
+    }
+    /**
+     * <pre>
+     * By convention, "op" in node_def is resolved by consulting with a
+     * user-defined library first. If not resolved, "func" is assumed to
+     * be a builtin op.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.NodeDef node_def = 3;</code>
+     */
+    public org.tensorflow.proto.framework.NodeDef.Builder addNodeDefBuilder(
+        int index) {
+      return getNodeDefFieldBuilder().addBuilder(
+          index, org.tensorflow.proto.framework.NodeDef.getDefaultInstance());
+    }
+    /**
+     * <pre>
+     * By convention, "op" in node_def is resolved by consulting with a
+     * user-defined library first. If not resolved, "func" is assumed to
+     * be a builtin op.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.NodeDef node_def = 3;</code>
+     */
+    public java.util.List<org.tensorflow.proto.framework.NodeDef.Builder> 
+         getNodeDefBuilderList() {
+      return getNodeDefFieldBuilder().getBuilderList();
+    }
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.framework.NodeDef, org.tensorflow.proto.framework.NodeDef.Builder, org.tensorflow.proto.framework.NodeDefOrBuilder> 
+        getNodeDefFieldBuilder() {
+      if (nodeDefBuilder_ == null) {
+        nodeDefBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
+            org.tensorflow.proto.framework.NodeDef, org.tensorflow.proto.framework.NodeDef.Builder, org.tensorflow.proto.framework.NodeDefOrBuilder>(
+                nodeDef_,
+                ((bitField0_ & 0x00000008) != 0),
+                getParentForChildren(),
+                isClean());
+        nodeDef_ = null;
+      }
+      return nodeDefBuilder_;
+    }
+
+    private com.google.protobuf.MapField<
+        java.lang.String, java.lang.String> ret_;
+    private com.google.protobuf.MapField<java.lang.String, java.lang.String>
+    internalGetRet() {
+      if (ret_ == null) {
+        return com.google.protobuf.MapField.emptyMapField(
+            RetDefaultEntryHolder.defaultEntry);
+      }
+      return ret_;
+    }
+    private com.google.protobuf.MapField<java.lang.String, java.lang.String>
+    internalGetMutableRet() {
+      onChanged();;
+      if (ret_ == null) {
+        ret_ = com.google.protobuf.MapField.newMapField(
+            RetDefaultEntryHolder.defaultEntry);
+      }
+      if (!ret_.isMutable()) {
+        ret_ = ret_.copy();
+      }
+      return ret_;
+    }
+
+    public int getRetCount() {
+      return internalGetRet().getMap().size();
+    }
+    /**
+     * <pre>
+     * A mapping from the output arg names from `signature` to the
+     * outputs from `node_def` that should be returned by the function.
+     * </pre>
+     *
+     * <code>map&lt;string, string&gt; ret = 4;</code>
+     */
+
+    public boolean containsRet(
+        java.lang.String key) {
+      if (key == null) { throw new java.lang.NullPointerException(); }
+      return internalGetRet().getMap().containsKey(key);
+    }
+    /**
+     * Use {@link #getRetMap()} instead.
+     */
+    @java.lang.Deprecated
+    public java.util.Map<java.lang.String, java.lang.String> getRet() {
+      return getRetMap();
+    }
+    /**
+     * <pre>
+     * A mapping from the output arg names from `signature` to the
+     * outputs from `node_def` that should be returned by the function.
+     * </pre>
+     *
+     * <code>map&lt;string, string&gt; ret = 4;</code>
+     */
+
+    public java.util.Map<java.lang.String, java.lang.String> getRetMap() {
+      return internalGetRet().getMap();
+    }
+    /**
+     * <pre>
+     * A mapping from the output arg names from `signature` to the
+     * outputs from `node_def` that should be returned by the function.
+     * </pre>
+     *
+     * <code>map&lt;string, string&gt; ret = 4;</code>
+     */
+
+    public java.lang.String getRetOrDefault(
+        java.lang.String key,
+        java.lang.String defaultValue) {
+      if (key == null) { throw new java.lang.NullPointerException(); }
+      java.util.Map<java.lang.String, java.lang.String> map =
+          internalGetRet().getMap();
+      return map.containsKey(key) ? map.get(key) : defaultValue;
+    }
+    /**
+     * <pre>
+     * A mapping from the output arg names from `signature` to the
+     * outputs from `node_def` that should be returned by the function.
+     * </pre>
+     *
+     * <code>map&lt;string, string&gt; ret = 4;</code>
+     */
+
+    public java.lang.String getRetOrThrow(
+        java.lang.String key) {
+      if (key == null) { throw new java.lang.NullPointerException(); }
+      java.util.Map<java.lang.String, java.lang.String> map =
+          internalGetRet().getMap();
+      if (!map.containsKey(key)) {
+        throw new java.lang.IllegalArgumentException();
+      }
+      return map.get(key);
+    }
+
+    public Builder clearRet() {
+      internalGetMutableRet().getMutableMap()
+          .clear();
+      return this;
+    }
+    /**
+     * <pre>
+     * A mapping from the output arg names from `signature` to the
+     * outputs from `node_def` that should be returned by the function.
+     * </pre>
+     *
+     * <code>map&lt;string, string&gt; ret = 4;</code>
+     */
+
+    public Builder removeRet(
+        java.lang.String key) {
+      if (key == null) { throw new java.lang.NullPointerException(); }
+      internalGetMutableRet().getMutableMap()
+          .remove(key);
+      return this;
+    }
+    /**
+     * Use alternate mutation accessors instead.
+     */
+    @java.lang.Deprecated
+    public java.util.Map<java.lang.String, java.lang.String>
+    getMutableRet() {
+      return internalGetMutableRet().getMutableMap();
+    }
+    /**
+     * <pre>
+     * A mapping from the output arg names from `signature` to the
+     * outputs from `node_def` that should be returned by the function.
+     * </pre>
+     *
+     * <code>map&lt;string, string&gt; ret = 4;</code>
+     */
+    public Builder putRet(
+        java.lang.String key,
+        java.lang.String value) {
+      if (key == null) { throw new java.lang.NullPointerException(); }
+      if (value == null) { throw new java.lang.NullPointerException(); }
+      internalGetMutableRet().getMutableMap()
+          .put(key, value);
+      return this;
+    }
+    /**
+     * <pre>
+     * A mapping from the output arg names from `signature` to the
+     * outputs from `node_def` that should be returned by the function.
+     * </pre>
+     *
+     * <code>map&lt;string, string&gt; ret = 4;</code>
+     */
+
+    public Builder putAllRet(
+        java.util.Map<java.lang.String, java.lang.String> values) {
+      internalGetMutableRet().getMutableMap()
+          .putAll(values);
+      return this;
+    }
+
+    private com.google.protobuf.MapField<
+        java.lang.String, java.lang.String> controlRet_;
+    private com.google.protobuf.MapField<java.lang.String, java.lang.String>
+    internalGetControlRet() {
+      if (controlRet_ == null) {
+        return com.google.protobuf.MapField.emptyMapField(
+            ControlRetDefaultEntryHolder.defaultEntry);
+      }
+      return controlRet_;
+    }
+    private com.google.protobuf.MapField<java.lang.String, java.lang.String>
+    internalGetMutableControlRet() {
+      onChanged();;
+      if (controlRet_ == null) {
+        controlRet_ = com.google.protobuf.MapField.newMapField(
+            ControlRetDefaultEntryHolder.defaultEntry);
+      }
+      if (!controlRet_.isMutable()) {
+        controlRet_ = controlRet_.copy();
+      }
+      return controlRet_;
+    }
+
+    public int getControlRetCount() {
+      return internalGetControlRet().getMap().size();
+    }
+    /**
+     * <pre>
+     * A mapping from control output names from `signature` to node names in
+     * `node_def` which should be control outputs of this function.
+     * </pre>
+     *
+     * <code>map&lt;string, string&gt; control_ret = 6;</code>
+     */
+
+    public boolean containsControlRet(
+        java.lang.String key) {
+      if (key == null) { throw new java.lang.NullPointerException(); }
+      return internalGetControlRet().getMap().containsKey(key);
+    }
+    /**
+     * Use {@link #getControlRetMap()} instead.
+     */
+    @java.lang.Deprecated
+    public java.util.Map<java.lang.String, java.lang.String> getControlRet() {
+      return getControlRetMap();
+    }
+    /**
+     * <pre>
+     * A mapping from control output names from `signature` to node names in
+     * `node_def` which should be control outputs of this function.
+     * </pre>
+     *
+     * <code>map&lt;string, string&gt; control_ret = 6;</code>
+     */
+
+    public java.util.Map<java.lang.String, java.lang.String> getControlRetMap() {
+      return internalGetControlRet().getMap();
+    }
+    /**
+     * <pre>
+     * A mapping from control output names from `signature` to node names in
+     * `node_def` which should be control outputs of this function.
+     * </pre>
+     *
+     * <code>map&lt;string, string&gt; control_ret = 6;</code>
+     */
+
+    public java.lang.String getControlRetOrDefault(
+        java.lang.String key,
+        java.lang.String defaultValue) {
+      if (key == null) { throw new java.lang.NullPointerException(); }
+      java.util.Map<java.lang.String, java.lang.String> map =
+          internalGetControlRet().getMap();
+      return map.containsKey(key) ? map.get(key) : defaultValue;
+    }
+    /**
+     * <pre>
+     * A mapping from control output names from `signature` to node names in
+     * `node_def` which should be control outputs of this function.
+     * </pre>
+     *
+     * <code>map&lt;string, string&gt; control_ret = 6;</code>
+     */
+
+    public java.lang.String getControlRetOrThrow(
+        java.lang.String key) {
+      if (key == null) { throw new java.lang.NullPointerException(); }
+      java.util.Map<java.lang.String, java.lang.String> map =
+          internalGetControlRet().getMap();
+      if (!map.containsKey(key)) {
+        throw new java.lang.IllegalArgumentException();
+      }
+      return map.get(key);
+    }
+
+    public Builder clearControlRet() {
+      internalGetMutableControlRet().getMutableMap()
+          .clear();
+      return this;
+    }
+    /**
+     * <pre>
+     * A mapping from control output names from `signature` to node names in
+     * `node_def` which should be control outputs of this function.
+     * </pre>
+     *
+     * <code>map&lt;string, string&gt; control_ret = 6;</code>
+     */
+
+    public Builder removeControlRet(
+        java.lang.String key) {
+      if (key == null) { throw new java.lang.NullPointerException(); }
+      internalGetMutableControlRet().getMutableMap()
+          .remove(key);
+      return this;
+    }
+    /**
+     * Use alternate mutation accessors instead.
+     */
+    @java.lang.Deprecated
+    public java.util.Map<java.lang.String, java.lang.String>
+    getMutableControlRet() {
+      return internalGetMutableControlRet().getMutableMap();
+    }
+    /**
+     * <pre>
+     * A mapping from control output names from `signature` to node names in
+     * `node_def` which should be control outputs of this function.
+     * </pre>
+     *
+     * <code>map&lt;string, string&gt; control_ret = 6;</code>
+     */
+    public Builder putControlRet(
+        java.lang.String key,
+        java.lang.String value) {
+      if (key == null) { throw new java.lang.NullPointerException(); }
+      if (value == null) { throw new java.lang.NullPointerException(); }
+      internalGetMutableControlRet().getMutableMap()
+          .put(key, value);
+      return this;
+    }
+    /**
+     * <pre>
+     * A mapping from control output names from `signature` to node names in
+     * `node_def` which should be control outputs of this function.
+     * </pre>
+     *
+     * <code>map&lt;string, string&gt; control_ret = 6;</code>
+     */
+
+    public Builder putAllControlRet(
+        java.util.Map<java.lang.String, java.lang.String> values) {
+      internalGetMutableControlRet().getMutableMap()
+          .putAll(values);
+      return this;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.FunctionDef)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.FunctionDef)
+  private static final org.tensorflow.proto.framework.FunctionDef DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.FunctionDef();
+  }
+
+  public static org.tensorflow.proto.framework.FunctionDef getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<FunctionDef>
+      PARSER = new com.google.protobuf.AbstractParser<FunctionDef>() {
+    @java.lang.Override
+    public FunctionDef parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new FunctionDef(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<FunctionDef> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<FunctionDef> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.framework.FunctionDef getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/FunctionDefLibrary.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/FunctionDefLibrary.java
new file mode 100644
index 00000000000..a4ad715f17c
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/FunctionDefLibrary.java
@@ -0,0 +1,1116 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/framework/function.proto
+
+package org.tensorflow.proto.framework;
+
+/**
+ * <pre>
+ * A library is a set of named functions.
+ * </pre>
+ *
+ * Protobuf type {@code tensorflow.FunctionDefLibrary}
+ */
+public  final class FunctionDefLibrary extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.FunctionDefLibrary)
+    FunctionDefLibraryOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use FunctionDefLibrary.newBuilder() to construct.
+  private FunctionDefLibrary(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private FunctionDefLibrary() {
+    function_ = java.util.Collections.emptyList();
+    gradient_ = java.util.Collections.emptyList();
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new FunctionDefLibrary();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private FunctionDefLibrary(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    int mutable_bitField0_ = 0;
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 10: {
+            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
+              function_ = new java.util.ArrayList<org.tensorflow.proto.framework.FunctionDef>();
+              mutable_bitField0_ |= 0x00000001;
+            }
+            function_.add(
+                input.readMessage(org.tensorflow.proto.framework.FunctionDef.parser(), extensionRegistry));
+            break;
+          }
+          case 18: {
+            if (!((mutable_bitField0_ & 0x00000002) != 0)) {
+              gradient_ = new java.util.ArrayList<org.tensorflow.proto.framework.GradientDef>();
+              mutable_bitField0_ |= 0x00000002;
+            }
+            gradient_.add(
+                input.readMessage(org.tensorflow.proto.framework.GradientDef.parser(), extensionRegistry));
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      if (((mutable_bitField0_ & 0x00000001) != 0)) {
+        function_ = java.util.Collections.unmodifiableList(function_);
+      }
+      if (((mutable_bitField0_ & 0x00000002) != 0)) {
+        gradient_ = java.util.Collections.unmodifiableList(gradient_);
+      }
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.framework.FunctionProtos.internal_static_tensorflow_FunctionDefLibrary_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.framework.FunctionProtos.internal_static_tensorflow_FunctionDefLibrary_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.framework.FunctionDefLibrary.class, org.tensorflow.proto.framework.FunctionDefLibrary.Builder.class);
+  }
+
+  public static final int FUNCTION_FIELD_NUMBER = 1;
+  private java.util.List<org.tensorflow.proto.framework.FunctionDef> function_;
+  /**
+   * <code>repeated .tensorflow.FunctionDef function = 1;</code>
+   */
+  public java.util.List<org.tensorflow.proto.framework.FunctionDef> getFunctionList() {
+    return function_;
+  }
+  /**
+   * <code>repeated .tensorflow.FunctionDef function = 1;</code>
+   */
+  public java.util.List<? extends org.tensorflow.proto.framework.FunctionDefOrBuilder> 
+      getFunctionOrBuilderList() {
+    return function_;
+  }
+  /**
+   * <code>repeated .tensorflow.FunctionDef function = 1;</code>
+   */
+  public int getFunctionCount() {
+    return function_.size();
+  }
+  /**
+   * <code>repeated .tensorflow.FunctionDef function = 1;</code>
+   */
+  public org.tensorflow.proto.framework.FunctionDef getFunction(int index) {
+    return function_.get(index);
+  }
+  /**
+   * <code>repeated .tensorflow.FunctionDef function = 1;</code>
+   */
+  public org.tensorflow.proto.framework.FunctionDefOrBuilder getFunctionOrBuilder(
+      int index) {
+    return function_.get(index);
+  }
+
+  public static final int GRADIENT_FIELD_NUMBER = 2;
+  private java.util.List<org.tensorflow.proto.framework.GradientDef> gradient_;
+  /**
+   * <code>repeated .tensorflow.GradientDef gradient = 2;</code>
+   */
+  public java.util.List<org.tensorflow.proto.framework.GradientDef> getGradientList() {
+    return gradient_;
+  }
+  /**
+   * <code>repeated .tensorflow.GradientDef gradient = 2;</code>
+   */
+  public java.util.List<? extends org.tensorflow.proto.framework.GradientDefOrBuilder> 
+      getGradientOrBuilderList() {
+    return gradient_;
+  }
+  /**
+   * <code>repeated .tensorflow.GradientDef gradient = 2;</code>
+   */
+  public int getGradientCount() {
+    return gradient_.size();
+  }
+  /**
+   * <code>repeated .tensorflow.GradientDef gradient = 2;</code>
+   */
+  public org.tensorflow.proto.framework.GradientDef getGradient(int index) {
+    return gradient_.get(index);
+  }
+  /**
+   * <code>repeated .tensorflow.GradientDef gradient = 2;</code>
+   */
+  public org.tensorflow.proto.framework.GradientDefOrBuilder getGradientOrBuilder(
+      int index) {
+    return gradient_.get(index);
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    for (int i = 0; i < function_.size(); i++) {
+      output.writeMessage(1, function_.get(i));
+    }
+    for (int i = 0; i < gradient_.size(); i++) {
+      output.writeMessage(2, gradient_.get(i));
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    for (int i = 0; i < function_.size(); i++) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(1, function_.get(i));
+    }
+    for (int i = 0; i < gradient_.size(); i++) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(2, gradient_.get(i));
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.framework.FunctionDefLibrary)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.framework.FunctionDefLibrary other = (org.tensorflow.proto.framework.FunctionDefLibrary) obj;
+
+    if (!getFunctionList()
+        .equals(other.getFunctionList())) return false;
+    if (!getGradientList()
+        .equals(other.getGradientList())) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    if (getFunctionCount() > 0) {
+      hash = (37 * hash) + FUNCTION_FIELD_NUMBER;
+      hash = (53 * hash) + getFunctionList().hashCode();
+    }
+    if (getGradientCount() > 0) {
+      hash = (37 * hash) + GRADIENT_FIELD_NUMBER;
+      hash = (53 * hash) + getGradientList().hashCode();
+    }
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.framework.FunctionDefLibrary parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.FunctionDefLibrary parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.FunctionDefLibrary parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.FunctionDefLibrary parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.FunctionDefLibrary parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.FunctionDefLibrary parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.FunctionDefLibrary parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.FunctionDefLibrary parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.FunctionDefLibrary parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.FunctionDefLibrary parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.FunctionDefLibrary parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.FunctionDefLibrary parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.framework.FunctionDefLibrary prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * <pre>
+   * A library is a set of named functions.
+   * </pre>
+   *
+   * Protobuf type {@code tensorflow.FunctionDefLibrary}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.FunctionDefLibrary)
+      org.tensorflow.proto.framework.FunctionDefLibraryOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.FunctionProtos.internal_static_tensorflow_FunctionDefLibrary_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.FunctionProtos.internal_static_tensorflow_FunctionDefLibrary_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.FunctionDefLibrary.class, org.tensorflow.proto.framework.FunctionDefLibrary.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.framework.FunctionDefLibrary.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+        getFunctionFieldBuilder();
+        getGradientFieldBuilder();
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      if (functionBuilder_ == null) {
+        function_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000001);
+      } else {
+        functionBuilder_.clear();
+      }
+      if (gradientBuilder_ == null) {
+        gradient_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000002);
+      } else {
+        gradientBuilder_.clear();
+      }
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.framework.FunctionProtos.internal_static_tensorflow_FunctionDefLibrary_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.FunctionDefLibrary getDefaultInstanceForType() {
+      return org.tensorflow.proto.framework.FunctionDefLibrary.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.FunctionDefLibrary build() {
+      org.tensorflow.proto.framework.FunctionDefLibrary result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.FunctionDefLibrary buildPartial() {
+      org.tensorflow.proto.framework.FunctionDefLibrary result = new org.tensorflow.proto.framework.FunctionDefLibrary(this);
+      int from_bitField0_ = bitField0_;
+      if (functionBuilder_ == null) {
+        if (((bitField0_ & 0x00000001) != 0)) {
+          function_ = java.util.Collections.unmodifiableList(function_);
+          bitField0_ = (bitField0_ & ~0x00000001);
+        }
+        result.function_ = function_;
+      } else {
+        result.function_ = functionBuilder_.build();
+      }
+      if (gradientBuilder_ == null) {
+        if (((bitField0_ & 0x00000002) != 0)) {
+          gradient_ = java.util.Collections.unmodifiableList(gradient_);
+          bitField0_ = (bitField0_ & ~0x00000002);
+        }
+        result.gradient_ = gradient_;
+      } else {
+        result.gradient_ = gradientBuilder_.build();
+      }
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.framework.FunctionDefLibrary) {
+        return mergeFrom((org.tensorflow.proto.framework.FunctionDefLibrary)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.framework.FunctionDefLibrary other) {
+      if (other == org.tensorflow.proto.framework.FunctionDefLibrary.getDefaultInstance()) return this;
+      if (functionBuilder_ == null) {
+        if (!other.function_.isEmpty()) {
+          if (function_.isEmpty()) {
+            function_ = other.function_;
+            bitField0_ = (bitField0_ & ~0x00000001);
+          } else {
+            ensureFunctionIsMutable();
+            function_.addAll(other.function_);
+          }
+          onChanged();
+        }
+      } else {
+        if (!other.function_.isEmpty()) {
+          if (functionBuilder_.isEmpty()) {
+            functionBuilder_.dispose();
+            functionBuilder_ = null;
+            function_ = other.function_;
+            bitField0_ = (bitField0_ & ~0x00000001);
+            functionBuilder_ = 
+              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
+                 getFunctionFieldBuilder() : null;
+          } else {
+            functionBuilder_.addAllMessages(other.function_);
+          }
+        }
+      }
+      if (gradientBuilder_ == null) {
+        if (!other.gradient_.isEmpty()) {
+          if (gradient_.isEmpty()) {
+            gradient_ = other.gradient_;
+            bitField0_ = (bitField0_ & ~0x00000002);
+          } else {
+            ensureGradientIsMutable();
+            gradient_.addAll(other.gradient_);
+          }
+          onChanged();
+        }
+      } else {
+        if (!other.gradient_.isEmpty()) {
+          if (gradientBuilder_.isEmpty()) {
+            gradientBuilder_.dispose();
+            gradientBuilder_ = null;
+            gradient_ = other.gradient_;
+            bitField0_ = (bitField0_ & ~0x00000002);
+            gradientBuilder_ = 
+              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
+                 getGradientFieldBuilder() : null;
+          } else {
+            gradientBuilder_.addAllMessages(other.gradient_);
+          }
+        }
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.framework.FunctionDefLibrary parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.framework.FunctionDefLibrary) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+    private int bitField0_;
+
+    private java.util.List<org.tensorflow.proto.framework.FunctionDef> function_ =
+      java.util.Collections.emptyList();
+    private void ensureFunctionIsMutable() {
+      if (!((bitField0_ & 0x00000001) != 0)) {
+        function_ = new java.util.ArrayList<org.tensorflow.proto.framework.FunctionDef>(function_);
+        bitField0_ |= 0x00000001;
+       }
+    }
+
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.framework.FunctionDef, org.tensorflow.proto.framework.FunctionDef.Builder, org.tensorflow.proto.framework.FunctionDefOrBuilder> functionBuilder_;
+
+    /**
+     * <code>repeated .tensorflow.FunctionDef function = 1;</code>
+     */
+    public java.util.List<org.tensorflow.proto.framework.FunctionDef> getFunctionList() {
+      if (functionBuilder_ == null) {
+        return java.util.Collections.unmodifiableList(function_);
+      } else {
+        return functionBuilder_.getMessageList();
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.FunctionDef function = 1;</code>
+     */
+    public int getFunctionCount() {
+      if (functionBuilder_ == null) {
+        return function_.size();
+      } else {
+        return functionBuilder_.getCount();
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.FunctionDef function = 1;</code>
+     */
+    public org.tensorflow.proto.framework.FunctionDef getFunction(int index) {
+      if (functionBuilder_ == null) {
+        return function_.get(index);
+      } else {
+        return functionBuilder_.getMessage(index);
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.FunctionDef function = 1;</code>
+     */
+    public Builder setFunction(
+        int index, org.tensorflow.proto.framework.FunctionDef value) {
+      if (functionBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureFunctionIsMutable();
+        function_.set(index, value);
+        onChanged();
+      } else {
+        functionBuilder_.setMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.FunctionDef function = 1;</code>
+     */
+    public Builder setFunction(
+        int index, org.tensorflow.proto.framework.FunctionDef.Builder builderForValue) {
+      if (functionBuilder_ == null) {
+        ensureFunctionIsMutable();
+        function_.set(index, builderForValue.build());
+        onChanged();
+      } else {
+        functionBuilder_.setMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.FunctionDef function = 1;</code>
+     */
+    public Builder addFunction(org.tensorflow.proto.framework.FunctionDef value) {
+      if (functionBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureFunctionIsMutable();
+        function_.add(value);
+        onChanged();
+      } else {
+        functionBuilder_.addMessage(value);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.FunctionDef function = 1;</code>
+     */
+    public Builder addFunction(
+        int index, org.tensorflow.proto.framework.FunctionDef value) {
+      if (functionBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureFunctionIsMutable();
+        function_.add(index, value);
+        onChanged();
+      } else {
+        functionBuilder_.addMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.FunctionDef function = 1;</code>
+     */
+    public Builder addFunction(
+        org.tensorflow.proto.framework.FunctionDef.Builder builderForValue) {
+      if (functionBuilder_ == null) {
+        ensureFunctionIsMutable();
+        function_.add(builderForValue.build());
+        onChanged();
+      } else {
+        functionBuilder_.addMessage(builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.FunctionDef function = 1;</code>
+     */
+    public Builder addFunction(
+        int index, org.tensorflow.proto.framework.FunctionDef.Builder builderForValue) {
+      if (functionBuilder_ == null) {
+        ensureFunctionIsMutable();
+        function_.add(index, builderForValue.build());
+        onChanged();
+      } else {
+        functionBuilder_.addMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.FunctionDef function = 1;</code>
+     */
+    public Builder addAllFunction(
+        java.lang.Iterable<? extends org.tensorflow.proto.framework.FunctionDef> values) {
+      if (functionBuilder_ == null) {
+        ensureFunctionIsMutable();
+        com.google.protobuf.AbstractMessageLite.Builder.addAll(
+            values, function_);
+        onChanged();
+      } else {
+        functionBuilder_.addAllMessages(values);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.FunctionDef function = 1;</code>
+     */
+    public Builder clearFunction() {
+      if (functionBuilder_ == null) {
+        function_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000001);
+        onChanged();
+      } else {
+        functionBuilder_.clear();
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.FunctionDef function = 1;</code>
+     */
+    public Builder removeFunction(int index) {
+      if (functionBuilder_ == null) {
+        ensureFunctionIsMutable();
+        function_.remove(index);
+        onChanged();
+      } else {
+        functionBuilder_.remove(index);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.FunctionDef function = 1;</code>
+     */
+    public org.tensorflow.proto.framework.FunctionDef.Builder getFunctionBuilder(
+        int index) {
+      return getFunctionFieldBuilder().getBuilder(index);
+    }
+    /**
+     * <code>repeated .tensorflow.FunctionDef function = 1;</code>
+     */
+    public org.tensorflow.proto.framework.FunctionDefOrBuilder getFunctionOrBuilder(
+        int index) {
+      if (functionBuilder_ == null) {
+        return function_.get(index);  } else {
+        return functionBuilder_.getMessageOrBuilder(index);
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.FunctionDef function = 1;</code>
+     */
+    public java.util.List<? extends org.tensorflow.proto.framework.FunctionDefOrBuilder> 
+         getFunctionOrBuilderList() {
+      if (functionBuilder_ != null) {
+        return functionBuilder_.getMessageOrBuilderList();
+      } else {
+        return java.util.Collections.unmodifiableList(function_);
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.FunctionDef function = 1;</code>
+     */
+    public org.tensorflow.proto.framework.FunctionDef.Builder addFunctionBuilder() {
+      return getFunctionFieldBuilder().addBuilder(
+          org.tensorflow.proto.framework.FunctionDef.getDefaultInstance());
+    }
+    /**
+     * <code>repeated .tensorflow.FunctionDef function = 1;</code>
+     */
+    public org.tensorflow.proto.framework.FunctionDef.Builder addFunctionBuilder(
+        int index) {
+      return getFunctionFieldBuilder().addBuilder(
+          index, org.tensorflow.proto.framework.FunctionDef.getDefaultInstance());
+    }
+    /**
+     * <code>repeated .tensorflow.FunctionDef function = 1;</code>
+     */
+    public java.util.List<org.tensorflow.proto.framework.FunctionDef.Builder> 
+         getFunctionBuilderList() {
+      return getFunctionFieldBuilder().getBuilderList();
+    }
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.framework.FunctionDef, org.tensorflow.proto.framework.FunctionDef.Builder, org.tensorflow.proto.framework.FunctionDefOrBuilder> 
+        getFunctionFieldBuilder() {
+      if (functionBuilder_ == null) {
+        functionBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
+            org.tensorflow.proto.framework.FunctionDef, org.tensorflow.proto.framework.FunctionDef.Builder, org.tensorflow.proto.framework.FunctionDefOrBuilder>(
+                function_,
+                ((bitField0_ & 0x00000001) != 0),
+                getParentForChildren(),
+                isClean());
+        function_ = null;
+      }
+      return functionBuilder_;
+    }
+
+    private java.util.List<org.tensorflow.proto.framework.GradientDef> gradient_ =
+      java.util.Collections.emptyList();
+    private void ensureGradientIsMutable() {
+      if (!((bitField0_ & 0x00000002) != 0)) {
+        gradient_ = new java.util.ArrayList<org.tensorflow.proto.framework.GradientDef>(gradient_);
+        bitField0_ |= 0x00000002;
+       }
+    }
+
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.framework.GradientDef, org.tensorflow.proto.framework.GradientDef.Builder, org.tensorflow.proto.framework.GradientDefOrBuilder> gradientBuilder_;
+
+    /**
+     * <code>repeated .tensorflow.GradientDef gradient = 2;</code>
+     */
+    public java.util.List<org.tensorflow.proto.framework.GradientDef> getGradientList() {
+      if (gradientBuilder_ == null) {
+        return java.util.Collections.unmodifiableList(gradient_);
+      } else {
+        return gradientBuilder_.getMessageList();
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.GradientDef gradient = 2;</code>
+     */
+    public int getGradientCount() {
+      if (gradientBuilder_ == null) {
+        return gradient_.size();
+      } else {
+        return gradientBuilder_.getCount();
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.GradientDef gradient = 2;</code>
+     */
+    public org.tensorflow.proto.framework.GradientDef getGradient(int index) {
+      if (gradientBuilder_ == null) {
+        return gradient_.get(index);
+      } else {
+        return gradientBuilder_.getMessage(index);
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.GradientDef gradient = 2;</code>
+     */
+    public Builder setGradient(
+        int index, org.tensorflow.proto.framework.GradientDef value) {
+      if (gradientBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureGradientIsMutable();
+        gradient_.set(index, value);
+        onChanged();
+      } else {
+        gradientBuilder_.setMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.GradientDef gradient = 2;</code>
+     */
+    public Builder setGradient(
+        int index, org.tensorflow.proto.framework.GradientDef.Builder builderForValue) {
+      if (gradientBuilder_ == null) {
+        ensureGradientIsMutable();
+        gradient_.set(index, builderForValue.build());
+        onChanged();
+      } else {
+        gradientBuilder_.setMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.GradientDef gradient = 2;</code>
+     */
+    public Builder addGradient(org.tensorflow.proto.framework.GradientDef value) {
+      if (gradientBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureGradientIsMutable();
+        gradient_.add(value);
+        onChanged();
+      } else {
+        gradientBuilder_.addMessage(value);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.GradientDef gradient = 2;</code>
+     */
+    public Builder addGradient(
+        int index, org.tensorflow.proto.framework.GradientDef value) {
+      if (gradientBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureGradientIsMutable();
+        gradient_.add(index, value);
+        onChanged();
+      } else {
+        gradientBuilder_.addMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.GradientDef gradient = 2;</code>
+     */
+    public Builder addGradient(
+        org.tensorflow.proto.framework.GradientDef.Builder builderForValue) {
+      if (gradientBuilder_ == null) {
+        ensureGradientIsMutable();
+        gradient_.add(builderForValue.build());
+        onChanged();
+      } else {
+        gradientBuilder_.addMessage(builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.GradientDef gradient = 2;</code>
+     */
+    public Builder addGradient(
+        int index, org.tensorflow.proto.framework.GradientDef.Builder builderForValue) {
+      if (gradientBuilder_ == null) {
+        ensureGradientIsMutable();
+        gradient_.add(index, builderForValue.build());
+        onChanged();
+      } else {
+        gradientBuilder_.addMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.GradientDef gradient = 2;</code>
+     */
+    public Builder addAllGradient(
+        java.lang.Iterable<? extends org.tensorflow.proto.framework.GradientDef> values) {
+      if (gradientBuilder_ == null) {
+        ensureGradientIsMutable();
+        com.google.protobuf.AbstractMessageLite.Builder.addAll(
+            values, gradient_);
+        onChanged();
+      } else {
+        gradientBuilder_.addAllMessages(values);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.GradientDef gradient = 2;</code>
+     */
+    public Builder clearGradient() {
+      if (gradientBuilder_ == null) {
+        gradient_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000002);
+        onChanged();
+      } else {
+        gradientBuilder_.clear();
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.GradientDef gradient = 2;</code>
+     */
+    public Builder removeGradient(int index) {
+      if (gradientBuilder_ == null) {
+        ensureGradientIsMutable();
+        gradient_.remove(index);
+        onChanged();
+      } else {
+        gradientBuilder_.remove(index);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.GradientDef gradient = 2;</code>
+     */
+    public org.tensorflow.proto.framework.GradientDef.Builder getGradientBuilder(
+        int index) {
+      return getGradientFieldBuilder().getBuilder(index);
+    }
+    /**
+     * <code>repeated .tensorflow.GradientDef gradient = 2;</code>
+     */
+    public org.tensorflow.proto.framework.GradientDefOrBuilder getGradientOrBuilder(
+        int index) {
+      if (gradientBuilder_ == null) {
+        return gradient_.get(index);  } else {
+        return gradientBuilder_.getMessageOrBuilder(index);
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.GradientDef gradient = 2;</code>
+     */
+    public java.util.List<? extends org.tensorflow.proto.framework.GradientDefOrBuilder> 
+         getGradientOrBuilderList() {
+      if (gradientBuilder_ != null) {
+        return gradientBuilder_.getMessageOrBuilderList();
+      } else {
+        return java.util.Collections.unmodifiableList(gradient_);
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.GradientDef gradient = 2;</code>
+     */
+    public org.tensorflow.proto.framework.GradientDef.Builder addGradientBuilder() {
+      return getGradientFieldBuilder().addBuilder(
+          org.tensorflow.proto.framework.GradientDef.getDefaultInstance());
+    }
+    /**
+     * <code>repeated .tensorflow.GradientDef gradient = 2;</code>
+     */
+    public org.tensorflow.proto.framework.GradientDef.Builder addGradientBuilder(
+        int index) {
+      return getGradientFieldBuilder().addBuilder(
+          index, org.tensorflow.proto.framework.GradientDef.getDefaultInstance());
+    }
+    /**
+     * <code>repeated .tensorflow.GradientDef gradient = 2;</code>
+     */
+    public java.util.List<org.tensorflow.proto.framework.GradientDef.Builder> 
+         getGradientBuilderList() {
+      return getGradientFieldBuilder().getBuilderList();
+    }
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.framework.GradientDef, org.tensorflow.proto.framework.GradientDef.Builder, org.tensorflow.proto.framework.GradientDefOrBuilder> 
+        getGradientFieldBuilder() {
+      if (gradientBuilder_ == null) {
+        gradientBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
+            org.tensorflow.proto.framework.GradientDef, org.tensorflow.proto.framework.GradientDef.Builder, org.tensorflow.proto.framework.GradientDefOrBuilder>(
+                gradient_,
+                ((bitField0_ & 0x00000002) != 0),
+                getParentForChildren(),
+                isClean());
+        gradient_ = null;
+      }
+      return gradientBuilder_;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.FunctionDefLibrary)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.FunctionDefLibrary)
+  private static final org.tensorflow.proto.framework.FunctionDefLibrary DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.FunctionDefLibrary();
+  }
+
+  public static org.tensorflow.proto.framework.FunctionDefLibrary getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<FunctionDefLibrary>
+      PARSER = new com.google.protobuf.AbstractParser<FunctionDefLibrary>() {
+    @java.lang.Override
+    public FunctionDefLibrary parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new FunctionDefLibrary(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<FunctionDefLibrary> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<FunctionDefLibrary> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.framework.FunctionDefLibrary getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/FunctionDefLibraryOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/FunctionDefLibraryOrBuilder.java
new file mode 100644
index 00000000000..cfb61e977ba
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/FunctionDefLibraryOrBuilder.java
@@ -0,0 +1,57 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/framework/function.proto
+
+package org.tensorflow.proto.framework;
+
+public interface FunctionDefLibraryOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.FunctionDefLibrary)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <code>repeated .tensorflow.FunctionDef function = 1;</code>
+   */
+  java.util.List<org.tensorflow.proto.framework.FunctionDef> 
+      getFunctionList();
+  /**
+   * <code>repeated .tensorflow.FunctionDef function = 1;</code>
+   */
+  org.tensorflow.proto.framework.FunctionDef getFunction(int index);
+  /**
+   * <code>repeated .tensorflow.FunctionDef function = 1;</code>
+   */
+  int getFunctionCount();
+  /**
+   * <code>repeated .tensorflow.FunctionDef function = 1;</code>
+   */
+  java.util.List<? extends org.tensorflow.proto.framework.FunctionDefOrBuilder> 
+      getFunctionOrBuilderList();
+  /**
+   * <code>repeated .tensorflow.FunctionDef function = 1;</code>
+   */
+  org.tensorflow.proto.framework.FunctionDefOrBuilder getFunctionOrBuilder(
+      int index);
+
+  /**
+   * <code>repeated .tensorflow.GradientDef gradient = 2;</code>
+   */
+  java.util.List<org.tensorflow.proto.framework.GradientDef> 
+      getGradientList();
+  /**
+   * <code>repeated .tensorflow.GradientDef gradient = 2;</code>
+   */
+  org.tensorflow.proto.framework.GradientDef getGradient(int index);
+  /**
+   * <code>repeated .tensorflow.GradientDef gradient = 2;</code>
+   */
+  int getGradientCount();
+  /**
+   * <code>repeated .tensorflow.GradientDef gradient = 2;</code>
+   */
+  java.util.List<? extends org.tensorflow.proto.framework.GradientDefOrBuilder> 
+      getGradientOrBuilderList();
+  /**
+   * <code>repeated .tensorflow.GradientDef gradient = 2;</code>
+   */
+  org.tensorflow.proto.framework.GradientDefOrBuilder getGradientOrBuilder(
+      int index);
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/FunctionDefOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/FunctionDefOrBuilder.java
new file mode 100644
index 00000000000..ebae0875f50
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/FunctionDefOrBuilder.java
@@ -0,0 +1,381 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/framework/function.proto
+
+package org.tensorflow.proto.framework;
+
+public interface FunctionDefOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.FunctionDef)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <pre>
+   * The definition of the function's name, arguments, return values,
+   * attrs etc.
+   * </pre>
+   *
+   * <code>.tensorflow.OpDef signature = 1;</code>
+   */
+  boolean hasSignature();
+  /**
+   * <pre>
+   * The definition of the function's name, arguments, return values,
+   * attrs etc.
+   * </pre>
+   *
+   * <code>.tensorflow.OpDef signature = 1;</code>
+   */
+  org.tensorflow.proto.framework.OpDef getSignature();
+  /**
+   * <pre>
+   * The definition of the function's name, arguments, return values,
+   * attrs etc.
+   * </pre>
+   *
+   * <code>.tensorflow.OpDef signature = 1;</code>
+   */
+  org.tensorflow.proto.framework.OpDefOrBuilder getSignatureOrBuilder();
+
+  /**
+   * <pre>
+   * Attributes specific to this function definition.
+   * </pre>
+   *
+   * <code>map&lt;string, .tensorflow.AttrValue&gt; attr = 5;</code>
+   */
+  int getAttrCount();
+  /**
+   * <pre>
+   * Attributes specific to this function definition.
+   * </pre>
+   *
+   * <code>map&lt;string, .tensorflow.AttrValue&gt; attr = 5;</code>
+   */
+  boolean containsAttr(
+      java.lang.String key);
+  /**
+   * Use {@link #getAttrMap()} instead.
+   */
+  @java.lang.Deprecated
+  java.util.Map<java.lang.String, org.tensorflow.proto.framework.AttrValue>
+  getAttr();
+  /**
+   * <pre>
+   * Attributes specific to this function definition.
+   * </pre>
+   *
+   * <code>map&lt;string, .tensorflow.AttrValue&gt; attr = 5;</code>
+   */
+  java.util.Map<java.lang.String, org.tensorflow.proto.framework.AttrValue>
+  getAttrMap();
+  /**
+   * <pre>
+   * Attributes specific to this function definition.
+   * </pre>
+   *
+   * <code>map&lt;string, .tensorflow.AttrValue&gt; attr = 5;</code>
+   */
+
+  org.tensorflow.proto.framework.AttrValue getAttrOrDefault(
+      java.lang.String key,
+      org.tensorflow.proto.framework.AttrValue defaultValue);
+  /**
+   * <pre>
+   * Attributes specific to this function definition.
+   * </pre>
+   *
+   * <code>map&lt;string, .tensorflow.AttrValue&gt; attr = 5;</code>
+   */
+
+  org.tensorflow.proto.framework.AttrValue getAttrOrThrow(
+      java.lang.String key);
+
+  /**
+   * <code>map&lt;uint32, .tensorflow.FunctionDef.ArgAttrs&gt; arg_attr = 7;</code>
+   */
+  int getArgAttrCount();
+  /**
+   * <code>map&lt;uint32, .tensorflow.FunctionDef.ArgAttrs&gt; arg_attr = 7;</code>
+   */
+  boolean containsArgAttr(
+      int key);
+  /**
+   * Use {@link #getArgAttrMap()} instead.
+   */
+  @java.lang.Deprecated
+  java.util.Map<java.lang.Integer, org.tensorflow.proto.framework.FunctionDef.ArgAttrs>
+  getArgAttr();
+  /**
+   * <code>map&lt;uint32, .tensorflow.FunctionDef.ArgAttrs&gt; arg_attr = 7;</code>
+   */
+  java.util.Map<java.lang.Integer, org.tensorflow.proto.framework.FunctionDef.ArgAttrs>
+  getArgAttrMap();
+  /**
+   * <code>map&lt;uint32, .tensorflow.FunctionDef.ArgAttrs&gt; arg_attr = 7;</code>
+   */
+
+  org.tensorflow.proto.framework.FunctionDef.ArgAttrs getArgAttrOrDefault(
+      int key,
+      org.tensorflow.proto.framework.FunctionDef.ArgAttrs defaultValue);
+  /**
+   * <code>map&lt;uint32, .tensorflow.FunctionDef.ArgAttrs&gt; arg_attr = 7;</code>
+   */
+
+  org.tensorflow.proto.framework.FunctionDef.ArgAttrs getArgAttrOrThrow(
+      int key);
+
+  /**
+   * <pre>
+   * Unique IDs for each resource argument, used to track aliasing resources. If
+   * Argument A and Argument B alias each other, then
+   * resource_arg_unique_ids[A.index] == resource_arg_unique_ids[B.index].
+   * If this field is empty, none of the arguments could alias; otherwise, every
+   * resource argument should have an entry in this field.
+   * When instantiated, the unique IDs will be attached to the _Arg nodes'
+   * "_resource_arg_unique_id" attribute.
+   * </pre>
+   *
+   * <code>map&lt;uint32, uint32&gt; resource_arg_unique_id = 8;</code>
+   */
+  int getResourceArgUniqueIdCount();
+  /**
+   * <pre>
+   * Unique IDs for each resource argument, used to track aliasing resources. If
+   * Argument A and Argument B alias each other, then
+   * resource_arg_unique_ids[A.index] == resource_arg_unique_ids[B.index].
+   * If this field is empty, none of the arguments could alias; otherwise, every
+   * resource argument should have an entry in this field.
+   * When instantiated, the unique IDs will be attached to the _Arg nodes'
+   * "_resource_arg_unique_id" attribute.
+   * </pre>
+   *
+   * <code>map&lt;uint32, uint32&gt; resource_arg_unique_id = 8;</code>
+   */
+  boolean containsResourceArgUniqueId(
+      int key);
+  /**
+   * Use {@link #getResourceArgUniqueIdMap()} instead.
+   */
+  @java.lang.Deprecated
+  java.util.Map<java.lang.Integer, java.lang.Integer>
+  getResourceArgUniqueId();
+  /**
+   * <pre>
+   * Unique IDs for each resource argument, used to track aliasing resources. If
+   * Argument A and Argument B alias each other, then
+   * resource_arg_unique_ids[A.index] == resource_arg_unique_ids[B.index].
+   * If this field is empty, none of the arguments could alias; otherwise, every
+   * resource argument should have an entry in this field.
+   * When instantiated, the unique IDs will be attached to the _Arg nodes'
+   * "_resource_arg_unique_id" attribute.
+   * </pre>
+   *
+   * <code>map&lt;uint32, uint32&gt; resource_arg_unique_id = 8;</code>
+   */
+  java.util.Map<java.lang.Integer, java.lang.Integer>
+  getResourceArgUniqueIdMap();
+  /**
+   * <pre>
+   * Unique IDs for each resource argument, used to track aliasing resources. If
+   * Argument A and Argument B alias each other, then
+   * resource_arg_unique_ids[A.index] == resource_arg_unique_ids[B.index].
+   * If this field is empty, none of the arguments could alias; otherwise, every
+   * resource argument should have an entry in this field.
+   * When instantiated, the unique IDs will be attached to the _Arg nodes'
+   * "_resource_arg_unique_id" attribute.
+   * </pre>
+   *
+   * <code>map&lt;uint32, uint32&gt; resource_arg_unique_id = 8;</code>
+   */
+
+  int getResourceArgUniqueIdOrDefault(
+      int key,
+      int defaultValue);
+  /**
+   * <pre>
+   * Unique IDs for each resource argument, used to track aliasing resources. If
+   * Argument A and Argument B alias each other, then
+   * resource_arg_unique_ids[A.index] == resource_arg_unique_ids[B.index].
+   * If this field is empty, none of the arguments could alias; otherwise, every
+   * resource argument should have an entry in this field.
+   * When instantiated, the unique IDs will be attached to the _Arg nodes'
+   * "_resource_arg_unique_id" attribute.
+   * </pre>
+   *
+   * <code>map&lt;uint32, uint32&gt; resource_arg_unique_id = 8;</code>
+   */
+
+  int getResourceArgUniqueIdOrThrow(
+      int key);
+
+  /**
+   * <pre>
+   * By convention, "op" in node_def is resolved by consulting with a
+   * user-defined library first. If not resolved, "func" is assumed to
+   * be a builtin op.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.NodeDef node_def = 3;</code>
+   */
+  java.util.List<org.tensorflow.proto.framework.NodeDef> 
+      getNodeDefList();
+  /**
+   * <pre>
+   * By convention, "op" in node_def is resolved by consulting with a
+   * user-defined library first. If not resolved, "func" is assumed to
+   * be a builtin op.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.NodeDef node_def = 3;</code>
+   */
+  org.tensorflow.proto.framework.NodeDef getNodeDef(int index);
+  /**
+   * <pre>
+   * By convention, "op" in node_def is resolved by consulting with a
+   * user-defined library first. If not resolved, "func" is assumed to
+   * be a builtin op.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.NodeDef node_def = 3;</code>
+   */
+  int getNodeDefCount();
+  /**
+   * <pre>
+   * By convention, "op" in node_def is resolved by consulting with a
+   * user-defined library first. If not resolved, "func" is assumed to
+   * be a builtin op.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.NodeDef node_def = 3;</code>
+   */
+  java.util.List<? extends org.tensorflow.proto.framework.NodeDefOrBuilder> 
+      getNodeDefOrBuilderList();
+  /**
+   * <pre>
+   * By convention, "op" in node_def is resolved by consulting with a
+   * user-defined library first. If not resolved, "func" is assumed to
+   * be a builtin op.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.NodeDef node_def = 3;</code>
+   */
+  org.tensorflow.proto.framework.NodeDefOrBuilder getNodeDefOrBuilder(
+      int index);
+
+  /**
+   * <pre>
+   * A mapping from the output arg names from `signature` to the
+   * outputs from `node_def` that should be returned by the function.
+   * </pre>
+   *
+   * <code>map&lt;string, string&gt; ret = 4;</code>
+   */
+  int getRetCount();
+  /**
+   * <pre>
+   * A mapping from the output arg names from `signature` to the
+   * outputs from `node_def` that should be returned by the function.
+   * </pre>
+   *
+   * <code>map&lt;string, string&gt; ret = 4;</code>
+   */
+  boolean containsRet(
+      java.lang.String key);
+  /**
+   * Use {@link #getRetMap()} instead.
+   */
+  @java.lang.Deprecated
+  java.util.Map<java.lang.String, java.lang.String>
+  getRet();
+  /**
+   * <pre>
+   * A mapping from the output arg names from `signature` to the
+   * outputs from `node_def` that should be returned by the function.
+   * </pre>
+   *
+   * <code>map&lt;string, string&gt; ret = 4;</code>
+   */
+  java.util.Map<java.lang.String, java.lang.String>
+  getRetMap();
+  /**
+   * <pre>
+   * A mapping from the output arg names from `signature` to the
+   * outputs from `node_def` that should be returned by the function.
+   * </pre>
+   *
+   * <code>map&lt;string, string&gt; ret = 4;</code>
+   */
+
+  java.lang.String getRetOrDefault(
+      java.lang.String key,
+      java.lang.String defaultValue);
+  /**
+   * <pre>
+   * A mapping from the output arg names from `signature` to the
+   * outputs from `node_def` that should be returned by the function.
+   * </pre>
+   *
+   * <code>map&lt;string, string&gt; ret = 4;</code>
+   */
+
+  java.lang.String getRetOrThrow(
+      java.lang.String key);
+
+  /**
+   * <pre>
+   * A mapping from control output names from `signature` to node names in
+   * `node_def` which should be control outputs of this function.
+   * </pre>
+   *
+   * <code>map&lt;string, string&gt; control_ret = 6;</code>
+   */
+  int getControlRetCount();
+  /**
+   * <pre>
+   * A mapping from control output names from `signature` to node names in
+   * `node_def` which should be control outputs of this function.
+   * </pre>
+   *
+   * <code>map&lt;string, string&gt; control_ret = 6;</code>
+   */
+  boolean containsControlRet(
+      java.lang.String key);
+  /**
+   * Use {@link #getControlRetMap()} instead.
+   */
+  @java.lang.Deprecated
+  java.util.Map<java.lang.String, java.lang.String>
+  getControlRet();
+  /**
+   * <pre>
+   * A mapping from control output names from `signature` to node names in
+   * `node_def` which should be control outputs of this function.
+   * </pre>
+   *
+   * <code>map&lt;string, string&gt; control_ret = 6;</code>
+   */
+  java.util.Map<java.lang.String, java.lang.String>
+  getControlRetMap();
+  /**
+   * <pre>
+   * A mapping from control output names from `signature` to node names in
+   * `node_def` which should be control outputs of this function.
+   * </pre>
+   *
+   * <code>map&lt;string, string&gt; control_ret = 6;</code>
+   */
+
+  java.lang.String getControlRetOrDefault(
+      java.lang.String key,
+      java.lang.String defaultValue);
+  /**
+   * <pre>
+   * A mapping from control output names from `signature` to node names in
+   * `node_def` which should be control outputs of this function.
+   * </pre>
+   *
+   * <code>map&lt;string, string&gt; control_ret = 6;</code>
+   */
+
+  java.lang.String getControlRetOrThrow(
+      java.lang.String key);
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/FunctionProtos.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/FunctionProtos.java
new file mode 100644
index 00000000000..e33df603ca8
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/FunctionProtos.java
@@ -0,0 +1,184 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/framework/function.proto
+
+package org.tensorflow.proto.framework;
+
+public final class FunctionProtos {
+  private FunctionProtos() {}
+  public static void registerAllExtensions(
+      com.google.protobuf.ExtensionRegistryLite registry) {
+  }
+
+  public static void registerAllExtensions(
+      com.google.protobuf.ExtensionRegistry registry) {
+    registerAllExtensions(
+        (com.google.protobuf.ExtensionRegistryLite) registry);
+  }
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_FunctionDefLibrary_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_FunctionDefLibrary_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_FunctionDef_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_FunctionDef_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_FunctionDef_AttrEntry_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_FunctionDef_AttrEntry_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_FunctionDef_ArgAttrs_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_FunctionDef_ArgAttrs_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_FunctionDef_ArgAttrs_AttrEntry_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_FunctionDef_ArgAttrs_AttrEntry_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_FunctionDef_ArgAttrEntry_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_FunctionDef_ArgAttrEntry_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_FunctionDef_ResourceArgUniqueIdEntry_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_FunctionDef_ResourceArgUniqueIdEntry_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_FunctionDef_RetEntry_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_FunctionDef_RetEntry_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_FunctionDef_ControlRetEntry_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_FunctionDef_ControlRetEntry_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_GradientDef_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_GradientDef_fieldAccessorTable;
+
+  public static com.google.protobuf.Descriptors.FileDescriptor
+      getDescriptor() {
+    return descriptor;
+  }
+  private static  com.google.protobuf.Descriptors.FileDescriptor
+      descriptor;
+  static {
+    java.lang.String[] descriptorData = {
+      "\n(tensorflow/core/framework/function.pro" +
+      "to\022\ntensorflow\032*tensorflow/core/framewor" +
+      "k/attr_value.proto\032(tensorflow/core/fram" +
+      "ework/node_def.proto\032&tensorflow/core/fr" +
+      "amework/op_def.proto\"j\n\022FunctionDefLibra" +
+      "ry\022)\n\010function\030\001 \003(\0132\027.tensorflow.Functi" +
+      "onDef\022)\n\010gradient\030\002 \003(\0132\027.tensorflow.Gra" +
+      "dientDef\"\304\006\n\013FunctionDef\022$\n\tsignature\030\001 " +
+      "\001(\0132\021.tensorflow.OpDef\022/\n\004attr\030\005 \003(\0132!.t" +
+      "ensorflow.FunctionDef.AttrEntry\0226\n\010arg_a" +
+      "ttr\030\007 \003(\0132$.tensorflow.FunctionDef.ArgAt" +
+      "trEntry\022P\n\026resource_arg_unique_id\030\010 \003(\0132" +
+      "0.tensorflow.FunctionDef.ResourceArgUniq" +
+      "ueIdEntry\022%\n\010node_def\030\003 \003(\0132\023.tensorflow" +
+      ".NodeDef\022-\n\003ret\030\004 \003(\0132 .tensorflow.Funct" +
+      "ionDef.RetEntry\022<\n\013control_ret\030\006 \003(\0132\'.t" +
+      "ensorflow.FunctionDef.ControlRetEntry\032B\n" +
+      "\tAttrEntry\022\013\n\003key\030\001 \001(\t\022$\n\005value\030\002 \001(\0132\025" +
+      ".tensorflow.AttrValue:\0028\001\032\210\001\n\010ArgAttrs\0228" +
+      "\n\004attr\030\001 \003(\0132*.tensorflow.FunctionDef.Ar" +
+      "gAttrs.AttrEntry\032B\n\tAttrEntry\022\013\n\003key\030\001 \001" +
+      "(\t\022$\n\005value\030\002 \001(\0132\025.tensorflow.AttrValue" +
+      ":\0028\001\032P\n\014ArgAttrEntry\022\013\n\003key\030\001 \001(\r\022/\n\005val" +
+      "ue\030\002 \001(\0132 .tensorflow.FunctionDef.ArgAtt" +
+      "rs:\0028\001\032:\n\030ResourceArgUniqueIdEntry\022\013\n\003ke" +
+      "y\030\001 \001(\r\022\r\n\005value\030\002 \001(\r:\0028\001\032*\n\010RetEntry\022\013" +
+      "\n\003key\030\001 \001(\t\022\r\n\005value\030\002 \001(\t:\0028\001\0321\n\017Contro" +
+      "lRetEntry\022\013\n\003key\030\001 \001(\t\022\r\n\005value\030\002 \001(\t:\0028" +
+      "\001J\004\010\002\020\003\";\n\013GradientDef\022\025\n\rfunction_name\030" +
+      "\001 \001(\t\022\025\n\rgradient_func\030\002 \001(\tB\206\001\n\036org.ten" +
+      "sorflow.proto.frameworkB\016FunctionProtosP" +
+      "\001ZOgithub.com/tensorflow/tensorflow/tens" +
+      "orflow/go/core/framework/function_go_pro" +
+      "to\370\001\001b\006proto3"
+    };
+    descriptor = com.google.protobuf.Descriptors.FileDescriptor
+      .internalBuildGeneratedFileFrom(descriptorData,
+        new com.google.protobuf.Descriptors.FileDescriptor[] {
+          org.tensorflow.proto.framework.AttrValueProtos.getDescriptor(),
+          org.tensorflow.proto.framework.NodeProto.getDescriptor(),
+          org.tensorflow.proto.framework.OpDefProtos.getDescriptor(),
+        });
+    internal_static_tensorflow_FunctionDefLibrary_descriptor =
+      getDescriptor().getMessageTypes().get(0);
+    internal_static_tensorflow_FunctionDefLibrary_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_FunctionDefLibrary_descriptor,
+        new java.lang.String[] { "Function", "Gradient", });
+    internal_static_tensorflow_FunctionDef_descriptor =
+      getDescriptor().getMessageTypes().get(1);
+    internal_static_tensorflow_FunctionDef_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_FunctionDef_descriptor,
+        new java.lang.String[] { "Signature", "Attr", "ArgAttr", "ResourceArgUniqueId", "NodeDef", "Ret", "ControlRet", });
+    internal_static_tensorflow_FunctionDef_AttrEntry_descriptor =
+      internal_static_tensorflow_FunctionDef_descriptor.getNestedTypes().get(0);
+    internal_static_tensorflow_FunctionDef_AttrEntry_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_FunctionDef_AttrEntry_descriptor,
+        new java.lang.String[] { "Key", "Value", });
+    internal_static_tensorflow_FunctionDef_ArgAttrs_descriptor =
+      internal_static_tensorflow_FunctionDef_descriptor.getNestedTypes().get(1);
+    internal_static_tensorflow_FunctionDef_ArgAttrs_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_FunctionDef_ArgAttrs_descriptor,
+        new java.lang.String[] { "Attr", });
+    internal_static_tensorflow_FunctionDef_ArgAttrs_AttrEntry_descriptor =
+      internal_static_tensorflow_FunctionDef_ArgAttrs_descriptor.getNestedTypes().get(0);
+    internal_static_tensorflow_FunctionDef_ArgAttrs_AttrEntry_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_FunctionDef_ArgAttrs_AttrEntry_descriptor,
+        new java.lang.String[] { "Key", "Value", });
+    internal_static_tensorflow_FunctionDef_ArgAttrEntry_descriptor =
+      internal_static_tensorflow_FunctionDef_descriptor.getNestedTypes().get(2);
+    internal_static_tensorflow_FunctionDef_ArgAttrEntry_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_FunctionDef_ArgAttrEntry_descriptor,
+        new java.lang.String[] { "Key", "Value", });
+    internal_static_tensorflow_FunctionDef_ResourceArgUniqueIdEntry_descriptor =
+      internal_static_tensorflow_FunctionDef_descriptor.getNestedTypes().get(3);
+    internal_static_tensorflow_FunctionDef_ResourceArgUniqueIdEntry_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_FunctionDef_ResourceArgUniqueIdEntry_descriptor,
+        new java.lang.String[] { "Key", "Value", });
+    internal_static_tensorflow_FunctionDef_RetEntry_descriptor =
+      internal_static_tensorflow_FunctionDef_descriptor.getNestedTypes().get(4);
+    internal_static_tensorflow_FunctionDef_RetEntry_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_FunctionDef_RetEntry_descriptor,
+        new java.lang.String[] { "Key", "Value", });
+    internal_static_tensorflow_FunctionDef_ControlRetEntry_descriptor =
+      internal_static_tensorflow_FunctionDef_descriptor.getNestedTypes().get(5);
+    internal_static_tensorflow_FunctionDef_ControlRetEntry_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_FunctionDef_ControlRetEntry_descriptor,
+        new java.lang.String[] { "Key", "Value", });
+    internal_static_tensorflow_GradientDef_descriptor =
+      getDescriptor().getMessageTypes().get(2);
+    internal_static_tensorflow_GradientDef_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_GradientDef_descriptor,
+        new java.lang.String[] { "FunctionName", "GradientFunc", });
+    org.tensorflow.proto.framework.AttrValueProtos.getDescriptor();
+    org.tensorflow.proto.framework.NodeProto.getDescriptor();
+    org.tensorflow.proto.framework.OpDefProtos.getDescriptor();
+  }
+
+  // @@protoc_insertion_point(outer_class_scope)
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/FunctionSpec.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/FunctionSpec.java
new file mode 100644
index 00000000000..0feaaf41951
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/FunctionSpec.java
@@ -0,0 +1,961 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/saved_object_graph.proto
+
+package org.tensorflow.proto.framework;
+
+/**
+ * <pre>
+ * Represents `FunctionSpec` used in `Function`. This represents a
+ * function that has been wrapped as a TensorFlow `Function`.
+ * </pre>
+ *
+ * Protobuf type {@code tensorflow.FunctionSpec}
+ */
+public  final class FunctionSpec extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.FunctionSpec)
+    FunctionSpecOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use FunctionSpec.newBuilder() to construct.
+  private FunctionSpec(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private FunctionSpec() {
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new FunctionSpec();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private FunctionSpec(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 10: {
+            org.tensorflow.proto.framework.StructuredValue.Builder subBuilder = null;
+            if (fullargspec_ != null) {
+              subBuilder = fullargspec_.toBuilder();
+            }
+            fullargspec_ = input.readMessage(org.tensorflow.proto.framework.StructuredValue.parser(), extensionRegistry);
+            if (subBuilder != null) {
+              subBuilder.mergeFrom(fullargspec_);
+              fullargspec_ = subBuilder.buildPartial();
+            }
+
+            break;
+          }
+          case 16: {
+
+            isMethod_ = input.readBool();
+            break;
+          }
+          case 42: {
+            org.tensorflow.proto.framework.StructuredValue.Builder subBuilder = null;
+            if (inputSignature_ != null) {
+              subBuilder = inputSignature_.toBuilder();
+            }
+            inputSignature_ = input.readMessage(org.tensorflow.proto.framework.StructuredValue.parser(), extensionRegistry);
+            if (subBuilder != null) {
+              subBuilder.mergeFrom(inputSignature_);
+              inputSignature_ = subBuilder.buildPartial();
+            }
+
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.framework.SavedObjectGraphProtos.internal_static_tensorflow_FunctionSpec_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.framework.SavedObjectGraphProtos.internal_static_tensorflow_FunctionSpec_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.framework.FunctionSpec.class, org.tensorflow.proto.framework.FunctionSpec.Builder.class);
+  }
+
+  public static final int FULLARGSPEC_FIELD_NUMBER = 1;
+  private org.tensorflow.proto.framework.StructuredValue fullargspec_;
+  /**
+   * <pre>
+   * Full arg spec from inspect.getfullargspec().
+   * </pre>
+   *
+   * <code>.tensorflow.StructuredValue fullargspec = 1;</code>
+   */
+  public boolean hasFullargspec() {
+    return fullargspec_ != null;
+  }
+  /**
+   * <pre>
+   * Full arg spec from inspect.getfullargspec().
+   * </pre>
+   *
+   * <code>.tensorflow.StructuredValue fullargspec = 1;</code>
+   */
+  public org.tensorflow.proto.framework.StructuredValue getFullargspec() {
+    return fullargspec_ == null ? org.tensorflow.proto.framework.StructuredValue.getDefaultInstance() : fullargspec_;
+  }
+  /**
+   * <pre>
+   * Full arg spec from inspect.getfullargspec().
+   * </pre>
+   *
+   * <code>.tensorflow.StructuredValue fullargspec = 1;</code>
+   */
+  public org.tensorflow.proto.framework.StructuredValueOrBuilder getFullargspecOrBuilder() {
+    return getFullargspec();
+  }
+
+  public static final int IS_METHOD_FIELD_NUMBER = 2;
+  private boolean isMethod_;
+  /**
+   * <pre>
+   * Whether this represents a class method.
+   * </pre>
+   *
+   * <code>bool is_method = 2;</code>
+   */
+  public boolean getIsMethod() {
+    return isMethod_;
+  }
+
+  public static final int INPUT_SIGNATURE_FIELD_NUMBER = 5;
+  private org.tensorflow.proto.framework.StructuredValue inputSignature_;
+  /**
+   * <pre>
+   * The input signature, if specified.
+   * </pre>
+   *
+   * <code>.tensorflow.StructuredValue input_signature = 5;</code>
+   */
+  public boolean hasInputSignature() {
+    return inputSignature_ != null;
+  }
+  /**
+   * <pre>
+   * The input signature, if specified.
+   * </pre>
+   *
+   * <code>.tensorflow.StructuredValue input_signature = 5;</code>
+   */
+  public org.tensorflow.proto.framework.StructuredValue getInputSignature() {
+    return inputSignature_ == null ? org.tensorflow.proto.framework.StructuredValue.getDefaultInstance() : inputSignature_;
+  }
+  /**
+   * <pre>
+   * The input signature, if specified.
+   * </pre>
+   *
+   * <code>.tensorflow.StructuredValue input_signature = 5;</code>
+   */
+  public org.tensorflow.proto.framework.StructuredValueOrBuilder getInputSignatureOrBuilder() {
+    return getInputSignature();
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    if (fullargspec_ != null) {
+      output.writeMessage(1, getFullargspec());
+    }
+    if (isMethod_ != false) {
+      output.writeBool(2, isMethod_);
+    }
+    if (inputSignature_ != null) {
+      output.writeMessage(5, getInputSignature());
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (fullargspec_ != null) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(1, getFullargspec());
+    }
+    if (isMethod_ != false) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeBoolSize(2, isMethod_);
+    }
+    if (inputSignature_ != null) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(5, getInputSignature());
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.framework.FunctionSpec)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.framework.FunctionSpec other = (org.tensorflow.proto.framework.FunctionSpec) obj;
+
+    if (hasFullargspec() != other.hasFullargspec()) return false;
+    if (hasFullargspec()) {
+      if (!getFullargspec()
+          .equals(other.getFullargspec())) return false;
+    }
+    if (getIsMethod()
+        != other.getIsMethod()) return false;
+    if (hasInputSignature() != other.hasInputSignature()) return false;
+    if (hasInputSignature()) {
+      if (!getInputSignature()
+          .equals(other.getInputSignature())) return false;
+    }
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    if (hasFullargspec()) {
+      hash = (37 * hash) + FULLARGSPEC_FIELD_NUMBER;
+      hash = (53 * hash) + getFullargspec().hashCode();
+    }
+    hash = (37 * hash) + IS_METHOD_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashBoolean(
+        getIsMethod());
+    if (hasInputSignature()) {
+      hash = (37 * hash) + INPUT_SIGNATURE_FIELD_NUMBER;
+      hash = (53 * hash) + getInputSignature().hashCode();
+    }
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.framework.FunctionSpec parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.FunctionSpec parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.FunctionSpec parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.FunctionSpec parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.FunctionSpec parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.FunctionSpec parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.FunctionSpec parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.FunctionSpec parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.FunctionSpec parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.FunctionSpec parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.FunctionSpec parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.FunctionSpec parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.framework.FunctionSpec prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * <pre>
+   * Represents `FunctionSpec` used in `Function`. This represents a
+   * function that has been wrapped as a TensorFlow `Function`.
+   * </pre>
+   *
+   * Protobuf type {@code tensorflow.FunctionSpec}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.FunctionSpec)
+      org.tensorflow.proto.framework.FunctionSpecOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.SavedObjectGraphProtos.internal_static_tensorflow_FunctionSpec_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.SavedObjectGraphProtos.internal_static_tensorflow_FunctionSpec_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.FunctionSpec.class, org.tensorflow.proto.framework.FunctionSpec.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.framework.FunctionSpec.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      if (fullargspecBuilder_ == null) {
+        fullargspec_ = null;
+      } else {
+        fullargspec_ = null;
+        fullargspecBuilder_ = null;
+      }
+      isMethod_ = false;
+
+      if (inputSignatureBuilder_ == null) {
+        inputSignature_ = null;
+      } else {
+        inputSignature_ = null;
+        inputSignatureBuilder_ = null;
+      }
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.framework.SavedObjectGraphProtos.internal_static_tensorflow_FunctionSpec_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.FunctionSpec getDefaultInstanceForType() {
+      return org.tensorflow.proto.framework.FunctionSpec.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.FunctionSpec build() {
+      org.tensorflow.proto.framework.FunctionSpec result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.FunctionSpec buildPartial() {
+      org.tensorflow.proto.framework.FunctionSpec result = new org.tensorflow.proto.framework.FunctionSpec(this);
+      if (fullargspecBuilder_ == null) {
+        result.fullargspec_ = fullargspec_;
+      } else {
+        result.fullargspec_ = fullargspecBuilder_.build();
+      }
+      result.isMethod_ = isMethod_;
+      if (inputSignatureBuilder_ == null) {
+        result.inputSignature_ = inputSignature_;
+      } else {
+        result.inputSignature_ = inputSignatureBuilder_.build();
+      }
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.framework.FunctionSpec) {
+        return mergeFrom((org.tensorflow.proto.framework.FunctionSpec)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.framework.FunctionSpec other) {
+      if (other == org.tensorflow.proto.framework.FunctionSpec.getDefaultInstance()) return this;
+      if (other.hasFullargspec()) {
+        mergeFullargspec(other.getFullargspec());
+      }
+      if (other.getIsMethod() != false) {
+        setIsMethod(other.getIsMethod());
+      }
+      if (other.hasInputSignature()) {
+        mergeInputSignature(other.getInputSignature());
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.framework.FunctionSpec parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.framework.FunctionSpec) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+
+    private org.tensorflow.proto.framework.StructuredValue fullargspec_;
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.StructuredValue, org.tensorflow.proto.framework.StructuredValue.Builder, org.tensorflow.proto.framework.StructuredValueOrBuilder> fullargspecBuilder_;
+    /**
+     * <pre>
+     * Full arg spec from inspect.getfullargspec().
+     * </pre>
+     *
+     * <code>.tensorflow.StructuredValue fullargspec = 1;</code>
+     */
+    public boolean hasFullargspec() {
+      return fullargspecBuilder_ != null || fullargspec_ != null;
+    }
+    /**
+     * <pre>
+     * Full arg spec from inspect.getfullargspec().
+     * </pre>
+     *
+     * <code>.tensorflow.StructuredValue fullargspec = 1;</code>
+     */
+    public org.tensorflow.proto.framework.StructuredValue getFullargspec() {
+      if (fullargspecBuilder_ == null) {
+        return fullargspec_ == null ? org.tensorflow.proto.framework.StructuredValue.getDefaultInstance() : fullargspec_;
+      } else {
+        return fullargspecBuilder_.getMessage();
+      }
+    }
+    /**
+     * <pre>
+     * Full arg spec from inspect.getfullargspec().
+     * </pre>
+     *
+     * <code>.tensorflow.StructuredValue fullargspec = 1;</code>
+     */
+    public Builder setFullargspec(org.tensorflow.proto.framework.StructuredValue value) {
+      if (fullargspecBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        fullargspec_ = value;
+        onChanged();
+      } else {
+        fullargspecBuilder_.setMessage(value);
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Full arg spec from inspect.getfullargspec().
+     * </pre>
+     *
+     * <code>.tensorflow.StructuredValue fullargspec = 1;</code>
+     */
+    public Builder setFullargspec(
+        org.tensorflow.proto.framework.StructuredValue.Builder builderForValue) {
+      if (fullargspecBuilder_ == null) {
+        fullargspec_ = builderForValue.build();
+        onChanged();
+      } else {
+        fullargspecBuilder_.setMessage(builderForValue.build());
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Full arg spec from inspect.getfullargspec().
+     * </pre>
+     *
+     * <code>.tensorflow.StructuredValue fullargspec = 1;</code>
+     */
+    public Builder mergeFullargspec(org.tensorflow.proto.framework.StructuredValue value) {
+      if (fullargspecBuilder_ == null) {
+        if (fullargspec_ != null) {
+          fullargspec_ =
+            org.tensorflow.proto.framework.StructuredValue.newBuilder(fullargspec_).mergeFrom(value).buildPartial();
+        } else {
+          fullargspec_ = value;
+        }
+        onChanged();
+      } else {
+        fullargspecBuilder_.mergeFrom(value);
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Full arg spec from inspect.getfullargspec().
+     * </pre>
+     *
+     * <code>.tensorflow.StructuredValue fullargspec = 1;</code>
+     */
+    public Builder clearFullargspec() {
+      if (fullargspecBuilder_ == null) {
+        fullargspec_ = null;
+        onChanged();
+      } else {
+        fullargspec_ = null;
+        fullargspecBuilder_ = null;
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Full arg spec from inspect.getfullargspec().
+     * </pre>
+     *
+     * <code>.tensorflow.StructuredValue fullargspec = 1;</code>
+     */
+    public org.tensorflow.proto.framework.StructuredValue.Builder getFullargspecBuilder() {
+      
+      onChanged();
+      return getFullargspecFieldBuilder().getBuilder();
+    }
+    /**
+     * <pre>
+     * Full arg spec from inspect.getfullargspec().
+     * </pre>
+     *
+     * <code>.tensorflow.StructuredValue fullargspec = 1;</code>
+     */
+    public org.tensorflow.proto.framework.StructuredValueOrBuilder getFullargspecOrBuilder() {
+      if (fullargspecBuilder_ != null) {
+        return fullargspecBuilder_.getMessageOrBuilder();
+      } else {
+        return fullargspec_ == null ?
+            org.tensorflow.proto.framework.StructuredValue.getDefaultInstance() : fullargspec_;
+      }
+    }
+    /**
+     * <pre>
+     * Full arg spec from inspect.getfullargspec().
+     * </pre>
+     *
+     * <code>.tensorflow.StructuredValue fullargspec = 1;</code>
+     */
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.StructuredValue, org.tensorflow.proto.framework.StructuredValue.Builder, org.tensorflow.proto.framework.StructuredValueOrBuilder> 
+        getFullargspecFieldBuilder() {
+      if (fullargspecBuilder_ == null) {
+        fullargspecBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+            org.tensorflow.proto.framework.StructuredValue, org.tensorflow.proto.framework.StructuredValue.Builder, org.tensorflow.proto.framework.StructuredValueOrBuilder>(
+                getFullargspec(),
+                getParentForChildren(),
+                isClean());
+        fullargspec_ = null;
+      }
+      return fullargspecBuilder_;
+    }
+
+    private boolean isMethod_ ;
+    /**
+     * <pre>
+     * Whether this represents a class method.
+     * </pre>
+     *
+     * <code>bool is_method = 2;</code>
+     */
+    public boolean getIsMethod() {
+      return isMethod_;
+    }
+    /**
+     * <pre>
+     * Whether this represents a class method.
+     * </pre>
+     *
+     * <code>bool is_method = 2;</code>
+     */
+    public Builder setIsMethod(boolean value) {
+      
+      isMethod_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Whether this represents a class method.
+     * </pre>
+     *
+     * <code>bool is_method = 2;</code>
+     */
+    public Builder clearIsMethod() {
+      
+      isMethod_ = false;
+      onChanged();
+      return this;
+    }
+
+    private org.tensorflow.proto.framework.StructuredValue inputSignature_;
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.StructuredValue, org.tensorflow.proto.framework.StructuredValue.Builder, org.tensorflow.proto.framework.StructuredValueOrBuilder> inputSignatureBuilder_;
+    /**
+     * <pre>
+     * The input signature, if specified.
+     * </pre>
+     *
+     * <code>.tensorflow.StructuredValue input_signature = 5;</code>
+     */
+    public boolean hasInputSignature() {
+      return inputSignatureBuilder_ != null || inputSignature_ != null;
+    }
+    /**
+     * <pre>
+     * The input signature, if specified.
+     * </pre>
+     *
+     * <code>.tensorflow.StructuredValue input_signature = 5;</code>
+     */
+    public org.tensorflow.proto.framework.StructuredValue getInputSignature() {
+      if (inputSignatureBuilder_ == null) {
+        return inputSignature_ == null ? org.tensorflow.proto.framework.StructuredValue.getDefaultInstance() : inputSignature_;
+      } else {
+        return inputSignatureBuilder_.getMessage();
+      }
+    }
+    /**
+     * <pre>
+     * The input signature, if specified.
+     * </pre>
+     *
+     * <code>.tensorflow.StructuredValue input_signature = 5;</code>
+     */
+    public Builder setInputSignature(org.tensorflow.proto.framework.StructuredValue value) {
+      if (inputSignatureBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        inputSignature_ = value;
+        onChanged();
+      } else {
+        inputSignatureBuilder_.setMessage(value);
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * The input signature, if specified.
+     * </pre>
+     *
+     * <code>.tensorflow.StructuredValue input_signature = 5;</code>
+     */
+    public Builder setInputSignature(
+        org.tensorflow.proto.framework.StructuredValue.Builder builderForValue) {
+      if (inputSignatureBuilder_ == null) {
+        inputSignature_ = builderForValue.build();
+        onChanged();
+      } else {
+        inputSignatureBuilder_.setMessage(builderForValue.build());
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * The input signature, if specified.
+     * </pre>
+     *
+     * <code>.tensorflow.StructuredValue input_signature = 5;</code>
+     */
+    public Builder mergeInputSignature(org.tensorflow.proto.framework.StructuredValue value) {
+      if (inputSignatureBuilder_ == null) {
+        if (inputSignature_ != null) {
+          inputSignature_ =
+            org.tensorflow.proto.framework.StructuredValue.newBuilder(inputSignature_).mergeFrom(value).buildPartial();
+        } else {
+          inputSignature_ = value;
+        }
+        onChanged();
+      } else {
+        inputSignatureBuilder_.mergeFrom(value);
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * The input signature, if specified.
+     * </pre>
+     *
+     * <code>.tensorflow.StructuredValue input_signature = 5;</code>
+     */
+    public Builder clearInputSignature() {
+      if (inputSignatureBuilder_ == null) {
+        inputSignature_ = null;
+        onChanged();
+      } else {
+        inputSignature_ = null;
+        inputSignatureBuilder_ = null;
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * The input signature, if specified.
+     * </pre>
+     *
+     * <code>.tensorflow.StructuredValue input_signature = 5;</code>
+     */
+    public org.tensorflow.proto.framework.StructuredValue.Builder getInputSignatureBuilder() {
+      
+      onChanged();
+      return getInputSignatureFieldBuilder().getBuilder();
+    }
+    /**
+     * <pre>
+     * The input signature, if specified.
+     * </pre>
+     *
+     * <code>.tensorflow.StructuredValue input_signature = 5;</code>
+     */
+    public org.tensorflow.proto.framework.StructuredValueOrBuilder getInputSignatureOrBuilder() {
+      if (inputSignatureBuilder_ != null) {
+        return inputSignatureBuilder_.getMessageOrBuilder();
+      } else {
+        return inputSignature_ == null ?
+            org.tensorflow.proto.framework.StructuredValue.getDefaultInstance() : inputSignature_;
+      }
+    }
+    /**
+     * <pre>
+     * The input signature, if specified.
+     * </pre>
+     *
+     * <code>.tensorflow.StructuredValue input_signature = 5;</code>
+     */
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.StructuredValue, org.tensorflow.proto.framework.StructuredValue.Builder, org.tensorflow.proto.framework.StructuredValueOrBuilder> 
+        getInputSignatureFieldBuilder() {
+      if (inputSignatureBuilder_ == null) {
+        inputSignatureBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+            org.tensorflow.proto.framework.StructuredValue, org.tensorflow.proto.framework.StructuredValue.Builder, org.tensorflow.proto.framework.StructuredValueOrBuilder>(
+                getInputSignature(),
+                getParentForChildren(),
+                isClean());
+        inputSignature_ = null;
+      }
+      return inputSignatureBuilder_;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.FunctionSpec)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.FunctionSpec)
+  private static final org.tensorflow.proto.framework.FunctionSpec DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.FunctionSpec();
+  }
+
+  public static org.tensorflow.proto.framework.FunctionSpec getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<FunctionSpec>
+      PARSER = new com.google.protobuf.AbstractParser<FunctionSpec>() {
+    @java.lang.Override
+    public FunctionSpec parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new FunctionSpec(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<FunctionSpec> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<FunctionSpec> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.framework.FunctionSpec getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/FunctionSpecOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/FunctionSpecOrBuilder.java
new file mode 100644
index 00000000000..8c200ad6323
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/FunctionSpecOrBuilder.java
@@ -0,0 +1,68 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/saved_object_graph.proto
+
+package org.tensorflow.proto.framework;
+
+public interface FunctionSpecOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.FunctionSpec)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <pre>
+   * Full arg spec from inspect.getfullargspec().
+   * </pre>
+   *
+   * <code>.tensorflow.StructuredValue fullargspec = 1;</code>
+   */
+  boolean hasFullargspec();
+  /**
+   * <pre>
+   * Full arg spec from inspect.getfullargspec().
+   * </pre>
+   *
+   * <code>.tensorflow.StructuredValue fullargspec = 1;</code>
+   */
+  org.tensorflow.proto.framework.StructuredValue getFullargspec();
+  /**
+   * <pre>
+   * Full arg spec from inspect.getfullargspec().
+   * </pre>
+   *
+   * <code>.tensorflow.StructuredValue fullargspec = 1;</code>
+   */
+  org.tensorflow.proto.framework.StructuredValueOrBuilder getFullargspecOrBuilder();
+
+  /**
+   * <pre>
+   * Whether this represents a class method.
+   * </pre>
+   *
+   * <code>bool is_method = 2;</code>
+   */
+  boolean getIsMethod();
+
+  /**
+   * <pre>
+   * The input signature, if specified.
+   * </pre>
+   *
+   * <code>.tensorflow.StructuredValue input_signature = 5;</code>
+   */
+  boolean hasInputSignature();
+  /**
+   * <pre>
+   * The input signature, if specified.
+   * </pre>
+   *
+   * <code>.tensorflow.StructuredValue input_signature = 5;</code>
+   */
+  org.tensorflow.proto.framework.StructuredValue getInputSignature();
+  /**
+   * <pre>
+   * The input signature, if specified.
+   * </pre>
+   *
+   * <code>.tensorflow.StructuredValue input_signature = 5;</code>
+   */
+  org.tensorflow.proto.framework.StructuredValueOrBuilder getInputSignatureOrBuilder();
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GPUOptions.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GPUOptions.java
new file mode 100644
index 00000000000..44b8a1c43e3
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GPUOptions.java
@@ -0,0 +1,5061 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/config.proto
+
+package org.tensorflow.proto.framework;
+
+/**
+ * Protobuf type {@code tensorflow.GPUOptions}
+ */
+public  final class GPUOptions extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.GPUOptions)
+    GPUOptionsOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use GPUOptions.newBuilder() to construct.
+  private GPUOptions(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private GPUOptions() {
+    allocatorType_ = "";
+    visibleDeviceList_ = "";
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new GPUOptions();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private GPUOptions(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 9: {
+
+            perProcessGpuMemoryFraction_ = input.readDouble();
+            break;
+          }
+          case 18: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            allocatorType_ = s;
+            break;
+          }
+          case 24: {
+
+            deferredDeletionBytes_ = input.readInt64();
+            break;
+          }
+          case 32: {
+
+            allowGrowth_ = input.readBool();
+            break;
+          }
+          case 42: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            visibleDeviceList_ = s;
+            break;
+          }
+          case 48: {
+
+            pollingActiveDelayUsecs_ = input.readInt32();
+            break;
+          }
+          case 56: {
+
+            pollingInactiveDelayMsecs_ = input.readInt32();
+            break;
+          }
+          case 64: {
+
+            forceGpuCompatible_ = input.readBool();
+            break;
+          }
+          case 74: {
+            org.tensorflow.proto.framework.GPUOptions.Experimental.Builder subBuilder = null;
+            if (experimental_ != null) {
+              subBuilder = experimental_.toBuilder();
+            }
+            experimental_ = input.readMessage(org.tensorflow.proto.framework.GPUOptions.Experimental.parser(), extensionRegistry);
+            if (subBuilder != null) {
+              subBuilder.mergeFrom(experimental_);
+              experimental_ = subBuilder.buildPartial();
+            }
+
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_GPUOptions_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_GPUOptions_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.framework.GPUOptions.class, org.tensorflow.proto.framework.GPUOptions.Builder.class);
+  }
+
+  public interface ExperimentalOrBuilder extends
+      // @@protoc_insertion_point(interface_extends:tensorflow.GPUOptions.Experimental)
+      com.google.protobuf.MessageOrBuilder {
+
+    /**
+     * <pre>
+     * The multi virtual device settings. If empty (not set), it will create
+     * single virtual device on each visible GPU, according to the settings
+     * in "visible_device_list" above. Otherwise, the number of elements in the
+     * list must be the same as the number of visible GPUs (after
+     * "visible_device_list" filtering if it is set), and the string represented
+     * device names (e.g. /device:GPU:&lt;id&gt;) will refer to the virtual
+     * devices and have the &lt;id&gt; field assigned sequentially starting from 0,
+     * according to the order they appear in this list and the "memory_limit"
+     * list inside each element. For example,
+     *   visible_device_list = "1,0"
+     *   virtual_devices { memory_limit: 1GB memory_limit: 2GB }
+     *   virtual_devices {}
+     * will create three virtual devices as:
+     *   /device:GPU:0 -&gt; visible GPU 1 with 1GB memory
+     *   /device:GPU:1 -&gt; visible GPU 1 with 2GB memory
+     *   /device:GPU:2 -&gt; visible GPU 0 with all available memory
+     * NOTE:
+     * 1. It's invalid to set both this and "per_process_gpu_memory_fraction"
+     *    at the same time.
+     * 2. Currently this setting is per-process, not per-session. Using
+     *    different settings in different sessions within same process will
+     *    result in undefined behavior.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.GPUOptions.Experimental.VirtualDevices virtual_devices = 1;</code>
+     */
+    java.util.List<org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevices> 
+        getVirtualDevicesList();
+    /**
+     * <pre>
+     * The multi virtual device settings. If empty (not set), it will create
+     * single virtual device on each visible GPU, according to the settings
+     * in "visible_device_list" above. Otherwise, the number of elements in the
+     * list must be the same as the number of visible GPUs (after
+     * "visible_device_list" filtering if it is set), and the string represented
+     * device names (e.g. /device:GPU:&lt;id&gt;) will refer to the virtual
+     * devices and have the &lt;id&gt; field assigned sequentially starting from 0,
+     * according to the order they appear in this list and the "memory_limit"
+     * list inside each element. For example,
+     *   visible_device_list = "1,0"
+     *   virtual_devices { memory_limit: 1GB memory_limit: 2GB }
+     *   virtual_devices {}
+     * will create three virtual devices as:
+     *   /device:GPU:0 -&gt; visible GPU 1 with 1GB memory
+     *   /device:GPU:1 -&gt; visible GPU 1 with 2GB memory
+     *   /device:GPU:2 -&gt; visible GPU 0 with all available memory
+     * NOTE:
+     * 1. It's invalid to set both this and "per_process_gpu_memory_fraction"
+     *    at the same time.
+     * 2. Currently this setting is per-process, not per-session. Using
+     *    different settings in different sessions within same process will
+     *    result in undefined behavior.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.GPUOptions.Experimental.VirtualDevices virtual_devices = 1;</code>
+     */
+    org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevices getVirtualDevices(int index);
+    /**
+     * <pre>
+     * The multi virtual device settings. If empty (not set), it will create
+     * single virtual device on each visible GPU, according to the settings
+     * in "visible_device_list" above. Otherwise, the number of elements in the
+     * list must be the same as the number of visible GPUs (after
+     * "visible_device_list" filtering if it is set), and the string represented
+     * device names (e.g. /device:GPU:&lt;id&gt;) will refer to the virtual
+     * devices and have the &lt;id&gt; field assigned sequentially starting from 0,
+     * according to the order they appear in this list and the "memory_limit"
+     * list inside each element. For example,
+     *   visible_device_list = "1,0"
+     *   virtual_devices { memory_limit: 1GB memory_limit: 2GB }
+     *   virtual_devices {}
+     * will create three virtual devices as:
+     *   /device:GPU:0 -&gt; visible GPU 1 with 1GB memory
+     *   /device:GPU:1 -&gt; visible GPU 1 with 2GB memory
+     *   /device:GPU:2 -&gt; visible GPU 0 with all available memory
+     * NOTE:
+     * 1. It's invalid to set both this and "per_process_gpu_memory_fraction"
+     *    at the same time.
+     * 2. Currently this setting is per-process, not per-session. Using
+     *    different settings in different sessions within same process will
+     *    result in undefined behavior.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.GPUOptions.Experimental.VirtualDevices virtual_devices = 1;</code>
+     */
+    int getVirtualDevicesCount();
+    /**
+     * <pre>
+     * The multi virtual device settings. If empty (not set), it will create
+     * single virtual device on each visible GPU, according to the settings
+     * in "visible_device_list" above. Otherwise, the number of elements in the
+     * list must be the same as the number of visible GPUs (after
+     * "visible_device_list" filtering if it is set), and the string represented
+     * device names (e.g. /device:GPU:&lt;id&gt;) will refer to the virtual
+     * devices and have the &lt;id&gt; field assigned sequentially starting from 0,
+     * according to the order they appear in this list and the "memory_limit"
+     * list inside each element. For example,
+     *   visible_device_list = "1,0"
+     *   virtual_devices { memory_limit: 1GB memory_limit: 2GB }
+     *   virtual_devices {}
+     * will create three virtual devices as:
+     *   /device:GPU:0 -&gt; visible GPU 1 with 1GB memory
+     *   /device:GPU:1 -&gt; visible GPU 1 with 2GB memory
+     *   /device:GPU:2 -&gt; visible GPU 0 with all available memory
+     * NOTE:
+     * 1. It's invalid to set both this and "per_process_gpu_memory_fraction"
+     *    at the same time.
+     * 2. Currently this setting is per-process, not per-session. Using
+     *    different settings in different sessions within same process will
+     *    result in undefined behavior.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.GPUOptions.Experimental.VirtualDevices virtual_devices = 1;</code>
+     */
+    java.util.List<? extends org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevicesOrBuilder> 
+        getVirtualDevicesOrBuilderList();
+    /**
+     * <pre>
+     * The multi virtual device settings. If empty (not set), it will create
+     * single virtual device on each visible GPU, according to the settings
+     * in "visible_device_list" above. Otherwise, the number of elements in the
+     * list must be the same as the number of visible GPUs (after
+     * "visible_device_list" filtering if it is set), and the string represented
+     * device names (e.g. /device:GPU:&lt;id&gt;) will refer to the virtual
+     * devices and have the &lt;id&gt; field assigned sequentially starting from 0,
+     * according to the order they appear in this list and the "memory_limit"
+     * list inside each element. For example,
+     *   visible_device_list = "1,0"
+     *   virtual_devices { memory_limit: 1GB memory_limit: 2GB }
+     *   virtual_devices {}
+     * will create three virtual devices as:
+     *   /device:GPU:0 -&gt; visible GPU 1 with 1GB memory
+     *   /device:GPU:1 -&gt; visible GPU 1 with 2GB memory
+     *   /device:GPU:2 -&gt; visible GPU 0 with all available memory
+     * NOTE:
+     * 1. It's invalid to set both this and "per_process_gpu_memory_fraction"
+     *    at the same time.
+     * 2. Currently this setting is per-process, not per-session. Using
+     *    different settings in different sessions within same process will
+     *    result in undefined behavior.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.GPUOptions.Experimental.VirtualDevices virtual_devices = 1;</code>
+     */
+    org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevicesOrBuilder getVirtualDevicesOrBuilder(
+        int index);
+
+    /**
+     * <pre>
+     * If true, uses CUDA unified memory for memory allocations. If
+     * per_process_gpu_memory_fraction option is greater than 1.0, then unified
+     * memory is used regardless of the value for this field. See comments for
+     * per_process_gpu_memory_fraction field for more details and requirements
+     * of the unified memory. This option is useful to oversubscribe memory if
+     * multiple processes are sharing a single GPU while individually using less
+     * than 1.0 per process memory fraction.
+     * </pre>
+     *
+     * <code>bool use_unified_memory = 2;</code>
+     */
+    boolean getUseUnifiedMemory();
+
+    /**
+     * <pre>
+     * If &gt; 1, the number of device-to-device copy streams to create
+     * for each GPUDevice.  Default value is 0, which is automatically
+     * converted to 1.
+     * </pre>
+     *
+     * <code>int32 num_dev_to_dev_copy_streams = 3;</code>
+     */
+    int getNumDevToDevCopyStreams();
+
+    /**
+     * <pre>
+     * If non-empty, defines a good GPU ring order on a single worker based on
+     * device interconnect.  This assumes that all workers have the same GPU
+     * topology.  Specify as a comma-separated string, e.g. "3,2,1,0,7,6,5,4".
+     * This ring order is used by the RingReducer implementation of
+     * CollectiveReduce, and serves as an override to automatic ring order
+     * generation in OrderTaskDeviceMap() during CollectiveParam resolution.
+     * </pre>
+     *
+     * <code>string collective_ring_order = 4;</code>
+     */
+    java.lang.String getCollectiveRingOrder();
+    /**
+     * <pre>
+     * If non-empty, defines a good GPU ring order on a single worker based on
+     * device interconnect.  This assumes that all workers have the same GPU
+     * topology.  Specify as a comma-separated string, e.g. "3,2,1,0,7,6,5,4".
+     * This ring order is used by the RingReducer implementation of
+     * CollectiveReduce, and serves as an override to automatic ring order
+     * generation in OrderTaskDeviceMap() during CollectiveParam resolution.
+     * </pre>
+     *
+     * <code>string collective_ring_order = 4;</code>
+     */
+    com.google.protobuf.ByteString
+        getCollectiveRingOrderBytes();
+
+    /**
+     * <pre>
+     * If true then extra work is done by GPUDevice and GPUBFCAllocator to
+     * keep track of when GPU memory is freed and when kernels actually
+     * complete so that we can know when a nominally free memory chunk
+     * is really not subject to pending use.
+     * </pre>
+     *
+     * <code>bool timestamped_allocator = 5;</code>
+     */
+    boolean getTimestampedAllocator();
+
+    /**
+     * <pre>
+     * Parameters for GPUKernelTracker.  By default no kernel tracking is done.
+     * Note that timestamped_allocator is only effective if some tracking is
+     * specified.
+     * If kernel_tracker_max_interval = n &gt; 0, then a tracking event
+     * is inserted after every n kernels without an event.
+     * </pre>
+     *
+     * <code>int32 kernel_tracker_max_interval = 7;</code>
+     */
+    int getKernelTrackerMaxInterval();
+
+    /**
+     * <pre>
+     * If kernel_tracker_max_bytes = n &gt; 0, then a tracking event is
+     * inserted after every series of kernels allocating a sum of
+     * memory &gt;= n.  If one kernel allocates b * n bytes, then one
+     * event will be inserted after it, but it will count as b against
+     * the pending limit.
+     * </pre>
+     *
+     * <code>int32 kernel_tracker_max_bytes = 8;</code>
+     */
+    int getKernelTrackerMaxBytes();
+
+    /**
+     * <pre>
+     * If kernel_tracker_max_pending &gt; 0 then no more than this many
+     * tracking events can be outstanding at a time.  An attempt to
+     * launch an additional kernel will stall until an event
+     * completes.
+     * </pre>
+     *
+     * <code>int32 kernel_tracker_max_pending = 9;</code>
+     */
+    int getKernelTrackerMaxPending();
+  }
+  /**
+   * Protobuf type {@code tensorflow.GPUOptions.Experimental}
+   */
+  public  static final class Experimental extends
+      com.google.protobuf.GeneratedMessageV3 implements
+      // @@protoc_insertion_point(message_implements:tensorflow.GPUOptions.Experimental)
+      ExperimentalOrBuilder {
+  private static final long serialVersionUID = 0L;
+    // Use Experimental.newBuilder() to construct.
+    private Experimental(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+      super(builder);
+    }
+    private Experimental() {
+      virtualDevices_ = java.util.Collections.emptyList();
+      collectiveRingOrder_ = "";
+    }
+
+    @java.lang.Override
+    @SuppressWarnings({"unused"})
+    protected java.lang.Object newInstance(
+        UnusedPrivateParameter unused) {
+      return new Experimental();
+    }
+
+    @java.lang.Override
+    public final com.google.protobuf.UnknownFieldSet
+    getUnknownFields() {
+      return this.unknownFields;
+    }
+    private Experimental(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      this();
+      if (extensionRegistry == null) {
+        throw new java.lang.NullPointerException();
+      }
+      int mutable_bitField0_ = 0;
+      com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+          com.google.protobuf.UnknownFieldSet.newBuilder();
+      try {
+        boolean done = false;
+        while (!done) {
+          int tag = input.readTag();
+          switch (tag) {
+            case 0:
+              done = true;
+              break;
+            case 10: {
+              if (!((mutable_bitField0_ & 0x00000001) != 0)) {
+                virtualDevices_ = new java.util.ArrayList<org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevices>();
+                mutable_bitField0_ |= 0x00000001;
+              }
+              virtualDevices_.add(
+                  input.readMessage(org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevices.parser(), extensionRegistry));
+              break;
+            }
+            case 16: {
+
+              useUnifiedMemory_ = input.readBool();
+              break;
+            }
+            case 24: {
+
+              numDevToDevCopyStreams_ = input.readInt32();
+              break;
+            }
+            case 34: {
+              java.lang.String s = input.readStringRequireUtf8();
+
+              collectiveRingOrder_ = s;
+              break;
+            }
+            case 40: {
+
+              timestampedAllocator_ = input.readBool();
+              break;
+            }
+            case 56: {
+
+              kernelTrackerMaxInterval_ = input.readInt32();
+              break;
+            }
+            case 64: {
+
+              kernelTrackerMaxBytes_ = input.readInt32();
+              break;
+            }
+            case 72: {
+
+              kernelTrackerMaxPending_ = input.readInt32();
+              break;
+            }
+            default: {
+              if (!parseUnknownField(
+                  input, unknownFields, extensionRegistry, tag)) {
+                done = true;
+              }
+              break;
+            }
+          }
+        }
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        throw e.setUnfinishedMessage(this);
+      } catch (java.io.IOException e) {
+        throw new com.google.protobuf.InvalidProtocolBufferException(
+            e).setUnfinishedMessage(this);
+      } finally {
+        if (((mutable_bitField0_ & 0x00000001) != 0)) {
+          virtualDevices_ = java.util.Collections.unmodifiableList(virtualDevices_);
+        }
+        this.unknownFields = unknownFields.build();
+        makeExtensionsImmutable();
+      }
+    }
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_GPUOptions_Experimental_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_GPUOptions_Experimental_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.GPUOptions.Experimental.class, org.tensorflow.proto.framework.GPUOptions.Experimental.Builder.class);
+    }
+
+    public interface VirtualDevicesOrBuilder extends
+        // @@protoc_insertion_point(interface_extends:tensorflow.GPUOptions.Experimental.VirtualDevices)
+        com.google.protobuf.MessageOrBuilder {
+
+      /**
+       * <pre>
+       * Per "virtual" device memory limit, in MB. The number of elements in
+       * the list is the number of virtual devices to create on the
+       * corresponding visible GPU (see "virtual_devices" below).
+       * If empty, it will create single virtual device taking all available
+       * memory from the device.
+       * For the concept of "visible" and "virtual" GPU, see the comments for
+       * "visible_device_list" above for more information.
+       * </pre>
+       *
+       * <code>repeated float memory_limit_mb = 1;</code>
+       */
+      java.util.List<java.lang.Float> getMemoryLimitMbList();
+      /**
+       * <pre>
+       * Per "virtual" device memory limit, in MB. The number of elements in
+       * the list is the number of virtual devices to create on the
+       * corresponding visible GPU (see "virtual_devices" below).
+       * If empty, it will create single virtual device taking all available
+       * memory from the device.
+       * For the concept of "visible" and "virtual" GPU, see the comments for
+       * "visible_device_list" above for more information.
+       * </pre>
+       *
+       * <code>repeated float memory_limit_mb = 1;</code>
+       */
+      int getMemoryLimitMbCount();
+      /**
+       * <pre>
+       * Per "virtual" device memory limit, in MB. The number of elements in
+       * the list is the number of virtual devices to create on the
+       * corresponding visible GPU (see "virtual_devices" below).
+       * If empty, it will create single virtual device taking all available
+       * memory from the device.
+       * For the concept of "visible" and "virtual" GPU, see the comments for
+       * "visible_device_list" above for more information.
+       * </pre>
+       *
+       * <code>repeated float memory_limit_mb = 1;</code>
+       */
+      float getMemoryLimitMb(int index);
+
+      /**
+       * <pre>
+       * Priority values to use with the virtual devices. Use the cuda function
+       * cudaDeviceGetStreamPriorityRange to query for valid range of values for
+       * priority.
+       * On a P4000 GPU with cuda 10.1, the priority range reported was 0 for
+       * least priority and -1 for greatest priority.
+       * If this field is not specified, then the virtual devices will be
+       * created with the default. If this field has values set, then the size
+       * of this must match with the above memory_limit_mb.
+       * </pre>
+       *
+       * <code>repeated int32 priority = 2;</code>
+       */
+      java.util.List<java.lang.Integer> getPriorityList();
+      /**
+       * <pre>
+       * Priority values to use with the virtual devices. Use the cuda function
+       * cudaDeviceGetStreamPriorityRange to query for valid range of values for
+       * priority.
+       * On a P4000 GPU with cuda 10.1, the priority range reported was 0 for
+       * least priority and -1 for greatest priority.
+       * If this field is not specified, then the virtual devices will be
+       * created with the default. If this field has values set, then the size
+       * of this must match with the above memory_limit_mb.
+       * </pre>
+       *
+       * <code>repeated int32 priority = 2;</code>
+       */
+      int getPriorityCount();
+      /**
+       * <pre>
+       * Priority values to use with the virtual devices. Use the cuda function
+       * cudaDeviceGetStreamPriorityRange to query for valid range of values for
+       * priority.
+       * On a P4000 GPU with cuda 10.1, the priority range reported was 0 for
+       * least priority and -1 for greatest priority.
+       * If this field is not specified, then the virtual devices will be
+       * created with the default. If this field has values set, then the size
+       * of this must match with the above memory_limit_mb.
+       * </pre>
+       *
+       * <code>repeated int32 priority = 2;</code>
+       */
+      int getPriority(int index);
+    }
+    /**
+     * <pre>
+     * Configuration for breaking down a visible GPU into multiple "virtual"
+     * devices.
+     * </pre>
+     *
+     * Protobuf type {@code tensorflow.GPUOptions.Experimental.VirtualDevices}
+     */
+    public  static final class VirtualDevices extends
+        com.google.protobuf.GeneratedMessageV3 implements
+        // @@protoc_insertion_point(message_implements:tensorflow.GPUOptions.Experimental.VirtualDevices)
+        VirtualDevicesOrBuilder {
+    private static final long serialVersionUID = 0L;
+      // Use VirtualDevices.newBuilder() to construct.
+      private VirtualDevices(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+        super(builder);
+      }
+      private VirtualDevices() {
+        memoryLimitMb_ = emptyFloatList();
+        priority_ = emptyIntList();
+      }
+
+      @java.lang.Override
+      @SuppressWarnings({"unused"})
+      protected java.lang.Object newInstance(
+          UnusedPrivateParameter unused) {
+        return new VirtualDevices();
+      }
+
+      @java.lang.Override
+      public final com.google.protobuf.UnknownFieldSet
+      getUnknownFields() {
+        return this.unknownFields;
+      }
+      private VirtualDevices(
+          com.google.protobuf.CodedInputStream input,
+          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+          throws com.google.protobuf.InvalidProtocolBufferException {
+        this();
+        if (extensionRegistry == null) {
+          throw new java.lang.NullPointerException();
+        }
+        int mutable_bitField0_ = 0;
+        com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+            com.google.protobuf.UnknownFieldSet.newBuilder();
+        try {
+          boolean done = false;
+          while (!done) {
+            int tag = input.readTag();
+            switch (tag) {
+              case 0:
+                done = true;
+                break;
+              case 13: {
+                if (!((mutable_bitField0_ & 0x00000001) != 0)) {
+                  memoryLimitMb_ = newFloatList();
+                  mutable_bitField0_ |= 0x00000001;
+                }
+                memoryLimitMb_.addFloat(input.readFloat());
+                break;
+              }
+              case 10: {
+                int length = input.readRawVarint32();
+                int limit = input.pushLimit(length);
+                if (!((mutable_bitField0_ & 0x00000001) != 0) && input.getBytesUntilLimit() > 0) {
+                  memoryLimitMb_ = newFloatList();
+                  mutable_bitField0_ |= 0x00000001;
+                }
+                while (input.getBytesUntilLimit() > 0) {
+                  memoryLimitMb_.addFloat(input.readFloat());
+                }
+                input.popLimit(limit);
+                break;
+              }
+              case 16: {
+                if (!((mutable_bitField0_ & 0x00000002) != 0)) {
+                  priority_ = newIntList();
+                  mutable_bitField0_ |= 0x00000002;
+                }
+                priority_.addInt(input.readInt32());
+                break;
+              }
+              case 18: {
+                int length = input.readRawVarint32();
+                int limit = input.pushLimit(length);
+                if (!((mutable_bitField0_ & 0x00000002) != 0) && input.getBytesUntilLimit() > 0) {
+                  priority_ = newIntList();
+                  mutable_bitField0_ |= 0x00000002;
+                }
+                while (input.getBytesUntilLimit() > 0) {
+                  priority_.addInt(input.readInt32());
+                }
+                input.popLimit(limit);
+                break;
+              }
+              default: {
+                if (!parseUnknownField(
+                    input, unknownFields, extensionRegistry, tag)) {
+                  done = true;
+                }
+                break;
+              }
+            }
+          }
+        } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+          throw e.setUnfinishedMessage(this);
+        } catch (java.io.IOException e) {
+          throw new com.google.protobuf.InvalidProtocolBufferException(
+              e).setUnfinishedMessage(this);
+        } finally {
+          if (((mutable_bitField0_ & 0x00000001) != 0)) {
+            memoryLimitMb_.makeImmutable(); // C
+          }
+          if (((mutable_bitField0_ & 0x00000002) != 0)) {
+            priority_.makeImmutable(); // C
+          }
+          this.unknownFields = unknownFields.build();
+          makeExtensionsImmutable();
+        }
+      }
+      public static final com.google.protobuf.Descriptors.Descriptor
+          getDescriptor() {
+        return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_GPUOptions_Experimental_VirtualDevices_descriptor;
+      }
+
+      @java.lang.Override
+      protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+          internalGetFieldAccessorTable() {
+        return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_GPUOptions_Experimental_VirtualDevices_fieldAccessorTable
+            .ensureFieldAccessorsInitialized(
+                org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevices.class, org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevices.Builder.class);
+      }
+
+      public static final int MEMORY_LIMIT_MB_FIELD_NUMBER = 1;
+      private com.google.protobuf.Internal.FloatList memoryLimitMb_;
+      /**
+       * <pre>
+       * Per "virtual" device memory limit, in MB. The number of elements in
+       * the list is the number of virtual devices to create on the
+       * corresponding visible GPU (see "virtual_devices" below).
+       * If empty, it will create single virtual device taking all available
+       * memory from the device.
+       * For the concept of "visible" and "virtual" GPU, see the comments for
+       * "visible_device_list" above for more information.
+       * </pre>
+       *
+       * <code>repeated float memory_limit_mb = 1;</code>
+       */
+      public java.util.List<java.lang.Float>
+          getMemoryLimitMbList() {
+        return memoryLimitMb_;
+      }
+      /**
+       * <pre>
+       * Per "virtual" device memory limit, in MB. The number of elements in
+       * the list is the number of virtual devices to create on the
+       * corresponding visible GPU (see "virtual_devices" below).
+       * If empty, it will create single virtual device taking all available
+       * memory from the device.
+       * For the concept of "visible" and "virtual" GPU, see the comments for
+       * "visible_device_list" above for more information.
+       * </pre>
+       *
+       * <code>repeated float memory_limit_mb = 1;</code>
+       */
+      public int getMemoryLimitMbCount() {
+        return memoryLimitMb_.size();
+      }
+      /**
+       * <pre>
+       * Per "virtual" device memory limit, in MB. The number of elements in
+       * the list is the number of virtual devices to create on the
+       * corresponding visible GPU (see "virtual_devices" below).
+       * If empty, it will create single virtual device taking all available
+       * memory from the device.
+       * For the concept of "visible" and "virtual" GPU, see the comments for
+       * "visible_device_list" above for more information.
+       * </pre>
+       *
+       * <code>repeated float memory_limit_mb = 1;</code>
+       */
+      public float getMemoryLimitMb(int index) {
+        return memoryLimitMb_.getFloat(index);
+      }
+      private int memoryLimitMbMemoizedSerializedSize = -1;
+
+      public static final int PRIORITY_FIELD_NUMBER = 2;
+      private com.google.protobuf.Internal.IntList priority_;
+      /**
+       * <pre>
+       * Priority values to use with the virtual devices. Use the cuda function
+       * cudaDeviceGetStreamPriorityRange to query for valid range of values for
+       * priority.
+       * On a P4000 GPU with cuda 10.1, the priority range reported was 0 for
+       * least priority and -1 for greatest priority.
+       * If this field is not specified, then the virtual devices will be
+       * created with the default. If this field has values set, then the size
+       * of this must match with the above memory_limit_mb.
+       * </pre>
+       *
+       * <code>repeated int32 priority = 2;</code>
+       */
+      public java.util.List<java.lang.Integer>
+          getPriorityList() {
+        return priority_;
+      }
+      /**
+       * <pre>
+       * Priority values to use with the virtual devices. Use the cuda function
+       * cudaDeviceGetStreamPriorityRange to query for valid range of values for
+       * priority.
+       * On a P4000 GPU with cuda 10.1, the priority range reported was 0 for
+       * least priority and -1 for greatest priority.
+       * If this field is not specified, then the virtual devices will be
+       * created with the default. If this field has values set, then the size
+       * of this must match with the above memory_limit_mb.
+       * </pre>
+       *
+       * <code>repeated int32 priority = 2;</code>
+       */
+      public int getPriorityCount() {
+        return priority_.size();
+      }
+      /**
+       * <pre>
+       * Priority values to use with the virtual devices. Use the cuda function
+       * cudaDeviceGetStreamPriorityRange to query for valid range of values for
+       * priority.
+       * On a P4000 GPU with cuda 10.1, the priority range reported was 0 for
+       * least priority and -1 for greatest priority.
+       * If this field is not specified, then the virtual devices will be
+       * created with the default. If this field has values set, then the size
+       * of this must match with the above memory_limit_mb.
+       * </pre>
+       *
+       * <code>repeated int32 priority = 2;</code>
+       */
+      public int getPriority(int index) {
+        return priority_.getInt(index);
+      }
+      private int priorityMemoizedSerializedSize = -1;
+
+      private byte memoizedIsInitialized = -1;
+      @java.lang.Override
+      public final boolean isInitialized() {
+        byte isInitialized = memoizedIsInitialized;
+        if (isInitialized == 1) return true;
+        if (isInitialized == 0) return false;
+
+        memoizedIsInitialized = 1;
+        return true;
+      }
+
+      @java.lang.Override
+      public void writeTo(com.google.protobuf.CodedOutputStream output)
+                          throws java.io.IOException {
+        getSerializedSize();
+        if (getMemoryLimitMbList().size() > 0) {
+          output.writeUInt32NoTag(10);
+          output.writeUInt32NoTag(memoryLimitMbMemoizedSerializedSize);
+        }
+        for (int i = 0; i < memoryLimitMb_.size(); i++) {
+          output.writeFloatNoTag(memoryLimitMb_.getFloat(i));
+        }
+        if (getPriorityList().size() > 0) {
+          output.writeUInt32NoTag(18);
+          output.writeUInt32NoTag(priorityMemoizedSerializedSize);
+        }
+        for (int i = 0; i < priority_.size(); i++) {
+          output.writeInt32NoTag(priority_.getInt(i));
+        }
+        unknownFields.writeTo(output);
+      }
+
+      @java.lang.Override
+      public int getSerializedSize() {
+        int size = memoizedSize;
+        if (size != -1) return size;
+
+        size = 0;
+        {
+          int dataSize = 0;
+          dataSize = 4 * getMemoryLimitMbList().size();
+          size += dataSize;
+          if (!getMemoryLimitMbList().isEmpty()) {
+            size += 1;
+            size += com.google.protobuf.CodedOutputStream
+                .computeInt32SizeNoTag(dataSize);
+          }
+          memoryLimitMbMemoizedSerializedSize = dataSize;
+        }
+        {
+          int dataSize = 0;
+          for (int i = 0; i < priority_.size(); i++) {
+            dataSize += com.google.protobuf.CodedOutputStream
+              .computeInt32SizeNoTag(priority_.getInt(i));
+          }
+          size += dataSize;
+          if (!getPriorityList().isEmpty()) {
+            size += 1;
+            size += com.google.protobuf.CodedOutputStream
+                .computeInt32SizeNoTag(dataSize);
+          }
+          priorityMemoizedSerializedSize = dataSize;
+        }
+        size += unknownFields.getSerializedSize();
+        memoizedSize = size;
+        return size;
+      }
+
+      @java.lang.Override
+      public boolean equals(final java.lang.Object obj) {
+        if (obj == this) {
+         return true;
+        }
+        if (!(obj instanceof org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevices)) {
+          return super.equals(obj);
+        }
+        org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevices other = (org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevices) obj;
+
+        if (!getMemoryLimitMbList()
+            .equals(other.getMemoryLimitMbList())) return false;
+        if (!getPriorityList()
+            .equals(other.getPriorityList())) return false;
+        if (!unknownFields.equals(other.unknownFields)) return false;
+        return true;
+      }
+
+      @java.lang.Override
+      public int hashCode() {
+        if (memoizedHashCode != 0) {
+          return memoizedHashCode;
+        }
+        int hash = 41;
+        hash = (19 * hash) + getDescriptor().hashCode();
+        if (getMemoryLimitMbCount() > 0) {
+          hash = (37 * hash) + MEMORY_LIMIT_MB_FIELD_NUMBER;
+          hash = (53 * hash) + getMemoryLimitMbList().hashCode();
+        }
+        if (getPriorityCount() > 0) {
+          hash = (37 * hash) + PRIORITY_FIELD_NUMBER;
+          hash = (53 * hash) + getPriorityList().hashCode();
+        }
+        hash = (29 * hash) + unknownFields.hashCode();
+        memoizedHashCode = hash;
+        return hash;
+      }
+
+      public static org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevices parseFrom(
+          java.nio.ByteBuffer data)
+          throws com.google.protobuf.InvalidProtocolBufferException {
+        return PARSER.parseFrom(data);
+      }
+      public static org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevices parseFrom(
+          java.nio.ByteBuffer data,
+          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+          throws com.google.protobuf.InvalidProtocolBufferException {
+        return PARSER.parseFrom(data, extensionRegistry);
+      }
+      public static org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevices parseFrom(
+          com.google.protobuf.ByteString data)
+          throws com.google.protobuf.InvalidProtocolBufferException {
+        return PARSER.parseFrom(data);
+      }
+      public static org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevices parseFrom(
+          com.google.protobuf.ByteString data,
+          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+          throws com.google.protobuf.InvalidProtocolBufferException {
+        return PARSER.parseFrom(data, extensionRegistry);
+      }
+      public static org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevices parseFrom(byte[] data)
+          throws com.google.protobuf.InvalidProtocolBufferException {
+        return PARSER.parseFrom(data);
+      }
+      public static org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevices parseFrom(
+          byte[] data,
+          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+          throws com.google.protobuf.InvalidProtocolBufferException {
+        return PARSER.parseFrom(data, extensionRegistry);
+      }
+      public static org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevices parseFrom(java.io.InputStream input)
+          throws java.io.IOException {
+        return com.google.protobuf.GeneratedMessageV3
+            .parseWithIOException(PARSER, input);
+      }
+      public static org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevices parseFrom(
+          java.io.InputStream input,
+          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+          throws java.io.IOException {
+        return com.google.protobuf.GeneratedMessageV3
+            .parseWithIOException(PARSER, input, extensionRegistry);
+      }
+      public static org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevices parseDelimitedFrom(java.io.InputStream input)
+          throws java.io.IOException {
+        return com.google.protobuf.GeneratedMessageV3
+            .parseDelimitedWithIOException(PARSER, input);
+      }
+      public static org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevices parseDelimitedFrom(
+          java.io.InputStream input,
+          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+          throws java.io.IOException {
+        return com.google.protobuf.GeneratedMessageV3
+            .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+      }
+      public static org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevices parseFrom(
+          com.google.protobuf.CodedInputStream input)
+          throws java.io.IOException {
+        return com.google.protobuf.GeneratedMessageV3
+            .parseWithIOException(PARSER, input);
+      }
+      public static org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevices parseFrom(
+          com.google.protobuf.CodedInputStream input,
+          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+          throws java.io.IOException {
+        return com.google.protobuf.GeneratedMessageV3
+            .parseWithIOException(PARSER, input, extensionRegistry);
+      }
+
+      @java.lang.Override
+      public Builder newBuilderForType() { return newBuilder(); }
+      public static Builder newBuilder() {
+        return DEFAULT_INSTANCE.toBuilder();
+      }
+      public static Builder newBuilder(org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevices prototype) {
+        return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+      }
+      @java.lang.Override
+      public Builder toBuilder() {
+        return this == DEFAULT_INSTANCE
+            ? new Builder() : new Builder().mergeFrom(this);
+      }
+
+      @java.lang.Override
+      protected Builder newBuilderForType(
+          com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+        Builder builder = new Builder(parent);
+        return builder;
+      }
+      /**
+       * <pre>
+       * Configuration for breaking down a visible GPU into multiple "virtual"
+       * devices.
+       * </pre>
+       *
+       * Protobuf type {@code tensorflow.GPUOptions.Experimental.VirtualDevices}
+       */
+      public static final class Builder extends
+          com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+          // @@protoc_insertion_point(builder_implements:tensorflow.GPUOptions.Experimental.VirtualDevices)
+          org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevicesOrBuilder {
+        public static final com.google.protobuf.Descriptors.Descriptor
+            getDescriptor() {
+          return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_GPUOptions_Experimental_VirtualDevices_descriptor;
+        }
+
+        @java.lang.Override
+        protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+            internalGetFieldAccessorTable() {
+          return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_GPUOptions_Experimental_VirtualDevices_fieldAccessorTable
+              .ensureFieldAccessorsInitialized(
+                  org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevices.class, org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevices.Builder.class);
+        }
+
+        // Construct using org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevices.newBuilder()
+        private Builder() {
+          maybeForceBuilderInitialization();
+        }
+
+        private Builder(
+            com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+          super(parent);
+          maybeForceBuilderInitialization();
+        }
+        private void maybeForceBuilderInitialization() {
+          if (com.google.protobuf.GeneratedMessageV3
+                  .alwaysUseFieldBuilders) {
+          }
+        }
+        @java.lang.Override
+        public Builder clear() {
+          super.clear();
+          memoryLimitMb_ = emptyFloatList();
+          bitField0_ = (bitField0_ & ~0x00000001);
+          priority_ = emptyIntList();
+          bitField0_ = (bitField0_ & ~0x00000002);
+          return this;
+        }
+
+        @java.lang.Override
+        public com.google.protobuf.Descriptors.Descriptor
+            getDescriptorForType() {
+          return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_GPUOptions_Experimental_VirtualDevices_descriptor;
+        }
+
+        @java.lang.Override
+        public org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevices getDefaultInstanceForType() {
+          return org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevices.getDefaultInstance();
+        }
+
+        @java.lang.Override
+        public org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevices build() {
+          org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevices result = buildPartial();
+          if (!result.isInitialized()) {
+            throw newUninitializedMessageException(result);
+          }
+          return result;
+        }
+
+        @java.lang.Override
+        public org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevices buildPartial() {
+          org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevices result = new org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevices(this);
+          int from_bitField0_ = bitField0_;
+          if (((bitField0_ & 0x00000001) != 0)) {
+            memoryLimitMb_.makeImmutable();
+            bitField0_ = (bitField0_ & ~0x00000001);
+          }
+          result.memoryLimitMb_ = memoryLimitMb_;
+          if (((bitField0_ & 0x00000002) != 0)) {
+            priority_.makeImmutable();
+            bitField0_ = (bitField0_ & ~0x00000002);
+          }
+          result.priority_ = priority_;
+          onBuilt();
+          return result;
+        }
+
+        @java.lang.Override
+        public Builder clone() {
+          return super.clone();
+        }
+        @java.lang.Override
+        public Builder setField(
+            com.google.protobuf.Descriptors.FieldDescriptor field,
+            java.lang.Object value) {
+          return super.setField(field, value);
+        }
+        @java.lang.Override
+        public Builder clearField(
+            com.google.protobuf.Descriptors.FieldDescriptor field) {
+          return super.clearField(field);
+        }
+        @java.lang.Override
+        public Builder clearOneof(
+            com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+          return super.clearOneof(oneof);
+        }
+        @java.lang.Override
+        public Builder setRepeatedField(
+            com.google.protobuf.Descriptors.FieldDescriptor field,
+            int index, java.lang.Object value) {
+          return super.setRepeatedField(field, index, value);
+        }
+        @java.lang.Override
+        public Builder addRepeatedField(
+            com.google.protobuf.Descriptors.FieldDescriptor field,
+            java.lang.Object value) {
+          return super.addRepeatedField(field, value);
+        }
+        @java.lang.Override
+        public Builder mergeFrom(com.google.protobuf.Message other) {
+          if (other instanceof org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevices) {
+            return mergeFrom((org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevices)other);
+          } else {
+            super.mergeFrom(other);
+            return this;
+          }
+        }
+
+        public Builder mergeFrom(org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevices other) {
+          if (other == org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevices.getDefaultInstance()) return this;
+          if (!other.memoryLimitMb_.isEmpty()) {
+            if (memoryLimitMb_.isEmpty()) {
+              memoryLimitMb_ = other.memoryLimitMb_;
+              bitField0_ = (bitField0_ & ~0x00000001);
+            } else {
+              ensureMemoryLimitMbIsMutable();
+              memoryLimitMb_.addAll(other.memoryLimitMb_);
+            }
+            onChanged();
+          }
+          if (!other.priority_.isEmpty()) {
+            if (priority_.isEmpty()) {
+              priority_ = other.priority_;
+              bitField0_ = (bitField0_ & ~0x00000002);
+            } else {
+              ensurePriorityIsMutable();
+              priority_.addAll(other.priority_);
+            }
+            onChanged();
+          }
+          this.mergeUnknownFields(other.unknownFields);
+          onChanged();
+          return this;
+        }
+
+        @java.lang.Override
+        public final boolean isInitialized() {
+          return true;
+        }
+
+        @java.lang.Override
+        public Builder mergeFrom(
+            com.google.protobuf.CodedInputStream input,
+            com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+            throws java.io.IOException {
+          org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevices parsedMessage = null;
+          try {
+            parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+          } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+            parsedMessage = (org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevices) e.getUnfinishedMessage();
+            throw e.unwrapIOException();
+          } finally {
+            if (parsedMessage != null) {
+              mergeFrom(parsedMessage);
+            }
+          }
+          return this;
+        }
+        private int bitField0_;
+
+        private com.google.protobuf.Internal.FloatList memoryLimitMb_ = emptyFloatList();
+        private void ensureMemoryLimitMbIsMutable() {
+          if (!((bitField0_ & 0x00000001) != 0)) {
+            memoryLimitMb_ = mutableCopy(memoryLimitMb_);
+            bitField0_ |= 0x00000001;
+           }
+        }
+        /**
+         * <pre>
+         * Per "virtual" device memory limit, in MB. The number of elements in
+         * the list is the number of virtual devices to create on the
+         * corresponding visible GPU (see "virtual_devices" below).
+         * If empty, it will create single virtual device taking all available
+         * memory from the device.
+         * For the concept of "visible" and "virtual" GPU, see the comments for
+         * "visible_device_list" above for more information.
+         * </pre>
+         *
+         * <code>repeated float memory_limit_mb = 1;</code>
+         */
+        public java.util.List<java.lang.Float>
+            getMemoryLimitMbList() {
+          return ((bitField0_ & 0x00000001) != 0) ?
+                   java.util.Collections.unmodifiableList(memoryLimitMb_) : memoryLimitMb_;
+        }
+        /**
+         * <pre>
+         * Per "virtual" device memory limit, in MB. The number of elements in
+         * the list is the number of virtual devices to create on the
+         * corresponding visible GPU (see "virtual_devices" below).
+         * If empty, it will create single virtual device taking all available
+         * memory from the device.
+         * For the concept of "visible" and "virtual" GPU, see the comments for
+         * "visible_device_list" above for more information.
+         * </pre>
+         *
+         * <code>repeated float memory_limit_mb = 1;</code>
+         */
+        public int getMemoryLimitMbCount() {
+          return memoryLimitMb_.size();
+        }
+        /**
+         * <pre>
+         * Per "virtual" device memory limit, in MB. The number of elements in
+         * the list is the number of virtual devices to create on the
+         * corresponding visible GPU (see "virtual_devices" below).
+         * If empty, it will create single virtual device taking all available
+         * memory from the device.
+         * For the concept of "visible" and "virtual" GPU, see the comments for
+         * "visible_device_list" above for more information.
+         * </pre>
+         *
+         * <code>repeated float memory_limit_mb = 1;</code>
+         */
+        public float getMemoryLimitMb(int index) {
+          return memoryLimitMb_.getFloat(index);
+        }
+        /**
+         * <pre>
+         * Per "virtual" device memory limit, in MB. The number of elements in
+         * the list is the number of virtual devices to create on the
+         * corresponding visible GPU (see "virtual_devices" below).
+         * If empty, it will create single virtual device taking all available
+         * memory from the device.
+         * For the concept of "visible" and "virtual" GPU, see the comments for
+         * "visible_device_list" above for more information.
+         * </pre>
+         *
+         * <code>repeated float memory_limit_mb = 1;</code>
+         */
+        public Builder setMemoryLimitMb(
+            int index, float value) {
+          ensureMemoryLimitMbIsMutable();
+          memoryLimitMb_.setFloat(index, value);
+          onChanged();
+          return this;
+        }
+        /**
+         * <pre>
+         * Per "virtual" device memory limit, in MB. The number of elements in
+         * the list is the number of virtual devices to create on the
+         * corresponding visible GPU (see "virtual_devices" below).
+         * If empty, it will create single virtual device taking all available
+         * memory from the device.
+         * For the concept of "visible" and "virtual" GPU, see the comments for
+         * "visible_device_list" above for more information.
+         * </pre>
+         *
+         * <code>repeated float memory_limit_mb = 1;</code>
+         */
+        public Builder addMemoryLimitMb(float value) {
+          ensureMemoryLimitMbIsMutable();
+          memoryLimitMb_.addFloat(value);
+          onChanged();
+          return this;
+        }
+        /**
+         * <pre>
+         * Per "virtual" device memory limit, in MB. The number of elements in
+         * the list is the number of virtual devices to create on the
+         * corresponding visible GPU (see "virtual_devices" below).
+         * If empty, it will create single virtual device taking all available
+         * memory from the device.
+         * For the concept of "visible" and "virtual" GPU, see the comments for
+         * "visible_device_list" above for more information.
+         * </pre>
+         *
+         * <code>repeated float memory_limit_mb = 1;</code>
+         */
+        public Builder addAllMemoryLimitMb(
+            java.lang.Iterable<? extends java.lang.Float> values) {
+          ensureMemoryLimitMbIsMutable();
+          com.google.protobuf.AbstractMessageLite.Builder.addAll(
+              values, memoryLimitMb_);
+          onChanged();
+          return this;
+        }
+        /**
+         * <pre>
+         * Per "virtual" device memory limit, in MB. The number of elements in
+         * the list is the number of virtual devices to create on the
+         * corresponding visible GPU (see "virtual_devices" below).
+         * If empty, it will create single virtual device taking all available
+         * memory from the device.
+         * For the concept of "visible" and "virtual" GPU, see the comments for
+         * "visible_device_list" above for more information.
+         * </pre>
+         *
+         * <code>repeated float memory_limit_mb = 1;</code>
+         */
+        public Builder clearMemoryLimitMb() {
+          memoryLimitMb_ = emptyFloatList();
+          bitField0_ = (bitField0_ & ~0x00000001);
+          onChanged();
+          return this;
+        }
+
+        private com.google.protobuf.Internal.IntList priority_ = emptyIntList();
+        private void ensurePriorityIsMutable() {
+          if (!((bitField0_ & 0x00000002) != 0)) {
+            priority_ = mutableCopy(priority_);
+            bitField0_ |= 0x00000002;
+           }
+        }
+        /**
+         * <pre>
+         * Priority values to use with the virtual devices. Use the cuda function
+         * cudaDeviceGetStreamPriorityRange to query for valid range of values for
+         * priority.
+         * On a P4000 GPU with cuda 10.1, the priority range reported was 0 for
+         * least priority and -1 for greatest priority.
+         * If this field is not specified, then the virtual devices will be
+         * created with the default. If this field has values set, then the size
+         * of this must match with the above memory_limit_mb.
+         * </pre>
+         *
+         * <code>repeated int32 priority = 2;</code>
+         */
+        public java.util.List<java.lang.Integer>
+            getPriorityList() {
+          return ((bitField0_ & 0x00000002) != 0) ?
+                   java.util.Collections.unmodifiableList(priority_) : priority_;
+        }
+        /**
+         * <pre>
+         * Priority values to use with the virtual devices. Use the cuda function
+         * cudaDeviceGetStreamPriorityRange to query for valid range of values for
+         * priority.
+         * On a P4000 GPU with cuda 10.1, the priority range reported was 0 for
+         * least priority and -1 for greatest priority.
+         * If this field is not specified, then the virtual devices will be
+         * created with the default. If this field has values set, then the size
+         * of this must match with the above memory_limit_mb.
+         * </pre>
+         *
+         * <code>repeated int32 priority = 2;</code>
+         */
+        public int getPriorityCount() {
+          return priority_.size();
+        }
+        /**
+         * <pre>
+         * Priority values to use with the virtual devices. Use the cuda function
+         * cudaDeviceGetStreamPriorityRange to query for valid range of values for
+         * priority.
+         * On a P4000 GPU with cuda 10.1, the priority range reported was 0 for
+         * least priority and -1 for greatest priority.
+         * If this field is not specified, then the virtual devices will be
+         * created with the default. If this field has values set, then the size
+         * of this must match with the above memory_limit_mb.
+         * </pre>
+         *
+         * <code>repeated int32 priority = 2;</code>
+         */
+        public int getPriority(int index) {
+          return priority_.getInt(index);
+        }
+        /**
+         * <pre>
+         * Priority values to use with the virtual devices. Use the cuda function
+         * cudaDeviceGetStreamPriorityRange to query for valid range of values for
+         * priority.
+         * On a P4000 GPU with cuda 10.1, the priority range reported was 0 for
+         * least priority and -1 for greatest priority.
+         * If this field is not specified, then the virtual devices will be
+         * created with the default. If this field has values set, then the size
+         * of this must match with the above memory_limit_mb.
+         * </pre>
+         *
+         * <code>repeated int32 priority = 2;</code>
+         */
+        public Builder setPriority(
+            int index, int value) {
+          ensurePriorityIsMutable();
+          priority_.setInt(index, value);
+          onChanged();
+          return this;
+        }
+        /**
+         * <pre>
+         * Priority values to use with the virtual devices. Use the cuda function
+         * cudaDeviceGetStreamPriorityRange to query for valid range of values for
+         * priority.
+         * On a P4000 GPU with cuda 10.1, the priority range reported was 0 for
+         * least priority and -1 for greatest priority.
+         * If this field is not specified, then the virtual devices will be
+         * created with the default. If this field has values set, then the size
+         * of this must match with the above memory_limit_mb.
+         * </pre>
+         *
+         * <code>repeated int32 priority = 2;</code>
+         */
+        public Builder addPriority(int value) {
+          ensurePriorityIsMutable();
+          priority_.addInt(value);
+          onChanged();
+          return this;
+        }
+        /**
+         * <pre>
+         * Priority values to use with the virtual devices. Use the cuda function
+         * cudaDeviceGetStreamPriorityRange to query for valid range of values for
+         * priority.
+         * On a P4000 GPU with cuda 10.1, the priority range reported was 0 for
+         * least priority and -1 for greatest priority.
+         * If this field is not specified, then the virtual devices will be
+         * created with the default. If this field has values set, then the size
+         * of this must match with the above memory_limit_mb.
+         * </pre>
+         *
+         * <code>repeated int32 priority = 2;</code>
+         */
+        public Builder addAllPriority(
+            java.lang.Iterable<? extends java.lang.Integer> values) {
+          ensurePriorityIsMutable();
+          com.google.protobuf.AbstractMessageLite.Builder.addAll(
+              values, priority_);
+          onChanged();
+          return this;
+        }
+        /**
+         * <pre>
+         * Priority values to use with the virtual devices. Use the cuda function
+         * cudaDeviceGetStreamPriorityRange to query for valid range of values for
+         * priority.
+         * On a P4000 GPU with cuda 10.1, the priority range reported was 0 for
+         * least priority and -1 for greatest priority.
+         * If this field is not specified, then the virtual devices will be
+         * created with the default. If this field has values set, then the size
+         * of this must match with the above memory_limit_mb.
+         * </pre>
+         *
+         * <code>repeated int32 priority = 2;</code>
+         */
+        public Builder clearPriority() {
+          priority_ = emptyIntList();
+          bitField0_ = (bitField0_ & ~0x00000002);
+          onChanged();
+          return this;
+        }
+        @java.lang.Override
+        public final Builder setUnknownFields(
+            final com.google.protobuf.UnknownFieldSet unknownFields) {
+          return super.setUnknownFields(unknownFields);
+        }
+
+        @java.lang.Override
+        public final Builder mergeUnknownFields(
+            final com.google.protobuf.UnknownFieldSet unknownFields) {
+          return super.mergeUnknownFields(unknownFields);
+        }
+
+
+        // @@protoc_insertion_point(builder_scope:tensorflow.GPUOptions.Experimental.VirtualDevices)
+      }
+
+      // @@protoc_insertion_point(class_scope:tensorflow.GPUOptions.Experimental.VirtualDevices)
+      private static final org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevices DEFAULT_INSTANCE;
+      static {
+        DEFAULT_INSTANCE = new org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevices();
+      }
+
+      public static org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevices getDefaultInstance() {
+        return DEFAULT_INSTANCE;
+      }
+
+      private static final com.google.protobuf.Parser<VirtualDevices>
+          PARSER = new com.google.protobuf.AbstractParser<VirtualDevices>() {
+        @java.lang.Override
+        public VirtualDevices parsePartialFrom(
+            com.google.protobuf.CodedInputStream input,
+            com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+            throws com.google.protobuf.InvalidProtocolBufferException {
+          return new VirtualDevices(input, extensionRegistry);
+        }
+      };
+
+      public static com.google.protobuf.Parser<VirtualDevices> parser() {
+        return PARSER;
+      }
+
+      @java.lang.Override
+      public com.google.protobuf.Parser<VirtualDevices> getParserForType() {
+        return PARSER;
+      }
+
+      @java.lang.Override
+      public org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevices getDefaultInstanceForType() {
+        return DEFAULT_INSTANCE;
+      }
+
+    }
+
+    public static final int VIRTUAL_DEVICES_FIELD_NUMBER = 1;
+    private java.util.List<org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevices> virtualDevices_;
+    /**
+     * <pre>
+     * The multi virtual device settings. If empty (not set), it will create
+     * single virtual device on each visible GPU, according to the settings
+     * in "visible_device_list" above. Otherwise, the number of elements in the
+     * list must be the same as the number of visible GPUs (after
+     * "visible_device_list" filtering if it is set), and the string represented
+     * device names (e.g. /device:GPU:&lt;id&gt;) will refer to the virtual
+     * devices and have the &lt;id&gt; field assigned sequentially starting from 0,
+     * according to the order they appear in this list and the "memory_limit"
+     * list inside each element. For example,
+     *   visible_device_list = "1,0"
+     *   virtual_devices { memory_limit: 1GB memory_limit: 2GB }
+     *   virtual_devices {}
+     * will create three virtual devices as:
+     *   /device:GPU:0 -&gt; visible GPU 1 with 1GB memory
+     *   /device:GPU:1 -&gt; visible GPU 1 with 2GB memory
+     *   /device:GPU:2 -&gt; visible GPU 0 with all available memory
+     * NOTE:
+     * 1. It's invalid to set both this and "per_process_gpu_memory_fraction"
+     *    at the same time.
+     * 2. Currently this setting is per-process, not per-session. Using
+     *    different settings in different sessions within same process will
+     *    result in undefined behavior.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.GPUOptions.Experimental.VirtualDevices virtual_devices = 1;</code>
+     */
+    public java.util.List<org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevices> getVirtualDevicesList() {
+      return virtualDevices_;
+    }
+    /**
+     * <pre>
+     * The multi virtual device settings. If empty (not set), it will create
+     * single virtual device on each visible GPU, according to the settings
+     * in "visible_device_list" above. Otherwise, the number of elements in the
+     * list must be the same as the number of visible GPUs (after
+     * "visible_device_list" filtering if it is set), and the string represented
+     * device names (e.g. /device:GPU:&lt;id&gt;) will refer to the virtual
+     * devices and have the &lt;id&gt; field assigned sequentially starting from 0,
+     * according to the order they appear in this list and the "memory_limit"
+     * list inside each element. For example,
+     *   visible_device_list = "1,0"
+     *   virtual_devices { memory_limit: 1GB memory_limit: 2GB }
+     *   virtual_devices {}
+     * will create three virtual devices as:
+     *   /device:GPU:0 -&gt; visible GPU 1 with 1GB memory
+     *   /device:GPU:1 -&gt; visible GPU 1 with 2GB memory
+     *   /device:GPU:2 -&gt; visible GPU 0 with all available memory
+     * NOTE:
+     * 1. It's invalid to set both this and "per_process_gpu_memory_fraction"
+     *    at the same time.
+     * 2. Currently this setting is per-process, not per-session. Using
+     *    different settings in different sessions within same process will
+     *    result in undefined behavior.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.GPUOptions.Experimental.VirtualDevices virtual_devices = 1;</code>
+     */
+    public java.util.List<? extends org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevicesOrBuilder> 
+        getVirtualDevicesOrBuilderList() {
+      return virtualDevices_;
+    }
+    /**
+     * <pre>
+     * The multi virtual device settings. If empty (not set), it will create
+     * single virtual device on each visible GPU, according to the settings
+     * in "visible_device_list" above. Otherwise, the number of elements in the
+     * list must be the same as the number of visible GPUs (after
+     * "visible_device_list" filtering if it is set), and the string represented
+     * device names (e.g. /device:GPU:&lt;id&gt;) will refer to the virtual
+     * devices and have the &lt;id&gt; field assigned sequentially starting from 0,
+     * according to the order they appear in this list and the "memory_limit"
+     * list inside each element. For example,
+     *   visible_device_list = "1,0"
+     *   virtual_devices { memory_limit: 1GB memory_limit: 2GB }
+     *   virtual_devices {}
+     * will create three virtual devices as:
+     *   /device:GPU:0 -&gt; visible GPU 1 with 1GB memory
+     *   /device:GPU:1 -&gt; visible GPU 1 with 2GB memory
+     *   /device:GPU:2 -&gt; visible GPU 0 with all available memory
+     * NOTE:
+     * 1. It's invalid to set both this and "per_process_gpu_memory_fraction"
+     *    at the same time.
+     * 2. Currently this setting is per-process, not per-session. Using
+     *    different settings in different sessions within same process will
+     *    result in undefined behavior.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.GPUOptions.Experimental.VirtualDevices virtual_devices = 1;</code>
+     */
+    public int getVirtualDevicesCount() {
+      return virtualDevices_.size();
+    }
+    /**
+     * <pre>
+     * The multi virtual device settings. If empty (not set), it will create
+     * single virtual device on each visible GPU, according to the settings
+     * in "visible_device_list" above. Otherwise, the number of elements in the
+     * list must be the same as the number of visible GPUs (after
+     * "visible_device_list" filtering if it is set), and the string represented
+     * device names (e.g. /device:GPU:&lt;id&gt;) will refer to the virtual
+     * devices and have the &lt;id&gt; field assigned sequentially starting from 0,
+     * according to the order they appear in this list and the "memory_limit"
+     * list inside each element. For example,
+     *   visible_device_list = "1,0"
+     *   virtual_devices { memory_limit: 1GB memory_limit: 2GB }
+     *   virtual_devices {}
+     * will create three virtual devices as:
+     *   /device:GPU:0 -&gt; visible GPU 1 with 1GB memory
+     *   /device:GPU:1 -&gt; visible GPU 1 with 2GB memory
+     *   /device:GPU:2 -&gt; visible GPU 0 with all available memory
+     * NOTE:
+     * 1. It's invalid to set both this and "per_process_gpu_memory_fraction"
+     *    at the same time.
+     * 2. Currently this setting is per-process, not per-session. Using
+     *    different settings in different sessions within same process will
+     *    result in undefined behavior.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.GPUOptions.Experimental.VirtualDevices virtual_devices = 1;</code>
+     */
+    public org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevices getVirtualDevices(int index) {
+      return virtualDevices_.get(index);
+    }
+    /**
+     * <pre>
+     * The multi virtual device settings. If empty (not set), it will create
+     * single virtual device on each visible GPU, according to the settings
+     * in "visible_device_list" above. Otherwise, the number of elements in the
+     * list must be the same as the number of visible GPUs (after
+     * "visible_device_list" filtering if it is set), and the string represented
+     * device names (e.g. /device:GPU:&lt;id&gt;) will refer to the virtual
+     * devices and have the &lt;id&gt; field assigned sequentially starting from 0,
+     * according to the order they appear in this list and the "memory_limit"
+     * list inside each element. For example,
+     *   visible_device_list = "1,0"
+     *   virtual_devices { memory_limit: 1GB memory_limit: 2GB }
+     *   virtual_devices {}
+     * will create three virtual devices as:
+     *   /device:GPU:0 -&gt; visible GPU 1 with 1GB memory
+     *   /device:GPU:1 -&gt; visible GPU 1 with 2GB memory
+     *   /device:GPU:2 -&gt; visible GPU 0 with all available memory
+     * NOTE:
+     * 1. It's invalid to set both this and "per_process_gpu_memory_fraction"
+     *    at the same time.
+     * 2. Currently this setting is per-process, not per-session. Using
+     *    different settings in different sessions within same process will
+     *    result in undefined behavior.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.GPUOptions.Experimental.VirtualDevices virtual_devices = 1;</code>
+     */
+    public org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevicesOrBuilder getVirtualDevicesOrBuilder(
+        int index) {
+      return virtualDevices_.get(index);
+    }
+
+    public static final int USE_UNIFIED_MEMORY_FIELD_NUMBER = 2;
+    private boolean useUnifiedMemory_;
+    /**
+     * <pre>
+     * If true, uses CUDA unified memory for memory allocations. If
+     * per_process_gpu_memory_fraction option is greater than 1.0, then unified
+     * memory is used regardless of the value for this field. See comments for
+     * per_process_gpu_memory_fraction field for more details and requirements
+     * of the unified memory. This option is useful to oversubscribe memory if
+     * multiple processes are sharing a single GPU while individually using less
+     * than 1.0 per process memory fraction.
+     * </pre>
+     *
+     * <code>bool use_unified_memory = 2;</code>
+     */
+    public boolean getUseUnifiedMemory() {
+      return useUnifiedMemory_;
+    }
+
+    public static final int NUM_DEV_TO_DEV_COPY_STREAMS_FIELD_NUMBER = 3;
+    private int numDevToDevCopyStreams_;
+    /**
+     * <pre>
+     * If &gt; 1, the number of device-to-device copy streams to create
+     * for each GPUDevice.  Default value is 0, which is automatically
+     * converted to 1.
+     * </pre>
+     *
+     * <code>int32 num_dev_to_dev_copy_streams = 3;</code>
+     */
+    public int getNumDevToDevCopyStreams() {
+      return numDevToDevCopyStreams_;
+    }
+
+    public static final int COLLECTIVE_RING_ORDER_FIELD_NUMBER = 4;
+    private volatile java.lang.Object collectiveRingOrder_;
+    /**
+     * <pre>
+     * If non-empty, defines a good GPU ring order on a single worker based on
+     * device interconnect.  This assumes that all workers have the same GPU
+     * topology.  Specify as a comma-separated string, e.g. "3,2,1,0,7,6,5,4".
+     * This ring order is used by the RingReducer implementation of
+     * CollectiveReduce, and serves as an override to automatic ring order
+     * generation in OrderTaskDeviceMap() during CollectiveParam resolution.
+     * </pre>
+     *
+     * <code>string collective_ring_order = 4;</code>
+     */
+    public java.lang.String getCollectiveRingOrder() {
+      java.lang.Object ref = collectiveRingOrder_;
+      if (ref instanceof java.lang.String) {
+        return (java.lang.String) ref;
+      } else {
+        com.google.protobuf.ByteString bs = 
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        collectiveRingOrder_ = s;
+        return s;
+      }
+    }
+    /**
+     * <pre>
+     * If non-empty, defines a good GPU ring order on a single worker based on
+     * device interconnect.  This assumes that all workers have the same GPU
+     * topology.  Specify as a comma-separated string, e.g. "3,2,1,0,7,6,5,4".
+     * This ring order is used by the RingReducer implementation of
+     * CollectiveReduce, and serves as an override to automatic ring order
+     * generation in OrderTaskDeviceMap() during CollectiveParam resolution.
+     * </pre>
+     *
+     * <code>string collective_ring_order = 4;</code>
+     */
+    public com.google.protobuf.ByteString
+        getCollectiveRingOrderBytes() {
+      java.lang.Object ref = collectiveRingOrder_;
+      if (ref instanceof java.lang.String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        collectiveRingOrder_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+
+    public static final int TIMESTAMPED_ALLOCATOR_FIELD_NUMBER = 5;
+    private boolean timestampedAllocator_;
+    /**
+     * <pre>
+     * If true then extra work is done by GPUDevice and GPUBFCAllocator to
+     * keep track of when GPU memory is freed and when kernels actually
+     * complete so that we can know when a nominally free memory chunk
+     * is really not subject to pending use.
+     * </pre>
+     *
+     * <code>bool timestamped_allocator = 5;</code>
+     */
+    public boolean getTimestampedAllocator() {
+      return timestampedAllocator_;
+    }
+
+    public static final int KERNEL_TRACKER_MAX_INTERVAL_FIELD_NUMBER = 7;
+    private int kernelTrackerMaxInterval_;
+    /**
+     * <pre>
+     * Parameters for GPUKernelTracker.  By default no kernel tracking is done.
+     * Note that timestamped_allocator is only effective if some tracking is
+     * specified.
+     * If kernel_tracker_max_interval = n &gt; 0, then a tracking event
+     * is inserted after every n kernels without an event.
+     * </pre>
+     *
+     * <code>int32 kernel_tracker_max_interval = 7;</code>
+     */
+    public int getKernelTrackerMaxInterval() {
+      return kernelTrackerMaxInterval_;
+    }
+
+    public static final int KERNEL_TRACKER_MAX_BYTES_FIELD_NUMBER = 8;
+    private int kernelTrackerMaxBytes_;
+    /**
+     * <pre>
+     * If kernel_tracker_max_bytes = n &gt; 0, then a tracking event is
+     * inserted after every series of kernels allocating a sum of
+     * memory &gt;= n.  If one kernel allocates b * n bytes, then one
+     * event will be inserted after it, but it will count as b against
+     * the pending limit.
+     * </pre>
+     *
+     * <code>int32 kernel_tracker_max_bytes = 8;</code>
+     */
+    public int getKernelTrackerMaxBytes() {
+      return kernelTrackerMaxBytes_;
+    }
+
+    public static final int KERNEL_TRACKER_MAX_PENDING_FIELD_NUMBER = 9;
+    private int kernelTrackerMaxPending_;
+    /**
+     * <pre>
+     * If kernel_tracker_max_pending &gt; 0 then no more than this many
+     * tracking events can be outstanding at a time.  An attempt to
+     * launch an additional kernel will stall until an event
+     * completes.
+     * </pre>
+     *
+     * <code>int32 kernel_tracker_max_pending = 9;</code>
+     */
+    public int getKernelTrackerMaxPending() {
+      return kernelTrackerMaxPending_;
+    }
+
+    private byte memoizedIsInitialized = -1;
+    @java.lang.Override
+    public final boolean isInitialized() {
+      byte isInitialized = memoizedIsInitialized;
+      if (isInitialized == 1) return true;
+      if (isInitialized == 0) return false;
+
+      memoizedIsInitialized = 1;
+      return true;
+    }
+
+    @java.lang.Override
+    public void writeTo(com.google.protobuf.CodedOutputStream output)
+                        throws java.io.IOException {
+      for (int i = 0; i < virtualDevices_.size(); i++) {
+        output.writeMessage(1, virtualDevices_.get(i));
+      }
+      if (useUnifiedMemory_ != false) {
+        output.writeBool(2, useUnifiedMemory_);
+      }
+      if (numDevToDevCopyStreams_ != 0) {
+        output.writeInt32(3, numDevToDevCopyStreams_);
+      }
+      if (!getCollectiveRingOrderBytes().isEmpty()) {
+        com.google.protobuf.GeneratedMessageV3.writeString(output, 4, collectiveRingOrder_);
+      }
+      if (timestampedAllocator_ != false) {
+        output.writeBool(5, timestampedAllocator_);
+      }
+      if (kernelTrackerMaxInterval_ != 0) {
+        output.writeInt32(7, kernelTrackerMaxInterval_);
+      }
+      if (kernelTrackerMaxBytes_ != 0) {
+        output.writeInt32(8, kernelTrackerMaxBytes_);
+      }
+      if (kernelTrackerMaxPending_ != 0) {
+        output.writeInt32(9, kernelTrackerMaxPending_);
+      }
+      unknownFields.writeTo(output);
+    }
+
+    @java.lang.Override
+    public int getSerializedSize() {
+      int size = memoizedSize;
+      if (size != -1) return size;
+
+      size = 0;
+      for (int i = 0; i < virtualDevices_.size(); i++) {
+        size += com.google.protobuf.CodedOutputStream
+          .computeMessageSize(1, virtualDevices_.get(i));
+      }
+      if (useUnifiedMemory_ != false) {
+        size += com.google.protobuf.CodedOutputStream
+          .computeBoolSize(2, useUnifiedMemory_);
+      }
+      if (numDevToDevCopyStreams_ != 0) {
+        size += com.google.protobuf.CodedOutputStream
+          .computeInt32Size(3, numDevToDevCopyStreams_);
+      }
+      if (!getCollectiveRingOrderBytes().isEmpty()) {
+        size += com.google.protobuf.GeneratedMessageV3.computeStringSize(4, collectiveRingOrder_);
+      }
+      if (timestampedAllocator_ != false) {
+        size += com.google.protobuf.CodedOutputStream
+          .computeBoolSize(5, timestampedAllocator_);
+      }
+      if (kernelTrackerMaxInterval_ != 0) {
+        size += com.google.protobuf.CodedOutputStream
+          .computeInt32Size(7, kernelTrackerMaxInterval_);
+      }
+      if (kernelTrackerMaxBytes_ != 0) {
+        size += com.google.protobuf.CodedOutputStream
+          .computeInt32Size(8, kernelTrackerMaxBytes_);
+      }
+      if (kernelTrackerMaxPending_ != 0) {
+        size += com.google.protobuf.CodedOutputStream
+          .computeInt32Size(9, kernelTrackerMaxPending_);
+      }
+      size += unknownFields.getSerializedSize();
+      memoizedSize = size;
+      return size;
+    }
+
+    @java.lang.Override
+    public boolean equals(final java.lang.Object obj) {
+      if (obj == this) {
+       return true;
+      }
+      if (!(obj instanceof org.tensorflow.proto.framework.GPUOptions.Experimental)) {
+        return super.equals(obj);
+      }
+      org.tensorflow.proto.framework.GPUOptions.Experimental other = (org.tensorflow.proto.framework.GPUOptions.Experimental) obj;
+
+      if (!getVirtualDevicesList()
+          .equals(other.getVirtualDevicesList())) return false;
+      if (getUseUnifiedMemory()
+          != other.getUseUnifiedMemory()) return false;
+      if (getNumDevToDevCopyStreams()
+          != other.getNumDevToDevCopyStreams()) return false;
+      if (!getCollectiveRingOrder()
+          .equals(other.getCollectiveRingOrder())) return false;
+      if (getTimestampedAllocator()
+          != other.getTimestampedAllocator()) return false;
+      if (getKernelTrackerMaxInterval()
+          != other.getKernelTrackerMaxInterval()) return false;
+      if (getKernelTrackerMaxBytes()
+          != other.getKernelTrackerMaxBytes()) return false;
+      if (getKernelTrackerMaxPending()
+          != other.getKernelTrackerMaxPending()) return false;
+      if (!unknownFields.equals(other.unknownFields)) return false;
+      return true;
+    }
+
+    @java.lang.Override
+    public int hashCode() {
+      if (memoizedHashCode != 0) {
+        return memoizedHashCode;
+      }
+      int hash = 41;
+      hash = (19 * hash) + getDescriptor().hashCode();
+      if (getVirtualDevicesCount() > 0) {
+        hash = (37 * hash) + VIRTUAL_DEVICES_FIELD_NUMBER;
+        hash = (53 * hash) + getVirtualDevicesList().hashCode();
+      }
+      hash = (37 * hash) + USE_UNIFIED_MEMORY_FIELD_NUMBER;
+      hash = (53 * hash) + com.google.protobuf.Internal.hashBoolean(
+          getUseUnifiedMemory());
+      hash = (37 * hash) + NUM_DEV_TO_DEV_COPY_STREAMS_FIELD_NUMBER;
+      hash = (53 * hash) + getNumDevToDevCopyStreams();
+      hash = (37 * hash) + COLLECTIVE_RING_ORDER_FIELD_NUMBER;
+      hash = (53 * hash) + getCollectiveRingOrder().hashCode();
+      hash = (37 * hash) + TIMESTAMPED_ALLOCATOR_FIELD_NUMBER;
+      hash = (53 * hash) + com.google.protobuf.Internal.hashBoolean(
+          getTimestampedAllocator());
+      hash = (37 * hash) + KERNEL_TRACKER_MAX_INTERVAL_FIELD_NUMBER;
+      hash = (53 * hash) + getKernelTrackerMaxInterval();
+      hash = (37 * hash) + KERNEL_TRACKER_MAX_BYTES_FIELD_NUMBER;
+      hash = (53 * hash) + getKernelTrackerMaxBytes();
+      hash = (37 * hash) + KERNEL_TRACKER_MAX_PENDING_FIELD_NUMBER;
+      hash = (53 * hash) + getKernelTrackerMaxPending();
+      hash = (29 * hash) + unknownFields.hashCode();
+      memoizedHashCode = hash;
+      return hash;
+    }
+
+    public static org.tensorflow.proto.framework.GPUOptions.Experimental parseFrom(
+        java.nio.ByteBuffer data)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data);
+    }
+    public static org.tensorflow.proto.framework.GPUOptions.Experimental parseFrom(
+        java.nio.ByteBuffer data,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.GPUOptions.Experimental parseFrom(
+        com.google.protobuf.ByteString data)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data);
+    }
+    public static org.tensorflow.proto.framework.GPUOptions.Experimental parseFrom(
+        com.google.protobuf.ByteString data,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.GPUOptions.Experimental parseFrom(byte[] data)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data);
+    }
+    public static org.tensorflow.proto.framework.GPUOptions.Experimental parseFrom(
+        byte[] data,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.GPUOptions.Experimental parseFrom(java.io.InputStream input)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input);
+    }
+    public static org.tensorflow.proto.framework.GPUOptions.Experimental parseFrom(
+        java.io.InputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.GPUOptions.Experimental parseDelimitedFrom(java.io.InputStream input)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseDelimitedWithIOException(PARSER, input);
+    }
+    public static org.tensorflow.proto.framework.GPUOptions.Experimental parseDelimitedFrom(
+        java.io.InputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.GPUOptions.Experimental parseFrom(
+        com.google.protobuf.CodedInputStream input)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input);
+    }
+    public static org.tensorflow.proto.framework.GPUOptions.Experimental parseFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input, extensionRegistry);
+    }
+
+    @java.lang.Override
+    public Builder newBuilderForType() { return newBuilder(); }
+    public static Builder newBuilder() {
+      return DEFAULT_INSTANCE.toBuilder();
+    }
+    public static Builder newBuilder(org.tensorflow.proto.framework.GPUOptions.Experimental prototype) {
+      return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+    }
+    @java.lang.Override
+    public Builder toBuilder() {
+      return this == DEFAULT_INSTANCE
+          ? new Builder() : new Builder().mergeFrom(this);
+    }
+
+    @java.lang.Override
+    protected Builder newBuilderForType(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      Builder builder = new Builder(parent);
+      return builder;
+    }
+    /**
+     * Protobuf type {@code tensorflow.GPUOptions.Experimental}
+     */
+    public static final class Builder extends
+        com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+        // @@protoc_insertion_point(builder_implements:tensorflow.GPUOptions.Experimental)
+        org.tensorflow.proto.framework.GPUOptions.ExperimentalOrBuilder {
+      public static final com.google.protobuf.Descriptors.Descriptor
+          getDescriptor() {
+        return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_GPUOptions_Experimental_descriptor;
+      }
+
+      @java.lang.Override
+      protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+          internalGetFieldAccessorTable() {
+        return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_GPUOptions_Experimental_fieldAccessorTable
+            .ensureFieldAccessorsInitialized(
+                org.tensorflow.proto.framework.GPUOptions.Experimental.class, org.tensorflow.proto.framework.GPUOptions.Experimental.Builder.class);
+      }
+
+      // Construct using org.tensorflow.proto.framework.GPUOptions.Experimental.newBuilder()
+      private Builder() {
+        maybeForceBuilderInitialization();
+      }
+
+      private Builder(
+          com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+        super(parent);
+        maybeForceBuilderInitialization();
+      }
+      private void maybeForceBuilderInitialization() {
+        if (com.google.protobuf.GeneratedMessageV3
+                .alwaysUseFieldBuilders) {
+          getVirtualDevicesFieldBuilder();
+        }
+      }
+      @java.lang.Override
+      public Builder clear() {
+        super.clear();
+        if (virtualDevicesBuilder_ == null) {
+          virtualDevices_ = java.util.Collections.emptyList();
+          bitField0_ = (bitField0_ & ~0x00000001);
+        } else {
+          virtualDevicesBuilder_.clear();
+        }
+        useUnifiedMemory_ = false;
+
+        numDevToDevCopyStreams_ = 0;
+
+        collectiveRingOrder_ = "";
+
+        timestampedAllocator_ = false;
+
+        kernelTrackerMaxInterval_ = 0;
+
+        kernelTrackerMaxBytes_ = 0;
+
+        kernelTrackerMaxPending_ = 0;
+
+        return this;
+      }
+
+      @java.lang.Override
+      public com.google.protobuf.Descriptors.Descriptor
+          getDescriptorForType() {
+        return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_GPUOptions_Experimental_descriptor;
+      }
+
+      @java.lang.Override
+      public org.tensorflow.proto.framework.GPUOptions.Experimental getDefaultInstanceForType() {
+        return org.tensorflow.proto.framework.GPUOptions.Experimental.getDefaultInstance();
+      }
+
+      @java.lang.Override
+      public org.tensorflow.proto.framework.GPUOptions.Experimental build() {
+        org.tensorflow.proto.framework.GPUOptions.Experimental result = buildPartial();
+        if (!result.isInitialized()) {
+          throw newUninitializedMessageException(result);
+        }
+        return result;
+      }
+
+      @java.lang.Override
+      public org.tensorflow.proto.framework.GPUOptions.Experimental buildPartial() {
+        org.tensorflow.proto.framework.GPUOptions.Experimental result = new org.tensorflow.proto.framework.GPUOptions.Experimental(this);
+        int from_bitField0_ = bitField0_;
+        if (virtualDevicesBuilder_ == null) {
+          if (((bitField0_ & 0x00000001) != 0)) {
+            virtualDevices_ = java.util.Collections.unmodifiableList(virtualDevices_);
+            bitField0_ = (bitField0_ & ~0x00000001);
+          }
+          result.virtualDevices_ = virtualDevices_;
+        } else {
+          result.virtualDevices_ = virtualDevicesBuilder_.build();
+        }
+        result.useUnifiedMemory_ = useUnifiedMemory_;
+        result.numDevToDevCopyStreams_ = numDevToDevCopyStreams_;
+        result.collectiveRingOrder_ = collectiveRingOrder_;
+        result.timestampedAllocator_ = timestampedAllocator_;
+        result.kernelTrackerMaxInterval_ = kernelTrackerMaxInterval_;
+        result.kernelTrackerMaxBytes_ = kernelTrackerMaxBytes_;
+        result.kernelTrackerMaxPending_ = kernelTrackerMaxPending_;
+        onBuilt();
+        return result;
+      }
+
+      @java.lang.Override
+      public Builder clone() {
+        return super.clone();
+      }
+      @java.lang.Override
+      public Builder setField(
+          com.google.protobuf.Descriptors.FieldDescriptor field,
+          java.lang.Object value) {
+        return super.setField(field, value);
+      }
+      @java.lang.Override
+      public Builder clearField(
+          com.google.protobuf.Descriptors.FieldDescriptor field) {
+        return super.clearField(field);
+      }
+      @java.lang.Override
+      public Builder clearOneof(
+          com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+        return super.clearOneof(oneof);
+      }
+      @java.lang.Override
+      public Builder setRepeatedField(
+          com.google.protobuf.Descriptors.FieldDescriptor field,
+          int index, java.lang.Object value) {
+        return super.setRepeatedField(field, index, value);
+      }
+      @java.lang.Override
+      public Builder addRepeatedField(
+          com.google.protobuf.Descriptors.FieldDescriptor field,
+          java.lang.Object value) {
+        return super.addRepeatedField(field, value);
+      }
+      @java.lang.Override
+      public Builder mergeFrom(com.google.protobuf.Message other) {
+        if (other instanceof org.tensorflow.proto.framework.GPUOptions.Experimental) {
+          return mergeFrom((org.tensorflow.proto.framework.GPUOptions.Experimental)other);
+        } else {
+          super.mergeFrom(other);
+          return this;
+        }
+      }
+
+      public Builder mergeFrom(org.tensorflow.proto.framework.GPUOptions.Experimental other) {
+        if (other == org.tensorflow.proto.framework.GPUOptions.Experimental.getDefaultInstance()) return this;
+        if (virtualDevicesBuilder_ == null) {
+          if (!other.virtualDevices_.isEmpty()) {
+            if (virtualDevices_.isEmpty()) {
+              virtualDevices_ = other.virtualDevices_;
+              bitField0_ = (bitField0_ & ~0x00000001);
+            } else {
+              ensureVirtualDevicesIsMutable();
+              virtualDevices_.addAll(other.virtualDevices_);
+            }
+            onChanged();
+          }
+        } else {
+          if (!other.virtualDevices_.isEmpty()) {
+            if (virtualDevicesBuilder_.isEmpty()) {
+              virtualDevicesBuilder_.dispose();
+              virtualDevicesBuilder_ = null;
+              virtualDevices_ = other.virtualDevices_;
+              bitField0_ = (bitField0_ & ~0x00000001);
+              virtualDevicesBuilder_ = 
+                com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
+                   getVirtualDevicesFieldBuilder() : null;
+            } else {
+              virtualDevicesBuilder_.addAllMessages(other.virtualDevices_);
+            }
+          }
+        }
+        if (other.getUseUnifiedMemory() != false) {
+          setUseUnifiedMemory(other.getUseUnifiedMemory());
+        }
+        if (other.getNumDevToDevCopyStreams() != 0) {
+          setNumDevToDevCopyStreams(other.getNumDevToDevCopyStreams());
+        }
+        if (!other.getCollectiveRingOrder().isEmpty()) {
+          collectiveRingOrder_ = other.collectiveRingOrder_;
+          onChanged();
+        }
+        if (other.getTimestampedAllocator() != false) {
+          setTimestampedAllocator(other.getTimestampedAllocator());
+        }
+        if (other.getKernelTrackerMaxInterval() != 0) {
+          setKernelTrackerMaxInterval(other.getKernelTrackerMaxInterval());
+        }
+        if (other.getKernelTrackerMaxBytes() != 0) {
+          setKernelTrackerMaxBytes(other.getKernelTrackerMaxBytes());
+        }
+        if (other.getKernelTrackerMaxPending() != 0) {
+          setKernelTrackerMaxPending(other.getKernelTrackerMaxPending());
+        }
+        this.mergeUnknownFields(other.unknownFields);
+        onChanged();
+        return this;
+      }
+
+      @java.lang.Override
+      public final boolean isInitialized() {
+        return true;
+      }
+
+      @java.lang.Override
+      public Builder mergeFrom(
+          com.google.protobuf.CodedInputStream input,
+          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+          throws java.io.IOException {
+        org.tensorflow.proto.framework.GPUOptions.Experimental parsedMessage = null;
+        try {
+          parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+        } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+          parsedMessage = (org.tensorflow.proto.framework.GPUOptions.Experimental) e.getUnfinishedMessage();
+          throw e.unwrapIOException();
+        } finally {
+          if (parsedMessage != null) {
+            mergeFrom(parsedMessage);
+          }
+        }
+        return this;
+      }
+      private int bitField0_;
+
+      private java.util.List<org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevices> virtualDevices_ =
+        java.util.Collections.emptyList();
+      private void ensureVirtualDevicesIsMutable() {
+        if (!((bitField0_ & 0x00000001) != 0)) {
+          virtualDevices_ = new java.util.ArrayList<org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevices>(virtualDevices_);
+          bitField0_ |= 0x00000001;
+         }
+      }
+
+      private com.google.protobuf.RepeatedFieldBuilderV3<
+          org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevices, org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevices.Builder, org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevicesOrBuilder> virtualDevicesBuilder_;
+
+      /**
+       * <pre>
+       * The multi virtual device settings. If empty (not set), it will create
+       * single virtual device on each visible GPU, according to the settings
+       * in "visible_device_list" above. Otherwise, the number of elements in the
+       * list must be the same as the number of visible GPUs (after
+       * "visible_device_list" filtering if it is set), and the string represented
+       * device names (e.g. /device:GPU:&lt;id&gt;) will refer to the virtual
+       * devices and have the &lt;id&gt; field assigned sequentially starting from 0,
+       * according to the order they appear in this list and the "memory_limit"
+       * list inside each element. For example,
+       *   visible_device_list = "1,0"
+       *   virtual_devices { memory_limit: 1GB memory_limit: 2GB }
+       *   virtual_devices {}
+       * will create three virtual devices as:
+       *   /device:GPU:0 -&gt; visible GPU 1 with 1GB memory
+       *   /device:GPU:1 -&gt; visible GPU 1 with 2GB memory
+       *   /device:GPU:2 -&gt; visible GPU 0 with all available memory
+       * NOTE:
+       * 1. It's invalid to set both this and "per_process_gpu_memory_fraction"
+       *    at the same time.
+       * 2. Currently this setting is per-process, not per-session. Using
+       *    different settings in different sessions within same process will
+       *    result in undefined behavior.
+       * </pre>
+       *
+       * <code>repeated .tensorflow.GPUOptions.Experimental.VirtualDevices virtual_devices = 1;</code>
+       */
+      public java.util.List<org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevices> getVirtualDevicesList() {
+        if (virtualDevicesBuilder_ == null) {
+          return java.util.Collections.unmodifiableList(virtualDevices_);
+        } else {
+          return virtualDevicesBuilder_.getMessageList();
+        }
+      }
+      /**
+       * <pre>
+       * The multi virtual device settings. If empty (not set), it will create
+       * single virtual device on each visible GPU, according to the settings
+       * in "visible_device_list" above. Otherwise, the number of elements in the
+       * list must be the same as the number of visible GPUs (after
+       * "visible_device_list" filtering if it is set), and the string represented
+       * device names (e.g. /device:GPU:&lt;id&gt;) will refer to the virtual
+       * devices and have the &lt;id&gt; field assigned sequentially starting from 0,
+       * according to the order they appear in this list and the "memory_limit"
+       * list inside each element. For example,
+       *   visible_device_list = "1,0"
+       *   virtual_devices { memory_limit: 1GB memory_limit: 2GB }
+       *   virtual_devices {}
+       * will create three virtual devices as:
+       *   /device:GPU:0 -&gt; visible GPU 1 with 1GB memory
+       *   /device:GPU:1 -&gt; visible GPU 1 with 2GB memory
+       *   /device:GPU:2 -&gt; visible GPU 0 with all available memory
+       * NOTE:
+       * 1. It's invalid to set both this and "per_process_gpu_memory_fraction"
+       *    at the same time.
+       * 2. Currently this setting is per-process, not per-session. Using
+       *    different settings in different sessions within same process will
+       *    result in undefined behavior.
+       * </pre>
+       *
+       * <code>repeated .tensorflow.GPUOptions.Experimental.VirtualDevices virtual_devices = 1;</code>
+       */
+      public int getVirtualDevicesCount() {
+        if (virtualDevicesBuilder_ == null) {
+          return virtualDevices_.size();
+        } else {
+          return virtualDevicesBuilder_.getCount();
+        }
+      }
+      /**
+       * <pre>
+       * The multi virtual device settings. If empty (not set), it will create
+       * single virtual device on each visible GPU, according to the settings
+       * in "visible_device_list" above. Otherwise, the number of elements in the
+       * list must be the same as the number of visible GPUs (after
+       * "visible_device_list" filtering if it is set), and the string represented
+       * device names (e.g. /device:GPU:&lt;id&gt;) will refer to the virtual
+       * devices and have the &lt;id&gt; field assigned sequentially starting from 0,
+       * according to the order they appear in this list and the "memory_limit"
+       * list inside each element. For example,
+       *   visible_device_list = "1,0"
+       *   virtual_devices { memory_limit: 1GB memory_limit: 2GB }
+       *   virtual_devices {}
+       * will create three virtual devices as:
+       *   /device:GPU:0 -&gt; visible GPU 1 with 1GB memory
+       *   /device:GPU:1 -&gt; visible GPU 1 with 2GB memory
+       *   /device:GPU:2 -&gt; visible GPU 0 with all available memory
+       * NOTE:
+       * 1. It's invalid to set both this and "per_process_gpu_memory_fraction"
+       *    at the same time.
+       * 2. Currently this setting is per-process, not per-session. Using
+       *    different settings in different sessions within same process will
+       *    result in undefined behavior.
+       * </pre>
+       *
+       * <code>repeated .tensorflow.GPUOptions.Experimental.VirtualDevices virtual_devices = 1;</code>
+       */
+      public org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevices getVirtualDevices(int index) {
+        if (virtualDevicesBuilder_ == null) {
+          return virtualDevices_.get(index);
+        } else {
+          return virtualDevicesBuilder_.getMessage(index);
+        }
+      }
+      /**
+       * <pre>
+       * The multi virtual device settings. If empty (not set), it will create
+       * single virtual device on each visible GPU, according to the settings
+       * in "visible_device_list" above. Otherwise, the number of elements in the
+       * list must be the same as the number of visible GPUs (after
+       * "visible_device_list" filtering if it is set), and the string represented
+       * device names (e.g. /device:GPU:&lt;id&gt;) will refer to the virtual
+       * devices and have the &lt;id&gt; field assigned sequentially starting from 0,
+       * according to the order they appear in this list and the "memory_limit"
+       * list inside each element. For example,
+       *   visible_device_list = "1,0"
+       *   virtual_devices { memory_limit: 1GB memory_limit: 2GB }
+       *   virtual_devices {}
+       * will create three virtual devices as:
+       *   /device:GPU:0 -&gt; visible GPU 1 with 1GB memory
+       *   /device:GPU:1 -&gt; visible GPU 1 with 2GB memory
+       *   /device:GPU:2 -&gt; visible GPU 0 with all available memory
+       * NOTE:
+       * 1. It's invalid to set both this and "per_process_gpu_memory_fraction"
+       *    at the same time.
+       * 2. Currently this setting is per-process, not per-session. Using
+       *    different settings in different sessions within same process will
+       *    result in undefined behavior.
+       * </pre>
+       *
+       * <code>repeated .tensorflow.GPUOptions.Experimental.VirtualDevices virtual_devices = 1;</code>
+       */
+      public Builder setVirtualDevices(
+          int index, org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevices value) {
+        if (virtualDevicesBuilder_ == null) {
+          if (value == null) {
+            throw new NullPointerException();
+          }
+          ensureVirtualDevicesIsMutable();
+          virtualDevices_.set(index, value);
+          onChanged();
+        } else {
+          virtualDevicesBuilder_.setMessage(index, value);
+        }
+        return this;
+      }
+      /**
+       * <pre>
+       * The multi virtual device settings. If empty (not set), it will create
+       * single virtual device on each visible GPU, according to the settings
+       * in "visible_device_list" above. Otherwise, the number of elements in the
+       * list must be the same as the number of visible GPUs (after
+       * "visible_device_list" filtering if it is set), and the string represented
+       * device names (e.g. /device:GPU:&lt;id&gt;) will refer to the virtual
+       * devices and have the &lt;id&gt; field assigned sequentially starting from 0,
+       * according to the order they appear in this list and the "memory_limit"
+       * list inside each element. For example,
+       *   visible_device_list = "1,0"
+       *   virtual_devices { memory_limit: 1GB memory_limit: 2GB }
+       *   virtual_devices {}
+       * will create three virtual devices as:
+       *   /device:GPU:0 -&gt; visible GPU 1 with 1GB memory
+       *   /device:GPU:1 -&gt; visible GPU 1 with 2GB memory
+       *   /device:GPU:2 -&gt; visible GPU 0 with all available memory
+       * NOTE:
+       * 1. It's invalid to set both this and "per_process_gpu_memory_fraction"
+       *    at the same time.
+       * 2. Currently this setting is per-process, not per-session. Using
+       *    different settings in different sessions within same process will
+       *    result in undefined behavior.
+       * </pre>
+       *
+       * <code>repeated .tensorflow.GPUOptions.Experimental.VirtualDevices virtual_devices = 1;</code>
+       */
+      public Builder setVirtualDevices(
+          int index, org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevices.Builder builderForValue) {
+        if (virtualDevicesBuilder_ == null) {
+          ensureVirtualDevicesIsMutable();
+          virtualDevices_.set(index, builderForValue.build());
+          onChanged();
+        } else {
+          virtualDevicesBuilder_.setMessage(index, builderForValue.build());
+        }
+        return this;
+      }
+      /**
+       * <pre>
+       * The multi virtual device settings. If empty (not set), it will create
+       * single virtual device on each visible GPU, according to the settings
+       * in "visible_device_list" above. Otherwise, the number of elements in the
+       * list must be the same as the number of visible GPUs (after
+       * "visible_device_list" filtering if it is set), and the string represented
+       * device names (e.g. /device:GPU:&lt;id&gt;) will refer to the virtual
+       * devices and have the &lt;id&gt; field assigned sequentially starting from 0,
+       * according to the order they appear in this list and the "memory_limit"
+       * list inside each element. For example,
+       *   visible_device_list = "1,0"
+       *   virtual_devices { memory_limit: 1GB memory_limit: 2GB }
+       *   virtual_devices {}
+       * will create three virtual devices as:
+       *   /device:GPU:0 -&gt; visible GPU 1 with 1GB memory
+       *   /device:GPU:1 -&gt; visible GPU 1 with 2GB memory
+       *   /device:GPU:2 -&gt; visible GPU 0 with all available memory
+       * NOTE:
+       * 1. It's invalid to set both this and "per_process_gpu_memory_fraction"
+       *    at the same time.
+       * 2. Currently this setting is per-process, not per-session. Using
+       *    different settings in different sessions within same process will
+       *    result in undefined behavior.
+       * </pre>
+       *
+       * <code>repeated .tensorflow.GPUOptions.Experimental.VirtualDevices virtual_devices = 1;</code>
+       */
+      public Builder addVirtualDevices(org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevices value) {
+        if (virtualDevicesBuilder_ == null) {
+          if (value == null) {
+            throw new NullPointerException();
+          }
+          ensureVirtualDevicesIsMutable();
+          virtualDevices_.add(value);
+          onChanged();
+        } else {
+          virtualDevicesBuilder_.addMessage(value);
+        }
+        return this;
+      }
+      /**
+       * <pre>
+       * The multi virtual device settings. If empty (not set), it will create
+       * single virtual device on each visible GPU, according to the settings
+       * in "visible_device_list" above. Otherwise, the number of elements in the
+       * list must be the same as the number of visible GPUs (after
+       * "visible_device_list" filtering if it is set), and the string represented
+       * device names (e.g. /device:GPU:&lt;id&gt;) will refer to the virtual
+       * devices and have the &lt;id&gt; field assigned sequentially starting from 0,
+       * according to the order they appear in this list and the "memory_limit"
+       * list inside each element. For example,
+       *   visible_device_list = "1,0"
+       *   virtual_devices { memory_limit: 1GB memory_limit: 2GB }
+       *   virtual_devices {}
+       * will create three virtual devices as:
+       *   /device:GPU:0 -&gt; visible GPU 1 with 1GB memory
+       *   /device:GPU:1 -&gt; visible GPU 1 with 2GB memory
+       *   /device:GPU:2 -&gt; visible GPU 0 with all available memory
+       * NOTE:
+       * 1. It's invalid to set both this and "per_process_gpu_memory_fraction"
+       *    at the same time.
+       * 2. Currently this setting is per-process, not per-session. Using
+       *    different settings in different sessions within same process will
+       *    result in undefined behavior.
+       * </pre>
+       *
+       * <code>repeated .tensorflow.GPUOptions.Experimental.VirtualDevices virtual_devices = 1;</code>
+       */
+      public Builder addVirtualDevices(
+          int index, org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevices value) {
+        if (virtualDevicesBuilder_ == null) {
+          if (value == null) {
+            throw new NullPointerException();
+          }
+          ensureVirtualDevicesIsMutable();
+          virtualDevices_.add(index, value);
+          onChanged();
+        } else {
+          virtualDevicesBuilder_.addMessage(index, value);
+        }
+        return this;
+      }
+      /**
+       * <pre>
+       * The multi virtual device settings. If empty (not set), it will create
+       * single virtual device on each visible GPU, according to the settings
+       * in "visible_device_list" above. Otherwise, the number of elements in the
+       * list must be the same as the number of visible GPUs (after
+       * "visible_device_list" filtering if it is set), and the string represented
+       * device names (e.g. /device:GPU:&lt;id&gt;) will refer to the virtual
+       * devices and have the &lt;id&gt; field assigned sequentially starting from 0,
+       * according to the order they appear in this list and the "memory_limit"
+       * list inside each element. For example,
+       *   visible_device_list = "1,0"
+       *   virtual_devices { memory_limit: 1GB memory_limit: 2GB }
+       *   virtual_devices {}
+       * will create three virtual devices as:
+       *   /device:GPU:0 -&gt; visible GPU 1 with 1GB memory
+       *   /device:GPU:1 -&gt; visible GPU 1 with 2GB memory
+       *   /device:GPU:2 -&gt; visible GPU 0 with all available memory
+       * NOTE:
+       * 1. It's invalid to set both this and "per_process_gpu_memory_fraction"
+       *    at the same time.
+       * 2. Currently this setting is per-process, not per-session. Using
+       *    different settings in different sessions within same process will
+       *    result in undefined behavior.
+       * </pre>
+       *
+       * <code>repeated .tensorflow.GPUOptions.Experimental.VirtualDevices virtual_devices = 1;</code>
+       */
+      public Builder addVirtualDevices(
+          org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevices.Builder builderForValue) {
+        if (virtualDevicesBuilder_ == null) {
+          ensureVirtualDevicesIsMutable();
+          virtualDevices_.add(builderForValue.build());
+          onChanged();
+        } else {
+          virtualDevicesBuilder_.addMessage(builderForValue.build());
+        }
+        return this;
+      }
+      /**
+       * <pre>
+       * The multi virtual device settings. If empty (not set), it will create
+       * single virtual device on each visible GPU, according to the settings
+       * in "visible_device_list" above. Otherwise, the number of elements in the
+       * list must be the same as the number of visible GPUs (after
+       * "visible_device_list" filtering if it is set), and the string represented
+       * device names (e.g. /device:GPU:&lt;id&gt;) will refer to the virtual
+       * devices and have the &lt;id&gt; field assigned sequentially starting from 0,
+       * according to the order they appear in this list and the "memory_limit"
+       * list inside each element. For example,
+       *   visible_device_list = "1,0"
+       *   virtual_devices { memory_limit: 1GB memory_limit: 2GB }
+       *   virtual_devices {}
+       * will create three virtual devices as:
+       *   /device:GPU:0 -&gt; visible GPU 1 with 1GB memory
+       *   /device:GPU:1 -&gt; visible GPU 1 with 2GB memory
+       *   /device:GPU:2 -&gt; visible GPU 0 with all available memory
+       * NOTE:
+       * 1. It's invalid to set both this and "per_process_gpu_memory_fraction"
+       *    at the same time.
+       * 2. Currently this setting is per-process, not per-session. Using
+       *    different settings in different sessions within same process will
+       *    result in undefined behavior.
+       * </pre>
+       *
+       * <code>repeated .tensorflow.GPUOptions.Experimental.VirtualDevices virtual_devices = 1;</code>
+       */
+      public Builder addVirtualDevices(
+          int index, org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevices.Builder builderForValue) {
+        if (virtualDevicesBuilder_ == null) {
+          ensureVirtualDevicesIsMutable();
+          virtualDevices_.add(index, builderForValue.build());
+          onChanged();
+        } else {
+          virtualDevicesBuilder_.addMessage(index, builderForValue.build());
+        }
+        return this;
+      }
+      /**
+       * <pre>
+       * The multi virtual device settings. If empty (not set), it will create
+       * single virtual device on each visible GPU, according to the settings
+       * in "visible_device_list" above. Otherwise, the number of elements in the
+       * list must be the same as the number of visible GPUs (after
+       * "visible_device_list" filtering if it is set), and the string represented
+       * device names (e.g. /device:GPU:&lt;id&gt;) will refer to the virtual
+       * devices and have the &lt;id&gt; field assigned sequentially starting from 0,
+       * according to the order they appear in this list and the "memory_limit"
+       * list inside each element. For example,
+       *   visible_device_list = "1,0"
+       *   virtual_devices { memory_limit: 1GB memory_limit: 2GB }
+       *   virtual_devices {}
+       * will create three virtual devices as:
+       *   /device:GPU:0 -&gt; visible GPU 1 with 1GB memory
+       *   /device:GPU:1 -&gt; visible GPU 1 with 2GB memory
+       *   /device:GPU:2 -&gt; visible GPU 0 with all available memory
+       * NOTE:
+       * 1. It's invalid to set both this and "per_process_gpu_memory_fraction"
+       *    at the same time.
+       * 2. Currently this setting is per-process, not per-session. Using
+       *    different settings in different sessions within same process will
+       *    result in undefined behavior.
+       * </pre>
+       *
+       * <code>repeated .tensorflow.GPUOptions.Experimental.VirtualDevices virtual_devices = 1;</code>
+       */
+      public Builder addAllVirtualDevices(
+          java.lang.Iterable<? extends org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevices> values) {
+        if (virtualDevicesBuilder_ == null) {
+          ensureVirtualDevicesIsMutable();
+          com.google.protobuf.AbstractMessageLite.Builder.addAll(
+              values, virtualDevices_);
+          onChanged();
+        } else {
+          virtualDevicesBuilder_.addAllMessages(values);
+        }
+        return this;
+      }
+      /**
+       * <pre>
+       * The multi virtual device settings. If empty (not set), it will create
+       * single virtual device on each visible GPU, according to the settings
+       * in "visible_device_list" above. Otherwise, the number of elements in the
+       * list must be the same as the number of visible GPUs (after
+       * "visible_device_list" filtering if it is set), and the string represented
+       * device names (e.g. /device:GPU:&lt;id&gt;) will refer to the virtual
+       * devices and have the &lt;id&gt; field assigned sequentially starting from 0,
+       * according to the order they appear in this list and the "memory_limit"
+       * list inside each element. For example,
+       *   visible_device_list = "1,0"
+       *   virtual_devices { memory_limit: 1GB memory_limit: 2GB }
+       *   virtual_devices {}
+       * will create three virtual devices as:
+       *   /device:GPU:0 -&gt; visible GPU 1 with 1GB memory
+       *   /device:GPU:1 -&gt; visible GPU 1 with 2GB memory
+       *   /device:GPU:2 -&gt; visible GPU 0 with all available memory
+       * NOTE:
+       * 1. It's invalid to set both this and "per_process_gpu_memory_fraction"
+       *    at the same time.
+       * 2. Currently this setting is per-process, not per-session. Using
+       *    different settings in different sessions within same process will
+       *    result in undefined behavior.
+       * </pre>
+       *
+       * <code>repeated .tensorflow.GPUOptions.Experimental.VirtualDevices virtual_devices = 1;</code>
+       */
+      public Builder clearVirtualDevices() {
+        if (virtualDevicesBuilder_ == null) {
+          virtualDevices_ = java.util.Collections.emptyList();
+          bitField0_ = (bitField0_ & ~0x00000001);
+          onChanged();
+        } else {
+          virtualDevicesBuilder_.clear();
+        }
+        return this;
+      }
+      /**
+       * <pre>
+       * The multi virtual device settings. If empty (not set), it will create
+       * single virtual device on each visible GPU, according to the settings
+       * in "visible_device_list" above. Otherwise, the number of elements in the
+       * list must be the same as the number of visible GPUs (after
+       * "visible_device_list" filtering if it is set), and the string represented
+       * device names (e.g. /device:GPU:&lt;id&gt;) will refer to the virtual
+       * devices and have the &lt;id&gt; field assigned sequentially starting from 0,
+       * according to the order they appear in this list and the "memory_limit"
+       * list inside each element. For example,
+       *   visible_device_list = "1,0"
+       *   virtual_devices { memory_limit: 1GB memory_limit: 2GB }
+       *   virtual_devices {}
+       * will create three virtual devices as:
+       *   /device:GPU:0 -&gt; visible GPU 1 with 1GB memory
+       *   /device:GPU:1 -&gt; visible GPU 1 with 2GB memory
+       *   /device:GPU:2 -&gt; visible GPU 0 with all available memory
+       * NOTE:
+       * 1. It's invalid to set both this and "per_process_gpu_memory_fraction"
+       *    at the same time.
+       * 2. Currently this setting is per-process, not per-session. Using
+       *    different settings in different sessions within same process will
+       *    result in undefined behavior.
+       * </pre>
+       *
+       * <code>repeated .tensorflow.GPUOptions.Experimental.VirtualDevices virtual_devices = 1;</code>
+       */
+      public Builder removeVirtualDevices(int index) {
+        if (virtualDevicesBuilder_ == null) {
+          ensureVirtualDevicesIsMutable();
+          virtualDevices_.remove(index);
+          onChanged();
+        } else {
+          virtualDevicesBuilder_.remove(index);
+        }
+        return this;
+      }
+      /**
+       * <pre>
+       * The multi virtual device settings. If empty (not set), it will create
+       * single virtual device on each visible GPU, according to the settings
+       * in "visible_device_list" above. Otherwise, the number of elements in the
+       * list must be the same as the number of visible GPUs (after
+       * "visible_device_list" filtering if it is set), and the string represented
+       * device names (e.g. /device:GPU:&lt;id&gt;) will refer to the virtual
+       * devices and have the &lt;id&gt; field assigned sequentially starting from 0,
+       * according to the order they appear in this list and the "memory_limit"
+       * list inside each element. For example,
+       *   visible_device_list = "1,0"
+       *   virtual_devices { memory_limit: 1GB memory_limit: 2GB }
+       *   virtual_devices {}
+       * will create three virtual devices as:
+       *   /device:GPU:0 -&gt; visible GPU 1 with 1GB memory
+       *   /device:GPU:1 -&gt; visible GPU 1 with 2GB memory
+       *   /device:GPU:2 -&gt; visible GPU 0 with all available memory
+       * NOTE:
+       * 1. It's invalid to set both this and "per_process_gpu_memory_fraction"
+       *    at the same time.
+       * 2. Currently this setting is per-process, not per-session. Using
+       *    different settings in different sessions within same process will
+       *    result in undefined behavior.
+       * </pre>
+       *
+       * <code>repeated .tensorflow.GPUOptions.Experimental.VirtualDevices virtual_devices = 1;</code>
+       */
+      public org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevices.Builder getVirtualDevicesBuilder(
+          int index) {
+        return getVirtualDevicesFieldBuilder().getBuilder(index);
+      }
+      /**
+       * <pre>
+       * The multi virtual device settings. If empty (not set), it will create
+       * single virtual device on each visible GPU, according to the settings
+       * in "visible_device_list" above. Otherwise, the number of elements in the
+       * list must be the same as the number of visible GPUs (after
+       * "visible_device_list" filtering if it is set), and the string represented
+       * device names (e.g. /device:GPU:&lt;id&gt;) will refer to the virtual
+       * devices and have the &lt;id&gt; field assigned sequentially starting from 0,
+       * according to the order they appear in this list and the "memory_limit"
+       * list inside each element. For example,
+       *   visible_device_list = "1,0"
+       *   virtual_devices { memory_limit: 1GB memory_limit: 2GB }
+       *   virtual_devices {}
+       * will create three virtual devices as:
+       *   /device:GPU:0 -&gt; visible GPU 1 with 1GB memory
+       *   /device:GPU:1 -&gt; visible GPU 1 with 2GB memory
+       *   /device:GPU:2 -&gt; visible GPU 0 with all available memory
+       * NOTE:
+       * 1. It's invalid to set both this and "per_process_gpu_memory_fraction"
+       *    at the same time.
+       * 2. Currently this setting is per-process, not per-session. Using
+       *    different settings in different sessions within same process will
+       *    result in undefined behavior.
+       * </pre>
+       *
+       * <code>repeated .tensorflow.GPUOptions.Experimental.VirtualDevices virtual_devices = 1;</code>
+       */
+      public org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevicesOrBuilder getVirtualDevicesOrBuilder(
+          int index) {
+        if (virtualDevicesBuilder_ == null) {
+          return virtualDevices_.get(index);  } else {
+          return virtualDevicesBuilder_.getMessageOrBuilder(index);
+        }
+      }
+      /**
+       * <pre>
+       * The multi virtual device settings. If empty (not set), it will create
+       * single virtual device on each visible GPU, according to the settings
+       * in "visible_device_list" above. Otherwise, the number of elements in the
+       * list must be the same as the number of visible GPUs (after
+       * "visible_device_list" filtering if it is set), and the string represented
+       * device names (e.g. /device:GPU:&lt;id&gt;) will refer to the virtual
+       * devices and have the &lt;id&gt; field assigned sequentially starting from 0,
+       * according to the order they appear in this list and the "memory_limit"
+       * list inside each element. For example,
+       *   visible_device_list = "1,0"
+       *   virtual_devices { memory_limit: 1GB memory_limit: 2GB }
+       *   virtual_devices {}
+       * will create three virtual devices as:
+       *   /device:GPU:0 -&gt; visible GPU 1 with 1GB memory
+       *   /device:GPU:1 -&gt; visible GPU 1 with 2GB memory
+       *   /device:GPU:2 -&gt; visible GPU 0 with all available memory
+       * NOTE:
+       * 1. It's invalid to set both this and "per_process_gpu_memory_fraction"
+       *    at the same time.
+       * 2. Currently this setting is per-process, not per-session. Using
+       *    different settings in different sessions within same process will
+       *    result in undefined behavior.
+       * </pre>
+       *
+       * <code>repeated .tensorflow.GPUOptions.Experimental.VirtualDevices virtual_devices = 1;</code>
+       */
+      public java.util.List<? extends org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevicesOrBuilder> 
+           getVirtualDevicesOrBuilderList() {
+        if (virtualDevicesBuilder_ != null) {
+          return virtualDevicesBuilder_.getMessageOrBuilderList();
+        } else {
+          return java.util.Collections.unmodifiableList(virtualDevices_);
+        }
+      }
+      /**
+       * <pre>
+       * The multi virtual device settings. If empty (not set), it will create
+       * single virtual device on each visible GPU, according to the settings
+       * in "visible_device_list" above. Otherwise, the number of elements in the
+       * list must be the same as the number of visible GPUs (after
+       * "visible_device_list" filtering if it is set), and the string represented
+       * device names (e.g. /device:GPU:&lt;id&gt;) will refer to the virtual
+       * devices and have the &lt;id&gt; field assigned sequentially starting from 0,
+       * according to the order they appear in this list and the "memory_limit"
+       * list inside each element. For example,
+       *   visible_device_list = "1,0"
+       *   virtual_devices { memory_limit: 1GB memory_limit: 2GB }
+       *   virtual_devices {}
+       * will create three virtual devices as:
+       *   /device:GPU:0 -&gt; visible GPU 1 with 1GB memory
+       *   /device:GPU:1 -&gt; visible GPU 1 with 2GB memory
+       *   /device:GPU:2 -&gt; visible GPU 0 with all available memory
+       * NOTE:
+       * 1. It's invalid to set both this and "per_process_gpu_memory_fraction"
+       *    at the same time.
+       * 2. Currently this setting is per-process, not per-session. Using
+       *    different settings in different sessions within same process will
+       *    result in undefined behavior.
+       * </pre>
+       *
+       * <code>repeated .tensorflow.GPUOptions.Experimental.VirtualDevices virtual_devices = 1;</code>
+       */
+      public org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevices.Builder addVirtualDevicesBuilder() {
+        return getVirtualDevicesFieldBuilder().addBuilder(
+            org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevices.getDefaultInstance());
+      }
+      /**
+       * <pre>
+       * The multi virtual device settings. If empty (not set), it will create
+       * single virtual device on each visible GPU, according to the settings
+       * in "visible_device_list" above. Otherwise, the number of elements in the
+       * list must be the same as the number of visible GPUs (after
+       * "visible_device_list" filtering if it is set), and the string represented
+       * device names (e.g. /device:GPU:&lt;id&gt;) will refer to the virtual
+       * devices and have the &lt;id&gt; field assigned sequentially starting from 0,
+       * according to the order they appear in this list and the "memory_limit"
+       * list inside each element. For example,
+       *   visible_device_list = "1,0"
+       *   virtual_devices { memory_limit: 1GB memory_limit: 2GB }
+       *   virtual_devices {}
+       * will create three virtual devices as:
+       *   /device:GPU:0 -&gt; visible GPU 1 with 1GB memory
+       *   /device:GPU:1 -&gt; visible GPU 1 with 2GB memory
+       *   /device:GPU:2 -&gt; visible GPU 0 with all available memory
+       * NOTE:
+       * 1. It's invalid to set both this and "per_process_gpu_memory_fraction"
+       *    at the same time.
+       * 2. Currently this setting is per-process, not per-session. Using
+       *    different settings in different sessions within same process will
+       *    result in undefined behavior.
+       * </pre>
+       *
+       * <code>repeated .tensorflow.GPUOptions.Experimental.VirtualDevices virtual_devices = 1;</code>
+       */
+      public org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevices.Builder addVirtualDevicesBuilder(
+          int index) {
+        return getVirtualDevicesFieldBuilder().addBuilder(
+            index, org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevices.getDefaultInstance());
+      }
+      /**
+       * <pre>
+       * The multi virtual device settings. If empty (not set), it will create
+       * single virtual device on each visible GPU, according to the settings
+       * in "visible_device_list" above. Otherwise, the number of elements in the
+       * list must be the same as the number of visible GPUs (after
+       * "visible_device_list" filtering if it is set), and the string represented
+       * device names (e.g. /device:GPU:&lt;id&gt;) will refer to the virtual
+       * devices and have the &lt;id&gt; field assigned sequentially starting from 0,
+       * according to the order they appear in this list and the "memory_limit"
+       * list inside each element. For example,
+       *   visible_device_list = "1,0"
+       *   virtual_devices { memory_limit: 1GB memory_limit: 2GB }
+       *   virtual_devices {}
+       * will create three virtual devices as:
+       *   /device:GPU:0 -&gt; visible GPU 1 with 1GB memory
+       *   /device:GPU:1 -&gt; visible GPU 1 with 2GB memory
+       *   /device:GPU:2 -&gt; visible GPU 0 with all available memory
+       * NOTE:
+       * 1. It's invalid to set both this and "per_process_gpu_memory_fraction"
+       *    at the same time.
+       * 2. Currently this setting is per-process, not per-session. Using
+       *    different settings in different sessions within same process will
+       *    result in undefined behavior.
+       * </pre>
+       *
+       * <code>repeated .tensorflow.GPUOptions.Experimental.VirtualDevices virtual_devices = 1;</code>
+       */
+      public java.util.List<org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevices.Builder> 
+           getVirtualDevicesBuilderList() {
+        return getVirtualDevicesFieldBuilder().getBuilderList();
+      }
+      private com.google.protobuf.RepeatedFieldBuilderV3<
+          org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevices, org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevices.Builder, org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevicesOrBuilder> 
+          getVirtualDevicesFieldBuilder() {
+        if (virtualDevicesBuilder_ == null) {
+          virtualDevicesBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
+              org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevices, org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevices.Builder, org.tensorflow.proto.framework.GPUOptions.Experimental.VirtualDevicesOrBuilder>(
+                  virtualDevices_,
+                  ((bitField0_ & 0x00000001) != 0),
+                  getParentForChildren(),
+                  isClean());
+          virtualDevices_ = null;
+        }
+        return virtualDevicesBuilder_;
+      }
+
+      private boolean useUnifiedMemory_ ;
+      /**
+       * <pre>
+       * If true, uses CUDA unified memory for memory allocations. If
+       * per_process_gpu_memory_fraction option is greater than 1.0, then unified
+       * memory is used regardless of the value for this field. See comments for
+       * per_process_gpu_memory_fraction field for more details and requirements
+       * of the unified memory. This option is useful to oversubscribe memory if
+       * multiple processes are sharing a single GPU while individually using less
+       * than 1.0 per process memory fraction.
+       * </pre>
+       *
+       * <code>bool use_unified_memory = 2;</code>
+       */
+      public boolean getUseUnifiedMemory() {
+        return useUnifiedMemory_;
+      }
+      /**
+       * <pre>
+       * If true, uses CUDA unified memory for memory allocations. If
+       * per_process_gpu_memory_fraction option is greater than 1.0, then unified
+       * memory is used regardless of the value for this field. See comments for
+       * per_process_gpu_memory_fraction field for more details and requirements
+       * of the unified memory. This option is useful to oversubscribe memory if
+       * multiple processes are sharing a single GPU while individually using less
+       * than 1.0 per process memory fraction.
+       * </pre>
+       *
+       * <code>bool use_unified_memory = 2;</code>
+       */
+      public Builder setUseUnifiedMemory(boolean value) {
+        
+        useUnifiedMemory_ = value;
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * If true, uses CUDA unified memory for memory allocations. If
+       * per_process_gpu_memory_fraction option is greater than 1.0, then unified
+       * memory is used regardless of the value for this field. See comments for
+       * per_process_gpu_memory_fraction field for more details and requirements
+       * of the unified memory. This option is useful to oversubscribe memory if
+       * multiple processes are sharing a single GPU while individually using less
+       * than 1.0 per process memory fraction.
+       * </pre>
+       *
+       * <code>bool use_unified_memory = 2;</code>
+       */
+      public Builder clearUseUnifiedMemory() {
+        
+        useUnifiedMemory_ = false;
+        onChanged();
+        return this;
+      }
+
+      private int numDevToDevCopyStreams_ ;
+      /**
+       * <pre>
+       * If &gt; 1, the number of device-to-device copy streams to create
+       * for each GPUDevice.  Default value is 0, which is automatically
+       * converted to 1.
+       * </pre>
+       *
+       * <code>int32 num_dev_to_dev_copy_streams = 3;</code>
+       */
+      public int getNumDevToDevCopyStreams() {
+        return numDevToDevCopyStreams_;
+      }
+      /**
+       * <pre>
+       * If &gt; 1, the number of device-to-device copy streams to create
+       * for each GPUDevice.  Default value is 0, which is automatically
+       * converted to 1.
+       * </pre>
+       *
+       * <code>int32 num_dev_to_dev_copy_streams = 3;</code>
+       */
+      public Builder setNumDevToDevCopyStreams(int value) {
+        
+        numDevToDevCopyStreams_ = value;
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * If &gt; 1, the number of device-to-device copy streams to create
+       * for each GPUDevice.  Default value is 0, which is automatically
+       * converted to 1.
+       * </pre>
+       *
+       * <code>int32 num_dev_to_dev_copy_streams = 3;</code>
+       */
+      public Builder clearNumDevToDevCopyStreams() {
+        
+        numDevToDevCopyStreams_ = 0;
+        onChanged();
+        return this;
+      }
+
+      private java.lang.Object collectiveRingOrder_ = "";
+      /**
+       * <pre>
+       * If non-empty, defines a good GPU ring order on a single worker based on
+       * device interconnect.  This assumes that all workers have the same GPU
+       * topology.  Specify as a comma-separated string, e.g. "3,2,1,0,7,6,5,4".
+       * This ring order is used by the RingReducer implementation of
+       * CollectiveReduce, and serves as an override to automatic ring order
+       * generation in OrderTaskDeviceMap() during CollectiveParam resolution.
+       * </pre>
+       *
+       * <code>string collective_ring_order = 4;</code>
+       */
+      public java.lang.String getCollectiveRingOrder() {
+        java.lang.Object ref = collectiveRingOrder_;
+        if (!(ref instanceof java.lang.String)) {
+          com.google.protobuf.ByteString bs =
+              (com.google.protobuf.ByteString) ref;
+          java.lang.String s = bs.toStringUtf8();
+          collectiveRingOrder_ = s;
+          return s;
+        } else {
+          return (java.lang.String) ref;
+        }
+      }
+      /**
+       * <pre>
+       * If non-empty, defines a good GPU ring order on a single worker based on
+       * device interconnect.  This assumes that all workers have the same GPU
+       * topology.  Specify as a comma-separated string, e.g. "3,2,1,0,7,6,5,4".
+       * This ring order is used by the RingReducer implementation of
+       * CollectiveReduce, and serves as an override to automatic ring order
+       * generation in OrderTaskDeviceMap() during CollectiveParam resolution.
+       * </pre>
+       *
+       * <code>string collective_ring_order = 4;</code>
+       */
+      public com.google.protobuf.ByteString
+          getCollectiveRingOrderBytes() {
+        java.lang.Object ref = collectiveRingOrder_;
+        if (ref instanceof String) {
+          com.google.protobuf.ByteString b = 
+              com.google.protobuf.ByteString.copyFromUtf8(
+                  (java.lang.String) ref);
+          collectiveRingOrder_ = b;
+          return b;
+        } else {
+          return (com.google.protobuf.ByteString) ref;
+        }
+      }
+      /**
+       * <pre>
+       * If non-empty, defines a good GPU ring order on a single worker based on
+       * device interconnect.  This assumes that all workers have the same GPU
+       * topology.  Specify as a comma-separated string, e.g. "3,2,1,0,7,6,5,4".
+       * This ring order is used by the RingReducer implementation of
+       * CollectiveReduce, and serves as an override to automatic ring order
+       * generation in OrderTaskDeviceMap() during CollectiveParam resolution.
+       * </pre>
+       *
+       * <code>string collective_ring_order = 4;</code>
+       */
+      public Builder setCollectiveRingOrder(
+          java.lang.String value) {
+        if (value == null) {
+    throw new NullPointerException();
+  }
+  
+        collectiveRingOrder_ = value;
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * If non-empty, defines a good GPU ring order on a single worker based on
+       * device interconnect.  This assumes that all workers have the same GPU
+       * topology.  Specify as a comma-separated string, e.g. "3,2,1,0,7,6,5,4".
+       * This ring order is used by the RingReducer implementation of
+       * CollectiveReduce, and serves as an override to automatic ring order
+       * generation in OrderTaskDeviceMap() during CollectiveParam resolution.
+       * </pre>
+       *
+       * <code>string collective_ring_order = 4;</code>
+       */
+      public Builder clearCollectiveRingOrder() {
+        
+        collectiveRingOrder_ = getDefaultInstance().getCollectiveRingOrder();
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * If non-empty, defines a good GPU ring order on a single worker based on
+       * device interconnect.  This assumes that all workers have the same GPU
+       * topology.  Specify as a comma-separated string, e.g. "3,2,1,0,7,6,5,4".
+       * This ring order is used by the RingReducer implementation of
+       * CollectiveReduce, and serves as an override to automatic ring order
+       * generation in OrderTaskDeviceMap() during CollectiveParam resolution.
+       * </pre>
+       *
+       * <code>string collective_ring_order = 4;</code>
+       */
+      public Builder setCollectiveRingOrderBytes(
+          com.google.protobuf.ByteString value) {
+        if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+        
+        collectiveRingOrder_ = value;
+        onChanged();
+        return this;
+      }
+
+      private boolean timestampedAllocator_ ;
+      /**
+       * <pre>
+       * If true then extra work is done by GPUDevice and GPUBFCAllocator to
+       * keep track of when GPU memory is freed and when kernels actually
+       * complete so that we can know when a nominally free memory chunk
+       * is really not subject to pending use.
+       * </pre>
+       *
+       * <code>bool timestamped_allocator = 5;</code>
+       */
+      public boolean getTimestampedAllocator() {
+        return timestampedAllocator_;
+      }
+      /**
+       * <pre>
+       * If true then extra work is done by GPUDevice and GPUBFCAllocator to
+       * keep track of when GPU memory is freed and when kernels actually
+       * complete so that we can know when a nominally free memory chunk
+       * is really not subject to pending use.
+       * </pre>
+       *
+       * <code>bool timestamped_allocator = 5;</code>
+       */
+      public Builder setTimestampedAllocator(boolean value) {
+        
+        timestampedAllocator_ = value;
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * If true then extra work is done by GPUDevice and GPUBFCAllocator to
+       * keep track of when GPU memory is freed and when kernels actually
+       * complete so that we can know when a nominally free memory chunk
+       * is really not subject to pending use.
+       * </pre>
+       *
+       * <code>bool timestamped_allocator = 5;</code>
+       */
+      public Builder clearTimestampedAllocator() {
+        
+        timestampedAllocator_ = false;
+        onChanged();
+        return this;
+      }
+
+      private int kernelTrackerMaxInterval_ ;
+      /**
+       * <pre>
+       * Parameters for GPUKernelTracker.  By default no kernel tracking is done.
+       * Note that timestamped_allocator is only effective if some tracking is
+       * specified.
+       * If kernel_tracker_max_interval = n &gt; 0, then a tracking event
+       * is inserted after every n kernels without an event.
+       * </pre>
+       *
+       * <code>int32 kernel_tracker_max_interval = 7;</code>
+       */
+      public int getKernelTrackerMaxInterval() {
+        return kernelTrackerMaxInterval_;
+      }
+      /**
+       * <pre>
+       * Parameters for GPUKernelTracker.  By default no kernel tracking is done.
+       * Note that timestamped_allocator is only effective if some tracking is
+       * specified.
+       * If kernel_tracker_max_interval = n &gt; 0, then a tracking event
+       * is inserted after every n kernels without an event.
+       * </pre>
+       *
+       * <code>int32 kernel_tracker_max_interval = 7;</code>
+       */
+      public Builder setKernelTrackerMaxInterval(int value) {
+        
+        kernelTrackerMaxInterval_ = value;
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * Parameters for GPUKernelTracker.  By default no kernel tracking is done.
+       * Note that timestamped_allocator is only effective if some tracking is
+       * specified.
+       * If kernel_tracker_max_interval = n &gt; 0, then a tracking event
+       * is inserted after every n kernels without an event.
+       * </pre>
+       *
+       * <code>int32 kernel_tracker_max_interval = 7;</code>
+       */
+      public Builder clearKernelTrackerMaxInterval() {
+        
+        kernelTrackerMaxInterval_ = 0;
+        onChanged();
+        return this;
+      }
+
+      private int kernelTrackerMaxBytes_ ;
+      /**
+       * <pre>
+       * If kernel_tracker_max_bytes = n &gt; 0, then a tracking event is
+       * inserted after every series of kernels allocating a sum of
+       * memory &gt;= n.  If one kernel allocates b * n bytes, then one
+       * event will be inserted after it, but it will count as b against
+       * the pending limit.
+       * </pre>
+       *
+       * <code>int32 kernel_tracker_max_bytes = 8;</code>
+       */
+      public int getKernelTrackerMaxBytes() {
+        return kernelTrackerMaxBytes_;
+      }
+      /**
+       * <pre>
+       * If kernel_tracker_max_bytes = n &gt; 0, then a tracking event is
+       * inserted after every series of kernels allocating a sum of
+       * memory &gt;= n.  If one kernel allocates b * n bytes, then one
+       * event will be inserted after it, but it will count as b against
+       * the pending limit.
+       * </pre>
+       *
+       * <code>int32 kernel_tracker_max_bytes = 8;</code>
+       */
+      public Builder setKernelTrackerMaxBytes(int value) {
+        
+        kernelTrackerMaxBytes_ = value;
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * If kernel_tracker_max_bytes = n &gt; 0, then a tracking event is
+       * inserted after every series of kernels allocating a sum of
+       * memory &gt;= n.  If one kernel allocates b * n bytes, then one
+       * event will be inserted after it, but it will count as b against
+       * the pending limit.
+       * </pre>
+       *
+       * <code>int32 kernel_tracker_max_bytes = 8;</code>
+       */
+      public Builder clearKernelTrackerMaxBytes() {
+        
+        kernelTrackerMaxBytes_ = 0;
+        onChanged();
+        return this;
+      }
+
+      private int kernelTrackerMaxPending_ ;
+      /**
+       * <pre>
+       * If kernel_tracker_max_pending &gt; 0 then no more than this many
+       * tracking events can be outstanding at a time.  An attempt to
+       * launch an additional kernel will stall until an event
+       * completes.
+       * </pre>
+       *
+       * <code>int32 kernel_tracker_max_pending = 9;</code>
+       */
+      public int getKernelTrackerMaxPending() {
+        return kernelTrackerMaxPending_;
+      }
+      /**
+       * <pre>
+       * If kernel_tracker_max_pending &gt; 0 then no more than this many
+       * tracking events can be outstanding at a time.  An attempt to
+       * launch an additional kernel will stall until an event
+       * completes.
+       * </pre>
+       *
+       * <code>int32 kernel_tracker_max_pending = 9;</code>
+       */
+      public Builder setKernelTrackerMaxPending(int value) {
+        
+        kernelTrackerMaxPending_ = value;
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * If kernel_tracker_max_pending &gt; 0 then no more than this many
+       * tracking events can be outstanding at a time.  An attempt to
+       * launch an additional kernel will stall until an event
+       * completes.
+       * </pre>
+       *
+       * <code>int32 kernel_tracker_max_pending = 9;</code>
+       */
+      public Builder clearKernelTrackerMaxPending() {
+        
+        kernelTrackerMaxPending_ = 0;
+        onChanged();
+        return this;
+      }
+      @java.lang.Override
+      public final Builder setUnknownFields(
+          final com.google.protobuf.UnknownFieldSet unknownFields) {
+        return super.setUnknownFields(unknownFields);
+      }
+
+      @java.lang.Override
+      public final Builder mergeUnknownFields(
+          final com.google.protobuf.UnknownFieldSet unknownFields) {
+        return super.mergeUnknownFields(unknownFields);
+      }
+
+
+      // @@protoc_insertion_point(builder_scope:tensorflow.GPUOptions.Experimental)
+    }
+
+    // @@protoc_insertion_point(class_scope:tensorflow.GPUOptions.Experimental)
+    private static final org.tensorflow.proto.framework.GPUOptions.Experimental DEFAULT_INSTANCE;
+    static {
+      DEFAULT_INSTANCE = new org.tensorflow.proto.framework.GPUOptions.Experimental();
+    }
+
+    public static org.tensorflow.proto.framework.GPUOptions.Experimental getDefaultInstance() {
+      return DEFAULT_INSTANCE;
+    }
+
+    private static final com.google.protobuf.Parser<Experimental>
+        PARSER = new com.google.protobuf.AbstractParser<Experimental>() {
+      @java.lang.Override
+      public Experimental parsePartialFrom(
+          com.google.protobuf.CodedInputStream input,
+          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+          throws com.google.protobuf.InvalidProtocolBufferException {
+        return new Experimental(input, extensionRegistry);
+      }
+    };
+
+    public static com.google.protobuf.Parser<Experimental> parser() {
+      return PARSER;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Parser<Experimental> getParserForType() {
+      return PARSER;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.GPUOptions.Experimental getDefaultInstanceForType() {
+      return DEFAULT_INSTANCE;
+    }
+
+  }
+
+  public static final int PER_PROCESS_GPU_MEMORY_FRACTION_FIELD_NUMBER = 1;
+  private double perProcessGpuMemoryFraction_;
+  /**
+   * <pre>
+   * Fraction of the available GPU memory to allocate for each process.
+   * 1 means to allocate all of the GPU memory, 0.5 means the process
+   * allocates up to ~50% of the available GPU memory.
+   * GPU memory is pre-allocated unless the allow_growth option is enabled.
+   * If greater than 1.0, uses CUDA unified memory to potentially oversubscribe
+   * the amount of memory available on the GPU device by using host memory as a
+   * swap space. Accessing memory not available on the device will be
+   * significantly slower as that would require memory transfer between the host
+   * and the device. Options to reduce the memory requirement should be
+   * considered before enabling this option as this may come with a negative
+   * performance impact. Oversubscription using the unified memory requires
+   * Pascal class or newer GPUs and it is currently only supported on the Linux
+   * operating system. See
+   * https://docs.nvidia.com/cuda/cuda-c-programming-guide/index.html#um-requirements
+   * for the detailed requirements.
+   * </pre>
+   *
+   * <code>double per_process_gpu_memory_fraction = 1;</code>
+   */
+  public double getPerProcessGpuMemoryFraction() {
+    return perProcessGpuMemoryFraction_;
+  }
+
+  public static final int ALLOW_GROWTH_FIELD_NUMBER = 4;
+  private boolean allowGrowth_;
+  /**
+   * <pre>
+   * If true, the allocator does not pre-allocate the entire specified
+   * GPU memory region, instead starting small and growing as needed.
+   * </pre>
+   *
+   * <code>bool allow_growth = 4;</code>
+   */
+  public boolean getAllowGrowth() {
+    return allowGrowth_;
+  }
+
+  public static final int ALLOCATOR_TYPE_FIELD_NUMBER = 2;
+  private volatile java.lang.Object allocatorType_;
+  /**
+   * <pre>
+   * The type of GPU allocation strategy to use.
+   * Allowed values:
+   * "": The empty string (default) uses a system-chosen default
+   *     which may change over time.
+   * "BFC": A "Best-fit with coalescing" algorithm, simplified from a
+   *        version of dlmalloc.
+   * </pre>
+   *
+   * <code>string allocator_type = 2;</code>
+   */
+  public java.lang.String getAllocatorType() {
+    java.lang.Object ref = allocatorType_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      allocatorType_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * The type of GPU allocation strategy to use.
+   * Allowed values:
+   * "": The empty string (default) uses a system-chosen default
+   *     which may change over time.
+   * "BFC": A "Best-fit with coalescing" algorithm, simplified from a
+   *        version of dlmalloc.
+   * </pre>
+   *
+   * <code>string allocator_type = 2;</code>
+   */
+  public com.google.protobuf.ByteString
+      getAllocatorTypeBytes() {
+    java.lang.Object ref = allocatorType_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      allocatorType_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int DEFERRED_DELETION_BYTES_FIELD_NUMBER = 3;
+  private long deferredDeletionBytes_;
+  /**
+   * <pre>
+   * Delay deletion of up to this many bytes to reduce the number of
+   * interactions with gpu driver code.  If 0, the system chooses
+   * a reasonable default (several MBs).
+   * </pre>
+   *
+   * <code>int64 deferred_deletion_bytes = 3;</code>
+   */
+  public long getDeferredDeletionBytes() {
+    return deferredDeletionBytes_;
+  }
+
+  public static final int VISIBLE_DEVICE_LIST_FIELD_NUMBER = 5;
+  private volatile java.lang.Object visibleDeviceList_;
+  /**
+   * <pre>
+   * A comma-separated list of GPU ids that determines the 'visible'
+   * to 'virtual' mapping of GPU devices.  For example, if TensorFlow
+   * can see 8 GPU devices in the process, and one wanted to map
+   * visible GPU devices 5 and 3 as "/device:GPU:0", and "/device:GPU:1",
+   * then one would specify this field as "5,3".  This field is similar in
+   * spirit to the CUDA_VISIBLE_DEVICES environment variable, except
+   * it applies to the visible GPU devices in the process.
+   * NOTE:
+   * 1. The GPU driver provides the process with the visible GPUs
+   *    in an order which is not guaranteed to have any correlation to
+   *    the *physical* GPU id in the machine.  This field is used for
+   *    remapping "visible" to "virtual", which means this operates only
+   *    after the process starts.  Users are required to use vendor
+   *    specific mechanisms (e.g., CUDA_VISIBLE_DEVICES) to control the
+   *    physical to visible device mapping prior to invoking TensorFlow.
+   * 2. In the code, the ids in this list are also called "platform GPU id"s,
+   *    and the 'virtual' ids of GPU devices (i.e. the ids in the device
+   *    name "/device:GPU:&lt;id&gt;") are also called "TF GPU id"s. Please
+   *    refer to third_party/tensorflow/core/common_runtime/gpu/gpu_id.h
+   *    for more information.
+   * </pre>
+   *
+   * <code>string visible_device_list = 5;</code>
+   */
+  public java.lang.String getVisibleDeviceList() {
+    java.lang.Object ref = visibleDeviceList_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      visibleDeviceList_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * A comma-separated list of GPU ids that determines the 'visible'
+   * to 'virtual' mapping of GPU devices.  For example, if TensorFlow
+   * can see 8 GPU devices in the process, and one wanted to map
+   * visible GPU devices 5 and 3 as "/device:GPU:0", and "/device:GPU:1",
+   * then one would specify this field as "5,3".  This field is similar in
+   * spirit to the CUDA_VISIBLE_DEVICES environment variable, except
+   * it applies to the visible GPU devices in the process.
+   * NOTE:
+   * 1. The GPU driver provides the process with the visible GPUs
+   *    in an order which is not guaranteed to have any correlation to
+   *    the *physical* GPU id in the machine.  This field is used for
+   *    remapping "visible" to "virtual", which means this operates only
+   *    after the process starts.  Users are required to use vendor
+   *    specific mechanisms (e.g., CUDA_VISIBLE_DEVICES) to control the
+   *    physical to visible device mapping prior to invoking TensorFlow.
+   * 2. In the code, the ids in this list are also called "platform GPU id"s,
+   *    and the 'virtual' ids of GPU devices (i.e. the ids in the device
+   *    name "/device:GPU:&lt;id&gt;") are also called "TF GPU id"s. Please
+   *    refer to third_party/tensorflow/core/common_runtime/gpu/gpu_id.h
+   *    for more information.
+   * </pre>
+   *
+   * <code>string visible_device_list = 5;</code>
+   */
+  public com.google.protobuf.ByteString
+      getVisibleDeviceListBytes() {
+    java.lang.Object ref = visibleDeviceList_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      visibleDeviceList_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int POLLING_ACTIVE_DELAY_USECS_FIELD_NUMBER = 6;
+  private int pollingActiveDelayUsecs_;
+  /**
+   * <pre>
+   * In the event polling loop sleep this many microseconds between
+   * PollEvents calls, when the queue is not empty.  If value is not
+   * set or set to 0, gets set to a non-zero default.
+   * </pre>
+   *
+   * <code>int32 polling_active_delay_usecs = 6;</code>
+   */
+  public int getPollingActiveDelayUsecs() {
+    return pollingActiveDelayUsecs_;
+  }
+
+  public static final int POLLING_INACTIVE_DELAY_MSECS_FIELD_NUMBER = 7;
+  private int pollingInactiveDelayMsecs_;
+  /**
+   * <pre>
+   * This field is deprecated and ignored.
+   * </pre>
+   *
+   * <code>int32 polling_inactive_delay_msecs = 7;</code>
+   */
+  public int getPollingInactiveDelayMsecs() {
+    return pollingInactiveDelayMsecs_;
+  }
+
+  public static final int FORCE_GPU_COMPATIBLE_FIELD_NUMBER = 8;
+  private boolean forceGpuCompatible_;
+  /**
+   * <pre>
+   * Force all tensors to be gpu_compatible. On a GPU-enabled TensorFlow,
+   * enabling this option forces all CPU tensors to be allocated with Cuda
+   * pinned memory. Normally, TensorFlow will infer which tensors should be
+   * allocated as the pinned memory. But in case where the inference is
+   * incomplete, this option can significantly speed up the cross-device memory
+   * copy performance as long as it fits the memory.
+   * Note that this option is not something that should be
+   * enabled by default for unknown or very large models, since all Cuda pinned
+   * memory is unpageable, having too much pinned memory might negatively impact
+   * the overall host system performance.
+   * </pre>
+   *
+   * <code>bool force_gpu_compatible = 8;</code>
+   */
+  public boolean getForceGpuCompatible() {
+    return forceGpuCompatible_;
+  }
+
+  public static final int EXPERIMENTAL_FIELD_NUMBER = 9;
+  private org.tensorflow.proto.framework.GPUOptions.Experimental experimental_;
+  /**
+   * <pre>
+   * Everything inside experimental is subject to change and is not subject
+   * to API stability guarantees in
+   * https://www.tensorflow.org/guide/version_compat.
+   * </pre>
+   *
+   * <code>.tensorflow.GPUOptions.Experimental experimental = 9;</code>
+   */
+  public boolean hasExperimental() {
+    return experimental_ != null;
+  }
+  /**
+   * <pre>
+   * Everything inside experimental is subject to change and is not subject
+   * to API stability guarantees in
+   * https://www.tensorflow.org/guide/version_compat.
+   * </pre>
+   *
+   * <code>.tensorflow.GPUOptions.Experimental experimental = 9;</code>
+   */
+  public org.tensorflow.proto.framework.GPUOptions.Experimental getExperimental() {
+    return experimental_ == null ? org.tensorflow.proto.framework.GPUOptions.Experimental.getDefaultInstance() : experimental_;
+  }
+  /**
+   * <pre>
+   * Everything inside experimental is subject to change and is not subject
+   * to API stability guarantees in
+   * https://www.tensorflow.org/guide/version_compat.
+   * </pre>
+   *
+   * <code>.tensorflow.GPUOptions.Experimental experimental = 9;</code>
+   */
+  public org.tensorflow.proto.framework.GPUOptions.ExperimentalOrBuilder getExperimentalOrBuilder() {
+    return getExperimental();
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    if (perProcessGpuMemoryFraction_ != 0D) {
+      output.writeDouble(1, perProcessGpuMemoryFraction_);
+    }
+    if (!getAllocatorTypeBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 2, allocatorType_);
+    }
+    if (deferredDeletionBytes_ != 0L) {
+      output.writeInt64(3, deferredDeletionBytes_);
+    }
+    if (allowGrowth_ != false) {
+      output.writeBool(4, allowGrowth_);
+    }
+    if (!getVisibleDeviceListBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 5, visibleDeviceList_);
+    }
+    if (pollingActiveDelayUsecs_ != 0) {
+      output.writeInt32(6, pollingActiveDelayUsecs_);
+    }
+    if (pollingInactiveDelayMsecs_ != 0) {
+      output.writeInt32(7, pollingInactiveDelayMsecs_);
+    }
+    if (forceGpuCompatible_ != false) {
+      output.writeBool(8, forceGpuCompatible_);
+    }
+    if (experimental_ != null) {
+      output.writeMessage(9, getExperimental());
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (perProcessGpuMemoryFraction_ != 0D) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeDoubleSize(1, perProcessGpuMemoryFraction_);
+    }
+    if (!getAllocatorTypeBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(2, allocatorType_);
+    }
+    if (deferredDeletionBytes_ != 0L) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt64Size(3, deferredDeletionBytes_);
+    }
+    if (allowGrowth_ != false) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeBoolSize(4, allowGrowth_);
+    }
+    if (!getVisibleDeviceListBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(5, visibleDeviceList_);
+    }
+    if (pollingActiveDelayUsecs_ != 0) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt32Size(6, pollingActiveDelayUsecs_);
+    }
+    if (pollingInactiveDelayMsecs_ != 0) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt32Size(7, pollingInactiveDelayMsecs_);
+    }
+    if (forceGpuCompatible_ != false) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeBoolSize(8, forceGpuCompatible_);
+    }
+    if (experimental_ != null) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(9, getExperimental());
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.framework.GPUOptions)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.framework.GPUOptions other = (org.tensorflow.proto.framework.GPUOptions) obj;
+
+    if (java.lang.Double.doubleToLongBits(getPerProcessGpuMemoryFraction())
+        != java.lang.Double.doubleToLongBits(
+            other.getPerProcessGpuMemoryFraction())) return false;
+    if (getAllowGrowth()
+        != other.getAllowGrowth()) return false;
+    if (!getAllocatorType()
+        .equals(other.getAllocatorType())) return false;
+    if (getDeferredDeletionBytes()
+        != other.getDeferredDeletionBytes()) return false;
+    if (!getVisibleDeviceList()
+        .equals(other.getVisibleDeviceList())) return false;
+    if (getPollingActiveDelayUsecs()
+        != other.getPollingActiveDelayUsecs()) return false;
+    if (getPollingInactiveDelayMsecs()
+        != other.getPollingInactiveDelayMsecs()) return false;
+    if (getForceGpuCompatible()
+        != other.getForceGpuCompatible()) return false;
+    if (hasExperimental() != other.hasExperimental()) return false;
+    if (hasExperimental()) {
+      if (!getExperimental()
+          .equals(other.getExperimental())) return false;
+    }
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    hash = (37 * hash) + PER_PROCESS_GPU_MEMORY_FRACTION_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+        java.lang.Double.doubleToLongBits(getPerProcessGpuMemoryFraction()));
+    hash = (37 * hash) + ALLOW_GROWTH_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashBoolean(
+        getAllowGrowth());
+    hash = (37 * hash) + ALLOCATOR_TYPE_FIELD_NUMBER;
+    hash = (53 * hash) + getAllocatorType().hashCode();
+    hash = (37 * hash) + DEFERRED_DELETION_BYTES_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+        getDeferredDeletionBytes());
+    hash = (37 * hash) + VISIBLE_DEVICE_LIST_FIELD_NUMBER;
+    hash = (53 * hash) + getVisibleDeviceList().hashCode();
+    hash = (37 * hash) + POLLING_ACTIVE_DELAY_USECS_FIELD_NUMBER;
+    hash = (53 * hash) + getPollingActiveDelayUsecs();
+    hash = (37 * hash) + POLLING_INACTIVE_DELAY_MSECS_FIELD_NUMBER;
+    hash = (53 * hash) + getPollingInactiveDelayMsecs();
+    hash = (37 * hash) + FORCE_GPU_COMPATIBLE_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashBoolean(
+        getForceGpuCompatible());
+    if (hasExperimental()) {
+      hash = (37 * hash) + EXPERIMENTAL_FIELD_NUMBER;
+      hash = (53 * hash) + getExperimental().hashCode();
+    }
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.framework.GPUOptions parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.GPUOptions parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.GPUOptions parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.GPUOptions parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.GPUOptions parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.GPUOptions parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.GPUOptions parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.GPUOptions parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.GPUOptions parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.GPUOptions parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.GPUOptions parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.GPUOptions parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.framework.GPUOptions prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * Protobuf type {@code tensorflow.GPUOptions}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.GPUOptions)
+      org.tensorflow.proto.framework.GPUOptionsOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_GPUOptions_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_GPUOptions_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.GPUOptions.class, org.tensorflow.proto.framework.GPUOptions.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.framework.GPUOptions.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      perProcessGpuMemoryFraction_ = 0D;
+
+      allowGrowth_ = false;
+
+      allocatorType_ = "";
+
+      deferredDeletionBytes_ = 0L;
+
+      visibleDeviceList_ = "";
+
+      pollingActiveDelayUsecs_ = 0;
+
+      pollingInactiveDelayMsecs_ = 0;
+
+      forceGpuCompatible_ = false;
+
+      if (experimentalBuilder_ == null) {
+        experimental_ = null;
+      } else {
+        experimental_ = null;
+        experimentalBuilder_ = null;
+      }
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_GPUOptions_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.GPUOptions getDefaultInstanceForType() {
+      return org.tensorflow.proto.framework.GPUOptions.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.GPUOptions build() {
+      org.tensorflow.proto.framework.GPUOptions result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.GPUOptions buildPartial() {
+      org.tensorflow.proto.framework.GPUOptions result = new org.tensorflow.proto.framework.GPUOptions(this);
+      result.perProcessGpuMemoryFraction_ = perProcessGpuMemoryFraction_;
+      result.allowGrowth_ = allowGrowth_;
+      result.allocatorType_ = allocatorType_;
+      result.deferredDeletionBytes_ = deferredDeletionBytes_;
+      result.visibleDeviceList_ = visibleDeviceList_;
+      result.pollingActiveDelayUsecs_ = pollingActiveDelayUsecs_;
+      result.pollingInactiveDelayMsecs_ = pollingInactiveDelayMsecs_;
+      result.forceGpuCompatible_ = forceGpuCompatible_;
+      if (experimentalBuilder_ == null) {
+        result.experimental_ = experimental_;
+      } else {
+        result.experimental_ = experimentalBuilder_.build();
+      }
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.framework.GPUOptions) {
+        return mergeFrom((org.tensorflow.proto.framework.GPUOptions)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.framework.GPUOptions other) {
+      if (other == org.tensorflow.proto.framework.GPUOptions.getDefaultInstance()) return this;
+      if (other.getPerProcessGpuMemoryFraction() != 0D) {
+        setPerProcessGpuMemoryFraction(other.getPerProcessGpuMemoryFraction());
+      }
+      if (other.getAllowGrowth() != false) {
+        setAllowGrowth(other.getAllowGrowth());
+      }
+      if (!other.getAllocatorType().isEmpty()) {
+        allocatorType_ = other.allocatorType_;
+        onChanged();
+      }
+      if (other.getDeferredDeletionBytes() != 0L) {
+        setDeferredDeletionBytes(other.getDeferredDeletionBytes());
+      }
+      if (!other.getVisibleDeviceList().isEmpty()) {
+        visibleDeviceList_ = other.visibleDeviceList_;
+        onChanged();
+      }
+      if (other.getPollingActiveDelayUsecs() != 0) {
+        setPollingActiveDelayUsecs(other.getPollingActiveDelayUsecs());
+      }
+      if (other.getPollingInactiveDelayMsecs() != 0) {
+        setPollingInactiveDelayMsecs(other.getPollingInactiveDelayMsecs());
+      }
+      if (other.getForceGpuCompatible() != false) {
+        setForceGpuCompatible(other.getForceGpuCompatible());
+      }
+      if (other.hasExperimental()) {
+        mergeExperimental(other.getExperimental());
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.framework.GPUOptions parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.framework.GPUOptions) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+
+    private double perProcessGpuMemoryFraction_ ;
+    /**
+     * <pre>
+     * Fraction of the available GPU memory to allocate for each process.
+     * 1 means to allocate all of the GPU memory, 0.5 means the process
+     * allocates up to ~50% of the available GPU memory.
+     * GPU memory is pre-allocated unless the allow_growth option is enabled.
+     * If greater than 1.0, uses CUDA unified memory to potentially oversubscribe
+     * the amount of memory available on the GPU device by using host memory as a
+     * swap space. Accessing memory not available on the device will be
+     * significantly slower as that would require memory transfer between the host
+     * and the device. Options to reduce the memory requirement should be
+     * considered before enabling this option as this may come with a negative
+     * performance impact. Oversubscription using the unified memory requires
+     * Pascal class or newer GPUs and it is currently only supported on the Linux
+     * operating system. See
+     * https://docs.nvidia.com/cuda/cuda-c-programming-guide/index.html#um-requirements
+     * for the detailed requirements.
+     * </pre>
+     *
+     * <code>double per_process_gpu_memory_fraction = 1;</code>
+     */
+    public double getPerProcessGpuMemoryFraction() {
+      return perProcessGpuMemoryFraction_;
+    }
+    /**
+     * <pre>
+     * Fraction of the available GPU memory to allocate for each process.
+     * 1 means to allocate all of the GPU memory, 0.5 means the process
+     * allocates up to ~50% of the available GPU memory.
+     * GPU memory is pre-allocated unless the allow_growth option is enabled.
+     * If greater than 1.0, uses CUDA unified memory to potentially oversubscribe
+     * the amount of memory available on the GPU device by using host memory as a
+     * swap space. Accessing memory not available on the device will be
+     * significantly slower as that would require memory transfer between the host
+     * and the device. Options to reduce the memory requirement should be
+     * considered before enabling this option as this may come with a negative
+     * performance impact. Oversubscription using the unified memory requires
+     * Pascal class or newer GPUs and it is currently only supported on the Linux
+     * operating system. See
+     * https://docs.nvidia.com/cuda/cuda-c-programming-guide/index.html#um-requirements
+     * for the detailed requirements.
+     * </pre>
+     *
+     * <code>double per_process_gpu_memory_fraction = 1;</code>
+     */
+    public Builder setPerProcessGpuMemoryFraction(double value) {
+      
+      perProcessGpuMemoryFraction_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Fraction of the available GPU memory to allocate for each process.
+     * 1 means to allocate all of the GPU memory, 0.5 means the process
+     * allocates up to ~50% of the available GPU memory.
+     * GPU memory is pre-allocated unless the allow_growth option is enabled.
+     * If greater than 1.0, uses CUDA unified memory to potentially oversubscribe
+     * the amount of memory available on the GPU device by using host memory as a
+     * swap space. Accessing memory not available on the device will be
+     * significantly slower as that would require memory transfer between the host
+     * and the device. Options to reduce the memory requirement should be
+     * considered before enabling this option as this may come with a negative
+     * performance impact. Oversubscription using the unified memory requires
+     * Pascal class or newer GPUs and it is currently only supported on the Linux
+     * operating system. See
+     * https://docs.nvidia.com/cuda/cuda-c-programming-guide/index.html#um-requirements
+     * for the detailed requirements.
+     * </pre>
+     *
+     * <code>double per_process_gpu_memory_fraction = 1;</code>
+     */
+    public Builder clearPerProcessGpuMemoryFraction() {
+      
+      perProcessGpuMemoryFraction_ = 0D;
+      onChanged();
+      return this;
+    }
+
+    private boolean allowGrowth_ ;
+    /**
+     * <pre>
+     * If true, the allocator does not pre-allocate the entire specified
+     * GPU memory region, instead starting small and growing as needed.
+     * </pre>
+     *
+     * <code>bool allow_growth = 4;</code>
+     */
+    public boolean getAllowGrowth() {
+      return allowGrowth_;
+    }
+    /**
+     * <pre>
+     * If true, the allocator does not pre-allocate the entire specified
+     * GPU memory region, instead starting small and growing as needed.
+     * </pre>
+     *
+     * <code>bool allow_growth = 4;</code>
+     */
+    public Builder setAllowGrowth(boolean value) {
+      
+      allowGrowth_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * If true, the allocator does not pre-allocate the entire specified
+     * GPU memory region, instead starting small and growing as needed.
+     * </pre>
+     *
+     * <code>bool allow_growth = 4;</code>
+     */
+    public Builder clearAllowGrowth() {
+      
+      allowGrowth_ = false;
+      onChanged();
+      return this;
+    }
+
+    private java.lang.Object allocatorType_ = "";
+    /**
+     * <pre>
+     * The type of GPU allocation strategy to use.
+     * Allowed values:
+     * "": The empty string (default) uses a system-chosen default
+     *     which may change over time.
+     * "BFC": A "Best-fit with coalescing" algorithm, simplified from a
+     *        version of dlmalloc.
+     * </pre>
+     *
+     * <code>string allocator_type = 2;</code>
+     */
+    public java.lang.String getAllocatorType() {
+      java.lang.Object ref = allocatorType_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        allocatorType_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * The type of GPU allocation strategy to use.
+     * Allowed values:
+     * "": The empty string (default) uses a system-chosen default
+     *     which may change over time.
+     * "BFC": A "Best-fit with coalescing" algorithm, simplified from a
+     *        version of dlmalloc.
+     * </pre>
+     *
+     * <code>string allocator_type = 2;</code>
+     */
+    public com.google.protobuf.ByteString
+        getAllocatorTypeBytes() {
+      java.lang.Object ref = allocatorType_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        allocatorType_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * The type of GPU allocation strategy to use.
+     * Allowed values:
+     * "": The empty string (default) uses a system-chosen default
+     *     which may change over time.
+     * "BFC": A "Best-fit with coalescing" algorithm, simplified from a
+     *        version of dlmalloc.
+     * </pre>
+     *
+     * <code>string allocator_type = 2;</code>
+     */
+    public Builder setAllocatorType(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      allocatorType_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * The type of GPU allocation strategy to use.
+     * Allowed values:
+     * "": The empty string (default) uses a system-chosen default
+     *     which may change over time.
+     * "BFC": A "Best-fit with coalescing" algorithm, simplified from a
+     *        version of dlmalloc.
+     * </pre>
+     *
+     * <code>string allocator_type = 2;</code>
+     */
+    public Builder clearAllocatorType() {
+      
+      allocatorType_ = getDefaultInstance().getAllocatorType();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * The type of GPU allocation strategy to use.
+     * Allowed values:
+     * "": The empty string (default) uses a system-chosen default
+     *     which may change over time.
+     * "BFC": A "Best-fit with coalescing" algorithm, simplified from a
+     *        version of dlmalloc.
+     * </pre>
+     *
+     * <code>string allocator_type = 2;</code>
+     */
+    public Builder setAllocatorTypeBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      allocatorType_ = value;
+      onChanged();
+      return this;
+    }
+
+    private long deferredDeletionBytes_ ;
+    /**
+     * <pre>
+     * Delay deletion of up to this many bytes to reduce the number of
+     * interactions with gpu driver code.  If 0, the system chooses
+     * a reasonable default (several MBs).
+     * </pre>
+     *
+     * <code>int64 deferred_deletion_bytes = 3;</code>
+     */
+    public long getDeferredDeletionBytes() {
+      return deferredDeletionBytes_;
+    }
+    /**
+     * <pre>
+     * Delay deletion of up to this many bytes to reduce the number of
+     * interactions with gpu driver code.  If 0, the system chooses
+     * a reasonable default (several MBs).
+     * </pre>
+     *
+     * <code>int64 deferred_deletion_bytes = 3;</code>
+     */
+    public Builder setDeferredDeletionBytes(long value) {
+      
+      deferredDeletionBytes_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Delay deletion of up to this many bytes to reduce the number of
+     * interactions with gpu driver code.  If 0, the system chooses
+     * a reasonable default (several MBs).
+     * </pre>
+     *
+     * <code>int64 deferred_deletion_bytes = 3;</code>
+     */
+    public Builder clearDeferredDeletionBytes() {
+      
+      deferredDeletionBytes_ = 0L;
+      onChanged();
+      return this;
+    }
+
+    private java.lang.Object visibleDeviceList_ = "";
+    /**
+     * <pre>
+     * A comma-separated list of GPU ids that determines the 'visible'
+     * to 'virtual' mapping of GPU devices.  For example, if TensorFlow
+     * can see 8 GPU devices in the process, and one wanted to map
+     * visible GPU devices 5 and 3 as "/device:GPU:0", and "/device:GPU:1",
+     * then one would specify this field as "5,3".  This field is similar in
+     * spirit to the CUDA_VISIBLE_DEVICES environment variable, except
+     * it applies to the visible GPU devices in the process.
+     * NOTE:
+     * 1. The GPU driver provides the process with the visible GPUs
+     *    in an order which is not guaranteed to have any correlation to
+     *    the *physical* GPU id in the machine.  This field is used for
+     *    remapping "visible" to "virtual", which means this operates only
+     *    after the process starts.  Users are required to use vendor
+     *    specific mechanisms (e.g., CUDA_VISIBLE_DEVICES) to control the
+     *    physical to visible device mapping prior to invoking TensorFlow.
+     * 2. In the code, the ids in this list are also called "platform GPU id"s,
+     *    and the 'virtual' ids of GPU devices (i.e. the ids in the device
+     *    name "/device:GPU:&lt;id&gt;") are also called "TF GPU id"s. Please
+     *    refer to third_party/tensorflow/core/common_runtime/gpu/gpu_id.h
+     *    for more information.
+     * </pre>
+     *
+     * <code>string visible_device_list = 5;</code>
+     */
+    public java.lang.String getVisibleDeviceList() {
+      java.lang.Object ref = visibleDeviceList_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        visibleDeviceList_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * A comma-separated list of GPU ids that determines the 'visible'
+     * to 'virtual' mapping of GPU devices.  For example, if TensorFlow
+     * can see 8 GPU devices in the process, and one wanted to map
+     * visible GPU devices 5 and 3 as "/device:GPU:0", and "/device:GPU:1",
+     * then one would specify this field as "5,3".  This field is similar in
+     * spirit to the CUDA_VISIBLE_DEVICES environment variable, except
+     * it applies to the visible GPU devices in the process.
+     * NOTE:
+     * 1. The GPU driver provides the process with the visible GPUs
+     *    in an order which is not guaranteed to have any correlation to
+     *    the *physical* GPU id in the machine.  This field is used for
+     *    remapping "visible" to "virtual", which means this operates only
+     *    after the process starts.  Users are required to use vendor
+     *    specific mechanisms (e.g., CUDA_VISIBLE_DEVICES) to control the
+     *    physical to visible device mapping prior to invoking TensorFlow.
+     * 2. In the code, the ids in this list are also called "platform GPU id"s,
+     *    and the 'virtual' ids of GPU devices (i.e. the ids in the device
+     *    name "/device:GPU:&lt;id&gt;") are also called "TF GPU id"s. Please
+     *    refer to third_party/tensorflow/core/common_runtime/gpu/gpu_id.h
+     *    for more information.
+     * </pre>
+     *
+     * <code>string visible_device_list = 5;</code>
+     */
+    public com.google.protobuf.ByteString
+        getVisibleDeviceListBytes() {
+      java.lang.Object ref = visibleDeviceList_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        visibleDeviceList_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * A comma-separated list of GPU ids that determines the 'visible'
+     * to 'virtual' mapping of GPU devices.  For example, if TensorFlow
+     * can see 8 GPU devices in the process, and one wanted to map
+     * visible GPU devices 5 and 3 as "/device:GPU:0", and "/device:GPU:1",
+     * then one would specify this field as "5,3".  This field is similar in
+     * spirit to the CUDA_VISIBLE_DEVICES environment variable, except
+     * it applies to the visible GPU devices in the process.
+     * NOTE:
+     * 1. The GPU driver provides the process with the visible GPUs
+     *    in an order which is not guaranteed to have any correlation to
+     *    the *physical* GPU id in the machine.  This field is used for
+     *    remapping "visible" to "virtual", which means this operates only
+     *    after the process starts.  Users are required to use vendor
+     *    specific mechanisms (e.g., CUDA_VISIBLE_DEVICES) to control the
+     *    physical to visible device mapping prior to invoking TensorFlow.
+     * 2. In the code, the ids in this list are also called "platform GPU id"s,
+     *    and the 'virtual' ids of GPU devices (i.e. the ids in the device
+     *    name "/device:GPU:&lt;id&gt;") are also called "TF GPU id"s. Please
+     *    refer to third_party/tensorflow/core/common_runtime/gpu/gpu_id.h
+     *    for more information.
+     * </pre>
+     *
+     * <code>string visible_device_list = 5;</code>
+     */
+    public Builder setVisibleDeviceList(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      visibleDeviceList_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * A comma-separated list of GPU ids that determines the 'visible'
+     * to 'virtual' mapping of GPU devices.  For example, if TensorFlow
+     * can see 8 GPU devices in the process, and one wanted to map
+     * visible GPU devices 5 and 3 as "/device:GPU:0", and "/device:GPU:1",
+     * then one would specify this field as "5,3".  This field is similar in
+     * spirit to the CUDA_VISIBLE_DEVICES environment variable, except
+     * it applies to the visible GPU devices in the process.
+     * NOTE:
+     * 1. The GPU driver provides the process with the visible GPUs
+     *    in an order which is not guaranteed to have any correlation to
+     *    the *physical* GPU id in the machine.  This field is used for
+     *    remapping "visible" to "virtual", which means this operates only
+     *    after the process starts.  Users are required to use vendor
+     *    specific mechanisms (e.g., CUDA_VISIBLE_DEVICES) to control the
+     *    physical to visible device mapping prior to invoking TensorFlow.
+     * 2. In the code, the ids in this list are also called "platform GPU id"s,
+     *    and the 'virtual' ids of GPU devices (i.e. the ids in the device
+     *    name "/device:GPU:&lt;id&gt;") are also called "TF GPU id"s. Please
+     *    refer to third_party/tensorflow/core/common_runtime/gpu/gpu_id.h
+     *    for more information.
+     * </pre>
+     *
+     * <code>string visible_device_list = 5;</code>
+     */
+    public Builder clearVisibleDeviceList() {
+      
+      visibleDeviceList_ = getDefaultInstance().getVisibleDeviceList();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * A comma-separated list of GPU ids that determines the 'visible'
+     * to 'virtual' mapping of GPU devices.  For example, if TensorFlow
+     * can see 8 GPU devices in the process, and one wanted to map
+     * visible GPU devices 5 and 3 as "/device:GPU:0", and "/device:GPU:1",
+     * then one would specify this field as "5,3".  This field is similar in
+     * spirit to the CUDA_VISIBLE_DEVICES environment variable, except
+     * it applies to the visible GPU devices in the process.
+     * NOTE:
+     * 1. The GPU driver provides the process with the visible GPUs
+     *    in an order which is not guaranteed to have any correlation to
+     *    the *physical* GPU id in the machine.  This field is used for
+     *    remapping "visible" to "virtual", which means this operates only
+     *    after the process starts.  Users are required to use vendor
+     *    specific mechanisms (e.g., CUDA_VISIBLE_DEVICES) to control the
+     *    physical to visible device mapping prior to invoking TensorFlow.
+     * 2. In the code, the ids in this list are also called "platform GPU id"s,
+     *    and the 'virtual' ids of GPU devices (i.e. the ids in the device
+     *    name "/device:GPU:&lt;id&gt;") are also called "TF GPU id"s. Please
+     *    refer to third_party/tensorflow/core/common_runtime/gpu/gpu_id.h
+     *    for more information.
+     * </pre>
+     *
+     * <code>string visible_device_list = 5;</code>
+     */
+    public Builder setVisibleDeviceListBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      visibleDeviceList_ = value;
+      onChanged();
+      return this;
+    }
+
+    private int pollingActiveDelayUsecs_ ;
+    /**
+     * <pre>
+     * In the event polling loop sleep this many microseconds between
+     * PollEvents calls, when the queue is not empty.  If value is not
+     * set or set to 0, gets set to a non-zero default.
+     * </pre>
+     *
+     * <code>int32 polling_active_delay_usecs = 6;</code>
+     */
+    public int getPollingActiveDelayUsecs() {
+      return pollingActiveDelayUsecs_;
+    }
+    /**
+     * <pre>
+     * In the event polling loop sleep this many microseconds between
+     * PollEvents calls, when the queue is not empty.  If value is not
+     * set or set to 0, gets set to a non-zero default.
+     * </pre>
+     *
+     * <code>int32 polling_active_delay_usecs = 6;</code>
+     */
+    public Builder setPollingActiveDelayUsecs(int value) {
+      
+      pollingActiveDelayUsecs_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * In the event polling loop sleep this many microseconds between
+     * PollEvents calls, when the queue is not empty.  If value is not
+     * set or set to 0, gets set to a non-zero default.
+     * </pre>
+     *
+     * <code>int32 polling_active_delay_usecs = 6;</code>
+     */
+    public Builder clearPollingActiveDelayUsecs() {
+      
+      pollingActiveDelayUsecs_ = 0;
+      onChanged();
+      return this;
+    }
+
+    private int pollingInactiveDelayMsecs_ ;
+    /**
+     * <pre>
+     * This field is deprecated and ignored.
+     * </pre>
+     *
+     * <code>int32 polling_inactive_delay_msecs = 7;</code>
+     */
+    public int getPollingInactiveDelayMsecs() {
+      return pollingInactiveDelayMsecs_;
+    }
+    /**
+     * <pre>
+     * This field is deprecated and ignored.
+     * </pre>
+     *
+     * <code>int32 polling_inactive_delay_msecs = 7;</code>
+     */
+    public Builder setPollingInactiveDelayMsecs(int value) {
+      
+      pollingInactiveDelayMsecs_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * This field is deprecated and ignored.
+     * </pre>
+     *
+     * <code>int32 polling_inactive_delay_msecs = 7;</code>
+     */
+    public Builder clearPollingInactiveDelayMsecs() {
+      
+      pollingInactiveDelayMsecs_ = 0;
+      onChanged();
+      return this;
+    }
+
+    private boolean forceGpuCompatible_ ;
+    /**
+     * <pre>
+     * Force all tensors to be gpu_compatible. On a GPU-enabled TensorFlow,
+     * enabling this option forces all CPU tensors to be allocated with Cuda
+     * pinned memory. Normally, TensorFlow will infer which tensors should be
+     * allocated as the pinned memory. But in case where the inference is
+     * incomplete, this option can significantly speed up the cross-device memory
+     * copy performance as long as it fits the memory.
+     * Note that this option is not something that should be
+     * enabled by default for unknown or very large models, since all Cuda pinned
+     * memory is unpageable, having too much pinned memory might negatively impact
+     * the overall host system performance.
+     * </pre>
+     *
+     * <code>bool force_gpu_compatible = 8;</code>
+     */
+    public boolean getForceGpuCompatible() {
+      return forceGpuCompatible_;
+    }
+    /**
+     * <pre>
+     * Force all tensors to be gpu_compatible. On a GPU-enabled TensorFlow,
+     * enabling this option forces all CPU tensors to be allocated with Cuda
+     * pinned memory. Normally, TensorFlow will infer which tensors should be
+     * allocated as the pinned memory. But in case where the inference is
+     * incomplete, this option can significantly speed up the cross-device memory
+     * copy performance as long as it fits the memory.
+     * Note that this option is not something that should be
+     * enabled by default for unknown or very large models, since all Cuda pinned
+     * memory is unpageable, having too much pinned memory might negatively impact
+     * the overall host system performance.
+     * </pre>
+     *
+     * <code>bool force_gpu_compatible = 8;</code>
+     */
+    public Builder setForceGpuCompatible(boolean value) {
+      
+      forceGpuCompatible_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Force all tensors to be gpu_compatible. On a GPU-enabled TensorFlow,
+     * enabling this option forces all CPU tensors to be allocated with Cuda
+     * pinned memory. Normally, TensorFlow will infer which tensors should be
+     * allocated as the pinned memory. But in case where the inference is
+     * incomplete, this option can significantly speed up the cross-device memory
+     * copy performance as long as it fits the memory.
+     * Note that this option is not something that should be
+     * enabled by default for unknown or very large models, since all Cuda pinned
+     * memory is unpageable, having too much pinned memory might negatively impact
+     * the overall host system performance.
+     * </pre>
+     *
+     * <code>bool force_gpu_compatible = 8;</code>
+     */
+    public Builder clearForceGpuCompatible() {
+      
+      forceGpuCompatible_ = false;
+      onChanged();
+      return this;
+    }
+
+    private org.tensorflow.proto.framework.GPUOptions.Experimental experimental_;
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.GPUOptions.Experimental, org.tensorflow.proto.framework.GPUOptions.Experimental.Builder, org.tensorflow.proto.framework.GPUOptions.ExperimentalOrBuilder> experimentalBuilder_;
+    /**
+     * <pre>
+     * Everything inside experimental is subject to change and is not subject
+     * to API stability guarantees in
+     * https://www.tensorflow.org/guide/version_compat.
+     * </pre>
+     *
+     * <code>.tensorflow.GPUOptions.Experimental experimental = 9;</code>
+     */
+    public boolean hasExperimental() {
+      return experimentalBuilder_ != null || experimental_ != null;
+    }
+    /**
+     * <pre>
+     * Everything inside experimental is subject to change and is not subject
+     * to API stability guarantees in
+     * https://www.tensorflow.org/guide/version_compat.
+     * </pre>
+     *
+     * <code>.tensorflow.GPUOptions.Experimental experimental = 9;</code>
+     */
+    public org.tensorflow.proto.framework.GPUOptions.Experimental getExperimental() {
+      if (experimentalBuilder_ == null) {
+        return experimental_ == null ? org.tensorflow.proto.framework.GPUOptions.Experimental.getDefaultInstance() : experimental_;
+      } else {
+        return experimentalBuilder_.getMessage();
+      }
+    }
+    /**
+     * <pre>
+     * Everything inside experimental is subject to change and is not subject
+     * to API stability guarantees in
+     * https://www.tensorflow.org/guide/version_compat.
+     * </pre>
+     *
+     * <code>.tensorflow.GPUOptions.Experimental experimental = 9;</code>
+     */
+    public Builder setExperimental(org.tensorflow.proto.framework.GPUOptions.Experimental value) {
+      if (experimentalBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        experimental_ = value;
+        onChanged();
+      } else {
+        experimentalBuilder_.setMessage(value);
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Everything inside experimental is subject to change and is not subject
+     * to API stability guarantees in
+     * https://www.tensorflow.org/guide/version_compat.
+     * </pre>
+     *
+     * <code>.tensorflow.GPUOptions.Experimental experimental = 9;</code>
+     */
+    public Builder setExperimental(
+        org.tensorflow.proto.framework.GPUOptions.Experimental.Builder builderForValue) {
+      if (experimentalBuilder_ == null) {
+        experimental_ = builderForValue.build();
+        onChanged();
+      } else {
+        experimentalBuilder_.setMessage(builderForValue.build());
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Everything inside experimental is subject to change and is not subject
+     * to API stability guarantees in
+     * https://www.tensorflow.org/guide/version_compat.
+     * </pre>
+     *
+     * <code>.tensorflow.GPUOptions.Experimental experimental = 9;</code>
+     */
+    public Builder mergeExperimental(org.tensorflow.proto.framework.GPUOptions.Experimental value) {
+      if (experimentalBuilder_ == null) {
+        if (experimental_ != null) {
+          experimental_ =
+            org.tensorflow.proto.framework.GPUOptions.Experimental.newBuilder(experimental_).mergeFrom(value).buildPartial();
+        } else {
+          experimental_ = value;
+        }
+        onChanged();
+      } else {
+        experimentalBuilder_.mergeFrom(value);
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Everything inside experimental is subject to change and is not subject
+     * to API stability guarantees in
+     * https://www.tensorflow.org/guide/version_compat.
+     * </pre>
+     *
+     * <code>.tensorflow.GPUOptions.Experimental experimental = 9;</code>
+     */
+    public Builder clearExperimental() {
+      if (experimentalBuilder_ == null) {
+        experimental_ = null;
+        onChanged();
+      } else {
+        experimental_ = null;
+        experimentalBuilder_ = null;
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Everything inside experimental is subject to change and is not subject
+     * to API stability guarantees in
+     * https://www.tensorflow.org/guide/version_compat.
+     * </pre>
+     *
+     * <code>.tensorflow.GPUOptions.Experimental experimental = 9;</code>
+     */
+    public org.tensorflow.proto.framework.GPUOptions.Experimental.Builder getExperimentalBuilder() {
+      
+      onChanged();
+      return getExperimentalFieldBuilder().getBuilder();
+    }
+    /**
+     * <pre>
+     * Everything inside experimental is subject to change and is not subject
+     * to API stability guarantees in
+     * https://www.tensorflow.org/guide/version_compat.
+     * </pre>
+     *
+     * <code>.tensorflow.GPUOptions.Experimental experimental = 9;</code>
+     */
+    public org.tensorflow.proto.framework.GPUOptions.ExperimentalOrBuilder getExperimentalOrBuilder() {
+      if (experimentalBuilder_ != null) {
+        return experimentalBuilder_.getMessageOrBuilder();
+      } else {
+        return experimental_ == null ?
+            org.tensorflow.proto.framework.GPUOptions.Experimental.getDefaultInstance() : experimental_;
+      }
+    }
+    /**
+     * <pre>
+     * Everything inside experimental is subject to change and is not subject
+     * to API stability guarantees in
+     * https://www.tensorflow.org/guide/version_compat.
+     * </pre>
+     *
+     * <code>.tensorflow.GPUOptions.Experimental experimental = 9;</code>
+     */
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.GPUOptions.Experimental, org.tensorflow.proto.framework.GPUOptions.Experimental.Builder, org.tensorflow.proto.framework.GPUOptions.ExperimentalOrBuilder> 
+        getExperimentalFieldBuilder() {
+      if (experimentalBuilder_ == null) {
+        experimentalBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+            org.tensorflow.proto.framework.GPUOptions.Experimental, org.tensorflow.proto.framework.GPUOptions.Experimental.Builder, org.tensorflow.proto.framework.GPUOptions.ExperimentalOrBuilder>(
+                getExperimental(),
+                getParentForChildren(),
+                isClean());
+        experimental_ = null;
+      }
+      return experimentalBuilder_;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.GPUOptions)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.GPUOptions)
+  private static final org.tensorflow.proto.framework.GPUOptions DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.GPUOptions();
+  }
+
+  public static org.tensorflow.proto.framework.GPUOptions getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<GPUOptions>
+      PARSER = new com.google.protobuf.AbstractParser<GPUOptions>() {
+    @java.lang.Override
+    public GPUOptions parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new GPUOptions(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<GPUOptions> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<GPUOptions> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.framework.GPUOptions getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GPUOptionsOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GPUOptionsOrBuilder.java
new file mode 100644
index 00000000000..6f11472d49a
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GPUOptionsOrBuilder.java
@@ -0,0 +1,206 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/config.proto
+
+package org.tensorflow.proto.framework;
+
+public interface GPUOptionsOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.GPUOptions)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <pre>
+   * Fraction of the available GPU memory to allocate for each process.
+   * 1 means to allocate all of the GPU memory, 0.5 means the process
+   * allocates up to ~50% of the available GPU memory.
+   * GPU memory is pre-allocated unless the allow_growth option is enabled.
+   * If greater than 1.0, uses CUDA unified memory to potentially oversubscribe
+   * the amount of memory available on the GPU device by using host memory as a
+   * swap space. Accessing memory not available on the device will be
+   * significantly slower as that would require memory transfer between the host
+   * and the device. Options to reduce the memory requirement should be
+   * considered before enabling this option as this may come with a negative
+   * performance impact. Oversubscription using the unified memory requires
+   * Pascal class or newer GPUs and it is currently only supported on the Linux
+   * operating system. See
+   * https://docs.nvidia.com/cuda/cuda-c-programming-guide/index.html#um-requirements
+   * for the detailed requirements.
+   * </pre>
+   *
+   * <code>double per_process_gpu_memory_fraction = 1;</code>
+   */
+  double getPerProcessGpuMemoryFraction();
+
+  /**
+   * <pre>
+   * If true, the allocator does not pre-allocate the entire specified
+   * GPU memory region, instead starting small and growing as needed.
+   * </pre>
+   *
+   * <code>bool allow_growth = 4;</code>
+   */
+  boolean getAllowGrowth();
+
+  /**
+   * <pre>
+   * The type of GPU allocation strategy to use.
+   * Allowed values:
+   * "": The empty string (default) uses a system-chosen default
+   *     which may change over time.
+   * "BFC": A "Best-fit with coalescing" algorithm, simplified from a
+   *        version of dlmalloc.
+   * </pre>
+   *
+   * <code>string allocator_type = 2;</code>
+   */
+  java.lang.String getAllocatorType();
+  /**
+   * <pre>
+   * The type of GPU allocation strategy to use.
+   * Allowed values:
+   * "": The empty string (default) uses a system-chosen default
+   *     which may change over time.
+   * "BFC": A "Best-fit with coalescing" algorithm, simplified from a
+   *        version of dlmalloc.
+   * </pre>
+   *
+   * <code>string allocator_type = 2;</code>
+   */
+  com.google.protobuf.ByteString
+      getAllocatorTypeBytes();
+
+  /**
+   * <pre>
+   * Delay deletion of up to this many bytes to reduce the number of
+   * interactions with gpu driver code.  If 0, the system chooses
+   * a reasonable default (several MBs).
+   * </pre>
+   *
+   * <code>int64 deferred_deletion_bytes = 3;</code>
+   */
+  long getDeferredDeletionBytes();
+
+  /**
+   * <pre>
+   * A comma-separated list of GPU ids that determines the 'visible'
+   * to 'virtual' mapping of GPU devices.  For example, if TensorFlow
+   * can see 8 GPU devices in the process, and one wanted to map
+   * visible GPU devices 5 and 3 as "/device:GPU:0", and "/device:GPU:1",
+   * then one would specify this field as "5,3".  This field is similar in
+   * spirit to the CUDA_VISIBLE_DEVICES environment variable, except
+   * it applies to the visible GPU devices in the process.
+   * NOTE:
+   * 1. The GPU driver provides the process with the visible GPUs
+   *    in an order which is not guaranteed to have any correlation to
+   *    the *physical* GPU id in the machine.  This field is used for
+   *    remapping "visible" to "virtual", which means this operates only
+   *    after the process starts.  Users are required to use vendor
+   *    specific mechanisms (e.g., CUDA_VISIBLE_DEVICES) to control the
+   *    physical to visible device mapping prior to invoking TensorFlow.
+   * 2. In the code, the ids in this list are also called "platform GPU id"s,
+   *    and the 'virtual' ids of GPU devices (i.e. the ids in the device
+   *    name "/device:GPU:&lt;id&gt;") are also called "TF GPU id"s. Please
+   *    refer to third_party/tensorflow/core/common_runtime/gpu/gpu_id.h
+   *    for more information.
+   * </pre>
+   *
+   * <code>string visible_device_list = 5;</code>
+   */
+  java.lang.String getVisibleDeviceList();
+  /**
+   * <pre>
+   * A comma-separated list of GPU ids that determines the 'visible'
+   * to 'virtual' mapping of GPU devices.  For example, if TensorFlow
+   * can see 8 GPU devices in the process, and one wanted to map
+   * visible GPU devices 5 and 3 as "/device:GPU:0", and "/device:GPU:1",
+   * then one would specify this field as "5,3".  This field is similar in
+   * spirit to the CUDA_VISIBLE_DEVICES environment variable, except
+   * it applies to the visible GPU devices in the process.
+   * NOTE:
+   * 1. The GPU driver provides the process with the visible GPUs
+   *    in an order which is not guaranteed to have any correlation to
+   *    the *physical* GPU id in the machine.  This field is used for
+   *    remapping "visible" to "virtual", which means this operates only
+   *    after the process starts.  Users are required to use vendor
+   *    specific mechanisms (e.g., CUDA_VISIBLE_DEVICES) to control the
+   *    physical to visible device mapping prior to invoking TensorFlow.
+   * 2. In the code, the ids in this list are also called "platform GPU id"s,
+   *    and the 'virtual' ids of GPU devices (i.e. the ids in the device
+   *    name "/device:GPU:&lt;id&gt;") are also called "TF GPU id"s. Please
+   *    refer to third_party/tensorflow/core/common_runtime/gpu/gpu_id.h
+   *    for more information.
+   * </pre>
+   *
+   * <code>string visible_device_list = 5;</code>
+   */
+  com.google.protobuf.ByteString
+      getVisibleDeviceListBytes();
+
+  /**
+   * <pre>
+   * In the event polling loop sleep this many microseconds between
+   * PollEvents calls, when the queue is not empty.  If value is not
+   * set or set to 0, gets set to a non-zero default.
+   * </pre>
+   *
+   * <code>int32 polling_active_delay_usecs = 6;</code>
+   */
+  int getPollingActiveDelayUsecs();
+
+  /**
+   * <pre>
+   * This field is deprecated and ignored.
+   * </pre>
+   *
+   * <code>int32 polling_inactive_delay_msecs = 7;</code>
+   */
+  int getPollingInactiveDelayMsecs();
+
+  /**
+   * <pre>
+   * Force all tensors to be gpu_compatible. On a GPU-enabled TensorFlow,
+   * enabling this option forces all CPU tensors to be allocated with Cuda
+   * pinned memory. Normally, TensorFlow will infer which tensors should be
+   * allocated as the pinned memory. But in case where the inference is
+   * incomplete, this option can significantly speed up the cross-device memory
+   * copy performance as long as it fits the memory.
+   * Note that this option is not something that should be
+   * enabled by default for unknown or very large models, since all Cuda pinned
+   * memory is unpageable, having too much pinned memory might negatively impact
+   * the overall host system performance.
+   * </pre>
+   *
+   * <code>bool force_gpu_compatible = 8;</code>
+   */
+  boolean getForceGpuCompatible();
+
+  /**
+   * <pre>
+   * Everything inside experimental is subject to change and is not subject
+   * to API stability guarantees in
+   * https://www.tensorflow.org/guide/version_compat.
+   * </pre>
+   *
+   * <code>.tensorflow.GPUOptions.Experimental experimental = 9;</code>
+   */
+  boolean hasExperimental();
+  /**
+   * <pre>
+   * Everything inside experimental is subject to change and is not subject
+   * to API stability guarantees in
+   * https://www.tensorflow.org/guide/version_compat.
+   * </pre>
+   *
+   * <code>.tensorflow.GPUOptions.Experimental experimental = 9;</code>
+   */
+  org.tensorflow.proto.framework.GPUOptions.Experimental getExperimental();
+  /**
+   * <pre>
+   * Everything inside experimental is subject to change and is not subject
+   * to API stability guarantees in
+   * https://www.tensorflow.org/guide/version_compat.
+   * </pre>
+   *
+   * <code>.tensorflow.GPUOptions.Experimental experimental = 9;</code>
+   */
+  org.tensorflow.proto.framework.GPUOptions.ExperimentalOrBuilder getExperimentalOrBuilder();
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GradientDef.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GradientDef.java
new file mode 100644
index 00000000000..fe1126cb2ef
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GradientDef.java
@@ -0,0 +1,765 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/framework/function.proto
+
+package org.tensorflow.proto.framework;
+
+/**
+ * <pre>
+ * GradientDef defines the gradient function of a function defined in
+ * a function library.
+ * A gradient function g (specified by gradient_func) for a function f
+ * (specified by function_name) must follow the following:
+ * The function 'f' must be a numerical function which takes N inputs
+ * and produces M outputs. Its gradient function 'g', which is a
+ * function taking N + M inputs and produces N outputs.
+ * I.e. if we have
+ *    (y1, y2, ..., y_M) = f(x1, x2, ..., x_N),
+ * then, g is
+ *    (dL/dx1, dL/dx2, ..., dL/dx_N) = g(x1, x2, ..., x_N,
+ *                                      dL/dy1, dL/dy2, ..., dL/dy_M),
+ * where L is a scalar-value function of (x1, x2, ..., xN) (e.g., the
+ * loss function). dL/dx_i is the partial derivative of L with respect
+ * to x_i.
+ * </pre>
+ *
+ * Protobuf type {@code tensorflow.GradientDef}
+ */
+public  final class GradientDef extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.GradientDef)
+    GradientDefOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use GradientDef.newBuilder() to construct.
+  private GradientDef(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private GradientDef() {
+    functionName_ = "";
+    gradientFunc_ = "";
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new GradientDef();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private GradientDef(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 10: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            functionName_ = s;
+            break;
+          }
+          case 18: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            gradientFunc_ = s;
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.framework.FunctionProtos.internal_static_tensorflow_GradientDef_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.framework.FunctionProtos.internal_static_tensorflow_GradientDef_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.framework.GradientDef.class, org.tensorflow.proto.framework.GradientDef.Builder.class);
+  }
+
+  public static final int FUNCTION_NAME_FIELD_NUMBER = 1;
+  private volatile java.lang.Object functionName_;
+  /**
+   * <pre>
+   * The function name.
+   * </pre>
+   *
+   * <code>string function_name = 1;</code>
+   */
+  public java.lang.String getFunctionName() {
+    java.lang.Object ref = functionName_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      functionName_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * The function name.
+   * </pre>
+   *
+   * <code>string function_name = 1;</code>
+   */
+  public com.google.protobuf.ByteString
+      getFunctionNameBytes() {
+    java.lang.Object ref = functionName_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      functionName_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int GRADIENT_FUNC_FIELD_NUMBER = 2;
+  private volatile java.lang.Object gradientFunc_;
+  /**
+   * <pre>
+   * The gradient function's name.
+   * </pre>
+   *
+   * <code>string gradient_func = 2;</code>
+   */
+  public java.lang.String getGradientFunc() {
+    java.lang.Object ref = gradientFunc_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      gradientFunc_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * The gradient function's name.
+   * </pre>
+   *
+   * <code>string gradient_func = 2;</code>
+   */
+  public com.google.protobuf.ByteString
+      getGradientFuncBytes() {
+    java.lang.Object ref = gradientFunc_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      gradientFunc_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    if (!getFunctionNameBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 1, functionName_);
+    }
+    if (!getGradientFuncBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 2, gradientFunc_);
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (!getFunctionNameBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, functionName_);
+    }
+    if (!getGradientFuncBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(2, gradientFunc_);
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.framework.GradientDef)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.framework.GradientDef other = (org.tensorflow.proto.framework.GradientDef) obj;
+
+    if (!getFunctionName()
+        .equals(other.getFunctionName())) return false;
+    if (!getGradientFunc()
+        .equals(other.getGradientFunc())) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    hash = (37 * hash) + FUNCTION_NAME_FIELD_NUMBER;
+    hash = (53 * hash) + getFunctionName().hashCode();
+    hash = (37 * hash) + GRADIENT_FUNC_FIELD_NUMBER;
+    hash = (53 * hash) + getGradientFunc().hashCode();
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.framework.GradientDef parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.GradientDef parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.GradientDef parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.GradientDef parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.GradientDef parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.GradientDef parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.GradientDef parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.GradientDef parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.GradientDef parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.GradientDef parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.GradientDef parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.GradientDef parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.framework.GradientDef prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * <pre>
+   * GradientDef defines the gradient function of a function defined in
+   * a function library.
+   * A gradient function g (specified by gradient_func) for a function f
+   * (specified by function_name) must follow the following:
+   * The function 'f' must be a numerical function which takes N inputs
+   * and produces M outputs. Its gradient function 'g', which is a
+   * function taking N + M inputs and produces N outputs.
+   * I.e. if we have
+   *    (y1, y2, ..., y_M) = f(x1, x2, ..., x_N),
+   * then, g is
+   *    (dL/dx1, dL/dx2, ..., dL/dx_N) = g(x1, x2, ..., x_N,
+   *                                      dL/dy1, dL/dy2, ..., dL/dy_M),
+   * where L is a scalar-value function of (x1, x2, ..., xN) (e.g., the
+   * loss function). dL/dx_i is the partial derivative of L with respect
+   * to x_i.
+   * </pre>
+   *
+   * Protobuf type {@code tensorflow.GradientDef}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.GradientDef)
+      org.tensorflow.proto.framework.GradientDefOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.FunctionProtos.internal_static_tensorflow_GradientDef_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.FunctionProtos.internal_static_tensorflow_GradientDef_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.GradientDef.class, org.tensorflow.proto.framework.GradientDef.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.framework.GradientDef.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      functionName_ = "";
+
+      gradientFunc_ = "";
+
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.framework.FunctionProtos.internal_static_tensorflow_GradientDef_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.GradientDef getDefaultInstanceForType() {
+      return org.tensorflow.proto.framework.GradientDef.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.GradientDef build() {
+      org.tensorflow.proto.framework.GradientDef result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.GradientDef buildPartial() {
+      org.tensorflow.proto.framework.GradientDef result = new org.tensorflow.proto.framework.GradientDef(this);
+      result.functionName_ = functionName_;
+      result.gradientFunc_ = gradientFunc_;
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.framework.GradientDef) {
+        return mergeFrom((org.tensorflow.proto.framework.GradientDef)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.framework.GradientDef other) {
+      if (other == org.tensorflow.proto.framework.GradientDef.getDefaultInstance()) return this;
+      if (!other.getFunctionName().isEmpty()) {
+        functionName_ = other.functionName_;
+        onChanged();
+      }
+      if (!other.getGradientFunc().isEmpty()) {
+        gradientFunc_ = other.gradientFunc_;
+        onChanged();
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.framework.GradientDef parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.framework.GradientDef) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+
+    private java.lang.Object functionName_ = "";
+    /**
+     * <pre>
+     * The function name.
+     * </pre>
+     *
+     * <code>string function_name = 1;</code>
+     */
+    public java.lang.String getFunctionName() {
+      java.lang.Object ref = functionName_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        functionName_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * The function name.
+     * </pre>
+     *
+     * <code>string function_name = 1;</code>
+     */
+    public com.google.protobuf.ByteString
+        getFunctionNameBytes() {
+      java.lang.Object ref = functionName_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        functionName_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * The function name.
+     * </pre>
+     *
+     * <code>string function_name = 1;</code>
+     */
+    public Builder setFunctionName(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      functionName_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * The function name.
+     * </pre>
+     *
+     * <code>string function_name = 1;</code>
+     */
+    public Builder clearFunctionName() {
+      
+      functionName_ = getDefaultInstance().getFunctionName();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * The function name.
+     * </pre>
+     *
+     * <code>string function_name = 1;</code>
+     */
+    public Builder setFunctionNameBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      functionName_ = value;
+      onChanged();
+      return this;
+    }
+
+    private java.lang.Object gradientFunc_ = "";
+    /**
+     * <pre>
+     * The gradient function's name.
+     * </pre>
+     *
+     * <code>string gradient_func = 2;</code>
+     */
+    public java.lang.String getGradientFunc() {
+      java.lang.Object ref = gradientFunc_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        gradientFunc_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * The gradient function's name.
+     * </pre>
+     *
+     * <code>string gradient_func = 2;</code>
+     */
+    public com.google.protobuf.ByteString
+        getGradientFuncBytes() {
+      java.lang.Object ref = gradientFunc_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        gradientFunc_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * The gradient function's name.
+     * </pre>
+     *
+     * <code>string gradient_func = 2;</code>
+     */
+    public Builder setGradientFunc(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      gradientFunc_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * The gradient function's name.
+     * </pre>
+     *
+     * <code>string gradient_func = 2;</code>
+     */
+    public Builder clearGradientFunc() {
+      
+      gradientFunc_ = getDefaultInstance().getGradientFunc();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * The gradient function's name.
+     * </pre>
+     *
+     * <code>string gradient_func = 2;</code>
+     */
+    public Builder setGradientFuncBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      gradientFunc_ = value;
+      onChanged();
+      return this;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.GradientDef)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.GradientDef)
+  private static final org.tensorflow.proto.framework.GradientDef DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.GradientDef();
+  }
+
+  public static org.tensorflow.proto.framework.GradientDef getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<GradientDef>
+      PARSER = new com.google.protobuf.AbstractParser<GradientDef>() {
+    @java.lang.Override
+    public GradientDef parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new GradientDef(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<GradientDef> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<GradientDef> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.framework.GradientDef getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GradientDefOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GradientDefOrBuilder.java
new file mode 100644
index 00000000000..9437941de35
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GradientDefOrBuilder.java
@@ -0,0 +1,45 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/framework/function.proto
+
+package org.tensorflow.proto.framework;
+
+public interface GradientDefOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.GradientDef)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <pre>
+   * The function name.
+   * </pre>
+   *
+   * <code>string function_name = 1;</code>
+   */
+  java.lang.String getFunctionName();
+  /**
+   * <pre>
+   * The function name.
+   * </pre>
+   *
+   * <code>string function_name = 1;</code>
+   */
+  com.google.protobuf.ByteString
+      getFunctionNameBytes();
+
+  /**
+   * <pre>
+   * The gradient function's name.
+   * </pre>
+   *
+   * <code>string gradient_func = 2;</code>
+   */
+  java.lang.String getGradientFunc();
+  /**
+   * <pre>
+   * The gradient function's name.
+   * </pre>
+   *
+   * <code>string gradient_func = 2;</code>
+   */
+  com.google.protobuf.ByteString
+      getGradientFuncBytes();
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphDebugInfo.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphDebugInfo.java
new file mode 100644
index 00000000000..165c76f3ce7
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphDebugInfo.java
@@ -0,0 +1,3004 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/graph_debug_info.proto
+
+package org.tensorflow.proto.framework;
+
+/**
+ * Protobuf type {@code tensorflow.GraphDebugInfo}
+ */
+public  final class GraphDebugInfo extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.GraphDebugInfo)
+    GraphDebugInfoOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use GraphDebugInfo.newBuilder() to construct.
+  private GraphDebugInfo(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private GraphDebugInfo() {
+    files_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new GraphDebugInfo();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private GraphDebugInfo(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    int mutable_bitField0_ = 0;
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 10: {
+            java.lang.String s = input.readStringRequireUtf8();
+            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
+              files_ = new com.google.protobuf.LazyStringArrayList();
+              mutable_bitField0_ |= 0x00000001;
+            }
+            files_.add(s);
+            break;
+          }
+          case 18: {
+            if (!((mutable_bitField0_ & 0x00000002) != 0)) {
+              traces_ = com.google.protobuf.MapField.newMapField(
+                  TracesDefaultEntryHolder.defaultEntry);
+              mutable_bitField0_ |= 0x00000002;
+            }
+            com.google.protobuf.MapEntry<java.lang.String, org.tensorflow.proto.framework.GraphDebugInfo.StackTrace>
+            traces__ = input.readMessage(
+                TracesDefaultEntryHolder.defaultEntry.getParserForType(), extensionRegistry);
+            traces_.getMutableMap().put(
+                traces__.getKey(), traces__.getValue());
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      if (((mutable_bitField0_ & 0x00000001) != 0)) {
+        files_ = files_.getUnmodifiableView();
+      }
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.framework.GraphDebugInfoProtos.internal_static_tensorflow_GraphDebugInfo_descriptor;
+  }
+
+  @SuppressWarnings({"rawtypes"})
+  @java.lang.Override
+  protected com.google.protobuf.MapField internalGetMapField(
+      int number) {
+    switch (number) {
+      case 2:
+        return internalGetTraces();
+      default:
+        throw new RuntimeException(
+            "Invalid map field number: " + number);
+    }
+  }
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.framework.GraphDebugInfoProtos.internal_static_tensorflow_GraphDebugInfo_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.framework.GraphDebugInfo.class, org.tensorflow.proto.framework.GraphDebugInfo.Builder.class);
+  }
+
+  public interface FileLineColOrBuilder extends
+      // @@protoc_insertion_point(interface_extends:tensorflow.GraphDebugInfo.FileLineCol)
+      com.google.protobuf.MessageOrBuilder {
+
+    /**
+     * <pre>
+     * File name index, which can be used to retrieve the file name string from
+     * `files`. The value should be between 0 and (len(files)-1)
+     * </pre>
+     *
+     * <code>int32 file_index = 1;</code>
+     */
+    int getFileIndex();
+
+    /**
+     * <pre>
+     * Line number in the file.
+     * </pre>
+     *
+     * <code>int32 line = 2;</code>
+     */
+    int getLine();
+
+    /**
+     * <pre>
+     * Col number in the file line.
+     * </pre>
+     *
+     * <code>int32 col = 3;</code>
+     */
+    int getCol();
+
+    /**
+     * <pre>
+     * Name of function contains the file line.
+     * </pre>
+     *
+     * <code>string func = 4;</code>
+     */
+    java.lang.String getFunc();
+    /**
+     * <pre>
+     * Name of function contains the file line.
+     * </pre>
+     *
+     * <code>string func = 4;</code>
+     */
+    com.google.protobuf.ByteString
+        getFuncBytes();
+
+    /**
+     * <pre>
+     * Source code contained in this file line.
+     * </pre>
+     *
+     * <code>string code = 5;</code>
+     */
+    java.lang.String getCode();
+    /**
+     * <pre>
+     * Source code contained in this file line.
+     * </pre>
+     *
+     * <code>string code = 5;</code>
+     */
+    com.google.protobuf.ByteString
+        getCodeBytes();
+  }
+  /**
+   * <pre>
+   * This represents a file/line location in the source code.
+   * </pre>
+   *
+   * Protobuf type {@code tensorflow.GraphDebugInfo.FileLineCol}
+   */
+  public  static final class FileLineCol extends
+      com.google.protobuf.GeneratedMessageV3 implements
+      // @@protoc_insertion_point(message_implements:tensorflow.GraphDebugInfo.FileLineCol)
+      FileLineColOrBuilder {
+  private static final long serialVersionUID = 0L;
+    // Use FileLineCol.newBuilder() to construct.
+    private FileLineCol(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+      super(builder);
+    }
+    private FileLineCol() {
+      func_ = "";
+      code_ = "";
+    }
+
+    @java.lang.Override
+    @SuppressWarnings({"unused"})
+    protected java.lang.Object newInstance(
+        UnusedPrivateParameter unused) {
+      return new FileLineCol();
+    }
+
+    @java.lang.Override
+    public final com.google.protobuf.UnknownFieldSet
+    getUnknownFields() {
+      return this.unknownFields;
+    }
+    private FileLineCol(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      this();
+      if (extensionRegistry == null) {
+        throw new java.lang.NullPointerException();
+      }
+      com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+          com.google.protobuf.UnknownFieldSet.newBuilder();
+      try {
+        boolean done = false;
+        while (!done) {
+          int tag = input.readTag();
+          switch (tag) {
+            case 0:
+              done = true;
+              break;
+            case 8: {
+
+              fileIndex_ = input.readInt32();
+              break;
+            }
+            case 16: {
+
+              line_ = input.readInt32();
+              break;
+            }
+            case 24: {
+
+              col_ = input.readInt32();
+              break;
+            }
+            case 34: {
+              java.lang.String s = input.readStringRequireUtf8();
+
+              func_ = s;
+              break;
+            }
+            case 42: {
+              java.lang.String s = input.readStringRequireUtf8();
+
+              code_ = s;
+              break;
+            }
+            default: {
+              if (!parseUnknownField(
+                  input, unknownFields, extensionRegistry, tag)) {
+                done = true;
+              }
+              break;
+            }
+          }
+        }
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        throw e.setUnfinishedMessage(this);
+      } catch (java.io.IOException e) {
+        throw new com.google.protobuf.InvalidProtocolBufferException(
+            e).setUnfinishedMessage(this);
+      } finally {
+        this.unknownFields = unknownFields.build();
+        makeExtensionsImmutable();
+      }
+    }
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.GraphDebugInfoProtos.internal_static_tensorflow_GraphDebugInfo_FileLineCol_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.GraphDebugInfoProtos.internal_static_tensorflow_GraphDebugInfo_FileLineCol_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol.class, org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol.Builder.class);
+    }
+
+    public static final int FILE_INDEX_FIELD_NUMBER = 1;
+    private int fileIndex_;
+    /**
+     * <pre>
+     * File name index, which can be used to retrieve the file name string from
+     * `files`. The value should be between 0 and (len(files)-1)
+     * </pre>
+     *
+     * <code>int32 file_index = 1;</code>
+     */
+    public int getFileIndex() {
+      return fileIndex_;
+    }
+
+    public static final int LINE_FIELD_NUMBER = 2;
+    private int line_;
+    /**
+     * <pre>
+     * Line number in the file.
+     * </pre>
+     *
+     * <code>int32 line = 2;</code>
+     */
+    public int getLine() {
+      return line_;
+    }
+
+    public static final int COL_FIELD_NUMBER = 3;
+    private int col_;
+    /**
+     * <pre>
+     * Col number in the file line.
+     * </pre>
+     *
+     * <code>int32 col = 3;</code>
+     */
+    public int getCol() {
+      return col_;
+    }
+
+    public static final int FUNC_FIELD_NUMBER = 4;
+    private volatile java.lang.Object func_;
+    /**
+     * <pre>
+     * Name of function contains the file line.
+     * </pre>
+     *
+     * <code>string func = 4;</code>
+     */
+    public java.lang.String getFunc() {
+      java.lang.Object ref = func_;
+      if (ref instanceof java.lang.String) {
+        return (java.lang.String) ref;
+      } else {
+        com.google.protobuf.ByteString bs = 
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        func_ = s;
+        return s;
+      }
+    }
+    /**
+     * <pre>
+     * Name of function contains the file line.
+     * </pre>
+     *
+     * <code>string func = 4;</code>
+     */
+    public com.google.protobuf.ByteString
+        getFuncBytes() {
+      java.lang.Object ref = func_;
+      if (ref instanceof java.lang.String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        func_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+
+    public static final int CODE_FIELD_NUMBER = 5;
+    private volatile java.lang.Object code_;
+    /**
+     * <pre>
+     * Source code contained in this file line.
+     * </pre>
+     *
+     * <code>string code = 5;</code>
+     */
+    public java.lang.String getCode() {
+      java.lang.Object ref = code_;
+      if (ref instanceof java.lang.String) {
+        return (java.lang.String) ref;
+      } else {
+        com.google.protobuf.ByteString bs = 
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        code_ = s;
+        return s;
+      }
+    }
+    /**
+     * <pre>
+     * Source code contained in this file line.
+     * </pre>
+     *
+     * <code>string code = 5;</code>
+     */
+    public com.google.protobuf.ByteString
+        getCodeBytes() {
+      java.lang.Object ref = code_;
+      if (ref instanceof java.lang.String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        code_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+
+    private byte memoizedIsInitialized = -1;
+    @java.lang.Override
+    public final boolean isInitialized() {
+      byte isInitialized = memoizedIsInitialized;
+      if (isInitialized == 1) return true;
+      if (isInitialized == 0) return false;
+
+      memoizedIsInitialized = 1;
+      return true;
+    }
+
+    @java.lang.Override
+    public void writeTo(com.google.protobuf.CodedOutputStream output)
+                        throws java.io.IOException {
+      if (fileIndex_ != 0) {
+        output.writeInt32(1, fileIndex_);
+      }
+      if (line_ != 0) {
+        output.writeInt32(2, line_);
+      }
+      if (col_ != 0) {
+        output.writeInt32(3, col_);
+      }
+      if (!getFuncBytes().isEmpty()) {
+        com.google.protobuf.GeneratedMessageV3.writeString(output, 4, func_);
+      }
+      if (!getCodeBytes().isEmpty()) {
+        com.google.protobuf.GeneratedMessageV3.writeString(output, 5, code_);
+      }
+      unknownFields.writeTo(output);
+    }
+
+    @java.lang.Override
+    public int getSerializedSize() {
+      int size = memoizedSize;
+      if (size != -1) return size;
+
+      size = 0;
+      if (fileIndex_ != 0) {
+        size += com.google.protobuf.CodedOutputStream
+          .computeInt32Size(1, fileIndex_);
+      }
+      if (line_ != 0) {
+        size += com.google.protobuf.CodedOutputStream
+          .computeInt32Size(2, line_);
+      }
+      if (col_ != 0) {
+        size += com.google.protobuf.CodedOutputStream
+          .computeInt32Size(3, col_);
+      }
+      if (!getFuncBytes().isEmpty()) {
+        size += com.google.protobuf.GeneratedMessageV3.computeStringSize(4, func_);
+      }
+      if (!getCodeBytes().isEmpty()) {
+        size += com.google.protobuf.GeneratedMessageV3.computeStringSize(5, code_);
+      }
+      size += unknownFields.getSerializedSize();
+      memoizedSize = size;
+      return size;
+    }
+
+    @java.lang.Override
+    public boolean equals(final java.lang.Object obj) {
+      if (obj == this) {
+       return true;
+      }
+      if (!(obj instanceof org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol)) {
+        return super.equals(obj);
+      }
+      org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol other = (org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol) obj;
+
+      if (getFileIndex()
+          != other.getFileIndex()) return false;
+      if (getLine()
+          != other.getLine()) return false;
+      if (getCol()
+          != other.getCol()) return false;
+      if (!getFunc()
+          .equals(other.getFunc())) return false;
+      if (!getCode()
+          .equals(other.getCode())) return false;
+      if (!unknownFields.equals(other.unknownFields)) return false;
+      return true;
+    }
+
+    @java.lang.Override
+    public int hashCode() {
+      if (memoizedHashCode != 0) {
+        return memoizedHashCode;
+      }
+      int hash = 41;
+      hash = (19 * hash) + getDescriptor().hashCode();
+      hash = (37 * hash) + FILE_INDEX_FIELD_NUMBER;
+      hash = (53 * hash) + getFileIndex();
+      hash = (37 * hash) + LINE_FIELD_NUMBER;
+      hash = (53 * hash) + getLine();
+      hash = (37 * hash) + COL_FIELD_NUMBER;
+      hash = (53 * hash) + getCol();
+      hash = (37 * hash) + FUNC_FIELD_NUMBER;
+      hash = (53 * hash) + getFunc().hashCode();
+      hash = (37 * hash) + CODE_FIELD_NUMBER;
+      hash = (53 * hash) + getCode().hashCode();
+      hash = (29 * hash) + unknownFields.hashCode();
+      memoizedHashCode = hash;
+      return hash;
+    }
+
+    public static org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol parseFrom(
+        java.nio.ByteBuffer data)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data);
+    }
+    public static org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol parseFrom(
+        java.nio.ByteBuffer data,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol parseFrom(
+        com.google.protobuf.ByteString data)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data);
+    }
+    public static org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol parseFrom(
+        com.google.protobuf.ByteString data,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol parseFrom(byte[] data)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data);
+    }
+    public static org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol parseFrom(
+        byte[] data,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol parseFrom(java.io.InputStream input)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input);
+    }
+    public static org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol parseFrom(
+        java.io.InputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol parseDelimitedFrom(java.io.InputStream input)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseDelimitedWithIOException(PARSER, input);
+    }
+    public static org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol parseDelimitedFrom(
+        java.io.InputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol parseFrom(
+        com.google.protobuf.CodedInputStream input)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input);
+    }
+    public static org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol parseFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input, extensionRegistry);
+    }
+
+    @java.lang.Override
+    public Builder newBuilderForType() { return newBuilder(); }
+    public static Builder newBuilder() {
+      return DEFAULT_INSTANCE.toBuilder();
+    }
+    public static Builder newBuilder(org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol prototype) {
+      return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+    }
+    @java.lang.Override
+    public Builder toBuilder() {
+      return this == DEFAULT_INSTANCE
+          ? new Builder() : new Builder().mergeFrom(this);
+    }
+
+    @java.lang.Override
+    protected Builder newBuilderForType(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      Builder builder = new Builder(parent);
+      return builder;
+    }
+    /**
+     * <pre>
+     * This represents a file/line location in the source code.
+     * </pre>
+     *
+     * Protobuf type {@code tensorflow.GraphDebugInfo.FileLineCol}
+     */
+    public static final class Builder extends
+        com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+        // @@protoc_insertion_point(builder_implements:tensorflow.GraphDebugInfo.FileLineCol)
+        org.tensorflow.proto.framework.GraphDebugInfo.FileLineColOrBuilder {
+      public static final com.google.protobuf.Descriptors.Descriptor
+          getDescriptor() {
+        return org.tensorflow.proto.framework.GraphDebugInfoProtos.internal_static_tensorflow_GraphDebugInfo_FileLineCol_descriptor;
+      }
+
+      @java.lang.Override
+      protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+          internalGetFieldAccessorTable() {
+        return org.tensorflow.proto.framework.GraphDebugInfoProtos.internal_static_tensorflow_GraphDebugInfo_FileLineCol_fieldAccessorTable
+            .ensureFieldAccessorsInitialized(
+                org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol.class, org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol.Builder.class);
+      }
+
+      // Construct using org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol.newBuilder()
+      private Builder() {
+        maybeForceBuilderInitialization();
+      }
+
+      private Builder(
+          com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+        super(parent);
+        maybeForceBuilderInitialization();
+      }
+      private void maybeForceBuilderInitialization() {
+        if (com.google.protobuf.GeneratedMessageV3
+                .alwaysUseFieldBuilders) {
+        }
+      }
+      @java.lang.Override
+      public Builder clear() {
+        super.clear();
+        fileIndex_ = 0;
+
+        line_ = 0;
+
+        col_ = 0;
+
+        func_ = "";
+
+        code_ = "";
+
+        return this;
+      }
+
+      @java.lang.Override
+      public com.google.protobuf.Descriptors.Descriptor
+          getDescriptorForType() {
+        return org.tensorflow.proto.framework.GraphDebugInfoProtos.internal_static_tensorflow_GraphDebugInfo_FileLineCol_descriptor;
+      }
+
+      @java.lang.Override
+      public org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol getDefaultInstanceForType() {
+        return org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol.getDefaultInstance();
+      }
+
+      @java.lang.Override
+      public org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol build() {
+        org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol result = buildPartial();
+        if (!result.isInitialized()) {
+          throw newUninitializedMessageException(result);
+        }
+        return result;
+      }
+
+      @java.lang.Override
+      public org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol buildPartial() {
+        org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol result = new org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol(this);
+        result.fileIndex_ = fileIndex_;
+        result.line_ = line_;
+        result.col_ = col_;
+        result.func_ = func_;
+        result.code_ = code_;
+        onBuilt();
+        return result;
+      }
+
+      @java.lang.Override
+      public Builder clone() {
+        return super.clone();
+      }
+      @java.lang.Override
+      public Builder setField(
+          com.google.protobuf.Descriptors.FieldDescriptor field,
+          java.lang.Object value) {
+        return super.setField(field, value);
+      }
+      @java.lang.Override
+      public Builder clearField(
+          com.google.protobuf.Descriptors.FieldDescriptor field) {
+        return super.clearField(field);
+      }
+      @java.lang.Override
+      public Builder clearOneof(
+          com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+        return super.clearOneof(oneof);
+      }
+      @java.lang.Override
+      public Builder setRepeatedField(
+          com.google.protobuf.Descriptors.FieldDescriptor field,
+          int index, java.lang.Object value) {
+        return super.setRepeatedField(field, index, value);
+      }
+      @java.lang.Override
+      public Builder addRepeatedField(
+          com.google.protobuf.Descriptors.FieldDescriptor field,
+          java.lang.Object value) {
+        return super.addRepeatedField(field, value);
+      }
+      @java.lang.Override
+      public Builder mergeFrom(com.google.protobuf.Message other) {
+        if (other instanceof org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol) {
+          return mergeFrom((org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol)other);
+        } else {
+          super.mergeFrom(other);
+          return this;
+        }
+      }
+
+      public Builder mergeFrom(org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol other) {
+        if (other == org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol.getDefaultInstance()) return this;
+        if (other.getFileIndex() != 0) {
+          setFileIndex(other.getFileIndex());
+        }
+        if (other.getLine() != 0) {
+          setLine(other.getLine());
+        }
+        if (other.getCol() != 0) {
+          setCol(other.getCol());
+        }
+        if (!other.getFunc().isEmpty()) {
+          func_ = other.func_;
+          onChanged();
+        }
+        if (!other.getCode().isEmpty()) {
+          code_ = other.code_;
+          onChanged();
+        }
+        this.mergeUnknownFields(other.unknownFields);
+        onChanged();
+        return this;
+      }
+
+      @java.lang.Override
+      public final boolean isInitialized() {
+        return true;
+      }
+
+      @java.lang.Override
+      public Builder mergeFrom(
+          com.google.protobuf.CodedInputStream input,
+          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+          throws java.io.IOException {
+        org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol parsedMessage = null;
+        try {
+          parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+        } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+          parsedMessage = (org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol) e.getUnfinishedMessage();
+          throw e.unwrapIOException();
+        } finally {
+          if (parsedMessage != null) {
+            mergeFrom(parsedMessage);
+          }
+        }
+        return this;
+      }
+
+      private int fileIndex_ ;
+      /**
+       * <pre>
+       * File name index, which can be used to retrieve the file name string from
+       * `files`. The value should be between 0 and (len(files)-1)
+       * </pre>
+       *
+       * <code>int32 file_index = 1;</code>
+       */
+      public int getFileIndex() {
+        return fileIndex_;
+      }
+      /**
+       * <pre>
+       * File name index, which can be used to retrieve the file name string from
+       * `files`. The value should be between 0 and (len(files)-1)
+       * </pre>
+       *
+       * <code>int32 file_index = 1;</code>
+       */
+      public Builder setFileIndex(int value) {
+        
+        fileIndex_ = value;
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * File name index, which can be used to retrieve the file name string from
+       * `files`. The value should be between 0 and (len(files)-1)
+       * </pre>
+       *
+       * <code>int32 file_index = 1;</code>
+       */
+      public Builder clearFileIndex() {
+        
+        fileIndex_ = 0;
+        onChanged();
+        return this;
+      }
+
+      private int line_ ;
+      /**
+       * <pre>
+       * Line number in the file.
+       * </pre>
+       *
+       * <code>int32 line = 2;</code>
+       */
+      public int getLine() {
+        return line_;
+      }
+      /**
+       * <pre>
+       * Line number in the file.
+       * </pre>
+       *
+       * <code>int32 line = 2;</code>
+       */
+      public Builder setLine(int value) {
+        
+        line_ = value;
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * Line number in the file.
+       * </pre>
+       *
+       * <code>int32 line = 2;</code>
+       */
+      public Builder clearLine() {
+        
+        line_ = 0;
+        onChanged();
+        return this;
+      }
+
+      private int col_ ;
+      /**
+       * <pre>
+       * Col number in the file line.
+       * </pre>
+       *
+       * <code>int32 col = 3;</code>
+       */
+      public int getCol() {
+        return col_;
+      }
+      /**
+       * <pre>
+       * Col number in the file line.
+       * </pre>
+       *
+       * <code>int32 col = 3;</code>
+       */
+      public Builder setCol(int value) {
+        
+        col_ = value;
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * Col number in the file line.
+       * </pre>
+       *
+       * <code>int32 col = 3;</code>
+       */
+      public Builder clearCol() {
+        
+        col_ = 0;
+        onChanged();
+        return this;
+      }
+
+      private java.lang.Object func_ = "";
+      /**
+       * <pre>
+       * Name of function contains the file line.
+       * </pre>
+       *
+       * <code>string func = 4;</code>
+       */
+      public java.lang.String getFunc() {
+        java.lang.Object ref = func_;
+        if (!(ref instanceof java.lang.String)) {
+          com.google.protobuf.ByteString bs =
+              (com.google.protobuf.ByteString) ref;
+          java.lang.String s = bs.toStringUtf8();
+          func_ = s;
+          return s;
+        } else {
+          return (java.lang.String) ref;
+        }
+      }
+      /**
+       * <pre>
+       * Name of function contains the file line.
+       * </pre>
+       *
+       * <code>string func = 4;</code>
+       */
+      public com.google.protobuf.ByteString
+          getFuncBytes() {
+        java.lang.Object ref = func_;
+        if (ref instanceof String) {
+          com.google.protobuf.ByteString b = 
+              com.google.protobuf.ByteString.copyFromUtf8(
+                  (java.lang.String) ref);
+          func_ = b;
+          return b;
+        } else {
+          return (com.google.protobuf.ByteString) ref;
+        }
+      }
+      /**
+       * <pre>
+       * Name of function contains the file line.
+       * </pre>
+       *
+       * <code>string func = 4;</code>
+       */
+      public Builder setFunc(
+          java.lang.String value) {
+        if (value == null) {
+    throw new NullPointerException();
+  }
+  
+        func_ = value;
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * Name of function contains the file line.
+       * </pre>
+       *
+       * <code>string func = 4;</code>
+       */
+      public Builder clearFunc() {
+        
+        func_ = getDefaultInstance().getFunc();
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * Name of function contains the file line.
+       * </pre>
+       *
+       * <code>string func = 4;</code>
+       */
+      public Builder setFuncBytes(
+          com.google.protobuf.ByteString value) {
+        if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+        
+        func_ = value;
+        onChanged();
+        return this;
+      }
+
+      private java.lang.Object code_ = "";
+      /**
+       * <pre>
+       * Source code contained in this file line.
+       * </pre>
+       *
+       * <code>string code = 5;</code>
+       */
+      public java.lang.String getCode() {
+        java.lang.Object ref = code_;
+        if (!(ref instanceof java.lang.String)) {
+          com.google.protobuf.ByteString bs =
+              (com.google.protobuf.ByteString) ref;
+          java.lang.String s = bs.toStringUtf8();
+          code_ = s;
+          return s;
+        } else {
+          return (java.lang.String) ref;
+        }
+      }
+      /**
+       * <pre>
+       * Source code contained in this file line.
+       * </pre>
+       *
+       * <code>string code = 5;</code>
+       */
+      public com.google.protobuf.ByteString
+          getCodeBytes() {
+        java.lang.Object ref = code_;
+        if (ref instanceof String) {
+          com.google.protobuf.ByteString b = 
+              com.google.protobuf.ByteString.copyFromUtf8(
+                  (java.lang.String) ref);
+          code_ = b;
+          return b;
+        } else {
+          return (com.google.protobuf.ByteString) ref;
+        }
+      }
+      /**
+       * <pre>
+       * Source code contained in this file line.
+       * </pre>
+       *
+       * <code>string code = 5;</code>
+       */
+      public Builder setCode(
+          java.lang.String value) {
+        if (value == null) {
+    throw new NullPointerException();
+  }
+  
+        code_ = value;
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * Source code contained in this file line.
+       * </pre>
+       *
+       * <code>string code = 5;</code>
+       */
+      public Builder clearCode() {
+        
+        code_ = getDefaultInstance().getCode();
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * Source code contained in this file line.
+       * </pre>
+       *
+       * <code>string code = 5;</code>
+       */
+      public Builder setCodeBytes(
+          com.google.protobuf.ByteString value) {
+        if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+        
+        code_ = value;
+        onChanged();
+        return this;
+      }
+      @java.lang.Override
+      public final Builder setUnknownFields(
+          final com.google.protobuf.UnknownFieldSet unknownFields) {
+        return super.setUnknownFields(unknownFields);
+      }
+
+      @java.lang.Override
+      public final Builder mergeUnknownFields(
+          final com.google.protobuf.UnknownFieldSet unknownFields) {
+        return super.mergeUnknownFields(unknownFields);
+      }
+
+
+      // @@protoc_insertion_point(builder_scope:tensorflow.GraphDebugInfo.FileLineCol)
+    }
+
+    // @@protoc_insertion_point(class_scope:tensorflow.GraphDebugInfo.FileLineCol)
+    private static final org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol DEFAULT_INSTANCE;
+    static {
+      DEFAULT_INSTANCE = new org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol();
+    }
+
+    public static org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol getDefaultInstance() {
+      return DEFAULT_INSTANCE;
+    }
+
+    private static final com.google.protobuf.Parser<FileLineCol>
+        PARSER = new com.google.protobuf.AbstractParser<FileLineCol>() {
+      @java.lang.Override
+      public FileLineCol parsePartialFrom(
+          com.google.protobuf.CodedInputStream input,
+          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+          throws com.google.protobuf.InvalidProtocolBufferException {
+        return new FileLineCol(input, extensionRegistry);
+      }
+    };
+
+    public static com.google.protobuf.Parser<FileLineCol> parser() {
+      return PARSER;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Parser<FileLineCol> getParserForType() {
+      return PARSER;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol getDefaultInstanceForType() {
+      return DEFAULT_INSTANCE;
+    }
+
+  }
+
+  public interface StackTraceOrBuilder extends
+      // @@protoc_insertion_point(interface_extends:tensorflow.GraphDebugInfo.StackTrace)
+      com.google.protobuf.MessageOrBuilder {
+
+    /**
+     * <pre>
+     * Each line in the stack trace.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.GraphDebugInfo.FileLineCol file_line_cols = 1;</code>
+     */
+    java.util.List<org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol> 
+        getFileLineColsList();
+    /**
+     * <pre>
+     * Each line in the stack trace.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.GraphDebugInfo.FileLineCol file_line_cols = 1;</code>
+     */
+    org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol getFileLineCols(int index);
+    /**
+     * <pre>
+     * Each line in the stack trace.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.GraphDebugInfo.FileLineCol file_line_cols = 1;</code>
+     */
+    int getFileLineColsCount();
+    /**
+     * <pre>
+     * Each line in the stack trace.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.GraphDebugInfo.FileLineCol file_line_cols = 1;</code>
+     */
+    java.util.List<? extends org.tensorflow.proto.framework.GraphDebugInfo.FileLineColOrBuilder> 
+        getFileLineColsOrBuilderList();
+    /**
+     * <pre>
+     * Each line in the stack trace.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.GraphDebugInfo.FileLineCol file_line_cols = 1;</code>
+     */
+    org.tensorflow.proto.framework.GraphDebugInfo.FileLineColOrBuilder getFileLineColsOrBuilder(
+        int index);
+  }
+  /**
+   * <pre>
+   * This represents a stack trace which is a ordered list of `FileLineCol`.
+   * </pre>
+   *
+   * Protobuf type {@code tensorflow.GraphDebugInfo.StackTrace}
+   */
+  public  static final class StackTrace extends
+      com.google.protobuf.GeneratedMessageV3 implements
+      // @@protoc_insertion_point(message_implements:tensorflow.GraphDebugInfo.StackTrace)
+      StackTraceOrBuilder {
+  private static final long serialVersionUID = 0L;
+    // Use StackTrace.newBuilder() to construct.
+    private StackTrace(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+      super(builder);
+    }
+    private StackTrace() {
+      fileLineCols_ = java.util.Collections.emptyList();
+    }
+
+    @java.lang.Override
+    @SuppressWarnings({"unused"})
+    protected java.lang.Object newInstance(
+        UnusedPrivateParameter unused) {
+      return new StackTrace();
+    }
+
+    @java.lang.Override
+    public final com.google.protobuf.UnknownFieldSet
+    getUnknownFields() {
+      return this.unknownFields;
+    }
+    private StackTrace(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      this();
+      if (extensionRegistry == null) {
+        throw new java.lang.NullPointerException();
+      }
+      int mutable_bitField0_ = 0;
+      com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+          com.google.protobuf.UnknownFieldSet.newBuilder();
+      try {
+        boolean done = false;
+        while (!done) {
+          int tag = input.readTag();
+          switch (tag) {
+            case 0:
+              done = true;
+              break;
+            case 10: {
+              if (!((mutable_bitField0_ & 0x00000001) != 0)) {
+                fileLineCols_ = new java.util.ArrayList<org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol>();
+                mutable_bitField0_ |= 0x00000001;
+              }
+              fileLineCols_.add(
+                  input.readMessage(org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol.parser(), extensionRegistry));
+              break;
+            }
+            default: {
+              if (!parseUnknownField(
+                  input, unknownFields, extensionRegistry, tag)) {
+                done = true;
+              }
+              break;
+            }
+          }
+        }
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        throw e.setUnfinishedMessage(this);
+      } catch (java.io.IOException e) {
+        throw new com.google.protobuf.InvalidProtocolBufferException(
+            e).setUnfinishedMessage(this);
+      } finally {
+        if (((mutable_bitField0_ & 0x00000001) != 0)) {
+          fileLineCols_ = java.util.Collections.unmodifiableList(fileLineCols_);
+        }
+        this.unknownFields = unknownFields.build();
+        makeExtensionsImmutable();
+      }
+    }
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.GraphDebugInfoProtos.internal_static_tensorflow_GraphDebugInfo_StackTrace_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.GraphDebugInfoProtos.internal_static_tensorflow_GraphDebugInfo_StackTrace_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.GraphDebugInfo.StackTrace.class, org.tensorflow.proto.framework.GraphDebugInfo.StackTrace.Builder.class);
+    }
+
+    public static final int FILE_LINE_COLS_FIELD_NUMBER = 1;
+    private java.util.List<org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol> fileLineCols_;
+    /**
+     * <pre>
+     * Each line in the stack trace.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.GraphDebugInfo.FileLineCol file_line_cols = 1;</code>
+     */
+    public java.util.List<org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol> getFileLineColsList() {
+      return fileLineCols_;
+    }
+    /**
+     * <pre>
+     * Each line in the stack trace.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.GraphDebugInfo.FileLineCol file_line_cols = 1;</code>
+     */
+    public java.util.List<? extends org.tensorflow.proto.framework.GraphDebugInfo.FileLineColOrBuilder> 
+        getFileLineColsOrBuilderList() {
+      return fileLineCols_;
+    }
+    /**
+     * <pre>
+     * Each line in the stack trace.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.GraphDebugInfo.FileLineCol file_line_cols = 1;</code>
+     */
+    public int getFileLineColsCount() {
+      return fileLineCols_.size();
+    }
+    /**
+     * <pre>
+     * Each line in the stack trace.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.GraphDebugInfo.FileLineCol file_line_cols = 1;</code>
+     */
+    public org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol getFileLineCols(int index) {
+      return fileLineCols_.get(index);
+    }
+    /**
+     * <pre>
+     * Each line in the stack trace.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.GraphDebugInfo.FileLineCol file_line_cols = 1;</code>
+     */
+    public org.tensorflow.proto.framework.GraphDebugInfo.FileLineColOrBuilder getFileLineColsOrBuilder(
+        int index) {
+      return fileLineCols_.get(index);
+    }
+
+    private byte memoizedIsInitialized = -1;
+    @java.lang.Override
+    public final boolean isInitialized() {
+      byte isInitialized = memoizedIsInitialized;
+      if (isInitialized == 1) return true;
+      if (isInitialized == 0) return false;
+
+      memoizedIsInitialized = 1;
+      return true;
+    }
+
+    @java.lang.Override
+    public void writeTo(com.google.protobuf.CodedOutputStream output)
+                        throws java.io.IOException {
+      for (int i = 0; i < fileLineCols_.size(); i++) {
+        output.writeMessage(1, fileLineCols_.get(i));
+      }
+      unknownFields.writeTo(output);
+    }
+
+    @java.lang.Override
+    public int getSerializedSize() {
+      int size = memoizedSize;
+      if (size != -1) return size;
+
+      size = 0;
+      for (int i = 0; i < fileLineCols_.size(); i++) {
+        size += com.google.protobuf.CodedOutputStream
+          .computeMessageSize(1, fileLineCols_.get(i));
+      }
+      size += unknownFields.getSerializedSize();
+      memoizedSize = size;
+      return size;
+    }
+
+    @java.lang.Override
+    public boolean equals(final java.lang.Object obj) {
+      if (obj == this) {
+       return true;
+      }
+      if (!(obj instanceof org.tensorflow.proto.framework.GraphDebugInfo.StackTrace)) {
+        return super.equals(obj);
+      }
+      org.tensorflow.proto.framework.GraphDebugInfo.StackTrace other = (org.tensorflow.proto.framework.GraphDebugInfo.StackTrace) obj;
+
+      if (!getFileLineColsList()
+          .equals(other.getFileLineColsList())) return false;
+      if (!unknownFields.equals(other.unknownFields)) return false;
+      return true;
+    }
+
+    @java.lang.Override
+    public int hashCode() {
+      if (memoizedHashCode != 0) {
+        return memoizedHashCode;
+      }
+      int hash = 41;
+      hash = (19 * hash) + getDescriptor().hashCode();
+      if (getFileLineColsCount() > 0) {
+        hash = (37 * hash) + FILE_LINE_COLS_FIELD_NUMBER;
+        hash = (53 * hash) + getFileLineColsList().hashCode();
+      }
+      hash = (29 * hash) + unknownFields.hashCode();
+      memoizedHashCode = hash;
+      return hash;
+    }
+
+    public static org.tensorflow.proto.framework.GraphDebugInfo.StackTrace parseFrom(
+        java.nio.ByteBuffer data)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data);
+    }
+    public static org.tensorflow.proto.framework.GraphDebugInfo.StackTrace parseFrom(
+        java.nio.ByteBuffer data,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.GraphDebugInfo.StackTrace parseFrom(
+        com.google.protobuf.ByteString data)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data);
+    }
+    public static org.tensorflow.proto.framework.GraphDebugInfo.StackTrace parseFrom(
+        com.google.protobuf.ByteString data,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.GraphDebugInfo.StackTrace parseFrom(byte[] data)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data);
+    }
+    public static org.tensorflow.proto.framework.GraphDebugInfo.StackTrace parseFrom(
+        byte[] data,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.GraphDebugInfo.StackTrace parseFrom(java.io.InputStream input)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input);
+    }
+    public static org.tensorflow.proto.framework.GraphDebugInfo.StackTrace parseFrom(
+        java.io.InputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.GraphDebugInfo.StackTrace parseDelimitedFrom(java.io.InputStream input)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseDelimitedWithIOException(PARSER, input);
+    }
+    public static org.tensorflow.proto.framework.GraphDebugInfo.StackTrace parseDelimitedFrom(
+        java.io.InputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.GraphDebugInfo.StackTrace parseFrom(
+        com.google.protobuf.CodedInputStream input)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input);
+    }
+    public static org.tensorflow.proto.framework.GraphDebugInfo.StackTrace parseFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input, extensionRegistry);
+    }
+
+    @java.lang.Override
+    public Builder newBuilderForType() { return newBuilder(); }
+    public static Builder newBuilder() {
+      return DEFAULT_INSTANCE.toBuilder();
+    }
+    public static Builder newBuilder(org.tensorflow.proto.framework.GraphDebugInfo.StackTrace prototype) {
+      return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+    }
+    @java.lang.Override
+    public Builder toBuilder() {
+      return this == DEFAULT_INSTANCE
+          ? new Builder() : new Builder().mergeFrom(this);
+    }
+
+    @java.lang.Override
+    protected Builder newBuilderForType(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      Builder builder = new Builder(parent);
+      return builder;
+    }
+    /**
+     * <pre>
+     * This represents a stack trace which is a ordered list of `FileLineCol`.
+     * </pre>
+     *
+     * Protobuf type {@code tensorflow.GraphDebugInfo.StackTrace}
+     */
+    public static final class Builder extends
+        com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+        // @@protoc_insertion_point(builder_implements:tensorflow.GraphDebugInfo.StackTrace)
+        org.tensorflow.proto.framework.GraphDebugInfo.StackTraceOrBuilder {
+      public static final com.google.protobuf.Descriptors.Descriptor
+          getDescriptor() {
+        return org.tensorflow.proto.framework.GraphDebugInfoProtos.internal_static_tensorflow_GraphDebugInfo_StackTrace_descriptor;
+      }
+
+      @java.lang.Override
+      protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+          internalGetFieldAccessorTable() {
+        return org.tensorflow.proto.framework.GraphDebugInfoProtos.internal_static_tensorflow_GraphDebugInfo_StackTrace_fieldAccessorTable
+            .ensureFieldAccessorsInitialized(
+                org.tensorflow.proto.framework.GraphDebugInfo.StackTrace.class, org.tensorflow.proto.framework.GraphDebugInfo.StackTrace.Builder.class);
+      }
+
+      // Construct using org.tensorflow.proto.framework.GraphDebugInfo.StackTrace.newBuilder()
+      private Builder() {
+        maybeForceBuilderInitialization();
+      }
+
+      private Builder(
+          com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+        super(parent);
+        maybeForceBuilderInitialization();
+      }
+      private void maybeForceBuilderInitialization() {
+        if (com.google.protobuf.GeneratedMessageV3
+                .alwaysUseFieldBuilders) {
+          getFileLineColsFieldBuilder();
+        }
+      }
+      @java.lang.Override
+      public Builder clear() {
+        super.clear();
+        if (fileLineColsBuilder_ == null) {
+          fileLineCols_ = java.util.Collections.emptyList();
+          bitField0_ = (bitField0_ & ~0x00000001);
+        } else {
+          fileLineColsBuilder_.clear();
+        }
+        return this;
+      }
+
+      @java.lang.Override
+      public com.google.protobuf.Descriptors.Descriptor
+          getDescriptorForType() {
+        return org.tensorflow.proto.framework.GraphDebugInfoProtos.internal_static_tensorflow_GraphDebugInfo_StackTrace_descriptor;
+      }
+
+      @java.lang.Override
+      public org.tensorflow.proto.framework.GraphDebugInfo.StackTrace getDefaultInstanceForType() {
+        return org.tensorflow.proto.framework.GraphDebugInfo.StackTrace.getDefaultInstance();
+      }
+
+      @java.lang.Override
+      public org.tensorflow.proto.framework.GraphDebugInfo.StackTrace build() {
+        org.tensorflow.proto.framework.GraphDebugInfo.StackTrace result = buildPartial();
+        if (!result.isInitialized()) {
+          throw newUninitializedMessageException(result);
+        }
+        return result;
+      }
+
+      @java.lang.Override
+      public org.tensorflow.proto.framework.GraphDebugInfo.StackTrace buildPartial() {
+        org.tensorflow.proto.framework.GraphDebugInfo.StackTrace result = new org.tensorflow.proto.framework.GraphDebugInfo.StackTrace(this);
+        int from_bitField0_ = bitField0_;
+        if (fileLineColsBuilder_ == null) {
+          if (((bitField0_ & 0x00000001) != 0)) {
+            fileLineCols_ = java.util.Collections.unmodifiableList(fileLineCols_);
+            bitField0_ = (bitField0_ & ~0x00000001);
+          }
+          result.fileLineCols_ = fileLineCols_;
+        } else {
+          result.fileLineCols_ = fileLineColsBuilder_.build();
+        }
+        onBuilt();
+        return result;
+      }
+
+      @java.lang.Override
+      public Builder clone() {
+        return super.clone();
+      }
+      @java.lang.Override
+      public Builder setField(
+          com.google.protobuf.Descriptors.FieldDescriptor field,
+          java.lang.Object value) {
+        return super.setField(field, value);
+      }
+      @java.lang.Override
+      public Builder clearField(
+          com.google.protobuf.Descriptors.FieldDescriptor field) {
+        return super.clearField(field);
+      }
+      @java.lang.Override
+      public Builder clearOneof(
+          com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+        return super.clearOneof(oneof);
+      }
+      @java.lang.Override
+      public Builder setRepeatedField(
+          com.google.protobuf.Descriptors.FieldDescriptor field,
+          int index, java.lang.Object value) {
+        return super.setRepeatedField(field, index, value);
+      }
+      @java.lang.Override
+      public Builder addRepeatedField(
+          com.google.protobuf.Descriptors.FieldDescriptor field,
+          java.lang.Object value) {
+        return super.addRepeatedField(field, value);
+      }
+      @java.lang.Override
+      public Builder mergeFrom(com.google.protobuf.Message other) {
+        if (other instanceof org.tensorflow.proto.framework.GraphDebugInfo.StackTrace) {
+          return mergeFrom((org.tensorflow.proto.framework.GraphDebugInfo.StackTrace)other);
+        } else {
+          super.mergeFrom(other);
+          return this;
+        }
+      }
+
+      public Builder mergeFrom(org.tensorflow.proto.framework.GraphDebugInfo.StackTrace other) {
+        if (other == org.tensorflow.proto.framework.GraphDebugInfo.StackTrace.getDefaultInstance()) return this;
+        if (fileLineColsBuilder_ == null) {
+          if (!other.fileLineCols_.isEmpty()) {
+            if (fileLineCols_.isEmpty()) {
+              fileLineCols_ = other.fileLineCols_;
+              bitField0_ = (bitField0_ & ~0x00000001);
+            } else {
+              ensureFileLineColsIsMutable();
+              fileLineCols_.addAll(other.fileLineCols_);
+            }
+            onChanged();
+          }
+        } else {
+          if (!other.fileLineCols_.isEmpty()) {
+            if (fileLineColsBuilder_.isEmpty()) {
+              fileLineColsBuilder_.dispose();
+              fileLineColsBuilder_ = null;
+              fileLineCols_ = other.fileLineCols_;
+              bitField0_ = (bitField0_ & ~0x00000001);
+              fileLineColsBuilder_ = 
+                com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
+                   getFileLineColsFieldBuilder() : null;
+            } else {
+              fileLineColsBuilder_.addAllMessages(other.fileLineCols_);
+            }
+          }
+        }
+        this.mergeUnknownFields(other.unknownFields);
+        onChanged();
+        return this;
+      }
+
+      @java.lang.Override
+      public final boolean isInitialized() {
+        return true;
+      }
+
+      @java.lang.Override
+      public Builder mergeFrom(
+          com.google.protobuf.CodedInputStream input,
+          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+          throws java.io.IOException {
+        org.tensorflow.proto.framework.GraphDebugInfo.StackTrace parsedMessage = null;
+        try {
+          parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+        } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+          parsedMessage = (org.tensorflow.proto.framework.GraphDebugInfo.StackTrace) e.getUnfinishedMessage();
+          throw e.unwrapIOException();
+        } finally {
+          if (parsedMessage != null) {
+            mergeFrom(parsedMessage);
+          }
+        }
+        return this;
+      }
+      private int bitField0_;
+
+      private java.util.List<org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol> fileLineCols_ =
+        java.util.Collections.emptyList();
+      private void ensureFileLineColsIsMutable() {
+        if (!((bitField0_ & 0x00000001) != 0)) {
+          fileLineCols_ = new java.util.ArrayList<org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol>(fileLineCols_);
+          bitField0_ |= 0x00000001;
+         }
+      }
+
+      private com.google.protobuf.RepeatedFieldBuilderV3<
+          org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol, org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol.Builder, org.tensorflow.proto.framework.GraphDebugInfo.FileLineColOrBuilder> fileLineColsBuilder_;
+
+      /**
+       * <pre>
+       * Each line in the stack trace.
+       * </pre>
+       *
+       * <code>repeated .tensorflow.GraphDebugInfo.FileLineCol file_line_cols = 1;</code>
+       */
+      public java.util.List<org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol> getFileLineColsList() {
+        if (fileLineColsBuilder_ == null) {
+          return java.util.Collections.unmodifiableList(fileLineCols_);
+        } else {
+          return fileLineColsBuilder_.getMessageList();
+        }
+      }
+      /**
+       * <pre>
+       * Each line in the stack trace.
+       * </pre>
+       *
+       * <code>repeated .tensorflow.GraphDebugInfo.FileLineCol file_line_cols = 1;</code>
+       */
+      public int getFileLineColsCount() {
+        if (fileLineColsBuilder_ == null) {
+          return fileLineCols_.size();
+        } else {
+          return fileLineColsBuilder_.getCount();
+        }
+      }
+      /**
+       * <pre>
+       * Each line in the stack trace.
+       * </pre>
+       *
+       * <code>repeated .tensorflow.GraphDebugInfo.FileLineCol file_line_cols = 1;</code>
+       */
+      public org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol getFileLineCols(int index) {
+        if (fileLineColsBuilder_ == null) {
+          return fileLineCols_.get(index);
+        } else {
+          return fileLineColsBuilder_.getMessage(index);
+        }
+      }
+      /**
+       * <pre>
+       * Each line in the stack trace.
+       * </pre>
+       *
+       * <code>repeated .tensorflow.GraphDebugInfo.FileLineCol file_line_cols = 1;</code>
+       */
+      public Builder setFileLineCols(
+          int index, org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol value) {
+        if (fileLineColsBuilder_ == null) {
+          if (value == null) {
+            throw new NullPointerException();
+          }
+          ensureFileLineColsIsMutable();
+          fileLineCols_.set(index, value);
+          onChanged();
+        } else {
+          fileLineColsBuilder_.setMessage(index, value);
+        }
+        return this;
+      }
+      /**
+       * <pre>
+       * Each line in the stack trace.
+       * </pre>
+       *
+       * <code>repeated .tensorflow.GraphDebugInfo.FileLineCol file_line_cols = 1;</code>
+       */
+      public Builder setFileLineCols(
+          int index, org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol.Builder builderForValue) {
+        if (fileLineColsBuilder_ == null) {
+          ensureFileLineColsIsMutable();
+          fileLineCols_.set(index, builderForValue.build());
+          onChanged();
+        } else {
+          fileLineColsBuilder_.setMessage(index, builderForValue.build());
+        }
+        return this;
+      }
+      /**
+       * <pre>
+       * Each line in the stack trace.
+       * </pre>
+       *
+       * <code>repeated .tensorflow.GraphDebugInfo.FileLineCol file_line_cols = 1;</code>
+       */
+      public Builder addFileLineCols(org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol value) {
+        if (fileLineColsBuilder_ == null) {
+          if (value == null) {
+            throw new NullPointerException();
+          }
+          ensureFileLineColsIsMutable();
+          fileLineCols_.add(value);
+          onChanged();
+        } else {
+          fileLineColsBuilder_.addMessage(value);
+        }
+        return this;
+      }
+      /**
+       * <pre>
+       * Each line in the stack trace.
+       * </pre>
+       *
+       * <code>repeated .tensorflow.GraphDebugInfo.FileLineCol file_line_cols = 1;</code>
+       */
+      public Builder addFileLineCols(
+          int index, org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol value) {
+        if (fileLineColsBuilder_ == null) {
+          if (value == null) {
+            throw new NullPointerException();
+          }
+          ensureFileLineColsIsMutable();
+          fileLineCols_.add(index, value);
+          onChanged();
+        } else {
+          fileLineColsBuilder_.addMessage(index, value);
+        }
+        return this;
+      }
+      /**
+       * <pre>
+       * Each line in the stack trace.
+       * </pre>
+       *
+       * <code>repeated .tensorflow.GraphDebugInfo.FileLineCol file_line_cols = 1;</code>
+       */
+      public Builder addFileLineCols(
+          org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol.Builder builderForValue) {
+        if (fileLineColsBuilder_ == null) {
+          ensureFileLineColsIsMutable();
+          fileLineCols_.add(builderForValue.build());
+          onChanged();
+        } else {
+          fileLineColsBuilder_.addMessage(builderForValue.build());
+        }
+        return this;
+      }
+      /**
+       * <pre>
+       * Each line in the stack trace.
+       * </pre>
+       *
+       * <code>repeated .tensorflow.GraphDebugInfo.FileLineCol file_line_cols = 1;</code>
+       */
+      public Builder addFileLineCols(
+          int index, org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol.Builder builderForValue) {
+        if (fileLineColsBuilder_ == null) {
+          ensureFileLineColsIsMutable();
+          fileLineCols_.add(index, builderForValue.build());
+          onChanged();
+        } else {
+          fileLineColsBuilder_.addMessage(index, builderForValue.build());
+        }
+        return this;
+      }
+      /**
+       * <pre>
+       * Each line in the stack trace.
+       * </pre>
+       *
+       * <code>repeated .tensorflow.GraphDebugInfo.FileLineCol file_line_cols = 1;</code>
+       */
+      public Builder addAllFileLineCols(
+          java.lang.Iterable<? extends org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol> values) {
+        if (fileLineColsBuilder_ == null) {
+          ensureFileLineColsIsMutable();
+          com.google.protobuf.AbstractMessageLite.Builder.addAll(
+              values, fileLineCols_);
+          onChanged();
+        } else {
+          fileLineColsBuilder_.addAllMessages(values);
+        }
+        return this;
+      }
+      /**
+       * <pre>
+       * Each line in the stack trace.
+       * </pre>
+       *
+       * <code>repeated .tensorflow.GraphDebugInfo.FileLineCol file_line_cols = 1;</code>
+       */
+      public Builder clearFileLineCols() {
+        if (fileLineColsBuilder_ == null) {
+          fileLineCols_ = java.util.Collections.emptyList();
+          bitField0_ = (bitField0_ & ~0x00000001);
+          onChanged();
+        } else {
+          fileLineColsBuilder_.clear();
+        }
+        return this;
+      }
+      /**
+       * <pre>
+       * Each line in the stack trace.
+       * </pre>
+       *
+       * <code>repeated .tensorflow.GraphDebugInfo.FileLineCol file_line_cols = 1;</code>
+       */
+      public Builder removeFileLineCols(int index) {
+        if (fileLineColsBuilder_ == null) {
+          ensureFileLineColsIsMutable();
+          fileLineCols_.remove(index);
+          onChanged();
+        } else {
+          fileLineColsBuilder_.remove(index);
+        }
+        return this;
+      }
+      /**
+       * <pre>
+       * Each line in the stack trace.
+       * </pre>
+       *
+       * <code>repeated .tensorflow.GraphDebugInfo.FileLineCol file_line_cols = 1;</code>
+       */
+      public org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol.Builder getFileLineColsBuilder(
+          int index) {
+        return getFileLineColsFieldBuilder().getBuilder(index);
+      }
+      /**
+       * <pre>
+       * Each line in the stack trace.
+       * </pre>
+       *
+       * <code>repeated .tensorflow.GraphDebugInfo.FileLineCol file_line_cols = 1;</code>
+       */
+      public org.tensorflow.proto.framework.GraphDebugInfo.FileLineColOrBuilder getFileLineColsOrBuilder(
+          int index) {
+        if (fileLineColsBuilder_ == null) {
+          return fileLineCols_.get(index);  } else {
+          return fileLineColsBuilder_.getMessageOrBuilder(index);
+        }
+      }
+      /**
+       * <pre>
+       * Each line in the stack trace.
+       * </pre>
+       *
+       * <code>repeated .tensorflow.GraphDebugInfo.FileLineCol file_line_cols = 1;</code>
+       */
+      public java.util.List<? extends org.tensorflow.proto.framework.GraphDebugInfo.FileLineColOrBuilder> 
+           getFileLineColsOrBuilderList() {
+        if (fileLineColsBuilder_ != null) {
+          return fileLineColsBuilder_.getMessageOrBuilderList();
+        } else {
+          return java.util.Collections.unmodifiableList(fileLineCols_);
+        }
+      }
+      /**
+       * <pre>
+       * Each line in the stack trace.
+       * </pre>
+       *
+       * <code>repeated .tensorflow.GraphDebugInfo.FileLineCol file_line_cols = 1;</code>
+       */
+      public org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol.Builder addFileLineColsBuilder() {
+        return getFileLineColsFieldBuilder().addBuilder(
+            org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol.getDefaultInstance());
+      }
+      /**
+       * <pre>
+       * Each line in the stack trace.
+       * </pre>
+       *
+       * <code>repeated .tensorflow.GraphDebugInfo.FileLineCol file_line_cols = 1;</code>
+       */
+      public org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol.Builder addFileLineColsBuilder(
+          int index) {
+        return getFileLineColsFieldBuilder().addBuilder(
+            index, org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol.getDefaultInstance());
+      }
+      /**
+       * <pre>
+       * Each line in the stack trace.
+       * </pre>
+       *
+       * <code>repeated .tensorflow.GraphDebugInfo.FileLineCol file_line_cols = 1;</code>
+       */
+      public java.util.List<org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol.Builder> 
+           getFileLineColsBuilderList() {
+        return getFileLineColsFieldBuilder().getBuilderList();
+      }
+      private com.google.protobuf.RepeatedFieldBuilderV3<
+          org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol, org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol.Builder, org.tensorflow.proto.framework.GraphDebugInfo.FileLineColOrBuilder> 
+          getFileLineColsFieldBuilder() {
+        if (fileLineColsBuilder_ == null) {
+          fileLineColsBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
+              org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol, org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol.Builder, org.tensorflow.proto.framework.GraphDebugInfo.FileLineColOrBuilder>(
+                  fileLineCols_,
+                  ((bitField0_ & 0x00000001) != 0),
+                  getParentForChildren(),
+                  isClean());
+          fileLineCols_ = null;
+        }
+        return fileLineColsBuilder_;
+      }
+      @java.lang.Override
+      public final Builder setUnknownFields(
+          final com.google.protobuf.UnknownFieldSet unknownFields) {
+        return super.setUnknownFields(unknownFields);
+      }
+
+      @java.lang.Override
+      public final Builder mergeUnknownFields(
+          final com.google.protobuf.UnknownFieldSet unknownFields) {
+        return super.mergeUnknownFields(unknownFields);
+      }
+
+
+      // @@protoc_insertion_point(builder_scope:tensorflow.GraphDebugInfo.StackTrace)
+    }
+
+    // @@protoc_insertion_point(class_scope:tensorflow.GraphDebugInfo.StackTrace)
+    private static final org.tensorflow.proto.framework.GraphDebugInfo.StackTrace DEFAULT_INSTANCE;
+    static {
+      DEFAULT_INSTANCE = new org.tensorflow.proto.framework.GraphDebugInfo.StackTrace();
+    }
+
+    public static org.tensorflow.proto.framework.GraphDebugInfo.StackTrace getDefaultInstance() {
+      return DEFAULT_INSTANCE;
+    }
+
+    private static final com.google.protobuf.Parser<StackTrace>
+        PARSER = new com.google.protobuf.AbstractParser<StackTrace>() {
+      @java.lang.Override
+      public StackTrace parsePartialFrom(
+          com.google.protobuf.CodedInputStream input,
+          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+          throws com.google.protobuf.InvalidProtocolBufferException {
+        return new StackTrace(input, extensionRegistry);
+      }
+    };
+
+    public static com.google.protobuf.Parser<StackTrace> parser() {
+      return PARSER;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Parser<StackTrace> getParserForType() {
+      return PARSER;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.GraphDebugInfo.StackTrace getDefaultInstanceForType() {
+      return DEFAULT_INSTANCE;
+    }
+
+  }
+
+  public static final int FILES_FIELD_NUMBER = 1;
+  private com.google.protobuf.LazyStringList files_;
+  /**
+   * <pre>
+   * This stores all the source code file names and can be indexed by the
+   * `file_index`.
+   * </pre>
+   *
+   * <code>repeated string files = 1;</code>
+   */
+  public com.google.protobuf.ProtocolStringList
+      getFilesList() {
+    return files_;
+  }
+  /**
+   * <pre>
+   * This stores all the source code file names and can be indexed by the
+   * `file_index`.
+   * </pre>
+   *
+   * <code>repeated string files = 1;</code>
+   */
+  public int getFilesCount() {
+    return files_.size();
+  }
+  /**
+   * <pre>
+   * This stores all the source code file names and can be indexed by the
+   * `file_index`.
+   * </pre>
+   *
+   * <code>repeated string files = 1;</code>
+   */
+  public java.lang.String getFiles(int index) {
+    return files_.get(index);
+  }
+  /**
+   * <pre>
+   * This stores all the source code file names and can be indexed by the
+   * `file_index`.
+   * </pre>
+   *
+   * <code>repeated string files = 1;</code>
+   */
+  public com.google.protobuf.ByteString
+      getFilesBytes(int index) {
+    return files_.getByteString(index);
+  }
+
+  public static final int TRACES_FIELD_NUMBER = 2;
+  private static final class TracesDefaultEntryHolder {
+    static final com.google.protobuf.MapEntry<
+        java.lang.String, org.tensorflow.proto.framework.GraphDebugInfo.StackTrace> defaultEntry =
+            com.google.protobuf.MapEntry
+            .<java.lang.String, org.tensorflow.proto.framework.GraphDebugInfo.StackTrace>newDefaultInstance(
+                org.tensorflow.proto.framework.GraphDebugInfoProtos.internal_static_tensorflow_GraphDebugInfo_TracesEntry_descriptor, 
+                com.google.protobuf.WireFormat.FieldType.STRING,
+                "",
+                com.google.protobuf.WireFormat.FieldType.MESSAGE,
+                org.tensorflow.proto.framework.GraphDebugInfo.StackTrace.getDefaultInstance());
+  }
+  private com.google.protobuf.MapField<
+      java.lang.String, org.tensorflow.proto.framework.GraphDebugInfo.StackTrace> traces_;
+  private com.google.protobuf.MapField<java.lang.String, org.tensorflow.proto.framework.GraphDebugInfo.StackTrace>
+  internalGetTraces() {
+    if (traces_ == null) {
+      return com.google.protobuf.MapField.emptyMapField(
+          TracesDefaultEntryHolder.defaultEntry);
+    }
+    return traces_;
+  }
+
+  public int getTracesCount() {
+    return internalGetTraces().getMap().size();
+  }
+  /**
+   * <pre>
+   * This maps a node name to a stack trace in the source code.
+   * The map key is a mangling of the containing function and op name with
+   * syntax:
+   *   op.name '&#64;' func_name
+   * For ops in the top-level graph, the func_name is the empty string.
+   * Note that op names are restricted to a small number of characters which
+   * exclude '&#64;', making it impossible to collide keys of this form. Function
+   * names accept a much wider set of characters.
+   * It would be preferable to avoid mangling and use a tuple key of (op.name,
+   * func_name), but this is not supported with protocol buffers.
+   * </pre>
+   *
+   * <code>map&lt;string, .tensorflow.GraphDebugInfo.StackTrace&gt; traces = 2;</code>
+   */
+
+  public boolean containsTraces(
+      java.lang.String key) {
+    if (key == null) { throw new java.lang.NullPointerException(); }
+    return internalGetTraces().getMap().containsKey(key);
+  }
+  /**
+   * Use {@link #getTracesMap()} instead.
+   */
+  @java.lang.Deprecated
+  public java.util.Map<java.lang.String, org.tensorflow.proto.framework.GraphDebugInfo.StackTrace> getTraces() {
+    return getTracesMap();
+  }
+  /**
+   * <pre>
+   * This maps a node name to a stack trace in the source code.
+   * The map key is a mangling of the containing function and op name with
+   * syntax:
+   *   op.name '&#64;' func_name
+   * For ops in the top-level graph, the func_name is the empty string.
+   * Note that op names are restricted to a small number of characters which
+   * exclude '&#64;', making it impossible to collide keys of this form. Function
+   * names accept a much wider set of characters.
+   * It would be preferable to avoid mangling and use a tuple key of (op.name,
+   * func_name), but this is not supported with protocol buffers.
+   * </pre>
+   *
+   * <code>map&lt;string, .tensorflow.GraphDebugInfo.StackTrace&gt; traces = 2;</code>
+   */
+
+  public java.util.Map<java.lang.String, org.tensorflow.proto.framework.GraphDebugInfo.StackTrace> getTracesMap() {
+    return internalGetTraces().getMap();
+  }
+  /**
+   * <pre>
+   * This maps a node name to a stack trace in the source code.
+   * The map key is a mangling of the containing function and op name with
+   * syntax:
+   *   op.name '&#64;' func_name
+   * For ops in the top-level graph, the func_name is the empty string.
+   * Note that op names are restricted to a small number of characters which
+   * exclude '&#64;', making it impossible to collide keys of this form. Function
+   * names accept a much wider set of characters.
+   * It would be preferable to avoid mangling and use a tuple key of (op.name,
+   * func_name), but this is not supported with protocol buffers.
+   * </pre>
+   *
+   * <code>map&lt;string, .tensorflow.GraphDebugInfo.StackTrace&gt; traces = 2;</code>
+   */
+
+  public org.tensorflow.proto.framework.GraphDebugInfo.StackTrace getTracesOrDefault(
+      java.lang.String key,
+      org.tensorflow.proto.framework.GraphDebugInfo.StackTrace defaultValue) {
+    if (key == null) { throw new java.lang.NullPointerException(); }
+    java.util.Map<java.lang.String, org.tensorflow.proto.framework.GraphDebugInfo.StackTrace> map =
+        internalGetTraces().getMap();
+    return map.containsKey(key) ? map.get(key) : defaultValue;
+  }
+  /**
+   * <pre>
+   * This maps a node name to a stack trace in the source code.
+   * The map key is a mangling of the containing function and op name with
+   * syntax:
+   *   op.name '&#64;' func_name
+   * For ops in the top-level graph, the func_name is the empty string.
+   * Note that op names are restricted to a small number of characters which
+   * exclude '&#64;', making it impossible to collide keys of this form. Function
+   * names accept a much wider set of characters.
+   * It would be preferable to avoid mangling and use a tuple key of (op.name,
+   * func_name), but this is not supported with protocol buffers.
+   * </pre>
+   *
+   * <code>map&lt;string, .tensorflow.GraphDebugInfo.StackTrace&gt; traces = 2;</code>
+   */
+
+  public org.tensorflow.proto.framework.GraphDebugInfo.StackTrace getTracesOrThrow(
+      java.lang.String key) {
+    if (key == null) { throw new java.lang.NullPointerException(); }
+    java.util.Map<java.lang.String, org.tensorflow.proto.framework.GraphDebugInfo.StackTrace> map =
+        internalGetTraces().getMap();
+    if (!map.containsKey(key)) {
+      throw new java.lang.IllegalArgumentException();
+    }
+    return map.get(key);
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    for (int i = 0; i < files_.size(); i++) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 1, files_.getRaw(i));
+    }
+    com.google.protobuf.GeneratedMessageV3
+      .serializeStringMapTo(
+        output,
+        internalGetTraces(),
+        TracesDefaultEntryHolder.defaultEntry,
+        2);
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    {
+      int dataSize = 0;
+      for (int i = 0; i < files_.size(); i++) {
+        dataSize += computeStringSizeNoTag(files_.getRaw(i));
+      }
+      size += dataSize;
+      size += 1 * getFilesList().size();
+    }
+    for (java.util.Map.Entry<java.lang.String, org.tensorflow.proto.framework.GraphDebugInfo.StackTrace> entry
+         : internalGetTraces().getMap().entrySet()) {
+      com.google.protobuf.MapEntry<java.lang.String, org.tensorflow.proto.framework.GraphDebugInfo.StackTrace>
+      traces__ = TracesDefaultEntryHolder.defaultEntry.newBuilderForType()
+          .setKey(entry.getKey())
+          .setValue(entry.getValue())
+          .build();
+      size += com.google.protobuf.CodedOutputStream
+          .computeMessageSize(2, traces__);
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.framework.GraphDebugInfo)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.framework.GraphDebugInfo other = (org.tensorflow.proto.framework.GraphDebugInfo) obj;
+
+    if (!getFilesList()
+        .equals(other.getFilesList())) return false;
+    if (!internalGetTraces().equals(
+        other.internalGetTraces())) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    if (getFilesCount() > 0) {
+      hash = (37 * hash) + FILES_FIELD_NUMBER;
+      hash = (53 * hash) + getFilesList().hashCode();
+    }
+    if (!internalGetTraces().getMap().isEmpty()) {
+      hash = (37 * hash) + TRACES_FIELD_NUMBER;
+      hash = (53 * hash) + internalGetTraces().hashCode();
+    }
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.framework.GraphDebugInfo parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.GraphDebugInfo parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.GraphDebugInfo parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.GraphDebugInfo parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.GraphDebugInfo parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.GraphDebugInfo parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.GraphDebugInfo parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.GraphDebugInfo parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.GraphDebugInfo parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.GraphDebugInfo parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.GraphDebugInfo parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.GraphDebugInfo parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.framework.GraphDebugInfo prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * Protobuf type {@code tensorflow.GraphDebugInfo}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.GraphDebugInfo)
+      org.tensorflow.proto.framework.GraphDebugInfoOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.GraphDebugInfoProtos.internal_static_tensorflow_GraphDebugInfo_descriptor;
+    }
+
+    @SuppressWarnings({"rawtypes"})
+    protected com.google.protobuf.MapField internalGetMapField(
+        int number) {
+      switch (number) {
+        case 2:
+          return internalGetTraces();
+        default:
+          throw new RuntimeException(
+              "Invalid map field number: " + number);
+      }
+    }
+    @SuppressWarnings({"rawtypes"})
+    protected com.google.protobuf.MapField internalGetMutableMapField(
+        int number) {
+      switch (number) {
+        case 2:
+          return internalGetMutableTraces();
+        default:
+          throw new RuntimeException(
+              "Invalid map field number: " + number);
+      }
+    }
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.GraphDebugInfoProtos.internal_static_tensorflow_GraphDebugInfo_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.GraphDebugInfo.class, org.tensorflow.proto.framework.GraphDebugInfo.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.framework.GraphDebugInfo.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      files_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+      bitField0_ = (bitField0_ & ~0x00000001);
+      internalGetMutableTraces().clear();
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.framework.GraphDebugInfoProtos.internal_static_tensorflow_GraphDebugInfo_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.GraphDebugInfo getDefaultInstanceForType() {
+      return org.tensorflow.proto.framework.GraphDebugInfo.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.GraphDebugInfo build() {
+      org.tensorflow.proto.framework.GraphDebugInfo result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.GraphDebugInfo buildPartial() {
+      org.tensorflow.proto.framework.GraphDebugInfo result = new org.tensorflow.proto.framework.GraphDebugInfo(this);
+      int from_bitField0_ = bitField0_;
+      if (((bitField0_ & 0x00000001) != 0)) {
+        files_ = files_.getUnmodifiableView();
+        bitField0_ = (bitField0_ & ~0x00000001);
+      }
+      result.files_ = files_;
+      result.traces_ = internalGetTraces();
+      result.traces_.makeImmutable();
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.framework.GraphDebugInfo) {
+        return mergeFrom((org.tensorflow.proto.framework.GraphDebugInfo)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.framework.GraphDebugInfo other) {
+      if (other == org.tensorflow.proto.framework.GraphDebugInfo.getDefaultInstance()) return this;
+      if (!other.files_.isEmpty()) {
+        if (files_.isEmpty()) {
+          files_ = other.files_;
+          bitField0_ = (bitField0_ & ~0x00000001);
+        } else {
+          ensureFilesIsMutable();
+          files_.addAll(other.files_);
+        }
+        onChanged();
+      }
+      internalGetMutableTraces().mergeFrom(
+          other.internalGetTraces());
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.framework.GraphDebugInfo parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.framework.GraphDebugInfo) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+    private int bitField0_;
+
+    private com.google.protobuf.LazyStringList files_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+    private void ensureFilesIsMutable() {
+      if (!((bitField0_ & 0x00000001) != 0)) {
+        files_ = new com.google.protobuf.LazyStringArrayList(files_);
+        bitField0_ |= 0x00000001;
+       }
+    }
+    /**
+     * <pre>
+     * This stores all the source code file names and can be indexed by the
+     * `file_index`.
+     * </pre>
+     *
+     * <code>repeated string files = 1;</code>
+     */
+    public com.google.protobuf.ProtocolStringList
+        getFilesList() {
+      return files_.getUnmodifiableView();
+    }
+    /**
+     * <pre>
+     * This stores all the source code file names and can be indexed by the
+     * `file_index`.
+     * </pre>
+     *
+     * <code>repeated string files = 1;</code>
+     */
+    public int getFilesCount() {
+      return files_.size();
+    }
+    /**
+     * <pre>
+     * This stores all the source code file names and can be indexed by the
+     * `file_index`.
+     * </pre>
+     *
+     * <code>repeated string files = 1;</code>
+     */
+    public java.lang.String getFiles(int index) {
+      return files_.get(index);
+    }
+    /**
+     * <pre>
+     * This stores all the source code file names and can be indexed by the
+     * `file_index`.
+     * </pre>
+     *
+     * <code>repeated string files = 1;</code>
+     */
+    public com.google.protobuf.ByteString
+        getFilesBytes(int index) {
+      return files_.getByteString(index);
+    }
+    /**
+     * <pre>
+     * This stores all the source code file names and can be indexed by the
+     * `file_index`.
+     * </pre>
+     *
+     * <code>repeated string files = 1;</code>
+     */
+    public Builder setFiles(
+        int index, java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  ensureFilesIsMutable();
+      files_.set(index, value);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * This stores all the source code file names and can be indexed by the
+     * `file_index`.
+     * </pre>
+     *
+     * <code>repeated string files = 1;</code>
+     */
+    public Builder addFiles(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  ensureFilesIsMutable();
+      files_.add(value);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * This stores all the source code file names and can be indexed by the
+     * `file_index`.
+     * </pre>
+     *
+     * <code>repeated string files = 1;</code>
+     */
+    public Builder addAllFiles(
+        java.lang.Iterable<java.lang.String> values) {
+      ensureFilesIsMutable();
+      com.google.protobuf.AbstractMessageLite.Builder.addAll(
+          values, files_);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * This stores all the source code file names and can be indexed by the
+     * `file_index`.
+     * </pre>
+     *
+     * <code>repeated string files = 1;</code>
+     */
+    public Builder clearFiles() {
+      files_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+      bitField0_ = (bitField0_ & ~0x00000001);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * This stores all the source code file names and can be indexed by the
+     * `file_index`.
+     * </pre>
+     *
+     * <code>repeated string files = 1;</code>
+     */
+    public Builder addFilesBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      ensureFilesIsMutable();
+      files_.add(value);
+      onChanged();
+      return this;
+    }
+
+    private com.google.protobuf.MapField<
+        java.lang.String, org.tensorflow.proto.framework.GraphDebugInfo.StackTrace> traces_;
+    private com.google.protobuf.MapField<java.lang.String, org.tensorflow.proto.framework.GraphDebugInfo.StackTrace>
+    internalGetTraces() {
+      if (traces_ == null) {
+        return com.google.protobuf.MapField.emptyMapField(
+            TracesDefaultEntryHolder.defaultEntry);
+      }
+      return traces_;
+    }
+    private com.google.protobuf.MapField<java.lang.String, org.tensorflow.proto.framework.GraphDebugInfo.StackTrace>
+    internalGetMutableTraces() {
+      onChanged();;
+      if (traces_ == null) {
+        traces_ = com.google.protobuf.MapField.newMapField(
+            TracesDefaultEntryHolder.defaultEntry);
+      }
+      if (!traces_.isMutable()) {
+        traces_ = traces_.copy();
+      }
+      return traces_;
+    }
+
+    public int getTracesCount() {
+      return internalGetTraces().getMap().size();
+    }
+    /**
+     * <pre>
+     * This maps a node name to a stack trace in the source code.
+     * The map key is a mangling of the containing function and op name with
+     * syntax:
+     *   op.name '&#64;' func_name
+     * For ops in the top-level graph, the func_name is the empty string.
+     * Note that op names are restricted to a small number of characters which
+     * exclude '&#64;', making it impossible to collide keys of this form. Function
+     * names accept a much wider set of characters.
+     * It would be preferable to avoid mangling and use a tuple key of (op.name,
+     * func_name), but this is not supported with protocol buffers.
+     * </pre>
+     *
+     * <code>map&lt;string, .tensorflow.GraphDebugInfo.StackTrace&gt; traces = 2;</code>
+     */
+
+    public boolean containsTraces(
+        java.lang.String key) {
+      if (key == null) { throw new java.lang.NullPointerException(); }
+      return internalGetTraces().getMap().containsKey(key);
+    }
+    /**
+     * Use {@link #getTracesMap()} instead.
+     */
+    @java.lang.Deprecated
+    public java.util.Map<java.lang.String, org.tensorflow.proto.framework.GraphDebugInfo.StackTrace> getTraces() {
+      return getTracesMap();
+    }
+    /**
+     * <pre>
+     * This maps a node name to a stack trace in the source code.
+     * The map key is a mangling of the containing function and op name with
+     * syntax:
+     *   op.name '&#64;' func_name
+     * For ops in the top-level graph, the func_name is the empty string.
+     * Note that op names are restricted to a small number of characters which
+     * exclude '&#64;', making it impossible to collide keys of this form. Function
+     * names accept a much wider set of characters.
+     * It would be preferable to avoid mangling and use a tuple key of (op.name,
+     * func_name), but this is not supported with protocol buffers.
+     * </pre>
+     *
+     * <code>map&lt;string, .tensorflow.GraphDebugInfo.StackTrace&gt; traces = 2;</code>
+     */
+
+    public java.util.Map<java.lang.String, org.tensorflow.proto.framework.GraphDebugInfo.StackTrace> getTracesMap() {
+      return internalGetTraces().getMap();
+    }
+    /**
+     * <pre>
+     * This maps a node name to a stack trace in the source code.
+     * The map key is a mangling of the containing function and op name with
+     * syntax:
+     *   op.name '&#64;' func_name
+     * For ops in the top-level graph, the func_name is the empty string.
+     * Note that op names are restricted to a small number of characters which
+     * exclude '&#64;', making it impossible to collide keys of this form. Function
+     * names accept a much wider set of characters.
+     * It would be preferable to avoid mangling and use a tuple key of (op.name,
+     * func_name), but this is not supported with protocol buffers.
+     * </pre>
+     *
+     * <code>map&lt;string, .tensorflow.GraphDebugInfo.StackTrace&gt; traces = 2;</code>
+     */
+
+    public org.tensorflow.proto.framework.GraphDebugInfo.StackTrace getTracesOrDefault(
+        java.lang.String key,
+        org.tensorflow.proto.framework.GraphDebugInfo.StackTrace defaultValue) {
+      if (key == null) { throw new java.lang.NullPointerException(); }
+      java.util.Map<java.lang.String, org.tensorflow.proto.framework.GraphDebugInfo.StackTrace> map =
+          internalGetTraces().getMap();
+      return map.containsKey(key) ? map.get(key) : defaultValue;
+    }
+    /**
+     * <pre>
+     * This maps a node name to a stack trace in the source code.
+     * The map key is a mangling of the containing function and op name with
+     * syntax:
+     *   op.name '&#64;' func_name
+     * For ops in the top-level graph, the func_name is the empty string.
+     * Note that op names are restricted to a small number of characters which
+     * exclude '&#64;', making it impossible to collide keys of this form. Function
+     * names accept a much wider set of characters.
+     * It would be preferable to avoid mangling and use a tuple key of (op.name,
+     * func_name), but this is not supported with protocol buffers.
+     * </pre>
+     *
+     * <code>map&lt;string, .tensorflow.GraphDebugInfo.StackTrace&gt; traces = 2;</code>
+     */
+
+    public org.tensorflow.proto.framework.GraphDebugInfo.StackTrace getTracesOrThrow(
+        java.lang.String key) {
+      if (key == null) { throw new java.lang.NullPointerException(); }
+      java.util.Map<java.lang.String, org.tensorflow.proto.framework.GraphDebugInfo.StackTrace> map =
+          internalGetTraces().getMap();
+      if (!map.containsKey(key)) {
+        throw new java.lang.IllegalArgumentException();
+      }
+      return map.get(key);
+    }
+
+    public Builder clearTraces() {
+      internalGetMutableTraces().getMutableMap()
+          .clear();
+      return this;
+    }
+    /**
+     * <pre>
+     * This maps a node name to a stack trace in the source code.
+     * The map key is a mangling of the containing function and op name with
+     * syntax:
+     *   op.name '&#64;' func_name
+     * For ops in the top-level graph, the func_name is the empty string.
+     * Note that op names are restricted to a small number of characters which
+     * exclude '&#64;', making it impossible to collide keys of this form. Function
+     * names accept a much wider set of characters.
+     * It would be preferable to avoid mangling and use a tuple key of (op.name,
+     * func_name), but this is not supported with protocol buffers.
+     * </pre>
+     *
+     * <code>map&lt;string, .tensorflow.GraphDebugInfo.StackTrace&gt; traces = 2;</code>
+     */
+
+    public Builder removeTraces(
+        java.lang.String key) {
+      if (key == null) { throw new java.lang.NullPointerException(); }
+      internalGetMutableTraces().getMutableMap()
+          .remove(key);
+      return this;
+    }
+    /**
+     * Use alternate mutation accessors instead.
+     */
+    @java.lang.Deprecated
+    public java.util.Map<java.lang.String, org.tensorflow.proto.framework.GraphDebugInfo.StackTrace>
+    getMutableTraces() {
+      return internalGetMutableTraces().getMutableMap();
+    }
+    /**
+     * <pre>
+     * This maps a node name to a stack trace in the source code.
+     * The map key is a mangling of the containing function and op name with
+     * syntax:
+     *   op.name '&#64;' func_name
+     * For ops in the top-level graph, the func_name is the empty string.
+     * Note that op names are restricted to a small number of characters which
+     * exclude '&#64;', making it impossible to collide keys of this form. Function
+     * names accept a much wider set of characters.
+     * It would be preferable to avoid mangling and use a tuple key of (op.name,
+     * func_name), but this is not supported with protocol buffers.
+     * </pre>
+     *
+     * <code>map&lt;string, .tensorflow.GraphDebugInfo.StackTrace&gt; traces = 2;</code>
+     */
+    public Builder putTraces(
+        java.lang.String key,
+        org.tensorflow.proto.framework.GraphDebugInfo.StackTrace value) {
+      if (key == null) { throw new java.lang.NullPointerException(); }
+      if (value == null) { throw new java.lang.NullPointerException(); }
+      internalGetMutableTraces().getMutableMap()
+          .put(key, value);
+      return this;
+    }
+    /**
+     * <pre>
+     * This maps a node name to a stack trace in the source code.
+     * The map key is a mangling of the containing function and op name with
+     * syntax:
+     *   op.name '&#64;' func_name
+     * For ops in the top-level graph, the func_name is the empty string.
+     * Note that op names are restricted to a small number of characters which
+     * exclude '&#64;', making it impossible to collide keys of this form. Function
+     * names accept a much wider set of characters.
+     * It would be preferable to avoid mangling and use a tuple key of (op.name,
+     * func_name), but this is not supported with protocol buffers.
+     * </pre>
+     *
+     * <code>map&lt;string, .tensorflow.GraphDebugInfo.StackTrace&gt; traces = 2;</code>
+     */
+
+    public Builder putAllTraces(
+        java.util.Map<java.lang.String, org.tensorflow.proto.framework.GraphDebugInfo.StackTrace> values) {
+      internalGetMutableTraces().getMutableMap()
+          .putAll(values);
+      return this;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.GraphDebugInfo)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.GraphDebugInfo)
+  private static final org.tensorflow.proto.framework.GraphDebugInfo DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.GraphDebugInfo();
+  }
+
+  public static org.tensorflow.proto.framework.GraphDebugInfo getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<GraphDebugInfo>
+      PARSER = new com.google.protobuf.AbstractParser<GraphDebugInfo>() {
+    @java.lang.Override
+    public GraphDebugInfo parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new GraphDebugInfo(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<GraphDebugInfo> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<GraphDebugInfo> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.framework.GraphDebugInfo getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphDebugInfoOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphDebugInfoOrBuilder.java
new file mode 100644
index 00000000000..86c48de7b70
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphDebugInfoOrBuilder.java
@@ -0,0 +1,147 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/graph_debug_info.proto
+
+package org.tensorflow.proto.framework;
+
+public interface GraphDebugInfoOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.GraphDebugInfo)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <pre>
+   * This stores all the source code file names and can be indexed by the
+   * `file_index`.
+   * </pre>
+   *
+   * <code>repeated string files = 1;</code>
+   */
+  java.util.List<java.lang.String>
+      getFilesList();
+  /**
+   * <pre>
+   * This stores all the source code file names and can be indexed by the
+   * `file_index`.
+   * </pre>
+   *
+   * <code>repeated string files = 1;</code>
+   */
+  int getFilesCount();
+  /**
+   * <pre>
+   * This stores all the source code file names and can be indexed by the
+   * `file_index`.
+   * </pre>
+   *
+   * <code>repeated string files = 1;</code>
+   */
+  java.lang.String getFiles(int index);
+  /**
+   * <pre>
+   * This stores all the source code file names and can be indexed by the
+   * `file_index`.
+   * </pre>
+   *
+   * <code>repeated string files = 1;</code>
+   */
+  com.google.protobuf.ByteString
+      getFilesBytes(int index);
+
+  /**
+   * <pre>
+   * This maps a node name to a stack trace in the source code.
+   * The map key is a mangling of the containing function and op name with
+   * syntax:
+   *   op.name '&#64;' func_name
+   * For ops in the top-level graph, the func_name is the empty string.
+   * Note that op names are restricted to a small number of characters which
+   * exclude '&#64;', making it impossible to collide keys of this form. Function
+   * names accept a much wider set of characters.
+   * It would be preferable to avoid mangling and use a tuple key of (op.name,
+   * func_name), but this is not supported with protocol buffers.
+   * </pre>
+   *
+   * <code>map&lt;string, .tensorflow.GraphDebugInfo.StackTrace&gt; traces = 2;</code>
+   */
+  int getTracesCount();
+  /**
+   * <pre>
+   * This maps a node name to a stack trace in the source code.
+   * The map key is a mangling of the containing function and op name with
+   * syntax:
+   *   op.name '&#64;' func_name
+   * For ops in the top-level graph, the func_name is the empty string.
+   * Note that op names are restricted to a small number of characters which
+   * exclude '&#64;', making it impossible to collide keys of this form. Function
+   * names accept a much wider set of characters.
+   * It would be preferable to avoid mangling and use a tuple key of (op.name,
+   * func_name), but this is not supported with protocol buffers.
+   * </pre>
+   *
+   * <code>map&lt;string, .tensorflow.GraphDebugInfo.StackTrace&gt; traces = 2;</code>
+   */
+  boolean containsTraces(
+      java.lang.String key);
+  /**
+   * Use {@link #getTracesMap()} instead.
+   */
+  @java.lang.Deprecated
+  java.util.Map<java.lang.String, org.tensorflow.proto.framework.GraphDebugInfo.StackTrace>
+  getTraces();
+  /**
+   * <pre>
+   * This maps a node name to a stack trace in the source code.
+   * The map key is a mangling of the containing function and op name with
+   * syntax:
+   *   op.name '&#64;' func_name
+   * For ops in the top-level graph, the func_name is the empty string.
+   * Note that op names are restricted to a small number of characters which
+   * exclude '&#64;', making it impossible to collide keys of this form. Function
+   * names accept a much wider set of characters.
+   * It would be preferable to avoid mangling and use a tuple key of (op.name,
+   * func_name), but this is not supported with protocol buffers.
+   * </pre>
+   *
+   * <code>map&lt;string, .tensorflow.GraphDebugInfo.StackTrace&gt; traces = 2;</code>
+   */
+  java.util.Map<java.lang.String, org.tensorflow.proto.framework.GraphDebugInfo.StackTrace>
+  getTracesMap();
+  /**
+   * <pre>
+   * This maps a node name to a stack trace in the source code.
+   * The map key is a mangling of the containing function and op name with
+   * syntax:
+   *   op.name '&#64;' func_name
+   * For ops in the top-level graph, the func_name is the empty string.
+   * Note that op names are restricted to a small number of characters which
+   * exclude '&#64;', making it impossible to collide keys of this form. Function
+   * names accept a much wider set of characters.
+   * It would be preferable to avoid mangling and use a tuple key of (op.name,
+   * func_name), but this is not supported with protocol buffers.
+   * </pre>
+   *
+   * <code>map&lt;string, .tensorflow.GraphDebugInfo.StackTrace&gt; traces = 2;</code>
+   */
+
+  org.tensorflow.proto.framework.GraphDebugInfo.StackTrace getTracesOrDefault(
+      java.lang.String key,
+      org.tensorflow.proto.framework.GraphDebugInfo.StackTrace defaultValue);
+  /**
+   * <pre>
+   * This maps a node name to a stack trace in the source code.
+   * The map key is a mangling of the containing function and op name with
+   * syntax:
+   *   op.name '&#64;' func_name
+   * For ops in the top-level graph, the func_name is the empty string.
+   * Note that op names are restricted to a small number of characters which
+   * exclude '&#64;', making it impossible to collide keys of this form. Function
+   * names accept a much wider set of characters.
+   * It would be preferable to avoid mangling and use a tuple key of (op.name,
+   * func_name), but this is not supported with protocol buffers.
+   * </pre>
+   *
+   * <code>map&lt;string, .tensorflow.GraphDebugInfo.StackTrace&gt; traces = 2;</code>
+   */
+
+  org.tensorflow.proto.framework.GraphDebugInfo.StackTrace getTracesOrThrow(
+      java.lang.String key);
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphDebugInfoProtos.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphDebugInfoProtos.java
new file mode 100644
index 00000000000..c206eb2d9cf
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphDebugInfoProtos.java
@@ -0,0 +1,92 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/graph_debug_info.proto
+
+package org.tensorflow.proto.framework;
+
+public final class GraphDebugInfoProtos {
+  private GraphDebugInfoProtos() {}
+  public static void registerAllExtensions(
+      com.google.protobuf.ExtensionRegistryLite registry) {
+  }
+
+  public static void registerAllExtensions(
+      com.google.protobuf.ExtensionRegistry registry) {
+    registerAllExtensions(
+        (com.google.protobuf.ExtensionRegistryLite) registry);
+  }
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_GraphDebugInfo_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_GraphDebugInfo_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_GraphDebugInfo_FileLineCol_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_GraphDebugInfo_FileLineCol_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_GraphDebugInfo_StackTrace_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_GraphDebugInfo_StackTrace_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_GraphDebugInfo_TracesEntry_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_GraphDebugInfo_TracesEntry_fieldAccessorTable;
+
+  public static com.google.protobuf.Descriptors.FileDescriptor
+      getDescriptor() {
+    return descriptor;
+  }
+  private static  com.google.protobuf.Descriptors.FileDescriptor
+      descriptor;
+  static {
+    java.lang.String[] descriptorData = {
+      "\n/tensorflow/core/protobuf/graph_debug_i" +
+      "nfo.proto\022\ntensorflow\"\325\002\n\016GraphDebugInfo" +
+      "\022\r\n\005files\030\001 \003(\t\0226\n\006traces\030\002 \003(\0132&.tensor" +
+      "flow.GraphDebugInfo.TracesEntry\032X\n\013FileL" +
+      "ineCol\022\022\n\nfile_index\030\001 \001(\005\022\014\n\004line\030\002 \001(\005" +
+      "\022\013\n\003col\030\003 \001(\005\022\014\n\004func\030\004 \001(\t\022\014\n\004code\030\005 \001(" +
+      "\t\032L\n\nStackTrace\022>\n\016file_line_cols\030\001 \003(\0132" +
+      "&.tensorflow.GraphDebugInfo.FileLineCol\032" +
+      "T\n\013TracesEntry\022\013\n\003key\030\001 \001(\t\0224\n\005value\030\002 \001" +
+      "(\0132%.tensorflow.GraphDebugInfo.StackTrac" +
+      "e:\0028\001B\205\001\n\036org.tensorflow.proto.framework" +
+      "B\024GraphDebugInfoProtosP\001ZHgithub.com/ten" +
+      "sorflow/tensorflow/tensorflow/go/core/co" +
+      "re_protos_go_proto\370\001\001b\006proto3"
+    };
+    descriptor = com.google.protobuf.Descriptors.FileDescriptor
+      .internalBuildGeneratedFileFrom(descriptorData,
+        new com.google.protobuf.Descriptors.FileDescriptor[] {
+        });
+    internal_static_tensorflow_GraphDebugInfo_descriptor =
+      getDescriptor().getMessageTypes().get(0);
+    internal_static_tensorflow_GraphDebugInfo_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_GraphDebugInfo_descriptor,
+        new java.lang.String[] { "Files", "Traces", });
+    internal_static_tensorflow_GraphDebugInfo_FileLineCol_descriptor =
+      internal_static_tensorflow_GraphDebugInfo_descriptor.getNestedTypes().get(0);
+    internal_static_tensorflow_GraphDebugInfo_FileLineCol_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_GraphDebugInfo_FileLineCol_descriptor,
+        new java.lang.String[] { "FileIndex", "Line", "Col", "Func", "Code", });
+    internal_static_tensorflow_GraphDebugInfo_StackTrace_descriptor =
+      internal_static_tensorflow_GraphDebugInfo_descriptor.getNestedTypes().get(1);
+    internal_static_tensorflow_GraphDebugInfo_StackTrace_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_GraphDebugInfo_StackTrace_descriptor,
+        new java.lang.String[] { "FileLineCols", });
+    internal_static_tensorflow_GraphDebugInfo_TracesEntry_descriptor =
+      internal_static_tensorflow_GraphDebugInfo_descriptor.getNestedTypes().get(2);
+    internal_static_tensorflow_GraphDebugInfo_TracesEntry_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_GraphDebugInfo_TracesEntry_descriptor,
+        new java.lang.String[] { "Key", "Value", });
+  }
+
+  // @@protoc_insertion_point(outer_class_scope)
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphDef.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphDef.java
new file mode 100644
index 00000000000..18e75b5e768
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphDef.java
@@ -0,0 +1,1588 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/framework/graph.proto
+
+package org.tensorflow.proto.framework;
+
+/**
+ * <pre>
+ * Represents the graph of operations
+ * </pre>
+ *
+ * Protobuf type {@code tensorflow.GraphDef}
+ */
+public  final class GraphDef extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.GraphDef)
+    GraphDefOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use GraphDef.newBuilder() to construct.
+  private GraphDef(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private GraphDef() {
+    node_ = java.util.Collections.emptyList();
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new GraphDef();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private GraphDef(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    int mutable_bitField0_ = 0;
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 10: {
+            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
+              node_ = new java.util.ArrayList<org.tensorflow.proto.framework.NodeDef>();
+              mutable_bitField0_ |= 0x00000001;
+            }
+            node_.add(
+                input.readMessage(org.tensorflow.proto.framework.NodeDef.parser(), extensionRegistry));
+            break;
+          }
+          case 18: {
+            org.tensorflow.proto.framework.FunctionDefLibrary.Builder subBuilder = null;
+            if (library_ != null) {
+              subBuilder = library_.toBuilder();
+            }
+            library_ = input.readMessage(org.tensorflow.proto.framework.FunctionDefLibrary.parser(), extensionRegistry);
+            if (subBuilder != null) {
+              subBuilder.mergeFrom(library_);
+              library_ = subBuilder.buildPartial();
+            }
+
+            break;
+          }
+          case 24: {
+
+            version_ = input.readInt32();
+            break;
+          }
+          case 34: {
+            org.tensorflow.proto.framework.VersionDef.Builder subBuilder = null;
+            if (versions_ != null) {
+              subBuilder = versions_.toBuilder();
+            }
+            versions_ = input.readMessage(org.tensorflow.proto.framework.VersionDef.parser(), extensionRegistry);
+            if (subBuilder != null) {
+              subBuilder.mergeFrom(versions_);
+              versions_ = subBuilder.buildPartial();
+            }
+
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      if (((mutable_bitField0_ & 0x00000001) != 0)) {
+        node_ = java.util.Collections.unmodifiableList(node_);
+      }
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.framework.GraphProtos.internal_static_tensorflow_GraphDef_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.framework.GraphProtos.internal_static_tensorflow_GraphDef_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.framework.GraphDef.class, org.tensorflow.proto.framework.GraphDef.Builder.class);
+  }
+
+  public static final int NODE_FIELD_NUMBER = 1;
+  private java.util.List<org.tensorflow.proto.framework.NodeDef> node_;
+  /**
+   * <code>repeated .tensorflow.NodeDef node = 1;</code>
+   */
+  public java.util.List<org.tensorflow.proto.framework.NodeDef> getNodeList() {
+    return node_;
+  }
+  /**
+   * <code>repeated .tensorflow.NodeDef node = 1;</code>
+   */
+  public java.util.List<? extends org.tensorflow.proto.framework.NodeDefOrBuilder> 
+      getNodeOrBuilderList() {
+    return node_;
+  }
+  /**
+   * <code>repeated .tensorflow.NodeDef node = 1;</code>
+   */
+  public int getNodeCount() {
+    return node_.size();
+  }
+  /**
+   * <code>repeated .tensorflow.NodeDef node = 1;</code>
+   */
+  public org.tensorflow.proto.framework.NodeDef getNode(int index) {
+    return node_.get(index);
+  }
+  /**
+   * <code>repeated .tensorflow.NodeDef node = 1;</code>
+   */
+  public org.tensorflow.proto.framework.NodeDefOrBuilder getNodeOrBuilder(
+      int index) {
+    return node_.get(index);
+  }
+
+  public static final int VERSIONS_FIELD_NUMBER = 4;
+  private org.tensorflow.proto.framework.VersionDef versions_;
+  /**
+   * <pre>
+   * Compatibility versions of the graph.  See core/public/version.h for version
+   * history.  The GraphDef version is distinct from the TensorFlow version, and
+   * each release of TensorFlow will support a range of GraphDef versions.
+   * </pre>
+   *
+   * <code>.tensorflow.VersionDef versions = 4;</code>
+   */
+  public boolean hasVersions() {
+    return versions_ != null;
+  }
+  /**
+   * <pre>
+   * Compatibility versions of the graph.  See core/public/version.h for version
+   * history.  The GraphDef version is distinct from the TensorFlow version, and
+   * each release of TensorFlow will support a range of GraphDef versions.
+   * </pre>
+   *
+   * <code>.tensorflow.VersionDef versions = 4;</code>
+   */
+  public org.tensorflow.proto.framework.VersionDef getVersions() {
+    return versions_ == null ? org.tensorflow.proto.framework.VersionDef.getDefaultInstance() : versions_;
+  }
+  /**
+   * <pre>
+   * Compatibility versions of the graph.  See core/public/version.h for version
+   * history.  The GraphDef version is distinct from the TensorFlow version, and
+   * each release of TensorFlow will support a range of GraphDef versions.
+   * </pre>
+   *
+   * <code>.tensorflow.VersionDef versions = 4;</code>
+   */
+  public org.tensorflow.proto.framework.VersionDefOrBuilder getVersionsOrBuilder() {
+    return getVersions();
+  }
+
+  public static final int VERSION_FIELD_NUMBER = 3;
+  private int version_;
+  /**
+   * <pre>
+   * Deprecated single version field; use versions above instead.  Since all
+   * GraphDef changes before "versions" was introduced were forward
+   * compatible, this field is entirely ignored.
+   * </pre>
+   *
+   * <code>int32 version = 3 [deprecated = true];</code>
+   */
+  @java.lang.Deprecated public int getVersion() {
+    return version_;
+  }
+
+  public static final int LIBRARY_FIELD_NUMBER = 2;
+  private org.tensorflow.proto.framework.FunctionDefLibrary library_;
+  /**
+   * <pre>
+   * EXPERIMENTAL. DO NOT USE OR DEPEND ON THIS YET.
+   * "library" provides user-defined functions.
+   * Naming:
+   *   * library.function.name are in a flat namespace.
+   *     NOTE: We may need to change it to be hierarchical to support
+   *     different orgs. E.g.,
+   *     { "/google/nn", { ... }},
+   *     { "/google/vision", { ... }}
+   *     { "/org_foo/module_bar", { ... }}
+   *     map&lt;string, FunctionDefLib&gt; named_lib;
+   *   * If node[i].op is the name of one function in "library",
+   *     node[i] is deemed as a function call. Otherwise, node[i].op
+   *     must be a primitive operation supported by the runtime.
+   * Function call semantics:
+   *   * The callee may start execution as soon as some of its inputs
+   *     are ready. The caller may want to use Tuple() mechanism to
+   *     ensure all inputs are ready in the same time.
+   *   * The consumer of return values may start executing as soon as
+   *     the return values the consumer depends on are ready.  The
+   *     consumer may want to use Tuple() mechanism to ensure the
+   *     consumer does not start until all return values of the callee
+   *     function are ready.
+   * </pre>
+   *
+   * <code>.tensorflow.FunctionDefLibrary library = 2;</code>
+   */
+  public boolean hasLibrary() {
+    return library_ != null;
+  }
+  /**
+   * <pre>
+   * EXPERIMENTAL. DO NOT USE OR DEPEND ON THIS YET.
+   * "library" provides user-defined functions.
+   * Naming:
+   *   * library.function.name are in a flat namespace.
+   *     NOTE: We may need to change it to be hierarchical to support
+   *     different orgs. E.g.,
+   *     { "/google/nn", { ... }},
+   *     { "/google/vision", { ... }}
+   *     { "/org_foo/module_bar", { ... }}
+   *     map&lt;string, FunctionDefLib&gt; named_lib;
+   *   * If node[i].op is the name of one function in "library",
+   *     node[i] is deemed as a function call. Otherwise, node[i].op
+   *     must be a primitive operation supported by the runtime.
+   * Function call semantics:
+   *   * The callee may start execution as soon as some of its inputs
+   *     are ready. The caller may want to use Tuple() mechanism to
+   *     ensure all inputs are ready in the same time.
+   *   * The consumer of return values may start executing as soon as
+   *     the return values the consumer depends on are ready.  The
+   *     consumer may want to use Tuple() mechanism to ensure the
+   *     consumer does not start until all return values of the callee
+   *     function are ready.
+   * </pre>
+   *
+   * <code>.tensorflow.FunctionDefLibrary library = 2;</code>
+   */
+  public org.tensorflow.proto.framework.FunctionDefLibrary getLibrary() {
+    return library_ == null ? org.tensorflow.proto.framework.FunctionDefLibrary.getDefaultInstance() : library_;
+  }
+  /**
+   * <pre>
+   * EXPERIMENTAL. DO NOT USE OR DEPEND ON THIS YET.
+   * "library" provides user-defined functions.
+   * Naming:
+   *   * library.function.name are in a flat namespace.
+   *     NOTE: We may need to change it to be hierarchical to support
+   *     different orgs. E.g.,
+   *     { "/google/nn", { ... }},
+   *     { "/google/vision", { ... }}
+   *     { "/org_foo/module_bar", { ... }}
+   *     map&lt;string, FunctionDefLib&gt; named_lib;
+   *   * If node[i].op is the name of one function in "library",
+   *     node[i] is deemed as a function call. Otherwise, node[i].op
+   *     must be a primitive operation supported by the runtime.
+   * Function call semantics:
+   *   * The callee may start execution as soon as some of its inputs
+   *     are ready. The caller may want to use Tuple() mechanism to
+   *     ensure all inputs are ready in the same time.
+   *   * The consumer of return values may start executing as soon as
+   *     the return values the consumer depends on are ready.  The
+   *     consumer may want to use Tuple() mechanism to ensure the
+   *     consumer does not start until all return values of the callee
+   *     function are ready.
+   * </pre>
+   *
+   * <code>.tensorflow.FunctionDefLibrary library = 2;</code>
+   */
+  public org.tensorflow.proto.framework.FunctionDefLibraryOrBuilder getLibraryOrBuilder() {
+    return getLibrary();
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    for (int i = 0; i < node_.size(); i++) {
+      output.writeMessage(1, node_.get(i));
+    }
+    if (library_ != null) {
+      output.writeMessage(2, getLibrary());
+    }
+    if (version_ != 0) {
+      output.writeInt32(3, version_);
+    }
+    if (versions_ != null) {
+      output.writeMessage(4, getVersions());
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    for (int i = 0; i < node_.size(); i++) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(1, node_.get(i));
+    }
+    if (library_ != null) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(2, getLibrary());
+    }
+    if (version_ != 0) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt32Size(3, version_);
+    }
+    if (versions_ != null) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(4, getVersions());
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.framework.GraphDef)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.framework.GraphDef other = (org.tensorflow.proto.framework.GraphDef) obj;
+
+    if (!getNodeList()
+        .equals(other.getNodeList())) return false;
+    if (hasVersions() != other.hasVersions()) return false;
+    if (hasVersions()) {
+      if (!getVersions()
+          .equals(other.getVersions())) return false;
+    }
+    if (getVersion()
+        != other.getVersion()) return false;
+    if (hasLibrary() != other.hasLibrary()) return false;
+    if (hasLibrary()) {
+      if (!getLibrary()
+          .equals(other.getLibrary())) return false;
+    }
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    if (getNodeCount() > 0) {
+      hash = (37 * hash) + NODE_FIELD_NUMBER;
+      hash = (53 * hash) + getNodeList().hashCode();
+    }
+    if (hasVersions()) {
+      hash = (37 * hash) + VERSIONS_FIELD_NUMBER;
+      hash = (53 * hash) + getVersions().hashCode();
+    }
+    hash = (37 * hash) + VERSION_FIELD_NUMBER;
+    hash = (53 * hash) + getVersion();
+    if (hasLibrary()) {
+      hash = (37 * hash) + LIBRARY_FIELD_NUMBER;
+      hash = (53 * hash) + getLibrary().hashCode();
+    }
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.framework.GraphDef parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.GraphDef parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.GraphDef parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.GraphDef parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.GraphDef parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.GraphDef parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.GraphDef parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.GraphDef parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.GraphDef parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.GraphDef parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.GraphDef parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.GraphDef parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.framework.GraphDef prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * <pre>
+   * Represents the graph of operations
+   * </pre>
+   *
+   * Protobuf type {@code tensorflow.GraphDef}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.GraphDef)
+      org.tensorflow.proto.framework.GraphDefOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.GraphProtos.internal_static_tensorflow_GraphDef_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.GraphProtos.internal_static_tensorflow_GraphDef_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.GraphDef.class, org.tensorflow.proto.framework.GraphDef.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.framework.GraphDef.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+        getNodeFieldBuilder();
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      if (nodeBuilder_ == null) {
+        node_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000001);
+      } else {
+        nodeBuilder_.clear();
+      }
+      if (versionsBuilder_ == null) {
+        versions_ = null;
+      } else {
+        versions_ = null;
+        versionsBuilder_ = null;
+      }
+      version_ = 0;
+
+      if (libraryBuilder_ == null) {
+        library_ = null;
+      } else {
+        library_ = null;
+        libraryBuilder_ = null;
+      }
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.framework.GraphProtos.internal_static_tensorflow_GraphDef_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.GraphDef getDefaultInstanceForType() {
+      return org.tensorflow.proto.framework.GraphDef.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.GraphDef build() {
+      org.tensorflow.proto.framework.GraphDef result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.GraphDef buildPartial() {
+      org.tensorflow.proto.framework.GraphDef result = new org.tensorflow.proto.framework.GraphDef(this);
+      int from_bitField0_ = bitField0_;
+      if (nodeBuilder_ == null) {
+        if (((bitField0_ & 0x00000001) != 0)) {
+          node_ = java.util.Collections.unmodifiableList(node_);
+          bitField0_ = (bitField0_ & ~0x00000001);
+        }
+        result.node_ = node_;
+      } else {
+        result.node_ = nodeBuilder_.build();
+      }
+      if (versionsBuilder_ == null) {
+        result.versions_ = versions_;
+      } else {
+        result.versions_ = versionsBuilder_.build();
+      }
+      result.version_ = version_;
+      if (libraryBuilder_ == null) {
+        result.library_ = library_;
+      } else {
+        result.library_ = libraryBuilder_.build();
+      }
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.framework.GraphDef) {
+        return mergeFrom((org.tensorflow.proto.framework.GraphDef)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.framework.GraphDef other) {
+      if (other == org.tensorflow.proto.framework.GraphDef.getDefaultInstance()) return this;
+      if (nodeBuilder_ == null) {
+        if (!other.node_.isEmpty()) {
+          if (node_.isEmpty()) {
+            node_ = other.node_;
+            bitField0_ = (bitField0_ & ~0x00000001);
+          } else {
+            ensureNodeIsMutable();
+            node_.addAll(other.node_);
+          }
+          onChanged();
+        }
+      } else {
+        if (!other.node_.isEmpty()) {
+          if (nodeBuilder_.isEmpty()) {
+            nodeBuilder_.dispose();
+            nodeBuilder_ = null;
+            node_ = other.node_;
+            bitField0_ = (bitField0_ & ~0x00000001);
+            nodeBuilder_ = 
+              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
+                 getNodeFieldBuilder() : null;
+          } else {
+            nodeBuilder_.addAllMessages(other.node_);
+          }
+        }
+      }
+      if (other.hasVersions()) {
+        mergeVersions(other.getVersions());
+      }
+      if (other.getVersion() != 0) {
+        setVersion(other.getVersion());
+      }
+      if (other.hasLibrary()) {
+        mergeLibrary(other.getLibrary());
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.framework.GraphDef parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.framework.GraphDef) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+    private int bitField0_;
+
+    private java.util.List<org.tensorflow.proto.framework.NodeDef> node_ =
+      java.util.Collections.emptyList();
+    private void ensureNodeIsMutable() {
+      if (!((bitField0_ & 0x00000001) != 0)) {
+        node_ = new java.util.ArrayList<org.tensorflow.proto.framework.NodeDef>(node_);
+        bitField0_ |= 0x00000001;
+       }
+    }
+
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.framework.NodeDef, org.tensorflow.proto.framework.NodeDef.Builder, org.tensorflow.proto.framework.NodeDefOrBuilder> nodeBuilder_;
+
+    /**
+     * <code>repeated .tensorflow.NodeDef node = 1;</code>
+     */
+    public java.util.List<org.tensorflow.proto.framework.NodeDef> getNodeList() {
+      if (nodeBuilder_ == null) {
+        return java.util.Collections.unmodifiableList(node_);
+      } else {
+        return nodeBuilder_.getMessageList();
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.NodeDef node = 1;</code>
+     */
+    public int getNodeCount() {
+      if (nodeBuilder_ == null) {
+        return node_.size();
+      } else {
+        return nodeBuilder_.getCount();
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.NodeDef node = 1;</code>
+     */
+    public org.tensorflow.proto.framework.NodeDef getNode(int index) {
+      if (nodeBuilder_ == null) {
+        return node_.get(index);
+      } else {
+        return nodeBuilder_.getMessage(index);
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.NodeDef node = 1;</code>
+     */
+    public Builder setNode(
+        int index, org.tensorflow.proto.framework.NodeDef value) {
+      if (nodeBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureNodeIsMutable();
+        node_.set(index, value);
+        onChanged();
+      } else {
+        nodeBuilder_.setMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.NodeDef node = 1;</code>
+     */
+    public Builder setNode(
+        int index, org.tensorflow.proto.framework.NodeDef.Builder builderForValue) {
+      if (nodeBuilder_ == null) {
+        ensureNodeIsMutable();
+        node_.set(index, builderForValue.build());
+        onChanged();
+      } else {
+        nodeBuilder_.setMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.NodeDef node = 1;</code>
+     */
+    public Builder addNode(org.tensorflow.proto.framework.NodeDef value) {
+      if (nodeBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureNodeIsMutable();
+        node_.add(value);
+        onChanged();
+      } else {
+        nodeBuilder_.addMessage(value);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.NodeDef node = 1;</code>
+     */
+    public Builder addNode(
+        int index, org.tensorflow.proto.framework.NodeDef value) {
+      if (nodeBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureNodeIsMutable();
+        node_.add(index, value);
+        onChanged();
+      } else {
+        nodeBuilder_.addMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.NodeDef node = 1;</code>
+     */
+    public Builder addNode(
+        org.tensorflow.proto.framework.NodeDef.Builder builderForValue) {
+      if (nodeBuilder_ == null) {
+        ensureNodeIsMutable();
+        node_.add(builderForValue.build());
+        onChanged();
+      } else {
+        nodeBuilder_.addMessage(builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.NodeDef node = 1;</code>
+     */
+    public Builder addNode(
+        int index, org.tensorflow.proto.framework.NodeDef.Builder builderForValue) {
+      if (nodeBuilder_ == null) {
+        ensureNodeIsMutable();
+        node_.add(index, builderForValue.build());
+        onChanged();
+      } else {
+        nodeBuilder_.addMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.NodeDef node = 1;</code>
+     */
+    public Builder addAllNode(
+        java.lang.Iterable<? extends org.tensorflow.proto.framework.NodeDef> values) {
+      if (nodeBuilder_ == null) {
+        ensureNodeIsMutable();
+        com.google.protobuf.AbstractMessageLite.Builder.addAll(
+            values, node_);
+        onChanged();
+      } else {
+        nodeBuilder_.addAllMessages(values);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.NodeDef node = 1;</code>
+     */
+    public Builder clearNode() {
+      if (nodeBuilder_ == null) {
+        node_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000001);
+        onChanged();
+      } else {
+        nodeBuilder_.clear();
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.NodeDef node = 1;</code>
+     */
+    public Builder removeNode(int index) {
+      if (nodeBuilder_ == null) {
+        ensureNodeIsMutable();
+        node_.remove(index);
+        onChanged();
+      } else {
+        nodeBuilder_.remove(index);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.NodeDef node = 1;</code>
+     */
+    public org.tensorflow.proto.framework.NodeDef.Builder getNodeBuilder(
+        int index) {
+      return getNodeFieldBuilder().getBuilder(index);
+    }
+    /**
+     * <code>repeated .tensorflow.NodeDef node = 1;</code>
+     */
+    public org.tensorflow.proto.framework.NodeDefOrBuilder getNodeOrBuilder(
+        int index) {
+      if (nodeBuilder_ == null) {
+        return node_.get(index);  } else {
+        return nodeBuilder_.getMessageOrBuilder(index);
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.NodeDef node = 1;</code>
+     */
+    public java.util.List<? extends org.tensorflow.proto.framework.NodeDefOrBuilder> 
+         getNodeOrBuilderList() {
+      if (nodeBuilder_ != null) {
+        return nodeBuilder_.getMessageOrBuilderList();
+      } else {
+        return java.util.Collections.unmodifiableList(node_);
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.NodeDef node = 1;</code>
+     */
+    public org.tensorflow.proto.framework.NodeDef.Builder addNodeBuilder() {
+      return getNodeFieldBuilder().addBuilder(
+          org.tensorflow.proto.framework.NodeDef.getDefaultInstance());
+    }
+    /**
+     * <code>repeated .tensorflow.NodeDef node = 1;</code>
+     */
+    public org.tensorflow.proto.framework.NodeDef.Builder addNodeBuilder(
+        int index) {
+      return getNodeFieldBuilder().addBuilder(
+          index, org.tensorflow.proto.framework.NodeDef.getDefaultInstance());
+    }
+    /**
+     * <code>repeated .tensorflow.NodeDef node = 1;</code>
+     */
+    public java.util.List<org.tensorflow.proto.framework.NodeDef.Builder> 
+         getNodeBuilderList() {
+      return getNodeFieldBuilder().getBuilderList();
+    }
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.framework.NodeDef, org.tensorflow.proto.framework.NodeDef.Builder, org.tensorflow.proto.framework.NodeDefOrBuilder> 
+        getNodeFieldBuilder() {
+      if (nodeBuilder_ == null) {
+        nodeBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
+            org.tensorflow.proto.framework.NodeDef, org.tensorflow.proto.framework.NodeDef.Builder, org.tensorflow.proto.framework.NodeDefOrBuilder>(
+                node_,
+                ((bitField0_ & 0x00000001) != 0),
+                getParentForChildren(),
+                isClean());
+        node_ = null;
+      }
+      return nodeBuilder_;
+    }
+
+    private org.tensorflow.proto.framework.VersionDef versions_;
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.VersionDef, org.tensorflow.proto.framework.VersionDef.Builder, org.tensorflow.proto.framework.VersionDefOrBuilder> versionsBuilder_;
+    /**
+     * <pre>
+     * Compatibility versions of the graph.  See core/public/version.h for version
+     * history.  The GraphDef version is distinct from the TensorFlow version, and
+     * each release of TensorFlow will support a range of GraphDef versions.
+     * </pre>
+     *
+     * <code>.tensorflow.VersionDef versions = 4;</code>
+     */
+    public boolean hasVersions() {
+      return versionsBuilder_ != null || versions_ != null;
+    }
+    /**
+     * <pre>
+     * Compatibility versions of the graph.  See core/public/version.h for version
+     * history.  The GraphDef version is distinct from the TensorFlow version, and
+     * each release of TensorFlow will support a range of GraphDef versions.
+     * </pre>
+     *
+     * <code>.tensorflow.VersionDef versions = 4;</code>
+     */
+    public org.tensorflow.proto.framework.VersionDef getVersions() {
+      if (versionsBuilder_ == null) {
+        return versions_ == null ? org.tensorflow.proto.framework.VersionDef.getDefaultInstance() : versions_;
+      } else {
+        return versionsBuilder_.getMessage();
+      }
+    }
+    /**
+     * <pre>
+     * Compatibility versions of the graph.  See core/public/version.h for version
+     * history.  The GraphDef version is distinct from the TensorFlow version, and
+     * each release of TensorFlow will support a range of GraphDef versions.
+     * </pre>
+     *
+     * <code>.tensorflow.VersionDef versions = 4;</code>
+     */
+    public Builder setVersions(org.tensorflow.proto.framework.VersionDef value) {
+      if (versionsBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        versions_ = value;
+        onChanged();
+      } else {
+        versionsBuilder_.setMessage(value);
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Compatibility versions of the graph.  See core/public/version.h for version
+     * history.  The GraphDef version is distinct from the TensorFlow version, and
+     * each release of TensorFlow will support a range of GraphDef versions.
+     * </pre>
+     *
+     * <code>.tensorflow.VersionDef versions = 4;</code>
+     */
+    public Builder setVersions(
+        org.tensorflow.proto.framework.VersionDef.Builder builderForValue) {
+      if (versionsBuilder_ == null) {
+        versions_ = builderForValue.build();
+        onChanged();
+      } else {
+        versionsBuilder_.setMessage(builderForValue.build());
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Compatibility versions of the graph.  See core/public/version.h for version
+     * history.  The GraphDef version is distinct from the TensorFlow version, and
+     * each release of TensorFlow will support a range of GraphDef versions.
+     * </pre>
+     *
+     * <code>.tensorflow.VersionDef versions = 4;</code>
+     */
+    public Builder mergeVersions(org.tensorflow.proto.framework.VersionDef value) {
+      if (versionsBuilder_ == null) {
+        if (versions_ != null) {
+          versions_ =
+            org.tensorflow.proto.framework.VersionDef.newBuilder(versions_).mergeFrom(value).buildPartial();
+        } else {
+          versions_ = value;
+        }
+        onChanged();
+      } else {
+        versionsBuilder_.mergeFrom(value);
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Compatibility versions of the graph.  See core/public/version.h for version
+     * history.  The GraphDef version is distinct from the TensorFlow version, and
+     * each release of TensorFlow will support a range of GraphDef versions.
+     * </pre>
+     *
+     * <code>.tensorflow.VersionDef versions = 4;</code>
+     */
+    public Builder clearVersions() {
+      if (versionsBuilder_ == null) {
+        versions_ = null;
+        onChanged();
+      } else {
+        versions_ = null;
+        versionsBuilder_ = null;
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Compatibility versions of the graph.  See core/public/version.h for version
+     * history.  The GraphDef version is distinct from the TensorFlow version, and
+     * each release of TensorFlow will support a range of GraphDef versions.
+     * </pre>
+     *
+     * <code>.tensorflow.VersionDef versions = 4;</code>
+     */
+    public org.tensorflow.proto.framework.VersionDef.Builder getVersionsBuilder() {
+      
+      onChanged();
+      return getVersionsFieldBuilder().getBuilder();
+    }
+    /**
+     * <pre>
+     * Compatibility versions of the graph.  See core/public/version.h for version
+     * history.  The GraphDef version is distinct from the TensorFlow version, and
+     * each release of TensorFlow will support a range of GraphDef versions.
+     * </pre>
+     *
+     * <code>.tensorflow.VersionDef versions = 4;</code>
+     */
+    public org.tensorflow.proto.framework.VersionDefOrBuilder getVersionsOrBuilder() {
+      if (versionsBuilder_ != null) {
+        return versionsBuilder_.getMessageOrBuilder();
+      } else {
+        return versions_ == null ?
+            org.tensorflow.proto.framework.VersionDef.getDefaultInstance() : versions_;
+      }
+    }
+    /**
+     * <pre>
+     * Compatibility versions of the graph.  See core/public/version.h for version
+     * history.  The GraphDef version is distinct from the TensorFlow version, and
+     * each release of TensorFlow will support a range of GraphDef versions.
+     * </pre>
+     *
+     * <code>.tensorflow.VersionDef versions = 4;</code>
+     */
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.VersionDef, org.tensorflow.proto.framework.VersionDef.Builder, org.tensorflow.proto.framework.VersionDefOrBuilder> 
+        getVersionsFieldBuilder() {
+      if (versionsBuilder_ == null) {
+        versionsBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+            org.tensorflow.proto.framework.VersionDef, org.tensorflow.proto.framework.VersionDef.Builder, org.tensorflow.proto.framework.VersionDefOrBuilder>(
+                getVersions(),
+                getParentForChildren(),
+                isClean());
+        versions_ = null;
+      }
+      return versionsBuilder_;
+    }
+
+    private int version_ ;
+    /**
+     * <pre>
+     * Deprecated single version field; use versions above instead.  Since all
+     * GraphDef changes before "versions" was introduced were forward
+     * compatible, this field is entirely ignored.
+     * </pre>
+     *
+     * <code>int32 version = 3 [deprecated = true];</code>
+     */
+    @java.lang.Deprecated public int getVersion() {
+      return version_;
+    }
+    /**
+     * <pre>
+     * Deprecated single version field; use versions above instead.  Since all
+     * GraphDef changes before "versions" was introduced were forward
+     * compatible, this field is entirely ignored.
+     * </pre>
+     *
+     * <code>int32 version = 3 [deprecated = true];</code>
+     */
+    @java.lang.Deprecated public Builder setVersion(int value) {
+      
+      version_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Deprecated single version field; use versions above instead.  Since all
+     * GraphDef changes before "versions" was introduced were forward
+     * compatible, this field is entirely ignored.
+     * </pre>
+     *
+     * <code>int32 version = 3 [deprecated = true];</code>
+     */
+    @java.lang.Deprecated public Builder clearVersion() {
+      
+      version_ = 0;
+      onChanged();
+      return this;
+    }
+
+    private org.tensorflow.proto.framework.FunctionDefLibrary library_;
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.FunctionDefLibrary, org.tensorflow.proto.framework.FunctionDefLibrary.Builder, org.tensorflow.proto.framework.FunctionDefLibraryOrBuilder> libraryBuilder_;
+    /**
+     * <pre>
+     * EXPERIMENTAL. DO NOT USE OR DEPEND ON THIS YET.
+     * "library" provides user-defined functions.
+     * Naming:
+     *   * library.function.name are in a flat namespace.
+     *     NOTE: We may need to change it to be hierarchical to support
+     *     different orgs. E.g.,
+     *     { "/google/nn", { ... }},
+     *     { "/google/vision", { ... }}
+     *     { "/org_foo/module_bar", { ... }}
+     *     map&lt;string, FunctionDefLib&gt; named_lib;
+     *   * If node[i].op is the name of one function in "library",
+     *     node[i] is deemed as a function call. Otherwise, node[i].op
+     *     must be a primitive operation supported by the runtime.
+     * Function call semantics:
+     *   * The callee may start execution as soon as some of its inputs
+     *     are ready. The caller may want to use Tuple() mechanism to
+     *     ensure all inputs are ready in the same time.
+     *   * The consumer of return values may start executing as soon as
+     *     the return values the consumer depends on are ready.  The
+     *     consumer may want to use Tuple() mechanism to ensure the
+     *     consumer does not start until all return values of the callee
+     *     function are ready.
+     * </pre>
+     *
+     * <code>.tensorflow.FunctionDefLibrary library = 2;</code>
+     */
+    public boolean hasLibrary() {
+      return libraryBuilder_ != null || library_ != null;
+    }
+    /**
+     * <pre>
+     * EXPERIMENTAL. DO NOT USE OR DEPEND ON THIS YET.
+     * "library" provides user-defined functions.
+     * Naming:
+     *   * library.function.name are in a flat namespace.
+     *     NOTE: We may need to change it to be hierarchical to support
+     *     different orgs. E.g.,
+     *     { "/google/nn", { ... }},
+     *     { "/google/vision", { ... }}
+     *     { "/org_foo/module_bar", { ... }}
+     *     map&lt;string, FunctionDefLib&gt; named_lib;
+     *   * If node[i].op is the name of one function in "library",
+     *     node[i] is deemed as a function call. Otherwise, node[i].op
+     *     must be a primitive operation supported by the runtime.
+     * Function call semantics:
+     *   * The callee may start execution as soon as some of its inputs
+     *     are ready. The caller may want to use Tuple() mechanism to
+     *     ensure all inputs are ready in the same time.
+     *   * The consumer of return values may start executing as soon as
+     *     the return values the consumer depends on are ready.  The
+     *     consumer may want to use Tuple() mechanism to ensure the
+     *     consumer does not start until all return values of the callee
+     *     function are ready.
+     * </pre>
+     *
+     * <code>.tensorflow.FunctionDefLibrary library = 2;</code>
+     */
+    public org.tensorflow.proto.framework.FunctionDefLibrary getLibrary() {
+      if (libraryBuilder_ == null) {
+        return library_ == null ? org.tensorflow.proto.framework.FunctionDefLibrary.getDefaultInstance() : library_;
+      } else {
+        return libraryBuilder_.getMessage();
+      }
+    }
+    /**
+     * <pre>
+     * EXPERIMENTAL. DO NOT USE OR DEPEND ON THIS YET.
+     * "library" provides user-defined functions.
+     * Naming:
+     *   * library.function.name are in a flat namespace.
+     *     NOTE: We may need to change it to be hierarchical to support
+     *     different orgs. E.g.,
+     *     { "/google/nn", { ... }},
+     *     { "/google/vision", { ... }}
+     *     { "/org_foo/module_bar", { ... }}
+     *     map&lt;string, FunctionDefLib&gt; named_lib;
+     *   * If node[i].op is the name of one function in "library",
+     *     node[i] is deemed as a function call. Otherwise, node[i].op
+     *     must be a primitive operation supported by the runtime.
+     * Function call semantics:
+     *   * The callee may start execution as soon as some of its inputs
+     *     are ready. The caller may want to use Tuple() mechanism to
+     *     ensure all inputs are ready in the same time.
+     *   * The consumer of return values may start executing as soon as
+     *     the return values the consumer depends on are ready.  The
+     *     consumer may want to use Tuple() mechanism to ensure the
+     *     consumer does not start until all return values of the callee
+     *     function are ready.
+     * </pre>
+     *
+     * <code>.tensorflow.FunctionDefLibrary library = 2;</code>
+     */
+    public Builder setLibrary(org.tensorflow.proto.framework.FunctionDefLibrary value) {
+      if (libraryBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        library_ = value;
+        onChanged();
+      } else {
+        libraryBuilder_.setMessage(value);
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * EXPERIMENTAL. DO NOT USE OR DEPEND ON THIS YET.
+     * "library" provides user-defined functions.
+     * Naming:
+     *   * library.function.name are in a flat namespace.
+     *     NOTE: We may need to change it to be hierarchical to support
+     *     different orgs. E.g.,
+     *     { "/google/nn", { ... }},
+     *     { "/google/vision", { ... }}
+     *     { "/org_foo/module_bar", { ... }}
+     *     map&lt;string, FunctionDefLib&gt; named_lib;
+     *   * If node[i].op is the name of one function in "library",
+     *     node[i] is deemed as a function call. Otherwise, node[i].op
+     *     must be a primitive operation supported by the runtime.
+     * Function call semantics:
+     *   * The callee may start execution as soon as some of its inputs
+     *     are ready. The caller may want to use Tuple() mechanism to
+     *     ensure all inputs are ready in the same time.
+     *   * The consumer of return values may start executing as soon as
+     *     the return values the consumer depends on are ready.  The
+     *     consumer may want to use Tuple() mechanism to ensure the
+     *     consumer does not start until all return values of the callee
+     *     function are ready.
+     * </pre>
+     *
+     * <code>.tensorflow.FunctionDefLibrary library = 2;</code>
+     */
+    public Builder setLibrary(
+        org.tensorflow.proto.framework.FunctionDefLibrary.Builder builderForValue) {
+      if (libraryBuilder_ == null) {
+        library_ = builderForValue.build();
+        onChanged();
+      } else {
+        libraryBuilder_.setMessage(builderForValue.build());
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * EXPERIMENTAL. DO NOT USE OR DEPEND ON THIS YET.
+     * "library" provides user-defined functions.
+     * Naming:
+     *   * library.function.name are in a flat namespace.
+     *     NOTE: We may need to change it to be hierarchical to support
+     *     different orgs. E.g.,
+     *     { "/google/nn", { ... }},
+     *     { "/google/vision", { ... }}
+     *     { "/org_foo/module_bar", { ... }}
+     *     map&lt;string, FunctionDefLib&gt; named_lib;
+     *   * If node[i].op is the name of one function in "library",
+     *     node[i] is deemed as a function call. Otherwise, node[i].op
+     *     must be a primitive operation supported by the runtime.
+     * Function call semantics:
+     *   * The callee may start execution as soon as some of its inputs
+     *     are ready. The caller may want to use Tuple() mechanism to
+     *     ensure all inputs are ready in the same time.
+     *   * The consumer of return values may start executing as soon as
+     *     the return values the consumer depends on are ready.  The
+     *     consumer may want to use Tuple() mechanism to ensure the
+     *     consumer does not start until all return values of the callee
+     *     function are ready.
+     * </pre>
+     *
+     * <code>.tensorflow.FunctionDefLibrary library = 2;</code>
+     */
+    public Builder mergeLibrary(org.tensorflow.proto.framework.FunctionDefLibrary value) {
+      if (libraryBuilder_ == null) {
+        if (library_ != null) {
+          library_ =
+            org.tensorflow.proto.framework.FunctionDefLibrary.newBuilder(library_).mergeFrom(value).buildPartial();
+        } else {
+          library_ = value;
+        }
+        onChanged();
+      } else {
+        libraryBuilder_.mergeFrom(value);
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * EXPERIMENTAL. DO NOT USE OR DEPEND ON THIS YET.
+     * "library" provides user-defined functions.
+     * Naming:
+     *   * library.function.name are in a flat namespace.
+     *     NOTE: We may need to change it to be hierarchical to support
+     *     different orgs. E.g.,
+     *     { "/google/nn", { ... }},
+     *     { "/google/vision", { ... }}
+     *     { "/org_foo/module_bar", { ... }}
+     *     map&lt;string, FunctionDefLib&gt; named_lib;
+     *   * If node[i].op is the name of one function in "library",
+     *     node[i] is deemed as a function call. Otherwise, node[i].op
+     *     must be a primitive operation supported by the runtime.
+     * Function call semantics:
+     *   * The callee may start execution as soon as some of its inputs
+     *     are ready. The caller may want to use Tuple() mechanism to
+     *     ensure all inputs are ready in the same time.
+     *   * The consumer of return values may start executing as soon as
+     *     the return values the consumer depends on are ready.  The
+     *     consumer may want to use Tuple() mechanism to ensure the
+     *     consumer does not start until all return values of the callee
+     *     function are ready.
+     * </pre>
+     *
+     * <code>.tensorflow.FunctionDefLibrary library = 2;</code>
+     */
+    public Builder clearLibrary() {
+      if (libraryBuilder_ == null) {
+        library_ = null;
+        onChanged();
+      } else {
+        library_ = null;
+        libraryBuilder_ = null;
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * EXPERIMENTAL. DO NOT USE OR DEPEND ON THIS YET.
+     * "library" provides user-defined functions.
+     * Naming:
+     *   * library.function.name are in a flat namespace.
+     *     NOTE: We may need to change it to be hierarchical to support
+     *     different orgs. E.g.,
+     *     { "/google/nn", { ... }},
+     *     { "/google/vision", { ... }}
+     *     { "/org_foo/module_bar", { ... }}
+     *     map&lt;string, FunctionDefLib&gt; named_lib;
+     *   * If node[i].op is the name of one function in "library",
+     *     node[i] is deemed as a function call. Otherwise, node[i].op
+     *     must be a primitive operation supported by the runtime.
+     * Function call semantics:
+     *   * The callee may start execution as soon as some of its inputs
+     *     are ready. The caller may want to use Tuple() mechanism to
+     *     ensure all inputs are ready in the same time.
+     *   * The consumer of return values may start executing as soon as
+     *     the return values the consumer depends on are ready.  The
+     *     consumer may want to use Tuple() mechanism to ensure the
+     *     consumer does not start until all return values of the callee
+     *     function are ready.
+     * </pre>
+     *
+     * <code>.tensorflow.FunctionDefLibrary library = 2;</code>
+     */
+    public org.tensorflow.proto.framework.FunctionDefLibrary.Builder getLibraryBuilder() {
+      
+      onChanged();
+      return getLibraryFieldBuilder().getBuilder();
+    }
+    /**
+     * <pre>
+     * EXPERIMENTAL. DO NOT USE OR DEPEND ON THIS YET.
+     * "library" provides user-defined functions.
+     * Naming:
+     *   * library.function.name are in a flat namespace.
+     *     NOTE: We may need to change it to be hierarchical to support
+     *     different orgs. E.g.,
+     *     { "/google/nn", { ... }},
+     *     { "/google/vision", { ... }}
+     *     { "/org_foo/module_bar", { ... }}
+     *     map&lt;string, FunctionDefLib&gt; named_lib;
+     *   * If node[i].op is the name of one function in "library",
+     *     node[i] is deemed as a function call. Otherwise, node[i].op
+     *     must be a primitive operation supported by the runtime.
+     * Function call semantics:
+     *   * The callee may start execution as soon as some of its inputs
+     *     are ready. The caller may want to use Tuple() mechanism to
+     *     ensure all inputs are ready in the same time.
+     *   * The consumer of return values may start executing as soon as
+     *     the return values the consumer depends on are ready.  The
+     *     consumer may want to use Tuple() mechanism to ensure the
+     *     consumer does not start until all return values of the callee
+     *     function are ready.
+     * </pre>
+     *
+     * <code>.tensorflow.FunctionDefLibrary library = 2;</code>
+     */
+    public org.tensorflow.proto.framework.FunctionDefLibraryOrBuilder getLibraryOrBuilder() {
+      if (libraryBuilder_ != null) {
+        return libraryBuilder_.getMessageOrBuilder();
+      } else {
+        return library_ == null ?
+            org.tensorflow.proto.framework.FunctionDefLibrary.getDefaultInstance() : library_;
+      }
+    }
+    /**
+     * <pre>
+     * EXPERIMENTAL. DO NOT USE OR DEPEND ON THIS YET.
+     * "library" provides user-defined functions.
+     * Naming:
+     *   * library.function.name are in a flat namespace.
+     *     NOTE: We may need to change it to be hierarchical to support
+     *     different orgs. E.g.,
+     *     { "/google/nn", { ... }},
+     *     { "/google/vision", { ... }}
+     *     { "/org_foo/module_bar", { ... }}
+     *     map&lt;string, FunctionDefLib&gt; named_lib;
+     *   * If node[i].op is the name of one function in "library",
+     *     node[i] is deemed as a function call. Otherwise, node[i].op
+     *     must be a primitive operation supported by the runtime.
+     * Function call semantics:
+     *   * The callee may start execution as soon as some of its inputs
+     *     are ready. The caller may want to use Tuple() mechanism to
+     *     ensure all inputs are ready in the same time.
+     *   * The consumer of return values may start executing as soon as
+     *     the return values the consumer depends on are ready.  The
+     *     consumer may want to use Tuple() mechanism to ensure the
+     *     consumer does not start until all return values of the callee
+     *     function are ready.
+     * </pre>
+     *
+     * <code>.tensorflow.FunctionDefLibrary library = 2;</code>
+     */
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.FunctionDefLibrary, org.tensorflow.proto.framework.FunctionDefLibrary.Builder, org.tensorflow.proto.framework.FunctionDefLibraryOrBuilder> 
+        getLibraryFieldBuilder() {
+      if (libraryBuilder_ == null) {
+        libraryBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+            org.tensorflow.proto.framework.FunctionDefLibrary, org.tensorflow.proto.framework.FunctionDefLibrary.Builder, org.tensorflow.proto.framework.FunctionDefLibraryOrBuilder>(
+                getLibrary(),
+                getParentForChildren(),
+                isClean());
+        library_ = null;
+      }
+      return libraryBuilder_;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.GraphDef)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.GraphDef)
+  private static final org.tensorflow.proto.framework.GraphDef DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.GraphDef();
+  }
+
+  public static org.tensorflow.proto.framework.GraphDef getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<GraphDef>
+      PARSER = new com.google.protobuf.AbstractParser<GraphDef>() {
+    @java.lang.Override
+    public GraphDef parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new GraphDef(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<GraphDef> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<GraphDef> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.framework.GraphDef getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphDefOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphDefOrBuilder.java
new file mode 100644
index 00000000000..0e4ac4dca6e
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphDefOrBuilder.java
@@ -0,0 +1,163 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/framework/graph.proto
+
+package org.tensorflow.proto.framework;
+
+public interface GraphDefOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.GraphDef)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <code>repeated .tensorflow.NodeDef node = 1;</code>
+   */
+  java.util.List<org.tensorflow.proto.framework.NodeDef> 
+      getNodeList();
+  /**
+   * <code>repeated .tensorflow.NodeDef node = 1;</code>
+   */
+  org.tensorflow.proto.framework.NodeDef getNode(int index);
+  /**
+   * <code>repeated .tensorflow.NodeDef node = 1;</code>
+   */
+  int getNodeCount();
+  /**
+   * <code>repeated .tensorflow.NodeDef node = 1;</code>
+   */
+  java.util.List<? extends org.tensorflow.proto.framework.NodeDefOrBuilder> 
+      getNodeOrBuilderList();
+  /**
+   * <code>repeated .tensorflow.NodeDef node = 1;</code>
+   */
+  org.tensorflow.proto.framework.NodeDefOrBuilder getNodeOrBuilder(
+      int index);
+
+  /**
+   * <pre>
+   * Compatibility versions of the graph.  See core/public/version.h for version
+   * history.  The GraphDef version is distinct from the TensorFlow version, and
+   * each release of TensorFlow will support a range of GraphDef versions.
+   * </pre>
+   *
+   * <code>.tensorflow.VersionDef versions = 4;</code>
+   */
+  boolean hasVersions();
+  /**
+   * <pre>
+   * Compatibility versions of the graph.  See core/public/version.h for version
+   * history.  The GraphDef version is distinct from the TensorFlow version, and
+   * each release of TensorFlow will support a range of GraphDef versions.
+   * </pre>
+   *
+   * <code>.tensorflow.VersionDef versions = 4;</code>
+   */
+  org.tensorflow.proto.framework.VersionDef getVersions();
+  /**
+   * <pre>
+   * Compatibility versions of the graph.  See core/public/version.h for version
+   * history.  The GraphDef version is distinct from the TensorFlow version, and
+   * each release of TensorFlow will support a range of GraphDef versions.
+   * </pre>
+   *
+   * <code>.tensorflow.VersionDef versions = 4;</code>
+   */
+  org.tensorflow.proto.framework.VersionDefOrBuilder getVersionsOrBuilder();
+
+  /**
+   * <pre>
+   * Deprecated single version field; use versions above instead.  Since all
+   * GraphDef changes before "versions" was introduced were forward
+   * compatible, this field is entirely ignored.
+   * </pre>
+   *
+   * <code>int32 version = 3 [deprecated = true];</code>
+   */
+  @java.lang.Deprecated int getVersion();
+
+  /**
+   * <pre>
+   * EXPERIMENTAL. DO NOT USE OR DEPEND ON THIS YET.
+   * "library" provides user-defined functions.
+   * Naming:
+   *   * library.function.name are in a flat namespace.
+   *     NOTE: We may need to change it to be hierarchical to support
+   *     different orgs. E.g.,
+   *     { "/google/nn", { ... }},
+   *     { "/google/vision", { ... }}
+   *     { "/org_foo/module_bar", { ... }}
+   *     map&lt;string, FunctionDefLib&gt; named_lib;
+   *   * If node[i].op is the name of one function in "library",
+   *     node[i] is deemed as a function call. Otherwise, node[i].op
+   *     must be a primitive operation supported by the runtime.
+   * Function call semantics:
+   *   * The callee may start execution as soon as some of its inputs
+   *     are ready. The caller may want to use Tuple() mechanism to
+   *     ensure all inputs are ready in the same time.
+   *   * The consumer of return values may start executing as soon as
+   *     the return values the consumer depends on are ready.  The
+   *     consumer may want to use Tuple() mechanism to ensure the
+   *     consumer does not start until all return values of the callee
+   *     function are ready.
+   * </pre>
+   *
+   * <code>.tensorflow.FunctionDefLibrary library = 2;</code>
+   */
+  boolean hasLibrary();
+  /**
+   * <pre>
+   * EXPERIMENTAL. DO NOT USE OR DEPEND ON THIS YET.
+   * "library" provides user-defined functions.
+   * Naming:
+   *   * library.function.name are in a flat namespace.
+   *     NOTE: We may need to change it to be hierarchical to support
+   *     different orgs. E.g.,
+   *     { "/google/nn", { ... }},
+   *     { "/google/vision", { ... }}
+   *     { "/org_foo/module_bar", { ... }}
+   *     map&lt;string, FunctionDefLib&gt; named_lib;
+   *   * If node[i].op is the name of one function in "library",
+   *     node[i] is deemed as a function call. Otherwise, node[i].op
+   *     must be a primitive operation supported by the runtime.
+   * Function call semantics:
+   *   * The callee may start execution as soon as some of its inputs
+   *     are ready. The caller may want to use Tuple() mechanism to
+   *     ensure all inputs are ready in the same time.
+   *   * The consumer of return values may start executing as soon as
+   *     the return values the consumer depends on are ready.  The
+   *     consumer may want to use Tuple() mechanism to ensure the
+   *     consumer does not start until all return values of the callee
+   *     function are ready.
+   * </pre>
+   *
+   * <code>.tensorflow.FunctionDefLibrary library = 2;</code>
+   */
+  org.tensorflow.proto.framework.FunctionDefLibrary getLibrary();
+  /**
+   * <pre>
+   * EXPERIMENTAL. DO NOT USE OR DEPEND ON THIS YET.
+   * "library" provides user-defined functions.
+   * Naming:
+   *   * library.function.name are in a flat namespace.
+   *     NOTE: We may need to change it to be hierarchical to support
+   *     different orgs. E.g.,
+   *     { "/google/nn", { ... }},
+   *     { "/google/vision", { ... }}
+   *     { "/org_foo/module_bar", { ... }}
+   *     map&lt;string, FunctionDefLib&gt; named_lib;
+   *   * If node[i].op is the name of one function in "library",
+   *     node[i] is deemed as a function call. Otherwise, node[i].op
+   *     must be a primitive operation supported by the runtime.
+   * Function call semantics:
+   *   * The callee may start execution as soon as some of its inputs
+   *     are ready. The caller may want to use Tuple() mechanism to
+   *     ensure all inputs are ready in the same time.
+   *   * The consumer of return values may start executing as soon as
+   *     the return values the consumer depends on are ready.  The
+   *     consumer may want to use Tuple() mechanism to ensure the
+   *     consumer does not start until all return values of the callee
+   *     function are ready.
+   * </pre>
+   *
+   * <code>.tensorflow.FunctionDefLibrary library = 2;</code>
+   */
+  org.tensorflow.proto.framework.FunctionDefLibraryOrBuilder getLibraryOrBuilder();
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphOptions.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphOptions.java
new file mode 100644
index 00000000000..5916d82ca1c
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphOptions.java
@@ -0,0 +1,1462 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/config.proto
+
+package org.tensorflow.proto.framework;
+
+/**
+ * Protobuf type {@code tensorflow.GraphOptions}
+ */
+public  final class GraphOptions extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.GraphOptions)
+    GraphOptionsOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use GraphOptions.newBuilder() to construct.
+  private GraphOptions(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private GraphOptions() {
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new GraphOptions();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private GraphOptions(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 16: {
+
+            enableRecvScheduling_ = input.readBool();
+            break;
+          }
+          case 26: {
+            org.tensorflow.proto.framework.OptimizerOptions.Builder subBuilder = null;
+            if (optimizerOptions_ != null) {
+              subBuilder = optimizerOptions_.toBuilder();
+            }
+            optimizerOptions_ = input.readMessage(org.tensorflow.proto.framework.OptimizerOptions.parser(), extensionRegistry);
+            if (subBuilder != null) {
+              subBuilder.mergeFrom(optimizerOptions_);
+              optimizerOptions_ = subBuilder.buildPartial();
+            }
+
+            break;
+          }
+          case 32: {
+
+            buildCostModel_ = input.readInt64();
+            break;
+          }
+          case 40: {
+
+            inferShapes_ = input.readBool();
+            break;
+          }
+          case 48: {
+
+            placePrunedGraph_ = input.readBool();
+            break;
+          }
+          case 56: {
+
+            enableBfloat16Sendrecv_ = input.readBool();
+            break;
+          }
+          case 64: {
+
+            timelineStep_ = input.readInt32();
+            break;
+          }
+          case 72: {
+
+            buildCostModelAfter_ = input.readInt64();
+            break;
+          }
+          case 82: {
+            org.tensorflow.proto.framework.RewriterConfig.Builder subBuilder = null;
+            if (rewriteOptions_ != null) {
+              subBuilder = rewriteOptions_.toBuilder();
+            }
+            rewriteOptions_ = input.readMessage(org.tensorflow.proto.framework.RewriterConfig.parser(), extensionRegistry);
+            if (subBuilder != null) {
+              subBuilder.mergeFrom(rewriteOptions_);
+              rewriteOptions_ = subBuilder.buildPartial();
+            }
+
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_GraphOptions_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_GraphOptions_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.framework.GraphOptions.class, org.tensorflow.proto.framework.GraphOptions.Builder.class);
+  }
+
+  public static final int ENABLE_RECV_SCHEDULING_FIELD_NUMBER = 2;
+  private boolean enableRecvScheduling_;
+  /**
+   * <pre>
+   * If true, use control flow to schedule the activation of Recv nodes.
+   * (Currently ignored.)
+   * </pre>
+   *
+   * <code>bool enable_recv_scheduling = 2;</code>
+   */
+  public boolean getEnableRecvScheduling() {
+    return enableRecvScheduling_;
+  }
+
+  public static final int OPTIMIZER_OPTIONS_FIELD_NUMBER = 3;
+  private org.tensorflow.proto.framework.OptimizerOptions optimizerOptions_;
+  /**
+   * <pre>
+   * Options controlling how graph is optimized.
+   * </pre>
+   *
+   * <code>.tensorflow.OptimizerOptions optimizer_options = 3;</code>
+   */
+  public boolean hasOptimizerOptions() {
+    return optimizerOptions_ != null;
+  }
+  /**
+   * <pre>
+   * Options controlling how graph is optimized.
+   * </pre>
+   *
+   * <code>.tensorflow.OptimizerOptions optimizer_options = 3;</code>
+   */
+  public org.tensorflow.proto.framework.OptimizerOptions getOptimizerOptions() {
+    return optimizerOptions_ == null ? org.tensorflow.proto.framework.OptimizerOptions.getDefaultInstance() : optimizerOptions_;
+  }
+  /**
+   * <pre>
+   * Options controlling how graph is optimized.
+   * </pre>
+   *
+   * <code>.tensorflow.OptimizerOptions optimizer_options = 3;</code>
+   */
+  public org.tensorflow.proto.framework.OptimizerOptionsOrBuilder getOptimizerOptionsOrBuilder() {
+    return getOptimizerOptions();
+  }
+
+  public static final int BUILD_COST_MODEL_FIELD_NUMBER = 4;
+  private long buildCostModel_;
+  /**
+   * <pre>
+   * The number of steps to run before returning a cost model detailing
+   * the memory usage and performance of each node of the graph. 0 means
+   * no cost model.
+   * </pre>
+   *
+   * <code>int64 build_cost_model = 4;</code>
+   */
+  public long getBuildCostModel() {
+    return buildCostModel_;
+  }
+
+  public static final int BUILD_COST_MODEL_AFTER_FIELD_NUMBER = 9;
+  private long buildCostModelAfter_;
+  /**
+   * <pre>
+   * The number of steps to skip before collecting statistics for the
+   * cost model.
+   * </pre>
+   *
+   * <code>int64 build_cost_model_after = 9;</code>
+   */
+  public long getBuildCostModelAfter() {
+    return buildCostModelAfter_;
+  }
+
+  public static final int INFER_SHAPES_FIELD_NUMBER = 5;
+  private boolean inferShapes_;
+  /**
+   * <pre>
+   * Annotate each Node with Op output shape data, to the extent it can
+   * be statically inferred.
+   * </pre>
+   *
+   * <code>bool infer_shapes = 5;</code>
+   */
+  public boolean getInferShapes() {
+    return inferShapes_;
+  }
+
+  public static final int PLACE_PRUNED_GRAPH_FIELD_NUMBER = 6;
+  private boolean placePrunedGraph_;
+  /**
+   * <pre>
+   * Only place the subgraphs that are run, rather than the entire graph.
+   * This is useful for interactive graph building, where one might
+   * produce graphs that cannot be placed during the debugging
+   * process.  In particular, it allows the client to continue work in
+   * a session after adding a node to a graph whose placement
+   * constraints are unsatisfiable.
+   * </pre>
+   *
+   * <code>bool place_pruned_graph = 6;</code>
+   */
+  public boolean getPlacePrunedGraph() {
+    return placePrunedGraph_;
+  }
+
+  public static final int ENABLE_BFLOAT16_SENDRECV_FIELD_NUMBER = 7;
+  private boolean enableBfloat16Sendrecv_;
+  /**
+   * <pre>
+   * If true, transfer float values between processes as bfloat16.
+   * </pre>
+   *
+   * <code>bool enable_bfloat16_sendrecv = 7;</code>
+   */
+  public boolean getEnableBfloat16Sendrecv() {
+    return enableBfloat16Sendrecv_;
+  }
+
+  public static final int TIMELINE_STEP_FIELD_NUMBER = 8;
+  private int timelineStep_;
+  /**
+   * <pre>
+   * If &gt; 0, record a timeline every this many steps.
+   * EXPERIMENTAL: This currently has no effect in MasterSession.
+   * </pre>
+   *
+   * <code>int32 timeline_step = 8;</code>
+   */
+  public int getTimelineStep() {
+    return timelineStep_;
+  }
+
+  public static final int REWRITE_OPTIONS_FIELD_NUMBER = 10;
+  private org.tensorflow.proto.framework.RewriterConfig rewriteOptions_;
+  /**
+   * <pre>
+   * Options that control the type and amount of graph rewriting.
+   * Not currently configurable via the public Python API (i.e. there is no API
+   * stability guarantee if you import RewriterConfig explicitly).
+   * </pre>
+   *
+   * <code>.tensorflow.RewriterConfig rewrite_options = 10;</code>
+   */
+  public boolean hasRewriteOptions() {
+    return rewriteOptions_ != null;
+  }
+  /**
+   * <pre>
+   * Options that control the type and amount of graph rewriting.
+   * Not currently configurable via the public Python API (i.e. there is no API
+   * stability guarantee if you import RewriterConfig explicitly).
+   * </pre>
+   *
+   * <code>.tensorflow.RewriterConfig rewrite_options = 10;</code>
+   */
+  public org.tensorflow.proto.framework.RewriterConfig getRewriteOptions() {
+    return rewriteOptions_ == null ? org.tensorflow.proto.framework.RewriterConfig.getDefaultInstance() : rewriteOptions_;
+  }
+  /**
+   * <pre>
+   * Options that control the type and amount of graph rewriting.
+   * Not currently configurable via the public Python API (i.e. there is no API
+   * stability guarantee if you import RewriterConfig explicitly).
+   * </pre>
+   *
+   * <code>.tensorflow.RewriterConfig rewrite_options = 10;</code>
+   */
+  public org.tensorflow.proto.framework.RewriterConfigOrBuilder getRewriteOptionsOrBuilder() {
+    return getRewriteOptions();
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    if (enableRecvScheduling_ != false) {
+      output.writeBool(2, enableRecvScheduling_);
+    }
+    if (optimizerOptions_ != null) {
+      output.writeMessage(3, getOptimizerOptions());
+    }
+    if (buildCostModel_ != 0L) {
+      output.writeInt64(4, buildCostModel_);
+    }
+    if (inferShapes_ != false) {
+      output.writeBool(5, inferShapes_);
+    }
+    if (placePrunedGraph_ != false) {
+      output.writeBool(6, placePrunedGraph_);
+    }
+    if (enableBfloat16Sendrecv_ != false) {
+      output.writeBool(7, enableBfloat16Sendrecv_);
+    }
+    if (timelineStep_ != 0) {
+      output.writeInt32(8, timelineStep_);
+    }
+    if (buildCostModelAfter_ != 0L) {
+      output.writeInt64(9, buildCostModelAfter_);
+    }
+    if (rewriteOptions_ != null) {
+      output.writeMessage(10, getRewriteOptions());
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (enableRecvScheduling_ != false) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeBoolSize(2, enableRecvScheduling_);
+    }
+    if (optimizerOptions_ != null) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(3, getOptimizerOptions());
+    }
+    if (buildCostModel_ != 0L) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt64Size(4, buildCostModel_);
+    }
+    if (inferShapes_ != false) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeBoolSize(5, inferShapes_);
+    }
+    if (placePrunedGraph_ != false) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeBoolSize(6, placePrunedGraph_);
+    }
+    if (enableBfloat16Sendrecv_ != false) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeBoolSize(7, enableBfloat16Sendrecv_);
+    }
+    if (timelineStep_ != 0) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt32Size(8, timelineStep_);
+    }
+    if (buildCostModelAfter_ != 0L) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt64Size(9, buildCostModelAfter_);
+    }
+    if (rewriteOptions_ != null) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(10, getRewriteOptions());
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.framework.GraphOptions)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.framework.GraphOptions other = (org.tensorflow.proto.framework.GraphOptions) obj;
+
+    if (getEnableRecvScheduling()
+        != other.getEnableRecvScheduling()) return false;
+    if (hasOptimizerOptions() != other.hasOptimizerOptions()) return false;
+    if (hasOptimizerOptions()) {
+      if (!getOptimizerOptions()
+          .equals(other.getOptimizerOptions())) return false;
+    }
+    if (getBuildCostModel()
+        != other.getBuildCostModel()) return false;
+    if (getBuildCostModelAfter()
+        != other.getBuildCostModelAfter()) return false;
+    if (getInferShapes()
+        != other.getInferShapes()) return false;
+    if (getPlacePrunedGraph()
+        != other.getPlacePrunedGraph()) return false;
+    if (getEnableBfloat16Sendrecv()
+        != other.getEnableBfloat16Sendrecv()) return false;
+    if (getTimelineStep()
+        != other.getTimelineStep()) return false;
+    if (hasRewriteOptions() != other.hasRewriteOptions()) return false;
+    if (hasRewriteOptions()) {
+      if (!getRewriteOptions()
+          .equals(other.getRewriteOptions())) return false;
+    }
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    hash = (37 * hash) + ENABLE_RECV_SCHEDULING_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashBoolean(
+        getEnableRecvScheduling());
+    if (hasOptimizerOptions()) {
+      hash = (37 * hash) + OPTIMIZER_OPTIONS_FIELD_NUMBER;
+      hash = (53 * hash) + getOptimizerOptions().hashCode();
+    }
+    hash = (37 * hash) + BUILD_COST_MODEL_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+        getBuildCostModel());
+    hash = (37 * hash) + BUILD_COST_MODEL_AFTER_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+        getBuildCostModelAfter());
+    hash = (37 * hash) + INFER_SHAPES_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashBoolean(
+        getInferShapes());
+    hash = (37 * hash) + PLACE_PRUNED_GRAPH_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashBoolean(
+        getPlacePrunedGraph());
+    hash = (37 * hash) + ENABLE_BFLOAT16_SENDRECV_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashBoolean(
+        getEnableBfloat16Sendrecv());
+    hash = (37 * hash) + TIMELINE_STEP_FIELD_NUMBER;
+    hash = (53 * hash) + getTimelineStep();
+    if (hasRewriteOptions()) {
+      hash = (37 * hash) + REWRITE_OPTIONS_FIELD_NUMBER;
+      hash = (53 * hash) + getRewriteOptions().hashCode();
+    }
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.framework.GraphOptions parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.GraphOptions parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.GraphOptions parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.GraphOptions parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.GraphOptions parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.GraphOptions parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.GraphOptions parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.GraphOptions parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.GraphOptions parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.GraphOptions parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.GraphOptions parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.GraphOptions parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.framework.GraphOptions prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * Protobuf type {@code tensorflow.GraphOptions}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.GraphOptions)
+      org.tensorflow.proto.framework.GraphOptionsOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_GraphOptions_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_GraphOptions_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.GraphOptions.class, org.tensorflow.proto.framework.GraphOptions.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.framework.GraphOptions.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      enableRecvScheduling_ = false;
+
+      if (optimizerOptionsBuilder_ == null) {
+        optimizerOptions_ = null;
+      } else {
+        optimizerOptions_ = null;
+        optimizerOptionsBuilder_ = null;
+      }
+      buildCostModel_ = 0L;
+
+      buildCostModelAfter_ = 0L;
+
+      inferShapes_ = false;
+
+      placePrunedGraph_ = false;
+
+      enableBfloat16Sendrecv_ = false;
+
+      timelineStep_ = 0;
+
+      if (rewriteOptionsBuilder_ == null) {
+        rewriteOptions_ = null;
+      } else {
+        rewriteOptions_ = null;
+        rewriteOptionsBuilder_ = null;
+      }
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_GraphOptions_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.GraphOptions getDefaultInstanceForType() {
+      return org.tensorflow.proto.framework.GraphOptions.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.GraphOptions build() {
+      org.tensorflow.proto.framework.GraphOptions result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.GraphOptions buildPartial() {
+      org.tensorflow.proto.framework.GraphOptions result = new org.tensorflow.proto.framework.GraphOptions(this);
+      result.enableRecvScheduling_ = enableRecvScheduling_;
+      if (optimizerOptionsBuilder_ == null) {
+        result.optimizerOptions_ = optimizerOptions_;
+      } else {
+        result.optimizerOptions_ = optimizerOptionsBuilder_.build();
+      }
+      result.buildCostModel_ = buildCostModel_;
+      result.buildCostModelAfter_ = buildCostModelAfter_;
+      result.inferShapes_ = inferShapes_;
+      result.placePrunedGraph_ = placePrunedGraph_;
+      result.enableBfloat16Sendrecv_ = enableBfloat16Sendrecv_;
+      result.timelineStep_ = timelineStep_;
+      if (rewriteOptionsBuilder_ == null) {
+        result.rewriteOptions_ = rewriteOptions_;
+      } else {
+        result.rewriteOptions_ = rewriteOptionsBuilder_.build();
+      }
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.framework.GraphOptions) {
+        return mergeFrom((org.tensorflow.proto.framework.GraphOptions)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.framework.GraphOptions other) {
+      if (other == org.tensorflow.proto.framework.GraphOptions.getDefaultInstance()) return this;
+      if (other.getEnableRecvScheduling() != false) {
+        setEnableRecvScheduling(other.getEnableRecvScheduling());
+      }
+      if (other.hasOptimizerOptions()) {
+        mergeOptimizerOptions(other.getOptimizerOptions());
+      }
+      if (other.getBuildCostModel() != 0L) {
+        setBuildCostModel(other.getBuildCostModel());
+      }
+      if (other.getBuildCostModelAfter() != 0L) {
+        setBuildCostModelAfter(other.getBuildCostModelAfter());
+      }
+      if (other.getInferShapes() != false) {
+        setInferShapes(other.getInferShapes());
+      }
+      if (other.getPlacePrunedGraph() != false) {
+        setPlacePrunedGraph(other.getPlacePrunedGraph());
+      }
+      if (other.getEnableBfloat16Sendrecv() != false) {
+        setEnableBfloat16Sendrecv(other.getEnableBfloat16Sendrecv());
+      }
+      if (other.getTimelineStep() != 0) {
+        setTimelineStep(other.getTimelineStep());
+      }
+      if (other.hasRewriteOptions()) {
+        mergeRewriteOptions(other.getRewriteOptions());
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.framework.GraphOptions parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.framework.GraphOptions) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+
+    private boolean enableRecvScheduling_ ;
+    /**
+     * <pre>
+     * If true, use control flow to schedule the activation of Recv nodes.
+     * (Currently ignored.)
+     * </pre>
+     *
+     * <code>bool enable_recv_scheduling = 2;</code>
+     */
+    public boolean getEnableRecvScheduling() {
+      return enableRecvScheduling_;
+    }
+    /**
+     * <pre>
+     * If true, use control flow to schedule the activation of Recv nodes.
+     * (Currently ignored.)
+     * </pre>
+     *
+     * <code>bool enable_recv_scheduling = 2;</code>
+     */
+    public Builder setEnableRecvScheduling(boolean value) {
+      
+      enableRecvScheduling_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * If true, use control flow to schedule the activation of Recv nodes.
+     * (Currently ignored.)
+     * </pre>
+     *
+     * <code>bool enable_recv_scheduling = 2;</code>
+     */
+    public Builder clearEnableRecvScheduling() {
+      
+      enableRecvScheduling_ = false;
+      onChanged();
+      return this;
+    }
+
+    private org.tensorflow.proto.framework.OptimizerOptions optimizerOptions_;
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.OptimizerOptions, org.tensorflow.proto.framework.OptimizerOptions.Builder, org.tensorflow.proto.framework.OptimizerOptionsOrBuilder> optimizerOptionsBuilder_;
+    /**
+     * <pre>
+     * Options controlling how graph is optimized.
+     * </pre>
+     *
+     * <code>.tensorflow.OptimizerOptions optimizer_options = 3;</code>
+     */
+    public boolean hasOptimizerOptions() {
+      return optimizerOptionsBuilder_ != null || optimizerOptions_ != null;
+    }
+    /**
+     * <pre>
+     * Options controlling how graph is optimized.
+     * </pre>
+     *
+     * <code>.tensorflow.OptimizerOptions optimizer_options = 3;</code>
+     */
+    public org.tensorflow.proto.framework.OptimizerOptions getOptimizerOptions() {
+      if (optimizerOptionsBuilder_ == null) {
+        return optimizerOptions_ == null ? org.tensorflow.proto.framework.OptimizerOptions.getDefaultInstance() : optimizerOptions_;
+      } else {
+        return optimizerOptionsBuilder_.getMessage();
+      }
+    }
+    /**
+     * <pre>
+     * Options controlling how graph is optimized.
+     * </pre>
+     *
+     * <code>.tensorflow.OptimizerOptions optimizer_options = 3;</code>
+     */
+    public Builder setOptimizerOptions(org.tensorflow.proto.framework.OptimizerOptions value) {
+      if (optimizerOptionsBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        optimizerOptions_ = value;
+        onChanged();
+      } else {
+        optimizerOptionsBuilder_.setMessage(value);
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Options controlling how graph is optimized.
+     * </pre>
+     *
+     * <code>.tensorflow.OptimizerOptions optimizer_options = 3;</code>
+     */
+    public Builder setOptimizerOptions(
+        org.tensorflow.proto.framework.OptimizerOptions.Builder builderForValue) {
+      if (optimizerOptionsBuilder_ == null) {
+        optimizerOptions_ = builderForValue.build();
+        onChanged();
+      } else {
+        optimizerOptionsBuilder_.setMessage(builderForValue.build());
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Options controlling how graph is optimized.
+     * </pre>
+     *
+     * <code>.tensorflow.OptimizerOptions optimizer_options = 3;</code>
+     */
+    public Builder mergeOptimizerOptions(org.tensorflow.proto.framework.OptimizerOptions value) {
+      if (optimizerOptionsBuilder_ == null) {
+        if (optimizerOptions_ != null) {
+          optimizerOptions_ =
+            org.tensorflow.proto.framework.OptimizerOptions.newBuilder(optimizerOptions_).mergeFrom(value).buildPartial();
+        } else {
+          optimizerOptions_ = value;
+        }
+        onChanged();
+      } else {
+        optimizerOptionsBuilder_.mergeFrom(value);
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Options controlling how graph is optimized.
+     * </pre>
+     *
+     * <code>.tensorflow.OptimizerOptions optimizer_options = 3;</code>
+     */
+    public Builder clearOptimizerOptions() {
+      if (optimizerOptionsBuilder_ == null) {
+        optimizerOptions_ = null;
+        onChanged();
+      } else {
+        optimizerOptions_ = null;
+        optimizerOptionsBuilder_ = null;
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Options controlling how graph is optimized.
+     * </pre>
+     *
+     * <code>.tensorflow.OptimizerOptions optimizer_options = 3;</code>
+     */
+    public org.tensorflow.proto.framework.OptimizerOptions.Builder getOptimizerOptionsBuilder() {
+      
+      onChanged();
+      return getOptimizerOptionsFieldBuilder().getBuilder();
+    }
+    /**
+     * <pre>
+     * Options controlling how graph is optimized.
+     * </pre>
+     *
+     * <code>.tensorflow.OptimizerOptions optimizer_options = 3;</code>
+     */
+    public org.tensorflow.proto.framework.OptimizerOptionsOrBuilder getOptimizerOptionsOrBuilder() {
+      if (optimizerOptionsBuilder_ != null) {
+        return optimizerOptionsBuilder_.getMessageOrBuilder();
+      } else {
+        return optimizerOptions_ == null ?
+            org.tensorflow.proto.framework.OptimizerOptions.getDefaultInstance() : optimizerOptions_;
+      }
+    }
+    /**
+     * <pre>
+     * Options controlling how graph is optimized.
+     * </pre>
+     *
+     * <code>.tensorflow.OptimizerOptions optimizer_options = 3;</code>
+     */
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.OptimizerOptions, org.tensorflow.proto.framework.OptimizerOptions.Builder, org.tensorflow.proto.framework.OptimizerOptionsOrBuilder> 
+        getOptimizerOptionsFieldBuilder() {
+      if (optimizerOptionsBuilder_ == null) {
+        optimizerOptionsBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+            org.tensorflow.proto.framework.OptimizerOptions, org.tensorflow.proto.framework.OptimizerOptions.Builder, org.tensorflow.proto.framework.OptimizerOptionsOrBuilder>(
+                getOptimizerOptions(),
+                getParentForChildren(),
+                isClean());
+        optimizerOptions_ = null;
+      }
+      return optimizerOptionsBuilder_;
+    }
+
+    private long buildCostModel_ ;
+    /**
+     * <pre>
+     * The number of steps to run before returning a cost model detailing
+     * the memory usage and performance of each node of the graph. 0 means
+     * no cost model.
+     * </pre>
+     *
+     * <code>int64 build_cost_model = 4;</code>
+     */
+    public long getBuildCostModel() {
+      return buildCostModel_;
+    }
+    /**
+     * <pre>
+     * The number of steps to run before returning a cost model detailing
+     * the memory usage and performance of each node of the graph. 0 means
+     * no cost model.
+     * </pre>
+     *
+     * <code>int64 build_cost_model = 4;</code>
+     */
+    public Builder setBuildCostModel(long value) {
+      
+      buildCostModel_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * The number of steps to run before returning a cost model detailing
+     * the memory usage and performance of each node of the graph. 0 means
+     * no cost model.
+     * </pre>
+     *
+     * <code>int64 build_cost_model = 4;</code>
+     */
+    public Builder clearBuildCostModel() {
+      
+      buildCostModel_ = 0L;
+      onChanged();
+      return this;
+    }
+
+    private long buildCostModelAfter_ ;
+    /**
+     * <pre>
+     * The number of steps to skip before collecting statistics for the
+     * cost model.
+     * </pre>
+     *
+     * <code>int64 build_cost_model_after = 9;</code>
+     */
+    public long getBuildCostModelAfter() {
+      return buildCostModelAfter_;
+    }
+    /**
+     * <pre>
+     * The number of steps to skip before collecting statistics for the
+     * cost model.
+     * </pre>
+     *
+     * <code>int64 build_cost_model_after = 9;</code>
+     */
+    public Builder setBuildCostModelAfter(long value) {
+      
+      buildCostModelAfter_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * The number of steps to skip before collecting statistics for the
+     * cost model.
+     * </pre>
+     *
+     * <code>int64 build_cost_model_after = 9;</code>
+     */
+    public Builder clearBuildCostModelAfter() {
+      
+      buildCostModelAfter_ = 0L;
+      onChanged();
+      return this;
+    }
+
+    private boolean inferShapes_ ;
+    /**
+     * <pre>
+     * Annotate each Node with Op output shape data, to the extent it can
+     * be statically inferred.
+     * </pre>
+     *
+     * <code>bool infer_shapes = 5;</code>
+     */
+    public boolean getInferShapes() {
+      return inferShapes_;
+    }
+    /**
+     * <pre>
+     * Annotate each Node with Op output shape data, to the extent it can
+     * be statically inferred.
+     * </pre>
+     *
+     * <code>bool infer_shapes = 5;</code>
+     */
+    public Builder setInferShapes(boolean value) {
+      
+      inferShapes_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Annotate each Node with Op output shape data, to the extent it can
+     * be statically inferred.
+     * </pre>
+     *
+     * <code>bool infer_shapes = 5;</code>
+     */
+    public Builder clearInferShapes() {
+      
+      inferShapes_ = false;
+      onChanged();
+      return this;
+    }
+
+    private boolean placePrunedGraph_ ;
+    /**
+     * <pre>
+     * Only place the subgraphs that are run, rather than the entire graph.
+     * This is useful for interactive graph building, where one might
+     * produce graphs that cannot be placed during the debugging
+     * process.  In particular, it allows the client to continue work in
+     * a session after adding a node to a graph whose placement
+     * constraints are unsatisfiable.
+     * </pre>
+     *
+     * <code>bool place_pruned_graph = 6;</code>
+     */
+    public boolean getPlacePrunedGraph() {
+      return placePrunedGraph_;
+    }
+    /**
+     * <pre>
+     * Only place the subgraphs that are run, rather than the entire graph.
+     * This is useful for interactive graph building, where one might
+     * produce graphs that cannot be placed during the debugging
+     * process.  In particular, it allows the client to continue work in
+     * a session after adding a node to a graph whose placement
+     * constraints are unsatisfiable.
+     * </pre>
+     *
+     * <code>bool place_pruned_graph = 6;</code>
+     */
+    public Builder setPlacePrunedGraph(boolean value) {
+      
+      placePrunedGraph_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Only place the subgraphs that are run, rather than the entire graph.
+     * This is useful for interactive graph building, where one might
+     * produce graphs that cannot be placed during the debugging
+     * process.  In particular, it allows the client to continue work in
+     * a session after adding a node to a graph whose placement
+     * constraints are unsatisfiable.
+     * </pre>
+     *
+     * <code>bool place_pruned_graph = 6;</code>
+     */
+    public Builder clearPlacePrunedGraph() {
+      
+      placePrunedGraph_ = false;
+      onChanged();
+      return this;
+    }
+
+    private boolean enableBfloat16Sendrecv_ ;
+    /**
+     * <pre>
+     * If true, transfer float values between processes as bfloat16.
+     * </pre>
+     *
+     * <code>bool enable_bfloat16_sendrecv = 7;</code>
+     */
+    public boolean getEnableBfloat16Sendrecv() {
+      return enableBfloat16Sendrecv_;
+    }
+    /**
+     * <pre>
+     * If true, transfer float values between processes as bfloat16.
+     * </pre>
+     *
+     * <code>bool enable_bfloat16_sendrecv = 7;</code>
+     */
+    public Builder setEnableBfloat16Sendrecv(boolean value) {
+      
+      enableBfloat16Sendrecv_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * If true, transfer float values between processes as bfloat16.
+     * </pre>
+     *
+     * <code>bool enable_bfloat16_sendrecv = 7;</code>
+     */
+    public Builder clearEnableBfloat16Sendrecv() {
+      
+      enableBfloat16Sendrecv_ = false;
+      onChanged();
+      return this;
+    }
+
+    private int timelineStep_ ;
+    /**
+     * <pre>
+     * If &gt; 0, record a timeline every this many steps.
+     * EXPERIMENTAL: This currently has no effect in MasterSession.
+     * </pre>
+     *
+     * <code>int32 timeline_step = 8;</code>
+     */
+    public int getTimelineStep() {
+      return timelineStep_;
+    }
+    /**
+     * <pre>
+     * If &gt; 0, record a timeline every this many steps.
+     * EXPERIMENTAL: This currently has no effect in MasterSession.
+     * </pre>
+     *
+     * <code>int32 timeline_step = 8;</code>
+     */
+    public Builder setTimelineStep(int value) {
+      
+      timelineStep_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * If &gt; 0, record a timeline every this many steps.
+     * EXPERIMENTAL: This currently has no effect in MasterSession.
+     * </pre>
+     *
+     * <code>int32 timeline_step = 8;</code>
+     */
+    public Builder clearTimelineStep() {
+      
+      timelineStep_ = 0;
+      onChanged();
+      return this;
+    }
+
+    private org.tensorflow.proto.framework.RewriterConfig rewriteOptions_;
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.RewriterConfig, org.tensorflow.proto.framework.RewriterConfig.Builder, org.tensorflow.proto.framework.RewriterConfigOrBuilder> rewriteOptionsBuilder_;
+    /**
+     * <pre>
+     * Options that control the type and amount of graph rewriting.
+     * Not currently configurable via the public Python API (i.e. there is no API
+     * stability guarantee if you import RewriterConfig explicitly).
+     * </pre>
+     *
+     * <code>.tensorflow.RewriterConfig rewrite_options = 10;</code>
+     */
+    public boolean hasRewriteOptions() {
+      return rewriteOptionsBuilder_ != null || rewriteOptions_ != null;
+    }
+    /**
+     * <pre>
+     * Options that control the type and amount of graph rewriting.
+     * Not currently configurable via the public Python API (i.e. there is no API
+     * stability guarantee if you import RewriterConfig explicitly).
+     * </pre>
+     *
+     * <code>.tensorflow.RewriterConfig rewrite_options = 10;</code>
+     */
+    public org.tensorflow.proto.framework.RewriterConfig getRewriteOptions() {
+      if (rewriteOptionsBuilder_ == null) {
+        return rewriteOptions_ == null ? org.tensorflow.proto.framework.RewriterConfig.getDefaultInstance() : rewriteOptions_;
+      } else {
+        return rewriteOptionsBuilder_.getMessage();
+      }
+    }
+    /**
+     * <pre>
+     * Options that control the type and amount of graph rewriting.
+     * Not currently configurable via the public Python API (i.e. there is no API
+     * stability guarantee if you import RewriterConfig explicitly).
+     * </pre>
+     *
+     * <code>.tensorflow.RewriterConfig rewrite_options = 10;</code>
+     */
+    public Builder setRewriteOptions(org.tensorflow.proto.framework.RewriterConfig value) {
+      if (rewriteOptionsBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        rewriteOptions_ = value;
+        onChanged();
+      } else {
+        rewriteOptionsBuilder_.setMessage(value);
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Options that control the type and amount of graph rewriting.
+     * Not currently configurable via the public Python API (i.e. there is no API
+     * stability guarantee if you import RewriterConfig explicitly).
+     * </pre>
+     *
+     * <code>.tensorflow.RewriterConfig rewrite_options = 10;</code>
+     */
+    public Builder setRewriteOptions(
+        org.tensorflow.proto.framework.RewriterConfig.Builder builderForValue) {
+      if (rewriteOptionsBuilder_ == null) {
+        rewriteOptions_ = builderForValue.build();
+        onChanged();
+      } else {
+        rewriteOptionsBuilder_.setMessage(builderForValue.build());
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Options that control the type and amount of graph rewriting.
+     * Not currently configurable via the public Python API (i.e. there is no API
+     * stability guarantee if you import RewriterConfig explicitly).
+     * </pre>
+     *
+     * <code>.tensorflow.RewriterConfig rewrite_options = 10;</code>
+     */
+    public Builder mergeRewriteOptions(org.tensorflow.proto.framework.RewriterConfig value) {
+      if (rewriteOptionsBuilder_ == null) {
+        if (rewriteOptions_ != null) {
+          rewriteOptions_ =
+            org.tensorflow.proto.framework.RewriterConfig.newBuilder(rewriteOptions_).mergeFrom(value).buildPartial();
+        } else {
+          rewriteOptions_ = value;
+        }
+        onChanged();
+      } else {
+        rewriteOptionsBuilder_.mergeFrom(value);
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Options that control the type and amount of graph rewriting.
+     * Not currently configurable via the public Python API (i.e. there is no API
+     * stability guarantee if you import RewriterConfig explicitly).
+     * </pre>
+     *
+     * <code>.tensorflow.RewriterConfig rewrite_options = 10;</code>
+     */
+    public Builder clearRewriteOptions() {
+      if (rewriteOptionsBuilder_ == null) {
+        rewriteOptions_ = null;
+        onChanged();
+      } else {
+        rewriteOptions_ = null;
+        rewriteOptionsBuilder_ = null;
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Options that control the type and amount of graph rewriting.
+     * Not currently configurable via the public Python API (i.e. there is no API
+     * stability guarantee if you import RewriterConfig explicitly).
+     * </pre>
+     *
+     * <code>.tensorflow.RewriterConfig rewrite_options = 10;</code>
+     */
+    public org.tensorflow.proto.framework.RewriterConfig.Builder getRewriteOptionsBuilder() {
+      
+      onChanged();
+      return getRewriteOptionsFieldBuilder().getBuilder();
+    }
+    /**
+     * <pre>
+     * Options that control the type and amount of graph rewriting.
+     * Not currently configurable via the public Python API (i.e. there is no API
+     * stability guarantee if you import RewriterConfig explicitly).
+     * </pre>
+     *
+     * <code>.tensorflow.RewriterConfig rewrite_options = 10;</code>
+     */
+    public org.tensorflow.proto.framework.RewriterConfigOrBuilder getRewriteOptionsOrBuilder() {
+      if (rewriteOptionsBuilder_ != null) {
+        return rewriteOptionsBuilder_.getMessageOrBuilder();
+      } else {
+        return rewriteOptions_ == null ?
+            org.tensorflow.proto.framework.RewriterConfig.getDefaultInstance() : rewriteOptions_;
+      }
+    }
+    /**
+     * <pre>
+     * Options that control the type and amount of graph rewriting.
+     * Not currently configurable via the public Python API (i.e. there is no API
+     * stability guarantee if you import RewriterConfig explicitly).
+     * </pre>
+     *
+     * <code>.tensorflow.RewriterConfig rewrite_options = 10;</code>
+     */
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.RewriterConfig, org.tensorflow.proto.framework.RewriterConfig.Builder, org.tensorflow.proto.framework.RewriterConfigOrBuilder> 
+        getRewriteOptionsFieldBuilder() {
+      if (rewriteOptionsBuilder_ == null) {
+        rewriteOptionsBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+            org.tensorflow.proto.framework.RewriterConfig, org.tensorflow.proto.framework.RewriterConfig.Builder, org.tensorflow.proto.framework.RewriterConfigOrBuilder>(
+                getRewriteOptions(),
+                getParentForChildren(),
+                isClean());
+        rewriteOptions_ = null;
+      }
+      return rewriteOptionsBuilder_;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.GraphOptions)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.GraphOptions)
+  private static final org.tensorflow.proto.framework.GraphOptions DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.GraphOptions();
+  }
+
+  public static org.tensorflow.proto.framework.GraphOptions getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<GraphOptions>
+      PARSER = new com.google.protobuf.AbstractParser<GraphOptions>() {
+    @java.lang.Override
+    public GraphOptions parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new GraphOptions(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<GraphOptions> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<GraphOptions> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.framework.GraphOptions getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphOptionsOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphOptionsOrBuilder.java
new file mode 100644
index 00000000000..bc373a1f956
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphOptionsOrBuilder.java
@@ -0,0 +1,139 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/config.proto
+
+package org.tensorflow.proto.framework;
+
+public interface GraphOptionsOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.GraphOptions)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <pre>
+   * If true, use control flow to schedule the activation of Recv nodes.
+   * (Currently ignored.)
+   * </pre>
+   *
+   * <code>bool enable_recv_scheduling = 2;</code>
+   */
+  boolean getEnableRecvScheduling();
+
+  /**
+   * <pre>
+   * Options controlling how graph is optimized.
+   * </pre>
+   *
+   * <code>.tensorflow.OptimizerOptions optimizer_options = 3;</code>
+   */
+  boolean hasOptimizerOptions();
+  /**
+   * <pre>
+   * Options controlling how graph is optimized.
+   * </pre>
+   *
+   * <code>.tensorflow.OptimizerOptions optimizer_options = 3;</code>
+   */
+  org.tensorflow.proto.framework.OptimizerOptions getOptimizerOptions();
+  /**
+   * <pre>
+   * Options controlling how graph is optimized.
+   * </pre>
+   *
+   * <code>.tensorflow.OptimizerOptions optimizer_options = 3;</code>
+   */
+  org.tensorflow.proto.framework.OptimizerOptionsOrBuilder getOptimizerOptionsOrBuilder();
+
+  /**
+   * <pre>
+   * The number of steps to run before returning a cost model detailing
+   * the memory usage and performance of each node of the graph. 0 means
+   * no cost model.
+   * </pre>
+   *
+   * <code>int64 build_cost_model = 4;</code>
+   */
+  long getBuildCostModel();
+
+  /**
+   * <pre>
+   * The number of steps to skip before collecting statistics for the
+   * cost model.
+   * </pre>
+   *
+   * <code>int64 build_cost_model_after = 9;</code>
+   */
+  long getBuildCostModelAfter();
+
+  /**
+   * <pre>
+   * Annotate each Node with Op output shape data, to the extent it can
+   * be statically inferred.
+   * </pre>
+   *
+   * <code>bool infer_shapes = 5;</code>
+   */
+  boolean getInferShapes();
+
+  /**
+   * <pre>
+   * Only place the subgraphs that are run, rather than the entire graph.
+   * This is useful for interactive graph building, where one might
+   * produce graphs that cannot be placed during the debugging
+   * process.  In particular, it allows the client to continue work in
+   * a session after adding a node to a graph whose placement
+   * constraints are unsatisfiable.
+   * </pre>
+   *
+   * <code>bool place_pruned_graph = 6;</code>
+   */
+  boolean getPlacePrunedGraph();
+
+  /**
+   * <pre>
+   * If true, transfer float values between processes as bfloat16.
+   * </pre>
+   *
+   * <code>bool enable_bfloat16_sendrecv = 7;</code>
+   */
+  boolean getEnableBfloat16Sendrecv();
+
+  /**
+   * <pre>
+   * If &gt; 0, record a timeline every this many steps.
+   * EXPERIMENTAL: This currently has no effect in MasterSession.
+   * </pre>
+   *
+   * <code>int32 timeline_step = 8;</code>
+   */
+  int getTimelineStep();
+
+  /**
+   * <pre>
+   * Options that control the type and amount of graph rewriting.
+   * Not currently configurable via the public Python API (i.e. there is no API
+   * stability guarantee if you import RewriterConfig explicitly).
+   * </pre>
+   *
+   * <code>.tensorflow.RewriterConfig rewrite_options = 10;</code>
+   */
+  boolean hasRewriteOptions();
+  /**
+   * <pre>
+   * Options that control the type and amount of graph rewriting.
+   * Not currently configurable via the public Python API (i.e. there is no API
+   * stability guarantee if you import RewriterConfig explicitly).
+   * </pre>
+   *
+   * <code>.tensorflow.RewriterConfig rewrite_options = 10;</code>
+   */
+  org.tensorflow.proto.framework.RewriterConfig getRewriteOptions();
+  /**
+   * <pre>
+   * Options that control the type and amount of graph rewriting.
+   * Not currently configurable via the public Python API (i.e. there is no API
+   * stability guarantee if you import RewriterConfig explicitly).
+   * </pre>
+   *
+   * <code>.tensorflow.RewriterConfig rewrite_options = 10;</code>
+   */
+  org.tensorflow.proto.framework.RewriterConfigOrBuilder getRewriteOptionsOrBuilder();
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphProtos.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphProtos.java
new file mode 100644
index 00000000000..43eba4ee9bd
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphProtos.java
@@ -0,0 +1,63 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/framework/graph.proto
+
+package org.tensorflow.proto.framework;
+
+public final class GraphProtos {
+  private GraphProtos() {}
+  public static void registerAllExtensions(
+      com.google.protobuf.ExtensionRegistryLite registry) {
+  }
+
+  public static void registerAllExtensions(
+      com.google.protobuf.ExtensionRegistry registry) {
+    registerAllExtensions(
+        (com.google.protobuf.ExtensionRegistryLite) registry);
+  }
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_GraphDef_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_GraphDef_fieldAccessorTable;
+
+  public static com.google.protobuf.Descriptors.FileDescriptor
+      getDescriptor() {
+    return descriptor;
+  }
+  private static  com.google.protobuf.Descriptors.FileDescriptor
+      descriptor;
+  static {
+    java.lang.String[] descriptorData = {
+      "\n%tensorflow/core/framework/graph.proto\022" +
+      "\ntensorflow\032(tensorflow/core/framework/f" +
+      "unction.proto\032(tensorflow/core/framework" +
+      "/node_def.proto\032(tensorflow/core/framewo" +
+      "rk/versions.proto\"\235\001\n\010GraphDef\022!\n\004node\030\001" +
+      " \003(\0132\023.tensorflow.NodeDef\022(\n\010versions\030\004 " +
+      "\001(\0132\026.tensorflow.VersionDef\022\023\n\007version\030\003" +
+      " \001(\005B\002\030\001\022/\n\007library\030\002 \001(\0132\036.tensorflow.F" +
+      "unctionDefLibraryB\200\001\n\036org.tensorflow.pro" +
+      "to.frameworkB\013GraphProtosP\001ZLgithub.com/" +
+      "tensorflow/tensorflow/tensorflow/go/core" +
+      "/framework/graph_go_proto\370\001\001b\006proto3"
+    };
+    descriptor = com.google.protobuf.Descriptors.FileDescriptor
+      .internalBuildGeneratedFileFrom(descriptorData,
+        new com.google.protobuf.Descriptors.FileDescriptor[] {
+          org.tensorflow.proto.framework.FunctionProtos.getDescriptor(),
+          org.tensorflow.proto.framework.NodeProto.getDescriptor(),
+          org.tensorflow.proto.framework.VersionsProtos.getDescriptor(),
+        });
+    internal_static_tensorflow_GraphDef_descriptor =
+      getDescriptor().getMessageTypes().get(0);
+    internal_static_tensorflow_GraphDef_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_GraphDef_descriptor,
+        new java.lang.String[] { "Node", "Versions", "Version", "Library", });
+    org.tensorflow.proto.framework.FunctionProtos.getDescriptor();
+    org.tensorflow.proto.framework.NodeProto.getDescriptor();
+    org.tensorflow.proto.framework.VersionsProtos.getDescriptor();
+  }
+
+  // @@protoc_insertion_point(outer_class_scope)
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphTransferConstNodeInfo.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphTransferConstNodeInfo.java
new file mode 100644
index 00000000000..a1564a2090a
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphTransferConstNodeInfo.java
@@ -0,0 +1,912 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/framework/graph_transfer_info.proto
+
+package org.tensorflow.proto.framework;
+
+/**
+ * Protobuf type {@code tensorflow.GraphTransferConstNodeInfo}
+ */
+public  final class GraphTransferConstNodeInfo extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.GraphTransferConstNodeInfo)
+    GraphTransferConstNodeInfoOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use GraphTransferConstNodeInfo.newBuilder() to construct.
+  private GraphTransferConstNodeInfo(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private GraphTransferConstNodeInfo() {
+    name_ = "";
+    shape_ = emptyLongList();
+    data_ = com.google.protobuf.ByteString.EMPTY;
+    dtype_ = 0;
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new GraphTransferConstNodeInfo();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private GraphTransferConstNodeInfo(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    int mutable_bitField0_ = 0;
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 10: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            name_ = s;
+            break;
+          }
+          case 16: {
+
+            nodeId_ = input.readInt32();
+            break;
+          }
+          case 24: {
+            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
+              shape_ = newLongList();
+              mutable_bitField0_ |= 0x00000001;
+            }
+            shape_.addLong(input.readInt64());
+            break;
+          }
+          case 26: {
+            int length = input.readRawVarint32();
+            int limit = input.pushLimit(length);
+            if (!((mutable_bitField0_ & 0x00000001) != 0) && input.getBytesUntilLimit() > 0) {
+              shape_ = newLongList();
+              mutable_bitField0_ |= 0x00000001;
+            }
+            while (input.getBytesUntilLimit() > 0) {
+              shape_.addLong(input.readInt64());
+            }
+            input.popLimit(limit);
+            break;
+          }
+          case 34: {
+
+            data_ = input.readBytes();
+            break;
+          }
+          case 40: {
+            int rawValue = input.readEnum();
+
+            dtype_ = rawValue;
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      if (((mutable_bitField0_ & 0x00000001) != 0)) {
+        shape_.makeImmutable(); // C
+      }
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.framework.GraphTransferInfoProto.internal_static_tensorflow_GraphTransferConstNodeInfo_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.framework.GraphTransferInfoProto.internal_static_tensorflow_GraphTransferConstNodeInfo_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.framework.GraphTransferConstNodeInfo.class, org.tensorflow.proto.framework.GraphTransferConstNodeInfo.Builder.class);
+  }
+
+  public static final int NAME_FIELD_NUMBER = 1;
+  private volatile java.lang.Object name_;
+  /**
+   * <code>string name = 1;</code>
+   */
+  public java.lang.String getName() {
+    java.lang.Object ref = name_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      name_ = s;
+      return s;
+    }
+  }
+  /**
+   * <code>string name = 1;</code>
+   */
+  public com.google.protobuf.ByteString
+      getNameBytes() {
+    java.lang.Object ref = name_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      name_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int NODE_ID_FIELD_NUMBER = 2;
+  private int nodeId_;
+  /**
+   * <code>int32 node_id = 2;</code>
+   */
+  public int getNodeId() {
+    return nodeId_;
+  }
+
+  public static final int SHAPE_FIELD_NUMBER = 3;
+  private com.google.protobuf.Internal.LongList shape_;
+  /**
+   * <code>repeated int64 shape = 3;</code>
+   */
+  public java.util.List<java.lang.Long>
+      getShapeList() {
+    return shape_;
+  }
+  /**
+   * <code>repeated int64 shape = 3;</code>
+   */
+  public int getShapeCount() {
+    return shape_.size();
+  }
+  /**
+   * <code>repeated int64 shape = 3;</code>
+   */
+  public long getShape(int index) {
+    return shape_.getLong(index);
+  }
+  private int shapeMemoizedSerializedSize = -1;
+
+  public static final int DATA_FIELD_NUMBER = 4;
+  private com.google.protobuf.ByteString data_;
+  /**
+   * <code>bytes data = 4;</code>
+   */
+  public com.google.protobuf.ByteString getData() {
+    return data_;
+  }
+
+  public static final int DTYPE_FIELD_NUMBER = 5;
+  private int dtype_;
+  /**
+   * <code>.tensorflow.DataType dtype = 5;</code>
+   */
+  public int getDtypeValue() {
+    return dtype_;
+  }
+  /**
+   * <code>.tensorflow.DataType dtype = 5;</code>
+   */
+  public org.tensorflow.proto.framework.DataType getDtype() {
+    @SuppressWarnings("deprecation")
+    org.tensorflow.proto.framework.DataType result = org.tensorflow.proto.framework.DataType.valueOf(dtype_);
+    return result == null ? org.tensorflow.proto.framework.DataType.UNRECOGNIZED : result;
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    getSerializedSize();
+    if (!getNameBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 1, name_);
+    }
+    if (nodeId_ != 0) {
+      output.writeInt32(2, nodeId_);
+    }
+    if (getShapeList().size() > 0) {
+      output.writeUInt32NoTag(26);
+      output.writeUInt32NoTag(shapeMemoizedSerializedSize);
+    }
+    for (int i = 0; i < shape_.size(); i++) {
+      output.writeInt64NoTag(shape_.getLong(i));
+    }
+    if (!data_.isEmpty()) {
+      output.writeBytes(4, data_);
+    }
+    if (dtype_ != org.tensorflow.proto.framework.DataType.DT_INVALID.getNumber()) {
+      output.writeEnum(5, dtype_);
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (!getNameBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, name_);
+    }
+    if (nodeId_ != 0) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt32Size(2, nodeId_);
+    }
+    {
+      int dataSize = 0;
+      for (int i = 0; i < shape_.size(); i++) {
+        dataSize += com.google.protobuf.CodedOutputStream
+          .computeInt64SizeNoTag(shape_.getLong(i));
+      }
+      size += dataSize;
+      if (!getShapeList().isEmpty()) {
+        size += 1;
+        size += com.google.protobuf.CodedOutputStream
+            .computeInt32SizeNoTag(dataSize);
+      }
+      shapeMemoizedSerializedSize = dataSize;
+    }
+    if (!data_.isEmpty()) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeBytesSize(4, data_);
+    }
+    if (dtype_ != org.tensorflow.proto.framework.DataType.DT_INVALID.getNumber()) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeEnumSize(5, dtype_);
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.framework.GraphTransferConstNodeInfo)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.framework.GraphTransferConstNodeInfo other = (org.tensorflow.proto.framework.GraphTransferConstNodeInfo) obj;
+
+    if (!getName()
+        .equals(other.getName())) return false;
+    if (getNodeId()
+        != other.getNodeId()) return false;
+    if (!getShapeList()
+        .equals(other.getShapeList())) return false;
+    if (!getData()
+        .equals(other.getData())) return false;
+    if (dtype_ != other.dtype_) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    hash = (37 * hash) + NAME_FIELD_NUMBER;
+    hash = (53 * hash) + getName().hashCode();
+    hash = (37 * hash) + NODE_ID_FIELD_NUMBER;
+    hash = (53 * hash) + getNodeId();
+    if (getShapeCount() > 0) {
+      hash = (37 * hash) + SHAPE_FIELD_NUMBER;
+      hash = (53 * hash) + getShapeList().hashCode();
+    }
+    hash = (37 * hash) + DATA_FIELD_NUMBER;
+    hash = (53 * hash) + getData().hashCode();
+    hash = (37 * hash) + DTYPE_FIELD_NUMBER;
+    hash = (53 * hash) + dtype_;
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.framework.GraphTransferConstNodeInfo parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.GraphTransferConstNodeInfo parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.GraphTransferConstNodeInfo parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.GraphTransferConstNodeInfo parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.GraphTransferConstNodeInfo parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.GraphTransferConstNodeInfo parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.GraphTransferConstNodeInfo parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.GraphTransferConstNodeInfo parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.GraphTransferConstNodeInfo parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.GraphTransferConstNodeInfo parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.GraphTransferConstNodeInfo parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.GraphTransferConstNodeInfo parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.framework.GraphTransferConstNodeInfo prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * Protobuf type {@code tensorflow.GraphTransferConstNodeInfo}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.GraphTransferConstNodeInfo)
+      org.tensorflow.proto.framework.GraphTransferConstNodeInfoOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.GraphTransferInfoProto.internal_static_tensorflow_GraphTransferConstNodeInfo_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.GraphTransferInfoProto.internal_static_tensorflow_GraphTransferConstNodeInfo_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.GraphTransferConstNodeInfo.class, org.tensorflow.proto.framework.GraphTransferConstNodeInfo.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.framework.GraphTransferConstNodeInfo.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      name_ = "";
+
+      nodeId_ = 0;
+
+      shape_ = emptyLongList();
+      bitField0_ = (bitField0_ & ~0x00000001);
+      data_ = com.google.protobuf.ByteString.EMPTY;
+
+      dtype_ = 0;
+
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.framework.GraphTransferInfoProto.internal_static_tensorflow_GraphTransferConstNodeInfo_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.GraphTransferConstNodeInfo getDefaultInstanceForType() {
+      return org.tensorflow.proto.framework.GraphTransferConstNodeInfo.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.GraphTransferConstNodeInfo build() {
+      org.tensorflow.proto.framework.GraphTransferConstNodeInfo result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.GraphTransferConstNodeInfo buildPartial() {
+      org.tensorflow.proto.framework.GraphTransferConstNodeInfo result = new org.tensorflow.proto.framework.GraphTransferConstNodeInfo(this);
+      int from_bitField0_ = bitField0_;
+      result.name_ = name_;
+      result.nodeId_ = nodeId_;
+      if (((bitField0_ & 0x00000001) != 0)) {
+        shape_.makeImmutable();
+        bitField0_ = (bitField0_ & ~0x00000001);
+      }
+      result.shape_ = shape_;
+      result.data_ = data_;
+      result.dtype_ = dtype_;
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.framework.GraphTransferConstNodeInfo) {
+        return mergeFrom((org.tensorflow.proto.framework.GraphTransferConstNodeInfo)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.framework.GraphTransferConstNodeInfo other) {
+      if (other == org.tensorflow.proto.framework.GraphTransferConstNodeInfo.getDefaultInstance()) return this;
+      if (!other.getName().isEmpty()) {
+        name_ = other.name_;
+        onChanged();
+      }
+      if (other.getNodeId() != 0) {
+        setNodeId(other.getNodeId());
+      }
+      if (!other.shape_.isEmpty()) {
+        if (shape_.isEmpty()) {
+          shape_ = other.shape_;
+          bitField0_ = (bitField0_ & ~0x00000001);
+        } else {
+          ensureShapeIsMutable();
+          shape_.addAll(other.shape_);
+        }
+        onChanged();
+      }
+      if (other.getData() != com.google.protobuf.ByteString.EMPTY) {
+        setData(other.getData());
+      }
+      if (other.dtype_ != 0) {
+        setDtypeValue(other.getDtypeValue());
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.framework.GraphTransferConstNodeInfo parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.framework.GraphTransferConstNodeInfo) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+    private int bitField0_;
+
+    private java.lang.Object name_ = "";
+    /**
+     * <code>string name = 1;</code>
+     */
+    public java.lang.String getName() {
+      java.lang.Object ref = name_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        name_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <code>string name = 1;</code>
+     */
+    public com.google.protobuf.ByteString
+        getNameBytes() {
+      java.lang.Object ref = name_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        name_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <code>string name = 1;</code>
+     */
+    public Builder setName(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      name_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>string name = 1;</code>
+     */
+    public Builder clearName() {
+      
+      name_ = getDefaultInstance().getName();
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>string name = 1;</code>
+     */
+    public Builder setNameBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      name_ = value;
+      onChanged();
+      return this;
+    }
+
+    private int nodeId_ ;
+    /**
+     * <code>int32 node_id = 2;</code>
+     */
+    public int getNodeId() {
+      return nodeId_;
+    }
+    /**
+     * <code>int32 node_id = 2;</code>
+     */
+    public Builder setNodeId(int value) {
+      
+      nodeId_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>int32 node_id = 2;</code>
+     */
+    public Builder clearNodeId() {
+      
+      nodeId_ = 0;
+      onChanged();
+      return this;
+    }
+
+    private com.google.protobuf.Internal.LongList shape_ = emptyLongList();
+    private void ensureShapeIsMutable() {
+      if (!((bitField0_ & 0x00000001) != 0)) {
+        shape_ = mutableCopy(shape_);
+        bitField0_ |= 0x00000001;
+       }
+    }
+    /**
+     * <code>repeated int64 shape = 3;</code>
+     */
+    public java.util.List<java.lang.Long>
+        getShapeList() {
+      return ((bitField0_ & 0x00000001) != 0) ?
+               java.util.Collections.unmodifiableList(shape_) : shape_;
+    }
+    /**
+     * <code>repeated int64 shape = 3;</code>
+     */
+    public int getShapeCount() {
+      return shape_.size();
+    }
+    /**
+     * <code>repeated int64 shape = 3;</code>
+     */
+    public long getShape(int index) {
+      return shape_.getLong(index);
+    }
+    /**
+     * <code>repeated int64 shape = 3;</code>
+     */
+    public Builder setShape(
+        int index, long value) {
+      ensureShapeIsMutable();
+      shape_.setLong(index, value);
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>repeated int64 shape = 3;</code>
+     */
+    public Builder addShape(long value) {
+      ensureShapeIsMutable();
+      shape_.addLong(value);
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>repeated int64 shape = 3;</code>
+     */
+    public Builder addAllShape(
+        java.lang.Iterable<? extends java.lang.Long> values) {
+      ensureShapeIsMutable();
+      com.google.protobuf.AbstractMessageLite.Builder.addAll(
+          values, shape_);
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>repeated int64 shape = 3;</code>
+     */
+    public Builder clearShape() {
+      shape_ = emptyLongList();
+      bitField0_ = (bitField0_ & ~0x00000001);
+      onChanged();
+      return this;
+    }
+
+    private com.google.protobuf.ByteString data_ = com.google.protobuf.ByteString.EMPTY;
+    /**
+     * <code>bytes data = 4;</code>
+     */
+    public com.google.protobuf.ByteString getData() {
+      return data_;
+    }
+    /**
+     * <code>bytes data = 4;</code>
+     */
+    public Builder setData(com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      data_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>bytes data = 4;</code>
+     */
+    public Builder clearData() {
+      
+      data_ = getDefaultInstance().getData();
+      onChanged();
+      return this;
+    }
+
+    private int dtype_ = 0;
+    /**
+     * <code>.tensorflow.DataType dtype = 5;</code>
+     */
+    public int getDtypeValue() {
+      return dtype_;
+    }
+    /**
+     * <code>.tensorflow.DataType dtype = 5;</code>
+     */
+    public Builder setDtypeValue(int value) {
+      dtype_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>.tensorflow.DataType dtype = 5;</code>
+     */
+    public org.tensorflow.proto.framework.DataType getDtype() {
+      @SuppressWarnings("deprecation")
+      org.tensorflow.proto.framework.DataType result = org.tensorflow.proto.framework.DataType.valueOf(dtype_);
+      return result == null ? org.tensorflow.proto.framework.DataType.UNRECOGNIZED : result;
+    }
+    /**
+     * <code>.tensorflow.DataType dtype = 5;</code>
+     */
+    public Builder setDtype(org.tensorflow.proto.framework.DataType value) {
+      if (value == null) {
+        throw new NullPointerException();
+      }
+      
+      dtype_ = value.getNumber();
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>.tensorflow.DataType dtype = 5;</code>
+     */
+    public Builder clearDtype() {
+      
+      dtype_ = 0;
+      onChanged();
+      return this;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.GraphTransferConstNodeInfo)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.GraphTransferConstNodeInfo)
+  private static final org.tensorflow.proto.framework.GraphTransferConstNodeInfo DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.GraphTransferConstNodeInfo();
+  }
+
+  public static org.tensorflow.proto.framework.GraphTransferConstNodeInfo getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<GraphTransferConstNodeInfo>
+      PARSER = new com.google.protobuf.AbstractParser<GraphTransferConstNodeInfo>() {
+    @java.lang.Override
+    public GraphTransferConstNodeInfo parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new GraphTransferConstNodeInfo(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<GraphTransferConstNodeInfo> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<GraphTransferConstNodeInfo> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.framework.GraphTransferConstNodeInfo getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphTransferConstNodeInfoOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphTransferConstNodeInfoOrBuilder.java
new file mode 100644
index 00000000000..ec0b9e60f90
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphTransferConstNodeInfoOrBuilder.java
@@ -0,0 +1,51 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/framework/graph_transfer_info.proto
+
+package org.tensorflow.proto.framework;
+
+public interface GraphTransferConstNodeInfoOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.GraphTransferConstNodeInfo)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <code>string name = 1;</code>
+   */
+  java.lang.String getName();
+  /**
+   * <code>string name = 1;</code>
+   */
+  com.google.protobuf.ByteString
+      getNameBytes();
+
+  /**
+   * <code>int32 node_id = 2;</code>
+   */
+  int getNodeId();
+
+  /**
+   * <code>repeated int64 shape = 3;</code>
+   */
+  java.util.List<java.lang.Long> getShapeList();
+  /**
+   * <code>repeated int64 shape = 3;</code>
+   */
+  int getShapeCount();
+  /**
+   * <code>repeated int64 shape = 3;</code>
+   */
+  long getShape(int index);
+
+  /**
+   * <code>bytes data = 4;</code>
+   */
+  com.google.protobuf.ByteString getData();
+
+  /**
+   * <code>.tensorflow.DataType dtype = 5;</code>
+   */
+  int getDtypeValue();
+  /**
+   * <code>.tensorflow.DataType dtype = 5;</code>
+   */
+  org.tensorflow.proto.framework.DataType getDtype();
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphTransferGraphInputNodeInfo.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphTransferGraphInputNodeInfo.java
new file mode 100644
index 00000000000..b4d12c201b5
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphTransferGraphInputNodeInfo.java
@@ -0,0 +1,794 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/framework/graph_transfer_info.proto
+
+package org.tensorflow.proto.framework;
+
+/**
+ * Protobuf type {@code tensorflow.GraphTransferGraphInputNodeInfo}
+ */
+public  final class GraphTransferGraphInputNodeInfo extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.GraphTransferGraphInputNodeInfo)
+    GraphTransferGraphInputNodeInfoOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use GraphTransferGraphInputNodeInfo.newBuilder() to construct.
+  private GraphTransferGraphInputNodeInfo(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private GraphTransferGraphInputNodeInfo() {
+    name_ = "";
+    shape_ = emptyLongList();
+    dtype_ = 0;
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new GraphTransferGraphInputNodeInfo();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private GraphTransferGraphInputNodeInfo(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    int mutable_bitField0_ = 0;
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 10: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            name_ = s;
+            break;
+          }
+          case 16: {
+            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
+              shape_ = newLongList();
+              mutable_bitField0_ |= 0x00000001;
+            }
+            shape_.addLong(input.readInt64());
+            break;
+          }
+          case 18: {
+            int length = input.readRawVarint32();
+            int limit = input.pushLimit(length);
+            if (!((mutable_bitField0_ & 0x00000001) != 0) && input.getBytesUntilLimit() > 0) {
+              shape_ = newLongList();
+              mutable_bitField0_ |= 0x00000001;
+            }
+            while (input.getBytesUntilLimit() > 0) {
+              shape_.addLong(input.readInt64());
+            }
+            input.popLimit(limit);
+            break;
+          }
+          case 24: {
+            int rawValue = input.readEnum();
+
+            dtype_ = rawValue;
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      if (((mutable_bitField0_ & 0x00000001) != 0)) {
+        shape_.makeImmutable(); // C
+      }
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.framework.GraphTransferInfoProto.internal_static_tensorflow_GraphTransferGraphInputNodeInfo_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.framework.GraphTransferInfoProto.internal_static_tensorflow_GraphTransferGraphInputNodeInfo_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfo.class, org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfo.Builder.class);
+  }
+
+  public static final int NAME_FIELD_NUMBER = 1;
+  private volatile java.lang.Object name_;
+  /**
+   * <code>string name = 1;</code>
+   */
+  public java.lang.String getName() {
+    java.lang.Object ref = name_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      name_ = s;
+      return s;
+    }
+  }
+  /**
+   * <code>string name = 1;</code>
+   */
+  public com.google.protobuf.ByteString
+      getNameBytes() {
+    java.lang.Object ref = name_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      name_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int SHAPE_FIELD_NUMBER = 2;
+  private com.google.protobuf.Internal.LongList shape_;
+  /**
+   * <code>repeated int64 shape = 2;</code>
+   */
+  public java.util.List<java.lang.Long>
+      getShapeList() {
+    return shape_;
+  }
+  /**
+   * <code>repeated int64 shape = 2;</code>
+   */
+  public int getShapeCount() {
+    return shape_.size();
+  }
+  /**
+   * <code>repeated int64 shape = 2;</code>
+   */
+  public long getShape(int index) {
+    return shape_.getLong(index);
+  }
+  private int shapeMemoizedSerializedSize = -1;
+
+  public static final int DTYPE_FIELD_NUMBER = 3;
+  private int dtype_;
+  /**
+   * <code>.tensorflow.DataType dtype = 3;</code>
+   */
+  public int getDtypeValue() {
+    return dtype_;
+  }
+  /**
+   * <code>.tensorflow.DataType dtype = 3;</code>
+   */
+  public org.tensorflow.proto.framework.DataType getDtype() {
+    @SuppressWarnings("deprecation")
+    org.tensorflow.proto.framework.DataType result = org.tensorflow.proto.framework.DataType.valueOf(dtype_);
+    return result == null ? org.tensorflow.proto.framework.DataType.UNRECOGNIZED : result;
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    getSerializedSize();
+    if (!getNameBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 1, name_);
+    }
+    if (getShapeList().size() > 0) {
+      output.writeUInt32NoTag(18);
+      output.writeUInt32NoTag(shapeMemoizedSerializedSize);
+    }
+    for (int i = 0; i < shape_.size(); i++) {
+      output.writeInt64NoTag(shape_.getLong(i));
+    }
+    if (dtype_ != org.tensorflow.proto.framework.DataType.DT_INVALID.getNumber()) {
+      output.writeEnum(3, dtype_);
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (!getNameBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, name_);
+    }
+    {
+      int dataSize = 0;
+      for (int i = 0; i < shape_.size(); i++) {
+        dataSize += com.google.protobuf.CodedOutputStream
+          .computeInt64SizeNoTag(shape_.getLong(i));
+      }
+      size += dataSize;
+      if (!getShapeList().isEmpty()) {
+        size += 1;
+        size += com.google.protobuf.CodedOutputStream
+            .computeInt32SizeNoTag(dataSize);
+      }
+      shapeMemoizedSerializedSize = dataSize;
+    }
+    if (dtype_ != org.tensorflow.proto.framework.DataType.DT_INVALID.getNumber()) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeEnumSize(3, dtype_);
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfo)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfo other = (org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfo) obj;
+
+    if (!getName()
+        .equals(other.getName())) return false;
+    if (!getShapeList()
+        .equals(other.getShapeList())) return false;
+    if (dtype_ != other.dtype_) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    hash = (37 * hash) + NAME_FIELD_NUMBER;
+    hash = (53 * hash) + getName().hashCode();
+    if (getShapeCount() > 0) {
+      hash = (37 * hash) + SHAPE_FIELD_NUMBER;
+      hash = (53 * hash) + getShapeList().hashCode();
+    }
+    hash = (37 * hash) + DTYPE_FIELD_NUMBER;
+    hash = (53 * hash) + dtype_;
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfo parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfo parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfo parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfo parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfo parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfo parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfo parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfo parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfo parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfo parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfo parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfo parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfo prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * Protobuf type {@code tensorflow.GraphTransferGraphInputNodeInfo}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.GraphTransferGraphInputNodeInfo)
+      org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfoOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.GraphTransferInfoProto.internal_static_tensorflow_GraphTransferGraphInputNodeInfo_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.GraphTransferInfoProto.internal_static_tensorflow_GraphTransferGraphInputNodeInfo_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfo.class, org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfo.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfo.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      name_ = "";
+
+      shape_ = emptyLongList();
+      bitField0_ = (bitField0_ & ~0x00000001);
+      dtype_ = 0;
+
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.framework.GraphTransferInfoProto.internal_static_tensorflow_GraphTransferGraphInputNodeInfo_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfo getDefaultInstanceForType() {
+      return org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfo.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfo build() {
+      org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfo result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfo buildPartial() {
+      org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfo result = new org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfo(this);
+      int from_bitField0_ = bitField0_;
+      result.name_ = name_;
+      if (((bitField0_ & 0x00000001) != 0)) {
+        shape_.makeImmutable();
+        bitField0_ = (bitField0_ & ~0x00000001);
+      }
+      result.shape_ = shape_;
+      result.dtype_ = dtype_;
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfo) {
+        return mergeFrom((org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfo)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfo other) {
+      if (other == org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfo.getDefaultInstance()) return this;
+      if (!other.getName().isEmpty()) {
+        name_ = other.name_;
+        onChanged();
+      }
+      if (!other.shape_.isEmpty()) {
+        if (shape_.isEmpty()) {
+          shape_ = other.shape_;
+          bitField0_ = (bitField0_ & ~0x00000001);
+        } else {
+          ensureShapeIsMutable();
+          shape_.addAll(other.shape_);
+        }
+        onChanged();
+      }
+      if (other.dtype_ != 0) {
+        setDtypeValue(other.getDtypeValue());
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfo parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfo) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+    private int bitField0_;
+
+    private java.lang.Object name_ = "";
+    /**
+     * <code>string name = 1;</code>
+     */
+    public java.lang.String getName() {
+      java.lang.Object ref = name_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        name_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <code>string name = 1;</code>
+     */
+    public com.google.protobuf.ByteString
+        getNameBytes() {
+      java.lang.Object ref = name_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        name_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <code>string name = 1;</code>
+     */
+    public Builder setName(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      name_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>string name = 1;</code>
+     */
+    public Builder clearName() {
+      
+      name_ = getDefaultInstance().getName();
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>string name = 1;</code>
+     */
+    public Builder setNameBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      name_ = value;
+      onChanged();
+      return this;
+    }
+
+    private com.google.protobuf.Internal.LongList shape_ = emptyLongList();
+    private void ensureShapeIsMutable() {
+      if (!((bitField0_ & 0x00000001) != 0)) {
+        shape_ = mutableCopy(shape_);
+        bitField0_ |= 0x00000001;
+       }
+    }
+    /**
+     * <code>repeated int64 shape = 2;</code>
+     */
+    public java.util.List<java.lang.Long>
+        getShapeList() {
+      return ((bitField0_ & 0x00000001) != 0) ?
+               java.util.Collections.unmodifiableList(shape_) : shape_;
+    }
+    /**
+     * <code>repeated int64 shape = 2;</code>
+     */
+    public int getShapeCount() {
+      return shape_.size();
+    }
+    /**
+     * <code>repeated int64 shape = 2;</code>
+     */
+    public long getShape(int index) {
+      return shape_.getLong(index);
+    }
+    /**
+     * <code>repeated int64 shape = 2;</code>
+     */
+    public Builder setShape(
+        int index, long value) {
+      ensureShapeIsMutable();
+      shape_.setLong(index, value);
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>repeated int64 shape = 2;</code>
+     */
+    public Builder addShape(long value) {
+      ensureShapeIsMutable();
+      shape_.addLong(value);
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>repeated int64 shape = 2;</code>
+     */
+    public Builder addAllShape(
+        java.lang.Iterable<? extends java.lang.Long> values) {
+      ensureShapeIsMutable();
+      com.google.protobuf.AbstractMessageLite.Builder.addAll(
+          values, shape_);
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>repeated int64 shape = 2;</code>
+     */
+    public Builder clearShape() {
+      shape_ = emptyLongList();
+      bitField0_ = (bitField0_ & ~0x00000001);
+      onChanged();
+      return this;
+    }
+
+    private int dtype_ = 0;
+    /**
+     * <code>.tensorflow.DataType dtype = 3;</code>
+     */
+    public int getDtypeValue() {
+      return dtype_;
+    }
+    /**
+     * <code>.tensorflow.DataType dtype = 3;</code>
+     */
+    public Builder setDtypeValue(int value) {
+      dtype_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>.tensorflow.DataType dtype = 3;</code>
+     */
+    public org.tensorflow.proto.framework.DataType getDtype() {
+      @SuppressWarnings("deprecation")
+      org.tensorflow.proto.framework.DataType result = org.tensorflow.proto.framework.DataType.valueOf(dtype_);
+      return result == null ? org.tensorflow.proto.framework.DataType.UNRECOGNIZED : result;
+    }
+    /**
+     * <code>.tensorflow.DataType dtype = 3;</code>
+     */
+    public Builder setDtype(org.tensorflow.proto.framework.DataType value) {
+      if (value == null) {
+        throw new NullPointerException();
+      }
+      
+      dtype_ = value.getNumber();
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>.tensorflow.DataType dtype = 3;</code>
+     */
+    public Builder clearDtype() {
+      
+      dtype_ = 0;
+      onChanged();
+      return this;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.GraphTransferGraphInputNodeInfo)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.GraphTransferGraphInputNodeInfo)
+  private static final org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfo DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfo();
+  }
+
+  public static org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfo getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<GraphTransferGraphInputNodeInfo>
+      PARSER = new com.google.protobuf.AbstractParser<GraphTransferGraphInputNodeInfo>() {
+    @java.lang.Override
+    public GraphTransferGraphInputNodeInfo parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new GraphTransferGraphInputNodeInfo(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<GraphTransferGraphInputNodeInfo> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<GraphTransferGraphInputNodeInfo> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfo getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphTransferGraphInputNodeInfoOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphTransferGraphInputNodeInfoOrBuilder.java
new file mode 100644
index 00000000000..c121353b111
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphTransferGraphInputNodeInfoOrBuilder.java
@@ -0,0 +1,41 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/framework/graph_transfer_info.proto
+
+package org.tensorflow.proto.framework;
+
+public interface GraphTransferGraphInputNodeInfoOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.GraphTransferGraphInputNodeInfo)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <code>string name = 1;</code>
+   */
+  java.lang.String getName();
+  /**
+   * <code>string name = 1;</code>
+   */
+  com.google.protobuf.ByteString
+      getNameBytes();
+
+  /**
+   * <code>repeated int64 shape = 2;</code>
+   */
+  java.util.List<java.lang.Long> getShapeList();
+  /**
+   * <code>repeated int64 shape = 2;</code>
+   */
+  int getShapeCount();
+  /**
+   * <code>repeated int64 shape = 2;</code>
+   */
+  long getShape(int index);
+
+  /**
+   * <code>.tensorflow.DataType dtype = 3;</code>
+   */
+  int getDtypeValue();
+  /**
+   * <code>.tensorflow.DataType dtype = 3;</code>
+   */
+  org.tensorflow.proto.framework.DataType getDtype();
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphTransferGraphOutputNodeInfo.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphTransferGraphOutputNodeInfo.java
new file mode 100644
index 00000000000..78bf270776e
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphTransferGraphOutputNodeInfo.java
@@ -0,0 +1,794 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/framework/graph_transfer_info.proto
+
+package org.tensorflow.proto.framework;
+
+/**
+ * Protobuf type {@code tensorflow.GraphTransferGraphOutputNodeInfo}
+ */
+public  final class GraphTransferGraphOutputNodeInfo extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.GraphTransferGraphOutputNodeInfo)
+    GraphTransferGraphOutputNodeInfoOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use GraphTransferGraphOutputNodeInfo.newBuilder() to construct.
+  private GraphTransferGraphOutputNodeInfo(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private GraphTransferGraphOutputNodeInfo() {
+    name_ = "";
+    shape_ = emptyLongList();
+    dtype_ = 0;
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new GraphTransferGraphOutputNodeInfo();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private GraphTransferGraphOutputNodeInfo(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    int mutable_bitField0_ = 0;
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 10: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            name_ = s;
+            break;
+          }
+          case 16: {
+            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
+              shape_ = newLongList();
+              mutable_bitField0_ |= 0x00000001;
+            }
+            shape_.addLong(input.readInt64());
+            break;
+          }
+          case 18: {
+            int length = input.readRawVarint32();
+            int limit = input.pushLimit(length);
+            if (!((mutable_bitField0_ & 0x00000001) != 0) && input.getBytesUntilLimit() > 0) {
+              shape_ = newLongList();
+              mutable_bitField0_ |= 0x00000001;
+            }
+            while (input.getBytesUntilLimit() > 0) {
+              shape_.addLong(input.readInt64());
+            }
+            input.popLimit(limit);
+            break;
+          }
+          case 24: {
+            int rawValue = input.readEnum();
+
+            dtype_ = rawValue;
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      if (((mutable_bitField0_ & 0x00000001) != 0)) {
+        shape_.makeImmutable(); // C
+      }
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.framework.GraphTransferInfoProto.internal_static_tensorflow_GraphTransferGraphOutputNodeInfo_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.framework.GraphTransferInfoProto.internal_static_tensorflow_GraphTransferGraphOutputNodeInfo_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfo.class, org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfo.Builder.class);
+  }
+
+  public static final int NAME_FIELD_NUMBER = 1;
+  private volatile java.lang.Object name_;
+  /**
+   * <code>string name = 1;</code>
+   */
+  public java.lang.String getName() {
+    java.lang.Object ref = name_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      name_ = s;
+      return s;
+    }
+  }
+  /**
+   * <code>string name = 1;</code>
+   */
+  public com.google.protobuf.ByteString
+      getNameBytes() {
+    java.lang.Object ref = name_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      name_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int SHAPE_FIELD_NUMBER = 2;
+  private com.google.protobuf.Internal.LongList shape_;
+  /**
+   * <code>repeated int64 shape = 2;</code>
+   */
+  public java.util.List<java.lang.Long>
+      getShapeList() {
+    return shape_;
+  }
+  /**
+   * <code>repeated int64 shape = 2;</code>
+   */
+  public int getShapeCount() {
+    return shape_.size();
+  }
+  /**
+   * <code>repeated int64 shape = 2;</code>
+   */
+  public long getShape(int index) {
+    return shape_.getLong(index);
+  }
+  private int shapeMemoizedSerializedSize = -1;
+
+  public static final int DTYPE_FIELD_NUMBER = 3;
+  private int dtype_;
+  /**
+   * <code>.tensorflow.DataType dtype = 3;</code>
+   */
+  public int getDtypeValue() {
+    return dtype_;
+  }
+  /**
+   * <code>.tensorflow.DataType dtype = 3;</code>
+   */
+  public org.tensorflow.proto.framework.DataType getDtype() {
+    @SuppressWarnings("deprecation")
+    org.tensorflow.proto.framework.DataType result = org.tensorflow.proto.framework.DataType.valueOf(dtype_);
+    return result == null ? org.tensorflow.proto.framework.DataType.UNRECOGNIZED : result;
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    getSerializedSize();
+    if (!getNameBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 1, name_);
+    }
+    if (getShapeList().size() > 0) {
+      output.writeUInt32NoTag(18);
+      output.writeUInt32NoTag(shapeMemoizedSerializedSize);
+    }
+    for (int i = 0; i < shape_.size(); i++) {
+      output.writeInt64NoTag(shape_.getLong(i));
+    }
+    if (dtype_ != org.tensorflow.proto.framework.DataType.DT_INVALID.getNumber()) {
+      output.writeEnum(3, dtype_);
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (!getNameBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, name_);
+    }
+    {
+      int dataSize = 0;
+      for (int i = 0; i < shape_.size(); i++) {
+        dataSize += com.google.protobuf.CodedOutputStream
+          .computeInt64SizeNoTag(shape_.getLong(i));
+      }
+      size += dataSize;
+      if (!getShapeList().isEmpty()) {
+        size += 1;
+        size += com.google.protobuf.CodedOutputStream
+            .computeInt32SizeNoTag(dataSize);
+      }
+      shapeMemoizedSerializedSize = dataSize;
+    }
+    if (dtype_ != org.tensorflow.proto.framework.DataType.DT_INVALID.getNumber()) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeEnumSize(3, dtype_);
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfo)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfo other = (org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfo) obj;
+
+    if (!getName()
+        .equals(other.getName())) return false;
+    if (!getShapeList()
+        .equals(other.getShapeList())) return false;
+    if (dtype_ != other.dtype_) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    hash = (37 * hash) + NAME_FIELD_NUMBER;
+    hash = (53 * hash) + getName().hashCode();
+    if (getShapeCount() > 0) {
+      hash = (37 * hash) + SHAPE_FIELD_NUMBER;
+      hash = (53 * hash) + getShapeList().hashCode();
+    }
+    hash = (37 * hash) + DTYPE_FIELD_NUMBER;
+    hash = (53 * hash) + dtype_;
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfo parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfo parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfo parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfo parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfo parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfo parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfo parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfo parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfo parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfo parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfo parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfo parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfo prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * Protobuf type {@code tensorflow.GraphTransferGraphOutputNodeInfo}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.GraphTransferGraphOutputNodeInfo)
+      org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfoOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.GraphTransferInfoProto.internal_static_tensorflow_GraphTransferGraphOutputNodeInfo_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.GraphTransferInfoProto.internal_static_tensorflow_GraphTransferGraphOutputNodeInfo_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfo.class, org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfo.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfo.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      name_ = "";
+
+      shape_ = emptyLongList();
+      bitField0_ = (bitField0_ & ~0x00000001);
+      dtype_ = 0;
+
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.framework.GraphTransferInfoProto.internal_static_tensorflow_GraphTransferGraphOutputNodeInfo_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfo getDefaultInstanceForType() {
+      return org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfo.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfo build() {
+      org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfo result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfo buildPartial() {
+      org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfo result = new org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfo(this);
+      int from_bitField0_ = bitField0_;
+      result.name_ = name_;
+      if (((bitField0_ & 0x00000001) != 0)) {
+        shape_.makeImmutable();
+        bitField0_ = (bitField0_ & ~0x00000001);
+      }
+      result.shape_ = shape_;
+      result.dtype_ = dtype_;
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfo) {
+        return mergeFrom((org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfo)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfo other) {
+      if (other == org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfo.getDefaultInstance()) return this;
+      if (!other.getName().isEmpty()) {
+        name_ = other.name_;
+        onChanged();
+      }
+      if (!other.shape_.isEmpty()) {
+        if (shape_.isEmpty()) {
+          shape_ = other.shape_;
+          bitField0_ = (bitField0_ & ~0x00000001);
+        } else {
+          ensureShapeIsMutable();
+          shape_.addAll(other.shape_);
+        }
+        onChanged();
+      }
+      if (other.dtype_ != 0) {
+        setDtypeValue(other.getDtypeValue());
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfo parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfo) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+    private int bitField0_;
+
+    private java.lang.Object name_ = "";
+    /**
+     * <code>string name = 1;</code>
+     */
+    public java.lang.String getName() {
+      java.lang.Object ref = name_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        name_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <code>string name = 1;</code>
+     */
+    public com.google.protobuf.ByteString
+        getNameBytes() {
+      java.lang.Object ref = name_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        name_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <code>string name = 1;</code>
+     */
+    public Builder setName(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      name_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>string name = 1;</code>
+     */
+    public Builder clearName() {
+      
+      name_ = getDefaultInstance().getName();
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>string name = 1;</code>
+     */
+    public Builder setNameBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      name_ = value;
+      onChanged();
+      return this;
+    }
+
+    private com.google.protobuf.Internal.LongList shape_ = emptyLongList();
+    private void ensureShapeIsMutable() {
+      if (!((bitField0_ & 0x00000001) != 0)) {
+        shape_ = mutableCopy(shape_);
+        bitField0_ |= 0x00000001;
+       }
+    }
+    /**
+     * <code>repeated int64 shape = 2;</code>
+     */
+    public java.util.List<java.lang.Long>
+        getShapeList() {
+      return ((bitField0_ & 0x00000001) != 0) ?
+               java.util.Collections.unmodifiableList(shape_) : shape_;
+    }
+    /**
+     * <code>repeated int64 shape = 2;</code>
+     */
+    public int getShapeCount() {
+      return shape_.size();
+    }
+    /**
+     * <code>repeated int64 shape = 2;</code>
+     */
+    public long getShape(int index) {
+      return shape_.getLong(index);
+    }
+    /**
+     * <code>repeated int64 shape = 2;</code>
+     */
+    public Builder setShape(
+        int index, long value) {
+      ensureShapeIsMutable();
+      shape_.setLong(index, value);
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>repeated int64 shape = 2;</code>
+     */
+    public Builder addShape(long value) {
+      ensureShapeIsMutable();
+      shape_.addLong(value);
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>repeated int64 shape = 2;</code>
+     */
+    public Builder addAllShape(
+        java.lang.Iterable<? extends java.lang.Long> values) {
+      ensureShapeIsMutable();
+      com.google.protobuf.AbstractMessageLite.Builder.addAll(
+          values, shape_);
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>repeated int64 shape = 2;</code>
+     */
+    public Builder clearShape() {
+      shape_ = emptyLongList();
+      bitField0_ = (bitField0_ & ~0x00000001);
+      onChanged();
+      return this;
+    }
+
+    private int dtype_ = 0;
+    /**
+     * <code>.tensorflow.DataType dtype = 3;</code>
+     */
+    public int getDtypeValue() {
+      return dtype_;
+    }
+    /**
+     * <code>.tensorflow.DataType dtype = 3;</code>
+     */
+    public Builder setDtypeValue(int value) {
+      dtype_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>.tensorflow.DataType dtype = 3;</code>
+     */
+    public org.tensorflow.proto.framework.DataType getDtype() {
+      @SuppressWarnings("deprecation")
+      org.tensorflow.proto.framework.DataType result = org.tensorflow.proto.framework.DataType.valueOf(dtype_);
+      return result == null ? org.tensorflow.proto.framework.DataType.UNRECOGNIZED : result;
+    }
+    /**
+     * <code>.tensorflow.DataType dtype = 3;</code>
+     */
+    public Builder setDtype(org.tensorflow.proto.framework.DataType value) {
+      if (value == null) {
+        throw new NullPointerException();
+      }
+      
+      dtype_ = value.getNumber();
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>.tensorflow.DataType dtype = 3;</code>
+     */
+    public Builder clearDtype() {
+      
+      dtype_ = 0;
+      onChanged();
+      return this;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.GraphTransferGraphOutputNodeInfo)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.GraphTransferGraphOutputNodeInfo)
+  private static final org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfo DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfo();
+  }
+
+  public static org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfo getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<GraphTransferGraphOutputNodeInfo>
+      PARSER = new com.google.protobuf.AbstractParser<GraphTransferGraphOutputNodeInfo>() {
+    @java.lang.Override
+    public GraphTransferGraphOutputNodeInfo parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new GraphTransferGraphOutputNodeInfo(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<GraphTransferGraphOutputNodeInfo> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<GraphTransferGraphOutputNodeInfo> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfo getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphTransferGraphOutputNodeInfoOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphTransferGraphOutputNodeInfoOrBuilder.java
new file mode 100644
index 00000000000..3fcf3f3ae7e
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphTransferGraphOutputNodeInfoOrBuilder.java
@@ -0,0 +1,41 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/framework/graph_transfer_info.proto
+
+package org.tensorflow.proto.framework;
+
+public interface GraphTransferGraphOutputNodeInfoOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.GraphTransferGraphOutputNodeInfo)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <code>string name = 1;</code>
+   */
+  java.lang.String getName();
+  /**
+   * <code>string name = 1;</code>
+   */
+  com.google.protobuf.ByteString
+      getNameBytes();
+
+  /**
+   * <code>repeated int64 shape = 2;</code>
+   */
+  java.util.List<java.lang.Long> getShapeList();
+  /**
+   * <code>repeated int64 shape = 2;</code>
+   */
+  int getShapeCount();
+  /**
+   * <code>repeated int64 shape = 2;</code>
+   */
+  long getShape(int index);
+
+  /**
+   * <code>.tensorflow.DataType dtype = 3;</code>
+   */
+  int getDtypeValue();
+  /**
+   * <code>.tensorflow.DataType dtype = 3;</code>
+   */
+  org.tensorflow.proto.framework.DataType getDtype();
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphTransferInfo.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphTransferInfo.java
new file mode 100644
index 00000000000..f9d53390a8b
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphTransferInfo.java
@@ -0,0 +1,2795 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/framework/graph_transfer_info.proto
+
+package org.tensorflow.proto.framework;
+
+/**
+ * <pre>
+ * Protocol buffer representing a handle to a tensorflow resource. Handles are
+ * not valid across executions, but can be serialized back and forth from within
+ * a single run.
+ * </pre>
+ *
+ * Protobuf type {@code tensorflow.GraphTransferInfo}
+ */
+public  final class GraphTransferInfo extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.GraphTransferInfo)
+    GraphTransferInfoOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use GraphTransferInfo.newBuilder() to construct.
+  private GraphTransferInfo(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private GraphTransferInfo() {
+    nodeInfo_ = java.util.Collections.emptyList();
+    constNodeInfo_ = java.util.Collections.emptyList();
+    nodeInputInfo_ = java.util.Collections.emptyList();
+    nodeOutputInfo_ = java.util.Collections.emptyList();
+    graphInputNodeInfo_ = java.util.Collections.emptyList();
+    graphOutputNodeInfo_ = java.util.Collections.emptyList();
+    destination_ = 0;
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new GraphTransferInfo();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private GraphTransferInfo(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    int mutable_bitField0_ = 0;
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 10: {
+            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
+              nodeInfo_ = new java.util.ArrayList<org.tensorflow.proto.framework.GraphTransferNodeInfo>();
+              mutable_bitField0_ |= 0x00000001;
+            }
+            nodeInfo_.add(
+                input.readMessage(org.tensorflow.proto.framework.GraphTransferNodeInfo.parser(), extensionRegistry));
+            break;
+          }
+          case 18: {
+            if (!((mutable_bitField0_ & 0x00000002) != 0)) {
+              constNodeInfo_ = new java.util.ArrayList<org.tensorflow.proto.framework.GraphTransferConstNodeInfo>();
+              mutable_bitField0_ |= 0x00000002;
+            }
+            constNodeInfo_.add(
+                input.readMessage(org.tensorflow.proto.framework.GraphTransferConstNodeInfo.parser(), extensionRegistry));
+            break;
+          }
+          case 26: {
+            if (!((mutable_bitField0_ & 0x00000004) != 0)) {
+              nodeInputInfo_ = new java.util.ArrayList<org.tensorflow.proto.framework.GraphTransferNodeInputInfo>();
+              mutable_bitField0_ |= 0x00000004;
+            }
+            nodeInputInfo_.add(
+                input.readMessage(org.tensorflow.proto.framework.GraphTransferNodeInputInfo.parser(), extensionRegistry));
+            break;
+          }
+          case 34: {
+            if (!((mutable_bitField0_ & 0x00000008) != 0)) {
+              nodeOutputInfo_ = new java.util.ArrayList<org.tensorflow.proto.framework.GraphTransferNodeOutputInfo>();
+              mutable_bitField0_ |= 0x00000008;
+            }
+            nodeOutputInfo_.add(
+                input.readMessage(org.tensorflow.proto.framework.GraphTransferNodeOutputInfo.parser(), extensionRegistry));
+            break;
+          }
+          case 42: {
+            if (!((mutable_bitField0_ & 0x00000010) != 0)) {
+              graphInputNodeInfo_ = new java.util.ArrayList<org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfo>();
+              mutable_bitField0_ |= 0x00000010;
+            }
+            graphInputNodeInfo_.add(
+                input.readMessage(org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfo.parser(), extensionRegistry));
+            break;
+          }
+          case 50: {
+            if (!((mutable_bitField0_ & 0x00000020) != 0)) {
+              graphOutputNodeInfo_ = new java.util.ArrayList<org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfo>();
+              mutable_bitField0_ |= 0x00000020;
+            }
+            graphOutputNodeInfo_.add(
+                input.readMessage(org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfo.parser(), extensionRegistry));
+            break;
+          }
+          case 56: {
+            int rawValue = input.readEnum();
+
+            destination_ = rawValue;
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      if (((mutable_bitField0_ & 0x00000001) != 0)) {
+        nodeInfo_ = java.util.Collections.unmodifiableList(nodeInfo_);
+      }
+      if (((mutable_bitField0_ & 0x00000002) != 0)) {
+        constNodeInfo_ = java.util.Collections.unmodifiableList(constNodeInfo_);
+      }
+      if (((mutable_bitField0_ & 0x00000004) != 0)) {
+        nodeInputInfo_ = java.util.Collections.unmodifiableList(nodeInputInfo_);
+      }
+      if (((mutable_bitField0_ & 0x00000008) != 0)) {
+        nodeOutputInfo_ = java.util.Collections.unmodifiableList(nodeOutputInfo_);
+      }
+      if (((mutable_bitField0_ & 0x00000010) != 0)) {
+        graphInputNodeInfo_ = java.util.Collections.unmodifiableList(graphInputNodeInfo_);
+      }
+      if (((mutable_bitField0_ & 0x00000020) != 0)) {
+        graphOutputNodeInfo_ = java.util.Collections.unmodifiableList(graphOutputNodeInfo_);
+      }
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.framework.GraphTransferInfoProto.internal_static_tensorflow_GraphTransferInfo_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.framework.GraphTransferInfoProto.internal_static_tensorflow_GraphTransferInfo_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.framework.GraphTransferInfo.class, org.tensorflow.proto.framework.GraphTransferInfo.Builder.class);
+  }
+
+  /**
+   * Protobuf enum {@code tensorflow.GraphTransferInfo.Destination}
+   */
+  public enum Destination
+      implements com.google.protobuf.ProtocolMessageEnum {
+    /**
+     * <code>NOP = 0;</code>
+     */
+    NOP(0),
+    /**
+     * <code>HEXAGON = 1;</code>
+     */
+    HEXAGON(1),
+    UNRECOGNIZED(-1),
+    ;
+
+    /**
+     * <code>NOP = 0;</code>
+     */
+    public static final int NOP_VALUE = 0;
+    /**
+     * <code>HEXAGON = 1;</code>
+     */
+    public static final int HEXAGON_VALUE = 1;
+
+
+    public final int getNumber() {
+      if (this == UNRECOGNIZED) {
+        throw new java.lang.IllegalArgumentException(
+            "Can't get the number of an unknown enum value.");
+      }
+      return value;
+    }
+
+    /**
+     * @deprecated Use {@link #forNumber(int)} instead.
+     */
+    @java.lang.Deprecated
+    public static Destination valueOf(int value) {
+      return forNumber(value);
+    }
+
+    public static Destination forNumber(int value) {
+      switch (value) {
+        case 0: return NOP;
+        case 1: return HEXAGON;
+        default: return null;
+      }
+    }
+
+    public static com.google.protobuf.Internal.EnumLiteMap<Destination>
+        internalGetValueMap() {
+      return internalValueMap;
+    }
+    private static final com.google.protobuf.Internal.EnumLiteMap<
+        Destination> internalValueMap =
+          new com.google.protobuf.Internal.EnumLiteMap<Destination>() {
+            public Destination findValueByNumber(int number) {
+              return Destination.forNumber(number);
+            }
+          };
+
+    public final com.google.protobuf.Descriptors.EnumValueDescriptor
+        getValueDescriptor() {
+      return getDescriptor().getValues().get(ordinal());
+    }
+    public final com.google.protobuf.Descriptors.EnumDescriptor
+        getDescriptorForType() {
+      return getDescriptor();
+    }
+    public static final com.google.protobuf.Descriptors.EnumDescriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.GraphTransferInfo.getDescriptor().getEnumTypes().get(0);
+    }
+
+    private static final Destination[] VALUES = values();
+
+    public static Destination valueOf(
+        com.google.protobuf.Descriptors.EnumValueDescriptor desc) {
+      if (desc.getType() != getDescriptor()) {
+        throw new java.lang.IllegalArgumentException(
+          "EnumValueDescriptor is not for this type.");
+      }
+      if (desc.getIndex() == -1) {
+        return UNRECOGNIZED;
+      }
+      return VALUES[desc.getIndex()];
+    }
+
+    private final int value;
+
+    private Destination(int value) {
+      this.value = value;
+    }
+
+    // @@protoc_insertion_point(enum_scope:tensorflow.GraphTransferInfo.Destination)
+  }
+
+  public static final int NODE_INFO_FIELD_NUMBER = 1;
+  private java.util.List<org.tensorflow.proto.framework.GraphTransferNodeInfo> nodeInfo_;
+  /**
+   * <code>repeated .tensorflow.GraphTransferNodeInfo node_info = 1;</code>
+   */
+  public java.util.List<org.tensorflow.proto.framework.GraphTransferNodeInfo> getNodeInfoList() {
+    return nodeInfo_;
+  }
+  /**
+   * <code>repeated .tensorflow.GraphTransferNodeInfo node_info = 1;</code>
+   */
+  public java.util.List<? extends org.tensorflow.proto.framework.GraphTransferNodeInfoOrBuilder> 
+      getNodeInfoOrBuilderList() {
+    return nodeInfo_;
+  }
+  /**
+   * <code>repeated .tensorflow.GraphTransferNodeInfo node_info = 1;</code>
+   */
+  public int getNodeInfoCount() {
+    return nodeInfo_.size();
+  }
+  /**
+   * <code>repeated .tensorflow.GraphTransferNodeInfo node_info = 1;</code>
+   */
+  public org.tensorflow.proto.framework.GraphTransferNodeInfo getNodeInfo(int index) {
+    return nodeInfo_.get(index);
+  }
+  /**
+   * <code>repeated .tensorflow.GraphTransferNodeInfo node_info = 1;</code>
+   */
+  public org.tensorflow.proto.framework.GraphTransferNodeInfoOrBuilder getNodeInfoOrBuilder(
+      int index) {
+    return nodeInfo_.get(index);
+  }
+
+  public static final int CONST_NODE_INFO_FIELD_NUMBER = 2;
+  private java.util.List<org.tensorflow.proto.framework.GraphTransferConstNodeInfo> constNodeInfo_;
+  /**
+   * <code>repeated .tensorflow.GraphTransferConstNodeInfo const_node_info = 2;</code>
+   */
+  public java.util.List<org.tensorflow.proto.framework.GraphTransferConstNodeInfo> getConstNodeInfoList() {
+    return constNodeInfo_;
+  }
+  /**
+   * <code>repeated .tensorflow.GraphTransferConstNodeInfo const_node_info = 2;</code>
+   */
+  public java.util.List<? extends org.tensorflow.proto.framework.GraphTransferConstNodeInfoOrBuilder> 
+      getConstNodeInfoOrBuilderList() {
+    return constNodeInfo_;
+  }
+  /**
+   * <code>repeated .tensorflow.GraphTransferConstNodeInfo const_node_info = 2;</code>
+   */
+  public int getConstNodeInfoCount() {
+    return constNodeInfo_.size();
+  }
+  /**
+   * <code>repeated .tensorflow.GraphTransferConstNodeInfo const_node_info = 2;</code>
+   */
+  public org.tensorflow.proto.framework.GraphTransferConstNodeInfo getConstNodeInfo(int index) {
+    return constNodeInfo_.get(index);
+  }
+  /**
+   * <code>repeated .tensorflow.GraphTransferConstNodeInfo const_node_info = 2;</code>
+   */
+  public org.tensorflow.proto.framework.GraphTransferConstNodeInfoOrBuilder getConstNodeInfoOrBuilder(
+      int index) {
+    return constNodeInfo_.get(index);
+  }
+
+  public static final int NODE_INPUT_INFO_FIELD_NUMBER = 3;
+  private java.util.List<org.tensorflow.proto.framework.GraphTransferNodeInputInfo> nodeInputInfo_;
+  /**
+   * <code>repeated .tensorflow.GraphTransferNodeInputInfo node_input_info = 3;</code>
+   */
+  public java.util.List<org.tensorflow.proto.framework.GraphTransferNodeInputInfo> getNodeInputInfoList() {
+    return nodeInputInfo_;
+  }
+  /**
+   * <code>repeated .tensorflow.GraphTransferNodeInputInfo node_input_info = 3;</code>
+   */
+  public java.util.List<? extends org.tensorflow.proto.framework.GraphTransferNodeInputInfoOrBuilder> 
+      getNodeInputInfoOrBuilderList() {
+    return nodeInputInfo_;
+  }
+  /**
+   * <code>repeated .tensorflow.GraphTransferNodeInputInfo node_input_info = 3;</code>
+   */
+  public int getNodeInputInfoCount() {
+    return nodeInputInfo_.size();
+  }
+  /**
+   * <code>repeated .tensorflow.GraphTransferNodeInputInfo node_input_info = 3;</code>
+   */
+  public org.tensorflow.proto.framework.GraphTransferNodeInputInfo getNodeInputInfo(int index) {
+    return nodeInputInfo_.get(index);
+  }
+  /**
+   * <code>repeated .tensorflow.GraphTransferNodeInputInfo node_input_info = 3;</code>
+   */
+  public org.tensorflow.proto.framework.GraphTransferNodeInputInfoOrBuilder getNodeInputInfoOrBuilder(
+      int index) {
+    return nodeInputInfo_.get(index);
+  }
+
+  public static final int NODE_OUTPUT_INFO_FIELD_NUMBER = 4;
+  private java.util.List<org.tensorflow.proto.framework.GraphTransferNodeOutputInfo> nodeOutputInfo_;
+  /**
+   * <code>repeated .tensorflow.GraphTransferNodeOutputInfo node_output_info = 4;</code>
+   */
+  public java.util.List<org.tensorflow.proto.framework.GraphTransferNodeOutputInfo> getNodeOutputInfoList() {
+    return nodeOutputInfo_;
+  }
+  /**
+   * <code>repeated .tensorflow.GraphTransferNodeOutputInfo node_output_info = 4;</code>
+   */
+  public java.util.List<? extends org.tensorflow.proto.framework.GraphTransferNodeOutputInfoOrBuilder> 
+      getNodeOutputInfoOrBuilderList() {
+    return nodeOutputInfo_;
+  }
+  /**
+   * <code>repeated .tensorflow.GraphTransferNodeOutputInfo node_output_info = 4;</code>
+   */
+  public int getNodeOutputInfoCount() {
+    return nodeOutputInfo_.size();
+  }
+  /**
+   * <code>repeated .tensorflow.GraphTransferNodeOutputInfo node_output_info = 4;</code>
+   */
+  public org.tensorflow.proto.framework.GraphTransferNodeOutputInfo getNodeOutputInfo(int index) {
+    return nodeOutputInfo_.get(index);
+  }
+  /**
+   * <code>repeated .tensorflow.GraphTransferNodeOutputInfo node_output_info = 4;</code>
+   */
+  public org.tensorflow.proto.framework.GraphTransferNodeOutputInfoOrBuilder getNodeOutputInfoOrBuilder(
+      int index) {
+    return nodeOutputInfo_.get(index);
+  }
+
+  public static final int GRAPH_INPUT_NODE_INFO_FIELD_NUMBER = 5;
+  private java.util.List<org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfo> graphInputNodeInfo_;
+  /**
+   * <pre>
+   * Input Node parameters of transferred graph
+   * </pre>
+   *
+   * <code>repeated .tensorflow.GraphTransferGraphInputNodeInfo graph_input_node_info = 5;</code>
+   */
+  public java.util.List<org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfo> getGraphInputNodeInfoList() {
+    return graphInputNodeInfo_;
+  }
+  /**
+   * <pre>
+   * Input Node parameters of transferred graph
+   * </pre>
+   *
+   * <code>repeated .tensorflow.GraphTransferGraphInputNodeInfo graph_input_node_info = 5;</code>
+   */
+  public java.util.List<? extends org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfoOrBuilder> 
+      getGraphInputNodeInfoOrBuilderList() {
+    return graphInputNodeInfo_;
+  }
+  /**
+   * <pre>
+   * Input Node parameters of transferred graph
+   * </pre>
+   *
+   * <code>repeated .tensorflow.GraphTransferGraphInputNodeInfo graph_input_node_info = 5;</code>
+   */
+  public int getGraphInputNodeInfoCount() {
+    return graphInputNodeInfo_.size();
+  }
+  /**
+   * <pre>
+   * Input Node parameters of transferred graph
+   * </pre>
+   *
+   * <code>repeated .tensorflow.GraphTransferGraphInputNodeInfo graph_input_node_info = 5;</code>
+   */
+  public org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfo getGraphInputNodeInfo(int index) {
+    return graphInputNodeInfo_.get(index);
+  }
+  /**
+   * <pre>
+   * Input Node parameters of transferred graph
+   * </pre>
+   *
+   * <code>repeated .tensorflow.GraphTransferGraphInputNodeInfo graph_input_node_info = 5;</code>
+   */
+  public org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfoOrBuilder getGraphInputNodeInfoOrBuilder(
+      int index) {
+    return graphInputNodeInfo_.get(index);
+  }
+
+  public static final int GRAPH_OUTPUT_NODE_INFO_FIELD_NUMBER = 6;
+  private java.util.List<org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfo> graphOutputNodeInfo_;
+  /**
+   * <code>repeated .tensorflow.GraphTransferGraphOutputNodeInfo graph_output_node_info = 6;</code>
+   */
+  public java.util.List<org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfo> getGraphOutputNodeInfoList() {
+    return graphOutputNodeInfo_;
+  }
+  /**
+   * <code>repeated .tensorflow.GraphTransferGraphOutputNodeInfo graph_output_node_info = 6;</code>
+   */
+  public java.util.List<? extends org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfoOrBuilder> 
+      getGraphOutputNodeInfoOrBuilderList() {
+    return graphOutputNodeInfo_;
+  }
+  /**
+   * <code>repeated .tensorflow.GraphTransferGraphOutputNodeInfo graph_output_node_info = 6;</code>
+   */
+  public int getGraphOutputNodeInfoCount() {
+    return graphOutputNodeInfo_.size();
+  }
+  /**
+   * <code>repeated .tensorflow.GraphTransferGraphOutputNodeInfo graph_output_node_info = 6;</code>
+   */
+  public org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfo getGraphOutputNodeInfo(int index) {
+    return graphOutputNodeInfo_.get(index);
+  }
+  /**
+   * <code>repeated .tensorflow.GraphTransferGraphOutputNodeInfo graph_output_node_info = 6;</code>
+   */
+  public org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfoOrBuilder getGraphOutputNodeInfoOrBuilder(
+      int index) {
+    return graphOutputNodeInfo_.get(index);
+  }
+
+  public static final int DESTINATION_FIELD_NUMBER = 7;
+  private int destination_;
+  /**
+   * <pre>
+   * Destination of graph transfer
+   * </pre>
+   *
+   * <code>.tensorflow.GraphTransferInfo.Destination destination = 7;</code>
+   */
+  public int getDestinationValue() {
+    return destination_;
+  }
+  /**
+   * <pre>
+   * Destination of graph transfer
+   * </pre>
+   *
+   * <code>.tensorflow.GraphTransferInfo.Destination destination = 7;</code>
+   */
+  public org.tensorflow.proto.framework.GraphTransferInfo.Destination getDestination() {
+    @SuppressWarnings("deprecation")
+    org.tensorflow.proto.framework.GraphTransferInfo.Destination result = org.tensorflow.proto.framework.GraphTransferInfo.Destination.valueOf(destination_);
+    return result == null ? org.tensorflow.proto.framework.GraphTransferInfo.Destination.UNRECOGNIZED : result;
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    for (int i = 0; i < nodeInfo_.size(); i++) {
+      output.writeMessage(1, nodeInfo_.get(i));
+    }
+    for (int i = 0; i < constNodeInfo_.size(); i++) {
+      output.writeMessage(2, constNodeInfo_.get(i));
+    }
+    for (int i = 0; i < nodeInputInfo_.size(); i++) {
+      output.writeMessage(3, nodeInputInfo_.get(i));
+    }
+    for (int i = 0; i < nodeOutputInfo_.size(); i++) {
+      output.writeMessage(4, nodeOutputInfo_.get(i));
+    }
+    for (int i = 0; i < graphInputNodeInfo_.size(); i++) {
+      output.writeMessage(5, graphInputNodeInfo_.get(i));
+    }
+    for (int i = 0; i < graphOutputNodeInfo_.size(); i++) {
+      output.writeMessage(6, graphOutputNodeInfo_.get(i));
+    }
+    if (destination_ != org.tensorflow.proto.framework.GraphTransferInfo.Destination.NOP.getNumber()) {
+      output.writeEnum(7, destination_);
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    for (int i = 0; i < nodeInfo_.size(); i++) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(1, nodeInfo_.get(i));
+    }
+    for (int i = 0; i < constNodeInfo_.size(); i++) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(2, constNodeInfo_.get(i));
+    }
+    for (int i = 0; i < nodeInputInfo_.size(); i++) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(3, nodeInputInfo_.get(i));
+    }
+    for (int i = 0; i < nodeOutputInfo_.size(); i++) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(4, nodeOutputInfo_.get(i));
+    }
+    for (int i = 0; i < graphInputNodeInfo_.size(); i++) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(5, graphInputNodeInfo_.get(i));
+    }
+    for (int i = 0; i < graphOutputNodeInfo_.size(); i++) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(6, graphOutputNodeInfo_.get(i));
+    }
+    if (destination_ != org.tensorflow.proto.framework.GraphTransferInfo.Destination.NOP.getNumber()) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeEnumSize(7, destination_);
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.framework.GraphTransferInfo)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.framework.GraphTransferInfo other = (org.tensorflow.proto.framework.GraphTransferInfo) obj;
+
+    if (!getNodeInfoList()
+        .equals(other.getNodeInfoList())) return false;
+    if (!getConstNodeInfoList()
+        .equals(other.getConstNodeInfoList())) return false;
+    if (!getNodeInputInfoList()
+        .equals(other.getNodeInputInfoList())) return false;
+    if (!getNodeOutputInfoList()
+        .equals(other.getNodeOutputInfoList())) return false;
+    if (!getGraphInputNodeInfoList()
+        .equals(other.getGraphInputNodeInfoList())) return false;
+    if (!getGraphOutputNodeInfoList()
+        .equals(other.getGraphOutputNodeInfoList())) return false;
+    if (destination_ != other.destination_) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    if (getNodeInfoCount() > 0) {
+      hash = (37 * hash) + NODE_INFO_FIELD_NUMBER;
+      hash = (53 * hash) + getNodeInfoList().hashCode();
+    }
+    if (getConstNodeInfoCount() > 0) {
+      hash = (37 * hash) + CONST_NODE_INFO_FIELD_NUMBER;
+      hash = (53 * hash) + getConstNodeInfoList().hashCode();
+    }
+    if (getNodeInputInfoCount() > 0) {
+      hash = (37 * hash) + NODE_INPUT_INFO_FIELD_NUMBER;
+      hash = (53 * hash) + getNodeInputInfoList().hashCode();
+    }
+    if (getNodeOutputInfoCount() > 0) {
+      hash = (37 * hash) + NODE_OUTPUT_INFO_FIELD_NUMBER;
+      hash = (53 * hash) + getNodeOutputInfoList().hashCode();
+    }
+    if (getGraphInputNodeInfoCount() > 0) {
+      hash = (37 * hash) + GRAPH_INPUT_NODE_INFO_FIELD_NUMBER;
+      hash = (53 * hash) + getGraphInputNodeInfoList().hashCode();
+    }
+    if (getGraphOutputNodeInfoCount() > 0) {
+      hash = (37 * hash) + GRAPH_OUTPUT_NODE_INFO_FIELD_NUMBER;
+      hash = (53 * hash) + getGraphOutputNodeInfoList().hashCode();
+    }
+    hash = (37 * hash) + DESTINATION_FIELD_NUMBER;
+    hash = (53 * hash) + destination_;
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.framework.GraphTransferInfo parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.GraphTransferInfo parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.GraphTransferInfo parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.GraphTransferInfo parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.GraphTransferInfo parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.GraphTransferInfo parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.GraphTransferInfo parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.GraphTransferInfo parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.GraphTransferInfo parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.GraphTransferInfo parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.GraphTransferInfo parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.GraphTransferInfo parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.framework.GraphTransferInfo prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * <pre>
+   * Protocol buffer representing a handle to a tensorflow resource. Handles are
+   * not valid across executions, but can be serialized back and forth from within
+   * a single run.
+   * </pre>
+   *
+   * Protobuf type {@code tensorflow.GraphTransferInfo}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.GraphTransferInfo)
+      org.tensorflow.proto.framework.GraphTransferInfoOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.GraphTransferInfoProto.internal_static_tensorflow_GraphTransferInfo_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.GraphTransferInfoProto.internal_static_tensorflow_GraphTransferInfo_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.GraphTransferInfo.class, org.tensorflow.proto.framework.GraphTransferInfo.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.framework.GraphTransferInfo.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+        getNodeInfoFieldBuilder();
+        getConstNodeInfoFieldBuilder();
+        getNodeInputInfoFieldBuilder();
+        getNodeOutputInfoFieldBuilder();
+        getGraphInputNodeInfoFieldBuilder();
+        getGraphOutputNodeInfoFieldBuilder();
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      if (nodeInfoBuilder_ == null) {
+        nodeInfo_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000001);
+      } else {
+        nodeInfoBuilder_.clear();
+      }
+      if (constNodeInfoBuilder_ == null) {
+        constNodeInfo_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000002);
+      } else {
+        constNodeInfoBuilder_.clear();
+      }
+      if (nodeInputInfoBuilder_ == null) {
+        nodeInputInfo_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000004);
+      } else {
+        nodeInputInfoBuilder_.clear();
+      }
+      if (nodeOutputInfoBuilder_ == null) {
+        nodeOutputInfo_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000008);
+      } else {
+        nodeOutputInfoBuilder_.clear();
+      }
+      if (graphInputNodeInfoBuilder_ == null) {
+        graphInputNodeInfo_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000010);
+      } else {
+        graphInputNodeInfoBuilder_.clear();
+      }
+      if (graphOutputNodeInfoBuilder_ == null) {
+        graphOutputNodeInfo_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000020);
+      } else {
+        graphOutputNodeInfoBuilder_.clear();
+      }
+      destination_ = 0;
+
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.framework.GraphTransferInfoProto.internal_static_tensorflow_GraphTransferInfo_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.GraphTransferInfo getDefaultInstanceForType() {
+      return org.tensorflow.proto.framework.GraphTransferInfo.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.GraphTransferInfo build() {
+      org.tensorflow.proto.framework.GraphTransferInfo result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.GraphTransferInfo buildPartial() {
+      org.tensorflow.proto.framework.GraphTransferInfo result = new org.tensorflow.proto.framework.GraphTransferInfo(this);
+      int from_bitField0_ = bitField0_;
+      if (nodeInfoBuilder_ == null) {
+        if (((bitField0_ & 0x00000001) != 0)) {
+          nodeInfo_ = java.util.Collections.unmodifiableList(nodeInfo_);
+          bitField0_ = (bitField0_ & ~0x00000001);
+        }
+        result.nodeInfo_ = nodeInfo_;
+      } else {
+        result.nodeInfo_ = nodeInfoBuilder_.build();
+      }
+      if (constNodeInfoBuilder_ == null) {
+        if (((bitField0_ & 0x00000002) != 0)) {
+          constNodeInfo_ = java.util.Collections.unmodifiableList(constNodeInfo_);
+          bitField0_ = (bitField0_ & ~0x00000002);
+        }
+        result.constNodeInfo_ = constNodeInfo_;
+      } else {
+        result.constNodeInfo_ = constNodeInfoBuilder_.build();
+      }
+      if (nodeInputInfoBuilder_ == null) {
+        if (((bitField0_ & 0x00000004) != 0)) {
+          nodeInputInfo_ = java.util.Collections.unmodifiableList(nodeInputInfo_);
+          bitField0_ = (bitField0_ & ~0x00000004);
+        }
+        result.nodeInputInfo_ = nodeInputInfo_;
+      } else {
+        result.nodeInputInfo_ = nodeInputInfoBuilder_.build();
+      }
+      if (nodeOutputInfoBuilder_ == null) {
+        if (((bitField0_ & 0x00000008) != 0)) {
+          nodeOutputInfo_ = java.util.Collections.unmodifiableList(nodeOutputInfo_);
+          bitField0_ = (bitField0_ & ~0x00000008);
+        }
+        result.nodeOutputInfo_ = nodeOutputInfo_;
+      } else {
+        result.nodeOutputInfo_ = nodeOutputInfoBuilder_.build();
+      }
+      if (graphInputNodeInfoBuilder_ == null) {
+        if (((bitField0_ & 0x00000010) != 0)) {
+          graphInputNodeInfo_ = java.util.Collections.unmodifiableList(graphInputNodeInfo_);
+          bitField0_ = (bitField0_ & ~0x00000010);
+        }
+        result.graphInputNodeInfo_ = graphInputNodeInfo_;
+      } else {
+        result.graphInputNodeInfo_ = graphInputNodeInfoBuilder_.build();
+      }
+      if (graphOutputNodeInfoBuilder_ == null) {
+        if (((bitField0_ & 0x00000020) != 0)) {
+          graphOutputNodeInfo_ = java.util.Collections.unmodifiableList(graphOutputNodeInfo_);
+          bitField0_ = (bitField0_ & ~0x00000020);
+        }
+        result.graphOutputNodeInfo_ = graphOutputNodeInfo_;
+      } else {
+        result.graphOutputNodeInfo_ = graphOutputNodeInfoBuilder_.build();
+      }
+      result.destination_ = destination_;
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.framework.GraphTransferInfo) {
+        return mergeFrom((org.tensorflow.proto.framework.GraphTransferInfo)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.framework.GraphTransferInfo other) {
+      if (other == org.tensorflow.proto.framework.GraphTransferInfo.getDefaultInstance()) return this;
+      if (nodeInfoBuilder_ == null) {
+        if (!other.nodeInfo_.isEmpty()) {
+          if (nodeInfo_.isEmpty()) {
+            nodeInfo_ = other.nodeInfo_;
+            bitField0_ = (bitField0_ & ~0x00000001);
+          } else {
+            ensureNodeInfoIsMutable();
+            nodeInfo_.addAll(other.nodeInfo_);
+          }
+          onChanged();
+        }
+      } else {
+        if (!other.nodeInfo_.isEmpty()) {
+          if (nodeInfoBuilder_.isEmpty()) {
+            nodeInfoBuilder_.dispose();
+            nodeInfoBuilder_ = null;
+            nodeInfo_ = other.nodeInfo_;
+            bitField0_ = (bitField0_ & ~0x00000001);
+            nodeInfoBuilder_ = 
+              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
+                 getNodeInfoFieldBuilder() : null;
+          } else {
+            nodeInfoBuilder_.addAllMessages(other.nodeInfo_);
+          }
+        }
+      }
+      if (constNodeInfoBuilder_ == null) {
+        if (!other.constNodeInfo_.isEmpty()) {
+          if (constNodeInfo_.isEmpty()) {
+            constNodeInfo_ = other.constNodeInfo_;
+            bitField0_ = (bitField0_ & ~0x00000002);
+          } else {
+            ensureConstNodeInfoIsMutable();
+            constNodeInfo_.addAll(other.constNodeInfo_);
+          }
+          onChanged();
+        }
+      } else {
+        if (!other.constNodeInfo_.isEmpty()) {
+          if (constNodeInfoBuilder_.isEmpty()) {
+            constNodeInfoBuilder_.dispose();
+            constNodeInfoBuilder_ = null;
+            constNodeInfo_ = other.constNodeInfo_;
+            bitField0_ = (bitField0_ & ~0x00000002);
+            constNodeInfoBuilder_ = 
+              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
+                 getConstNodeInfoFieldBuilder() : null;
+          } else {
+            constNodeInfoBuilder_.addAllMessages(other.constNodeInfo_);
+          }
+        }
+      }
+      if (nodeInputInfoBuilder_ == null) {
+        if (!other.nodeInputInfo_.isEmpty()) {
+          if (nodeInputInfo_.isEmpty()) {
+            nodeInputInfo_ = other.nodeInputInfo_;
+            bitField0_ = (bitField0_ & ~0x00000004);
+          } else {
+            ensureNodeInputInfoIsMutable();
+            nodeInputInfo_.addAll(other.nodeInputInfo_);
+          }
+          onChanged();
+        }
+      } else {
+        if (!other.nodeInputInfo_.isEmpty()) {
+          if (nodeInputInfoBuilder_.isEmpty()) {
+            nodeInputInfoBuilder_.dispose();
+            nodeInputInfoBuilder_ = null;
+            nodeInputInfo_ = other.nodeInputInfo_;
+            bitField0_ = (bitField0_ & ~0x00000004);
+            nodeInputInfoBuilder_ = 
+              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
+                 getNodeInputInfoFieldBuilder() : null;
+          } else {
+            nodeInputInfoBuilder_.addAllMessages(other.nodeInputInfo_);
+          }
+        }
+      }
+      if (nodeOutputInfoBuilder_ == null) {
+        if (!other.nodeOutputInfo_.isEmpty()) {
+          if (nodeOutputInfo_.isEmpty()) {
+            nodeOutputInfo_ = other.nodeOutputInfo_;
+            bitField0_ = (bitField0_ & ~0x00000008);
+          } else {
+            ensureNodeOutputInfoIsMutable();
+            nodeOutputInfo_.addAll(other.nodeOutputInfo_);
+          }
+          onChanged();
+        }
+      } else {
+        if (!other.nodeOutputInfo_.isEmpty()) {
+          if (nodeOutputInfoBuilder_.isEmpty()) {
+            nodeOutputInfoBuilder_.dispose();
+            nodeOutputInfoBuilder_ = null;
+            nodeOutputInfo_ = other.nodeOutputInfo_;
+            bitField0_ = (bitField0_ & ~0x00000008);
+            nodeOutputInfoBuilder_ = 
+              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
+                 getNodeOutputInfoFieldBuilder() : null;
+          } else {
+            nodeOutputInfoBuilder_.addAllMessages(other.nodeOutputInfo_);
+          }
+        }
+      }
+      if (graphInputNodeInfoBuilder_ == null) {
+        if (!other.graphInputNodeInfo_.isEmpty()) {
+          if (graphInputNodeInfo_.isEmpty()) {
+            graphInputNodeInfo_ = other.graphInputNodeInfo_;
+            bitField0_ = (bitField0_ & ~0x00000010);
+          } else {
+            ensureGraphInputNodeInfoIsMutable();
+            graphInputNodeInfo_.addAll(other.graphInputNodeInfo_);
+          }
+          onChanged();
+        }
+      } else {
+        if (!other.graphInputNodeInfo_.isEmpty()) {
+          if (graphInputNodeInfoBuilder_.isEmpty()) {
+            graphInputNodeInfoBuilder_.dispose();
+            graphInputNodeInfoBuilder_ = null;
+            graphInputNodeInfo_ = other.graphInputNodeInfo_;
+            bitField0_ = (bitField0_ & ~0x00000010);
+            graphInputNodeInfoBuilder_ = 
+              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
+                 getGraphInputNodeInfoFieldBuilder() : null;
+          } else {
+            graphInputNodeInfoBuilder_.addAllMessages(other.graphInputNodeInfo_);
+          }
+        }
+      }
+      if (graphOutputNodeInfoBuilder_ == null) {
+        if (!other.graphOutputNodeInfo_.isEmpty()) {
+          if (graphOutputNodeInfo_.isEmpty()) {
+            graphOutputNodeInfo_ = other.graphOutputNodeInfo_;
+            bitField0_ = (bitField0_ & ~0x00000020);
+          } else {
+            ensureGraphOutputNodeInfoIsMutable();
+            graphOutputNodeInfo_.addAll(other.graphOutputNodeInfo_);
+          }
+          onChanged();
+        }
+      } else {
+        if (!other.graphOutputNodeInfo_.isEmpty()) {
+          if (graphOutputNodeInfoBuilder_.isEmpty()) {
+            graphOutputNodeInfoBuilder_.dispose();
+            graphOutputNodeInfoBuilder_ = null;
+            graphOutputNodeInfo_ = other.graphOutputNodeInfo_;
+            bitField0_ = (bitField0_ & ~0x00000020);
+            graphOutputNodeInfoBuilder_ = 
+              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
+                 getGraphOutputNodeInfoFieldBuilder() : null;
+          } else {
+            graphOutputNodeInfoBuilder_.addAllMessages(other.graphOutputNodeInfo_);
+          }
+        }
+      }
+      if (other.destination_ != 0) {
+        setDestinationValue(other.getDestinationValue());
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.framework.GraphTransferInfo parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.framework.GraphTransferInfo) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+    private int bitField0_;
+
+    private java.util.List<org.tensorflow.proto.framework.GraphTransferNodeInfo> nodeInfo_ =
+      java.util.Collections.emptyList();
+    private void ensureNodeInfoIsMutable() {
+      if (!((bitField0_ & 0x00000001) != 0)) {
+        nodeInfo_ = new java.util.ArrayList<org.tensorflow.proto.framework.GraphTransferNodeInfo>(nodeInfo_);
+        bitField0_ |= 0x00000001;
+       }
+    }
+
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.framework.GraphTransferNodeInfo, org.tensorflow.proto.framework.GraphTransferNodeInfo.Builder, org.tensorflow.proto.framework.GraphTransferNodeInfoOrBuilder> nodeInfoBuilder_;
+
+    /**
+     * <code>repeated .tensorflow.GraphTransferNodeInfo node_info = 1;</code>
+     */
+    public java.util.List<org.tensorflow.proto.framework.GraphTransferNodeInfo> getNodeInfoList() {
+      if (nodeInfoBuilder_ == null) {
+        return java.util.Collections.unmodifiableList(nodeInfo_);
+      } else {
+        return nodeInfoBuilder_.getMessageList();
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.GraphTransferNodeInfo node_info = 1;</code>
+     */
+    public int getNodeInfoCount() {
+      if (nodeInfoBuilder_ == null) {
+        return nodeInfo_.size();
+      } else {
+        return nodeInfoBuilder_.getCount();
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.GraphTransferNodeInfo node_info = 1;</code>
+     */
+    public org.tensorflow.proto.framework.GraphTransferNodeInfo getNodeInfo(int index) {
+      if (nodeInfoBuilder_ == null) {
+        return nodeInfo_.get(index);
+      } else {
+        return nodeInfoBuilder_.getMessage(index);
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.GraphTransferNodeInfo node_info = 1;</code>
+     */
+    public Builder setNodeInfo(
+        int index, org.tensorflow.proto.framework.GraphTransferNodeInfo value) {
+      if (nodeInfoBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureNodeInfoIsMutable();
+        nodeInfo_.set(index, value);
+        onChanged();
+      } else {
+        nodeInfoBuilder_.setMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.GraphTransferNodeInfo node_info = 1;</code>
+     */
+    public Builder setNodeInfo(
+        int index, org.tensorflow.proto.framework.GraphTransferNodeInfo.Builder builderForValue) {
+      if (nodeInfoBuilder_ == null) {
+        ensureNodeInfoIsMutable();
+        nodeInfo_.set(index, builderForValue.build());
+        onChanged();
+      } else {
+        nodeInfoBuilder_.setMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.GraphTransferNodeInfo node_info = 1;</code>
+     */
+    public Builder addNodeInfo(org.tensorflow.proto.framework.GraphTransferNodeInfo value) {
+      if (nodeInfoBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureNodeInfoIsMutable();
+        nodeInfo_.add(value);
+        onChanged();
+      } else {
+        nodeInfoBuilder_.addMessage(value);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.GraphTransferNodeInfo node_info = 1;</code>
+     */
+    public Builder addNodeInfo(
+        int index, org.tensorflow.proto.framework.GraphTransferNodeInfo value) {
+      if (nodeInfoBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureNodeInfoIsMutable();
+        nodeInfo_.add(index, value);
+        onChanged();
+      } else {
+        nodeInfoBuilder_.addMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.GraphTransferNodeInfo node_info = 1;</code>
+     */
+    public Builder addNodeInfo(
+        org.tensorflow.proto.framework.GraphTransferNodeInfo.Builder builderForValue) {
+      if (nodeInfoBuilder_ == null) {
+        ensureNodeInfoIsMutable();
+        nodeInfo_.add(builderForValue.build());
+        onChanged();
+      } else {
+        nodeInfoBuilder_.addMessage(builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.GraphTransferNodeInfo node_info = 1;</code>
+     */
+    public Builder addNodeInfo(
+        int index, org.tensorflow.proto.framework.GraphTransferNodeInfo.Builder builderForValue) {
+      if (nodeInfoBuilder_ == null) {
+        ensureNodeInfoIsMutable();
+        nodeInfo_.add(index, builderForValue.build());
+        onChanged();
+      } else {
+        nodeInfoBuilder_.addMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.GraphTransferNodeInfo node_info = 1;</code>
+     */
+    public Builder addAllNodeInfo(
+        java.lang.Iterable<? extends org.tensorflow.proto.framework.GraphTransferNodeInfo> values) {
+      if (nodeInfoBuilder_ == null) {
+        ensureNodeInfoIsMutable();
+        com.google.protobuf.AbstractMessageLite.Builder.addAll(
+            values, nodeInfo_);
+        onChanged();
+      } else {
+        nodeInfoBuilder_.addAllMessages(values);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.GraphTransferNodeInfo node_info = 1;</code>
+     */
+    public Builder clearNodeInfo() {
+      if (nodeInfoBuilder_ == null) {
+        nodeInfo_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000001);
+        onChanged();
+      } else {
+        nodeInfoBuilder_.clear();
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.GraphTransferNodeInfo node_info = 1;</code>
+     */
+    public Builder removeNodeInfo(int index) {
+      if (nodeInfoBuilder_ == null) {
+        ensureNodeInfoIsMutable();
+        nodeInfo_.remove(index);
+        onChanged();
+      } else {
+        nodeInfoBuilder_.remove(index);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.GraphTransferNodeInfo node_info = 1;</code>
+     */
+    public org.tensorflow.proto.framework.GraphTransferNodeInfo.Builder getNodeInfoBuilder(
+        int index) {
+      return getNodeInfoFieldBuilder().getBuilder(index);
+    }
+    /**
+     * <code>repeated .tensorflow.GraphTransferNodeInfo node_info = 1;</code>
+     */
+    public org.tensorflow.proto.framework.GraphTransferNodeInfoOrBuilder getNodeInfoOrBuilder(
+        int index) {
+      if (nodeInfoBuilder_ == null) {
+        return nodeInfo_.get(index);  } else {
+        return nodeInfoBuilder_.getMessageOrBuilder(index);
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.GraphTransferNodeInfo node_info = 1;</code>
+     */
+    public java.util.List<? extends org.tensorflow.proto.framework.GraphTransferNodeInfoOrBuilder> 
+         getNodeInfoOrBuilderList() {
+      if (nodeInfoBuilder_ != null) {
+        return nodeInfoBuilder_.getMessageOrBuilderList();
+      } else {
+        return java.util.Collections.unmodifiableList(nodeInfo_);
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.GraphTransferNodeInfo node_info = 1;</code>
+     */
+    public org.tensorflow.proto.framework.GraphTransferNodeInfo.Builder addNodeInfoBuilder() {
+      return getNodeInfoFieldBuilder().addBuilder(
+          org.tensorflow.proto.framework.GraphTransferNodeInfo.getDefaultInstance());
+    }
+    /**
+     * <code>repeated .tensorflow.GraphTransferNodeInfo node_info = 1;</code>
+     */
+    public org.tensorflow.proto.framework.GraphTransferNodeInfo.Builder addNodeInfoBuilder(
+        int index) {
+      return getNodeInfoFieldBuilder().addBuilder(
+          index, org.tensorflow.proto.framework.GraphTransferNodeInfo.getDefaultInstance());
+    }
+    /**
+     * <code>repeated .tensorflow.GraphTransferNodeInfo node_info = 1;</code>
+     */
+    public java.util.List<org.tensorflow.proto.framework.GraphTransferNodeInfo.Builder> 
+         getNodeInfoBuilderList() {
+      return getNodeInfoFieldBuilder().getBuilderList();
+    }
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.framework.GraphTransferNodeInfo, org.tensorflow.proto.framework.GraphTransferNodeInfo.Builder, org.tensorflow.proto.framework.GraphTransferNodeInfoOrBuilder> 
+        getNodeInfoFieldBuilder() {
+      if (nodeInfoBuilder_ == null) {
+        nodeInfoBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
+            org.tensorflow.proto.framework.GraphTransferNodeInfo, org.tensorflow.proto.framework.GraphTransferNodeInfo.Builder, org.tensorflow.proto.framework.GraphTransferNodeInfoOrBuilder>(
+                nodeInfo_,
+                ((bitField0_ & 0x00000001) != 0),
+                getParentForChildren(),
+                isClean());
+        nodeInfo_ = null;
+      }
+      return nodeInfoBuilder_;
+    }
+
+    private java.util.List<org.tensorflow.proto.framework.GraphTransferConstNodeInfo> constNodeInfo_ =
+      java.util.Collections.emptyList();
+    private void ensureConstNodeInfoIsMutable() {
+      if (!((bitField0_ & 0x00000002) != 0)) {
+        constNodeInfo_ = new java.util.ArrayList<org.tensorflow.proto.framework.GraphTransferConstNodeInfo>(constNodeInfo_);
+        bitField0_ |= 0x00000002;
+       }
+    }
+
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.framework.GraphTransferConstNodeInfo, org.tensorflow.proto.framework.GraphTransferConstNodeInfo.Builder, org.tensorflow.proto.framework.GraphTransferConstNodeInfoOrBuilder> constNodeInfoBuilder_;
+
+    /**
+     * <code>repeated .tensorflow.GraphTransferConstNodeInfo const_node_info = 2;</code>
+     */
+    public java.util.List<org.tensorflow.proto.framework.GraphTransferConstNodeInfo> getConstNodeInfoList() {
+      if (constNodeInfoBuilder_ == null) {
+        return java.util.Collections.unmodifiableList(constNodeInfo_);
+      } else {
+        return constNodeInfoBuilder_.getMessageList();
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.GraphTransferConstNodeInfo const_node_info = 2;</code>
+     */
+    public int getConstNodeInfoCount() {
+      if (constNodeInfoBuilder_ == null) {
+        return constNodeInfo_.size();
+      } else {
+        return constNodeInfoBuilder_.getCount();
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.GraphTransferConstNodeInfo const_node_info = 2;</code>
+     */
+    public org.tensorflow.proto.framework.GraphTransferConstNodeInfo getConstNodeInfo(int index) {
+      if (constNodeInfoBuilder_ == null) {
+        return constNodeInfo_.get(index);
+      } else {
+        return constNodeInfoBuilder_.getMessage(index);
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.GraphTransferConstNodeInfo const_node_info = 2;</code>
+     */
+    public Builder setConstNodeInfo(
+        int index, org.tensorflow.proto.framework.GraphTransferConstNodeInfo value) {
+      if (constNodeInfoBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureConstNodeInfoIsMutable();
+        constNodeInfo_.set(index, value);
+        onChanged();
+      } else {
+        constNodeInfoBuilder_.setMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.GraphTransferConstNodeInfo const_node_info = 2;</code>
+     */
+    public Builder setConstNodeInfo(
+        int index, org.tensorflow.proto.framework.GraphTransferConstNodeInfo.Builder builderForValue) {
+      if (constNodeInfoBuilder_ == null) {
+        ensureConstNodeInfoIsMutable();
+        constNodeInfo_.set(index, builderForValue.build());
+        onChanged();
+      } else {
+        constNodeInfoBuilder_.setMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.GraphTransferConstNodeInfo const_node_info = 2;</code>
+     */
+    public Builder addConstNodeInfo(org.tensorflow.proto.framework.GraphTransferConstNodeInfo value) {
+      if (constNodeInfoBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureConstNodeInfoIsMutable();
+        constNodeInfo_.add(value);
+        onChanged();
+      } else {
+        constNodeInfoBuilder_.addMessage(value);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.GraphTransferConstNodeInfo const_node_info = 2;</code>
+     */
+    public Builder addConstNodeInfo(
+        int index, org.tensorflow.proto.framework.GraphTransferConstNodeInfo value) {
+      if (constNodeInfoBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureConstNodeInfoIsMutable();
+        constNodeInfo_.add(index, value);
+        onChanged();
+      } else {
+        constNodeInfoBuilder_.addMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.GraphTransferConstNodeInfo const_node_info = 2;</code>
+     */
+    public Builder addConstNodeInfo(
+        org.tensorflow.proto.framework.GraphTransferConstNodeInfo.Builder builderForValue) {
+      if (constNodeInfoBuilder_ == null) {
+        ensureConstNodeInfoIsMutable();
+        constNodeInfo_.add(builderForValue.build());
+        onChanged();
+      } else {
+        constNodeInfoBuilder_.addMessage(builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.GraphTransferConstNodeInfo const_node_info = 2;</code>
+     */
+    public Builder addConstNodeInfo(
+        int index, org.tensorflow.proto.framework.GraphTransferConstNodeInfo.Builder builderForValue) {
+      if (constNodeInfoBuilder_ == null) {
+        ensureConstNodeInfoIsMutable();
+        constNodeInfo_.add(index, builderForValue.build());
+        onChanged();
+      } else {
+        constNodeInfoBuilder_.addMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.GraphTransferConstNodeInfo const_node_info = 2;</code>
+     */
+    public Builder addAllConstNodeInfo(
+        java.lang.Iterable<? extends org.tensorflow.proto.framework.GraphTransferConstNodeInfo> values) {
+      if (constNodeInfoBuilder_ == null) {
+        ensureConstNodeInfoIsMutable();
+        com.google.protobuf.AbstractMessageLite.Builder.addAll(
+            values, constNodeInfo_);
+        onChanged();
+      } else {
+        constNodeInfoBuilder_.addAllMessages(values);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.GraphTransferConstNodeInfo const_node_info = 2;</code>
+     */
+    public Builder clearConstNodeInfo() {
+      if (constNodeInfoBuilder_ == null) {
+        constNodeInfo_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000002);
+        onChanged();
+      } else {
+        constNodeInfoBuilder_.clear();
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.GraphTransferConstNodeInfo const_node_info = 2;</code>
+     */
+    public Builder removeConstNodeInfo(int index) {
+      if (constNodeInfoBuilder_ == null) {
+        ensureConstNodeInfoIsMutable();
+        constNodeInfo_.remove(index);
+        onChanged();
+      } else {
+        constNodeInfoBuilder_.remove(index);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.GraphTransferConstNodeInfo const_node_info = 2;</code>
+     */
+    public org.tensorflow.proto.framework.GraphTransferConstNodeInfo.Builder getConstNodeInfoBuilder(
+        int index) {
+      return getConstNodeInfoFieldBuilder().getBuilder(index);
+    }
+    /**
+     * <code>repeated .tensorflow.GraphTransferConstNodeInfo const_node_info = 2;</code>
+     */
+    public org.tensorflow.proto.framework.GraphTransferConstNodeInfoOrBuilder getConstNodeInfoOrBuilder(
+        int index) {
+      if (constNodeInfoBuilder_ == null) {
+        return constNodeInfo_.get(index);  } else {
+        return constNodeInfoBuilder_.getMessageOrBuilder(index);
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.GraphTransferConstNodeInfo const_node_info = 2;</code>
+     */
+    public java.util.List<? extends org.tensorflow.proto.framework.GraphTransferConstNodeInfoOrBuilder> 
+         getConstNodeInfoOrBuilderList() {
+      if (constNodeInfoBuilder_ != null) {
+        return constNodeInfoBuilder_.getMessageOrBuilderList();
+      } else {
+        return java.util.Collections.unmodifiableList(constNodeInfo_);
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.GraphTransferConstNodeInfo const_node_info = 2;</code>
+     */
+    public org.tensorflow.proto.framework.GraphTransferConstNodeInfo.Builder addConstNodeInfoBuilder() {
+      return getConstNodeInfoFieldBuilder().addBuilder(
+          org.tensorflow.proto.framework.GraphTransferConstNodeInfo.getDefaultInstance());
+    }
+    /**
+     * <code>repeated .tensorflow.GraphTransferConstNodeInfo const_node_info = 2;</code>
+     */
+    public org.tensorflow.proto.framework.GraphTransferConstNodeInfo.Builder addConstNodeInfoBuilder(
+        int index) {
+      return getConstNodeInfoFieldBuilder().addBuilder(
+          index, org.tensorflow.proto.framework.GraphTransferConstNodeInfo.getDefaultInstance());
+    }
+    /**
+     * <code>repeated .tensorflow.GraphTransferConstNodeInfo const_node_info = 2;</code>
+     */
+    public java.util.List<org.tensorflow.proto.framework.GraphTransferConstNodeInfo.Builder> 
+         getConstNodeInfoBuilderList() {
+      return getConstNodeInfoFieldBuilder().getBuilderList();
+    }
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.framework.GraphTransferConstNodeInfo, org.tensorflow.proto.framework.GraphTransferConstNodeInfo.Builder, org.tensorflow.proto.framework.GraphTransferConstNodeInfoOrBuilder> 
+        getConstNodeInfoFieldBuilder() {
+      if (constNodeInfoBuilder_ == null) {
+        constNodeInfoBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
+            org.tensorflow.proto.framework.GraphTransferConstNodeInfo, org.tensorflow.proto.framework.GraphTransferConstNodeInfo.Builder, org.tensorflow.proto.framework.GraphTransferConstNodeInfoOrBuilder>(
+                constNodeInfo_,
+                ((bitField0_ & 0x00000002) != 0),
+                getParentForChildren(),
+                isClean());
+        constNodeInfo_ = null;
+      }
+      return constNodeInfoBuilder_;
+    }
+
+    private java.util.List<org.tensorflow.proto.framework.GraphTransferNodeInputInfo> nodeInputInfo_ =
+      java.util.Collections.emptyList();
+    private void ensureNodeInputInfoIsMutable() {
+      if (!((bitField0_ & 0x00000004) != 0)) {
+        nodeInputInfo_ = new java.util.ArrayList<org.tensorflow.proto.framework.GraphTransferNodeInputInfo>(nodeInputInfo_);
+        bitField0_ |= 0x00000004;
+       }
+    }
+
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.framework.GraphTransferNodeInputInfo, org.tensorflow.proto.framework.GraphTransferNodeInputInfo.Builder, org.tensorflow.proto.framework.GraphTransferNodeInputInfoOrBuilder> nodeInputInfoBuilder_;
+
+    /**
+     * <code>repeated .tensorflow.GraphTransferNodeInputInfo node_input_info = 3;</code>
+     */
+    public java.util.List<org.tensorflow.proto.framework.GraphTransferNodeInputInfo> getNodeInputInfoList() {
+      if (nodeInputInfoBuilder_ == null) {
+        return java.util.Collections.unmodifiableList(nodeInputInfo_);
+      } else {
+        return nodeInputInfoBuilder_.getMessageList();
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.GraphTransferNodeInputInfo node_input_info = 3;</code>
+     */
+    public int getNodeInputInfoCount() {
+      if (nodeInputInfoBuilder_ == null) {
+        return nodeInputInfo_.size();
+      } else {
+        return nodeInputInfoBuilder_.getCount();
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.GraphTransferNodeInputInfo node_input_info = 3;</code>
+     */
+    public org.tensorflow.proto.framework.GraphTransferNodeInputInfo getNodeInputInfo(int index) {
+      if (nodeInputInfoBuilder_ == null) {
+        return nodeInputInfo_.get(index);
+      } else {
+        return nodeInputInfoBuilder_.getMessage(index);
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.GraphTransferNodeInputInfo node_input_info = 3;</code>
+     */
+    public Builder setNodeInputInfo(
+        int index, org.tensorflow.proto.framework.GraphTransferNodeInputInfo value) {
+      if (nodeInputInfoBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureNodeInputInfoIsMutable();
+        nodeInputInfo_.set(index, value);
+        onChanged();
+      } else {
+        nodeInputInfoBuilder_.setMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.GraphTransferNodeInputInfo node_input_info = 3;</code>
+     */
+    public Builder setNodeInputInfo(
+        int index, org.tensorflow.proto.framework.GraphTransferNodeInputInfo.Builder builderForValue) {
+      if (nodeInputInfoBuilder_ == null) {
+        ensureNodeInputInfoIsMutable();
+        nodeInputInfo_.set(index, builderForValue.build());
+        onChanged();
+      } else {
+        nodeInputInfoBuilder_.setMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.GraphTransferNodeInputInfo node_input_info = 3;</code>
+     */
+    public Builder addNodeInputInfo(org.tensorflow.proto.framework.GraphTransferNodeInputInfo value) {
+      if (nodeInputInfoBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureNodeInputInfoIsMutable();
+        nodeInputInfo_.add(value);
+        onChanged();
+      } else {
+        nodeInputInfoBuilder_.addMessage(value);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.GraphTransferNodeInputInfo node_input_info = 3;</code>
+     */
+    public Builder addNodeInputInfo(
+        int index, org.tensorflow.proto.framework.GraphTransferNodeInputInfo value) {
+      if (nodeInputInfoBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureNodeInputInfoIsMutable();
+        nodeInputInfo_.add(index, value);
+        onChanged();
+      } else {
+        nodeInputInfoBuilder_.addMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.GraphTransferNodeInputInfo node_input_info = 3;</code>
+     */
+    public Builder addNodeInputInfo(
+        org.tensorflow.proto.framework.GraphTransferNodeInputInfo.Builder builderForValue) {
+      if (nodeInputInfoBuilder_ == null) {
+        ensureNodeInputInfoIsMutable();
+        nodeInputInfo_.add(builderForValue.build());
+        onChanged();
+      } else {
+        nodeInputInfoBuilder_.addMessage(builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.GraphTransferNodeInputInfo node_input_info = 3;</code>
+     */
+    public Builder addNodeInputInfo(
+        int index, org.tensorflow.proto.framework.GraphTransferNodeInputInfo.Builder builderForValue) {
+      if (nodeInputInfoBuilder_ == null) {
+        ensureNodeInputInfoIsMutable();
+        nodeInputInfo_.add(index, builderForValue.build());
+        onChanged();
+      } else {
+        nodeInputInfoBuilder_.addMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.GraphTransferNodeInputInfo node_input_info = 3;</code>
+     */
+    public Builder addAllNodeInputInfo(
+        java.lang.Iterable<? extends org.tensorflow.proto.framework.GraphTransferNodeInputInfo> values) {
+      if (nodeInputInfoBuilder_ == null) {
+        ensureNodeInputInfoIsMutable();
+        com.google.protobuf.AbstractMessageLite.Builder.addAll(
+            values, nodeInputInfo_);
+        onChanged();
+      } else {
+        nodeInputInfoBuilder_.addAllMessages(values);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.GraphTransferNodeInputInfo node_input_info = 3;</code>
+     */
+    public Builder clearNodeInputInfo() {
+      if (nodeInputInfoBuilder_ == null) {
+        nodeInputInfo_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000004);
+        onChanged();
+      } else {
+        nodeInputInfoBuilder_.clear();
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.GraphTransferNodeInputInfo node_input_info = 3;</code>
+     */
+    public Builder removeNodeInputInfo(int index) {
+      if (nodeInputInfoBuilder_ == null) {
+        ensureNodeInputInfoIsMutable();
+        nodeInputInfo_.remove(index);
+        onChanged();
+      } else {
+        nodeInputInfoBuilder_.remove(index);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.GraphTransferNodeInputInfo node_input_info = 3;</code>
+     */
+    public org.tensorflow.proto.framework.GraphTransferNodeInputInfo.Builder getNodeInputInfoBuilder(
+        int index) {
+      return getNodeInputInfoFieldBuilder().getBuilder(index);
+    }
+    /**
+     * <code>repeated .tensorflow.GraphTransferNodeInputInfo node_input_info = 3;</code>
+     */
+    public org.tensorflow.proto.framework.GraphTransferNodeInputInfoOrBuilder getNodeInputInfoOrBuilder(
+        int index) {
+      if (nodeInputInfoBuilder_ == null) {
+        return nodeInputInfo_.get(index);  } else {
+        return nodeInputInfoBuilder_.getMessageOrBuilder(index);
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.GraphTransferNodeInputInfo node_input_info = 3;</code>
+     */
+    public java.util.List<? extends org.tensorflow.proto.framework.GraphTransferNodeInputInfoOrBuilder> 
+         getNodeInputInfoOrBuilderList() {
+      if (nodeInputInfoBuilder_ != null) {
+        return nodeInputInfoBuilder_.getMessageOrBuilderList();
+      } else {
+        return java.util.Collections.unmodifiableList(nodeInputInfo_);
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.GraphTransferNodeInputInfo node_input_info = 3;</code>
+     */
+    public org.tensorflow.proto.framework.GraphTransferNodeInputInfo.Builder addNodeInputInfoBuilder() {
+      return getNodeInputInfoFieldBuilder().addBuilder(
+          org.tensorflow.proto.framework.GraphTransferNodeInputInfo.getDefaultInstance());
+    }
+    /**
+     * <code>repeated .tensorflow.GraphTransferNodeInputInfo node_input_info = 3;</code>
+     */
+    public org.tensorflow.proto.framework.GraphTransferNodeInputInfo.Builder addNodeInputInfoBuilder(
+        int index) {
+      return getNodeInputInfoFieldBuilder().addBuilder(
+          index, org.tensorflow.proto.framework.GraphTransferNodeInputInfo.getDefaultInstance());
+    }
+    /**
+     * <code>repeated .tensorflow.GraphTransferNodeInputInfo node_input_info = 3;</code>
+     */
+    public java.util.List<org.tensorflow.proto.framework.GraphTransferNodeInputInfo.Builder> 
+         getNodeInputInfoBuilderList() {
+      return getNodeInputInfoFieldBuilder().getBuilderList();
+    }
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.framework.GraphTransferNodeInputInfo, org.tensorflow.proto.framework.GraphTransferNodeInputInfo.Builder, org.tensorflow.proto.framework.GraphTransferNodeInputInfoOrBuilder> 
+        getNodeInputInfoFieldBuilder() {
+      if (nodeInputInfoBuilder_ == null) {
+        nodeInputInfoBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
+            org.tensorflow.proto.framework.GraphTransferNodeInputInfo, org.tensorflow.proto.framework.GraphTransferNodeInputInfo.Builder, org.tensorflow.proto.framework.GraphTransferNodeInputInfoOrBuilder>(
+                nodeInputInfo_,
+                ((bitField0_ & 0x00000004) != 0),
+                getParentForChildren(),
+                isClean());
+        nodeInputInfo_ = null;
+      }
+      return nodeInputInfoBuilder_;
+    }
+
+    private java.util.List<org.tensorflow.proto.framework.GraphTransferNodeOutputInfo> nodeOutputInfo_ =
+      java.util.Collections.emptyList();
+    private void ensureNodeOutputInfoIsMutable() {
+      if (!((bitField0_ & 0x00000008) != 0)) {
+        nodeOutputInfo_ = new java.util.ArrayList<org.tensorflow.proto.framework.GraphTransferNodeOutputInfo>(nodeOutputInfo_);
+        bitField0_ |= 0x00000008;
+       }
+    }
+
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.framework.GraphTransferNodeOutputInfo, org.tensorflow.proto.framework.GraphTransferNodeOutputInfo.Builder, org.tensorflow.proto.framework.GraphTransferNodeOutputInfoOrBuilder> nodeOutputInfoBuilder_;
+
+    /**
+     * <code>repeated .tensorflow.GraphTransferNodeOutputInfo node_output_info = 4;</code>
+     */
+    public java.util.List<org.tensorflow.proto.framework.GraphTransferNodeOutputInfo> getNodeOutputInfoList() {
+      if (nodeOutputInfoBuilder_ == null) {
+        return java.util.Collections.unmodifiableList(nodeOutputInfo_);
+      } else {
+        return nodeOutputInfoBuilder_.getMessageList();
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.GraphTransferNodeOutputInfo node_output_info = 4;</code>
+     */
+    public int getNodeOutputInfoCount() {
+      if (nodeOutputInfoBuilder_ == null) {
+        return nodeOutputInfo_.size();
+      } else {
+        return nodeOutputInfoBuilder_.getCount();
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.GraphTransferNodeOutputInfo node_output_info = 4;</code>
+     */
+    public org.tensorflow.proto.framework.GraphTransferNodeOutputInfo getNodeOutputInfo(int index) {
+      if (nodeOutputInfoBuilder_ == null) {
+        return nodeOutputInfo_.get(index);
+      } else {
+        return nodeOutputInfoBuilder_.getMessage(index);
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.GraphTransferNodeOutputInfo node_output_info = 4;</code>
+     */
+    public Builder setNodeOutputInfo(
+        int index, org.tensorflow.proto.framework.GraphTransferNodeOutputInfo value) {
+      if (nodeOutputInfoBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureNodeOutputInfoIsMutable();
+        nodeOutputInfo_.set(index, value);
+        onChanged();
+      } else {
+        nodeOutputInfoBuilder_.setMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.GraphTransferNodeOutputInfo node_output_info = 4;</code>
+     */
+    public Builder setNodeOutputInfo(
+        int index, org.tensorflow.proto.framework.GraphTransferNodeOutputInfo.Builder builderForValue) {
+      if (nodeOutputInfoBuilder_ == null) {
+        ensureNodeOutputInfoIsMutable();
+        nodeOutputInfo_.set(index, builderForValue.build());
+        onChanged();
+      } else {
+        nodeOutputInfoBuilder_.setMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.GraphTransferNodeOutputInfo node_output_info = 4;</code>
+     */
+    public Builder addNodeOutputInfo(org.tensorflow.proto.framework.GraphTransferNodeOutputInfo value) {
+      if (nodeOutputInfoBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureNodeOutputInfoIsMutable();
+        nodeOutputInfo_.add(value);
+        onChanged();
+      } else {
+        nodeOutputInfoBuilder_.addMessage(value);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.GraphTransferNodeOutputInfo node_output_info = 4;</code>
+     */
+    public Builder addNodeOutputInfo(
+        int index, org.tensorflow.proto.framework.GraphTransferNodeOutputInfo value) {
+      if (nodeOutputInfoBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureNodeOutputInfoIsMutable();
+        nodeOutputInfo_.add(index, value);
+        onChanged();
+      } else {
+        nodeOutputInfoBuilder_.addMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.GraphTransferNodeOutputInfo node_output_info = 4;</code>
+     */
+    public Builder addNodeOutputInfo(
+        org.tensorflow.proto.framework.GraphTransferNodeOutputInfo.Builder builderForValue) {
+      if (nodeOutputInfoBuilder_ == null) {
+        ensureNodeOutputInfoIsMutable();
+        nodeOutputInfo_.add(builderForValue.build());
+        onChanged();
+      } else {
+        nodeOutputInfoBuilder_.addMessage(builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.GraphTransferNodeOutputInfo node_output_info = 4;</code>
+     */
+    public Builder addNodeOutputInfo(
+        int index, org.tensorflow.proto.framework.GraphTransferNodeOutputInfo.Builder builderForValue) {
+      if (nodeOutputInfoBuilder_ == null) {
+        ensureNodeOutputInfoIsMutable();
+        nodeOutputInfo_.add(index, builderForValue.build());
+        onChanged();
+      } else {
+        nodeOutputInfoBuilder_.addMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.GraphTransferNodeOutputInfo node_output_info = 4;</code>
+     */
+    public Builder addAllNodeOutputInfo(
+        java.lang.Iterable<? extends org.tensorflow.proto.framework.GraphTransferNodeOutputInfo> values) {
+      if (nodeOutputInfoBuilder_ == null) {
+        ensureNodeOutputInfoIsMutable();
+        com.google.protobuf.AbstractMessageLite.Builder.addAll(
+            values, nodeOutputInfo_);
+        onChanged();
+      } else {
+        nodeOutputInfoBuilder_.addAllMessages(values);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.GraphTransferNodeOutputInfo node_output_info = 4;</code>
+     */
+    public Builder clearNodeOutputInfo() {
+      if (nodeOutputInfoBuilder_ == null) {
+        nodeOutputInfo_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000008);
+        onChanged();
+      } else {
+        nodeOutputInfoBuilder_.clear();
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.GraphTransferNodeOutputInfo node_output_info = 4;</code>
+     */
+    public Builder removeNodeOutputInfo(int index) {
+      if (nodeOutputInfoBuilder_ == null) {
+        ensureNodeOutputInfoIsMutable();
+        nodeOutputInfo_.remove(index);
+        onChanged();
+      } else {
+        nodeOutputInfoBuilder_.remove(index);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.GraphTransferNodeOutputInfo node_output_info = 4;</code>
+     */
+    public org.tensorflow.proto.framework.GraphTransferNodeOutputInfo.Builder getNodeOutputInfoBuilder(
+        int index) {
+      return getNodeOutputInfoFieldBuilder().getBuilder(index);
+    }
+    /**
+     * <code>repeated .tensorflow.GraphTransferNodeOutputInfo node_output_info = 4;</code>
+     */
+    public org.tensorflow.proto.framework.GraphTransferNodeOutputInfoOrBuilder getNodeOutputInfoOrBuilder(
+        int index) {
+      if (nodeOutputInfoBuilder_ == null) {
+        return nodeOutputInfo_.get(index);  } else {
+        return nodeOutputInfoBuilder_.getMessageOrBuilder(index);
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.GraphTransferNodeOutputInfo node_output_info = 4;</code>
+     */
+    public java.util.List<? extends org.tensorflow.proto.framework.GraphTransferNodeOutputInfoOrBuilder> 
+         getNodeOutputInfoOrBuilderList() {
+      if (nodeOutputInfoBuilder_ != null) {
+        return nodeOutputInfoBuilder_.getMessageOrBuilderList();
+      } else {
+        return java.util.Collections.unmodifiableList(nodeOutputInfo_);
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.GraphTransferNodeOutputInfo node_output_info = 4;</code>
+     */
+    public org.tensorflow.proto.framework.GraphTransferNodeOutputInfo.Builder addNodeOutputInfoBuilder() {
+      return getNodeOutputInfoFieldBuilder().addBuilder(
+          org.tensorflow.proto.framework.GraphTransferNodeOutputInfo.getDefaultInstance());
+    }
+    /**
+     * <code>repeated .tensorflow.GraphTransferNodeOutputInfo node_output_info = 4;</code>
+     */
+    public org.tensorflow.proto.framework.GraphTransferNodeOutputInfo.Builder addNodeOutputInfoBuilder(
+        int index) {
+      return getNodeOutputInfoFieldBuilder().addBuilder(
+          index, org.tensorflow.proto.framework.GraphTransferNodeOutputInfo.getDefaultInstance());
+    }
+    /**
+     * <code>repeated .tensorflow.GraphTransferNodeOutputInfo node_output_info = 4;</code>
+     */
+    public java.util.List<org.tensorflow.proto.framework.GraphTransferNodeOutputInfo.Builder> 
+         getNodeOutputInfoBuilderList() {
+      return getNodeOutputInfoFieldBuilder().getBuilderList();
+    }
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.framework.GraphTransferNodeOutputInfo, org.tensorflow.proto.framework.GraphTransferNodeOutputInfo.Builder, org.tensorflow.proto.framework.GraphTransferNodeOutputInfoOrBuilder> 
+        getNodeOutputInfoFieldBuilder() {
+      if (nodeOutputInfoBuilder_ == null) {
+        nodeOutputInfoBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
+            org.tensorflow.proto.framework.GraphTransferNodeOutputInfo, org.tensorflow.proto.framework.GraphTransferNodeOutputInfo.Builder, org.tensorflow.proto.framework.GraphTransferNodeOutputInfoOrBuilder>(
+                nodeOutputInfo_,
+                ((bitField0_ & 0x00000008) != 0),
+                getParentForChildren(),
+                isClean());
+        nodeOutputInfo_ = null;
+      }
+      return nodeOutputInfoBuilder_;
+    }
+
+    private java.util.List<org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfo> graphInputNodeInfo_ =
+      java.util.Collections.emptyList();
+    private void ensureGraphInputNodeInfoIsMutable() {
+      if (!((bitField0_ & 0x00000010) != 0)) {
+        graphInputNodeInfo_ = new java.util.ArrayList<org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfo>(graphInputNodeInfo_);
+        bitField0_ |= 0x00000010;
+       }
+    }
+
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfo, org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfo.Builder, org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfoOrBuilder> graphInputNodeInfoBuilder_;
+
+    /**
+     * <pre>
+     * Input Node parameters of transferred graph
+     * </pre>
+     *
+     * <code>repeated .tensorflow.GraphTransferGraphInputNodeInfo graph_input_node_info = 5;</code>
+     */
+    public java.util.List<org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfo> getGraphInputNodeInfoList() {
+      if (graphInputNodeInfoBuilder_ == null) {
+        return java.util.Collections.unmodifiableList(graphInputNodeInfo_);
+      } else {
+        return graphInputNodeInfoBuilder_.getMessageList();
+      }
+    }
+    /**
+     * <pre>
+     * Input Node parameters of transferred graph
+     * </pre>
+     *
+     * <code>repeated .tensorflow.GraphTransferGraphInputNodeInfo graph_input_node_info = 5;</code>
+     */
+    public int getGraphInputNodeInfoCount() {
+      if (graphInputNodeInfoBuilder_ == null) {
+        return graphInputNodeInfo_.size();
+      } else {
+        return graphInputNodeInfoBuilder_.getCount();
+      }
+    }
+    /**
+     * <pre>
+     * Input Node parameters of transferred graph
+     * </pre>
+     *
+     * <code>repeated .tensorflow.GraphTransferGraphInputNodeInfo graph_input_node_info = 5;</code>
+     */
+    public org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfo getGraphInputNodeInfo(int index) {
+      if (graphInputNodeInfoBuilder_ == null) {
+        return graphInputNodeInfo_.get(index);
+      } else {
+        return graphInputNodeInfoBuilder_.getMessage(index);
+      }
+    }
+    /**
+     * <pre>
+     * Input Node parameters of transferred graph
+     * </pre>
+     *
+     * <code>repeated .tensorflow.GraphTransferGraphInputNodeInfo graph_input_node_info = 5;</code>
+     */
+    public Builder setGraphInputNodeInfo(
+        int index, org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfo value) {
+      if (graphInputNodeInfoBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureGraphInputNodeInfoIsMutable();
+        graphInputNodeInfo_.set(index, value);
+        onChanged();
+      } else {
+        graphInputNodeInfoBuilder_.setMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Input Node parameters of transferred graph
+     * </pre>
+     *
+     * <code>repeated .tensorflow.GraphTransferGraphInputNodeInfo graph_input_node_info = 5;</code>
+     */
+    public Builder setGraphInputNodeInfo(
+        int index, org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfo.Builder builderForValue) {
+      if (graphInputNodeInfoBuilder_ == null) {
+        ensureGraphInputNodeInfoIsMutable();
+        graphInputNodeInfo_.set(index, builderForValue.build());
+        onChanged();
+      } else {
+        graphInputNodeInfoBuilder_.setMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Input Node parameters of transferred graph
+     * </pre>
+     *
+     * <code>repeated .tensorflow.GraphTransferGraphInputNodeInfo graph_input_node_info = 5;</code>
+     */
+    public Builder addGraphInputNodeInfo(org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfo value) {
+      if (graphInputNodeInfoBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureGraphInputNodeInfoIsMutable();
+        graphInputNodeInfo_.add(value);
+        onChanged();
+      } else {
+        graphInputNodeInfoBuilder_.addMessage(value);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Input Node parameters of transferred graph
+     * </pre>
+     *
+     * <code>repeated .tensorflow.GraphTransferGraphInputNodeInfo graph_input_node_info = 5;</code>
+     */
+    public Builder addGraphInputNodeInfo(
+        int index, org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfo value) {
+      if (graphInputNodeInfoBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureGraphInputNodeInfoIsMutable();
+        graphInputNodeInfo_.add(index, value);
+        onChanged();
+      } else {
+        graphInputNodeInfoBuilder_.addMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Input Node parameters of transferred graph
+     * </pre>
+     *
+     * <code>repeated .tensorflow.GraphTransferGraphInputNodeInfo graph_input_node_info = 5;</code>
+     */
+    public Builder addGraphInputNodeInfo(
+        org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfo.Builder builderForValue) {
+      if (graphInputNodeInfoBuilder_ == null) {
+        ensureGraphInputNodeInfoIsMutable();
+        graphInputNodeInfo_.add(builderForValue.build());
+        onChanged();
+      } else {
+        graphInputNodeInfoBuilder_.addMessage(builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Input Node parameters of transferred graph
+     * </pre>
+     *
+     * <code>repeated .tensorflow.GraphTransferGraphInputNodeInfo graph_input_node_info = 5;</code>
+     */
+    public Builder addGraphInputNodeInfo(
+        int index, org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfo.Builder builderForValue) {
+      if (graphInputNodeInfoBuilder_ == null) {
+        ensureGraphInputNodeInfoIsMutable();
+        graphInputNodeInfo_.add(index, builderForValue.build());
+        onChanged();
+      } else {
+        graphInputNodeInfoBuilder_.addMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Input Node parameters of transferred graph
+     * </pre>
+     *
+     * <code>repeated .tensorflow.GraphTransferGraphInputNodeInfo graph_input_node_info = 5;</code>
+     */
+    public Builder addAllGraphInputNodeInfo(
+        java.lang.Iterable<? extends org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfo> values) {
+      if (graphInputNodeInfoBuilder_ == null) {
+        ensureGraphInputNodeInfoIsMutable();
+        com.google.protobuf.AbstractMessageLite.Builder.addAll(
+            values, graphInputNodeInfo_);
+        onChanged();
+      } else {
+        graphInputNodeInfoBuilder_.addAllMessages(values);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Input Node parameters of transferred graph
+     * </pre>
+     *
+     * <code>repeated .tensorflow.GraphTransferGraphInputNodeInfo graph_input_node_info = 5;</code>
+     */
+    public Builder clearGraphInputNodeInfo() {
+      if (graphInputNodeInfoBuilder_ == null) {
+        graphInputNodeInfo_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000010);
+        onChanged();
+      } else {
+        graphInputNodeInfoBuilder_.clear();
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Input Node parameters of transferred graph
+     * </pre>
+     *
+     * <code>repeated .tensorflow.GraphTransferGraphInputNodeInfo graph_input_node_info = 5;</code>
+     */
+    public Builder removeGraphInputNodeInfo(int index) {
+      if (graphInputNodeInfoBuilder_ == null) {
+        ensureGraphInputNodeInfoIsMutable();
+        graphInputNodeInfo_.remove(index);
+        onChanged();
+      } else {
+        graphInputNodeInfoBuilder_.remove(index);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Input Node parameters of transferred graph
+     * </pre>
+     *
+     * <code>repeated .tensorflow.GraphTransferGraphInputNodeInfo graph_input_node_info = 5;</code>
+     */
+    public org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfo.Builder getGraphInputNodeInfoBuilder(
+        int index) {
+      return getGraphInputNodeInfoFieldBuilder().getBuilder(index);
+    }
+    /**
+     * <pre>
+     * Input Node parameters of transferred graph
+     * </pre>
+     *
+     * <code>repeated .tensorflow.GraphTransferGraphInputNodeInfo graph_input_node_info = 5;</code>
+     */
+    public org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfoOrBuilder getGraphInputNodeInfoOrBuilder(
+        int index) {
+      if (graphInputNodeInfoBuilder_ == null) {
+        return graphInputNodeInfo_.get(index);  } else {
+        return graphInputNodeInfoBuilder_.getMessageOrBuilder(index);
+      }
+    }
+    /**
+     * <pre>
+     * Input Node parameters of transferred graph
+     * </pre>
+     *
+     * <code>repeated .tensorflow.GraphTransferGraphInputNodeInfo graph_input_node_info = 5;</code>
+     */
+    public java.util.List<? extends org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfoOrBuilder> 
+         getGraphInputNodeInfoOrBuilderList() {
+      if (graphInputNodeInfoBuilder_ != null) {
+        return graphInputNodeInfoBuilder_.getMessageOrBuilderList();
+      } else {
+        return java.util.Collections.unmodifiableList(graphInputNodeInfo_);
+      }
+    }
+    /**
+     * <pre>
+     * Input Node parameters of transferred graph
+     * </pre>
+     *
+     * <code>repeated .tensorflow.GraphTransferGraphInputNodeInfo graph_input_node_info = 5;</code>
+     */
+    public org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfo.Builder addGraphInputNodeInfoBuilder() {
+      return getGraphInputNodeInfoFieldBuilder().addBuilder(
+          org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfo.getDefaultInstance());
+    }
+    /**
+     * <pre>
+     * Input Node parameters of transferred graph
+     * </pre>
+     *
+     * <code>repeated .tensorflow.GraphTransferGraphInputNodeInfo graph_input_node_info = 5;</code>
+     */
+    public org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfo.Builder addGraphInputNodeInfoBuilder(
+        int index) {
+      return getGraphInputNodeInfoFieldBuilder().addBuilder(
+          index, org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfo.getDefaultInstance());
+    }
+    /**
+     * <pre>
+     * Input Node parameters of transferred graph
+     * </pre>
+     *
+     * <code>repeated .tensorflow.GraphTransferGraphInputNodeInfo graph_input_node_info = 5;</code>
+     */
+    public java.util.List<org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfo.Builder> 
+         getGraphInputNodeInfoBuilderList() {
+      return getGraphInputNodeInfoFieldBuilder().getBuilderList();
+    }
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfo, org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfo.Builder, org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfoOrBuilder> 
+        getGraphInputNodeInfoFieldBuilder() {
+      if (graphInputNodeInfoBuilder_ == null) {
+        graphInputNodeInfoBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
+            org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfo, org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfo.Builder, org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfoOrBuilder>(
+                graphInputNodeInfo_,
+                ((bitField0_ & 0x00000010) != 0),
+                getParentForChildren(),
+                isClean());
+        graphInputNodeInfo_ = null;
+      }
+      return graphInputNodeInfoBuilder_;
+    }
+
+    private java.util.List<org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfo> graphOutputNodeInfo_ =
+      java.util.Collections.emptyList();
+    private void ensureGraphOutputNodeInfoIsMutable() {
+      if (!((bitField0_ & 0x00000020) != 0)) {
+        graphOutputNodeInfo_ = new java.util.ArrayList<org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfo>(graphOutputNodeInfo_);
+        bitField0_ |= 0x00000020;
+       }
+    }
+
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfo, org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfo.Builder, org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfoOrBuilder> graphOutputNodeInfoBuilder_;
+
+    /**
+     * <code>repeated .tensorflow.GraphTransferGraphOutputNodeInfo graph_output_node_info = 6;</code>
+     */
+    public java.util.List<org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfo> getGraphOutputNodeInfoList() {
+      if (graphOutputNodeInfoBuilder_ == null) {
+        return java.util.Collections.unmodifiableList(graphOutputNodeInfo_);
+      } else {
+        return graphOutputNodeInfoBuilder_.getMessageList();
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.GraphTransferGraphOutputNodeInfo graph_output_node_info = 6;</code>
+     */
+    public int getGraphOutputNodeInfoCount() {
+      if (graphOutputNodeInfoBuilder_ == null) {
+        return graphOutputNodeInfo_.size();
+      } else {
+        return graphOutputNodeInfoBuilder_.getCount();
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.GraphTransferGraphOutputNodeInfo graph_output_node_info = 6;</code>
+     */
+    public org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfo getGraphOutputNodeInfo(int index) {
+      if (graphOutputNodeInfoBuilder_ == null) {
+        return graphOutputNodeInfo_.get(index);
+      } else {
+        return graphOutputNodeInfoBuilder_.getMessage(index);
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.GraphTransferGraphOutputNodeInfo graph_output_node_info = 6;</code>
+     */
+    public Builder setGraphOutputNodeInfo(
+        int index, org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfo value) {
+      if (graphOutputNodeInfoBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureGraphOutputNodeInfoIsMutable();
+        graphOutputNodeInfo_.set(index, value);
+        onChanged();
+      } else {
+        graphOutputNodeInfoBuilder_.setMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.GraphTransferGraphOutputNodeInfo graph_output_node_info = 6;</code>
+     */
+    public Builder setGraphOutputNodeInfo(
+        int index, org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfo.Builder builderForValue) {
+      if (graphOutputNodeInfoBuilder_ == null) {
+        ensureGraphOutputNodeInfoIsMutable();
+        graphOutputNodeInfo_.set(index, builderForValue.build());
+        onChanged();
+      } else {
+        graphOutputNodeInfoBuilder_.setMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.GraphTransferGraphOutputNodeInfo graph_output_node_info = 6;</code>
+     */
+    public Builder addGraphOutputNodeInfo(org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfo value) {
+      if (graphOutputNodeInfoBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureGraphOutputNodeInfoIsMutable();
+        graphOutputNodeInfo_.add(value);
+        onChanged();
+      } else {
+        graphOutputNodeInfoBuilder_.addMessage(value);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.GraphTransferGraphOutputNodeInfo graph_output_node_info = 6;</code>
+     */
+    public Builder addGraphOutputNodeInfo(
+        int index, org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfo value) {
+      if (graphOutputNodeInfoBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureGraphOutputNodeInfoIsMutable();
+        graphOutputNodeInfo_.add(index, value);
+        onChanged();
+      } else {
+        graphOutputNodeInfoBuilder_.addMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.GraphTransferGraphOutputNodeInfo graph_output_node_info = 6;</code>
+     */
+    public Builder addGraphOutputNodeInfo(
+        org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfo.Builder builderForValue) {
+      if (graphOutputNodeInfoBuilder_ == null) {
+        ensureGraphOutputNodeInfoIsMutable();
+        graphOutputNodeInfo_.add(builderForValue.build());
+        onChanged();
+      } else {
+        graphOutputNodeInfoBuilder_.addMessage(builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.GraphTransferGraphOutputNodeInfo graph_output_node_info = 6;</code>
+     */
+    public Builder addGraphOutputNodeInfo(
+        int index, org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfo.Builder builderForValue) {
+      if (graphOutputNodeInfoBuilder_ == null) {
+        ensureGraphOutputNodeInfoIsMutable();
+        graphOutputNodeInfo_.add(index, builderForValue.build());
+        onChanged();
+      } else {
+        graphOutputNodeInfoBuilder_.addMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.GraphTransferGraphOutputNodeInfo graph_output_node_info = 6;</code>
+     */
+    public Builder addAllGraphOutputNodeInfo(
+        java.lang.Iterable<? extends org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfo> values) {
+      if (graphOutputNodeInfoBuilder_ == null) {
+        ensureGraphOutputNodeInfoIsMutable();
+        com.google.protobuf.AbstractMessageLite.Builder.addAll(
+            values, graphOutputNodeInfo_);
+        onChanged();
+      } else {
+        graphOutputNodeInfoBuilder_.addAllMessages(values);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.GraphTransferGraphOutputNodeInfo graph_output_node_info = 6;</code>
+     */
+    public Builder clearGraphOutputNodeInfo() {
+      if (graphOutputNodeInfoBuilder_ == null) {
+        graphOutputNodeInfo_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000020);
+        onChanged();
+      } else {
+        graphOutputNodeInfoBuilder_.clear();
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.GraphTransferGraphOutputNodeInfo graph_output_node_info = 6;</code>
+     */
+    public Builder removeGraphOutputNodeInfo(int index) {
+      if (graphOutputNodeInfoBuilder_ == null) {
+        ensureGraphOutputNodeInfoIsMutable();
+        graphOutputNodeInfo_.remove(index);
+        onChanged();
+      } else {
+        graphOutputNodeInfoBuilder_.remove(index);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.GraphTransferGraphOutputNodeInfo graph_output_node_info = 6;</code>
+     */
+    public org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfo.Builder getGraphOutputNodeInfoBuilder(
+        int index) {
+      return getGraphOutputNodeInfoFieldBuilder().getBuilder(index);
+    }
+    /**
+     * <code>repeated .tensorflow.GraphTransferGraphOutputNodeInfo graph_output_node_info = 6;</code>
+     */
+    public org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfoOrBuilder getGraphOutputNodeInfoOrBuilder(
+        int index) {
+      if (graphOutputNodeInfoBuilder_ == null) {
+        return graphOutputNodeInfo_.get(index);  } else {
+        return graphOutputNodeInfoBuilder_.getMessageOrBuilder(index);
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.GraphTransferGraphOutputNodeInfo graph_output_node_info = 6;</code>
+     */
+    public java.util.List<? extends org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfoOrBuilder> 
+         getGraphOutputNodeInfoOrBuilderList() {
+      if (graphOutputNodeInfoBuilder_ != null) {
+        return graphOutputNodeInfoBuilder_.getMessageOrBuilderList();
+      } else {
+        return java.util.Collections.unmodifiableList(graphOutputNodeInfo_);
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.GraphTransferGraphOutputNodeInfo graph_output_node_info = 6;</code>
+     */
+    public org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfo.Builder addGraphOutputNodeInfoBuilder() {
+      return getGraphOutputNodeInfoFieldBuilder().addBuilder(
+          org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfo.getDefaultInstance());
+    }
+    /**
+     * <code>repeated .tensorflow.GraphTransferGraphOutputNodeInfo graph_output_node_info = 6;</code>
+     */
+    public org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfo.Builder addGraphOutputNodeInfoBuilder(
+        int index) {
+      return getGraphOutputNodeInfoFieldBuilder().addBuilder(
+          index, org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfo.getDefaultInstance());
+    }
+    /**
+     * <code>repeated .tensorflow.GraphTransferGraphOutputNodeInfo graph_output_node_info = 6;</code>
+     */
+    public java.util.List<org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfo.Builder> 
+         getGraphOutputNodeInfoBuilderList() {
+      return getGraphOutputNodeInfoFieldBuilder().getBuilderList();
+    }
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfo, org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfo.Builder, org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfoOrBuilder> 
+        getGraphOutputNodeInfoFieldBuilder() {
+      if (graphOutputNodeInfoBuilder_ == null) {
+        graphOutputNodeInfoBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
+            org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfo, org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfo.Builder, org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfoOrBuilder>(
+                graphOutputNodeInfo_,
+                ((bitField0_ & 0x00000020) != 0),
+                getParentForChildren(),
+                isClean());
+        graphOutputNodeInfo_ = null;
+      }
+      return graphOutputNodeInfoBuilder_;
+    }
+
+    private int destination_ = 0;
+    /**
+     * <pre>
+     * Destination of graph transfer
+     * </pre>
+     *
+     * <code>.tensorflow.GraphTransferInfo.Destination destination = 7;</code>
+     */
+    public int getDestinationValue() {
+      return destination_;
+    }
+    /**
+     * <pre>
+     * Destination of graph transfer
+     * </pre>
+     *
+     * <code>.tensorflow.GraphTransferInfo.Destination destination = 7;</code>
+     */
+    public Builder setDestinationValue(int value) {
+      destination_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Destination of graph transfer
+     * </pre>
+     *
+     * <code>.tensorflow.GraphTransferInfo.Destination destination = 7;</code>
+     */
+    public org.tensorflow.proto.framework.GraphTransferInfo.Destination getDestination() {
+      @SuppressWarnings("deprecation")
+      org.tensorflow.proto.framework.GraphTransferInfo.Destination result = org.tensorflow.proto.framework.GraphTransferInfo.Destination.valueOf(destination_);
+      return result == null ? org.tensorflow.proto.framework.GraphTransferInfo.Destination.UNRECOGNIZED : result;
+    }
+    /**
+     * <pre>
+     * Destination of graph transfer
+     * </pre>
+     *
+     * <code>.tensorflow.GraphTransferInfo.Destination destination = 7;</code>
+     */
+    public Builder setDestination(org.tensorflow.proto.framework.GraphTransferInfo.Destination value) {
+      if (value == null) {
+        throw new NullPointerException();
+      }
+      
+      destination_ = value.getNumber();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Destination of graph transfer
+     * </pre>
+     *
+     * <code>.tensorflow.GraphTransferInfo.Destination destination = 7;</code>
+     */
+    public Builder clearDestination() {
+      
+      destination_ = 0;
+      onChanged();
+      return this;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.GraphTransferInfo)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.GraphTransferInfo)
+  private static final org.tensorflow.proto.framework.GraphTransferInfo DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.GraphTransferInfo();
+  }
+
+  public static org.tensorflow.proto.framework.GraphTransferInfo getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<GraphTransferInfo>
+      PARSER = new com.google.protobuf.AbstractParser<GraphTransferInfo>() {
+    @java.lang.Override
+    public GraphTransferInfo parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new GraphTransferInfo(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<GraphTransferInfo> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<GraphTransferInfo> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.framework.GraphTransferInfo getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphTransferInfoOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphTransferInfoOrBuilder.java
new file mode 100644
index 00000000000..d999e1d9d48
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphTransferInfoOrBuilder.java
@@ -0,0 +1,190 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/framework/graph_transfer_info.proto
+
+package org.tensorflow.proto.framework;
+
+public interface GraphTransferInfoOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.GraphTransferInfo)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <code>repeated .tensorflow.GraphTransferNodeInfo node_info = 1;</code>
+   */
+  java.util.List<org.tensorflow.proto.framework.GraphTransferNodeInfo> 
+      getNodeInfoList();
+  /**
+   * <code>repeated .tensorflow.GraphTransferNodeInfo node_info = 1;</code>
+   */
+  org.tensorflow.proto.framework.GraphTransferNodeInfo getNodeInfo(int index);
+  /**
+   * <code>repeated .tensorflow.GraphTransferNodeInfo node_info = 1;</code>
+   */
+  int getNodeInfoCount();
+  /**
+   * <code>repeated .tensorflow.GraphTransferNodeInfo node_info = 1;</code>
+   */
+  java.util.List<? extends org.tensorflow.proto.framework.GraphTransferNodeInfoOrBuilder> 
+      getNodeInfoOrBuilderList();
+  /**
+   * <code>repeated .tensorflow.GraphTransferNodeInfo node_info = 1;</code>
+   */
+  org.tensorflow.proto.framework.GraphTransferNodeInfoOrBuilder getNodeInfoOrBuilder(
+      int index);
+
+  /**
+   * <code>repeated .tensorflow.GraphTransferConstNodeInfo const_node_info = 2;</code>
+   */
+  java.util.List<org.tensorflow.proto.framework.GraphTransferConstNodeInfo> 
+      getConstNodeInfoList();
+  /**
+   * <code>repeated .tensorflow.GraphTransferConstNodeInfo const_node_info = 2;</code>
+   */
+  org.tensorflow.proto.framework.GraphTransferConstNodeInfo getConstNodeInfo(int index);
+  /**
+   * <code>repeated .tensorflow.GraphTransferConstNodeInfo const_node_info = 2;</code>
+   */
+  int getConstNodeInfoCount();
+  /**
+   * <code>repeated .tensorflow.GraphTransferConstNodeInfo const_node_info = 2;</code>
+   */
+  java.util.List<? extends org.tensorflow.proto.framework.GraphTransferConstNodeInfoOrBuilder> 
+      getConstNodeInfoOrBuilderList();
+  /**
+   * <code>repeated .tensorflow.GraphTransferConstNodeInfo const_node_info = 2;</code>
+   */
+  org.tensorflow.proto.framework.GraphTransferConstNodeInfoOrBuilder getConstNodeInfoOrBuilder(
+      int index);
+
+  /**
+   * <code>repeated .tensorflow.GraphTransferNodeInputInfo node_input_info = 3;</code>
+   */
+  java.util.List<org.tensorflow.proto.framework.GraphTransferNodeInputInfo> 
+      getNodeInputInfoList();
+  /**
+   * <code>repeated .tensorflow.GraphTransferNodeInputInfo node_input_info = 3;</code>
+   */
+  org.tensorflow.proto.framework.GraphTransferNodeInputInfo getNodeInputInfo(int index);
+  /**
+   * <code>repeated .tensorflow.GraphTransferNodeInputInfo node_input_info = 3;</code>
+   */
+  int getNodeInputInfoCount();
+  /**
+   * <code>repeated .tensorflow.GraphTransferNodeInputInfo node_input_info = 3;</code>
+   */
+  java.util.List<? extends org.tensorflow.proto.framework.GraphTransferNodeInputInfoOrBuilder> 
+      getNodeInputInfoOrBuilderList();
+  /**
+   * <code>repeated .tensorflow.GraphTransferNodeInputInfo node_input_info = 3;</code>
+   */
+  org.tensorflow.proto.framework.GraphTransferNodeInputInfoOrBuilder getNodeInputInfoOrBuilder(
+      int index);
+
+  /**
+   * <code>repeated .tensorflow.GraphTransferNodeOutputInfo node_output_info = 4;</code>
+   */
+  java.util.List<org.tensorflow.proto.framework.GraphTransferNodeOutputInfo> 
+      getNodeOutputInfoList();
+  /**
+   * <code>repeated .tensorflow.GraphTransferNodeOutputInfo node_output_info = 4;</code>
+   */
+  org.tensorflow.proto.framework.GraphTransferNodeOutputInfo getNodeOutputInfo(int index);
+  /**
+   * <code>repeated .tensorflow.GraphTransferNodeOutputInfo node_output_info = 4;</code>
+   */
+  int getNodeOutputInfoCount();
+  /**
+   * <code>repeated .tensorflow.GraphTransferNodeOutputInfo node_output_info = 4;</code>
+   */
+  java.util.List<? extends org.tensorflow.proto.framework.GraphTransferNodeOutputInfoOrBuilder> 
+      getNodeOutputInfoOrBuilderList();
+  /**
+   * <code>repeated .tensorflow.GraphTransferNodeOutputInfo node_output_info = 4;</code>
+   */
+  org.tensorflow.proto.framework.GraphTransferNodeOutputInfoOrBuilder getNodeOutputInfoOrBuilder(
+      int index);
+
+  /**
+   * <pre>
+   * Input Node parameters of transferred graph
+   * </pre>
+   *
+   * <code>repeated .tensorflow.GraphTransferGraphInputNodeInfo graph_input_node_info = 5;</code>
+   */
+  java.util.List<org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfo> 
+      getGraphInputNodeInfoList();
+  /**
+   * <pre>
+   * Input Node parameters of transferred graph
+   * </pre>
+   *
+   * <code>repeated .tensorflow.GraphTransferGraphInputNodeInfo graph_input_node_info = 5;</code>
+   */
+  org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfo getGraphInputNodeInfo(int index);
+  /**
+   * <pre>
+   * Input Node parameters of transferred graph
+   * </pre>
+   *
+   * <code>repeated .tensorflow.GraphTransferGraphInputNodeInfo graph_input_node_info = 5;</code>
+   */
+  int getGraphInputNodeInfoCount();
+  /**
+   * <pre>
+   * Input Node parameters of transferred graph
+   * </pre>
+   *
+   * <code>repeated .tensorflow.GraphTransferGraphInputNodeInfo graph_input_node_info = 5;</code>
+   */
+  java.util.List<? extends org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfoOrBuilder> 
+      getGraphInputNodeInfoOrBuilderList();
+  /**
+   * <pre>
+   * Input Node parameters of transferred graph
+   * </pre>
+   *
+   * <code>repeated .tensorflow.GraphTransferGraphInputNodeInfo graph_input_node_info = 5;</code>
+   */
+  org.tensorflow.proto.framework.GraphTransferGraphInputNodeInfoOrBuilder getGraphInputNodeInfoOrBuilder(
+      int index);
+
+  /**
+   * <code>repeated .tensorflow.GraphTransferGraphOutputNodeInfo graph_output_node_info = 6;</code>
+   */
+  java.util.List<org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfo> 
+      getGraphOutputNodeInfoList();
+  /**
+   * <code>repeated .tensorflow.GraphTransferGraphOutputNodeInfo graph_output_node_info = 6;</code>
+   */
+  org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfo getGraphOutputNodeInfo(int index);
+  /**
+   * <code>repeated .tensorflow.GraphTransferGraphOutputNodeInfo graph_output_node_info = 6;</code>
+   */
+  int getGraphOutputNodeInfoCount();
+  /**
+   * <code>repeated .tensorflow.GraphTransferGraphOutputNodeInfo graph_output_node_info = 6;</code>
+   */
+  java.util.List<? extends org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfoOrBuilder> 
+      getGraphOutputNodeInfoOrBuilderList();
+  /**
+   * <code>repeated .tensorflow.GraphTransferGraphOutputNodeInfo graph_output_node_info = 6;</code>
+   */
+  org.tensorflow.proto.framework.GraphTransferGraphOutputNodeInfoOrBuilder getGraphOutputNodeInfoOrBuilder(
+      int index);
+
+  /**
+   * <pre>
+   * Destination of graph transfer
+   * </pre>
+   *
+   * <code>.tensorflow.GraphTransferInfo.Destination destination = 7;</code>
+   */
+  int getDestinationValue();
+  /**
+   * <pre>
+   * Destination of graph transfer
+   * </pre>
+   *
+   * <code>.tensorflow.GraphTransferInfo.Destination destination = 7;</code>
+   */
+  org.tensorflow.proto.framework.GraphTransferInfo.Destination getDestination();
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphTransferInfoProto.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphTransferInfoProto.java
new file mode 100644
index 00000000000..750526c8897
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphTransferInfoProto.java
@@ -0,0 +1,163 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/framework/graph_transfer_info.proto
+
+package org.tensorflow.proto.framework;
+
+public final class GraphTransferInfoProto {
+  private GraphTransferInfoProto() {}
+  public static void registerAllExtensions(
+      com.google.protobuf.ExtensionRegistryLite registry) {
+  }
+
+  public static void registerAllExtensions(
+      com.google.protobuf.ExtensionRegistry registry) {
+    registerAllExtensions(
+        (com.google.protobuf.ExtensionRegistryLite) registry);
+  }
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_GraphTransferNodeInput_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_GraphTransferNodeInput_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_GraphTransferNodeInfo_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_GraphTransferNodeInfo_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_GraphTransferConstNodeInfo_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_GraphTransferConstNodeInfo_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_GraphTransferNodeInputInfo_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_GraphTransferNodeInputInfo_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_GraphTransferNodeOutputInfo_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_GraphTransferNodeOutputInfo_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_GraphTransferGraphInputNodeInfo_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_GraphTransferGraphInputNodeInfo_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_GraphTransferGraphOutputNodeInfo_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_GraphTransferGraphOutputNodeInfo_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_GraphTransferInfo_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_GraphTransferInfo_fieldAccessorTable;
+
+  public static com.google.protobuf.Descriptors.FileDescriptor
+      getDescriptor() {
+    return descriptor;
+  }
+  private static  com.google.protobuf.Descriptors.FileDescriptor
+      descriptor;
+  static {
+    java.lang.String[] descriptorData = {
+      "\n3tensorflow/core/framework/graph_transf" +
+      "er_info.proto\022\ntensorflow\032%tensorflow/co" +
+      "re/framework/types.proto\">\n\026GraphTransfe" +
+      "rNodeInput\022\017\n\007node_id\030\001 \001(\005\022\023\n\013output_po" +
+      "rt\030\002 \001(\005\"\233\001\n\025GraphTransferNodeInfo\022\014\n\004na" +
+      "me\030\001 \001(\t\022\017\n\007node_id\030\002 \001(\005\022\021\n\ttype_name\030\003" +
+      " \001(\t\022\021\n\tsoc_op_id\030\004 \001(\005\022\022\n\npadding_id\030\005 " +
+      "\001(\005\022\023\n\013input_count\030\006 \001(\005\022\024\n\014output_count" +
+      "\030\007 \001(\005\"}\n\032GraphTransferConstNodeInfo\022\014\n\004" +
+      "name\030\001 \001(\t\022\017\n\007node_id\030\002 \001(\005\022\r\n\005shape\030\003 \003" +
+      "(\003\022\014\n\004data\030\004 \001(\014\022#\n\005dtype\030\005 \001(\0162\024.tensor" +
+      "flow.DataType\"e\n\032GraphTransferNodeInputI" +
+      "nfo\022\017\n\007node_id\030\001 \001(\005\0226\n\nnode_input\030\002 \003(\013" +
+      "2\".tensorflow.GraphTransferNodeInput\"E\n\033" +
+      "GraphTransferNodeOutputInfo\022\017\n\007node_id\030\001" +
+      " \001(\005\022\025\n\rmax_byte_size\030\002 \003(\005\"c\n\037GraphTran" +
+      "sferGraphInputNodeInfo\022\014\n\004name\030\001 \001(\t\022\r\n\005" +
+      "shape\030\002 \003(\003\022#\n\005dtype\030\003 \001(\0162\024.tensorflow." +
+      "DataType\"d\n GraphTransferGraphOutputNode" +
+      "Info\022\014\n\004name\030\001 \001(\t\022\r\n\005shape\030\002 \003(\003\022#\n\005dty" +
+      "pe\030\003 \001(\0162\024.tensorflow.DataType\"\215\004\n\021Graph" +
+      "TransferInfo\0224\n\tnode_info\030\001 \003(\0132!.tensor" +
+      "flow.GraphTransferNodeInfo\022?\n\017const_node" +
+      "_info\030\002 \003(\0132&.tensorflow.GraphTransferCo" +
+      "nstNodeInfo\022?\n\017node_input_info\030\003 \003(\0132&.t" +
+      "ensorflow.GraphTransferNodeInputInfo\022A\n\020" +
+      "node_output_info\030\004 \003(\0132\'.tensorflow.Grap" +
+      "hTransferNodeOutputInfo\022J\n\025graph_input_n" +
+      "ode_info\030\005 \003(\0132+.tensorflow.GraphTransfe" +
+      "rGraphInputNodeInfo\022L\n\026graph_output_node" +
+      "_info\030\006 \003(\0132,.tensorflow.GraphTransferGr" +
+      "aphOutputNodeInfo\022>\n\013destination\030\007 \001(\0162)" +
+      ".tensorflow.GraphTransferInfo.Destinatio" +
+      "n\"#\n\013Destination\022\007\n\003NOP\020\000\022\013\n\007HEXAGON\020\001B\231" +
+      "\001\n\036org.tensorflow.proto.frameworkB\026Graph" +
+      "TransferInfoProtoP\001ZZgithub.com/tensorfl" +
+      "ow/tensorflow/tensorflow/go/core/framewo" +
+      "rk/graph_transfer_info_go_proto\370\001\001b\006prot" +
+      "o3"
+    };
+    descriptor = com.google.protobuf.Descriptors.FileDescriptor
+      .internalBuildGeneratedFileFrom(descriptorData,
+        new com.google.protobuf.Descriptors.FileDescriptor[] {
+          org.tensorflow.proto.framework.TypesProtos.getDescriptor(),
+        });
+    internal_static_tensorflow_GraphTransferNodeInput_descriptor =
+      getDescriptor().getMessageTypes().get(0);
+    internal_static_tensorflow_GraphTransferNodeInput_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_GraphTransferNodeInput_descriptor,
+        new java.lang.String[] { "NodeId", "OutputPort", });
+    internal_static_tensorflow_GraphTransferNodeInfo_descriptor =
+      getDescriptor().getMessageTypes().get(1);
+    internal_static_tensorflow_GraphTransferNodeInfo_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_GraphTransferNodeInfo_descriptor,
+        new java.lang.String[] { "Name", "NodeId", "TypeName", "SocOpId", "PaddingId", "InputCount", "OutputCount", });
+    internal_static_tensorflow_GraphTransferConstNodeInfo_descriptor =
+      getDescriptor().getMessageTypes().get(2);
+    internal_static_tensorflow_GraphTransferConstNodeInfo_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_GraphTransferConstNodeInfo_descriptor,
+        new java.lang.String[] { "Name", "NodeId", "Shape", "Data", "Dtype", });
+    internal_static_tensorflow_GraphTransferNodeInputInfo_descriptor =
+      getDescriptor().getMessageTypes().get(3);
+    internal_static_tensorflow_GraphTransferNodeInputInfo_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_GraphTransferNodeInputInfo_descriptor,
+        new java.lang.String[] { "NodeId", "NodeInput", });
+    internal_static_tensorflow_GraphTransferNodeOutputInfo_descriptor =
+      getDescriptor().getMessageTypes().get(4);
+    internal_static_tensorflow_GraphTransferNodeOutputInfo_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_GraphTransferNodeOutputInfo_descriptor,
+        new java.lang.String[] { "NodeId", "MaxByteSize", });
+    internal_static_tensorflow_GraphTransferGraphInputNodeInfo_descriptor =
+      getDescriptor().getMessageTypes().get(5);
+    internal_static_tensorflow_GraphTransferGraphInputNodeInfo_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_GraphTransferGraphInputNodeInfo_descriptor,
+        new java.lang.String[] { "Name", "Shape", "Dtype", });
+    internal_static_tensorflow_GraphTransferGraphOutputNodeInfo_descriptor =
+      getDescriptor().getMessageTypes().get(6);
+    internal_static_tensorflow_GraphTransferGraphOutputNodeInfo_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_GraphTransferGraphOutputNodeInfo_descriptor,
+        new java.lang.String[] { "Name", "Shape", "Dtype", });
+    internal_static_tensorflow_GraphTransferInfo_descriptor =
+      getDescriptor().getMessageTypes().get(7);
+    internal_static_tensorflow_GraphTransferInfo_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_GraphTransferInfo_descriptor,
+        new java.lang.String[] { "NodeInfo", "ConstNodeInfo", "NodeInputInfo", "NodeOutputInfo", "GraphInputNodeInfo", "GraphOutputNodeInfo", "Destination", });
+    org.tensorflow.proto.framework.TypesProtos.getDescriptor();
+  }
+
+  // @@protoc_insertion_point(outer_class_scope)
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphTransferNodeInfo.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphTransferNodeInfo.java
new file mode 100644
index 00000000000..d144359008c
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphTransferNodeInfo.java
@@ -0,0 +1,958 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/framework/graph_transfer_info.proto
+
+package org.tensorflow.proto.framework;
+
+/**
+ * Protobuf type {@code tensorflow.GraphTransferNodeInfo}
+ */
+public  final class GraphTransferNodeInfo extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.GraphTransferNodeInfo)
+    GraphTransferNodeInfoOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use GraphTransferNodeInfo.newBuilder() to construct.
+  private GraphTransferNodeInfo(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private GraphTransferNodeInfo() {
+    name_ = "";
+    typeName_ = "";
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new GraphTransferNodeInfo();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private GraphTransferNodeInfo(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 10: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            name_ = s;
+            break;
+          }
+          case 16: {
+
+            nodeId_ = input.readInt32();
+            break;
+          }
+          case 26: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            typeName_ = s;
+            break;
+          }
+          case 32: {
+
+            socOpId_ = input.readInt32();
+            break;
+          }
+          case 40: {
+
+            paddingId_ = input.readInt32();
+            break;
+          }
+          case 48: {
+
+            inputCount_ = input.readInt32();
+            break;
+          }
+          case 56: {
+
+            outputCount_ = input.readInt32();
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.framework.GraphTransferInfoProto.internal_static_tensorflow_GraphTransferNodeInfo_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.framework.GraphTransferInfoProto.internal_static_tensorflow_GraphTransferNodeInfo_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.framework.GraphTransferNodeInfo.class, org.tensorflow.proto.framework.GraphTransferNodeInfo.Builder.class);
+  }
+
+  public static final int NAME_FIELD_NUMBER = 1;
+  private volatile java.lang.Object name_;
+  /**
+   * <code>string name = 1;</code>
+   */
+  public java.lang.String getName() {
+    java.lang.Object ref = name_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      name_ = s;
+      return s;
+    }
+  }
+  /**
+   * <code>string name = 1;</code>
+   */
+  public com.google.protobuf.ByteString
+      getNameBytes() {
+    java.lang.Object ref = name_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      name_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int NODE_ID_FIELD_NUMBER = 2;
+  private int nodeId_;
+  /**
+   * <code>int32 node_id = 2;</code>
+   */
+  public int getNodeId() {
+    return nodeId_;
+  }
+
+  public static final int TYPE_NAME_FIELD_NUMBER = 3;
+  private volatile java.lang.Object typeName_;
+  /**
+   * <code>string type_name = 3;</code>
+   */
+  public java.lang.String getTypeName() {
+    java.lang.Object ref = typeName_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      typeName_ = s;
+      return s;
+    }
+  }
+  /**
+   * <code>string type_name = 3;</code>
+   */
+  public com.google.protobuf.ByteString
+      getTypeNameBytes() {
+    java.lang.Object ref = typeName_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      typeName_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int SOC_OP_ID_FIELD_NUMBER = 4;
+  private int socOpId_;
+  /**
+   * <code>int32 soc_op_id = 4;</code>
+   */
+  public int getSocOpId() {
+    return socOpId_;
+  }
+
+  public static final int PADDING_ID_FIELD_NUMBER = 5;
+  private int paddingId_;
+  /**
+   * <code>int32 padding_id = 5;</code>
+   */
+  public int getPaddingId() {
+    return paddingId_;
+  }
+
+  public static final int INPUT_COUNT_FIELD_NUMBER = 6;
+  private int inputCount_;
+  /**
+   * <code>int32 input_count = 6;</code>
+   */
+  public int getInputCount() {
+    return inputCount_;
+  }
+
+  public static final int OUTPUT_COUNT_FIELD_NUMBER = 7;
+  private int outputCount_;
+  /**
+   * <code>int32 output_count = 7;</code>
+   */
+  public int getOutputCount() {
+    return outputCount_;
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    if (!getNameBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 1, name_);
+    }
+    if (nodeId_ != 0) {
+      output.writeInt32(2, nodeId_);
+    }
+    if (!getTypeNameBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 3, typeName_);
+    }
+    if (socOpId_ != 0) {
+      output.writeInt32(4, socOpId_);
+    }
+    if (paddingId_ != 0) {
+      output.writeInt32(5, paddingId_);
+    }
+    if (inputCount_ != 0) {
+      output.writeInt32(6, inputCount_);
+    }
+    if (outputCount_ != 0) {
+      output.writeInt32(7, outputCount_);
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (!getNameBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, name_);
+    }
+    if (nodeId_ != 0) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt32Size(2, nodeId_);
+    }
+    if (!getTypeNameBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(3, typeName_);
+    }
+    if (socOpId_ != 0) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt32Size(4, socOpId_);
+    }
+    if (paddingId_ != 0) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt32Size(5, paddingId_);
+    }
+    if (inputCount_ != 0) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt32Size(6, inputCount_);
+    }
+    if (outputCount_ != 0) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt32Size(7, outputCount_);
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.framework.GraphTransferNodeInfo)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.framework.GraphTransferNodeInfo other = (org.tensorflow.proto.framework.GraphTransferNodeInfo) obj;
+
+    if (!getName()
+        .equals(other.getName())) return false;
+    if (getNodeId()
+        != other.getNodeId()) return false;
+    if (!getTypeName()
+        .equals(other.getTypeName())) return false;
+    if (getSocOpId()
+        != other.getSocOpId()) return false;
+    if (getPaddingId()
+        != other.getPaddingId()) return false;
+    if (getInputCount()
+        != other.getInputCount()) return false;
+    if (getOutputCount()
+        != other.getOutputCount()) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    hash = (37 * hash) + NAME_FIELD_NUMBER;
+    hash = (53 * hash) + getName().hashCode();
+    hash = (37 * hash) + NODE_ID_FIELD_NUMBER;
+    hash = (53 * hash) + getNodeId();
+    hash = (37 * hash) + TYPE_NAME_FIELD_NUMBER;
+    hash = (53 * hash) + getTypeName().hashCode();
+    hash = (37 * hash) + SOC_OP_ID_FIELD_NUMBER;
+    hash = (53 * hash) + getSocOpId();
+    hash = (37 * hash) + PADDING_ID_FIELD_NUMBER;
+    hash = (53 * hash) + getPaddingId();
+    hash = (37 * hash) + INPUT_COUNT_FIELD_NUMBER;
+    hash = (53 * hash) + getInputCount();
+    hash = (37 * hash) + OUTPUT_COUNT_FIELD_NUMBER;
+    hash = (53 * hash) + getOutputCount();
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.framework.GraphTransferNodeInfo parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.GraphTransferNodeInfo parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.GraphTransferNodeInfo parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.GraphTransferNodeInfo parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.GraphTransferNodeInfo parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.GraphTransferNodeInfo parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.GraphTransferNodeInfo parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.GraphTransferNodeInfo parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.GraphTransferNodeInfo parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.GraphTransferNodeInfo parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.GraphTransferNodeInfo parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.GraphTransferNodeInfo parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.framework.GraphTransferNodeInfo prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * Protobuf type {@code tensorflow.GraphTransferNodeInfo}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.GraphTransferNodeInfo)
+      org.tensorflow.proto.framework.GraphTransferNodeInfoOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.GraphTransferInfoProto.internal_static_tensorflow_GraphTransferNodeInfo_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.GraphTransferInfoProto.internal_static_tensorflow_GraphTransferNodeInfo_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.GraphTransferNodeInfo.class, org.tensorflow.proto.framework.GraphTransferNodeInfo.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.framework.GraphTransferNodeInfo.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      name_ = "";
+
+      nodeId_ = 0;
+
+      typeName_ = "";
+
+      socOpId_ = 0;
+
+      paddingId_ = 0;
+
+      inputCount_ = 0;
+
+      outputCount_ = 0;
+
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.framework.GraphTransferInfoProto.internal_static_tensorflow_GraphTransferNodeInfo_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.GraphTransferNodeInfo getDefaultInstanceForType() {
+      return org.tensorflow.proto.framework.GraphTransferNodeInfo.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.GraphTransferNodeInfo build() {
+      org.tensorflow.proto.framework.GraphTransferNodeInfo result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.GraphTransferNodeInfo buildPartial() {
+      org.tensorflow.proto.framework.GraphTransferNodeInfo result = new org.tensorflow.proto.framework.GraphTransferNodeInfo(this);
+      result.name_ = name_;
+      result.nodeId_ = nodeId_;
+      result.typeName_ = typeName_;
+      result.socOpId_ = socOpId_;
+      result.paddingId_ = paddingId_;
+      result.inputCount_ = inputCount_;
+      result.outputCount_ = outputCount_;
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.framework.GraphTransferNodeInfo) {
+        return mergeFrom((org.tensorflow.proto.framework.GraphTransferNodeInfo)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.framework.GraphTransferNodeInfo other) {
+      if (other == org.tensorflow.proto.framework.GraphTransferNodeInfo.getDefaultInstance()) return this;
+      if (!other.getName().isEmpty()) {
+        name_ = other.name_;
+        onChanged();
+      }
+      if (other.getNodeId() != 0) {
+        setNodeId(other.getNodeId());
+      }
+      if (!other.getTypeName().isEmpty()) {
+        typeName_ = other.typeName_;
+        onChanged();
+      }
+      if (other.getSocOpId() != 0) {
+        setSocOpId(other.getSocOpId());
+      }
+      if (other.getPaddingId() != 0) {
+        setPaddingId(other.getPaddingId());
+      }
+      if (other.getInputCount() != 0) {
+        setInputCount(other.getInputCount());
+      }
+      if (other.getOutputCount() != 0) {
+        setOutputCount(other.getOutputCount());
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.framework.GraphTransferNodeInfo parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.framework.GraphTransferNodeInfo) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+
+    private java.lang.Object name_ = "";
+    /**
+     * <code>string name = 1;</code>
+     */
+    public java.lang.String getName() {
+      java.lang.Object ref = name_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        name_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <code>string name = 1;</code>
+     */
+    public com.google.protobuf.ByteString
+        getNameBytes() {
+      java.lang.Object ref = name_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        name_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <code>string name = 1;</code>
+     */
+    public Builder setName(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      name_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>string name = 1;</code>
+     */
+    public Builder clearName() {
+      
+      name_ = getDefaultInstance().getName();
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>string name = 1;</code>
+     */
+    public Builder setNameBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      name_ = value;
+      onChanged();
+      return this;
+    }
+
+    private int nodeId_ ;
+    /**
+     * <code>int32 node_id = 2;</code>
+     */
+    public int getNodeId() {
+      return nodeId_;
+    }
+    /**
+     * <code>int32 node_id = 2;</code>
+     */
+    public Builder setNodeId(int value) {
+      
+      nodeId_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>int32 node_id = 2;</code>
+     */
+    public Builder clearNodeId() {
+      
+      nodeId_ = 0;
+      onChanged();
+      return this;
+    }
+
+    private java.lang.Object typeName_ = "";
+    /**
+     * <code>string type_name = 3;</code>
+     */
+    public java.lang.String getTypeName() {
+      java.lang.Object ref = typeName_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        typeName_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <code>string type_name = 3;</code>
+     */
+    public com.google.protobuf.ByteString
+        getTypeNameBytes() {
+      java.lang.Object ref = typeName_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        typeName_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <code>string type_name = 3;</code>
+     */
+    public Builder setTypeName(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      typeName_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>string type_name = 3;</code>
+     */
+    public Builder clearTypeName() {
+      
+      typeName_ = getDefaultInstance().getTypeName();
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>string type_name = 3;</code>
+     */
+    public Builder setTypeNameBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      typeName_ = value;
+      onChanged();
+      return this;
+    }
+
+    private int socOpId_ ;
+    /**
+     * <code>int32 soc_op_id = 4;</code>
+     */
+    public int getSocOpId() {
+      return socOpId_;
+    }
+    /**
+     * <code>int32 soc_op_id = 4;</code>
+     */
+    public Builder setSocOpId(int value) {
+      
+      socOpId_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>int32 soc_op_id = 4;</code>
+     */
+    public Builder clearSocOpId() {
+      
+      socOpId_ = 0;
+      onChanged();
+      return this;
+    }
+
+    private int paddingId_ ;
+    /**
+     * <code>int32 padding_id = 5;</code>
+     */
+    public int getPaddingId() {
+      return paddingId_;
+    }
+    /**
+     * <code>int32 padding_id = 5;</code>
+     */
+    public Builder setPaddingId(int value) {
+      
+      paddingId_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>int32 padding_id = 5;</code>
+     */
+    public Builder clearPaddingId() {
+      
+      paddingId_ = 0;
+      onChanged();
+      return this;
+    }
+
+    private int inputCount_ ;
+    /**
+     * <code>int32 input_count = 6;</code>
+     */
+    public int getInputCount() {
+      return inputCount_;
+    }
+    /**
+     * <code>int32 input_count = 6;</code>
+     */
+    public Builder setInputCount(int value) {
+      
+      inputCount_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>int32 input_count = 6;</code>
+     */
+    public Builder clearInputCount() {
+      
+      inputCount_ = 0;
+      onChanged();
+      return this;
+    }
+
+    private int outputCount_ ;
+    /**
+     * <code>int32 output_count = 7;</code>
+     */
+    public int getOutputCount() {
+      return outputCount_;
+    }
+    /**
+     * <code>int32 output_count = 7;</code>
+     */
+    public Builder setOutputCount(int value) {
+      
+      outputCount_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>int32 output_count = 7;</code>
+     */
+    public Builder clearOutputCount() {
+      
+      outputCount_ = 0;
+      onChanged();
+      return this;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.GraphTransferNodeInfo)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.GraphTransferNodeInfo)
+  private static final org.tensorflow.proto.framework.GraphTransferNodeInfo DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.GraphTransferNodeInfo();
+  }
+
+  public static org.tensorflow.proto.framework.GraphTransferNodeInfo getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<GraphTransferNodeInfo>
+      PARSER = new com.google.protobuf.AbstractParser<GraphTransferNodeInfo>() {
+    @java.lang.Override
+    public GraphTransferNodeInfo parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new GraphTransferNodeInfo(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<GraphTransferNodeInfo> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<GraphTransferNodeInfo> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.framework.GraphTransferNodeInfo getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphTransferNodeInfoOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphTransferNodeInfoOrBuilder.java
new file mode 100644
index 00000000000..98656c6ee2c
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphTransferNodeInfoOrBuilder.java
@@ -0,0 +1,54 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/framework/graph_transfer_info.proto
+
+package org.tensorflow.proto.framework;
+
+public interface GraphTransferNodeInfoOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.GraphTransferNodeInfo)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <code>string name = 1;</code>
+   */
+  java.lang.String getName();
+  /**
+   * <code>string name = 1;</code>
+   */
+  com.google.protobuf.ByteString
+      getNameBytes();
+
+  /**
+   * <code>int32 node_id = 2;</code>
+   */
+  int getNodeId();
+
+  /**
+   * <code>string type_name = 3;</code>
+   */
+  java.lang.String getTypeName();
+  /**
+   * <code>string type_name = 3;</code>
+   */
+  com.google.protobuf.ByteString
+      getTypeNameBytes();
+
+  /**
+   * <code>int32 soc_op_id = 4;</code>
+   */
+  int getSocOpId();
+
+  /**
+   * <code>int32 padding_id = 5;</code>
+   */
+  int getPaddingId();
+
+  /**
+   * <code>int32 input_count = 6;</code>
+   */
+  int getInputCount();
+
+  /**
+   * <code>int32 output_count = 7;</code>
+   */
+  int getOutputCount();
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphTransferNodeInput.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphTransferNodeInput.java
new file mode 100644
index 00000000000..77ad6b87dc8
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphTransferNodeInput.java
@@ -0,0 +1,533 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/framework/graph_transfer_info.proto
+
+package org.tensorflow.proto.framework;
+
+/**
+ * Protobuf type {@code tensorflow.GraphTransferNodeInput}
+ */
+public  final class GraphTransferNodeInput extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.GraphTransferNodeInput)
+    GraphTransferNodeInputOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use GraphTransferNodeInput.newBuilder() to construct.
+  private GraphTransferNodeInput(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private GraphTransferNodeInput() {
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new GraphTransferNodeInput();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private GraphTransferNodeInput(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 8: {
+
+            nodeId_ = input.readInt32();
+            break;
+          }
+          case 16: {
+
+            outputPort_ = input.readInt32();
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.framework.GraphTransferInfoProto.internal_static_tensorflow_GraphTransferNodeInput_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.framework.GraphTransferInfoProto.internal_static_tensorflow_GraphTransferNodeInput_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.framework.GraphTransferNodeInput.class, org.tensorflow.proto.framework.GraphTransferNodeInput.Builder.class);
+  }
+
+  public static final int NODE_ID_FIELD_NUMBER = 1;
+  private int nodeId_;
+  /**
+   * <code>int32 node_id = 1;</code>
+   */
+  public int getNodeId() {
+    return nodeId_;
+  }
+
+  public static final int OUTPUT_PORT_FIELD_NUMBER = 2;
+  private int outputPort_;
+  /**
+   * <code>int32 output_port = 2;</code>
+   */
+  public int getOutputPort() {
+    return outputPort_;
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    if (nodeId_ != 0) {
+      output.writeInt32(1, nodeId_);
+    }
+    if (outputPort_ != 0) {
+      output.writeInt32(2, outputPort_);
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (nodeId_ != 0) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt32Size(1, nodeId_);
+    }
+    if (outputPort_ != 0) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt32Size(2, outputPort_);
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.framework.GraphTransferNodeInput)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.framework.GraphTransferNodeInput other = (org.tensorflow.proto.framework.GraphTransferNodeInput) obj;
+
+    if (getNodeId()
+        != other.getNodeId()) return false;
+    if (getOutputPort()
+        != other.getOutputPort()) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    hash = (37 * hash) + NODE_ID_FIELD_NUMBER;
+    hash = (53 * hash) + getNodeId();
+    hash = (37 * hash) + OUTPUT_PORT_FIELD_NUMBER;
+    hash = (53 * hash) + getOutputPort();
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.framework.GraphTransferNodeInput parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.GraphTransferNodeInput parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.GraphTransferNodeInput parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.GraphTransferNodeInput parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.GraphTransferNodeInput parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.GraphTransferNodeInput parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.GraphTransferNodeInput parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.GraphTransferNodeInput parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.GraphTransferNodeInput parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.GraphTransferNodeInput parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.GraphTransferNodeInput parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.GraphTransferNodeInput parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.framework.GraphTransferNodeInput prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * Protobuf type {@code tensorflow.GraphTransferNodeInput}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.GraphTransferNodeInput)
+      org.tensorflow.proto.framework.GraphTransferNodeInputOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.GraphTransferInfoProto.internal_static_tensorflow_GraphTransferNodeInput_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.GraphTransferInfoProto.internal_static_tensorflow_GraphTransferNodeInput_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.GraphTransferNodeInput.class, org.tensorflow.proto.framework.GraphTransferNodeInput.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.framework.GraphTransferNodeInput.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      nodeId_ = 0;
+
+      outputPort_ = 0;
+
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.framework.GraphTransferInfoProto.internal_static_tensorflow_GraphTransferNodeInput_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.GraphTransferNodeInput getDefaultInstanceForType() {
+      return org.tensorflow.proto.framework.GraphTransferNodeInput.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.GraphTransferNodeInput build() {
+      org.tensorflow.proto.framework.GraphTransferNodeInput result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.GraphTransferNodeInput buildPartial() {
+      org.tensorflow.proto.framework.GraphTransferNodeInput result = new org.tensorflow.proto.framework.GraphTransferNodeInput(this);
+      result.nodeId_ = nodeId_;
+      result.outputPort_ = outputPort_;
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.framework.GraphTransferNodeInput) {
+        return mergeFrom((org.tensorflow.proto.framework.GraphTransferNodeInput)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.framework.GraphTransferNodeInput other) {
+      if (other == org.tensorflow.proto.framework.GraphTransferNodeInput.getDefaultInstance()) return this;
+      if (other.getNodeId() != 0) {
+        setNodeId(other.getNodeId());
+      }
+      if (other.getOutputPort() != 0) {
+        setOutputPort(other.getOutputPort());
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.framework.GraphTransferNodeInput parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.framework.GraphTransferNodeInput) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+
+    private int nodeId_ ;
+    /**
+     * <code>int32 node_id = 1;</code>
+     */
+    public int getNodeId() {
+      return nodeId_;
+    }
+    /**
+     * <code>int32 node_id = 1;</code>
+     */
+    public Builder setNodeId(int value) {
+      
+      nodeId_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>int32 node_id = 1;</code>
+     */
+    public Builder clearNodeId() {
+      
+      nodeId_ = 0;
+      onChanged();
+      return this;
+    }
+
+    private int outputPort_ ;
+    /**
+     * <code>int32 output_port = 2;</code>
+     */
+    public int getOutputPort() {
+      return outputPort_;
+    }
+    /**
+     * <code>int32 output_port = 2;</code>
+     */
+    public Builder setOutputPort(int value) {
+      
+      outputPort_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>int32 output_port = 2;</code>
+     */
+    public Builder clearOutputPort() {
+      
+      outputPort_ = 0;
+      onChanged();
+      return this;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.GraphTransferNodeInput)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.GraphTransferNodeInput)
+  private static final org.tensorflow.proto.framework.GraphTransferNodeInput DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.GraphTransferNodeInput();
+  }
+
+  public static org.tensorflow.proto.framework.GraphTransferNodeInput getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<GraphTransferNodeInput>
+      PARSER = new com.google.protobuf.AbstractParser<GraphTransferNodeInput>() {
+    @java.lang.Override
+    public GraphTransferNodeInput parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new GraphTransferNodeInput(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<GraphTransferNodeInput> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<GraphTransferNodeInput> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.framework.GraphTransferNodeInput getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphTransferNodeInputInfo.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphTransferNodeInputInfo.java
new file mode 100644
index 00000000000..71d0552eacd
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphTransferNodeInputInfo.java
@@ -0,0 +1,822 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/framework/graph_transfer_info.proto
+
+package org.tensorflow.proto.framework;
+
+/**
+ * Protobuf type {@code tensorflow.GraphTransferNodeInputInfo}
+ */
+public  final class GraphTransferNodeInputInfo extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.GraphTransferNodeInputInfo)
+    GraphTransferNodeInputInfoOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use GraphTransferNodeInputInfo.newBuilder() to construct.
+  private GraphTransferNodeInputInfo(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private GraphTransferNodeInputInfo() {
+    nodeInput_ = java.util.Collections.emptyList();
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new GraphTransferNodeInputInfo();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private GraphTransferNodeInputInfo(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    int mutable_bitField0_ = 0;
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 8: {
+
+            nodeId_ = input.readInt32();
+            break;
+          }
+          case 18: {
+            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
+              nodeInput_ = new java.util.ArrayList<org.tensorflow.proto.framework.GraphTransferNodeInput>();
+              mutable_bitField0_ |= 0x00000001;
+            }
+            nodeInput_.add(
+                input.readMessage(org.tensorflow.proto.framework.GraphTransferNodeInput.parser(), extensionRegistry));
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      if (((mutable_bitField0_ & 0x00000001) != 0)) {
+        nodeInput_ = java.util.Collections.unmodifiableList(nodeInput_);
+      }
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.framework.GraphTransferInfoProto.internal_static_tensorflow_GraphTransferNodeInputInfo_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.framework.GraphTransferInfoProto.internal_static_tensorflow_GraphTransferNodeInputInfo_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.framework.GraphTransferNodeInputInfo.class, org.tensorflow.proto.framework.GraphTransferNodeInputInfo.Builder.class);
+  }
+
+  public static final int NODE_ID_FIELD_NUMBER = 1;
+  private int nodeId_;
+  /**
+   * <code>int32 node_id = 1;</code>
+   */
+  public int getNodeId() {
+    return nodeId_;
+  }
+
+  public static final int NODE_INPUT_FIELD_NUMBER = 2;
+  private java.util.List<org.tensorflow.proto.framework.GraphTransferNodeInput> nodeInput_;
+  /**
+   * <code>repeated .tensorflow.GraphTransferNodeInput node_input = 2;</code>
+   */
+  public java.util.List<org.tensorflow.proto.framework.GraphTransferNodeInput> getNodeInputList() {
+    return nodeInput_;
+  }
+  /**
+   * <code>repeated .tensorflow.GraphTransferNodeInput node_input = 2;</code>
+   */
+  public java.util.List<? extends org.tensorflow.proto.framework.GraphTransferNodeInputOrBuilder> 
+      getNodeInputOrBuilderList() {
+    return nodeInput_;
+  }
+  /**
+   * <code>repeated .tensorflow.GraphTransferNodeInput node_input = 2;</code>
+   */
+  public int getNodeInputCount() {
+    return nodeInput_.size();
+  }
+  /**
+   * <code>repeated .tensorflow.GraphTransferNodeInput node_input = 2;</code>
+   */
+  public org.tensorflow.proto.framework.GraphTransferNodeInput getNodeInput(int index) {
+    return nodeInput_.get(index);
+  }
+  /**
+   * <code>repeated .tensorflow.GraphTransferNodeInput node_input = 2;</code>
+   */
+  public org.tensorflow.proto.framework.GraphTransferNodeInputOrBuilder getNodeInputOrBuilder(
+      int index) {
+    return nodeInput_.get(index);
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    if (nodeId_ != 0) {
+      output.writeInt32(1, nodeId_);
+    }
+    for (int i = 0; i < nodeInput_.size(); i++) {
+      output.writeMessage(2, nodeInput_.get(i));
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (nodeId_ != 0) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt32Size(1, nodeId_);
+    }
+    for (int i = 0; i < nodeInput_.size(); i++) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(2, nodeInput_.get(i));
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.framework.GraphTransferNodeInputInfo)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.framework.GraphTransferNodeInputInfo other = (org.tensorflow.proto.framework.GraphTransferNodeInputInfo) obj;
+
+    if (getNodeId()
+        != other.getNodeId()) return false;
+    if (!getNodeInputList()
+        .equals(other.getNodeInputList())) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    hash = (37 * hash) + NODE_ID_FIELD_NUMBER;
+    hash = (53 * hash) + getNodeId();
+    if (getNodeInputCount() > 0) {
+      hash = (37 * hash) + NODE_INPUT_FIELD_NUMBER;
+      hash = (53 * hash) + getNodeInputList().hashCode();
+    }
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.framework.GraphTransferNodeInputInfo parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.GraphTransferNodeInputInfo parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.GraphTransferNodeInputInfo parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.GraphTransferNodeInputInfo parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.GraphTransferNodeInputInfo parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.GraphTransferNodeInputInfo parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.GraphTransferNodeInputInfo parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.GraphTransferNodeInputInfo parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.GraphTransferNodeInputInfo parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.GraphTransferNodeInputInfo parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.GraphTransferNodeInputInfo parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.GraphTransferNodeInputInfo parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.framework.GraphTransferNodeInputInfo prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * Protobuf type {@code tensorflow.GraphTransferNodeInputInfo}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.GraphTransferNodeInputInfo)
+      org.tensorflow.proto.framework.GraphTransferNodeInputInfoOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.GraphTransferInfoProto.internal_static_tensorflow_GraphTransferNodeInputInfo_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.GraphTransferInfoProto.internal_static_tensorflow_GraphTransferNodeInputInfo_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.GraphTransferNodeInputInfo.class, org.tensorflow.proto.framework.GraphTransferNodeInputInfo.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.framework.GraphTransferNodeInputInfo.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+        getNodeInputFieldBuilder();
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      nodeId_ = 0;
+
+      if (nodeInputBuilder_ == null) {
+        nodeInput_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000001);
+      } else {
+        nodeInputBuilder_.clear();
+      }
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.framework.GraphTransferInfoProto.internal_static_tensorflow_GraphTransferNodeInputInfo_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.GraphTransferNodeInputInfo getDefaultInstanceForType() {
+      return org.tensorflow.proto.framework.GraphTransferNodeInputInfo.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.GraphTransferNodeInputInfo build() {
+      org.tensorflow.proto.framework.GraphTransferNodeInputInfo result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.GraphTransferNodeInputInfo buildPartial() {
+      org.tensorflow.proto.framework.GraphTransferNodeInputInfo result = new org.tensorflow.proto.framework.GraphTransferNodeInputInfo(this);
+      int from_bitField0_ = bitField0_;
+      result.nodeId_ = nodeId_;
+      if (nodeInputBuilder_ == null) {
+        if (((bitField0_ & 0x00000001) != 0)) {
+          nodeInput_ = java.util.Collections.unmodifiableList(nodeInput_);
+          bitField0_ = (bitField0_ & ~0x00000001);
+        }
+        result.nodeInput_ = nodeInput_;
+      } else {
+        result.nodeInput_ = nodeInputBuilder_.build();
+      }
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.framework.GraphTransferNodeInputInfo) {
+        return mergeFrom((org.tensorflow.proto.framework.GraphTransferNodeInputInfo)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.framework.GraphTransferNodeInputInfo other) {
+      if (other == org.tensorflow.proto.framework.GraphTransferNodeInputInfo.getDefaultInstance()) return this;
+      if (other.getNodeId() != 0) {
+        setNodeId(other.getNodeId());
+      }
+      if (nodeInputBuilder_ == null) {
+        if (!other.nodeInput_.isEmpty()) {
+          if (nodeInput_.isEmpty()) {
+            nodeInput_ = other.nodeInput_;
+            bitField0_ = (bitField0_ & ~0x00000001);
+          } else {
+            ensureNodeInputIsMutable();
+            nodeInput_.addAll(other.nodeInput_);
+          }
+          onChanged();
+        }
+      } else {
+        if (!other.nodeInput_.isEmpty()) {
+          if (nodeInputBuilder_.isEmpty()) {
+            nodeInputBuilder_.dispose();
+            nodeInputBuilder_ = null;
+            nodeInput_ = other.nodeInput_;
+            bitField0_ = (bitField0_ & ~0x00000001);
+            nodeInputBuilder_ = 
+              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
+                 getNodeInputFieldBuilder() : null;
+          } else {
+            nodeInputBuilder_.addAllMessages(other.nodeInput_);
+          }
+        }
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.framework.GraphTransferNodeInputInfo parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.framework.GraphTransferNodeInputInfo) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+    private int bitField0_;
+
+    private int nodeId_ ;
+    /**
+     * <code>int32 node_id = 1;</code>
+     */
+    public int getNodeId() {
+      return nodeId_;
+    }
+    /**
+     * <code>int32 node_id = 1;</code>
+     */
+    public Builder setNodeId(int value) {
+      
+      nodeId_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>int32 node_id = 1;</code>
+     */
+    public Builder clearNodeId() {
+      
+      nodeId_ = 0;
+      onChanged();
+      return this;
+    }
+
+    private java.util.List<org.tensorflow.proto.framework.GraphTransferNodeInput> nodeInput_ =
+      java.util.Collections.emptyList();
+    private void ensureNodeInputIsMutable() {
+      if (!((bitField0_ & 0x00000001) != 0)) {
+        nodeInput_ = new java.util.ArrayList<org.tensorflow.proto.framework.GraphTransferNodeInput>(nodeInput_);
+        bitField0_ |= 0x00000001;
+       }
+    }
+
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.framework.GraphTransferNodeInput, org.tensorflow.proto.framework.GraphTransferNodeInput.Builder, org.tensorflow.proto.framework.GraphTransferNodeInputOrBuilder> nodeInputBuilder_;
+
+    /**
+     * <code>repeated .tensorflow.GraphTransferNodeInput node_input = 2;</code>
+     */
+    public java.util.List<org.tensorflow.proto.framework.GraphTransferNodeInput> getNodeInputList() {
+      if (nodeInputBuilder_ == null) {
+        return java.util.Collections.unmodifiableList(nodeInput_);
+      } else {
+        return nodeInputBuilder_.getMessageList();
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.GraphTransferNodeInput node_input = 2;</code>
+     */
+    public int getNodeInputCount() {
+      if (nodeInputBuilder_ == null) {
+        return nodeInput_.size();
+      } else {
+        return nodeInputBuilder_.getCount();
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.GraphTransferNodeInput node_input = 2;</code>
+     */
+    public org.tensorflow.proto.framework.GraphTransferNodeInput getNodeInput(int index) {
+      if (nodeInputBuilder_ == null) {
+        return nodeInput_.get(index);
+      } else {
+        return nodeInputBuilder_.getMessage(index);
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.GraphTransferNodeInput node_input = 2;</code>
+     */
+    public Builder setNodeInput(
+        int index, org.tensorflow.proto.framework.GraphTransferNodeInput value) {
+      if (nodeInputBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureNodeInputIsMutable();
+        nodeInput_.set(index, value);
+        onChanged();
+      } else {
+        nodeInputBuilder_.setMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.GraphTransferNodeInput node_input = 2;</code>
+     */
+    public Builder setNodeInput(
+        int index, org.tensorflow.proto.framework.GraphTransferNodeInput.Builder builderForValue) {
+      if (nodeInputBuilder_ == null) {
+        ensureNodeInputIsMutable();
+        nodeInput_.set(index, builderForValue.build());
+        onChanged();
+      } else {
+        nodeInputBuilder_.setMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.GraphTransferNodeInput node_input = 2;</code>
+     */
+    public Builder addNodeInput(org.tensorflow.proto.framework.GraphTransferNodeInput value) {
+      if (nodeInputBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureNodeInputIsMutable();
+        nodeInput_.add(value);
+        onChanged();
+      } else {
+        nodeInputBuilder_.addMessage(value);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.GraphTransferNodeInput node_input = 2;</code>
+     */
+    public Builder addNodeInput(
+        int index, org.tensorflow.proto.framework.GraphTransferNodeInput value) {
+      if (nodeInputBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureNodeInputIsMutable();
+        nodeInput_.add(index, value);
+        onChanged();
+      } else {
+        nodeInputBuilder_.addMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.GraphTransferNodeInput node_input = 2;</code>
+     */
+    public Builder addNodeInput(
+        org.tensorflow.proto.framework.GraphTransferNodeInput.Builder builderForValue) {
+      if (nodeInputBuilder_ == null) {
+        ensureNodeInputIsMutable();
+        nodeInput_.add(builderForValue.build());
+        onChanged();
+      } else {
+        nodeInputBuilder_.addMessage(builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.GraphTransferNodeInput node_input = 2;</code>
+     */
+    public Builder addNodeInput(
+        int index, org.tensorflow.proto.framework.GraphTransferNodeInput.Builder builderForValue) {
+      if (nodeInputBuilder_ == null) {
+        ensureNodeInputIsMutable();
+        nodeInput_.add(index, builderForValue.build());
+        onChanged();
+      } else {
+        nodeInputBuilder_.addMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.GraphTransferNodeInput node_input = 2;</code>
+     */
+    public Builder addAllNodeInput(
+        java.lang.Iterable<? extends org.tensorflow.proto.framework.GraphTransferNodeInput> values) {
+      if (nodeInputBuilder_ == null) {
+        ensureNodeInputIsMutable();
+        com.google.protobuf.AbstractMessageLite.Builder.addAll(
+            values, nodeInput_);
+        onChanged();
+      } else {
+        nodeInputBuilder_.addAllMessages(values);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.GraphTransferNodeInput node_input = 2;</code>
+     */
+    public Builder clearNodeInput() {
+      if (nodeInputBuilder_ == null) {
+        nodeInput_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000001);
+        onChanged();
+      } else {
+        nodeInputBuilder_.clear();
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.GraphTransferNodeInput node_input = 2;</code>
+     */
+    public Builder removeNodeInput(int index) {
+      if (nodeInputBuilder_ == null) {
+        ensureNodeInputIsMutable();
+        nodeInput_.remove(index);
+        onChanged();
+      } else {
+        nodeInputBuilder_.remove(index);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.GraphTransferNodeInput node_input = 2;</code>
+     */
+    public org.tensorflow.proto.framework.GraphTransferNodeInput.Builder getNodeInputBuilder(
+        int index) {
+      return getNodeInputFieldBuilder().getBuilder(index);
+    }
+    /**
+     * <code>repeated .tensorflow.GraphTransferNodeInput node_input = 2;</code>
+     */
+    public org.tensorflow.proto.framework.GraphTransferNodeInputOrBuilder getNodeInputOrBuilder(
+        int index) {
+      if (nodeInputBuilder_ == null) {
+        return nodeInput_.get(index);  } else {
+        return nodeInputBuilder_.getMessageOrBuilder(index);
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.GraphTransferNodeInput node_input = 2;</code>
+     */
+    public java.util.List<? extends org.tensorflow.proto.framework.GraphTransferNodeInputOrBuilder> 
+         getNodeInputOrBuilderList() {
+      if (nodeInputBuilder_ != null) {
+        return nodeInputBuilder_.getMessageOrBuilderList();
+      } else {
+        return java.util.Collections.unmodifiableList(nodeInput_);
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.GraphTransferNodeInput node_input = 2;</code>
+     */
+    public org.tensorflow.proto.framework.GraphTransferNodeInput.Builder addNodeInputBuilder() {
+      return getNodeInputFieldBuilder().addBuilder(
+          org.tensorflow.proto.framework.GraphTransferNodeInput.getDefaultInstance());
+    }
+    /**
+     * <code>repeated .tensorflow.GraphTransferNodeInput node_input = 2;</code>
+     */
+    public org.tensorflow.proto.framework.GraphTransferNodeInput.Builder addNodeInputBuilder(
+        int index) {
+      return getNodeInputFieldBuilder().addBuilder(
+          index, org.tensorflow.proto.framework.GraphTransferNodeInput.getDefaultInstance());
+    }
+    /**
+     * <code>repeated .tensorflow.GraphTransferNodeInput node_input = 2;</code>
+     */
+    public java.util.List<org.tensorflow.proto.framework.GraphTransferNodeInput.Builder> 
+         getNodeInputBuilderList() {
+      return getNodeInputFieldBuilder().getBuilderList();
+    }
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.framework.GraphTransferNodeInput, org.tensorflow.proto.framework.GraphTransferNodeInput.Builder, org.tensorflow.proto.framework.GraphTransferNodeInputOrBuilder> 
+        getNodeInputFieldBuilder() {
+      if (nodeInputBuilder_ == null) {
+        nodeInputBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
+            org.tensorflow.proto.framework.GraphTransferNodeInput, org.tensorflow.proto.framework.GraphTransferNodeInput.Builder, org.tensorflow.proto.framework.GraphTransferNodeInputOrBuilder>(
+                nodeInput_,
+                ((bitField0_ & 0x00000001) != 0),
+                getParentForChildren(),
+                isClean());
+        nodeInput_ = null;
+      }
+      return nodeInputBuilder_;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.GraphTransferNodeInputInfo)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.GraphTransferNodeInputInfo)
+  private static final org.tensorflow.proto.framework.GraphTransferNodeInputInfo DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.GraphTransferNodeInputInfo();
+  }
+
+  public static org.tensorflow.proto.framework.GraphTransferNodeInputInfo getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<GraphTransferNodeInputInfo>
+      PARSER = new com.google.protobuf.AbstractParser<GraphTransferNodeInputInfo>() {
+    @java.lang.Override
+    public GraphTransferNodeInputInfo parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new GraphTransferNodeInputInfo(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<GraphTransferNodeInputInfo> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<GraphTransferNodeInputInfo> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.framework.GraphTransferNodeInputInfo getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphTransferNodeInputInfoOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphTransferNodeInputInfoOrBuilder.java
new file mode 100644
index 00000000000..552a182d67b
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphTransferNodeInputInfoOrBuilder.java
@@ -0,0 +1,38 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/framework/graph_transfer_info.proto
+
+package org.tensorflow.proto.framework;
+
+public interface GraphTransferNodeInputInfoOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.GraphTransferNodeInputInfo)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <code>int32 node_id = 1;</code>
+   */
+  int getNodeId();
+
+  /**
+   * <code>repeated .tensorflow.GraphTransferNodeInput node_input = 2;</code>
+   */
+  java.util.List<org.tensorflow.proto.framework.GraphTransferNodeInput> 
+      getNodeInputList();
+  /**
+   * <code>repeated .tensorflow.GraphTransferNodeInput node_input = 2;</code>
+   */
+  org.tensorflow.proto.framework.GraphTransferNodeInput getNodeInput(int index);
+  /**
+   * <code>repeated .tensorflow.GraphTransferNodeInput node_input = 2;</code>
+   */
+  int getNodeInputCount();
+  /**
+   * <code>repeated .tensorflow.GraphTransferNodeInput node_input = 2;</code>
+   */
+  java.util.List<? extends org.tensorflow.proto.framework.GraphTransferNodeInputOrBuilder> 
+      getNodeInputOrBuilderList();
+  /**
+   * <code>repeated .tensorflow.GraphTransferNodeInput node_input = 2;</code>
+   */
+  org.tensorflow.proto.framework.GraphTransferNodeInputOrBuilder getNodeInputOrBuilder(
+      int index);
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphTransferNodeInputOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphTransferNodeInputOrBuilder.java
new file mode 100644
index 00000000000..971ffed8443
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphTransferNodeInputOrBuilder.java
@@ -0,0 +1,19 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/framework/graph_transfer_info.proto
+
+package org.tensorflow.proto.framework;
+
+public interface GraphTransferNodeInputOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.GraphTransferNodeInput)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <code>int32 node_id = 1;</code>
+   */
+  int getNodeId();
+
+  /**
+   * <code>int32 output_port = 2;</code>
+   */
+  int getOutputPort();
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphTransferNodeOutputInfo.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphTransferNodeOutputInfo.java
new file mode 100644
index 00000000000..c39062bebdb
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphTransferNodeOutputInfo.java
@@ -0,0 +1,639 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/framework/graph_transfer_info.proto
+
+package org.tensorflow.proto.framework;
+
+/**
+ * Protobuf type {@code tensorflow.GraphTransferNodeOutputInfo}
+ */
+public  final class GraphTransferNodeOutputInfo extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.GraphTransferNodeOutputInfo)
+    GraphTransferNodeOutputInfoOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use GraphTransferNodeOutputInfo.newBuilder() to construct.
+  private GraphTransferNodeOutputInfo(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private GraphTransferNodeOutputInfo() {
+    maxByteSize_ = emptyIntList();
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new GraphTransferNodeOutputInfo();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private GraphTransferNodeOutputInfo(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    int mutable_bitField0_ = 0;
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 8: {
+
+            nodeId_ = input.readInt32();
+            break;
+          }
+          case 16: {
+            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
+              maxByteSize_ = newIntList();
+              mutable_bitField0_ |= 0x00000001;
+            }
+            maxByteSize_.addInt(input.readInt32());
+            break;
+          }
+          case 18: {
+            int length = input.readRawVarint32();
+            int limit = input.pushLimit(length);
+            if (!((mutable_bitField0_ & 0x00000001) != 0) && input.getBytesUntilLimit() > 0) {
+              maxByteSize_ = newIntList();
+              mutable_bitField0_ |= 0x00000001;
+            }
+            while (input.getBytesUntilLimit() > 0) {
+              maxByteSize_.addInt(input.readInt32());
+            }
+            input.popLimit(limit);
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      if (((mutable_bitField0_ & 0x00000001) != 0)) {
+        maxByteSize_.makeImmutable(); // C
+      }
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.framework.GraphTransferInfoProto.internal_static_tensorflow_GraphTransferNodeOutputInfo_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.framework.GraphTransferInfoProto.internal_static_tensorflow_GraphTransferNodeOutputInfo_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.framework.GraphTransferNodeOutputInfo.class, org.tensorflow.proto.framework.GraphTransferNodeOutputInfo.Builder.class);
+  }
+
+  public static final int NODE_ID_FIELD_NUMBER = 1;
+  private int nodeId_;
+  /**
+   * <code>int32 node_id = 1;</code>
+   */
+  public int getNodeId() {
+    return nodeId_;
+  }
+
+  public static final int MAX_BYTE_SIZE_FIELD_NUMBER = 2;
+  private com.google.protobuf.Internal.IntList maxByteSize_;
+  /**
+   * <code>repeated int32 max_byte_size = 2;</code>
+   */
+  public java.util.List<java.lang.Integer>
+      getMaxByteSizeList() {
+    return maxByteSize_;
+  }
+  /**
+   * <code>repeated int32 max_byte_size = 2;</code>
+   */
+  public int getMaxByteSizeCount() {
+    return maxByteSize_.size();
+  }
+  /**
+   * <code>repeated int32 max_byte_size = 2;</code>
+   */
+  public int getMaxByteSize(int index) {
+    return maxByteSize_.getInt(index);
+  }
+  private int maxByteSizeMemoizedSerializedSize = -1;
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    getSerializedSize();
+    if (nodeId_ != 0) {
+      output.writeInt32(1, nodeId_);
+    }
+    if (getMaxByteSizeList().size() > 0) {
+      output.writeUInt32NoTag(18);
+      output.writeUInt32NoTag(maxByteSizeMemoizedSerializedSize);
+    }
+    for (int i = 0; i < maxByteSize_.size(); i++) {
+      output.writeInt32NoTag(maxByteSize_.getInt(i));
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (nodeId_ != 0) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt32Size(1, nodeId_);
+    }
+    {
+      int dataSize = 0;
+      for (int i = 0; i < maxByteSize_.size(); i++) {
+        dataSize += com.google.protobuf.CodedOutputStream
+          .computeInt32SizeNoTag(maxByteSize_.getInt(i));
+      }
+      size += dataSize;
+      if (!getMaxByteSizeList().isEmpty()) {
+        size += 1;
+        size += com.google.protobuf.CodedOutputStream
+            .computeInt32SizeNoTag(dataSize);
+      }
+      maxByteSizeMemoizedSerializedSize = dataSize;
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.framework.GraphTransferNodeOutputInfo)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.framework.GraphTransferNodeOutputInfo other = (org.tensorflow.proto.framework.GraphTransferNodeOutputInfo) obj;
+
+    if (getNodeId()
+        != other.getNodeId()) return false;
+    if (!getMaxByteSizeList()
+        .equals(other.getMaxByteSizeList())) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    hash = (37 * hash) + NODE_ID_FIELD_NUMBER;
+    hash = (53 * hash) + getNodeId();
+    if (getMaxByteSizeCount() > 0) {
+      hash = (37 * hash) + MAX_BYTE_SIZE_FIELD_NUMBER;
+      hash = (53 * hash) + getMaxByteSizeList().hashCode();
+    }
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.framework.GraphTransferNodeOutputInfo parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.GraphTransferNodeOutputInfo parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.GraphTransferNodeOutputInfo parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.GraphTransferNodeOutputInfo parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.GraphTransferNodeOutputInfo parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.GraphTransferNodeOutputInfo parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.GraphTransferNodeOutputInfo parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.GraphTransferNodeOutputInfo parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.GraphTransferNodeOutputInfo parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.GraphTransferNodeOutputInfo parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.GraphTransferNodeOutputInfo parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.GraphTransferNodeOutputInfo parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.framework.GraphTransferNodeOutputInfo prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * Protobuf type {@code tensorflow.GraphTransferNodeOutputInfo}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.GraphTransferNodeOutputInfo)
+      org.tensorflow.proto.framework.GraphTransferNodeOutputInfoOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.GraphTransferInfoProto.internal_static_tensorflow_GraphTransferNodeOutputInfo_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.GraphTransferInfoProto.internal_static_tensorflow_GraphTransferNodeOutputInfo_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.GraphTransferNodeOutputInfo.class, org.tensorflow.proto.framework.GraphTransferNodeOutputInfo.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.framework.GraphTransferNodeOutputInfo.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      nodeId_ = 0;
+
+      maxByteSize_ = emptyIntList();
+      bitField0_ = (bitField0_ & ~0x00000001);
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.framework.GraphTransferInfoProto.internal_static_tensorflow_GraphTransferNodeOutputInfo_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.GraphTransferNodeOutputInfo getDefaultInstanceForType() {
+      return org.tensorflow.proto.framework.GraphTransferNodeOutputInfo.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.GraphTransferNodeOutputInfo build() {
+      org.tensorflow.proto.framework.GraphTransferNodeOutputInfo result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.GraphTransferNodeOutputInfo buildPartial() {
+      org.tensorflow.proto.framework.GraphTransferNodeOutputInfo result = new org.tensorflow.proto.framework.GraphTransferNodeOutputInfo(this);
+      int from_bitField0_ = bitField0_;
+      result.nodeId_ = nodeId_;
+      if (((bitField0_ & 0x00000001) != 0)) {
+        maxByteSize_.makeImmutable();
+        bitField0_ = (bitField0_ & ~0x00000001);
+      }
+      result.maxByteSize_ = maxByteSize_;
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.framework.GraphTransferNodeOutputInfo) {
+        return mergeFrom((org.tensorflow.proto.framework.GraphTransferNodeOutputInfo)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.framework.GraphTransferNodeOutputInfo other) {
+      if (other == org.tensorflow.proto.framework.GraphTransferNodeOutputInfo.getDefaultInstance()) return this;
+      if (other.getNodeId() != 0) {
+        setNodeId(other.getNodeId());
+      }
+      if (!other.maxByteSize_.isEmpty()) {
+        if (maxByteSize_.isEmpty()) {
+          maxByteSize_ = other.maxByteSize_;
+          bitField0_ = (bitField0_ & ~0x00000001);
+        } else {
+          ensureMaxByteSizeIsMutable();
+          maxByteSize_.addAll(other.maxByteSize_);
+        }
+        onChanged();
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.framework.GraphTransferNodeOutputInfo parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.framework.GraphTransferNodeOutputInfo) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+    private int bitField0_;
+
+    private int nodeId_ ;
+    /**
+     * <code>int32 node_id = 1;</code>
+     */
+    public int getNodeId() {
+      return nodeId_;
+    }
+    /**
+     * <code>int32 node_id = 1;</code>
+     */
+    public Builder setNodeId(int value) {
+      
+      nodeId_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>int32 node_id = 1;</code>
+     */
+    public Builder clearNodeId() {
+      
+      nodeId_ = 0;
+      onChanged();
+      return this;
+    }
+
+    private com.google.protobuf.Internal.IntList maxByteSize_ = emptyIntList();
+    private void ensureMaxByteSizeIsMutable() {
+      if (!((bitField0_ & 0x00000001) != 0)) {
+        maxByteSize_ = mutableCopy(maxByteSize_);
+        bitField0_ |= 0x00000001;
+       }
+    }
+    /**
+     * <code>repeated int32 max_byte_size = 2;</code>
+     */
+    public java.util.List<java.lang.Integer>
+        getMaxByteSizeList() {
+      return ((bitField0_ & 0x00000001) != 0) ?
+               java.util.Collections.unmodifiableList(maxByteSize_) : maxByteSize_;
+    }
+    /**
+     * <code>repeated int32 max_byte_size = 2;</code>
+     */
+    public int getMaxByteSizeCount() {
+      return maxByteSize_.size();
+    }
+    /**
+     * <code>repeated int32 max_byte_size = 2;</code>
+     */
+    public int getMaxByteSize(int index) {
+      return maxByteSize_.getInt(index);
+    }
+    /**
+     * <code>repeated int32 max_byte_size = 2;</code>
+     */
+    public Builder setMaxByteSize(
+        int index, int value) {
+      ensureMaxByteSizeIsMutable();
+      maxByteSize_.setInt(index, value);
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>repeated int32 max_byte_size = 2;</code>
+     */
+    public Builder addMaxByteSize(int value) {
+      ensureMaxByteSizeIsMutable();
+      maxByteSize_.addInt(value);
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>repeated int32 max_byte_size = 2;</code>
+     */
+    public Builder addAllMaxByteSize(
+        java.lang.Iterable<? extends java.lang.Integer> values) {
+      ensureMaxByteSizeIsMutable();
+      com.google.protobuf.AbstractMessageLite.Builder.addAll(
+          values, maxByteSize_);
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>repeated int32 max_byte_size = 2;</code>
+     */
+    public Builder clearMaxByteSize() {
+      maxByteSize_ = emptyIntList();
+      bitField0_ = (bitField0_ & ~0x00000001);
+      onChanged();
+      return this;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.GraphTransferNodeOutputInfo)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.GraphTransferNodeOutputInfo)
+  private static final org.tensorflow.proto.framework.GraphTransferNodeOutputInfo DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.GraphTransferNodeOutputInfo();
+  }
+
+  public static org.tensorflow.proto.framework.GraphTransferNodeOutputInfo getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<GraphTransferNodeOutputInfo>
+      PARSER = new com.google.protobuf.AbstractParser<GraphTransferNodeOutputInfo>() {
+    @java.lang.Override
+    public GraphTransferNodeOutputInfo parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new GraphTransferNodeOutputInfo(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<GraphTransferNodeOutputInfo> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<GraphTransferNodeOutputInfo> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.framework.GraphTransferNodeOutputInfo getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphTransferNodeOutputInfoOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphTransferNodeOutputInfoOrBuilder.java
new file mode 100644
index 00000000000..7721bc2dc6a
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/GraphTransferNodeOutputInfoOrBuilder.java
@@ -0,0 +1,27 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/framework/graph_transfer_info.proto
+
+package org.tensorflow.proto.framework;
+
+public interface GraphTransferNodeOutputInfoOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.GraphTransferNodeOutputInfo)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <code>int32 node_id = 1;</code>
+   */
+  int getNodeId();
+
+  /**
+   * <code>repeated int32 max_byte_size = 2;</code>
+   */
+  java.util.List<java.lang.Integer> getMaxByteSizeList();
+  /**
+   * <code>repeated int32 max_byte_size = 2;</code>
+   */
+  int getMaxByteSizeCount();
+  /**
+   * <code>repeated int32 max_byte_size = 2;</code>
+   */
+  int getMaxByteSize(int index);
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/HistogramProto.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/HistogramProto.java
new file mode 100644
index 00000000000..e5e19354670
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/HistogramProto.java
@@ -0,0 +1,1120 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/framework/summary.proto
+
+package org.tensorflow.proto.framework;
+
+/**
+ * <pre>
+ * Serialization format for histogram module in
+ * core/lib/histogram/histogram.h
+ * </pre>
+ *
+ * Protobuf type {@code tensorflow.HistogramProto}
+ */
+public  final class HistogramProto extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.HistogramProto)
+    HistogramProtoOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use HistogramProto.newBuilder() to construct.
+  private HistogramProto(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private HistogramProto() {
+    bucketLimit_ = emptyDoubleList();
+    bucket_ = emptyDoubleList();
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new HistogramProto();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private HistogramProto(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    int mutable_bitField0_ = 0;
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 9: {
+
+            min_ = input.readDouble();
+            break;
+          }
+          case 17: {
+
+            max_ = input.readDouble();
+            break;
+          }
+          case 25: {
+
+            num_ = input.readDouble();
+            break;
+          }
+          case 33: {
+
+            sum_ = input.readDouble();
+            break;
+          }
+          case 41: {
+
+            sumSquares_ = input.readDouble();
+            break;
+          }
+          case 49: {
+            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
+              bucketLimit_ = newDoubleList();
+              mutable_bitField0_ |= 0x00000001;
+            }
+            bucketLimit_.addDouble(input.readDouble());
+            break;
+          }
+          case 50: {
+            int length = input.readRawVarint32();
+            int limit = input.pushLimit(length);
+            if (!((mutable_bitField0_ & 0x00000001) != 0) && input.getBytesUntilLimit() > 0) {
+              bucketLimit_ = newDoubleList();
+              mutable_bitField0_ |= 0x00000001;
+            }
+            while (input.getBytesUntilLimit() > 0) {
+              bucketLimit_.addDouble(input.readDouble());
+            }
+            input.popLimit(limit);
+            break;
+          }
+          case 57: {
+            if (!((mutable_bitField0_ & 0x00000002) != 0)) {
+              bucket_ = newDoubleList();
+              mutable_bitField0_ |= 0x00000002;
+            }
+            bucket_.addDouble(input.readDouble());
+            break;
+          }
+          case 58: {
+            int length = input.readRawVarint32();
+            int limit = input.pushLimit(length);
+            if (!((mutable_bitField0_ & 0x00000002) != 0) && input.getBytesUntilLimit() > 0) {
+              bucket_ = newDoubleList();
+              mutable_bitField0_ |= 0x00000002;
+            }
+            while (input.getBytesUntilLimit() > 0) {
+              bucket_.addDouble(input.readDouble());
+            }
+            input.popLimit(limit);
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      if (((mutable_bitField0_ & 0x00000001) != 0)) {
+        bucketLimit_.makeImmutable(); // C
+      }
+      if (((mutable_bitField0_ & 0x00000002) != 0)) {
+        bucket_.makeImmutable(); // C
+      }
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.framework.SummaryProtos.internal_static_tensorflow_HistogramProto_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.framework.SummaryProtos.internal_static_tensorflow_HistogramProto_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.framework.HistogramProto.class, org.tensorflow.proto.framework.HistogramProto.Builder.class);
+  }
+
+  public static final int MIN_FIELD_NUMBER = 1;
+  private double min_;
+  /**
+   * <code>double min = 1;</code>
+   */
+  public double getMin() {
+    return min_;
+  }
+
+  public static final int MAX_FIELD_NUMBER = 2;
+  private double max_;
+  /**
+   * <code>double max = 2;</code>
+   */
+  public double getMax() {
+    return max_;
+  }
+
+  public static final int NUM_FIELD_NUMBER = 3;
+  private double num_;
+  /**
+   * <code>double num = 3;</code>
+   */
+  public double getNum() {
+    return num_;
+  }
+
+  public static final int SUM_FIELD_NUMBER = 4;
+  private double sum_;
+  /**
+   * <code>double sum = 4;</code>
+   */
+  public double getSum() {
+    return sum_;
+  }
+
+  public static final int SUM_SQUARES_FIELD_NUMBER = 5;
+  private double sumSquares_;
+  /**
+   * <code>double sum_squares = 5;</code>
+   */
+  public double getSumSquares() {
+    return sumSquares_;
+  }
+
+  public static final int BUCKET_LIMIT_FIELD_NUMBER = 6;
+  private com.google.protobuf.Internal.DoubleList bucketLimit_;
+  /**
+   * <pre>
+   * Parallel arrays encoding the bucket boundaries and the bucket values.
+   * bucket(i) is the count for the bucket i.  The range for
+   * a bucket is:
+   *   i == 0:  -DBL_MAX .. bucket_limit(0)
+   *   i != 0:  bucket_limit(i-1) .. bucket_limit(i)
+   * </pre>
+   *
+   * <code>repeated double bucket_limit = 6 [packed = true];</code>
+   */
+  public java.util.List<java.lang.Double>
+      getBucketLimitList() {
+    return bucketLimit_;
+  }
+  /**
+   * <pre>
+   * Parallel arrays encoding the bucket boundaries and the bucket values.
+   * bucket(i) is the count for the bucket i.  The range for
+   * a bucket is:
+   *   i == 0:  -DBL_MAX .. bucket_limit(0)
+   *   i != 0:  bucket_limit(i-1) .. bucket_limit(i)
+   * </pre>
+   *
+   * <code>repeated double bucket_limit = 6 [packed = true];</code>
+   */
+  public int getBucketLimitCount() {
+    return bucketLimit_.size();
+  }
+  /**
+   * <pre>
+   * Parallel arrays encoding the bucket boundaries and the bucket values.
+   * bucket(i) is the count for the bucket i.  The range for
+   * a bucket is:
+   *   i == 0:  -DBL_MAX .. bucket_limit(0)
+   *   i != 0:  bucket_limit(i-1) .. bucket_limit(i)
+   * </pre>
+   *
+   * <code>repeated double bucket_limit = 6 [packed = true];</code>
+   */
+  public double getBucketLimit(int index) {
+    return bucketLimit_.getDouble(index);
+  }
+  private int bucketLimitMemoizedSerializedSize = -1;
+
+  public static final int BUCKET_FIELD_NUMBER = 7;
+  private com.google.protobuf.Internal.DoubleList bucket_;
+  /**
+   * <code>repeated double bucket = 7 [packed = true];</code>
+   */
+  public java.util.List<java.lang.Double>
+      getBucketList() {
+    return bucket_;
+  }
+  /**
+   * <code>repeated double bucket = 7 [packed = true];</code>
+   */
+  public int getBucketCount() {
+    return bucket_.size();
+  }
+  /**
+   * <code>repeated double bucket = 7 [packed = true];</code>
+   */
+  public double getBucket(int index) {
+    return bucket_.getDouble(index);
+  }
+  private int bucketMemoizedSerializedSize = -1;
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    getSerializedSize();
+    if (min_ != 0D) {
+      output.writeDouble(1, min_);
+    }
+    if (max_ != 0D) {
+      output.writeDouble(2, max_);
+    }
+    if (num_ != 0D) {
+      output.writeDouble(3, num_);
+    }
+    if (sum_ != 0D) {
+      output.writeDouble(4, sum_);
+    }
+    if (sumSquares_ != 0D) {
+      output.writeDouble(5, sumSquares_);
+    }
+    if (getBucketLimitList().size() > 0) {
+      output.writeUInt32NoTag(50);
+      output.writeUInt32NoTag(bucketLimitMemoizedSerializedSize);
+    }
+    for (int i = 0; i < bucketLimit_.size(); i++) {
+      output.writeDoubleNoTag(bucketLimit_.getDouble(i));
+    }
+    if (getBucketList().size() > 0) {
+      output.writeUInt32NoTag(58);
+      output.writeUInt32NoTag(bucketMemoizedSerializedSize);
+    }
+    for (int i = 0; i < bucket_.size(); i++) {
+      output.writeDoubleNoTag(bucket_.getDouble(i));
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (min_ != 0D) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeDoubleSize(1, min_);
+    }
+    if (max_ != 0D) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeDoubleSize(2, max_);
+    }
+    if (num_ != 0D) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeDoubleSize(3, num_);
+    }
+    if (sum_ != 0D) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeDoubleSize(4, sum_);
+    }
+    if (sumSquares_ != 0D) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeDoubleSize(5, sumSquares_);
+    }
+    {
+      int dataSize = 0;
+      dataSize = 8 * getBucketLimitList().size();
+      size += dataSize;
+      if (!getBucketLimitList().isEmpty()) {
+        size += 1;
+        size += com.google.protobuf.CodedOutputStream
+            .computeInt32SizeNoTag(dataSize);
+      }
+      bucketLimitMemoizedSerializedSize = dataSize;
+    }
+    {
+      int dataSize = 0;
+      dataSize = 8 * getBucketList().size();
+      size += dataSize;
+      if (!getBucketList().isEmpty()) {
+        size += 1;
+        size += com.google.protobuf.CodedOutputStream
+            .computeInt32SizeNoTag(dataSize);
+      }
+      bucketMemoizedSerializedSize = dataSize;
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.framework.HistogramProto)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.framework.HistogramProto other = (org.tensorflow.proto.framework.HistogramProto) obj;
+
+    if (java.lang.Double.doubleToLongBits(getMin())
+        != java.lang.Double.doubleToLongBits(
+            other.getMin())) return false;
+    if (java.lang.Double.doubleToLongBits(getMax())
+        != java.lang.Double.doubleToLongBits(
+            other.getMax())) return false;
+    if (java.lang.Double.doubleToLongBits(getNum())
+        != java.lang.Double.doubleToLongBits(
+            other.getNum())) return false;
+    if (java.lang.Double.doubleToLongBits(getSum())
+        != java.lang.Double.doubleToLongBits(
+            other.getSum())) return false;
+    if (java.lang.Double.doubleToLongBits(getSumSquares())
+        != java.lang.Double.doubleToLongBits(
+            other.getSumSquares())) return false;
+    if (!getBucketLimitList()
+        .equals(other.getBucketLimitList())) return false;
+    if (!getBucketList()
+        .equals(other.getBucketList())) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    hash = (37 * hash) + MIN_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+        java.lang.Double.doubleToLongBits(getMin()));
+    hash = (37 * hash) + MAX_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+        java.lang.Double.doubleToLongBits(getMax()));
+    hash = (37 * hash) + NUM_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+        java.lang.Double.doubleToLongBits(getNum()));
+    hash = (37 * hash) + SUM_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+        java.lang.Double.doubleToLongBits(getSum()));
+    hash = (37 * hash) + SUM_SQUARES_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+        java.lang.Double.doubleToLongBits(getSumSquares()));
+    if (getBucketLimitCount() > 0) {
+      hash = (37 * hash) + BUCKET_LIMIT_FIELD_NUMBER;
+      hash = (53 * hash) + getBucketLimitList().hashCode();
+    }
+    if (getBucketCount() > 0) {
+      hash = (37 * hash) + BUCKET_FIELD_NUMBER;
+      hash = (53 * hash) + getBucketList().hashCode();
+    }
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.framework.HistogramProto parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.HistogramProto parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.HistogramProto parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.HistogramProto parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.HistogramProto parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.HistogramProto parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.HistogramProto parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.HistogramProto parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.HistogramProto parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.HistogramProto parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.HistogramProto parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.HistogramProto parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.framework.HistogramProto prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * <pre>
+   * Serialization format for histogram module in
+   * core/lib/histogram/histogram.h
+   * </pre>
+   *
+   * Protobuf type {@code tensorflow.HistogramProto}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.HistogramProto)
+      org.tensorflow.proto.framework.HistogramProtoOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.SummaryProtos.internal_static_tensorflow_HistogramProto_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.SummaryProtos.internal_static_tensorflow_HistogramProto_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.HistogramProto.class, org.tensorflow.proto.framework.HistogramProto.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.framework.HistogramProto.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      min_ = 0D;
+
+      max_ = 0D;
+
+      num_ = 0D;
+
+      sum_ = 0D;
+
+      sumSquares_ = 0D;
+
+      bucketLimit_ = emptyDoubleList();
+      bitField0_ = (bitField0_ & ~0x00000001);
+      bucket_ = emptyDoubleList();
+      bitField0_ = (bitField0_ & ~0x00000002);
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.framework.SummaryProtos.internal_static_tensorflow_HistogramProto_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.HistogramProto getDefaultInstanceForType() {
+      return org.tensorflow.proto.framework.HistogramProto.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.HistogramProto build() {
+      org.tensorflow.proto.framework.HistogramProto result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.HistogramProto buildPartial() {
+      org.tensorflow.proto.framework.HistogramProto result = new org.tensorflow.proto.framework.HistogramProto(this);
+      int from_bitField0_ = bitField0_;
+      result.min_ = min_;
+      result.max_ = max_;
+      result.num_ = num_;
+      result.sum_ = sum_;
+      result.sumSquares_ = sumSquares_;
+      if (((bitField0_ & 0x00000001) != 0)) {
+        bucketLimit_.makeImmutable();
+        bitField0_ = (bitField0_ & ~0x00000001);
+      }
+      result.bucketLimit_ = bucketLimit_;
+      if (((bitField0_ & 0x00000002) != 0)) {
+        bucket_.makeImmutable();
+        bitField0_ = (bitField0_ & ~0x00000002);
+      }
+      result.bucket_ = bucket_;
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.framework.HistogramProto) {
+        return mergeFrom((org.tensorflow.proto.framework.HistogramProto)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.framework.HistogramProto other) {
+      if (other == org.tensorflow.proto.framework.HistogramProto.getDefaultInstance()) return this;
+      if (other.getMin() != 0D) {
+        setMin(other.getMin());
+      }
+      if (other.getMax() != 0D) {
+        setMax(other.getMax());
+      }
+      if (other.getNum() != 0D) {
+        setNum(other.getNum());
+      }
+      if (other.getSum() != 0D) {
+        setSum(other.getSum());
+      }
+      if (other.getSumSquares() != 0D) {
+        setSumSquares(other.getSumSquares());
+      }
+      if (!other.bucketLimit_.isEmpty()) {
+        if (bucketLimit_.isEmpty()) {
+          bucketLimit_ = other.bucketLimit_;
+          bitField0_ = (bitField0_ & ~0x00000001);
+        } else {
+          ensureBucketLimitIsMutable();
+          bucketLimit_.addAll(other.bucketLimit_);
+        }
+        onChanged();
+      }
+      if (!other.bucket_.isEmpty()) {
+        if (bucket_.isEmpty()) {
+          bucket_ = other.bucket_;
+          bitField0_ = (bitField0_ & ~0x00000002);
+        } else {
+          ensureBucketIsMutable();
+          bucket_.addAll(other.bucket_);
+        }
+        onChanged();
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.framework.HistogramProto parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.framework.HistogramProto) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+    private int bitField0_;
+
+    private double min_ ;
+    /**
+     * <code>double min = 1;</code>
+     */
+    public double getMin() {
+      return min_;
+    }
+    /**
+     * <code>double min = 1;</code>
+     */
+    public Builder setMin(double value) {
+      
+      min_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>double min = 1;</code>
+     */
+    public Builder clearMin() {
+      
+      min_ = 0D;
+      onChanged();
+      return this;
+    }
+
+    private double max_ ;
+    /**
+     * <code>double max = 2;</code>
+     */
+    public double getMax() {
+      return max_;
+    }
+    /**
+     * <code>double max = 2;</code>
+     */
+    public Builder setMax(double value) {
+      
+      max_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>double max = 2;</code>
+     */
+    public Builder clearMax() {
+      
+      max_ = 0D;
+      onChanged();
+      return this;
+    }
+
+    private double num_ ;
+    /**
+     * <code>double num = 3;</code>
+     */
+    public double getNum() {
+      return num_;
+    }
+    /**
+     * <code>double num = 3;</code>
+     */
+    public Builder setNum(double value) {
+      
+      num_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>double num = 3;</code>
+     */
+    public Builder clearNum() {
+      
+      num_ = 0D;
+      onChanged();
+      return this;
+    }
+
+    private double sum_ ;
+    /**
+     * <code>double sum = 4;</code>
+     */
+    public double getSum() {
+      return sum_;
+    }
+    /**
+     * <code>double sum = 4;</code>
+     */
+    public Builder setSum(double value) {
+      
+      sum_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>double sum = 4;</code>
+     */
+    public Builder clearSum() {
+      
+      sum_ = 0D;
+      onChanged();
+      return this;
+    }
+
+    private double sumSquares_ ;
+    /**
+     * <code>double sum_squares = 5;</code>
+     */
+    public double getSumSquares() {
+      return sumSquares_;
+    }
+    /**
+     * <code>double sum_squares = 5;</code>
+     */
+    public Builder setSumSquares(double value) {
+      
+      sumSquares_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>double sum_squares = 5;</code>
+     */
+    public Builder clearSumSquares() {
+      
+      sumSquares_ = 0D;
+      onChanged();
+      return this;
+    }
+
+    private com.google.protobuf.Internal.DoubleList bucketLimit_ = emptyDoubleList();
+    private void ensureBucketLimitIsMutable() {
+      if (!((bitField0_ & 0x00000001) != 0)) {
+        bucketLimit_ = mutableCopy(bucketLimit_);
+        bitField0_ |= 0x00000001;
+       }
+    }
+    /**
+     * <pre>
+     * Parallel arrays encoding the bucket boundaries and the bucket values.
+     * bucket(i) is the count for the bucket i.  The range for
+     * a bucket is:
+     *   i == 0:  -DBL_MAX .. bucket_limit(0)
+     *   i != 0:  bucket_limit(i-1) .. bucket_limit(i)
+     * </pre>
+     *
+     * <code>repeated double bucket_limit = 6 [packed = true];</code>
+     */
+    public java.util.List<java.lang.Double>
+        getBucketLimitList() {
+      return ((bitField0_ & 0x00000001) != 0) ?
+               java.util.Collections.unmodifiableList(bucketLimit_) : bucketLimit_;
+    }
+    /**
+     * <pre>
+     * Parallel arrays encoding the bucket boundaries and the bucket values.
+     * bucket(i) is the count for the bucket i.  The range for
+     * a bucket is:
+     *   i == 0:  -DBL_MAX .. bucket_limit(0)
+     *   i != 0:  bucket_limit(i-1) .. bucket_limit(i)
+     * </pre>
+     *
+     * <code>repeated double bucket_limit = 6 [packed = true];</code>
+     */
+    public int getBucketLimitCount() {
+      return bucketLimit_.size();
+    }
+    /**
+     * <pre>
+     * Parallel arrays encoding the bucket boundaries and the bucket values.
+     * bucket(i) is the count for the bucket i.  The range for
+     * a bucket is:
+     *   i == 0:  -DBL_MAX .. bucket_limit(0)
+     *   i != 0:  bucket_limit(i-1) .. bucket_limit(i)
+     * </pre>
+     *
+     * <code>repeated double bucket_limit = 6 [packed = true];</code>
+     */
+    public double getBucketLimit(int index) {
+      return bucketLimit_.getDouble(index);
+    }
+    /**
+     * <pre>
+     * Parallel arrays encoding the bucket boundaries and the bucket values.
+     * bucket(i) is the count for the bucket i.  The range for
+     * a bucket is:
+     *   i == 0:  -DBL_MAX .. bucket_limit(0)
+     *   i != 0:  bucket_limit(i-1) .. bucket_limit(i)
+     * </pre>
+     *
+     * <code>repeated double bucket_limit = 6 [packed = true];</code>
+     */
+    public Builder setBucketLimit(
+        int index, double value) {
+      ensureBucketLimitIsMutable();
+      bucketLimit_.setDouble(index, value);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Parallel arrays encoding the bucket boundaries and the bucket values.
+     * bucket(i) is the count for the bucket i.  The range for
+     * a bucket is:
+     *   i == 0:  -DBL_MAX .. bucket_limit(0)
+     *   i != 0:  bucket_limit(i-1) .. bucket_limit(i)
+     * </pre>
+     *
+     * <code>repeated double bucket_limit = 6 [packed = true];</code>
+     */
+    public Builder addBucketLimit(double value) {
+      ensureBucketLimitIsMutable();
+      bucketLimit_.addDouble(value);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Parallel arrays encoding the bucket boundaries and the bucket values.
+     * bucket(i) is the count for the bucket i.  The range for
+     * a bucket is:
+     *   i == 0:  -DBL_MAX .. bucket_limit(0)
+     *   i != 0:  bucket_limit(i-1) .. bucket_limit(i)
+     * </pre>
+     *
+     * <code>repeated double bucket_limit = 6 [packed = true];</code>
+     */
+    public Builder addAllBucketLimit(
+        java.lang.Iterable<? extends java.lang.Double> values) {
+      ensureBucketLimitIsMutable();
+      com.google.protobuf.AbstractMessageLite.Builder.addAll(
+          values, bucketLimit_);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Parallel arrays encoding the bucket boundaries and the bucket values.
+     * bucket(i) is the count for the bucket i.  The range for
+     * a bucket is:
+     *   i == 0:  -DBL_MAX .. bucket_limit(0)
+     *   i != 0:  bucket_limit(i-1) .. bucket_limit(i)
+     * </pre>
+     *
+     * <code>repeated double bucket_limit = 6 [packed = true];</code>
+     */
+    public Builder clearBucketLimit() {
+      bucketLimit_ = emptyDoubleList();
+      bitField0_ = (bitField0_ & ~0x00000001);
+      onChanged();
+      return this;
+    }
+
+    private com.google.protobuf.Internal.DoubleList bucket_ = emptyDoubleList();
+    private void ensureBucketIsMutable() {
+      if (!((bitField0_ & 0x00000002) != 0)) {
+        bucket_ = mutableCopy(bucket_);
+        bitField0_ |= 0x00000002;
+       }
+    }
+    /**
+     * <code>repeated double bucket = 7 [packed = true];</code>
+     */
+    public java.util.List<java.lang.Double>
+        getBucketList() {
+      return ((bitField0_ & 0x00000002) != 0) ?
+               java.util.Collections.unmodifiableList(bucket_) : bucket_;
+    }
+    /**
+     * <code>repeated double bucket = 7 [packed = true];</code>
+     */
+    public int getBucketCount() {
+      return bucket_.size();
+    }
+    /**
+     * <code>repeated double bucket = 7 [packed = true];</code>
+     */
+    public double getBucket(int index) {
+      return bucket_.getDouble(index);
+    }
+    /**
+     * <code>repeated double bucket = 7 [packed = true];</code>
+     */
+    public Builder setBucket(
+        int index, double value) {
+      ensureBucketIsMutable();
+      bucket_.setDouble(index, value);
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>repeated double bucket = 7 [packed = true];</code>
+     */
+    public Builder addBucket(double value) {
+      ensureBucketIsMutable();
+      bucket_.addDouble(value);
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>repeated double bucket = 7 [packed = true];</code>
+     */
+    public Builder addAllBucket(
+        java.lang.Iterable<? extends java.lang.Double> values) {
+      ensureBucketIsMutable();
+      com.google.protobuf.AbstractMessageLite.Builder.addAll(
+          values, bucket_);
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>repeated double bucket = 7 [packed = true];</code>
+     */
+    public Builder clearBucket() {
+      bucket_ = emptyDoubleList();
+      bitField0_ = (bitField0_ & ~0x00000002);
+      onChanged();
+      return this;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.HistogramProto)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.HistogramProto)
+  private static final org.tensorflow.proto.framework.HistogramProto DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.HistogramProto();
+  }
+
+  public static org.tensorflow.proto.framework.HistogramProto getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<HistogramProto>
+      PARSER = new com.google.protobuf.AbstractParser<HistogramProto>() {
+    @java.lang.Override
+    public HistogramProto parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new HistogramProto(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<HistogramProto> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<HistogramProto> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.framework.HistogramProto getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/HistogramProtoOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/HistogramProtoOrBuilder.java
new file mode 100644
index 00000000000..d76afe24d19
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/HistogramProtoOrBuilder.java
@@ -0,0 +1,84 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/framework/summary.proto
+
+package org.tensorflow.proto.framework;
+
+public interface HistogramProtoOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.HistogramProto)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <code>double min = 1;</code>
+   */
+  double getMin();
+
+  /**
+   * <code>double max = 2;</code>
+   */
+  double getMax();
+
+  /**
+   * <code>double num = 3;</code>
+   */
+  double getNum();
+
+  /**
+   * <code>double sum = 4;</code>
+   */
+  double getSum();
+
+  /**
+   * <code>double sum_squares = 5;</code>
+   */
+  double getSumSquares();
+
+  /**
+   * <pre>
+   * Parallel arrays encoding the bucket boundaries and the bucket values.
+   * bucket(i) is the count for the bucket i.  The range for
+   * a bucket is:
+   *   i == 0:  -DBL_MAX .. bucket_limit(0)
+   *   i != 0:  bucket_limit(i-1) .. bucket_limit(i)
+   * </pre>
+   *
+   * <code>repeated double bucket_limit = 6 [packed = true];</code>
+   */
+  java.util.List<java.lang.Double> getBucketLimitList();
+  /**
+   * <pre>
+   * Parallel arrays encoding the bucket boundaries and the bucket values.
+   * bucket(i) is the count for the bucket i.  The range for
+   * a bucket is:
+   *   i == 0:  -DBL_MAX .. bucket_limit(0)
+   *   i != 0:  bucket_limit(i-1) .. bucket_limit(i)
+   * </pre>
+   *
+   * <code>repeated double bucket_limit = 6 [packed = true];</code>
+   */
+  int getBucketLimitCount();
+  /**
+   * <pre>
+   * Parallel arrays encoding the bucket boundaries and the bucket values.
+   * bucket(i) is the count for the bucket i.  The range for
+   * a bucket is:
+   *   i == 0:  -DBL_MAX .. bucket_limit(0)
+   *   i != 0:  bucket_limit(i-1) .. bucket_limit(i)
+   * </pre>
+   *
+   * <code>repeated double bucket_limit = 6 [packed = true];</code>
+   */
+  double getBucketLimit(int index);
+
+  /**
+   * <code>repeated double bucket = 7 [packed = true];</code>
+   */
+  java.util.List<java.lang.Double> getBucketList();
+  /**
+   * <code>repeated double bucket = 7 [packed = true];</code>
+   */
+  int getBucketCount();
+  /**
+   * <code>repeated double bucket = 7 [packed = true];</code>
+   */
+  double getBucket(int index);
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/InterconnectLink.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/InterconnectLink.java
new file mode 100644
index 00000000000..89fbd7a8dec
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/InterconnectLink.java
@@ -0,0 +1,660 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/framework/device_attributes.proto
+
+package org.tensorflow.proto.framework;
+
+/**
+ * Protobuf type {@code tensorflow.InterconnectLink}
+ */
+public  final class InterconnectLink extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.InterconnectLink)
+    InterconnectLinkOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use InterconnectLink.newBuilder() to construct.
+  private InterconnectLink(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private InterconnectLink() {
+    type_ = "";
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new InterconnectLink();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private InterconnectLink(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 8: {
+
+            deviceId_ = input.readInt32();
+            break;
+          }
+          case 18: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            type_ = s;
+            break;
+          }
+          case 24: {
+
+            strength_ = input.readInt32();
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.framework.DeviceAttributesProtos.internal_static_tensorflow_InterconnectLink_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.framework.DeviceAttributesProtos.internal_static_tensorflow_InterconnectLink_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.framework.InterconnectLink.class, org.tensorflow.proto.framework.InterconnectLink.Builder.class);
+  }
+
+  public static final int DEVICE_ID_FIELD_NUMBER = 1;
+  private int deviceId_;
+  /**
+   * <code>int32 device_id = 1;</code>
+   */
+  public int getDeviceId() {
+    return deviceId_;
+  }
+
+  public static final int TYPE_FIELD_NUMBER = 2;
+  private volatile java.lang.Object type_;
+  /**
+   * <code>string type = 2;</code>
+   */
+  public java.lang.String getType() {
+    java.lang.Object ref = type_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      type_ = s;
+      return s;
+    }
+  }
+  /**
+   * <code>string type = 2;</code>
+   */
+  public com.google.protobuf.ByteString
+      getTypeBytes() {
+    java.lang.Object ref = type_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      type_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int STRENGTH_FIELD_NUMBER = 3;
+  private int strength_;
+  /**
+   * <code>int32 strength = 3;</code>
+   */
+  public int getStrength() {
+    return strength_;
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    if (deviceId_ != 0) {
+      output.writeInt32(1, deviceId_);
+    }
+    if (!getTypeBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 2, type_);
+    }
+    if (strength_ != 0) {
+      output.writeInt32(3, strength_);
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (deviceId_ != 0) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt32Size(1, deviceId_);
+    }
+    if (!getTypeBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(2, type_);
+    }
+    if (strength_ != 0) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt32Size(3, strength_);
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.framework.InterconnectLink)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.framework.InterconnectLink other = (org.tensorflow.proto.framework.InterconnectLink) obj;
+
+    if (getDeviceId()
+        != other.getDeviceId()) return false;
+    if (!getType()
+        .equals(other.getType())) return false;
+    if (getStrength()
+        != other.getStrength()) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    hash = (37 * hash) + DEVICE_ID_FIELD_NUMBER;
+    hash = (53 * hash) + getDeviceId();
+    hash = (37 * hash) + TYPE_FIELD_NUMBER;
+    hash = (53 * hash) + getType().hashCode();
+    hash = (37 * hash) + STRENGTH_FIELD_NUMBER;
+    hash = (53 * hash) + getStrength();
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.framework.InterconnectLink parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.InterconnectLink parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.InterconnectLink parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.InterconnectLink parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.InterconnectLink parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.InterconnectLink parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.InterconnectLink parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.InterconnectLink parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.InterconnectLink parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.InterconnectLink parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.InterconnectLink parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.InterconnectLink parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.framework.InterconnectLink prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * Protobuf type {@code tensorflow.InterconnectLink}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.InterconnectLink)
+      org.tensorflow.proto.framework.InterconnectLinkOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.DeviceAttributesProtos.internal_static_tensorflow_InterconnectLink_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.DeviceAttributesProtos.internal_static_tensorflow_InterconnectLink_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.InterconnectLink.class, org.tensorflow.proto.framework.InterconnectLink.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.framework.InterconnectLink.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      deviceId_ = 0;
+
+      type_ = "";
+
+      strength_ = 0;
+
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.framework.DeviceAttributesProtos.internal_static_tensorflow_InterconnectLink_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.InterconnectLink getDefaultInstanceForType() {
+      return org.tensorflow.proto.framework.InterconnectLink.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.InterconnectLink build() {
+      org.tensorflow.proto.framework.InterconnectLink result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.InterconnectLink buildPartial() {
+      org.tensorflow.proto.framework.InterconnectLink result = new org.tensorflow.proto.framework.InterconnectLink(this);
+      result.deviceId_ = deviceId_;
+      result.type_ = type_;
+      result.strength_ = strength_;
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.framework.InterconnectLink) {
+        return mergeFrom((org.tensorflow.proto.framework.InterconnectLink)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.framework.InterconnectLink other) {
+      if (other == org.tensorflow.proto.framework.InterconnectLink.getDefaultInstance()) return this;
+      if (other.getDeviceId() != 0) {
+        setDeviceId(other.getDeviceId());
+      }
+      if (!other.getType().isEmpty()) {
+        type_ = other.type_;
+        onChanged();
+      }
+      if (other.getStrength() != 0) {
+        setStrength(other.getStrength());
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.framework.InterconnectLink parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.framework.InterconnectLink) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+
+    private int deviceId_ ;
+    /**
+     * <code>int32 device_id = 1;</code>
+     */
+    public int getDeviceId() {
+      return deviceId_;
+    }
+    /**
+     * <code>int32 device_id = 1;</code>
+     */
+    public Builder setDeviceId(int value) {
+      
+      deviceId_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>int32 device_id = 1;</code>
+     */
+    public Builder clearDeviceId() {
+      
+      deviceId_ = 0;
+      onChanged();
+      return this;
+    }
+
+    private java.lang.Object type_ = "";
+    /**
+     * <code>string type = 2;</code>
+     */
+    public java.lang.String getType() {
+      java.lang.Object ref = type_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        type_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <code>string type = 2;</code>
+     */
+    public com.google.protobuf.ByteString
+        getTypeBytes() {
+      java.lang.Object ref = type_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        type_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <code>string type = 2;</code>
+     */
+    public Builder setType(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      type_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>string type = 2;</code>
+     */
+    public Builder clearType() {
+      
+      type_ = getDefaultInstance().getType();
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>string type = 2;</code>
+     */
+    public Builder setTypeBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      type_ = value;
+      onChanged();
+      return this;
+    }
+
+    private int strength_ ;
+    /**
+     * <code>int32 strength = 3;</code>
+     */
+    public int getStrength() {
+      return strength_;
+    }
+    /**
+     * <code>int32 strength = 3;</code>
+     */
+    public Builder setStrength(int value) {
+      
+      strength_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>int32 strength = 3;</code>
+     */
+    public Builder clearStrength() {
+      
+      strength_ = 0;
+      onChanged();
+      return this;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.InterconnectLink)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.InterconnectLink)
+  private static final org.tensorflow.proto.framework.InterconnectLink DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.InterconnectLink();
+  }
+
+  public static org.tensorflow.proto.framework.InterconnectLink getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<InterconnectLink>
+      PARSER = new com.google.protobuf.AbstractParser<InterconnectLink>() {
+    @java.lang.Override
+    public InterconnectLink parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new InterconnectLink(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<InterconnectLink> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<InterconnectLink> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.framework.InterconnectLink getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/InterconnectLinkOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/InterconnectLinkOrBuilder.java
new file mode 100644
index 00000000000..61316cfa8cc
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/InterconnectLinkOrBuilder.java
@@ -0,0 +1,29 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/framework/device_attributes.proto
+
+package org.tensorflow.proto.framework;
+
+public interface InterconnectLinkOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.InterconnectLink)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <code>int32 device_id = 1;</code>
+   */
+  int getDeviceId();
+
+  /**
+   * <code>string type = 2;</code>
+   */
+  java.lang.String getType();
+  /**
+   * <code>string type = 2;</code>
+   */
+  com.google.protobuf.ByteString
+      getTypeBytes();
+
+  /**
+   * <code>int32 strength = 3;</code>
+   */
+  int getStrength();
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/KernelDef.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/KernelDef.java
new file mode 100644
index 00000000000..76637537c9e
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/KernelDef.java
@@ -0,0 +1,2420 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/framework/kernel_def.proto
+
+package org.tensorflow.proto.framework;
+
+/**
+ * Protobuf type {@code tensorflow.KernelDef}
+ */
+public  final class KernelDef extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.KernelDef)
+    KernelDefOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use KernelDef.newBuilder() to construct.
+  private KernelDef(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private KernelDef() {
+    op_ = "";
+    deviceType_ = "";
+    constraint_ = java.util.Collections.emptyList();
+    hostMemoryArg_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+    label_ = "";
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new KernelDef();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private KernelDef(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    int mutable_bitField0_ = 0;
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 10: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            op_ = s;
+            break;
+          }
+          case 18: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            deviceType_ = s;
+            break;
+          }
+          case 26: {
+            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
+              constraint_ = new java.util.ArrayList<org.tensorflow.proto.framework.KernelDef.AttrConstraint>();
+              mutable_bitField0_ |= 0x00000001;
+            }
+            constraint_.add(
+                input.readMessage(org.tensorflow.proto.framework.KernelDef.AttrConstraint.parser(), extensionRegistry));
+            break;
+          }
+          case 34: {
+            java.lang.String s = input.readStringRequireUtf8();
+            if (!((mutable_bitField0_ & 0x00000002) != 0)) {
+              hostMemoryArg_ = new com.google.protobuf.LazyStringArrayList();
+              mutable_bitField0_ |= 0x00000002;
+            }
+            hostMemoryArg_.add(s);
+            break;
+          }
+          case 42: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            label_ = s;
+            break;
+          }
+          case 48: {
+
+            priority_ = input.readInt32();
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      if (((mutable_bitField0_ & 0x00000001) != 0)) {
+        constraint_ = java.util.Collections.unmodifiableList(constraint_);
+      }
+      if (((mutable_bitField0_ & 0x00000002) != 0)) {
+        hostMemoryArg_ = hostMemoryArg_.getUnmodifiableView();
+      }
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.framework.KernelDefProtos.internal_static_tensorflow_KernelDef_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.framework.KernelDefProtos.internal_static_tensorflow_KernelDef_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.framework.KernelDef.class, org.tensorflow.proto.framework.KernelDef.Builder.class);
+  }
+
+  public interface AttrConstraintOrBuilder extends
+      // @@protoc_insertion_point(interface_extends:tensorflow.KernelDef.AttrConstraint)
+      com.google.protobuf.MessageOrBuilder {
+
+    /**
+     * <pre>
+     * Name of an attr from the Op.
+     * </pre>
+     *
+     * <code>string name = 1;</code>
+     */
+    java.lang.String getName();
+    /**
+     * <pre>
+     * Name of an attr from the Op.
+     * </pre>
+     *
+     * <code>string name = 1;</code>
+     */
+    com.google.protobuf.ByteString
+        getNameBytes();
+
+    /**
+     * <pre>
+     * A list of values that this kernel supports for this attr.
+     * Like OpDef.AttrDef.allowed_values, except for kernels instead of Ops.
+     * </pre>
+     *
+     * <code>.tensorflow.AttrValue allowed_values = 2;</code>
+     */
+    boolean hasAllowedValues();
+    /**
+     * <pre>
+     * A list of values that this kernel supports for this attr.
+     * Like OpDef.AttrDef.allowed_values, except for kernels instead of Ops.
+     * </pre>
+     *
+     * <code>.tensorflow.AttrValue allowed_values = 2;</code>
+     */
+    org.tensorflow.proto.framework.AttrValue getAllowedValues();
+    /**
+     * <pre>
+     * A list of values that this kernel supports for this attr.
+     * Like OpDef.AttrDef.allowed_values, except for kernels instead of Ops.
+     * </pre>
+     *
+     * <code>.tensorflow.AttrValue allowed_values = 2;</code>
+     */
+    org.tensorflow.proto.framework.AttrValueOrBuilder getAllowedValuesOrBuilder();
+  }
+  /**
+   * Protobuf type {@code tensorflow.KernelDef.AttrConstraint}
+   */
+  public  static final class AttrConstraint extends
+      com.google.protobuf.GeneratedMessageV3 implements
+      // @@protoc_insertion_point(message_implements:tensorflow.KernelDef.AttrConstraint)
+      AttrConstraintOrBuilder {
+  private static final long serialVersionUID = 0L;
+    // Use AttrConstraint.newBuilder() to construct.
+    private AttrConstraint(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+      super(builder);
+    }
+    private AttrConstraint() {
+      name_ = "";
+    }
+
+    @java.lang.Override
+    @SuppressWarnings({"unused"})
+    protected java.lang.Object newInstance(
+        UnusedPrivateParameter unused) {
+      return new AttrConstraint();
+    }
+
+    @java.lang.Override
+    public final com.google.protobuf.UnknownFieldSet
+    getUnknownFields() {
+      return this.unknownFields;
+    }
+    private AttrConstraint(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      this();
+      if (extensionRegistry == null) {
+        throw new java.lang.NullPointerException();
+      }
+      com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+          com.google.protobuf.UnknownFieldSet.newBuilder();
+      try {
+        boolean done = false;
+        while (!done) {
+          int tag = input.readTag();
+          switch (tag) {
+            case 0:
+              done = true;
+              break;
+            case 10: {
+              java.lang.String s = input.readStringRequireUtf8();
+
+              name_ = s;
+              break;
+            }
+            case 18: {
+              org.tensorflow.proto.framework.AttrValue.Builder subBuilder = null;
+              if (allowedValues_ != null) {
+                subBuilder = allowedValues_.toBuilder();
+              }
+              allowedValues_ = input.readMessage(org.tensorflow.proto.framework.AttrValue.parser(), extensionRegistry);
+              if (subBuilder != null) {
+                subBuilder.mergeFrom(allowedValues_);
+                allowedValues_ = subBuilder.buildPartial();
+              }
+
+              break;
+            }
+            default: {
+              if (!parseUnknownField(
+                  input, unknownFields, extensionRegistry, tag)) {
+                done = true;
+              }
+              break;
+            }
+          }
+        }
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        throw e.setUnfinishedMessage(this);
+      } catch (java.io.IOException e) {
+        throw new com.google.protobuf.InvalidProtocolBufferException(
+            e).setUnfinishedMessage(this);
+      } finally {
+        this.unknownFields = unknownFields.build();
+        makeExtensionsImmutable();
+      }
+    }
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.KernelDefProtos.internal_static_tensorflow_KernelDef_AttrConstraint_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.KernelDefProtos.internal_static_tensorflow_KernelDef_AttrConstraint_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.KernelDef.AttrConstraint.class, org.tensorflow.proto.framework.KernelDef.AttrConstraint.Builder.class);
+    }
+
+    public static final int NAME_FIELD_NUMBER = 1;
+    private volatile java.lang.Object name_;
+    /**
+     * <pre>
+     * Name of an attr from the Op.
+     * </pre>
+     *
+     * <code>string name = 1;</code>
+     */
+    public java.lang.String getName() {
+      java.lang.Object ref = name_;
+      if (ref instanceof java.lang.String) {
+        return (java.lang.String) ref;
+      } else {
+        com.google.protobuf.ByteString bs = 
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        name_ = s;
+        return s;
+      }
+    }
+    /**
+     * <pre>
+     * Name of an attr from the Op.
+     * </pre>
+     *
+     * <code>string name = 1;</code>
+     */
+    public com.google.protobuf.ByteString
+        getNameBytes() {
+      java.lang.Object ref = name_;
+      if (ref instanceof java.lang.String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        name_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+
+    public static final int ALLOWED_VALUES_FIELD_NUMBER = 2;
+    private org.tensorflow.proto.framework.AttrValue allowedValues_;
+    /**
+     * <pre>
+     * A list of values that this kernel supports for this attr.
+     * Like OpDef.AttrDef.allowed_values, except for kernels instead of Ops.
+     * </pre>
+     *
+     * <code>.tensorflow.AttrValue allowed_values = 2;</code>
+     */
+    public boolean hasAllowedValues() {
+      return allowedValues_ != null;
+    }
+    /**
+     * <pre>
+     * A list of values that this kernel supports for this attr.
+     * Like OpDef.AttrDef.allowed_values, except for kernels instead of Ops.
+     * </pre>
+     *
+     * <code>.tensorflow.AttrValue allowed_values = 2;</code>
+     */
+    public org.tensorflow.proto.framework.AttrValue getAllowedValues() {
+      return allowedValues_ == null ? org.tensorflow.proto.framework.AttrValue.getDefaultInstance() : allowedValues_;
+    }
+    /**
+     * <pre>
+     * A list of values that this kernel supports for this attr.
+     * Like OpDef.AttrDef.allowed_values, except for kernels instead of Ops.
+     * </pre>
+     *
+     * <code>.tensorflow.AttrValue allowed_values = 2;</code>
+     */
+    public org.tensorflow.proto.framework.AttrValueOrBuilder getAllowedValuesOrBuilder() {
+      return getAllowedValues();
+    }
+
+    private byte memoizedIsInitialized = -1;
+    @java.lang.Override
+    public final boolean isInitialized() {
+      byte isInitialized = memoizedIsInitialized;
+      if (isInitialized == 1) return true;
+      if (isInitialized == 0) return false;
+
+      memoizedIsInitialized = 1;
+      return true;
+    }
+
+    @java.lang.Override
+    public void writeTo(com.google.protobuf.CodedOutputStream output)
+                        throws java.io.IOException {
+      if (!getNameBytes().isEmpty()) {
+        com.google.protobuf.GeneratedMessageV3.writeString(output, 1, name_);
+      }
+      if (allowedValues_ != null) {
+        output.writeMessage(2, getAllowedValues());
+      }
+      unknownFields.writeTo(output);
+    }
+
+    @java.lang.Override
+    public int getSerializedSize() {
+      int size = memoizedSize;
+      if (size != -1) return size;
+
+      size = 0;
+      if (!getNameBytes().isEmpty()) {
+        size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, name_);
+      }
+      if (allowedValues_ != null) {
+        size += com.google.protobuf.CodedOutputStream
+          .computeMessageSize(2, getAllowedValues());
+      }
+      size += unknownFields.getSerializedSize();
+      memoizedSize = size;
+      return size;
+    }
+
+    @java.lang.Override
+    public boolean equals(final java.lang.Object obj) {
+      if (obj == this) {
+       return true;
+      }
+      if (!(obj instanceof org.tensorflow.proto.framework.KernelDef.AttrConstraint)) {
+        return super.equals(obj);
+      }
+      org.tensorflow.proto.framework.KernelDef.AttrConstraint other = (org.tensorflow.proto.framework.KernelDef.AttrConstraint) obj;
+
+      if (!getName()
+          .equals(other.getName())) return false;
+      if (hasAllowedValues() != other.hasAllowedValues()) return false;
+      if (hasAllowedValues()) {
+        if (!getAllowedValues()
+            .equals(other.getAllowedValues())) return false;
+      }
+      if (!unknownFields.equals(other.unknownFields)) return false;
+      return true;
+    }
+
+    @java.lang.Override
+    public int hashCode() {
+      if (memoizedHashCode != 0) {
+        return memoizedHashCode;
+      }
+      int hash = 41;
+      hash = (19 * hash) + getDescriptor().hashCode();
+      hash = (37 * hash) + NAME_FIELD_NUMBER;
+      hash = (53 * hash) + getName().hashCode();
+      if (hasAllowedValues()) {
+        hash = (37 * hash) + ALLOWED_VALUES_FIELD_NUMBER;
+        hash = (53 * hash) + getAllowedValues().hashCode();
+      }
+      hash = (29 * hash) + unknownFields.hashCode();
+      memoizedHashCode = hash;
+      return hash;
+    }
+
+    public static org.tensorflow.proto.framework.KernelDef.AttrConstraint parseFrom(
+        java.nio.ByteBuffer data)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data);
+    }
+    public static org.tensorflow.proto.framework.KernelDef.AttrConstraint parseFrom(
+        java.nio.ByteBuffer data,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.KernelDef.AttrConstraint parseFrom(
+        com.google.protobuf.ByteString data)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data);
+    }
+    public static org.tensorflow.proto.framework.KernelDef.AttrConstraint parseFrom(
+        com.google.protobuf.ByteString data,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.KernelDef.AttrConstraint parseFrom(byte[] data)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data);
+    }
+    public static org.tensorflow.proto.framework.KernelDef.AttrConstraint parseFrom(
+        byte[] data,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.KernelDef.AttrConstraint parseFrom(java.io.InputStream input)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input);
+    }
+    public static org.tensorflow.proto.framework.KernelDef.AttrConstraint parseFrom(
+        java.io.InputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.KernelDef.AttrConstraint parseDelimitedFrom(java.io.InputStream input)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseDelimitedWithIOException(PARSER, input);
+    }
+    public static org.tensorflow.proto.framework.KernelDef.AttrConstraint parseDelimitedFrom(
+        java.io.InputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.KernelDef.AttrConstraint parseFrom(
+        com.google.protobuf.CodedInputStream input)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input);
+    }
+    public static org.tensorflow.proto.framework.KernelDef.AttrConstraint parseFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input, extensionRegistry);
+    }
+
+    @java.lang.Override
+    public Builder newBuilderForType() { return newBuilder(); }
+    public static Builder newBuilder() {
+      return DEFAULT_INSTANCE.toBuilder();
+    }
+    public static Builder newBuilder(org.tensorflow.proto.framework.KernelDef.AttrConstraint prototype) {
+      return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+    }
+    @java.lang.Override
+    public Builder toBuilder() {
+      return this == DEFAULT_INSTANCE
+          ? new Builder() : new Builder().mergeFrom(this);
+    }
+
+    @java.lang.Override
+    protected Builder newBuilderForType(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      Builder builder = new Builder(parent);
+      return builder;
+    }
+    /**
+     * Protobuf type {@code tensorflow.KernelDef.AttrConstraint}
+     */
+    public static final class Builder extends
+        com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+        // @@protoc_insertion_point(builder_implements:tensorflow.KernelDef.AttrConstraint)
+        org.tensorflow.proto.framework.KernelDef.AttrConstraintOrBuilder {
+      public static final com.google.protobuf.Descriptors.Descriptor
+          getDescriptor() {
+        return org.tensorflow.proto.framework.KernelDefProtos.internal_static_tensorflow_KernelDef_AttrConstraint_descriptor;
+      }
+
+      @java.lang.Override
+      protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+          internalGetFieldAccessorTable() {
+        return org.tensorflow.proto.framework.KernelDefProtos.internal_static_tensorflow_KernelDef_AttrConstraint_fieldAccessorTable
+            .ensureFieldAccessorsInitialized(
+                org.tensorflow.proto.framework.KernelDef.AttrConstraint.class, org.tensorflow.proto.framework.KernelDef.AttrConstraint.Builder.class);
+      }
+
+      // Construct using org.tensorflow.proto.framework.KernelDef.AttrConstraint.newBuilder()
+      private Builder() {
+        maybeForceBuilderInitialization();
+      }
+
+      private Builder(
+          com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+        super(parent);
+        maybeForceBuilderInitialization();
+      }
+      private void maybeForceBuilderInitialization() {
+        if (com.google.protobuf.GeneratedMessageV3
+                .alwaysUseFieldBuilders) {
+        }
+      }
+      @java.lang.Override
+      public Builder clear() {
+        super.clear();
+        name_ = "";
+
+        if (allowedValuesBuilder_ == null) {
+          allowedValues_ = null;
+        } else {
+          allowedValues_ = null;
+          allowedValuesBuilder_ = null;
+        }
+        return this;
+      }
+
+      @java.lang.Override
+      public com.google.protobuf.Descriptors.Descriptor
+          getDescriptorForType() {
+        return org.tensorflow.proto.framework.KernelDefProtos.internal_static_tensorflow_KernelDef_AttrConstraint_descriptor;
+      }
+
+      @java.lang.Override
+      public org.tensorflow.proto.framework.KernelDef.AttrConstraint getDefaultInstanceForType() {
+        return org.tensorflow.proto.framework.KernelDef.AttrConstraint.getDefaultInstance();
+      }
+
+      @java.lang.Override
+      public org.tensorflow.proto.framework.KernelDef.AttrConstraint build() {
+        org.tensorflow.proto.framework.KernelDef.AttrConstraint result = buildPartial();
+        if (!result.isInitialized()) {
+          throw newUninitializedMessageException(result);
+        }
+        return result;
+      }
+
+      @java.lang.Override
+      public org.tensorflow.proto.framework.KernelDef.AttrConstraint buildPartial() {
+        org.tensorflow.proto.framework.KernelDef.AttrConstraint result = new org.tensorflow.proto.framework.KernelDef.AttrConstraint(this);
+        result.name_ = name_;
+        if (allowedValuesBuilder_ == null) {
+          result.allowedValues_ = allowedValues_;
+        } else {
+          result.allowedValues_ = allowedValuesBuilder_.build();
+        }
+        onBuilt();
+        return result;
+      }
+
+      @java.lang.Override
+      public Builder clone() {
+        return super.clone();
+      }
+      @java.lang.Override
+      public Builder setField(
+          com.google.protobuf.Descriptors.FieldDescriptor field,
+          java.lang.Object value) {
+        return super.setField(field, value);
+      }
+      @java.lang.Override
+      public Builder clearField(
+          com.google.protobuf.Descriptors.FieldDescriptor field) {
+        return super.clearField(field);
+      }
+      @java.lang.Override
+      public Builder clearOneof(
+          com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+        return super.clearOneof(oneof);
+      }
+      @java.lang.Override
+      public Builder setRepeatedField(
+          com.google.protobuf.Descriptors.FieldDescriptor field,
+          int index, java.lang.Object value) {
+        return super.setRepeatedField(field, index, value);
+      }
+      @java.lang.Override
+      public Builder addRepeatedField(
+          com.google.protobuf.Descriptors.FieldDescriptor field,
+          java.lang.Object value) {
+        return super.addRepeatedField(field, value);
+      }
+      @java.lang.Override
+      public Builder mergeFrom(com.google.protobuf.Message other) {
+        if (other instanceof org.tensorflow.proto.framework.KernelDef.AttrConstraint) {
+          return mergeFrom((org.tensorflow.proto.framework.KernelDef.AttrConstraint)other);
+        } else {
+          super.mergeFrom(other);
+          return this;
+        }
+      }
+
+      public Builder mergeFrom(org.tensorflow.proto.framework.KernelDef.AttrConstraint other) {
+        if (other == org.tensorflow.proto.framework.KernelDef.AttrConstraint.getDefaultInstance()) return this;
+        if (!other.getName().isEmpty()) {
+          name_ = other.name_;
+          onChanged();
+        }
+        if (other.hasAllowedValues()) {
+          mergeAllowedValues(other.getAllowedValues());
+        }
+        this.mergeUnknownFields(other.unknownFields);
+        onChanged();
+        return this;
+      }
+
+      @java.lang.Override
+      public final boolean isInitialized() {
+        return true;
+      }
+
+      @java.lang.Override
+      public Builder mergeFrom(
+          com.google.protobuf.CodedInputStream input,
+          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+          throws java.io.IOException {
+        org.tensorflow.proto.framework.KernelDef.AttrConstraint parsedMessage = null;
+        try {
+          parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+        } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+          parsedMessage = (org.tensorflow.proto.framework.KernelDef.AttrConstraint) e.getUnfinishedMessage();
+          throw e.unwrapIOException();
+        } finally {
+          if (parsedMessage != null) {
+            mergeFrom(parsedMessage);
+          }
+        }
+        return this;
+      }
+
+      private java.lang.Object name_ = "";
+      /**
+       * <pre>
+       * Name of an attr from the Op.
+       * </pre>
+       *
+       * <code>string name = 1;</code>
+       */
+      public java.lang.String getName() {
+        java.lang.Object ref = name_;
+        if (!(ref instanceof java.lang.String)) {
+          com.google.protobuf.ByteString bs =
+              (com.google.protobuf.ByteString) ref;
+          java.lang.String s = bs.toStringUtf8();
+          name_ = s;
+          return s;
+        } else {
+          return (java.lang.String) ref;
+        }
+      }
+      /**
+       * <pre>
+       * Name of an attr from the Op.
+       * </pre>
+       *
+       * <code>string name = 1;</code>
+       */
+      public com.google.protobuf.ByteString
+          getNameBytes() {
+        java.lang.Object ref = name_;
+        if (ref instanceof String) {
+          com.google.protobuf.ByteString b = 
+              com.google.protobuf.ByteString.copyFromUtf8(
+                  (java.lang.String) ref);
+          name_ = b;
+          return b;
+        } else {
+          return (com.google.protobuf.ByteString) ref;
+        }
+      }
+      /**
+       * <pre>
+       * Name of an attr from the Op.
+       * </pre>
+       *
+       * <code>string name = 1;</code>
+       */
+      public Builder setName(
+          java.lang.String value) {
+        if (value == null) {
+    throw new NullPointerException();
+  }
+  
+        name_ = value;
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * Name of an attr from the Op.
+       * </pre>
+       *
+       * <code>string name = 1;</code>
+       */
+      public Builder clearName() {
+        
+        name_ = getDefaultInstance().getName();
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * Name of an attr from the Op.
+       * </pre>
+       *
+       * <code>string name = 1;</code>
+       */
+      public Builder setNameBytes(
+          com.google.protobuf.ByteString value) {
+        if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+        
+        name_ = value;
+        onChanged();
+        return this;
+      }
+
+      private org.tensorflow.proto.framework.AttrValue allowedValues_;
+      private com.google.protobuf.SingleFieldBuilderV3<
+          org.tensorflow.proto.framework.AttrValue, org.tensorflow.proto.framework.AttrValue.Builder, org.tensorflow.proto.framework.AttrValueOrBuilder> allowedValuesBuilder_;
+      /**
+       * <pre>
+       * A list of values that this kernel supports for this attr.
+       * Like OpDef.AttrDef.allowed_values, except for kernels instead of Ops.
+       * </pre>
+       *
+       * <code>.tensorflow.AttrValue allowed_values = 2;</code>
+       */
+      public boolean hasAllowedValues() {
+        return allowedValuesBuilder_ != null || allowedValues_ != null;
+      }
+      /**
+       * <pre>
+       * A list of values that this kernel supports for this attr.
+       * Like OpDef.AttrDef.allowed_values, except for kernels instead of Ops.
+       * </pre>
+       *
+       * <code>.tensorflow.AttrValue allowed_values = 2;</code>
+       */
+      public org.tensorflow.proto.framework.AttrValue getAllowedValues() {
+        if (allowedValuesBuilder_ == null) {
+          return allowedValues_ == null ? org.tensorflow.proto.framework.AttrValue.getDefaultInstance() : allowedValues_;
+        } else {
+          return allowedValuesBuilder_.getMessage();
+        }
+      }
+      /**
+       * <pre>
+       * A list of values that this kernel supports for this attr.
+       * Like OpDef.AttrDef.allowed_values, except for kernels instead of Ops.
+       * </pre>
+       *
+       * <code>.tensorflow.AttrValue allowed_values = 2;</code>
+       */
+      public Builder setAllowedValues(org.tensorflow.proto.framework.AttrValue value) {
+        if (allowedValuesBuilder_ == null) {
+          if (value == null) {
+            throw new NullPointerException();
+          }
+          allowedValues_ = value;
+          onChanged();
+        } else {
+          allowedValuesBuilder_.setMessage(value);
+        }
+
+        return this;
+      }
+      /**
+       * <pre>
+       * A list of values that this kernel supports for this attr.
+       * Like OpDef.AttrDef.allowed_values, except for kernels instead of Ops.
+       * </pre>
+       *
+       * <code>.tensorflow.AttrValue allowed_values = 2;</code>
+       */
+      public Builder setAllowedValues(
+          org.tensorflow.proto.framework.AttrValue.Builder builderForValue) {
+        if (allowedValuesBuilder_ == null) {
+          allowedValues_ = builderForValue.build();
+          onChanged();
+        } else {
+          allowedValuesBuilder_.setMessage(builderForValue.build());
+        }
+
+        return this;
+      }
+      /**
+       * <pre>
+       * A list of values that this kernel supports for this attr.
+       * Like OpDef.AttrDef.allowed_values, except for kernels instead of Ops.
+       * </pre>
+       *
+       * <code>.tensorflow.AttrValue allowed_values = 2;</code>
+       */
+      public Builder mergeAllowedValues(org.tensorflow.proto.framework.AttrValue value) {
+        if (allowedValuesBuilder_ == null) {
+          if (allowedValues_ != null) {
+            allowedValues_ =
+              org.tensorflow.proto.framework.AttrValue.newBuilder(allowedValues_).mergeFrom(value).buildPartial();
+          } else {
+            allowedValues_ = value;
+          }
+          onChanged();
+        } else {
+          allowedValuesBuilder_.mergeFrom(value);
+        }
+
+        return this;
+      }
+      /**
+       * <pre>
+       * A list of values that this kernel supports for this attr.
+       * Like OpDef.AttrDef.allowed_values, except for kernels instead of Ops.
+       * </pre>
+       *
+       * <code>.tensorflow.AttrValue allowed_values = 2;</code>
+       */
+      public Builder clearAllowedValues() {
+        if (allowedValuesBuilder_ == null) {
+          allowedValues_ = null;
+          onChanged();
+        } else {
+          allowedValues_ = null;
+          allowedValuesBuilder_ = null;
+        }
+
+        return this;
+      }
+      /**
+       * <pre>
+       * A list of values that this kernel supports for this attr.
+       * Like OpDef.AttrDef.allowed_values, except for kernels instead of Ops.
+       * </pre>
+       *
+       * <code>.tensorflow.AttrValue allowed_values = 2;</code>
+       */
+      public org.tensorflow.proto.framework.AttrValue.Builder getAllowedValuesBuilder() {
+        
+        onChanged();
+        return getAllowedValuesFieldBuilder().getBuilder();
+      }
+      /**
+       * <pre>
+       * A list of values that this kernel supports for this attr.
+       * Like OpDef.AttrDef.allowed_values, except for kernels instead of Ops.
+       * </pre>
+       *
+       * <code>.tensorflow.AttrValue allowed_values = 2;</code>
+       */
+      public org.tensorflow.proto.framework.AttrValueOrBuilder getAllowedValuesOrBuilder() {
+        if (allowedValuesBuilder_ != null) {
+          return allowedValuesBuilder_.getMessageOrBuilder();
+        } else {
+          return allowedValues_ == null ?
+              org.tensorflow.proto.framework.AttrValue.getDefaultInstance() : allowedValues_;
+        }
+      }
+      /**
+       * <pre>
+       * A list of values that this kernel supports for this attr.
+       * Like OpDef.AttrDef.allowed_values, except for kernels instead of Ops.
+       * </pre>
+       *
+       * <code>.tensorflow.AttrValue allowed_values = 2;</code>
+       */
+      private com.google.protobuf.SingleFieldBuilderV3<
+          org.tensorflow.proto.framework.AttrValue, org.tensorflow.proto.framework.AttrValue.Builder, org.tensorflow.proto.framework.AttrValueOrBuilder> 
+          getAllowedValuesFieldBuilder() {
+        if (allowedValuesBuilder_ == null) {
+          allowedValuesBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+              org.tensorflow.proto.framework.AttrValue, org.tensorflow.proto.framework.AttrValue.Builder, org.tensorflow.proto.framework.AttrValueOrBuilder>(
+                  getAllowedValues(),
+                  getParentForChildren(),
+                  isClean());
+          allowedValues_ = null;
+        }
+        return allowedValuesBuilder_;
+      }
+      @java.lang.Override
+      public final Builder setUnknownFields(
+          final com.google.protobuf.UnknownFieldSet unknownFields) {
+        return super.setUnknownFields(unknownFields);
+      }
+
+      @java.lang.Override
+      public final Builder mergeUnknownFields(
+          final com.google.protobuf.UnknownFieldSet unknownFields) {
+        return super.mergeUnknownFields(unknownFields);
+      }
+
+
+      // @@protoc_insertion_point(builder_scope:tensorflow.KernelDef.AttrConstraint)
+    }
+
+    // @@protoc_insertion_point(class_scope:tensorflow.KernelDef.AttrConstraint)
+    private static final org.tensorflow.proto.framework.KernelDef.AttrConstraint DEFAULT_INSTANCE;
+    static {
+      DEFAULT_INSTANCE = new org.tensorflow.proto.framework.KernelDef.AttrConstraint();
+    }
+
+    public static org.tensorflow.proto.framework.KernelDef.AttrConstraint getDefaultInstance() {
+      return DEFAULT_INSTANCE;
+    }
+
+    private static final com.google.protobuf.Parser<AttrConstraint>
+        PARSER = new com.google.protobuf.AbstractParser<AttrConstraint>() {
+      @java.lang.Override
+      public AttrConstraint parsePartialFrom(
+          com.google.protobuf.CodedInputStream input,
+          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+          throws com.google.protobuf.InvalidProtocolBufferException {
+        return new AttrConstraint(input, extensionRegistry);
+      }
+    };
+
+    public static com.google.protobuf.Parser<AttrConstraint> parser() {
+      return PARSER;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Parser<AttrConstraint> getParserForType() {
+      return PARSER;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.KernelDef.AttrConstraint getDefaultInstanceForType() {
+      return DEFAULT_INSTANCE;
+    }
+
+  }
+
+  public static final int OP_FIELD_NUMBER = 1;
+  private volatile java.lang.Object op_;
+  /**
+   * <pre>
+   * Must match the name of an Op.
+   * </pre>
+   *
+   * <code>string op = 1;</code>
+   */
+  public java.lang.String getOp() {
+    java.lang.Object ref = op_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      op_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * Must match the name of an Op.
+   * </pre>
+   *
+   * <code>string op = 1;</code>
+   */
+  public com.google.protobuf.ByteString
+      getOpBytes() {
+    java.lang.Object ref = op_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      op_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int DEVICE_TYPE_FIELD_NUMBER = 2;
+  private volatile java.lang.Object deviceType_;
+  /**
+   * <pre>
+   * Type of device this kernel runs on.
+   * </pre>
+   *
+   * <code>string device_type = 2;</code>
+   */
+  public java.lang.String getDeviceType() {
+    java.lang.Object ref = deviceType_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      deviceType_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * Type of device this kernel runs on.
+   * </pre>
+   *
+   * <code>string device_type = 2;</code>
+   */
+  public com.google.protobuf.ByteString
+      getDeviceTypeBytes() {
+    java.lang.Object ref = deviceType_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      deviceType_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int CONSTRAINT_FIELD_NUMBER = 3;
+  private java.util.List<org.tensorflow.proto.framework.KernelDef.AttrConstraint> constraint_;
+  /**
+   * <code>repeated .tensorflow.KernelDef.AttrConstraint constraint = 3;</code>
+   */
+  public java.util.List<org.tensorflow.proto.framework.KernelDef.AttrConstraint> getConstraintList() {
+    return constraint_;
+  }
+  /**
+   * <code>repeated .tensorflow.KernelDef.AttrConstraint constraint = 3;</code>
+   */
+  public java.util.List<? extends org.tensorflow.proto.framework.KernelDef.AttrConstraintOrBuilder> 
+      getConstraintOrBuilderList() {
+    return constraint_;
+  }
+  /**
+   * <code>repeated .tensorflow.KernelDef.AttrConstraint constraint = 3;</code>
+   */
+  public int getConstraintCount() {
+    return constraint_.size();
+  }
+  /**
+   * <code>repeated .tensorflow.KernelDef.AttrConstraint constraint = 3;</code>
+   */
+  public org.tensorflow.proto.framework.KernelDef.AttrConstraint getConstraint(int index) {
+    return constraint_.get(index);
+  }
+  /**
+   * <code>repeated .tensorflow.KernelDef.AttrConstraint constraint = 3;</code>
+   */
+  public org.tensorflow.proto.framework.KernelDef.AttrConstraintOrBuilder getConstraintOrBuilder(
+      int index) {
+    return constraint_.get(index);
+  }
+
+  public static final int HOST_MEMORY_ARG_FIELD_NUMBER = 4;
+  private com.google.protobuf.LazyStringList hostMemoryArg_;
+  /**
+   * <pre>
+   * Names of the Op's input_/output_args that reside in host memory
+   * instead of device memory.
+   * </pre>
+   *
+   * <code>repeated string host_memory_arg = 4;</code>
+   */
+  public com.google.protobuf.ProtocolStringList
+      getHostMemoryArgList() {
+    return hostMemoryArg_;
+  }
+  /**
+   * <pre>
+   * Names of the Op's input_/output_args that reside in host memory
+   * instead of device memory.
+   * </pre>
+   *
+   * <code>repeated string host_memory_arg = 4;</code>
+   */
+  public int getHostMemoryArgCount() {
+    return hostMemoryArg_.size();
+  }
+  /**
+   * <pre>
+   * Names of the Op's input_/output_args that reside in host memory
+   * instead of device memory.
+   * </pre>
+   *
+   * <code>repeated string host_memory_arg = 4;</code>
+   */
+  public java.lang.String getHostMemoryArg(int index) {
+    return hostMemoryArg_.get(index);
+  }
+  /**
+   * <pre>
+   * Names of the Op's input_/output_args that reside in host memory
+   * instead of device memory.
+   * </pre>
+   *
+   * <code>repeated string host_memory_arg = 4;</code>
+   */
+  public com.google.protobuf.ByteString
+      getHostMemoryArgBytes(int index) {
+    return hostMemoryArg_.getByteString(index);
+  }
+
+  public static final int LABEL_FIELD_NUMBER = 5;
+  private volatile java.lang.Object label_;
+  /**
+   * <pre>
+   * This allows experimental kernels to be registered for an op that
+   * won't be used unless the user specifies a "_kernel" attr with
+   * value matching this.
+   * </pre>
+   *
+   * <code>string label = 5;</code>
+   */
+  public java.lang.String getLabel() {
+    java.lang.Object ref = label_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      label_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * This allows experimental kernels to be registered for an op that
+   * won't be used unless the user specifies a "_kernel" attr with
+   * value matching this.
+   * </pre>
+   *
+   * <code>string label = 5;</code>
+   */
+  public com.google.protobuf.ByteString
+      getLabelBytes() {
+    java.lang.Object ref = label_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      label_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int PRIORITY_FIELD_NUMBER = 6;
+  private int priority_;
+  /**
+   * <pre>
+   * Prioritization of kernel amongst different devices. By default we assume
+   * priority is 0. The higher the priority the better. By default (i.e. if
+   * this is not set), we prefer GPU kernels over CPU.
+   * </pre>
+   *
+   * <code>int32 priority = 6;</code>
+   */
+  public int getPriority() {
+    return priority_;
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    if (!getOpBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 1, op_);
+    }
+    if (!getDeviceTypeBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 2, deviceType_);
+    }
+    for (int i = 0; i < constraint_.size(); i++) {
+      output.writeMessage(3, constraint_.get(i));
+    }
+    for (int i = 0; i < hostMemoryArg_.size(); i++) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 4, hostMemoryArg_.getRaw(i));
+    }
+    if (!getLabelBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 5, label_);
+    }
+    if (priority_ != 0) {
+      output.writeInt32(6, priority_);
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (!getOpBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, op_);
+    }
+    if (!getDeviceTypeBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(2, deviceType_);
+    }
+    for (int i = 0; i < constraint_.size(); i++) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(3, constraint_.get(i));
+    }
+    {
+      int dataSize = 0;
+      for (int i = 0; i < hostMemoryArg_.size(); i++) {
+        dataSize += computeStringSizeNoTag(hostMemoryArg_.getRaw(i));
+      }
+      size += dataSize;
+      size += 1 * getHostMemoryArgList().size();
+    }
+    if (!getLabelBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(5, label_);
+    }
+    if (priority_ != 0) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt32Size(6, priority_);
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.framework.KernelDef)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.framework.KernelDef other = (org.tensorflow.proto.framework.KernelDef) obj;
+
+    if (!getOp()
+        .equals(other.getOp())) return false;
+    if (!getDeviceType()
+        .equals(other.getDeviceType())) return false;
+    if (!getConstraintList()
+        .equals(other.getConstraintList())) return false;
+    if (!getHostMemoryArgList()
+        .equals(other.getHostMemoryArgList())) return false;
+    if (!getLabel()
+        .equals(other.getLabel())) return false;
+    if (getPriority()
+        != other.getPriority()) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    hash = (37 * hash) + OP_FIELD_NUMBER;
+    hash = (53 * hash) + getOp().hashCode();
+    hash = (37 * hash) + DEVICE_TYPE_FIELD_NUMBER;
+    hash = (53 * hash) + getDeviceType().hashCode();
+    if (getConstraintCount() > 0) {
+      hash = (37 * hash) + CONSTRAINT_FIELD_NUMBER;
+      hash = (53 * hash) + getConstraintList().hashCode();
+    }
+    if (getHostMemoryArgCount() > 0) {
+      hash = (37 * hash) + HOST_MEMORY_ARG_FIELD_NUMBER;
+      hash = (53 * hash) + getHostMemoryArgList().hashCode();
+    }
+    hash = (37 * hash) + LABEL_FIELD_NUMBER;
+    hash = (53 * hash) + getLabel().hashCode();
+    hash = (37 * hash) + PRIORITY_FIELD_NUMBER;
+    hash = (53 * hash) + getPriority();
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.framework.KernelDef parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.KernelDef parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.KernelDef parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.KernelDef parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.KernelDef parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.KernelDef parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.KernelDef parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.KernelDef parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.KernelDef parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.KernelDef parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.KernelDef parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.KernelDef parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.framework.KernelDef prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * Protobuf type {@code tensorflow.KernelDef}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.KernelDef)
+      org.tensorflow.proto.framework.KernelDefOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.KernelDefProtos.internal_static_tensorflow_KernelDef_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.KernelDefProtos.internal_static_tensorflow_KernelDef_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.KernelDef.class, org.tensorflow.proto.framework.KernelDef.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.framework.KernelDef.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+        getConstraintFieldBuilder();
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      op_ = "";
+
+      deviceType_ = "";
+
+      if (constraintBuilder_ == null) {
+        constraint_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000001);
+      } else {
+        constraintBuilder_.clear();
+      }
+      hostMemoryArg_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+      bitField0_ = (bitField0_ & ~0x00000002);
+      label_ = "";
+
+      priority_ = 0;
+
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.framework.KernelDefProtos.internal_static_tensorflow_KernelDef_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.KernelDef getDefaultInstanceForType() {
+      return org.tensorflow.proto.framework.KernelDef.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.KernelDef build() {
+      org.tensorflow.proto.framework.KernelDef result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.KernelDef buildPartial() {
+      org.tensorflow.proto.framework.KernelDef result = new org.tensorflow.proto.framework.KernelDef(this);
+      int from_bitField0_ = bitField0_;
+      result.op_ = op_;
+      result.deviceType_ = deviceType_;
+      if (constraintBuilder_ == null) {
+        if (((bitField0_ & 0x00000001) != 0)) {
+          constraint_ = java.util.Collections.unmodifiableList(constraint_);
+          bitField0_ = (bitField0_ & ~0x00000001);
+        }
+        result.constraint_ = constraint_;
+      } else {
+        result.constraint_ = constraintBuilder_.build();
+      }
+      if (((bitField0_ & 0x00000002) != 0)) {
+        hostMemoryArg_ = hostMemoryArg_.getUnmodifiableView();
+        bitField0_ = (bitField0_ & ~0x00000002);
+      }
+      result.hostMemoryArg_ = hostMemoryArg_;
+      result.label_ = label_;
+      result.priority_ = priority_;
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.framework.KernelDef) {
+        return mergeFrom((org.tensorflow.proto.framework.KernelDef)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.framework.KernelDef other) {
+      if (other == org.tensorflow.proto.framework.KernelDef.getDefaultInstance()) return this;
+      if (!other.getOp().isEmpty()) {
+        op_ = other.op_;
+        onChanged();
+      }
+      if (!other.getDeviceType().isEmpty()) {
+        deviceType_ = other.deviceType_;
+        onChanged();
+      }
+      if (constraintBuilder_ == null) {
+        if (!other.constraint_.isEmpty()) {
+          if (constraint_.isEmpty()) {
+            constraint_ = other.constraint_;
+            bitField0_ = (bitField0_ & ~0x00000001);
+          } else {
+            ensureConstraintIsMutable();
+            constraint_.addAll(other.constraint_);
+          }
+          onChanged();
+        }
+      } else {
+        if (!other.constraint_.isEmpty()) {
+          if (constraintBuilder_.isEmpty()) {
+            constraintBuilder_.dispose();
+            constraintBuilder_ = null;
+            constraint_ = other.constraint_;
+            bitField0_ = (bitField0_ & ~0x00000001);
+            constraintBuilder_ = 
+              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
+                 getConstraintFieldBuilder() : null;
+          } else {
+            constraintBuilder_.addAllMessages(other.constraint_);
+          }
+        }
+      }
+      if (!other.hostMemoryArg_.isEmpty()) {
+        if (hostMemoryArg_.isEmpty()) {
+          hostMemoryArg_ = other.hostMemoryArg_;
+          bitField0_ = (bitField0_ & ~0x00000002);
+        } else {
+          ensureHostMemoryArgIsMutable();
+          hostMemoryArg_.addAll(other.hostMemoryArg_);
+        }
+        onChanged();
+      }
+      if (!other.getLabel().isEmpty()) {
+        label_ = other.label_;
+        onChanged();
+      }
+      if (other.getPriority() != 0) {
+        setPriority(other.getPriority());
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.framework.KernelDef parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.framework.KernelDef) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+    private int bitField0_;
+
+    private java.lang.Object op_ = "";
+    /**
+     * <pre>
+     * Must match the name of an Op.
+     * </pre>
+     *
+     * <code>string op = 1;</code>
+     */
+    public java.lang.String getOp() {
+      java.lang.Object ref = op_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        op_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Must match the name of an Op.
+     * </pre>
+     *
+     * <code>string op = 1;</code>
+     */
+    public com.google.protobuf.ByteString
+        getOpBytes() {
+      java.lang.Object ref = op_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        op_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Must match the name of an Op.
+     * </pre>
+     *
+     * <code>string op = 1;</code>
+     */
+    public Builder setOp(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      op_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Must match the name of an Op.
+     * </pre>
+     *
+     * <code>string op = 1;</code>
+     */
+    public Builder clearOp() {
+      
+      op_ = getDefaultInstance().getOp();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Must match the name of an Op.
+     * </pre>
+     *
+     * <code>string op = 1;</code>
+     */
+    public Builder setOpBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      op_ = value;
+      onChanged();
+      return this;
+    }
+
+    private java.lang.Object deviceType_ = "";
+    /**
+     * <pre>
+     * Type of device this kernel runs on.
+     * </pre>
+     *
+     * <code>string device_type = 2;</code>
+     */
+    public java.lang.String getDeviceType() {
+      java.lang.Object ref = deviceType_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        deviceType_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Type of device this kernel runs on.
+     * </pre>
+     *
+     * <code>string device_type = 2;</code>
+     */
+    public com.google.protobuf.ByteString
+        getDeviceTypeBytes() {
+      java.lang.Object ref = deviceType_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        deviceType_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Type of device this kernel runs on.
+     * </pre>
+     *
+     * <code>string device_type = 2;</code>
+     */
+    public Builder setDeviceType(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      deviceType_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Type of device this kernel runs on.
+     * </pre>
+     *
+     * <code>string device_type = 2;</code>
+     */
+    public Builder clearDeviceType() {
+      
+      deviceType_ = getDefaultInstance().getDeviceType();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Type of device this kernel runs on.
+     * </pre>
+     *
+     * <code>string device_type = 2;</code>
+     */
+    public Builder setDeviceTypeBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      deviceType_ = value;
+      onChanged();
+      return this;
+    }
+
+    private java.util.List<org.tensorflow.proto.framework.KernelDef.AttrConstraint> constraint_ =
+      java.util.Collections.emptyList();
+    private void ensureConstraintIsMutable() {
+      if (!((bitField0_ & 0x00000001) != 0)) {
+        constraint_ = new java.util.ArrayList<org.tensorflow.proto.framework.KernelDef.AttrConstraint>(constraint_);
+        bitField0_ |= 0x00000001;
+       }
+    }
+
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.framework.KernelDef.AttrConstraint, org.tensorflow.proto.framework.KernelDef.AttrConstraint.Builder, org.tensorflow.proto.framework.KernelDef.AttrConstraintOrBuilder> constraintBuilder_;
+
+    /**
+     * <code>repeated .tensorflow.KernelDef.AttrConstraint constraint = 3;</code>
+     */
+    public java.util.List<org.tensorflow.proto.framework.KernelDef.AttrConstraint> getConstraintList() {
+      if (constraintBuilder_ == null) {
+        return java.util.Collections.unmodifiableList(constraint_);
+      } else {
+        return constraintBuilder_.getMessageList();
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.KernelDef.AttrConstraint constraint = 3;</code>
+     */
+    public int getConstraintCount() {
+      if (constraintBuilder_ == null) {
+        return constraint_.size();
+      } else {
+        return constraintBuilder_.getCount();
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.KernelDef.AttrConstraint constraint = 3;</code>
+     */
+    public org.tensorflow.proto.framework.KernelDef.AttrConstraint getConstraint(int index) {
+      if (constraintBuilder_ == null) {
+        return constraint_.get(index);
+      } else {
+        return constraintBuilder_.getMessage(index);
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.KernelDef.AttrConstraint constraint = 3;</code>
+     */
+    public Builder setConstraint(
+        int index, org.tensorflow.proto.framework.KernelDef.AttrConstraint value) {
+      if (constraintBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureConstraintIsMutable();
+        constraint_.set(index, value);
+        onChanged();
+      } else {
+        constraintBuilder_.setMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.KernelDef.AttrConstraint constraint = 3;</code>
+     */
+    public Builder setConstraint(
+        int index, org.tensorflow.proto.framework.KernelDef.AttrConstraint.Builder builderForValue) {
+      if (constraintBuilder_ == null) {
+        ensureConstraintIsMutable();
+        constraint_.set(index, builderForValue.build());
+        onChanged();
+      } else {
+        constraintBuilder_.setMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.KernelDef.AttrConstraint constraint = 3;</code>
+     */
+    public Builder addConstraint(org.tensorflow.proto.framework.KernelDef.AttrConstraint value) {
+      if (constraintBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureConstraintIsMutable();
+        constraint_.add(value);
+        onChanged();
+      } else {
+        constraintBuilder_.addMessage(value);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.KernelDef.AttrConstraint constraint = 3;</code>
+     */
+    public Builder addConstraint(
+        int index, org.tensorflow.proto.framework.KernelDef.AttrConstraint value) {
+      if (constraintBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureConstraintIsMutable();
+        constraint_.add(index, value);
+        onChanged();
+      } else {
+        constraintBuilder_.addMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.KernelDef.AttrConstraint constraint = 3;</code>
+     */
+    public Builder addConstraint(
+        org.tensorflow.proto.framework.KernelDef.AttrConstraint.Builder builderForValue) {
+      if (constraintBuilder_ == null) {
+        ensureConstraintIsMutable();
+        constraint_.add(builderForValue.build());
+        onChanged();
+      } else {
+        constraintBuilder_.addMessage(builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.KernelDef.AttrConstraint constraint = 3;</code>
+     */
+    public Builder addConstraint(
+        int index, org.tensorflow.proto.framework.KernelDef.AttrConstraint.Builder builderForValue) {
+      if (constraintBuilder_ == null) {
+        ensureConstraintIsMutable();
+        constraint_.add(index, builderForValue.build());
+        onChanged();
+      } else {
+        constraintBuilder_.addMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.KernelDef.AttrConstraint constraint = 3;</code>
+     */
+    public Builder addAllConstraint(
+        java.lang.Iterable<? extends org.tensorflow.proto.framework.KernelDef.AttrConstraint> values) {
+      if (constraintBuilder_ == null) {
+        ensureConstraintIsMutable();
+        com.google.protobuf.AbstractMessageLite.Builder.addAll(
+            values, constraint_);
+        onChanged();
+      } else {
+        constraintBuilder_.addAllMessages(values);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.KernelDef.AttrConstraint constraint = 3;</code>
+     */
+    public Builder clearConstraint() {
+      if (constraintBuilder_ == null) {
+        constraint_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000001);
+        onChanged();
+      } else {
+        constraintBuilder_.clear();
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.KernelDef.AttrConstraint constraint = 3;</code>
+     */
+    public Builder removeConstraint(int index) {
+      if (constraintBuilder_ == null) {
+        ensureConstraintIsMutable();
+        constraint_.remove(index);
+        onChanged();
+      } else {
+        constraintBuilder_.remove(index);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.KernelDef.AttrConstraint constraint = 3;</code>
+     */
+    public org.tensorflow.proto.framework.KernelDef.AttrConstraint.Builder getConstraintBuilder(
+        int index) {
+      return getConstraintFieldBuilder().getBuilder(index);
+    }
+    /**
+     * <code>repeated .tensorflow.KernelDef.AttrConstraint constraint = 3;</code>
+     */
+    public org.tensorflow.proto.framework.KernelDef.AttrConstraintOrBuilder getConstraintOrBuilder(
+        int index) {
+      if (constraintBuilder_ == null) {
+        return constraint_.get(index);  } else {
+        return constraintBuilder_.getMessageOrBuilder(index);
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.KernelDef.AttrConstraint constraint = 3;</code>
+     */
+    public java.util.List<? extends org.tensorflow.proto.framework.KernelDef.AttrConstraintOrBuilder> 
+         getConstraintOrBuilderList() {
+      if (constraintBuilder_ != null) {
+        return constraintBuilder_.getMessageOrBuilderList();
+      } else {
+        return java.util.Collections.unmodifiableList(constraint_);
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.KernelDef.AttrConstraint constraint = 3;</code>
+     */
+    public org.tensorflow.proto.framework.KernelDef.AttrConstraint.Builder addConstraintBuilder() {
+      return getConstraintFieldBuilder().addBuilder(
+          org.tensorflow.proto.framework.KernelDef.AttrConstraint.getDefaultInstance());
+    }
+    /**
+     * <code>repeated .tensorflow.KernelDef.AttrConstraint constraint = 3;</code>
+     */
+    public org.tensorflow.proto.framework.KernelDef.AttrConstraint.Builder addConstraintBuilder(
+        int index) {
+      return getConstraintFieldBuilder().addBuilder(
+          index, org.tensorflow.proto.framework.KernelDef.AttrConstraint.getDefaultInstance());
+    }
+    /**
+     * <code>repeated .tensorflow.KernelDef.AttrConstraint constraint = 3;</code>
+     */
+    public java.util.List<org.tensorflow.proto.framework.KernelDef.AttrConstraint.Builder> 
+         getConstraintBuilderList() {
+      return getConstraintFieldBuilder().getBuilderList();
+    }
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.framework.KernelDef.AttrConstraint, org.tensorflow.proto.framework.KernelDef.AttrConstraint.Builder, org.tensorflow.proto.framework.KernelDef.AttrConstraintOrBuilder> 
+        getConstraintFieldBuilder() {
+      if (constraintBuilder_ == null) {
+        constraintBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
+            org.tensorflow.proto.framework.KernelDef.AttrConstraint, org.tensorflow.proto.framework.KernelDef.AttrConstraint.Builder, org.tensorflow.proto.framework.KernelDef.AttrConstraintOrBuilder>(
+                constraint_,
+                ((bitField0_ & 0x00000001) != 0),
+                getParentForChildren(),
+                isClean());
+        constraint_ = null;
+      }
+      return constraintBuilder_;
+    }
+
+    private com.google.protobuf.LazyStringList hostMemoryArg_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+    private void ensureHostMemoryArgIsMutable() {
+      if (!((bitField0_ & 0x00000002) != 0)) {
+        hostMemoryArg_ = new com.google.protobuf.LazyStringArrayList(hostMemoryArg_);
+        bitField0_ |= 0x00000002;
+       }
+    }
+    /**
+     * <pre>
+     * Names of the Op's input_/output_args that reside in host memory
+     * instead of device memory.
+     * </pre>
+     *
+     * <code>repeated string host_memory_arg = 4;</code>
+     */
+    public com.google.protobuf.ProtocolStringList
+        getHostMemoryArgList() {
+      return hostMemoryArg_.getUnmodifiableView();
+    }
+    /**
+     * <pre>
+     * Names of the Op's input_/output_args that reside in host memory
+     * instead of device memory.
+     * </pre>
+     *
+     * <code>repeated string host_memory_arg = 4;</code>
+     */
+    public int getHostMemoryArgCount() {
+      return hostMemoryArg_.size();
+    }
+    /**
+     * <pre>
+     * Names of the Op's input_/output_args that reside in host memory
+     * instead of device memory.
+     * </pre>
+     *
+     * <code>repeated string host_memory_arg = 4;</code>
+     */
+    public java.lang.String getHostMemoryArg(int index) {
+      return hostMemoryArg_.get(index);
+    }
+    /**
+     * <pre>
+     * Names of the Op's input_/output_args that reside in host memory
+     * instead of device memory.
+     * </pre>
+     *
+     * <code>repeated string host_memory_arg = 4;</code>
+     */
+    public com.google.protobuf.ByteString
+        getHostMemoryArgBytes(int index) {
+      return hostMemoryArg_.getByteString(index);
+    }
+    /**
+     * <pre>
+     * Names of the Op's input_/output_args that reside in host memory
+     * instead of device memory.
+     * </pre>
+     *
+     * <code>repeated string host_memory_arg = 4;</code>
+     */
+    public Builder setHostMemoryArg(
+        int index, java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  ensureHostMemoryArgIsMutable();
+      hostMemoryArg_.set(index, value);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Names of the Op's input_/output_args that reside in host memory
+     * instead of device memory.
+     * </pre>
+     *
+     * <code>repeated string host_memory_arg = 4;</code>
+     */
+    public Builder addHostMemoryArg(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  ensureHostMemoryArgIsMutable();
+      hostMemoryArg_.add(value);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Names of the Op's input_/output_args that reside in host memory
+     * instead of device memory.
+     * </pre>
+     *
+     * <code>repeated string host_memory_arg = 4;</code>
+     */
+    public Builder addAllHostMemoryArg(
+        java.lang.Iterable<java.lang.String> values) {
+      ensureHostMemoryArgIsMutable();
+      com.google.protobuf.AbstractMessageLite.Builder.addAll(
+          values, hostMemoryArg_);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Names of the Op's input_/output_args that reside in host memory
+     * instead of device memory.
+     * </pre>
+     *
+     * <code>repeated string host_memory_arg = 4;</code>
+     */
+    public Builder clearHostMemoryArg() {
+      hostMemoryArg_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+      bitField0_ = (bitField0_ & ~0x00000002);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Names of the Op's input_/output_args that reside in host memory
+     * instead of device memory.
+     * </pre>
+     *
+     * <code>repeated string host_memory_arg = 4;</code>
+     */
+    public Builder addHostMemoryArgBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      ensureHostMemoryArgIsMutable();
+      hostMemoryArg_.add(value);
+      onChanged();
+      return this;
+    }
+
+    private java.lang.Object label_ = "";
+    /**
+     * <pre>
+     * This allows experimental kernels to be registered for an op that
+     * won't be used unless the user specifies a "_kernel" attr with
+     * value matching this.
+     * </pre>
+     *
+     * <code>string label = 5;</code>
+     */
+    public java.lang.String getLabel() {
+      java.lang.Object ref = label_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        label_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * This allows experimental kernels to be registered for an op that
+     * won't be used unless the user specifies a "_kernel" attr with
+     * value matching this.
+     * </pre>
+     *
+     * <code>string label = 5;</code>
+     */
+    public com.google.protobuf.ByteString
+        getLabelBytes() {
+      java.lang.Object ref = label_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        label_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * This allows experimental kernels to be registered for an op that
+     * won't be used unless the user specifies a "_kernel" attr with
+     * value matching this.
+     * </pre>
+     *
+     * <code>string label = 5;</code>
+     */
+    public Builder setLabel(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      label_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * This allows experimental kernels to be registered for an op that
+     * won't be used unless the user specifies a "_kernel" attr with
+     * value matching this.
+     * </pre>
+     *
+     * <code>string label = 5;</code>
+     */
+    public Builder clearLabel() {
+      
+      label_ = getDefaultInstance().getLabel();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * This allows experimental kernels to be registered for an op that
+     * won't be used unless the user specifies a "_kernel" attr with
+     * value matching this.
+     * </pre>
+     *
+     * <code>string label = 5;</code>
+     */
+    public Builder setLabelBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      label_ = value;
+      onChanged();
+      return this;
+    }
+
+    private int priority_ ;
+    /**
+     * <pre>
+     * Prioritization of kernel amongst different devices. By default we assume
+     * priority is 0. The higher the priority the better. By default (i.e. if
+     * this is not set), we prefer GPU kernels over CPU.
+     * </pre>
+     *
+     * <code>int32 priority = 6;</code>
+     */
+    public int getPriority() {
+      return priority_;
+    }
+    /**
+     * <pre>
+     * Prioritization of kernel amongst different devices. By default we assume
+     * priority is 0. The higher the priority the better. By default (i.e. if
+     * this is not set), we prefer GPU kernels over CPU.
+     * </pre>
+     *
+     * <code>int32 priority = 6;</code>
+     */
+    public Builder setPriority(int value) {
+      
+      priority_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Prioritization of kernel amongst different devices. By default we assume
+     * priority is 0. The higher the priority the better. By default (i.e. if
+     * this is not set), we prefer GPU kernels over CPU.
+     * </pre>
+     *
+     * <code>int32 priority = 6;</code>
+     */
+    public Builder clearPriority() {
+      
+      priority_ = 0;
+      onChanged();
+      return this;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.KernelDef)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.KernelDef)
+  private static final org.tensorflow.proto.framework.KernelDef DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.KernelDef();
+  }
+
+  public static org.tensorflow.proto.framework.KernelDef getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<KernelDef>
+      PARSER = new com.google.protobuf.AbstractParser<KernelDef>() {
+    @java.lang.Override
+    public KernelDef parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new KernelDef(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<KernelDef> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<KernelDef> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.framework.KernelDef getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/KernelDefOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/KernelDefOrBuilder.java
new file mode 100644
index 00000000000..2d62f81acc3
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/KernelDefOrBuilder.java
@@ -0,0 +1,141 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/framework/kernel_def.proto
+
+package org.tensorflow.proto.framework;
+
+public interface KernelDefOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.KernelDef)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <pre>
+   * Must match the name of an Op.
+   * </pre>
+   *
+   * <code>string op = 1;</code>
+   */
+  java.lang.String getOp();
+  /**
+   * <pre>
+   * Must match the name of an Op.
+   * </pre>
+   *
+   * <code>string op = 1;</code>
+   */
+  com.google.protobuf.ByteString
+      getOpBytes();
+
+  /**
+   * <pre>
+   * Type of device this kernel runs on.
+   * </pre>
+   *
+   * <code>string device_type = 2;</code>
+   */
+  java.lang.String getDeviceType();
+  /**
+   * <pre>
+   * Type of device this kernel runs on.
+   * </pre>
+   *
+   * <code>string device_type = 2;</code>
+   */
+  com.google.protobuf.ByteString
+      getDeviceTypeBytes();
+
+  /**
+   * <code>repeated .tensorflow.KernelDef.AttrConstraint constraint = 3;</code>
+   */
+  java.util.List<org.tensorflow.proto.framework.KernelDef.AttrConstraint> 
+      getConstraintList();
+  /**
+   * <code>repeated .tensorflow.KernelDef.AttrConstraint constraint = 3;</code>
+   */
+  org.tensorflow.proto.framework.KernelDef.AttrConstraint getConstraint(int index);
+  /**
+   * <code>repeated .tensorflow.KernelDef.AttrConstraint constraint = 3;</code>
+   */
+  int getConstraintCount();
+  /**
+   * <code>repeated .tensorflow.KernelDef.AttrConstraint constraint = 3;</code>
+   */
+  java.util.List<? extends org.tensorflow.proto.framework.KernelDef.AttrConstraintOrBuilder> 
+      getConstraintOrBuilderList();
+  /**
+   * <code>repeated .tensorflow.KernelDef.AttrConstraint constraint = 3;</code>
+   */
+  org.tensorflow.proto.framework.KernelDef.AttrConstraintOrBuilder getConstraintOrBuilder(
+      int index);
+
+  /**
+   * <pre>
+   * Names of the Op's input_/output_args that reside in host memory
+   * instead of device memory.
+   * </pre>
+   *
+   * <code>repeated string host_memory_arg = 4;</code>
+   */
+  java.util.List<java.lang.String>
+      getHostMemoryArgList();
+  /**
+   * <pre>
+   * Names of the Op's input_/output_args that reside in host memory
+   * instead of device memory.
+   * </pre>
+   *
+   * <code>repeated string host_memory_arg = 4;</code>
+   */
+  int getHostMemoryArgCount();
+  /**
+   * <pre>
+   * Names of the Op's input_/output_args that reside in host memory
+   * instead of device memory.
+   * </pre>
+   *
+   * <code>repeated string host_memory_arg = 4;</code>
+   */
+  java.lang.String getHostMemoryArg(int index);
+  /**
+   * <pre>
+   * Names of the Op's input_/output_args that reside in host memory
+   * instead of device memory.
+   * </pre>
+   *
+   * <code>repeated string host_memory_arg = 4;</code>
+   */
+  com.google.protobuf.ByteString
+      getHostMemoryArgBytes(int index);
+
+  /**
+   * <pre>
+   * This allows experimental kernels to be registered for an op that
+   * won't be used unless the user specifies a "_kernel" attr with
+   * value matching this.
+   * </pre>
+   *
+   * <code>string label = 5;</code>
+   */
+  java.lang.String getLabel();
+  /**
+   * <pre>
+   * This allows experimental kernels to be registered for an op that
+   * won't be used unless the user specifies a "_kernel" attr with
+   * value matching this.
+   * </pre>
+   *
+   * <code>string label = 5;</code>
+   */
+  com.google.protobuf.ByteString
+      getLabelBytes();
+
+  /**
+   * <pre>
+   * Prioritization of kernel amongst different devices. By default we assume
+   * priority is 0. The higher the priority the better. By default (i.e. if
+   * this is not set), we prefer GPU kernels over CPU.
+   * </pre>
+   *
+   * <code>int32 priority = 6;</code>
+   */
+  int getPriority();
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/KernelDefProtos.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/KernelDefProtos.java
new file mode 100644
index 00000000000..50b235b600f
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/KernelDefProtos.java
@@ -0,0 +1,83 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/framework/kernel_def.proto
+
+package org.tensorflow.proto.framework;
+
+public final class KernelDefProtos {
+  private KernelDefProtos() {}
+  public static void registerAllExtensions(
+      com.google.protobuf.ExtensionRegistryLite registry) {
+  }
+
+  public static void registerAllExtensions(
+      com.google.protobuf.ExtensionRegistry registry) {
+    registerAllExtensions(
+        (com.google.protobuf.ExtensionRegistryLite) registry);
+  }
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_KernelDef_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_KernelDef_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_KernelDef_AttrConstraint_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_KernelDef_AttrConstraint_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_KernelList_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_KernelList_fieldAccessorTable;
+
+  public static com.google.protobuf.Descriptors.FileDescriptor
+      getDescriptor() {
+    return descriptor;
+  }
+  private static  com.google.protobuf.Descriptors.FileDescriptor
+      descriptor;
+  static {
+    java.lang.String[] descriptorData = {
+      "\n*tensorflow/core/framework/kernel_def.p" +
+      "roto\022\ntensorflow\032*tensorflow/core/framew" +
+      "ork/attr_value.proto\"\357\001\n\tKernelDef\022\n\n\002op" +
+      "\030\001 \001(\t\022\023\n\013device_type\030\002 \001(\t\0228\n\nconstrain" +
+      "t\030\003 \003(\0132$.tensorflow.KernelDef.AttrConst" +
+      "raint\022\027\n\017host_memory_arg\030\004 \003(\t\022\r\n\005label\030" +
+      "\005 \001(\t\022\020\n\010priority\030\006 \001(\005\032M\n\016AttrConstrain" +
+      "t\022\014\n\004name\030\001 \001(\t\022-\n\016allowed_values\030\002 \001(\0132" +
+      "\025.tensorflow.AttrValue\"3\n\nKernelList\022%\n\006" +
+      "kernel\030\001 \003(\0132\025.tensorflow.KernelDefB\211\001\n\036" +
+      "org.tensorflow.proto.frameworkB\017KernelDe" +
+      "fProtosP\001ZQgithub.com/tensorflow/tensorf" +
+      "low/tensorflow/go/core/framework/kernel_" +
+      "def_go_proto\370\001\001b\006proto3"
+    };
+    descriptor = com.google.protobuf.Descriptors.FileDescriptor
+      .internalBuildGeneratedFileFrom(descriptorData,
+        new com.google.protobuf.Descriptors.FileDescriptor[] {
+          org.tensorflow.proto.framework.AttrValueProtos.getDescriptor(),
+        });
+    internal_static_tensorflow_KernelDef_descriptor =
+      getDescriptor().getMessageTypes().get(0);
+    internal_static_tensorflow_KernelDef_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_KernelDef_descriptor,
+        new java.lang.String[] { "Op", "DeviceType", "Constraint", "HostMemoryArg", "Label", "Priority", });
+    internal_static_tensorflow_KernelDef_AttrConstraint_descriptor =
+      internal_static_tensorflow_KernelDef_descriptor.getNestedTypes().get(0);
+    internal_static_tensorflow_KernelDef_AttrConstraint_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_KernelDef_AttrConstraint_descriptor,
+        new java.lang.String[] { "Name", "AllowedValues", });
+    internal_static_tensorflow_KernelList_descriptor =
+      getDescriptor().getMessageTypes().get(1);
+    internal_static_tensorflow_KernelList_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_KernelList_descriptor,
+        new java.lang.String[] { "Kernel", });
+    org.tensorflow.proto.framework.AttrValueProtos.getDescriptor();
+  }
+
+  // @@protoc_insertion_point(outer_class_scope)
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/KernelList.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/KernelList.java
new file mode 100644
index 00000000000..06c6e5a22f2
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/KernelList.java
@@ -0,0 +1,773 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/framework/kernel_def.proto
+
+package org.tensorflow.proto.framework;
+
+/**
+ * <pre>
+ * A collection of KernelDefs
+ * </pre>
+ *
+ * Protobuf type {@code tensorflow.KernelList}
+ */
+public  final class KernelList extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.KernelList)
+    KernelListOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use KernelList.newBuilder() to construct.
+  private KernelList(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private KernelList() {
+    kernel_ = java.util.Collections.emptyList();
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new KernelList();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private KernelList(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    int mutable_bitField0_ = 0;
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 10: {
+            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
+              kernel_ = new java.util.ArrayList<org.tensorflow.proto.framework.KernelDef>();
+              mutable_bitField0_ |= 0x00000001;
+            }
+            kernel_.add(
+                input.readMessage(org.tensorflow.proto.framework.KernelDef.parser(), extensionRegistry));
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      if (((mutable_bitField0_ & 0x00000001) != 0)) {
+        kernel_ = java.util.Collections.unmodifiableList(kernel_);
+      }
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.framework.KernelDefProtos.internal_static_tensorflow_KernelList_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.framework.KernelDefProtos.internal_static_tensorflow_KernelList_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.framework.KernelList.class, org.tensorflow.proto.framework.KernelList.Builder.class);
+  }
+
+  public static final int KERNEL_FIELD_NUMBER = 1;
+  private java.util.List<org.tensorflow.proto.framework.KernelDef> kernel_;
+  /**
+   * <code>repeated .tensorflow.KernelDef kernel = 1;</code>
+   */
+  public java.util.List<org.tensorflow.proto.framework.KernelDef> getKernelList() {
+    return kernel_;
+  }
+  /**
+   * <code>repeated .tensorflow.KernelDef kernel = 1;</code>
+   */
+  public java.util.List<? extends org.tensorflow.proto.framework.KernelDefOrBuilder> 
+      getKernelOrBuilderList() {
+    return kernel_;
+  }
+  /**
+   * <code>repeated .tensorflow.KernelDef kernel = 1;</code>
+   */
+  public int getKernelCount() {
+    return kernel_.size();
+  }
+  /**
+   * <code>repeated .tensorflow.KernelDef kernel = 1;</code>
+   */
+  public org.tensorflow.proto.framework.KernelDef getKernel(int index) {
+    return kernel_.get(index);
+  }
+  /**
+   * <code>repeated .tensorflow.KernelDef kernel = 1;</code>
+   */
+  public org.tensorflow.proto.framework.KernelDefOrBuilder getKernelOrBuilder(
+      int index) {
+    return kernel_.get(index);
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    for (int i = 0; i < kernel_.size(); i++) {
+      output.writeMessage(1, kernel_.get(i));
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    for (int i = 0; i < kernel_.size(); i++) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(1, kernel_.get(i));
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.framework.KernelList)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.framework.KernelList other = (org.tensorflow.proto.framework.KernelList) obj;
+
+    if (!getKernelList()
+        .equals(other.getKernelList())) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    if (getKernelCount() > 0) {
+      hash = (37 * hash) + KERNEL_FIELD_NUMBER;
+      hash = (53 * hash) + getKernelList().hashCode();
+    }
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.framework.KernelList parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.KernelList parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.KernelList parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.KernelList parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.KernelList parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.KernelList parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.KernelList parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.KernelList parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.KernelList parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.KernelList parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.KernelList parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.KernelList parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.framework.KernelList prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * <pre>
+   * A collection of KernelDefs
+   * </pre>
+   *
+   * Protobuf type {@code tensorflow.KernelList}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.KernelList)
+      org.tensorflow.proto.framework.KernelListOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.KernelDefProtos.internal_static_tensorflow_KernelList_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.KernelDefProtos.internal_static_tensorflow_KernelList_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.KernelList.class, org.tensorflow.proto.framework.KernelList.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.framework.KernelList.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+        getKernelFieldBuilder();
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      if (kernelBuilder_ == null) {
+        kernel_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000001);
+      } else {
+        kernelBuilder_.clear();
+      }
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.framework.KernelDefProtos.internal_static_tensorflow_KernelList_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.KernelList getDefaultInstanceForType() {
+      return org.tensorflow.proto.framework.KernelList.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.KernelList build() {
+      org.tensorflow.proto.framework.KernelList result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.KernelList buildPartial() {
+      org.tensorflow.proto.framework.KernelList result = new org.tensorflow.proto.framework.KernelList(this);
+      int from_bitField0_ = bitField0_;
+      if (kernelBuilder_ == null) {
+        if (((bitField0_ & 0x00000001) != 0)) {
+          kernel_ = java.util.Collections.unmodifiableList(kernel_);
+          bitField0_ = (bitField0_ & ~0x00000001);
+        }
+        result.kernel_ = kernel_;
+      } else {
+        result.kernel_ = kernelBuilder_.build();
+      }
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.framework.KernelList) {
+        return mergeFrom((org.tensorflow.proto.framework.KernelList)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.framework.KernelList other) {
+      if (other == org.tensorflow.proto.framework.KernelList.getDefaultInstance()) return this;
+      if (kernelBuilder_ == null) {
+        if (!other.kernel_.isEmpty()) {
+          if (kernel_.isEmpty()) {
+            kernel_ = other.kernel_;
+            bitField0_ = (bitField0_ & ~0x00000001);
+          } else {
+            ensureKernelIsMutable();
+            kernel_.addAll(other.kernel_);
+          }
+          onChanged();
+        }
+      } else {
+        if (!other.kernel_.isEmpty()) {
+          if (kernelBuilder_.isEmpty()) {
+            kernelBuilder_.dispose();
+            kernelBuilder_ = null;
+            kernel_ = other.kernel_;
+            bitField0_ = (bitField0_ & ~0x00000001);
+            kernelBuilder_ = 
+              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
+                 getKernelFieldBuilder() : null;
+          } else {
+            kernelBuilder_.addAllMessages(other.kernel_);
+          }
+        }
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.framework.KernelList parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.framework.KernelList) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+    private int bitField0_;
+
+    private java.util.List<org.tensorflow.proto.framework.KernelDef> kernel_ =
+      java.util.Collections.emptyList();
+    private void ensureKernelIsMutable() {
+      if (!((bitField0_ & 0x00000001) != 0)) {
+        kernel_ = new java.util.ArrayList<org.tensorflow.proto.framework.KernelDef>(kernel_);
+        bitField0_ |= 0x00000001;
+       }
+    }
+
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.framework.KernelDef, org.tensorflow.proto.framework.KernelDef.Builder, org.tensorflow.proto.framework.KernelDefOrBuilder> kernelBuilder_;
+
+    /**
+     * <code>repeated .tensorflow.KernelDef kernel = 1;</code>
+     */
+    public java.util.List<org.tensorflow.proto.framework.KernelDef> getKernelList() {
+      if (kernelBuilder_ == null) {
+        return java.util.Collections.unmodifiableList(kernel_);
+      } else {
+        return kernelBuilder_.getMessageList();
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.KernelDef kernel = 1;</code>
+     */
+    public int getKernelCount() {
+      if (kernelBuilder_ == null) {
+        return kernel_.size();
+      } else {
+        return kernelBuilder_.getCount();
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.KernelDef kernel = 1;</code>
+     */
+    public org.tensorflow.proto.framework.KernelDef getKernel(int index) {
+      if (kernelBuilder_ == null) {
+        return kernel_.get(index);
+      } else {
+        return kernelBuilder_.getMessage(index);
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.KernelDef kernel = 1;</code>
+     */
+    public Builder setKernel(
+        int index, org.tensorflow.proto.framework.KernelDef value) {
+      if (kernelBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureKernelIsMutable();
+        kernel_.set(index, value);
+        onChanged();
+      } else {
+        kernelBuilder_.setMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.KernelDef kernel = 1;</code>
+     */
+    public Builder setKernel(
+        int index, org.tensorflow.proto.framework.KernelDef.Builder builderForValue) {
+      if (kernelBuilder_ == null) {
+        ensureKernelIsMutable();
+        kernel_.set(index, builderForValue.build());
+        onChanged();
+      } else {
+        kernelBuilder_.setMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.KernelDef kernel = 1;</code>
+     */
+    public Builder addKernel(org.tensorflow.proto.framework.KernelDef value) {
+      if (kernelBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureKernelIsMutable();
+        kernel_.add(value);
+        onChanged();
+      } else {
+        kernelBuilder_.addMessage(value);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.KernelDef kernel = 1;</code>
+     */
+    public Builder addKernel(
+        int index, org.tensorflow.proto.framework.KernelDef value) {
+      if (kernelBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureKernelIsMutable();
+        kernel_.add(index, value);
+        onChanged();
+      } else {
+        kernelBuilder_.addMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.KernelDef kernel = 1;</code>
+     */
+    public Builder addKernel(
+        org.tensorflow.proto.framework.KernelDef.Builder builderForValue) {
+      if (kernelBuilder_ == null) {
+        ensureKernelIsMutable();
+        kernel_.add(builderForValue.build());
+        onChanged();
+      } else {
+        kernelBuilder_.addMessage(builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.KernelDef kernel = 1;</code>
+     */
+    public Builder addKernel(
+        int index, org.tensorflow.proto.framework.KernelDef.Builder builderForValue) {
+      if (kernelBuilder_ == null) {
+        ensureKernelIsMutable();
+        kernel_.add(index, builderForValue.build());
+        onChanged();
+      } else {
+        kernelBuilder_.addMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.KernelDef kernel = 1;</code>
+     */
+    public Builder addAllKernel(
+        java.lang.Iterable<? extends org.tensorflow.proto.framework.KernelDef> values) {
+      if (kernelBuilder_ == null) {
+        ensureKernelIsMutable();
+        com.google.protobuf.AbstractMessageLite.Builder.addAll(
+            values, kernel_);
+        onChanged();
+      } else {
+        kernelBuilder_.addAllMessages(values);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.KernelDef kernel = 1;</code>
+     */
+    public Builder clearKernel() {
+      if (kernelBuilder_ == null) {
+        kernel_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000001);
+        onChanged();
+      } else {
+        kernelBuilder_.clear();
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.KernelDef kernel = 1;</code>
+     */
+    public Builder removeKernel(int index) {
+      if (kernelBuilder_ == null) {
+        ensureKernelIsMutable();
+        kernel_.remove(index);
+        onChanged();
+      } else {
+        kernelBuilder_.remove(index);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.KernelDef kernel = 1;</code>
+     */
+    public org.tensorflow.proto.framework.KernelDef.Builder getKernelBuilder(
+        int index) {
+      return getKernelFieldBuilder().getBuilder(index);
+    }
+    /**
+     * <code>repeated .tensorflow.KernelDef kernel = 1;</code>
+     */
+    public org.tensorflow.proto.framework.KernelDefOrBuilder getKernelOrBuilder(
+        int index) {
+      if (kernelBuilder_ == null) {
+        return kernel_.get(index);  } else {
+        return kernelBuilder_.getMessageOrBuilder(index);
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.KernelDef kernel = 1;</code>
+     */
+    public java.util.List<? extends org.tensorflow.proto.framework.KernelDefOrBuilder> 
+         getKernelOrBuilderList() {
+      if (kernelBuilder_ != null) {
+        return kernelBuilder_.getMessageOrBuilderList();
+      } else {
+        return java.util.Collections.unmodifiableList(kernel_);
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.KernelDef kernel = 1;</code>
+     */
+    public org.tensorflow.proto.framework.KernelDef.Builder addKernelBuilder() {
+      return getKernelFieldBuilder().addBuilder(
+          org.tensorflow.proto.framework.KernelDef.getDefaultInstance());
+    }
+    /**
+     * <code>repeated .tensorflow.KernelDef kernel = 1;</code>
+     */
+    public org.tensorflow.proto.framework.KernelDef.Builder addKernelBuilder(
+        int index) {
+      return getKernelFieldBuilder().addBuilder(
+          index, org.tensorflow.proto.framework.KernelDef.getDefaultInstance());
+    }
+    /**
+     * <code>repeated .tensorflow.KernelDef kernel = 1;</code>
+     */
+    public java.util.List<org.tensorflow.proto.framework.KernelDef.Builder> 
+         getKernelBuilderList() {
+      return getKernelFieldBuilder().getBuilderList();
+    }
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.framework.KernelDef, org.tensorflow.proto.framework.KernelDef.Builder, org.tensorflow.proto.framework.KernelDefOrBuilder> 
+        getKernelFieldBuilder() {
+      if (kernelBuilder_ == null) {
+        kernelBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
+            org.tensorflow.proto.framework.KernelDef, org.tensorflow.proto.framework.KernelDef.Builder, org.tensorflow.proto.framework.KernelDefOrBuilder>(
+                kernel_,
+                ((bitField0_ & 0x00000001) != 0),
+                getParentForChildren(),
+                isClean());
+        kernel_ = null;
+      }
+      return kernelBuilder_;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.KernelList)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.KernelList)
+  private static final org.tensorflow.proto.framework.KernelList DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.KernelList();
+  }
+
+  public static org.tensorflow.proto.framework.KernelList getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<KernelList>
+      PARSER = new com.google.protobuf.AbstractParser<KernelList>() {
+    @java.lang.Override
+    public KernelList parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new KernelList(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<KernelList> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<KernelList> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.framework.KernelList getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/KernelListOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/KernelListOrBuilder.java
new file mode 100644
index 00000000000..31c95b5de9d
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/KernelListOrBuilder.java
@@ -0,0 +1,33 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/framework/kernel_def.proto
+
+package org.tensorflow.proto.framework;
+
+public interface KernelListOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.KernelList)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <code>repeated .tensorflow.KernelDef kernel = 1;</code>
+   */
+  java.util.List<org.tensorflow.proto.framework.KernelDef> 
+      getKernelList();
+  /**
+   * <code>repeated .tensorflow.KernelDef kernel = 1;</code>
+   */
+  org.tensorflow.proto.framework.KernelDef getKernel(int index);
+  /**
+   * <code>repeated .tensorflow.KernelDef kernel = 1;</code>
+   */
+  int getKernelCount();
+  /**
+   * <code>repeated .tensorflow.KernelDef kernel = 1;</code>
+   */
+  java.util.List<? extends org.tensorflow.proto.framework.KernelDefOrBuilder> 
+      getKernelOrBuilderList();
+  /**
+   * <code>repeated .tensorflow.KernelDef kernel = 1;</code>
+   */
+  org.tensorflow.proto.framework.KernelDefOrBuilder getKernelOrBuilder(
+      int index);
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ListValue.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ListValue.java
new file mode 100644
index 00000000000..56101ea237c
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ListValue.java
@@ -0,0 +1,773 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/struct.proto
+
+package org.tensorflow.proto.framework;
+
+/**
+ * <pre>
+ * Represents a Python list.
+ * </pre>
+ *
+ * Protobuf type {@code tensorflow.ListValue}
+ */
+public  final class ListValue extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.ListValue)
+    ListValueOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use ListValue.newBuilder() to construct.
+  private ListValue(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private ListValue() {
+    values_ = java.util.Collections.emptyList();
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new ListValue();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private ListValue(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    int mutable_bitField0_ = 0;
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 10: {
+            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
+              values_ = new java.util.ArrayList<org.tensorflow.proto.framework.StructuredValue>();
+              mutable_bitField0_ |= 0x00000001;
+            }
+            values_.add(
+                input.readMessage(org.tensorflow.proto.framework.StructuredValue.parser(), extensionRegistry));
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      if (((mutable_bitField0_ & 0x00000001) != 0)) {
+        values_ = java.util.Collections.unmodifiableList(values_);
+      }
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.framework.StructProtos.internal_static_tensorflow_ListValue_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.framework.StructProtos.internal_static_tensorflow_ListValue_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.framework.ListValue.class, org.tensorflow.proto.framework.ListValue.Builder.class);
+  }
+
+  public static final int VALUES_FIELD_NUMBER = 1;
+  private java.util.List<org.tensorflow.proto.framework.StructuredValue> values_;
+  /**
+   * <code>repeated .tensorflow.StructuredValue values = 1;</code>
+   */
+  public java.util.List<org.tensorflow.proto.framework.StructuredValue> getValuesList() {
+    return values_;
+  }
+  /**
+   * <code>repeated .tensorflow.StructuredValue values = 1;</code>
+   */
+  public java.util.List<? extends org.tensorflow.proto.framework.StructuredValueOrBuilder> 
+      getValuesOrBuilderList() {
+    return values_;
+  }
+  /**
+   * <code>repeated .tensorflow.StructuredValue values = 1;</code>
+   */
+  public int getValuesCount() {
+    return values_.size();
+  }
+  /**
+   * <code>repeated .tensorflow.StructuredValue values = 1;</code>
+   */
+  public org.tensorflow.proto.framework.StructuredValue getValues(int index) {
+    return values_.get(index);
+  }
+  /**
+   * <code>repeated .tensorflow.StructuredValue values = 1;</code>
+   */
+  public org.tensorflow.proto.framework.StructuredValueOrBuilder getValuesOrBuilder(
+      int index) {
+    return values_.get(index);
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    for (int i = 0; i < values_.size(); i++) {
+      output.writeMessage(1, values_.get(i));
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    for (int i = 0; i < values_.size(); i++) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(1, values_.get(i));
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.framework.ListValue)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.framework.ListValue other = (org.tensorflow.proto.framework.ListValue) obj;
+
+    if (!getValuesList()
+        .equals(other.getValuesList())) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    if (getValuesCount() > 0) {
+      hash = (37 * hash) + VALUES_FIELD_NUMBER;
+      hash = (53 * hash) + getValuesList().hashCode();
+    }
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.framework.ListValue parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.ListValue parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.ListValue parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.ListValue parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.ListValue parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.ListValue parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.ListValue parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.ListValue parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.ListValue parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.ListValue parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.ListValue parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.ListValue parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.framework.ListValue prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * <pre>
+   * Represents a Python list.
+   * </pre>
+   *
+   * Protobuf type {@code tensorflow.ListValue}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.ListValue)
+      org.tensorflow.proto.framework.ListValueOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.StructProtos.internal_static_tensorflow_ListValue_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.StructProtos.internal_static_tensorflow_ListValue_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.ListValue.class, org.tensorflow.proto.framework.ListValue.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.framework.ListValue.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+        getValuesFieldBuilder();
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      if (valuesBuilder_ == null) {
+        values_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000001);
+      } else {
+        valuesBuilder_.clear();
+      }
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.framework.StructProtos.internal_static_tensorflow_ListValue_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.ListValue getDefaultInstanceForType() {
+      return org.tensorflow.proto.framework.ListValue.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.ListValue build() {
+      org.tensorflow.proto.framework.ListValue result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.ListValue buildPartial() {
+      org.tensorflow.proto.framework.ListValue result = new org.tensorflow.proto.framework.ListValue(this);
+      int from_bitField0_ = bitField0_;
+      if (valuesBuilder_ == null) {
+        if (((bitField0_ & 0x00000001) != 0)) {
+          values_ = java.util.Collections.unmodifiableList(values_);
+          bitField0_ = (bitField0_ & ~0x00000001);
+        }
+        result.values_ = values_;
+      } else {
+        result.values_ = valuesBuilder_.build();
+      }
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.framework.ListValue) {
+        return mergeFrom((org.tensorflow.proto.framework.ListValue)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.framework.ListValue other) {
+      if (other == org.tensorflow.proto.framework.ListValue.getDefaultInstance()) return this;
+      if (valuesBuilder_ == null) {
+        if (!other.values_.isEmpty()) {
+          if (values_.isEmpty()) {
+            values_ = other.values_;
+            bitField0_ = (bitField0_ & ~0x00000001);
+          } else {
+            ensureValuesIsMutable();
+            values_.addAll(other.values_);
+          }
+          onChanged();
+        }
+      } else {
+        if (!other.values_.isEmpty()) {
+          if (valuesBuilder_.isEmpty()) {
+            valuesBuilder_.dispose();
+            valuesBuilder_ = null;
+            values_ = other.values_;
+            bitField0_ = (bitField0_ & ~0x00000001);
+            valuesBuilder_ = 
+              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
+                 getValuesFieldBuilder() : null;
+          } else {
+            valuesBuilder_.addAllMessages(other.values_);
+          }
+        }
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.framework.ListValue parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.framework.ListValue) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+    private int bitField0_;
+
+    private java.util.List<org.tensorflow.proto.framework.StructuredValue> values_ =
+      java.util.Collections.emptyList();
+    private void ensureValuesIsMutable() {
+      if (!((bitField0_ & 0x00000001) != 0)) {
+        values_ = new java.util.ArrayList<org.tensorflow.proto.framework.StructuredValue>(values_);
+        bitField0_ |= 0x00000001;
+       }
+    }
+
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.framework.StructuredValue, org.tensorflow.proto.framework.StructuredValue.Builder, org.tensorflow.proto.framework.StructuredValueOrBuilder> valuesBuilder_;
+
+    /**
+     * <code>repeated .tensorflow.StructuredValue values = 1;</code>
+     */
+    public java.util.List<org.tensorflow.proto.framework.StructuredValue> getValuesList() {
+      if (valuesBuilder_ == null) {
+        return java.util.Collections.unmodifiableList(values_);
+      } else {
+        return valuesBuilder_.getMessageList();
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.StructuredValue values = 1;</code>
+     */
+    public int getValuesCount() {
+      if (valuesBuilder_ == null) {
+        return values_.size();
+      } else {
+        return valuesBuilder_.getCount();
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.StructuredValue values = 1;</code>
+     */
+    public org.tensorflow.proto.framework.StructuredValue getValues(int index) {
+      if (valuesBuilder_ == null) {
+        return values_.get(index);
+      } else {
+        return valuesBuilder_.getMessage(index);
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.StructuredValue values = 1;</code>
+     */
+    public Builder setValues(
+        int index, org.tensorflow.proto.framework.StructuredValue value) {
+      if (valuesBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureValuesIsMutable();
+        values_.set(index, value);
+        onChanged();
+      } else {
+        valuesBuilder_.setMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.StructuredValue values = 1;</code>
+     */
+    public Builder setValues(
+        int index, org.tensorflow.proto.framework.StructuredValue.Builder builderForValue) {
+      if (valuesBuilder_ == null) {
+        ensureValuesIsMutable();
+        values_.set(index, builderForValue.build());
+        onChanged();
+      } else {
+        valuesBuilder_.setMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.StructuredValue values = 1;</code>
+     */
+    public Builder addValues(org.tensorflow.proto.framework.StructuredValue value) {
+      if (valuesBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureValuesIsMutable();
+        values_.add(value);
+        onChanged();
+      } else {
+        valuesBuilder_.addMessage(value);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.StructuredValue values = 1;</code>
+     */
+    public Builder addValues(
+        int index, org.tensorflow.proto.framework.StructuredValue value) {
+      if (valuesBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureValuesIsMutable();
+        values_.add(index, value);
+        onChanged();
+      } else {
+        valuesBuilder_.addMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.StructuredValue values = 1;</code>
+     */
+    public Builder addValues(
+        org.tensorflow.proto.framework.StructuredValue.Builder builderForValue) {
+      if (valuesBuilder_ == null) {
+        ensureValuesIsMutable();
+        values_.add(builderForValue.build());
+        onChanged();
+      } else {
+        valuesBuilder_.addMessage(builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.StructuredValue values = 1;</code>
+     */
+    public Builder addValues(
+        int index, org.tensorflow.proto.framework.StructuredValue.Builder builderForValue) {
+      if (valuesBuilder_ == null) {
+        ensureValuesIsMutable();
+        values_.add(index, builderForValue.build());
+        onChanged();
+      } else {
+        valuesBuilder_.addMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.StructuredValue values = 1;</code>
+     */
+    public Builder addAllValues(
+        java.lang.Iterable<? extends org.tensorflow.proto.framework.StructuredValue> values) {
+      if (valuesBuilder_ == null) {
+        ensureValuesIsMutable();
+        com.google.protobuf.AbstractMessageLite.Builder.addAll(
+            values, values_);
+        onChanged();
+      } else {
+        valuesBuilder_.addAllMessages(values);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.StructuredValue values = 1;</code>
+     */
+    public Builder clearValues() {
+      if (valuesBuilder_ == null) {
+        values_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000001);
+        onChanged();
+      } else {
+        valuesBuilder_.clear();
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.StructuredValue values = 1;</code>
+     */
+    public Builder removeValues(int index) {
+      if (valuesBuilder_ == null) {
+        ensureValuesIsMutable();
+        values_.remove(index);
+        onChanged();
+      } else {
+        valuesBuilder_.remove(index);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.StructuredValue values = 1;</code>
+     */
+    public org.tensorflow.proto.framework.StructuredValue.Builder getValuesBuilder(
+        int index) {
+      return getValuesFieldBuilder().getBuilder(index);
+    }
+    /**
+     * <code>repeated .tensorflow.StructuredValue values = 1;</code>
+     */
+    public org.tensorflow.proto.framework.StructuredValueOrBuilder getValuesOrBuilder(
+        int index) {
+      if (valuesBuilder_ == null) {
+        return values_.get(index);  } else {
+        return valuesBuilder_.getMessageOrBuilder(index);
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.StructuredValue values = 1;</code>
+     */
+    public java.util.List<? extends org.tensorflow.proto.framework.StructuredValueOrBuilder> 
+         getValuesOrBuilderList() {
+      if (valuesBuilder_ != null) {
+        return valuesBuilder_.getMessageOrBuilderList();
+      } else {
+        return java.util.Collections.unmodifiableList(values_);
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.StructuredValue values = 1;</code>
+     */
+    public org.tensorflow.proto.framework.StructuredValue.Builder addValuesBuilder() {
+      return getValuesFieldBuilder().addBuilder(
+          org.tensorflow.proto.framework.StructuredValue.getDefaultInstance());
+    }
+    /**
+     * <code>repeated .tensorflow.StructuredValue values = 1;</code>
+     */
+    public org.tensorflow.proto.framework.StructuredValue.Builder addValuesBuilder(
+        int index) {
+      return getValuesFieldBuilder().addBuilder(
+          index, org.tensorflow.proto.framework.StructuredValue.getDefaultInstance());
+    }
+    /**
+     * <code>repeated .tensorflow.StructuredValue values = 1;</code>
+     */
+    public java.util.List<org.tensorflow.proto.framework.StructuredValue.Builder> 
+         getValuesBuilderList() {
+      return getValuesFieldBuilder().getBuilderList();
+    }
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.framework.StructuredValue, org.tensorflow.proto.framework.StructuredValue.Builder, org.tensorflow.proto.framework.StructuredValueOrBuilder> 
+        getValuesFieldBuilder() {
+      if (valuesBuilder_ == null) {
+        valuesBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
+            org.tensorflow.proto.framework.StructuredValue, org.tensorflow.proto.framework.StructuredValue.Builder, org.tensorflow.proto.framework.StructuredValueOrBuilder>(
+                values_,
+                ((bitField0_ & 0x00000001) != 0),
+                getParentForChildren(),
+                isClean());
+        values_ = null;
+      }
+      return valuesBuilder_;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.ListValue)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.ListValue)
+  private static final org.tensorflow.proto.framework.ListValue DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.ListValue();
+  }
+
+  public static org.tensorflow.proto.framework.ListValue getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<ListValue>
+      PARSER = new com.google.protobuf.AbstractParser<ListValue>() {
+    @java.lang.Override
+    public ListValue parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new ListValue(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<ListValue> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<ListValue> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.framework.ListValue getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ListValueOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ListValueOrBuilder.java
new file mode 100644
index 00000000000..e4b2d7076d9
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ListValueOrBuilder.java
@@ -0,0 +1,33 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/struct.proto
+
+package org.tensorflow.proto.framework;
+
+public interface ListValueOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.ListValue)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <code>repeated .tensorflow.StructuredValue values = 1;</code>
+   */
+  java.util.List<org.tensorflow.proto.framework.StructuredValue> 
+      getValuesList();
+  /**
+   * <code>repeated .tensorflow.StructuredValue values = 1;</code>
+   */
+  org.tensorflow.proto.framework.StructuredValue getValues(int index);
+  /**
+   * <code>repeated .tensorflow.StructuredValue values = 1;</code>
+   */
+  int getValuesCount();
+  /**
+   * <code>repeated .tensorflow.StructuredValue values = 1;</code>
+   */
+  java.util.List<? extends org.tensorflow.proto.framework.StructuredValueOrBuilder> 
+      getValuesOrBuilderList();
+  /**
+   * <code>repeated .tensorflow.StructuredValue values = 1;</code>
+   */
+  org.tensorflow.proto.framework.StructuredValueOrBuilder getValuesOrBuilder(
+      int index);
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/LocalLinks.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/LocalLinks.java
new file mode 100644
index 00000000000..2219a69600d
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/LocalLinks.java
@@ -0,0 +1,765 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/framework/device_attributes.proto
+
+package org.tensorflow.proto.framework;
+
+/**
+ * Protobuf type {@code tensorflow.LocalLinks}
+ */
+public  final class LocalLinks extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.LocalLinks)
+    LocalLinksOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use LocalLinks.newBuilder() to construct.
+  private LocalLinks(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private LocalLinks() {
+    link_ = java.util.Collections.emptyList();
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new LocalLinks();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private LocalLinks(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    int mutable_bitField0_ = 0;
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 10: {
+            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
+              link_ = new java.util.ArrayList<org.tensorflow.proto.framework.InterconnectLink>();
+              mutable_bitField0_ |= 0x00000001;
+            }
+            link_.add(
+                input.readMessage(org.tensorflow.proto.framework.InterconnectLink.parser(), extensionRegistry));
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      if (((mutable_bitField0_ & 0x00000001) != 0)) {
+        link_ = java.util.Collections.unmodifiableList(link_);
+      }
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.framework.DeviceAttributesProtos.internal_static_tensorflow_LocalLinks_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.framework.DeviceAttributesProtos.internal_static_tensorflow_LocalLinks_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.framework.LocalLinks.class, org.tensorflow.proto.framework.LocalLinks.Builder.class);
+  }
+
+  public static final int LINK_FIELD_NUMBER = 1;
+  private java.util.List<org.tensorflow.proto.framework.InterconnectLink> link_;
+  /**
+   * <code>repeated .tensorflow.InterconnectLink link = 1;</code>
+   */
+  public java.util.List<org.tensorflow.proto.framework.InterconnectLink> getLinkList() {
+    return link_;
+  }
+  /**
+   * <code>repeated .tensorflow.InterconnectLink link = 1;</code>
+   */
+  public java.util.List<? extends org.tensorflow.proto.framework.InterconnectLinkOrBuilder> 
+      getLinkOrBuilderList() {
+    return link_;
+  }
+  /**
+   * <code>repeated .tensorflow.InterconnectLink link = 1;</code>
+   */
+  public int getLinkCount() {
+    return link_.size();
+  }
+  /**
+   * <code>repeated .tensorflow.InterconnectLink link = 1;</code>
+   */
+  public org.tensorflow.proto.framework.InterconnectLink getLink(int index) {
+    return link_.get(index);
+  }
+  /**
+   * <code>repeated .tensorflow.InterconnectLink link = 1;</code>
+   */
+  public org.tensorflow.proto.framework.InterconnectLinkOrBuilder getLinkOrBuilder(
+      int index) {
+    return link_.get(index);
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    for (int i = 0; i < link_.size(); i++) {
+      output.writeMessage(1, link_.get(i));
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    for (int i = 0; i < link_.size(); i++) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(1, link_.get(i));
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.framework.LocalLinks)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.framework.LocalLinks other = (org.tensorflow.proto.framework.LocalLinks) obj;
+
+    if (!getLinkList()
+        .equals(other.getLinkList())) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    if (getLinkCount() > 0) {
+      hash = (37 * hash) + LINK_FIELD_NUMBER;
+      hash = (53 * hash) + getLinkList().hashCode();
+    }
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.framework.LocalLinks parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.LocalLinks parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.LocalLinks parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.LocalLinks parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.LocalLinks parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.LocalLinks parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.LocalLinks parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.LocalLinks parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.LocalLinks parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.LocalLinks parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.LocalLinks parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.LocalLinks parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.framework.LocalLinks prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * Protobuf type {@code tensorflow.LocalLinks}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.LocalLinks)
+      org.tensorflow.proto.framework.LocalLinksOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.DeviceAttributesProtos.internal_static_tensorflow_LocalLinks_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.DeviceAttributesProtos.internal_static_tensorflow_LocalLinks_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.LocalLinks.class, org.tensorflow.proto.framework.LocalLinks.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.framework.LocalLinks.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+        getLinkFieldBuilder();
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      if (linkBuilder_ == null) {
+        link_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000001);
+      } else {
+        linkBuilder_.clear();
+      }
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.framework.DeviceAttributesProtos.internal_static_tensorflow_LocalLinks_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.LocalLinks getDefaultInstanceForType() {
+      return org.tensorflow.proto.framework.LocalLinks.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.LocalLinks build() {
+      org.tensorflow.proto.framework.LocalLinks result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.LocalLinks buildPartial() {
+      org.tensorflow.proto.framework.LocalLinks result = new org.tensorflow.proto.framework.LocalLinks(this);
+      int from_bitField0_ = bitField0_;
+      if (linkBuilder_ == null) {
+        if (((bitField0_ & 0x00000001) != 0)) {
+          link_ = java.util.Collections.unmodifiableList(link_);
+          bitField0_ = (bitField0_ & ~0x00000001);
+        }
+        result.link_ = link_;
+      } else {
+        result.link_ = linkBuilder_.build();
+      }
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.framework.LocalLinks) {
+        return mergeFrom((org.tensorflow.proto.framework.LocalLinks)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.framework.LocalLinks other) {
+      if (other == org.tensorflow.proto.framework.LocalLinks.getDefaultInstance()) return this;
+      if (linkBuilder_ == null) {
+        if (!other.link_.isEmpty()) {
+          if (link_.isEmpty()) {
+            link_ = other.link_;
+            bitField0_ = (bitField0_ & ~0x00000001);
+          } else {
+            ensureLinkIsMutable();
+            link_.addAll(other.link_);
+          }
+          onChanged();
+        }
+      } else {
+        if (!other.link_.isEmpty()) {
+          if (linkBuilder_.isEmpty()) {
+            linkBuilder_.dispose();
+            linkBuilder_ = null;
+            link_ = other.link_;
+            bitField0_ = (bitField0_ & ~0x00000001);
+            linkBuilder_ = 
+              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
+                 getLinkFieldBuilder() : null;
+          } else {
+            linkBuilder_.addAllMessages(other.link_);
+          }
+        }
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.framework.LocalLinks parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.framework.LocalLinks) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+    private int bitField0_;
+
+    private java.util.List<org.tensorflow.proto.framework.InterconnectLink> link_ =
+      java.util.Collections.emptyList();
+    private void ensureLinkIsMutable() {
+      if (!((bitField0_ & 0x00000001) != 0)) {
+        link_ = new java.util.ArrayList<org.tensorflow.proto.framework.InterconnectLink>(link_);
+        bitField0_ |= 0x00000001;
+       }
+    }
+
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.framework.InterconnectLink, org.tensorflow.proto.framework.InterconnectLink.Builder, org.tensorflow.proto.framework.InterconnectLinkOrBuilder> linkBuilder_;
+
+    /**
+     * <code>repeated .tensorflow.InterconnectLink link = 1;</code>
+     */
+    public java.util.List<org.tensorflow.proto.framework.InterconnectLink> getLinkList() {
+      if (linkBuilder_ == null) {
+        return java.util.Collections.unmodifiableList(link_);
+      } else {
+        return linkBuilder_.getMessageList();
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.InterconnectLink link = 1;</code>
+     */
+    public int getLinkCount() {
+      if (linkBuilder_ == null) {
+        return link_.size();
+      } else {
+        return linkBuilder_.getCount();
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.InterconnectLink link = 1;</code>
+     */
+    public org.tensorflow.proto.framework.InterconnectLink getLink(int index) {
+      if (linkBuilder_ == null) {
+        return link_.get(index);
+      } else {
+        return linkBuilder_.getMessage(index);
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.InterconnectLink link = 1;</code>
+     */
+    public Builder setLink(
+        int index, org.tensorflow.proto.framework.InterconnectLink value) {
+      if (linkBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureLinkIsMutable();
+        link_.set(index, value);
+        onChanged();
+      } else {
+        linkBuilder_.setMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.InterconnectLink link = 1;</code>
+     */
+    public Builder setLink(
+        int index, org.tensorflow.proto.framework.InterconnectLink.Builder builderForValue) {
+      if (linkBuilder_ == null) {
+        ensureLinkIsMutable();
+        link_.set(index, builderForValue.build());
+        onChanged();
+      } else {
+        linkBuilder_.setMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.InterconnectLink link = 1;</code>
+     */
+    public Builder addLink(org.tensorflow.proto.framework.InterconnectLink value) {
+      if (linkBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureLinkIsMutable();
+        link_.add(value);
+        onChanged();
+      } else {
+        linkBuilder_.addMessage(value);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.InterconnectLink link = 1;</code>
+     */
+    public Builder addLink(
+        int index, org.tensorflow.proto.framework.InterconnectLink value) {
+      if (linkBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureLinkIsMutable();
+        link_.add(index, value);
+        onChanged();
+      } else {
+        linkBuilder_.addMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.InterconnectLink link = 1;</code>
+     */
+    public Builder addLink(
+        org.tensorflow.proto.framework.InterconnectLink.Builder builderForValue) {
+      if (linkBuilder_ == null) {
+        ensureLinkIsMutable();
+        link_.add(builderForValue.build());
+        onChanged();
+      } else {
+        linkBuilder_.addMessage(builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.InterconnectLink link = 1;</code>
+     */
+    public Builder addLink(
+        int index, org.tensorflow.proto.framework.InterconnectLink.Builder builderForValue) {
+      if (linkBuilder_ == null) {
+        ensureLinkIsMutable();
+        link_.add(index, builderForValue.build());
+        onChanged();
+      } else {
+        linkBuilder_.addMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.InterconnectLink link = 1;</code>
+     */
+    public Builder addAllLink(
+        java.lang.Iterable<? extends org.tensorflow.proto.framework.InterconnectLink> values) {
+      if (linkBuilder_ == null) {
+        ensureLinkIsMutable();
+        com.google.protobuf.AbstractMessageLite.Builder.addAll(
+            values, link_);
+        onChanged();
+      } else {
+        linkBuilder_.addAllMessages(values);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.InterconnectLink link = 1;</code>
+     */
+    public Builder clearLink() {
+      if (linkBuilder_ == null) {
+        link_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000001);
+        onChanged();
+      } else {
+        linkBuilder_.clear();
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.InterconnectLink link = 1;</code>
+     */
+    public Builder removeLink(int index) {
+      if (linkBuilder_ == null) {
+        ensureLinkIsMutable();
+        link_.remove(index);
+        onChanged();
+      } else {
+        linkBuilder_.remove(index);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.InterconnectLink link = 1;</code>
+     */
+    public org.tensorflow.proto.framework.InterconnectLink.Builder getLinkBuilder(
+        int index) {
+      return getLinkFieldBuilder().getBuilder(index);
+    }
+    /**
+     * <code>repeated .tensorflow.InterconnectLink link = 1;</code>
+     */
+    public org.tensorflow.proto.framework.InterconnectLinkOrBuilder getLinkOrBuilder(
+        int index) {
+      if (linkBuilder_ == null) {
+        return link_.get(index);  } else {
+        return linkBuilder_.getMessageOrBuilder(index);
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.InterconnectLink link = 1;</code>
+     */
+    public java.util.List<? extends org.tensorflow.proto.framework.InterconnectLinkOrBuilder> 
+         getLinkOrBuilderList() {
+      if (linkBuilder_ != null) {
+        return linkBuilder_.getMessageOrBuilderList();
+      } else {
+        return java.util.Collections.unmodifiableList(link_);
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.InterconnectLink link = 1;</code>
+     */
+    public org.tensorflow.proto.framework.InterconnectLink.Builder addLinkBuilder() {
+      return getLinkFieldBuilder().addBuilder(
+          org.tensorflow.proto.framework.InterconnectLink.getDefaultInstance());
+    }
+    /**
+     * <code>repeated .tensorflow.InterconnectLink link = 1;</code>
+     */
+    public org.tensorflow.proto.framework.InterconnectLink.Builder addLinkBuilder(
+        int index) {
+      return getLinkFieldBuilder().addBuilder(
+          index, org.tensorflow.proto.framework.InterconnectLink.getDefaultInstance());
+    }
+    /**
+     * <code>repeated .tensorflow.InterconnectLink link = 1;</code>
+     */
+    public java.util.List<org.tensorflow.proto.framework.InterconnectLink.Builder> 
+         getLinkBuilderList() {
+      return getLinkFieldBuilder().getBuilderList();
+    }
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.framework.InterconnectLink, org.tensorflow.proto.framework.InterconnectLink.Builder, org.tensorflow.proto.framework.InterconnectLinkOrBuilder> 
+        getLinkFieldBuilder() {
+      if (linkBuilder_ == null) {
+        linkBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
+            org.tensorflow.proto.framework.InterconnectLink, org.tensorflow.proto.framework.InterconnectLink.Builder, org.tensorflow.proto.framework.InterconnectLinkOrBuilder>(
+                link_,
+                ((bitField0_ & 0x00000001) != 0),
+                getParentForChildren(),
+                isClean());
+        link_ = null;
+      }
+      return linkBuilder_;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.LocalLinks)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.LocalLinks)
+  private static final org.tensorflow.proto.framework.LocalLinks DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.LocalLinks();
+  }
+
+  public static org.tensorflow.proto.framework.LocalLinks getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<LocalLinks>
+      PARSER = new com.google.protobuf.AbstractParser<LocalLinks>() {
+    @java.lang.Override
+    public LocalLinks parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new LocalLinks(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<LocalLinks> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<LocalLinks> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.framework.LocalLinks getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/LocalLinksOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/LocalLinksOrBuilder.java
new file mode 100644
index 00000000000..e1cd5854d3b
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/LocalLinksOrBuilder.java
@@ -0,0 +1,33 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/framework/device_attributes.proto
+
+package org.tensorflow.proto.framework;
+
+public interface LocalLinksOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.LocalLinks)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <code>repeated .tensorflow.InterconnectLink link = 1;</code>
+   */
+  java.util.List<org.tensorflow.proto.framework.InterconnectLink> 
+      getLinkList();
+  /**
+   * <code>repeated .tensorflow.InterconnectLink link = 1;</code>
+   */
+  org.tensorflow.proto.framework.InterconnectLink getLink(int index);
+  /**
+   * <code>repeated .tensorflow.InterconnectLink link = 1;</code>
+   */
+  int getLinkCount();
+  /**
+   * <code>repeated .tensorflow.InterconnectLink link = 1;</code>
+   */
+  java.util.List<? extends org.tensorflow.proto.framework.InterconnectLinkOrBuilder> 
+      getLinkOrBuilderList();
+  /**
+   * <code>repeated .tensorflow.InterconnectLink link = 1;</code>
+   */
+  org.tensorflow.proto.framework.InterconnectLinkOrBuilder getLinkOrBuilder(
+      int index);
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/LogMemoryProtos.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/LogMemoryProtos.java
new file mode 100644
index 00000000000..4cef19ad6fe
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/LogMemoryProtos.java
@@ -0,0 +1,125 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/framework/log_memory.proto
+
+package org.tensorflow.proto.framework;
+
+public final class LogMemoryProtos {
+  private LogMemoryProtos() {}
+  public static void registerAllExtensions(
+      com.google.protobuf.ExtensionRegistryLite registry) {
+  }
+
+  public static void registerAllExtensions(
+      com.google.protobuf.ExtensionRegistry registry) {
+    registerAllExtensions(
+        (com.google.protobuf.ExtensionRegistryLite) registry);
+  }
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_MemoryLogStep_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_MemoryLogStep_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_MemoryLogTensorAllocation_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_MemoryLogTensorAllocation_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_MemoryLogTensorDeallocation_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_MemoryLogTensorDeallocation_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_MemoryLogTensorOutput_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_MemoryLogTensorOutput_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_MemoryLogRawAllocation_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_MemoryLogRawAllocation_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_MemoryLogRawDeallocation_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_MemoryLogRawDeallocation_fieldAccessorTable;
+
+  public static com.google.protobuf.Descriptors.FileDescriptor
+      getDescriptor() {
+    return descriptor;
+  }
+  private static  com.google.protobuf.Descriptors.FileDescriptor
+      descriptor;
+  static {
+    java.lang.String[] descriptorData = {
+      "\n*tensorflow/core/framework/log_memory.p" +
+      "roto\022\ntensorflow\0322tensorflow/core/framew" +
+      "ork/tensor_description.proto\"0\n\rMemoryLo" +
+      "gStep\022\017\n\007step_id\030\001 \001(\003\022\016\n\006handle\030\002 \001(\t\"p" +
+      "\n\031MemoryLogTensorAllocation\022\017\n\007step_id\030\001" +
+      " \001(\003\022\023\n\013kernel_name\030\002 \001(\t\022-\n\006tensor\030\003 \001(" +
+      "\0132\035.tensorflow.TensorDescription\"L\n\033Memo" +
+      "ryLogTensorDeallocation\022\025\n\rallocation_id" +
+      "\030\001 \001(\003\022\026\n\016allocator_name\030\002 \001(\t\"{\n\025Memory" +
+      "LogTensorOutput\022\017\n\007step_id\030\001 \001(\003\022\023\n\013kern" +
+      "el_name\030\002 \001(\t\022\r\n\005index\030\003 \001(\005\022-\n\006tensor\030\004" +
+      " \001(\0132\035.tensorflow.TensorDescription\"\213\001\n\026" +
+      "MemoryLogRawAllocation\022\017\n\007step_id\030\001 \001(\003\022" +
+      "\021\n\toperation\030\002 \001(\t\022\021\n\tnum_bytes\030\003 \001(\003\022\013\n" +
+      "\003ptr\030\004 \001(\004\022\025\n\rallocation_id\030\005 \001(\003\022\026\n\016all" +
+      "ocator_name\030\006 \001(\t\"\177\n\030MemoryLogRawDealloc" +
+      "ation\022\017\n\007step_id\030\001 \001(\003\022\021\n\toperation\030\002 \001(" +
+      "\t\022\025\n\rallocation_id\030\003 \001(\003\022\026\n\016allocator_na" +
+      "me\030\004 \001(\t\022\020\n\010deferred\030\005 \001(\010B\211\001\n\036org.tenso" +
+      "rflow.proto.frameworkB\017LogMemoryProtosP\001" +
+      "ZQgithub.com/tensorflow/tensorflow/tenso" +
+      "rflow/go/core/framework/log_memory_go_pr" +
+      "oto\370\001\001b\006proto3"
+    };
+    descriptor = com.google.protobuf.Descriptors.FileDescriptor
+      .internalBuildGeneratedFileFrom(descriptorData,
+        new com.google.protobuf.Descriptors.FileDescriptor[] {
+          org.tensorflow.proto.framework.TensorDescriptionProtos.getDescriptor(),
+        });
+    internal_static_tensorflow_MemoryLogStep_descriptor =
+      getDescriptor().getMessageTypes().get(0);
+    internal_static_tensorflow_MemoryLogStep_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_MemoryLogStep_descriptor,
+        new java.lang.String[] { "StepId", "Handle", });
+    internal_static_tensorflow_MemoryLogTensorAllocation_descriptor =
+      getDescriptor().getMessageTypes().get(1);
+    internal_static_tensorflow_MemoryLogTensorAllocation_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_MemoryLogTensorAllocation_descriptor,
+        new java.lang.String[] { "StepId", "KernelName", "Tensor", });
+    internal_static_tensorflow_MemoryLogTensorDeallocation_descriptor =
+      getDescriptor().getMessageTypes().get(2);
+    internal_static_tensorflow_MemoryLogTensorDeallocation_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_MemoryLogTensorDeallocation_descriptor,
+        new java.lang.String[] { "AllocationId", "AllocatorName", });
+    internal_static_tensorflow_MemoryLogTensorOutput_descriptor =
+      getDescriptor().getMessageTypes().get(3);
+    internal_static_tensorflow_MemoryLogTensorOutput_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_MemoryLogTensorOutput_descriptor,
+        new java.lang.String[] { "StepId", "KernelName", "Index", "Tensor", });
+    internal_static_tensorflow_MemoryLogRawAllocation_descriptor =
+      getDescriptor().getMessageTypes().get(4);
+    internal_static_tensorflow_MemoryLogRawAllocation_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_MemoryLogRawAllocation_descriptor,
+        new java.lang.String[] { "StepId", "Operation", "NumBytes", "Ptr", "AllocationId", "AllocatorName", });
+    internal_static_tensorflow_MemoryLogRawDeallocation_descriptor =
+      getDescriptor().getMessageTypes().get(5);
+    internal_static_tensorflow_MemoryLogRawDeallocation_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_MemoryLogRawDeallocation_descriptor,
+        new java.lang.String[] { "StepId", "Operation", "AllocationId", "AllocatorName", "Deferred", });
+    org.tensorflow.proto.framework.TensorDescriptionProtos.getDescriptor();
+  }
+
+  // @@protoc_insertion_point(outer_class_scope)
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/MemoryLogRawAllocation.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/MemoryLogRawAllocation.java
new file mode 100644
index 00000000000..db40403eba4
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/MemoryLogRawAllocation.java
@@ -0,0 +1,1029 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/framework/log_memory.proto
+
+package org.tensorflow.proto.framework;
+
+/**
+ * Protobuf type {@code tensorflow.MemoryLogRawAllocation}
+ */
+public  final class MemoryLogRawAllocation extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.MemoryLogRawAllocation)
+    MemoryLogRawAllocationOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use MemoryLogRawAllocation.newBuilder() to construct.
+  private MemoryLogRawAllocation(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private MemoryLogRawAllocation() {
+    operation_ = "";
+    allocatorName_ = "";
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new MemoryLogRawAllocation();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private MemoryLogRawAllocation(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 8: {
+
+            stepId_ = input.readInt64();
+            break;
+          }
+          case 18: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            operation_ = s;
+            break;
+          }
+          case 24: {
+
+            numBytes_ = input.readInt64();
+            break;
+          }
+          case 32: {
+
+            ptr_ = input.readUInt64();
+            break;
+          }
+          case 40: {
+
+            allocationId_ = input.readInt64();
+            break;
+          }
+          case 50: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            allocatorName_ = s;
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.framework.LogMemoryProtos.internal_static_tensorflow_MemoryLogRawAllocation_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.framework.LogMemoryProtos.internal_static_tensorflow_MemoryLogRawAllocation_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.framework.MemoryLogRawAllocation.class, org.tensorflow.proto.framework.MemoryLogRawAllocation.Builder.class);
+  }
+
+  public static final int STEP_ID_FIELD_NUMBER = 1;
+  private long stepId_;
+  /**
+   * <pre>
+   * Process-unique step id.
+   * </pre>
+   *
+   * <code>int64 step_id = 1;</code>
+   */
+  public long getStepId() {
+    return stepId_;
+  }
+
+  public static final int OPERATION_FIELD_NUMBER = 2;
+  private volatile java.lang.Object operation_;
+  /**
+   * <pre>
+   * Name of the operation making the allocation.
+   * </pre>
+   *
+   * <code>string operation = 2;</code>
+   */
+  public java.lang.String getOperation() {
+    java.lang.Object ref = operation_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      operation_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * Name of the operation making the allocation.
+   * </pre>
+   *
+   * <code>string operation = 2;</code>
+   */
+  public com.google.protobuf.ByteString
+      getOperationBytes() {
+    java.lang.Object ref = operation_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      operation_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int NUM_BYTES_FIELD_NUMBER = 3;
+  private long numBytes_;
+  /**
+   * <pre>
+   * Number of bytes in the allocation.
+   * </pre>
+   *
+   * <code>int64 num_bytes = 3;</code>
+   */
+  public long getNumBytes() {
+    return numBytes_;
+  }
+
+  public static final int PTR_FIELD_NUMBER = 4;
+  private long ptr_;
+  /**
+   * <pre>
+   * Address of the allocation.
+   * </pre>
+   *
+   * <code>uint64 ptr = 4;</code>
+   */
+  public long getPtr() {
+    return ptr_;
+  }
+
+  public static final int ALLOCATION_ID_FIELD_NUMBER = 5;
+  private long allocationId_;
+  /**
+   * <pre>
+   * Id of the tensor buffer being allocated, used to match to a
+   * corresponding deallocation.
+   * </pre>
+   *
+   * <code>int64 allocation_id = 5;</code>
+   */
+  public long getAllocationId() {
+    return allocationId_;
+  }
+
+  public static final int ALLOCATOR_NAME_FIELD_NUMBER = 6;
+  private volatile java.lang.Object allocatorName_;
+  /**
+   * <pre>
+   * Name of the allocator used.
+   * </pre>
+   *
+   * <code>string allocator_name = 6;</code>
+   */
+  public java.lang.String getAllocatorName() {
+    java.lang.Object ref = allocatorName_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      allocatorName_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * Name of the allocator used.
+   * </pre>
+   *
+   * <code>string allocator_name = 6;</code>
+   */
+  public com.google.protobuf.ByteString
+      getAllocatorNameBytes() {
+    java.lang.Object ref = allocatorName_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      allocatorName_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    if (stepId_ != 0L) {
+      output.writeInt64(1, stepId_);
+    }
+    if (!getOperationBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 2, operation_);
+    }
+    if (numBytes_ != 0L) {
+      output.writeInt64(3, numBytes_);
+    }
+    if (ptr_ != 0L) {
+      output.writeUInt64(4, ptr_);
+    }
+    if (allocationId_ != 0L) {
+      output.writeInt64(5, allocationId_);
+    }
+    if (!getAllocatorNameBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 6, allocatorName_);
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (stepId_ != 0L) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt64Size(1, stepId_);
+    }
+    if (!getOperationBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(2, operation_);
+    }
+    if (numBytes_ != 0L) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt64Size(3, numBytes_);
+    }
+    if (ptr_ != 0L) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeUInt64Size(4, ptr_);
+    }
+    if (allocationId_ != 0L) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt64Size(5, allocationId_);
+    }
+    if (!getAllocatorNameBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(6, allocatorName_);
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.framework.MemoryLogRawAllocation)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.framework.MemoryLogRawAllocation other = (org.tensorflow.proto.framework.MemoryLogRawAllocation) obj;
+
+    if (getStepId()
+        != other.getStepId()) return false;
+    if (!getOperation()
+        .equals(other.getOperation())) return false;
+    if (getNumBytes()
+        != other.getNumBytes()) return false;
+    if (getPtr()
+        != other.getPtr()) return false;
+    if (getAllocationId()
+        != other.getAllocationId()) return false;
+    if (!getAllocatorName()
+        .equals(other.getAllocatorName())) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    hash = (37 * hash) + STEP_ID_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+        getStepId());
+    hash = (37 * hash) + OPERATION_FIELD_NUMBER;
+    hash = (53 * hash) + getOperation().hashCode();
+    hash = (37 * hash) + NUM_BYTES_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+        getNumBytes());
+    hash = (37 * hash) + PTR_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+        getPtr());
+    hash = (37 * hash) + ALLOCATION_ID_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+        getAllocationId());
+    hash = (37 * hash) + ALLOCATOR_NAME_FIELD_NUMBER;
+    hash = (53 * hash) + getAllocatorName().hashCode();
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.framework.MemoryLogRawAllocation parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.MemoryLogRawAllocation parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.MemoryLogRawAllocation parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.MemoryLogRawAllocation parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.MemoryLogRawAllocation parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.MemoryLogRawAllocation parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.MemoryLogRawAllocation parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.MemoryLogRawAllocation parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.MemoryLogRawAllocation parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.MemoryLogRawAllocation parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.MemoryLogRawAllocation parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.MemoryLogRawAllocation parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.framework.MemoryLogRawAllocation prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * Protobuf type {@code tensorflow.MemoryLogRawAllocation}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.MemoryLogRawAllocation)
+      org.tensorflow.proto.framework.MemoryLogRawAllocationOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.LogMemoryProtos.internal_static_tensorflow_MemoryLogRawAllocation_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.LogMemoryProtos.internal_static_tensorflow_MemoryLogRawAllocation_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.MemoryLogRawAllocation.class, org.tensorflow.proto.framework.MemoryLogRawAllocation.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.framework.MemoryLogRawAllocation.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      stepId_ = 0L;
+
+      operation_ = "";
+
+      numBytes_ = 0L;
+
+      ptr_ = 0L;
+
+      allocationId_ = 0L;
+
+      allocatorName_ = "";
+
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.framework.LogMemoryProtos.internal_static_tensorflow_MemoryLogRawAllocation_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.MemoryLogRawAllocation getDefaultInstanceForType() {
+      return org.tensorflow.proto.framework.MemoryLogRawAllocation.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.MemoryLogRawAllocation build() {
+      org.tensorflow.proto.framework.MemoryLogRawAllocation result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.MemoryLogRawAllocation buildPartial() {
+      org.tensorflow.proto.framework.MemoryLogRawAllocation result = new org.tensorflow.proto.framework.MemoryLogRawAllocation(this);
+      result.stepId_ = stepId_;
+      result.operation_ = operation_;
+      result.numBytes_ = numBytes_;
+      result.ptr_ = ptr_;
+      result.allocationId_ = allocationId_;
+      result.allocatorName_ = allocatorName_;
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.framework.MemoryLogRawAllocation) {
+        return mergeFrom((org.tensorflow.proto.framework.MemoryLogRawAllocation)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.framework.MemoryLogRawAllocation other) {
+      if (other == org.tensorflow.proto.framework.MemoryLogRawAllocation.getDefaultInstance()) return this;
+      if (other.getStepId() != 0L) {
+        setStepId(other.getStepId());
+      }
+      if (!other.getOperation().isEmpty()) {
+        operation_ = other.operation_;
+        onChanged();
+      }
+      if (other.getNumBytes() != 0L) {
+        setNumBytes(other.getNumBytes());
+      }
+      if (other.getPtr() != 0L) {
+        setPtr(other.getPtr());
+      }
+      if (other.getAllocationId() != 0L) {
+        setAllocationId(other.getAllocationId());
+      }
+      if (!other.getAllocatorName().isEmpty()) {
+        allocatorName_ = other.allocatorName_;
+        onChanged();
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.framework.MemoryLogRawAllocation parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.framework.MemoryLogRawAllocation) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+
+    private long stepId_ ;
+    /**
+     * <pre>
+     * Process-unique step id.
+     * </pre>
+     *
+     * <code>int64 step_id = 1;</code>
+     */
+    public long getStepId() {
+      return stepId_;
+    }
+    /**
+     * <pre>
+     * Process-unique step id.
+     * </pre>
+     *
+     * <code>int64 step_id = 1;</code>
+     */
+    public Builder setStepId(long value) {
+      
+      stepId_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Process-unique step id.
+     * </pre>
+     *
+     * <code>int64 step_id = 1;</code>
+     */
+    public Builder clearStepId() {
+      
+      stepId_ = 0L;
+      onChanged();
+      return this;
+    }
+
+    private java.lang.Object operation_ = "";
+    /**
+     * <pre>
+     * Name of the operation making the allocation.
+     * </pre>
+     *
+     * <code>string operation = 2;</code>
+     */
+    public java.lang.String getOperation() {
+      java.lang.Object ref = operation_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        operation_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Name of the operation making the allocation.
+     * </pre>
+     *
+     * <code>string operation = 2;</code>
+     */
+    public com.google.protobuf.ByteString
+        getOperationBytes() {
+      java.lang.Object ref = operation_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        operation_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Name of the operation making the allocation.
+     * </pre>
+     *
+     * <code>string operation = 2;</code>
+     */
+    public Builder setOperation(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      operation_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Name of the operation making the allocation.
+     * </pre>
+     *
+     * <code>string operation = 2;</code>
+     */
+    public Builder clearOperation() {
+      
+      operation_ = getDefaultInstance().getOperation();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Name of the operation making the allocation.
+     * </pre>
+     *
+     * <code>string operation = 2;</code>
+     */
+    public Builder setOperationBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      operation_ = value;
+      onChanged();
+      return this;
+    }
+
+    private long numBytes_ ;
+    /**
+     * <pre>
+     * Number of bytes in the allocation.
+     * </pre>
+     *
+     * <code>int64 num_bytes = 3;</code>
+     */
+    public long getNumBytes() {
+      return numBytes_;
+    }
+    /**
+     * <pre>
+     * Number of bytes in the allocation.
+     * </pre>
+     *
+     * <code>int64 num_bytes = 3;</code>
+     */
+    public Builder setNumBytes(long value) {
+      
+      numBytes_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Number of bytes in the allocation.
+     * </pre>
+     *
+     * <code>int64 num_bytes = 3;</code>
+     */
+    public Builder clearNumBytes() {
+      
+      numBytes_ = 0L;
+      onChanged();
+      return this;
+    }
+
+    private long ptr_ ;
+    /**
+     * <pre>
+     * Address of the allocation.
+     * </pre>
+     *
+     * <code>uint64 ptr = 4;</code>
+     */
+    public long getPtr() {
+      return ptr_;
+    }
+    /**
+     * <pre>
+     * Address of the allocation.
+     * </pre>
+     *
+     * <code>uint64 ptr = 4;</code>
+     */
+    public Builder setPtr(long value) {
+      
+      ptr_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Address of the allocation.
+     * </pre>
+     *
+     * <code>uint64 ptr = 4;</code>
+     */
+    public Builder clearPtr() {
+      
+      ptr_ = 0L;
+      onChanged();
+      return this;
+    }
+
+    private long allocationId_ ;
+    /**
+     * <pre>
+     * Id of the tensor buffer being allocated, used to match to a
+     * corresponding deallocation.
+     * </pre>
+     *
+     * <code>int64 allocation_id = 5;</code>
+     */
+    public long getAllocationId() {
+      return allocationId_;
+    }
+    /**
+     * <pre>
+     * Id of the tensor buffer being allocated, used to match to a
+     * corresponding deallocation.
+     * </pre>
+     *
+     * <code>int64 allocation_id = 5;</code>
+     */
+    public Builder setAllocationId(long value) {
+      
+      allocationId_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Id of the tensor buffer being allocated, used to match to a
+     * corresponding deallocation.
+     * </pre>
+     *
+     * <code>int64 allocation_id = 5;</code>
+     */
+    public Builder clearAllocationId() {
+      
+      allocationId_ = 0L;
+      onChanged();
+      return this;
+    }
+
+    private java.lang.Object allocatorName_ = "";
+    /**
+     * <pre>
+     * Name of the allocator used.
+     * </pre>
+     *
+     * <code>string allocator_name = 6;</code>
+     */
+    public java.lang.String getAllocatorName() {
+      java.lang.Object ref = allocatorName_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        allocatorName_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Name of the allocator used.
+     * </pre>
+     *
+     * <code>string allocator_name = 6;</code>
+     */
+    public com.google.protobuf.ByteString
+        getAllocatorNameBytes() {
+      java.lang.Object ref = allocatorName_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        allocatorName_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Name of the allocator used.
+     * </pre>
+     *
+     * <code>string allocator_name = 6;</code>
+     */
+    public Builder setAllocatorName(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      allocatorName_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Name of the allocator used.
+     * </pre>
+     *
+     * <code>string allocator_name = 6;</code>
+     */
+    public Builder clearAllocatorName() {
+      
+      allocatorName_ = getDefaultInstance().getAllocatorName();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Name of the allocator used.
+     * </pre>
+     *
+     * <code>string allocator_name = 6;</code>
+     */
+    public Builder setAllocatorNameBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      allocatorName_ = value;
+      onChanged();
+      return this;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.MemoryLogRawAllocation)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.MemoryLogRawAllocation)
+  private static final org.tensorflow.proto.framework.MemoryLogRawAllocation DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.MemoryLogRawAllocation();
+  }
+
+  public static org.tensorflow.proto.framework.MemoryLogRawAllocation getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<MemoryLogRawAllocation>
+      PARSER = new com.google.protobuf.AbstractParser<MemoryLogRawAllocation>() {
+    @java.lang.Override
+    public MemoryLogRawAllocation parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new MemoryLogRawAllocation(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<MemoryLogRawAllocation> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<MemoryLogRawAllocation> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.framework.MemoryLogRawAllocation getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/MemoryLogRawAllocationOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/MemoryLogRawAllocationOrBuilder.java
new file mode 100644
index 00000000000..d8cabb6c9d4
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/MemoryLogRawAllocationOrBuilder.java
@@ -0,0 +1,82 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/framework/log_memory.proto
+
+package org.tensorflow.proto.framework;
+
+public interface MemoryLogRawAllocationOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.MemoryLogRawAllocation)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <pre>
+   * Process-unique step id.
+   * </pre>
+   *
+   * <code>int64 step_id = 1;</code>
+   */
+  long getStepId();
+
+  /**
+   * <pre>
+   * Name of the operation making the allocation.
+   * </pre>
+   *
+   * <code>string operation = 2;</code>
+   */
+  java.lang.String getOperation();
+  /**
+   * <pre>
+   * Name of the operation making the allocation.
+   * </pre>
+   *
+   * <code>string operation = 2;</code>
+   */
+  com.google.protobuf.ByteString
+      getOperationBytes();
+
+  /**
+   * <pre>
+   * Number of bytes in the allocation.
+   * </pre>
+   *
+   * <code>int64 num_bytes = 3;</code>
+   */
+  long getNumBytes();
+
+  /**
+   * <pre>
+   * Address of the allocation.
+   * </pre>
+   *
+   * <code>uint64 ptr = 4;</code>
+   */
+  long getPtr();
+
+  /**
+   * <pre>
+   * Id of the tensor buffer being allocated, used to match to a
+   * corresponding deallocation.
+   * </pre>
+   *
+   * <code>int64 allocation_id = 5;</code>
+   */
+  long getAllocationId();
+
+  /**
+   * <pre>
+   * Name of the allocator used.
+   * </pre>
+   *
+   * <code>string allocator_name = 6;</code>
+   */
+  java.lang.String getAllocatorName();
+  /**
+   * <pre>
+   * Name of the allocator used.
+   * </pre>
+   *
+   * <code>string allocator_name = 6;</code>
+   */
+  com.google.protobuf.ByteString
+      getAllocatorNameBytes();
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/MemoryLogRawDeallocation.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/MemoryLogRawDeallocation.java
new file mode 100644
index 00000000000..ae8c55fcf31
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/MemoryLogRawDeallocation.java
@@ -0,0 +1,959 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/framework/log_memory.proto
+
+package org.tensorflow.proto.framework;
+
+/**
+ * Protobuf type {@code tensorflow.MemoryLogRawDeallocation}
+ */
+public  final class MemoryLogRawDeallocation extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.MemoryLogRawDeallocation)
+    MemoryLogRawDeallocationOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use MemoryLogRawDeallocation.newBuilder() to construct.
+  private MemoryLogRawDeallocation(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private MemoryLogRawDeallocation() {
+    operation_ = "";
+    allocatorName_ = "";
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new MemoryLogRawDeallocation();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private MemoryLogRawDeallocation(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 8: {
+
+            stepId_ = input.readInt64();
+            break;
+          }
+          case 18: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            operation_ = s;
+            break;
+          }
+          case 24: {
+
+            allocationId_ = input.readInt64();
+            break;
+          }
+          case 34: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            allocatorName_ = s;
+            break;
+          }
+          case 40: {
+
+            deferred_ = input.readBool();
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.framework.LogMemoryProtos.internal_static_tensorflow_MemoryLogRawDeallocation_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.framework.LogMemoryProtos.internal_static_tensorflow_MemoryLogRawDeallocation_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.framework.MemoryLogRawDeallocation.class, org.tensorflow.proto.framework.MemoryLogRawDeallocation.Builder.class);
+  }
+
+  public static final int STEP_ID_FIELD_NUMBER = 1;
+  private long stepId_;
+  /**
+   * <pre>
+   * Process-unique step id.
+   * </pre>
+   *
+   * <code>int64 step_id = 1;</code>
+   */
+  public long getStepId() {
+    return stepId_;
+  }
+
+  public static final int OPERATION_FIELD_NUMBER = 2;
+  private volatile java.lang.Object operation_;
+  /**
+   * <pre>
+   * Name of the operation making the deallocation.
+   * </pre>
+   *
+   * <code>string operation = 2;</code>
+   */
+  public java.lang.String getOperation() {
+    java.lang.Object ref = operation_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      operation_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * Name of the operation making the deallocation.
+   * </pre>
+   *
+   * <code>string operation = 2;</code>
+   */
+  public com.google.protobuf.ByteString
+      getOperationBytes() {
+    java.lang.Object ref = operation_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      operation_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int ALLOCATION_ID_FIELD_NUMBER = 3;
+  private long allocationId_;
+  /**
+   * <pre>
+   * Id of the tensor buffer being deallocated, used to match to a
+   * corresponding allocation.
+   * </pre>
+   *
+   * <code>int64 allocation_id = 3;</code>
+   */
+  public long getAllocationId() {
+    return allocationId_;
+  }
+
+  public static final int ALLOCATOR_NAME_FIELD_NUMBER = 4;
+  private volatile java.lang.Object allocatorName_;
+  /**
+   * <pre>
+   * Name of the allocator used.
+   * </pre>
+   *
+   * <code>string allocator_name = 4;</code>
+   */
+  public java.lang.String getAllocatorName() {
+    java.lang.Object ref = allocatorName_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      allocatorName_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * Name of the allocator used.
+   * </pre>
+   *
+   * <code>string allocator_name = 4;</code>
+   */
+  public com.google.protobuf.ByteString
+      getAllocatorNameBytes() {
+    java.lang.Object ref = allocatorName_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      allocatorName_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int DEFERRED_FIELD_NUMBER = 5;
+  private boolean deferred_;
+  /**
+   * <pre>
+   * True if the deallocation is queued and will be performed later,
+   * e.g. for GPU lazy freeing of buffers.
+   * </pre>
+   *
+   * <code>bool deferred = 5;</code>
+   */
+  public boolean getDeferred() {
+    return deferred_;
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    if (stepId_ != 0L) {
+      output.writeInt64(1, stepId_);
+    }
+    if (!getOperationBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 2, operation_);
+    }
+    if (allocationId_ != 0L) {
+      output.writeInt64(3, allocationId_);
+    }
+    if (!getAllocatorNameBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 4, allocatorName_);
+    }
+    if (deferred_ != false) {
+      output.writeBool(5, deferred_);
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (stepId_ != 0L) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt64Size(1, stepId_);
+    }
+    if (!getOperationBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(2, operation_);
+    }
+    if (allocationId_ != 0L) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt64Size(3, allocationId_);
+    }
+    if (!getAllocatorNameBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(4, allocatorName_);
+    }
+    if (deferred_ != false) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeBoolSize(5, deferred_);
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.framework.MemoryLogRawDeallocation)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.framework.MemoryLogRawDeallocation other = (org.tensorflow.proto.framework.MemoryLogRawDeallocation) obj;
+
+    if (getStepId()
+        != other.getStepId()) return false;
+    if (!getOperation()
+        .equals(other.getOperation())) return false;
+    if (getAllocationId()
+        != other.getAllocationId()) return false;
+    if (!getAllocatorName()
+        .equals(other.getAllocatorName())) return false;
+    if (getDeferred()
+        != other.getDeferred()) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    hash = (37 * hash) + STEP_ID_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+        getStepId());
+    hash = (37 * hash) + OPERATION_FIELD_NUMBER;
+    hash = (53 * hash) + getOperation().hashCode();
+    hash = (37 * hash) + ALLOCATION_ID_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+        getAllocationId());
+    hash = (37 * hash) + ALLOCATOR_NAME_FIELD_NUMBER;
+    hash = (53 * hash) + getAllocatorName().hashCode();
+    hash = (37 * hash) + DEFERRED_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashBoolean(
+        getDeferred());
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.framework.MemoryLogRawDeallocation parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.MemoryLogRawDeallocation parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.MemoryLogRawDeallocation parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.MemoryLogRawDeallocation parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.MemoryLogRawDeallocation parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.MemoryLogRawDeallocation parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.MemoryLogRawDeallocation parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.MemoryLogRawDeallocation parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.MemoryLogRawDeallocation parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.MemoryLogRawDeallocation parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.MemoryLogRawDeallocation parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.MemoryLogRawDeallocation parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.framework.MemoryLogRawDeallocation prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * Protobuf type {@code tensorflow.MemoryLogRawDeallocation}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.MemoryLogRawDeallocation)
+      org.tensorflow.proto.framework.MemoryLogRawDeallocationOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.LogMemoryProtos.internal_static_tensorflow_MemoryLogRawDeallocation_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.LogMemoryProtos.internal_static_tensorflow_MemoryLogRawDeallocation_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.MemoryLogRawDeallocation.class, org.tensorflow.proto.framework.MemoryLogRawDeallocation.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.framework.MemoryLogRawDeallocation.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      stepId_ = 0L;
+
+      operation_ = "";
+
+      allocationId_ = 0L;
+
+      allocatorName_ = "";
+
+      deferred_ = false;
+
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.framework.LogMemoryProtos.internal_static_tensorflow_MemoryLogRawDeallocation_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.MemoryLogRawDeallocation getDefaultInstanceForType() {
+      return org.tensorflow.proto.framework.MemoryLogRawDeallocation.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.MemoryLogRawDeallocation build() {
+      org.tensorflow.proto.framework.MemoryLogRawDeallocation result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.MemoryLogRawDeallocation buildPartial() {
+      org.tensorflow.proto.framework.MemoryLogRawDeallocation result = new org.tensorflow.proto.framework.MemoryLogRawDeallocation(this);
+      result.stepId_ = stepId_;
+      result.operation_ = operation_;
+      result.allocationId_ = allocationId_;
+      result.allocatorName_ = allocatorName_;
+      result.deferred_ = deferred_;
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.framework.MemoryLogRawDeallocation) {
+        return mergeFrom((org.tensorflow.proto.framework.MemoryLogRawDeallocation)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.framework.MemoryLogRawDeallocation other) {
+      if (other == org.tensorflow.proto.framework.MemoryLogRawDeallocation.getDefaultInstance()) return this;
+      if (other.getStepId() != 0L) {
+        setStepId(other.getStepId());
+      }
+      if (!other.getOperation().isEmpty()) {
+        operation_ = other.operation_;
+        onChanged();
+      }
+      if (other.getAllocationId() != 0L) {
+        setAllocationId(other.getAllocationId());
+      }
+      if (!other.getAllocatorName().isEmpty()) {
+        allocatorName_ = other.allocatorName_;
+        onChanged();
+      }
+      if (other.getDeferred() != false) {
+        setDeferred(other.getDeferred());
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.framework.MemoryLogRawDeallocation parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.framework.MemoryLogRawDeallocation) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+
+    private long stepId_ ;
+    /**
+     * <pre>
+     * Process-unique step id.
+     * </pre>
+     *
+     * <code>int64 step_id = 1;</code>
+     */
+    public long getStepId() {
+      return stepId_;
+    }
+    /**
+     * <pre>
+     * Process-unique step id.
+     * </pre>
+     *
+     * <code>int64 step_id = 1;</code>
+     */
+    public Builder setStepId(long value) {
+      
+      stepId_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Process-unique step id.
+     * </pre>
+     *
+     * <code>int64 step_id = 1;</code>
+     */
+    public Builder clearStepId() {
+      
+      stepId_ = 0L;
+      onChanged();
+      return this;
+    }
+
+    private java.lang.Object operation_ = "";
+    /**
+     * <pre>
+     * Name of the operation making the deallocation.
+     * </pre>
+     *
+     * <code>string operation = 2;</code>
+     */
+    public java.lang.String getOperation() {
+      java.lang.Object ref = operation_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        operation_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Name of the operation making the deallocation.
+     * </pre>
+     *
+     * <code>string operation = 2;</code>
+     */
+    public com.google.protobuf.ByteString
+        getOperationBytes() {
+      java.lang.Object ref = operation_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        operation_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Name of the operation making the deallocation.
+     * </pre>
+     *
+     * <code>string operation = 2;</code>
+     */
+    public Builder setOperation(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      operation_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Name of the operation making the deallocation.
+     * </pre>
+     *
+     * <code>string operation = 2;</code>
+     */
+    public Builder clearOperation() {
+      
+      operation_ = getDefaultInstance().getOperation();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Name of the operation making the deallocation.
+     * </pre>
+     *
+     * <code>string operation = 2;</code>
+     */
+    public Builder setOperationBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      operation_ = value;
+      onChanged();
+      return this;
+    }
+
+    private long allocationId_ ;
+    /**
+     * <pre>
+     * Id of the tensor buffer being deallocated, used to match to a
+     * corresponding allocation.
+     * </pre>
+     *
+     * <code>int64 allocation_id = 3;</code>
+     */
+    public long getAllocationId() {
+      return allocationId_;
+    }
+    /**
+     * <pre>
+     * Id of the tensor buffer being deallocated, used to match to a
+     * corresponding allocation.
+     * </pre>
+     *
+     * <code>int64 allocation_id = 3;</code>
+     */
+    public Builder setAllocationId(long value) {
+      
+      allocationId_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Id of the tensor buffer being deallocated, used to match to a
+     * corresponding allocation.
+     * </pre>
+     *
+     * <code>int64 allocation_id = 3;</code>
+     */
+    public Builder clearAllocationId() {
+      
+      allocationId_ = 0L;
+      onChanged();
+      return this;
+    }
+
+    private java.lang.Object allocatorName_ = "";
+    /**
+     * <pre>
+     * Name of the allocator used.
+     * </pre>
+     *
+     * <code>string allocator_name = 4;</code>
+     */
+    public java.lang.String getAllocatorName() {
+      java.lang.Object ref = allocatorName_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        allocatorName_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Name of the allocator used.
+     * </pre>
+     *
+     * <code>string allocator_name = 4;</code>
+     */
+    public com.google.protobuf.ByteString
+        getAllocatorNameBytes() {
+      java.lang.Object ref = allocatorName_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        allocatorName_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Name of the allocator used.
+     * </pre>
+     *
+     * <code>string allocator_name = 4;</code>
+     */
+    public Builder setAllocatorName(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      allocatorName_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Name of the allocator used.
+     * </pre>
+     *
+     * <code>string allocator_name = 4;</code>
+     */
+    public Builder clearAllocatorName() {
+      
+      allocatorName_ = getDefaultInstance().getAllocatorName();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Name of the allocator used.
+     * </pre>
+     *
+     * <code>string allocator_name = 4;</code>
+     */
+    public Builder setAllocatorNameBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      allocatorName_ = value;
+      onChanged();
+      return this;
+    }
+
+    private boolean deferred_ ;
+    /**
+     * <pre>
+     * True if the deallocation is queued and will be performed later,
+     * e.g. for GPU lazy freeing of buffers.
+     * </pre>
+     *
+     * <code>bool deferred = 5;</code>
+     */
+    public boolean getDeferred() {
+      return deferred_;
+    }
+    /**
+     * <pre>
+     * True if the deallocation is queued and will be performed later,
+     * e.g. for GPU lazy freeing of buffers.
+     * </pre>
+     *
+     * <code>bool deferred = 5;</code>
+     */
+    public Builder setDeferred(boolean value) {
+      
+      deferred_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * True if the deallocation is queued and will be performed later,
+     * e.g. for GPU lazy freeing of buffers.
+     * </pre>
+     *
+     * <code>bool deferred = 5;</code>
+     */
+    public Builder clearDeferred() {
+      
+      deferred_ = false;
+      onChanged();
+      return this;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.MemoryLogRawDeallocation)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.MemoryLogRawDeallocation)
+  private static final org.tensorflow.proto.framework.MemoryLogRawDeallocation DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.MemoryLogRawDeallocation();
+  }
+
+  public static org.tensorflow.proto.framework.MemoryLogRawDeallocation getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<MemoryLogRawDeallocation>
+      PARSER = new com.google.protobuf.AbstractParser<MemoryLogRawDeallocation>() {
+    @java.lang.Override
+    public MemoryLogRawDeallocation parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new MemoryLogRawDeallocation(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<MemoryLogRawDeallocation> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<MemoryLogRawDeallocation> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.framework.MemoryLogRawDeallocation getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/MemoryLogRawDeallocationOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/MemoryLogRawDeallocationOrBuilder.java
new file mode 100644
index 00000000000..e8f66fff55c
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/MemoryLogRawDeallocationOrBuilder.java
@@ -0,0 +1,74 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/framework/log_memory.proto
+
+package org.tensorflow.proto.framework;
+
+public interface MemoryLogRawDeallocationOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.MemoryLogRawDeallocation)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <pre>
+   * Process-unique step id.
+   * </pre>
+   *
+   * <code>int64 step_id = 1;</code>
+   */
+  long getStepId();
+
+  /**
+   * <pre>
+   * Name of the operation making the deallocation.
+   * </pre>
+   *
+   * <code>string operation = 2;</code>
+   */
+  java.lang.String getOperation();
+  /**
+   * <pre>
+   * Name of the operation making the deallocation.
+   * </pre>
+   *
+   * <code>string operation = 2;</code>
+   */
+  com.google.protobuf.ByteString
+      getOperationBytes();
+
+  /**
+   * <pre>
+   * Id of the tensor buffer being deallocated, used to match to a
+   * corresponding allocation.
+   * </pre>
+   *
+   * <code>int64 allocation_id = 3;</code>
+   */
+  long getAllocationId();
+
+  /**
+   * <pre>
+   * Name of the allocator used.
+   * </pre>
+   *
+   * <code>string allocator_name = 4;</code>
+   */
+  java.lang.String getAllocatorName();
+  /**
+   * <pre>
+   * Name of the allocator used.
+   * </pre>
+   *
+   * <code>string allocator_name = 4;</code>
+   */
+  com.google.protobuf.ByteString
+      getAllocatorNameBytes();
+
+  /**
+   * <pre>
+   * True if the deallocation is queued and will be performed later,
+   * e.g. for GPU lazy freeing of buffers.
+   * </pre>
+   *
+   * <code>bool deferred = 5;</code>
+   */
+  boolean getDeferred();
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/MemoryLogStep.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/MemoryLogStep.java
new file mode 100644
index 00000000000..73d400a36e0
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/MemoryLogStep.java
@@ -0,0 +1,648 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/framework/log_memory.proto
+
+package org.tensorflow.proto.framework;
+
+/**
+ * Protobuf type {@code tensorflow.MemoryLogStep}
+ */
+public  final class MemoryLogStep extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.MemoryLogStep)
+    MemoryLogStepOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use MemoryLogStep.newBuilder() to construct.
+  private MemoryLogStep(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private MemoryLogStep() {
+    handle_ = "";
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new MemoryLogStep();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private MemoryLogStep(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 8: {
+
+            stepId_ = input.readInt64();
+            break;
+          }
+          case 18: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            handle_ = s;
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.framework.LogMemoryProtos.internal_static_tensorflow_MemoryLogStep_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.framework.LogMemoryProtos.internal_static_tensorflow_MemoryLogStep_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.framework.MemoryLogStep.class, org.tensorflow.proto.framework.MemoryLogStep.Builder.class);
+  }
+
+  public static final int STEP_ID_FIELD_NUMBER = 1;
+  private long stepId_;
+  /**
+   * <pre>
+   * Process-unique step id.
+   * </pre>
+   *
+   * <code>int64 step_id = 1;</code>
+   */
+  public long getStepId() {
+    return stepId_;
+  }
+
+  public static final int HANDLE_FIELD_NUMBER = 2;
+  private volatile java.lang.Object handle_;
+  /**
+   * <pre>
+   * Handle describing the feeds and fetches of the step.
+   * </pre>
+   *
+   * <code>string handle = 2;</code>
+   */
+  public java.lang.String getHandle() {
+    java.lang.Object ref = handle_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      handle_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * Handle describing the feeds and fetches of the step.
+   * </pre>
+   *
+   * <code>string handle = 2;</code>
+   */
+  public com.google.protobuf.ByteString
+      getHandleBytes() {
+    java.lang.Object ref = handle_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      handle_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    if (stepId_ != 0L) {
+      output.writeInt64(1, stepId_);
+    }
+    if (!getHandleBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 2, handle_);
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (stepId_ != 0L) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt64Size(1, stepId_);
+    }
+    if (!getHandleBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(2, handle_);
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.framework.MemoryLogStep)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.framework.MemoryLogStep other = (org.tensorflow.proto.framework.MemoryLogStep) obj;
+
+    if (getStepId()
+        != other.getStepId()) return false;
+    if (!getHandle()
+        .equals(other.getHandle())) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    hash = (37 * hash) + STEP_ID_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+        getStepId());
+    hash = (37 * hash) + HANDLE_FIELD_NUMBER;
+    hash = (53 * hash) + getHandle().hashCode();
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.framework.MemoryLogStep parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.MemoryLogStep parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.MemoryLogStep parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.MemoryLogStep parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.MemoryLogStep parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.MemoryLogStep parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.MemoryLogStep parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.MemoryLogStep parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.MemoryLogStep parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.MemoryLogStep parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.MemoryLogStep parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.MemoryLogStep parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.framework.MemoryLogStep prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * Protobuf type {@code tensorflow.MemoryLogStep}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.MemoryLogStep)
+      org.tensorflow.proto.framework.MemoryLogStepOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.LogMemoryProtos.internal_static_tensorflow_MemoryLogStep_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.LogMemoryProtos.internal_static_tensorflow_MemoryLogStep_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.MemoryLogStep.class, org.tensorflow.proto.framework.MemoryLogStep.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.framework.MemoryLogStep.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      stepId_ = 0L;
+
+      handle_ = "";
+
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.framework.LogMemoryProtos.internal_static_tensorflow_MemoryLogStep_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.MemoryLogStep getDefaultInstanceForType() {
+      return org.tensorflow.proto.framework.MemoryLogStep.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.MemoryLogStep build() {
+      org.tensorflow.proto.framework.MemoryLogStep result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.MemoryLogStep buildPartial() {
+      org.tensorflow.proto.framework.MemoryLogStep result = new org.tensorflow.proto.framework.MemoryLogStep(this);
+      result.stepId_ = stepId_;
+      result.handle_ = handle_;
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.framework.MemoryLogStep) {
+        return mergeFrom((org.tensorflow.proto.framework.MemoryLogStep)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.framework.MemoryLogStep other) {
+      if (other == org.tensorflow.proto.framework.MemoryLogStep.getDefaultInstance()) return this;
+      if (other.getStepId() != 0L) {
+        setStepId(other.getStepId());
+      }
+      if (!other.getHandle().isEmpty()) {
+        handle_ = other.handle_;
+        onChanged();
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.framework.MemoryLogStep parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.framework.MemoryLogStep) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+
+    private long stepId_ ;
+    /**
+     * <pre>
+     * Process-unique step id.
+     * </pre>
+     *
+     * <code>int64 step_id = 1;</code>
+     */
+    public long getStepId() {
+      return stepId_;
+    }
+    /**
+     * <pre>
+     * Process-unique step id.
+     * </pre>
+     *
+     * <code>int64 step_id = 1;</code>
+     */
+    public Builder setStepId(long value) {
+      
+      stepId_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Process-unique step id.
+     * </pre>
+     *
+     * <code>int64 step_id = 1;</code>
+     */
+    public Builder clearStepId() {
+      
+      stepId_ = 0L;
+      onChanged();
+      return this;
+    }
+
+    private java.lang.Object handle_ = "";
+    /**
+     * <pre>
+     * Handle describing the feeds and fetches of the step.
+     * </pre>
+     *
+     * <code>string handle = 2;</code>
+     */
+    public java.lang.String getHandle() {
+      java.lang.Object ref = handle_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        handle_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Handle describing the feeds and fetches of the step.
+     * </pre>
+     *
+     * <code>string handle = 2;</code>
+     */
+    public com.google.protobuf.ByteString
+        getHandleBytes() {
+      java.lang.Object ref = handle_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        handle_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Handle describing the feeds and fetches of the step.
+     * </pre>
+     *
+     * <code>string handle = 2;</code>
+     */
+    public Builder setHandle(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      handle_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Handle describing the feeds and fetches of the step.
+     * </pre>
+     *
+     * <code>string handle = 2;</code>
+     */
+    public Builder clearHandle() {
+      
+      handle_ = getDefaultInstance().getHandle();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Handle describing the feeds and fetches of the step.
+     * </pre>
+     *
+     * <code>string handle = 2;</code>
+     */
+    public Builder setHandleBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      handle_ = value;
+      onChanged();
+      return this;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.MemoryLogStep)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.MemoryLogStep)
+  private static final org.tensorflow.proto.framework.MemoryLogStep DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.MemoryLogStep();
+  }
+
+  public static org.tensorflow.proto.framework.MemoryLogStep getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<MemoryLogStep>
+      PARSER = new com.google.protobuf.AbstractParser<MemoryLogStep>() {
+    @java.lang.Override
+    public MemoryLogStep parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new MemoryLogStep(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<MemoryLogStep> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<MemoryLogStep> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.framework.MemoryLogStep getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/MemoryLogStepOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/MemoryLogStepOrBuilder.java
new file mode 100644
index 00000000000..3e8a13645c1
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/MemoryLogStepOrBuilder.java
@@ -0,0 +1,36 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/framework/log_memory.proto
+
+package org.tensorflow.proto.framework;
+
+public interface MemoryLogStepOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.MemoryLogStep)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <pre>
+   * Process-unique step id.
+   * </pre>
+   *
+   * <code>int64 step_id = 1;</code>
+   */
+  long getStepId();
+
+  /**
+   * <pre>
+   * Handle describing the feeds and fetches of the step.
+   * </pre>
+   *
+   * <code>string handle = 2;</code>
+   */
+  java.lang.String getHandle();
+  /**
+   * <pre>
+   * Handle describing the feeds and fetches of the step.
+   * </pre>
+   *
+   * <code>string handle = 2;</code>
+   */
+  com.google.protobuf.ByteString
+      getHandleBytes();
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/MemoryLogTensorAllocation.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/MemoryLogTensorAllocation.java
new file mode 100644
index 00000000000..bf59aacf6ed
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/MemoryLogTensorAllocation.java
@@ -0,0 +1,884 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/framework/log_memory.proto
+
+package org.tensorflow.proto.framework;
+
+/**
+ * Protobuf type {@code tensorflow.MemoryLogTensorAllocation}
+ */
+public  final class MemoryLogTensorAllocation extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.MemoryLogTensorAllocation)
+    MemoryLogTensorAllocationOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use MemoryLogTensorAllocation.newBuilder() to construct.
+  private MemoryLogTensorAllocation(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private MemoryLogTensorAllocation() {
+    kernelName_ = "";
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new MemoryLogTensorAllocation();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private MemoryLogTensorAllocation(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 8: {
+
+            stepId_ = input.readInt64();
+            break;
+          }
+          case 18: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            kernelName_ = s;
+            break;
+          }
+          case 26: {
+            org.tensorflow.proto.framework.TensorDescription.Builder subBuilder = null;
+            if (tensor_ != null) {
+              subBuilder = tensor_.toBuilder();
+            }
+            tensor_ = input.readMessage(org.tensorflow.proto.framework.TensorDescription.parser(), extensionRegistry);
+            if (subBuilder != null) {
+              subBuilder.mergeFrom(tensor_);
+              tensor_ = subBuilder.buildPartial();
+            }
+
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.framework.LogMemoryProtos.internal_static_tensorflow_MemoryLogTensorAllocation_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.framework.LogMemoryProtos.internal_static_tensorflow_MemoryLogTensorAllocation_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.framework.MemoryLogTensorAllocation.class, org.tensorflow.proto.framework.MemoryLogTensorAllocation.Builder.class);
+  }
+
+  public static final int STEP_ID_FIELD_NUMBER = 1;
+  private long stepId_;
+  /**
+   * <pre>
+   * Process-unique step id.
+   * </pre>
+   *
+   * <code>int64 step_id = 1;</code>
+   */
+  public long getStepId() {
+    return stepId_;
+  }
+
+  public static final int KERNEL_NAME_FIELD_NUMBER = 2;
+  private volatile java.lang.Object kernelName_;
+  /**
+   * <pre>
+   * Name of the kernel making the allocation as set in GraphDef,
+   * e.g., "affine2/weights/Assign".
+   * </pre>
+   *
+   * <code>string kernel_name = 2;</code>
+   */
+  public java.lang.String getKernelName() {
+    java.lang.Object ref = kernelName_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      kernelName_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * Name of the kernel making the allocation as set in GraphDef,
+   * e.g., "affine2/weights/Assign".
+   * </pre>
+   *
+   * <code>string kernel_name = 2;</code>
+   */
+  public com.google.protobuf.ByteString
+      getKernelNameBytes() {
+    java.lang.Object ref = kernelName_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      kernelName_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int TENSOR_FIELD_NUMBER = 3;
+  private org.tensorflow.proto.framework.TensorDescription tensor_;
+  /**
+   * <pre>
+   * Allocated tensor details.
+   * </pre>
+   *
+   * <code>.tensorflow.TensorDescription tensor = 3;</code>
+   */
+  public boolean hasTensor() {
+    return tensor_ != null;
+  }
+  /**
+   * <pre>
+   * Allocated tensor details.
+   * </pre>
+   *
+   * <code>.tensorflow.TensorDescription tensor = 3;</code>
+   */
+  public org.tensorflow.proto.framework.TensorDescription getTensor() {
+    return tensor_ == null ? org.tensorflow.proto.framework.TensorDescription.getDefaultInstance() : tensor_;
+  }
+  /**
+   * <pre>
+   * Allocated tensor details.
+   * </pre>
+   *
+   * <code>.tensorflow.TensorDescription tensor = 3;</code>
+   */
+  public org.tensorflow.proto.framework.TensorDescriptionOrBuilder getTensorOrBuilder() {
+    return getTensor();
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    if (stepId_ != 0L) {
+      output.writeInt64(1, stepId_);
+    }
+    if (!getKernelNameBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 2, kernelName_);
+    }
+    if (tensor_ != null) {
+      output.writeMessage(3, getTensor());
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (stepId_ != 0L) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt64Size(1, stepId_);
+    }
+    if (!getKernelNameBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(2, kernelName_);
+    }
+    if (tensor_ != null) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(3, getTensor());
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.framework.MemoryLogTensorAllocation)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.framework.MemoryLogTensorAllocation other = (org.tensorflow.proto.framework.MemoryLogTensorAllocation) obj;
+
+    if (getStepId()
+        != other.getStepId()) return false;
+    if (!getKernelName()
+        .equals(other.getKernelName())) return false;
+    if (hasTensor() != other.hasTensor()) return false;
+    if (hasTensor()) {
+      if (!getTensor()
+          .equals(other.getTensor())) return false;
+    }
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    hash = (37 * hash) + STEP_ID_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+        getStepId());
+    hash = (37 * hash) + KERNEL_NAME_FIELD_NUMBER;
+    hash = (53 * hash) + getKernelName().hashCode();
+    if (hasTensor()) {
+      hash = (37 * hash) + TENSOR_FIELD_NUMBER;
+      hash = (53 * hash) + getTensor().hashCode();
+    }
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.framework.MemoryLogTensorAllocation parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.MemoryLogTensorAllocation parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.MemoryLogTensorAllocation parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.MemoryLogTensorAllocation parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.MemoryLogTensorAllocation parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.MemoryLogTensorAllocation parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.MemoryLogTensorAllocation parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.MemoryLogTensorAllocation parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.MemoryLogTensorAllocation parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.MemoryLogTensorAllocation parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.MemoryLogTensorAllocation parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.MemoryLogTensorAllocation parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.framework.MemoryLogTensorAllocation prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * Protobuf type {@code tensorflow.MemoryLogTensorAllocation}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.MemoryLogTensorAllocation)
+      org.tensorflow.proto.framework.MemoryLogTensorAllocationOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.LogMemoryProtos.internal_static_tensorflow_MemoryLogTensorAllocation_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.LogMemoryProtos.internal_static_tensorflow_MemoryLogTensorAllocation_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.MemoryLogTensorAllocation.class, org.tensorflow.proto.framework.MemoryLogTensorAllocation.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.framework.MemoryLogTensorAllocation.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      stepId_ = 0L;
+
+      kernelName_ = "";
+
+      if (tensorBuilder_ == null) {
+        tensor_ = null;
+      } else {
+        tensor_ = null;
+        tensorBuilder_ = null;
+      }
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.framework.LogMemoryProtos.internal_static_tensorflow_MemoryLogTensorAllocation_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.MemoryLogTensorAllocation getDefaultInstanceForType() {
+      return org.tensorflow.proto.framework.MemoryLogTensorAllocation.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.MemoryLogTensorAllocation build() {
+      org.tensorflow.proto.framework.MemoryLogTensorAllocation result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.MemoryLogTensorAllocation buildPartial() {
+      org.tensorflow.proto.framework.MemoryLogTensorAllocation result = new org.tensorflow.proto.framework.MemoryLogTensorAllocation(this);
+      result.stepId_ = stepId_;
+      result.kernelName_ = kernelName_;
+      if (tensorBuilder_ == null) {
+        result.tensor_ = tensor_;
+      } else {
+        result.tensor_ = tensorBuilder_.build();
+      }
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.framework.MemoryLogTensorAllocation) {
+        return mergeFrom((org.tensorflow.proto.framework.MemoryLogTensorAllocation)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.framework.MemoryLogTensorAllocation other) {
+      if (other == org.tensorflow.proto.framework.MemoryLogTensorAllocation.getDefaultInstance()) return this;
+      if (other.getStepId() != 0L) {
+        setStepId(other.getStepId());
+      }
+      if (!other.getKernelName().isEmpty()) {
+        kernelName_ = other.kernelName_;
+        onChanged();
+      }
+      if (other.hasTensor()) {
+        mergeTensor(other.getTensor());
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.framework.MemoryLogTensorAllocation parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.framework.MemoryLogTensorAllocation) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+
+    private long stepId_ ;
+    /**
+     * <pre>
+     * Process-unique step id.
+     * </pre>
+     *
+     * <code>int64 step_id = 1;</code>
+     */
+    public long getStepId() {
+      return stepId_;
+    }
+    /**
+     * <pre>
+     * Process-unique step id.
+     * </pre>
+     *
+     * <code>int64 step_id = 1;</code>
+     */
+    public Builder setStepId(long value) {
+      
+      stepId_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Process-unique step id.
+     * </pre>
+     *
+     * <code>int64 step_id = 1;</code>
+     */
+    public Builder clearStepId() {
+      
+      stepId_ = 0L;
+      onChanged();
+      return this;
+    }
+
+    private java.lang.Object kernelName_ = "";
+    /**
+     * <pre>
+     * Name of the kernel making the allocation as set in GraphDef,
+     * e.g., "affine2/weights/Assign".
+     * </pre>
+     *
+     * <code>string kernel_name = 2;</code>
+     */
+    public java.lang.String getKernelName() {
+      java.lang.Object ref = kernelName_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        kernelName_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Name of the kernel making the allocation as set in GraphDef,
+     * e.g., "affine2/weights/Assign".
+     * </pre>
+     *
+     * <code>string kernel_name = 2;</code>
+     */
+    public com.google.protobuf.ByteString
+        getKernelNameBytes() {
+      java.lang.Object ref = kernelName_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        kernelName_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Name of the kernel making the allocation as set in GraphDef,
+     * e.g., "affine2/weights/Assign".
+     * </pre>
+     *
+     * <code>string kernel_name = 2;</code>
+     */
+    public Builder setKernelName(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      kernelName_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Name of the kernel making the allocation as set in GraphDef,
+     * e.g., "affine2/weights/Assign".
+     * </pre>
+     *
+     * <code>string kernel_name = 2;</code>
+     */
+    public Builder clearKernelName() {
+      
+      kernelName_ = getDefaultInstance().getKernelName();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Name of the kernel making the allocation as set in GraphDef,
+     * e.g., "affine2/weights/Assign".
+     * </pre>
+     *
+     * <code>string kernel_name = 2;</code>
+     */
+    public Builder setKernelNameBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      kernelName_ = value;
+      onChanged();
+      return this;
+    }
+
+    private org.tensorflow.proto.framework.TensorDescription tensor_;
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.TensorDescription, org.tensorflow.proto.framework.TensorDescription.Builder, org.tensorflow.proto.framework.TensorDescriptionOrBuilder> tensorBuilder_;
+    /**
+     * <pre>
+     * Allocated tensor details.
+     * </pre>
+     *
+     * <code>.tensorflow.TensorDescription tensor = 3;</code>
+     */
+    public boolean hasTensor() {
+      return tensorBuilder_ != null || tensor_ != null;
+    }
+    /**
+     * <pre>
+     * Allocated tensor details.
+     * </pre>
+     *
+     * <code>.tensorflow.TensorDescription tensor = 3;</code>
+     */
+    public org.tensorflow.proto.framework.TensorDescription getTensor() {
+      if (tensorBuilder_ == null) {
+        return tensor_ == null ? org.tensorflow.proto.framework.TensorDescription.getDefaultInstance() : tensor_;
+      } else {
+        return tensorBuilder_.getMessage();
+      }
+    }
+    /**
+     * <pre>
+     * Allocated tensor details.
+     * </pre>
+     *
+     * <code>.tensorflow.TensorDescription tensor = 3;</code>
+     */
+    public Builder setTensor(org.tensorflow.proto.framework.TensorDescription value) {
+      if (tensorBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        tensor_ = value;
+        onChanged();
+      } else {
+        tensorBuilder_.setMessage(value);
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Allocated tensor details.
+     * </pre>
+     *
+     * <code>.tensorflow.TensorDescription tensor = 3;</code>
+     */
+    public Builder setTensor(
+        org.tensorflow.proto.framework.TensorDescription.Builder builderForValue) {
+      if (tensorBuilder_ == null) {
+        tensor_ = builderForValue.build();
+        onChanged();
+      } else {
+        tensorBuilder_.setMessage(builderForValue.build());
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Allocated tensor details.
+     * </pre>
+     *
+     * <code>.tensorflow.TensorDescription tensor = 3;</code>
+     */
+    public Builder mergeTensor(org.tensorflow.proto.framework.TensorDescription value) {
+      if (tensorBuilder_ == null) {
+        if (tensor_ != null) {
+          tensor_ =
+            org.tensorflow.proto.framework.TensorDescription.newBuilder(tensor_).mergeFrom(value).buildPartial();
+        } else {
+          tensor_ = value;
+        }
+        onChanged();
+      } else {
+        tensorBuilder_.mergeFrom(value);
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Allocated tensor details.
+     * </pre>
+     *
+     * <code>.tensorflow.TensorDescription tensor = 3;</code>
+     */
+    public Builder clearTensor() {
+      if (tensorBuilder_ == null) {
+        tensor_ = null;
+        onChanged();
+      } else {
+        tensor_ = null;
+        tensorBuilder_ = null;
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Allocated tensor details.
+     * </pre>
+     *
+     * <code>.tensorflow.TensorDescription tensor = 3;</code>
+     */
+    public org.tensorflow.proto.framework.TensorDescription.Builder getTensorBuilder() {
+      
+      onChanged();
+      return getTensorFieldBuilder().getBuilder();
+    }
+    /**
+     * <pre>
+     * Allocated tensor details.
+     * </pre>
+     *
+     * <code>.tensorflow.TensorDescription tensor = 3;</code>
+     */
+    public org.tensorflow.proto.framework.TensorDescriptionOrBuilder getTensorOrBuilder() {
+      if (tensorBuilder_ != null) {
+        return tensorBuilder_.getMessageOrBuilder();
+      } else {
+        return tensor_ == null ?
+            org.tensorflow.proto.framework.TensorDescription.getDefaultInstance() : tensor_;
+      }
+    }
+    /**
+     * <pre>
+     * Allocated tensor details.
+     * </pre>
+     *
+     * <code>.tensorflow.TensorDescription tensor = 3;</code>
+     */
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.TensorDescription, org.tensorflow.proto.framework.TensorDescription.Builder, org.tensorflow.proto.framework.TensorDescriptionOrBuilder> 
+        getTensorFieldBuilder() {
+      if (tensorBuilder_ == null) {
+        tensorBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+            org.tensorflow.proto.framework.TensorDescription, org.tensorflow.proto.framework.TensorDescription.Builder, org.tensorflow.proto.framework.TensorDescriptionOrBuilder>(
+                getTensor(),
+                getParentForChildren(),
+                isClean());
+        tensor_ = null;
+      }
+      return tensorBuilder_;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.MemoryLogTensorAllocation)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.MemoryLogTensorAllocation)
+  private static final org.tensorflow.proto.framework.MemoryLogTensorAllocation DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.MemoryLogTensorAllocation();
+  }
+
+  public static org.tensorflow.proto.framework.MemoryLogTensorAllocation getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<MemoryLogTensorAllocation>
+      PARSER = new com.google.protobuf.AbstractParser<MemoryLogTensorAllocation>() {
+    @java.lang.Override
+    public MemoryLogTensorAllocation parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new MemoryLogTensorAllocation(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<MemoryLogTensorAllocation> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<MemoryLogTensorAllocation> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.framework.MemoryLogTensorAllocation getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/MemoryLogTensorAllocationOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/MemoryLogTensorAllocationOrBuilder.java
new file mode 100644
index 00000000000..1b3a3732149
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/MemoryLogTensorAllocationOrBuilder.java
@@ -0,0 +1,63 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/framework/log_memory.proto
+
+package org.tensorflow.proto.framework;
+
+public interface MemoryLogTensorAllocationOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.MemoryLogTensorAllocation)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <pre>
+   * Process-unique step id.
+   * </pre>
+   *
+   * <code>int64 step_id = 1;</code>
+   */
+  long getStepId();
+
+  /**
+   * <pre>
+   * Name of the kernel making the allocation as set in GraphDef,
+   * e.g., "affine2/weights/Assign".
+   * </pre>
+   *
+   * <code>string kernel_name = 2;</code>
+   */
+  java.lang.String getKernelName();
+  /**
+   * <pre>
+   * Name of the kernel making the allocation as set in GraphDef,
+   * e.g., "affine2/weights/Assign".
+   * </pre>
+   *
+   * <code>string kernel_name = 2;</code>
+   */
+  com.google.protobuf.ByteString
+      getKernelNameBytes();
+
+  /**
+   * <pre>
+   * Allocated tensor details.
+   * </pre>
+   *
+   * <code>.tensorflow.TensorDescription tensor = 3;</code>
+   */
+  boolean hasTensor();
+  /**
+   * <pre>
+   * Allocated tensor details.
+   * </pre>
+   *
+   * <code>.tensorflow.TensorDescription tensor = 3;</code>
+   */
+  org.tensorflow.proto.framework.TensorDescription getTensor();
+  /**
+   * <pre>
+   * Allocated tensor details.
+   * </pre>
+   *
+   * <code>.tensorflow.TensorDescription tensor = 3;</code>
+   */
+  org.tensorflow.proto.framework.TensorDescriptionOrBuilder getTensorOrBuilder();
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/MemoryLogTensorDeallocation.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/MemoryLogTensorDeallocation.java
new file mode 100644
index 00000000000..f0061b53c50
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/MemoryLogTensorDeallocation.java
@@ -0,0 +1,652 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/framework/log_memory.proto
+
+package org.tensorflow.proto.framework;
+
+/**
+ * Protobuf type {@code tensorflow.MemoryLogTensorDeallocation}
+ */
+public  final class MemoryLogTensorDeallocation extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.MemoryLogTensorDeallocation)
+    MemoryLogTensorDeallocationOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use MemoryLogTensorDeallocation.newBuilder() to construct.
+  private MemoryLogTensorDeallocation(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private MemoryLogTensorDeallocation() {
+    allocatorName_ = "";
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new MemoryLogTensorDeallocation();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private MemoryLogTensorDeallocation(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 8: {
+
+            allocationId_ = input.readInt64();
+            break;
+          }
+          case 18: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            allocatorName_ = s;
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.framework.LogMemoryProtos.internal_static_tensorflow_MemoryLogTensorDeallocation_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.framework.LogMemoryProtos.internal_static_tensorflow_MemoryLogTensorDeallocation_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.framework.MemoryLogTensorDeallocation.class, org.tensorflow.proto.framework.MemoryLogTensorDeallocation.Builder.class);
+  }
+
+  public static final int ALLOCATION_ID_FIELD_NUMBER = 1;
+  private long allocationId_;
+  /**
+   * <pre>
+   * Id of the tensor buffer being deallocated, used to match to a
+   * corresponding allocation.
+   * </pre>
+   *
+   * <code>int64 allocation_id = 1;</code>
+   */
+  public long getAllocationId() {
+    return allocationId_;
+  }
+
+  public static final int ALLOCATOR_NAME_FIELD_NUMBER = 2;
+  private volatile java.lang.Object allocatorName_;
+  /**
+   * <pre>
+   * Name of the allocator used.
+   * </pre>
+   *
+   * <code>string allocator_name = 2;</code>
+   */
+  public java.lang.String getAllocatorName() {
+    java.lang.Object ref = allocatorName_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      allocatorName_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * Name of the allocator used.
+   * </pre>
+   *
+   * <code>string allocator_name = 2;</code>
+   */
+  public com.google.protobuf.ByteString
+      getAllocatorNameBytes() {
+    java.lang.Object ref = allocatorName_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      allocatorName_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    if (allocationId_ != 0L) {
+      output.writeInt64(1, allocationId_);
+    }
+    if (!getAllocatorNameBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 2, allocatorName_);
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (allocationId_ != 0L) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt64Size(1, allocationId_);
+    }
+    if (!getAllocatorNameBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(2, allocatorName_);
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.framework.MemoryLogTensorDeallocation)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.framework.MemoryLogTensorDeallocation other = (org.tensorflow.proto.framework.MemoryLogTensorDeallocation) obj;
+
+    if (getAllocationId()
+        != other.getAllocationId()) return false;
+    if (!getAllocatorName()
+        .equals(other.getAllocatorName())) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    hash = (37 * hash) + ALLOCATION_ID_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+        getAllocationId());
+    hash = (37 * hash) + ALLOCATOR_NAME_FIELD_NUMBER;
+    hash = (53 * hash) + getAllocatorName().hashCode();
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.framework.MemoryLogTensorDeallocation parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.MemoryLogTensorDeallocation parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.MemoryLogTensorDeallocation parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.MemoryLogTensorDeallocation parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.MemoryLogTensorDeallocation parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.MemoryLogTensorDeallocation parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.MemoryLogTensorDeallocation parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.MemoryLogTensorDeallocation parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.MemoryLogTensorDeallocation parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.MemoryLogTensorDeallocation parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.MemoryLogTensorDeallocation parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.MemoryLogTensorDeallocation parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.framework.MemoryLogTensorDeallocation prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * Protobuf type {@code tensorflow.MemoryLogTensorDeallocation}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.MemoryLogTensorDeallocation)
+      org.tensorflow.proto.framework.MemoryLogTensorDeallocationOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.LogMemoryProtos.internal_static_tensorflow_MemoryLogTensorDeallocation_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.LogMemoryProtos.internal_static_tensorflow_MemoryLogTensorDeallocation_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.MemoryLogTensorDeallocation.class, org.tensorflow.proto.framework.MemoryLogTensorDeallocation.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.framework.MemoryLogTensorDeallocation.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      allocationId_ = 0L;
+
+      allocatorName_ = "";
+
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.framework.LogMemoryProtos.internal_static_tensorflow_MemoryLogTensorDeallocation_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.MemoryLogTensorDeallocation getDefaultInstanceForType() {
+      return org.tensorflow.proto.framework.MemoryLogTensorDeallocation.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.MemoryLogTensorDeallocation build() {
+      org.tensorflow.proto.framework.MemoryLogTensorDeallocation result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.MemoryLogTensorDeallocation buildPartial() {
+      org.tensorflow.proto.framework.MemoryLogTensorDeallocation result = new org.tensorflow.proto.framework.MemoryLogTensorDeallocation(this);
+      result.allocationId_ = allocationId_;
+      result.allocatorName_ = allocatorName_;
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.framework.MemoryLogTensorDeallocation) {
+        return mergeFrom((org.tensorflow.proto.framework.MemoryLogTensorDeallocation)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.framework.MemoryLogTensorDeallocation other) {
+      if (other == org.tensorflow.proto.framework.MemoryLogTensorDeallocation.getDefaultInstance()) return this;
+      if (other.getAllocationId() != 0L) {
+        setAllocationId(other.getAllocationId());
+      }
+      if (!other.getAllocatorName().isEmpty()) {
+        allocatorName_ = other.allocatorName_;
+        onChanged();
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.framework.MemoryLogTensorDeallocation parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.framework.MemoryLogTensorDeallocation) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+
+    private long allocationId_ ;
+    /**
+     * <pre>
+     * Id of the tensor buffer being deallocated, used to match to a
+     * corresponding allocation.
+     * </pre>
+     *
+     * <code>int64 allocation_id = 1;</code>
+     */
+    public long getAllocationId() {
+      return allocationId_;
+    }
+    /**
+     * <pre>
+     * Id of the tensor buffer being deallocated, used to match to a
+     * corresponding allocation.
+     * </pre>
+     *
+     * <code>int64 allocation_id = 1;</code>
+     */
+    public Builder setAllocationId(long value) {
+      
+      allocationId_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Id of the tensor buffer being deallocated, used to match to a
+     * corresponding allocation.
+     * </pre>
+     *
+     * <code>int64 allocation_id = 1;</code>
+     */
+    public Builder clearAllocationId() {
+      
+      allocationId_ = 0L;
+      onChanged();
+      return this;
+    }
+
+    private java.lang.Object allocatorName_ = "";
+    /**
+     * <pre>
+     * Name of the allocator used.
+     * </pre>
+     *
+     * <code>string allocator_name = 2;</code>
+     */
+    public java.lang.String getAllocatorName() {
+      java.lang.Object ref = allocatorName_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        allocatorName_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Name of the allocator used.
+     * </pre>
+     *
+     * <code>string allocator_name = 2;</code>
+     */
+    public com.google.protobuf.ByteString
+        getAllocatorNameBytes() {
+      java.lang.Object ref = allocatorName_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        allocatorName_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Name of the allocator used.
+     * </pre>
+     *
+     * <code>string allocator_name = 2;</code>
+     */
+    public Builder setAllocatorName(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      allocatorName_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Name of the allocator used.
+     * </pre>
+     *
+     * <code>string allocator_name = 2;</code>
+     */
+    public Builder clearAllocatorName() {
+      
+      allocatorName_ = getDefaultInstance().getAllocatorName();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Name of the allocator used.
+     * </pre>
+     *
+     * <code>string allocator_name = 2;</code>
+     */
+    public Builder setAllocatorNameBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      allocatorName_ = value;
+      onChanged();
+      return this;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.MemoryLogTensorDeallocation)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.MemoryLogTensorDeallocation)
+  private static final org.tensorflow.proto.framework.MemoryLogTensorDeallocation DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.MemoryLogTensorDeallocation();
+  }
+
+  public static org.tensorflow.proto.framework.MemoryLogTensorDeallocation getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<MemoryLogTensorDeallocation>
+      PARSER = new com.google.protobuf.AbstractParser<MemoryLogTensorDeallocation>() {
+    @java.lang.Override
+    public MemoryLogTensorDeallocation parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new MemoryLogTensorDeallocation(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<MemoryLogTensorDeallocation> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<MemoryLogTensorDeallocation> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.framework.MemoryLogTensorDeallocation getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/MemoryLogTensorDeallocationOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/MemoryLogTensorDeallocationOrBuilder.java
new file mode 100644
index 00000000000..7d45248a17a
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/MemoryLogTensorDeallocationOrBuilder.java
@@ -0,0 +1,37 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/framework/log_memory.proto
+
+package org.tensorflow.proto.framework;
+
+public interface MemoryLogTensorDeallocationOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.MemoryLogTensorDeallocation)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <pre>
+   * Id of the tensor buffer being deallocated, used to match to a
+   * corresponding allocation.
+   * </pre>
+   *
+   * <code>int64 allocation_id = 1;</code>
+   */
+  long getAllocationId();
+
+  /**
+   * <pre>
+   * Name of the allocator used.
+   * </pre>
+   *
+   * <code>string allocator_name = 2;</code>
+   */
+  java.lang.String getAllocatorName();
+  /**
+   * <pre>
+   * Name of the allocator used.
+   * </pre>
+   *
+   * <code>string allocator_name = 2;</code>
+   */
+  com.google.protobuf.ByteString
+      getAllocatorNameBytes();
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/MemoryLogTensorOutput.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/MemoryLogTensorOutput.java
new file mode 100644
index 00000000000..a8aaf7dc7d9
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/MemoryLogTensorOutput.java
@@ -0,0 +1,957 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/framework/log_memory.proto
+
+package org.tensorflow.proto.framework;
+
+/**
+ * Protobuf type {@code tensorflow.MemoryLogTensorOutput}
+ */
+public  final class MemoryLogTensorOutput extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.MemoryLogTensorOutput)
+    MemoryLogTensorOutputOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use MemoryLogTensorOutput.newBuilder() to construct.
+  private MemoryLogTensorOutput(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private MemoryLogTensorOutput() {
+    kernelName_ = "";
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new MemoryLogTensorOutput();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private MemoryLogTensorOutput(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 8: {
+
+            stepId_ = input.readInt64();
+            break;
+          }
+          case 18: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            kernelName_ = s;
+            break;
+          }
+          case 24: {
+
+            index_ = input.readInt32();
+            break;
+          }
+          case 34: {
+            org.tensorflow.proto.framework.TensorDescription.Builder subBuilder = null;
+            if (tensor_ != null) {
+              subBuilder = tensor_.toBuilder();
+            }
+            tensor_ = input.readMessage(org.tensorflow.proto.framework.TensorDescription.parser(), extensionRegistry);
+            if (subBuilder != null) {
+              subBuilder.mergeFrom(tensor_);
+              tensor_ = subBuilder.buildPartial();
+            }
+
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.framework.LogMemoryProtos.internal_static_tensorflow_MemoryLogTensorOutput_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.framework.LogMemoryProtos.internal_static_tensorflow_MemoryLogTensorOutput_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.framework.MemoryLogTensorOutput.class, org.tensorflow.proto.framework.MemoryLogTensorOutput.Builder.class);
+  }
+
+  public static final int STEP_ID_FIELD_NUMBER = 1;
+  private long stepId_;
+  /**
+   * <pre>
+   * Process-unique step id.
+   * </pre>
+   *
+   * <code>int64 step_id = 1;</code>
+   */
+  public long getStepId() {
+    return stepId_;
+  }
+
+  public static final int KERNEL_NAME_FIELD_NUMBER = 2;
+  private volatile java.lang.Object kernelName_;
+  /**
+   * <pre>
+   * Name of the kernel producing an output as set in GraphDef, e.g.,
+   * "affine2/weights/Assign".
+   * </pre>
+   *
+   * <code>string kernel_name = 2;</code>
+   */
+  public java.lang.String getKernelName() {
+    java.lang.Object ref = kernelName_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      kernelName_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * Name of the kernel producing an output as set in GraphDef, e.g.,
+   * "affine2/weights/Assign".
+   * </pre>
+   *
+   * <code>string kernel_name = 2;</code>
+   */
+  public com.google.protobuf.ByteString
+      getKernelNameBytes() {
+    java.lang.Object ref = kernelName_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      kernelName_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int INDEX_FIELD_NUMBER = 3;
+  private int index_;
+  /**
+   * <pre>
+   * Index of the output being set.
+   * </pre>
+   *
+   * <code>int32 index = 3;</code>
+   */
+  public int getIndex() {
+    return index_;
+  }
+
+  public static final int TENSOR_FIELD_NUMBER = 4;
+  private org.tensorflow.proto.framework.TensorDescription tensor_;
+  /**
+   * <pre>
+   * Output tensor details.
+   * </pre>
+   *
+   * <code>.tensorflow.TensorDescription tensor = 4;</code>
+   */
+  public boolean hasTensor() {
+    return tensor_ != null;
+  }
+  /**
+   * <pre>
+   * Output tensor details.
+   * </pre>
+   *
+   * <code>.tensorflow.TensorDescription tensor = 4;</code>
+   */
+  public org.tensorflow.proto.framework.TensorDescription getTensor() {
+    return tensor_ == null ? org.tensorflow.proto.framework.TensorDescription.getDefaultInstance() : tensor_;
+  }
+  /**
+   * <pre>
+   * Output tensor details.
+   * </pre>
+   *
+   * <code>.tensorflow.TensorDescription tensor = 4;</code>
+   */
+  public org.tensorflow.proto.framework.TensorDescriptionOrBuilder getTensorOrBuilder() {
+    return getTensor();
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    if (stepId_ != 0L) {
+      output.writeInt64(1, stepId_);
+    }
+    if (!getKernelNameBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 2, kernelName_);
+    }
+    if (index_ != 0) {
+      output.writeInt32(3, index_);
+    }
+    if (tensor_ != null) {
+      output.writeMessage(4, getTensor());
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (stepId_ != 0L) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt64Size(1, stepId_);
+    }
+    if (!getKernelNameBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(2, kernelName_);
+    }
+    if (index_ != 0) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt32Size(3, index_);
+    }
+    if (tensor_ != null) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(4, getTensor());
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.framework.MemoryLogTensorOutput)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.framework.MemoryLogTensorOutput other = (org.tensorflow.proto.framework.MemoryLogTensorOutput) obj;
+
+    if (getStepId()
+        != other.getStepId()) return false;
+    if (!getKernelName()
+        .equals(other.getKernelName())) return false;
+    if (getIndex()
+        != other.getIndex()) return false;
+    if (hasTensor() != other.hasTensor()) return false;
+    if (hasTensor()) {
+      if (!getTensor()
+          .equals(other.getTensor())) return false;
+    }
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    hash = (37 * hash) + STEP_ID_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+        getStepId());
+    hash = (37 * hash) + KERNEL_NAME_FIELD_NUMBER;
+    hash = (53 * hash) + getKernelName().hashCode();
+    hash = (37 * hash) + INDEX_FIELD_NUMBER;
+    hash = (53 * hash) + getIndex();
+    if (hasTensor()) {
+      hash = (37 * hash) + TENSOR_FIELD_NUMBER;
+      hash = (53 * hash) + getTensor().hashCode();
+    }
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.framework.MemoryLogTensorOutput parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.MemoryLogTensorOutput parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.MemoryLogTensorOutput parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.MemoryLogTensorOutput parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.MemoryLogTensorOutput parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.MemoryLogTensorOutput parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.MemoryLogTensorOutput parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.MemoryLogTensorOutput parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.MemoryLogTensorOutput parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.MemoryLogTensorOutput parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.MemoryLogTensorOutput parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.MemoryLogTensorOutput parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.framework.MemoryLogTensorOutput prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * Protobuf type {@code tensorflow.MemoryLogTensorOutput}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.MemoryLogTensorOutput)
+      org.tensorflow.proto.framework.MemoryLogTensorOutputOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.LogMemoryProtos.internal_static_tensorflow_MemoryLogTensorOutput_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.LogMemoryProtos.internal_static_tensorflow_MemoryLogTensorOutput_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.MemoryLogTensorOutput.class, org.tensorflow.proto.framework.MemoryLogTensorOutput.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.framework.MemoryLogTensorOutput.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      stepId_ = 0L;
+
+      kernelName_ = "";
+
+      index_ = 0;
+
+      if (tensorBuilder_ == null) {
+        tensor_ = null;
+      } else {
+        tensor_ = null;
+        tensorBuilder_ = null;
+      }
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.framework.LogMemoryProtos.internal_static_tensorflow_MemoryLogTensorOutput_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.MemoryLogTensorOutput getDefaultInstanceForType() {
+      return org.tensorflow.proto.framework.MemoryLogTensorOutput.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.MemoryLogTensorOutput build() {
+      org.tensorflow.proto.framework.MemoryLogTensorOutput result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.MemoryLogTensorOutput buildPartial() {
+      org.tensorflow.proto.framework.MemoryLogTensorOutput result = new org.tensorflow.proto.framework.MemoryLogTensorOutput(this);
+      result.stepId_ = stepId_;
+      result.kernelName_ = kernelName_;
+      result.index_ = index_;
+      if (tensorBuilder_ == null) {
+        result.tensor_ = tensor_;
+      } else {
+        result.tensor_ = tensorBuilder_.build();
+      }
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.framework.MemoryLogTensorOutput) {
+        return mergeFrom((org.tensorflow.proto.framework.MemoryLogTensorOutput)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.framework.MemoryLogTensorOutput other) {
+      if (other == org.tensorflow.proto.framework.MemoryLogTensorOutput.getDefaultInstance()) return this;
+      if (other.getStepId() != 0L) {
+        setStepId(other.getStepId());
+      }
+      if (!other.getKernelName().isEmpty()) {
+        kernelName_ = other.kernelName_;
+        onChanged();
+      }
+      if (other.getIndex() != 0) {
+        setIndex(other.getIndex());
+      }
+      if (other.hasTensor()) {
+        mergeTensor(other.getTensor());
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.framework.MemoryLogTensorOutput parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.framework.MemoryLogTensorOutput) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+
+    private long stepId_ ;
+    /**
+     * <pre>
+     * Process-unique step id.
+     * </pre>
+     *
+     * <code>int64 step_id = 1;</code>
+     */
+    public long getStepId() {
+      return stepId_;
+    }
+    /**
+     * <pre>
+     * Process-unique step id.
+     * </pre>
+     *
+     * <code>int64 step_id = 1;</code>
+     */
+    public Builder setStepId(long value) {
+      
+      stepId_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Process-unique step id.
+     * </pre>
+     *
+     * <code>int64 step_id = 1;</code>
+     */
+    public Builder clearStepId() {
+      
+      stepId_ = 0L;
+      onChanged();
+      return this;
+    }
+
+    private java.lang.Object kernelName_ = "";
+    /**
+     * <pre>
+     * Name of the kernel producing an output as set in GraphDef, e.g.,
+     * "affine2/weights/Assign".
+     * </pre>
+     *
+     * <code>string kernel_name = 2;</code>
+     */
+    public java.lang.String getKernelName() {
+      java.lang.Object ref = kernelName_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        kernelName_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Name of the kernel producing an output as set in GraphDef, e.g.,
+     * "affine2/weights/Assign".
+     * </pre>
+     *
+     * <code>string kernel_name = 2;</code>
+     */
+    public com.google.protobuf.ByteString
+        getKernelNameBytes() {
+      java.lang.Object ref = kernelName_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        kernelName_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Name of the kernel producing an output as set in GraphDef, e.g.,
+     * "affine2/weights/Assign".
+     * </pre>
+     *
+     * <code>string kernel_name = 2;</code>
+     */
+    public Builder setKernelName(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      kernelName_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Name of the kernel producing an output as set in GraphDef, e.g.,
+     * "affine2/weights/Assign".
+     * </pre>
+     *
+     * <code>string kernel_name = 2;</code>
+     */
+    public Builder clearKernelName() {
+      
+      kernelName_ = getDefaultInstance().getKernelName();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Name of the kernel producing an output as set in GraphDef, e.g.,
+     * "affine2/weights/Assign".
+     * </pre>
+     *
+     * <code>string kernel_name = 2;</code>
+     */
+    public Builder setKernelNameBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      kernelName_ = value;
+      onChanged();
+      return this;
+    }
+
+    private int index_ ;
+    /**
+     * <pre>
+     * Index of the output being set.
+     * </pre>
+     *
+     * <code>int32 index = 3;</code>
+     */
+    public int getIndex() {
+      return index_;
+    }
+    /**
+     * <pre>
+     * Index of the output being set.
+     * </pre>
+     *
+     * <code>int32 index = 3;</code>
+     */
+    public Builder setIndex(int value) {
+      
+      index_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Index of the output being set.
+     * </pre>
+     *
+     * <code>int32 index = 3;</code>
+     */
+    public Builder clearIndex() {
+      
+      index_ = 0;
+      onChanged();
+      return this;
+    }
+
+    private org.tensorflow.proto.framework.TensorDescription tensor_;
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.TensorDescription, org.tensorflow.proto.framework.TensorDescription.Builder, org.tensorflow.proto.framework.TensorDescriptionOrBuilder> tensorBuilder_;
+    /**
+     * <pre>
+     * Output tensor details.
+     * </pre>
+     *
+     * <code>.tensorflow.TensorDescription tensor = 4;</code>
+     */
+    public boolean hasTensor() {
+      return tensorBuilder_ != null || tensor_ != null;
+    }
+    /**
+     * <pre>
+     * Output tensor details.
+     * </pre>
+     *
+     * <code>.tensorflow.TensorDescription tensor = 4;</code>
+     */
+    public org.tensorflow.proto.framework.TensorDescription getTensor() {
+      if (tensorBuilder_ == null) {
+        return tensor_ == null ? org.tensorflow.proto.framework.TensorDescription.getDefaultInstance() : tensor_;
+      } else {
+        return tensorBuilder_.getMessage();
+      }
+    }
+    /**
+     * <pre>
+     * Output tensor details.
+     * </pre>
+     *
+     * <code>.tensorflow.TensorDescription tensor = 4;</code>
+     */
+    public Builder setTensor(org.tensorflow.proto.framework.TensorDescription value) {
+      if (tensorBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        tensor_ = value;
+        onChanged();
+      } else {
+        tensorBuilder_.setMessage(value);
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Output tensor details.
+     * </pre>
+     *
+     * <code>.tensorflow.TensorDescription tensor = 4;</code>
+     */
+    public Builder setTensor(
+        org.tensorflow.proto.framework.TensorDescription.Builder builderForValue) {
+      if (tensorBuilder_ == null) {
+        tensor_ = builderForValue.build();
+        onChanged();
+      } else {
+        tensorBuilder_.setMessage(builderForValue.build());
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Output tensor details.
+     * </pre>
+     *
+     * <code>.tensorflow.TensorDescription tensor = 4;</code>
+     */
+    public Builder mergeTensor(org.tensorflow.proto.framework.TensorDescription value) {
+      if (tensorBuilder_ == null) {
+        if (tensor_ != null) {
+          tensor_ =
+            org.tensorflow.proto.framework.TensorDescription.newBuilder(tensor_).mergeFrom(value).buildPartial();
+        } else {
+          tensor_ = value;
+        }
+        onChanged();
+      } else {
+        tensorBuilder_.mergeFrom(value);
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Output tensor details.
+     * </pre>
+     *
+     * <code>.tensorflow.TensorDescription tensor = 4;</code>
+     */
+    public Builder clearTensor() {
+      if (tensorBuilder_ == null) {
+        tensor_ = null;
+        onChanged();
+      } else {
+        tensor_ = null;
+        tensorBuilder_ = null;
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Output tensor details.
+     * </pre>
+     *
+     * <code>.tensorflow.TensorDescription tensor = 4;</code>
+     */
+    public org.tensorflow.proto.framework.TensorDescription.Builder getTensorBuilder() {
+      
+      onChanged();
+      return getTensorFieldBuilder().getBuilder();
+    }
+    /**
+     * <pre>
+     * Output tensor details.
+     * </pre>
+     *
+     * <code>.tensorflow.TensorDescription tensor = 4;</code>
+     */
+    public org.tensorflow.proto.framework.TensorDescriptionOrBuilder getTensorOrBuilder() {
+      if (tensorBuilder_ != null) {
+        return tensorBuilder_.getMessageOrBuilder();
+      } else {
+        return tensor_ == null ?
+            org.tensorflow.proto.framework.TensorDescription.getDefaultInstance() : tensor_;
+      }
+    }
+    /**
+     * <pre>
+     * Output tensor details.
+     * </pre>
+     *
+     * <code>.tensorflow.TensorDescription tensor = 4;</code>
+     */
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.TensorDescription, org.tensorflow.proto.framework.TensorDescription.Builder, org.tensorflow.proto.framework.TensorDescriptionOrBuilder> 
+        getTensorFieldBuilder() {
+      if (tensorBuilder_ == null) {
+        tensorBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+            org.tensorflow.proto.framework.TensorDescription, org.tensorflow.proto.framework.TensorDescription.Builder, org.tensorflow.proto.framework.TensorDescriptionOrBuilder>(
+                getTensor(),
+                getParentForChildren(),
+                isClean());
+        tensor_ = null;
+      }
+      return tensorBuilder_;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.MemoryLogTensorOutput)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.MemoryLogTensorOutput)
+  private static final org.tensorflow.proto.framework.MemoryLogTensorOutput DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.MemoryLogTensorOutput();
+  }
+
+  public static org.tensorflow.proto.framework.MemoryLogTensorOutput getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<MemoryLogTensorOutput>
+      PARSER = new com.google.protobuf.AbstractParser<MemoryLogTensorOutput>() {
+    @java.lang.Override
+    public MemoryLogTensorOutput parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new MemoryLogTensorOutput(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<MemoryLogTensorOutput> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<MemoryLogTensorOutput> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.framework.MemoryLogTensorOutput getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/MemoryLogTensorOutputOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/MemoryLogTensorOutputOrBuilder.java
new file mode 100644
index 00000000000..b9275e22f7e
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/MemoryLogTensorOutputOrBuilder.java
@@ -0,0 +1,72 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/framework/log_memory.proto
+
+package org.tensorflow.proto.framework;
+
+public interface MemoryLogTensorOutputOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.MemoryLogTensorOutput)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <pre>
+   * Process-unique step id.
+   * </pre>
+   *
+   * <code>int64 step_id = 1;</code>
+   */
+  long getStepId();
+
+  /**
+   * <pre>
+   * Name of the kernel producing an output as set in GraphDef, e.g.,
+   * "affine2/weights/Assign".
+   * </pre>
+   *
+   * <code>string kernel_name = 2;</code>
+   */
+  java.lang.String getKernelName();
+  /**
+   * <pre>
+   * Name of the kernel producing an output as set in GraphDef, e.g.,
+   * "affine2/weights/Assign".
+   * </pre>
+   *
+   * <code>string kernel_name = 2;</code>
+   */
+  com.google.protobuf.ByteString
+      getKernelNameBytes();
+
+  /**
+   * <pre>
+   * Index of the output being set.
+   * </pre>
+   *
+   * <code>int32 index = 3;</code>
+   */
+  int getIndex();
+
+  /**
+   * <pre>
+   * Output tensor details.
+   * </pre>
+   *
+   * <code>.tensorflow.TensorDescription tensor = 4;</code>
+   */
+  boolean hasTensor();
+  /**
+   * <pre>
+   * Output tensor details.
+   * </pre>
+   *
+   * <code>.tensorflow.TensorDescription tensor = 4;</code>
+   */
+  org.tensorflow.proto.framework.TensorDescription getTensor();
+  /**
+   * <pre>
+   * Output tensor details.
+   * </pre>
+   *
+   * <code>.tensorflow.TensorDescription tensor = 4;</code>
+   */
+  org.tensorflow.proto.framework.TensorDescriptionOrBuilder getTensorOrBuilder();
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/MemoryStats.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/MemoryStats.java
new file mode 100644
index 00000000000..e351d0be2b0
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/MemoryStats.java
@@ -0,0 +1,981 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/framework/step_stats.proto
+
+package org.tensorflow.proto.framework;
+
+/**
+ * <pre>
+ * For memory tracking.
+ * </pre>
+ *
+ * Protobuf type {@code tensorflow.MemoryStats}
+ */
+public  final class MemoryStats extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.MemoryStats)
+    MemoryStatsOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use MemoryStats.newBuilder() to construct.
+  private MemoryStats(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private MemoryStats() {
+    persistentTensorAllocIds_ = emptyLongList();
+    devicePersistentTensorAllocIds_ = emptyLongList();
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new MemoryStats();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private MemoryStats(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    int mutable_bitField0_ = 0;
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 8: {
+
+            tempMemorySize_ = input.readInt64();
+            break;
+          }
+          case 16: {
+
+            deviceTempMemorySize_ = input.readInt64();
+            break;
+          }
+          case 24: {
+
+            persistentMemorySize_ = input.readInt64();
+            break;
+          }
+          case 32: {
+
+            devicePersistentMemorySize_ = input.readInt64();
+            break;
+          }
+          case 40: {
+            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
+              persistentTensorAllocIds_ = newLongList();
+              mutable_bitField0_ |= 0x00000001;
+            }
+            persistentTensorAllocIds_.addLong(input.readInt64());
+            break;
+          }
+          case 42: {
+            int length = input.readRawVarint32();
+            int limit = input.pushLimit(length);
+            if (!((mutable_bitField0_ & 0x00000001) != 0) && input.getBytesUntilLimit() > 0) {
+              persistentTensorAllocIds_ = newLongList();
+              mutable_bitField0_ |= 0x00000001;
+            }
+            while (input.getBytesUntilLimit() > 0) {
+              persistentTensorAllocIds_.addLong(input.readInt64());
+            }
+            input.popLimit(limit);
+            break;
+          }
+          case 48: {
+            if (!((mutable_bitField0_ & 0x00000002) != 0)) {
+              devicePersistentTensorAllocIds_ = newLongList();
+              mutable_bitField0_ |= 0x00000002;
+            }
+            devicePersistentTensorAllocIds_.addLong(input.readInt64());
+            break;
+          }
+          case 50: {
+            int length = input.readRawVarint32();
+            int limit = input.pushLimit(length);
+            if (!((mutable_bitField0_ & 0x00000002) != 0) && input.getBytesUntilLimit() > 0) {
+              devicePersistentTensorAllocIds_ = newLongList();
+              mutable_bitField0_ |= 0x00000002;
+            }
+            while (input.getBytesUntilLimit() > 0) {
+              devicePersistentTensorAllocIds_.addLong(input.readInt64());
+            }
+            input.popLimit(limit);
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      if (((mutable_bitField0_ & 0x00000001) != 0)) {
+        persistentTensorAllocIds_.makeImmutable(); // C
+      }
+      if (((mutable_bitField0_ & 0x00000002) != 0)) {
+        devicePersistentTensorAllocIds_.makeImmutable(); // C
+      }
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.framework.StepStatsProtos.internal_static_tensorflow_MemoryStats_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.framework.StepStatsProtos.internal_static_tensorflow_MemoryStats_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.framework.MemoryStats.class, org.tensorflow.proto.framework.MemoryStats.Builder.class);
+  }
+
+  public static final int TEMP_MEMORY_SIZE_FIELD_NUMBER = 1;
+  private long tempMemorySize_;
+  /**
+   * <code>int64 temp_memory_size = 1;</code>
+   */
+  public long getTempMemorySize() {
+    return tempMemorySize_;
+  }
+
+  public static final int PERSISTENT_MEMORY_SIZE_FIELD_NUMBER = 3;
+  private long persistentMemorySize_;
+  /**
+   * <code>int64 persistent_memory_size = 3;</code>
+   */
+  public long getPersistentMemorySize() {
+    return persistentMemorySize_;
+  }
+
+  public static final int PERSISTENT_TENSOR_ALLOC_IDS_FIELD_NUMBER = 5;
+  private com.google.protobuf.Internal.LongList persistentTensorAllocIds_;
+  /**
+   * <code>repeated int64 persistent_tensor_alloc_ids = 5;</code>
+   */
+  public java.util.List<java.lang.Long>
+      getPersistentTensorAllocIdsList() {
+    return persistentTensorAllocIds_;
+  }
+  /**
+   * <code>repeated int64 persistent_tensor_alloc_ids = 5;</code>
+   */
+  public int getPersistentTensorAllocIdsCount() {
+    return persistentTensorAllocIds_.size();
+  }
+  /**
+   * <code>repeated int64 persistent_tensor_alloc_ids = 5;</code>
+   */
+  public long getPersistentTensorAllocIds(int index) {
+    return persistentTensorAllocIds_.getLong(index);
+  }
+  private int persistentTensorAllocIdsMemoizedSerializedSize = -1;
+
+  public static final int DEVICE_TEMP_MEMORY_SIZE_FIELD_NUMBER = 2;
+  private long deviceTempMemorySize_;
+  /**
+   * <code>int64 device_temp_memory_size = 2 [deprecated = true];</code>
+   */
+  @java.lang.Deprecated public long getDeviceTempMemorySize() {
+    return deviceTempMemorySize_;
+  }
+
+  public static final int DEVICE_PERSISTENT_MEMORY_SIZE_FIELD_NUMBER = 4;
+  private long devicePersistentMemorySize_;
+  /**
+   * <code>int64 device_persistent_memory_size = 4 [deprecated = true];</code>
+   */
+  @java.lang.Deprecated public long getDevicePersistentMemorySize() {
+    return devicePersistentMemorySize_;
+  }
+
+  public static final int DEVICE_PERSISTENT_TENSOR_ALLOC_IDS_FIELD_NUMBER = 6;
+  private com.google.protobuf.Internal.LongList devicePersistentTensorAllocIds_;
+  /**
+   * <code>repeated int64 device_persistent_tensor_alloc_ids = 6 [deprecated = true];</code>
+   */
+  @java.lang.Deprecated public java.util.List<java.lang.Long>
+      getDevicePersistentTensorAllocIdsList() {
+    return devicePersistentTensorAllocIds_;
+  }
+  /**
+   * <code>repeated int64 device_persistent_tensor_alloc_ids = 6 [deprecated = true];</code>
+   */
+  @java.lang.Deprecated public int getDevicePersistentTensorAllocIdsCount() {
+    return devicePersistentTensorAllocIds_.size();
+  }
+  /**
+   * <code>repeated int64 device_persistent_tensor_alloc_ids = 6 [deprecated = true];</code>
+   */
+  @java.lang.Deprecated public long getDevicePersistentTensorAllocIds(int index) {
+    return devicePersistentTensorAllocIds_.getLong(index);
+  }
+  private int devicePersistentTensorAllocIdsMemoizedSerializedSize = -1;
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    getSerializedSize();
+    if (tempMemorySize_ != 0L) {
+      output.writeInt64(1, tempMemorySize_);
+    }
+    if (deviceTempMemorySize_ != 0L) {
+      output.writeInt64(2, deviceTempMemorySize_);
+    }
+    if (persistentMemorySize_ != 0L) {
+      output.writeInt64(3, persistentMemorySize_);
+    }
+    if (devicePersistentMemorySize_ != 0L) {
+      output.writeInt64(4, devicePersistentMemorySize_);
+    }
+    if (getPersistentTensorAllocIdsList().size() > 0) {
+      output.writeUInt32NoTag(42);
+      output.writeUInt32NoTag(persistentTensorAllocIdsMemoizedSerializedSize);
+    }
+    for (int i = 0; i < persistentTensorAllocIds_.size(); i++) {
+      output.writeInt64NoTag(persistentTensorAllocIds_.getLong(i));
+    }
+    if (getDevicePersistentTensorAllocIdsList().size() > 0) {
+      output.writeUInt32NoTag(50);
+      output.writeUInt32NoTag(devicePersistentTensorAllocIdsMemoizedSerializedSize);
+    }
+    for (int i = 0; i < devicePersistentTensorAllocIds_.size(); i++) {
+      output.writeInt64NoTag(devicePersistentTensorAllocIds_.getLong(i));
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (tempMemorySize_ != 0L) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt64Size(1, tempMemorySize_);
+    }
+    if (deviceTempMemorySize_ != 0L) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt64Size(2, deviceTempMemorySize_);
+    }
+    if (persistentMemorySize_ != 0L) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt64Size(3, persistentMemorySize_);
+    }
+    if (devicePersistentMemorySize_ != 0L) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt64Size(4, devicePersistentMemorySize_);
+    }
+    {
+      int dataSize = 0;
+      for (int i = 0; i < persistentTensorAllocIds_.size(); i++) {
+        dataSize += com.google.protobuf.CodedOutputStream
+          .computeInt64SizeNoTag(persistentTensorAllocIds_.getLong(i));
+      }
+      size += dataSize;
+      if (!getPersistentTensorAllocIdsList().isEmpty()) {
+        size += 1;
+        size += com.google.protobuf.CodedOutputStream
+            .computeInt32SizeNoTag(dataSize);
+      }
+      persistentTensorAllocIdsMemoizedSerializedSize = dataSize;
+    }
+    {
+      int dataSize = 0;
+      for (int i = 0; i < devicePersistentTensorAllocIds_.size(); i++) {
+        dataSize += com.google.protobuf.CodedOutputStream
+          .computeInt64SizeNoTag(devicePersistentTensorAllocIds_.getLong(i));
+      }
+      size += dataSize;
+      if (!getDevicePersistentTensorAllocIdsList().isEmpty()) {
+        size += 1;
+        size += com.google.protobuf.CodedOutputStream
+            .computeInt32SizeNoTag(dataSize);
+      }
+      devicePersistentTensorAllocIdsMemoizedSerializedSize = dataSize;
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.framework.MemoryStats)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.framework.MemoryStats other = (org.tensorflow.proto.framework.MemoryStats) obj;
+
+    if (getTempMemorySize()
+        != other.getTempMemorySize()) return false;
+    if (getPersistentMemorySize()
+        != other.getPersistentMemorySize()) return false;
+    if (!getPersistentTensorAllocIdsList()
+        .equals(other.getPersistentTensorAllocIdsList())) return false;
+    if (getDeviceTempMemorySize()
+        != other.getDeviceTempMemorySize()) return false;
+    if (getDevicePersistentMemorySize()
+        != other.getDevicePersistentMemorySize()) return false;
+    if (!getDevicePersistentTensorAllocIdsList()
+        .equals(other.getDevicePersistentTensorAllocIdsList())) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    hash = (37 * hash) + TEMP_MEMORY_SIZE_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+        getTempMemorySize());
+    hash = (37 * hash) + PERSISTENT_MEMORY_SIZE_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+        getPersistentMemorySize());
+    if (getPersistentTensorAllocIdsCount() > 0) {
+      hash = (37 * hash) + PERSISTENT_TENSOR_ALLOC_IDS_FIELD_NUMBER;
+      hash = (53 * hash) + getPersistentTensorAllocIdsList().hashCode();
+    }
+    hash = (37 * hash) + DEVICE_TEMP_MEMORY_SIZE_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+        getDeviceTempMemorySize());
+    hash = (37 * hash) + DEVICE_PERSISTENT_MEMORY_SIZE_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+        getDevicePersistentMemorySize());
+    if (getDevicePersistentTensorAllocIdsCount() > 0) {
+      hash = (37 * hash) + DEVICE_PERSISTENT_TENSOR_ALLOC_IDS_FIELD_NUMBER;
+      hash = (53 * hash) + getDevicePersistentTensorAllocIdsList().hashCode();
+    }
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.framework.MemoryStats parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.MemoryStats parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.MemoryStats parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.MemoryStats parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.MemoryStats parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.MemoryStats parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.MemoryStats parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.MemoryStats parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.MemoryStats parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.MemoryStats parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.MemoryStats parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.MemoryStats parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.framework.MemoryStats prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * <pre>
+   * For memory tracking.
+   * </pre>
+   *
+   * Protobuf type {@code tensorflow.MemoryStats}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.MemoryStats)
+      org.tensorflow.proto.framework.MemoryStatsOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.StepStatsProtos.internal_static_tensorflow_MemoryStats_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.StepStatsProtos.internal_static_tensorflow_MemoryStats_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.MemoryStats.class, org.tensorflow.proto.framework.MemoryStats.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.framework.MemoryStats.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      tempMemorySize_ = 0L;
+
+      persistentMemorySize_ = 0L;
+
+      persistentTensorAllocIds_ = emptyLongList();
+      bitField0_ = (bitField0_ & ~0x00000001);
+      deviceTempMemorySize_ = 0L;
+
+      devicePersistentMemorySize_ = 0L;
+
+      devicePersistentTensorAllocIds_ = emptyLongList();
+      bitField0_ = (bitField0_ & ~0x00000002);
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.framework.StepStatsProtos.internal_static_tensorflow_MemoryStats_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.MemoryStats getDefaultInstanceForType() {
+      return org.tensorflow.proto.framework.MemoryStats.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.MemoryStats build() {
+      org.tensorflow.proto.framework.MemoryStats result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.MemoryStats buildPartial() {
+      org.tensorflow.proto.framework.MemoryStats result = new org.tensorflow.proto.framework.MemoryStats(this);
+      int from_bitField0_ = bitField0_;
+      result.tempMemorySize_ = tempMemorySize_;
+      result.persistentMemorySize_ = persistentMemorySize_;
+      if (((bitField0_ & 0x00000001) != 0)) {
+        persistentTensorAllocIds_.makeImmutable();
+        bitField0_ = (bitField0_ & ~0x00000001);
+      }
+      result.persistentTensorAllocIds_ = persistentTensorAllocIds_;
+      result.deviceTempMemorySize_ = deviceTempMemorySize_;
+      result.devicePersistentMemorySize_ = devicePersistentMemorySize_;
+      if (((bitField0_ & 0x00000002) != 0)) {
+        devicePersistentTensorAllocIds_.makeImmutable();
+        bitField0_ = (bitField0_ & ~0x00000002);
+      }
+      result.devicePersistentTensorAllocIds_ = devicePersistentTensorAllocIds_;
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.framework.MemoryStats) {
+        return mergeFrom((org.tensorflow.proto.framework.MemoryStats)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.framework.MemoryStats other) {
+      if (other == org.tensorflow.proto.framework.MemoryStats.getDefaultInstance()) return this;
+      if (other.getTempMemorySize() != 0L) {
+        setTempMemorySize(other.getTempMemorySize());
+      }
+      if (other.getPersistentMemorySize() != 0L) {
+        setPersistentMemorySize(other.getPersistentMemorySize());
+      }
+      if (!other.persistentTensorAllocIds_.isEmpty()) {
+        if (persistentTensorAllocIds_.isEmpty()) {
+          persistentTensorAllocIds_ = other.persistentTensorAllocIds_;
+          bitField0_ = (bitField0_ & ~0x00000001);
+        } else {
+          ensurePersistentTensorAllocIdsIsMutable();
+          persistentTensorAllocIds_.addAll(other.persistentTensorAllocIds_);
+        }
+        onChanged();
+      }
+      if (other.getDeviceTempMemorySize() != 0L) {
+        setDeviceTempMemorySize(other.getDeviceTempMemorySize());
+      }
+      if (other.getDevicePersistentMemorySize() != 0L) {
+        setDevicePersistentMemorySize(other.getDevicePersistentMemorySize());
+      }
+      if (!other.devicePersistentTensorAllocIds_.isEmpty()) {
+        if (devicePersistentTensorAllocIds_.isEmpty()) {
+          devicePersistentTensorAllocIds_ = other.devicePersistentTensorAllocIds_;
+          bitField0_ = (bitField0_ & ~0x00000002);
+        } else {
+          ensureDevicePersistentTensorAllocIdsIsMutable();
+          devicePersistentTensorAllocIds_.addAll(other.devicePersistentTensorAllocIds_);
+        }
+        onChanged();
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.framework.MemoryStats parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.framework.MemoryStats) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+    private int bitField0_;
+
+    private long tempMemorySize_ ;
+    /**
+     * <code>int64 temp_memory_size = 1;</code>
+     */
+    public long getTempMemorySize() {
+      return tempMemorySize_;
+    }
+    /**
+     * <code>int64 temp_memory_size = 1;</code>
+     */
+    public Builder setTempMemorySize(long value) {
+      
+      tempMemorySize_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>int64 temp_memory_size = 1;</code>
+     */
+    public Builder clearTempMemorySize() {
+      
+      tempMemorySize_ = 0L;
+      onChanged();
+      return this;
+    }
+
+    private long persistentMemorySize_ ;
+    /**
+     * <code>int64 persistent_memory_size = 3;</code>
+     */
+    public long getPersistentMemorySize() {
+      return persistentMemorySize_;
+    }
+    /**
+     * <code>int64 persistent_memory_size = 3;</code>
+     */
+    public Builder setPersistentMemorySize(long value) {
+      
+      persistentMemorySize_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>int64 persistent_memory_size = 3;</code>
+     */
+    public Builder clearPersistentMemorySize() {
+      
+      persistentMemorySize_ = 0L;
+      onChanged();
+      return this;
+    }
+
+    private com.google.protobuf.Internal.LongList persistentTensorAllocIds_ = emptyLongList();
+    private void ensurePersistentTensorAllocIdsIsMutable() {
+      if (!((bitField0_ & 0x00000001) != 0)) {
+        persistentTensorAllocIds_ = mutableCopy(persistentTensorAllocIds_);
+        bitField0_ |= 0x00000001;
+       }
+    }
+    /**
+     * <code>repeated int64 persistent_tensor_alloc_ids = 5;</code>
+     */
+    public java.util.List<java.lang.Long>
+        getPersistentTensorAllocIdsList() {
+      return ((bitField0_ & 0x00000001) != 0) ?
+               java.util.Collections.unmodifiableList(persistentTensorAllocIds_) : persistentTensorAllocIds_;
+    }
+    /**
+     * <code>repeated int64 persistent_tensor_alloc_ids = 5;</code>
+     */
+    public int getPersistentTensorAllocIdsCount() {
+      return persistentTensorAllocIds_.size();
+    }
+    /**
+     * <code>repeated int64 persistent_tensor_alloc_ids = 5;</code>
+     */
+    public long getPersistentTensorAllocIds(int index) {
+      return persistentTensorAllocIds_.getLong(index);
+    }
+    /**
+     * <code>repeated int64 persistent_tensor_alloc_ids = 5;</code>
+     */
+    public Builder setPersistentTensorAllocIds(
+        int index, long value) {
+      ensurePersistentTensorAllocIdsIsMutable();
+      persistentTensorAllocIds_.setLong(index, value);
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>repeated int64 persistent_tensor_alloc_ids = 5;</code>
+     */
+    public Builder addPersistentTensorAllocIds(long value) {
+      ensurePersistentTensorAllocIdsIsMutable();
+      persistentTensorAllocIds_.addLong(value);
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>repeated int64 persistent_tensor_alloc_ids = 5;</code>
+     */
+    public Builder addAllPersistentTensorAllocIds(
+        java.lang.Iterable<? extends java.lang.Long> values) {
+      ensurePersistentTensorAllocIdsIsMutable();
+      com.google.protobuf.AbstractMessageLite.Builder.addAll(
+          values, persistentTensorAllocIds_);
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>repeated int64 persistent_tensor_alloc_ids = 5;</code>
+     */
+    public Builder clearPersistentTensorAllocIds() {
+      persistentTensorAllocIds_ = emptyLongList();
+      bitField0_ = (bitField0_ & ~0x00000001);
+      onChanged();
+      return this;
+    }
+
+    private long deviceTempMemorySize_ ;
+    /**
+     * <code>int64 device_temp_memory_size = 2 [deprecated = true];</code>
+     */
+    @java.lang.Deprecated public long getDeviceTempMemorySize() {
+      return deviceTempMemorySize_;
+    }
+    /**
+     * <code>int64 device_temp_memory_size = 2 [deprecated = true];</code>
+     */
+    @java.lang.Deprecated public Builder setDeviceTempMemorySize(long value) {
+      
+      deviceTempMemorySize_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>int64 device_temp_memory_size = 2 [deprecated = true];</code>
+     */
+    @java.lang.Deprecated public Builder clearDeviceTempMemorySize() {
+      
+      deviceTempMemorySize_ = 0L;
+      onChanged();
+      return this;
+    }
+
+    private long devicePersistentMemorySize_ ;
+    /**
+     * <code>int64 device_persistent_memory_size = 4 [deprecated = true];</code>
+     */
+    @java.lang.Deprecated public long getDevicePersistentMemorySize() {
+      return devicePersistentMemorySize_;
+    }
+    /**
+     * <code>int64 device_persistent_memory_size = 4 [deprecated = true];</code>
+     */
+    @java.lang.Deprecated public Builder setDevicePersistentMemorySize(long value) {
+      
+      devicePersistentMemorySize_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>int64 device_persistent_memory_size = 4 [deprecated = true];</code>
+     */
+    @java.lang.Deprecated public Builder clearDevicePersistentMemorySize() {
+      
+      devicePersistentMemorySize_ = 0L;
+      onChanged();
+      return this;
+    }
+
+    private com.google.protobuf.Internal.LongList devicePersistentTensorAllocIds_ = emptyLongList();
+    private void ensureDevicePersistentTensorAllocIdsIsMutable() {
+      if (!((bitField0_ & 0x00000002) != 0)) {
+        devicePersistentTensorAllocIds_ = mutableCopy(devicePersistentTensorAllocIds_);
+        bitField0_ |= 0x00000002;
+       }
+    }
+    /**
+     * <code>repeated int64 device_persistent_tensor_alloc_ids = 6 [deprecated = true];</code>
+     */
+    @java.lang.Deprecated public java.util.List<java.lang.Long>
+        getDevicePersistentTensorAllocIdsList() {
+      return ((bitField0_ & 0x00000002) != 0) ?
+               java.util.Collections.unmodifiableList(devicePersistentTensorAllocIds_) : devicePersistentTensorAllocIds_;
+    }
+    /**
+     * <code>repeated int64 device_persistent_tensor_alloc_ids = 6 [deprecated = true];</code>
+     */
+    @java.lang.Deprecated public int getDevicePersistentTensorAllocIdsCount() {
+      return devicePersistentTensorAllocIds_.size();
+    }
+    /**
+     * <code>repeated int64 device_persistent_tensor_alloc_ids = 6 [deprecated = true];</code>
+     */
+    @java.lang.Deprecated public long getDevicePersistentTensorAllocIds(int index) {
+      return devicePersistentTensorAllocIds_.getLong(index);
+    }
+    /**
+     * <code>repeated int64 device_persistent_tensor_alloc_ids = 6 [deprecated = true];</code>
+     */
+    @java.lang.Deprecated public Builder setDevicePersistentTensorAllocIds(
+        int index, long value) {
+      ensureDevicePersistentTensorAllocIdsIsMutable();
+      devicePersistentTensorAllocIds_.setLong(index, value);
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>repeated int64 device_persistent_tensor_alloc_ids = 6 [deprecated = true];</code>
+     */
+    @java.lang.Deprecated public Builder addDevicePersistentTensorAllocIds(long value) {
+      ensureDevicePersistentTensorAllocIdsIsMutable();
+      devicePersistentTensorAllocIds_.addLong(value);
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>repeated int64 device_persistent_tensor_alloc_ids = 6 [deprecated = true];</code>
+     */
+    @java.lang.Deprecated public Builder addAllDevicePersistentTensorAllocIds(
+        java.lang.Iterable<? extends java.lang.Long> values) {
+      ensureDevicePersistentTensorAllocIdsIsMutable();
+      com.google.protobuf.AbstractMessageLite.Builder.addAll(
+          values, devicePersistentTensorAllocIds_);
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>repeated int64 device_persistent_tensor_alloc_ids = 6 [deprecated = true];</code>
+     */
+    @java.lang.Deprecated public Builder clearDevicePersistentTensorAllocIds() {
+      devicePersistentTensorAllocIds_ = emptyLongList();
+      bitField0_ = (bitField0_ & ~0x00000002);
+      onChanged();
+      return this;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.MemoryStats)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.MemoryStats)
+  private static final org.tensorflow.proto.framework.MemoryStats DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.MemoryStats();
+  }
+
+  public static org.tensorflow.proto.framework.MemoryStats getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<MemoryStats>
+      PARSER = new com.google.protobuf.AbstractParser<MemoryStats>() {
+    @java.lang.Override
+    public MemoryStats parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new MemoryStats(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<MemoryStats> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<MemoryStats> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.framework.MemoryStats getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/MemoryStatsOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/MemoryStatsOrBuilder.java
new file mode 100644
index 00000000000..db3c896352c
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/MemoryStatsOrBuilder.java
@@ -0,0 +1,55 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/framework/step_stats.proto
+
+package org.tensorflow.proto.framework;
+
+public interface MemoryStatsOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.MemoryStats)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <code>int64 temp_memory_size = 1;</code>
+   */
+  long getTempMemorySize();
+
+  /**
+   * <code>int64 persistent_memory_size = 3;</code>
+   */
+  long getPersistentMemorySize();
+
+  /**
+   * <code>repeated int64 persistent_tensor_alloc_ids = 5;</code>
+   */
+  java.util.List<java.lang.Long> getPersistentTensorAllocIdsList();
+  /**
+   * <code>repeated int64 persistent_tensor_alloc_ids = 5;</code>
+   */
+  int getPersistentTensorAllocIdsCount();
+  /**
+   * <code>repeated int64 persistent_tensor_alloc_ids = 5;</code>
+   */
+  long getPersistentTensorAllocIds(int index);
+
+  /**
+   * <code>int64 device_temp_memory_size = 2 [deprecated = true];</code>
+   */
+  @java.lang.Deprecated long getDeviceTempMemorySize();
+
+  /**
+   * <code>int64 device_persistent_memory_size = 4 [deprecated = true];</code>
+   */
+  @java.lang.Deprecated long getDevicePersistentMemorySize();
+
+  /**
+   * <code>repeated int64 device_persistent_tensor_alloc_ids = 6 [deprecated = true];</code>
+   */
+  @java.lang.Deprecated java.util.List<java.lang.Long> getDevicePersistentTensorAllocIdsList();
+  /**
+   * <code>repeated int64 device_persistent_tensor_alloc_ids = 6 [deprecated = true];</code>
+   */
+  @java.lang.Deprecated int getDevicePersistentTensorAllocIdsCount();
+  /**
+   * <code>repeated int64 device_persistent_tensor_alloc_ids = 6 [deprecated = true];</code>
+   */
+  @java.lang.Deprecated long getDevicePersistentTensorAllocIds(int index);
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/MetaGraphDef.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/MetaGraphDef.java
new file mode 100644
index 00000000000..b071cf84bf8
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/MetaGraphDef.java
@@ -0,0 +1,4702 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/meta_graph.proto
+
+package org.tensorflow.proto.framework;
+
+/**
+ * <pre>
+ * NOTE: This protocol buffer is evolving, and will go through revisions in the
+ * coming months.
+ * Protocol buffer containing the following which are necessary to restart
+ * training, run inference. It can be used to serialize/de-serialize memory
+ * objects necessary for running computation in a graph when crossing the
+ * process boundary. It can be used for long term storage of graphs,
+ * cross-language execution of graphs, etc.
+ *   MetaInfoDef
+ *   GraphDef
+ *   SaverDef
+ *   CollectionDef
+ *   TensorInfo
+ *   SignatureDef
+ * </pre>
+ *
+ * Protobuf type {@code tensorflow.MetaGraphDef}
+ */
+public  final class MetaGraphDef extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.MetaGraphDef)
+    MetaGraphDefOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use MetaGraphDef.newBuilder() to construct.
+  private MetaGraphDef(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private MetaGraphDef() {
+    assetFileDef_ = java.util.Collections.emptyList();
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new MetaGraphDef();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private MetaGraphDef(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    int mutable_bitField0_ = 0;
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 10: {
+            org.tensorflow.proto.framework.MetaGraphDef.MetaInfoDef.Builder subBuilder = null;
+            if (metaInfoDef_ != null) {
+              subBuilder = metaInfoDef_.toBuilder();
+            }
+            metaInfoDef_ = input.readMessage(org.tensorflow.proto.framework.MetaGraphDef.MetaInfoDef.parser(), extensionRegistry);
+            if (subBuilder != null) {
+              subBuilder.mergeFrom(metaInfoDef_);
+              metaInfoDef_ = subBuilder.buildPartial();
+            }
+
+            break;
+          }
+          case 18: {
+            org.tensorflow.proto.framework.GraphDef.Builder subBuilder = null;
+            if (graphDef_ != null) {
+              subBuilder = graphDef_.toBuilder();
+            }
+            graphDef_ = input.readMessage(org.tensorflow.proto.framework.GraphDef.parser(), extensionRegistry);
+            if (subBuilder != null) {
+              subBuilder.mergeFrom(graphDef_);
+              graphDef_ = subBuilder.buildPartial();
+            }
+
+            break;
+          }
+          case 26: {
+            org.tensorflow.proto.util.SaverDef.Builder subBuilder = null;
+            if (saverDef_ != null) {
+              subBuilder = saverDef_.toBuilder();
+            }
+            saverDef_ = input.readMessage(org.tensorflow.proto.util.SaverDef.parser(), extensionRegistry);
+            if (subBuilder != null) {
+              subBuilder.mergeFrom(saverDef_);
+              saverDef_ = subBuilder.buildPartial();
+            }
+
+            break;
+          }
+          case 34: {
+            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
+              collectionDef_ = com.google.protobuf.MapField.newMapField(
+                  CollectionDefDefaultEntryHolder.defaultEntry);
+              mutable_bitField0_ |= 0x00000001;
+            }
+            com.google.protobuf.MapEntry<java.lang.String, org.tensorflow.proto.framework.CollectionDef>
+            collectionDef__ = input.readMessage(
+                CollectionDefDefaultEntryHolder.defaultEntry.getParserForType(), extensionRegistry);
+            collectionDef_.getMutableMap().put(
+                collectionDef__.getKey(), collectionDef__.getValue());
+            break;
+          }
+          case 42: {
+            if (!((mutable_bitField0_ & 0x00000002) != 0)) {
+              signatureDef_ = com.google.protobuf.MapField.newMapField(
+                  SignatureDefDefaultEntryHolder.defaultEntry);
+              mutable_bitField0_ |= 0x00000002;
+            }
+            com.google.protobuf.MapEntry<java.lang.String, org.tensorflow.proto.framework.SignatureDef>
+            signatureDef__ = input.readMessage(
+                SignatureDefDefaultEntryHolder.defaultEntry.getParserForType(), extensionRegistry);
+            signatureDef_.getMutableMap().put(
+                signatureDef__.getKey(), signatureDef__.getValue());
+            break;
+          }
+          case 50: {
+            if (!((mutable_bitField0_ & 0x00000004) != 0)) {
+              assetFileDef_ = new java.util.ArrayList<org.tensorflow.proto.framework.AssetFileDef>();
+              mutable_bitField0_ |= 0x00000004;
+            }
+            assetFileDef_.add(
+                input.readMessage(org.tensorflow.proto.framework.AssetFileDef.parser(), extensionRegistry));
+            break;
+          }
+          case 58: {
+            org.tensorflow.proto.framework.SavedObjectGraph.Builder subBuilder = null;
+            if (objectGraphDef_ != null) {
+              subBuilder = objectGraphDef_.toBuilder();
+            }
+            objectGraphDef_ = input.readMessage(org.tensorflow.proto.framework.SavedObjectGraph.parser(), extensionRegistry);
+            if (subBuilder != null) {
+              subBuilder.mergeFrom(objectGraphDef_);
+              objectGraphDef_ = subBuilder.buildPartial();
+            }
+
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      if (((mutable_bitField0_ & 0x00000004) != 0)) {
+        assetFileDef_ = java.util.Collections.unmodifiableList(assetFileDef_);
+      }
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.framework.MetaGraphProtos.internal_static_tensorflow_MetaGraphDef_descriptor;
+  }
+
+  @SuppressWarnings({"rawtypes"})
+  @java.lang.Override
+  protected com.google.protobuf.MapField internalGetMapField(
+      int number) {
+    switch (number) {
+      case 4:
+        return internalGetCollectionDef();
+      case 5:
+        return internalGetSignatureDef();
+      default:
+        throw new RuntimeException(
+            "Invalid map field number: " + number);
+    }
+  }
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.framework.MetaGraphProtos.internal_static_tensorflow_MetaGraphDef_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.framework.MetaGraphDef.class, org.tensorflow.proto.framework.MetaGraphDef.Builder.class);
+  }
+
+  public interface MetaInfoDefOrBuilder extends
+      // @@protoc_insertion_point(interface_extends:tensorflow.MetaGraphDef.MetaInfoDef)
+      com.google.protobuf.MessageOrBuilder {
+
+    /**
+     * <pre>
+     * User specified Version string. Can be the name of the model and revision,
+     * steps this model has been trained to, etc.
+     * </pre>
+     *
+     * <code>string meta_graph_version = 1;</code>
+     */
+    java.lang.String getMetaGraphVersion();
+    /**
+     * <pre>
+     * User specified Version string. Can be the name of the model and revision,
+     * steps this model has been trained to, etc.
+     * </pre>
+     *
+     * <code>string meta_graph_version = 1;</code>
+     */
+    com.google.protobuf.ByteString
+        getMetaGraphVersionBytes();
+
+    /**
+     * <pre>
+     * A copy of the OpDefs used by the producer of this graph_def.
+     * Descriptions and Ops not used in graph_def are stripped out.
+     * </pre>
+     *
+     * <code>.tensorflow.OpList stripped_op_list = 2;</code>
+     */
+    boolean hasStrippedOpList();
+    /**
+     * <pre>
+     * A copy of the OpDefs used by the producer of this graph_def.
+     * Descriptions and Ops not used in graph_def are stripped out.
+     * </pre>
+     *
+     * <code>.tensorflow.OpList stripped_op_list = 2;</code>
+     */
+    org.tensorflow.proto.framework.OpList getStrippedOpList();
+    /**
+     * <pre>
+     * A copy of the OpDefs used by the producer of this graph_def.
+     * Descriptions and Ops not used in graph_def are stripped out.
+     * </pre>
+     *
+     * <code>.tensorflow.OpList stripped_op_list = 2;</code>
+     */
+    org.tensorflow.proto.framework.OpListOrBuilder getStrippedOpListOrBuilder();
+
+    /**
+     * <pre>
+     * A serialized protobuf. Can be the time this meta graph is created, or
+     * modified, or name of the model.
+     * </pre>
+     *
+     * <code>.google.protobuf.Any any_info = 3;</code>
+     */
+    boolean hasAnyInfo();
+    /**
+     * <pre>
+     * A serialized protobuf. Can be the time this meta graph is created, or
+     * modified, or name of the model.
+     * </pre>
+     *
+     * <code>.google.protobuf.Any any_info = 3;</code>
+     */
+    com.google.protobuf.Any getAnyInfo();
+    /**
+     * <pre>
+     * A serialized protobuf. Can be the time this meta graph is created, or
+     * modified, or name of the model.
+     * </pre>
+     *
+     * <code>.google.protobuf.Any any_info = 3;</code>
+     */
+    com.google.protobuf.AnyOrBuilder getAnyInfoOrBuilder();
+
+    /**
+     * <pre>
+     * User supplied tag(s) on the meta_graph and included graph_def.
+     * MetaGraphDefs should be tagged with their capabilities or use-cases.
+     * Examples: "train", "serve", "gpu", "tpu", etc.
+     * These tags enable loaders to access the MetaGraph(s) appropriate for a
+     * specific use-case or runtime environment.
+     * </pre>
+     *
+     * <code>repeated string tags = 4;</code>
+     */
+    java.util.List<java.lang.String>
+        getTagsList();
+    /**
+     * <pre>
+     * User supplied tag(s) on the meta_graph and included graph_def.
+     * MetaGraphDefs should be tagged with their capabilities or use-cases.
+     * Examples: "train", "serve", "gpu", "tpu", etc.
+     * These tags enable loaders to access the MetaGraph(s) appropriate for a
+     * specific use-case or runtime environment.
+     * </pre>
+     *
+     * <code>repeated string tags = 4;</code>
+     */
+    int getTagsCount();
+    /**
+     * <pre>
+     * User supplied tag(s) on the meta_graph and included graph_def.
+     * MetaGraphDefs should be tagged with their capabilities or use-cases.
+     * Examples: "train", "serve", "gpu", "tpu", etc.
+     * These tags enable loaders to access the MetaGraph(s) appropriate for a
+     * specific use-case or runtime environment.
+     * </pre>
+     *
+     * <code>repeated string tags = 4;</code>
+     */
+    java.lang.String getTags(int index);
+    /**
+     * <pre>
+     * User supplied tag(s) on the meta_graph and included graph_def.
+     * MetaGraphDefs should be tagged with their capabilities or use-cases.
+     * Examples: "train", "serve", "gpu", "tpu", etc.
+     * These tags enable loaders to access the MetaGraph(s) appropriate for a
+     * specific use-case or runtime environment.
+     * </pre>
+     *
+     * <code>repeated string tags = 4;</code>
+     */
+    com.google.protobuf.ByteString
+        getTagsBytes(int index);
+
+    /**
+     * <pre>
+     * The __version__ string of the tensorflow build used to write this graph.
+     * This will be populated by the framework, which will overwrite any user
+     * supplied value.
+     * </pre>
+     *
+     * <code>string tensorflow_version = 5;</code>
+     */
+    java.lang.String getTensorflowVersion();
+    /**
+     * <pre>
+     * The __version__ string of the tensorflow build used to write this graph.
+     * This will be populated by the framework, which will overwrite any user
+     * supplied value.
+     * </pre>
+     *
+     * <code>string tensorflow_version = 5;</code>
+     */
+    com.google.protobuf.ByteString
+        getTensorflowVersionBytes();
+
+    /**
+     * <pre>
+     * The __git_version__ string of the tensorflow build used to write this
+     * graph. This will be populated by the framework, which will overwrite any
+     * user supplied value.
+     * </pre>
+     *
+     * <code>string tensorflow_git_version = 6;</code>
+     */
+    java.lang.String getTensorflowGitVersion();
+    /**
+     * <pre>
+     * The __git_version__ string of the tensorflow build used to write this
+     * graph. This will be populated by the framework, which will overwrite any
+     * user supplied value.
+     * </pre>
+     *
+     * <code>string tensorflow_git_version = 6;</code>
+     */
+    com.google.protobuf.ByteString
+        getTensorflowGitVersionBytes();
+
+    /**
+     * <pre>
+     * A flag to denote whether default-valued attrs have been stripped from
+     * the nodes in this graph_def.
+     * </pre>
+     *
+     * <code>bool stripped_default_attrs = 7;</code>
+     */
+    boolean getStrippedDefaultAttrs();
+
+    /**
+     * <pre>
+     * FunctionDef name to aliases mapping.
+     * </pre>
+     *
+     * <code>map&lt;string, string&gt; function_aliases = 8;</code>
+     */
+    int getFunctionAliasesCount();
+    /**
+     * <pre>
+     * FunctionDef name to aliases mapping.
+     * </pre>
+     *
+     * <code>map&lt;string, string&gt; function_aliases = 8;</code>
+     */
+    boolean containsFunctionAliases(
+        java.lang.String key);
+    /**
+     * Use {@link #getFunctionAliasesMap()} instead.
+     */
+    @java.lang.Deprecated
+    java.util.Map<java.lang.String, java.lang.String>
+    getFunctionAliases();
+    /**
+     * <pre>
+     * FunctionDef name to aliases mapping.
+     * </pre>
+     *
+     * <code>map&lt;string, string&gt; function_aliases = 8;</code>
+     */
+    java.util.Map<java.lang.String, java.lang.String>
+    getFunctionAliasesMap();
+    /**
+     * <pre>
+     * FunctionDef name to aliases mapping.
+     * </pre>
+     *
+     * <code>map&lt;string, string&gt; function_aliases = 8;</code>
+     */
+
+    java.lang.String getFunctionAliasesOrDefault(
+        java.lang.String key,
+        java.lang.String defaultValue);
+    /**
+     * <pre>
+     * FunctionDef name to aliases mapping.
+     * </pre>
+     *
+     * <code>map&lt;string, string&gt; function_aliases = 8;</code>
+     */
+
+    java.lang.String getFunctionAliasesOrThrow(
+        java.lang.String key);
+  }
+  /**
+   * <pre>
+   * Meta information regarding the graph to be exported.  To be used by users
+   * of this protocol buffer to encode information regarding their meta graph.
+   * </pre>
+   *
+   * Protobuf type {@code tensorflow.MetaGraphDef.MetaInfoDef}
+   */
+  public  static final class MetaInfoDef extends
+      com.google.protobuf.GeneratedMessageV3 implements
+      // @@protoc_insertion_point(message_implements:tensorflow.MetaGraphDef.MetaInfoDef)
+      MetaInfoDefOrBuilder {
+  private static final long serialVersionUID = 0L;
+    // Use MetaInfoDef.newBuilder() to construct.
+    private MetaInfoDef(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+      super(builder);
+    }
+    private MetaInfoDef() {
+      metaGraphVersion_ = "";
+      tags_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+      tensorflowVersion_ = "";
+      tensorflowGitVersion_ = "";
+    }
+
+    @java.lang.Override
+    @SuppressWarnings({"unused"})
+    protected java.lang.Object newInstance(
+        UnusedPrivateParameter unused) {
+      return new MetaInfoDef();
+    }
+
+    @java.lang.Override
+    public final com.google.protobuf.UnknownFieldSet
+    getUnknownFields() {
+      return this.unknownFields;
+    }
+    private MetaInfoDef(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      this();
+      if (extensionRegistry == null) {
+        throw new java.lang.NullPointerException();
+      }
+      int mutable_bitField0_ = 0;
+      com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+          com.google.protobuf.UnknownFieldSet.newBuilder();
+      try {
+        boolean done = false;
+        while (!done) {
+          int tag = input.readTag();
+          switch (tag) {
+            case 0:
+              done = true;
+              break;
+            case 10: {
+              java.lang.String s = input.readStringRequireUtf8();
+
+              metaGraphVersion_ = s;
+              break;
+            }
+            case 18: {
+              org.tensorflow.proto.framework.OpList.Builder subBuilder = null;
+              if (strippedOpList_ != null) {
+                subBuilder = strippedOpList_.toBuilder();
+              }
+              strippedOpList_ = input.readMessage(org.tensorflow.proto.framework.OpList.parser(), extensionRegistry);
+              if (subBuilder != null) {
+                subBuilder.mergeFrom(strippedOpList_);
+                strippedOpList_ = subBuilder.buildPartial();
+              }
+
+              break;
+            }
+            case 26: {
+              com.google.protobuf.Any.Builder subBuilder = null;
+              if (anyInfo_ != null) {
+                subBuilder = anyInfo_.toBuilder();
+              }
+              anyInfo_ = input.readMessage(com.google.protobuf.Any.parser(), extensionRegistry);
+              if (subBuilder != null) {
+                subBuilder.mergeFrom(anyInfo_);
+                anyInfo_ = subBuilder.buildPartial();
+              }
+
+              break;
+            }
+            case 34: {
+              java.lang.String s = input.readStringRequireUtf8();
+              if (!((mutable_bitField0_ & 0x00000001) != 0)) {
+                tags_ = new com.google.protobuf.LazyStringArrayList();
+                mutable_bitField0_ |= 0x00000001;
+              }
+              tags_.add(s);
+              break;
+            }
+            case 42: {
+              java.lang.String s = input.readStringRequireUtf8();
+
+              tensorflowVersion_ = s;
+              break;
+            }
+            case 50: {
+              java.lang.String s = input.readStringRequireUtf8();
+
+              tensorflowGitVersion_ = s;
+              break;
+            }
+            case 56: {
+
+              strippedDefaultAttrs_ = input.readBool();
+              break;
+            }
+            case 66: {
+              if (!((mutable_bitField0_ & 0x00000002) != 0)) {
+                functionAliases_ = com.google.protobuf.MapField.newMapField(
+                    FunctionAliasesDefaultEntryHolder.defaultEntry);
+                mutable_bitField0_ |= 0x00000002;
+              }
+              com.google.protobuf.MapEntry<java.lang.String, java.lang.String>
+              functionAliases__ = input.readMessage(
+                  FunctionAliasesDefaultEntryHolder.defaultEntry.getParserForType(), extensionRegistry);
+              functionAliases_.getMutableMap().put(
+                  functionAliases__.getKey(), functionAliases__.getValue());
+              break;
+            }
+            default: {
+              if (!parseUnknownField(
+                  input, unknownFields, extensionRegistry, tag)) {
+                done = true;
+              }
+              break;
+            }
+          }
+        }
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        throw e.setUnfinishedMessage(this);
+      } catch (java.io.IOException e) {
+        throw new com.google.protobuf.InvalidProtocolBufferException(
+            e).setUnfinishedMessage(this);
+      } finally {
+        if (((mutable_bitField0_ & 0x00000001) != 0)) {
+          tags_ = tags_.getUnmodifiableView();
+        }
+        this.unknownFields = unknownFields.build();
+        makeExtensionsImmutable();
+      }
+    }
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.MetaGraphProtos.internal_static_tensorflow_MetaGraphDef_MetaInfoDef_descriptor;
+    }
+
+    @SuppressWarnings({"rawtypes"})
+    @java.lang.Override
+    protected com.google.protobuf.MapField internalGetMapField(
+        int number) {
+      switch (number) {
+        case 8:
+          return internalGetFunctionAliases();
+        default:
+          throw new RuntimeException(
+              "Invalid map field number: " + number);
+      }
+    }
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.MetaGraphProtos.internal_static_tensorflow_MetaGraphDef_MetaInfoDef_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.MetaGraphDef.MetaInfoDef.class, org.tensorflow.proto.framework.MetaGraphDef.MetaInfoDef.Builder.class);
+    }
+
+    public static final int META_GRAPH_VERSION_FIELD_NUMBER = 1;
+    private volatile java.lang.Object metaGraphVersion_;
+    /**
+     * <pre>
+     * User specified Version string. Can be the name of the model and revision,
+     * steps this model has been trained to, etc.
+     * </pre>
+     *
+     * <code>string meta_graph_version = 1;</code>
+     */
+    public java.lang.String getMetaGraphVersion() {
+      java.lang.Object ref = metaGraphVersion_;
+      if (ref instanceof java.lang.String) {
+        return (java.lang.String) ref;
+      } else {
+        com.google.protobuf.ByteString bs = 
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        metaGraphVersion_ = s;
+        return s;
+      }
+    }
+    /**
+     * <pre>
+     * User specified Version string. Can be the name of the model and revision,
+     * steps this model has been trained to, etc.
+     * </pre>
+     *
+     * <code>string meta_graph_version = 1;</code>
+     */
+    public com.google.protobuf.ByteString
+        getMetaGraphVersionBytes() {
+      java.lang.Object ref = metaGraphVersion_;
+      if (ref instanceof java.lang.String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        metaGraphVersion_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+
+    public static final int STRIPPED_OP_LIST_FIELD_NUMBER = 2;
+    private org.tensorflow.proto.framework.OpList strippedOpList_;
+    /**
+     * <pre>
+     * A copy of the OpDefs used by the producer of this graph_def.
+     * Descriptions and Ops not used in graph_def are stripped out.
+     * </pre>
+     *
+     * <code>.tensorflow.OpList stripped_op_list = 2;</code>
+     */
+    public boolean hasStrippedOpList() {
+      return strippedOpList_ != null;
+    }
+    /**
+     * <pre>
+     * A copy of the OpDefs used by the producer of this graph_def.
+     * Descriptions and Ops not used in graph_def are stripped out.
+     * </pre>
+     *
+     * <code>.tensorflow.OpList stripped_op_list = 2;</code>
+     */
+    public org.tensorflow.proto.framework.OpList getStrippedOpList() {
+      return strippedOpList_ == null ? org.tensorflow.proto.framework.OpList.getDefaultInstance() : strippedOpList_;
+    }
+    /**
+     * <pre>
+     * A copy of the OpDefs used by the producer of this graph_def.
+     * Descriptions and Ops not used in graph_def are stripped out.
+     * </pre>
+     *
+     * <code>.tensorflow.OpList stripped_op_list = 2;</code>
+     */
+    public org.tensorflow.proto.framework.OpListOrBuilder getStrippedOpListOrBuilder() {
+      return getStrippedOpList();
+    }
+
+    public static final int ANY_INFO_FIELD_NUMBER = 3;
+    private com.google.protobuf.Any anyInfo_;
+    /**
+     * <pre>
+     * A serialized protobuf. Can be the time this meta graph is created, or
+     * modified, or name of the model.
+     * </pre>
+     *
+     * <code>.google.protobuf.Any any_info = 3;</code>
+     */
+    public boolean hasAnyInfo() {
+      return anyInfo_ != null;
+    }
+    /**
+     * <pre>
+     * A serialized protobuf. Can be the time this meta graph is created, or
+     * modified, or name of the model.
+     * </pre>
+     *
+     * <code>.google.protobuf.Any any_info = 3;</code>
+     */
+    public com.google.protobuf.Any getAnyInfo() {
+      return anyInfo_ == null ? com.google.protobuf.Any.getDefaultInstance() : anyInfo_;
+    }
+    /**
+     * <pre>
+     * A serialized protobuf. Can be the time this meta graph is created, or
+     * modified, or name of the model.
+     * </pre>
+     *
+     * <code>.google.protobuf.Any any_info = 3;</code>
+     */
+    public com.google.protobuf.AnyOrBuilder getAnyInfoOrBuilder() {
+      return getAnyInfo();
+    }
+
+    public static final int TAGS_FIELD_NUMBER = 4;
+    private com.google.protobuf.LazyStringList tags_;
+    /**
+     * <pre>
+     * User supplied tag(s) on the meta_graph and included graph_def.
+     * MetaGraphDefs should be tagged with their capabilities or use-cases.
+     * Examples: "train", "serve", "gpu", "tpu", etc.
+     * These tags enable loaders to access the MetaGraph(s) appropriate for a
+     * specific use-case or runtime environment.
+     * </pre>
+     *
+     * <code>repeated string tags = 4;</code>
+     */
+    public com.google.protobuf.ProtocolStringList
+        getTagsList() {
+      return tags_;
+    }
+    /**
+     * <pre>
+     * User supplied tag(s) on the meta_graph and included graph_def.
+     * MetaGraphDefs should be tagged with their capabilities or use-cases.
+     * Examples: "train", "serve", "gpu", "tpu", etc.
+     * These tags enable loaders to access the MetaGraph(s) appropriate for a
+     * specific use-case or runtime environment.
+     * </pre>
+     *
+     * <code>repeated string tags = 4;</code>
+     */
+    public int getTagsCount() {
+      return tags_.size();
+    }
+    /**
+     * <pre>
+     * User supplied tag(s) on the meta_graph and included graph_def.
+     * MetaGraphDefs should be tagged with their capabilities or use-cases.
+     * Examples: "train", "serve", "gpu", "tpu", etc.
+     * These tags enable loaders to access the MetaGraph(s) appropriate for a
+     * specific use-case or runtime environment.
+     * </pre>
+     *
+     * <code>repeated string tags = 4;</code>
+     */
+    public java.lang.String getTags(int index) {
+      return tags_.get(index);
+    }
+    /**
+     * <pre>
+     * User supplied tag(s) on the meta_graph and included graph_def.
+     * MetaGraphDefs should be tagged with their capabilities or use-cases.
+     * Examples: "train", "serve", "gpu", "tpu", etc.
+     * These tags enable loaders to access the MetaGraph(s) appropriate for a
+     * specific use-case or runtime environment.
+     * </pre>
+     *
+     * <code>repeated string tags = 4;</code>
+     */
+    public com.google.protobuf.ByteString
+        getTagsBytes(int index) {
+      return tags_.getByteString(index);
+    }
+
+    public static final int TENSORFLOW_VERSION_FIELD_NUMBER = 5;
+    private volatile java.lang.Object tensorflowVersion_;
+    /**
+     * <pre>
+     * The __version__ string of the tensorflow build used to write this graph.
+     * This will be populated by the framework, which will overwrite any user
+     * supplied value.
+     * </pre>
+     *
+     * <code>string tensorflow_version = 5;</code>
+     */
+    public java.lang.String getTensorflowVersion() {
+      java.lang.Object ref = tensorflowVersion_;
+      if (ref instanceof java.lang.String) {
+        return (java.lang.String) ref;
+      } else {
+        com.google.protobuf.ByteString bs = 
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        tensorflowVersion_ = s;
+        return s;
+      }
+    }
+    /**
+     * <pre>
+     * The __version__ string of the tensorflow build used to write this graph.
+     * This will be populated by the framework, which will overwrite any user
+     * supplied value.
+     * </pre>
+     *
+     * <code>string tensorflow_version = 5;</code>
+     */
+    public com.google.protobuf.ByteString
+        getTensorflowVersionBytes() {
+      java.lang.Object ref = tensorflowVersion_;
+      if (ref instanceof java.lang.String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        tensorflowVersion_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+
+    public static final int TENSORFLOW_GIT_VERSION_FIELD_NUMBER = 6;
+    private volatile java.lang.Object tensorflowGitVersion_;
+    /**
+     * <pre>
+     * The __git_version__ string of the tensorflow build used to write this
+     * graph. This will be populated by the framework, which will overwrite any
+     * user supplied value.
+     * </pre>
+     *
+     * <code>string tensorflow_git_version = 6;</code>
+     */
+    public java.lang.String getTensorflowGitVersion() {
+      java.lang.Object ref = tensorflowGitVersion_;
+      if (ref instanceof java.lang.String) {
+        return (java.lang.String) ref;
+      } else {
+        com.google.protobuf.ByteString bs = 
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        tensorflowGitVersion_ = s;
+        return s;
+      }
+    }
+    /**
+     * <pre>
+     * The __git_version__ string of the tensorflow build used to write this
+     * graph. This will be populated by the framework, which will overwrite any
+     * user supplied value.
+     * </pre>
+     *
+     * <code>string tensorflow_git_version = 6;</code>
+     */
+    public com.google.protobuf.ByteString
+        getTensorflowGitVersionBytes() {
+      java.lang.Object ref = tensorflowGitVersion_;
+      if (ref instanceof java.lang.String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        tensorflowGitVersion_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+
+    public static final int STRIPPED_DEFAULT_ATTRS_FIELD_NUMBER = 7;
+    private boolean strippedDefaultAttrs_;
+    /**
+     * <pre>
+     * A flag to denote whether default-valued attrs have been stripped from
+     * the nodes in this graph_def.
+     * </pre>
+     *
+     * <code>bool stripped_default_attrs = 7;</code>
+     */
+    public boolean getStrippedDefaultAttrs() {
+      return strippedDefaultAttrs_;
+    }
+
+    public static final int FUNCTION_ALIASES_FIELD_NUMBER = 8;
+    private static final class FunctionAliasesDefaultEntryHolder {
+      static final com.google.protobuf.MapEntry<
+          java.lang.String, java.lang.String> defaultEntry =
+              com.google.protobuf.MapEntry
+              .<java.lang.String, java.lang.String>newDefaultInstance(
+                  org.tensorflow.proto.framework.MetaGraphProtos.internal_static_tensorflow_MetaGraphDef_MetaInfoDef_FunctionAliasesEntry_descriptor, 
+                  com.google.protobuf.WireFormat.FieldType.STRING,
+                  "",
+                  com.google.protobuf.WireFormat.FieldType.STRING,
+                  "");
+    }
+    private com.google.protobuf.MapField<
+        java.lang.String, java.lang.String> functionAliases_;
+    private com.google.protobuf.MapField<java.lang.String, java.lang.String>
+    internalGetFunctionAliases() {
+      if (functionAliases_ == null) {
+        return com.google.protobuf.MapField.emptyMapField(
+            FunctionAliasesDefaultEntryHolder.defaultEntry);
+      }
+      return functionAliases_;
+    }
+
+    public int getFunctionAliasesCount() {
+      return internalGetFunctionAliases().getMap().size();
+    }
+    /**
+     * <pre>
+     * FunctionDef name to aliases mapping.
+     * </pre>
+     *
+     * <code>map&lt;string, string&gt; function_aliases = 8;</code>
+     */
+
+    public boolean containsFunctionAliases(
+        java.lang.String key) {
+      if (key == null) { throw new java.lang.NullPointerException(); }
+      return internalGetFunctionAliases().getMap().containsKey(key);
+    }
+    /**
+     * Use {@link #getFunctionAliasesMap()} instead.
+     */
+    @java.lang.Deprecated
+    public java.util.Map<java.lang.String, java.lang.String> getFunctionAliases() {
+      return getFunctionAliasesMap();
+    }
+    /**
+     * <pre>
+     * FunctionDef name to aliases mapping.
+     * </pre>
+     *
+     * <code>map&lt;string, string&gt; function_aliases = 8;</code>
+     */
+
+    public java.util.Map<java.lang.String, java.lang.String> getFunctionAliasesMap() {
+      return internalGetFunctionAliases().getMap();
+    }
+    /**
+     * <pre>
+     * FunctionDef name to aliases mapping.
+     * </pre>
+     *
+     * <code>map&lt;string, string&gt; function_aliases = 8;</code>
+     */
+
+    public java.lang.String getFunctionAliasesOrDefault(
+        java.lang.String key,
+        java.lang.String defaultValue) {
+      if (key == null) { throw new java.lang.NullPointerException(); }
+      java.util.Map<java.lang.String, java.lang.String> map =
+          internalGetFunctionAliases().getMap();
+      return map.containsKey(key) ? map.get(key) : defaultValue;
+    }
+    /**
+     * <pre>
+     * FunctionDef name to aliases mapping.
+     * </pre>
+     *
+     * <code>map&lt;string, string&gt; function_aliases = 8;</code>
+     */
+
+    public java.lang.String getFunctionAliasesOrThrow(
+        java.lang.String key) {
+      if (key == null) { throw new java.lang.NullPointerException(); }
+      java.util.Map<java.lang.String, java.lang.String> map =
+          internalGetFunctionAliases().getMap();
+      if (!map.containsKey(key)) {
+        throw new java.lang.IllegalArgumentException();
+      }
+      return map.get(key);
+    }
+
+    private byte memoizedIsInitialized = -1;
+    @java.lang.Override
+    public final boolean isInitialized() {
+      byte isInitialized = memoizedIsInitialized;
+      if (isInitialized == 1) return true;
+      if (isInitialized == 0) return false;
+
+      memoizedIsInitialized = 1;
+      return true;
+    }
+
+    @java.lang.Override
+    public void writeTo(com.google.protobuf.CodedOutputStream output)
+                        throws java.io.IOException {
+      if (!getMetaGraphVersionBytes().isEmpty()) {
+        com.google.protobuf.GeneratedMessageV3.writeString(output, 1, metaGraphVersion_);
+      }
+      if (strippedOpList_ != null) {
+        output.writeMessage(2, getStrippedOpList());
+      }
+      if (anyInfo_ != null) {
+        output.writeMessage(3, getAnyInfo());
+      }
+      for (int i = 0; i < tags_.size(); i++) {
+        com.google.protobuf.GeneratedMessageV3.writeString(output, 4, tags_.getRaw(i));
+      }
+      if (!getTensorflowVersionBytes().isEmpty()) {
+        com.google.protobuf.GeneratedMessageV3.writeString(output, 5, tensorflowVersion_);
+      }
+      if (!getTensorflowGitVersionBytes().isEmpty()) {
+        com.google.protobuf.GeneratedMessageV3.writeString(output, 6, tensorflowGitVersion_);
+      }
+      if (strippedDefaultAttrs_ != false) {
+        output.writeBool(7, strippedDefaultAttrs_);
+      }
+      com.google.protobuf.GeneratedMessageV3
+        .serializeStringMapTo(
+          output,
+          internalGetFunctionAliases(),
+          FunctionAliasesDefaultEntryHolder.defaultEntry,
+          8);
+      unknownFields.writeTo(output);
+    }
+
+    @java.lang.Override
+    public int getSerializedSize() {
+      int size = memoizedSize;
+      if (size != -1) return size;
+
+      size = 0;
+      if (!getMetaGraphVersionBytes().isEmpty()) {
+        size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, metaGraphVersion_);
+      }
+      if (strippedOpList_ != null) {
+        size += com.google.protobuf.CodedOutputStream
+          .computeMessageSize(2, getStrippedOpList());
+      }
+      if (anyInfo_ != null) {
+        size += com.google.protobuf.CodedOutputStream
+          .computeMessageSize(3, getAnyInfo());
+      }
+      {
+        int dataSize = 0;
+        for (int i = 0; i < tags_.size(); i++) {
+          dataSize += computeStringSizeNoTag(tags_.getRaw(i));
+        }
+        size += dataSize;
+        size += 1 * getTagsList().size();
+      }
+      if (!getTensorflowVersionBytes().isEmpty()) {
+        size += com.google.protobuf.GeneratedMessageV3.computeStringSize(5, tensorflowVersion_);
+      }
+      if (!getTensorflowGitVersionBytes().isEmpty()) {
+        size += com.google.protobuf.GeneratedMessageV3.computeStringSize(6, tensorflowGitVersion_);
+      }
+      if (strippedDefaultAttrs_ != false) {
+        size += com.google.protobuf.CodedOutputStream
+          .computeBoolSize(7, strippedDefaultAttrs_);
+      }
+      for (java.util.Map.Entry<java.lang.String, java.lang.String> entry
+           : internalGetFunctionAliases().getMap().entrySet()) {
+        com.google.protobuf.MapEntry<java.lang.String, java.lang.String>
+        functionAliases__ = FunctionAliasesDefaultEntryHolder.defaultEntry.newBuilderForType()
+            .setKey(entry.getKey())
+            .setValue(entry.getValue())
+            .build();
+        size += com.google.protobuf.CodedOutputStream
+            .computeMessageSize(8, functionAliases__);
+      }
+      size += unknownFields.getSerializedSize();
+      memoizedSize = size;
+      return size;
+    }
+
+    @java.lang.Override
+    public boolean equals(final java.lang.Object obj) {
+      if (obj == this) {
+       return true;
+      }
+      if (!(obj instanceof org.tensorflow.proto.framework.MetaGraphDef.MetaInfoDef)) {
+        return super.equals(obj);
+      }
+      org.tensorflow.proto.framework.MetaGraphDef.MetaInfoDef other = (org.tensorflow.proto.framework.MetaGraphDef.MetaInfoDef) obj;
+
+      if (!getMetaGraphVersion()
+          .equals(other.getMetaGraphVersion())) return false;
+      if (hasStrippedOpList() != other.hasStrippedOpList()) return false;
+      if (hasStrippedOpList()) {
+        if (!getStrippedOpList()
+            .equals(other.getStrippedOpList())) return false;
+      }
+      if (hasAnyInfo() != other.hasAnyInfo()) return false;
+      if (hasAnyInfo()) {
+        if (!getAnyInfo()
+            .equals(other.getAnyInfo())) return false;
+      }
+      if (!getTagsList()
+          .equals(other.getTagsList())) return false;
+      if (!getTensorflowVersion()
+          .equals(other.getTensorflowVersion())) return false;
+      if (!getTensorflowGitVersion()
+          .equals(other.getTensorflowGitVersion())) return false;
+      if (getStrippedDefaultAttrs()
+          != other.getStrippedDefaultAttrs()) return false;
+      if (!internalGetFunctionAliases().equals(
+          other.internalGetFunctionAliases())) return false;
+      if (!unknownFields.equals(other.unknownFields)) return false;
+      return true;
+    }
+
+    @java.lang.Override
+    public int hashCode() {
+      if (memoizedHashCode != 0) {
+        return memoizedHashCode;
+      }
+      int hash = 41;
+      hash = (19 * hash) + getDescriptor().hashCode();
+      hash = (37 * hash) + META_GRAPH_VERSION_FIELD_NUMBER;
+      hash = (53 * hash) + getMetaGraphVersion().hashCode();
+      if (hasStrippedOpList()) {
+        hash = (37 * hash) + STRIPPED_OP_LIST_FIELD_NUMBER;
+        hash = (53 * hash) + getStrippedOpList().hashCode();
+      }
+      if (hasAnyInfo()) {
+        hash = (37 * hash) + ANY_INFO_FIELD_NUMBER;
+        hash = (53 * hash) + getAnyInfo().hashCode();
+      }
+      if (getTagsCount() > 0) {
+        hash = (37 * hash) + TAGS_FIELD_NUMBER;
+        hash = (53 * hash) + getTagsList().hashCode();
+      }
+      hash = (37 * hash) + TENSORFLOW_VERSION_FIELD_NUMBER;
+      hash = (53 * hash) + getTensorflowVersion().hashCode();
+      hash = (37 * hash) + TENSORFLOW_GIT_VERSION_FIELD_NUMBER;
+      hash = (53 * hash) + getTensorflowGitVersion().hashCode();
+      hash = (37 * hash) + STRIPPED_DEFAULT_ATTRS_FIELD_NUMBER;
+      hash = (53 * hash) + com.google.protobuf.Internal.hashBoolean(
+          getStrippedDefaultAttrs());
+      if (!internalGetFunctionAliases().getMap().isEmpty()) {
+        hash = (37 * hash) + FUNCTION_ALIASES_FIELD_NUMBER;
+        hash = (53 * hash) + internalGetFunctionAliases().hashCode();
+      }
+      hash = (29 * hash) + unknownFields.hashCode();
+      memoizedHashCode = hash;
+      return hash;
+    }
+
+    public static org.tensorflow.proto.framework.MetaGraphDef.MetaInfoDef parseFrom(
+        java.nio.ByteBuffer data)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data);
+    }
+    public static org.tensorflow.proto.framework.MetaGraphDef.MetaInfoDef parseFrom(
+        java.nio.ByteBuffer data,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.MetaGraphDef.MetaInfoDef parseFrom(
+        com.google.protobuf.ByteString data)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data);
+    }
+    public static org.tensorflow.proto.framework.MetaGraphDef.MetaInfoDef parseFrom(
+        com.google.protobuf.ByteString data,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.MetaGraphDef.MetaInfoDef parseFrom(byte[] data)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data);
+    }
+    public static org.tensorflow.proto.framework.MetaGraphDef.MetaInfoDef parseFrom(
+        byte[] data,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.MetaGraphDef.MetaInfoDef parseFrom(java.io.InputStream input)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input);
+    }
+    public static org.tensorflow.proto.framework.MetaGraphDef.MetaInfoDef parseFrom(
+        java.io.InputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.MetaGraphDef.MetaInfoDef parseDelimitedFrom(java.io.InputStream input)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseDelimitedWithIOException(PARSER, input);
+    }
+    public static org.tensorflow.proto.framework.MetaGraphDef.MetaInfoDef parseDelimitedFrom(
+        java.io.InputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.MetaGraphDef.MetaInfoDef parseFrom(
+        com.google.protobuf.CodedInputStream input)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input);
+    }
+    public static org.tensorflow.proto.framework.MetaGraphDef.MetaInfoDef parseFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input, extensionRegistry);
+    }
+
+    @java.lang.Override
+    public Builder newBuilderForType() { return newBuilder(); }
+    public static Builder newBuilder() {
+      return DEFAULT_INSTANCE.toBuilder();
+    }
+    public static Builder newBuilder(org.tensorflow.proto.framework.MetaGraphDef.MetaInfoDef prototype) {
+      return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+    }
+    @java.lang.Override
+    public Builder toBuilder() {
+      return this == DEFAULT_INSTANCE
+          ? new Builder() : new Builder().mergeFrom(this);
+    }
+
+    @java.lang.Override
+    protected Builder newBuilderForType(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      Builder builder = new Builder(parent);
+      return builder;
+    }
+    /**
+     * <pre>
+     * Meta information regarding the graph to be exported.  To be used by users
+     * of this protocol buffer to encode information regarding their meta graph.
+     * </pre>
+     *
+     * Protobuf type {@code tensorflow.MetaGraphDef.MetaInfoDef}
+     */
+    public static final class Builder extends
+        com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+        // @@protoc_insertion_point(builder_implements:tensorflow.MetaGraphDef.MetaInfoDef)
+        org.tensorflow.proto.framework.MetaGraphDef.MetaInfoDefOrBuilder {
+      public static final com.google.protobuf.Descriptors.Descriptor
+          getDescriptor() {
+        return org.tensorflow.proto.framework.MetaGraphProtos.internal_static_tensorflow_MetaGraphDef_MetaInfoDef_descriptor;
+      }
+
+      @SuppressWarnings({"rawtypes"})
+      protected com.google.protobuf.MapField internalGetMapField(
+          int number) {
+        switch (number) {
+          case 8:
+            return internalGetFunctionAliases();
+          default:
+            throw new RuntimeException(
+                "Invalid map field number: " + number);
+        }
+      }
+      @SuppressWarnings({"rawtypes"})
+      protected com.google.protobuf.MapField internalGetMutableMapField(
+          int number) {
+        switch (number) {
+          case 8:
+            return internalGetMutableFunctionAliases();
+          default:
+            throw new RuntimeException(
+                "Invalid map field number: " + number);
+        }
+      }
+      @java.lang.Override
+      protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+          internalGetFieldAccessorTable() {
+        return org.tensorflow.proto.framework.MetaGraphProtos.internal_static_tensorflow_MetaGraphDef_MetaInfoDef_fieldAccessorTable
+            .ensureFieldAccessorsInitialized(
+                org.tensorflow.proto.framework.MetaGraphDef.MetaInfoDef.class, org.tensorflow.proto.framework.MetaGraphDef.MetaInfoDef.Builder.class);
+      }
+
+      // Construct using org.tensorflow.proto.framework.MetaGraphDef.MetaInfoDef.newBuilder()
+      private Builder() {
+        maybeForceBuilderInitialization();
+      }
+
+      private Builder(
+          com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+        super(parent);
+        maybeForceBuilderInitialization();
+      }
+      private void maybeForceBuilderInitialization() {
+        if (com.google.protobuf.GeneratedMessageV3
+                .alwaysUseFieldBuilders) {
+        }
+      }
+      @java.lang.Override
+      public Builder clear() {
+        super.clear();
+        metaGraphVersion_ = "";
+
+        if (strippedOpListBuilder_ == null) {
+          strippedOpList_ = null;
+        } else {
+          strippedOpList_ = null;
+          strippedOpListBuilder_ = null;
+        }
+        if (anyInfoBuilder_ == null) {
+          anyInfo_ = null;
+        } else {
+          anyInfo_ = null;
+          anyInfoBuilder_ = null;
+        }
+        tags_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+        bitField0_ = (bitField0_ & ~0x00000001);
+        tensorflowVersion_ = "";
+
+        tensorflowGitVersion_ = "";
+
+        strippedDefaultAttrs_ = false;
+
+        internalGetMutableFunctionAliases().clear();
+        return this;
+      }
+
+      @java.lang.Override
+      public com.google.protobuf.Descriptors.Descriptor
+          getDescriptorForType() {
+        return org.tensorflow.proto.framework.MetaGraphProtos.internal_static_tensorflow_MetaGraphDef_MetaInfoDef_descriptor;
+      }
+
+      @java.lang.Override
+      public org.tensorflow.proto.framework.MetaGraphDef.MetaInfoDef getDefaultInstanceForType() {
+        return org.tensorflow.proto.framework.MetaGraphDef.MetaInfoDef.getDefaultInstance();
+      }
+
+      @java.lang.Override
+      public org.tensorflow.proto.framework.MetaGraphDef.MetaInfoDef build() {
+        org.tensorflow.proto.framework.MetaGraphDef.MetaInfoDef result = buildPartial();
+        if (!result.isInitialized()) {
+          throw newUninitializedMessageException(result);
+        }
+        return result;
+      }
+
+      @java.lang.Override
+      public org.tensorflow.proto.framework.MetaGraphDef.MetaInfoDef buildPartial() {
+        org.tensorflow.proto.framework.MetaGraphDef.MetaInfoDef result = new org.tensorflow.proto.framework.MetaGraphDef.MetaInfoDef(this);
+        int from_bitField0_ = bitField0_;
+        result.metaGraphVersion_ = metaGraphVersion_;
+        if (strippedOpListBuilder_ == null) {
+          result.strippedOpList_ = strippedOpList_;
+        } else {
+          result.strippedOpList_ = strippedOpListBuilder_.build();
+        }
+        if (anyInfoBuilder_ == null) {
+          result.anyInfo_ = anyInfo_;
+        } else {
+          result.anyInfo_ = anyInfoBuilder_.build();
+        }
+        if (((bitField0_ & 0x00000001) != 0)) {
+          tags_ = tags_.getUnmodifiableView();
+          bitField0_ = (bitField0_ & ~0x00000001);
+        }
+        result.tags_ = tags_;
+        result.tensorflowVersion_ = tensorflowVersion_;
+        result.tensorflowGitVersion_ = tensorflowGitVersion_;
+        result.strippedDefaultAttrs_ = strippedDefaultAttrs_;
+        result.functionAliases_ = internalGetFunctionAliases();
+        result.functionAliases_.makeImmutable();
+        onBuilt();
+        return result;
+      }
+
+      @java.lang.Override
+      public Builder clone() {
+        return super.clone();
+      }
+      @java.lang.Override
+      public Builder setField(
+          com.google.protobuf.Descriptors.FieldDescriptor field,
+          java.lang.Object value) {
+        return super.setField(field, value);
+      }
+      @java.lang.Override
+      public Builder clearField(
+          com.google.protobuf.Descriptors.FieldDescriptor field) {
+        return super.clearField(field);
+      }
+      @java.lang.Override
+      public Builder clearOneof(
+          com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+        return super.clearOneof(oneof);
+      }
+      @java.lang.Override
+      public Builder setRepeatedField(
+          com.google.protobuf.Descriptors.FieldDescriptor field,
+          int index, java.lang.Object value) {
+        return super.setRepeatedField(field, index, value);
+      }
+      @java.lang.Override
+      public Builder addRepeatedField(
+          com.google.protobuf.Descriptors.FieldDescriptor field,
+          java.lang.Object value) {
+        return super.addRepeatedField(field, value);
+      }
+      @java.lang.Override
+      public Builder mergeFrom(com.google.protobuf.Message other) {
+        if (other instanceof org.tensorflow.proto.framework.MetaGraphDef.MetaInfoDef) {
+          return mergeFrom((org.tensorflow.proto.framework.MetaGraphDef.MetaInfoDef)other);
+        } else {
+          super.mergeFrom(other);
+          return this;
+        }
+      }
+
+      public Builder mergeFrom(org.tensorflow.proto.framework.MetaGraphDef.MetaInfoDef other) {
+        if (other == org.tensorflow.proto.framework.MetaGraphDef.MetaInfoDef.getDefaultInstance()) return this;
+        if (!other.getMetaGraphVersion().isEmpty()) {
+          metaGraphVersion_ = other.metaGraphVersion_;
+          onChanged();
+        }
+        if (other.hasStrippedOpList()) {
+          mergeStrippedOpList(other.getStrippedOpList());
+        }
+        if (other.hasAnyInfo()) {
+          mergeAnyInfo(other.getAnyInfo());
+        }
+        if (!other.tags_.isEmpty()) {
+          if (tags_.isEmpty()) {
+            tags_ = other.tags_;
+            bitField0_ = (bitField0_ & ~0x00000001);
+          } else {
+            ensureTagsIsMutable();
+            tags_.addAll(other.tags_);
+          }
+          onChanged();
+        }
+        if (!other.getTensorflowVersion().isEmpty()) {
+          tensorflowVersion_ = other.tensorflowVersion_;
+          onChanged();
+        }
+        if (!other.getTensorflowGitVersion().isEmpty()) {
+          tensorflowGitVersion_ = other.tensorflowGitVersion_;
+          onChanged();
+        }
+        if (other.getStrippedDefaultAttrs() != false) {
+          setStrippedDefaultAttrs(other.getStrippedDefaultAttrs());
+        }
+        internalGetMutableFunctionAliases().mergeFrom(
+            other.internalGetFunctionAliases());
+        this.mergeUnknownFields(other.unknownFields);
+        onChanged();
+        return this;
+      }
+
+      @java.lang.Override
+      public final boolean isInitialized() {
+        return true;
+      }
+
+      @java.lang.Override
+      public Builder mergeFrom(
+          com.google.protobuf.CodedInputStream input,
+          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+          throws java.io.IOException {
+        org.tensorflow.proto.framework.MetaGraphDef.MetaInfoDef parsedMessage = null;
+        try {
+          parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+        } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+          parsedMessage = (org.tensorflow.proto.framework.MetaGraphDef.MetaInfoDef) e.getUnfinishedMessage();
+          throw e.unwrapIOException();
+        } finally {
+          if (parsedMessage != null) {
+            mergeFrom(parsedMessage);
+          }
+        }
+        return this;
+      }
+      private int bitField0_;
+
+      private java.lang.Object metaGraphVersion_ = "";
+      /**
+       * <pre>
+       * User specified Version string. Can be the name of the model and revision,
+       * steps this model has been trained to, etc.
+       * </pre>
+       *
+       * <code>string meta_graph_version = 1;</code>
+       */
+      public java.lang.String getMetaGraphVersion() {
+        java.lang.Object ref = metaGraphVersion_;
+        if (!(ref instanceof java.lang.String)) {
+          com.google.protobuf.ByteString bs =
+              (com.google.protobuf.ByteString) ref;
+          java.lang.String s = bs.toStringUtf8();
+          metaGraphVersion_ = s;
+          return s;
+        } else {
+          return (java.lang.String) ref;
+        }
+      }
+      /**
+       * <pre>
+       * User specified Version string. Can be the name of the model and revision,
+       * steps this model has been trained to, etc.
+       * </pre>
+       *
+       * <code>string meta_graph_version = 1;</code>
+       */
+      public com.google.protobuf.ByteString
+          getMetaGraphVersionBytes() {
+        java.lang.Object ref = metaGraphVersion_;
+        if (ref instanceof String) {
+          com.google.protobuf.ByteString b = 
+              com.google.protobuf.ByteString.copyFromUtf8(
+                  (java.lang.String) ref);
+          metaGraphVersion_ = b;
+          return b;
+        } else {
+          return (com.google.protobuf.ByteString) ref;
+        }
+      }
+      /**
+       * <pre>
+       * User specified Version string. Can be the name of the model and revision,
+       * steps this model has been trained to, etc.
+       * </pre>
+       *
+       * <code>string meta_graph_version = 1;</code>
+       */
+      public Builder setMetaGraphVersion(
+          java.lang.String value) {
+        if (value == null) {
+    throw new NullPointerException();
+  }
+  
+        metaGraphVersion_ = value;
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * User specified Version string. Can be the name of the model and revision,
+       * steps this model has been trained to, etc.
+       * </pre>
+       *
+       * <code>string meta_graph_version = 1;</code>
+       */
+      public Builder clearMetaGraphVersion() {
+        
+        metaGraphVersion_ = getDefaultInstance().getMetaGraphVersion();
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * User specified Version string. Can be the name of the model and revision,
+       * steps this model has been trained to, etc.
+       * </pre>
+       *
+       * <code>string meta_graph_version = 1;</code>
+       */
+      public Builder setMetaGraphVersionBytes(
+          com.google.protobuf.ByteString value) {
+        if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+        
+        metaGraphVersion_ = value;
+        onChanged();
+        return this;
+      }
+
+      private org.tensorflow.proto.framework.OpList strippedOpList_;
+      private com.google.protobuf.SingleFieldBuilderV3<
+          org.tensorflow.proto.framework.OpList, org.tensorflow.proto.framework.OpList.Builder, org.tensorflow.proto.framework.OpListOrBuilder> strippedOpListBuilder_;
+      /**
+       * <pre>
+       * A copy of the OpDefs used by the producer of this graph_def.
+       * Descriptions and Ops not used in graph_def are stripped out.
+       * </pre>
+       *
+       * <code>.tensorflow.OpList stripped_op_list = 2;</code>
+       */
+      public boolean hasStrippedOpList() {
+        return strippedOpListBuilder_ != null || strippedOpList_ != null;
+      }
+      /**
+       * <pre>
+       * A copy of the OpDefs used by the producer of this graph_def.
+       * Descriptions and Ops not used in graph_def are stripped out.
+       * </pre>
+       *
+       * <code>.tensorflow.OpList stripped_op_list = 2;</code>
+       */
+      public org.tensorflow.proto.framework.OpList getStrippedOpList() {
+        if (strippedOpListBuilder_ == null) {
+          return strippedOpList_ == null ? org.tensorflow.proto.framework.OpList.getDefaultInstance() : strippedOpList_;
+        } else {
+          return strippedOpListBuilder_.getMessage();
+        }
+      }
+      /**
+       * <pre>
+       * A copy of the OpDefs used by the producer of this graph_def.
+       * Descriptions and Ops not used in graph_def are stripped out.
+       * </pre>
+       *
+       * <code>.tensorflow.OpList stripped_op_list = 2;</code>
+       */
+      public Builder setStrippedOpList(org.tensorflow.proto.framework.OpList value) {
+        if (strippedOpListBuilder_ == null) {
+          if (value == null) {
+            throw new NullPointerException();
+          }
+          strippedOpList_ = value;
+          onChanged();
+        } else {
+          strippedOpListBuilder_.setMessage(value);
+        }
+
+        return this;
+      }
+      /**
+       * <pre>
+       * A copy of the OpDefs used by the producer of this graph_def.
+       * Descriptions and Ops not used in graph_def are stripped out.
+       * </pre>
+       *
+       * <code>.tensorflow.OpList stripped_op_list = 2;</code>
+       */
+      public Builder setStrippedOpList(
+          org.tensorflow.proto.framework.OpList.Builder builderForValue) {
+        if (strippedOpListBuilder_ == null) {
+          strippedOpList_ = builderForValue.build();
+          onChanged();
+        } else {
+          strippedOpListBuilder_.setMessage(builderForValue.build());
+        }
+
+        return this;
+      }
+      /**
+       * <pre>
+       * A copy of the OpDefs used by the producer of this graph_def.
+       * Descriptions and Ops not used in graph_def are stripped out.
+       * </pre>
+       *
+       * <code>.tensorflow.OpList stripped_op_list = 2;</code>
+       */
+      public Builder mergeStrippedOpList(org.tensorflow.proto.framework.OpList value) {
+        if (strippedOpListBuilder_ == null) {
+          if (strippedOpList_ != null) {
+            strippedOpList_ =
+              org.tensorflow.proto.framework.OpList.newBuilder(strippedOpList_).mergeFrom(value).buildPartial();
+          } else {
+            strippedOpList_ = value;
+          }
+          onChanged();
+        } else {
+          strippedOpListBuilder_.mergeFrom(value);
+        }
+
+        return this;
+      }
+      /**
+       * <pre>
+       * A copy of the OpDefs used by the producer of this graph_def.
+       * Descriptions and Ops not used in graph_def are stripped out.
+       * </pre>
+       *
+       * <code>.tensorflow.OpList stripped_op_list = 2;</code>
+       */
+      public Builder clearStrippedOpList() {
+        if (strippedOpListBuilder_ == null) {
+          strippedOpList_ = null;
+          onChanged();
+        } else {
+          strippedOpList_ = null;
+          strippedOpListBuilder_ = null;
+        }
+
+        return this;
+      }
+      /**
+       * <pre>
+       * A copy of the OpDefs used by the producer of this graph_def.
+       * Descriptions and Ops not used in graph_def are stripped out.
+       * </pre>
+       *
+       * <code>.tensorflow.OpList stripped_op_list = 2;</code>
+       */
+      public org.tensorflow.proto.framework.OpList.Builder getStrippedOpListBuilder() {
+        
+        onChanged();
+        return getStrippedOpListFieldBuilder().getBuilder();
+      }
+      /**
+       * <pre>
+       * A copy of the OpDefs used by the producer of this graph_def.
+       * Descriptions and Ops not used in graph_def are stripped out.
+       * </pre>
+       *
+       * <code>.tensorflow.OpList stripped_op_list = 2;</code>
+       */
+      public org.tensorflow.proto.framework.OpListOrBuilder getStrippedOpListOrBuilder() {
+        if (strippedOpListBuilder_ != null) {
+          return strippedOpListBuilder_.getMessageOrBuilder();
+        } else {
+          return strippedOpList_ == null ?
+              org.tensorflow.proto.framework.OpList.getDefaultInstance() : strippedOpList_;
+        }
+      }
+      /**
+       * <pre>
+       * A copy of the OpDefs used by the producer of this graph_def.
+       * Descriptions and Ops not used in graph_def are stripped out.
+       * </pre>
+       *
+       * <code>.tensorflow.OpList stripped_op_list = 2;</code>
+       */
+      private com.google.protobuf.SingleFieldBuilderV3<
+          org.tensorflow.proto.framework.OpList, org.tensorflow.proto.framework.OpList.Builder, org.tensorflow.proto.framework.OpListOrBuilder> 
+          getStrippedOpListFieldBuilder() {
+        if (strippedOpListBuilder_ == null) {
+          strippedOpListBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+              org.tensorflow.proto.framework.OpList, org.tensorflow.proto.framework.OpList.Builder, org.tensorflow.proto.framework.OpListOrBuilder>(
+                  getStrippedOpList(),
+                  getParentForChildren(),
+                  isClean());
+          strippedOpList_ = null;
+        }
+        return strippedOpListBuilder_;
+      }
+
+      private com.google.protobuf.Any anyInfo_;
+      private com.google.protobuf.SingleFieldBuilderV3<
+          com.google.protobuf.Any, com.google.protobuf.Any.Builder, com.google.protobuf.AnyOrBuilder> anyInfoBuilder_;
+      /**
+       * <pre>
+       * A serialized protobuf. Can be the time this meta graph is created, or
+       * modified, or name of the model.
+       * </pre>
+       *
+       * <code>.google.protobuf.Any any_info = 3;</code>
+       */
+      public boolean hasAnyInfo() {
+        return anyInfoBuilder_ != null || anyInfo_ != null;
+      }
+      /**
+       * <pre>
+       * A serialized protobuf. Can be the time this meta graph is created, or
+       * modified, or name of the model.
+       * </pre>
+       *
+       * <code>.google.protobuf.Any any_info = 3;</code>
+       */
+      public com.google.protobuf.Any getAnyInfo() {
+        if (anyInfoBuilder_ == null) {
+          return anyInfo_ == null ? com.google.protobuf.Any.getDefaultInstance() : anyInfo_;
+        } else {
+          return anyInfoBuilder_.getMessage();
+        }
+      }
+      /**
+       * <pre>
+       * A serialized protobuf. Can be the time this meta graph is created, or
+       * modified, or name of the model.
+       * </pre>
+       *
+       * <code>.google.protobuf.Any any_info = 3;</code>
+       */
+      public Builder setAnyInfo(com.google.protobuf.Any value) {
+        if (anyInfoBuilder_ == null) {
+          if (value == null) {
+            throw new NullPointerException();
+          }
+          anyInfo_ = value;
+          onChanged();
+        } else {
+          anyInfoBuilder_.setMessage(value);
+        }
+
+        return this;
+      }
+      /**
+       * <pre>
+       * A serialized protobuf. Can be the time this meta graph is created, or
+       * modified, or name of the model.
+       * </pre>
+       *
+       * <code>.google.protobuf.Any any_info = 3;</code>
+       */
+      public Builder setAnyInfo(
+          com.google.protobuf.Any.Builder builderForValue) {
+        if (anyInfoBuilder_ == null) {
+          anyInfo_ = builderForValue.build();
+          onChanged();
+        } else {
+          anyInfoBuilder_.setMessage(builderForValue.build());
+        }
+
+        return this;
+      }
+      /**
+       * <pre>
+       * A serialized protobuf. Can be the time this meta graph is created, or
+       * modified, or name of the model.
+       * </pre>
+       *
+       * <code>.google.protobuf.Any any_info = 3;</code>
+       */
+      public Builder mergeAnyInfo(com.google.protobuf.Any value) {
+        if (anyInfoBuilder_ == null) {
+          if (anyInfo_ != null) {
+            anyInfo_ =
+              com.google.protobuf.Any.newBuilder(anyInfo_).mergeFrom(value).buildPartial();
+          } else {
+            anyInfo_ = value;
+          }
+          onChanged();
+        } else {
+          anyInfoBuilder_.mergeFrom(value);
+        }
+
+        return this;
+      }
+      /**
+       * <pre>
+       * A serialized protobuf. Can be the time this meta graph is created, or
+       * modified, or name of the model.
+       * </pre>
+       *
+       * <code>.google.protobuf.Any any_info = 3;</code>
+       */
+      public Builder clearAnyInfo() {
+        if (anyInfoBuilder_ == null) {
+          anyInfo_ = null;
+          onChanged();
+        } else {
+          anyInfo_ = null;
+          anyInfoBuilder_ = null;
+        }
+
+        return this;
+      }
+      /**
+       * <pre>
+       * A serialized protobuf. Can be the time this meta graph is created, or
+       * modified, or name of the model.
+       * </pre>
+       *
+       * <code>.google.protobuf.Any any_info = 3;</code>
+       */
+      public com.google.protobuf.Any.Builder getAnyInfoBuilder() {
+        
+        onChanged();
+        return getAnyInfoFieldBuilder().getBuilder();
+      }
+      /**
+       * <pre>
+       * A serialized protobuf. Can be the time this meta graph is created, or
+       * modified, or name of the model.
+       * </pre>
+       *
+       * <code>.google.protobuf.Any any_info = 3;</code>
+       */
+      public com.google.protobuf.AnyOrBuilder getAnyInfoOrBuilder() {
+        if (anyInfoBuilder_ != null) {
+          return anyInfoBuilder_.getMessageOrBuilder();
+        } else {
+          return anyInfo_ == null ?
+              com.google.protobuf.Any.getDefaultInstance() : anyInfo_;
+        }
+      }
+      /**
+       * <pre>
+       * A serialized protobuf. Can be the time this meta graph is created, or
+       * modified, or name of the model.
+       * </pre>
+       *
+       * <code>.google.protobuf.Any any_info = 3;</code>
+       */
+      private com.google.protobuf.SingleFieldBuilderV3<
+          com.google.protobuf.Any, com.google.protobuf.Any.Builder, com.google.protobuf.AnyOrBuilder> 
+          getAnyInfoFieldBuilder() {
+        if (anyInfoBuilder_ == null) {
+          anyInfoBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+              com.google.protobuf.Any, com.google.protobuf.Any.Builder, com.google.protobuf.AnyOrBuilder>(
+                  getAnyInfo(),
+                  getParentForChildren(),
+                  isClean());
+          anyInfo_ = null;
+        }
+        return anyInfoBuilder_;
+      }
+
+      private com.google.protobuf.LazyStringList tags_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+      private void ensureTagsIsMutable() {
+        if (!((bitField0_ & 0x00000001) != 0)) {
+          tags_ = new com.google.protobuf.LazyStringArrayList(tags_);
+          bitField0_ |= 0x00000001;
+         }
+      }
+      /**
+       * <pre>
+       * User supplied tag(s) on the meta_graph and included graph_def.
+       * MetaGraphDefs should be tagged with their capabilities or use-cases.
+       * Examples: "train", "serve", "gpu", "tpu", etc.
+       * These tags enable loaders to access the MetaGraph(s) appropriate for a
+       * specific use-case or runtime environment.
+       * </pre>
+       *
+       * <code>repeated string tags = 4;</code>
+       */
+      public com.google.protobuf.ProtocolStringList
+          getTagsList() {
+        return tags_.getUnmodifiableView();
+      }
+      /**
+       * <pre>
+       * User supplied tag(s) on the meta_graph and included graph_def.
+       * MetaGraphDefs should be tagged with their capabilities or use-cases.
+       * Examples: "train", "serve", "gpu", "tpu", etc.
+       * These tags enable loaders to access the MetaGraph(s) appropriate for a
+       * specific use-case or runtime environment.
+       * </pre>
+       *
+       * <code>repeated string tags = 4;</code>
+       */
+      public int getTagsCount() {
+        return tags_.size();
+      }
+      /**
+       * <pre>
+       * User supplied tag(s) on the meta_graph and included graph_def.
+       * MetaGraphDefs should be tagged with their capabilities or use-cases.
+       * Examples: "train", "serve", "gpu", "tpu", etc.
+       * These tags enable loaders to access the MetaGraph(s) appropriate for a
+       * specific use-case or runtime environment.
+       * </pre>
+       *
+       * <code>repeated string tags = 4;</code>
+       */
+      public java.lang.String getTags(int index) {
+        return tags_.get(index);
+      }
+      /**
+       * <pre>
+       * User supplied tag(s) on the meta_graph and included graph_def.
+       * MetaGraphDefs should be tagged with their capabilities or use-cases.
+       * Examples: "train", "serve", "gpu", "tpu", etc.
+       * These tags enable loaders to access the MetaGraph(s) appropriate for a
+       * specific use-case or runtime environment.
+       * </pre>
+       *
+       * <code>repeated string tags = 4;</code>
+       */
+      public com.google.protobuf.ByteString
+          getTagsBytes(int index) {
+        return tags_.getByteString(index);
+      }
+      /**
+       * <pre>
+       * User supplied tag(s) on the meta_graph and included graph_def.
+       * MetaGraphDefs should be tagged with their capabilities or use-cases.
+       * Examples: "train", "serve", "gpu", "tpu", etc.
+       * These tags enable loaders to access the MetaGraph(s) appropriate for a
+       * specific use-case or runtime environment.
+       * </pre>
+       *
+       * <code>repeated string tags = 4;</code>
+       */
+      public Builder setTags(
+          int index, java.lang.String value) {
+        if (value == null) {
+    throw new NullPointerException();
+  }
+  ensureTagsIsMutable();
+        tags_.set(index, value);
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * User supplied tag(s) on the meta_graph and included graph_def.
+       * MetaGraphDefs should be tagged with their capabilities or use-cases.
+       * Examples: "train", "serve", "gpu", "tpu", etc.
+       * These tags enable loaders to access the MetaGraph(s) appropriate for a
+       * specific use-case or runtime environment.
+       * </pre>
+       *
+       * <code>repeated string tags = 4;</code>
+       */
+      public Builder addTags(
+          java.lang.String value) {
+        if (value == null) {
+    throw new NullPointerException();
+  }
+  ensureTagsIsMutable();
+        tags_.add(value);
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * User supplied tag(s) on the meta_graph and included graph_def.
+       * MetaGraphDefs should be tagged with their capabilities or use-cases.
+       * Examples: "train", "serve", "gpu", "tpu", etc.
+       * These tags enable loaders to access the MetaGraph(s) appropriate for a
+       * specific use-case or runtime environment.
+       * </pre>
+       *
+       * <code>repeated string tags = 4;</code>
+       */
+      public Builder addAllTags(
+          java.lang.Iterable<java.lang.String> values) {
+        ensureTagsIsMutable();
+        com.google.protobuf.AbstractMessageLite.Builder.addAll(
+            values, tags_);
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * User supplied tag(s) on the meta_graph and included graph_def.
+       * MetaGraphDefs should be tagged with their capabilities or use-cases.
+       * Examples: "train", "serve", "gpu", "tpu", etc.
+       * These tags enable loaders to access the MetaGraph(s) appropriate for a
+       * specific use-case or runtime environment.
+       * </pre>
+       *
+       * <code>repeated string tags = 4;</code>
+       */
+      public Builder clearTags() {
+        tags_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+        bitField0_ = (bitField0_ & ~0x00000001);
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * User supplied tag(s) on the meta_graph and included graph_def.
+       * MetaGraphDefs should be tagged with their capabilities or use-cases.
+       * Examples: "train", "serve", "gpu", "tpu", etc.
+       * These tags enable loaders to access the MetaGraph(s) appropriate for a
+       * specific use-case or runtime environment.
+       * </pre>
+       *
+       * <code>repeated string tags = 4;</code>
+       */
+      public Builder addTagsBytes(
+          com.google.protobuf.ByteString value) {
+        if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+        ensureTagsIsMutable();
+        tags_.add(value);
+        onChanged();
+        return this;
+      }
+
+      private java.lang.Object tensorflowVersion_ = "";
+      /**
+       * <pre>
+       * The __version__ string of the tensorflow build used to write this graph.
+       * This will be populated by the framework, which will overwrite any user
+       * supplied value.
+       * </pre>
+       *
+       * <code>string tensorflow_version = 5;</code>
+       */
+      public java.lang.String getTensorflowVersion() {
+        java.lang.Object ref = tensorflowVersion_;
+        if (!(ref instanceof java.lang.String)) {
+          com.google.protobuf.ByteString bs =
+              (com.google.protobuf.ByteString) ref;
+          java.lang.String s = bs.toStringUtf8();
+          tensorflowVersion_ = s;
+          return s;
+        } else {
+          return (java.lang.String) ref;
+        }
+      }
+      /**
+       * <pre>
+       * The __version__ string of the tensorflow build used to write this graph.
+       * This will be populated by the framework, which will overwrite any user
+       * supplied value.
+       * </pre>
+       *
+       * <code>string tensorflow_version = 5;</code>
+       */
+      public com.google.protobuf.ByteString
+          getTensorflowVersionBytes() {
+        java.lang.Object ref = tensorflowVersion_;
+        if (ref instanceof String) {
+          com.google.protobuf.ByteString b = 
+              com.google.protobuf.ByteString.copyFromUtf8(
+                  (java.lang.String) ref);
+          tensorflowVersion_ = b;
+          return b;
+        } else {
+          return (com.google.protobuf.ByteString) ref;
+        }
+      }
+      /**
+       * <pre>
+       * The __version__ string of the tensorflow build used to write this graph.
+       * This will be populated by the framework, which will overwrite any user
+       * supplied value.
+       * </pre>
+       *
+       * <code>string tensorflow_version = 5;</code>
+       */
+      public Builder setTensorflowVersion(
+          java.lang.String value) {
+        if (value == null) {
+    throw new NullPointerException();
+  }
+  
+        tensorflowVersion_ = value;
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * The __version__ string of the tensorflow build used to write this graph.
+       * This will be populated by the framework, which will overwrite any user
+       * supplied value.
+       * </pre>
+       *
+       * <code>string tensorflow_version = 5;</code>
+       */
+      public Builder clearTensorflowVersion() {
+        
+        tensorflowVersion_ = getDefaultInstance().getTensorflowVersion();
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * The __version__ string of the tensorflow build used to write this graph.
+       * This will be populated by the framework, which will overwrite any user
+       * supplied value.
+       * </pre>
+       *
+       * <code>string tensorflow_version = 5;</code>
+       */
+      public Builder setTensorflowVersionBytes(
+          com.google.protobuf.ByteString value) {
+        if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+        
+        tensorflowVersion_ = value;
+        onChanged();
+        return this;
+      }
+
+      private java.lang.Object tensorflowGitVersion_ = "";
+      /**
+       * <pre>
+       * The __git_version__ string of the tensorflow build used to write this
+       * graph. This will be populated by the framework, which will overwrite any
+       * user supplied value.
+       * </pre>
+       *
+       * <code>string tensorflow_git_version = 6;</code>
+       */
+      public java.lang.String getTensorflowGitVersion() {
+        java.lang.Object ref = tensorflowGitVersion_;
+        if (!(ref instanceof java.lang.String)) {
+          com.google.protobuf.ByteString bs =
+              (com.google.protobuf.ByteString) ref;
+          java.lang.String s = bs.toStringUtf8();
+          tensorflowGitVersion_ = s;
+          return s;
+        } else {
+          return (java.lang.String) ref;
+        }
+      }
+      /**
+       * <pre>
+       * The __git_version__ string of the tensorflow build used to write this
+       * graph. This will be populated by the framework, which will overwrite any
+       * user supplied value.
+       * </pre>
+       *
+       * <code>string tensorflow_git_version = 6;</code>
+       */
+      public com.google.protobuf.ByteString
+          getTensorflowGitVersionBytes() {
+        java.lang.Object ref = tensorflowGitVersion_;
+        if (ref instanceof String) {
+          com.google.protobuf.ByteString b = 
+              com.google.protobuf.ByteString.copyFromUtf8(
+                  (java.lang.String) ref);
+          tensorflowGitVersion_ = b;
+          return b;
+        } else {
+          return (com.google.protobuf.ByteString) ref;
+        }
+      }
+      /**
+       * <pre>
+       * The __git_version__ string of the tensorflow build used to write this
+       * graph. This will be populated by the framework, which will overwrite any
+       * user supplied value.
+       * </pre>
+       *
+       * <code>string tensorflow_git_version = 6;</code>
+       */
+      public Builder setTensorflowGitVersion(
+          java.lang.String value) {
+        if (value == null) {
+    throw new NullPointerException();
+  }
+  
+        tensorflowGitVersion_ = value;
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * The __git_version__ string of the tensorflow build used to write this
+       * graph. This will be populated by the framework, which will overwrite any
+       * user supplied value.
+       * </pre>
+       *
+       * <code>string tensorflow_git_version = 6;</code>
+       */
+      public Builder clearTensorflowGitVersion() {
+        
+        tensorflowGitVersion_ = getDefaultInstance().getTensorflowGitVersion();
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * The __git_version__ string of the tensorflow build used to write this
+       * graph. This will be populated by the framework, which will overwrite any
+       * user supplied value.
+       * </pre>
+       *
+       * <code>string tensorflow_git_version = 6;</code>
+       */
+      public Builder setTensorflowGitVersionBytes(
+          com.google.protobuf.ByteString value) {
+        if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+        
+        tensorflowGitVersion_ = value;
+        onChanged();
+        return this;
+      }
+
+      private boolean strippedDefaultAttrs_ ;
+      /**
+       * <pre>
+       * A flag to denote whether default-valued attrs have been stripped from
+       * the nodes in this graph_def.
+       * </pre>
+       *
+       * <code>bool stripped_default_attrs = 7;</code>
+       */
+      public boolean getStrippedDefaultAttrs() {
+        return strippedDefaultAttrs_;
+      }
+      /**
+       * <pre>
+       * A flag to denote whether default-valued attrs have been stripped from
+       * the nodes in this graph_def.
+       * </pre>
+       *
+       * <code>bool stripped_default_attrs = 7;</code>
+       */
+      public Builder setStrippedDefaultAttrs(boolean value) {
+        
+        strippedDefaultAttrs_ = value;
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * A flag to denote whether default-valued attrs have been stripped from
+       * the nodes in this graph_def.
+       * </pre>
+       *
+       * <code>bool stripped_default_attrs = 7;</code>
+       */
+      public Builder clearStrippedDefaultAttrs() {
+        
+        strippedDefaultAttrs_ = false;
+        onChanged();
+        return this;
+      }
+
+      private com.google.protobuf.MapField<
+          java.lang.String, java.lang.String> functionAliases_;
+      private com.google.protobuf.MapField<java.lang.String, java.lang.String>
+      internalGetFunctionAliases() {
+        if (functionAliases_ == null) {
+          return com.google.protobuf.MapField.emptyMapField(
+              FunctionAliasesDefaultEntryHolder.defaultEntry);
+        }
+        return functionAliases_;
+      }
+      private com.google.protobuf.MapField<java.lang.String, java.lang.String>
+      internalGetMutableFunctionAliases() {
+        onChanged();;
+        if (functionAliases_ == null) {
+          functionAliases_ = com.google.protobuf.MapField.newMapField(
+              FunctionAliasesDefaultEntryHolder.defaultEntry);
+        }
+        if (!functionAliases_.isMutable()) {
+          functionAliases_ = functionAliases_.copy();
+        }
+        return functionAliases_;
+      }
+
+      public int getFunctionAliasesCount() {
+        return internalGetFunctionAliases().getMap().size();
+      }
+      /**
+       * <pre>
+       * FunctionDef name to aliases mapping.
+       * </pre>
+       *
+       * <code>map&lt;string, string&gt; function_aliases = 8;</code>
+       */
+
+      public boolean containsFunctionAliases(
+          java.lang.String key) {
+        if (key == null) { throw new java.lang.NullPointerException(); }
+        return internalGetFunctionAliases().getMap().containsKey(key);
+      }
+      /**
+       * Use {@link #getFunctionAliasesMap()} instead.
+       */
+      @java.lang.Deprecated
+      public java.util.Map<java.lang.String, java.lang.String> getFunctionAliases() {
+        return getFunctionAliasesMap();
+      }
+      /**
+       * <pre>
+       * FunctionDef name to aliases mapping.
+       * </pre>
+       *
+       * <code>map&lt;string, string&gt; function_aliases = 8;</code>
+       */
+
+      public java.util.Map<java.lang.String, java.lang.String> getFunctionAliasesMap() {
+        return internalGetFunctionAliases().getMap();
+      }
+      /**
+       * <pre>
+       * FunctionDef name to aliases mapping.
+       * </pre>
+       *
+       * <code>map&lt;string, string&gt; function_aliases = 8;</code>
+       */
+
+      public java.lang.String getFunctionAliasesOrDefault(
+          java.lang.String key,
+          java.lang.String defaultValue) {
+        if (key == null) { throw new java.lang.NullPointerException(); }
+        java.util.Map<java.lang.String, java.lang.String> map =
+            internalGetFunctionAliases().getMap();
+        return map.containsKey(key) ? map.get(key) : defaultValue;
+      }
+      /**
+       * <pre>
+       * FunctionDef name to aliases mapping.
+       * </pre>
+       *
+       * <code>map&lt;string, string&gt; function_aliases = 8;</code>
+       */
+
+      public java.lang.String getFunctionAliasesOrThrow(
+          java.lang.String key) {
+        if (key == null) { throw new java.lang.NullPointerException(); }
+        java.util.Map<java.lang.String, java.lang.String> map =
+            internalGetFunctionAliases().getMap();
+        if (!map.containsKey(key)) {
+          throw new java.lang.IllegalArgumentException();
+        }
+        return map.get(key);
+      }
+
+      public Builder clearFunctionAliases() {
+        internalGetMutableFunctionAliases().getMutableMap()
+            .clear();
+        return this;
+      }
+      /**
+       * <pre>
+       * FunctionDef name to aliases mapping.
+       * </pre>
+       *
+       * <code>map&lt;string, string&gt; function_aliases = 8;</code>
+       */
+
+      public Builder removeFunctionAliases(
+          java.lang.String key) {
+        if (key == null) { throw new java.lang.NullPointerException(); }
+        internalGetMutableFunctionAliases().getMutableMap()
+            .remove(key);
+        return this;
+      }
+      /**
+       * Use alternate mutation accessors instead.
+       */
+      @java.lang.Deprecated
+      public java.util.Map<java.lang.String, java.lang.String>
+      getMutableFunctionAliases() {
+        return internalGetMutableFunctionAliases().getMutableMap();
+      }
+      /**
+       * <pre>
+       * FunctionDef name to aliases mapping.
+       * </pre>
+       *
+       * <code>map&lt;string, string&gt; function_aliases = 8;</code>
+       */
+      public Builder putFunctionAliases(
+          java.lang.String key,
+          java.lang.String value) {
+        if (key == null) { throw new java.lang.NullPointerException(); }
+        if (value == null) { throw new java.lang.NullPointerException(); }
+        internalGetMutableFunctionAliases().getMutableMap()
+            .put(key, value);
+        return this;
+      }
+      /**
+       * <pre>
+       * FunctionDef name to aliases mapping.
+       * </pre>
+       *
+       * <code>map&lt;string, string&gt; function_aliases = 8;</code>
+       */
+
+      public Builder putAllFunctionAliases(
+          java.util.Map<java.lang.String, java.lang.String> values) {
+        internalGetMutableFunctionAliases().getMutableMap()
+            .putAll(values);
+        return this;
+      }
+      @java.lang.Override
+      public final Builder setUnknownFields(
+          final com.google.protobuf.UnknownFieldSet unknownFields) {
+        return super.setUnknownFields(unknownFields);
+      }
+
+      @java.lang.Override
+      public final Builder mergeUnknownFields(
+          final com.google.protobuf.UnknownFieldSet unknownFields) {
+        return super.mergeUnknownFields(unknownFields);
+      }
+
+
+      // @@protoc_insertion_point(builder_scope:tensorflow.MetaGraphDef.MetaInfoDef)
+    }
+
+    // @@protoc_insertion_point(class_scope:tensorflow.MetaGraphDef.MetaInfoDef)
+    private static final org.tensorflow.proto.framework.MetaGraphDef.MetaInfoDef DEFAULT_INSTANCE;
+    static {
+      DEFAULT_INSTANCE = new org.tensorflow.proto.framework.MetaGraphDef.MetaInfoDef();
+    }
+
+    public static org.tensorflow.proto.framework.MetaGraphDef.MetaInfoDef getDefaultInstance() {
+      return DEFAULT_INSTANCE;
+    }
+
+    private static final com.google.protobuf.Parser<MetaInfoDef>
+        PARSER = new com.google.protobuf.AbstractParser<MetaInfoDef>() {
+      @java.lang.Override
+      public MetaInfoDef parsePartialFrom(
+          com.google.protobuf.CodedInputStream input,
+          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+          throws com.google.protobuf.InvalidProtocolBufferException {
+        return new MetaInfoDef(input, extensionRegistry);
+      }
+    };
+
+    public static com.google.protobuf.Parser<MetaInfoDef> parser() {
+      return PARSER;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Parser<MetaInfoDef> getParserForType() {
+      return PARSER;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.MetaGraphDef.MetaInfoDef getDefaultInstanceForType() {
+      return DEFAULT_INSTANCE;
+    }
+
+  }
+
+  public static final int META_INFO_DEF_FIELD_NUMBER = 1;
+  private org.tensorflow.proto.framework.MetaGraphDef.MetaInfoDef metaInfoDef_;
+  /**
+   * <code>.tensorflow.MetaGraphDef.MetaInfoDef meta_info_def = 1;</code>
+   */
+  public boolean hasMetaInfoDef() {
+    return metaInfoDef_ != null;
+  }
+  /**
+   * <code>.tensorflow.MetaGraphDef.MetaInfoDef meta_info_def = 1;</code>
+   */
+  public org.tensorflow.proto.framework.MetaGraphDef.MetaInfoDef getMetaInfoDef() {
+    return metaInfoDef_ == null ? org.tensorflow.proto.framework.MetaGraphDef.MetaInfoDef.getDefaultInstance() : metaInfoDef_;
+  }
+  /**
+   * <code>.tensorflow.MetaGraphDef.MetaInfoDef meta_info_def = 1;</code>
+   */
+  public org.tensorflow.proto.framework.MetaGraphDef.MetaInfoDefOrBuilder getMetaInfoDefOrBuilder() {
+    return getMetaInfoDef();
+  }
+
+  public static final int GRAPH_DEF_FIELD_NUMBER = 2;
+  private org.tensorflow.proto.framework.GraphDef graphDef_;
+  /**
+   * <pre>
+   * GraphDef.
+   * </pre>
+   *
+   * <code>.tensorflow.GraphDef graph_def = 2;</code>
+   */
+  public boolean hasGraphDef() {
+    return graphDef_ != null;
+  }
+  /**
+   * <pre>
+   * GraphDef.
+   * </pre>
+   *
+   * <code>.tensorflow.GraphDef graph_def = 2;</code>
+   */
+  public org.tensorflow.proto.framework.GraphDef getGraphDef() {
+    return graphDef_ == null ? org.tensorflow.proto.framework.GraphDef.getDefaultInstance() : graphDef_;
+  }
+  /**
+   * <pre>
+   * GraphDef.
+   * </pre>
+   *
+   * <code>.tensorflow.GraphDef graph_def = 2;</code>
+   */
+  public org.tensorflow.proto.framework.GraphDefOrBuilder getGraphDefOrBuilder() {
+    return getGraphDef();
+  }
+
+  public static final int SAVER_DEF_FIELD_NUMBER = 3;
+  private org.tensorflow.proto.util.SaverDef saverDef_;
+  /**
+   * <pre>
+   * SaverDef.
+   * </pre>
+   *
+   * <code>.tensorflow.SaverDef saver_def = 3;</code>
+   */
+  public boolean hasSaverDef() {
+    return saverDef_ != null;
+  }
+  /**
+   * <pre>
+   * SaverDef.
+   * </pre>
+   *
+   * <code>.tensorflow.SaverDef saver_def = 3;</code>
+   */
+  public org.tensorflow.proto.util.SaverDef getSaverDef() {
+    return saverDef_ == null ? org.tensorflow.proto.util.SaverDef.getDefaultInstance() : saverDef_;
+  }
+  /**
+   * <pre>
+   * SaverDef.
+   * </pre>
+   *
+   * <code>.tensorflow.SaverDef saver_def = 3;</code>
+   */
+  public org.tensorflow.proto.util.SaverDefOrBuilder getSaverDefOrBuilder() {
+    return getSaverDef();
+  }
+
+  public static final int COLLECTION_DEF_FIELD_NUMBER = 4;
+  private static final class CollectionDefDefaultEntryHolder {
+    static final com.google.protobuf.MapEntry<
+        java.lang.String, org.tensorflow.proto.framework.CollectionDef> defaultEntry =
+            com.google.protobuf.MapEntry
+            .<java.lang.String, org.tensorflow.proto.framework.CollectionDef>newDefaultInstance(
+                org.tensorflow.proto.framework.MetaGraphProtos.internal_static_tensorflow_MetaGraphDef_CollectionDefEntry_descriptor, 
+                com.google.protobuf.WireFormat.FieldType.STRING,
+                "",
+                com.google.protobuf.WireFormat.FieldType.MESSAGE,
+                org.tensorflow.proto.framework.CollectionDef.getDefaultInstance());
+  }
+  private com.google.protobuf.MapField<
+      java.lang.String, org.tensorflow.proto.framework.CollectionDef> collectionDef_;
+  private com.google.protobuf.MapField<java.lang.String, org.tensorflow.proto.framework.CollectionDef>
+  internalGetCollectionDef() {
+    if (collectionDef_ == null) {
+      return com.google.protobuf.MapField.emptyMapField(
+          CollectionDefDefaultEntryHolder.defaultEntry);
+    }
+    return collectionDef_;
+  }
+
+  public int getCollectionDefCount() {
+    return internalGetCollectionDef().getMap().size();
+  }
+  /**
+   * <pre>
+   * collection_def: Map from collection name to collections.
+   * See CollectionDef section for details.
+   * </pre>
+   *
+   * <code>map&lt;string, .tensorflow.CollectionDef&gt; collection_def = 4;</code>
+   */
+
+  public boolean containsCollectionDef(
+      java.lang.String key) {
+    if (key == null) { throw new java.lang.NullPointerException(); }
+    return internalGetCollectionDef().getMap().containsKey(key);
+  }
+  /**
+   * Use {@link #getCollectionDefMap()} instead.
+   */
+  @java.lang.Deprecated
+  public java.util.Map<java.lang.String, org.tensorflow.proto.framework.CollectionDef> getCollectionDef() {
+    return getCollectionDefMap();
+  }
+  /**
+   * <pre>
+   * collection_def: Map from collection name to collections.
+   * See CollectionDef section for details.
+   * </pre>
+   *
+   * <code>map&lt;string, .tensorflow.CollectionDef&gt; collection_def = 4;</code>
+   */
+
+  public java.util.Map<java.lang.String, org.tensorflow.proto.framework.CollectionDef> getCollectionDefMap() {
+    return internalGetCollectionDef().getMap();
+  }
+  /**
+   * <pre>
+   * collection_def: Map from collection name to collections.
+   * See CollectionDef section for details.
+   * </pre>
+   *
+   * <code>map&lt;string, .tensorflow.CollectionDef&gt; collection_def = 4;</code>
+   */
+
+  public org.tensorflow.proto.framework.CollectionDef getCollectionDefOrDefault(
+      java.lang.String key,
+      org.tensorflow.proto.framework.CollectionDef defaultValue) {
+    if (key == null) { throw new java.lang.NullPointerException(); }
+    java.util.Map<java.lang.String, org.tensorflow.proto.framework.CollectionDef> map =
+        internalGetCollectionDef().getMap();
+    return map.containsKey(key) ? map.get(key) : defaultValue;
+  }
+  /**
+   * <pre>
+   * collection_def: Map from collection name to collections.
+   * See CollectionDef section for details.
+   * </pre>
+   *
+   * <code>map&lt;string, .tensorflow.CollectionDef&gt; collection_def = 4;</code>
+   */
+
+  public org.tensorflow.proto.framework.CollectionDef getCollectionDefOrThrow(
+      java.lang.String key) {
+    if (key == null) { throw new java.lang.NullPointerException(); }
+    java.util.Map<java.lang.String, org.tensorflow.proto.framework.CollectionDef> map =
+        internalGetCollectionDef().getMap();
+    if (!map.containsKey(key)) {
+      throw new java.lang.IllegalArgumentException();
+    }
+    return map.get(key);
+  }
+
+  public static final int SIGNATURE_DEF_FIELD_NUMBER = 5;
+  private static final class SignatureDefDefaultEntryHolder {
+    static final com.google.protobuf.MapEntry<
+        java.lang.String, org.tensorflow.proto.framework.SignatureDef> defaultEntry =
+            com.google.protobuf.MapEntry
+            .<java.lang.String, org.tensorflow.proto.framework.SignatureDef>newDefaultInstance(
+                org.tensorflow.proto.framework.MetaGraphProtos.internal_static_tensorflow_MetaGraphDef_SignatureDefEntry_descriptor, 
+                com.google.protobuf.WireFormat.FieldType.STRING,
+                "",
+                com.google.protobuf.WireFormat.FieldType.MESSAGE,
+                org.tensorflow.proto.framework.SignatureDef.getDefaultInstance());
+  }
+  private com.google.protobuf.MapField<
+      java.lang.String, org.tensorflow.proto.framework.SignatureDef> signatureDef_;
+  private com.google.protobuf.MapField<java.lang.String, org.tensorflow.proto.framework.SignatureDef>
+  internalGetSignatureDef() {
+    if (signatureDef_ == null) {
+      return com.google.protobuf.MapField.emptyMapField(
+          SignatureDefDefaultEntryHolder.defaultEntry);
+    }
+    return signatureDef_;
+  }
+
+  public int getSignatureDefCount() {
+    return internalGetSignatureDef().getMap().size();
+  }
+  /**
+   * <pre>
+   * signature_def: Map from user supplied key for a signature to a single
+   * SignatureDef.
+   * </pre>
+   *
+   * <code>map&lt;string, .tensorflow.SignatureDef&gt; signature_def = 5;</code>
+   */
+
+  public boolean containsSignatureDef(
+      java.lang.String key) {
+    if (key == null) { throw new java.lang.NullPointerException(); }
+    return internalGetSignatureDef().getMap().containsKey(key);
+  }
+  /**
+   * Use {@link #getSignatureDefMap()} instead.
+   */
+  @java.lang.Deprecated
+  public java.util.Map<java.lang.String, org.tensorflow.proto.framework.SignatureDef> getSignatureDef() {
+    return getSignatureDefMap();
+  }
+  /**
+   * <pre>
+   * signature_def: Map from user supplied key for a signature to a single
+   * SignatureDef.
+   * </pre>
+   *
+   * <code>map&lt;string, .tensorflow.SignatureDef&gt; signature_def = 5;</code>
+   */
+
+  public java.util.Map<java.lang.String, org.tensorflow.proto.framework.SignatureDef> getSignatureDefMap() {
+    return internalGetSignatureDef().getMap();
+  }
+  /**
+   * <pre>
+   * signature_def: Map from user supplied key for a signature to a single
+   * SignatureDef.
+   * </pre>
+   *
+   * <code>map&lt;string, .tensorflow.SignatureDef&gt; signature_def = 5;</code>
+   */
+
+  public org.tensorflow.proto.framework.SignatureDef getSignatureDefOrDefault(
+      java.lang.String key,
+      org.tensorflow.proto.framework.SignatureDef defaultValue) {
+    if (key == null) { throw new java.lang.NullPointerException(); }
+    java.util.Map<java.lang.String, org.tensorflow.proto.framework.SignatureDef> map =
+        internalGetSignatureDef().getMap();
+    return map.containsKey(key) ? map.get(key) : defaultValue;
+  }
+  /**
+   * <pre>
+   * signature_def: Map from user supplied key for a signature to a single
+   * SignatureDef.
+   * </pre>
+   *
+   * <code>map&lt;string, .tensorflow.SignatureDef&gt; signature_def = 5;</code>
+   */
+
+  public org.tensorflow.proto.framework.SignatureDef getSignatureDefOrThrow(
+      java.lang.String key) {
+    if (key == null) { throw new java.lang.NullPointerException(); }
+    java.util.Map<java.lang.String, org.tensorflow.proto.framework.SignatureDef> map =
+        internalGetSignatureDef().getMap();
+    if (!map.containsKey(key)) {
+      throw new java.lang.IllegalArgumentException();
+    }
+    return map.get(key);
+  }
+
+  public static final int ASSET_FILE_DEF_FIELD_NUMBER = 6;
+  private java.util.List<org.tensorflow.proto.framework.AssetFileDef> assetFileDef_;
+  /**
+   * <pre>
+   * Asset file def to be used with the defined graph.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.AssetFileDef asset_file_def = 6;</code>
+   */
+  public java.util.List<org.tensorflow.proto.framework.AssetFileDef> getAssetFileDefList() {
+    return assetFileDef_;
+  }
+  /**
+   * <pre>
+   * Asset file def to be used with the defined graph.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.AssetFileDef asset_file_def = 6;</code>
+   */
+  public java.util.List<? extends org.tensorflow.proto.framework.AssetFileDefOrBuilder> 
+      getAssetFileDefOrBuilderList() {
+    return assetFileDef_;
+  }
+  /**
+   * <pre>
+   * Asset file def to be used with the defined graph.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.AssetFileDef asset_file_def = 6;</code>
+   */
+  public int getAssetFileDefCount() {
+    return assetFileDef_.size();
+  }
+  /**
+   * <pre>
+   * Asset file def to be used with the defined graph.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.AssetFileDef asset_file_def = 6;</code>
+   */
+  public org.tensorflow.proto.framework.AssetFileDef getAssetFileDef(int index) {
+    return assetFileDef_.get(index);
+  }
+  /**
+   * <pre>
+   * Asset file def to be used with the defined graph.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.AssetFileDef asset_file_def = 6;</code>
+   */
+  public org.tensorflow.proto.framework.AssetFileDefOrBuilder getAssetFileDefOrBuilder(
+      int index) {
+    return assetFileDef_.get(index);
+  }
+
+  public static final int OBJECT_GRAPH_DEF_FIELD_NUMBER = 7;
+  private org.tensorflow.proto.framework.SavedObjectGraph objectGraphDef_;
+  /**
+   * <pre>
+   * Extra information about the structure of functions and stateful objects.
+   * </pre>
+   *
+   * <code>.tensorflow.SavedObjectGraph object_graph_def = 7;</code>
+   */
+  public boolean hasObjectGraphDef() {
+    return objectGraphDef_ != null;
+  }
+  /**
+   * <pre>
+   * Extra information about the structure of functions and stateful objects.
+   * </pre>
+   *
+   * <code>.tensorflow.SavedObjectGraph object_graph_def = 7;</code>
+   */
+  public org.tensorflow.proto.framework.SavedObjectGraph getObjectGraphDef() {
+    return objectGraphDef_ == null ? org.tensorflow.proto.framework.SavedObjectGraph.getDefaultInstance() : objectGraphDef_;
+  }
+  /**
+   * <pre>
+   * Extra information about the structure of functions and stateful objects.
+   * </pre>
+   *
+   * <code>.tensorflow.SavedObjectGraph object_graph_def = 7;</code>
+   */
+  public org.tensorflow.proto.framework.SavedObjectGraphOrBuilder getObjectGraphDefOrBuilder() {
+    return getObjectGraphDef();
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    if (metaInfoDef_ != null) {
+      output.writeMessage(1, getMetaInfoDef());
+    }
+    if (graphDef_ != null) {
+      output.writeMessage(2, getGraphDef());
+    }
+    if (saverDef_ != null) {
+      output.writeMessage(3, getSaverDef());
+    }
+    com.google.protobuf.GeneratedMessageV3
+      .serializeStringMapTo(
+        output,
+        internalGetCollectionDef(),
+        CollectionDefDefaultEntryHolder.defaultEntry,
+        4);
+    com.google.protobuf.GeneratedMessageV3
+      .serializeStringMapTo(
+        output,
+        internalGetSignatureDef(),
+        SignatureDefDefaultEntryHolder.defaultEntry,
+        5);
+    for (int i = 0; i < assetFileDef_.size(); i++) {
+      output.writeMessage(6, assetFileDef_.get(i));
+    }
+    if (objectGraphDef_ != null) {
+      output.writeMessage(7, getObjectGraphDef());
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (metaInfoDef_ != null) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(1, getMetaInfoDef());
+    }
+    if (graphDef_ != null) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(2, getGraphDef());
+    }
+    if (saverDef_ != null) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(3, getSaverDef());
+    }
+    for (java.util.Map.Entry<java.lang.String, org.tensorflow.proto.framework.CollectionDef> entry
+         : internalGetCollectionDef().getMap().entrySet()) {
+      com.google.protobuf.MapEntry<java.lang.String, org.tensorflow.proto.framework.CollectionDef>
+      collectionDef__ = CollectionDefDefaultEntryHolder.defaultEntry.newBuilderForType()
+          .setKey(entry.getKey())
+          .setValue(entry.getValue())
+          .build();
+      size += com.google.protobuf.CodedOutputStream
+          .computeMessageSize(4, collectionDef__);
+    }
+    for (java.util.Map.Entry<java.lang.String, org.tensorflow.proto.framework.SignatureDef> entry
+         : internalGetSignatureDef().getMap().entrySet()) {
+      com.google.protobuf.MapEntry<java.lang.String, org.tensorflow.proto.framework.SignatureDef>
+      signatureDef__ = SignatureDefDefaultEntryHolder.defaultEntry.newBuilderForType()
+          .setKey(entry.getKey())
+          .setValue(entry.getValue())
+          .build();
+      size += com.google.protobuf.CodedOutputStream
+          .computeMessageSize(5, signatureDef__);
+    }
+    for (int i = 0; i < assetFileDef_.size(); i++) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(6, assetFileDef_.get(i));
+    }
+    if (objectGraphDef_ != null) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(7, getObjectGraphDef());
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.framework.MetaGraphDef)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.framework.MetaGraphDef other = (org.tensorflow.proto.framework.MetaGraphDef) obj;
+
+    if (hasMetaInfoDef() != other.hasMetaInfoDef()) return false;
+    if (hasMetaInfoDef()) {
+      if (!getMetaInfoDef()
+          .equals(other.getMetaInfoDef())) return false;
+    }
+    if (hasGraphDef() != other.hasGraphDef()) return false;
+    if (hasGraphDef()) {
+      if (!getGraphDef()
+          .equals(other.getGraphDef())) return false;
+    }
+    if (hasSaverDef() != other.hasSaverDef()) return false;
+    if (hasSaverDef()) {
+      if (!getSaverDef()
+          .equals(other.getSaverDef())) return false;
+    }
+    if (!internalGetCollectionDef().equals(
+        other.internalGetCollectionDef())) return false;
+    if (!internalGetSignatureDef().equals(
+        other.internalGetSignatureDef())) return false;
+    if (!getAssetFileDefList()
+        .equals(other.getAssetFileDefList())) return false;
+    if (hasObjectGraphDef() != other.hasObjectGraphDef()) return false;
+    if (hasObjectGraphDef()) {
+      if (!getObjectGraphDef()
+          .equals(other.getObjectGraphDef())) return false;
+    }
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    if (hasMetaInfoDef()) {
+      hash = (37 * hash) + META_INFO_DEF_FIELD_NUMBER;
+      hash = (53 * hash) + getMetaInfoDef().hashCode();
+    }
+    if (hasGraphDef()) {
+      hash = (37 * hash) + GRAPH_DEF_FIELD_NUMBER;
+      hash = (53 * hash) + getGraphDef().hashCode();
+    }
+    if (hasSaverDef()) {
+      hash = (37 * hash) + SAVER_DEF_FIELD_NUMBER;
+      hash = (53 * hash) + getSaverDef().hashCode();
+    }
+    if (!internalGetCollectionDef().getMap().isEmpty()) {
+      hash = (37 * hash) + COLLECTION_DEF_FIELD_NUMBER;
+      hash = (53 * hash) + internalGetCollectionDef().hashCode();
+    }
+    if (!internalGetSignatureDef().getMap().isEmpty()) {
+      hash = (37 * hash) + SIGNATURE_DEF_FIELD_NUMBER;
+      hash = (53 * hash) + internalGetSignatureDef().hashCode();
+    }
+    if (getAssetFileDefCount() > 0) {
+      hash = (37 * hash) + ASSET_FILE_DEF_FIELD_NUMBER;
+      hash = (53 * hash) + getAssetFileDefList().hashCode();
+    }
+    if (hasObjectGraphDef()) {
+      hash = (37 * hash) + OBJECT_GRAPH_DEF_FIELD_NUMBER;
+      hash = (53 * hash) + getObjectGraphDef().hashCode();
+    }
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.framework.MetaGraphDef parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.MetaGraphDef parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.MetaGraphDef parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.MetaGraphDef parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.MetaGraphDef parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.MetaGraphDef parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.MetaGraphDef parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.MetaGraphDef parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.MetaGraphDef parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.MetaGraphDef parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.MetaGraphDef parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.MetaGraphDef parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.framework.MetaGraphDef prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * <pre>
+   * NOTE: This protocol buffer is evolving, and will go through revisions in the
+   * coming months.
+   * Protocol buffer containing the following which are necessary to restart
+   * training, run inference. It can be used to serialize/de-serialize memory
+   * objects necessary for running computation in a graph when crossing the
+   * process boundary. It can be used for long term storage of graphs,
+   * cross-language execution of graphs, etc.
+   *   MetaInfoDef
+   *   GraphDef
+   *   SaverDef
+   *   CollectionDef
+   *   TensorInfo
+   *   SignatureDef
+   * </pre>
+   *
+   * Protobuf type {@code tensorflow.MetaGraphDef}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.MetaGraphDef)
+      org.tensorflow.proto.framework.MetaGraphDefOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.MetaGraphProtos.internal_static_tensorflow_MetaGraphDef_descriptor;
+    }
+
+    @SuppressWarnings({"rawtypes"})
+    protected com.google.protobuf.MapField internalGetMapField(
+        int number) {
+      switch (number) {
+        case 4:
+          return internalGetCollectionDef();
+        case 5:
+          return internalGetSignatureDef();
+        default:
+          throw new RuntimeException(
+              "Invalid map field number: " + number);
+      }
+    }
+    @SuppressWarnings({"rawtypes"})
+    protected com.google.protobuf.MapField internalGetMutableMapField(
+        int number) {
+      switch (number) {
+        case 4:
+          return internalGetMutableCollectionDef();
+        case 5:
+          return internalGetMutableSignatureDef();
+        default:
+          throw new RuntimeException(
+              "Invalid map field number: " + number);
+      }
+    }
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.MetaGraphProtos.internal_static_tensorflow_MetaGraphDef_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.MetaGraphDef.class, org.tensorflow.proto.framework.MetaGraphDef.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.framework.MetaGraphDef.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+        getAssetFileDefFieldBuilder();
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      if (metaInfoDefBuilder_ == null) {
+        metaInfoDef_ = null;
+      } else {
+        metaInfoDef_ = null;
+        metaInfoDefBuilder_ = null;
+      }
+      if (graphDefBuilder_ == null) {
+        graphDef_ = null;
+      } else {
+        graphDef_ = null;
+        graphDefBuilder_ = null;
+      }
+      if (saverDefBuilder_ == null) {
+        saverDef_ = null;
+      } else {
+        saverDef_ = null;
+        saverDefBuilder_ = null;
+      }
+      internalGetMutableCollectionDef().clear();
+      internalGetMutableSignatureDef().clear();
+      if (assetFileDefBuilder_ == null) {
+        assetFileDef_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000004);
+      } else {
+        assetFileDefBuilder_.clear();
+      }
+      if (objectGraphDefBuilder_ == null) {
+        objectGraphDef_ = null;
+      } else {
+        objectGraphDef_ = null;
+        objectGraphDefBuilder_ = null;
+      }
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.framework.MetaGraphProtos.internal_static_tensorflow_MetaGraphDef_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.MetaGraphDef getDefaultInstanceForType() {
+      return org.tensorflow.proto.framework.MetaGraphDef.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.MetaGraphDef build() {
+      org.tensorflow.proto.framework.MetaGraphDef result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.MetaGraphDef buildPartial() {
+      org.tensorflow.proto.framework.MetaGraphDef result = new org.tensorflow.proto.framework.MetaGraphDef(this);
+      int from_bitField0_ = bitField0_;
+      if (metaInfoDefBuilder_ == null) {
+        result.metaInfoDef_ = metaInfoDef_;
+      } else {
+        result.metaInfoDef_ = metaInfoDefBuilder_.build();
+      }
+      if (graphDefBuilder_ == null) {
+        result.graphDef_ = graphDef_;
+      } else {
+        result.graphDef_ = graphDefBuilder_.build();
+      }
+      if (saverDefBuilder_ == null) {
+        result.saverDef_ = saverDef_;
+      } else {
+        result.saverDef_ = saverDefBuilder_.build();
+      }
+      result.collectionDef_ = internalGetCollectionDef();
+      result.collectionDef_.makeImmutable();
+      result.signatureDef_ = internalGetSignatureDef();
+      result.signatureDef_.makeImmutable();
+      if (assetFileDefBuilder_ == null) {
+        if (((bitField0_ & 0x00000004) != 0)) {
+          assetFileDef_ = java.util.Collections.unmodifiableList(assetFileDef_);
+          bitField0_ = (bitField0_ & ~0x00000004);
+        }
+        result.assetFileDef_ = assetFileDef_;
+      } else {
+        result.assetFileDef_ = assetFileDefBuilder_.build();
+      }
+      if (objectGraphDefBuilder_ == null) {
+        result.objectGraphDef_ = objectGraphDef_;
+      } else {
+        result.objectGraphDef_ = objectGraphDefBuilder_.build();
+      }
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.framework.MetaGraphDef) {
+        return mergeFrom((org.tensorflow.proto.framework.MetaGraphDef)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.framework.MetaGraphDef other) {
+      if (other == org.tensorflow.proto.framework.MetaGraphDef.getDefaultInstance()) return this;
+      if (other.hasMetaInfoDef()) {
+        mergeMetaInfoDef(other.getMetaInfoDef());
+      }
+      if (other.hasGraphDef()) {
+        mergeGraphDef(other.getGraphDef());
+      }
+      if (other.hasSaverDef()) {
+        mergeSaverDef(other.getSaverDef());
+      }
+      internalGetMutableCollectionDef().mergeFrom(
+          other.internalGetCollectionDef());
+      internalGetMutableSignatureDef().mergeFrom(
+          other.internalGetSignatureDef());
+      if (assetFileDefBuilder_ == null) {
+        if (!other.assetFileDef_.isEmpty()) {
+          if (assetFileDef_.isEmpty()) {
+            assetFileDef_ = other.assetFileDef_;
+            bitField0_ = (bitField0_ & ~0x00000004);
+          } else {
+            ensureAssetFileDefIsMutable();
+            assetFileDef_.addAll(other.assetFileDef_);
+          }
+          onChanged();
+        }
+      } else {
+        if (!other.assetFileDef_.isEmpty()) {
+          if (assetFileDefBuilder_.isEmpty()) {
+            assetFileDefBuilder_.dispose();
+            assetFileDefBuilder_ = null;
+            assetFileDef_ = other.assetFileDef_;
+            bitField0_ = (bitField0_ & ~0x00000004);
+            assetFileDefBuilder_ = 
+              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
+                 getAssetFileDefFieldBuilder() : null;
+          } else {
+            assetFileDefBuilder_.addAllMessages(other.assetFileDef_);
+          }
+        }
+      }
+      if (other.hasObjectGraphDef()) {
+        mergeObjectGraphDef(other.getObjectGraphDef());
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.framework.MetaGraphDef parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.framework.MetaGraphDef) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+    private int bitField0_;
+
+    private org.tensorflow.proto.framework.MetaGraphDef.MetaInfoDef metaInfoDef_;
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.MetaGraphDef.MetaInfoDef, org.tensorflow.proto.framework.MetaGraphDef.MetaInfoDef.Builder, org.tensorflow.proto.framework.MetaGraphDef.MetaInfoDefOrBuilder> metaInfoDefBuilder_;
+    /**
+     * <code>.tensorflow.MetaGraphDef.MetaInfoDef meta_info_def = 1;</code>
+     */
+    public boolean hasMetaInfoDef() {
+      return metaInfoDefBuilder_ != null || metaInfoDef_ != null;
+    }
+    /**
+     * <code>.tensorflow.MetaGraphDef.MetaInfoDef meta_info_def = 1;</code>
+     */
+    public org.tensorflow.proto.framework.MetaGraphDef.MetaInfoDef getMetaInfoDef() {
+      if (metaInfoDefBuilder_ == null) {
+        return metaInfoDef_ == null ? org.tensorflow.proto.framework.MetaGraphDef.MetaInfoDef.getDefaultInstance() : metaInfoDef_;
+      } else {
+        return metaInfoDefBuilder_.getMessage();
+      }
+    }
+    /**
+     * <code>.tensorflow.MetaGraphDef.MetaInfoDef meta_info_def = 1;</code>
+     */
+    public Builder setMetaInfoDef(org.tensorflow.proto.framework.MetaGraphDef.MetaInfoDef value) {
+      if (metaInfoDefBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        metaInfoDef_ = value;
+        onChanged();
+      } else {
+        metaInfoDefBuilder_.setMessage(value);
+      }
+
+      return this;
+    }
+    /**
+     * <code>.tensorflow.MetaGraphDef.MetaInfoDef meta_info_def = 1;</code>
+     */
+    public Builder setMetaInfoDef(
+        org.tensorflow.proto.framework.MetaGraphDef.MetaInfoDef.Builder builderForValue) {
+      if (metaInfoDefBuilder_ == null) {
+        metaInfoDef_ = builderForValue.build();
+        onChanged();
+      } else {
+        metaInfoDefBuilder_.setMessage(builderForValue.build());
+      }
+
+      return this;
+    }
+    /**
+     * <code>.tensorflow.MetaGraphDef.MetaInfoDef meta_info_def = 1;</code>
+     */
+    public Builder mergeMetaInfoDef(org.tensorflow.proto.framework.MetaGraphDef.MetaInfoDef value) {
+      if (metaInfoDefBuilder_ == null) {
+        if (metaInfoDef_ != null) {
+          metaInfoDef_ =
+            org.tensorflow.proto.framework.MetaGraphDef.MetaInfoDef.newBuilder(metaInfoDef_).mergeFrom(value).buildPartial();
+        } else {
+          metaInfoDef_ = value;
+        }
+        onChanged();
+      } else {
+        metaInfoDefBuilder_.mergeFrom(value);
+      }
+
+      return this;
+    }
+    /**
+     * <code>.tensorflow.MetaGraphDef.MetaInfoDef meta_info_def = 1;</code>
+     */
+    public Builder clearMetaInfoDef() {
+      if (metaInfoDefBuilder_ == null) {
+        metaInfoDef_ = null;
+        onChanged();
+      } else {
+        metaInfoDef_ = null;
+        metaInfoDefBuilder_ = null;
+      }
+
+      return this;
+    }
+    /**
+     * <code>.tensorflow.MetaGraphDef.MetaInfoDef meta_info_def = 1;</code>
+     */
+    public org.tensorflow.proto.framework.MetaGraphDef.MetaInfoDef.Builder getMetaInfoDefBuilder() {
+      
+      onChanged();
+      return getMetaInfoDefFieldBuilder().getBuilder();
+    }
+    /**
+     * <code>.tensorflow.MetaGraphDef.MetaInfoDef meta_info_def = 1;</code>
+     */
+    public org.tensorflow.proto.framework.MetaGraphDef.MetaInfoDefOrBuilder getMetaInfoDefOrBuilder() {
+      if (metaInfoDefBuilder_ != null) {
+        return metaInfoDefBuilder_.getMessageOrBuilder();
+      } else {
+        return metaInfoDef_ == null ?
+            org.tensorflow.proto.framework.MetaGraphDef.MetaInfoDef.getDefaultInstance() : metaInfoDef_;
+      }
+    }
+    /**
+     * <code>.tensorflow.MetaGraphDef.MetaInfoDef meta_info_def = 1;</code>
+     */
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.MetaGraphDef.MetaInfoDef, org.tensorflow.proto.framework.MetaGraphDef.MetaInfoDef.Builder, org.tensorflow.proto.framework.MetaGraphDef.MetaInfoDefOrBuilder> 
+        getMetaInfoDefFieldBuilder() {
+      if (metaInfoDefBuilder_ == null) {
+        metaInfoDefBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+            org.tensorflow.proto.framework.MetaGraphDef.MetaInfoDef, org.tensorflow.proto.framework.MetaGraphDef.MetaInfoDef.Builder, org.tensorflow.proto.framework.MetaGraphDef.MetaInfoDefOrBuilder>(
+                getMetaInfoDef(),
+                getParentForChildren(),
+                isClean());
+        metaInfoDef_ = null;
+      }
+      return metaInfoDefBuilder_;
+    }
+
+    private org.tensorflow.proto.framework.GraphDef graphDef_;
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.GraphDef, org.tensorflow.proto.framework.GraphDef.Builder, org.tensorflow.proto.framework.GraphDefOrBuilder> graphDefBuilder_;
+    /**
+     * <pre>
+     * GraphDef.
+     * </pre>
+     *
+     * <code>.tensorflow.GraphDef graph_def = 2;</code>
+     */
+    public boolean hasGraphDef() {
+      return graphDefBuilder_ != null || graphDef_ != null;
+    }
+    /**
+     * <pre>
+     * GraphDef.
+     * </pre>
+     *
+     * <code>.tensorflow.GraphDef graph_def = 2;</code>
+     */
+    public org.tensorflow.proto.framework.GraphDef getGraphDef() {
+      if (graphDefBuilder_ == null) {
+        return graphDef_ == null ? org.tensorflow.proto.framework.GraphDef.getDefaultInstance() : graphDef_;
+      } else {
+        return graphDefBuilder_.getMessage();
+      }
+    }
+    /**
+     * <pre>
+     * GraphDef.
+     * </pre>
+     *
+     * <code>.tensorflow.GraphDef graph_def = 2;</code>
+     */
+    public Builder setGraphDef(org.tensorflow.proto.framework.GraphDef value) {
+      if (graphDefBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        graphDef_ = value;
+        onChanged();
+      } else {
+        graphDefBuilder_.setMessage(value);
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * GraphDef.
+     * </pre>
+     *
+     * <code>.tensorflow.GraphDef graph_def = 2;</code>
+     */
+    public Builder setGraphDef(
+        org.tensorflow.proto.framework.GraphDef.Builder builderForValue) {
+      if (graphDefBuilder_ == null) {
+        graphDef_ = builderForValue.build();
+        onChanged();
+      } else {
+        graphDefBuilder_.setMessage(builderForValue.build());
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * GraphDef.
+     * </pre>
+     *
+     * <code>.tensorflow.GraphDef graph_def = 2;</code>
+     */
+    public Builder mergeGraphDef(org.tensorflow.proto.framework.GraphDef value) {
+      if (graphDefBuilder_ == null) {
+        if (graphDef_ != null) {
+          graphDef_ =
+            org.tensorflow.proto.framework.GraphDef.newBuilder(graphDef_).mergeFrom(value).buildPartial();
+        } else {
+          graphDef_ = value;
+        }
+        onChanged();
+      } else {
+        graphDefBuilder_.mergeFrom(value);
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * GraphDef.
+     * </pre>
+     *
+     * <code>.tensorflow.GraphDef graph_def = 2;</code>
+     */
+    public Builder clearGraphDef() {
+      if (graphDefBuilder_ == null) {
+        graphDef_ = null;
+        onChanged();
+      } else {
+        graphDef_ = null;
+        graphDefBuilder_ = null;
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * GraphDef.
+     * </pre>
+     *
+     * <code>.tensorflow.GraphDef graph_def = 2;</code>
+     */
+    public org.tensorflow.proto.framework.GraphDef.Builder getGraphDefBuilder() {
+      
+      onChanged();
+      return getGraphDefFieldBuilder().getBuilder();
+    }
+    /**
+     * <pre>
+     * GraphDef.
+     * </pre>
+     *
+     * <code>.tensorflow.GraphDef graph_def = 2;</code>
+     */
+    public org.tensorflow.proto.framework.GraphDefOrBuilder getGraphDefOrBuilder() {
+      if (graphDefBuilder_ != null) {
+        return graphDefBuilder_.getMessageOrBuilder();
+      } else {
+        return graphDef_ == null ?
+            org.tensorflow.proto.framework.GraphDef.getDefaultInstance() : graphDef_;
+      }
+    }
+    /**
+     * <pre>
+     * GraphDef.
+     * </pre>
+     *
+     * <code>.tensorflow.GraphDef graph_def = 2;</code>
+     */
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.GraphDef, org.tensorflow.proto.framework.GraphDef.Builder, org.tensorflow.proto.framework.GraphDefOrBuilder> 
+        getGraphDefFieldBuilder() {
+      if (graphDefBuilder_ == null) {
+        graphDefBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+            org.tensorflow.proto.framework.GraphDef, org.tensorflow.proto.framework.GraphDef.Builder, org.tensorflow.proto.framework.GraphDefOrBuilder>(
+                getGraphDef(),
+                getParentForChildren(),
+                isClean());
+        graphDef_ = null;
+      }
+      return graphDefBuilder_;
+    }
+
+    private org.tensorflow.proto.util.SaverDef saverDef_;
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.util.SaverDef, org.tensorflow.proto.util.SaverDef.Builder, org.tensorflow.proto.util.SaverDefOrBuilder> saverDefBuilder_;
+    /**
+     * <pre>
+     * SaverDef.
+     * </pre>
+     *
+     * <code>.tensorflow.SaverDef saver_def = 3;</code>
+     */
+    public boolean hasSaverDef() {
+      return saverDefBuilder_ != null || saverDef_ != null;
+    }
+    /**
+     * <pre>
+     * SaverDef.
+     * </pre>
+     *
+     * <code>.tensorflow.SaverDef saver_def = 3;</code>
+     */
+    public org.tensorflow.proto.util.SaverDef getSaverDef() {
+      if (saverDefBuilder_ == null) {
+        return saverDef_ == null ? org.tensorflow.proto.util.SaverDef.getDefaultInstance() : saverDef_;
+      } else {
+        return saverDefBuilder_.getMessage();
+      }
+    }
+    /**
+     * <pre>
+     * SaverDef.
+     * </pre>
+     *
+     * <code>.tensorflow.SaverDef saver_def = 3;</code>
+     */
+    public Builder setSaverDef(org.tensorflow.proto.util.SaverDef value) {
+      if (saverDefBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        saverDef_ = value;
+        onChanged();
+      } else {
+        saverDefBuilder_.setMessage(value);
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * SaverDef.
+     * </pre>
+     *
+     * <code>.tensorflow.SaverDef saver_def = 3;</code>
+     */
+    public Builder setSaverDef(
+        org.tensorflow.proto.util.SaverDef.Builder builderForValue) {
+      if (saverDefBuilder_ == null) {
+        saverDef_ = builderForValue.build();
+        onChanged();
+      } else {
+        saverDefBuilder_.setMessage(builderForValue.build());
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * SaverDef.
+     * </pre>
+     *
+     * <code>.tensorflow.SaverDef saver_def = 3;</code>
+     */
+    public Builder mergeSaverDef(org.tensorflow.proto.util.SaverDef value) {
+      if (saverDefBuilder_ == null) {
+        if (saverDef_ != null) {
+          saverDef_ =
+            org.tensorflow.proto.util.SaverDef.newBuilder(saverDef_).mergeFrom(value).buildPartial();
+        } else {
+          saverDef_ = value;
+        }
+        onChanged();
+      } else {
+        saverDefBuilder_.mergeFrom(value);
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * SaverDef.
+     * </pre>
+     *
+     * <code>.tensorflow.SaverDef saver_def = 3;</code>
+     */
+    public Builder clearSaverDef() {
+      if (saverDefBuilder_ == null) {
+        saverDef_ = null;
+        onChanged();
+      } else {
+        saverDef_ = null;
+        saverDefBuilder_ = null;
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * SaverDef.
+     * </pre>
+     *
+     * <code>.tensorflow.SaverDef saver_def = 3;</code>
+     */
+    public org.tensorflow.proto.util.SaverDef.Builder getSaverDefBuilder() {
+      
+      onChanged();
+      return getSaverDefFieldBuilder().getBuilder();
+    }
+    /**
+     * <pre>
+     * SaverDef.
+     * </pre>
+     *
+     * <code>.tensorflow.SaverDef saver_def = 3;</code>
+     */
+    public org.tensorflow.proto.util.SaverDefOrBuilder getSaverDefOrBuilder() {
+      if (saverDefBuilder_ != null) {
+        return saverDefBuilder_.getMessageOrBuilder();
+      } else {
+        return saverDef_ == null ?
+            org.tensorflow.proto.util.SaverDef.getDefaultInstance() : saverDef_;
+      }
+    }
+    /**
+     * <pre>
+     * SaverDef.
+     * </pre>
+     *
+     * <code>.tensorflow.SaverDef saver_def = 3;</code>
+     */
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.util.SaverDef, org.tensorflow.proto.util.SaverDef.Builder, org.tensorflow.proto.util.SaverDefOrBuilder> 
+        getSaverDefFieldBuilder() {
+      if (saverDefBuilder_ == null) {
+        saverDefBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+            org.tensorflow.proto.util.SaverDef, org.tensorflow.proto.util.SaverDef.Builder, org.tensorflow.proto.util.SaverDefOrBuilder>(
+                getSaverDef(),
+                getParentForChildren(),
+                isClean());
+        saverDef_ = null;
+      }
+      return saverDefBuilder_;
+    }
+
+    private com.google.protobuf.MapField<
+        java.lang.String, org.tensorflow.proto.framework.CollectionDef> collectionDef_;
+    private com.google.protobuf.MapField<java.lang.String, org.tensorflow.proto.framework.CollectionDef>
+    internalGetCollectionDef() {
+      if (collectionDef_ == null) {
+        return com.google.protobuf.MapField.emptyMapField(
+            CollectionDefDefaultEntryHolder.defaultEntry);
+      }
+      return collectionDef_;
+    }
+    private com.google.protobuf.MapField<java.lang.String, org.tensorflow.proto.framework.CollectionDef>
+    internalGetMutableCollectionDef() {
+      onChanged();;
+      if (collectionDef_ == null) {
+        collectionDef_ = com.google.protobuf.MapField.newMapField(
+            CollectionDefDefaultEntryHolder.defaultEntry);
+      }
+      if (!collectionDef_.isMutable()) {
+        collectionDef_ = collectionDef_.copy();
+      }
+      return collectionDef_;
+    }
+
+    public int getCollectionDefCount() {
+      return internalGetCollectionDef().getMap().size();
+    }
+    /**
+     * <pre>
+     * collection_def: Map from collection name to collections.
+     * See CollectionDef section for details.
+     * </pre>
+     *
+     * <code>map&lt;string, .tensorflow.CollectionDef&gt; collection_def = 4;</code>
+     */
+
+    public boolean containsCollectionDef(
+        java.lang.String key) {
+      if (key == null) { throw new java.lang.NullPointerException(); }
+      return internalGetCollectionDef().getMap().containsKey(key);
+    }
+    /**
+     * Use {@link #getCollectionDefMap()} instead.
+     */
+    @java.lang.Deprecated
+    public java.util.Map<java.lang.String, org.tensorflow.proto.framework.CollectionDef> getCollectionDef() {
+      return getCollectionDefMap();
+    }
+    /**
+     * <pre>
+     * collection_def: Map from collection name to collections.
+     * See CollectionDef section for details.
+     * </pre>
+     *
+     * <code>map&lt;string, .tensorflow.CollectionDef&gt; collection_def = 4;</code>
+     */
+
+    public java.util.Map<java.lang.String, org.tensorflow.proto.framework.CollectionDef> getCollectionDefMap() {
+      return internalGetCollectionDef().getMap();
+    }
+    /**
+     * <pre>
+     * collection_def: Map from collection name to collections.
+     * See CollectionDef section for details.
+     * </pre>
+     *
+     * <code>map&lt;string, .tensorflow.CollectionDef&gt; collection_def = 4;</code>
+     */
+
+    public org.tensorflow.proto.framework.CollectionDef getCollectionDefOrDefault(
+        java.lang.String key,
+        org.tensorflow.proto.framework.CollectionDef defaultValue) {
+      if (key == null) { throw new java.lang.NullPointerException(); }
+      java.util.Map<java.lang.String, org.tensorflow.proto.framework.CollectionDef> map =
+          internalGetCollectionDef().getMap();
+      return map.containsKey(key) ? map.get(key) : defaultValue;
+    }
+    /**
+     * <pre>
+     * collection_def: Map from collection name to collections.
+     * See CollectionDef section for details.
+     * </pre>
+     *
+     * <code>map&lt;string, .tensorflow.CollectionDef&gt; collection_def = 4;</code>
+     */
+
+    public org.tensorflow.proto.framework.CollectionDef getCollectionDefOrThrow(
+        java.lang.String key) {
+      if (key == null) { throw new java.lang.NullPointerException(); }
+      java.util.Map<java.lang.String, org.tensorflow.proto.framework.CollectionDef> map =
+          internalGetCollectionDef().getMap();
+      if (!map.containsKey(key)) {
+        throw new java.lang.IllegalArgumentException();
+      }
+      return map.get(key);
+    }
+
+    public Builder clearCollectionDef() {
+      internalGetMutableCollectionDef().getMutableMap()
+          .clear();
+      return this;
+    }
+    /**
+     * <pre>
+     * collection_def: Map from collection name to collections.
+     * See CollectionDef section for details.
+     * </pre>
+     *
+     * <code>map&lt;string, .tensorflow.CollectionDef&gt; collection_def = 4;</code>
+     */
+
+    public Builder removeCollectionDef(
+        java.lang.String key) {
+      if (key == null) { throw new java.lang.NullPointerException(); }
+      internalGetMutableCollectionDef().getMutableMap()
+          .remove(key);
+      return this;
+    }
+    /**
+     * Use alternate mutation accessors instead.
+     */
+    @java.lang.Deprecated
+    public java.util.Map<java.lang.String, org.tensorflow.proto.framework.CollectionDef>
+    getMutableCollectionDef() {
+      return internalGetMutableCollectionDef().getMutableMap();
+    }
+    /**
+     * <pre>
+     * collection_def: Map from collection name to collections.
+     * See CollectionDef section for details.
+     * </pre>
+     *
+     * <code>map&lt;string, .tensorflow.CollectionDef&gt; collection_def = 4;</code>
+     */
+    public Builder putCollectionDef(
+        java.lang.String key,
+        org.tensorflow.proto.framework.CollectionDef value) {
+      if (key == null) { throw new java.lang.NullPointerException(); }
+      if (value == null) { throw new java.lang.NullPointerException(); }
+      internalGetMutableCollectionDef().getMutableMap()
+          .put(key, value);
+      return this;
+    }
+    /**
+     * <pre>
+     * collection_def: Map from collection name to collections.
+     * See CollectionDef section for details.
+     * </pre>
+     *
+     * <code>map&lt;string, .tensorflow.CollectionDef&gt; collection_def = 4;</code>
+     */
+
+    public Builder putAllCollectionDef(
+        java.util.Map<java.lang.String, org.tensorflow.proto.framework.CollectionDef> values) {
+      internalGetMutableCollectionDef().getMutableMap()
+          .putAll(values);
+      return this;
+    }
+
+    private com.google.protobuf.MapField<
+        java.lang.String, org.tensorflow.proto.framework.SignatureDef> signatureDef_;
+    private com.google.protobuf.MapField<java.lang.String, org.tensorflow.proto.framework.SignatureDef>
+    internalGetSignatureDef() {
+      if (signatureDef_ == null) {
+        return com.google.protobuf.MapField.emptyMapField(
+            SignatureDefDefaultEntryHolder.defaultEntry);
+      }
+      return signatureDef_;
+    }
+    private com.google.protobuf.MapField<java.lang.String, org.tensorflow.proto.framework.SignatureDef>
+    internalGetMutableSignatureDef() {
+      onChanged();;
+      if (signatureDef_ == null) {
+        signatureDef_ = com.google.protobuf.MapField.newMapField(
+            SignatureDefDefaultEntryHolder.defaultEntry);
+      }
+      if (!signatureDef_.isMutable()) {
+        signatureDef_ = signatureDef_.copy();
+      }
+      return signatureDef_;
+    }
+
+    public int getSignatureDefCount() {
+      return internalGetSignatureDef().getMap().size();
+    }
+    /**
+     * <pre>
+     * signature_def: Map from user supplied key for a signature to a single
+     * SignatureDef.
+     * </pre>
+     *
+     * <code>map&lt;string, .tensorflow.SignatureDef&gt; signature_def = 5;</code>
+     */
+
+    public boolean containsSignatureDef(
+        java.lang.String key) {
+      if (key == null) { throw new java.lang.NullPointerException(); }
+      return internalGetSignatureDef().getMap().containsKey(key);
+    }
+    /**
+     * Use {@link #getSignatureDefMap()} instead.
+     */
+    @java.lang.Deprecated
+    public java.util.Map<java.lang.String, org.tensorflow.proto.framework.SignatureDef> getSignatureDef() {
+      return getSignatureDefMap();
+    }
+    /**
+     * <pre>
+     * signature_def: Map from user supplied key for a signature to a single
+     * SignatureDef.
+     * </pre>
+     *
+     * <code>map&lt;string, .tensorflow.SignatureDef&gt; signature_def = 5;</code>
+     */
+
+    public java.util.Map<java.lang.String, org.tensorflow.proto.framework.SignatureDef> getSignatureDefMap() {
+      return internalGetSignatureDef().getMap();
+    }
+    /**
+     * <pre>
+     * signature_def: Map from user supplied key for a signature to a single
+     * SignatureDef.
+     * </pre>
+     *
+     * <code>map&lt;string, .tensorflow.SignatureDef&gt; signature_def = 5;</code>
+     */
+
+    public org.tensorflow.proto.framework.SignatureDef getSignatureDefOrDefault(
+        java.lang.String key,
+        org.tensorflow.proto.framework.SignatureDef defaultValue) {
+      if (key == null) { throw new java.lang.NullPointerException(); }
+      java.util.Map<java.lang.String, org.tensorflow.proto.framework.SignatureDef> map =
+          internalGetSignatureDef().getMap();
+      return map.containsKey(key) ? map.get(key) : defaultValue;
+    }
+    /**
+     * <pre>
+     * signature_def: Map from user supplied key for a signature to a single
+     * SignatureDef.
+     * </pre>
+     *
+     * <code>map&lt;string, .tensorflow.SignatureDef&gt; signature_def = 5;</code>
+     */
+
+    public org.tensorflow.proto.framework.SignatureDef getSignatureDefOrThrow(
+        java.lang.String key) {
+      if (key == null) { throw new java.lang.NullPointerException(); }
+      java.util.Map<java.lang.String, org.tensorflow.proto.framework.SignatureDef> map =
+          internalGetSignatureDef().getMap();
+      if (!map.containsKey(key)) {
+        throw new java.lang.IllegalArgumentException();
+      }
+      return map.get(key);
+    }
+
+    public Builder clearSignatureDef() {
+      internalGetMutableSignatureDef().getMutableMap()
+          .clear();
+      return this;
+    }
+    /**
+     * <pre>
+     * signature_def: Map from user supplied key for a signature to a single
+     * SignatureDef.
+     * </pre>
+     *
+     * <code>map&lt;string, .tensorflow.SignatureDef&gt; signature_def = 5;</code>
+     */
+
+    public Builder removeSignatureDef(
+        java.lang.String key) {
+      if (key == null) { throw new java.lang.NullPointerException(); }
+      internalGetMutableSignatureDef().getMutableMap()
+          .remove(key);
+      return this;
+    }
+    /**
+     * Use alternate mutation accessors instead.
+     */
+    @java.lang.Deprecated
+    public java.util.Map<java.lang.String, org.tensorflow.proto.framework.SignatureDef>
+    getMutableSignatureDef() {
+      return internalGetMutableSignatureDef().getMutableMap();
+    }
+    /**
+     * <pre>
+     * signature_def: Map from user supplied key for a signature to a single
+     * SignatureDef.
+     * </pre>
+     *
+     * <code>map&lt;string, .tensorflow.SignatureDef&gt; signature_def = 5;</code>
+     */
+    public Builder putSignatureDef(
+        java.lang.String key,
+        org.tensorflow.proto.framework.SignatureDef value) {
+      if (key == null) { throw new java.lang.NullPointerException(); }
+      if (value == null) { throw new java.lang.NullPointerException(); }
+      internalGetMutableSignatureDef().getMutableMap()
+          .put(key, value);
+      return this;
+    }
+    /**
+     * <pre>
+     * signature_def: Map from user supplied key for a signature to a single
+     * SignatureDef.
+     * </pre>
+     *
+     * <code>map&lt;string, .tensorflow.SignatureDef&gt; signature_def = 5;</code>
+     */
+
+    public Builder putAllSignatureDef(
+        java.util.Map<java.lang.String, org.tensorflow.proto.framework.SignatureDef> values) {
+      internalGetMutableSignatureDef().getMutableMap()
+          .putAll(values);
+      return this;
+    }
+
+    private java.util.List<org.tensorflow.proto.framework.AssetFileDef> assetFileDef_ =
+      java.util.Collections.emptyList();
+    private void ensureAssetFileDefIsMutable() {
+      if (!((bitField0_ & 0x00000004) != 0)) {
+        assetFileDef_ = new java.util.ArrayList<org.tensorflow.proto.framework.AssetFileDef>(assetFileDef_);
+        bitField0_ |= 0x00000004;
+       }
+    }
+
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.framework.AssetFileDef, org.tensorflow.proto.framework.AssetFileDef.Builder, org.tensorflow.proto.framework.AssetFileDefOrBuilder> assetFileDefBuilder_;
+
+    /**
+     * <pre>
+     * Asset file def to be used with the defined graph.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.AssetFileDef asset_file_def = 6;</code>
+     */
+    public java.util.List<org.tensorflow.proto.framework.AssetFileDef> getAssetFileDefList() {
+      if (assetFileDefBuilder_ == null) {
+        return java.util.Collections.unmodifiableList(assetFileDef_);
+      } else {
+        return assetFileDefBuilder_.getMessageList();
+      }
+    }
+    /**
+     * <pre>
+     * Asset file def to be used with the defined graph.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.AssetFileDef asset_file_def = 6;</code>
+     */
+    public int getAssetFileDefCount() {
+      if (assetFileDefBuilder_ == null) {
+        return assetFileDef_.size();
+      } else {
+        return assetFileDefBuilder_.getCount();
+      }
+    }
+    /**
+     * <pre>
+     * Asset file def to be used with the defined graph.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.AssetFileDef asset_file_def = 6;</code>
+     */
+    public org.tensorflow.proto.framework.AssetFileDef getAssetFileDef(int index) {
+      if (assetFileDefBuilder_ == null) {
+        return assetFileDef_.get(index);
+      } else {
+        return assetFileDefBuilder_.getMessage(index);
+      }
+    }
+    /**
+     * <pre>
+     * Asset file def to be used with the defined graph.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.AssetFileDef asset_file_def = 6;</code>
+     */
+    public Builder setAssetFileDef(
+        int index, org.tensorflow.proto.framework.AssetFileDef value) {
+      if (assetFileDefBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureAssetFileDefIsMutable();
+        assetFileDef_.set(index, value);
+        onChanged();
+      } else {
+        assetFileDefBuilder_.setMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Asset file def to be used with the defined graph.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.AssetFileDef asset_file_def = 6;</code>
+     */
+    public Builder setAssetFileDef(
+        int index, org.tensorflow.proto.framework.AssetFileDef.Builder builderForValue) {
+      if (assetFileDefBuilder_ == null) {
+        ensureAssetFileDefIsMutable();
+        assetFileDef_.set(index, builderForValue.build());
+        onChanged();
+      } else {
+        assetFileDefBuilder_.setMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Asset file def to be used with the defined graph.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.AssetFileDef asset_file_def = 6;</code>
+     */
+    public Builder addAssetFileDef(org.tensorflow.proto.framework.AssetFileDef value) {
+      if (assetFileDefBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureAssetFileDefIsMutable();
+        assetFileDef_.add(value);
+        onChanged();
+      } else {
+        assetFileDefBuilder_.addMessage(value);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Asset file def to be used with the defined graph.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.AssetFileDef asset_file_def = 6;</code>
+     */
+    public Builder addAssetFileDef(
+        int index, org.tensorflow.proto.framework.AssetFileDef value) {
+      if (assetFileDefBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureAssetFileDefIsMutable();
+        assetFileDef_.add(index, value);
+        onChanged();
+      } else {
+        assetFileDefBuilder_.addMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Asset file def to be used with the defined graph.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.AssetFileDef asset_file_def = 6;</code>
+     */
+    public Builder addAssetFileDef(
+        org.tensorflow.proto.framework.AssetFileDef.Builder builderForValue) {
+      if (assetFileDefBuilder_ == null) {
+        ensureAssetFileDefIsMutable();
+        assetFileDef_.add(builderForValue.build());
+        onChanged();
+      } else {
+        assetFileDefBuilder_.addMessage(builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Asset file def to be used with the defined graph.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.AssetFileDef asset_file_def = 6;</code>
+     */
+    public Builder addAssetFileDef(
+        int index, org.tensorflow.proto.framework.AssetFileDef.Builder builderForValue) {
+      if (assetFileDefBuilder_ == null) {
+        ensureAssetFileDefIsMutable();
+        assetFileDef_.add(index, builderForValue.build());
+        onChanged();
+      } else {
+        assetFileDefBuilder_.addMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Asset file def to be used with the defined graph.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.AssetFileDef asset_file_def = 6;</code>
+     */
+    public Builder addAllAssetFileDef(
+        java.lang.Iterable<? extends org.tensorflow.proto.framework.AssetFileDef> values) {
+      if (assetFileDefBuilder_ == null) {
+        ensureAssetFileDefIsMutable();
+        com.google.protobuf.AbstractMessageLite.Builder.addAll(
+            values, assetFileDef_);
+        onChanged();
+      } else {
+        assetFileDefBuilder_.addAllMessages(values);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Asset file def to be used with the defined graph.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.AssetFileDef asset_file_def = 6;</code>
+     */
+    public Builder clearAssetFileDef() {
+      if (assetFileDefBuilder_ == null) {
+        assetFileDef_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000004);
+        onChanged();
+      } else {
+        assetFileDefBuilder_.clear();
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Asset file def to be used with the defined graph.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.AssetFileDef asset_file_def = 6;</code>
+     */
+    public Builder removeAssetFileDef(int index) {
+      if (assetFileDefBuilder_ == null) {
+        ensureAssetFileDefIsMutable();
+        assetFileDef_.remove(index);
+        onChanged();
+      } else {
+        assetFileDefBuilder_.remove(index);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Asset file def to be used with the defined graph.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.AssetFileDef asset_file_def = 6;</code>
+     */
+    public org.tensorflow.proto.framework.AssetFileDef.Builder getAssetFileDefBuilder(
+        int index) {
+      return getAssetFileDefFieldBuilder().getBuilder(index);
+    }
+    /**
+     * <pre>
+     * Asset file def to be used with the defined graph.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.AssetFileDef asset_file_def = 6;</code>
+     */
+    public org.tensorflow.proto.framework.AssetFileDefOrBuilder getAssetFileDefOrBuilder(
+        int index) {
+      if (assetFileDefBuilder_ == null) {
+        return assetFileDef_.get(index);  } else {
+        return assetFileDefBuilder_.getMessageOrBuilder(index);
+      }
+    }
+    /**
+     * <pre>
+     * Asset file def to be used with the defined graph.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.AssetFileDef asset_file_def = 6;</code>
+     */
+    public java.util.List<? extends org.tensorflow.proto.framework.AssetFileDefOrBuilder> 
+         getAssetFileDefOrBuilderList() {
+      if (assetFileDefBuilder_ != null) {
+        return assetFileDefBuilder_.getMessageOrBuilderList();
+      } else {
+        return java.util.Collections.unmodifiableList(assetFileDef_);
+      }
+    }
+    /**
+     * <pre>
+     * Asset file def to be used with the defined graph.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.AssetFileDef asset_file_def = 6;</code>
+     */
+    public org.tensorflow.proto.framework.AssetFileDef.Builder addAssetFileDefBuilder() {
+      return getAssetFileDefFieldBuilder().addBuilder(
+          org.tensorflow.proto.framework.AssetFileDef.getDefaultInstance());
+    }
+    /**
+     * <pre>
+     * Asset file def to be used with the defined graph.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.AssetFileDef asset_file_def = 6;</code>
+     */
+    public org.tensorflow.proto.framework.AssetFileDef.Builder addAssetFileDefBuilder(
+        int index) {
+      return getAssetFileDefFieldBuilder().addBuilder(
+          index, org.tensorflow.proto.framework.AssetFileDef.getDefaultInstance());
+    }
+    /**
+     * <pre>
+     * Asset file def to be used with the defined graph.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.AssetFileDef asset_file_def = 6;</code>
+     */
+    public java.util.List<org.tensorflow.proto.framework.AssetFileDef.Builder> 
+         getAssetFileDefBuilderList() {
+      return getAssetFileDefFieldBuilder().getBuilderList();
+    }
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.framework.AssetFileDef, org.tensorflow.proto.framework.AssetFileDef.Builder, org.tensorflow.proto.framework.AssetFileDefOrBuilder> 
+        getAssetFileDefFieldBuilder() {
+      if (assetFileDefBuilder_ == null) {
+        assetFileDefBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
+            org.tensorflow.proto.framework.AssetFileDef, org.tensorflow.proto.framework.AssetFileDef.Builder, org.tensorflow.proto.framework.AssetFileDefOrBuilder>(
+                assetFileDef_,
+                ((bitField0_ & 0x00000004) != 0),
+                getParentForChildren(),
+                isClean());
+        assetFileDef_ = null;
+      }
+      return assetFileDefBuilder_;
+    }
+
+    private org.tensorflow.proto.framework.SavedObjectGraph objectGraphDef_;
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.SavedObjectGraph, org.tensorflow.proto.framework.SavedObjectGraph.Builder, org.tensorflow.proto.framework.SavedObjectGraphOrBuilder> objectGraphDefBuilder_;
+    /**
+     * <pre>
+     * Extra information about the structure of functions and stateful objects.
+     * </pre>
+     *
+     * <code>.tensorflow.SavedObjectGraph object_graph_def = 7;</code>
+     */
+    public boolean hasObjectGraphDef() {
+      return objectGraphDefBuilder_ != null || objectGraphDef_ != null;
+    }
+    /**
+     * <pre>
+     * Extra information about the structure of functions and stateful objects.
+     * </pre>
+     *
+     * <code>.tensorflow.SavedObjectGraph object_graph_def = 7;</code>
+     */
+    public org.tensorflow.proto.framework.SavedObjectGraph getObjectGraphDef() {
+      if (objectGraphDefBuilder_ == null) {
+        return objectGraphDef_ == null ? org.tensorflow.proto.framework.SavedObjectGraph.getDefaultInstance() : objectGraphDef_;
+      } else {
+        return objectGraphDefBuilder_.getMessage();
+      }
+    }
+    /**
+     * <pre>
+     * Extra information about the structure of functions and stateful objects.
+     * </pre>
+     *
+     * <code>.tensorflow.SavedObjectGraph object_graph_def = 7;</code>
+     */
+    public Builder setObjectGraphDef(org.tensorflow.proto.framework.SavedObjectGraph value) {
+      if (objectGraphDefBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        objectGraphDef_ = value;
+        onChanged();
+      } else {
+        objectGraphDefBuilder_.setMessage(value);
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Extra information about the structure of functions and stateful objects.
+     * </pre>
+     *
+     * <code>.tensorflow.SavedObjectGraph object_graph_def = 7;</code>
+     */
+    public Builder setObjectGraphDef(
+        org.tensorflow.proto.framework.SavedObjectGraph.Builder builderForValue) {
+      if (objectGraphDefBuilder_ == null) {
+        objectGraphDef_ = builderForValue.build();
+        onChanged();
+      } else {
+        objectGraphDefBuilder_.setMessage(builderForValue.build());
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Extra information about the structure of functions and stateful objects.
+     * </pre>
+     *
+     * <code>.tensorflow.SavedObjectGraph object_graph_def = 7;</code>
+     */
+    public Builder mergeObjectGraphDef(org.tensorflow.proto.framework.SavedObjectGraph value) {
+      if (objectGraphDefBuilder_ == null) {
+        if (objectGraphDef_ != null) {
+          objectGraphDef_ =
+            org.tensorflow.proto.framework.SavedObjectGraph.newBuilder(objectGraphDef_).mergeFrom(value).buildPartial();
+        } else {
+          objectGraphDef_ = value;
+        }
+        onChanged();
+      } else {
+        objectGraphDefBuilder_.mergeFrom(value);
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Extra information about the structure of functions and stateful objects.
+     * </pre>
+     *
+     * <code>.tensorflow.SavedObjectGraph object_graph_def = 7;</code>
+     */
+    public Builder clearObjectGraphDef() {
+      if (objectGraphDefBuilder_ == null) {
+        objectGraphDef_ = null;
+        onChanged();
+      } else {
+        objectGraphDef_ = null;
+        objectGraphDefBuilder_ = null;
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Extra information about the structure of functions and stateful objects.
+     * </pre>
+     *
+     * <code>.tensorflow.SavedObjectGraph object_graph_def = 7;</code>
+     */
+    public org.tensorflow.proto.framework.SavedObjectGraph.Builder getObjectGraphDefBuilder() {
+      
+      onChanged();
+      return getObjectGraphDefFieldBuilder().getBuilder();
+    }
+    /**
+     * <pre>
+     * Extra information about the structure of functions and stateful objects.
+     * </pre>
+     *
+     * <code>.tensorflow.SavedObjectGraph object_graph_def = 7;</code>
+     */
+    public org.tensorflow.proto.framework.SavedObjectGraphOrBuilder getObjectGraphDefOrBuilder() {
+      if (objectGraphDefBuilder_ != null) {
+        return objectGraphDefBuilder_.getMessageOrBuilder();
+      } else {
+        return objectGraphDef_ == null ?
+            org.tensorflow.proto.framework.SavedObjectGraph.getDefaultInstance() : objectGraphDef_;
+      }
+    }
+    /**
+     * <pre>
+     * Extra information about the structure of functions and stateful objects.
+     * </pre>
+     *
+     * <code>.tensorflow.SavedObjectGraph object_graph_def = 7;</code>
+     */
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.SavedObjectGraph, org.tensorflow.proto.framework.SavedObjectGraph.Builder, org.tensorflow.proto.framework.SavedObjectGraphOrBuilder> 
+        getObjectGraphDefFieldBuilder() {
+      if (objectGraphDefBuilder_ == null) {
+        objectGraphDefBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+            org.tensorflow.proto.framework.SavedObjectGraph, org.tensorflow.proto.framework.SavedObjectGraph.Builder, org.tensorflow.proto.framework.SavedObjectGraphOrBuilder>(
+                getObjectGraphDef(),
+                getParentForChildren(),
+                isClean());
+        objectGraphDef_ = null;
+      }
+      return objectGraphDefBuilder_;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.MetaGraphDef)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.MetaGraphDef)
+  private static final org.tensorflow.proto.framework.MetaGraphDef DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.MetaGraphDef();
+  }
+
+  public static org.tensorflow.proto.framework.MetaGraphDef getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<MetaGraphDef>
+      PARSER = new com.google.protobuf.AbstractParser<MetaGraphDef>() {
+    @java.lang.Override
+    public MetaGraphDef parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new MetaGraphDef(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<MetaGraphDef> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<MetaGraphDef> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.framework.MetaGraphDef getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/MetaGraphDefOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/MetaGraphDefOrBuilder.java
new file mode 100644
index 00000000000..ca9cf06f97b
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/MetaGraphDefOrBuilder.java
@@ -0,0 +1,259 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/meta_graph.proto
+
+package org.tensorflow.proto.framework;
+
+public interface MetaGraphDefOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.MetaGraphDef)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <code>.tensorflow.MetaGraphDef.MetaInfoDef meta_info_def = 1;</code>
+   */
+  boolean hasMetaInfoDef();
+  /**
+   * <code>.tensorflow.MetaGraphDef.MetaInfoDef meta_info_def = 1;</code>
+   */
+  org.tensorflow.proto.framework.MetaGraphDef.MetaInfoDef getMetaInfoDef();
+  /**
+   * <code>.tensorflow.MetaGraphDef.MetaInfoDef meta_info_def = 1;</code>
+   */
+  org.tensorflow.proto.framework.MetaGraphDef.MetaInfoDefOrBuilder getMetaInfoDefOrBuilder();
+
+  /**
+   * <pre>
+   * GraphDef.
+   * </pre>
+   *
+   * <code>.tensorflow.GraphDef graph_def = 2;</code>
+   */
+  boolean hasGraphDef();
+  /**
+   * <pre>
+   * GraphDef.
+   * </pre>
+   *
+   * <code>.tensorflow.GraphDef graph_def = 2;</code>
+   */
+  org.tensorflow.proto.framework.GraphDef getGraphDef();
+  /**
+   * <pre>
+   * GraphDef.
+   * </pre>
+   *
+   * <code>.tensorflow.GraphDef graph_def = 2;</code>
+   */
+  org.tensorflow.proto.framework.GraphDefOrBuilder getGraphDefOrBuilder();
+
+  /**
+   * <pre>
+   * SaverDef.
+   * </pre>
+   *
+   * <code>.tensorflow.SaverDef saver_def = 3;</code>
+   */
+  boolean hasSaverDef();
+  /**
+   * <pre>
+   * SaverDef.
+   * </pre>
+   *
+   * <code>.tensorflow.SaverDef saver_def = 3;</code>
+   */
+  org.tensorflow.proto.util.SaverDef getSaverDef();
+  /**
+   * <pre>
+   * SaverDef.
+   * </pre>
+   *
+   * <code>.tensorflow.SaverDef saver_def = 3;</code>
+   */
+  org.tensorflow.proto.util.SaverDefOrBuilder getSaverDefOrBuilder();
+
+  /**
+   * <pre>
+   * collection_def: Map from collection name to collections.
+   * See CollectionDef section for details.
+   * </pre>
+   *
+   * <code>map&lt;string, .tensorflow.CollectionDef&gt; collection_def = 4;</code>
+   */
+  int getCollectionDefCount();
+  /**
+   * <pre>
+   * collection_def: Map from collection name to collections.
+   * See CollectionDef section for details.
+   * </pre>
+   *
+   * <code>map&lt;string, .tensorflow.CollectionDef&gt; collection_def = 4;</code>
+   */
+  boolean containsCollectionDef(
+      java.lang.String key);
+  /**
+   * Use {@link #getCollectionDefMap()} instead.
+   */
+  @java.lang.Deprecated
+  java.util.Map<java.lang.String, org.tensorflow.proto.framework.CollectionDef>
+  getCollectionDef();
+  /**
+   * <pre>
+   * collection_def: Map from collection name to collections.
+   * See CollectionDef section for details.
+   * </pre>
+   *
+   * <code>map&lt;string, .tensorflow.CollectionDef&gt; collection_def = 4;</code>
+   */
+  java.util.Map<java.lang.String, org.tensorflow.proto.framework.CollectionDef>
+  getCollectionDefMap();
+  /**
+   * <pre>
+   * collection_def: Map from collection name to collections.
+   * See CollectionDef section for details.
+   * </pre>
+   *
+   * <code>map&lt;string, .tensorflow.CollectionDef&gt; collection_def = 4;</code>
+   */
+
+  org.tensorflow.proto.framework.CollectionDef getCollectionDefOrDefault(
+      java.lang.String key,
+      org.tensorflow.proto.framework.CollectionDef defaultValue);
+  /**
+   * <pre>
+   * collection_def: Map from collection name to collections.
+   * See CollectionDef section for details.
+   * </pre>
+   *
+   * <code>map&lt;string, .tensorflow.CollectionDef&gt; collection_def = 4;</code>
+   */
+
+  org.tensorflow.proto.framework.CollectionDef getCollectionDefOrThrow(
+      java.lang.String key);
+
+  /**
+   * <pre>
+   * signature_def: Map from user supplied key for a signature to a single
+   * SignatureDef.
+   * </pre>
+   *
+   * <code>map&lt;string, .tensorflow.SignatureDef&gt; signature_def = 5;</code>
+   */
+  int getSignatureDefCount();
+  /**
+   * <pre>
+   * signature_def: Map from user supplied key for a signature to a single
+   * SignatureDef.
+   * </pre>
+   *
+   * <code>map&lt;string, .tensorflow.SignatureDef&gt; signature_def = 5;</code>
+   */
+  boolean containsSignatureDef(
+      java.lang.String key);
+  /**
+   * Use {@link #getSignatureDefMap()} instead.
+   */
+  @java.lang.Deprecated
+  java.util.Map<java.lang.String, org.tensorflow.proto.framework.SignatureDef>
+  getSignatureDef();
+  /**
+   * <pre>
+   * signature_def: Map from user supplied key for a signature to a single
+   * SignatureDef.
+   * </pre>
+   *
+   * <code>map&lt;string, .tensorflow.SignatureDef&gt; signature_def = 5;</code>
+   */
+  java.util.Map<java.lang.String, org.tensorflow.proto.framework.SignatureDef>
+  getSignatureDefMap();
+  /**
+   * <pre>
+   * signature_def: Map from user supplied key for a signature to a single
+   * SignatureDef.
+   * </pre>
+   *
+   * <code>map&lt;string, .tensorflow.SignatureDef&gt; signature_def = 5;</code>
+   */
+
+  org.tensorflow.proto.framework.SignatureDef getSignatureDefOrDefault(
+      java.lang.String key,
+      org.tensorflow.proto.framework.SignatureDef defaultValue);
+  /**
+   * <pre>
+   * signature_def: Map from user supplied key for a signature to a single
+   * SignatureDef.
+   * </pre>
+   *
+   * <code>map&lt;string, .tensorflow.SignatureDef&gt; signature_def = 5;</code>
+   */
+
+  org.tensorflow.proto.framework.SignatureDef getSignatureDefOrThrow(
+      java.lang.String key);
+
+  /**
+   * <pre>
+   * Asset file def to be used with the defined graph.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.AssetFileDef asset_file_def = 6;</code>
+   */
+  java.util.List<org.tensorflow.proto.framework.AssetFileDef> 
+      getAssetFileDefList();
+  /**
+   * <pre>
+   * Asset file def to be used with the defined graph.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.AssetFileDef asset_file_def = 6;</code>
+   */
+  org.tensorflow.proto.framework.AssetFileDef getAssetFileDef(int index);
+  /**
+   * <pre>
+   * Asset file def to be used with the defined graph.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.AssetFileDef asset_file_def = 6;</code>
+   */
+  int getAssetFileDefCount();
+  /**
+   * <pre>
+   * Asset file def to be used with the defined graph.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.AssetFileDef asset_file_def = 6;</code>
+   */
+  java.util.List<? extends org.tensorflow.proto.framework.AssetFileDefOrBuilder> 
+      getAssetFileDefOrBuilderList();
+  /**
+   * <pre>
+   * Asset file def to be used with the defined graph.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.AssetFileDef asset_file_def = 6;</code>
+   */
+  org.tensorflow.proto.framework.AssetFileDefOrBuilder getAssetFileDefOrBuilder(
+      int index);
+
+  /**
+   * <pre>
+   * Extra information about the structure of functions and stateful objects.
+   * </pre>
+   *
+   * <code>.tensorflow.SavedObjectGraph object_graph_def = 7;</code>
+   */
+  boolean hasObjectGraphDef();
+  /**
+   * <pre>
+   * Extra information about the structure of functions and stateful objects.
+   * </pre>
+   *
+   * <code>.tensorflow.SavedObjectGraph object_graph_def = 7;</code>
+   */
+  org.tensorflow.proto.framework.SavedObjectGraph getObjectGraphDef();
+  /**
+   * <pre>
+   * Extra information about the structure of functions and stateful objects.
+   * </pre>
+   *
+   * <code>.tensorflow.SavedObjectGraph object_graph_def = 7;</code>
+   */
+  org.tensorflow.proto.framework.SavedObjectGraphOrBuilder getObjectGraphDefOrBuilder();
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/MetaGraphProtos.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/MetaGraphProtos.java
new file mode 100644
index 00000000000..cd79fd2f986
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/MetaGraphProtos.java
@@ -0,0 +1,318 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/meta_graph.proto
+
+package org.tensorflow.proto.framework;
+
+public final class MetaGraphProtos {
+  private MetaGraphProtos() {}
+  public static void registerAllExtensions(
+      com.google.protobuf.ExtensionRegistryLite registry) {
+  }
+
+  public static void registerAllExtensions(
+      com.google.protobuf.ExtensionRegistry registry) {
+    registerAllExtensions(
+        (com.google.protobuf.ExtensionRegistryLite) registry);
+  }
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_MetaGraphDef_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_MetaGraphDef_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_MetaGraphDef_MetaInfoDef_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_MetaGraphDef_MetaInfoDef_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_MetaGraphDef_MetaInfoDef_FunctionAliasesEntry_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_MetaGraphDef_MetaInfoDef_FunctionAliasesEntry_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_MetaGraphDef_CollectionDefEntry_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_MetaGraphDef_CollectionDefEntry_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_MetaGraphDef_SignatureDefEntry_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_MetaGraphDef_SignatureDefEntry_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_CollectionDef_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_CollectionDef_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_CollectionDef_NodeList_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_CollectionDef_NodeList_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_CollectionDef_BytesList_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_CollectionDef_BytesList_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_CollectionDef_Int64List_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_CollectionDef_Int64List_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_CollectionDef_FloatList_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_CollectionDef_FloatList_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_CollectionDef_AnyList_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_CollectionDef_AnyList_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_TensorInfo_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_TensorInfo_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_TensorInfo_CooSparse_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_TensorInfo_CooSparse_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_TensorInfo_CompositeTensor_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_TensorInfo_CompositeTensor_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_SignatureDef_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_SignatureDef_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_SignatureDef_InputsEntry_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_SignatureDef_InputsEntry_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_SignatureDef_OutputsEntry_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_SignatureDef_OutputsEntry_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_AssetFileDef_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_AssetFileDef_fieldAccessorTable;
+
+  public static com.google.protobuf.Descriptors.FileDescriptor
+      getDescriptor() {
+    return descriptor;
+  }
+  private static  com.google.protobuf.Descriptors.FileDescriptor
+      descriptor;
+  static {
+    java.lang.String[] descriptorData = {
+      "\n)tensorflow/core/protobuf/meta_graph.pr" +
+      "oto\022\ntensorflow\032\031google/protobuf/any.pro" +
+      "to\032%tensorflow/core/framework/graph.prot" +
+      "o\032&tensorflow/core/framework/op_def.prot" +
+      "o\032,tensorflow/core/framework/tensor_shap" +
+      "e.proto\032%tensorflow/core/framework/types" +
+      ".proto\0321tensorflow/core/protobuf/saved_o" +
+      "bject_graph.proto\032$tensorflow/core/proto" +
+      "buf/saver.proto\032%tensorflow/core/protobu" +
+      "f/struct.proto\"\250\007\n\014MetaGraphDef\022;\n\rmeta_" +
+      "info_def\030\001 \001(\0132$.tensorflow.MetaGraphDef" +
+      ".MetaInfoDef\022\'\n\tgraph_def\030\002 \001(\0132\024.tensor" +
+      "flow.GraphDef\022\'\n\tsaver_def\030\003 \001(\0132\024.tenso" +
+      "rflow.SaverDef\022C\n\016collection_def\030\004 \003(\0132+" +
+      ".tensorflow.MetaGraphDef.CollectionDefEn" +
+      "try\022A\n\rsignature_def\030\005 \003(\0132*.tensorflow." +
+      "MetaGraphDef.SignatureDefEntry\0220\n\016asset_" +
+      "file_def\030\006 \003(\0132\030.tensorflow.AssetFileDef" +
+      "\0226\n\020object_graph_def\030\007 \001(\0132\034.tensorflow." +
+      "SavedObjectGraph\032\366\002\n\013MetaInfoDef\022\032\n\022meta" +
+      "_graph_version\030\001 \001(\t\022,\n\020stripped_op_list" +
+      "\030\002 \001(\0132\022.tensorflow.OpList\022&\n\010any_info\030\003" +
+      " \001(\0132\024.google.protobuf.Any\022\014\n\004tags\030\004 \003(\t" +
+      "\022\032\n\022tensorflow_version\030\005 \001(\t\022\036\n\026tensorfl" +
+      "ow_git_version\030\006 \001(\t\022\036\n\026stripped_default" +
+      "_attrs\030\007 \001(\010\022S\n\020function_aliases\030\010 \003(\01329" +
+      ".tensorflow.MetaGraphDef.MetaInfoDef.Fun" +
+      "ctionAliasesEntry\0326\n\024FunctionAliasesEntr" +
+      "y\022\013\n\003key\030\001 \001(\t\022\r\n\005value\030\002 \001(\t:\0028\001\032O\n\022Col" +
+      "lectionDefEntry\022\013\n\003key\030\001 \001(\t\022(\n\005value\030\002 " +
+      "\001(\0132\031.tensorflow.CollectionDef:\0028\001\032M\n\021Si" +
+      "gnatureDefEntry\022\013\n\003key\030\001 \001(\t\022\'\n\005value\030\002 " +
+      "\001(\0132\030.tensorflow.SignatureDef:\0028\001\"\337\003\n\rCo" +
+      "llectionDef\0227\n\tnode_list\030\001 \001(\0132\".tensorf" +
+      "low.CollectionDef.NodeListH\000\0229\n\nbytes_li" +
+      "st\030\002 \001(\0132#.tensorflow.CollectionDef.Byte" +
+      "sListH\000\0229\n\nint64_list\030\003 \001(\0132#.tensorflow" +
+      ".CollectionDef.Int64ListH\000\0229\n\nfloat_list" +
+      "\030\004 \001(\0132#.tensorflow.CollectionDef.FloatL" +
+      "istH\000\0225\n\010any_list\030\005 \001(\0132!.tensorflow.Col" +
+      "lectionDef.AnyListH\000\032\031\n\010NodeList\022\r\n\005valu" +
+      "e\030\001 \003(\t\032\032\n\tBytesList\022\r\n\005value\030\001 \003(\014\032\036\n\tI" +
+      "nt64List\022\021\n\005value\030\001 \003(\003B\002\020\001\032\036\n\tFloatList" +
+      "\022\021\n\005value\030\001 \003(\002B\002\020\001\032.\n\007AnyList\022#\n\005value\030" +
+      "\001 \003(\0132\024.google.protobuf.AnyB\006\n\004kind\"\321\003\n\n" +
+      "TensorInfo\022\016\n\004name\030\001 \001(\tH\000\0226\n\ncoo_sparse" +
+      "\030\004 \001(\0132 .tensorflow.TensorInfo.CooSparse" +
+      "H\000\022B\n\020composite_tensor\030\005 \001(\0132&.tensorflo" +
+      "w.TensorInfo.CompositeTensorH\000\022#\n\005dtype\030" +
+      "\002 \001(\0162\024.tensorflow.DataType\0222\n\014tensor_sh" +
+      "ape\030\003 \001(\0132\034.tensorflow.TensorShapeProto\032" +
+      "e\n\tCooSparse\022\032\n\022values_tensor_name\030\001 \001(\t" +
+      "\022\033\n\023indices_tensor_name\030\002 \001(\t\022\037\n\027dense_s" +
+      "hape_tensor_name\030\003 \001(\t\032k\n\017CompositeTenso" +
+      "r\022,\n\ttype_spec\030\001 \001(\0132\031.tensorflow.TypeSp" +
+      "ecProto\022*\n\ncomponents\030\002 \003(\0132\026.tensorflow" +
+      ".TensorInfoB\n\n\010encoding\"\240\002\n\014SignatureDef" +
+      "\0224\n\006inputs\030\001 \003(\0132$.tensorflow.SignatureD" +
+      "ef.InputsEntry\0226\n\007outputs\030\002 \003(\0132%.tensor" +
+      "flow.SignatureDef.OutputsEntry\022\023\n\013method" +
+      "_name\030\003 \001(\t\032E\n\013InputsEntry\022\013\n\003key\030\001 \001(\t\022" +
+      "%\n\005value\030\002 \001(\0132\026.tensorflow.TensorInfo:\002" +
+      "8\001\032F\n\014OutputsEntry\022\013\n\003key\030\001 \001(\t\022%\n\005value" +
+      "\030\002 \001(\0132\026.tensorflow.TensorInfo:\0028\001\"M\n\014As" +
+      "setFileDef\022+\n\013tensor_info\030\001 \001(\0132\026.tensor" +
+      "flow.TensorInfo\022\020\n\010filename\030\002 \001(\tB\200\001\n\036or" +
+      "g.tensorflow.proto.frameworkB\017MetaGraphP" +
+      "rotosP\001ZHgithub.com/tensorflow/tensorflo" +
+      "w/tensorflow/go/core/core_protos_go_prot" +
+      "o\370\001\001b\006proto3"
+    };
+    descriptor = com.google.protobuf.Descriptors.FileDescriptor
+      .internalBuildGeneratedFileFrom(descriptorData,
+        new com.google.protobuf.Descriptors.FileDescriptor[] {
+          com.google.protobuf.AnyProto.getDescriptor(),
+          org.tensorflow.proto.framework.GraphProtos.getDescriptor(),
+          org.tensorflow.proto.framework.OpDefProtos.getDescriptor(),
+          org.tensorflow.proto.framework.TensorShapeProtos.getDescriptor(),
+          org.tensorflow.proto.framework.TypesProtos.getDescriptor(),
+          org.tensorflow.proto.framework.SavedObjectGraphProtos.getDescriptor(),
+          org.tensorflow.proto.util.SaverProtos.getDescriptor(),
+          org.tensorflow.proto.framework.StructProtos.getDescriptor(),
+        });
+    internal_static_tensorflow_MetaGraphDef_descriptor =
+      getDescriptor().getMessageTypes().get(0);
+    internal_static_tensorflow_MetaGraphDef_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_MetaGraphDef_descriptor,
+        new java.lang.String[] { "MetaInfoDef", "GraphDef", "SaverDef", "CollectionDef", "SignatureDef", "AssetFileDef", "ObjectGraphDef", });
+    internal_static_tensorflow_MetaGraphDef_MetaInfoDef_descriptor =
+      internal_static_tensorflow_MetaGraphDef_descriptor.getNestedTypes().get(0);
+    internal_static_tensorflow_MetaGraphDef_MetaInfoDef_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_MetaGraphDef_MetaInfoDef_descriptor,
+        new java.lang.String[] { "MetaGraphVersion", "StrippedOpList", "AnyInfo", "Tags", "TensorflowVersion", "TensorflowGitVersion", "StrippedDefaultAttrs", "FunctionAliases", });
+    internal_static_tensorflow_MetaGraphDef_MetaInfoDef_FunctionAliasesEntry_descriptor =
+      internal_static_tensorflow_MetaGraphDef_MetaInfoDef_descriptor.getNestedTypes().get(0);
+    internal_static_tensorflow_MetaGraphDef_MetaInfoDef_FunctionAliasesEntry_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_MetaGraphDef_MetaInfoDef_FunctionAliasesEntry_descriptor,
+        new java.lang.String[] { "Key", "Value", });
+    internal_static_tensorflow_MetaGraphDef_CollectionDefEntry_descriptor =
+      internal_static_tensorflow_MetaGraphDef_descriptor.getNestedTypes().get(1);
+    internal_static_tensorflow_MetaGraphDef_CollectionDefEntry_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_MetaGraphDef_CollectionDefEntry_descriptor,
+        new java.lang.String[] { "Key", "Value", });
+    internal_static_tensorflow_MetaGraphDef_SignatureDefEntry_descriptor =
+      internal_static_tensorflow_MetaGraphDef_descriptor.getNestedTypes().get(2);
+    internal_static_tensorflow_MetaGraphDef_SignatureDefEntry_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_MetaGraphDef_SignatureDefEntry_descriptor,
+        new java.lang.String[] { "Key", "Value", });
+    internal_static_tensorflow_CollectionDef_descriptor =
+      getDescriptor().getMessageTypes().get(1);
+    internal_static_tensorflow_CollectionDef_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_CollectionDef_descriptor,
+        new java.lang.String[] { "NodeList", "BytesList", "Int64List", "FloatList", "AnyList", "Kind", });
+    internal_static_tensorflow_CollectionDef_NodeList_descriptor =
+      internal_static_tensorflow_CollectionDef_descriptor.getNestedTypes().get(0);
+    internal_static_tensorflow_CollectionDef_NodeList_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_CollectionDef_NodeList_descriptor,
+        new java.lang.String[] { "Value", });
+    internal_static_tensorflow_CollectionDef_BytesList_descriptor =
+      internal_static_tensorflow_CollectionDef_descriptor.getNestedTypes().get(1);
+    internal_static_tensorflow_CollectionDef_BytesList_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_CollectionDef_BytesList_descriptor,
+        new java.lang.String[] { "Value", });
+    internal_static_tensorflow_CollectionDef_Int64List_descriptor =
+      internal_static_tensorflow_CollectionDef_descriptor.getNestedTypes().get(2);
+    internal_static_tensorflow_CollectionDef_Int64List_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_CollectionDef_Int64List_descriptor,
+        new java.lang.String[] { "Value", });
+    internal_static_tensorflow_CollectionDef_FloatList_descriptor =
+      internal_static_tensorflow_CollectionDef_descriptor.getNestedTypes().get(3);
+    internal_static_tensorflow_CollectionDef_FloatList_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_CollectionDef_FloatList_descriptor,
+        new java.lang.String[] { "Value", });
+    internal_static_tensorflow_CollectionDef_AnyList_descriptor =
+      internal_static_tensorflow_CollectionDef_descriptor.getNestedTypes().get(4);
+    internal_static_tensorflow_CollectionDef_AnyList_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_CollectionDef_AnyList_descriptor,
+        new java.lang.String[] { "Value", });
+    internal_static_tensorflow_TensorInfo_descriptor =
+      getDescriptor().getMessageTypes().get(2);
+    internal_static_tensorflow_TensorInfo_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_TensorInfo_descriptor,
+        new java.lang.String[] { "Name", "CooSparse", "CompositeTensor", "Dtype", "TensorShape", "Encoding", });
+    internal_static_tensorflow_TensorInfo_CooSparse_descriptor =
+      internal_static_tensorflow_TensorInfo_descriptor.getNestedTypes().get(0);
+    internal_static_tensorflow_TensorInfo_CooSparse_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_TensorInfo_CooSparse_descriptor,
+        new java.lang.String[] { "ValuesTensorName", "IndicesTensorName", "DenseShapeTensorName", });
+    internal_static_tensorflow_TensorInfo_CompositeTensor_descriptor =
+      internal_static_tensorflow_TensorInfo_descriptor.getNestedTypes().get(1);
+    internal_static_tensorflow_TensorInfo_CompositeTensor_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_TensorInfo_CompositeTensor_descriptor,
+        new java.lang.String[] { "TypeSpec", "Components", });
+    internal_static_tensorflow_SignatureDef_descriptor =
+      getDescriptor().getMessageTypes().get(3);
+    internal_static_tensorflow_SignatureDef_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_SignatureDef_descriptor,
+        new java.lang.String[] { "Inputs", "Outputs", "MethodName", });
+    internal_static_tensorflow_SignatureDef_InputsEntry_descriptor =
+      internal_static_tensorflow_SignatureDef_descriptor.getNestedTypes().get(0);
+    internal_static_tensorflow_SignatureDef_InputsEntry_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_SignatureDef_InputsEntry_descriptor,
+        new java.lang.String[] { "Key", "Value", });
+    internal_static_tensorflow_SignatureDef_OutputsEntry_descriptor =
+      internal_static_tensorflow_SignatureDef_descriptor.getNestedTypes().get(1);
+    internal_static_tensorflow_SignatureDef_OutputsEntry_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_SignatureDef_OutputsEntry_descriptor,
+        new java.lang.String[] { "Key", "Value", });
+    internal_static_tensorflow_AssetFileDef_descriptor =
+      getDescriptor().getMessageTypes().get(4);
+    internal_static_tensorflow_AssetFileDef_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_AssetFileDef_descriptor,
+        new java.lang.String[] { "TensorInfo", "Filename", });
+    com.google.protobuf.AnyProto.getDescriptor();
+    org.tensorflow.proto.framework.GraphProtos.getDescriptor();
+    org.tensorflow.proto.framework.OpDefProtos.getDescriptor();
+    org.tensorflow.proto.framework.TensorShapeProtos.getDescriptor();
+    org.tensorflow.proto.framework.TypesProtos.getDescriptor();
+    org.tensorflow.proto.framework.SavedObjectGraphProtos.getDescriptor();
+    org.tensorflow.proto.util.SaverProtos.getDescriptor();
+    org.tensorflow.proto.framework.StructProtos.getDescriptor();
+  }
+
+  // @@protoc_insertion_point(outer_class_scope)
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NameAttrList.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NameAttrList.java
new file mode 100644
index 00000000000..a17b31f8aa1
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NameAttrList.java
@@ -0,0 +1,832 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/framework/attr_value.proto
+
+package org.tensorflow.proto.framework;
+
+/**
+ * <pre>
+ * A list of attr names and their values. The whole list is attached
+ * with a string name.  E.g., MatMul[T=float].
+ * </pre>
+ *
+ * Protobuf type {@code tensorflow.NameAttrList}
+ */
+public  final class NameAttrList extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.NameAttrList)
+    NameAttrListOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use NameAttrList.newBuilder() to construct.
+  private NameAttrList(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private NameAttrList() {
+    name_ = "";
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new NameAttrList();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private NameAttrList(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    int mutable_bitField0_ = 0;
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 10: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            name_ = s;
+            break;
+          }
+          case 18: {
+            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
+              attr_ = com.google.protobuf.MapField.newMapField(
+                  AttrDefaultEntryHolder.defaultEntry);
+              mutable_bitField0_ |= 0x00000001;
+            }
+            com.google.protobuf.MapEntry<java.lang.String, org.tensorflow.proto.framework.AttrValue>
+            attr__ = input.readMessage(
+                AttrDefaultEntryHolder.defaultEntry.getParserForType(), extensionRegistry);
+            attr_.getMutableMap().put(
+                attr__.getKey(), attr__.getValue());
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.framework.AttrValueProtos.internal_static_tensorflow_NameAttrList_descriptor;
+  }
+
+  @SuppressWarnings({"rawtypes"})
+  @java.lang.Override
+  protected com.google.protobuf.MapField internalGetMapField(
+      int number) {
+    switch (number) {
+      case 2:
+        return internalGetAttr();
+      default:
+        throw new RuntimeException(
+            "Invalid map field number: " + number);
+    }
+  }
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.framework.AttrValueProtos.internal_static_tensorflow_NameAttrList_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.framework.NameAttrList.class, org.tensorflow.proto.framework.NameAttrList.Builder.class);
+  }
+
+  public static final int NAME_FIELD_NUMBER = 1;
+  private volatile java.lang.Object name_;
+  /**
+   * <code>string name = 1;</code>
+   */
+  public java.lang.String getName() {
+    java.lang.Object ref = name_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      name_ = s;
+      return s;
+    }
+  }
+  /**
+   * <code>string name = 1;</code>
+   */
+  public com.google.protobuf.ByteString
+      getNameBytes() {
+    java.lang.Object ref = name_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      name_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int ATTR_FIELD_NUMBER = 2;
+  private static final class AttrDefaultEntryHolder {
+    static final com.google.protobuf.MapEntry<
+        java.lang.String, org.tensorflow.proto.framework.AttrValue> defaultEntry =
+            com.google.protobuf.MapEntry
+            .<java.lang.String, org.tensorflow.proto.framework.AttrValue>newDefaultInstance(
+                org.tensorflow.proto.framework.AttrValueProtos.internal_static_tensorflow_NameAttrList_AttrEntry_descriptor, 
+                com.google.protobuf.WireFormat.FieldType.STRING,
+                "",
+                com.google.protobuf.WireFormat.FieldType.MESSAGE,
+                org.tensorflow.proto.framework.AttrValue.getDefaultInstance());
+  }
+  private com.google.protobuf.MapField<
+      java.lang.String, org.tensorflow.proto.framework.AttrValue> attr_;
+  private com.google.protobuf.MapField<java.lang.String, org.tensorflow.proto.framework.AttrValue>
+  internalGetAttr() {
+    if (attr_ == null) {
+      return com.google.protobuf.MapField.emptyMapField(
+          AttrDefaultEntryHolder.defaultEntry);
+    }
+    return attr_;
+  }
+
+  public int getAttrCount() {
+    return internalGetAttr().getMap().size();
+  }
+  /**
+   * <code>map&lt;string, .tensorflow.AttrValue&gt; attr = 2;</code>
+   */
+
+  public boolean containsAttr(
+      java.lang.String key) {
+    if (key == null) { throw new java.lang.NullPointerException(); }
+    return internalGetAttr().getMap().containsKey(key);
+  }
+  /**
+   * Use {@link #getAttrMap()} instead.
+   */
+  @java.lang.Deprecated
+  public java.util.Map<java.lang.String, org.tensorflow.proto.framework.AttrValue> getAttr() {
+    return getAttrMap();
+  }
+  /**
+   * <code>map&lt;string, .tensorflow.AttrValue&gt; attr = 2;</code>
+   */
+
+  public java.util.Map<java.lang.String, org.tensorflow.proto.framework.AttrValue> getAttrMap() {
+    return internalGetAttr().getMap();
+  }
+  /**
+   * <code>map&lt;string, .tensorflow.AttrValue&gt; attr = 2;</code>
+   */
+
+  public org.tensorflow.proto.framework.AttrValue getAttrOrDefault(
+      java.lang.String key,
+      org.tensorflow.proto.framework.AttrValue defaultValue) {
+    if (key == null) { throw new java.lang.NullPointerException(); }
+    java.util.Map<java.lang.String, org.tensorflow.proto.framework.AttrValue> map =
+        internalGetAttr().getMap();
+    return map.containsKey(key) ? map.get(key) : defaultValue;
+  }
+  /**
+   * <code>map&lt;string, .tensorflow.AttrValue&gt; attr = 2;</code>
+   */
+
+  public org.tensorflow.proto.framework.AttrValue getAttrOrThrow(
+      java.lang.String key) {
+    if (key == null) { throw new java.lang.NullPointerException(); }
+    java.util.Map<java.lang.String, org.tensorflow.proto.framework.AttrValue> map =
+        internalGetAttr().getMap();
+    if (!map.containsKey(key)) {
+      throw new java.lang.IllegalArgumentException();
+    }
+    return map.get(key);
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    if (!getNameBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 1, name_);
+    }
+    com.google.protobuf.GeneratedMessageV3
+      .serializeStringMapTo(
+        output,
+        internalGetAttr(),
+        AttrDefaultEntryHolder.defaultEntry,
+        2);
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (!getNameBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, name_);
+    }
+    for (java.util.Map.Entry<java.lang.String, org.tensorflow.proto.framework.AttrValue> entry
+         : internalGetAttr().getMap().entrySet()) {
+      com.google.protobuf.MapEntry<java.lang.String, org.tensorflow.proto.framework.AttrValue>
+      attr__ = AttrDefaultEntryHolder.defaultEntry.newBuilderForType()
+          .setKey(entry.getKey())
+          .setValue(entry.getValue())
+          .build();
+      size += com.google.protobuf.CodedOutputStream
+          .computeMessageSize(2, attr__);
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.framework.NameAttrList)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.framework.NameAttrList other = (org.tensorflow.proto.framework.NameAttrList) obj;
+
+    if (!getName()
+        .equals(other.getName())) return false;
+    if (!internalGetAttr().equals(
+        other.internalGetAttr())) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    hash = (37 * hash) + NAME_FIELD_NUMBER;
+    hash = (53 * hash) + getName().hashCode();
+    if (!internalGetAttr().getMap().isEmpty()) {
+      hash = (37 * hash) + ATTR_FIELD_NUMBER;
+      hash = (53 * hash) + internalGetAttr().hashCode();
+    }
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.framework.NameAttrList parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.NameAttrList parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.NameAttrList parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.NameAttrList parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.NameAttrList parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.NameAttrList parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.NameAttrList parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.NameAttrList parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.NameAttrList parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.NameAttrList parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.NameAttrList parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.NameAttrList parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.framework.NameAttrList prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * <pre>
+   * A list of attr names and their values. The whole list is attached
+   * with a string name.  E.g., MatMul[T=float].
+   * </pre>
+   *
+   * Protobuf type {@code tensorflow.NameAttrList}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.NameAttrList)
+      org.tensorflow.proto.framework.NameAttrListOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.AttrValueProtos.internal_static_tensorflow_NameAttrList_descriptor;
+    }
+
+    @SuppressWarnings({"rawtypes"})
+    protected com.google.protobuf.MapField internalGetMapField(
+        int number) {
+      switch (number) {
+        case 2:
+          return internalGetAttr();
+        default:
+          throw new RuntimeException(
+              "Invalid map field number: " + number);
+      }
+    }
+    @SuppressWarnings({"rawtypes"})
+    protected com.google.protobuf.MapField internalGetMutableMapField(
+        int number) {
+      switch (number) {
+        case 2:
+          return internalGetMutableAttr();
+        default:
+          throw new RuntimeException(
+              "Invalid map field number: " + number);
+      }
+    }
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.AttrValueProtos.internal_static_tensorflow_NameAttrList_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.NameAttrList.class, org.tensorflow.proto.framework.NameAttrList.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.framework.NameAttrList.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      name_ = "";
+
+      internalGetMutableAttr().clear();
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.framework.AttrValueProtos.internal_static_tensorflow_NameAttrList_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.NameAttrList getDefaultInstanceForType() {
+      return org.tensorflow.proto.framework.NameAttrList.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.NameAttrList build() {
+      org.tensorflow.proto.framework.NameAttrList result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.NameAttrList buildPartial() {
+      org.tensorflow.proto.framework.NameAttrList result = new org.tensorflow.proto.framework.NameAttrList(this);
+      int from_bitField0_ = bitField0_;
+      result.name_ = name_;
+      result.attr_ = internalGetAttr();
+      result.attr_.makeImmutable();
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.framework.NameAttrList) {
+        return mergeFrom((org.tensorflow.proto.framework.NameAttrList)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.framework.NameAttrList other) {
+      if (other == org.tensorflow.proto.framework.NameAttrList.getDefaultInstance()) return this;
+      if (!other.getName().isEmpty()) {
+        name_ = other.name_;
+        onChanged();
+      }
+      internalGetMutableAttr().mergeFrom(
+          other.internalGetAttr());
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.framework.NameAttrList parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.framework.NameAttrList) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+    private int bitField0_;
+
+    private java.lang.Object name_ = "";
+    /**
+     * <code>string name = 1;</code>
+     */
+    public java.lang.String getName() {
+      java.lang.Object ref = name_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        name_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <code>string name = 1;</code>
+     */
+    public com.google.protobuf.ByteString
+        getNameBytes() {
+      java.lang.Object ref = name_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        name_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <code>string name = 1;</code>
+     */
+    public Builder setName(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      name_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>string name = 1;</code>
+     */
+    public Builder clearName() {
+      
+      name_ = getDefaultInstance().getName();
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>string name = 1;</code>
+     */
+    public Builder setNameBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      name_ = value;
+      onChanged();
+      return this;
+    }
+
+    private com.google.protobuf.MapField<
+        java.lang.String, org.tensorflow.proto.framework.AttrValue> attr_;
+    private com.google.protobuf.MapField<java.lang.String, org.tensorflow.proto.framework.AttrValue>
+    internalGetAttr() {
+      if (attr_ == null) {
+        return com.google.protobuf.MapField.emptyMapField(
+            AttrDefaultEntryHolder.defaultEntry);
+      }
+      return attr_;
+    }
+    private com.google.protobuf.MapField<java.lang.String, org.tensorflow.proto.framework.AttrValue>
+    internalGetMutableAttr() {
+      onChanged();;
+      if (attr_ == null) {
+        attr_ = com.google.protobuf.MapField.newMapField(
+            AttrDefaultEntryHolder.defaultEntry);
+      }
+      if (!attr_.isMutable()) {
+        attr_ = attr_.copy();
+      }
+      return attr_;
+    }
+
+    public int getAttrCount() {
+      return internalGetAttr().getMap().size();
+    }
+    /**
+     * <code>map&lt;string, .tensorflow.AttrValue&gt; attr = 2;</code>
+     */
+
+    public boolean containsAttr(
+        java.lang.String key) {
+      if (key == null) { throw new java.lang.NullPointerException(); }
+      return internalGetAttr().getMap().containsKey(key);
+    }
+    /**
+     * Use {@link #getAttrMap()} instead.
+     */
+    @java.lang.Deprecated
+    public java.util.Map<java.lang.String, org.tensorflow.proto.framework.AttrValue> getAttr() {
+      return getAttrMap();
+    }
+    /**
+     * <code>map&lt;string, .tensorflow.AttrValue&gt; attr = 2;</code>
+     */
+
+    public java.util.Map<java.lang.String, org.tensorflow.proto.framework.AttrValue> getAttrMap() {
+      return internalGetAttr().getMap();
+    }
+    /**
+     * <code>map&lt;string, .tensorflow.AttrValue&gt; attr = 2;</code>
+     */
+
+    public org.tensorflow.proto.framework.AttrValue getAttrOrDefault(
+        java.lang.String key,
+        org.tensorflow.proto.framework.AttrValue defaultValue) {
+      if (key == null) { throw new java.lang.NullPointerException(); }
+      java.util.Map<java.lang.String, org.tensorflow.proto.framework.AttrValue> map =
+          internalGetAttr().getMap();
+      return map.containsKey(key) ? map.get(key) : defaultValue;
+    }
+    /**
+     * <code>map&lt;string, .tensorflow.AttrValue&gt; attr = 2;</code>
+     */
+
+    public org.tensorflow.proto.framework.AttrValue getAttrOrThrow(
+        java.lang.String key) {
+      if (key == null) { throw new java.lang.NullPointerException(); }
+      java.util.Map<java.lang.String, org.tensorflow.proto.framework.AttrValue> map =
+          internalGetAttr().getMap();
+      if (!map.containsKey(key)) {
+        throw new java.lang.IllegalArgumentException();
+      }
+      return map.get(key);
+    }
+
+    public Builder clearAttr() {
+      internalGetMutableAttr().getMutableMap()
+          .clear();
+      return this;
+    }
+    /**
+     * <code>map&lt;string, .tensorflow.AttrValue&gt; attr = 2;</code>
+     */
+
+    public Builder removeAttr(
+        java.lang.String key) {
+      if (key == null) { throw new java.lang.NullPointerException(); }
+      internalGetMutableAttr().getMutableMap()
+          .remove(key);
+      return this;
+    }
+    /**
+     * Use alternate mutation accessors instead.
+     */
+    @java.lang.Deprecated
+    public java.util.Map<java.lang.String, org.tensorflow.proto.framework.AttrValue>
+    getMutableAttr() {
+      return internalGetMutableAttr().getMutableMap();
+    }
+    /**
+     * <code>map&lt;string, .tensorflow.AttrValue&gt; attr = 2;</code>
+     */
+    public Builder putAttr(
+        java.lang.String key,
+        org.tensorflow.proto.framework.AttrValue value) {
+      if (key == null) { throw new java.lang.NullPointerException(); }
+      if (value == null) { throw new java.lang.NullPointerException(); }
+      internalGetMutableAttr().getMutableMap()
+          .put(key, value);
+      return this;
+    }
+    /**
+     * <code>map&lt;string, .tensorflow.AttrValue&gt; attr = 2;</code>
+     */
+
+    public Builder putAllAttr(
+        java.util.Map<java.lang.String, org.tensorflow.proto.framework.AttrValue> values) {
+      internalGetMutableAttr().getMutableMap()
+          .putAll(values);
+      return this;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.NameAttrList)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.NameAttrList)
+  private static final org.tensorflow.proto.framework.NameAttrList DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.NameAttrList();
+  }
+
+  public static org.tensorflow.proto.framework.NameAttrList getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<NameAttrList>
+      PARSER = new com.google.protobuf.AbstractParser<NameAttrList>() {
+    @java.lang.Override
+    public NameAttrList parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new NameAttrList(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<NameAttrList> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<NameAttrList> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.framework.NameAttrList getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NameAttrListOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NameAttrListOrBuilder.java
new file mode 100644
index 00000000000..293b4278408
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NameAttrListOrBuilder.java
@@ -0,0 +1,53 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/framework/attr_value.proto
+
+package org.tensorflow.proto.framework;
+
+public interface NameAttrListOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.NameAttrList)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <code>string name = 1;</code>
+   */
+  java.lang.String getName();
+  /**
+   * <code>string name = 1;</code>
+   */
+  com.google.protobuf.ByteString
+      getNameBytes();
+
+  /**
+   * <code>map&lt;string, .tensorflow.AttrValue&gt; attr = 2;</code>
+   */
+  int getAttrCount();
+  /**
+   * <code>map&lt;string, .tensorflow.AttrValue&gt; attr = 2;</code>
+   */
+  boolean containsAttr(
+      java.lang.String key);
+  /**
+   * Use {@link #getAttrMap()} instead.
+   */
+  @java.lang.Deprecated
+  java.util.Map<java.lang.String, org.tensorflow.proto.framework.AttrValue>
+  getAttr();
+  /**
+   * <code>map&lt;string, .tensorflow.AttrValue&gt; attr = 2;</code>
+   */
+  java.util.Map<java.lang.String, org.tensorflow.proto.framework.AttrValue>
+  getAttrMap();
+  /**
+   * <code>map&lt;string, .tensorflow.AttrValue&gt; attr = 2;</code>
+   */
+
+  org.tensorflow.proto.framework.AttrValue getAttrOrDefault(
+      java.lang.String key,
+      org.tensorflow.proto.framework.AttrValue defaultValue);
+  /**
+   * <code>map&lt;string, .tensorflow.AttrValue&gt; attr = 2;</code>
+   */
+
+  org.tensorflow.proto.framework.AttrValue getAttrOrThrow(
+      java.lang.String key);
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NamedDevice.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NamedDevice.java
new file mode 100644
index 00000000000..02a1dc9f6c7
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NamedDevice.java
@@ -0,0 +1,727 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/device_properties.proto
+
+package org.tensorflow.proto.framework;
+
+/**
+ * Protobuf type {@code tensorflow.NamedDevice}
+ */
+public  final class NamedDevice extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.NamedDevice)
+    NamedDeviceOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use NamedDevice.newBuilder() to construct.
+  private NamedDevice(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private NamedDevice() {
+    name_ = "";
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new NamedDevice();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private NamedDevice(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 10: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            name_ = s;
+            break;
+          }
+          case 18: {
+            org.tensorflow.proto.framework.DeviceProperties.Builder subBuilder = null;
+            if (properties_ != null) {
+              subBuilder = properties_.toBuilder();
+            }
+            properties_ = input.readMessage(org.tensorflow.proto.framework.DeviceProperties.parser(), extensionRegistry);
+            if (subBuilder != null) {
+              subBuilder.mergeFrom(properties_);
+              properties_ = subBuilder.buildPartial();
+            }
+
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.framework.DevicePropertiesProtos.internal_static_tensorflow_NamedDevice_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.framework.DevicePropertiesProtos.internal_static_tensorflow_NamedDevice_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.framework.NamedDevice.class, org.tensorflow.proto.framework.NamedDevice.Builder.class);
+  }
+
+  public static final int NAME_FIELD_NUMBER = 1;
+  private volatile java.lang.Object name_;
+  /**
+   * <code>string name = 1;</code>
+   */
+  public java.lang.String getName() {
+    java.lang.Object ref = name_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      name_ = s;
+      return s;
+    }
+  }
+  /**
+   * <code>string name = 1;</code>
+   */
+  public com.google.protobuf.ByteString
+      getNameBytes() {
+    java.lang.Object ref = name_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      name_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int PROPERTIES_FIELD_NUMBER = 2;
+  private org.tensorflow.proto.framework.DeviceProperties properties_;
+  /**
+   * <code>.tensorflow.DeviceProperties properties = 2;</code>
+   */
+  public boolean hasProperties() {
+    return properties_ != null;
+  }
+  /**
+   * <code>.tensorflow.DeviceProperties properties = 2;</code>
+   */
+  public org.tensorflow.proto.framework.DeviceProperties getProperties() {
+    return properties_ == null ? org.tensorflow.proto.framework.DeviceProperties.getDefaultInstance() : properties_;
+  }
+  /**
+   * <code>.tensorflow.DeviceProperties properties = 2;</code>
+   */
+  public org.tensorflow.proto.framework.DevicePropertiesOrBuilder getPropertiesOrBuilder() {
+    return getProperties();
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    if (!getNameBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 1, name_);
+    }
+    if (properties_ != null) {
+      output.writeMessage(2, getProperties());
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (!getNameBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, name_);
+    }
+    if (properties_ != null) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(2, getProperties());
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.framework.NamedDevice)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.framework.NamedDevice other = (org.tensorflow.proto.framework.NamedDevice) obj;
+
+    if (!getName()
+        .equals(other.getName())) return false;
+    if (hasProperties() != other.hasProperties()) return false;
+    if (hasProperties()) {
+      if (!getProperties()
+          .equals(other.getProperties())) return false;
+    }
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    hash = (37 * hash) + NAME_FIELD_NUMBER;
+    hash = (53 * hash) + getName().hashCode();
+    if (hasProperties()) {
+      hash = (37 * hash) + PROPERTIES_FIELD_NUMBER;
+      hash = (53 * hash) + getProperties().hashCode();
+    }
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.framework.NamedDevice parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.NamedDevice parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.NamedDevice parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.NamedDevice parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.NamedDevice parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.NamedDevice parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.NamedDevice parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.NamedDevice parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.NamedDevice parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.NamedDevice parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.NamedDevice parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.NamedDevice parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.framework.NamedDevice prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * Protobuf type {@code tensorflow.NamedDevice}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.NamedDevice)
+      org.tensorflow.proto.framework.NamedDeviceOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.DevicePropertiesProtos.internal_static_tensorflow_NamedDevice_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.DevicePropertiesProtos.internal_static_tensorflow_NamedDevice_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.NamedDevice.class, org.tensorflow.proto.framework.NamedDevice.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.framework.NamedDevice.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      name_ = "";
+
+      if (propertiesBuilder_ == null) {
+        properties_ = null;
+      } else {
+        properties_ = null;
+        propertiesBuilder_ = null;
+      }
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.framework.DevicePropertiesProtos.internal_static_tensorflow_NamedDevice_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.NamedDevice getDefaultInstanceForType() {
+      return org.tensorflow.proto.framework.NamedDevice.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.NamedDevice build() {
+      org.tensorflow.proto.framework.NamedDevice result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.NamedDevice buildPartial() {
+      org.tensorflow.proto.framework.NamedDevice result = new org.tensorflow.proto.framework.NamedDevice(this);
+      result.name_ = name_;
+      if (propertiesBuilder_ == null) {
+        result.properties_ = properties_;
+      } else {
+        result.properties_ = propertiesBuilder_.build();
+      }
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.framework.NamedDevice) {
+        return mergeFrom((org.tensorflow.proto.framework.NamedDevice)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.framework.NamedDevice other) {
+      if (other == org.tensorflow.proto.framework.NamedDevice.getDefaultInstance()) return this;
+      if (!other.getName().isEmpty()) {
+        name_ = other.name_;
+        onChanged();
+      }
+      if (other.hasProperties()) {
+        mergeProperties(other.getProperties());
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.framework.NamedDevice parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.framework.NamedDevice) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+
+    private java.lang.Object name_ = "";
+    /**
+     * <code>string name = 1;</code>
+     */
+    public java.lang.String getName() {
+      java.lang.Object ref = name_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        name_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <code>string name = 1;</code>
+     */
+    public com.google.protobuf.ByteString
+        getNameBytes() {
+      java.lang.Object ref = name_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        name_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <code>string name = 1;</code>
+     */
+    public Builder setName(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      name_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>string name = 1;</code>
+     */
+    public Builder clearName() {
+      
+      name_ = getDefaultInstance().getName();
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>string name = 1;</code>
+     */
+    public Builder setNameBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      name_ = value;
+      onChanged();
+      return this;
+    }
+
+    private org.tensorflow.proto.framework.DeviceProperties properties_;
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.DeviceProperties, org.tensorflow.proto.framework.DeviceProperties.Builder, org.tensorflow.proto.framework.DevicePropertiesOrBuilder> propertiesBuilder_;
+    /**
+     * <code>.tensorflow.DeviceProperties properties = 2;</code>
+     */
+    public boolean hasProperties() {
+      return propertiesBuilder_ != null || properties_ != null;
+    }
+    /**
+     * <code>.tensorflow.DeviceProperties properties = 2;</code>
+     */
+    public org.tensorflow.proto.framework.DeviceProperties getProperties() {
+      if (propertiesBuilder_ == null) {
+        return properties_ == null ? org.tensorflow.proto.framework.DeviceProperties.getDefaultInstance() : properties_;
+      } else {
+        return propertiesBuilder_.getMessage();
+      }
+    }
+    /**
+     * <code>.tensorflow.DeviceProperties properties = 2;</code>
+     */
+    public Builder setProperties(org.tensorflow.proto.framework.DeviceProperties value) {
+      if (propertiesBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        properties_ = value;
+        onChanged();
+      } else {
+        propertiesBuilder_.setMessage(value);
+      }
+
+      return this;
+    }
+    /**
+     * <code>.tensorflow.DeviceProperties properties = 2;</code>
+     */
+    public Builder setProperties(
+        org.tensorflow.proto.framework.DeviceProperties.Builder builderForValue) {
+      if (propertiesBuilder_ == null) {
+        properties_ = builderForValue.build();
+        onChanged();
+      } else {
+        propertiesBuilder_.setMessage(builderForValue.build());
+      }
+
+      return this;
+    }
+    /**
+     * <code>.tensorflow.DeviceProperties properties = 2;</code>
+     */
+    public Builder mergeProperties(org.tensorflow.proto.framework.DeviceProperties value) {
+      if (propertiesBuilder_ == null) {
+        if (properties_ != null) {
+          properties_ =
+            org.tensorflow.proto.framework.DeviceProperties.newBuilder(properties_).mergeFrom(value).buildPartial();
+        } else {
+          properties_ = value;
+        }
+        onChanged();
+      } else {
+        propertiesBuilder_.mergeFrom(value);
+      }
+
+      return this;
+    }
+    /**
+     * <code>.tensorflow.DeviceProperties properties = 2;</code>
+     */
+    public Builder clearProperties() {
+      if (propertiesBuilder_ == null) {
+        properties_ = null;
+        onChanged();
+      } else {
+        properties_ = null;
+        propertiesBuilder_ = null;
+      }
+
+      return this;
+    }
+    /**
+     * <code>.tensorflow.DeviceProperties properties = 2;</code>
+     */
+    public org.tensorflow.proto.framework.DeviceProperties.Builder getPropertiesBuilder() {
+      
+      onChanged();
+      return getPropertiesFieldBuilder().getBuilder();
+    }
+    /**
+     * <code>.tensorflow.DeviceProperties properties = 2;</code>
+     */
+    public org.tensorflow.proto.framework.DevicePropertiesOrBuilder getPropertiesOrBuilder() {
+      if (propertiesBuilder_ != null) {
+        return propertiesBuilder_.getMessageOrBuilder();
+      } else {
+        return properties_ == null ?
+            org.tensorflow.proto.framework.DeviceProperties.getDefaultInstance() : properties_;
+      }
+    }
+    /**
+     * <code>.tensorflow.DeviceProperties properties = 2;</code>
+     */
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.DeviceProperties, org.tensorflow.proto.framework.DeviceProperties.Builder, org.tensorflow.proto.framework.DevicePropertiesOrBuilder> 
+        getPropertiesFieldBuilder() {
+      if (propertiesBuilder_ == null) {
+        propertiesBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+            org.tensorflow.proto.framework.DeviceProperties, org.tensorflow.proto.framework.DeviceProperties.Builder, org.tensorflow.proto.framework.DevicePropertiesOrBuilder>(
+                getProperties(),
+                getParentForChildren(),
+                isClean());
+        properties_ = null;
+      }
+      return propertiesBuilder_;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.NamedDevice)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.NamedDevice)
+  private static final org.tensorflow.proto.framework.NamedDevice DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.NamedDevice();
+  }
+
+  public static org.tensorflow.proto.framework.NamedDevice getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<NamedDevice>
+      PARSER = new com.google.protobuf.AbstractParser<NamedDevice>() {
+    @java.lang.Override
+    public NamedDevice parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new NamedDevice(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<NamedDevice> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<NamedDevice> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.framework.NamedDevice getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NamedDeviceOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NamedDeviceOrBuilder.java
new file mode 100644
index 00000000000..52eed19dd38
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NamedDeviceOrBuilder.java
@@ -0,0 +1,32 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/device_properties.proto
+
+package org.tensorflow.proto.framework;
+
+public interface NamedDeviceOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.NamedDevice)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <code>string name = 1;</code>
+   */
+  java.lang.String getName();
+  /**
+   * <code>string name = 1;</code>
+   */
+  com.google.protobuf.ByteString
+      getNameBytes();
+
+  /**
+   * <code>.tensorflow.DeviceProperties properties = 2;</code>
+   */
+  boolean hasProperties();
+  /**
+   * <code>.tensorflow.DeviceProperties properties = 2;</code>
+   */
+  org.tensorflow.proto.framework.DeviceProperties getProperties();
+  /**
+   * <code>.tensorflow.DeviceProperties properties = 2;</code>
+   */
+  org.tensorflow.proto.framework.DevicePropertiesOrBuilder getPropertiesOrBuilder();
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NamedTensorProto.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NamedTensorProto.java
new file mode 100644
index 00000000000..c5b5514e2e3
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NamedTensorProto.java
@@ -0,0 +1,859 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/named_tensor.proto
+
+package org.tensorflow.proto.framework;
+
+/**
+ * <pre>
+ * A pair of tensor name and tensor values.
+ * </pre>
+ *
+ * Protobuf type {@code tensorflow.NamedTensorProto}
+ */
+public  final class NamedTensorProto extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.NamedTensorProto)
+    NamedTensorProtoOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use NamedTensorProto.newBuilder() to construct.
+  private NamedTensorProto(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private NamedTensorProto() {
+    name_ = "";
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new NamedTensorProto();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private NamedTensorProto(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 10: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            name_ = s;
+            break;
+          }
+          case 18: {
+            org.tensorflow.proto.framework.TensorProto.Builder subBuilder = null;
+            if (tensor_ != null) {
+              subBuilder = tensor_.toBuilder();
+            }
+            tensor_ = input.readMessage(org.tensorflow.proto.framework.TensorProto.parser(), extensionRegistry);
+            if (subBuilder != null) {
+              subBuilder.mergeFrom(tensor_);
+              tensor_ = subBuilder.buildPartial();
+            }
+
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.framework.NamedTensorProtos.internal_static_tensorflow_NamedTensorProto_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.framework.NamedTensorProtos.internal_static_tensorflow_NamedTensorProto_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.framework.NamedTensorProto.class, org.tensorflow.proto.framework.NamedTensorProto.Builder.class);
+  }
+
+  public static final int NAME_FIELD_NUMBER = 1;
+  private volatile java.lang.Object name_;
+  /**
+   * <pre>
+   * Name of the tensor.
+   * </pre>
+   *
+   * <code>string name = 1;</code>
+   */
+  public java.lang.String getName() {
+    java.lang.Object ref = name_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      name_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * Name of the tensor.
+   * </pre>
+   *
+   * <code>string name = 1;</code>
+   */
+  public com.google.protobuf.ByteString
+      getNameBytes() {
+    java.lang.Object ref = name_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      name_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int TENSOR_FIELD_NUMBER = 2;
+  private org.tensorflow.proto.framework.TensorProto tensor_;
+  /**
+   * <pre>
+   * The client can populate a TensorProto using a tensorflow::Tensor`, or
+   * directly using the protobuf field accessors.
+   * The client specifies whether the returned tensor values should be
+   * filled tensor fields (float_val, int_val, etc.) or encoded in a
+   * compact form in tensor.tensor_content.
+   * </pre>
+   *
+   * <code>.tensorflow.TensorProto tensor = 2;</code>
+   */
+  public boolean hasTensor() {
+    return tensor_ != null;
+  }
+  /**
+   * <pre>
+   * The client can populate a TensorProto using a tensorflow::Tensor`, or
+   * directly using the protobuf field accessors.
+   * The client specifies whether the returned tensor values should be
+   * filled tensor fields (float_val, int_val, etc.) or encoded in a
+   * compact form in tensor.tensor_content.
+   * </pre>
+   *
+   * <code>.tensorflow.TensorProto tensor = 2;</code>
+   */
+  public org.tensorflow.proto.framework.TensorProto getTensor() {
+    return tensor_ == null ? org.tensorflow.proto.framework.TensorProto.getDefaultInstance() : tensor_;
+  }
+  /**
+   * <pre>
+   * The client can populate a TensorProto using a tensorflow::Tensor`, or
+   * directly using the protobuf field accessors.
+   * The client specifies whether the returned tensor values should be
+   * filled tensor fields (float_val, int_val, etc.) or encoded in a
+   * compact form in tensor.tensor_content.
+   * </pre>
+   *
+   * <code>.tensorflow.TensorProto tensor = 2;</code>
+   */
+  public org.tensorflow.proto.framework.TensorProtoOrBuilder getTensorOrBuilder() {
+    return getTensor();
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    if (!getNameBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 1, name_);
+    }
+    if (tensor_ != null) {
+      output.writeMessage(2, getTensor());
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (!getNameBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, name_);
+    }
+    if (tensor_ != null) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(2, getTensor());
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.framework.NamedTensorProto)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.framework.NamedTensorProto other = (org.tensorflow.proto.framework.NamedTensorProto) obj;
+
+    if (!getName()
+        .equals(other.getName())) return false;
+    if (hasTensor() != other.hasTensor()) return false;
+    if (hasTensor()) {
+      if (!getTensor()
+          .equals(other.getTensor())) return false;
+    }
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    hash = (37 * hash) + NAME_FIELD_NUMBER;
+    hash = (53 * hash) + getName().hashCode();
+    if (hasTensor()) {
+      hash = (37 * hash) + TENSOR_FIELD_NUMBER;
+      hash = (53 * hash) + getTensor().hashCode();
+    }
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.framework.NamedTensorProto parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.NamedTensorProto parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.NamedTensorProto parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.NamedTensorProto parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.NamedTensorProto parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.NamedTensorProto parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.NamedTensorProto parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.NamedTensorProto parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.NamedTensorProto parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.NamedTensorProto parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.NamedTensorProto parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.NamedTensorProto parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.framework.NamedTensorProto prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * <pre>
+   * A pair of tensor name and tensor values.
+   * </pre>
+   *
+   * Protobuf type {@code tensorflow.NamedTensorProto}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.NamedTensorProto)
+      org.tensorflow.proto.framework.NamedTensorProtoOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.NamedTensorProtos.internal_static_tensorflow_NamedTensorProto_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.NamedTensorProtos.internal_static_tensorflow_NamedTensorProto_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.NamedTensorProto.class, org.tensorflow.proto.framework.NamedTensorProto.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.framework.NamedTensorProto.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      name_ = "";
+
+      if (tensorBuilder_ == null) {
+        tensor_ = null;
+      } else {
+        tensor_ = null;
+        tensorBuilder_ = null;
+      }
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.framework.NamedTensorProtos.internal_static_tensorflow_NamedTensorProto_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.NamedTensorProto getDefaultInstanceForType() {
+      return org.tensorflow.proto.framework.NamedTensorProto.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.NamedTensorProto build() {
+      org.tensorflow.proto.framework.NamedTensorProto result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.NamedTensorProto buildPartial() {
+      org.tensorflow.proto.framework.NamedTensorProto result = new org.tensorflow.proto.framework.NamedTensorProto(this);
+      result.name_ = name_;
+      if (tensorBuilder_ == null) {
+        result.tensor_ = tensor_;
+      } else {
+        result.tensor_ = tensorBuilder_.build();
+      }
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.framework.NamedTensorProto) {
+        return mergeFrom((org.tensorflow.proto.framework.NamedTensorProto)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.framework.NamedTensorProto other) {
+      if (other == org.tensorflow.proto.framework.NamedTensorProto.getDefaultInstance()) return this;
+      if (!other.getName().isEmpty()) {
+        name_ = other.name_;
+        onChanged();
+      }
+      if (other.hasTensor()) {
+        mergeTensor(other.getTensor());
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.framework.NamedTensorProto parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.framework.NamedTensorProto) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+
+    private java.lang.Object name_ = "";
+    /**
+     * <pre>
+     * Name of the tensor.
+     * </pre>
+     *
+     * <code>string name = 1;</code>
+     */
+    public java.lang.String getName() {
+      java.lang.Object ref = name_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        name_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Name of the tensor.
+     * </pre>
+     *
+     * <code>string name = 1;</code>
+     */
+    public com.google.protobuf.ByteString
+        getNameBytes() {
+      java.lang.Object ref = name_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        name_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Name of the tensor.
+     * </pre>
+     *
+     * <code>string name = 1;</code>
+     */
+    public Builder setName(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      name_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Name of the tensor.
+     * </pre>
+     *
+     * <code>string name = 1;</code>
+     */
+    public Builder clearName() {
+      
+      name_ = getDefaultInstance().getName();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Name of the tensor.
+     * </pre>
+     *
+     * <code>string name = 1;</code>
+     */
+    public Builder setNameBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      name_ = value;
+      onChanged();
+      return this;
+    }
+
+    private org.tensorflow.proto.framework.TensorProto tensor_;
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.TensorProto, org.tensorflow.proto.framework.TensorProto.Builder, org.tensorflow.proto.framework.TensorProtoOrBuilder> tensorBuilder_;
+    /**
+     * <pre>
+     * The client can populate a TensorProto using a tensorflow::Tensor`, or
+     * directly using the protobuf field accessors.
+     * The client specifies whether the returned tensor values should be
+     * filled tensor fields (float_val, int_val, etc.) or encoded in a
+     * compact form in tensor.tensor_content.
+     * </pre>
+     *
+     * <code>.tensorflow.TensorProto tensor = 2;</code>
+     */
+    public boolean hasTensor() {
+      return tensorBuilder_ != null || tensor_ != null;
+    }
+    /**
+     * <pre>
+     * The client can populate a TensorProto using a tensorflow::Tensor`, or
+     * directly using the protobuf field accessors.
+     * The client specifies whether the returned tensor values should be
+     * filled tensor fields (float_val, int_val, etc.) or encoded in a
+     * compact form in tensor.tensor_content.
+     * </pre>
+     *
+     * <code>.tensorflow.TensorProto tensor = 2;</code>
+     */
+    public org.tensorflow.proto.framework.TensorProto getTensor() {
+      if (tensorBuilder_ == null) {
+        return tensor_ == null ? org.tensorflow.proto.framework.TensorProto.getDefaultInstance() : tensor_;
+      } else {
+        return tensorBuilder_.getMessage();
+      }
+    }
+    /**
+     * <pre>
+     * The client can populate a TensorProto using a tensorflow::Tensor`, or
+     * directly using the protobuf field accessors.
+     * The client specifies whether the returned tensor values should be
+     * filled tensor fields (float_val, int_val, etc.) or encoded in a
+     * compact form in tensor.tensor_content.
+     * </pre>
+     *
+     * <code>.tensorflow.TensorProto tensor = 2;</code>
+     */
+    public Builder setTensor(org.tensorflow.proto.framework.TensorProto value) {
+      if (tensorBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        tensor_ = value;
+        onChanged();
+      } else {
+        tensorBuilder_.setMessage(value);
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * The client can populate a TensorProto using a tensorflow::Tensor`, or
+     * directly using the protobuf field accessors.
+     * The client specifies whether the returned tensor values should be
+     * filled tensor fields (float_val, int_val, etc.) or encoded in a
+     * compact form in tensor.tensor_content.
+     * </pre>
+     *
+     * <code>.tensorflow.TensorProto tensor = 2;</code>
+     */
+    public Builder setTensor(
+        org.tensorflow.proto.framework.TensorProto.Builder builderForValue) {
+      if (tensorBuilder_ == null) {
+        tensor_ = builderForValue.build();
+        onChanged();
+      } else {
+        tensorBuilder_.setMessage(builderForValue.build());
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * The client can populate a TensorProto using a tensorflow::Tensor`, or
+     * directly using the protobuf field accessors.
+     * The client specifies whether the returned tensor values should be
+     * filled tensor fields (float_val, int_val, etc.) or encoded in a
+     * compact form in tensor.tensor_content.
+     * </pre>
+     *
+     * <code>.tensorflow.TensorProto tensor = 2;</code>
+     */
+    public Builder mergeTensor(org.tensorflow.proto.framework.TensorProto value) {
+      if (tensorBuilder_ == null) {
+        if (tensor_ != null) {
+          tensor_ =
+            org.tensorflow.proto.framework.TensorProto.newBuilder(tensor_).mergeFrom(value).buildPartial();
+        } else {
+          tensor_ = value;
+        }
+        onChanged();
+      } else {
+        tensorBuilder_.mergeFrom(value);
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * The client can populate a TensorProto using a tensorflow::Tensor`, or
+     * directly using the protobuf field accessors.
+     * The client specifies whether the returned tensor values should be
+     * filled tensor fields (float_val, int_val, etc.) or encoded in a
+     * compact form in tensor.tensor_content.
+     * </pre>
+     *
+     * <code>.tensorflow.TensorProto tensor = 2;</code>
+     */
+    public Builder clearTensor() {
+      if (tensorBuilder_ == null) {
+        tensor_ = null;
+        onChanged();
+      } else {
+        tensor_ = null;
+        tensorBuilder_ = null;
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * The client can populate a TensorProto using a tensorflow::Tensor`, or
+     * directly using the protobuf field accessors.
+     * The client specifies whether the returned tensor values should be
+     * filled tensor fields (float_val, int_val, etc.) or encoded in a
+     * compact form in tensor.tensor_content.
+     * </pre>
+     *
+     * <code>.tensorflow.TensorProto tensor = 2;</code>
+     */
+    public org.tensorflow.proto.framework.TensorProto.Builder getTensorBuilder() {
+      
+      onChanged();
+      return getTensorFieldBuilder().getBuilder();
+    }
+    /**
+     * <pre>
+     * The client can populate a TensorProto using a tensorflow::Tensor`, or
+     * directly using the protobuf field accessors.
+     * The client specifies whether the returned tensor values should be
+     * filled tensor fields (float_val, int_val, etc.) or encoded in a
+     * compact form in tensor.tensor_content.
+     * </pre>
+     *
+     * <code>.tensorflow.TensorProto tensor = 2;</code>
+     */
+    public org.tensorflow.proto.framework.TensorProtoOrBuilder getTensorOrBuilder() {
+      if (tensorBuilder_ != null) {
+        return tensorBuilder_.getMessageOrBuilder();
+      } else {
+        return tensor_ == null ?
+            org.tensorflow.proto.framework.TensorProto.getDefaultInstance() : tensor_;
+      }
+    }
+    /**
+     * <pre>
+     * The client can populate a TensorProto using a tensorflow::Tensor`, or
+     * directly using the protobuf field accessors.
+     * The client specifies whether the returned tensor values should be
+     * filled tensor fields (float_val, int_val, etc.) or encoded in a
+     * compact form in tensor.tensor_content.
+     * </pre>
+     *
+     * <code>.tensorflow.TensorProto tensor = 2;</code>
+     */
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.TensorProto, org.tensorflow.proto.framework.TensorProto.Builder, org.tensorflow.proto.framework.TensorProtoOrBuilder> 
+        getTensorFieldBuilder() {
+      if (tensorBuilder_ == null) {
+        tensorBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+            org.tensorflow.proto.framework.TensorProto, org.tensorflow.proto.framework.TensorProto.Builder, org.tensorflow.proto.framework.TensorProtoOrBuilder>(
+                getTensor(),
+                getParentForChildren(),
+                isClean());
+        tensor_ = null;
+      }
+      return tensorBuilder_;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.NamedTensorProto)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.NamedTensorProto)
+  private static final org.tensorflow.proto.framework.NamedTensorProto DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.NamedTensorProto();
+  }
+
+  public static org.tensorflow.proto.framework.NamedTensorProto getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<NamedTensorProto>
+      PARSER = new com.google.protobuf.AbstractParser<NamedTensorProto>() {
+    @java.lang.Override
+    public NamedTensorProto parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new NamedTensorProto(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<NamedTensorProto> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<NamedTensorProto> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.framework.NamedTensorProto getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NamedTensorProtoOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NamedTensorProtoOrBuilder.java
new file mode 100644
index 00000000000..41b758b2f91
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NamedTensorProtoOrBuilder.java
@@ -0,0 +1,64 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/named_tensor.proto
+
+package org.tensorflow.proto.framework;
+
+public interface NamedTensorProtoOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.NamedTensorProto)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <pre>
+   * Name of the tensor.
+   * </pre>
+   *
+   * <code>string name = 1;</code>
+   */
+  java.lang.String getName();
+  /**
+   * <pre>
+   * Name of the tensor.
+   * </pre>
+   *
+   * <code>string name = 1;</code>
+   */
+  com.google.protobuf.ByteString
+      getNameBytes();
+
+  /**
+   * <pre>
+   * The client can populate a TensorProto using a tensorflow::Tensor`, or
+   * directly using the protobuf field accessors.
+   * The client specifies whether the returned tensor values should be
+   * filled tensor fields (float_val, int_val, etc.) or encoded in a
+   * compact form in tensor.tensor_content.
+   * </pre>
+   *
+   * <code>.tensorflow.TensorProto tensor = 2;</code>
+   */
+  boolean hasTensor();
+  /**
+   * <pre>
+   * The client can populate a TensorProto using a tensorflow::Tensor`, or
+   * directly using the protobuf field accessors.
+   * The client specifies whether the returned tensor values should be
+   * filled tensor fields (float_val, int_val, etc.) or encoded in a
+   * compact form in tensor.tensor_content.
+   * </pre>
+   *
+   * <code>.tensorflow.TensorProto tensor = 2;</code>
+   */
+  org.tensorflow.proto.framework.TensorProto getTensor();
+  /**
+   * <pre>
+   * The client can populate a TensorProto using a tensorflow::Tensor`, or
+   * directly using the protobuf field accessors.
+   * The client specifies whether the returned tensor values should be
+   * filled tensor fields (float_val, int_val, etc.) or encoded in a
+   * compact form in tensor.tensor_content.
+   * </pre>
+   *
+   * <code>.tensorflow.TensorProto tensor = 2;</code>
+   */
+  org.tensorflow.proto.framework.TensorProtoOrBuilder getTensorOrBuilder();
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NamedTensorProtos.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NamedTensorProtos.java
new file mode 100644
index 00000000000..98685b73c33
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NamedTensorProtos.java
@@ -0,0 +1,55 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/named_tensor.proto
+
+package org.tensorflow.proto.framework;
+
+public final class NamedTensorProtos {
+  private NamedTensorProtos() {}
+  public static void registerAllExtensions(
+      com.google.protobuf.ExtensionRegistryLite registry) {
+  }
+
+  public static void registerAllExtensions(
+      com.google.protobuf.ExtensionRegistry registry) {
+    registerAllExtensions(
+        (com.google.protobuf.ExtensionRegistryLite) registry);
+  }
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_NamedTensorProto_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_NamedTensorProto_fieldAccessorTable;
+
+  public static com.google.protobuf.Descriptors.FileDescriptor
+      getDescriptor() {
+    return descriptor;
+  }
+  private static  com.google.protobuf.Descriptors.FileDescriptor
+      descriptor;
+  static {
+    java.lang.String[] descriptorData = {
+      "\n+tensorflow/core/protobuf/named_tensor." +
+      "proto\022\ntensorflow\032&tensorflow/core/frame" +
+      "work/tensor.proto\"I\n\020NamedTensorProto\022\014\n" +
+      "\004name\030\001 \001(\t\022\'\n\006tensor\030\002 \001(\0132\027.tensorflow" +
+      ".TensorProtoB\202\001\n\036org.tensorflow.proto.fr" +
+      "ameworkB\021NamedTensorProtosP\001ZHgithub.com" +
+      "/tensorflow/tensorflow/tensorflow/go/cor" +
+      "e/core_protos_go_proto\370\001\001b\006proto3"
+    };
+    descriptor = com.google.protobuf.Descriptors.FileDescriptor
+      .internalBuildGeneratedFileFrom(descriptorData,
+        new com.google.protobuf.Descriptors.FileDescriptor[] {
+          org.tensorflow.proto.framework.TensorProtos.getDescriptor(),
+        });
+    internal_static_tensorflow_NamedTensorProto_descriptor =
+      getDescriptor().getMessageTypes().get(0);
+    internal_static_tensorflow_NamedTensorProto_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_NamedTensorProto_descriptor,
+        new java.lang.String[] { "Name", "Tensor", });
+    org.tensorflow.proto.framework.TensorProtos.getDescriptor();
+  }
+
+  // @@protoc_insertion_point(outer_class_scope)
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NamedTupleValue.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NamedTupleValue.java
new file mode 100644
index 00000000000..e1adfe3a508
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NamedTupleValue.java
@@ -0,0 +1,900 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/struct.proto
+
+package org.tensorflow.proto.framework;
+
+/**
+ * <pre>
+ * Represents Python's namedtuple.
+ * </pre>
+ *
+ * Protobuf type {@code tensorflow.NamedTupleValue}
+ */
+public  final class NamedTupleValue extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.NamedTupleValue)
+    NamedTupleValueOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use NamedTupleValue.newBuilder() to construct.
+  private NamedTupleValue(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private NamedTupleValue() {
+    name_ = "";
+    values_ = java.util.Collections.emptyList();
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new NamedTupleValue();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private NamedTupleValue(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    int mutable_bitField0_ = 0;
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 10: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            name_ = s;
+            break;
+          }
+          case 18: {
+            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
+              values_ = new java.util.ArrayList<org.tensorflow.proto.framework.PairValue>();
+              mutable_bitField0_ |= 0x00000001;
+            }
+            values_.add(
+                input.readMessage(org.tensorflow.proto.framework.PairValue.parser(), extensionRegistry));
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      if (((mutable_bitField0_ & 0x00000001) != 0)) {
+        values_ = java.util.Collections.unmodifiableList(values_);
+      }
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.framework.StructProtos.internal_static_tensorflow_NamedTupleValue_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.framework.StructProtos.internal_static_tensorflow_NamedTupleValue_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.framework.NamedTupleValue.class, org.tensorflow.proto.framework.NamedTupleValue.Builder.class);
+  }
+
+  public static final int NAME_FIELD_NUMBER = 1;
+  private volatile java.lang.Object name_;
+  /**
+   * <code>string name = 1;</code>
+   */
+  public java.lang.String getName() {
+    java.lang.Object ref = name_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      name_ = s;
+      return s;
+    }
+  }
+  /**
+   * <code>string name = 1;</code>
+   */
+  public com.google.protobuf.ByteString
+      getNameBytes() {
+    java.lang.Object ref = name_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      name_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int VALUES_FIELD_NUMBER = 2;
+  private java.util.List<org.tensorflow.proto.framework.PairValue> values_;
+  /**
+   * <code>repeated .tensorflow.PairValue values = 2;</code>
+   */
+  public java.util.List<org.tensorflow.proto.framework.PairValue> getValuesList() {
+    return values_;
+  }
+  /**
+   * <code>repeated .tensorflow.PairValue values = 2;</code>
+   */
+  public java.util.List<? extends org.tensorflow.proto.framework.PairValueOrBuilder> 
+      getValuesOrBuilderList() {
+    return values_;
+  }
+  /**
+   * <code>repeated .tensorflow.PairValue values = 2;</code>
+   */
+  public int getValuesCount() {
+    return values_.size();
+  }
+  /**
+   * <code>repeated .tensorflow.PairValue values = 2;</code>
+   */
+  public org.tensorflow.proto.framework.PairValue getValues(int index) {
+    return values_.get(index);
+  }
+  /**
+   * <code>repeated .tensorflow.PairValue values = 2;</code>
+   */
+  public org.tensorflow.proto.framework.PairValueOrBuilder getValuesOrBuilder(
+      int index) {
+    return values_.get(index);
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    if (!getNameBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 1, name_);
+    }
+    for (int i = 0; i < values_.size(); i++) {
+      output.writeMessage(2, values_.get(i));
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (!getNameBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, name_);
+    }
+    for (int i = 0; i < values_.size(); i++) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(2, values_.get(i));
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.framework.NamedTupleValue)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.framework.NamedTupleValue other = (org.tensorflow.proto.framework.NamedTupleValue) obj;
+
+    if (!getName()
+        .equals(other.getName())) return false;
+    if (!getValuesList()
+        .equals(other.getValuesList())) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    hash = (37 * hash) + NAME_FIELD_NUMBER;
+    hash = (53 * hash) + getName().hashCode();
+    if (getValuesCount() > 0) {
+      hash = (37 * hash) + VALUES_FIELD_NUMBER;
+      hash = (53 * hash) + getValuesList().hashCode();
+    }
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.framework.NamedTupleValue parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.NamedTupleValue parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.NamedTupleValue parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.NamedTupleValue parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.NamedTupleValue parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.NamedTupleValue parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.NamedTupleValue parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.NamedTupleValue parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.NamedTupleValue parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.NamedTupleValue parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.NamedTupleValue parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.NamedTupleValue parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.framework.NamedTupleValue prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * <pre>
+   * Represents Python's namedtuple.
+   * </pre>
+   *
+   * Protobuf type {@code tensorflow.NamedTupleValue}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.NamedTupleValue)
+      org.tensorflow.proto.framework.NamedTupleValueOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.StructProtos.internal_static_tensorflow_NamedTupleValue_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.StructProtos.internal_static_tensorflow_NamedTupleValue_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.NamedTupleValue.class, org.tensorflow.proto.framework.NamedTupleValue.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.framework.NamedTupleValue.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+        getValuesFieldBuilder();
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      name_ = "";
+
+      if (valuesBuilder_ == null) {
+        values_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000001);
+      } else {
+        valuesBuilder_.clear();
+      }
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.framework.StructProtos.internal_static_tensorflow_NamedTupleValue_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.NamedTupleValue getDefaultInstanceForType() {
+      return org.tensorflow.proto.framework.NamedTupleValue.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.NamedTupleValue build() {
+      org.tensorflow.proto.framework.NamedTupleValue result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.NamedTupleValue buildPartial() {
+      org.tensorflow.proto.framework.NamedTupleValue result = new org.tensorflow.proto.framework.NamedTupleValue(this);
+      int from_bitField0_ = bitField0_;
+      result.name_ = name_;
+      if (valuesBuilder_ == null) {
+        if (((bitField0_ & 0x00000001) != 0)) {
+          values_ = java.util.Collections.unmodifiableList(values_);
+          bitField0_ = (bitField0_ & ~0x00000001);
+        }
+        result.values_ = values_;
+      } else {
+        result.values_ = valuesBuilder_.build();
+      }
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.framework.NamedTupleValue) {
+        return mergeFrom((org.tensorflow.proto.framework.NamedTupleValue)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.framework.NamedTupleValue other) {
+      if (other == org.tensorflow.proto.framework.NamedTupleValue.getDefaultInstance()) return this;
+      if (!other.getName().isEmpty()) {
+        name_ = other.name_;
+        onChanged();
+      }
+      if (valuesBuilder_ == null) {
+        if (!other.values_.isEmpty()) {
+          if (values_.isEmpty()) {
+            values_ = other.values_;
+            bitField0_ = (bitField0_ & ~0x00000001);
+          } else {
+            ensureValuesIsMutable();
+            values_.addAll(other.values_);
+          }
+          onChanged();
+        }
+      } else {
+        if (!other.values_.isEmpty()) {
+          if (valuesBuilder_.isEmpty()) {
+            valuesBuilder_.dispose();
+            valuesBuilder_ = null;
+            values_ = other.values_;
+            bitField0_ = (bitField0_ & ~0x00000001);
+            valuesBuilder_ = 
+              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
+                 getValuesFieldBuilder() : null;
+          } else {
+            valuesBuilder_.addAllMessages(other.values_);
+          }
+        }
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.framework.NamedTupleValue parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.framework.NamedTupleValue) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+    private int bitField0_;
+
+    private java.lang.Object name_ = "";
+    /**
+     * <code>string name = 1;</code>
+     */
+    public java.lang.String getName() {
+      java.lang.Object ref = name_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        name_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <code>string name = 1;</code>
+     */
+    public com.google.protobuf.ByteString
+        getNameBytes() {
+      java.lang.Object ref = name_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        name_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <code>string name = 1;</code>
+     */
+    public Builder setName(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      name_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>string name = 1;</code>
+     */
+    public Builder clearName() {
+      
+      name_ = getDefaultInstance().getName();
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>string name = 1;</code>
+     */
+    public Builder setNameBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      name_ = value;
+      onChanged();
+      return this;
+    }
+
+    private java.util.List<org.tensorflow.proto.framework.PairValue> values_ =
+      java.util.Collections.emptyList();
+    private void ensureValuesIsMutable() {
+      if (!((bitField0_ & 0x00000001) != 0)) {
+        values_ = new java.util.ArrayList<org.tensorflow.proto.framework.PairValue>(values_);
+        bitField0_ |= 0x00000001;
+       }
+    }
+
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.framework.PairValue, org.tensorflow.proto.framework.PairValue.Builder, org.tensorflow.proto.framework.PairValueOrBuilder> valuesBuilder_;
+
+    /**
+     * <code>repeated .tensorflow.PairValue values = 2;</code>
+     */
+    public java.util.List<org.tensorflow.proto.framework.PairValue> getValuesList() {
+      if (valuesBuilder_ == null) {
+        return java.util.Collections.unmodifiableList(values_);
+      } else {
+        return valuesBuilder_.getMessageList();
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.PairValue values = 2;</code>
+     */
+    public int getValuesCount() {
+      if (valuesBuilder_ == null) {
+        return values_.size();
+      } else {
+        return valuesBuilder_.getCount();
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.PairValue values = 2;</code>
+     */
+    public org.tensorflow.proto.framework.PairValue getValues(int index) {
+      if (valuesBuilder_ == null) {
+        return values_.get(index);
+      } else {
+        return valuesBuilder_.getMessage(index);
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.PairValue values = 2;</code>
+     */
+    public Builder setValues(
+        int index, org.tensorflow.proto.framework.PairValue value) {
+      if (valuesBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureValuesIsMutable();
+        values_.set(index, value);
+        onChanged();
+      } else {
+        valuesBuilder_.setMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.PairValue values = 2;</code>
+     */
+    public Builder setValues(
+        int index, org.tensorflow.proto.framework.PairValue.Builder builderForValue) {
+      if (valuesBuilder_ == null) {
+        ensureValuesIsMutable();
+        values_.set(index, builderForValue.build());
+        onChanged();
+      } else {
+        valuesBuilder_.setMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.PairValue values = 2;</code>
+     */
+    public Builder addValues(org.tensorflow.proto.framework.PairValue value) {
+      if (valuesBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureValuesIsMutable();
+        values_.add(value);
+        onChanged();
+      } else {
+        valuesBuilder_.addMessage(value);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.PairValue values = 2;</code>
+     */
+    public Builder addValues(
+        int index, org.tensorflow.proto.framework.PairValue value) {
+      if (valuesBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureValuesIsMutable();
+        values_.add(index, value);
+        onChanged();
+      } else {
+        valuesBuilder_.addMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.PairValue values = 2;</code>
+     */
+    public Builder addValues(
+        org.tensorflow.proto.framework.PairValue.Builder builderForValue) {
+      if (valuesBuilder_ == null) {
+        ensureValuesIsMutable();
+        values_.add(builderForValue.build());
+        onChanged();
+      } else {
+        valuesBuilder_.addMessage(builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.PairValue values = 2;</code>
+     */
+    public Builder addValues(
+        int index, org.tensorflow.proto.framework.PairValue.Builder builderForValue) {
+      if (valuesBuilder_ == null) {
+        ensureValuesIsMutable();
+        values_.add(index, builderForValue.build());
+        onChanged();
+      } else {
+        valuesBuilder_.addMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.PairValue values = 2;</code>
+     */
+    public Builder addAllValues(
+        java.lang.Iterable<? extends org.tensorflow.proto.framework.PairValue> values) {
+      if (valuesBuilder_ == null) {
+        ensureValuesIsMutable();
+        com.google.protobuf.AbstractMessageLite.Builder.addAll(
+            values, values_);
+        onChanged();
+      } else {
+        valuesBuilder_.addAllMessages(values);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.PairValue values = 2;</code>
+     */
+    public Builder clearValues() {
+      if (valuesBuilder_ == null) {
+        values_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000001);
+        onChanged();
+      } else {
+        valuesBuilder_.clear();
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.PairValue values = 2;</code>
+     */
+    public Builder removeValues(int index) {
+      if (valuesBuilder_ == null) {
+        ensureValuesIsMutable();
+        values_.remove(index);
+        onChanged();
+      } else {
+        valuesBuilder_.remove(index);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.PairValue values = 2;</code>
+     */
+    public org.tensorflow.proto.framework.PairValue.Builder getValuesBuilder(
+        int index) {
+      return getValuesFieldBuilder().getBuilder(index);
+    }
+    /**
+     * <code>repeated .tensorflow.PairValue values = 2;</code>
+     */
+    public org.tensorflow.proto.framework.PairValueOrBuilder getValuesOrBuilder(
+        int index) {
+      if (valuesBuilder_ == null) {
+        return values_.get(index);  } else {
+        return valuesBuilder_.getMessageOrBuilder(index);
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.PairValue values = 2;</code>
+     */
+    public java.util.List<? extends org.tensorflow.proto.framework.PairValueOrBuilder> 
+         getValuesOrBuilderList() {
+      if (valuesBuilder_ != null) {
+        return valuesBuilder_.getMessageOrBuilderList();
+      } else {
+        return java.util.Collections.unmodifiableList(values_);
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.PairValue values = 2;</code>
+     */
+    public org.tensorflow.proto.framework.PairValue.Builder addValuesBuilder() {
+      return getValuesFieldBuilder().addBuilder(
+          org.tensorflow.proto.framework.PairValue.getDefaultInstance());
+    }
+    /**
+     * <code>repeated .tensorflow.PairValue values = 2;</code>
+     */
+    public org.tensorflow.proto.framework.PairValue.Builder addValuesBuilder(
+        int index) {
+      return getValuesFieldBuilder().addBuilder(
+          index, org.tensorflow.proto.framework.PairValue.getDefaultInstance());
+    }
+    /**
+     * <code>repeated .tensorflow.PairValue values = 2;</code>
+     */
+    public java.util.List<org.tensorflow.proto.framework.PairValue.Builder> 
+         getValuesBuilderList() {
+      return getValuesFieldBuilder().getBuilderList();
+    }
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.framework.PairValue, org.tensorflow.proto.framework.PairValue.Builder, org.tensorflow.proto.framework.PairValueOrBuilder> 
+        getValuesFieldBuilder() {
+      if (valuesBuilder_ == null) {
+        valuesBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
+            org.tensorflow.proto.framework.PairValue, org.tensorflow.proto.framework.PairValue.Builder, org.tensorflow.proto.framework.PairValueOrBuilder>(
+                values_,
+                ((bitField0_ & 0x00000001) != 0),
+                getParentForChildren(),
+                isClean());
+        values_ = null;
+      }
+      return valuesBuilder_;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.NamedTupleValue)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.NamedTupleValue)
+  private static final org.tensorflow.proto.framework.NamedTupleValue DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.NamedTupleValue();
+  }
+
+  public static org.tensorflow.proto.framework.NamedTupleValue getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<NamedTupleValue>
+      PARSER = new com.google.protobuf.AbstractParser<NamedTupleValue>() {
+    @java.lang.Override
+    public NamedTupleValue parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new NamedTupleValue(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<NamedTupleValue> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<NamedTupleValue> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.framework.NamedTupleValue getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NamedTupleValueOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NamedTupleValueOrBuilder.java
new file mode 100644
index 00000000000..6829f4aa2ab
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NamedTupleValueOrBuilder.java
@@ -0,0 +1,43 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/struct.proto
+
+package org.tensorflow.proto.framework;
+
+public interface NamedTupleValueOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.NamedTupleValue)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <code>string name = 1;</code>
+   */
+  java.lang.String getName();
+  /**
+   * <code>string name = 1;</code>
+   */
+  com.google.protobuf.ByteString
+      getNameBytes();
+
+  /**
+   * <code>repeated .tensorflow.PairValue values = 2;</code>
+   */
+  java.util.List<org.tensorflow.proto.framework.PairValue> 
+      getValuesList();
+  /**
+   * <code>repeated .tensorflow.PairValue values = 2;</code>
+   */
+  org.tensorflow.proto.framework.PairValue getValues(int index);
+  /**
+   * <code>repeated .tensorflow.PairValue values = 2;</code>
+   */
+  int getValuesCount();
+  /**
+   * <code>repeated .tensorflow.PairValue values = 2;</code>
+   */
+  java.util.List<? extends org.tensorflow.proto.framework.PairValueOrBuilder> 
+      getValuesOrBuilderList();
+  /**
+   * <code>repeated .tensorflow.PairValue values = 2;</code>
+   */
+  org.tensorflow.proto.framework.PairValueOrBuilder getValuesOrBuilder(
+      int index);
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NodeDef.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NodeDef.java
new file mode 100644
index 00000000000..9bfd5c5c1f3
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NodeDef.java
@@ -0,0 +1,3076 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/framework/node_def.proto
+
+package org.tensorflow.proto.framework;
+
+/**
+ * Protobuf type {@code tensorflow.NodeDef}
+ */
+public  final class NodeDef extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.NodeDef)
+    NodeDefOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use NodeDef.newBuilder() to construct.
+  private NodeDef(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private NodeDef() {
+    name_ = "";
+    op_ = "";
+    input_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+    device_ = "";
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new NodeDef();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private NodeDef(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    int mutable_bitField0_ = 0;
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 10: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            name_ = s;
+            break;
+          }
+          case 18: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            op_ = s;
+            break;
+          }
+          case 26: {
+            java.lang.String s = input.readStringRequireUtf8();
+            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
+              input_ = new com.google.protobuf.LazyStringArrayList();
+              mutable_bitField0_ |= 0x00000001;
+            }
+            input_.add(s);
+            break;
+          }
+          case 34: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            device_ = s;
+            break;
+          }
+          case 42: {
+            if (!((mutable_bitField0_ & 0x00000002) != 0)) {
+              attr_ = com.google.protobuf.MapField.newMapField(
+                  AttrDefaultEntryHolder.defaultEntry);
+              mutable_bitField0_ |= 0x00000002;
+            }
+            com.google.protobuf.MapEntry<java.lang.String, org.tensorflow.proto.framework.AttrValue>
+            attr__ = input.readMessage(
+                AttrDefaultEntryHolder.defaultEntry.getParserForType(), extensionRegistry);
+            attr_.getMutableMap().put(
+                attr__.getKey(), attr__.getValue());
+            break;
+          }
+          case 50: {
+            org.tensorflow.proto.framework.NodeDef.ExperimentalDebugInfo.Builder subBuilder = null;
+            if (experimentalDebugInfo_ != null) {
+              subBuilder = experimentalDebugInfo_.toBuilder();
+            }
+            experimentalDebugInfo_ = input.readMessage(org.tensorflow.proto.framework.NodeDef.ExperimentalDebugInfo.parser(), extensionRegistry);
+            if (subBuilder != null) {
+              subBuilder.mergeFrom(experimentalDebugInfo_);
+              experimentalDebugInfo_ = subBuilder.buildPartial();
+            }
+
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      if (((mutable_bitField0_ & 0x00000001) != 0)) {
+        input_ = input_.getUnmodifiableView();
+      }
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.framework.NodeProto.internal_static_tensorflow_NodeDef_descriptor;
+  }
+
+  @SuppressWarnings({"rawtypes"})
+  @java.lang.Override
+  protected com.google.protobuf.MapField internalGetMapField(
+      int number) {
+    switch (number) {
+      case 5:
+        return internalGetAttr();
+      default:
+        throw new RuntimeException(
+            "Invalid map field number: " + number);
+    }
+  }
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.framework.NodeProto.internal_static_tensorflow_NodeDef_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.framework.NodeDef.class, org.tensorflow.proto.framework.NodeDef.Builder.class);
+  }
+
+  public interface ExperimentalDebugInfoOrBuilder extends
+      // @@protoc_insertion_point(interface_extends:tensorflow.NodeDef.ExperimentalDebugInfo)
+      com.google.protobuf.MessageOrBuilder {
+
+    /**
+     * <pre>
+     * Opaque string inserted into error messages created by the runtime.
+     * This is intended to store the list of names of the nodes from the
+     * original graph that this node was derived. For example if this node, say
+     * C, was result of a fusion of 2 nodes A and B, then 'original_node' would
+     * be {A, B}. This information can be used to map errors originating at the
+     * current node to some top level source code.
+     * </pre>
+     *
+     * <code>repeated string original_node_names = 1;</code>
+     */
+    java.util.List<java.lang.String>
+        getOriginalNodeNamesList();
+    /**
+     * <pre>
+     * Opaque string inserted into error messages created by the runtime.
+     * This is intended to store the list of names of the nodes from the
+     * original graph that this node was derived. For example if this node, say
+     * C, was result of a fusion of 2 nodes A and B, then 'original_node' would
+     * be {A, B}. This information can be used to map errors originating at the
+     * current node to some top level source code.
+     * </pre>
+     *
+     * <code>repeated string original_node_names = 1;</code>
+     */
+    int getOriginalNodeNamesCount();
+    /**
+     * <pre>
+     * Opaque string inserted into error messages created by the runtime.
+     * This is intended to store the list of names of the nodes from the
+     * original graph that this node was derived. For example if this node, say
+     * C, was result of a fusion of 2 nodes A and B, then 'original_node' would
+     * be {A, B}. This information can be used to map errors originating at the
+     * current node to some top level source code.
+     * </pre>
+     *
+     * <code>repeated string original_node_names = 1;</code>
+     */
+    java.lang.String getOriginalNodeNames(int index);
+    /**
+     * <pre>
+     * Opaque string inserted into error messages created by the runtime.
+     * This is intended to store the list of names of the nodes from the
+     * original graph that this node was derived. For example if this node, say
+     * C, was result of a fusion of 2 nodes A and B, then 'original_node' would
+     * be {A, B}. This information can be used to map errors originating at the
+     * current node to some top level source code.
+     * </pre>
+     *
+     * <code>repeated string original_node_names = 1;</code>
+     */
+    com.google.protobuf.ByteString
+        getOriginalNodeNamesBytes(int index);
+
+    /**
+     * <pre>
+     * This is intended to store the list of names of the functions from the
+     * original graph that this node was derived. For example if this node, say
+     * C, was result of a fusion of node A in function FA and node B in function
+     * FB, then `original_funcs` would be {FA, FB}. If the node is in the top
+     * level graph, the `original_func` is empty. This information, with the
+     * `original_node_names` can be used to map errors originating at the
+     * current ndoe to some top level source code.
+     * </pre>
+     *
+     * <code>repeated string original_func_names = 2;</code>
+     */
+    java.util.List<java.lang.String>
+        getOriginalFuncNamesList();
+    /**
+     * <pre>
+     * This is intended to store the list of names of the functions from the
+     * original graph that this node was derived. For example if this node, say
+     * C, was result of a fusion of node A in function FA and node B in function
+     * FB, then `original_funcs` would be {FA, FB}. If the node is in the top
+     * level graph, the `original_func` is empty. This information, with the
+     * `original_node_names` can be used to map errors originating at the
+     * current ndoe to some top level source code.
+     * </pre>
+     *
+     * <code>repeated string original_func_names = 2;</code>
+     */
+    int getOriginalFuncNamesCount();
+    /**
+     * <pre>
+     * This is intended to store the list of names of the functions from the
+     * original graph that this node was derived. For example if this node, say
+     * C, was result of a fusion of node A in function FA and node B in function
+     * FB, then `original_funcs` would be {FA, FB}. If the node is in the top
+     * level graph, the `original_func` is empty. This information, with the
+     * `original_node_names` can be used to map errors originating at the
+     * current ndoe to some top level source code.
+     * </pre>
+     *
+     * <code>repeated string original_func_names = 2;</code>
+     */
+    java.lang.String getOriginalFuncNames(int index);
+    /**
+     * <pre>
+     * This is intended to store the list of names of the functions from the
+     * original graph that this node was derived. For example if this node, say
+     * C, was result of a fusion of node A in function FA and node B in function
+     * FB, then `original_funcs` would be {FA, FB}. If the node is in the top
+     * level graph, the `original_func` is empty. This information, with the
+     * `original_node_names` can be used to map errors originating at the
+     * current ndoe to some top level source code.
+     * </pre>
+     *
+     * <code>repeated string original_func_names = 2;</code>
+     */
+    com.google.protobuf.ByteString
+        getOriginalFuncNamesBytes(int index);
+  }
+  /**
+   * Protobuf type {@code tensorflow.NodeDef.ExperimentalDebugInfo}
+   */
+  public  static final class ExperimentalDebugInfo extends
+      com.google.protobuf.GeneratedMessageV3 implements
+      // @@protoc_insertion_point(message_implements:tensorflow.NodeDef.ExperimentalDebugInfo)
+      ExperimentalDebugInfoOrBuilder {
+  private static final long serialVersionUID = 0L;
+    // Use ExperimentalDebugInfo.newBuilder() to construct.
+    private ExperimentalDebugInfo(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+      super(builder);
+    }
+    private ExperimentalDebugInfo() {
+      originalNodeNames_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+      originalFuncNames_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+    }
+
+    @java.lang.Override
+    @SuppressWarnings({"unused"})
+    protected java.lang.Object newInstance(
+        UnusedPrivateParameter unused) {
+      return new ExperimentalDebugInfo();
+    }
+
+    @java.lang.Override
+    public final com.google.protobuf.UnknownFieldSet
+    getUnknownFields() {
+      return this.unknownFields;
+    }
+    private ExperimentalDebugInfo(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      this();
+      if (extensionRegistry == null) {
+        throw new java.lang.NullPointerException();
+      }
+      int mutable_bitField0_ = 0;
+      com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+          com.google.protobuf.UnknownFieldSet.newBuilder();
+      try {
+        boolean done = false;
+        while (!done) {
+          int tag = input.readTag();
+          switch (tag) {
+            case 0:
+              done = true;
+              break;
+            case 10: {
+              java.lang.String s = input.readStringRequireUtf8();
+              if (!((mutable_bitField0_ & 0x00000001) != 0)) {
+                originalNodeNames_ = new com.google.protobuf.LazyStringArrayList();
+                mutable_bitField0_ |= 0x00000001;
+              }
+              originalNodeNames_.add(s);
+              break;
+            }
+            case 18: {
+              java.lang.String s = input.readStringRequireUtf8();
+              if (!((mutable_bitField0_ & 0x00000002) != 0)) {
+                originalFuncNames_ = new com.google.protobuf.LazyStringArrayList();
+                mutable_bitField0_ |= 0x00000002;
+              }
+              originalFuncNames_.add(s);
+              break;
+            }
+            default: {
+              if (!parseUnknownField(
+                  input, unknownFields, extensionRegistry, tag)) {
+                done = true;
+              }
+              break;
+            }
+          }
+        }
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        throw e.setUnfinishedMessage(this);
+      } catch (java.io.IOException e) {
+        throw new com.google.protobuf.InvalidProtocolBufferException(
+            e).setUnfinishedMessage(this);
+      } finally {
+        if (((mutable_bitField0_ & 0x00000001) != 0)) {
+          originalNodeNames_ = originalNodeNames_.getUnmodifiableView();
+        }
+        if (((mutable_bitField0_ & 0x00000002) != 0)) {
+          originalFuncNames_ = originalFuncNames_.getUnmodifiableView();
+        }
+        this.unknownFields = unknownFields.build();
+        makeExtensionsImmutable();
+      }
+    }
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.NodeProto.internal_static_tensorflow_NodeDef_ExperimentalDebugInfo_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.NodeProto.internal_static_tensorflow_NodeDef_ExperimentalDebugInfo_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.NodeDef.ExperimentalDebugInfo.class, org.tensorflow.proto.framework.NodeDef.ExperimentalDebugInfo.Builder.class);
+    }
+
+    public static final int ORIGINAL_NODE_NAMES_FIELD_NUMBER = 1;
+    private com.google.protobuf.LazyStringList originalNodeNames_;
+    /**
+     * <pre>
+     * Opaque string inserted into error messages created by the runtime.
+     * This is intended to store the list of names of the nodes from the
+     * original graph that this node was derived. For example if this node, say
+     * C, was result of a fusion of 2 nodes A and B, then 'original_node' would
+     * be {A, B}. This information can be used to map errors originating at the
+     * current node to some top level source code.
+     * </pre>
+     *
+     * <code>repeated string original_node_names = 1;</code>
+     */
+    public com.google.protobuf.ProtocolStringList
+        getOriginalNodeNamesList() {
+      return originalNodeNames_;
+    }
+    /**
+     * <pre>
+     * Opaque string inserted into error messages created by the runtime.
+     * This is intended to store the list of names of the nodes from the
+     * original graph that this node was derived. For example if this node, say
+     * C, was result of a fusion of 2 nodes A and B, then 'original_node' would
+     * be {A, B}. This information can be used to map errors originating at the
+     * current node to some top level source code.
+     * </pre>
+     *
+     * <code>repeated string original_node_names = 1;</code>
+     */
+    public int getOriginalNodeNamesCount() {
+      return originalNodeNames_.size();
+    }
+    /**
+     * <pre>
+     * Opaque string inserted into error messages created by the runtime.
+     * This is intended to store the list of names of the nodes from the
+     * original graph that this node was derived. For example if this node, say
+     * C, was result of a fusion of 2 nodes A and B, then 'original_node' would
+     * be {A, B}. This information can be used to map errors originating at the
+     * current node to some top level source code.
+     * </pre>
+     *
+     * <code>repeated string original_node_names = 1;</code>
+     */
+    public java.lang.String getOriginalNodeNames(int index) {
+      return originalNodeNames_.get(index);
+    }
+    /**
+     * <pre>
+     * Opaque string inserted into error messages created by the runtime.
+     * This is intended to store the list of names of the nodes from the
+     * original graph that this node was derived. For example if this node, say
+     * C, was result of a fusion of 2 nodes A and B, then 'original_node' would
+     * be {A, B}. This information can be used to map errors originating at the
+     * current node to some top level source code.
+     * </pre>
+     *
+     * <code>repeated string original_node_names = 1;</code>
+     */
+    public com.google.protobuf.ByteString
+        getOriginalNodeNamesBytes(int index) {
+      return originalNodeNames_.getByteString(index);
+    }
+
+    public static final int ORIGINAL_FUNC_NAMES_FIELD_NUMBER = 2;
+    private com.google.protobuf.LazyStringList originalFuncNames_;
+    /**
+     * <pre>
+     * This is intended to store the list of names of the functions from the
+     * original graph that this node was derived. For example if this node, say
+     * C, was result of a fusion of node A in function FA and node B in function
+     * FB, then `original_funcs` would be {FA, FB}. If the node is in the top
+     * level graph, the `original_func` is empty. This information, with the
+     * `original_node_names` can be used to map errors originating at the
+     * current ndoe to some top level source code.
+     * </pre>
+     *
+     * <code>repeated string original_func_names = 2;</code>
+     */
+    public com.google.protobuf.ProtocolStringList
+        getOriginalFuncNamesList() {
+      return originalFuncNames_;
+    }
+    /**
+     * <pre>
+     * This is intended to store the list of names of the functions from the
+     * original graph that this node was derived. For example if this node, say
+     * C, was result of a fusion of node A in function FA and node B in function
+     * FB, then `original_funcs` would be {FA, FB}. If the node is in the top
+     * level graph, the `original_func` is empty. This information, with the
+     * `original_node_names` can be used to map errors originating at the
+     * current ndoe to some top level source code.
+     * </pre>
+     *
+     * <code>repeated string original_func_names = 2;</code>
+     */
+    public int getOriginalFuncNamesCount() {
+      return originalFuncNames_.size();
+    }
+    /**
+     * <pre>
+     * This is intended to store the list of names of the functions from the
+     * original graph that this node was derived. For example if this node, say
+     * C, was result of a fusion of node A in function FA and node B in function
+     * FB, then `original_funcs` would be {FA, FB}. If the node is in the top
+     * level graph, the `original_func` is empty. This information, with the
+     * `original_node_names` can be used to map errors originating at the
+     * current ndoe to some top level source code.
+     * </pre>
+     *
+     * <code>repeated string original_func_names = 2;</code>
+     */
+    public java.lang.String getOriginalFuncNames(int index) {
+      return originalFuncNames_.get(index);
+    }
+    /**
+     * <pre>
+     * This is intended to store the list of names of the functions from the
+     * original graph that this node was derived. For example if this node, say
+     * C, was result of a fusion of node A in function FA and node B in function
+     * FB, then `original_funcs` would be {FA, FB}. If the node is in the top
+     * level graph, the `original_func` is empty. This information, with the
+     * `original_node_names` can be used to map errors originating at the
+     * current ndoe to some top level source code.
+     * </pre>
+     *
+     * <code>repeated string original_func_names = 2;</code>
+     */
+    public com.google.protobuf.ByteString
+        getOriginalFuncNamesBytes(int index) {
+      return originalFuncNames_.getByteString(index);
+    }
+
+    private byte memoizedIsInitialized = -1;
+    @java.lang.Override
+    public final boolean isInitialized() {
+      byte isInitialized = memoizedIsInitialized;
+      if (isInitialized == 1) return true;
+      if (isInitialized == 0) return false;
+
+      memoizedIsInitialized = 1;
+      return true;
+    }
+
+    @java.lang.Override
+    public void writeTo(com.google.protobuf.CodedOutputStream output)
+                        throws java.io.IOException {
+      for (int i = 0; i < originalNodeNames_.size(); i++) {
+        com.google.protobuf.GeneratedMessageV3.writeString(output, 1, originalNodeNames_.getRaw(i));
+      }
+      for (int i = 0; i < originalFuncNames_.size(); i++) {
+        com.google.protobuf.GeneratedMessageV3.writeString(output, 2, originalFuncNames_.getRaw(i));
+      }
+      unknownFields.writeTo(output);
+    }
+
+    @java.lang.Override
+    public int getSerializedSize() {
+      int size = memoizedSize;
+      if (size != -1) return size;
+
+      size = 0;
+      {
+        int dataSize = 0;
+        for (int i = 0; i < originalNodeNames_.size(); i++) {
+          dataSize += computeStringSizeNoTag(originalNodeNames_.getRaw(i));
+        }
+        size += dataSize;
+        size += 1 * getOriginalNodeNamesList().size();
+      }
+      {
+        int dataSize = 0;
+        for (int i = 0; i < originalFuncNames_.size(); i++) {
+          dataSize += computeStringSizeNoTag(originalFuncNames_.getRaw(i));
+        }
+        size += dataSize;
+        size += 1 * getOriginalFuncNamesList().size();
+      }
+      size += unknownFields.getSerializedSize();
+      memoizedSize = size;
+      return size;
+    }
+
+    @java.lang.Override
+    public boolean equals(final java.lang.Object obj) {
+      if (obj == this) {
+       return true;
+      }
+      if (!(obj instanceof org.tensorflow.proto.framework.NodeDef.ExperimentalDebugInfo)) {
+        return super.equals(obj);
+      }
+      org.tensorflow.proto.framework.NodeDef.ExperimentalDebugInfo other = (org.tensorflow.proto.framework.NodeDef.ExperimentalDebugInfo) obj;
+
+      if (!getOriginalNodeNamesList()
+          .equals(other.getOriginalNodeNamesList())) return false;
+      if (!getOriginalFuncNamesList()
+          .equals(other.getOriginalFuncNamesList())) return false;
+      if (!unknownFields.equals(other.unknownFields)) return false;
+      return true;
+    }
+
+    @java.lang.Override
+    public int hashCode() {
+      if (memoizedHashCode != 0) {
+        return memoizedHashCode;
+      }
+      int hash = 41;
+      hash = (19 * hash) + getDescriptor().hashCode();
+      if (getOriginalNodeNamesCount() > 0) {
+        hash = (37 * hash) + ORIGINAL_NODE_NAMES_FIELD_NUMBER;
+        hash = (53 * hash) + getOriginalNodeNamesList().hashCode();
+      }
+      if (getOriginalFuncNamesCount() > 0) {
+        hash = (37 * hash) + ORIGINAL_FUNC_NAMES_FIELD_NUMBER;
+        hash = (53 * hash) + getOriginalFuncNamesList().hashCode();
+      }
+      hash = (29 * hash) + unknownFields.hashCode();
+      memoizedHashCode = hash;
+      return hash;
+    }
+
+    public static org.tensorflow.proto.framework.NodeDef.ExperimentalDebugInfo parseFrom(
+        java.nio.ByteBuffer data)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data);
+    }
+    public static org.tensorflow.proto.framework.NodeDef.ExperimentalDebugInfo parseFrom(
+        java.nio.ByteBuffer data,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.NodeDef.ExperimentalDebugInfo parseFrom(
+        com.google.protobuf.ByteString data)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data);
+    }
+    public static org.tensorflow.proto.framework.NodeDef.ExperimentalDebugInfo parseFrom(
+        com.google.protobuf.ByteString data,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.NodeDef.ExperimentalDebugInfo parseFrom(byte[] data)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data);
+    }
+    public static org.tensorflow.proto.framework.NodeDef.ExperimentalDebugInfo parseFrom(
+        byte[] data,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.NodeDef.ExperimentalDebugInfo parseFrom(java.io.InputStream input)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input);
+    }
+    public static org.tensorflow.proto.framework.NodeDef.ExperimentalDebugInfo parseFrom(
+        java.io.InputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.NodeDef.ExperimentalDebugInfo parseDelimitedFrom(java.io.InputStream input)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseDelimitedWithIOException(PARSER, input);
+    }
+    public static org.tensorflow.proto.framework.NodeDef.ExperimentalDebugInfo parseDelimitedFrom(
+        java.io.InputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.NodeDef.ExperimentalDebugInfo parseFrom(
+        com.google.protobuf.CodedInputStream input)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input);
+    }
+    public static org.tensorflow.proto.framework.NodeDef.ExperimentalDebugInfo parseFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input, extensionRegistry);
+    }
+
+    @java.lang.Override
+    public Builder newBuilderForType() { return newBuilder(); }
+    public static Builder newBuilder() {
+      return DEFAULT_INSTANCE.toBuilder();
+    }
+    public static Builder newBuilder(org.tensorflow.proto.framework.NodeDef.ExperimentalDebugInfo prototype) {
+      return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+    }
+    @java.lang.Override
+    public Builder toBuilder() {
+      return this == DEFAULT_INSTANCE
+          ? new Builder() : new Builder().mergeFrom(this);
+    }
+
+    @java.lang.Override
+    protected Builder newBuilderForType(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      Builder builder = new Builder(parent);
+      return builder;
+    }
+    /**
+     * Protobuf type {@code tensorflow.NodeDef.ExperimentalDebugInfo}
+     */
+    public static final class Builder extends
+        com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+        // @@protoc_insertion_point(builder_implements:tensorflow.NodeDef.ExperimentalDebugInfo)
+        org.tensorflow.proto.framework.NodeDef.ExperimentalDebugInfoOrBuilder {
+      public static final com.google.protobuf.Descriptors.Descriptor
+          getDescriptor() {
+        return org.tensorflow.proto.framework.NodeProto.internal_static_tensorflow_NodeDef_ExperimentalDebugInfo_descriptor;
+      }
+
+      @java.lang.Override
+      protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+          internalGetFieldAccessorTable() {
+        return org.tensorflow.proto.framework.NodeProto.internal_static_tensorflow_NodeDef_ExperimentalDebugInfo_fieldAccessorTable
+            .ensureFieldAccessorsInitialized(
+                org.tensorflow.proto.framework.NodeDef.ExperimentalDebugInfo.class, org.tensorflow.proto.framework.NodeDef.ExperimentalDebugInfo.Builder.class);
+      }
+
+      // Construct using org.tensorflow.proto.framework.NodeDef.ExperimentalDebugInfo.newBuilder()
+      private Builder() {
+        maybeForceBuilderInitialization();
+      }
+
+      private Builder(
+          com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+        super(parent);
+        maybeForceBuilderInitialization();
+      }
+      private void maybeForceBuilderInitialization() {
+        if (com.google.protobuf.GeneratedMessageV3
+                .alwaysUseFieldBuilders) {
+        }
+      }
+      @java.lang.Override
+      public Builder clear() {
+        super.clear();
+        originalNodeNames_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+        bitField0_ = (bitField0_ & ~0x00000001);
+        originalFuncNames_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+        bitField0_ = (bitField0_ & ~0x00000002);
+        return this;
+      }
+
+      @java.lang.Override
+      public com.google.protobuf.Descriptors.Descriptor
+          getDescriptorForType() {
+        return org.tensorflow.proto.framework.NodeProto.internal_static_tensorflow_NodeDef_ExperimentalDebugInfo_descriptor;
+      }
+
+      @java.lang.Override
+      public org.tensorflow.proto.framework.NodeDef.ExperimentalDebugInfo getDefaultInstanceForType() {
+        return org.tensorflow.proto.framework.NodeDef.ExperimentalDebugInfo.getDefaultInstance();
+      }
+
+      @java.lang.Override
+      public org.tensorflow.proto.framework.NodeDef.ExperimentalDebugInfo build() {
+        org.tensorflow.proto.framework.NodeDef.ExperimentalDebugInfo result = buildPartial();
+        if (!result.isInitialized()) {
+          throw newUninitializedMessageException(result);
+        }
+        return result;
+      }
+
+      @java.lang.Override
+      public org.tensorflow.proto.framework.NodeDef.ExperimentalDebugInfo buildPartial() {
+        org.tensorflow.proto.framework.NodeDef.ExperimentalDebugInfo result = new org.tensorflow.proto.framework.NodeDef.ExperimentalDebugInfo(this);
+        int from_bitField0_ = bitField0_;
+        if (((bitField0_ & 0x00000001) != 0)) {
+          originalNodeNames_ = originalNodeNames_.getUnmodifiableView();
+          bitField0_ = (bitField0_ & ~0x00000001);
+        }
+        result.originalNodeNames_ = originalNodeNames_;
+        if (((bitField0_ & 0x00000002) != 0)) {
+          originalFuncNames_ = originalFuncNames_.getUnmodifiableView();
+          bitField0_ = (bitField0_ & ~0x00000002);
+        }
+        result.originalFuncNames_ = originalFuncNames_;
+        onBuilt();
+        return result;
+      }
+
+      @java.lang.Override
+      public Builder clone() {
+        return super.clone();
+      }
+      @java.lang.Override
+      public Builder setField(
+          com.google.protobuf.Descriptors.FieldDescriptor field,
+          java.lang.Object value) {
+        return super.setField(field, value);
+      }
+      @java.lang.Override
+      public Builder clearField(
+          com.google.protobuf.Descriptors.FieldDescriptor field) {
+        return super.clearField(field);
+      }
+      @java.lang.Override
+      public Builder clearOneof(
+          com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+        return super.clearOneof(oneof);
+      }
+      @java.lang.Override
+      public Builder setRepeatedField(
+          com.google.protobuf.Descriptors.FieldDescriptor field,
+          int index, java.lang.Object value) {
+        return super.setRepeatedField(field, index, value);
+      }
+      @java.lang.Override
+      public Builder addRepeatedField(
+          com.google.protobuf.Descriptors.FieldDescriptor field,
+          java.lang.Object value) {
+        return super.addRepeatedField(field, value);
+      }
+      @java.lang.Override
+      public Builder mergeFrom(com.google.protobuf.Message other) {
+        if (other instanceof org.tensorflow.proto.framework.NodeDef.ExperimentalDebugInfo) {
+          return mergeFrom((org.tensorflow.proto.framework.NodeDef.ExperimentalDebugInfo)other);
+        } else {
+          super.mergeFrom(other);
+          return this;
+        }
+      }
+
+      public Builder mergeFrom(org.tensorflow.proto.framework.NodeDef.ExperimentalDebugInfo other) {
+        if (other == org.tensorflow.proto.framework.NodeDef.ExperimentalDebugInfo.getDefaultInstance()) return this;
+        if (!other.originalNodeNames_.isEmpty()) {
+          if (originalNodeNames_.isEmpty()) {
+            originalNodeNames_ = other.originalNodeNames_;
+            bitField0_ = (bitField0_ & ~0x00000001);
+          } else {
+            ensureOriginalNodeNamesIsMutable();
+            originalNodeNames_.addAll(other.originalNodeNames_);
+          }
+          onChanged();
+        }
+        if (!other.originalFuncNames_.isEmpty()) {
+          if (originalFuncNames_.isEmpty()) {
+            originalFuncNames_ = other.originalFuncNames_;
+            bitField0_ = (bitField0_ & ~0x00000002);
+          } else {
+            ensureOriginalFuncNamesIsMutable();
+            originalFuncNames_.addAll(other.originalFuncNames_);
+          }
+          onChanged();
+        }
+        this.mergeUnknownFields(other.unknownFields);
+        onChanged();
+        return this;
+      }
+
+      @java.lang.Override
+      public final boolean isInitialized() {
+        return true;
+      }
+
+      @java.lang.Override
+      public Builder mergeFrom(
+          com.google.protobuf.CodedInputStream input,
+          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+          throws java.io.IOException {
+        org.tensorflow.proto.framework.NodeDef.ExperimentalDebugInfo parsedMessage = null;
+        try {
+          parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+        } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+          parsedMessage = (org.tensorflow.proto.framework.NodeDef.ExperimentalDebugInfo) e.getUnfinishedMessage();
+          throw e.unwrapIOException();
+        } finally {
+          if (parsedMessage != null) {
+            mergeFrom(parsedMessage);
+          }
+        }
+        return this;
+      }
+      private int bitField0_;
+
+      private com.google.protobuf.LazyStringList originalNodeNames_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+      private void ensureOriginalNodeNamesIsMutable() {
+        if (!((bitField0_ & 0x00000001) != 0)) {
+          originalNodeNames_ = new com.google.protobuf.LazyStringArrayList(originalNodeNames_);
+          bitField0_ |= 0x00000001;
+         }
+      }
+      /**
+       * <pre>
+       * Opaque string inserted into error messages created by the runtime.
+       * This is intended to store the list of names of the nodes from the
+       * original graph that this node was derived. For example if this node, say
+       * C, was result of a fusion of 2 nodes A and B, then 'original_node' would
+       * be {A, B}. This information can be used to map errors originating at the
+       * current node to some top level source code.
+       * </pre>
+       *
+       * <code>repeated string original_node_names = 1;</code>
+       */
+      public com.google.protobuf.ProtocolStringList
+          getOriginalNodeNamesList() {
+        return originalNodeNames_.getUnmodifiableView();
+      }
+      /**
+       * <pre>
+       * Opaque string inserted into error messages created by the runtime.
+       * This is intended to store the list of names of the nodes from the
+       * original graph that this node was derived. For example if this node, say
+       * C, was result of a fusion of 2 nodes A and B, then 'original_node' would
+       * be {A, B}. This information can be used to map errors originating at the
+       * current node to some top level source code.
+       * </pre>
+       *
+       * <code>repeated string original_node_names = 1;</code>
+       */
+      public int getOriginalNodeNamesCount() {
+        return originalNodeNames_.size();
+      }
+      /**
+       * <pre>
+       * Opaque string inserted into error messages created by the runtime.
+       * This is intended to store the list of names of the nodes from the
+       * original graph that this node was derived. For example if this node, say
+       * C, was result of a fusion of 2 nodes A and B, then 'original_node' would
+       * be {A, B}. This information can be used to map errors originating at the
+       * current node to some top level source code.
+       * </pre>
+       *
+       * <code>repeated string original_node_names = 1;</code>
+       */
+      public java.lang.String getOriginalNodeNames(int index) {
+        return originalNodeNames_.get(index);
+      }
+      /**
+       * <pre>
+       * Opaque string inserted into error messages created by the runtime.
+       * This is intended to store the list of names of the nodes from the
+       * original graph that this node was derived. For example if this node, say
+       * C, was result of a fusion of 2 nodes A and B, then 'original_node' would
+       * be {A, B}. This information can be used to map errors originating at the
+       * current node to some top level source code.
+       * </pre>
+       *
+       * <code>repeated string original_node_names = 1;</code>
+       */
+      public com.google.protobuf.ByteString
+          getOriginalNodeNamesBytes(int index) {
+        return originalNodeNames_.getByteString(index);
+      }
+      /**
+       * <pre>
+       * Opaque string inserted into error messages created by the runtime.
+       * This is intended to store the list of names of the nodes from the
+       * original graph that this node was derived. For example if this node, say
+       * C, was result of a fusion of 2 nodes A and B, then 'original_node' would
+       * be {A, B}. This information can be used to map errors originating at the
+       * current node to some top level source code.
+       * </pre>
+       *
+       * <code>repeated string original_node_names = 1;</code>
+       */
+      public Builder setOriginalNodeNames(
+          int index, java.lang.String value) {
+        if (value == null) {
+    throw new NullPointerException();
+  }
+  ensureOriginalNodeNamesIsMutable();
+        originalNodeNames_.set(index, value);
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * Opaque string inserted into error messages created by the runtime.
+       * This is intended to store the list of names of the nodes from the
+       * original graph that this node was derived. For example if this node, say
+       * C, was result of a fusion of 2 nodes A and B, then 'original_node' would
+       * be {A, B}. This information can be used to map errors originating at the
+       * current node to some top level source code.
+       * </pre>
+       *
+       * <code>repeated string original_node_names = 1;</code>
+       */
+      public Builder addOriginalNodeNames(
+          java.lang.String value) {
+        if (value == null) {
+    throw new NullPointerException();
+  }
+  ensureOriginalNodeNamesIsMutable();
+        originalNodeNames_.add(value);
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * Opaque string inserted into error messages created by the runtime.
+       * This is intended to store the list of names of the nodes from the
+       * original graph that this node was derived. For example if this node, say
+       * C, was result of a fusion of 2 nodes A and B, then 'original_node' would
+       * be {A, B}. This information can be used to map errors originating at the
+       * current node to some top level source code.
+       * </pre>
+       *
+       * <code>repeated string original_node_names = 1;</code>
+       */
+      public Builder addAllOriginalNodeNames(
+          java.lang.Iterable<java.lang.String> values) {
+        ensureOriginalNodeNamesIsMutable();
+        com.google.protobuf.AbstractMessageLite.Builder.addAll(
+            values, originalNodeNames_);
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * Opaque string inserted into error messages created by the runtime.
+       * This is intended to store the list of names of the nodes from the
+       * original graph that this node was derived. For example if this node, say
+       * C, was result of a fusion of 2 nodes A and B, then 'original_node' would
+       * be {A, B}. This information can be used to map errors originating at the
+       * current node to some top level source code.
+       * </pre>
+       *
+       * <code>repeated string original_node_names = 1;</code>
+       */
+      public Builder clearOriginalNodeNames() {
+        originalNodeNames_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+        bitField0_ = (bitField0_ & ~0x00000001);
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * Opaque string inserted into error messages created by the runtime.
+       * This is intended to store the list of names of the nodes from the
+       * original graph that this node was derived. For example if this node, say
+       * C, was result of a fusion of 2 nodes A and B, then 'original_node' would
+       * be {A, B}. This information can be used to map errors originating at the
+       * current node to some top level source code.
+       * </pre>
+       *
+       * <code>repeated string original_node_names = 1;</code>
+       */
+      public Builder addOriginalNodeNamesBytes(
+          com.google.protobuf.ByteString value) {
+        if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+        ensureOriginalNodeNamesIsMutable();
+        originalNodeNames_.add(value);
+        onChanged();
+        return this;
+      }
+
+      private com.google.protobuf.LazyStringList originalFuncNames_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+      private void ensureOriginalFuncNamesIsMutable() {
+        if (!((bitField0_ & 0x00000002) != 0)) {
+          originalFuncNames_ = new com.google.protobuf.LazyStringArrayList(originalFuncNames_);
+          bitField0_ |= 0x00000002;
+         }
+      }
+      /**
+       * <pre>
+       * This is intended to store the list of names of the functions from the
+       * original graph that this node was derived. For example if this node, say
+       * C, was result of a fusion of node A in function FA and node B in function
+       * FB, then `original_funcs` would be {FA, FB}. If the node is in the top
+       * level graph, the `original_func` is empty. This information, with the
+       * `original_node_names` can be used to map errors originating at the
+       * current ndoe to some top level source code.
+       * </pre>
+       *
+       * <code>repeated string original_func_names = 2;</code>
+       */
+      public com.google.protobuf.ProtocolStringList
+          getOriginalFuncNamesList() {
+        return originalFuncNames_.getUnmodifiableView();
+      }
+      /**
+       * <pre>
+       * This is intended to store the list of names of the functions from the
+       * original graph that this node was derived. For example if this node, say
+       * C, was result of a fusion of node A in function FA and node B in function
+       * FB, then `original_funcs` would be {FA, FB}. If the node is in the top
+       * level graph, the `original_func` is empty. This information, with the
+       * `original_node_names` can be used to map errors originating at the
+       * current ndoe to some top level source code.
+       * </pre>
+       *
+       * <code>repeated string original_func_names = 2;</code>
+       */
+      public int getOriginalFuncNamesCount() {
+        return originalFuncNames_.size();
+      }
+      /**
+       * <pre>
+       * This is intended to store the list of names of the functions from the
+       * original graph that this node was derived. For example if this node, say
+       * C, was result of a fusion of node A in function FA and node B in function
+       * FB, then `original_funcs` would be {FA, FB}. If the node is in the top
+       * level graph, the `original_func` is empty. This information, with the
+       * `original_node_names` can be used to map errors originating at the
+       * current ndoe to some top level source code.
+       * </pre>
+       *
+       * <code>repeated string original_func_names = 2;</code>
+       */
+      public java.lang.String getOriginalFuncNames(int index) {
+        return originalFuncNames_.get(index);
+      }
+      /**
+       * <pre>
+       * This is intended to store the list of names of the functions from the
+       * original graph that this node was derived. For example if this node, say
+       * C, was result of a fusion of node A in function FA and node B in function
+       * FB, then `original_funcs` would be {FA, FB}. If the node is in the top
+       * level graph, the `original_func` is empty. This information, with the
+       * `original_node_names` can be used to map errors originating at the
+       * current ndoe to some top level source code.
+       * </pre>
+       *
+       * <code>repeated string original_func_names = 2;</code>
+       */
+      public com.google.protobuf.ByteString
+          getOriginalFuncNamesBytes(int index) {
+        return originalFuncNames_.getByteString(index);
+      }
+      /**
+       * <pre>
+       * This is intended to store the list of names of the functions from the
+       * original graph that this node was derived. For example if this node, say
+       * C, was result of a fusion of node A in function FA and node B in function
+       * FB, then `original_funcs` would be {FA, FB}. If the node is in the top
+       * level graph, the `original_func` is empty. This information, with the
+       * `original_node_names` can be used to map errors originating at the
+       * current ndoe to some top level source code.
+       * </pre>
+       *
+       * <code>repeated string original_func_names = 2;</code>
+       */
+      public Builder setOriginalFuncNames(
+          int index, java.lang.String value) {
+        if (value == null) {
+    throw new NullPointerException();
+  }
+  ensureOriginalFuncNamesIsMutable();
+        originalFuncNames_.set(index, value);
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * This is intended to store the list of names of the functions from the
+       * original graph that this node was derived. For example if this node, say
+       * C, was result of a fusion of node A in function FA and node B in function
+       * FB, then `original_funcs` would be {FA, FB}. If the node is in the top
+       * level graph, the `original_func` is empty. This information, with the
+       * `original_node_names` can be used to map errors originating at the
+       * current ndoe to some top level source code.
+       * </pre>
+       *
+       * <code>repeated string original_func_names = 2;</code>
+       */
+      public Builder addOriginalFuncNames(
+          java.lang.String value) {
+        if (value == null) {
+    throw new NullPointerException();
+  }
+  ensureOriginalFuncNamesIsMutable();
+        originalFuncNames_.add(value);
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * This is intended to store the list of names of the functions from the
+       * original graph that this node was derived. For example if this node, say
+       * C, was result of a fusion of node A in function FA and node B in function
+       * FB, then `original_funcs` would be {FA, FB}. If the node is in the top
+       * level graph, the `original_func` is empty. This information, with the
+       * `original_node_names` can be used to map errors originating at the
+       * current ndoe to some top level source code.
+       * </pre>
+       *
+       * <code>repeated string original_func_names = 2;</code>
+       */
+      public Builder addAllOriginalFuncNames(
+          java.lang.Iterable<java.lang.String> values) {
+        ensureOriginalFuncNamesIsMutable();
+        com.google.protobuf.AbstractMessageLite.Builder.addAll(
+            values, originalFuncNames_);
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * This is intended to store the list of names of the functions from the
+       * original graph that this node was derived. For example if this node, say
+       * C, was result of a fusion of node A in function FA and node B in function
+       * FB, then `original_funcs` would be {FA, FB}. If the node is in the top
+       * level graph, the `original_func` is empty. This information, with the
+       * `original_node_names` can be used to map errors originating at the
+       * current ndoe to some top level source code.
+       * </pre>
+       *
+       * <code>repeated string original_func_names = 2;</code>
+       */
+      public Builder clearOriginalFuncNames() {
+        originalFuncNames_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+        bitField0_ = (bitField0_ & ~0x00000002);
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * This is intended to store the list of names of the functions from the
+       * original graph that this node was derived. For example if this node, say
+       * C, was result of a fusion of node A in function FA and node B in function
+       * FB, then `original_funcs` would be {FA, FB}. If the node is in the top
+       * level graph, the `original_func` is empty. This information, with the
+       * `original_node_names` can be used to map errors originating at the
+       * current ndoe to some top level source code.
+       * </pre>
+       *
+       * <code>repeated string original_func_names = 2;</code>
+       */
+      public Builder addOriginalFuncNamesBytes(
+          com.google.protobuf.ByteString value) {
+        if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+        ensureOriginalFuncNamesIsMutable();
+        originalFuncNames_.add(value);
+        onChanged();
+        return this;
+      }
+      @java.lang.Override
+      public final Builder setUnknownFields(
+          final com.google.protobuf.UnknownFieldSet unknownFields) {
+        return super.setUnknownFields(unknownFields);
+      }
+
+      @java.lang.Override
+      public final Builder mergeUnknownFields(
+          final com.google.protobuf.UnknownFieldSet unknownFields) {
+        return super.mergeUnknownFields(unknownFields);
+      }
+
+
+      // @@protoc_insertion_point(builder_scope:tensorflow.NodeDef.ExperimentalDebugInfo)
+    }
+
+    // @@protoc_insertion_point(class_scope:tensorflow.NodeDef.ExperimentalDebugInfo)
+    private static final org.tensorflow.proto.framework.NodeDef.ExperimentalDebugInfo DEFAULT_INSTANCE;
+    static {
+      DEFAULT_INSTANCE = new org.tensorflow.proto.framework.NodeDef.ExperimentalDebugInfo();
+    }
+
+    public static org.tensorflow.proto.framework.NodeDef.ExperimentalDebugInfo getDefaultInstance() {
+      return DEFAULT_INSTANCE;
+    }
+
+    private static final com.google.protobuf.Parser<ExperimentalDebugInfo>
+        PARSER = new com.google.protobuf.AbstractParser<ExperimentalDebugInfo>() {
+      @java.lang.Override
+      public ExperimentalDebugInfo parsePartialFrom(
+          com.google.protobuf.CodedInputStream input,
+          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+          throws com.google.protobuf.InvalidProtocolBufferException {
+        return new ExperimentalDebugInfo(input, extensionRegistry);
+      }
+    };
+
+    public static com.google.protobuf.Parser<ExperimentalDebugInfo> parser() {
+      return PARSER;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Parser<ExperimentalDebugInfo> getParserForType() {
+      return PARSER;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.NodeDef.ExperimentalDebugInfo getDefaultInstanceForType() {
+      return DEFAULT_INSTANCE;
+    }
+
+  }
+
+  public static final int NAME_FIELD_NUMBER = 1;
+  private volatile java.lang.Object name_;
+  /**
+   * <pre>
+   * The name given to this operator. Used for naming inputs,
+   * logging, visualization, etc.  Unique within a single GraphDef.
+   * Must match the regexp "[A-Za-z0-9.][A-Za-z0-9_&gt;./]*".
+   * </pre>
+   *
+   * <code>string name = 1;</code>
+   */
+  public java.lang.String getName() {
+    java.lang.Object ref = name_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      name_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * The name given to this operator. Used for naming inputs,
+   * logging, visualization, etc.  Unique within a single GraphDef.
+   * Must match the regexp "[A-Za-z0-9.][A-Za-z0-9_&gt;./]*".
+   * </pre>
+   *
+   * <code>string name = 1;</code>
+   */
+  public com.google.protobuf.ByteString
+      getNameBytes() {
+    java.lang.Object ref = name_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      name_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int OP_FIELD_NUMBER = 2;
+  private volatile java.lang.Object op_;
+  /**
+   * <pre>
+   * The operation name.  There may be custom parameters in attrs.
+   * Op names starting with an underscore are reserved for internal use.
+   * </pre>
+   *
+   * <code>string op = 2;</code>
+   */
+  public java.lang.String getOp() {
+    java.lang.Object ref = op_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      op_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * The operation name.  There may be custom parameters in attrs.
+   * Op names starting with an underscore are reserved for internal use.
+   * </pre>
+   *
+   * <code>string op = 2;</code>
+   */
+  public com.google.protobuf.ByteString
+      getOpBytes() {
+    java.lang.Object ref = op_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      op_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int INPUT_FIELD_NUMBER = 3;
+  private com.google.protobuf.LazyStringList input_;
+  /**
+   * <pre>
+   * Each input is "node:src_output" with "node" being a string name and
+   * "src_output" indicating which output tensor to use from "node". If
+   * "src_output" is 0 the ":0" suffix can be omitted.  Regular inputs
+   * may optionally be followed by control inputs that have the format
+   * "^node".
+   * </pre>
+   *
+   * <code>repeated string input = 3;</code>
+   */
+  public com.google.protobuf.ProtocolStringList
+      getInputList() {
+    return input_;
+  }
+  /**
+   * <pre>
+   * Each input is "node:src_output" with "node" being a string name and
+   * "src_output" indicating which output tensor to use from "node". If
+   * "src_output" is 0 the ":0" suffix can be omitted.  Regular inputs
+   * may optionally be followed by control inputs that have the format
+   * "^node".
+   * </pre>
+   *
+   * <code>repeated string input = 3;</code>
+   */
+  public int getInputCount() {
+    return input_.size();
+  }
+  /**
+   * <pre>
+   * Each input is "node:src_output" with "node" being a string name and
+   * "src_output" indicating which output tensor to use from "node". If
+   * "src_output" is 0 the ":0" suffix can be omitted.  Regular inputs
+   * may optionally be followed by control inputs that have the format
+   * "^node".
+   * </pre>
+   *
+   * <code>repeated string input = 3;</code>
+   */
+  public java.lang.String getInput(int index) {
+    return input_.get(index);
+  }
+  /**
+   * <pre>
+   * Each input is "node:src_output" with "node" being a string name and
+   * "src_output" indicating which output tensor to use from "node". If
+   * "src_output" is 0 the ":0" suffix can be omitted.  Regular inputs
+   * may optionally be followed by control inputs that have the format
+   * "^node".
+   * </pre>
+   *
+   * <code>repeated string input = 3;</code>
+   */
+  public com.google.protobuf.ByteString
+      getInputBytes(int index) {
+    return input_.getByteString(index);
+  }
+
+  public static final int DEVICE_FIELD_NUMBER = 4;
+  private volatile java.lang.Object device_;
+  /**
+   * <pre>
+   * A (possibly partial) specification for the device on which this
+   * node should be placed.
+   * The expected syntax for this string is as follows:
+   * DEVICE_SPEC ::= PARTIAL_SPEC
+   * PARTIAL_SPEC ::= ("/" CONSTRAINT) *
+   * CONSTRAINT ::= ("job:" JOB_NAME)
+   *              | ("replica:" [1-9][0-9]*)
+   *              | ("task:" [1-9][0-9]*)
+   *              | ("device:" [A-Za-z]* ":" ([1-9][0-9]* | "*") )
+   * Valid values for this string include:
+   * * "/job:worker/replica:0/task:1/device:GPU:3"  (full specification)
+   * * "/job:worker/device:GPU:3"                   (partial specification)
+   * * ""                                    (no specification)
+   * If the constraints do not resolve to a single device (or if this
+   * field is empty or not present), the runtime will attempt to
+   * choose a device automatically.
+   * </pre>
+   *
+   * <code>string device = 4;</code>
+   */
+  public java.lang.String getDevice() {
+    java.lang.Object ref = device_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      device_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * A (possibly partial) specification for the device on which this
+   * node should be placed.
+   * The expected syntax for this string is as follows:
+   * DEVICE_SPEC ::= PARTIAL_SPEC
+   * PARTIAL_SPEC ::= ("/" CONSTRAINT) *
+   * CONSTRAINT ::= ("job:" JOB_NAME)
+   *              | ("replica:" [1-9][0-9]*)
+   *              | ("task:" [1-9][0-9]*)
+   *              | ("device:" [A-Za-z]* ":" ([1-9][0-9]* | "*") )
+   * Valid values for this string include:
+   * * "/job:worker/replica:0/task:1/device:GPU:3"  (full specification)
+   * * "/job:worker/device:GPU:3"                   (partial specification)
+   * * ""                                    (no specification)
+   * If the constraints do not resolve to a single device (or if this
+   * field is empty or not present), the runtime will attempt to
+   * choose a device automatically.
+   * </pre>
+   *
+   * <code>string device = 4;</code>
+   */
+  public com.google.protobuf.ByteString
+      getDeviceBytes() {
+    java.lang.Object ref = device_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      device_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int ATTR_FIELD_NUMBER = 5;
+  private static final class AttrDefaultEntryHolder {
+    static final com.google.protobuf.MapEntry<
+        java.lang.String, org.tensorflow.proto.framework.AttrValue> defaultEntry =
+            com.google.protobuf.MapEntry
+            .<java.lang.String, org.tensorflow.proto.framework.AttrValue>newDefaultInstance(
+                org.tensorflow.proto.framework.NodeProto.internal_static_tensorflow_NodeDef_AttrEntry_descriptor, 
+                com.google.protobuf.WireFormat.FieldType.STRING,
+                "",
+                com.google.protobuf.WireFormat.FieldType.MESSAGE,
+                org.tensorflow.proto.framework.AttrValue.getDefaultInstance());
+  }
+  private com.google.protobuf.MapField<
+      java.lang.String, org.tensorflow.proto.framework.AttrValue> attr_;
+  private com.google.protobuf.MapField<java.lang.String, org.tensorflow.proto.framework.AttrValue>
+  internalGetAttr() {
+    if (attr_ == null) {
+      return com.google.protobuf.MapField.emptyMapField(
+          AttrDefaultEntryHolder.defaultEntry);
+    }
+    return attr_;
+  }
+
+  public int getAttrCount() {
+    return internalGetAttr().getMap().size();
+  }
+  /**
+   * <pre>
+   * Operation-specific graph-construction-time configuration.
+   * Note that this should include all attrs defined in the
+   * corresponding OpDef, including those with a value matching
+   * the default -- this allows the default to change and makes
+   * NodeDefs easier to interpret on their own.  However, if
+   * an attr with a default is not specified in this list, the
+   * default will be used.
+   * The "names" (keys) must match the regexp "[a-z][a-z0-9_]+" (and
+   * one of the names from the corresponding OpDef's attr field).
+   * The values must have a type matching the corresponding OpDef
+   * attr's type field.
+   * TODO(josh11b): Add some examples here showing best practices.
+   * </pre>
+   *
+   * <code>map&lt;string, .tensorflow.AttrValue&gt; attr = 5;</code>
+   */
+
+  public boolean containsAttr(
+      java.lang.String key) {
+    if (key == null) { throw new java.lang.NullPointerException(); }
+    return internalGetAttr().getMap().containsKey(key);
+  }
+  /**
+   * Use {@link #getAttrMap()} instead.
+   */
+  @java.lang.Deprecated
+  public java.util.Map<java.lang.String, org.tensorflow.proto.framework.AttrValue> getAttr() {
+    return getAttrMap();
+  }
+  /**
+   * <pre>
+   * Operation-specific graph-construction-time configuration.
+   * Note that this should include all attrs defined in the
+   * corresponding OpDef, including those with a value matching
+   * the default -- this allows the default to change and makes
+   * NodeDefs easier to interpret on their own.  However, if
+   * an attr with a default is not specified in this list, the
+   * default will be used.
+   * The "names" (keys) must match the regexp "[a-z][a-z0-9_]+" (and
+   * one of the names from the corresponding OpDef's attr field).
+   * The values must have a type matching the corresponding OpDef
+   * attr's type field.
+   * TODO(josh11b): Add some examples here showing best practices.
+   * </pre>
+   *
+   * <code>map&lt;string, .tensorflow.AttrValue&gt; attr = 5;</code>
+   */
+
+  public java.util.Map<java.lang.String, org.tensorflow.proto.framework.AttrValue> getAttrMap() {
+    return internalGetAttr().getMap();
+  }
+  /**
+   * <pre>
+   * Operation-specific graph-construction-time configuration.
+   * Note that this should include all attrs defined in the
+   * corresponding OpDef, including those with a value matching
+   * the default -- this allows the default to change and makes
+   * NodeDefs easier to interpret on their own.  However, if
+   * an attr with a default is not specified in this list, the
+   * default will be used.
+   * The "names" (keys) must match the regexp "[a-z][a-z0-9_]+" (and
+   * one of the names from the corresponding OpDef's attr field).
+   * The values must have a type matching the corresponding OpDef
+   * attr's type field.
+   * TODO(josh11b): Add some examples here showing best practices.
+   * </pre>
+   *
+   * <code>map&lt;string, .tensorflow.AttrValue&gt; attr = 5;</code>
+   */
+
+  public org.tensorflow.proto.framework.AttrValue getAttrOrDefault(
+      java.lang.String key,
+      org.tensorflow.proto.framework.AttrValue defaultValue) {
+    if (key == null) { throw new java.lang.NullPointerException(); }
+    java.util.Map<java.lang.String, org.tensorflow.proto.framework.AttrValue> map =
+        internalGetAttr().getMap();
+    return map.containsKey(key) ? map.get(key) : defaultValue;
+  }
+  /**
+   * <pre>
+   * Operation-specific graph-construction-time configuration.
+   * Note that this should include all attrs defined in the
+   * corresponding OpDef, including those with a value matching
+   * the default -- this allows the default to change and makes
+   * NodeDefs easier to interpret on their own.  However, if
+   * an attr with a default is not specified in this list, the
+   * default will be used.
+   * The "names" (keys) must match the regexp "[a-z][a-z0-9_]+" (and
+   * one of the names from the corresponding OpDef's attr field).
+   * The values must have a type matching the corresponding OpDef
+   * attr's type field.
+   * TODO(josh11b): Add some examples here showing best practices.
+   * </pre>
+   *
+   * <code>map&lt;string, .tensorflow.AttrValue&gt; attr = 5;</code>
+   */
+
+  public org.tensorflow.proto.framework.AttrValue getAttrOrThrow(
+      java.lang.String key) {
+    if (key == null) { throw new java.lang.NullPointerException(); }
+    java.util.Map<java.lang.String, org.tensorflow.proto.framework.AttrValue> map =
+        internalGetAttr().getMap();
+    if (!map.containsKey(key)) {
+      throw new java.lang.IllegalArgumentException();
+    }
+    return map.get(key);
+  }
+
+  public static final int EXPERIMENTAL_DEBUG_INFO_FIELD_NUMBER = 6;
+  private org.tensorflow.proto.framework.NodeDef.ExperimentalDebugInfo experimentalDebugInfo_;
+  /**
+   * <pre>
+   * This stores debug information associated with the node.
+   * </pre>
+   *
+   * <code>.tensorflow.NodeDef.ExperimentalDebugInfo experimental_debug_info = 6;</code>
+   */
+  public boolean hasExperimentalDebugInfo() {
+    return experimentalDebugInfo_ != null;
+  }
+  /**
+   * <pre>
+   * This stores debug information associated with the node.
+   * </pre>
+   *
+   * <code>.tensorflow.NodeDef.ExperimentalDebugInfo experimental_debug_info = 6;</code>
+   */
+  public org.tensorflow.proto.framework.NodeDef.ExperimentalDebugInfo getExperimentalDebugInfo() {
+    return experimentalDebugInfo_ == null ? org.tensorflow.proto.framework.NodeDef.ExperimentalDebugInfo.getDefaultInstance() : experimentalDebugInfo_;
+  }
+  /**
+   * <pre>
+   * This stores debug information associated with the node.
+   * </pre>
+   *
+   * <code>.tensorflow.NodeDef.ExperimentalDebugInfo experimental_debug_info = 6;</code>
+   */
+  public org.tensorflow.proto.framework.NodeDef.ExperimentalDebugInfoOrBuilder getExperimentalDebugInfoOrBuilder() {
+    return getExperimentalDebugInfo();
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    if (!getNameBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 1, name_);
+    }
+    if (!getOpBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 2, op_);
+    }
+    for (int i = 0; i < input_.size(); i++) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 3, input_.getRaw(i));
+    }
+    if (!getDeviceBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 4, device_);
+    }
+    com.google.protobuf.GeneratedMessageV3
+      .serializeStringMapTo(
+        output,
+        internalGetAttr(),
+        AttrDefaultEntryHolder.defaultEntry,
+        5);
+    if (experimentalDebugInfo_ != null) {
+      output.writeMessage(6, getExperimentalDebugInfo());
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (!getNameBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, name_);
+    }
+    if (!getOpBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(2, op_);
+    }
+    {
+      int dataSize = 0;
+      for (int i = 0; i < input_.size(); i++) {
+        dataSize += computeStringSizeNoTag(input_.getRaw(i));
+      }
+      size += dataSize;
+      size += 1 * getInputList().size();
+    }
+    if (!getDeviceBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(4, device_);
+    }
+    for (java.util.Map.Entry<java.lang.String, org.tensorflow.proto.framework.AttrValue> entry
+         : internalGetAttr().getMap().entrySet()) {
+      com.google.protobuf.MapEntry<java.lang.String, org.tensorflow.proto.framework.AttrValue>
+      attr__ = AttrDefaultEntryHolder.defaultEntry.newBuilderForType()
+          .setKey(entry.getKey())
+          .setValue(entry.getValue())
+          .build();
+      size += com.google.protobuf.CodedOutputStream
+          .computeMessageSize(5, attr__);
+    }
+    if (experimentalDebugInfo_ != null) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(6, getExperimentalDebugInfo());
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.framework.NodeDef)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.framework.NodeDef other = (org.tensorflow.proto.framework.NodeDef) obj;
+
+    if (!getName()
+        .equals(other.getName())) return false;
+    if (!getOp()
+        .equals(other.getOp())) return false;
+    if (!getInputList()
+        .equals(other.getInputList())) return false;
+    if (!getDevice()
+        .equals(other.getDevice())) return false;
+    if (!internalGetAttr().equals(
+        other.internalGetAttr())) return false;
+    if (hasExperimentalDebugInfo() != other.hasExperimentalDebugInfo()) return false;
+    if (hasExperimentalDebugInfo()) {
+      if (!getExperimentalDebugInfo()
+          .equals(other.getExperimentalDebugInfo())) return false;
+    }
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    hash = (37 * hash) + NAME_FIELD_NUMBER;
+    hash = (53 * hash) + getName().hashCode();
+    hash = (37 * hash) + OP_FIELD_NUMBER;
+    hash = (53 * hash) + getOp().hashCode();
+    if (getInputCount() > 0) {
+      hash = (37 * hash) + INPUT_FIELD_NUMBER;
+      hash = (53 * hash) + getInputList().hashCode();
+    }
+    hash = (37 * hash) + DEVICE_FIELD_NUMBER;
+    hash = (53 * hash) + getDevice().hashCode();
+    if (!internalGetAttr().getMap().isEmpty()) {
+      hash = (37 * hash) + ATTR_FIELD_NUMBER;
+      hash = (53 * hash) + internalGetAttr().hashCode();
+    }
+    if (hasExperimentalDebugInfo()) {
+      hash = (37 * hash) + EXPERIMENTAL_DEBUG_INFO_FIELD_NUMBER;
+      hash = (53 * hash) + getExperimentalDebugInfo().hashCode();
+    }
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.framework.NodeDef parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.NodeDef parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.NodeDef parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.NodeDef parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.NodeDef parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.NodeDef parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.NodeDef parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.NodeDef parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.NodeDef parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.NodeDef parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.NodeDef parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.NodeDef parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.framework.NodeDef prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * Protobuf type {@code tensorflow.NodeDef}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.NodeDef)
+      org.tensorflow.proto.framework.NodeDefOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.NodeProto.internal_static_tensorflow_NodeDef_descriptor;
+    }
+
+    @SuppressWarnings({"rawtypes"})
+    protected com.google.protobuf.MapField internalGetMapField(
+        int number) {
+      switch (number) {
+        case 5:
+          return internalGetAttr();
+        default:
+          throw new RuntimeException(
+              "Invalid map field number: " + number);
+      }
+    }
+    @SuppressWarnings({"rawtypes"})
+    protected com.google.protobuf.MapField internalGetMutableMapField(
+        int number) {
+      switch (number) {
+        case 5:
+          return internalGetMutableAttr();
+        default:
+          throw new RuntimeException(
+              "Invalid map field number: " + number);
+      }
+    }
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.NodeProto.internal_static_tensorflow_NodeDef_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.NodeDef.class, org.tensorflow.proto.framework.NodeDef.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.framework.NodeDef.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      name_ = "";
+
+      op_ = "";
+
+      input_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+      bitField0_ = (bitField0_ & ~0x00000001);
+      device_ = "";
+
+      internalGetMutableAttr().clear();
+      if (experimentalDebugInfoBuilder_ == null) {
+        experimentalDebugInfo_ = null;
+      } else {
+        experimentalDebugInfo_ = null;
+        experimentalDebugInfoBuilder_ = null;
+      }
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.framework.NodeProto.internal_static_tensorflow_NodeDef_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.NodeDef getDefaultInstanceForType() {
+      return org.tensorflow.proto.framework.NodeDef.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.NodeDef build() {
+      org.tensorflow.proto.framework.NodeDef result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.NodeDef buildPartial() {
+      org.tensorflow.proto.framework.NodeDef result = new org.tensorflow.proto.framework.NodeDef(this);
+      int from_bitField0_ = bitField0_;
+      result.name_ = name_;
+      result.op_ = op_;
+      if (((bitField0_ & 0x00000001) != 0)) {
+        input_ = input_.getUnmodifiableView();
+        bitField0_ = (bitField0_ & ~0x00000001);
+      }
+      result.input_ = input_;
+      result.device_ = device_;
+      result.attr_ = internalGetAttr();
+      result.attr_.makeImmutable();
+      if (experimentalDebugInfoBuilder_ == null) {
+        result.experimentalDebugInfo_ = experimentalDebugInfo_;
+      } else {
+        result.experimentalDebugInfo_ = experimentalDebugInfoBuilder_.build();
+      }
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.framework.NodeDef) {
+        return mergeFrom((org.tensorflow.proto.framework.NodeDef)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.framework.NodeDef other) {
+      if (other == org.tensorflow.proto.framework.NodeDef.getDefaultInstance()) return this;
+      if (!other.getName().isEmpty()) {
+        name_ = other.name_;
+        onChanged();
+      }
+      if (!other.getOp().isEmpty()) {
+        op_ = other.op_;
+        onChanged();
+      }
+      if (!other.input_.isEmpty()) {
+        if (input_.isEmpty()) {
+          input_ = other.input_;
+          bitField0_ = (bitField0_ & ~0x00000001);
+        } else {
+          ensureInputIsMutable();
+          input_.addAll(other.input_);
+        }
+        onChanged();
+      }
+      if (!other.getDevice().isEmpty()) {
+        device_ = other.device_;
+        onChanged();
+      }
+      internalGetMutableAttr().mergeFrom(
+          other.internalGetAttr());
+      if (other.hasExperimentalDebugInfo()) {
+        mergeExperimentalDebugInfo(other.getExperimentalDebugInfo());
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.framework.NodeDef parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.framework.NodeDef) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+    private int bitField0_;
+
+    private java.lang.Object name_ = "";
+    /**
+     * <pre>
+     * The name given to this operator. Used for naming inputs,
+     * logging, visualization, etc.  Unique within a single GraphDef.
+     * Must match the regexp "[A-Za-z0-9.][A-Za-z0-9_&gt;./]*".
+     * </pre>
+     *
+     * <code>string name = 1;</code>
+     */
+    public java.lang.String getName() {
+      java.lang.Object ref = name_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        name_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * The name given to this operator. Used for naming inputs,
+     * logging, visualization, etc.  Unique within a single GraphDef.
+     * Must match the regexp "[A-Za-z0-9.][A-Za-z0-9_&gt;./]*".
+     * </pre>
+     *
+     * <code>string name = 1;</code>
+     */
+    public com.google.protobuf.ByteString
+        getNameBytes() {
+      java.lang.Object ref = name_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        name_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * The name given to this operator. Used for naming inputs,
+     * logging, visualization, etc.  Unique within a single GraphDef.
+     * Must match the regexp "[A-Za-z0-9.][A-Za-z0-9_&gt;./]*".
+     * </pre>
+     *
+     * <code>string name = 1;</code>
+     */
+    public Builder setName(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      name_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * The name given to this operator. Used for naming inputs,
+     * logging, visualization, etc.  Unique within a single GraphDef.
+     * Must match the regexp "[A-Za-z0-9.][A-Za-z0-9_&gt;./]*".
+     * </pre>
+     *
+     * <code>string name = 1;</code>
+     */
+    public Builder clearName() {
+      
+      name_ = getDefaultInstance().getName();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * The name given to this operator. Used for naming inputs,
+     * logging, visualization, etc.  Unique within a single GraphDef.
+     * Must match the regexp "[A-Za-z0-9.][A-Za-z0-9_&gt;./]*".
+     * </pre>
+     *
+     * <code>string name = 1;</code>
+     */
+    public Builder setNameBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      name_ = value;
+      onChanged();
+      return this;
+    }
+
+    private java.lang.Object op_ = "";
+    /**
+     * <pre>
+     * The operation name.  There may be custom parameters in attrs.
+     * Op names starting with an underscore are reserved for internal use.
+     * </pre>
+     *
+     * <code>string op = 2;</code>
+     */
+    public java.lang.String getOp() {
+      java.lang.Object ref = op_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        op_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * The operation name.  There may be custom parameters in attrs.
+     * Op names starting with an underscore are reserved for internal use.
+     * </pre>
+     *
+     * <code>string op = 2;</code>
+     */
+    public com.google.protobuf.ByteString
+        getOpBytes() {
+      java.lang.Object ref = op_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        op_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * The operation name.  There may be custom parameters in attrs.
+     * Op names starting with an underscore are reserved for internal use.
+     * </pre>
+     *
+     * <code>string op = 2;</code>
+     */
+    public Builder setOp(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      op_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * The operation name.  There may be custom parameters in attrs.
+     * Op names starting with an underscore are reserved for internal use.
+     * </pre>
+     *
+     * <code>string op = 2;</code>
+     */
+    public Builder clearOp() {
+      
+      op_ = getDefaultInstance().getOp();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * The operation name.  There may be custom parameters in attrs.
+     * Op names starting with an underscore are reserved for internal use.
+     * </pre>
+     *
+     * <code>string op = 2;</code>
+     */
+    public Builder setOpBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      op_ = value;
+      onChanged();
+      return this;
+    }
+
+    private com.google.protobuf.LazyStringList input_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+    private void ensureInputIsMutable() {
+      if (!((bitField0_ & 0x00000001) != 0)) {
+        input_ = new com.google.protobuf.LazyStringArrayList(input_);
+        bitField0_ |= 0x00000001;
+       }
+    }
+    /**
+     * <pre>
+     * Each input is "node:src_output" with "node" being a string name and
+     * "src_output" indicating which output tensor to use from "node". If
+     * "src_output" is 0 the ":0" suffix can be omitted.  Regular inputs
+     * may optionally be followed by control inputs that have the format
+     * "^node".
+     * </pre>
+     *
+     * <code>repeated string input = 3;</code>
+     */
+    public com.google.protobuf.ProtocolStringList
+        getInputList() {
+      return input_.getUnmodifiableView();
+    }
+    /**
+     * <pre>
+     * Each input is "node:src_output" with "node" being a string name and
+     * "src_output" indicating which output tensor to use from "node". If
+     * "src_output" is 0 the ":0" suffix can be omitted.  Regular inputs
+     * may optionally be followed by control inputs that have the format
+     * "^node".
+     * </pre>
+     *
+     * <code>repeated string input = 3;</code>
+     */
+    public int getInputCount() {
+      return input_.size();
+    }
+    /**
+     * <pre>
+     * Each input is "node:src_output" with "node" being a string name and
+     * "src_output" indicating which output tensor to use from "node". If
+     * "src_output" is 0 the ":0" suffix can be omitted.  Regular inputs
+     * may optionally be followed by control inputs that have the format
+     * "^node".
+     * </pre>
+     *
+     * <code>repeated string input = 3;</code>
+     */
+    public java.lang.String getInput(int index) {
+      return input_.get(index);
+    }
+    /**
+     * <pre>
+     * Each input is "node:src_output" with "node" being a string name and
+     * "src_output" indicating which output tensor to use from "node". If
+     * "src_output" is 0 the ":0" suffix can be omitted.  Regular inputs
+     * may optionally be followed by control inputs that have the format
+     * "^node".
+     * </pre>
+     *
+     * <code>repeated string input = 3;</code>
+     */
+    public com.google.protobuf.ByteString
+        getInputBytes(int index) {
+      return input_.getByteString(index);
+    }
+    /**
+     * <pre>
+     * Each input is "node:src_output" with "node" being a string name and
+     * "src_output" indicating which output tensor to use from "node". If
+     * "src_output" is 0 the ":0" suffix can be omitted.  Regular inputs
+     * may optionally be followed by control inputs that have the format
+     * "^node".
+     * </pre>
+     *
+     * <code>repeated string input = 3;</code>
+     */
+    public Builder setInput(
+        int index, java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  ensureInputIsMutable();
+      input_.set(index, value);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Each input is "node:src_output" with "node" being a string name and
+     * "src_output" indicating which output tensor to use from "node". If
+     * "src_output" is 0 the ":0" suffix can be omitted.  Regular inputs
+     * may optionally be followed by control inputs that have the format
+     * "^node".
+     * </pre>
+     *
+     * <code>repeated string input = 3;</code>
+     */
+    public Builder addInput(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  ensureInputIsMutable();
+      input_.add(value);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Each input is "node:src_output" with "node" being a string name and
+     * "src_output" indicating which output tensor to use from "node". If
+     * "src_output" is 0 the ":0" suffix can be omitted.  Regular inputs
+     * may optionally be followed by control inputs that have the format
+     * "^node".
+     * </pre>
+     *
+     * <code>repeated string input = 3;</code>
+     */
+    public Builder addAllInput(
+        java.lang.Iterable<java.lang.String> values) {
+      ensureInputIsMutable();
+      com.google.protobuf.AbstractMessageLite.Builder.addAll(
+          values, input_);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Each input is "node:src_output" with "node" being a string name and
+     * "src_output" indicating which output tensor to use from "node". If
+     * "src_output" is 0 the ":0" suffix can be omitted.  Regular inputs
+     * may optionally be followed by control inputs that have the format
+     * "^node".
+     * </pre>
+     *
+     * <code>repeated string input = 3;</code>
+     */
+    public Builder clearInput() {
+      input_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+      bitField0_ = (bitField0_ & ~0x00000001);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Each input is "node:src_output" with "node" being a string name and
+     * "src_output" indicating which output tensor to use from "node". If
+     * "src_output" is 0 the ":0" suffix can be omitted.  Regular inputs
+     * may optionally be followed by control inputs that have the format
+     * "^node".
+     * </pre>
+     *
+     * <code>repeated string input = 3;</code>
+     */
+    public Builder addInputBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      ensureInputIsMutable();
+      input_.add(value);
+      onChanged();
+      return this;
+    }
+
+    private java.lang.Object device_ = "";
+    /**
+     * <pre>
+     * A (possibly partial) specification for the device on which this
+     * node should be placed.
+     * The expected syntax for this string is as follows:
+     * DEVICE_SPEC ::= PARTIAL_SPEC
+     * PARTIAL_SPEC ::= ("/" CONSTRAINT) *
+     * CONSTRAINT ::= ("job:" JOB_NAME)
+     *              | ("replica:" [1-9][0-9]*)
+     *              | ("task:" [1-9][0-9]*)
+     *              | ("device:" [A-Za-z]* ":" ([1-9][0-9]* | "*") )
+     * Valid values for this string include:
+     * * "/job:worker/replica:0/task:1/device:GPU:3"  (full specification)
+     * * "/job:worker/device:GPU:3"                   (partial specification)
+     * * ""                                    (no specification)
+     * If the constraints do not resolve to a single device (or if this
+     * field is empty or not present), the runtime will attempt to
+     * choose a device automatically.
+     * </pre>
+     *
+     * <code>string device = 4;</code>
+     */
+    public java.lang.String getDevice() {
+      java.lang.Object ref = device_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        device_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * A (possibly partial) specification for the device on which this
+     * node should be placed.
+     * The expected syntax for this string is as follows:
+     * DEVICE_SPEC ::= PARTIAL_SPEC
+     * PARTIAL_SPEC ::= ("/" CONSTRAINT) *
+     * CONSTRAINT ::= ("job:" JOB_NAME)
+     *              | ("replica:" [1-9][0-9]*)
+     *              | ("task:" [1-9][0-9]*)
+     *              | ("device:" [A-Za-z]* ":" ([1-9][0-9]* | "*") )
+     * Valid values for this string include:
+     * * "/job:worker/replica:0/task:1/device:GPU:3"  (full specification)
+     * * "/job:worker/device:GPU:3"                   (partial specification)
+     * * ""                                    (no specification)
+     * If the constraints do not resolve to a single device (or if this
+     * field is empty or not present), the runtime will attempt to
+     * choose a device automatically.
+     * </pre>
+     *
+     * <code>string device = 4;</code>
+     */
+    public com.google.protobuf.ByteString
+        getDeviceBytes() {
+      java.lang.Object ref = device_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        device_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * A (possibly partial) specification for the device on which this
+     * node should be placed.
+     * The expected syntax for this string is as follows:
+     * DEVICE_SPEC ::= PARTIAL_SPEC
+     * PARTIAL_SPEC ::= ("/" CONSTRAINT) *
+     * CONSTRAINT ::= ("job:" JOB_NAME)
+     *              | ("replica:" [1-9][0-9]*)
+     *              | ("task:" [1-9][0-9]*)
+     *              | ("device:" [A-Za-z]* ":" ([1-9][0-9]* | "*") )
+     * Valid values for this string include:
+     * * "/job:worker/replica:0/task:1/device:GPU:3"  (full specification)
+     * * "/job:worker/device:GPU:3"                   (partial specification)
+     * * ""                                    (no specification)
+     * If the constraints do not resolve to a single device (or if this
+     * field is empty or not present), the runtime will attempt to
+     * choose a device automatically.
+     * </pre>
+     *
+     * <code>string device = 4;</code>
+     */
+    public Builder setDevice(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      device_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * A (possibly partial) specification for the device on which this
+     * node should be placed.
+     * The expected syntax for this string is as follows:
+     * DEVICE_SPEC ::= PARTIAL_SPEC
+     * PARTIAL_SPEC ::= ("/" CONSTRAINT) *
+     * CONSTRAINT ::= ("job:" JOB_NAME)
+     *              | ("replica:" [1-9][0-9]*)
+     *              | ("task:" [1-9][0-9]*)
+     *              | ("device:" [A-Za-z]* ":" ([1-9][0-9]* | "*") )
+     * Valid values for this string include:
+     * * "/job:worker/replica:0/task:1/device:GPU:3"  (full specification)
+     * * "/job:worker/device:GPU:3"                   (partial specification)
+     * * ""                                    (no specification)
+     * If the constraints do not resolve to a single device (or if this
+     * field is empty or not present), the runtime will attempt to
+     * choose a device automatically.
+     * </pre>
+     *
+     * <code>string device = 4;</code>
+     */
+    public Builder clearDevice() {
+      
+      device_ = getDefaultInstance().getDevice();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * A (possibly partial) specification for the device on which this
+     * node should be placed.
+     * The expected syntax for this string is as follows:
+     * DEVICE_SPEC ::= PARTIAL_SPEC
+     * PARTIAL_SPEC ::= ("/" CONSTRAINT) *
+     * CONSTRAINT ::= ("job:" JOB_NAME)
+     *              | ("replica:" [1-9][0-9]*)
+     *              | ("task:" [1-9][0-9]*)
+     *              | ("device:" [A-Za-z]* ":" ([1-9][0-9]* | "*") )
+     * Valid values for this string include:
+     * * "/job:worker/replica:0/task:1/device:GPU:3"  (full specification)
+     * * "/job:worker/device:GPU:3"                   (partial specification)
+     * * ""                                    (no specification)
+     * If the constraints do not resolve to a single device (or if this
+     * field is empty or not present), the runtime will attempt to
+     * choose a device automatically.
+     * </pre>
+     *
+     * <code>string device = 4;</code>
+     */
+    public Builder setDeviceBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      device_ = value;
+      onChanged();
+      return this;
+    }
+
+    private com.google.protobuf.MapField<
+        java.lang.String, org.tensorflow.proto.framework.AttrValue> attr_;
+    private com.google.protobuf.MapField<java.lang.String, org.tensorflow.proto.framework.AttrValue>
+    internalGetAttr() {
+      if (attr_ == null) {
+        return com.google.protobuf.MapField.emptyMapField(
+            AttrDefaultEntryHolder.defaultEntry);
+      }
+      return attr_;
+    }
+    private com.google.protobuf.MapField<java.lang.String, org.tensorflow.proto.framework.AttrValue>
+    internalGetMutableAttr() {
+      onChanged();;
+      if (attr_ == null) {
+        attr_ = com.google.protobuf.MapField.newMapField(
+            AttrDefaultEntryHolder.defaultEntry);
+      }
+      if (!attr_.isMutable()) {
+        attr_ = attr_.copy();
+      }
+      return attr_;
+    }
+
+    public int getAttrCount() {
+      return internalGetAttr().getMap().size();
+    }
+    /**
+     * <pre>
+     * Operation-specific graph-construction-time configuration.
+     * Note that this should include all attrs defined in the
+     * corresponding OpDef, including those with a value matching
+     * the default -- this allows the default to change and makes
+     * NodeDefs easier to interpret on their own.  However, if
+     * an attr with a default is not specified in this list, the
+     * default will be used.
+     * The "names" (keys) must match the regexp "[a-z][a-z0-9_]+" (and
+     * one of the names from the corresponding OpDef's attr field).
+     * The values must have a type matching the corresponding OpDef
+     * attr's type field.
+     * TODO(josh11b): Add some examples here showing best practices.
+     * </pre>
+     *
+     * <code>map&lt;string, .tensorflow.AttrValue&gt; attr = 5;</code>
+     */
+
+    public boolean containsAttr(
+        java.lang.String key) {
+      if (key == null) { throw new java.lang.NullPointerException(); }
+      return internalGetAttr().getMap().containsKey(key);
+    }
+    /**
+     * Use {@link #getAttrMap()} instead.
+     */
+    @java.lang.Deprecated
+    public java.util.Map<java.lang.String, org.tensorflow.proto.framework.AttrValue> getAttr() {
+      return getAttrMap();
+    }
+    /**
+     * <pre>
+     * Operation-specific graph-construction-time configuration.
+     * Note that this should include all attrs defined in the
+     * corresponding OpDef, including those with a value matching
+     * the default -- this allows the default to change and makes
+     * NodeDefs easier to interpret on their own.  However, if
+     * an attr with a default is not specified in this list, the
+     * default will be used.
+     * The "names" (keys) must match the regexp "[a-z][a-z0-9_]+" (and
+     * one of the names from the corresponding OpDef's attr field).
+     * The values must have a type matching the corresponding OpDef
+     * attr's type field.
+     * TODO(josh11b): Add some examples here showing best practices.
+     * </pre>
+     *
+     * <code>map&lt;string, .tensorflow.AttrValue&gt; attr = 5;</code>
+     */
+
+    public java.util.Map<java.lang.String, org.tensorflow.proto.framework.AttrValue> getAttrMap() {
+      return internalGetAttr().getMap();
+    }
+    /**
+     * <pre>
+     * Operation-specific graph-construction-time configuration.
+     * Note that this should include all attrs defined in the
+     * corresponding OpDef, including those with a value matching
+     * the default -- this allows the default to change and makes
+     * NodeDefs easier to interpret on their own.  However, if
+     * an attr with a default is not specified in this list, the
+     * default will be used.
+     * The "names" (keys) must match the regexp "[a-z][a-z0-9_]+" (and
+     * one of the names from the corresponding OpDef's attr field).
+     * The values must have a type matching the corresponding OpDef
+     * attr's type field.
+     * TODO(josh11b): Add some examples here showing best practices.
+     * </pre>
+     *
+     * <code>map&lt;string, .tensorflow.AttrValue&gt; attr = 5;</code>
+     */
+
+    public org.tensorflow.proto.framework.AttrValue getAttrOrDefault(
+        java.lang.String key,
+        org.tensorflow.proto.framework.AttrValue defaultValue) {
+      if (key == null) { throw new java.lang.NullPointerException(); }
+      java.util.Map<java.lang.String, org.tensorflow.proto.framework.AttrValue> map =
+          internalGetAttr().getMap();
+      return map.containsKey(key) ? map.get(key) : defaultValue;
+    }
+    /**
+     * <pre>
+     * Operation-specific graph-construction-time configuration.
+     * Note that this should include all attrs defined in the
+     * corresponding OpDef, including those with a value matching
+     * the default -- this allows the default to change and makes
+     * NodeDefs easier to interpret on their own.  However, if
+     * an attr with a default is not specified in this list, the
+     * default will be used.
+     * The "names" (keys) must match the regexp "[a-z][a-z0-9_]+" (and
+     * one of the names from the corresponding OpDef's attr field).
+     * The values must have a type matching the corresponding OpDef
+     * attr's type field.
+     * TODO(josh11b): Add some examples here showing best practices.
+     * </pre>
+     *
+     * <code>map&lt;string, .tensorflow.AttrValue&gt; attr = 5;</code>
+     */
+
+    public org.tensorflow.proto.framework.AttrValue getAttrOrThrow(
+        java.lang.String key) {
+      if (key == null) { throw new java.lang.NullPointerException(); }
+      java.util.Map<java.lang.String, org.tensorflow.proto.framework.AttrValue> map =
+          internalGetAttr().getMap();
+      if (!map.containsKey(key)) {
+        throw new java.lang.IllegalArgumentException();
+      }
+      return map.get(key);
+    }
+
+    public Builder clearAttr() {
+      internalGetMutableAttr().getMutableMap()
+          .clear();
+      return this;
+    }
+    /**
+     * <pre>
+     * Operation-specific graph-construction-time configuration.
+     * Note that this should include all attrs defined in the
+     * corresponding OpDef, including those with a value matching
+     * the default -- this allows the default to change and makes
+     * NodeDefs easier to interpret on their own.  However, if
+     * an attr with a default is not specified in this list, the
+     * default will be used.
+     * The "names" (keys) must match the regexp "[a-z][a-z0-9_]+" (and
+     * one of the names from the corresponding OpDef's attr field).
+     * The values must have a type matching the corresponding OpDef
+     * attr's type field.
+     * TODO(josh11b): Add some examples here showing best practices.
+     * </pre>
+     *
+     * <code>map&lt;string, .tensorflow.AttrValue&gt; attr = 5;</code>
+     */
+
+    public Builder removeAttr(
+        java.lang.String key) {
+      if (key == null) { throw new java.lang.NullPointerException(); }
+      internalGetMutableAttr().getMutableMap()
+          .remove(key);
+      return this;
+    }
+    /**
+     * Use alternate mutation accessors instead.
+     */
+    @java.lang.Deprecated
+    public java.util.Map<java.lang.String, org.tensorflow.proto.framework.AttrValue>
+    getMutableAttr() {
+      return internalGetMutableAttr().getMutableMap();
+    }
+    /**
+     * <pre>
+     * Operation-specific graph-construction-time configuration.
+     * Note that this should include all attrs defined in the
+     * corresponding OpDef, including those with a value matching
+     * the default -- this allows the default to change and makes
+     * NodeDefs easier to interpret on their own.  However, if
+     * an attr with a default is not specified in this list, the
+     * default will be used.
+     * The "names" (keys) must match the regexp "[a-z][a-z0-9_]+" (and
+     * one of the names from the corresponding OpDef's attr field).
+     * The values must have a type matching the corresponding OpDef
+     * attr's type field.
+     * TODO(josh11b): Add some examples here showing best practices.
+     * </pre>
+     *
+     * <code>map&lt;string, .tensorflow.AttrValue&gt; attr = 5;</code>
+     */
+    public Builder putAttr(
+        java.lang.String key,
+        org.tensorflow.proto.framework.AttrValue value) {
+      if (key == null) { throw new java.lang.NullPointerException(); }
+      if (value == null) { throw new java.lang.NullPointerException(); }
+      internalGetMutableAttr().getMutableMap()
+          .put(key, value);
+      return this;
+    }
+    /**
+     * <pre>
+     * Operation-specific graph-construction-time configuration.
+     * Note that this should include all attrs defined in the
+     * corresponding OpDef, including those with a value matching
+     * the default -- this allows the default to change and makes
+     * NodeDefs easier to interpret on their own.  However, if
+     * an attr with a default is not specified in this list, the
+     * default will be used.
+     * The "names" (keys) must match the regexp "[a-z][a-z0-9_]+" (and
+     * one of the names from the corresponding OpDef's attr field).
+     * The values must have a type matching the corresponding OpDef
+     * attr's type field.
+     * TODO(josh11b): Add some examples here showing best practices.
+     * </pre>
+     *
+     * <code>map&lt;string, .tensorflow.AttrValue&gt; attr = 5;</code>
+     */
+
+    public Builder putAllAttr(
+        java.util.Map<java.lang.String, org.tensorflow.proto.framework.AttrValue> values) {
+      internalGetMutableAttr().getMutableMap()
+          .putAll(values);
+      return this;
+    }
+
+    private org.tensorflow.proto.framework.NodeDef.ExperimentalDebugInfo experimentalDebugInfo_;
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.NodeDef.ExperimentalDebugInfo, org.tensorflow.proto.framework.NodeDef.ExperimentalDebugInfo.Builder, org.tensorflow.proto.framework.NodeDef.ExperimentalDebugInfoOrBuilder> experimentalDebugInfoBuilder_;
+    /**
+     * <pre>
+     * This stores debug information associated with the node.
+     * </pre>
+     *
+     * <code>.tensorflow.NodeDef.ExperimentalDebugInfo experimental_debug_info = 6;</code>
+     */
+    public boolean hasExperimentalDebugInfo() {
+      return experimentalDebugInfoBuilder_ != null || experimentalDebugInfo_ != null;
+    }
+    /**
+     * <pre>
+     * This stores debug information associated with the node.
+     * </pre>
+     *
+     * <code>.tensorflow.NodeDef.ExperimentalDebugInfo experimental_debug_info = 6;</code>
+     */
+    public org.tensorflow.proto.framework.NodeDef.ExperimentalDebugInfo getExperimentalDebugInfo() {
+      if (experimentalDebugInfoBuilder_ == null) {
+        return experimentalDebugInfo_ == null ? org.tensorflow.proto.framework.NodeDef.ExperimentalDebugInfo.getDefaultInstance() : experimentalDebugInfo_;
+      } else {
+        return experimentalDebugInfoBuilder_.getMessage();
+      }
+    }
+    /**
+     * <pre>
+     * This stores debug information associated with the node.
+     * </pre>
+     *
+     * <code>.tensorflow.NodeDef.ExperimentalDebugInfo experimental_debug_info = 6;</code>
+     */
+    public Builder setExperimentalDebugInfo(org.tensorflow.proto.framework.NodeDef.ExperimentalDebugInfo value) {
+      if (experimentalDebugInfoBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        experimentalDebugInfo_ = value;
+        onChanged();
+      } else {
+        experimentalDebugInfoBuilder_.setMessage(value);
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * This stores debug information associated with the node.
+     * </pre>
+     *
+     * <code>.tensorflow.NodeDef.ExperimentalDebugInfo experimental_debug_info = 6;</code>
+     */
+    public Builder setExperimentalDebugInfo(
+        org.tensorflow.proto.framework.NodeDef.ExperimentalDebugInfo.Builder builderForValue) {
+      if (experimentalDebugInfoBuilder_ == null) {
+        experimentalDebugInfo_ = builderForValue.build();
+        onChanged();
+      } else {
+        experimentalDebugInfoBuilder_.setMessage(builderForValue.build());
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * This stores debug information associated with the node.
+     * </pre>
+     *
+     * <code>.tensorflow.NodeDef.ExperimentalDebugInfo experimental_debug_info = 6;</code>
+     */
+    public Builder mergeExperimentalDebugInfo(org.tensorflow.proto.framework.NodeDef.ExperimentalDebugInfo value) {
+      if (experimentalDebugInfoBuilder_ == null) {
+        if (experimentalDebugInfo_ != null) {
+          experimentalDebugInfo_ =
+            org.tensorflow.proto.framework.NodeDef.ExperimentalDebugInfo.newBuilder(experimentalDebugInfo_).mergeFrom(value).buildPartial();
+        } else {
+          experimentalDebugInfo_ = value;
+        }
+        onChanged();
+      } else {
+        experimentalDebugInfoBuilder_.mergeFrom(value);
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * This stores debug information associated with the node.
+     * </pre>
+     *
+     * <code>.tensorflow.NodeDef.ExperimentalDebugInfo experimental_debug_info = 6;</code>
+     */
+    public Builder clearExperimentalDebugInfo() {
+      if (experimentalDebugInfoBuilder_ == null) {
+        experimentalDebugInfo_ = null;
+        onChanged();
+      } else {
+        experimentalDebugInfo_ = null;
+        experimentalDebugInfoBuilder_ = null;
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * This stores debug information associated with the node.
+     * </pre>
+     *
+     * <code>.tensorflow.NodeDef.ExperimentalDebugInfo experimental_debug_info = 6;</code>
+     */
+    public org.tensorflow.proto.framework.NodeDef.ExperimentalDebugInfo.Builder getExperimentalDebugInfoBuilder() {
+      
+      onChanged();
+      return getExperimentalDebugInfoFieldBuilder().getBuilder();
+    }
+    /**
+     * <pre>
+     * This stores debug information associated with the node.
+     * </pre>
+     *
+     * <code>.tensorflow.NodeDef.ExperimentalDebugInfo experimental_debug_info = 6;</code>
+     */
+    public org.tensorflow.proto.framework.NodeDef.ExperimentalDebugInfoOrBuilder getExperimentalDebugInfoOrBuilder() {
+      if (experimentalDebugInfoBuilder_ != null) {
+        return experimentalDebugInfoBuilder_.getMessageOrBuilder();
+      } else {
+        return experimentalDebugInfo_ == null ?
+            org.tensorflow.proto.framework.NodeDef.ExperimentalDebugInfo.getDefaultInstance() : experimentalDebugInfo_;
+      }
+    }
+    /**
+     * <pre>
+     * This stores debug information associated with the node.
+     * </pre>
+     *
+     * <code>.tensorflow.NodeDef.ExperimentalDebugInfo experimental_debug_info = 6;</code>
+     */
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.NodeDef.ExperimentalDebugInfo, org.tensorflow.proto.framework.NodeDef.ExperimentalDebugInfo.Builder, org.tensorflow.proto.framework.NodeDef.ExperimentalDebugInfoOrBuilder> 
+        getExperimentalDebugInfoFieldBuilder() {
+      if (experimentalDebugInfoBuilder_ == null) {
+        experimentalDebugInfoBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+            org.tensorflow.proto.framework.NodeDef.ExperimentalDebugInfo, org.tensorflow.proto.framework.NodeDef.ExperimentalDebugInfo.Builder, org.tensorflow.proto.framework.NodeDef.ExperimentalDebugInfoOrBuilder>(
+                getExperimentalDebugInfo(),
+                getParentForChildren(),
+                isClean());
+        experimentalDebugInfo_ = null;
+      }
+      return experimentalDebugInfoBuilder_;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.NodeDef)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.NodeDef)
+  private static final org.tensorflow.proto.framework.NodeDef DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.NodeDef();
+  }
+
+  public static org.tensorflow.proto.framework.NodeDef getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<NodeDef>
+      PARSER = new com.google.protobuf.AbstractParser<NodeDef>() {
+    @java.lang.Override
+    public NodeDef parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new NodeDef(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<NodeDef> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<NodeDef> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.framework.NodeDef getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NodeDefOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NodeDefOrBuilder.java
new file mode 100644
index 00000000000..43971913d97
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NodeDefOrBuilder.java
@@ -0,0 +1,284 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/framework/node_def.proto
+
+package org.tensorflow.proto.framework;
+
+public interface NodeDefOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.NodeDef)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <pre>
+   * The name given to this operator. Used for naming inputs,
+   * logging, visualization, etc.  Unique within a single GraphDef.
+   * Must match the regexp "[A-Za-z0-9.][A-Za-z0-9_&gt;./]*".
+   * </pre>
+   *
+   * <code>string name = 1;</code>
+   */
+  java.lang.String getName();
+  /**
+   * <pre>
+   * The name given to this operator. Used for naming inputs,
+   * logging, visualization, etc.  Unique within a single GraphDef.
+   * Must match the regexp "[A-Za-z0-9.][A-Za-z0-9_&gt;./]*".
+   * </pre>
+   *
+   * <code>string name = 1;</code>
+   */
+  com.google.protobuf.ByteString
+      getNameBytes();
+
+  /**
+   * <pre>
+   * The operation name.  There may be custom parameters in attrs.
+   * Op names starting with an underscore are reserved for internal use.
+   * </pre>
+   *
+   * <code>string op = 2;</code>
+   */
+  java.lang.String getOp();
+  /**
+   * <pre>
+   * The operation name.  There may be custom parameters in attrs.
+   * Op names starting with an underscore are reserved for internal use.
+   * </pre>
+   *
+   * <code>string op = 2;</code>
+   */
+  com.google.protobuf.ByteString
+      getOpBytes();
+
+  /**
+   * <pre>
+   * Each input is "node:src_output" with "node" being a string name and
+   * "src_output" indicating which output tensor to use from "node". If
+   * "src_output" is 0 the ":0" suffix can be omitted.  Regular inputs
+   * may optionally be followed by control inputs that have the format
+   * "^node".
+   * </pre>
+   *
+   * <code>repeated string input = 3;</code>
+   */
+  java.util.List<java.lang.String>
+      getInputList();
+  /**
+   * <pre>
+   * Each input is "node:src_output" with "node" being a string name and
+   * "src_output" indicating which output tensor to use from "node". If
+   * "src_output" is 0 the ":0" suffix can be omitted.  Regular inputs
+   * may optionally be followed by control inputs that have the format
+   * "^node".
+   * </pre>
+   *
+   * <code>repeated string input = 3;</code>
+   */
+  int getInputCount();
+  /**
+   * <pre>
+   * Each input is "node:src_output" with "node" being a string name and
+   * "src_output" indicating which output tensor to use from "node". If
+   * "src_output" is 0 the ":0" suffix can be omitted.  Regular inputs
+   * may optionally be followed by control inputs that have the format
+   * "^node".
+   * </pre>
+   *
+   * <code>repeated string input = 3;</code>
+   */
+  java.lang.String getInput(int index);
+  /**
+   * <pre>
+   * Each input is "node:src_output" with "node" being a string name and
+   * "src_output" indicating which output tensor to use from "node". If
+   * "src_output" is 0 the ":0" suffix can be omitted.  Regular inputs
+   * may optionally be followed by control inputs that have the format
+   * "^node".
+   * </pre>
+   *
+   * <code>repeated string input = 3;</code>
+   */
+  com.google.protobuf.ByteString
+      getInputBytes(int index);
+
+  /**
+   * <pre>
+   * A (possibly partial) specification for the device on which this
+   * node should be placed.
+   * The expected syntax for this string is as follows:
+   * DEVICE_SPEC ::= PARTIAL_SPEC
+   * PARTIAL_SPEC ::= ("/" CONSTRAINT) *
+   * CONSTRAINT ::= ("job:" JOB_NAME)
+   *              | ("replica:" [1-9][0-9]*)
+   *              | ("task:" [1-9][0-9]*)
+   *              | ("device:" [A-Za-z]* ":" ([1-9][0-9]* | "*") )
+   * Valid values for this string include:
+   * * "/job:worker/replica:0/task:1/device:GPU:3"  (full specification)
+   * * "/job:worker/device:GPU:3"                   (partial specification)
+   * * ""                                    (no specification)
+   * If the constraints do not resolve to a single device (or if this
+   * field is empty or not present), the runtime will attempt to
+   * choose a device automatically.
+   * </pre>
+   *
+   * <code>string device = 4;</code>
+   */
+  java.lang.String getDevice();
+  /**
+   * <pre>
+   * A (possibly partial) specification for the device on which this
+   * node should be placed.
+   * The expected syntax for this string is as follows:
+   * DEVICE_SPEC ::= PARTIAL_SPEC
+   * PARTIAL_SPEC ::= ("/" CONSTRAINT) *
+   * CONSTRAINT ::= ("job:" JOB_NAME)
+   *              | ("replica:" [1-9][0-9]*)
+   *              | ("task:" [1-9][0-9]*)
+   *              | ("device:" [A-Za-z]* ":" ([1-9][0-9]* | "*") )
+   * Valid values for this string include:
+   * * "/job:worker/replica:0/task:1/device:GPU:3"  (full specification)
+   * * "/job:worker/device:GPU:3"                   (partial specification)
+   * * ""                                    (no specification)
+   * If the constraints do not resolve to a single device (or if this
+   * field is empty or not present), the runtime will attempt to
+   * choose a device automatically.
+   * </pre>
+   *
+   * <code>string device = 4;</code>
+   */
+  com.google.protobuf.ByteString
+      getDeviceBytes();
+
+  /**
+   * <pre>
+   * Operation-specific graph-construction-time configuration.
+   * Note that this should include all attrs defined in the
+   * corresponding OpDef, including those with a value matching
+   * the default -- this allows the default to change and makes
+   * NodeDefs easier to interpret on their own.  However, if
+   * an attr with a default is not specified in this list, the
+   * default will be used.
+   * The "names" (keys) must match the regexp "[a-z][a-z0-9_]+" (and
+   * one of the names from the corresponding OpDef's attr field).
+   * The values must have a type matching the corresponding OpDef
+   * attr's type field.
+   * TODO(josh11b): Add some examples here showing best practices.
+   * </pre>
+   *
+   * <code>map&lt;string, .tensorflow.AttrValue&gt; attr = 5;</code>
+   */
+  int getAttrCount();
+  /**
+   * <pre>
+   * Operation-specific graph-construction-time configuration.
+   * Note that this should include all attrs defined in the
+   * corresponding OpDef, including those with a value matching
+   * the default -- this allows the default to change and makes
+   * NodeDefs easier to interpret on their own.  However, if
+   * an attr with a default is not specified in this list, the
+   * default will be used.
+   * The "names" (keys) must match the regexp "[a-z][a-z0-9_]+" (and
+   * one of the names from the corresponding OpDef's attr field).
+   * The values must have a type matching the corresponding OpDef
+   * attr's type field.
+   * TODO(josh11b): Add some examples here showing best practices.
+   * </pre>
+   *
+   * <code>map&lt;string, .tensorflow.AttrValue&gt; attr = 5;</code>
+   */
+  boolean containsAttr(
+      java.lang.String key);
+  /**
+   * Use {@link #getAttrMap()} instead.
+   */
+  @java.lang.Deprecated
+  java.util.Map<java.lang.String, org.tensorflow.proto.framework.AttrValue>
+  getAttr();
+  /**
+   * <pre>
+   * Operation-specific graph-construction-time configuration.
+   * Note that this should include all attrs defined in the
+   * corresponding OpDef, including those with a value matching
+   * the default -- this allows the default to change and makes
+   * NodeDefs easier to interpret on their own.  However, if
+   * an attr with a default is not specified in this list, the
+   * default will be used.
+   * The "names" (keys) must match the regexp "[a-z][a-z0-9_]+" (and
+   * one of the names from the corresponding OpDef's attr field).
+   * The values must have a type matching the corresponding OpDef
+   * attr's type field.
+   * TODO(josh11b): Add some examples here showing best practices.
+   * </pre>
+   *
+   * <code>map&lt;string, .tensorflow.AttrValue&gt; attr = 5;</code>
+   */
+  java.util.Map<java.lang.String, org.tensorflow.proto.framework.AttrValue>
+  getAttrMap();
+  /**
+   * <pre>
+   * Operation-specific graph-construction-time configuration.
+   * Note that this should include all attrs defined in the
+   * corresponding OpDef, including those with a value matching
+   * the default -- this allows the default to change and makes
+   * NodeDefs easier to interpret on their own.  However, if
+   * an attr with a default is not specified in this list, the
+   * default will be used.
+   * The "names" (keys) must match the regexp "[a-z][a-z0-9_]+" (and
+   * one of the names from the corresponding OpDef's attr field).
+   * The values must have a type matching the corresponding OpDef
+   * attr's type field.
+   * TODO(josh11b): Add some examples here showing best practices.
+   * </pre>
+   *
+   * <code>map&lt;string, .tensorflow.AttrValue&gt; attr = 5;</code>
+   */
+
+  org.tensorflow.proto.framework.AttrValue getAttrOrDefault(
+      java.lang.String key,
+      org.tensorflow.proto.framework.AttrValue defaultValue);
+  /**
+   * <pre>
+   * Operation-specific graph-construction-time configuration.
+   * Note that this should include all attrs defined in the
+   * corresponding OpDef, including those with a value matching
+   * the default -- this allows the default to change and makes
+   * NodeDefs easier to interpret on their own.  However, if
+   * an attr with a default is not specified in this list, the
+   * default will be used.
+   * The "names" (keys) must match the regexp "[a-z][a-z0-9_]+" (and
+   * one of the names from the corresponding OpDef's attr field).
+   * The values must have a type matching the corresponding OpDef
+   * attr's type field.
+   * TODO(josh11b): Add some examples here showing best practices.
+   * </pre>
+   *
+   * <code>map&lt;string, .tensorflow.AttrValue&gt; attr = 5;</code>
+   */
+
+  org.tensorflow.proto.framework.AttrValue getAttrOrThrow(
+      java.lang.String key);
+
+  /**
+   * <pre>
+   * This stores debug information associated with the node.
+   * </pre>
+   *
+   * <code>.tensorflow.NodeDef.ExperimentalDebugInfo experimental_debug_info = 6;</code>
+   */
+  boolean hasExperimentalDebugInfo();
+  /**
+   * <pre>
+   * This stores debug information associated with the node.
+   * </pre>
+   *
+   * <code>.tensorflow.NodeDef.ExperimentalDebugInfo experimental_debug_info = 6;</code>
+   */
+  org.tensorflow.proto.framework.NodeDef.ExperimentalDebugInfo getExperimentalDebugInfo();
+  /**
+   * <pre>
+   * This stores debug information associated with the node.
+   * </pre>
+   *
+   * <code>.tensorflow.NodeDef.ExperimentalDebugInfo experimental_debug_info = 6;</code>
+   */
+  org.tensorflow.proto.framework.NodeDef.ExperimentalDebugInfoOrBuilder getExperimentalDebugInfoOrBuilder();
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NodeExecStats.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NodeExecStats.java
new file mode 100644
index 00000000000..93e1e3df32c
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NodeExecStats.java
@@ -0,0 +1,2580 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/framework/step_stats.proto
+
+package org.tensorflow.proto.framework;
+
+/**
+ * <pre>
+ * Time/size stats recorded for a single execution of a graph node.
+ * </pre>
+ *
+ * Protobuf type {@code tensorflow.NodeExecStats}
+ */
+public  final class NodeExecStats extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.NodeExecStats)
+    NodeExecStatsOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use NodeExecStats.newBuilder() to construct.
+  private NodeExecStats(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private NodeExecStats() {
+    nodeName_ = "";
+    memory_ = java.util.Collections.emptyList();
+    output_ = java.util.Collections.emptyList();
+    timelineLabel_ = "";
+    referencedTensor_ = java.util.Collections.emptyList();
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new NodeExecStats();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private NodeExecStats(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    int mutable_bitField0_ = 0;
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 10: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            nodeName_ = s;
+            break;
+          }
+          case 16: {
+
+            allStartMicros_ = input.readInt64();
+            break;
+          }
+          case 24: {
+
+            opStartRelMicros_ = input.readInt64();
+            break;
+          }
+          case 32: {
+
+            opEndRelMicros_ = input.readInt64();
+            break;
+          }
+          case 40: {
+
+            allEndRelMicros_ = input.readInt64();
+            break;
+          }
+          case 50: {
+            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
+              memory_ = new java.util.ArrayList<org.tensorflow.proto.framework.AllocatorMemoryUsed>();
+              mutable_bitField0_ |= 0x00000001;
+            }
+            memory_.add(
+                input.readMessage(org.tensorflow.proto.framework.AllocatorMemoryUsed.parser(), extensionRegistry));
+            break;
+          }
+          case 58: {
+            if (!((mutable_bitField0_ & 0x00000002) != 0)) {
+              output_ = new java.util.ArrayList<org.tensorflow.proto.framework.NodeOutput>();
+              mutable_bitField0_ |= 0x00000002;
+            }
+            output_.add(
+                input.readMessage(org.tensorflow.proto.framework.NodeOutput.parser(), extensionRegistry));
+            break;
+          }
+          case 66: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            timelineLabel_ = s;
+            break;
+          }
+          case 72: {
+
+            scheduledMicros_ = input.readInt64();
+            break;
+          }
+          case 80: {
+
+            threadId_ = input.readUInt32();
+            break;
+          }
+          case 90: {
+            if (!((mutable_bitField0_ & 0x00000004) != 0)) {
+              referencedTensor_ = new java.util.ArrayList<org.tensorflow.proto.framework.AllocationDescription>();
+              mutable_bitField0_ |= 0x00000004;
+            }
+            referencedTensor_.add(
+                input.readMessage(org.tensorflow.proto.framework.AllocationDescription.parser(), extensionRegistry));
+            break;
+          }
+          case 98: {
+            org.tensorflow.proto.framework.MemoryStats.Builder subBuilder = null;
+            if (memoryStats_ != null) {
+              subBuilder = memoryStats_.toBuilder();
+            }
+            memoryStats_ = input.readMessage(org.tensorflow.proto.framework.MemoryStats.parser(), extensionRegistry);
+            if (subBuilder != null) {
+              subBuilder.mergeFrom(memoryStats_);
+              memoryStats_ = subBuilder.buildPartial();
+            }
+
+            break;
+          }
+          case 104: {
+
+            allStartNanos_ = input.readInt64();
+            break;
+          }
+          case 112: {
+
+            opStartRelNanos_ = input.readInt64();
+            break;
+          }
+          case 120: {
+
+            opEndRelNanos_ = input.readInt64();
+            break;
+          }
+          case 128: {
+
+            allEndRelNanos_ = input.readInt64();
+            break;
+          }
+          case 136: {
+
+            scheduledNanos_ = input.readInt64();
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      if (((mutable_bitField0_ & 0x00000001) != 0)) {
+        memory_ = java.util.Collections.unmodifiableList(memory_);
+      }
+      if (((mutable_bitField0_ & 0x00000002) != 0)) {
+        output_ = java.util.Collections.unmodifiableList(output_);
+      }
+      if (((mutable_bitField0_ & 0x00000004) != 0)) {
+        referencedTensor_ = java.util.Collections.unmodifiableList(referencedTensor_);
+      }
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.framework.StepStatsProtos.internal_static_tensorflow_NodeExecStats_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.framework.StepStatsProtos.internal_static_tensorflow_NodeExecStats_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.framework.NodeExecStats.class, org.tensorflow.proto.framework.NodeExecStats.Builder.class);
+  }
+
+  public static final int NODE_NAME_FIELD_NUMBER = 1;
+  private volatile java.lang.Object nodeName_;
+  /**
+   * <pre>
+   * TODO(tucker): Use some more compact form of node identity than
+   * the full string name.  Either all processes should agree on a
+   * global id (cost_id?) for each node, or we should use a hash of
+   * the name.
+   * </pre>
+   *
+   * <code>string node_name = 1;</code>
+   */
+  public java.lang.String getNodeName() {
+    java.lang.Object ref = nodeName_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      nodeName_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * TODO(tucker): Use some more compact form of node identity than
+   * the full string name.  Either all processes should agree on a
+   * global id (cost_id?) for each node, or we should use a hash of
+   * the name.
+   * </pre>
+   *
+   * <code>string node_name = 1;</code>
+   */
+  public com.google.protobuf.ByteString
+      getNodeNameBytes() {
+    java.lang.Object ref = nodeName_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      nodeName_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int ALL_START_MICROS_FIELD_NUMBER = 2;
+  private long allStartMicros_;
+  /**
+   * <code>int64 all_start_micros = 2;</code>
+   */
+  public long getAllStartMicros() {
+    return allStartMicros_;
+  }
+
+  public static final int OP_START_REL_MICROS_FIELD_NUMBER = 3;
+  private long opStartRelMicros_;
+  /**
+   * <code>int64 op_start_rel_micros = 3;</code>
+   */
+  public long getOpStartRelMicros() {
+    return opStartRelMicros_;
+  }
+
+  public static final int OP_END_REL_MICROS_FIELD_NUMBER = 4;
+  private long opEndRelMicros_;
+  /**
+   * <code>int64 op_end_rel_micros = 4;</code>
+   */
+  public long getOpEndRelMicros() {
+    return opEndRelMicros_;
+  }
+
+  public static final int ALL_END_REL_MICROS_FIELD_NUMBER = 5;
+  private long allEndRelMicros_;
+  /**
+   * <code>int64 all_end_rel_micros = 5;</code>
+   */
+  public long getAllEndRelMicros() {
+    return allEndRelMicros_;
+  }
+
+  public static final int MEMORY_FIELD_NUMBER = 6;
+  private java.util.List<org.tensorflow.proto.framework.AllocatorMemoryUsed> memory_;
+  /**
+   * <code>repeated .tensorflow.AllocatorMemoryUsed memory = 6;</code>
+   */
+  public java.util.List<org.tensorflow.proto.framework.AllocatorMemoryUsed> getMemoryList() {
+    return memory_;
+  }
+  /**
+   * <code>repeated .tensorflow.AllocatorMemoryUsed memory = 6;</code>
+   */
+  public java.util.List<? extends org.tensorflow.proto.framework.AllocatorMemoryUsedOrBuilder> 
+      getMemoryOrBuilderList() {
+    return memory_;
+  }
+  /**
+   * <code>repeated .tensorflow.AllocatorMemoryUsed memory = 6;</code>
+   */
+  public int getMemoryCount() {
+    return memory_.size();
+  }
+  /**
+   * <code>repeated .tensorflow.AllocatorMemoryUsed memory = 6;</code>
+   */
+  public org.tensorflow.proto.framework.AllocatorMemoryUsed getMemory(int index) {
+    return memory_.get(index);
+  }
+  /**
+   * <code>repeated .tensorflow.AllocatorMemoryUsed memory = 6;</code>
+   */
+  public org.tensorflow.proto.framework.AllocatorMemoryUsedOrBuilder getMemoryOrBuilder(
+      int index) {
+    return memory_.get(index);
+  }
+
+  public static final int OUTPUT_FIELD_NUMBER = 7;
+  private java.util.List<org.tensorflow.proto.framework.NodeOutput> output_;
+  /**
+   * <code>repeated .tensorflow.NodeOutput output = 7;</code>
+   */
+  public java.util.List<org.tensorflow.proto.framework.NodeOutput> getOutputList() {
+    return output_;
+  }
+  /**
+   * <code>repeated .tensorflow.NodeOutput output = 7;</code>
+   */
+  public java.util.List<? extends org.tensorflow.proto.framework.NodeOutputOrBuilder> 
+      getOutputOrBuilderList() {
+    return output_;
+  }
+  /**
+   * <code>repeated .tensorflow.NodeOutput output = 7;</code>
+   */
+  public int getOutputCount() {
+    return output_.size();
+  }
+  /**
+   * <code>repeated .tensorflow.NodeOutput output = 7;</code>
+   */
+  public org.tensorflow.proto.framework.NodeOutput getOutput(int index) {
+    return output_.get(index);
+  }
+  /**
+   * <code>repeated .tensorflow.NodeOutput output = 7;</code>
+   */
+  public org.tensorflow.proto.framework.NodeOutputOrBuilder getOutputOrBuilder(
+      int index) {
+    return output_.get(index);
+  }
+
+  public static final int TIMELINE_LABEL_FIELD_NUMBER = 8;
+  private volatile java.lang.Object timelineLabel_;
+  /**
+   * <code>string timeline_label = 8;</code>
+   */
+  public java.lang.String getTimelineLabel() {
+    java.lang.Object ref = timelineLabel_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      timelineLabel_ = s;
+      return s;
+    }
+  }
+  /**
+   * <code>string timeline_label = 8;</code>
+   */
+  public com.google.protobuf.ByteString
+      getTimelineLabelBytes() {
+    java.lang.Object ref = timelineLabel_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      timelineLabel_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int SCHEDULED_MICROS_FIELD_NUMBER = 9;
+  private long scheduledMicros_;
+  /**
+   * <code>int64 scheduled_micros = 9;</code>
+   */
+  public long getScheduledMicros() {
+    return scheduledMicros_;
+  }
+
+  public static final int THREAD_ID_FIELD_NUMBER = 10;
+  private int threadId_;
+  /**
+   * <code>uint32 thread_id = 10;</code>
+   */
+  public int getThreadId() {
+    return threadId_;
+  }
+
+  public static final int REFERENCED_TENSOR_FIELD_NUMBER = 11;
+  private java.util.List<org.tensorflow.proto.framework.AllocationDescription> referencedTensor_;
+  /**
+   * <code>repeated .tensorflow.AllocationDescription referenced_tensor = 11;</code>
+   */
+  public java.util.List<org.tensorflow.proto.framework.AllocationDescription> getReferencedTensorList() {
+    return referencedTensor_;
+  }
+  /**
+   * <code>repeated .tensorflow.AllocationDescription referenced_tensor = 11;</code>
+   */
+  public java.util.List<? extends org.tensorflow.proto.framework.AllocationDescriptionOrBuilder> 
+      getReferencedTensorOrBuilderList() {
+    return referencedTensor_;
+  }
+  /**
+   * <code>repeated .tensorflow.AllocationDescription referenced_tensor = 11;</code>
+   */
+  public int getReferencedTensorCount() {
+    return referencedTensor_.size();
+  }
+  /**
+   * <code>repeated .tensorflow.AllocationDescription referenced_tensor = 11;</code>
+   */
+  public org.tensorflow.proto.framework.AllocationDescription getReferencedTensor(int index) {
+    return referencedTensor_.get(index);
+  }
+  /**
+   * <code>repeated .tensorflow.AllocationDescription referenced_tensor = 11;</code>
+   */
+  public org.tensorflow.proto.framework.AllocationDescriptionOrBuilder getReferencedTensorOrBuilder(
+      int index) {
+    return referencedTensor_.get(index);
+  }
+
+  public static final int MEMORY_STATS_FIELD_NUMBER = 12;
+  private org.tensorflow.proto.framework.MemoryStats memoryStats_;
+  /**
+   * <code>.tensorflow.MemoryStats memory_stats = 12;</code>
+   */
+  public boolean hasMemoryStats() {
+    return memoryStats_ != null;
+  }
+  /**
+   * <code>.tensorflow.MemoryStats memory_stats = 12;</code>
+   */
+  public org.tensorflow.proto.framework.MemoryStats getMemoryStats() {
+    return memoryStats_ == null ? org.tensorflow.proto.framework.MemoryStats.getDefaultInstance() : memoryStats_;
+  }
+  /**
+   * <code>.tensorflow.MemoryStats memory_stats = 12;</code>
+   */
+  public org.tensorflow.proto.framework.MemoryStatsOrBuilder getMemoryStatsOrBuilder() {
+    return getMemoryStats();
+  }
+
+  public static final int ALL_START_NANOS_FIELD_NUMBER = 13;
+  private long allStartNanos_;
+  /**
+   * <code>int64 all_start_nanos = 13;</code>
+   */
+  public long getAllStartNanos() {
+    return allStartNanos_;
+  }
+
+  public static final int OP_START_REL_NANOS_FIELD_NUMBER = 14;
+  private long opStartRelNanos_;
+  /**
+   * <code>int64 op_start_rel_nanos = 14;</code>
+   */
+  public long getOpStartRelNanos() {
+    return opStartRelNanos_;
+  }
+
+  public static final int OP_END_REL_NANOS_FIELD_NUMBER = 15;
+  private long opEndRelNanos_;
+  /**
+   * <code>int64 op_end_rel_nanos = 15;</code>
+   */
+  public long getOpEndRelNanos() {
+    return opEndRelNanos_;
+  }
+
+  public static final int ALL_END_REL_NANOS_FIELD_NUMBER = 16;
+  private long allEndRelNanos_;
+  /**
+   * <code>int64 all_end_rel_nanos = 16;</code>
+   */
+  public long getAllEndRelNanos() {
+    return allEndRelNanos_;
+  }
+
+  public static final int SCHEDULED_NANOS_FIELD_NUMBER = 17;
+  private long scheduledNanos_;
+  /**
+   * <code>int64 scheduled_nanos = 17;</code>
+   */
+  public long getScheduledNanos() {
+    return scheduledNanos_;
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    if (!getNodeNameBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 1, nodeName_);
+    }
+    if (allStartMicros_ != 0L) {
+      output.writeInt64(2, allStartMicros_);
+    }
+    if (opStartRelMicros_ != 0L) {
+      output.writeInt64(3, opStartRelMicros_);
+    }
+    if (opEndRelMicros_ != 0L) {
+      output.writeInt64(4, opEndRelMicros_);
+    }
+    if (allEndRelMicros_ != 0L) {
+      output.writeInt64(5, allEndRelMicros_);
+    }
+    for (int i = 0; i < memory_.size(); i++) {
+      output.writeMessage(6, memory_.get(i));
+    }
+    for (int i = 0; i < output_.size(); i++) {
+      output.writeMessage(7, output_.get(i));
+    }
+    if (!getTimelineLabelBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 8, timelineLabel_);
+    }
+    if (scheduledMicros_ != 0L) {
+      output.writeInt64(9, scheduledMicros_);
+    }
+    if (threadId_ != 0) {
+      output.writeUInt32(10, threadId_);
+    }
+    for (int i = 0; i < referencedTensor_.size(); i++) {
+      output.writeMessage(11, referencedTensor_.get(i));
+    }
+    if (memoryStats_ != null) {
+      output.writeMessage(12, getMemoryStats());
+    }
+    if (allStartNanos_ != 0L) {
+      output.writeInt64(13, allStartNanos_);
+    }
+    if (opStartRelNanos_ != 0L) {
+      output.writeInt64(14, opStartRelNanos_);
+    }
+    if (opEndRelNanos_ != 0L) {
+      output.writeInt64(15, opEndRelNanos_);
+    }
+    if (allEndRelNanos_ != 0L) {
+      output.writeInt64(16, allEndRelNanos_);
+    }
+    if (scheduledNanos_ != 0L) {
+      output.writeInt64(17, scheduledNanos_);
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (!getNodeNameBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, nodeName_);
+    }
+    if (allStartMicros_ != 0L) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt64Size(2, allStartMicros_);
+    }
+    if (opStartRelMicros_ != 0L) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt64Size(3, opStartRelMicros_);
+    }
+    if (opEndRelMicros_ != 0L) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt64Size(4, opEndRelMicros_);
+    }
+    if (allEndRelMicros_ != 0L) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt64Size(5, allEndRelMicros_);
+    }
+    for (int i = 0; i < memory_.size(); i++) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(6, memory_.get(i));
+    }
+    for (int i = 0; i < output_.size(); i++) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(7, output_.get(i));
+    }
+    if (!getTimelineLabelBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(8, timelineLabel_);
+    }
+    if (scheduledMicros_ != 0L) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt64Size(9, scheduledMicros_);
+    }
+    if (threadId_ != 0) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeUInt32Size(10, threadId_);
+    }
+    for (int i = 0; i < referencedTensor_.size(); i++) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(11, referencedTensor_.get(i));
+    }
+    if (memoryStats_ != null) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(12, getMemoryStats());
+    }
+    if (allStartNanos_ != 0L) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt64Size(13, allStartNanos_);
+    }
+    if (opStartRelNanos_ != 0L) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt64Size(14, opStartRelNanos_);
+    }
+    if (opEndRelNanos_ != 0L) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt64Size(15, opEndRelNanos_);
+    }
+    if (allEndRelNanos_ != 0L) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt64Size(16, allEndRelNanos_);
+    }
+    if (scheduledNanos_ != 0L) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt64Size(17, scheduledNanos_);
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.framework.NodeExecStats)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.framework.NodeExecStats other = (org.tensorflow.proto.framework.NodeExecStats) obj;
+
+    if (!getNodeName()
+        .equals(other.getNodeName())) return false;
+    if (getAllStartMicros()
+        != other.getAllStartMicros()) return false;
+    if (getOpStartRelMicros()
+        != other.getOpStartRelMicros()) return false;
+    if (getOpEndRelMicros()
+        != other.getOpEndRelMicros()) return false;
+    if (getAllEndRelMicros()
+        != other.getAllEndRelMicros()) return false;
+    if (!getMemoryList()
+        .equals(other.getMemoryList())) return false;
+    if (!getOutputList()
+        .equals(other.getOutputList())) return false;
+    if (!getTimelineLabel()
+        .equals(other.getTimelineLabel())) return false;
+    if (getScheduledMicros()
+        != other.getScheduledMicros()) return false;
+    if (getThreadId()
+        != other.getThreadId()) return false;
+    if (!getReferencedTensorList()
+        .equals(other.getReferencedTensorList())) return false;
+    if (hasMemoryStats() != other.hasMemoryStats()) return false;
+    if (hasMemoryStats()) {
+      if (!getMemoryStats()
+          .equals(other.getMemoryStats())) return false;
+    }
+    if (getAllStartNanos()
+        != other.getAllStartNanos()) return false;
+    if (getOpStartRelNanos()
+        != other.getOpStartRelNanos()) return false;
+    if (getOpEndRelNanos()
+        != other.getOpEndRelNanos()) return false;
+    if (getAllEndRelNanos()
+        != other.getAllEndRelNanos()) return false;
+    if (getScheduledNanos()
+        != other.getScheduledNanos()) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    hash = (37 * hash) + NODE_NAME_FIELD_NUMBER;
+    hash = (53 * hash) + getNodeName().hashCode();
+    hash = (37 * hash) + ALL_START_MICROS_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+        getAllStartMicros());
+    hash = (37 * hash) + OP_START_REL_MICROS_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+        getOpStartRelMicros());
+    hash = (37 * hash) + OP_END_REL_MICROS_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+        getOpEndRelMicros());
+    hash = (37 * hash) + ALL_END_REL_MICROS_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+        getAllEndRelMicros());
+    if (getMemoryCount() > 0) {
+      hash = (37 * hash) + MEMORY_FIELD_NUMBER;
+      hash = (53 * hash) + getMemoryList().hashCode();
+    }
+    if (getOutputCount() > 0) {
+      hash = (37 * hash) + OUTPUT_FIELD_NUMBER;
+      hash = (53 * hash) + getOutputList().hashCode();
+    }
+    hash = (37 * hash) + TIMELINE_LABEL_FIELD_NUMBER;
+    hash = (53 * hash) + getTimelineLabel().hashCode();
+    hash = (37 * hash) + SCHEDULED_MICROS_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+        getScheduledMicros());
+    hash = (37 * hash) + THREAD_ID_FIELD_NUMBER;
+    hash = (53 * hash) + getThreadId();
+    if (getReferencedTensorCount() > 0) {
+      hash = (37 * hash) + REFERENCED_TENSOR_FIELD_NUMBER;
+      hash = (53 * hash) + getReferencedTensorList().hashCode();
+    }
+    if (hasMemoryStats()) {
+      hash = (37 * hash) + MEMORY_STATS_FIELD_NUMBER;
+      hash = (53 * hash) + getMemoryStats().hashCode();
+    }
+    hash = (37 * hash) + ALL_START_NANOS_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+        getAllStartNanos());
+    hash = (37 * hash) + OP_START_REL_NANOS_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+        getOpStartRelNanos());
+    hash = (37 * hash) + OP_END_REL_NANOS_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+        getOpEndRelNanos());
+    hash = (37 * hash) + ALL_END_REL_NANOS_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+        getAllEndRelNanos());
+    hash = (37 * hash) + SCHEDULED_NANOS_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+        getScheduledNanos());
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.framework.NodeExecStats parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.NodeExecStats parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.NodeExecStats parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.NodeExecStats parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.NodeExecStats parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.NodeExecStats parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.NodeExecStats parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.NodeExecStats parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.NodeExecStats parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.NodeExecStats parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.NodeExecStats parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.NodeExecStats parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.framework.NodeExecStats prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * <pre>
+   * Time/size stats recorded for a single execution of a graph node.
+   * </pre>
+   *
+   * Protobuf type {@code tensorflow.NodeExecStats}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.NodeExecStats)
+      org.tensorflow.proto.framework.NodeExecStatsOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.StepStatsProtos.internal_static_tensorflow_NodeExecStats_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.StepStatsProtos.internal_static_tensorflow_NodeExecStats_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.NodeExecStats.class, org.tensorflow.proto.framework.NodeExecStats.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.framework.NodeExecStats.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+        getMemoryFieldBuilder();
+        getOutputFieldBuilder();
+        getReferencedTensorFieldBuilder();
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      nodeName_ = "";
+
+      allStartMicros_ = 0L;
+
+      opStartRelMicros_ = 0L;
+
+      opEndRelMicros_ = 0L;
+
+      allEndRelMicros_ = 0L;
+
+      if (memoryBuilder_ == null) {
+        memory_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000001);
+      } else {
+        memoryBuilder_.clear();
+      }
+      if (outputBuilder_ == null) {
+        output_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000002);
+      } else {
+        outputBuilder_.clear();
+      }
+      timelineLabel_ = "";
+
+      scheduledMicros_ = 0L;
+
+      threadId_ = 0;
+
+      if (referencedTensorBuilder_ == null) {
+        referencedTensor_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000004);
+      } else {
+        referencedTensorBuilder_.clear();
+      }
+      if (memoryStatsBuilder_ == null) {
+        memoryStats_ = null;
+      } else {
+        memoryStats_ = null;
+        memoryStatsBuilder_ = null;
+      }
+      allStartNanos_ = 0L;
+
+      opStartRelNanos_ = 0L;
+
+      opEndRelNanos_ = 0L;
+
+      allEndRelNanos_ = 0L;
+
+      scheduledNanos_ = 0L;
+
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.framework.StepStatsProtos.internal_static_tensorflow_NodeExecStats_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.NodeExecStats getDefaultInstanceForType() {
+      return org.tensorflow.proto.framework.NodeExecStats.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.NodeExecStats build() {
+      org.tensorflow.proto.framework.NodeExecStats result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.NodeExecStats buildPartial() {
+      org.tensorflow.proto.framework.NodeExecStats result = new org.tensorflow.proto.framework.NodeExecStats(this);
+      int from_bitField0_ = bitField0_;
+      result.nodeName_ = nodeName_;
+      result.allStartMicros_ = allStartMicros_;
+      result.opStartRelMicros_ = opStartRelMicros_;
+      result.opEndRelMicros_ = opEndRelMicros_;
+      result.allEndRelMicros_ = allEndRelMicros_;
+      if (memoryBuilder_ == null) {
+        if (((bitField0_ & 0x00000001) != 0)) {
+          memory_ = java.util.Collections.unmodifiableList(memory_);
+          bitField0_ = (bitField0_ & ~0x00000001);
+        }
+        result.memory_ = memory_;
+      } else {
+        result.memory_ = memoryBuilder_.build();
+      }
+      if (outputBuilder_ == null) {
+        if (((bitField0_ & 0x00000002) != 0)) {
+          output_ = java.util.Collections.unmodifiableList(output_);
+          bitField0_ = (bitField0_ & ~0x00000002);
+        }
+        result.output_ = output_;
+      } else {
+        result.output_ = outputBuilder_.build();
+      }
+      result.timelineLabel_ = timelineLabel_;
+      result.scheduledMicros_ = scheduledMicros_;
+      result.threadId_ = threadId_;
+      if (referencedTensorBuilder_ == null) {
+        if (((bitField0_ & 0x00000004) != 0)) {
+          referencedTensor_ = java.util.Collections.unmodifiableList(referencedTensor_);
+          bitField0_ = (bitField0_ & ~0x00000004);
+        }
+        result.referencedTensor_ = referencedTensor_;
+      } else {
+        result.referencedTensor_ = referencedTensorBuilder_.build();
+      }
+      if (memoryStatsBuilder_ == null) {
+        result.memoryStats_ = memoryStats_;
+      } else {
+        result.memoryStats_ = memoryStatsBuilder_.build();
+      }
+      result.allStartNanos_ = allStartNanos_;
+      result.opStartRelNanos_ = opStartRelNanos_;
+      result.opEndRelNanos_ = opEndRelNanos_;
+      result.allEndRelNanos_ = allEndRelNanos_;
+      result.scheduledNanos_ = scheduledNanos_;
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.framework.NodeExecStats) {
+        return mergeFrom((org.tensorflow.proto.framework.NodeExecStats)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.framework.NodeExecStats other) {
+      if (other == org.tensorflow.proto.framework.NodeExecStats.getDefaultInstance()) return this;
+      if (!other.getNodeName().isEmpty()) {
+        nodeName_ = other.nodeName_;
+        onChanged();
+      }
+      if (other.getAllStartMicros() != 0L) {
+        setAllStartMicros(other.getAllStartMicros());
+      }
+      if (other.getOpStartRelMicros() != 0L) {
+        setOpStartRelMicros(other.getOpStartRelMicros());
+      }
+      if (other.getOpEndRelMicros() != 0L) {
+        setOpEndRelMicros(other.getOpEndRelMicros());
+      }
+      if (other.getAllEndRelMicros() != 0L) {
+        setAllEndRelMicros(other.getAllEndRelMicros());
+      }
+      if (memoryBuilder_ == null) {
+        if (!other.memory_.isEmpty()) {
+          if (memory_.isEmpty()) {
+            memory_ = other.memory_;
+            bitField0_ = (bitField0_ & ~0x00000001);
+          } else {
+            ensureMemoryIsMutable();
+            memory_.addAll(other.memory_);
+          }
+          onChanged();
+        }
+      } else {
+        if (!other.memory_.isEmpty()) {
+          if (memoryBuilder_.isEmpty()) {
+            memoryBuilder_.dispose();
+            memoryBuilder_ = null;
+            memory_ = other.memory_;
+            bitField0_ = (bitField0_ & ~0x00000001);
+            memoryBuilder_ = 
+              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
+                 getMemoryFieldBuilder() : null;
+          } else {
+            memoryBuilder_.addAllMessages(other.memory_);
+          }
+        }
+      }
+      if (outputBuilder_ == null) {
+        if (!other.output_.isEmpty()) {
+          if (output_.isEmpty()) {
+            output_ = other.output_;
+            bitField0_ = (bitField0_ & ~0x00000002);
+          } else {
+            ensureOutputIsMutable();
+            output_.addAll(other.output_);
+          }
+          onChanged();
+        }
+      } else {
+        if (!other.output_.isEmpty()) {
+          if (outputBuilder_.isEmpty()) {
+            outputBuilder_.dispose();
+            outputBuilder_ = null;
+            output_ = other.output_;
+            bitField0_ = (bitField0_ & ~0x00000002);
+            outputBuilder_ = 
+              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
+                 getOutputFieldBuilder() : null;
+          } else {
+            outputBuilder_.addAllMessages(other.output_);
+          }
+        }
+      }
+      if (!other.getTimelineLabel().isEmpty()) {
+        timelineLabel_ = other.timelineLabel_;
+        onChanged();
+      }
+      if (other.getScheduledMicros() != 0L) {
+        setScheduledMicros(other.getScheduledMicros());
+      }
+      if (other.getThreadId() != 0) {
+        setThreadId(other.getThreadId());
+      }
+      if (referencedTensorBuilder_ == null) {
+        if (!other.referencedTensor_.isEmpty()) {
+          if (referencedTensor_.isEmpty()) {
+            referencedTensor_ = other.referencedTensor_;
+            bitField0_ = (bitField0_ & ~0x00000004);
+          } else {
+            ensureReferencedTensorIsMutable();
+            referencedTensor_.addAll(other.referencedTensor_);
+          }
+          onChanged();
+        }
+      } else {
+        if (!other.referencedTensor_.isEmpty()) {
+          if (referencedTensorBuilder_.isEmpty()) {
+            referencedTensorBuilder_.dispose();
+            referencedTensorBuilder_ = null;
+            referencedTensor_ = other.referencedTensor_;
+            bitField0_ = (bitField0_ & ~0x00000004);
+            referencedTensorBuilder_ = 
+              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
+                 getReferencedTensorFieldBuilder() : null;
+          } else {
+            referencedTensorBuilder_.addAllMessages(other.referencedTensor_);
+          }
+        }
+      }
+      if (other.hasMemoryStats()) {
+        mergeMemoryStats(other.getMemoryStats());
+      }
+      if (other.getAllStartNanos() != 0L) {
+        setAllStartNanos(other.getAllStartNanos());
+      }
+      if (other.getOpStartRelNanos() != 0L) {
+        setOpStartRelNanos(other.getOpStartRelNanos());
+      }
+      if (other.getOpEndRelNanos() != 0L) {
+        setOpEndRelNanos(other.getOpEndRelNanos());
+      }
+      if (other.getAllEndRelNanos() != 0L) {
+        setAllEndRelNanos(other.getAllEndRelNanos());
+      }
+      if (other.getScheduledNanos() != 0L) {
+        setScheduledNanos(other.getScheduledNanos());
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.framework.NodeExecStats parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.framework.NodeExecStats) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+    private int bitField0_;
+
+    private java.lang.Object nodeName_ = "";
+    /**
+     * <pre>
+     * TODO(tucker): Use some more compact form of node identity than
+     * the full string name.  Either all processes should agree on a
+     * global id (cost_id?) for each node, or we should use a hash of
+     * the name.
+     * </pre>
+     *
+     * <code>string node_name = 1;</code>
+     */
+    public java.lang.String getNodeName() {
+      java.lang.Object ref = nodeName_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        nodeName_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * TODO(tucker): Use some more compact form of node identity than
+     * the full string name.  Either all processes should agree on a
+     * global id (cost_id?) for each node, or we should use a hash of
+     * the name.
+     * </pre>
+     *
+     * <code>string node_name = 1;</code>
+     */
+    public com.google.protobuf.ByteString
+        getNodeNameBytes() {
+      java.lang.Object ref = nodeName_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        nodeName_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * TODO(tucker): Use some more compact form of node identity than
+     * the full string name.  Either all processes should agree on a
+     * global id (cost_id?) for each node, or we should use a hash of
+     * the name.
+     * </pre>
+     *
+     * <code>string node_name = 1;</code>
+     */
+    public Builder setNodeName(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      nodeName_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * TODO(tucker): Use some more compact form of node identity than
+     * the full string name.  Either all processes should agree on a
+     * global id (cost_id?) for each node, or we should use a hash of
+     * the name.
+     * </pre>
+     *
+     * <code>string node_name = 1;</code>
+     */
+    public Builder clearNodeName() {
+      
+      nodeName_ = getDefaultInstance().getNodeName();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * TODO(tucker): Use some more compact form of node identity than
+     * the full string name.  Either all processes should agree on a
+     * global id (cost_id?) for each node, or we should use a hash of
+     * the name.
+     * </pre>
+     *
+     * <code>string node_name = 1;</code>
+     */
+    public Builder setNodeNameBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      nodeName_ = value;
+      onChanged();
+      return this;
+    }
+
+    private long allStartMicros_ ;
+    /**
+     * <code>int64 all_start_micros = 2;</code>
+     */
+    public long getAllStartMicros() {
+      return allStartMicros_;
+    }
+    /**
+     * <code>int64 all_start_micros = 2;</code>
+     */
+    public Builder setAllStartMicros(long value) {
+      
+      allStartMicros_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>int64 all_start_micros = 2;</code>
+     */
+    public Builder clearAllStartMicros() {
+      
+      allStartMicros_ = 0L;
+      onChanged();
+      return this;
+    }
+
+    private long opStartRelMicros_ ;
+    /**
+     * <code>int64 op_start_rel_micros = 3;</code>
+     */
+    public long getOpStartRelMicros() {
+      return opStartRelMicros_;
+    }
+    /**
+     * <code>int64 op_start_rel_micros = 3;</code>
+     */
+    public Builder setOpStartRelMicros(long value) {
+      
+      opStartRelMicros_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>int64 op_start_rel_micros = 3;</code>
+     */
+    public Builder clearOpStartRelMicros() {
+      
+      opStartRelMicros_ = 0L;
+      onChanged();
+      return this;
+    }
+
+    private long opEndRelMicros_ ;
+    /**
+     * <code>int64 op_end_rel_micros = 4;</code>
+     */
+    public long getOpEndRelMicros() {
+      return opEndRelMicros_;
+    }
+    /**
+     * <code>int64 op_end_rel_micros = 4;</code>
+     */
+    public Builder setOpEndRelMicros(long value) {
+      
+      opEndRelMicros_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>int64 op_end_rel_micros = 4;</code>
+     */
+    public Builder clearOpEndRelMicros() {
+      
+      opEndRelMicros_ = 0L;
+      onChanged();
+      return this;
+    }
+
+    private long allEndRelMicros_ ;
+    /**
+     * <code>int64 all_end_rel_micros = 5;</code>
+     */
+    public long getAllEndRelMicros() {
+      return allEndRelMicros_;
+    }
+    /**
+     * <code>int64 all_end_rel_micros = 5;</code>
+     */
+    public Builder setAllEndRelMicros(long value) {
+      
+      allEndRelMicros_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>int64 all_end_rel_micros = 5;</code>
+     */
+    public Builder clearAllEndRelMicros() {
+      
+      allEndRelMicros_ = 0L;
+      onChanged();
+      return this;
+    }
+
+    private java.util.List<org.tensorflow.proto.framework.AllocatorMemoryUsed> memory_ =
+      java.util.Collections.emptyList();
+    private void ensureMemoryIsMutable() {
+      if (!((bitField0_ & 0x00000001) != 0)) {
+        memory_ = new java.util.ArrayList<org.tensorflow.proto.framework.AllocatorMemoryUsed>(memory_);
+        bitField0_ |= 0x00000001;
+       }
+    }
+
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.framework.AllocatorMemoryUsed, org.tensorflow.proto.framework.AllocatorMemoryUsed.Builder, org.tensorflow.proto.framework.AllocatorMemoryUsedOrBuilder> memoryBuilder_;
+
+    /**
+     * <code>repeated .tensorflow.AllocatorMemoryUsed memory = 6;</code>
+     */
+    public java.util.List<org.tensorflow.proto.framework.AllocatorMemoryUsed> getMemoryList() {
+      if (memoryBuilder_ == null) {
+        return java.util.Collections.unmodifiableList(memory_);
+      } else {
+        return memoryBuilder_.getMessageList();
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.AllocatorMemoryUsed memory = 6;</code>
+     */
+    public int getMemoryCount() {
+      if (memoryBuilder_ == null) {
+        return memory_.size();
+      } else {
+        return memoryBuilder_.getCount();
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.AllocatorMemoryUsed memory = 6;</code>
+     */
+    public org.tensorflow.proto.framework.AllocatorMemoryUsed getMemory(int index) {
+      if (memoryBuilder_ == null) {
+        return memory_.get(index);
+      } else {
+        return memoryBuilder_.getMessage(index);
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.AllocatorMemoryUsed memory = 6;</code>
+     */
+    public Builder setMemory(
+        int index, org.tensorflow.proto.framework.AllocatorMemoryUsed value) {
+      if (memoryBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureMemoryIsMutable();
+        memory_.set(index, value);
+        onChanged();
+      } else {
+        memoryBuilder_.setMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.AllocatorMemoryUsed memory = 6;</code>
+     */
+    public Builder setMemory(
+        int index, org.tensorflow.proto.framework.AllocatorMemoryUsed.Builder builderForValue) {
+      if (memoryBuilder_ == null) {
+        ensureMemoryIsMutable();
+        memory_.set(index, builderForValue.build());
+        onChanged();
+      } else {
+        memoryBuilder_.setMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.AllocatorMemoryUsed memory = 6;</code>
+     */
+    public Builder addMemory(org.tensorflow.proto.framework.AllocatorMemoryUsed value) {
+      if (memoryBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureMemoryIsMutable();
+        memory_.add(value);
+        onChanged();
+      } else {
+        memoryBuilder_.addMessage(value);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.AllocatorMemoryUsed memory = 6;</code>
+     */
+    public Builder addMemory(
+        int index, org.tensorflow.proto.framework.AllocatorMemoryUsed value) {
+      if (memoryBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureMemoryIsMutable();
+        memory_.add(index, value);
+        onChanged();
+      } else {
+        memoryBuilder_.addMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.AllocatorMemoryUsed memory = 6;</code>
+     */
+    public Builder addMemory(
+        org.tensorflow.proto.framework.AllocatorMemoryUsed.Builder builderForValue) {
+      if (memoryBuilder_ == null) {
+        ensureMemoryIsMutable();
+        memory_.add(builderForValue.build());
+        onChanged();
+      } else {
+        memoryBuilder_.addMessage(builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.AllocatorMemoryUsed memory = 6;</code>
+     */
+    public Builder addMemory(
+        int index, org.tensorflow.proto.framework.AllocatorMemoryUsed.Builder builderForValue) {
+      if (memoryBuilder_ == null) {
+        ensureMemoryIsMutable();
+        memory_.add(index, builderForValue.build());
+        onChanged();
+      } else {
+        memoryBuilder_.addMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.AllocatorMemoryUsed memory = 6;</code>
+     */
+    public Builder addAllMemory(
+        java.lang.Iterable<? extends org.tensorflow.proto.framework.AllocatorMemoryUsed> values) {
+      if (memoryBuilder_ == null) {
+        ensureMemoryIsMutable();
+        com.google.protobuf.AbstractMessageLite.Builder.addAll(
+            values, memory_);
+        onChanged();
+      } else {
+        memoryBuilder_.addAllMessages(values);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.AllocatorMemoryUsed memory = 6;</code>
+     */
+    public Builder clearMemory() {
+      if (memoryBuilder_ == null) {
+        memory_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000001);
+        onChanged();
+      } else {
+        memoryBuilder_.clear();
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.AllocatorMemoryUsed memory = 6;</code>
+     */
+    public Builder removeMemory(int index) {
+      if (memoryBuilder_ == null) {
+        ensureMemoryIsMutable();
+        memory_.remove(index);
+        onChanged();
+      } else {
+        memoryBuilder_.remove(index);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.AllocatorMemoryUsed memory = 6;</code>
+     */
+    public org.tensorflow.proto.framework.AllocatorMemoryUsed.Builder getMemoryBuilder(
+        int index) {
+      return getMemoryFieldBuilder().getBuilder(index);
+    }
+    /**
+     * <code>repeated .tensorflow.AllocatorMemoryUsed memory = 6;</code>
+     */
+    public org.tensorflow.proto.framework.AllocatorMemoryUsedOrBuilder getMemoryOrBuilder(
+        int index) {
+      if (memoryBuilder_ == null) {
+        return memory_.get(index);  } else {
+        return memoryBuilder_.getMessageOrBuilder(index);
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.AllocatorMemoryUsed memory = 6;</code>
+     */
+    public java.util.List<? extends org.tensorflow.proto.framework.AllocatorMemoryUsedOrBuilder> 
+         getMemoryOrBuilderList() {
+      if (memoryBuilder_ != null) {
+        return memoryBuilder_.getMessageOrBuilderList();
+      } else {
+        return java.util.Collections.unmodifiableList(memory_);
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.AllocatorMemoryUsed memory = 6;</code>
+     */
+    public org.tensorflow.proto.framework.AllocatorMemoryUsed.Builder addMemoryBuilder() {
+      return getMemoryFieldBuilder().addBuilder(
+          org.tensorflow.proto.framework.AllocatorMemoryUsed.getDefaultInstance());
+    }
+    /**
+     * <code>repeated .tensorflow.AllocatorMemoryUsed memory = 6;</code>
+     */
+    public org.tensorflow.proto.framework.AllocatorMemoryUsed.Builder addMemoryBuilder(
+        int index) {
+      return getMemoryFieldBuilder().addBuilder(
+          index, org.tensorflow.proto.framework.AllocatorMemoryUsed.getDefaultInstance());
+    }
+    /**
+     * <code>repeated .tensorflow.AllocatorMemoryUsed memory = 6;</code>
+     */
+    public java.util.List<org.tensorflow.proto.framework.AllocatorMemoryUsed.Builder> 
+         getMemoryBuilderList() {
+      return getMemoryFieldBuilder().getBuilderList();
+    }
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.framework.AllocatorMemoryUsed, org.tensorflow.proto.framework.AllocatorMemoryUsed.Builder, org.tensorflow.proto.framework.AllocatorMemoryUsedOrBuilder> 
+        getMemoryFieldBuilder() {
+      if (memoryBuilder_ == null) {
+        memoryBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
+            org.tensorflow.proto.framework.AllocatorMemoryUsed, org.tensorflow.proto.framework.AllocatorMemoryUsed.Builder, org.tensorflow.proto.framework.AllocatorMemoryUsedOrBuilder>(
+                memory_,
+                ((bitField0_ & 0x00000001) != 0),
+                getParentForChildren(),
+                isClean());
+        memory_ = null;
+      }
+      return memoryBuilder_;
+    }
+
+    private java.util.List<org.tensorflow.proto.framework.NodeOutput> output_ =
+      java.util.Collections.emptyList();
+    private void ensureOutputIsMutable() {
+      if (!((bitField0_ & 0x00000002) != 0)) {
+        output_ = new java.util.ArrayList<org.tensorflow.proto.framework.NodeOutput>(output_);
+        bitField0_ |= 0x00000002;
+       }
+    }
+
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.framework.NodeOutput, org.tensorflow.proto.framework.NodeOutput.Builder, org.tensorflow.proto.framework.NodeOutputOrBuilder> outputBuilder_;
+
+    /**
+     * <code>repeated .tensorflow.NodeOutput output = 7;</code>
+     */
+    public java.util.List<org.tensorflow.proto.framework.NodeOutput> getOutputList() {
+      if (outputBuilder_ == null) {
+        return java.util.Collections.unmodifiableList(output_);
+      } else {
+        return outputBuilder_.getMessageList();
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.NodeOutput output = 7;</code>
+     */
+    public int getOutputCount() {
+      if (outputBuilder_ == null) {
+        return output_.size();
+      } else {
+        return outputBuilder_.getCount();
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.NodeOutput output = 7;</code>
+     */
+    public org.tensorflow.proto.framework.NodeOutput getOutput(int index) {
+      if (outputBuilder_ == null) {
+        return output_.get(index);
+      } else {
+        return outputBuilder_.getMessage(index);
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.NodeOutput output = 7;</code>
+     */
+    public Builder setOutput(
+        int index, org.tensorflow.proto.framework.NodeOutput value) {
+      if (outputBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureOutputIsMutable();
+        output_.set(index, value);
+        onChanged();
+      } else {
+        outputBuilder_.setMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.NodeOutput output = 7;</code>
+     */
+    public Builder setOutput(
+        int index, org.tensorflow.proto.framework.NodeOutput.Builder builderForValue) {
+      if (outputBuilder_ == null) {
+        ensureOutputIsMutable();
+        output_.set(index, builderForValue.build());
+        onChanged();
+      } else {
+        outputBuilder_.setMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.NodeOutput output = 7;</code>
+     */
+    public Builder addOutput(org.tensorflow.proto.framework.NodeOutput value) {
+      if (outputBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureOutputIsMutable();
+        output_.add(value);
+        onChanged();
+      } else {
+        outputBuilder_.addMessage(value);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.NodeOutput output = 7;</code>
+     */
+    public Builder addOutput(
+        int index, org.tensorflow.proto.framework.NodeOutput value) {
+      if (outputBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureOutputIsMutable();
+        output_.add(index, value);
+        onChanged();
+      } else {
+        outputBuilder_.addMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.NodeOutput output = 7;</code>
+     */
+    public Builder addOutput(
+        org.tensorflow.proto.framework.NodeOutput.Builder builderForValue) {
+      if (outputBuilder_ == null) {
+        ensureOutputIsMutable();
+        output_.add(builderForValue.build());
+        onChanged();
+      } else {
+        outputBuilder_.addMessage(builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.NodeOutput output = 7;</code>
+     */
+    public Builder addOutput(
+        int index, org.tensorflow.proto.framework.NodeOutput.Builder builderForValue) {
+      if (outputBuilder_ == null) {
+        ensureOutputIsMutable();
+        output_.add(index, builderForValue.build());
+        onChanged();
+      } else {
+        outputBuilder_.addMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.NodeOutput output = 7;</code>
+     */
+    public Builder addAllOutput(
+        java.lang.Iterable<? extends org.tensorflow.proto.framework.NodeOutput> values) {
+      if (outputBuilder_ == null) {
+        ensureOutputIsMutable();
+        com.google.protobuf.AbstractMessageLite.Builder.addAll(
+            values, output_);
+        onChanged();
+      } else {
+        outputBuilder_.addAllMessages(values);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.NodeOutput output = 7;</code>
+     */
+    public Builder clearOutput() {
+      if (outputBuilder_ == null) {
+        output_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000002);
+        onChanged();
+      } else {
+        outputBuilder_.clear();
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.NodeOutput output = 7;</code>
+     */
+    public Builder removeOutput(int index) {
+      if (outputBuilder_ == null) {
+        ensureOutputIsMutable();
+        output_.remove(index);
+        onChanged();
+      } else {
+        outputBuilder_.remove(index);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.NodeOutput output = 7;</code>
+     */
+    public org.tensorflow.proto.framework.NodeOutput.Builder getOutputBuilder(
+        int index) {
+      return getOutputFieldBuilder().getBuilder(index);
+    }
+    /**
+     * <code>repeated .tensorflow.NodeOutput output = 7;</code>
+     */
+    public org.tensorflow.proto.framework.NodeOutputOrBuilder getOutputOrBuilder(
+        int index) {
+      if (outputBuilder_ == null) {
+        return output_.get(index);  } else {
+        return outputBuilder_.getMessageOrBuilder(index);
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.NodeOutput output = 7;</code>
+     */
+    public java.util.List<? extends org.tensorflow.proto.framework.NodeOutputOrBuilder> 
+         getOutputOrBuilderList() {
+      if (outputBuilder_ != null) {
+        return outputBuilder_.getMessageOrBuilderList();
+      } else {
+        return java.util.Collections.unmodifiableList(output_);
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.NodeOutput output = 7;</code>
+     */
+    public org.tensorflow.proto.framework.NodeOutput.Builder addOutputBuilder() {
+      return getOutputFieldBuilder().addBuilder(
+          org.tensorflow.proto.framework.NodeOutput.getDefaultInstance());
+    }
+    /**
+     * <code>repeated .tensorflow.NodeOutput output = 7;</code>
+     */
+    public org.tensorflow.proto.framework.NodeOutput.Builder addOutputBuilder(
+        int index) {
+      return getOutputFieldBuilder().addBuilder(
+          index, org.tensorflow.proto.framework.NodeOutput.getDefaultInstance());
+    }
+    /**
+     * <code>repeated .tensorflow.NodeOutput output = 7;</code>
+     */
+    public java.util.List<org.tensorflow.proto.framework.NodeOutput.Builder> 
+         getOutputBuilderList() {
+      return getOutputFieldBuilder().getBuilderList();
+    }
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.framework.NodeOutput, org.tensorflow.proto.framework.NodeOutput.Builder, org.tensorflow.proto.framework.NodeOutputOrBuilder> 
+        getOutputFieldBuilder() {
+      if (outputBuilder_ == null) {
+        outputBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
+            org.tensorflow.proto.framework.NodeOutput, org.tensorflow.proto.framework.NodeOutput.Builder, org.tensorflow.proto.framework.NodeOutputOrBuilder>(
+                output_,
+                ((bitField0_ & 0x00000002) != 0),
+                getParentForChildren(),
+                isClean());
+        output_ = null;
+      }
+      return outputBuilder_;
+    }
+
+    private java.lang.Object timelineLabel_ = "";
+    /**
+     * <code>string timeline_label = 8;</code>
+     */
+    public java.lang.String getTimelineLabel() {
+      java.lang.Object ref = timelineLabel_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        timelineLabel_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <code>string timeline_label = 8;</code>
+     */
+    public com.google.protobuf.ByteString
+        getTimelineLabelBytes() {
+      java.lang.Object ref = timelineLabel_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        timelineLabel_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <code>string timeline_label = 8;</code>
+     */
+    public Builder setTimelineLabel(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      timelineLabel_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>string timeline_label = 8;</code>
+     */
+    public Builder clearTimelineLabel() {
+      
+      timelineLabel_ = getDefaultInstance().getTimelineLabel();
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>string timeline_label = 8;</code>
+     */
+    public Builder setTimelineLabelBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      timelineLabel_ = value;
+      onChanged();
+      return this;
+    }
+
+    private long scheduledMicros_ ;
+    /**
+     * <code>int64 scheduled_micros = 9;</code>
+     */
+    public long getScheduledMicros() {
+      return scheduledMicros_;
+    }
+    /**
+     * <code>int64 scheduled_micros = 9;</code>
+     */
+    public Builder setScheduledMicros(long value) {
+      
+      scheduledMicros_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>int64 scheduled_micros = 9;</code>
+     */
+    public Builder clearScheduledMicros() {
+      
+      scheduledMicros_ = 0L;
+      onChanged();
+      return this;
+    }
+
+    private int threadId_ ;
+    /**
+     * <code>uint32 thread_id = 10;</code>
+     */
+    public int getThreadId() {
+      return threadId_;
+    }
+    /**
+     * <code>uint32 thread_id = 10;</code>
+     */
+    public Builder setThreadId(int value) {
+      
+      threadId_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>uint32 thread_id = 10;</code>
+     */
+    public Builder clearThreadId() {
+      
+      threadId_ = 0;
+      onChanged();
+      return this;
+    }
+
+    private java.util.List<org.tensorflow.proto.framework.AllocationDescription> referencedTensor_ =
+      java.util.Collections.emptyList();
+    private void ensureReferencedTensorIsMutable() {
+      if (!((bitField0_ & 0x00000004) != 0)) {
+        referencedTensor_ = new java.util.ArrayList<org.tensorflow.proto.framework.AllocationDescription>(referencedTensor_);
+        bitField0_ |= 0x00000004;
+       }
+    }
+
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.framework.AllocationDescription, org.tensorflow.proto.framework.AllocationDescription.Builder, org.tensorflow.proto.framework.AllocationDescriptionOrBuilder> referencedTensorBuilder_;
+
+    /**
+     * <code>repeated .tensorflow.AllocationDescription referenced_tensor = 11;</code>
+     */
+    public java.util.List<org.tensorflow.proto.framework.AllocationDescription> getReferencedTensorList() {
+      if (referencedTensorBuilder_ == null) {
+        return java.util.Collections.unmodifiableList(referencedTensor_);
+      } else {
+        return referencedTensorBuilder_.getMessageList();
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.AllocationDescription referenced_tensor = 11;</code>
+     */
+    public int getReferencedTensorCount() {
+      if (referencedTensorBuilder_ == null) {
+        return referencedTensor_.size();
+      } else {
+        return referencedTensorBuilder_.getCount();
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.AllocationDescription referenced_tensor = 11;</code>
+     */
+    public org.tensorflow.proto.framework.AllocationDescription getReferencedTensor(int index) {
+      if (referencedTensorBuilder_ == null) {
+        return referencedTensor_.get(index);
+      } else {
+        return referencedTensorBuilder_.getMessage(index);
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.AllocationDescription referenced_tensor = 11;</code>
+     */
+    public Builder setReferencedTensor(
+        int index, org.tensorflow.proto.framework.AllocationDescription value) {
+      if (referencedTensorBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureReferencedTensorIsMutable();
+        referencedTensor_.set(index, value);
+        onChanged();
+      } else {
+        referencedTensorBuilder_.setMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.AllocationDescription referenced_tensor = 11;</code>
+     */
+    public Builder setReferencedTensor(
+        int index, org.tensorflow.proto.framework.AllocationDescription.Builder builderForValue) {
+      if (referencedTensorBuilder_ == null) {
+        ensureReferencedTensorIsMutable();
+        referencedTensor_.set(index, builderForValue.build());
+        onChanged();
+      } else {
+        referencedTensorBuilder_.setMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.AllocationDescription referenced_tensor = 11;</code>
+     */
+    public Builder addReferencedTensor(org.tensorflow.proto.framework.AllocationDescription value) {
+      if (referencedTensorBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureReferencedTensorIsMutable();
+        referencedTensor_.add(value);
+        onChanged();
+      } else {
+        referencedTensorBuilder_.addMessage(value);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.AllocationDescription referenced_tensor = 11;</code>
+     */
+    public Builder addReferencedTensor(
+        int index, org.tensorflow.proto.framework.AllocationDescription value) {
+      if (referencedTensorBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureReferencedTensorIsMutable();
+        referencedTensor_.add(index, value);
+        onChanged();
+      } else {
+        referencedTensorBuilder_.addMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.AllocationDescription referenced_tensor = 11;</code>
+     */
+    public Builder addReferencedTensor(
+        org.tensorflow.proto.framework.AllocationDescription.Builder builderForValue) {
+      if (referencedTensorBuilder_ == null) {
+        ensureReferencedTensorIsMutable();
+        referencedTensor_.add(builderForValue.build());
+        onChanged();
+      } else {
+        referencedTensorBuilder_.addMessage(builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.AllocationDescription referenced_tensor = 11;</code>
+     */
+    public Builder addReferencedTensor(
+        int index, org.tensorflow.proto.framework.AllocationDescription.Builder builderForValue) {
+      if (referencedTensorBuilder_ == null) {
+        ensureReferencedTensorIsMutable();
+        referencedTensor_.add(index, builderForValue.build());
+        onChanged();
+      } else {
+        referencedTensorBuilder_.addMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.AllocationDescription referenced_tensor = 11;</code>
+     */
+    public Builder addAllReferencedTensor(
+        java.lang.Iterable<? extends org.tensorflow.proto.framework.AllocationDescription> values) {
+      if (referencedTensorBuilder_ == null) {
+        ensureReferencedTensorIsMutable();
+        com.google.protobuf.AbstractMessageLite.Builder.addAll(
+            values, referencedTensor_);
+        onChanged();
+      } else {
+        referencedTensorBuilder_.addAllMessages(values);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.AllocationDescription referenced_tensor = 11;</code>
+     */
+    public Builder clearReferencedTensor() {
+      if (referencedTensorBuilder_ == null) {
+        referencedTensor_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000004);
+        onChanged();
+      } else {
+        referencedTensorBuilder_.clear();
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.AllocationDescription referenced_tensor = 11;</code>
+     */
+    public Builder removeReferencedTensor(int index) {
+      if (referencedTensorBuilder_ == null) {
+        ensureReferencedTensorIsMutable();
+        referencedTensor_.remove(index);
+        onChanged();
+      } else {
+        referencedTensorBuilder_.remove(index);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.AllocationDescription referenced_tensor = 11;</code>
+     */
+    public org.tensorflow.proto.framework.AllocationDescription.Builder getReferencedTensorBuilder(
+        int index) {
+      return getReferencedTensorFieldBuilder().getBuilder(index);
+    }
+    /**
+     * <code>repeated .tensorflow.AllocationDescription referenced_tensor = 11;</code>
+     */
+    public org.tensorflow.proto.framework.AllocationDescriptionOrBuilder getReferencedTensorOrBuilder(
+        int index) {
+      if (referencedTensorBuilder_ == null) {
+        return referencedTensor_.get(index);  } else {
+        return referencedTensorBuilder_.getMessageOrBuilder(index);
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.AllocationDescription referenced_tensor = 11;</code>
+     */
+    public java.util.List<? extends org.tensorflow.proto.framework.AllocationDescriptionOrBuilder> 
+         getReferencedTensorOrBuilderList() {
+      if (referencedTensorBuilder_ != null) {
+        return referencedTensorBuilder_.getMessageOrBuilderList();
+      } else {
+        return java.util.Collections.unmodifiableList(referencedTensor_);
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.AllocationDescription referenced_tensor = 11;</code>
+     */
+    public org.tensorflow.proto.framework.AllocationDescription.Builder addReferencedTensorBuilder() {
+      return getReferencedTensorFieldBuilder().addBuilder(
+          org.tensorflow.proto.framework.AllocationDescription.getDefaultInstance());
+    }
+    /**
+     * <code>repeated .tensorflow.AllocationDescription referenced_tensor = 11;</code>
+     */
+    public org.tensorflow.proto.framework.AllocationDescription.Builder addReferencedTensorBuilder(
+        int index) {
+      return getReferencedTensorFieldBuilder().addBuilder(
+          index, org.tensorflow.proto.framework.AllocationDescription.getDefaultInstance());
+    }
+    /**
+     * <code>repeated .tensorflow.AllocationDescription referenced_tensor = 11;</code>
+     */
+    public java.util.List<org.tensorflow.proto.framework.AllocationDescription.Builder> 
+         getReferencedTensorBuilderList() {
+      return getReferencedTensorFieldBuilder().getBuilderList();
+    }
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.framework.AllocationDescription, org.tensorflow.proto.framework.AllocationDescription.Builder, org.tensorflow.proto.framework.AllocationDescriptionOrBuilder> 
+        getReferencedTensorFieldBuilder() {
+      if (referencedTensorBuilder_ == null) {
+        referencedTensorBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
+            org.tensorflow.proto.framework.AllocationDescription, org.tensorflow.proto.framework.AllocationDescription.Builder, org.tensorflow.proto.framework.AllocationDescriptionOrBuilder>(
+                referencedTensor_,
+                ((bitField0_ & 0x00000004) != 0),
+                getParentForChildren(),
+                isClean());
+        referencedTensor_ = null;
+      }
+      return referencedTensorBuilder_;
+    }
+
+    private org.tensorflow.proto.framework.MemoryStats memoryStats_;
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.MemoryStats, org.tensorflow.proto.framework.MemoryStats.Builder, org.tensorflow.proto.framework.MemoryStatsOrBuilder> memoryStatsBuilder_;
+    /**
+     * <code>.tensorflow.MemoryStats memory_stats = 12;</code>
+     */
+    public boolean hasMemoryStats() {
+      return memoryStatsBuilder_ != null || memoryStats_ != null;
+    }
+    /**
+     * <code>.tensorflow.MemoryStats memory_stats = 12;</code>
+     */
+    public org.tensorflow.proto.framework.MemoryStats getMemoryStats() {
+      if (memoryStatsBuilder_ == null) {
+        return memoryStats_ == null ? org.tensorflow.proto.framework.MemoryStats.getDefaultInstance() : memoryStats_;
+      } else {
+        return memoryStatsBuilder_.getMessage();
+      }
+    }
+    /**
+     * <code>.tensorflow.MemoryStats memory_stats = 12;</code>
+     */
+    public Builder setMemoryStats(org.tensorflow.proto.framework.MemoryStats value) {
+      if (memoryStatsBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        memoryStats_ = value;
+        onChanged();
+      } else {
+        memoryStatsBuilder_.setMessage(value);
+      }
+
+      return this;
+    }
+    /**
+     * <code>.tensorflow.MemoryStats memory_stats = 12;</code>
+     */
+    public Builder setMemoryStats(
+        org.tensorflow.proto.framework.MemoryStats.Builder builderForValue) {
+      if (memoryStatsBuilder_ == null) {
+        memoryStats_ = builderForValue.build();
+        onChanged();
+      } else {
+        memoryStatsBuilder_.setMessage(builderForValue.build());
+      }
+
+      return this;
+    }
+    /**
+     * <code>.tensorflow.MemoryStats memory_stats = 12;</code>
+     */
+    public Builder mergeMemoryStats(org.tensorflow.proto.framework.MemoryStats value) {
+      if (memoryStatsBuilder_ == null) {
+        if (memoryStats_ != null) {
+          memoryStats_ =
+            org.tensorflow.proto.framework.MemoryStats.newBuilder(memoryStats_).mergeFrom(value).buildPartial();
+        } else {
+          memoryStats_ = value;
+        }
+        onChanged();
+      } else {
+        memoryStatsBuilder_.mergeFrom(value);
+      }
+
+      return this;
+    }
+    /**
+     * <code>.tensorflow.MemoryStats memory_stats = 12;</code>
+     */
+    public Builder clearMemoryStats() {
+      if (memoryStatsBuilder_ == null) {
+        memoryStats_ = null;
+        onChanged();
+      } else {
+        memoryStats_ = null;
+        memoryStatsBuilder_ = null;
+      }
+
+      return this;
+    }
+    /**
+     * <code>.tensorflow.MemoryStats memory_stats = 12;</code>
+     */
+    public org.tensorflow.proto.framework.MemoryStats.Builder getMemoryStatsBuilder() {
+      
+      onChanged();
+      return getMemoryStatsFieldBuilder().getBuilder();
+    }
+    /**
+     * <code>.tensorflow.MemoryStats memory_stats = 12;</code>
+     */
+    public org.tensorflow.proto.framework.MemoryStatsOrBuilder getMemoryStatsOrBuilder() {
+      if (memoryStatsBuilder_ != null) {
+        return memoryStatsBuilder_.getMessageOrBuilder();
+      } else {
+        return memoryStats_ == null ?
+            org.tensorflow.proto.framework.MemoryStats.getDefaultInstance() : memoryStats_;
+      }
+    }
+    /**
+     * <code>.tensorflow.MemoryStats memory_stats = 12;</code>
+     */
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.MemoryStats, org.tensorflow.proto.framework.MemoryStats.Builder, org.tensorflow.proto.framework.MemoryStatsOrBuilder> 
+        getMemoryStatsFieldBuilder() {
+      if (memoryStatsBuilder_ == null) {
+        memoryStatsBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+            org.tensorflow.proto.framework.MemoryStats, org.tensorflow.proto.framework.MemoryStats.Builder, org.tensorflow.proto.framework.MemoryStatsOrBuilder>(
+                getMemoryStats(),
+                getParentForChildren(),
+                isClean());
+        memoryStats_ = null;
+      }
+      return memoryStatsBuilder_;
+    }
+
+    private long allStartNanos_ ;
+    /**
+     * <code>int64 all_start_nanos = 13;</code>
+     */
+    public long getAllStartNanos() {
+      return allStartNanos_;
+    }
+    /**
+     * <code>int64 all_start_nanos = 13;</code>
+     */
+    public Builder setAllStartNanos(long value) {
+      
+      allStartNanos_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>int64 all_start_nanos = 13;</code>
+     */
+    public Builder clearAllStartNanos() {
+      
+      allStartNanos_ = 0L;
+      onChanged();
+      return this;
+    }
+
+    private long opStartRelNanos_ ;
+    /**
+     * <code>int64 op_start_rel_nanos = 14;</code>
+     */
+    public long getOpStartRelNanos() {
+      return opStartRelNanos_;
+    }
+    /**
+     * <code>int64 op_start_rel_nanos = 14;</code>
+     */
+    public Builder setOpStartRelNanos(long value) {
+      
+      opStartRelNanos_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>int64 op_start_rel_nanos = 14;</code>
+     */
+    public Builder clearOpStartRelNanos() {
+      
+      opStartRelNanos_ = 0L;
+      onChanged();
+      return this;
+    }
+
+    private long opEndRelNanos_ ;
+    /**
+     * <code>int64 op_end_rel_nanos = 15;</code>
+     */
+    public long getOpEndRelNanos() {
+      return opEndRelNanos_;
+    }
+    /**
+     * <code>int64 op_end_rel_nanos = 15;</code>
+     */
+    public Builder setOpEndRelNanos(long value) {
+      
+      opEndRelNanos_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>int64 op_end_rel_nanos = 15;</code>
+     */
+    public Builder clearOpEndRelNanos() {
+      
+      opEndRelNanos_ = 0L;
+      onChanged();
+      return this;
+    }
+
+    private long allEndRelNanos_ ;
+    /**
+     * <code>int64 all_end_rel_nanos = 16;</code>
+     */
+    public long getAllEndRelNanos() {
+      return allEndRelNanos_;
+    }
+    /**
+     * <code>int64 all_end_rel_nanos = 16;</code>
+     */
+    public Builder setAllEndRelNanos(long value) {
+      
+      allEndRelNanos_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>int64 all_end_rel_nanos = 16;</code>
+     */
+    public Builder clearAllEndRelNanos() {
+      
+      allEndRelNanos_ = 0L;
+      onChanged();
+      return this;
+    }
+
+    private long scheduledNanos_ ;
+    /**
+     * <code>int64 scheduled_nanos = 17;</code>
+     */
+    public long getScheduledNanos() {
+      return scheduledNanos_;
+    }
+    /**
+     * <code>int64 scheduled_nanos = 17;</code>
+     */
+    public Builder setScheduledNanos(long value) {
+      
+      scheduledNanos_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>int64 scheduled_nanos = 17;</code>
+     */
+    public Builder clearScheduledNanos() {
+      
+      scheduledNanos_ = 0L;
+      onChanged();
+      return this;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.NodeExecStats)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.NodeExecStats)
+  private static final org.tensorflow.proto.framework.NodeExecStats DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.NodeExecStats();
+  }
+
+  public static org.tensorflow.proto.framework.NodeExecStats getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<NodeExecStats>
+      PARSER = new com.google.protobuf.AbstractParser<NodeExecStats>() {
+    @java.lang.Override
+    public NodeExecStats parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new NodeExecStats(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<NodeExecStats> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<NodeExecStats> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.framework.NodeExecStats getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NodeExecStatsOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NodeExecStatsOrBuilder.java
new file mode 100644
index 00000000000..60be2e49270
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NodeExecStatsOrBuilder.java
@@ -0,0 +1,183 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/framework/step_stats.proto
+
+package org.tensorflow.proto.framework;
+
+public interface NodeExecStatsOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.NodeExecStats)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <pre>
+   * TODO(tucker): Use some more compact form of node identity than
+   * the full string name.  Either all processes should agree on a
+   * global id (cost_id?) for each node, or we should use a hash of
+   * the name.
+   * </pre>
+   *
+   * <code>string node_name = 1;</code>
+   */
+  java.lang.String getNodeName();
+  /**
+   * <pre>
+   * TODO(tucker): Use some more compact form of node identity than
+   * the full string name.  Either all processes should agree on a
+   * global id (cost_id?) for each node, or we should use a hash of
+   * the name.
+   * </pre>
+   *
+   * <code>string node_name = 1;</code>
+   */
+  com.google.protobuf.ByteString
+      getNodeNameBytes();
+
+  /**
+   * <code>int64 all_start_micros = 2;</code>
+   */
+  long getAllStartMicros();
+
+  /**
+   * <code>int64 op_start_rel_micros = 3;</code>
+   */
+  long getOpStartRelMicros();
+
+  /**
+   * <code>int64 op_end_rel_micros = 4;</code>
+   */
+  long getOpEndRelMicros();
+
+  /**
+   * <code>int64 all_end_rel_micros = 5;</code>
+   */
+  long getAllEndRelMicros();
+
+  /**
+   * <code>repeated .tensorflow.AllocatorMemoryUsed memory = 6;</code>
+   */
+  java.util.List<org.tensorflow.proto.framework.AllocatorMemoryUsed> 
+      getMemoryList();
+  /**
+   * <code>repeated .tensorflow.AllocatorMemoryUsed memory = 6;</code>
+   */
+  org.tensorflow.proto.framework.AllocatorMemoryUsed getMemory(int index);
+  /**
+   * <code>repeated .tensorflow.AllocatorMemoryUsed memory = 6;</code>
+   */
+  int getMemoryCount();
+  /**
+   * <code>repeated .tensorflow.AllocatorMemoryUsed memory = 6;</code>
+   */
+  java.util.List<? extends org.tensorflow.proto.framework.AllocatorMemoryUsedOrBuilder> 
+      getMemoryOrBuilderList();
+  /**
+   * <code>repeated .tensorflow.AllocatorMemoryUsed memory = 6;</code>
+   */
+  org.tensorflow.proto.framework.AllocatorMemoryUsedOrBuilder getMemoryOrBuilder(
+      int index);
+
+  /**
+   * <code>repeated .tensorflow.NodeOutput output = 7;</code>
+   */
+  java.util.List<org.tensorflow.proto.framework.NodeOutput> 
+      getOutputList();
+  /**
+   * <code>repeated .tensorflow.NodeOutput output = 7;</code>
+   */
+  org.tensorflow.proto.framework.NodeOutput getOutput(int index);
+  /**
+   * <code>repeated .tensorflow.NodeOutput output = 7;</code>
+   */
+  int getOutputCount();
+  /**
+   * <code>repeated .tensorflow.NodeOutput output = 7;</code>
+   */
+  java.util.List<? extends org.tensorflow.proto.framework.NodeOutputOrBuilder> 
+      getOutputOrBuilderList();
+  /**
+   * <code>repeated .tensorflow.NodeOutput output = 7;</code>
+   */
+  org.tensorflow.proto.framework.NodeOutputOrBuilder getOutputOrBuilder(
+      int index);
+
+  /**
+   * <code>string timeline_label = 8;</code>
+   */
+  java.lang.String getTimelineLabel();
+  /**
+   * <code>string timeline_label = 8;</code>
+   */
+  com.google.protobuf.ByteString
+      getTimelineLabelBytes();
+
+  /**
+   * <code>int64 scheduled_micros = 9;</code>
+   */
+  long getScheduledMicros();
+
+  /**
+   * <code>uint32 thread_id = 10;</code>
+   */
+  int getThreadId();
+
+  /**
+   * <code>repeated .tensorflow.AllocationDescription referenced_tensor = 11;</code>
+   */
+  java.util.List<org.tensorflow.proto.framework.AllocationDescription> 
+      getReferencedTensorList();
+  /**
+   * <code>repeated .tensorflow.AllocationDescription referenced_tensor = 11;</code>
+   */
+  org.tensorflow.proto.framework.AllocationDescription getReferencedTensor(int index);
+  /**
+   * <code>repeated .tensorflow.AllocationDescription referenced_tensor = 11;</code>
+   */
+  int getReferencedTensorCount();
+  /**
+   * <code>repeated .tensorflow.AllocationDescription referenced_tensor = 11;</code>
+   */
+  java.util.List<? extends org.tensorflow.proto.framework.AllocationDescriptionOrBuilder> 
+      getReferencedTensorOrBuilderList();
+  /**
+   * <code>repeated .tensorflow.AllocationDescription referenced_tensor = 11;</code>
+   */
+  org.tensorflow.proto.framework.AllocationDescriptionOrBuilder getReferencedTensorOrBuilder(
+      int index);
+
+  /**
+   * <code>.tensorflow.MemoryStats memory_stats = 12;</code>
+   */
+  boolean hasMemoryStats();
+  /**
+   * <code>.tensorflow.MemoryStats memory_stats = 12;</code>
+   */
+  org.tensorflow.proto.framework.MemoryStats getMemoryStats();
+  /**
+   * <code>.tensorflow.MemoryStats memory_stats = 12;</code>
+   */
+  org.tensorflow.proto.framework.MemoryStatsOrBuilder getMemoryStatsOrBuilder();
+
+  /**
+   * <code>int64 all_start_nanos = 13;</code>
+   */
+  long getAllStartNanos();
+
+  /**
+   * <code>int64 op_start_rel_nanos = 14;</code>
+   */
+  long getOpStartRelNanos();
+
+  /**
+   * <code>int64 op_end_rel_nanos = 15;</code>
+   */
+  long getOpEndRelNanos();
+
+  /**
+   * <code>int64 all_end_rel_nanos = 16;</code>
+   */
+  long getAllEndRelNanos();
+
+  /**
+   * <code>int64 scheduled_nanos = 17;</code>
+   */
+  long getScheduledNanos();
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NodeOutput.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NodeOutput.java
new file mode 100644
index 00000000000..c6e38aaec46
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NodeOutput.java
@@ -0,0 +1,665 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/framework/step_stats.proto
+
+package org.tensorflow.proto.framework;
+
+/**
+ * <pre>
+ * Output sizes recorded for a single execution of a graph node.
+ * </pre>
+ *
+ * Protobuf type {@code tensorflow.NodeOutput}
+ */
+public  final class NodeOutput extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.NodeOutput)
+    NodeOutputOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use NodeOutput.newBuilder() to construct.
+  private NodeOutput(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private NodeOutput() {
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new NodeOutput();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private NodeOutput(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 8: {
+
+            slot_ = input.readInt32();
+            break;
+          }
+          case 26: {
+            org.tensorflow.proto.framework.TensorDescription.Builder subBuilder = null;
+            if (tensorDescription_ != null) {
+              subBuilder = tensorDescription_.toBuilder();
+            }
+            tensorDescription_ = input.readMessage(org.tensorflow.proto.framework.TensorDescription.parser(), extensionRegistry);
+            if (subBuilder != null) {
+              subBuilder.mergeFrom(tensorDescription_);
+              tensorDescription_ = subBuilder.buildPartial();
+            }
+
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.framework.StepStatsProtos.internal_static_tensorflow_NodeOutput_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.framework.StepStatsProtos.internal_static_tensorflow_NodeOutput_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.framework.NodeOutput.class, org.tensorflow.proto.framework.NodeOutput.Builder.class);
+  }
+
+  public static final int SLOT_FIELD_NUMBER = 1;
+  private int slot_;
+  /**
+   * <code>int32 slot = 1;</code>
+   */
+  public int getSlot() {
+    return slot_;
+  }
+
+  public static final int TENSOR_DESCRIPTION_FIELD_NUMBER = 3;
+  private org.tensorflow.proto.framework.TensorDescription tensorDescription_;
+  /**
+   * <code>.tensorflow.TensorDescription tensor_description = 3;</code>
+   */
+  public boolean hasTensorDescription() {
+    return tensorDescription_ != null;
+  }
+  /**
+   * <code>.tensorflow.TensorDescription tensor_description = 3;</code>
+   */
+  public org.tensorflow.proto.framework.TensorDescription getTensorDescription() {
+    return tensorDescription_ == null ? org.tensorflow.proto.framework.TensorDescription.getDefaultInstance() : tensorDescription_;
+  }
+  /**
+   * <code>.tensorflow.TensorDescription tensor_description = 3;</code>
+   */
+  public org.tensorflow.proto.framework.TensorDescriptionOrBuilder getTensorDescriptionOrBuilder() {
+    return getTensorDescription();
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    if (slot_ != 0) {
+      output.writeInt32(1, slot_);
+    }
+    if (tensorDescription_ != null) {
+      output.writeMessage(3, getTensorDescription());
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (slot_ != 0) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt32Size(1, slot_);
+    }
+    if (tensorDescription_ != null) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(3, getTensorDescription());
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.framework.NodeOutput)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.framework.NodeOutput other = (org.tensorflow.proto.framework.NodeOutput) obj;
+
+    if (getSlot()
+        != other.getSlot()) return false;
+    if (hasTensorDescription() != other.hasTensorDescription()) return false;
+    if (hasTensorDescription()) {
+      if (!getTensorDescription()
+          .equals(other.getTensorDescription())) return false;
+    }
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    hash = (37 * hash) + SLOT_FIELD_NUMBER;
+    hash = (53 * hash) + getSlot();
+    if (hasTensorDescription()) {
+      hash = (37 * hash) + TENSOR_DESCRIPTION_FIELD_NUMBER;
+      hash = (53 * hash) + getTensorDescription().hashCode();
+    }
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.framework.NodeOutput parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.NodeOutput parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.NodeOutput parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.NodeOutput parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.NodeOutput parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.NodeOutput parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.NodeOutput parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.NodeOutput parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.NodeOutput parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.NodeOutput parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.NodeOutput parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.NodeOutput parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.framework.NodeOutput prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * <pre>
+   * Output sizes recorded for a single execution of a graph node.
+   * </pre>
+   *
+   * Protobuf type {@code tensorflow.NodeOutput}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.NodeOutput)
+      org.tensorflow.proto.framework.NodeOutputOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.StepStatsProtos.internal_static_tensorflow_NodeOutput_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.StepStatsProtos.internal_static_tensorflow_NodeOutput_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.NodeOutput.class, org.tensorflow.proto.framework.NodeOutput.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.framework.NodeOutput.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      slot_ = 0;
+
+      if (tensorDescriptionBuilder_ == null) {
+        tensorDescription_ = null;
+      } else {
+        tensorDescription_ = null;
+        tensorDescriptionBuilder_ = null;
+      }
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.framework.StepStatsProtos.internal_static_tensorflow_NodeOutput_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.NodeOutput getDefaultInstanceForType() {
+      return org.tensorflow.proto.framework.NodeOutput.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.NodeOutput build() {
+      org.tensorflow.proto.framework.NodeOutput result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.NodeOutput buildPartial() {
+      org.tensorflow.proto.framework.NodeOutput result = new org.tensorflow.proto.framework.NodeOutput(this);
+      result.slot_ = slot_;
+      if (tensorDescriptionBuilder_ == null) {
+        result.tensorDescription_ = tensorDescription_;
+      } else {
+        result.tensorDescription_ = tensorDescriptionBuilder_.build();
+      }
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.framework.NodeOutput) {
+        return mergeFrom((org.tensorflow.proto.framework.NodeOutput)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.framework.NodeOutput other) {
+      if (other == org.tensorflow.proto.framework.NodeOutput.getDefaultInstance()) return this;
+      if (other.getSlot() != 0) {
+        setSlot(other.getSlot());
+      }
+      if (other.hasTensorDescription()) {
+        mergeTensorDescription(other.getTensorDescription());
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.framework.NodeOutput parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.framework.NodeOutput) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+
+    private int slot_ ;
+    /**
+     * <code>int32 slot = 1;</code>
+     */
+    public int getSlot() {
+      return slot_;
+    }
+    /**
+     * <code>int32 slot = 1;</code>
+     */
+    public Builder setSlot(int value) {
+      
+      slot_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>int32 slot = 1;</code>
+     */
+    public Builder clearSlot() {
+      
+      slot_ = 0;
+      onChanged();
+      return this;
+    }
+
+    private org.tensorflow.proto.framework.TensorDescription tensorDescription_;
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.TensorDescription, org.tensorflow.proto.framework.TensorDescription.Builder, org.tensorflow.proto.framework.TensorDescriptionOrBuilder> tensorDescriptionBuilder_;
+    /**
+     * <code>.tensorflow.TensorDescription tensor_description = 3;</code>
+     */
+    public boolean hasTensorDescription() {
+      return tensorDescriptionBuilder_ != null || tensorDescription_ != null;
+    }
+    /**
+     * <code>.tensorflow.TensorDescription tensor_description = 3;</code>
+     */
+    public org.tensorflow.proto.framework.TensorDescription getTensorDescription() {
+      if (tensorDescriptionBuilder_ == null) {
+        return tensorDescription_ == null ? org.tensorflow.proto.framework.TensorDescription.getDefaultInstance() : tensorDescription_;
+      } else {
+        return tensorDescriptionBuilder_.getMessage();
+      }
+    }
+    /**
+     * <code>.tensorflow.TensorDescription tensor_description = 3;</code>
+     */
+    public Builder setTensorDescription(org.tensorflow.proto.framework.TensorDescription value) {
+      if (tensorDescriptionBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        tensorDescription_ = value;
+        onChanged();
+      } else {
+        tensorDescriptionBuilder_.setMessage(value);
+      }
+
+      return this;
+    }
+    /**
+     * <code>.tensorflow.TensorDescription tensor_description = 3;</code>
+     */
+    public Builder setTensorDescription(
+        org.tensorflow.proto.framework.TensorDescription.Builder builderForValue) {
+      if (tensorDescriptionBuilder_ == null) {
+        tensorDescription_ = builderForValue.build();
+        onChanged();
+      } else {
+        tensorDescriptionBuilder_.setMessage(builderForValue.build());
+      }
+
+      return this;
+    }
+    /**
+     * <code>.tensorflow.TensorDescription tensor_description = 3;</code>
+     */
+    public Builder mergeTensorDescription(org.tensorflow.proto.framework.TensorDescription value) {
+      if (tensorDescriptionBuilder_ == null) {
+        if (tensorDescription_ != null) {
+          tensorDescription_ =
+            org.tensorflow.proto.framework.TensorDescription.newBuilder(tensorDescription_).mergeFrom(value).buildPartial();
+        } else {
+          tensorDescription_ = value;
+        }
+        onChanged();
+      } else {
+        tensorDescriptionBuilder_.mergeFrom(value);
+      }
+
+      return this;
+    }
+    /**
+     * <code>.tensorflow.TensorDescription tensor_description = 3;</code>
+     */
+    public Builder clearTensorDescription() {
+      if (tensorDescriptionBuilder_ == null) {
+        tensorDescription_ = null;
+        onChanged();
+      } else {
+        tensorDescription_ = null;
+        tensorDescriptionBuilder_ = null;
+      }
+
+      return this;
+    }
+    /**
+     * <code>.tensorflow.TensorDescription tensor_description = 3;</code>
+     */
+    public org.tensorflow.proto.framework.TensorDescription.Builder getTensorDescriptionBuilder() {
+      
+      onChanged();
+      return getTensorDescriptionFieldBuilder().getBuilder();
+    }
+    /**
+     * <code>.tensorflow.TensorDescription tensor_description = 3;</code>
+     */
+    public org.tensorflow.proto.framework.TensorDescriptionOrBuilder getTensorDescriptionOrBuilder() {
+      if (tensorDescriptionBuilder_ != null) {
+        return tensorDescriptionBuilder_.getMessageOrBuilder();
+      } else {
+        return tensorDescription_ == null ?
+            org.tensorflow.proto.framework.TensorDescription.getDefaultInstance() : tensorDescription_;
+      }
+    }
+    /**
+     * <code>.tensorflow.TensorDescription tensor_description = 3;</code>
+     */
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.TensorDescription, org.tensorflow.proto.framework.TensorDescription.Builder, org.tensorflow.proto.framework.TensorDescriptionOrBuilder> 
+        getTensorDescriptionFieldBuilder() {
+      if (tensorDescriptionBuilder_ == null) {
+        tensorDescriptionBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+            org.tensorflow.proto.framework.TensorDescription, org.tensorflow.proto.framework.TensorDescription.Builder, org.tensorflow.proto.framework.TensorDescriptionOrBuilder>(
+                getTensorDescription(),
+                getParentForChildren(),
+                isClean());
+        tensorDescription_ = null;
+      }
+      return tensorDescriptionBuilder_;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.NodeOutput)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.NodeOutput)
+  private static final org.tensorflow.proto.framework.NodeOutput DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.NodeOutput();
+  }
+
+  public static org.tensorflow.proto.framework.NodeOutput getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<NodeOutput>
+      PARSER = new com.google.protobuf.AbstractParser<NodeOutput>() {
+    @java.lang.Override
+    public NodeOutput parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new NodeOutput(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<NodeOutput> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<NodeOutput> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.framework.NodeOutput getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NodeOutputOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NodeOutputOrBuilder.java
new file mode 100644
index 00000000000..940f8defed7
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NodeOutputOrBuilder.java
@@ -0,0 +1,27 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/framework/step_stats.proto
+
+package org.tensorflow.proto.framework;
+
+public interface NodeOutputOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.NodeOutput)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <code>int32 slot = 1;</code>
+   */
+  int getSlot();
+
+  /**
+   * <code>.tensorflow.TensorDescription tensor_description = 3;</code>
+   */
+  boolean hasTensorDescription();
+  /**
+   * <code>.tensorflow.TensorDescription tensor_description = 3;</code>
+   */
+  org.tensorflow.proto.framework.TensorDescription getTensorDescription();
+  /**
+   * <code>.tensorflow.TensorDescription tensor_description = 3;</code>
+   */
+  org.tensorflow.proto.framework.TensorDescriptionOrBuilder getTensorDescriptionOrBuilder();
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NodeProto.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NodeProto.java
new file mode 100644
index 00000000000..ef4be3729ea
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NodeProto.java
@@ -0,0 +1,84 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/framework/node_def.proto
+
+package org.tensorflow.proto.framework;
+
+public final class NodeProto {
+  private NodeProto() {}
+  public static void registerAllExtensions(
+      com.google.protobuf.ExtensionRegistryLite registry) {
+  }
+
+  public static void registerAllExtensions(
+      com.google.protobuf.ExtensionRegistry registry) {
+    registerAllExtensions(
+        (com.google.protobuf.ExtensionRegistryLite) registry);
+  }
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_NodeDef_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_NodeDef_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_NodeDef_AttrEntry_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_NodeDef_AttrEntry_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_NodeDef_ExperimentalDebugInfo_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_NodeDef_ExperimentalDebugInfo_fieldAccessorTable;
+
+  public static com.google.protobuf.Descriptors.FileDescriptor
+      getDescriptor() {
+    return descriptor;
+  }
+  private static  com.google.protobuf.Descriptors.FileDescriptor
+      descriptor;
+  static {
+    java.lang.String[] descriptorData = {
+      "\n(tensorflow/core/framework/node_def.pro" +
+      "to\022\ntensorflow\032*tensorflow/core/framewor" +
+      "k/attr_value.proto\"\322\002\n\007NodeDef\022\014\n\004name\030\001" +
+      " \001(\t\022\n\n\002op\030\002 \001(\t\022\r\n\005input\030\003 \003(\t\022\016\n\006devic" +
+      "e\030\004 \001(\t\022+\n\004attr\030\005 \003(\0132\035.tensorflow.NodeD" +
+      "ef.AttrEntry\022J\n\027experimental_debug_info\030" +
+      "\006 \001(\0132).tensorflow.NodeDef.ExperimentalD" +
+      "ebugInfo\032B\n\tAttrEntry\022\013\n\003key\030\001 \001(\t\022$\n\005va" +
+      "lue\030\002 \001(\0132\025.tensorflow.AttrValue:\0028\001\032Q\n\025" +
+      "ExperimentalDebugInfo\022\033\n\023original_node_n" +
+      "ames\030\001 \003(\t\022\033\n\023original_func_names\030\002 \003(\tB" +
+      "\201\001\n\036org.tensorflow.proto.frameworkB\tNode" +
+      "ProtoP\001ZOgithub.com/tensorflow/tensorflo" +
+      "w/tensorflow/go/core/framework/node_def_" +
+      "go_proto\370\001\001b\006proto3"
+    };
+    descriptor = com.google.protobuf.Descriptors.FileDescriptor
+      .internalBuildGeneratedFileFrom(descriptorData,
+        new com.google.protobuf.Descriptors.FileDescriptor[] {
+          org.tensorflow.proto.framework.AttrValueProtos.getDescriptor(),
+        });
+    internal_static_tensorflow_NodeDef_descriptor =
+      getDescriptor().getMessageTypes().get(0);
+    internal_static_tensorflow_NodeDef_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_NodeDef_descriptor,
+        new java.lang.String[] { "Name", "Op", "Input", "Device", "Attr", "ExperimentalDebugInfo", });
+    internal_static_tensorflow_NodeDef_AttrEntry_descriptor =
+      internal_static_tensorflow_NodeDef_descriptor.getNestedTypes().get(0);
+    internal_static_tensorflow_NodeDef_AttrEntry_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_NodeDef_AttrEntry_descriptor,
+        new java.lang.String[] { "Key", "Value", });
+    internal_static_tensorflow_NodeDef_ExperimentalDebugInfo_descriptor =
+      internal_static_tensorflow_NodeDef_descriptor.getNestedTypes().get(1);
+    internal_static_tensorflow_NodeDef_ExperimentalDebugInfo_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_NodeDef_ExperimentalDebugInfo_descriptor,
+        new java.lang.String[] { "OriginalNodeNames", "OriginalFuncNames", });
+    org.tensorflow.proto.framework.AttrValueProtos.getDescriptor();
+  }
+
+  // @@protoc_insertion_point(outer_class_scope)
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NoneValue.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NoneValue.java
new file mode 100644
index 00000000000..7aaba25f2d2
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NoneValue.java
@@ -0,0 +1,427 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/struct.proto
+
+package org.tensorflow.proto.framework;
+
+/**
+ * <pre>
+ * Represents None.
+ * </pre>
+ *
+ * Protobuf type {@code tensorflow.NoneValue}
+ */
+public  final class NoneValue extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.NoneValue)
+    NoneValueOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use NoneValue.newBuilder() to construct.
+  private NoneValue(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private NoneValue() {
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new NoneValue();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private NoneValue(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.framework.StructProtos.internal_static_tensorflow_NoneValue_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.framework.StructProtos.internal_static_tensorflow_NoneValue_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.framework.NoneValue.class, org.tensorflow.proto.framework.NoneValue.Builder.class);
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.framework.NoneValue)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.framework.NoneValue other = (org.tensorflow.proto.framework.NoneValue) obj;
+
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.framework.NoneValue parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.NoneValue parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.NoneValue parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.NoneValue parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.NoneValue parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.NoneValue parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.NoneValue parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.NoneValue parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.NoneValue parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.NoneValue parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.NoneValue parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.NoneValue parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.framework.NoneValue prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * <pre>
+   * Represents None.
+   * </pre>
+   *
+   * Protobuf type {@code tensorflow.NoneValue}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.NoneValue)
+      org.tensorflow.proto.framework.NoneValueOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.StructProtos.internal_static_tensorflow_NoneValue_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.StructProtos.internal_static_tensorflow_NoneValue_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.NoneValue.class, org.tensorflow.proto.framework.NoneValue.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.framework.NoneValue.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.framework.StructProtos.internal_static_tensorflow_NoneValue_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.NoneValue getDefaultInstanceForType() {
+      return org.tensorflow.proto.framework.NoneValue.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.NoneValue build() {
+      org.tensorflow.proto.framework.NoneValue result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.NoneValue buildPartial() {
+      org.tensorflow.proto.framework.NoneValue result = new org.tensorflow.proto.framework.NoneValue(this);
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.framework.NoneValue) {
+        return mergeFrom((org.tensorflow.proto.framework.NoneValue)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.framework.NoneValue other) {
+      if (other == org.tensorflow.proto.framework.NoneValue.getDefaultInstance()) return this;
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.framework.NoneValue parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.framework.NoneValue) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.NoneValue)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.NoneValue)
+  private static final org.tensorflow.proto.framework.NoneValue DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.NoneValue();
+  }
+
+  public static org.tensorflow.proto.framework.NoneValue getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<NoneValue>
+      PARSER = new com.google.protobuf.AbstractParser<NoneValue>() {
+    @java.lang.Override
+    public NoneValue parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new NoneValue(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<NoneValue> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<NoneValue> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.framework.NoneValue getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NoneValueOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NoneValueOrBuilder.java
new file mode 100644
index 00000000000..3d720d0c269
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/NoneValueOrBuilder.java
@@ -0,0 +1,9 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/struct.proto
+
+package org.tensorflow.proto.framework;
+
+public interface NoneValueOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.NoneValue)
+    com.google.protobuf.MessageOrBuilder {
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/OpDef.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/OpDef.java
new file mode 100644
index 00000000000..80839f48dcc
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/OpDef.java
@@ -0,0 +1,6284 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/framework/op_def.proto
+
+package org.tensorflow.proto.framework;
+
+/**
+ * <pre>
+ * Defines an operation. A NodeDef in a GraphDef specifies an Op by
+ * using the "op" field which should match the name of a OpDef.
+ * LINT.IfChange
+ * </pre>
+ *
+ * Protobuf type {@code tensorflow.OpDef}
+ */
+public  final class OpDef extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.OpDef)
+    OpDefOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use OpDef.newBuilder() to construct.
+  private OpDef(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private OpDef() {
+    name_ = "";
+    inputArg_ = java.util.Collections.emptyList();
+    outputArg_ = java.util.Collections.emptyList();
+    controlOutput_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+    attr_ = java.util.Collections.emptyList();
+    summary_ = "";
+    description_ = "";
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new OpDef();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private OpDef(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    int mutable_bitField0_ = 0;
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 10: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            name_ = s;
+            break;
+          }
+          case 18: {
+            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
+              inputArg_ = new java.util.ArrayList<org.tensorflow.proto.framework.OpDef.ArgDef>();
+              mutable_bitField0_ |= 0x00000001;
+            }
+            inputArg_.add(
+                input.readMessage(org.tensorflow.proto.framework.OpDef.ArgDef.parser(), extensionRegistry));
+            break;
+          }
+          case 26: {
+            if (!((mutable_bitField0_ & 0x00000002) != 0)) {
+              outputArg_ = new java.util.ArrayList<org.tensorflow.proto.framework.OpDef.ArgDef>();
+              mutable_bitField0_ |= 0x00000002;
+            }
+            outputArg_.add(
+                input.readMessage(org.tensorflow.proto.framework.OpDef.ArgDef.parser(), extensionRegistry));
+            break;
+          }
+          case 34: {
+            if (!((mutable_bitField0_ & 0x00000008) != 0)) {
+              attr_ = new java.util.ArrayList<org.tensorflow.proto.framework.OpDef.AttrDef>();
+              mutable_bitField0_ |= 0x00000008;
+            }
+            attr_.add(
+                input.readMessage(org.tensorflow.proto.framework.OpDef.AttrDef.parser(), extensionRegistry));
+            break;
+          }
+          case 42: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            summary_ = s;
+            break;
+          }
+          case 50: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            description_ = s;
+            break;
+          }
+          case 66: {
+            org.tensorflow.proto.framework.OpDeprecation.Builder subBuilder = null;
+            if (deprecation_ != null) {
+              subBuilder = deprecation_.toBuilder();
+            }
+            deprecation_ = input.readMessage(org.tensorflow.proto.framework.OpDeprecation.parser(), extensionRegistry);
+            if (subBuilder != null) {
+              subBuilder.mergeFrom(deprecation_);
+              deprecation_ = subBuilder.buildPartial();
+            }
+
+            break;
+          }
+          case 128: {
+
+            isAggregate_ = input.readBool();
+            break;
+          }
+          case 136: {
+
+            isStateful_ = input.readBool();
+            break;
+          }
+          case 144: {
+
+            isCommutative_ = input.readBool();
+            break;
+          }
+          case 152: {
+
+            allowsUninitializedInput_ = input.readBool();
+            break;
+          }
+          case 162: {
+            java.lang.String s = input.readStringRequireUtf8();
+            if (!((mutable_bitField0_ & 0x00000004) != 0)) {
+              controlOutput_ = new com.google.protobuf.LazyStringArrayList();
+              mutable_bitField0_ |= 0x00000004;
+            }
+            controlOutput_.add(s);
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      if (((mutable_bitField0_ & 0x00000001) != 0)) {
+        inputArg_ = java.util.Collections.unmodifiableList(inputArg_);
+      }
+      if (((mutable_bitField0_ & 0x00000002) != 0)) {
+        outputArg_ = java.util.Collections.unmodifiableList(outputArg_);
+      }
+      if (((mutable_bitField0_ & 0x00000008) != 0)) {
+        attr_ = java.util.Collections.unmodifiableList(attr_);
+      }
+      if (((mutable_bitField0_ & 0x00000004) != 0)) {
+        controlOutput_ = controlOutput_.getUnmodifiableView();
+      }
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.framework.OpDefProtos.internal_static_tensorflow_OpDef_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.framework.OpDefProtos.internal_static_tensorflow_OpDef_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.framework.OpDef.class, org.tensorflow.proto.framework.OpDef.Builder.class);
+  }
+
+  public interface ArgDefOrBuilder extends
+      // @@protoc_insertion_point(interface_extends:tensorflow.OpDef.ArgDef)
+      com.google.protobuf.MessageOrBuilder {
+
+    /**
+     * <pre>
+     * Name for the input/output.  Should match the regexp "[a-z][a-z0-9_]*".
+     * </pre>
+     *
+     * <code>string name = 1;</code>
+     */
+    java.lang.String getName();
+    /**
+     * <pre>
+     * Name for the input/output.  Should match the regexp "[a-z][a-z0-9_]*".
+     * </pre>
+     *
+     * <code>string name = 1;</code>
+     */
+    com.google.protobuf.ByteString
+        getNameBytes();
+
+    /**
+     * <pre>
+     * Human readable description.
+     * </pre>
+     *
+     * <code>string description = 2;</code>
+     */
+    java.lang.String getDescription();
+    /**
+     * <pre>
+     * Human readable description.
+     * </pre>
+     *
+     * <code>string description = 2;</code>
+     */
+    com.google.protobuf.ByteString
+        getDescriptionBytes();
+
+    /**
+     * <pre>
+     * Describes the type of one or more tensors that are accepted/produced
+     * by this input/output arg.  The only legal combinations are:
+     * * For a single tensor: either the "type" field is set or the
+     *   "type_attr" field is set to the name of an attr with type "type".
+     * * For a sequence of tensors with the same type: the "number_attr"
+     *   field will be set to the name of an attr with type "int", and
+     *   either the "type" or "type_attr" field will be set as for
+     *   single tensors.
+     * * For a sequence of tensors, the "type_list_attr" field will be set
+     *   to the name of an attr with type "list(type)".
+     * </pre>
+     *
+     * <code>.tensorflow.DataType type = 3;</code>
+     */
+    int getTypeValue();
+    /**
+     * <pre>
+     * Describes the type of one or more tensors that are accepted/produced
+     * by this input/output arg.  The only legal combinations are:
+     * * For a single tensor: either the "type" field is set or the
+     *   "type_attr" field is set to the name of an attr with type "type".
+     * * For a sequence of tensors with the same type: the "number_attr"
+     *   field will be set to the name of an attr with type "int", and
+     *   either the "type" or "type_attr" field will be set as for
+     *   single tensors.
+     * * For a sequence of tensors, the "type_list_attr" field will be set
+     *   to the name of an attr with type "list(type)".
+     * </pre>
+     *
+     * <code>.tensorflow.DataType type = 3;</code>
+     */
+    org.tensorflow.proto.framework.DataType getType();
+
+    /**
+     * <pre>
+     * if specified, attr must have type "type"
+     * </pre>
+     *
+     * <code>string type_attr = 4;</code>
+     */
+    java.lang.String getTypeAttr();
+    /**
+     * <pre>
+     * if specified, attr must have type "type"
+     * </pre>
+     *
+     * <code>string type_attr = 4;</code>
+     */
+    com.google.protobuf.ByteString
+        getTypeAttrBytes();
+
+    /**
+     * <pre>
+     * if specified, attr must have type "int"
+     * </pre>
+     *
+     * <code>string number_attr = 5;</code>
+     */
+    java.lang.String getNumberAttr();
+    /**
+     * <pre>
+     * if specified, attr must have type "int"
+     * </pre>
+     *
+     * <code>string number_attr = 5;</code>
+     */
+    com.google.protobuf.ByteString
+        getNumberAttrBytes();
+
+    /**
+     * <pre>
+     * If specified, attr must have type "list(type)", and none of
+     * type, type_attr, and number_attr may be specified.
+     * </pre>
+     *
+     * <code>string type_list_attr = 6;</code>
+     */
+    java.lang.String getTypeListAttr();
+    /**
+     * <pre>
+     * If specified, attr must have type "list(type)", and none of
+     * type, type_attr, and number_attr may be specified.
+     * </pre>
+     *
+     * <code>string type_list_attr = 6;</code>
+     */
+    com.google.protobuf.ByteString
+        getTypeListAttrBytes();
+
+    /**
+     * <pre>
+     * For inputs: if true, the inputs are required to be refs.
+     *   By default, inputs can be either refs or non-refs.
+     * For outputs: if true, outputs are refs, otherwise they are not.
+     * </pre>
+     *
+     * <code>bool is_ref = 16;</code>
+     */
+    boolean getIsRef();
+  }
+  /**
+   * <pre>
+   * For describing inputs and outputs.
+   * </pre>
+   *
+   * Protobuf type {@code tensorflow.OpDef.ArgDef}
+   */
+  public  static final class ArgDef extends
+      com.google.protobuf.GeneratedMessageV3 implements
+      // @@protoc_insertion_point(message_implements:tensorflow.OpDef.ArgDef)
+      ArgDefOrBuilder {
+  private static final long serialVersionUID = 0L;
+    // Use ArgDef.newBuilder() to construct.
+    private ArgDef(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+      super(builder);
+    }
+    private ArgDef() {
+      name_ = "";
+      description_ = "";
+      type_ = 0;
+      typeAttr_ = "";
+      numberAttr_ = "";
+      typeListAttr_ = "";
+    }
+
+    @java.lang.Override
+    @SuppressWarnings({"unused"})
+    protected java.lang.Object newInstance(
+        UnusedPrivateParameter unused) {
+      return new ArgDef();
+    }
+
+    @java.lang.Override
+    public final com.google.protobuf.UnknownFieldSet
+    getUnknownFields() {
+      return this.unknownFields;
+    }
+    private ArgDef(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      this();
+      if (extensionRegistry == null) {
+        throw new java.lang.NullPointerException();
+      }
+      com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+          com.google.protobuf.UnknownFieldSet.newBuilder();
+      try {
+        boolean done = false;
+        while (!done) {
+          int tag = input.readTag();
+          switch (tag) {
+            case 0:
+              done = true;
+              break;
+            case 10: {
+              java.lang.String s = input.readStringRequireUtf8();
+
+              name_ = s;
+              break;
+            }
+            case 18: {
+              java.lang.String s = input.readStringRequireUtf8();
+
+              description_ = s;
+              break;
+            }
+            case 24: {
+              int rawValue = input.readEnum();
+
+              type_ = rawValue;
+              break;
+            }
+            case 34: {
+              java.lang.String s = input.readStringRequireUtf8();
+
+              typeAttr_ = s;
+              break;
+            }
+            case 42: {
+              java.lang.String s = input.readStringRequireUtf8();
+
+              numberAttr_ = s;
+              break;
+            }
+            case 50: {
+              java.lang.String s = input.readStringRequireUtf8();
+
+              typeListAttr_ = s;
+              break;
+            }
+            case 128: {
+
+              isRef_ = input.readBool();
+              break;
+            }
+            default: {
+              if (!parseUnknownField(
+                  input, unknownFields, extensionRegistry, tag)) {
+                done = true;
+              }
+              break;
+            }
+          }
+        }
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        throw e.setUnfinishedMessage(this);
+      } catch (java.io.IOException e) {
+        throw new com.google.protobuf.InvalidProtocolBufferException(
+            e).setUnfinishedMessage(this);
+      } finally {
+        this.unknownFields = unknownFields.build();
+        makeExtensionsImmutable();
+      }
+    }
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.OpDefProtos.internal_static_tensorflow_OpDef_ArgDef_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.OpDefProtos.internal_static_tensorflow_OpDef_ArgDef_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.OpDef.ArgDef.class, org.tensorflow.proto.framework.OpDef.ArgDef.Builder.class);
+    }
+
+    public static final int NAME_FIELD_NUMBER = 1;
+    private volatile java.lang.Object name_;
+    /**
+     * <pre>
+     * Name for the input/output.  Should match the regexp "[a-z][a-z0-9_]*".
+     * </pre>
+     *
+     * <code>string name = 1;</code>
+     */
+    public java.lang.String getName() {
+      java.lang.Object ref = name_;
+      if (ref instanceof java.lang.String) {
+        return (java.lang.String) ref;
+      } else {
+        com.google.protobuf.ByteString bs = 
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        name_ = s;
+        return s;
+      }
+    }
+    /**
+     * <pre>
+     * Name for the input/output.  Should match the regexp "[a-z][a-z0-9_]*".
+     * </pre>
+     *
+     * <code>string name = 1;</code>
+     */
+    public com.google.protobuf.ByteString
+        getNameBytes() {
+      java.lang.Object ref = name_;
+      if (ref instanceof java.lang.String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        name_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+
+    public static final int DESCRIPTION_FIELD_NUMBER = 2;
+    private volatile java.lang.Object description_;
+    /**
+     * <pre>
+     * Human readable description.
+     * </pre>
+     *
+     * <code>string description = 2;</code>
+     */
+    public java.lang.String getDescription() {
+      java.lang.Object ref = description_;
+      if (ref instanceof java.lang.String) {
+        return (java.lang.String) ref;
+      } else {
+        com.google.protobuf.ByteString bs = 
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        description_ = s;
+        return s;
+      }
+    }
+    /**
+     * <pre>
+     * Human readable description.
+     * </pre>
+     *
+     * <code>string description = 2;</code>
+     */
+    public com.google.protobuf.ByteString
+        getDescriptionBytes() {
+      java.lang.Object ref = description_;
+      if (ref instanceof java.lang.String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        description_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+
+    public static final int TYPE_FIELD_NUMBER = 3;
+    private int type_;
+    /**
+     * <pre>
+     * Describes the type of one or more tensors that are accepted/produced
+     * by this input/output arg.  The only legal combinations are:
+     * * For a single tensor: either the "type" field is set or the
+     *   "type_attr" field is set to the name of an attr with type "type".
+     * * For a sequence of tensors with the same type: the "number_attr"
+     *   field will be set to the name of an attr with type "int", and
+     *   either the "type" or "type_attr" field will be set as for
+     *   single tensors.
+     * * For a sequence of tensors, the "type_list_attr" field will be set
+     *   to the name of an attr with type "list(type)".
+     * </pre>
+     *
+     * <code>.tensorflow.DataType type = 3;</code>
+     */
+    public int getTypeValue() {
+      return type_;
+    }
+    /**
+     * <pre>
+     * Describes the type of one or more tensors that are accepted/produced
+     * by this input/output arg.  The only legal combinations are:
+     * * For a single tensor: either the "type" field is set or the
+     *   "type_attr" field is set to the name of an attr with type "type".
+     * * For a sequence of tensors with the same type: the "number_attr"
+     *   field will be set to the name of an attr with type "int", and
+     *   either the "type" or "type_attr" field will be set as for
+     *   single tensors.
+     * * For a sequence of tensors, the "type_list_attr" field will be set
+     *   to the name of an attr with type "list(type)".
+     * </pre>
+     *
+     * <code>.tensorflow.DataType type = 3;</code>
+     */
+    public org.tensorflow.proto.framework.DataType getType() {
+      @SuppressWarnings("deprecation")
+      org.tensorflow.proto.framework.DataType result = org.tensorflow.proto.framework.DataType.valueOf(type_);
+      return result == null ? org.tensorflow.proto.framework.DataType.UNRECOGNIZED : result;
+    }
+
+    public static final int TYPE_ATTR_FIELD_NUMBER = 4;
+    private volatile java.lang.Object typeAttr_;
+    /**
+     * <pre>
+     * if specified, attr must have type "type"
+     * </pre>
+     *
+     * <code>string type_attr = 4;</code>
+     */
+    public java.lang.String getTypeAttr() {
+      java.lang.Object ref = typeAttr_;
+      if (ref instanceof java.lang.String) {
+        return (java.lang.String) ref;
+      } else {
+        com.google.protobuf.ByteString bs = 
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        typeAttr_ = s;
+        return s;
+      }
+    }
+    /**
+     * <pre>
+     * if specified, attr must have type "type"
+     * </pre>
+     *
+     * <code>string type_attr = 4;</code>
+     */
+    public com.google.protobuf.ByteString
+        getTypeAttrBytes() {
+      java.lang.Object ref = typeAttr_;
+      if (ref instanceof java.lang.String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        typeAttr_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+
+    public static final int NUMBER_ATTR_FIELD_NUMBER = 5;
+    private volatile java.lang.Object numberAttr_;
+    /**
+     * <pre>
+     * if specified, attr must have type "int"
+     * </pre>
+     *
+     * <code>string number_attr = 5;</code>
+     */
+    public java.lang.String getNumberAttr() {
+      java.lang.Object ref = numberAttr_;
+      if (ref instanceof java.lang.String) {
+        return (java.lang.String) ref;
+      } else {
+        com.google.protobuf.ByteString bs = 
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        numberAttr_ = s;
+        return s;
+      }
+    }
+    /**
+     * <pre>
+     * if specified, attr must have type "int"
+     * </pre>
+     *
+     * <code>string number_attr = 5;</code>
+     */
+    public com.google.protobuf.ByteString
+        getNumberAttrBytes() {
+      java.lang.Object ref = numberAttr_;
+      if (ref instanceof java.lang.String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        numberAttr_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+
+    public static final int TYPE_LIST_ATTR_FIELD_NUMBER = 6;
+    private volatile java.lang.Object typeListAttr_;
+    /**
+     * <pre>
+     * If specified, attr must have type "list(type)", and none of
+     * type, type_attr, and number_attr may be specified.
+     * </pre>
+     *
+     * <code>string type_list_attr = 6;</code>
+     */
+    public java.lang.String getTypeListAttr() {
+      java.lang.Object ref = typeListAttr_;
+      if (ref instanceof java.lang.String) {
+        return (java.lang.String) ref;
+      } else {
+        com.google.protobuf.ByteString bs = 
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        typeListAttr_ = s;
+        return s;
+      }
+    }
+    /**
+     * <pre>
+     * If specified, attr must have type "list(type)", and none of
+     * type, type_attr, and number_attr may be specified.
+     * </pre>
+     *
+     * <code>string type_list_attr = 6;</code>
+     */
+    public com.google.protobuf.ByteString
+        getTypeListAttrBytes() {
+      java.lang.Object ref = typeListAttr_;
+      if (ref instanceof java.lang.String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        typeListAttr_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+
+    public static final int IS_REF_FIELD_NUMBER = 16;
+    private boolean isRef_;
+    /**
+     * <pre>
+     * For inputs: if true, the inputs are required to be refs.
+     *   By default, inputs can be either refs or non-refs.
+     * For outputs: if true, outputs are refs, otherwise they are not.
+     * </pre>
+     *
+     * <code>bool is_ref = 16;</code>
+     */
+    public boolean getIsRef() {
+      return isRef_;
+    }
+
+    private byte memoizedIsInitialized = -1;
+    @java.lang.Override
+    public final boolean isInitialized() {
+      byte isInitialized = memoizedIsInitialized;
+      if (isInitialized == 1) return true;
+      if (isInitialized == 0) return false;
+
+      memoizedIsInitialized = 1;
+      return true;
+    }
+
+    @java.lang.Override
+    public void writeTo(com.google.protobuf.CodedOutputStream output)
+                        throws java.io.IOException {
+      if (!getNameBytes().isEmpty()) {
+        com.google.protobuf.GeneratedMessageV3.writeString(output, 1, name_);
+      }
+      if (!getDescriptionBytes().isEmpty()) {
+        com.google.protobuf.GeneratedMessageV3.writeString(output, 2, description_);
+      }
+      if (type_ != org.tensorflow.proto.framework.DataType.DT_INVALID.getNumber()) {
+        output.writeEnum(3, type_);
+      }
+      if (!getTypeAttrBytes().isEmpty()) {
+        com.google.protobuf.GeneratedMessageV3.writeString(output, 4, typeAttr_);
+      }
+      if (!getNumberAttrBytes().isEmpty()) {
+        com.google.protobuf.GeneratedMessageV3.writeString(output, 5, numberAttr_);
+      }
+      if (!getTypeListAttrBytes().isEmpty()) {
+        com.google.protobuf.GeneratedMessageV3.writeString(output, 6, typeListAttr_);
+      }
+      if (isRef_ != false) {
+        output.writeBool(16, isRef_);
+      }
+      unknownFields.writeTo(output);
+    }
+
+    @java.lang.Override
+    public int getSerializedSize() {
+      int size = memoizedSize;
+      if (size != -1) return size;
+
+      size = 0;
+      if (!getNameBytes().isEmpty()) {
+        size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, name_);
+      }
+      if (!getDescriptionBytes().isEmpty()) {
+        size += com.google.protobuf.GeneratedMessageV3.computeStringSize(2, description_);
+      }
+      if (type_ != org.tensorflow.proto.framework.DataType.DT_INVALID.getNumber()) {
+        size += com.google.protobuf.CodedOutputStream
+          .computeEnumSize(3, type_);
+      }
+      if (!getTypeAttrBytes().isEmpty()) {
+        size += com.google.protobuf.GeneratedMessageV3.computeStringSize(4, typeAttr_);
+      }
+      if (!getNumberAttrBytes().isEmpty()) {
+        size += com.google.protobuf.GeneratedMessageV3.computeStringSize(5, numberAttr_);
+      }
+      if (!getTypeListAttrBytes().isEmpty()) {
+        size += com.google.protobuf.GeneratedMessageV3.computeStringSize(6, typeListAttr_);
+      }
+      if (isRef_ != false) {
+        size += com.google.protobuf.CodedOutputStream
+          .computeBoolSize(16, isRef_);
+      }
+      size += unknownFields.getSerializedSize();
+      memoizedSize = size;
+      return size;
+    }
+
+    @java.lang.Override
+    public boolean equals(final java.lang.Object obj) {
+      if (obj == this) {
+       return true;
+      }
+      if (!(obj instanceof org.tensorflow.proto.framework.OpDef.ArgDef)) {
+        return super.equals(obj);
+      }
+      org.tensorflow.proto.framework.OpDef.ArgDef other = (org.tensorflow.proto.framework.OpDef.ArgDef) obj;
+
+      if (!getName()
+          .equals(other.getName())) return false;
+      if (!getDescription()
+          .equals(other.getDescription())) return false;
+      if (type_ != other.type_) return false;
+      if (!getTypeAttr()
+          .equals(other.getTypeAttr())) return false;
+      if (!getNumberAttr()
+          .equals(other.getNumberAttr())) return false;
+      if (!getTypeListAttr()
+          .equals(other.getTypeListAttr())) return false;
+      if (getIsRef()
+          != other.getIsRef()) return false;
+      if (!unknownFields.equals(other.unknownFields)) return false;
+      return true;
+    }
+
+    @java.lang.Override
+    public int hashCode() {
+      if (memoizedHashCode != 0) {
+        return memoizedHashCode;
+      }
+      int hash = 41;
+      hash = (19 * hash) + getDescriptor().hashCode();
+      hash = (37 * hash) + NAME_FIELD_NUMBER;
+      hash = (53 * hash) + getName().hashCode();
+      hash = (37 * hash) + DESCRIPTION_FIELD_NUMBER;
+      hash = (53 * hash) + getDescription().hashCode();
+      hash = (37 * hash) + TYPE_FIELD_NUMBER;
+      hash = (53 * hash) + type_;
+      hash = (37 * hash) + TYPE_ATTR_FIELD_NUMBER;
+      hash = (53 * hash) + getTypeAttr().hashCode();
+      hash = (37 * hash) + NUMBER_ATTR_FIELD_NUMBER;
+      hash = (53 * hash) + getNumberAttr().hashCode();
+      hash = (37 * hash) + TYPE_LIST_ATTR_FIELD_NUMBER;
+      hash = (53 * hash) + getTypeListAttr().hashCode();
+      hash = (37 * hash) + IS_REF_FIELD_NUMBER;
+      hash = (53 * hash) + com.google.protobuf.Internal.hashBoolean(
+          getIsRef());
+      hash = (29 * hash) + unknownFields.hashCode();
+      memoizedHashCode = hash;
+      return hash;
+    }
+
+    public static org.tensorflow.proto.framework.OpDef.ArgDef parseFrom(
+        java.nio.ByteBuffer data)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data);
+    }
+    public static org.tensorflow.proto.framework.OpDef.ArgDef parseFrom(
+        java.nio.ByteBuffer data,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.OpDef.ArgDef parseFrom(
+        com.google.protobuf.ByteString data)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data);
+    }
+    public static org.tensorflow.proto.framework.OpDef.ArgDef parseFrom(
+        com.google.protobuf.ByteString data,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.OpDef.ArgDef parseFrom(byte[] data)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data);
+    }
+    public static org.tensorflow.proto.framework.OpDef.ArgDef parseFrom(
+        byte[] data,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.OpDef.ArgDef parseFrom(java.io.InputStream input)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input);
+    }
+    public static org.tensorflow.proto.framework.OpDef.ArgDef parseFrom(
+        java.io.InputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.OpDef.ArgDef parseDelimitedFrom(java.io.InputStream input)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseDelimitedWithIOException(PARSER, input);
+    }
+    public static org.tensorflow.proto.framework.OpDef.ArgDef parseDelimitedFrom(
+        java.io.InputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.OpDef.ArgDef parseFrom(
+        com.google.protobuf.CodedInputStream input)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input);
+    }
+    public static org.tensorflow.proto.framework.OpDef.ArgDef parseFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input, extensionRegistry);
+    }
+
+    @java.lang.Override
+    public Builder newBuilderForType() { return newBuilder(); }
+    public static Builder newBuilder() {
+      return DEFAULT_INSTANCE.toBuilder();
+    }
+    public static Builder newBuilder(org.tensorflow.proto.framework.OpDef.ArgDef prototype) {
+      return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+    }
+    @java.lang.Override
+    public Builder toBuilder() {
+      return this == DEFAULT_INSTANCE
+          ? new Builder() : new Builder().mergeFrom(this);
+    }
+
+    @java.lang.Override
+    protected Builder newBuilderForType(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      Builder builder = new Builder(parent);
+      return builder;
+    }
+    /**
+     * <pre>
+     * For describing inputs and outputs.
+     * </pre>
+     *
+     * Protobuf type {@code tensorflow.OpDef.ArgDef}
+     */
+    public static final class Builder extends
+        com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+        // @@protoc_insertion_point(builder_implements:tensorflow.OpDef.ArgDef)
+        org.tensorflow.proto.framework.OpDef.ArgDefOrBuilder {
+      public static final com.google.protobuf.Descriptors.Descriptor
+          getDescriptor() {
+        return org.tensorflow.proto.framework.OpDefProtos.internal_static_tensorflow_OpDef_ArgDef_descriptor;
+      }
+
+      @java.lang.Override
+      protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+          internalGetFieldAccessorTable() {
+        return org.tensorflow.proto.framework.OpDefProtos.internal_static_tensorflow_OpDef_ArgDef_fieldAccessorTable
+            .ensureFieldAccessorsInitialized(
+                org.tensorflow.proto.framework.OpDef.ArgDef.class, org.tensorflow.proto.framework.OpDef.ArgDef.Builder.class);
+      }
+
+      // Construct using org.tensorflow.proto.framework.OpDef.ArgDef.newBuilder()
+      private Builder() {
+        maybeForceBuilderInitialization();
+      }
+
+      private Builder(
+          com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+        super(parent);
+        maybeForceBuilderInitialization();
+      }
+      private void maybeForceBuilderInitialization() {
+        if (com.google.protobuf.GeneratedMessageV3
+                .alwaysUseFieldBuilders) {
+        }
+      }
+      @java.lang.Override
+      public Builder clear() {
+        super.clear();
+        name_ = "";
+
+        description_ = "";
+
+        type_ = 0;
+
+        typeAttr_ = "";
+
+        numberAttr_ = "";
+
+        typeListAttr_ = "";
+
+        isRef_ = false;
+
+        return this;
+      }
+
+      @java.lang.Override
+      public com.google.protobuf.Descriptors.Descriptor
+          getDescriptorForType() {
+        return org.tensorflow.proto.framework.OpDefProtos.internal_static_tensorflow_OpDef_ArgDef_descriptor;
+      }
+
+      @java.lang.Override
+      public org.tensorflow.proto.framework.OpDef.ArgDef getDefaultInstanceForType() {
+        return org.tensorflow.proto.framework.OpDef.ArgDef.getDefaultInstance();
+      }
+
+      @java.lang.Override
+      public org.tensorflow.proto.framework.OpDef.ArgDef build() {
+        org.tensorflow.proto.framework.OpDef.ArgDef result = buildPartial();
+        if (!result.isInitialized()) {
+          throw newUninitializedMessageException(result);
+        }
+        return result;
+      }
+
+      @java.lang.Override
+      public org.tensorflow.proto.framework.OpDef.ArgDef buildPartial() {
+        org.tensorflow.proto.framework.OpDef.ArgDef result = new org.tensorflow.proto.framework.OpDef.ArgDef(this);
+        result.name_ = name_;
+        result.description_ = description_;
+        result.type_ = type_;
+        result.typeAttr_ = typeAttr_;
+        result.numberAttr_ = numberAttr_;
+        result.typeListAttr_ = typeListAttr_;
+        result.isRef_ = isRef_;
+        onBuilt();
+        return result;
+      }
+
+      @java.lang.Override
+      public Builder clone() {
+        return super.clone();
+      }
+      @java.lang.Override
+      public Builder setField(
+          com.google.protobuf.Descriptors.FieldDescriptor field,
+          java.lang.Object value) {
+        return super.setField(field, value);
+      }
+      @java.lang.Override
+      public Builder clearField(
+          com.google.protobuf.Descriptors.FieldDescriptor field) {
+        return super.clearField(field);
+      }
+      @java.lang.Override
+      public Builder clearOneof(
+          com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+        return super.clearOneof(oneof);
+      }
+      @java.lang.Override
+      public Builder setRepeatedField(
+          com.google.protobuf.Descriptors.FieldDescriptor field,
+          int index, java.lang.Object value) {
+        return super.setRepeatedField(field, index, value);
+      }
+      @java.lang.Override
+      public Builder addRepeatedField(
+          com.google.protobuf.Descriptors.FieldDescriptor field,
+          java.lang.Object value) {
+        return super.addRepeatedField(field, value);
+      }
+      @java.lang.Override
+      public Builder mergeFrom(com.google.protobuf.Message other) {
+        if (other instanceof org.tensorflow.proto.framework.OpDef.ArgDef) {
+          return mergeFrom((org.tensorflow.proto.framework.OpDef.ArgDef)other);
+        } else {
+          super.mergeFrom(other);
+          return this;
+        }
+      }
+
+      public Builder mergeFrom(org.tensorflow.proto.framework.OpDef.ArgDef other) {
+        if (other == org.tensorflow.proto.framework.OpDef.ArgDef.getDefaultInstance()) return this;
+        if (!other.getName().isEmpty()) {
+          name_ = other.name_;
+          onChanged();
+        }
+        if (!other.getDescription().isEmpty()) {
+          description_ = other.description_;
+          onChanged();
+        }
+        if (other.type_ != 0) {
+          setTypeValue(other.getTypeValue());
+        }
+        if (!other.getTypeAttr().isEmpty()) {
+          typeAttr_ = other.typeAttr_;
+          onChanged();
+        }
+        if (!other.getNumberAttr().isEmpty()) {
+          numberAttr_ = other.numberAttr_;
+          onChanged();
+        }
+        if (!other.getTypeListAttr().isEmpty()) {
+          typeListAttr_ = other.typeListAttr_;
+          onChanged();
+        }
+        if (other.getIsRef() != false) {
+          setIsRef(other.getIsRef());
+        }
+        this.mergeUnknownFields(other.unknownFields);
+        onChanged();
+        return this;
+      }
+
+      @java.lang.Override
+      public final boolean isInitialized() {
+        return true;
+      }
+
+      @java.lang.Override
+      public Builder mergeFrom(
+          com.google.protobuf.CodedInputStream input,
+          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+          throws java.io.IOException {
+        org.tensorflow.proto.framework.OpDef.ArgDef parsedMessage = null;
+        try {
+          parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+        } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+          parsedMessage = (org.tensorflow.proto.framework.OpDef.ArgDef) e.getUnfinishedMessage();
+          throw e.unwrapIOException();
+        } finally {
+          if (parsedMessage != null) {
+            mergeFrom(parsedMessage);
+          }
+        }
+        return this;
+      }
+
+      private java.lang.Object name_ = "";
+      /**
+       * <pre>
+       * Name for the input/output.  Should match the regexp "[a-z][a-z0-9_]*".
+       * </pre>
+       *
+       * <code>string name = 1;</code>
+       */
+      public java.lang.String getName() {
+        java.lang.Object ref = name_;
+        if (!(ref instanceof java.lang.String)) {
+          com.google.protobuf.ByteString bs =
+              (com.google.protobuf.ByteString) ref;
+          java.lang.String s = bs.toStringUtf8();
+          name_ = s;
+          return s;
+        } else {
+          return (java.lang.String) ref;
+        }
+      }
+      /**
+       * <pre>
+       * Name for the input/output.  Should match the regexp "[a-z][a-z0-9_]*".
+       * </pre>
+       *
+       * <code>string name = 1;</code>
+       */
+      public com.google.protobuf.ByteString
+          getNameBytes() {
+        java.lang.Object ref = name_;
+        if (ref instanceof String) {
+          com.google.protobuf.ByteString b = 
+              com.google.protobuf.ByteString.copyFromUtf8(
+                  (java.lang.String) ref);
+          name_ = b;
+          return b;
+        } else {
+          return (com.google.protobuf.ByteString) ref;
+        }
+      }
+      /**
+       * <pre>
+       * Name for the input/output.  Should match the regexp "[a-z][a-z0-9_]*".
+       * </pre>
+       *
+       * <code>string name = 1;</code>
+       */
+      public Builder setName(
+          java.lang.String value) {
+        if (value == null) {
+    throw new NullPointerException();
+  }
+  
+        name_ = value;
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * Name for the input/output.  Should match the regexp "[a-z][a-z0-9_]*".
+       * </pre>
+       *
+       * <code>string name = 1;</code>
+       */
+      public Builder clearName() {
+        
+        name_ = getDefaultInstance().getName();
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * Name for the input/output.  Should match the regexp "[a-z][a-z0-9_]*".
+       * </pre>
+       *
+       * <code>string name = 1;</code>
+       */
+      public Builder setNameBytes(
+          com.google.protobuf.ByteString value) {
+        if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+        
+        name_ = value;
+        onChanged();
+        return this;
+      }
+
+      private java.lang.Object description_ = "";
+      /**
+       * <pre>
+       * Human readable description.
+       * </pre>
+       *
+       * <code>string description = 2;</code>
+       */
+      public java.lang.String getDescription() {
+        java.lang.Object ref = description_;
+        if (!(ref instanceof java.lang.String)) {
+          com.google.protobuf.ByteString bs =
+              (com.google.protobuf.ByteString) ref;
+          java.lang.String s = bs.toStringUtf8();
+          description_ = s;
+          return s;
+        } else {
+          return (java.lang.String) ref;
+        }
+      }
+      /**
+       * <pre>
+       * Human readable description.
+       * </pre>
+       *
+       * <code>string description = 2;</code>
+       */
+      public com.google.protobuf.ByteString
+          getDescriptionBytes() {
+        java.lang.Object ref = description_;
+        if (ref instanceof String) {
+          com.google.protobuf.ByteString b = 
+              com.google.protobuf.ByteString.copyFromUtf8(
+                  (java.lang.String) ref);
+          description_ = b;
+          return b;
+        } else {
+          return (com.google.protobuf.ByteString) ref;
+        }
+      }
+      /**
+       * <pre>
+       * Human readable description.
+       * </pre>
+       *
+       * <code>string description = 2;</code>
+       */
+      public Builder setDescription(
+          java.lang.String value) {
+        if (value == null) {
+    throw new NullPointerException();
+  }
+  
+        description_ = value;
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * Human readable description.
+       * </pre>
+       *
+       * <code>string description = 2;</code>
+       */
+      public Builder clearDescription() {
+        
+        description_ = getDefaultInstance().getDescription();
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * Human readable description.
+       * </pre>
+       *
+       * <code>string description = 2;</code>
+       */
+      public Builder setDescriptionBytes(
+          com.google.protobuf.ByteString value) {
+        if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+        
+        description_ = value;
+        onChanged();
+        return this;
+      }
+
+      private int type_ = 0;
+      /**
+       * <pre>
+       * Describes the type of one or more tensors that are accepted/produced
+       * by this input/output arg.  The only legal combinations are:
+       * * For a single tensor: either the "type" field is set or the
+       *   "type_attr" field is set to the name of an attr with type "type".
+       * * For a sequence of tensors with the same type: the "number_attr"
+       *   field will be set to the name of an attr with type "int", and
+       *   either the "type" or "type_attr" field will be set as for
+       *   single tensors.
+       * * For a sequence of tensors, the "type_list_attr" field will be set
+       *   to the name of an attr with type "list(type)".
+       * </pre>
+       *
+       * <code>.tensorflow.DataType type = 3;</code>
+       */
+      public int getTypeValue() {
+        return type_;
+      }
+      /**
+       * <pre>
+       * Describes the type of one or more tensors that are accepted/produced
+       * by this input/output arg.  The only legal combinations are:
+       * * For a single tensor: either the "type" field is set or the
+       *   "type_attr" field is set to the name of an attr with type "type".
+       * * For a sequence of tensors with the same type: the "number_attr"
+       *   field will be set to the name of an attr with type "int", and
+       *   either the "type" or "type_attr" field will be set as for
+       *   single tensors.
+       * * For a sequence of tensors, the "type_list_attr" field will be set
+       *   to the name of an attr with type "list(type)".
+       * </pre>
+       *
+       * <code>.tensorflow.DataType type = 3;</code>
+       */
+      public Builder setTypeValue(int value) {
+        type_ = value;
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * Describes the type of one or more tensors that are accepted/produced
+       * by this input/output arg.  The only legal combinations are:
+       * * For a single tensor: either the "type" field is set or the
+       *   "type_attr" field is set to the name of an attr with type "type".
+       * * For a sequence of tensors with the same type: the "number_attr"
+       *   field will be set to the name of an attr with type "int", and
+       *   either the "type" or "type_attr" field will be set as for
+       *   single tensors.
+       * * For a sequence of tensors, the "type_list_attr" field will be set
+       *   to the name of an attr with type "list(type)".
+       * </pre>
+       *
+       * <code>.tensorflow.DataType type = 3;</code>
+       */
+      public org.tensorflow.proto.framework.DataType getType() {
+        @SuppressWarnings("deprecation")
+        org.tensorflow.proto.framework.DataType result = org.tensorflow.proto.framework.DataType.valueOf(type_);
+        return result == null ? org.tensorflow.proto.framework.DataType.UNRECOGNIZED : result;
+      }
+      /**
+       * <pre>
+       * Describes the type of one or more tensors that are accepted/produced
+       * by this input/output arg.  The only legal combinations are:
+       * * For a single tensor: either the "type" field is set or the
+       *   "type_attr" field is set to the name of an attr with type "type".
+       * * For a sequence of tensors with the same type: the "number_attr"
+       *   field will be set to the name of an attr with type "int", and
+       *   either the "type" or "type_attr" field will be set as for
+       *   single tensors.
+       * * For a sequence of tensors, the "type_list_attr" field will be set
+       *   to the name of an attr with type "list(type)".
+       * </pre>
+       *
+       * <code>.tensorflow.DataType type = 3;</code>
+       */
+      public Builder setType(org.tensorflow.proto.framework.DataType value) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        
+        type_ = value.getNumber();
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * Describes the type of one or more tensors that are accepted/produced
+       * by this input/output arg.  The only legal combinations are:
+       * * For a single tensor: either the "type" field is set or the
+       *   "type_attr" field is set to the name of an attr with type "type".
+       * * For a sequence of tensors with the same type: the "number_attr"
+       *   field will be set to the name of an attr with type "int", and
+       *   either the "type" or "type_attr" field will be set as for
+       *   single tensors.
+       * * For a sequence of tensors, the "type_list_attr" field will be set
+       *   to the name of an attr with type "list(type)".
+       * </pre>
+       *
+       * <code>.tensorflow.DataType type = 3;</code>
+       */
+      public Builder clearType() {
+        
+        type_ = 0;
+        onChanged();
+        return this;
+      }
+
+      private java.lang.Object typeAttr_ = "";
+      /**
+       * <pre>
+       * if specified, attr must have type "type"
+       * </pre>
+       *
+       * <code>string type_attr = 4;</code>
+       */
+      public java.lang.String getTypeAttr() {
+        java.lang.Object ref = typeAttr_;
+        if (!(ref instanceof java.lang.String)) {
+          com.google.protobuf.ByteString bs =
+              (com.google.protobuf.ByteString) ref;
+          java.lang.String s = bs.toStringUtf8();
+          typeAttr_ = s;
+          return s;
+        } else {
+          return (java.lang.String) ref;
+        }
+      }
+      /**
+       * <pre>
+       * if specified, attr must have type "type"
+       * </pre>
+       *
+       * <code>string type_attr = 4;</code>
+       */
+      public com.google.protobuf.ByteString
+          getTypeAttrBytes() {
+        java.lang.Object ref = typeAttr_;
+        if (ref instanceof String) {
+          com.google.protobuf.ByteString b = 
+              com.google.protobuf.ByteString.copyFromUtf8(
+                  (java.lang.String) ref);
+          typeAttr_ = b;
+          return b;
+        } else {
+          return (com.google.protobuf.ByteString) ref;
+        }
+      }
+      /**
+       * <pre>
+       * if specified, attr must have type "type"
+       * </pre>
+       *
+       * <code>string type_attr = 4;</code>
+       */
+      public Builder setTypeAttr(
+          java.lang.String value) {
+        if (value == null) {
+    throw new NullPointerException();
+  }
+  
+        typeAttr_ = value;
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * if specified, attr must have type "type"
+       * </pre>
+       *
+       * <code>string type_attr = 4;</code>
+       */
+      public Builder clearTypeAttr() {
+        
+        typeAttr_ = getDefaultInstance().getTypeAttr();
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * if specified, attr must have type "type"
+       * </pre>
+       *
+       * <code>string type_attr = 4;</code>
+       */
+      public Builder setTypeAttrBytes(
+          com.google.protobuf.ByteString value) {
+        if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+        
+        typeAttr_ = value;
+        onChanged();
+        return this;
+      }
+
+      private java.lang.Object numberAttr_ = "";
+      /**
+       * <pre>
+       * if specified, attr must have type "int"
+       * </pre>
+       *
+       * <code>string number_attr = 5;</code>
+       */
+      public java.lang.String getNumberAttr() {
+        java.lang.Object ref = numberAttr_;
+        if (!(ref instanceof java.lang.String)) {
+          com.google.protobuf.ByteString bs =
+              (com.google.protobuf.ByteString) ref;
+          java.lang.String s = bs.toStringUtf8();
+          numberAttr_ = s;
+          return s;
+        } else {
+          return (java.lang.String) ref;
+        }
+      }
+      /**
+       * <pre>
+       * if specified, attr must have type "int"
+       * </pre>
+       *
+       * <code>string number_attr = 5;</code>
+       */
+      public com.google.protobuf.ByteString
+          getNumberAttrBytes() {
+        java.lang.Object ref = numberAttr_;
+        if (ref instanceof String) {
+          com.google.protobuf.ByteString b = 
+              com.google.protobuf.ByteString.copyFromUtf8(
+                  (java.lang.String) ref);
+          numberAttr_ = b;
+          return b;
+        } else {
+          return (com.google.protobuf.ByteString) ref;
+        }
+      }
+      /**
+       * <pre>
+       * if specified, attr must have type "int"
+       * </pre>
+       *
+       * <code>string number_attr = 5;</code>
+       */
+      public Builder setNumberAttr(
+          java.lang.String value) {
+        if (value == null) {
+    throw new NullPointerException();
+  }
+  
+        numberAttr_ = value;
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * if specified, attr must have type "int"
+       * </pre>
+       *
+       * <code>string number_attr = 5;</code>
+       */
+      public Builder clearNumberAttr() {
+        
+        numberAttr_ = getDefaultInstance().getNumberAttr();
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * if specified, attr must have type "int"
+       * </pre>
+       *
+       * <code>string number_attr = 5;</code>
+       */
+      public Builder setNumberAttrBytes(
+          com.google.protobuf.ByteString value) {
+        if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+        
+        numberAttr_ = value;
+        onChanged();
+        return this;
+      }
+
+      private java.lang.Object typeListAttr_ = "";
+      /**
+       * <pre>
+       * If specified, attr must have type "list(type)", and none of
+       * type, type_attr, and number_attr may be specified.
+       * </pre>
+       *
+       * <code>string type_list_attr = 6;</code>
+       */
+      public java.lang.String getTypeListAttr() {
+        java.lang.Object ref = typeListAttr_;
+        if (!(ref instanceof java.lang.String)) {
+          com.google.protobuf.ByteString bs =
+              (com.google.protobuf.ByteString) ref;
+          java.lang.String s = bs.toStringUtf8();
+          typeListAttr_ = s;
+          return s;
+        } else {
+          return (java.lang.String) ref;
+        }
+      }
+      /**
+       * <pre>
+       * If specified, attr must have type "list(type)", and none of
+       * type, type_attr, and number_attr may be specified.
+       * </pre>
+       *
+       * <code>string type_list_attr = 6;</code>
+       */
+      public com.google.protobuf.ByteString
+          getTypeListAttrBytes() {
+        java.lang.Object ref = typeListAttr_;
+        if (ref instanceof String) {
+          com.google.protobuf.ByteString b = 
+              com.google.protobuf.ByteString.copyFromUtf8(
+                  (java.lang.String) ref);
+          typeListAttr_ = b;
+          return b;
+        } else {
+          return (com.google.protobuf.ByteString) ref;
+        }
+      }
+      /**
+       * <pre>
+       * If specified, attr must have type "list(type)", and none of
+       * type, type_attr, and number_attr may be specified.
+       * </pre>
+       *
+       * <code>string type_list_attr = 6;</code>
+       */
+      public Builder setTypeListAttr(
+          java.lang.String value) {
+        if (value == null) {
+    throw new NullPointerException();
+  }
+  
+        typeListAttr_ = value;
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * If specified, attr must have type "list(type)", and none of
+       * type, type_attr, and number_attr may be specified.
+       * </pre>
+       *
+       * <code>string type_list_attr = 6;</code>
+       */
+      public Builder clearTypeListAttr() {
+        
+        typeListAttr_ = getDefaultInstance().getTypeListAttr();
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * If specified, attr must have type "list(type)", and none of
+       * type, type_attr, and number_attr may be specified.
+       * </pre>
+       *
+       * <code>string type_list_attr = 6;</code>
+       */
+      public Builder setTypeListAttrBytes(
+          com.google.protobuf.ByteString value) {
+        if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+        
+        typeListAttr_ = value;
+        onChanged();
+        return this;
+      }
+
+      private boolean isRef_ ;
+      /**
+       * <pre>
+       * For inputs: if true, the inputs are required to be refs.
+       *   By default, inputs can be either refs or non-refs.
+       * For outputs: if true, outputs are refs, otherwise they are not.
+       * </pre>
+       *
+       * <code>bool is_ref = 16;</code>
+       */
+      public boolean getIsRef() {
+        return isRef_;
+      }
+      /**
+       * <pre>
+       * For inputs: if true, the inputs are required to be refs.
+       *   By default, inputs can be either refs or non-refs.
+       * For outputs: if true, outputs are refs, otherwise they are not.
+       * </pre>
+       *
+       * <code>bool is_ref = 16;</code>
+       */
+      public Builder setIsRef(boolean value) {
+        
+        isRef_ = value;
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * For inputs: if true, the inputs are required to be refs.
+       *   By default, inputs can be either refs or non-refs.
+       * For outputs: if true, outputs are refs, otherwise they are not.
+       * </pre>
+       *
+       * <code>bool is_ref = 16;</code>
+       */
+      public Builder clearIsRef() {
+        
+        isRef_ = false;
+        onChanged();
+        return this;
+      }
+      @java.lang.Override
+      public final Builder setUnknownFields(
+          final com.google.protobuf.UnknownFieldSet unknownFields) {
+        return super.setUnknownFields(unknownFields);
+      }
+
+      @java.lang.Override
+      public final Builder mergeUnknownFields(
+          final com.google.protobuf.UnknownFieldSet unknownFields) {
+        return super.mergeUnknownFields(unknownFields);
+      }
+
+
+      // @@protoc_insertion_point(builder_scope:tensorflow.OpDef.ArgDef)
+    }
+
+    // @@protoc_insertion_point(class_scope:tensorflow.OpDef.ArgDef)
+    private static final org.tensorflow.proto.framework.OpDef.ArgDef DEFAULT_INSTANCE;
+    static {
+      DEFAULT_INSTANCE = new org.tensorflow.proto.framework.OpDef.ArgDef();
+    }
+
+    public static org.tensorflow.proto.framework.OpDef.ArgDef getDefaultInstance() {
+      return DEFAULT_INSTANCE;
+    }
+
+    private static final com.google.protobuf.Parser<ArgDef>
+        PARSER = new com.google.protobuf.AbstractParser<ArgDef>() {
+      @java.lang.Override
+      public ArgDef parsePartialFrom(
+          com.google.protobuf.CodedInputStream input,
+          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+          throws com.google.protobuf.InvalidProtocolBufferException {
+        return new ArgDef(input, extensionRegistry);
+      }
+    };
+
+    public static com.google.protobuf.Parser<ArgDef> parser() {
+      return PARSER;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Parser<ArgDef> getParserForType() {
+      return PARSER;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.OpDef.ArgDef getDefaultInstanceForType() {
+      return DEFAULT_INSTANCE;
+    }
+
+  }
+
+  public interface AttrDefOrBuilder extends
+      // @@protoc_insertion_point(interface_extends:tensorflow.OpDef.AttrDef)
+      com.google.protobuf.MessageOrBuilder {
+
+    /**
+     * <pre>
+     * A descriptive name for the argument.  May be used, e.g. by the
+     * Python client, as a keyword argument name, and so should match
+     * the regexp "[a-z][a-z0-9_]+".
+     * </pre>
+     *
+     * <code>string name = 1;</code>
+     */
+    java.lang.String getName();
+    /**
+     * <pre>
+     * A descriptive name for the argument.  May be used, e.g. by the
+     * Python client, as a keyword argument name, and so should match
+     * the regexp "[a-z][a-z0-9_]+".
+     * </pre>
+     *
+     * <code>string name = 1;</code>
+     */
+    com.google.protobuf.ByteString
+        getNameBytes();
+
+    /**
+     * <pre>
+     * One of the type names from attr_value.proto ("string", "list(string)",
+     * "int", etc.).
+     * </pre>
+     *
+     * <code>string type = 2;</code>
+     */
+    java.lang.String getType();
+    /**
+     * <pre>
+     * One of the type names from attr_value.proto ("string", "list(string)",
+     * "int", etc.).
+     * </pre>
+     *
+     * <code>string type = 2;</code>
+     */
+    com.google.protobuf.ByteString
+        getTypeBytes();
+
+    /**
+     * <pre>
+     * A reasonable default for this attribute if the user does not supply
+     * a value.  If not specified, the user must supply a value.
+     * </pre>
+     *
+     * <code>.tensorflow.AttrValue default_value = 3;</code>
+     */
+    boolean hasDefaultValue();
+    /**
+     * <pre>
+     * A reasonable default for this attribute if the user does not supply
+     * a value.  If not specified, the user must supply a value.
+     * </pre>
+     *
+     * <code>.tensorflow.AttrValue default_value = 3;</code>
+     */
+    org.tensorflow.proto.framework.AttrValue getDefaultValue();
+    /**
+     * <pre>
+     * A reasonable default for this attribute if the user does not supply
+     * a value.  If not specified, the user must supply a value.
+     * </pre>
+     *
+     * <code>.tensorflow.AttrValue default_value = 3;</code>
+     */
+    org.tensorflow.proto.framework.AttrValueOrBuilder getDefaultValueOrBuilder();
+
+    /**
+     * <pre>
+     * Human-readable description.
+     * </pre>
+     *
+     * <code>string description = 4;</code>
+     */
+    java.lang.String getDescription();
+    /**
+     * <pre>
+     * Human-readable description.
+     * </pre>
+     *
+     * <code>string description = 4;</code>
+     */
+    com.google.protobuf.ByteString
+        getDescriptionBytes();
+
+    /**
+     * <pre>
+     * For type == "int", this is a minimum value.  For "list(___)"
+     * types, this is the minimum length.
+     * </pre>
+     *
+     * <code>bool has_minimum = 5;</code>
+     */
+    boolean getHasMinimum();
+
+    /**
+     * <code>int64 minimum = 6;</code>
+     */
+    long getMinimum();
+
+    /**
+     * <pre>
+     * The set of allowed values.  Has type that is the "list" version
+     * of the "type" field above (uses the "list" field of AttrValue).
+     * If type == "type" or "list(type)" above, then the "type" field
+     * of "allowed_values.list" has the set of allowed DataTypes.
+     * If type == "string" or "list(string)", then the "s" field of
+     * "allowed_values.list" has the set of allowed strings.
+     * </pre>
+     *
+     * <code>.tensorflow.AttrValue allowed_values = 7;</code>
+     */
+    boolean hasAllowedValues();
+    /**
+     * <pre>
+     * The set of allowed values.  Has type that is the "list" version
+     * of the "type" field above (uses the "list" field of AttrValue).
+     * If type == "type" or "list(type)" above, then the "type" field
+     * of "allowed_values.list" has the set of allowed DataTypes.
+     * If type == "string" or "list(string)", then the "s" field of
+     * "allowed_values.list" has the set of allowed strings.
+     * </pre>
+     *
+     * <code>.tensorflow.AttrValue allowed_values = 7;</code>
+     */
+    org.tensorflow.proto.framework.AttrValue getAllowedValues();
+    /**
+     * <pre>
+     * The set of allowed values.  Has type that is the "list" version
+     * of the "type" field above (uses the "list" field of AttrValue).
+     * If type == "type" or "list(type)" above, then the "type" field
+     * of "allowed_values.list" has the set of allowed DataTypes.
+     * If type == "string" or "list(string)", then the "s" field of
+     * "allowed_values.list" has the set of allowed strings.
+     * </pre>
+     *
+     * <code>.tensorflow.AttrValue allowed_values = 7;</code>
+     */
+    org.tensorflow.proto.framework.AttrValueOrBuilder getAllowedValuesOrBuilder();
+  }
+  /**
+   * <pre>
+   * Description of the graph-construction-time configuration of this
+   * Op.  That is to say, this describes the attr fields that will
+   * be specified in the NodeDef.
+   * </pre>
+   *
+   * Protobuf type {@code tensorflow.OpDef.AttrDef}
+   */
+  public  static final class AttrDef extends
+      com.google.protobuf.GeneratedMessageV3 implements
+      // @@protoc_insertion_point(message_implements:tensorflow.OpDef.AttrDef)
+      AttrDefOrBuilder {
+  private static final long serialVersionUID = 0L;
+    // Use AttrDef.newBuilder() to construct.
+    private AttrDef(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+      super(builder);
+    }
+    private AttrDef() {
+      name_ = "";
+      type_ = "";
+      description_ = "";
+    }
+
+    @java.lang.Override
+    @SuppressWarnings({"unused"})
+    protected java.lang.Object newInstance(
+        UnusedPrivateParameter unused) {
+      return new AttrDef();
+    }
+
+    @java.lang.Override
+    public final com.google.protobuf.UnknownFieldSet
+    getUnknownFields() {
+      return this.unknownFields;
+    }
+    private AttrDef(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      this();
+      if (extensionRegistry == null) {
+        throw new java.lang.NullPointerException();
+      }
+      com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+          com.google.protobuf.UnknownFieldSet.newBuilder();
+      try {
+        boolean done = false;
+        while (!done) {
+          int tag = input.readTag();
+          switch (tag) {
+            case 0:
+              done = true;
+              break;
+            case 10: {
+              java.lang.String s = input.readStringRequireUtf8();
+
+              name_ = s;
+              break;
+            }
+            case 18: {
+              java.lang.String s = input.readStringRequireUtf8();
+
+              type_ = s;
+              break;
+            }
+            case 26: {
+              org.tensorflow.proto.framework.AttrValue.Builder subBuilder = null;
+              if (defaultValue_ != null) {
+                subBuilder = defaultValue_.toBuilder();
+              }
+              defaultValue_ = input.readMessage(org.tensorflow.proto.framework.AttrValue.parser(), extensionRegistry);
+              if (subBuilder != null) {
+                subBuilder.mergeFrom(defaultValue_);
+                defaultValue_ = subBuilder.buildPartial();
+              }
+
+              break;
+            }
+            case 34: {
+              java.lang.String s = input.readStringRequireUtf8();
+
+              description_ = s;
+              break;
+            }
+            case 40: {
+
+              hasMinimum_ = input.readBool();
+              break;
+            }
+            case 48: {
+
+              minimum_ = input.readInt64();
+              break;
+            }
+            case 58: {
+              org.tensorflow.proto.framework.AttrValue.Builder subBuilder = null;
+              if (allowedValues_ != null) {
+                subBuilder = allowedValues_.toBuilder();
+              }
+              allowedValues_ = input.readMessage(org.tensorflow.proto.framework.AttrValue.parser(), extensionRegistry);
+              if (subBuilder != null) {
+                subBuilder.mergeFrom(allowedValues_);
+                allowedValues_ = subBuilder.buildPartial();
+              }
+
+              break;
+            }
+            default: {
+              if (!parseUnknownField(
+                  input, unknownFields, extensionRegistry, tag)) {
+                done = true;
+              }
+              break;
+            }
+          }
+        }
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        throw e.setUnfinishedMessage(this);
+      } catch (java.io.IOException e) {
+        throw new com.google.protobuf.InvalidProtocolBufferException(
+            e).setUnfinishedMessage(this);
+      } finally {
+        this.unknownFields = unknownFields.build();
+        makeExtensionsImmutable();
+      }
+    }
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.OpDefProtos.internal_static_tensorflow_OpDef_AttrDef_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.OpDefProtos.internal_static_tensorflow_OpDef_AttrDef_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.OpDef.AttrDef.class, org.tensorflow.proto.framework.OpDef.AttrDef.Builder.class);
+    }
+
+    public static final int NAME_FIELD_NUMBER = 1;
+    private volatile java.lang.Object name_;
+    /**
+     * <pre>
+     * A descriptive name for the argument.  May be used, e.g. by the
+     * Python client, as a keyword argument name, and so should match
+     * the regexp "[a-z][a-z0-9_]+".
+     * </pre>
+     *
+     * <code>string name = 1;</code>
+     */
+    public java.lang.String getName() {
+      java.lang.Object ref = name_;
+      if (ref instanceof java.lang.String) {
+        return (java.lang.String) ref;
+      } else {
+        com.google.protobuf.ByteString bs = 
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        name_ = s;
+        return s;
+      }
+    }
+    /**
+     * <pre>
+     * A descriptive name for the argument.  May be used, e.g. by the
+     * Python client, as a keyword argument name, and so should match
+     * the regexp "[a-z][a-z0-9_]+".
+     * </pre>
+     *
+     * <code>string name = 1;</code>
+     */
+    public com.google.protobuf.ByteString
+        getNameBytes() {
+      java.lang.Object ref = name_;
+      if (ref instanceof java.lang.String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        name_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+
+    public static final int TYPE_FIELD_NUMBER = 2;
+    private volatile java.lang.Object type_;
+    /**
+     * <pre>
+     * One of the type names from attr_value.proto ("string", "list(string)",
+     * "int", etc.).
+     * </pre>
+     *
+     * <code>string type = 2;</code>
+     */
+    public java.lang.String getType() {
+      java.lang.Object ref = type_;
+      if (ref instanceof java.lang.String) {
+        return (java.lang.String) ref;
+      } else {
+        com.google.protobuf.ByteString bs = 
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        type_ = s;
+        return s;
+      }
+    }
+    /**
+     * <pre>
+     * One of the type names from attr_value.proto ("string", "list(string)",
+     * "int", etc.).
+     * </pre>
+     *
+     * <code>string type = 2;</code>
+     */
+    public com.google.protobuf.ByteString
+        getTypeBytes() {
+      java.lang.Object ref = type_;
+      if (ref instanceof java.lang.String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        type_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+
+    public static final int DEFAULT_VALUE_FIELD_NUMBER = 3;
+    private org.tensorflow.proto.framework.AttrValue defaultValue_;
+    /**
+     * <pre>
+     * A reasonable default for this attribute if the user does not supply
+     * a value.  If not specified, the user must supply a value.
+     * </pre>
+     *
+     * <code>.tensorflow.AttrValue default_value = 3;</code>
+     */
+    public boolean hasDefaultValue() {
+      return defaultValue_ != null;
+    }
+    /**
+     * <pre>
+     * A reasonable default for this attribute if the user does not supply
+     * a value.  If not specified, the user must supply a value.
+     * </pre>
+     *
+     * <code>.tensorflow.AttrValue default_value = 3;</code>
+     */
+    public org.tensorflow.proto.framework.AttrValue getDefaultValue() {
+      return defaultValue_ == null ? org.tensorflow.proto.framework.AttrValue.getDefaultInstance() : defaultValue_;
+    }
+    /**
+     * <pre>
+     * A reasonable default for this attribute if the user does not supply
+     * a value.  If not specified, the user must supply a value.
+     * </pre>
+     *
+     * <code>.tensorflow.AttrValue default_value = 3;</code>
+     */
+    public org.tensorflow.proto.framework.AttrValueOrBuilder getDefaultValueOrBuilder() {
+      return getDefaultValue();
+    }
+
+    public static final int DESCRIPTION_FIELD_NUMBER = 4;
+    private volatile java.lang.Object description_;
+    /**
+     * <pre>
+     * Human-readable description.
+     * </pre>
+     *
+     * <code>string description = 4;</code>
+     */
+    public java.lang.String getDescription() {
+      java.lang.Object ref = description_;
+      if (ref instanceof java.lang.String) {
+        return (java.lang.String) ref;
+      } else {
+        com.google.protobuf.ByteString bs = 
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        description_ = s;
+        return s;
+      }
+    }
+    /**
+     * <pre>
+     * Human-readable description.
+     * </pre>
+     *
+     * <code>string description = 4;</code>
+     */
+    public com.google.protobuf.ByteString
+        getDescriptionBytes() {
+      java.lang.Object ref = description_;
+      if (ref instanceof java.lang.String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        description_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+
+    public static final int HAS_MINIMUM_FIELD_NUMBER = 5;
+    private boolean hasMinimum_;
+    /**
+     * <pre>
+     * For type == "int", this is a minimum value.  For "list(___)"
+     * types, this is the minimum length.
+     * </pre>
+     *
+     * <code>bool has_minimum = 5;</code>
+     */
+    public boolean getHasMinimum() {
+      return hasMinimum_;
+    }
+
+    public static final int MINIMUM_FIELD_NUMBER = 6;
+    private long minimum_;
+    /**
+     * <code>int64 minimum = 6;</code>
+     */
+    public long getMinimum() {
+      return minimum_;
+    }
+
+    public static final int ALLOWED_VALUES_FIELD_NUMBER = 7;
+    private org.tensorflow.proto.framework.AttrValue allowedValues_;
+    /**
+     * <pre>
+     * The set of allowed values.  Has type that is the "list" version
+     * of the "type" field above (uses the "list" field of AttrValue).
+     * If type == "type" or "list(type)" above, then the "type" field
+     * of "allowed_values.list" has the set of allowed DataTypes.
+     * If type == "string" or "list(string)", then the "s" field of
+     * "allowed_values.list" has the set of allowed strings.
+     * </pre>
+     *
+     * <code>.tensorflow.AttrValue allowed_values = 7;</code>
+     */
+    public boolean hasAllowedValues() {
+      return allowedValues_ != null;
+    }
+    /**
+     * <pre>
+     * The set of allowed values.  Has type that is the "list" version
+     * of the "type" field above (uses the "list" field of AttrValue).
+     * If type == "type" or "list(type)" above, then the "type" field
+     * of "allowed_values.list" has the set of allowed DataTypes.
+     * If type == "string" or "list(string)", then the "s" field of
+     * "allowed_values.list" has the set of allowed strings.
+     * </pre>
+     *
+     * <code>.tensorflow.AttrValue allowed_values = 7;</code>
+     */
+    public org.tensorflow.proto.framework.AttrValue getAllowedValues() {
+      return allowedValues_ == null ? org.tensorflow.proto.framework.AttrValue.getDefaultInstance() : allowedValues_;
+    }
+    /**
+     * <pre>
+     * The set of allowed values.  Has type that is the "list" version
+     * of the "type" field above (uses the "list" field of AttrValue).
+     * If type == "type" or "list(type)" above, then the "type" field
+     * of "allowed_values.list" has the set of allowed DataTypes.
+     * If type == "string" or "list(string)", then the "s" field of
+     * "allowed_values.list" has the set of allowed strings.
+     * </pre>
+     *
+     * <code>.tensorflow.AttrValue allowed_values = 7;</code>
+     */
+    public org.tensorflow.proto.framework.AttrValueOrBuilder getAllowedValuesOrBuilder() {
+      return getAllowedValues();
+    }
+
+    private byte memoizedIsInitialized = -1;
+    @java.lang.Override
+    public final boolean isInitialized() {
+      byte isInitialized = memoizedIsInitialized;
+      if (isInitialized == 1) return true;
+      if (isInitialized == 0) return false;
+
+      memoizedIsInitialized = 1;
+      return true;
+    }
+
+    @java.lang.Override
+    public void writeTo(com.google.protobuf.CodedOutputStream output)
+                        throws java.io.IOException {
+      if (!getNameBytes().isEmpty()) {
+        com.google.protobuf.GeneratedMessageV3.writeString(output, 1, name_);
+      }
+      if (!getTypeBytes().isEmpty()) {
+        com.google.protobuf.GeneratedMessageV3.writeString(output, 2, type_);
+      }
+      if (defaultValue_ != null) {
+        output.writeMessage(3, getDefaultValue());
+      }
+      if (!getDescriptionBytes().isEmpty()) {
+        com.google.protobuf.GeneratedMessageV3.writeString(output, 4, description_);
+      }
+      if (hasMinimum_ != false) {
+        output.writeBool(5, hasMinimum_);
+      }
+      if (minimum_ != 0L) {
+        output.writeInt64(6, minimum_);
+      }
+      if (allowedValues_ != null) {
+        output.writeMessage(7, getAllowedValues());
+      }
+      unknownFields.writeTo(output);
+    }
+
+    @java.lang.Override
+    public int getSerializedSize() {
+      int size = memoizedSize;
+      if (size != -1) return size;
+
+      size = 0;
+      if (!getNameBytes().isEmpty()) {
+        size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, name_);
+      }
+      if (!getTypeBytes().isEmpty()) {
+        size += com.google.protobuf.GeneratedMessageV3.computeStringSize(2, type_);
+      }
+      if (defaultValue_ != null) {
+        size += com.google.protobuf.CodedOutputStream
+          .computeMessageSize(3, getDefaultValue());
+      }
+      if (!getDescriptionBytes().isEmpty()) {
+        size += com.google.protobuf.GeneratedMessageV3.computeStringSize(4, description_);
+      }
+      if (hasMinimum_ != false) {
+        size += com.google.protobuf.CodedOutputStream
+          .computeBoolSize(5, hasMinimum_);
+      }
+      if (minimum_ != 0L) {
+        size += com.google.protobuf.CodedOutputStream
+          .computeInt64Size(6, minimum_);
+      }
+      if (allowedValues_ != null) {
+        size += com.google.protobuf.CodedOutputStream
+          .computeMessageSize(7, getAllowedValues());
+      }
+      size += unknownFields.getSerializedSize();
+      memoizedSize = size;
+      return size;
+    }
+
+    @java.lang.Override
+    public boolean equals(final java.lang.Object obj) {
+      if (obj == this) {
+       return true;
+      }
+      if (!(obj instanceof org.tensorflow.proto.framework.OpDef.AttrDef)) {
+        return super.equals(obj);
+      }
+      org.tensorflow.proto.framework.OpDef.AttrDef other = (org.tensorflow.proto.framework.OpDef.AttrDef) obj;
+
+      if (!getName()
+          .equals(other.getName())) return false;
+      if (!getType()
+          .equals(other.getType())) return false;
+      if (hasDefaultValue() != other.hasDefaultValue()) return false;
+      if (hasDefaultValue()) {
+        if (!getDefaultValue()
+            .equals(other.getDefaultValue())) return false;
+      }
+      if (!getDescription()
+          .equals(other.getDescription())) return false;
+      if (getHasMinimum()
+          != other.getHasMinimum()) return false;
+      if (getMinimum()
+          != other.getMinimum()) return false;
+      if (hasAllowedValues() != other.hasAllowedValues()) return false;
+      if (hasAllowedValues()) {
+        if (!getAllowedValues()
+            .equals(other.getAllowedValues())) return false;
+      }
+      if (!unknownFields.equals(other.unknownFields)) return false;
+      return true;
+    }
+
+    @java.lang.Override
+    public int hashCode() {
+      if (memoizedHashCode != 0) {
+        return memoizedHashCode;
+      }
+      int hash = 41;
+      hash = (19 * hash) + getDescriptor().hashCode();
+      hash = (37 * hash) + NAME_FIELD_NUMBER;
+      hash = (53 * hash) + getName().hashCode();
+      hash = (37 * hash) + TYPE_FIELD_NUMBER;
+      hash = (53 * hash) + getType().hashCode();
+      if (hasDefaultValue()) {
+        hash = (37 * hash) + DEFAULT_VALUE_FIELD_NUMBER;
+        hash = (53 * hash) + getDefaultValue().hashCode();
+      }
+      hash = (37 * hash) + DESCRIPTION_FIELD_NUMBER;
+      hash = (53 * hash) + getDescription().hashCode();
+      hash = (37 * hash) + HAS_MINIMUM_FIELD_NUMBER;
+      hash = (53 * hash) + com.google.protobuf.Internal.hashBoolean(
+          getHasMinimum());
+      hash = (37 * hash) + MINIMUM_FIELD_NUMBER;
+      hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+          getMinimum());
+      if (hasAllowedValues()) {
+        hash = (37 * hash) + ALLOWED_VALUES_FIELD_NUMBER;
+        hash = (53 * hash) + getAllowedValues().hashCode();
+      }
+      hash = (29 * hash) + unknownFields.hashCode();
+      memoizedHashCode = hash;
+      return hash;
+    }
+
+    public static org.tensorflow.proto.framework.OpDef.AttrDef parseFrom(
+        java.nio.ByteBuffer data)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data);
+    }
+    public static org.tensorflow.proto.framework.OpDef.AttrDef parseFrom(
+        java.nio.ByteBuffer data,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.OpDef.AttrDef parseFrom(
+        com.google.protobuf.ByteString data)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data);
+    }
+    public static org.tensorflow.proto.framework.OpDef.AttrDef parseFrom(
+        com.google.protobuf.ByteString data,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.OpDef.AttrDef parseFrom(byte[] data)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data);
+    }
+    public static org.tensorflow.proto.framework.OpDef.AttrDef parseFrom(
+        byte[] data,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.OpDef.AttrDef parseFrom(java.io.InputStream input)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input);
+    }
+    public static org.tensorflow.proto.framework.OpDef.AttrDef parseFrom(
+        java.io.InputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.OpDef.AttrDef parseDelimitedFrom(java.io.InputStream input)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseDelimitedWithIOException(PARSER, input);
+    }
+    public static org.tensorflow.proto.framework.OpDef.AttrDef parseDelimitedFrom(
+        java.io.InputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.OpDef.AttrDef parseFrom(
+        com.google.protobuf.CodedInputStream input)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input);
+    }
+    public static org.tensorflow.proto.framework.OpDef.AttrDef parseFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input, extensionRegistry);
+    }
+
+    @java.lang.Override
+    public Builder newBuilderForType() { return newBuilder(); }
+    public static Builder newBuilder() {
+      return DEFAULT_INSTANCE.toBuilder();
+    }
+    public static Builder newBuilder(org.tensorflow.proto.framework.OpDef.AttrDef prototype) {
+      return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+    }
+    @java.lang.Override
+    public Builder toBuilder() {
+      return this == DEFAULT_INSTANCE
+          ? new Builder() : new Builder().mergeFrom(this);
+    }
+
+    @java.lang.Override
+    protected Builder newBuilderForType(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      Builder builder = new Builder(parent);
+      return builder;
+    }
+    /**
+     * <pre>
+     * Description of the graph-construction-time configuration of this
+     * Op.  That is to say, this describes the attr fields that will
+     * be specified in the NodeDef.
+     * </pre>
+     *
+     * Protobuf type {@code tensorflow.OpDef.AttrDef}
+     */
+    public static final class Builder extends
+        com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+        // @@protoc_insertion_point(builder_implements:tensorflow.OpDef.AttrDef)
+        org.tensorflow.proto.framework.OpDef.AttrDefOrBuilder {
+      public static final com.google.protobuf.Descriptors.Descriptor
+          getDescriptor() {
+        return org.tensorflow.proto.framework.OpDefProtos.internal_static_tensorflow_OpDef_AttrDef_descriptor;
+      }
+
+      @java.lang.Override
+      protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+          internalGetFieldAccessorTable() {
+        return org.tensorflow.proto.framework.OpDefProtos.internal_static_tensorflow_OpDef_AttrDef_fieldAccessorTable
+            .ensureFieldAccessorsInitialized(
+                org.tensorflow.proto.framework.OpDef.AttrDef.class, org.tensorflow.proto.framework.OpDef.AttrDef.Builder.class);
+      }
+
+      // Construct using org.tensorflow.proto.framework.OpDef.AttrDef.newBuilder()
+      private Builder() {
+        maybeForceBuilderInitialization();
+      }
+
+      private Builder(
+          com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+        super(parent);
+        maybeForceBuilderInitialization();
+      }
+      private void maybeForceBuilderInitialization() {
+        if (com.google.protobuf.GeneratedMessageV3
+                .alwaysUseFieldBuilders) {
+        }
+      }
+      @java.lang.Override
+      public Builder clear() {
+        super.clear();
+        name_ = "";
+
+        type_ = "";
+
+        if (defaultValueBuilder_ == null) {
+          defaultValue_ = null;
+        } else {
+          defaultValue_ = null;
+          defaultValueBuilder_ = null;
+        }
+        description_ = "";
+
+        hasMinimum_ = false;
+
+        minimum_ = 0L;
+
+        if (allowedValuesBuilder_ == null) {
+          allowedValues_ = null;
+        } else {
+          allowedValues_ = null;
+          allowedValuesBuilder_ = null;
+        }
+        return this;
+      }
+
+      @java.lang.Override
+      public com.google.protobuf.Descriptors.Descriptor
+          getDescriptorForType() {
+        return org.tensorflow.proto.framework.OpDefProtos.internal_static_tensorflow_OpDef_AttrDef_descriptor;
+      }
+
+      @java.lang.Override
+      public org.tensorflow.proto.framework.OpDef.AttrDef getDefaultInstanceForType() {
+        return org.tensorflow.proto.framework.OpDef.AttrDef.getDefaultInstance();
+      }
+
+      @java.lang.Override
+      public org.tensorflow.proto.framework.OpDef.AttrDef build() {
+        org.tensorflow.proto.framework.OpDef.AttrDef result = buildPartial();
+        if (!result.isInitialized()) {
+          throw newUninitializedMessageException(result);
+        }
+        return result;
+      }
+
+      @java.lang.Override
+      public org.tensorflow.proto.framework.OpDef.AttrDef buildPartial() {
+        org.tensorflow.proto.framework.OpDef.AttrDef result = new org.tensorflow.proto.framework.OpDef.AttrDef(this);
+        result.name_ = name_;
+        result.type_ = type_;
+        if (defaultValueBuilder_ == null) {
+          result.defaultValue_ = defaultValue_;
+        } else {
+          result.defaultValue_ = defaultValueBuilder_.build();
+        }
+        result.description_ = description_;
+        result.hasMinimum_ = hasMinimum_;
+        result.minimum_ = minimum_;
+        if (allowedValuesBuilder_ == null) {
+          result.allowedValues_ = allowedValues_;
+        } else {
+          result.allowedValues_ = allowedValuesBuilder_.build();
+        }
+        onBuilt();
+        return result;
+      }
+
+      @java.lang.Override
+      public Builder clone() {
+        return super.clone();
+      }
+      @java.lang.Override
+      public Builder setField(
+          com.google.protobuf.Descriptors.FieldDescriptor field,
+          java.lang.Object value) {
+        return super.setField(field, value);
+      }
+      @java.lang.Override
+      public Builder clearField(
+          com.google.protobuf.Descriptors.FieldDescriptor field) {
+        return super.clearField(field);
+      }
+      @java.lang.Override
+      public Builder clearOneof(
+          com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+        return super.clearOneof(oneof);
+      }
+      @java.lang.Override
+      public Builder setRepeatedField(
+          com.google.protobuf.Descriptors.FieldDescriptor field,
+          int index, java.lang.Object value) {
+        return super.setRepeatedField(field, index, value);
+      }
+      @java.lang.Override
+      public Builder addRepeatedField(
+          com.google.protobuf.Descriptors.FieldDescriptor field,
+          java.lang.Object value) {
+        return super.addRepeatedField(field, value);
+      }
+      @java.lang.Override
+      public Builder mergeFrom(com.google.protobuf.Message other) {
+        if (other instanceof org.tensorflow.proto.framework.OpDef.AttrDef) {
+          return mergeFrom((org.tensorflow.proto.framework.OpDef.AttrDef)other);
+        } else {
+          super.mergeFrom(other);
+          return this;
+        }
+      }
+
+      public Builder mergeFrom(org.tensorflow.proto.framework.OpDef.AttrDef other) {
+        if (other == org.tensorflow.proto.framework.OpDef.AttrDef.getDefaultInstance()) return this;
+        if (!other.getName().isEmpty()) {
+          name_ = other.name_;
+          onChanged();
+        }
+        if (!other.getType().isEmpty()) {
+          type_ = other.type_;
+          onChanged();
+        }
+        if (other.hasDefaultValue()) {
+          mergeDefaultValue(other.getDefaultValue());
+        }
+        if (!other.getDescription().isEmpty()) {
+          description_ = other.description_;
+          onChanged();
+        }
+        if (other.getHasMinimum() != false) {
+          setHasMinimum(other.getHasMinimum());
+        }
+        if (other.getMinimum() != 0L) {
+          setMinimum(other.getMinimum());
+        }
+        if (other.hasAllowedValues()) {
+          mergeAllowedValues(other.getAllowedValues());
+        }
+        this.mergeUnknownFields(other.unknownFields);
+        onChanged();
+        return this;
+      }
+
+      @java.lang.Override
+      public final boolean isInitialized() {
+        return true;
+      }
+
+      @java.lang.Override
+      public Builder mergeFrom(
+          com.google.protobuf.CodedInputStream input,
+          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+          throws java.io.IOException {
+        org.tensorflow.proto.framework.OpDef.AttrDef parsedMessage = null;
+        try {
+          parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+        } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+          parsedMessage = (org.tensorflow.proto.framework.OpDef.AttrDef) e.getUnfinishedMessage();
+          throw e.unwrapIOException();
+        } finally {
+          if (parsedMessage != null) {
+            mergeFrom(parsedMessage);
+          }
+        }
+        return this;
+      }
+
+      private java.lang.Object name_ = "";
+      /**
+       * <pre>
+       * A descriptive name for the argument.  May be used, e.g. by the
+       * Python client, as a keyword argument name, and so should match
+       * the regexp "[a-z][a-z0-9_]+".
+       * </pre>
+       *
+       * <code>string name = 1;</code>
+       */
+      public java.lang.String getName() {
+        java.lang.Object ref = name_;
+        if (!(ref instanceof java.lang.String)) {
+          com.google.protobuf.ByteString bs =
+              (com.google.protobuf.ByteString) ref;
+          java.lang.String s = bs.toStringUtf8();
+          name_ = s;
+          return s;
+        } else {
+          return (java.lang.String) ref;
+        }
+      }
+      /**
+       * <pre>
+       * A descriptive name for the argument.  May be used, e.g. by the
+       * Python client, as a keyword argument name, and so should match
+       * the regexp "[a-z][a-z0-9_]+".
+       * </pre>
+       *
+       * <code>string name = 1;</code>
+       */
+      public com.google.protobuf.ByteString
+          getNameBytes() {
+        java.lang.Object ref = name_;
+        if (ref instanceof String) {
+          com.google.protobuf.ByteString b = 
+              com.google.protobuf.ByteString.copyFromUtf8(
+                  (java.lang.String) ref);
+          name_ = b;
+          return b;
+        } else {
+          return (com.google.protobuf.ByteString) ref;
+        }
+      }
+      /**
+       * <pre>
+       * A descriptive name for the argument.  May be used, e.g. by the
+       * Python client, as a keyword argument name, and so should match
+       * the regexp "[a-z][a-z0-9_]+".
+       * </pre>
+       *
+       * <code>string name = 1;</code>
+       */
+      public Builder setName(
+          java.lang.String value) {
+        if (value == null) {
+    throw new NullPointerException();
+  }
+  
+        name_ = value;
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * A descriptive name for the argument.  May be used, e.g. by the
+       * Python client, as a keyword argument name, and so should match
+       * the regexp "[a-z][a-z0-9_]+".
+       * </pre>
+       *
+       * <code>string name = 1;</code>
+       */
+      public Builder clearName() {
+        
+        name_ = getDefaultInstance().getName();
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * A descriptive name for the argument.  May be used, e.g. by the
+       * Python client, as a keyword argument name, and so should match
+       * the regexp "[a-z][a-z0-9_]+".
+       * </pre>
+       *
+       * <code>string name = 1;</code>
+       */
+      public Builder setNameBytes(
+          com.google.protobuf.ByteString value) {
+        if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+        
+        name_ = value;
+        onChanged();
+        return this;
+      }
+
+      private java.lang.Object type_ = "";
+      /**
+       * <pre>
+       * One of the type names from attr_value.proto ("string", "list(string)",
+       * "int", etc.).
+       * </pre>
+       *
+       * <code>string type = 2;</code>
+       */
+      public java.lang.String getType() {
+        java.lang.Object ref = type_;
+        if (!(ref instanceof java.lang.String)) {
+          com.google.protobuf.ByteString bs =
+              (com.google.protobuf.ByteString) ref;
+          java.lang.String s = bs.toStringUtf8();
+          type_ = s;
+          return s;
+        } else {
+          return (java.lang.String) ref;
+        }
+      }
+      /**
+       * <pre>
+       * One of the type names from attr_value.proto ("string", "list(string)",
+       * "int", etc.).
+       * </pre>
+       *
+       * <code>string type = 2;</code>
+       */
+      public com.google.protobuf.ByteString
+          getTypeBytes() {
+        java.lang.Object ref = type_;
+        if (ref instanceof String) {
+          com.google.protobuf.ByteString b = 
+              com.google.protobuf.ByteString.copyFromUtf8(
+                  (java.lang.String) ref);
+          type_ = b;
+          return b;
+        } else {
+          return (com.google.protobuf.ByteString) ref;
+        }
+      }
+      /**
+       * <pre>
+       * One of the type names from attr_value.proto ("string", "list(string)",
+       * "int", etc.).
+       * </pre>
+       *
+       * <code>string type = 2;</code>
+       */
+      public Builder setType(
+          java.lang.String value) {
+        if (value == null) {
+    throw new NullPointerException();
+  }
+  
+        type_ = value;
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * One of the type names from attr_value.proto ("string", "list(string)",
+       * "int", etc.).
+       * </pre>
+       *
+       * <code>string type = 2;</code>
+       */
+      public Builder clearType() {
+        
+        type_ = getDefaultInstance().getType();
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * One of the type names from attr_value.proto ("string", "list(string)",
+       * "int", etc.).
+       * </pre>
+       *
+       * <code>string type = 2;</code>
+       */
+      public Builder setTypeBytes(
+          com.google.protobuf.ByteString value) {
+        if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+        
+        type_ = value;
+        onChanged();
+        return this;
+      }
+
+      private org.tensorflow.proto.framework.AttrValue defaultValue_;
+      private com.google.protobuf.SingleFieldBuilderV3<
+          org.tensorflow.proto.framework.AttrValue, org.tensorflow.proto.framework.AttrValue.Builder, org.tensorflow.proto.framework.AttrValueOrBuilder> defaultValueBuilder_;
+      /**
+       * <pre>
+       * A reasonable default for this attribute if the user does not supply
+       * a value.  If not specified, the user must supply a value.
+       * </pre>
+       *
+       * <code>.tensorflow.AttrValue default_value = 3;</code>
+       */
+      public boolean hasDefaultValue() {
+        return defaultValueBuilder_ != null || defaultValue_ != null;
+      }
+      /**
+       * <pre>
+       * A reasonable default for this attribute if the user does not supply
+       * a value.  If not specified, the user must supply a value.
+       * </pre>
+       *
+       * <code>.tensorflow.AttrValue default_value = 3;</code>
+       */
+      public org.tensorflow.proto.framework.AttrValue getDefaultValue() {
+        if (defaultValueBuilder_ == null) {
+          return defaultValue_ == null ? org.tensorflow.proto.framework.AttrValue.getDefaultInstance() : defaultValue_;
+        } else {
+          return defaultValueBuilder_.getMessage();
+        }
+      }
+      /**
+       * <pre>
+       * A reasonable default for this attribute if the user does not supply
+       * a value.  If not specified, the user must supply a value.
+       * </pre>
+       *
+       * <code>.tensorflow.AttrValue default_value = 3;</code>
+       */
+      public Builder setDefaultValue(org.tensorflow.proto.framework.AttrValue value) {
+        if (defaultValueBuilder_ == null) {
+          if (value == null) {
+            throw new NullPointerException();
+          }
+          defaultValue_ = value;
+          onChanged();
+        } else {
+          defaultValueBuilder_.setMessage(value);
+        }
+
+        return this;
+      }
+      /**
+       * <pre>
+       * A reasonable default for this attribute if the user does not supply
+       * a value.  If not specified, the user must supply a value.
+       * </pre>
+       *
+       * <code>.tensorflow.AttrValue default_value = 3;</code>
+       */
+      public Builder setDefaultValue(
+          org.tensorflow.proto.framework.AttrValue.Builder builderForValue) {
+        if (defaultValueBuilder_ == null) {
+          defaultValue_ = builderForValue.build();
+          onChanged();
+        } else {
+          defaultValueBuilder_.setMessage(builderForValue.build());
+        }
+
+        return this;
+      }
+      /**
+       * <pre>
+       * A reasonable default for this attribute if the user does not supply
+       * a value.  If not specified, the user must supply a value.
+       * </pre>
+       *
+       * <code>.tensorflow.AttrValue default_value = 3;</code>
+       */
+      public Builder mergeDefaultValue(org.tensorflow.proto.framework.AttrValue value) {
+        if (defaultValueBuilder_ == null) {
+          if (defaultValue_ != null) {
+            defaultValue_ =
+              org.tensorflow.proto.framework.AttrValue.newBuilder(defaultValue_).mergeFrom(value).buildPartial();
+          } else {
+            defaultValue_ = value;
+          }
+          onChanged();
+        } else {
+          defaultValueBuilder_.mergeFrom(value);
+        }
+
+        return this;
+      }
+      /**
+       * <pre>
+       * A reasonable default for this attribute if the user does not supply
+       * a value.  If not specified, the user must supply a value.
+       * </pre>
+       *
+       * <code>.tensorflow.AttrValue default_value = 3;</code>
+       */
+      public Builder clearDefaultValue() {
+        if (defaultValueBuilder_ == null) {
+          defaultValue_ = null;
+          onChanged();
+        } else {
+          defaultValue_ = null;
+          defaultValueBuilder_ = null;
+        }
+
+        return this;
+      }
+      /**
+       * <pre>
+       * A reasonable default for this attribute if the user does not supply
+       * a value.  If not specified, the user must supply a value.
+       * </pre>
+       *
+       * <code>.tensorflow.AttrValue default_value = 3;</code>
+       */
+      public org.tensorflow.proto.framework.AttrValue.Builder getDefaultValueBuilder() {
+        
+        onChanged();
+        return getDefaultValueFieldBuilder().getBuilder();
+      }
+      /**
+       * <pre>
+       * A reasonable default for this attribute if the user does not supply
+       * a value.  If not specified, the user must supply a value.
+       * </pre>
+       *
+       * <code>.tensorflow.AttrValue default_value = 3;</code>
+       */
+      public org.tensorflow.proto.framework.AttrValueOrBuilder getDefaultValueOrBuilder() {
+        if (defaultValueBuilder_ != null) {
+          return defaultValueBuilder_.getMessageOrBuilder();
+        } else {
+          return defaultValue_ == null ?
+              org.tensorflow.proto.framework.AttrValue.getDefaultInstance() : defaultValue_;
+        }
+      }
+      /**
+       * <pre>
+       * A reasonable default for this attribute if the user does not supply
+       * a value.  If not specified, the user must supply a value.
+       * </pre>
+       *
+       * <code>.tensorflow.AttrValue default_value = 3;</code>
+       */
+      private com.google.protobuf.SingleFieldBuilderV3<
+          org.tensorflow.proto.framework.AttrValue, org.tensorflow.proto.framework.AttrValue.Builder, org.tensorflow.proto.framework.AttrValueOrBuilder> 
+          getDefaultValueFieldBuilder() {
+        if (defaultValueBuilder_ == null) {
+          defaultValueBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+              org.tensorflow.proto.framework.AttrValue, org.tensorflow.proto.framework.AttrValue.Builder, org.tensorflow.proto.framework.AttrValueOrBuilder>(
+                  getDefaultValue(),
+                  getParentForChildren(),
+                  isClean());
+          defaultValue_ = null;
+        }
+        return defaultValueBuilder_;
+      }
+
+      private java.lang.Object description_ = "";
+      /**
+       * <pre>
+       * Human-readable description.
+       * </pre>
+       *
+       * <code>string description = 4;</code>
+       */
+      public java.lang.String getDescription() {
+        java.lang.Object ref = description_;
+        if (!(ref instanceof java.lang.String)) {
+          com.google.protobuf.ByteString bs =
+              (com.google.protobuf.ByteString) ref;
+          java.lang.String s = bs.toStringUtf8();
+          description_ = s;
+          return s;
+        } else {
+          return (java.lang.String) ref;
+        }
+      }
+      /**
+       * <pre>
+       * Human-readable description.
+       * </pre>
+       *
+       * <code>string description = 4;</code>
+       */
+      public com.google.protobuf.ByteString
+          getDescriptionBytes() {
+        java.lang.Object ref = description_;
+        if (ref instanceof String) {
+          com.google.protobuf.ByteString b = 
+              com.google.protobuf.ByteString.copyFromUtf8(
+                  (java.lang.String) ref);
+          description_ = b;
+          return b;
+        } else {
+          return (com.google.protobuf.ByteString) ref;
+        }
+      }
+      /**
+       * <pre>
+       * Human-readable description.
+       * </pre>
+       *
+       * <code>string description = 4;</code>
+       */
+      public Builder setDescription(
+          java.lang.String value) {
+        if (value == null) {
+    throw new NullPointerException();
+  }
+  
+        description_ = value;
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * Human-readable description.
+       * </pre>
+       *
+       * <code>string description = 4;</code>
+       */
+      public Builder clearDescription() {
+        
+        description_ = getDefaultInstance().getDescription();
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * Human-readable description.
+       * </pre>
+       *
+       * <code>string description = 4;</code>
+       */
+      public Builder setDescriptionBytes(
+          com.google.protobuf.ByteString value) {
+        if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+        
+        description_ = value;
+        onChanged();
+        return this;
+      }
+
+      private boolean hasMinimum_ ;
+      /**
+       * <pre>
+       * For type == "int", this is a minimum value.  For "list(___)"
+       * types, this is the minimum length.
+       * </pre>
+       *
+       * <code>bool has_minimum = 5;</code>
+       */
+      public boolean getHasMinimum() {
+        return hasMinimum_;
+      }
+      /**
+       * <pre>
+       * For type == "int", this is a minimum value.  For "list(___)"
+       * types, this is the minimum length.
+       * </pre>
+       *
+       * <code>bool has_minimum = 5;</code>
+       */
+      public Builder setHasMinimum(boolean value) {
+        
+        hasMinimum_ = value;
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * For type == "int", this is a minimum value.  For "list(___)"
+       * types, this is the minimum length.
+       * </pre>
+       *
+       * <code>bool has_minimum = 5;</code>
+       */
+      public Builder clearHasMinimum() {
+        
+        hasMinimum_ = false;
+        onChanged();
+        return this;
+      }
+
+      private long minimum_ ;
+      /**
+       * <code>int64 minimum = 6;</code>
+       */
+      public long getMinimum() {
+        return minimum_;
+      }
+      /**
+       * <code>int64 minimum = 6;</code>
+       */
+      public Builder setMinimum(long value) {
+        
+        minimum_ = value;
+        onChanged();
+        return this;
+      }
+      /**
+       * <code>int64 minimum = 6;</code>
+       */
+      public Builder clearMinimum() {
+        
+        minimum_ = 0L;
+        onChanged();
+        return this;
+      }
+
+      private org.tensorflow.proto.framework.AttrValue allowedValues_;
+      private com.google.protobuf.SingleFieldBuilderV3<
+          org.tensorflow.proto.framework.AttrValue, org.tensorflow.proto.framework.AttrValue.Builder, org.tensorflow.proto.framework.AttrValueOrBuilder> allowedValuesBuilder_;
+      /**
+       * <pre>
+       * The set of allowed values.  Has type that is the "list" version
+       * of the "type" field above (uses the "list" field of AttrValue).
+       * If type == "type" or "list(type)" above, then the "type" field
+       * of "allowed_values.list" has the set of allowed DataTypes.
+       * If type == "string" or "list(string)", then the "s" field of
+       * "allowed_values.list" has the set of allowed strings.
+       * </pre>
+       *
+       * <code>.tensorflow.AttrValue allowed_values = 7;</code>
+       */
+      public boolean hasAllowedValues() {
+        return allowedValuesBuilder_ != null || allowedValues_ != null;
+      }
+      /**
+       * <pre>
+       * The set of allowed values.  Has type that is the "list" version
+       * of the "type" field above (uses the "list" field of AttrValue).
+       * If type == "type" or "list(type)" above, then the "type" field
+       * of "allowed_values.list" has the set of allowed DataTypes.
+       * If type == "string" or "list(string)", then the "s" field of
+       * "allowed_values.list" has the set of allowed strings.
+       * </pre>
+       *
+       * <code>.tensorflow.AttrValue allowed_values = 7;</code>
+       */
+      public org.tensorflow.proto.framework.AttrValue getAllowedValues() {
+        if (allowedValuesBuilder_ == null) {
+          return allowedValues_ == null ? org.tensorflow.proto.framework.AttrValue.getDefaultInstance() : allowedValues_;
+        } else {
+          return allowedValuesBuilder_.getMessage();
+        }
+      }
+      /**
+       * <pre>
+       * The set of allowed values.  Has type that is the "list" version
+       * of the "type" field above (uses the "list" field of AttrValue).
+       * If type == "type" or "list(type)" above, then the "type" field
+       * of "allowed_values.list" has the set of allowed DataTypes.
+       * If type == "string" or "list(string)", then the "s" field of
+       * "allowed_values.list" has the set of allowed strings.
+       * </pre>
+       *
+       * <code>.tensorflow.AttrValue allowed_values = 7;</code>
+       */
+      public Builder setAllowedValues(org.tensorflow.proto.framework.AttrValue value) {
+        if (allowedValuesBuilder_ == null) {
+          if (value == null) {
+            throw new NullPointerException();
+          }
+          allowedValues_ = value;
+          onChanged();
+        } else {
+          allowedValuesBuilder_.setMessage(value);
+        }
+
+        return this;
+      }
+      /**
+       * <pre>
+       * The set of allowed values.  Has type that is the "list" version
+       * of the "type" field above (uses the "list" field of AttrValue).
+       * If type == "type" or "list(type)" above, then the "type" field
+       * of "allowed_values.list" has the set of allowed DataTypes.
+       * If type == "string" or "list(string)", then the "s" field of
+       * "allowed_values.list" has the set of allowed strings.
+       * </pre>
+       *
+       * <code>.tensorflow.AttrValue allowed_values = 7;</code>
+       */
+      public Builder setAllowedValues(
+          org.tensorflow.proto.framework.AttrValue.Builder builderForValue) {
+        if (allowedValuesBuilder_ == null) {
+          allowedValues_ = builderForValue.build();
+          onChanged();
+        } else {
+          allowedValuesBuilder_.setMessage(builderForValue.build());
+        }
+
+        return this;
+      }
+      /**
+       * <pre>
+       * The set of allowed values.  Has type that is the "list" version
+       * of the "type" field above (uses the "list" field of AttrValue).
+       * If type == "type" or "list(type)" above, then the "type" field
+       * of "allowed_values.list" has the set of allowed DataTypes.
+       * If type == "string" or "list(string)", then the "s" field of
+       * "allowed_values.list" has the set of allowed strings.
+       * </pre>
+       *
+       * <code>.tensorflow.AttrValue allowed_values = 7;</code>
+       */
+      public Builder mergeAllowedValues(org.tensorflow.proto.framework.AttrValue value) {
+        if (allowedValuesBuilder_ == null) {
+          if (allowedValues_ != null) {
+            allowedValues_ =
+              org.tensorflow.proto.framework.AttrValue.newBuilder(allowedValues_).mergeFrom(value).buildPartial();
+          } else {
+            allowedValues_ = value;
+          }
+          onChanged();
+        } else {
+          allowedValuesBuilder_.mergeFrom(value);
+        }
+
+        return this;
+      }
+      /**
+       * <pre>
+       * The set of allowed values.  Has type that is the "list" version
+       * of the "type" field above (uses the "list" field of AttrValue).
+       * If type == "type" or "list(type)" above, then the "type" field
+       * of "allowed_values.list" has the set of allowed DataTypes.
+       * If type == "string" or "list(string)", then the "s" field of
+       * "allowed_values.list" has the set of allowed strings.
+       * </pre>
+       *
+       * <code>.tensorflow.AttrValue allowed_values = 7;</code>
+       */
+      public Builder clearAllowedValues() {
+        if (allowedValuesBuilder_ == null) {
+          allowedValues_ = null;
+          onChanged();
+        } else {
+          allowedValues_ = null;
+          allowedValuesBuilder_ = null;
+        }
+
+        return this;
+      }
+      /**
+       * <pre>
+       * The set of allowed values.  Has type that is the "list" version
+       * of the "type" field above (uses the "list" field of AttrValue).
+       * If type == "type" or "list(type)" above, then the "type" field
+       * of "allowed_values.list" has the set of allowed DataTypes.
+       * If type == "string" or "list(string)", then the "s" field of
+       * "allowed_values.list" has the set of allowed strings.
+       * </pre>
+       *
+       * <code>.tensorflow.AttrValue allowed_values = 7;</code>
+       */
+      public org.tensorflow.proto.framework.AttrValue.Builder getAllowedValuesBuilder() {
+        
+        onChanged();
+        return getAllowedValuesFieldBuilder().getBuilder();
+      }
+      /**
+       * <pre>
+       * The set of allowed values.  Has type that is the "list" version
+       * of the "type" field above (uses the "list" field of AttrValue).
+       * If type == "type" or "list(type)" above, then the "type" field
+       * of "allowed_values.list" has the set of allowed DataTypes.
+       * If type == "string" or "list(string)", then the "s" field of
+       * "allowed_values.list" has the set of allowed strings.
+       * </pre>
+       *
+       * <code>.tensorflow.AttrValue allowed_values = 7;</code>
+       */
+      public org.tensorflow.proto.framework.AttrValueOrBuilder getAllowedValuesOrBuilder() {
+        if (allowedValuesBuilder_ != null) {
+          return allowedValuesBuilder_.getMessageOrBuilder();
+        } else {
+          return allowedValues_ == null ?
+              org.tensorflow.proto.framework.AttrValue.getDefaultInstance() : allowedValues_;
+        }
+      }
+      /**
+       * <pre>
+       * The set of allowed values.  Has type that is the "list" version
+       * of the "type" field above (uses the "list" field of AttrValue).
+       * If type == "type" or "list(type)" above, then the "type" field
+       * of "allowed_values.list" has the set of allowed DataTypes.
+       * If type == "string" or "list(string)", then the "s" field of
+       * "allowed_values.list" has the set of allowed strings.
+       * </pre>
+       *
+       * <code>.tensorflow.AttrValue allowed_values = 7;</code>
+       */
+      private com.google.protobuf.SingleFieldBuilderV3<
+          org.tensorflow.proto.framework.AttrValue, org.tensorflow.proto.framework.AttrValue.Builder, org.tensorflow.proto.framework.AttrValueOrBuilder> 
+          getAllowedValuesFieldBuilder() {
+        if (allowedValuesBuilder_ == null) {
+          allowedValuesBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+              org.tensorflow.proto.framework.AttrValue, org.tensorflow.proto.framework.AttrValue.Builder, org.tensorflow.proto.framework.AttrValueOrBuilder>(
+                  getAllowedValues(),
+                  getParentForChildren(),
+                  isClean());
+          allowedValues_ = null;
+        }
+        return allowedValuesBuilder_;
+      }
+      @java.lang.Override
+      public final Builder setUnknownFields(
+          final com.google.protobuf.UnknownFieldSet unknownFields) {
+        return super.setUnknownFields(unknownFields);
+      }
+
+      @java.lang.Override
+      public final Builder mergeUnknownFields(
+          final com.google.protobuf.UnknownFieldSet unknownFields) {
+        return super.mergeUnknownFields(unknownFields);
+      }
+
+
+      // @@protoc_insertion_point(builder_scope:tensorflow.OpDef.AttrDef)
+    }
+
+    // @@protoc_insertion_point(class_scope:tensorflow.OpDef.AttrDef)
+    private static final org.tensorflow.proto.framework.OpDef.AttrDef DEFAULT_INSTANCE;
+    static {
+      DEFAULT_INSTANCE = new org.tensorflow.proto.framework.OpDef.AttrDef();
+    }
+
+    public static org.tensorflow.proto.framework.OpDef.AttrDef getDefaultInstance() {
+      return DEFAULT_INSTANCE;
+    }
+
+    private static final com.google.protobuf.Parser<AttrDef>
+        PARSER = new com.google.protobuf.AbstractParser<AttrDef>() {
+      @java.lang.Override
+      public AttrDef parsePartialFrom(
+          com.google.protobuf.CodedInputStream input,
+          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+          throws com.google.protobuf.InvalidProtocolBufferException {
+        return new AttrDef(input, extensionRegistry);
+      }
+    };
+
+    public static com.google.protobuf.Parser<AttrDef> parser() {
+      return PARSER;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Parser<AttrDef> getParserForType() {
+      return PARSER;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.OpDef.AttrDef getDefaultInstanceForType() {
+      return DEFAULT_INSTANCE;
+    }
+
+  }
+
+  public static final int NAME_FIELD_NUMBER = 1;
+  private volatile java.lang.Object name_;
+  /**
+   * <pre>
+   * Op names starting with an underscore are reserved for internal use.
+   * Names should be CamelCase and match the regexp "[A-Z][a-zA-Z0-9&gt;_]*".
+   * </pre>
+   *
+   * <code>string name = 1;</code>
+   */
+  public java.lang.String getName() {
+    java.lang.Object ref = name_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      name_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * Op names starting with an underscore are reserved for internal use.
+   * Names should be CamelCase and match the regexp "[A-Z][a-zA-Z0-9&gt;_]*".
+   * </pre>
+   *
+   * <code>string name = 1;</code>
+   */
+  public com.google.protobuf.ByteString
+      getNameBytes() {
+    java.lang.Object ref = name_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      name_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int INPUT_ARG_FIELD_NUMBER = 2;
+  private java.util.List<org.tensorflow.proto.framework.OpDef.ArgDef> inputArg_;
+  /**
+   * <pre>
+   * Description of the input(s).
+   * </pre>
+   *
+   * <code>repeated .tensorflow.OpDef.ArgDef input_arg = 2;</code>
+   */
+  public java.util.List<org.tensorflow.proto.framework.OpDef.ArgDef> getInputArgList() {
+    return inputArg_;
+  }
+  /**
+   * <pre>
+   * Description of the input(s).
+   * </pre>
+   *
+   * <code>repeated .tensorflow.OpDef.ArgDef input_arg = 2;</code>
+   */
+  public java.util.List<? extends org.tensorflow.proto.framework.OpDef.ArgDefOrBuilder> 
+      getInputArgOrBuilderList() {
+    return inputArg_;
+  }
+  /**
+   * <pre>
+   * Description of the input(s).
+   * </pre>
+   *
+   * <code>repeated .tensorflow.OpDef.ArgDef input_arg = 2;</code>
+   */
+  public int getInputArgCount() {
+    return inputArg_.size();
+  }
+  /**
+   * <pre>
+   * Description of the input(s).
+   * </pre>
+   *
+   * <code>repeated .tensorflow.OpDef.ArgDef input_arg = 2;</code>
+   */
+  public org.tensorflow.proto.framework.OpDef.ArgDef getInputArg(int index) {
+    return inputArg_.get(index);
+  }
+  /**
+   * <pre>
+   * Description of the input(s).
+   * </pre>
+   *
+   * <code>repeated .tensorflow.OpDef.ArgDef input_arg = 2;</code>
+   */
+  public org.tensorflow.proto.framework.OpDef.ArgDefOrBuilder getInputArgOrBuilder(
+      int index) {
+    return inputArg_.get(index);
+  }
+
+  public static final int OUTPUT_ARG_FIELD_NUMBER = 3;
+  private java.util.List<org.tensorflow.proto.framework.OpDef.ArgDef> outputArg_;
+  /**
+   * <pre>
+   * Description of the output(s).
+   * </pre>
+   *
+   * <code>repeated .tensorflow.OpDef.ArgDef output_arg = 3;</code>
+   */
+  public java.util.List<org.tensorflow.proto.framework.OpDef.ArgDef> getOutputArgList() {
+    return outputArg_;
+  }
+  /**
+   * <pre>
+   * Description of the output(s).
+   * </pre>
+   *
+   * <code>repeated .tensorflow.OpDef.ArgDef output_arg = 3;</code>
+   */
+  public java.util.List<? extends org.tensorflow.proto.framework.OpDef.ArgDefOrBuilder> 
+      getOutputArgOrBuilderList() {
+    return outputArg_;
+  }
+  /**
+   * <pre>
+   * Description of the output(s).
+   * </pre>
+   *
+   * <code>repeated .tensorflow.OpDef.ArgDef output_arg = 3;</code>
+   */
+  public int getOutputArgCount() {
+    return outputArg_.size();
+  }
+  /**
+   * <pre>
+   * Description of the output(s).
+   * </pre>
+   *
+   * <code>repeated .tensorflow.OpDef.ArgDef output_arg = 3;</code>
+   */
+  public org.tensorflow.proto.framework.OpDef.ArgDef getOutputArg(int index) {
+    return outputArg_.get(index);
+  }
+  /**
+   * <pre>
+   * Description of the output(s).
+   * </pre>
+   *
+   * <code>repeated .tensorflow.OpDef.ArgDef output_arg = 3;</code>
+   */
+  public org.tensorflow.proto.framework.OpDef.ArgDefOrBuilder getOutputArgOrBuilder(
+      int index) {
+    return outputArg_.get(index);
+  }
+
+  public static final int CONTROL_OUTPUT_FIELD_NUMBER = 20;
+  private com.google.protobuf.LazyStringList controlOutput_;
+  /**
+   * <pre>
+   * Named control outputs for this operation. Useful only for composite
+   * operations (i.e. functions) which want to name different control outputs.
+   * </pre>
+   *
+   * <code>repeated string control_output = 20;</code>
+   */
+  public com.google.protobuf.ProtocolStringList
+      getControlOutputList() {
+    return controlOutput_;
+  }
+  /**
+   * <pre>
+   * Named control outputs for this operation. Useful only for composite
+   * operations (i.e. functions) which want to name different control outputs.
+   * </pre>
+   *
+   * <code>repeated string control_output = 20;</code>
+   */
+  public int getControlOutputCount() {
+    return controlOutput_.size();
+  }
+  /**
+   * <pre>
+   * Named control outputs for this operation. Useful only for composite
+   * operations (i.e. functions) which want to name different control outputs.
+   * </pre>
+   *
+   * <code>repeated string control_output = 20;</code>
+   */
+  public java.lang.String getControlOutput(int index) {
+    return controlOutput_.get(index);
+  }
+  /**
+   * <pre>
+   * Named control outputs for this operation. Useful only for composite
+   * operations (i.e. functions) which want to name different control outputs.
+   * </pre>
+   *
+   * <code>repeated string control_output = 20;</code>
+   */
+  public com.google.protobuf.ByteString
+      getControlOutputBytes(int index) {
+    return controlOutput_.getByteString(index);
+  }
+
+  public static final int ATTR_FIELD_NUMBER = 4;
+  private java.util.List<org.tensorflow.proto.framework.OpDef.AttrDef> attr_;
+  /**
+   * <code>repeated .tensorflow.OpDef.AttrDef attr = 4;</code>
+   */
+  public java.util.List<org.tensorflow.proto.framework.OpDef.AttrDef> getAttrList() {
+    return attr_;
+  }
+  /**
+   * <code>repeated .tensorflow.OpDef.AttrDef attr = 4;</code>
+   */
+  public java.util.List<? extends org.tensorflow.proto.framework.OpDef.AttrDefOrBuilder> 
+      getAttrOrBuilderList() {
+    return attr_;
+  }
+  /**
+   * <code>repeated .tensorflow.OpDef.AttrDef attr = 4;</code>
+   */
+  public int getAttrCount() {
+    return attr_.size();
+  }
+  /**
+   * <code>repeated .tensorflow.OpDef.AttrDef attr = 4;</code>
+   */
+  public org.tensorflow.proto.framework.OpDef.AttrDef getAttr(int index) {
+    return attr_.get(index);
+  }
+  /**
+   * <code>repeated .tensorflow.OpDef.AttrDef attr = 4;</code>
+   */
+  public org.tensorflow.proto.framework.OpDef.AttrDefOrBuilder getAttrOrBuilder(
+      int index) {
+    return attr_.get(index);
+  }
+
+  public static final int DEPRECATION_FIELD_NUMBER = 8;
+  private org.tensorflow.proto.framework.OpDeprecation deprecation_;
+  /**
+   * <pre>
+   * Optional deprecation based on GraphDef versions.
+   * </pre>
+   *
+   * <code>.tensorflow.OpDeprecation deprecation = 8;</code>
+   */
+  public boolean hasDeprecation() {
+    return deprecation_ != null;
+  }
+  /**
+   * <pre>
+   * Optional deprecation based on GraphDef versions.
+   * </pre>
+   *
+   * <code>.tensorflow.OpDeprecation deprecation = 8;</code>
+   */
+  public org.tensorflow.proto.framework.OpDeprecation getDeprecation() {
+    return deprecation_ == null ? org.tensorflow.proto.framework.OpDeprecation.getDefaultInstance() : deprecation_;
+  }
+  /**
+   * <pre>
+   * Optional deprecation based on GraphDef versions.
+   * </pre>
+   *
+   * <code>.tensorflow.OpDeprecation deprecation = 8;</code>
+   */
+  public org.tensorflow.proto.framework.OpDeprecationOrBuilder getDeprecationOrBuilder() {
+    return getDeprecation();
+  }
+
+  public static final int SUMMARY_FIELD_NUMBER = 5;
+  private volatile java.lang.Object summary_;
+  /**
+   * <pre>
+   * One-line human-readable description of what the Op does.
+   * </pre>
+   *
+   * <code>string summary = 5;</code>
+   */
+  public java.lang.String getSummary() {
+    java.lang.Object ref = summary_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      summary_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * One-line human-readable description of what the Op does.
+   * </pre>
+   *
+   * <code>string summary = 5;</code>
+   */
+  public com.google.protobuf.ByteString
+      getSummaryBytes() {
+    java.lang.Object ref = summary_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      summary_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int DESCRIPTION_FIELD_NUMBER = 6;
+  private volatile java.lang.Object description_;
+  /**
+   * <pre>
+   * Additional, longer human-readable description of what the Op does.
+   * </pre>
+   *
+   * <code>string description = 6;</code>
+   */
+  public java.lang.String getDescription() {
+    java.lang.Object ref = description_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      description_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * Additional, longer human-readable description of what the Op does.
+   * </pre>
+   *
+   * <code>string description = 6;</code>
+   */
+  public com.google.protobuf.ByteString
+      getDescriptionBytes() {
+    java.lang.Object ref = description_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      description_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int IS_COMMUTATIVE_FIELD_NUMBER = 18;
+  private boolean isCommutative_;
+  /**
+   * <pre>
+   * True if the operation is commutative ("op(a,b) == op(b,a)" for all inputs)
+   * </pre>
+   *
+   * <code>bool is_commutative = 18;</code>
+   */
+  public boolean getIsCommutative() {
+    return isCommutative_;
+  }
+
+  public static final int IS_AGGREGATE_FIELD_NUMBER = 16;
+  private boolean isAggregate_;
+  /**
+   * <pre>
+   * If is_aggregate is true, then this operation accepts N &gt;= 2
+   * inputs and produces 1 output all of the same type.  Should be
+   * associative and commutative, and produce output with the same
+   * shape as the input.  The optimizer may replace an aggregate op
+   * taking input from multiple devices with a tree of aggregate ops
+   * that aggregate locally within each device (and possibly within
+   * groups of nearby devices) before communicating.
+   * TODO(josh11b): Implement that optimization.
+   * </pre>
+   *
+   * <code>bool is_aggregate = 16;</code>
+   */
+  public boolean getIsAggregate() {
+    return isAggregate_;
+  }
+
+  public static final int IS_STATEFUL_FIELD_NUMBER = 17;
+  private boolean isStateful_;
+  /**
+   * <pre>
+   * Ops are marked as stateful if their behavior depends on some state beyond
+   * their input tensors (e.g. variable reading op) or if they have
+   * a side-effect (e.g. printing or asserting ops). Equivalently, stateless ops
+   * must always produce the same output for the same input and have
+   * no side-effects.
+   * By default Ops may be moved between devices.  Stateful ops should
+   * either not be moved, or should only be moved if that state can also
+   * be moved (e.g. via some sort of save / restore).
+   * Stateful ops are guaranteed to never be optimized away by Common
+   * Subexpression Elimination (CSE).
+   * </pre>
+   *
+   * <code>bool is_stateful = 17;</code>
+   */
+  public boolean getIsStateful() {
+    return isStateful_;
+  }
+
+  public static final int ALLOWS_UNINITIALIZED_INPUT_FIELD_NUMBER = 19;
+  private boolean allowsUninitializedInput_;
+  /**
+   * <pre>
+   * By default, all inputs to an Op must be initialized Tensors.  Ops
+   * that may initialize tensors for the first time should set this
+   * field to true, to allow the Op to take an uninitialized Tensor as
+   * input.
+   * </pre>
+   *
+   * <code>bool allows_uninitialized_input = 19;</code>
+   */
+  public boolean getAllowsUninitializedInput() {
+    return allowsUninitializedInput_;
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    if (!getNameBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 1, name_);
+    }
+    for (int i = 0; i < inputArg_.size(); i++) {
+      output.writeMessage(2, inputArg_.get(i));
+    }
+    for (int i = 0; i < outputArg_.size(); i++) {
+      output.writeMessage(3, outputArg_.get(i));
+    }
+    for (int i = 0; i < attr_.size(); i++) {
+      output.writeMessage(4, attr_.get(i));
+    }
+    if (!getSummaryBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 5, summary_);
+    }
+    if (!getDescriptionBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 6, description_);
+    }
+    if (deprecation_ != null) {
+      output.writeMessage(8, getDeprecation());
+    }
+    if (isAggregate_ != false) {
+      output.writeBool(16, isAggregate_);
+    }
+    if (isStateful_ != false) {
+      output.writeBool(17, isStateful_);
+    }
+    if (isCommutative_ != false) {
+      output.writeBool(18, isCommutative_);
+    }
+    if (allowsUninitializedInput_ != false) {
+      output.writeBool(19, allowsUninitializedInput_);
+    }
+    for (int i = 0; i < controlOutput_.size(); i++) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 20, controlOutput_.getRaw(i));
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (!getNameBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, name_);
+    }
+    for (int i = 0; i < inputArg_.size(); i++) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(2, inputArg_.get(i));
+    }
+    for (int i = 0; i < outputArg_.size(); i++) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(3, outputArg_.get(i));
+    }
+    for (int i = 0; i < attr_.size(); i++) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(4, attr_.get(i));
+    }
+    if (!getSummaryBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(5, summary_);
+    }
+    if (!getDescriptionBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(6, description_);
+    }
+    if (deprecation_ != null) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(8, getDeprecation());
+    }
+    if (isAggregate_ != false) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeBoolSize(16, isAggregate_);
+    }
+    if (isStateful_ != false) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeBoolSize(17, isStateful_);
+    }
+    if (isCommutative_ != false) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeBoolSize(18, isCommutative_);
+    }
+    if (allowsUninitializedInput_ != false) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeBoolSize(19, allowsUninitializedInput_);
+    }
+    {
+      int dataSize = 0;
+      for (int i = 0; i < controlOutput_.size(); i++) {
+        dataSize += computeStringSizeNoTag(controlOutput_.getRaw(i));
+      }
+      size += dataSize;
+      size += 2 * getControlOutputList().size();
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.framework.OpDef)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.framework.OpDef other = (org.tensorflow.proto.framework.OpDef) obj;
+
+    if (!getName()
+        .equals(other.getName())) return false;
+    if (!getInputArgList()
+        .equals(other.getInputArgList())) return false;
+    if (!getOutputArgList()
+        .equals(other.getOutputArgList())) return false;
+    if (!getControlOutputList()
+        .equals(other.getControlOutputList())) return false;
+    if (!getAttrList()
+        .equals(other.getAttrList())) return false;
+    if (hasDeprecation() != other.hasDeprecation()) return false;
+    if (hasDeprecation()) {
+      if (!getDeprecation()
+          .equals(other.getDeprecation())) return false;
+    }
+    if (!getSummary()
+        .equals(other.getSummary())) return false;
+    if (!getDescription()
+        .equals(other.getDescription())) return false;
+    if (getIsCommutative()
+        != other.getIsCommutative()) return false;
+    if (getIsAggregate()
+        != other.getIsAggregate()) return false;
+    if (getIsStateful()
+        != other.getIsStateful()) return false;
+    if (getAllowsUninitializedInput()
+        != other.getAllowsUninitializedInput()) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    hash = (37 * hash) + NAME_FIELD_NUMBER;
+    hash = (53 * hash) + getName().hashCode();
+    if (getInputArgCount() > 0) {
+      hash = (37 * hash) + INPUT_ARG_FIELD_NUMBER;
+      hash = (53 * hash) + getInputArgList().hashCode();
+    }
+    if (getOutputArgCount() > 0) {
+      hash = (37 * hash) + OUTPUT_ARG_FIELD_NUMBER;
+      hash = (53 * hash) + getOutputArgList().hashCode();
+    }
+    if (getControlOutputCount() > 0) {
+      hash = (37 * hash) + CONTROL_OUTPUT_FIELD_NUMBER;
+      hash = (53 * hash) + getControlOutputList().hashCode();
+    }
+    if (getAttrCount() > 0) {
+      hash = (37 * hash) + ATTR_FIELD_NUMBER;
+      hash = (53 * hash) + getAttrList().hashCode();
+    }
+    if (hasDeprecation()) {
+      hash = (37 * hash) + DEPRECATION_FIELD_NUMBER;
+      hash = (53 * hash) + getDeprecation().hashCode();
+    }
+    hash = (37 * hash) + SUMMARY_FIELD_NUMBER;
+    hash = (53 * hash) + getSummary().hashCode();
+    hash = (37 * hash) + DESCRIPTION_FIELD_NUMBER;
+    hash = (53 * hash) + getDescription().hashCode();
+    hash = (37 * hash) + IS_COMMUTATIVE_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashBoolean(
+        getIsCommutative());
+    hash = (37 * hash) + IS_AGGREGATE_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashBoolean(
+        getIsAggregate());
+    hash = (37 * hash) + IS_STATEFUL_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashBoolean(
+        getIsStateful());
+    hash = (37 * hash) + ALLOWS_UNINITIALIZED_INPUT_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashBoolean(
+        getAllowsUninitializedInput());
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.framework.OpDef parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.OpDef parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.OpDef parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.OpDef parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.OpDef parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.OpDef parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.OpDef parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.OpDef parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.OpDef parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.OpDef parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.OpDef parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.OpDef parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.framework.OpDef prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * <pre>
+   * Defines an operation. A NodeDef in a GraphDef specifies an Op by
+   * using the "op" field which should match the name of a OpDef.
+   * LINT.IfChange
+   * </pre>
+   *
+   * Protobuf type {@code tensorflow.OpDef}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.OpDef)
+      org.tensorflow.proto.framework.OpDefOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.OpDefProtos.internal_static_tensorflow_OpDef_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.OpDefProtos.internal_static_tensorflow_OpDef_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.OpDef.class, org.tensorflow.proto.framework.OpDef.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.framework.OpDef.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+        getInputArgFieldBuilder();
+        getOutputArgFieldBuilder();
+        getAttrFieldBuilder();
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      name_ = "";
+
+      if (inputArgBuilder_ == null) {
+        inputArg_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000001);
+      } else {
+        inputArgBuilder_.clear();
+      }
+      if (outputArgBuilder_ == null) {
+        outputArg_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000002);
+      } else {
+        outputArgBuilder_.clear();
+      }
+      controlOutput_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+      bitField0_ = (bitField0_ & ~0x00000004);
+      if (attrBuilder_ == null) {
+        attr_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000008);
+      } else {
+        attrBuilder_.clear();
+      }
+      if (deprecationBuilder_ == null) {
+        deprecation_ = null;
+      } else {
+        deprecation_ = null;
+        deprecationBuilder_ = null;
+      }
+      summary_ = "";
+
+      description_ = "";
+
+      isCommutative_ = false;
+
+      isAggregate_ = false;
+
+      isStateful_ = false;
+
+      allowsUninitializedInput_ = false;
+
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.framework.OpDefProtos.internal_static_tensorflow_OpDef_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.OpDef getDefaultInstanceForType() {
+      return org.tensorflow.proto.framework.OpDef.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.OpDef build() {
+      org.tensorflow.proto.framework.OpDef result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.OpDef buildPartial() {
+      org.tensorflow.proto.framework.OpDef result = new org.tensorflow.proto.framework.OpDef(this);
+      int from_bitField0_ = bitField0_;
+      result.name_ = name_;
+      if (inputArgBuilder_ == null) {
+        if (((bitField0_ & 0x00000001) != 0)) {
+          inputArg_ = java.util.Collections.unmodifiableList(inputArg_);
+          bitField0_ = (bitField0_ & ~0x00000001);
+        }
+        result.inputArg_ = inputArg_;
+      } else {
+        result.inputArg_ = inputArgBuilder_.build();
+      }
+      if (outputArgBuilder_ == null) {
+        if (((bitField0_ & 0x00000002) != 0)) {
+          outputArg_ = java.util.Collections.unmodifiableList(outputArg_);
+          bitField0_ = (bitField0_ & ~0x00000002);
+        }
+        result.outputArg_ = outputArg_;
+      } else {
+        result.outputArg_ = outputArgBuilder_.build();
+      }
+      if (((bitField0_ & 0x00000004) != 0)) {
+        controlOutput_ = controlOutput_.getUnmodifiableView();
+        bitField0_ = (bitField0_ & ~0x00000004);
+      }
+      result.controlOutput_ = controlOutput_;
+      if (attrBuilder_ == null) {
+        if (((bitField0_ & 0x00000008) != 0)) {
+          attr_ = java.util.Collections.unmodifiableList(attr_);
+          bitField0_ = (bitField0_ & ~0x00000008);
+        }
+        result.attr_ = attr_;
+      } else {
+        result.attr_ = attrBuilder_.build();
+      }
+      if (deprecationBuilder_ == null) {
+        result.deprecation_ = deprecation_;
+      } else {
+        result.deprecation_ = deprecationBuilder_.build();
+      }
+      result.summary_ = summary_;
+      result.description_ = description_;
+      result.isCommutative_ = isCommutative_;
+      result.isAggregate_ = isAggregate_;
+      result.isStateful_ = isStateful_;
+      result.allowsUninitializedInput_ = allowsUninitializedInput_;
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.framework.OpDef) {
+        return mergeFrom((org.tensorflow.proto.framework.OpDef)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.framework.OpDef other) {
+      if (other == org.tensorflow.proto.framework.OpDef.getDefaultInstance()) return this;
+      if (!other.getName().isEmpty()) {
+        name_ = other.name_;
+        onChanged();
+      }
+      if (inputArgBuilder_ == null) {
+        if (!other.inputArg_.isEmpty()) {
+          if (inputArg_.isEmpty()) {
+            inputArg_ = other.inputArg_;
+            bitField0_ = (bitField0_ & ~0x00000001);
+          } else {
+            ensureInputArgIsMutable();
+            inputArg_.addAll(other.inputArg_);
+          }
+          onChanged();
+        }
+      } else {
+        if (!other.inputArg_.isEmpty()) {
+          if (inputArgBuilder_.isEmpty()) {
+            inputArgBuilder_.dispose();
+            inputArgBuilder_ = null;
+            inputArg_ = other.inputArg_;
+            bitField0_ = (bitField0_ & ~0x00000001);
+            inputArgBuilder_ = 
+              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
+                 getInputArgFieldBuilder() : null;
+          } else {
+            inputArgBuilder_.addAllMessages(other.inputArg_);
+          }
+        }
+      }
+      if (outputArgBuilder_ == null) {
+        if (!other.outputArg_.isEmpty()) {
+          if (outputArg_.isEmpty()) {
+            outputArg_ = other.outputArg_;
+            bitField0_ = (bitField0_ & ~0x00000002);
+          } else {
+            ensureOutputArgIsMutable();
+            outputArg_.addAll(other.outputArg_);
+          }
+          onChanged();
+        }
+      } else {
+        if (!other.outputArg_.isEmpty()) {
+          if (outputArgBuilder_.isEmpty()) {
+            outputArgBuilder_.dispose();
+            outputArgBuilder_ = null;
+            outputArg_ = other.outputArg_;
+            bitField0_ = (bitField0_ & ~0x00000002);
+            outputArgBuilder_ = 
+              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
+                 getOutputArgFieldBuilder() : null;
+          } else {
+            outputArgBuilder_.addAllMessages(other.outputArg_);
+          }
+        }
+      }
+      if (!other.controlOutput_.isEmpty()) {
+        if (controlOutput_.isEmpty()) {
+          controlOutput_ = other.controlOutput_;
+          bitField0_ = (bitField0_ & ~0x00000004);
+        } else {
+          ensureControlOutputIsMutable();
+          controlOutput_.addAll(other.controlOutput_);
+        }
+        onChanged();
+      }
+      if (attrBuilder_ == null) {
+        if (!other.attr_.isEmpty()) {
+          if (attr_.isEmpty()) {
+            attr_ = other.attr_;
+            bitField0_ = (bitField0_ & ~0x00000008);
+          } else {
+            ensureAttrIsMutable();
+            attr_.addAll(other.attr_);
+          }
+          onChanged();
+        }
+      } else {
+        if (!other.attr_.isEmpty()) {
+          if (attrBuilder_.isEmpty()) {
+            attrBuilder_.dispose();
+            attrBuilder_ = null;
+            attr_ = other.attr_;
+            bitField0_ = (bitField0_ & ~0x00000008);
+            attrBuilder_ = 
+              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
+                 getAttrFieldBuilder() : null;
+          } else {
+            attrBuilder_.addAllMessages(other.attr_);
+          }
+        }
+      }
+      if (other.hasDeprecation()) {
+        mergeDeprecation(other.getDeprecation());
+      }
+      if (!other.getSummary().isEmpty()) {
+        summary_ = other.summary_;
+        onChanged();
+      }
+      if (!other.getDescription().isEmpty()) {
+        description_ = other.description_;
+        onChanged();
+      }
+      if (other.getIsCommutative() != false) {
+        setIsCommutative(other.getIsCommutative());
+      }
+      if (other.getIsAggregate() != false) {
+        setIsAggregate(other.getIsAggregate());
+      }
+      if (other.getIsStateful() != false) {
+        setIsStateful(other.getIsStateful());
+      }
+      if (other.getAllowsUninitializedInput() != false) {
+        setAllowsUninitializedInput(other.getAllowsUninitializedInput());
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.framework.OpDef parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.framework.OpDef) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+    private int bitField0_;
+
+    private java.lang.Object name_ = "";
+    /**
+     * <pre>
+     * Op names starting with an underscore are reserved for internal use.
+     * Names should be CamelCase and match the regexp "[A-Z][a-zA-Z0-9&gt;_]*".
+     * </pre>
+     *
+     * <code>string name = 1;</code>
+     */
+    public java.lang.String getName() {
+      java.lang.Object ref = name_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        name_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Op names starting with an underscore are reserved for internal use.
+     * Names should be CamelCase and match the regexp "[A-Z][a-zA-Z0-9&gt;_]*".
+     * </pre>
+     *
+     * <code>string name = 1;</code>
+     */
+    public com.google.protobuf.ByteString
+        getNameBytes() {
+      java.lang.Object ref = name_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        name_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Op names starting with an underscore are reserved for internal use.
+     * Names should be CamelCase and match the regexp "[A-Z][a-zA-Z0-9&gt;_]*".
+     * </pre>
+     *
+     * <code>string name = 1;</code>
+     */
+    public Builder setName(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      name_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Op names starting with an underscore are reserved for internal use.
+     * Names should be CamelCase and match the regexp "[A-Z][a-zA-Z0-9&gt;_]*".
+     * </pre>
+     *
+     * <code>string name = 1;</code>
+     */
+    public Builder clearName() {
+      
+      name_ = getDefaultInstance().getName();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Op names starting with an underscore are reserved for internal use.
+     * Names should be CamelCase and match the regexp "[A-Z][a-zA-Z0-9&gt;_]*".
+     * </pre>
+     *
+     * <code>string name = 1;</code>
+     */
+    public Builder setNameBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      name_ = value;
+      onChanged();
+      return this;
+    }
+
+    private java.util.List<org.tensorflow.proto.framework.OpDef.ArgDef> inputArg_ =
+      java.util.Collections.emptyList();
+    private void ensureInputArgIsMutable() {
+      if (!((bitField0_ & 0x00000001) != 0)) {
+        inputArg_ = new java.util.ArrayList<org.tensorflow.proto.framework.OpDef.ArgDef>(inputArg_);
+        bitField0_ |= 0x00000001;
+       }
+    }
+
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.framework.OpDef.ArgDef, org.tensorflow.proto.framework.OpDef.ArgDef.Builder, org.tensorflow.proto.framework.OpDef.ArgDefOrBuilder> inputArgBuilder_;
+
+    /**
+     * <pre>
+     * Description of the input(s).
+     * </pre>
+     *
+     * <code>repeated .tensorflow.OpDef.ArgDef input_arg = 2;</code>
+     */
+    public java.util.List<org.tensorflow.proto.framework.OpDef.ArgDef> getInputArgList() {
+      if (inputArgBuilder_ == null) {
+        return java.util.Collections.unmodifiableList(inputArg_);
+      } else {
+        return inputArgBuilder_.getMessageList();
+      }
+    }
+    /**
+     * <pre>
+     * Description of the input(s).
+     * </pre>
+     *
+     * <code>repeated .tensorflow.OpDef.ArgDef input_arg = 2;</code>
+     */
+    public int getInputArgCount() {
+      if (inputArgBuilder_ == null) {
+        return inputArg_.size();
+      } else {
+        return inputArgBuilder_.getCount();
+      }
+    }
+    /**
+     * <pre>
+     * Description of the input(s).
+     * </pre>
+     *
+     * <code>repeated .tensorflow.OpDef.ArgDef input_arg = 2;</code>
+     */
+    public org.tensorflow.proto.framework.OpDef.ArgDef getInputArg(int index) {
+      if (inputArgBuilder_ == null) {
+        return inputArg_.get(index);
+      } else {
+        return inputArgBuilder_.getMessage(index);
+      }
+    }
+    /**
+     * <pre>
+     * Description of the input(s).
+     * </pre>
+     *
+     * <code>repeated .tensorflow.OpDef.ArgDef input_arg = 2;</code>
+     */
+    public Builder setInputArg(
+        int index, org.tensorflow.proto.framework.OpDef.ArgDef value) {
+      if (inputArgBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureInputArgIsMutable();
+        inputArg_.set(index, value);
+        onChanged();
+      } else {
+        inputArgBuilder_.setMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Description of the input(s).
+     * </pre>
+     *
+     * <code>repeated .tensorflow.OpDef.ArgDef input_arg = 2;</code>
+     */
+    public Builder setInputArg(
+        int index, org.tensorflow.proto.framework.OpDef.ArgDef.Builder builderForValue) {
+      if (inputArgBuilder_ == null) {
+        ensureInputArgIsMutable();
+        inputArg_.set(index, builderForValue.build());
+        onChanged();
+      } else {
+        inputArgBuilder_.setMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Description of the input(s).
+     * </pre>
+     *
+     * <code>repeated .tensorflow.OpDef.ArgDef input_arg = 2;</code>
+     */
+    public Builder addInputArg(org.tensorflow.proto.framework.OpDef.ArgDef value) {
+      if (inputArgBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureInputArgIsMutable();
+        inputArg_.add(value);
+        onChanged();
+      } else {
+        inputArgBuilder_.addMessage(value);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Description of the input(s).
+     * </pre>
+     *
+     * <code>repeated .tensorflow.OpDef.ArgDef input_arg = 2;</code>
+     */
+    public Builder addInputArg(
+        int index, org.tensorflow.proto.framework.OpDef.ArgDef value) {
+      if (inputArgBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureInputArgIsMutable();
+        inputArg_.add(index, value);
+        onChanged();
+      } else {
+        inputArgBuilder_.addMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Description of the input(s).
+     * </pre>
+     *
+     * <code>repeated .tensorflow.OpDef.ArgDef input_arg = 2;</code>
+     */
+    public Builder addInputArg(
+        org.tensorflow.proto.framework.OpDef.ArgDef.Builder builderForValue) {
+      if (inputArgBuilder_ == null) {
+        ensureInputArgIsMutable();
+        inputArg_.add(builderForValue.build());
+        onChanged();
+      } else {
+        inputArgBuilder_.addMessage(builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Description of the input(s).
+     * </pre>
+     *
+     * <code>repeated .tensorflow.OpDef.ArgDef input_arg = 2;</code>
+     */
+    public Builder addInputArg(
+        int index, org.tensorflow.proto.framework.OpDef.ArgDef.Builder builderForValue) {
+      if (inputArgBuilder_ == null) {
+        ensureInputArgIsMutable();
+        inputArg_.add(index, builderForValue.build());
+        onChanged();
+      } else {
+        inputArgBuilder_.addMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Description of the input(s).
+     * </pre>
+     *
+     * <code>repeated .tensorflow.OpDef.ArgDef input_arg = 2;</code>
+     */
+    public Builder addAllInputArg(
+        java.lang.Iterable<? extends org.tensorflow.proto.framework.OpDef.ArgDef> values) {
+      if (inputArgBuilder_ == null) {
+        ensureInputArgIsMutable();
+        com.google.protobuf.AbstractMessageLite.Builder.addAll(
+            values, inputArg_);
+        onChanged();
+      } else {
+        inputArgBuilder_.addAllMessages(values);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Description of the input(s).
+     * </pre>
+     *
+     * <code>repeated .tensorflow.OpDef.ArgDef input_arg = 2;</code>
+     */
+    public Builder clearInputArg() {
+      if (inputArgBuilder_ == null) {
+        inputArg_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000001);
+        onChanged();
+      } else {
+        inputArgBuilder_.clear();
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Description of the input(s).
+     * </pre>
+     *
+     * <code>repeated .tensorflow.OpDef.ArgDef input_arg = 2;</code>
+     */
+    public Builder removeInputArg(int index) {
+      if (inputArgBuilder_ == null) {
+        ensureInputArgIsMutable();
+        inputArg_.remove(index);
+        onChanged();
+      } else {
+        inputArgBuilder_.remove(index);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Description of the input(s).
+     * </pre>
+     *
+     * <code>repeated .tensorflow.OpDef.ArgDef input_arg = 2;</code>
+     */
+    public org.tensorflow.proto.framework.OpDef.ArgDef.Builder getInputArgBuilder(
+        int index) {
+      return getInputArgFieldBuilder().getBuilder(index);
+    }
+    /**
+     * <pre>
+     * Description of the input(s).
+     * </pre>
+     *
+     * <code>repeated .tensorflow.OpDef.ArgDef input_arg = 2;</code>
+     */
+    public org.tensorflow.proto.framework.OpDef.ArgDefOrBuilder getInputArgOrBuilder(
+        int index) {
+      if (inputArgBuilder_ == null) {
+        return inputArg_.get(index);  } else {
+        return inputArgBuilder_.getMessageOrBuilder(index);
+      }
+    }
+    /**
+     * <pre>
+     * Description of the input(s).
+     * </pre>
+     *
+     * <code>repeated .tensorflow.OpDef.ArgDef input_arg = 2;</code>
+     */
+    public java.util.List<? extends org.tensorflow.proto.framework.OpDef.ArgDefOrBuilder> 
+         getInputArgOrBuilderList() {
+      if (inputArgBuilder_ != null) {
+        return inputArgBuilder_.getMessageOrBuilderList();
+      } else {
+        return java.util.Collections.unmodifiableList(inputArg_);
+      }
+    }
+    /**
+     * <pre>
+     * Description of the input(s).
+     * </pre>
+     *
+     * <code>repeated .tensorflow.OpDef.ArgDef input_arg = 2;</code>
+     */
+    public org.tensorflow.proto.framework.OpDef.ArgDef.Builder addInputArgBuilder() {
+      return getInputArgFieldBuilder().addBuilder(
+          org.tensorflow.proto.framework.OpDef.ArgDef.getDefaultInstance());
+    }
+    /**
+     * <pre>
+     * Description of the input(s).
+     * </pre>
+     *
+     * <code>repeated .tensorflow.OpDef.ArgDef input_arg = 2;</code>
+     */
+    public org.tensorflow.proto.framework.OpDef.ArgDef.Builder addInputArgBuilder(
+        int index) {
+      return getInputArgFieldBuilder().addBuilder(
+          index, org.tensorflow.proto.framework.OpDef.ArgDef.getDefaultInstance());
+    }
+    /**
+     * <pre>
+     * Description of the input(s).
+     * </pre>
+     *
+     * <code>repeated .tensorflow.OpDef.ArgDef input_arg = 2;</code>
+     */
+    public java.util.List<org.tensorflow.proto.framework.OpDef.ArgDef.Builder> 
+         getInputArgBuilderList() {
+      return getInputArgFieldBuilder().getBuilderList();
+    }
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.framework.OpDef.ArgDef, org.tensorflow.proto.framework.OpDef.ArgDef.Builder, org.tensorflow.proto.framework.OpDef.ArgDefOrBuilder> 
+        getInputArgFieldBuilder() {
+      if (inputArgBuilder_ == null) {
+        inputArgBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
+            org.tensorflow.proto.framework.OpDef.ArgDef, org.tensorflow.proto.framework.OpDef.ArgDef.Builder, org.tensorflow.proto.framework.OpDef.ArgDefOrBuilder>(
+                inputArg_,
+                ((bitField0_ & 0x00000001) != 0),
+                getParentForChildren(),
+                isClean());
+        inputArg_ = null;
+      }
+      return inputArgBuilder_;
+    }
+
+    private java.util.List<org.tensorflow.proto.framework.OpDef.ArgDef> outputArg_ =
+      java.util.Collections.emptyList();
+    private void ensureOutputArgIsMutable() {
+      if (!((bitField0_ & 0x00000002) != 0)) {
+        outputArg_ = new java.util.ArrayList<org.tensorflow.proto.framework.OpDef.ArgDef>(outputArg_);
+        bitField0_ |= 0x00000002;
+       }
+    }
+
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.framework.OpDef.ArgDef, org.tensorflow.proto.framework.OpDef.ArgDef.Builder, org.tensorflow.proto.framework.OpDef.ArgDefOrBuilder> outputArgBuilder_;
+
+    /**
+     * <pre>
+     * Description of the output(s).
+     * </pre>
+     *
+     * <code>repeated .tensorflow.OpDef.ArgDef output_arg = 3;</code>
+     */
+    public java.util.List<org.tensorflow.proto.framework.OpDef.ArgDef> getOutputArgList() {
+      if (outputArgBuilder_ == null) {
+        return java.util.Collections.unmodifiableList(outputArg_);
+      } else {
+        return outputArgBuilder_.getMessageList();
+      }
+    }
+    /**
+     * <pre>
+     * Description of the output(s).
+     * </pre>
+     *
+     * <code>repeated .tensorflow.OpDef.ArgDef output_arg = 3;</code>
+     */
+    public int getOutputArgCount() {
+      if (outputArgBuilder_ == null) {
+        return outputArg_.size();
+      } else {
+        return outputArgBuilder_.getCount();
+      }
+    }
+    /**
+     * <pre>
+     * Description of the output(s).
+     * </pre>
+     *
+     * <code>repeated .tensorflow.OpDef.ArgDef output_arg = 3;</code>
+     */
+    public org.tensorflow.proto.framework.OpDef.ArgDef getOutputArg(int index) {
+      if (outputArgBuilder_ == null) {
+        return outputArg_.get(index);
+      } else {
+        return outputArgBuilder_.getMessage(index);
+      }
+    }
+    /**
+     * <pre>
+     * Description of the output(s).
+     * </pre>
+     *
+     * <code>repeated .tensorflow.OpDef.ArgDef output_arg = 3;</code>
+     */
+    public Builder setOutputArg(
+        int index, org.tensorflow.proto.framework.OpDef.ArgDef value) {
+      if (outputArgBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureOutputArgIsMutable();
+        outputArg_.set(index, value);
+        onChanged();
+      } else {
+        outputArgBuilder_.setMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Description of the output(s).
+     * </pre>
+     *
+     * <code>repeated .tensorflow.OpDef.ArgDef output_arg = 3;</code>
+     */
+    public Builder setOutputArg(
+        int index, org.tensorflow.proto.framework.OpDef.ArgDef.Builder builderForValue) {
+      if (outputArgBuilder_ == null) {
+        ensureOutputArgIsMutable();
+        outputArg_.set(index, builderForValue.build());
+        onChanged();
+      } else {
+        outputArgBuilder_.setMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Description of the output(s).
+     * </pre>
+     *
+     * <code>repeated .tensorflow.OpDef.ArgDef output_arg = 3;</code>
+     */
+    public Builder addOutputArg(org.tensorflow.proto.framework.OpDef.ArgDef value) {
+      if (outputArgBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureOutputArgIsMutable();
+        outputArg_.add(value);
+        onChanged();
+      } else {
+        outputArgBuilder_.addMessage(value);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Description of the output(s).
+     * </pre>
+     *
+     * <code>repeated .tensorflow.OpDef.ArgDef output_arg = 3;</code>
+     */
+    public Builder addOutputArg(
+        int index, org.tensorflow.proto.framework.OpDef.ArgDef value) {
+      if (outputArgBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureOutputArgIsMutable();
+        outputArg_.add(index, value);
+        onChanged();
+      } else {
+        outputArgBuilder_.addMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Description of the output(s).
+     * </pre>
+     *
+     * <code>repeated .tensorflow.OpDef.ArgDef output_arg = 3;</code>
+     */
+    public Builder addOutputArg(
+        org.tensorflow.proto.framework.OpDef.ArgDef.Builder builderForValue) {
+      if (outputArgBuilder_ == null) {
+        ensureOutputArgIsMutable();
+        outputArg_.add(builderForValue.build());
+        onChanged();
+      } else {
+        outputArgBuilder_.addMessage(builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Description of the output(s).
+     * </pre>
+     *
+     * <code>repeated .tensorflow.OpDef.ArgDef output_arg = 3;</code>
+     */
+    public Builder addOutputArg(
+        int index, org.tensorflow.proto.framework.OpDef.ArgDef.Builder builderForValue) {
+      if (outputArgBuilder_ == null) {
+        ensureOutputArgIsMutable();
+        outputArg_.add(index, builderForValue.build());
+        onChanged();
+      } else {
+        outputArgBuilder_.addMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Description of the output(s).
+     * </pre>
+     *
+     * <code>repeated .tensorflow.OpDef.ArgDef output_arg = 3;</code>
+     */
+    public Builder addAllOutputArg(
+        java.lang.Iterable<? extends org.tensorflow.proto.framework.OpDef.ArgDef> values) {
+      if (outputArgBuilder_ == null) {
+        ensureOutputArgIsMutable();
+        com.google.protobuf.AbstractMessageLite.Builder.addAll(
+            values, outputArg_);
+        onChanged();
+      } else {
+        outputArgBuilder_.addAllMessages(values);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Description of the output(s).
+     * </pre>
+     *
+     * <code>repeated .tensorflow.OpDef.ArgDef output_arg = 3;</code>
+     */
+    public Builder clearOutputArg() {
+      if (outputArgBuilder_ == null) {
+        outputArg_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000002);
+        onChanged();
+      } else {
+        outputArgBuilder_.clear();
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Description of the output(s).
+     * </pre>
+     *
+     * <code>repeated .tensorflow.OpDef.ArgDef output_arg = 3;</code>
+     */
+    public Builder removeOutputArg(int index) {
+      if (outputArgBuilder_ == null) {
+        ensureOutputArgIsMutable();
+        outputArg_.remove(index);
+        onChanged();
+      } else {
+        outputArgBuilder_.remove(index);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Description of the output(s).
+     * </pre>
+     *
+     * <code>repeated .tensorflow.OpDef.ArgDef output_arg = 3;</code>
+     */
+    public org.tensorflow.proto.framework.OpDef.ArgDef.Builder getOutputArgBuilder(
+        int index) {
+      return getOutputArgFieldBuilder().getBuilder(index);
+    }
+    /**
+     * <pre>
+     * Description of the output(s).
+     * </pre>
+     *
+     * <code>repeated .tensorflow.OpDef.ArgDef output_arg = 3;</code>
+     */
+    public org.tensorflow.proto.framework.OpDef.ArgDefOrBuilder getOutputArgOrBuilder(
+        int index) {
+      if (outputArgBuilder_ == null) {
+        return outputArg_.get(index);  } else {
+        return outputArgBuilder_.getMessageOrBuilder(index);
+      }
+    }
+    /**
+     * <pre>
+     * Description of the output(s).
+     * </pre>
+     *
+     * <code>repeated .tensorflow.OpDef.ArgDef output_arg = 3;</code>
+     */
+    public java.util.List<? extends org.tensorflow.proto.framework.OpDef.ArgDefOrBuilder> 
+         getOutputArgOrBuilderList() {
+      if (outputArgBuilder_ != null) {
+        return outputArgBuilder_.getMessageOrBuilderList();
+      } else {
+        return java.util.Collections.unmodifiableList(outputArg_);
+      }
+    }
+    /**
+     * <pre>
+     * Description of the output(s).
+     * </pre>
+     *
+     * <code>repeated .tensorflow.OpDef.ArgDef output_arg = 3;</code>
+     */
+    public org.tensorflow.proto.framework.OpDef.ArgDef.Builder addOutputArgBuilder() {
+      return getOutputArgFieldBuilder().addBuilder(
+          org.tensorflow.proto.framework.OpDef.ArgDef.getDefaultInstance());
+    }
+    /**
+     * <pre>
+     * Description of the output(s).
+     * </pre>
+     *
+     * <code>repeated .tensorflow.OpDef.ArgDef output_arg = 3;</code>
+     */
+    public org.tensorflow.proto.framework.OpDef.ArgDef.Builder addOutputArgBuilder(
+        int index) {
+      return getOutputArgFieldBuilder().addBuilder(
+          index, org.tensorflow.proto.framework.OpDef.ArgDef.getDefaultInstance());
+    }
+    /**
+     * <pre>
+     * Description of the output(s).
+     * </pre>
+     *
+     * <code>repeated .tensorflow.OpDef.ArgDef output_arg = 3;</code>
+     */
+    public java.util.List<org.tensorflow.proto.framework.OpDef.ArgDef.Builder> 
+         getOutputArgBuilderList() {
+      return getOutputArgFieldBuilder().getBuilderList();
+    }
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.framework.OpDef.ArgDef, org.tensorflow.proto.framework.OpDef.ArgDef.Builder, org.tensorflow.proto.framework.OpDef.ArgDefOrBuilder> 
+        getOutputArgFieldBuilder() {
+      if (outputArgBuilder_ == null) {
+        outputArgBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
+            org.tensorflow.proto.framework.OpDef.ArgDef, org.tensorflow.proto.framework.OpDef.ArgDef.Builder, org.tensorflow.proto.framework.OpDef.ArgDefOrBuilder>(
+                outputArg_,
+                ((bitField0_ & 0x00000002) != 0),
+                getParentForChildren(),
+                isClean());
+        outputArg_ = null;
+      }
+      return outputArgBuilder_;
+    }
+
+    private com.google.protobuf.LazyStringList controlOutput_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+    private void ensureControlOutputIsMutable() {
+      if (!((bitField0_ & 0x00000004) != 0)) {
+        controlOutput_ = new com.google.protobuf.LazyStringArrayList(controlOutput_);
+        bitField0_ |= 0x00000004;
+       }
+    }
+    /**
+     * <pre>
+     * Named control outputs for this operation. Useful only for composite
+     * operations (i.e. functions) which want to name different control outputs.
+     * </pre>
+     *
+     * <code>repeated string control_output = 20;</code>
+     */
+    public com.google.protobuf.ProtocolStringList
+        getControlOutputList() {
+      return controlOutput_.getUnmodifiableView();
+    }
+    /**
+     * <pre>
+     * Named control outputs for this operation. Useful only for composite
+     * operations (i.e. functions) which want to name different control outputs.
+     * </pre>
+     *
+     * <code>repeated string control_output = 20;</code>
+     */
+    public int getControlOutputCount() {
+      return controlOutput_.size();
+    }
+    /**
+     * <pre>
+     * Named control outputs for this operation. Useful only for composite
+     * operations (i.e. functions) which want to name different control outputs.
+     * </pre>
+     *
+     * <code>repeated string control_output = 20;</code>
+     */
+    public java.lang.String getControlOutput(int index) {
+      return controlOutput_.get(index);
+    }
+    /**
+     * <pre>
+     * Named control outputs for this operation. Useful only for composite
+     * operations (i.e. functions) which want to name different control outputs.
+     * </pre>
+     *
+     * <code>repeated string control_output = 20;</code>
+     */
+    public com.google.protobuf.ByteString
+        getControlOutputBytes(int index) {
+      return controlOutput_.getByteString(index);
+    }
+    /**
+     * <pre>
+     * Named control outputs for this operation. Useful only for composite
+     * operations (i.e. functions) which want to name different control outputs.
+     * </pre>
+     *
+     * <code>repeated string control_output = 20;</code>
+     */
+    public Builder setControlOutput(
+        int index, java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  ensureControlOutputIsMutable();
+      controlOutput_.set(index, value);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Named control outputs for this operation. Useful only for composite
+     * operations (i.e. functions) which want to name different control outputs.
+     * </pre>
+     *
+     * <code>repeated string control_output = 20;</code>
+     */
+    public Builder addControlOutput(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  ensureControlOutputIsMutable();
+      controlOutput_.add(value);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Named control outputs for this operation. Useful only for composite
+     * operations (i.e. functions) which want to name different control outputs.
+     * </pre>
+     *
+     * <code>repeated string control_output = 20;</code>
+     */
+    public Builder addAllControlOutput(
+        java.lang.Iterable<java.lang.String> values) {
+      ensureControlOutputIsMutable();
+      com.google.protobuf.AbstractMessageLite.Builder.addAll(
+          values, controlOutput_);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Named control outputs for this operation. Useful only for composite
+     * operations (i.e. functions) which want to name different control outputs.
+     * </pre>
+     *
+     * <code>repeated string control_output = 20;</code>
+     */
+    public Builder clearControlOutput() {
+      controlOutput_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+      bitField0_ = (bitField0_ & ~0x00000004);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Named control outputs for this operation. Useful only for composite
+     * operations (i.e. functions) which want to name different control outputs.
+     * </pre>
+     *
+     * <code>repeated string control_output = 20;</code>
+     */
+    public Builder addControlOutputBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      ensureControlOutputIsMutable();
+      controlOutput_.add(value);
+      onChanged();
+      return this;
+    }
+
+    private java.util.List<org.tensorflow.proto.framework.OpDef.AttrDef> attr_ =
+      java.util.Collections.emptyList();
+    private void ensureAttrIsMutable() {
+      if (!((bitField0_ & 0x00000008) != 0)) {
+        attr_ = new java.util.ArrayList<org.tensorflow.proto.framework.OpDef.AttrDef>(attr_);
+        bitField0_ |= 0x00000008;
+       }
+    }
+
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.framework.OpDef.AttrDef, org.tensorflow.proto.framework.OpDef.AttrDef.Builder, org.tensorflow.proto.framework.OpDef.AttrDefOrBuilder> attrBuilder_;
+
+    /**
+     * <code>repeated .tensorflow.OpDef.AttrDef attr = 4;</code>
+     */
+    public java.util.List<org.tensorflow.proto.framework.OpDef.AttrDef> getAttrList() {
+      if (attrBuilder_ == null) {
+        return java.util.Collections.unmodifiableList(attr_);
+      } else {
+        return attrBuilder_.getMessageList();
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.OpDef.AttrDef attr = 4;</code>
+     */
+    public int getAttrCount() {
+      if (attrBuilder_ == null) {
+        return attr_.size();
+      } else {
+        return attrBuilder_.getCount();
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.OpDef.AttrDef attr = 4;</code>
+     */
+    public org.tensorflow.proto.framework.OpDef.AttrDef getAttr(int index) {
+      if (attrBuilder_ == null) {
+        return attr_.get(index);
+      } else {
+        return attrBuilder_.getMessage(index);
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.OpDef.AttrDef attr = 4;</code>
+     */
+    public Builder setAttr(
+        int index, org.tensorflow.proto.framework.OpDef.AttrDef value) {
+      if (attrBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureAttrIsMutable();
+        attr_.set(index, value);
+        onChanged();
+      } else {
+        attrBuilder_.setMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.OpDef.AttrDef attr = 4;</code>
+     */
+    public Builder setAttr(
+        int index, org.tensorflow.proto.framework.OpDef.AttrDef.Builder builderForValue) {
+      if (attrBuilder_ == null) {
+        ensureAttrIsMutable();
+        attr_.set(index, builderForValue.build());
+        onChanged();
+      } else {
+        attrBuilder_.setMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.OpDef.AttrDef attr = 4;</code>
+     */
+    public Builder addAttr(org.tensorflow.proto.framework.OpDef.AttrDef value) {
+      if (attrBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureAttrIsMutable();
+        attr_.add(value);
+        onChanged();
+      } else {
+        attrBuilder_.addMessage(value);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.OpDef.AttrDef attr = 4;</code>
+     */
+    public Builder addAttr(
+        int index, org.tensorflow.proto.framework.OpDef.AttrDef value) {
+      if (attrBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureAttrIsMutable();
+        attr_.add(index, value);
+        onChanged();
+      } else {
+        attrBuilder_.addMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.OpDef.AttrDef attr = 4;</code>
+     */
+    public Builder addAttr(
+        org.tensorflow.proto.framework.OpDef.AttrDef.Builder builderForValue) {
+      if (attrBuilder_ == null) {
+        ensureAttrIsMutable();
+        attr_.add(builderForValue.build());
+        onChanged();
+      } else {
+        attrBuilder_.addMessage(builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.OpDef.AttrDef attr = 4;</code>
+     */
+    public Builder addAttr(
+        int index, org.tensorflow.proto.framework.OpDef.AttrDef.Builder builderForValue) {
+      if (attrBuilder_ == null) {
+        ensureAttrIsMutable();
+        attr_.add(index, builderForValue.build());
+        onChanged();
+      } else {
+        attrBuilder_.addMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.OpDef.AttrDef attr = 4;</code>
+     */
+    public Builder addAllAttr(
+        java.lang.Iterable<? extends org.tensorflow.proto.framework.OpDef.AttrDef> values) {
+      if (attrBuilder_ == null) {
+        ensureAttrIsMutable();
+        com.google.protobuf.AbstractMessageLite.Builder.addAll(
+            values, attr_);
+        onChanged();
+      } else {
+        attrBuilder_.addAllMessages(values);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.OpDef.AttrDef attr = 4;</code>
+     */
+    public Builder clearAttr() {
+      if (attrBuilder_ == null) {
+        attr_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000008);
+        onChanged();
+      } else {
+        attrBuilder_.clear();
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.OpDef.AttrDef attr = 4;</code>
+     */
+    public Builder removeAttr(int index) {
+      if (attrBuilder_ == null) {
+        ensureAttrIsMutable();
+        attr_.remove(index);
+        onChanged();
+      } else {
+        attrBuilder_.remove(index);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.OpDef.AttrDef attr = 4;</code>
+     */
+    public org.tensorflow.proto.framework.OpDef.AttrDef.Builder getAttrBuilder(
+        int index) {
+      return getAttrFieldBuilder().getBuilder(index);
+    }
+    /**
+     * <code>repeated .tensorflow.OpDef.AttrDef attr = 4;</code>
+     */
+    public org.tensorflow.proto.framework.OpDef.AttrDefOrBuilder getAttrOrBuilder(
+        int index) {
+      if (attrBuilder_ == null) {
+        return attr_.get(index);  } else {
+        return attrBuilder_.getMessageOrBuilder(index);
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.OpDef.AttrDef attr = 4;</code>
+     */
+    public java.util.List<? extends org.tensorflow.proto.framework.OpDef.AttrDefOrBuilder> 
+         getAttrOrBuilderList() {
+      if (attrBuilder_ != null) {
+        return attrBuilder_.getMessageOrBuilderList();
+      } else {
+        return java.util.Collections.unmodifiableList(attr_);
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.OpDef.AttrDef attr = 4;</code>
+     */
+    public org.tensorflow.proto.framework.OpDef.AttrDef.Builder addAttrBuilder() {
+      return getAttrFieldBuilder().addBuilder(
+          org.tensorflow.proto.framework.OpDef.AttrDef.getDefaultInstance());
+    }
+    /**
+     * <code>repeated .tensorflow.OpDef.AttrDef attr = 4;</code>
+     */
+    public org.tensorflow.proto.framework.OpDef.AttrDef.Builder addAttrBuilder(
+        int index) {
+      return getAttrFieldBuilder().addBuilder(
+          index, org.tensorflow.proto.framework.OpDef.AttrDef.getDefaultInstance());
+    }
+    /**
+     * <code>repeated .tensorflow.OpDef.AttrDef attr = 4;</code>
+     */
+    public java.util.List<org.tensorflow.proto.framework.OpDef.AttrDef.Builder> 
+         getAttrBuilderList() {
+      return getAttrFieldBuilder().getBuilderList();
+    }
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.framework.OpDef.AttrDef, org.tensorflow.proto.framework.OpDef.AttrDef.Builder, org.tensorflow.proto.framework.OpDef.AttrDefOrBuilder> 
+        getAttrFieldBuilder() {
+      if (attrBuilder_ == null) {
+        attrBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
+            org.tensorflow.proto.framework.OpDef.AttrDef, org.tensorflow.proto.framework.OpDef.AttrDef.Builder, org.tensorflow.proto.framework.OpDef.AttrDefOrBuilder>(
+                attr_,
+                ((bitField0_ & 0x00000008) != 0),
+                getParentForChildren(),
+                isClean());
+        attr_ = null;
+      }
+      return attrBuilder_;
+    }
+
+    private org.tensorflow.proto.framework.OpDeprecation deprecation_;
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.OpDeprecation, org.tensorflow.proto.framework.OpDeprecation.Builder, org.tensorflow.proto.framework.OpDeprecationOrBuilder> deprecationBuilder_;
+    /**
+     * <pre>
+     * Optional deprecation based on GraphDef versions.
+     * </pre>
+     *
+     * <code>.tensorflow.OpDeprecation deprecation = 8;</code>
+     */
+    public boolean hasDeprecation() {
+      return deprecationBuilder_ != null || deprecation_ != null;
+    }
+    /**
+     * <pre>
+     * Optional deprecation based on GraphDef versions.
+     * </pre>
+     *
+     * <code>.tensorflow.OpDeprecation deprecation = 8;</code>
+     */
+    public org.tensorflow.proto.framework.OpDeprecation getDeprecation() {
+      if (deprecationBuilder_ == null) {
+        return deprecation_ == null ? org.tensorflow.proto.framework.OpDeprecation.getDefaultInstance() : deprecation_;
+      } else {
+        return deprecationBuilder_.getMessage();
+      }
+    }
+    /**
+     * <pre>
+     * Optional deprecation based on GraphDef versions.
+     * </pre>
+     *
+     * <code>.tensorflow.OpDeprecation deprecation = 8;</code>
+     */
+    public Builder setDeprecation(org.tensorflow.proto.framework.OpDeprecation value) {
+      if (deprecationBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        deprecation_ = value;
+        onChanged();
+      } else {
+        deprecationBuilder_.setMessage(value);
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Optional deprecation based on GraphDef versions.
+     * </pre>
+     *
+     * <code>.tensorflow.OpDeprecation deprecation = 8;</code>
+     */
+    public Builder setDeprecation(
+        org.tensorflow.proto.framework.OpDeprecation.Builder builderForValue) {
+      if (deprecationBuilder_ == null) {
+        deprecation_ = builderForValue.build();
+        onChanged();
+      } else {
+        deprecationBuilder_.setMessage(builderForValue.build());
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Optional deprecation based on GraphDef versions.
+     * </pre>
+     *
+     * <code>.tensorflow.OpDeprecation deprecation = 8;</code>
+     */
+    public Builder mergeDeprecation(org.tensorflow.proto.framework.OpDeprecation value) {
+      if (deprecationBuilder_ == null) {
+        if (deprecation_ != null) {
+          deprecation_ =
+            org.tensorflow.proto.framework.OpDeprecation.newBuilder(deprecation_).mergeFrom(value).buildPartial();
+        } else {
+          deprecation_ = value;
+        }
+        onChanged();
+      } else {
+        deprecationBuilder_.mergeFrom(value);
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Optional deprecation based on GraphDef versions.
+     * </pre>
+     *
+     * <code>.tensorflow.OpDeprecation deprecation = 8;</code>
+     */
+    public Builder clearDeprecation() {
+      if (deprecationBuilder_ == null) {
+        deprecation_ = null;
+        onChanged();
+      } else {
+        deprecation_ = null;
+        deprecationBuilder_ = null;
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Optional deprecation based on GraphDef versions.
+     * </pre>
+     *
+     * <code>.tensorflow.OpDeprecation deprecation = 8;</code>
+     */
+    public org.tensorflow.proto.framework.OpDeprecation.Builder getDeprecationBuilder() {
+      
+      onChanged();
+      return getDeprecationFieldBuilder().getBuilder();
+    }
+    /**
+     * <pre>
+     * Optional deprecation based on GraphDef versions.
+     * </pre>
+     *
+     * <code>.tensorflow.OpDeprecation deprecation = 8;</code>
+     */
+    public org.tensorflow.proto.framework.OpDeprecationOrBuilder getDeprecationOrBuilder() {
+      if (deprecationBuilder_ != null) {
+        return deprecationBuilder_.getMessageOrBuilder();
+      } else {
+        return deprecation_ == null ?
+            org.tensorflow.proto.framework.OpDeprecation.getDefaultInstance() : deprecation_;
+      }
+    }
+    /**
+     * <pre>
+     * Optional deprecation based on GraphDef versions.
+     * </pre>
+     *
+     * <code>.tensorflow.OpDeprecation deprecation = 8;</code>
+     */
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.OpDeprecation, org.tensorflow.proto.framework.OpDeprecation.Builder, org.tensorflow.proto.framework.OpDeprecationOrBuilder> 
+        getDeprecationFieldBuilder() {
+      if (deprecationBuilder_ == null) {
+        deprecationBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+            org.tensorflow.proto.framework.OpDeprecation, org.tensorflow.proto.framework.OpDeprecation.Builder, org.tensorflow.proto.framework.OpDeprecationOrBuilder>(
+                getDeprecation(),
+                getParentForChildren(),
+                isClean());
+        deprecation_ = null;
+      }
+      return deprecationBuilder_;
+    }
+
+    private java.lang.Object summary_ = "";
+    /**
+     * <pre>
+     * One-line human-readable description of what the Op does.
+     * </pre>
+     *
+     * <code>string summary = 5;</code>
+     */
+    public java.lang.String getSummary() {
+      java.lang.Object ref = summary_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        summary_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * One-line human-readable description of what the Op does.
+     * </pre>
+     *
+     * <code>string summary = 5;</code>
+     */
+    public com.google.protobuf.ByteString
+        getSummaryBytes() {
+      java.lang.Object ref = summary_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        summary_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * One-line human-readable description of what the Op does.
+     * </pre>
+     *
+     * <code>string summary = 5;</code>
+     */
+    public Builder setSummary(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      summary_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * One-line human-readable description of what the Op does.
+     * </pre>
+     *
+     * <code>string summary = 5;</code>
+     */
+    public Builder clearSummary() {
+      
+      summary_ = getDefaultInstance().getSummary();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * One-line human-readable description of what the Op does.
+     * </pre>
+     *
+     * <code>string summary = 5;</code>
+     */
+    public Builder setSummaryBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      summary_ = value;
+      onChanged();
+      return this;
+    }
+
+    private java.lang.Object description_ = "";
+    /**
+     * <pre>
+     * Additional, longer human-readable description of what the Op does.
+     * </pre>
+     *
+     * <code>string description = 6;</code>
+     */
+    public java.lang.String getDescription() {
+      java.lang.Object ref = description_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        description_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Additional, longer human-readable description of what the Op does.
+     * </pre>
+     *
+     * <code>string description = 6;</code>
+     */
+    public com.google.protobuf.ByteString
+        getDescriptionBytes() {
+      java.lang.Object ref = description_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        description_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Additional, longer human-readable description of what the Op does.
+     * </pre>
+     *
+     * <code>string description = 6;</code>
+     */
+    public Builder setDescription(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      description_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Additional, longer human-readable description of what the Op does.
+     * </pre>
+     *
+     * <code>string description = 6;</code>
+     */
+    public Builder clearDescription() {
+      
+      description_ = getDefaultInstance().getDescription();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Additional, longer human-readable description of what the Op does.
+     * </pre>
+     *
+     * <code>string description = 6;</code>
+     */
+    public Builder setDescriptionBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      description_ = value;
+      onChanged();
+      return this;
+    }
+
+    private boolean isCommutative_ ;
+    /**
+     * <pre>
+     * True if the operation is commutative ("op(a,b) == op(b,a)" for all inputs)
+     * </pre>
+     *
+     * <code>bool is_commutative = 18;</code>
+     */
+    public boolean getIsCommutative() {
+      return isCommutative_;
+    }
+    /**
+     * <pre>
+     * True if the operation is commutative ("op(a,b) == op(b,a)" for all inputs)
+     * </pre>
+     *
+     * <code>bool is_commutative = 18;</code>
+     */
+    public Builder setIsCommutative(boolean value) {
+      
+      isCommutative_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * True if the operation is commutative ("op(a,b) == op(b,a)" for all inputs)
+     * </pre>
+     *
+     * <code>bool is_commutative = 18;</code>
+     */
+    public Builder clearIsCommutative() {
+      
+      isCommutative_ = false;
+      onChanged();
+      return this;
+    }
+
+    private boolean isAggregate_ ;
+    /**
+     * <pre>
+     * If is_aggregate is true, then this operation accepts N &gt;= 2
+     * inputs and produces 1 output all of the same type.  Should be
+     * associative and commutative, and produce output with the same
+     * shape as the input.  The optimizer may replace an aggregate op
+     * taking input from multiple devices with a tree of aggregate ops
+     * that aggregate locally within each device (and possibly within
+     * groups of nearby devices) before communicating.
+     * TODO(josh11b): Implement that optimization.
+     * </pre>
+     *
+     * <code>bool is_aggregate = 16;</code>
+     */
+    public boolean getIsAggregate() {
+      return isAggregate_;
+    }
+    /**
+     * <pre>
+     * If is_aggregate is true, then this operation accepts N &gt;= 2
+     * inputs and produces 1 output all of the same type.  Should be
+     * associative and commutative, and produce output with the same
+     * shape as the input.  The optimizer may replace an aggregate op
+     * taking input from multiple devices with a tree of aggregate ops
+     * that aggregate locally within each device (and possibly within
+     * groups of nearby devices) before communicating.
+     * TODO(josh11b): Implement that optimization.
+     * </pre>
+     *
+     * <code>bool is_aggregate = 16;</code>
+     */
+    public Builder setIsAggregate(boolean value) {
+      
+      isAggregate_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * If is_aggregate is true, then this operation accepts N &gt;= 2
+     * inputs and produces 1 output all of the same type.  Should be
+     * associative and commutative, and produce output with the same
+     * shape as the input.  The optimizer may replace an aggregate op
+     * taking input from multiple devices with a tree of aggregate ops
+     * that aggregate locally within each device (and possibly within
+     * groups of nearby devices) before communicating.
+     * TODO(josh11b): Implement that optimization.
+     * </pre>
+     *
+     * <code>bool is_aggregate = 16;</code>
+     */
+    public Builder clearIsAggregate() {
+      
+      isAggregate_ = false;
+      onChanged();
+      return this;
+    }
+
+    private boolean isStateful_ ;
+    /**
+     * <pre>
+     * Ops are marked as stateful if their behavior depends on some state beyond
+     * their input tensors (e.g. variable reading op) or if they have
+     * a side-effect (e.g. printing or asserting ops). Equivalently, stateless ops
+     * must always produce the same output for the same input and have
+     * no side-effects.
+     * By default Ops may be moved between devices.  Stateful ops should
+     * either not be moved, or should only be moved if that state can also
+     * be moved (e.g. via some sort of save / restore).
+     * Stateful ops are guaranteed to never be optimized away by Common
+     * Subexpression Elimination (CSE).
+     * </pre>
+     *
+     * <code>bool is_stateful = 17;</code>
+     */
+    public boolean getIsStateful() {
+      return isStateful_;
+    }
+    /**
+     * <pre>
+     * Ops are marked as stateful if their behavior depends on some state beyond
+     * their input tensors (e.g. variable reading op) or if they have
+     * a side-effect (e.g. printing or asserting ops). Equivalently, stateless ops
+     * must always produce the same output for the same input and have
+     * no side-effects.
+     * By default Ops may be moved between devices.  Stateful ops should
+     * either not be moved, or should only be moved if that state can also
+     * be moved (e.g. via some sort of save / restore).
+     * Stateful ops are guaranteed to never be optimized away by Common
+     * Subexpression Elimination (CSE).
+     * </pre>
+     *
+     * <code>bool is_stateful = 17;</code>
+     */
+    public Builder setIsStateful(boolean value) {
+      
+      isStateful_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Ops are marked as stateful if their behavior depends on some state beyond
+     * their input tensors (e.g. variable reading op) or if they have
+     * a side-effect (e.g. printing or asserting ops). Equivalently, stateless ops
+     * must always produce the same output for the same input and have
+     * no side-effects.
+     * By default Ops may be moved between devices.  Stateful ops should
+     * either not be moved, or should only be moved if that state can also
+     * be moved (e.g. via some sort of save / restore).
+     * Stateful ops are guaranteed to never be optimized away by Common
+     * Subexpression Elimination (CSE).
+     * </pre>
+     *
+     * <code>bool is_stateful = 17;</code>
+     */
+    public Builder clearIsStateful() {
+      
+      isStateful_ = false;
+      onChanged();
+      return this;
+    }
+
+    private boolean allowsUninitializedInput_ ;
+    /**
+     * <pre>
+     * By default, all inputs to an Op must be initialized Tensors.  Ops
+     * that may initialize tensors for the first time should set this
+     * field to true, to allow the Op to take an uninitialized Tensor as
+     * input.
+     * </pre>
+     *
+     * <code>bool allows_uninitialized_input = 19;</code>
+     */
+    public boolean getAllowsUninitializedInput() {
+      return allowsUninitializedInput_;
+    }
+    /**
+     * <pre>
+     * By default, all inputs to an Op must be initialized Tensors.  Ops
+     * that may initialize tensors for the first time should set this
+     * field to true, to allow the Op to take an uninitialized Tensor as
+     * input.
+     * </pre>
+     *
+     * <code>bool allows_uninitialized_input = 19;</code>
+     */
+    public Builder setAllowsUninitializedInput(boolean value) {
+      
+      allowsUninitializedInput_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * By default, all inputs to an Op must be initialized Tensors.  Ops
+     * that may initialize tensors for the first time should set this
+     * field to true, to allow the Op to take an uninitialized Tensor as
+     * input.
+     * </pre>
+     *
+     * <code>bool allows_uninitialized_input = 19;</code>
+     */
+    public Builder clearAllowsUninitializedInput() {
+      
+      allowsUninitializedInput_ = false;
+      onChanged();
+      return this;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.OpDef)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.OpDef)
+  private static final org.tensorflow.proto.framework.OpDef DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.OpDef();
+  }
+
+  public static org.tensorflow.proto.framework.OpDef getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<OpDef>
+      PARSER = new com.google.protobuf.AbstractParser<OpDef>() {
+    @java.lang.Override
+    public OpDef parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new OpDef(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<OpDef> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<OpDef> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.framework.OpDef getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/OpDefOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/OpDefOrBuilder.java
new file mode 100644
index 00000000000..6ac64c0ed97
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/OpDefOrBuilder.java
@@ -0,0 +1,296 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/framework/op_def.proto
+
+package org.tensorflow.proto.framework;
+
+public interface OpDefOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.OpDef)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <pre>
+   * Op names starting with an underscore are reserved for internal use.
+   * Names should be CamelCase and match the regexp "[A-Z][a-zA-Z0-9&gt;_]*".
+   * </pre>
+   *
+   * <code>string name = 1;</code>
+   */
+  java.lang.String getName();
+  /**
+   * <pre>
+   * Op names starting with an underscore are reserved for internal use.
+   * Names should be CamelCase and match the regexp "[A-Z][a-zA-Z0-9&gt;_]*".
+   * </pre>
+   *
+   * <code>string name = 1;</code>
+   */
+  com.google.protobuf.ByteString
+      getNameBytes();
+
+  /**
+   * <pre>
+   * Description of the input(s).
+   * </pre>
+   *
+   * <code>repeated .tensorflow.OpDef.ArgDef input_arg = 2;</code>
+   */
+  java.util.List<org.tensorflow.proto.framework.OpDef.ArgDef> 
+      getInputArgList();
+  /**
+   * <pre>
+   * Description of the input(s).
+   * </pre>
+   *
+   * <code>repeated .tensorflow.OpDef.ArgDef input_arg = 2;</code>
+   */
+  org.tensorflow.proto.framework.OpDef.ArgDef getInputArg(int index);
+  /**
+   * <pre>
+   * Description of the input(s).
+   * </pre>
+   *
+   * <code>repeated .tensorflow.OpDef.ArgDef input_arg = 2;</code>
+   */
+  int getInputArgCount();
+  /**
+   * <pre>
+   * Description of the input(s).
+   * </pre>
+   *
+   * <code>repeated .tensorflow.OpDef.ArgDef input_arg = 2;</code>
+   */
+  java.util.List<? extends org.tensorflow.proto.framework.OpDef.ArgDefOrBuilder> 
+      getInputArgOrBuilderList();
+  /**
+   * <pre>
+   * Description of the input(s).
+   * </pre>
+   *
+   * <code>repeated .tensorflow.OpDef.ArgDef input_arg = 2;</code>
+   */
+  org.tensorflow.proto.framework.OpDef.ArgDefOrBuilder getInputArgOrBuilder(
+      int index);
+
+  /**
+   * <pre>
+   * Description of the output(s).
+   * </pre>
+   *
+   * <code>repeated .tensorflow.OpDef.ArgDef output_arg = 3;</code>
+   */
+  java.util.List<org.tensorflow.proto.framework.OpDef.ArgDef> 
+      getOutputArgList();
+  /**
+   * <pre>
+   * Description of the output(s).
+   * </pre>
+   *
+   * <code>repeated .tensorflow.OpDef.ArgDef output_arg = 3;</code>
+   */
+  org.tensorflow.proto.framework.OpDef.ArgDef getOutputArg(int index);
+  /**
+   * <pre>
+   * Description of the output(s).
+   * </pre>
+   *
+   * <code>repeated .tensorflow.OpDef.ArgDef output_arg = 3;</code>
+   */
+  int getOutputArgCount();
+  /**
+   * <pre>
+   * Description of the output(s).
+   * </pre>
+   *
+   * <code>repeated .tensorflow.OpDef.ArgDef output_arg = 3;</code>
+   */
+  java.util.List<? extends org.tensorflow.proto.framework.OpDef.ArgDefOrBuilder> 
+      getOutputArgOrBuilderList();
+  /**
+   * <pre>
+   * Description of the output(s).
+   * </pre>
+   *
+   * <code>repeated .tensorflow.OpDef.ArgDef output_arg = 3;</code>
+   */
+  org.tensorflow.proto.framework.OpDef.ArgDefOrBuilder getOutputArgOrBuilder(
+      int index);
+
+  /**
+   * <pre>
+   * Named control outputs for this operation. Useful only for composite
+   * operations (i.e. functions) which want to name different control outputs.
+   * </pre>
+   *
+   * <code>repeated string control_output = 20;</code>
+   */
+  java.util.List<java.lang.String>
+      getControlOutputList();
+  /**
+   * <pre>
+   * Named control outputs for this operation. Useful only for composite
+   * operations (i.e. functions) which want to name different control outputs.
+   * </pre>
+   *
+   * <code>repeated string control_output = 20;</code>
+   */
+  int getControlOutputCount();
+  /**
+   * <pre>
+   * Named control outputs for this operation. Useful only for composite
+   * operations (i.e. functions) which want to name different control outputs.
+   * </pre>
+   *
+   * <code>repeated string control_output = 20;</code>
+   */
+  java.lang.String getControlOutput(int index);
+  /**
+   * <pre>
+   * Named control outputs for this operation. Useful only for composite
+   * operations (i.e. functions) which want to name different control outputs.
+   * </pre>
+   *
+   * <code>repeated string control_output = 20;</code>
+   */
+  com.google.protobuf.ByteString
+      getControlOutputBytes(int index);
+
+  /**
+   * <code>repeated .tensorflow.OpDef.AttrDef attr = 4;</code>
+   */
+  java.util.List<org.tensorflow.proto.framework.OpDef.AttrDef> 
+      getAttrList();
+  /**
+   * <code>repeated .tensorflow.OpDef.AttrDef attr = 4;</code>
+   */
+  org.tensorflow.proto.framework.OpDef.AttrDef getAttr(int index);
+  /**
+   * <code>repeated .tensorflow.OpDef.AttrDef attr = 4;</code>
+   */
+  int getAttrCount();
+  /**
+   * <code>repeated .tensorflow.OpDef.AttrDef attr = 4;</code>
+   */
+  java.util.List<? extends org.tensorflow.proto.framework.OpDef.AttrDefOrBuilder> 
+      getAttrOrBuilderList();
+  /**
+   * <code>repeated .tensorflow.OpDef.AttrDef attr = 4;</code>
+   */
+  org.tensorflow.proto.framework.OpDef.AttrDefOrBuilder getAttrOrBuilder(
+      int index);
+
+  /**
+   * <pre>
+   * Optional deprecation based on GraphDef versions.
+   * </pre>
+   *
+   * <code>.tensorflow.OpDeprecation deprecation = 8;</code>
+   */
+  boolean hasDeprecation();
+  /**
+   * <pre>
+   * Optional deprecation based on GraphDef versions.
+   * </pre>
+   *
+   * <code>.tensorflow.OpDeprecation deprecation = 8;</code>
+   */
+  org.tensorflow.proto.framework.OpDeprecation getDeprecation();
+  /**
+   * <pre>
+   * Optional deprecation based on GraphDef versions.
+   * </pre>
+   *
+   * <code>.tensorflow.OpDeprecation deprecation = 8;</code>
+   */
+  org.tensorflow.proto.framework.OpDeprecationOrBuilder getDeprecationOrBuilder();
+
+  /**
+   * <pre>
+   * One-line human-readable description of what the Op does.
+   * </pre>
+   *
+   * <code>string summary = 5;</code>
+   */
+  java.lang.String getSummary();
+  /**
+   * <pre>
+   * One-line human-readable description of what the Op does.
+   * </pre>
+   *
+   * <code>string summary = 5;</code>
+   */
+  com.google.protobuf.ByteString
+      getSummaryBytes();
+
+  /**
+   * <pre>
+   * Additional, longer human-readable description of what the Op does.
+   * </pre>
+   *
+   * <code>string description = 6;</code>
+   */
+  java.lang.String getDescription();
+  /**
+   * <pre>
+   * Additional, longer human-readable description of what the Op does.
+   * </pre>
+   *
+   * <code>string description = 6;</code>
+   */
+  com.google.protobuf.ByteString
+      getDescriptionBytes();
+
+  /**
+   * <pre>
+   * True if the operation is commutative ("op(a,b) == op(b,a)" for all inputs)
+   * </pre>
+   *
+   * <code>bool is_commutative = 18;</code>
+   */
+  boolean getIsCommutative();
+
+  /**
+   * <pre>
+   * If is_aggregate is true, then this operation accepts N &gt;= 2
+   * inputs and produces 1 output all of the same type.  Should be
+   * associative and commutative, and produce output with the same
+   * shape as the input.  The optimizer may replace an aggregate op
+   * taking input from multiple devices with a tree of aggregate ops
+   * that aggregate locally within each device (and possibly within
+   * groups of nearby devices) before communicating.
+   * TODO(josh11b): Implement that optimization.
+   * </pre>
+   *
+   * <code>bool is_aggregate = 16;</code>
+   */
+  boolean getIsAggregate();
+
+  /**
+   * <pre>
+   * Ops are marked as stateful if their behavior depends on some state beyond
+   * their input tensors (e.g. variable reading op) or if they have
+   * a side-effect (e.g. printing or asserting ops). Equivalently, stateless ops
+   * must always produce the same output for the same input and have
+   * no side-effects.
+   * By default Ops may be moved between devices.  Stateful ops should
+   * either not be moved, or should only be moved if that state can also
+   * be moved (e.g. via some sort of save / restore).
+   * Stateful ops are guaranteed to never be optimized away by Common
+   * Subexpression Elimination (CSE).
+   * </pre>
+   *
+   * <code>bool is_stateful = 17;</code>
+   */
+  boolean getIsStateful();
+
+  /**
+   * <pre>
+   * By default, all inputs to an Op must be initialized Tensors.  Ops
+   * that may initialize tensors for the first time should set this
+   * field to true, to allow the Op to take an uninitialized Tensor as
+   * input.
+   * </pre>
+   *
+   * <code>bool allows_uninitialized_input = 19;</code>
+   */
+  boolean getAllowsUninitializedInput();
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/OpDefProtos.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/OpDefProtos.java
new file mode 100644
index 00000000000..8beef161f83
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/OpDefProtos.java
@@ -0,0 +1,121 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/framework/op_def.proto
+
+package org.tensorflow.proto.framework;
+
+public final class OpDefProtos {
+  private OpDefProtos() {}
+  public static void registerAllExtensions(
+      com.google.protobuf.ExtensionRegistryLite registry) {
+  }
+
+  public static void registerAllExtensions(
+      com.google.protobuf.ExtensionRegistry registry) {
+    registerAllExtensions(
+        (com.google.protobuf.ExtensionRegistryLite) registry);
+  }
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_OpDef_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_OpDef_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_OpDef_ArgDef_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_OpDef_ArgDef_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_OpDef_AttrDef_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_OpDef_AttrDef_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_OpDeprecation_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_OpDeprecation_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_OpList_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_OpList_fieldAccessorTable;
+
+  public static com.google.protobuf.Descriptors.FileDescriptor
+      getDescriptor() {
+    return descriptor;
+  }
+  private static  com.google.protobuf.Descriptors.FileDescriptor
+      descriptor;
+  static {
+    java.lang.String[] descriptorData = {
+      "\n&tensorflow/core/framework/op_def.proto" +
+      "\022\ntensorflow\032*tensorflow/core/framework/" +
+      "attr_value.proto\032%tensorflow/core/framew" +
+      "ork/types.proto\"\320\005\n\005OpDef\022\014\n\004name\030\001 \001(\t\022" +
+      "+\n\tinput_arg\030\002 \003(\0132\030.tensorflow.OpDef.Ar" +
+      "gDef\022,\n\noutput_arg\030\003 \003(\0132\030.tensorflow.Op" +
+      "Def.ArgDef\022\026\n\016control_output\030\024 \003(\t\022\'\n\004at" +
+      "tr\030\004 \003(\0132\031.tensorflow.OpDef.AttrDef\022.\n\013d" +
+      "eprecation\030\010 \001(\0132\031.tensorflow.OpDeprecat" +
+      "ion\022\017\n\007summary\030\005 \001(\t\022\023\n\013description\030\006 \001(" +
+      "\t\022\026\n\016is_commutative\030\022 \001(\010\022\024\n\014is_aggregat" +
+      "e\030\020 \001(\010\022\023\n\013is_stateful\030\021 \001(\010\022\"\n\032allows_u" +
+      "ninitialized_input\030\023 \001(\010\032\237\001\n\006ArgDef\022\014\n\004n" +
+      "ame\030\001 \001(\t\022\023\n\013description\030\002 \001(\t\022\"\n\004type\030\003" +
+      " \001(\0162\024.tensorflow.DataType\022\021\n\ttype_attr\030" +
+      "\004 \001(\t\022\023\n\013number_attr\030\005 \001(\t\022\026\n\016type_list_" +
+      "attr\030\006 \001(\t\022\016\n\006is_ref\030\020 \001(\010\032\275\001\n\007AttrDef\022\014" +
+      "\n\004name\030\001 \001(\t\022\014\n\004type\030\002 \001(\t\022,\n\rdefault_va" +
+      "lue\030\003 \001(\0132\025.tensorflow.AttrValue\022\023\n\013desc" +
+      "ription\030\004 \001(\t\022\023\n\013has_minimum\030\005 \001(\010\022\017\n\007mi" +
+      "nimum\030\006 \001(\003\022-\n\016allowed_values\030\007 \001(\0132\025.te" +
+      "nsorflow.AttrValue\"5\n\rOpDeprecation\022\017\n\007v" +
+      "ersion\030\001 \001(\005\022\023\n\013explanation\030\002 \001(\t\"\'\n\006OpL" +
+      "ist\022\035\n\002op\030\001 \003(\0132\021.tensorflow.OpDefB\201\001\n\036o" +
+      "rg.tensorflow.proto.frameworkB\013OpDefProt" +
+      "osP\001ZMgithub.com/tensorflow/tensorflow/t" +
+      "ensorflow/go/core/framework/op_def_go_pr" +
+      "oto\370\001\001b\006proto3"
+    };
+    descriptor = com.google.protobuf.Descriptors.FileDescriptor
+      .internalBuildGeneratedFileFrom(descriptorData,
+        new com.google.protobuf.Descriptors.FileDescriptor[] {
+          org.tensorflow.proto.framework.AttrValueProtos.getDescriptor(),
+          org.tensorflow.proto.framework.TypesProtos.getDescriptor(),
+        });
+    internal_static_tensorflow_OpDef_descriptor =
+      getDescriptor().getMessageTypes().get(0);
+    internal_static_tensorflow_OpDef_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_OpDef_descriptor,
+        new java.lang.String[] { "Name", "InputArg", "OutputArg", "ControlOutput", "Attr", "Deprecation", "Summary", "Description", "IsCommutative", "IsAggregate", "IsStateful", "AllowsUninitializedInput", });
+    internal_static_tensorflow_OpDef_ArgDef_descriptor =
+      internal_static_tensorflow_OpDef_descriptor.getNestedTypes().get(0);
+    internal_static_tensorflow_OpDef_ArgDef_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_OpDef_ArgDef_descriptor,
+        new java.lang.String[] { "Name", "Description", "Type", "TypeAttr", "NumberAttr", "TypeListAttr", "IsRef", });
+    internal_static_tensorflow_OpDef_AttrDef_descriptor =
+      internal_static_tensorflow_OpDef_descriptor.getNestedTypes().get(1);
+    internal_static_tensorflow_OpDef_AttrDef_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_OpDef_AttrDef_descriptor,
+        new java.lang.String[] { "Name", "Type", "DefaultValue", "Description", "HasMinimum", "Minimum", "AllowedValues", });
+    internal_static_tensorflow_OpDeprecation_descriptor =
+      getDescriptor().getMessageTypes().get(1);
+    internal_static_tensorflow_OpDeprecation_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_OpDeprecation_descriptor,
+        new java.lang.String[] { "Version", "Explanation", });
+    internal_static_tensorflow_OpList_descriptor =
+      getDescriptor().getMessageTypes().get(2);
+    internal_static_tensorflow_OpList_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_OpList_descriptor,
+        new java.lang.String[] { "Op", });
+    org.tensorflow.proto.framework.AttrValueProtos.getDescriptor();
+    org.tensorflow.proto.framework.TypesProtos.getDescriptor();
+  }
+
+  // @@protoc_insertion_point(outer_class_scope)
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/OpDeprecation.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/OpDeprecation.java
new file mode 100644
index 00000000000..f33a34c9052
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/OpDeprecation.java
@@ -0,0 +1,655 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/framework/op_def.proto
+
+package org.tensorflow.proto.framework;
+
+/**
+ * <pre>
+ * Information about version-dependent deprecation of an op
+ * </pre>
+ *
+ * Protobuf type {@code tensorflow.OpDeprecation}
+ */
+public  final class OpDeprecation extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.OpDeprecation)
+    OpDeprecationOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use OpDeprecation.newBuilder() to construct.
+  private OpDeprecation(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private OpDeprecation() {
+    explanation_ = "";
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new OpDeprecation();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private OpDeprecation(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 8: {
+
+            version_ = input.readInt32();
+            break;
+          }
+          case 18: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            explanation_ = s;
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.framework.OpDefProtos.internal_static_tensorflow_OpDeprecation_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.framework.OpDefProtos.internal_static_tensorflow_OpDeprecation_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.framework.OpDeprecation.class, org.tensorflow.proto.framework.OpDeprecation.Builder.class);
+  }
+
+  public static final int VERSION_FIELD_NUMBER = 1;
+  private int version_;
+  /**
+   * <pre>
+   * First GraphDef version at which the op is disallowed.
+   * </pre>
+   *
+   * <code>int32 version = 1;</code>
+   */
+  public int getVersion() {
+    return version_;
+  }
+
+  public static final int EXPLANATION_FIELD_NUMBER = 2;
+  private volatile java.lang.Object explanation_;
+  /**
+   * <pre>
+   * Explanation of why it was deprecated and what to use instead.
+   * </pre>
+   *
+   * <code>string explanation = 2;</code>
+   */
+  public java.lang.String getExplanation() {
+    java.lang.Object ref = explanation_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      explanation_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * Explanation of why it was deprecated and what to use instead.
+   * </pre>
+   *
+   * <code>string explanation = 2;</code>
+   */
+  public com.google.protobuf.ByteString
+      getExplanationBytes() {
+    java.lang.Object ref = explanation_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      explanation_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    if (version_ != 0) {
+      output.writeInt32(1, version_);
+    }
+    if (!getExplanationBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 2, explanation_);
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (version_ != 0) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt32Size(1, version_);
+    }
+    if (!getExplanationBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(2, explanation_);
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.framework.OpDeprecation)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.framework.OpDeprecation other = (org.tensorflow.proto.framework.OpDeprecation) obj;
+
+    if (getVersion()
+        != other.getVersion()) return false;
+    if (!getExplanation()
+        .equals(other.getExplanation())) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    hash = (37 * hash) + VERSION_FIELD_NUMBER;
+    hash = (53 * hash) + getVersion();
+    hash = (37 * hash) + EXPLANATION_FIELD_NUMBER;
+    hash = (53 * hash) + getExplanation().hashCode();
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.framework.OpDeprecation parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.OpDeprecation parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.OpDeprecation parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.OpDeprecation parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.OpDeprecation parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.OpDeprecation parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.OpDeprecation parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.OpDeprecation parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.OpDeprecation parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.OpDeprecation parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.OpDeprecation parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.OpDeprecation parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.framework.OpDeprecation prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * <pre>
+   * Information about version-dependent deprecation of an op
+   * </pre>
+   *
+   * Protobuf type {@code tensorflow.OpDeprecation}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.OpDeprecation)
+      org.tensorflow.proto.framework.OpDeprecationOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.OpDefProtos.internal_static_tensorflow_OpDeprecation_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.OpDefProtos.internal_static_tensorflow_OpDeprecation_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.OpDeprecation.class, org.tensorflow.proto.framework.OpDeprecation.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.framework.OpDeprecation.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      version_ = 0;
+
+      explanation_ = "";
+
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.framework.OpDefProtos.internal_static_tensorflow_OpDeprecation_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.OpDeprecation getDefaultInstanceForType() {
+      return org.tensorflow.proto.framework.OpDeprecation.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.OpDeprecation build() {
+      org.tensorflow.proto.framework.OpDeprecation result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.OpDeprecation buildPartial() {
+      org.tensorflow.proto.framework.OpDeprecation result = new org.tensorflow.proto.framework.OpDeprecation(this);
+      result.version_ = version_;
+      result.explanation_ = explanation_;
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.framework.OpDeprecation) {
+        return mergeFrom((org.tensorflow.proto.framework.OpDeprecation)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.framework.OpDeprecation other) {
+      if (other == org.tensorflow.proto.framework.OpDeprecation.getDefaultInstance()) return this;
+      if (other.getVersion() != 0) {
+        setVersion(other.getVersion());
+      }
+      if (!other.getExplanation().isEmpty()) {
+        explanation_ = other.explanation_;
+        onChanged();
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.framework.OpDeprecation parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.framework.OpDeprecation) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+
+    private int version_ ;
+    /**
+     * <pre>
+     * First GraphDef version at which the op is disallowed.
+     * </pre>
+     *
+     * <code>int32 version = 1;</code>
+     */
+    public int getVersion() {
+      return version_;
+    }
+    /**
+     * <pre>
+     * First GraphDef version at which the op is disallowed.
+     * </pre>
+     *
+     * <code>int32 version = 1;</code>
+     */
+    public Builder setVersion(int value) {
+      
+      version_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * First GraphDef version at which the op is disallowed.
+     * </pre>
+     *
+     * <code>int32 version = 1;</code>
+     */
+    public Builder clearVersion() {
+      
+      version_ = 0;
+      onChanged();
+      return this;
+    }
+
+    private java.lang.Object explanation_ = "";
+    /**
+     * <pre>
+     * Explanation of why it was deprecated and what to use instead.
+     * </pre>
+     *
+     * <code>string explanation = 2;</code>
+     */
+    public java.lang.String getExplanation() {
+      java.lang.Object ref = explanation_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        explanation_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Explanation of why it was deprecated and what to use instead.
+     * </pre>
+     *
+     * <code>string explanation = 2;</code>
+     */
+    public com.google.protobuf.ByteString
+        getExplanationBytes() {
+      java.lang.Object ref = explanation_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        explanation_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Explanation of why it was deprecated and what to use instead.
+     * </pre>
+     *
+     * <code>string explanation = 2;</code>
+     */
+    public Builder setExplanation(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      explanation_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Explanation of why it was deprecated and what to use instead.
+     * </pre>
+     *
+     * <code>string explanation = 2;</code>
+     */
+    public Builder clearExplanation() {
+      
+      explanation_ = getDefaultInstance().getExplanation();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Explanation of why it was deprecated and what to use instead.
+     * </pre>
+     *
+     * <code>string explanation = 2;</code>
+     */
+    public Builder setExplanationBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      explanation_ = value;
+      onChanged();
+      return this;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.OpDeprecation)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.OpDeprecation)
+  private static final org.tensorflow.proto.framework.OpDeprecation DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.OpDeprecation();
+  }
+
+  public static org.tensorflow.proto.framework.OpDeprecation getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<OpDeprecation>
+      PARSER = new com.google.protobuf.AbstractParser<OpDeprecation>() {
+    @java.lang.Override
+    public OpDeprecation parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new OpDeprecation(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<OpDeprecation> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<OpDeprecation> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.framework.OpDeprecation getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/OpDeprecationOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/OpDeprecationOrBuilder.java
new file mode 100644
index 00000000000..3248db05508
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/OpDeprecationOrBuilder.java
@@ -0,0 +1,36 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/framework/op_def.proto
+
+package org.tensorflow.proto.framework;
+
+public interface OpDeprecationOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.OpDeprecation)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <pre>
+   * First GraphDef version at which the op is disallowed.
+   * </pre>
+   *
+   * <code>int32 version = 1;</code>
+   */
+  int getVersion();
+
+  /**
+   * <pre>
+   * Explanation of why it was deprecated and what to use instead.
+   * </pre>
+   *
+   * <code>string explanation = 2;</code>
+   */
+  java.lang.String getExplanation();
+  /**
+   * <pre>
+   * Explanation of why it was deprecated and what to use instead.
+   * </pre>
+   *
+   * <code>string explanation = 2;</code>
+   */
+  com.google.protobuf.ByteString
+      getExplanationBytes();
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/OpList.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/OpList.java
new file mode 100644
index 00000000000..1be81195b29
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/OpList.java
@@ -0,0 +1,773 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/framework/op_def.proto
+
+package org.tensorflow.proto.framework;
+
+/**
+ * <pre>
+ * A collection of OpDefs
+ * </pre>
+ *
+ * Protobuf type {@code tensorflow.OpList}
+ */
+public  final class OpList extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.OpList)
+    OpListOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use OpList.newBuilder() to construct.
+  private OpList(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private OpList() {
+    op_ = java.util.Collections.emptyList();
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new OpList();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private OpList(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    int mutable_bitField0_ = 0;
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 10: {
+            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
+              op_ = new java.util.ArrayList<org.tensorflow.proto.framework.OpDef>();
+              mutable_bitField0_ |= 0x00000001;
+            }
+            op_.add(
+                input.readMessage(org.tensorflow.proto.framework.OpDef.parser(), extensionRegistry));
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      if (((mutable_bitField0_ & 0x00000001) != 0)) {
+        op_ = java.util.Collections.unmodifiableList(op_);
+      }
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.framework.OpDefProtos.internal_static_tensorflow_OpList_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.framework.OpDefProtos.internal_static_tensorflow_OpList_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.framework.OpList.class, org.tensorflow.proto.framework.OpList.Builder.class);
+  }
+
+  public static final int OP_FIELD_NUMBER = 1;
+  private java.util.List<org.tensorflow.proto.framework.OpDef> op_;
+  /**
+   * <code>repeated .tensorflow.OpDef op = 1;</code>
+   */
+  public java.util.List<org.tensorflow.proto.framework.OpDef> getOpList() {
+    return op_;
+  }
+  /**
+   * <code>repeated .tensorflow.OpDef op = 1;</code>
+   */
+  public java.util.List<? extends org.tensorflow.proto.framework.OpDefOrBuilder> 
+      getOpOrBuilderList() {
+    return op_;
+  }
+  /**
+   * <code>repeated .tensorflow.OpDef op = 1;</code>
+   */
+  public int getOpCount() {
+    return op_.size();
+  }
+  /**
+   * <code>repeated .tensorflow.OpDef op = 1;</code>
+   */
+  public org.tensorflow.proto.framework.OpDef getOp(int index) {
+    return op_.get(index);
+  }
+  /**
+   * <code>repeated .tensorflow.OpDef op = 1;</code>
+   */
+  public org.tensorflow.proto.framework.OpDefOrBuilder getOpOrBuilder(
+      int index) {
+    return op_.get(index);
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    for (int i = 0; i < op_.size(); i++) {
+      output.writeMessage(1, op_.get(i));
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    for (int i = 0; i < op_.size(); i++) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(1, op_.get(i));
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.framework.OpList)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.framework.OpList other = (org.tensorflow.proto.framework.OpList) obj;
+
+    if (!getOpList()
+        .equals(other.getOpList())) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    if (getOpCount() > 0) {
+      hash = (37 * hash) + OP_FIELD_NUMBER;
+      hash = (53 * hash) + getOpList().hashCode();
+    }
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.framework.OpList parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.OpList parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.OpList parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.OpList parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.OpList parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.OpList parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.OpList parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.OpList parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.OpList parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.OpList parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.OpList parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.OpList parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.framework.OpList prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * <pre>
+   * A collection of OpDefs
+   * </pre>
+   *
+   * Protobuf type {@code tensorflow.OpList}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.OpList)
+      org.tensorflow.proto.framework.OpListOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.OpDefProtos.internal_static_tensorflow_OpList_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.OpDefProtos.internal_static_tensorflow_OpList_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.OpList.class, org.tensorflow.proto.framework.OpList.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.framework.OpList.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+        getOpFieldBuilder();
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      if (opBuilder_ == null) {
+        op_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000001);
+      } else {
+        opBuilder_.clear();
+      }
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.framework.OpDefProtos.internal_static_tensorflow_OpList_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.OpList getDefaultInstanceForType() {
+      return org.tensorflow.proto.framework.OpList.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.OpList build() {
+      org.tensorflow.proto.framework.OpList result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.OpList buildPartial() {
+      org.tensorflow.proto.framework.OpList result = new org.tensorflow.proto.framework.OpList(this);
+      int from_bitField0_ = bitField0_;
+      if (opBuilder_ == null) {
+        if (((bitField0_ & 0x00000001) != 0)) {
+          op_ = java.util.Collections.unmodifiableList(op_);
+          bitField0_ = (bitField0_ & ~0x00000001);
+        }
+        result.op_ = op_;
+      } else {
+        result.op_ = opBuilder_.build();
+      }
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.framework.OpList) {
+        return mergeFrom((org.tensorflow.proto.framework.OpList)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.framework.OpList other) {
+      if (other == org.tensorflow.proto.framework.OpList.getDefaultInstance()) return this;
+      if (opBuilder_ == null) {
+        if (!other.op_.isEmpty()) {
+          if (op_.isEmpty()) {
+            op_ = other.op_;
+            bitField0_ = (bitField0_ & ~0x00000001);
+          } else {
+            ensureOpIsMutable();
+            op_.addAll(other.op_);
+          }
+          onChanged();
+        }
+      } else {
+        if (!other.op_.isEmpty()) {
+          if (opBuilder_.isEmpty()) {
+            opBuilder_.dispose();
+            opBuilder_ = null;
+            op_ = other.op_;
+            bitField0_ = (bitField0_ & ~0x00000001);
+            opBuilder_ = 
+              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
+                 getOpFieldBuilder() : null;
+          } else {
+            opBuilder_.addAllMessages(other.op_);
+          }
+        }
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.framework.OpList parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.framework.OpList) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+    private int bitField0_;
+
+    private java.util.List<org.tensorflow.proto.framework.OpDef> op_ =
+      java.util.Collections.emptyList();
+    private void ensureOpIsMutable() {
+      if (!((bitField0_ & 0x00000001) != 0)) {
+        op_ = new java.util.ArrayList<org.tensorflow.proto.framework.OpDef>(op_);
+        bitField0_ |= 0x00000001;
+       }
+    }
+
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.framework.OpDef, org.tensorflow.proto.framework.OpDef.Builder, org.tensorflow.proto.framework.OpDefOrBuilder> opBuilder_;
+
+    /**
+     * <code>repeated .tensorflow.OpDef op = 1;</code>
+     */
+    public java.util.List<org.tensorflow.proto.framework.OpDef> getOpList() {
+      if (opBuilder_ == null) {
+        return java.util.Collections.unmodifiableList(op_);
+      } else {
+        return opBuilder_.getMessageList();
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.OpDef op = 1;</code>
+     */
+    public int getOpCount() {
+      if (opBuilder_ == null) {
+        return op_.size();
+      } else {
+        return opBuilder_.getCount();
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.OpDef op = 1;</code>
+     */
+    public org.tensorflow.proto.framework.OpDef getOp(int index) {
+      if (opBuilder_ == null) {
+        return op_.get(index);
+      } else {
+        return opBuilder_.getMessage(index);
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.OpDef op = 1;</code>
+     */
+    public Builder setOp(
+        int index, org.tensorflow.proto.framework.OpDef value) {
+      if (opBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureOpIsMutable();
+        op_.set(index, value);
+        onChanged();
+      } else {
+        opBuilder_.setMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.OpDef op = 1;</code>
+     */
+    public Builder setOp(
+        int index, org.tensorflow.proto.framework.OpDef.Builder builderForValue) {
+      if (opBuilder_ == null) {
+        ensureOpIsMutable();
+        op_.set(index, builderForValue.build());
+        onChanged();
+      } else {
+        opBuilder_.setMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.OpDef op = 1;</code>
+     */
+    public Builder addOp(org.tensorflow.proto.framework.OpDef value) {
+      if (opBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureOpIsMutable();
+        op_.add(value);
+        onChanged();
+      } else {
+        opBuilder_.addMessage(value);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.OpDef op = 1;</code>
+     */
+    public Builder addOp(
+        int index, org.tensorflow.proto.framework.OpDef value) {
+      if (opBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureOpIsMutable();
+        op_.add(index, value);
+        onChanged();
+      } else {
+        opBuilder_.addMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.OpDef op = 1;</code>
+     */
+    public Builder addOp(
+        org.tensorflow.proto.framework.OpDef.Builder builderForValue) {
+      if (opBuilder_ == null) {
+        ensureOpIsMutable();
+        op_.add(builderForValue.build());
+        onChanged();
+      } else {
+        opBuilder_.addMessage(builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.OpDef op = 1;</code>
+     */
+    public Builder addOp(
+        int index, org.tensorflow.proto.framework.OpDef.Builder builderForValue) {
+      if (opBuilder_ == null) {
+        ensureOpIsMutable();
+        op_.add(index, builderForValue.build());
+        onChanged();
+      } else {
+        opBuilder_.addMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.OpDef op = 1;</code>
+     */
+    public Builder addAllOp(
+        java.lang.Iterable<? extends org.tensorflow.proto.framework.OpDef> values) {
+      if (opBuilder_ == null) {
+        ensureOpIsMutable();
+        com.google.protobuf.AbstractMessageLite.Builder.addAll(
+            values, op_);
+        onChanged();
+      } else {
+        opBuilder_.addAllMessages(values);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.OpDef op = 1;</code>
+     */
+    public Builder clearOp() {
+      if (opBuilder_ == null) {
+        op_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000001);
+        onChanged();
+      } else {
+        opBuilder_.clear();
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.OpDef op = 1;</code>
+     */
+    public Builder removeOp(int index) {
+      if (opBuilder_ == null) {
+        ensureOpIsMutable();
+        op_.remove(index);
+        onChanged();
+      } else {
+        opBuilder_.remove(index);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.OpDef op = 1;</code>
+     */
+    public org.tensorflow.proto.framework.OpDef.Builder getOpBuilder(
+        int index) {
+      return getOpFieldBuilder().getBuilder(index);
+    }
+    /**
+     * <code>repeated .tensorflow.OpDef op = 1;</code>
+     */
+    public org.tensorflow.proto.framework.OpDefOrBuilder getOpOrBuilder(
+        int index) {
+      if (opBuilder_ == null) {
+        return op_.get(index);  } else {
+        return opBuilder_.getMessageOrBuilder(index);
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.OpDef op = 1;</code>
+     */
+    public java.util.List<? extends org.tensorflow.proto.framework.OpDefOrBuilder> 
+         getOpOrBuilderList() {
+      if (opBuilder_ != null) {
+        return opBuilder_.getMessageOrBuilderList();
+      } else {
+        return java.util.Collections.unmodifiableList(op_);
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.OpDef op = 1;</code>
+     */
+    public org.tensorflow.proto.framework.OpDef.Builder addOpBuilder() {
+      return getOpFieldBuilder().addBuilder(
+          org.tensorflow.proto.framework.OpDef.getDefaultInstance());
+    }
+    /**
+     * <code>repeated .tensorflow.OpDef op = 1;</code>
+     */
+    public org.tensorflow.proto.framework.OpDef.Builder addOpBuilder(
+        int index) {
+      return getOpFieldBuilder().addBuilder(
+          index, org.tensorflow.proto.framework.OpDef.getDefaultInstance());
+    }
+    /**
+     * <code>repeated .tensorflow.OpDef op = 1;</code>
+     */
+    public java.util.List<org.tensorflow.proto.framework.OpDef.Builder> 
+         getOpBuilderList() {
+      return getOpFieldBuilder().getBuilderList();
+    }
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.framework.OpDef, org.tensorflow.proto.framework.OpDef.Builder, org.tensorflow.proto.framework.OpDefOrBuilder> 
+        getOpFieldBuilder() {
+      if (opBuilder_ == null) {
+        opBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
+            org.tensorflow.proto.framework.OpDef, org.tensorflow.proto.framework.OpDef.Builder, org.tensorflow.proto.framework.OpDefOrBuilder>(
+                op_,
+                ((bitField0_ & 0x00000001) != 0),
+                getParentForChildren(),
+                isClean());
+        op_ = null;
+      }
+      return opBuilder_;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.OpList)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.OpList)
+  private static final org.tensorflow.proto.framework.OpList DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.OpList();
+  }
+
+  public static org.tensorflow.proto.framework.OpList getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<OpList>
+      PARSER = new com.google.protobuf.AbstractParser<OpList>() {
+    @java.lang.Override
+    public OpList parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new OpList(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<OpList> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<OpList> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.framework.OpList getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/OpListOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/OpListOrBuilder.java
new file mode 100644
index 00000000000..a3fc1fa9856
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/OpListOrBuilder.java
@@ -0,0 +1,33 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/framework/op_def.proto
+
+package org.tensorflow.proto.framework;
+
+public interface OpListOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.OpList)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <code>repeated .tensorflow.OpDef op = 1;</code>
+   */
+  java.util.List<org.tensorflow.proto.framework.OpDef> 
+      getOpList();
+  /**
+   * <code>repeated .tensorflow.OpDef op = 1;</code>
+   */
+  org.tensorflow.proto.framework.OpDef getOp(int index);
+  /**
+   * <code>repeated .tensorflow.OpDef op = 1;</code>
+   */
+  int getOpCount();
+  /**
+   * <code>repeated .tensorflow.OpDef op = 1;</code>
+   */
+  java.util.List<? extends org.tensorflow.proto.framework.OpDefOrBuilder> 
+      getOpOrBuilderList();
+  /**
+   * <code>repeated .tensorflow.OpDef op = 1;</code>
+   */
+  org.tensorflow.proto.framework.OpDefOrBuilder getOpOrBuilder(
+      int index);
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/OptimizerOptions.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/OptimizerOptions.java
new file mode 100644
index 00000000000..2afa5a96067
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/OptimizerOptions.java
@@ -0,0 +1,1210 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/config.proto
+
+package org.tensorflow.proto.framework;
+
+/**
+ * <pre>
+ * Options passed to the graph optimizer
+ * </pre>
+ *
+ * Protobuf type {@code tensorflow.OptimizerOptions}
+ */
+public  final class OptimizerOptions extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.OptimizerOptions)
+    OptimizerOptionsOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use OptimizerOptions.newBuilder() to construct.
+  private OptimizerOptions(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private OptimizerOptions() {
+    optLevel_ = 0;
+    globalJitLevel_ = 0;
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new OptimizerOptions();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private OptimizerOptions(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 8: {
+
+            doCommonSubexpressionElimination_ = input.readBool();
+            break;
+          }
+          case 16: {
+
+            doConstantFolding_ = input.readBool();
+            break;
+          }
+          case 24: {
+            int rawValue = input.readEnum();
+
+            optLevel_ = rawValue;
+            break;
+          }
+          case 32: {
+
+            doFunctionInlining_ = input.readBool();
+            break;
+          }
+          case 40: {
+            int rawValue = input.readEnum();
+
+            globalJitLevel_ = rawValue;
+            break;
+          }
+          case 48: {
+
+            maxFoldedConstantInBytes_ = input.readInt64();
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_OptimizerOptions_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_OptimizerOptions_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.framework.OptimizerOptions.class, org.tensorflow.proto.framework.OptimizerOptions.Builder.class);
+  }
+
+  /**
+   * <pre>
+   * Optimization level
+   * </pre>
+   *
+   * Protobuf enum {@code tensorflow.OptimizerOptions.Level}
+   */
+  public enum Level
+      implements com.google.protobuf.ProtocolMessageEnum {
+    /**
+     * <pre>
+     * L1 is the default level.
+     * Optimization performed at L1 :
+     * 1. Common subexpression elimination
+     * 2. Constant folding
+     * </pre>
+     *
+     * <code>L1 = 0;</code>
+     */
+    L1(0),
+    /**
+     * <pre>
+     * No optimizations
+     * </pre>
+     *
+     * <code>L0 = -1;</code>
+     */
+    L0(-1),
+    UNRECOGNIZED(-1),
+    ;
+
+    /**
+     * <pre>
+     * L1 is the default level.
+     * Optimization performed at L1 :
+     * 1. Common subexpression elimination
+     * 2. Constant folding
+     * </pre>
+     *
+     * <code>L1 = 0;</code>
+     */
+    public static final int L1_VALUE = 0;
+    /**
+     * <pre>
+     * No optimizations
+     * </pre>
+     *
+     * <code>L0 = -1;</code>
+     */
+    public static final int L0_VALUE = -1;
+
+
+    public final int getNumber() {
+      if (this == UNRECOGNIZED) {
+        throw new java.lang.IllegalArgumentException(
+            "Can't get the number of an unknown enum value.");
+      }
+      return value;
+    }
+
+    /**
+     * @deprecated Use {@link #forNumber(int)} instead.
+     */
+    @java.lang.Deprecated
+    public static Level valueOf(int value) {
+      return forNumber(value);
+    }
+
+    public static Level forNumber(int value) {
+      switch (value) {
+        case 0: return L1;
+        case -1: return L0;
+        default: return null;
+      }
+    }
+
+    public static com.google.protobuf.Internal.EnumLiteMap<Level>
+        internalGetValueMap() {
+      return internalValueMap;
+    }
+    private static final com.google.protobuf.Internal.EnumLiteMap<
+        Level> internalValueMap =
+          new com.google.protobuf.Internal.EnumLiteMap<Level>() {
+            public Level findValueByNumber(int number) {
+              return Level.forNumber(number);
+            }
+          };
+
+    public final com.google.protobuf.Descriptors.EnumValueDescriptor
+        getValueDescriptor() {
+      return getDescriptor().getValues().get(ordinal());
+    }
+    public final com.google.protobuf.Descriptors.EnumDescriptor
+        getDescriptorForType() {
+      return getDescriptor();
+    }
+    public static final com.google.protobuf.Descriptors.EnumDescriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.OptimizerOptions.getDescriptor().getEnumTypes().get(0);
+    }
+
+    private static final Level[] VALUES = values();
+
+    public static Level valueOf(
+        com.google.protobuf.Descriptors.EnumValueDescriptor desc) {
+      if (desc.getType() != getDescriptor()) {
+        throw new java.lang.IllegalArgumentException(
+          "EnumValueDescriptor is not for this type.");
+      }
+      if (desc.getIndex() == -1) {
+        return UNRECOGNIZED;
+      }
+      return VALUES[desc.getIndex()];
+    }
+
+    private final int value;
+
+    private Level(int value) {
+      this.value = value;
+    }
+
+    // @@protoc_insertion_point(enum_scope:tensorflow.OptimizerOptions.Level)
+  }
+
+  /**
+   * <pre>
+   * Control the use of the compiler/jit.  Experimental.
+   * </pre>
+   *
+   * Protobuf enum {@code tensorflow.OptimizerOptions.GlobalJitLevel}
+   */
+  public enum GlobalJitLevel
+      implements com.google.protobuf.ProtocolMessageEnum {
+    /**
+     * <pre>
+     * Default setting ("off" now, but later expected to be "on")
+     * </pre>
+     *
+     * <code>DEFAULT = 0;</code>
+     */
+    DEFAULT(0),
+    /**
+     * <code>OFF = -1;</code>
+     */
+    OFF(-1),
+    /**
+     * <pre>
+     * The following settings turn on compilation, with higher values being
+     * more aggressive.  Higher values may reduce opportunities for parallelism
+     * and may use more memory.  (At present, there is no distinction, but this
+     * is expected to change.)
+     * </pre>
+     *
+     * <code>ON_1 = 1;</code>
+     */
+    ON_1(1),
+    /**
+     * <code>ON_2 = 2;</code>
+     */
+    ON_2(2),
+    UNRECOGNIZED(-1),
+    ;
+
+    /**
+     * <pre>
+     * Default setting ("off" now, but later expected to be "on")
+     * </pre>
+     *
+     * <code>DEFAULT = 0;</code>
+     */
+    public static final int DEFAULT_VALUE = 0;
+    /**
+     * <code>OFF = -1;</code>
+     */
+    public static final int OFF_VALUE = -1;
+    /**
+     * <pre>
+     * The following settings turn on compilation, with higher values being
+     * more aggressive.  Higher values may reduce opportunities for parallelism
+     * and may use more memory.  (At present, there is no distinction, but this
+     * is expected to change.)
+     * </pre>
+     *
+     * <code>ON_1 = 1;</code>
+     */
+    public static final int ON_1_VALUE = 1;
+    /**
+     * <code>ON_2 = 2;</code>
+     */
+    public static final int ON_2_VALUE = 2;
+
+
+    public final int getNumber() {
+      if (this == UNRECOGNIZED) {
+        throw new java.lang.IllegalArgumentException(
+            "Can't get the number of an unknown enum value.");
+      }
+      return value;
+    }
+
+    /**
+     * @deprecated Use {@link #forNumber(int)} instead.
+     */
+    @java.lang.Deprecated
+    public static GlobalJitLevel valueOf(int value) {
+      return forNumber(value);
+    }
+
+    public static GlobalJitLevel forNumber(int value) {
+      switch (value) {
+        case 0: return DEFAULT;
+        case -1: return OFF;
+        case 1: return ON_1;
+        case 2: return ON_2;
+        default: return null;
+      }
+    }
+
+    public static com.google.protobuf.Internal.EnumLiteMap<GlobalJitLevel>
+        internalGetValueMap() {
+      return internalValueMap;
+    }
+    private static final com.google.protobuf.Internal.EnumLiteMap<
+        GlobalJitLevel> internalValueMap =
+          new com.google.protobuf.Internal.EnumLiteMap<GlobalJitLevel>() {
+            public GlobalJitLevel findValueByNumber(int number) {
+              return GlobalJitLevel.forNumber(number);
+            }
+          };
+
+    public final com.google.protobuf.Descriptors.EnumValueDescriptor
+        getValueDescriptor() {
+      return getDescriptor().getValues().get(ordinal());
+    }
+    public final com.google.protobuf.Descriptors.EnumDescriptor
+        getDescriptorForType() {
+      return getDescriptor();
+    }
+    public static final com.google.protobuf.Descriptors.EnumDescriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.OptimizerOptions.getDescriptor().getEnumTypes().get(1);
+    }
+
+    private static final GlobalJitLevel[] VALUES = values();
+
+    public static GlobalJitLevel valueOf(
+        com.google.protobuf.Descriptors.EnumValueDescriptor desc) {
+      if (desc.getType() != getDescriptor()) {
+        throw new java.lang.IllegalArgumentException(
+          "EnumValueDescriptor is not for this type.");
+      }
+      if (desc.getIndex() == -1) {
+        return UNRECOGNIZED;
+      }
+      return VALUES[desc.getIndex()];
+    }
+
+    private final int value;
+
+    private GlobalJitLevel(int value) {
+      this.value = value;
+    }
+
+    // @@protoc_insertion_point(enum_scope:tensorflow.OptimizerOptions.GlobalJitLevel)
+  }
+
+  public static final int DO_COMMON_SUBEXPRESSION_ELIMINATION_FIELD_NUMBER = 1;
+  private boolean doCommonSubexpressionElimination_;
+  /**
+   * <pre>
+   * If true, optimize the graph using common subexpression elimination.
+   * </pre>
+   *
+   * <code>bool do_common_subexpression_elimination = 1;</code>
+   */
+  public boolean getDoCommonSubexpressionElimination() {
+    return doCommonSubexpressionElimination_;
+  }
+
+  public static final int DO_CONSTANT_FOLDING_FIELD_NUMBER = 2;
+  private boolean doConstantFolding_;
+  /**
+   * <pre>
+   * If true, perform constant folding optimization on the graph.
+   * </pre>
+   *
+   * <code>bool do_constant_folding = 2;</code>
+   */
+  public boolean getDoConstantFolding() {
+    return doConstantFolding_;
+  }
+
+  public static final int MAX_FOLDED_CONSTANT_IN_BYTES_FIELD_NUMBER = 6;
+  private long maxFoldedConstantInBytes_;
+  /**
+   * <pre>
+   * Constant folding optimization replaces tensors whose values can be
+   * predetermined, with constant nodes. To avoid inserting too large constants,
+   * the size of each constant created can be limited. If this value is zero, a
+   * default limit of 10 MiB will be applied. If constant folding optimization
+   * is disabled, this value is ignored.
+   * </pre>
+   *
+   * <code>int64 max_folded_constant_in_bytes = 6;</code>
+   */
+  public long getMaxFoldedConstantInBytes() {
+    return maxFoldedConstantInBytes_;
+  }
+
+  public static final int DO_FUNCTION_INLINING_FIELD_NUMBER = 4;
+  private boolean doFunctionInlining_;
+  /**
+   * <pre>
+   * If true, perform function inlining on the graph.
+   * </pre>
+   *
+   * <code>bool do_function_inlining = 4;</code>
+   */
+  public boolean getDoFunctionInlining() {
+    return doFunctionInlining_;
+  }
+
+  public static final int OPT_LEVEL_FIELD_NUMBER = 3;
+  private int optLevel_;
+  /**
+   * <pre>
+   * Overall optimization level. The actual optimizations applied will be the
+   * logical OR of the flags that this level implies and any flags already set.
+   * </pre>
+   *
+   * <code>.tensorflow.OptimizerOptions.Level opt_level = 3;</code>
+   */
+  public int getOptLevelValue() {
+    return optLevel_;
+  }
+  /**
+   * <pre>
+   * Overall optimization level. The actual optimizations applied will be the
+   * logical OR of the flags that this level implies and any flags already set.
+   * </pre>
+   *
+   * <code>.tensorflow.OptimizerOptions.Level opt_level = 3;</code>
+   */
+  public org.tensorflow.proto.framework.OptimizerOptions.Level getOptLevel() {
+    @SuppressWarnings("deprecation")
+    org.tensorflow.proto.framework.OptimizerOptions.Level result = org.tensorflow.proto.framework.OptimizerOptions.Level.valueOf(optLevel_);
+    return result == null ? org.tensorflow.proto.framework.OptimizerOptions.Level.UNRECOGNIZED : result;
+  }
+
+  public static final int GLOBAL_JIT_LEVEL_FIELD_NUMBER = 5;
+  private int globalJitLevel_;
+  /**
+   * <code>.tensorflow.OptimizerOptions.GlobalJitLevel global_jit_level = 5;</code>
+   */
+  public int getGlobalJitLevelValue() {
+    return globalJitLevel_;
+  }
+  /**
+   * <code>.tensorflow.OptimizerOptions.GlobalJitLevel global_jit_level = 5;</code>
+   */
+  public org.tensorflow.proto.framework.OptimizerOptions.GlobalJitLevel getGlobalJitLevel() {
+    @SuppressWarnings("deprecation")
+    org.tensorflow.proto.framework.OptimizerOptions.GlobalJitLevel result = org.tensorflow.proto.framework.OptimizerOptions.GlobalJitLevel.valueOf(globalJitLevel_);
+    return result == null ? org.tensorflow.proto.framework.OptimizerOptions.GlobalJitLevel.UNRECOGNIZED : result;
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    if (doCommonSubexpressionElimination_ != false) {
+      output.writeBool(1, doCommonSubexpressionElimination_);
+    }
+    if (doConstantFolding_ != false) {
+      output.writeBool(2, doConstantFolding_);
+    }
+    if (optLevel_ != org.tensorflow.proto.framework.OptimizerOptions.Level.L1.getNumber()) {
+      output.writeEnum(3, optLevel_);
+    }
+    if (doFunctionInlining_ != false) {
+      output.writeBool(4, doFunctionInlining_);
+    }
+    if (globalJitLevel_ != org.tensorflow.proto.framework.OptimizerOptions.GlobalJitLevel.DEFAULT.getNumber()) {
+      output.writeEnum(5, globalJitLevel_);
+    }
+    if (maxFoldedConstantInBytes_ != 0L) {
+      output.writeInt64(6, maxFoldedConstantInBytes_);
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (doCommonSubexpressionElimination_ != false) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeBoolSize(1, doCommonSubexpressionElimination_);
+    }
+    if (doConstantFolding_ != false) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeBoolSize(2, doConstantFolding_);
+    }
+    if (optLevel_ != org.tensorflow.proto.framework.OptimizerOptions.Level.L1.getNumber()) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeEnumSize(3, optLevel_);
+    }
+    if (doFunctionInlining_ != false) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeBoolSize(4, doFunctionInlining_);
+    }
+    if (globalJitLevel_ != org.tensorflow.proto.framework.OptimizerOptions.GlobalJitLevel.DEFAULT.getNumber()) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeEnumSize(5, globalJitLevel_);
+    }
+    if (maxFoldedConstantInBytes_ != 0L) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt64Size(6, maxFoldedConstantInBytes_);
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.framework.OptimizerOptions)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.framework.OptimizerOptions other = (org.tensorflow.proto.framework.OptimizerOptions) obj;
+
+    if (getDoCommonSubexpressionElimination()
+        != other.getDoCommonSubexpressionElimination()) return false;
+    if (getDoConstantFolding()
+        != other.getDoConstantFolding()) return false;
+    if (getMaxFoldedConstantInBytes()
+        != other.getMaxFoldedConstantInBytes()) return false;
+    if (getDoFunctionInlining()
+        != other.getDoFunctionInlining()) return false;
+    if (optLevel_ != other.optLevel_) return false;
+    if (globalJitLevel_ != other.globalJitLevel_) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    hash = (37 * hash) + DO_COMMON_SUBEXPRESSION_ELIMINATION_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashBoolean(
+        getDoCommonSubexpressionElimination());
+    hash = (37 * hash) + DO_CONSTANT_FOLDING_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashBoolean(
+        getDoConstantFolding());
+    hash = (37 * hash) + MAX_FOLDED_CONSTANT_IN_BYTES_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+        getMaxFoldedConstantInBytes());
+    hash = (37 * hash) + DO_FUNCTION_INLINING_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashBoolean(
+        getDoFunctionInlining());
+    hash = (37 * hash) + OPT_LEVEL_FIELD_NUMBER;
+    hash = (53 * hash) + optLevel_;
+    hash = (37 * hash) + GLOBAL_JIT_LEVEL_FIELD_NUMBER;
+    hash = (53 * hash) + globalJitLevel_;
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.framework.OptimizerOptions parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.OptimizerOptions parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.OptimizerOptions parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.OptimizerOptions parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.OptimizerOptions parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.OptimizerOptions parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.OptimizerOptions parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.OptimizerOptions parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.OptimizerOptions parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.OptimizerOptions parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.OptimizerOptions parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.OptimizerOptions parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.framework.OptimizerOptions prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * <pre>
+   * Options passed to the graph optimizer
+   * </pre>
+   *
+   * Protobuf type {@code tensorflow.OptimizerOptions}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.OptimizerOptions)
+      org.tensorflow.proto.framework.OptimizerOptionsOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_OptimizerOptions_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_OptimizerOptions_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.OptimizerOptions.class, org.tensorflow.proto.framework.OptimizerOptions.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.framework.OptimizerOptions.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      doCommonSubexpressionElimination_ = false;
+
+      doConstantFolding_ = false;
+
+      maxFoldedConstantInBytes_ = 0L;
+
+      doFunctionInlining_ = false;
+
+      optLevel_ = 0;
+
+      globalJitLevel_ = 0;
+
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_OptimizerOptions_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.OptimizerOptions getDefaultInstanceForType() {
+      return org.tensorflow.proto.framework.OptimizerOptions.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.OptimizerOptions build() {
+      org.tensorflow.proto.framework.OptimizerOptions result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.OptimizerOptions buildPartial() {
+      org.tensorflow.proto.framework.OptimizerOptions result = new org.tensorflow.proto.framework.OptimizerOptions(this);
+      result.doCommonSubexpressionElimination_ = doCommonSubexpressionElimination_;
+      result.doConstantFolding_ = doConstantFolding_;
+      result.maxFoldedConstantInBytes_ = maxFoldedConstantInBytes_;
+      result.doFunctionInlining_ = doFunctionInlining_;
+      result.optLevel_ = optLevel_;
+      result.globalJitLevel_ = globalJitLevel_;
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.framework.OptimizerOptions) {
+        return mergeFrom((org.tensorflow.proto.framework.OptimizerOptions)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.framework.OptimizerOptions other) {
+      if (other == org.tensorflow.proto.framework.OptimizerOptions.getDefaultInstance()) return this;
+      if (other.getDoCommonSubexpressionElimination() != false) {
+        setDoCommonSubexpressionElimination(other.getDoCommonSubexpressionElimination());
+      }
+      if (other.getDoConstantFolding() != false) {
+        setDoConstantFolding(other.getDoConstantFolding());
+      }
+      if (other.getMaxFoldedConstantInBytes() != 0L) {
+        setMaxFoldedConstantInBytes(other.getMaxFoldedConstantInBytes());
+      }
+      if (other.getDoFunctionInlining() != false) {
+        setDoFunctionInlining(other.getDoFunctionInlining());
+      }
+      if (other.optLevel_ != 0) {
+        setOptLevelValue(other.getOptLevelValue());
+      }
+      if (other.globalJitLevel_ != 0) {
+        setGlobalJitLevelValue(other.getGlobalJitLevelValue());
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.framework.OptimizerOptions parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.framework.OptimizerOptions) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+
+    private boolean doCommonSubexpressionElimination_ ;
+    /**
+     * <pre>
+     * If true, optimize the graph using common subexpression elimination.
+     * </pre>
+     *
+     * <code>bool do_common_subexpression_elimination = 1;</code>
+     */
+    public boolean getDoCommonSubexpressionElimination() {
+      return doCommonSubexpressionElimination_;
+    }
+    /**
+     * <pre>
+     * If true, optimize the graph using common subexpression elimination.
+     * </pre>
+     *
+     * <code>bool do_common_subexpression_elimination = 1;</code>
+     */
+    public Builder setDoCommonSubexpressionElimination(boolean value) {
+      
+      doCommonSubexpressionElimination_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * If true, optimize the graph using common subexpression elimination.
+     * </pre>
+     *
+     * <code>bool do_common_subexpression_elimination = 1;</code>
+     */
+    public Builder clearDoCommonSubexpressionElimination() {
+      
+      doCommonSubexpressionElimination_ = false;
+      onChanged();
+      return this;
+    }
+
+    private boolean doConstantFolding_ ;
+    /**
+     * <pre>
+     * If true, perform constant folding optimization on the graph.
+     * </pre>
+     *
+     * <code>bool do_constant_folding = 2;</code>
+     */
+    public boolean getDoConstantFolding() {
+      return doConstantFolding_;
+    }
+    /**
+     * <pre>
+     * If true, perform constant folding optimization on the graph.
+     * </pre>
+     *
+     * <code>bool do_constant_folding = 2;</code>
+     */
+    public Builder setDoConstantFolding(boolean value) {
+      
+      doConstantFolding_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * If true, perform constant folding optimization on the graph.
+     * </pre>
+     *
+     * <code>bool do_constant_folding = 2;</code>
+     */
+    public Builder clearDoConstantFolding() {
+      
+      doConstantFolding_ = false;
+      onChanged();
+      return this;
+    }
+
+    private long maxFoldedConstantInBytes_ ;
+    /**
+     * <pre>
+     * Constant folding optimization replaces tensors whose values can be
+     * predetermined, with constant nodes. To avoid inserting too large constants,
+     * the size of each constant created can be limited. If this value is zero, a
+     * default limit of 10 MiB will be applied. If constant folding optimization
+     * is disabled, this value is ignored.
+     * </pre>
+     *
+     * <code>int64 max_folded_constant_in_bytes = 6;</code>
+     */
+    public long getMaxFoldedConstantInBytes() {
+      return maxFoldedConstantInBytes_;
+    }
+    /**
+     * <pre>
+     * Constant folding optimization replaces tensors whose values can be
+     * predetermined, with constant nodes. To avoid inserting too large constants,
+     * the size of each constant created can be limited. If this value is zero, a
+     * default limit of 10 MiB will be applied. If constant folding optimization
+     * is disabled, this value is ignored.
+     * </pre>
+     *
+     * <code>int64 max_folded_constant_in_bytes = 6;</code>
+     */
+    public Builder setMaxFoldedConstantInBytes(long value) {
+      
+      maxFoldedConstantInBytes_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Constant folding optimization replaces tensors whose values can be
+     * predetermined, with constant nodes. To avoid inserting too large constants,
+     * the size of each constant created can be limited. If this value is zero, a
+     * default limit of 10 MiB will be applied. If constant folding optimization
+     * is disabled, this value is ignored.
+     * </pre>
+     *
+     * <code>int64 max_folded_constant_in_bytes = 6;</code>
+     */
+    public Builder clearMaxFoldedConstantInBytes() {
+      
+      maxFoldedConstantInBytes_ = 0L;
+      onChanged();
+      return this;
+    }
+
+    private boolean doFunctionInlining_ ;
+    /**
+     * <pre>
+     * If true, perform function inlining on the graph.
+     * </pre>
+     *
+     * <code>bool do_function_inlining = 4;</code>
+     */
+    public boolean getDoFunctionInlining() {
+      return doFunctionInlining_;
+    }
+    /**
+     * <pre>
+     * If true, perform function inlining on the graph.
+     * </pre>
+     *
+     * <code>bool do_function_inlining = 4;</code>
+     */
+    public Builder setDoFunctionInlining(boolean value) {
+      
+      doFunctionInlining_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * If true, perform function inlining on the graph.
+     * </pre>
+     *
+     * <code>bool do_function_inlining = 4;</code>
+     */
+    public Builder clearDoFunctionInlining() {
+      
+      doFunctionInlining_ = false;
+      onChanged();
+      return this;
+    }
+
+    private int optLevel_ = 0;
+    /**
+     * <pre>
+     * Overall optimization level. The actual optimizations applied will be the
+     * logical OR of the flags that this level implies and any flags already set.
+     * </pre>
+     *
+     * <code>.tensorflow.OptimizerOptions.Level opt_level = 3;</code>
+     */
+    public int getOptLevelValue() {
+      return optLevel_;
+    }
+    /**
+     * <pre>
+     * Overall optimization level. The actual optimizations applied will be the
+     * logical OR of the flags that this level implies and any flags already set.
+     * </pre>
+     *
+     * <code>.tensorflow.OptimizerOptions.Level opt_level = 3;</code>
+     */
+    public Builder setOptLevelValue(int value) {
+      optLevel_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Overall optimization level. The actual optimizations applied will be the
+     * logical OR of the flags that this level implies and any flags already set.
+     * </pre>
+     *
+     * <code>.tensorflow.OptimizerOptions.Level opt_level = 3;</code>
+     */
+    public org.tensorflow.proto.framework.OptimizerOptions.Level getOptLevel() {
+      @SuppressWarnings("deprecation")
+      org.tensorflow.proto.framework.OptimizerOptions.Level result = org.tensorflow.proto.framework.OptimizerOptions.Level.valueOf(optLevel_);
+      return result == null ? org.tensorflow.proto.framework.OptimizerOptions.Level.UNRECOGNIZED : result;
+    }
+    /**
+     * <pre>
+     * Overall optimization level. The actual optimizations applied will be the
+     * logical OR of the flags that this level implies and any flags already set.
+     * </pre>
+     *
+     * <code>.tensorflow.OptimizerOptions.Level opt_level = 3;</code>
+     */
+    public Builder setOptLevel(org.tensorflow.proto.framework.OptimizerOptions.Level value) {
+      if (value == null) {
+        throw new NullPointerException();
+      }
+      
+      optLevel_ = value.getNumber();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Overall optimization level. The actual optimizations applied will be the
+     * logical OR of the flags that this level implies and any flags already set.
+     * </pre>
+     *
+     * <code>.tensorflow.OptimizerOptions.Level opt_level = 3;</code>
+     */
+    public Builder clearOptLevel() {
+      
+      optLevel_ = 0;
+      onChanged();
+      return this;
+    }
+
+    private int globalJitLevel_ = 0;
+    /**
+     * <code>.tensorflow.OptimizerOptions.GlobalJitLevel global_jit_level = 5;</code>
+     */
+    public int getGlobalJitLevelValue() {
+      return globalJitLevel_;
+    }
+    /**
+     * <code>.tensorflow.OptimizerOptions.GlobalJitLevel global_jit_level = 5;</code>
+     */
+    public Builder setGlobalJitLevelValue(int value) {
+      globalJitLevel_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>.tensorflow.OptimizerOptions.GlobalJitLevel global_jit_level = 5;</code>
+     */
+    public org.tensorflow.proto.framework.OptimizerOptions.GlobalJitLevel getGlobalJitLevel() {
+      @SuppressWarnings("deprecation")
+      org.tensorflow.proto.framework.OptimizerOptions.GlobalJitLevel result = org.tensorflow.proto.framework.OptimizerOptions.GlobalJitLevel.valueOf(globalJitLevel_);
+      return result == null ? org.tensorflow.proto.framework.OptimizerOptions.GlobalJitLevel.UNRECOGNIZED : result;
+    }
+    /**
+     * <code>.tensorflow.OptimizerOptions.GlobalJitLevel global_jit_level = 5;</code>
+     */
+    public Builder setGlobalJitLevel(org.tensorflow.proto.framework.OptimizerOptions.GlobalJitLevel value) {
+      if (value == null) {
+        throw new NullPointerException();
+      }
+      
+      globalJitLevel_ = value.getNumber();
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>.tensorflow.OptimizerOptions.GlobalJitLevel global_jit_level = 5;</code>
+     */
+    public Builder clearGlobalJitLevel() {
+      
+      globalJitLevel_ = 0;
+      onChanged();
+      return this;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.OptimizerOptions)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.OptimizerOptions)
+  private static final org.tensorflow.proto.framework.OptimizerOptions DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.OptimizerOptions();
+  }
+
+  public static org.tensorflow.proto.framework.OptimizerOptions getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<OptimizerOptions>
+      PARSER = new com.google.protobuf.AbstractParser<OptimizerOptions>() {
+    @java.lang.Override
+    public OptimizerOptions parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new OptimizerOptions(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<OptimizerOptions> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<OptimizerOptions> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.framework.OptimizerOptions getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/OptimizerOptionsOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/OptimizerOptionsOrBuilder.java
new file mode 100644
index 00000000000..255ace49e05
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/OptimizerOptionsOrBuilder.java
@@ -0,0 +1,77 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/config.proto
+
+package org.tensorflow.proto.framework;
+
+public interface OptimizerOptionsOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.OptimizerOptions)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <pre>
+   * If true, optimize the graph using common subexpression elimination.
+   * </pre>
+   *
+   * <code>bool do_common_subexpression_elimination = 1;</code>
+   */
+  boolean getDoCommonSubexpressionElimination();
+
+  /**
+   * <pre>
+   * If true, perform constant folding optimization on the graph.
+   * </pre>
+   *
+   * <code>bool do_constant_folding = 2;</code>
+   */
+  boolean getDoConstantFolding();
+
+  /**
+   * <pre>
+   * Constant folding optimization replaces tensors whose values can be
+   * predetermined, with constant nodes. To avoid inserting too large constants,
+   * the size of each constant created can be limited. If this value is zero, a
+   * default limit of 10 MiB will be applied. If constant folding optimization
+   * is disabled, this value is ignored.
+   * </pre>
+   *
+   * <code>int64 max_folded_constant_in_bytes = 6;</code>
+   */
+  long getMaxFoldedConstantInBytes();
+
+  /**
+   * <pre>
+   * If true, perform function inlining on the graph.
+   * </pre>
+   *
+   * <code>bool do_function_inlining = 4;</code>
+   */
+  boolean getDoFunctionInlining();
+
+  /**
+   * <pre>
+   * Overall optimization level. The actual optimizations applied will be the
+   * logical OR of the flags that this level implies and any flags already set.
+   * </pre>
+   *
+   * <code>.tensorflow.OptimizerOptions.Level opt_level = 3;</code>
+   */
+  int getOptLevelValue();
+  /**
+   * <pre>
+   * Overall optimization level. The actual optimizations applied will be the
+   * logical OR of the flags that this level implies and any flags already set.
+   * </pre>
+   *
+   * <code>.tensorflow.OptimizerOptions.Level opt_level = 3;</code>
+   */
+  org.tensorflow.proto.framework.OptimizerOptions.Level getOptLevel();
+
+  /**
+   * <code>.tensorflow.OptimizerOptions.GlobalJitLevel global_jit_level = 5;</code>
+   */
+  int getGlobalJitLevelValue();
+  /**
+   * <code>.tensorflow.OptimizerOptions.GlobalJitLevel global_jit_level = 5;</code>
+   */
+  org.tensorflow.proto.framework.OptimizerOptions.GlobalJitLevel getGlobalJitLevel();
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/PairValue.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/PairValue.java
new file mode 100644
index 00000000000..f9972971b17
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/PairValue.java
@@ -0,0 +1,735 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/struct.proto
+
+package org.tensorflow.proto.framework;
+
+/**
+ * <pre>
+ * Represents a (key, value) pair.
+ * </pre>
+ *
+ * Protobuf type {@code tensorflow.PairValue}
+ */
+public  final class PairValue extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.PairValue)
+    PairValueOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use PairValue.newBuilder() to construct.
+  private PairValue(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private PairValue() {
+    key_ = "";
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new PairValue();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private PairValue(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 10: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            key_ = s;
+            break;
+          }
+          case 18: {
+            org.tensorflow.proto.framework.StructuredValue.Builder subBuilder = null;
+            if (value_ != null) {
+              subBuilder = value_.toBuilder();
+            }
+            value_ = input.readMessage(org.tensorflow.proto.framework.StructuredValue.parser(), extensionRegistry);
+            if (subBuilder != null) {
+              subBuilder.mergeFrom(value_);
+              value_ = subBuilder.buildPartial();
+            }
+
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.framework.StructProtos.internal_static_tensorflow_PairValue_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.framework.StructProtos.internal_static_tensorflow_PairValue_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.framework.PairValue.class, org.tensorflow.proto.framework.PairValue.Builder.class);
+  }
+
+  public static final int KEY_FIELD_NUMBER = 1;
+  private volatile java.lang.Object key_;
+  /**
+   * <code>string key = 1;</code>
+   */
+  public java.lang.String getKey() {
+    java.lang.Object ref = key_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      key_ = s;
+      return s;
+    }
+  }
+  /**
+   * <code>string key = 1;</code>
+   */
+  public com.google.protobuf.ByteString
+      getKeyBytes() {
+    java.lang.Object ref = key_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      key_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int VALUE_FIELD_NUMBER = 2;
+  private org.tensorflow.proto.framework.StructuredValue value_;
+  /**
+   * <code>.tensorflow.StructuredValue value = 2;</code>
+   */
+  public boolean hasValue() {
+    return value_ != null;
+  }
+  /**
+   * <code>.tensorflow.StructuredValue value = 2;</code>
+   */
+  public org.tensorflow.proto.framework.StructuredValue getValue() {
+    return value_ == null ? org.tensorflow.proto.framework.StructuredValue.getDefaultInstance() : value_;
+  }
+  /**
+   * <code>.tensorflow.StructuredValue value = 2;</code>
+   */
+  public org.tensorflow.proto.framework.StructuredValueOrBuilder getValueOrBuilder() {
+    return getValue();
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    if (!getKeyBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 1, key_);
+    }
+    if (value_ != null) {
+      output.writeMessage(2, getValue());
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (!getKeyBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, key_);
+    }
+    if (value_ != null) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(2, getValue());
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.framework.PairValue)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.framework.PairValue other = (org.tensorflow.proto.framework.PairValue) obj;
+
+    if (!getKey()
+        .equals(other.getKey())) return false;
+    if (hasValue() != other.hasValue()) return false;
+    if (hasValue()) {
+      if (!getValue()
+          .equals(other.getValue())) return false;
+    }
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    hash = (37 * hash) + KEY_FIELD_NUMBER;
+    hash = (53 * hash) + getKey().hashCode();
+    if (hasValue()) {
+      hash = (37 * hash) + VALUE_FIELD_NUMBER;
+      hash = (53 * hash) + getValue().hashCode();
+    }
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.framework.PairValue parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.PairValue parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.PairValue parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.PairValue parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.PairValue parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.PairValue parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.PairValue parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.PairValue parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.PairValue parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.PairValue parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.PairValue parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.PairValue parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.framework.PairValue prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * <pre>
+   * Represents a (key, value) pair.
+   * </pre>
+   *
+   * Protobuf type {@code tensorflow.PairValue}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.PairValue)
+      org.tensorflow.proto.framework.PairValueOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.StructProtos.internal_static_tensorflow_PairValue_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.StructProtos.internal_static_tensorflow_PairValue_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.PairValue.class, org.tensorflow.proto.framework.PairValue.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.framework.PairValue.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      key_ = "";
+
+      if (valueBuilder_ == null) {
+        value_ = null;
+      } else {
+        value_ = null;
+        valueBuilder_ = null;
+      }
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.framework.StructProtos.internal_static_tensorflow_PairValue_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.PairValue getDefaultInstanceForType() {
+      return org.tensorflow.proto.framework.PairValue.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.PairValue build() {
+      org.tensorflow.proto.framework.PairValue result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.PairValue buildPartial() {
+      org.tensorflow.proto.framework.PairValue result = new org.tensorflow.proto.framework.PairValue(this);
+      result.key_ = key_;
+      if (valueBuilder_ == null) {
+        result.value_ = value_;
+      } else {
+        result.value_ = valueBuilder_.build();
+      }
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.framework.PairValue) {
+        return mergeFrom((org.tensorflow.proto.framework.PairValue)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.framework.PairValue other) {
+      if (other == org.tensorflow.proto.framework.PairValue.getDefaultInstance()) return this;
+      if (!other.getKey().isEmpty()) {
+        key_ = other.key_;
+        onChanged();
+      }
+      if (other.hasValue()) {
+        mergeValue(other.getValue());
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.framework.PairValue parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.framework.PairValue) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+
+    private java.lang.Object key_ = "";
+    /**
+     * <code>string key = 1;</code>
+     */
+    public java.lang.String getKey() {
+      java.lang.Object ref = key_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        key_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <code>string key = 1;</code>
+     */
+    public com.google.protobuf.ByteString
+        getKeyBytes() {
+      java.lang.Object ref = key_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        key_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <code>string key = 1;</code>
+     */
+    public Builder setKey(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      key_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>string key = 1;</code>
+     */
+    public Builder clearKey() {
+      
+      key_ = getDefaultInstance().getKey();
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>string key = 1;</code>
+     */
+    public Builder setKeyBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      key_ = value;
+      onChanged();
+      return this;
+    }
+
+    private org.tensorflow.proto.framework.StructuredValue value_;
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.StructuredValue, org.tensorflow.proto.framework.StructuredValue.Builder, org.tensorflow.proto.framework.StructuredValueOrBuilder> valueBuilder_;
+    /**
+     * <code>.tensorflow.StructuredValue value = 2;</code>
+     */
+    public boolean hasValue() {
+      return valueBuilder_ != null || value_ != null;
+    }
+    /**
+     * <code>.tensorflow.StructuredValue value = 2;</code>
+     */
+    public org.tensorflow.proto.framework.StructuredValue getValue() {
+      if (valueBuilder_ == null) {
+        return value_ == null ? org.tensorflow.proto.framework.StructuredValue.getDefaultInstance() : value_;
+      } else {
+        return valueBuilder_.getMessage();
+      }
+    }
+    /**
+     * <code>.tensorflow.StructuredValue value = 2;</code>
+     */
+    public Builder setValue(org.tensorflow.proto.framework.StructuredValue value) {
+      if (valueBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        value_ = value;
+        onChanged();
+      } else {
+        valueBuilder_.setMessage(value);
+      }
+
+      return this;
+    }
+    /**
+     * <code>.tensorflow.StructuredValue value = 2;</code>
+     */
+    public Builder setValue(
+        org.tensorflow.proto.framework.StructuredValue.Builder builderForValue) {
+      if (valueBuilder_ == null) {
+        value_ = builderForValue.build();
+        onChanged();
+      } else {
+        valueBuilder_.setMessage(builderForValue.build());
+      }
+
+      return this;
+    }
+    /**
+     * <code>.tensorflow.StructuredValue value = 2;</code>
+     */
+    public Builder mergeValue(org.tensorflow.proto.framework.StructuredValue value) {
+      if (valueBuilder_ == null) {
+        if (value_ != null) {
+          value_ =
+            org.tensorflow.proto.framework.StructuredValue.newBuilder(value_).mergeFrom(value).buildPartial();
+        } else {
+          value_ = value;
+        }
+        onChanged();
+      } else {
+        valueBuilder_.mergeFrom(value);
+      }
+
+      return this;
+    }
+    /**
+     * <code>.tensorflow.StructuredValue value = 2;</code>
+     */
+    public Builder clearValue() {
+      if (valueBuilder_ == null) {
+        value_ = null;
+        onChanged();
+      } else {
+        value_ = null;
+        valueBuilder_ = null;
+      }
+
+      return this;
+    }
+    /**
+     * <code>.tensorflow.StructuredValue value = 2;</code>
+     */
+    public org.tensorflow.proto.framework.StructuredValue.Builder getValueBuilder() {
+      
+      onChanged();
+      return getValueFieldBuilder().getBuilder();
+    }
+    /**
+     * <code>.tensorflow.StructuredValue value = 2;</code>
+     */
+    public org.tensorflow.proto.framework.StructuredValueOrBuilder getValueOrBuilder() {
+      if (valueBuilder_ != null) {
+        return valueBuilder_.getMessageOrBuilder();
+      } else {
+        return value_ == null ?
+            org.tensorflow.proto.framework.StructuredValue.getDefaultInstance() : value_;
+      }
+    }
+    /**
+     * <code>.tensorflow.StructuredValue value = 2;</code>
+     */
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.StructuredValue, org.tensorflow.proto.framework.StructuredValue.Builder, org.tensorflow.proto.framework.StructuredValueOrBuilder> 
+        getValueFieldBuilder() {
+      if (valueBuilder_ == null) {
+        valueBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+            org.tensorflow.proto.framework.StructuredValue, org.tensorflow.proto.framework.StructuredValue.Builder, org.tensorflow.proto.framework.StructuredValueOrBuilder>(
+                getValue(),
+                getParentForChildren(),
+                isClean());
+        value_ = null;
+      }
+      return valueBuilder_;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.PairValue)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.PairValue)
+  private static final org.tensorflow.proto.framework.PairValue DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.PairValue();
+  }
+
+  public static org.tensorflow.proto.framework.PairValue getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<PairValue>
+      PARSER = new com.google.protobuf.AbstractParser<PairValue>() {
+    @java.lang.Override
+    public PairValue parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new PairValue(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<PairValue> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<PairValue> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.framework.PairValue getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/PairValueOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/PairValueOrBuilder.java
new file mode 100644
index 00000000000..0e35d82c1af
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/PairValueOrBuilder.java
@@ -0,0 +1,32 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/struct.proto
+
+package org.tensorflow.proto.framework;
+
+public interface PairValueOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.PairValue)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <code>string key = 1;</code>
+   */
+  java.lang.String getKey();
+  /**
+   * <code>string key = 1;</code>
+   */
+  com.google.protobuf.ByteString
+      getKeyBytes();
+
+  /**
+   * <code>.tensorflow.StructuredValue value = 2;</code>
+   */
+  boolean hasValue();
+  /**
+   * <code>.tensorflow.StructuredValue value = 2;</code>
+   */
+  org.tensorflow.proto.framework.StructuredValue getValue();
+  /**
+   * <code>.tensorflow.StructuredValue value = 2;</code>
+   */
+  org.tensorflow.proto.framework.StructuredValueOrBuilder getValueOrBuilder();
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/QueueRunnerDef.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/QueueRunnerDef.java
new file mode 100644
index 00000000000..8ca69c8f7f2
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/QueueRunnerDef.java
@@ -0,0 +1,1435 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/queue_runner.proto
+
+package org.tensorflow.proto.framework;
+
+/**
+ * <pre>
+ * Protocol buffer representing a QueueRunner.
+ * </pre>
+ *
+ * Protobuf type {@code tensorflow.QueueRunnerDef}
+ */
+public  final class QueueRunnerDef extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.QueueRunnerDef)
+    QueueRunnerDefOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use QueueRunnerDef.newBuilder() to construct.
+  private QueueRunnerDef(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private QueueRunnerDef() {
+    queueName_ = "";
+    enqueueOpName_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+    closeOpName_ = "";
+    cancelOpName_ = "";
+    queueClosedExceptionTypes_ = java.util.Collections.emptyList();
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new QueueRunnerDef();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private QueueRunnerDef(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    int mutable_bitField0_ = 0;
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 10: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            queueName_ = s;
+            break;
+          }
+          case 18: {
+            java.lang.String s = input.readStringRequireUtf8();
+            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
+              enqueueOpName_ = new com.google.protobuf.LazyStringArrayList();
+              mutable_bitField0_ |= 0x00000001;
+            }
+            enqueueOpName_.add(s);
+            break;
+          }
+          case 26: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            closeOpName_ = s;
+            break;
+          }
+          case 34: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            cancelOpName_ = s;
+            break;
+          }
+          case 40: {
+            int rawValue = input.readEnum();
+            if (!((mutable_bitField0_ & 0x00000002) != 0)) {
+              queueClosedExceptionTypes_ = new java.util.ArrayList<java.lang.Integer>();
+              mutable_bitField0_ |= 0x00000002;
+            }
+            queueClosedExceptionTypes_.add(rawValue);
+            break;
+          }
+          case 42: {
+            int length = input.readRawVarint32();
+            int oldLimit = input.pushLimit(length);
+            while(input.getBytesUntilLimit() > 0) {
+              int rawValue = input.readEnum();
+              if (!((mutable_bitField0_ & 0x00000002) != 0)) {
+                queueClosedExceptionTypes_ = new java.util.ArrayList<java.lang.Integer>();
+                mutable_bitField0_ |= 0x00000002;
+              }
+              queueClosedExceptionTypes_.add(rawValue);
+            }
+            input.popLimit(oldLimit);
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      if (((mutable_bitField0_ & 0x00000001) != 0)) {
+        enqueueOpName_ = enqueueOpName_.getUnmodifiableView();
+      }
+      if (((mutable_bitField0_ & 0x00000002) != 0)) {
+        queueClosedExceptionTypes_ = java.util.Collections.unmodifiableList(queueClosedExceptionTypes_);
+      }
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.framework.QueueRunnerProtos.internal_static_tensorflow_QueueRunnerDef_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.framework.QueueRunnerProtos.internal_static_tensorflow_QueueRunnerDef_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.framework.QueueRunnerDef.class, org.tensorflow.proto.framework.QueueRunnerDef.Builder.class);
+  }
+
+  public static final int QUEUE_NAME_FIELD_NUMBER = 1;
+  private volatile java.lang.Object queueName_;
+  /**
+   * <pre>
+   * Queue name.
+   * </pre>
+   *
+   * <code>string queue_name = 1;</code>
+   */
+  public java.lang.String getQueueName() {
+    java.lang.Object ref = queueName_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      queueName_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * Queue name.
+   * </pre>
+   *
+   * <code>string queue_name = 1;</code>
+   */
+  public com.google.protobuf.ByteString
+      getQueueNameBytes() {
+    java.lang.Object ref = queueName_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      queueName_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int ENQUEUE_OP_NAME_FIELD_NUMBER = 2;
+  private com.google.protobuf.LazyStringList enqueueOpName_;
+  /**
+   * <pre>
+   * A list of enqueue operations.
+   * </pre>
+   *
+   * <code>repeated string enqueue_op_name = 2;</code>
+   */
+  public com.google.protobuf.ProtocolStringList
+      getEnqueueOpNameList() {
+    return enqueueOpName_;
+  }
+  /**
+   * <pre>
+   * A list of enqueue operations.
+   * </pre>
+   *
+   * <code>repeated string enqueue_op_name = 2;</code>
+   */
+  public int getEnqueueOpNameCount() {
+    return enqueueOpName_.size();
+  }
+  /**
+   * <pre>
+   * A list of enqueue operations.
+   * </pre>
+   *
+   * <code>repeated string enqueue_op_name = 2;</code>
+   */
+  public java.lang.String getEnqueueOpName(int index) {
+    return enqueueOpName_.get(index);
+  }
+  /**
+   * <pre>
+   * A list of enqueue operations.
+   * </pre>
+   *
+   * <code>repeated string enqueue_op_name = 2;</code>
+   */
+  public com.google.protobuf.ByteString
+      getEnqueueOpNameBytes(int index) {
+    return enqueueOpName_.getByteString(index);
+  }
+
+  public static final int CLOSE_OP_NAME_FIELD_NUMBER = 3;
+  private volatile java.lang.Object closeOpName_;
+  /**
+   * <pre>
+   * The operation to run to close the queue.
+   * </pre>
+   *
+   * <code>string close_op_name = 3;</code>
+   */
+  public java.lang.String getCloseOpName() {
+    java.lang.Object ref = closeOpName_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      closeOpName_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * The operation to run to close the queue.
+   * </pre>
+   *
+   * <code>string close_op_name = 3;</code>
+   */
+  public com.google.protobuf.ByteString
+      getCloseOpNameBytes() {
+    java.lang.Object ref = closeOpName_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      closeOpName_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int CANCEL_OP_NAME_FIELD_NUMBER = 4;
+  private volatile java.lang.Object cancelOpName_;
+  /**
+   * <pre>
+   * The operation to run to cancel the queue.
+   * </pre>
+   *
+   * <code>string cancel_op_name = 4;</code>
+   */
+  public java.lang.String getCancelOpName() {
+    java.lang.Object ref = cancelOpName_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      cancelOpName_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * The operation to run to cancel the queue.
+   * </pre>
+   *
+   * <code>string cancel_op_name = 4;</code>
+   */
+  public com.google.protobuf.ByteString
+      getCancelOpNameBytes() {
+    java.lang.Object ref = cancelOpName_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      cancelOpName_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int QUEUE_CLOSED_EXCEPTION_TYPES_FIELD_NUMBER = 5;
+  private java.util.List<java.lang.Integer> queueClosedExceptionTypes_;
+  private static final com.google.protobuf.Internal.ListAdapter.Converter<
+      java.lang.Integer, org.tensorflow.proto.framework.Code> queueClosedExceptionTypes_converter_ =
+          new com.google.protobuf.Internal.ListAdapter.Converter<
+              java.lang.Integer, org.tensorflow.proto.framework.Code>() {
+            public org.tensorflow.proto.framework.Code convert(java.lang.Integer from) {
+              @SuppressWarnings("deprecation")
+              org.tensorflow.proto.framework.Code result = org.tensorflow.proto.framework.Code.valueOf(from);
+              return result == null ? org.tensorflow.proto.framework.Code.UNRECOGNIZED : result;
+            }
+          };
+  /**
+   * <pre>
+   * A list of exception types considered to signal a safely closed queue
+   * if raised during enqueue operations.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.error.Code queue_closed_exception_types = 5;</code>
+   */
+  public java.util.List<org.tensorflow.proto.framework.Code> getQueueClosedExceptionTypesList() {
+    return new com.google.protobuf.Internal.ListAdapter<
+        java.lang.Integer, org.tensorflow.proto.framework.Code>(queueClosedExceptionTypes_, queueClosedExceptionTypes_converter_);
+  }
+  /**
+   * <pre>
+   * A list of exception types considered to signal a safely closed queue
+   * if raised during enqueue operations.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.error.Code queue_closed_exception_types = 5;</code>
+   */
+  public int getQueueClosedExceptionTypesCount() {
+    return queueClosedExceptionTypes_.size();
+  }
+  /**
+   * <pre>
+   * A list of exception types considered to signal a safely closed queue
+   * if raised during enqueue operations.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.error.Code queue_closed_exception_types = 5;</code>
+   */
+  public org.tensorflow.proto.framework.Code getQueueClosedExceptionTypes(int index) {
+    return queueClosedExceptionTypes_converter_.convert(queueClosedExceptionTypes_.get(index));
+  }
+  /**
+   * <pre>
+   * A list of exception types considered to signal a safely closed queue
+   * if raised during enqueue operations.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.error.Code queue_closed_exception_types = 5;</code>
+   */
+  public java.util.List<java.lang.Integer>
+  getQueueClosedExceptionTypesValueList() {
+    return queueClosedExceptionTypes_;
+  }
+  /**
+   * <pre>
+   * A list of exception types considered to signal a safely closed queue
+   * if raised during enqueue operations.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.error.Code queue_closed_exception_types = 5;</code>
+   */
+  public int getQueueClosedExceptionTypesValue(int index) {
+    return queueClosedExceptionTypes_.get(index);
+  }
+  private int queueClosedExceptionTypesMemoizedSerializedSize;
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    getSerializedSize();
+    if (!getQueueNameBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 1, queueName_);
+    }
+    for (int i = 0; i < enqueueOpName_.size(); i++) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 2, enqueueOpName_.getRaw(i));
+    }
+    if (!getCloseOpNameBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 3, closeOpName_);
+    }
+    if (!getCancelOpNameBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 4, cancelOpName_);
+    }
+    if (getQueueClosedExceptionTypesList().size() > 0) {
+      output.writeUInt32NoTag(42);
+      output.writeUInt32NoTag(queueClosedExceptionTypesMemoizedSerializedSize);
+    }
+    for (int i = 0; i < queueClosedExceptionTypes_.size(); i++) {
+      output.writeEnumNoTag(queueClosedExceptionTypes_.get(i));
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (!getQueueNameBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, queueName_);
+    }
+    {
+      int dataSize = 0;
+      for (int i = 0; i < enqueueOpName_.size(); i++) {
+        dataSize += computeStringSizeNoTag(enqueueOpName_.getRaw(i));
+      }
+      size += dataSize;
+      size += 1 * getEnqueueOpNameList().size();
+    }
+    if (!getCloseOpNameBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(3, closeOpName_);
+    }
+    if (!getCancelOpNameBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(4, cancelOpName_);
+    }
+    {
+      int dataSize = 0;
+      for (int i = 0; i < queueClosedExceptionTypes_.size(); i++) {
+        dataSize += com.google.protobuf.CodedOutputStream
+          .computeEnumSizeNoTag(queueClosedExceptionTypes_.get(i));
+      }
+      size += dataSize;
+      if (!getQueueClosedExceptionTypesList().isEmpty()) {  size += 1;
+        size += com.google.protobuf.CodedOutputStream
+          .computeUInt32SizeNoTag(dataSize);
+      }queueClosedExceptionTypesMemoizedSerializedSize = dataSize;
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.framework.QueueRunnerDef)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.framework.QueueRunnerDef other = (org.tensorflow.proto.framework.QueueRunnerDef) obj;
+
+    if (!getQueueName()
+        .equals(other.getQueueName())) return false;
+    if (!getEnqueueOpNameList()
+        .equals(other.getEnqueueOpNameList())) return false;
+    if (!getCloseOpName()
+        .equals(other.getCloseOpName())) return false;
+    if (!getCancelOpName()
+        .equals(other.getCancelOpName())) return false;
+    if (!queueClosedExceptionTypes_.equals(other.queueClosedExceptionTypes_)) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    hash = (37 * hash) + QUEUE_NAME_FIELD_NUMBER;
+    hash = (53 * hash) + getQueueName().hashCode();
+    if (getEnqueueOpNameCount() > 0) {
+      hash = (37 * hash) + ENQUEUE_OP_NAME_FIELD_NUMBER;
+      hash = (53 * hash) + getEnqueueOpNameList().hashCode();
+    }
+    hash = (37 * hash) + CLOSE_OP_NAME_FIELD_NUMBER;
+    hash = (53 * hash) + getCloseOpName().hashCode();
+    hash = (37 * hash) + CANCEL_OP_NAME_FIELD_NUMBER;
+    hash = (53 * hash) + getCancelOpName().hashCode();
+    if (getQueueClosedExceptionTypesCount() > 0) {
+      hash = (37 * hash) + QUEUE_CLOSED_EXCEPTION_TYPES_FIELD_NUMBER;
+      hash = (53 * hash) + queueClosedExceptionTypes_.hashCode();
+    }
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.framework.QueueRunnerDef parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.QueueRunnerDef parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.QueueRunnerDef parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.QueueRunnerDef parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.QueueRunnerDef parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.QueueRunnerDef parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.QueueRunnerDef parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.QueueRunnerDef parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.QueueRunnerDef parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.QueueRunnerDef parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.QueueRunnerDef parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.QueueRunnerDef parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.framework.QueueRunnerDef prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * <pre>
+   * Protocol buffer representing a QueueRunner.
+   * </pre>
+   *
+   * Protobuf type {@code tensorflow.QueueRunnerDef}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.QueueRunnerDef)
+      org.tensorflow.proto.framework.QueueRunnerDefOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.QueueRunnerProtos.internal_static_tensorflow_QueueRunnerDef_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.QueueRunnerProtos.internal_static_tensorflow_QueueRunnerDef_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.QueueRunnerDef.class, org.tensorflow.proto.framework.QueueRunnerDef.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.framework.QueueRunnerDef.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      queueName_ = "";
+
+      enqueueOpName_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+      bitField0_ = (bitField0_ & ~0x00000001);
+      closeOpName_ = "";
+
+      cancelOpName_ = "";
+
+      queueClosedExceptionTypes_ = java.util.Collections.emptyList();
+      bitField0_ = (bitField0_ & ~0x00000002);
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.framework.QueueRunnerProtos.internal_static_tensorflow_QueueRunnerDef_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.QueueRunnerDef getDefaultInstanceForType() {
+      return org.tensorflow.proto.framework.QueueRunnerDef.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.QueueRunnerDef build() {
+      org.tensorflow.proto.framework.QueueRunnerDef result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.QueueRunnerDef buildPartial() {
+      org.tensorflow.proto.framework.QueueRunnerDef result = new org.tensorflow.proto.framework.QueueRunnerDef(this);
+      int from_bitField0_ = bitField0_;
+      result.queueName_ = queueName_;
+      if (((bitField0_ & 0x00000001) != 0)) {
+        enqueueOpName_ = enqueueOpName_.getUnmodifiableView();
+        bitField0_ = (bitField0_ & ~0x00000001);
+      }
+      result.enqueueOpName_ = enqueueOpName_;
+      result.closeOpName_ = closeOpName_;
+      result.cancelOpName_ = cancelOpName_;
+      if (((bitField0_ & 0x00000002) != 0)) {
+        queueClosedExceptionTypes_ = java.util.Collections.unmodifiableList(queueClosedExceptionTypes_);
+        bitField0_ = (bitField0_ & ~0x00000002);
+      }
+      result.queueClosedExceptionTypes_ = queueClosedExceptionTypes_;
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.framework.QueueRunnerDef) {
+        return mergeFrom((org.tensorflow.proto.framework.QueueRunnerDef)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.framework.QueueRunnerDef other) {
+      if (other == org.tensorflow.proto.framework.QueueRunnerDef.getDefaultInstance()) return this;
+      if (!other.getQueueName().isEmpty()) {
+        queueName_ = other.queueName_;
+        onChanged();
+      }
+      if (!other.enqueueOpName_.isEmpty()) {
+        if (enqueueOpName_.isEmpty()) {
+          enqueueOpName_ = other.enqueueOpName_;
+          bitField0_ = (bitField0_ & ~0x00000001);
+        } else {
+          ensureEnqueueOpNameIsMutable();
+          enqueueOpName_.addAll(other.enqueueOpName_);
+        }
+        onChanged();
+      }
+      if (!other.getCloseOpName().isEmpty()) {
+        closeOpName_ = other.closeOpName_;
+        onChanged();
+      }
+      if (!other.getCancelOpName().isEmpty()) {
+        cancelOpName_ = other.cancelOpName_;
+        onChanged();
+      }
+      if (!other.queueClosedExceptionTypes_.isEmpty()) {
+        if (queueClosedExceptionTypes_.isEmpty()) {
+          queueClosedExceptionTypes_ = other.queueClosedExceptionTypes_;
+          bitField0_ = (bitField0_ & ~0x00000002);
+        } else {
+          ensureQueueClosedExceptionTypesIsMutable();
+          queueClosedExceptionTypes_.addAll(other.queueClosedExceptionTypes_);
+        }
+        onChanged();
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.framework.QueueRunnerDef parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.framework.QueueRunnerDef) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+    private int bitField0_;
+
+    private java.lang.Object queueName_ = "";
+    /**
+     * <pre>
+     * Queue name.
+     * </pre>
+     *
+     * <code>string queue_name = 1;</code>
+     */
+    public java.lang.String getQueueName() {
+      java.lang.Object ref = queueName_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        queueName_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Queue name.
+     * </pre>
+     *
+     * <code>string queue_name = 1;</code>
+     */
+    public com.google.protobuf.ByteString
+        getQueueNameBytes() {
+      java.lang.Object ref = queueName_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        queueName_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Queue name.
+     * </pre>
+     *
+     * <code>string queue_name = 1;</code>
+     */
+    public Builder setQueueName(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      queueName_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Queue name.
+     * </pre>
+     *
+     * <code>string queue_name = 1;</code>
+     */
+    public Builder clearQueueName() {
+      
+      queueName_ = getDefaultInstance().getQueueName();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Queue name.
+     * </pre>
+     *
+     * <code>string queue_name = 1;</code>
+     */
+    public Builder setQueueNameBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      queueName_ = value;
+      onChanged();
+      return this;
+    }
+
+    private com.google.protobuf.LazyStringList enqueueOpName_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+    private void ensureEnqueueOpNameIsMutable() {
+      if (!((bitField0_ & 0x00000001) != 0)) {
+        enqueueOpName_ = new com.google.protobuf.LazyStringArrayList(enqueueOpName_);
+        bitField0_ |= 0x00000001;
+       }
+    }
+    /**
+     * <pre>
+     * A list of enqueue operations.
+     * </pre>
+     *
+     * <code>repeated string enqueue_op_name = 2;</code>
+     */
+    public com.google.protobuf.ProtocolStringList
+        getEnqueueOpNameList() {
+      return enqueueOpName_.getUnmodifiableView();
+    }
+    /**
+     * <pre>
+     * A list of enqueue operations.
+     * </pre>
+     *
+     * <code>repeated string enqueue_op_name = 2;</code>
+     */
+    public int getEnqueueOpNameCount() {
+      return enqueueOpName_.size();
+    }
+    /**
+     * <pre>
+     * A list of enqueue operations.
+     * </pre>
+     *
+     * <code>repeated string enqueue_op_name = 2;</code>
+     */
+    public java.lang.String getEnqueueOpName(int index) {
+      return enqueueOpName_.get(index);
+    }
+    /**
+     * <pre>
+     * A list of enqueue operations.
+     * </pre>
+     *
+     * <code>repeated string enqueue_op_name = 2;</code>
+     */
+    public com.google.protobuf.ByteString
+        getEnqueueOpNameBytes(int index) {
+      return enqueueOpName_.getByteString(index);
+    }
+    /**
+     * <pre>
+     * A list of enqueue operations.
+     * </pre>
+     *
+     * <code>repeated string enqueue_op_name = 2;</code>
+     */
+    public Builder setEnqueueOpName(
+        int index, java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  ensureEnqueueOpNameIsMutable();
+      enqueueOpName_.set(index, value);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * A list of enqueue operations.
+     * </pre>
+     *
+     * <code>repeated string enqueue_op_name = 2;</code>
+     */
+    public Builder addEnqueueOpName(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  ensureEnqueueOpNameIsMutable();
+      enqueueOpName_.add(value);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * A list of enqueue operations.
+     * </pre>
+     *
+     * <code>repeated string enqueue_op_name = 2;</code>
+     */
+    public Builder addAllEnqueueOpName(
+        java.lang.Iterable<java.lang.String> values) {
+      ensureEnqueueOpNameIsMutable();
+      com.google.protobuf.AbstractMessageLite.Builder.addAll(
+          values, enqueueOpName_);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * A list of enqueue operations.
+     * </pre>
+     *
+     * <code>repeated string enqueue_op_name = 2;</code>
+     */
+    public Builder clearEnqueueOpName() {
+      enqueueOpName_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+      bitField0_ = (bitField0_ & ~0x00000001);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * A list of enqueue operations.
+     * </pre>
+     *
+     * <code>repeated string enqueue_op_name = 2;</code>
+     */
+    public Builder addEnqueueOpNameBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      ensureEnqueueOpNameIsMutable();
+      enqueueOpName_.add(value);
+      onChanged();
+      return this;
+    }
+
+    private java.lang.Object closeOpName_ = "";
+    /**
+     * <pre>
+     * The operation to run to close the queue.
+     * </pre>
+     *
+     * <code>string close_op_name = 3;</code>
+     */
+    public java.lang.String getCloseOpName() {
+      java.lang.Object ref = closeOpName_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        closeOpName_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * The operation to run to close the queue.
+     * </pre>
+     *
+     * <code>string close_op_name = 3;</code>
+     */
+    public com.google.protobuf.ByteString
+        getCloseOpNameBytes() {
+      java.lang.Object ref = closeOpName_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        closeOpName_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * The operation to run to close the queue.
+     * </pre>
+     *
+     * <code>string close_op_name = 3;</code>
+     */
+    public Builder setCloseOpName(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      closeOpName_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * The operation to run to close the queue.
+     * </pre>
+     *
+     * <code>string close_op_name = 3;</code>
+     */
+    public Builder clearCloseOpName() {
+      
+      closeOpName_ = getDefaultInstance().getCloseOpName();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * The operation to run to close the queue.
+     * </pre>
+     *
+     * <code>string close_op_name = 3;</code>
+     */
+    public Builder setCloseOpNameBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      closeOpName_ = value;
+      onChanged();
+      return this;
+    }
+
+    private java.lang.Object cancelOpName_ = "";
+    /**
+     * <pre>
+     * The operation to run to cancel the queue.
+     * </pre>
+     *
+     * <code>string cancel_op_name = 4;</code>
+     */
+    public java.lang.String getCancelOpName() {
+      java.lang.Object ref = cancelOpName_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        cancelOpName_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * The operation to run to cancel the queue.
+     * </pre>
+     *
+     * <code>string cancel_op_name = 4;</code>
+     */
+    public com.google.protobuf.ByteString
+        getCancelOpNameBytes() {
+      java.lang.Object ref = cancelOpName_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        cancelOpName_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * The operation to run to cancel the queue.
+     * </pre>
+     *
+     * <code>string cancel_op_name = 4;</code>
+     */
+    public Builder setCancelOpName(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      cancelOpName_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * The operation to run to cancel the queue.
+     * </pre>
+     *
+     * <code>string cancel_op_name = 4;</code>
+     */
+    public Builder clearCancelOpName() {
+      
+      cancelOpName_ = getDefaultInstance().getCancelOpName();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * The operation to run to cancel the queue.
+     * </pre>
+     *
+     * <code>string cancel_op_name = 4;</code>
+     */
+    public Builder setCancelOpNameBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      cancelOpName_ = value;
+      onChanged();
+      return this;
+    }
+
+    private java.util.List<java.lang.Integer> queueClosedExceptionTypes_ =
+      java.util.Collections.emptyList();
+    private void ensureQueueClosedExceptionTypesIsMutable() {
+      if (!((bitField0_ & 0x00000002) != 0)) {
+        queueClosedExceptionTypes_ = new java.util.ArrayList<java.lang.Integer>(queueClosedExceptionTypes_);
+        bitField0_ |= 0x00000002;
+      }
+    }
+    /**
+     * <pre>
+     * A list of exception types considered to signal a safely closed queue
+     * if raised during enqueue operations.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.error.Code queue_closed_exception_types = 5;</code>
+     */
+    public java.util.List<org.tensorflow.proto.framework.Code> getQueueClosedExceptionTypesList() {
+      return new com.google.protobuf.Internal.ListAdapter<
+          java.lang.Integer, org.tensorflow.proto.framework.Code>(queueClosedExceptionTypes_, queueClosedExceptionTypes_converter_);
+    }
+    /**
+     * <pre>
+     * A list of exception types considered to signal a safely closed queue
+     * if raised during enqueue operations.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.error.Code queue_closed_exception_types = 5;</code>
+     */
+    public int getQueueClosedExceptionTypesCount() {
+      return queueClosedExceptionTypes_.size();
+    }
+    /**
+     * <pre>
+     * A list of exception types considered to signal a safely closed queue
+     * if raised during enqueue operations.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.error.Code queue_closed_exception_types = 5;</code>
+     */
+    public org.tensorflow.proto.framework.Code getQueueClosedExceptionTypes(int index) {
+      return queueClosedExceptionTypes_converter_.convert(queueClosedExceptionTypes_.get(index));
+    }
+    /**
+     * <pre>
+     * A list of exception types considered to signal a safely closed queue
+     * if raised during enqueue operations.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.error.Code queue_closed_exception_types = 5;</code>
+     */
+    public Builder setQueueClosedExceptionTypes(
+        int index, org.tensorflow.proto.framework.Code value) {
+      if (value == null) {
+        throw new NullPointerException();
+      }
+      ensureQueueClosedExceptionTypesIsMutable();
+      queueClosedExceptionTypes_.set(index, value.getNumber());
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * A list of exception types considered to signal a safely closed queue
+     * if raised during enqueue operations.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.error.Code queue_closed_exception_types = 5;</code>
+     */
+    public Builder addQueueClosedExceptionTypes(org.tensorflow.proto.framework.Code value) {
+      if (value == null) {
+        throw new NullPointerException();
+      }
+      ensureQueueClosedExceptionTypesIsMutable();
+      queueClosedExceptionTypes_.add(value.getNumber());
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * A list of exception types considered to signal a safely closed queue
+     * if raised during enqueue operations.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.error.Code queue_closed_exception_types = 5;</code>
+     */
+    public Builder addAllQueueClosedExceptionTypes(
+        java.lang.Iterable<? extends org.tensorflow.proto.framework.Code> values) {
+      ensureQueueClosedExceptionTypesIsMutable();
+      for (org.tensorflow.proto.framework.Code value : values) {
+        queueClosedExceptionTypes_.add(value.getNumber());
+      }
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * A list of exception types considered to signal a safely closed queue
+     * if raised during enqueue operations.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.error.Code queue_closed_exception_types = 5;</code>
+     */
+    public Builder clearQueueClosedExceptionTypes() {
+      queueClosedExceptionTypes_ = java.util.Collections.emptyList();
+      bitField0_ = (bitField0_ & ~0x00000002);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * A list of exception types considered to signal a safely closed queue
+     * if raised during enqueue operations.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.error.Code queue_closed_exception_types = 5;</code>
+     */
+    public java.util.List<java.lang.Integer>
+    getQueueClosedExceptionTypesValueList() {
+      return java.util.Collections.unmodifiableList(queueClosedExceptionTypes_);
+    }
+    /**
+     * <pre>
+     * A list of exception types considered to signal a safely closed queue
+     * if raised during enqueue operations.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.error.Code queue_closed_exception_types = 5;</code>
+     */
+    public int getQueueClosedExceptionTypesValue(int index) {
+      return queueClosedExceptionTypes_.get(index);
+    }
+    /**
+     * <pre>
+     * A list of exception types considered to signal a safely closed queue
+     * if raised during enqueue operations.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.error.Code queue_closed_exception_types = 5;</code>
+     */
+    public Builder setQueueClosedExceptionTypesValue(
+        int index, int value) {
+      ensureQueueClosedExceptionTypesIsMutable();
+      queueClosedExceptionTypes_.set(index, value);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * A list of exception types considered to signal a safely closed queue
+     * if raised during enqueue operations.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.error.Code queue_closed_exception_types = 5;</code>
+     */
+    public Builder addQueueClosedExceptionTypesValue(int value) {
+      ensureQueueClosedExceptionTypesIsMutable();
+      queueClosedExceptionTypes_.add(value);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * A list of exception types considered to signal a safely closed queue
+     * if raised during enqueue operations.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.error.Code queue_closed_exception_types = 5;</code>
+     */
+    public Builder addAllQueueClosedExceptionTypesValue(
+        java.lang.Iterable<java.lang.Integer> values) {
+      ensureQueueClosedExceptionTypesIsMutable();
+      for (int value : values) {
+        queueClosedExceptionTypes_.add(value);
+      }
+      onChanged();
+      return this;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.QueueRunnerDef)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.QueueRunnerDef)
+  private static final org.tensorflow.proto.framework.QueueRunnerDef DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.QueueRunnerDef();
+  }
+
+  public static org.tensorflow.proto.framework.QueueRunnerDef getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<QueueRunnerDef>
+      PARSER = new com.google.protobuf.AbstractParser<QueueRunnerDef>() {
+    @java.lang.Override
+    public QueueRunnerDef parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new QueueRunnerDef(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<QueueRunnerDef> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<QueueRunnerDef> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.framework.QueueRunnerDef getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/QueueRunnerDefOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/QueueRunnerDefOrBuilder.java
new file mode 100644
index 00000000000..61a20e767a3
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/QueueRunnerDefOrBuilder.java
@@ -0,0 +1,145 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/queue_runner.proto
+
+package org.tensorflow.proto.framework;
+
+public interface QueueRunnerDefOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.QueueRunnerDef)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <pre>
+   * Queue name.
+   * </pre>
+   *
+   * <code>string queue_name = 1;</code>
+   */
+  java.lang.String getQueueName();
+  /**
+   * <pre>
+   * Queue name.
+   * </pre>
+   *
+   * <code>string queue_name = 1;</code>
+   */
+  com.google.protobuf.ByteString
+      getQueueNameBytes();
+
+  /**
+   * <pre>
+   * A list of enqueue operations.
+   * </pre>
+   *
+   * <code>repeated string enqueue_op_name = 2;</code>
+   */
+  java.util.List<java.lang.String>
+      getEnqueueOpNameList();
+  /**
+   * <pre>
+   * A list of enqueue operations.
+   * </pre>
+   *
+   * <code>repeated string enqueue_op_name = 2;</code>
+   */
+  int getEnqueueOpNameCount();
+  /**
+   * <pre>
+   * A list of enqueue operations.
+   * </pre>
+   *
+   * <code>repeated string enqueue_op_name = 2;</code>
+   */
+  java.lang.String getEnqueueOpName(int index);
+  /**
+   * <pre>
+   * A list of enqueue operations.
+   * </pre>
+   *
+   * <code>repeated string enqueue_op_name = 2;</code>
+   */
+  com.google.protobuf.ByteString
+      getEnqueueOpNameBytes(int index);
+
+  /**
+   * <pre>
+   * The operation to run to close the queue.
+   * </pre>
+   *
+   * <code>string close_op_name = 3;</code>
+   */
+  java.lang.String getCloseOpName();
+  /**
+   * <pre>
+   * The operation to run to close the queue.
+   * </pre>
+   *
+   * <code>string close_op_name = 3;</code>
+   */
+  com.google.protobuf.ByteString
+      getCloseOpNameBytes();
+
+  /**
+   * <pre>
+   * The operation to run to cancel the queue.
+   * </pre>
+   *
+   * <code>string cancel_op_name = 4;</code>
+   */
+  java.lang.String getCancelOpName();
+  /**
+   * <pre>
+   * The operation to run to cancel the queue.
+   * </pre>
+   *
+   * <code>string cancel_op_name = 4;</code>
+   */
+  com.google.protobuf.ByteString
+      getCancelOpNameBytes();
+
+  /**
+   * <pre>
+   * A list of exception types considered to signal a safely closed queue
+   * if raised during enqueue operations.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.error.Code queue_closed_exception_types = 5;</code>
+   */
+  java.util.List<org.tensorflow.proto.framework.Code> getQueueClosedExceptionTypesList();
+  /**
+   * <pre>
+   * A list of exception types considered to signal a safely closed queue
+   * if raised during enqueue operations.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.error.Code queue_closed_exception_types = 5;</code>
+   */
+  int getQueueClosedExceptionTypesCount();
+  /**
+   * <pre>
+   * A list of exception types considered to signal a safely closed queue
+   * if raised during enqueue operations.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.error.Code queue_closed_exception_types = 5;</code>
+   */
+  org.tensorflow.proto.framework.Code getQueueClosedExceptionTypes(int index);
+  /**
+   * <pre>
+   * A list of exception types considered to signal a safely closed queue
+   * if raised during enqueue operations.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.error.Code queue_closed_exception_types = 5;</code>
+   */
+  java.util.List<java.lang.Integer>
+  getQueueClosedExceptionTypesValueList();
+  /**
+   * <pre>
+   * A list of exception types considered to signal a safely closed queue
+   * if raised during enqueue operations.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.error.Code queue_closed_exception_types = 5;</code>
+   */
+  int getQueueClosedExceptionTypesValue(int index);
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/QueueRunnerProtos.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/QueueRunnerProtos.java
new file mode 100644
index 00000000000..37a45c16673
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/QueueRunnerProtos.java
@@ -0,0 +1,58 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/queue_runner.proto
+
+package org.tensorflow.proto.framework;
+
+public final class QueueRunnerProtos {
+  private QueueRunnerProtos() {}
+  public static void registerAllExtensions(
+      com.google.protobuf.ExtensionRegistryLite registry) {
+  }
+
+  public static void registerAllExtensions(
+      com.google.protobuf.ExtensionRegistry registry) {
+    registerAllExtensions(
+        (com.google.protobuf.ExtensionRegistryLite) registry);
+  }
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_QueueRunnerDef_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_QueueRunnerDef_fieldAccessorTable;
+
+  public static com.google.protobuf.Descriptors.FileDescriptor
+      getDescriptor() {
+    return descriptor;
+  }
+  private static  com.google.protobuf.Descriptors.FileDescriptor
+      descriptor;
+  static {
+    java.lang.String[] descriptorData = {
+      "\n+tensorflow/core/protobuf/queue_runner." +
+      "proto\022\ntensorflow\032*tensorflow/core/proto" +
+      "buf/error_codes.proto\"\252\001\n\016QueueRunnerDef" +
+      "\022\022\n\nqueue_name\030\001 \001(\t\022\027\n\017enqueue_op_name\030" +
+      "\002 \003(\t\022\025\n\rclose_op_name\030\003 \001(\t\022\026\n\016cancel_o" +
+      "p_name\030\004 \001(\t\022<\n\034queue_closed_exception_t" +
+      "ypes\030\005 \003(\0162\026.tensorflow.error.CodeB\202\001\n\036o" +
+      "rg.tensorflow.proto.frameworkB\021QueueRunn" +
+      "erProtosP\001ZHgithub.com/tensorflow/tensor" +
+      "flow/tensorflow/go/core/core_protos_go_p" +
+      "roto\370\001\001b\006proto3"
+    };
+    descriptor = com.google.protobuf.Descriptors.FileDescriptor
+      .internalBuildGeneratedFileFrom(descriptorData,
+        new com.google.protobuf.Descriptors.FileDescriptor[] {
+          org.tensorflow.proto.framework.ErrorCodesProtos.getDescriptor(),
+        });
+    internal_static_tensorflow_QueueRunnerDef_descriptor =
+      getDescriptor().getMessageTypes().get(0);
+    internal_static_tensorflow_QueueRunnerDef_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_QueueRunnerDef_descriptor,
+        new java.lang.String[] { "QueueName", "EnqueueOpName", "CloseOpName", "CancelOpName", "QueueClosedExceptionTypes", });
+    org.tensorflow.proto.framework.ErrorCodesProtos.getDescriptor();
+  }
+
+  // @@protoc_insertion_point(outer_class_scope)
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/RPCOptions.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/RPCOptions.java
new file mode 100644
index 00000000000..6c1a717876d
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/RPCOptions.java
@@ -0,0 +1,905 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/config.proto
+
+package org.tensorflow.proto.framework;
+
+/**
+ * Protobuf type {@code tensorflow.RPCOptions}
+ */
+public  final class RPCOptions extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.RPCOptions)
+    RPCOptionsOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use RPCOptions.newBuilder() to construct.
+  private RPCOptions(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private RPCOptions() {
+    compressionAlgorithm_ = "";
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new RPCOptions();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private RPCOptions(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 8: {
+
+            useRpcForInprocessMaster_ = input.readBool();
+            break;
+          }
+          case 18: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            compressionAlgorithm_ = s;
+            break;
+          }
+          case 24: {
+
+            compressionLevel_ = input.readInt32();
+            break;
+          }
+          case 32: {
+
+            cacheRpcResponse_ = input.readBool();
+            break;
+          }
+          case 40: {
+
+            disableSessionConnectionSharing_ = input.readBool();
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_RPCOptions_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_RPCOptions_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.framework.RPCOptions.class, org.tensorflow.proto.framework.RPCOptions.Builder.class);
+  }
+
+  public static final int USE_RPC_FOR_INPROCESS_MASTER_FIELD_NUMBER = 1;
+  private boolean useRpcForInprocessMaster_;
+  /**
+   * <pre>
+   * If true, always use RPC to contact the session target.
+   * If false (the default option), TensorFlow may use an optimized
+   * transport for client-master communication that avoids the RPC
+   * stack. This option is primarily for used testing the RPC stack.
+   * </pre>
+   *
+   * <code>bool use_rpc_for_inprocess_master = 1;</code>
+   */
+  public boolean getUseRpcForInprocessMaster() {
+    return useRpcForInprocessMaster_;
+  }
+
+  public static final int COMPRESSION_ALGORITHM_FIELD_NUMBER = 2;
+  private volatile java.lang.Object compressionAlgorithm_;
+  /**
+   * <pre>
+   * The compression algorithm to be used. One of "deflate", "gzip".
+   * </pre>
+   *
+   * <code>string compression_algorithm = 2;</code>
+   */
+  public java.lang.String getCompressionAlgorithm() {
+    java.lang.Object ref = compressionAlgorithm_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      compressionAlgorithm_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * The compression algorithm to be used. One of "deflate", "gzip".
+   * </pre>
+   *
+   * <code>string compression_algorithm = 2;</code>
+   */
+  public com.google.protobuf.ByteString
+      getCompressionAlgorithmBytes() {
+    java.lang.Object ref = compressionAlgorithm_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      compressionAlgorithm_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int COMPRESSION_LEVEL_FIELD_NUMBER = 3;
+  private int compressionLevel_;
+  /**
+   * <pre>
+   * If compression_algorithm is set, the compression level to be used.
+   * From 0 (no compression), up to 3.
+   * </pre>
+   *
+   * <code>int32 compression_level = 3;</code>
+   */
+  public int getCompressionLevel() {
+    return compressionLevel_;
+  }
+
+  public static final int CACHE_RPC_RESPONSE_FIELD_NUMBER = 4;
+  private boolean cacheRpcResponse_;
+  /**
+   * <pre>
+   * Setting cache_rpc_response to true will enable sender side caching of
+   * response for RecvTensorAsync and RecvBufAsync to allow receiver to retry
+   * requests . This is only necessary when the network fabric is experiencing a
+   * significant error rate.  Without it we'll fail a step on an network error,
+   * while with it we'll be able to complete long steps (like complex
+   * initializations) in the face of some network errors during RecvTensor.
+   * </pre>
+   *
+   * <code>bool cache_rpc_response = 4;</code>
+   */
+  public boolean getCacheRpcResponse() {
+    return cacheRpcResponse_;
+  }
+
+  public static final int DISABLE_SESSION_CONNECTION_SHARING_FIELD_NUMBER = 5;
+  private boolean disableSessionConnectionSharing_;
+  /**
+   * <pre>
+   * Disables TCP connection sharing when opening a new RPC channel.
+   * </pre>
+   *
+   * <code>bool disable_session_connection_sharing = 5;</code>
+   */
+  public boolean getDisableSessionConnectionSharing() {
+    return disableSessionConnectionSharing_;
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    if (useRpcForInprocessMaster_ != false) {
+      output.writeBool(1, useRpcForInprocessMaster_);
+    }
+    if (!getCompressionAlgorithmBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 2, compressionAlgorithm_);
+    }
+    if (compressionLevel_ != 0) {
+      output.writeInt32(3, compressionLevel_);
+    }
+    if (cacheRpcResponse_ != false) {
+      output.writeBool(4, cacheRpcResponse_);
+    }
+    if (disableSessionConnectionSharing_ != false) {
+      output.writeBool(5, disableSessionConnectionSharing_);
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (useRpcForInprocessMaster_ != false) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeBoolSize(1, useRpcForInprocessMaster_);
+    }
+    if (!getCompressionAlgorithmBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(2, compressionAlgorithm_);
+    }
+    if (compressionLevel_ != 0) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt32Size(3, compressionLevel_);
+    }
+    if (cacheRpcResponse_ != false) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeBoolSize(4, cacheRpcResponse_);
+    }
+    if (disableSessionConnectionSharing_ != false) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeBoolSize(5, disableSessionConnectionSharing_);
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.framework.RPCOptions)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.framework.RPCOptions other = (org.tensorflow.proto.framework.RPCOptions) obj;
+
+    if (getUseRpcForInprocessMaster()
+        != other.getUseRpcForInprocessMaster()) return false;
+    if (!getCompressionAlgorithm()
+        .equals(other.getCompressionAlgorithm())) return false;
+    if (getCompressionLevel()
+        != other.getCompressionLevel()) return false;
+    if (getCacheRpcResponse()
+        != other.getCacheRpcResponse()) return false;
+    if (getDisableSessionConnectionSharing()
+        != other.getDisableSessionConnectionSharing()) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    hash = (37 * hash) + USE_RPC_FOR_INPROCESS_MASTER_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashBoolean(
+        getUseRpcForInprocessMaster());
+    hash = (37 * hash) + COMPRESSION_ALGORITHM_FIELD_NUMBER;
+    hash = (53 * hash) + getCompressionAlgorithm().hashCode();
+    hash = (37 * hash) + COMPRESSION_LEVEL_FIELD_NUMBER;
+    hash = (53 * hash) + getCompressionLevel();
+    hash = (37 * hash) + CACHE_RPC_RESPONSE_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashBoolean(
+        getCacheRpcResponse());
+    hash = (37 * hash) + DISABLE_SESSION_CONNECTION_SHARING_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashBoolean(
+        getDisableSessionConnectionSharing());
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.framework.RPCOptions parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.RPCOptions parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.RPCOptions parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.RPCOptions parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.RPCOptions parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.RPCOptions parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.RPCOptions parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.RPCOptions parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.RPCOptions parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.RPCOptions parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.RPCOptions parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.RPCOptions parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.framework.RPCOptions prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * Protobuf type {@code tensorflow.RPCOptions}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.RPCOptions)
+      org.tensorflow.proto.framework.RPCOptionsOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_RPCOptions_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_RPCOptions_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.RPCOptions.class, org.tensorflow.proto.framework.RPCOptions.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.framework.RPCOptions.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      useRpcForInprocessMaster_ = false;
+
+      compressionAlgorithm_ = "";
+
+      compressionLevel_ = 0;
+
+      cacheRpcResponse_ = false;
+
+      disableSessionConnectionSharing_ = false;
+
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_RPCOptions_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.RPCOptions getDefaultInstanceForType() {
+      return org.tensorflow.proto.framework.RPCOptions.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.RPCOptions build() {
+      org.tensorflow.proto.framework.RPCOptions result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.RPCOptions buildPartial() {
+      org.tensorflow.proto.framework.RPCOptions result = new org.tensorflow.proto.framework.RPCOptions(this);
+      result.useRpcForInprocessMaster_ = useRpcForInprocessMaster_;
+      result.compressionAlgorithm_ = compressionAlgorithm_;
+      result.compressionLevel_ = compressionLevel_;
+      result.cacheRpcResponse_ = cacheRpcResponse_;
+      result.disableSessionConnectionSharing_ = disableSessionConnectionSharing_;
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.framework.RPCOptions) {
+        return mergeFrom((org.tensorflow.proto.framework.RPCOptions)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.framework.RPCOptions other) {
+      if (other == org.tensorflow.proto.framework.RPCOptions.getDefaultInstance()) return this;
+      if (other.getUseRpcForInprocessMaster() != false) {
+        setUseRpcForInprocessMaster(other.getUseRpcForInprocessMaster());
+      }
+      if (!other.getCompressionAlgorithm().isEmpty()) {
+        compressionAlgorithm_ = other.compressionAlgorithm_;
+        onChanged();
+      }
+      if (other.getCompressionLevel() != 0) {
+        setCompressionLevel(other.getCompressionLevel());
+      }
+      if (other.getCacheRpcResponse() != false) {
+        setCacheRpcResponse(other.getCacheRpcResponse());
+      }
+      if (other.getDisableSessionConnectionSharing() != false) {
+        setDisableSessionConnectionSharing(other.getDisableSessionConnectionSharing());
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.framework.RPCOptions parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.framework.RPCOptions) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+
+    private boolean useRpcForInprocessMaster_ ;
+    /**
+     * <pre>
+     * If true, always use RPC to contact the session target.
+     * If false (the default option), TensorFlow may use an optimized
+     * transport for client-master communication that avoids the RPC
+     * stack. This option is primarily for used testing the RPC stack.
+     * </pre>
+     *
+     * <code>bool use_rpc_for_inprocess_master = 1;</code>
+     */
+    public boolean getUseRpcForInprocessMaster() {
+      return useRpcForInprocessMaster_;
+    }
+    /**
+     * <pre>
+     * If true, always use RPC to contact the session target.
+     * If false (the default option), TensorFlow may use an optimized
+     * transport for client-master communication that avoids the RPC
+     * stack. This option is primarily for used testing the RPC stack.
+     * </pre>
+     *
+     * <code>bool use_rpc_for_inprocess_master = 1;</code>
+     */
+    public Builder setUseRpcForInprocessMaster(boolean value) {
+      
+      useRpcForInprocessMaster_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * If true, always use RPC to contact the session target.
+     * If false (the default option), TensorFlow may use an optimized
+     * transport for client-master communication that avoids the RPC
+     * stack. This option is primarily for used testing the RPC stack.
+     * </pre>
+     *
+     * <code>bool use_rpc_for_inprocess_master = 1;</code>
+     */
+    public Builder clearUseRpcForInprocessMaster() {
+      
+      useRpcForInprocessMaster_ = false;
+      onChanged();
+      return this;
+    }
+
+    private java.lang.Object compressionAlgorithm_ = "";
+    /**
+     * <pre>
+     * The compression algorithm to be used. One of "deflate", "gzip".
+     * </pre>
+     *
+     * <code>string compression_algorithm = 2;</code>
+     */
+    public java.lang.String getCompressionAlgorithm() {
+      java.lang.Object ref = compressionAlgorithm_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        compressionAlgorithm_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * The compression algorithm to be used. One of "deflate", "gzip".
+     * </pre>
+     *
+     * <code>string compression_algorithm = 2;</code>
+     */
+    public com.google.protobuf.ByteString
+        getCompressionAlgorithmBytes() {
+      java.lang.Object ref = compressionAlgorithm_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        compressionAlgorithm_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * The compression algorithm to be used. One of "deflate", "gzip".
+     * </pre>
+     *
+     * <code>string compression_algorithm = 2;</code>
+     */
+    public Builder setCompressionAlgorithm(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      compressionAlgorithm_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * The compression algorithm to be used. One of "deflate", "gzip".
+     * </pre>
+     *
+     * <code>string compression_algorithm = 2;</code>
+     */
+    public Builder clearCompressionAlgorithm() {
+      
+      compressionAlgorithm_ = getDefaultInstance().getCompressionAlgorithm();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * The compression algorithm to be used. One of "deflate", "gzip".
+     * </pre>
+     *
+     * <code>string compression_algorithm = 2;</code>
+     */
+    public Builder setCompressionAlgorithmBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      compressionAlgorithm_ = value;
+      onChanged();
+      return this;
+    }
+
+    private int compressionLevel_ ;
+    /**
+     * <pre>
+     * If compression_algorithm is set, the compression level to be used.
+     * From 0 (no compression), up to 3.
+     * </pre>
+     *
+     * <code>int32 compression_level = 3;</code>
+     */
+    public int getCompressionLevel() {
+      return compressionLevel_;
+    }
+    /**
+     * <pre>
+     * If compression_algorithm is set, the compression level to be used.
+     * From 0 (no compression), up to 3.
+     * </pre>
+     *
+     * <code>int32 compression_level = 3;</code>
+     */
+    public Builder setCompressionLevel(int value) {
+      
+      compressionLevel_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * If compression_algorithm is set, the compression level to be used.
+     * From 0 (no compression), up to 3.
+     * </pre>
+     *
+     * <code>int32 compression_level = 3;</code>
+     */
+    public Builder clearCompressionLevel() {
+      
+      compressionLevel_ = 0;
+      onChanged();
+      return this;
+    }
+
+    private boolean cacheRpcResponse_ ;
+    /**
+     * <pre>
+     * Setting cache_rpc_response to true will enable sender side caching of
+     * response for RecvTensorAsync and RecvBufAsync to allow receiver to retry
+     * requests . This is only necessary when the network fabric is experiencing a
+     * significant error rate.  Without it we'll fail a step on an network error,
+     * while with it we'll be able to complete long steps (like complex
+     * initializations) in the face of some network errors during RecvTensor.
+     * </pre>
+     *
+     * <code>bool cache_rpc_response = 4;</code>
+     */
+    public boolean getCacheRpcResponse() {
+      return cacheRpcResponse_;
+    }
+    /**
+     * <pre>
+     * Setting cache_rpc_response to true will enable sender side caching of
+     * response for RecvTensorAsync and RecvBufAsync to allow receiver to retry
+     * requests . This is only necessary when the network fabric is experiencing a
+     * significant error rate.  Without it we'll fail a step on an network error,
+     * while with it we'll be able to complete long steps (like complex
+     * initializations) in the face of some network errors during RecvTensor.
+     * </pre>
+     *
+     * <code>bool cache_rpc_response = 4;</code>
+     */
+    public Builder setCacheRpcResponse(boolean value) {
+      
+      cacheRpcResponse_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Setting cache_rpc_response to true will enable sender side caching of
+     * response for RecvTensorAsync and RecvBufAsync to allow receiver to retry
+     * requests . This is only necessary when the network fabric is experiencing a
+     * significant error rate.  Without it we'll fail a step on an network error,
+     * while with it we'll be able to complete long steps (like complex
+     * initializations) in the face of some network errors during RecvTensor.
+     * </pre>
+     *
+     * <code>bool cache_rpc_response = 4;</code>
+     */
+    public Builder clearCacheRpcResponse() {
+      
+      cacheRpcResponse_ = false;
+      onChanged();
+      return this;
+    }
+
+    private boolean disableSessionConnectionSharing_ ;
+    /**
+     * <pre>
+     * Disables TCP connection sharing when opening a new RPC channel.
+     * </pre>
+     *
+     * <code>bool disable_session_connection_sharing = 5;</code>
+     */
+    public boolean getDisableSessionConnectionSharing() {
+      return disableSessionConnectionSharing_;
+    }
+    /**
+     * <pre>
+     * Disables TCP connection sharing when opening a new RPC channel.
+     * </pre>
+     *
+     * <code>bool disable_session_connection_sharing = 5;</code>
+     */
+    public Builder setDisableSessionConnectionSharing(boolean value) {
+      
+      disableSessionConnectionSharing_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Disables TCP connection sharing when opening a new RPC channel.
+     * </pre>
+     *
+     * <code>bool disable_session_connection_sharing = 5;</code>
+     */
+    public Builder clearDisableSessionConnectionSharing() {
+      
+      disableSessionConnectionSharing_ = false;
+      onChanged();
+      return this;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.RPCOptions)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.RPCOptions)
+  private static final org.tensorflow.proto.framework.RPCOptions DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.RPCOptions();
+  }
+
+  public static org.tensorflow.proto.framework.RPCOptions getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<RPCOptions>
+      PARSER = new com.google.protobuf.AbstractParser<RPCOptions>() {
+    @java.lang.Override
+    public RPCOptions parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new RPCOptions(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<RPCOptions> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<RPCOptions> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.framework.RPCOptions getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/RPCOptionsOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/RPCOptionsOrBuilder.java
new file mode 100644
index 00000000000..e7178532c17
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/RPCOptionsOrBuilder.java
@@ -0,0 +1,72 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/config.proto
+
+package org.tensorflow.proto.framework;
+
+public interface RPCOptionsOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.RPCOptions)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <pre>
+   * If true, always use RPC to contact the session target.
+   * If false (the default option), TensorFlow may use an optimized
+   * transport for client-master communication that avoids the RPC
+   * stack. This option is primarily for used testing the RPC stack.
+   * </pre>
+   *
+   * <code>bool use_rpc_for_inprocess_master = 1;</code>
+   */
+  boolean getUseRpcForInprocessMaster();
+
+  /**
+   * <pre>
+   * The compression algorithm to be used. One of "deflate", "gzip".
+   * </pre>
+   *
+   * <code>string compression_algorithm = 2;</code>
+   */
+  java.lang.String getCompressionAlgorithm();
+  /**
+   * <pre>
+   * The compression algorithm to be used. One of "deflate", "gzip".
+   * </pre>
+   *
+   * <code>string compression_algorithm = 2;</code>
+   */
+  com.google.protobuf.ByteString
+      getCompressionAlgorithmBytes();
+
+  /**
+   * <pre>
+   * If compression_algorithm is set, the compression level to be used.
+   * From 0 (no compression), up to 3.
+   * </pre>
+   *
+   * <code>int32 compression_level = 3;</code>
+   */
+  int getCompressionLevel();
+
+  /**
+   * <pre>
+   * Setting cache_rpc_response to true will enable sender side caching of
+   * response for RecvTensorAsync and RecvBufAsync to allow receiver to retry
+   * requests . This is only necessary when the network fabric is experiencing a
+   * significant error rate.  Without it we'll fail a step on an network error,
+   * while with it we'll be able to complete long steps (like complex
+   * initializations) in the face of some network errors during RecvTensor.
+   * </pre>
+   *
+   * <code>bool cache_rpc_response = 4;</code>
+   */
+  boolean getCacheRpcResponse();
+
+  /**
+   * <pre>
+   * Disables TCP connection sharing when opening a new RPC channel.
+   * </pre>
+   *
+   * <code>bool disable_session_connection_sharing = 5;</code>
+   */
+  boolean getDisableSessionConnectionSharing();
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ReaderBaseProtos.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ReaderBaseProtos.java
new file mode 100644
index 00000000000..3c4e600440b
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ReaderBaseProtos.java
@@ -0,0 +1,54 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/framework/reader_base.proto
+
+package org.tensorflow.proto.framework;
+
+public final class ReaderBaseProtos {
+  private ReaderBaseProtos() {}
+  public static void registerAllExtensions(
+      com.google.protobuf.ExtensionRegistryLite registry) {
+  }
+
+  public static void registerAllExtensions(
+      com.google.protobuf.ExtensionRegistry registry) {
+    registerAllExtensions(
+        (com.google.protobuf.ExtensionRegistryLite) registry);
+  }
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_ReaderBaseState_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_ReaderBaseState_fieldAccessorTable;
+
+  public static com.google.protobuf.Descriptors.FileDescriptor
+      getDescriptor() {
+    return descriptor;
+  }
+  private static  com.google.protobuf.Descriptors.FileDescriptor
+      descriptor;
+  static {
+    java.lang.String[] descriptorData = {
+      "\n+tensorflow/core/framework/reader_base." +
+      "proto\022\ntensorflow\"r\n\017ReaderBaseState\022\024\n\014" +
+      "work_started\030\001 \001(\003\022\025\n\rwork_finished\030\002 \001(" +
+      "\003\022\034\n\024num_records_produced\030\003 \001(\003\022\024\n\014curre" +
+      "nt_work\030\004 \001(\014B\213\001\n\036org.tensorflow.proto.f" +
+      "rameworkB\020ReaderBaseProtosP\001ZRgithub.com" +
+      "/tensorflow/tensorflow/tensorflow/go/cor" +
+      "e/framework/reader_base_go_proto\370\001\001b\006pro" +
+      "to3"
+    };
+    descriptor = com.google.protobuf.Descriptors.FileDescriptor
+      .internalBuildGeneratedFileFrom(descriptorData,
+        new com.google.protobuf.Descriptors.FileDescriptor[] {
+        });
+    internal_static_tensorflow_ReaderBaseState_descriptor =
+      getDescriptor().getMessageTypes().get(0);
+    internal_static_tensorflow_ReaderBaseState_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_ReaderBaseState_descriptor,
+        new java.lang.String[] { "WorkStarted", "WorkFinished", "NumRecordsProduced", "CurrentWork", });
+  }
+
+  // @@protoc_insertion_point(outer_class_scope)
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ReaderBaseState.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ReaderBaseState.java
new file mode 100644
index 00000000000..4602bd786a5
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ReaderBaseState.java
@@ -0,0 +1,664 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/framework/reader_base.proto
+
+package org.tensorflow.proto.framework;
+
+/**
+ * <pre>
+ * For serializing and restoring the state of ReaderBase, see
+ * reader_base.h for details.
+ * </pre>
+ *
+ * Protobuf type {@code tensorflow.ReaderBaseState}
+ */
+public  final class ReaderBaseState extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.ReaderBaseState)
+    ReaderBaseStateOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use ReaderBaseState.newBuilder() to construct.
+  private ReaderBaseState(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private ReaderBaseState() {
+    currentWork_ = com.google.protobuf.ByteString.EMPTY;
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new ReaderBaseState();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private ReaderBaseState(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 8: {
+
+            workStarted_ = input.readInt64();
+            break;
+          }
+          case 16: {
+
+            workFinished_ = input.readInt64();
+            break;
+          }
+          case 24: {
+
+            numRecordsProduced_ = input.readInt64();
+            break;
+          }
+          case 34: {
+
+            currentWork_ = input.readBytes();
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.framework.ReaderBaseProtos.internal_static_tensorflow_ReaderBaseState_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.framework.ReaderBaseProtos.internal_static_tensorflow_ReaderBaseState_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.framework.ReaderBaseState.class, org.tensorflow.proto.framework.ReaderBaseState.Builder.class);
+  }
+
+  public static final int WORK_STARTED_FIELD_NUMBER = 1;
+  private long workStarted_;
+  /**
+   * <code>int64 work_started = 1;</code>
+   */
+  public long getWorkStarted() {
+    return workStarted_;
+  }
+
+  public static final int WORK_FINISHED_FIELD_NUMBER = 2;
+  private long workFinished_;
+  /**
+   * <code>int64 work_finished = 2;</code>
+   */
+  public long getWorkFinished() {
+    return workFinished_;
+  }
+
+  public static final int NUM_RECORDS_PRODUCED_FIELD_NUMBER = 3;
+  private long numRecordsProduced_;
+  /**
+   * <code>int64 num_records_produced = 3;</code>
+   */
+  public long getNumRecordsProduced() {
+    return numRecordsProduced_;
+  }
+
+  public static final int CURRENT_WORK_FIELD_NUMBER = 4;
+  private com.google.protobuf.ByteString currentWork_;
+  /**
+   * <code>bytes current_work = 4;</code>
+   */
+  public com.google.protobuf.ByteString getCurrentWork() {
+    return currentWork_;
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    if (workStarted_ != 0L) {
+      output.writeInt64(1, workStarted_);
+    }
+    if (workFinished_ != 0L) {
+      output.writeInt64(2, workFinished_);
+    }
+    if (numRecordsProduced_ != 0L) {
+      output.writeInt64(3, numRecordsProduced_);
+    }
+    if (!currentWork_.isEmpty()) {
+      output.writeBytes(4, currentWork_);
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (workStarted_ != 0L) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt64Size(1, workStarted_);
+    }
+    if (workFinished_ != 0L) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt64Size(2, workFinished_);
+    }
+    if (numRecordsProduced_ != 0L) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt64Size(3, numRecordsProduced_);
+    }
+    if (!currentWork_.isEmpty()) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeBytesSize(4, currentWork_);
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.framework.ReaderBaseState)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.framework.ReaderBaseState other = (org.tensorflow.proto.framework.ReaderBaseState) obj;
+
+    if (getWorkStarted()
+        != other.getWorkStarted()) return false;
+    if (getWorkFinished()
+        != other.getWorkFinished()) return false;
+    if (getNumRecordsProduced()
+        != other.getNumRecordsProduced()) return false;
+    if (!getCurrentWork()
+        .equals(other.getCurrentWork())) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    hash = (37 * hash) + WORK_STARTED_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+        getWorkStarted());
+    hash = (37 * hash) + WORK_FINISHED_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+        getWorkFinished());
+    hash = (37 * hash) + NUM_RECORDS_PRODUCED_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+        getNumRecordsProduced());
+    hash = (37 * hash) + CURRENT_WORK_FIELD_NUMBER;
+    hash = (53 * hash) + getCurrentWork().hashCode();
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.framework.ReaderBaseState parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.ReaderBaseState parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.ReaderBaseState parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.ReaderBaseState parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.ReaderBaseState parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.ReaderBaseState parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.ReaderBaseState parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.ReaderBaseState parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.ReaderBaseState parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.ReaderBaseState parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.ReaderBaseState parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.ReaderBaseState parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.framework.ReaderBaseState prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * <pre>
+   * For serializing and restoring the state of ReaderBase, see
+   * reader_base.h for details.
+   * </pre>
+   *
+   * Protobuf type {@code tensorflow.ReaderBaseState}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.ReaderBaseState)
+      org.tensorflow.proto.framework.ReaderBaseStateOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.ReaderBaseProtos.internal_static_tensorflow_ReaderBaseState_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.ReaderBaseProtos.internal_static_tensorflow_ReaderBaseState_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.ReaderBaseState.class, org.tensorflow.proto.framework.ReaderBaseState.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.framework.ReaderBaseState.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      workStarted_ = 0L;
+
+      workFinished_ = 0L;
+
+      numRecordsProduced_ = 0L;
+
+      currentWork_ = com.google.protobuf.ByteString.EMPTY;
+
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.framework.ReaderBaseProtos.internal_static_tensorflow_ReaderBaseState_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.ReaderBaseState getDefaultInstanceForType() {
+      return org.tensorflow.proto.framework.ReaderBaseState.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.ReaderBaseState build() {
+      org.tensorflow.proto.framework.ReaderBaseState result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.ReaderBaseState buildPartial() {
+      org.tensorflow.proto.framework.ReaderBaseState result = new org.tensorflow.proto.framework.ReaderBaseState(this);
+      result.workStarted_ = workStarted_;
+      result.workFinished_ = workFinished_;
+      result.numRecordsProduced_ = numRecordsProduced_;
+      result.currentWork_ = currentWork_;
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.framework.ReaderBaseState) {
+        return mergeFrom((org.tensorflow.proto.framework.ReaderBaseState)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.framework.ReaderBaseState other) {
+      if (other == org.tensorflow.proto.framework.ReaderBaseState.getDefaultInstance()) return this;
+      if (other.getWorkStarted() != 0L) {
+        setWorkStarted(other.getWorkStarted());
+      }
+      if (other.getWorkFinished() != 0L) {
+        setWorkFinished(other.getWorkFinished());
+      }
+      if (other.getNumRecordsProduced() != 0L) {
+        setNumRecordsProduced(other.getNumRecordsProduced());
+      }
+      if (other.getCurrentWork() != com.google.protobuf.ByteString.EMPTY) {
+        setCurrentWork(other.getCurrentWork());
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.framework.ReaderBaseState parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.framework.ReaderBaseState) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+
+    private long workStarted_ ;
+    /**
+     * <code>int64 work_started = 1;</code>
+     */
+    public long getWorkStarted() {
+      return workStarted_;
+    }
+    /**
+     * <code>int64 work_started = 1;</code>
+     */
+    public Builder setWorkStarted(long value) {
+      
+      workStarted_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>int64 work_started = 1;</code>
+     */
+    public Builder clearWorkStarted() {
+      
+      workStarted_ = 0L;
+      onChanged();
+      return this;
+    }
+
+    private long workFinished_ ;
+    /**
+     * <code>int64 work_finished = 2;</code>
+     */
+    public long getWorkFinished() {
+      return workFinished_;
+    }
+    /**
+     * <code>int64 work_finished = 2;</code>
+     */
+    public Builder setWorkFinished(long value) {
+      
+      workFinished_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>int64 work_finished = 2;</code>
+     */
+    public Builder clearWorkFinished() {
+      
+      workFinished_ = 0L;
+      onChanged();
+      return this;
+    }
+
+    private long numRecordsProduced_ ;
+    /**
+     * <code>int64 num_records_produced = 3;</code>
+     */
+    public long getNumRecordsProduced() {
+      return numRecordsProduced_;
+    }
+    /**
+     * <code>int64 num_records_produced = 3;</code>
+     */
+    public Builder setNumRecordsProduced(long value) {
+      
+      numRecordsProduced_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>int64 num_records_produced = 3;</code>
+     */
+    public Builder clearNumRecordsProduced() {
+      
+      numRecordsProduced_ = 0L;
+      onChanged();
+      return this;
+    }
+
+    private com.google.protobuf.ByteString currentWork_ = com.google.protobuf.ByteString.EMPTY;
+    /**
+     * <code>bytes current_work = 4;</code>
+     */
+    public com.google.protobuf.ByteString getCurrentWork() {
+      return currentWork_;
+    }
+    /**
+     * <code>bytes current_work = 4;</code>
+     */
+    public Builder setCurrentWork(com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      currentWork_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>bytes current_work = 4;</code>
+     */
+    public Builder clearCurrentWork() {
+      
+      currentWork_ = getDefaultInstance().getCurrentWork();
+      onChanged();
+      return this;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.ReaderBaseState)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.ReaderBaseState)
+  private static final org.tensorflow.proto.framework.ReaderBaseState DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.ReaderBaseState();
+  }
+
+  public static org.tensorflow.proto.framework.ReaderBaseState getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<ReaderBaseState>
+      PARSER = new com.google.protobuf.AbstractParser<ReaderBaseState>() {
+    @java.lang.Override
+    public ReaderBaseState parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new ReaderBaseState(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<ReaderBaseState> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<ReaderBaseState> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.framework.ReaderBaseState getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ReaderBaseStateOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ReaderBaseStateOrBuilder.java
new file mode 100644
index 00000000000..72f0c6c0922
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ReaderBaseStateOrBuilder.java
@@ -0,0 +1,29 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/framework/reader_base.proto
+
+package org.tensorflow.proto.framework;
+
+public interface ReaderBaseStateOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.ReaderBaseState)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <code>int64 work_started = 1;</code>
+   */
+  long getWorkStarted();
+
+  /**
+   * <code>int64 work_finished = 2;</code>
+   */
+  long getWorkFinished();
+
+  /**
+   * <code>int64 num_records_produced = 3;</code>
+   */
+  long getNumRecordsProduced();
+
+  /**
+   * <code>bytes current_work = 4;</code>
+   */
+  com.google.protobuf.ByteString getCurrentWork();
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/RemoteFusedGraphExecuteInfo.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/RemoteFusedGraphExecuteInfo.java
new file mode 100644
index 00000000000..bc2e6840de2
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/RemoteFusedGraphExecuteInfo.java
@@ -0,0 +1,3007 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/framework/remote_fused_graph_execute_info.proto
+
+package org.tensorflow.proto.framework;
+
+/**
+ * <pre>
+ * Protocol buffer representing a handle to a tensorflow resource. Handles are
+ * not valid across executions, but can be serialized back and forth from within
+ * a single run.
+ * </pre>
+ *
+ * Protobuf type {@code tensorflow.RemoteFusedGraphExecuteInfo}
+ */
+public  final class RemoteFusedGraphExecuteInfo extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.RemoteFusedGraphExecuteInfo)
+    RemoteFusedGraphExecuteInfoOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use RemoteFusedGraphExecuteInfo.newBuilder() to construct.
+  private RemoteFusedGraphExecuteInfo(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private RemoteFusedGraphExecuteInfo() {
+    graphInputNodeName_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+    graphOutputNodeName_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+    executorName_ = "";
+    serializedExecutorParameters_ = com.google.protobuf.ByteString.EMPTY;
+    defaultGraphInputTensorShape_ = java.util.Collections.emptyList();
+    defaultGraphOutputTensorShape_ = java.util.Collections.emptyList();
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new RemoteFusedGraphExecuteInfo();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private RemoteFusedGraphExecuteInfo(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    int mutable_bitField0_ = 0;
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 10: {
+            org.tensorflow.proto.framework.GraphDef.Builder subBuilder = null;
+            if (remoteGraph_ != null) {
+              subBuilder = remoteGraph_.toBuilder();
+            }
+            remoteGraph_ = input.readMessage(org.tensorflow.proto.framework.GraphDef.parser(), extensionRegistry);
+            if (subBuilder != null) {
+              subBuilder.mergeFrom(remoteGraph_);
+              remoteGraph_ = subBuilder.buildPartial();
+            }
+
+            break;
+          }
+          case 18: {
+            java.lang.String s = input.readStringRequireUtf8();
+            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
+              graphInputNodeName_ = new com.google.protobuf.LazyStringArrayList();
+              mutable_bitField0_ |= 0x00000001;
+            }
+            graphInputNodeName_.add(s);
+            break;
+          }
+          case 26: {
+            java.lang.String s = input.readStringRequireUtf8();
+            if (!((mutable_bitField0_ & 0x00000002) != 0)) {
+              graphOutputNodeName_ = new com.google.protobuf.LazyStringArrayList();
+              mutable_bitField0_ |= 0x00000002;
+            }
+            graphOutputNodeName_.add(s);
+            break;
+          }
+          case 34: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            executorName_ = s;
+            break;
+          }
+          case 42: {
+
+            serializedExecutorParameters_ = input.readBytes();
+            break;
+          }
+          case 50: {
+            if (!((mutable_bitField0_ & 0x00000004) != 0)) {
+              defaultGraphInputTensorShape_ = new java.util.ArrayList<org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto>();
+              mutable_bitField0_ |= 0x00000004;
+            }
+            defaultGraphInputTensorShape_.add(
+                input.readMessage(org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto.parser(), extensionRegistry));
+            break;
+          }
+          case 58: {
+            if (!((mutable_bitField0_ & 0x00000008) != 0)) {
+              defaultGraphOutputTensorShape_ = new java.util.ArrayList<org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto>();
+              mutable_bitField0_ |= 0x00000008;
+            }
+            defaultGraphOutputTensorShape_.add(
+                input.readMessage(org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto.parser(), extensionRegistry));
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      if (((mutable_bitField0_ & 0x00000001) != 0)) {
+        graphInputNodeName_ = graphInputNodeName_.getUnmodifiableView();
+      }
+      if (((mutable_bitField0_ & 0x00000002) != 0)) {
+        graphOutputNodeName_ = graphOutputNodeName_.getUnmodifiableView();
+      }
+      if (((mutable_bitField0_ & 0x00000004) != 0)) {
+        defaultGraphInputTensorShape_ = java.util.Collections.unmodifiableList(defaultGraphInputTensorShape_);
+      }
+      if (((mutable_bitField0_ & 0x00000008) != 0)) {
+        defaultGraphOutputTensorShape_ = java.util.Collections.unmodifiableList(defaultGraphOutputTensorShape_);
+      }
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfoProto.internal_static_tensorflow_RemoteFusedGraphExecuteInfo_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfoProto.internal_static_tensorflow_RemoteFusedGraphExecuteInfo_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.class, org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.Builder.class);
+  }
+
+  public interface TensorShapeTypeProtoOrBuilder extends
+      // @@protoc_insertion_point(interface_extends:tensorflow.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto)
+      com.google.protobuf.MessageOrBuilder {
+
+    /**
+     * <code>.tensorflow.DataType dtype = 1;</code>
+     */
+    int getDtypeValue();
+    /**
+     * <code>.tensorflow.DataType dtype = 1;</code>
+     */
+    org.tensorflow.proto.framework.DataType getDtype();
+
+    /**
+     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+     */
+    boolean hasShape();
+    /**
+     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+     */
+    org.tensorflow.proto.framework.TensorShapeProto getShape();
+    /**
+     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+     */
+    org.tensorflow.proto.framework.TensorShapeProtoOrBuilder getShapeOrBuilder();
+  }
+  /**
+   * Protobuf type {@code tensorflow.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto}
+   */
+  public  static final class TensorShapeTypeProto extends
+      com.google.protobuf.GeneratedMessageV3 implements
+      // @@protoc_insertion_point(message_implements:tensorflow.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto)
+      TensorShapeTypeProtoOrBuilder {
+  private static final long serialVersionUID = 0L;
+    // Use TensorShapeTypeProto.newBuilder() to construct.
+    private TensorShapeTypeProto(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+      super(builder);
+    }
+    private TensorShapeTypeProto() {
+      dtype_ = 0;
+    }
+
+    @java.lang.Override
+    @SuppressWarnings({"unused"})
+    protected java.lang.Object newInstance(
+        UnusedPrivateParameter unused) {
+      return new TensorShapeTypeProto();
+    }
+
+    @java.lang.Override
+    public final com.google.protobuf.UnknownFieldSet
+    getUnknownFields() {
+      return this.unknownFields;
+    }
+    private TensorShapeTypeProto(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      this();
+      if (extensionRegistry == null) {
+        throw new java.lang.NullPointerException();
+      }
+      com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+          com.google.protobuf.UnknownFieldSet.newBuilder();
+      try {
+        boolean done = false;
+        while (!done) {
+          int tag = input.readTag();
+          switch (tag) {
+            case 0:
+              done = true;
+              break;
+            case 8: {
+              int rawValue = input.readEnum();
+
+              dtype_ = rawValue;
+              break;
+            }
+            case 18: {
+              org.tensorflow.proto.framework.TensorShapeProto.Builder subBuilder = null;
+              if (shape_ != null) {
+                subBuilder = shape_.toBuilder();
+              }
+              shape_ = input.readMessage(org.tensorflow.proto.framework.TensorShapeProto.parser(), extensionRegistry);
+              if (subBuilder != null) {
+                subBuilder.mergeFrom(shape_);
+                shape_ = subBuilder.buildPartial();
+              }
+
+              break;
+            }
+            default: {
+              if (!parseUnknownField(
+                  input, unknownFields, extensionRegistry, tag)) {
+                done = true;
+              }
+              break;
+            }
+          }
+        }
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        throw e.setUnfinishedMessage(this);
+      } catch (java.io.IOException e) {
+        throw new com.google.protobuf.InvalidProtocolBufferException(
+            e).setUnfinishedMessage(this);
+      } finally {
+        this.unknownFields = unknownFields.build();
+        makeExtensionsImmutable();
+      }
+    }
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfoProto.internal_static_tensorflow_RemoteFusedGraphExecuteInfo_TensorShapeTypeProto_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfoProto.internal_static_tensorflow_RemoteFusedGraphExecuteInfo_TensorShapeTypeProto_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto.class, org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto.Builder.class);
+    }
+
+    public static final int DTYPE_FIELD_NUMBER = 1;
+    private int dtype_;
+    /**
+     * <code>.tensorflow.DataType dtype = 1;</code>
+     */
+    public int getDtypeValue() {
+      return dtype_;
+    }
+    /**
+     * <code>.tensorflow.DataType dtype = 1;</code>
+     */
+    public org.tensorflow.proto.framework.DataType getDtype() {
+      @SuppressWarnings("deprecation")
+      org.tensorflow.proto.framework.DataType result = org.tensorflow.proto.framework.DataType.valueOf(dtype_);
+      return result == null ? org.tensorflow.proto.framework.DataType.UNRECOGNIZED : result;
+    }
+
+    public static final int SHAPE_FIELD_NUMBER = 2;
+    private org.tensorflow.proto.framework.TensorShapeProto shape_;
+    /**
+     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+     */
+    public boolean hasShape() {
+      return shape_ != null;
+    }
+    /**
+     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+     */
+    public org.tensorflow.proto.framework.TensorShapeProto getShape() {
+      return shape_ == null ? org.tensorflow.proto.framework.TensorShapeProto.getDefaultInstance() : shape_;
+    }
+    /**
+     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+     */
+    public org.tensorflow.proto.framework.TensorShapeProtoOrBuilder getShapeOrBuilder() {
+      return getShape();
+    }
+
+    private byte memoizedIsInitialized = -1;
+    @java.lang.Override
+    public final boolean isInitialized() {
+      byte isInitialized = memoizedIsInitialized;
+      if (isInitialized == 1) return true;
+      if (isInitialized == 0) return false;
+
+      memoizedIsInitialized = 1;
+      return true;
+    }
+
+    @java.lang.Override
+    public void writeTo(com.google.protobuf.CodedOutputStream output)
+                        throws java.io.IOException {
+      if (dtype_ != org.tensorflow.proto.framework.DataType.DT_INVALID.getNumber()) {
+        output.writeEnum(1, dtype_);
+      }
+      if (shape_ != null) {
+        output.writeMessage(2, getShape());
+      }
+      unknownFields.writeTo(output);
+    }
+
+    @java.lang.Override
+    public int getSerializedSize() {
+      int size = memoizedSize;
+      if (size != -1) return size;
+
+      size = 0;
+      if (dtype_ != org.tensorflow.proto.framework.DataType.DT_INVALID.getNumber()) {
+        size += com.google.protobuf.CodedOutputStream
+          .computeEnumSize(1, dtype_);
+      }
+      if (shape_ != null) {
+        size += com.google.protobuf.CodedOutputStream
+          .computeMessageSize(2, getShape());
+      }
+      size += unknownFields.getSerializedSize();
+      memoizedSize = size;
+      return size;
+    }
+
+    @java.lang.Override
+    public boolean equals(final java.lang.Object obj) {
+      if (obj == this) {
+       return true;
+      }
+      if (!(obj instanceof org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto)) {
+        return super.equals(obj);
+      }
+      org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto other = (org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto) obj;
+
+      if (dtype_ != other.dtype_) return false;
+      if (hasShape() != other.hasShape()) return false;
+      if (hasShape()) {
+        if (!getShape()
+            .equals(other.getShape())) return false;
+      }
+      if (!unknownFields.equals(other.unknownFields)) return false;
+      return true;
+    }
+
+    @java.lang.Override
+    public int hashCode() {
+      if (memoizedHashCode != 0) {
+        return memoizedHashCode;
+      }
+      int hash = 41;
+      hash = (19 * hash) + getDescriptor().hashCode();
+      hash = (37 * hash) + DTYPE_FIELD_NUMBER;
+      hash = (53 * hash) + dtype_;
+      if (hasShape()) {
+        hash = (37 * hash) + SHAPE_FIELD_NUMBER;
+        hash = (53 * hash) + getShape().hashCode();
+      }
+      hash = (29 * hash) + unknownFields.hashCode();
+      memoizedHashCode = hash;
+      return hash;
+    }
+
+    public static org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto parseFrom(
+        java.nio.ByteBuffer data)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data);
+    }
+    public static org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto parseFrom(
+        java.nio.ByteBuffer data,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto parseFrom(
+        com.google.protobuf.ByteString data)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data);
+    }
+    public static org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto parseFrom(
+        com.google.protobuf.ByteString data,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto parseFrom(byte[] data)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data);
+    }
+    public static org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto parseFrom(
+        byte[] data,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto parseFrom(java.io.InputStream input)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input);
+    }
+    public static org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto parseFrom(
+        java.io.InputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto parseDelimitedFrom(java.io.InputStream input)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseDelimitedWithIOException(PARSER, input);
+    }
+    public static org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto parseDelimitedFrom(
+        java.io.InputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto parseFrom(
+        com.google.protobuf.CodedInputStream input)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input);
+    }
+    public static org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto parseFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input, extensionRegistry);
+    }
+
+    @java.lang.Override
+    public Builder newBuilderForType() { return newBuilder(); }
+    public static Builder newBuilder() {
+      return DEFAULT_INSTANCE.toBuilder();
+    }
+    public static Builder newBuilder(org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto prototype) {
+      return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+    }
+    @java.lang.Override
+    public Builder toBuilder() {
+      return this == DEFAULT_INSTANCE
+          ? new Builder() : new Builder().mergeFrom(this);
+    }
+
+    @java.lang.Override
+    protected Builder newBuilderForType(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      Builder builder = new Builder(parent);
+      return builder;
+    }
+    /**
+     * Protobuf type {@code tensorflow.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto}
+     */
+    public static final class Builder extends
+        com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+        // @@protoc_insertion_point(builder_implements:tensorflow.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto)
+        org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProtoOrBuilder {
+      public static final com.google.protobuf.Descriptors.Descriptor
+          getDescriptor() {
+        return org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfoProto.internal_static_tensorflow_RemoteFusedGraphExecuteInfo_TensorShapeTypeProto_descriptor;
+      }
+
+      @java.lang.Override
+      protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+          internalGetFieldAccessorTable() {
+        return org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfoProto.internal_static_tensorflow_RemoteFusedGraphExecuteInfo_TensorShapeTypeProto_fieldAccessorTable
+            .ensureFieldAccessorsInitialized(
+                org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto.class, org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto.Builder.class);
+      }
+
+      // Construct using org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto.newBuilder()
+      private Builder() {
+        maybeForceBuilderInitialization();
+      }
+
+      private Builder(
+          com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+        super(parent);
+        maybeForceBuilderInitialization();
+      }
+      private void maybeForceBuilderInitialization() {
+        if (com.google.protobuf.GeneratedMessageV3
+                .alwaysUseFieldBuilders) {
+        }
+      }
+      @java.lang.Override
+      public Builder clear() {
+        super.clear();
+        dtype_ = 0;
+
+        if (shapeBuilder_ == null) {
+          shape_ = null;
+        } else {
+          shape_ = null;
+          shapeBuilder_ = null;
+        }
+        return this;
+      }
+
+      @java.lang.Override
+      public com.google.protobuf.Descriptors.Descriptor
+          getDescriptorForType() {
+        return org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfoProto.internal_static_tensorflow_RemoteFusedGraphExecuteInfo_TensorShapeTypeProto_descriptor;
+      }
+
+      @java.lang.Override
+      public org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto getDefaultInstanceForType() {
+        return org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto.getDefaultInstance();
+      }
+
+      @java.lang.Override
+      public org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto build() {
+        org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto result = buildPartial();
+        if (!result.isInitialized()) {
+          throw newUninitializedMessageException(result);
+        }
+        return result;
+      }
+
+      @java.lang.Override
+      public org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto buildPartial() {
+        org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto result = new org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto(this);
+        result.dtype_ = dtype_;
+        if (shapeBuilder_ == null) {
+          result.shape_ = shape_;
+        } else {
+          result.shape_ = shapeBuilder_.build();
+        }
+        onBuilt();
+        return result;
+      }
+
+      @java.lang.Override
+      public Builder clone() {
+        return super.clone();
+      }
+      @java.lang.Override
+      public Builder setField(
+          com.google.protobuf.Descriptors.FieldDescriptor field,
+          java.lang.Object value) {
+        return super.setField(field, value);
+      }
+      @java.lang.Override
+      public Builder clearField(
+          com.google.protobuf.Descriptors.FieldDescriptor field) {
+        return super.clearField(field);
+      }
+      @java.lang.Override
+      public Builder clearOneof(
+          com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+        return super.clearOneof(oneof);
+      }
+      @java.lang.Override
+      public Builder setRepeatedField(
+          com.google.protobuf.Descriptors.FieldDescriptor field,
+          int index, java.lang.Object value) {
+        return super.setRepeatedField(field, index, value);
+      }
+      @java.lang.Override
+      public Builder addRepeatedField(
+          com.google.protobuf.Descriptors.FieldDescriptor field,
+          java.lang.Object value) {
+        return super.addRepeatedField(field, value);
+      }
+      @java.lang.Override
+      public Builder mergeFrom(com.google.protobuf.Message other) {
+        if (other instanceof org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto) {
+          return mergeFrom((org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto)other);
+        } else {
+          super.mergeFrom(other);
+          return this;
+        }
+      }
+
+      public Builder mergeFrom(org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto other) {
+        if (other == org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto.getDefaultInstance()) return this;
+        if (other.dtype_ != 0) {
+          setDtypeValue(other.getDtypeValue());
+        }
+        if (other.hasShape()) {
+          mergeShape(other.getShape());
+        }
+        this.mergeUnknownFields(other.unknownFields);
+        onChanged();
+        return this;
+      }
+
+      @java.lang.Override
+      public final boolean isInitialized() {
+        return true;
+      }
+
+      @java.lang.Override
+      public Builder mergeFrom(
+          com.google.protobuf.CodedInputStream input,
+          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+          throws java.io.IOException {
+        org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto parsedMessage = null;
+        try {
+          parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+        } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+          parsedMessage = (org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto) e.getUnfinishedMessage();
+          throw e.unwrapIOException();
+        } finally {
+          if (parsedMessage != null) {
+            mergeFrom(parsedMessage);
+          }
+        }
+        return this;
+      }
+
+      private int dtype_ = 0;
+      /**
+       * <code>.tensorflow.DataType dtype = 1;</code>
+       */
+      public int getDtypeValue() {
+        return dtype_;
+      }
+      /**
+       * <code>.tensorflow.DataType dtype = 1;</code>
+       */
+      public Builder setDtypeValue(int value) {
+        dtype_ = value;
+        onChanged();
+        return this;
+      }
+      /**
+       * <code>.tensorflow.DataType dtype = 1;</code>
+       */
+      public org.tensorflow.proto.framework.DataType getDtype() {
+        @SuppressWarnings("deprecation")
+        org.tensorflow.proto.framework.DataType result = org.tensorflow.proto.framework.DataType.valueOf(dtype_);
+        return result == null ? org.tensorflow.proto.framework.DataType.UNRECOGNIZED : result;
+      }
+      /**
+       * <code>.tensorflow.DataType dtype = 1;</code>
+       */
+      public Builder setDtype(org.tensorflow.proto.framework.DataType value) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        
+        dtype_ = value.getNumber();
+        onChanged();
+        return this;
+      }
+      /**
+       * <code>.tensorflow.DataType dtype = 1;</code>
+       */
+      public Builder clearDtype() {
+        
+        dtype_ = 0;
+        onChanged();
+        return this;
+      }
+
+      private org.tensorflow.proto.framework.TensorShapeProto shape_;
+      private com.google.protobuf.SingleFieldBuilderV3<
+          org.tensorflow.proto.framework.TensorShapeProto, org.tensorflow.proto.framework.TensorShapeProto.Builder, org.tensorflow.proto.framework.TensorShapeProtoOrBuilder> shapeBuilder_;
+      /**
+       * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+       */
+      public boolean hasShape() {
+        return shapeBuilder_ != null || shape_ != null;
+      }
+      /**
+       * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+       */
+      public org.tensorflow.proto.framework.TensorShapeProto getShape() {
+        if (shapeBuilder_ == null) {
+          return shape_ == null ? org.tensorflow.proto.framework.TensorShapeProto.getDefaultInstance() : shape_;
+        } else {
+          return shapeBuilder_.getMessage();
+        }
+      }
+      /**
+       * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+       */
+      public Builder setShape(org.tensorflow.proto.framework.TensorShapeProto value) {
+        if (shapeBuilder_ == null) {
+          if (value == null) {
+            throw new NullPointerException();
+          }
+          shape_ = value;
+          onChanged();
+        } else {
+          shapeBuilder_.setMessage(value);
+        }
+
+        return this;
+      }
+      /**
+       * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+       */
+      public Builder setShape(
+          org.tensorflow.proto.framework.TensorShapeProto.Builder builderForValue) {
+        if (shapeBuilder_ == null) {
+          shape_ = builderForValue.build();
+          onChanged();
+        } else {
+          shapeBuilder_.setMessage(builderForValue.build());
+        }
+
+        return this;
+      }
+      /**
+       * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+       */
+      public Builder mergeShape(org.tensorflow.proto.framework.TensorShapeProto value) {
+        if (shapeBuilder_ == null) {
+          if (shape_ != null) {
+            shape_ =
+              org.tensorflow.proto.framework.TensorShapeProto.newBuilder(shape_).mergeFrom(value).buildPartial();
+          } else {
+            shape_ = value;
+          }
+          onChanged();
+        } else {
+          shapeBuilder_.mergeFrom(value);
+        }
+
+        return this;
+      }
+      /**
+       * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+       */
+      public Builder clearShape() {
+        if (shapeBuilder_ == null) {
+          shape_ = null;
+          onChanged();
+        } else {
+          shape_ = null;
+          shapeBuilder_ = null;
+        }
+
+        return this;
+      }
+      /**
+       * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+       */
+      public org.tensorflow.proto.framework.TensorShapeProto.Builder getShapeBuilder() {
+        
+        onChanged();
+        return getShapeFieldBuilder().getBuilder();
+      }
+      /**
+       * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+       */
+      public org.tensorflow.proto.framework.TensorShapeProtoOrBuilder getShapeOrBuilder() {
+        if (shapeBuilder_ != null) {
+          return shapeBuilder_.getMessageOrBuilder();
+        } else {
+          return shape_ == null ?
+              org.tensorflow.proto.framework.TensorShapeProto.getDefaultInstance() : shape_;
+        }
+      }
+      /**
+       * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+       */
+      private com.google.protobuf.SingleFieldBuilderV3<
+          org.tensorflow.proto.framework.TensorShapeProto, org.tensorflow.proto.framework.TensorShapeProto.Builder, org.tensorflow.proto.framework.TensorShapeProtoOrBuilder> 
+          getShapeFieldBuilder() {
+        if (shapeBuilder_ == null) {
+          shapeBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+              org.tensorflow.proto.framework.TensorShapeProto, org.tensorflow.proto.framework.TensorShapeProto.Builder, org.tensorflow.proto.framework.TensorShapeProtoOrBuilder>(
+                  getShape(),
+                  getParentForChildren(),
+                  isClean());
+          shape_ = null;
+        }
+        return shapeBuilder_;
+      }
+      @java.lang.Override
+      public final Builder setUnknownFields(
+          final com.google.protobuf.UnknownFieldSet unknownFields) {
+        return super.setUnknownFields(unknownFields);
+      }
+
+      @java.lang.Override
+      public final Builder mergeUnknownFields(
+          final com.google.protobuf.UnknownFieldSet unknownFields) {
+        return super.mergeUnknownFields(unknownFields);
+      }
+
+
+      // @@protoc_insertion_point(builder_scope:tensorflow.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto)
+    }
+
+    // @@protoc_insertion_point(class_scope:tensorflow.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto)
+    private static final org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto DEFAULT_INSTANCE;
+    static {
+      DEFAULT_INSTANCE = new org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto();
+    }
+
+    public static org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto getDefaultInstance() {
+      return DEFAULT_INSTANCE;
+    }
+
+    private static final com.google.protobuf.Parser<TensorShapeTypeProto>
+        PARSER = new com.google.protobuf.AbstractParser<TensorShapeTypeProto>() {
+      @java.lang.Override
+      public TensorShapeTypeProto parsePartialFrom(
+          com.google.protobuf.CodedInputStream input,
+          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+          throws com.google.protobuf.InvalidProtocolBufferException {
+        return new TensorShapeTypeProto(input, extensionRegistry);
+      }
+    };
+
+    public static com.google.protobuf.Parser<TensorShapeTypeProto> parser() {
+      return PARSER;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Parser<TensorShapeTypeProto> getParserForType() {
+      return PARSER;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto getDefaultInstanceForType() {
+      return DEFAULT_INSTANCE;
+    }
+
+  }
+
+  public static final int REMOTE_GRAPH_FIELD_NUMBER = 1;
+  private org.tensorflow.proto.framework.GraphDef remoteGraph_;
+  /**
+   * <pre>
+   * Definition of remote graph
+   * </pre>
+   *
+   * <code>.tensorflow.GraphDef remote_graph = 1;</code>
+   */
+  public boolean hasRemoteGraph() {
+    return remoteGraph_ != null;
+  }
+  /**
+   * <pre>
+   * Definition of remote graph
+   * </pre>
+   *
+   * <code>.tensorflow.GraphDef remote_graph = 1;</code>
+   */
+  public org.tensorflow.proto.framework.GraphDef getRemoteGraph() {
+    return remoteGraph_ == null ? org.tensorflow.proto.framework.GraphDef.getDefaultInstance() : remoteGraph_;
+  }
+  /**
+   * <pre>
+   * Definition of remote graph
+   * </pre>
+   *
+   * <code>.tensorflow.GraphDef remote_graph = 1;</code>
+   */
+  public org.tensorflow.proto.framework.GraphDefOrBuilder getRemoteGraphOrBuilder() {
+    return getRemoteGraph();
+  }
+
+  public static final int GRAPH_INPUT_NODE_NAME_FIELD_NUMBER = 2;
+  private com.google.protobuf.LazyStringList graphInputNodeName_;
+  /**
+   * <pre>
+   * Remote fused graph input node name
+   * </pre>
+   *
+   * <code>repeated string graph_input_node_name = 2;</code>
+   */
+  public com.google.protobuf.ProtocolStringList
+      getGraphInputNodeNameList() {
+    return graphInputNodeName_;
+  }
+  /**
+   * <pre>
+   * Remote fused graph input node name
+   * </pre>
+   *
+   * <code>repeated string graph_input_node_name = 2;</code>
+   */
+  public int getGraphInputNodeNameCount() {
+    return graphInputNodeName_.size();
+  }
+  /**
+   * <pre>
+   * Remote fused graph input node name
+   * </pre>
+   *
+   * <code>repeated string graph_input_node_name = 2;</code>
+   */
+  public java.lang.String getGraphInputNodeName(int index) {
+    return graphInputNodeName_.get(index);
+  }
+  /**
+   * <pre>
+   * Remote fused graph input node name
+   * </pre>
+   *
+   * <code>repeated string graph_input_node_name = 2;</code>
+   */
+  public com.google.protobuf.ByteString
+      getGraphInputNodeNameBytes(int index) {
+    return graphInputNodeName_.getByteString(index);
+  }
+
+  public static final int GRAPH_OUTPUT_NODE_NAME_FIELD_NUMBER = 3;
+  private com.google.protobuf.LazyStringList graphOutputNodeName_;
+  /**
+   * <pre>
+   * Remote fused graph output node name
+   * </pre>
+   *
+   * <code>repeated string graph_output_node_name = 3;</code>
+   */
+  public com.google.protobuf.ProtocolStringList
+      getGraphOutputNodeNameList() {
+    return graphOutputNodeName_;
+  }
+  /**
+   * <pre>
+   * Remote fused graph output node name
+   * </pre>
+   *
+   * <code>repeated string graph_output_node_name = 3;</code>
+   */
+  public int getGraphOutputNodeNameCount() {
+    return graphOutputNodeName_.size();
+  }
+  /**
+   * <pre>
+   * Remote fused graph output node name
+   * </pre>
+   *
+   * <code>repeated string graph_output_node_name = 3;</code>
+   */
+  public java.lang.String getGraphOutputNodeName(int index) {
+    return graphOutputNodeName_.get(index);
+  }
+  /**
+   * <pre>
+   * Remote fused graph output node name
+   * </pre>
+   *
+   * <code>repeated string graph_output_node_name = 3;</code>
+   */
+  public com.google.protobuf.ByteString
+      getGraphOutputNodeNameBytes(int index) {
+    return graphOutputNodeName_.getByteString(index);
+  }
+
+  public static final int EXECUTOR_NAME_FIELD_NUMBER = 4;
+  private volatile java.lang.Object executorName_;
+  /**
+   * <pre>
+   * Executor's name
+   * </pre>
+   *
+   * <code>string executor_name = 4;</code>
+   */
+  public java.lang.String getExecutorName() {
+    java.lang.Object ref = executorName_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      executorName_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * Executor's name
+   * </pre>
+   *
+   * <code>string executor_name = 4;</code>
+   */
+  public com.google.protobuf.ByteString
+      getExecutorNameBytes() {
+    java.lang.Object ref = executorName_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      executorName_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int SERIALIZED_EXECUTOR_PARAMETERS_FIELD_NUMBER = 5;
+  private com.google.protobuf.ByteString serializedExecutorParameters_;
+  /**
+   * <pre>
+   * Optional: Parameters given to the executor
+   * </pre>
+   *
+   * <code>bytes serialized_executor_parameters = 5;</code>
+   */
+  public com.google.protobuf.ByteString getSerializedExecutorParameters() {
+    return serializedExecutorParameters_;
+  }
+
+  public static final int DEFAULT_GRAPH_INPUT_TENSOR_SHAPE_FIELD_NUMBER = 6;
+  private java.util.List<org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto> defaultGraphInputTensorShape_;
+  /**
+   * <pre>
+   * Optional: Default graph input tensor shape used to allocate memory
+   * before executing op
+   * </pre>
+   *
+   * <code>repeated .tensorflow.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto default_graph_input_tensor_shape = 6;</code>
+   */
+  public java.util.List<org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto> getDefaultGraphInputTensorShapeList() {
+    return defaultGraphInputTensorShape_;
+  }
+  /**
+   * <pre>
+   * Optional: Default graph input tensor shape used to allocate memory
+   * before executing op
+   * </pre>
+   *
+   * <code>repeated .tensorflow.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto default_graph_input_tensor_shape = 6;</code>
+   */
+  public java.util.List<? extends org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProtoOrBuilder> 
+      getDefaultGraphInputTensorShapeOrBuilderList() {
+    return defaultGraphInputTensorShape_;
+  }
+  /**
+   * <pre>
+   * Optional: Default graph input tensor shape used to allocate memory
+   * before executing op
+   * </pre>
+   *
+   * <code>repeated .tensorflow.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto default_graph_input_tensor_shape = 6;</code>
+   */
+  public int getDefaultGraphInputTensorShapeCount() {
+    return defaultGraphInputTensorShape_.size();
+  }
+  /**
+   * <pre>
+   * Optional: Default graph input tensor shape used to allocate memory
+   * before executing op
+   * </pre>
+   *
+   * <code>repeated .tensorflow.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto default_graph_input_tensor_shape = 6;</code>
+   */
+  public org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto getDefaultGraphInputTensorShape(int index) {
+    return defaultGraphInputTensorShape_.get(index);
+  }
+  /**
+   * <pre>
+   * Optional: Default graph input tensor shape used to allocate memory
+   * before executing op
+   * </pre>
+   *
+   * <code>repeated .tensorflow.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto default_graph_input_tensor_shape = 6;</code>
+   */
+  public org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProtoOrBuilder getDefaultGraphInputTensorShapeOrBuilder(
+      int index) {
+    return defaultGraphInputTensorShape_.get(index);
+  }
+
+  public static final int DEFAULT_GRAPH_OUTPUT_TENSOR_SHAPE_FIELD_NUMBER = 7;
+  private java.util.List<org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto> defaultGraphOutputTensorShape_;
+  /**
+   * <pre>
+   * Optional: Default graph input tensor shape used to allocate memory
+   * before executing op
+   * TODO(satok): Remote output tensor shape once shape information is stored
+   * in NodeDef
+   * </pre>
+   *
+   * <code>repeated .tensorflow.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto default_graph_output_tensor_shape = 7;</code>
+   */
+  public java.util.List<org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto> getDefaultGraphOutputTensorShapeList() {
+    return defaultGraphOutputTensorShape_;
+  }
+  /**
+   * <pre>
+   * Optional: Default graph input tensor shape used to allocate memory
+   * before executing op
+   * TODO(satok): Remote output tensor shape once shape information is stored
+   * in NodeDef
+   * </pre>
+   *
+   * <code>repeated .tensorflow.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto default_graph_output_tensor_shape = 7;</code>
+   */
+  public java.util.List<? extends org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProtoOrBuilder> 
+      getDefaultGraphOutputTensorShapeOrBuilderList() {
+    return defaultGraphOutputTensorShape_;
+  }
+  /**
+   * <pre>
+   * Optional: Default graph input tensor shape used to allocate memory
+   * before executing op
+   * TODO(satok): Remote output tensor shape once shape information is stored
+   * in NodeDef
+   * </pre>
+   *
+   * <code>repeated .tensorflow.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto default_graph_output_tensor_shape = 7;</code>
+   */
+  public int getDefaultGraphOutputTensorShapeCount() {
+    return defaultGraphOutputTensorShape_.size();
+  }
+  /**
+   * <pre>
+   * Optional: Default graph input tensor shape used to allocate memory
+   * before executing op
+   * TODO(satok): Remote output tensor shape once shape information is stored
+   * in NodeDef
+   * </pre>
+   *
+   * <code>repeated .tensorflow.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto default_graph_output_tensor_shape = 7;</code>
+   */
+  public org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto getDefaultGraphOutputTensorShape(int index) {
+    return defaultGraphOutputTensorShape_.get(index);
+  }
+  /**
+   * <pre>
+   * Optional: Default graph input tensor shape used to allocate memory
+   * before executing op
+   * TODO(satok): Remote output tensor shape once shape information is stored
+   * in NodeDef
+   * </pre>
+   *
+   * <code>repeated .tensorflow.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto default_graph_output_tensor_shape = 7;</code>
+   */
+  public org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProtoOrBuilder getDefaultGraphOutputTensorShapeOrBuilder(
+      int index) {
+    return defaultGraphOutputTensorShape_.get(index);
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    if (remoteGraph_ != null) {
+      output.writeMessage(1, getRemoteGraph());
+    }
+    for (int i = 0; i < graphInputNodeName_.size(); i++) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 2, graphInputNodeName_.getRaw(i));
+    }
+    for (int i = 0; i < graphOutputNodeName_.size(); i++) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 3, graphOutputNodeName_.getRaw(i));
+    }
+    if (!getExecutorNameBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 4, executorName_);
+    }
+    if (!serializedExecutorParameters_.isEmpty()) {
+      output.writeBytes(5, serializedExecutorParameters_);
+    }
+    for (int i = 0; i < defaultGraphInputTensorShape_.size(); i++) {
+      output.writeMessage(6, defaultGraphInputTensorShape_.get(i));
+    }
+    for (int i = 0; i < defaultGraphOutputTensorShape_.size(); i++) {
+      output.writeMessage(7, defaultGraphOutputTensorShape_.get(i));
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (remoteGraph_ != null) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(1, getRemoteGraph());
+    }
+    {
+      int dataSize = 0;
+      for (int i = 0; i < graphInputNodeName_.size(); i++) {
+        dataSize += computeStringSizeNoTag(graphInputNodeName_.getRaw(i));
+      }
+      size += dataSize;
+      size += 1 * getGraphInputNodeNameList().size();
+    }
+    {
+      int dataSize = 0;
+      for (int i = 0; i < graphOutputNodeName_.size(); i++) {
+        dataSize += computeStringSizeNoTag(graphOutputNodeName_.getRaw(i));
+      }
+      size += dataSize;
+      size += 1 * getGraphOutputNodeNameList().size();
+    }
+    if (!getExecutorNameBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(4, executorName_);
+    }
+    if (!serializedExecutorParameters_.isEmpty()) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeBytesSize(5, serializedExecutorParameters_);
+    }
+    for (int i = 0; i < defaultGraphInputTensorShape_.size(); i++) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(6, defaultGraphInputTensorShape_.get(i));
+    }
+    for (int i = 0; i < defaultGraphOutputTensorShape_.size(); i++) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(7, defaultGraphOutputTensorShape_.get(i));
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo other = (org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo) obj;
+
+    if (hasRemoteGraph() != other.hasRemoteGraph()) return false;
+    if (hasRemoteGraph()) {
+      if (!getRemoteGraph()
+          .equals(other.getRemoteGraph())) return false;
+    }
+    if (!getGraphInputNodeNameList()
+        .equals(other.getGraphInputNodeNameList())) return false;
+    if (!getGraphOutputNodeNameList()
+        .equals(other.getGraphOutputNodeNameList())) return false;
+    if (!getExecutorName()
+        .equals(other.getExecutorName())) return false;
+    if (!getSerializedExecutorParameters()
+        .equals(other.getSerializedExecutorParameters())) return false;
+    if (!getDefaultGraphInputTensorShapeList()
+        .equals(other.getDefaultGraphInputTensorShapeList())) return false;
+    if (!getDefaultGraphOutputTensorShapeList()
+        .equals(other.getDefaultGraphOutputTensorShapeList())) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    if (hasRemoteGraph()) {
+      hash = (37 * hash) + REMOTE_GRAPH_FIELD_NUMBER;
+      hash = (53 * hash) + getRemoteGraph().hashCode();
+    }
+    if (getGraphInputNodeNameCount() > 0) {
+      hash = (37 * hash) + GRAPH_INPUT_NODE_NAME_FIELD_NUMBER;
+      hash = (53 * hash) + getGraphInputNodeNameList().hashCode();
+    }
+    if (getGraphOutputNodeNameCount() > 0) {
+      hash = (37 * hash) + GRAPH_OUTPUT_NODE_NAME_FIELD_NUMBER;
+      hash = (53 * hash) + getGraphOutputNodeNameList().hashCode();
+    }
+    hash = (37 * hash) + EXECUTOR_NAME_FIELD_NUMBER;
+    hash = (53 * hash) + getExecutorName().hashCode();
+    hash = (37 * hash) + SERIALIZED_EXECUTOR_PARAMETERS_FIELD_NUMBER;
+    hash = (53 * hash) + getSerializedExecutorParameters().hashCode();
+    if (getDefaultGraphInputTensorShapeCount() > 0) {
+      hash = (37 * hash) + DEFAULT_GRAPH_INPUT_TENSOR_SHAPE_FIELD_NUMBER;
+      hash = (53 * hash) + getDefaultGraphInputTensorShapeList().hashCode();
+    }
+    if (getDefaultGraphOutputTensorShapeCount() > 0) {
+      hash = (37 * hash) + DEFAULT_GRAPH_OUTPUT_TENSOR_SHAPE_FIELD_NUMBER;
+      hash = (53 * hash) + getDefaultGraphOutputTensorShapeList().hashCode();
+    }
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * <pre>
+   * Protocol buffer representing a handle to a tensorflow resource. Handles are
+   * not valid across executions, but can be serialized back and forth from within
+   * a single run.
+   * </pre>
+   *
+   * Protobuf type {@code tensorflow.RemoteFusedGraphExecuteInfo}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.RemoteFusedGraphExecuteInfo)
+      org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfoOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfoProto.internal_static_tensorflow_RemoteFusedGraphExecuteInfo_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfoProto.internal_static_tensorflow_RemoteFusedGraphExecuteInfo_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.class, org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+        getDefaultGraphInputTensorShapeFieldBuilder();
+        getDefaultGraphOutputTensorShapeFieldBuilder();
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      if (remoteGraphBuilder_ == null) {
+        remoteGraph_ = null;
+      } else {
+        remoteGraph_ = null;
+        remoteGraphBuilder_ = null;
+      }
+      graphInputNodeName_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+      bitField0_ = (bitField0_ & ~0x00000001);
+      graphOutputNodeName_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+      bitField0_ = (bitField0_ & ~0x00000002);
+      executorName_ = "";
+
+      serializedExecutorParameters_ = com.google.protobuf.ByteString.EMPTY;
+
+      if (defaultGraphInputTensorShapeBuilder_ == null) {
+        defaultGraphInputTensorShape_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000004);
+      } else {
+        defaultGraphInputTensorShapeBuilder_.clear();
+      }
+      if (defaultGraphOutputTensorShapeBuilder_ == null) {
+        defaultGraphOutputTensorShape_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000008);
+      } else {
+        defaultGraphOutputTensorShapeBuilder_.clear();
+      }
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfoProto.internal_static_tensorflow_RemoteFusedGraphExecuteInfo_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo getDefaultInstanceForType() {
+      return org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo build() {
+      org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo buildPartial() {
+      org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo result = new org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo(this);
+      int from_bitField0_ = bitField0_;
+      if (remoteGraphBuilder_ == null) {
+        result.remoteGraph_ = remoteGraph_;
+      } else {
+        result.remoteGraph_ = remoteGraphBuilder_.build();
+      }
+      if (((bitField0_ & 0x00000001) != 0)) {
+        graphInputNodeName_ = graphInputNodeName_.getUnmodifiableView();
+        bitField0_ = (bitField0_ & ~0x00000001);
+      }
+      result.graphInputNodeName_ = graphInputNodeName_;
+      if (((bitField0_ & 0x00000002) != 0)) {
+        graphOutputNodeName_ = graphOutputNodeName_.getUnmodifiableView();
+        bitField0_ = (bitField0_ & ~0x00000002);
+      }
+      result.graphOutputNodeName_ = graphOutputNodeName_;
+      result.executorName_ = executorName_;
+      result.serializedExecutorParameters_ = serializedExecutorParameters_;
+      if (defaultGraphInputTensorShapeBuilder_ == null) {
+        if (((bitField0_ & 0x00000004) != 0)) {
+          defaultGraphInputTensorShape_ = java.util.Collections.unmodifiableList(defaultGraphInputTensorShape_);
+          bitField0_ = (bitField0_ & ~0x00000004);
+        }
+        result.defaultGraphInputTensorShape_ = defaultGraphInputTensorShape_;
+      } else {
+        result.defaultGraphInputTensorShape_ = defaultGraphInputTensorShapeBuilder_.build();
+      }
+      if (defaultGraphOutputTensorShapeBuilder_ == null) {
+        if (((bitField0_ & 0x00000008) != 0)) {
+          defaultGraphOutputTensorShape_ = java.util.Collections.unmodifiableList(defaultGraphOutputTensorShape_);
+          bitField0_ = (bitField0_ & ~0x00000008);
+        }
+        result.defaultGraphOutputTensorShape_ = defaultGraphOutputTensorShape_;
+      } else {
+        result.defaultGraphOutputTensorShape_ = defaultGraphOutputTensorShapeBuilder_.build();
+      }
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo) {
+        return mergeFrom((org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo other) {
+      if (other == org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.getDefaultInstance()) return this;
+      if (other.hasRemoteGraph()) {
+        mergeRemoteGraph(other.getRemoteGraph());
+      }
+      if (!other.graphInputNodeName_.isEmpty()) {
+        if (graphInputNodeName_.isEmpty()) {
+          graphInputNodeName_ = other.graphInputNodeName_;
+          bitField0_ = (bitField0_ & ~0x00000001);
+        } else {
+          ensureGraphInputNodeNameIsMutable();
+          graphInputNodeName_.addAll(other.graphInputNodeName_);
+        }
+        onChanged();
+      }
+      if (!other.graphOutputNodeName_.isEmpty()) {
+        if (graphOutputNodeName_.isEmpty()) {
+          graphOutputNodeName_ = other.graphOutputNodeName_;
+          bitField0_ = (bitField0_ & ~0x00000002);
+        } else {
+          ensureGraphOutputNodeNameIsMutable();
+          graphOutputNodeName_.addAll(other.graphOutputNodeName_);
+        }
+        onChanged();
+      }
+      if (!other.getExecutorName().isEmpty()) {
+        executorName_ = other.executorName_;
+        onChanged();
+      }
+      if (other.getSerializedExecutorParameters() != com.google.protobuf.ByteString.EMPTY) {
+        setSerializedExecutorParameters(other.getSerializedExecutorParameters());
+      }
+      if (defaultGraphInputTensorShapeBuilder_ == null) {
+        if (!other.defaultGraphInputTensorShape_.isEmpty()) {
+          if (defaultGraphInputTensorShape_.isEmpty()) {
+            defaultGraphInputTensorShape_ = other.defaultGraphInputTensorShape_;
+            bitField0_ = (bitField0_ & ~0x00000004);
+          } else {
+            ensureDefaultGraphInputTensorShapeIsMutable();
+            defaultGraphInputTensorShape_.addAll(other.defaultGraphInputTensorShape_);
+          }
+          onChanged();
+        }
+      } else {
+        if (!other.defaultGraphInputTensorShape_.isEmpty()) {
+          if (defaultGraphInputTensorShapeBuilder_.isEmpty()) {
+            defaultGraphInputTensorShapeBuilder_.dispose();
+            defaultGraphInputTensorShapeBuilder_ = null;
+            defaultGraphInputTensorShape_ = other.defaultGraphInputTensorShape_;
+            bitField0_ = (bitField0_ & ~0x00000004);
+            defaultGraphInputTensorShapeBuilder_ = 
+              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
+                 getDefaultGraphInputTensorShapeFieldBuilder() : null;
+          } else {
+            defaultGraphInputTensorShapeBuilder_.addAllMessages(other.defaultGraphInputTensorShape_);
+          }
+        }
+      }
+      if (defaultGraphOutputTensorShapeBuilder_ == null) {
+        if (!other.defaultGraphOutputTensorShape_.isEmpty()) {
+          if (defaultGraphOutputTensorShape_.isEmpty()) {
+            defaultGraphOutputTensorShape_ = other.defaultGraphOutputTensorShape_;
+            bitField0_ = (bitField0_ & ~0x00000008);
+          } else {
+            ensureDefaultGraphOutputTensorShapeIsMutable();
+            defaultGraphOutputTensorShape_.addAll(other.defaultGraphOutputTensorShape_);
+          }
+          onChanged();
+        }
+      } else {
+        if (!other.defaultGraphOutputTensorShape_.isEmpty()) {
+          if (defaultGraphOutputTensorShapeBuilder_.isEmpty()) {
+            defaultGraphOutputTensorShapeBuilder_.dispose();
+            defaultGraphOutputTensorShapeBuilder_ = null;
+            defaultGraphOutputTensorShape_ = other.defaultGraphOutputTensorShape_;
+            bitField0_ = (bitField0_ & ~0x00000008);
+            defaultGraphOutputTensorShapeBuilder_ = 
+              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
+                 getDefaultGraphOutputTensorShapeFieldBuilder() : null;
+          } else {
+            defaultGraphOutputTensorShapeBuilder_.addAllMessages(other.defaultGraphOutputTensorShape_);
+          }
+        }
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+    private int bitField0_;
+
+    private org.tensorflow.proto.framework.GraphDef remoteGraph_;
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.GraphDef, org.tensorflow.proto.framework.GraphDef.Builder, org.tensorflow.proto.framework.GraphDefOrBuilder> remoteGraphBuilder_;
+    /**
+     * <pre>
+     * Definition of remote graph
+     * </pre>
+     *
+     * <code>.tensorflow.GraphDef remote_graph = 1;</code>
+     */
+    public boolean hasRemoteGraph() {
+      return remoteGraphBuilder_ != null || remoteGraph_ != null;
+    }
+    /**
+     * <pre>
+     * Definition of remote graph
+     * </pre>
+     *
+     * <code>.tensorflow.GraphDef remote_graph = 1;</code>
+     */
+    public org.tensorflow.proto.framework.GraphDef getRemoteGraph() {
+      if (remoteGraphBuilder_ == null) {
+        return remoteGraph_ == null ? org.tensorflow.proto.framework.GraphDef.getDefaultInstance() : remoteGraph_;
+      } else {
+        return remoteGraphBuilder_.getMessage();
+      }
+    }
+    /**
+     * <pre>
+     * Definition of remote graph
+     * </pre>
+     *
+     * <code>.tensorflow.GraphDef remote_graph = 1;</code>
+     */
+    public Builder setRemoteGraph(org.tensorflow.proto.framework.GraphDef value) {
+      if (remoteGraphBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        remoteGraph_ = value;
+        onChanged();
+      } else {
+        remoteGraphBuilder_.setMessage(value);
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Definition of remote graph
+     * </pre>
+     *
+     * <code>.tensorflow.GraphDef remote_graph = 1;</code>
+     */
+    public Builder setRemoteGraph(
+        org.tensorflow.proto.framework.GraphDef.Builder builderForValue) {
+      if (remoteGraphBuilder_ == null) {
+        remoteGraph_ = builderForValue.build();
+        onChanged();
+      } else {
+        remoteGraphBuilder_.setMessage(builderForValue.build());
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Definition of remote graph
+     * </pre>
+     *
+     * <code>.tensorflow.GraphDef remote_graph = 1;</code>
+     */
+    public Builder mergeRemoteGraph(org.tensorflow.proto.framework.GraphDef value) {
+      if (remoteGraphBuilder_ == null) {
+        if (remoteGraph_ != null) {
+          remoteGraph_ =
+            org.tensorflow.proto.framework.GraphDef.newBuilder(remoteGraph_).mergeFrom(value).buildPartial();
+        } else {
+          remoteGraph_ = value;
+        }
+        onChanged();
+      } else {
+        remoteGraphBuilder_.mergeFrom(value);
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Definition of remote graph
+     * </pre>
+     *
+     * <code>.tensorflow.GraphDef remote_graph = 1;</code>
+     */
+    public Builder clearRemoteGraph() {
+      if (remoteGraphBuilder_ == null) {
+        remoteGraph_ = null;
+        onChanged();
+      } else {
+        remoteGraph_ = null;
+        remoteGraphBuilder_ = null;
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Definition of remote graph
+     * </pre>
+     *
+     * <code>.tensorflow.GraphDef remote_graph = 1;</code>
+     */
+    public org.tensorflow.proto.framework.GraphDef.Builder getRemoteGraphBuilder() {
+      
+      onChanged();
+      return getRemoteGraphFieldBuilder().getBuilder();
+    }
+    /**
+     * <pre>
+     * Definition of remote graph
+     * </pre>
+     *
+     * <code>.tensorflow.GraphDef remote_graph = 1;</code>
+     */
+    public org.tensorflow.proto.framework.GraphDefOrBuilder getRemoteGraphOrBuilder() {
+      if (remoteGraphBuilder_ != null) {
+        return remoteGraphBuilder_.getMessageOrBuilder();
+      } else {
+        return remoteGraph_ == null ?
+            org.tensorflow.proto.framework.GraphDef.getDefaultInstance() : remoteGraph_;
+      }
+    }
+    /**
+     * <pre>
+     * Definition of remote graph
+     * </pre>
+     *
+     * <code>.tensorflow.GraphDef remote_graph = 1;</code>
+     */
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.GraphDef, org.tensorflow.proto.framework.GraphDef.Builder, org.tensorflow.proto.framework.GraphDefOrBuilder> 
+        getRemoteGraphFieldBuilder() {
+      if (remoteGraphBuilder_ == null) {
+        remoteGraphBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+            org.tensorflow.proto.framework.GraphDef, org.tensorflow.proto.framework.GraphDef.Builder, org.tensorflow.proto.framework.GraphDefOrBuilder>(
+                getRemoteGraph(),
+                getParentForChildren(),
+                isClean());
+        remoteGraph_ = null;
+      }
+      return remoteGraphBuilder_;
+    }
+
+    private com.google.protobuf.LazyStringList graphInputNodeName_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+    private void ensureGraphInputNodeNameIsMutable() {
+      if (!((bitField0_ & 0x00000001) != 0)) {
+        graphInputNodeName_ = new com.google.protobuf.LazyStringArrayList(graphInputNodeName_);
+        bitField0_ |= 0x00000001;
+       }
+    }
+    /**
+     * <pre>
+     * Remote fused graph input node name
+     * </pre>
+     *
+     * <code>repeated string graph_input_node_name = 2;</code>
+     */
+    public com.google.protobuf.ProtocolStringList
+        getGraphInputNodeNameList() {
+      return graphInputNodeName_.getUnmodifiableView();
+    }
+    /**
+     * <pre>
+     * Remote fused graph input node name
+     * </pre>
+     *
+     * <code>repeated string graph_input_node_name = 2;</code>
+     */
+    public int getGraphInputNodeNameCount() {
+      return graphInputNodeName_.size();
+    }
+    /**
+     * <pre>
+     * Remote fused graph input node name
+     * </pre>
+     *
+     * <code>repeated string graph_input_node_name = 2;</code>
+     */
+    public java.lang.String getGraphInputNodeName(int index) {
+      return graphInputNodeName_.get(index);
+    }
+    /**
+     * <pre>
+     * Remote fused graph input node name
+     * </pre>
+     *
+     * <code>repeated string graph_input_node_name = 2;</code>
+     */
+    public com.google.protobuf.ByteString
+        getGraphInputNodeNameBytes(int index) {
+      return graphInputNodeName_.getByteString(index);
+    }
+    /**
+     * <pre>
+     * Remote fused graph input node name
+     * </pre>
+     *
+     * <code>repeated string graph_input_node_name = 2;</code>
+     */
+    public Builder setGraphInputNodeName(
+        int index, java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  ensureGraphInputNodeNameIsMutable();
+      graphInputNodeName_.set(index, value);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Remote fused graph input node name
+     * </pre>
+     *
+     * <code>repeated string graph_input_node_name = 2;</code>
+     */
+    public Builder addGraphInputNodeName(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  ensureGraphInputNodeNameIsMutable();
+      graphInputNodeName_.add(value);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Remote fused graph input node name
+     * </pre>
+     *
+     * <code>repeated string graph_input_node_name = 2;</code>
+     */
+    public Builder addAllGraphInputNodeName(
+        java.lang.Iterable<java.lang.String> values) {
+      ensureGraphInputNodeNameIsMutable();
+      com.google.protobuf.AbstractMessageLite.Builder.addAll(
+          values, graphInputNodeName_);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Remote fused graph input node name
+     * </pre>
+     *
+     * <code>repeated string graph_input_node_name = 2;</code>
+     */
+    public Builder clearGraphInputNodeName() {
+      graphInputNodeName_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+      bitField0_ = (bitField0_ & ~0x00000001);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Remote fused graph input node name
+     * </pre>
+     *
+     * <code>repeated string graph_input_node_name = 2;</code>
+     */
+    public Builder addGraphInputNodeNameBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      ensureGraphInputNodeNameIsMutable();
+      graphInputNodeName_.add(value);
+      onChanged();
+      return this;
+    }
+
+    private com.google.protobuf.LazyStringList graphOutputNodeName_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+    private void ensureGraphOutputNodeNameIsMutable() {
+      if (!((bitField0_ & 0x00000002) != 0)) {
+        graphOutputNodeName_ = new com.google.protobuf.LazyStringArrayList(graphOutputNodeName_);
+        bitField0_ |= 0x00000002;
+       }
+    }
+    /**
+     * <pre>
+     * Remote fused graph output node name
+     * </pre>
+     *
+     * <code>repeated string graph_output_node_name = 3;</code>
+     */
+    public com.google.protobuf.ProtocolStringList
+        getGraphOutputNodeNameList() {
+      return graphOutputNodeName_.getUnmodifiableView();
+    }
+    /**
+     * <pre>
+     * Remote fused graph output node name
+     * </pre>
+     *
+     * <code>repeated string graph_output_node_name = 3;</code>
+     */
+    public int getGraphOutputNodeNameCount() {
+      return graphOutputNodeName_.size();
+    }
+    /**
+     * <pre>
+     * Remote fused graph output node name
+     * </pre>
+     *
+     * <code>repeated string graph_output_node_name = 3;</code>
+     */
+    public java.lang.String getGraphOutputNodeName(int index) {
+      return graphOutputNodeName_.get(index);
+    }
+    /**
+     * <pre>
+     * Remote fused graph output node name
+     * </pre>
+     *
+     * <code>repeated string graph_output_node_name = 3;</code>
+     */
+    public com.google.protobuf.ByteString
+        getGraphOutputNodeNameBytes(int index) {
+      return graphOutputNodeName_.getByteString(index);
+    }
+    /**
+     * <pre>
+     * Remote fused graph output node name
+     * </pre>
+     *
+     * <code>repeated string graph_output_node_name = 3;</code>
+     */
+    public Builder setGraphOutputNodeName(
+        int index, java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  ensureGraphOutputNodeNameIsMutable();
+      graphOutputNodeName_.set(index, value);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Remote fused graph output node name
+     * </pre>
+     *
+     * <code>repeated string graph_output_node_name = 3;</code>
+     */
+    public Builder addGraphOutputNodeName(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  ensureGraphOutputNodeNameIsMutable();
+      graphOutputNodeName_.add(value);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Remote fused graph output node name
+     * </pre>
+     *
+     * <code>repeated string graph_output_node_name = 3;</code>
+     */
+    public Builder addAllGraphOutputNodeName(
+        java.lang.Iterable<java.lang.String> values) {
+      ensureGraphOutputNodeNameIsMutable();
+      com.google.protobuf.AbstractMessageLite.Builder.addAll(
+          values, graphOutputNodeName_);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Remote fused graph output node name
+     * </pre>
+     *
+     * <code>repeated string graph_output_node_name = 3;</code>
+     */
+    public Builder clearGraphOutputNodeName() {
+      graphOutputNodeName_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+      bitField0_ = (bitField0_ & ~0x00000002);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Remote fused graph output node name
+     * </pre>
+     *
+     * <code>repeated string graph_output_node_name = 3;</code>
+     */
+    public Builder addGraphOutputNodeNameBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      ensureGraphOutputNodeNameIsMutable();
+      graphOutputNodeName_.add(value);
+      onChanged();
+      return this;
+    }
+
+    private java.lang.Object executorName_ = "";
+    /**
+     * <pre>
+     * Executor's name
+     * </pre>
+     *
+     * <code>string executor_name = 4;</code>
+     */
+    public java.lang.String getExecutorName() {
+      java.lang.Object ref = executorName_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        executorName_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Executor's name
+     * </pre>
+     *
+     * <code>string executor_name = 4;</code>
+     */
+    public com.google.protobuf.ByteString
+        getExecutorNameBytes() {
+      java.lang.Object ref = executorName_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        executorName_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Executor's name
+     * </pre>
+     *
+     * <code>string executor_name = 4;</code>
+     */
+    public Builder setExecutorName(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      executorName_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Executor's name
+     * </pre>
+     *
+     * <code>string executor_name = 4;</code>
+     */
+    public Builder clearExecutorName() {
+      
+      executorName_ = getDefaultInstance().getExecutorName();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Executor's name
+     * </pre>
+     *
+     * <code>string executor_name = 4;</code>
+     */
+    public Builder setExecutorNameBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      executorName_ = value;
+      onChanged();
+      return this;
+    }
+
+    private com.google.protobuf.ByteString serializedExecutorParameters_ = com.google.protobuf.ByteString.EMPTY;
+    /**
+     * <pre>
+     * Optional: Parameters given to the executor
+     * </pre>
+     *
+     * <code>bytes serialized_executor_parameters = 5;</code>
+     */
+    public com.google.protobuf.ByteString getSerializedExecutorParameters() {
+      return serializedExecutorParameters_;
+    }
+    /**
+     * <pre>
+     * Optional: Parameters given to the executor
+     * </pre>
+     *
+     * <code>bytes serialized_executor_parameters = 5;</code>
+     */
+    public Builder setSerializedExecutorParameters(com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      serializedExecutorParameters_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Optional: Parameters given to the executor
+     * </pre>
+     *
+     * <code>bytes serialized_executor_parameters = 5;</code>
+     */
+    public Builder clearSerializedExecutorParameters() {
+      
+      serializedExecutorParameters_ = getDefaultInstance().getSerializedExecutorParameters();
+      onChanged();
+      return this;
+    }
+
+    private java.util.List<org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto> defaultGraphInputTensorShape_ =
+      java.util.Collections.emptyList();
+    private void ensureDefaultGraphInputTensorShapeIsMutable() {
+      if (!((bitField0_ & 0x00000004) != 0)) {
+        defaultGraphInputTensorShape_ = new java.util.ArrayList<org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto>(defaultGraphInputTensorShape_);
+        bitField0_ |= 0x00000004;
+       }
+    }
+
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto, org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto.Builder, org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProtoOrBuilder> defaultGraphInputTensorShapeBuilder_;
+
+    /**
+     * <pre>
+     * Optional: Default graph input tensor shape used to allocate memory
+     * before executing op
+     * </pre>
+     *
+     * <code>repeated .tensorflow.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto default_graph_input_tensor_shape = 6;</code>
+     */
+    public java.util.List<org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto> getDefaultGraphInputTensorShapeList() {
+      if (defaultGraphInputTensorShapeBuilder_ == null) {
+        return java.util.Collections.unmodifiableList(defaultGraphInputTensorShape_);
+      } else {
+        return defaultGraphInputTensorShapeBuilder_.getMessageList();
+      }
+    }
+    /**
+     * <pre>
+     * Optional: Default graph input tensor shape used to allocate memory
+     * before executing op
+     * </pre>
+     *
+     * <code>repeated .tensorflow.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto default_graph_input_tensor_shape = 6;</code>
+     */
+    public int getDefaultGraphInputTensorShapeCount() {
+      if (defaultGraphInputTensorShapeBuilder_ == null) {
+        return defaultGraphInputTensorShape_.size();
+      } else {
+        return defaultGraphInputTensorShapeBuilder_.getCount();
+      }
+    }
+    /**
+     * <pre>
+     * Optional: Default graph input tensor shape used to allocate memory
+     * before executing op
+     * </pre>
+     *
+     * <code>repeated .tensorflow.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto default_graph_input_tensor_shape = 6;</code>
+     */
+    public org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto getDefaultGraphInputTensorShape(int index) {
+      if (defaultGraphInputTensorShapeBuilder_ == null) {
+        return defaultGraphInputTensorShape_.get(index);
+      } else {
+        return defaultGraphInputTensorShapeBuilder_.getMessage(index);
+      }
+    }
+    /**
+     * <pre>
+     * Optional: Default graph input tensor shape used to allocate memory
+     * before executing op
+     * </pre>
+     *
+     * <code>repeated .tensorflow.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto default_graph_input_tensor_shape = 6;</code>
+     */
+    public Builder setDefaultGraphInputTensorShape(
+        int index, org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto value) {
+      if (defaultGraphInputTensorShapeBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureDefaultGraphInputTensorShapeIsMutable();
+        defaultGraphInputTensorShape_.set(index, value);
+        onChanged();
+      } else {
+        defaultGraphInputTensorShapeBuilder_.setMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Optional: Default graph input tensor shape used to allocate memory
+     * before executing op
+     * </pre>
+     *
+     * <code>repeated .tensorflow.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto default_graph_input_tensor_shape = 6;</code>
+     */
+    public Builder setDefaultGraphInputTensorShape(
+        int index, org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto.Builder builderForValue) {
+      if (defaultGraphInputTensorShapeBuilder_ == null) {
+        ensureDefaultGraphInputTensorShapeIsMutable();
+        defaultGraphInputTensorShape_.set(index, builderForValue.build());
+        onChanged();
+      } else {
+        defaultGraphInputTensorShapeBuilder_.setMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Optional: Default graph input tensor shape used to allocate memory
+     * before executing op
+     * </pre>
+     *
+     * <code>repeated .tensorflow.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto default_graph_input_tensor_shape = 6;</code>
+     */
+    public Builder addDefaultGraphInputTensorShape(org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto value) {
+      if (defaultGraphInputTensorShapeBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureDefaultGraphInputTensorShapeIsMutable();
+        defaultGraphInputTensorShape_.add(value);
+        onChanged();
+      } else {
+        defaultGraphInputTensorShapeBuilder_.addMessage(value);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Optional: Default graph input tensor shape used to allocate memory
+     * before executing op
+     * </pre>
+     *
+     * <code>repeated .tensorflow.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto default_graph_input_tensor_shape = 6;</code>
+     */
+    public Builder addDefaultGraphInputTensorShape(
+        int index, org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto value) {
+      if (defaultGraphInputTensorShapeBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureDefaultGraphInputTensorShapeIsMutable();
+        defaultGraphInputTensorShape_.add(index, value);
+        onChanged();
+      } else {
+        defaultGraphInputTensorShapeBuilder_.addMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Optional: Default graph input tensor shape used to allocate memory
+     * before executing op
+     * </pre>
+     *
+     * <code>repeated .tensorflow.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto default_graph_input_tensor_shape = 6;</code>
+     */
+    public Builder addDefaultGraphInputTensorShape(
+        org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto.Builder builderForValue) {
+      if (defaultGraphInputTensorShapeBuilder_ == null) {
+        ensureDefaultGraphInputTensorShapeIsMutable();
+        defaultGraphInputTensorShape_.add(builderForValue.build());
+        onChanged();
+      } else {
+        defaultGraphInputTensorShapeBuilder_.addMessage(builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Optional: Default graph input tensor shape used to allocate memory
+     * before executing op
+     * </pre>
+     *
+     * <code>repeated .tensorflow.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto default_graph_input_tensor_shape = 6;</code>
+     */
+    public Builder addDefaultGraphInputTensorShape(
+        int index, org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto.Builder builderForValue) {
+      if (defaultGraphInputTensorShapeBuilder_ == null) {
+        ensureDefaultGraphInputTensorShapeIsMutable();
+        defaultGraphInputTensorShape_.add(index, builderForValue.build());
+        onChanged();
+      } else {
+        defaultGraphInputTensorShapeBuilder_.addMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Optional: Default graph input tensor shape used to allocate memory
+     * before executing op
+     * </pre>
+     *
+     * <code>repeated .tensorflow.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto default_graph_input_tensor_shape = 6;</code>
+     */
+    public Builder addAllDefaultGraphInputTensorShape(
+        java.lang.Iterable<? extends org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto> values) {
+      if (defaultGraphInputTensorShapeBuilder_ == null) {
+        ensureDefaultGraphInputTensorShapeIsMutable();
+        com.google.protobuf.AbstractMessageLite.Builder.addAll(
+            values, defaultGraphInputTensorShape_);
+        onChanged();
+      } else {
+        defaultGraphInputTensorShapeBuilder_.addAllMessages(values);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Optional: Default graph input tensor shape used to allocate memory
+     * before executing op
+     * </pre>
+     *
+     * <code>repeated .tensorflow.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto default_graph_input_tensor_shape = 6;</code>
+     */
+    public Builder clearDefaultGraphInputTensorShape() {
+      if (defaultGraphInputTensorShapeBuilder_ == null) {
+        defaultGraphInputTensorShape_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000004);
+        onChanged();
+      } else {
+        defaultGraphInputTensorShapeBuilder_.clear();
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Optional: Default graph input tensor shape used to allocate memory
+     * before executing op
+     * </pre>
+     *
+     * <code>repeated .tensorflow.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto default_graph_input_tensor_shape = 6;</code>
+     */
+    public Builder removeDefaultGraphInputTensorShape(int index) {
+      if (defaultGraphInputTensorShapeBuilder_ == null) {
+        ensureDefaultGraphInputTensorShapeIsMutable();
+        defaultGraphInputTensorShape_.remove(index);
+        onChanged();
+      } else {
+        defaultGraphInputTensorShapeBuilder_.remove(index);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Optional: Default graph input tensor shape used to allocate memory
+     * before executing op
+     * </pre>
+     *
+     * <code>repeated .tensorflow.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto default_graph_input_tensor_shape = 6;</code>
+     */
+    public org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto.Builder getDefaultGraphInputTensorShapeBuilder(
+        int index) {
+      return getDefaultGraphInputTensorShapeFieldBuilder().getBuilder(index);
+    }
+    /**
+     * <pre>
+     * Optional: Default graph input tensor shape used to allocate memory
+     * before executing op
+     * </pre>
+     *
+     * <code>repeated .tensorflow.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto default_graph_input_tensor_shape = 6;</code>
+     */
+    public org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProtoOrBuilder getDefaultGraphInputTensorShapeOrBuilder(
+        int index) {
+      if (defaultGraphInputTensorShapeBuilder_ == null) {
+        return defaultGraphInputTensorShape_.get(index);  } else {
+        return defaultGraphInputTensorShapeBuilder_.getMessageOrBuilder(index);
+      }
+    }
+    /**
+     * <pre>
+     * Optional: Default graph input tensor shape used to allocate memory
+     * before executing op
+     * </pre>
+     *
+     * <code>repeated .tensorflow.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto default_graph_input_tensor_shape = 6;</code>
+     */
+    public java.util.List<? extends org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProtoOrBuilder> 
+         getDefaultGraphInputTensorShapeOrBuilderList() {
+      if (defaultGraphInputTensorShapeBuilder_ != null) {
+        return defaultGraphInputTensorShapeBuilder_.getMessageOrBuilderList();
+      } else {
+        return java.util.Collections.unmodifiableList(defaultGraphInputTensorShape_);
+      }
+    }
+    /**
+     * <pre>
+     * Optional: Default graph input tensor shape used to allocate memory
+     * before executing op
+     * </pre>
+     *
+     * <code>repeated .tensorflow.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto default_graph_input_tensor_shape = 6;</code>
+     */
+    public org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto.Builder addDefaultGraphInputTensorShapeBuilder() {
+      return getDefaultGraphInputTensorShapeFieldBuilder().addBuilder(
+          org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto.getDefaultInstance());
+    }
+    /**
+     * <pre>
+     * Optional: Default graph input tensor shape used to allocate memory
+     * before executing op
+     * </pre>
+     *
+     * <code>repeated .tensorflow.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto default_graph_input_tensor_shape = 6;</code>
+     */
+    public org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto.Builder addDefaultGraphInputTensorShapeBuilder(
+        int index) {
+      return getDefaultGraphInputTensorShapeFieldBuilder().addBuilder(
+          index, org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto.getDefaultInstance());
+    }
+    /**
+     * <pre>
+     * Optional: Default graph input tensor shape used to allocate memory
+     * before executing op
+     * </pre>
+     *
+     * <code>repeated .tensorflow.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto default_graph_input_tensor_shape = 6;</code>
+     */
+    public java.util.List<org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto.Builder> 
+         getDefaultGraphInputTensorShapeBuilderList() {
+      return getDefaultGraphInputTensorShapeFieldBuilder().getBuilderList();
+    }
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto, org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto.Builder, org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProtoOrBuilder> 
+        getDefaultGraphInputTensorShapeFieldBuilder() {
+      if (defaultGraphInputTensorShapeBuilder_ == null) {
+        defaultGraphInputTensorShapeBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
+            org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto, org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto.Builder, org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProtoOrBuilder>(
+                defaultGraphInputTensorShape_,
+                ((bitField0_ & 0x00000004) != 0),
+                getParentForChildren(),
+                isClean());
+        defaultGraphInputTensorShape_ = null;
+      }
+      return defaultGraphInputTensorShapeBuilder_;
+    }
+
+    private java.util.List<org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto> defaultGraphOutputTensorShape_ =
+      java.util.Collections.emptyList();
+    private void ensureDefaultGraphOutputTensorShapeIsMutable() {
+      if (!((bitField0_ & 0x00000008) != 0)) {
+        defaultGraphOutputTensorShape_ = new java.util.ArrayList<org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto>(defaultGraphOutputTensorShape_);
+        bitField0_ |= 0x00000008;
+       }
+    }
+
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto, org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto.Builder, org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProtoOrBuilder> defaultGraphOutputTensorShapeBuilder_;
+
+    /**
+     * <pre>
+     * Optional: Default graph input tensor shape used to allocate memory
+     * before executing op
+     * TODO(satok): Remote output tensor shape once shape information is stored
+     * in NodeDef
+     * </pre>
+     *
+     * <code>repeated .tensorflow.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto default_graph_output_tensor_shape = 7;</code>
+     */
+    public java.util.List<org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto> getDefaultGraphOutputTensorShapeList() {
+      if (defaultGraphOutputTensorShapeBuilder_ == null) {
+        return java.util.Collections.unmodifiableList(defaultGraphOutputTensorShape_);
+      } else {
+        return defaultGraphOutputTensorShapeBuilder_.getMessageList();
+      }
+    }
+    /**
+     * <pre>
+     * Optional: Default graph input tensor shape used to allocate memory
+     * before executing op
+     * TODO(satok): Remote output tensor shape once shape information is stored
+     * in NodeDef
+     * </pre>
+     *
+     * <code>repeated .tensorflow.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto default_graph_output_tensor_shape = 7;</code>
+     */
+    public int getDefaultGraphOutputTensorShapeCount() {
+      if (defaultGraphOutputTensorShapeBuilder_ == null) {
+        return defaultGraphOutputTensorShape_.size();
+      } else {
+        return defaultGraphOutputTensorShapeBuilder_.getCount();
+      }
+    }
+    /**
+     * <pre>
+     * Optional: Default graph input tensor shape used to allocate memory
+     * before executing op
+     * TODO(satok): Remote output tensor shape once shape information is stored
+     * in NodeDef
+     * </pre>
+     *
+     * <code>repeated .tensorflow.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto default_graph_output_tensor_shape = 7;</code>
+     */
+    public org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto getDefaultGraphOutputTensorShape(int index) {
+      if (defaultGraphOutputTensorShapeBuilder_ == null) {
+        return defaultGraphOutputTensorShape_.get(index);
+      } else {
+        return defaultGraphOutputTensorShapeBuilder_.getMessage(index);
+      }
+    }
+    /**
+     * <pre>
+     * Optional: Default graph input tensor shape used to allocate memory
+     * before executing op
+     * TODO(satok): Remote output tensor shape once shape information is stored
+     * in NodeDef
+     * </pre>
+     *
+     * <code>repeated .tensorflow.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto default_graph_output_tensor_shape = 7;</code>
+     */
+    public Builder setDefaultGraphOutputTensorShape(
+        int index, org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto value) {
+      if (defaultGraphOutputTensorShapeBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureDefaultGraphOutputTensorShapeIsMutable();
+        defaultGraphOutputTensorShape_.set(index, value);
+        onChanged();
+      } else {
+        defaultGraphOutputTensorShapeBuilder_.setMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Optional: Default graph input tensor shape used to allocate memory
+     * before executing op
+     * TODO(satok): Remote output tensor shape once shape information is stored
+     * in NodeDef
+     * </pre>
+     *
+     * <code>repeated .tensorflow.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto default_graph_output_tensor_shape = 7;</code>
+     */
+    public Builder setDefaultGraphOutputTensorShape(
+        int index, org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto.Builder builderForValue) {
+      if (defaultGraphOutputTensorShapeBuilder_ == null) {
+        ensureDefaultGraphOutputTensorShapeIsMutable();
+        defaultGraphOutputTensorShape_.set(index, builderForValue.build());
+        onChanged();
+      } else {
+        defaultGraphOutputTensorShapeBuilder_.setMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Optional: Default graph input tensor shape used to allocate memory
+     * before executing op
+     * TODO(satok): Remote output tensor shape once shape information is stored
+     * in NodeDef
+     * </pre>
+     *
+     * <code>repeated .tensorflow.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto default_graph_output_tensor_shape = 7;</code>
+     */
+    public Builder addDefaultGraphOutputTensorShape(org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto value) {
+      if (defaultGraphOutputTensorShapeBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureDefaultGraphOutputTensorShapeIsMutable();
+        defaultGraphOutputTensorShape_.add(value);
+        onChanged();
+      } else {
+        defaultGraphOutputTensorShapeBuilder_.addMessage(value);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Optional: Default graph input tensor shape used to allocate memory
+     * before executing op
+     * TODO(satok): Remote output tensor shape once shape information is stored
+     * in NodeDef
+     * </pre>
+     *
+     * <code>repeated .tensorflow.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto default_graph_output_tensor_shape = 7;</code>
+     */
+    public Builder addDefaultGraphOutputTensorShape(
+        int index, org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto value) {
+      if (defaultGraphOutputTensorShapeBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureDefaultGraphOutputTensorShapeIsMutable();
+        defaultGraphOutputTensorShape_.add(index, value);
+        onChanged();
+      } else {
+        defaultGraphOutputTensorShapeBuilder_.addMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Optional: Default graph input tensor shape used to allocate memory
+     * before executing op
+     * TODO(satok): Remote output tensor shape once shape information is stored
+     * in NodeDef
+     * </pre>
+     *
+     * <code>repeated .tensorflow.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto default_graph_output_tensor_shape = 7;</code>
+     */
+    public Builder addDefaultGraphOutputTensorShape(
+        org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto.Builder builderForValue) {
+      if (defaultGraphOutputTensorShapeBuilder_ == null) {
+        ensureDefaultGraphOutputTensorShapeIsMutable();
+        defaultGraphOutputTensorShape_.add(builderForValue.build());
+        onChanged();
+      } else {
+        defaultGraphOutputTensorShapeBuilder_.addMessage(builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Optional: Default graph input tensor shape used to allocate memory
+     * before executing op
+     * TODO(satok): Remote output tensor shape once shape information is stored
+     * in NodeDef
+     * </pre>
+     *
+     * <code>repeated .tensorflow.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto default_graph_output_tensor_shape = 7;</code>
+     */
+    public Builder addDefaultGraphOutputTensorShape(
+        int index, org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto.Builder builderForValue) {
+      if (defaultGraphOutputTensorShapeBuilder_ == null) {
+        ensureDefaultGraphOutputTensorShapeIsMutable();
+        defaultGraphOutputTensorShape_.add(index, builderForValue.build());
+        onChanged();
+      } else {
+        defaultGraphOutputTensorShapeBuilder_.addMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Optional: Default graph input tensor shape used to allocate memory
+     * before executing op
+     * TODO(satok): Remote output tensor shape once shape information is stored
+     * in NodeDef
+     * </pre>
+     *
+     * <code>repeated .tensorflow.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto default_graph_output_tensor_shape = 7;</code>
+     */
+    public Builder addAllDefaultGraphOutputTensorShape(
+        java.lang.Iterable<? extends org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto> values) {
+      if (defaultGraphOutputTensorShapeBuilder_ == null) {
+        ensureDefaultGraphOutputTensorShapeIsMutable();
+        com.google.protobuf.AbstractMessageLite.Builder.addAll(
+            values, defaultGraphOutputTensorShape_);
+        onChanged();
+      } else {
+        defaultGraphOutputTensorShapeBuilder_.addAllMessages(values);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Optional: Default graph input tensor shape used to allocate memory
+     * before executing op
+     * TODO(satok): Remote output tensor shape once shape information is stored
+     * in NodeDef
+     * </pre>
+     *
+     * <code>repeated .tensorflow.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto default_graph_output_tensor_shape = 7;</code>
+     */
+    public Builder clearDefaultGraphOutputTensorShape() {
+      if (defaultGraphOutputTensorShapeBuilder_ == null) {
+        defaultGraphOutputTensorShape_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000008);
+        onChanged();
+      } else {
+        defaultGraphOutputTensorShapeBuilder_.clear();
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Optional: Default graph input tensor shape used to allocate memory
+     * before executing op
+     * TODO(satok): Remote output tensor shape once shape information is stored
+     * in NodeDef
+     * </pre>
+     *
+     * <code>repeated .tensorflow.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto default_graph_output_tensor_shape = 7;</code>
+     */
+    public Builder removeDefaultGraphOutputTensorShape(int index) {
+      if (defaultGraphOutputTensorShapeBuilder_ == null) {
+        ensureDefaultGraphOutputTensorShapeIsMutable();
+        defaultGraphOutputTensorShape_.remove(index);
+        onChanged();
+      } else {
+        defaultGraphOutputTensorShapeBuilder_.remove(index);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Optional: Default graph input tensor shape used to allocate memory
+     * before executing op
+     * TODO(satok): Remote output tensor shape once shape information is stored
+     * in NodeDef
+     * </pre>
+     *
+     * <code>repeated .tensorflow.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto default_graph_output_tensor_shape = 7;</code>
+     */
+    public org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto.Builder getDefaultGraphOutputTensorShapeBuilder(
+        int index) {
+      return getDefaultGraphOutputTensorShapeFieldBuilder().getBuilder(index);
+    }
+    /**
+     * <pre>
+     * Optional: Default graph input tensor shape used to allocate memory
+     * before executing op
+     * TODO(satok): Remote output tensor shape once shape information is stored
+     * in NodeDef
+     * </pre>
+     *
+     * <code>repeated .tensorflow.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto default_graph_output_tensor_shape = 7;</code>
+     */
+    public org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProtoOrBuilder getDefaultGraphOutputTensorShapeOrBuilder(
+        int index) {
+      if (defaultGraphOutputTensorShapeBuilder_ == null) {
+        return defaultGraphOutputTensorShape_.get(index);  } else {
+        return defaultGraphOutputTensorShapeBuilder_.getMessageOrBuilder(index);
+      }
+    }
+    /**
+     * <pre>
+     * Optional: Default graph input tensor shape used to allocate memory
+     * before executing op
+     * TODO(satok): Remote output tensor shape once shape information is stored
+     * in NodeDef
+     * </pre>
+     *
+     * <code>repeated .tensorflow.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto default_graph_output_tensor_shape = 7;</code>
+     */
+    public java.util.List<? extends org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProtoOrBuilder> 
+         getDefaultGraphOutputTensorShapeOrBuilderList() {
+      if (defaultGraphOutputTensorShapeBuilder_ != null) {
+        return defaultGraphOutputTensorShapeBuilder_.getMessageOrBuilderList();
+      } else {
+        return java.util.Collections.unmodifiableList(defaultGraphOutputTensorShape_);
+      }
+    }
+    /**
+     * <pre>
+     * Optional: Default graph input tensor shape used to allocate memory
+     * before executing op
+     * TODO(satok): Remote output tensor shape once shape information is stored
+     * in NodeDef
+     * </pre>
+     *
+     * <code>repeated .tensorflow.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto default_graph_output_tensor_shape = 7;</code>
+     */
+    public org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto.Builder addDefaultGraphOutputTensorShapeBuilder() {
+      return getDefaultGraphOutputTensorShapeFieldBuilder().addBuilder(
+          org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto.getDefaultInstance());
+    }
+    /**
+     * <pre>
+     * Optional: Default graph input tensor shape used to allocate memory
+     * before executing op
+     * TODO(satok): Remote output tensor shape once shape information is stored
+     * in NodeDef
+     * </pre>
+     *
+     * <code>repeated .tensorflow.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto default_graph_output_tensor_shape = 7;</code>
+     */
+    public org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto.Builder addDefaultGraphOutputTensorShapeBuilder(
+        int index) {
+      return getDefaultGraphOutputTensorShapeFieldBuilder().addBuilder(
+          index, org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto.getDefaultInstance());
+    }
+    /**
+     * <pre>
+     * Optional: Default graph input tensor shape used to allocate memory
+     * before executing op
+     * TODO(satok): Remote output tensor shape once shape information is stored
+     * in NodeDef
+     * </pre>
+     *
+     * <code>repeated .tensorflow.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto default_graph_output_tensor_shape = 7;</code>
+     */
+    public java.util.List<org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto.Builder> 
+         getDefaultGraphOutputTensorShapeBuilderList() {
+      return getDefaultGraphOutputTensorShapeFieldBuilder().getBuilderList();
+    }
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto, org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto.Builder, org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProtoOrBuilder> 
+        getDefaultGraphOutputTensorShapeFieldBuilder() {
+      if (defaultGraphOutputTensorShapeBuilder_ == null) {
+        defaultGraphOutputTensorShapeBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
+            org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto, org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto.Builder, org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProtoOrBuilder>(
+                defaultGraphOutputTensorShape_,
+                ((bitField0_ & 0x00000008) != 0),
+                getParentForChildren(),
+                isClean());
+        defaultGraphOutputTensorShape_ = null;
+      }
+      return defaultGraphOutputTensorShapeBuilder_;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.RemoteFusedGraphExecuteInfo)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.RemoteFusedGraphExecuteInfo)
+  private static final org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo();
+  }
+
+  public static org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<RemoteFusedGraphExecuteInfo>
+      PARSER = new com.google.protobuf.AbstractParser<RemoteFusedGraphExecuteInfo>() {
+    @java.lang.Override
+    public RemoteFusedGraphExecuteInfo parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new RemoteFusedGraphExecuteInfo(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<RemoteFusedGraphExecuteInfo> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<RemoteFusedGraphExecuteInfo> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/RemoteFusedGraphExecuteInfoOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/RemoteFusedGraphExecuteInfoOrBuilder.java
new file mode 100644
index 00000000000..6e14ad8a2e0
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/RemoteFusedGraphExecuteInfoOrBuilder.java
@@ -0,0 +1,239 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/framework/remote_fused_graph_execute_info.proto
+
+package org.tensorflow.proto.framework;
+
+public interface RemoteFusedGraphExecuteInfoOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.RemoteFusedGraphExecuteInfo)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <pre>
+   * Definition of remote graph
+   * </pre>
+   *
+   * <code>.tensorflow.GraphDef remote_graph = 1;</code>
+   */
+  boolean hasRemoteGraph();
+  /**
+   * <pre>
+   * Definition of remote graph
+   * </pre>
+   *
+   * <code>.tensorflow.GraphDef remote_graph = 1;</code>
+   */
+  org.tensorflow.proto.framework.GraphDef getRemoteGraph();
+  /**
+   * <pre>
+   * Definition of remote graph
+   * </pre>
+   *
+   * <code>.tensorflow.GraphDef remote_graph = 1;</code>
+   */
+  org.tensorflow.proto.framework.GraphDefOrBuilder getRemoteGraphOrBuilder();
+
+  /**
+   * <pre>
+   * Remote fused graph input node name
+   * </pre>
+   *
+   * <code>repeated string graph_input_node_name = 2;</code>
+   */
+  java.util.List<java.lang.String>
+      getGraphInputNodeNameList();
+  /**
+   * <pre>
+   * Remote fused graph input node name
+   * </pre>
+   *
+   * <code>repeated string graph_input_node_name = 2;</code>
+   */
+  int getGraphInputNodeNameCount();
+  /**
+   * <pre>
+   * Remote fused graph input node name
+   * </pre>
+   *
+   * <code>repeated string graph_input_node_name = 2;</code>
+   */
+  java.lang.String getGraphInputNodeName(int index);
+  /**
+   * <pre>
+   * Remote fused graph input node name
+   * </pre>
+   *
+   * <code>repeated string graph_input_node_name = 2;</code>
+   */
+  com.google.protobuf.ByteString
+      getGraphInputNodeNameBytes(int index);
+
+  /**
+   * <pre>
+   * Remote fused graph output node name
+   * </pre>
+   *
+   * <code>repeated string graph_output_node_name = 3;</code>
+   */
+  java.util.List<java.lang.String>
+      getGraphOutputNodeNameList();
+  /**
+   * <pre>
+   * Remote fused graph output node name
+   * </pre>
+   *
+   * <code>repeated string graph_output_node_name = 3;</code>
+   */
+  int getGraphOutputNodeNameCount();
+  /**
+   * <pre>
+   * Remote fused graph output node name
+   * </pre>
+   *
+   * <code>repeated string graph_output_node_name = 3;</code>
+   */
+  java.lang.String getGraphOutputNodeName(int index);
+  /**
+   * <pre>
+   * Remote fused graph output node name
+   * </pre>
+   *
+   * <code>repeated string graph_output_node_name = 3;</code>
+   */
+  com.google.protobuf.ByteString
+      getGraphOutputNodeNameBytes(int index);
+
+  /**
+   * <pre>
+   * Executor's name
+   * </pre>
+   *
+   * <code>string executor_name = 4;</code>
+   */
+  java.lang.String getExecutorName();
+  /**
+   * <pre>
+   * Executor's name
+   * </pre>
+   *
+   * <code>string executor_name = 4;</code>
+   */
+  com.google.protobuf.ByteString
+      getExecutorNameBytes();
+
+  /**
+   * <pre>
+   * Optional: Parameters given to the executor
+   * </pre>
+   *
+   * <code>bytes serialized_executor_parameters = 5;</code>
+   */
+  com.google.protobuf.ByteString getSerializedExecutorParameters();
+
+  /**
+   * <pre>
+   * Optional: Default graph input tensor shape used to allocate memory
+   * before executing op
+   * </pre>
+   *
+   * <code>repeated .tensorflow.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto default_graph_input_tensor_shape = 6;</code>
+   */
+  java.util.List<org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto> 
+      getDefaultGraphInputTensorShapeList();
+  /**
+   * <pre>
+   * Optional: Default graph input tensor shape used to allocate memory
+   * before executing op
+   * </pre>
+   *
+   * <code>repeated .tensorflow.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto default_graph_input_tensor_shape = 6;</code>
+   */
+  org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto getDefaultGraphInputTensorShape(int index);
+  /**
+   * <pre>
+   * Optional: Default graph input tensor shape used to allocate memory
+   * before executing op
+   * </pre>
+   *
+   * <code>repeated .tensorflow.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto default_graph_input_tensor_shape = 6;</code>
+   */
+  int getDefaultGraphInputTensorShapeCount();
+  /**
+   * <pre>
+   * Optional: Default graph input tensor shape used to allocate memory
+   * before executing op
+   * </pre>
+   *
+   * <code>repeated .tensorflow.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto default_graph_input_tensor_shape = 6;</code>
+   */
+  java.util.List<? extends org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProtoOrBuilder> 
+      getDefaultGraphInputTensorShapeOrBuilderList();
+  /**
+   * <pre>
+   * Optional: Default graph input tensor shape used to allocate memory
+   * before executing op
+   * </pre>
+   *
+   * <code>repeated .tensorflow.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto default_graph_input_tensor_shape = 6;</code>
+   */
+  org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProtoOrBuilder getDefaultGraphInputTensorShapeOrBuilder(
+      int index);
+
+  /**
+   * <pre>
+   * Optional: Default graph input tensor shape used to allocate memory
+   * before executing op
+   * TODO(satok): Remote output tensor shape once shape information is stored
+   * in NodeDef
+   * </pre>
+   *
+   * <code>repeated .tensorflow.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto default_graph_output_tensor_shape = 7;</code>
+   */
+  java.util.List<org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto> 
+      getDefaultGraphOutputTensorShapeList();
+  /**
+   * <pre>
+   * Optional: Default graph input tensor shape used to allocate memory
+   * before executing op
+   * TODO(satok): Remote output tensor shape once shape information is stored
+   * in NodeDef
+   * </pre>
+   *
+   * <code>repeated .tensorflow.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto default_graph_output_tensor_shape = 7;</code>
+   */
+  org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto getDefaultGraphOutputTensorShape(int index);
+  /**
+   * <pre>
+   * Optional: Default graph input tensor shape used to allocate memory
+   * before executing op
+   * TODO(satok): Remote output tensor shape once shape information is stored
+   * in NodeDef
+   * </pre>
+   *
+   * <code>repeated .tensorflow.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto default_graph_output_tensor_shape = 7;</code>
+   */
+  int getDefaultGraphOutputTensorShapeCount();
+  /**
+   * <pre>
+   * Optional: Default graph input tensor shape used to allocate memory
+   * before executing op
+   * TODO(satok): Remote output tensor shape once shape information is stored
+   * in NodeDef
+   * </pre>
+   *
+   * <code>repeated .tensorflow.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto default_graph_output_tensor_shape = 7;</code>
+   */
+  java.util.List<? extends org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProtoOrBuilder> 
+      getDefaultGraphOutputTensorShapeOrBuilderList();
+  /**
+   * <pre>
+   * Optional: Default graph input tensor shape used to allocate memory
+   * before executing op
+   * TODO(satok): Remote output tensor shape once shape information is stored
+   * in NodeDef
+   * </pre>
+   *
+   * <code>repeated .tensorflow.RemoteFusedGraphExecuteInfo.TensorShapeTypeProto default_graph_output_tensor_shape = 7;</code>
+   */
+  org.tensorflow.proto.framework.RemoteFusedGraphExecuteInfo.TensorShapeTypeProtoOrBuilder getDefaultGraphOutputTensorShapeOrBuilder(
+      int index);
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/RemoteFusedGraphExecuteInfoProto.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/RemoteFusedGraphExecuteInfoProto.java
new file mode 100644
index 00000000000..95aeddaf025
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/RemoteFusedGraphExecuteInfoProto.java
@@ -0,0 +1,85 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/framework/remote_fused_graph_execute_info.proto
+
+package org.tensorflow.proto.framework;
+
+public final class RemoteFusedGraphExecuteInfoProto {
+  private RemoteFusedGraphExecuteInfoProto() {}
+  public static void registerAllExtensions(
+      com.google.protobuf.ExtensionRegistryLite registry) {
+  }
+
+  public static void registerAllExtensions(
+      com.google.protobuf.ExtensionRegistry registry) {
+    registerAllExtensions(
+        (com.google.protobuf.ExtensionRegistryLite) registry);
+  }
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_RemoteFusedGraphExecuteInfo_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_RemoteFusedGraphExecuteInfo_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_RemoteFusedGraphExecuteInfo_TensorShapeTypeProto_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_RemoteFusedGraphExecuteInfo_TensorShapeTypeProto_fieldAccessorTable;
+
+  public static com.google.protobuf.Descriptors.FileDescriptor
+      getDescriptor() {
+    return descriptor;
+  }
+  private static  com.google.protobuf.Descriptors.FileDescriptor
+      descriptor;
+  static {
+    java.lang.String[] descriptorData = {
+      "\n?tensorflow/core/framework/remote_fused" +
+      "_graph_execute_info.proto\022\ntensorflow\032%t" +
+      "ensorflow/core/framework/graph.proto\032,te" +
+      "nsorflow/core/framework/tensor_shape.pro" +
+      "to\032%tensorflow/core/framework/types.prot" +
+      "o\"\202\004\n\033RemoteFusedGraphExecuteInfo\022*\n\014rem" +
+      "ote_graph\030\001 \001(\0132\024.tensorflow.GraphDef\022\035\n" +
+      "\025graph_input_node_name\030\002 \003(\t\022\036\n\026graph_ou" +
+      "tput_node_name\030\003 \003(\t\022\025\n\rexecutor_name\030\004 " +
+      "\001(\t\022&\n\036serialized_executor_parameters\030\005 " +
+      "\001(\014\022f\n default_graph_input_tensor_shape\030" +
+      "\006 \003(\0132<.tensorflow.RemoteFusedGraphExecu" +
+      "teInfo.TensorShapeTypeProto\022g\n!default_g" +
+      "raph_output_tensor_shape\030\007 \003(\0132<.tensorf" +
+      "low.RemoteFusedGraphExecuteInfo.TensorSh" +
+      "apeTypeProto\032h\n\024TensorShapeTypeProto\022#\n\005" +
+      "dtype\030\001 \001(\0162\024.tensorflow.DataType\022+\n\005sha" +
+      "pe\030\002 \001(\0132\034.tensorflow.TensorShapeProtoB\257" +
+      "\001\n\036org.tensorflow.proto.frameworkB Remot" +
+      "eFusedGraphExecuteInfoProtoP\001Zfgithub.co" +
+      "m/tensorflow/tensorflow/tensorflow/go/co" +
+      "re/framework/remote_fused_graph_execute_" +
+      "info_go_proto\370\001\001b\006proto3"
+    };
+    descriptor = com.google.protobuf.Descriptors.FileDescriptor
+      .internalBuildGeneratedFileFrom(descriptorData,
+        new com.google.protobuf.Descriptors.FileDescriptor[] {
+          org.tensorflow.proto.framework.GraphProtos.getDescriptor(),
+          org.tensorflow.proto.framework.TensorShapeProtos.getDescriptor(),
+          org.tensorflow.proto.framework.TypesProtos.getDescriptor(),
+        });
+    internal_static_tensorflow_RemoteFusedGraphExecuteInfo_descriptor =
+      getDescriptor().getMessageTypes().get(0);
+    internal_static_tensorflow_RemoteFusedGraphExecuteInfo_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_RemoteFusedGraphExecuteInfo_descriptor,
+        new java.lang.String[] { "RemoteGraph", "GraphInputNodeName", "GraphOutputNodeName", "ExecutorName", "SerializedExecutorParameters", "DefaultGraphInputTensorShape", "DefaultGraphOutputTensorShape", });
+    internal_static_tensorflow_RemoteFusedGraphExecuteInfo_TensorShapeTypeProto_descriptor =
+      internal_static_tensorflow_RemoteFusedGraphExecuteInfo_descriptor.getNestedTypes().get(0);
+    internal_static_tensorflow_RemoteFusedGraphExecuteInfo_TensorShapeTypeProto_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_RemoteFusedGraphExecuteInfo_TensorShapeTypeProto_descriptor,
+        new java.lang.String[] { "Dtype", "Shape", });
+    org.tensorflow.proto.framework.GraphProtos.getDescriptor();
+    org.tensorflow.proto.framework.TensorShapeProtos.getDescriptor();
+    org.tensorflow.proto.framework.TypesProtos.getDescriptor();
+  }
+
+  // @@protoc_insertion_point(outer_class_scope)
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/RemoteTensorHandle.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/RemoteTensorHandle.java
new file mode 100644
index 00000000000..85a3d489f33
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/RemoteTensorHandle.java
@@ -0,0 +1,1441 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/remote_tensor_handle.proto
+
+package org.tensorflow.proto.framework;
+
+/**
+ * Protobuf type {@code tensorflow.eager.RemoteTensorHandle}
+ */
+public  final class RemoteTensorHandle extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.eager.RemoteTensorHandle)
+    RemoteTensorHandleOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use RemoteTensorHandle.newBuilder() to construct.
+  private RemoteTensorHandle(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private RemoteTensorHandle() {
+    device_ = "";
+    opDevice_ = "";
+    dtype_ = 0;
+    resourceDtypesAndShapes_ = java.util.Collections.emptyList();
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new RemoteTensorHandle();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private RemoteTensorHandle(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    int mutable_bitField0_ = 0;
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 8: {
+
+            opId_ = input.readInt64();
+            break;
+          }
+          case 16: {
+
+            outputNum_ = input.readInt32();
+            break;
+          }
+          case 26: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            device_ = s;
+            break;
+          }
+          case 34: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            opDevice_ = s;
+            break;
+          }
+          case 40: {
+            int rawValue = input.readEnum();
+
+            dtype_ = rawValue;
+            break;
+          }
+          case 50: {
+            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
+              resourceDtypesAndShapes_ = new java.util.ArrayList<org.tensorflow.proto.framework.ResourceDtypeAndShape>();
+              mutable_bitField0_ |= 0x00000001;
+            }
+            resourceDtypesAndShapes_.add(
+                input.readMessage(org.tensorflow.proto.framework.ResourceDtypeAndShape.parser(), extensionRegistry));
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      if (((mutable_bitField0_ & 0x00000001) != 0)) {
+        resourceDtypesAndShapes_ = java.util.Collections.unmodifiableList(resourceDtypesAndShapes_);
+      }
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.framework.RemoteTensorHandleProtos.internal_static_tensorflow_eager_RemoteTensorHandle_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.framework.RemoteTensorHandleProtos.internal_static_tensorflow_eager_RemoteTensorHandle_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.framework.RemoteTensorHandle.class, org.tensorflow.proto.framework.RemoteTensorHandle.Builder.class);
+  }
+
+  public static final int OP_ID_FIELD_NUMBER = 1;
+  private long opId_;
+  /**
+   * <pre>
+   * The ID of the operation that produced this tensor.
+   * </pre>
+   *
+   * <code>int64 op_id = 1;</code>
+   */
+  public long getOpId() {
+    return opId_;
+  }
+
+  public static final int OUTPUT_NUM_FIELD_NUMBER = 2;
+  private int outputNum_;
+  /**
+   * <pre>
+   * The index into the outputs of the operation that produced this tensor.
+   * </pre>
+   *
+   * <code>int32 output_num = 2;</code>
+   */
+  public int getOutputNum() {
+    return outputNum_;
+  }
+
+  public static final int DEVICE_FIELD_NUMBER = 3;
+  private volatile java.lang.Object device_;
+  /**
+   * <pre>
+   * Device where the tensor is located. Cannot be empty.
+   * For multi-device functions, it's the default device passed to placer.
+   * </pre>
+   *
+   * <code>string device = 3;</code>
+   */
+  public java.lang.String getDevice() {
+    java.lang.Object ref = device_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      device_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * Device where the tensor is located. Cannot be empty.
+   * For multi-device functions, it's the default device passed to placer.
+   * </pre>
+   *
+   * <code>string device = 3;</code>
+   */
+  public com.google.protobuf.ByteString
+      getDeviceBytes() {
+    java.lang.Object ref = device_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      device_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int OP_DEVICE_FIELD_NUMBER = 4;
+  private volatile java.lang.Object opDevice_;
+  /**
+   * <pre>
+   * Device of the operation producing this tensor. Can be empty if the
+   * operation producing this tensor is a multi-device function.
+   * </pre>
+   *
+   * <code>string op_device = 4;</code>
+   */
+  public java.lang.String getOpDevice() {
+    java.lang.Object ref = opDevice_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      opDevice_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * Device of the operation producing this tensor. Can be empty if the
+   * operation producing this tensor is a multi-device function.
+   * </pre>
+   *
+   * <code>string op_device = 4;</code>
+   */
+  public com.google.protobuf.ByteString
+      getOpDeviceBytes() {
+    java.lang.Object ref = opDevice_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      opDevice_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int DTYPE_FIELD_NUMBER = 5;
+  private int dtype_;
+  /**
+   * <pre>
+   * Tensor type.
+   * </pre>
+   *
+   * <code>.tensorflow.DataType dtype = 5;</code>
+   */
+  public int getDtypeValue() {
+    return dtype_;
+  }
+  /**
+   * <pre>
+   * Tensor type.
+   * </pre>
+   *
+   * <code>.tensorflow.DataType dtype = 5;</code>
+   */
+  public org.tensorflow.proto.framework.DataType getDtype() {
+    @SuppressWarnings("deprecation")
+    org.tensorflow.proto.framework.DataType result = org.tensorflow.proto.framework.DataType.valueOf(dtype_);
+    return result == null ? org.tensorflow.proto.framework.DataType.UNRECOGNIZED : result;
+  }
+
+  public static final int RESOURCE_DTYPES_AND_SHAPES_FIELD_NUMBER = 6;
+  private java.util.List<org.tensorflow.proto.framework.ResourceDtypeAndShape> resourceDtypesAndShapes_;
+  /**
+   * <pre>
+   * Optional data types and shapes of a remote resource variable.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.eager.ResourceDtypeAndShape resource_dtypes_and_shapes = 6;</code>
+   */
+  public java.util.List<org.tensorflow.proto.framework.ResourceDtypeAndShape> getResourceDtypesAndShapesList() {
+    return resourceDtypesAndShapes_;
+  }
+  /**
+   * <pre>
+   * Optional data types and shapes of a remote resource variable.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.eager.ResourceDtypeAndShape resource_dtypes_and_shapes = 6;</code>
+   */
+  public java.util.List<? extends org.tensorflow.proto.framework.ResourceDtypeAndShapeOrBuilder> 
+      getResourceDtypesAndShapesOrBuilderList() {
+    return resourceDtypesAndShapes_;
+  }
+  /**
+   * <pre>
+   * Optional data types and shapes of a remote resource variable.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.eager.ResourceDtypeAndShape resource_dtypes_and_shapes = 6;</code>
+   */
+  public int getResourceDtypesAndShapesCount() {
+    return resourceDtypesAndShapes_.size();
+  }
+  /**
+   * <pre>
+   * Optional data types and shapes of a remote resource variable.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.eager.ResourceDtypeAndShape resource_dtypes_and_shapes = 6;</code>
+   */
+  public org.tensorflow.proto.framework.ResourceDtypeAndShape getResourceDtypesAndShapes(int index) {
+    return resourceDtypesAndShapes_.get(index);
+  }
+  /**
+   * <pre>
+   * Optional data types and shapes of a remote resource variable.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.eager.ResourceDtypeAndShape resource_dtypes_and_shapes = 6;</code>
+   */
+  public org.tensorflow.proto.framework.ResourceDtypeAndShapeOrBuilder getResourceDtypesAndShapesOrBuilder(
+      int index) {
+    return resourceDtypesAndShapes_.get(index);
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    if (opId_ != 0L) {
+      output.writeInt64(1, opId_);
+    }
+    if (outputNum_ != 0) {
+      output.writeInt32(2, outputNum_);
+    }
+    if (!getDeviceBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 3, device_);
+    }
+    if (!getOpDeviceBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 4, opDevice_);
+    }
+    if (dtype_ != org.tensorflow.proto.framework.DataType.DT_INVALID.getNumber()) {
+      output.writeEnum(5, dtype_);
+    }
+    for (int i = 0; i < resourceDtypesAndShapes_.size(); i++) {
+      output.writeMessage(6, resourceDtypesAndShapes_.get(i));
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (opId_ != 0L) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt64Size(1, opId_);
+    }
+    if (outputNum_ != 0) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt32Size(2, outputNum_);
+    }
+    if (!getDeviceBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(3, device_);
+    }
+    if (!getOpDeviceBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(4, opDevice_);
+    }
+    if (dtype_ != org.tensorflow.proto.framework.DataType.DT_INVALID.getNumber()) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeEnumSize(5, dtype_);
+    }
+    for (int i = 0; i < resourceDtypesAndShapes_.size(); i++) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(6, resourceDtypesAndShapes_.get(i));
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.framework.RemoteTensorHandle)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.framework.RemoteTensorHandle other = (org.tensorflow.proto.framework.RemoteTensorHandle) obj;
+
+    if (getOpId()
+        != other.getOpId()) return false;
+    if (getOutputNum()
+        != other.getOutputNum()) return false;
+    if (!getDevice()
+        .equals(other.getDevice())) return false;
+    if (!getOpDevice()
+        .equals(other.getOpDevice())) return false;
+    if (dtype_ != other.dtype_) return false;
+    if (!getResourceDtypesAndShapesList()
+        .equals(other.getResourceDtypesAndShapesList())) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    hash = (37 * hash) + OP_ID_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+        getOpId());
+    hash = (37 * hash) + OUTPUT_NUM_FIELD_NUMBER;
+    hash = (53 * hash) + getOutputNum();
+    hash = (37 * hash) + DEVICE_FIELD_NUMBER;
+    hash = (53 * hash) + getDevice().hashCode();
+    hash = (37 * hash) + OP_DEVICE_FIELD_NUMBER;
+    hash = (53 * hash) + getOpDevice().hashCode();
+    hash = (37 * hash) + DTYPE_FIELD_NUMBER;
+    hash = (53 * hash) + dtype_;
+    if (getResourceDtypesAndShapesCount() > 0) {
+      hash = (37 * hash) + RESOURCE_DTYPES_AND_SHAPES_FIELD_NUMBER;
+      hash = (53 * hash) + getResourceDtypesAndShapesList().hashCode();
+    }
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.framework.RemoteTensorHandle parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.RemoteTensorHandle parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.RemoteTensorHandle parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.RemoteTensorHandle parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.RemoteTensorHandle parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.RemoteTensorHandle parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.RemoteTensorHandle parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.RemoteTensorHandle parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.RemoteTensorHandle parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.RemoteTensorHandle parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.RemoteTensorHandle parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.RemoteTensorHandle parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.framework.RemoteTensorHandle prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * Protobuf type {@code tensorflow.eager.RemoteTensorHandle}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.eager.RemoteTensorHandle)
+      org.tensorflow.proto.framework.RemoteTensorHandleOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.RemoteTensorHandleProtos.internal_static_tensorflow_eager_RemoteTensorHandle_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.RemoteTensorHandleProtos.internal_static_tensorflow_eager_RemoteTensorHandle_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.RemoteTensorHandle.class, org.tensorflow.proto.framework.RemoteTensorHandle.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.framework.RemoteTensorHandle.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+        getResourceDtypesAndShapesFieldBuilder();
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      opId_ = 0L;
+
+      outputNum_ = 0;
+
+      device_ = "";
+
+      opDevice_ = "";
+
+      dtype_ = 0;
+
+      if (resourceDtypesAndShapesBuilder_ == null) {
+        resourceDtypesAndShapes_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000001);
+      } else {
+        resourceDtypesAndShapesBuilder_.clear();
+      }
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.framework.RemoteTensorHandleProtos.internal_static_tensorflow_eager_RemoteTensorHandle_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.RemoteTensorHandle getDefaultInstanceForType() {
+      return org.tensorflow.proto.framework.RemoteTensorHandle.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.RemoteTensorHandle build() {
+      org.tensorflow.proto.framework.RemoteTensorHandle result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.RemoteTensorHandle buildPartial() {
+      org.tensorflow.proto.framework.RemoteTensorHandle result = new org.tensorflow.proto.framework.RemoteTensorHandle(this);
+      int from_bitField0_ = bitField0_;
+      result.opId_ = opId_;
+      result.outputNum_ = outputNum_;
+      result.device_ = device_;
+      result.opDevice_ = opDevice_;
+      result.dtype_ = dtype_;
+      if (resourceDtypesAndShapesBuilder_ == null) {
+        if (((bitField0_ & 0x00000001) != 0)) {
+          resourceDtypesAndShapes_ = java.util.Collections.unmodifiableList(resourceDtypesAndShapes_);
+          bitField0_ = (bitField0_ & ~0x00000001);
+        }
+        result.resourceDtypesAndShapes_ = resourceDtypesAndShapes_;
+      } else {
+        result.resourceDtypesAndShapes_ = resourceDtypesAndShapesBuilder_.build();
+      }
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.framework.RemoteTensorHandle) {
+        return mergeFrom((org.tensorflow.proto.framework.RemoteTensorHandle)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.framework.RemoteTensorHandle other) {
+      if (other == org.tensorflow.proto.framework.RemoteTensorHandle.getDefaultInstance()) return this;
+      if (other.getOpId() != 0L) {
+        setOpId(other.getOpId());
+      }
+      if (other.getOutputNum() != 0) {
+        setOutputNum(other.getOutputNum());
+      }
+      if (!other.getDevice().isEmpty()) {
+        device_ = other.device_;
+        onChanged();
+      }
+      if (!other.getOpDevice().isEmpty()) {
+        opDevice_ = other.opDevice_;
+        onChanged();
+      }
+      if (other.dtype_ != 0) {
+        setDtypeValue(other.getDtypeValue());
+      }
+      if (resourceDtypesAndShapesBuilder_ == null) {
+        if (!other.resourceDtypesAndShapes_.isEmpty()) {
+          if (resourceDtypesAndShapes_.isEmpty()) {
+            resourceDtypesAndShapes_ = other.resourceDtypesAndShapes_;
+            bitField0_ = (bitField0_ & ~0x00000001);
+          } else {
+            ensureResourceDtypesAndShapesIsMutable();
+            resourceDtypesAndShapes_.addAll(other.resourceDtypesAndShapes_);
+          }
+          onChanged();
+        }
+      } else {
+        if (!other.resourceDtypesAndShapes_.isEmpty()) {
+          if (resourceDtypesAndShapesBuilder_.isEmpty()) {
+            resourceDtypesAndShapesBuilder_.dispose();
+            resourceDtypesAndShapesBuilder_ = null;
+            resourceDtypesAndShapes_ = other.resourceDtypesAndShapes_;
+            bitField0_ = (bitField0_ & ~0x00000001);
+            resourceDtypesAndShapesBuilder_ = 
+              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
+                 getResourceDtypesAndShapesFieldBuilder() : null;
+          } else {
+            resourceDtypesAndShapesBuilder_.addAllMessages(other.resourceDtypesAndShapes_);
+          }
+        }
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.framework.RemoteTensorHandle parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.framework.RemoteTensorHandle) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+    private int bitField0_;
+
+    private long opId_ ;
+    /**
+     * <pre>
+     * The ID of the operation that produced this tensor.
+     * </pre>
+     *
+     * <code>int64 op_id = 1;</code>
+     */
+    public long getOpId() {
+      return opId_;
+    }
+    /**
+     * <pre>
+     * The ID of the operation that produced this tensor.
+     * </pre>
+     *
+     * <code>int64 op_id = 1;</code>
+     */
+    public Builder setOpId(long value) {
+      
+      opId_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * The ID of the operation that produced this tensor.
+     * </pre>
+     *
+     * <code>int64 op_id = 1;</code>
+     */
+    public Builder clearOpId() {
+      
+      opId_ = 0L;
+      onChanged();
+      return this;
+    }
+
+    private int outputNum_ ;
+    /**
+     * <pre>
+     * The index into the outputs of the operation that produced this tensor.
+     * </pre>
+     *
+     * <code>int32 output_num = 2;</code>
+     */
+    public int getOutputNum() {
+      return outputNum_;
+    }
+    /**
+     * <pre>
+     * The index into the outputs of the operation that produced this tensor.
+     * </pre>
+     *
+     * <code>int32 output_num = 2;</code>
+     */
+    public Builder setOutputNum(int value) {
+      
+      outputNum_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * The index into the outputs of the operation that produced this tensor.
+     * </pre>
+     *
+     * <code>int32 output_num = 2;</code>
+     */
+    public Builder clearOutputNum() {
+      
+      outputNum_ = 0;
+      onChanged();
+      return this;
+    }
+
+    private java.lang.Object device_ = "";
+    /**
+     * <pre>
+     * Device where the tensor is located. Cannot be empty.
+     * For multi-device functions, it's the default device passed to placer.
+     * </pre>
+     *
+     * <code>string device = 3;</code>
+     */
+    public java.lang.String getDevice() {
+      java.lang.Object ref = device_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        device_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Device where the tensor is located. Cannot be empty.
+     * For multi-device functions, it's the default device passed to placer.
+     * </pre>
+     *
+     * <code>string device = 3;</code>
+     */
+    public com.google.protobuf.ByteString
+        getDeviceBytes() {
+      java.lang.Object ref = device_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        device_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Device where the tensor is located. Cannot be empty.
+     * For multi-device functions, it's the default device passed to placer.
+     * </pre>
+     *
+     * <code>string device = 3;</code>
+     */
+    public Builder setDevice(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      device_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Device where the tensor is located. Cannot be empty.
+     * For multi-device functions, it's the default device passed to placer.
+     * </pre>
+     *
+     * <code>string device = 3;</code>
+     */
+    public Builder clearDevice() {
+      
+      device_ = getDefaultInstance().getDevice();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Device where the tensor is located. Cannot be empty.
+     * For multi-device functions, it's the default device passed to placer.
+     * </pre>
+     *
+     * <code>string device = 3;</code>
+     */
+    public Builder setDeviceBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      device_ = value;
+      onChanged();
+      return this;
+    }
+
+    private java.lang.Object opDevice_ = "";
+    /**
+     * <pre>
+     * Device of the operation producing this tensor. Can be empty if the
+     * operation producing this tensor is a multi-device function.
+     * </pre>
+     *
+     * <code>string op_device = 4;</code>
+     */
+    public java.lang.String getOpDevice() {
+      java.lang.Object ref = opDevice_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        opDevice_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Device of the operation producing this tensor. Can be empty if the
+     * operation producing this tensor is a multi-device function.
+     * </pre>
+     *
+     * <code>string op_device = 4;</code>
+     */
+    public com.google.protobuf.ByteString
+        getOpDeviceBytes() {
+      java.lang.Object ref = opDevice_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        opDevice_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Device of the operation producing this tensor. Can be empty if the
+     * operation producing this tensor is a multi-device function.
+     * </pre>
+     *
+     * <code>string op_device = 4;</code>
+     */
+    public Builder setOpDevice(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      opDevice_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Device of the operation producing this tensor. Can be empty if the
+     * operation producing this tensor is a multi-device function.
+     * </pre>
+     *
+     * <code>string op_device = 4;</code>
+     */
+    public Builder clearOpDevice() {
+      
+      opDevice_ = getDefaultInstance().getOpDevice();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Device of the operation producing this tensor. Can be empty if the
+     * operation producing this tensor is a multi-device function.
+     * </pre>
+     *
+     * <code>string op_device = 4;</code>
+     */
+    public Builder setOpDeviceBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      opDevice_ = value;
+      onChanged();
+      return this;
+    }
+
+    private int dtype_ = 0;
+    /**
+     * <pre>
+     * Tensor type.
+     * </pre>
+     *
+     * <code>.tensorflow.DataType dtype = 5;</code>
+     */
+    public int getDtypeValue() {
+      return dtype_;
+    }
+    /**
+     * <pre>
+     * Tensor type.
+     * </pre>
+     *
+     * <code>.tensorflow.DataType dtype = 5;</code>
+     */
+    public Builder setDtypeValue(int value) {
+      dtype_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Tensor type.
+     * </pre>
+     *
+     * <code>.tensorflow.DataType dtype = 5;</code>
+     */
+    public org.tensorflow.proto.framework.DataType getDtype() {
+      @SuppressWarnings("deprecation")
+      org.tensorflow.proto.framework.DataType result = org.tensorflow.proto.framework.DataType.valueOf(dtype_);
+      return result == null ? org.tensorflow.proto.framework.DataType.UNRECOGNIZED : result;
+    }
+    /**
+     * <pre>
+     * Tensor type.
+     * </pre>
+     *
+     * <code>.tensorflow.DataType dtype = 5;</code>
+     */
+    public Builder setDtype(org.tensorflow.proto.framework.DataType value) {
+      if (value == null) {
+        throw new NullPointerException();
+      }
+      
+      dtype_ = value.getNumber();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Tensor type.
+     * </pre>
+     *
+     * <code>.tensorflow.DataType dtype = 5;</code>
+     */
+    public Builder clearDtype() {
+      
+      dtype_ = 0;
+      onChanged();
+      return this;
+    }
+
+    private java.util.List<org.tensorflow.proto.framework.ResourceDtypeAndShape> resourceDtypesAndShapes_ =
+      java.util.Collections.emptyList();
+    private void ensureResourceDtypesAndShapesIsMutable() {
+      if (!((bitField0_ & 0x00000001) != 0)) {
+        resourceDtypesAndShapes_ = new java.util.ArrayList<org.tensorflow.proto.framework.ResourceDtypeAndShape>(resourceDtypesAndShapes_);
+        bitField0_ |= 0x00000001;
+       }
+    }
+
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.framework.ResourceDtypeAndShape, org.tensorflow.proto.framework.ResourceDtypeAndShape.Builder, org.tensorflow.proto.framework.ResourceDtypeAndShapeOrBuilder> resourceDtypesAndShapesBuilder_;
+
+    /**
+     * <pre>
+     * Optional data types and shapes of a remote resource variable.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.eager.ResourceDtypeAndShape resource_dtypes_and_shapes = 6;</code>
+     */
+    public java.util.List<org.tensorflow.proto.framework.ResourceDtypeAndShape> getResourceDtypesAndShapesList() {
+      if (resourceDtypesAndShapesBuilder_ == null) {
+        return java.util.Collections.unmodifiableList(resourceDtypesAndShapes_);
+      } else {
+        return resourceDtypesAndShapesBuilder_.getMessageList();
+      }
+    }
+    /**
+     * <pre>
+     * Optional data types and shapes of a remote resource variable.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.eager.ResourceDtypeAndShape resource_dtypes_and_shapes = 6;</code>
+     */
+    public int getResourceDtypesAndShapesCount() {
+      if (resourceDtypesAndShapesBuilder_ == null) {
+        return resourceDtypesAndShapes_.size();
+      } else {
+        return resourceDtypesAndShapesBuilder_.getCount();
+      }
+    }
+    /**
+     * <pre>
+     * Optional data types and shapes of a remote resource variable.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.eager.ResourceDtypeAndShape resource_dtypes_and_shapes = 6;</code>
+     */
+    public org.tensorflow.proto.framework.ResourceDtypeAndShape getResourceDtypesAndShapes(int index) {
+      if (resourceDtypesAndShapesBuilder_ == null) {
+        return resourceDtypesAndShapes_.get(index);
+      } else {
+        return resourceDtypesAndShapesBuilder_.getMessage(index);
+      }
+    }
+    /**
+     * <pre>
+     * Optional data types and shapes of a remote resource variable.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.eager.ResourceDtypeAndShape resource_dtypes_and_shapes = 6;</code>
+     */
+    public Builder setResourceDtypesAndShapes(
+        int index, org.tensorflow.proto.framework.ResourceDtypeAndShape value) {
+      if (resourceDtypesAndShapesBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureResourceDtypesAndShapesIsMutable();
+        resourceDtypesAndShapes_.set(index, value);
+        onChanged();
+      } else {
+        resourceDtypesAndShapesBuilder_.setMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Optional data types and shapes of a remote resource variable.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.eager.ResourceDtypeAndShape resource_dtypes_and_shapes = 6;</code>
+     */
+    public Builder setResourceDtypesAndShapes(
+        int index, org.tensorflow.proto.framework.ResourceDtypeAndShape.Builder builderForValue) {
+      if (resourceDtypesAndShapesBuilder_ == null) {
+        ensureResourceDtypesAndShapesIsMutable();
+        resourceDtypesAndShapes_.set(index, builderForValue.build());
+        onChanged();
+      } else {
+        resourceDtypesAndShapesBuilder_.setMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Optional data types and shapes of a remote resource variable.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.eager.ResourceDtypeAndShape resource_dtypes_and_shapes = 6;</code>
+     */
+    public Builder addResourceDtypesAndShapes(org.tensorflow.proto.framework.ResourceDtypeAndShape value) {
+      if (resourceDtypesAndShapesBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureResourceDtypesAndShapesIsMutable();
+        resourceDtypesAndShapes_.add(value);
+        onChanged();
+      } else {
+        resourceDtypesAndShapesBuilder_.addMessage(value);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Optional data types and shapes of a remote resource variable.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.eager.ResourceDtypeAndShape resource_dtypes_and_shapes = 6;</code>
+     */
+    public Builder addResourceDtypesAndShapes(
+        int index, org.tensorflow.proto.framework.ResourceDtypeAndShape value) {
+      if (resourceDtypesAndShapesBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureResourceDtypesAndShapesIsMutable();
+        resourceDtypesAndShapes_.add(index, value);
+        onChanged();
+      } else {
+        resourceDtypesAndShapesBuilder_.addMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Optional data types and shapes of a remote resource variable.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.eager.ResourceDtypeAndShape resource_dtypes_and_shapes = 6;</code>
+     */
+    public Builder addResourceDtypesAndShapes(
+        org.tensorflow.proto.framework.ResourceDtypeAndShape.Builder builderForValue) {
+      if (resourceDtypesAndShapesBuilder_ == null) {
+        ensureResourceDtypesAndShapesIsMutable();
+        resourceDtypesAndShapes_.add(builderForValue.build());
+        onChanged();
+      } else {
+        resourceDtypesAndShapesBuilder_.addMessage(builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Optional data types and shapes of a remote resource variable.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.eager.ResourceDtypeAndShape resource_dtypes_and_shapes = 6;</code>
+     */
+    public Builder addResourceDtypesAndShapes(
+        int index, org.tensorflow.proto.framework.ResourceDtypeAndShape.Builder builderForValue) {
+      if (resourceDtypesAndShapesBuilder_ == null) {
+        ensureResourceDtypesAndShapesIsMutable();
+        resourceDtypesAndShapes_.add(index, builderForValue.build());
+        onChanged();
+      } else {
+        resourceDtypesAndShapesBuilder_.addMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Optional data types and shapes of a remote resource variable.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.eager.ResourceDtypeAndShape resource_dtypes_and_shapes = 6;</code>
+     */
+    public Builder addAllResourceDtypesAndShapes(
+        java.lang.Iterable<? extends org.tensorflow.proto.framework.ResourceDtypeAndShape> values) {
+      if (resourceDtypesAndShapesBuilder_ == null) {
+        ensureResourceDtypesAndShapesIsMutable();
+        com.google.protobuf.AbstractMessageLite.Builder.addAll(
+            values, resourceDtypesAndShapes_);
+        onChanged();
+      } else {
+        resourceDtypesAndShapesBuilder_.addAllMessages(values);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Optional data types and shapes of a remote resource variable.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.eager.ResourceDtypeAndShape resource_dtypes_and_shapes = 6;</code>
+     */
+    public Builder clearResourceDtypesAndShapes() {
+      if (resourceDtypesAndShapesBuilder_ == null) {
+        resourceDtypesAndShapes_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000001);
+        onChanged();
+      } else {
+        resourceDtypesAndShapesBuilder_.clear();
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Optional data types and shapes of a remote resource variable.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.eager.ResourceDtypeAndShape resource_dtypes_and_shapes = 6;</code>
+     */
+    public Builder removeResourceDtypesAndShapes(int index) {
+      if (resourceDtypesAndShapesBuilder_ == null) {
+        ensureResourceDtypesAndShapesIsMutable();
+        resourceDtypesAndShapes_.remove(index);
+        onChanged();
+      } else {
+        resourceDtypesAndShapesBuilder_.remove(index);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Optional data types and shapes of a remote resource variable.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.eager.ResourceDtypeAndShape resource_dtypes_and_shapes = 6;</code>
+     */
+    public org.tensorflow.proto.framework.ResourceDtypeAndShape.Builder getResourceDtypesAndShapesBuilder(
+        int index) {
+      return getResourceDtypesAndShapesFieldBuilder().getBuilder(index);
+    }
+    /**
+     * <pre>
+     * Optional data types and shapes of a remote resource variable.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.eager.ResourceDtypeAndShape resource_dtypes_and_shapes = 6;</code>
+     */
+    public org.tensorflow.proto.framework.ResourceDtypeAndShapeOrBuilder getResourceDtypesAndShapesOrBuilder(
+        int index) {
+      if (resourceDtypesAndShapesBuilder_ == null) {
+        return resourceDtypesAndShapes_.get(index);  } else {
+        return resourceDtypesAndShapesBuilder_.getMessageOrBuilder(index);
+      }
+    }
+    /**
+     * <pre>
+     * Optional data types and shapes of a remote resource variable.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.eager.ResourceDtypeAndShape resource_dtypes_and_shapes = 6;</code>
+     */
+    public java.util.List<? extends org.tensorflow.proto.framework.ResourceDtypeAndShapeOrBuilder> 
+         getResourceDtypesAndShapesOrBuilderList() {
+      if (resourceDtypesAndShapesBuilder_ != null) {
+        return resourceDtypesAndShapesBuilder_.getMessageOrBuilderList();
+      } else {
+        return java.util.Collections.unmodifiableList(resourceDtypesAndShapes_);
+      }
+    }
+    /**
+     * <pre>
+     * Optional data types and shapes of a remote resource variable.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.eager.ResourceDtypeAndShape resource_dtypes_and_shapes = 6;</code>
+     */
+    public org.tensorflow.proto.framework.ResourceDtypeAndShape.Builder addResourceDtypesAndShapesBuilder() {
+      return getResourceDtypesAndShapesFieldBuilder().addBuilder(
+          org.tensorflow.proto.framework.ResourceDtypeAndShape.getDefaultInstance());
+    }
+    /**
+     * <pre>
+     * Optional data types and shapes of a remote resource variable.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.eager.ResourceDtypeAndShape resource_dtypes_and_shapes = 6;</code>
+     */
+    public org.tensorflow.proto.framework.ResourceDtypeAndShape.Builder addResourceDtypesAndShapesBuilder(
+        int index) {
+      return getResourceDtypesAndShapesFieldBuilder().addBuilder(
+          index, org.tensorflow.proto.framework.ResourceDtypeAndShape.getDefaultInstance());
+    }
+    /**
+     * <pre>
+     * Optional data types and shapes of a remote resource variable.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.eager.ResourceDtypeAndShape resource_dtypes_and_shapes = 6;</code>
+     */
+    public java.util.List<org.tensorflow.proto.framework.ResourceDtypeAndShape.Builder> 
+         getResourceDtypesAndShapesBuilderList() {
+      return getResourceDtypesAndShapesFieldBuilder().getBuilderList();
+    }
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.framework.ResourceDtypeAndShape, org.tensorflow.proto.framework.ResourceDtypeAndShape.Builder, org.tensorflow.proto.framework.ResourceDtypeAndShapeOrBuilder> 
+        getResourceDtypesAndShapesFieldBuilder() {
+      if (resourceDtypesAndShapesBuilder_ == null) {
+        resourceDtypesAndShapesBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
+            org.tensorflow.proto.framework.ResourceDtypeAndShape, org.tensorflow.proto.framework.ResourceDtypeAndShape.Builder, org.tensorflow.proto.framework.ResourceDtypeAndShapeOrBuilder>(
+                resourceDtypesAndShapes_,
+                ((bitField0_ & 0x00000001) != 0),
+                getParentForChildren(),
+                isClean());
+        resourceDtypesAndShapes_ = null;
+      }
+      return resourceDtypesAndShapesBuilder_;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.eager.RemoteTensorHandle)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.eager.RemoteTensorHandle)
+  private static final org.tensorflow.proto.framework.RemoteTensorHandle DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.RemoteTensorHandle();
+  }
+
+  public static org.tensorflow.proto.framework.RemoteTensorHandle getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<RemoteTensorHandle>
+      PARSER = new com.google.protobuf.AbstractParser<RemoteTensorHandle>() {
+    @java.lang.Override
+    public RemoteTensorHandle parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new RemoteTensorHandle(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<RemoteTensorHandle> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<RemoteTensorHandle> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.framework.RemoteTensorHandle getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/RemoteTensorHandleOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/RemoteTensorHandleOrBuilder.java
new file mode 100644
index 00000000000..e9f1098f494
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/RemoteTensorHandleOrBuilder.java
@@ -0,0 +1,128 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/remote_tensor_handle.proto
+
+package org.tensorflow.proto.framework;
+
+public interface RemoteTensorHandleOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.eager.RemoteTensorHandle)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <pre>
+   * The ID of the operation that produced this tensor.
+   * </pre>
+   *
+   * <code>int64 op_id = 1;</code>
+   */
+  long getOpId();
+
+  /**
+   * <pre>
+   * The index into the outputs of the operation that produced this tensor.
+   * </pre>
+   *
+   * <code>int32 output_num = 2;</code>
+   */
+  int getOutputNum();
+
+  /**
+   * <pre>
+   * Device where the tensor is located. Cannot be empty.
+   * For multi-device functions, it's the default device passed to placer.
+   * </pre>
+   *
+   * <code>string device = 3;</code>
+   */
+  java.lang.String getDevice();
+  /**
+   * <pre>
+   * Device where the tensor is located. Cannot be empty.
+   * For multi-device functions, it's the default device passed to placer.
+   * </pre>
+   *
+   * <code>string device = 3;</code>
+   */
+  com.google.protobuf.ByteString
+      getDeviceBytes();
+
+  /**
+   * <pre>
+   * Device of the operation producing this tensor. Can be empty if the
+   * operation producing this tensor is a multi-device function.
+   * </pre>
+   *
+   * <code>string op_device = 4;</code>
+   */
+  java.lang.String getOpDevice();
+  /**
+   * <pre>
+   * Device of the operation producing this tensor. Can be empty if the
+   * operation producing this tensor is a multi-device function.
+   * </pre>
+   *
+   * <code>string op_device = 4;</code>
+   */
+  com.google.protobuf.ByteString
+      getOpDeviceBytes();
+
+  /**
+   * <pre>
+   * Tensor type.
+   * </pre>
+   *
+   * <code>.tensorflow.DataType dtype = 5;</code>
+   */
+  int getDtypeValue();
+  /**
+   * <pre>
+   * Tensor type.
+   * </pre>
+   *
+   * <code>.tensorflow.DataType dtype = 5;</code>
+   */
+  org.tensorflow.proto.framework.DataType getDtype();
+
+  /**
+   * <pre>
+   * Optional data types and shapes of a remote resource variable.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.eager.ResourceDtypeAndShape resource_dtypes_and_shapes = 6;</code>
+   */
+  java.util.List<org.tensorflow.proto.framework.ResourceDtypeAndShape> 
+      getResourceDtypesAndShapesList();
+  /**
+   * <pre>
+   * Optional data types and shapes of a remote resource variable.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.eager.ResourceDtypeAndShape resource_dtypes_and_shapes = 6;</code>
+   */
+  org.tensorflow.proto.framework.ResourceDtypeAndShape getResourceDtypesAndShapes(int index);
+  /**
+   * <pre>
+   * Optional data types and shapes of a remote resource variable.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.eager.ResourceDtypeAndShape resource_dtypes_and_shapes = 6;</code>
+   */
+  int getResourceDtypesAndShapesCount();
+  /**
+   * <pre>
+   * Optional data types and shapes of a remote resource variable.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.eager.ResourceDtypeAndShape resource_dtypes_and_shapes = 6;</code>
+   */
+  java.util.List<? extends org.tensorflow.proto.framework.ResourceDtypeAndShapeOrBuilder> 
+      getResourceDtypesAndShapesOrBuilderList();
+  /**
+   * <pre>
+   * Optional data types and shapes of a remote resource variable.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.eager.ResourceDtypeAndShape resource_dtypes_and_shapes = 6;</code>
+   */
+  org.tensorflow.proto.framework.ResourceDtypeAndShapeOrBuilder getResourceDtypesAndShapesOrBuilder(
+      int index);
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/RemoteTensorHandleProtos.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/RemoteTensorHandleProtos.java
new file mode 100644
index 00000000000..82e7ea5bc2c
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/RemoteTensorHandleProtos.java
@@ -0,0 +1,76 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/remote_tensor_handle.proto
+
+package org.tensorflow.proto.framework;
+
+public final class RemoteTensorHandleProtos {
+  private RemoteTensorHandleProtos() {}
+  public static void registerAllExtensions(
+      com.google.protobuf.ExtensionRegistryLite registry) {
+  }
+
+  public static void registerAllExtensions(
+      com.google.protobuf.ExtensionRegistry registry) {
+    registerAllExtensions(
+        (com.google.protobuf.ExtensionRegistryLite) registry);
+  }
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_eager_ResourceDtypeAndShape_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_eager_ResourceDtypeAndShape_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_eager_RemoteTensorHandle_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_eager_RemoteTensorHandle_fieldAccessorTable;
+
+  public static com.google.protobuf.Descriptors.FileDescriptor
+      getDescriptor() {
+    return descriptor;
+  }
+  private static  com.google.protobuf.Descriptors.FileDescriptor
+      descriptor;
+  static {
+    java.lang.String[] descriptorData = {
+      "\n3tensorflow/core/protobuf/remote_tensor" +
+      "_handle.proto\022\020tensorflow.eager\032,tensorf" +
+      "low/core/framework/tensor_shape.proto\032%t" +
+      "ensorflow/core/framework/types.proto\"i\n\025" +
+      "ResourceDtypeAndShape\022#\n\005dtype\030\001 \001(\0162\024.t" +
+      "ensorflow.DataType\022+\n\005shape\030\002 \001(\0132\034.tens" +
+      "orflow.TensorShapeProto\"\314\001\n\022RemoteTensor" +
+      "Handle\022\r\n\005op_id\030\001 \001(\003\022\022\n\noutput_num\030\002 \001(" +
+      "\005\022\016\n\006device\030\003 \001(\t\022\021\n\top_device\030\004 \001(\t\022#\n\005" +
+      "dtype\030\005 \001(\0162\024.tensorflow.DataType\022K\n\032res" +
+      "ource_dtypes_and_shapes\030\006 \003(\0132\'.tensorfl" +
+      "ow.eager.ResourceDtypeAndShapeB\211\001\n\036org.t" +
+      "ensorflow.proto.frameworkB\030RemoteTensorH" +
+      "andleProtosP\001ZHgithub.com/tensorflow/ten" +
+      "sorflow/tensorflow/go/core/core_protos_g" +
+      "o_proto\370\001\001b\006proto3"
+    };
+    descriptor = com.google.protobuf.Descriptors.FileDescriptor
+      .internalBuildGeneratedFileFrom(descriptorData,
+        new com.google.protobuf.Descriptors.FileDescriptor[] {
+          org.tensorflow.proto.framework.TensorShapeProtos.getDescriptor(),
+          org.tensorflow.proto.framework.TypesProtos.getDescriptor(),
+        });
+    internal_static_tensorflow_eager_ResourceDtypeAndShape_descriptor =
+      getDescriptor().getMessageTypes().get(0);
+    internal_static_tensorflow_eager_ResourceDtypeAndShape_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_eager_ResourceDtypeAndShape_descriptor,
+        new java.lang.String[] { "Dtype", "Shape", });
+    internal_static_tensorflow_eager_RemoteTensorHandle_descriptor =
+      getDescriptor().getMessageTypes().get(1);
+    internal_static_tensorflow_eager_RemoteTensorHandle_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_eager_RemoteTensorHandle_descriptor,
+        new java.lang.String[] { "OpId", "OutputNum", "Device", "OpDevice", "Dtype", "ResourceDtypesAndShapes", });
+    org.tensorflow.proto.framework.TensorShapeProtos.getDescriptor();
+    org.tensorflow.proto.framework.TypesProtos.getDescriptor();
+  }
+
+  // @@protoc_insertion_point(outer_class_scope)
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ResourceDtypeAndShape.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ResourceDtypeAndShape.java
new file mode 100644
index 00000000000..2586a7e626a
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ResourceDtypeAndShape.java
@@ -0,0 +1,685 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/remote_tensor_handle.proto
+
+package org.tensorflow.proto.framework;
+
+/**
+ * Protobuf type {@code tensorflow.eager.ResourceDtypeAndShape}
+ */
+public  final class ResourceDtypeAndShape extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.eager.ResourceDtypeAndShape)
+    ResourceDtypeAndShapeOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use ResourceDtypeAndShape.newBuilder() to construct.
+  private ResourceDtypeAndShape(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private ResourceDtypeAndShape() {
+    dtype_ = 0;
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new ResourceDtypeAndShape();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private ResourceDtypeAndShape(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 8: {
+            int rawValue = input.readEnum();
+
+            dtype_ = rawValue;
+            break;
+          }
+          case 18: {
+            org.tensorflow.proto.framework.TensorShapeProto.Builder subBuilder = null;
+            if (shape_ != null) {
+              subBuilder = shape_.toBuilder();
+            }
+            shape_ = input.readMessage(org.tensorflow.proto.framework.TensorShapeProto.parser(), extensionRegistry);
+            if (subBuilder != null) {
+              subBuilder.mergeFrom(shape_);
+              shape_ = subBuilder.buildPartial();
+            }
+
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.framework.RemoteTensorHandleProtos.internal_static_tensorflow_eager_ResourceDtypeAndShape_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.framework.RemoteTensorHandleProtos.internal_static_tensorflow_eager_ResourceDtypeAndShape_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.framework.ResourceDtypeAndShape.class, org.tensorflow.proto.framework.ResourceDtypeAndShape.Builder.class);
+  }
+
+  public static final int DTYPE_FIELD_NUMBER = 1;
+  private int dtype_;
+  /**
+   * <code>.tensorflow.DataType dtype = 1;</code>
+   */
+  public int getDtypeValue() {
+    return dtype_;
+  }
+  /**
+   * <code>.tensorflow.DataType dtype = 1;</code>
+   */
+  public org.tensorflow.proto.framework.DataType getDtype() {
+    @SuppressWarnings("deprecation")
+    org.tensorflow.proto.framework.DataType result = org.tensorflow.proto.framework.DataType.valueOf(dtype_);
+    return result == null ? org.tensorflow.proto.framework.DataType.UNRECOGNIZED : result;
+  }
+
+  public static final int SHAPE_FIELD_NUMBER = 2;
+  private org.tensorflow.proto.framework.TensorShapeProto shape_;
+  /**
+   * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+   */
+  public boolean hasShape() {
+    return shape_ != null;
+  }
+  /**
+   * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+   */
+  public org.tensorflow.proto.framework.TensorShapeProto getShape() {
+    return shape_ == null ? org.tensorflow.proto.framework.TensorShapeProto.getDefaultInstance() : shape_;
+  }
+  /**
+   * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+   */
+  public org.tensorflow.proto.framework.TensorShapeProtoOrBuilder getShapeOrBuilder() {
+    return getShape();
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    if (dtype_ != org.tensorflow.proto.framework.DataType.DT_INVALID.getNumber()) {
+      output.writeEnum(1, dtype_);
+    }
+    if (shape_ != null) {
+      output.writeMessage(2, getShape());
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (dtype_ != org.tensorflow.proto.framework.DataType.DT_INVALID.getNumber()) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeEnumSize(1, dtype_);
+    }
+    if (shape_ != null) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(2, getShape());
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.framework.ResourceDtypeAndShape)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.framework.ResourceDtypeAndShape other = (org.tensorflow.proto.framework.ResourceDtypeAndShape) obj;
+
+    if (dtype_ != other.dtype_) return false;
+    if (hasShape() != other.hasShape()) return false;
+    if (hasShape()) {
+      if (!getShape()
+          .equals(other.getShape())) return false;
+    }
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    hash = (37 * hash) + DTYPE_FIELD_NUMBER;
+    hash = (53 * hash) + dtype_;
+    if (hasShape()) {
+      hash = (37 * hash) + SHAPE_FIELD_NUMBER;
+      hash = (53 * hash) + getShape().hashCode();
+    }
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.framework.ResourceDtypeAndShape parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.ResourceDtypeAndShape parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.ResourceDtypeAndShape parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.ResourceDtypeAndShape parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.ResourceDtypeAndShape parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.ResourceDtypeAndShape parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.ResourceDtypeAndShape parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.ResourceDtypeAndShape parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.ResourceDtypeAndShape parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.ResourceDtypeAndShape parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.ResourceDtypeAndShape parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.ResourceDtypeAndShape parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.framework.ResourceDtypeAndShape prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * Protobuf type {@code tensorflow.eager.ResourceDtypeAndShape}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.eager.ResourceDtypeAndShape)
+      org.tensorflow.proto.framework.ResourceDtypeAndShapeOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.RemoteTensorHandleProtos.internal_static_tensorflow_eager_ResourceDtypeAndShape_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.RemoteTensorHandleProtos.internal_static_tensorflow_eager_ResourceDtypeAndShape_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.ResourceDtypeAndShape.class, org.tensorflow.proto.framework.ResourceDtypeAndShape.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.framework.ResourceDtypeAndShape.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      dtype_ = 0;
+
+      if (shapeBuilder_ == null) {
+        shape_ = null;
+      } else {
+        shape_ = null;
+        shapeBuilder_ = null;
+      }
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.framework.RemoteTensorHandleProtos.internal_static_tensorflow_eager_ResourceDtypeAndShape_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.ResourceDtypeAndShape getDefaultInstanceForType() {
+      return org.tensorflow.proto.framework.ResourceDtypeAndShape.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.ResourceDtypeAndShape build() {
+      org.tensorflow.proto.framework.ResourceDtypeAndShape result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.ResourceDtypeAndShape buildPartial() {
+      org.tensorflow.proto.framework.ResourceDtypeAndShape result = new org.tensorflow.proto.framework.ResourceDtypeAndShape(this);
+      result.dtype_ = dtype_;
+      if (shapeBuilder_ == null) {
+        result.shape_ = shape_;
+      } else {
+        result.shape_ = shapeBuilder_.build();
+      }
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.framework.ResourceDtypeAndShape) {
+        return mergeFrom((org.tensorflow.proto.framework.ResourceDtypeAndShape)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.framework.ResourceDtypeAndShape other) {
+      if (other == org.tensorflow.proto.framework.ResourceDtypeAndShape.getDefaultInstance()) return this;
+      if (other.dtype_ != 0) {
+        setDtypeValue(other.getDtypeValue());
+      }
+      if (other.hasShape()) {
+        mergeShape(other.getShape());
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.framework.ResourceDtypeAndShape parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.framework.ResourceDtypeAndShape) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+
+    private int dtype_ = 0;
+    /**
+     * <code>.tensorflow.DataType dtype = 1;</code>
+     */
+    public int getDtypeValue() {
+      return dtype_;
+    }
+    /**
+     * <code>.tensorflow.DataType dtype = 1;</code>
+     */
+    public Builder setDtypeValue(int value) {
+      dtype_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>.tensorflow.DataType dtype = 1;</code>
+     */
+    public org.tensorflow.proto.framework.DataType getDtype() {
+      @SuppressWarnings("deprecation")
+      org.tensorflow.proto.framework.DataType result = org.tensorflow.proto.framework.DataType.valueOf(dtype_);
+      return result == null ? org.tensorflow.proto.framework.DataType.UNRECOGNIZED : result;
+    }
+    /**
+     * <code>.tensorflow.DataType dtype = 1;</code>
+     */
+    public Builder setDtype(org.tensorflow.proto.framework.DataType value) {
+      if (value == null) {
+        throw new NullPointerException();
+      }
+      
+      dtype_ = value.getNumber();
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>.tensorflow.DataType dtype = 1;</code>
+     */
+    public Builder clearDtype() {
+      
+      dtype_ = 0;
+      onChanged();
+      return this;
+    }
+
+    private org.tensorflow.proto.framework.TensorShapeProto shape_;
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.TensorShapeProto, org.tensorflow.proto.framework.TensorShapeProto.Builder, org.tensorflow.proto.framework.TensorShapeProtoOrBuilder> shapeBuilder_;
+    /**
+     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+     */
+    public boolean hasShape() {
+      return shapeBuilder_ != null || shape_ != null;
+    }
+    /**
+     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+     */
+    public org.tensorflow.proto.framework.TensorShapeProto getShape() {
+      if (shapeBuilder_ == null) {
+        return shape_ == null ? org.tensorflow.proto.framework.TensorShapeProto.getDefaultInstance() : shape_;
+      } else {
+        return shapeBuilder_.getMessage();
+      }
+    }
+    /**
+     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+     */
+    public Builder setShape(org.tensorflow.proto.framework.TensorShapeProto value) {
+      if (shapeBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        shape_ = value;
+        onChanged();
+      } else {
+        shapeBuilder_.setMessage(value);
+      }
+
+      return this;
+    }
+    /**
+     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+     */
+    public Builder setShape(
+        org.tensorflow.proto.framework.TensorShapeProto.Builder builderForValue) {
+      if (shapeBuilder_ == null) {
+        shape_ = builderForValue.build();
+        onChanged();
+      } else {
+        shapeBuilder_.setMessage(builderForValue.build());
+      }
+
+      return this;
+    }
+    /**
+     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+     */
+    public Builder mergeShape(org.tensorflow.proto.framework.TensorShapeProto value) {
+      if (shapeBuilder_ == null) {
+        if (shape_ != null) {
+          shape_ =
+            org.tensorflow.proto.framework.TensorShapeProto.newBuilder(shape_).mergeFrom(value).buildPartial();
+        } else {
+          shape_ = value;
+        }
+        onChanged();
+      } else {
+        shapeBuilder_.mergeFrom(value);
+      }
+
+      return this;
+    }
+    /**
+     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+     */
+    public Builder clearShape() {
+      if (shapeBuilder_ == null) {
+        shape_ = null;
+        onChanged();
+      } else {
+        shape_ = null;
+        shapeBuilder_ = null;
+      }
+
+      return this;
+    }
+    /**
+     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+     */
+    public org.tensorflow.proto.framework.TensorShapeProto.Builder getShapeBuilder() {
+      
+      onChanged();
+      return getShapeFieldBuilder().getBuilder();
+    }
+    /**
+     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+     */
+    public org.tensorflow.proto.framework.TensorShapeProtoOrBuilder getShapeOrBuilder() {
+      if (shapeBuilder_ != null) {
+        return shapeBuilder_.getMessageOrBuilder();
+      } else {
+        return shape_ == null ?
+            org.tensorflow.proto.framework.TensorShapeProto.getDefaultInstance() : shape_;
+      }
+    }
+    /**
+     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+     */
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.TensorShapeProto, org.tensorflow.proto.framework.TensorShapeProto.Builder, org.tensorflow.proto.framework.TensorShapeProtoOrBuilder> 
+        getShapeFieldBuilder() {
+      if (shapeBuilder_ == null) {
+        shapeBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+            org.tensorflow.proto.framework.TensorShapeProto, org.tensorflow.proto.framework.TensorShapeProto.Builder, org.tensorflow.proto.framework.TensorShapeProtoOrBuilder>(
+                getShape(),
+                getParentForChildren(),
+                isClean());
+        shape_ = null;
+      }
+      return shapeBuilder_;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.eager.ResourceDtypeAndShape)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.eager.ResourceDtypeAndShape)
+  private static final org.tensorflow.proto.framework.ResourceDtypeAndShape DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.ResourceDtypeAndShape();
+  }
+
+  public static org.tensorflow.proto.framework.ResourceDtypeAndShape getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<ResourceDtypeAndShape>
+      PARSER = new com.google.protobuf.AbstractParser<ResourceDtypeAndShape>() {
+    @java.lang.Override
+    public ResourceDtypeAndShape parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new ResourceDtypeAndShape(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<ResourceDtypeAndShape> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<ResourceDtypeAndShape> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.framework.ResourceDtypeAndShape getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ResourceDtypeAndShapeOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ResourceDtypeAndShapeOrBuilder.java
new file mode 100644
index 00000000000..67ab4da1b8d
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ResourceDtypeAndShapeOrBuilder.java
@@ -0,0 +1,31 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/remote_tensor_handle.proto
+
+package org.tensorflow.proto.framework;
+
+public interface ResourceDtypeAndShapeOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.eager.ResourceDtypeAndShape)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <code>.tensorflow.DataType dtype = 1;</code>
+   */
+  int getDtypeValue();
+  /**
+   * <code>.tensorflow.DataType dtype = 1;</code>
+   */
+  org.tensorflow.proto.framework.DataType getDtype();
+
+  /**
+   * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+   */
+  boolean hasShape();
+  /**
+   * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+   */
+  org.tensorflow.proto.framework.TensorShapeProto getShape();
+  /**
+   * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+   */
+  org.tensorflow.proto.framework.TensorShapeProtoOrBuilder getShapeOrBuilder();
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ResourceHandle.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ResourceHandle.java
new file mode 100644
index 00000000000..5a96505f0eb
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ResourceHandle.java
@@ -0,0 +1,75 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/framework/resource_handle.proto
+
+package org.tensorflow.proto.framework;
+
+public final class ResourceHandle {
+  private ResourceHandle() {}
+  public static void registerAllExtensions(
+      com.google.protobuf.ExtensionRegistryLite registry) {
+  }
+
+  public static void registerAllExtensions(
+      com.google.protobuf.ExtensionRegistry registry) {
+    registerAllExtensions(
+        (com.google.protobuf.ExtensionRegistryLite) registry);
+  }
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_ResourceHandleProto_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_ResourceHandleProto_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_ResourceHandleProto_DtypeAndShape_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_ResourceHandleProto_DtypeAndShape_fieldAccessorTable;
+
+  public static com.google.protobuf.Descriptors.FileDescriptor
+      getDescriptor() {
+    return descriptor;
+  }
+  private static  com.google.protobuf.Descriptors.FileDescriptor
+      descriptor;
+  static {
+    java.lang.String[] descriptorData = {
+      "\n/tensorflow/core/framework/resource_han" +
+      "dle.proto\022\ntensorflow\032,tensorflow/core/f" +
+      "ramework/tensor_shape.proto\032%tensorflow/" +
+      "core/framework/types.proto\"\245\002\n\023ResourceH" +
+      "andleProto\022\016\n\006device\030\001 \001(\t\022\021\n\tcontainer\030" +
+      "\002 \001(\t\022\014\n\004name\030\003 \001(\t\022\021\n\thash_code\030\004 \001(\004\022\027" +
+      "\n\017maybe_type_name\030\005 \001(\t\022H\n\021dtypes_and_sh" +
+      "apes\030\006 \003(\0132-.tensorflow.ResourceHandlePr" +
+      "oto.DtypeAndShape\032a\n\rDtypeAndShape\022#\n\005dt" +
+      "ype\030\001 \001(\0162\024.tensorflow.DataType\022+\n\005shape" +
+      "\030\002 \001(\0132\034.tensorflow.TensorShapeProtoJ\004\010\007" +
+      "\020\010B\215\001\n\036org.tensorflow.proto.frameworkB\016R" +
+      "esourceHandleP\001ZVgithub.com/tensorflow/t" +
+      "ensorflow/tensorflow/go/core/framework/r" +
+      "esource_handle_go_proto\370\001\001b\006proto3"
+    };
+    descriptor = com.google.protobuf.Descriptors.FileDescriptor
+      .internalBuildGeneratedFileFrom(descriptorData,
+        new com.google.protobuf.Descriptors.FileDescriptor[] {
+          org.tensorflow.proto.framework.TensorShapeProtos.getDescriptor(),
+          org.tensorflow.proto.framework.TypesProtos.getDescriptor(),
+        });
+    internal_static_tensorflow_ResourceHandleProto_descriptor =
+      getDescriptor().getMessageTypes().get(0);
+    internal_static_tensorflow_ResourceHandleProto_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_ResourceHandleProto_descriptor,
+        new java.lang.String[] { "Device", "Container", "Name", "HashCode", "MaybeTypeName", "DtypesAndShapes", });
+    internal_static_tensorflow_ResourceHandleProto_DtypeAndShape_descriptor =
+      internal_static_tensorflow_ResourceHandleProto_descriptor.getNestedTypes().get(0);
+    internal_static_tensorflow_ResourceHandleProto_DtypeAndShape_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_ResourceHandleProto_DtypeAndShape_descriptor,
+        new java.lang.String[] { "Dtype", "Shape", });
+    org.tensorflow.proto.framework.TensorShapeProtos.getDescriptor();
+    org.tensorflow.proto.framework.TypesProtos.getDescriptor();
+  }
+
+  // @@protoc_insertion_point(outer_class_scope)
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ResourceHandleProto.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ResourceHandleProto.java
new file mode 100644
index 00000000000..3d31206b794
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ResourceHandleProto.java
@@ -0,0 +1,2288 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/framework/resource_handle.proto
+
+package org.tensorflow.proto.framework;
+
+/**
+ * <pre>
+ * Protocol buffer representing a handle to a tensorflow resource. Handles are
+ * not valid across executions, but can be serialized back and forth from within
+ * a single run.
+ * </pre>
+ *
+ * Protobuf type {@code tensorflow.ResourceHandleProto}
+ */
+public  final class ResourceHandleProto extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.ResourceHandleProto)
+    ResourceHandleProtoOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use ResourceHandleProto.newBuilder() to construct.
+  private ResourceHandleProto(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private ResourceHandleProto() {
+    device_ = "";
+    container_ = "";
+    name_ = "";
+    maybeTypeName_ = "";
+    dtypesAndShapes_ = java.util.Collections.emptyList();
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new ResourceHandleProto();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private ResourceHandleProto(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    int mutable_bitField0_ = 0;
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 10: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            device_ = s;
+            break;
+          }
+          case 18: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            container_ = s;
+            break;
+          }
+          case 26: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            name_ = s;
+            break;
+          }
+          case 32: {
+
+            hashCode_ = input.readUInt64();
+            break;
+          }
+          case 42: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            maybeTypeName_ = s;
+            break;
+          }
+          case 50: {
+            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
+              dtypesAndShapes_ = new java.util.ArrayList<org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShape>();
+              mutable_bitField0_ |= 0x00000001;
+            }
+            dtypesAndShapes_.add(
+                input.readMessage(org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShape.parser(), extensionRegistry));
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      if (((mutable_bitField0_ & 0x00000001) != 0)) {
+        dtypesAndShapes_ = java.util.Collections.unmodifiableList(dtypesAndShapes_);
+      }
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.framework.ResourceHandle.internal_static_tensorflow_ResourceHandleProto_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.framework.ResourceHandle.internal_static_tensorflow_ResourceHandleProto_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.framework.ResourceHandleProto.class, org.tensorflow.proto.framework.ResourceHandleProto.Builder.class);
+  }
+
+  public interface DtypeAndShapeOrBuilder extends
+      // @@protoc_insertion_point(interface_extends:tensorflow.ResourceHandleProto.DtypeAndShape)
+      com.google.protobuf.MessageOrBuilder {
+
+    /**
+     * <code>.tensorflow.DataType dtype = 1;</code>
+     */
+    int getDtypeValue();
+    /**
+     * <code>.tensorflow.DataType dtype = 1;</code>
+     */
+    org.tensorflow.proto.framework.DataType getDtype();
+
+    /**
+     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+     */
+    boolean hasShape();
+    /**
+     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+     */
+    org.tensorflow.proto.framework.TensorShapeProto getShape();
+    /**
+     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+     */
+    org.tensorflow.proto.framework.TensorShapeProtoOrBuilder getShapeOrBuilder();
+  }
+  /**
+   * <pre>
+   * Protocol buffer representing a pair of (data type, tensor shape).
+   * </pre>
+   *
+   * Protobuf type {@code tensorflow.ResourceHandleProto.DtypeAndShape}
+   */
+  public  static final class DtypeAndShape extends
+      com.google.protobuf.GeneratedMessageV3 implements
+      // @@protoc_insertion_point(message_implements:tensorflow.ResourceHandleProto.DtypeAndShape)
+      DtypeAndShapeOrBuilder {
+  private static final long serialVersionUID = 0L;
+    // Use DtypeAndShape.newBuilder() to construct.
+    private DtypeAndShape(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+      super(builder);
+    }
+    private DtypeAndShape() {
+      dtype_ = 0;
+    }
+
+    @java.lang.Override
+    @SuppressWarnings({"unused"})
+    protected java.lang.Object newInstance(
+        UnusedPrivateParameter unused) {
+      return new DtypeAndShape();
+    }
+
+    @java.lang.Override
+    public final com.google.protobuf.UnknownFieldSet
+    getUnknownFields() {
+      return this.unknownFields;
+    }
+    private DtypeAndShape(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      this();
+      if (extensionRegistry == null) {
+        throw new java.lang.NullPointerException();
+      }
+      com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+          com.google.protobuf.UnknownFieldSet.newBuilder();
+      try {
+        boolean done = false;
+        while (!done) {
+          int tag = input.readTag();
+          switch (tag) {
+            case 0:
+              done = true;
+              break;
+            case 8: {
+              int rawValue = input.readEnum();
+
+              dtype_ = rawValue;
+              break;
+            }
+            case 18: {
+              org.tensorflow.proto.framework.TensorShapeProto.Builder subBuilder = null;
+              if (shape_ != null) {
+                subBuilder = shape_.toBuilder();
+              }
+              shape_ = input.readMessage(org.tensorflow.proto.framework.TensorShapeProto.parser(), extensionRegistry);
+              if (subBuilder != null) {
+                subBuilder.mergeFrom(shape_);
+                shape_ = subBuilder.buildPartial();
+              }
+
+              break;
+            }
+            default: {
+              if (!parseUnknownField(
+                  input, unknownFields, extensionRegistry, tag)) {
+                done = true;
+              }
+              break;
+            }
+          }
+        }
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        throw e.setUnfinishedMessage(this);
+      } catch (java.io.IOException e) {
+        throw new com.google.protobuf.InvalidProtocolBufferException(
+            e).setUnfinishedMessage(this);
+      } finally {
+        this.unknownFields = unknownFields.build();
+        makeExtensionsImmutable();
+      }
+    }
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.ResourceHandle.internal_static_tensorflow_ResourceHandleProto_DtypeAndShape_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.ResourceHandle.internal_static_tensorflow_ResourceHandleProto_DtypeAndShape_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShape.class, org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShape.Builder.class);
+    }
+
+    public static final int DTYPE_FIELD_NUMBER = 1;
+    private int dtype_;
+    /**
+     * <code>.tensorflow.DataType dtype = 1;</code>
+     */
+    public int getDtypeValue() {
+      return dtype_;
+    }
+    /**
+     * <code>.tensorflow.DataType dtype = 1;</code>
+     */
+    public org.tensorflow.proto.framework.DataType getDtype() {
+      @SuppressWarnings("deprecation")
+      org.tensorflow.proto.framework.DataType result = org.tensorflow.proto.framework.DataType.valueOf(dtype_);
+      return result == null ? org.tensorflow.proto.framework.DataType.UNRECOGNIZED : result;
+    }
+
+    public static final int SHAPE_FIELD_NUMBER = 2;
+    private org.tensorflow.proto.framework.TensorShapeProto shape_;
+    /**
+     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+     */
+    public boolean hasShape() {
+      return shape_ != null;
+    }
+    /**
+     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+     */
+    public org.tensorflow.proto.framework.TensorShapeProto getShape() {
+      return shape_ == null ? org.tensorflow.proto.framework.TensorShapeProto.getDefaultInstance() : shape_;
+    }
+    /**
+     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+     */
+    public org.tensorflow.proto.framework.TensorShapeProtoOrBuilder getShapeOrBuilder() {
+      return getShape();
+    }
+
+    private byte memoizedIsInitialized = -1;
+    @java.lang.Override
+    public final boolean isInitialized() {
+      byte isInitialized = memoizedIsInitialized;
+      if (isInitialized == 1) return true;
+      if (isInitialized == 0) return false;
+
+      memoizedIsInitialized = 1;
+      return true;
+    }
+
+    @java.lang.Override
+    public void writeTo(com.google.protobuf.CodedOutputStream output)
+                        throws java.io.IOException {
+      if (dtype_ != org.tensorflow.proto.framework.DataType.DT_INVALID.getNumber()) {
+        output.writeEnum(1, dtype_);
+      }
+      if (shape_ != null) {
+        output.writeMessage(2, getShape());
+      }
+      unknownFields.writeTo(output);
+    }
+
+    @java.lang.Override
+    public int getSerializedSize() {
+      int size = memoizedSize;
+      if (size != -1) return size;
+
+      size = 0;
+      if (dtype_ != org.tensorflow.proto.framework.DataType.DT_INVALID.getNumber()) {
+        size += com.google.protobuf.CodedOutputStream
+          .computeEnumSize(1, dtype_);
+      }
+      if (shape_ != null) {
+        size += com.google.protobuf.CodedOutputStream
+          .computeMessageSize(2, getShape());
+      }
+      size += unknownFields.getSerializedSize();
+      memoizedSize = size;
+      return size;
+    }
+
+    @java.lang.Override
+    public boolean equals(final java.lang.Object obj) {
+      if (obj == this) {
+       return true;
+      }
+      if (!(obj instanceof org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShape)) {
+        return super.equals(obj);
+      }
+      org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShape other = (org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShape) obj;
+
+      if (dtype_ != other.dtype_) return false;
+      if (hasShape() != other.hasShape()) return false;
+      if (hasShape()) {
+        if (!getShape()
+            .equals(other.getShape())) return false;
+      }
+      if (!unknownFields.equals(other.unknownFields)) return false;
+      return true;
+    }
+
+    @java.lang.Override
+    public int hashCode() {
+      if (memoizedHashCode != 0) {
+        return memoizedHashCode;
+      }
+      int hash = 41;
+      hash = (19 * hash) + getDescriptor().hashCode();
+      hash = (37 * hash) + DTYPE_FIELD_NUMBER;
+      hash = (53 * hash) + dtype_;
+      if (hasShape()) {
+        hash = (37 * hash) + SHAPE_FIELD_NUMBER;
+        hash = (53 * hash) + getShape().hashCode();
+      }
+      hash = (29 * hash) + unknownFields.hashCode();
+      memoizedHashCode = hash;
+      return hash;
+    }
+
+    public static org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShape parseFrom(
+        java.nio.ByteBuffer data)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data);
+    }
+    public static org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShape parseFrom(
+        java.nio.ByteBuffer data,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShape parseFrom(
+        com.google.protobuf.ByteString data)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data);
+    }
+    public static org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShape parseFrom(
+        com.google.protobuf.ByteString data,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShape parseFrom(byte[] data)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data);
+    }
+    public static org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShape parseFrom(
+        byte[] data,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShape parseFrom(java.io.InputStream input)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input);
+    }
+    public static org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShape parseFrom(
+        java.io.InputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShape parseDelimitedFrom(java.io.InputStream input)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseDelimitedWithIOException(PARSER, input);
+    }
+    public static org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShape parseDelimitedFrom(
+        java.io.InputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShape parseFrom(
+        com.google.protobuf.CodedInputStream input)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input);
+    }
+    public static org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShape parseFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input, extensionRegistry);
+    }
+
+    @java.lang.Override
+    public Builder newBuilderForType() { return newBuilder(); }
+    public static Builder newBuilder() {
+      return DEFAULT_INSTANCE.toBuilder();
+    }
+    public static Builder newBuilder(org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShape prototype) {
+      return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+    }
+    @java.lang.Override
+    public Builder toBuilder() {
+      return this == DEFAULT_INSTANCE
+          ? new Builder() : new Builder().mergeFrom(this);
+    }
+
+    @java.lang.Override
+    protected Builder newBuilderForType(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      Builder builder = new Builder(parent);
+      return builder;
+    }
+    /**
+     * <pre>
+     * Protocol buffer representing a pair of (data type, tensor shape).
+     * </pre>
+     *
+     * Protobuf type {@code tensorflow.ResourceHandleProto.DtypeAndShape}
+     */
+    public static final class Builder extends
+        com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+        // @@protoc_insertion_point(builder_implements:tensorflow.ResourceHandleProto.DtypeAndShape)
+        org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShapeOrBuilder {
+      public static final com.google.protobuf.Descriptors.Descriptor
+          getDescriptor() {
+        return org.tensorflow.proto.framework.ResourceHandle.internal_static_tensorflow_ResourceHandleProto_DtypeAndShape_descriptor;
+      }
+
+      @java.lang.Override
+      protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+          internalGetFieldAccessorTable() {
+        return org.tensorflow.proto.framework.ResourceHandle.internal_static_tensorflow_ResourceHandleProto_DtypeAndShape_fieldAccessorTable
+            .ensureFieldAccessorsInitialized(
+                org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShape.class, org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShape.Builder.class);
+      }
+
+      // Construct using org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShape.newBuilder()
+      private Builder() {
+        maybeForceBuilderInitialization();
+      }
+
+      private Builder(
+          com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+        super(parent);
+        maybeForceBuilderInitialization();
+      }
+      private void maybeForceBuilderInitialization() {
+        if (com.google.protobuf.GeneratedMessageV3
+                .alwaysUseFieldBuilders) {
+        }
+      }
+      @java.lang.Override
+      public Builder clear() {
+        super.clear();
+        dtype_ = 0;
+
+        if (shapeBuilder_ == null) {
+          shape_ = null;
+        } else {
+          shape_ = null;
+          shapeBuilder_ = null;
+        }
+        return this;
+      }
+
+      @java.lang.Override
+      public com.google.protobuf.Descriptors.Descriptor
+          getDescriptorForType() {
+        return org.tensorflow.proto.framework.ResourceHandle.internal_static_tensorflow_ResourceHandleProto_DtypeAndShape_descriptor;
+      }
+
+      @java.lang.Override
+      public org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShape getDefaultInstanceForType() {
+        return org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShape.getDefaultInstance();
+      }
+
+      @java.lang.Override
+      public org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShape build() {
+        org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShape result = buildPartial();
+        if (!result.isInitialized()) {
+          throw newUninitializedMessageException(result);
+        }
+        return result;
+      }
+
+      @java.lang.Override
+      public org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShape buildPartial() {
+        org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShape result = new org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShape(this);
+        result.dtype_ = dtype_;
+        if (shapeBuilder_ == null) {
+          result.shape_ = shape_;
+        } else {
+          result.shape_ = shapeBuilder_.build();
+        }
+        onBuilt();
+        return result;
+      }
+
+      @java.lang.Override
+      public Builder clone() {
+        return super.clone();
+      }
+      @java.lang.Override
+      public Builder setField(
+          com.google.protobuf.Descriptors.FieldDescriptor field,
+          java.lang.Object value) {
+        return super.setField(field, value);
+      }
+      @java.lang.Override
+      public Builder clearField(
+          com.google.protobuf.Descriptors.FieldDescriptor field) {
+        return super.clearField(field);
+      }
+      @java.lang.Override
+      public Builder clearOneof(
+          com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+        return super.clearOneof(oneof);
+      }
+      @java.lang.Override
+      public Builder setRepeatedField(
+          com.google.protobuf.Descriptors.FieldDescriptor field,
+          int index, java.lang.Object value) {
+        return super.setRepeatedField(field, index, value);
+      }
+      @java.lang.Override
+      public Builder addRepeatedField(
+          com.google.protobuf.Descriptors.FieldDescriptor field,
+          java.lang.Object value) {
+        return super.addRepeatedField(field, value);
+      }
+      @java.lang.Override
+      public Builder mergeFrom(com.google.protobuf.Message other) {
+        if (other instanceof org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShape) {
+          return mergeFrom((org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShape)other);
+        } else {
+          super.mergeFrom(other);
+          return this;
+        }
+      }
+
+      public Builder mergeFrom(org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShape other) {
+        if (other == org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShape.getDefaultInstance()) return this;
+        if (other.dtype_ != 0) {
+          setDtypeValue(other.getDtypeValue());
+        }
+        if (other.hasShape()) {
+          mergeShape(other.getShape());
+        }
+        this.mergeUnknownFields(other.unknownFields);
+        onChanged();
+        return this;
+      }
+
+      @java.lang.Override
+      public final boolean isInitialized() {
+        return true;
+      }
+
+      @java.lang.Override
+      public Builder mergeFrom(
+          com.google.protobuf.CodedInputStream input,
+          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+          throws java.io.IOException {
+        org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShape parsedMessage = null;
+        try {
+          parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+        } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+          parsedMessage = (org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShape) e.getUnfinishedMessage();
+          throw e.unwrapIOException();
+        } finally {
+          if (parsedMessage != null) {
+            mergeFrom(parsedMessage);
+          }
+        }
+        return this;
+      }
+
+      private int dtype_ = 0;
+      /**
+       * <code>.tensorflow.DataType dtype = 1;</code>
+       */
+      public int getDtypeValue() {
+        return dtype_;
+      }
+      /**
+       * <code>.tensorflow.DataType dtype = 1;</code>
+       */
+      public Builder setDtypeValue(int value) {
+        dtype_ = value;
+        onChanged();
+        return this;
+      }
+      /**
+       * <code>.tensorflow.DataType dtype = 1;</code>
+       */
+      public org.tensorflow.proto.framework.DataType getDtype() {
+        @SuppressWarnings("deprecation")
+        org.tensorflow.proto.framework.DataType result = org.tensorflow.proto.framework.DataType.valueOf(dtype_);
+        return result == null ? org.tensorflow.proto.framework.DataType.UNRECOGNIZED : result;
+      }
+      /**
+       * <code>.tensorflow.DataType dtype = 1;</code>
+       */
+      public Builder setDtype(org.tensorflow.proto.framework.DataType value) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        
+        dtype_ = value.getNumber();
+        onChanged();
+        return this;
+      }
+      /**
+       * <code>.tensorflow.DataType dtype = 1;</code>
+       */
+      public Builder clearDtype() {
+        
+        dtype_ = 0;
+        onChanged();
+        return this;
+      }
+
+      private org.tensorflow.proto.framework.TensorShapeProto shape_;
+      private com.google.protobuf.SingleFieldBuilderV3<
+          org.tensorflow.proto.framework.TensorShapeProto, org.tensorflow.proto.framework.TensorShapeProto.Builder, org.tensorflow.proto.framework.TensorShapeProtoOrBuilder> shapeBuilder_;
+      /**
+       * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+       */
+      public boolean hasShape() {
+        return shapeBuilder_ != null || shape_ != null;
+      }
+      /**
+       * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+       */
+      public org.tensorflow.proto.framework.TensorShapeProto getShape() {
+        if (shapeBuilder_ == null) {
+          return shape_ == null ? org.tensorflow.proto.framework.TensorShapeProto.getDefaultInstance() : shape_;
+        } else {
+          return shapeBuilder_.getMessage();
+        }
+      }
+      /**
+       * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+       */
+      public Builder setShape(org.tensorflow.proto.framework.TensorShapeProto value) {
+        if (shapeBuilder_ == null) {
+          if (value == null) {
+            throw new NullPointerException();
+          }
+          shape_ = value;
+          onChanged();
+        } else {
+          shapeBuilder_.setMessage(value);
+        }
+
+        return this;
+      }
+      /**
+       * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+       */
+      public Builder setShape(
+          org.tensorflow.proto.framework.TensorShapeProto.Builder builderForValue) {
+        if (shapeBuilder_ == null) {
+          shape_ = builderForValue.build();
+          onChanged();
+        } else {
+          shapeBuilder_.setMessage(builderForValue.build());
+        }
+
+        return this;
+      }
+      /**
+       * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+       */
+      public Builder mergeShape(org.tensorflow.proto.framework.TensorShapeProto value) {
+        if (shapeBuilder_ == null) {
+          if (shape_ != null) {
+            shape_ =
+              org.tensorflow.proto.framework.TensorShapeProto.newBuilder(shape_).mergeFrom(value).buildPartial();
+          } else {
+            shape_ = value;
+          }
+          onChanged();
+        } else {
+          shapeBuilder_.mergeFrom(value);
+        }
+
+        return this;
+      }
+      /**
+       * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+       */
+      public Builder clearShape() {
+        if (shapeBuilder_ == null) {
+          shape_ = null;
+          onChanged();
+        } else {
+          shape_ = null;
+          shapeBuilder_ = null;
+        }
+
+        return this;
+      }
+      /**
+       * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+       */
+      public org.tensorflow.proto.framework.TensorShapeProto.Builder getShapeBuilder() {
+        
+        onChanged();
+        return getShapeFieldBuilder().getBuilder();
+      }
+      /**
+       * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+       */
+      public org.tensorflow.proto.framework.TensorShapeProtoOrBuilder getShapeOrBuilder() {
+        if (shapeBuilder_ != null) {
+          return shapeBuilder_.getMessageOrBuilder();
+        } else {
+          return shape_ == null ?
+              org.tensorflow.proto.framework.TensorShapeProto.getDefaultInstance() : shape_;
+        }
+      }
+      /**
+       * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+       */
+      private com.google.protobuf.SingleFieldBuilderV3<
+          org.tensorflow.proto.framework.TensorShapeProto, org.tensorflow.proto.framework.TensorShapeProto.Builder, org.tensorflow.proto.framework.TensorShapeProtoOrBuilder> 
+          getShapeFieldBuilder() {
+        if (shapeBuilder_ == null) {
+          shapeBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+              org.tensorflow.proto.framework.TensorShapeProto, org.tensorflow.proto.framework.TensorShapeProto.Builder, org.tensorflow.proto.framework.TensorShapeProtoOrBuilder>(
+                  getShape(),
+                  getParentForChildren(),
+                  isClean());
+          shape_ = null;
+        }
+        return shapeBuilder_;
+      }
+      @java.lang.Override
+      public final Builder setUnknownFields(
+          final com.google.protobuf.UnknownFieldSet unknownFields) {
+        return super.setUnknownFields(unknownFields);
+      }
+
+      @java.lang.Override
+      public final Builder mergeUnknownFields(
+          final com.google.protobuf.UnknownFieldSet unknownFields) {
+        return super.mergeUnknownFields(unknownFields);
+      }
+
+
+      // @@protoc_insertion_point(builder_scope:tensorflow.ResourceHandleProto.DtypeAndShape)
+    }
+
+    // @@protoc_insertion_point(class_scope:tensorflow.ResourceHandleProto.DtypeAndShape)
+    private static final org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShape DEFAULT_INSTANCE;
+    static {
+      DEFAULT_INSTANCE = new org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShape();
+    }
+
+    public static org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShape getDefaultInstance() {
+      return DEFAULT_INSTANCE;
+    }
+
+    private static final com.google.protobuf.Parser<DtypeAndShape>
+        PARSER = new com.google.protobuf.AbstractParser<DtypeAndShape>() {
+      @java.lang.Override
+      public DtypeAndShape parsePartialFrom(
+          com.google.protobuf.CodedInputStream input,
+          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+          throws com.google.protobuf.InvalidProtocolBufferException {
+        return new DtypeAndShape(input, extensionRegistry);
+      }
+    };
+
+    public static com.google.protobuf.Parser<DtypeAndShape> parser() {
+      return PARSER;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Parser<DtypeAndShape> getParserForType() {
+      return PARSER;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShape getDefaultInstanceForType() {
+      return DEFAULT_INSTANCE;
+    }
+
+  }
+
+  public static final int DEVICE_FIELD_NUMBER = 1;
+  private volatile java.lang.Object device_;
+  /**
+   * <pre>
+   * Unique name for the device containing the resource.
+   * </pre>
+   *
+   * <code>string device = 1;</code>
+   */
+  public java.lang.String getDevice() {
+    java.lang.Object ref = device_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      device_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * Unique name for the device containing the resource.
+   * </pre>
+   *
+   * <code>string device = 1;</code>
+   */
+  public com.google.protobuf.ByteString
+      getDeviceBytes() {
+    java.lang.Object ref = device_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      device_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int CONTAINER_FIELD_NUMBER = 2;
+  private volatile java.lang.Object container_;
+  /**
+   * <pre>
+   * Container in which this resource is placed.
+   * </pre>
+   *
+   * <code>string container = 2;</code>
+   */
+  public java.lang.String getContainer() {
+    java.lang.Object ref = container_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      container_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * Container in which this resource is placed.
+   * </pre>
+   *
+   * <code>string container = 2;</code>
+   */
+  public com.google.protobuf.ByteString
+      getContainerBytes() {
+    java.lang.Object ref = container_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      container_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int NAME_FIELD_NUMBER = 3;
+  private volatile java.lang.Object name_;
+  /**
+   * <pre>
+   * Unique name of this resource.
+   * </pre>
+   *
+   * <code>string name = 3;</code>
+   */
+  public java.lang.String getName() {
+    java.lang.Object ref = name_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      name_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * Unique name of this resource.
+   * </pre>
+   *
+   * <code>string name = 3;</code>
+   */
+  public com.google.protobuf.ByteString
+      getNameBytes() {
+    java.lang.Object ref = name_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      name_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int HASH_CODE_FIELD_NUMBER = 4;
+  private long hashCode_;
+  /**
+   * <pre>
+   * Hash code for the type of the resource. Is only valid in the same device
+   * and in the same execution.
+   * </pre>
+   *
+   * <code>uint64 hash_code = 4;</code>
+   */
+  public long getHashCode() {
+    return hashCode_;
+  }
+
+  public static final int MAYBE_TYPE_NAME_FIELD_NUMBER = 5;
+  private volatile java.lang.Object maybeTypeName_;
+  /**
+   * <pre>
+   * For debug-only, the name of the type pointed to by this handle, if
+   * available.
+   * </pre>
+   *
+   * <code>string maybe_type_name = 5;</code>
+   */
+  public java.lang.String getMaybeTypeName() {
+    java.lang.Object ref = maybeTypeName_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      maybeTypeName_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * For debug-only, the name of the type pointed to by this handle, if
+   * available.
+   * </pre>
+   *
+   * <code>string maybe_type_name = 5;</code>
+   */
+  public com.google.protobuf.ByteString
+      getMaybeTypeNameBytes() {
+    java.lang.Object ref = maybeTypeName_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      maybeTypeName_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int DTYPES_AND_SHAPES_FIELD_NUMBER = 6;
+  private java.util.List<org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShape> dtypesAndShapes_;
+  /**
+   * <pre>
+   * Data types and shapes for the underlying resource.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.ResourceHandleProto.DtypeAndShape dtypes_and_shapes = 6;</code>
+   */
+  public java.util.List<org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShape> getDtypesAndShapesList() {
+    return dtypesAndShapes_;
+  }
+  /**
+   * <pre>
+   * Data types and shapes for the underlying resource.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.ResourceHandleProto.DtypeAndShape dtypes_and_shapes = 6;</code>
+   */
+  public java.util.List<? extends org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShapeOrBuilder> 
+      getDtypesAndShapesOrBuilderList() {
+    return dtypesAndShapes_;
+  }
+  /**
+   * <pre>
+   * Data types and shapes for the underlying resource.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.ResourceHandleProto.DtypeAndShape dtypes_and_shapes = 6;</code>
+   */
+  public int getDtypesAndShapesCount() {
+    return dtypesAndShapes_.size();
+  }
+  /**
+   * <pre>
+   * Data types and shapes for the underlying resource.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.ResourceHandleProto.DtypeAndShape dtypes_and_shapes = 6;</code>
+   */
+  public org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShape getDtypesAndShapes(int index) {
+    return dtypesAndShapes_.get(index);
+  }
+  /**
+   * <pre>
+   * Data types and shapes for the underlying resource.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.ResourceHandleProto.DtypeAndShape dtypes_and_shapes = 6;</code>
+   */
+  public org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShapeOrBuilder getDtypesAndShapesOrBuilder(
+      int index) {
+    return dtypesAndShapes_.get(index);
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    if (!getDeviceBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 1, device_);
+    }
+    if (!getContainerBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 2, container_);
+    }
+    if (!getNameBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 3, name_);
+    }
+    if (hashCode_ != 0L) {
+      output.writeUInt64(4, hashCode_);
+    }
+    if (!getMaybeTypeNameBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 5, maybeTypeName_);
+    }
+    for (int i = 0; i < dtypesAndShapes_.size(); i++) {
+      output.writeMessage(6, dtypesAndShapes_.get(i));
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (!getDeviceBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, device_);
+    }
+    if (!getContainerBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(2, container_);
+    }
+    if (!getNameBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(3, name_);
+    }
+    if (hashCode_ != 0L) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeUInt64Size(4, hashCode_);
+    }
+    if (!getMaybeTypeNameBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(5, maybeTypeName_);
+    }
+    for (int i = 0; i < dtypesAndShapes_.size(); i++) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(6, dtypesAndShapes_.get(i));
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.framework.ResourceHandleProto)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.framework.ResourceHandleProto other = (org.tensorflow.proto.framework.ResourceHandleProto) obj;
+
+    if (!getDevice()
+        .equals(other.getDevice())) return false;
+    if (!getContainer()
+        .equals(other.getContainer())) return false;
+    if (!getName()
+        .equals(other.getName())) return false;
+    if (getHashCode()
+        != other.getHashCode()) return false;
+    if (!getMaybeTypeName()
+        .equals(other.getMaybeTypeName())) return false;
+    if (!getDtypesAndShapesList()
+        .equals(other.getDtypesAndShapesList())) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    hash = (37 * hash) + DEVICE_FIELD_NUMBER;
+    hash = (53 * hash) + getDevice().hashCode();
+    hash = (37 * hash) + CONTAINER_FIELD_NUMBER;
+    hash = (53 * hash) + getContainer().hashCode();
+    hash = (37 * hash) + NAME_FIELD_NUMBER;
+    hash = (53 * hash) + getName().hashCode();
+    hash = (37 * hash) + HASH_CODE_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+        getHashCode());
+    hash = (37 * hash) + MAYBE_TYPE_NAME_FIELD_NUMBER;
+    hash = (53 * hash) + getMaybeTypeName().hashCode();
+    if (getDtypesAndShapesCount() > 0) {
+      hash = (37 * hash) + DTYPES_AND_SHAPES_FIELD_NUMBER;
+      hash = (53 * hash) + getDtypesAndShapesList().hashCode();
+    }
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.framework.ResourceHandleProto parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.ResourceHandleProto parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.ResourceHandleProto parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.ResourceHandleProto parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.ResourceHandleProto parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.ResourceHandleProto parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.ResourceHandleProto parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.ResourceHandleProto parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.ResourceHandleProto parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.ResourceHandleProto parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.ResourceHandleProto parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.ResourceHandleProto parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.framework.ResourceHandleProto prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * <pre>
+   * Protocol buffer representing a handle to a tensorflow resource. Handles are
+   * not valid across executions, but can be serialized back and forth from within
+   * a single run.
+   * </pre>
+   *
+   * Protobuf type {@code tensorflow.ResourceHandleProto}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.ResourceHandleProto)
+      org.tensorflow.proto.framework.ResourceHandleProtoOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.ResourceHandle.internal_static_tensorflow_ResourceHandleProto_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.ResourceHandle.internal_static_tensorflow_ResourceHandleProto_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.ResourceHandleProto.class, org.tensorflow.proto.framework.ResourceHandleProto.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.framework.ResourceHandleProto.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+        getDtypesAndShapesFieldBuilder();
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      device_ = "";
+
+      container_ = "";
+
+      name_ = "";
+
+      hashCode_ = 0L;
+
+      maybeTypeName_ = "";
+
+      if (dtypesAndShapesBuilder_ == null) {
+        dtypesAndShapes_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000001);
+      } else {
+        dtypesAndShapesBuilder_.clear();
+      }
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.framework.ResourceHandle.internal_static_tensorflow_ResourceHandleProto_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.ResourceHandleProto getDefaultInstanceForType() {
+      return org.tensorflow.proto.framework.ResourceHandleProto.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.ResourceHandleProto build() {
+      org.tensorflow.proto.framework.ResourceHandleProto result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.ResourceHandleProto buildPartial() {
+      org.tensorflow.proto.framework.ResourceHandleProto result = new org.tensorflow.proto.framework.ResourceHandleProto(this);
+      int from_bitField0_ = bitField0_;
+      result.device_ = device_;
+      result.container_ = container_;
+      result.name_ = name_;
+      result.hashCode_ = hashCode_;
+      result.maybeTypeName_ = maybeTypeName_;
+      if (dtypesAndShapesBuilder_ == null) {
+        if (((bitField0_ & 0x00000001) != 0)) {
+          dtypesAndShapes_ = java.util.Collections.unmodifiableList(dtypesAndShapes_);
+          bitField0_ = (bitField0_ & ~0x00000001);
+        }
+        result.dtypesAndShapes_ = dtypesAndShapes_;
+      } else {
+        result.dtypesAndShapes_ = dtypesAndShapesBuilder_.build();
+      }
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.framework.ResourceHandleProto) {
+        return mergeFrom((org.tensorflow.proto.framework.ResourceHandleProto)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.framework.ResourceHandleProto other) {
+      if (other == org.tensorflow.proto.framework.ResourceHandleProto.getDefaultInstance()) return this;
+      if (!other.getDevice().isEmpty()) {
+        device_ = other.device_;
+        onChanged();
+      }
+      if (!other.getContainer().isEmpty()) {
+        container_ = other.container_;
+        onChanged();
+      }
+      if (!other.getName().isEmpty()) {
+        name_ = other.name_;
+        onChanged();
+      }
+      if (other.getHashCode() != 0L) {
+        setHashCode(other.getHashCode());
+      }
+      if (!other.getMaybeTypeName().isEmpty()) {
+        maybeTypeName_ = other.maybeTypeName_;
+        onChanged();
+      }
+      if (dtypesAndShapesBuilder_ == null) {
+        if (!other.dtypesAndShapes_.isEmpty()) {
+          if (dtypesAndShapes_.isEmpty()) {
+            dtypesAndShapes_ = other.dtypesAndShapes_;
+            bitField0_ = (bitField0_ & ~0x00000001);
+          } else {
+            ensureDtypesAndShapesIsMutable();
+            dtypesAndShapes_.addAll(other.dtypesAndShapes_);
+          }
+          onChanged();
+        }
+      } else {
+        if (!other.dtypesAndShapes_.isEmpty()) {
+          if (dtypesAndShapesBuilder_.isEmpty()) {
+            dtypesAndShapesBuilder_.dispose();
+            dtypesAndShapesBuilder_ = null;
+            dtypesAndShapes_ = other.dtypesAndShapes_;
+            bitField0_ = (bitField0_ & ~0x00000001);
+            dtypesAndShapesBuilder_ = 
+              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
+                 getDtypesAndShapesFieldBuilder() : null;
+          } else {
+            dtypesAndShapesBuilder_.addAllMessages(other.dtypesAndShapes_);
+          }
+        }
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.framework.ResourceHandleProto parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.framework.ResourceHandleProto) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+    private int bitField0_;
+
+    private java.lang.Object device_ = "";
+    /**
+     * <pre>
+     * Unique name for the device containing the resource.
+     * </pre>
+     *
+     * <code>string device = 1;</code>
+     */
+    public java.lang.String getDevice() {
+      java.lang.Object ref = device_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        device_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Unique name for the device containing the resource.
+     * </pre>
+     *
+     * <code>string device = 1;</code>
+     */
+    public com.google.protobuf.ByteString
+        getDeviceBytes() {
+      java.lang.Object ref = device_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        device_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Unique name for the device containing the resource.
+     * </pre>
+     *
+     * <code>string device = 1;</code>
+     */
+    public Builder setDevice(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      device_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Unique name for the device containing the resource.
+     * </pre>
+     *
+     * <code>string device = 1;</code>
+     */
+    public Builder clearDevice() {
+      
+      device_ = getDefaultInstance().getDevice();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Unique name for the device containing the resource.
+     * </pre>
+     *
+     * <code>string device = 1;</code>
+     */
+    public Builder setDeviceBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      device_ = value;
+      onChanged();
+      return this;
+    }
+
+    private java.lang.Object container_ = "";
+    /**
+     * <pre>
+     * Container in which this resource is placed.
+     * </pre>
+     *
+     * <code>string container = 2;</code>
+     */
+    public java.lang.String getContainer() {
+      java.lang.Object ref = container_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        container_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Container in which this resource is placed.
+     * </pre>
+     *
+     * <code>string container = 2;</code>
+     */
+    public com.google.protobuf.ByteString
+        getContainerBytes() {
+      java.lang.Object ref = container_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        container_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Container in which this resource is placed.
+     * </pre>
+     *
+     * <code>string container = 2;</code>
+     */
+    public Builder setContainer(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      container_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Container in which this resource is placed.
+     * </pre>
+     *
+     * <code>string container = 2;</code>
+     */
+    public Builder clearContainer() {
+      
+      container_ = getDefaultInstance().getContainer();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Container in which this resource is placed.
+     * </pre>
+     *
+     * <code>string container = 2;</code>
+     */
+    public Builder setContainerBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      container_ = value;
+      onChanged();
+      return this;
+    }
+
+    private java.lang.Object name_ = "";
+    /**
+     * <pre>
+     * Unique name of this resource.
+     * </pre>
+     *
+     * <code>string name = 3;</code>
+     */
+    public java.lang.String getName() {
+      java.lang.Object ref = name_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        name_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Unique name of this resource.
+     * </pre>
+     *
+     * <code>string name = 3;</code>
+     */
+    public com.google.protobuf.ByteString
+        getNameBytes() {
+      java.lang.Object ref = name_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        name_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Unique name of this resource.
+     * </pre>
+     *
+     * <code>string name = 3;</code>
+     */
+    public Builder setName(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      name_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Unique name of this resource.
+     * </pre>
+     *
+     * <code>string name = 3;</code>
+     */
+    public Builder clearName() {
+      
+      name_ = getDefaultInstance().getName();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Unique name of this resource.
+     * </pre>
+     *
+     * <code>string name = 3;</code>
+     */
+    public Builder setNameBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      name_ = value;
+      onChanged();
+      return this;
+    }
+
+    private long hashCode_ ;
+    /**
+     * <pre>
+     * Hash code for the type of the resource. Is only valid in the same device
+     * and in the same execution.
+     * </pre>
+     *
+     * <code>uint64 hash_code = 4;</code>
+     */
+    public long getHashCode() {
+      return hashCode_;
+    }
+    /**
+     * <pre>
+     * Hash code for the type of the resource. Is only valid in the same device
+     * and in the same execution.
+     * </pre>
+     *
+     * <code>uint64 hash_code = 4;</code>
+     */
+    public Builder setHashCode(long value) {
+      
+      hashCode_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Hash code for the type of the resource. Is only valid in the same device
+     * and in the same execution.
+     * </pre>
+     *
+     * <code>uint64 hash_code = 4;</code>
+     */
+    public Builder clearHashCode() {
+      
+      hashCode_ = 0L;
+      onChanged();
+      return this;
+    }
+
+    private java.lang.Object maybeTypeName_ = "";
+    /**
+     * <pre>
+     * For debug-only, the name of the type pointed to by this handle, if
+     * available.
+     * </pre>
+     *
+     * <code>string maybe_type_name = 5;</code>
+     */
+    public java.lang.String getMaybeTypeName() {
+      java.lang.Object ref = maybeTypeName_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        maybeTypeName_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * For debug-only, the name of the type pointed to by this handle, if
+     * available.
+     * </pre>
+     *
+     * <code>string maybe_type_name = 5;</code>
+     */
+    public com.google.protobuf.ByteString
+        getMaybeTypeNameBytes() {
+      java.lang.Object ref = maybeTypeName_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        maybeTypeName_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * For debug-only, the name of the type pointed to by this handle, if
+     * available.
+     * </pre>
+     *
+     * <code>string maybe_type_name = 5;</code>
+     */
+    public Builder setMaybeTypeName(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      maybeTypeName_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * For debug-only, the name of the type pointed to by this handle, if
+     * available.
+     * </pre>
+     *
+     * <code>string maybe_type_name = 5;</code>
+     */
+    public Builder clearMaybeTypeName() {
+      
+      maybeTypeName_ = getDefaultInstance().getMaybeTypeName();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * For debug-only, the name of the type pointed to by this handle, if
+     * available.
+     * </pre>
+     *
+     * <code>string maybe_type_name = 5;</code>
+     */
+    public Builder setMaybeTypeNameBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      maybeTypeName_ = value;
+      onChanged();
+      return this;
+    }
+
+    private java.util.List<org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShape> dtypesAndShapes_ =
+      java.util.Collections.emptyList();
+    private void ensureDtypesAndShapesIsMutable() {
+      if (!((bitField0_ & 0x00000001) != 0)) {
+        dtypesAndShapes_ = new java.util.ArrayList<org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShape>(dtypesAndShapes_);
+        bitField0_ |= 0x00000001;
+       }
+    }
+
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShape, org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShape.Builder, org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShapeOrBuilder> dtypesAndShapesBuilder_;
+
+    /**
+     * <pre>
+     * Data types and shapes for the underlying resource.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.ResourceHandleProto.DtypeAndShape dtypes_and_shapes = 6;</code>
+     */
+    public java.util.List<org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShape> getDtypesAndShapesList() {
+      if (dtypesAndShapesBuilder_ == null) {
+        return java.util.Collections.unmodifiableList(dtypesAndShapes_);
+      } else {
+        return dtypesAndShapesBuilder_.getMessageList();
+      }
+    }
+    /**
+     * <pre>
+     * Data types and shapes for the underlying resource.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.ResourceHandleProto.DtypeAndShape dtypes_and_shapes = 6;</code>
+     */
+    public int getDtypesAndShapesCount() {
+      if (dtypesAndShapesBuilder_ == null) {
+        return dtypesAndShapes_.size();
+      } else {
+        return dtypesAndShapesBuilder_.getCount();
+      }
+    }
+    /**
+     * <pre>
+     * Data types and shapes for the underlying resource.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.ResourceHandleProto.DtypeAndShape dtypes_and_shapes = 6;</code>
+     */
+    public org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShape getDtypesAndShapes(int index) {
+      if (dtypesAndShapesBuilder_ == null) {
+        return dtypesAndShapes_.get(index);
+      } else {
+        return dtypesAndShapesBuilder_.getMessage(index);
+      }
+    }
+    /**
+     * <pre>
+     * Data types and shapes for the underlying resource.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.ResourceHandleProto.DtypeAndShape dtypes_and_shapes = 6;</code>
+     */
+    public Builder setDtypesAndShapes(
+        int index, org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShape value) {
+      if (dtypesAndShapesBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureDtypesAndShapesIsMutable();
+        dtypesAndShapes_.set(index, value);
+        onChanged();
+      } else {
+        dtypesAndShapesBuilder_.setMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Data types and shapes for the underlying resource.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.ResourceHandleProto.DtypeAndShape dtypes_and_shapes = 6;</code>
+     */
+    public Builder setDtypesAndShapes(
+        int index, org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShape.Builder builderForValue) {
+      if (dtypesAndShapesBuilder_ == null) {
+        ensureDtypesAndShapesIsMutable();
+        dtypesAndShapes_.set(index, builderForValue.build());
+        onChanged();
+      } else {
+        dtypesAndShapesBuilder_.setMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Data types and shapes for the underlying resource.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.ResourceHandleProto.DtypeAndShape dtypes_and_shapes = 6;</code>
+     */
+    public Builder addDtypesAndShapes(org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShape value) {
+      if (dtypesAndShapesBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureDtypesAndShapesIsMutable();
+        dtypesAndShapes_.add(value);
+        onChanged();
+      } else {
+        dtypesAndShapesBuilder_.addMessage(value);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Data types and shapes for the underlying resource.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.ResourceHandleProto.DtypeAndShape dtypes_and_shapes = 6;</code>
+     */
+    public Builder addDtypesAndShapes(
+        int index, org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShape value) {
+      if (dtypesAndShapesBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureDtypesAndShapesIsMutable();
+        dtypesAndShapes_.add(index, value);
+        onChanged();
+      } else {
+        dtypesAndShapesBuilder_.addMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Data types and shapes for the underlying resource.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.ResourceHandleProto.DtypeAndShape dtypes_and_shapes = 6;</code>
+     */
+    public Builder addDtypesAndShapes(
+        org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShape.Builder builderForValue) {
+      if (dtypesAndShapesBuilder_ == null) {
+        ensureDtypesAndShapesIsMutable();
+        dtypesAndShapes_.add(builderForValue.build());
+        onChanged();
+      } else {
+        dtypesAndShapesBuilder_.addMessage(builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Data types and shapes for the underlying resource.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.ResourceHandleProto.DtypeAndShape dtypes_and_shapes = 6;</code>
+     */
+    public Builder addDtypesAndShapes(
+        int index, org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShape.Builder builderForValue) {
+      if (dtypesAndShapesBuilder_ == null) {
+        ensureDtypesAndShapesIsMutable();
+        dtypesAndShapes_.add(index, builderForValue.build());
+        onChanged();
+      } else {
+        dtypesAndShapesBuilder_.addMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Data types and shapes for the underlying resource.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.ResourceHandleProto.DtypeAndShape dtypes_and_shapes = 6;</code>
+     */
+    public Builder addAllDtypesAndShapes(
+        java.lang.Iterable<? extends org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShape> values) {
+      if (dtypesAndShapesBuilder_ == null) {
+        ensureDtypesAndShapesIsMutable();
+        com.google.protobuf.AbstractMessageLite.Builder.addAll(
+            values, dtypesAndShapes_);
+        onChanged();
+      } else {
+        dtypesAndShapesBuilder_.addAllMessages(values);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Data types and shapes for the underlying resource.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.ResourceHandleProto.DtypeAndShape dtypes_and_shapes = 6;</code>
+     */
+    public Builder clearDtypesAndShapes() {
+      if (dtypesAndShapesBuilder_ == null) {
+        dtypesAndShapes_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000001);
+        onChanged();
+      } else {
+        dtypesAndShapesBuilder_.clear();
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Data types and shapes for the underlying resource.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.ResourceHandleProto.DtypeAndShape dtypes_and_shapes = 6;</code>
+     */
+    public Builder removeDtypesAndShapes(int index) {
+      if (dtypesAndShapesBuilder_ == null) {
+        ensureDtypesAndShapesIsMutable();
+        dtypesAndShapes_.remove(index);
+        onChanged();
+      } else {
+        dtypesAndShapesBuilder_.remove(index);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Data types and shapes for the underlying resource.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.ResourceHandleProto.DtypeAndShape dtypes_and_shapes = 6;</code>
+     */
+    public org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShape.Builder getDtypesAndShapesBuilder(
+        int index) {
+      return getDtypesAndShapesFieldBuilder().getBuilder(index);
+    }
+    /**
+     * <pre>
+     * Data types and shapes for the underlying resource.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.ResourceHandleProto.DtypeAndShape dtypes_and_shapes = 6;</code>
+     */
+    public org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShapeOrBuilder getDtypesAndShapesOrBuilder(
+        int index) {
+      if (dtypesAndShapesBuilder_ == null) {
+        return dtypesAndShapes_.get(index);  } else {
+        return dtypesAndShapesBuilder_.getMessageOrBuilder(index);
+      }
+    }
+    /**
+     * <pre>
+     * Data types and shapes for the underlying resource.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.ResourceHandleProto.DtypeAndShape dtypes_and_shapes = 6;</code>
+     */
+    public java.util.List<? extends org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShapeOrBuilder> 
+         getDtypesAndShapesOrBuilderList() {
+      if (dtypesAndShapesBuilder_ != null) {
+        return dtypesAndShapesBuilder_.getMessageOrBuilderList();
+      } else {
+        return java.util.Collections.unmodifiableList(dtypesAndShapes_);
+      }
+    }
+    /**
+     * <pre>
+     * Data types and shapes for the underlying resource.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.ResourceHandleProto.DtypeAndShape dtypes_and_shapes = 6;</code>
+     */
+    public org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShape.Builder addDtypesAndShapesBuilder() {
+      return getDtypesAndShapesFieldBuilder().addBuilder(
+          org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShape.getDefaultInstance());
+    }
+    /**
+     * <pre>
+     * Data types and shapes for the underlying resource.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.ResourceHandleProto.DtypeAndShape dtypes_and_shapes = 6;</code>
+     */
+    public org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShape.Builder addDtypesAndShapesBuilder(
+        int index) {
+      return getDtypesAndShapesFieldBuilder().addBuilder(
+          index, org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShape.getDefaultInstance());
+    }
+    /**
+     * <pre>
+     * Data types and shapes for the underlying resource.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.ResourceHandleProto.DtypeAndShape dtypes_and_shapes = 6;</code>
+     */
+    public java.util.List<org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShape.Builder> 
+         getDtypesAndShapesBuilderList() {
+      return getDtypesAndShapesFieldBuilder().getBuilderList();
+    }
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShape, org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShape.Builder, org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShapeOrBuilder> 
+        getDtypesAndShapesFieldBuilder() {
+      if (dtypesAndShapesBuilder_ == null) {
+        dtypesAndShapesBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
+            org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShape, org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShape.Builder, org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShapeOrBuilder>(
+                dtypesAndShapes_,
+                ((bitField0_ & 0x00000001) != 0),
+                getParentForChildren(),
+                isClean());
+        dtypesAndShapes_ = null;
+      }
+      return dtypesAndShapesBuilder_;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.ResourceHandleProto)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.ResourceHandleProto)
+  private static final org.tensorflow.proto.framework.ResourceHandleProto DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.ResourceHandleProto();
+  }
+
+  public static org.tensorflow.proto.framework.ResourceHandleProto getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<ResourceHandleProto>
+      PARSER = new com.google.protobuf.AbstractParser<ResourceHandleProto>() {
+    @java.lang.Override
+    public ResourceHandleProto parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new ResourceHandleProto(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<ResourceHandleProto> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<ResourceHandleProto> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.framework.ResourceHandleProto getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ResourceHandleProtoOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ResourceHandleProtoOrBuilder.java
new file mode 100644
index 00000000000..461b2de1dba
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ResourceHandleProtoOrBuilder.java
@@ -0,0 +1,137 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/framework/resource_handle.proto
+
+package org.tensorflow.proto.framework;
+
+public interface ResourceHandleProtoOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.ResourceHandleProto)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <pre>
+   * Unique name for the device containing the resource.
+   * </pre>
+   *
+   * <code>string device = 1;</code>
+   */
+  java.lang.String getDevice();
+  /**
+   * <pre>
+   * Unique name for the device containing the resource.
+   * </pre>
+   *
+   * <code>string device = 1;</code>
+   */
+  com.google.protobuf.ByteString
+      getDeviceBytes();
+
+  /**
+   * <pre>
+   * Container in which this resource is placed.
+   * </pre>
+   *
+   * <code>string container = 2;</code>
+   */
+  java.lang.String getContainer();
+  /**
+   * <pre>
+   * Container in which this resource is placed.
+   * </pre>
+   *
+   * <code>string container = 2;</code>
+   */
+  com.google.protobuf.ByteString
+      getContainerBytes();
+
+  /**
+   * <pre>
+   * Unique name of this resource.
+   * </pre>
+   *
+   * <code>string name = 3;</code>
+   */
+  java.lang.String getName();
+  /**
+   * <pre>
+   * Unique name of this resource.
+   * </pre>
+   *
+   * <code>string name = 3;</code>
+   */
+  com.google.protobuf.ByteString
+      getNameBytes();
+
+  /**
+   * <pre>
+   * Hash code for the type of the resource. Is only valid in the same device
+   * and in the same execution.
+   * </pre>
+   *
+   * <code>uint64 hash_code = 4;</code>
+   */
+  long getHashCode();
+
+  /**
+   * <pre>
+   * For debug-only, the name of the type pointed to by this handle, if
+   * available.
+   * </pre>
+   *
+   * <code>string maybe_type_name = 5;</code>
+   */
+  java.lang.String getMaybeTypeName();
+  /**
+   * <pre>
+   * For debug-only, the name of the type pointed to by this handle, if
+   * available.
+   * </pre>
+   *
+   * <code>string maybe_type_name = 5;</code>
+   */
+  com.google.protobuf.ByteString
+      getMaybeTypeNameBytes();
+
+  /**
+   * <pre>
+   * Data types and shapes for the underlying resource.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.ResourceHandleProto.DtypeAndShape dtypes_and_shapes = 6;</code>
+   */
+  java.util.List<org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShape> 
+      getDtypesAndShapesList();
+  /**
+   * <pre>
+   * Data types and shapes for the underlying resource.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.ResourceHandleProto.DtypeAndShape dtypes_and_shapes = 6;</code>
+   */
+  org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShape getDtypesAndShapes(int index);
+  /**
+   * <pre>
+   * Data types and shapes for the underlying resource.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.ResourceHandleProto.DtypeAndShape dtypes_and_shapes = 6;</code>
+   */
+  int getDtypesAndShapesCount();
+  /**
+   * <pre>
+   * Data types and shapes for the underlying resource.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.ResourceHandleProto.DtypeAndShape dtypes_and_shapes = 6;</code>
+   */
+  java.util.List<? extends org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShapeOrBuilder> 
+      getDtypesAndShapesOrBuilderList();
+  /**
+   * <pre>
+   * Data types and shapes for the underlying resource.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.ResourceHandleProto.DtypeAndShape dtypes_and_shapes = 6;</code>
+   */
+  org.tensorflow.proto.framework.ResourceHandleProto.DtypeAndShapeOrBuilder getDtypesAndShapesOrBuilder(
+      int index);
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/RewriterConfig.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/RewriterConfig.java
new file mode 100644
index 00000000000..fe1571856ee
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/RewriterConfig.java
@@ -0,0 +1,6064 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/rewriter_config.proto
+
+package org.tensorflow.proto.framework;
+
+/**
+ * <pre>
+ * Graph rewriting is experimental and subject to change, not covered by any
+ * API stability guarantees.
+ * </pre>
+ *
+ * Protobuf type {@code tensorflow.RewriterConfig}
+ */
+public  final class RewriterConfig extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.RewriterConfig)
+    RewriterConfigOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use RewriterConfig.newBuilder() to construct.
+  private RewriterConfig(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private RewriterConfig() {
+    layoutOptimizer_ = 0;
+    constantFolding_ = 0;
+    shapeOptimization_ = 0;
+    remapping_ = 0;
+    commonSubgraphElimination_ = 0;
+    arithmeticOptimization_ = 0;
+    dependencyOptimization_ = 0;
+    loopOptimization_ = 0;
+    functionOptimization_ = 0;
+    debugStripper_ = 0;
+    scopedAllocatorOptimization_ = 0;
+    pinToHostOptimization_ = 0;
+    implementationSelector_ = 0;
+    autoMixedPrecision_ = 0;
+    autoMixedPrecisionMkl_ = 0;
+    metaOptimizerIterations_ = 0;
+    memoryOptimization_ = 0;
+    memoryOptimizerTargetNodeNameScope_ = "";
+    optimizers_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+    customOptimizers_ = java.util.Collections.emptyList();
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new RewriterConfig();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private RewriterConfig(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    int mutable_bitField0_ = 0;
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 8: {
+            int rawValue = input.readEnum();
+
+            layoutOptimizer_ = rawValue;
+            break;
+          }
+          case 16: {
+
+            disableModelPruning_ = input.readBool();
+            break;
+          }
+          case 24: {
+            int rawValue = input.readEnum();
+
+            constantFolding_ = rawValue;
+            break;
+          }
+          case 32: {
+            int rawValue = input.readEnum();
+
+            memoryOptimization_ = rawValue;
+            break;
+          }
+          case 42: {
+            org.tensorflow.proto.framework.AutoParallelOptions.Builder subBuilder = null;
+            if (autoParallel_ != null) {
+              subBuilder = autoParallel_.toBuilder();
+            }
+            autoParallel_ = input.readMessage(org.tensorflow.proto.framework.AutoParallelOptions.parser(), extensionRegistry);
+            if (subBuilder != null) {
+              subBuilder.mergeFrom(autoParallel_);
+              autoParallel_ = subBuilder.buildPartial();
+            }
+
+            break;
+          }
+          case 50: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            memoryOptimizerTargetNodeNameScope_ = s;
+            break;
+          }
+          case 56: {
+            int rawValue = input.readEnum();
+
+            arithmeticOptimization_ = rawValue;
+            break;
+          }
+          case 64: {
+            int rawValue = input.readEnum();
+
+            dependencyOptimization_ = rawValue;
+            break;
+          }
+          case 72: {
+            int rawValue = input.readEnum();
+
+            loopOptimization_ = rawValue;
+            break;
+          }
+          case 80: {
+            int rawValue = input.readEnum();
+
+            functionOptimization_ = rawValue;
+            break;
+          }
+          case 88: {
+            int rawValue = input.readEnum();
+
+            debugStripper_ = rawValue;
+            break;
+          }
+          case 96: {
+            int rawValue = input.readEnum();
+
+            metaOptimizerIterations_ = rawValue;
+            break;
+          }
+          case 104: {
+            int rawValue = input.readEnum();
+
+            shapeOptimization_ = rawValue;
+            break;
+          }
+          case 112: {
+            int rawValue = input.readEnum();
+
+            remapping_ = rawValue;
+            break;
+          }
+          case 120: {
+            int rawValue = input.readEnum();
+
+            scopedAllocatorOptimization_ = rawValue;
+            break;
+          }
+          case 130: {
+            org.tensorflow.proto.framework.ScopedAllocatorOptions.Builder subBuilder = null;
+            if (scopedAllocatorOpts_ != null) {
+              subBuilder = scopedAllocatorOpts_.toBuilder();
+            }
+            scopedAllocatorOpts_ = input.readMessage(org.tensorflow.proto.framework.ScopedAllocatorOptions.parser(), extensionRegistry);
+            if (subBuilder != null) {
+              subBuilder.mergeFrom(scopedAllocatorOpts_);
+              scopedAllocatorOpts_ = subBuilder.buildPartial();
+            }
+
+            break;
+          }
+          case 136: {
+
+            minGraphNodes_ = input.readInt32();
+            break;
+          }
+          case 144: {
+            int rawValue = input.readEnum();
+
+            pinToHostOptimization_ = rawValue;
+            break;
+          }
+          case 152: {
+
+            disableMetaOptimizer_ = input.readBool();
+            break;
+          }
+          case 160: {
+
+            metaOptimizerTimeoutMs_ = input.readInt64();
+            break;
+          }
+          case 168: {
+
+            failOnOptimizerErrors_ = input.readBool();
+            break;
+          }
+          case 176: {
+            int rawValue = input.readEnum();
+
+            implementationSelector_ = rawValue;
+            break;
+          }
+          case 184: {
+            int rawValue = input.readEnum();
+
+            autoMixedPrecision_ = rawValue;
+            break;
+          }
+          case 192: {
+            int rawValue = input.readEnum();
+
+            commonSubgraphElimination_ = rawValue;
+            break;
+          }
+          case 200: {
+            int rawValue = input.readEnum();
+
+            autoMixedPrecisionMkl_ = rawValue;
+            break;
+          }
+          case 802: {
+            java.lang.String s = input.readStringRequireUtf8();
+            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
+              optimizers_ = new com.google.protobuf.LazyStringArrayList();
+              mutable_bitField0_ |= 0x00000001;
+            }
+            optimizers_.add(s);
+            break;
+          }
+          case 1602: {
+            if (!((mutable_bitField0_ & 0x00000002) != 0)) {
+              customOptimizers_ = new java.util.ArrayList<org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizer>();
+              mutable_bitField0_ |= 0x00000002;
+            }
+            customOptimizers_.add(
+                input.readMessage(org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizer.parser(), extensionRegistry));
+            break;
+          }
+          case 2402: {
+            org.tensorflow.proto.framework.VerifierConfig.Builder subBuilder = null;
+            if (interOptimizerVerifierConfig_ != null) {
+              subBuilder = interOptimizerVerifierConfig_.toBuilder();
+            }
+            interOptimizerVerifierConfig_ = input.readMessage(org.tensorflow.proto.framework.VerifierConfig.parser(), extensionRegistry);
+            if (subBuilder != null) {
+              subBuilder.mergeFrom(interOptimizerVerifierConfig_);
+              interOptimizerVerifierConfig_ = subBuilder.buildPartial();
+            }
+
+            break;
+          }
+          case 2410: {
+            org.tensorflow.proto.framework.VerifierConfig.Builder subBuilder = null;
+            if (postOptimizationVerifierConfig_ != null) {
+              subBuilder = postOptimizationVerifierConfig_.toBuilder();
+            }
+            postOptimizationVerifierConfig_ = input.readMessage(org.tensorflow.proto.framework.VerifierConfig.parser(), extensionRegistry);
+            if (subBuilder != null) {
+              subBuilder.mergeFrom(postOptimizationVerifierConfig_);
+              postOptimizationVerifierConfig_ = subBuilder.buildPartial();
+            }
+
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      if (((mutable_bitField0_ & 0x00000001) != 0)) {
+        optimizers_ = optimizers_.getUnmodifiableView();
+      }
+      if (((mutable_bitField0_ & 0x00000002) != 0)) {
+        customOptimizers_ = java.util.Collections.unmodifiableList(customOptimizers_);
+      }
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.framework.RewriterConfigProtos.internal_static_tensorflow_RewriterConfig_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.framework.RewriterConfigProtos.internal_static_tensorflow_RewriterConfig_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.framework.RewriterConfig.class, org.tensorflow.proto.framework.RewriterConfig.Builder.class);
+  }
+
+  /**
+   * Protobuf enum {@code tensorflow.RewriterConfig.Toggle}
+   */
+  public enum Toggle
+      implements com.google.protobuf.ProtocolMessageEnum {
+    /**
+     * <code>DEFAULT = 0;</code>
+     */
+    DEFAULT(0),
+    /**
+     * <code>ON = 1;</code>
+     */
+    ON(1),
+    /**
+     * <code>OFF = 2;</code>
+     */
+    OFF(2),
+    /**
+     * <pre>
+     * Enable some aggressive optimizations that use assumptions that TF graphs
+     * may break. For example, assume the shape of a placeholder matches its
+     * actual feed.
+     * </pre>
+     *
+     * <code>AGGRESSIVE = 3;</code>
+     */
+    AGGRESSIVE(3),
+    UNRECOGNIZED(-1),
+    ;
+
+    /**
+     * <code>DEFAULT = 0;</code>
+     */
+    public static final int DEFAULT_VALUE = 0;
+    /**
+     * <code>ON = 1;</code>
+     */
+    public static final int ON_VALUE = 1;
+    /**
+     * <code>OFF = 2;</code>
+     */
+    public static final int OFF_VALUE = 2;
+    /**
+     * <pre>
+     * Enable some aggressive optimizations that use assumptions that TF graphs
+     * may break. For example, assume the shape of a placeholder matches its
+     * actual feed.
+     * </pre>
+     *
+     * <code>AGGRESSIVE = 3;</code>
+     */
+    public static final int AGGRESSIVE_VALUE = 3;
+
+
+    public final int getNumber() {
+      if (this == UNRECOGNIZED) {
+        throw new java.lang.IllegalArgumentException(
+            "Can't get the number of an unknown enum value.");
+      }
+      return value;
+    }
+
+    /**
+     * @deprecated Use {@link #forNumber(int)} instead.
+     */
+    @java.lang.Deprecated
+    public static Toggle valueOf(int value) {
+      return forNumber(value);
+    }
+
+    public static Toggle forNumber(int value) {
+      switch (value) {
+        case 0: return DEFAULT;
+        case 1: return ON;
+        case 2: return OFF;
+        case 3: return AGGRESSIVE;
+        default: return null;
+      }
+    }
+
+    public static com.google.protobuf.Internal.EnumLiteMap<Toggle>
+        internalGetValueMap() {
+      return internalValueMap;
+    }
+    private static final com.google.protobuf.Internal.EnumLiteMap<
+        Toggle> internalValueMap =
+          new com.google.protobuf.Internal.EnumLiteMap<Toggle>() {
+            public Toggle findValueByNumber(int number) {
+              return Toggle.forNumber(number);
+            }
+          };
+
+    public final com.google.protobuf.Descriptors.EnumValueDescriptor
+        getValueDescriptor() {
+      return getDescriptor().getValues().get(ordinal());
+    }
+    public final com.google.protobuf.Descriptors.EnumDescriptor
+        getDescriptorForType() {
+      return getDescriptor();
+    }
+    public static final com.google.protobuf.Descriptors.EnumDescriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.RewriterConfig.getDescriptor().getEnumTypes().get(0);
+    }
+
+    private static final Toggle[] VALUES = values();
+
+    public static Toggle valueOf(
+        com.google.protobuf.Descriptors.EnumValueDescriptor desc) {
+      if (desc.getType() != getDescriptor()) {
+        throw new java.lang.IllegalArgumentException(
+          "EnumValueDescriptor is not for this type.");
+      }
+      if (desc.getIndex() == -1) {
+        return UNRECOGNIZED;
+      }
+      return VALUES[desc.getIndex()];
+    }
+
+    private final int value;
+
+    private Toggle(int value) {
+      this.value = value;
+    }
+
+    // @@protoc_insertion_point(enum_scope:tensorflow.RewriterConfig.Toggle)
+  }
+
+  /**
+   * <pre>
+   * Enum controlling the number of times to run optimizers. The default is to
+   * run them twice.
+   * </pre>
+   *
+   * Protobuf enum {@code tensorflow.RewriterConfig.NumIterationsType}
+   */
+  public enum NumIterationsType
+      implements com.google.protobuf.ProtocolMessageEnum {
+    /**
+     * <code>DEFAULT_NUM_ITERS = 0;</code>
+     */
+    DEFAULT_NUM_ITERS(0),
+    /**
+     * <code>ONE = 1;</code>
+     */
+    ONE(1),
+    /**
+     * <code>TWO = 2;</code>
+     */
+    TWO(2),
+    UNRECOGNIZED(-1),
+    ;
+
+    /**
+     * <code>DEFAULT_NUM_ITERS = 0;</code>
+     */
+    public static final int DEFAULT_NUM_ITERS_VALUE = 0;
+    /**
+     * <code>ONE = 1;</code>
+     */
+    public static final int ONE_VALUE = 1;
+    /**
+     * <code>TWO = 2;</code>
+     */
+    public static final int TWO_VALUE = 2;
+
+
+    public final int getNumber() {
+      if (this == UNRECOGNIZED) {
+        throw new java.lang.IllegalArgumentException(
+            "Can't get the number of an unknown enum value.");
+      }
+      return value;
+    }
+
+    /**
+     * @deprecated Use {@link #forNumber(int)} instead.
+     */
+    @java.lang.Deprecated
+    public static NumIterationsType valueOf(int value) {
+      return forNumber(value);
+    }
+
+    public static NumIterationsType forNumber(int value) {
+      switch (value) {
+        case 0: return DEFAULT_NUM_ITERS;
+        case 1: return ONE;
+        case 2: return TWO;
+        default: return null;
+      }
+    }
+
+    public static com.google.protobuf.Internal.EnumLiteMap<NumIterationsType>
+        internalGetValueMap() {
+      return internalValueMap;
+    }
+    private static final com.google.protobuf.Internal.EnumLiteMap<
+        NumIterationsType> internalValueMap =
+          new com.google.protobuf.Internal.EnumLiteMap<NumIterationsType>() {
+            public NumIterationsType findValueByNumber(int number) {
+              return NumIterationsType.forNumber(number);
+            }
+          };
+
+    public final com.google.protobuf.Descriptors.EnumValueDescriptor
+        getValueDescriptor() {
+      return getDescriptor().getValues().get(ordinal());
+    }
+    public final com.google.protobuf.Descriptors.EnumDescriptor
+        getDescriptorForType() {
+      return getDescriptor();
+    }
+    public static final com.google.protobuf.Descriptors.EnumDescriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.RewriterConfig.getDescriptor().getEnumTypes().get(1);
+    }
+
+    private static final NumIterationsType[] VALUES = values();
+
+    public static NumIterationsType valueOf(
+        com.google.protobuf.Descriptors.EnumValueDescriptor desc) {
+      if (desc.getType() != getDescriptor()) {
+        throw new java.lang.IllegalArgumentException(
+          "EnumValueDescriptor is not for this type.");
+      }
+      if (desc.getIndex() == -1) {
+        return UNRECOGNIZED;
+      }
+      return VALUES[desc.getIndex()];
+    }
+
+    private final int value;
+
+    private NumIterationsType(int value) {
+      this.value = value;
+    }
+
+    // @@protoc_insertion_point(enum_scope:tensorflow.RewriterConfig.NumIterationsType)
+  }
+
+  /**
+   * Protobuf enum {@code tensorflow.RewriterConfig.MemOptType}
+   */
+  public enum MemOptType
+      implements com.google.protobuf.ProtocolMessageEnum {
+    /**
+     * <pre>
+     * The default setting (SCHEDULING and SWAPPING HEURISTICS only)
+     * </pre>
+     *
+     * <code>DEFAULT_MEM_OPT = 0;</code>
+     */
+    DEFAULT_MEM_OPT(0),
+    /**
+     * <pre>
+     * Disabled in the meta-optimizer.
+     * </pre>
+     *
+     * <code>NO_MEM_OPT = 1;</code>
+     */
+    NO_MEM_OPT(1),
+    /**
+     * <pre>
+     * Driven by manual op-level annotations.
+     * </pre>
+     *
+     * <code>MANUAL = 2;</code>
+     */
+    MANUAL(2),
+    /**
+     * <pre>
+     * Swapping heuristic will move a tensor from the GPU to the CPU and move
+     * it back when needed to reduce peak memory usage.
+     * </pre>
+     *
+     * <code>SWAPPING_HEURISTICS = 4;</code>
+     */
+    SWAPPING_HEURISTICS(4),
+    /**
+     * <pre>
+     * Recomputation heuristics will recompute ops (such as Relu activation)
+     * during backprop instead of storing them, reducing peak memory usage.
+     * </pre>
+     *
+     * <code>RECOMPUTATION_HEURISTICS = 5;</code>
+     */
+    RECOMPUTATION_HEURISTICS(5),
+    /**
+     * <pre>
+     * Scheduling will split big ops such as AddN and try to enforce a schedule
+     * of the new computations that decreases peak memory usage.
+     * </pre>
+     *
+     * <code>SCHEDULING_HEURISTICS = 6;</code>
+     */
+    SCHEDULING_HEURISTICS(6),
+    /**
+     * <pre>
+     * Use any combination of swapping and recomputation heuristics.
+     * </pre>
+     *
+     * <code>HEURISTICS = 3;</code>
+     */
+    HEURISTICS(3),
+    UNRECOGNIZED(-1),
+    ;
+
+    /**
+     * <pre>
+     * The default setting (SCHEDULING and SWAPPING HEURISTICS only)
+     * </pre>
+     *
+     * <code>DEFAULT_MEM_OPT = 0;</code>
+     */
+    public static final int DEFAULT_MEM_OPT_VALUE = 0;
+    /**
+     * <pre>
+     * Disabled in the meta-optimizer.
+     * </pre>
+     *
+     * <code>NO_MEM_OPT = 1;</code>
+     */
+    public static final int NO_MEM_OPT_VALUE = 1;
+    /**
+     * <pre>
+     * Driven by manual op-level annotations.
+     * </pre>
+     *
+     * <code>MANUAL = 2;</code>
+     */
+    public static final int MANUAL_VALUE = 2;
+    /**
+     * <pre>
+     * Swapping heuristic will move a tensor from the GPU to the CPU and move
+     * it back when needed to reduce peak memory usage.
+     * </pre>
+     *
+     * <code>SWAPPING_HEURISTICS = 4;</code>
+     */
+    public static final int SWAPPING_HEURISTICS_VALUE = 4;
+    /**
+     * <pre>
+     * Recomputation heuristics will recompute ops (such as Relu activation)
+     * during backprop instead of storing them, reducing peak memory usage.
+     * </pre>
+     *
+     * <code>RECOMPUTATION_HEURISTICS = 5;</code>
+     */
+    public static final int RECOMPUTATION_HEURISTICS_VALUE = 5;
+    /**
+     * <pre>
+     * Scheduling will split big ops such as AddN and try to enforce a schedule
+     * of the new computations that decreases peak memory usage.
+     * </pre>
+     *
+     * <code>SCHEDULING_HEURISTICS = 6;</code>
+     */
+    public static final int SCHEDULING_HEURISTICS_VALUE = 6;
+    /**
+     * <pre>
+     * Use any combination of swapping and recomputation heuristics.
+     * </pre>
+     *
+     * <code>HEURISTICS = 3;</code>
+     */
+    public static final int HEURISTICS_VALUE = 3;
+
+
+    public final int getNumber() {
+      if (this == UNRECOGNIZED) {
+        throw new java.lang.IllegalArgumentException(
+            "Can't get the number of an unknown enum value.");
+      }
+      return value;
+    }
+
+    /**
+     * @deprecated Use {@link #forNumber(int)} instead.
+     */
+    @java.lang.Deprecated
+    public static MemOptType valueOf(int value) {
+      return forNumber(value);
+    }
+
+    public static MemOptType forNumber(int value) {
+      switch (value) {
+        case 0: return DEFAULT_MEM_OPT;
+        case 1: return NO_MEM_OPT;
+        case 2: return MANUAL;
+        case 4: return SWAPPING_HEURISTICS;
+        case 5: return RECOMPUTATION_HEURISTICS;
+        case 6: return SCHEDULING_HEURISTICS;
+        case 3: return HEURISTICS;
+        default: return null;
+      }
+    }
+
+    public static com.google.protobuf.Internal.EnumLiteMap<MemOptType>
+        internalGetValueMap() {
+      return internalValueMap;
+    }
+    private static final com.google.protobuf.Internal.EnumLiteMap<
+        MemOptType> internalValueMap =
+          new com.google.protobuf.Internal.EnumLiteMap<MemOptType>() {
+            public MemOptType findValueByNumber(int number) {
+              return MemOptType.forNumber(number);
+            }
+          };
+
+    public final com.google.protobuf.Descriptors.EnumValueDescriptor
+        getValueDescriptor() {
+      return getDescriptor().getValues().get(ordinal());
+    }
+    public final com.google.protobuf.Descriptors.EnumDescriptor
+        getDescriptorForType() {
+      return getDescriptor();
+    }
+    public static final com.google.protobuf.Descriptors.EnumDescriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.RewriterConfig.getDescriptor().getEnumTypes().get(2);
+    }
+
+    private static final MemOptType[] VALUES = values();
+
+    public static MemOptType valueOf(
+        com.google.protobuf.Descriptors.EnumValueDescriptor desc) {
+      if (desc.getType() != getDescriptor()) {
+        throw new java.lang.IllegalArgumentException(
+          "EnumValueDescriptor is not for this type.");
+      }
+      if (desc.getIndex() == -1) {
+        return UNRECOGNIZED;
+      }
+      return VALUES[desc.getIndex()];
+    }
+
+    private final int value;
+
+    private MemOptType(int value) {
+      this.value = value;
+    }
+
+    // @@protoc_insertion_point(enum_scope:tensorflow.RewriterConfig.MemOptType)
+  }
+
+  public interface CustomGraphOptimizerOrBuilder extends
+      // @@protoc_insertion_point(interface_extends:tensorflow.RewriterConfig.CustomGraphOptimizer)
+      com.google.protobuf.MessageOrBuilder {
+
+    /**
+     * <code>string name = 1;</code>
+     */
+    java.lang.String getName();
+    /**
+     * <code>string name = 1;</code>
+     */
+    com.google.protobuf.ByteString
+        getNameBytes();
+
+    /**
+     * <code>map&lt;string, .tensorflow.AttrValue&gt; parameter_map = 2;</code>
+     */
+    int getParameterMapCount();
+    /**
+     * <code>map&lt;string, .tensorflow.AttrValue&gt; parameter_map = 2;</code>
+     */
+    boolean containsParameterMap(
+        java.lang.String key);
+    /**
+     * Use {@link #getParameterMapMap()} instead.
+     */
+    @java.lang.Deprecated
+    java.util.Map<java.lang.String, org.tensorflow.proto.framework.AttrValue>
+    getParameterMap();
+    /**
+     * <code>map&lt;string, .tensorflow.AttrValue&gt; parameter_map = 2;</code>
+     */
+    java.util.Map<java.lang.String, org.tensorflow.proto.framework.AttrValue>
+    getParameterMapMap();
+    /**
+     * <code>map&lt;string, .tensorflow.AttrValue&gt; parameter_map = 2;</code>
+     */
+
+    org.tensorflow.proto.framework.AttrValue getParameterMapOrDefault(
+        java.lang.String key,
+        org.tensorflow.proto.framework.AttrValue defaultValue);
+    /**
+     * <code>map&lt;string, .tensorflow.AttrValue&gt; parameter_map = 2;</code>
+     */
+
+    org.tensorflow.proto.framework.AttrValue getParameterMapOrThrow(
+        java.lang.String key);
+  }
+  /**
+   * <pre>
+   * Message to describe custom graph optimizer and its parameters
+   * </pre>
+   *
+   * Protobuf type {@code tensorflow.RewriterConfig.CustomGraphOptimizer}
+   */
+  public  static final class CustomGraphOptimizer extends
+      com.google.protobuf.GeneratedMessageV3 implements
+      // @@protoc_insertion_point(message_implements:tensorflow.RewriterConfig.CustomGraphOptimizer)
+      CustomGraphOptimizerOrBuilder {
+  private static final long serialVersionUID = 0L;
+    // Use CustomGraphOptimizer.newBuilder() to construct.
+    private CustomGraphOptimizer(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+      super(builder);
+    }
+    private CustomGraphOptimizer() {
+      name_ = "";
+    }
+
+    @java.lang.Override
+    @SuppressWarnings({"unused"})
+    protected java.lang.Object newInstance(
+        UnusedPrivateParameter unused) {
+      return new CustomGraphOptimizer();
+    }
+
+    @java.lang.Override
+    public final com.google.protobuf.UnknownFieldSet
+    getUnknownFields() {
+      return this.unknownFields;
+    }
+    private CustomGraphOptimizer(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      this();
+      if (extensionRegistry == null) {
+        throw new java.lang.NullPointerException();
+      }
+      int mutable_bitField0_ = 0;
+      com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+          com.google.protobuf.UnknownFieldSet.newBuilder();
+      try {
+        boolean done = false;
+        while (!done) {
+          int tag = input.readTag();
+          switch (tag) {
+            case 0:
+              done = true;
+              break;
+            case 10: {
+              java.lang.String s = input.readStringRequireUtf8();
+
+              name_ = s;
+              break;
+            }
+            case 18: {
+              if (!((mutable_bitField0_ & 0x00000001) != 0)) {
+                parameterMap_ = com.google.protobuf.MapField.newMapField(
+                    ParameterMapDefaultEntryHolder.defaultEntry);
+                mutable_bitField0_ |= 0x00000001;
+              }
+              com.google.protobuf.MapEntry<java.lang.String, org.tensorflow.proto.framework.AttrValue>
+              parameterMap__ = input.readMessage(
+                  ParameterMapDefaultEntryHolder.defaultEntry.getParserForType(), extensionRegistry);
+              parameterMap_.getMutableMap().put(
+                  parameterMap__.getKey(), parameterMap__.getValue());
+              break;
+            }
+            default: {
+              if (!parseUnknownField(
+                  input, unknownFields, extensionRegistry, tag)) {
+                done = true;
+              }
+              break;
+            }
+          }
+        }
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        throw e.setUnfinishedMessage(this);
+      } catch (java.io.IOException e) {
+        throw new com.google.protobuf.InvalidProtocolBufferException(
+            e).setUnfinishedMessage(this);
+      } finally {
+        this.unknownFields = unknownFields.build();
+        makeExtensionsImmutable();
+      }
+    }
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.RewriterConfigProtos.internal_static_tensorflow_RewriterConfig_CustomGraphOptimizer_descriptor;
+    }
+
+    @SuppressWarnings({"rawtypes"})
+    @java.lang.Override
+    protected com.google.protobuf.MapField internalGetMapField(
+        int number) {
+      switch (number) {
+        case 2:
+          return internalGetParameterMap();
+        default:
+          throw new RuntimeException(
+              "Invalid map field number: " + number);
+      }
+    }
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.RewriterConfigProtos.internal_static_tensorflow_RewriterConfig_CustomGraphOptimizer_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizer.class, org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizer.Builder.class);
+    }
+
+    public static final int NAME_FIELD_NUMBER = 1;
+    private volatile java.lang.Object name_;
+    /**
+     * <code>string name = 1;</code>
+     */
+    public java.lang.String getName() {
+      java.lang.Object ref = name_;
+      if (ref instanceof java.lang.String) {
+        return (java.lang.String) ref;
+      } else {
+        com.google.protobuf.ByteString bs = 
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        name_ = s;
+        return s;
+      }
+    }
+    /**
+     * <code>string name = 1;</code>
+     */
+    public com.google.protobuf.ByteString
+        getNameBytes() {
+      java.lang.Object ref = name_;
+      if (ref instanceof java.lang.String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        name_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+
+    public static final int PARAMETER_MAP_FIELD_NUMBER = 2;
+    private static final class ParameterMapDefaultEntryHolder {
+      static final com.google.protobuf.MapEntry<
+          java.lang.String, org.tensorflow.proto.framework.AttrValue> defaultEntry =
+              com.google.protobuf.MapEntry
+              .<java.lang.String, org.tensorflow.proto.framework.AttrValue>newDefaultInstance(
+                  org.tensorflow.proto.framework.RewriterConfigProtos.internal_static_tensorflow_RewriterConfig_CustomGraphOptimizer_ParameterMapEntry_descriptor, 
+                  com.google.protobuf.WireFormat.FieldType.STRING,
+                  "",
+                  com.google.protobuf.WireFormat.FieldType.MESSAGE,
+                  org.tensorflow.proto.framework.AttrValue.getDefaultInstance());
+    }
+    private com.google.protobuf.MapField<
+        java.lang.String, org.tensorflow.proto.framework.AttrValue> parameterMap_;
+    private com.google.protobuf.MapField<java.lang.String, org.tensorflow.proto.framework.AttrValue>
+    internalGetParameterMap() {
+      if (parameterMap_ == null) {
+        return com.google.protobuf.MapField.emptyMapField(
+            ParameterMapDefaultEntryHolder.defaultEntry);
+      }
+      return parameterMap_;
+    }
+
+    public int getParameterMapCount() {
+      return internalGetParameterMap().getMap().size();
+    }
+    /**
+     * <code>map&lt;string, .tensorflow.AttrValue&gt; parameter_map = 2;</code>
+     */
+
+    public boolean containsParameterMap(
+        java.lang.String key) {
+      if (key == null) { throw new java.lang.NullPointerException(); }
+      return internalGetParameterMap().getMap().containsKey(key);
+    }
+    /**
+     * Use {@link #getParameterMapMap()} instead.
+     */
+    @java.lang.Deprecated
+    public java.util.Map<java.lang.String, org.tensorflow.proto.framework.AttrValue> getParameterMap() {
+      return getParameterMapMap();
+    }
+    /**
+     * <code>map&lt;string, .tensorflow.AttrValue&gt; parameter_map = 2;</code>
+     */
+
+    public java.util.Map<java.lang.String, org.tensorflow.proto.framework.AttrValue> getParameterMapMap() {
+      return internalGetParameterMap().getMap();
+    }
+    /**
+     * <code>map&lt;string, .tensorflow.AttrValue&gt; parameter_map = 2;</code>
+     */
+
+    public org.tensorflow.proto.framework.AttrValue getParameterMapOrDefault(
+        java.lang.String key,
+        org.tensorflow.proto.framework.AttrValue defaultValue) {
+      if (key == null) { throw new java.lang.NullPointerException(); }
+      java.util.Map<java.lang.String, org.tensorflow.proto.framework.AttrValue> map =
+          internalGetParameterMap().getMap();
+      return map.containsKey(key) ? map.get(key) : defaultValue;
+    }
+    /**
+     * <code>map&lt;string, .tensorflow.AttrValue&gt; parameter_map = 2;</code>
+     */
+
+    public org.tensorflow.proto.framework.AttrValue getParameterMapOrThrow(
+        java.lang.String key) {
+      if (key == null) { throw new java.lang.NullPointerException(); }
+      java.util.Map<java.lang.String, org.tensorflow.proto.framework.AttrValue> map =
+          internalGetParameterMap().getMap();
+      if (!map.containsKey(key)) {
+        throw new java.lang.IllegalArgumentException();
+      }
+      return map.get(key);
+    }
+
+    private byte memoizedIsInitialized = -1;
+    @java.lang.Override
+    public final boolean isInitialized() {
+      byte isInitialized = memoizedIsInitialized;
+      if (isInitialized == 1) return true;
+      if (isInitialized == 0) return false;
+
+      memoizedIsInitialized = 1;
+      return true;
+    }
+
+    @java.lang.Override
+    public void writeTo(com.google.protobuf.CodedOutputStream output)
+                        throws java.io.IOException {
+      if (!getNameBytes().isEmpty()) {
+        com.google.protobuf.GeneratedMessageV3.writeString(output, 1, name_);
+      }
+      com.google.protobuf.GeneratedMessageV3
+        .serializeStringMapTo(
+          output,
+          internalGetParameterMap(),
+          ParameterMapDefaultEntryHolder.defaultEntry,
+          2);
+      unknownFields.writeTo(output);
+    }
+
+    @java.lang.Override
+    public int getSerializedSize() {
+      int size = memoizedSize;
+      if (size != -1) return size;
+
+      size = 0;
+      if (!getNameBytes().isEmpty()) {
+        size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, name_);
+      }
+      for (java.util.Map.Entry<java.lang.String, org.tensorflow.proto.framework.AttrValue> entry
+           : internalGetParameterMap().getMap().entrySet()) {
+        com.google.protobuf.MapEntry<java.lang.String, org.tensorflow.proto.framework.AttrValue>
+        parameterMap__ = ParameterMapDefaultEntryHolder.defaultEntry.newBuilderForType()
+            .setKey(entry.getKey())
+            .setValue(entry.getValue())
+            .build();
+        size += com.google.protobuf.CodedOutputStream
+            .computeMessageSize(2, parameterMap__);
+      }
+      size += unknownFields.getSerializedSize();
+      memoizedSize = size;
+      return size;
+    }
+
+    @java.lang.Override
+    public boolean equals(final java.lang.Object obj) {
+      if (obj == this) {
+       return true;
+      }
+      if (!(obj instanceof org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizer)) {
+        return super.equals(obj);
+      }
+      org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizer other = (org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizer) obj;
+
+      if (!getName()
+          .equals(other.getName())) return false;
+      if (!internalGetParameterMap().equals(
+          other.internalGetParameterMap())) return false;
+      if (!unknownFields.equals(other.unknownFields)) return false;
+      return true;
+    }
+
+    @java.lang.Override
+    public int hashCode() {
+      if (memoizedHashCode != 0) {
+        return memoizedHashCode;
+      }
+      int hash = 41;
+      hash = (19 * hash) + getDescriptor().hashCode();
+      hash = (37 * hash) + NAME_FIELD_NUMBER;
+      hash = (53 * hash) + getName().hashCode();
+      if (!internalGetParameterMap().getMap().isEmpty()) {
+        hash = (37 * hash) + PARAMETER_MAP_FIELD_NUMBER;
+        hash = (53 * hash) + internalGetParameterMap().hashCode();
+      }
+      hash = (29 * hash) + unknownFields.hashCode();
+      memoizedHashCode = hash;
+      return hash;
+    }
+
+    public static org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizer parseFrom(
+        java.nio.ByteBuffer data)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data);
+    }
+    public static org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizer parseFrom(
+        java.nio.ByteBuffer data,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizer parseFrom(
+        com.google.protobuf.ByteString data)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data);
+    }
+    public static org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizer parseFrom(
+        com.google.protobuf.ByteString data,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizer parseFrom(byte[] data)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data);
+    }
+    public static org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizer parseFrom(
+        byte[] data,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizer parseFrom(java.io.InputStream input)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input);
+    }
+    public static org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizer parseFrom(
+        java.io.InputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizer parseDelimitedFrom(java.io.InputStream input)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseDelimitedWithIOException(PARSER, input);
+    }
+    public static org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizer parseDelimitedFrom(
+        java.io.InputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizer parseFrom(
+        com.google.protobuf.CodedInputStream input)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input);
+    }
+    public static org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizer parseFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input, extensionRegistry);
+    }
+
+    @java.lang.Override
+    public Builder newBuilderForType() { return newBuilder(); }
+    public static Builder newBuilder() {
+      return DEFAULT_INSTANCE.toBuilder();
+    }
+    public static Builder newBuilder(org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizer prototype) {
+      return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+    }
+    @java.lang.Override
+    public Builder toBuilder() {
+      return this == DEFAULT_INSTANCE
+          ? new Builder() : new Builder().mergeFrom(this);
+    }
+
+    @java.lang.Override
+    protected Builder newBuilderForType(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      Builder builder = new Builder(parent);
+      return builder;
+    }
+    /**
+     * <pre>
+     * Message to describe custom graph optimizer and its parameters
+     * </pre>
+     *
+     * Protobuf type {@code tensorflow.RewriterConfig.CustomGraphOptimizer}
+     */
+    public static final class Builder extends
+        com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+        // @@protoc_insertion_point(builder_implements:tensorflow.RewriterConfig.CustomGraphOptimizer)
+        org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizerOrBuilder {
+      public static final com.google.protobuf.Descriptors.Descriptor
+          getDescriptor() {
+        return org.tensorflow.proto.framework.RewriterConfigProtos.internal_static_tensorflow_RewriterConfig_CustomGraphOptimizer_descriptor;
+      }
+
+      @SuppressWarnings({"rawtypes"})
+      protected com.google.protobuf.MapField internalGetMapField(
+          int number) {
+        switch (number) {
+          case 2:
+            return internalGetParameterMap();
+          default:
+            throw new RuntimeException(
+                "Invalid map field number: " + number);
+        }
+      }
+      @SuppressWarnings({"rawtypes"})
+      protected com.google.protobuf.MapField internalGetMutableMapField(
+          int number) {
+        switch (number) {
+          case 2:
+            return internalGetMutableParameterMap();
+          default:
+            throw new RuntimeException(
+                "Invalid map field number: " + number);
+        }
+      }
+      @java.lang.Override
+      protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+          internalGetFieldAccessorTable() {
+        return org.tensorflow.proto.framework.RewriterConfigProtos.internal_static_tensorflow_RewriterConfig_CustomGraphOptimizer_fieldAccessorTable
+            .ensureFieldAccessorsInitialized(
+                org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizer.class, org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizer.Builder.class);
+      }
+
+      // Construct using org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizer.newBuilder()
+      private Builder() {
+        maybeForceBuilderInitialization();
+      }
+
+      private Builder(
+          com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+        super(parent);
+        maybeForceBuilderInitialization();
+      }
+      private void maybeForceBuilderInitialization() {
+        if (com.google.protobuf.GeneratedMessageV3
+                .alwaysUseFieldBuilders) {
+        }
+      }
+      @java.lang.Override
+      public Builder clear() {
+        super.clear();
+        name_ = "";
+
+        internalGetMutableParameterMap().clear();
+        return this;
+      }
+
+      @java.lang.Override
+      public com.google.protobuf.Descriptors.Descriptor
+          getDescriptorForType() {
+        return org.tensorflow.proto.framework.RewriterConfigProtos.internal_static_tensorflow_RewriterConfig_CustomGraphOptimizer_descriptor;
+      }
+
+      @java.lang.Override
+      public org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizer getDefaultInstanceForType() {
+        return org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizer.getDefaultInstance();
+      }
+
+      @java.lang.Override
+      public org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizer build() {
+        org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizer result = buildPartial();
+        if (!result.isInitialized()) {
+          throw newUninitializedMessageException(result);
+        }
+        return result;
+      }
+
+      @java.lang.Override
+      public org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizer buildPartial() {
+        org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizer result = new org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizer(this);
+        int from_bitField0_ = bitField0_;
+        result.name_ = name_;
+        result.parameterMap_ = internalGetParameterMap();
+        result.parameterMap_.makeImmutable();
+        onBuilt();
+        return result;
+      }
+
+      @java.lang.Override
+      public Builder clone() {
+        return super.clone();
+      }
+      @java.lang.Override
+      public Builder setField(
+          com.google.protobuf.Descriptors.FieldDescriptor field,
+          java.lang.Object value) {
+        return super.setField(field, value);
+      }
+      @java.lang.Override
+      public Builder clearField(
+          com.google.protobuf.Descriptors.FieldDescriptor field) {
+        return super.clearField(field);
+      }
+      @java.lang.Override
+      public Builder clearOneof(
+          com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+        return super.clearOneof(oneof);
+      }
+      @java.lang.Override
+      public Builder setRepeatedField(
+          com.google.protobuf.Descriptors.FieldDescriptor field,
+          int index, java.lang.Object value) {
+        return super.setRepeatedField(field, index, value);
+      }
+      @java.lang.Override
+      public Builder addRepeatedField(
+          com.google.protobuf.Descriptors.FieldDescriptor field,
+          java.lang.Object value) {
+        return super.addRepeatedField(field, value);
+      }
+      @java.lang.Override
+      public Builder mergeFrom(com.google.protobuf.Message other) {
+        if (other instanceof org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizer) {
+          return mergeFrom((org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizer)other);
+        } else {
+          super.mergeFrom(other);
+          return this;
+        }
+      }
+
+      public Builder mergeFrom(org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizer other) {
+        if (other == org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizer.getDefaultInstance()) return this;
+        if (!other.getName().isEmpty()) {
+          name_ = other.name_;
+          onChanged();
+        }
+        internalGetMutableParameterMap().mergeFrom(
+            other.internalGetParameterMap());
+        this.mergeUnknownFields(other.unknownFields);
+        onChanged();
+        return this;
+      }
+
+      @java.lang.Override
+      public final boolean isInitialized() {
+        return true;
+      }
+
+      @java.lang.Override
+      public Builder mergeFrom(
+          com.google.protobuf.CodedInputStream input,
+          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+          throws java.io.IOException {
+        org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizer parsedMessage = null;
+        try {
+          parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+        } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+          parsedMessage = (org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizer) e.getUnfinishedMessage();
+          throw e.unwrapIOException();
+        } finally {
+          if (parsedMessage != null) {
+            mergeFrom(parsedMessage);
+          }
+        }
+        return this;
+      }
+      private int bitField0_;
+
+      private java.lang.Object name_ = "";
+      /**
+       * <code>string name = 1;</code>
+       */
+      public java.lang.String getName() {
+        java.lang.Object ref = name_;
+        if (!(ref instanceof java.lang.String)) {
+          com.google.protobuf.ByteString bs =
+              (com.google.protobuf.ByteString) ref;
+          java.lang.String s = bs.toStringUtf8();
+          name_ = s;
+          return s;
+        } else {
+          return (java.lang.String) ref;
+        }
+      }
+      /**
+       * <code>string name = 1;</code>
+       */
+      public com.google.protobuf.ByteString
+          getNameBytes() {
+        java.lang.Object ref = name_;
+        if (ref instanceof String) {
+          com.google.protobuf.ByteString b = 
+              com.google.protobuf.ByteString.copyFromUtf8(
+                  (java.lang.String) ref);
+          name_ = b;
+          return b;
+        } else {
+          return (com.google.protobuf.ByteString) ref;
+        }
+      }
+      /**
+       * <code>string name = 1;</code>
+       */
+      public Builder setName(
+          java.lang.String value) {
+        if (value == null) {
+    throw new NullPointerException();
+  }
+  
+        name_ = value;
+        onChanged();
+        return this;
+      }
+      /**
+       * <code>string name = 1;</code>
+       */
+      public Builder clearName() {
+        
+        name_ = getDefaultInstance().getName();
+        onChanged();
+        return this;
+      }
+      /**
+       * <code>string name = 1;</code>
+       */
+      public Builder setNameBytes(
+          com.google.protobuf.ByteString value) {
+        if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+        
+        name_ = value;
+        onChanged();
+        return this;
+      }
+
+      private com.google.protobuf.MapField<
+          java.lang.String, org.tensorflow.proto.framework.AttrValue> parameterMap_;
+      private com.google.protobuf.MapField<java.lang.String, org.tensorflow.proto.framework.AttrValue>
+      internalGetParameterMap() {
+        if (parameterMap_ == null) {
+          return com.google.protobuf.MapField.emptyMapField(
+              ParameterMapDefaultEntryHolder.defaultEntry);
+        }
+        return parameterMap_;
+      }
+      private com.google.protobuf.MapField<java.lang.String, org.tensorflow.proto.framework.AttrValue>
+      internalGetMutableParameterMap() {
+        onChanged();;
+        if (parameterMap_ == null) {
+          parameterMap_ = com.google.protobuf.MapField.newMapField(
+              ParameterMapDefaultEntryHolder.defaultEntry);
+        }
+        if (!parameterMap_.isMutable()) {
+          parameterMap_ = parameterMap_.copy();
+        }
+        return parameterMap_;
+      }
+
+      public int getParameterMapCount() {
+        return internalGetParameterMap().getMap().size();
+      }
+      /**
+       * <code>map&lt;string, .tensorflow.AttrValue&gt; parameter_map = 2;</code>
+       */
+
+      public boolean containsParameterMap(
+          java.lang.String key) {
+        if (key == null) { throw new java.lang.NullPointerException(); }
+        return internalGetParameterMap().getMap().containsKey(key);
+      }
+      /**
+       * Use {@link #getParameterMapMap()} instead.
+       */
+      @java.lang.Deprecated
+      public java.util.Map<java.lang.String, org.tensorflow.proto.framework.AttrValue> getParameterMap() {
+        return getParameterMapMap();
+      }
+      /**
+       * <code>map&lt;string, .tensorflow.AttrValue&gt; parameter_map = 2;</code>
+       */
+
+      public java.util.Map<java.lang.String, org.tensorflow.proto.framework.AttrValue> getParameterMapMap() {
+        return internalGetParameterMap().getMap();
+      }
+      /**
+       * <code>map&lt;string, .tensorflow.AttrValue&gt; parameter_map = 2;</code>
+       */
+
+      public org.tensorflow.proto.framework.AttrValue getParameterMapOrDefault(
+          java.lang.String key,
+          org.tensorflow.proto.framework.AttrValue defaultValue) {
+        if (key == null) { throw new java.lang.NullPointerException(); }
+        java.util.Map<java.lang.String, org.tensorflow.proto.framework.AttrValue> map =
+            internalGetParameterMap().getMap();
+        return map.containsKey(key) ? map.get(key) : defaultValue;
+      }
+      /**
+       * <code>map&lt;string, .tensorflow.AttrValue&gt; parameter_map = 2;</code>
+       */
+
+      public org.tensorflow.proto.framework.AttrValue getParameterMapOrThrow(
+          java.lang.String key) {
+        if (key == null) { throw new java.lang.NullPointerException(); }
+        java.util.Map<java.lang.String, org.tensorflow.proto.framework.AttrValue> map =
+            internalGetParameterMap().getMap();
+        if (!map.containsKey(key)) {
+          throw new java.lang.IllegalArgumentException();
+        }
+        return map.get(key);
+      }
+
+      public Builder clearParameterMap() {
+        internalGetMutableParameterMap().getMutableMap()
+            .clear();
+        return this;
+      }
+      /**
+       * <code>map&lt;string, .tensorflow.AttrValue&gt; parameter_map = 2;</code>
+       */
+
+      public Builder removeParameterMap(
+          java.lang.String key) {
+        if (key == null) { throw new java.lang.NullPointerException(); }
+        internalGetMutableParameterMap().getMutableMap()
+            .remove(key);
+        return this;
+      }
+      /**
+       * Use alternate mutation accessors instead.
+       */
+      @java.lang.Deprecated
+      public java.util.Map<java.lang.String, org.tensorflow.proto.framework.AttrValue>
+      getMutableParameterMap() {
+        return internalGetMutableParameterMap().getMutableMap();
+      }
+      /**
+       * <code>map&lt;string, .tensorflow.AttrValue&gt; parameter_map = 2;</code>
+       */
+      public Builder putParameterMap(
+          java.lang.String key,
+          org.tensorflow.proto.framework.AttrValue value) {
+        if (key == null) { throw new java.lang.NullPointerException(); }
+        if (value == null) { throw new java.lang.NullPointerException(); }
+        internalGetMutableParameterMap().getMutableMap()
+            .put(key, value);
+        return this;
+      }
+      /**
+       * <code>map&lt;string, .tensorflow.AttrValue&gt; parameter_map = 2;</code>
+       */
+
+      public Builder putAllParameterMap(
+          java.util.Map<java.lang.String, org.tensorflow.proto.framework.AttrValue> values) {
+        internalGetMutableParameterMap().getMutableMap()
+            .putAll(values);
+        return this;
+      }
+      @java.lang.Override
+      public final Builder setUnknownFields(
+          final com.google.protobuf.UnknownFieldSet unknownFields) {
+        return super.setUnknownFields(unknownFields);
+      }
+
+      @java.lang.Override
+      public final Builder mergeUnknownFields(
+          final com.google.protobuf.UnknownFieldSet unknownFields) {
+        return super.mergeUnknownFields(unknownFields);
+      }
+
+
+      // @@protoc_insertion_point(builder_scope:tensorflow.RewriterConfig.CustomGraphOptimizer)
+    }
+
+    // @@protoc_insertion_point(class_scope:tensorflow.RewriterConfig.CustomGraphOptimizer)
+    private static final org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizer DEFAULT_INSTANCE;
+    static {
+      DEFAULT_INSTANCE = new org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizer();
+    }
+
+    public static org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizer getDefaultInstance() {
+      return DEFAULT_INSTANCE;
+    }
+
+    private static final com.google.protobuf.Parser<CustomGraphOptimizer>
+        PARSER = new com.google.protobuf.AbstractParser<CustomGraphOptimizer>() {
+      @java.lang.Override
+      public CustomGraphOptimizer parsePartialFrom(
+          com.google.protobuf.CodedInputStream input,
+          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+          throws com.google.protobuf.InvalidProtocolBufferException {
+        return new CustomGraphOptimizer(input, extensionRegistry);
+      }
+    };
+
+    public static com.google.protobuf.Parser<CustomGraphOptimizer> parser() {
+      return PARSER;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Parser<CustomGraphOptimizer> getParserForType() {
+      return PARSER;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizer getDefaultInstanceForType() {
+      return DEFAULT_INSTANCE;
+    }
+
+  }
+
+  public static final int LAYOUT_OPTIMIZER_FIELD_NUMBER = 1;
+  private int layoutOptimizer_;
+  /**
+   * <pre>
+   * Optimize tensor layouts (default is ON)
+   * e.g. This will try to use NCHW layout on GPU which is faster.
+   * </pre>
+   *
+   * <code>.tensorflow.RewriterConfig.Toggle layout_optimizer = 1;</code>
+   */
+  public int getLayoutOptimizerValue() {
+    return layoutOptimizer_;
+  }
+  /**
+   * <pre>
+   * Optimize tensor layouts (default is ON)
+   * e.g. This will try to use NCHW layout on GPU which is faster.
+   * </pre>
+   *
+   * <code>.tensorflow.RewriterConfig.Toggle layout_optimizer = 1;</code>
+   */
+  public org.tensorflow.proto.framework.RewriterConfig.Toggle getLayoutOptimizer() {
+    @SuppressWarnings("deprecation")
+    org.tensorflow.proto.framework.RewriterConfig.Toggle result = org.tensorflow.proto.framework.RewriterConfig.Toggle.valueOf(layoutOptimizer_);
+    return result == null ? org.tensorflow.proto.framework.RewriterConfig.Toggle.UNRECOGNIZED : result;
+  }
+
+  public static final int CONSTANT_FOLDING_FIELD_NUMBER = 3;
+  private int constantFolding_;
+  /**
+   * <pre>
+   * Fold constants (default is ON)
+   * Statically infer the value of tensors when possible, and materialize the
+   * result using constants.
+   * </pre>
+   *
+   * <code>.tensorflow.RewriterConfig.Toggle constant_folding = 3;</code>
+   */
+  public int getConstantFoldingValue() {
+    return constantFolding_;
+  }
+  /**
+   * <pre>
+   * Fold constants (default is ON)
+   * Statically infer the value of tensors when possible, and materialize the
+   * result using constants.
+   * </pre>
+   *
+   * <code>.tensorflow.RewriterConfig.Toggle constant_folding = 3;</code>
+   */
+  public org.tensorflow.proto.framework.RewriterConfig.Toggle getConstantFolding() {
+    @SuppressWarnings("deprecation")
+    org.tensorflow.proto.framework.RewriterConfig.Toggle result = org.tensorflow.proto.framework.RewriterConfig.Toggle.valueOf(constantFolding_);
+    return result == null ? org.tensorflow.proto.framework.RewriterConfig.Toggle.UNRECOGNIZED : result;
+  }
+
+  public static final int SHAPE_OPTIMIZATION_FIELD_NUMBER = 13;
+  private int shapeOptimization_;
+  /**
+   * <pre>
+   * Shape optimizations (default is ON)
+   * Simplify computations made on shapes.
+   * </pre>
+   *
+   * <code>.tensorflow.RewriterConfig.Toggle shape_optimization = 13;</code>
+   */
+  public int getShapeOptimizationValue() {
+    return shapeOptimization_;
+  }
+  /**
+   * <pre>
+   * Shape optimizations (default is ON)
+   * Simplify computations made on shapes.
+   * </pre>
+   *
+   * <code>.tensorflow.RewriterConfig.Toggle shape_optimization = 13;</code>
+   */
+  public org.tensorflow.proto.framework.RewriterConfig.Toggle getShapeOptimization() {
+    @SuppressWarnings("deprecation")
+    org.tensorflow.proto.framework.RewriterConfig.Toggle result = org.tensorflow.proto.framework.RewriterConfig.Toggle.valueOf(shapeOptimization_);
+    return result == null ? org.tensorflow.proto.framework.RewriterConfig.Toggle.UNRECOGNIZED : result;
+  }
+
+  public static final int REMAPPING_FIELD_NUMBER = 14;
+  private int remapping_;
+  /**
+   * <pre>
+   * Remapping (default is ON)
+   * Remap subgraphs onto more efficient implementations.
+   * </pre>
+   *
+   * <code>.tensorflow.RewriterConfig.Toggle remapping = 14;</code>
+   */
+  public int getRemappingValue() {
+    return remapping_;
+  }
+  /**
+   * <pre>
+   * Remapping (default is ON)
+   * Remap subgraphs onto more efficient implementations.
+   * </pre>
+   *
+   * <code>.tensorflow.RewriterConfig.Toggle remapping = 14;</code>
+   */
+  public org.tensorflow.proto.framework.RewriterConfig.Toggle getRemapping() {
+    @SuppressWarnings("deprecation")
+    org.tensorflow.proto.framework.RewriterConfig.Toggle result = org.tensorflow.proto.framework.RewriterConfig.Toggle.valueOf(remapping_);
+    return result == null ? org.tensorflow.proto.framework.RewriterConfig.Toggle.UNRECOGNIZED : result;
+  }
+
+  public static final int COMMON_SUBGRAPH_ELIMINATION_FIELD_NUMBER = 24;
+  private int commonSubgraphElimination_;
+  /**
+   * <pre>
+   * Common subgraph elimination (default is ON)
+   * e.g. Simplify arithmetic ops; merge ops with same value (like constants).
+   * </pre>
+   *
+   * <code>.tensorflow.RewriterConfig.Toggle common_subgraph_elimination = 24;</code>
+   */
+  public int getCommonSubgraphEliminationValue() {
+    return commonSubgraphElimination_;
+  }
+  /**
+   * <pre>
+   * Common subgraph elimination (default is ON)
+   * e.g. Simplify arithmetic ops; merge ops with same value (like constants).
+   * </pre>
+   *
+   * <code>.tensorflow.RewriterConfig.Toggle common_subgraph_elimination = 24;</code>
+   */
+  public org.tensorflow.proto.framework.RewriterConfig.Toggle getCommonSubgraphElimination() {
+    @SuppressWarnings("deprecation")
+    org.tensorflow.proto.framework.RewriterConfig.Toggle result = org.tensorflow.proto.framework.RewriterConfig.Toggle.valueOf(commonSubgraphElimination_);
+    return result == null ? org.tensorflow.proto.framework.RewriterConfig.Toggle.UNRECOGNIZED : result;
+  }
+
+  public static final int ARITHMETIC_OPTIMIZATION_FIELD_NUMBER = 7;
+  private int arithmeticOptimization_;
+  /**
+   * <pre>
+   * Arithmetic optimizations (default is ON)
+   * e.g. Simplify arithmetic ops; merge ops with same value (like constants).
+   * </pre>
+   *
+   * <code>.tensorflow.RewriterConfig.Toggle arithmetic_optimization = 7;</code>
+   */
+  public int getArithmeticOptimizationValue() {
+    return arithmeticOptimization_;
+  }
+  /**
+   * <pre>
+   * Arithmetic optimizations (default is ON)
+   * e.g. Simplify arithmetic ops; merge ops with same value (like constants).
+   * </pre>
+   *
+   * <code>.tensorflow.RewriterConfig.Toggle arithmetic_optimization = 7;</code>
+   */
+  public org.tensorflow.proto.framework.RewriterConfig.Toggle getArithmeticOptimization() {
+    @SuppressWarnings("deprecation")
+    org.tensorflow.proto.framework.RewriterConfig.Toggle result = org.tensorflow.proto.framework.RewriterConfig.Toggle.valueOf(arithmeticOptimization_);
+    return result == null ? org.tensorflow.proto.framework.RewriterConfig.Toggle.UNRECOGNIZED : result;
+  }
+
+  public static final int DEPENDENCY_OPTIMIZATION_FIELD_NUMBER = 8;
+  private int dependencyOptimization_;
+  /**
+   * <pre>
+   * Control dependency optimizations (default is ON).
+   * Remove redundant control dependencies, which may enable other optimization.
+   * </pre>
+   *
+   * <code>.tensorflow.RewriterConfig.Toggle dependency_optimization = 8;</code>
+   */
+  public int getDependencyOptimizationValue() {
+    return dependencyOptimization_;
+  }
+  /**
+   * <pre>
+   * Control dependency optimizations (default is ON).
+   * Remove redundant control dependencies, which may enable other optimization.
+   * </pre>
+   *
+   * <code>.tensorflow.RewriterConfig.Toggle dependency_optimization = 8;</code>
+   */
+  public org.tensorflow.proto.framework.RewriterConfig.Toggle getDependencyOptimization() {
+    @SuppressWarnings("deprecation")
+    org.tensorflow.proto.framework.RewriterConfig.Toggle result = org.tensorflow.proto.framework.RewriterConfig.Toggle.valueOf(dependencyOptimization_);
+    return result == null ? org.tensorflow.proto.framework.RewriterConfig.Toggle.UNRECOGNIZED : result;
+  }
+
+  public static final int LOOP_OPTIMIZATION_FIELD_NUMBER = 9;
+  private int loopOptimization_;
+  /**
+   * <pre>
+   * Loop optimizations (default is ON).
+   * </pre>
+   *
+   * <code>.tensorflow.RewriterConfig.Toggle loop_optimization = 9;</code>
+   */
+  public int getLoopOptimizationValue() {
+    return loopOptimization_;
+  }
+  /**
+   * <pre>
+   * Loop optimizations (default is ON).
+   * </pre>
+   *
+   * <code>.tensorflow.RewriterConfig.Toggle loop_optimization = 9;</code>
+   */
+  public org.tensorflow.proto.framework.RewriterConfig.Toggle getLoopOptimization() {
+    @SuppressWarnings("deprecation")
+    org.tensorflow.proto.framework.RewriterConfig.Toggle result = org.tensorflow.proto.framework.RewriterConfig.Toggle.valueOf(loopOptimization_);
+    return result == null ? org.tensorflow.proto.framework.RewriterConfig.Toggle.UNRECOGNIZED : result;
+  }
+
+  public static final int FUNCTION_OPTIMIZATION_FIELD_NUMBER = 10;
+  private int functionOptimization_;
+  /**
+   * <pre>
+   * Function optimizations (default is ON).
+   * </pre>
+   *
+   * <code>.tensorflow.RewriterConfig.Toggle function_optimization = 10;</code>
+   */
+  public int getFunctionOptimizationValue() {
+    return functionOptimization_;
+  }
+  /**
+   * <pre>
+   * Function optimizations (default is ON).
+   * </pre>
+   *
+   * <code>.tensorflow.RewriterConfig.Toggle function_optimization = 10;</code>
+   */
+  public org.tensorflow.proto.framework.RewriterConfig.Toggle getFunctionOptimization() {
+    @SuppressWarnings("deprecation")
+    org.tensorflow.proto.framework.RewriterConfig.Toggle result = org.tensorflow.proto.framework.RewriterConfig.Toggle.valueOf(functionOptimization_);
+    return result == null ? org.tensorflow.proto.framework.RewriterConfig.Toggle.UNRECOGNIZED : result;
+  }
+
+  public static final int DEBUG_STRIPPER_FIELD_NUMBER = 11;
+  private int debugStripper_;
+  /**
+   * <pre>
+   * Strips debug-related nodes from the graph (off by default).
+   * </pre>
+   *
+   * <code>.tensorflow.RewriterConfig.Toggle debug_stripper = 11;</code>
+   */
+  public int getDebugStripperValue() {
+    return debugStripper_;
+  }
+  /**
+   * <pre>
+   * Strips debug-related nodes from the graph (off by default).
+   * </pre>
+   *
+   * <code>.tensorflow.RewriterConfig.Toggle debug_stripper = 11;</code>
+   */
+  public org.tensorflow.proto.framework.RewriterConfig.Toggle getDebugStripper() {
+    @SuppressWarnings("deprecation")
+    org.tensorflow.proto.framework.RewriterConfig.Toggle result = org.tensorflow.proto.framework.RewriterConfig.Toggle.valueOf(debugStripper_);
+    return result == null ? org.tensorflow.proto.framework.RewriterConfig.Toggle.UNRECOGNIZED : result;
+  }
+
+  public static final int DISABLE_MODEL_PRUNING_FIELD_NUMBER = 2;
+  private boolean disableModelPruning_;
+  /**
+   * <pre>
+   * If true, don't remove unnecessary ops from the graph
+   * </pre>
+   *
+   * <code>bool disable_model_pruning = 2;</code>
+   */
+  public boolean getDisableModelPruning() {
+    return disableModelPruning_;
+  }
+
+  public static final int SCOPED_ALLOCATOR_OPTIMIZATION_FIELD_NUMBER = 15;
+  private int scopedAllocatorOptimization_;
+  /**
+   * <pre>
+   * Try to allocate some independent Op outputs contiguously in order to
+   * merge or eliminate downstream Ops (off by default).
+   * </pre>
+   *
+   * <code>.tensorflow.RewriterConfig.Toggle scoped_allocator_optimization = 15;</code>
+   */
+  public int getScopedAllocatorOptimizationValue() {
+    return scopedAllocatorOptimization_;
+  }
+  /**
+   * <pre>
+   * Try to allocate some independent Op outputs contiguously in order to
+   * merge or eliminate downstream Ops (off by default).
+   * </pre>
+   *
+   * <code>.tensorflow.RewriterConfig.Toggle scoped_allocator_optimization = 15;</code>
+   */
+  public org.tensorflow.proto.framework.RewriterConfig.Toggle getScopedAllocatorOptimization() {
+    @SuppressWarnings("deprecation")
+    org.tensorflow.proto.framework.RewriterConfig.Toggle result = org.tensorflow.proto.framework.RewriterConfig.Toggle.valueOf(scopedAllocatorOptimization_);
+    return result == null ? org.tensorflow.proto.framework.RewriterConfig.Toggle.UNRECOGNIZED : result;
+  }
+
+  public static final int PIN_TO_HOST_OPTIMIZATION_FIELD_NUMBER = 18;
+  private int pinToHostOptimization_;
+  /**
+   * <pre>
+   * Force small ops onto the CPU (default is OFF).
+   * </pre>
+   *
+   * <code>.tensorflow.RewriterConfig.Toggle pin_to_host_optimization = 18;</code>
+   */
+  public int getPinToHostOptimizationValue() {
+    return pinToHostOptimization_;
+  }
+  /**
+   * <pre>
+   * Force small ops onto the CPU (default is OFF).
+   * </pre>
+   *
+   * <code>.tensorflow.RewriterConfig.Toggle pin_to_host_optimization = 18;</code>
+   */
+  public org.tensorflow.proto.framework.RewriterConfig.Toggle getPinToHostOptimization() {
+    @SuppressWarnings("deprecation")
+    org.tensorflow.proto.framework.RewriterConfig.Toggle result = org.tensorflow.proto.framework.RewriterConfig.Toggle.valueOf(pinToHostOptimization_);
+    return result == null ? org.tensorflow.proto.framework.RewriterConfig.Toggle.UNRECOGNIZED : result;
+  }
+
+  public static final int IMPLEMENTATION_SELECTOR_FIELD_NUMBER = 22;
+  private int implementationSelector_;
+  /**
+   * <pre>
+   * Enable the swap of kernel implementations based on the device placement
+   * (default is ON).
+   * </pre>
+   *
+   * <code>.tensorflow.RewriterConfig.Toggle implementation_selector = 22;</code>
+   */
+  public int getImplementationSelectorValue() {
+    return implementationSelector_;
+  }
+  /**
+   * <pre>
+   * Enable the swap of kernel implementations based on the device placement
+   * (default is ON).
+   * </pre>
+   *
+   * <code>.tensorflow.RewriterConfig.Toggle implementation_selector = 22;</code>
+   */
+  public org.tensorflow.proto.framework.RewriterConfig.Toggle getImplementationSelector() {
+    @SuppressWarnings("deprecation")
+    org.tensorflow.proto.framework.RewriterConfig.Toggle result = org.tensorflow.proto.framework.RewriterConfig.Toggle.valueOf(implementationSelector_);
+    return result == null ? org.tensorflow.proto.framework.RewriterConfig.Toggle.UNRECOGNIZED : result;
+  }
+
+  public static final int AUTO_MIXED_PRECISION_FIELD_NUMBER = 23;
+  private int autoMixedPrecision_;
+  /**
+   * <pre>
+   * Optimize data types for CUDA (default is OFF).
+   * This will try to use float16 on GPU which is faster.
+   * Note that this can change the numerical stability of the graph and may
+   * require the use of loss scaling to maintain model convergence.
+   * </pre>
+   *
+   * <code>.tensorflow.RewriterConfig.Toggle auto_mixed_precision = 23;</code>
+   */
+  public int getAutoMixedPrecisionValue() {
+    return autoMixedPrecision_;
+  }
+  /**
+   * <pre>
+   * Optimize data types for CUDA (default is OFF).
+   * This will try to use float16 on GPU which is faster.
+   * Note that this can change the numerical stability of the graph and may
+   * require the use of loss scaling to maintain model convergence.
+   * </pre>
+   *
+   * <code>.tensorflow.RewriterConfig.Toggle auto_mixed_precision = 23;</code>
+   */
+  public org.tensorflow.proto.framework.RewriterConfig.Toggle getAutoMixedPrecision() {
+    @SuppressWarnings("deprecation")
+    org.tensorflow.proto.framework.RewriterConfig.Toggle result = org.tensorflow.proto.framework.RewriterConfig.Toggle.valueOf(autoMixedPrecision_);
+    return result == null ? org.tensorflow.proto.framework.RewriterConfig.Toggle.UNRECOGNIZED : result;
+  }
+
+  public static final int AUTO_MIXED_PRECISION_MKL_FIELD_NUMBER = 25;
+  private int autoMixedPrecisionMkl_;
+  /**
+   * <pre>
+   * Optimize data types for MKL (default is OFF).
+   * This will try to use bfloat16 on CPUs, which is faster.
+   * Note that this can change the numerical stability of the graph.
+   * </pre>
+   *
+   * <code>.tensorflow.RewriterConfig.Toggle auto_mixed_precision_mkl = 25;</code>
+   */
+  public int getAutoMixedPrecisionMklValue() {
+    return autoMixedPrecisionMkl_;
+  }
+  /**
+   * <pre>
+   * Optimize data types for MKL (default is OFF).
+   * This will try to use bfloat16 on CPUs, which is faster.
+   * Note that this can change the numerical stability of the graph.
+   * </pre>
+   *
+   * <code>.tensorflow.RewriterConfig.Toggle auto_mixed_precision_mkl = 25;</code>
+   */
+  public org.tensorflow.proto.framework.RewriterConfig.Toggle getAutoMixedPrecisionMkl() {
+    @SuppressWarnings("deprecation")
+    org.tensorflow.proto.framework.RewriterConfig.Toggle result = org.tensorflow.proto.framework.RewriterConfig.Toggle.valueOf(autoMixedPrecisionMkl_);
+    return result == null ? org.tensorflow.proto.framework.RewriterConfig.Toggle.UNRECOGNIZED : result;
+  }
+
+  public static final int DISABLE_META_OPTIMIZER_FIELD_NUMBER = 19;
+  private boolean disableMetaOptimizer_;
+  /**
+   * <pre>
+   * Disable the entire meta optimizer (off by default).
+   * </pre>
+   *
+   * <code>bool disable_meta_optimizer = 19;</code>
+   */
+  public boolean getDisableMetaOptimizer() {
+    return disableMetaOptimizer_;
+  }
+
+  public static final int META_OPTIMIZER_ITERATIONS_FIELD_NUMBER = 12;
+  private int metaOptimizerIterations_;
+  /**
+   * <pre>
+   * Controls how many times we run the optimizers in meta optimizer (default
+   * is once).
+   * </pre>
+   *
+   * <code>.tensorflow.RewriterConfig.NumIterationsType meta_optimizer_iterations = 12;</code>
+   */
+  public int getMetaOptimizerIterationsValue() {
+    return metaOptimizerIterations_;
+  }
+  /**
+   * <pre>
+   * Controls how many times we run the optimizers in meta optimizer (default
+   * is once).
+   * </pre>
+   *
+   * <code>.tensorflow.RewriterConfig.NumIterationsType meta_optimizer_iterations = 12;</code>
+   */
+  public org.tensorflow.proto.framework.RewriterConfig.NumIterationsType getMetaOptimizerIterations() {
+    @SuppressWarnings("deprecation")
+    org.tensorflow.proto.framework.RewriterConfig.NumIterationsType result = org.tensorflow.proto.framework.RewriterConfig.NumIterationsType.valueOf(metaOptimizerIterations_);
+    return result == null ? org.tensorflow.proto.framework.RewriterConfig.NumIterationsType.UNRECOGNIZED : result;
+  }
+
+  public static final int MIN_GRAPH_NODES_FIELD_NUMBER = 17;
+  private int minGraphNodes_;
+  /**
+   * <pre>
+   * The minimum number of nodes in a graph to optimizer. For smaller graphs,
+   * optimization is skipped.
+   * 0 means the system picks an appropriate number.
+   * &lt; 0 means do not skip optimization.
+   * </pre>
+   *
+   * <code>int32 min_graph_nodes = 17;</code>
+   */
+  public int getMinGraphNodes() {
+    return minGraphNodes_;
+  }
+
+  public static final int MEMORY_OPTIMIZATION_FIELD_NUMBER = 4;
+  private int memoryOptimization_;
+  /**
+   * <pre>
+   * Configures memory optimization passes through the meta-optimizer. Has no
+   * effect on manually requested memory optimization passes in the optimizers
+   * field.
+   * </pre>
+   *
+   * <code>.tensorflow.RewriterConfig.MemOptType memory_optimization = 4;</code>
+   */
+  public int getMemoryOptimizationValue() {
+    return memoryOptimization_;
+  }
+  /**
+   * <pre>
+   * Configures memory optimization passes through the meta-optimizer. Has no
+   * effect on manually requested memory optimization passes in the optimizers
+   * field.
+   * </pre>
+   *
+   * <code>.tensorflow.RewriterConfig.MemOptType memory_optimization = 4;</code>
+   */
+  public org.tensorflow.proto.framework.RewriterConfig.MemOptType getMemoryOptimization() {
+    @SuppressWarnings("deprecation")
+    org.tensorflow.proto.framework.RewriterConfig.MemOptType result = org.tensorflow.proto.framework.RewriterConfig.MemOptType.valueOf(memoryOptimization_);
+    return result == null ? org.tensorflow.proto.framework.RewriterConfig.MemOptType.UNRECOGNIZED : result;
+  }
+
+  public static final int MEMORY_OPTIMIZER_TARGET_NODE_NAME_SCOPE_FIELD_NUMBER = 6;
+  private volatile java.lang.Object memoryOptimizerTargetNodeNameScope_;
+  /**
+   * <pre>
+   * A node name scope for node names which are valid outputs of recomputations.
+   * Inputs to nodes that match this scope may be recomputed (subject either to
+   * manual annotation of those input nodes or to manual annotation and
+   * heuristics depending on memory_optimization), but the nodes themselves will
+   * not be recomputed. This matches any sub-scopes as well, meaning the scope
+   * can appear not just as a top-level scope. For example, if the value is
+   * "gradients/", the default, it will match node name "gradients/foo",
+   * "foo/gradients/bar", but not "foo_gradients/"
+   * </pre>
+   *
+   * <code>string memory_optimizer_target_node_name_scope = 6;</code>
+   */
+  public java.lang.String getMemoryOptimizerTargetNodeNameScope() {
+    java.lang.Object ref = memoryOptimizerTargetNodeNameScope_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      memoryOptimizerTargetNodeNameScope_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * A node name scope for node names which are valid outputs of recomputations.
+   * Inputs to nodes that match this scope may be recomputed (subject either to
+   * manual annotation of those input nodes or to manual annotation and
+   * heuristics depending on memory_optimization), but the nodes themselves will
+   * not be recomputed. This matches any sub-scopes as well, meaning the scope
+   * can appear not just as a top-level scope. For example, if the value is
+   * "gradients/", the default, it will match node name "gradients/foo",
+   * "foo/gradients/bar", but not "foo_gradients/"
+   * </pre>
+   *
+   * <code>string memory_optimizer_target_node_name_scope = 6;</code>
+   */
+  public com.google.protobuf.ByteString
+      getMemoryOptimizerTargetNodeNameScopeBytes() {
+    java.lang.Object ref = memoryOptimizerTargetNodeNameScope_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      memoryOptimizerTargetNodeNameScope_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int META_OPTIMIZER_TIMEOUT_MS_FIELD_NUMBER = 20;
+  private long metaOptimizerTimeoutMs_;
+  /**
+   * <pre>
+   * Maximum number of milliseconds to spend optimizing a single graph before
+   * timing out. If equal to 0 the system picks a default (currently 5 minutes).
+   * If less than 0 the optimizer will never time out.
+   * </pre>
+   *
+   * <code>int64 meta_optimizer_timeout_ms = 20;</code>
+   */
+  public long getMetaOptimizerTimeoutMs() {
+    return metaOptimizerTimeoutMs_;
+  }
+
+  public static final int AUTO_PARALLEL_FIELD_NUMBER = 5;
+  private org.tensorflow.proto.framework.AutoParallelOptions autoParallel_;
+  /**
+   * <pre>
+   * Configures AutoParallel optimization passes either through the
+   * meta-optimizer or when manually specified through the optimizers field.
+   * </pre>
+   *
+   * <code>.tensorflow.AutoParallelOptions auto_parallel = 5;</code>
+   */
+  public boolean hasAutoParallel() {
+    return autoParallel_ != null;
+  }
+  /**
+   * <pre>
+   * Configures AutoParallel optimization passes either through the
+   * meta-optimizer or when manually specified through the optimizers field.
+   * </pre>
+   *
+   * <code>.tensorflow.AutoParallelOptions auto_parallel = 5;</code>
+   */
+  public org.tensorflow.proto.framework.AutoParallelOptions getAutoParallel() {
+    return autoParallel_ == null ? org.tensorflow.proto.framework.AutoParallelOptions.getDefaultInstance() : autoParallel_;
+  }
+  /**
+   * <pre>
+   * Configures AutoParallel optimization passes either through the
+   * meta-optimizer or when manually specified through the optimizers field.
+   * </pre>
+   *
+   * <code>.tensorflow.AutoParallelOptions auto_parallel = 5;</code>
+   */
+  public org.tensorflow.proto.framework.AutoParallelOptionsOrBuilder getAutoParallelOrBuilder() {
+    return getAutoParallel();
+  }
+
+  public static final int FAIL_ON_OPTIMIZER_ERRORS_FIELD_NUMBER = 21;
+  private boolean failOnOptimizerErrors_;
+  /**
+   * <pre>
+   * If true, any optimization pass failing will cause the MetaOptimizer to
+   * stop with an error. By default - or when set to false, failing passes are
+   * skipped silently.
+   * </pre>
+   *
+   * <code>bool fail_on_optimizer_errors = 21;</code>
+   */
+  public boolean getFailOnOptimizerErrors() {
+    return failOnOptimizerErrors_;
+  }
+
+  public static final int SCOPED_ALLOCATOR_OPTS_FIELD_NUMBER = 16;
+  private org.tensorflow.proto.framework.ScopedAllocatorOptions scopedAllocatorOpts_;
+  /**
+   * <code>.tensorflow.ScopedAllocatorOptions scoped_allocator_opts = 16;</code>
+   */
+  public boolean hasScopedAllocatorOpts() {
+    return scopedAllocatorOpts_ != null;
+  }
+  /**
+   * <code>.tensorflow.ScopedAllocatorOptions scoped_allocator_opts = 16;</code>
+   */
+  public org.tensorflow.proto.framework.ScopedAllocatorOptions getScopedAllocatorOpts() {
+    return scopedAllocatorOpts_ == null ? org.tensorflow.proto.framework.ScopedAllocatorOptions.getDefaultInstance() : scopedAllocatorOpts_;
+  }
+  /**
+   * <code>.tensorflow.ScopedAllocatorOptions scoped_allocator_opts = 16;</code>
+   */
+  public org.tensorflow.proto.framework.ScopedAllocatorOptionsOrBuilder getScopedAllocatorOptsOrBuilder() {
+    return getScopedAllocatorOpts();
+  }
+
+  public static final int OPTIMIZERS_FIELD_NUMBER = 100;
+  private com.google.protobuf.LazyStringList optimizers_;
+  /**
+   * <pre>
+   * If non-empty, will use this as an alternative way to specify a list of
+   * optimizations to turn on and the order of the optimizations (replacing the
+   * meta-optimizer).
+   * Of the RewriterConfig options, only the AutoParallel configuration options
+   * (the auto_parallel field) apply to manually requested optimization passes
+   * ("autoparallel"). Memory optimization passes ("memory") invoked here are
+   * not configurable (in contrast to memory optimization passes through the
+   * meta-optimizer) and act only on manual op annotations.
+   * Custom optimizers (see custom_optimizers) that are not part of this
+   * schedule will be run after - in the order that they were specified.
+   * </pre>
+   *
+   * <code>repeated string optimizers = 100;</code>
+   */
+  public com.google.protobuf.ProtocolStringList
+      getOptimizersList() {
+    return optimizers_;
+  }
+  /**
+   * <pre>
+   * If non-empty, will use this as an alternative way to specify a list of
+   * optimizations to turn on and the order of the optimizations (replacing the
+   * meta-optimizer).
+   * Of the RewriterConfig options, only the AutoParallel configuration options
+   * (the auto_parallel field) apply to manually requested optimization passes
+   * ("autoparallel"). Memory optimization passes ("memory") invoked here are
+   * not configurable (in contrast to memory optimization passes through the
+   * meta-optimizer) and act only on manual op annotations.
+   * Custom optimizers (see custom_optimizers) that are not part of this
+   * schedule will be run after - in the order that they were specified.
+   * </pre>
+   *
+   * <code>repeated string optimizers = 100;</code>
+   */
+  public int getOptimizersCount() {
+    return optimizers_.size();
+  }
+  /**
+   * <pre>
+   * If non-empty, will use this as an alternative way to specify a list of
+   * optimizations to turn on and the order of the optimizations (replacing the
+   * meta-optimizer).
+   * Of the RewriterConfig options, only the AutoParallel configuration options
+   * (the auto_parallel field) apply to manually requested optimization passes
+   * ("autoparallel"). Memory optimization passes ("memory") invoked here are
+   * not configurable (in contrast to memory optimization passes through the
+   * meta-optimizer) and act only on manual op annotations.
+   * Custom optimizers (see custom_optimizers) that are not part of this
+   * schedule will be run after - in the order that they were specified.
+   * </pre>
+   *
+   * <code>repeated string optimizers = 100;</code>
+   */
+  public java.lang.String getOptimizers(int index) {
+    return optimizers_.get(index);
+  }
+  /**
+   * <pre>
+   * If non-empty, will use this as an alternative way to specify a list of
+   * optimizations to turn on and the order of the optimizations (replacing the
+   * meta-optimizer).
+   * Of the RewriterConfig options, only the AutoParallel configuration options
+   * (the auto_parallel field) apply to manually requested optimization passes
+   * ("autoparallel"). Memory optimization passes ("memory") invoked here are
+   * not configurable (in contrast to memory optimization passes through the
+   * meta-optimizer) and act only on manual op annotations.
+   * Custom optimizers (see custom_optimizers) that are not part of this
+   * schedule will be run after - in the order that they were specified.
+   * </pre>
+   *
+   * <code>repeated string optimizers = 100;</code>
+   */
+  public com.google.protobuf.ByteString
+      getOptimizersBytes(int index) {
+    return optimizers_.getByteString(index);
+  }
+
+  public static final int CUSTOM_OPTIMIZERS_FIELD_NUMBER = 200;
+  private java.util.List<org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizer> customOptimizers_;
+  /**
+   * <pre>
+   * list of CustomGraphOptimizers to apply.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.RewriterConfig.CustomGraphOptimizer custom_optimizers = 200;</code>
+   */
+  public java.util.List<org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizer> getCustomOptimizersList() {
+    return customOptimizers_;
+  }
+  /**
+   * <pre>
+   * list of CustomGraphOptimizers to apply.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.RewriterConfig.CustomGraphOptimizer custom_optimizers = 200;</code>
+   */
+  public java.util.List<? extends org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizerOrBuilder> 
+      getCustomOptimizersOrBuilderList() {
+    return customOptimizers_;
+  }
+  /**
+   * <pre>
+   * list of CustomGraphOptimizers to apply.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.RewriterConfig.CustomGraphOptimizer custom_optimizers = 200;</code>
+   */
+  public int getCustomOptimizersCount() {
+    return customOptimizers_.size();
+  }
+  /**
+   * <pre>
+   * list of CustomGraphOptimizers to apply.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.RewriterConfig.CustomGraphOptimizer custom_optimizers = 200;</code>
+   */
+  public org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizer getCustomOptimizers(int index) {
+    return customOptimizers_.get(index);
+  }
+  /**
+   * <pre>
+   * list of CustomGraphOptimizers to apply.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.RewriterConfig.CustomGraphOptimizer custom_optimizers = 200;</code>
+   */
+  public org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizerOrBuilder getCustomOptimizersOrBuilder(
+      int index) {
+    return customOptimizers_.get(index);
+  }
+
+  public static final int INTER_OPTIMIZER_VERIFIER_CONFIG_FIELD_NUMBER = 300;
+  private org.tensorflow.proto.framework.VerifierConfig interOptimizerVerifierConfig_;
+  /**
+   * <pre>
+   * VerifierConfig specifying the verifiers to be run after every optimizer.
+   * </pre>
+   *
+   * <code>.tensorflow.VerifierConfig inter_optimizer_verifier_config = 300;</code>
+   */
+  public boolean hasInterOptimizerVerifierConfig() {
+    return interOptimizerVerifierConfig_ != null;
+  }
+  /**
+   * <pre>
+   * VerifierConfig specifying the verifiers to be run after every optimizer.
+   * </pre>
+   *
+   * <code>.tensorflow.VerifierConfig inter_optimizer_verifier_config = 300;</code>
+   */
+  public org.tensorflow.proto.framework.VerifierConfig getInterOptimizerVerifierConfig() {
+    return interOptimizerVerifierConfig_ == null ? org.tensorflow.proto.framework.VerifierConfig.getDefaultInstance() : interOptimizerVerifierConfig_;
+  }
+  /**
+   * <pre>
+   * VerifierConfig specifying the verifiers to be run after every optimizer.
+   * </pre>
+   *
+   * <code>.tensorflow.VerifierConfig inter_optimizer_verifier_config = 300;</code>
+   */
+  public org.tensorflow.proto.framework.VerifierConfigOrBuilder getInterOptimizerVerifierConfigOrBuilder() {
+    return getInterOptimizerVerifierConfig();
+  }
+
+  public static final int POST_OPTIMIZATION_VERIFIER_CONFIG_FIELD_NUMBER = 301;
+  private org.tensorflow.proto.framework.VerifierConfig postOptimizationVerifierConfig_;
+  /**
+   * <pre>
+   * VerifierConfig specifying the verifiers to be run at the end, after all
+   * optimizers have run.
+   * </pre>
+   *
+   * <code>.tensorflow.VerifierConfig post_optimization_verifier_config = 301;</code>
+   */
+  public boolean hasPostOptimizationVerifierConfig() {
+    return postOptimizationVerifierConfig_ != null;
+  }
+  /**
+   * <pre>
+   * VerifierConfig specifying the verifiers to be run at the end, after all
+   * optimizers have run.
+   * </pre>
+   *
+   * <code>.tensorflow.VerifierConfig post_optimization_verifier_config = 301;</code>
+   */
+  public org.tensorflow.proto.framework.VerifierConfig getPostOptimizationVerifierConfig() {
+    return postOptimizationVerifierConfig_ == null ? org.tensorflow.proto.framework.VerifierConfig.getDefaultInstance() : postOptimizationVerifierConfig_;
+  }
+  /**
+   * <pre>
+   * VerifierConfig specifying the verifiers to be run at the end, after all
+   * optimizers have run.
+   * </pre>
+   *
+   * <code>.tensorflow.VerifierConfig post_optimization_verifier_config = 301;</code>
+   */
+  public org.tensorflow.proto.framework.VerifierConfigOrBuilder getPostOptimizationVerifierConfigOrBuilder() {
+    return getPostOptimizationVerifierConfig();
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    if (layoutOptimizer_ != org.tensorflow.proto.framework.RewriterConfig.Toggle.DEFAULT.getNumber()) {
+      output.writeEnum(1, layoutOptimizer_);
+    }
+    if (disableModelPruning_ != false) {
+      output.writeBool(2, disableModelPruning_);
+    }
+    if (constantFolding_ != org.tensorflow.proto.framework.RewriterConfig.Toggle.DEFAULT.getNumber()) {
+      output.writeEnum(3, constantFolding_);
+    }
+    if (memoryOptimization_ != org.tensorflow.proto.framework.RewriterConfig.MemOptType.DEFAULT_MEM_OPT.getNumber()) {
+      output.writeEnum(4, memoryOptimization_);
+    }
+    if (autoParallel_ != null) {
+      output.writeMessage(5, getAutoParallel());
+    }
+    if (!getMemoryOptimizerTargetNodeNameScopeBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 6, memoryOptimizerTargetNodeNameScope_);
+    }
+    if (arithmeticOptimization_ != org.tensorflow.proto.framework.RewriterConfig.Toggle.DEFAULT.getNumber()) {
+      output.writeEnum(7, arithmeticOptimization_);
+    }
+    if (dependencyOptimization_ != org.tensorflow.proto.framework.RewriterConfig.Toggle.DEFAULT.getNumber()) {
+      output.writeEnum(8, dependencyOptimization_);
+    }
+    if (loopOptimization_ != org.tensorflow.proto.framework.RewriterConfig.Toggle.DEFAULT.getNumber()) {
+      output.writeEnum(9, loopOptimization_);
+    }
+    if (functionOptimization_ != org.tensorflow.proto.framework.RewriterConfig.Toggle.DEFAULT.getNumber()) {
+      output.writeEnum(10, functionOptimization_);
+    }
+    if (debugStripper_ != org.tensorflow.proto.framework.RewriterConfig.Toggle.DEFAULT.getNumber()) {
+      output.writeEnum(11, debugStripper_);
+    }
+    if (metaOptimizerIterations_ != org.tensorflow.proto.framework.RewriterConfig.NumIterationsType.DEFAULT_NUM_ITERS.getNumber()) {
+      output.writeEnum(12, metaOptimizerIterations_);
+    }
+    if (shapeOptimization_ != org.tensorflow.proto.framework.RewriterConfig.Toggle.DEFAULT.getNumber()) {
+      output.writeEnum(13, shapeOptimization_);
+    }
+    if (remapping_ != org.tensorflow.proto.framework.RewriterConfig.Toggle.DEFAULT.getNumber()) {
+      output.writeEnum(14, remapping_);
+    }
+    if (scopedAllocatorOptimization_ != org.tensorflow.proto.framework.RewriterConfig.Toggle.DEFAULT.getNumber()) {
+      output.writeEnum(15, scopedAllocatorOptimization_);
+    }
+    if (scopedAllocatorOpts_ != null) {
+      output.writeMessage(16, getScopedAllocatorOpts());
+    }
+    if (minGraphNodes_ != 0) {
+      output.writeInt32(17, minGraphNodes_);
+    }
+    if (pinToHostOptimization_ != org.tensorflow.proto.framework.RewriterConfig.Toggle.DEFAULT.getNumber()) {
+      output.writeEnum(18, pinToHostOptimization_);
+    }
+    if (disableMetaOptimizer_ != false) {
+      output.writeBool(19, disableMetaOptimizer_);
+    }
+    if (metaOptimizerTimeoutMs_ != 0L) {
+      output.writeInt64(20, metaOptimizerTimeoutMs_);
+    }
+    if (failOnOptimizerErrors_ != false) {
+      output.writeBool(21, failOnOptimizerErrors_);
+    }
+    if (implementationSelector_ != org.tensorflow.proto.framework.RewriterConfig.Toggle.DEFAULT.getNumber()) {
+      output.writeEnum(22, implementationSelector_);
+    }
+    if (autoMixedPrecision_ != org.tensorflow.proto.framework.RewriterConfig.Toggle.DEFAULT.getNumber()) {
+      output.writeEnum(23, autoMixedPrecision_);
+    }
+    if (commonSubgraphElimination_ != org.tensorflow.proto.framework.RewriterConfig.Toggle.DEFAULT.getNumber()) {
+      output.writeEnum(24, commonSubgraphElimination_);
+    }
+    if (autoMixedPrecisionMkl_ != org.tensorflow.proto.framework.RewriterConfig.Toggle.DEFAULT.getNumber()) {
+      output.writeEnum(25, autoMixedPrecisionMkl_);
+    }
+    for (int i = 0; i < optimizers_.size(); i++) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 100, optimizers_.getRaw(i));
+    }
+    for (int i = 0; i < customOptimizers_.size(); i++) {
+      output.writeMessage(200, customOptimizers_.get(i));
+    }
+    if (interOptimizerVerifierConfig_ != null) {
+      output.writeMessage(300, getInterOptimizerVerifierConfig());
+    }
+    if (postOptimizationVerifierConfig_ != null) {
+      output.writeMessage(301, getPostOptimizationVerifierConfig());
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (layoutOptimizer_ != org.tensorflow.proto.framework.RewriterConfig.Toggle.DEFAULT.getNumber()) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeEnumSize(1, layoutOptimizer_);
+    }
+    if (disableModelPruning_ != false) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeBoolSize(2, disableModelPruning_);
+    }
+    if (constantFolding_ != org.tensorflow.proto.framework.RewriterConfig.Toggle.DEFAULT.getNumber()) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeEnumSize(3, constantFolding_);
+    }
+    if (memoryOptimization_ != org.tensorflow.proto.framework.RewriterConfig.MemOptType.DEFAULT_MEM_OPT.getNumber()) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeEnumSize(4, memoryOptimization_);
+    }
+    if (autoParallel_ != null) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(5, getAutoParallel());
+    }
+    if (!getMemoryOptimizerTargetNodeNameScopeBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(6, memoryOptimizerTargetNodeNameScope_);
+    }
+    if (arithmeticOptimization_ != org.tensorflow.proto.framework.RewriterConfig.Toggle.DEFAULT.getNumber()) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeEnumSize(7, arithmeticOptimization_);
+    }
+    if (dependencyOptimization_ != org.tensorflow.proto.framework.RewriterConfig.Toggle.DEFAULT.getNumber()) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeEnumSize(8, dependencyOptimization_);
+    }
+    if (loopOptimization_ != org.tensorflow.proto.framework.RewriterConfig.Toggle.DEFAULT.getNumber()) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeEnumSize(9, loopOptimization_);
+    }
+    if (functionOptimization_ != org.tensorflow.proto.framework.RewriterConfig.Toggle.DEFAULT.getNumber()) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeEnumSize(10, functionOptimization_);
+    }
+    if (debugStripper_ != org.tensorflow.proto.framework.RewriterConfig.Toggle.DEFAULT.getNumber()) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeEnumSize(11, debugStripper_);
+    }
+    if (metaOptimizerIterations_ != org.tensorflow.proto.framework.RewriterConfig.NumIterationsType.DEFAULT_NUM_ITERS.getNumber()) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeEnumSize(12, metaOptimizerIterations_);
+    }
+    if (shapeOptimization_ != org.tensorflow.proto.framework.RewriterConfig.Toggle.DEFAULT.getNumber()) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeEnumSize(13, shapeOptimization_);
+    }
+    if (remapping_ != org.tensorflow.proto.framework.RewriterConfig.Toggle.DEFAULT.getNumber()) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeEnumSize(14, remapping_);
+    }
+    if (scopedAllocatorOptimization_ != org.tensorflow.proto.framework.RewriterConfig.Toggle.DEFAULT.getNumber()) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeEnumSize(15, scopedAllocatorOptimization_);
+    }
+    if (scopedAllocatorOpts_ != null) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(16, getScopedAllocatorOpts());
+    }
+    if (minGraphNodes_ != 0) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt32Size(17, minGraphNodes_);
+    }
+    if (pinToHostOptimization_ != org.tensorflow.proto.framework.RewriterConfig.Toggle.DEFAULT.getNumber()) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeEnumSize(18, pinToHostOptimization_);
+    }
+    if (disableMetaOptimizer_ != false) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeBoolSize(19, disableMetaOptimizer_);
+    }
+    if (metaOptimizerTimeoutMs_ != 0L) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt64Size(20, metaOptimizerTimeoutMs_);
+    }
+    if (failOnOptimizerErrors_ != false) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeBoolSize(21, failOnOptimizerErrors_);
+    }
+    if (implementationSelector_ != org.tensorflow.proto.framework.RewriterConfig.Toggle.DEFAULT.getNumber()) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeEnumSize(22, implementationSelector_);
+    }
+    if (autoMixedPrecision_ != org.tensorflow.proto.framework.RewriterConfig.Toggle.DEFAULT.getNumber()) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeEnumSize(23, autoMixedPrecision_);
+    }
+    if (commonSubgraphElimination_ != org.tensorflow.proto.framework.RewriterConfig.Toggle.DEFAULT.getNumber()) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeEnumSize(24, commonSubgraphElimination_);
+    }
+    if (autoMixedPrecisionMkl_ != org.tensorflow.proto.framework.RewriterConfig.Toggle.DEFAULT.getNumber()) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeEnumSize(25, autoMixedPrecisionMkl_);
+    }
+    {
+      int dataSize = 0;
+      for (int i = 0; i < optimizers_.size(); i++) {
+        dataSize += computeStringSizeNoTag(optimizers_.getRaw(i));
+      }
+      size += dataSize;
+      size += 2 * getOptimizersList().size();
+    }
+    for (int i = 0; i < customOptimizers_.size(); i++) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(200, customOptimizers_.get(i));
+    }
+    if (interOptimizerVerifierConfig_ != null) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(300, getInterOptimizerVerifierConfig());
+    }
+    if (postOptimizationVerifierConfig_ != null) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(301, getPostOptimizationVerifierConfig());
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.framework.RewriterConfig)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.framework.RewriterConfig other = (org.tensorflow.proto.framework.RewriterConfig) obj;
+
+    if (layoutOptimizer_ != other.layoutOptimizer_) return false;
+    if (constantFolding_ != other.constantFolding_) return false;
+    if (shapeOptimization_ != other.shapeOptimization_) return false;
+    if (remapping_ != other.remapping_) return false;
+    if (commonSubgraphElimination_ != other.commonSubgraphElimination_) return false;
+    if (arithmeticOptimization_ != other.arithmeticOptimization_) return false;
+    if (dependencyOptimization_ != other.dependencyOptimization_) return false;
+    if (loopOptimization_ != other.loopOptimization_) return false;
+    if (functionOptimization_ != other.functionOptimization_) return false;
+    if (debugStripper_ != other.debugStripper_) return false;
+    if (getDisableModelPruning()
+        != other.getDisableModelPruning()) return false;
+    if (scopedAllocatorOptimization_ != other.scopedAllocatorOptimization_) return false;
+    if (pinToHostOptimization_ != other.pinToHostOptimization_) return false;
+    if (implementationSelector_ != other.implementationSelector_) return false;
+    if (autoMixedPrecision_ != other.autoMixedPrecision_) return false;
+    if (autoMixedPrecisionMkl_ != other.autoMixedPrecisionMkl_) return false;
+    if (getDisableMetaOptimizer()
+        != other.getDisableMetaOptimizer()) return false;
+    if (metaOptimizerIterations_ != other.metaOptimizerIterations_) return false;
+    if (getMinGraphNodes()
+        != other.getMinGraphNodes()) return false;
+    if (memoryOptimization_ != other.memoryOptimization_) return false;
+    if (!getMemoryOptimizerTargetNodeNameScope()
+        .equals(other.getMemoryOptimizerTargetNodeNameScope())) return false;
+    if (getMetaOptimizerTimeoutMs()
+        != other.getMetaOptimizerTimeoutMs()) return false;
+    if (hasAutoParallel() != other.hasAutoParallel()) return false;
+    if (hasAutoParallel()) {
+      if (!getAutoParallel()
+          .equals(other.getAutoParallel())) return false;
+    }
+    if (getFailOnOptimizerErrors()
+        != other.getFailOnOptimizerErrors()) return false;
+    if (hasScopedAllocatorOpts() != other.hasScopedAllocatorOpts()) return false;
+    if (hasScopedAllocatorOpts()) {
+      if (!getScopedAllocatorOpts()
+          .equals(other.getScopedAllocatorOpts())) return false;
+    }
+    if (!getOptimizersList()
+        .equals(other.getOptimizersList())) return false;
+    if (!getCustomOptimizersList()
+        .equals(other.getCustomOptimizersList())) return false;
+    if (hasInterOptimizerVerifierConfig() != other.hasInterOptimizerVerifierConfig()) return false;
+    if (hasInterOptimizerVerifierConfig()) {
+      if (!getInterOptimizerVerifierConfig()
+          .equals(other.getInterOptimizerVerifierConfig())) return false;
+    }
+    if (hasPostOptimizationVerifierConfig() != other.hasPostOptimizationVerifierConfig()) return false;
+    if (hasPostOptimizationVerifierConfig()) {
+      if (!getPostOptimizationVerifierConfig()
+          .equals(other.getPostOptimizationVerifierConfig())) return false;
+    }
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    hash = (37 * hash) + LAYOUT_OPTIMIZER_FIELD_NUMBER;
+    hash = (53 * hash) + layoutOptimizer_;
+    hash = (37 * hash) + CONSTANT_FOLDING_FIELD_NUMBER;
+    hash = (53 * hash) + constantFolding_;
+    hash = (37 * hash) + SHAPE_OPTIMIZATION_FIELD_NUMBER;
+    hash = (53 * hash) + shapeOptimization_;
+    hash = (37 * hash) + REMAPPING_FIELD_NUMBER;
+    hash = (53 * hash) + remapping_;
+    hash = (37 * hash) + COMMON_SUBGRAPH_ELIMINATION_FIELD_NUMBER;
+    hash = (53 * hash) + commonSubgraphElimination_;
+    hash = (37 * hash) + ARITHMETIC_OPTIMIZATION_FIELD_NUMBER;
+    hash = (53 * hash) + arithmeticOptimization_;
+    hash = (37 * hash) + DEPENDENCY_OPTIMIZATION_FIELD_NUMBER;
+    hash = (53 * hash) + dependencyOptimization_;
+    hash = (37 * hash) + LOOP_OPTIMIZATION_FIELD_NUMBER;
+    hash = (53 * hash) + loopOptimization_;
+    hash = (37 * hash) + FUNCTION_OPTIMIZATION_FIELD_NUMBER;
+    hash = (53 * hash) + functionOptimization_;
+    hash = (37 * hash) + DEBUG_STRIPPER_FIELD_NUMBER;
+    hash = (53 * hash) + debugStripper_;
+    hash = (37 * hash) + DISABLE_MODEL_PRUNING_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashBoolean(
+        getDisableModelPruning());
+    hash = (37 * hash) + SCOPED_ALLOCATOR_OPTIMIZATION_FIELD_NUMBER;
+    hash = (53 * hash) + scopedAllocatorOptimization_;
+    hash = (37 * hash) + PIN_TO_HOST_OPTIMIZATION_FIELD_NUMBER;
+    hash = (53 * hash) + pinToHostOptimization_;
+    hash = (37 * hash) + IMPLEMENTATION_SELECTOR_FIELD_NUMBER;
+    hash = (53 * hash) + implementationSelector_;
+    hash = (37 * hash) + AUTO_MIXED_PRECISION_FIELD_NUMBER;
+    hash = (53 * hash) + autoMixedPrecision_;
+    hash = (37 * hash) + AUTO_MIXED_PRECISION_MKL_FIELD_NUMBER;
+    hash = (53 * hash) + autoMixedPrecisionMkl_;
+    hash = (37 * hash) + DISABLE_META_OPTIMIZER_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashBoolean(
+        getDisableMetaOptimizer());
+    hash = (37 * hash) + META_OPTIMIZER_ITERATIONS_FIELD_NUMBER;
+    hash = (53 * hash) + metaOptimizerIterations_;
+    hash = (37 * hash) + MIN_GRAPH_NODES_FIELD_NUMBER;
+    hash = (53 * hash) + getMinGraphNodes();
+    hash = (37 * hash) + MEMORY_OPTIMIZATION_FIELD_NUMBER;
+    hash = (53 * hash) + memoryOptimization_;
+    hash = (37 * hash) + MEMORY_OPTIMIZER_TARGET_NODE_NAME_SCOPE_FIELD_NUMBER;
+    hash = (53 * hash) + getMemoryOptimizerTargetNodeNameScope().hashCode();
+    hash = (37 * hash) + META_OPTIMIZER_TIMEOUT_MS_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+        getMetaOptimizerTimeoutMs());
+    if (hasAutoParallel()) {
+      hash = (37 * hash) + AUTO_PARALLEL_FIELD_NUMBER;
+      hash = (53 * hash) + getAutoParallel().hashCode();
+    }
+    hash = (37 * hash) + FAIL_ON_OPTIMIZER_ERRORS_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashBoolean(
+        getFailOnOptimizerErrors());
+    if (hasScopedAllocatorOpts()) {
+      hash = (37 * hash) + SCOPED_ALLOCATOR_OPTS_FIELD_NUMBER;
+      hash = (53 * hash) + getScopedAllocatorOpts().hashCode();
+    }
+    if (getOptimizersCount() > 0) {
+      hash = (37 * hash) + OPTIMIZERS_FIELD_NUMBER;
+      hash = (53 * hash) + getOptimizersList().hashCode();
+    }
+    if (getCustomOptimizersCount() > 0) {
+      hash = (37 * hash) + CUSTOM_OPTIMIZERS_FIELD_NUMBER;
+      hash = (53 * hash) + getCustomOptimizersList().hashCode();
+    }
+    if (hasInterOptimizerVerifierConfig()) {
+      hash = (37 * hash) + INTER_OPTIMIZER_VERIFIER_CONFIG_FIELD_NUMBER;
+      hash = (53 * hash) + getInterOptimizerVerifierConfig().hashCode();
+    }
+    if (hasPostOptimizationVerifierConfig()) {
+      hash = (37 * hash) + POST_OPTIMIZATION_VERIFIER_CONFIG_FIELD_NUMBER;
+      hash = (53 * hash) + getPostOptimizationVerifierConfig().hashCode();
+    }
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.framework.RewriterConfig parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.RewriterConfig parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.RewriterConfig parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.RewriterConfig parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.RewriterConfig parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.RewriterConfig parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.RewriterConfig parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.RewriterConfig parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.RewriterConfig parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.RewriterConfig parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.RewriterConfig parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.RewriterConfig parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.framework.RewriterConfig prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * <pre>
+   * Graph rewriting is experimental and subject to change, not covered by any
+   * API stability guarantees.
+   * </pre>
+   *
+   * Protobuf type {@code tensorflow.RewriterConfig}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.RewriterConfig)
+      org.tensorflow.proto.framework.RewriterConfigOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.RewriterConfigProtos.internal_static_tensorflow_RewriterConfig_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.RewriterConfigProtos.internal_static_tensorflow_RewriterConfig_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.RewriterConfig.class, org.tensorflow.proto.framework.RewriterConfig.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.framework.RewriterConfig.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+        getCustomOptimizersFieldBuilder();
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      layoutOptimizer_ = 0;
+
+      constantFolding_ = 0;
+
+      shapeOptimization_ = 0;
+
+      remapping_ = 0;
+
+      commonSubgraphElimination_ = 0;
+
+      arithmeticOptimization_ = 0;
+
+      dependencyOptimization_ = 0;
+
+      loopOptimization_ = 0;
+
+      functionOptimization_ = 0;
+
+      debugStripper_ = 0;
+
+      disableModelPruning_ = false;
+
+      scopedAllocatorOptimization_ = 0;
+
+      pinToHostOptimization_ = 0;
+
+      implementationSelector_ = 0;
+
+      autoMixedPrecision_ = 0;
+
+      autoMixedPrecisionMkl_ = 0;
+
+      disableMetaOptimizer_ = false;
+
+      metaOptimizerIterations_ = 0;
+
+      minGraphNodes_ = 0;
+
+      memoryOptimization_ = 0;
+
+      memoryOptimizerTargetNodeNameScope_ = "";
+
+      metaOptimizerTimeoutMs_ = 0L;
+
+      if (autoParallelBuilder_ == null) {
+        autoParallel_ = null;
+      } else {
+        autoParallel_ = null;
+        autoParallelBuilder_ = null;
+      }
+      failOnOptimizerErrors_ = false;
+
+      if (scopedAllocatorOptsBuilder_ == null) {
+        scopedAllocatorOpts_ = null;
+      } else {
+        scopedAllocatorOpts_ = null;
+        scopedAllocatorOptsBuilder_ = null;
+      }
+      optimizers_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+      bitField0_ = (bitField0_ & ~0x00000001);
+      if (customOptimizersBuilder_ == null) {
+        customOptimizers_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000002);
+      } else {
+        customOptimizersBuilder_.clear();
+      }
+      if (interOptimizerVerifierConfigBuilder_ == null) {
+        interOptimizerVerifierConfig_ = null;
+      } else {
+        interOptimizerVerifierConfig_ = null;
+        interOptimizerVerifierConfigBuilder_ = null;
+      }
+      if (postOptimizationVerifierConfigBuilder_ == null) {
+        postOptimizationVerifierConfig_ = null;
+      } else {
+        postOptimizationVerifierConfig_ = null;
+        postOptimizationVerifierConfigBuilder_ = null;
+      }
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.framework.RewriterConfigProtos.internal_static_tensorflow_RewriterConfig_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.RewriterConfig getDefaultInstanceForType() {
+      return org.tensorflow.proto.framework.RewriterConfig.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.RewriterConfig build() {
+      org.tensorflow.proto.framework.RewriterConfig result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.RewriterConfig buildPartial() {
+      org.tensorflow.proto.framework.RewriterConfig result = new org.tensorflow.proto.framework.RewriterConfig(this);
+      int from_bitField0_ = bitField0_;
+      result.layoutOptimizer_ = layoutOptimizer_;
+      result.constantFolding_ = constantFolding_;
+      result.shapeOptimization_ = shapeOptimization_;
+      result.remapping_ = remapping_;
+      result.commonSubgraphElimination_ = commonSubgraphElimination_;
+      result.arithmeticOptimization_ = arithmeticOptimization_;
+      result.dependencyOptimization_ = dependencyOptimization_;
+      result.loopOptimization_ = loopOptimization_;
+      result.functionOptimization_ = functionOptimization_;
+      result.debugStripper_ = debugStripper_;
+      result.disableModelPruning_ = disableModelPruning_;
+      result.scopedAllocatorOptimization_ = scopedAllocatorOptimization_;
+      result.pinToHostOptimization_ = pinToHostOptimization_;
+      result.implementationSelector_ = implementationSelector_;
+      result.autoMixedPrecision_ = autoMixedPrecision_;
+      result.autoMixedPrecisionMkl_ = autoMixedPrecisionMkl_;
+      result.disableMetaOptimizer_ = disableMetaOptimizer_;
+      result.metaOptimizerIterations_ = metaOptimizerIterations_;
+      result.minGraphNodes_ = minGraphNodes_;
+      result.memoryOptimization_ = memoryOptimization_;
+      result.memoryOptimizerTargetNodeNameScope_ = memoryOptimizerTargetNodeNameScope_;
+      result.metaOptimizerTimeoutMs_ = metaOptimizerTimeoutMs_;
+      if (autoParallelBuilder_ == null) {
+        result.autoParallel_ = autoParallel_;
+      } else {
+        result.autoParallel_ = autoParallelBuilder_.build();
+      }
+      result.failOnOptimizerErrors_ = failOnOptimizerErrors_;
+      if (scopedAllocatorOptsBuilder_ == null) {
+        result.scopedAllocatorOpts_ = scopedAllocatorOpts_;
+      } else {
+        result.scopedAllocatorOpts_ = scopedAllocatorOptsBuilder_.build();
+      }
+      if (((bitField0_ & 0x00000001) != 0)) {
+        optimizers_ = optimizers_.getUnmodifiableView();
+        bitField0_ = (bitField0_ & ~0x00000001);
+      }
+      result.optimizers_ = optimizers_;
+      if (customOptimizersBuilder_ == null) {
+        if (((bitField0_ & 0x00000002) != 0)) {
+          customOptimizers_ = java.util.Collections.unmodifiableList(customOptimizers_);
+          bitField0_ = (bitField0_ & ~0x00000002);
+        }
+        result.customOptimizers_ = customOptimizers_;
+      } else {
+        result.customOptimizers_ = customOptimizersBuilder_.build();
+      }
+      if (interOptimizerVerifierConfigBuilder_ == null) {
+        result.interOptimizerVerifierConfig_ = interOptimizerVerifierConfig_;
+      } else {
+        result.interOptimizerVerifierConfig_ = interOptimizerVerifierConfigBuilder_.build();
+      }
+      if (postOptimizationVerifierConfigBuilder_ == null) {
+        result.postOptimizationVerifierConfig_ = postOptimizationVerifierConfig_;
+      } else {
+        result.postOptimizationVerifierConfig_ = postOptimizationVerifierConfigBuilder_.build();
+      }
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.framework.RewriterConfig) {
+        return mergeFrom((org.tensorflow.proto.framework.RewriterConfig)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.framework.RewriterConfig other) {
+      if (other == org.tensorflow.proto.framework.RewriterConfig.getDefaultInstance()) return this;
+      if (other.layoutOptimizer_ != 0) {
+        setLayoutOptimizerValue(other.getLayoutOptimizerValue());
+      }
+      if (other.constantFolding_ != 0) {
+        setConstantFoldingValue(other.getConstantFoldingValue());
+      }
+      if (other.shapeOptimization_ != 0) {
+        setShapeOptimizationValue(other.getShapeOptimizationValue());
+      }
+      if (other.remapping_ != 0) {
+        setRemappingValue(other.getRemappingValue());
+      }
+      if (other.commonSubgraphElimination_ != 0) {
+        setCommonSubgraphEliminationValue(other.getCommonSubgraphEliminationValue());
+      }
+      if (other.arithmeticOptimization_ != 0) {
+        setArithmeticOptimizationValue(other.getArithmeticOptimizationValue());
+      }
+      if (other.dependencyOptimization_ != 0) {
+        setDependencyOptimizationValue(other.getDependencyOptimizationValue());
+      }
+      if (other.loopOptimization_ != 0) {
+        setLoopOptimizationValue(other.getLoopOptimizationValue());
+      }
+      if (other.functionOptimization_ != 0) {
+        setFunctionOptimizationValue(other.getFunctionOptimizationValue());
+      }
+      if (other.debugStripper_ != 0) {
+        setDebugStripperValue(other.getDebugStripperValue());
+      }
+      if (other.getDisableModelPruning() != false) {
+        setDisableModelPruning(other.getDisableModelPruning());
+      }
+      if (other.scopedAllocatorOptimization_ != 0) {
+        setScopedAllocatorOptimizationValue(other.getScopedAllocatorOptimizationValue());
+      }
+      if (other.pinToHostOptimization_ != 0) {
+        setPinToHostOptimizationValue(other.getPinToHostOptimizationValue());
+      }
+      if (other.implementationSelector_ != 0) {
+        setImplementationSelectorValue(other.getImplementationSelectorValue());
+      }
+      if (other.autoMixedPrecision_ != 0) {
+        setAutoMixedPrecisionValue(other.getAutoMixedPrecisionValue());
+      }
+      if (other.autoMixedPrecisionMkl_ != 0) {
+        setAutoMixedPrecisionMklValue(other.getAutoMixedPrecisionMklValue());
+      }
+      if (other.getDisableMetaOptimizer() != false) {
+        setDisableMetaOptimizer(other.getDisableMetaOptimizer());
+      }
+      if (other.metaOptimizerIterations_ != 0) {
+        setMetaOptimizerIterationsValue(other.getMetaOptimizerIterationsValue());
+      }
+      if (other.getMinGraphNodes() != 0) {
+        setMinGraphNodes(other.getMinGraphNodes());
+      }
+      if (other.memoryOptimization_ != 0) {
+        setMemoryOptimizationValue(other.getMemoryOptimizationValue());
+      }
+      if (!other.getMemoryOptimizerTargetNodeNameScope().isEmpty()) {
+        memoryOptimizerTargetNodeNameScope_ = other.memoryOptimizerTargetNodeNameScope_;
+        onChanged();
+      }
+      if (other.getMetaOptimizerTimeoutMs() != 0L) {
+        setMetaOptimizerTimeoutMs(other.getMetaOptimizerTimeoutMs());
+      }
+      if (other.hasAutoParallel()) {
+        mergeAutoParallel(other.getAutoParallel());
+      }
+      if (other.getFailOnOptimizerErrors() != false) {
+        setFailOnOptimizerErrors(other.getFailOnOptimizerErrors());
+      }
+      if (other.hasScopedAllocatorOpts()) {
+        mergeScopedAllocatorOpts(other.getScopedAllocatorOpts());
+      }
+      if (!other.optimizers_.isEmpty()) {
+        if (optimizers_.isEmpty()) {
+          optimizers_ = other.optimizers_;
+          bitField0_ = (bitField0_ & ~0x00000001);
+        } else {
+          ensureOptimizersIsMutable();
+          optimizers_.addAll(other.optimizers_);
+        }
+        onChanged();
+      }
+      if (customOptimizersBuilder_ == null) {
+        if (!other.customOptimizers_.isEmpty()) {
+          if (customOptimizers_.isEmpty()) {
+            customOptimizers_ = other.customOptimizers_;
+            bitField0_ = (bitField0_ & ~0x00000002);
+          } else {
+            ensureCustomOptimizersIsMutable();
+            customOptimizers_.addAll(other.customOptimizers_);
+          }
+          onChanged();
+        }
+      } else {
+        if (!other.customOptimizers_.isEmpty()) {
+          if (customOptimizersBuilder_.isEmpty()) {
+            customOptimizersBuilder_.dispose();
+            customOptimizersBuilder_ = null;
+            customOptimizers_ = other.customOptimizers_;
+            bitField0_ = (bitField0_ & ~0x00000002);
+            customOptimizersBuilder_ = 
+              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
+                 getCustomOptimizersFieldBuilder() : null;
+          } else {
+            customOptimizersBuilder_.addAllMessages(other.customOptimizers_);
+          }
+        }
+      }
+      if (other.hasInterOptimizerVerifierConfig()) {
+        mergeInterOptimizerVerifierConfig(other.getInterOptimizerVerifierConfig());
+      }
+      if (other.hasPostOptimizationVerifierConfig()) {
+        mergePostOptimizationVerifierConfig(other.getPostOptimizationVerifierConfig());
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.framework.RewriterConfig parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.framework.RewriterConfig) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+    private int bitField0_;
+
+    private int layoutOptimizer_ = 0;
+    /**
+     * <pre>
+     * Optimize tensor layouts (default is ON)
+     * e.g. This will try to use NCHW layout on GPU which is faster.
+     * </pre>
+     *
+     * <code>.tensorflow.RewriterConfig.Toggle layout_optimizer = 1;</code>
+     */
+    public int getLayoutOptimizerValue() {
+      return layoutOptimizer_;
+    }
+    /**
+     * <pre>
+     * Optimize tensor layouts (default is ON)
+     * e.g. This will try to use NCHW layout on GPU which is faster.
+     * </pre>
+     *
+     * <code>.tensorflow.RewriterConfig.Toggle layout_optimizer = 1;</code>
+     */
+    public Builder setLayoutOptimizerValue(int value) {
+      layoutOptimizer_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Optimize tensor layouts (default is ON)
+     * e.g. This will try to use NCHW layout on GPU which is faster.
+     * </pre>
+     *
+     * <code>.tensorflow.RewriterConfig.Toggle layout_optimizer = 1;</code>
+     */
+    public org.tensorflow.proto.framework.RewriterConfig.Toggle getLayoutOptimizer() {
+      @SuppressWarnings("deprecation")
+      org.tensorflow.proto.framework.RewriterConfig.Toggle result = org.tensorflow.proto.framework.RewriterConfig.Toggle.valueOf(layoutOptimizer_);
+      return result == null ? org.tensorflow.proto.framework.RewriterConfig.Toggle.UNRECOGNIZED : result;
+    }
+    /**
+     * <pre>
+     * Optimize tensor layouts (default is ON)
+     * e.g. This will try to use NCHW layout on GPU which is faster.
+     * </pre>
+     *
+     * <code>.tensorflow.RewriterConfig.Toggle layout_optimizer = 1;</code>
+     */
+    public Builder setLayoutOptimizer(org.tensorflow.proto.framework.RewriterConfig.Toggle value) {
+      if (value == null) {
+        throw new NullPointerException();
+      }
+      
+      layoutOptimizer_ = value.getNumber();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Optimize tensor layouts (default is ON)
+     * e.g. This will try to use NCHW layout on GPU which is faster.
+     * </pre>
+     *
+     * <code>.tensorflow.RewriterConfig.Toggle layout_optimizer = 1;</code>
+     */
+    public Builder clearLayoutOptimizer() {
+      
+      layoutOptimizer_ = 0;
+      onChanged();
+      return this;
+    }
+
+    private int constantFolding_ = 0;
+    /**
+     * <pre>
+     * Fold constants (default is ON)
+     * Statically infer the value of tensors when possible, and materialize the
+     * result using constants.
+     * </pre>
+     *
+     * <code>.tensorflow.RewriterConfig.Toggle constant_folding = 3;</code>
+     */
+    public int getConstantFoldingValue() {
+      return constantFolding_;
+    }
+    /**
+     * <pre>
+     * Fold constants (default is ON)
+     * Statically infer the value of tensors when possible, and materialize the
+     * result using constants.
+     * </pre>
+     *
+     * <code>.tensorflow.RewriterConfig.Toggle constant_folding = 3;</code>
+     */
+    public Builder setConstantFoldingValue(int value) {
+      constantFolding_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Fold constants (default is ON)
+     * Statically infer the value of tensors when possible, and materialize the
+     * result using constants.
+     * </pre>
+     *
+     * <code>.tensorflow.RewriterConfig.Toggle constant_folding = 3;</code>
+     */
+    public org.tensorflow.proto.framework.RewriterConfig.Toggle getConstantFolding() {
+      @SuppressWarnings("deprecation")
+      org.tensorflow.proto.framework.RewriterConfig.Toggle result = org.tensorflow.proto.framework.RewriterConfig.Toggle.valueOf(constantFolding_);
+      return result == null ? org.tensorflow.proto.framework.RewriterConfig.Toggle.UNRECOGNIZED : result;
+    }
+    /**
+     * <pre>
+     * Fold constants (default is ON)
+     * Statically infer the value of tensors when possible, and materialize the
+     * result using constants.
+     * </pre>
+     *
+     * <code>.tensorflow.RewriterConfig.Toggle constant_folding = 3;</code>
+     */
+    public Builder setConstantFolding(org.tensorflow.proto.framework.RewriterConfig.Toggle value) {
+      if (value == null) {
+        throw new NullPointerException();
+      }
+      
+      constantFolding_ = value.getNumber();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Fold constants (default is ON)
+     * Statically infer the value of tensors when possible, and materialize the
+     * result using constants.
+     * </pre>
+     *
+     * <code>.tensorflow.RewriterConfig.Toggle constant_folding = 3;</code>
+     */
+    public Builder clearConstantFolding() {
+      
+      constantFolding_ = 0;
+      onChanged();
+      return this;
+    }
+
+    private int shapeOptimization_ = 0;
+    /**
+     * <pre>
+     * Shape optimizations (default is ON)
+     * Simplify computations made on shapes.
+     * </pre>
+     *
+     * <code>.tensorflow.RewriterConfig.Toggle shape_optimization = 13;</code>
+     */
+    public int getShapeOptimizationValue() {
+      return shapeOptimization_;
+    }
+    /**
+     * <pre>
+     * Shape optimizations (default is ON)
+     * Simplify computations made on shapes.
+     * </pre>
+     *
+     * <code>.tensorflow.RewriterConfig.Toggle shape_optimization = 13;</code>
+     */
+    public Builder setShapeOptimizationValue(int value) {
+      shapeOptimization_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Shape optimizations (default is ON)
+     * Simplify computations made on shapes.
+     * </pre>
+     *
+     * <code>.tensorflow.RewriterConfig.Toggle shape_optimization = 13;</code>
+     */
+    public org.tensorflow.proto.framework.RewriterConfig.Toggle getShapeOptimization() {
+      @SuppressWarnings("deprecation")
+      org.tensorflow.proto.framework.RewriterConfig.Toggle result = org.tensorflow.proto.framework.RewriterConfig.Toggle.valueOf(shapeOptimization_);
+      return result == null ? org.tensorflow.proto.framework.RewriterConfig.Toggle.UNRECOGNIZED : result;
+    }
+    /**
+     * <pre>
+     * Shape optimizations (default is ON)
+     * Simplify computations made on shapes.
+     * </pre>
+     *
+     * <code>.tensorflow.RewriterConfig.Toggle shape_optimization = 13;</code>
+     */
+    public Builder setShapeOptimization(org.tensorflow.proto.framework.RewriterConfig.Toggle value) {
+      if (value == null) {
+        throw new NullPointerException();
+      }
+      
+      shapeOptimization_ = value.getNumber();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Shape optimizations (default is ON)
+     * Simplify computations made on shapes.
+     * </pre>
+     *
+     * <code>.tensorflow.RewriterConfig.Toggle shape_optimization = 13;</code>
+     */
+    public Builder clearShapeOptimization() {
+      
+      shapeOptimization_ = 0;
+      onChanged();
+      return this;
+    }
+
+    private int remapping_ = 0;
+    /**
+     * <pre>
+     * Remapping (default is ON)
+     * Remap subgraphs onto more efficient implementations.
+     * </pre>
+     *
+     * <code>.tensorflow.RewriterConfig.Toggle remapping = 14;</code>
+     */
+    public int getRemappingValue() {
+      return remapping_;
+    }
+    /**
+     * <pre>
+     * Remapping (default is ON)
+     * Remap subgraphs onto more efficient implementations.
+     * </pre>
+     *
+     * <code>.tensorflow.RewriterConfig.Toggle remapping = 14;</code>
+     */
+    public Builder setRemappingValue(int value) {
+      remapping_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Remapping (default is ON)
+     * Remap subgraphs onto more efficient implementations.
+     * </pre>
+     *
+     * <code>.tensorflow.RewriterConfig.Toggle remapping = 14;</code>
+     */
+    public org.tensorflow.proto.framework.RewriterConfig.Toggle getRemapping() {
+      @SuppressWarnings("deprecation")
+      org.tensorflow.proto.framework.RewriterConfig.Toggle result = org.tensorflow.proto.framework.RewriterConfig.Toggle.valueOf(remapping_);
+      return result == null ? org.tensorflow.proto.framework.RewriterConfig.Toggle.UNRECOGNIZED : result;
+    }
+    /**
+     * <pre>
+     * Remapping (default is ON)
+     * Remap subgraphs onto more efficient implementations.
+     * </pre>
+     *
+     * <code>.tensorflow.RewriterConfig.Toggle remapping = 14;</code>
+     */
+    public Builder setRemapping(org.tensorflow.proto.framework.RewriterConfig.Toggle value) {
+      if (value == null) {
+        throw new NullPointerException();
+      }
+      
+      remapping_ = value.getNumber();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Remapping (default is ON)
+     * Remap subgraphs onto more efficient implementations.
+     * </pre>
+     *
+     * <code>.tensorflow.RewriterConfig.Toggle remapping = 14;</code>
+     */
+    public Builder clearRemapping() {
+      
+      remapping_ = 0;
+      onChanged();
+      return this;
+    }
+
+    private int commonSubgraphElimination_ = 0;
+    /**
+     * <pre>
+     * Common subgraph elimination (default is ON)
+     * e.g. Simplify arithmetic ops; merge ops with same value (like constants).
+     * </pre>
+     *
+     * <code>.tensorflow.RewriterConfig.Toggle common_subgraph_elimination = 24;</code>
+     */
+    public int getCommonSubgraphEliminationValue() {
+      return commonSubgraphElimination_;
+    }
+    /**
+     * <pre>
+     * Common subgraph elimination (default is ON)
+     * e.g. Simplify arithmetic ops; merge ops with same value (like constants).
+     * </pre>
+     *
+     * <code>.tensorflow.RewriterConfig.Toggle common_subgraph_elimination = 24;</code>
+     */
+    public Builder setCommonSubgraphEliminationValue(int value) {
+      commonSubgraphElimination_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Common subgraph elimination (default is ON)
+     * e.g. Simplify arithmetic ops; merge ops with same value (like constants).
+     * </pre>
+     *
+     * <code>.tensorflow.RewriterConfig.Toggle common_subgraph_elimination = 24;</code>
+     */
+    public org.tensorflow.proto.framework.RewriterConfig.Toggle getCommonSubgraphElimination() {
+      @SuppressWarnings("deprecation")
+      org.tensorflow.proto.framework.RewriterConfig.Toggle result = org.tensorflow.proto.framework.RewriterConfig.Toggle.valueOf(commonSubgraphElimination_);
+      return result == null ? org.tensorflow.proto.framework.RewriterConfig.Toggle.UNRECOGNIZED : result;
+    }
+    /**
+     * <pre>
+     * Common subgraph elimination (default is ON)
+     * e.g. Simplify arithmetic ops; merge ops with same value (like constants).
+     * </pre>
+     *
+     * <code>.tensorflow.RewriterConfig.Toggle common_subgraph_elimination = 24;</code>
+     */
+    public Builder setCommonSubgraphElimination(org.tensorflow.proto.framework.RewriterConfig.Toggle value) {
+      if (value == null) {
+        throw new NullPointerException();
+      }
+      
+      commonSubgraphElimination_ = value.getNumber();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Common subgraph elimination (default is ON)
+     * e.g. Simplify arithmetic ops; merge ops with same value (like constants).
+     * </pre>
+     *
+     * <code>.tensorflow.RewriterConfig.Toggle common_subgraph_elimination = 24;</code>
+     */
+    public Builder clearCommonSubgraphElimination() {
+      
+      commonSubgraphElimination_ = 0;
+      onChanged();
+      return this;
+    }
+
+    private int arithmeticOptimization_ = 0;
+    /**
+     * <pre>
+     * Arithmetic optimizations (default is ON)
+     * e.g. Simplify arithmetic ops; merge ops with same value (like constants).
+     * </pre>
+     *
+     * <code>.tensorflow.RewriterConfig.Toggle arithmetic_optimization = 7;</code>
+     */
+    public int getArithmeticOptimizationValue() {
+      return arithmeticOptimization_;
+    }
+    /**
+     * <pre>
+     * Arithmetic optimizations (default is ON)
+     * e.g. Simplify arithmetic ops; merge ops with same value (like constants).
+     * </pre>
+     *
+     * <code>.tensorflow.RewriterConfig.Toggle arithmetic_optimization = 7;</code>
+     */
+    public Builder setArithmeticOptimizationValue(int value) {
+      arithmeticOptimization_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Arithmetic optimizations (default is ON)
+     * e.g. Simplify arithmetic ops; merge ops with same value (like constants).
+     * </pre>
+     *
+     * <code>.tensorflow.RewriterConfig.Toggle arithmetic_optimization = 7;</code>
+     */
+    public org.tensorflow.proto.framework.RewriterConfig.Toggle getArithmeticOptimization() {
+      @SuppressWarnings("deprecation")
+      org.tensorflow.proto.framework.RewriterConfig.Toggle result = org.tensorflow.proto.framework.RewriterConfig.Toggle.valueOf(arithmeticOptimization_);
+      return result == null ? org.tensorflow.proto.framework.RewriterConfig.Toggle.UNRECOGNIZED : result;
+    }
+    /**
+     * <pre>
+     * Arithmetic optimizations (default is ON)
+     * e.g. Simplify arithmetic ops; merge ops with same value (like constants).
+     * </pre>
+     *
+     * <code>.tensorflow.RewriterConfig.Toggle arithmetic_optimization = 7;</code>
+     */
+    public Builder setArithmeticOptimization(org.tensorflow.proto.framework.RewriterConfig.Toggle value) {
+      if (value == null) {
+        throw new NullPointerException();
+      }
+      
+      arithmeticOptimization_ = value.getNumber();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Arithmetic optimizations (default is ON)
+     * e.g. Simplify arithmetic ops; merge ops with same value (like constants).
+     * </pre>
+     *
+     * <code>.tensorflow.RewriterConfig.Toggle arithmetic_optimization = 7;</code>
+     */
+    public Builder clearArithmeticOptimization() {
+      
+      arithmeticOptimization_ = 0;
+      onChanged();
+      return this;
+    }
+
+    private int dependencyOptimization_ = 0;
+    /**
+     * <pre>
+     * Control dependency optimizations (default is ON).
+     * Remove redundant control dependencies, which may enable other optimization.
+     * </pre>
+     *
+     * <code>.tensorflow.RewriterConfig.Toggle dependency_optimization = 8;</code>
+     */
+    public int getDependencyOptimizationValue() {
+      return dependencyOptimization_;
+    }
+    /**
+     * <pre>
+     * Control dependency optimizations (default is ON).
+     * Remove redundant control dependencies, which may enable other optimization.
+     * </pre>
+     *
+     * <code>.tensorflow.RewriterConfig.Toggle dependency_optimization = 8;</code>
+     */
+    public Builder setDependencyOptimizationValue(int value) {
+      dependencyOptimization_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Control dependency optimizations (default is ON).
+     * Remove redundant control dependencies, which may enable other optimization.
+     * </pre>
+     *
+     * <code>.tensorflow.RewriterConfig.Toggle dependency_optimization = 8;</code>
+     */
+    public org.tensorflow.proto.framework.RewriterConfig.Toggle getDependencyOptimization() {
+      @SuppressWarnings("deprecation")
+      org.tensorflow.proto.framework.RewriterConfig.Toggle result = org.tensorflow.proto.framework.RewriterConfig.Toggle.valueOf(dependencyOptimization_);
+      return result == null ? org.tensorflow.proto.framework.RewriterConfig.Toggle.UNRECOGNIZED : result;
+    }
+    /**
+     * <pre>
+     * Control dependency optimizations (default is ON).
+     * Remove redundant control dependencies, which may enable other optimization.
+     * </pre>
+     *
+     * <code>.tensorflow.RewriterConfig.Toggle dependency_optimization = 8;</code>
+     */
+    public Builder setDependencyOptimization(org.tensorflow.proto.framework.RewriterConfig.Toggle value) {
+      if (value == null) {
+        throw new NullPointerException();
+      }
+      
+      dependencyOptimization_ = value.getNumber();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Control dependency optimizations (default is ON).
+     * Remove redundant control dependencies, which may enable other optimization.
+     * </pre>
+     *
+     * <code>.tensorflow.RewriterConfig.Toggle dependency_optimization = 8;</code>
+     */
+    public Builder clearDependencyOptimization() {
+      
+      dependencyOptimization_ = 0;
+      onChanged();
+      return this;
+    }
+
+    private int loopOptimization_ = 0;
+    /**
+     * <pre>
+     * Loop optimizations (default is ON).
+     * </pre>
+     *
+     * <code>.tensorflow.RewriterConfig.Toggle loop_optimization = 9;</code>
+     */
+    public int getLoopOptimizationValue() {
+      return loopOptimization_;
+    }
+    /**
+     * <pre>
+     * Loop optimizations (default is ON).
+     * </pre>
+     *
+     * <code>.tensorflow.RewriterConfig.Toggle loop_optimization = 9;</code>
+     */
+    public Builder setLoopOptimizationValue(int value) {
+      loopOptimization_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Loop optimizations (default is ON).
+     * </pre>
+     *
+     * <code>.tensorflow.RewriterConfig.Toggle loop_optimization = 9;</code>
+     */
+    public org.tensorflow.proto.framework.RewriterConfig.Toggle getLoopOptimization() {
+      @SuppressWarnings("deprecation")
+      org.tensorflow.proto.framework.RewriterConfig.Toggle result = org.tensorflow.proto.framework.RewriterConfig.Toggle.valueOf(loopOptimization_);
+      return result == null ? org.tensorflow.proto.framework.RewriterConfig.Toggle.UNRECOGNIZED : result;
+    }
+    /**
+     * <pre>
+     * Loop optimizations (default is ON).
+     * </pre>
+     *
+     * <code>.tensorflow.RewriterConfig.Toggle loop_optimization = 9;</code>
+     */
+    public Builder setLoopOptimization(org.tensorflow.proto.framework.RewriterConfig.Toggle value) {
+      if (value == null) {
+        throw new NullPointerException();
+      }
+      
+      loopOptimization_ = value.getNumber();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Loop optimizations (default is ON).
+     * </pre>
+     *
+     * <code>.tensorflow.RewriterConfig.Toggle loop_optimization = 9;</code>
+     */
+    public Builder clearLoopOptimization() {
+      
+      loopOptimization_ = 0;
+      onChanged();
+      return this;
+    }
+
+    private int functionOptimization_ = 0;
+    /**
+     * <pre>
+     * Function optimizations (default is ON).
+     * </pre>
+     *
+     * <code>.tensorflow.RewriterConfig.Toggle function_optimization = 10;</code>
+     */
+    public int getFunctionOptimizationValue() {
+      return functionOptimization_;
+    }
+    /**
+     * <pre>
+     * Function optimizations (default is ON).
+     * </pre>
+     *
+     * <code>.tensorflow.RewriterConfig.Toggle function_optimization = 10;</code>
+     */
+    public Builder setFunctionOptimizationValue(int value) {
+      functionOptimization_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Function optimizations (default is ON).
+     * </pre>
+     *
+     * <code>.tensorflow.RewriterConfig.Toggle function_optimization = 10;</code>
+     */
+    public org.tensorflow.proto.framework.RewriterConfig.Toggle getFunctionOptimization() {
+      @SuppressWarnings("deprecation")
+      org.tensorflow.proto.framework.RewriterConfig.Toggle result = org.tensorflow.proto.framework.RewriterConfig.Toggle.valueOf(functionOptimization_);
+      return result == null ? org.tensorflow.proto.framework.RewriterConfig.Toggle.UNRECOGNIZED : result;
+    }
+    /**
+     * <pre>
+     * Function optimizations (default is ON).
+     * </pre>
+     *
+     * <code>.tensorflow.RewriterConfig.Toggle function_optimization = 10;</code>
+     */
+    public Builder setFunctionOptimization(org.tensorflow.proto.framework.RewriterConfig.Toggle value) {
+      if (value == null) {
+        throw new NullPointerException();
+      }
+      
+      functionOptimization_ = value.getNumber();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Function optimizations (default is ON).
+     * </pre>
+     *
+     * <code>.tensorflow.RewriterConfig.Toggle function_optimization = 10;</code>
+     */
+    public Builder clearFunctionOptimization() {
+      
+      functionOptimization_ = 0;
+      onChanged();
+      return this;
+    }
+
+    private int debugStripper_ = 0;
+    /**
+     * <pre>
+     * Strips debug-related nodes from the graph (off by default).
+     * </pre>
+     *
+     * <code>.tensorflow.RewriterConfig.Toggle debug_stripper = 11;</code>
+     */
+    public int getDebugStripperValue() {
+      return debugStripper_;
+    }
+    /**
+     * <pre>
+     * Strips debug-related nodes from the graph (off by default).
+     * </pre>
+     *
+     * <code>.tensorflow.RewriterConfig.Toggle debug_stripper = 11;</code>
+     */
+    public Builder setDebugStripperValue(int value) {
+      debugStripper_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Strips debug-related nodes from the graph (off by default).
+     * </pre>
+     *
+     * <code>.tensorflow.RewriterConfig.Toggle debug_stripper = 11;</code>
+     */
+    public org.tensorflow.proto.framework.RewriterConfig.Toggle getDebugStripper() {
+      @SuppressWarnings("deprecation")
+      org.tensorflow.proto.framework.RewriterConfig.Toggle result = org.tensorflow.proto.framework.RewriterConfig.Toggle.valueOf(debugStripper_);
+      return result == null ? org.tensorflow.proto.framework.RewriterConfig.Toggle.UNRECOGNIZED : result;
+    }
+    /**
+     * <pre>
+     * Strips debug-related nodes from the graph (off by default).
+     * </pre>
+     *
+     * <code>.tensorflow.RewriterConfig.Toggle debug_stripper = 11;</code>
+     */
+    public Builder setDebugStripper(org.tensorflow.proto.framework.RewriterConfig.Toggle value) {
+      if (value == null) {
+        throw new NullPointerException();
+      }
+      
+      debugStripper_ = value.getNumber();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Strips debug-related nodes from the graph (off by default).
+     * </pre>
+     *
+     * <code>.tensorflow.RewriterConfig.Toggle debug_stripper = 11;</code>
+     */
+    public Builder clearDebugStripper() {
+      
+      debugStripper_ = 0;
+      onChanged();
+      return this;
+    }
+
+    private boolean disableModelPruning_ ;
+    /**
+     * <pre>
+     * If true, don't remove unnecessary ops from the graph
+     * </pre>
+     *
+     * <code>bool disable_model_pruning = 2;</code>
+     */
+    public boolean getDisableModelPruning() {
+      return disableModelPruning_;
+    }
+    /**
+     * <pre>
+     * If true, don't remove unnecessary ops from the graph
+     * </pre>
+     *
+     * <code>bool disable_model_pruning = 2;</code>
+     */
+    public Builder setDisableModelPruning(boolean value) {
+      
+      disableModelPruning_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * If true, don't remove unnecessary ops from the graph
+     * </pre>
+     *
+     * <code>bool disable_model_pruning = 2;</code>
+     */
+    public Builder clearDisableModelPruning() {
+      
+      disableModelPruning_ = false;
+      onChanged();
+      return this;
+    }
+
+    private int scopedAllocatorOptimization_ = 0;
+    /**
+     * <pre>
+     * Try to allocate some independent Op outputs contiguously in order to
+     * merge or eliminate downstream Ops (off by default).
+     * </pre>
+     *
+     * <code>.tensorflow.RewriterConfig.Toggle scoped_allocator_optimization = 15;</code>
+     */
+    public int getScopedAllocatorOptimizationValue() {
+      return scopedAllocatorOptimization_;
+    }
+    /**
+     * <pre>
+     * Try to allocate some independent Op outputs contiguously in order to
+     * merge or eliminate downstream Ops (off by default).
+     * </pre>
+     *
+     * <code>.tensorflow.RewriterConfig.Toggle scoped_allocator_optimization = 15;</code>
+     */
+    public Builder setScopedAllocatorOptimizationValue(int value) {
+      scopedAllocatorOptimization_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Try to allocate some independent Op outputs contiguously in order to
+     * merge or eliminate downstream Ops (off by default).
+     * </pre>
+     *
+     * <code>.tensorflow.RewriterConfig.Toggle scoped_allocator_optimization = 15;</code>
+     */
+    public org.tensorflow.proto.framework.RewriterConfig.Toggle getScopedAllocatorOptimization() {
+      @SuppressWarnings("deprecation")
+      org.tensorflow.proto.framework.RewriterConfig.Toggle result = org.tensorflow.proto.framework.RewriterConfig.Toggle.valueOf(scopedAllocatorOptimization_);
+      return result == null ? org.tensorflow.proto.framework.RewriterConfig.Toggle.UNRECOGNIZED : result;
+    }
+    /**
+     * <pre>
+     * Try to allocate some independent Op outputs contiguously in order to
+     * merge or eliminate downstream Ops (off by default).
+     * </pre>
+     *
+     * <code>.tensorflow.RewriterConfig.Toggle scoped_allocator_optimization = 15;</code>
+     */
+    public Builder setScopedAllocatorOptimization(org.tensorflow.proto.framework.RewriterConfig.Toggle value) {
+      if (value == null) {
+        throw new NullPointerException();
+      }
+      
+      scopedAllocatorOptimization_ = value.getNumber();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Try to allocate some independent Op outputs contiguously in order to
+     * merge or eliminate downstream Ops (off by default).
+     * </pre>
+     *
+     * <code>.tensorflow.RewriterConfig.Toggle scoped_allocator_optimization = 15;</code>
+     */
+    public Builder clearScopedAllocatorOptimization() {
+      
+      scopedAllocatorOptimization_ = 0;
+      onChanged();
+      return this;
+    }
+
+    private int pinToHostOptimization_ = 0;
+    /**
+     * <pre>
+     * Force small ops onto the CPU (default is OFF).
+     * </pre>
+     *
+     * <code>.tensorflow.RewriterConfig.Toggle pin_to_host_optimization = 18;</code>
+     */
+    public int getPinToHostOptimizationValue() {
+      return pinToHostOptimization_;
+    }
+    /**
+     * <pre>
+     * Force small ops onto the CPU (default is OFF).
+     * </pre>
+     *
+     * <code>.tensorflow.RewriterConfig.Toggle pin_to_host_optimization = 18;</code>
+     */
+    public Builder setPinToHostOptimizationValue(int value) {
+      pinToHostOptimization_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Force small ops onto the CPU (default is OFF).
+     * </pre>
+     *
+     * <code>.tensorflow.RewriterConfig.Toggle pin_to_host_optimization = 18;</code>
+     */
+    public org.tensorflow.proto.framework.RewriterConfig.Toggle getPinToHostOptimization() {
+      @SuppressWarnings("deprecation")
+      org.tensorflow.proto.framework.RewriterConfig.Toggle result = org.tensorflow.proto.framework.RewriterConfig.Toggle.valueOf(pinToHostOptimization_);
+      return result == null ? org.tensorflow.proto.framework.RewriterConfig.Toggle.UNRECOGNIZED : result;
+    }
+    /**
+     * <pre>
+     * Force small ops onto the CPU (default is OFF).
+     * </pre>
+     *
+     * <code>.tensorflow.RewriterConfig.Toggle pin_to_host_optimization = 18;</code>
+     */
+    public Builder setPinToHostOptimization(org.tensorflow.proto.framework.RewriterConfig.Toggle value) {
+      if (value == null) {
+        throw new NullPointerException();
+      }
+      
+      pinToHostOptimization_ = value.getNumber();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Force small ops onto the CPU (default is OFF).
+     * </pre>
+     *
+     * <code>.tensorflow.RewriterConfig.Toggle pin_to_host_optimization = 18;</code>
+     */
+    public Builder clearPinToHostOptimization() {
+      
+      pinToHostOptimization_ = 0;
+      onChanged();
+      return this;
+    }
+
+    private int implementationSelector_ = 0;
+    /**
+     * <pre>
+     * Enable the swap of kernel implementations based on the device placement
+     * (default is ON).
+     * </pre>
+     *
+     * <code>.tensorflow.RewriterConfig.Toggle implementation_selector = 22;</code>
+     */
+    public int getImplementationSelectorValue() {
+      return implementationSelector_;
+    }
+    /**
+     * <pre>
+     * Enable the swap of kernel implementations based on the device placement
+     * (default is ON).
+     * </pre>
+     *
+     * <code>.tensorflow.RewriterConfig.Toggle implementation_selector = 22;</code>
+     */
+    public Builder setImplementationSelectorValue(int value) {
+      implementationSelector_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Enable the swap of kernel implementations based on the device placement
+     * (default is ON).
+     * </pre>
+     *
+     * <code>.tensorflow.RewriterConfig.Toggle implementation_selector = 22;</code>
+     */
+    public org.tensorflow.proto.framework.RewriterConfig.Toggle getImplementationSelector() {
+      @SuppressWarnings("deprecation")
+      org.tensorflow.proto.framework.RewriterConfig.Toggle result = org.tensorflow.proto.framework.RewriterConfig.Toggle.valueOf(implementationSelector_);
+      return result == null ? org.tensorflow.proto.framework.RewriterConfig.Toggle.UNRECOGNIZED : result;
+    }
+    /**
+     * <pre>
+     * Enable the swap of kernel implementations based on the device placement
+     * (default is ON).
+     * </pre>
+     *
+     * <code>.tensorflow.RewriterConfig.Toggle implementation_selector = 22;</code>
+     */
+    public Builder setImplementationSelector(org.tensorflow.proto.framework.RewriterConfig.Toggle value) {
+      if (value == null) {
+        throw new NullPointerException();
+      }
+      
+      implementationSelector_ = value.getNumber();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Enable the swap of kernel implementations based on the device placement
+     * (default is ON).
+     * </pre>
+     *
+     * <code>.tensorflow.RewriterConfig.Toggle implementation_selector = 22;</code>
+     */
+    public Builder clearImplementationSelector() {
+      
+      implementationSelector_ = 0;
+      onChanged();
+      return this;
+    }
+
+    private int autoMixedPrecision_ = 0;
+    /**
+     * <pre>
+     * Optimize data types for CUDA (default is OFF).
+     * This will try to use float16 on GPU which is faster.
+     * Note that this can change the numerical stability of the graph and may
+     * require the use of loss scaling to maintain model convergence.
+     * </pre>
+     *
+     * <code>.tensorflow.RewriterConfig.Toggle auto_mixed_precision = 23;</code>
+     */
+    public int getAutoMixedPrecisionValue() {
+      return autoMixedPrecision_;
+    }
+    /**
+     * <pre>
+     * Optimize data types for CUDA (default is OFF).
+     * This will try to use float16 on GPU which is faster.
+     * Note that this can change the numerical stability of the graph and may
+     * require the use of loss scaling to maintain model convergence.
+     * </pre>
+     *
+     * <code>.tensorflow.RewriterConfig.Toggle auto_mixed_precision = 23;</code>
+     */
+    public Builder setAutoMixedPrecisionValue(int value) {
+      autoMixedPrecision_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Optimize data types for CUDA (default is OFF).
+     * This will try to use float16 on GPU which is faster.
+     * Note that this can change the numerical stability of the graph and may
+     * require the use of loss scaling to maintain model convergence.
+     * </pre>
+     *
+     * <code>.tensorflow.RewriterConfig.Toggle auto_mixed_precision = 23;</code>
+     */
+    public org.tensorflow.proto.framework.RewriterConfig.Toggle getAutoMixedPrecision() {
+      @SuppressWarnings("deprecation")
+      org.tensorflow.proto.framework.RewriterConfig.Toggle result = org.tensorflow.proto.framework.RewriterConfig.Toggle.valueOf(autoMixedPrecision_);
+      return result == null ? org.tensorflow.proto.framework.RewriterConfig.Toggle.UNRECOGNIZED : result;
+    }
+    /**
+     * <pre>
+     * Optimize data types for CUDA (default is OFF).
+     * This will try to use float16 on GPU which is faster.
+     * Note that this can change the numerical stability of the graph and may
+     * require the use of loss scaling to maintain model convergence.
+     * </pre>
+     *
+     * <code>.tensorflow.RewriterConfig.Toggle auto_mixed_precision = 23;</code>
+     */
+    public Builder setAutoMixedPrecision(org.tensorflow.proto.framework.RewriterConfig.Toggle value) {
+      if (value == null) {
+        throw new NullPointerException();
+      }
+      
+      autoMixedPrecision_ = value.getNumber();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Optimize data types for CUDA (default is OFF).
+     * This will try to use float16 on GPU which is faster.
+     * Note that this can change the numerical stability of the graph and may
+     * require the use of loss scaling to maintain model convergence.
+     * </pre>
+     *
+     * <code>.tensorflow.RewriterConfig.Toggle auto_mixed_precision = 23;</code>
+     */
+    public Builder clearAutoMixedPrecision() {
+      
+      autoMixedPrecision_ = 0;
+      onChanged();
+      return this;
+    }
+
+    private int autoMixedPrecisionMkl_ = 0;
+    /**
+     * <pre>
+     * Optimize data types for MKL (default is OFF).
+     * This will try to use bfloat16 on CPUs, which is faster.
+     * Note that this can change the numerical stability of the graph.
+     * </pre>
+     *
+     * <code>.tensorflow.RewriterConfig.Toggle auto_mixed_precision_mkl = 25;</code>
+     */
+    public int getAutoMixedPrecisionMklValue() {
+      return autoMixedPrecisionMkl_;
+    }
+    /**
+     * <pre>
+     * Optimize data types for MKL (default is OFF).
+     * This will try to use bfloat16 on CPUs, which is faster.
+     * Note that this can change the numerical stability of the graph.
+     * </pre>
+     *
+     * <code>.tensorflow.RewriterConfig.Toggle auto_mixed_precision_mkl = 25;</code>
+     */
+    public Builder setAutoMixedPrecisionMklValue(int value) {
+      autoMixedPrecisionMkl_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Optimize data types for MKL (default is OFF).
+     * This will try to use bfloat16 on CPUs, which is faster.
+     * Note that this can change the numerical stability of the graph.
+     * </pre>
+     *
+     * <code>.tensorflow.RewriterConfig.Toggle auto_mixed_precision_mkl = 25;</code>
+     */
+    public org.tensorflow.proto.framework.RewriterConfig.Toggle getAutoMixedPrecisionMkl() {
+      @SuppressWarnings("deprecation")
+      org.tensorflow.proto.framework.RewriterConfig.Toggle result = org.tensorflow.proto.framework.RewriterConfig.Toggle.valueOf(autoMixedPrecisionMkl_);
+      return result == null ? org.tensorflow.proto.framework.RewriterConfig.Toggle.UNRECOGNIZED : result;
+    }
+    /**
+     * <pre>
+     * Optimize data types for MKL (default is OFF).
+     * This will try to use bfloat16 on CPUs, which is faster.
+     * Note that this can change the numerical stability of the graph.
+     * </pre>
+     *
+     * <code>.tensorflow.RewriterConfig.Toggle auto_mixed_precision_mkl = 25;</code>
+     */
+    public Builder setAutoMixedPrecisionMkl(org.tensorflow.proto.framework.RewriterConfig.Toggle value) {
+      if (value == null) {
+        throw new NullPointerException();
+      }
+      
+      autoMixedPrecisionMkl_ = value.getNumber();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Optimize data types for MKL (default is OFF).
+     * This will try to use bfloat16 on CPUs, which is faster.
+     * Note that this can change the numerical stability of the graph.
+     * </pre>
+     *
+     * <code>.tensorflow.RewriterConfig.Toggle auto_mixed_precision_mkl = 25;</code>
+     */
+    public Builder clearAutoMixedPrecisionMkl() {
+      
+      autoMixedPrecisionMkl_ = 0;
+      onChanged();
+      return this;
+    }
+
+    private boolean disableMetaOptimizer_ ;
+    /**
+     * <pre>
+     * Disable the entire meta optimizer (off by default).
+     * </pre>
+     *
+     * <code>bool disable_meta_optimizer = 19;</code>
+     */
+    public boolean getDisableMetaOptimizer() {
+      return disableMetaOptimizer_;
+    }
+    /**
+     * <pre>
+     * Disable the entire meta optimizer (off by default).
+     * </pre>
+     *
+     * <code>bool disable_meta_optimizer = 19;</code>
+     */
+    public Builder setDisableMetaOptimizer(boolean value) {
+      
+      disableMetaOptimizer_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Disable the entire meta optimizer (off by default).
+     * </pre>
+     *
+     * <code>bool disable_meta_optimizer = 19;</code>
+     */
+    public Builder clearDisableMetaOptimizer() {
+      
+      disableMetaOptimizer_ = false;
+      onChanged();
+      return this;
+    }
+
+    private int metaOptimizerIterations_ = 0;
+    /**
+     * <pre>
+     * Controls how many times we run the optimizers in meta optimizer (default
+     * is once).
+     * </pre>
+     *
+     * <code>.tensorflow.RewriterConfig.NumIterationsType meta_optimizer_iterations = 12;</code>
+     */
+    public int getMetaOptimizerIterationsValue() {
+      return metaOptimizerIterations_;
+    }
+    /**
+     * <pre>
+     * Controls how many times we run the optimizers in meta optimizer (default
+     * is once).
+     * </pre>
+     *
+     * <code>.tensorflow.RewriterConfig.NumIterationsType meta_optimizer_iterations = 12;</code>
+     */
+    public Builder setMetaOptimizerIterationsValue(int value) {
+      metaOptimizerIterations_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Controls how many times we run the optimizers in meta optimizer (default
+     * is once).
+     * </pre>
+     *
+     * <code>.tensorflow.RewriterConfig.NumIterationsType meta_optimizer_iterations = 12;</code>
+     */
+    public org.tensorflow.proto.framework.RewriterConfig.NumIterationsType getMetaOptimizerIterations() {
+      @SuppressWarnings("deprecation")
+      org.tensorflow.proto.framework.RewriterConfig.NumIterationsType result = org.tensorflow.proto.framework.RewriterConfig.NumIterationsType.valueOf(metaOptimizerIterations_);
+      return result == null ? org.tensorflow.proto.framework.RewriterConfig.NumIterationsType.UNRECOGNIZED : result;
+    }
+    /**
+     * <pre>
+     * Controls how many times we run the optimizers in meta optimizer (default
+     * is once).
+     * </pre>
+     *
+     * <code>.tensorflow.RewriterConfig.NumIterationsType meta_optimizer_iterations = 12;</code>
+     */
+    public Builder setMetaOptimizerIterations(org.tensorflow.proto.framework.RewriterConfig.NumIterationsType value) {
+      if (value == null) {
+        throw new NullPointerException();
+      }
+      
+      metaOptimizerIterations_ = value.getNumber();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Controls how many times we run the optimizers in meta optimizer (default
+     * is once).
+     * </pre>
+     *
+     * <code>.tensorflow.RewriterConfig.NumIterationsType meta_optimizer_iterations = 12;</code>
+     */
+    public Builder clearMetaOptimizerIterations() {
+      
+      metaOptimizerIterations_ = 0;
+      onChanged();
+      return this;
+    }
+
+    private int minGraphNodes_ ;
+    /**
+     * <pre>
+     * The minimum number of nodes in a graph to optimizer. For smaller graphs,
+     * optimization is skipped.
+     * 0 means the system picks an appropriate number.
+     * &lt; 0 means do not skip optimization.
+     * </pre>
+     *
+     * <code>int32 min_graph_nodes = 17;</code>
+     */
+    public int getMinGraphNodes() {
+      return minGraphNodes_;
+    }
+    /**
+     * <pre>
+     * The minimum number of nodes in a graph to optimizer. For smaller graphs,
+     * optimization is skipped.
+     * 0 means the system picks an appropriate number.
+     * &lt; 0 means do not skip optimization.
+     * </pre>
+     *
+     * <code>int32 min_graph_nodes = 17;</code>
+     */
+    public Builder setMinGraphNodes(int value) {
+      
+      minGraphNodes_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * The minimum number of nodes in a graph to optimizer. For smaller graphs,
+     * optimization is skipped.
+     * 0 means the system picks an appropriate number.
+     * &lt; 0 means do not skip optimization.
+     * </pre>
+     *
+     * <code>int32 min_graph_nodes = 17;</code>
+     */
+    public Builder clearMinGraphNodes() {
+      
+      minGraphNodes_ = 0;
+      onChanged();
+      return this;
+    }
+
+    private int memoryOptimization_ = 0;
+    /**
+     * <pre>
+     * Configures memory optimization passes through the meta-optimizer. Has no
+     * effect on manually requested memory optimization passes in the optimizers
+     * field.
+     * </pre>
+     *
+     * <code>.tensorflow.RewriterConfig.MemOptType memory_optimization = 4;</code>
+     */
+    public int getMemoryOptimizationValue() {
+      return memoryOptimization_;
+    }
+    /**
+     * <pre>
+     * Configures memory optimization passes through the meta-optimizer. Has no
+     * effect on manually requested memory optimization passes in the optimizers
+     * field.
+     * </pre>
+     *
+     * <code>.tensorflow.RewriterConfig.MemOptType memory_optimization = 4;</code>
+     */
+    public Builder setMemoryOptimizationValue(int value) {
+      memoryOptimization_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Configures memory optimization passes through the meta-optimizer. Has no
+     * effect on manually requested memory optimization passes in the optimizers
+     * field.
+     * </pre>
+     *
+     * <code>.tensorflow.RewriterConfig.MemOptType memory_optimization = 4;</code>
+     */
+    public org.tensorflow.proto.framework.RewriterConfig.MemOptType getMemoryOptimization() {
+      @SuppressWarnings("deprecation")
+      org.tensorflow.proto.framework.RewriterConfig.MemOptType result = org.tensorflow.proto.framework.RewriterConfig.MemOptType.valueOf(memoryOptimization_);
+      return result == null ? org.tensorflow.proto.framework.RewriterConfig.MemOptType.UNRECOGNIZED : result;
+    }
+    /**
+     * <pre>
+     * Configures memory optimization passes through the meta-optimizer. Has no
+     * effect on manually requested memory optimization passes in the optimizers
+     * field.
+     * </pre>
+     *
+     * <code>.tensorflow.RewriterConfig.MemOptType memory_optimization = 4;</code>
+     */
+    public Builder setMemoryOptimization(org.tensorflow.proto.framework.RewriterConfig.MemOptType value) {
+      if (value == null) {
+        throw new NullPointerException();
+      }
+      
+      memoryOptimization_ = value.getNumber();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Configures memory optimization passes through the meta-optimizer. Has no
+     * effect on manually requested memory optimization passes in the optimizers
+     * field.
+     * </pre>
+     *
+     * <code>.tensorflow.RewriterConfig.MemOptType memory_optimization = 4;</code>
+     */
+    public Builder clearMemoryOptimization() {
+      
+      memoryOptimization_ = 0;
+      onChanged();
+      return this;
+    }
+
+    private java.lang.Object memoryOptimizerTargetNodeNameScope_ = "";
+    /**
+     * <pre>
+     * A node name scope for node names which are valid outputs of recomputations.
+     * Inputs to nodes that match this scope may be recomputed (subject either to
+     * manual annotation of those input nodes or to manual annotation and
+     * heuristics depending on memory_optimization), but the nodes themselves will
+     * not be recomputed. This matches any sub-scopes as well, meaning the scope
+     * can appear not just as a top-level scope. For example, if the value is
+     * "gradients/", the default, it will match node name "gradients/foo",
+     * "foo/gradients/bar", but not "foo_gradients/"
+     * </pre>
+     *
+     * <code>string memory_optimizer_target_node_name_scope = 6;</code>
+     */
+    public java.lang.String getMemoryOptimizerTargetNodeNameScope() {
+      java.lang.Object ref = memoryOptimizerTargetNodeNameScope_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        memoryOptimizerTargetNodeNameScope_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * A node name scope for node names which are valid outputs of recomputations.
+     * Inputs to nodes that match this scope may be recomputed (subject either to
+     * manual annotation of those input nodes or to manual annotation and
+     * heuristics depending on memory_optimization), but the nodes themselves will
+     * not be recomputed. This matches any sub-scopes as well, meaning the scope
+     * can appear not just as a top-level scope. For example, if the value is
+     * "gradients/", the default, it will match node name "gradients/foo",
+     * "foo/gradients/bar", but not "foo_gradients/"
+     * </pre>
+     *
+     * <code>string memory_optimizer_target_node_name_scope = 6;</code>
+     */
+    public com.google.protobuf.ByteString
+        getMemoryOptimizerTargetNodeNameScopeBytes() {
+      java.lang.Object ref = memoryOptimizerTargetNodeNameScope_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        memoryOptimizerTargetNodeNameScope_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * A node name scope for node names which are valid outputs of recomputations.
+     * Inputs to nodes that match this scope may be recomputed (subject either to
+     * manual annotation of those input nodes or to manual annotation and
+     * heuristics depending on memory_optimization), but the nodes themselves will
+     * not be recomputed. This matches any sub-scopes as well, meaning the scope
+     * can appear not just as a top-level scope. For example, if the value is
+     * "gradients/", the default, it will match node name "gradients/foo",
+     * "foo/gradients/bar", but not "foo_gradients/"
+     * </pre>
+     *
+     * <code>string memory_optimizer_target_node_name_scope = 6;</code>
+     */
+    public Builder setMemoryOptimizerTargetNodeNameScope(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      memoryOptimizerTargetNodeNameScope_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * A node name scope for node names which are valid outputs of recomputations.
+     * Inputs to nodes that match this scope may be recomputed (subject either to
+     * manual annotation of those input nodes or to manual annotation and
+     * heuristics depending on memory_optimization), but the nodes themselves will
+     * not be recomputed. This matches any sub-scopes as well, meaning the scope
+     * can appear not just as a top-level scope. For example, if the value is
+     * "gradients/", the default, it will match node name "gradients/foo",
+     * "foo/gradients/bar", but not "foo_gradients/"
+     * </pre>
+     *
+     * <code>string memory_optimizer_target_node_name_scope = 6;</code>
+     */
+    public Builder clearMemoryOptimizerTargetNodeNameScope() {
+      
+      memoryOptimizerTargetNodeNameScope_ = getDefaultInstance().getMemoryOptimizerTargetNodeNameScope();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * A node name scope for node names which are valid outputs of recomputations.
+     * Inputs to nodes that match this scope may be recomputed (subject either to
+     * manual annotation of those input nodes or to manual annotation and
+     * heuristics depending on memory_optimization), but the nodes themselves will
+     * not be recomputed. This matches any sub-scopes as well, meaning the scope
+     * can appear not just as a top-level scope. For example, if the value is
+     * "gradients/", the default, it will match node name "gradients/foo",
+     * "foo/gradients/bar", but not "foo_gradients/"
+     * </pre>
+     *
+     * <code>string memory_optimizer_target_node_name_scope = 6;</code>
+     */
+    public Builder setMemoryOptimizerTargetNodeNameScopeBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      memoryOptimizerTargetNodeNameScope_ = value;
+      onChanged();
+      return this;
+    }
+
+    private long metaOptimizerTimeoutMs_ ;
+    /**
+     * <pre>
+     * Maximum number of milliseconds to spend optimizing a single graph before
+     * timing out. If equal to 0 the system picks a default (currently 5 minutes).
+     * If less than 0 the optimizer will never time out.
+     * </pre>
+     *
+     * <code>int64 meta_optimizer_timeout_ms = 20;</code>
+     */
+    public long getMetaOptimizerTimeoutMs() {
+      return metaOptimizerTimeoutMs_;
+    }
+    /**
+     * <pre>
+     * Maximum number of milliseconds to spend optimizing a single graph before
+     * timing out. If equal to 0 the system picks a default (currently 5 minutes).
+     * If less than 0 the optimizer will never time out.
+     * </pre>
+     *
+     * <code>int64 meta_optimizer_timeout_ms = 20;</code>
+     */
+    public Builder setMetaOptimizerTimeoutMs(long value) {
+      
+      metaOptimizerTimeoutMs_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Maximum number of milliseconds to spend optimizing a single graph before
+     * timing out. If equal to 0 the system picks a default (currently 5 minutes).
+     * If less than 0 the optimizer will never time out.
+     * </pre>
+     *
+     * <code>int64 meta_optimizer_timeout_ms = 20;</code>
+     */
+    public Builder clearMetaOptimizerTimeoutMs() {
+      
+      metaOptimizerTimeoutMs_ = 0L;
+      onChanged();
+      return this;
+    }
+
+    private org.tensorflow.proto.framework.AutoParallelOptions autoParallel_;
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.AutoParallelOptions, org.tensorflow.proto.framework.AutoParallelOptions.Builder, org.tensorflow.proto.framework.AutoParallelOptionsOrBuilder> autoParallelBuilder_;
+    /**
+     * <pre>
+     * Configures AutoParallel optimization passes either through the
+     * meta-optimizer or when manually specified through the optimizers field.
+     * </pre>
+     *
+     * <code>.tensorflow.AutoParallelOptions auto_parallel = 5;</code>
+     */
+    public boolean hasAutoParallel() {
+      return autoParallelBuilder_ != null || autoParallel_ != null;
+    }
+    /**
+     * <pre>
+     * Configures AutoParallel optimization passes either through the
+     * meta-optimizer or when manually specified through the optimizers field.
+     * </pre>
+     *
+     * <code>.tensorflow.AutoParallelOptions auto_parallel = 5;</code>
+     */
+    public org.tensorflow.proto.framework.AutoParallelOptions getAutoParallel() {
+      if (autoParallelBuilder_ == null) {
+        return autoParallel_ == null ? org.tensorflow.proto.framework.AutoParallelOptions.getDefaultInstance() : autoParallel_;
+      } else {
+        return autoParallelBuilder_.getMessage();
+      }
+    }
+    /**
+     * <pre>
+     * Configures AutoParallel optimization passes either through the
+     * meta-optimizer or when manually specified through the optimizers field.
+     * </pre>
+     *
+     * <code>.tensorflow.AutoParallelOptions auto_parallel = 5;</code>
+     */
+    public Builder setAutoParallel(org.tensorflow.proto.framework.AutoParallelOptions value) {
+      if (autoParallelBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        autoParallel_ = value;
+        onChanged();
+      } else {
+        autoParallelBuilder_.setMessage(value);
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Configures AutoParallel optimization passes either through the
+     * meta-optimizer or when manually specified through the optimizers field.
+     * </pre>
+     *
+     * <code>.tensorflow.AutoParallelOptions auto_parallel = 5;</code>
+     */
+    public Builder setAutoParallel(
+        org.tensorflow.proto.framework.AutoParallelOptions.Builder builderForValue) {
+      if (autoParallelBuilder_ == null) {
+        autoParallel_ = builderForValue.build();
+        onChanged();
+      } else {
+        autoParallelBuilder_.setMessage(builderForValue.build());
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Configures AutoParallel optimization passes either through the
+     * meta-optimizer or when manually specified through the optimizers field.
+     * </pre>
+     *
+     * <code>.tensorflow.AutoParallelOptions auto_parallel = 5;</code>
+     */
+    public Builder mergeAutoParallel(org.tensorflow.proto.framework.AutoParallelOptions value) {
+      if (autoParallelBuilder_ == null) {
+        if (autoParallel_ != null) {
+          autoParallel_ =
+            org.tensorflow.proto.framework.AutoParallelOptions.newBuilder(autoParallel_).mergeFrom(value).buildPartial();
+        } else {
+          autoParallel_ = value;
+        }
+        onChanged();
+      } else {
+        autoParallelBuilder_.mergeFrom(value);
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Configures AutoParallel optimization passes either through the
+     * meta-optimizer or when manually specified through the optimizers field.
+     * </pre>
+     *
+     * <code>.tensorflow.AutoParallelOptions auto_parallel = 5;</code>
+     */
+    public Builder clearAutoParallel() {
+      if (autoParallelBuilder_ == null) {
+        autoParallel_ = null;
+        onChanged();
+      } else {
+        autoParallel_ = null;
+        autoParallelBuilder_ = null;
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Configures AutoParallel optimization passes either through the
+     * meta-optimizer or when manually specified through the optimizers field.
+     * </pre>
+     *
+     * <code>.tensorflow.AutoParallelOptions auto_parallel = 5;</code>
+     */
+    public org.tensorflow.proto.framework.AutoParallelOptions.Builder getAutoParallelBuilder() {
+      
+      onChanged();
+      return getAutoParallelFieldBuilder().getBuilder();
+    }
+    /**
+     * <pre>
+     * Configures AutoParallel optimization passes either through the
+     * meta-optimizer or when manually specified through the optimizers field.
+     * </pre>
+     *
+     * <code>.tensorflow.AutoParallelOptions auto_parallel = 5;</code>
+     */
+    public org.tensorflow.proto.framework.AutoParallelOptionsOrBuilder getAutoParallelOrBuilder() {
+      if (autoParallelBuilder_ != null) {
+        return autoParallelBuilder_.getMessageOrBuilder();
+      } else {
+        return autoParallel_ == null ?
+            org.tensorflow.proto.framework.AutoParallelOptions.getDefaultInstance() : autoParallel_;
+      }
+    }
+    /**
+     * <pre>
+     * Configures AutoParallel optimization passes either through the
+     * meta-optimizer or when manually specified through the optimizers field.
+     * </pre>
+     *
+     * <code>.tensorflow.AutoParallelOptions auto_parallel = 5;</code>
+     */
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.AutoParallelOptions, org.tensorflow.proto.framework.AutoParallelOptions.Builder, org.tensorflow.proto.framework.AutoParallelOptionsOrBuilder> 
+        getAutoParallelFieldBuilder() {
+      if (autoParallelBuilder_ == null) {
+        autoParallelBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+            org.tensorflow.proto.framework.AutoParallelOptions, org.tensorflow.proto.framework.AutoParallelOptions.Builder, org.tensorflow.proto.framework.AutoParallelOptionsOrBuilder>(
+                getAutoParallel(),
+                getParentForChildren(),
+                isClean());
+        autoParallel_ = null;
+      }
+      return autoParallelBuilder_;
+    }
+
+    private boolean failOnOptimizerErrors_ ;
+    /**
+     * <pre>
+     * If true, any optimization pass failing will cause the MetaOptimizer to
+     * stop with an error. By default - or when set to false, failing passes are
+     * skipped silently.
+     * </pre>
+     *
+     * <code>bool fail_on_optimizer_errors = 21;</code>
+     */
+    public boolean getFailOnOptimizerErrors() {
+      return failOnOptimizerErrors_;
+    }
+    /**
+     * <pre>
+     * If true, any optimization pass failing will cause the MetaOptimizer to
+     * stop with an error. By default - or when set to false, failing passes are
+     * skipped silently.
+     * </pre>
+     *
+     * <code>bool fail_on_optimizer_errors = 21;</code>
+     */
+    public Builder setFailOnOptimizerErrors(boolean value) {
+      
+      failOnOptimizerErrors_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * If true, any optimization pass failing will cause the MetaOptimizer to
+     * stop with an error. By default - or when set to false, failing passes are
+     * skipped silently.
+     * </pre>
+     *
+     * <code>bool fail_on_optimizer_errors = 21;</code>
+     */
+    public Builder clearFailOnOptimizerErrors() {
+      
+      failOnOptimizerErrors_ = false;
+      onChanged();
+      return this;
+    }
+
+    private org.tensorflow.proto.framework.ScopedAllocatorOptions scopedAllocatorOpts_;
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.ScopedAllocatorOptions, org.tensorflow.proto.framework.ScopedAllocatorOptions.Builder, org.tensorflow.proto.framework.ScopedAllocatorOptionsOrBuilder> scopedAllocatorOptsBuilder_;
+    /**
+     * <code>.tensorflow.ScopedAllocatorOptions scoped_allocator_opts = 16;</code>
+     */
+    public boolean hasScopedAllocatorOpts() {
+      return scopedAllocatorOptsBuilder_ != null || scopedAllocatorOpts_ != null;
+    }
+    /**
+     * <code>.tensorflow.ScopedAllocatorOptions scoped_allocator_opts = 16;</code>
+     */
+    public org.tensorflow.proto.framework.ScopedAllocatorOptions getScopedAllocatorOpts() {
+      if (scopedAllocatorOptsBuilder_ == null) {
+        return scopedAllocatorOpts_ == null ? org.tensorflow.proto.framework.ScopedAllocatorOptions.getDefaultInstance() : scopedAllocatorOpts_;
+      } else {
+        return scopedAllocatorOptsBuilder_.getMessage();
+      }
+    }
+    /**
+     * <code>.tensorflow.ScopedAllocatorOptions scoped_allocator_opts = 16;</code>
+     */
+    public Builder setScopedAllocatorOpts(org.tensorflow.proto.framework.ScopedAllocatorOptions value) {
+      if (scopedAllocatorOptsBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        scopedAllocatorOpts_ = value;
+        onChanged();
+      } else {
+        scopedAllocatorOptsBuilder_.setMessage(value);
+      }
+
+      return this;
+    }
+    /**
+     * <code>.tensorflow.ScopedAllocatorOptions scoped_allocator_opts = 16;</code>
+     */
+    public Builder setScopedAllocatorOpts(
+        org.tensorflow.proto.framework.ScopedAllocatorOptions.Builder builderForValue) {
+      if (scopedAllocatorOptsBuilder_ == null) {
+        scopedAllocatorOpts_ = builderForValue.build();
+        onChanged();
+      } else {
+        scopedAllocatorOptsBuilder_.setMessage(builderForValue.build());
+      }
+
+      return this;
+    }
+    /**
+     * <code>.tensorflow.ScopedAllocatorOptions scoped_allocator_opts = 16;</code>
+     */
+    public Builder mergeScopedAllocatorOpts(org.tensorflow.proto.framework.ScopedAllocatorOptions value) {
+      if (scopedAllocatorOptsBuilder_ == null) {
+        if (scopedAllocatorOpts_ != null) {
+          scopedAllocatorOpts_ =
+            org.tensorflow.proto.framework.ScopedAllocatorOptions.newBuilder(scopedAllocatorOpts_).mergeFrom(value).buildPartial();
+        } else {
+          scopedAllocatorOpts_ = value;
+        }
+        onChanged();
+      } else {
+        scopedAllocatorOptsBuilder_.mergeFrom(value);
+      }
+
+      return this;
+    }
+    /**
+     * <code>.tensorflow.ScopedAllocatorOptions scoped_allocator_opts = 16;</code>
+     */
+    public Builder clearScopedAllocatorOpts() {
+      if (scopedAllocatorOptsBuilder_ == null) {
+        scopedAllocatorOpts_ = null;
+        onChanged();
+      } else {
+        scopedAllocatorOpts_ = null;
+        scopedAllocatorOptsBuilder_ = null;
+      }
+
+      return this;
+    }
+    /**
+     * <code>.tensorflow.ScopedAllocatorOptions scoped_allocator_opts = 16;</code>
+     */
+    public org.tensorflow.proto.framework.ScopedAllocatorOptions.Builder getScopedAllocatorOptsBuilder() {
+      
+      onChanged();
+      return getScopedAllocatorOptsFieldBuilder().getBuilder();
+    }
+    /**
+     * <code>.tensorflow.ScopedAllocatorOptions scoped_allocator_opts = 16;</code>
+     */
+    public org.tensorflow.proto.framework.ScopedAllocatorOptionsOrBuilder getScopedAllocatorOptsOrBuilder() {
+      if (scopedAllocatorOptsBuilder_ != null) {
+        return scopedAllocatorOptsBuilder_.getMessageOrBuilder();
+      } else {
+        return scopedAllocatorOpts_ == null ?
+            org.tensorflow.proto.framework.ScopedAllocatorOptions.getDefaultInstance() : scopedAllocatorOpts_;
+      }
+    }
+    /**
+     * <code>.tensorflow.ScopedAllocatorOptions scoped_allocator_opts = 16;</code>
+     */
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.ScopedAllocatorOptions, org.tensorflow.proto.framework.ScopedAllocatorOptions.Builder, org.tensorflow.proto.framework.ScopedAllocatorOptionsOrBuilder> 
+        getScopedAllocatorOptsFieldBuilder() {
+      if (scopedAllocatorOptsBuilder_ == null) {
+        scopedAllocatorOptsBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+            org.tensorflow.proto.framework.ScopedAllocatorOptions, org.tensorflow.proto.framework.ScopedAllocatorOptions.Builder, org.tensorflow.proto.framework.ScopedAllocatorOptionsOrBuilder>(
+                getScopedAllocatorOpts(),
+                getParentForChildren(),
+                isClean());
+        scopedAllocatorOpts_ = null;
+      }
+      return scopedAllocatorOptsBuilder_;
+    }
+
+    private com.google.protobuf.LazyStringList optimizers_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+    private void ensureOptimizersIsMutable() {
+      if (!((bitField0_ & 0x00000001) != 0)) {
+        optimizers_ = new com.google.protobuf.LazyStringArrayList(optimizers_);
+        bitField0_ |= 0x00000001;
+       }
+    }
+    /**
+     * <pre>
+     * If non-empty, will use this as an alternative way to specify a list of
+     * optimizations to turn on and the order of the optimizations (replacing the
+     * meta-optimizer).
+     * Of the RewriterConfig options, only the AutoParallel configuration options
+     * (the auto_parallel field) apply to manually requested optimization passes
+     * ("autoparallel"). Memory optimization passes ("memory") invoked here are
+     * not configurable (in contrast to memory optimization passes through the
+     * meta-optimizer) and act only on manual op annotations.
+     * Custom optimizers (see custom_optimizers) that are not part of this
+     * schedule will be run after - in the order that they were specified.
+     * </pre>
+     *
+     * <code>repeated string optimizers = 100;</code>
+     */
+    public com.google.protobuf.ProtocolStringList
+        getOptimizersList() {
+      return optimizers_.getUnmodifiableView();
+    }
+    /**
+     * <pre>
+     * If non-empty, will use this as an alternative way to specify a list of
+     * optimizations to turn on and the order of the optimizations (replacing the
+     * meta-optimizer).
+     * Of the RewriterConfig options, only the AutoParallel configuration options
+     * (the auto_parallel field) apply to manually requested optimization passes
+     * ("autoparallel"). Memory optimization passes ("memory") invoked here are
+     * not configurable (in contrast to memory optimization passes through the
+     * meta-optimizer) and act only on manual op annotations.
+     * Custom optimizers (see custom_optimizers) that are not part of this
+     * schedule will be run after - in the order that they were specified.
+     * </pre>
+     *
+     * <code>repeated string optimizers = 100;</code>
+     */
+    public int getOptimizersCount() {
+      return optimizers_.size();
+    }
+    /**
+     * <pre>
+     * If non-empty, will use this as an alternative way to specify a list of
+     * optimizations to turn on and the order of the optimizations (replacing the
+     * meta-optimizer).
+     * Of the RewriterConfig options, only the AutoParallel configuration options
+     * (the auto_parallel field) apply to manually requested optimization passes
+     * ("autoparallel"). Memory optimization passes ("memory") invoked here are
+     * not configurable (in contrast to memory optimization passes through the
+     * meta-optimizer) and act only on manual op annotations.
+     * Custom optimizers (see custom_optimizers) that are not part of this
+     * schedule will be run after - in the order that they were specified.
+     * </pre>
+     *
+     * <code>repeated string optimizers = 100;</code>
+     */
+    public java.lang.String getOptimizers(int index) {
+      return optimizers_.get(index);
+    }
+    /**
+     * <pre>
+     * If non-empty, will use this as an alternative way to specify a list of
+     * optimizations to turn on and the order of the optimizations (replacing the
+     * meta-optimizer).
+     * Of the RewriterConfig options, only the AutoParallel configuration options
+     * (the auto_parallel field) apply to manually requested optimization passes
+     * ("autoparallel"). Memory optimization passes ("memory") invoked here are
+     * not configurable (in contrast to memory optimization passes through the
+     * meta-optimizer) and act only on manual op annotations.
+     * Custom optimizers (see custom_optimizers) that are not part of this
+     * schedule will be run after - in the order that they were specified.
+     * </pre>
+     *
+     * <code>repeated string optimizers = 100;</code>
+     */
+    public com.google.protobuf.ByteString
+        getOptimizersBytes(int index) {
+      return optimizers_.getByteString(index);
+    }
+    /**
+     * <pre>
+     * If non-empty, will use this as an alternative way to specify a list of
+     * optimizations to turn on and the order of the optimizations (replacing the
+     * meta-optimizer).
+     * Of the RewriterConfig options, only the AutoParallel configuration options
+     * (the auto_parallel field) apply to manually requested optimization passes
+     * ("autoparallel"). Memory optimization passes ("memory") invoked here are
+     * not configurable (in contrast to memory optimization passes through the
+     * meta-optimizer) and act only on manual op annotations.
+     * Custom optimizers (see custom_optimizers) that are not part of this
+     * schedule will be run after - in the order that they were specified.
+     * </pre>
+     *
+     * <code>repeated string optimizers = 100;</code>
+     */
+    public Builder setOptimizers(
+        int index, java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  ensureOptimizersIsMutable();
+      optimizers_.set(index, value);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * If non-empty, will use this as an alternative way to specify a list of
+     * optimizations to turn on and the order of the optimizations (replacing the
+     * meta-optimizer).
+     * Of the RewriterConfig options, only the AutoParallel configuration options
+     * (the auto_parallel field) apply to manually requested optimization passes
+     * ("autoparallel"). Memory optimization passes ("memory") invoked here are
+     * not configurable (in contrast to memory optimization passes through the
+     * meta-optimizer) and act only on manual op annotations.
+     * Custom optimizers (see custom_optimizers) that are not part of this
+     * schedule will be run after - in the order that they were specified.
+     * </pre>
+     *
+     * <code>repeated string optimizers = 100;</code>
+     */
+    public Builder addOptimizers(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  ensureOptimizersIsMutable();
+      optimizers_.add(value);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * If non-empty, will use this as an alternative way to specify a list of
+     * optimizations to turn on and the order of the optimizations (replacing the
+     * meta-optimizer).
+     * Of the RewriterConfig options, only the AutoParallel configuration options
+     * (the auto_parallel field) apply to manually requested optimization passes
+     * ("autoparallel"). Memory optimization passes ("memory") invoked here are
+     * not configurable (in contrast to memory optimization passes through the
+     * meta-optimizer) and act only on manual op annotations.
+     * Custom optimizers (see custom_optimizers) that are not part of this
+     * schedule will be run after - in the order that they were specified.
+     * </pre>
+     *
+     * <code>repeated string optimizers = 100;</code>
+     */
+    public Builder addAllOptimizers(
+        java.lang.Iterable<java.lang.String> values) {
+      ensureOptimizersIsMutable();
+      com.google.protobuf.AbstractMessageLite.Builder.addAll(
+          values, optimizers_);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * If non-empty, will use this as an alternative way to specify a list of
+     * optimizations to turn on and the order of the optimizations (replacing the
+     * meta-optimizer).
+     * Of the RewriterConfig options, only the AutoParallel configuration options
+     * (the auto_parallel field) apply to manually requested optimization passes
+     * ("autoparallel"). Memory optimization passes ("memory") invoked here are
+     * not configurable (in contrast to memory optimization passes through the
+     * meta-optimizer) and act only on manual op annotations.
+     * Custom optimizers (see custom_optimizers) that are not part of this
+     * schedule will be run after - in the order that they were specified.
+     * </pre>
+     *
+     * <code>repeated string optimizers = 100;</code>
+     */
+    public Builder clearOptimizers() {
+      optimizers_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+      bitField0_ = (bitField0_ & ~0x00000001);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * If non-empty, will use this as an alternative way to specify a list of
+     * optimizations to turn on and the order of the optimizations (replacing the
+     * meta-optimizer).
+     * Of the RewriterConfig options, only the AutoParallel configuration options
+     * (the auto_parallel field) apply to manually requested optimization passes
+     * ("autoparallel"). Memory optimization passes ("memory") invoked here are
+     * not configurable (in contrast to memory optimization passes through the
+     * meta-optimizer) and act only on manual op annotations.
+     * Custom optimizers (see custom_optimizers) that are not part of this
+     * schedule will be run after - in the order that they were specified.
+     * </pre>
+     *
+     * <code>repeated string optimizers = 100;</code>
+     */
+    public Builder addOptimizersBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      ensureOptimizersIsMutable();
+      optimizers_.add(value);
+      onChanged();
+      return this;
+    }
+
+    private java.util.List<org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizer> customOptimizers_ =
+      java.util.Collections.emptyList();
+    private void ensureCustomOptimizersIsMutable() {
+      if (!((bitField0_ & 0x00000002) != 0)) {
+        customOptimizers_ = new java.util.ArrayList<org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizer>(customOptimizers_);
+        bitField0_ |= 0x00000002;
+       }
+    }
+
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizer, org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizer.Builder, org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizerOrBuilder> customOptimizersBuilder_;
+
+    /**
+     * <pre>
+     * list of CustomGraphOptimizers to apply.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.RewriterConfig.CustomGraphOptimizer custom_optimizers = 200;</code>
+     */
+    public java.util.List<org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizer> getCustomOptimizersList() {
+      if (customOptimizersBuilder_ == null) {
+        return java.util.Collections.unmodifiableList(customOptimizers_);
+      } else {
+        return customOptimizersBuilder_.getMessageList();
+      }
+    }
+    /**
+     * <pre>
+     * list of CustomGraphOptimizers to apply.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.RewriterConfig.CustomGraphOptimizer custom_optimizers = 200;</code>
+     */
+    public int getCustomOptimizersCount() {
+      if (customOptimizersBuilder_ == null) {
+        return customOptimizers_.size();
+      } else {
+        return customOptimizersBuilder_.getCount();
+      }
+    }
+    /**
+     * <pre>
+     * list of CustomGraphOptimizers to apply.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.RewriterConfig.CustomGraphOptimizer custom_optimizers = 200;</code>
+     */
+    public org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizer getCustomOptimizers(int index) {
+      if (customOptimizersBuilder_ == null) {
+        return customOptimizers_.get(index);
+      } else {
+        return customOptimizersBuilder_.getMessage(index);
+      }
+    }
+    /**
+     * <pre>
+     * list of CustomGraphOptimizers to apply.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.RewriterConfig.CustomGraphOptimizer custom_optimizers = 200;</code>
+     */
+    public Builder setCustomOptimizers(
+        int index, org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizer value) {
+      if (customOptimizersBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureCustomOptimizersIsMutable();
+        customOptimizers_.set(index, value);
+        onChanged();
+      } else {
+        customOptimizersBuilder_.setMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * list of CustomGraphOptimizers to apply.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.RewriterConfig.CustomGraphOptimizer custom_optimizers = 200;</code>
+     */
+    public Builder setCustomOptimizers(
+        int index, org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizer.Builder builderForValue) {
+      if (customOptimizersBuilder_ == null) {
+        ensureCustomOptimizersIsMutable();
+        customOptimizers_.set(index, builderForValue.build());
+        onChanged();
+      } else {
+        customOptimizersBuilder_.setMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * list of CustomGraphOptimizers to apply.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.RewriterConfig.CustomGraphOptimizer custom_optimizers = 200;</code>
+     */
+    public Builder addCustomOptimizers(org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizer value) {
+      if (customOptimizersBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureCustomOptimizersIsMutable();
+        customOptimizers_.add(value);
+        onChanged();
+      } else {
+        customOptimizersBuilder_.addMessage(value);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * list of CustomGraphOptimizers to apply.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.RewriterConfig.CustomGraphOptimizer custom_optimizers = 200;</code>
+     */
+    public Builder addCustomOptimizers(
+        int index, org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizer value) {
+      if (customOptimizersBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureCustomOptimizersIsMutable();
+        customOptimizers_.add(index, value);
+        onChanged();
+      } else {
+        customOptimizersBuilder_.addMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * list of CustomGraphOptimizers to apply.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.RewriterConfig.CustomGraphOptimizer custom_optimizers = 200;</code>
+     */
+    public Builder addCustomOptimizers(
+        org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizer.Builder builderForValue) {
+      if (customOptimizersBuilder_ == null) {
+        ensureCustomOptimizersIsMutable();
+        customOptimizers_.add(builderForValue.build());
+        onChanged();
+      } else {
+        customOptimizersBuilder_.addMessage(builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * list of CustomGraphOptimizers to apply.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.RewriterConfig.CustomGraphOptimizer custom_optimizers = 200;</code>
+     */
+    public Builder addCustomOptimizers(
+        int index, org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizer.Builder builderForValue) {
+      if (customOptimizersBuilder_ == null) {
+        ensureCustomOptimizersIsMutable();
+        customOptimizers_.add(index, builderForValue.build());
+        onChanged();
+      } else {
+        customOptimizersBuilder_.addMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * list of CustomGraphOptimizers to apply.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.RewriterConfig.CustomGraphOptimizer custom_optimizers = 200;</code>
+     */
+    public Builder addAllCustomOptimizers(
+        java.lang.Iterable<? extends org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizer> values) {
+      if (customOptimizersBuilder_ == null) {
+        ensureCustomOptimizersIsMutable();
+        com.google.protobuf.AbstractMessageLite.Builder.addAll(
+            values, customOptimizers_);
+        onChanged();
+      } else {
+        customOptimizersBuilder_.addAllMessages(values);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * list of CustomGraphOptimizers to apply.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.RewriterConfig.CustomGraphOptimizer custom_optimizers = 200;</code>
+     */
+    public Builder clearCustomOptimizers() {
+      if (customOptimizersBuilder_ == null) {
+        customOptimizers_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000002);
+        onChanged();
+      } else {
+        customOptimizersBuilder_.clear();
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * list of CustomGraphOptimizers to apply.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.RewriterConfig.CustomGraphOptimizer custom_optimizers = 200;</code>
+     */
+    public Builder removeCustomOptimizers(int index) {
+      if (customOptimizersBuilder_ == null) {
+        ensureCustomOptimizersIsMutable();
+        customOptimizers_.remove(index);
+        onChanged();
+      } else {
+        customOptimizersBuilder_.remove(index);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * list of CustomGraphOptimizers to apply.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.RewriterConfig.CustomGraphOptimizer custom_optimizers = 200;</code>
+     */
+    public org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizer.Builder getCustomOptimizersBuilder(
+        int index) {
+      return getCustomOptimizersFieldBuilder().getBuilder(index);
+    }
+    /**
+     * <pre>
+     * list of CustomGraphOptimizers to apply.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.RewriterConfig.CustomGraphOptimizer custom_optimizers = 200;</code>
+     */
+    public org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizerOrBuilder getCustomOptimizersOrBuilder(
+        int index) {
+      if (customOptimizersBuilder_ == null) {
+        return customOptimizers_.get(index);  } else {
+        return customOptimizersBuilder_.getMessageOrBuilder(index);
+      }
+    }
+    /**
+     * <pre>
+     * list of CustomGraphOptimizers to apply.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.RewriterConfig.CustomGraphOptimizer custom_optimizers = 200;</code>
+     */
+    public java.util.List<? extends org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizerOrBuilder> 
+         getCustomOptimizersOrBuilderList() {
+      if (customOptimizersBuilder_ != null) {
+        return customOptimizersBuilder_.getMessageOrBuilderList();
+      } else {
+        return java.util.Collections.unmodifiableList(customOptimizers_);
+      }
+    }
+    /**
+     * <pre>
+     * list of CustomGraphOptimizers to apply.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.RewriterConfig.CustomGraphOptimizer custom_optimizers = 200;</code>
+     */
+    public org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizer.Builder addCustomOptimizersBuilder() {
+      return getCustomOptimizersFieldBuilder().addBuilder(
+          org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizer.getDefaultInstance());
+    }
+    /**
+     * <pre>
+     * list of CustomGraphOptimizers to apply.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.RewriterConfig.CustomGraphOptimizer custom_optimizers = 200;</code>
+     */
+    public org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizer.Builder addCustomOptimizersBuilder(
+        int index) {
+      return getCustomOptimizersFieldBuilder().addBuilder(
+          index, org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizer.getDefaultInstance());
+    }
+    /**
+     * <pre>
+     * list of CustomGraphOptimizers to apply.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.RewriterConfig.CustomGraphOptimizer custom_optimizers = 200;</code>
+     */
+    public java.util.List<org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizer.Builder> 
+         getCustomOptimizersBuilderList() {
+      return getCustomOptimizersFieldBuilder().getBuilderList();
+    }
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizer, org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizer.Builder, org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizerOrBuilder> 
+        getCustomOptimizersFieldBuilder() {
+      if (customOptimizersBuilder_ == null) {
+        customOptimizersBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
+            org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizer, org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizer.Builder, org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizerOrBuilder>(
+                customOptimizers_,
+                ((bitField0_ & 0x00000002) != 0),
+                getParentForChildren(),
+                isClean());
+        customOptimizers_ = null;
+      }
+      return customOptimizersBuilder_;
+    }
+
+    private org.tensorflow.proto.framework.VerifierConfig interOptimizerVerifierConfig_;
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.VerifierConfig, org.tensorflow.proto.framework.VerifierConfig.Builder, org.tensorflow.proto.framework.VerifierConfigOrBuilder> interOptimizerVerifierConfigBuilder_;
+    /**
+     * <pre>
+     * VerifierConfig specifying the verifiers to be run after every optimizer.
+     * </pre>
+     *
+     * <code>.tensorflow.VerifierConfig inter_optimizer_verifier_config = 300;</code>
+     */
+    public boolean hasInterOptimizerVerifierConfig() {
+      return interOptimizerVerifierConfigBuilder_ != null || interOptimizerVerifierConfig_ != null;
+    }
+    /**
+     * <pre>
+     * VerifierConfig specifying the verifiers to be run after every optimizer.
+     * </pre>
+     *
+     * <code>.tensorflow.VerifierConfig inter_optimizer_verifier_config = 300;</code>
+     */
+    public org.tensorflow.proto.framework.VerifierConfig getInterOptimizerVerifierConfig() {
+      if (interOptimizerVerifierConfigBuilder_ == null) {
+        return interOptimizerVerifierConfig_ == null ? org.tensorflow.proto.framework.VerifierConfig.getDefaultInstance() : interOptimizerVerifierConfig_;
+      } else {
+        return interOptimizerVerifierConfigBuilder_.getMessage();
+      }
+    }
+    /**
+     * <pre>
+     * VerifierConfig specifying the verifiers to be run after every optimizer.
+     * </pre>
+     *
+     * <code>.tensorflow.VerifierConfig inter_optimizer_verifier_config = 300;</code>
+     */
+    public Builder setInterOptimizerVerifierConfig(org.tensorflow.proto.framework.VerifierConfig value) {
+      if (interOptimizerVerifierConfigBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        interOptimizerVerifierConfig_ = value;
+        onChanged();
+      } else {
+        interOptimizerVerifierConfigBuilder_.setMessage(value);
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * VerifierConfig specifying the verifiers to be run after every optimizer.
+     * </pre>
+     *
+     * <code>.tensorflow.VerifierConfig inter_optimizer_verifier_config = 300;</code>
+     */
+    public Builder setInterOptimizerVerifierConfig(
+        org.tensorflow.proto.framework.VerifierConfig.Builder builderForValue) {
+      if (interOptimizerVerifierConfigBuilder_ == null) {
+        interOptimizerVerifierConfig_ = builderForValue.build();
+        onChanged();
+      } else {
+        interOptimizerVerifierConfigBuilder_.setMessage(builderForValue.build());
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * VerifierConfig specifying the verifiers to be run after every optimizer.
+     * </pre>
+     *
+     * <code>.tensorflow.VerifierConfig inter_optimizer_verifier_config = 300;</code>
+     */
+    public Builder mergeInterOptimizerVerifierConfig(org.tensorflow.proto.framework.VerifierConfig value) {
+      if (interOptimizerVerifierConfigBuilder_ == null) {
+        if (interOptimizerVerifierConfig_ != null) {
+          interOptimizerVerifierConfig_ =
+            org.tensorflow.proto.framework.VerifierConfig.newBuilder(interOptimizerVerifierConfig_).mergeFrom(value).buildPartial();
+        } else {
+          interOptimizerVerifierConfig_ = value;
+        }
+        onChanged();
+      } else {
+        interOptimizerVerifierConfigBuilder_.mergeFrom(value);
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * VerifierConfig specifying the verifiers to be run after every optimizer.
+     * </pre>
+     *
+     * <code>.tensorflow.VerifierConfig inter_optimizer_verifier_config = 300;</code>
+     */
+    public Builder clearInterOptimizerVerifierConfig() {
+      if (interOptimizerVerifierConfigBuilder_ == null) {
+        interOptimizerVerifierConfig_ = null;
+        onChanged();
+      } else {
+        interOptimizerVerifierConfig_ = null;
+        interOptimizerVerifierConfigBuilder_ = null;
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * VerifierConfig specifying the verifiers to be run after every optimizer.
+     * </pre>
+     *
+     * <code>.tensorflow.VerifierConfig inter_optimizer_verifier_config = 300;</code>
+     */
+    public org.tensorflow.proto.framework.VerifierConfig.Builder getInterOptimizerVerifierConfigBuilder() {
+      
+      onChanged();
+      return getInterOptimizerVerifierConfigFieldBuilder().getBuilder();
+    }
+    /**
+     * <pre>
+     * VerifierConfig specifying the verifiers to be run after every optimizer.
+     * </pre>
+     *
+     * <code>.tensorflow.VerifierConfig inter_optimizer_verifier_config = 300;</code>
+     */
+    public org.tensorflow.proto.framework.VerifierConfigOrBuilder getInterOptimizerVerifierConfigOrBuilder() {
+      if (interOptimizerVerifierConfigBuilder_ != null) {
+        return interOptimizerVerifierConfigBuilder_.getMessageOrBuilder();
+      } else {
+        return interOptimizerVerifierConfig_ == null ?
+            org.tensorflow.proto.framework.VerifierConfig.getDefaultInstance() : interOptimizerVerifierConfig_;
+      }
+    }
+    /**
+     * <pre>
+     * VerifierConfig specifying the verifiers to be run after every optimizer.
+     * </pre>
+     *
+     * <code>.tensorflow.VerifierConfig inter_optimizer_verifier_config = 300;</code>
+     */
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.VerifierConfig, org.tensorflow.proto.framework.VerifierConfig.Builder, org.tensorflow.proto.framework.VerifierConfigOrBuilder> 
+        getInterOptimizerVerifierConfigFieldBuilder() {
+      if (interOptimizerVerifierConfigBuilder_ == null) {
+        interOptimizerVerifierConfigBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+            org.tensorflow.proto.framework.VerifierConfig, org.tensorflow.proto.framework.VerifierConfig.Builder, org.tensorflow.proto.framework.VerifierConfigOrBuilder>(
+                getInterOptimizerVerifierConfig(),
+                getParentForChildren(),
+                isClean());
+        interOptimizerVerifierConfig_ = null;
+      }
+      return interOptimizerVerifierConfigBuilder_;
+    }
+
+    private org.tensorflow.proto.framework.VerifierConfig postOptimizationVerifierConfig_;
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.VerifierConfig, org.tensorflow.proto.framework.VerifierConfig.Builder, org.tensorflow.proto.framework.VerifierConfigOrBuilder> postOptimizationVerifierConfigBuilder_;
+    /**
+     * <pre>
+     * VerifierConfig specifying the verifiers to be run at the end, after all
+     * optimizers have run.
+     * </pre>
+     *
+     * <code>.tensorflow.VerifierConfig post_optimization_verifier_config = 301;</code>
+     */
+    public boolean hasPostOptimizationVerifierConfig() {
+      return postOptimizationVerifierConfigBuilder_ != null || postOptimizationVerifierConfig_ != null;
+    }
+    /**
+     * <pre>
+     * VerifierConfig specifying the verifiers to be run at the end, after all
+     * optimizers have run.
+     * </pre>
+     *
+     * <code>.tensorflow.VerifierConfig post_optimization_verifier_config = 301;</code>
+     */
+    public org.tensorflow.proto.framework.VerifierConfig getPostOptimizationVerifierConfig() {
+      if (postOptimizationVerifierConfigBuilder_ == null) {
+        return postOptimizationVerifierConfig_ == null ? org.tensorflow.proto.framework.VerifierConfig.getDefaultInstance() : postOptimizationVerifierConfig_;
+      } else {
+        return postOptimizationVerifierConfigBuilder_.getMessage();
+      }
+    }
+    /**
+     * <pre>
+     * VerifierConfig specifying the verifiers to be run at the end, after all
+     * optimizers have run.
+     * </pre>
+     *
+     * <code>.tensorflow.VerifierConfig post_optimization_verifier_config = 301;</code>
+     */
+    public Builder setPostOptimizationVerifierConfig(org.tensorflow.proto.framework.VerifierConfig value) {
+      if (postOptimizationVerifierConfigBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        postOptimizationVerifierConfig_ = value;
+        onChanged();
+      } else {
+        postOptimizationVerifierConfigBuilder_.setMessage(value);
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * VerifierConfig specifying the verifiers to be run at the end, after all
+     * optimizers have run.
+     * </pre>
+     *
+     * <code>.tensorflow.VerifierConfig post_optimization_verifier_config = 301;</code>
+     */
+    public Builder setPostOptimizationVerifierConfig(
+        org.tensorflow.proto.framework.VerifierConfig.Builder builderForValue) {
+      if (postOptimizationVerifierConfigBuilder_ == null) {
+        postOptimizationVerifierConfig_ = builderForValue.build();
+        onChanged();
+      } else {
+        postOptimizationVerifierConfigBuilder_.setMessage(builderForValue.build());
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * VerifierConfig specifying the verifiers to be run at the end, after all
+     * optimizers have run.
+     * </pre>
+     *
+     * <code>.tensorflow.VerifierConfig post_optimization_verifier_config = 301;</code>
+     */
+    public Builder mergePostOptimizationVerifierConfig(org.tensorflow.proto.framework.VerifierConfig value) {
+      if (postOptimizationVerifierConfigBuilder_ == null) {
+        if (postOptimizationVerifierConfig_ != null) {
+          postOptimizationVerifierConfig_ =
+            org.tensorflow.proto.framework.VerifierConfig.newBuilder(postOptimizationVerifierConfig_).mergeFrom(value).buildPartial();
+        } else {
+          postOptimizationVerifierConfig_ = value;
+        }
+        onChanged();
+      } else {
+        postOptimizationVerifierConfigBuilder_.mergeFrom(value);
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * VerifierConfig specifying the verifiers to be run at the end, after all
+     * optimizers have run.
+     * </pre>
+     *
+     * <code>.tensorflow.VerifierConfig post_optimization_verifier_config = 301;</code>
+     */
+    public Builder clearPostOptimizationVerifierConfig() {
+      if (postOptimizationVerifierConfigBuilder_ == null) {
+        postOptimizationVerifierConfig_ = null;
+        onChanged();
+      } else {
+        postOptimizationVerifierConfig_ = null;
+        postOptimizationVerifierConfigBuilder_ = null;
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * VerifierConfig specifying the verifiers to be run at the end, after all
+     * optimizers have run.
+     * </pre>
+     *
+     * <code>.tensorflow.VerifierConfig post_optimization_verifier_config = 301;</code>
+     */
+    public org.tensorflow.proto.framework.VerifierConfig.Builder getPostOptimizationVerifierConfigBuilder() {
+      
+      onChanged();
+      return getPostOptimizationVerifierConfigFieldBuilder().getBuilder();
+    }
+    /**
+     * <pre>
+     * VerifierConfig specifying the verifiers to be run at the end, after all
+     * optimizers have run.
+     * </pre>
+     *
+     * <code>.tensorflow.VerifierConfig post_optimization_verifier_config = 301;</code>
+     */
+    public org.tensorflow.proto.framework.VerifierConfigOrBuilder getPostOptimizationVerifierConfigOrBuilder() {
+      if (postOptimizationVerifierConfigBuilder_ != null) {
+        return postOptimizationVerifierConfigBuilder_.getMessageOrBuilder();
+      } else {
+        return postOptimizationVerifierConfig_ == null ?
+            org.tensorflow.proto.framework.VerifierConfig.getDefaultInstance() : postOptimizationVerifierConfig_;
+      }
+    }
+    /**
+     * <pre>
+     * VerifierConfig specifying the verifiers to be run at the end, after all
+     * optimizers have run.
+     * </pre>
+     *
+     * <code>.tensorflow.VerifierConfig post_optimization_verifier_config = 301;</code>
+     */
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.VerifierConfig, org.tensorflow.proto.framework.VerifierConfig.Builder, org.tensorflow.proto.framework.VerifierConfigOrBuilder> 
+        getPostOptimizationVerifierConfigFieldBuilder() {
+      if (postOptimizationVerifierConfigBuilder_ == null) {
+        postOptimizationVerifierConfigBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+            org.tensorflow.proto.framework.VerifierConfig, org.tensorflow.proto.framework.VerifierConfig.Builder, org.tensorflow.proto.framework.VerifierConfigOrBuilder>(
+                getPostOptimizationVerifierConfig(),
+                getParentForChildren(),
+                isClean());
+        postOptimizationVerifierConfig_ = null;
+      }
+      return postOptimizationVerifierConfigBuilder_;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.RewriterConfig)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.RewriterConfig)
+  private static final org.tensorflow.proto.framework.RewriterConfig DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.RewriterConfig();
+  }
+
+  public static org.tensorflow.proto.framework.RewriterConfig getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<RewriterConfig>
+      PARSER = new com.google.protobuf.AbstractParser<RewriterConfig>() {
+    @java.lang.Override
+    public RewriterConfig parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new RewriterConfig(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<RewriterConfig> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<RewriterConfig> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.framework.RewriterConfig getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/RewriterConfigOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/RewriterConfigOrBuilder.java
new file mode 100644
index 00000000000..a8a1c6f1ed0
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/RewriterConfigOrBuilder.java
@@ -0,0 +1,627 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/rewriter_config.proto
+
+package org.tensorflow.proto.framework;
+
+public interface RewriterConfigOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.RewriterConfig)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <pre>
+   * Optimize tensor layouts (default is ON)
+   * e.g. This will try to use NCHW layout on GPU which is faster.
+   * </pre>
+   *
+   * <code>.tensorflow.RewriterConfig.Toggle layout_optimizer = 1;</code>
+   */
+  int getLayoutOptimizerValue();
+  /**
+   * <pre>
+   * Optimize tensor layouts (default is ON)
+   * e.g. This will try to use NCHW layout on GPU which is faster.
+   * </pre>
+   *
+   * <code>.tensorflow.RewriterConfig.Toggle layout_optimizer = 1;</code>
+   */
+  org.tensorflow.proto.framework.RewriterConfig.Toggle getLayoutOptimizer();
+
+  /**
+   * <pre>
+   * Fold constants (default is ON)
+   * Statically infer the value of tensors when possible, and materialize the
+   * result using constants.
+   * </pre>
+   *
+   * <code>.tensorflow.RewriterConfig.Toggle constant_folding = 3;</code>
+   */
+  int getConstantFoldingValue();
+  /**
+   * <pre>
+   * Fold constants (default is ON)
+   * Statically infer the value of tensors when possible, and materialize the
+   * result using constants.
+   * </pre>
+   *
+   * <code>.tensorflow.RewriterConfig.Toggle constant_folding = 3;</code>
+   */
+  org.tensorflow.proto.framework.RewriterConfig.Toggle getConstantFolding();
+
+  /**
+   * <pre>
+   * Shape optimizations (default is ON)
+   * Simplify computations made on shapes.
+   * </pre>
+   *
+   * <code>.tensorflow.RewriterConfig.Toggle shape_optimization = 13;</code>
+   */
+  int getShapeOptimizationValue();
+  /**
+   * <pre>
+   * Shape optimizations (default is ON)
+   * Simplify computations made on shapes.
+   * </pre>
+   *
+   * <code>.tensorflow.RewriterConfig.Toggle shape_optimization = 13;</code>
+   */
+  org.tensorflow.proto.framework.RewriterConfig.Toggle getShapeOptimization();
+
+  /**
+   * <pre>
+   * Remapping (default is ON)
+   * Remap subgraphs onto more efficient implementations.
+   * </pre>
+   *
+   * <code>.tensorflow.RewriterConfig.Toggle remapping = 14;</code>
+   */
+  int getRemappingValue();
+  /**
+   * <pre>
+   * Remapping (default is ON)
+   * Remap subgraphs onto more efficient implementations.
+   * </pre>
+   *
+   * <code>.tensorflow.RewriterConfig.Toggle remapping = 14;</code>
+   */
+  org.tensorflow.proto.framework.RewriterConfig.Toggle getRemapping();
+
+  /**
+   * <pre>
+   * Common subgraph elimination (default is ON)
+   * e.g. Simplify arithmetic ops; merge ops with same value (like constants).
+   * </pre>
+   *
+   * <code>.tensorflow.RewriterConfig.Toggle common_subgraph_elimination = 24;</code>
+   */
+  int getCommonSubgraphEliminationValue();
+  /**
+   * <pre>
+   * Common subgraph elimination (default is ON)
+   * e.g. Simplify arithmetic ops; merge ops with same value (like constants).
+   * </pre>
+   *
+   * <code>.tensorflow.RewriterConfig.Toggle common_subgraph_elimination = 24;</code>
+   */
+  org.tensorflow.proto.framework.RewriterConfig.Toggle getCommonSubgraphElimination();
+
+  /**
+   * <pre>
+   * Arithmetic optimizations (default is ON)
+   * e.g. Simplify arithmetic ops; merge ops with same value (like constants).
+   * </pre>
+   *
+   * <code>.tensorflow.RewriterConfig.Toggle arithmetic_optimization = 7;</code>
+   */
+  int getArithmeticOptimizationValue();
+  /**
+   * <pre>
+   * Arithmetic optimizations (default is ON)
+   * e.g. Simplify arithmetic ops; merge ops with same value (like constants).
+   * </pre>
+   *
+   * <code>.tensorflow.RewriterConfig.Toggle arithmetic_optimization = 7;</code>
+   */
+  org.tensorflow.proto.framework.RewriterConfig.Toggle getArithmeticOptimization();
+
+  /**
+   * <pre>
+   * Control dependency optimizations (default is ON).
+   * Remove redundant control dependencies, which may enable other optimization.
+   * </pre>
+   *
+   * <code>.tensorflow.RewriterConfig.Toggle dependency_optimization = 8;</code>
+   */
+  int getDependencyOptimizationValue();
+  /**
+   * <pre>
+   * Control dependency optimizations (default is ON).
+   * Remove redundant control dependencies, which may enable other optimization.
+   * </pre>
+   *
+   * <code>.tensorflow.RewriterConfig.Toggle dependency_optimization = 8;</code>
+   */
+  org.tensorflow.proto.framework.RewriterConfig.Toggle getDependencyOptimization();
+
+  /**
+   * <pre>
+   * Loop optimizations (default is ON).
+   * </pre>
+   *
+   * <code>.tensorflow.RewriterConfig.Toggle loop_optimization = 9;</code>
+   */
+  int getLoopOptimizationValue();
+  /**
+   * <pre>
+   * Loop optimizations (default is ON).
+   * </pre>
+   *
+   * <code>.tensorflow.RewriterConfig.Toggle loop_optimization = 9;</code>
+   */
+  org.tensorflow.proto.framework.RewriterConfig.Toggle getLoopOptimization();
+
+  /**
+   * <pre>
+   * Function optimizations (default is ON).
+   * </pre>
+   *
+   * <code>.tensorflow.RewriterConfig.Toggle function_optimization = 10;</code>
+   */
+  int getFunctionOptimizationValue();
+  /**
+   * <pre>
+   * Function optimizations (default is ON).
+   * </pre>
+   *
+   * <code>.tensorflow.RewriterConfig.Toggle function_optimization = 10;</code>
+   */
+  org.tensorflow.proto.framework.RewriterConfig.Toggle getFunctionOptimization();
+
+  /**
+   * <pre>
+   * Strips debug-related nodes from the graph (off by default).
+   * </pre>
+   *
+   * <code>.tensorflow.RewriterConfig.Toggle debug_stripper = 11;</code>
+   */
+  int getDebugStripperValue();
+  /**
+   * <pre>
+   * Strips debug-related nodes from the graph (off by default).
+   * </pre>
+   *
+   * <code>.tensorflow.RewriterConfig.Toggle debug_stripper = 11;</code>
+   */
+  org.tensorflow.proto.framework.RewriterConfig.Toggle getDebugStripper();
+
+  /**
+   * <pre>
+   * If true, don't remove unnecessary ops from the graph
+   * </pre>
+   *
+   * <code>bool disable_model_pruning = 2;</code>
+   */
+  boolean getDisableModelPruning();
+
+  /**
+   * <pre>
+   * Try to allocate some independent Op outputs contiguously in order to
+   * merge or eliminate downstream Ops (off by default).
+   * </pre>
+   *
+   * <code>.tensorflow.RewriterConfig.Toggle scoped_allocator_optimization = 15;</code>
+   */
+  int getScopedAllocatorOptimizationValue();
+  /**
+   * <pre>
+   * Try to allocate some independent Op outputs contiguously in order to
+   * merge or eliminate downstream Ops (off by default).
+   * </pre>
+   *
+   * <code>.tensorflow.RewriterConfig.Toggle scoped_allocator_optimization = 15;</code>
+   */
+  org.tensorflow.proto.framework.RewriterConfig.Toggle getScopedAllocatorOptimization();
+
+  /**
+   * <pre>
+   * Force small ops onto the CPU (default is OFF).
+   * </pre>
+   *
+   * <code>.tensorflow.RewriterConfig.Toggle pin_to_host_optimization = 18;</code>
+   */
+  int getPinToHostOptimizationValue();
+  /**
+   * <pre>
+   * Force small ops onto the CPU (default is OFF).
+   * </pre>
+   *
+   * <code>.tensorflow.RewriterConfig.Toggle pin_to_host_optimization = 18;</code>
+   */
+  org.tensorflow.proto.framework.RewriterConfig.Toggle getPinToHostOptimization();
+
+  /**
+   * <pre>
+   * Enable the swap of kernel implementations based on the device placement
+   * (default is ON).
+   * </pre>
+   *
+   * <code>.tensorflow.RewriterConfig.Toggle implementation_selector = 22;</code>
+   */
+  int getImplementationSelectorValue();
+  /**
+   * <pre>
+   * Enable the swap of kernel implementations based on the device placement
+   * (default is ON).
+   * </pre>
+   *
+   * <code>.tensorflow.RewriterConfig.Toggle implementation_selector = 22;</code>
+   */
+  org.tensorflow.proto.framework.RewriterConfig.Toggle getImplementationSelector();
+
+  /**
+   * <pre>
+   * Optimize data types for CUDA (default is OFF).
+   * This will try to use float16 on GPU which is faster.
+   * Note that this can change the numerical stability of the graph and may
+   * require the use of loss scaling to maintain model convergence.
+   * </pre>
+   *
+   * <code>.tensorflow.RewriterConfig.Toggle auto_mixed_precision = 23;</code>
+   */
+  int getAutoMixedPrecisionValue();
+  /**
+   * <pre>
+   * Optimize data types for CUDA (default is OFF).
+   * This will try to use float16 on GPU which is faster.
+   * Note that this can change the numerical stability of the graph and may
+   * require the use of loss scaling to maintain model convergence.
+   * </pre>
+   *
+   * <code>.tensorflow.RewriterConfig.Toggle auto_mixed_precision = 23;</code>
+   */
+  org.tensorflow.proto.framework.RewriterConfig.Toggle getAutoMixedPrecision();
+
+  /**
+   * <pre>
+   * Optimize data types for MKL (default is OFF).
+   * This will try to use bfloat16 on CPUs, which is faster.
+   * Note that this can change the numerical stability of the graph.
+   * </pre>
+   *
+   * <code>.tensorflow.RewriterConfig.Toggle auto_mixed_precision_mkl = 25;</code>
+   */
+  int getAutoMixedPrecisionMklValue();
+  /**
+   * <pre>
+   * Optimize data types for MKL (default is OFF).
+   * This will try to use bfloat16 on CPUs, which is faster.
+   * Note that this can change the numerical stability of the graph.
+   * </pre>
+   *
+   * <code>.tensorflow.RewriterConfig.Toggle auto_mixed_precision_mkl = 25;</code>
+   */
+  org.tensorflow.proto.framework.RewriterConfig.Toggle getAutoMixedPrecisionMkl();
+
+  /**
+   * <pre>
+   * Disable the entire meta optimizer (off by default).
+   * </pre>
+   *
+   * <code>bool disable_meta_optimizer = 19;</code>
+   */
+  boolean getDisableMetaOptimizer();
+
+  /**
+   * <pre>
+   * Controls how many times we run the optimizers in meta optimizer (default
+   * is once).
+   * </pre>
+   *
+   * <code>.tensorflow.RewriterConfig.NumIterationsType meta_optimizer_iterations = 12;</code>
+   */
+  int getMetaOptimizerIterationsValue();
+  /**
+   * <pre>
+   * Controls how many times we run the optimizers in meta optimizer (default
+   * is once).
+   * </pre>
+   *
+   * <code>.tensorflow.RewriterConfig.NumIterationsType meta_optimizer_iterations = 12;</code>
+   */
+  org.tensorflow.proto.framework.RewriterConfig.NumIterationsType getMetaOptimizerIterations();
+
+  /**
+   * <pre>
+   * The minimum number of nodes in a graph to optimizer. For smaller graphs,
+   * optimization is skipped.
+   * 0 means the system picks an appropriate number.
+   * &lt; 0 means do not skip optimization.
+   * </pre>
+   *
+   * <code>int32 min_graph_nodes = 17;</code>
+   */
+  int getMinGraphNodes();
+
+  /**
+   * <pre>
+   * Configures memory optimization passes through the meta-optimizer. Has no
+   * effect on manually requested memory optimization passes in the optimizers
+   * field.
+   * </pre>
+   *
+   * <code>.tensorflow.RewriterConfig.MemOptType memory_optimization = 4;</code>
+   */
+  int getMemoryOptimizationValue();
+  /**
+   * <pre>
+   * Configures memory optimization passes through the meta-optimizer. Has no
+   * effect on manually requested memory optimization passes in the optimizers
+   * field.
+   * </pre>
+   *
+   * <code>.tensorflow.RewriterConfig.MemOptType memory_optimization = 4;</code>
+   */
+  org.tensorflow.proto.framework.RewriterConfig.MemOptType getMemoryOptimization();
+
+  /**
+   * <pre>
+   * A node name scope for node names which are valid outputs of recomputations.
+   * Inputs to nodes that match this scope may be recomputed (subject either to
+   * manual annotation of those input nodes or to manual annotation and
+   * heuristics depending on memory_optimization), but the nodes themselves will
+   * not be recomputed. This matches any sub-scopes as well, meaning the scope
+   * can appear not just as a top-level scope. For example, if the value is
+   * "gradients/", the default, it will match node name "gradients/foo",
+   * "foo/gradients/bar", but not "foo_gradients/"
+   * </pre>
+   *
+   * <code>string memory_optimizer_target_node_name_scope = 6;</code>
+   */
+  java.lang.String getMemoryOptimizerTargetNodeNameScope();
+  /**
+   * <pre>
+   * A node name scope for node names which are valid outputs of recomputations.
+   * Inputs to nodes that match this scope may be recomputed (subject either to
+   * manual annotation of those input nodes or to manual annotation and
+   * heuristics depending on memory_optimization), but the nodes themselves will
+   * not be recomputed. This matches any sub-scopes as well, meaning the scope
+   * can appear not just as a top-level scope. For example, if the value is
+   * "gradients/", the default, it will match node name "gradients/foo",
+   * "foo/gradients/bar", but not "foo_gradients/"
+   * </pre>
+   *
+   * <code>string memory_optimizer_target_node_name_scope = 6;</code>
+   */
+  com.google.protobuf.ByteString
+      getMemoryOptimizerTargetNodeNameScopeBytes();
+
+  /**
+   * <pre>
+   * Maximum number of milliseconds to spend optimizing a single graph before
+   * timing out. If equal to 0 the system picks a default (currently 5 minutes).
+   * If less than 0 the optimizer will never time out.
+   * </pre>
+   *
+   * <code>int64 meta_optimizer_timeout_ms = 20;</code>
+   */
+  long getMetaOptimizerTimeoutMs();
+
+  /**
+   * <pre>
+   * Configures AutoParallel optimization passes either through the
+   * meta-optimizer or when manually specified through the optimizers field.
+   * </pre>
+   *
+   * <code>.tensorflow.AutoParallelOptions auto_parallel = 5;</code>
+   */
+  boolean hasAutoParallel();
+  /**
+   * <pre>
+   * Configures AutoParallel optimization passes either through the
+   * meta-optimizer or when manually specified through the optimizers field.
+   * </pre>
+   *
+   * <code>.tensorflow.AutoParallelOptions auto_parallel = 5;</code>
+   */
+  org.tensorflow.proto.framework.AutoParallelOptions getAutoParallel();
+  /**
+   * <pre>
+   * Configures AutoParallel optimization passes either through the
+   * meta-optimizer or when manually specified through the optimizers field.
+   * </pre>
+   *
+   * <code>.tensorflow.AutoParallelOptions auto_parallel = 5;</code>
+   */
+  org.tensorflow.proto.framework.AutoParallelOptionsOrBuilder getAutoParallelOrBuilder();
+
+  /**
+   * <pre>
+   * If true, any optimization pass failing will cause the MetaOptimizer to
+   * stop with an error. By default - or when set to false, failing passes are
+   * skipped silently.
+   * </pre>
+   *
+   * <code>bool fail_on_optimizer_errors = 21;</code>
+   */
+  boolean getFailOnOptimizerErrors();
+
+  /**
+   * <code>.tensorflow.ScopedAllocatorOptions scoped_allocator_opts = 16;</code>
+   */
+  boolean hasScopedAllocatorOpts();
+  /**
+   * <code>.tensorflow.ScopedAllocatorOptions scoped_allocator_opts = 16;</code>
+   */
+  org.tensorflow.proto.framework.ScopedAllocatorOptions getScopedAllocatorOpts();
+  /**
+   * <code>.tensorflow.ScopedAllocatorOptions scoped_allocator_opts = 16;</code>
+   */
+  org.tensorflow.proto.framework.ScopedAllocatorOptionsOrBuilder getScopedAllocatorOptsOrBuilder();
+
+  /**
+   * <pre>
+   * If non-empty, will use this as an alternative way to specify a list of
+   * optimizations to turn on and the order of the optimizations (replacing the
+   * meta-optimizer).
+   * Of the RewriterConfig options, only the AutoParallel configuration options
+   * (the auto_parallel field) apply to manually requested optimization passes
+   * ("autoparallel"). Memory optimization passes ("memory") invoked here are
+   * not configurable (in contrast to memory optimization passes through the
+   * meta-optimizer) and act only on manual op annotations.
+   * Custom optimizers (see custom_optimizers) that are not part of this
+   * schedule will be run after - in the order that they were specified.
+   * </pre>
+   *
+   * <code>repeated string optimizers = 100;</code>
+   */
+  java.util.List<java.lang.String>
+      getOptimizersList();
+  /**
+   * <pre>
+   * If non-empty, will use this as an alternative way to specify a list of
+   * optimizations to turn on and the order of the optimizations (replacing the
+   * meta-optimizer).
+   * Of the RewriterConfig options, only the AutoParallel configuration options
+   * (the auto_parallel field) apply to manually requested optimization passes
+   * ("autoparallel"). Memory optimization passes ("memory") invoked here are
+   * not configurable (in contrast to memory optimization passes through the
+   * meta-optimizer) and act only on manual op annotations.
+   * Custom optimizers (see custom_optimizers) that are not part of this
+   * schedule will be run after - in the order that they were specified.
+   * </pre>
+   *
+   * <code>repeated string optimizers = 100;</code>
+   */
+  int getOptimizersCount();
+  /**
+   * <pre>
+   * If non-empty, will use this as an alternative way to specify a list of
+   * optimizations to turn on and the order of the optimizations (replacing the
+   * meta-optimizer).
+   * Of the RewriterConfig options, only the AutoParallel configuration options
+   * (the auto_parallel field) apply to manually requested optimization passes
+   * ("autoparallel"). Memory optimization passes ("memory") invoked here are
+   * not configurable (in contrast to memory optimization passes through the
+   * meta-optimizer) and act only on manual op annotations.
+   * Custom optimizers (see custom_optimizers) that are not part of this
+   * schedule will be run after - in the order that they were specified.
+   * </pre>
+   *
+   * <code>repeated string optimizers = 100;</code>
+   */
+  java.lang.String getOptimizers(int index);
+  /**
+   * <pre>
+   * If non-empty, will use this as an alternative way to specify a list of
+   * optimizations to turn on and the order of the optimizations (replacing the
+   * meta-optimizer).
+   * Of the RewriterConfig options, only the AutoParallel configuration options
+   * (the auto_parallel field) apply to manually requested optimization passes
+   * ("autoparallel"). Memory optimization passes ("memory") invoked here are
+   * not configurable (in contrast to memory optimization passes through the
+   * meta-optimizer) and act only on manual op annotations.
+   * Custom optimizers (see custom_optimizers) that are not part of this
+   * schedule will be run after - in the order that they were specified.
+   * </pre>
+   *
+   * <code>repeated string optimizers = 100;</code>
+   */
+  com.google.protobuf.ByteString
+      getOptimizersBytes(int index);
+
+  /**
+   * <pre>
+   * list of CustomGraphOptimizers to apply.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.RewriterConfig.CustomGraphOptimizer custom_optimizers = 200;</code>
+   */
+  java.util.List<org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizer> 
+      getCustomOptimizersList();
+  /**
+   * <pre>
+   * list of CustomGraphOptimizers to apply.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.RewriterConfig.CustomGraphOptimizer custom_optimizers = 200;</code>
+   */
+  org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizer getCustomOptimizers(int index);
+  /**
+   * <pre>
+   * list of CustomGraphOptimizers to apply.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.RewriterConfig.CustomGraphOptimizer custom_optimizers = 200;</code>
+   */
+  int getCustomOptimizersCount();
+  /**
+   * <pre>
+   * list of CustomGraphOptimizers to apply.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.RewriterConfig.CustomGraphOptimizer custom_optimizers = 200;</code>
+   */
+  java.util.List<? extends org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizerOrBuilder> 
+      getCustomOptimizersOrBuilderList();
+  /**
+   * <pre>
+   * list of CustomGraphOptimizers to apply.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.RewriterConfig.CustomGraphOptimizer custom_optimizers = 200;</code>
+   */
+  org.tensorflow.proto.framework.RewriterConfig.CustomGraphOptimizerOrBuilder getCustomOptimizersOrBuilder(
+      int index);
+
+  /**
+   * <pre>
+   * VerifierConfig specifying the verifiers to be run after every optimizer.
+   * </pre>
+   *
+   * <code>.tensorflow.VerifierConfig inter_optimizer_verifier_config = 300;</code>
+   */
+  boolean hasInterOptimizerVerifierConfig();
+  /**
+   * <pre>
+   * VerifierConfig specifying the verifiers to be run after every optimizer.
+   * </pre>
+   *
+   * <code>.tensorflow.VerifierConfig inter_optimizer_verifier_config = 300;</code>
+   */
+  org.tensorflow.proto.framework.VerifierConfig getInterOptimizerVerifierConfig();
+  /**
+   * <pre>
+   * VerifierConfig specifying the verifiers to be run after every optimizer.
+   * </pre>
+   *
+   * <code>.tensorflow.VerifierConfig inter_optimizer_verifier_config = 300;</code>
+   */
+  org.tensorflow.proto.framework.VerifierConfigOrBuilder getInterOptimizerVerifierConfigOrBuilder();
+
+  /**
+   * <pre>
+   * VerifierConfig specifying the verifiers to be run at the end, after all
+   * optimizers have run.
+   * </pre>
+   *
+   * <code>.tensorflow.VerifierConfig post_optimization_verifier_config = 301;</code>
+   */
+  boolean hasPostOptimizationVerifierConfig();
+  /**
+   * <pre>
+   * VerifierConfig specifying the verifiers to be run at the end, after all
+   * optimizers have run.
+   * </pre>
+   *
+   * <code>.tensorflow.VerifierConfig post_optimization_verifier_config = 301;</code>
+   */
+  org.tensorflow.proto.framework.VerifierConfig getPostOptimizationVerifierConfig();
+  /**
+   * <pre>
+   * VerifierConfig specifying the verifiers to be run at the end, after all
+   * optimizers have run.
+   * </pre>
+   *
+   * <code>.tensorflow.VerifierConfig post_optimization_verifier_config = 301;</code>
+   */
+  org.tensorflow.proto.framework.VerifierConfigOrBuilder getPostOptimizationVerifierConfigOrBuilder();
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/RewriterConfigProtos.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/RewriterConfigProtos.java
new file mode 100644
index 00000000000..01debd9ea7f
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/RewriterConfigProtos.java
@@ -0,0 +1,159 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/rewriter_config.proto
+
+package org.tensorflow.proto.framework;
+
+public final class RewriterConfigProtos {
+  private RewriterConfigProtos() {}
+  public static void registerAllExtensions(
+      com.google.protobuf.ExtensionRegistryLite registry) {
+  }
+
+  public static void registerAllExtensions(
+      com.google.protobuf.ExtensionRegistry registry) {
+    registerAllExtensions(
+        (com.google.protobuf.ExtensionRegistryLite) registry);
+  }
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_AutoParallelOptions_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_AutoParallelOptions_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_ScopedAllocatorOptions_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_ScopedAllocatorOptions_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_RewriterConfig_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_RewriterConfig_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_RewriterConfig_CustomGraphOptimizer_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_RewriterConfig_CustomGraphOptimizer_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_RewriterConfig_CustomGraphOptimizer_ParameterMapEntry_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_RewriterConfig_CustomGraphOptimizer_ParameterMapEntry_fieldAccessorTable;
+
+  public static com.google.protobuf.Descriptors.FileDescriptor
+      getDescriptor() {
+    return descriptor;
+  }
+  private static  com.google.protobuf.Descriptors.FileDescriptor
+      descriptor;
+  static {
+    java.lang.String[] descriptorData = {
+      "\n.tensorflow/core/protobuf/rewriter_conf" +
+      "ig.proto\022\ntensorflow\032*tensorflow/core/fr" +
+      "amework/attr_value.proto\032.tensorflow/cor" +
+      "e/protobuf/verifier_config.proto\";\n\023Auto" +
+      "ParallelOptions\022\016\n\006enable\030\001 \001(\010\022\024\n\014num_r" +
+      "eplicas\030\002 \001(\005\"+\n\026ScopedAllocatorOptions\022" +
+      "\021\n\tenable_op\030\001 \003(\t\"\225\021\n\016RewriterConfig\022;\n" +
+      "\020layout_optimizer\030\001 \001(\0162!.tensorflow.Rew" +
+      "riterConfig.Toggle\022;\n\020constant_folding\030\003" +
+      " \001(\0162!.tensorflow.RewriterConfig.Toggle\022" +
+      "=\n\022shape_optimization\030\r \001(\0162!.tensorflow" +
+      ".RewriterConfig.Toggle\0224\n\tremapping\030\016 \001(" +
+      "\0162!.tensorflow.RewriterConfig.Toggle\022F\n\033" +
+      "common_subgraph_elimination\030\030 \001(\0162!.tens" +
+      "orflow.RewriterConfig.Toggle\022B\n\027arithmet" +
+      "ic_optimization\030\007 \001(\0162!.tensorflow.Rewri" +
+      "terConfig.Toggle\022B\n\027dependency_optimizat" +
+      "ion\030\010 \001(\0162!.tensorflow.RewriterConfig.To" +
+      "ggle\022<\n\021loop_optimization\030\t \001(\0162!.tensor" +
+      "flow.RewriterConfig.Toggle\022@\n\025function_o" +
+      "ptimization\030\n \001(\0162!.tensorflow.RewriterC" +
+      "onfig.Toggle\0229\n\016debug_stripper\030\013 \001(\0162!.t" +
+      "ensorflow.RewriterConfig.Toggle\022\035\n\025disab" +
+      "le_model_pruning\030\002 \001(\010\022H\n\035scoped_allocat" +
+      "or_optimization\030\017 \001(\0162!.tensorflow.Rewri" +
+      "terConfig.Toggle\022C\n\030pin_to_host_optimiza" +
+      "tion\030\022 \001(\0162!.tensorflow.RewriterConfig.T" +
+      "oggle\022B\n\027implementation_selector\030\026 \001(\0162!" +
+      ".tensorflow.RewriterConfig.Toggle\022?\n\024aut" +
+      "o_mixed_precision\030\027 \001(\0162!.tensorflow.Rew" +
+      "riterConfig.Toggle\022C\n\030auto_mixed_precisi" +
+      "on_mkl\030\031 \001(\0162!.tensorflow.RewriterConfig" +
+      ".Toggle\022\036\n\026disable_meta_optimizer\030\023 \001(\010\022" +
+      "O\n\031meta_optimizer_iterations\030\014 \001(\0162,.ten" +
+      "sorflow.RewriterConfig.NumIterationsType" +
+      "\022\027\n\017min_graph_nodes\030\021 \001(\005\022B\n\023memory_opti" +
+      "mization\030\004 \001(\0162%.tensorflow.RewriterConf" +
+      "ig.MemOptType\022/\n\'memory_optimizer_target" +
+      "_node_name_scope\030\006 \001(\t\022!\n\031meta_optimizer" +
+      "_timeout_ms\030\024 \001(\003\0226\n\rauto_parallel\030\005 \001(\013" +
+      "2\037.tensorflow.AutoParallelOptions\022 \n\030fai" +
+      "l_on_optimizer_errors\030\025 \001(\010\022A\n\025scoped_al" +
+      "locator_opts\030\020 \001(\0132\".tensorflow.ScopedAl" +
+      "locatorOptions\022\022\n\noptimizers\030d \003(\t\022K\n\021cu" +
+      "stom_optimizers\030\310\001 \003(\0132/.tensorflow.Rewr" +
+      "iterConfig.CustomGraphOptimizer\022D\n\037inter" +
+      "_optimizer_verifier_config\030\254\002 \001(\0132\032.tens" +
+      "orflow.VerifierConfig\022F\n!post_optimizati" +
+      "on_verifier_config\030\255\002 \001(\0132\032.tensorflow.V" +
+      "erifierConfig\032\312\001\n\024CustomGraphOptimizer\022\014" +
+      "\n\004name\030\001 \001(\t\022X\n\rparameter_map\030\002 \003(\0132A.te" +
+      "nsorflow.RewriterConfig.CustomGraphOptim" +
+      "izer.ParameterMapEntry\032J\n\021ParameterMapEn" +
+      "try\022\013\n\003key\030\001 \001(\t\022$\n\005value\030\002 \001(\0132\025.tensor" +
+      "flow.AttrValue:\0028\001\"6\n\006Toggle\022\013\n\007DEFAULT\020" +
+      "\000\022\006\n\002ON\020\001\022\007\n\003OFF\020\002\022\016\n\nAGGRESSIVE\020\003\"<\n\021Nu" +
+      "mIterationsType\022\025\n\021DEFAULT_NUM_ITERS\020\000\022\007" +
+      "\n\003ONE\020\001\022\007\n\003TWO\020\002\"\237\001\n\nMemOptType\022\023\n\017DEFAU" +
+      "LT_MEM_OPT\020\000\022\016\n\nNO_MEM_OPT\020\001\022\n\n\006MANUAL\020\002" +
+      "\022\027\n\023SWAPPING_HEURISTICS\020\004\022\034\n\030RECOMPUTATI" +
+      "ON_HEURISTICS\020\005\022\031\n\025SCHEDULING_HEURISTICS" +
+      "\020\006\022\016\n\nHEURISTICS\020\003B\205\001\n\036org.tensorflow.pr" +
+      "oto.frameworkB\024RewriterConfigProtosP\001ZHg" +
+      "ithub.com/tensorflow/tensorflow/tensorfl" +
+      "ow/go/core/core_protos_go_proto\370\001\001b\006prot" +
+      "o3"
+    };
+    descriptor = com.google.protobuf.Descriptors.FileDescriptor
+      .internalBuildGeneratedFileFrom(descriptorData,
+        new com.google.protobuf.Descriptors.FileDescriptor[] {
+          org.tensorflow.proto.framework.AttrValueProtos.getDescriptor(),
+          org.tensorflow.proto.framework.VerifierConfigProtos.getDescriptor(),
+        });
+    internal_static_tensorflow_AutoParallelOptions_descriptor =
+      getDescriptor().getMessageTypes().get(0);
+    internal_static_tensorflow_AutoParallelOptions_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_AutoParallelOptions_descriptor,
+        new java.lang.String[] { "Enable", "NumReplicas", });
+    internal_static_tensorflow_ScopedAllocatorOptions_descriptor =
+      getDescriptor().getMessageTypes().get(1);
+    internal_static_tensorflow_ScopedAllocatorOptions_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_ScopedAllocatorOptions_descriptor,
+        new java.lang.String[] { "EnableOp", });
+    internal_static_tensorflow_RewriterConfig_descriptor =
+      getDescriptor().getMessageTypes().get(2);
+    internal_static_tensorflow_RewriterConfig_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_RewriterConfig_descriptor,
+        new java.lang.String[] { "LayoutOptimizer", "ConstantFolding", "ShapeOptimization", "Remapping", "CommonSubgraphElimination", "ArithmeticOptimization", "DependencyOptimization", "LoopOptimization", "FunctionOptimization", "DebugStripper", "DisableModelPruning", "ScopedAllocatorOptimization", "PinToHostOptimization", "ImplementationSelector", "AutoMixedPrecision", "AutoMixedPrecisionMkl", "DisableMetaOptimizer", "MetaOptimizerIterations", "MinGraphNodes", "MemoryOptimization", "MemoryOptimizerTargetNodeNameScope", "MetaOptimizerTimeoutMs", "AutoParallel", "FailOnOptimizerErrors", "ScopedAllocatorOpts", "Optimizers", "CustomOptimizers", "InterOptimizerVerifierConfig", "PostOptimizationVerifierConfig", });
+    internal_static_tensorflow_RewriterConfig_CustomGraphOptimizer_descriptor =
+      internal_static_tensorflow_RewriterConfig_descriptor.getNestedTypes().get(0);
+    internal_static_tensorflow_RewriterConfig_CustomGraphOptimizer_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_RewriterConfig_CustomGraphOptimizer_descriptor,
+        new java.lang.String[] { "Name", "ParameterMap", });
+    internal_static_tensorflow_RewriterConfig_CustomGraphOptimizer_ParameterMapEntry_descriptor =
+      internal_static_tensorflow_RewriterConfig_CustomGraphOptimizer_descriptor.getNestedTypes().get(0);
+    internal_static_tensorflow_RewriterConfig_CustomGraphOptimizer_ParameterMapEntry_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_RewriterConfig_CustomGraphOptimizer_ParameterMapEntry_descriptor,
+        new java.lang.String[] { "Key", "Value", });
+    org.tensorflow.proto.framework.AttrValueProtos.getDescriptor();
+    org.tensorflow.proto.framework.VerifierConfigProtos.getDescriptor();
+  }
+
+  // @@protoc_insertion_point(outer_class_scope)
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/RunMetadata.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/RunMetadata.java
new file mode 100644
index 00000000000..47f4aa27082
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/RunMetadata.java
@@ -0,0 +1,3277 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/config.proto
+
+package org.tensorflow.proto.framework;
+
+/**
+ * <pre>
+ * Metadata output (i.e., non-Tensor) for a single Run() call.
+ * </pre>
+ *
+ * Protobuf type {@code tensorflow.RunMetadata}
+ */
+public  final class RunMetadata extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.RunMetadata)
+    RunMetadataOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use RunMetadata.newBuilder() to construct.
+  private RunMetadata(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private RunMetadata() {
+    partitionGraphs_ = java.util.Collections.emptyList();
+    functionGraphs_ = java.util.Collections.emptyList();
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new RunMetadata();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private RunMetadata(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    int mutable_bitField0_ = 0;
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 10: {
+            org.tensorflow.proto.framework.StepStats.Builder subBuilder = null;
+            if (stepStats_ != null) {
+              subBuilder = stepStats_.toBuilder();
+            }
+            stepStats_ = input.readMessage(org.tensorflow.proto.framework.StepStats.parser(), extensionRegistry);
+            if (subBuilder != null) {
+              subBuilder.mergeFrom(stepStats_);
+              stepStats_ = subBuilder.buildPartial();
+            }
+
+            break;
+          }
+          case 18: {
+            org.tensorflow.proto.framework.CostGraphDef.Builder subBuilder = null;
+            if (costGraph_ != null) {
+              subBuilder = costGraph_.toBuilder();
+            }
+            costGraph_ = input.readMessage(org.tensorflow.proto.framework.CostGraphDef.parser(), extensionRegistry);
+            if (subBuilder != null) {
+              subBuilder.mergeFrom(costGraph_);
+              costGraph_ = subBuilder.buildPartial();
+            }
+
+            break;
+          }
+          case 26: {
+            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
+              partitionGraphs_ = new java.util.ArrayList<org.tensorflow.proto.framework.GraphDef>();
+              mutable_bitField0_ |= 0x00000001;
+            }
+            partitionGraphs_.add(
+                input.readMessage(org.tensorflow.proto.framework.GraphDef.parser(), extensionRegistry));
+            break;
+          }
+          case 34: {
+            if (!((mutable_bitField0_ & 0x00000002) != 0)) {
+              functionGraphs_ = new java.util.ArrayList<org.tensorflow.proto.framework.RunMetadata.FunctionGraphs>();
+              mutable_bitField0_ |= 0x00000002;
+            }
+            functionGraphs_.add(
+                input.readMessage(org.tensorflow.proto.framework.RunMetadata.FunctionGraphs.parser(), extensionRegistry));
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      if (((mutable_bitField0_ & 0x00000001) != 0)) {
+        partitionGraphs_ = java.util.Collections.unmodifiableList(partitionGraphs_);
+      }
+      if (((mutable_bitField0_ & 0x00000002) != 0)) {
+        functionGraphs_ = java.util.Collections.unmodifiableList(functionGraphs_);
+      }
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_RunMetadata_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_RunMetadata_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.framework.RunMetadata.class, org.tensorflow.proto.framework.RunMetadata.Builder.class);
+  }
+
+  public interface FunctionGraphsOrBuilder extends
+      // @@protoc_insertion_point(interface_extends:tensorflow.RunMetadata.FunctionGraphs)
+      com.google.protobuf.MessageOrBuilder {
+
+    /**
+     * <pre>
+     * TODO(nareshmodi): Include some sort of function/cache-key identifier?
+     * </pre>
+     *
+     * <code>repeated .tensorflow.GraphDef partition_graphs = 1;</code>
+     */
+    java.util.List<org.tensorflow.proto.framework.GraphDef> 
+        getPartitionGraphsList();
+    /**
+     * <pre>
+     * TODO(nareshmodi): Include some sort of function/cache-key identifier?
+     * </pre>
+     *
+     * <code>repeated .tensorflow.GraphDef partition_graphs = 1;</code>
+     */
+    org.tensorflow.proto.framework.GraphDef getPartitionGraphs(int index);
+    /**
+     * <pre>
+     * TODO(nareshmodi): Include some sort of function/cache-key identifier?
+     * </pre>
+     *
+     * <code>repeated .tensorflow.GraphDef partition_graphs = 1;</code>
+     */
+    int getPartitionGraphsCount();
+    /**
+     * <pre>
+     * TODO(nareshmodi): Include some sort of function/cache-key identifier?
+     * </pre>
+     *
+     * <code>repeated .tensorflow.GraphDef partition_graphs = 1;</code>
+     */
+    java.util.List<? extends org.tensorflow.proto.framework.GraphDefOrBuilder> 
+        getPartitionGraphsOrBuilderList();
+    /**
+     * <pre>
+     * TODO(nareshmodi): Include some sort of function/cache-key identifier?
+     * </pre>
+     *
+     * <code>repeated .tensorflow.GraphDef partition_graphs = 1;</code>
+     */
+    org.tensorflow.proto.framework.GraphDefOrBuilder getPartitionGraphsOrBuilder(
+        int index);
+
+    /**
+     * <code>.tensorflow.GraphDef pre_optimization_graph = 2;</code>
+     */
+    boolean hasPreOptimizationGraph();
+    /**
+     * <code>.tensorflow.GraphDef pre_optimization_graph = 2;</code>
+     */
+    org.tensorflow.proto.framework.GraphDef getPreOptimizationGraph();
+    /**
+     * <code>.tensorflow.GraphDef pre_optimization_graph = 2;</code>
+     */
+    org.tensorflow.proto.framework.GraphDefOrBuilder getPreOptimizationGraphOrBuilder();
+
+    /**
+     * <code>.tensorflow.GraphDef post_optimization_graph = 3;</code>
+     */
+    boolean hasPostOptimizationGraph();
+    /**
+     * <code>.tensorflow.GraphDef post_optimization_graph = 3;</code>
+     */
+    org.tensorflow.proto.framework.GraphDef getPostOptimizationGraph();
+    /**
+     * <code>.tensorflow.GraphDef post_optimization_graph = 3;</code>
+     */
+    org.tensorflow.proto.framework.GraphDefOrBuilder getPostOptimizationGraphOrBuilder();
+  }
+  /**
+   * Protobuf type {@code tensorflow.RunMetadata.FunctionGraphs}
+   */
+  public  static final class FunctionGraphs extends
+      com.google.protobuf.GeneratedMessageV3 implements
+      // @@protoc_insertion_point(message_implements:tensorflow.RunMetadata.FunctionGraphs)
+      FunctionGraphsOrBuilder {
+  private static final long serialVersionUID = 0L;
+    // Use FunctionGraphs.newBuilder() to construct.
+    private FunctionGraphs(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+      super(builder);
+    }
+    private FunctionGraphs() {
+      partitionGraphs_ = java.util.Collections.emptyList();
+    }
+
+    @java.lang.Override
+    @SuppressWarnings({"unused"})
+    protected java.lang.Object newInstance(
+        UnusedPrivateParameter unused) {
+      return new FunctionGraphs();
+    }
+
+    @java.lang.Override
+    public final com.google.protobuf.UnknownFieldSet
+    getUnknownFields() {
+      return this.unknownFields;
+    }
+    private FunctionGraphs(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      this();
+      if (extensionRegistry == null) {
+        throw new java.lang.NullPointerException();
+      }
+      int mutable_bitField0_ = 0;
+      com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+          com.google.protobuf.UnknownFieldSet.newBuilder();
+      try {
+        boolean done = false;
+        while (!done) {
+          int tag = input.readTag();
+          switch (tag) {
+            case 0:
+              done = true;
+              break;
+            case 10: {
+              if (!((mutable_bitField0_ & 0x00000001) != 0)) {
+                partitionGraphs_ = new java.util.ArrayList<org.tensorflow.proto.framework.GraphDef>();
+                mutable_bitField0_ |= 0x00000001;
+              }
+              partitionGraphs_.add(
+                  input.readMessage(org.tensorflow.proto.framework.GraphDef.parser(), extensionRegistry));
+              break;
+            }
+            case 18: {
+              org.tensorflow.proto.framework.GraphDef.Builder subBuilder = null;
+              if (preOptimizationGraph_ != null) {
+                subBuilder = preOptimizationGraph_.toBuilder();
+              }
+              preOptimizationGraph_ = input.readMessage(org.tensorflow.proto.framework.GraphDef.parser(), extensionRegistry);
+              if (subBuilder != null) {
+                subBuilder.mergeFrom(preOptimizationGraph_);
+                preOptimizationGraph_ = subBuilder.buildPartial();
+              }
+
+              break;
+            }
+            case 26: {
+              org.tensorflow.proto.framework.GraphDef.Builder subBuilder = null;
+              if (postOptimizationGraph_ != null) {
+                subBuilder = postOptimizationGraph_.toBuilder();
+              }
+              postOptimizationGraph_ = input.readMessage(org.tensorflow.proto.framework.GraphDef.parser(), extensionRegistry);
+              if (subBuilder != null) {
+                subBuilder.mergeFrom(postOptimizationGraph_);
+                postOptimizationGraph_ = subBuilder.buildPartial();
+              }
+
+              break;
+            }
+            default: {
+              if (!parseUnknownField(
+                  input, unknownFields, extensionRegistry, tag)) {
+                done = true;
+              }
+              break;
+            }
+          }
+        }
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        throw e.setUnfinishedMessage(this);
+      } catch (java.io.IOException e) {
+        throw new com.google.protobuf.InvalidProtocolBufferException(
+            e).setUnfinishedMessage(this);
+      } finally {
+        if (((mutable_bitField0_ & 0x00000001) != 0)) {
+          partitionGraphs_ = java.util.Collections.unmodifiableList(partitionGraphs_);
+        }
+        this.unknownFields = unknownFields.build();
+        makeExtensionsImmutable();
+      }
+    }
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_RunMetadata_FunctionGraphs_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_RunMetadata_FunctionGraphs_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.RunMetadata.FunctionGraphs.class, org.tensorflow.proto.framework.RunMetadata.FunctionGraphs.Builder.class);
+    }
+
+    public static final int PARTITION_GRAPHS_FIELD_NUMBER = 1;
+    private java.util.List<org.tensorflow.proto.framework.GraphDef> partitionGraphs_;
+    /**
+     * <pre>
+     * TODO(nareshmodi): Include some sort of function/cache-key identifier?
+     * </pre>
+     *
+     * <code>repeated .tensorflow.GraphDef partition_graphs = 1;</code>
+     */
+    public java.util.List<org.tensorflow.proto.framework.GraphDef> getPartitionGraphsList() {
+      return partitionGraphs_;
+    }
+    /**
+     * <pre>
+     * TODO(nareshmodi): Include some sort of function/cache-key identifier?
+     * </pre>
+     *
+     * <code>repeated .tensorflow.GraphDef partition_graphs = 1;</code>
+     */
+    public java.util.List<? extends org.tensorflow.proto.framework.GraphDefOrBuilder> 
+        getPartitionGraphsOrBuilderList() {
+      return partitionGraphs_;
+    }
+    /**
+     * <pre>
+     * TODO(nareshmodi): Include some sort of function/cache-key identifier?
+     * </pre>
+     *
+     * <code>repeated .tensorflow.GraphDef partition_graphs = 1;</code>
+     */
+    public int getPartitionGraphsCount() {
+      return partitionGraphs_.size();
+    }
+    /**
+     * <pre>
+     * TODO(nareshmodi): Include some sort of function/cache-key identifier?
+     * </pre>
+     *
+     * <code>repeated .tensorflow.GraphDef partition_graphs = 1;</code>
+     */
+    public org.tensorflow.proto.framework.GraphDef getPartitionGraphs(int index) {
+      return partitionGraphs_.get(index);
+    }
+    /**
+     * <pre>
+     * TODO(nareshmodi): Include some sort of function/cache-key identifier?
+     * </pre>
+     *
+     * <code>repeated .tensorflow.GraphDef partition_graphs = 1;</code>
+     */
+    public org.tensorflow.proto.framework.GraphDefOrBuilder getPartitionGraphsOrBuilder(
+        int index) {
+      return partitionGraphs_.get(index);
+    }
+
+    public static final int PRE_OPTIMIZATION_GRAPH_FIELD_NUMBER = 2;
+    private org.tensorflow.proto.framework.GraphDef preOptimizationGraph_;
+    /**
+     * <code>.tensorflow.GraphDef pre_optimization_graph = 2;</code>
+     */
+    public boolean hasPreOptimizationGraph() {
+      return preOptimizationGraph_ != null;
+    }
+    /**
+     * <code>.tensorflow.GraphDef pre_optimization_graph = 2;</code>
+     */
+    public org.tensorflow.proto.framework.GraphDef getPreOptimizationGraph() {
+      return preOptimizationGraph_ == null ? org.tensorflow.proto.framework.GraphDef.getDefaultInstance() : preOptimizationGraph_;
+    }
+    /**
+     * <code>.tensorflow.GraphDef pre_optimization_graph = 2;</code>
+     */
+    public org.tensorflow.proto.framework.GraphDefOrBuilder getPreOptimizationGraphOrBuilder() {
+      return getPreOptimizationGraph();
+    }
+
+    public static final int POST_OPTIMIZATION_GRAPH_FIELD_NUMBER = 3;
+    private org.tensorflow.proto.framework.GraphDef postOptimizationGraph_;
+    /**
+     * <code>.tensorflow.GraphDef post_optimization_graph = 3;</code>
+     */
+    public boolean hasPostOptimizationGraph() {
+      return postOptimizationGraph_ != null;
+    }
+    /**
+     * <code>.tensorflow.GraphDef post_optimization_graph = 3;</code>
+     */
+    public org.tensorflow.proto.framework.GraphDef getPostOptimizationGraph() {
+      return postOptimizationGraph_ == null ? org.tensorflow.proto.framework.GraphDef.getDefaultInstance() : postOptimizationGraph_;
+    }
+    /**
+     * <code>.tensorflow.GraphDef post_optimization_graph = 3;</code>
+     */
+    public org.tensorflow.proto.framework.GraphDefOrBuilder getPostOptimizationGraphOrBuilder() {
+      return getPostOptimizationGraph();
+    }
+
+    private byte memoizedIsInitialized = -1;
+    @java.lang.Override
+    public final boolean isInitialized() {
+      byte isInitialized = memoizedIsInitialized;
+      if (isInitialized == 1) return true;
+      if (isInitialized == 0) return false;
+
+      memoizedIsInitialized = 1;
+      return true;
+    }
+
+    @java.lang.Override
+    public void writeTo(com.google.protobuf.CodedOutputStream output)
+                        throws java.io.IOException {
+      for (int i = 0; i < partitionGraphs_.size(); i++) {
+        output.writeMessage(1, partitionGraphs_.get(i));
+      }
+      if (preOptimizationGraph_ != null) {
+        output.writeMessage(2, getPreOptimizationGraph());
+      }
+      if (postOptimizationGraph_ != null) {
+        output.writeMessage(3, getPostOptimizationGraph());
+      }
+      unknownFields.writeTo(output);
+    }
+
+    @java.lang.Override
+    public int getSerializedSize() {
+      int size = memoizedSize;
+      if (size != -1) return size;
+
+      size = 0;
+      for (int i = 0; i < partitionGraphs_.size(); i++) {
+        size += com.google.protobuf.CodedOutputStream
+          .computeMessageSize(1, partitionGraphs_.get(i));
+      }
+      if (preOptimizationGraph_ != null) {
+        size += com.google.protobuf.CodedOutputStream
+          .computeMessageSize(2, getPreOptimizationGraph());
+      }
+      if (postOptimizationGraph_ != null) {
+        size += com.google.protobuf.CodedOutputStream
+          .computeMessageSize(3, getPostOptimizationGraph());
+      }
+      size += unknownFields.getSerializedSize();
+      memoizedSize = size;
+      return size;
+    }
+
+    @java.lang.Override
+    public boolean equals(final java.lang.Object obj) {
+      if (obj == this) {
+       return true;
+      }
+      if (!(obj instanceof org.tensorflow.proto.framework.RunMetadata.FunctionGraphs)) {
+        return super.equals(obj);
+      }
+      org.tensorflow.proto.framework.RunMetadata.FunctionGraphs other = (org.tensorflow.proto.framework.RunMetadata.FunctionGraphs) obj;
+
+      if (!getPartitionGraphsList()
+          .equals(other.getPartitionGraphsList())) return false;
+      if (hasPreOptimizationGraph() != other.hasPreOptimizationGraph()) return false;
+      if (hasPreOptimizationGraph()) {
+        if (!getPreOptimizationGraph()
+            .equals(other.getPreOptimizationGraph())) return false;
+      }
+      if (hasPostOptimizationGraph() != other.hasPostOptimizationGraph()) return false;
+      if (hasPostOptimizationGraph()) {
+        if (!getPostOptimizationGraph()
+            .equals(other.getPostOptimizationGraph())) return false;
+      }
+      if (!unknownFields.equals(other.unknownFields)) return false;
+      return true;
+    }
+
+    @java.lang.Override
+    public int hashCode() {
+      if (memoizedHashCode != 0) {
+        return memoizedHashCode;
+      }
+      int hash = 41;
+      hash = (19 * hash) + getDescriptor().hashCode();
+      if (getPartitionGraphsCount() > 0) {
+        hash = (37 * hash) + PARTITION_GRAPHS_FIELD_NUMBER;
+        hash = (53 * hash) + getPartitionGraphsList().hashCode();
+      }
+      if (hasPreOptimizationGraph()) {
+        hash = (37 * hash) + PRE_OPTIMIZATION_GRAPH_FIELD_NUMBER;
+        hash = (53 * hash) + getPreOptimizationGraph().hashCode();
+      }
+      if (hasPostOptimizationGraph()) {
+        hash = (37 * hash) + POST_OPTIMIZATION_GRAPH_FIELD_NUMBER;
+        hash = (53 * hash) + getPostOptimizationGraph().hashCode();
+      }
+      hash = (29 * hash) + unknownFields.hashCode();
+      memoizedHashCode = hash;
+      return hash;
+    }
+
+    public static org.tensorflow.proto.framework.RunMetadata.FunctionGraphs parseFrom(
+        java.nio.ByteBuffer data)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data);
+    }
+    public static org.tensorflow.proto.framework.RunMetadata.FunctionGraphs parseFrom(
+        java.nio.ByteBuffer data,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.RunMetadata.FunctionGraphs parseFrom(
+        com.google.protobuf.ByteString data)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data);
+    }
+    public static org.tensorflow.proto.framework.RunMetadata.FunctionGraphs parseFrom(
+        com.google.protobuf.ByteString data,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.RunMetadata.FunctionGraphs parseFrom(byte[] data)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data);
+    }
+    public static org.tensorflow.proto.framework.RunMetadata.FunctionGraphs parseFrom(
+        byte[] data,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.RunMetadata.FunctionGraphs parseFrom(java.io.InputStream input)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input);
+    }
+    public static org.tensorflow.proto.framework.RunMetadata.FunctionGraphs parseFrom(
+        java.io.InputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.RunMetadata.FunctionGraphs parseDelimitedFrom(java.io.InputStream input)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseDelimitedWithIOException(PARSER, input);
+    }
+    public static org.tensorflow.proto.framework.RunMetadata.FunctionGraphs parseDelimitedFrom(
+        java.io.InputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.RunMetadata.FunctionGraphs parseFrom(
+        com.google.protobuf.CodedInputStream input)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input);
+    }
+    public static org.tensorflow.proto.framework.RunMetadata.FunctionGraphs parseFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input, extensionRegistry);
+    }
+
+    @java.lang.Override
+    public Builder newBuilderForType() { return newBuilder(); }
+    public static Builder newBuilder() {
+      return DEFAULT_INSTANCE.toBuilder();
+    }
+    public static Builder newBuilder(org.tensorflow.proto.framework.RunMetadata.FunctionGraphs prototype) {
+      return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+    }
+    @java.lang.Override
+    public Builder toBuilder() {
+      return this == DEFAULT_INSTANCE
+          ? new Builder() : new Builder().mergeFrom(this);
+    }
+
+    @java.lang.Override
+    protected Builder newBuilderForType(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      Builder builder = new Builder(parent);
+      return builder;
+    }
+    /**
+     * Protobuf type {@code tensorflow.RunMetadata.FunctionGraphs}
+     */
+    public static final class Builder extends
+        com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+        // @@protoc_insertion_point(builder_implements:tensorflow.RunMetadata.FunctionGraphs)
+        org.tensorflow.proto.framework.RunMetadata.FunctionGraphsOrBuilder {
+      public static final com.google.protobuf.Descriptors.Descriptor
+          getDescriptor() {
+        return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_RunMetadata_FunctionGraphs_descriptor;
+      }
+
+      @java.lang.Override
+      protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+          internalGetFieldAccessorTable() {
+        return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_RunMetadata_FunctionGraphs_fieldAccessorTable
+            .ensureFieldAccessorsInitialized(
+                org.tensorflow.proto.framework.RunMetadata.FunctionGraphs.class, org.tensorflow.proto.framework.RunMetadata.FunctionGraphs.Builder.class);
+      }
+
+      // Construct using org.tensorflow.proto.framework.RunMetadata.FunctionGraphs.newBuilder()
+      private Builder() {
+        maybeForceBuilderInitialization();
+      }
+
+      private Builder(
+          com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+        super(parent);
+        maybeForceBuilderInitialization();
+      }
+      private void maybeForceBuilderInitialization() {
+        if (com.google.protobuf.GeneratedMessageV3
+                .alwaysUseFieldBuilders) {
+          getPartitionGraphsFieldBuilder();
+        }
+      }
+      @java.lang.Override
+      public Builder clear() {
+        super.clear();
+        if (partitionGraphsBuilder_ == null) {
+          partitionGraphs_ = java.util.Collections.emptyList();
+          bitField0_ = (bitField0_ & ~0x00000001);
+        } else {
+          partitionGraphsBuilder_.clear();
+        }
+        if (preOptimizationGraphBuilder_ == null) {
+          preOptimizationGraph_ = null;
+        } else {
+          preOptimizationGraph_ = null;
+          preOptimizationGraphBuilder_ = null;
+        }
+        if (postOptimizationGraphBuilder_ == null) {
+          postOptimizationGraph_ = null;
+        } else {
+          postOptimizationGraph_ = null;
+          postOptimizationGraphBuilder_ = null;
+        }
+        return this;
+      }
+
+      @java.lang.Override
+      public com.google.protobuf.Descriptors.Descriptor
+          getDescriptorForType() {
+        return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_RunMetadata_FunctionGraphs_descriptor;
+      }
+
+      @java.lang.Override
+      public org.tensorflow.proto.framework.RunMetadata.FunctionGraphs getDefaultInstanceForType() {
+        return org.tensorflow.proto.framework.RunMetadata.FunctionGraphs.getDefaultInstance();
+      }
+
+      @java.lang.Override
+      public org.tensorflow.proto.framework.RunMetadata.FunctionGraphs build() {
+        org.tensorflow.proto.framework.RunMetadata.FunctionGraphs result = buildPartial();
+        if (!result.isInitialized()) {
+          throw newUninitializedMessageException(result);
+        }
+        return result;
+      }
+
+      @java.lang.Override
+      public org.tensorflow.proto.framework.RunMetadata.FunctionGraphs buildPartial() {
+        org.tensorflow.proto.framework.RunMetadata.FunctionGraphs result = new org.tensorflow.proto.framework.RunMetadata.FunctionGraphs(this);
+        int from_bitField0_ = bitField0_;
+        if (partitionGraphsBuilder_ == null) {
+          if (((bitField0_ & 0x00000001) != 0)) {
+            partitionGraphs_ = java.util.Collections.unmodifiableList(partitionGraphs_);
+            bitField0_ = (bitField0_ & ~0x00000001);
+          }
+          result.partitionGraphs_ = partitionGraphs_;
+        } else {
+          result.partitionGraphs_ = partitionGraphsBuilder_.build();
+        }
+        if (preOptimizationGraphBuilder_ == null) {
+          result.preOptimizationGraph_ = preOptimizationGraph_;
+        } else {
+          result.preOptimizationGraph_ = preOptimizationGraphBuilder_.build();
+        }
+        if (postOptimizationGraphBuilder_ == null) {
+          result.postOptimizationGraph_ = postOptimizationGraph_;
+        } else {
+          result.postOptimizationGraph_ = postOptimizationGraphBuilder_.build();
+        }
+        onBuilt();
+        return result;
+      }
+
+      @java.lang.Override
+      public Builder clone() {
+        return super.clone();
+      }
+      @java.lang.Override
+      public Builder setField(
+          com.google.protobuf.Descriptors.FieldDescriptor field,
+          java.lang.Object value) {
+        return super.setField(field, value);
+      }
+      @java.lang.Override
+      public Builder clearField(
+          com.google.protobuf.Descriptors.FieldDescriptor field) {
+        return super.clearField(field);
+      }
+      @java.lang.Override
+      public Builder clearOneof(
+          com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+        return super.clearOneof(oneof);
+      }
+      @java.lang.Override
+      public Builder setRepeatedField(
+          com.google.protobuf.Descriptors.FieldDescriptor field,
+          int index, java.lang.Object value) {
+        return super.setRepeatedField(field, index, value);
+      }
+      @java.lang.Override
+      public Builder addRepeatedField(
+          com.google.protobuf.Descriptors.FieldDescriptor field,
+          java.lang.Object value) {
+        return super.addRepeatedField(field, value);
+      }
+      @java.lang.Override
+      public Builder mergeFrom(com.google.protobuf.Message other) {
+        if (other instanceof org.tensorflow.proto.framework.RunMetadata.FunctionGraphs) {
+          return mergeFrom((org.tensorflow.proto.framework.RunMetadata.FunctionGraphs)other);
+        } else {
+          super.mergeFrom(other);
+          return this;
+        }
+      }
+
+      public Builder mergeFrom(org.tensorflow.proto.framework.RunMetadata.FunctionGraphs other) {
+        if (other == org.tensorflow.proto.framework.RunMetadata.FunctionGraphs.getDefaultInstance()) return this;
+        if (partitionGraphsBuilder_ == null) {
+          if (!other.partitionGraphs_.isEmpty()) {
+            if (partitionGraphs_.isEmpty()) {
+              partitionGraphs_ = other.partitionGraphs_;
+              bitField0_ = (bitField0_ & ~0x00000001);
+            } else {
+              ensurePartitionGraphsIsMutable();
+              partitionGraphs_.addAll(other.partitionGraphs_);
+            }
+            onChanged();
+          }
+        } else {
+          if (!other.partitionGraphs_.isEmpty()) {
+            if (partitionGraphsBuilder_.isEmpty()) {
+              partitionGraphsBuilder_.dispose();
+              partitionGraphsBuilder_ = null;
+              partitionGraphs_ = other.partitionGraphs_;
+              bitField0_ = (bitField0_ & ~0x00000001);
+              partitionGraphsBuilder_ = 
+                com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
+                   getPartitionGraphsFieldBuilder() : null;
+            } else {
+              partitionGraphsBuilder_.addAllMessages(other.partitionGraphs_);
+            }
+          }
+        }
+        if (other.hasPreOptimizationGraph()) {
+          mergePreOptimizationGraph(other.getPreOptimizationGraph());
+        }
+        if (other.hasPostOptimizationGraph()) {
+          mergePostOptimizationGraph(other.getPostOptimizationGraph());
+        }
+        this.mergeUnknownFields(other.unknownFields);
+        onChanged();
+        return this;
+      }
+
+      @java.lang.Override
+      public final boolean isInitialized() {
+        return true;
+      }
+
+      @java.lang.Override
+      public Builder mergeFrom(
+          com.google.protobuf.CodedInputStream input,
+          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+          throws java.io.IOException {
+        org.tensorflow.proto.framework.RunMetadata.FunctionGraphs parsedMessage = null;
+        try {
+          parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+        } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+          parsedMessage = (org.tensorflow.proto.framework.RunMetadata.FunctionGraphs) e.getUnfinishedMessage();
+          throw e.unwrapIOException();
+        } finally {
+          if (parsedMessage != null) {
+            mergeFrom(parsedMessage);
+          }
+        }
+        return this;
+      }
+      private int bitField0_;
+
+      private java.util.List<org.tensorflow.proto.framework.GraphDef> partitionGraphs_ =
+        java.util.Collections.emptyList();
+      private void ensurePartitionGraphsIsMutable() {
+        if (!((bitField0_ & 0x00000001) != 0)) {
+          partitionGraphs_ = new java.util.ArrayList<org.tensorflow.proto.framework.GraphDef>(partitionGraphs_);
+          bitField0_ |= 0x00000001;
+         }
+      }
+
+      private com.google.protobuf.RepeatedFieldBuilderV3<
+          org.tensorflow.proto.framework.GraphDef, org.tensorflow.proto.framework.GraphDef.Builder, org.tensorflow.proto.framework.GraphDefOrBuilder> partitionGraphsBuilder_;
+
+      /**
+       * <pre>
+       * TODO(nareshmodi): Include some sort of function/cache-key identifier?
+       * </pre>
+       *
+       * <code>repeated .tensorflow.GraphDef partition_graphs = 1;</code>
+       */
+      public java.util.List<org.tensorflow.proto.framework.GraphDef> getPartitionGraphsList() {
+        if (partitionGraphsBuilder_ == null) {
+          return java.util.Collections.unmodifiableList(partitionGraphs_);
+        } else {
+          return partitionGraphsBuilder_.getMessageList();
+        }
+      }
+      /**
+       * <pre>
+       * TODO(nareshmodi): Include some sort of function/cache-key identifier?
+       * </pre>
+       *
+       * <code>repeated .tensorflow.GraphDef partition_graphs = 1;</code>
+       */
+      public int getPartitionGraphsCount() {
+        if (partitionGraphsBuilder_ == null) {
+          return partitionGraphs_.size();
+        } else {
+          return partitionGraphsBuilder_.getCount();
+        }
+      }
+      /**
+       * <pre>
+       * TODO(nareshmodi): Include some sort of function/cache-key identifier?
+       * </pre>
+       *
+       * <code>repeated .tensorflow.GraphDef partition_graphs = 1;</code>
+       */
+      public org.tensorflow.proto.framework.GraphDef getPartitionGraphs(int index) {
+        if (partitionGraphsBuilder_ == null) {
+          return partitionGraphs_.get(index);
+        } else {
+          return partitionGraphsBuilder_.getMessage(index);
+        }
+      }
+      /**
+       * <pre>
+       * TODO(nareshmodi): Include some sort of function/cache-key identifier?
+       * </pre>
+       *
+       * <code>repeated .tensorflow.GraphDef partition_graphs = 1;</code>
+       */
+      public Builder setPartitionGraphs(
+          int index, org.tensorflow.proto.framework.GraphDef value) {
+        if (partitionGraphsBuilder_ == null) {
+          if (value == null) {
+            throw new NullPointerException();
+          }
+          ensurePartitionGraphsIsMutable();
+          partitionGraphs_.set(index, value);
+          onChanged();
+        } else {
+          partitionGraphsBuilder_.setMessage(index, value);
+        }
+        return this;
+      }
+      /**
+       * <pre>
+       * TODO(nareshmodi): Include some sort of function/cache-key identifier?
+       * </pre>
+       *
+       * <code>repeated .tensorflow.GraphDef partition_graphs = 1;</code>
+       */
+      public Builder setPartitionGraphs(
+          int index, org.tensorflow.proto.framework.GraphDef.Builder builderForValue) {
+        if (partitionGraphsBuilder_ == null) {
+          ensurePartitionGraphsIsMutable();
+          partitionGraphs_.set(index, builderForValue.build());
+          onChanged();
+        } else {
+          partitionGraphsBuilder_.setMessage(index, builderForValue.build());
+        }
+        return this;
+      }
+      /**
+       * <pre>
+       * TODO(nareshmodi): Include some sort of function/cache-key identifier?
+       * </pre>
+       *
+       * <code>repeated .tensorflow.GraphDef partition_graphs = 1;</code>
+       */
+      public Builder addPartitionGraphs(org.tensorflow.proto.framework.GraphDef value) {
+        if (partitionGraphsBuilder_ == null) {
+          if (value == null) {
+            throw new NullPointerException();
+          }
+          ensurePartitionGraphsIsMutable();
+          partitionGraphs_.add(value);
+          onChanged();
+        } else {
+          partitionGraphsBuilder_.addMessage(value);
+        }
+        return this;
+      }
+      /**
+       * <pre>
+       * TODO(nareshmodi): Include some sort of function/cache-key identifier?
+       * </pre>
+       *
+       * <code>repeated .tensorflow.GraphDef partition_graphs = 1;</code>
+       */
+      public Builder addPartitionGraphs(
+          int index, org.tensorflow.proto.framework.GraphDef value) {
+        if (partitionGraphsBuilder_ == null) {
+          if (value == null) {
+            throw new NullPointerException();
+          }
+          ensurePartitionGraphsIsMutable();
+          partitionGraphs_.add(index, value);
+          onChanged();
+        } else {
+          partitionGraphsBuilder_.addMessage(index, value);
+        }
+        return this;
+      }
+      /**
+       * <pre>
+       * TODO(nareshmodi): Include some sort of function/cache-key identifier?
+       * </pre>
+       *
+       * <code>repeated .tensorflow.GraphDef partition_graphs = 1;</code>
+       */
+      public Builder addPartitionGraphs(
+          org.tensorflow.proto.framework.GraphDef.Builder builderForValue) {
+        if (partitionGraphsBuilder_ == null) {
+          ensurePartitionGraphsIsMutable();
+          partitionGraphs_.add(builderForValue.build());
+          onChanged();
+        } else {
+          partitionGraphsBuilder_.addMessage(builderForValue.build());
+        }
+        return this;
+      }
+      /**
+       * <pre>
+       * TODO(nareshmodi): Include some sort of function/cache-key identifier?
+       * </pre>
+       *
+       * <code>repeated .tensorflow.GraphDef partition_graphs = 1;</code>
+       */
+      public Builder addPartitionGraphs(
+          int index, org.tensorflow.proto.framework.GraphDef.Builder builderForValue) {
+        if (partitionGraphsBuilder_ == null) {
+          ensurePartitionGraphsIsMutable();
+          partitionGraphs_.add(index, builderForValue.build());
+          onChanged();
+        } else {
+          partitionGraphsBuilder_.addMessage(index, builderForValue.build());
+        }
+        return this;
+      }
+      /**
+       * <pre>
+       * TODO(nareshmodi): Include some sort of function/cache-key identifier?
+       * </pre>
+       *
+       * <code>repeated .tensorflow.GraphDef partition_graphs = 1;</code>
+       */
+      public Builder addAllPartitionGraphs(
+          java.lang.Iterable<? extends org.tensorflow.proto.framework.GraphDef> values) {
+        if (partitionGraphsBuilder_ == null) {
+          ensurePartitionGraphsIsMutable();
+          com.google.protobuf.AbstractMessageLite.Builder.addAll(
+              values, partitionGraphs_);
+          onChanged();
+        } else {
+          partitionGraphsBuilder_.addAllMessages(values);
+        }
+        return this;
+      }
+      /**
+       * <pre>
+       * TODO(nareshmodi): Include some sort of function/cache-key identifier?
+       * </pre>
+       *
+       * <code>repeated .tensorflow.GraphDef partition_graphs = 1;</code>
+       */
+      public Builder clearPartitionGraphs() {
+        if (partitionGraphsBuilder_ == null) {
+          partitionGraphs_ = java.util.Collections.emptyList();
+          bitField0_ = (bitField0_ & ~0x00000001);
+          onChanged();
+        } else {
+          partitionGraphsBuilder_.clear();
+        }
+        return this;
+      }
+      /**
+       * <pre>
+       * TODO(nareshmodi): Include some sort of function/cache-key identifier?
+       * </pre>
+       *
+       * <code>repeated .tensorflow.GraphDef partition_graphs = 1;</code>
+       */
+      public Builder removePartitionGraphs(int index) {
+        if (partitionGraphsBuilder_ == null) {
+          ensurePartitionGraphsIsMutable();
+          partitionGraphs_.remove(index);
+          onChanged();
+        } else {
+          partitionGraphsBuilder_.remove(index);
+        }
+        return this;
+      }
+      /**
+       * <pre>
+       * TODO(nareshmodi): Include some sort of function/cache-key identifier?
+       * </pre>
+       *
+       * <code>repeated .tensorflow.GraphDef partition_graphs = 1;</code>
+       */
+      public org.tensorflow.proto.framework.GraphDef.Builder getPartitionGraphsBuilder(
+          int index) {
+        return getPartitionGraphsFieldBuilder().getBuilder(index);
+      }
+      /**
+       * <pre>
+       * TODO(nareshmodi): Include some sort of function/cache-key identifier?
+       * </pre>
+       *
+       * <code>repeated .tensorflow.GraphDef partition_graphs = 1;</code>
+       */
+      public org.tensorflow.proto.framework.GraphDefOrBuilder getPartitionGraphsOrBuilder(
+          int index) {
+        if (partitionGraphsBuilder_ == null) {
+          return partitionGraphs_.get(index);  } else {
+          return partitionGraphsBuilder_.getMessageOrBuilder(index);
+        }
+      }
+      /**
+       * <pre>
+       * TODO(nareshmodi): Include some sort of function/cache-key identifier?
+       * </pre>
+       *
+       * <code>repeated .tensorflow.GraphDef partition_graphs = 1;</code>
+       */
+      public java.util.List<? extends org.tensorflow.proto.framework.GraphDefOrBuilder> 
+           getPartitionGraphsOrBuilderList() {
+        if (partitionGraphsBuilder_ != null) {
+          return partitionGraphsBuilder_.getMessageOrBuilderList();
+        } else {
+          return java.util.Collections.unmodifiableList(partitionGraphs_);
+        }
+      }
+      /**
+       * <pre>
+       * TODO(nareshmodi): Include some sort of function/cache-key identifier?
+       * </pre>
+       *
+       * <code>repeated .tensorflow.GraphDef partition_graphs = 1;</code>
+       */
+      public org.tensorflow.proto.framework.GraphDef.Builder addPartitionGraphsBuilder() {
+        return getPartitionGraphsFieldBuilder().addBuilder(
+            org.tensorflow.proto.framework.GraphDef.getDefaultInstance());
+      }
+      /**
+       * <pre>
+       * TODO(nareshmodi): Include some sort of function/cache-key identifier?
+       * </pre>
+       *
+       * <code>repeated .tensorflow.GraphDef partition_graphs = 1;</code>
+       */
+      public org.tensorflow.proto.framework.GraphDef.Builder addPartitionGraphsBuilder(
+          int index) {
+        return getPartitionGraphsFieldBuilder().addBuilder(
+            index, org.tensorflow.proto.framework.GraphDef.getDefaultInstance());
+      }
+      /**
+       * <pre>
+       * TODO(nareshmodi): Include some sort of function/cache-key identifier?
+       * </pre>
+       *
+       * <code>repeated .tensorflow.GraphDef partition_graphs = 1;</code>
+       */
+      public java.util.List<org.tensorflow.proto.framework.GraphDef.Builder> 
+           getPartitionGraphsBuilderList() {
+        return getPartitionGraphsFieldBuilder().getBuilderList();
+      }
+      private com.google.protobuf.RepeatedFieldBuilderV3<
+          org.tensorflow.proto.framework.GraphDef, org.tensorflow.proto.framework.GraphDef.Builder, org.tensorflow.proto.framework.GraphDefOrBuilder> 
+          getPartitionGraphsFieldBuilder() {
+        if (partitionGraphsBuilder_ == null) {
+          partitionGraphsBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
+              org.tensorflow.proto.framework.GraphDef, org.tensorflow.proto.framework.GraphDef.Builder, org.tensorflow.proto.framework.GraphDefOrBuilder>(
+                  partitionGraphs_,
+                  ((bitField0_ & 0x00000001) != 0),
+                  getParentForChildren(),
+                  isClean());
+          partitionGraphs_ = null;
+        }
+        return partitionGraphsBuilder_;
+      }
+
+      private org.tensorflow.proto.framework.GraphDef preOptimizationGraph_;
+      private com.google.protobuf.SingleFieldBuilderV3<
+          org.tensorflow.proto.framework.GraphDef, org.tensorflow.proto.framework.GraphDef.Builder, org.tensorflow.proto.framework.GraphDefOrBuilder> preOptimizationGraphBuilder_;
+      /**
+       * <code>.tensorflow.GraphDef pre_optimization_graph = 2;</code>
+       */
+      public boolean hasPreOptimizationGraph() {
+        return preOptimizationGraphBuilder_ != null || preOptimizationGraph_ != null;
+      }
+      /**
+       * <code>.tensorflow.GraphDef pre_optimization_graph = 2;</code>
+       */
+      public org.tensorflow.proto.framework.GraphDef getPreOptimizationGraph() {
+        if (preOptimizationGraphBuilder_ == null) {
+          return preOptimizationGraph_ == null ? org.tensorflow.proto.framework.GraphDef.getDefaultInstance() : preOptimizationGraph_;
+        } else {
+          return preOptimizationGraphBuilder_.getMessage();
+        }
+      }
+      /**
+       * <code>.tensorflow.GraphDef pre_optimization_graph = 2;</code>
+       */
+      public Builder setPreOptimizationGraph(org.tensorflow.proto.framework.GraphDef value) {
+        if (preOptimizationGraphBuilder_ == null) {
+          if (value == null) {
+            throw new NullPointerException();
+          }
+          preOptimizationGraph_ = value;
+          onChanged();
+        } else {
+          preOptimizationGraphBuilder_.setMessage(value);
+        }
+
+        return this;
+      }
+      /**
+       * <code>.tensorflow.GraphDef pre_optimization_graph = 2;</code>
+       */
+      public Builder setPreOptimizationGraph(
+          org.tensorflow.proto.framework.GraphDef.Builder builderForValue) {
+        if (preOptimizationGraphBuilder_ == null) {
+          preOptimizationGraph_ = builderForValue.build();
+          onChanged();
+        } else {
+          preOptimizationGraphBuilder_.setMessage(builderForValue.build());
+        }
+
+        return this;
+      }
+      /**
+       * <code>.tensorflow.GraphDef pre_optimization_graph = 2;</code>
+       */
+      public Builder mergePreOptimizationGraph(org.tensorflow.proto.framework.GraphDef value) {
+        if (preOptimizationGraphBuilder_ == null) {
+          if (preOptimizationGraph_ != null) {
+            preOptimizationGraph_ =
+              org.tensorflow.proto.framework.GraphDef.newBuilder(preOptimizationGraph_).mergeFrom(value).buildPartial();
+          } else {
+            preOptimizationGraph_ = value;
+          }
+          onChanged();
+        } else {
+          preOptimizationGraphBuilder_.mergeFrom(value);
+        }
+
+        return this;
+      }
+      /**
+       * <code>.tensorflow.GraphDef pre_optimization_graph = 2;</code>
+       */
+      public Builder clearPreOptimizationGraph() {
+        if (preOptimizationGraphBuilder_ == null) {
+          preOptimizationGraph_ = null;
+          onChanged();
+        } else {
+          preOptimizationGraph_ = null;
+          preOptimizationGraphBuilder_ = null;
+        }
+
+        return this;
+      }
+      /**
+       * <code>.tensorflow.GraphDef pre_optimization_graph = 2;</code>
+       */
+      public org.tensorflow.proto.framework.GraphDef.Builder getPreOptimizationGraphBuilder() {
+        
+        onChanged();
+        return getPreOptimizationGraphFieldBuilder().getBuilder();
+      }
+      /**
+       * <code>.tensorflow.GraphDef pre_optimization_graph = 2;</code>
+       */
+      public org.tensorflow.proto.framework.GraphDefOrBuilder getPreOptimizationGraphOrBuilder() {
+        if (preOptimizationGraphBuilder_ != null) {
+          return preOptimizationGraphBuilder_.getMessageOrBuilder();
+        } else {
+          return preOptimizationGraph_ == null ?
+              org.tensorflow.proto.framework.GraphDef.getDefaultInstance() : preOptimizationGraph_;
+        }
+      }
+      /**
+       * <code>.tensorflow.GraphDef pre_optimization_graph = 2;</code>
+       */
+      private com.google.protobuf.SingleFieldBuilderV3<
+          org.tensorflow.proto.framework.GraphDef, org.tensorflow.proto.framework.GraphDef.Builder, org.tensorflow.proto.framework.GraphDefOrBuilder> 
+          getPreOptimizationGraphFieldBuilder() {
+        if (preOptimizationGraphBuilder_ == null) {
+          preOptimizationGraphBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+              org.tensorflow.proto.framework.GraphDef, org.tensorflow.proto.framework.GraphDef.Builder, org.tensorflow.proto.framework.GraphDefOrBuilder>(
+                  getPreOptimizationGraph(),
+                  getParentForChildren(),
+                  isClean());
+          preOptimizationGraph_ = null;
+        }
+        return preOptimizationGraphBuilder_;
+      }
+
+      private org.tensorflow.proto.framework.GraphDef postOptimizationGraph_;
+      private com.google.protobuf.SingleFieldBuilderV3<
+          org.tensorflow.proto.framework.GraphDef, org.tensorflow.proto.framework.GraphDef.Builder, org.tensorflow.proto.framework.GraphDefOrBuilder> postOptimizationGraphBuilder_;
+      /**
+       * <code>.tensorflow.GraphDef post_optimization_graph = 3;</code>
+       */
+      public boolean hasPostOptimizationGraph() {
+        return postOptimizationGraphBuilder_ != null || postOptimizationGraph_ != null;
+      }
+      /**
+       * <code>.tensorflow.GraphDef post_optimization_graph = 3;</code>
+       */
+      public org.tensorflow.proto.framework.GraphDef getPostOptimizationGraph() {
+        if (postOptimizationGraphBuilder_ == null) {
+          return postOptimizationGraph_ == null ? org.tensorflow.proto.framework.GraphDef.getDefaultInstance() : postOptimizationGraph_;
+        } else {
+          return postOptimizationGraphBuilder_.getMessage();
+        }
+      }
+      /**
+       * <code>.tensorflow.GraphDef post_optimization_graph = 3;</code>
+       */
+      public Builder setPostOptimizationGraph(org.tensorflow.proto.framework.GraphDef value) {
+        if (postOptimizationGraphBuilder_ == null) {
+          if (value == null) {
+            throw new NullPointerException();
+          }
+          postOptimizationGraph_ = value;
+          onChanged();
+        } else {
+          postOptimizationGraphBuilder_.setMessage(value);
+        }
+
+        return this;
+      }
+      /**
+       * <code>.tensorflow.GraphDef post_optimization_graph = 3;</code>
+       */
+      public Builder setPostOptimizationGraph(
+          org.tensorflow.proto.framework.GraphDef.Builder builderForValue) {
+        if (postOptimizationGraphBuilder_ == null) {
+          postOptimizationGraph_ = builderForValue.build();
+          onChanged();
+        } else {
+          postOptimizationGraphBuilder_.setMessage(builderForValue.build());
+        }
+
+        return this;
+      }
+      /**
+       * <code>.tensorflow.GraphDef post_optimization_graph = 3;</code>
+       */
+      public Builder mergePostOptimizationGraph(org.tensorflow.proto.framework.GraphDef value) {
+        if (postOptimizationGraphBuilder_ == null) {
+          if (postOptimizationGraph_ != null) {
+            postOptimizationGraph_ =
+              org.tensorflow.proto.framework.GraphDef.newBuilder(postOptimizationGraph_).mergeFrom(value).buildPartial();
+          } else {
+            postOptimizationGraph_ = value;
+          }
+          onChanged();
+        } else {
+          postOptimizationGraphBuilder_.mergeFrom(value);
+        }
+
+        return this;
+      }
+      /**
+       * <code>.tensorflow.GraphDef post_optimization_graph = 3;</code>
+       */
+      public Builder clearPostOptimizationGraph() {
+        if (postOptimizationGraphBuilder_ == null) {
+          postOptimizationGraph_ = null;
+          onChanged();
+        } else {
+          postOptimizationGraph_ = null;
+          postOptimizationGraphBuilder_ = null;
+        }
+
+        return this;
+      }
+      /**
+       * <code>.tensorflow.GraphDef post_optimization_graph = 3;</code>
+       */
+      public org.tensorflow.proto.framework.GraphDef.Builder getPostOptimizationGraphBuilder() {
+        
+        onChanged();
+        return getPostOptimizationGraphFieldBuilder().getBuilder();
+      }
+      /**
+       * <code>.tensorflow.GraphDef post_optimization_graph = 3;</code>
+       */
+      public org.tensorflow.proto.framework.GraphDefOrBuilder getPostOptimizationGraphOrBuilder() {
+        if (postOptimizationGraphBuilder_ != null) {
+          return postOptimizationGraphBuilder_.getMessageOrBuilder();
+        } else {
+          return postOptimizationGraph_ == null ?
+              org.tensorflow.proto.framework.GraphDef.getDefaultInstance() : postOptimizationGraph_;
+        }
+      }
+      /**
+       * <code>.tensorflow.GraphDef post_optimization_graph = 3;</code>
+       */
+      private com.google.protobuf.SingleFieldBuilderV3<
+          org.tensorflow.proto.framework.GraphDef, org.tensorflow.proto.framework.GraphDef.Builder, org.tensorflow.proto.framework.GraphDefOrBuilder> 
+          getPostOptimizationGraphFieldBuilder() {
+        if (postOptimizationGraphBuilder_ == null) {
+          postOptimizationGraphBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+              org.tensorflow.proto.framework.GraphDef, org.tensorflow.proto.framework.GraphDef.Builder, org.tensorflow.proto.framework.GraphDefOrBuilder>(
+                  getPostOptimizationGraph(),
+                  getParentForChildren(),
+                  isClean());
+          postOptimizationGraph_ = null;
+        }
+        return postOptimizationGraphBuilder_;
+      }
+      @java.lang.Override
+      public final Builder setUnknownFields(
+          final com.google.protobuf.UnknownFieldSet unknownFields) {
+        return super.setUnknownFields(unknownFields);
+      }
+
+      @java.lang.Override
+      public final Builder mergeUnknownFields(
+          final com.google.protobuf.UnknownFieldSet unknownFields) {
+        return super.mergeUnknownFields(unknownFields);
+      }
+
+
+      // @@protoc_insertion_point(builder_scope:tensorflow.RunMetadata.FunctionGraphs)
+    }
+
+    // @@protoc_insertion_point(class_scope:tensorflow.RunMetadata.FunctionGraphs)
+    private static final org.tensorflow.proto.framework.RunMetadata.FunctionGraphs DEFAULT_INSTANCE;
+    static {
+      DEFAULT_INSTANCE = new org.tensorflow.proto.framework.RunMetadata.FunctionGraphs();
+    }
+
+    public static org.tensorflow.proto.framework.RunMetadata.FunctionGraphs getDefaultInstance() {
+      return DEFAULT_INSTANCE;
+    }
+
+    private static final com.google.protobuf.Parser<FunctionGraphs>
+        PARSER = new com.google.protobuf.AbstractParser<FunctionGraphs>() {
+      @java.lang.Override
+      public FunctionGraphs parsePartialFrom(
+          com.google.protobuf.CodedInputStream input,
+          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+          throws com.google.protobuf.InvalidProtocolBufferException {
+        return new FunctionGraphs(input, extensionRegistry);
+      }
+    };
+
+    public static com.google.protobuf.Parser<FunctionGraphs> parser() {
+      return PARSER;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Parser<FunctionGraphs> getParserForType() {
+      return PARSER;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.RunMetadata.FunctionGraphs getDefaultInstanceForType() {
+      return DEFAULT_INSTANCE;
+    }
+
+  }
+
+  public static final int STEP_STATS_FIELD_NUMBER = 1;
+  private org.tensorflow.proto.framework.StepStats stepStats_;
+  /**
+   * <pre>
+   * Statistics traced for this step. Populated if tracing is turned on via the
+   * "RunOptions" proto.
+   * EXPERIMENTAL: The format and set of events may change in future versions.
+   * </pre>
+   *
+   * <code>.tensorflow.StepStats step_stats = 1;</code>
+   */
+  public boolean hasStepStats() {
+    return stepStats_ != null;
+  }
+  /**
+   * <pre>
+   * Statistics traced for this step. Populated if tracing is turned on via the
+   * "RunOptions" proto.
+   * EXPERIMENTAL: The format and set of events may change in future versions.
+   * </pre>
+   *
+   * <code>.tensorflow.StepStats step_stats = 1;</code>
+   */
+  public org.tensorflow.proto.framework.StepStats getStepStats() {
+    return stepStats_ == null ? org.tensorflow.proto.framework.StepStats.getDefaultInstance() : stepStats_;
+  }
+  /**
+   * <pre>
+   * Statistics traced for this step. Populated if tracing is turned on via the
+   * "RunOptions" proto.
+   * EXPERIMENTAL: The format and set of events may change in future versions.
+   * </pre>
+   *
+   * <code>.tensorflow.StepStats step_stats = 1;</code>
+   */
+  public org.tensorflow.proto.framework.StepStatsOrBuilder getStepStatsOrBuilder() {
+    return getStepStats();
+  }
+
+  public static final int COST_GRAPH_FIELD_NUMBER = 2;
+  private org.tensorflow.proto.framework.CostGraphDef costGraph_;
+  /**
+   * <pre>
+   * The cost graph for the computation defined by the run call.
+   * </pre>
+   *
+   * <code>.tensorflow.CostGraphDef cost_graph = 2;</code>
+   */
+  public boolean hasCostGraph() {
+    return costGraph_ != null;
+  }
+  /**
+   * <pre>
+   * The cost graph for the computation defined by the run call.
+   * </pre>
+   *
+   * <code>.tensorflow.CostGraphDef cost_graph = 2;</code>
+   */
+  public org.tensorflow.proto.framework.CostGraphDef getCostGraph() {
+    return costGraph_ == null ? org.tensorflow.proto.framework.CostGraphDef.getDefaultInstance() : costGraph_;
+  }
+  /**
+   * <pre>
+   * The cost graph for the computation defined by the run call.
+   * </pre>
+   *
+   * <code>.tensorflow.CostGraphDef cost_graph = 2;</code>
+   */
+  public org.tensorflow.proto.framework.CostGraphDefOrBuilder getCostGraphOrBuilder() {
+    return getCostGraph();
+  }
+
+  public static final int PARTITION_GRAPHS_FIELD_NUMBER = 3;
+  private java.util.List<org.tensorflow.proto.framework.GraphDef> partitionGraphs_;
+  /**
+   * <pre>
+   * Graphs of the partitions executed by executors.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.GraphDef partition_graphs = 3;</code>
+   */
+  public java.util.List<org.tensorflow.proto.framework.GraphDef> getPartitionGraphsList() {
+    return partitionGraphs_;
+  }
+  /**
+   * <pre>
+   * Graphs of the partitions executed by executors.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.GraphDef partition_graphs = 3;</code>
+   */
+  public java.util.List<? extends org.tensorflow.proto.framework.GraphDefOrBuilder> 
+      getPartitionGraphsOrBuilderList() {
+    return partitionGraphs_;
+  }
+  /**
+   * <pre>
+   * Graphs of the partitions executed by executors.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.GraphDef partition_graphs = 3;</code>
+   */
+  public int getPartitionGraphsCount() {
+    return partitionGraphs_.size();
+  }
+  /**
+   * <pre>
+   * Graphs of the partitions executed by executors.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.GraphDef partition_graphs = 3;</code>
+   */
+  public org.tensorflow.proto.framework.GraphDef getPartitionGraphs(int index) {
+    return partitionGraphs_.get(index);
+  }
+  /**
+   * <pre>
+   * Graphs of the partitions executed by executors.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.GraphDef partition_graphs = 3;</code>
+   */
+  public org.tensorflow.proto.framework.GraphDefOrBuilder getPartitionGraphsOrBuilder(
+      int index) {
+    return partitionGraphs_.get(index);
+  }
+
+  public static final int FUNCTION_GRAPHS_FIELD_NUMBER = 4;
+  private java.util.List<org.tensorflow.proto.framework.RunMetadata.FunctionGraphs> functionGraphs_;
+  /**
+   * <pre>
+   * This is only populated for graphs that are run as functions in TensorFlow
+   * V2. There will be an entry below for each function that is traced.
+   * The main use cases of the post_optimization_graph and the partition_graphs
+   * is to give the caller insight into the graphs that were actually run by the
+   * runtime. Additional information (such as those in step_stats) will match
+   * these graphs.
+   * We also include the pre_optimization_graph since it is usually easier to
+   * read, and is helpful in situations where the caller wants to get a high
+   * level idea of what the built graph looks like (since the various graph
+   * optimization passes might change the structure of the graph significantly).
+   * </pre>
+   *
+   * <code>repeated .tensorflow.RunMetadata.FunctionGraphs function_graphs = 4;</code>
+   */
+  public java.util.List<org.tensorflow.proto.framework.RunMetadata.FunctionGraphs> getFunctionGraphsList() {
+    return functionGraphs_;
+  }
+  /**
+   * <pre>
+   * This is only populated for graphs that are run as functions in TensorFlow
+   * V2. There will be an entry below for each function that is traced.
+   * The main use cases of the post_optimization_graph and the partition_graphs
+   * is to give the caller insight into the graphs that were actually run by the
+   * runtime. Additional information (such as those in step_stats) will match
+   * these graphs.
+   * We also include the pre_optimization_graph since it is usually easier to
+   * read, and is helpful in situations where the caller wants to get a high
+   * level idea of what the built graph looks like (since the various graph
+   * optimization passes might change the structure of the graph significantly).
+   * </pre>
+   *
+   * <code>repeated .tensorflow.RunMetadata.FunctionGraphs function_graphs = 4;</code>
+   */
+  public java.util.List<? extends org.tensorflow.proto.framework.RunMetadata.FunctionGraphsOrBuilder> 
+      getFunctionGraphsOrBuilderList() {
+    return functionGraphs_;
+  }
+  /**
+   * <pre>
+   * This is only populated for graphs that are run as functions in TensorFlow
+   * V2. There will be an entry below for each function that is traced.
+   * The main use cases of the post_optimization_graph and the partition_graphs
+   * is to give the caller insight into the graphs that were actually run by the
+   * runtime. Additional information (such as those in step_stats) will match
+   * these graphs.
+   * We also include the pre_optimization_graph since it is usually easier to
+   * read, and is helpful in situations where the caller wants to get a high
+   * level idea of what the built graph looks like (since the various graph
+   * optimization passes might change the structure of the graph significantly).
+   * </pre>
+   *
+   * <code>repeated .tensorflow.RunMetadata.FunctionGraphs function_graphs = 4;</code>
+   */
+  public int getFunctionGraphsCount() {
+    return functionGraphs_.size();
+  }
+  /**
+   * <pre>
+   * This is only populated for graphs that are run as functions in TensorFlow
+   * V2. There will be an entry below for each function that is traced.
+   * The main use cases of the post_optimization_graph and the partition_graphs
+   * is to give the caller insight into the graphs that were actually run by the
+   * runtime. Additional information (such as those in step_stats) will match
+   * these graphs.
+   * We also include the pre_optimization_graph since it is usually easier to
+   * read, and is helpful in situations where the caller wants to get a high
+   * level idea of what the built graph looks like (since the various graph
+   * optimization passes might change the structure of the graph significantly).
+   * </pre>
+   *
+   * <code>repeated .tensorflow.RunMetadata.FunctionGraphs function_graphs = 4;</code>
+   */
+  public org.tensorflow.proto.framework.RunMetadata.FunctionGraphs getFunctionGraphs(int index) {
+    return functionGraphs_.get(index);
+  }
+  /**
+   * <pre>
+   * This is only populated for graphs that are run as functions in TensorFlow
+   * V2. There will be an entry below for each function that is traced.
+   * The main use cases of the post_optimization_graph and the partition_graphs
+   * is to give the caller insight into the graphs that were actually run by the
+   * runtime. Additional information (such as those in step_stats) will match
+   * these graphs.
+   * We also include the pre_optimization_graph since it is usually easier to
+   * read, and is helpful in situations where the caller wants to get a high
+   * level idea of what the built graph looks like (since the various graph
+   * optimization passes might change the structure of the graph significantly).
+   * </pre>
+   *
+   * <code>repeated .tensorflow.RunMetadata.FunctionGraphs function_graphs = 4;</code>
+   */
+  public org.tensorflow.proto.framework.RunMetadata.FunctionGraphsOrBuilder getFunctionGraphsOrBuilder(
+      int index) {
+    return functionGraphs_.get(index);
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    if (stepStats_ != null) {
+      output.writeMessage(1, getStepStats());
+    }
+    if (costGraph_ != null) {
+      output.writeMessage(2, getCostGraph());
+    }
+    for (int i = 0; i < partitionGraphs_.size(); i++) {
+      output.writeMessage(3, partitionGraphs_.get(i));
+    }
+    for (int i = 0; i < functionGraphs_.size(); i++) {
+      output.writeMessage(4, functionGraphs_.get(i));
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (stepStats_ != null) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(1, getStepStats());
+    }
+    if (costGraph_ != null) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(2, getCostGraph());
+    }
+    for (int i = 0; i < partitionGraphs_.size(); i++) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(3, partitionGraphs_.get(i));
+    }
+    for (int i = 0; i < functionGraphs_.size(); i++) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(4, functionGraphs_.get(i));
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.framework.RunMetadata)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.framework.RunMetadata other = (org.tensorflow.proto.framework.RunMetadata) obj;
+
+    if (hasStepStats() != other.hasStepStats()) return false;
+    if (hasStepStats()) {
+      if (!getStepStats()
+          .equals(other.getStepStats())) return false;
+    }
+    if (hasCostGraph() != other.hasCostGraph()) return false;
+    if (hasCostGraph()) {
+      if (!getCostGraph()
+          .equals(other.getCostGraph())) return false;
+    }
+    if (!getPartitionGraphsList()
+        .equals(other.getPartitionGraphsList())) return false;
+    if (!getFunctionGraphsList()
+        .equals(other.getFunctionGraphsList())) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    if (hasStepStats()) {
+      hash = (37 * hash) + STEP_STATS_FIELD_NUMBER;
+      hash = (53 * hash) + getStepStats().hashCode();
+    }
+    if (hasCostGraph()) {
+      hash = (37 * hash) + COST_GRAPH_FIELD_NUMBER;
+      hash = (53 * hash) + getCostGraph().hashCode();
+    }
+    if (getPartitionGraphsCount() > 0) {
+      hash = (37 * hash) + PARTITION_GRAPHS_FIELD_NUMBER;
+      hash = (53 * hash) + getPartitionGraphsList().hashCode();
+    }
+    if (getFunctionGraphsCount() > 0) {
+      hash = (37 * hash) + FUNCTION_GRAPHS_FIELD_NUMBER;
+      hash = (53 * hash) + getFunctionGraphsList().hashCode();
+    }
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.framework.RunMetadata parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.RunMetadata parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.RunMetadata parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.RunMetadata parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.RunMetadata parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.RunMetadata parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.RunMetadata parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.RunMetadata parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.RunMetadata parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.RunMetadata parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.RunMetadata parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.RunMetadata parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.framework.RunMetadata prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * <pre>
+   * Metadata output (i.e., non-Tensor) for a single Run() call.
+   * </pre>
+   *
+   * Protobuf type {@code tensorflow.RunMetadata}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.RunMetadata)
+      org.tensorflow.proto.framework.RunMetadataOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_RunMetadata_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_RunMetadata_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.RunMetadata.class, org.tensorflow.proto.framework.RunMetadata.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.framework.RunMetadata.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+        getPartitionGraphsFieldBuilder();
+        getFunctionGraphsFieldBuilder();
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      if (stepStatsBuilder_ == null) {
+        stepStats_ = null;
+      } else {
+        stepStats_ = null;
+        stepStatsBuilder_ = null;
+      }
+      if (costGraphBuilder_ == null) {
+        costGraph_ = null;
+      } else {
+        costGraph_ = null;
+        costGraphBuilder_ = null;
+      }
+      if (partitionGraphsBuilder_ == null) {
+        partitionGraphs_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000001);
+      } else {
+        partitionGraphsBuilder_.clear();
+      }
+      if (functionGraphsBuilder_ == null) {
+        functionGraphs_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000002);
+      } else {
+        functionGraphsBuilder_.clear();
+      }
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_RunMetadata_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.RunMetadata getDefaultInstanceForType() {
+      return org.tensorflow.proto.framework.RunMetadata.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.RunMetadata build() {
+      org.tensorflow.proto.framework.RunMetadata result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.RunMetadata buildPartial() {
+      org.tensorflow.proto.framework.RunMetadata result = new org.tensorflow.proto.framework.RunMetadata(this);
+      int from_bitField0_ = bitField0_;
+      if (stepStatsBuilder_ == null) {
+        result.stepStats_ = stepStats_;
+      } else {
+        result.stepStats_ = stepStatsBuilder_.build();
+      }
+      if (costGraphBuilder_ == null) {
+        result.costGraph_ = costGraph_;
+      } else {
+        result.costGraph_ = costGraphBuilder_.build();
+      }
+      if (partitionGraphsBuilder_ == null) {
+        if (((bitField0_ & 0x00000001) != 0)) {
+          partitionGraphs_ = java.util.Collections.unmodifiableList(partitionGraphs_);
+          bitField0_ = (bitField0_ & ~0x00000001);
+        }
+        result.partitionGraphs_ = partitionGraphs_;
+      } else {
+        result.partitionGraphs_ = partitionGraphsBuilder_.build();
+      }
+      if (functionGraphsBuilder_ == null) {
+        if (((bitField0_ & 0x00000002) != 0)) {
+          functionGraphs_ = java.util.Collections.unmodifiableList(functionGraphs_);
+          bitField0_ = (bitField0_ & ~0x00000002);
+        }
+        result.functionGraphs_ = functionGraphs_;
+      } else {
+        result.functionGraphs_ = functionGraphsBuilder_.build();
+      }
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.framework.RunMetadata) {
+        return mergeFrom((org.tensorflow.proto.framework.RunMetadata)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.framework.RunMetadata other) {
+      if (other == org.tensorflow.proto.framework.RunMetadata.getDefaultInstance()) return this;
+      if (other.hasStepStats()) {
+        mergeStepStats(other.getStepStats());
+      }
+      if (other.hasCostGraph()) {
+        mergeCostGraph(other.getCostGraph());
+      }
+      if (partitionGraphsBuilder_ == null) {
+        if (!other.partitionGraphs_.isEmpty()) {
+          if (partitionGraphs_.isEmpty()) {
+            partitionGraphs_ = other.partitionGraphs_;
+            bitField0_ = (bitField0_ & ~0x00000001);
+          } else {
+            ensurePartitionGraphsIsMutable();
+            partitionGraphs_.addAll(other.partitionGraphs_);
+          }
+          onChanged();
+        }
+      } else {
+        if (!other.partitionGraphs_.isEmpty()) {
+          if (partitionGraphsBuilder_.isEmpty()) {
+            partitionGraphsBuilder_.dispose();
+            partitionGraphsBuilder_ = null;
+            partitionGraphs_ = other.partitionGraphs_;
+            bitField0_ = (bitField0_ & ~0x00000001);
+            partitionGraphsBuilder_ = 
+              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
+                 getPartitionGraphsFieldBuilder() : null;
+          } else {
+            partitionGraphsBuilder_.addAllMessages(other.partitionGraphs_);
+          }
+        }
+      }
+      if (functionGraphsBuilder_ == null) {
+        if (!other.functionGraphs_.isEmpty()) {
+          if (functionGraphs_.isEmpty()) {
+            functionGraphs_ = other.functionGraphs_;
+            bitField0_ = (bitField0_ & ~0x00000002);
+          } else {
+            ensureFunctionGraphsIsMutable();
+            functionGraphs_.addAll(other.functionGraphs_);
+          }
+          onChanged();
+        }
+      } else {
+        if (!other.functionGraphs_.isEmpty()) {
+          if (functionGraphsBuilder_.isEmpty()) {
+            functionGraphsBuilder_.dispose();
+            functionGraphsBuilder_ = null;
+            functionGraphs_ = other.functionGraphs_;
+            bitField0_ = (bitField0_ & ~0x00000002);
+            functionGraphsBuilder_ = 
+              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
+                 getFunctionGraphsFieldBuilder() : null;
+          } else {
+            functionGraphsBuilder_.addAllMessages(other.functionGraphs_);
+          }
+        }
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.framework.RunMetadata parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.framework.RunMetadata) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+    private int bitField0_;
+
+    private org.tensorflow.proto.framework.StepStats stepStats_;
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.StepStats, org.tensorflow.proto.framework.StepStats.Builder, org.tensorflow.proto.framework.StepStatsOrBuilder> stepStatsBuilder_;
+    /**
+     * <pre>
+     * Statistics traced for this step. Populated if tracing is turned on via the
+     * "RunOptions" proto.
+     * EXPERIMENTAL: The format and set of events may change in future versions.
+     * </pre>
+     *
+     * <code>.tensorflow.StepStats step_stats = 1;</code>
+     */
+    public boolean hasStepStats() {
+      return stepStatsBuilder_ != null || stepStats_ != null;
+    }
+    /**
+     * <pre>
+     * Statistics traced for this step. Populated if tracing is turned on via the
+     * "RunOptions" proto.
+     * EXPERIMENTAL: The format and set of events may change in future versions.
+     * </pre>
+     *
+     * <code>.tensorflow.StepStats step_stats = 1;</code>
+     */
+    public org.tensorflow.proto.framework.StepStats getStepStats() {
+      if (stepStatsBuilder_ == null) {
+        return stepStats_ == null ? org.tensorflow.proto.framework.StepStats.getDefaultInstance() : stepStats_;
+      } else {
+        return stepStatsBuilder_.getMessage();
+      }
+    }
+    /**
+     * <pre>
+     * Statistics traced for this step. Populated if tracing is turned on via the
+     * "RunOptions" proto.
+     * EXPERIMENTAL: The format and set of events may change in future versions.
+     * </pre>
+     *
+     * <code>.tensorflow.StepStats step_stats = 1;</code>
+     */
+    public Builder setStepStats(org.tensorflow.proto.framework.StepStats value) {
+      if (stepStatsBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        stepStats_ = value;
+        onChanged();
+      } else {
+        stepStatsBuilder_.setMessage(value);
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Statistics traced for this step. Populated if tracing is turned on via the
+     * "RunOptions" proto.
+     * EXPERIMENTAL: The format and set of events may change in future versions.
+     * </pre>
+     *
+     * <code>.tensorflow.StepStats step_stats = 1;</code>
+     */
+    public Builder setStepStats(
+        org.tensorflow.proto.framework.StepStats.Builder builderForValue) {
+      if (stepStatsBuilder_ == null) {
+        stepStats_ = builderForValue.build();
+        onChanged();
+      } else {
+        stepStatsBuilder_.setMessage(builderForValue.build());
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Statistics traced for this step. Populated if tracing is turned on via the
+     * "RunOptions" proto.
+     * EXPERIMENTAL: The format and set of events may change in future versions.
+     * </pre>
+     *
+     * <code>.tensorflow.StepStats step_stats = 1;</code>
+     */
+    public Builder mergeStepStats(org.tensorflow.proto.framework.StepStats value) {
+      if (stepStatsBuilder_ == null) {
+        if (stepStats_ != null) {
+          stepStats_ =
+            org.tensorflow.proto.framework.StepStats.newBuilder(stepStats_).mergeFrom(value).buildPartial();
+        } else {
+          stepStats_ = value;
+        }
+        onChanged();
+      } else {
+        stepStatsBuilder_.mergeFrom(value);
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Statistics traced for this step. Populated if tracing is turned on via the
+     * "RunOptions" proto.
+     * EXPERIMENTAL: The format and set of events may change in future versions.
+     * </pre>
+     *
+     * <code>.tensorflow.StepStats step_stats = 1;</code>
+     */
+    public Builder clearStepStats() {
+      if (stepStatsBuilder_ == null) {
+        stepStats_ = null;
+        onChanged();
+      } else {
+        stepStats_ = null;
+        stepStatsBuilder_ = null;
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Statistics traced for this step. Populated if tracing is turned on via the
+     * "RunOptions" proto.
+     * EXPERIMENTAL: The format and set of events may change in future versions.
+     * </pre>
+     *
+     * <code>.tensorflow.StepStats step_stats = 1;</code>
+     */
+    public org.tensorflow.proto.framework.StepStats.Builder getStepStatsBuilder() {
+      
+      onChanged();
+      return getStepStatsFieldBuilder().getBuilder();
+    }
+    /**
+     * <pre>
+     * Statistics traced for this step. Populated if tracing is turned on via the
+     * "RunOptions" proto.
+     * EXPERIMENTAL: The format and set of events may change in future versions.
+     * </pre>
+     *
+     * <code>.tensorflow.StepStats step_stats = 1;</code>
+     */
+    public org.tensorflow.proto.framework.StepStatsOrBuilder getStepStatsOrBuilder() {
+      if (stepStatsBuilder_ != null) {
+        return stepStatsBuilder_.getMessageOrBuilder();
+      } else {
+        return stepStats_ == null ?
+            org.tensorflow.proto.framework.StepStats.getDefaultInstance() : stepStats_;
+      }
+    }
+    /**
+     * <pre>
+     * Statistics traced for this step. Populated if tracing is turned on via the
+     * "RunOptions" proto.
+     * EXPERIMENTAL: The format and set of events may change in future versions.
+     * </pre>
+     *
+     * <code>.tensorflow.StepStats step_stats = 1;</code>
+     */
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.StepStats, org.tensorflow.proto.framework.StepStats.Builder, org.tensorflow.proto.framework.StepStatsOrBuilder> 
+        getStepStatsFieldBuilder() {
+      if (stepStatsBuilder_ == null) {
+        stepStatsBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+            org.tensorflow.proto.framework.StepStats, org.tensorflow.proto.framework.StepStats.Builder, org.tensorflow.proto.framework.StepStatsOrBuilder>(
+                getStepStats(),
+                getParentForChildren(),
+                isClean());
+        stepStats_ = null;
+      }
+      return stepStatsBuilder_;
+    }
+
+    private org.tensorflow.proto.framework.CostGraphDef costGraph_;
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.CostGraphDef, org.tensorflow.proto.framework.CostGraphDef.Builder, org.tensorflow.proto.framework.CostGraphDefOrBuilder> costGraphBuilder_;
+    /**
+     * <pre>
+     * The cost graph for the computation defined by the run call.
+     * </pre>
+     *
+     * <code>.tensorflow.CostGraphDef cost_graph = 2;</code>
+     */
+    public boolean hasCostGraph() {
+      return costGraphBuilder_ != null || costGraph_ != null;
+    }
+    /**
+     * <pre>
+     * The cost graph for the computation defined by the run call.
+     * </pre>
+     *
+     * <code>.tensorflow.CostGraphDef cost_graph = 2;</code>
+     */
+    public org.tensorflow.proto.framework.CostGraphDef getCostGraph() {
+      if (costGraphBuilder_ == null) {
+        return costGraph_ == null ? org.tensorflow.proto.framework.CostGraphDef.getDefaultInstance() : costGraph_;
+      } else {
+        return costGraphBuilder_.getMessage();
+      }
+    }
+    /**
+     * <pre>
+     * The cost graph for the computation defined by the run call.
+     * </pre>
+     *
+     * <code>.tensorflow.CostGraphDef cost_graph = 2;</code>
+     */
+    public Builder setCostGraph(org.tensorflow.proto.framework.CostGraphDef value) {
+      if (costGraphBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        costGraph_ = value;
+        onChanged();
+      } else {
+        costGraphBuilder_.setMessage(value);
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * The cost graph for the computation defined by the run call.
+     * </pre>
+     *
+     * <code>.tensorflow.CostGraphDef cost_graph = 2;</code>
+     */
+    public Builder setCostGraph(
+        org.tensorflow.proto.framework.CostGraphDef.Builder builderForValue) {
+      if (costGraphBuilder_ == null) {
+        costGraph_ = builderForValue.build();
+        onChanged();
+      } else {
+        costGraphBuilder_.setMessage(builderForValue.build());
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * The cost graph for the computation defined by the run call.
+     * </pre>
+     *
+     * <code>.tensorflow.CostGraphDef cost_graph = 2;</code>
+     */
+    public Builder mergeCostGraph(org.tensorflow.proto.framework.CostGraphDef value) {
+      if (costGraphBuilder_ == null) {
+        if (costGraph_ != null) {
+          costGraph_ =
+            org.tensorflow.proto.framework.CostGraphDef.newBuilder(costGraph_).mergeFrom(value).buildPartial();
+        } else {
+          costGraph_ = value;
+        }
+        onChanged();
+      } else {
+        costGraphBuilder_.mergeFrom(value);
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * The cost graph for the computation defined by the run call.
+     * </pre>
+     *
+     * <code>.tensorflow.CostGraphDef cost_graph = 2;</code>
+     */
+    public Builder clearCostGraph() {
+      if (costGraphBuilder_ == null) {
+        costGraph_ = null;
+        onChanged();
+      } else {
+        costGraph_ = null;
+        costGraphBuilder_ = null;
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * The cost graph for the computation defined by the run call.
+     * </pre>
+     *
+     * <code>.tensorflow.CostGraphDef cost_graph = 2;</code>
+     */
+    public org.tensorflow.proto.framework.CostGraphDef.Builder getCostGraphBuilder() {
+      
+      onChanged();
+      return getCostGraphFieldBuilder().getBuilder();
+    }
+    /**
+     * <pre>
+     * The cost graph for the computation defined by the run call.
+     * </pre>
+     *
+     * <code>.tensorflow.CostGraphDef cost_graph = 2;</code>
+     */
+    public org.tensorflow.proto.framework.CostGraphDefOrBuilder getCostGraphOrBuilder() {
+      if (costGraphBuilder_ != null) {
+        return costGraphBuilder_.getMessageOrBuilder();
+      } else {
+        return costGraph_ == null ?
+            org.tensorflow.proto.framework.CostGraphDef.getDefaultInstance() : costGraph_;
+      }
+    }
+    /**
+     * <pre>
+     * The cost graph for the computation defined by the run call.
+     * </pre>
+     *
+     * <code>.tensorflow.CostGraphDef cost_graph = 2;</code>
+     */
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.CostGraphDef, org.tensorflow.proto.framework.CostGraphDef.Builder, org.tensorflow.proto.framework.CostGraphDefOrBuilder> 
+        getCostGraphFieldBuilder() {
+      if (costGraphBuilder_ == null) {
+        costGraphBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+            org.tensorflow.proto.framework.CostGraphDef, org.tensorflow.proto.framework.CostGraphDef.Builder, org.tensorflow.proto.framework.CostGraphDefOrBuilder>(
+                getCostGraph(),
+                getParentForChildren(),
+                isClean());
+        costGraph_ = null;
+      }
+      return costGraphBuilder_;
+    }
+
+    private java.util.List<org.tensorflow.proto.framework.GraphDef> partitionGraphs_ =
+      java.util.Collections.emptyList();
+    private void ensurePartitionGraphsIsMutable() {
+      if (!((bitField0_ & 0x00000001) != 0)) {
+        partitionGraphs_ = new java.util.ArrayList<org.tensorflow.proto.framework.GraphDef>(partitionGraphs_);
+        bitField0_ |= 0x00000001;
+       }
+    }
+
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.framework.GraphDef, org.tensorflow.proto.framework.GraphDef.Builder, org.tensorflow.proto.framework.GraphDefOrBuilder> partitionGraphsBuilder_;
+
+    /**
+     * <pre>
+     * Graphs of the partitions executed by executors.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.GraphDef partition_graphs = 3;</code>
+     */
+    public java.util.List<org.tensorflow.proto.framework.GraphDef> getPartitionGraphsList() {
+      if (partitionGraphsBuilder_ == null) {
+        return java.util.Collections.unmodifiableList(partitionGraphs_);
+      } else {
+        return partitionGraphsBuilder_.getMessageList();
+      }
+    }
+    /**
+     * <pre>
+     * Graphs of the partitions executed by executors.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.GraphDef partition_graphs = 3;</code>
+     */
+    public int getPartitionGraphsCount() {
+      if (partitionGraphsBuilder_ == null) {
+        return partitionGraphs_.size();
+      } else {
+        return partitionGraphsBuilder_.getCount();
+      }
+    }
+    /**
+     * <pre>
+     * Graphs of the partitions executed by executors.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.GraphDef partition_graphs = 3;</code>
+     */
+    public org.tensorflow.proto.framework.GraphDef getPartitionGraphs(int index) {
+      if (partitionGraphsBuilder_ == null) {
+        return partitionGraphs_.get(index);
+      } else {
+        return partitionGraphsBuilder_.getMessage(index);
+      }
+    }
+    /**
+     * <pre>
+     * Graphs of the partitions executed by executors.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.GraphDef partition_graphs = 3;</code>
+     */
+    public Builder setPartitionGraphs(
+        int index, org.tensorflow.proto.framework.GraphDef value) {
+      if (partitionGraphsBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensurePartitionGraphsIsMutable();
+        partitionGraphs_.set(index, value);
+        onChanged();
+      } else {
+        partitionGraphsBuilder_.setMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Graphs of the partitions executed by executors.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.GraphDef partition_graphs = 3;</code>
+     */
+    public Builder setPartitionGraphs(
+        int index, org.tensorflow.proto.framework.GraphDef.Builder builderForValue) {
+      if (partitionGraphsBuilder_ == null) {
+        ensurePartitionGraphsIsMutable();
+        partitionGraphs_.set(index, builderForValue.build());
+        onChanged();
+      } else {
+        partitionGraphsBuilder_.setMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Graphs of the partitions executed by executors.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.GraphDef partition_graphs = 3;</code>
+     */
+    public Builder addPartitionGraphs(org.tensorflow.proto.framework.GraphDef value) {
+      if (partitionGraphsBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensurePartitionGraphsIsMutable();
+        partitionGraphs_.add(value);
+        onChanged();
+      } else {
+        partitionGraphsBuilder_.addMessage(value);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Graphs of the partitions executed by executors.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.GraphDef partition_graphs = 3;</code>
+     */
+    public Builder addPartitionGraphs(
+        int index, org.tensorflow.proto.framework.GraphDef value) {
+      if (partitionGraphsBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensurePartitionGraphsIsMutable();
+        partitionGraphs_.add(index, value);
+        onChanged();
+      } else {
+        partitionGraphsBuilder_.addMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Graphs of the partitions executed by executors.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.GraphDef partition_graphs = 3;</code>
+     */
+    public Builder addPartitionGraphs(
+        org.tensorflow.proto.framework.GraphDef.Builder builderForValue) {
+      if (partitionGraphsBuilder_ == null) {
+        ensurePartitionGraphsIsMutable();
+        partitionGraphs_.add(builderForValue.build());
+        onChanged();
+      } else {
+        partitionGraphsBuilder_.addMessage(builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Graphs of the partitions executed by executors.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.GraphDef partition_graphs = 3;</code>
+     */
+    public Builder addPartitionGraphs(
+        int index, org.tensorflow.proto.framework.GraphDef.Builder builderForValue) {
+      if (partitionGraphsBuilder_ == null) {
+        ensurePartitionGraphsIsMutable();
+        partitionGraphs_.add(index, builderForValue.build());
+        onChanged();
+      } else {
+        partitionGraphsBuilder_.addMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Graphs of the partitions executed by executors.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.GraphDef partition_graphs = 3;</code>
+     */
+    public Builder addAllPartitionGraphs(
+        java.lang.Iterable<? extends org.tensorflow.proto.framework.GraphDef> values) {
+      if (partitionGraphsBuilder_ == null) {
+        ensurePartitionGraphsIsMutable();
+        com.google.protobuf.AbstractMessageLite.Builder.addAll(
+            values, partitionGraphs_);
+        onChanged();
+      } else {
+        partitionGraphsBuilder_.addAllMessages(values);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Graphs of the partitions executed by executors.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.GraphDef partition_graphs = 3;</code>
+     */
+    public Builder clearPartitionGraphs() {
+      if (partitionGraphsBuilder_ == null) {
+        partitionGraphs_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000001);
+        onChanged();
+      } else {
+        partitionGraphsBuilder_.clear();
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Graphs of the partitions executed by executors.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.GraphDef partition_graphs = 3;</code>
+     */
+    public Builder removePartitionGraphs(int index) {
+      if (partitionGraphsBuilder_ == null) {
+        ensurePartitionGraphsIsMutable();
+        partitionGraphs_.remove(index);
+        onChanged();
+      } else {
+        partitionGraphsBuilder_.remove(index);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Graphs of the partitions executed by executors.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.GraphDef partition_graphs = 3;</code>
+     */
+    public org.tensorflow.proto.framework.GraphDef.Builder getPartitionGraphsBuilder(
+        int index) {
+      return getPartitionGraphsFieldBuilder().getBuilder(index);
+    }
+    /**
+     * <pre>
+     * Graphs of the partitions executed by executors.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.GraphDef partition_graphs = 3;</code>
+     */
+    public org.tensorflow.proto.framework.GraphDefOrBuilder getPartitionGraphsOrBuilder(
+        int index) {
+      if (partitionGraphsBuilder_ == null) {
+        return partitionGraphs_.get(index);  } else {
+        return partitionGraphsBuilder_.getMessageOrBuilder(index);
+      }
+    }
+    /**
+     * <pre>
+     * Graphs of the partitions executed by executors.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.GraphDef partition_graphs = 3;</code>
+     */
+    public java.util.List<? extends org.tensorflow.proto.framework.GraphDefOrBuilder> 
+         getPartitionGraphsOrBuilderList() {
+      if (partitionGraphsBuilder_ != null) {
+        return partitionGraphsBuilder_.getMessageOrBuilderList();
+      } else {
+        return java.util.Collections.unmodifiableList(partitionGraphs_);
+      }
+    }
+    /**
+     * <pre>
+     * Graphs of the partitions executed by executors.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.GraphDef partition_graphs = 3;</code>
+     */
+    public org.tensorflow.proto.framework.GraphDef.Builder addPartitionGraphsBuilder() {
+      return getPartitionGraphsFieldBuilder().addBuilder(
+          org.tensorflow.proto.framework.GraphDef.getDefaultInstance());
+    }
+    /**
+     * <pre>
+     * Graphs of the partitions executed by executors.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.GraphDef partition_graphs = 3;</code>
+     */
+    public org.tensorflow.proto.framework.GraphDef.Builder addPartitionGraphsBuilder(
+        int index) {
+      return getPartitionGraphsFieldBuilder().addBuilder(
+          index, org.tensorflow.proto.framework.GraphDef.getDefaultInstance());
+    }
+    /**
+     * <pre>
+     * Graphs of the partitions executed by executors.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.GraphDef partition_graphs = 3;</code>
+     */
+    public java.util.List<org.tensorflow.proto.framework.GraphDef.Builder> 
+         getPartitionGraphsBuilderList() {
+      return getPartitionGraphsFieldBuilder().getBuilderList();
+    }
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.framework.GraphDef, org.tensorflow.proto.framework.GraphDef.Builder, org.tensorflow.proto.framework.GraphDefOrBuilder> 
+        getPartitionGraphsFieldBuilder() {
+      if (partitionGraphsBuilder_ == null) {
+        partitionGraphsBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
+            org.tensorflow.proto.framework.GraphDef, org.tensorflow.proto.framework.GraphDef.Builder, org.tensorflow.proto.framework.GraphDefOrBuilder>(
+                partitionGraphs_,
+                ((bitField0_ & 0x00000001) != 0),
+                getParentForChildren(),
+                isClean());
+        partitionGraphs_ = null;
+      }
+      return partitionGraphsBuilder_;
+    }
+
+    private java.util.List<org.tensorflow.proto.framework.RunMetadata.FunctionGraphs> functionGraphs_ =
+      java.util.Collections.emptyList();
+    private void ensureFunctionGraphsIsMutable() {
+      if (!((bitField0_ & 0x00000002) != 0)) {
+        functionGraphs_ = new java.util.ArrayList<org.tensorflow.proto.framework.RunMetadata.FunctionGraphs>(functionGraphs_);
+        bitField0_ |= 0x00000002;
+       }
+    }
+
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.framework.RunMetadata.FunctionGraphs, org.tensorflow.proto.framework.RunMetadata.FunctionGraphs.Builder, org.tensorflow.proto.framework.RunMetadata.FunctionGraphsOrBuilder> functionGraphsBuilder_;
+
+    /**
+     * <pre>
+     * This is only populated for graphs that are run as functions in TensorFlow
+     * V2. There will be an entry below for each function that is traced.
+     * The main use cases of the post_optimization_graph and the partition_graphs
+     * is to give the caller insight into the graphs that were actually run by the
+     * runtime. Additional information (such as those in step_stats) will match
+     * these graphs.
+     * We also include the pre_optimization_graph since it is usually easier to
+     * read, and is helpful in situations where the caller wants to get a high
+     * level idea of what the built graph looks like (since the various graph
+     * optimization passes might change the structure of the graph significantly).
+     * </pre>
+     *
+     * <code>repeated .tensorflow.RunMetadata.FunctionGraphs function_graphs = 4;</code>
+     */
+    public java.util.List<org.tensorflow.proto.framework.RunMetadata.FunctionGraphs> getFunctionGraphsList() {
+      if (functionGraphsBuilder_ == null) {
+        return java.util.Collections.unmodifiableList(functionGraphs_);
+      } else {
+        return functionGraphsBuilder_.getMessageList();
+      }
+    }
+    /**
+     * <pre>
+     * This is only populated for graphs that are run as functions in TensorFlow
+     * V2. There will be an entry below for each function that is traced.
+     * The main use cases of the post_optimization_graph and the partition_graphs
+     * is to give the caller insight into the graphs that were actually run by the
+     * runtime. Additional information (such as those in step_stats) will match
+     * these graphs.
+     * We also include the pre_optimization_graph since it is usually easier to
+     * read, and is helpful in situations where the caller wants to get a high
+     * level idea of what the built graph looks like (since the various graph
+     * optimization passes might change the structure of the graph significantly).
+     * </pre>
+     *
+     * <code>repeated .tensorflow.RunMetadata.FunctionGraphs function_graphs = 4;</code>
+     */
+    public int getFunctionGraphsCount() {
+      if (functionGraphsBuilder_ == null) {
+        return functionGraphs_.size();
+      } else {
+        return functionGraphsBuilder_.getCount();
+      }
+    }
+    /**
+     * <pre>
+     * This is only populated for graphs that are run as functions in TensorFlow
+     * V2. There will be an entry below for each function that is traced.
+     * The main use cases of the post_optimization_graph and the partition_graphs
+     * is to give the caller insight into the graphs that were actually run by the
+     * runtime. Additional information (such as those in step_stats) will match
+     * these graphs.
+     * We also include the pre_optimization_graph since it is usually easier to
+     * read, and is helpful in situations where the caller wants to get a high
+     * level idea of what the built graph looks like (since the various graph
+     * optimization passes might change the structure of the graph significantly).
+     * </pre>
+     *
+     * <code>repeated .tensorflow.RunMetadata.FunctionGraphs function_graphs = 4;</code>
+     */
+    public org.tensorflow.proto.framework.RunMetadata.FunctionGraphs getFunctionGraphs(int index) {
+      if (functionGraphsBuilder_ == null) {
+        return functionGraphs_.get(index);
+      } else {
+        return functionGraphsBuilder_.getMessage(index);
+      }
+    }
+    /**
+     * <pre>
+     * This is only populated for graphs that are run as functions in TensorFlow
+     * V2. There will be an entry below for each function that is traced.
+     * The main use cases of the post_optimization_graph and the partition_graphs
+     * is to give the caller insight into the graphs that were actually run by the
+     * runtime. Additional information (such as those in step_stats) will match
+     * these graphs.
+     * We also include the pre_optimization_graph since it is usually easier to
+     * read, and is helpful in situations where the caller wants to get a high
+     * level idea of what the built graph looks like (since the various graph
+     * optimization passes might change the structure of the graph significantly).
+     * </pre>
+     *
+     * <code>repeated .tensorflow.RunMetadata.FunctionGraphs function_graphs = 4;</code>
+     */
+    public Builder setFunctionGraphs(
+        int index, org.tensorflow.proto.framework.RunMetadata.FunctionGraphs value) {
+      if (functionGraphsBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureFunctionGraphsIsMutable();
+        functionGraphs_.set(index, value);
+        onChanged();
+      } else {
+        functionGraphsBuilder_.setMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * This is only populated for graphs that are run as functions in TensorFlow
+     * V2. There will be an entry below for each function that is traced.
+     * The main use cases of the post_optimization_graph and the partition_graphs
+     * is to give the caller insight into the graphs that were actually run by the
+     * runtime. Additional information (such as those in step_stats) will match
+     * these graphs.
+     * We also include the pre_optimization_graph since it is usually easier to
+     * read, and is helpful in situations where the caller wants to get a high
+     * level idea of what the built graph looks like (since the various graph
+     * optimization passes might change the structure of the graph significantly).
+     * </pre>
+     *
+     * <code>repeated .tensorflow.RunMetadata.FunctionGraphs function_graphs = 4;</code>
+     */
+    public Builder setFunctionGraphs(
+        int index, org.tensorflow.proto.framework.RunMetadata.FunctionGraphs.Builder builderForValue) {
+      if (functionGraphsBuilder_ == null) {
+        ensureFunctionGraphsIsMutable();
+        functionGraphs_.set(index, builderForValue.build());
+        onChanged();
+      } else {
+        functionGraphsBuilder_.setMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * This is only populated for graphs that are run as functions in TensorFlow
+     * V2. There will be an entry below for each function that is traced.
+     * The main use cases of the post_optimization_graph and the partition_graphs
+     * is to give the caller insight into the graphs that were actually run by the
+     * runtime. Additional information (such as those in step_stats) will match
+     * these graphs.
+     * We also include the pre_optimization_graph since it is usually easier to
+     * read, and is helpful in situations where the caller wants to get a high
+     * level idea of what the built graph looks like (since the various graph
+     * optimization passes might change the structure of the graph significantly).
+     * </pre>
+     *
+     * <code>repeated .tensorflow.RunMetadata.FunctionGraphs function_graphs = 4;</code>
+     */
+    public Builder addFunctionGraphs(org.tensorflow.proto.framework.RunMetadata.FunctionGraphs value) {
+      if (functionGraphsBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureFunctionGraphsIsMutable();
+        functionGraphs_.add(value);
+        onChanged();
+      } else {
+        functionGraphsBuilder_.addMessage(value);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * This is only populated for graphs that are run as functions in TensorFlow
+     * V2. There will be an entry below for each function that is traced.
+     * The main use cases of the post_optimization_graph and the partition_graphs
+     * is to give the caller insight into the graphs that were actually run by the
+     * runtime. Additional information (such as those in step_stats) will match
+     * these graphs.
+     * We also include the pre_optimization_graph since it is usually easier to
+     * read, and is helpful in situations where the caller wants to get a high
+     * level idea of what the built graph looks like (since the various graph
+     * optimization passes might change the structure of the graph significantly).
+     * </pre>
+     *
+     * <code>repeated .tensorflow.RunMetadata.FunctionGraphs function_graphs = 4;</code>
+     */
+    public Builder addFunctionGraphs(
+        int index, org.tensorflow.proto.framework.RunMetadata.FunctionGraphs value) {
+      if (functionGraphsBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureFunctionGraphsIsMutable();
+        functionGraphs_.add(index, value);
+        onChanged();
+      } else {
+        functionGraphsBuilder_.addMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * This is only populated for graphs that are run as functions in TensorFlow
+     * V2. There will be an entry below for each function that is traced.
+     * The main use cases of the post_optimization_graph and the partition_graphs
+     * is to give the caller insight into the graphs that were actually run by the
+     * runtime. Additional information (such as those in step_stats) will match
+     * these graphs.
+     * We also include the pre_optimization_graph since it is usually easier to
+     * read, and is helpful in situations where the caller wants to get a high
+     * level idea of what the built graph looks like (since the various graph
+     * optimization passes might change the structure of the graph significantly).
+     * </pre>
+     *
+     * <code>repeated .tensorflow.RunMetadata.FunctionGraphs function_graphs = 4;</code>
+     */
+    public Builder addFunctionGraphs(
+        org.tensorflow.proto.framework.RunMetadata.FunctionGraphs.Builder builderForValue) {
+      if (functionGraphsBuilder_ == null) {
+        ensureFunctionGraphsIsMutable();
+        functionGraphs_.add(builderForValue.build());
+        onChanged();
+      } else {
+        functionGraphsBuilder_.addMessage(builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * This is only populated for graphs that are run as functions in TensorFlow
+     * V2. There will be an entry below for each function that is traced.
+     * The main use cases of the post_optimization_graph and the partition_graphs
+     * is to give the caller insight into the graphs that were actually run by the
+     * runtime. Additional information (such as those in step_stats) will match
+     * these graphs.
+     * We also include the pre_optimization_graph since it is usually easier to
+     * read, and is helpful in situations where the caller wants to get a high
+     * level idea of what the built graph looks like (since the various graph
+     * optimization passes might change the structure of the graph significantly).
+     * </pre>
+     *
+     * <code>repeated .tensorflow.RunMetadata.FunctionGraphs function_graphs = 4;</code>
+     */
+    public Builder addFunctionGraphs(
+        int index, org.tensorflow.proto.framework.RunMetadata.FunctionGraphs.Builder builderForValue) {
+      if (functionGraphsBuilder_ == null) {
+        ensureFunctionGraphsIsMutable();
+        functionGraphs_.add(index, builderForValue.build());
+        onChanged();
+      } else {
+        functionGraphsBuilder_.addMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * This is only populated for graphs that are run as functions in TensorFlow
+     * V2. There will be an entry below for each function that is traced.
+     * The main use cases of the post_optimization_graph and the partition_graphs
+     * is to give the caller insight into the graphs that were actually run by the
+     * runtime. Additional information (such as those in step_stats) will match
+     * these graphs.
+     * We also include the pre_optimization_graph since it is usually easier to
+     * read, and is helpful in situations where the caller wants to get a high
+     * level idea of what the built graph looks like (since the various graph
+     * optimization passes might change the structure of the graph significantly).
+     * </pre>
+     *
+     * <code>repeated .tensorflow.RunMetadata.FunctionGraphs function_graphs = 4;</code>
+     */
+    public Builder addAllFunctionGraphs(
+        java.lang.Iterable<? extends org.tensorflow.proto.framework.RunMetadata.FunctionGraphs> values) {
+      if (functionGraphsBuilder_ == null) {
+        ensureFunctionGraphsIsMutable();
+        com.google.protobuf.AbstractMessageLite.Builder.addAll(
+            values, functionGraphs_);
+        onChanged();
+      } else {
+        functionGraphsBuilder_.addAllMessages(values);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * This is only populated for graphs that are run as functions in TensorFlow
+     * V2. There will be an entry below for each function that is traced.
+     * The main use cases of the post_optimization_graph and the partition_graphs
+     * is to give the caller insight into the graphs that were actually run by the
+     * runtime. Additional information (such as those in step_stats) will match
+     * these graphs.
+     * We also include the pre_optimization_graph since it is usually easier to
+     * read, and is helpful in situations where the caller wants to get a high
+     * level idea of what the built graph looks like (since the various graph
+     * optimization passes might change the structure of the graph significantly).
+     * </pre>
+     *
+     * <code>repeated .tensorflow.RunMetadata.FunctionGraphs function_graphs = 4;</code>
+     */
+    public Builder clearFunctionGraphs() {
+      if (functionGraphsBuilder_ == null) {
+        functionGraphs_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000002);
+        onChanged();
+      } else {
+        functionGraphsBuilder_.clear();
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * This is only populated for graphs that are run as functions in TensorFlow
+     * V2. There will be an entry below for each function that is traced.
+     * The main use cases of the post_optimization_graph and the partition_graphs
+     * is to give the caller insight into the graphs that were actually run by the
+     * runtime. Additional information (such as those in step_stats) will match
+     * these graphs.
+     * We also include the pre_optimization_graph since it is usually easier to
+     * read, and is helpful in situations where the caller wants to get a high
+     * level idea of what the built graph looks like (since the various graph
+     * optimization passes might change the structure of the graph significantly).
+     * </pre>
+     *
+     * <code>repeated .tensorflow.RunMetadata.FunctionGraphs function_graphs = 4;</code>
+     */
+    public Builder removeFunctionGraphs(int index) {
+      if (functionGraphsBuilder_ == null) {
+        ensureFunctionGraphsIsMutable();
+        functionGraphs_.remove(index);
+        onChanged();
+      } else {
+        functionGraphsBuilder_.remove(index);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * This is only populated for graphs that are run as functions in TensorFlow
+     * V2. There will be an entry below for each function that is traced.
+     * The main use cases of the post_optimization_graph and the partition_graphs
+     * is to give the caller insight into the graphs that were actually run by the
+     * runtime. Additional information (such as those in step_stats) will match
+     * these graphs.
+     * We also include the pre_optimization_graph since it is usually easier to
+     * read, and is helpful in situations where the caller wants to get a high
+     * level idea of what the built graph looks like (since the various graph
+     * optimization passes might change the structure of the graph significantly).
+     * </pre>
+     *
+     * <code>repeated .tensorflow.RunMetadata.FunctionGraphs function_graphs = 4;</code>
+     */
+    public org.tensorflow.proto.framework.RunMetadata.FunctionGraphs.Builder getFunctionGraphsBuilder(
+        int index) {
+      return getFunctionGraphsFieldBuilder().getBuilder(index);
+    }
+    /**
+     * <pre>
+     * This is only populated for graphs that are run as functions in TensorFlow
+     * V2. There will be an entry below for each function that is traced.
+     * The main use cases of the post_optimization_graph and the partition_graphs
+     * is to give the caller insight into the graphs that were actually run by the
+     * runtime. Additional information (such as those in step_stats) will match
+     * these graphs.
+     * We also include the pre_optimization_graph since it is usually easier to
+     * read, and is helpful in situations where the caller wants to get a high
+     * level idea of what the built graph looks like (since the various graph
+     * optimization passes might change the structure of the graph significantly).
+     * </pre>
+     *
+     * <code>repeated .tensorflow.RunMetadata.FunctionGraphs function_graphs = 4;</code>
+     */
+    public org.tensorflow.proto.framework.RunMetadata.FunctionGraphsOrBuilder getFunctionGraphsOrBuilder(
+        int index) {
+      if (functionGraphsBuilder_ == null) {
+        return functionGraphs_.get(index);  } else {
+        return functionGraphsBuilder_.getMessageOrBuilder(index);
+      }
+    }
+    /**
+     * <pre>
+     * This is only populated for graphs that are run as functions in TensorFlow
+     * V2. There will be an entry below for each function that is traced.
+     * The main use cases of the post_optimization_graph and the partition_graphs
+     * is to give the caller insight into the graphs that were actually run by the
+     * runtime. Additional information (such as those in step_stats) will match
+     * these graphs.
+     * We also include the pre_optimization_graph since it is usually easier to
+     * read, and is helpful in situations where the caller wants to get a high
+     * level idea of what the built graph looks like (since the various graph
+     * optimization passes might change the structure of the graph significantly).
+     * </pre>
+     *
+     * <code>repeated .tensorflow.RunMetadata.FunctionGraphs function_graphs = 4;</code>
+     */
+    public java.util.List<? extends org.tensorflow.proto.framework.RunMetadata.FunctionGraphsOrBuilder> 
+         getFunctionGraphsOrBuilderList() {
+      if (functionGraphsBuilder_ != null) {
+        return functionGraphsBuilder_.getMessageOrBuilderList();
+      } else {
+        return java.util.Collections.unmodifiableList(functionGraphs_);
+      }
+    }
+    /**
+     * <pre>
+     * This is only populated for graphs that are run as functions in TensorFlow
+     * V2. There will be an entry below for each function that is traced.
+     * The main use cases of the post_optimization_graph and the partition_graphs
+     * is to give the caller insight into the graphs that were actually run by the
+     * runtime. Additional information (such as those in step_stats) will match
+     * these graphs.
+     * We also include the pre_optimization_graph since it is usually easier to
+     * read, and is helpful in situations where the caller wants to get a high
+     * level idea of what the built graph looks like (since the various graph
+     * optimization passes might change the structure of the graph significantly).
+     * </pre>
+     *
+     * <code>repeated .tensorflow.RunMetadata.FunctionGraphs function_graphs = 4;</code>
+     */
+    public org.tensorflow.proto.framework.RunMetadata.FunctionGraphs.Builder addFunctionGraphsBuilder() {
+      return getFunctionGraphsFieldBuilder().addBuilder(
+          org.tensorflow.proto.framework.RunMetadata.FunctionGraphs.getDefaultInstance());
+    }
+    /**
+     * <pre>
+     * This is only populated for graphs that are run as functions in TensorFlow
+     * V2. There will be an entry below for each function that is traced.
+     * The main use cases of the post_optimization_graph and the partition_graphs
+     * is to give the caller insight into the graphs that were actually run by the
+     * runtime. Additional information (such as those in step_stats) will match
+     * these graphs.
+     * We also include the pre_optimization_graph since it is usually easier to
+     * read, and is helpful in situations where the caller wants to get a high
+     * level idea of what the built graph looks like (since the various graph
+     * optimization passes might change the structure of the graph significantly).
+     * </pre>
+     *
+     * <code>repeated .tensorflow.RunMetadata.FunctionGraphs function_graphs = 4;</code>
+     */
+    public org.tensorflow.proto.framework.RunMetadata.FunctionGraphs.Builder addFunctionGraphsBuilder(
+        int index) {
+      return getFunctionGraphsFieldBuilder().addBuilder(
+          index, org.tensorflow.proto.framework.RunMetadata.FunctionGraphs.getDefaultInstance());
+    }
+    /**
+     * <pre>
+     * This is only populated for graphs that are run as functions in TensorFlow
+     * V2. There will be an entry below for each function that is traced.
+     * The main use cases of the post_optimization_graph and the partition_graphs
+     * is to give the caller insight into the graphs that were actually run by the
+     * runtime. Additional information (such as those in step_stats) will match
+     * these graphs.
+     * We also include the pre_optimization_graph since it is usually easier to
+     * read, and is helpful in situations where the caller wants to get a high
+     * level idea of what the built graph looks like (since the various graph
+     * optimization passes might change the structure of the graph significantly).
+     * </pre>
+     *
+     * <code>repeated .tensorflow.RunMetadata.FunctionGraphs function_graphs = 4;</code>
+     */
+    public java.util.List<org.tensorflow.proto.framework.RunMetadata.FunctionGraphs.Builder> 
+         getFunctionGraphsBuilderList() {
+      return getFunctionGraphsFieldBuilder().getBuilderList();
+    }
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.framework.RunMetadata.FunctionGraphs, org.tensorflow.proto.framework.RunMetadata.FunctionGraphs.Builder, org.tensorflow.proto.framework.RunMetadata.FunctionGraphsOrBuilder> 
+        getFunctionGraphsFieldBuilder() {
+      if (functionGraphsBuilder_ == null) {
+        functionGraphsBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
+            org.tensorflow.proto.framework.RunMetadata.FunctionGraphs, org.tensorflow.proto.framework.RunMetadata.FunctionGraphs.Builder, org.tensorflow.proto.framework.RunMetadata.FunctionGraphsOrBuilder>(
+                functionGraphs_,
+                ((bitField0_ & 0x00000002) != 0),
+                getParentForChildren(),
+                isClean());
+        functionGraphs_ = null;
+      }
+      return functionGraphsBuilder_;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.RunMetadata)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.RunMetadata)
+  private static final org.tensorflow.proto.framework.RunMetadata DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.RunMetadata();
+  }
+
+  public static org.tensorflow.proto.framework.RunMetadata getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<RunMetadata>
+      PARSER = new com.google.protobuf.AbstractParser<RunMetadata>() {
+    @java.lang.Override
+    public RunMetadata parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new RunMetadata(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<RunMetadata> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<RunMetadata> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.framework.RunMetadata getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/RunMetadataOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/RunMetadataOrBuilder.java
new file mode 100644
index 00000000000..21e15741316
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/RunMetadataOrBuilder.java
@@ -0,0 +1,198 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/config.proto
+
+package org.tensorflow.proto.framework;
+
+public interface RunMetadataOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.RunMetadata)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <pre>
+   * Statistics traced for this step. Populated if tracing is turned on via the
+   * "RunOptions" proto.
+   * EXPERIMENTAL: The format and set of events may change in future versions.
+   * </pre>
+   *
+   * <code>.tensorflow.StepStats step_stats = 1;</code>
+   */
+  boolean hasStepStats();
+  /**
+   * <pre>
+   * Statistics traced for this step. Populated if tracing is turned on via the
+   * "RunOptions" proto.
+   * EXPERIMENTAL: The format and set of events may change in future versions.
+   * </pre>
+   *
+   * <code>.tensorflow.StepStats step_stats = 1;</code>
+   */
+  org.tensorflow.proto.framework.StepStats getStepStats();
+  /**
+   * <pre>
+   * Statistics traced for this step. Populated if tracing is turned on via the
+   * "RunOptions" proto.
+   * EXPERIMENTAL: The format and set of events may change in future versions.
+   * </pre>
+   *
+   * <code>.tensorflow.StepStats step_stats = 1;</code>
+   */
+  org.tensorflow.proto.framework.StepStatsOrBuilder getStepStatsOrBuilder();
+
+  /**
+   * <pre>
+   * The cost graph for the computation defined by the run call.
+   * </pre>
+   *
+   * <code>.tensorflow.CostGraphDef cost_graph = 2;</code>
+   */
+  boolean hasCostGraph();
+  /**
+   * <pre>
+   * The cost graph for the computation defined by the run call.
+   * </pre>
+   *
+   * <code>.tensorflow.CostGraphDef cost_graph = 2;</code>
+   */
+  org.tensorflow.proto.framework.CostGraphDef getCostGraph();
+  /**
+   * <pre>
+   * The cost graph for the computation defined by the run call.
+   * </pre>
+   *
+   * <code>.tensorflow.CostGraphDef cost_graph = 2;</code>
+   */
+  org.tensorflow.proto.framework.CostGraphDefOrBuilder getCostGraphOrBuilder();
+
+  /**
+   * <pre>
+   * Graphs of the partitions executed by executors.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.GraphDef partition_graphs = 3;</code>
+   */
+  java.util.List<org.tensorflow.proto.framework.GraphDef> 
+      getPartitionGraphsList();
+  /**
+   * <pre>
+   * Graphs of the partitions executed by executors.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.GraphDef partition_graphs = 3;</code>
+   */
+  org.tensorflow.proto.framework.GraphDef getPartitionGraphs(int index);
+  /**
+   * <pre>
+   * Graphs of the partitions executed by executors.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.GraphDef partition_graphs = 3;</code>
+   */
+  int getPartitionGraphsCount();
+  /**
+   * <pre>
+   * Graphs of the partitions executed by executors.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.GraphDef partition_graphs = 3;</code>
+   */
+  java.util.List<? extends org.tensorflow.proto.framework.GraphDefOrBuilder> 
+      getPartitionGraphsOrBuilderList();
+  /**
+   * <pre>
+   * Graphs of the partitions executed by executors.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.GraphDef partition_graphs = 3;</code>
+   */
+  org.tensorflow.proto.framework.GraphDefOrBuilder getPartitionGraphsOrBuilder(
+      int index);
+
+  /**
+   * <pre>
+   * This is only populated for graphs that are run as functions in TensorFlow
+   * V2. There will be an entry below for each function that is traced.
+   * The main use cases of the post_optimization_graph and the partition_graphs
+   * is to give the caller insight into the graphs that were actually run by the
+   * runtime. Additional information (such as those in step_stats) will match
+   * these graphs.
+   * We also include the pre_optimization_graph since it is usually easier to
+   * read, and is helpful in situations where the caller wants to get a high
+   * level idea of what the built graph looks like (since the various graph
+   * optimization passes might change the structure of the graph significantly).
+   * </pre>
+   *
+   * <code>repeated .tensorflow.RunMetadata.FunctionGraphs function_graphs = 4;</code>
+   */
+  java.util.List<org.tensorflow.proto.framework.RunMetadata.FunctionGraphs> 
+      getFunctionGraphsList();
+  /**
+   * <pre>
+   * This is only populated for graphs that are run as functions in TensorFlow
+   * V2. There will be an entry below for each function that is traced.
+   * The main use cases of the post_optimization_graph and the partition_graphs
+   * is to give the caller insight into the graphs that were actually run by the
+   * runtime. Additional information (such as those in step_stats) will match
+   * these graphs.
+   * We also include the pre_optimization_graph since it is usually easier to
+   * read, and is helpful in situations where the caller wants to get a high
+   * level idea of what the built graph looks like (since the various graph
+   * optimization passes might change the structure of the graph significantly).
+   * </pre>
+   *
+   * <code>repeated .tensorflow.RunMetadata.FunctionGraphs function_graphs = 4;</code>
+   */
+  org.tensorflow.proto.framework.RunMetadata.FunctionGraphs getFunctionGraphs(int index);
+  /**
+   * <pre>
+   * This is only populated for graphs that are run as functions in TensorFlow
+   * V2. There will be an entry below for each function that is traced.
+   * The main use cases of the post_optimization_graph and the partition_graphs
+   * is to give the caller insight into the graphs that were actually run by the
+   * runtime. Additional information (such as those in step_stats) will match
+   * these graphs.
+   * We also include the pre_optimization_graph since it is usually easier to
+   * read, and is helpful in situations where the caller wants to get a high
+   * level idea of what the built graph looks like (since the various graph
+   * optimization passes might change the structure of the graph significantly).
+   * </pre>
+   *
+   * <code>repeated .tensorflow.RunMetadata.FunctionGraphs function_graphs = 4;</code>
+   */
+  int getFunctionGraphsCount();
+  /**
+   * <pre>
+   * This is only populated for graphs that are run as functions in TensorFlow
+   * V2. There will be an entry below for each function that is traced.
+   * The main use cases of the post_optimization_graph and the partition_graphs
+   * is to give the caller insight into the graphs that were actually run by the
+   * runtime. Additional information (such as those in step_stats) will match
+   * these graphs.
+   * We also include the pre_optimization_graph since it is usually easier to
+   * read, and is helpful in situations where the caller wants to get a high
+   * level idea of what the built graph looks like (since the various graph
+   * optimization passes might change the structure of the graph significantly).
+   * </pre>
+   *
+   * <code>repeated .tensorflow.RunMetadata.FunctionGraphs function_graphs = 4;</code>
+   */
+  java.util.List<? extends org.tensorflow.proto.framework.RunMetadata.FunctionGraphsOrBuilder> 
+      getFunctionGraphsOrBuilderList();
+  /**
+   * <pre>
+   * This is only populated for graphs that are run as functions in TensorFlow
+   * V2. There will be an entry below for each function that is traced.
+   * The main use cases of the post_optimization_graph and the partition_graphs
+   * is to give the caller insight into the graphs that were actually run by the
+   * runtime. Additional information (such as those in step_stats) will match
+   * these graphs.
+   * We also include the pre_optimization_graph since it is usually easier to
+   * read, and is helpful in situations where the caller wants to get a high
+   * level idea of what the built graph looks like (since the various graph
+   * optimization passes might change the structure of the graph significantly).
+   * </pre>
+   *
+   * <code>repeated .tensorflow.RunMetadata.FunctionGraphs function_graphs = 4;</code>
+   */
+  org.tensorflow.proto.framework.RunMetadata.FunctionGraphsOrBuilder getFunctionGraphsOrBuilder(
+      int index);
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/RunOptions.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/RunOptions.java
new file mode 100644
index 00000000000..9ccdb06d2b6
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/RunOptions.java
@@ -0,0 +1,2708 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/config.proto
+
+package org.tensorflow.proto.framework;
+
+/**
+ * <pre>
+ * Options for a single Run() call.
+ * </pre>
+ *
+ * Protobuf type {@code tensorflow.RunOptions}
+ */
+public  final class RunOptions extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.RunOptions)
+    RunOptionsOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use RunOptions.newBuilder() to construct.
+  private RunOptions(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private RunOptions() {
+    traceLevel_ = 0;
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new RunOptions();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private RunOptions(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 8: {
+            int rawValue = input.readEnum();
+
+            traceLevel_ = rawValue;
+            break;
+          }
+          case 16: {
+
+            timeoutInMs_ = input.readInt64();
+            break;
+          }
+          case 24: {
+
+            interOpThreadPool_ = input.readInt32();
+            break;
+          }
+          case 40: {
+
+            outputPartitionGraphs_ = input.readBool();
+            break;
+          }
+          case 50: {
+            org.tensorflow.proto.framework.DebugOptions.Builder subBuilder = null;
+            if (debugOptions_ != null) {
+              subBuilder = debugOptions_.toBuilder();
+            }
+            debugOptions_ = input.readMessage(org.tensorflow.proto.framework.DebugOptions.parser(), extensionRegistry);
+            if (subBuilder != null) {
+              subBuilder.mergeFrom(debugOptions_);
+              debugOptions_ = subBuilder.buildPartial();
+            }
+
+            break;
+          }
+          case 56: {
+
+            reportTensorAllocationsUponOom_ = input.readBool();
+            break;
+          }
+          case 66: {
+            org.tensorflow.proto.framework.RunOptions.Experimental.Builder subBuilder = null;
+            if (experimental_ != null) {
+              subBuilder = experimental_.toBuilder();
+            }
+            experimental_ = input.readMessage(org.tensorflow.proto.framework.RunOptions.Experimental.parser(), extensionRegistry);
+            if (subBuilder != null) {
+              subBuilder.mergeFrom(experimental_);
+              experimental_ = subBuilder.buildPartial();
+            }
+
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_RunOptions_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_RunOptions_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.framework.RunOptions.class, org.tensorflow.proto.framework.RunOptions.Builder.class);
+  }
+
+  /**
+   * <pre>
+   * TODO(pbar) Turn this into a TraceOptions proto which allows
+   * tracing to be controlled in a more orthogonal manner?
+   * </pre>
+   *
+   * Protobuf enum {@code tensorflow.RunOptions.TraceLevel}
+   */
+  public enum TraceLevel
+      implements com.google.protobuf.ProtocolMessageEnum {
+    /**
+     * <code>NO_TRACE = 0;</code>
+     */
+    NO_TRACE(0),
+    /**
+     * <code>SOFTWARE_TRACE = 1;</code>
+     */
+    SOFTWARE_TRACE(1),
+    /**
+     * <code>HARDWARE_TRACE = 2;</code>
+     */
+    HARDWARE_TRACE(2),
+    /**
+     * <code>FULL_TRACE = 3;</code>
+     */
+    FULL_TRACE(3),
+    UNRECOGNIZED(-1),
+    ;
+
+    /**
+     * <code>NO_TRACE = 0;</code>
+     */
+    public static final int NO_TRACE_VALUE = 0;
+    /**
+     * <code>SOFTWARE_TRACE = 1;</code>
+     */
+    public static final int SOFTWARE_TRACE_VALUE = 1;
+    /**
+     * <code>HARDWARE_TRACE = 2;</code>
+     */
+    public static final int HARDWARE_TRACE_VALUE = 2;
+    /**
+     * <code>FULL_TRACE = 3;</code>
+     */
+    public static final int FULL_TRACE_VALUE = 3;
+
+
+    public final int getNumber() {
+      if (this == UNRECOGNIZED) {
+        throw new java.lang.IllegalArgumentException(
+            "Can't get the number of an unknown enum value.");
+      }
+      return value;
+    }
+
+    /**
+     * @deprecated Use {@link #forNumber(int)} instead.
+     */
+    @java.lang.Deprecated
+    public static TraceLevel valueOf(int value) {
+      return forNumber(value);
+    }
+
+    public static TraceLevel forNumber(int value) {
+      switch (value) {
+        case 0: return NO_TRACE;
+        case 1: return SOFTWARE_TRACE;
+        case 2: return HARDWARE_TRACE;
+        case 3: return FULL_TRACE;
+        default: return null;
+      }
+    }
+
+    public static com.google.protobuf.Internal.EnumLiteMap<TraceLevel>
+        internalGetValueMap() {
+      return internalValueMap;
+    }
+    private static final com.google.protobuf.Internal.EnumLiteMap<
+        TraceLevel> internalValueMap =
+          new com.google.protobuf.Internal.EnumLiteMap<TraceLevel>() {
+            public TraceLevel findValueByNumber(int number) {
+              return TraceLevel.forNumber(number);
+            }
+          };
+
+    public final com.google.protobuf.Descriptors.EnumValueDescriptor
+        getValueDescriptor() {
+      return getDescriptor().getValues().get(ordinal());
+    }
+    public final com.google.protobuf.Descriptors.EnumDescriptor
+        getDescriptorForType() {
+      return getDescriptor();
+    }
+    public static final com.google.protobuf.Descriptors.EnumDescriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.RunOptions.getDescriptor().getEnumTypes().get(0);
+    }
+
+    private static final TraceLevel[] VALUES = values();
+
+    public static TraceLevel valueOf(
+        com.google.protobuf.Descriptors.EnumValueDescriptor desc) {
+      if (desc.getType() != getDescriptor()) {
+        throw new java.lang.IllegalArgumentException(
+          "EnumValueDescriptor is not for this type.");
+      }
+      if (desc.getIndex() == -1) {
+        return UNRECOGNIZED;
+      }
+      return VALUES[desc.getIndex()];
+    }
+
+    private final int value;
+
+    private TraceLevel(int value) {
+      this.value = value;
+    }
+
+    // @@protoc_insertion_point(enum_scope:tensorflow.RunOptions.TraceLevel)
+  }
+
+  public interface ExperimentalOrBuilder extends
+      // @@protoc_insertion_point(interface_extends:tensorflow.RunOptions.Experimental)
+      com.google.protobuf.MessageOrBuilder {
+
+    /**
+     * <pre>
+     * If non-zero, declares that this graph is going to use collective
+     * ops and must synchronize step_ids with any other graph with this
+     * same group_key value (in a distributed computation where tasks
+     * run disjoint graphs).
+     * </pre>
+     *
+     * <code>int64 collective_graph_key = 1;</code>
+     */
+    long getCollectiveGraphKey();
+
+    /**
+     * <pre>
+     * If true, then operations (using the inter-op pool) across all
+     * session::run() calls will be centrally scheduled, optimizing for (median
+     * and tail) latency.
+     * Consider using this option for CPU-bound workloads like inference.
+     * </pre>
+     *
+     * <code>bool use_run_handler_pool = 2;</code>
+     */
+    boolean getUseRunHandlerPool();
+
+    /**
+     * <code>.tensorflow.RunOptions.Experimental.RunHandlerPoolOptions run_handler_pool_options = 3;</code>
+     */
+    boolean hasRunHandlerPoolOptions();
+    /**
+     * <code>.tensorflow.RunOptions.Experimental.RunHandlerPoolOptions run_handler_pool_options = 3;</code>
+     */
+    org.tensorflow.proto.framework.RunOptions.Experimental.RunHandlerPoolOptions getRunHandlerPoolOptions();
+    /**
+     * <code>.tensorflow.RunOptions.Experimental.RunHandlerPoolOptions run_handler_pool_options = 3;</code>
+     */
+    org.tensorflow.proto.framework.RunOptions.Experimental.RunHandlerPoolOptionsOrBuilder getRunHandlerPoolOptionsOrBuilder();
+  }
+  /**
+   * <pre>
+   * Everything inside Experimental is subject to change and is not subject
+   * to API stability guarantees in
+   * https://www.tensorflow.org/guide/version_compat.
+   * </pre>
+   *
+   * Protobuf type {@code tensorflow.RunOptions.Experimental}
+   */
+  public  static final class Experimental extends
+      com.google.protobuf.GeneratedMessageV3 implements
+      // @@protoc_insertion_point(message_implements:tensorflow.RunOptions.Experimental)
+      ExperimentalOrBuilder {
+  private static final long serialVersionUID = 0L;
+    // Use Experimental.newBuilder() to construct.
+    private Experimental(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+      super(builder);
+    }
+    private Experimental() {
+    }
+
+    @java.lang.Override
+    @SuppressWarnings({"unused"})
+    protected java.lang.Object newInstance(
+        UnusedPrivateParameter unused) {
+      return new Experimental();
+    }
+
+    @java.lang.Override
+    public final com.google.protobuf.UnknownFieldSet
+    getUnknownFields() {
+      return this.unknownFields;
+    }
+    private Experimental(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      this();
+      if (extensionRegistry == null) {
+        throw new java.lang.NullPointerException();
+      }
+      com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+          com.google.protobuf.UnknownFieldSet.newBuilder();
+      try {
+        boolean done = false;
+        while (!done) {
+          int tag = input.readTag();
+          switch (tag) {
+            case 0:
+              done = true;
+              break;
+            case 8: {
+
+              collectiveGraphKey_ = input.readInt64();
+              break;
+            }
+            case 16: {
+
+              useRunHandlerPool_ = input.readBool();
+              break;
+            }
+            case 26: {
+              org.tensorflow.proto.framework.RunOptions.Experimental.RunHandlerPoolOptions.Builder subBuilder = null;
+              if (runHandlerPoolOptions_ != null) {
+                subBuilder = runHandlerPoolOptions_.toBuilder();
+              }
+              runHandlerPoolOptions_ = input.readMessage(org.tensorflow.proto.framework.RunOptions.Experimental.RunHandlerPoolOptions.parser(), extensionRegistry);
+              if (subBuilder != null) {
+                subBuilder.mergeFrom(runHandlerPoolOptions_);
+                runHandlerPoolOptions_ = subBuilder.buildPartial();
+              }
+
+              break;
+            }
+            default: {
+              if (!parseUnknownField(
+                  input, unknownFields, extensionRegistry, tag)) {
+                done = true;
+              }
+              break;
+            }
+          }
+        }
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        throw e.setUnfinishedMessage(this);
+      } catch (java.io.IOException e) {
+        throw new com.google.protobuf.InvalidProtocolBufferException(
+            e).setUnfinishedMessage(this);
+      } finally {
+        this.unknownFields = unknownFields.build();
+        makeExtensionsImmutable();
+      }
+    }
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_RunOptions_Experimental_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_RunOptions_Experimental_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.RunOptions.Experimental.class, org.tensorflow.proto.framework.RunOptions.Experimental.Builder.class);
+    }
+
+    public interface RunHandlerPoolOptionsOrBuilder extends
+        // @@protoc_insertion_point(interface_extends:tensorflow.RunOptions.Experimental.RunHandlerPoolOptions)
+        com.google.protobuf.MessageOrBuilder {
+
+      /**
+       * <pre>
+       * Priority of the request. The run handler thread pool will schedule ops
+       * based on the priority number. The larger number means higher priority.
+       * </pre>
+       *
+       * <code>int64 priority = 1;</code>
+       */
+      long getPriority();
+    }
+    /**
+     * <pre>
+     * Options for run handler thread pool.
+     * </pre>
+     *
+     * Protobuf type {@code tensorflow.RunOptions.Experimental.RunHandlerPoolOptions}
+     */
+    public  static final class RunHandlerPoolOptions extends
+        com.google.protobuf.GeneratedMessageV3 implements
+        // @@protoc_insertion_point(message_implements:tensorflow.RunOptions.Experimental.RunHandlerPoolOptions)
+        RunHandlerPoolOptionsOrBuilder {
+    private static final long serialVersionUID = 0L;
+      // Use RunHandlerPoolOptions.newBuilder() to construct.
+      private RunHandlerPoolOptions(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+        super(builder);
+      }
+      private RunHandlerPoolOptions() {
+      }
+
+      @java.lang.Override
+      @SuppressWarnings({"unused"})
+      protected java.lang.Object newInstance(
+          UnusedPrivateParameter unused) {
+        return new RunHandlerPoolOptions();
+      }
+
+      @java.lang.Override
+      public final com.google.protobuf.UnknownFieldSet
+      getUnknownFields() {
+        return this.unknownFields;
+      }
+      private RunHandlerPoolOptions(
+          com.google.protobuf.CodedInputStream input,
+          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+          throws com.google.protobuf.InvalidProtocolBufferException {
+        this();
+        if (extensionRegistry == null) {
+          throw new java.lang.NullPointerException();
+        }
+        com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+            com.google.protobuf.UnknownFieldSet.newBuilder();
+        try {
+          boolean done = false;
+          while (!done) {
+            int tag = input.readTag();
+            switch (tag) {
+              case 0:
+                done = true;
+                break;
+              case 8: {
+
+                priority_ = input.readInt64();
+                break;
+              }
+              default: {
+                if (!parseUnknownField(
+                    input, unknownFields, extensionRegistry, tag)) {
+                  done = true;
+                }
+                break;
+              }
+            }
+          }
+        } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+          throw e.setUnfinishedMessage(this);
+        } catch (java.io.IOException e) {
+          throw new com.google.protobuf.InvalidProtocolBufferException(
+              e).setUnfinishedMessage(this);
+        } finally {
+          this.unknownFields = unknownFields.build();
+          makeExtensionsImmutable();
+        }
+      }
+      public static final com.google.protobuf.Descriptors.Descriptor
+          getDescriptor() {
+        return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_RunOptions_Experimental_RunHandlerPoolOptions_descriptor;
+      }
+
+      @java.lang.Override
+      protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+          internalGetFieldAccessorTable() {
+        return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_RunOptions_Experimental_RunHandlerPoolOptions_fieldAccessorTable
+            .ensureFieldAccessorsInitialized(
+                org.tensorflow.proto.framework.RunOptions.Experimental.RunHandlerPoolOptions.class, org.tensorflow.proto.framework.RunOptions.Experimental.RunHandlerPoolOptions.Builder.class);
+      }
+
+      public static final int PRIORITY_FIELD_NUMBER = 1;
+      private long priority_;
+      /**
+       * <pre>
+       * Priority of the request. The run handler thread pool will schedule ops
+       * based on the priority number. The larger number means higher priority.
+       * </pre>
+       *
+       * <code>int64 priority = 1;</code>
+       */
+      public long getPriority() {
+        return priority_;
+      }
+
+      private byte memoizedIsInitialized = -1;
+      @java.lang.Override
+      public final boolean isInitialized() {
+        byte isInitialized = memoizedIsInitialized;
+        if (isInitialized == 1) return true;
+        if (isInitialized == 0) return false;
+
+        memoizedIsInitialized = 1;
+        return true;
+      }
+
+      @java.lang.Override
+      public void writeTo(com.google.protobuf.CodedOutputStream output)
+                          throws java.io.IOException {
+        if (priority_ != 0L) {
+          output.writeInt64(1, priority_);
+        }
+        unknownFields.writeTo(output);
+      }
+
+      @java.lang.Override
+      public int getSerializedSize() {
+        int size = memoizedSize;
+        if (size != -1) return size;
+
+        size = 0;
+        if (priority_ != 0L) {
+          size += com.google.protobuf.CodedOutputStream
+            .computeInt64Size(1, priority_);
+        }
+        size += unknownFields.getSerializedSize();
+        memoizedSize = size;
+        return size;
+      }
+
+      @java.lang.Override
+      public boolean equals(final java.lang.Object obj) {
+        if (obj == this) {
+         return true;
+        }
+        if (!(obj instanceof org.tensorflow.proto.framework.RunOptions.Experimental.RunHandlerPoolOptions)) {
+          return super.equals(obj);
+        }
+        org.tensorflow.proto.framework.RunOptions.Experimental.RunHandlerPoolOptions other = (org.tensorflow.proto.framework.RunOptions.Experimental.RunHandlerPoolOptions) obj;
+
+        if (getPriority()
+            != other.getPriority()) return false;
+        if (!unknownFields.equals(other.unknownFields)) return false;
+        return true;
+      }
+
+      @java.lang.Override
+      public int hashCode() {
+        if (memoizedHashCode != 0) {
+          return memoizedHashCode;
+        }
+        int hash = 41;
+        hash = (19 * hash) + getDescriptor().hashCode();
+        hash = (37 * hash) + PRIORITY_FIELD_NUMBER;
+        hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+            getPriority());
+        hash = (29 * hash) + unknownFields.hashCode();
+        memoizedHashCode = hash;
+        return hash;
+      }
+
+      public static org.tensorflow.proto.framework.RunOptions.Experimental.RunHandlerPoolOptions parseFrom(
+          java.nio.ByteBuffer data)
+          throws com.google.protobuf.InvalidProtocolBufferException {
+        return PARSER.parseFrom(data);
+      }
+      public static org.tensorflow.proto.framework.RunOptions.Experimental.RunHandlerPoolOptions parseFrom(
+          java.nio.ByteBuffer data,
+          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+          throws com.google.protobuf.InvalidProtocolBufferException {
+        return PARSER.parseFrom(data, extensionRegistry);
+      }
+      public static org.tensorflow.proto.framework.RunOptions.Experimental.RunHandlerPoolOptions parseFrom(
+          com.google.protobuf.ByteString data)
+          throws com.google.protobuf.InvalidProtocolBufferException {
+        return PARSER.parseFrom(data);
+      }
+      public static org.tensorflow.proto.framework.RunOptions.Experimental.RunHandlerPoolOptions parseFrom(
+          com.google.protobuf.ByteString data,
+          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+          throws com.google.protobuf.InvalidProtocolBufferException {
+        return PARSER.parseFrom(data, extensionRegistry);
+      }
+      public static org.tensorflow.proto.framework.RunOptions.Experimental.RunHandlerPoolOptions parseFrom(byte[] data)
+          throws com.google.protobuf.InvalidProtocolBufferException {
+        return PARSER.parseFrom(data);
+      }
+      public static org.tensorflow.proto.framework.RunOptions.Experimental.RunHandlerPoolOptions parseFrom(
+          byte[] data,
+          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+          throws com.google.protobuf.InvalidProtocolBufferException {
+        return PARSER.parseFrom(data, extensionRegistry);
+      }
+      public static org.tensorflow.proto.framework.RunOptions.Experimental.RunHandlerPoolOptions parseFrom(java.io.InputStream input)
+          throws java.io.IOException {
+        return com.google.protobuf.GeneratedMessageV3
+            .parseWithIOException(PARSER, input);
+      }
+      public static org.tensorflow.proto.framework.RunOptions.Experimental.RunHandlerPoolOptions parseFrom(
+          java.io.InputStream input,
+          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+          throws java.io.IOException {
+        return com.google.protobuf.GeneratedMessageV3
+            .parseWithIOException(PARSER, input, extensionRegistry);
+      }
+      public static org.tensorflow.proto.framework.RunOptions.Experimental.RunHandlerPoolOptions parseDelimitedFrom(java.io.InputStream input)
+          throws java.io.IOException {
+        return com.google.protobuf.GeneratedMessageV3
+            .parseDelimitedWithIOException(PARSER, input);
+      }
+      public static org.tensorflow.proto.framework.RunOptions.Experimental.RunHandlerPoolOptions parseDelimitedFrom(
+          java.io.InputStream input,
+          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+          throws java.io.IOException {
+        return com.google.protobuf.GeneratedMessageV3
+            .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+      }
+      public static org.tensorflow.proto.framework.RunOptions.Experimental.RunHandlerPoolOptions parseFrom(
+          com.google.protobuf.CodedInputStream input)
+          throws java.io.IOException {
+        return com.google.protobuf.GeneratedMessageV3
+            .parseWithIOException(PARSER, input);
+      }
+      public static org.tensorflow.proto.framework.RunOptions.Experimental.RunHandlerPoolOptions parseFrom(
+          com.google.protobuf.CodedInputStream input,
+          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+          throws java.io.IOException {
+        return com.google.protobuf.GeneratedMessageV3
+            .parseWithIOException(PARSER, input, extensionRegistry);
+      }
+
+      @java.lang.Override
+      public Builder newBuilderForType() { return newBuilder(); }
+      public static Builder newBuilder() {
+        return DEFAULT_INSTANCE.toBuilder();
+      }
+      public static Builder newBuilder(org.tensorflow.proto.framework.RunOptions.Experimental.RunHandlerPoolOptions prototype) {
+        return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+      }
+      @java.lang.Override
+      public Builder toBuilder() {
+        return this == DEFAULT_INSTANCE
+            ? new Builder() : new Builder().mergeFrom(this);
+      }
+
+      @java.lang.Override
+      protected Builder newBuilderForType(
+          com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+        Builder builder = new Builder(parent);
+        return builder;
+      }
+      /**
+       * <pre>
+       * Options for run handler thread pool.
+       * </pre>
+       *
+       * Protobuf type {@code tensorflow.RunOptions.Experimental.RunHandlerPoolOptions}
+       */
+      public static final class Builder extends
+          com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+          // @@protoc_insertion_point(builder_implements:tensorflow.RunOptions.Experimental.RunHandlerPoolOptions)
+          org.tensorflow.proto.framework.RunOptions.Experimental.RunHandlerPoolOptionsOrBuilder {
+        public static final com.google.protobuf.Descriptors.Descriptor
+            getDescriptor() {
+          return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_RunOptions_Experimental_RunHandlerPoolOptions_descriptor;
+        }
+
+        @java.lang.Override
+        protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+            internalGetFieldAccessorTable() {
+          return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_RunOptions_Experimental_RunHandlerPoolOptions_fieldAccessorTable
+              .ensureFieldAccessorsInitialized(
+                  org.tensorflow.proto.framework.RunOptions.Experimental.RunHandlerPoolOptions.class, org.tensorflow.proto.framework.RunOptions.Experimental.RunHandlerPoolOptions.Builder.class);
+        }
+
+        // Construct using org.tensorflow.proto.framework.RunOptions.Experimental.RunHandlerPoolOptions.newBuilder()
+        private Builder() {
+          maybeForceBuilderInitialization();
+        }
+
+        private Builder(
+            com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+          super(parent);
+          maybeForceBuilderInitialization();
+        }
+        private void maybeForceBuilderInitialization() {
+          if (com.google.protobuf.GeneratedMessageV3
+                  .alwaysUseFieldBuilders) {
+          }
+        }
+        @java.lang.Override
+        public Builder clear() {
+          super.clear();
+          priority_ = 0L;
+
+          return this;
+        }
+
+        @java.lang.Override
+        public com.google.protobuf.Descriptors.Descriptor
+            getDescriptorForType() {
+          return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_RunOptions_Experimental_RunHandlerPoolOptions_descriptor;
+        }
+
+        @java.lang.Override
+        public org.tensorflow.proto.framework.RunOptions.Experimental.RunHandlerPoolOptions getDefaultInstanceForType() {
+          return org.tensorflow.proto.framework.RunOptions.Experimental.RunHandlerPoolOptions.getDefaultInstance();
+        }
+
+        @java.lang.Override
+        public org.tensorflow.proto.framework.RunOptions.Experimental.RunHandlerPoolOptions build() {
+          org.tensorflow.proto.framework.RunOptions.Experimental.RunHandlerPoolOptions result = buildPartial();
+          if (!result.isInitialized()) {
+            throw newUninitializedMessageException(result);
+          }
+          return result;
+        }
+
+        @java.lang.Override
+        public org.tensorflow.proto.framework.RunOptions.Experimental.RunHandlerPoolOptions buildPartial() {
+          org.tensorflow.proto.framework.RunOptions.Experimental.RunHandlerPoolOptions result = new org.tensorflow.proto.framework.RunOptions.Experimental.RunHandlerPoolOptions(this);
+          result.priority_ = priority_;
+          onBuilt();
+          return result;
+        }
+
+        @java.lang.Override
+        public Builder clone() {
+          return super.clone();
+        }
+        @java.lang.Override
+        public Builder setField(
+            com.google.protobuf.Descriptors.FieldDescriptor field,
+            java.lang.Object value) {
+          return super.setField(field, value);
+        }
+        @java.lang.Override
+        public Builder clearField(
+            com.google.protobuf.Descriptors.FieldDescriptor field) {
+          return super.clearField(field);
+        }
+        @java.lang.Override
+        public Builder clearOneof(
+            com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+          return super.clearOneof(oneof);
+        }
+        @java.lang.Override
+        public Builder setRepeatedField(
+            com.google.protobuf.Descriptors.FieldDescriptor field,
+            int index, java.lang.Object value) {
+          return super.setRepeatedField(field, index, value);
+        }
+        @java.lang.Override
+        public Builder addRepeatedField(
+            com.google.protobuf.Descriptors.FieldDescriptor field,
+            java.lang.Object value) {
+          return super.addRepeatedField(field, value);
+        }
+        @java.lang.Override
+        public Builder mergeFrom(com.google.protobuf.Message other) {
+          if (other instanceof org.tensorflow.proto.framework.RunOptions.Experimental.RunHandlerPoolOptions) {
+            return mergeFrom((org.tensorflow.proto.framework.RunOptions.Experimental.RunHandlerPoolOptions)other);
+          } else {
+            super.mergeFrom(other);
+            return this;
+          }
+        }
+
+        public Builder mergeFrom(org.tensorflow.proto.framework.RunOptions.Experimental.RunHandlerPoolOptions other) {
+          if (other == org.tensorflow.proto.framework.RunOptions.Experimental.RunHandlerPoolOptions.getDefaultInstance()) return this;
+          if (other.getPriority() != 0L) {
+            setPriority(other.getPriority());
+          }
+          this.mergeUnknownFields(other.unknownFields);
+          onChanged();
+          return this;
+        }
+
+        @java.lang.Override
+        public final boolean isInitialized() {
+          return true;
+        }
+
+        @java.lang.Override
+        public Builder mergeFrom(
+            com.google.protobuf.CodedInputStream input,
+            com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+            throws java.io.IOException {
+          org.tensorflow.proto.framework.RunOptions.Experimental.RunHandlerPoolOptions parsedMessage = null;
+          try {
+            parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+          } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+            parsedMessage = (org.tensorflow.proto.framework.RunOptions.Experimental.RunHandlerPoolOptions) e.getUnfinishedMessage();
+            throw e.unwrapIOException();
+          } finally {
+            if (parsedMessage != null) {
+              mergeFrom(parsedMessage);
+            }
+          }
+          return this;
+        }
+
+        private long priority_ ;
+        /**
+         * <pre>
+         * Priority of the request. The run handler thread pool will schedule ops
+         * based on the priority number. The larger number means higher priority.
+         * </pre>
+         *
+         * <code>int64 priority = 1;</code>
+         */
+        public long getPriority() {
+          return priority_;
+        }
+        /**
+         * <pre>
+         * Priority of the request. The run handler thread pool will schedule ops
+         * based on the priority number. The larger number means higher priority.
+         * </pre>
+         *
+         * <code>int64 priority = 1;</code>
+         */
+        public Builder setPriority(long value) {
+          
+          priority_ = value;
+          onChanged();
+          return this;
+        }
+        /**
+         * <pre>
+         * Priority of the request. The run handler thread pool will schedule ops
+         * based on the priority number. The larger number means higher priority.
+         * </pre>
+         *
+         * <code>int64 priority = 1;</code>
+         */
+        public Builder clearPriority() {
+          
+          priority_ = 0L;
+          onChanged();
+          return this;
+        }
+        @java.lang.Override
+        public final Builder setUnknownFields(
+            final com.google.protobuf.UnknownFieldSet unknownFields) {
+          return super.setUnknownFields(unknownFields);
+        }
+
+        @java.lang.Override
+        public final Builder mergeUnknownFields(
+            final com.google.protobuf.UnknownFieldSet unknownFields) {
+          return super.mergeUnknownFields(unknownFields);
+        }
+
+
+        // @@protoc_insertion_point(builder_scope:tensorflow.RunOptions.Experimental.RunHandlerPoolOptions)
+      }
+
+      // @@protoc_insertion_point(class_scope:tensorflow.RunOptions.Experimental.RunHandlerPoolOptions)
+      private static final org.tensorflow.proto.framework.RunOptions.Experimental.RunHandlerPoolOptions DEFAULT_INSTANCE;
+      static {
+        DEFAULT_INSTANCE = new org.tensorflow.proto.framework.RunOptions.Experimental.RunHandlerPoolOptions();
+      }
+
+      public static org.tensorflow.proto.framework.RunOptions.Experimental.RunHandlerPoolOptions getDefaultInstance() {
+        return DEFAULT_INSTANCE;
+      }
+
+      private static final com.google.protobuf.Parser<RunHandlerPoolOptions>
+          PARSER = new com.google.protobuf.AbstractParser<RunHandlerPoolOptions>() {
+        @java.lang.Override
+        public RunHandlerPoolOptions parsePartialFrom(
+            com.google.protobuf.CodedInputStream input,
+            com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+            throws com.google.protobuf.InvalidProtocolBufferException {
+          return new RunHandlerPoolOptions(input, extensionRegistry);
+        }
+      };
+
+      public static com.google.protobuf.Parser<RunHandlerPoolOptions> parser() {
+        return PARSER;
+      }
+
+      @java.lang.Override
+      public com.google.protobuf.Parser<RunHandlerPoolOptions> getParserForType() {
+        return PARSER;
+      }
+
+      @java.lang.Override
+      public org.tensorflow.proto.framework.RunOptions.Experimental.RunHandlerPoolOptions getDefaultInstanceForType() {
+        return DEFAULT_INSTANCE;
+      }
+
+    }
+
+    public static final int COLLECTIVE_GRAPH_KEY_FIELD_NUMBER = 1;
+    private long collectiveGraphKey_;
+    /**
+     * <pre>
+     * If non-zero, declares that this graph is going to use collective
+     * ops and must synchronize step_ids with any other graph with this
+     * same group_key value (in a distributed computation where tasks
+     * run disjoint graphs).
+     * </pre>
+     *
+     * <code>int64 collective_graph_key = 1;</code>
+     */
+    public long getCollectiveGraphKey() {
+      return collectiveGraphKey_;
+    }
+
+    public static final int USE_RUN_HANDLER_POOL_FIELD_NUMBER = 2;
+    private boolean useRunHandlerPool_;
+    /**
+     * <pre>
+     * If true, then operations (using the inter-op pool) across all
+     * session::run() calls will be centrally scheduled, optimizing for (median
+     * and tail) latency.
+     * Consider using this option for CPU-bound workloads like inference.
+     * </pre>
+     *
+     * <code>bool use_run_handler_pool = 2;</code>
+     */
+    public boolean getUseRunHandlerPool() {
+      return useRunHandlerPool_;
+    }
+
+    public static final int RUN_HANDLER_POOL_OPTIONS_FIELD_NUMBER = 3;
+    private org.tensorflow.proto.framework.RunOptions.Experimental.RunHandlerPoolOptions runHandlerPoolOptions_;
+    /**
+     * <code>.tensorflow.RunOptions.Experimental.RunHandlerPoolOptions run_handler_pool_options = 3;</code>
+     */
+    public boolean hasRunHandlerPoolOptions() {
+      return runHandlerPoolOptions_ != null;
+    }
+    /**
+     * <code>.tensorflow.RunOptions.Experimental.RunHandlerPoolOptions run_handler_pool_options = 3;</code>
+     */
+    public org.tensorflow.proto.framework.RunOptions.Experimental.RunHandlerPoolOptions getRunHandlerPoolOptions() {
+      return runHandlerPoolOptions_ == null ? org.tensorflow.proto.framework.RunOptions.Experimental.RunHandlerPoolOptions.getDefaultInstance() : runHandlerPoolOptions_;
+    }
+    /**
+     * <code>.tensorflow.RunOptions.Experimental.RunHandlerPoolOptions run_handler_pool_options = 3;</code>
+     */
+    public org.tensorflow.proto.framework.RunOptions.Experimental.RunHandlerPoolOptionsOrBuilder getRunHandlerPoolOptionsOrBuilder() {
+      return getRunHandlerPoolOptions();
+    }
+
+    private byte memoizedIsInitialized = -1;
+    @java.lang.Override
+    public final boolean isInitialized() {
+      byte isInitialized = memoizedIsInitialized;
+      if (isInitialized == 1) return true;
+      if (isInitialized == 0) return false;
+
+      memoizedIsInitialized = 1;
+      return true;
+    }
+
+    @java.lang.Override
+    public void writeTo(com.google.protobuf.CodedOutputStream output)
+                        throws java.io.IOException {
+      if (collectiveGraphKey_ != 0L) {
+        output.writeInt64(1, collectiveGraphKey_);
+      }
+      if (useRunHandlerPool_ != false) {
+        output.writeBool(2, useRunHandlerPool_);
+      }
+      if (runHandlerPoolOptions_ != null) {
+        output.writeMessage(3, getRunHandlerPoolOptions());
+      }
+      unknownFields.writeTo(output);
+    }
+
+    @java.lang.Override
+    public int getSerializedSize() {
+      int size = memoizedSize;
+      if (size != -1) return size;
+
+      size = 0;
+      if (collectiveGraphKey_ != 0L) {
+        size += com.google.protobuf.CodedOutputStream
+          .computeInt64Size(1, collectiveGraphKey_);
+      }
+      if (useRunHandlerPool_ != false) {
+        size += com.google.protobuf.CodedOutputStream
+          .computeBoolSize(2, useRunHandlerPool_);
+      }
+      if (runHandlerPoolOptions_ != null) {
+        size += com.google.protobuf.CodedOutputStream
+          .computeMessageSize(3, getRunHandlerPoolOptions());
+      }
+      size += unknownFields.getSerializedSize();
+      memoizedSize = size;
+      return size;
+    }
+
+    @java.lang.Override
+    public boolean equals(final java.lang.Object obj) {
+      if (obj == this) {
+       return true;
+      }
+      if (!(obj instanceof org.tensorflow.proto.framework.RunOptions.Experimental)) {
+        return super.equals(obj);
+      }
+      org.tensorflow.proto.framework.RunOptions.Experimental other = (org.tensorflow.proto.framework.RunOptions.Experimental) obj;
+
+      if (getCollectiveGraphKey()
+          != other.getCollectiveGraphKey()) return false;
+      if (getUseRunHandlerPool()
+          != other.getUseRunHandlerPool()) return false;
+      if (hasRunHandlerPoolOptions() != other.hasRunHandlerPoolOptions()) return false;
+      if (hasRunHandlerPoolOptions()) {
+        if (!getRunHandlerPoolOptions()
+            .equals(other.getRunHandlerPoolOptions())) return false;
+      }
+      if (!unknownFields.equals(other.unknownFields)) return false;
+      return true;
+    }
+
+    @java.lang.Override
+    public int hashCode() {
+      if (memoizedHashCode != 0) {
+        return memoizedHashCode;
+      }
+      int hash = 41;
+      hash = (19 * hash) + getDescriptor().hashCode();
+      hash = (37 * hash) + COLLECTIVE_GRAPH_KEY_FIELD_NUMBER;
+      hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+          getCollectiveGraphKey());
+      hash = (37 * hash) + USE_RUN_HANDLER_POOL_FIELD_NUMBER;
+      hash = (53 * hash) + com.google.protobuf.Internal.hashBoolean(
+          getUseRunHandlerPool());
+      if (hasRunHandlerPoolOptions()) {
+        hash = (37 * hash) + RUN_HANDLER_POOL_OPTIONS_FIELD_NUMBER;
+        hash = (53 * hash) + getRunHandlerPoolOptions().hashCode();
+      }
+      hash = (29 * hash) + unknownFields.hashCode();
+      memoizedHashCode = hash;
+      return hash;
+    }
+
+    public static org.tensorflow.proto.framework.RunOptions.Experimental parseFrom(
+        java.nio.ByteBuffer data)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data);
+    }
+    public static org.tensorflow.proto.framework.RunOptions.Experimental parseFrom(
+        java.nio.ByteBuffer data,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.RunOptions.Experimental parseFrom(
+        com.google.protobuf.ByteString data)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data);
+    }
+    public static org.tensorflow.proto.framework.RunOptions.Experimental parseFrom(
+        com.google.protobuf.ByteString data,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.RunOptions.Experimental parseFrom(byte[] data)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data);
+    }
+    public static org.tensorflow.proto.framework.RunOptions.Experimental parseFrom(
+        byte[] data,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.RunOptions.Experimental parseFrom(java.io.InputStream input)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input);
+    }
+    public static org.tensorflow.proto.framework.RunOptions.Experimental parseFrom(
+        java.io.InputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.RunOptions.Experimental parseDelimitedFrom(java.io.InputStream input)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseDelimitedWithIOException(PARSER, input);
+    }
+    public static org.tensorflow.proto.framework.RunOptions.Experimental parseDelimitedFrom(
+        java.io.InputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.RunOptions.Experimental parseFrom(
+        com.google.protobuf.CodedInputStream input)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input);
+    }
+    public static org.tensorflow.proto.framework.RunOptions.Experimental parseFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input, extensionRegistry);
+    }
+
+    @java.lang.Override
+    public Builder newBuilderForType() { return newBuilder(); }
+    public static Builder newBuilder() {
+      return DEFAULT_INSTANCE.toBuilder();
+    }
+    public static Builder newBuilder(org.tensorflow.proto.framework.RunOptions.Experimental prototype) {
+      return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+    }
+    @java.lang.Override
+    public Builder toBuilder() {
+      return this == DEFAULT_INSTANCE
+          ? new Builder() : new Builder().mergeFrom(this);
+    }
+
+    @java.lang.Override
+    protected Builder newBuilderForType(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      Builder builder = new Builder(parent);
+      return builder;
+    }
+    /**
+     * <pre>
+     * Everything inside Experimental is subject to change and is not subject
+     * to API stability guarantees in
+     * https://www.tensorflow.org/guide/version_compat.
+     * </pre>
+     *
+     * Protobuf type {@code tensorflow.RunOptions.Experimental}
+     */
+    public static final class Builder extends
+        com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+        // @@protoc_insertion_point(builder_implements:tensorflow.RunOptions.Experimental)
+        org.tensorflow.proto.framework.RunOptions.ExperimentalOrBuilder {
+      public static final com.google.protobuf.Descriptors.Descriptor
+          getDescriptor() {
+        return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_RunOptions_Experimental_descriptor;
+      }
+
+      @java.lang.Override
+      protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+          internalGetFieldAccessorTable() {
+        return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_RunOptions_Experimental_fieldAccessorTable
+            .ensureFieldAccessorsInitialized(
+                org.tensorflow.proto.framework.RunOptions.Experimental.class, org.tensorflow.proto.framework.RunOptions.Experimental.Builder.class);
+      }
+
+      // Construct using org.tensorflow.proto.framework.RunOptions.Experimental.newBuilder()
+      private Builder() {
+        maybeForceBuilderInitialization();
+      }
+
+      private Builder(
+          com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+        super(parent);
+        maybeForceBuilderInitialization();
+      }
+      private void maybeForceBuilderInitialization() {
+        if (com.google.protobuf.GeneratedMessageV3
+                .alwaysUseFieldBuilders) {
+        }
+      }
+      @java.lang.Override
+      public Builder clear() {
+        super.clear();
+        collectiveGraphKey_ = 0L;
+
+        useRunHandlerPool_ = false;
+
+        if (runHandlerPoolOptionsBuilder_ == null) {
+          runHandlerPoolOptions_ = null;
+        } else {
+          runHandlerPoolOptions_ = null;
+          runHandlerPoolOptionsBuilder_ = null;
+        }
+        return this;
+      }
+
+      @java.lang.Override
+      public com.google.protobuf.Descriptors.Descriptor
+          getDescriptorForType() {
+        return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_RunOptions_Experimental_descriptor;
+      }
+
+      @java.lang.Override
+      public org.tensorflow.proto.framework.RunOptions.Experimental getDefaultInstanceForType() {
+        return org.tensorflow.proto.framework.RunOptions.Experimental.getDefaultInstance();
+      }
+
+      @java.lang.Override
+      public org.tensorflow.proto.framework.RunOptions.Experimental build() {
+        org.tensorflow.proto.framework.RunOptions.Experimental result = buildPartial();
+        if (!result.isInitialized()) {
+          throw newUninitializedMessageException(result);
+        }
+        return result;
+      }
+
+      @java.lang.Override
+      public org.tensorflow.proto.framework.RunOptions.Experimental buildPartial() {
+        org.tensorflow.proto.framework.RunOptions.Experimental result = new org.tensorflow.proto.framework.RunOptions.Experimental(this);
+        result.collectiveGraphKey_ = collectiveGraphKey_;
+        result.useRunHandlerPool_ = useRunHandlerPool_;
+        if (runHandlerPoolOptionsBuilder_ == null) {
+          result.runHandlerPoolOptions_ = runHandlerPoolOptions_;
+        } else {
+          result.runHandlerPoolOptions_ = runHandlerPoolOptionsBuilder_.build();
+        }
+        onBuilt();
+        return result;
+      }
+
+      @java.lang.Override
+      public Builder clone() {
+        return super.clone();
+      }
+      @java.lang.Override
+      public Builder setField(
+          com.google.protobuf.Descriptors.FieldDescriptor field,
+          java.lang.Object value) {
+        return super.setField(field, value);
+      }
+      @java.lang.Override
+      public Builder clearField(
+          com.google.protobuf.Descriptors.FieldDescriptor field) {
+        return super.clearField(field);
+      }
+      @java.lang.Override
+      public Builder clearOneof(
+          com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+        return super.clearOneof(oneof);
+      }
+      @java.lang.Override
+      public Builder setRepeatedField(
+          com.google.protobuf.Descriptors.FieldDescriptor field,
+          int index, java.lang.Object value) {
+        return super.setRepeatedField(field, index, value);
+      }
+      @java.lang.Override
+      public Builder addRepeatedField(
+          com.google.protobuf.Descriptors.FieldDescriptor field,
+          java.lang.Object value) {
+        return super.addRepeatedField(field, value);
+      }
+      @java.lang.Override
+      public Builder mergeFrom(com.google.protobuf.Message other) {
+        if (other instanceof org.tensorflow.proto.framework.RunOptions.Experimental) {
+          return mergeFrom((org.tensorflow.proto.framework.RunOptions.Experimental)other);
+        } else {
+          super.mergeFrom(other);
+          return this;
+        }
+      }
+
+      public Builder mergeFrom(org.tensorflow.proto.framework.RunOptions.Experimental other) {
+        if (other == org.tensorflow.proto.framework.RunOptions.Experimental.getDefaultInstance()) return this;
+        if (other.getCollectiveGraphKey() != 0L) {
+          setCollectiveGraphKey(other.getCollectiveGraphKey());
+        }
+        if (other.getUseRunHandlerPool() != false) {
+          setUseRunHandlerPool(other.getUseRunHandlerPool());
+        }
+        if (other.hasRunHandlerPoolOptions()) {
+          mergeRunHandlerPoolOptions(other.getRunHandlerPoolOptions());
+        }
+        this.mergeUnknownFields(other.unknownFields);
+        onChanged();
+        return this;
+      }
+
+      @java.lang.Override
+      public final boolean isInitialized() {
+        return true;
+      }
+
+      @java.lang.Override
+      public Builder mergeFrom(
+          com.google.protobuf.CodedInputStream input,
+          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+          throws java.io.IOException {
+        org.tensorflow.proto.framework.RunOptions.Experimental parsedMessage = null;
+        try {
+          parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+        } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+          parsedMessage = (org.tensorflow.proto.framework.RunOptions.Experimental) e.getUnfinishedMessage();
+          throw e.unwrapIOException();
+        } finally {
+          if (parsedMessage != null) {
+            mergeFrom(parsedMessage);
+          }
+        }
+        return this;
+      }
+
+      private long collectiveGraphKey_ ;
+      /**
+       * <pre>
+       * If non-zero, declares that this graph is going to use collective
+       * ops and must synchronize step_ids with any other graph with this
+       * same group_key value (in a distributed computation where tasks
+       * run disjoint graphs).
+       * </pre>
+       *
+       * <code>int64 collective_graph_key = 1;</code>
+       */
+      public long getCollectiveGraphKey() {
+        return collectiveGraphKey_;
+      }
+      /**
+       * <pre>
+       * If non-zero, declares that this graph is going to use collective
+       * ops and must synchronize step_ids with any other graph with this
+       * same group_key value (in a distributed computation where tasks
+       * run disjoint graphs).
+       * </pre>
+       *
+       * <code>int64 collective_graph_key = 1;</code>
+       */
+      public Builder setCollectiveGraphKey(long value) {
+        
+        collectiveGraphKey_ = value;
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * If non-zero, declares that this graph is going to use collective
+       * ops and must synchronize step_ids with any other graph with this
+       * same group_key value (in a distributed computation where tasks
+       * run disjoint graphs).
+       * </pre>
+       *
+       * <code>int64 collective_graph_key = 1;</code>
+       */
+      public Builder clearCollectiveGraphKey() {
+        
+        collectiveGraphKey_ = 0L;
+        onChanged();
+        return this;
+      }
+
+      private boolean useRunHandlerPool_ ;
+      /**
+       * <pre>
+       * If true, then operations (using the inter-op pool) across all
+       * session::run() calls will be centrally scheduled, optimizing for (median
+       * and tail) latency.
+       * Consider using this option for CPU-bound workloads like inference.
+       * </pre>
+       *
+       * <code>bool use_run_handler_pool = 2;</code>
+       */
+      public boolean getUseRunHandlerPool() {
+        return useRunHandlerPool_;
+      }
+      /**
+       * <pre>
+       * If true, then operations (using the inter-op pool) across all
+       * session::run() calls will be centrally scheduled, optimizing for (median
+       * and tail) latency.
+       * Consider using this option for CPU-bound workloads like inference.
+       * </pre>
+       *
+       * <code>bool use_run_handler_pool = 2;</code>
+       */
+      public Builder setUseRunHandlerPool(boolean value) {
+        
+        useRunHandlerPool_ = value;
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * If true, then operations (using the inter-op pool) across all
+       * session::run() calls will be centrally scheduled, optimizing for (median
+       * and tail) latency.
+       * Consider using this option for CPU-bound workloads like inference.
+       * </pre>
+       *
+       * <code>bool use_run_handler_pool = 2;</code>
+       */
+      public Builder clearUseRunHandlerPool() {
+        
+        useRunHandlerPool_ = false;
+        onChanged();
+        return this;
+      }
+
+      private org.tensorflow.proto.framework.RunOptions.Experimental.RunHandlerPoolOptions runHandlerPoolOptions_;
+      private com.google.protobuf.SingleFieldBuilderV3<
+          org.tensorflow.proto.framework.RunOptions.Experimental.RunHandlerPoolOptions, org.tensorflow.proto.framework.RunOptions.Experimental.RunHandlerPoolOptions.Builder, org.tensorflow.proto.framework.RunOptions.Experimental.RunHandlerPoolOptionsOrBuilder> runHandlerPoolOptionsBuilder_;
+      /**
+       * <code>.tensorflow.RunOptions.Experimental.RunHandlerPoolOptions run_handler_pool_options = 3;</code>
+       */
+      public boolean hasRunHandlerPoolOptions() {
+        return runHandlerPoolOptionsBuilder_ != null || runHandlerPoolOptions_ != null;
+      }
+      /**
+       * <code>.tensorflow.RunOptions.Experimental.RunHandlerPoolOptions run_handler_pool_options = 3;</code>
+       */
+      public org.tensorflow.proto.framework.RunOptions.Experimental.RunHandlerPoolOptions getRunHandlerPoolOptions() {
+        if (runHandlerPoolOptionsBuilder_ == null) {
+          return runHandlerPoolOptions_ == null ? org.tensorflow.proto.framework.RunOptions.Experimental.RunHandlerPoolOptions.getDefaultInstance() : runHandlerPoolOptions_;
+        } else {
+          return runHandlerPoolOptionsBuilder_.getMessage();
+        }
+      }
+      /**
+       * <code>.tensorflow.RunOptions.Experimental.RunHandlerPoolOptions run_handler_pool_options = 3;</code>
+       */
+      public Builder setRunHandlerPoolOptions(org.tensorflow.proto.framework.RunOptions.Experimental.RunHandlerPoolOptions value) {
+        if (runHandlerPoolOptionsBuilder_ == null) {
+          if (value == null) {
+            throw new NullPointerException();
+          }
+          runHandlerPoolOptions_ = value;
+          onChanged();
+        } else {
+          runHandlerPoolOptionsBuilder_.setMessage(value);
+        }
+
+        return this;
+      }
+      /**
+       * <code>.tensorflow.RunOptions.Experimental.RunHandlerPoolOptions run_handler_pool_options = 3;</code>
+       */
+      public Builder setRunHandlerPoolOptions(
+          org.tensorflow.proto.framework.RunOptions.Experimental.RunHandlerPoolOptions.Builder builderForValue) {
+        if (runHandlerPoolOptionsBuilder_ == null) {
+          runHandlerPoolOptions_ = builderForValue.build();
+          onChanged();
+        } else {
+          runHandlerPoolOptionsBuilder_.setMessage(builderForValue.build());
+        }
+
+        return this;
+      }
+      /**
+       * <code>.tensorflow.RunOptions.Experimental.RunHandlerPoolOptions run_handler_pool_options = 3;</code>
+       */
+      public Builder mergeRunHandlerPoolOptions(org.tensorflow.proto.framework.RunOptions.Experimental.RunHandlerPoolOptions value) {
+        if (runHandlerPoolOptionsBuilder_ == null) {
+          if (runHandlerPoolOptions_ != null) {
+            runHandlerPoolOptions_ =
+              org.tensorflow.proto.framework.RunOptions.Experimental.RunHandlerPoolOptions.newBuilder(runHandlerPoolOptions_).mergeFrom(value).buildPartial();
+          } else {
+            runHandlerPoolOptions_ = value;
+          }
+          onChanged();
+        } else {
+          runHandlerPoolOptionsBuilder_.mergeFrom(value);
+        }
+
+        return this;
+      }
+      /**
+       * <code>.tensorflow.RunOptions.Experimental.RunHandlerPoolOptions run_handler_pool_options = 3;</code>
+       */
+      public Builder clearRunHandlerPoolOptions() {
+        if (runHandlerPoolOptionsBuilder_ == null) {
+          runHandlerPoolOptions_ = null;
+          onChanged();
+        } else {
+          runHandlerPoolOptions_ = null;
+          runHandlerPoolOptionsBuilder_ = null;
+        }
+
+        return this;
+      }
+      /**
+       * <code>.tensorflow.RunOptions.Experimental.RunHandlerPoolOptions run_handler_pool_options = 3;</code>
+       */
+      public org.tensorflow.proto.framework.RunOptions.Experimental.RunHandlerPoolOptions.Builder getRunHandlerPoolOptionsBuilder() {
+        
+        onChanged();
+        return getRunHandlerPoolOptionsFieldBuilder().getBuilder();
+      }
+      /**
+       * <code>.tensorflow.RunOptions.Experimental.RunHandlerPoolOptions run_handler_pool_options = 3;</code>
+       */
+      public org.tensorflow.proto.framework.RunOptions.Experimental.RunHandlerPoolOptionsOrBuilder getRunHandlerPoolOptionsOrBuilder() {
+        if (runHandlerPoolOptionsBuilder_ != null) {
+          return runHandlerPoolOptionsBuilder_.getMessageOrBuilder();
+        } else {
+          return runHandlerPoolOptions_ == null ?
+              org.tensorflow.proto.framework.RunOptions.Experimental.RunHandlerPoolOptions.getDefaultInstance() : runHandlerPoolOptions_;
+        }
+      }
+      /**
+       * <code>.tensorflow.RunOptions.Experimental.RunHandlerPoolOptions run_handler_pool_options = 3;</code>
+       */
+      private com.google.protobuf.SingleFieldBuilderV3<
+          org.tensorflow.proto.framework.RunOptions.Experimental.RunHandlerPoolOptions, org.tensorflow.proto.framework.RunOptions.Experimental.RunHandlerPoolOptions.Builder, org.tensorflow.proto.framework.RunOptions.Experimental.RunHandlerPoolOptionsOrBuilder> 
+          getRunHandlerPoolOptionsFieldBuilder() {
+        if (runHandlerPoolOptionsBuilder_ == null) {
+          runHandlerPoolOptionsBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+              org.tensorflow.proto.framework.RunOptions.Experimental.RunHandlerPoolOptions, org.tensorflow.proto.framework.RunOptions.Experimental.RunHandlerPoolOptions.Builder, org.tensorflow.proto.framework.RunOptions.Experimental.RunHandlerPoolOptionsOrBuilder>(
+                  getRunHandlerPoolOptions(),
+                  getParentForChildren(),
+                  isClean());
+          runHandlerPoolOptions_ = null;
+        }
+        return runHandlerPoolOptionsBuilder_;
+      }
+      @java.lang.Override
+      public final Builder setUnknownFields(
+          final com.google.protobuf.UnknownFieldSet unknownFields) {
+        return super.setUnknownFields(unknownFields);
+      }
+
+      @java.lang.Override
+      public final Builder mergeUnknownFields(
+          final com.google.protobuf.UnknownFieldSet unknownFields) {
+        return super.mergeUnknownFields(unknownFields);
+      }
+
+
+      // @@protoc_insertion_point(builder_scope:tensorflow.RunOptions.Experimental)
+    }
+
+    // @@protoc_insertion_point(class_scope:tensorflow.RunOptions.Experimental)
+    private static final org.tensorflow.proto.framework.RunOptions.Experimental DEFAULT_INSTANCE;
+    static {
+      DEFAULT_INSTANCE = new org.tensorflow.proto.framework.RunOptions.Experimental();
+    }
+
+    public static org.tensorflow.proto.framework.RunOptions.Experimental getDefaultInstance() {
+      return DEFAULT_INSTANCE;
+    }
+
+    private static final com.google.protobuf.Parser<Experimental>
+        PARSER = new com.google.protobuf.AbstractParser<Experimental>() {
+      @java.lang.Override
+      public Experimental parsePartialFrom(
+          com.google.protobuf.CodedInputStream input,
+          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+          throws com.google.protobuf.InvalidProtocolBufferException {
+        return new Experimental(input, extensionRegistry);
+      }
+    };
+
+    public static com.google.protobuf.Parser<Experimental> parser() {
+      return PARSER;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Parser<Experimental> getParserForType() {
+      return PARSER;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.RunOptions.Experimental getDefaultInstanceForType() {
+      return DEFAULT_INSTANCE;
+    }
+
+  }
+
+  public static final int TRACE_LEVEL_FIELD_NUMBER = 1;
+  private int traceLevel_;
+  /**
+   * <code>.tensorflow.RunOptions.TraceLevel trace_level = 1;</code>
+   */
+  public int getTraceLevelValue() {
+    return traceLevel_;
+  }
+  /**
+   * <code>.tensorflow.RunOptions.TraceLevel trace_level = 1;</code>
+   */
+  public org.tensorflow.proto.framework.RunOptions.TraceLevel getTraceLevel() {
+    @SuppressWarnings("deprecation")
+    org.tensorflow.proto.framework.RunOptions.TraceLevel result = org.tensorflow.proto.framework.RunOptions.TraceLevel.valueOf(traceLevel_);
+    return result == null ? org.tensorflow.proto.framework.RunOptions.TraceLevel.UNRECOGNIZED : result;
+  }
+
+  public static final int TIMEOUT_IN_MS_FIELD_NUMBER = 2;
+  private long timeoutInMs_;
+  /**
+   * <pre>
+   * Time to wait for operation to complete in milliseconds.
+   * </pre>
+   *
+   * <code>int64 timeout_in_ms = 2;</code>
+   */
+  public long getTimeoutInMs() {
+    return timeoutInMs_;
+  }
+
+  public static final int INTER_OP_THREAD_POOL_FIELD_NUMBER = 3;
+  private int interOpThreadPool_;
+  /**
+   * <pre>
+   * The thread pool to use, if session_inter_op_thread_pool is configured.
+   * To use the caller thread set this to -1 - this uses the caller thread
+   * to execute Session::Run() and thus avoids a context switch. Using the
+   * caller thread to execute Session::Run() should be done ONLY for simple
+   * graphs, where the overhead of an additional context switch is
+   * comparable with the overhead of Session::Run().
+   * </pre>
+   *
+   * <code>int32 inter_op_thread_pool = 3;</code>
+   */
+  public int getInterOpThreadPool() {
+    return interOpThreadPool_;
+  }
+
+  public static final int OUTPUT_PARTITION_GRAPHS_FIELD_NUMBER = 5;
+  private boolean outputPartitionGraphs_;
+  /**
+   * <pre>
+   * Whether the partition graph(s) executed by the executor(s) should be
+   * outputted via RunMetadata.
+   * </pre>
+   *
+   * <code>bool output_partition_graphs = 5;</code>
+   */
+  public boolean getOutputPartitionGraphs() {
+    return outputPartitionGraphs_;
+  }
+
+  public static final int DEBUG_OPTIONS_FIELD_NUMBER = 6;
+  private org.tensorflow.proto.framework.DebugOptions debugOptions_;
+  /**
+   * <pre>
+   * EXPERIMENTAL.  Options used to initialize DebuggerState, if enabled.
+   * </pre>
+   *
+   * <code>.tensorflow.DebugOptions debug_options = 6;</code>
+   */
+  public boolean hasDebugOptions() {
+    return debugOptions_ != null;
+  }
+  /**
+   * <pre>
+   * EXPERIMENTAL.  Options used to initialize DebuggerState, if enabled.
+   * </pre>
+   *
+   * <code>.tensorflow.DebugOptions debug_options = 6;</code>
+   */
+  public org.tensorflow.proto.framework.DebugOptions getDebugOptions() {
+    return debugOptions_ == null ? org.tensorflow.proto.framework.DebugOptions.getDefaultInstance() : debugOptions_;
+  }
+  /**
+   * <pre>
+   * EXPERIMENTAL.  Options used to initialize DebuggerState, if enabled.
+   * </pre>
+   *
+   * <code>.tensorflow.DebugOptions debug_options = 6;</code>
+   */
+  public org.tensorflow.proto.framework.DebugOptionsOrBuilder getDebugOptionsOrBuilder() {
+    return getDebugOptions();
+  }
+
+  public static final int REPORT_TENSOR_ALLOCATIONS_UPON_OOM_FIELD_NUMBER = 7;
+  private boolean reportTensorAllocationsUponOom_;
+  /**
+   * <pre>
+   * When enabled, causes tensor allocation information to be included in
+   * the error message when the Run() call fails because the allocator ran
+   * out of memory (OOM).
+   * Enabling this option can slow down the Run() call.
+   * </pre>
+   *
+   * <code>bool report_tensor_allocations_upon_oom = 7;</code>
+   */
+  public boolean getReportTensorAllocationsUponOom() {
+    return reportTensorAllocationsUponOom_;
+  }
+
+  public static final int EXPERIMENTAL_FIELD_NUMBER = 8;
+  private org.tensorflow.proto.framework.RunOptions.Experimental experimental_;
+  /**
+   * <code>.tensorflow.RunOptions.Experimental experimental = 8;</code>
+   */
+  public boolean hasExperimental() {
+    return experimental_ != null;
+  }
+  /**
+   * <code>.tensorflow.RunOptions.Experimental experimental = 8;</code>
+   */
+  public org.tensorflow.proto.framework.RunOptions.Experimental getExperimental() {
+    return experimental_ == null ? org.tensorflow.proto.framework.RunOptions.Experimental.getDefaultInstance() : experimental_;
+  }
+  /**
+   * <code>.tensorflow.RunOptions.Experimental experimental = 8;</code>
+   */
+  public org.tensorflow.proto.framework.RunOptions.ExperimentalOrBuilder getExperimentalOrBuilder() {
+    return getExperimental();
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    if (traceLevel_ != org.tensorflow.proto.framework.RunOptions.TraceLevel.NO_TRACE.getNumber()) {
+      output.writeEnum(1, traceLevel_);
+    }
+    if (timeoutInMs_ != 0L) {
+      output.writeInt64(2, timeoutInMs_);
+    }
+    if (interOpThreadPool_ != 0) {
+      output.writeInt32(3, interOpThreadPool_);
+    }
+    if (outputPartitionGraphs_ != false) {
+      output.writeBool(5, outputPartitionGraphs_);
+    }
+    if (debugOptions_ != null) {
+      output.writeMessage(6, getDebugOptions());
+    }
+    if (reportTensorAllocationsUponOom_ != false) {
+      output.writeBool(7, reportTensorAllocationsUponOom_);
+    }
+    if (experimental_ != null) {
+      output.writeMessage(8, getExperimental());
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (traceLevel_ != org.tensorflow.proto.framework.RunOptions.TraceLevel.NO_TRACE.getNumber()) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeEnumSize(1, traceLevel_);
+    }
+    if (timeoutInMs_ != 0L) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt64Size(2, timeoutInMs_);
+    }
+    if (interOpThreadPool_ != 0) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt32Size(3, interOpThreadPool_);
+    }
+    if (outputPartitionGraphs_ != false) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeBoolSize(5, outputPartitionGraphs_);
+    }
+    if (debugOptions_ != null) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(6, getDebugOptions());
+    }
+    if (reportTensorAllocationsUponOom_ != false) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeBoolSize(7, reportTensorAllocationsUponOom_);
+    }
+    if (experimental_ != null) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(8, getExperimental());
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.framework.RunOptions)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.framework.RunOptions other = (org.tensorflow.proto.framework.RunOptions) obj;
+
+    if (traceLevel_ != other.traceLevel_) return false;
+    if (getTimeoutInMs()
+        != other.getTimeoutInMs()) return false;
+    if (getInterOpThreadPool()
+        != other.getInterOpThreadPool()) return false;
+    if (getOutputPartitionGraphs()
+        != other.getOutputPartitionGraphs()) return false;
+    if (hasDebugOptions() != other.hasDebugOptions()) return false;
+    if (hasDebugOptions()) {
+      if (!getDebugOptions()
+          .equals(other.getDebugOptions())) return false;
+    }
+    if (getReportTensorAllocationsUponOom()
+        != other.getReportTensorAllocationsUponOom()) return false;
+    if (hasExperimental() != other.hasExperimental()) return false;
+    if (hasExperimental()) {
+      if (!getExperimental()
+          .equals(other.getExperimental())) return false;
+    }
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    hash = (37 * hash) + TRACE_LEVEL_FIELD_NUMBER;
+    hash = (53 * hash) + traceLevel_;
+    hash = (37 * hash) + TIMEOUT_IN_MS_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+        getTimeoutInMs());
+    hash = (37 * hash) + INTER_OP_THREAD_POOL_FIELD_NUMBER;
+    hash = (53 * hash) + getInterOpThreadPool();
+    hash = (37 * hash) + OUTPUT_PARTITION_GRAPHS_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashBoolean(
+        getOutputPartitionGraphs());
+    if (hasDebugOptions()) {
+      hash = (37 * hash) + DEBUG_OPTIONS_FIELD_NUMBER;
+      hash = (53 * hash) + getDebugOptions().hashCode();
+    }
+    hash = (37 * hash) + REPORT_TENSOR_ALLOCATIONS_UPON_OOM_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashBoolean(
+        getReportTensorAllocationsUponOom());
+    if (hasExperimental()) {
+      hash = (37 * hash) + EXPERIMENTAL_FIELD_NUMBER;
+      hash = (53 * hash) + getExperimental().hashCode();
+    }
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.framework.RunOptions parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.RunOptions parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.RunOptions parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.RunOptions parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.RunOptions parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.RunOptions parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.RunOptions parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.RunOptions parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.RunOptions parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.RunOptions parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.RunOptions parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.RunOptions parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.framework.RunOptions prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * <pre>
+   * Options for a single Run() call.
+   * </pre>
+   *
+   * Protobuf type {@code tensorflow.RunOptions}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.RunOptions)
+      org.tensorflow.proto.framework.RunOptionsOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_RunOptions_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_RunOptions_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.RunOptions.class, org.tensorflow.proto.framework.RunOptions.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.framework.RunOptions.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      traceLevel_ = 0;
+
+      timeoutInMs_ = 0L;
+
+      interOpThreadPool_ = 0;
+
+      outputPartitionGraphs_ = false;
+
+      if (debugOptionsBuilder_ == null) {
+        debugOptions_ = null;
+      } else {
+        debugOptions_ = null;
+        debugOptionsBuilder_ = null;
+      }
+      reportTensorAllocationsUponOom_ = false;
+
+      if (experimentalBuilder_ == null) {
+        experimental_ = null;
+      } else {
+        experimental_ = null;
+        experimentalBuilder_ = null;
+      }
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_RunOptions_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.RunOptions getDefaultInstanceForType() {
+      return org.tensorflow.proto.framework.RunOptions.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.RunOptions build() {
+      org.tensorflow.proto.framework.RunOptions result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.RunOptions buildPartial() {
+      org.tensorflow.proto.framework.RunOptions result = new org.tensorflow.proto.framework.RunOptions(this);
+      result.traceLevel_ = traceLevel_;
+      result.timeoutInMs_ = timeoutInMs_;
+      result.interOpThreadPool_ = interOpThreadPool_;
+      result.outputPartitionGraphs_ = outputPartitionGraphs_;
+      if (debugOptionsBuilder_ == null) {
+        result.debugOptions_ = debugOptions_;
+      } else {
+        result.debugOptions_ = debugOptionsBuilder_.build();
+      }
+      result.reportTensorAllocationsUponOom_ = reportTensorAllocationsUponOom_;
+      if (experimentalBuilder_ == null) {
+        result.experimental_ = experimental_;
+      } else {
+        result.experimental_ = experimentalBuilder_.build();
+      }
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.framework.RunOptions) {
+        return mergeFrom((org.tensorflow.proto.framework.RunOptions)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.framework.RunOptions other) {
+      if (other == org.tensorflow.proto.framework.RunOptions.getDefaultInstance()) return this;
+      if (other.traceLevel_ != 0) {
+        setTraceLevelValue(other.getTraceLevelValue());
+      }
+      if (other.getTimeoutInMs() != 0L) {
+        setTimeoutInMs(other.getTimeoutInMs());
+      }
+      if (other.getInterOpThreadPool() != 0) {
+        setInterOpThreadPool(other.getInterOpThreadPool());
+      }
+      if (other.getOutputPartitionGraphs() != false) {
+        setOutputPartitionGraphs(other.getOutputPartitionGraphs());
+      }
+      if (other.hasDebugOptions()) {
+        mergeDebugOptions(other.getDebugOptions());
+      }
+      if (other.getReportTensorAllocationsUponOom() != false) {
+        setReportTensorAllocationsUponOom(other.getReportTensorAllocationsUponOom());
+      }
+      if (other.hasExperimental()) {
+        mergeExperimental(other.getExperimental());
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.framework.RunOptions parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.framework.RunOptions) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+
+    private int traceLevel_ = 0;
+    /**
+     * <code>.tensorflow.RunOptions.TraceLevel trace_level = 1;</code>
+     */
+    public int getTraceLevelValue() {
+      return traceLevel_;
+    }
+    /**
+     * <code>.tensorflow.RunOptions.TraceLevel trace_level = 1;</code>
+     */
+    public Builder setTraceLevelValue(int value) {
+      traceLevel_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>.tensorflow.RunOptions.TraceLevel trace_level = 1;</code>
+     */
+    public org.tensorflow.proto.framework.RunOptions.TraceLevel getTraceLevel() {
+      @SuppressWarnings("deprecation")
+      org.tensorflow.proto.framework.RunOptions.TraceLevel result = org.tensorflow.proto.framework.RunOptions.TraceLevel.valueOf(traceLevel_);
+      return result == null ? org.tensorflow.proto.framework.RunOptions.TraceLevel.UNRECOGNIZED : result;
+    }
+    /**
+     * <code>.tensorflow.RunOptions.TraceLevel trace_level = 1;</code>
+     */
+    public Builder setTraceLevel(org.tensorflow.proto.framework.RunOptions.TraceLevel value) {
+      if (value == null) {
+        throw new NullPointerException();
+      }
+      
+      traceLevel_ = value.getNumber();
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>.tensorflow.RunOptions.TraceLevel trace_level = 1;</code>
+     */
+    public Builder clearTraceLevel() {
+      
+      traceLevel_ = 0;
+      onChanged();
+      return this;
+    }
+
+    private long timeoutInMs_ ;
+    /**
+     * <pre>
+     * Time to wait for operation to complete in milliseconds.
+     * </pre>
+     *
+     * <code>int64 timeout_in_ms = 2;</code>
+     */
+    public long getTimeoutInMs() {
+      return timeoutInMs_;
+    }
+    /**
+     * <pre>
+     * Time to wait for operation to complete in milliseconds.
+     * </pre>
+     *
+     * <code>int64 timeout_in_ms = 2;</code>
+     */
+    public Builder setTimeoutInMs(long value) {
+      
+      timeoutInMs_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Time to wait for operation to complete in milliseconds.
+     * </pre>
+     *
+     * <code>int64 timeout_in_ms = 2;</code>
+     */
+    public Builder clearTimeoutInMs() {
+      
+      timeoutInMs_ = 0L;
+      onChanged();
+      return this;
+    }
+
+    private int interOpThreadPool_ ;
+    /**
+     * <pre>
+     * The thread pool to use, if session_inter_op_thread_pool is configured.
+     * To use the caller thread set this to -1 - this uses the caller thread
+     * to execute Session::Run() and thus avoids a context switch. Using the
+     * caller thread to execute Session::Run() should be done ONLY for simple
+     * graphs, where the overhead of an additional context switch is
+     * comparable with the overhead of Session::Run().
+     * </pre>
+     *
+     * <code>int32 inter_op_thread_pool = 3;</code>
+     */
+    public int getInterOpThreadPool() {
+      return interOpThreadPool_;
+    }
+    /**
+     * <pre>
+     * The thread pool to use, if session_inter_op_thread_pool is configured.
+     * To use the caller thread set this to -1 - this uses the caller thread
+     * to execute Session::Run() and thus avoids a context switch. Using the
+     * caller thread to execute Session::Run() should be done ONLY for simple
+     * graphs, where the overhead of an additional context switch is
+     * comparable with the overhead of Session::Run().
+     * </pre>
+     *
+     * <code>int32 inter_op_thread_pool = 3;</code>
+     */
+    public Builder setInterOpThreadPool(int value) {
+      
+      interOpThreadPool_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * The thread pool to use, if session_inter_op_thread_pool is configured.
+     * To use the caller thread set this to -1 - this uses the caller thread
+     * to execute Session::Run() and thus avoids a context switch. Using the
+     * caller thread to execute Session::Run() should be done ONLY for simple
+     * graphs, where the overhead of an additional context switch is
+     * comparable with the overhead of Session::Run().
+     * </pre>
+     *
+     * <code>int32 inter_op_thread_pool = 3;</code>
+     */
+    public Builder clearInterOpThreadPool() {
+      
+      interOpThreadPool_ = 0;
+      onChanged();
+      return this;
+    }
+
+    private boolean outputPartitionGraphs_ ;
+    /**
+     * <pre>
+     * Whether the partition graph(s) executed by the executor(s) should be
+     * outputted via RunMetadata.
+     * </pre>
+     *
+     * <code>bool output_partition_graphs = 5;</code>
+     */
+    public boolean getOutputPartitionGraphs() {
+      return outputPartitionGraphs_;
+    }
+    /**
+     * <pre>
+     * Whether the partition graph(s) executed by the executor(s) should be
+     * outputted via RunMetadata.
+     * </pre>
+     *
+     * <code>bool output_partition_graphs = 5;</code>
+     */
+    public Builder setOutputPartitionGraphs(boolean value) {
+      
+      outputPartitionGraphs_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Whether the partition graph(s) executed by the executor(s) should be
+     * outputted via RunMetadata.
+     * </pre>
+     *
+     * <code>bool output_partition_graphs = 5;</code>
+     */
+    public Builder clearOutputPartitionGraphs() {
+      
+      outputPartitionGraphs_ = false;
+      onChanged();
+      return this;
+    }
+
+    private org.tensorflow.proto.framework.DebugOptions debugOptions_;
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.DebugOptions, org.tensorflow.proto.framework.DebugOptions.Builder, org.tensorflow.proto.framework.DebugOptionsOrBuilder> debugOptionsBuilder_;
+    /**
+     * <pre>
+     * EXPERIMENTAL.  Options used to initialize DebuggerState, if enabled.
+     * </pre>
+     *
+     * <code>.tensorflow.DebugOptions debug_options = 6;</code>
+     */
+    public boolean hasDebugOptions() {
+      return debugOptionsBuilder_ != null || debugOptions_ != null;
+    }
+    /**
+     * <pre>
+     * EXPERIMENTAL.  Options used to initialize DebuggerState, if enabled.
+     * </pre>
+     *
+     * <code>.tensorflow.DebugOptions debug_options = 6;</code>
+     */
+    public org.tensorflow.proto.framework.DebugOptions getDebugOptions() {
+      if (debugOptionsBuilder_ == null) {
+        return debugOptions_ == null ? org.tensorflow.proto.framework.DebugOptions.getDefaultInstance() : debugOptions_;
+      } else {
+        return debugOptionsBuilder_.getMessage();
+      }
+    }
+    /**
+     * <pre>
+     * EXPERIMENTAL.  Options used to initialize DebuggerState, if enabled.
+     * </pre>
+     *
+     * <code>.tensorflow.DebugOptions debug_options = 6;</code>
+     */
+    public Builder setDebugOptions(org.tensorflow.proto.framework.DebugOptions value) {
+      if (debugOptionsBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        debugOptions_ = value;
+        onChanged();
+      } else {
+        debugOptionsBuilder_.setMessage(value);
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * EXPERIMENTAL.  Options used to initialize DebuggerState, if enabled.
+     * </pre>
+     *
+     * <code>.tensorflow.DebugOptions debug_options = 6;</code>
+     */
+    public Builder setDebugOptions(
+        org.tensorflow.proto.framework.DebugOptions.Builder builderForValue) {
+      if (debugOptionsBuilder_ == null) {
+        debugOptions_ = builderForValue.build();
+        onChanged();
+      } else {
+        debugOptionsBuilder_.setMessage(builderForValue.build());
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * EXPERIMENTAL.  Options used to initialize DebuggerState, if enabled.
+     * </pre>
+     *
+     * <code>.tensorflow.DebugOptions debug_options = 6;</code>
+     */
+    public Builder mergeDebugOptions(org.tensorflow.proto.framework.DebugOptions value) {
+      if (debugOptionsBuilder_ == null) {
+        if (debugOptions_ != null) {
+          debugOptions_ =
+            org.tensorflow.proto.framework.DebugOptions.newBuilder(debugOptions_).mergeFrom(value).buildPartial();
+        } else {
+          debugOptions_ = value;
+        }
+        onChanged();
+      } else {
+        debugOptionsBuilder_.mergeFrom(value);
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * EXPERIMENTAL.  Options used to initialize DebuggerState, if enabled.
+     * </pre>
+     *
+     * <code>.tensorflow.DebugOptions debug_options = 6;</code>
+     */
+    public Builder clearDebugOptions() {
+      if (debugOptionsBuilder_ == null) {
+        debugOptions_ = null;
+        onChanged();
+      } else {
+        debugOptions_ = null;
+        debugOptionsBuilder_ = null;
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * EXPERIMENTAL.  Options used to initialize DebuggerState, if enabled.
+     * </pre>
+     *
+     * <code>.tensorflow.DebugOptions debug_options = 6;</code>
+     */
+    public org.tensorflow.proto.framework.DebugOptions.Builder getDebugOptionsBuilder() {
+      
+      onChanged();
+      return getDebugOptionsFieldBuilder().getBuilder();
+    }
+    /**
+     * <pre>
+     * EXPERIMENTAL.  Options used to initialize DebuggerState, if enabled.
+     * </pre>
+     *
+     * <code>.tensorflow.DebugOptions debug_options = 6;</code>
+     */
+    public org.tensorflow.proto.framework.DebugOptionsOrBuilder getDebugOptionsOrBuilder() {
+      if (debugOptionsBuilder_ != null) {
+        return debugOptionsBuilder_.getMessageOrBuilder();
+      } else {
+        return debugOptions_ == null ?
+            org.tensorflow.proto.framework.DebugOptions.getDefaultInstance() : debugOptions_;
+      }
+    }
+    /**
+     * <pre>
+     * EXPERIMENTAL.  Options used to initialize DebuggerState, if enabled.
+     * </pre>
+     *
+     * <code>.tensorflow.DebugOptions debug_options = 6;</code>
+     */
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.DebugOptions, org.tensorflow.proto.framework.DebugOptions.Builder, org.tensorflow.proto.framework.DebugOptionsOrBuilder> 
+        getDebugOptionsFieldBuilder() {
+      if (debugOptionsBuilder_ == null) {
+        debugOptionsBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+            org.tensorflow.proto.framework.DebugOptions, org.tensorflow.proto.framework.DebugOptions.Builder, org.tensorflow.proto.framework.DebugOptionsOrBuilder>(
+                getDebugOptions(),
+                getParentForChildren(),
+                isClean());
+        debugOptions_ = null;
+      }
+      return debugOptionsBuilder_;
+    }
+
+    private boolean reportTensorAllocationsUponOom_ ;
+    /**
+     * <pre>
+     * When enabled, causes tensor allocation information to be included in
+     * the error message when the Run() call fails because the allocator ran
+     * out of memory (OOM).
+     * Enabling this option can slow down the Run() call.
+     * </pre>
+     *
+     * <code>bool report_tensor_allocations_upon_oom = 7;</code>
+     */
+    public boolean getReportTensorAllocationsUponOom() {
+      return reportTensorAllocationsUponOom_;
+    }
+    /**
+     * <pre>
+     * When enabled, causes tensor allocation information to be included in
+     * the error message when the Run() call fails because the allocator ran
+     * out of memory (OOM).
+     * Enabling this option can slow down the Run() call.
+     * </pre>
+     *
+     * <code>bool report_tensor_allocations_upon_oom = 7;</code>
+     */
+    public Builder setReportTensorAllocationsUponOom(boolean value) {
+      
+      reportTensorAllocationsUponOom_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * When enabled, causes tensor allocation information to be included in
+     * the error message when the Run() call fails because the allocator ran
+     * out of memory (OOM).
+     * Enabling this option can slow down the Run() call.
+     * </pre>
+     *
+     * <code>bool report_tensor_allocations_upon_oom = 7;</code>
+     */
+    public Builder clearReportTensorAllocationsUponOom() {
+      
+      reportTensorAllocationsUponOom_ = false;
+      onChanged();
+      return this;
+    }
+
+    private org.tensorflow.proto.framework.RunOptions.Experimental experimental_;
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.RunOptions.Experimental, org.tensorflow.proto.framework.RunOptions.Experimental.Builder, org.tensorflow.proto.framework.RunOptions.ExperimentalOrBuilder> experimentalBuilder_;
+    /**
+     * <code>.tensorflow.RunOptions.Experimental experimental = 8;</code>
+     */
+    public boolean hasExperimental() {
+      return experimentalBuilder_ != null || experimental_ != null;
+    }
+    /**
+     * <code>.tensorflow.RunOptions.Experimental experimental = 8;</code>
+     */
+    public org.tensorflow.proto.framework.RunOptions.Experimental getExperimental() {
+      if (experimentalBuilder_ == null) {
+        return experimental_ == null ? org.tensorflow.proto.framework.RunOptions.Experimental.getDefaultInstance() : experimental_;
+      } else {
+        return experimentalBuilder_.getMessage();
+      }
+    }
+    /**
+     * <code>.tensorflow.RunOptions.Experimental experimental = 8;</code>
+     */
+    public Builder setExperimental(org.tensorflow.proto.framework.RunOptions.Experimental value) {
+      if (experimentalBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        experimental_ = value;
+        onChanged();
+      } else {
+        experimentalBuilder_.setMessage(value);
+      }
+
+      return this;
+    }
+    /**
+     * <code>.tensorflow.RunOptions.Experimental experimental = 8;</code>
+     */
+    public Builder setExperimental(
+        org.tensorflow.proto.framework.RunOptions.Experimental.Builder builderForValue) {
+      if (experimentalBuilder_ == null) {
+        experimental_ = builderForValue.build();
+        onChanged();
+      } else {
+        experimentalBuilder_.setMessage(builderForValue.build());
+      }
+
+      return this;
+    }
+    /**
+     * <code>.tensorflow.RunOptions.Experimental experimental = 8;</code>
+     */
+    public Builder mergeExperimental(org.tensorflow.proto.framework.RunOptions.Experimental value) {
+      if (experimentalBuilder_ == null) {
+        if (experimental_ != null) {
+          experimental_ =
+            org.tensorflow.proto.framework.RunOptions.Experimental.newBuilder(experimental_).mergeFrom(value).buildPartial();
+        } else {
+          experimental_ = value;
+        }
+        onChanged();
+      } else {
+        experimentalBuilder_.mergeFrom(value);
+      }
+
+      return this;
+    }
+    /**
+     * <code>.tensorflow.RunOptions.Experimental experimental = 8;</code>
+     */
+    public Builder clearExperimental() {
+      if (experimentalBuilder_ == null) {
+        experimental_ = null;
+        onChanged();
+      } else {
+        experimental_ = null;
+        experimentalBuilder_ = null;
+      }
+
+      return this;
+    }
+    /**
+     * <code>.tensorflow.RunOptions.Experimental experimental = 8;</code>
+     */
+    public org.tensorflow.proto.framework.RunOptions.Experimental.Builder getExperimentalBuilder() {
+      
+      onChanged();
+      return getExperimentalFieldBuilder().getBuilder();
+    }
+    /**
+     * <code>.tensorflow.RunOptions.Experimental experimental = 8;</code>
+     */
+    public org.tensorflow.proto.framework.RunOptions.ExperimentalOrBuilder getExperimentalOrBuilder() {
+      if (experimentalBuilder_ != null) {
+        return experimentalBuilder_.getMessageOrBuilder();
+      } else {
+        return experimental_ == null ?
+            org.tensorflow.proto.framework.RunOptions.Experimental.getDefaultInstance() : experimental_;
+      }
+    }
+    /**
+     * <code>.tensorflow.RunOptions.Experimental experimental = 8;</code>
+     */
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.RunOptions.Experimental, org.tensorflow.proto.framework.RunOptions.Experimental.Builder, org.tensorflow.proto.framework.RunOptions.ExperimentalOrBuilder> 
+        getExperimentalFieldBuilder() {
+      if (experimentalBuilder_ == null) {
+        experimentalBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+            org.tensorflow.proto.framework.RunOptions.Experimental, org.tensorflow.proto.framework.RunOptions.Experimental.Builder, org.tensorflow.proto.framework.RunOptions.ExperimentalOrBuilder>(
+                getExperimental(),
+                getParentForChildren(),
+                isClean());
+        experimental_ = null;
+      }
+      return experimentalBuilder_;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.RunOptions)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.RunOptions)
+  private static final org.tensorflow.proto.framework.RunOptions DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.RunOptions();
+  }
+
+  public static org.tensorflow.proto.framework.RunOptions getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<RunOptions>
+      PARSER = new com.google.protobuf.AbstractParser<RunOptions>() {
+    @java.lang.Override
+    public RunOptions parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new RunOptions(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<RunOptions> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<RunOptions> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.framework.RunOptions getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/RunOptionsOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/RunOptionsOrBuilder.java
new file mode 100644
index 00000000000..c3f4a9d7205
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/RunOptionsOrBuilder.java
@@ -0,0 +1,101 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/config.proto
+
+package org.tensorflow.proto.framework;
+
+public interface RunOptionsOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.RunOptions)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <code>.tensorflow.RunOptions.TraceLevel trace_level = 1;</code>
+   */
+  int getTraceLevelValue();
+  /**
+   * <code>.tensorflow.RunOptions.TraceLevel trace_level = 1;</code>
+   */
+  org.tensorflow.proto.framework.RunOptions.TraceLevel getTraceLevel();
+
+  /**
+   * <pre>
+   * Time to wait for operation to complete in milliseconds.
+   * </pre>
+   *
+   * <code>int64 timeout_in_ms = 2;</code>
+   */
+  long getTimeoutInMs();
+
+  /**
+   * <pre>
+   * The thread pool to use, if session_inter_op_thread_pool is configured.
+   * To use the caller thread set this to -1 - this uses the caller thread
+   * to execute Session::Run() and thus avoids a context switch. Using the
+   * caller thread to execute Session::Run() should be done ONLY for simple
+   * graphs, where the overhead of an additional context switch is
+   * comparable with the overhead of Session::Run().
+   * </pre>
+   *
+   * <code>int32 inter_op_thread_pool = 3;</code>
+   */
+  int getInterOpThreadPool();
+
+  /**
+   * <pre>
+   * Whether the partition graph(s) executed by the executor(s) should be
+   * outputted via RunMetadata.
+   * </pre>
+   *
+   * <code>bool output_partition_graphs = 5;</code>
+   */
+  boolean getOutputPartitionGraphs();
+
+  /**
+   * <pre>
+   * EXPERIMENTAL.  Options used to initialize DebuggerState, if enabled.
+   * </pre>
+   *
+   * <code>.tensorflow.DebugOptions debug_options = 6;</code>
+   */
+  boolean hasDebugOptions();
+  /**
+   * <pre>
+   * EXPERIMENTAL.  Options used to initialize DebuggerState, if enabled.
+   * </pre>
+   *
+   * <code>.tensorflow.DebugOptions debug_options = 6;</code>
+   */
+  org.tensorflow.proto.framework.DebugOptions getDebugOptions();
+  /**
+   * <pre>
+   * EXPERIMENTAL.  Options used to initialize DebuggerState, if enabled.
+   * </pre>
+   *
+   * <code>.tensorflow.DebugOptions debug_options = 6;</code>
+   */
+  org.tensorflow.proto.framework.DebugOptionsOrBuilder getDebugOptionsOrBuilder();
+
+  /**
+   * <pre>
+   * When enabled, causes tensor allocation information to be included in
+   * the error message when the Run() call fails because the allocator ran
+   * out of memory (OOM).
+   * Enabling this option can slow down the Run() call.
+   * </pre>
+   *
+   * <code>bool report_tensor_allocations_upon_oom = 7;</code>
+   */
+  boolean getReportTensorAllocationsUponOom();
+
+  /**
+   * <code>.tensorflow.RunOptions.Experimental experimental = 8;</code>
+   */
+  boolean hasExperimental();
+  /**
+   * <code>.tensorflow.RunOptions.Experimental experimental = 8;</code>
+   */
+  org.tensorflow.proto.framework.RunOptions.Experimental getExperimental();
+  /**
+   * <code>.tensorflow.RunOptions.Experimental experimental = 8;</code>
+   */
+  org.tensorflow.proto.framework.RunOptions.ExperimentalOrBuilder getExperimentalOrBuilder();
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SaveSliceInfoDef.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SaveSliceInfoDef.java
new file mode 100644
index 00000000000..3cf8b1b4b12
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SaveSliceInfoDef.java
@@ -0,0 +1,1175 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/framework/variable.proto
+
+package org.tensorflow.proto.framework;
+
+/**
+ * Protobuf type {@code tensorflow.SaveSliceInfoDef}
+ */
+public  final class SaveSliceInfoDef extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.SaveSliceInfoDef)
+    SaveSliceInfoDefOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use SaveSliceInfoDef.newBuilder() to construct.
+  private SaveSliceInfoDef(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private SaveSliceInfoDef() {
+    fullName_ = "";
+    fullShape_ = emptyLongList();
+    varOffset_ = emptyLongList();
+    varShape_ = emptyLongList();
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new SaveSliceInfoDef();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private SaveSliceInfoDef(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    int mutable_bitField0_ = 0;
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 10: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            fullName_ = s;
+            break;
+          }
+          case 16: {
+            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
+              fullShape_ = newLongList();
+              mutable_bitField0_ |= 0x00000001;
+            }
+            fullShape_.addLong(input.readInt64());
+            break;
+          }
+          case 18: {
+            int length = input.readRawVarint32();
+            int limit = input.pushLimit(length);
+            if (!((mutable_bitField0_ & 0x00000001) != 0) && input.getBytesUntilLimit() > 0) {
+              fullShape_ = newLongList();
+              mutable_bitField0_ |= 0x00000001;
+            }
+            while (input.getBytesUntilLimit() > 0) {
+              fullShape_.addLong(input.readInt64());
+            }
+            input.popLimit(limit);
+            break;
+          }
+          case 24: {
+            if (!((mutable_bitField0_ & 0x00000002) != 0)) {
+              varOffset_ = newLongList();
+              mutable_bitField0_ |= 0x00000002;
+            }
+            varOffset_.addLong(input.readInt64());
+            break;
+          }
+          case 26: {
+            int length = input.readRawVarint32();
+            int limit = input.pushLimit(length);
+            if (!((mutable_bitField0_ & 0x00000002) != 0) && input.getBytesUntilLimit() > 0) {
+              varOffset_ = newLongList();
+              mutable_bitField0_ |= 0x00000002;
+            }
+            while (input.getBytesUntilLimit() > 0) {
+              varOffset_.addLong(input.readInt64());
+            }
+            input.popLimit(limit);
+            break;
+          }
+          case 32: {
+            if (!((mutable_bitField0_ & 0x00000004) != 0)) {
+              varShape_ = newLongList();
+              mutable_bitField0_ |= 0x00000004;
+            }
+            varShape_.addLong(input.readInt64());
+            break;
+          }
+          case 34: {
+            int length = input.readRawVarint32();
+            int limit = input.pushLimit(length);
+            if (!((mutable_bitField0_ & 0x00000004) != 0) && input.getBytesUntilLimit() > 0) {
+              varShape_ = newLongList();
+              mutable_bitField0_ |= 0x00000004;
+            }
+            while (input.getBytesUntilLimit() > 0) {
+              varShape_.addLong(input.readInt64());
+            }
+            input.popLimit(limit);
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      if (((mutable_bitField0_ & 0x00000001) != 0)) {
+        fullShape_.makeImmutable(); // C
+      }
+      if (((mutable_bitField0_ & 0x00000002) != 0)) {
+        varOffset_.makeImmutable(); // C
+      }
+      if (((mutable_bitField0_ & 0x00000004) != 0)) {
+        varShape_.makeImmutable(); // C
+      }
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.framework.VariableProtos.internal_static_tensorflow_SaveSliceInfoDef_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.framework.VariableProtos.internal_static_tensorflow_SaveSliceInfoDef_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.framework.SaveSliceInfoDef.class, org.tensorflow.proto.framework.SaveSliceInfoDef.Builder.class);
+  }
+
+  public static final int FULL_NAME_FIELD_NUMBER = 1;
+  private volatile java.lang.Object fullName_;
+  /**
+   * <pre>
+   * Name of the full variable of which this is a slice.
+   * </pre>
+   *
+   * <code>string full_name = 1;</code>
+   */
+  public java.lang.String getFullName() {
+    java.lang.Object ref = fullName_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      fullName_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * Name of the full variable of which this is a slice.
+   * </pre>
+   *
+   * <code>string full_name = 1;</code>
+   */
+  public com.google.protobuf.ByteString
+      getFullNameBytes() {
+    java.lang.Object ref = fullName_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      fullName_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int FULL_SHAPE_FIELD_NUMBER = 2;
+  private com.google.protobuf.Internal.LongList fullShape_;
+  /**
+   * <pre>
+   * Shape of the full variable.
+   * </pre>
+   *
+   * <code>repeated int64 full_shape = 2;</code>
+   */
+  public java.util.List<java.lang.Long>
+      getFullShapeList() {
+    return fullShape_;
+  }
+  /**
+   * <pre>
+   * Shape of the full variable.
+   * </pre>
+   *
+   * <code>repeated int64 full_shape = 2;</code>
+   */
+  public int getFullShapeCount() {
+    return fullShape_.size();
+  }
+  /**
+   * <pre>
+   * Shape of the full variable.
+   * </pre>
+   *
+   * <code>repeated int64 full_shape = 2;</code>
+   */
+  public long getFullShape(int index) {
+    return fullShape_.getLong(index);
+  }
+  private int fullShapeMemoizedSerializedSize = -1;
+
+  public static final int VAR_OFFSET_FIELD_NUMBER = 3;
+  private com.google.protobuf.Internal.LongList varOffset_;
+  /**
+   * <pre>
+   * Offset of this variable into the full variable.
+   * </pre>
+   *
+   * <code>repeated int64 var_offset = 3;</code>
+   */
+  public java.util.List<java.lang.Long>
+      getVarOffsetList() {
+    return varOffset_;
+  }
+  /**
+   * <pre>
+   * Offset of this variable into the full variable.
+   * </pre>
+   *
+   * <code>repeated int64 var_offset = 3;</code>
+   */
+  public int getVarOffsetCount() {
+    return varOffset_.size();
+  }
+  /**
+   * <pre>
+   * Offset of this variable into the full variable.
+   * </pre>
+   *
+   * <code>repeated int64 var_offset = 3;</code>
+   */
+  public long getVarOffset(int index) {
+    return varOffset_.getLong(index);
+  }
+  private int varOffsetMemoizedSerializedSize = -1;
+
+  public static final int VAR_SHAPE_FIELD_NUMBER = 4;
+  private com.google.protobuf.Internal.LongList varShape_;
+  /**
+   * <pre>
+   * Shape of this variable.
+   * </pre>
+   *
+   * <code>repeated int64 var_shape = 4;</code>
+   */
+  public java.util.List<java.lang.Long>
+      getVarShapeList() {
+    return varShape_;
+  }
+  /**
+   * <pre>
+   * Shape of this variable.
+   * </pre>
+   *
+   * <code>repeated int64 var_shape = 4;</code>
+   */
+  public int getVarShapeCount() {
+    return varShape_.size();
+  }
+  /**
+   * <pre>
+   * Shape of this variable.
+   * </pre>
+   *
+   * <code>repeated int64 var_shape = 4;</code>
+   */
+  public long getVarShape(int index) {
+    return varShape_.getLong(index);
+  }
+  private int varShapeMemoizedSerializedSize = -1;
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    getSerializedSize();
+    if (!getFullNameBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 1, fullName_);
+    }
+    if (getFullShapeList().size() > 0) {
+      output.writeUInt32NoTag(18);
+      output.writeUInt32NoTag(fullShapeMemoizedSerializedSize);
+    }
+    for (int i = 0; i < fullShape_.size(); i++) {
+      output.writeInt64NoTag(fullShape_.getLong(i));
+    }
+    if (getVarOffsetList().size() > 0) {
+      output.writeUInt32NoTag(26);
+      output.writeUInt32NoTag(varOffsetMemoizedSerializedSize);
+    }
+    for (int i = 0; i < varOffset_.size(); i++) {
+      output.writeInt64NoTag(varOffset_.getLong(i));
+    }
+    if (getVarShapeList().size() > 0) {
+      output.writeUInt32NoTag(34);
+      output.writeUInt32NoTag(varShapeMemoizedSerializedSize);
+    }
+    for (int i = 0; i < varShape_.size(); i++) {
+      output.writeInt64NoTag(varShape_.getLong(i));
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (!getFullNameBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, fullName_);
+    }
+    {
+      int dataSize = 0;
+      for (int i = 0; i < fullShape_.size(); i++) {
+        dataSize += com.google.protobuf.CodedOutputStream
+          .computeInt64SizeNoTag(fullShape_.getLong(i));
+      }
+      size += dataSize;
+      if (!getFullShapeList().isEmpty()) {
+        size += 1;
+        size += com.google.protobuf.CodedOutputStream
+            .computeInt32SizeNoTag(dataSize);
+      }
+      fullShapeMemoizedSerializedSize = dataSize;
+    }
+    {
+      int dataSize = 0;
+      for (int i = 0; i < varOffset_.size(); i++) {
+        dataSize += com.google.protobuf.CodedOutputStream
+          .computeInt64SizeNoTag(varOffset_.getLong(i));
+      }
+      size += dataSize;
+      if (!getVarOffsetList().isEmpty()) {
+        size += 1;
+        size += com.google.protobuf.CodedOutputStream
+            .computeInt32SizeNoTag(dataSize);
+      }
+      varOffsetMemoizedSerializedSize = dataSize;
+    }
+    {
+      int dataSize = 0;
+      for (int i = 0; i < varShape_.size(); i++) {
+        dataSize += com.google.protobuf.CodedOutputStream
+          .computeInt64SizeNoTag(varShape_.getLong(i));
+      }
+      size += dataSize;
+      if (!getVarShapeList().isEmpty()) {
+        size += 1;
+        size += com.google.protobuf.CodedOutputStream
+            .computeInt32SizeNoTag(dataSize);
+      }
+      varShapeMemoizedSerializedSize = dataSize;
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.framework.SaveSliceInfoDef)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.framework.SaveSliceInfoDef other = (org.tensorflow.proto.framework.SaveSliceInfoDef) obj;
+
+    if (!getFullName()
+        .equals(other.getFullName())) return false;
+    if (!getFullShapeList()
+        .equals(other.getFullShapeList())) return false;
+    if (!getVarOffsetList()
+        .equals(other.getVarOffsetList())) return false;
+    if (!getVarShapeList()
+        .equals(other.getVarShapeList())) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    hash = (37 * hash) + FULL_NAME_FIELD_NUMBER;
+    hash = (53 * hash) + getFullName().hashCode();
+    if (getFullShapeCount() > 0) {
+      hash = (37 * hash) + FULL_SHAPE_FIELD_NUMBER;
+      hash = (53 * hash) + getFullShapeList().hashCode();
+    }
+    if (getVarOffsetCount() > 0) {
+      hash = (37 * hash) + VAR_OFFSET_FIELD_NUMBER;
+      hash = (53 * hash) + getVarOffsetList().hashCode();
+    }
+    if (getVarShapeCount() > 0) {
+      hash = (37 * hash) + VAR_SHAPE_FIELD_NUMBER;
+      hash = (53 * hash) + getVarShapeList().hashCode();
+    }
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.framework.SaveSliceInfoDef parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.SaveSliceInfoDef parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.SaveSliceInfoDef parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.SaveSliceInfoDef parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.SaveSliceInfoDef parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.SaveSliceInfoDef parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.SaveSliceInfoDef parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.SaveSliceInfoDef parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.SaveSliceInfoDef parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.SaveSliceInfoDef parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.SaveSliceInfoDef parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.SaveSliceInfoDef parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.framework.SaveSliceInfoDef prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * Protobuf type {@code tensorflow.SaveSliceInfoDef}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.SaveSliceInfoDef)
+      org.tensorflow.proto.framework.SaveSliceInfoDefOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.VariableProtos.internal_static_tensorflow_SaveSliceInfoDef_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.VariableProtos.internal_static_tensorflow_SaveSliceInfoDef_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.SaveSliceInfoDef.class, org.tensorflow.proto.framework.SaveSliceInfoDef.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.framework.SaveSliceInfoDef.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      fullName_ = "";
+
+      fullShape_ = emptyLongList();
+      bitField0_ = (bitField0_ & ~0x00000001);
+      varOffset_ = emptyLongList();
+      bitField0_ = (bitField0_ & ~0x00000002);
+      varShape_ = emptyLongList();
+      bitField0_ = (bitField0_ & ~0x00000004);
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.framework.VariableProtos.internal_static_tensorflow_SaveSliceInfoDef_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.SaveSliceInfoDef getDefaultInstanceForType() {
+      return org.tensorflow.proto.framework.SaveSliceInfoDef.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.SaveSliceInfoDef build() {
+      org.tensorflow.proto.framework.SaveSliceInfoDef result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.SaveSliceInfoDef buildPartial() {
+      org.tensorflow.proto.framework.SaveSliceInfoDef result = new org.tensorflow.proto.framework.SaveSliceInfoDef(this);
+      int from_bitField0_ = bitField0_;
+      result.fullName_ = fullName_;
+      if (((bitField0_ & 0x00000001) != 0)) {
+        fullShape_.makeImmutable();
+        bitField0_ = (bitField0_ & ~0x00000001);
+      }
+      result.fullShape_ = fullShape_;
+      if (((bitField0_ & 0x00000002) != 0)) {
+        varOffset_.makeImmutable();
+        bitField0_ = (bitField0_ & ~0x00000002);
+      }
+      result.varOffset_ = varOffset_;
+      if (((bitField0_ & 0x00000004) != 0)) {
+        varShape_.makeImmutable();
+        bitField0_ = (bitField0_ & ~0x00000004);
+      }
+      result.varShape_ = varShape_;
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.framework.SaveSliceInfoDef) {
+        return mergeFrom((org.tensorflow.proto.framework.SaveSliceInfoDef)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.framework.SaveSliceInfoDef other) {
+      if (other == org.tensorflow.proto.framework.SaveSliceInfoDef.getDefaultInstance()) return this;
+      if (!other.getFullName().isEmpty()) {
+        fullName_ = other.fullName_;
+        onChanged();
+      }
+      if (!other.fullShape_.isEmpty()) {
+        if (fullShape_.isEmpty()) {
+          fullShape_ = other.fullShape_;
+          bitField0_ = (bitField0_ & ~0x00000001);
+        } else {
+          ensureFullShapeIsMutable();
+          fullShape_.addAll(other.fullShape_);
+        }
+        onChanged();
+      }
+      if (!other.varOffset_.isEmpty()) {
+        if (varOffset_.isEmpty()) {
+          varOffset_ = other.varOffset_;
+          bitField0_ = (bitField0_ & ~0x00000002);
+        } else {
+          ensureVarOffsetIsMutable();
+          varOffset_.addAll(other.varOffset_);
+        }
+        onChanged();
+      }
+      if (!other.varShape_.isEmpty()) {
+        if (varShape_.isEmpty()) {
+          varShape_ = other.varShape_;
+          bitField0_ = (bitField0_ & ~0x00000004);
+        } else {
+          ensureVarShapeIsMutable();
+          varShape_.addAll(other.varShape_);
+        }
+        onChanged();
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.framework.SaveSliceInfoDef parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.framework.SaveSliceInfoDef) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+    private int bitField0_;
+
+    private java.lang.Object fullName_ = "";
+    /**
+     * <pre>
+     * Name of the full variable of which this is a slice.
+     * </pre>
+     *
+     * <code>string full_name = 1;</code>
+     */
+    public java.lang.String getFullName() {
+      java.lang.Object ref = fullName_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        fullName_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Name of the full variable of which this is a slice.
+     * </pre>
+     *
+     * <code>string full_name = 1;</code>
+     */
+    public com.google.protobuf.ByteString
+        getFullNameBytes() {
+      java.lang.Object ref = fullName_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        fullName_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Name of the full variable of which this is a slice.
+     * </pre>
+     *
+     * <code>string full_name = 1;</code>
+     */
+    public Builder setFullName(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      fullName_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Name of the full variable of which this is a slice.
+     * </pre>
+     *
+     * <code>string full_name = 1;</code>
+     */
+    public Builder clearFullName() {
+      
+      fullName_ = getDefaultInstance().getFullName();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Name of the full variable of which this is a slice.
+     * </pre>
+     *
+     * <code>string full_name = 1;</code>
+     */
+    public Builder setFullNameBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      fullName_ = value;
+      onChanged();
+      return this;
+    }
+
+    private com.google.protobuf.Internal.LongList fullShape_ = emptyLongList();
+    private void ensureFullShapeIsMutable() {
+      if (!((bitField0_ & 0x00000001) != 0)) {
+        fullShape_ = mutableCopy(fullShape_);
+        bitField0_ |= 0x00000001;
+       }
+    }
+    /**
+     * <pre>
+     * Shape of the full variable.
+     * </pre>
+     *
+     * <code>repeated int64 full_shape = 2;</code>
+     */
+    public java.util.List<java.lang.Long>
+        getFullShapeList() {
+      return ((bitField0_ & 0x00000001) != 0) ?
+               java.util.Collections.unmodifiableList(fullShape_) : fullShape_;
+    }
+    /**
+     * <pre>
+     * Shape of the full variable.
+     * </pre>
+     *
+     * <code>repeated int64 full_shape = 2;</code>
+     */
+    public int getFullShapeCount() {
+      return fullShape_.size();
+    }
+    /**
+     * <pre>
+     * Shape of the full variable.
+     * </pre>
+     *
+     * <code>repeated int64 full_shape = 2;</code>
+     */
+    public long getFullShape(int index) {
+      return fullShape_.getLong(index);
+    }
+    /**
+     * <pre>
+     * Shape of the full variable.
+     * </pre>
+     *
+     * <code>repeated int64 full_shape = 2;</code>
+     */
+    public Builder setFullShape(
+        int index, long value) {
+      ensureFullShapeIsMutable();
+      fullShape_.setLong(index, value);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Shape of the full variable.
+     * </pre>
+     *
+     * <code>repeated int64 full_shape = 2;</code>
+     */
+    public Builder addFullShape(long value) {
+      ensureFullShapeIsMutable();
+      fullShape_.addLong(value);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Shape of the full variable.
+     * </pre>
+     *
+     * <code>repeated int64 full_shape = 2;</code>
+     */
+    public Builder addAllFullShape(
+        java.lang.Iterable<? extends java.lang.Long> values) {
+      ensureFullShapeIsMutable();
+      com.google.protobuf.AbstractMessageLite.Builder.addAll(
+          values, fullShape_);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Shape of the full variable.
+     * </pre>
+     *
+     * <code>repeated int64 full_shape = 2;</code>
+     */
+    public Builder clearFullShape() {
+      fullShape_ = emptyLongList();
+      bitField0_ = (bitField0_ & ~0x00000001);
+      onChanged();
+      return this;
+    }
+
+    private com.google.protobuf.Internal.LongList varOffset_ = emptyLongList();
+    private void ensureVarOffsetIsMutable() {
+      if (!((bitField0_ & 0x00000002) != 0)) {
+        varOffset_ = mutableCopy(varOffset_);
+        bitField0_ |= 0x00000002;
+       }
+    }
+    /**
+     * <pre>
+     * Offset of this variable into the full variable.
+     * </pre>
+     *
+     * <code>repeated int64 var_offset = 3;</code>
+     */
+    public java.util.List<java.lang.Long>
+        getVarOffsetList() {
+      return ((bitField0_ & 0x00000002) != 0) ?
+               java.util.Collections.unmodifiableList(varOffset_) : varOffset_;
+    }
+    /**
+     * <pre>
+     * Offset of this variable into the full variable.
+     * </pre>
+     *
+     * <code>repeated int64 var_offset = 3;</code>
+     */
+    public int getVarOffsetCount() {
+      return varOffset_.size();
+    }
+    /**
+     * <pre>
+     * Offset of this variable into the full variable.
+     * </pre>
+     *
+     * <code>repeated int64 var_offset = 3;</code>
+     */
+    public long getVarOffset(int index) {
+      return varOffset_.getLong(index);
+    }
+    /**
+     * <pre>
+     * Offset of this variable into the full variable.
+     * </pre>
+     *
+     * <code>repeated int64 var_offset = 3;</code>
+     */
+    public Builder setVarOffset(
+        int index, long value) {
+      ensureVarOffsetIsMutable();
+      varOffset_.setLong(index, value);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Offset of this variable into the full variable.
+     * </pre>
+     *
+     * <code>repeated int64 var_offset = 3;</code>
+     */
+    public Builder addVarOffset(long value) {
+      ensureVarOffsetIsMutable();
+      varOffset_.addLong(value);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Offset of this variable into the full variable.
+     * </pre>
+     *
+     * <code>repeated int64 var_offset = 3;</code>
+     */
+    public Builder addAllVarOffset(
+        java.lang.Iterable<? extends java.lang.Long> values) {
+      ensureVarOffsetIsMutable();
+      com.google.protobuf.AbstractMessageLite.Builder.addAll(
+          values, varOffset_);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Offset of this variable into the full variable.
+     * </pre>
+     *
+     * <code>repeated int64 var_offset = 3;</code>
+     */
+    public Builder clearVarOffset() {
+      varOffset_ = emptyLongList();
+      bitField0_ = (bitField0_ & ~0x00000002);
+      onChanged();
+      return this;
+    }
+
+    private com.google.protobuf.Internal.LongList varShape_ = emptyLongList();
+    private void ensureVarShapeIsMutable() {
+      if (!((bitField0_ & 0x00000004) != 0)) {
+        varShape_ = mutableCopy(varShape_);
+        bitField0_ |= 0x00000004;
+       }
+    }
+    /**
+     * <pre>
+     * Shape of this variable.
+     * </pre>
+     *
+     * <code>repeated int64 var_shape = 4;</code>
+     */
+    public java.util.List<java.lang.Long>
+        getVarShapeList() {
+      return ((bitField0_ & 0x00000004) != 0) ?
+               java.util.Collections.unmodifiableList(varShape_) : varShape_;
+    }
+    /**
+     * <pre>
+     * Shape of this variable.
+     * </pre>
+     *
+     * <code>repeated int64 var_shape = 4;</code>
+     */
+    public int getVarShapeCount() {
+      return varShape_.size();
+    }
+    /**
+     * <pre>
+     * Shape of this variable.
+     * </pre>
+     *
+     * <code>repeated int64 var_shape = 4;</code>
+     */
+    public long getVarShape(int index) {
+      return varShape_.getLong(index);
+    }
+    /**
+     * <pre>
+     * Shape of this variable.
+     * </pre>
+     *
+     * <code>repeated int64 var_shape = 4;</code>
+     */
+    public Builder setVarShape(
+        int index, long value) {
+      ensureVarShapeIsMutable();
+      varShape_.setLong(index, value);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Shape of this variable.
+     * </pre>
+     *
+     * <code>repeated int64 var_shape = 4;</code>
+     */
+    public Builder addVarShape(long value) {
+      ensureVarShapeIsMutable();
+      varShape_.addLong(value);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Shape of this variable.
+     * </pre>
+     *
+     * <code>repeated int64 var_shape = 4;</code>
+     */
+    public Builder addAllVarShape(
+        java.lang.Iterable<? extends java.lang.Long> values) {
+      ensureVarShapeIsMutable();
+      com.google.protobuf.AbstractMessageLite.Builder.addAll(
+          values, varShape_);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Shape of this variable.
+     * </pre>
+     *
+     * <code>repeated int64 var_shape = 4;</code>
+     */
+    public Builder clearVarShape() {
+      varShape_ = emptyLongList();
+      bitField0_ = (bitField0_ & ~0x00000004);
+      onChanged();
+      return this;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.SaveSliceInfoDef)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.SaveSliceInfoDef)
+  private static final org.tensorflow.proto.framework.SaveSliceInfoDef DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.SaveSliceInfoDef();
+  }
+
+  public static org.tensorflow.proto.framework.SaveSliceInfoDef getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<SaveSliceInfoDef>
+      PARSER = new com.google.protobuf.AbstractParser<SaveSliceInfoDef>() {
+    @java.lang.Override
+    public SaveSliceInfoDef parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new SaveSliceInfoDef(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<SaveSliceInfoDef> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<SaveSliceInfoDef> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.framework.SaveSliceInfoDef getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SaveSliceInfoDefOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SaveSliceInfoDefOrBuilder.java
new file mode 100644
index 00000000000..c06b53fbb61
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SaveSliceInfoDefOrBuilder.java
@@ -0,0 +1,102 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/framework/variable.proto
+
+package org.tensorflow.proto.framework;
+
+public interface SaveSliceInfoDefOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.SaveSliceInfoDef)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <pre>
+   * Name of the full variable of which this is a slice.
+   * </pre>
+   *
+   * <code>string full_name = 1;</code>
+   */
+  java.lang.String getFullName();
+  /**
+   * <pre>
+   * Name of the full variable of which this is a slice.
+   * </pre>
+   *
+   * <code>string full_name = 1;</code>
+   */
+  com.google.protobuf.ByteString
+      getFullNameBytes();
+
+  /**
+   * <pre>
+   * Shape of the full variable.
+   * </pre>
+   *
+   * <code>repeated int64 full_shape = 2;</code>
+   */
+  java.util.List<java.lang.Long> getFullShapeList();
+  /**
+   * <pre>
+   * Shape of the full variable.
+   * </pre>
+   *
+   * <code>repeated int64 full_shape = 2;</code>
+   */
+  int getFullShapeCount();
+  /**
+   * <pre>
+   * Shape of the full variable.
+   * </pre>
+   *
+   * <code>repeated int64 full_shape = 2;</code>
+   */
+  long getFullShape(int index);
+
+  /**
+   * <pre>
+   * Offset of this variable into the full variable.
+   * </pre>
+   *
+   * <code>repeated int64 var_offset = 3;</code>
+   */
+  java.util.List<java.lang.Long> getVarOffsetList();
+  /**
+   * <pre>
+   * Offset of this variable into the full variable.
+   * </pre>
+   *
+   * <code>repeated int64 var_offset = 3;</code>
+   */
+  int getVarOffsetCount();
+  /**
+   * <pre>
+   * Offset of this variable into the full variable.
+   * </pre>
+   *
+   * <code>repeated int64 var_offset = 3;</code>
+   */
+  long getVarOffset(int index);
+
+  /**
+   * <pre>
+   * Shape of this variable.
+   * </pre>
+   *
+   * <code>repeated int64 var_shape = 4;</code>
+   */
+  java.util.List<java.lang.Long> getVarShapeList();
+  /**
+   * <pre>
+   * Shape of this variable.
+   * </pre>
+   *
+   * <code>repeated int64 var_shape = 4;</code>
+   */
+  int getVarShapeCount();
+  /**
+   * <pre>
+   * Shape of this variable.
+   * </pre>
+   *
+   * <code>repeated int64 var_shape = 4;</code>
+   */
+  long getVarShape(int index);
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SaveableObject.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SaveableObject.java
new file mode 100644
index 00000000000..c2b8ae241d9
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SaveableObject.java
@@ -0,0 +1,549 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/saved_object_graph.proto
+
+package org.tensorflow.proto.framework;
+
+/**
+ * Protobuf type {@code tensorflow.SaveableObject}
+ */
+public  final class SaveableObject extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.SaveableObject)
+    SaveableObjectOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use SaveableObject.newBuilder() to construct.
+  private SaveableObject(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private SaveableObject() {
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new SaveableObject();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private SaveableObject(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 16: {
+
+            saveFunction_ = input.readInt32();
+            break;
+          }
+          case 24: {
+
+            restoreFunction_ = input.readInt32();
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.framework.SavedObjectGraphProtos.internal_static_tensorflow_SaveableObject_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.framework.SavedObjectGraphProtos.internal_static_tensorflow_SaveableObject_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.framework.SaveableObject.class, org.tensorflow.proto.framework.SaveableObject.Builder.class);
+  }
+
+  public static final int SAVE_FUNCTION_FIELD_NUMBER = 2;
+  private int saveFunction_;
+  /**
+   * <pre>
+   * Node ids of concrete functions for saving and loading from a checkpoint.
+   * </pre>
+   *
+   * <code>int32 save_function = 2;</code>
+   */
+  public int getSaveFunction() {
+    return saveFunction_;
+  }
+
+  public static final int RESTORE_FUNCTION_FIELD_NUMBER = 3;
+  private int restoreFunction_;
+  /**
+   * <code>int32 restore_function = 3;</code>
+   */
+  public int getRestoreFunction() {
+    return restoreFunction_;
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    if (saveFunction_ != 0) {
+      output.writeInt32(2, saveFunction_);
+    }
+    if (restoreFunction_ != 0) {
+      output.writeInt32(3, restoreFunction_);
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (saveFunction_ != 0) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt32Size(2, saveFunction_);
+    }
+    if (restoreFunction_ != 0) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt32Size(3, restoreFunction_);
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.framework.SaveableObject)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.framework.SaveableObject other = (org.tensorflow.proto.framework.SaveableObject) obj;
+
+    if (getSaveFunction()
+        != other.getSaveFunction()) return false;
+    if (getRestoreFunction()
+        != other.getRestoreFunction()) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    hash = (37 * hash) + SAVE_FUNCTION_FIELD_NUMBER;
+    hash = (53 * hash) + getSaveFunction();
+    hash = (37 * hash) + RESTORE_FUNCTION_FIELD_NUMBER;
+    hash = (53 * hash) + getRestoreFunction();
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.framework.SaveableObject parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.SaveableObject parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.SaveableObject parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.SaveableObject parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.SaveableObject parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.SaveableObject parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.SaveableObject parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.SaveableObject parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.SaveableObject parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.SaveableObject parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.SaveableObject parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.SaveableObject parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.framework.SaveableObject prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * Protobuf type {@code tensorflow.SaveableObject}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.SaveableObject)
+      org.tensorflow.proto.framework.SaveableObjectOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.SavedObjectGraphProtos.internal_static_tensorflow_SaveableObject_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.SavedObjectGraphProtos.internal_static_tensorflow_SaveableObject_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.SaveableObject.class, org.tensorflow.proto.framework.SaveableObject.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.framework.SaveableObject.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      saveFunction_ = 0;
+
+      restoreFunction_ = 0;
+
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.framework.SavedObjectGraphProtos.internal_static_tensorflow_SaveableObject_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.SaveableObject getDefaultInstanceForType() {
+      return org.tensorflow.proto.framework.SaveableObject.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.SaveableObject build() {
+      org.tensorflow.proto.framework.SaveableObject result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.SaveableObject buildPartial() {
+      org.tensorflow.proto.framework.SaveableObject result = new org.tensorflow.proto.framework.SaveableObject(this);
+      result.saveFunction_ = saveFunction_;
+      result.restoreFunction_ = restoreFunction_;
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.framework.SaveableObject) {
+        return mergeFrom((org.tensorflow.proto.framework.SaveableObject)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.framework.SaveableObject other) {
+      if (other == org.tensorflow.proto.framework.SaveableObject.getDefaultInstance()) return this;
+      if (other.getSaveFunction() != 0) {
+        setSaveFunction(other.getSaveFunction());
+      }
+      if (other.getRestoreFunction() != 0) {
+        setRestoreFunction(other.getRestoreFunction());
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.framework.SaveableObject parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.framework.SaveableObject) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+
+    private int saveFunction_ ;
+    /**
+     * <pre>
+     * Node ids of concrete functions for saving and loading from a checkpoint.
+     * </pre>
+     *
+     * <code>int32 save_function = 2;</code>
+     */
+    public int getSaveFunction() {
+      return saveFunction_;
+    }
+    /**
+     * <pre>
+     * Node ids of concrete functions for saving and loading from a checkpoint.
+     * </pre>
+     *
+     * <code>int32 save_function = 2;</code>
+     */
+    public Builder setSaveFunction(int value) {
+      
+      saveFunction_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Node ids of concrete functions for saving and loading from a checkpoint.
+     * </pre>
+     *
+     * <code>int32 save_function = 2;</code>
+     */
+    public Builder clearSaveFunction() {
+      
+      saveFunction_ = 0;
+      onChanged();
+      return this;
+    }
+
+    private int restoreFunction_ ;
+    /**
+     * <code>int32 restore_function = 3;</code>
+     */
+    public int getRestoreFunction() {
+      return restoreFunction_;
+    }
+    /**
+     * <code>int32 restore_function = 3;</code>
+     */
+    public Builder setRestoreFunction(int value) {
+      
+      restoreFunction_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>int32 restore_function = 3;</code>
+     */
+    public Builder clearRestoreFunction() {
+      
+      restoreFunction_ = 0;
+      onChanged();
+      return this;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.SaveableObject)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.SaveableObject)
+  private static final org.tensorflow.proto.framework.SaveableObject DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.SaveableObject();
+  }
+
+  public static org.tensorflow.proto.framework.SaveableObject getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<SaveableObject>
+      PARSER = new com.google.protobuf.AbstractParser<SaveableObject>() {
+    @java.lang.Override
+    public SaveableObject parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new SaveableObject(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<SaveableObject> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<SaveableObject> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.framework.SaveableObject getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SaveableObjectOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SaveableObjectOrBuilder.java
new file mode 100644
index 00000000000..896dcfa8b34
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SaveableObjectOrBuilder.java
@@ -0,0 +1,23 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/saved_object_graph.proto
+
+package org.tensorflow.proto.framework;
+
+public interface SaveableObjectOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.SaveableObject)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <pre>
+   * Node ids of concrete functions for saving and loading from a checkpoint.
+   * </pre>
+   *
+   * <code>int32 save_function = 2;</code>
+   */
+  int getSaveFunction();
+
+  /**
+   * <code>int32 restore_function = 3;</code>
+   */
+  int getRestoreFunction();
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedAsset.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedAsset.java
new file mode 100644
index 00000000000..2558719f7c3
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedAsset.java
@@ -0,0 +1,514 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/saved_object_graph.proto
+
+package org.tensorflow.proto.framework;
+
+/**
+ * <pre>
+ * A SavedAsset points to an asset in the MetaGraph.
+ * When bound to a function this object evaluates to a tensor with the absolute
+ * filename. Users should not depend on a particular part of the filename to
+ * remain stable (e.g. basename could be changed).
+ * </pre>
+ *
+ * Protobuf type {@code tensorflow.SavedAsset}
+ */
+public  final class SavedAsset extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.SavedAsset)
+    SavedAssetOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use SavedAsset.newBuilder() to construct.
+  private SavedAsset(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private SavedAsset() {
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new SavedAsset();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private SavedAsset(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 8: {
+
+            assetFileDefIndex_ = input.readInt32();
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.framework.SavedObjectGraphProtos.internal_static_tensorflow_SavedAsset_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.framework.SavedObjectGraphProtos.internal_static_tensorflow_SavedAsset_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.framework.SavedAsset.class, org.tensorflow.proto.framework.SavedAsset.Builder.class);
+  }
+
+  public static final int ASSET_FILE_DEF_INDEX_FIELD_NUMBER = 1;
+  private int assetFileDefIndex_;
+  /**
+   * <pre>
+   * Index into `MetaGraphDef.asset_file_def[]` that describes the Asset.
+   * Only the field `AssetFileDef.filename` is used. Other fields, such as
+   * `AssetFileDef.tensor_info`, MUST be ignored.
+   * </pre>
+   *
+   * <code>int32 asset_file_def_index = 1;</code>
+   */
+  public int getAssetFileDefIndex() {
+    return assetFileDefIndex_;
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    if (assetFileDefIndex_ != 0) {
+      output.writeInt32(1, assetFileDefIndex_);
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (assetFileDefIndex_ != 0) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt32Size(1, assetFileDefIndex_);
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.framework.SavedAsset)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.framework.SavedAsset other = (org.tensorflow.proto.framework.SavedAsset) obj;
+
+    if (getAssetFileDefIndex()
+        != other.getAssetFileDefIndex()) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    hash = (37 * hash) + ASSET_FILE_DEF_INDEX_FIELD_NUMBER;
+    hash = (53 * hash) + getAssetFileDefIndex();
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.framework.SavedAsset parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.SavedAsset parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.SavedAsset parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.SavedAsset parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.SavedAsset parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.SavedAsset parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.SavedAsset parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.SavedAsset parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.SavedAsset parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.SavedAsset parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.SavedAsset parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.SavedAsset parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.framework.SavedAsset prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * <pre>
+   * A SavedAsset points to an asset in the MetaGraph.
+   * When bound to a function this object evaluates to a tensor with the absolute
+   * filename. Users should not depend on a particular part of the filename to
+   * remain stable (e.g. basename could be changed).
+   * </pre>
+   *
+   * Protobuf type {@code tensorflow.SavedAsset}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.SavedAsset)
+      org.tensorflow.proto.framework.SavedAssetOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.SavedObjectGraphProtos.internal_static_tensorflow_SavedAsset_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.SavedObjectGraphProtos.internal_static_tensorflow_SavedAsset_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.SavedAsset.class, org.tensorflow.proto.framework.SavedAsset.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.framework.SavedAsset.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      assetFileDefIndex_ = 0;
+
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.framework.SavedObjectGraphProtos.internal_static_tensorflow_SavedAsset_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.SavedAsset getDefaultInstanceForType() {
+      return org.tensorflow.proto.framework.SavedAsset.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.SavedAsset build() {
+      org.tensorflow.proto.framework.SavedAsset result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.SavedAsset buildPartial() {
+      org.tensorflow.proto.framework.SavedAsset result = new org.tensorflow.proto.framework.SavedAsset(this);
+      result.assetFileDefIndex_ = assetFileDefIndex_;
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.framework.SavedAsset) {
+        return mergeFrom((org.tensorflow.proto.framework.SavedAsset)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.framework.SavedAsset other) {
+      if (other == org.tensorflow.proto.framework.SavedAsset.getDefaultInstance()) return this;
+      if (other.getAssetFileDefIndex() != 0) {
+        setAssetFileDefIndex(other.getAssetFileDefIndex());
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.framework.SavedAsset parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.framework.SavedAsset) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+
+    private int assetFileDefIndex_ ;
+    /**
+     * <pre>
+     * Index into `MetaGraphDef.asset_file_def[]` that describes the Asset.
+     * Only the field `AssetFileDef.filename` is used. Other fields, such as
+     * `AssetFileDef.tensor_info`, MUST be ignored.
+     * </pre>
+     *
+     * <code>int32 asset_file_def_index = 1;</code>
+     */
+    public int getAssetFileDefIndex() {
+      return assetFileDefIndex_;
+    }
+    /**
+     * <pre>
+     * Index into `MetaGraphDef.asset_file_def[]` that describes the Asset.
+     * Only the field `AssetFileDef.filename` is used. Other fields, such as
+     * `AssetFileDef.tensor_info`, MUST be ignored.
+     * </pre>
+     *
+     * <code>int32 asset_file_def_index = 1;</code>
+     */
+    public Builder setAssetFileDefIndex(int value) {
+      
+      assetFileDefIndex_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Index into `MetaGraphDef.asset_file_def[]` that describes the Asset.
+     * Only the field `AssetFileDef.filename` is used. Other fields, such as
+     * `AssetFileDef.tensor_info`, MUST be ignored.
+     * </pre>
+     *
+     * <code>int32 asset_file_def_index = 1;</code>
+     */
+    public Builder clearAssetFileDefIndex() {
+      
+      assetFileDefIndex_ = 0;
+      onChanged();
+      return this;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.SavedAsset)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.SavedAsset)
+  private static final org.tensorflow.proto.framework.SavedAsset DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.SavedAsset();
+  }
+
+  public static org.tensorflow.proto.framework.SavedAsset getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<SavedAsset>
+      PARSER = new com.google.protobuf.AbstractParser<SavedAsset>() {
+    @java.lang.Override
+    public SavedAsset parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new SavedAsset(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<SavedAsset> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<SavedAsset> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.framework.SavedAsset getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedAssetOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedAssetOrBuilder.java
new file mode 100644
index 00000000000..a12a5ff6ad0
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedAssetOrBuilder.java
@@ -0,0 +1,20 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/saved_object_graph.proto
+
+package org.tensorflow.proto.framework;
+
+public interface SavedAssetOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.SavedAsset)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <pre>
+   * Index into `MetaGraphDef.asset_file_def[]` that describes the Asset.
+   * Only the field `AssetFileDef.filename` is used. Other fields, such as
+   * `AssetFileDef.tensor_info`, MUST be ignored.
+   * </pre>
+   *
+   * <code>int32 asset_file_def_index = 1;</code>
+   */
+  int getAssetFileDefIndex();
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedBareConcreteFunction.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedBareConcreteFunction.java
new file mode 100644
index 00000000000..e7fadf9f362
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedBareConcreteFunction.java
@@ -0,0 +1,873 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/saved_object_graph.proto
+
+package org.tensorflow.proto.framework;
+
+/**
+ * Protobuf type {@code tensorflow.SavedBareConcreteFunction}
+ */
+public  final class SavedBareConcreteFunction extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.SavedBareConcreteFunction)
+    SavedBareConcreteFunctionOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use SavedBareConcreteFunction.newBuilder() to construct.
+  private SavedBareConcreteFunction(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private SavedBareConcreteFunction() {
+    concreteFunctionName_ = "";
+    argumentKeywords_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new SavedBareConcreteFunction();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private SavedBareConcreteFunction(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    int mutable_bitField0_ = 0;
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 10: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            concreteFunctionName_ = s;
+            break;
+          }
+          case 18: {
+            java.lang.String s = input.readStringRequireUtf8();
+            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
+              argumentKeywords_ = new com.google.protobuf.LazyStringArrayList();
+              mutable_bitField0_ |= 0x00000001;
+            }
+            argumentKeywords_.add(s);
+            break;
+          }
+          case 24: {
+
+            allowedPositionalArguments_ = input.readInt64();
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      if (((mutable_bitField0_ & 0x00000001) != 0)) {
+        argumentKeywords_ = argumentKeywords_.getUnmodifiableView();
+      }
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.framework.SavedObjectGraphProtos.internal_static_tensorflow_SavedBareConcreteFunction_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.framework.SavedObjectGraphProtos.internal_static_tensorflow_SavedBareConcreteFunction_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.framework.SavedBareConcreteFunction.class, org.tensorflow.proto.framework.SavedBareConcreteFunction.Builder.class);
+  }
+
+  public static final int CONCRETE_FUNCTION_NAME_FIELD_NUMBER = 1;
+  private volatile java.lang.Object concreteFunctionName_;
+  /**
+   * <pre>
+   * Identifies a SavedConcreteFunction.
+   * </pre>
+   *
+   * <code>string concrete_function_name = 1;</code>
+   */
+  public java.lang.String getConcreteFunctionName() {
+    java.lang.Object ref = concreteFunctionName_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      concreteFunctionName_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * Identifies a SavedConcreteFunction.
+   * </pre>
+   *
+   * <code>string concrete_function_name = 1;</code>
+   */
+  public com.google.protobuf.ByteString
+      getConcreteFunctionNameBytes() {
+    java.lang.Object ref = concreteFunctionName_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      concreteFunctionName_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int ARGUMENT_KEYWORDS_FIELD_NUMBER = 2;
+  private com.google.protobuf.LazyStringList argumentKeywords_;
+  /**
+   * <pre>
+   * A sequence of unique strings, one per Tensor argument.
+   * </pre>
+   *
+   * <code>repeated string argument_keywords = 2;</code>
+   */
+  public com.google.protobuf.ProtocolStringList
+      getArgumentKeywordsList() {
+    return argumentKeywords_;
+  }
+  /**
+   * <pre>
+   * A sequence of unique strings, one per Tensor argument.
+   * </pre>
+   *
+   * <code>repeated string argument_keywords = 2;</code>
+   */
+  public int getArgumentKeywordsCount() {
+    return argumentKeywords_.size();
+  }
+  /**
+   * <pre>
+   * A sequence of unique strings, one per Tensor argument.
+   * </pre>
+   *
+   * <code>repeated string argument_keywords = 2;</code>
+   */
+  public java.lang.String getArgumentKeywords(int index) {
+    return argumentKeywords_.get(index);
+  }
+  /**
+   * <pre>
+   * A sequence of unique strings, one per Tensor argument.
+   * </pre>
+   *
+   * <code>repeated string argument_keywords = 2;</code>
+   */
+  public com.google.protobuf.ByteString
+      getArgumentKeywordsBytes(int index) {
+    return argumentKeywords_.getByteString(index);
+  }
+
+  public static final int ALLOWED_POSITIONAL_ARGUMENTS_FIELD_NUMBER = 3;
+  private long allowedPositionalArguments_;
+  /**
+   * <pre>
+   * The prefix of `argument_keywords` which may be identified by position.
+   * </pre>
+   *
+   * <code>int64 allowed_positional_arguments = 3;</code>
+   */
+  public long getAllowedPositionalArguments() {
+    return allowedPositionalArguments_;
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    if (!getConcreteFunctionNameBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 1, concreteFunctionName_);
+    }
+    for (int i = 0; i < argumentKeywords_.size(); i++) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 2, argumentKeywords_.getRaw(i));
+    }
+    if (allowedPositionalArguments_ != 0L) {
+      output.writeInt64(3, allowedPositionalArguments_);
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (!getConcreteFunctionNameBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, concreteFunctionName_);
+    }
+    {
+      int dataSize = 0;
+      for (int i = 0; i < argumentKeywords_.size(); i++) {
+        dataSize += computeStringSizeNoTag(argumentKeywords_.getRaw(i));
+      }
+      size += dataSize;
+      size += 1 * getArgumentKeywordsList().size();
+    }
+    if (allowedPositionalArguments_ != 0L) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt64Size(3, allowedPositionalArguments_);
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.framework.SavedBareConcreteFunction)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.framework.SavedBareConcreteFunction other = (org.tensorflow.proto.framework.SavedBareConcreteFunction) obj;
+
+    if (!getConcreteFunctionName()
+        .equals(other.getConcreteFunctionName())) return false;
+    if (!getArgumentKeywordsList()
+        .equals(other.getArgumentKeywordsList())) return false;
+    if (getAllowedPositionalArguments()
+        != other.getAllowedPositionalArguments()) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    hash = (37 * hash) + CONCRETE_FUNCTION_NAME_FIELD_NUMBER;
+    hash = (53 * hash) + getConcreteFunctionName().hashCode();
+    if (getArgumentKeywordsCount() > 0) {
+      hash = (37 * hash) + ARGUMENT_KEYWORDS_FIELD_NUMBER;
+      hash = (53 * hash) + getArgumentKeywordsList().hashCode();
+    }
+    hash = (37 * hash) + ALLOWED_POSITIONAL_ARGUMENTS_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+        getAllowedPositionalArguments());
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.framework.SavedBareConcreteFunction parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.SavedBareConcreteFunction parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.SavedBareConcreteFunction parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.SavedBareConcreteFunction parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.SavedBareConcreteFunction parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.SavedBareConcreteFunction parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.SavedBareConcreteFunction parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.SavedBareConcreteFunction parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.SavedBareConcreteFunction parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.SavedBareConcreteFunction parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.SavedBareConcreteFunction parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.SavedBareConcreteFunction parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.framework.SavedBareConcreteFunction prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * Protobuf type {@code tensorflow.SavedBareConcreteFunction}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.SavedBareConcreteFunction)
+      org.tensorflow.proto.framework.SavedBareConcreteFunctionOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.SavedObjectGraphProtos.internal_static_tensorflow_SavedBareConcreteFunction_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.SavedObjectGraphProtos.internal_static_tensorflow_SavedBareConcreteFunction_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.SavedBareConcreteFunction.class, org.tensorflow.proto.framework.SavedBareConcreteFunction.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.framework.SavedBareConcreteFunction.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      concreteFunctionName_ = "";
+
+      argumentKeywords_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+      bitField0_ = (bitField0_ & ~0x00000001);
+      allowedPositionalArguments_ = 0L;
+
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.framework.SavedObjectGraphProtos.internal_static_tensorflow_SavedBareConcreteFunction_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.SavedBareConcreteFunction getDefaultInstanceForType() {
+      return org.tensorflow.proto.framework.SavedBareConcreteFunction.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.SavedBareConcreteFunction build() {
+      org.tensorflow.proto.framework.SavedBareConcreteFunction result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.SavedBareConcreteFunction buildPartial() {
+      org.tensorflow.proto.framework.SavedBareConcreteFunction result = new org.tensorflow.proto.framework.SavedBareConcreteFunction(this);
+      int from_bitField0_ = bitField0_;
+      result.concreteFunctionName_ = concreteFunctionName_;
+      if (((bitField0_ & 0x00000001) != 0)) {
+        argumentKeywords_ = argumentKeywords_.getUnmodifiableView();
+        bitField0_ = (bitField0_ & ~0x00000001);
+      }
+      result.argumentKeywords_ = argumentKeywords_;
+      result.allowedPositionalArguments_ = allowedPositionalArguments_;
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.framework.SavedBareConcreteFunction) {
+        return mergeFrom((org.tensorflow.proto.framework.SavedBareConcreteFunction)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.framework.SavedBareConcreteFunction other) {
+      if (other == org.tensorflow.proto.framework.SavedBareConcreteFunction.getDefaultInstance()) return this;
+      if (!other.getConcreteFunctionName().isEmpty()) {
+        concreteFunctionName_ = other.concreteFunctionName_;
+        onChanged();
+      }
+      if (!other.argumentKeywords_.isEmpty()) {
+        if (argumentKeywords_.isEmpty()) {
+          argumentKeywords_ = other.argumentKeywords_;
+          bitField0_ = (bitField0_ & ~0x00000001);
+        } else {
+          ensureArgumentKeywordsIsMutable();
+          argumentKeywords_.addAll(other.argumentKeywords_);
+        }
+        onChanged();
+      }
+      if (other.getAllowedPositionalArguments() != 0L) {
+        setAllowedPositionalArguments(other.getAllowedPositionalArguments());
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.framework.SavedBareConcreteFunction parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.framework.SavedBareConcreteFunction) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+    private int bitField0_;
+
+    private java.lang.Object concreteFunctionName_ = "";
+    /**
+     * <pre>
+     * Identifies a SavedConcreteFunction.
+     * </pre>
+     *
+     * <code>string concrete_function_name = 1;</code>
+     */
+    public java.lang.String getConcreteFunctionName() {
+      java.lang.Object ref = concreteFunctionName_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        concreteFunctionName_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Identifies a SavedConcreteFunction.
+     * </pre>
+     *
+     * <code>string concrete_function_name = 1;</code>
+     */
+    public com.google.protobuf.ByteString
+        getConcreteFunctionNameBytes() {
+      java.lang.Object ref = concreteFunctionName_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        concreteFunctionName_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Identifies a SavedConcreteFunction.
+     * </pre>
+     *
+     * <code>string concrete_function_name = 1;</code>
+     */
+    public Builder setConcreteFunctionName(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      concreteFunctionName_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Identifies a SavedConcreteFunction.
+     * </pre>
+     *
+     * <code>string concrete_function_name = 1;</code>
+     */
+    public Builder clearConcreteFunctionName() {
+      
+      concreteFunctionName_ = getDefaultInstance().getConcreteFunctionName();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Identifies a SavedConcreteFunction.
+     * </pre>
+     *
+     * <code>string concrete_function_name = 1;</code>
+     */
+    public Builder setConcreteFunctionNameBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      concreteFunctionName_ = value;
+      onChanged();
+      return this;
+    }
+
+    private com.google.protobuf.LazyStringList argumentKeywords_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+    private void ensureArgumentKeywordsIsMutable() {
+      if (!((bitField0_ & 0x00000001) != 0)) {
+        argumentKeywords_ = new com.google.protobuf.LazyStringArrayList(argumentKeywords_);
+        bitField0_ |= 0x00000001;
+       }
+    }
+    /**
+     * <pre>
+     * A sequence of unique strings, one per Tensor argument.
+     * </pre>
+     *
+     * <code>repeated string argument_keywords = 2;</code>
+     */
+    public com.google.protobuf.ProtocolStringList
+        getArgumentKeywordsList() {
+      return argumentKeywords_.getUnmodifiableView();
+    }
+    /**
+     * <pre>
+     * A sequence of unique strings, one per Tensor argument.
+     * </pre>
+     *
+     * <code>repeated string argument_keywords = 2;</code>
+     */
+    public int getArgumentKeywordsCount() {
+      return argumentKeywords_.size();
+    }
+    /**
+     * <pre>
+     * A sequence of unique strings, one per Tensor argument.
+     * </pre>
+     *
+     * <code>repeated string argument_keywords = 2;</code>
+     */
+    public java.lang.String getArgumentKeywords(int index) {
+      return argumentKeywords_.get(index);
+    }
+    /**
+     * <pre>
+     * A sequence of unique strings, one per Tensor argument.
+     * </pre>
+     *
+     * <code>repeated string argument_keywords = 2;</code>
+     */
+    public com.google.protobuf.ByteString
+        getArgumentKeywordsBytes(int index) {
+      return argumentKeywords_.getByteString(index);
+    }
+    /**
+     * <pre>
+     * A sequence of unique strings, one per Tensor argument.
+     * </pre>
+     *
+     * <code>repeated string argument_keywords = 2;</code>
+     */
+    public Builder setArgumentKeywords(
+        int index, java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  ensureArgumentKeywordsIsMutable();
+      argumentKeywords_.set(index, value);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * A sequence of unique strings, one per Tensor argument.
+     * </pre>
+     *
+     * <code>repeated string argument_keywords = 2;</code>
+     */
+    public Builder addArgumentKeywords(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  ensureArgumentKeywordsIsMutable();
+      argumentKeywords_.add(value);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * A sequence of unique strings, one per Tensor argument.
+     * </pre>
+     *
+     * <code>repeated string argument_keywords = 2;</code>
+     */
+    public Builder addAllArgumentKeywords(
+        java.lang.Iterable<java.lang.String> values) {
+      ensureArgumentKeywordsIsMutable();
+      com.google.protobuf.AbstractMessageLite.Builder.addAll(
+          values, argumentKeywords_);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * A sequence of unique strings, one per Tensor argument.
+     * </pre>
+     *
+     * <code>repeated string argument_keywords = 2;</code>
+     */
+    public Builder clearArgumentKeywords() {
+      argumentKeywords_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+      bitField0_ = (bitField0_ & ~0x00000001);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * A sequence of unique strings, one per Tensor argument.
+     * </pre>
+     *
+     * <code>repeated string argument_keywords = 2;</code>
+     */
+    public Builder addArgumentKeywordsBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      ensureArgumentKeywordsIsMutable();
+      argumentKeywords_.add(value);
+      onChanged();
+      return this;
+    }
+
+    private long allowedPositionalArguments_ ;
+    /**
+     * <pre>
+     * The prefix of `argument_keywords` which may be identified by position.
+     * </pre>
+     *
+     * <code>int64 allowed_positional_arguments = 3;</code>
+     */
+    public long getAllowedPositionalArguments() {
+      return allowedPositionalArguments_;
+    }
+    /**
+     * <pre>
+     * The prefix of `argument_keywords` which may be identified by position.
+     * </pre>
+     *
+     * <code>int64 allowed_positional_arguments = 3;</code>
+     */
+    public Builder setAllowedPositionalArguments(long value) {
+      
+      allowedPositionalArguments_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * The prefix of `argument_keywords` which may be identified by position.
+     * </pre>
+     *
+     * <code>int64 allowed_positional_arguments = 3;</code>
+     */
+    public Builder clearAllowedPositionalArguments() {
+      
+      allowedPositionalArguments_ = 0L;
+      onChanged();
+      return this;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.SavedBareConcreteFunction)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.SavedBareConcreteFunction)
+  private static final org.tensorflow.proto.framework.SavedBareConcreteFunction DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.SavedBareConcreteFunction();
+  }
+
+  public static org.tensorflow.proto.framework.SavedBareConcreteFunction getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<SavedBareConcreteFunction>
+      PARSER = new com.google.protobuf.AbstractParser<SavedBareConcreteFunction>() {
+    @java.lang.Override
+    public SavedBareConcreteFunction parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new SavedBareConcreteFunction(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<SavedBareConcreteFunction> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<SavedBareConcreteFunction> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.framework.SavedBareConcreteFunction getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedBareConcreteFunctionOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedBareConcreteFunctionOrBuilder.java
new file mode 100644
index 00000000000..cf9a625e8b3
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedBareConcreteFunctionOrBuilder.java
@@ -0,0 +1,71 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/saved_object_graph.proto
+
+package org.tensorflow.proto.framework;
+
+public interface SavedBareConcreteFunctionOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.SavedBareConcreteFunction)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <pre>
+   * Identifies a SavedConcreteFunction.
+   * </pre>
+   *
+   * <code>string concrete_function_name = 1;</code>
+   */
+  java.lang.String getConcreteFunctionName();
+  /**
+   * <pre>
+   * Identifies a SavedConcreteFunction.
+   * </pre>
+   *
+   * <code>string concrete_function_name = 1;</code>
+   */
+  com.google.protobuf.ByteString
+      getConcreteFunctionNameBytes();
+
+  /**
+   * <pre>
+   * A sequence of unique strings, one per Tensor argument.
+   * </pre>
+   *
+   * <code>repeated string argument_keywords = 2;</code>
+   */
+  java.util.List<java.lang.String>
+      getArgumentKeywordsList();
+  /**
+   * <pre>
+   * A sequence of unique strings, one per Tensor argument.
+   * </pre>
+   *
+   * <code>repeated string argument_keywords = 2;</code>
+   */
+  int getArgumentKeywordsCount();
+  /**
+   * <pre>
+   * A sequence of unique strings, one per Tensor argument.
+   * </pre>
+   *
+   * <code>repeated string argument_keywords = 2;</code>
+   */
+  java.lang.String getArgumentKeywords(int index);
+  /**
+   * <pre>
+   * A sequence of unique strings, one per Tensor argument.
+   * </pre>
+   *
+   * <code>repeated string argument_keywords = 2;</code>
+   */
+  com.google.protobuf.ByteString
+      getArgumentKeywordsBytes(int index);
+
+  /**
+   * <pre>
+   * The prefix of `argument_keywords` which may be identified by position.
+   * </pre>
+   *
+   * <code>int64 allowed_positional_arguments = 3;</code>
+   */
+  long getAllowedPositionalArguments();
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedConcreteFunction.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedConcreteFunction.java
new file mode 100644
index 00000000000..62106345581
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedConcreteFunction.java
@@ -0,0 +1,1156 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/saved_object_graph.proto
+
+package org.tensorflow.proto.framework;
+
+/**
+ * <pre>
+ * Stores low-level information about a concrete function. Referenced in either
+ * a SavedFunction or a SavedBareConcreteFunction.
+ * </pre>
+ *
+ * Protobuf type {@code tensorflow.SavedConcreteFunction}
+ */
+public  final class SavedConcreteFunction extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.SavedConcreteFunction)
+    SavedConcreteFunctionOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use SavedConcreteFunction.newBuilder() to construct.
+  private SavedConcreteFunction(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private SavedConcreteFunction() {
+    boundInputs_ = emptyIntList();
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new SavedConcreteFunction();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private SavedConcreteFunction(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    int mutable_bitField0_ = 0;
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 16: {
+            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
+              boundInputs_ = newIntList();
+              mutable_bitField0_ |= 0x00000001;
+            }
+            boundInputs_.addInt(input.readInt32());
+            break;
+          }
+          case 18: {
+            int length = input.readRawVarint32();
+            int limit = input.pushLimit(length);
+            if (!((mutable_bitField0_ & 0x00000001) != 0) && input.getBytesUntilLimit() > 0) {
+              boundInputs_ = newIntList();
+              mutable_bitField0_ |= 0x00000001;
+            }
+            while (input.getBytesUntilLimit() > 0) {
+              boundInputs_.addInt(input.readInt32());
+            }
+            input.popLimit(limit);
+            break;
+          }
+          case 26: {
+            org.tensorflow.proto.framework.StructuredValue.Builder subBuilder = null;
+            if (canonicalizedInputSignature_ != null) {
+              subBuilder = canonicalizedInputSignature_.toBuilder();
+            }
+            canonicalizedInputSignature_ = input.readMessage(org.tensorflow.proto.framework.StructuredValue.parser(), extensionRegistry);
+            if (subBuilder != null) {
+              subBuilder.mergeFrom(canonicalizedInputSignature_);
+              canonicalizedInputSignature_ = subBuilder.buildPartial();
+            }
+
+            break;
+          }
+          case 34: {
+            org.tensorflow.proto.framework.StructuredValue.Builder subBuilder = null;
+            if (outputSignature_ != null) {
+              subBuilder = outputSignature_.toBuilder();
+            }
+            outputSignature_ = input.readMessage(org.tensorflow.proto.framework.StructuredValue.parser(), extensionRegistry);
+            if (subBuilder != null) {
+              subBuilder.mergeFrom(outputSignature_);
+              outputSignature_ = subBuilder.buildPartial();
+            }
+
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      if (((mutable_bitField0_ & 0x00000001) != 0)) {
+        boundInputs_.makeImmutable(); // C
+      }
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.framework.SavedObjectGraphProtos.internal_static_tensorflow_SavedConcreteFunction_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.framework.SavedObjectGraphProtos.internal_static_tensorflow_SavedConcreteFunction_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.framework.SavedConcreteFunction.class, org.tensorflow.proto.framework.SavedConcreteFunction.Builder.class);
+  }
+
+  public static final int BOUND_INPUTS_FIELD_NUMBER = 2;
+  private com.google.protobuf.Internal.IntList boundInputs_;
+  /**
+   * <pre>
+   * Bound inputs to the function. The SavedObjects identified by the node ids
+   * given here are appended as extra inputs to the caller-supplied inputs.
+   * The only types of SavedObjects valid here are SavedVariable, SavedResource
+   * and SavedAsset.
+   * </pre>
+   *
+   * <code>repeated int32 bound_inputs = 2;</code>
+   */
+  public java.util.List<java.lang.Integer>
+      getBoundInputsList() {
+    return boundInputs_;
+  }
+  /**
+   * <pre>
+   * Bound inputs to the function. The SavedObjects identified by the node ids
+   * given here are appended as extra inputs to the caller-supplied inputs.
+   * The only types of SavedObjects valid here are SavedVariable, SavedResource
+   * and SavedAsset.
+   * </pre>
+   *
+   * <code>repeated int32 bound_inputs = 2;</code>
+   */
+  public int getBoundInputsCount() {
+    return boundInputs_.size();
+  }
+  /**
+   * <pre>
+   * Bound inputs to the function. The SavedObjects identified by the node ids
+   * given here are appended as extra inputs to the caller-supplied inputs.
+   * The only types of SavedObjects valid here are SavedVariable, SavedResource
+   * and SavedAsset.
+   * </pre>
+   *
+   * <code>repeated int32 bound_inputs = 2;</code>
+   */
+  public int getBoundInputs(int index) {
+    return boundInputs_.getInt(index);
+  }
+  private int boundInputsMemoizedSerializedSize = -1;
+
+  public static final int CANONICALIZED_INPUT_SIGNATURE_FIELD_NUMBER = 3;
+  private org.tensorflow.proto.framework.StructuredValue canonicalizedInputSignature_;
+  /**
+   * <pre>
+   * Input in canonicalized form that was received to create this concrete
+   * function.
+   * </pre>
+   *
+   * <code>.tensorflow.StructuredValue canonicalized_input_signature = 3;</code>
+   */
+  public boolean hasCanonicalizedInputSignature() {
+    return canonicalizedInputSignature_ != null;
+  }
+  /**
+   * <pre>
+   * Input in canonicalized form that was received to create this concrete
+   * function.
+   * </pre>
+   *
+   * <code>.tensorflow.StructuredValue canonicalized_input_signature = 3;</code>
+   */
+  public org.tensorflow.proto.framework.StructuredValue getCanonicalizedInputSignature() {
+    return canonicalizedInputSignature_ == null ? org.tensorflow.proto.framework.StructuredValue.getDefaultInstance() : canonicalizedInputSignature_;
+  }
+  /**
+   * <pre>
+   * Input in canonicalized form that was received to create this concrete
+   * function.
+   * </pre>
+   *
+   * <code>.tensorflow.StructuredValue canonicalized_input_signature = 3;</code>
+   */
+  public org.tensorflow.proto.framework.StructuredValueOrBuilder getCanonicalizedInputSignatureOrBuilder() {
+    return getCanonicalizedInputSignature();
+  }
+
+  public static final int OUTPUT_SIGNATURE_FIELD_NUMBER = 4;
+  private org.tensorflow.proto.framework.StructuredValue outputSignature_;
+  /**
+   * <pre>
+   * Output that was the return value of this function after replacing all
+   * Tensors with TensorSpecs. This can be an arbitrary nested function and will
+   * be used to reconstruct the full structure from pure tensors.
+   * </pre>
+   *
+   * <code>.tensorflow.StructuredValue output_signature = 4;</code>
+   */
+  public boolean hasOutputSignature() {
+    return outputSignature_ != null;
+  }
+  /**
+   * <pre>
+   * Output that was the return value of this function after replacing all
+   * Tensors with TensorSpecs. This can be an arbitrary nested function and will
+   * be used to reconstruct the full structure from pure tensors.
+   * </pre>
+   *
+   * <code>.tensorflow.StructuredValue output_signature = 4;</code>
+   */
+  public org.tensorflow.proto.framework.StructuredValue getOutputSignature() {
+    return outputSignature_ == null ? org.tensorflow.proto.framework.StructuredValue.getDefaultInstance() : outputSignature_;
+  }
+  /**
+   * <pre>
+   * Output that was the return value of this function after replacing all
+   * Tensors with TensorSpecs. This can be an arbitrary nested function and will
+   * be used to reconstruct the full structure from pure tensors.
+   * </pre>
+   *
+   * <code>.tensorflow.StructuredValue output_signature = 4;</code>
+   */
+  public org.tensorflow.proto.framework.StructuredValueOrBuilder getOutputSignatureOrBuilder() {
+    return getOutputSignature();
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    getSerializedSize();
+    if (getBoundInputsList().size() > 0) {
+      output.writeUInt32NoTag(18);
+      output.writeUInt32NoTag(boundInputsMemoizedSerializedSize);
+    }
+    for (int i = 0; i < boundInputs_.size(); i++) {
+      output.writeInt32NoTag(boundInputs_.getInt(i));
+    }
+    if (canonicalizedInputSignature_ != null) {
+      output.writeMessage(3, getCanonicalizedInputSignature());
+    }
+    if (outputSignature_ != null) {
+      output.writeMessage(4, getOutputSignature());
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    {
+      int dataSize = 0;
+      for (int i = 0; i < boundInputs_.size(); i++) {
+        dataSize += com.google.protobuf.CodedOutputStream
+          .computeInt32SizeNoTag(boundInputs_.getInt(i));
+      }
+      size += dataSize;
+      if (!getBoundInputsList().isEmpty()) {
+        size += 1;
+        size += com.google.protobuf.CodedOutputStream
+            .computeInt32SizeNoTag(dataSize);
+      }
+      boundInputsMemoizedSerializedSize = dataSize;
+    }
+    if (canonicalizedInputSignature_ != null) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(3, getCanonicalizedInputSignature());
+    }
+    if (outputSignature_ != null) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(4, getOutputSignature());
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.framework.SavedConcreteFunction)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.framework.SavedConcreteFunction other = (org.tensorflow.proto.framework.SavedConcreteFunction) obj;
+
+    if (!getBoundInputsList()
+        .equals(other.getBoundInputsList())) return false;
+    if (hasCanonicalizedInputSignature() != other.hasCanonicalizedInputSignature()) return false;
+    if (hasCanonicalizedInputSignature()) {
+      if (!getCanonicalizedInputSignature()
+          .equals(other.getCanonicalizedInputSignature())) return false;
+    }
+    if (hasOutputSignature() != other.hasOutputSignature()) return false;
+    if (hasOutputSignature()) {
+      if (!getOutputSignature()
+          .equals(other.getOutputSignature())) return false;
+    }
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    if (getBoundInputsCount() > 0) {
+      hash = (37 * hash) + BOUND_INPUTS_FIELD_NUMBER;
+      hash = (53 * hash) + getBoundInputsList().hashCode();
+    }
+    if (hasCanonicalizedInputSignature()) {
+      hash = (37 * hash) + CANONICALIZED_INPUT_SIGNATURE_FIELD_NUMBER;
+      hash = (53 * hash) + getCanonicalizedInputSignature().hashCode();
+    }
+    if (hasOutputSignature()) {
+      hash = (37 * hash) + OUTPUT_SIGNATURE_FIELD_NUMBER;
+      hash = (53 * hash) + getOutputSignature().hashCode();
+    }
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.framework.SavedConcreteFunction parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.SavedConcreteFunction parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.SavedConcreteFunction parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.SavedConcreteFunction parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.SavedConcreteFunction parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.SavedConcreteFunction parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.SavedConcreteFunction parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.SavedConcreteFunction parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.SavedConcreteFunction parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.SavedConcreteFunction parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.SavedConcreteFunction parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.SavedConcreteFunction parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.framework.SavedConcreteFunction prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * <pre>
+   * Stores low-level information about a concrete function. Referenced in either
+   * a SavedFunction or a SavedBareConcreteFunction.
+   * </pre>
+   *
+   * Protobuf type {@code tensorflow.SavedConcreteFunction}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.SavedConcreteFunction)
+      org.tensorflow.proto.framework.SavedConcreteFunctionOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.SavedObjectGraphProtos.internal_static_tensorflow_SavedConcreteFunction_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.SavedObjectGraphProtos.internal_static_tensorflow_SavedConcreteFunction_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.SavedConcreteFunction.class, org.tensorflow.proto.framework.SavedConcreteFunction.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.framework.SavedConcreteFunction.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      boundInputs_ = emptyIntList();
+      bitField0_ = (bitField0_ & ~0x00000001);
+      if (canonicalizedInputSignatureBuilder_ == null) {
+        canonicalizedInputSignature_ = null;
+      } else {
+        canonicalizedInputSignature_ = null;
+        canonicalizedInputSignatureBuilder_ = null;
+      }
+      if (outputSignatureBuilder_ == null) {
+        outputSignature_ = null;
+      } else {
+        outputSignature_ = null;
+        outputSignatureBuilder_ = null;
+      }
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.framework.SavedObjectGraphProtos.internal_static_tensorflow_SavedConcreteFunction_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.SavedConcreteFunction getDefaultInstanceForType() {
+      return org.tensorflow.proto.framework.SavedConcreteFunction.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.SavedConcreteFunction build() {
+      org.tensorflow.proto.framework.SavedConcreteFunction result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.SavedConcreteFunction buildPartial() {
+      org.tensorflow.proto.framework.SavedConcreteFunction result = new org.tensorflow.proto.framework.SavedConcreteFunction(this);
+      int from_bitField0_ = bitField0_;
+      if (((bitField0_ & 0x00000001) != 0)) {
+        boundInputs_.makeImmutable();
+        bitField0_ = (bitField0_ & ~0x00000001);
+      }
+      result.boundInputs_ = boundInputs_;
+      if (canonicalizedInputSignatureBuilder_ == null) {
+        result.canonicalizedInputSignature_ = canonicalizedInputSignature_;
+      } else {
+        result.canonicalizedInputSignature_ = canonicalizedInputSignatureBuilder_.build();
+      }
+      if (outputSignatureBuilder_ == null) {
+        result.outputSignature_ = outputSignature_;
+      } else {
+        result.outputSignature_ = outputSignatureBuilder_.build();
+      }
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.framework.SavedConcreteFunction) {
+        return mergeFrom((org.tensorflow.proto.framework.SavedConcreteFunction)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.framework.SavedConcreteFunction other) {
+      if (other == org.tensorflow.proto.framework.SavedConcreteFunction.getDefaultInstance()) return this;
+      if (!other.boundInputs_.isEmpty()) {
+        if (boundInputs_.isEmpty()) {
+          boundInputs_ = other.boundInputs_;
+          bitField0_ = (bitField0_ & ~0x00000001);
+        } else {
+          ensureBoundInputsIsMutable();
+          boundInputs_.addAll(other.boundInputs_);
+        }
+        onChanged();
+      }
+      if (other.hasCanonicalizedInputSignature()) {
+        mergeCanonicalizedInputSignature(other.getCanonicalizedInputSignature());
+      }
+      if (other.hasOutputSignature()) {
+        mergeOutputSignature(other.getOutputSignature());
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.framework.SavedConcreteFunction parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.framework.SavedConcreteFunction) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+    private int bitField0_;
+
+    private com.google.protobuf.Internal.IntList boundInputs_ = emptyIntList();
+    private void ensureBoundInputsIsMutable() {
+      if (!((bitField0_ & 0x00000001) != 0)) {
+        boundInputs_ = mutableCopy(boundInputs_);
+        bitField0_ |= 0x00000001;
+       }
+    }
+    /**
+     * <pre>
+     * Bound inputs to the function. The SavedObjects identified by the node ids
+     * given here are appended as extra inputs to the caller-supplied inputs.
+     * The only types of SavedObjects valid here are SavedVariable, SavedResource
+     * and SavedAsset.
+     * </pre>
+     *
+     * <code>repeated int32 bound_inputs = 2;</code>
+     */
+    public java.util.List<java.lang.Integer>
+        getBoundInputsList() {
+      return ((bitField0_ & 0x00000001) != 0) ?
+               java.util.Collections.unmodifiableList(boundInputs_) : boundInputs_;
+    }
+    /**
+     * <pre>
+     * Bound inputs to the function. The SavedObjects identified by the node ids
+     * given here are appended as extra inputs to the caller-supplied inputs.
+     * The only types of SavedObjects valid here are SavedVariable, SavedResource
+     * and SavedAsset.
+     * </pre>
+     *
+     * <code>repeated int32 bound_inputs = 2;</code>
+     */
+    public int getBoundInputsCount() {
+      return boundInputs_.size();
+    }
+    /**
+     * <pre>
+     * Bound inputs to the function. The SavedObjects identified by the node ids
+     * given here are appended as extra inputs to the caller-supplied inputs.
+     * The only types of SavedObjects valid here are SavedVariable, SavedResource
+     * and SavedAsset.
+     * </pre>
+     *
+     * <code>repeated int32 bound_inputs = 2;</code>
+     */
+    public int getBoundInputs(int index) {
+      return boundInputs_.getInt(index);
+    }
+    /**
+     * <pre>
+     * Bound inputs to the function. The SavedObjects identified by the node ids
+     * given here are appended as extra inputs to the caller-supplied inputs.
+     * The only types of SavedObjects valid here are SavedVariable, SavedResource
+     * and SavedAsset.
+     * </pre>
+     *
+     * <code>repeated int32 bound_inputs = 2;</code>
+     */
+    public Builder setBoundInputs(
+        int index, int value) {
+      ensureBoundInputsIsMutable();
+      boundInputs_.setInt(index, value);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Bound inputs to the function. The SavedObjects identified by the node ids
+     * given here are appended as extra inputs to the caller-supplied inputs.
+     * The only types of SavedObjects valid here are SavedVariable, SavedResource
+     * and SavedAsset.
+     * </pre>
+     *
+     * <code>repeated int32 bound_inputs = 2;</code>
+     */
+    public Builder addBoundInputs(int value) {
+      ensureBoundInputsIsMutable();
+      boundInputs_.addInt(value);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Bound inputs to the function. The SavedObjects identified by the node ids
+     * given here are appended as extra inputs to the caller-supplied inputs.
+     * The only types of SavedObjects valid here are SavedVariable, SavedResource
+     * and SavedAsset.
+     * </pre>
+     *
+     * <code>repeated int32 bound_inputs = 2;</code>
+     */
+    public Builder addAllBoundInputs(
+        java.lang.Iterable<? extends java.lang.Integer> values) {
+      ensureBoundInputsIsMutable();
+      com.google.protobuf.AbstractMessageLite.Builder.addAll(
+          values, boundInputs_);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Bound inputs to the function. The SavedObjects identified by the node ids
+     * given here are appended as extra inputs to the caller-supplied inputs.
+     * The only types of SavedObjects valid here are SavedVariable, SavedResource
+     * and SavedAsset.
+     * </pre>
+     *
+     * <code>repeated int32 bound_inputs = 2;</code>
+     */
+    public Builder clearBoundInputs() {
+      boundInputs_ = emptyIntList();
+      bitField0_ = (bitField0_ & ~0x00000001);
+      onChanged();
+      return this;
+    }
+
+    private org.tensorflow.proto.framework.StructuredValue canonicalizedInputSignature_;
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.StructuredValue, org.tensorflow.proto.framework.StructuredValue.Builder, org.tensorflow.proto.framework.StructuredValueOrBuilder> canonicalizedInputSignatureBuilder_;
+    /**
+     * <pre>
+     * Input in canonicalized form that was received to create this concrete
+     * function.
+     * </pre>
+     *
+     * <code>.tensorflow.StructuredValue canonicalized_input_signature = 3;</code>
+     */
+    public boolean hasCanonicalizedInputSignature() {
+      return canonicalizedInputSignatureBuilder_ != null || canonicalizedInputSignature_ != null;
+    }
+    /**
+     * <pre>
+     * Input in canonicalized form that was received to create this concrete
+     * function.
+     * </pre>
+     *
+     * <code>.tensorflow.StructuredValue canonicalized_input_signature = 3;</code>
+     */
+    public org.tensorflow.proto.framework.StructuredValue getCanonicalizedInputSignature() {
+      if (canonicalizedInputSignatureBuilder_ == null) {
+        return canonicalizedInputSignature_ == null ? org.tensorflow.proto.framework.StructuredValue.getDefaultInstance() : canonicalizedInputSignature_;
+      } else {
+        return canonicalizedInputSignatureBuilder_.getMessage();
+      }
+    }
+    /**
+     * <pre>
+     * Input in canonicalized form that was received to create this concrete
+     * function.
+     * </pre>
+     *
+     * <code>.tensorflow.StructuredValue canonicalized_input_signature = 3;</code>
+     */
+    public Builder setCanonicalizedInputSignature(org.tensorflow.proto.framework.StructuredValue value) {
+      if (canonicalizedInputSignatureBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        canonicalizedInputSignature_ = value;
+        onChanged();
+      } else {
+        canonicalizedInputSignatureBuilder_.setMessage(value);
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Input in canonicalized form that was received to create this concrete
+     * function.
+     * </pre>
+     *
+     * <code>.tensorflow.StructuredValue canonicalized_input_signature = 3;</code>
+     */
+    public Builder setCanonicalizedInputSignature(
+        org.tensorflow.proto.framework.StructuredValue.Builder builderForValue) {
+      if (canonicalizedInputSignatureBuilder_ == null) {
+        canonicalizedInputSignature_ = builderForValue.build();
+        onChanged();
+      } else {
+        canonicalizedInputSignatureBuilder_.setMessage(builderForValue.build());
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Input in canonicalized form that was received to create this concrete
+     * function.
+     * </pre>
+     *
+     * <code>.tensorflow.StructuredValue canonicalized_input_signature = 3;</code>
+     */
+    public Builder mergeCanonicalizedInputSignature(org.tensorflow.proto.framework.StructuredValue value) {
+      if (canonicalizedInputSignatureBuilder_ == null) {
+        if (canonicalizedInputSignature_ != null) {
+          canonicalizedInputSignature_ =
+            org.tensorflow.proto.framework.StructuredValue.newBuilder(canonicalizedInputSignature_).mergeFrom(value).buildPartial();
+        } else {
+          canonicalizedInputSignature_ = value;
+        }
+        onChanged();
+      } else {
+        canonicalizedInputSignatureBuilder_.mergeFrom(value);
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Input in canonicalized form that was received to create this concrete
+     * function.
+     * </pre>
+     *
+     * <code>.tensorflow.StructuredValue canonicalized_input_signature = 3;</code>
+     */
+    public Builder clearCanonicalizedInputSignature() {
+      if (canonicalizedInputSignatureBuilder_ == null) {
+        canonicalizedInputSignature_ = null;
+        onChanged();
+      } else {
+        canonicalizedInputSignature_ = null;
+        canonicalizedInputSignatureBuilder_ = null;
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Input in canonicalized form that was received to create this concrete
+     * function.
+     * </pre>
+     *
+     * <code>.tensorflow.StructuredValue canonicalized_input_signature = 3;</code>
+     */
+    public org.tensorflow.proto.framework.StructuredValue.Builder getCanonicalizedInputSignatureBuilder() {
+      
+      onChanged();
+      return getCanonicalizedInputSignatureFieldBuilder().getBuilder();
+    }
+    /**
+     * <pre>
+     * Input in canonicalized form that was received to create this concrete
+     * function.
+     * </pre>
+     *
+     * <code>.tensorflow.StructuredValue canonicalized_input_signature = 3;</code>
+     */
+    public org.tensorflow.proto.framework.StructuredValueOrBuilder getCanonicalizedInputSignatureOrBuilder() {
+      if (canonicalizedInputSignatureBuilder_ != null) {
+        return canonicalizedInputSignatureBuilder_.getMessageOrBuilder();
+      } else {
+        return canonicalizedInputSignature_ == null ?
+            org.tensorflow.proto.framework.StructuredValue.getDefaultInstance() : canonicalizedInputSignature_;
+      }
+    }
+    /**
+     * <pre>
+     * Input in canonicalized form that was received to create this concrete
+     * function.
+     * </pre>
+     *
+     * <code>.tensorflow.StructuredValue canonicalized_input_signature = 3;</code>
+     */
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.StructuredValue, org.tensorflow.proto.framework.StructuredValue.Builder, org.tensorflow.proto.framework.StructuredValueOrBuilder> 
+        getCanonicalizedInputSignatureFieldBuilder() {
+      if (canonicalizedInputSignatureBuilder_ == null) {
+        canonicalizedInputSignatureBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+            org.tensorflow.proto.framework.StructuredValue, org.tensorflow.proto.framework.StructuredValue.Builder, org.tensorflow.proto.framework.StructuredValueOrBuilder>(
+                getCanonicalizedInputSignature(),
+                getParentForChildren(),
+                isClean());
+        canonicalizedInputSignature_ = null;
+      }
+      return canonicalizedInputSignatureBuilder_;
+    }
+
+    private org.tensorflow.proto.framework.StructuredValue outputSignature_;
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.StructuredValue, org.tensorflow.proto.framework.StructuredValue.Builder, org.tensorflow.proto.framework.StructuredValueOrBuilder> outputSignatureBuilder_;
+    /**
+     * <pre>
+     * Output that was the return value of this function after replacing all
+     * Tensors with TensorSpecs. This can be an arbitrary nested function and will
+     * be used to reconstruct the full structure from pure tensors.
+     * </pre>
+     *
+     * <code>.tensorflow.StructuredValue output_signature = 4;</code>
+     */
+    public boolean hasOutputSignature() {
+      return outputSignatureBuilder_ != null || outputSignature_ != null;
+    }
+    /**
+     * <pre>
+     * Output that was the return value of this function after replacing all
+     * Tensors with TensorSpecs. This can be an arbitrary nested function and will
+     * be used to reconstruct the full structure from pure tensors.
+     * </pre>
+     *
+     * <code>.tensorflow.StructuredValue output_signature = 4;</code>
+     */
+    public org.tensorflow.proto.framework.StructuredValue getOutputSignature() {
+      if (outputSignatureBuilder_ == null) {
+        return outputSignature_ == null ? org.tensorflow.proto.framework.StructuredValue.getDefaultInstance() : outputSignature_;
+      } else {
+        return outputSignatureBuilder_.getMessage();
+      }
+    }
+    /**
+     * <pre>
+     * Output that was the return value of this function after replacing all
+     * Tensors with TensorSpecs. This can be an arbitrary nested function and will
+     * be used to reconstruct the full structure from pure tensors.
+     * </pre>
+     *
+     * <code>.tensorflow.StructuredValue output_signature = 4;</code>
+     */
+    public Builder setOutputSignature(org.tensorflow.proto.framework.StructuredValue value) {
+      if (outputSignatureBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        outputSignature_ = value;
+        onChanged();
+      } else {
+        outputSignatureBuilder_.setMessage(value);
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Output that was the return value of this function after replacing all
+     * Tensors with TensorSpecs. This can be an arbitrary nested function and will
+     * be used to reconstruct the full structure from pure tensors.
+     * </pre>
+     *
+     * <code>.tensorflow.StructuredValue output_signature = 4;</code>
+     */
+    public Builder setOutputSignature(
+        org.tensorflow.proto.framework.StructuredValue.Builder builderForValue) {
+      if (outputSignatureBuilder_ == null) {
+        outputSignature_ = builderForValue.build();
+        onChanged();
+      } else {
+        outputSignatureBuilder_.setMessage(builderForValue.build());
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Output that was the return value of this function after replacing all
+     * Tensors with TensorSpecs. This can be an arbitrary nested function and will
+     * be used to reconstruct the full structure from pure tensors.
+     * </pre>
+     *
+     * <code>.tensorflow.StructuredValue output_signature = 4;</code>
+     */
+    public Builder mergeOutputSignature(org.tensorflow.proto.framework.StructuredValue value) {
+      if (outputSignatureBuilder_ == null) {
+        if (outputSignature_ != null) {
+          outputSignature_ =
+            org.tensorflow.proto.framework.StructuredValue.newBuilder(outputSignature_).mergeFrom(value).buildPartial();
+        } else {
+          outputSignature_ = value;
+        }
+        onChanged();
+      } else {
+        outputSignatureBuilder_.mergeFrom(value);
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Output that was the return value of this function after replacing all
+     * Tensors with TensorSpecs. This can be an arbitrary nested function and will
+     * be used to reconstruct the full structure from pure tensors.
+     * </pre>
+     *
+     * <code>.tensorflow.StructuredValue output_signature = 4;</code>
+     */
+    public Builder clearOutputSignature() {
+      if (outputSignatureBuilder_ == null) {
+        outputSignature_ = null;
+        onChanged();
+      } else {
+        outputSignature_ = null;
+        outputSignatureBuilder_ = null;
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Output that was the return value of this function after replacing all
+     * Tensors with TensorSpecs. This can be an arbitrary nested function and will
+     * be used to reconstruct the full structure from pure tensors.
+     * </pre>
+     *
+     * <code>.tensorflow.StructuredValue output_signature = 4;</code>
+     */
+    public org.tensorflow.proto.framework.StructuredValue.Builder getOutputSignatureBuilder() {
+      
+      onChanged();
+      return getOutputSignatureFieldBuilder().getBuilder();
+    }
+    /**
+     * <pre>
+     * Output that was the return value of this function after replacing all
+     * Tensors with TensorSpecs. This can be an arbitrary nested function and will
+     * be used to reconstruct the full structure from pure tensors.
+     * </pre>
+     *
+     * <code>.tensorflow.StructuredValue output_signature = 4;</code>
+     */
+    public org.tensorflow.proto.framework.StructuredValueOrBuilder getOutputSignatureOrBuilder() {
+      if (outputSignatureBuilder_ != null) {
+        return outputSignatureBuilder_.getMessageOrBuilder();
+      } else {
+        return outputSignature_ == null ?
+            org.tensorflow.proto.framework.StructuredValue.getDefaultInstance() : outputSignature_;
+      }
+    }
+    /**
+     * <pre>
+     * Output that was the return value of this function after replacing all
+     * Tensors with TensorSpecs. This can be an arbitrary nested function and will
+     * be used to reconstruct the full structure from pure tensors.
+     * </pre>
+     *
+     * <code>.tensorflow.StructuredValue output_signature = 4;</code>
+     */
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.StructuredValue, org.tensorflow.proto.framework.StructuredValue.Builder, org.tensorflow.proto.framework.StructuredValueOrBuilder> 
+        getOutputSignatureFieldBuilder() {
+      if (outputSignatureBuilder_ == null) {
+        outputSignatureBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+            org.tensorflow.proto.framework.StructuredValue, org.tensorflow.proto.framework.StructuredValue.Builder, org.tensorflow.proto.framework.StructuredValueOrBuilder>(
+                getOutputSignature(),
+                getParentForChildren(),
+                isClean());
+        outputSignature_ = null;
+      }
+      return outputSignatureBuilder_;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.SavedConcreteFunction)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.SavedConcreteFunction)
+  private static final org.tensorflow.proto.framework.SavedConcreteFunction DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.SavedConcreteFunction();
+  }
+
+  public static org.tensorflow.proto.framework.SavedConcreteFunction getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<SavedConcreteFunction>
+      PARSER = new com.google.protobuf.AbstractParser<SavedConcreteFunction>() {
+    @java.lang.Override
+    public SavedConcreteFunction parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new SavedConcreteFunction(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<SavedConcreteFunction> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<SavedConcreteFunction> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.framework.SavedConcreteFunction getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedConcreteFunctionOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedConcreteFunctionOrBuilder.java
new file mode 100644
index 00000000000..49145b8ca01
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedConcreteFunctionOrBuilder.java
@@ -0,0 +1,102 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/saved_object_graph.proto
+
+package org.tensorflow.proto.framework;
+
+public interface SavedConcreteFunctionOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.SavedConcreteFunction)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <pre>
+   * Bound inputs to the function. The SavedObjects identified by the node ids
+   * given here are appended as extra inputs to the caller-supplied inputs.
+   * The only types of SavedObjects valid here are SavedVariable, SavedResource
+   * and SavedAsset.
+   * </pre>
+   *
+   * <code>repeated int32 bound_inputs = 2;</code>
+   */
+  java.util.List<java.lang.Integer> getBoundInputsList();
+  /**
+   * <pre>
+   * Bound inputs to the function. The SavedObjects identified by the node ids
+   * given here are appended as extra inputs to the caller-supplied inputs.
+   * The only types of SavedObjects valid here are SavedVariable, SavedResource
+   * and SavedAsset.
+   * </pre>
+   *
+   * <code>repeated int32 bound_inputs = 2;</code>
+   */
+  int getBoundInputsCount();
+  /**
+   * <pre>
+   * Bound inputs to the function. The SavedObjects identified by the node ids
+   * given here are appended as extra inputs to the caller-supplied inputs.
+   * The only types of SavedObjects valid here are SavedVariable, SavedResource
+   * and SavedAsset.
+   * </pre>
+   *
+   * <code>repeated int32 bound_inputs = 2;</code>
+   */
+  int getBoundInputs(int index);
+
+  /**
+   * <pre>
+   * Input in canonicalized form that was received to create this concrete
+   * function.
+   * </pre>
+   *
+   * <code>.tensorflow.StructuredValue canonicalized_input_signature = 3;</code>
+   */
+  boolean hasCanonicalizedInputSignature();
+  /**
+   * <pre>
+   * Input in canonicalized form that was received to create this concrete
+   * function.
+   * </pre>
+   *
+   * <code>.tensorflow.StructuredValue canonicalized_input_signature = 3;</code>
+   */
+  org.tensorflow.proto.framework.StructuredValue getCanonicalizedInputSignature();
+  /**
+   * <pre>
+   * Input in canonicalized form that was received to create this concrete
+   * function.
+   * </pre>
+   *
+   * <code>.tensorflow.StructuredValue canonicalized_input_signature = 3;</code>
+   */
+  org.tensorflow.proto.framework.StructuredValueOrBuilder getCanonicalizedInputSignatureOrBuilder();
+
+  /**
+   * <pre>
+   * Output that was the return value of this function after replacing all
+   * Tensors with TensorSpecs. This can be an arbitrary nested function and will
+   * be used to reconstruct the full structure from pure tensors.
+   * </pre>
+   *
+   * <code>.tensorflow.StructuredValue output_signature = 4;</code>
+   */
+  boolean hasOutputSignature();
+  /**
+   * <pre>
+   * Output that was the return value of this function after replacing all
+   * Tensors with TensorSpecs. This can be an arbitrary nested function and will
+   * be used to reconstruct the full structure from pure tensors.
+   * </pre>
+   *
+   * <code>.tensorflow.StructuredValue output_signature = 4;</code>
+   */
+  org.tensorflow.proto.framework.StructuredValue getOutputSignature();
+  /**
+   * <pre>
+   * Output that was the return value of this function after replacing all
+   * Tensors with TensorSpecs. This can be an arbitrary nested function and will
+   * be used to reconstruct the full structure from pure tensors.
+   * </pre>
+   *
+   * <code>.tensorflow.StructuredValue output_signature = 4;</code>
+   */
+  org.tensorflow.proto.framework.StructuredValueOrBuilder getOutputSignatureOrBuilder();
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedConstant.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedConstant.java
new file mode 100644
index 00000000000..4ba709c9c98
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedConstant.java
@@ -0,0 +1,574 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/saved_object_graph.proto
+
+package org.tensorflow.proto.framework;
+
+/**
+ * Protobuf type {@code tensorflow.SavedConstant}
+ */
+public  final class SavedConstant extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.SavedConstant)
+    SavedConstantOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use SavedConstant.newBuilder() to construct.
+  private SavedConstant(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private SavedConstant() {
+    operation_ = "";
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new SavedConstant();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private SavedConstant(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 10: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            operation_ = s;
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.framework.SavedObjectGraphProtos.internal_static_tensorflow_SavedConstant_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.framework.SavedObjectGraphProtos.internal_static_tensorflow_SavedConstant_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.framework.SavedConstant.class, org.tensorflow.proto.framework.SavedConstant.Builder.class);
+  }
+
+  public static final int OPERATION_FIELD_NUMBER = 1;
+  private volatile java.lang.Object operation_;
+  /**
+   * <pre>
+   * An Operation name for a ConstantOp in this SavedObjectGraph's MetaGraph.
+   * </pre>
+   *
+   * <code>string operation = 1;</code>
+   */
+  public java.lang.String getOperation() {
+    java.lang.Object ref = operation_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      operation_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * An Operation name for a ConstantOp in this SavedObjectGraph's MetaGraph.
+   * </pre>
+   *
+   * <code>string operation = 1;</code>
+   */
+  public com.google.protobuf.ByteString
+      getOperationBytes() {
+    java.lang.Object ref = operation_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      operation_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    if (!getOperationBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 1, operation_);
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (!getOperationBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, operation_);
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.framework.SavedConstant)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.framework.SavedConstant other = (org.tensorflow.proto.framework.SavedConstant) obj;
+
+    if (!getOperation()
+        .equals(other.getOperation())) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    hash = (37 * hash) + OPERATION_FIELD_NUMBER;
+    hash = (53 * hash) + getOperation().hashCode();
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.framework.SavedConstant parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.SavedConstant parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.SavedConstant parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.SavedConstant parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.SavedConstant parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.SavedConstant parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.SavedConstant parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.SavedConstant parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.SavedConstant parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.SavedConstant parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.SavedConstant parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.SavedConstant parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.framework.SavedConstant prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * Protobuf type {@code tensorflow.SavedConstant}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.SavedConstant)
+      org.tensorflow.proto.framework.SavedConstantOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.SavedObjectGraphProtos.internal_static_tensorflow_SavedConstant_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.SavedObjectGraphProtos.internal_static_tensorflow_SavedConstant_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.SavedConstant.class, org.tensorflow.proto.framework.SavedConstant.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.framework.SavedConstant.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      operation_ = "";
+
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.framework.SavedObjectGraphProtos.internal_static_tensorflow_SavedConstant_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.SavedConstant getDefaultInstanceForType() {
+      return org.tensorflow.proto.framework.SavedConstant.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.SavedConstant build() {
+      org.tensorflow.proto.framework.SavedConstant result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.SavedConstant buildPartial() {
+      org.tensorflow.proto.framework.SavedConstant result = new org.tensorflow.proto.framework.SavedConstant(this);
+      result.operation_ = operation_;
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.framework.SavedConstant) {
+        return mergeFrom((org.tensorflow.proto.framework.SavedConstant)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.framework.SavedConstant other) {
+      if (other == org.tensorflow.proto.framework.SavedConstant.getDefaultInstance()) return this;
+      if (!other.getOperation().isEmpty()) {
+        operation_ = other.operation_;
+        onChanged();
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.framework.SavedConstant parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.framework.SavedConstant) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+
+    private java.lang.Object operation_ = "";
+    /**
+     * <pre>
+     * An Operation name for a ConstantOp in this SavedObjectGraph's MetaGraph.
+     * </pre>
+     *
+     * <code>string operation = 1;</code>
+     */
+    public java.lang.String getOperation() {
+      java.lang.Object ref = operation_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        operation_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * An Operation name for a ConstantOp in this SavedObjectGraph's MetaGraph.
+     * </pre>
+     *
+     * <code>string operation = 1;</code>
+     */
+    public com.google.protobuf.ByteString
+        getOperationBytes() {
+      java.lang.Object ref = operation_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        operation_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * An Operation name for a ConstantOp in this SavedObjectGraph's MetaGraph.
+     * </pre>
+     *
+     * <code>string operation = 1;</code>
+     */
+    public Builder setOperation(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      operation_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * An Operation name for a ConstantOp in this SavedObjectGraph's MetaGraph.
+     * </pre>
+     *
+     * <code>string operation = 1;</code>
+     */
+    public Builder clearOperation() {
+      
+      operation_ = getDefaultInstance().getOperation();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * An Operation name for a ConstantOp in this SavedObjectGraph's MetaGraph.
+     * </pre>
+     *
+     * <code>string operation = 1;</code>
+     */
+    public Builder setOperationBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      operation_ = value;
+      onChanged();
+      return this;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.SavedConstant)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.SavedConstant)
+  private static final org.tensorflow.proto.framework.SavedConstant DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.SavedConstant();
+  }
+
+  public static org.tensorflow.proto.framework.SavedConstant getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<SavedConstant>
+      PARSER = new com.google.protobuf.AbstractParser<SavedConstant>() {
+    @java.lang.Override
+    public SavedConstant parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new SavedConstant(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<SavedConstant> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<SavedConstant> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.framework.SavedConstant getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedConstantOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedConstantOrBuilder.java
new file mode 100644
index 00000000000..ed435bc83ba
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedConstantOrBuilder.java
@@ -0,0 +1,27 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/saved_object_graph.proto
+
+package org.tensorflow.proto.framework;
+
+public interface SavedConstantOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.SavedConstant)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <pre>
+   * An Operation name for a ConstantOp in this SavedObjectGraph's MetaGraph.
+   * </pre>
+   *
+   * <code>string operation = 1;</code>
+   */
+  java.lang.String getOperation();
+  /**
+   * <pre>
+   * An Operation name for a ConstantOp in this SavedObjectGraph's MetaGraph.
+   * </pre>
+   *
+   * <code>string operation = 1;</code>
+   */
+  com.google.protobuf.ByteString
+      getOperationBytes();
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedFunction.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedFunction.java
new file mode 100644
index 00000000000..d873063784e
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedFunction.java
@@ -0,0 +1,781 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/saved_object_graph.proto
+
+package org.tensorflow.proto.framework;
+
+/**
+ * <pre>
+ * A function with multiple signatures, possibly with non-Tensor arguments.
+ * </pre>
+ *
+ * Protobuf type {@code tensorflow.SavedFunction}
+ */
+public  final class SavedFunction extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.SavedFunction)
+    SavedFunctionOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use SavedFunction.newBuilder() to construct.
+  private SavedFunction(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private SavedFunction() {
+    concreteFunctions_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new SavedFunction();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private SavedFunction(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    int mutable_bitField0_ = 0;
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 10: {
+            java.lang.String s = input.readStringRequireUtf8();
+            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
+              concreteFunctions_ = new com.google.protobuf.LazyStringArrayList();
+              mutable_bitField0_ |= 0x00000001;
+            }
+            concreteFunctions_.add(s);
+            break;
+          }
+          case 18: {
+            org.tensorflow.proto.framework.FunctionSpec.Builder subBuilder = null;
+            if (functionSpec_ != null) {
+              subBuilder = functionSpec_.toBuilder();
+            }
+            functionSpec_ = input.readMessage(org.tensorflow.proto.framework.FunctionSpec.parser(), extensionRegistry);
+            if (subBuilder != null) {
+              subBuilder.mergeFrom(functionSpec_);
+              functionSpec_ = subBuilder.buildPartial();
+            }
+
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      if (((mutable_bitField0_ & 0x00000001) != 0)) {
+        concreteFunctions_ = concreteFunctions_.getUnmodifiableView();
+      }
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.framework.SavedObjectGraphProtos.internal_static_tensorflow_SavedFunction_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.framework.SavedObjectGraphProtos.internal_static_tensorflow_SavedFunction_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.framework.SavedFunction.class, org.tensorflow.proto.framework.SavedFunction.Builder.class);
+  }
+
+  public static final int CONCRETE_FUNCTIONS_FIELD_NUMBER = 1;
+  private com.google.protobuf.LazyStringList concreteFunctions_;
+  /**
+   * <code>repeated string concrete_functions = 1;</code>
+   */
+  public com.google.protobuf.ProtocolStringList
+      getConcreteFunctionsList() {
+    return concreteFunctions_;
+  }
+  /**
+   * <code>repeated string concrete_functions = 1;</code>
+   */
+  public int getConcreteFunctionsCount() {
+    return concreteFunctions_.size();
+  }
+  /**
+   * <code>repeated string concrete_functions = 1;</code>
+   */
+  public java.lang.String getConcreteFunctions(int index) {
+    return concreteFunctions_.get(index);
+  }
+  /**
+   * <code>repeated string concrete_functions = 1;</code>
+   */
+  public com.google.protobuf.ByteString
+      getConcreteFunctionsBytes(int index) {
+    return concreteFunctions_.getByteString(index);
+  }
+
+  public static final int FUNCTION_SPEC_FIELD_NUMBER = 2;
+  private org.tensorflow.proto.framework.FunctionSpec functionSpec_;
+  /**
+   * <code>.tensorflow.FunctionSpec function_spec = 2;</code>
+   */
+  public boolean hasFunctionSpec() {
+    return functionSpec_ != null;
+  }
+  /**
+   * <code>.tensorflow.FunctionSpec function_spec = 2;</code>
+   */
+  public org.tensorflow.proto.framework.FunctionSpec getFunctionSpec() {
+    return functionSpec_ == null ? org.tensorflow.proto.framework.FunctionSpec.getDefaultInstance() : functionSpec_;
+  }
+  /**
+   * <code>.tensorflow.FunctionSpec function_spec = 2;</code>
+   */
+  public org.tensorflow.proto.framework.FunctionSpecOrBuilder getFunctionSpecOrBuilder() {
+    return getFunctionSpec();
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    for (int i = 0; i < concreteFunctions_.size(); i++) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 1, concreteFunctions_.getRaw(i));
+    }
+    if (functionSpec_ != null) {
+      output.writeMessage(2, getFunctionSpec());
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    {
+      int dataSize = 0;
+      for (int i = 0; i < concreteFunctions_.size(); i++) {
+        dataSize += computeStringSizeNoTag(concreteFunctions_.getRaw(i));
+      }
+      size += dataSize;
+      size += 1 * getConcreteFunctionsList().size();
+    }
+    if (functionSpec_ != null) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(2, getFunctionSpec());
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.framework.SavedFunction)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.framework.SavedFunction other = (org.tensorflow.proto.framework.SavedFunction) obj;
+
+    if (!getConcreteFunctionsList()
+        .equals(other.getConcreteFunctionsList())) return false;
+    if (hasFunctionSpec() != other.hasFunctionSpec()) return false;
+    if (hasFunctionSpec()) {
+      if (!getFunctionSpec()
+          .equals(other.getFunctionSpec())) return false;
+    }
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    if (getConcreteFunctionsCount() > 0) {
+      hash = (37 * hash) + CONCRETE_FUNCTIONS_FIELD_NUMBER;
+      hash = (53 * hash) + getConcreteFunctionsList().hashCode();
+    }
+    if (hasFunctionSpec()) {
+      hash = (37 * hash) + FUNCTION_SPEC_FIELD_NUMBER;
+      hash = (53 * hash) + getFunctionSpec().hashCode();
+    }
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.framework.SavedFunction parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.SavedFunction parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.SavedFunction parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.SavedFunction parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.SavedFunction parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.SavedFunction parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.SavedFunction parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.SavedFunction parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.SavedFunction parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.SavedFunction parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.SavedFunction parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.SavedFunction parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.framework.SavedFunction prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * <pre>
+   * A function with multiple signatures, possibly with non-Tensor arguments.
+   * </pre>
+   *
+   * Protobuf type {@code tensorflow.SavedFunction}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.SavedFunction)
+      org.tensorflow.proto.framework.SavedFunctionOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.SavedObjectGraphProtos.internal_static_tensorflow_SavedFunction_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.SavedObjectGraphProtos.internal_static_tensorflow_SavedFunction_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.SavedFunction.class, org.tensorflow.proto.framework.SavedFunction.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.framework.SavedFunction.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      concreteFunctions_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+      bitField0_ = (bitField0_ & ~0x00000001);
+      if (functionSpecBuilder_ == null) {
+        functionSpec_ = null;
+      } else {
+        functionSpec_ = null;
+        functionSpecBuilder_ = null;
+      }
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.framework.SavedObjectGraphProtos.internal_static_tensorflow_SavedFunction_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.SavedFunction getDefaultInstanceForType() {
+      return org.tensorflow.proto.framework.SavedFunction.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.SavedFunction build() {
+      org.tensorflow.proto.framework.SavedFunction result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.SavedFunction buildPartial() {
+      org.tensorflow.proto.framework.SavedFunction result = new org.tensorflow.proto.framework.SavedFunction(this);
+      int from_bitField0_ = bitField0_;
+      if (((bitField0_ & 0x00000001) != 0)) {
+        concreteFunctions_ = concreteFunctions_.getUnmodifiableView();
+        bitField0_ = (bitField0_ & ~0x00000001);
+      }
+      result.concreteFunctions_ = concreteFunctions_;
+      if (functionSpecBuilder_ == null) {
+        result.functionSpec_ = functionSpec_;
+      } else {
+        result.functionSpec_ = functionSpecBuilder_.build();
+      }
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.framework.SavedFunction) {
+        return mergeFrom((org.tensorflow.proto.framework.SavedFunction)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.framework.SavedFunction other) {
+      if (other == org.tensorflow.proto.framework.SavedFunction.getDefaultInstance()) return this;
+      if (!other.concreteFunctions_.isEmpty()) {
+        if (concreteFunctions_.isEmpty()) {
+          concreteFunctions_ = other.concreteFunctions_;
+          bitField0_ = (bitField0_ & ~0x00000001);
+        } else {
+          ensureConcreteFunctionsIsMutable();
+          concreteFunctions_.addAll(other.concreteFunctions_);
+        }
+        onChanged();
+      }
+      if (other.hasFunctionSpec()) {
+        mergeFunctionSpec(other.getFunctionSpec());
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.framework.SavedFunction parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.framework.SavedFunction) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+    private int bitField0_;
+
+    private com.google.protobuf.LazyStringList concreteFunctions_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+    private void ensureConcreteFunctionsIsMutable() {
+      if (!((bitField0_ & 0x00000001) != 0)) {
+        concreteFunctions_ = new com.google.protobuf.LazyStringArrayList(concreteFunctions_);
+        bitField0_ |= 0x00000001;
+       }
+    }
+    /**
+     * <code>repeated string concrete_functions = 1;</code>
+     */
+    public com.google.protobuf.ProtocolStringList
+        getConcreteFunctionsList() {
+      return concreteFunctions_.getUnmodifiableView();
+    }
+    /**
+     * <code>repeated string concrete_functions = 1;</code>
+     */
+    public int getConcreteFunctionsCount() {
+      return concreteFunctions_.size();
+    }
+    /**
+     * <code>repeated string concrete_functions = 1;</code>
+     */
+    public java.lang.String getConcreteFunctions(int index) {
+      return concreteFunctions_.get(index);
+    }
+    /**
+     * <code>repeated string concrete_functions = 1;</code>
+     */
+    public com.google.protobuf.ByteString
+        getConcreteFunctionsBytes(int index) {
+      return concreteFunctions_.getByteString(index);
+    }
+    /**
+     * <code>repeated string concrete_functions = 1;</code>
+     */
+    public Builder setConcreteFunctions(
+        int index, java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  ensureConcreteFunctionsIsMutable();
+      concreteFunctions_.set(index, value);
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>repeated string concrete_functions = 1;</code>
+     */
+    public Builder addConcreteFunctions(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  ensureConcreteFunctionsIsMutable();
+      concreteFunctions_.add(value);
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>repeated string concrete_functions = 1;</code>
+     */
+    public Builder addAllConcreteFunctions(
+        java.lang.Iterable<java.lang.String> values) {
+      ensureConcreteFunctionsIsMutable();
+      com.google.protobuf.AbstractMessageLite.Builder.addAll(
+          values, concreteFunctions_);
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>repeated string concrete_functions = 1;</code>
+     */
+    public Builder clearConcreteFunctions() {
+      concreteFunctions_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+      bitField0_ = (bitField0_ & ~0x00000001);
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>repeated string concrete_functions = 1;</code>
+     */
+    public Builder addConcreteFunctionsBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      ensureConcreteFunctionsIsMutable();
+      concreteFunctions_.add(value);
+      onChanged();
+      return this;
+    }
+
+    private org.tensorflow.proto.framework.FunctionSpec functionSpec_;
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.FunctionSpec, org.tensorflow.proto.framework.FunctionSpec.Builder, org.tensorflow.proto.framework.FunctionSpecOrBuilder> functionSpecBuilder_;
+    /**
+     * <code>.tensorflow.FunctionSpec function_spec = 2;</code>
+     */
+    public boolean hasFunctionSpec() {
+      return functionSpecBuilder_ != null || functionSpec_ != null;
+    }
+    /**
+     * <code>.tensorflow.FunctionSpec function_spec = 2;</code>
+     */
+    public org.tensorflow.proto.framework.FunctionSpec getFunctionSpec() {
+      if (functionSpecBuilder_ == null) {
+        return functionSpec_ == null ? org.tensorflow.proto.framework.FunctionSpec.getDefaultInstance() : functionSpec_;
+      } else {
+        return functionSpecBuilder_.getMessage();
+      }
+    }
+    /**
+     * <code>.tensorflow.FunctionSpec function_spec = 2;</code>
+     */
+    public Builder setFunctionSpec(org.tensorflow.proto.framework.FunctionSpec value) {
+      if (functionSpecBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        functionSpec_ = value;
+        onChanged();
+      } else {
+        functionSpecBuilder_.setMessage(value);
+      }
+
+      return this;
+    }
+    /**
+     * <code>.tensorflow.FunctionSpec function_spec = 2;</code>
+     */
+    public Builder setFunctionSpec(
+        org.tensorflow.proto.framework.FunctionSpec.Builder builderForValue) {
+      if (functionSpecBuilder_ == null) {
+        functionSpec_ = builderForValue.build();
+        onChanged();
+      } else {
+        functionSpecBuilder_.setMessage(builderForValue.build());
+      }
+
+      return this;
+    }
+    /**
+     * <code>.tensorflow.FunctionSpec function_spec = 2;</code>
+     */
+    public Builder mergeFunctionSpec(org.tensorflow.proto.framework.FunctionSpec value) {
+      if (functionSpecBuilder_ == null) {
+        if (functionSpec_ != null) {
+          functionSpec_ =
+            org.tensorflow.proto.framework.FunctionSpec.newBuilder(functionSpec_).mergeFrom(value).buildPartial();
+        } else {
+          functionSpec_ = value;
+        }
+        onChanged();
+      } else {
+        functionSpecBuilder_.mergeFrom(value);
+      }
+
+      return this;
+    }
+    /**
+     * <code>.tensorflow.FunctionSpec function_spec = 2;</code>
+     */
+    public Builder clearFunctionSpec() {
+      if (functionSpecBuilder_ == null) {
+        functionSpec_ = null;
+        onChanged();
+      } else {
+        functionSpec_ = null;
+        functionSpecBuilder_ = null;
+      }
+
+      return this;
+    }
+    /**
+     * <code>.tensorflow.FunctionSpec function_spec = 2;</code>
+     */
+    public org.tensorflow.proto.framework.FunctionSpec.Builder getFunctionSpecBuilder() {
+      
+      onChanged();
+      return getFunctionSpecFieldBuilder().getBuilder();
+    }
+    /**
+     * <code>.tensorflow.FunctionSpec function_spec = 2;</code>
+     */
+    public org.tensorflow.proto.framework.FunctionSpecOrBuilder getFunctionSpecOrBuilder() {
+      if (functionSpecBuilder_ != null) {
+        return functionSpecBuilder_.getMessageOrBuilder();
+      } else {
+        return functionSpec_ == null ?
+            org.tensorflow.proto.framework.FunctionSpec.getDefaultInstance() : functionSpec_;
+      }
+    }
+    /**
+     * <code>.tensorflow.FunctionSpec function_spec = 2;</code>
+     */
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.FunctionSpec, org.tensorflow.proto.framework.FunctionSpec.Builder, org.tensorflow.proto.framework.FunctionSpecOrBuilder> 
+        getFunctionSpecFieldBuilder() {
+      if (functionSpecBuilder_ == null) {
+        functionSpecBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+            org.tensorflow.proto.framework.FunctionSpec, org.tensorflow.proto.framework.FunctionSpec.Builder, org.tensorflow.proto.framework.FunctionSpecOrBuilder>(
+                getFunctionSpec(),
+                getParentForChildren(),
+                isClean());
+        functionSpec_ = null;
+      }
+      return functionSpecBuilder_;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.SavedFunction)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.SavedFunction)
+  private static final org.tensorflow.proto.framework.SavedFunction DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.SavedFunction();
+  }
+
+  public static org.tensorflow.proto.framework.SavedFunction getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<SavedFunction>
+      PARSER = new com.google.protobuf.AbstractParser<SavedFunction>() {
+    @java.lang.Override
+    public SavedFunction parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new SavedFunction(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<SavedFunction> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<SavedFunction> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.framework.SavedFunction getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedFunctionOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedFunctionOrBuilder.java
new file mode 100644
index 00000000000..19b27a54fb3
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedFunctionOrBuilder.java
@@ -0,0 +1,41 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/saved_object_graph.proto
+
+package org.tensorflow.proto.framework;
+
+public interface SavedFunctionOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.SavedFunction)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <code>repeated string concrete_functions = 1;</code>
+   */
+  java.util.List<java.lang.String>
+      getConcreteFunctionsList();
+  /**
+   * <code>repeated string concrete_functions = 1;</code>
+   */
+  int getConcreteFunctionsCount();
+  /**
+   * <code>repeated string concrete_functions = 1;</code>
+   */
+  java.lang.String getConcreteFunctions(int index);
+  /**
+   * <code>repeated string concrete_functions = 1;</code>
+   */
+  com.google.protobuf.ByteString
+      getConcreteFunctionsBytes(int index);
+
+  /**
+   * <code>.tensorflow.FunctionSpec function_spec = 2;</code>
+   */
+  boolean hasFunctionSpec();
+  /**
+   * <code>.tensorflow.FunctionSpec function_spec = 2;</code>
+   */
+  org.tensorflow.proto.framework.FunctionSpec getFunctionSpec();
+  /**
+   * <code>.tensorflow.FunctionSpec function_spec = 2;</code>
+   */
+  org.tensorflow.proto.framework.FunctionSpecOrBuilder getFunctionSpecOrBuilder();
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedModel.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedModel.java
new file mode 100644
index 00000000000..ac25416e31a
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedModel.java
@@ -0,0 +1,949 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/saved_model.proto
+
+package org.tensorflow.proto.framework;
+
+/**
+ * <pre>
+ * SavedModel is the high level serialization format for TensorFlow Models.
+ * See [todo: doc links, similar to session_bundle] for more information.
+ * </pre>
+ *
+ * Protobuf type {@code tensorflow.SavedModel}
+ */
+public  final class SavedModel extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.SavedModel)
+    SavedModelOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use SavedModel.newBuilder() to construct.
+  private SavedModel(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private SavedModel() {
+    metaGraphs_ = java.util.Collections.emptyList();
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new SavedModel();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private SavedModel(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    int mutable_bitField0_ = 0;
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 8: {
+
+            savedModelSchemaVersion_ = input.readInt64();
+            break;
+          }
+          case 18: {
+            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
+              metaGraphs_ = new java.util.ArrayList<org.tensorflow.proto.framework.MetaGraphDef>();
+              mutable_bitField0_ |= 0x00000001;
+            }
+            metaGraphs_.add(
+                input.readMessage(org.tensorflow.proto.framework.MetaGraphDef.parser(), extensionRegistry));
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      if (((mutable_bitField0_ & 0x00000001) != 0)) {
+        metaGraphs_ = java.util.Collections.unmodifiableList(metaGraphs_);
+      }
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.framework.SavedModelProtos.internal_static_tensorflow_SavedModel_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.framework.SavedModelProtos.internal_static_tensorflow_SavedModel_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.framework.SavedModel.class, org.tensorflow.proto.framework.SavedModel.Builder.class);
+  }
+
+  public static final int SAVED_MODEL_SCHEMA_VERSION_FIELD_NUMBER = 1;
+  private long savedModelSchemaVersion_;
+  /**
+   * <pre>
+   * The schema version of the SavedModel instance. Used for versioning when
+   * making future changes to the specification/implementation. Initial value
+   * at release will be 1.
+   * </pre>
+   *
+   * <code>int64 saved_model_schema_version = 1;</code>
+   */
+  public long getSavedModelSchemaVersion() {
+    return savedModelSchemaVersion_;
+  }
+
+  public static final int META_GRAPHS_FIELD_NUMBER = 2;
+  private java.util.List<org.tensorflow.proto.framework.MetaGraphDef> metaGraphs_;
+  /**
+   * <pre>
+   * One or more MetaGraphs.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.MetaGraphDef meta_graphs = 2;</code>
+   */
+  public java.util.List<org.tensorflow.proto.framework.MetaGraphDef> getMetaGraphsList() {
+    return metaGraphs_;
+  }
+  /**
+   * <pre>
+   * One or more MetaGraphs.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.MetaGraphDef meta_graphs = 2;</code>
+   */
+  public java.util.List<? extends org.tensorflow.proto.framework.MetaGraphDefOrBuilder> 
+      getMetaGraphsOrBuilderList() {
+    return metaGraphs_;
+  }
+  /**
+   * <pre>
+   * One or more MetaGraphs.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.MetaGraphDef meta_graphs = 2;</code>
+   */
+  public int getMetaGraphsCount() {
+    return metaGraphs_.size();
+  }
+  /**
+   * <pre>
+   * One or more MetaGraphs.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.MetaGraphDef meta_graphs = 2;</code>
+   */
+  public org.tensorflow.proto.framework.MetaGraphDef getMetaGraphs(int index) {
+    return metaGraphs_.get(index);
+  }
+  /**
+   * <pre>
+   * One or more MetaGraphs.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.MetaGraphDef meta_graphs = 2;</code>
+   */
+  public org.tensorflow.proto.framework.MetaGraphDefOrBuilder getMetaGraphsOrBuilder(
+      int index) {
+    return metaGraphs_.get(index);
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    if (savedModelSchemaVersion_ != 0L) {
+      output.writeInt64(1, savedModelSchemaVersion_);
+    }
+    for (int i = 0; i < metaGraphs_.size(); i++) {
+      output.writeMessage(2, metaGraphs_.get(i));
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (savedModelSchemaVersion_ != 0L) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt64Size(1, savedModelSchemaVersion_);
+    }
+    for (int i = 0; i < metaGraphs_.size(); i++) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(2, metaGraphs_.get(i));
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.framework.SavedModel)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.framework.SavedModel other = (org.tensorflow.proto.framework.SavedModel) obj;
+
+    if (getSavedModelSchemaVersion()
+        != other.getSavedModelSchemaVersion()) return false;
+    if (!getMetaGraphsList()
+        .equals(other.getMetaGraphsList())) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    hash = (37 * hash) + SAVED_MODEL_SCHEMA_VERSION_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+        getSavedModelSchemaVersion());
+    if (getMetaGraphsCount() > 0) {
+      hash = (37 * hash) + META_GRAPHS_FIELD_NUMBER;
+      hash = (53 * hash) + getMetaGraphsList().hashCode();
+    }
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.framework.SavedModel parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.SavedModel parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.SavedModel parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.SavedModel parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.SavedModel parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.SavedModel parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.SavedModel parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.SavedModel parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.SavedModel parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.SavedModel parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.SavedModel parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.SavedModel parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.framework.SavedModel prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * <pre>
+   * SavedModel is the high level serialization format for TensorFlow Models.
+   * See [todo: doc links, similar to session_bundle] for more information.
+   * </pre>
+   *
+   * Protobuf type {@code tensorflow.SavedModel}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.SavedModel)
+      org.tensorflow.proto.framework.SavedModelOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.SavedModelProtos.internal_static_tensorflow_SavedModel_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.SavedModelProtos.internal_static_tensorflow_SavedModel_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.SavedModel.class, org.tensorflow.proto.framework.SavedModel.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.framework.SavedModel.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+        getMetaGraphsFieldBuilder();
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      savedModelSchemaVersion_ = 0L;
+
+      if (metaGraphsBuilder_ == null) {
+        metaGraphs_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000001);
+      } else {
+        metaGraphsBuilder_.clear();
+      }
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.framework.SavedModelProtos.internal_static_tensorflow_SavedModel_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.SavedModel getDefaultInstanceForType() {
+      return org.tensorflow.proto.framework.SavedModel.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.SavedModel build() {
+      org.tensorflow.proto.framework.SavedModel result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.SavedModel buildPartial() {
+      org.tensorflow.proto.framework.SavedModel result = new org.tensorflow.proto.framework.SavedModel(this);
+      int from_bitField0_ = bitField0_;
+      result.savedModelSchemaVersion_ = savedModelSchemaVersion_;
+      if (metaGraphsBuilder_ == null) {
+        if (((bitField0_ & 0x00000001) != 0)) {
+          metaGraphs_ = java.util.Collections.unmodifiableList(metaGraphs_);
+          bitField0_ = (bitField0_ & ~0x00000001);
+        }
+        result.metaGraphs_ = metaGraphs_;
+      } else {
+        result.metaGraphs_ = metaGraphsBuilder_.build();
+      }
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.framework.SavedModel) {
+        return mergeFrom((org.tensorflow.proto.framework.SavedModel)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.framework.SavedModel other) {
+      if (other == org.tensorflow.proto.framework.SavedModel.getDefaultInstance()) return this;
+      if (other.getSavedModelSchemaVersion() != 0L) {
+        setSavedModelSchemaVersion(other.getSavedModelSchemaVersion());
+      }
+      if (metaGraphsBuilder_ == null) {
+        if (!other.metaGraphs_.isEmpty()) {
+          if (metaGraphs_.isEmpty()) {
+            metaGraphs_ = other.metaGraphs_;
+            bitField0_ = (bitField0_ & ~0x00000001);
+          } else {
+            ensureMetaGraphsIsMutable();
+            metaGraphs_.addAll(other.metaGraphs_);
+          }
+          onChanged();
+        }
+      } else {
+        if (!other.metaGraphs_.isEmpty()) {
+          if (metaGraphsBuilder_.isEmpty()) {
+            metaGraphsBuilder_.dispose();
+            metaGraphsBuilder_ = null;
+            metaGraphs_ = other.metaGraphs_;
+            bitField0_ = (bitField0_ & ~0x00000001);
+            metaGraphsBuilder_ = 
+              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
+                 getMetaGraphsFieldBuilder() : null;
+          } else {
+            metaGraphsBuilder_.addAllMessages(other.metaGraphs_);
+          }
+        }
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.framework.SavedModel parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.framework.SavedModel) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+    private int bitField0_;
+
+    private long savedModelSchemaVersion_ ;
+    /**
+     * <pre>
+     * The schema version of the SavedModel instance. Used for versioning when
+     * making future changes to the specification/implementation. Initial value
+     * at release will be 1.
+     * </pre>
+     *
+     * <code>int64 saved_model_schema_version = 1;</code>
+     */
+    public long getSavedModelSchemaVersion() {
+      return savedModelSchemaVersion_;
+    }
+    /**
+     * <pre>
+     * The schema version of the SavedModel instance. Used for versioning when
+     * making future changes to the specification/implementation. Initial value
+     * at release will be 1.
+     * </pre>
+     *
+     * <code>int64 saved_model_schema_version = 1;</code>
+     */
+    public Builder setSavedModelSchemaVersion(long value) {
+      
+      savedModelSchemaVersion_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * The schema version of the SavedModel instance. Used for versioning when
+     * making future changes to the specification/implementation. Initial value
+     * at release will be 1.
+     * </pre>
+     *
+     * <code>int64 saved_model_schema_version = 1;</code>
+     */
+    public Builder clearSavedModelSchemaVersion() {
+      
+      savedModelSchemaVersion_ = 0L;
+      onChanged();
+      return this;
+    }
+
+    private java.util.List<org.tensorflow.proto.framework.MetaGraphDef> metaGraphs_ =
+      java.util.Collections.emptyList();
+    private void ensureMetaGraphsIsMutable() {
+      if (!((bitField0_ & 0x00000001) != 0)) {
+        metaGraphs_ = new java.util.ArrayList<org.tensorflow.proto.framework.MetaGraphDef>(metaGraphs_);
+        bitField0_ |= 0x00000001;
+       }
+    }
+
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.framework.MetaGraphDef, org.tensorflow.proto.framework.MetaGraphDef.Builder, org.tensorflow.proto.framework.MetaGraphDefOrBuilder> metaGraphsBuilder_;
+
+    /**
+     * <pre>
+     * One or more MetaGraphs.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.MetaGraphDef meta_graphs = 2;</code>
+     */
+    public java.util.List<org.tensorflow.proto.framework.MetaGraphDef> getMetaGraphsList() {
+      if (metaGraphsBuilder_ == null) {
+        return java.util.Collections.unmodifiableList(metaGraphs_);
+      } else {
+        return metaGraphsBuilder_.getMessageList();
+      }
+    }
+    /**
+     * <pre>
+     * One or more MetaGraphs.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.MetaGraphDef meta_graphs = 2;</code>
+     */
+    public int getMetaGraphsCount() {
+      if (metaGraphsBuilder_ == null) {
+        return metaGraphs_.size();
+      } else {
+        return metaGraphsBuilder_.getCount();
+      }
+    }
+    /**
+     * <pre>
+     * One or more MetaGraphs.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.MetaGraphDef meta_graphs = 2;</code>
+     */
+    public org.tensorflow.proto.framework.MetaGraphDef getMetaGraphs(int index) {
+      if (metaGraphsBuilder_ == null) {
+        return metaGraphs_.get(index);
+      } else {
+        return metaGraphsBuilder_.getMessage(index);
+      }
+    }
+    /**
+     * <pre>
+     * One or more MetaGraphs.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.MetaGraphDef meta_graphs = 2;</code>
+     */
+    public Builder setMetaGraphs(
+        int index, org.tensorflow.proto.framework.MetaGraphDef value) {
+      if (metaGraphsBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureMetaGraphsIsMutable();
+        metaGraphs_.set(index, value);
+        onChanged();
+      } else {
+        metaGraphsBuilder_.setMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * One or more MetaGraphs.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.MetaGraphDef meta_graphs = 2;</code>
+     */
+    public Builder setMetaGraphs(
+        int index, org.tensorflow.proto.framework.MetaGraphDef.Builder builderForValue) {
+      if (metaGraphsBuilder_ == null) {
+        ensureMetaGraphsIsMutable();
+        metaGraphs_.set(index, builderForValue.build());
+        onChanged();
+      } else {
+        metaGraphsBuilder_.setMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * One or more MetaGraphs.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.MetaGraphDef meta_graphs = 2;</code>
+     */
+    public Builder addMetaGraphs(org.tensorflow.proto.framework.MetaGraphDef value) {
+      if (metaGraphsBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureMetaGraphsIsMutable();
+        metaGraphs_.add(value);
+        onChanged();
+      } else {
+        metaGraphsBuilder_.addMessage(value);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * One or more MetaGraphs.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.MetaGraphDef meta_graphs = 2;</code>
+     */
+    public Builder addMetaGraphs(
+        int index, org.tensorflow.proto.framework.MetaGraphDef value) {
+      if (metaGraphsBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureMetaGraphsIsMutable();
+        metaGraphs_.add(index, value);
+        onChanged();
+      } else {
+        metaGraphsBuilder_.addMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * One or more MetaGraphs.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.MetaGraphDef meta_graphs = 2;</code>
+     */
+    public Builder addMetaGraphs(
+        org.tensorflow.proto.framework.MetaGraphDef.Builder builderForValue) {
+      if (metaGraphsBuilder_ == null) {
+        ensureMetaGraphsIsMutable();
+        metaGraphs_.add(builderForValue.build());
+        onChanged();
+      } else {
+        metaGraphsBuilder_.addMessage(builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * One or more MetaGraphs.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.MetaGraphDef meta_graphs = 2;</code>
+     */
+    public Builder addMetaGraphs(
+        int index, org.tensorflow.proto.framework.MetaGraphDef.Builder builderForValue) {
+      if (metaGraphsBuilder_ == null) {
+        ensureMetaGraphsIsMutable();
+        metaGraphs_.add(index, builderForValue.build());
+        onChanged();
+      } else {
+        metaGraphsBuilder_.addMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * One or more MetaGraphs.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.MetaGraphDef meta_graphs = 2;</code>
+     */
+    public Builder addAllMetaGraphs(
+        java.lang.Iterable<? extends org.tensorflow.proto.framework.MetaGraphDef> values) {
+      if (metaGraphsBuilder_ == null) {
+        ensureMetaGraphsIsMutable();
+        com.google.protobuf.AbstractMessageLite.Builder.addAll(
+            values, metaGraphs_);
+        onChanged();
+      } else {
+        metaGraphsBuilder_.addAllMessages(values);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * One or more MetaGraphs.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.MetaGraphDef meta_graphs = 2;</code>
+     */
+    public Builder clearMetaGraphs() {
+      if (metaGraphsBuilder_ == null) {
+        metaGraphs_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000001);
+        onChanged();
+      } else {
+        metaGraphsBuilder_.clear();
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * One or more MetaGraphs.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.MetaGraphDef meta_graphs = 2;</code>
+     */
+    public Builder removeMetaGraphs(int index) {
+      if (metaGraphsBuilder_ == null) {
+        ensureMetaGraphsIsMutable();
+        metaGraphs_.remove(index);
+        onChanged();
+      } else {
+        metaGraphsBuilder_.remove(index);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * One or more MetaGraphs.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.MetaGraphDef meta_graphs = 2;</code>
+     */
+    public org.tensorflow.proto.framework.MetaGraphDef.Builder getMetaGraphsBuilder(
+        int index) {
+      return getMetaGraphsFieldBuilder().getBuilder(index);
+    }
+    /**
+     * <pre>
+     * One or more MetaGraphs.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.MetaGraphDef meta_graphs = 2;</code>
+     */
+    public org.tensorflow.proto.framework.MetaGraphDefOrBuilder getMetaGraphsOrBuilder(
+        int index) {
+      if (metaGraphsBuilder_ == null) {
+        return metaGraphs_.get(index);  } else {
+        return metaGraphsBuilder_.getMessageOrBuilder(index);
+      }
+    }
+    /**
+     * <pre>
+     * One or more MetaGraphs.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.MetaGraphDef meta_graphs = 2;</code>
+     */
+    public java.util.List<? extends org.tensorflow.proto.framework.MetaGraphDefOrBuilder> 
+         getMetaGraphsOrBuilderList() {
+      if (metaGraphsBuilder_ != null) {
+        return metaGraphsBuilder_.getMessageOrBuilderList();
+      } else {
+        return java.util.Collections.unmodifiableList(metaGraphs_);
+      }
+    }
+    /**
+     * <pre>
+     * One or more MetaGraphs.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.MetaGraphDef meta_graphs = 2;</code>
+     */
+    public org.tensorflow.proto.framework.MetaGraphDef.Builder addMetaGraphsBuilder() {
+      return getMetaGraphsFieldBuilder().addBuilder(
+          org.tensorflow.proto.framework.MetaGraphDef.getDefaultInstance());
+    }
+    /**
+     * <pre>
+     * One or more MetaGraphs.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.MetaGraphDef meta_graphs = 2;</code>
+     */
+    public org.tensorflow.proto.framework.MetaGraphDef.Builder addMetaGraphsBuilder(
+        int index) {
+      return getMetaGraphsFieldBuilder().addBuilder(
+          index, org.tensorflow.proto.framework.MetaGraphDef.getDefaultInstance());
+    }
+    /**
+     * <pre>
+     * One or more MetaGraphs.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.MetaGraphDef meta_graphs = 2;</code>
+     */
+    public java.util.List<org.tensorflow.proto.framework.MetaGraphDef.Builder> 
+         getMetaGraphsBuilderList() {
+      return getMetaGraphsFieldBuilder().getBuilderList();
+    }
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.framework.MetaGraphDef, org.tensorflow.proto.framework.MetaGraphDef.Builder, org.tensorflow.proto.framework.MetaGraphDefOrBuilder> 
+        getMetaGraphsFieldBuilder() {
+      if (metaGraphsBuilder_ == null) {
+        metaGraphsBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
+            org.tensorflow.proto.framework.MetaGraphDef, org.tensorflow.proto.framework.MetaGraphDef.Builder, org.tensorflow.proto.framework.MetaGraphDefOrBuilder>(
+                metaGraphs_,
+                ((bitField0_ & 0x00000001) != 0),
+                getParentForChildren(),
+                isClean());
+        metaGraphs_ = null;
+      }
+      return metaGraphsBuilder_;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.SavedModel)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.SavedModel)
+  private static final org.tensorflow.proto.framework.SavedModel DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.SavedModel();
+  }
+
+  public static org.tensorflow.proto.framework.SavedModel getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<SavedModel>
+      PARSER = new com.google.protobuf.AbstractParser<SavedModel>() {
+    @java.lang.Override
+    public SavedModel parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new SavedModel(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<SavedModel> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<SavedModel> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.framework.SavedModel getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedModelOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedModelOrBuilder.java
new file mode 100644
index 00000000000..9714aee34dd
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedModelOrBuilder.java
@@ -0,0 +1,64 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/saved_model.proto
+
+package org.tensorflow.proto.framework;
+
+public interface SavedModelOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.SavedModel)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <pre>
+   * The schema version of the SavedModel instance. Used for versioning when
+   * making future changes to the specification/implementation. Initial value
+   * at release will be 1.
+   * </pre>
+   *
+   * <code>int64 saved_model_schema_version = 1;</code>
+   */
+  long getSavedModelSchemaVersion();
+
+  /**
+   * <pre>
+   * One or more MetaGraphs.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.MetaGraphDef meta_graphs = 2;</code>
+   */
+  java.util.List<org.tensorflow.proto.framework.MetaGraphDef> 
+      getMetaGraphsList();
+  /**
+   * <pre>
+   * One or more MetaGraphs.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.MetaGraphDef meta_graphs = 2;</code>
+   */
+  org.tensorflow.proto.framework.MetaGraphDef getMetaGraphs(int index);
+  /**
+   * <pre>
+   * One or more MetaGraphs.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.MetaGraphDef meta_graphs = 2;</code>
+   */
+  int getMetaGraphsCount();
+  /**
+   * <pre>
+   * One or more MetaGraphs.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.MetaGraphDef meta_graphs = 2;</code>
+   */
+  java.util.List<? extends org.tensorflow.proto.framework.MetaGraphDefOrBuilder> 
+      getMetaGraphsOrBuilderList();
+  /**
+   * <pre>
+   * One or more MetaGraphs.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.MetaGraphDef meta_graphs = 2;</code>
+   */
+  org.tensorflow.proto.framework.MetaGraphDefOrBuilder getMetaGraphsOrBuilder(
+      int index);
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedModelProtos.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedModelProtos.java
new file mode 100644
index 00000000000..05b4edd36f5
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedModelProtos.java
@@ -0,0 +1,56 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/saved_model.proto
+
+package org.tensorflow.proto.framework;
+
+public final class SavedModelProtos {
+  private SavedModelProtos() {}
+  public static void registerAllExtensions(
+      com.google.protobuf.ExtensionRegistryLite registry) {
+  }
+
+  public static void registerAllExtensions(
+      com.google.protobuf.ExtensionRegistry registry) {
+    registerAllExtensions(
+        (com.google.protobuf.ExtensionRegistryLite) registry);
+  }
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_SavedModel_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_SavedModel_fieldAccessorTable;
+
+  public static com.google.protobuf.Descriptors.FileDescriptor
+      getDescriptor() {
+    return descriptor;
+  }
+  private static  com.google.protobuf.Descriptors.FileDescriptor
+      descriptor;
+  static {
+    java.lang.String[] descriptorData = {
+      "\n*tensorflow/core/protobuf/saved_model.p" +
+      "roto\022\ntensorflow\032)tensorflow/core/protob" +
+      "uf/meta_graph.proto\"_\n\nSavedModel\022\"\n\032sav" +
+      "ed_model_schema_version\030\001 \001(\003\022-\n\013meta_gr" +
+      "aphs\030\002 \003(\0132\030.tensorflow.MetaGraphDefB\201\001\n" +
+      "\036org.tensorflow.proto.frameworkB\020SavedMo" +
+      "delProtosP\001ZHgithub.com/tensorflow/tenso" +
+      "rflow/tensorflow/go/core/core_protos_go_" +
+      "proto\370\001\001b\006proto3"
+    };
+    descriptor = com.google.protobuf.Descriptors.FileDescriptor
+      .internalBuildGeneratedFileFrom(descriptorData,
+        new com.google.protobuf.Descriptors.FileDescriptor[] {
+          org.tensorflow.proto.framework.MetaGraphProtos.getDescriptor(),
+        });
+    internal_static_tensorflow_SavedModel_descriptor =
+      getDescriptor().getMessageTypes().get(0);
+    internal_static_tensorflow_SavedModel_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_SavedModel_descriptor,
+        new java.lang.String[] { "SavedModelSchemaVersion", "MetaGraphs", });
+    org.tensorflow.proto.framework.MetaGraphProtos.getDescriptor();
+  }
+
+  // @@protoc_insertion_point(outer_class_scope)
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedObject.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedObject.java
new file mode 100644
index 00000000000..d38d76b85a1
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedObject.java
@@ -0,0 +1,3174 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/saved_object_graph.proto
+
+package org.tensorflow.proto.framework;
+
+/**
+ * Protobuf type {@code tensorflow.SavedObject}
+ */
+public  final class SavedObject extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.SavedObject)
+    SavedObjectOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use SavedObject.newBuilder() to construct.
+  private SavedObject(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private SavedObject() {
+    children_ = java.util.Collections.emptyList();
+    slotVariables_ = java.util.Collections.emptyList();
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new SavedObject();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private SavedObject(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    int mutable_bitField0_ = 0;
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 10: {
+            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
+              children_ = new java.util.ArrayList<org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference>();
+              mutable_bitField0_ |= 0x00000001;
+            }
+            children_.add(
+                input.readMessage(org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference.parser(), extensionRegistry));
+            break;
+          }
+          case 26: {
+            if (!((mutable_bitField0_ & 0x00000002) != 0)) {
+              slotVariables_ = new java.util.ArrayList<org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference>();
+              mutable_bitField0_ |= 0x00000002;
+            }
+            slotVariables_.add(
+                input.readMessage(org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference.parser(), extensionRegistry));
+            break;
+          }
+          case 34: {
+            org.tensorflow.proto.framework.SavedUserObject.Builder subBuilder = null;
+            if (kindCase_ == 4) {
+              subBuilder = ((org.tensorflow.proto.framework.SavedUserObject) kind_).toBuilder();
+            }
+            kind_ =
+                input.readMessage(org.tensorflow.proto.framework.SavedUserObject.parser(), extensionRegistry);
+            if (subBuilder != null) {
+              subBuilder.mergeFrom((org.tensorflow.proto.framework.SavedUserObject) kind_);
+              kind_ = subBuilder.buildPartial();
+            }
+            kindCase_ = 4;
+            break;
+          }
+          case 42: {
+            org.tensorflow.proto.framework.SavedAsset.Builder subBuilder = null;
+            if (kindCase_ == 5) {
+              subBuilder = ((org.tensorflow.proto.framework.SavedAsset) kind_).toBuilder();
+            }
+            kind_ =
+                input.readMessage(org.tensorflow.proto.framework.SavedAsset.parser(), extensionRegistry);
+            if (subBuilder != null) {
+              subBuilder.mergeFrom((org.tensorflow.proto.framework.SavedAsset) kind_);
+              kind_ = subBuilder.buildPartial();
+            }
+            kindCase_ = 5;
+            break;
+          }
+          case 50: {
+            org.tensorflow.proto.framework.SavedFunction.Builder subBuilder = null;
+            if (kindCase_ == 6) {
+              subBuilder = ((org.tensorflow.proto.framework.SavedFunction) kind_).toBuilder();
+            }
+            kind_ =
+                input.readMessage(org.tensorflow.proto.framework.SavedFunction.parser(), extensionRegistry);
+            if (subBuilder != null) {
+              subBuilder.mergeFrom((org.tensorflow.proto.framework.SavedFunction) kind_);
+              kind_ = subBuilder.buildPartial();
+            }
+            kindCase_ = 6;
+            break;
+          }
+          case 58: {
+            org.tensorflow.proto.framework.SavedVariable.Builder subBuilder = null;
+            if (kindCase_ == 7) {
+              subBuilder = ((org.tensorflow.proto.framework.SavedVariable) kind_).toBuilder();
+            }
+            kind_ =
+                input.readMessage(org.tensorflow.proto.framework.SavedVariable.parser(), extensionRegistry);
+            if (subBuilder != null) {
+              subBuilder.mergeFrom((org.tensorflow.proto.framework.SavedVariable) kind_);
+              kind_ = subBuilder.buildPartial();
+            }
+            kindCase_ = 7;
+            break;
+          }
+          case 66: {
+            org.tensorflow.proto.framework.SavedBareConcreteFunction.Builder subBuilder = null;
+            if (kindCase_ == 8) {
+              subBuilder = ((org.tensorflow.proto.framework.SavedBareConcreteFunction) kind_).toBuilder();
+            }
+            kind_ =
+                input.readMessage(org.tensorflow.proto.framework.SavedBareConcreteFunction.parser(), extensionRegistry);
+            if (subBuilder != null) {
+              subBuilder.mergeFrom((org.tensorflow.proto.framework.SavedBareConcreteFunction) kind_);
+              kind_ = subBuilder.buildPartial();
+            }
+            kindCase_ = 8;
+            break;
+          }
+          case 74: {
+            org.tensorflow.proto.framework.SavedConstant.Builder subBuilder = null;
+            if (kindCase_ == 9) {
+              subBuilder = ((org.tensorflow.proto.framework.SavedConstant) kind_).toBuilder();
+            }
+            kind_ =
+                input.readMessage(org.tensorflow.proto.framework.SavedConstant.parser(), extensionRegistry);
+            if (subBuilder != null) {
+              subBuilder.mergeFrom((org.tensorflow.proto.framework.SavedConstant) kind_);
+              kind_ = subBuilder.buildPartial();
+            }
+            kindCase_ = 9;
+            break;
+          }
+          case 82: {
+            org.tensorflow.proto.framework.SavedResource.Builder subBuilder = null;
+            if (kindCase_ == 10) {
+              subBuilder = ((org.tensorflow.proto.framework.SavedResource) kind_).toBuilder();
+            }
+            kind_ =
+                input.readMessage(org.tensorflow.proto.framework.SavedResource.parser(), extensionRegistry);
+            if (subBuilder != null) {
+              subBuilder.mergeFrom((org.tensorflow.proto.framework.SavedResource) kind_);
+              kind_ = subBuilder.buildPartial();
+            }
+            kindCase_ = 10;
+            break;
+          }
+          case 90: {
+            if (!((mutable_bitField0_ & 0x00000004) != 0)) {
+              saveableObjects_ = com.google.protobuf.MapField.newMapField(
+                  SaveableObjectsDefaultEntryHolder.defaultEntry);
+              mutable_bitField0_ |= 0x00000004;
+            }
+            com.google.protobuf.MapEntry<java.lang.String, org.tensorflow.proto.framework.SaveableObject>
+            saveableObjects__ = input.readMessage(
+                SaveableObjectsDefaultEntryHolder.defaultEntry.getParserForType(), extensionRegistry);
+            saveableObjects_.getMutableMap().put(
+                saveableObjects__.getKey(), saveableObjects__.getValue());
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      if (((mutable_bitField0_ & 0x00000001) != 0)) {
+        children_ = java.util.Collections.unmodifiableList(children_);
+      }
+      if (((mutable_bitField0_ & 0x00000002) != 0)) {
+        slotVariables_ = java.util.Collections.unmodifiableList(slotVariables_);
+      }
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.framework.SavedObjectGraphProtos.internal_static_tensorflow_SavedObject_descriptor;
+  }
+
+  @SuppressWarnings({"rawtypes"})
+  @java.lang.Override
+  protected com.google.protobuf.MapField internalGetMapField(
+      int number) {
+    switch (number) {
+      case 11:
+        return internalGetSaveableObjects();
+      default:
+        throw new RuntimeException(
+            "Invalid map field number: " + number);
+    }
+  }
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.framework.SavedObjectGraphProtos.internal_static_tensorflow_SavedObject_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.framework.SavedObject.class, org.tensorflow.proto.framework.SavedObject.Builder.class);
+  }
+
+  private int kindCase_ = 0;
+  private java.lang.Object kind_;
+  public enum KindCase
+      implements com.google.protobuf.Internal.EnumLite {
+    USER_OBJECT(4),
+    ASSET(5),
+    FUNCTION(6),
+    VARIABLE(7),
+    BARE_CONCRETE_FUNCTION(8),
+    CONSTANT(9),
+    RESOURCE(10),
+    KIND_NOT_SET(0);
+    private final int value;
+    private KindCase(int value) {
+      this.value = value;
+    }
+    /**
+     * @deprecated Use {@link #forNumber(int)} instead.
+     */
+    @java.lang.Deprecated
+    public static KindCase valueOf(int value) {
+      return forNumber(value);
+    }
+
+    public static KindCase forNumber(int value) {
+      switch (value) {
+        case 4: return USER_OBJECT;
+        case 5: return ASSET;
+        case 6: return FUNCTION;
+        case 7: return VARIABLE;
+        case 8: return BARE_CONCRETE_FUNCTION;
+        case 9: return CONSTANT;
+        case 10: return RESOURCE;
+        case 0: return KIND_NOT_SET;
+        default: return null;
+      }
+    }
+    public int getNumber() {
+      return this.value;
+    }
+  };
+
+  public KindCase
+  getKindCase() {
+    return KindCase.forNumber(
+        kindCase_);
+  }
+
+  public static final int CHILDREN_FIELD_NUMBER = 1;
+  private java.util.List<org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference> children_;
+  /**
+   * <pre>
+   * Objects which this object depends on: named edges in the dependency
+   * graph.
+   * Note: currently only valid if kind == "user_object".
+   * </pre>
+   *
+   * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.ObjectReference children = 1;</code>
+   */
+  public java.util.List<org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference> getChildrenList() {
+    return children_;
+  }
+  /**
+   * <pre>
+   * Objects which this object depends on: named edges in the dependency
+   * graph.
+   * Note: currently only valid if kind == "user_object".
+   * </pre>
+   *
+   * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.ObjectReference children = 1;</code>
+   */
+  public java.util.List<? extends org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReferenceOrBuilder> 
+      getChildrenOrBuilderList() {
+    return children_;
+  }
+  /**
+   * <pre>
+   * Objects which this object depends on: named edges in the dependency
+   * graph.
+   * Note: currently only valid if kind == "user_object".
+   * </pre>
+   *
+   * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.ObjectReference children = 1;</code>
+   */
+  public int getChildrenCount() {
+    return children_.size();
+  }
+  /**
+   * <pre>
+   * Objects which this object depends on: named edges in the dependency
+   * graph.
+   * Note: currently only valid if kind == "user_object".
+   * </pre>
+   *
+   * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.ObjectReference children = 1;</code>
+   */
+  public org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference getChildren(int index) {
+    return children_.get(index);
+  }
+  /**
+   * <pre>
+   * Objects which this object depends on: named edges in the dependency
+   * graph.
+   * Note: currently only valid if kind == "user_object".
+   * </pre>
+   *
+   * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.ObjectReference children = 1;</code>
+   */
+  public org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReferenceOrBuilder getChildrenOrBuilder(
+      int index) {
+    return children_.get(index);
+  }
+
+  public static final int SLOT_VARIABLES_FIELD_NUMBER = 3;
+  private java.util.List<org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference> slotVariables_;
+  /**
+   * <pre>
+   * Slot variables owned by this object. This describes the three-way
+   * (optimizer, variable, slot variable) relationship; none of the three
+   * depend on the others directly.
+   * Note: currently only valid if kind == "user_object".
+   * </pre>
+   *
+   * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SlotVariableReference slot_variables = 3;</code>
+   */
+  public java.util.List<org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference> getSlotVariablesList() {
+    return slotVariables_;
+  }
+  /**
+   * <pre>
+   * Slot variables owned by this object. This describes the three-way
+   * (optimizer, variable, slot variable) relationship; none of the three
+   * depend on the others directly.
+   * Note: currently only valid if kind == "user_object".
+   * </pre>
+   *
+   * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SlotVariableReference slot_variables = 3;</code>
+   */
+  public java.util.List<? extends org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReferenceOrBuilder> 
+      getSlotVariablesOrBuilderList() {
+    return slotVariables_;
+  }
+  /**
+   * <pre>
+   * Slot variables owned by this object. This describes the three-way
+   * (optimizer, variable, slot variable) relationship; none of the three
+   * depend on the others directly.
+   * Note: currently only valid if kind == "user_object".
+   * </pre>
+   *
+   * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SlotVariableReference slot_variables = 3;</code>
+   */
+  public int getSlotVariablesCount() {
+    return slotVariables_.size();
+  }
+  /**
+   * <pre>
+   * Slot variables owned by this object. This describes the three-way
+   * (optimizer, variable, slot variable) relationship; none of the three
+   * depend on the others directly.
+   * Note: currently only valid if kind == "user_object".
+   * </pre>
+   *
+   * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SlotVariableReference slot_variables = 3;</code>
+   */
+  public org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference getSlotVariables(int index) {
+    return slotVariables_.get(index);
+  }
+  /**
+   * <pre>
+   * Slot variables owned by this object. This describes the three-way
+   * (optimizer, variable, slot variable) relationship; none of the three
+   * depend on the others directly.
+   * Note: currently only valid if kind == "user_object".
+   * </pre>
+   *
+   * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SlotVariableReference slot_variables = 3;</code>
+   */
+  public org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReferenceOrBuilder getSlotVariablesOrBuilder(
+      int index) {
+    return slotVariables_.get(index);
+  }
+
+  public static final int USER_OBJECT_FIELD_NUMBER = 4;
+  /**
+   * <code>.tensorflow.SavedUserObject user_object = 4;</code>
+   */
+  public boolean hasUserObject() {
+    return kindCase_ == 4;
+  }
+  /**
+   * <code>.tensorflow.SavedUserObject user_object = 4;</code>
+   */
+  public org.tensorflow.proto.framework.SavedUserObject getUserObject() {
+    if (kindCase_ == 4) {
+       return (org.tensorflow.proto.framework.SavedUserObject) kind_;
+    }
+    return org.tensorflow.proto.framework.SavedUserObject.getDefaultInstance();
+  }
+  /**
+   * <code>.tensorflow.SavedUserObject user_object = 4;</code>
+   */
+  public org.tensorflow.proto.framework.SavedUserObjectOrBuilder getUserObjectOrBuilder() {
+    if (kindCase_ == 4) {
+       return (org.tensorflow.proto.framework.SavedUserObject) kind_;
+    }
+    return org.tensorflow.proto.framework.SavedUserObject.getDefaultInstance();
+  }
+
+  public static final int ASSET_FIELD_NUMBER = 5;
+  /**
+   * <code>.tensorflow.SavedAsset asset = 5;</code>
+   */
+  public boolean hasAsset() {
+    return kindCase_ == 5;
+  }
+  /**
+   * <code>.tensorflow.SavedAsset asset = 5;</code>
+   */
+  public org.tensorflow.proto.framework.SavedAsset getAsset() {
+    if (kindCase_ == 5) {
+       return (org.tensorflow.proto.framework.SavedAsset) kind_;
+    }
+    return org.tensorflow.proto.framework.SavedAsset.getDefaultInstance();
+  }
+  /**
+   * <code>.tensorflow.SavedAsset asset = 5;</code>
+   */
+  public org.tensorflow.proto.framework.SavedAssetOrBuilder getAssetOrBuilder() {
+    if (kindCase_ == 5) {
+       return (org.tensorflow.proto.framework.SavedAsset) kind_;
+    }
+    return org.tensorflow.proto.framework.SavedAsset.getDefaultInstance();
+  }
+
+  public static final int FUNCTION_FIELD_NUMBER = 6;
+  /**
+   * <code>.tensorflow.SavedFunction function = 6;</code>
+   */
+  public boolean hasFunction() {
+    return kindCase_ == 6;
+  }
+  /**
+   * <code>.tensorflow.SavedFunction function = 6;</code>
+   */
+  public org.tensorflow.proto.framework.SavedFunction getFunction() {
+    if (kindCase_ == 6) {
+       return (org.tensorflow.proto.framework.SavedFunction) kind_;
+    }
+    return org.tensorflow.proto.framework.SavedFunction.getDefaultInstance();
+  }
+  /**
+   * <code>.tensorflow.SavedFunction function = 6;</code>
+   */
+  public org.tensorflow.proto.framework.SavedFunctionOrBuilder getFunctionOrBuilder() {
+    if (kindCase_ == 6) {
+       return (org.tensorflow.proto.framework.SavedFunction) kind_;
+    }
+    return org.tensorflow.proto.framework.SavedFunction.getDefaultInstance();
+  }
+
+  public static final int VARIABLE_FIELD_NUMBER = 7;
+  /**
+   * <code>.tensorflow.SavedVariable variable = 7;</code>
+   */
+  public boolean hasVariable() {
+    return kindCase_ == 7;
+  }
+  /**
+   * <code>.tensorflow.SavedVariable variable = 7;</code>
+   */
+  public org.tensorflow.proto.framework.SavedVariable getVariable() {
+    if (kindCase_ == 7) {
+       return (org.tensorflow.proto.framework.SavedVariable) kind_;
+    }
+    return org.tensorflow.proto.framework.SavedVariable.getDefaultInstance();
+  }
+  /**
+   * <code>.tensorflow.SavedVariable variable = 7;</code>
+   */
+  public org.tensorflow.proto.framework.SavedVariableOrBuilder getVariableOrBuilder() {
+    if (kindCase_ == 7) {
+       return (org.tensorflow.proto.framework.SavedVariable) kind_;
+    }
+    return org.tensorflow.proto.framework.SavedVariable.getDefaultInstance();
+  }
+
+  public static final int BARE_CONCRETE_FUNCTION_FIELD_NUMBER = 8;
+  /**
+   * <code>.tensorflow.SavedBareConcreteFunction bare_concrete_function = 8;</code>
+   */
+  public boolean hasBareConcreteFunction() {
+    return kindCase_ == 8;
+  }
+  /**
+   * <code>.tensorflow.SavedBareConcreteFunction bare_concrete_function = 8;</code>
+   */
+  public org.tensorflow.proto.framework.SavedBareConcreteFunction getBareConcreteFunction() {
+    if (kindCase_ == 8) {
+       return (org.tensorflow.proto.framework.SavedBareConcreteFunction) kind_;
+    }
+    return org.tensorflow.proto.framework.SavedBareConcreteFunction.getDefaultInstance();
+  }
+  /**
+   * <code>.tensorflow.SavedBareConcreteFunction bare_concrete_function = 8;</code>
+   */
+  public org.tensorflow.proto.framework.SavedBareConcreteFunctionOrBuilder getBareConcreteFunctionOrBuilder() {
+    if (kindCase_ == 8) {
+       return (org.tensorflow.proto.framework.SavedBareConcreteFunction) kind_;
+    }
+    return org.tensorflow.proto.framework.SavedBareConcreteFunction.getDefaultInstance();
+  }
+
+  public static final int CONSTANT_FIELD_NUMBER = 9;
+  /**
+   * <code>.tensorflow.SavedConstant constant = 9;</code>
+   */
+  public boolean hasConstant() {
+    return kindCase_ == 9;
+  }
+  /**
+   * <code>.tensorflow.SavedConstant constant = 9;</code>
+   */
+  public org.tensorflow.proto.framework.SavedConstant getConstant() {
+    if (kindCase_ == 9) {
+       return (org.tensorflow.proto.framework.SavedConstant) kind_;
+    }
+    return org.tensorflow.proto.framework.SavedConstant.getDefaultInstance();
+  }
+  /**
+   * <code>.tensorflow.SavedConstant constant = 9;</code>
+   */
+  public org.tensorflow.proto.framework.SavedConstantOrBuilder getConstantOrBuilder() {
+    if (kindCase_ == 9) {
+       return (org.tensorflow.proto.framework.SavedConstant) kind_;
+    }
+    return org.tensorflow.proto.framework.SavedConstant.getDefaultInstance();
+  }
+
+  public static final int RESOURCE_FIELD_NUMBER = 10;
+  /**
+   * <code>.tensorflow.SavedResource resource = 10;</code>
+   */
+  public boolean hasResource() {
+    return kindCase_ == 10;
+  }
+  /**
+   * <code>.tensorflow.SavedResource resource = 10;</code>
+   */
+  public org.tensorflow.proto.framework.SavedResource getResource() {
+    if (kindCase_ == 10) {
+       return (org.tensorflow.proto.framework.SavedResource) kind_;
+    }
+    return org.tensorflow.proto.framework.SavedResource.getDefaultInstance();
+  }
+  /**
+   * <code>.tensorflow.SavedResource resource = 10;</code>
+   */
+  public org.tensorflow.proto.framework.SavedResourceOrBuilder getResourceOrBuilder() {
+    if (kindCase_ == 10) {
+       return (org.tensorflow.proto.framework.SavedResource) kind_;
+    }
+    return org.tensorflow.proto.framework.SavedResource.getDefaultInstance();
+  }
+
+  public static final int SAVEABLE_OBJECTS_FIELD_NUMBER = 11;
+  private static final class SaveableObjectsDefaultEntryHolder {
+    static final com.google.protobuf.MapEntry<
+        java.lang.String, org.tensorflow.proto.framework.SaveableObject> defaultEntry =
+            com.google.protobuf.MapEntry
+            .<java.lang.String, org.tensorflow.proto.framework.SaveableObject>newDefaultInstance(
+                org.tensorflow.proto.framework.SavedObjectGraphProtos.internal_static_tensorflow_SavedObject_SaveableObjectsEntry_descriptor, 
+                com.google.protobuf.WireFormat.FieldType.STRING,
+                "",
+                com.google.protobuf.WireFormat.FieldType.MESSAGE,
+                org.tensorflow.proto.framework.SaveableObject.getDefaultInstance());
+  }
+  private com.google.protobuf.MapField<
+      java.lang.String, org.tensorflow.proto.framework.SaveableObject> saveableObjects_;
+  private com.google.protobuf.MapField<java.lang.String, org.tensorflow.proto.framework.SaveableObject>
+  internalGetSaveableObjects() {
+    if (saveableObjects_ == null) {
+      return com.google.protobuf.MapField.emptyMapField(
+          SaveableObjectsDefaultEntryHolder.defaultEntry);
+    }
+    return saveableObjects_;
+  }
+
+  public int getSaveableObjectsCount() {
+    return internalGetSaveableObjects().getMap().size();
+  }
+  /**
+   * <code>map&lt;string, .tensorflow.SaveableObject&gt; saveable_objects = 11;</code>
+   */
+
+  public boolean containsSaveableObjects(
+      java.lang.String key) {
+    if (key == null) { throw new java.lang.NullPointerException(); }
+    return internalGetSaveableObjects().getMap().containsKey(key);
+  }
+  /**
+   * Use {@link #getSaveableObjectsMap()} instead.
+   */
+  @java.lang.Deprecated
+  public java.util.Map<java.lang.String, org.tensorflow.proto.framework.SaveableObject> getSaveableObjects() {
+    return getSaveableObjectsMap();
+  }
+  /**
+   * <code>map&lt;string, .tensorflow.SaveableObject&gt; saveable_objects = 11;</code>
+   */
+
+  public java.util.Map<java.lang.String, org.tensorflow.proto.framework.SaveableObject> getSaveableObjectsMap() {
+    return internalGetSaveableObjects().getMap();
+  }
+  /**
+   * <code>map&lt;string, .tensorflow.SaveableObject&gt; saveable_objects = 11;</code>
+   */
+
+  public org.tensorflow.proto.framework.SaveableObject getSaveableObjectsOrDefault(
+      java.lang.String key,
+      org.tensorflow.proto.framework.SaveableObject defaultValue) {
+    if (key == null) { throw new java.lang.NullPointerException(); }
+    java.util.Map<java.lang.String, org.tensorflow.proto.framework.SaveableObject> map =
+        internalGetSaveableObjects().getMap();
+    return map.containsKey(key) ? map.get(key) : defaultValue;
+  }
+  /**
+   * <code>map&lt;string, .tensorflow.SaveableObject&gt; saveable_objects = 11;</code>
+   */
+
+  public org.tensorflow.proto.framework.SaveableObject getSaveableObjectsOrThrow(
+      java.lang.String key) {
+    if (key == null) { throw new java.lang.NullPointerException(); }
+    java.util.Map<java.lang.String, org.tensorflow.proto.framework.SaveableObject> map =
+        internalGetSaveableObjects().getMap();
+    if (!map.containsKey(key)) {
+      throw new java.lang.IllegalArgumentException();
+    }
+    return map.get(key);
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    for (int i = 0; i < children_.size(); i++) {
+      output.writeMessage(1, children_.get(i));
+    }
+    for (int i = 0; i < slotVariables_.size(); i++) {
+      output.writeMessage(3, slotVariables_.get(i));
+    }
+    if (kindCase_ == 4) {
+      output.writeMessage(4, (org.tensorflow.proto.framework.SavedUserObject) kind_);
+    }
+    if (kindCase_ == 5) {
+      output.writeMessage(5, (org.tensorflow.proto.framework.SavedAsset) kind_);
+    }
+    if (kindCase_ == 6) {
+      output.writeMessage(6, (org.tensorflow.proto.framework.SavedFunction) kind_);
+    }
+    if (kindCase_ == 7) {
+      output.writeMessage(7, (org.tensorflow.proto.framework.SavedVariable) kind_);
+    }
+    if (kindCase_ == 8) {
+      output.writeMessage(8, (org.tensorflow.proto.framework.SavedBareConcreteFunction) kind_);
+    }
+    if (kindCase_ == 9) {
+      output.writeMessage(9, (org.tensorflow.proto.framework.SavedConstant) kind_);
+    }
+    if (kindCase_ == 10) {
+      output.writeMessage(10, (org.tensorflow.proto.framework.SavedResource) kind_);
+    }
+    com.google.protobuf.GeneratedMessageV3
+      .serializeStringMapTo(
+        output,
+        internalGetSaveableObjects(),
+        SaveableObjectsDefaultEntryHolder.defaultEntry,
+        11);
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    for (int i = 0; i < children_.size(); i++) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(1, children_.get(i));
+    }
+    for (int i = 0; i < slotVariables_.size(); i++) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(3, slotVariables_.get(i));
+    }
+    if (kindCase_ == 4) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(4, (org.tensorflow.proto.framework.SavedUserObject) kind_);
+    }
+    if (kindCase_ == 5) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(5, (org.tensorflow.proto.framework.SavedAsset) kind_);
+    }
+    if (kindCase_ == 6) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(6, (org.tensorflow.proto.framework.SavedFunction) kind_);
+    }
+    if (kindCase_ == 7) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(7, (org.tensorflow.proto.framework.SavedVariable) kind_);
+    }
+    if (kindCase_ == 8) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(8, (org.tensorflow.proto.framework.SavedBareConcreteFunction) kind_);
+    }
+    if (kindCase_ == 9) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(9, (org.tensorflow.proto.framework.SavedConstant) kind_);
+    }
+    if (kindCase_ == 10) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(10, (org.tensorflow.proto.framework.SavedResource) kind_);
+    }
+    for (java.util.Map.Entry<java.lang.String, org.tensorflow.proto.framework.SaveableObject> entry
+         : internalGetSaveableObjects().getMap().entrySet()) {
+      com.google.protobuf.MapEntry<java.lang.String, org.tensorflow.proto.framework.SaveableObject>
+      saveableObjects__ = SaveableObjectsDefaultEntryHolder.defaultEntry.newBuilderForType()
+          .setKey(entry.getKey())
+          .setValue(entry.getValue())
+          .build();
+      size += com.google.protobuf.CodedOutputStream
+          .computeMessageSize(11, saveableObjects__);
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.framework.SavedObject)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.framework.SavedObject other = (org.tensorflow.proto.framework.SavedObject) obj;
+
+    if (!getChildrenList()
+        .equals(other.getChildrenList())) return false;
+    if (!getSlotVariablesList()
+        .equals(other.getSlotVariablesList())) return false;
+    if (!internalGetSaveableObjects().equals(
+        other.internalGetSaveableObjects())) return false;
+    if (!getKindCase().equals(other.getKindCase())) return false;
+    switch (kindCase_) {
+      case 4:
+        if (!getUserObject()
+            .equals(other.getUserObject())) return false;
+        break;
+      case 5:
+        if (!getAsset()
+            .equals(other.getAsset())) return false;
+        break;
+      case 6:
+        if (!getFunction()
+            .equals(other.getFunction())) return false;
+        break;
+      case 7:
+        if (!getVariable()
+            .equals(other.getVariable())) return false;
+        break;
+      case 8:
+        if (!getBareConcreteFunction()
+            .equals(other.getBareConcreteFunction())) return false;
+        break;
+      case 9:
+        if (!getConstant()
+            .equals(other.getConstant())) return false;
+        break;
+      case 10:
+        if (!getResource()
+            .equals(other.getResource())) return false;
+        break;
+      case 0:
+      default:
+    }
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    if (getChildrenCount() > 0) {
+      hash = (37 * hash) + CHILDREN_FIELD_NUMBER;
+      hash = (53 * hash) + getChildrenList().hashCode();
+    }
+    if (getSlotVariablesCount() > 0) {
+      hash = (37 * hash) + SLOT_VARIABLES_FIELD_NUMBER;
+      hash = (53 * hash) + getSlotVariablesList().hashCode();
+    }
+    if (!internalGetSaveableObjects().getMap().isEmpty()) {
+      hash = (37 * hash) + SAVEABLE_OBJECTS_FIELD_NUMBER;
+      hash = (53 * hash) + internalGetSaveableObjects().hashCode();
+    }
+    switch (kindCase_) {
+      case 4:
+        hash = (37 * hash) + USER_OBJECT_FIELD_NUMBER;
+        hash = (53 * hash) + getUserObject().hashCode();
+        break;
+      case 5:
+        hash = (37 * hash) + ASSET_FIELD_NUMBER;
+        hash = (53 * hash) + getAsset().hashCode();
+        break;
+      case 6:
+        hash = (37 * hash) + FUNCTION_FIELD_NUMBER;
+        hash = (53 * hash) + getFunction().hashCode();
+        break;
+      case 7:
+        hash = (37 * hash) + VARIABLE_FIELD_NUMBER;
+        hash = (53 * hash) + getVariable().hashCode();
+        break;
+      case 8:
+        hash = (37 * hash) + BARE_CONCRETE_FUNCTION_FIELD_NUMBER;
+        hash = (53 * hash) + getBareConcreteFunction().hashCode();
+        break;
+      case 9:
+        hash = (37 * hash) + CONSTANT_FIELD_NUMBER;
+        hash = (53 * hash) + getConstant().hashCode();
+        break;
+      case 10:
+        hash = (37 * hash) + RESOURCE_FIELD_NUMBER;
+        hash = (53 * hash) + getResource().hashCode();
+        break;
+      case 0:
+      default:
+    }
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.framework.SavedObject parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.SavedObject parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.SavedObject parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.SavedObject parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.SavedObject parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.SavedObject parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.SavedObject parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.SavedObject parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.SavedObject parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.SavedObject parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.SavedObject parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.SavedObject parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.framework.SavedObject prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * Protobuf type {@code tensorflow.SavedObject}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.SavedObject)
+      org.tensorflow.proto.framework.SavedObjectOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.SavedObjectGraphProtos.internal_static_tensorflow_SavedObject_descriptor;
+    }
+
+    @SuppressWarnings({"rawtypes"})
+    protected com.google.protobuf.MapField internalGetMapField(
+        int number) {
+      switch (number) {
+        case 11:
+          return internalGetSaveableObjects();
+        default:
+          throw new RuntimeException(
+              "Invalid map field number: " + number);
+      }
+    }
+    @SuppressWarnings({"rawtypes"})
+    protected com.google.protobuf.MapField internalGetMutableMapField(
+        int number) {
+      switch (number) {
+        case 11:
+          return internalGetMutableSaveableObjects();
+        default:
+          throw new RuntimeException(
+              "Invalid map field number: " + number);
+      }
+    }
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.SavedObjectGraphProtos.internal_static_tensorflow_SavedObject_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.SavedObject.class, org.tensorflow.proto.framework.SavedObject.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.framework.SavedObject.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+        getChildrenFieldBuilder();
+        getSlotVariablesFieldBuilder();
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      if (childrenBuilder_ == null) {
+        children_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000001);
+      } else {
+        childrenBuilder_.clear();
+      }
+      if (slotVariablesBuilder_ == null) {
+        slotVariables_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000002);
+      } else {
+        slotVariablesBuilder_.clear();
+      }
+      internalGetMutableSaveableObjects().clear();
+      kindCase_ = 0;
+      kind_ = null;
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.framework.SavedObjectGraphProtos.internal_static_tensorflow_SavedObject_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.SavedObject getDefaultInstanceForType() {
+      return org.tensorflow.proto.framework.SavedObject.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.SavedObject build() {
+      org.tensorflow.proto.framework.SavedObject result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.SavedObject buildPartial() {
+      org.tensorflow.proto.framework.SavedObject result = new org.tensorflow.proto.framework.SavedObject(this);
+      int from_bitField0_ = bitField0_;
+      if (childrenBuilder_ == null) {
+        if (((bitField0_ & 0x00000001) != 0)) {
+          children_ = java.util.Collections.unmodifiableList(children_);
+          bitField0_ = (bitField0_ & ~0x00000001);
+        }
+        result.children_ = children_;
+      } else {
+        result.children_ = childrenBuilder_.build();
+      }
+      if (slotVariablesBuilder_ == null) {
+        if (((bitField0_ & 0x00000002) != 0)) {
+          slotVariables_ = java.util.Collections.unmodifiableList(slotVariables_);
+          bitField0_ = (bitField0_ & ~0x00000002);
+        }
+        result.slotVariables_ = slotVariables_;
+      } else {
+        result.slotVariables_ = slotVariablesBuilder_.build();
+      }
+      if (kindCase_ == 4) {
+        if (userObjectBuilder_ == null) {
+          result.kind_ = kind_;
+        } else {
+          result.kind_ = userObjectBuilder_.build();
+        }
+      }
+      if (kindCase_ == 5) {
+        if (assetBuilder_ == null) {
+          result.kind_ = kind_;
+        } else {
+          result.kind_ = assetBuilder_.build();
+        }
+      }
+      if (kindCase_ == 6) {
+        if (functionBuilder_ == null) {
+          result.kind_ = kind_;
+        } else {
+          result.kind_ = functionBuilder_.build();
+        }
+      }
+      if (kindCase_ == 7) {
+        if (variableBuilder_ == null) {
+          result.kind_ = kind_;
+        } else {
+          result.kind_ = variableBuilder_.build();
+        }
+      }
+      if (kindCase_ == 8) {
+        if (bareConcreteFunctionBuilder_ == null) {
+          result.kind_ = kind_;
+        } else {
+          result.kind_ = bareConcreteFunctionBuilder_.build();
+        }
+      }
+      if (kindCase_ == 9) {
+        if (constantBuilder_ == null) {
+          result.kind_ = kind_;
+        } else {
+          result.kind_ = constantBuilder_.build();
+        }
+      }
+      if (kindCase_ == 10) {
+        if (resourceBuilder_ == null) {
+          result.kind_ = kind_;
+        } else {
+          result.kind_ = resourceBuilder_.build();
+        }
+      }
+      result.saveableObjects_ = internalGetSaveableObjects();
+      result.saveableObjects_.makeImmutable();
+      result.kindCase_ = kindCase_;
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.framework.SavedObject) {
+        return mergeFrom((org.tensorflow.proto.framework.SavedObject)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.framework.SavedObject other) {
+      if (other == org.tensorflow.proto.framework.SavedObject.getDefaultInstance()) return this;
+      if (childrenBuilder_ == null) {
+        if (!other.children_.isEmpty()) {
+          if (children_.isEmpty()) {
+            children_ = other.children_;
+            bitField0_ = (bitField0_ & ~0x00000001);
+          } else {
+            ensureChildrenIsMutable();
+            children_.addAll(other.children_);
+          }
+          onChanged();
+        }
+      } else {
+        if (!other.children_.isEmpty()) {
+          if (childrenBuilder_.isEmpty()) {
+            childrenBuilder_.dispose();
+            childrenBuilder_ = null;
+            children_ = other.children_;
+            bitField0_ = (bitField0_ & ~0x00000001);
+            childrenBuilder_ = 
+              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
+                 getChildrenFieldBuilder() : null;
+          } else {
+            childrenBuilder_.addAllMessages(other.children_);
+          }
+        }
+      }
+      if (slotVariablesBuilder_ == null) {
+        if (!other.slotVariables_.isEmpty()) {
+          if (slotVariables_.isEmpty()) {
+            slotVariables_ = other.slotVariables_;
+            bitField0_ = (bitField0_ & ~0x00000002);
+          } else {
+            ensureSlotVariablesIsMutable();
+            slotVariables_.addAll(other.slotVariables_);
+          }
+          onChanged();
+        }
+      } else {
+        if (!other.slotVariables_.isEmpty()) {
+          if (slotVariablesBuilder_.isEmpty()) {
+            slotVariablesBuilder_.dispose();
+            slotVariablesBuilder_ = null;
+            slotVariables_ = other.slotVariables_;
+            bitField0_ = (bitField0_ & ~0x00000002);
+            slotVariablesBuilder_ = 
+              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
+                 getSlotVariablesFieldBuilder() : null;
+          } else {
+            slotVariablesBuilder_.addAllMessages(other.slotVariables_);
+          }
+        }
+      }
+      internalGetMutableSaveableObjects().mergeFrom(
+          other.internalGetSaveableObjects());
+      switch (other.getKindCase()) {
+        case USER_OBJECT: {
+          mergeUserObject(other.getUserObject());
+          break;
+        }
+        case ASSET: {
+          mergeAsset(other.getAsset());
+          break;
+        }
+        case FUNCTION: {
+          mergeFunction(other.getFunction());
+          break;
+        }
+        case VARIABLE: {
+          mergeVariable(other.getVariable());
+          break;
+        }
+        case BARE_CONCRETE_FUNCTION: {
+          mergeBareConcreteFunction(other.getBareConcreteFunction());
+          break;
+        }
+        case CONSTANT: {
+          mergeConstant(other.getConstant());
+          break;
+        }
+        case RESOURCE: {
+          mergeResource(other.getResource());
+          break;
+        }
+        case KIND_NOT_SET: {
+          break;
+        }
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.framework.SavedObject parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.framework.SavedObject) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+    private int kindCase_ = 0;
+    private java.lang.Object kind_;
+    public KindCase
+        getKindCase() {
+      return KindCase.forNumber(
+          kindCase_);
+    }
+
+    public Builder clearKind() {
+      kindCase_ = 0;
+      kind_ = null;
+      onChanged();
+      return this;
+    }
+
+    private int bitField0_;
+
+    private java.util.List<org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference> children_ =
+      java.util.Collections.emptyList();
+    private void ensureChildrenIsMutable() {
+      if (!((bitField0_ & 0x00000001) != 0)) {
+        children_ = new java.util.ArrayList<org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference>(children_);
+        bitField0_ |= 0x00000001;
+       }
+    }
+
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference.Builder, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReferenceOrBuilder> childrenBuilder_;
+
+    /**
+     * <pre>
+     * Objects which this object depends on: named edges in the dependency
+     * graph.
+     * Note: currently only valid if kind == "user_object".
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.ObjectReference children = 1;</code>
+     */
+    public java.util.List<org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference> getChildrenList() {
+      if (childrenBuilder_ == null) {
+        return java.util.Collections.unmodifiableList(children_);
+      } else {
+        return childrenBuilder_.getMessageList();
+      }
+    }
+    /**
+     * <pre>
+     * Objects which this object depends on: named edges in the dependency
+     * graph.
+     * Note: currently only valid if kind == "user_object".
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.ObjectReference children = 1;</code>
+     */
+    public int getChildrenCount() {
+      if (childrenBuilder_ == null) {
+        return children_.size();
+      } else {
+        return childrenBuilder_.getCount();
+      }
+    }
+    /**
+     * <pre>
+     * Objects which this object depends on: named edges in the dependency
+     * graph.
+     * Note: currently only valid if kind == "user_object".
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.ObjectReference children = 1;</code>
+     */
+    public org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference getChildren(int index) {
+      if (childrenBuilder_ == null) {
+        return children_.get(index);
+      } else {
+        return childrenBuilder_.getMessage(index);
+      }
+    }
+    /**
+     * <pre>
+     * Objects which this object depends on: named edges in the dependency
+     * graph.
+     * Note: currently only valid if kind == "user_object".
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.ObjectReference children = 1;</code>
+     */
+    public Builder setChildren(
+        int index, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference value) {
+      if (childrenBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureChildrenIsMutable();
+        children_.set(index, value);
+        onChanged();
+      } else {
+        childrenBuilder_.setMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Objects which this object depends on: named edges in the dependency
+     * graph.
+     * Note: currently only valid if kind == "user_object".
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.ObjectReference children = 1;</code>
+     */
+    public Builder setChildren(
+        int index, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference.Builder builderForValue) {
+      if (childrenBuilder_ == null) {
+        ensureChildrenIsMutable();
+        children_.set(index, builderForValue.build());
+        onChanged();
+      } else {
+        childrenBuilder_.setMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Objects which this object depends on: named edges in the dependency
+     * graph.
+     * Note: currently only valid if kind == "user_object".
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.ObjectReference children = 1;</code>
+     */
+    public Builder addChildren(org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference value) {
+      if (childrenBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureChildrenIsMutable();
+        children_.add(value);
+        onChanged();
+      } else {
+        childrenBuilder_.addMessage(value);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Objects which this object depends on: named edges in the dependency
+     * graph.
+     * Note: currently only valid if kind == "user_object".
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.ObjectReference children = 1;</code>
+     */
+    public Builder addChildren(
+        int index, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference value) {
+      if (childrenBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureChildrenIsMutable();
+        children_.add(index, value);
+        onChanged();
+      } else {
+        childrenBuilder_.addMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Objects which this object depends on: named edges in the dependency
+     * graph.
+     * Note: currently only valid if kind == "user_object".
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.ObjectReference children = 1;</code>
+     */
+    public Builder addChildren(
+        org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference.Builder builderForValue) {
+      if (childrenBuilder_ == null) {
+        ensureChildrenIsMutable();
+        children_.add(builderForValue.build());
+        onChanged();
+      } else {
+        childrenBuilder_.addMessage(builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Objects which this object depends on: named edges in the dependency
+     * graph.
+     * Note: currently only valid if kind == "user_object".
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.ObjectReference children = 1;</code>
+     */
+    public Builder addChildren(
+        int index, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference.Builder builderForValue) {
+      if (childrenBuilder_ == null) {
+        ensureChildrenIsMutable();
+        children_.add(index, builderForValue.build());
+        onChanged();
+      } else {
+        childrenBuilder_.addMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Objects which this object depends on: named edges in the dependency
+     * graph.
+     * Note: currently only valid if kind == "user_object".
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.ObjectReference children = 1;</code>
+     */
+    public Builder addAllChildren(
+        java.lang.Iterable<? extends org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference> values) {
+      if (childrenBuilder_ == null) {
+        ensureChildrenIsMutable();
+        com.google.protobuf.AbstractMessageLite.Builder.addAll(
+            values, children_);
+        onChanged();
+      } else {
+        childrenBuilder_.addAllMessages(values);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Objects which this object depends on: named edges in the dependency
+     * graph.
+     * Note: currently only valid if kind == "user_object".
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.ObjectReference children = 1;</code>
+     */
+    public Builder clearChildren() {
+      if (childrenBuilder_ == null) {
+        children_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000001);
+        onChanged();
+      } else {
+        childrenBuilder_.clear();
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Objects which this object depends on: named edges in the dependency
+     * graph.
+     * Note: currently only valid if kind == "user_object".
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.ObjectReference children = 1;</code>
+     */
+    public Builder removeChildren(int index) {
+      if (childrenBuilder_ == null) {
+        ensureChildrenIsMutable();
+        children_.remove(index);
+        onChanged();
+      } else {
+        childrenBuilder_.remove(index);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Objects which this object depends on: named edges in the dependency
+     * graph.
+     * Note: currently only valid if kind == "user_object".
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.ObjectReference children = 1;</code>
+     */
+    public org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference.Builder getChildrenBuilder(
+        int index) {
+      return getChildrenFieldBuilder().getBuilder(index);
+    }
+    /**
+     * <pre>
+     * Objects which this object depends on: named edges in the dependency
+     * graph.
+     * Note: currently only valid if kind == "user_object".
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.ObjectReference children = 1;</code>
+     */
+    public org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReferenceOrBuilder getChildrenOrBuilder(
+        int index) {
+      if (childrenBuilder_ == null) {
+        return children_.get(index);  } else {
+        return childrenBuilder_.getMessageOrBuilder(index);
+      }
+    }
+    /**
+     * <pre>
+     * Objects which this object depends on: named edges in the dependency
+     * graph.
+     * Note: currently only valid if kind == "user_object".
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.ObjectReference children = 1;</code>
+     */
+    public java.util.List<? extends org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReferenceOrBuilder> 
+         getChildrenOrBuilderList() {
+      if (childrenBuilder_ != null) {
+        return childrenBuilder_.getMessageOrBuilderList();
+      } else {
+        return java.util.Collections.unmodifiableList(children_);
+      }
+    }
+    /**
+     * <pre>
+     * Objects which this object depends on: named edges in the dependency
+     * graph.
+     * Note: currently only valid if kind == "user_object".
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.ObjectReference children = 1;</code>
+     */
+    public org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference.Builder addChildrenBuilder() {
+      return getChildrenFieldBuilder().addBuilder(
+          org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference.getDefaultInstance());
+    }
+    /**
+     * <pre>
+     * Objects which this object depends on: named edges in the dependency
+     * graph.
+     * Note: currently only valid if kind == "user_object".
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.ObjectReference children = 1;</code>
+     */
+    public org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference.Builder addChildrenBuilder(
+        int index) {
+      return getChildrenFieldBuilder().addBuilder(
+          index, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference.getDefaultInstance());
+    }
+    /**
+     * <pre>
+     * Objects which this object depends on: named edges in the dependency
+     * graph.
+     * Note: currently only valid if kind == "user_object".
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.ObjectReference children = 1;</code>
+     */
+    public java.util.List<org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference.Builder> 
+         getChildrenBuilderList() {
+      return getChildrenFieldBuilder().getBuilderList();
+    }
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference.Builder, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReferenceOrBuilder> 
+        getChildrenFieldBuilder() {
+      if (childrenBuilder_ == null) {
+        childrenBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
+            org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference.Builder, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReferenceOrBuilder>(
+                children_,
+                ((bitField0_ & 0x00000001) != 0),
+                getParentForChildren(),
+                isClean());
+        children_ = null;
+      }
+      return childrenBuilder_;
+    }
+
+    private java.util.List<org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference> slotVariables_ =
+      java.util.Collections.emptyList();
+    private void ensureSlotVariablesIsMutable() {
+      if (!((bitField0_ & 0x00000002) != 0)) {
+        slotVariables_ = new java.util.ArrayList<org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference>(slotVariables_);
+        bitField0_ |= 0x00000002;
+       }
+    }
+
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference.Builder, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReferenceOrBuilder> slotVariablesBuilder_;
+
+    /**
+     * <pre>
+     * Slot variables owned by this object. This describes the three-way
+     * (optimizer, variable, slot variable) relationship; none of the three
+     * depend on the others directly.
+     * Note: currently only valid if kind == "user_object".
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SlotVariableReference slot_variables = 3;</code>
+     */
+    public java.util.List<org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference> getSlotVariablesList() {
+      if (slotVariablesBuilder_ == null) {
+        return java.util.Collections.unmodifiableList(slotVariables_);
+      } else {
+        return slotVariablesBuilder_.getMessageList();
+      }
+    }
+    /**
+     * <pre>
+     * Slot variables owned by this object. This describes the three-way
+     * (optimizer, variable, slot variable) relationship; none of the three
+     * depend on the others directly.
+     * Note: currently only valid if kind == "user_object".
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SlotVariableReference slot_variables = 3;</code>
+     */
+    public int getSlotVariablesCount() {
+      if (slotVariablesBuilder_ == null) {
+        return slotVariables_.size();
+      } else {
+        return slotVariablesBuilder_.getCount();
+      }
+    }
+    /**
+     * <pre>
+     * Slot variables owned by this object. This describes the three-way
+     * (optimizer, variable, slot variable) relationship; none of the three
+     * depend on the others directly.
+     * Note: currently only valid if kind == "user_object".
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SlotVariableReference slot_variables = 3;</code>
+     */
+    public org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference getSlotVariables(int index) {
+      if (slotVariablesBuilder_ == null) {
+        return slotVariables_.get(index);
+      } else {
+        return slotVariablesBuilder_.getMessage(index);
+      }
+    }
+    /**
+     * <pre>
+     * Slot variables owned by this object. This describes the three-way
+     * (optimizer, variable, slot variable) relationship; none of the three
+     * depend on the others directly.
+     * Note: currently only valid if kind == "user_object".
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SlotVariableReference slot_variables = 3;</code>
+     */
+    public Builder setSlotVariables(
+        int index, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference value) {
+      if (slotVariablesBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureSlotVariablesIsMutable();
+        slotVariables_.set(index, value);
+        onChanged();
+      } else {
+        slotVariablesBuilder_.setMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Slot variables owned by this object. This describes the three-way
+     * (optimizer, variable, slot variable) relationship; none of the three
+     * depend on the others directly.
+     * Note: currently only valid if kind == "user_object".
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SlotVariableReference slot_variables = 3;</code>
+     */
+    public Builder setSlotVariables(
+        int index, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference.Builder builderForValue) {
+      if (slotVariablesBuilder_ == null) {
+        ensureSlotVariablesIsMutable();
+        slotVariables_.set(index, builderForValue.build());
+        onChanged();
+      } else {
+        slotVariablesBuilder_.setMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Slot variables owned by this object. This describes the three-way
+     * (optimizer, variable, slot variable) relationship; none of the three
+     * depend on the others directly.
+     * Note: currently only valid if kind == "user_object".
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SlotVariableReference slot_variables = 3;</code>
+     */
+    public Builder addSlotVariables(org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference value) {
+      if (slotVariablesBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureSlotVariablesIsMutable();
+        slotVariables_.add(value);
+        onChanged();
+      } else {
+        slotVariablesBuilder_.addMessage(value);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Slot variables owned by this object. This describes the three-way
+     * (optimizer, variable, slot variable) relationship; none of the three
+     * depend on the others directly.
+     * Note: currently only valid if kind == "user_object".
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SlotVariableReference slot_variables = 3;</code>
+     */
+    public Builder addSlotVariables(
+        int index, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference value) {
+      if (slotVariablesBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureSlotVariablesIsMutable();
+        slotVariables_.add(index, value);
+        onChanged();
+      } else {
+        slotVariablesBuilder_.addMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Slot variables owned by this object. This describes the three-way
+     * (optimizer, variable, slot variable) relationship; none of the three
+     * depend on the others directly.
+     * Note: currently only valid if kind == "user_object".
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SlotVariableReference slot_variables = 3;</code>
+     */
+    public Builder addSlotVariables(
+        org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference.Builder builderForValue) {
+      if (slotVariablesBuilder_ == null) {
+        ensureSlotVariablesIsMutable();
+        slotVariables_.add(builderForValue.build());
+        onChanged();
+      } else {
+        slotVariablesBuilder_.addMessage(builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Slot variables owned by this object. This describes the three-way
+     * (optimizer, variable, slot variable) relationship; none of the three
+     * depend on the others directly.
+     * Note: currently only valid if kind == "user_object".
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SlotVariableReference slot_variables = 3;</code>
+     */
+    public Builder addSlotVariables(
+        int index, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference.Builder builderForValue) {
+      if (slotVariablesBuilder_ == null) {
+        ensureSlotVariablesIsMutable();
+        slotVariables_.add(index, builderForValue.build());
+        onChanged();
+      } else {
+        slotVariablesBuilder_.addMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Slot variables owned by this object. This describes the three-way
+     * (optimizer, variable, slot variable) relationship; none of the three
+     * depend on the others directly.
+     * Note: currently only valid if kind == "user_object".
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SlotVariableReference slot_variables = 3;</code>
+     */
+    public Builder addAllSlotVariables(
+        java.lang.Iterable<? extends org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference> values) {
+      if (slotVariablesBuilder_ == null) {
+        ensureSlotVariablesIsMutable();
+        com.google.protobuf.AbstractMessageLite.Builder.addAll(
+            values, slotVariables_);
+        onChanged();
+      } else {
+        slotVariablesBuilder_.addAllMessages(values);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Slot variables owned by this object. This describes the three-way
+     * (optimizer, variable, slot variable) relationship; none of the three
+     * depend on the others directly.
+     * Note: currently only valid if kind == "user_object".
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SlotVariableReference slot_variables = 3;</code>
+     */
+    public Builder clearSlotVariables() {
+      if (slotVariablesBuilder_ == null) {
+        slotVariables_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000002);
+        onChanged();
+      } else {
+        slotVariablesBuilder_.clear();
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Slot variables owned by this object. This describes the three-way
+     * (optimizer, variable, slot variable) relationship; none of the three
+     * depend on the others directly.
+     * Note: currently only valid if kind == "user_object".
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SlotVariableReference slot_variables = 3;</code>
+     */
+    public Builder removeSlotVariables(int index) {
+      if (slotVariablesBuilder_ == null) {
+        ensureSlotVariablesIsMutable();
+        slotVariables_.remove(index);
+        onChanged();
+      } else {
+        slotVariablesBuilder_.remove(index);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Slot variables owned by this object. This describes the three-way
+     * (optimizer, variable, slot variable) relationship; none of the three
+     * depend on the others directly.
+     * Note: currently only valid if kind == "user_object".
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SlotVariableReference slot_variables = 3;</code>
+     */
+    public org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference.Builder getSlotVariablesBuilder(
+        int index) {
+      return getSlotVariablesFieldBuilder().getBuilder(index);
+    }
+    /**
+     * <pre>
+     * Slot variables owned by this object. This describes the three-way
+     * (optimizer, variable, slot variable) relationship; none of the three
+     * depend on the others directly.
+     * Note: currently only valid if kind == "user_object".
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SlotVariableReference slot_variables = 3;</code>
+     */
+    public org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReferenceOrBuilder getSlotVariablesOrBuilder(
+        int index) {
+      if (slotVariablesBuilder_ == null) {
+        return slotVariables_.get(index);  } else {
+        return slotVariablesBuilder_.getMessageOrBuilder(index);
+      }
+    }
+    /**
+     * <pre>
+     * Slot variables owned by this object. This describes the three-way
+     * (optimizer, variable, slot variable) relationship; none of the three
+     * depend on the others directly.
+     * Note: currently only valid if kind == "user_object".
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SlotVariableReference slot_variables = 3;</code>
+     */
+    public java.util.List<? extends org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReferenceOrBuilder> 
+         getSlotVariablesOrBuilderList() {
+      if (slotVariablesBuilder_ != null) {
+        return slotVariablesBuilder_.getMessageOrBuilderList();
+      } else {
+        return java.util.Collections.unmodifiableList(slotVariables_);
+      }
+    }
+    /**
+     * <pre>
+     * Slot variables owned by this object. This describes the three-way
+     * (optimizer, variable, slot variable) relationship; none of the three
+     * depend on the others directly.
+     * Note: currently only valid if kind == "user_object".
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SlotVariableReference slot_variables = 3;</code>
+     */
+    public org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference.Builder addSlotVariablesBuilder() {
+      return getSlotVariablesFieldBuilder().addBuilder(
+          org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference.getDefaultInstance());
+    }
+    /**
+     * <pre>
+     * Slot variables owned by this object. This describes the three-way
+     * (optimizer, variable, slot variable) relationship; none of the three
+     * depend on the others directly.
+     * Note: currently only valid if kind == "user_object".
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SlotVariableReference slot_variables = 3;</code>
+     */
+    public org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference.Builder addSlotVariablesBuilder(
+        int index) {
+      return getSlotVariablesFieldBuilder().addBuilder(
+          index, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference.getDefaultInstance());
+    }
+    /**
+     * <pre>
+     * Slot variables owned by this object. This describes the three-way
+     * (optimizer, variable, slot variable) relationship; none of the three
+     * depend on the others directly.
+     * Note: currently only valid if kind == "user_object".
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SlotVariableReference slot_variables = 3;</code>
+     */
+    public java.util.List<org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference.Builder> 
+         getSlotVariablesBuilderList() {
+      return getSlotVariablesFieldBuilder().getBuilderList();
+    }
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference.Builder, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReferenceOrBuilder> 
+        getSlotVariablesFieldBuilder() {
+      if (slotVariablesBuilder_ == null) {
+        slotVariablesBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
+            org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference.Builder, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReferenceOrBuilder>(
+                slotVariables_,
+                ((bitField0_ & 0x00000002) != 0),
+                getParentForChildren(),
+                isClean());
+        slotVariables_ = null;
+      }
+      return slotVariablesBuilder_;
+    }
+
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.SavedUserObject, org.tensorflow.proto.framework.SavedUserObject.Builder, org.tensorflow.proto.framework.SavedUserObjectOrBuilder> userObjectBuilder_;
+    /**
+     * <code>.tensorflow.SavedUserObject user_object = 4;</code>
+     */
+    public boolean hasUserObject() {
+      return kindCase_ == 4;
+    }
+    /**
+     * <code>.tensorflow.SavedUserObject user_object = 4;</code>
+     */
+    public org.tensorflow.proto.framework.SavedUserObject getUserObject() {
+      if (userObjectBuilder_ == null) {
+        if (kindCase_ == 4) {
+          return (org.tensorflow.proto.framework.SavedUserObject) kind_;
+        }
+        return org.tensorflow.proto.framework.SavedUserObject.getDefaultInstance();
+      } else {
+        if (kindCase_ == 4) {
+          return userObjectBuilder_.getMessage();
+        }
+        return org.tensorflow.proto.framework.SavedUserObject.getDefaultInstance();
+      }
+    }
+    /**
+     * <code>.tensorflow.SavedUserObject user_object = 4;</code>
+     */
+    public Builder setUserObject(org.tensorflow.proto.framework.SavedUserObject value) {
+      if (userObjectBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        kind_ = value;
+        onChanged();
+      } else {
+        userObjectBuilder_.setMessage(value);
+      }
+      kindCase_ = 4;
+      return this;
+    }
+    /**
+     * <code>.tensorflow.SavedUserObject user_object = 4;</code>
+     */
+    public Builder setUserObject(
+        org.tensorflow.proto.framework.SavedUserObject.Builder builderForValue) {
+      if (userObjectBuilder_ == null) {
+        kind_ = builderForValue.build();
+        onChanged();
+      } else {
+        userObjectBuilder_.setMessage(builderForValue.build());
+      }
+      kindCase_ = 4;
+      return this;
+    }
+    /**
+     * <code>.tensorflow.SavedUserObject user_object = 4;</code>
+     */
+    public Builder mergeUserObject(org.tensorflow.proto.framework.SavedUserObject value) {
+      if (userObjectBuilder_ == null) {
+        if (kindCase_ == 4 &&
+            kind_ != org.tensorflow.proto.framework.SavedUserObject.getDefaultInstance()) {
+          kind_ = org.tensorflow.proto.framework.SavedUserObject.newBuilder((org.tensorflow.proto.framework.SavedUserObject) kind_)
+              .mergeFrom(value).buildPartial();
+        } else {
+          kind_ = value;
+        }
+        onChanged();
+      } else {
+        if (kindCase_ == 4) {
+          userObjectBuilder_.mergeFrom(value);
+        }
+        userObjectBuilder_.setMessage(value);
+      }
+      kindCase_ = 4;
+      return this;
+    }
+    /**
+     * <code>.tensorflow.SavedUserObject user_object = 4;</code>
+     */
+    public Builder clearUserObject() {
+      if (userObjectBuilder_ == null) {
+        if (kindCase_ == 4) {
+          kindCase_ = 0;
+          kind_ = null;
+          onChanged();
+        }
+      } else {
+        if (kindCase_ == 4) {
+          kindCase_ = 0;
+          kind_ = null;
+        }
+        userObjectBuilder_.clear();
+      }
+      return this;
+    }
+    /**
+     * <code>.tensorflow.SavedUserObject user_object = 4;</code>
+     */
+    public org.tensorflow.proto.framework.SavedUserObject.Builder getUserObjectBuilder() {
+      return getUserObjectFieldBuilder().getBuilder();
+    }
+    /**
+     * <code>.tensorflow.SavedUserObject user_object = 4;</code>
+     */
+    public org.tensorflow.proto.framework.SavedUserObjectOrBuilder getUserObjectOrBuilder() {
+      if ((kindCase_ == 4) && (userObjectBuilder_ != null)) {
+        return userObjectBuilder_.getMessageOrBuilder();
+      } else {
+        if (kindCase_ == 4) {
+          return (org.tensorflow.proto.framework.SavedUserObject) kind_;
+        }
+        return org.tensorflow.proto.framework.SavedUserObject.getDefaultInstance();
+      }
+    }
+    /**
+     * <code>.tensorflow.SavedUserObject user_object = 4;</code>
+     */
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.SavedUserObject, org.tensorflow.proto.framework.SavedUserObject.Builder, org.tensorflow.proto.framework.SavedUserObjectOrBuilder> 
+        getUserObjectFieldBuilder() {
+      if (userObjectBuilder_ == null) {
+        if (!(kindCase_ == 4)) {
+          kind_ = org.tensorflow.proto.framework.SavedUserObject.getDefaultInstance();
+        }
+        userObjectBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+            org.tensorflow.proto.framework.SavedUserObject, org.tensorflow.proto.framework.SavedUserObject.Builder, org.tensorflow.proto.framework.SavedUserObjectOrBuilder>(
+                (org.tensorflow.proto.framework.SavedUserObject) kind_,
+                getParentForChildren(),
+                isClean());
+        kind_ = null;
+      }
+      kindCase_ = 4;
+      onChanged();;
+      return userObjectBuilder_;
+    }
+
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.SavedAsset, org.tensorflow.proto.framework.SavedAsset.Builder, org.tensorflow.proto.framework.SavedAssetOrBuilder> assetBuilder_;
+    /**
+     * <code>.tensorflow.SavedAsset asset = 5;</code>
+     */
+    public boolean hasAsset() {
+      return kindCase_ == 5;
+    }
+    /**
+     * <code>.tensorflow.SavedAsset asset = 5;</code>
+     */
+    public org.tensorflow.proto.framework.SavedAsset getAsset() {
+      if (assetBuilder_ == null) {
+        if (kindCase_ == 5) {
+          return (org.tensorflow.proto.framework.SavedAsset) kind_;
+        }
+        return org.tensorflow.proto.framework.SavedAsset.getDefaultInstance();
+      } else {
+        if (kindCase_ == 5) {
+          return assetBuilder_.getMessage();
+        }
+        return org.tensorflow.proto.framework.SavedAsset.getDefaultInstance();
+      }
+    }
+    /**
+     * <code>.tensorflow.SavedAsset asset = 5;</code>
+     */
+    public Builder setAsset(org.tensorflow.proto.framework.SavedAsset value) {
+      if (assetBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        kind_ = value;
+        onChanged();
+      } else {
+        assetBuilder_.setMessage(value);
+      }
+      kindCase_ = 5;
+      return this;
+    }
+    /**
+     * <code>.tensorflow.SavedAsset asset = 5;</code>
+     */
+    public Builder setAsset(
+        org.tensorflow.proto.framework.SavedAsset.Builder builderForValue) {
+      if (assetBuilder_ == null) {
+        kind_ = builderForValue.build();
+        onChanged();
+      } else {
+        assetBuilder_.setMessage(builderForValue.build());
+      }
+      kindCase_ = 5;
+      return this;
+    }
+    /**
+     * <code>.tensorflow.SavedAsset asset = 5;</code>
+     */
+    public Builder mergeAsset(org.tensorflow.proto.framework.SavedAsset value) {
+      if (assetBuilder_ == null) {
+        if (kindCase_ == 5 &&
+            kind_ != org.tensorflow.proto.framework.SavedAsset.getDefaultInstance()) {
+          kind_ = org.tensorflow.proto.framework.SavedAsset.newBuilder((org.tensorflow.proto.framework.SavedAsset) kind_)
+              .mergeFrom(value).buildPartial();
+        } else {
+          kind_ = value;
+        }
+        onChanged();
+      } else {
+        if (kindCase_ == 5) {
+          assetBuilder_.mergeFrom(value);
+        }
+        assetBuilder_.setMessage(value);
+      }
+      kindCase_ = 5;
+      return this;
+    }
+    /**
+     * <code>.tensorflow.SavedAsset asset = 5;</code>
+     */
+    public Builder clearAsset() {
+      if (assetBuilder_ == null) {
+        if (kindCase_ == 5) {
+          kindCase_ = 0;
+          kind_ = null;
+          onChanged();
+        }
+      } else {
+        if (kindCase_ == 5) {
+          kindCase_ = 0;
+          kind_ = null;
+        }
+        assetBuilder_.clear();
+      }
+      return this;
+    }
+    /**
+     * <code>.tensorflow.SavedAsset asset = 5;</code>
+     */
+    public org.tensorflow.proto.framework.SavedAsset.Builder getAssetBuilder() {
+      return getAssetFieldBuilder().getBuilder();
+    }
+    /**
+     * <code>.tensorflow.SavedAsset asset = 5;</code>
+     */
+    public org.tensorflow.proto.framework.SavedAssetOrBuilder getAssetOrBuilder() {
+      if ((kindCase_ == 5) && (assetBuilder_ != null)) {
+        return assetBuilder_.getMessageOrBuilder();
+      } else {
+        if (kindCase_ == 5) {
+          return (org.tensorflow.proto.framework.SavedAsset) kind_;
+        }
+        return org.tensorflow.proto.framework.SavedAsset.getDefaultInstance();
+      }
+    }
+    /**
+     * <code>.tensorflow.SavedAsset asset = 5;</code>
+     */
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.SavedAsset, org.tensorflow.proto.framework.SavedAsset.Builder, org.tensorflow.proto.framework.SavedAssetOrBuilder> 
+        getAssetFieldBuilder() {
+      if (assetBuilder_ == null) {
+        if (!(kindCase_ == 5)) {
+          kind_ = org.tensorflow.proto.framework.SavedAsset.getDefaultInstance();
+        }
+        assetBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+            org.tensorflow.proto.framework.SavedAsset, org.tensorflow.proto.framework.SavedAsset.Builder, org.tensorflow.proto.framework.SavedAssetOrBuilder>(
+                (org.tensorflow.proto.framework.SavedAsset) kind_,
+                getParentForChildren(),
+                isClean());
+        kind_ = null;
+      }
+      kindCase_ = 5;
+      onChanged();;
+      return assetBuilder_;
+    }
+
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.SavedFunction, org.tensorflow.proto.framework.SavedFunction.Builder, org.tensorflow.proto.framework.SavedFunctionOrBuilder> functionBuilder_;
+    /**
+     * <code>.tensorflow.SavedFunction function = 6;</code>
+     */
+    public boolean hasFunction() {
+      return kindCase_ == 6;
+    }
+    /**
+     * <code>.tensorflow.SavedFunction function = 6;</code>
+     */
+    public org.tensorflow.proto.framework.SavedFunction getFunction() {
+      if (functionBuilder_ == null) {
+        if (kindCase_ == 6) {
+          return (org.tensorflow.proto.framework.SavedFunction) kind_;
+        }
+        return org.tensorflow.proto.framework.SavedFunction.getDefaultInstance();
+      } else {
+        if (kindCase_ == 6) {
+          return functionBuilder_.getMessage();
+        }
+        return org.tensorflow.proto.framework.SavedFunction.getDefaultInstance();
+      }
+    }
+    /**
+     * <code>.tensorflow.SavedFunction function = 6;</code>
+     */
+    public Builder setFunction(org.tensorflow.proto.framework.SavedFunction value) {
+      if (functionBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        kind_ = value;
+        onChanged();
+      } else {
+        functionBuilder_.setMessage(value);
+      }
+      kindCase_ = 6;
+      return this;
+    }
+    /**
+     * <code>.tensorflow.SavedFunction function = 6;</code>
+     */
+    public Builder setFunction(
+        org.tensorflow.proto.framework.SavedFunction.Builder builderForValue) {
+      if (functionBuilder_ == null) {
+        kind_ = builderForValue.build();
+        onChanged();
+      } else {
+        functionBuilder_.setMessage(builderForValue.build());
+      }
+      kindCase_ = 6;
+      return this;
+    }
+    /**
+     * <code>.tensorflow.SavedFunction function = 6;</code>
+     */
+    public Builder mergeFunction(org.tensorflow.proto.framework.SavedFunction value) {
+      if (functionBuilder_ == null) {
+        if (kindCase_ == 6 &&
+            kind_ != org.tensorflow.proto.framework.SavedFunction.getDefaultInstance()) {
+          kind_ = org.tensorflow.proto.framework.SavedFunction.newBuilder((org.tensorflow.proto.framework.SavedFunction) kind_)
+              .mergeFrom(value).buildPartial();
+        } else {
+          kind_ = value;
+        }
+        onChanged();
+      } else {
+        if (kindCase_ == 6) {
+          functionBuilder_.mergeFrom(value);
+        }
+        functionBuilder_.setMessage(value);
+      }
+      kindCase_ = 6;
+      return this;
+    }
+    /**
+     * <code>.tensorflow.SavedFunction function = 6;</code>
+     */
+    public Builder clearFunction() {
+      if (functionBuilder_ == null) {
+        if (kindCase_ == 6) {
+          kindCase_ = 0;
+          kind_ = null;
+          onChanged();
+        }
+      } else {
+        if (kindCase_ == 6) {
+          kindCase_ = 0;
+          kind_ = null;
+        }
+        functionBuilder_.clear();
+      }
+      return this;
+    }
+    /**
+     * <code>.tensorflow.SavedFunction function = 6;</code>
+     */
+    public org.tensorflow.proto.framework.SavedFunction.Builder getFunctionBuilder() {
+      return getFunctionFieldBuilder().getBuilder();
+    }
+    /**
+     * <code>.tensorflow.SavedFunction function = 6;</code>
+     */
+    public org.tensorflow.proto.framework.SavedFunctionOrBuilder getFunctionOrBuilder() {
+      if ((kindCase_ == 6) && (functionBuilder_ != null)) {
+        return functionBuilder_.getMessageOrBuilder();
+      } else {
+        if (kindCase_ == 6) {
+          return (org.tensorflow.proto.framework.SavedFunction) kind_;
+        }
+        return org.tensorflow.proto.framework.SavedFunction.getDefaultInstance();
+      }
+    }
+    /**
+     * <code>.tensorflow.SavedFunction function = 6;</code>
+     */
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.SavedFunction, org.tensorflow.proto.framework.SavedFunction.Builder, org.tensorflow.proto.framework.SavedFunctionOrBuilder> 
+        getFunctionFieldBuilder() {
+      if (functionBuilder_ == null) {
+        if (!(kindCase_ == 6)) {
+          kind_ = org.tensorflow.proto.framework.SavedFunction.getDefaultInstance();
+        }
+        functionBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+            org.tensorflow.proto.framework.SavedFunction, org.tensorflow.proto.framework.SavedFunction.Builder, org.tensorflow.proto.framework.SavedFunctionOrBuilder>(
+                (org.tensorflow.proto.framework.SavedFunction) kind_,
+                getParentForChildren(),
+                isClean());
+        kind_ = null;
+      }
+      kindCase_ = 6;
+      onChanged();;
+      return functionBuilder_;
+    }
+
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.SavedVariable, org.tensorflow.proto.framework.SavedVariable.Builder, org.tensorflow.proto.framework.SavedVariableOrBuilder> variableBuilder_;
+    /**
+     * <code>.tensorflow.SavedVariable variable = 7;</code>
+     */
+    public boolean hasVariable() {
+      return kindCase_ == 7;
+    }
+    /**
+     * <code>.tensorflow.SavedVariable variable = 7;</code>
+     */
+    public org.tensorflow.proto.framework.SavedVariable getVariable() {
+      if (variableBuilder_ == null) {
+        if (kindCase_ == 7) {
+          return (org.tensorflow.proto.framework.SavedVariable) kind_;
+        }
+        return org.tensorflow.proto.framework.SavedVariable.getDefaultInstance();
+      } else {
+        if (kindCase_ == 7) {
+          return variableBuilder_.getMessage();
+        }
+        return org.tensorflow.proto.framework.SavedVariable.getDefaultInstance();
+      }
+    }
+    /**
+     * <code>.tensorflow.SavedVariable variable = 7;</code>
+     */
+    public Builder setVariable(org.tensorflow.proto.framework.SavedVariable value) {
+      if (variableBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        kind_ = value;
+        onChanged();
+      } else {
+        variableBuilder_.setMessage(value);
+      }
+      kindCase_ = 7;
+      return this;
+    }
+    /**
+     * <code>.tensorflow.SavedVariable variable = 7;</code>
+     */
+    public Builder setVariable(
+        org.tensorflow.proto.framework.SavedVariable.Builder builderForValue) {
+      if (variableBuilder_ == null) {
+        kind_ = builderForValue.build();
+        onChanged();
+      } else {
+        variableBuilder_.setMessage(builderForValue.build());
+      }
+      kindCase_ = 7;
+      return this;
+    }
+    /**
+     * <code>.tensorflow.SavedVariable variable = 7;</code>
+     */
+    public Builder mergeVariable(org.tensorflow.proto.framework.SavedVariable value) {
+      if (variableBuilder_ == null) {
+        if (kindCase_ == 7 &&
+            kind_ != org.tensorflow.proto.framework.SavedVariable.getDefaultInstance()) {
+          kind_ = org.tensorflow.proto.framework.SavedVariable.newBuilder((org.tensorflow.proto.framework.SavedVariable) kind_)
+              .mergeFrom(value).buildPartial();
+        } else {
+          kind_ = value;
+        }
+        onChanged();
+      } else {
+        if (kindCase_ == 7) {
+          variableBuilder_.mergeFrom(value);
+        }
+        variableBuilder_.setMessage(value);
+      }
+      kindCase_ = 7;
+      return this;
+    }
+    /**
+     * <code>.tensorflow.SavedVariable variable = 7;</code>
+     */
+    public Builder clearVariable() {
+      if (variableBuilder_ == null) {
+        if (kindCase_ == 7) {
+          kindCase_ = 0;
+          kind_ = null;
+          onChanged();
+        }
+      } else {
+        if (kindCase_ == 7) {
+          kindCase_ = 0;
+          kind_ = null;
+        }
+        variableBuilder_.clear();
+      }
+      return this;
+    }
+    /**
+     * <code>.tensorflow.SavedVariable variable = 7;</code>
+     */
+    public org.tensorflow.proto.framework.SavedVariable.Builder getVariableBuilder() {
+      return getVariableFieldBuilder().getBuilder();
+    }
+    /**
+     * <code>.tensorflow.SavedVariable variable = 7;</code>
+     */
+    public org.tensorflow.proto.framework.SavedVariableOrBuilder getVariableOrBuilder() {
+      if ((kindCase_ == 7) && (variableBuilder_ != null)) {
+        return variableBuilder_.getMessageOrBuilder();
+      } else {
+        if (kindCase_ == 7) {
+          return (org.tensorflow.proto.framework.SavedVariable) kind_;
+        }
+        return org.tensorflow.proto.framework.SavedVariable.getDefaultInstance();
+      }
+    }
+    /**
+     * <code>.tensorflow.SavedVariable variable = 7;</code>
+     */
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.SavedVariable, org.tensorflow.proto.framework.SavedVariable.Builder, org.tensorflow.proto.framework.SavedVariableOrBuilder> 
+        getVariableFieldBuilder() {
+      if (variableBuilder_ == null) {
+        if (!(kindCase_ == 7)) {
+          kind_ = org.tensorflow.proto.framework.SavedVariable.getDefaultInstance();
+        }
+        variableBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+            org.tensorflow.proto.framework.SavedVariable, org.tensorflow.proto.framework.SavedVariable.Builder, org.tensorflow.proto.framework.SavedVariableOrBuilder>(
+                (org.tensorflow.proto.framework.SavedVariable) kind_,
+                getParentForChildren(),
+                isClean());
+        kind_ = null;
+      }
+      kindCase_ = 7;
+      onChanged();;
+      return variableBuilder_;
+    }
+
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.SavedBareConcreteFunction, org.tensorflow.proto.framework.SavedBareConcreteFunction.Builder, org.tensorflow.proto.framework.SavedBareConcreteFunctionOrBuilder> bareConcreteFunctionBuilder_;
+    /**
+     * <code>.tensorflow.SavedBareConcreteFunction bare_concrete_function = 8;</code>
+     */
+    public boolean hasBareConcreteFunction() {
+      return kindCase_ == 8;
+    }
+    /**
+     * <code>.tensorflow.SavedBareConcreteFunction bare_concrete_function = 8;</code>
+     */
+    public org.tensorflow.proto.framework.SavedBareConcreteFunction getBareConcreteFunction() {
+      if (bareConcreteFunctionBuilder_ == null) {
+        if (kindCase_ == 8) {
+          return (org.tensorflow.proto.framework.SavedBareConcreteFunction) kind_;
+        }
+        return org.tensorflow.proto.framework.SavedBareConcreteFunction.getDefaultInstance();
+      } else {
+        if (kindCase_ == 8) {
+          return bareConcreteFunctionBuilder_.getMessage();
+        }
+        return org.tensorflow.proto.framework.SavedBareConcreteFunction.getDefaultInstance();
+      }
+    }
+    /**
+     * <code>.tensorflow.SavedBareConcreteFunction bare_concrete_function = 8;</code>
+     */
+    public Builder setBareConcreteFunction(org.tensorflow.proto.framework.SavedBareConcreteFunction value) {
+      if (bareConcreteFunctionBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        kind_ = value;
+        onChanged();
+      } else {
+        bareConcreteFunctionBuilder_.setMessage(value);
+      }
+      kindCase_ = 8;
+      return this;
+    }
+    /**
+     * <code>.tensorflow.SavedBareConcreteFunction bare_concrete_function = 8;</code>
+     */
+    public Builder setBareConcreteFunction(
+        org.tensorflow.proto.framework.SavedBareConcreteFunction.Builder builderForValue) {
+      if (bareConcreteFunctionBuilder_ == null) {
+        kind_ = builderForValue.build();
+        onChanged();
+      } else {
+        bareConcreteFunctionBuilder_.setMessage(builderForValue.build());
+      }
+      kindCase_ = 8;
+      return this;
+    }
+    /**
+     * <code>.tensorflow.SavedBareConcreteFunction bare_concrete_function = 8;</code>
+     */
+    public Builder mergeBareConcreteFunction(org.tensorflow.proto.framework.SavedBareConcreteFunction value) {
+      if (bareConcreteFunctionBuilder_ == null) {
+        if (kindCase_ == 8 &&
+            kind_ != org.tensorflow.proto.framework.SavedBareConcreteFunction.getDefaultInstance()) {
+          kind_ = org.tensorflow.proto.framework.SavedBareConcreteFunction.newBuilder((org.tensorflow.proto.framework.SavedBareConcreteFunction) kind_)
+              .mergeFrom(value).buildPartial();
+        } else {
+          kind_ = value;
+        }
+        onChanged();
+      } else {
+        if (kindCase_ == 8) {
+          bareConcreteFunctionBuilder_.mergeFrom(value);
+        }
+        bareConcreteFunctionBuilder_.setMessage(value);
+      }
+      kindCase_ = 8;
+      return this;
+    }
+    /**
+     * <code>.tensorflow.SavedBareConcreteFunction bare_concrete_function = 8;</code>
+     */
+    public Builder clearBareConcreteFunction() {
+      if (bareConcreteFunctionBuilder_ == null) {
+        if (kindCase_ == 8) {
+          kindCase_ = 0;
+          kind_ = null;
+          onChanged();
+        }
+      } else {
+        if (kindCase_ == 8) {
+          kindCase_ = 0;
+          kind_ = null;
+        }
+        bareConcreteFunctionBuilder_.clear();
+      }
+      return this;
+    }
+    /**
+     * <code>.tensorflow.SavedBareConcreteFunction bare_concrete_function = 8;</code>
+     */
+    public org.tensorflow.proto.framework.SavedBareConcreteFunction.Builder getBareConcreteFunctionBuilder() {
+      return getBareConcreteFunctionFieldBuilder().getBuilder();
+    }
+    /**
+     * <code>.tensorflow.SavedBareConcreteFunction bare_concrete_function = 8;</code>
+     */
+    public org.tensorflow.proto.framework.SavedBareConcreteFunctionOrBuilder getBareConcreteFunctionOrBuilder() {
+      if ((kindCase_ == 8) && (bareConcreteFunctionBuilder_ != null)) {
+        return bareConcreteFunctionBuilder_.getMessageOrBuilder();
+      } else {
+        if (kindCase_ == 8) {
+          return (org.tensorflow.proto.framework.SavedBareConcreteFunction) kind_;
+        }
+        return org.tensorflow.proto.framework.SavedBareConcreteFunction.getDefaultInstance();
+      }
+    }
+    /**
+     * <code>.tensorflow.SavedBareConcreteFunction bare_concrete_function = 8;</code>
+     */
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.SavedBareConcreteFunction, org.tensorflow.proto.framework.SavedBareConcreteFunction.Builder, org.tensorflow.proto.framework.SavedBareConcreteFunctionOrBuilder> 
+        getBareConcreteFunctionFieldBuilder() {
+      if (bareConcreteFunctionBuilder_ == null) {
+        if (!(kindCase_ == 8)) {
+          kind_ = org.tensorflow.proto.framework.SavedBareConcreteFunction.getDefaultInstance();
+        }
+        bareConcreteFunctionBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+            org.tensorflow.proto.framework.SavedBareConcreteFunction, org.tensorflow.proto.framework.SavedBareConcreteFunction.Builder, org.tensorflow.proto.framework.SavedBareConcreteFunctionOrBuilder>(
+                (org.tensorflow.proto.framework.SavedBareConcreteFunction) kind_,
+                getParentForChildren(),
+                isClean());
+        kind_ = null;
+      }
+      kindCase_ = 8;
+      onChanged();;
+      return bareConcreteFunctionBuilder_;
+    }
+
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.SavedConstant, org.tensorflow.proto.framework.SavedConstant.Builder, org.tensorflow.proto.framework.SavedConstantOrBuilder> constantBuilder_;
+    /**
+     * <code>.tensorflow.SavedConstant constant = 9;</code>
+     */
+    public boolean hasConstant() {
+      return kindCase_ == 9;
+    }
+    /**
+     * <code>.tensorflow.SavedConstant constant = 9;</code>
+     */
+    public org.tensorflow.proto.framework.SavedConstant getConstant() {
+      if (constantBuilder_ == null) {
+        if (kindCase_ == 9) {
+          return (org.tensorflow.proto.framework.SavedConstant) kind_;
+        }
+        return org.tensorflow.proto.framework.SavedConstant.getDefaultInstance();
+      } else {
+        if (kindCase_ == 9) {
+          return constantBuilder_.getMessage();
+        }
+        return org.tensorflow.proto.framework.SavedConstant.getDefaultInstance();
+      }
+    }
+    /**
+     * <code>.tensorflow.SavedConstant constant = 9;</code>
+     */
+    public Builder setConstant(org.tensorflow.proto.framework.SavedConstant value) {
+      if (constantBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        kind_ = value;
+        onChanged();
+      } else {
+        constantBuilder_.setMessage(value);
+      }
+      kindCase_ = 9;
+      return this;
+    }
+    /**
+     * <code>.tensorflow.SavedConstant constant = 9;</code>
+     */
+    public Builder setConstant(
+        org.tensorflow.proto.framework.SavedConstant.Builder builderForValue) {
+      if (constantBuilder_ == null) {
+        kind_ = builderForValue.build();
+        onChanged();
+      } else {
+        constantBuilder_.setMessage(builderForValue.build());
+      }
+      kindCase_ = 9;
+      return this;
+    }
+    /**
+     * <code>.tensorflow.SavedConstant constant = 9;</code>
+     */
+    public Builder mergeConstant(org.tensorflow.proto.framework.SavedConstant value) {
+      if (constantBuilder_ == null) {
+        if (kindCase_ == 9 &&
+            kind_ != org.tensorflow.proto.framework.SavedConstant.getDefaultInstance()) {
+          kind_ = org.tensorflow.proto.framework.SavedConstant.newBuilder((org.tensorflow.proto.framework.SavedConstant) kind_)
+              .mergeFrom(value).buildPartial();
+        } else {
+          kind_ = value;
+        }
+        onChanged();
+      } else {
+        if (kindCase_ == 9) {
+          constantBuilder_.mergeFrom(value);
+        }
+        constantBuilder_.setMessage(value);
+      }
+      kindCase_ = 9;
+      return this;
+    }
+    /**
+     * <code>.tensorflow.SavedConstant constant = 9;</code>
+     */
+    public Builder clearConstant() {
+      if (constantBuilder_ == null) {
+        if (kindCase_ == 9) {
+          kindCase_ = 0;
+          kind_ = null;
+          onChanged();
+        }
+      } else {
+        if (kindCase_ == 9) {
+          kindCase_ = 0;
+          kind_ = null;
+        }
+        constantBuilder_.clear();
+      }
+      return this;
+    }
+    /**
+     * <code>.tensorflow.SavedConstant constant = 9;</code>
+     */
+    public org.tensorflow.proto.framework.SavedConstant.Builder getConstantBuilder() {
+      return getConstantFieldBuilder().getBuilder();
+    }
+    /**
+     * <code>.tensorflow.SavedConstant constant = 9;</code>
+     */
+    public org.tensorflow.proto.framework.SavedConstantOrBuilder getConstantOrBuilder() {
+      if ((kindCase_ == 9) && (constantBuilder_ != null)) {
+        return constantBuilder_.getMessageOrBuilder();
+      } else {
+        if (kindCase_ == 9) {
+          return (org.tensorflow.proto.framework.SavedConstant) kind_;
+        }
+        return org.tensorflow.proto.framework.SavedConstant.getDefaultInstance();
+      }
+    }
+    /**
+     * <code>.tensorflow.SavedConstant constant = 9;</code>
+     */
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.SavedConstant, org.tensorflow.proto.framework.SavedConstant.Builder, org.tensorflow.proto.framework.SavedConstantOrBuilder> 
+        getConstantFieldBuilder() {
+      if (constantBuilder_ == null) {
+        if (!(kindCase_ == 9)) {
+          kind_ = org.tensorflow.proto.framework.SavedConstant.getDefaultInstance();
+        }
+        constantBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+            org.tensorflow.proto.framework.SavedConstant, org.tensorflow.proto.framework.SavedConstant.Builder, org.tensorflow.proto.framework.SavedConstantOrBuilder>(
+                (org.tensorflow.proto.framework.SavedConstant) kind_,
+                getParentForChildren(),
+                isClean());
+        kind_ = null;
+      }
+      kindCase_ = 9;
+      onChanged();;
+      return constantBuilder_;
+    }
+
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.SavedResource, org.tensorflow.proto.framework.SavedResource.Builder, org.tensorflow.proto.framework.SavedResourceOrBuilder> resourceBuilder_;
+    /**
+     * <code>.tensorflow.SavedResource resource = 10;</code>
+     */
+    public boolean hasResource() {
+      return kindCase_ == 10;
+    }
+    /**
+     * <code>.tensorflow.SavedResource resource = 10;</code>
+     */
+    public org.tensorflow.proto.framework.SavedResource getResource() {
+      if (resourceBuilder_ == null) {
+        if (kindCase_ == 10) {
+          return (org.tensorflow.proto.framework.SavedResource) kind_;
+        }
+        return org.tensorflow.proto.framework.SavedResource.getDefaultInstance();
+      } else {
+        if (kindCase_ == 10) {
+          return resourceBuilder_.getMessage();
+        }
+        return org.tensorflow.proto.framework.SavedResource.getDefaultInstance();
+      }
+    }
+    /**
+     * <code>.tensorflow.SavedResource resource = 10;</code>
+     */
+    public Builder setResource(org.tensorflow.proto.framework.SavedResource value) {
+      if (resourceBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        kind_ = value;
+        onChanged();
+      } else {
+        resourceBuilder_.setMessage(value);
+      }
+      kindCase_ = 10;
+      return this;
+    }
+    /**
+     * <code>.tensorflow.SavedResource resource = 10;</code>
+     */
+    public Builder setResource(
+        org.tensorflow.proto.framework.SavedResource.Builder builderForValue) {
+      if (resourceBuilder_ == null) {
+        kind_ = builderForValue.build();
+        onChanged();
+      } else {
+        resourceBuilder_.setMessage(builderForValue.build());
+      }
+      kindCase_ = 10;
+      return this;
+    }
+    /**
+     * <code>.tensorflow.SavedResource resource = 10;</code>
+     */
+    public Builder mergeResource(org.tensorflow.proto.framework.SavedResource value) {
+      if (resourceBuilder_ == null) {
+        if (kindCase_ == 10 &&
+            kind_ != org.tensorflow.proto.framework.SavedResource.getDefaultInstance()) {
+          kind_ = org.tensorflow.proto.framework.SavedResource.newBuilder((org.tensorflow.proto.framework.SavedResource) kind_)
+              .mergeFrom(value).buildPartial();
+        } else {
+          kind_ = value;
+        }
+        onChanged();
+      } else {
+        if (kindCase_ == 10) {
+          resourceBuilder_.mergeFrom(value);
+        }
+        resourceBuilder_.setMessage(value);
+      }
+      kindCase_ = 10;
+      return this;
+    }
+    /**
+     * <code>.tensorflow.SavedResource resource = 10;</code>
+     */
+    public Builder clearResource() {
+      if (resourceBuilder_ == null) {
+        if (kindCase_ == 10) {
+          kindCase_ = 0;
+          kind_ = null;
+          onChanged();
+        }
+      } else {
+        if (kindCase_ == 10) {
+          kindCase_ = 0;
+          kind_ = null;
+        }
+        resourceBuilder_.clear();
+      }
+      return this;
+    }
+    /**
+     * <code>.tensorflow.SavedResource resource = 10;</code>
+     */
+    public org.tensorflow.proto.framework.SavedResource.Builder getResourceBuilder() {
+      return getResourceFieldBuilder().getBuilder();
+    }
+    /**
+     * <code>.tensorflow.SavedResource resource = 10;</code>
+     */
+    public org.tensorflow.proto.framework.SavedResourceOrBuilder getResourceOrBuilder() {
+      if ((kindCase_ == 10) && (resourceBuilder_ != null)) {
+        return resourceBuilder_.getMessageOrBuilder();
+      } else {
+        if (kindCase_ == 10) {
+          return (org.tensorflow.proto.framework.SavedResource) kind_;
+        }
+        return org.tensorflow.proto.framework.SavedResource.getDefaultInstance();
+      }
+    }
+    /**
+     * <code>.tensorflow.SavedResource resource = 10;</code>
+     */
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.SavedResource, org.tensorflow.proto.framework.SavedResource.Builder, org.tensorflow.proto.framework.SavedResourceOrBuilder> 
+        getResourceFieldBuilder() {
+      if (resourceBuilder_ == null) {
+        if (!(kindCase_ == 10)) {
+          kind_ = org.tensorflow.proto.framework.SavedResource.getDefaultInstance();
+        }
+        resourceBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+            org.tensorflow.proto.framework.SavedResource, org.tensorflow.proto.framework.SavedResource.Builder, org.tensorflow.proto.framework.SavedResourceOrBuilder>(
+                (org.tensorflow.proto.framework.SavedResource) kind_,
+                getParentForChildren(),
+                isClean());
+        kind_ = null;
+      }
+      kindCase_ = 10;
+      onChanged();;
+      return resourceBuilder_;
+    }
+
+    private com.google.protobuf.MapField<
+        java.lang.String, org.tensorflow.proto.framework.SaveableObject> saveableObjects_;
+    private com.google.protobuf.MapField<java.lang.String, org.tensorflow.proto.framework.SaveableObject>
+    internalGetSaveableObjects() {
+      if (saveableObjects_ == null) {
+        return com.google.protobuf.MapField.emptyMapField(
+            SaveableObjectsDefaultEntryHolder.defaultEntry);
+      }
+      return saveableObjects_;
+    }
+    private com.google.protobuf.MapField<java.lang.String, org.tensorflow.proto.framework.SaveableObject>
+    internalGetMutableSaveableObjects() {
+      onChanged();;
+      if (saveableObjects_ == null) {
+        saveableObjects_ = com.google.protobuf.MapField.newMapField(
+            SaveableObjectsDefaultEntryHolder.defaultEntry);
+      }
+      if (!saveableObjects_.isMutable()) {
+        saveableObjects_ = saveableObjects_.copy();
+      }
+      return saveableObjects_;
+    }
+
+    public int getSaveableObjectsCount() {
+      return internalGetSaveableObjects().getMap().size();
+    }
+    /**
+     * <code>map&lt;string, .tensorflow.SaveableObject&gt; saveable_objects = 11;</code>
+     */
+
+    public boolean containsSaveableObjects(
+        java.lang.String key) {
+      if (key == null) { throw new java.lang.NullPointerException(); }
+      return internalGetSaveableObjects().getMap().containsKey(key);
+    }
+    /**
+     * Use {@link #getSaveableObjectsMap()} instead.
+     */
+    @java.lang.Deprecated
+    public java.util.Map<java.lang.String, org.tensorflow.proto.framework.SaveableObject> getSaveableObjects() {
+      return getSaveableObjectsMap();
+    }
+    /**
+     * <code>map&lt;string, .tensorflow.SaveableObject&gt; saveable_objects = 11;</code>
+     */
+
+    public java.util.Map<java.lang.String, org.tensorflow.proto.framework.SaveableObject> getSaveableObjectsMap() {
+      return internalGetSaveableObjects().getMap();
+    }
+    /**
+     * <code>map&lt;string, .tensorflow.SaveableObject&gt; saveable_objects = 11;</code>
+     */
+
+    public org.tensorflow.proto.framework.SaveableObject getSaveableObjectsOrDefault(
+        java.lang.String key,
+        org.tensorflow.proto.framework.SaveableObject defaultValue) {
+      if (key == null) { throw new java.lang.NullPointerException(); }
+      java.util.Map<java.lang.String, org.tensorflow.proto.framework.SaveableObject> map =
+          internalGetSaveableObjects().getMap();
+      return map.containsKey(key) ? map.get(key) : defaultValue;
+    }
+    /**
+     * <code>map&lt;string, .tensorflow.SaveableObject&gt; saveable_objects = 11;</code>
+     */
+
+    public org.tensorflow.proto.framework.SaveableObject getSaveableObjectsOrThrow(
+        java.lang.String key) {
+      if (key == null) { throw new java.lang.NullPointerException(); }
+      java.util.Map<java.lang.String, org.tensorflow.proto.framework.SaveableObject> map =
+          internalGetSaveableObjects().getMap();
+      if (!map.containsKey(key)) {
+        throw new java.lang.IllegalArgumentException();
+      }
+      return map.get(key);
+    }
+
+    public Builder clearSaveableObjects() {
+      internalGetMutableSaveableObjects().getMutableMap()
+          .clear();
+      return this;
+    }
+    /**
+     * <code>map&lt;string, .tensorflow.SaveableObject&gt; saveable_objects = 11;</code>
+     */
+
+    public Builder removeSaveableObjects(
+        java.lang.String key) {
+      if (key == null) { throw new java.lang.NullPointerException(); }
+      internalGetMutableSaveableObjects().getMutableMap()
+          .remove(key);
+      return this;
+    }
+    /**
+     * Use alternate mutation accessors instead.
+     */
+    @java.lang.Deprecated
+    public java.util.Map<java.lang.String, org.tensorflow.proto.framework.SaveableObject>
+    getMutableSaveableObjects() {
+      return internalGetMutableSaveableObjects().getMutableMap();
+    }
+    /**
+     * <code>map&lt;string, .tensorflow.SaveableObject&gt; saveable_objects = 11;</code>
+     */
+    public Builder putSaveableObjects(
+        java.lang.String key,
+        org.tensorflow.proto.framework.SaveableObject value) {
+      if (key == null) { throw new java.lang.NullPointerException(); }
+      if (value == null) { throw new java.lang.NullPointerException(); }
+      internalGetMutableSaveableObjects().getMutableMap()
+          .put(key, value);
+      return this;
+    }
+    /**
+     * <code>map&lt;string, .tensorflow.SaveableObject&gt; saveable_objects = 11;</code>
+     */
+
+    public Builder putAllSaveableObjects(
+        java.util.Map<java.lang.String, org.tensorflow.proto.framework.SaveableObject> values) {
+      internalGetMutableSaveableObjects().getMutableMap()
+          .putAll(values);
+      return this;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.SavedObject)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.SavedObject)
+  private static final org.tensorflow.proto.framework.SavedObject DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.SavedObject();
+  }
+
+  public static org.tensorflow.proto.framework.SavedObject getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<SavedObject>
+      PARSER = new com.google.protobuf.AbstractParser<SavedObject>() {
+    @java.lang.Override
+    public SavedObject parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new SavedObject(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<SavedObject> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<SavedObject> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.framework.SavedObject getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedObjectGraph.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedObjectGraph.java
new file mode 100644
index 00000000000..40a1e497db4
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedObjectGraph.java
@@ -0,0 +1,1231 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/saved_object_graph.proto
+
+package org.tensorflow.proto.framework;
+
+/**
+ * Protobuf type {@code tensorflow.SavedObjectGraph}
+ */
+public  final class SavedObjectGraph extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.SavedObjectGraph)
+    SavedObjectGraphOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use SavedObjectGraph.newBuilder() to construct.
+  private SavedObjectGraph(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private SavedObjectGraph() {
+    nodes_ = java.util.Collections.emptyList();
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new SavedObjectGraph();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private SavedObjectGraph(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    int mutable_bitField0_ = 0;
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 10: {
+            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
+              nodes_ = new java.util.ArrayList<org.tensorflow.proto.framework.SavedObject>();
+              mutable_bitField0_ |= 0x00000001;
+            }
+            nodes_.add(
+                input.readMessage(org.tensorflow.proto.framework.SavedObject.parser(), extensionRegistry));
+            break;
+          }
+          case 18: {
+            if (!((mutable_bitField0_ & 0x00000002) != 0)) {
+              concreteFunctions_ = com.google.protobuf.MapField.newMapField(
+                  ConcreteFunctionsDefaultEntryHolder.defaultEntry);
+              mutable_bitField0_ |= 0x00000002;
+            }
+            com.google.protobuf.MapEntry<java.lang.String, org.tensorflow.proto.framework.SavedConcreteFunction>
+            concreteFunctions__ = input.readMessage(
+                ConcreteFunctionsDefaultEntryHolder.defaultEntry.getParserForType(), extensionRegistry);
+            concreteFunctions_.getMutableMap().put(
+                concreteFunctions__.getKey(), concreteFunctions__.getValue());
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      if (((mutable_bitField0_ & 0x00000001) != 0)) {
+        nodes_ = java.util.Collections.unmodifiableList(nodes_);
+      }
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.framework.SavedObjectGraphProtos.internal_static_tensorflow_SavedObjectGraph_descriptor;
+  }
+
+  @SuppressWarnings({"rawtypes"})
+  @java.lang.Override
+  protected com.google.protobuf.MapField internalGetMapField(
+      int number) {
+    switch (number) {
+      case 2:
+        return internalGetConcreteFunctions();
+      default:
+        throw new RuntimeException(
+            "Invalid map field number: " + number);
+    }
+  }
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.framework.SavedObjectGraphProtos.internal_static_tensorflow_SavedObjectGraph_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.framework.SavedObjectGraph.class, org.tensorflow.proto.framework.SavedObjectGraph.Builder.class);
+  }
+
+  public static final int NODES_FIELD_NUMBER = 1;
+  private java.util.List<org.tensorflow.proto.framework.SavedObject> nodes_;
+  /**
+   * <pre>
+   * Flattened list of objects in the object graph.
+   * The position of the object in this list indicates its id.
+   * Nodes[0] is considered the root node.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.SavedObject nodes = 1;</code>
+   */
+  public java.util.List<org.tensorflow.proto.framework.SavedObject> getNodesList() {
+    return nodes_;
+  }
+  /**
+   * <pre>
+   * Flattened list of objects in the object graph.
+   * The position of the object in this list indicates its id.
+   * Nodes[0] is considered the root node.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.SavedObject nodes = 1;</code>
+   */
+  public java.util.List<? extends org.tensorflow.proto.framework.SavedObjectOrBuilder> 
+      getNodesOrBuilderList() {
+    return nodes_;
+  }
+  /**
+   * <pre>
+   * Flattened list of objects in the object graph.
+   * The position of the object in this list indicates its id.
+   * Nodes[0] is considered the root node.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.SavedObject nodes = 1;</code>
+   */
+  public int getNodesCount() {
+    return nodes_.size();
+  }
+  /**
+   * <pre>
+   * Flattened list of objects in the object graph.
+   * The position of the object in this list indicates its id.
+   * Nodes[0] is considered the root node.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.SavedObject nodes = 1;</code>
+   */
+  public org.tensorflow.proto.framework.SavedObject getNodes(int index) {
+    return nodes_.get(index);
+  }
+  /**
+   * <pre>
+   * Flattened list of objects in the object graph.
+   * The position of the object in this list indicates its id.
+   * Nodes[0] is considered the root node.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.SavedObject nodes = 1;</code>
+   */
+  public org.tensorflow.proto.framework.SavedObjectOrBuilder getNodesOrBuilder(
+      int index) {
+    return nodes_.get(index);
+  }
+
+  public static final int CONCRETE_FUNCTIONS_FIELD_NUMBER = 2;
+  private static final class ConcreteFunctionsDefaultEntryHolder {
+    static final com.google.protobuf.MapEntry<
+        java.lang.String, org.tensorflow.proto.framework.SavedConcreteFunction> defaultEntry =
+            com.google.protobuf.MapEntry
+            .<java.lang.String, org.tensorflow.proto.framework.SavedConcreteFunction>newDefaultInstance(
+                org.tensorflow.proto.framework.SavedObjectGraphProtos.internal_static_tensorflow_SavedObjectGraph_ConcreteFunctionsEntry_descriptor, 
+                com.google.protobuf.WireFormat.FieldType.STRING,
+                "",
+                com.google.protobuf.WireFormat.FieldType.MESSAGE,
+                org.tensorflow.proto.framework.SavedConcreteFunction.getDefaultInstance());
+  }
+  private com.google.protobuf.MapField<
+      java.lang.String, org.tensorflow.proto.framework.SavedConcreteFunction> concreteFunctions_;
+  private com.google.protobuf.MapField<java.lang.String, org.tensorflow.proto.framework.SavedConcreteFunction>
+  internalGetConcreteFunctions() {
+    if (concreteFunctions_ == null) {
+      return com.google.protobuf.MapField.emptyMapField(
+          ConcreteFunctionsDefaultEntryHolder.defaultEntry);
+    }
+    return concreteFunctions_;
+  }
+
+  public int getConcreteFunctionsCount() {
+    return internalGetConcreteFunctions().getMap().size();
+  }
+  /**
+   * <pre>
+   * Information about captures and output structures in concrete functions.
+   * Referenced from SavedBareConcreteFunction and SavedFunction.
+   * </pre>
+   *
+   * <code>map&lt;string, .tensorflow.SavedConcreteFunction&gt; concrete_functions = 2;</code>
+   */
+
+  public boolean containsConcreteFunctions(
+      java.lang.String key) {
+    if (key == null) { throw new java.lang.NullPointerException(); }
+    return internalGetConcreteFunctions().getMap().containsKey(key);
+  }
+  /**
+   * Use {@link #getConcreteFunctionsMap()} instead.
+   */
+  @java.lang.Deprecated
+  public java.util.Map<java.lang.String, org.tensorflow.proto.framework.SavedConcreteFunction> getConcreteFunctions() {
+    return getConcreteFunctionsMap();
+  }
+  /**
+   * <pre>
+   * Information about captures and output structures in concrete functions.
+   * Referenced from SavedBareConcreteFunction and SavedFunction.
+   * </pre>
+   *
+   * <code>map&lt;string, .tensorflow.SavedConcreteFunction&gt; concrete_functions = 2;</code>
+   */
+
+  public java.util.Map<java.lang.String, org.tensorflow.proto.framework.SavedConcreteFunction> getConcreteFunctionsMap() {
+    return internalGetConcreteFunctions().getMap();
+  }
+  /**
+   * <pre>
+   * Information about captures and output structures in concrete functions.
+   * Referenced from SavedBareConcreteFunction and SavedFunction.
+   * </pre>
+   *
+   * <code>map&lt;string, .tensorflow.SavedConcreteFunction&gt; concrete_functions = 2;</code>
+   */
+
+  public org.tensorflow.proto.framework.SavedConcreteFunction getConcreteFunctionsOrDefault(
+      java.lang.String key,
+      org.tensorflow.proto.framework.SavedConcreteFunction defaultValue) {
+    if (key == null) { throw new java.lang.NullPointerException(); }
+    java.util.Map<java.lang.String, org.tensorflow.proto.framework.SavedConcreteFunction> map =
+        internalGetConcreteFunctions().getMap();
+    return map.containsKey(key) ? map.get(key) : defaultValue;
+  }
+  /**
+   * <pre>
+   * Information about captures and output structures in concrete functions.
+   * Referenced from SavedBareConcreteFunction and SavedFunction.
+   * </pre>
+   *
+   * <code>map&lt;string, .tensorflow.SavedConcreteFunction&gt; concrete_functions = 2;</code>
+   */
+
+  public org.tensorflow.proto.framework.SavedConcreteFunction getConcreteFunctionsOrThrow(
+      java.lang.String key) {
+    if (key == null) { throw new java.lang.NullPointerException(); }
+    java.util.Map<java.lang.String, org.tensorflow.proto.framework.SavedConcreteFunction> map =
+        internalGetConcreteFunctions().getMap();
+    if (!map.containsKey(key)) {
+      throw new java.lang.IllegalArgumentException();
+    }
+    return map.get(key);
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    for (int i = 0; i < nodes_.size(); i++) {
+      output.writeMessage(1, nodes_.get(i));
+    }
+    com.google.protobuf.GeneratedMessageV3
+      .serializeStringMapTo(
+        output,
+        internalGetConcreteFunctions(),
+        ConcreteFunctionsDefaultEntryHolder.defaultEntry,
+        2);
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    for (int i = 0; i < nodes_.size(); i++) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(1, nodes_.get(i));
+    }
+    for (java.util.Map.Entry<java.lang.String, org.tensorflow.proto.framework.SavedConcreteFunction> entry
+         : internalGetConcreteFunctions().getMap().entrySet()) {
+      com.google.protobuf.MapEntry<java.lang.String, org.tensorflow.proto.framework.SavedConcreteFunction>
+      concreteFunctions__ = ConcreteFunctionsDefaultEntryHolder.defaultEntry.newBuilderForType()
+          .setKey(entry.getKey())
+          .setValue(entry.getValue())
+          .build();
+      size += com.google.protobuf.CodedOutputStream
+          .computeMessageSize(2, concreteFunctions__);
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.framework.SavedObjectGraph)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.framework.SavedObjectGraph other = (org.tensorflow.proto.framework.SavedObjectGraph) obj;
+
+    if (!getNodesList()
+        .equals(other.getNodesList())) return false;
+    if (!internalGetConcreteFunctions().equals(
+        other.internalGetConcreteFunctions())) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    if (getNodesCount() > 0) {
+      hash = (37 * hash) + NODES_FIELD_NUMBER;
+      hash = (53 * hash) + getNodesList().hashCode();
+    }
+    if (!internalGetConcreteFunctions().getMap().isEmpty()) {
+      hash = (37 * hash) + CONCRETE_FUNCTIONS_FIELD_NUMBER;
+      hash = (53 * hash) + internalGetConcreteFunctions().hashCode();
+    }
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.framework.SavedObjectGraph parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.SavedObjectGraph parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.SavedObjectGraph parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.SavedObjectGraph parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.SavedObjectGraph parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.SavedObjectGraph parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.SavedObjectGraph parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.SavedObjectGraph parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.SavedObjectGraph parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.SavedObjectGraph parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.SavedObjectGraph parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.SavedObjectGraph parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.framework.SavedObjectGraph prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * Protobuf type {@code tensorflow.SavedObjectGraph}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.SavedObjectGraph)
+      org.tensorflow.proto.framework.SavedObjectGraphOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.SavedObjectGraphProtos.internal_static_tensorflow_SavedObjectGraph_descriptor;
+    }
+
+    @SuppressWarnings({"rawtypes"})
+    protected com.google.protobuf.MapField internalGetMapField(
+        int number) {
+      switch (number) {
+        case 2:
+          return internalGetConcreteFunctions();
+        default:
+          throw new RuntimeException(
+              "Invalid map field number: " + number);
+      }
+    }
+    @SuppressWarnings({"rawtypes"})
+    protected com.google.protobuf.MapField internalGetMutableMapField(
+        int number) {
+      switch (number) {
+        case 2:
+          return internalGetMutableConcreteFunctions();
+        default:
+          throw new RuntimeException(
+              "Invalid map field number: " + number);
+      }
+    }
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.SavedObjectGraphProtos.internal_static_tensorflow_SavedObjectGraph_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.SavedObjectGraph.class, org.tensorflow.proto.framework.SavedObjectGraph.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.framework.SavedObjectGraph.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+        getNodesFieldBuilder();
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      if (nodesBuilder_ == null) {
+        nodes_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000001);
+      } else {
+        nodesBuilder_.clear();
+      }
+      internalGetMutableConcreteFunctions().clear();
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.framework.SavedObjectGraphProtos.internal_static_tensorflow_SavedObjectGraph_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.SavedObjectGraph getDefaultInstanceForType() {
+      return org.tensorflow.proto.framework.SavedObjectGraph.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.SavedObjectGraph build() {
+      org.tensorflow.proto.framework.SavedObjectGraph result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.SavedObjectGraph buildPartial() {
+      org.tensorflow.proto.framework.SavedObjectGraph result = new org.tensorflow.proto.framework.SavedObjectGraph(this);
+      int from_bitField0_ = bitField0_;
+      if (nodesBuilder_ == null) {
+        if (((bitField0_ & 0x00000001) != 0)) {
+          nodes_ = java.util.Collections.unmodifiableList(nodes_);
+          bitField0_ = (bitField0_ & ~0x00000001);
+        }
+        result.nodes_ = nodes_;
+      } else {
+        result.nodes_ = nodesBuilder_.build();
+      }
+      result.concreteFunctions_ = internalGetConcreteFunctions();
+      result.concreteFunctions_.makeImmutable();
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.framework.SavedObjectGraph) {
+        return mergeFrom((org.tensorflow.proto.framework.SavedObjectGraph)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.framework.SavedObjectGraph other) {
+      if (other == org.tensorflow.proto.framework.SavedObjectGraph.getDefaultInstance()) return this;
+      if (nodesBuilder_ == null) {
+        if (!other.nodes_.isEmpty()) {
+          if (nodes_.isEmpty()) {
+            nodes_ = other.nodes_;
+            bitField0_ = (bitField0_ & ~0x00000001);
+          } else {
+            ensureNodesIsMutable();
+            nodes_.addAll(other.nodes_);
+          }
+          onChanged();
+        }
+      } else {
+        if (!other.nodes_.isEmpty()) {
+          if (nodesBuilder_.isEmpty()) {
+            nodesBuilder_.dispose();
+            nodesBuilder_ = null;
+            nodes_ = other.nodes_;
+            bitField0_ = (bitField0_ & ~0x00000001);
+            nodesBuilder_ = 
+              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
+                 getNodesFieldBuilder() : null;
+          } else {
+            nodesBuilder_.addAllMessages(other.nodes_);
+          }
+        }
+      }
+      internalGetMutableConcreteFunctions().mergeFrom(
+          other.internalGetConcreteFunctions());
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.framework.SavedObjectGraph parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.framework.SavedObjectGraph) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+    private int bitField0_;
+
+    private java.util.List<org.tensorflow.proto.framework.SavedObject> nodes_ =
+      java.util.Collections.emptyList();
+    private void ensureNodesIsMutable() {
+      if (!((bitField0_ & 0x00000001) != 0)) {
+        nodes_ = new java.util.ArrayList<org.tensorflow.proto.framework.SavedObject>(nodes_);
+        bitField0_ |= 0x00000001;
+       }
+    }
+
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.framework.SavedObject, org.tensorflow.proto.framework.SavedObject.Builder, org.tensorflow.proto.framework.SavedObjectOrBuilder> nodesBuilder_;
+
+    /**
+     * <pre>
+     * Flattened list of objects in the object graph.
+     * The position of the object in this list indicates its id.
+     * Nodes[0] is considered the root node.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.SavedObject nodes = 1;</code>
+     */
+    public java.util.List<org.tensorflow.proto.framework.SavedObject> getNodesList() {
+      if (nodesBuilder_ == null) {
+        return java.util.Collections.unmodifiableList(nodes_);
+      } else {
+        return nodesBuilder_.getMessageList();
+      }
+    }
+    /**
+     * <pre>
+     * Flattened list of objects in the object graph.
+     * The position of the object in this list indicates its id.
+     * Nodes[0] is considered the root node.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.SavedObject nodes = 1;</code>
+     */
+    public int getNodesCount() {
+      if (nodesBuilder_ == null) {
+        return nodes_.size();
+      } else {
+        return nodesBuilder_.getCount();
+      }
+    }
+    /**
+     * <pre>
+     * Flattened list of objects in the object graph.
+     * The position of the object in this list indicates its id.
+     * Nodes[0] is considered the root node.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.SavedObject nodes = 1;</code>
+     */
+    public org.tensorflow.proto.framework.SavedObject getNodes(int index) {
+      if (nodesBuilder_ == null) {
+        return nodes_.get(index);
+      } else {
+        return nodesBuilder_.getMessage(index);
+      }
+    }
+    /**
+     * <pre>
+     * Flattened list of objects in the object graph.
+     * The position of the object in this list indicates its id.
+     * Nodes[0] is considered the root node.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.SavedObject nodes = 1;</code>
+     */
+    public Builder setNodes(
+        int index, org.tensorflow.proto.framework.SavedObject value) {
+      if (nodesBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureNodesIsMutable();
+        nodes_.set(index, value);
+        onChanged();
+      } else {
+        nodesBuilder_.setMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Flattened list of objects in the object graph.
+     * The position of the object in this list indicates its id.
+     * Nodes[0] is considered the root node.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.SavedObject nodes = 1;</code>
+     */
+    public Builder setNodes(
+        int index, org.tensorflow.proto.framework.SavedObject.Builder builderForValue) {
+      if (nodesBuilder_ == null) {
+        ensureNodesIsMutable();
+        nodes_.set(index, builderForValue.build());
+        onChanged();
+      } else {
+        nodesBuilder_.setMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Flattened list of objects in the object graph.
+     * The position of the object in this list indicates its id.
+     * Nodes[0] is considered the root node.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.SavedObject nodes = 1;</code>
+     */
+    public Builder addNodes(org.tensorflow.proto.framework.SavedObject value) {
+      if (nodesBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureNodesIsMutable();
+        nodes_.add(value);
+        onChanged();
+      } else {
+        nodesBuilder_.addMessage(value);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Flattened list of objects in the object graph.
+     * The position of the object in this list indicates its id.
+     * Nodes[0] is considered the root node.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.SavedObject nodes = 1;</code>
+     */
+    public Builder addNodes(
+        int index, org.tensorflow.proto.framework.SavedObject value) {
+      if (nodesBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureNodesIsMutable();
+        nodes_.add(index, value);
+        onChanged();
+      } else {
+        nodesBuilder_.addMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Flattened list of objects in the object graph.
+     * The position of the object in this list indicates its id.
+     * Nodes[0] is considered the root node.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.SavedObject nodes = 1;</code>
+     */
+    public Builder addNodes(
+        org.tensorflow.proto.framework.SavedObject.Builder builderForValue) {
+      if (nodesBuilder_ == null) {
+        ensureNodesIsMutable();
+        nodes_.add(builderForValue.build());
+        onChanged();
+      } else {
+        nodesBuilder_.addMessage(builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Flattened list of objects in the object graph.
+     * The position of the object in this list indicates its id.
+     * Nodes[0] is considered the root node.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.SavedObject nodes = 1;</code>
+     */
+    public Builder addNodes(
+        int index, org.tensorflow.proto.framework.SavedObject.Builder builderForValue) {
+      if (nodesBuilder_ == null) {
+        ensureNodesIsMutable();
+        nodes_.add(index, builderForValue.build());
+        onChanged();
+      } else {
+        nodesBuilder_.addMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Flattened list of objects in the object graph.
+     * The position of the object in this list indicates its id.
+     * Nodes[0] is considered the root node.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.SavedObject nodes = 1;</code>
+     */
+    public Builder addAllNodes(
+        java.lang.Iterable<? extends org.tensorflow.proto.framework.SavedObject> values) {
+      if (nodesBuilder_ == null) {
+        ensureNodesIsMutable();
+        com.google.protobuf.AbstractMessageLite.Builder.addAll(
+            values, nodes_);
+        onChanged();
+      } else {
+        nodesBuilder_.addAllMessages(values);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Flattened list of objects in the object graph.
+     * The position of the object in this list indicates its id.
+     * Nodes[0] is considered the root node.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.SavedObject nodes = 1;</code>
+     */
+    public Builder clearNodes() {
+      if (nodesBuilder_ == null) {
+        nodes_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000001);
+        onChanged();
+      } else {
+        nodesBuilder_.clear();
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Flattened list of objects in the object graph.
+     * The position of the object in this list indicates its id.
+     * Nodes[0] is considered the root node.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.SavedObject nodes = 1;</code>
+     */
+    public Builder removeNodes(int index) {
+      if (nodesBuilder_ == null) {
+        ensureNodesIsMutable();
+        nodes_.remove(index);
+        onChanged();
+      } else {
+        nodesBuilder_.remove(index);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Flattened list of objects in the object graph.
+     * The position of the object in this list indicates its id.
+     * Nodes[0] is considered the root node.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.SavedObject nodes = 1;</code>
+     */
+    public org.tensorflow.proto.framework.SavedObject.Builder getNodesBuilder(
+        int index) {
+      return getNodesFieldBuilder().getBuilder(index);
+    }
+    /**
+     * <pre>
+     * Flattened list of objects in the object graph.
+     * The position of the object in this list indicates its id.
+     * Nodes[0] is considered the root node.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.SavedObject nodes = 1;</code>
+     */
+    public org.tensorflow.proto.framework.SavedObjectOrBuilder getNodesOrBuilder(
+        int index) {
+      if (nodesBuilder_ == null) {
+        return nodes_.get(index);  } else {
+        return nodesBuilder_.getMessageOrBuilder(index);
+      }
+    }
+    /**
+     * <pre>
+     * Flattened list of objects in the object graph.
+     * The position of the object in this list indicates its id.
+     * Nodes[0] is considered the root node.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.SavedObject nodes = 1;</code>
+     */
+    public java.util.List<? extends org.tensorflow.proto.framework.SavedObjectOrBuilder> 
+         getNodesOrBuilderList() {
+      if (nodesBuilder_ != null) {
+        return nodesBuilder_.getMessageOrBuilderList();
+      } else {
+        return java.util.Collections.unmodifiableList(nodes_);
+      }
+    }
+    /**
+     * <pre>
+     * Flattened list of objects in the object graph.
+     * The position of the object in this list indicates its id.
+     * Nodes[0] is considered the root node.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.SavedObject nodes = 1;</code>
+     */
+    public org.tensorflow.proto.framework.SavedObject.Builder addNodesBuilder() {
+      return getNodesFieldBuilder().addBuilder(
+          org.tensorflow.proto.framework.SavedObject.getDefaultInstance());
+    }
+    /**
+     * <pre>
+     * Flattened list of objects in the object graph.
+     * The position of the object in this list indicates its id.
+     * Nodes[0] is considered the root node.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.SavedObject nodes = 1;</code>
+     */
+    public org.tensorflow.proto.framework.SavedObject.Builder addNodesBuilder(
+        int index) {
+      return getNodesFieldBuilder().addBuilder(
+          index, org.tensorflow.proto.framework.SavedObject.getDefaultInstance());
+    }
+    /**
+     * <pre>
+     * Flattened list of objects in the object graph.
+     * The position of the object in this list indicates its id.
+     * Nodes[0] is considered the root node.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.SavedObject nodes = 1;</code>
+     */
+    public java.util.List<org.tensorflow.proto.framework.SavedObject.Builder> 
+         getNodesBuilderList() {
+      return getNodesFieldBuilder().getBuilderList();
+    }
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.framework.SavedObject, org.tensorflow.proto.framework.SavedObject.Builder, org.tensorflow.proto.framework.SavedObjectOrBuilder> 
+        getNodesFieldBuilder() {
+      if (nodesBuilder_ == null) {
+        nodesBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
+            org.tensorflow.proto.framework.SavedObject, org.tensorflow.proto.framework.SavedObject.Builder, org.tensorflow.proto.framework.SavedObjectOrBuilder>(
+                nodes_,
+                ((bitField0_ & 0x00000001) != 0),
+                getParentForChildren(),
+                isClean());
+        nodes_ = null;
+      }
+      return nodesBuilder_;
+    }
+
+    private com.google.protobuf.MapField<
+        java.lang.String, org.tensorflow.proto.framework.SavedConcreteFunction> concreteFunctions_;
+    private com.google.protobuf.MapField<java.lang.String, org.tensorflow.proto.framework.SavedConcreteFunction>
+    internalGetConcreteFunctions() {
+      if (concreteFunctions_ == null) {
+        return com.google.protobuf.MapField.emptyMapField(
+            ConcreteFunctionsDefaultEntryHolder.defaultEntry);
+      }
+      return concreteFunctions_;
+    }
+    private com.google.protobuf.MapField<java.lang.String, org.tensorflow.proto.framework.SavedConcreteFunction>
+    internalGetMutableConcreteFunctions() {
+      onChanged();;
+      if (concreteFunctions_ == null) {
+        concreteFunctions_ = com.google.protobuf.MapField.newMapField(
+            ConcreteFunctionsDefaultEntryHolder.defaultEntry);
+      }
+      if (!concreteFunctions_.isMutable()) {
+        concreteFunctions_ = concreteFunctions_.copy();
+      }
+      return concreteFunctions_;
+    }
+
+    public int getConcreteFunctionsCount() {
+      return internalGetConcreteFunctions().getMap().size();
+    }
+    /**
+     * <pre>
+     * Information about captures and output structures in concrete functions.
+     * Referenced from SavedBareConcreteFunction and SavedFunction.
+     * </pre>
+     *
+     * <code>map&lt;string, .tensorflow.SavedConcreteFunction&gt; concrete_functions = 2;</code>
+     */
+
+    public boolean containsConcreteFunctions(
+        java.lang.String key) {
+      if (key == null) { throw new java.lang.NullPointerException(); }
+      return internalGetConcreteFunctions().getMap().containsKey(key);
+    }
+    /**
+     * Use {@link #getConcreteFunctionsMap()} instead.
+     */
+    @java.lang.Deprecated
+    public java.util.Map<java.lang.String, org.tensorflow.proto.framework.SavedConcreteFunction> getConcreteFunctions() {
+      return getConcreteFunctionsMap();
+    }
+    /**
+     * <pre>
+     * Information about captures and output structures in concrete functions.
+     * Referenced from SavedBareConcreteFunction and SavedFunction.
+     * </pre>
+     *
+     * <code>map&lt;string, .tensorflow.SavedConcreteFunction&gt; concrete_functions = 2;</code>
+     */
+
+    public java.util.Map<java.lang.String, org.tensorflow.proto.framework.SavedConcreteFunction> getConcreteFunctionsMap() {
+      return internalGetConcreteFunctions().getMap();
+    }
+    /**
+     * <pre>
+     * Information about captures and output structures in concrete functions.
+     * Referenced from SavedBareConcreteFunction and SavedFunction.
+     * </pre>
+     *
+     * <code>map&lt;string, .tensorflow.SavedConcreteFunction&gt; concrete_functions = 2;</code>
+     */
+
+    public org.tensorflow.proto.framework.SavedConcreteFunction getConcreteFunctionsOrDefault(
+        java.lang.String key,
+        org.tensorflow.proto.framework.SavedConcreteFunction defaultValue) {
+      if (key == null) { throw new java.lang.NullPointerException(); }
+      java.util.Map<java.lang.String, org.tensorflow.proto.framework.SavedConcreteFunction> map =
+          internalGetConcreteFunctions().getMap();
+      return map.containsKey(key) ? map.get(key) : defaultValue;
+    }
+    /**
+     * <pre>
+     * Information about captures and output structures in concrete functions.
+     * Referenced from SavedBareConcreteFunction and SavedFunction.
+     * </pre>
+     *
+     * <code>map&lt;string, .tensorflow.SavedConcreteFunction&gt; concrete_functions = 2;</code>
+     */
+
+    public org.tensorflow.proto.framework.SavedConcreteFunction getConcreteFunctionsOrThrow(
+        java.lang.String key) {
+      if (key == null) { throw new java.lang.NullPointerException(); }
+      java.util.Map<java.lang.String, org.tensorflow.proto.framework.SavedConcreteFunction> map =
+          internalGetConcreteFunctions().getMap();
+      if (!map.containsKey(key)) {
+        throw new java.lang.IllegalArgumentException();
+      }
+      return map.get(key);
+    }
+
+    public Builder clearConcreteFunctions() {
+      internalGetMutableConcreteFunctions().getMutableMap()
+          .clear();
+      return this;
+    }
+    /**
+     * <pre>
+     * Information about captures and output structures in concrete functions.
+     * Referenced from SavedBareConcreteFunction and SavedFunction.
+     * </pre>
+     *
+     * <code>map&lt;string, .tensorflow.SavedConcreteFunction&gt; concrete_functions = 2;</code>
+     */
+
+    public Builder removeConcreteFunctions(
+        java.lang.String key) {
+      if (key == null) { throw new java.lang.NullPointerException(); }
+      internalGetMutableConcreteFunctions().getMutableMap()
+          .remove(key);
+      return this;
+    }
+    /**
+     * Use alternate mutation accessors instead.
+     */
+    @java.lang.Deprecated
+    public java.util.Map<java.lang.String, org.tensorflow.proto.framework.SavedConcreteFunction>
+    getMutableConcreteFunctions() {
+      return internalGetMutableConcreteFunctions().getMutableMap();
+    }
+    /**
+     * <pre>
+     * Information about captures and output structures in concrete functions.
+     * Referenced from SavedBareConcreteFunction and SavedFunction.
+     * </pre>
+     *
+     * <code>map&lt;string, .tensorflow.SavedConcreteFunction&gt; concrete_functions = 2;</code>
+     */
+    public Builder putConcreteFunctions(
+        java.lang.String key,
+        org.tensorflow.proto.framework.SavedConcreteFunction value) {
+      if (key == null) { throw new java.lang.NullPointerException(); }
+      if (value == null) { throw new java.lang.NullPointerException(); }
+      internalGetMutableConcreteFunctions().getMutableMap()
+          .put(key, value);
+      return this;
+    }
+    /**
+     * <pre>
+     * Information about captures and output structures in concrete functions.
+     * Referenced from SavedBareConcreteFunction and SavedFunction.
+     * </pre>
+     *
+     * <code>map&lt;string, .tensorflow.SavedConcreteFunction&gt; concrete_functions = 2;</code>
+     */
+
+    public Builder putAllConcreteFunctions(
+        java.util.Map<java.lang.String, org.tensorflow.proto.framework.SavedConcreteFunction> values) {
+      internalGetMutableConcreteFunctions().getMutableMap()
+          .putAll(values);
+      return this;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.SavedObjectGraph)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.SavedObjectGraph)
+  private static final org.tensorflow.proto.framework.SavedObjectGraph DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.SavedObjectGraph();
+  }
+
+  public static org.tensorflow.proto.framework.SavedObjectGraph getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<SavedObjectGraph>
+      PARSER = new com.google.protobuf.AbstractParser<SavedObjectGraph>() {
+    @java.lang.Override
+    public SavedObjectGraph parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new SavedObjectGraph(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<SavedObjectGraph> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<SavedObjectGraph> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.framework.SavedObjectGraph getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedObjectGraphOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedObjectGraphOrBuilder.java
new file mode 100644
index 00000000000..3d80fbc1306
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedObjectGraphOrBuilder.java
@@ -0,0 +1,122 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/saved_object_graph.proto
+
+package org.tensorflow.proto.framework;
+
+public interface SavedObjectGraphOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.SavedObjectGraph)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <pre>
+   * Flattened list of objects in the object graph.
+   * The position of the object in this list indicates its id.
+   * Nodes[0] is considered the root node.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.SavedObject nodes = 1;</code>
+   */
+  java.util.List<org.tensorflow.proto.framework.SavedObject> 
+      getNodesList();
+  /**
+   * <pre>
+   * Flattened list of objects in the object graph.
+   * The position of the object in this list indicates its id.
+   * Nodes[0] is considered the root node.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.SavedObject nodes = 1;</code>
+   */
+  org.tensorflow.proto.framework.SavedObject getNodes(int index);
+  /**
+   * <pre>
+   * Flattened list of objects in the object graph.
+   * The position of the object in this list indicates its id.
+   * Nodes[0] is considered the root node.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.SavedObject nodes = 1;</code>
+   */
+  int getNodesCount();
+  /**
+   * <pre>
+   * Flattened list of objects in the object graph.
+   * The position of the object in this list indicates its id.
+   * Nodes[0] is considered the root node.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.SavedObject nodes = 1;</code>
+   */
+  java.util.List<? extends org.tensorflow.proto.framework.SavedObjectOrBuilder> 
+      getNodesOrBuilderList();
+  /**
+   * <pre>
+   * Flattened list of objects in the object graph.
+   * The position of the object in this list indicates its id.
+   * Nodes[0] is considered the root node.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.SavedObject nodes = 1;</code>
+   */
+  org.tensorflow.proto.framework.SavedObjectOrBuilder getNodesOrBuilder(
+      int index);
+
+  /**
+   * <pre>
+   * Information about captures and output structures in concrete functions.
+   * Referenced from SavedBareConcreteFunction and SavedFunction.
+   * </pre>
+   *
+   * <code>map&lt;string, .tensorflow.SavedConcreteFunction&gt; concrete_functions = 2;</code>
+   */
+  int getConcreteFunctionsCount();
+  /**
+   * <pre>
+   * Information about captures and output structures in concrete functions.
+   * Referenced from SavedBareConcreteFunction and SavedFunction.
+   * </pre>
+   *
+   * <code>map&lt;string, .tensorflow.SavedConcreteFunction&gt; concrete_functions = 2;</code>
+   */
+  boolean containsConcreteFunctions(
+      java.lang.String key);
+  /**
+   * Use {@link #getConcreteFunctionsMap()} instead.
+   */
+  @java.lang.Deprecated
+  java.util.Map<java.lang.String, org.tensorflow.proto.framework.SavedConcreteFunction>
+  getConcreteFunctions();
+  /**
+   * <pre>
+   * Information about captures and output structures in concrete functions.
+   * Referenced from SavedBareConcreteFunction and SavedFunction.
+   * </pre>
+   *
+   * <code>map&lt;string, .tensorflow.SavedConcreteFunction&gt; concrete_functions = 2;</code>
+   */
+  java.util.Map<java.lang.String, org.tensorflow.proto.framework.SavedConcreteFunction>
+  getConcreteFunctionsMap();
+  /**
+   * <pre>
+   * Information about captures and output structures in concrete functions.
+   * Referenced from SavedBareConcreteFunction and SavedFunction.
+   * </pre>
+   *
+   * <code>map&lt;string, .tensorflow.SavedConcreteFunction&gt; concrete_functions = 2;</code>
+   */
+
+  org.tensorflow.proto.framework.SavedConcreteFunction getConcreteFunctionsOrDefault(
+      java.lang.String key,
+      org.tensorflow.proto.framework.SavedConcreteFunction defaultValue);
+  /**
+   * <pre>
+   * Information about captures and output structures in concrete functions.
+   * Referenced from SavedBareConcreteFunction and SavedFunction.
+   * </pre>
+   *
+   * <code>map&lt;string, .tensorflow.SavedConcreteFunction&gt; concrete_functions = 2;</code>
+   */
+
+  org.tensorflow.proto.framework.SavedConcreteFunction getConcreteFunctionsOrThrow(
+      java.lang.String key);
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedObjectGraphProtos.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedObjectGraphProtos.java
new file mode 100644
index 00000000000..a7baea2d222
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedObjectGraphProtos.java
@@ -0,0 +1,263 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/saved_object_graph.proto
+
+package org.tensorflow.proto.framework;
+
+public final class SavedObjectGraphProtos {
+  private SavedObjectGraphProtos() {}
+  public static void registerAllExtensions(
+      com.google.protobuf.ExtensionRegistryLite registry) {
+  }
+
+  public static void registerAllExtensions(
+      com.google.protobuf.ExtensionRegistry registry) {
+    registerAllExtensions(
+        (com.google.protobuf.ExtensionRegistryLite) registry);
+  }
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_SavedObjectGraph_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_SavedObjectGraph_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_SavedObjectGraph_ConcreteFunctionsEntry_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_SavedObjectGraph_ConcreteFunctionsEntry_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_SavedObject_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_SavedObject_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_SavedObject_SaveableObjectsEntry_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_SavedObject_SaveableObjectsEntry_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_SavedUserObject_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_SavedUserObject_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_SavedAsset_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_SavedAsset_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_SavedFunction_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_SavedFunction_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_SavedConcreteFunction_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_SavedConcreteFunction_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_SavedBareConcreteFunction_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_SavedBareConcreteFunction_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_SavedConstant_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_SavedConstant_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_SavedVariable_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_SavedVariable_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_FunctionSpec_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_FunctionSpec_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_SavedResource_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_SavedResource_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_SaveableObject_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_SaveableObject_fieldAccessorTable;
+
+  public static com.google.protobuf.Descriptors.FileDescriptor
+      getDescriptor() {
+    return descriptor;
+  }
+  private static  com.google.protobuf.Descriptors.FileDescriptor
+      descriptor;
+  static {
+    java.lang.String[] descriptorData = {
+      "\n1tensorflow/core/protobuf/saved_object_" +
+      "graph.proto\022\ntensorflow\032,tensorflow/core" +
+      "/framework/tensor_shape.proto\032%tensorflo" +
+      "w/core/framework/types.proto\032(tensorflow" +
+      "/core/framework/variable.proto\032(tensorfl" +
+      "ow/core/framework/versions.proto\032%tensor" +
+      "flow/core/protobuf/struct.proto\0325tensorf" +
+      "low/core/protobuf/trackable_object_graph" +
+      ".proto\"\350\001\n\020SavedObjectGraph\022&\n\005nodes\030\001 \003" +
+      "(\0132\027.tensorflow.SavedObject\022O\n\022concrete_" +
+      "functions\030\002 \003(\01323.tensorflow.SavedObject" +
+      "Graph.ConcreteFunctionsEntry\032[\n\026Concrete" +
+      "FunctionsEntry\022\013\n\003key\030\001 \001(\t\0220\n\005value\030\002 \001" +
+      "(\0132!.tensorflow.SavedConcreteFunction:\0028" +
+      "\001\"\331\005\n\013SavedObject\022R\n\010children\030\001 \003(\0132@.te" +
+      "nsorflow.TrackableObjectGraph.TrackableO" +
+      "bject.ObjectReference\022^\n\016slot_variables\030" +
+      "\003 \003(\0132F.tensorflow.TrackableObjectGraph." +
+      "TrackableObject.SlotVariableReference\0222\n" +
+      "\013user_object\030\004 \001(\0132\033.tensorflow.SavedUse" +
+      "rObjectH\000\022\'\n\005asset\030\005 \001(\0132\026.tensorflow.Sa" +
+      "vedAssetH\000\022-\n\010function\030\006 \001(\0132\031.tensorflo" +
+      "w.SavedFunctionH\000\022-\n\010variable\030\007 \001(\0132\031.te" +
+      "nsorflow.SavedVariableH\000\022G\n\026bare_concret" +
+      "e_function\030\010 \001(\0132%.tensorflow.SavedBareC" +
+      "oncreteFunctionH\000\022-\n\010constant\030\t \001(\0132\031.te" +
+      "nsorflow.SavedConstantH\000\022-\n\010resource\030\n \001" +
+      "(\0132\031.tensorflow.SavedResourceH\000\022F\n\020savea" +
+      "ble_objects\030\013 \003(\0132,.tensorflow.SavedObje" +
+      "ct.SaveableObjectsEntry\032R\n\024SaveableObjec" +
+      "tsEntry\022\013\n\003key\030\001 \001(\t\022)\n\005value\030\002 \001(\0132\032.te" +
+      "nsorflow.SaveableObject:\0028\001B\006\n\004kindJ\004\010\002\020" +
+      "\003R\nattributes\"`\n\017SavedUserObject\022\022\n\niden" +
+      "tifier\030\001 \001(\t\022\'\n\007version\030\002 \001(\0132\026.tensorfl" +
+      "ow.VersionDef\022\020\n\010metadata\030\003 \001(\t\"*\n\nSaved" +
+      "Asset\022\034\n\024asset_file_def_index\030\001 \001(\005\"\\\n\rS" +
+      "avedFunction\022\032\n\022concrete_functions\030\001 \003(\t" +
+      "\022/\n\rfunction_spec\030\002 \001(\0132\030.tensorflow.Fun" +
+      "ctionSpec\"\250\001\n\025SavedConcreteFunction\022\024\n\014b" +
+      "ound_inputs\030\002 \003(\005\022B\n\035canonicalized_input" +
+      "_signature\030\003 \001(\0132\033.tensorflow.Structured" +
+      "Value\0225\n\020output_signature\030\004 \001(\0132\033.tensor" +
+      "flow.StructuredValue\"|\n\031SavedBareConcret" +
+      "eFunction\022\036\n\026concrete_function_name\030\001 \001(" +
+      "\t\022\031\n\021argument_keywords\030\002 \003(\t\022$\n\034allowed_" +
+      "positional_arguments\030\003 \001(\003\"\"\n\rSavedConst" +
+      "ant\022\021\n\toperation\030\001 \001(\t\"\366\001\n\rSavedVariable" +
+      "\022#\n\005dtype\030\001 \001(\0162\024.tensorflow.DataType\022+\n" +
+      "\005shape\030\002 \001(\0132\034.tensorflow.TensorShapePro" +
+      "to\022\021\n\ttrainable\030\003 \001(\010\022<\n\017synchronization" +
+      "\030\004 \001(\0162#.tensorflow.VariableSynchronizat" +
+      "ion\0224\n\013aggregation\030\005 \001(\0162\037.tensorflow.Va" +
+      "riableAggregation\022\014\n\004name\030\006 \001(\t\"\225\001\n\014Func" +
+      "tionSpec\0220\n\013fullargspec\030\001 \001(\0132\033.tensorfl" +
+      "ow.StructuredValue\022\021\n\tis_method\030\002 \001(\010\0224\n" +
+      "\017input_signature\030\005 \001(\0132\033.tensorflow.Stru" +
+      "cturedValueJ\004\010\003\020\004J\004\010\004\020\005\"\037\n\rSavedResource" +
+      "\022\016\n\006device\030\001 \001(\t\"A\n\016SaveableObject\022\025\n\rsa" +
+      "ve_function\030\002 \001(\005\022\030\n\020restore_function\030\003 " +
+      "\001(\005B\207\001\n\036org.tensorflow.proto.frameworkB\026" +
+      "SavedObjectGraphProtosP\001ZHgithub.com/ten" +
+      "sorflow/tensorflow/tensorflow/go/core/co" +
+      "re_protos_go_proto\370\001\001b\006proto3"
+    };
+    descriptor = com.google.protobuf.Descriptors.FileDescriptor
+      .internalBuildGeneratedFileFrom(descriptorData,
+        new com.google.protobuf.Descriptors.FileDescriptor[] {
+          org.tensorflow.proto.framework.TensorShapeProtos.getDescriptor(),
+          org.tensorflow.proto.framework.TypesProtos.getDescriptor(),
+          org.tensorflow.proto.framework.VariableProtos.getDescriptor(),
+          org.tensorflow.proto.framework.VersionsProtos.getDescriptor(),
+          org.tensorflow.proto.framework.StructProtos.getDescriptor(),
+          org.tensorflow.proto.framework.TrackableObjectGraphProtos.getDescriptor(),
+        });
+    internal_static_tensorflow_SavedObjectGraph_descriptor =
+      getDescriptor().getMessageTypes().get(0);
+    internal_static_tensorflow_SavedObjectGraph_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_SavedObjectGraph_descriptor,
+        new java.lang.String[] { "Nodes", "ConcreteFunctions", });
+    internal_static_tensorflow_SavedObjectGraph_ConcreteFunctionsEntry_descriptor =
+      internal_static_tensorflow_SavedObjectGraph_descriptor.getNestedTypes().get(0);
+    internal_static_tensorflow_SavedObjectGraph_ConcreteFunctionsEntry_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_SavedObjectGraph_ConcreteFunctionsEntry_descriptor,
+        new java.lang.String[] { "Key", "Value", });
+    internal_static_tensorflow_SavedObject_descriptor =
+      getDescriptor().getMessageTypes().get(1);
+    internal_static_tensorflow_SavedObject_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_SavedObject_descriptor,
+        new java.lang.String[] { "Children", "SlotVariables", "UserObject", "Asset", "Function", "Variable", "BareConcreteFunction", "Constant", "Resource", "SaveableObjects", "Kind", });
+    internal_static_tensorflow_SavedObject_SaveableObjectsEntry_descriptor =
+      internal_static_tensorflow_SavedObject_descriptor.getNestedTypes().get(0);
+    internal_static_tensorflow_SavedObject_SaveableObjectsEntry_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_SavedObject_SaveableObjectsEntry_descriptor,
+        new java.lang.String[] { "Key", "Value", });
+    internal_static_tensorflow_SavedUserObject_descriptor =
+      getDescriptor().getMessageTypes().get(2);
+    internal_static_tensorflow_SavedUserObject_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_SavedUserObject_descriptor,
+        new java.lang.String[] { "Identifier", "Version", "Metadata", });
+    internal_static_tensorflow_SavedAsset_descriptor =
+      getDescriptor().getMessageTypes().get(3);
+    internal_static_tensorflow_SavedAsset_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_SavedAsset_descriptor,
+        new java.lang.String[] { "AssetFileDefIndex", });
+    internal_static_tensorflow_SavedFunction_descriptor =
+      getDescriptor().getMessageTypes().get(4);
+    internal_static_tensorflow_SavedFunction_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_SavedFunction_descriptor,
+        new java.lang.String[] { "ConcreteFunctions", "FunctionSpec", });
+    internal_static_tensorflow_SavedConcreteFunction_descriptor =
+      getDescriptor().getMessageTypes().get(5);
+    internal_static_tensorflow_SavedConcreteFunction_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_SavedConcreteFunction_descriptor,
+        new java.lang.String[] { "BoundInputs", "CanonicalizedInputSignature", "OutputSignature", });
+    internal_static_tensorflow_SavedBareConcreteFunction_descriptor =
+      getDescriptor().getMessageTypes().get(6);
+    internal_static_tensorflow_SavedBareConcreteFunction_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_SavedBareConcreteFunction_descriptor,
+        new java.lang.String[] { "ConcreteFunctionName", "ArgumentKeywords", "AllowedPositionalArguments", });
+    internal_static_tensorflow_SavedConstant_descriptor =
+      getDescriptor().getMessageTypes().get(7);
+    internal_static_tensorflow_SavedConstant_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_SavedConstant_descriptor,
+        new java.lang.String[] { "Operation", });
+    internal_static_tensorflow_SavedVariable_descriptor =
+      getDescriptor().getMessageTypes().get(8);
+    internal_static_tensorflow_SavedVariable_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_SavedVariable_descriptor,
+        new java.lang.String[] { "Dtype", "Shape", "Trainable", "Synchronization", "Aggregation", "Name", });
+    internal_static_tensorflow_FunctionSpec_descriptor =
+      getDescriptor().getMessageTypes().get(9);
+    internal_static_tensorflow_FunctionSpec_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_FunctionSpec_descriptor,
+        new java.lang.String[] { "Fullargspec", "IsMethod", "InputSignature", });
+    internal_static_tensorflow_SavedResource_descriptor =
+      getDescriptor().getMessageTypes().get(10);
+    internal_static_tensorflow_SavedResource_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_SavedResource_descriptor,
+        new java.lang.String[] { "Device", });
+    internal_static_tensorflow_SaveableObject_descriptor =
+      getDescriptor().getMessageTypes().get(11);
+    internal_static_tensorflow_SaveableObject_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_SaveableObject_descriptor,
+        new java.lang.String[] { "SaveFunction", "RestoreFunction", });
+    org.tensorflow.proto.framework.TensorShapeProtos.getDescriptor();
+    org.tensorflow.proto.framework.TypesProtos.getDescriptor();
+    org.tensorflow.proto.framework.VariableProtos.getDescriptor();
+    org.tensorflow.proto.framework.VersionsProtos.getDescriptor();
+    org.tensorflow.proto.framework.StructProtos.getDescriptor();
+    org.tensorflow.proto.framework.TrackableObjectGraphProtos.getDescriptor();
+  }
+
+  // @@protoc_insertion_point(outer_class_scope)
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedObjectOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedObjectOrBuilder.java
new file mode 100644
index 00000000000..269ee74c9f2
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedObjectOrBuilder.java
@@ -0,0 +1,249 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/saved_object_graph.proto
+
+package org.tensorflow.proto.framework;
+
+public interface SavedObjectOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.SavedObject)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <pre>
+   * Objects which this object depends on: named edges in the dependency
+   * graph.
+   * Note: currently only valid if kind == "user_object".
+   * </pre>
+   *
+   * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.ObjectReference children = 1;</code>
+   */
+  java.util.List<org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference> 
+      getChildrenList();
+  /**
+   * <pre>
+   * Objects which this object depends on: named edges in the dependency
+   * graph.
+   * Note: currently only valid if kind == "user_object".
+   * </pre>
+   *
+   * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.ObjectReference children = 1;</code>
+   */
+  org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference getChildren(int index);
+  /**
+   * <pre>
+   * Objects which this object depends on: named edges in the dependency
+   * graph.
+   * Note: currently only valid if kind == "user_object".
+   * </pre>
+   *
+   * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.ObjectReference children = 1;</code>
+   */
+  int getChildrenCount();
+  /**
+   * <pre>
+   * Objects which this object depends on: named edges in the dependency
+   * graph.
+   * Note: currently only valid if kind == "user_object".
+   * </pre>
+   *
+   * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.ObjectReference children = 1;</code>
+   */
+  java.util.List<? extends org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReferenceOrBuilder> 
+      getChildrenOrBuilderList();
+  /**
+   * <pre>
+   * Objects which this object depends on: named edges in the dependency
+   * graph.
+   * Note: currently only valid if kind == "user_object".
+   * </pre>
+   *
+   * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.ObjectReference children = 1;</code>
+   */
+  org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReferenceOrBuilder getChildrenOrBuilder(
+      int index);
+
+  /**
+   * <pre>
+   * Slot variables owned by this object. This describes the three-way
+   * (optimizer, variable, slot variable) relationship; none of the three
+   * depend on the others directly.
+   * Note: currently only valid if kind == "user_object".
+   * </pre>
+   *
+   * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SlotVariableReference slot_variables = 3;</code>
+   */
+  java.util.List<org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference> 
+      getSlotVariablesList();
+  /**
+   * <pre>
+   * Slot variables owned by this object. This describes the three-way
+   * (optimizer, variable, slot variable) relationship; none of the three
+   * depend on the others directly.
+   * Note: currently only valid if kind == "user_object".
+   * </pre>
+   *
+   * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SlotVariableReference slot_variables = 3;</code>
+   */
+  org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference getSlotVariables(int index);
+  /**
+   * <pre>
+   * Slot variables owned by this object. This describes the three-way
+   * (optimizer, variable, slot variable) relationship; none of the three
+   * depend on the others directly.
+   * Note: currently only valid if kind == "user_object".
+   * </pre>
+   *
+   * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SlotVariableReference slot_variables = 3;</code>
+   */
+  int getSlotVariablesCount();
+  /**
+   * <pre>
+   * Slot variables owned by this object. This describes the three-way
+   * (optimizer, variable, slot variable) relationship; none of the three
+   * depend on the others directly.
+   * Note: currently only valid if kind == "user_object".
+   * </pre>
+   *
+   * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SlotVariableReference slot_variables = 3;</code>
+   */
+  java.util.List<? extends org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReferenceOrBuilder> 
+      getSlotVariablesOrBuilderList();
+  /**
+   * <pre>
+   * Slot variables owned by this object. This describes the three-way
+   * (optimizer, variable, slot variable) relationship; none of the three
+   * depend on the others directly.
+   * Note: currently only valid if kind == "user_object".
+   * </pre>
+   *
+   * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SlotVariableReference slot_variables = 3;</code>
+   */
+  org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReferenceOrBuilder getSlotVariablesOrBuilder(
+      int index);
+
+  /**
+   * <code>.tensorflow.SavedUserObject user_object = 4;</code>
+   */
+  boolean hasUserObject();
+  /**
+   * <code>.tensorflow.SavedUserObject user_object = 4;</code>
+   */
+  org.tensorflow.proto.framework.SavedUserObject getUserObject();
+  /**
+   * <code>.tensorflow.SavedUserObject user_object = 4;</code>
+   */
+  org.tensorflow.proto.framework.SavedUserObjectOrBuilder getUserObjectOrBuilder();
+
+  /**
+   * <code>.tensorflow.SavedAsset asset = 5;</code>
+   */
+  boolean hasAsset();
+  /**
+   * <code>.tensorflow.SavedAsset asset = 5;</code>
+   */
+  org.tensorflow.proto.framework.SavedAsset getAsset();
+  /**
+   * <code>.tensorflow.SavedAsset asset = 5;</code>
+   */
+  org.tensorflow.proto.framework.SavedAssetOrBuilder getAssetOrBuilder();
+
+  /**
+   * <code>.tensorflow.SavedFunction function = 6;</code>
+   */
+  boolean hasFunction();
+  /**
+   * <code>.tensorflow.SavedFunction function = 6;</code>
+   */
+  org.tensorflow.proto.framework.SavedFunction getFunction();
+  /**
+   * <code>.tensorflow.SavedFunction function = 6;</code>
+   */
+  org.tensorflow.proto.framework.SavedFunctionOrBuilder getFunctionOrBuilder();
+
+  /**
+   * <code>.tensorflow.SavedVariable variable = 7;</code>
+   */
+  boolean hasVariable();
+  /**
+   * <code>.tensorflow.SavedVariable variable = 7;</code>
+   */
+  org.tensorflow.proto.framework.SavedVariable getVariable();
+  /**
+   * <code>.tensorflow.SavedVariable variable = 7;</code>
+   */
+  org.tensorflow.proto.framework.SavedVariableOrBuilder getVariableOrBuilder();
+
+  /**
+   * <code>.tensorflow.SavedBareConcreteFunction bare_concrete_function = 8;</code>
+   */
+  boolean hasBareConcreteFunction();
+  /**
+   * <code>.tensorflow.SavedBareConcreteFunction bare_concrete_function = 8;</code>
+   */
+  org.tensorflow.proto.framework.SavedBareConcreteFunction getBareConcreteFunction();
+  /**
+   * <code>.tensorflow.SavedBareConcreteFunction bare_concrete_function = 8;</code>
+   */
+  org.tensorflow.proto.framework.SavedBareConcreteFunctionOrBuilder getBareConcreteFunctionOrBuilder();
+
+  /**
+   * <code>.tensorflow.SavedConstant constant = 9;</code>
+   */
+  boolean hasConstant();
+  /**
+   * <code>.tensorflow.SavedConstant constant = 9;</code>
+   */
+  org.tensorflow.proto.framework.SavedConstant getConstant();
+  /**
+   * <code>.tensorflow.SavedConstant constant = 9;</code>
+   */
+  org.tensorflow.proto.framework.SavedConstantOrBuilder getConstantOrBuilder();
+
+  /**
+   * <code>.tensorflow.SavedResource resource = 10;</code>
+   */
+  boolean hasResource();
+  /**
+   * <code>.tensorflow.SavedResource resource = 10;</code>
+   */
+  org.tensorflow.proto.framework.SavedResource getResource();
+  /**
+   * <code>.tensorflow.SavedResource resource = 10;</code>
+   */
+  org.tensorflow.proto.framework.SavedResourceOrBuilder getResourceOrBuilder();
+
+  /**
+   * <code>map&lt;string, .tensorflow.SaveableObject&gt; saveable_objects = 11;</code>
+   */
+  int getSaveableObjectsCount();
+  /**
+   * <code>map&lt;string, .tensorflow.SaveableObject&gt; saveable_objects = 11;</code>
+   */
+  boolean containsSaveableObjects(
+      java.lang.String key);
+  /**
+   * Use {@link #getSaveableObjectsMap()} instead.
+   */
+  @java.lang.Deprecated
+  java.util.Map<java.lang.String, org.tensorflow.proto.framework.SaveableObject>
+  getSaveableObjects();
+  /**
+   * <code>map&lt;string, .tensorflow.SaveableObject&gt; saveable_objects = 11;</code>
+   */
+  java.util.Map<java.lang.String, org.tensorflow.proto.framework.SaveableObject>
+  getSaveableObjectsMap();
+  /**
+   * <code>map&lt;string, .tensorflow.SaveableObject&gt; saveable_objects = 11;</code>
+   */
+
+  org.tensorflow.proto.framework.SaveableObject getSaveableObjectsOrDefault(
+      java.lang.String key,
+      org.tensorflow.proto.framework.SaveableObject defaultValue);
+  /**
+   * <code>map&lt;string, .tensorflow.SaveableObject&gt; saveable_objects = 11;</code>
+   */
+
+  org.tensorflow.proto.framework.SaveableObject getSaveableObjectsOrThrow(
+      java.lang.String key);
+
+  public org.tensorflow.proto.framework.SavedObject.KindCase getKindCase();
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedResource.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedResource.java
new file mode 100644
index 00000000000..b45f7ff73a9
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedResource.java
@@ -0,0 +1,600 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/saved_object_graph.proto
+
+package org.tensorflow.proto.framework;
+
+/**
+ * <pre>
+ * A SavedResource represents a TF object that holds state during its lifetime.
+ * An object of this type can have a reference to a:
+ * create_resource() and an initialize() function.
+ * </pre>
+ *
+ * Protobuf type {@code tensorflow.SavedResource}
+ */
+public  final class SavedResource extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.SavedResource)
+    SavedResourceOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use SavedResource.newBuilder() to construct.
+  private SavedResource(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private SavedResource() {
+    device_ = "";
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new SavedResource();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private SavedResource(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 10: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            device_ = s;
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.framework.SavedObjectGraphProtos.internal_static_tensorflow_SavedResource_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.framework.SavedObjectGraphProtos.internal_static_tensorflow_SavedResource_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.framework.SavedResource.class, org.tensorflow.proto.framework.SavedResource.Builder.class);
+  }
+
+  public static final int DEVICE_FIELD_NUMBER = 1;
+  private volatile java.lang.Object device_;
+  /**
+   * <pre>
+   * A device specification indicating a required placement for the resource
+   * creation function, e.g. "CPU". An empty string allows the user to select a
+   * device.
+   * </pre>
+   *
+   * <code>string device = 1;</code>
+   */
+  public java.lang.String getDevice() {
+    java.lang.Object ref = device_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      device_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * A device specification indicating a required placement for the resource
+   * creation function, e.g. "CPU". An empty string allows the user to select a
+   * device.
+   * </pre>
+   *
+   * <code>string device = 1;</code>
+   */
+  public com.google.protobuf.ByteString
+      getDeviceBytes() {
+    java.lang.Object ref = device_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      device_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    if (!getDeviceBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 1, device_);
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (!getDeviceBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, device_);
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.framework.SavedResource)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.framework.SavedResource other = (org.tensorflow.proto.framework.SavedResource) obj;
+
+    if (!getDevice()
+        .equals(other.getDevice())) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    hash = (37 * hash) + DEVICE_FIELD_NUMBER;
+    hash = (53 * hash) + getDevice().hashCode();
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.framework.SavedResource parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.SavedResource parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.SavedResource parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.SavedResource parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.SavedResource parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.SavedResource parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.SavedResource parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.SavedResource parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.SavedResource parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.SavedResource parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.SavedResource parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.SavedResource parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.framework.SavedResource prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * <pre>
+   * A SavedResource represents a TF object that holds state during its lifetime.
+   * An object of this type can have a reference to a:
+   * create_resource() and an initialize() function.
+   * </pre>
+   *
+   * Protobuf type {@code tensorflow.SavedResource}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.SavedResource)
+      org.tensorflow.proto.framework.SavedResourceOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.SavedObjectGraphProtos.internal_static_tensorflow_SavedResource_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.SavedObjectGraphProtos.internal_static_tensorflow_SavedResource_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.SavedResource.class, org.tensorflow.proto.framework.SavedResource.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.framework.SavedResource.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      device_ = "";
+
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.framework.SavedObjectGraphProtos.internal_static_tensorflow_SavedResource_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.SavedResource getDefaultInstanceForType() {
+      return org.tensorflow.proto.framework.SavedResource.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.SavedResource build() {
+      org.tensorflow.proto.framework.SavedResource result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.SavedResource buildPartial() {
+      org.tensorflow.proto.framework.SavedResource result = new org.tensorflow.proto.framework.SavedResource(this);
+      result.device_ = device_;
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.framework.SavedResource) {
+        return mergeFrom((org.tensorflow.proto.framework.SavedResource)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.framework.SavedResource other) {
+      if (other == org.tensorflow.proto.framework.SavedResource.getDefaultInstance()) return this;
+      if (!other.getDevice().isEmpty()) {
+        device_ = other.device_;
+        onChanged();
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.framework.SavedResource parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.framework.SavedResource) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+
+    private java.lang.Object device_ = "";
+    /**
+     * <pre>
+     * A device specification indicating a required placement for the resource
+     * creation function, e.g. "CPU". An empty string allows the user to select a
+     * device.
+     * </pre>
+     *
+     * <code>string device = 1;</code>
+     */
+    public java.lang.String getDevice() {
+      java.lang.Object ref = device_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        device_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * A device specification indicating a required placement for the resource
+     * creation function, e.g. "CPU". An empty string allows the user to select a
+     * device.
+     * </pre>
+     *
+     * <code>string device = 1;</code>
+     */
+    public com.google.protobuf.ByteString
+        getDeviceBytes() {
+      java.lang.Object ref = device_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        device_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * A device specification indicating a required placement for the resource
+     * creation function, e.g. "CPU". An empty string allows the user to select a
+     * device.
+     * </pre>
+     *
+     * <code>string device = 1;</code>
+     */
+    public Builder setDevice(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      device_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * A device specification indicating a required placement for the resource
+     * creation function, e.g. "CPU". An empty string allows the user to select a
+     * device.
+     * </pre>
+     *
+     * <code>string device = 1;</code>
+     */
+    public Builder clearDevice() {
+      
+      device_ = getDefaultInstance().getDevice();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * A device specification indicating a required placement for the resource
+     * creation function, e.g. "CPU". An empty string allows the user to select a
+     * device.
+     * </pre>
+     *
+     * <code>string device = 1;</code>
+     */
+    public Builder setDeviceBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      device_ = value;
+      onChanged();
+      return this;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.SavedResource)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.SavedResource)
+  private static final org.tensorflow.proto.framework.SavedResource DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.SavedResource();
+  }
+
+  public static org.tensorflow.proto.framework.SavedResource getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<SavedResource>
+      PARSER = new com.google.protobuf.AbstractParser<SavedResource>() {
+    @java.lang.Override
+    public SavedResource parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new SavedResource(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<SavedResource> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<SavedResource> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.framework.SavedResource getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedResourceOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedResourceOrBuilder.java
new file mode 100644
index 00000000000..5d78c2eea9b
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedResourceOrBuilder.java
@@ -0,0 +1,31 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/saved_object_graph.proto
+
+package org.tensorflow.proto.framework;
+
+public interface SavedResourceOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.SavedResource)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <pre>
+   * A device specification indicating a required placement for the resource
+   * creation function, e.g. "CPU". An empty string allows the user to select a
+   * device.
+   * </pre>
+   *
+   * <code>string device = 1;</code>
+   */
+  java.lang.String getDevice();
+  /**
+   * <pre>
+   * A device specification indicating a required placement for the resource
+   * creation function, e.g. "CPU". An empty string allows the user to select a
+   * device.
+   * </pre>
+   *
+   * <code>string device = 1;</code>
+   */
+  com.google.protobuf.ByteString
+      getDeviceBytes();
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedUserObject.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedUserObject.java
new file mode 100644
index 00000000000..f66874096f5
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedUserObject.java
@@ -0,0 +1,974 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/saved_object_graph.proto
+
+package org.tensorflow.proto.framework;
+
+/**
+ * <pre>
+ * A SavedUserObject is an object (in the object-oriented language of the
+ * TensorFlow program) of some user- or framework-defined class other than
+ * those handled specifically by the other kinds of SavedObjects.
+ * This object cannot be evaluated as a tensor, and therefore cannot be bound
+ * to an input of a function.
+ * </pre>
+ *
+ * Protobuf type {@code tensorflow.SavedUserObject}
+ */
+public  final class SavedUserObject extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.SavedUserObject)
+    SavedUserObjectOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use SavedUserObject.newBuilder() to construct.
+  private SavedUserObject(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private SavedUserObject() {
+    identifier_ = "";
+    metadata_ = "";
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new SavedUserObject();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private SavedUserObject(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 10: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            identifier_ = s;
+            break;
+          }
+          case 18: {
+            org.tensorflow.proto.framework.VersionDef.Builder subBuilder = null;
+            if (version_ != null) {
+              subBuilder = version_.toBuilder();
+            }
+            version_ = input.readMessage(org.tensorflow.proto.framework.VersionDef.parser(), extensionRegistry);
+            if (subBuilder != null) {
+              subBuilder.mergeFrom(version_);
+              version_ = subBuilder.buildPartial();
+            }
+
+            break;
+          }
+          case 26: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            metadata_ = s;
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.framework.SavedObjectGraphProtos.internal_static_tensorflow_SavedUserObject_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.framework.SavedObjectGraphProtos.internal_static_tensorflow_SavedUserObject_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.framework.SavedUserObject.class, org.tensorflow.proto.framework.SavedUserObject.Builder.class);
+  }
+
+  public static final int IDENTIFIER_FIELD_NUMBER = 1;
+  private volatile java.lang.Object identifier_;
+  /**
+   * <pre>
+   * Corresponds to a registration of the type to use in the loading program.
+   * </pre>
+   *
+   * <code>string identifier = 1;</code>
+   */
+  public java.lang.String getIdentifier() {
+    java.lang.Object ref = identifier_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      identifier_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * Corresponds to a registration of the type to use in the loading program.
+   * </pre>
+   *
+   * <code>string identifier = 1;</code>
+   */
+  public com.google.protobuf.ByteString
+      getIdentifierBytes() {
+    java.lang.Object ref = identifier_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      identifier_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int VERSION_FIELD_NUMBER = 2;
+  private org.tensorflow.proto.framework.VersionDef version_;
+  /**
+   * <pre>
+   * Version information from the producer of this SavedUserObject.
+   * </pre>
+   *
+   * <code>.tensorflow.VersionDef version = 2;</code>
+   */
+  public boolean hasVersion() {
+    return version_ != null;
+  }
+  /**
+   * <pre>
+   * Version information from the producer of this SavedUserObject.
+   * </pre>
+   *
+   * <code>.tensorflow.VersionDef version = 2;</code>
+   */
+  public org.tensorflow.proto.framework.VersionDef getVersion() {
+    return version_ == null ? org.tensorflow.proto.framework.VersionDef.getDefaultInstance() : version_;
+  }
+  /**
+   * <pre>
+   * Version information from the producer of this SavedUserObject.
+   * </pre>
+   *
+   * <code>.tensorflow.VersionDef version = 2;</code>
+   */
+  public org.tensorflow.proto.framework.VersionDefOrBuilder getVersionOrBuilder() {
+    return getVersion();
+  }
+
+  public static final int METADATA_FIELD_NUMBER = 3;
+  private volatile java.lang.Object metadata_;
+  /**
+   * <pre>
+   * Initialization-related metadata.
+   * </pre>
+   *
+   * <code>string metadata = 3;</code>
+   */
+  public java.lang.String getMetadata() {
+    java.lang.Object ref = metadata_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      metadata_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * Initialization-related metadata.
+   * </pre>
+   *
+   * <code>string metadata = 3;</code>
+   */
+  public com.google.protobuf.ByteString
+      getMetadataBytes() {
+    java.lang.Object ref = metadata_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      metadata_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    if (!getIdentifierBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 1, identifier_);
+    }
+    if (version_ != null) {
+      output.writeMessage(2, getVersion());
+    }
+    if (!getMetadataBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 3, metadata_);
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (!getIdentifierBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, identifier_);
+    }
+    if (version_ != null) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(2, getVersion());
+    }
+    if (!getMetadataBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(3, metadata_);
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.framework.SavedUserObject)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.framework.SavedUserObject other = (org.tensorflow.proto.framework.SavedUserObject) obj;
+
+    if (!getIdentifier()
+        .equals(other.getIdentifier())) return false;
+    if (hasVersion() != other.hasVersion()) return false;
+    if (hasVersion()) {
+      if (!getVersion()
+          .equals(other.getVersion())) return false;
+    }
+    if (!getMetadata()
+        .equals(other.getMetadata())) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    hash = (37 * hash) + IDENTIFIER_FIELD_NUMBER;
+    hash = (53 * hash) + getIdentifier().hashCode();
+    if (hasVersion()) {
+      hash = (37 * hash) + VERSION_FIELD_NUMBER;
+      hash = (53 * hash) + getVersion().hashCode();
+    }
+    hash = (37 * hash) + METADATA_FIELD_NUMBER;
+    hash = (53 * hash) + getMetadata().hashCode();
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.framework.SavedUserObject parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.SavedUserObject parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.SavedUserObject parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.SavedUserObject parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.SavedUserObject parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.SavedUserObject parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.SavedUserObject parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.SavedUserObject parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.SavedUserObject parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.SavedUserObject parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.SavedUserObject parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.SavedUserObject parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.framework.SavedUserObject prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * <pre>
+   * A SavedUserObject is an object (in the object-oriented language of the
+   * TensorFlow program) of some user- or framework-defined class other than
+   * those handled specifically by the other kinds of SavedObjects.
+   * This object cannot be evaluated as a tensor, and therefore cannot be bound
+   * to an input of a function.
+   * </pre>
+   *
+   * Protobuf type {@code tensorflow.SavedUserObject}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.SavedUserObject)
+      org.tensorflow.proto.framework.SavedUserObjectOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.SavedObjectGraphProtos.internal_static_tensorflow_SavedUserObject_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.SavedObjectGraphProtos.internal_static_tensorflow_SavedUserObject_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.SavedUserObject.class, org.tensorflow.proto.framework.SavedUserObject.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.framework.SavedUserObject.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      identifier_ = "";
+
+      if (versionBuilder_ == null) {
+        version_ = null;
+      } else {
+        version_ = null;
+        versionBuilder_ = null;
+      }
+      metadata_ = "";
+
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.framework.SavedObjectGraphProtos.internal_static_tensorflow_SavedUserObject_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.SavedUserObject getDefaultInstanceForType() {
+      return org.tensorflow.proto.framework.SavedUserObject.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.SavedUserObject build() {
+      org.tensorflow.proto.framework.SavedUserObject result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.SavedUserObject buildPartial() {
+      org.tensorflow.proto.framework.SavedUserObject result = new org.tensorflow.proto.framework.SavedUserObject(this);
+      result.identifier_ = identifier_;
+      if (versionBuilder_ == null) {
+        result.version_ = version_;
+      } else {
+        result.version_ = versionBuilder_.build();
+      }
+      result.metadata_ = metadata_;
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.framework.SavedUserObject) {
+        return mergeFrom((org.tensorflow.proto.framework.SavedUserObject)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.framework.SavedUserObject other) {
+      if (other == org.tensorflow.proto.framework.SavedUserObject.getDefaultInstance()) return this;
+      if (!other.getIdentifier().isEmpty()) {
+        identifier_ = other.identifier_;
+        onChanged();
+      }
+      if (other.hasVersion()) {
+        mergeVersion(other.getVersion());
+      }
+      if (!other.getMetadata().isEmpty()) {
+        metadata_ = other.metadata_;
+        onChanged();
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.framework.SavedUserObject parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.framework.SavedUserObject) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+
+    private java.lang.Object identifier_ = "";
+    /**
+     * <pre>
+     * Corresponds to a registration of the type to use in the loading program.
+     * </pre>
+     *
+     * <code>string identifier = 1;</code>
+     */
+    public java.lang.String getIdentifier() {
+      java.lang.Object ref = identifier_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        identifier_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Corresponds to a registration of the type to use in the loading program.
+     * </pre>
+     *
+     * <code>string identifier = 1;</code>
+     */
+    public com.google.protobuf.ByteString
+        getIdentifierBytes() {
+      java.lang.Object ref = identifier_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        identifier_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Corresponds to a registration of the type to use in the loading program.
+     * </pre>
+     *
+     * <code>string identifier = 1;</code>
+     */
+    public Builder setIdentifier(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      identifier_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Corresponds to a registration of the type to use in the loading program.
+     * </pre>
+     *
+     * <code>string identifier = 1;</code>
+     */
+    public Builder clearIdentifier() {
+      
+      identifier_ = getDefaultInstance().getIdentifier();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Corresponds to a registration of the type to use in the loading program.
+     * </pre>
+     *
+     * <code>string identifier = 1;</code>
+     */
+    public Builder setIdentifierBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      identifier_ = value;
+      onChanged();
+      return this;
+    }
+
+    private org.tensorflow.proto.framework.VersionDef version_;
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.VersionDef, org.tensorflow.proto.framework.VersionDef.Builder, org.tensorflow.proto.framework.VersionDefOrBuilder> versionBuilder_;
+    /**
+     * <pre>
+     * Version information from the producer of this SavedUserObject.
+     * </pre>
+     *
+     * <code>.tensorflow.VersionDef version = 2;</code>
+     */
+    public boolean hasVersion() {
+      return versionBuilder_ != null || version_ != null;
+    }
+    /**
+     * <pre>
+     * Version information from the producer of this SavedUserObject.
+     * </pre>
+     *
+     * <code>.tensorflow.VersionDef version = 2;</code>
+     */
+    public org.tensorflow.proto.framework.VersionDef getVersion() {
+      if (versionBuilder_ == null) {
+        return version_ == null ? org.tensorflow.proto.framework.VersionDef.getDefaultInstance() : version_;
+      } else {
+        return versionBuilder_.getMessage();
+      }
+    }
+    /**
+     * <pre>
+     * Version information from the producer of this SavedUserObject.
+     * </pre>
+     *
+     * <code>.tensorflow.VersionDef version = 2;</code>
+     */
+    public Builder setVersion(org.tensorflow.proto.framework.VersionDef value) {
+      if (versionBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        version_ = value;
+        onChanged();
+      } else {
+        versionBuilder_.setMessage(value);
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Version information from the producer of this SavedUserObject.
+     * </pre>
+     *
+     * <code>.tensorflow.VersionDef version = 2;</code>
+     */
+    public Builder setVersion(
+        org.tensorflow.proto.framework.VersionDef.Builder builderForValue) {
+      if (versionBuilder_ == null) {
+        version_ = builderForValue.build();
+        onChanged();
+      } else {
+        versionBuilder_.setMessage(builderForValue.build());
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Version information from the producer of this SavedUserObject.
+     * </pre>
+     *
+     * <code>.tensorflow.VersionDef version = 2;</code>
+     */
+    public Builder mergeVersion(org.tensorflow.proto.framework.VersionDef value) {
+      if (versionBuilder_ == null) {
+        if (version_ != null) {
+          version_ =
+            org.tensorflow.proto.framework.VersionDef.newBuilder(version_).mergeFrom(value).buildPartial();
+        } else {
+          version_ = value;
+        }
+        onChanged();
+      } else {
+        versionBuilder_.mergeFrom(value);
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Version information from the producer of this SavedUserObject.
+     * </pre>
+     *
+     * <code>.tensorflow.VersionDef version = 2;</code>
+     */
+    public Builder clearVersion() {
+      if (versionBuilder_ == null) {
+        version_ = null;
+        onChanged();
+      } else {
+        version_ = null;
+        versionBuilder_ = null;
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Version information from the producer of this SavedUserObject.
+     * </pre>
+     *
+     * <code>.tensorflow.VersionDef version = 2;</code>
+     */
+    public org.tensorflow.proto.framework.VersionDef.Builder getVersionBuilder() {
+      
+      onChanged();
+      return getVersionFieldBuilder().getBuilder();
+    }
+    /**
+     * <pre>
+     * Version information from the producer of this SavedUserObject.
+     * </pre>
+     *
+     * <code>.tensorflow.VersionDef version = 2;</code>
+     */
+    public org.tensorflow.proto.framework.VersionDefOrBuilder getVersionOrBuilder() {
+      if (versionBuilder_ != null) {
+        return versionBuilder_.getMessageOrBuilder();
+      } else {
+        return version_ == null ?
+            org.tensorflow.proto.framework.VersionDef.getDefaultInstance() : version_;
+      }
+    }
+    /**
+     * <pre>
+     * Version information from the producer of this SavedUserObject.
+     * </pre>
+     *
+     * <code>.tensorflow.VersionDef version = 2;</code>
+     */
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.VersionDef, org.tensorflow.proto.framework.VersionDef.Builder, org.tensorflow.proto.framework.VersionDefOrBuilder> 
+        getVersionFieldBuilder() {
+      if (versionBuilder_ == null) {
+        versionBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+            org.tensorflow.proto.framework.VersionDef, org.tensorflow.proto.framework.VersionDef.Builder, org.tensorflow.proto.framework.VersionDefOrBuilder>(
+                getVersion(),
+                getParentForChildren(),
+                isClean());
+        version_ = null;
+      }
+      return versionBuilder_;
+    }
+
+    private java.lang.Object metadata_ = "";
+    /**
+     * <pre>
+     * Initialization-related metadata.
+     * </pre>
+     *
+     * <code>string metadata = 3;</code>
+     */
+    public java.lang.String getMetadata() {
+      java.lang.Object ref = metadata_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        metadata_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Initialization-related metadata.
+     * </pre>
+     *
+     * <code>string metadata = 3;</code>
+     */
+    public com.google.protobuf.ByteString
+        getMetadataBytes() {
+      java.lang.Object ref = metadata_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        metadata_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Initialization-related metadata.
+     * </pre>
+     *
+     * <code>string metadata = 3;</code>
+     */
+    public Builder setMetadata(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      metadata_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Initialization-related metadata.
+     * </pre>
+     *
+     * <code>string metadata = 3;</code>
+     */
+    public Builder clearMetadata() {
+      
+      metadata_ = getDefaultInstance().getMetadata();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Initialization-related metadata.
+     * </pre>
+     *
+     * <code>string metadata = 3;</code>
+     */
+    public Builder setMetadataBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      metadata_ = value;
+      onChanged();
+      return this;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.SavedUserObject)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.SavedUserObject)
+  private static final org.tensorflow.proto.framework.SavedUserObject DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.SavedUserObject();
+  }
+
+  public static org.tensorflow.proto.framework.SavedUserObject getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<SavedUserObject>
+      PARSER = new com.google.protobuf.AbstractParser<SavedUserObject>() {
+    @java.lang.Override
+    public SavedUserObject parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new SavedUserObject(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<SavedUserObject> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<SavedUserObject> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.framework.SavedUserObject getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedUserObjectOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedUserObjectOrBuilder.java
new file mode 100644
index 00000000000..bd7cbadcbd3
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedUserObjectOrBuilder.java
@@ -0,0 +1,70 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/saved_object_graph.proto
+
+package org.tensorflow.proto.framework;
+
+public interface SavedUserObjectOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.SavedUserObject)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <pre>
+   * Corresponds to a registration of the type to use in the loading program.
+   * </pre>
+   *
+   * <code>string identifier = 1;</code>
+   */
+  java.lang.String getIdentifier();
+  /**
+   * <pre>
+   * Corresponds to a registration of the type to use in the loading program.
+   * </pre>
+   *
+   * <code>string identifier = 1;</code>
+   */
+  com.google.protobuf.ByteString
+      getIdentifierBytes();
+
+  /**
+   * <pre>
+   * Version information from the producer of this SavedUserObject.
+   * </pre>
+   *
+   * <code>.tensorflow.VersionDef version = 2;</code>
+   */
+  boolean hasVersion();
+  /**
+   * <pre>
+   * Version information from the producer of this SavedUserObject.
+   * </pre>
+   *
+   * <code>.tensorflow.VersionDef version = 2;</code>
+   */
+  org.tensorflow.proto.framework.VersionDef getVersion();
+  /**
+   * <pre>
+   * Version information from the producer of this SavedUserObject.
+   * </pre>
+   *
+   * <code>.tensorflow.VersionDef version = 2;</code>
+   */
+  org.tensorflow.proto.framework.VersionDefOrBuilder getVersionOrBuilder();
+
+  /**
+   * <pre>
+   * Initialization-related metadata.
+   * </pre>
+   *
+   * <code>string metadata = 3;</code>
+   */
+  java.lang.String getMetadata();
+  /**
+   * <pre>
+   * Initialization-related metadata.
+   * </pre>
+   *
+   * <code>string metadata = 3;</code>
+   */
+  com.google.protobuf.ByteString
+      getMetadataBytes();
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedVariable.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedVariable.java
new file mode 100644
index 00000000000..17ce45af6d7
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedVariable.java
@@ -0,0 +1,1050 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/saved_object_graph.proto
+
+package org.tensorflow.proto.framework;
+
+/**
+ * <pre>
+ * Represents a Variable that is initialized by loading the contents from the
+ * checkpoint.
+ * </pre>
+ *
+ * Protobuf type {@code tensorflow.SavedVariable}
+ */
+public  final class SavedVariable extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.SavedVariable)
+    SavedVariableOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use SavedVariable.newBuilder() to construct.
+  private SavedVariable(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private SavedVariable() {
+    dtype_ = 0;
+    synchronization_ = 0;
+    aggregation_ = 0;
+    name_ = "";
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new SavedVariable();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private SavedVariable(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 8: {
+            int rawValue = input.readEnum();
+
+            dtype_ = rawValue;
+            break;
+          }
+          case 18: {
+            org.tensorflow.proto.framework.TensorShapeProto.Builder subBuilder = null;
+            if (shape_ != null) {
+              subBuilder = shape_.toBuilder();
+            }
+            shape_ = input.readMessage(org.tensorflow.proto.framework.TensorShapeProto.parser(), extensionRegistry);
+            if (subBuilder != null) {
+              subBuilder.mergeFrom(shape_);
+              shape_ = subBuilder.buildPartial();
+            }
+
+            break;
+          }
+          case 24: {
+
+            trainable_ = input.readBool();
+            break;
+          }
+          case 32: {
+            int rawValue = input.readEnum();
+
+            synchronization_ = rawValue;
+            break;
+          }
+          case 40: {
+            int rawValue = input.readEnum();
+
+            aggregation_ = rawValue;
+            break;
+          }
+          case 50: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            name_ = s;
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.framework.SavedObjectGraphProtos.internal_static_tensorflow_SavedVariable_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.framework.SavedObjectGraphProtos.internal_static_tensorflow_SavedVariable_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.framework.SavedVariable.class, org.tensorflow.proto.framework.SavedVariable.Builder.class);
+  }
+
+  public static final int DTYPE_FIELD_NUMBER = 1;
+  private int dtype_;
+  /**
+   * <code>.tensorflow.DataType dtype = 1;</code>
+   */
+  public int getDtypeValue() {
+    return dtype_;
+  }
+  /**
+   * <code>.tensorflow.DataType dtype = 1;</code>
+   */
+  public org.tensorflow.proto.framework.DataType getDtype() {
+    @SuppressWarnings("deprecation")
+    org.tensorflow.proto.framework.DataType result = org.tensorflow.proto.framework.DataType.valueOf(dtype_);
+    return result == null ? org.tensorflow.proto.framework.DataType.UNRECOGNIZED : result;
+  }
+
+  public static final int SHAPE_FIELD_NUMBER = 2;
+  private org.tensorflow.proto.framework.TensorShapeProto shape_;
+  /**
+   * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+   */
+  public boolean hasShape() {
+    return shape_ != null;
+  }
+  /**
+   * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+   */
+  public org.tensorflow.proto.framework.TensorShapeProto getShape() {
+    return shape_ == null ? org.tensorflow.proto.framework.TensorShapeProto.getDefaultInstance() : shape_;
+  }
+  /**
+   * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+   */
+  public org.tensorflow.proto.framework.TensorShapeProtoOrBuilder getShapeOrBuilder() {
+    return getShape();
+  }
+
+  public static final int TRAINABLE_FIELD_NUMBER = 3;
+  private boolean trainable_;
+  /**
+   * <code>bool trainable = 3;</code>
+   */
+  public boolean getTrainable() {
+    return trainable_;
+  }
+
+  public static final int SYNCHRONIZATION_FIELD_NUMBER = 4;
+  private int synchronization_;
+  /**
+   * <code>.tensorflow.VariableSynchronization synchronization = 4;</code>
+   */
+  public int getSynchronizationValue() {
+    return synchronization_;
+  }
+  /**
+   * <code>.tensorflow.VariableSynchronization synchronization = 4;</code>
+   */
+  public org.tensorflow.proto.framework.VariableSynchronization getSynchronization() {
+    @SuppressWarnings("deprecation")
+    org.tensorflow.proto.framework.VariableSynchronization result = org.tensorflow.proto.framework.VariableSynchronization.valueOf(synchronization_);
+    return result == null ? org.tensorflow.proto.framework.VariableSynchronization.UNRECOGNIZED : result;
+  }
+
+  public static final int AGGREGATION_FIELD_NUMBER = 5;
+  private int aggregation_;
+  /**
+   * <code>.tensorflow.VariableAggregation aggregation = 5;</code>
+   */
+  public int getAggregationValue() {
+    return aggregation_;
+  }
+  /**
+   * <code>.tensorflow.VariableAggregation aggregation = 5;</code>
+   */
+  public org.tensorflow.proto.framework.VariableAggregation getAggregation() {
+    @SuppressWarnings("deprecation")
+    org.tensorflow.proto.framework.VariableAggregation result = org.tensorflow.proto.framework.VariableAggregation.valueOf(aggregation_);
+    return result == null ? org.tensorflow.proto.framework.VariableAggregation.UNRECOGNIZED : result;
+  }
+
+  public static final int NAME_FIELD_NUMBER = 6;
+  private volatile java.lang.Object name_;
+  /**
+   * <code>string name = 6;</code>
+   */
+  public java.lang.String getName() {
+    java.lang.Object ref = name_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      name_ = s;
+      return s;
+    }
+  }
+  /**
+   * <code>string name = 6;</code>
+   */
+  public com.google.protobuf.ByteString
+      getNameBytes() {
+    java.lang.Object ref = name_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      name_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    if (dtype_ != org.tensorflow.proto.framework.DataType.DT_INVALID.getNumber()) {
+      output.writeEnum(1, dtype_);
+    }
+    if (shape_ != null) {
+      output.writeMessage(2, getShape());
+    }
+    if (trainable_ != false) {
+      output.writeBool(3, trainable_);
+    }
+    if (synchronization_ != org.tensorflow.proto.framework.VariableSynchronization.VARIABLE_SYNCHRONIZATION_AUTO.getNumber()) {
+      output.writeEnum(4, synchronization_);
+    }
+    if (aggregation_ != org.tensorflow.proto.framework.VariableAggregation.VARIABLE_AGGREGATION_NONE.getNumber()) {
+      output.writeEnum(5, aggregation_);
+    }
+    if (!getNameBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 6, name_);
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (dtype_ != org.tensorflow.proto.framework.DataType.DT_INVALID.getNumber()) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeEnumSize(1, dtype_);
+    }
+    if (shape_ != null) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(2, getShape());
+    }
+    if (trainable_ != false) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeBoolSize(3, trainable_);
+    }
+    if (synchronization_ != org.tensorflow.proto.framework.VariableSynchronization.VARIABLE_SYNCHRONIZATION_AUTO.getNumber()) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeEnumSize(4, synchronization_);
+    }
+    if (aggregation_ != org.tensorflow.proto.framework.VariableAggregation.VARIABLE_AGGREGATION_NONE.getNumber()) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeEnumSize(5, aggregation_);
+    }
+    if (!getNameBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(6, name_);
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.framework.SavedVariable)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.framework.SavedVariable other = (org.tensorflow.proto.framework.SavedVariable) obj;
+
+    if (dtype_ != other.dtype_) return false;
+    if (hasShape() != other.hasShape()) return false;
+    if (hasShape()) {
+      if (!getShape()
+          .equals(other.getShape())) return false;
+    }
+    if (getTrainable()
+        != other.getTrainable()) return false;
+    if (synchronization_ != other.synchronization_) return false;
+    if (aggregation_ != other.aggregation_) return false;
+    if (!getName()
+        .equals(other.getName())) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    hash = (37 * hash) + DTYPE_FIELD_NUMBER;
+    hash = (53 * hash) + dtype_;
+    if (hasShape()) {
+      hash = (37 * hash) + SHAPE_FIELD_NUMBER;
+      hash = (53 * hash) + getShape().hashCode();
+    }
+    hash = (37 * hash) + TRAINABLE_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashBoolean(
+        getTrainable());
+    hash = (37 * hash) + SYNCHRONIZATION_FIELD_NUMBER;
+    hash = (53 * hash) + synchronization_;
+    hash = (37 * hash) + AGGREGATION_FIELD_NUMBER;
+    hash = (53 * hash) + aggregation_;
+    hash = (37 * hash) + NAME_FIELD_NUMBER;
+    hash = (53 * hash) + getName().hashCode();
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.framework.SavedVariable parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.SavedVariable parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.SavedVariable parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.SavedVariable parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.SavedVariable parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.SavedVariable parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.SavedVariable parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.SavedVariable parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.SavedVariable parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.SavedVariable parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.SavedVariable parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.SavedVariable parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.framework.SavedVariable prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * <pre>
+   * Represents a Variable that is initialized by loading the contents from the
+   * checkpoint.
+   * </pre>
+   *
+   * Protobuf type {@code tensorflow.SavedVariable}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.SavedVariable)
+      org.tensorflow.proto.framework.SavedVariableOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.SavedObjectGraphProtos.internal_static_tensorflow_SavedVariable_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.SavedObjectGraphProtos.internal_static_tensorflow_SavedVariable_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.SavedVariable.class, org.tensorflow.proto.framework.SavedVariable.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.framework.SavedVariable.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      dtype_ = 0;
+
+      if (shapeBuilder_ == null) {
+        shape_ = null;
+      } else {
+        shape_ = null;
+        shapeBuilder_ = null;
+      }
+      trainable_ = false;
+
+      synchronization_ = 0;
+
+      aggregation_ = 0;
+
+      name_ = "";
+
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.framework.SavedObjectGraphProtos.internal_static_tensorflow_SavedVariable_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.SavedVariable getDefaultInstanceForType() {
+      return org.tensorflow.proto.framework.SavedVariable.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.SavedVariable build() {
+      org.tensorflow.proto.framework.SavedVariable result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.SavedVariable buildPartial() {
+      org.tensorflow.proto.framework.SavedVariable result = new org.tensorflow.proto.framework.SavedVariable(this);
+      result.dtype_ = dtype_;
+      if (shapeBuilder_ == null) {
+        result.shape_ = shape_;
+      } else {
+        result.shape_ = shapeBuilder_.build();
+      }
+      result.trainable_ = trainable_;
+      result.synchronization_ = synchronization_;
+      result.aggregation_ = aggregation_;
+      result.name_ = name_;
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.framework.SavedVariable) {
+        return mergeFrom((org.tensorflow.proto.framework.SavedVariable)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.framework.SavedVariable other) {
+      if (other == org.tensorflow.proto.framework.SavedVariable.getDefaultInstance()) return this;
+      if (other.dtype_ != 0) {
+        setDtypeValue(other.getDtypeValue());
+      }
+      if (other.hasShape()) {
+        mergeShape(other.getShape());
+      }
+      if (other.getTrainable() != false) {
+        setTrainable(other.getTrainable());
+      }
+      if (other.synchronization_ != 0) {
+        setSynchronizationValue(other.getSynchronizationValue());
+      }
+      if (other.aggregation_ != 0) {
+        setAggregationValue(other.getAggregationValue());
+      }
+      if (!other.getName().isEmpty()) {
+        name_ = other.name_;
+        onChanged();
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.framework.SavedVariable parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.framework.SavedVariable) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+
+    private int dtype_ = 0;
+    /**
+     * <code>.tensorflow.DataType dtype = 1;</code>
+     */
+    public int getDtypeValue() {
+      return dtype_;
+    }
+    /**
+     * <code>.tensorflow.DataType dtype = 1;</code>
+     */
+    public Builder setDtypeValue(int value) {
+      dtype_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>.tensorflow.DataType dtype = 1;</code>
+     */
+    public org.tensorflow.proto.framework.DataType getDtype() {
+      @SuppressWarnings("deprecation")
+      org.tensorflow.proto.framework.DataType result = org.tensorflow.proto.framework.DataType.valueOf(dtype_);
+      return result == null ? org.tensorflow.proto.framework.DataType.UNRECOGNIZED : result;
+    }
+    /**
+     * <code>.tensorflow.DataType dtype = 1;</code>
+     */
+    public Builder setDtype(org.tensorflow.proto.framework.DataType value) {
+      if (value == null) {
+        throw new NullPointerException();
+      }
+      
+      dtype_ = value.getNumber();
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>.tensorflow.DataType dtype = 1;</code>
+     */
+    public Builder clearDtype() {
+      
+      dtype_ = 0;
+      onChanged();
+      return this;
+    }
+
+    private org.tensorflow.proto.framework.TensorShapeProto shape_;
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.TensorShapeProto, org.tensorflow.proto.framework.TensorShapeProto.Builder, org.tensorflow.proto.framework.TensorShapeProtoOrBuilder> shapeBuilder_;
+    /**
+     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+     */
+    public boolean hasShape() {
+      return shapeBuilder_ != null || shape_ != null;
+    }
+    /**
+     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+     */
+    public org.tensorflow.proto.framework.TensorShapeProto getShape() {
+      if (shapeBuilder_ == null) {
+        return shape_ == null ? org.tensorflow.proto.framework.TensorShapeProto.getDefaultInstance() : shape_;
+      } else {
+        return shapeBuilder_.getMessage();
+      }
+    }
+    /**
+     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+     */
+    public Builder setShape(org.tensorflow.proto.framework.TensorShapeProto value) {
+      if (shapeBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        shape_ = value;
+        onChanged();
+      } else {
+        shapeBuilder_.setMessage(value);
+      }
+
+      return this;
+    }
+    /**
+     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+     */
+    public Builder setShape(
+        org.tensorflow.proto.framework.TensorShapeProto.Builder builderForValue) {
+      if (shapeBuilder_ == null) {
+        shape_ = builderForValue.build();
+        onChanged();
+      } else {
+        shapeBuilder_.setMessage(builderForValue.build());
+      }
+
+      return this;
+    }
+    /**
+     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+     */
+    public Builder mergeShape(org.tensorflow.proto.framework.TensorShapeProto value) {
+      if (shapeBuilder_ == null) {
+        if (shape_ != null) {
+          shape_ =
+            org.tensorflow.proto.framework.TensorShapeProto.newBuilder(shape_).mergeFrom(value).buildPartial();
+        } else {
+          shape_ = value;
+        }
+        onChanged();
+      } else {
+        shapeBuilder_.mergeFrom(value);
+      }
+
+      return this;
+    }
+    /**
+     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+     */
+    public Builder clearShape() {
+      if (shapeBuilder_ == null) {
+        shape_ = null;
+        onChanged();
+      } else {
+        shape_ = null;
+        shapeBuilder_ = null;
+      }
+
+      return this;
+    }
+    /**
+     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+     */
+    public org.tensorflow.proto.framework.TensorShapeProto.Builder getShapeBuilder() {
+      
+      onChanged();
+      return getShapeFieldBuilder().getBuilder();
+    }
+    /**
+     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+     */
+    public org.tensorflow.proto.framework.TensorShapeProtoOrBuilder getShapeOrBuilder() {
+      if (shapeBuilder_ != null) {
+        return shapeBuilder_.getMessageOrBuilder();
+      } else {
+        return shape_ == null ?
+            org.tensorflow.proto.framework.TensorShapeProto.getDefaultInstance() : shape_;
+      }
+    }
+    /**
+     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+     */
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.TensorShapeProto, org.tensorflow.proto.framework.TensorShapeProto.Builder, org.tensorflow.proto.framework.TensorShapeProtoOrBuilder> 
+        getShapeFieldBuilder() {
+      if (shapeBuilder_ == null) {
+        shapeBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+            org.tensorflow.proto.framework.TensorShapeProto, org.tensorflow.proto.framework.TensorShapeProto.Builder, org.tensorflow.proto.framework.TensorShapeProtoOrBuilder>(
+                getShape(),
+                getParentForChildren(),
+                isClean());
+        shape_ = null;
+      }
+      return shapeBuilder_;
+    }
+
+    private boolean trainable_ ;
+    /**
+     * <code>bool trainable = 3;</code>
+     */
+    public boolean getTrainable() {
+      return trainable_;
+    }
+    /**
+     * <code>bool trainable = 3;</code>
+     */
+    public Builder setTrainable(boolean value) {
+      
+      trainable_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>bool trainable = 3;</code>
+     */
+    public Builder clearTrainable() {
+      
+      trainable_ = false;
+      onChanged();
+      return this;
+    }
+
+    private int synchronization_ = 0;
+    /**
+     * <code>.tensorflow.VariableSynchronization synchronization = 4;</code>
+     */
+    public int getSynchronizationValue() {
+      return synchronization_;
+    }
+    /**
+     * <code>.tensorflow.VariableSynchronization synchronization = 4;</code>
+     */
+    public Builder setSynchronizationValue(int value) {
+      synchronization_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>.tensorflow.VariableSynchronization synchronization = 4;</code>
+     */
+    public org.tensorflow.proto.framework.VariableSynchronization getSynchronization() {
+      @SuppressWarnings("deprecation")
+      org.tensorflow.proto.framework.VariableSynchronization result = org.tensorflow.proto.framework.VariableSynchronization.valueOf(synchronization_);
+      return result == null ? org.tensorflow.proto.framework.VariableSynchronization.UNRECOGNIZED : result;
+    }
+    /**
+     * <code>.tensorflow.VariableSynchronization synchronization = 4;</code>
+     */
+    public Builder setSynchronization(org.tensorflow.proto.framework.VariableSynchronization value) {
+      if (value == null) {
+        throw new NullPointerException();
+      }
+      
+      synchronization_ = value.getNumber();
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>.tensorflow.VariableSynchronization synchronization = 4;</code>
+     */
+    public Builder clearSynchronization() {
+      
+      synchronization_ = 0;
+      onChanged();
+      return this;
+    }
+
+    private int aggregation_ = 0;
+    /**
+     * <code>.tensorflow.VariableAggregation aggregation = 5;</code>
+     */
+    public int getAggregationValue() {
+      return aggregation_;
+    }
+    /**
+     * <code>.tensorflow.VariableAggregation aggregation = 5;</code>
+     */
+    public Builder setAggregationValue(int value) {
+      aggregation_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>.tensorflow.VariableAggregation aggregation = 5;</code>
+     */
+    public org.tensorflow.proto.framework.VariableAggregation getAggregation() {
+      @SuppressWarnings("deprecation")
+      org.tensorflow.proto.framework.VariableAggregation result = org.tensorflow.proto.framework.VariableAggregation.valueOf(aggregation_);
+      return result == null ? org.tensorflow.proto.framework.VariableAggregation.UNRECOGNIZED : result;
+    }
+    /**
+     * <code>.tensorflow.VariableAggregation aggregation = 5;</code>
+     */
+    public Builder setAggregation(org.tensorflow.proto.framework.VariableAggregation value) {
+      if (value == null) {
+        throw new NullPointerException();
+      }
+      
+      aggregation_ = value.getNumber();
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>.tensorflow.VariableAggregation aggregation = 5;</code>
+     */
+    public Builder clearAggregation() {
+      
+      aggregation_ = 0;
+      onChanged();
+      return this;
+    }
+
+    private java.lang.Object name_ = "";
+    /**
+     * <code>string name = 6;</code>
+     */
+    public java.lang.String getName() {
+      java.lang.Object ref = name_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        name_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <code>string name = 6;</code>
+     */
+    public com.google.protobuf.ByteString
+        getNameBytes() {
+      java.lang.Object ref = name_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        name_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <code>string name = 6;</code>
+     */
+    public Builder setName(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      name_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>string name = 6;</code>
+     */
+    public Builder clearName() {
+      
+      name_ = getDefaultInstance().getName();
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>string name = 6;</code>
+     */
+    public Builder setNameBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      name_ = value;
+      onChanged();
+      return this;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.SavedVariable)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.SavedVariable)
+  private static final org.tensorflow.proto.framework.SavedVariable DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.SavedVariable();
+  }
+
+  public static org.tensorflow.proto.framework.SavedVariable getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<SavedVariable>
+      PARSER = new com.google.protobuf.AbstractParser<SavedVariable>() {
+    @java.lang.Override
+    public SavedVariable parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new SavedVariable(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<SavedVariable> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<SavedVariable> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.framework.SavedVariable getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedVariableOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedVariableOrBuilder.java
new file mode 100644
index 00000000000..27da7b6abfa
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SavedVariableOrBuilder.java
@@ -0,0 +1,64 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/saved_object_graph.proto
+
+package org.tensorflow.proto.framework;
+
+public interface SavedVariableOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.SavedVariable)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <code>.tensorflow.DataType dtype = 1;</code>
+   */
+  int getDtypeValue();
+  /**
+   * <code>.tensorflow.DataType dtype = 1;</code>
+   */
+  org.tensorflow.proto.framework.DataType getDtype();
+
+  /**
+   * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+   */
+  boolean hasShape();
+  /**
+   * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+   */
+  org.tensorflow.proto.framework.TensorShapeProto getShape();
+  /**
+   * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+   */
+  org.tensorflow.proto.framework.TensorShapeProtoOrBuilder getShapeOrBuilder();
+
+  /**
+   * <code>bool trainable = 3;</code>
+   */
+  boolean getTrainable();
+
+  /**
+   * <code>.tensorflow.VariableSynchronization synchronization = 4;</code>
+   */
+  int getSynchronizationValue();
+  /**
+   * <code>.tensorflow.VariableSynchronization synchronization = 4;</code>
+   */
+  org.tensorflow.proto.framework.VariableSynchronization getSynchronization();
+
+  /**
+   * <code>.tensorflow.VariableAggregation aggregation = 5;</code>
+   */
+  int getAggregationValue();
+  /**
+   * <code>.tensorflow.VariableAggregation aggregation = 5;</code>
+   */
+  org.tensorflow.proto.framework.VariableAggregation getAggregation();
+
+  /**
+   * <code>string name = 6;</code>
+   */
+  java.lang.String getName();
+  /**
+   * <code>string name = 6;</code>
+   */
+  com.google.protobuf.ByteString
+      getNameBytes();
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ScopedAllocatorOptions.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ScopedAllocatorOptions.java
new file mode 100644
index 00000000000..37bf6f9c4d1
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ScopedAllocatorOptions.java
@@ -0,0 +1,644 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/rewriter_config.proto
+
+package org.tensorflow.proto.framework;
+
+/**
+ * Protobuf type {@code tensorflow.ScopedAllocatorOptions}
+ */
+public  final class ScopedAllocatorOptions extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.ScopedAllocatorOptions)
+    ScopedAllocatorOptionsOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use ScopedAllocatorOptions.newBuilder() to construct.
+  private ScopedAllocatorOptions(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private ScopedAllocatorOptions() {
+    enableOp_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new ScopedAllocatorOptions();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private ScopedAllocatorOptions(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    int mutable_bitField0_ = 0;
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 10: {
+            java.lang.String s = input.readStringRequireUtf8();
+            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
+              enableOp_ = new com.google.protobuf.LazyStringArrayList();
+              mutable_bitField0_ |= 0x00000001;
+            }
+            enableOp_.add(s);
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      if (((mutable_bitField0_ & 0x00000001) != 0)) {
+        enableOp_ = enableOp_.getUnmodifiableView();
+      }
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.framework.RewriterConfigProtos.internal_static_tensorflow_ScopedAllocatorOptions_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.framework.RewriterConfigProtos.internal_static_tensorflow_ScopedAllocatorOptions_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.framework.ScopedAllocatorOptions.class, org.tensorflow.proto.framework.ScopedAllocatorOptions.Builder.class);
+  }
+
+  public static final int ENABLE_OP_FIELD_NUMBER = 1;
+  private com.google.protobuf.LazyStringList enableOp_;
+  /**
+   * <pre>
+   * If present, only perform optimization for these ops.
+   * </pre>
+   *
+   * <code>repeated string enable_op = 1;</code>
+   */
+  public com.google.protobuf.ProtocolStringList
+      getEnableOpList() {
+    return enableOp_;
+  }
+  /**
+   * <pre>
+   * If present, only perform optimization for these ops.
+   * </pre>
+   *
+   * <code>repeated string enable_op = 1;</code>
+   */
+  public int getEnableOpCount() {
+    return enableOp_.size();
+  }
+  /**
+   * <pre>
+   * If present, only perform optimization for these ops.
+   * </pre>
+   *
+   * <code>repeated string enable_op = 1;</code>
+   */
+  public java.lang.String getEnableOp(int index) {
+    return enableOp_.get(index);
+  }
+  /**
+   * <pre>
+   * If present, only perform optimization for these ops.
+   * </pre>
+   *
+   * <code>repeated string enable_op = 1;</code>
+   */
+  public com.google.protobuf.ByteString
+      getEnableOpBytes(int index) {
+    return enableOp_.getByteString(index);
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    for (int i = 0; i < enableOp_.size(); i++) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 1, enableOp_.getRaw(i));
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    {
+      int dataSize = 0;
+      for (int i = 0; i < enableOp_.size(); i++) {
+        dataSize += computeStringSizeNoTag(enableOp_.getRaw(i));
+      }
+      size += dataSize;
+      size += 1 * getEnableOpList().size();
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.framework.ScopedAllocatorOptions)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.framework.ScopedAllocatorOptions other = (org.tensorflow.proto.framework.ScopedAllocatorOptions) obj;
+
+    if (!getEnableOpList()
+        .equals(other.getEnableOpList())) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    if (getEnableOpCount() > 0) {
+      hash = (37 * hash) + ENABLE_OP_FIELD_NUMBER;
+      hash = (53 * hash) + getEnableOpList().hashCode();
+    }
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.framework.ScopedAllocatorOptions parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.ScopedAllocatorOptions parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.ScopedAllocatorOptions parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.ScopedAllocatorOptions parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.ScopedAllocatorOptions parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.ScopedAllocatorOptions parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.ScopedAllocatorOptions parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.ScopedAllocatorOptions parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.ScopedAllocatorOptions parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.ScopedAllocatorOptions parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.ScopedAllocatorOptions parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.ScopedAllocatorOptions parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.framework.ScopedAllocatorOptions prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * Protobuf type {@code tensorflow.ScopedAllocatorOptions}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.ScopedAllocatorOptions)
+      org.tensorflow.proto.framework.ScopedAllocatorOptionsOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.RewriterConfigProtos.internal_static_tensorflow_ScopedAllocatorOptions_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.RewriterConfigProtos.internal_static_tensorflow_ScopedAllocatorOptions_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.ScopedAllocatorOptions.class, org.tensorflow.proto.framework.ScopedAllocatorOptions.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.framework.ScopedAllocatorOptions.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      enableOp_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+      bitField0_ = (bitField0_ & ~0x00000001);
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.framework.RewriterConfigProtos.internal_static_tensorflow_ScopedAllocatorOptions_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.ScopedAllocatorOptions getDefaultInstanceForType() {
+      return org.tensorflow.proto.framework.ScopedAllocatorOptions.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.ScopedAllocatorOptions build() {
+      org.tensorflow.proto.framework.ScopedAllocatorOptions result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.ScopedAllocatorOptions buildPartial() {
+      org.tensorflow.proto.framework.ScopedAllocatorOptions result = new org.tensorflow.proto.framework.ScopedAllocatorOptions(this);
+      int from_bitField0_ = bitField0_;
+      if (((bitField0_ & 0x00000001) != 0)) {
+        enableOp_ = enableOp_.getUnmodifiableView();
+        bitField0_ = (bitField0_ & ~0x00000001);
+      }
+      result.enableOp_ = enableOp_;
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.framework.ScopedAllocatorOptions) {
+        return mergeFrom((org.tensorflow.proto.framework.ScopedAllocatorOptions)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.framework.ScopedAllocatorOptions other) {
+      if (other == org.tensorflow.proto.framework.ScopedAllocatorOptions.getDefaultInstance()) return this;
+      if (!other.enableOp_.isEmpty()) {
+        if (enableOp_.isEmpty()) {
+          enableOp_ = other.enableOp_;
+          bitField0_ = (bitField0_ & ~0x00000001);
+        } else {
+          ensureEnableOpIsMutable();
+          enableOp_.addAll(other.enableOp_);
+        }
+        onChanged();
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.framework.ScopedAllocatorOptions parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.framework.ScopedAllocatorOptions) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+    private int bitField0_;
+
+    private com.google.protobuf.LazyStringList enableOp_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+    private void ensureEnableOpIsMutable() {
+      if (!((bitField0_ & 0x00000001) != 0)) {
+        enableOp_ = new com.google.protobuf.LazyStringArrayList(enableOp_);
+        bitField0_ |= 0x00000001;
+       }
+    }
+    /**
+     * <pre>
+     * If present, only perform optimization for these ops.
+     * </pre>
+     *
+     * <code>repeated string enable_op = 1;</code>
+     */
+    public com.google.protobuf.ProtocolStringList
+        getEnableOpList() {
+      return enableOp_.getUnmodifiableView();
+    }
+    /**
+     * <pre>
+     * If present, only perform optimization for these ops.
+     * </pre>
+     *
+     * <code>repeated string enable_op = 1;</code>
+     */
+    public int getEnableOpCount() {
+      return enableOp_.size();
+    }
+    /**
+     * <pre>
+     * If present, only perform optimization for these ops.
+     * </pre>
+     *
+     * <code>repeated string enable_op = 1;</code>
+     */
+    public java.lang.String getEnableOp(int index) {
+      return enableOp_.get(index);
+    }
+    /**
+     * <pre>
+     * If present, only perform optimization for these ops.
+     * </pre>
+     *
+     * <code>repeated string enable_op = 1;</code>
+     */
+    public com.google.protobuf.ByteString
+        getEnableOpBytes(int index) {
+      return enableOp_.getByteString(index);
+    }
+    /**
+     * <pre>
+     * If present, only perform optimization for these ops.
+     * </pre>
+     *
+     * <code>repeated string enable_op = 1;</code>
+     */
+    public Builder setEnableOp(
+        int index, java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  ensureEnableOpIsMutable();
+      enableOp_.set(index, value);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * If present, only perform optimization for these ops.
+     * </pre>
+     *
+     * <code>repeated string enable_op = 1;</code>
+     */
+    public Builder addEnableOp(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  ensureEnableOpIsMutable();
+      enableOp_.add(value);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * If present, only perform optimization for these ops.
+     * </pre>
+     *
+     * <code>repeated string enable_op = 1;</code>
+     */
+    public Builder addAllEnableOp(
+        java.lang.Iterable<java.lang.String> values) {
+      ensureEnableOpIsMutable();
+      com.google.protobuf.AbstractMessageLite.Builder.addAll(
+          values, enableOp_);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * If present, only perform optimization for these ops.
+     * </pre>
+     *
+     * <code>repeated string enable_op = 1;</code>
+     */
+    public Builder clearEnableOp() {
+      enableOp_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+      bitField0_ = (bitField0_ & ~0x00000001);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * If present, only perform optimization for these ops.
+     * </pre>
+     *
+     * <code>repeated string enable_op = 1;</code>
+     */
+    public Builder addEnableOpBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      ensureEnableOpIsMutable();
+      enableOp_.add(value);
+      onChanged();
+      return this;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.ScopedAllocatorOptions)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.ScopedAllocatorOptions)
+  private static final org.tensorflow.proto.framework.ScopedAllocatorOptions DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.ScopedAllocatorOptions();
+  }
+
+  public static org.tensorflow.proto.framework.ScopedAllocatorOptions getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<ScopedAllocatorOptions>
+      PARSER = new com.google.protobuf.AbstractParser<ScopedAllocatorOptions>() {
+    @java.lang.Override
+    public ScopedAllocatorOptions parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new ScopedAllocatorOptions(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<ScopedAllocatorOptions> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<ScopedAllocatorOptions> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.framework.ScopedAllocatorOptions getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ScopedAllocatorOptionsOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ScopedAllocatorOptionsOrBuilder.java
new file mode 100644
index 00000000000..df536ed18a9
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ScopedAllocatorOptionsOrBuilder.java
@@ -0,0 +1,44 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/rewriter_config.proto
+
+package org.tensorflow.proto.framework;
+
+public interface ScopedAllocatorOptionsOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.ScopedAllocatorOptions)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <pre>
+   * If present, only perform optimization for these ops.
+   * </pre>
+   *
+   * <code>repeated string enable_op = 1;</code>
+   */
+  java.util.List<java.lang.String>
+      getEnableOpList();
+  /**
+   * <pre>
+   * If present, only perform optimization for these ops.
+   * </pre>
+   *
+   * <code>repeated string enable_op = 1;</code>
+   */
+  int getEnableOpCount();
+  /**
+   * <pre>
+   * If present, only perform optimization for these ops.
+   * </pre>
+   *
+   * <code>repeated string enable_op = 1;</code>
+   */
+  java.lang.String getEnableOp(int index);
+  /**
+   * <pre>
+   * If present, only perform optimization for these ops.
+   * </pre>
+   *
+   * <code>repeated string enable_op = 1;</code>
+   */
+  com.google.protobuf.ByteString
+      getEnableOpBytes(int index);
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SessionMetadata.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SessionMetadata.java
new file mode 100644
index 00000000000..a825437b468
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SessionMetadata.java
@@ -0,0 +1,636 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/config.proto
+
+package org.tensorflow.proto.framework;
+
+/**
+ * <pre>
+ * Metadata about the session.
+ * This can be used by the runtime and the Ops for debugging, monitoring, etc.
+ * The (name, version) tuple is expected to be a unique identifier for
+ * sessions within the same process.
+ * NOTE: This is currently used and propagated only by the direct session.
+ * </pre>
+ *
+ * Protobuf type {@code tensorflow.SessionMetadata}
+ */
+public  final class SessionMetadata extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.SessionMetadata)
+    SessionMetadataOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use SessionMetadata.newBuilder() to construct.
+  private SessionMetadata(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private SessionMetadata() {
+    name_ = "";
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new SessionMetadata();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private SessionMetadata(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 10: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            name_ = s;
+            break;
+          }
+          case 16: {
+
+            version_ = input.readInt64();
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_SessionMetadata_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_SessionMetadata_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.framework.SessionMetadata.class, org.tensorflow.proto.framework.SessionMetadata.Builder.class);
+  }
+
+  public static final int NAME_FIELD_NUMBER = 1;
+  private volatile java.lang.Object name_;
+  /**
+   * <code>string name = 1;</code>
+   */
+  public java.lang.String getName() {
+    java.lang.Object ref = name_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      name_ = s;
+      return s;
+    }
+  }
+  /**
+   * <code>string name = 1;</code>
+   */
+  public com.google.protobuf.ByteString
+      getNameBytes() {
+    java.lang.Object ref = name_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      name_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int VERSION_FIELD_NUMBER = 2;
+  private long version_;
+  /**
+   * <pre>
+   * The version is optional. If set, needs to be &gt;= 0.
+   * </pre>
+   *
+   * <code>int64 version = 2;</code>
+   */
+  public long getVersion() {
+    return version_;
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    if (!getNameBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 1, name_);
+    }
+    if (version_ != 0L) {
+      output.writeInt64(2, version_);
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (!getNameBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, name_);
+    }
+    if (version_ != 0L) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt64Size(2, version_);
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.framework.SessionMetadata)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.framework.SessionMetadata other = (org.tensorflow.proto.framework.SessionMetadata) obj;
+
+    if (!getName()
+        .equals(other.getName())) return false;
+    if (getVersion()
+        != other.getVersion()) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    hash = (37 * hash) + NAME_FIELD_NUMBER;
+    hash = (53 * hash) + getName().hashCode();
+    hash = (37 * hash) + VERSION_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+        getVersion());
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.framework.SessionMetadata parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.SessionMetadata parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.SessionMetadata parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.SessionMetadata parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.SessionMetadata parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.SessionMetadata parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.SessionMetadata parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.SessionMetadata parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.SessionMetadata parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.SessionMetadata parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.SessionMetadata parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.SessionMetadata parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.framework.SessionMetadata prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * <pre>
+   * Metadata about the session.
+   * This can be used by the runtime and the Ops for debugging, monitoring, etc.
+   * The (name, version) tuple is expected to be a unique identifier for
+   * sessions within the same process.
+   * NOTE: This is currently used and propagated only by the direct session.
+   * </pre>
+   *
+   * Protobuf type {@code tensorflow.SessionMetadata}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.SessionMetadata)
+      org.tensorflow.proto.framework.SessionMetadataOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_SessionMetadata_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_SessionMetadata_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.SessionMetadata.class, org.tensorflow.proto.framework.SessionMetadata.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.framework.SessionMetadata.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      name_ = "";
+
+      version_ = 0L;
+
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_SessionMetadata_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.SessionMetadata getDefaultInstanceForType() {
+      return org.tensorflow.proto.framework.SessionMetadata.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.SessionMetadata build() {
+      org.tensorflow.proto.framework.SessionMetadata result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.SessionMetadata buildPartial() {
+      org.tensorflow.proto.framework.SessionMetadata result = new org.tensorflow.proto.framework.SessionMetadata(this);
+      result.name_ = name_;
+      result.version_ = version_;
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.framework.SessionMetadata) {
+        return mergeFrom((org.tensorflow.proto.framework.SessionMetadata)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.framework.SessionMetadata other) {
+      if (other == org.tensorflow.proto.framework.SessionMetadata.getDefaultInstance()) return this;
+      if (!other.getName().isEmpty()) {
+        name_ = other.name_;
+        onChanged();
+      }
+      if (other.getVersion() != 0L) {
+        setVersion(other.getVersion());
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.framework.SessionMetadata parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.framework.SessionMetadata) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+
+    private java.lang.Object name_ = "";
+    /**
+     * <code>string name = 1;</code>
+     */
+    public java.lang.String getName() {
+      java.lang.Object ref = name_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        name_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <code>string name = 1;</code>
+     */
+    public com.google.protobuf.ByteString
+        getNameBytes() {
+      java.lang.Object ref = name_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        name_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <code>string name = 1;</code>
+     */
+    public Builder setName(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      name_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>string name = 1;</code>
+     */
+    public Builder clearName() {
+      
+      name_ = getDefaultInstance().getName();
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>string name = 1;</code>
+     */
+    public Builder setNameBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      name_ = value;
+      onChanged();
+      return this;
+    }
+
+    private long version_ ;
+    /**
+     * <pre>
+     * The version is optional. If set, needs to be &gt;= 0.
+     * </pre>
+     *
+     * <code>int64 version = 2;</code>
+     */
+    public long getVersion() {
+      return version_;
+    }
+    /**
+     * <pre>
+     * The version is optional. If set, needs to be &gt;= 0.
+     * </pre>
+     *
+     * <code>int64 version = 2;</code>
+     */
+    public Builder setVersion(long value) {
+      
+      version_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * The version is optional. If set, needs to be &gt;= 0.
+     * </pre>
+     *
+     * <code>int64 version = 2;</code>
+     */
+    public Builder clearVersion() {
+      
+      version_ = 0L;
+      onChanged();
+      return this;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.SessionMetadata)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.SessionMetadata)
+  private static final org.tensorflow.proto.framework.SessionMetadata DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.SessionMetadata();
+  }
+
+  public static org.tensorflow.proto.framework.SessionMetadata getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<SessionMetadata>
+      PARSER = new com.google.protobuf.AbstractParser<SessionMetadata>() {
+    @java.lang.Override
+    public SessionMetadata parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new SessionMetadata(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<SessionMetadata> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<SessionMetadata> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.framework.SessionMetadata getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SessionMetadataOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SessionMetadataOrBuilder.java
new file mode 100644
index 00000000000..eae7295e772
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SessionMetadataOrBuilder.java
@@ -0,0 +1,28 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/config.proto
+
+package org.tensorflow.proto.framework;
+
+public interface SessionMetadataOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.SessionMetadata)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <code>string name = 1;</code>
+   */
+  java.lang.String getName();
+  /**
+   * <code>string name = 1;</code>
+   */
+  com.google.protobuf.ByteString
+      getNameBytes();
+
+  /**
+   * <pre>
+   * The version is optional. If set, needs to be &gt;= 0.
+   * </pre>
+   *
+   * <code>int64 version = 2;</code>
+   */
+  long getVersion();
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SignatureDef.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SignatureDef.java
new file mode 100644
index 00000000000..fe588917c86
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SignatureDef.java
@@ -0,0 +1,1339 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/meta_graph.proto
+
+package org.tensorflow.proto.framework;
+
+/**
+ * <pre>
+ * SignatureDef defines the signature of a computation supported by a TensorFlow
+ * graph.
+ * For example, a model with two loss computations, sharing a single input,
+ * might have the following signature_def map.
+ * Note that across the two SignatureDefs "loss_A" and "loss_B", the input key,
+ * output key, and method_name are identical, and will be used by system(s) that
+ * implement or rely upon this particular loss method. The output tensor names
+ * differ, demonstrating how different outputs can exist for the same method.
+ * signature_def {
+ *   key: "loss_A"
+ *   value {
+ *     inputs {
+ *       key: "input"
+ *       value {
+ *         name: "input:0"
+ *         dtype: DT_STRING
+ *         tensor_shape: ...
+ *       }
+ *     }
+ *     outputs {
+ *       key: "loss_output"
+ *       value {
+ *         name: "loss_output_A:0"
+ *         dtype: DT_FLOAT
+ *         tensor_shape: ...
+ *       }
+ *     }
+ *   }
+ *   ...
+ *   method_name: "some/package/compute_loss"
+ * }
+ * signature_def {
+ *   key: "loss_B"
+ *   value {
+ *     inputs {
+ *       key: "input"
+ *       value {
+ *         name: "input:0"
+ *         dtype: DT_STRING
+ *         tensor_shape: ...
+ *       }
+ *     }
+ *     outputs {
+ *       key: "loss_output"
+ *       value {
+ *         name: "loss_output_B:0"
+ *         dtype: DT_FLOAT
+ *         tensor_shape: ...
+ *       }
+ *     }
+ *   }
+ *   ...
+ *   method_name: "some/package/compute_loss"
+ * }
+ * </pre>
+ *
+ * Protobuf type {@code tensorflow.SignatureDef}
+ */
+public  final class SignatureDef extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.SignatureDef)
+    SignatureDefOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use SignatureDef.newBuilder() to construct.
+  private SignatureDef(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private SignatureDef() {
+    methodName_ = "";
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new SignatureDef();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private SignatureDef(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    int mutable_bitField0_ = 0;
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 10: {
+            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
+              inputs_ = com.google.protobuf.MapField.newMapField(
+                  InputsDefaultEntryHolder.defaultEntry);
+              mutable_bitField0_ |= 0x00000001;
+            }
+            com.google.protobuf.MapEntry<java.lang.String, org.tensorflow.proto.framework.TensorInfo>
+            inputs__ = input.readMessage(
+                InputsDefaultEntryHolder.defaultEntry.getParserForType(), extensionRegistry);
+            inputs_.getMutableMap().put(
+                inputs__.getKey(), inputs__.getValue());
+            break;
+          }
+          case 18: {
+            if (!((mutable_bitField0_ & 0x00000002) != 0)) {
+              outputs_ = com.google.protobuf.MapField.newMapField(
+                  OutputsDefaultEntryHolder.defaultEntry);
+              mutable_bitField0_ |= 0x00000002;
+            }
+            com.google.protobuf.MapEntry<java.lang.String, org.tensorflow.proto.framework.TensorInfo>
+            outputs__ = input.readMessage(
+                OutputsDefaultEntryHolder.defaultEntry.getParserForType(), extensionRegistry);
+            outputs_.getMutableMap().put(
+                outputs__.getKey(), outputs__.getValue());
+            break;
+          }
+          case 26: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            methodName_ = s;
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.framework.MetaGraphProtos.internal_static_tensorflow_SignatureDef_descriptor;
+  }
+
+  @SuppressWarnings({"rawtypes"})
+  @java.lang.Override
+  protected com.google.protobuf.MapField internalGetMapField(
+      int number) {
+    switch (number) {
+      case 1:
+        return internalGetInputs();
+      case 2:
+        return internalGetOutputs();
+      default:
+        throw new RuntimeException(
+            "Invalid map field number: " + number);
+    }
+  }
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.framework.MetaGraphProtos.internal_static_tensorflow_SignatureDef_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.framework.SignatureDef.class, org.tensorflow.proto.framework.SignatureDef.Builder.class);
+  }
+
+  public static final int INPUTS_FIELD_NUMBER = 1;
+  private static final class InputsDefaultEntryHolder {
+    static final com.google.protobuf.MapEntry<
+        java.lang.String, org.tensorflow.proto.framework.TensorInfo> defaultEntry =
+            com.google.protobuf.MapEntry
+            .<java.lang.String, org.tensorflow.proto.framework.TensorInfo>newDefaultInstance(
+                org.tensorflow.proto.framework.MetaGraphProtos.internal_static_tensorflow_SignatureDef_InputsEntry_descriptor, 
+                com.google.protobuf.WireFormat.FieldType.STRING,
+                "",
+                com.google.protobuf.WireFormat.FieldType.MESSAGE,
+                org.tensorflow.proto.framework.TensorInfo.getDefaultInstance());
+  }
+  private com.google.protobuf.MapField<
+      java.lang.String, org.tensorflow.proto.framework.TensorInfo> inputs_;
+  private com.google.protobuf.MapField<java.lang.String, org.tensorflow.proto.framework.TensorInfo>
+  internalGetInputs() {
+    if (inputs_ == null) {
+      return com.google.protobuf.MapField.emptyMapField(
+          InputsDefaultEntryHolder.defaultEntry);
+    }
+    return inputs_;
+  }
+
+  public int getInputsCount() {
+    return internalGetInputs().getMap().size();
+  }
+  /**
+   * <pre>
+   * Named input parameters.
+   * </pre>
+   *
+   * <code>map&lt;string, .tensorflow.TensorInfo&gt; inputs = 1;</code>
+   */
+
+  public boolean containsInputs(
+      java.lang.String key) {
+    if (key == null) { throw new java.lang.NullPointerException(); }
+    return internalGetInputs().getMap().containsKey(key);
+  }
+  /**
+   * Use {@link #getInputsMap()} instead.
+   */
+  @java.lang.Deprecated
+  public java.util.Map<java.lang.String, org.tensorflow.proto.framework.TensorInfo> getInputs() {
+    return getInputsMap();
+  }
+  /**
+   * <pre>
+   * Named input parameters.
+   * </pre>
+   *
+   * <code>map&lt;string, .tensorflow.TensorInfo&gt; inputs = 1;</code>
+   */
+
+  public java.util.Map<java.lang.String, org.tensorflow.proto.framework.TensorInfo> getInputsMap() {
+    return internalGetInputs().getMap();
+  }
+  /**
+   * <pre>
+   * Named input parameters.
+   * </pre>
+   *
+   * <code>map&lt;string, .tensorflow.TensorInfo&gt; inputs = 1;</code>
+   */
+
+  public org.tensorflow.proto.framework.TensorInfo getInputsOrDefault(
+      java.lang.String key,
+      org.tensorflow.proto.framework.TensorInfo defaultValue) {
+    if (key == null) { throw new java.lang.NullPointerException(); }
+    java.util.Map<java.lang.String, org.tensorflow.proto.framework.TensorInfo> map =
+        internalGetInputs().getMap();
+    return map.containsKey(key) ? map.get(key) : defaultValue;
+  }
+  /**
+   * <pre>
+   * Named input parameters.
+   * </pre>
+   *
+   * <code>map&lt;string, .tensorflow.TensorInfo&gt; inputs = 1;</code>
+   */
+
+  public org.tensorflow.proto.framework.TensorInfo getInputsOrThrow(
+      java.lang.String key) {
+    if (key == null) { throw new java.lang.NullPointerException(); }
+    java.util.Map<java.lang.String, org.tensorflow.proto.framework.TensorInfo> map =
+        internalGetInputs().getMap();
+    if (!map.containsKey(key)) {
+      throw new java.lang.IllegalArgumentException();
+    }
+    return map.get(key);
+  }
+
+  public static final int OUTPUTS_FIELD_NUMBER = 2;
+  private static final class OutputsDefaultEntryHolder {
+    static final com.google.protobuf.MapEntry<
+        java.lang.String, org.tensorflow.proto.framework.TensorInfo> defaultEntry =
+            com.google.protobuf.MapEntry
+            .<java.lang.String, org.tensorflow.proto.framework.TensorInfo>newDefaultInstance(
+                org.tensorflow.proto.framework.MetaGraphProtos.internal_static_tensorflow_SignatureDef_OutputsEntry_descriptor, 
+                com.google.protobuf.WireFormat.FieldType.STRING,
+                "",
+                com.google.protobuf.WireFormat.FieldType.MESSAGE,
+                org.tensorflow.proto.framework.TensorInfo.getDefaultInstance());
+  }
+  private com.google.protobuf.MapField<
+      java.lang.String, org.tensorflow.proto.framework.TensorInfo> outputs_;
+  private com.google.protobuf.MapField<java.lang.String, org.tensorflow.proto.framework.TensorInfo>
+  internalGetOutputs() {
+    if (outputs_ == null) {
+      return com.google.protobuf.MapField.emptyMapField(
+          OutputsDefaultEntryHolder.defaultEntry);
+    }
+    return outputs_;
+  }
+
+  public int getOutputsCount() {
+    return internalGetOutputs().getMap().size();
+  }
+  /**
+   * <pre>
+   * Named output parameters.
+   * </pre>
+   *
+   * <code>map&lt;string, .tensorflow.TensorInfo&gt; outputs = 2;</code>
+   */
+
+  public boolean containsOutputs(
+      java.lang.String key) {
+    if (key == null) { throw new java.lang.NullPointerException(); }
+    return internalGetOutputs().getMap().containsKey(key);
+  }
+  /**
+   * Use {@link #getOutputsMap()} instead.
+   */
+  @java.lang.Deprecated
+  public java.util.Map<java.lang.String, org.tensorflow.proto.framework.TensorInfo> getOutputs() {
+    return getOutputsMap();
+  }
+  /**
+   * <pre>
+   * Named output parameters.
+   * </pre>
+   *
+   * <code>map&lt;string, .tensorflow.TensorInfo&gt; outputs = 2;</code>
+   */
+
+  public java.util.Map<java.lang.String, org.tensorflow.proto.framework.TensorInfo> getOutputsMap() {
+    return internalGetOutputs().getMap();
+  }
+  /**
+   * <pre>
+   * Named output parameters.
+   * </pre>
+   *
+   * <code>map&lt;string, .tensorflow.TensorInfo&gt; outputs = 2;</code>
+   */
+
+  public org.tensorflow.proto.framework.TensorInfo getOutputsOrDefault(
+      java.lang.String key,
+      org.tensorflow.proto.framework.TensorInfo defaultValue) {
+    if (key == null) { throw new java.lang.NullPointerException(); }
+    java.util.Map<java.lang.String, org.tensorflow.proto.framework.TensorInfo> map =
+        internalGetOutputs().getMap();
+    return map.containsKey(key) ? map.get(key) : defaultValue;
+  }
+  /**
+   * <pre>
+   * Named output parameters.
+   * </pre>
+   *
+   * <code>map&lt;string, .tensorflow.TensorInfo&gt; outputs = 2;</code>
+   */
+
+  public org.tensorflow.proto.framework.TensorInfo getOutputsOrThrow(
+      java.lang.String key) {
+    if (key == null) { throw new java.lang.NullPointerException(); }
+    java.util.Map<java.lang.String, org.tensorflow.proto.framework.TensorInfo> map =
+        internalGetOutputs().getMap();
+    if (!map.containsKey(key)) {
+      throw new java.lang.IllegalArgumentException();
+    }
+    return map.get(key);
+  }
+
+  public static final int METHOD_NAME_FIELD_NUMBER = 3;
+  private volatile java.lang.Object methodName_;
+  /**
+   * <pre>
+   * Extensible method_name information enabling third-party users to mark a
+   * SignatureDef as supporting a particular method. This enables producers and
+   * consumers of SignatureDefs, e.g. a model definition library and a serving
+   * library to have a clear hand-off regarding the semantics of a computation.
+   * Note that multiple SignatureDefs in a single MetaGraphDef may have the same
+   * method_name. This is commonly used to support multi-headed computation,
+   * where a single graph computation may return multiple results.
+   * </pre>
+   *
+   * <code>string method_name = 3;</code>
+   */
+  public java.lang.String getMethodName() {
+    java.lang.Object ref = methodName_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      methodName_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * Extensible method_name information enabling third-party users to mark a
+   * SignatureDef as supporting a particular method. This enables producers and
+   * consumers of SignatureDefs, e.g. a model definition library and a serving
+   * library to have a clear hand-off regarding the semantics of a computation.
+   * Note that multiple SignatureDefs in a single MetaGraphDef may have the same
+   * method_name. This is commonly used to support multi-headed computation,
+   * where a single graph computation may return multiple results.
+   * </pre>
+   *
+   * <code>string method_name = 3;</code>
+   */
+  public com.google.protobuf.ByteString
+      getMethodNameBytes() {
+    java.lang.Object ref = methodName_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      methodName_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    com.google.protobuf.GeneratedMessageV3
+      .serializeStringMapTo(
+        output,
+        internalGetInputs(),
+        InputsDefaultEntryHolder.defaultEntry,
+        1);
+    com.google.protobuf.GeneratedMessageV3
+      .serializeStringMapTo(
+        output,
+        internalGetOutputs(),
+        OutputsDefaultEntryHolder.defaultEntry,
+        2);
+    if (!getMethodNameBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 3, methodName_);
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    for (java.util.Map.Entry<java.lang.String, org.tensorflow.proto.framework.TensorInfo> entry
+         : internalGetInputs().getMap().entrySet()) {
+      com.google.protobuf.MapEntry<java.lang.String, org.tensorflow.proto.framework.TensorInfo>
+      inputs__ = InputsDefaultEntryHolder.defaultEntry.newBuilderForType()
+          .setKey(entry.getKey())
+          .setValue(entry.getValue())
+          .build();
+      size += com.google.protobuf.CodedOutputStream
+          .computeMessageSize(1, inputs__);
+    }
+    for (java.util.Map.Entry<java.lang.String, org.tensorflow.proto.framework.TensorInfo> entry
+         : internalGetOutputs().getMap().entrySet()) {
+      com.google.protobuf.MapEntry<java.lang.String, org.tensorflow.proto.framework.TensorInfo>
+      outputs__ = OutputsDefaultEntryHolder.defaultEntry.newBuilderForType()
+          .setKey(entry.getKey())
+          .setValue(entry.getValue())
+          .build();
+      size += com.google.protobuf.CodedOutputStream
+          .computeMessageSize(2, outputs__);
+    }
+    if (!getMethodNameBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(3, methodName_);
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.framework.SignatureDef)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.framework.SignatureDef other = (org.tensorflow.proto.framework.SignatureDef) obj;
+
+    if (!internalGetInputs().equals(
+        other.internalGetInputs())) return false;
+    if (!internalGetOutputs().equals(
+        other.internalGetOutputs())) return false;
+    if (!getMethodName()
+        .equals(other.getMethodName())) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    if (!internalGetInputs().getMap().isEmpty()) {
+      hash = (37 * hash) + INPUTS_FIELD_NUMBER;
+      hash = (53 * hash) + internalGetInputs().hashCode();
+    }
+    if (!internalGetOutputs().getMap().isEmpty()) {
+      hash = (37 * hash) + OUTPUTS_FIELD_NUMBER;
+      hash = (53 * hash) + internalGetOutputs().hashCode();
+    }
+    hash = (37 * hash) + METHOD_NAME_FIELD_NUMBER;
+    hash = (53 * hash) + getMethodName().hashCode();
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.framework.SignatureDef parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.SignatureDef parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.SignatureDef parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.SignatureDef parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.SignatureDef parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.SignatureDef parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.SignatureDef parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.SignatureDef parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.SignatureDef parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.SignatureDef parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.SignatureDef parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.SignatureDef parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.framework.SignatureDef prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * <pre>
+   * SignatureDef defines the signature of a computation supported by a TensorFlow
+   * graph.
+   * For example, a model with two loss computations, sharing a single input,
+   * might have the following signature_def map.
+   * Note that across the two SignatureDefs "loss_A" and "loss_B", the input key,
+   * output key, and method_name are identical, and will be used by system(s) that
+   * implement or rely upon this particular loss method. The output tensor names
+   * differ, demonstrating how different outputs can exist for the same method.
+   * signature_def {
+   *   key: "loss_A"
+   *   value {
+   *     inputs {
+   *       key: "input"
+   *       value {
+   *         name: "input:0"
+   *         dtype: DT_STRING
+   *         tensor_shape: ...
+   *       }
+   *     }
+   *     outputs {
+   *       key: "loss_output"
+   *       value {
+   *         name: "loss_output_A:0"
+   *         dtype: DT_FLOAT
+   *         tensor_shape: ...
+   *       }
+   *     }
+   *   }
+   *   ...
+   *   method_name: "some/package/compute_loss"
+   * }
+   * signature_def {
+   *   key: "loss_B"
+   *   value {
+   *     inputs {
+   *       key: "input"
+   *       value {
+   *         name: "input:0"
+   *         dtype: DT_STRING
+   *         tensor_shape: ...
+   *       }
+   *     }
+   *     outputs {
+   *       key: "loss_output"
+   *       value {
+   *         name: "loss_output_B:0"
+   *         dtype: DT_FLOAT
+   *         tensor_shape: ...
+   *       }
+   *     }
+   *   }
+   *   ...
+   *   method_name: "some/package/compute_loss"
+   * }
+   * </pre>
+   *
+   * Protobuf type {@code tensorflow.SignatureDef}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.SignatureDef)
+      org.tensorflow.proto.framework.SignatureDefOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.MetaGraphProtos.internal_static_tensorflow_SignatureDef_descriptor;
+    }
+
+    @SuppressWarnings({"rawtypes"})
+    protected com.google.protobuf.MapField internalGetMapField(
+        int number) {
+      switch (number) {
+        case 1:
+          return internalGetInputs();
+        case 2:
+          return internalGetOutputs();
+        default:
+          throw new RuntimeException(
+              "Invalid map field number: " + number);
+      }
+    }
+    @SuppressWarnings({"rawtypes"})
+    protected com.google.protobuf.MapField internalGetMutableMapField(
+        int number) {
+      switch (number) {
+        case 1:
+          return internalGetMutableInputs();
+        case 2:
+          return internalGetMutableOutputs();
+        default:
+          throw new RuntimeException(
+              "Invalid map field number: " + number);
+      }
+    }
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.MetaGraphProtos.internal_static_tensorflow_SignatureDef_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.SignatureDef.class, org.tensorflow.proto.framework.SignatureDef.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.framework.SignatureDef.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      internalGetMutableInputs().clear();
+      internalGetMutableOutputs().clear();
+      methodName_ = "";
+
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.framework.MetaGraphProtos.internal_static_tensorflow_SignatureDef_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.SignatureDef getDefaultInstanceForType() {
+      return org.tensorflow.proto.framework.SignatureDef.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.SignatureDef build() {
+      org.tensorflow.proto.framework.SignatureDef result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.SignatureDef buildPartial() {
+      org.tensorflow.proto.framework.SignatureDef result = new org.tensorflow.proto.framework.SignatureDef(this);
+      int from_bitField0_ = bitField0_;
+      result.inputs_ = internalGetInputs();
+      result.inputs_.makeImmutable();
+      result.outputs_ = internalGetOutputs();
+      result.outputs_.makeImmutable();
+      result.methodName_ = methodName_;
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.framework.SignatureDef) {
+        return mergeFrom((org.tensorflow.proto.framework.SignatureDef)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.framework.SignatureDef other) {
+      if (other == org.tensorflow.proto.framework.SignatureDef.getDefaultInstance()) return this;
+      internalGetMutableInputs().mergeFrom(
+          other.internalGetInputs());
+      internalGetMutableOutputs().mergeFrom(
+          other.internalGetOutputs());
+      if (!other.getMethodName().isEmpty()) {
+        methodName_ = other.methodName_;
+        onChanged();
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.framework.SignatureDef parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.framework.SignatureDef) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+    private int bitField0_;
+
+    private com.google.protobuf.MapField<
+        java.lang.String, org.tensorflow.proto.framework.TensorInfo> inputs_;
+    private com.google.protobuf.MapField<java.lang.String, org.tensorflow.proto.framework.TensorInfo>
+    internalGetInputs() {
+      if (inputs_ == null) {
+        return com.google.protobuf.MapField.emptyMapField(
+            InputsDefaultEntryHolder.defaultEntry);
+      }
+      return inputs_;
+    }
+    private com.google.protobuf.MapField<java.lang.String, org.tensorflow.proto.framework.TensorInfo>
+    internalGetMutableInputs() {
+      onChanged();;
+      if (inputs_ == null) {
+        inputs_ = com.google.protobuf.MapField.newMapField(
+            InputsDefaultEntryHolder.defaultEntry);
+      }
+      if (!inputs_.isMutable()) {
+        inputs_ = inputs_.copy();
+      }
+      return inputs_;
+    }
+
+    public int getInputsCount() {
+      return internalGetInputs().getMap().size();
+    }
+    /**
+     * <pre>
+     * Named input parameters.
+     * </pre>
+     *
+     * <code>map&lt;string, .tensorflow.TensorInfo&gt; inputs = 1;</code>
+     */
+
+    public boolean containsInputs(
+        java.lang.String key) {
+      if (key == null) { throw new java.lang.NullPointerException(); }
+      return internalGetInputs().getMap().containsKey(key);
+    }
+    /**
+     * Use {@link #getInputsMap()} instead.
+     */
+    @java.lang.Deprecated
+    public java.util.Map<java.lang.String, org.tensorflow.proto.framework.TensorInfo> getInputs() {
+      return getInputsMap();
+    }
+    /**
+     * <pre>
+     * Named input parameters.
+     * </pre>
+     *
+     * <code>map&lt;string, .tensorflow.TensorInfo&gt; inputs = 1;</code>
+     */
+
+    public java.util.Map<java.lang.String, org.tensorflow.proto.framework.TensorInfo> getInputsMap() {
+      return internalGetInputs().getMap();
+    }
+    /**
+     * <pre>
+     * Named input parameters.
+     * </pre>
+     *
+     * <code>map&lt;string, .tensorflow.TensorInfo&gt; inputs = 1;</code>
+     */
+
+    public org.tensorflow.proto.framework.TensorInfo getInputsOrDefault(
+        java.lang.String key,
+        org.tensorflow.proto.framework.TensorInfo defaultValue) {
+      if (key == null) { throw new java.lang.NullPointerException(); }
+      java.util.Map<java.lang.String, org.tensorflow.proto.framework.TensorInfo> map =
+          internalGetInputs().getMap();
+      return map.containsKey(key) ? map.get(key) : defaultValue;
+    }
+    /**
+     * <pre>
+     * Named input parameters.
+     * </pre>
+     *
+     * <code>map&lt;string, .tensorflow.TensorInfo&gt; inputs = 1;</code>
+     */
+
+    public org.tensorflow.proto.framework.TensorInfo getInputsOrThrow(
+        java.lang.String key) {
+      if (key == null) { throw new java.lang.NullPointerException(); }
+      java.util.Map<java.lang.String, org.tensorflow.proto.framework.TensorInfo> map =
+          internalGetInputs().getMap();
+      if (!map.containsKey(key)) {
+        throw new java.lang.IllegalArgumentException();
+      }
+      return map.get(key);
+    }
+
+    public Builder clearInputs() {
+      internalGetMutableInputs().getMutableMap()
+          .clear();
+      return this;
+    }
+    /**
+     * <pre>
+     * Named input parameters.
+     * </pre>
+     *
+     * <code>map&lt;string, .tensorflow.TensorInfo&gt; inputs = 1;</code>
+     */
+
+    public Builder removeInputs(
+        java.lang.String key) {
+      if (key == null) { throw new java.lang.NullPointerException(); }
+      internalGetMutableInputs().getMutableMap()
+          .remove(key);
+      return this;
+    }
+    /**
+     * Use alternate mutation accessors instead.
+     */
+    @java.lang.Deprecated
+    public java.util.Map<java.lang.String, org.tensorflow.proto.framework.TensorInfo>
+    getMutableInputs() {
+      return internalGetMutableInputs().getMutableMap();
+    }
+    /**
+     * <pre>
+     * Named input parameters.
+     * </pre>
+     *
+     * <code>map&lt;string, .tensorflow.TensorInfo&gt; inputs = 1;</code>
+     */
+    public Builder putInputs(
+        java.lang.String key,
+        org.tensorflow.proto.framework.TensorInfo value) {
+      if (key == null) { throw new java.lang.NullPointerException(); }
+      if (value == null) { throw new java.lang.NullPointerException(); }
+      internalGetMutableInputs().getMutableMap()
+          .put(key, value);
+      return this;
+    }
+    /**
+     * <pre>
+     * Named input parameters.
+     * </pre>
+     *
+     * <code>map&lt;string, .tensorflow.TensorInfo&gt; inputs = 1;</code>
+     */
+
+    public Builder putAllInputs(
+        java.util.Map<java.lang.String, org.tensorflow.proto.framework.TensorInfo> values) {
+      internalGetMutableInputs().getMutableMap()
+          .putAll(values);
+      return this;
+    }
+
+    private com.google.protobuf.MapField<
+        java.lang.String, org.tensorflow.proto.framework.TensorInfo> outputs_;
+    private com.google.protobuf.MapField<java.lang.String, org.tensorflow.proto.framework.TensorInfo>
+    internalGetOutputs() {
+      if (outputs_ == null) {
+        return com.google.protobuf.MapField.emptyMapField(
+            OutputsDefaultEntryHolder.defaultEntry);
+      }
+      return outputs_;
+    }
+    private com.google.protobuf.MapField<java.lang.String, org.tensorflow.proto.framework.TensorInfo>
+    internalGetMutableOutputs() {
+      onChanged();;
+      if (outputs_ == null) {
+        outputs_ = com.google.protobuf.MapField.newMapField(
+            OutputsDefaultEntryHolder.defaultEntry);
+      }
+      if (!outputs_.isMutable()) {
+        outputs_ = outputs_.copy();
+      }
+      return outputs_;
+    }
+
+    public int getOutputsCount() {
+      return internalGetOutputs().getMap().size();
+    }
+    /**
+     * <pre>
+     * Named output parameters.
+     * </pre>
+     *
+     * <code>map&lt;string, .tensorflow.TensorInfo&gt; outputs = 2;</code>
+     */
+
+    public boolean containsOutputs(
+        java.lang.String key) {
+      if (key == null) { throw new java.lang.NullPointerException(); }
+      return internalGetOutputs().getMap().containsKey(key);
+    }
+    /**
+     * Use {@link #getOutputsMap()} instead.
+     */
+    @java.lang.Deprecated
+    public java.util.Map<java.lang.String, org.tensorflow.proto.framework.TensorInfo> getOutputs() {
+      return getOutputsMap();
+    }
+    /**
+     * <pre>
+     * Named output parameters.
+     * </pre>
+     *
+     * <code>map&lt;string, .tensorflow.TensorInfo&gt; outputs = 2;</code>
+     */
+
+    public java.util.Map<java.lang.String, org.tensorflow.proto.framework.TensorInfo> getOutputsMap() {
+      return internalGetOutputs().getMap();
+    }
+    /**
+     * <pre>
+     * Named output parameters.
+     * </pre>
+     *
+     * <code>map&lt;string, .tensorflow.TensorInfo&gt; outputs = 2;</code>
+     */
+
+    public org.tensorflow.proto.framework.TensorInfo getOutputsOrDefault(
+        java.lang.String key,
+        org.tensorflow.proto.framework.TensorInfo defaultValue) {
+      if (key == null) { throw new java.lang.NullPointerException(); }
+      java.util.Map<java.lang.String, org.tensorflow.proto.framework.TensorInfo> map =
+          internalGetOutputs().getMap();
+      return map.containsKey(key) ? map.get(key) : defaultValue;
+    }
+    /**
+     * <pre>
+     * Named output parameters.
+     * </pre>
+     *
+     * <code>map&lt;string, .tensorflow.TensorInfo&gt; outputs = 2;</code>
+     */
+
+    public org.tensorflow.proto.framework.TensorInfo getOutputsOrThrow(
+        java.lang.String key) {
+      if (key == null) { throw new java.lang.NullPointerException(); }
+      java.util.Map<java.lang.String, org.tensorflow.proto.framework.TensorInfo> map =
+          internalGetOutputs().getMap();
+      if (!map.containsKey(key)) {
+        throw new java.lang.IllegalArgumentException();
+      }
+      return map.get(key);
+    }
+
+    public Builder clearOutputs() {
+      internalGetMutableOutputs().getMutableMap()
+          .clear();
+      return this;
+    }
+    /**
+     * <pre>
+     * Named output parameters.
+     * </pre>
+     *
+     * <code>map&lt;string, .tensorflow.TensorInfo&gt; outputs = 2;</code>
+     */
+
+    public Builder removeOutputs(
+        java.lang.String key) {
+      if (key == null) { throw new java.lang.NullPointerException(); }
+      internalGetMutableOutputs().getMutableMap()
+          .remove(key);
+      return this;
+    }
+    /**
+     * Use alternate mutation accessors instead.
+     */
+    @java.lang.Deprecated
+    public java.util.Map<java.lang.String, org.tensorflow.proto.framework.TensorInfo>
+    getMutableOutputs() {
+      return internalGetMutableOutputs().getMutableMap();
+    }
+    /**
+     * <pre>
+     * Named output parameters.
+     * </pre>
+     *
+     * <code>map&lt;string, .tensorflow.TensorInfo&gt; outputs = 2;</code>
+     */
+    public Builder putOutputs(
+        java.lang.String key,
+        org.tensorflow.proto.framework.TensorInfo value) {
+      if (key == null) { throw new java.lang.NullPointerException(); }
+      if (value == null) { throw new java.lang.NullPointerException(); }
+      internalGetMutableOutputs().getMutableMap()
+          .put(key, value);
+      return this;
+    }
+    /**
+     * <pre>
+     * Named output parameters.
+     * </pre>
+     *
+     * <code>map&lt;string, .tensorflow.TensorInfo&gt; outputs = 2;</code>
+     */
+
+    public Builder putAllOutputs(
+        java.util.Map<java.lang.String, org.tensorflow.proto.framework.TensorInfo> values) {
+      internalGetMutableOutputs().getMutableMap()
+          .putAll(values);
+      return this;
+    }
+
+    private java.lang.Object methodName_ = "";
+    /**
+     * <pre>
+     * Extensible method_name information enabling third-party users to mark a
+     * SignatureDef as supporting a particular method. This enables producers and
+     * consumers of SignatureDefs, e.g. a model definition library and a serving
+     * library to have a clear hand-off regarding the semantics of a computation.
+     * Note that multiple SignatureDefs in a single MetaGraphDef may have the same
+     * method_name. This is commonly used to support multi-headed computation,
+     * where a single graph computation may return multiple results.
+     * </pre>
+     *
+     * <code>string method_name = 3;</code>
+     */
+    public java.lang.String getMethodName() {
+      java.lang.Object ref = methodName_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        methodName_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Extensible method_name information enabling third-party users to mark a
+     * SignatureDef as supporting a particular method. This enables producers and
+     * consumers of SignatureDefs, e.g. a model definition library and a serving
+     * library to have a clear hand-off regarding the semantics of a computation.
+     * Note that multiple SignatureDefs in a single MetaGraphDef may have the same
+     * method_name. This is commonly used to support multi-headed computation,
+     * where a single graph computation may return multiple results.
+     * </pre>
+     *
+     * <code>string method_name = 3;</code>
+     */
+    public com.google.protobuf.ByteString
+        getMethodNameBytes() {
+      java.lang.Object ref = methodName_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        methodName_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Extensible method_name information enabling third-party users to mark a
+     * SignatureDef as supporting a particular method. This enables producers and
+     * consumers of SignatureDefs, e.g. a model definition library and a serving
+     * library to have a clear hand-off regarding the semantics of a computation.
+     * Note that multiple SignatureDefs in a single MetaGraphDef may have the same
+     * method_name. This is commonly used to support multi-headed computation,
+     * where a single graph computation may return multiple results.
+     * </pre>
+     *
+     * <code>string method_name = 3;</code>
+     */
+    public Builder setMethodName(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      methodName_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Extensible method_name information enabling third-party users to mark a
+     * SignatureDef as supporting a particular method. This enables producers and
+     * consumers of SignatureDefs, e.g. a model definition library and a serving
+     * library to have a clear hand-off regarding the semantics of a computation.
+     * Note that multiple SignatureDefs in a single MetaGraphDef may have the same
+     * method_name. This is commonly used to support multi-headed computation,
+     * where a single graph computation may return multiple results.
+     * </pre>
+     *
+     * <code>string method_name = 3;</code>
+     */
+    public Builder clearMethodName() {
+      
+      methodName_ = getDefaultInstance().getMethodName();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Extensible method_name information enabling third-party users to mark a
+     * SignatureDef as supporting a particular method. This enables producers and
+     * consumers of SignatureDefs, e.g. a model definition library and a serving
+     * library to have a clear hand-off regarding the semantics of a computation.
+     * Note that multiple SignatureDefs in a single MetaGraphDef may have the same
+     * method_name. This is commonly used to support multi-headed computation,
+     * where a single graph computation may return multiple results.
+     * </pre>
+     *
+     * <code>string method_name = 3;</code>
+     */
+    public Builder setMethodNameBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      methodName_ = value;
+      onChanged();
+      return this;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.SignatureDef)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.SignatureDef)
+  private static final org.tensorflow.proto.framework.SignatureDef DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.SignatureDef();
+  }
+
+  public static org.tensorflow.proto.framework.SignatureDef getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<SignatureDef>
+      PARSER = new com.google.protobuf.AbstractParser<SignatureDef>() {
+    @java.lang.Override
+    public SignatureDef parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new SignatureDef(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<SignatureDef> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<SignatureDef> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.framework.SignatureDef getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SignatureDefOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SignatureDefOrBuilder.java
new file mode 100644
index 00000000000..e9234adf5f7
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SignatureDefOrBuilder.java
@@ -0,0 +1,147 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/meta_graph.proto
+
+package org.tensorflow.proto.framework;
+
+public interface SignatureDefOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.SignatureDef)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <pre>
+   * Named input parameters.
+   * </pre>
+   *
+   * <code>map&lt;string, .tensorflow.TensorInfo&gt; inputs = 1;</code>
+   */
+  int getInputsCount();
+  /**
+   * <pre>
+   * Named input parameters.
+   * </pre>
+   *
+   * <code>map&lt;string, .tensorflow.TensorInfo&gt; inputs = 1;</code>
+   */
+  boolean containsInputs(
+      java.lang.String key);
+  /**
+   * Use {@link #getInputsMap()} instead.
+   */
+  @java.lang.Deprecated
+  java.util.Map<java.lang.String, org.tensorflow.proto.framework.TensorInfo>
+  getInputs();
+  /**
+   * <pre>
+   * Named input parameters.
+   * </pre>
+   *
+   * <code>map&lt;string, .tensorflow.TensorInfo&gt; inputs = 1;</code>
+   */
+  java.util.Map<java.lang.String, org.tensorflow.proto.framework.TensorInfo>
+  getInputsMap();
+  /**
+   * <pre>
+   * Named input parameters.
+   * </pre>
+   *
+   * <code>map&lt;string, .tensorflow.TensorInfo&gt; inputs = 1;</code>
+   */
+
+  org.tensorflow.proto.framework.TensorInfo getInputsOrDefault(
+      java.lang.String key,
+      org.tensorflow.proto.framework.TensorInfo defaultValue);
+  /**
+   * <pre>
+   * Named input parameters.
+   * </pre>
+   *
+   * <code>map&lt;string, .tensorflow.TensorInfo&gt; inputs = 1;</code>
+   */
+
+  org.tensorflow.proto.framework.TensorInfo getInputsOrThrow(
+      java.lang.String key);
+
+  /**
+   * <pre>
+   * Named output parameters.
+   * </pre>
+   *
+   * <code>map&lt;string, .tensorflow.TensorInfo&gt; outputs = 2;</code>
+   */
+  int getOutputsCount();
+  /**
+   * <pre>
+   * Named output parameters.
+   * </pre>
+   *
+   * <code>map&lt;string, .tensorflow.TensorInfo&gt; outputs = 2;</code>
+   */
+  boolean containsOutputs(
+      java.lang.String key);
+  /**
+   * Use {@link #getOutputsMap()} instead.
+   */
+  @java.lang.Deprecated
+  java.util.Map<java.lang.String, org.tensorflow.proto.framework.TensorInfo>
+  getOutputs();
+  /**
+   * <pre>
+   * Named output parameters.
+   * </pre>
+   *
+   * <code>map&lt;string, .tensorflow.TensorInfo&gt; outputs = 2;</code>
+   */
+  java.util.Map<java.lang.String, org.tensorflow.proto.framework.TensorInfo>
+  getOutputsMap();
+  /**
+   * <pre>
+   * Named output parameters.
+   * </pre>
+   *
+   * <code>map&lt;string, .tensorflow.TensorInfo&gt; outputs = 2;</code>
+   */
+
+  org.tensorflow.proto.framework.TensorInfo getOutputsOrDefault(
+      java.lang.String key,
+      org.tensorflow.proto.framework.TensorInfo defaultValue);
+  /**
+   * <pre>
+   * Named output parameters.
+   * </pre>
+   *
+   * <code>map&lt;string, .tensorflow.TensorInfo&gt; outputs = 2;</code>
+   */
+
+  org.tensorflow.proto.framework.TensorInfo getOutputsOrThrow(
+      java.lang.String key);
+
+  /**
+   * <pre>
+   * Extensible method_name information enabling third-party users to mark a
+   * SignatureDef as supporting a particular method. This enables producers and
+   * consumers of SignatureDefs, e.g. a model definition library and a serving
+   * library to have a clear hand-off regarding the semantics of a computation.
+   * Note that multiple SignatureDefs in a single MetaGraphDef may have the same
+   * method_name. This is commonly used to support multi-headed computation,
+   * where a single graph computation may return multiple results.
+   * </pre>
+   *
+   * <code>string method_name = 3;</code>
+   */
+  java.lang.String getMethodName();
+  /**
+   * <pre>
+   * Extensible method_name information enabling third-party users to mark a
+   * SignatureDef as supporting a particular method. This enables producers and
+   * consumers of SignatureDefs, e.g. a model definition library and a serving
+   * library to have a clear hand-off regarding the semantics of a computation.
+   * Note that multiple SignatureDefs in a single MetaGraphDef may have the same
+   * method_name. This is commonly used to support multi-headed computation,
+   * where a single graph computation may return multiple results.
+   * </pre>
+   *
+   * <code>string method_name = 3;</code>
+   */
+  com.google.protobuf.ByteString
+      getMethodNameBytes();
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/StepStats.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/StepStats.java
new file mode 100644
index 00000000000..01a979698bd
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/StepStats.java
@@ -0,0 +1,765 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/framework/step_stats.proto
+
+package org.tensorflow.proto.framework;
+
+/**
+ * Protobuf type {@code tensorflow.StepStats}
+ */
+public  final class StepStats extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.StepStats)
+    StepStatsOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use StepStats.newBuilder() to construct.
+  private StepStats(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private StepStats() {
+    devStats_ = java.util.Collections.emptyList();
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new StepStats();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private StepStats(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    int mutable_bitField0_ = 0;
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 10: {
+            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
+              devStats_ = new java.util.ArrayList<org.tensorflow.proto.framework.DeviceStepStats>();
+              mutable_bitField0_ |= 0x00000001;
+            }
+            devStats_.add(
+                input.readMessage(org.tensorflow.proto.framework.DeviceStepStats.parser(), extensionRegistry));
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      if (((mutable_bitField0_ & 0x00000001) != 0)) {
+        devStats_ = java.util.Collections.unmodifiableList(devStats_);
+      }
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.framework.StepStatsProtos.internal_static_tensorflow_StepStats_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.framework.StepStatsProtos.internal_static_tensorflow_StepStats_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.framework.StepStats.class, org.tensorflow.proto.framework.StepStats.Builder.class);
+  }
+
+  public static final int DEV_STATS_FIELD_NUMBER = 1;
+  private java.util.List<org.tensorflow.proto.framework.DeviceStepStats> devStats_;
+  /**
+   * <code>repeated .tensorflow.DeviceStepStats dev_stats = 1;</code>
+   */
+  public java.util.List<org.tensorflow.proto.framework.DeviceStepStats> getDevStatsList() {
+    return devStats_;
+  }
+  /**
+   * <code>repeated .tensorflow.DeviceStepStats dev_stats = 1;</code>
+   */
+  public java.util.List<? extends org.tensorflow.proto.framework.DeviceStepStatsOrBuilder> 
+      getDevStatsOrBuilderList() {
+    return devStats_;
+  }
+  /**
+   * <code>repeated .tensorflow.DeviceStepStats dev_stats = 1;</code>
+   */
+  public int getDevStatsCount() {
+    return devStats_.size();
+  }
+  /**
+   * <code>repeated .tensorflow.DeviceStepStats dev_stats = 1;</code>
+   */
+  public org.tensorflow.proto.framework.DeviceStepStats getDevStats(int index) {
+    return devStats_.get(index);
+  }
+  /**
+   * <code>repeated .tensorflow.DeviceStepStats dev_stats = 1;</code>
+   */
+  public org.tensorflow.proto.framework.DeviceStepStatsOrBuilder getDevStatsOrBuilder(
+      int index) {
+    return devStats_.get(index);
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    for (int i = 0; i < devStats_.size(); i++) {
+      output.writeMessage(1, devStats_.get(i));
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    for (int i = 0; i < devStats_.size(); i++) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(1, devStats_.get(i));
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.framework.StepStats)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.framework.StepStats other = (org.tensorflow.proto.framework.StepStats) obj;
+
+    if (!getDevStatsList()
+        .equals(other.getDevStatsList())) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    if (getDevStatsCount() > 0) {
+      hash = (37 * hash) + DEV_STATS_FIELD_NUMBER;
+      hash = (53 * hash) + getDevStatsList().hashCode();
+    }
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.framework.StepStats parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.StepStats parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.StepStats parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.StepStats parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.StepStats parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.StepStats parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.StepStats parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.StepStats parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.StepStats parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.StepStats parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.StepStats parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.StepStats parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.framework.StepStats prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * Protobuf type {@code tensorflow.StepStats}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.StepStats)
+      org.tensorflow.proto.framework.StepStatsOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.StepStatsProtos.internal_static_tensorflow_StepStats_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.StepStatsProtos.internal_static_tensorflow_StepStats_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.StepStats.class, org.tensorflow.proto.framework.StepStats.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.framework.StepStats.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+        getDevStatsFieldBuilder();
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      if (devStatsBuilder_ == null) {
+        devStats_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000001);
+      } else {
+        devStatsBuilder_.clear();
+      }
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.framework.StepStatsProtos.internal_static_tensorflow_StepStats_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.StepStats getDefaultInstanceForType() {
+      return org.tensorflow.proto.framework.StepStats.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.StepStats build() {
+      org.tensorflow.proto.framework.StepStats result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.StepStats buildPartial() {
+      org.tensorflow.proto.framework.StepStats result = new org.tensorflow.proto.framework.StepStats(this);
+      int from_bitField0_ = bitField0_;
+      if (devStatsBuilder_ == null) {
+        if (((bitField0_ & 0x00000001) != 0)) {
+          devStats_ = java.util.Collections.unmodifiableList(devStats_);
+          bitField0_ = (bitField0_ & ~0x00000001);
+        }
+        result.devStats_ = devStats_;
+      } else {
+        result.devStats_ = devStatsBuilder_.build();
+      }
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.framework.StepStats) {
+        return mergeFrom((org.tensorflow.proto.framework.StepStats)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.framework.StepStats other) {
+      if (other == org.tensorflow.proto.framework.StepStats.getDefaultInstance()) return this;
+      if (devStatsBuilder_ == null) {
+        if (!other.devStats_.isEmpty()) {
+          if (devStats_.isEmpty()) {
+            devStats_ = other.devStats_;
+            bitField0_ = (bitField0_ & ~0x00000001);
+          } else {
+            ensureDevStatsIsMutable();
+            devStats_.addAll(other.devStats_);
+          }
+          onChanged();
+        }
+      } else {
+        if (!other.devStats_.isEmpty()) {
+          if (devStatsBuilder_.isEmpty()) {
+            devStatsBuilder_.dispose();
+            devStatsBuilder_ = null;
+            devStats_ = other.devStats_;
+            bitField0_ = (bitField0_ & ~0x00000001);
+            devStatsBuilder_ = 
+              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
+                 getDevStatsFieldBuilder() : null;
+          } else {
+            devStatsBuilder_.addAllMessages(other.devStats_);
+          }
+        }
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.framework.StepStats parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.framework.StepStats) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+    private int bitField0_;
+
+    private java.util.List<org.tensorflow.proto.framework.DeviceStepStats> devStats_ =
+      java.util.Collections.emptyList();
+    private void ensureDevStatsIsMutable() {
+      if (!((bitField0_ & 0x00000001) != 0)) {
+        devStats_ = new java.util.ArrayList<org.tensorflow.proto.framework.DeviceStepStats>(devStats_);
+        bitField0_ |= 0x00000001;
+       }
+    }
+
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.framework.DeviceStepStats, org.tensorflow.proto.framework.DeviceStepStats.Builder, org.tensorflow.proto.framework.DeviceStepStatsOrBuilder> devStatsBuilder_;
+
+    /**
+     * <code>repeated .tensorflow.DeviceStepStats dev_stats = 1;</code>
+     */
+    public java.util.List<org.tensorflow.proto.framework.DeviceStepStats> getDevStatsList() {
+      if (devStatsBuilder_ == null) {
+        return java.util.Collections.unmodifiableList(devStats_);
+      } else {
+        return devStatsBuilder_.getMessageList();
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.DeviceStepStats dev_stats = 1;</code>
+     */
+    public int getDevStatsCount() {
+      if (devStatsBuilder_ == null) {
+        return devStats_.size();
+      } else {
+        return devStatsBuilder_.getCount();
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.DeviceStepStats dev_stats = 1;</code>
+     */
+    public org.tensorflow.proto.framework.DeviceStepStats getDevStats(int index) {
+      if (devStatsBuilder_ == null) {
+        return devStats_.get(index);
+      } else {
+        return devStatsBuilder_.getMessage(index);
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.DeviceStepStats dev_stats = 1;</code>
+     */
+    public Builder setDevStats(
+        int index, org.tensorflow.proto.framework.DeviceStepStats value) {
+      if (devStatsBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureDevStatsIsMutable();
+        devStats_.set(index, value);
+        onChanged();
+      } else {
+        devStatsBuilder_.setMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.DeviceStepStats dev_stats = 1;</code>
+     */
+    public Builder setDevStats(
+        int index, org.tensorflow.proto.framework.DeviceStepStats.Builder builderForValue) {
+      if (devStatsBuilder_ == null) {
+        ensureDevStatsIsMutable();
+        devStats_.set(index, builderForValue.build());
+        onChanged();
+      } else {
+        devStatsBuilder_.setMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.DeviceStepStats dev_stats = 1;</code>
+     */
+    public Builder addDevStats(org.tensorflow.proto.framework.DeviceStepStats value) {
+      if (devStatsBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureDevStatsIsMutable();
+        devStats_.add(value);
+        onChanged();
+      } else {
+        devStatsBuilder_.addMessage(value);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.DeviceStepStats dev_stats = 1;</code>
+     */
+    public Builder addDevStats(
+        int index, org.tensorflow.proto.framework.DeviceStepStats value) {
+      if (devStatsBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureDevStatsIsMutable();
+        devStats_.add(index, value);
+        onChanged();
+      } else {
+        devStatsBuilder_.addMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.DeviceStepStats dev_stats = 1;</code>
+     */
+    public Builder addDevStats(
+        org.tensorflow.proto.framework.DeviceStepStats.Builder builderForValue) {
+      if (devStatsBuilder_ == null) {
+        ensureDevStatsIsMutable();
+        devStats_.add(builderForValue.build());
+        onChanged();
+      } else {
+        devStatsBuilder_.addMessage(builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.DeviceStepStats dev_stats = 1;</code>
+     */
+    public Builder addDevStats(
+        int index, org.tensorflow.proto.framework.DeviceStepStats.Builder builderForValue) {
+      if (devStatsBuilder_ == null) {
+        ensureDevStatsIsMutable();
+        devStats_.add(index, builderForValue.build());
+        onChanged();
+      } else {
+        devStatsBuilder_.addMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.DeviceStepStats dev_stats = 1;</code>
+     */
+    public Builder addAllDevStats(
+        java.lang.Iterable<? extends org.tensorflow.proto.framework.DeviceStepStats> values) {
+      if (devStatsBuilder_ == null) {
+        ensureDevStatsIsMutable();
+        com.google.protobuf.AbstractMessageLite.Builder.addAll(
+            values, devStats_);
+        onChanged();
+      } else {
+        devStatsBuilder_.addAllMessages(values);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.DeviceStepStats dev_stats = 1;</code>
+     */
+    public Builder clearDevStats() {
+      if (devStatsBuilder_ == null) {
+        devStats_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000001);
+        onChanged();
+      } else {
+        devStatsBuilder_.clear();
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.DeviceStepStats dev_stats = 1;</code>
+     */
+    public Builder removeDevStats(int index) {
+      if (devStatsBuilder_ == null) {
+        ensureDevStatsIsMutable();
+        devStats_.remove(index);
+        onChanged();
+      } else {
+        devStatsBuilder_.remove(index);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.DeviceStepStats dev_stats = 1;</code>
+     */
+    public org.tensorflow.proto.framework.DeviceStepStats.Builder getDevStatsBuilder(
+        int index) {
+      return getDevStatsFieldBuilder().getBuilder(index);
+    }
+    /**
+     * <code>repeated .tensorflow.DeviceStepStats dev_stats = 1;</code>
+     */
+    public org.tensorflow.proto.framework.DeviceStepStatsOrBuilder getDevStatsOrBuilder(
+        int index) {
+      if (devStatsBuilder_ == null) {
+        return devStats_.get(index);  } else {
+        return devStatsBuilder_.getMessageOrBuilder(index);
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.DeviceStepStats dev_stats = 1;</code>
+     */
+    public java.util.List<? extends org.tensorflow.proto.framework.DeviceStepStatsOrBuilder> 
+         getDevStatsOrBuilderList() {
+      if (devStatsBuilder_ != null) {
+        return devStatsBuilder_.getMessageOrBuilderList();
+      } else {
+        return java.util.Collections.unmodifiableList(devStats_);
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.DeviceStepStats dev_stats = 1;</code>
+     */
+    public org.tensorflow.proto.framework.DeviceStepStats.Builder addDevStatsBuilder() {
+      return getDevStatsFieldBuilder().addBuilder(
+          org.tensorflow.proto.framework.DeviceStepStats.getDefaultInstance());
+    }
+    /**
+     * <code>repeated .tensorflow.DeviceStepStats dev_stats = 1;</code>
+     */
+    public org.tensorflow.proto.framework.DeviceStepStats.Builder addDevStatsBuilder(
+        int index) {
+      return getDevStatsFieldBuilder().addBuilder(
+          index, org.tensorflow.proto.framework.DeviceStepStats.getDefaultInstance());
+    }
+    /**
+     * <code>repeated .tensorflow.DeviceStepStats dev_stats = 1;</code>
+     */
+    public java.util.List<org.tensorflow.proto.framework.DeviceStepStats.Builder> 
+         getDevStatsBuilderList() {
+      return getDevStatsFieldBuilder().getBuilderList();
+    }
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.framework.DeviceStepStats, org.tensorflow.proto.framework.DeviceStepStats.Builder, org.tensorflow.proto.framework.DeviceStepStatsOrBuilder> 
+        getDevStatsFieldBuilder() {
+      if (devStatsBuilder_ == null) {
+        devStatsBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
+            org.tensorflow.proto.framework.DeviceStepStats, org.tensorflow.proto.framework.DeviceStepStats.Builder, org.tensorflow.proto.framework.DeviceStepStatsOrBuilder>(
+                devStats_,
+                ((bitField0_ & 0x00000001) != 0),
+                getParentForChildren(),
+                isClean());
+        devStats_ = null;
+      }
+      return devStatsBuilder_;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.StepStats)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.StepStats)
+  private static final org.tensorflow.proto.framework.StepStats DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.StepStats();
+  }
+
+  public static org.tensorflow.proto.framework.StepStats getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<StepStats>
+      PARSER = new com.google.protobuf.AbstractParser<StepStats>() {
+    @java.lang.Override
+    public StepStats parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new StepStats(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<StepStats> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<StepStats> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.framework.StepStats getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/StepStatsOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/StepStatsOrBuilder.java
new file mode 100644
index 00000000000..bb7cdf41046
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/StepStatsOrBuilder.java
@@ -0,0 +1,33 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/framework/step_stats.proto
+
+package org.tensorflow.proto.framework;
+
+public interface StepStatsOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.StepStats)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <code>repeated .tensorflow.DeviceStepStats dev_stats = 1;</code>
+   */
+  java.util.List<org.tensorflow.proto.framework.DeviceStepStats> 
+      getDevStatsList();
+  /**
+   * <code>repeated .tensorflow.DeviceStepStats dev_stats = 1;</code>
+   */
+  org.tensorflow.proto.framework.DeviceStepStats getDevStats(int index);
+  /**
+   * <code>repeated .tensorflow.DeviceStepStats dev_stats = 1;</code>
+   */
+  int getDevStatsCount();
+  /**
+   * <code>repeated .tensorflow.DeviceStepStats dev_stats = 1;</code>
+   */
+  java.util.List<? extends org.tensorflow.proto.framework.DeviceStepStatsOrBuilder> 
+      getDevStatsOrBuilderList();
+  /**
+   * <code>repeated .tensorflow.DeviceStepStats dev_stats = 1;</code>
+   */
+  org.tensorflow.proto.framework.DeviceStepStatsOrBuilder getDevStatsOrBuilder(
+      int index);
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/StepStatsProtos.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/StepStatsProtos.java
new file mode 100644
index 00000000000..15ef4a6b554
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/StepStatsProtos.java
@@ -0,0 +1,169 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/framework/step_stats.proto
+
+package org.tensorflow.proto.framework;
+
+public final class StepStatsProtos {
+  private StepStatsProtos() {}
+  public static void registerAllExtensions(
+      com.google.protobuf.ExtensionRegistryLite registry) {
+  }
+
+  public static void registerAllExtensions(
+      com.google.protobuf.ExtensionRegistry registry) {
+    registerAllExtensions(
+        (com.google.protobuf.ExtensionRegistryLite) registry);
+  }
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_AllocationRecord_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_AllocationRecord_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_AllocatorMemoryUsed_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_AllocatorMemoryUsed_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_NodeOutput_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_NodeOutput_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_MemoryStats_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_MemoryStats_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_NodeExecStats_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_NodeExecStats_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_DeviceStepStats_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_DeviceStepStats_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_DeviceStepStats_ThreadNamesEntry_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_DeviceStepStats_ThreadNamesEntry_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_StepStats_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_StepStats_fieldAccessorTable;
+
+  public static com.google.protobuf.Descriptors.FileDescriptor
+      getDescriptor() {
+    return descriptor;
+  }
+  private static  com.google.protobuf.Descriptors.FileDescriptor
+      descriptor;
+  static {
+    java.lang.String[] descriptorData = {
+      "\n*tensorflow/core/framework/step_stats.p" +
+      "roto\022\ntensorflow\0326tensorflow/core/framew" +
+      "ork/allocation_description.proto\0322tensor" +
+      "flow/core/framework/tensor_description.p" +
+      "roto\"=\n\020AllocationRecord\022\024\n\014alloc_micros" +
+      "\030\001 \001(\003\022\023\n\013alloc_bytes\030\002 \001(\003\"\304\001\n\023Allocato" +
+      "rMemoryUsed\022\026\n\016allocator_name\030\001 \001(\t\022\023\n\013t" +
+      "otal_bytes\030\002 \001(\003\022\022\n\npeak_bytes\030\003 \001(\003\022\022\n\n" +
+      "live_bytes\030\004 \001(\003\0228\n\022allocation_records\030\006" +
+      " \003(\0132\034.tensorflow.AllocationRecord\022\036\n\026al" +
+      "locator_bytes_in_use\030\005 \001(\003\"U\n\nNodeOutput" +
+      "\022\014\n\004slot\030\001 \001(\005\0229\n\022tensor_description\030\003 \001" +
+      "(\0132\035.tensorflow.TensorDescription\"\354\001\n\013Me" +
+      "moryStats\022\030\n\020temp_memory_size\030\001 \001(\003\022\036\n\026p" +
+      "ersistent_memory_size\030\003 \001(\003\022#\n\033persisten" +
+      "t_tensor_alloc_ids\030\005 \003(\003\022#\n\027device_temp_" +
+      "memory_size\030\002 \001(\003B\002\030\001\022)\n\035device_persiste" +
+      "nt_memory_size\030\004 \001(\003B\002\030\001\022.\n\"device_persi" +
+      "stent_tensor_alloc_ids\030\006 \003(\003B\002\030\001\"\236\004\n\rNod" +
+      "eExecStats\022\021\n\tnode_name\030\001 \001(\t\022\030\n\020all_sta" +
+      "rt_micros\030\002 \001(\003\022\033\n\023op_start_rel_micros\030\003" +
+      " \001(\003\022\031\n\021op_end_rel_micros\030\004 \001(\003\022\032\n\022all_e" +
+      "nd_rel_micros\030\005 \001(\003\022/\n\006memory\030\006 \003(\0132\037.te" +
+      "nsorflow.AllocatorMemoryUsed\022&\n\006output\030\007" +
+      " \003(\0132\026.tensorflow.NodeOutput\022\026\n\016timeline" +
+      "_label\030\010 \001(\t\022\030\n\020scheduled_micros\030\t \001(\003\022\021" +
+      "\n\tthread_id\030\n \001(\r\022<\n\021referenced_tensor\030\013" +
+      " \003(\0132!.tensorflow.AllocationDescription\022" +
+      "-\n\014memory_stats\030\014 \001(\0132\027.tensorflow.Memor" +
+      "yStats\022\027\n\017all_start_nanos\030\r \001(\003\022\032\n\022op_st" +
+      "art_rel_nanos\030\016 \001(\003\022\030\n\020op_end_rel_nanos\030" +
+      "\017 \001(\003\022\031\n\021all_end_rel_nanos\030\020 \001(\003\022\027\n\017sche" +
+      "duled_nanos\030\021 \001(\003\"\310\001\n\017DeviceStepStats\022\016\n" +
+      "\006device\030\001 \001(\t\022-\n\nnode_stats\030\002 \003(\0132\031.tens" +
+      "orflow.NodeExecStats\022B\n\014thread_names\030\003 \003" +
+      "(\0132,.tensorflow.DeviceStepStats.ThreadNa" +
+      "mesEntry\0322\n\020ThreadNamesEntry\022\013\n\003key\030\001 \001(" +
+      "\r\022\r\n\005value\030\002 \001(\t:\0028\001\";\n\tStepStats\022.\n\tdev" +
+      "_stats\030\001 \003(\0132\033.tensorflow.DeviceStepStat" +
+      "sB\211\001\n\036org.tensorflow.proto.frameworkB\017St" +
+      "epStatsProtosP\001ZQgithub.com/tensorflow/t" +
+      "ensorflow/tensorflow/go/core/framework/s" +
+      "tep_stats_go_proto\370\001\001b\006proto3"
+    };
+    descriptor = com.google.protobuf.Descriptors.FileDescriptor
+      .internalBuildGeneratedFileFrom(descriptorData,
+        new com.google.protobuf.Descriptors.FileDescriptor[] {
+          org.tensorflow.proto.framework.AllocationDescriptionProtos.getDescriptor(),
+          org.tensorflow.proto.framework.TensorDescriptionProtos.getDescriptor(),
+        });
+    internal_static_tensorflow_AllocationRecord_descriptor =
+      getDescriptor().getMessageTypes().get(0);
+    internal_static_tensorflow_AllocationRecord_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_AllocationRecord_descriptor,
+        new java.lang.String[] { "AllocMicros", "AllocBytes", });
+    internal_static_tensorflow_AllocatorMemoryUsed_descriptor =
+      getDescriptor().getMessageTypes().get(1);
+    internal_static_tensorflow_AllocatorMemoryUsed_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_AllocatorMemoryUsed_descriptor,
+        new java.lang.String[] { "AllocatorName", "TotalBytes", "PeakBytes", "LiveBytes", "AllocationRecords", "AllocatorBytesInUse", });
+    internal_static_tensorflow_NodeOutput_descriptor =
+      getDescriptor().getMessageTypes().get(2);
+    internal_static_tensorflow_NodeOutput_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_NodeOutput_descriptor,
+        new java.lang.String[] { "Slot", "TensorDescription", });
+    internal_static_tensorflow_MemoryStats_descriptor =
+      getDescriptor().getMessageTypes().get(3);
+    internal_static_tensorflow_MemoryStats_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_MemoryStats_descriptor,
+        new java.lang.String[] { "TempMemorySize", "PersistentMemorySize", "PersistentTensorAllocIds", "DeviceTempMemorySize", "DevicePersistentMemorySize", "DevicePersistentTensorAllocIds", });
+    internal_static_tensorflow_NodeExecStats_descriptor =
+      getDescriptor().getMessageTypes().get(4);
+    internal_static_tensorflow_NodeExecStats_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_NodeExecStats_descriptor,
+        new java.lang.String[] { "NodeName", "AllStartMicros", "OpStartRelMicros", "OpEndRelMicros", "AllEndRelMicros", "Memory", "Output", "TimelineLabel", "ScheduledMicros", "ThreadId", "ReferencedTensor", "MemoryStats", "AllStartNanos", "OpStartRelNanos", "OpEndRelNanos", "AllEndRelNanos", "ScheduledNanos", });
+    internal_static_tensorflow_DeviceStepStats_descriptor =
+      getDescriptor().getMessageTypes().get(5);
+    internal_static_tensorflow_DeviceStepStats_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_DeviceStepStats_descriptor,
+        new java.lang.String[] { "Device", "NodeStats", "ThreadNames", });
+    internal_static_tensorflow_DeviceStepStats_ThreadNamesEntry_descriptor =
+      internal_static_tensorflow_DeviceStepStats_descriptor.getNestedTypes().get(0);
+    internal_static_tensorflow_DeviceStepStats_ThreadNamesEntry_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_DeviceStepStats_ThreadNamesEntry_descriptor,
+        new java.lang.String[] { "Key", "Value", });
+    internal_static_tensorflow_StepStats_descriptor =
+      getDescriptor().getMessageTypes().get(6);
+    internal_static_tensorflow_StepStats_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_StepStats_descriptor,
+        new java.lang.String[] { "DevStats", });
+    org.tensorflow.proto.framework.AllocationDescriptionProtos.getDescriptor();
+    org.tensorflow.proto.framework.TensorDescriptionProtos.getDescriptor();
+  }
+
+  // @@protoc_insertion_point(outer_class_scope)
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/StructProtos.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/StructProtos.java
new file mode 100644
index 00000000000..6b25fbe0fa5
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/StructProtos.java
@@ -0,0 +1,215 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/struct.proto
+
+package org.tensorflow.proto.framework;
+
+public final class StructProtos {
+  private StructProtos() {}
+  public static void registerAllExtensions(
+      com.google.protobuf.ExtensionRegistryLite registry) {
+  }
+
+  public static void registerAllExtensions(
+      com.google.protobuf.ExtensionRegistry registry) {
+    registerAllExtensions(
+        (com.google.protobuf.ExtensionRegistryLite) registry);
+  }
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_StructuredValue_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_StructuredValue_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_NoneValue_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_NoneValue_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_ListValue_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_ListValue_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_TupleValue_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_TupleValue_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_DictValue_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_DictValue_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_DictValue_FieldsEntry_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_DictValue_FieldsEntry_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_PairValue_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_PairValue_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_NamedTupleValue_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_NamedTupleValue_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_TensorSpecProto_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_TensorSpecProto_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_BoundedTensorSpecProto_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_BoundedTensorSpecProto_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_TypeSpecProto_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_TypeSpecProto_fieldAccessorTable;
+
+  public static com.google.protobuf.Descriptors.FileDescriptor
+      getDescriptor() {
+    return descriptor;
+  }
+  private static  com.google.protobuf.Descriptors.FileDescriptor
+      descriptor;
+  static {
+    java.lang.String[] descriptorData = {
+      "\n%tensorflow/core/protobuf/struct.proto\022" +
+      "\ntensorflow\032&tensorflow/core/framework/t" +
+      "ensor.proto\032,tensorflow/core/framework/t" +
+      "ensor_shape.proto\032%tensorflow/core/frame" +
+      "work/types.proto\"\220\005\n\017StructuredValue\022+\n\n" +
+      "none_value\030\001 \001(\0132\025.tensorflow.NoneValueH" +
+      "\000\022\027\n\rfloat64_value\030\013 \001(\001H\000\022\025\n\013int64_valu" +
+      "e\030\014 \001(\022H\000\022\026\n\014string_value\030\r \001(\tH\000\022\024\n\nboo" +
+      "l_value\030\016 \001(\010H\000\022:\n\022tensor_shape_value\030\037 " +
+      "\001(\0132\034.tensorflow.TensorShapeProtoH\000\0222\n\022t" +
+      "ensor_dtype_value\030  \001(\0162\024.tensorflow.Dat" +
+      "aTypeH\000\0228\n\021tensor_spec_value\030! \001(\0132\033.ten" +
+      "sorflow.TensorSpecProtoH\000\0224\n\017type_spec_v" +
+      "alue\030\" \001(\0132\031.tensorflow.TypeSpecProtoH\000\022" +
+      "G\n\031bounded_tensor_spec_value\030# \001(\0132\".ten" +
+      "sorflow.BoundedTensorSpecProtoH\000\022+\n\nlist" +
+      "_value\0303 \001(\0132\025.tensorflow.ListValueH\000\022-\n" +
+      "\013tuple_value\0304 \001(\0132\026.tensorflow.TupleVal" +
+      "ueH\000\022+\n\ndict_value\0305 \001(\0132\025.tensorflow.Di" +
+      "ctValueH\000\0228\n\021named_tuple_value\0306 \001(\0132\033.t" +
+      "ensorflow.NamedTupleValueH\000B\006\n\004kind\"\013\n\tN" +
+      "oneValue\"8\n\tListValue\022+\n\006values\030\001 \003(\0132\033." +
+      "tensorflow.StructuredValue\"9\n\nTupleValue" +
+      "\022+\n\006values\030\001 \003(\0132\033.tensorflow.Structured" +
+      "Value\"\212\001\n\tDictValue\0221\n\006fields\030\001 \003(\0132!.te" +
+      "nsorflow.DictValue.FieldsEntry\032J\n\013Fields" +
+      "Entry\022\013\n\003key\030\001 \001(\t\022*\n\005value\030\002 \001(\0132\033.tens" +
+      "orflow.StructuredValue:\0028\001\"D\n\tPairValue\022" +
+      "\013\n\003key\030\001 \001(\t\022*\n\005value\030\002 \001(\0132\033.tensorflow" +
+      ".StructuredValue\"F\n\017NamedTupleValue\022\014\n\004n" +
+      "ame\030\001 \001(\t\022%\n\006values\030\002 \003(\0132\025.tensorflow.P" +
+      "airValue\"q\n\017TensorSpecProto\022\014\n\004name\030\001 \001(" +
+      "\t\022+\n\005shape\030\002 \001(\0132\034.tensorflow.TensorShap" +
+      "eProto\022#\n\005dtype\030\003 \001(\0162\024.tensorflow.DataT" +
+      "ype\"\314\001\n\026BoundedTensorSpecProto\022\014\n\004name\030\001" +
+      " \001(\t\022+\n\005shape\030\002 \001(\0132\034.tensorflow.TensorS" +
+      "hapeProto\022#\n\005dtype\030\003 \001(\0162\024.tensorflow.Da" +
+      "taType\022(\n\007minimum\030\004 \001(\0132\027.tensorflow.Ten" +
+      "sorProto\022(\n\007maximum\030\005 \001(\0132\027.tensorflow.T" +
+      "ensorProto\"\242\003\n\rTypeSpecProto\022@\n\017type_spe" +
+      "c_class\030\001 \001(\0162\'.tensorflow.TypeSpecProto" +
+      ".TypeSpecClass\022/\n\ntype_state\030\002 \001(\0132\033.ten" +
+      "sorflow.StructuredValue\022\034\n\024type_spec_cla" +
+      "ss_name\030\003 \001(\t\"\377\001\n\rTypeSpecClass\022\013\n\007UNKNO" +
+      "WN\020\000\022\026\n\022SPARSE_TENSOR_SPEC\020\001\022\027\n\023INDEXED_" +
+      "SLICES_SPEC\020\002\022\026\n\022RAGGED_TENSOR_SPEC\020\003\022\025\n" +
+      "\021TENSOR_ARRAY_SPEC\020\004\022\025\n\021DATA_DATASET_SPE" +
+      "C\020\005\022\026\n\022DATA_ITERATOR_SPEC\020\006\022\021\n\rOPTIONAL_" +
+      "SPEC\020\007\022\024\n\020PER_REPLICA_SPEC\020\010\022\021\n\rVARIABLE" +
+      "_SPEC\020\t\022\026\n\022ROW_PARTITION_SPEC\020\nBz\n\036org.t" +
+      "ensorflow.proto.frameworkB\014StructProtosP" +
+      "\001ZHgithub.com/tensorflow/tensorflow/tens" +
+      "orflow/go/core/core_protos_go_protob\006pro" +
+      "to3"
+    };
+    descriptor = com.google.protobuf.Descriptors.FileDescriptor
+      .internalBuildGeneratedFileFrom(descriptorData,
+        new com.google.protobuf.Descriptors.FileDescriptor[] {
+          org.tensorflow.proto.framework.TensorProtos.getDescriptor(),
+          org.tensorflow.proto.framework.TensorShapeProtos.getDescriptor(),
+          org.tensorflow.proto.framework.TypesProtos.getDescriptor(),
+        });
+    internal_static_tensorflow_StructuredValue_descriptor =
+      getDescriptor().getMessageTypes().get(0);
+    internal_static_tensorflow_StructuredValue_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_StructuredValue_descriptor,
+        new java.lang.String[] { "NoneValue", "Float64Value", "Int64Value", "StringValue", "BoolValue", "TensorShapeValue", "TensorDtypeValue", "TensorSpecValue", "TypeSpecValue", "BoundedTensorSpecValue", "ListValue", "TupleValue", "DictValue", "NamedTupleValue", "Kind", });
+    internal_static_tensorflow_NoneValue_descriptor =
+      getDescriptor().getMessageTypes().get(1);
+    internal_static_tensorflow_NoneValue_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_NoneValue_descriptor,
+        new java.lang.String[] { });
+    internal_static_tensorflow_ListValue_descriptor =
+      getDescriptor().getMessageTypes().get(2);
+    internal_static_tensorflow_ListValue_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_ListValue_descriptor,
+        new java.lang.String[] { "Values", });
+    internal_static_tensorflow_TupleValue_descriptor =
+      getDescriptor().getMessageTypes().get(3);
+    internal_static_tensorflow_TupleValue_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_TupleValue_descriptor,
+        new java.lang.String[] { "Values", });
+    internal_static_tensorflow_DictValue_descriptor =
+      getDescriptor().getMessageTypes().get(4);
+    internal_static_tensorflow_DictValue_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_DictValue_descriptor,
+        new java.lang.String[] { "Fields", });
+    internal_static_tensorflow_DictValue_FieldsEntry_descriptor =
+      internal_static_tensorflow_DictValue_descriptor.getNestedTypes().get(0);
+    internal_static_tensorflow_DictValue_FieldsEntry_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_DictValue_FieldsEntry_descriptor,
+        new java.lang.String[] { "Key", "Value", });
+    internal_static_tensorflow_PairValue_descriptor =
+      getDescriptor().getMessageTypes().get(5);
+    internal_static_tensorflow_PairValue_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_PairValue_descriptor,
+        new java.lang.String[] { "Key", "Value", });
+    internal_static_tensorflow_NamedTupleValue_descriptor =
+      getDescriptor().getMessageTypes().get(6);
+    internal_static_tensorflow_NamedTupleValue_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_NamedTupleValue_descriptor,
+        new java.lang.String[] { "Name", "Values", });
+    internal_static_tensorflow_TensorSpecProto_descriptor =
+      getDescriptor().getMessageTypes().get(7);
+    internal_static_tensorflow_TensorSpecProto_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_TensorSpecProto_descriptor,
+        new java.lang.String[] { "Name", "Shape", "Dtype", });
+    internal_static_tensorflow_BoundedTensorSpecProto_descriptor =
+      getDescriptor().getMessageTypes().get(8);
+    internal_static_tensorflow_BoundedTensorSpecProto_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_BoundedTensorSpecProto_descriptor,
+        new java.lang.String[] { "Name", "Shape", "Dtype", "Minimum", "Maximum", });
+    internal_static_tensorflow_TypeSpecProto_descriptor =
+      getDescriptor().getMessageTypes().get(9);
+    internal_static_tensorflow_TypeSpecProto_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_TypeSpecProto_descriptor,
+        new java.lang.String[] { "TypeSpecClass", "TypeState", "TypeSpecClassName", });
+    org.tensorflow.proto.framework.TensorProtos.getDescriptor();
+    org.tensorflow.proto.framework.TensorShapeProtos.getDescriptor();
+    org.tensorflow.proto.framework.TypesProtos.getDescriptor();
+  }
+
+  // @@protoc_insertion_point(outer_class_scope)
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/StructuredValue.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/StructuredValue.java
new file mode 100644
index 00000000000..327e919c2ab
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/StructuredValue.java
@@ -0,0 +1,3423 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/struct.proto
+
+package org.tensorflow.proto.framework;
+
+/**
+ * <pre>
+ * `StructuredValue` represents a dynamically typed value representing various
+ * data structures that are inspired by Python data structures typically used in
+ * TensorFlow functions as inputs and outputs.
+ * For example when saving a Layer there may be a `training` argument. If the
+ * user passes a boolean True/False, that switches between two concrete
+ * TensorFlow functions. In order to switch between them in the same way after
+ * loading the SavedModel, we need to represent "True" and "False".
+ * A more advanced example might be a function which takes a list of
+ * dictionaries mapping from strings to Tensors. In order to map from
+ * user-specified arguments `[{"a": tf.constant(1.)}, {"q": tf.constant(3.)}]`
+ * after load to the right saved TensorFlow function, we need to represent the
+ * nested structure and the strings, recording that we have a trace for anything
+ * matching `[{"a": tf.TensorSpec(None, tf.float32)}, {"q": tf.TensorSpec([],
+ * tf.float64)}]` as an example.
+ * Likewise functions may return nested structures of Tensors, for example
+ * returning a dictionary mapping from strings to Tensors. In order for the
+ * loaded function to return the same structure we need to serialize it.
+ * This is an ergonomic aid for working with loaded SavedModels, not a promise
+ * to serialize all possible function signatures. For example we do not expect
+ * to pickle generic Python objects, and ideally we'd stay language-agnostic.
+ * </pre>
+ *
+ * Protobuf type {@code tensorflow.StructuredValue}
+ */
+public  final class StructuredValue extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.StructuredValue)
+    StructuredValueOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use StructuredValue.newBuilder() to construct.
+  private StructuredValue(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private StructuredValue() {
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new StructuredValue();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private StructuredValue(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 10: {
+            org.tensorflow.proto.framework.NoneValue.Builder subBuilder = null;
+            if (kindCase_ == 1) {
+              subBuilder = ((org.tensorflow.proto.framework.NoneValue) kind_).toBuilder();
+            }
+            kind_ =
+                input.readMessage(org.tensorflow.proto.framework.NoneValue.parser(), extensionRegistry);
+            if (subBuilder != null) {
+              subBuilder.mergeFrom((org.tensorflow.proto.framework.NoneValue) kind_);
+              kind_ = subBuilder.buildPartial();
+            }
+            kindCase_ = 1;
+            break;
+          }
+          case 89: {
+            kindCase_ = 11;
+            kind_ = input.readDouble();
+            break;
+          }
+          case 96: {
+            kindCase_ = 12;
+            kind_ = input.readSInt64();
+            break;
+          }
+          case 106: {
+            java.lang.String s = input.readStringRequireUtf8();
+            kindCase_ = 13;
+            kind_ = s;
+            break;
+          }
+          case 112: {
+            kindCase_ = 14;
+            kind_ = input.readBool();
+            break;
+          }
+          case 250: {
+            org.tensorflow.proto.framework.TensorShapeProto.Builder subBuilder = null;
+            if (kindCase_ == 31) {
+              subBuilder = ((org.tensorflow.proto.framework.TensorShapeProto) kind_).toBuilder();
+            }
+            kind_ =
+                input.readMessage(org.tensorflow.proto.framework.TensorShapeProto.parser(), extensionRegistry);
+            if (subBuilder != null) {
+              subBuilder.mergeFrom((org.tensorflow.proto.framework.TensorShapeProto) kind_);
+              kind_ = subBuilder.buildPartial();
+            }
+            kindCase_ = 31;
+            break;
+          }
+          case 256: {
+            int rawValue = input.readEnum();
+            kindCase_ = 32;
+            kind_ = rawValue;
+            break;
+          }
+          case 266: {
+            org.tensorflow.proto.framework.TensorSpecProto.Builder subBuilder = null;
+            if (kindCase_ == 33) {
+              subBuilder = ((org.tensorflow.proto.framework.TensorSpecProto) kind_).toBuilder();
+            }
+            kind_ =
+                input.readMessage(org.tensorflow.proto.framework.TensorSpecProto.parser(), extensionRegistry);
+            if (subBuilder != null) {
+              subBuilder.mergeFrom((org.tensorflow.proto.framework.TensorSpecProto) kind_);
+              kind_ = subBuilder.buildPartial();
+            }
+            kindCase_ = 33;
+            break;
+          }
+          case 274: {
+            org.tensorflow.proto.framework.TypeSpecProto.Builder subBuilder = null;
+            if (kindCase_ == 34) {
+              subBuilder = ((org.tensorflow.proto.framework.TypeSpecProto) kind_).toBuilder();
+            }
+            kind_ =
+                input.readMessage(org.tensorflow.proto.framework.TypeSpecProto.parser(), extensionRegistry);
+            if (subBuilder != null) {
+              subBuilder.mergeFrom((org.tensorflow.proto.framework.TypeSpecProto) kind_);
+              kind_ = subBuilder.buildPartial();
+            }
+            kindCase_ = 34;
+            break;
+          }
+          case 282: {
+            org.tensorflow.proto.framework.BoundedTensorSpecProto.Builder subBuilder = null;
+            if (kindCase_ == 35) {
+              subBuilder = ((org.tensorflow.proto.framework.BoundedTensorSpecProto) kind_).toBuilder();
+            }
+            kind_ =
+                input.readMessage(org.tensorflow.proto.framework.BoundedTensorSpecProto.parser(), extensionRegistry);
+            if (subBuilder != null) {
+              subBuilder.mergeFrom((org.tensorflow.proto.framework.BoundedTensorSpecProto) kind_);
+              kind_ = subBuilder.buildPartial();
+            }
+            kindCase_ = 35;
+            break;
+          }
+          case 410: {
+            org.tensorflow.proto.framework.ListValue.Builder subBuilder = null;
+            if (kindCase_ == 51) {
+              subBuilder = ((org.tensorflow.proto.framework.ListValue) kind_).toBuilder();
+            }
+            kind_ =
+                input.readMessage(org.tensorflow.proto.framework.ListValue.parser(), extensionRegistry);
+            if (subBuilder != null) {
+              subBuilder.mergeFrom((org.tensorflow.proto.framework.ListValue) kind_);
+              kind_ = subBuilder.buildPartial();
+            }
+            kindCase_ = 51;
+            break;
+          }
+          case 418: {
+            org.tensorflow.proto.framework.TupleValue.Builder subBuilder = null;
+            if (kindCase_ == 52) {
+              subBuilder = ((org.tensorflow.proto.framework.TupleValue) kind_).toBuilder();
+            }
+            kind_ =
+                input.readMessage(org.tensorflow.proto.framework.TupleValue.parser(), extensionRegistry);
+            if (subBuilder != null) {
+              subBuilder.mergeFrom((org.tensorflow.proto.framework.TupleValue) kind_);
+              kind_ = subBuilder.buildPartial();
+            }
+            kindCase_ = 52;
+            break;
+          }
+          case 426: {
+            org.tensorflow.proto.framework.DictValue.Builder subBuilder = null;
+            if (kindCase_ == 53) {
+              subBuilder = ((org.tensorflow.proto.framework.DictValue) kind_).toBuilder();
+            }
+            kind_ =
+                input.readMessage(org.tensorflow.proto.framework.DictValue.parser(), extensionRegistry);
+            if (subBuilder != null) {
+              subBuilder.mergeFrom((org.tensorflow.proto.framework.DictValue) kind_);
+              kind_ = subBuilder.buildPartial();
+            }
+            kindCase_ = 53;
+            break;
+          }
+          case 434: {
+            org.tensorflow.proto.framework.NamedTupleValue.Builder subBuilder = null;
+            if (kindCase_ == 54) {
+              subBuilder = ((org.tensorflow.proto.framework.NamedTupleValue) kind_).toBuilder();
+            }
+            kind_ =
+                input.readMessage(org.tensorflow.proto.framework.NamedTupleValue.parser(), extensionRegistry);
+            if (subBuilder != null) {
+              subBuilder.mergeFrom((org.tensorflow.proto.framework.NamedTupleValue) kind_);
+              kind_ = subBuilder.buildPartial();
+            }
+            kindCase_ = 54;
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.framework.StructProtos.internal_static_tensorflow_StructuredValue_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.framework.StructProtos.internal_static_tensorflow_StructuredValue_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.framework.StructuredValue.class, org.tensorflow.proto.framework.StructuredValue.Builder.class);
+  }
+
+  private int kindCase_ = 0;
+  private java.lang.Object kind_;
+  public enum KindCase
+      implements com.google.protobuf.Internal.EnumLite {
+    NONE_VALUE(1),
+    FLOAT64_VALUE(11),
+    INT64_VALUE(12),
+    STRING_VALUE(13),
+    BOOL_VALUE(14),
+    TENSOR_SHAPE_VALUE(31),
+    TENSOR_DTYPE_VALUE(32),
+    TENSOR_SPEC_VALUE(33),
+    TYPE_SPEC_VALUE(34),
+    BOUNDED_TENSOR_SPEC_VALUE(35),
+    LIST_VALUE(51),
+    TUPLE_VALUE(52),
+    DICT_VALUE(53),
+    NAMED_TUPLE_VALUE(54),
+    KIND_NOT_SET(0);
+    private final int value;
+    private KindCase(int value) {
+      this.value = value;
+    }
+    /**
+     * @deprecated Use {@link #forNumber(int)} instead.
+     */
+    @java.lang.Deprecated
+    public static KindCase valueOf(int value) {
+      return forNumber(value);
+    }
+
+    public static KindCase forNumber(int value) {
+      switch (value) {
+        case 1: return NONE_VALUE;
+        case 11: return FLOAT64_VALUE;
+        case 12: return INT64_VALUE;
+        case 13: return STRING_VALUE;
+        case 14: return BOOL_VALUE;
+        case 31: return TENSOR_SHAPE_VALUE;
+        case 32: return TENSOR_DTYPE_VALUE;
+        case 33: return TENSOR_SPEC_VALUE;
+        case 34: return TYPE_SPEC_VALUE;
+        case 35: return BOUNDED_TENSOR_SPEC_VALUE;
+        case 51: return LIST_VALUE;
+        case 52: return TUPLE_VALUE;
+        case 53: return DICT_VALUE;
+        case 54: return NAMED_TUPLE_VALUE;
+        case 0: return KIND_NOT_SET;
+        default: return null;
+      }
+    }
+    public int getNumber() {
+      return this.value;
+    }
+  };
+
+  public KindCase
+  getKindCase() {
+    return KindCase.forNumber(
+        kindCase_);
+  }
+
+  public static final int NONE_VALUE_FIELD_NUMBER = 1;
+  /**
+   * <pre>
+   * Represents None.
+   * </pre>
+   *
+   * <code>.tensorflow.NoneValue none_value = 1;</code>
+   */
+  public boolean hasNoneValue() {
+    return kindCase_ == 1;
+  }
+  /**
+   * <pre>
+   * Represents None.
+   * </pre>
+   *
+   * <code>.tensorflow.NoneValue none_value = 1;</code>
+   */
+  public org.tensorflow.proto.framework.NoneValue getNoneValue() {
+    if (kindCase_ == 1) {
+       return (org.tensorflow.proto.framework.NoneValue) kind_;
+    }
+    return org.tensorflow.proto.framework.NoneValue.getDefaultInstance();
+  }
+  /**
+   * <pre>
+   * Represents None.
+   * </pre>
+   *
+   * <code>.tensorflow.NoneValue none_value = 1;</code>
+   */
+  public org.tensorflow.proto.framework.NoneValueOrBuilder getNoneValueOrBuilder() {
+    if (kindCase_ == 1) {
+       return (org.tensorflow.proto.framework.NoneValue) kind_;
+    }
+    return org.tensorflow.proto.framework.NoneValue.getDefaultInstance();
+  }
+
+  public static final int FLOAT64_VALUE_FIELD_NUMBER = 11;
+  /**
+   * <pre>
+   * Represents a double-precision floating-point value (a Python `float`).
+   * </pre>
+   *
+   * <code>double float64_value = 11;</code>
+   */
+  public double getFloat64Value() {
+    if (kindCase_ == 11) {
+      return (java.lang.Double) kind_;
+    }
+    return 0D;
+  }
+
+  public static final int INT64_VALUE_FIELD_NUMBER = 12;
+  /**
+   * <pre>
+   * Represents a signed integer value, limited to 64 bits.
+   * Larger values from Python's arbitrary-precision integers are unsupported.
+   * </pre>
+   *
+   * <code>sint64 int64_value = 12;</code>
+   */
+  public long getInt64Value() {
+    if (kindCase_ == 12) {
+      return (java.lang.Long) kind_;
+    }
+    return 0L;
+  }
+
+  public static final int STRING_VALUE_FIELD_NUMBER = 13;
+  /**
+   * <pre>
+   * Represents a string of Unicode characters stored in a Python `str`.
+   * In Python 3, this is exactly what type `str` is.
+   * In Python 2, this is the UTF-8 encoding of the characters.
+   * For strings with ASCII characters only (as often used in TensorFlow code)
+   * there is effectively no difference between the language versions.
+   * The obsolescent `unicode` type of Python 2 is not supported here.
+   * </pre>
+   *
+   * <code>string string_value = 13;</code>
+   */
+  public java.lang.String getStringValue() {
+    java.lang.Object ref = "";
+    if (kindCase_ == 13) {
+      ref = kind_;
+    }
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      if (kindCase_ == 13) {
+        kind_ = s;
+      }
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * Represents a string of Unicode characters stored in a Python `str`.
+   * In Python 3, this is exactly what type `str` is.
+   * In Python 2, this is the UTF-8 encoding of the characters.
+   * For strings with ASCII characters only (as often used in TensorFlow code)
+   * there is effectively no difference between the language versions.
+   * The obsolescent `unicode` type of Python 2 is not supported here.
+   * </pre>
+   *
+   * <code>string string_value = 13;</code>
+   */
+  public com.google.protobuf.ByteString
+      getStringValueBytes() {
+    java.lang.Object ref = "";
+    if (kindCase_ == 13) {
+      ref = kind_;
+    }
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      if (kindCase_ == 13) {
+        kind_ = b;
+      }
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int BOOL_VALUE_FIELD_NUMBER = 14;
+  /**
+   * <pre>
+   * Represents a boolean value.
+   * </pre>
+   *
+   * <code>bool bool_value = 14;</code>
+   */
+  public boolean getBoolValue() {
+    if (kindCase_ == 14) {
+      return (java.lang.Boolean) kind_;
+    }
+    return false;
+  }
+
+  public static final int TENSOR_SHAPE_VALUE_FIELD_NUMBER = 31;
+  /**
+   * <pre>
+   * Represents a TensorShape.
+   * </pre>
+   *
+   * <code>.tensorflow.TensorShapeProto tensor_shape_value = 31;</code>
+   */
+  public boolean hasTensorShapeValue() {
+    return kindCase_ == 31;
+  }
+  /**
+   * <pre>
+   * Represents a TensorShape.
+   * </pre>
+   *
+   * <code>.tensorflow.TensorShapeProto tensor_shape_value = 31;</code>
+   */
+  public org.tensorflow.proto.framework.TensorShapeProto getTensorShapeValue() {
+    if (kindCase_ == 31) {
+       return (org.tensorflow.proto.framework.TensorShapeProto) kind_;
+    }
+    return org.tensorflow.proto.framework.TensorShapeProto.getDefaultInstance();
+  }
+  /**
+   * <pre>
+   * Represents a TensorShape.
+   * </pre>
+   *
+   * <code>.tensorflow.TensorShapeProto tensor_shape_value = 31;</code>
+   */
+  public org.tensorflow.proto.framework.TensorShapeProtoOrBuilder getTensorShapeValueOrBuilder() {
+    if (kindCase_ == 31) {
+       return (org.tensorflow.proto.framework.TensorShapeProto) kind_;
+    }
+    return org.tensorflow.proto.framework.TensorShapeProto.getDefaultInstance();
+  }
+
+  public static final int TENSOR_DTYPE_VALUE_FIELD_NUMBER = 32;
+  /**
+   * <pre>
+   * Represents an enum value for dtype.
+   * </pre>
+   *
+   * <code>.tensorflow.DataType tensor_dtype_value = 32;</code>
+   */
+  public int getTensorDtypeValueValue() {
+    if (kindCase_ == 32) {
+      return (java.lang.Integer) kind_;
+    }
+    return 0;
+  }
+  /**
+   * <pre>
+   * Represents an enum value for dtype.
+   * </pre>
+   *
+   * <code>.tensorflow.DataType tensor_dtype_value = 32;</code>
+   */
+  public org.tensorflow.proto.framework.DataType getTensorDtypeValue() {
+    if (kindCase_ == 32) {
+      @SuppressWarnings("deprecation")
+      org.tensorflow.proto.framework.DataType result = org.tensorflow.proto.framework.DataType.valueOf(
+          (java.lang.Integer) kind_);
+      return result == null ? org.tensorflow.proto.framework.DataType.UNRECOGNIZED : result;
+    }
+    return org.tensorflow.proto.framework.DataType.DT_INVALID;
+  }
+
+  public static final int TENSOR_SPEC_VALUE_FIELD_NUMBER = 33;
+  /**
+   * <pre>
+   * Represents a value for tf.TensorSpec.
+   * </pre>
+   *
+   * <code>.tensorflow.TensorSpecProto tensor_spec_value = 33;</code>
+   */
+  public boolean hasTensorSpecValue() {
+    return kindCase_ == 33;
+  }
+  /**
+   * <pre>
+   * Represents a value for tf.TensorSpec.
+   * </pre>
+   *
+   * <code>.tensorflow.TensorSpecProto tensor_spec_value = 33;</code>
+   */
+  public org.tensorflow.proto.framework.TensorSpecProto getTensorSpecValue() {
+    if (kindCase_ == 33) {
+       return (org.tensorflow.proto.framework.TensorSpecProto) kind_;
+    }
+    return org.tensorflow.proto.framework.TensorSpecProto.getDefaultInstance();
+  }
+  /**
+   * <pre>
+   * Represents a value for tf.TensorSpec.
+   * </pre>
+   *
+   * <code>.tensorflow.TensorSpecProto tensor_spec_value = 33;</code>
+   */
+  public org.tensorflow.proto.framework.TensorSpecProtoOrBuilder getTensorSpecValueOrBuilder() {
+    if (kindCase_ == 33) {
+       return (org.tensorflow.proto.framework.TensorSpecProto) kind_;
+    }
+    return org.tensorflow.proto.framework.TensorSpecProto.getDefaultInstance();
+  }
+
+  public static final int TYPE_SPEC_VALUE_FIELD_NUMBER = 34;
+  /**
+   * <pre>
+   * Represents a value for tf.TypeSpec.
+   * </pre>
+   *
+   * <code>.tensorflow.TypeSpecProto type_spec_value = 34;</code>
+   */
+  public boolean hasTypeSpecValue() {
+    return kindCase_ == 34;
+  }
+  /**
+   * <pre>
+   * Represents a value for tf.TypeSpec.
+   * </pre>
+   *
+   * <code>.tensorflow.TypeSpecProto type_spec_value = 34;</code>
+   */
+  public org.tensorflow.proto.framework.TypeSpecProto getTypeSpecValue() {
+    if (kindCase_ == 34) {
+       return (org.tensorflow.proto.framework.TypeSpecProto) kind_;
+    }
+    return org.tensorflow.proto.framework.TypeSpecProto.getDefaultInstance();
+  }
+  /**
+   * <pre>
+   * Represents a value for tf.TypeSpec.
+   * </pre>
+   *
+   * <code>.tensorflow.TypeSpecProto type_spec_value = 34;</code>
+   */
+  public org.tensorflow.proto.framework.TypeSpecProtoOrBuilder getTypeSpecValueOrBuilder() {
+    if (kindCase_ == 34) {
+       return (org.tensorflow.proto.framework.TypeSpecProto) kind_;
+    }
+    return org.tensorflow.proto.framework.TypeSpecProto.getDefaultInstance();
+  }
+
+  public static final int BOUNDED_TENSOR_SPEC_VALUE_FIELD_NUMBER = 35;
+  /**
+   * <pre>
+   * Represents a value for tf.BoundedTensorSpec.
+   * </pre>
+   *
+   * <code>.tensorflow.BoundedTensorSpecProto bounded_tensor_spec_value = 35;</code>
+   */
+  public boolean hasBoundedTensorSpecValue() {
+    return kindCase_ == 35;
+  }
+  /**
+   * <pre>
+   * Represents a value for tf.BoundedTensorSpec.
+   * </pre>
+   *
+   * <code>.tensorflow.BoundedTensorSpecProto bounded_tensor_spec_value = 35;</code>
+   */
+  public org.tensorflow.proto.framework.BoundedTensorSpecProto getBoundedTensorSpecValue() {
+    if (kindCase_ == 35) {
+       return (org.tensorflow.proto.framework.BoundedTensorSpecProto) kind_;
+    }
+    return org.tensorflow.proto.framework.BoundedTensorSpecProto.getDefaultInstance();
+  }
+  /**
+   * <pre>
+   * Represents a value for tf.BoundedTensorSpec.
+   * </pre>
+   *
+   * <code>.tensorflow.BoundedTensorSpecProto bounded_tensor_spec_value = 35;</code>
+   */
+  public org.tensorflow.proto.framework.BoundedTensorSpecProtoOrBuilder getBoundedTensorSpecValueOrBuilder() {
+    if (kindCase_ == 35) {
+       return (org.tensorflow.proto.framework.BoundedTensorSpecProto) kind_;
+    }
+    return org.tensorflow.proto.framework.BoundedTensorSpecProto.getDefaultInstance();
+  }
+
+  public static final int LIST_VALUE_FIELD_NUMBER = 51;
+  /**
+   * <pre>
+   * Represents a list of `Value`.
+   * </pre>
+   *
+   * <code>.tensorflow.ListValue list_value = 51;</code>
+   */
+  public boolean hasListValue() {
+    return kindCase_ == 51;
+  }
+  /**
+   * <pre>
+   * Represents a list of `Value`.
+   * </pre>
+   *
+   * <code>.tensorflow.ListValue list_value = 51;</code>
+   */
+  public org.tensorflow.proto.framework.ListValue getListValue() {
+    if (kindCase_ == 51) {
+       return (org.tensorflow.proto.framework.ListValue) kind_;
+    }
+    return org.tensorflow.proto.framework.ListValue.getDefaultInstance();
+  }
+  /**
+   * <pre>
+   * Represents a list of `Value`.
+   * </pre>
+   *
+   * <code>.tensorflow.ListValue list_value = 51;</code>
+   */
+  public org.tensorflow.proto.framework.ListValueOrBuilder getListValueOrBuilder() {
+    if (kindCase_ == 51) {
+       return (org.tensorflow.proto.framework.ListValue) kind_;
+    }
+    return org.tensorflow.proto.framework.ListValue.getDefaultInstance();
+  }
+
+  public static final int TUPLE_VALUE_FIELD_NUMBER = 52;
+  /**
+   * <pre>
+   * Represents a tuple of `Value`.
+   * </pre>
+   *
+   * <code>.tensorflow.TupleValue tuple_value = 52;</code>
+   */
+  public boolean hasTupleValue() {
+    return kindCase_ == 52;
+  }
+  /**
+   * <pre>
+   * Represents a tuple of `Value`.
+   * </pre>
+   *
+   * <code>.tensorflow.TupleValue tuple_value = 52;</code>
+   */
+  public org.tensorflow.proto.framework.TupleValue getTupleValue() {
+    if (kindCase_ == 52) {
+       return (org.tensorflow.proto.framework.TupleValue) kind_;
+    }
+    return org.tensorflow.proto.framework.TupleValue.getDefaultInstance();
+  }
+  /**
+   * <pre>
+   * Represents a tuple of `Value`.
+   * </pre>
+   *
+   * <code>.tensorflow.TupleValue tuple_value = 52;</code>
+   */
+  public org.tensorflow.proto.framework.TupleValueOrBuilder getTupleValueOrBuilder() {
+    if (kindCase_ == 52) {
+       return (org.tensorflow.proto.framework.TupleValue) kind_;
+    }
+    return org.tensorflow.proto.framework.TupleValue.getDefaultInstance();
+  }
+
+  public static final int DICT_VALUE_FIELD_NUMBER = 53;
+  /**
+   * <pre>
+   * Represents a dict `Value`.
+   * </pre>
+   *
+   * <code>.tensorflow.DictValue dict_value = 53;</code>
+   */
+  public boolean hasDictValue() {
+    return kindCase_ == 53;
+  }
+  /**
+   * <pre>
+   * Represents a dict `Value`.
+   * </pre>
+   *
+   * <code>.tensorflow.DictValue dict_value = 53;</code>
+   */
+  public org.tensorflow.proto.framework.DictValue getDictValue() {
+    if (kindCase_ == 53) {
+       return (org.tensorflow.proto.framework.DictValue) kind_;
+    }
+    return org.tensorflow.proto.framework.DictValue.getDefaultInstance();
+  }
+  /**
+   * <pre>
+   * Represents a dict `Value`.
+   * </pre>
+   *
+   * <code>.tensorflow.DictValue dict_value = 53;</code>
+   */
+  public org.tensorflow.proto.framework.DictValueOrBuilder getDictValueOrBuilder() {
+    if (kindCase_ == 53) {
+       return (org.tensorflow.proto.framework.DictValue) kind_;
+    }
+    return org.tensorflow.proto.framework.DictValue.getDefaultInstance();
+  }
+
+  public static final int NAMED_TUPLE_VALUE_FIELD_NUMBER = 54;
+  /**
+   * <pre>
+   * Represents Python's namedtuple.
+   * </pre>
+   *
+   * <code>.tensorflow.NamedTupleValue named_tuple_value = 54;</code>
+   */
+  public boolean hasNamedTupleValue() {
+    return kindCase_ == 54;
+  }
+  /**
+   * <pre>
+   * Represents Python's namedtuple.
+   * </pre>
+   *
+   * <code>.tensorflow.NamedTupleValue named_tuple_value = 54;</code>
+   */
+  public org.tensorflow.proto.framework.NamedTupleValue getNamedTupleValue() {
+    if (kindCase_ == 54) {
+       return (org.tensorflow.proto.framework.NamedTupleValue) kind_;
+    }
+    return org.tensorflow.proto.framework.NamedTupleValue.getDefaultInstance();
+  }
+  /**
+   * <pre>
+   * Represents Python's namedtuple.
+   * </pre>
+   *
+   * <code>.tensorflow.NamedTupleValue named_tuple_value = 54;</code>
+   */
+  public org.tensorflow.proto.framework.NamedTupleValueOrBuilder getNamedTupleValueOrBuilder() {
+    if (kindCase_ == 54) {
+       return (org.tensorflow.proto.framework.NamedTupleValue) kind_;
+    }
+    return org.tensorflow.proto.framework.NamedTupleValue.getDefaultInstance();
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    if (kindCase_ == 1) {
+      output.writeMessage(1, (org.tensorflow.proto.framework.NoneValue) kind_);
+    }
+    if (kindCase_ == 11) {
+      output.writeDouble(
+          11, (double)((java.lang.Double) kind_));
+    }
+    if (kindCase_ == 12) {
+      output.writeSInt64(
+          12, (long)((java.lang.Long) kind_));
+    }
+    if (kindCase_ == 13) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 13, kind_);
+    }
+    if (kindCase_ == 14) {
+      output.writeBool(
+          14, (boolean)((java.lang.Boolean) kind_));
+    }
+    if (kindCase_ == 31) {
+      output.writeMessage(31, (org.tensorflow.proto.framework.TensorShapeProto) kind_);
+    }
+    if (kindCase_ == 32) {
+      output.writeEnum(32, ((java.lang.Integer) kind_));
+    }
+    if (kindCase_ == 33) {
+      output.writeMessage(33, (org.tensorflow.proto.framework.TensorSpecProto) kind_);
+    }
+    if (kindCase_ == 34) {
+      output.writeMessage(34, (org.tensorflow.proto.framework.TypeSpecProto) kind_);
+    }
+    if (kindCase_ == 35) {
+      output.writeMessage(35, (org.tensorflow.proto.framework.BoundedTensorSpecProto) kind_);
+    }
+    if (kindCase_ == 51) {
+      output.writeMessage(51, (org.tensorflow.proto.framework.ListValue) kind_);
+    }
+    if (kindCase_ == 52) {
+      output.writeMessage(52, (org.tensorflow.proto.framework.TupleValue) kind_);
+    }
+    if (kindCase_ == 53) {
+      output.writeMessage(53, (org.tensorflow.proto.framework.DictValue) kind_);
+    }
+    if (kindCase_ == 54) {
+      output.writeMessage(54, (org.tensorflow.proto.framework.NamedTupleValue) kind_);
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (kindCase_ == 1) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(1, (org.tensorflow.proto.framework.NoneValue) kind_);
+    }
+    if (kindCase_ == 11) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeDoubleSize(
+            11, (double)((java.lang.Double) kind_));
+    }
+    if (kindCase_ == 12) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeSInt64Size(
+            12, (long)((java.lang.Long) kind_));
+    }
+    if (kindCase_ == 13) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(13, kind_);
+    }
+    if (kindCase_ == 14) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeBoolSize(
+            14, (boolean)((java.lang.Boolean) kind_));
+    }
+    if (kindCase_ == 31) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(31, (org.tensorflow.proto.framework.TensorShapeProto) kind_);
+    }
+    if (kindCase_ == 32) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeEnumSize(32, ((java.lang.Integer) kind_));
+    }
+    if (kindCase_ == 33) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(33, (org.tensorflow.proto.framework.TensorSpecProto) kind_);
+    }
+    if (kindCase_ == 34) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(34, (org.tensorflow.proto.framework.TypeSpecProto) kind_);
+    }
+    if (kindCase_ == 35) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(35, (org.tensorflow.proto.framework.BoundedTensorSpecProto) kind_);
+    }
+    if (kindCase_ == 51) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(51, (org.tensorflow.proto.framework.ListValue) kind_);
+    }
+    if (kindCase_ == 52) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(52, (org.tensorflow.proto.framework.TupleValue) kind_);
+    }
+    if (kindCase_ == 53) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(53, (org.tensorflow.proto.framework.DictValue) kind_);
+    }
+    if (kindCase_ == 54) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(54, (org.tensorflow.proto.framework.NamedTupleValue) kind_);
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.framework.StructuredValue)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.framework.StructuredValue other = (org.tensorflow.proto.framework.StructuredValue) obj;
+
+    if (!getKindCase().equals(other.getKindCase())) return false;
+    switch (kindCase_) {
+      case 1:
+        if (!getNoneValue()
+            .equals(other.getNoneValue())) return false;
+        break;
+      case 11:
+        if (java.lang.Double.doubleToLongBits(getFloat64Value())
+            != java.lang.Double.doubleToLongBits(
+                other.getFloat64Value())) return false;
+        break;
+      case 12:
+        if (getInt64Value()
+            != other.getInt64Value()) return false;
+        break;
+      case 13:
+        if (!getStringValue()
+            .equals(other.getStringValue())) return false;
+        break;
+      case 14:
+        if (getBoolValue()
+            != other.getBoolValue()) return false;
+        break;
+      case 31:
+        if (!getTensorShapeValue()
+            .equals(other.getTensorShapeValue())) return false;
+        break;
+      case 32:
+        if (getTensorDtypeValueValue()
+            != other.getTensorDtypeValueValue()) return false;
+        break;
+      case 33:
+        if (!getTensorSpecValue()
+            .equals(other.getTensorSpecValue())) return false;
+        break;
+      case 34:
+        if (!getTypeSpecValue()
+            .equals(other.getTypeSpecValue())) return false;
+        break;
+      case 35:
+        if (!getBoundedTensorSpecValue()
+            .equals(other.getBoundedTensorSpecValue())) return false;
+        break;
+      case 51:
+        if (!getListValue()
+            .equals(other.getListValue())) return false;
+        break;
+      case 52:
+        if (!getTupleValue()
+            .equals(other.getTupleValue())) return false;
+        break;
+      case 53:
+        if (!getDictValue()
+            .equals(other.getDictValue())) return false;
+        break;
+      case 54:
+        if (!getNamedTupleValue()
+            .equals(other.getNamedTupleValue())) return false;
+        break;
+      case 0:
+      default:
+    }
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    switch (kindCase_) {
+      case 1:
+        hash = (37 * hash) + NONE_VALUE_FIELD_NUMBER;
+        hash = (53 * hash) + getNoneValue().hashCode();
+        break;
+      case 11:
+        hash = (37 * hash) + FLOAT64_VALUE_FIELD_NUMBER;
+        hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+            java.lang.Double.doubleToLongBits(getFloat64Value()));
+        break;
+      case 12:
+        hash = (37 * hash) + INT64_VALUE_FIELD_NUMBER;
+        hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+            getInt64Value());
+        break;
+      case 13:
+        hash = (37 * hash) + STRING_VALUE_FIELD_NUMBER;
+        hash = (53 * hash) + getStringValue().hashCode();
+        break;
+      case 14:
+        hash = (37 * hash) + BOOL_VALUE_FIELD_NUMBER;
+        hash = (53 * hash) + com.google.protobuf.Internal.hashBoolean(
+            getBoolValue());
+        break;
+      case 31:
+        hash = (37 * hash) + TENSOR_SHAPE_VALUE_FIELD_NUMBER;
+        hash = (53 * hash) + getTensorShapeValue().hashCode();
+        break;
+      case 32:
+        hash = (37 * hash) + TENSOR_DTYPE_VALUE_FIELD_NUMBER;
+        hash = (53 * hash) + getTensorDtypeValueValue();
+        break;
+      case 33:
+        hash = (37 * hash) + TENSOR_SPEC_VALUE_FIELD_NUMBER;
+        hash = (53 * hash) + getTensorSpecValue().hashCode();
+        break;
+      case 34:
+        hash = (37 * hash) + TYPE_SPEC_VALUE_FIELD_NUMBER;
+        hash = (53 * hash) + getTypeSpecValue().hashCode();
+        break;
+      case 35:
+        hash = (37 * hash) + BOUNDED_TENSOR_SPEC_VALUE_FIELD_NUMBER;
+        hash = (53 * hash) + getBoundedTensorSpecValue().hashCode();
+        break;
+      case 51:
+        hash = (37 * hash) + LIST_VALUE_FIELD_NUMBER;
+        hash = (53 * hash) + getListValue().hashCode();
+        break;
+      case 52:
+        hash = (37 * hash) + TUPLE_VALUE_FIELD_NUMBER;
+        hash = (53 * hash) + getTupleValue().hashCode();
+        break;
+      case 53:
+        hash = (37 * hash) + DICT_VALUE_FIELD_NUMBER;
+        hash = (53 * hash) + getDictValue().hashCode();
+        break;
+      case 54:
+        hash = (37 * hash) + NAMED_TUPLE_VALUE_FIELD_NUMBER;
+        hash = (53 * hash) + getNamedTupleValue().hashCode();
+        break;
+      case 0:
+      default:
+    }
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.framework.StructuredValue parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.StructuredValue parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.StructuredValue parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.StructuredValue parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.StructuredValue parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.StructuredValue parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.StructuredValue parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.StructuredValue parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.StructuredValue parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.StructuredValue parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.StructuredValue parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.StructuredValue parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.framework.StructuredValue prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * <pre>
+   * `StructuredValue` represents a dynamically typed value representing various
+   * data structures that are inspired by Python data structures typically used in
+   * TensorFlow functions as inputs and outputs.
+   * For example when saving a Layer there may be a `training` argument. If the
+   * user passes a boolean True/False, that switches between two concrete
+   * TensorFlow functions. In order to switch between them in the same way after
+   * loading the SavedModel, we need to represent "True" and "False".
+   * A more advanced example might be a function which takes a list of
+   * dictionaries mapping from strings to Tensors. In order to map from
+   * user-specified arguments `[{"a": tf.constant(1.)}, {"q": tf.constant(3.)}]`
+   * after load to the right saved TensorFlow function, we need to represent the
+   * nested structure and the strings, recording that we have a trace for anything
+   * matching `[{"a": tf.TensorSpec(None, tf.float32)}, {"q": tf.TensorSpec([],
+   * tf.float64)}]` as an example.
+   * Likewise functions may return nested structures of Tensors, for example
+   * returning a dictionary mapping from strings to Tensors. In order for the
+   * loaded function to return the same structure we need to serialize it.
+   * This is an ergonomic aid for working with loaded SavedModels, not a promise
+   * to serialize all possible function signatures. For example we do not expect
+   * to pickle generic Python objects, and ideally we'd stay language-agnostic.
+   * </pre>
+   *
+   * Protobuf type {@code tensorflow.StructuredValue}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.StructuredValue)
+      org.tensorflow.proto.framework.StructuredValueOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.StructProtos.internal_static_tensorflow_StructuredValue_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.StructProtos.internal_static_tensorflow_StructuredValue_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.StructuredValue.class, org.tensorflow.proto.framework.StructuredValue.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.framework.StructuredValue.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      kindCase_ = 0;
+      kind_ = null;
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.framework.StructProtos.internal_static_tensorflow_StructuredValue_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.StructuredValue getDefaultInstanceForType() {
+      return org.tensorflow.proto.framework.StructuredValue.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.StructuredValue build() {
+      org.tensorflow.proto.framework.StructuredValue result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.StructuredValue buildPartial() {
+      org.tensorflow.proto.framework.StructuredValue result = new org.tensorflow.proto.framework.StructuredValue(this);
+      if (kindCase_ == 1) {
+        if (noneValueBuilder_ == null) {
+          result.kind_ = kind_;
+        } else {
+          result.kind_ = noneValueBuilder_.build();
+        }
+      }
+      if (kindCase_ == 11) {
+        result.kind_ = kind_;
+      }
+      if (kindCase_ == 12) {
+        result.kind_ = kind_;
+      }
+      if (kindCase_ == 13) {
+        result.kind_ = kind_;
+      }
+      if (kindCase_ == 14) {
+        result.kind_ = kind_;
+      }
+      if (kindCase_ == 31) {
+        if (tensorShapeValueBuilder_ == null) {
+          result.kind_ = kind_;
+        } else {
+          result.kind_ = tensorShapeValueBuilder_.build();
+        }
+      }
+      if (kindCase_ == 32) {
+        result.kind_ = kind_;
+      }
+      if (kindCase_ == 33) {
+        if (tensorSpecValueBuilder_ == null) {
+          result.kind_ = kind_;
+        } else {
+          result.kind_ = tensorSpecValueBuilder_.build();
+        }
+      }
+      if (kindCase_ == 34) {
+        if (typeSpecValueBuilder_ == null) {
+          result.kind_ = kind_;
+        } else {
+          result.kind_ = typeSpecValueBuilder_.build();
+        }
+      }
+      if (kindCase_ == 35) {
+        if (boundedTensorSpecValueBuilder_ == null) {
+          result.kind_ = kind_;
+        } else {
+          result.kind_ = boundedTensorSpecValueBuilder_.build();
+        }
+      }
+      if (kindCase_ == 51) {
+        if (listValueBuilder_ == null) {
+          result.kind_ = kind_;
+        } else {
+          result.kind_ = listValueBuilder_.build();
+        }
+      }
+      if (kindCase_ == 52) {
+        if (tupleValueBuilder_ == null) {
+          result.kind_ = kind_;
+        } else {
+          result.kind_ = tupleValueBuilder_.build();
+        }
+      }
+      if (kindCase_ == 53) {
+        if (dictValueBuilder_ == null) {
+          result.kind_ = kind_;
+        } else {
+          result.kind_ = dictValueBuilder_.build();
+        }
+      }
+      if (kindCase_ == 54) {
+        if (namedTupleValueBuilder_ == null) {
+          result.kind_ = kind_;
+        } else {
+          result.kind_ = namedTupleValueBuilder_.build();
+        }
+      }
+      result.kindCase_ = kindCase_;
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.framework.StructuredValue) {
+        return mergeFrom((org.tensorflow.proto.framework.StructuredValue)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.framework.StructuredValue other) {
+      if (other == org.tensorflow.proto.framework.StructuredValue.getDefaultInstance()) return this;
+      switch (other.getKindCase()) {
+        case NONE_VALUE: {
+          mergeNoneValue(other.getNoneValue());
+          break;
+        }
+        case FLOAT64_VALUE: {
+          setFloat64Value(other.getFloat64Value());
+          break;
+        }
+        case INT64_VALUE: {
+          setInt64Value(other.getInt64Value());
+          break;
+        }
+        case STRING_VALUE: {
+          kindCase_ = 13;
+          kind_ = other.kind_;
+          onChanged();
+          break;
+        }
+        case BOOL_VALUE: {
+          setBoolValue(other.getBoolValue());
+          break;
+        }
+        case TENSOR_SHAPE_VALUE: {
+          mergeTensorShapeValue(other.getTensorShapeValue());
+          break;
+        }
+        case TENSOR_DTYPE_VALUE: {
+          setTensorDtypeValueValue(other.getTensorDtypeValueValue());
+          break;
+        }
+        case TENSOR_SPEC_VALUE: {
+          mergeTensorSpecValue(other.getTensorSpecValue());
+          break;
+        }
+        case TYPE_SPEC_VALUE: {
+          mergeTypeSpecValue(other.getTypeSpecValue());
+          break;
+        }
+        case BOUNDED_TENSOR_SPEC_VALUE: {
+          mergeBoundedTensorSpecValue(other.getBoundedTensorSpecValue());
+          break;
+        }
+        case LIST_VALUE: {
+          mergeListValue(other.getListValue());
+          break;
+        }
+        case TUPLE_VALUE: {
+          mergeTupleValue(other.getTupleValue());
+          break;
+        }
+        case DICT_VALUE: {
+          mergeDictValue(other.getDictValue());
+          break;
+        }
+        case NAMED_TUPLE_VALUE: {
+          mergeNamedTupleValue(other.getNamedTupleValue());
+          break;
+        }
+        case KIND_NOT_SET: {
+          break;
+        }
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.framework.StructuredValue parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.framework.StructuredValue) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+    private int kindCase_ = 0;
+    private java.lang.Object kind_;
+    public KindCase
+        getKindCase() {
+      return KindCase.forNumber(
+          kindCase_);
+    }
+
+    public Builder clearKind() {
+      kindCase_ = 0;
+      kind_ = null;
+      onChanged();
+      return this;
+    }
+
+
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.NoneValue, org.tensorflow.proto.framework.NoneValue.Builder, org.tensorflow.proto.framework.NoneValueOrBuilder> noneValueBuilder_;
+    /**
+     * <pre>
+     * Represents None.
+     * </pre>
+     *
+     * <code>.tensorflow.NoneValue none_value = 1;</code>
+     */
+    public boolean hasNoneValue() {
+      return kindCase_ == 1;
+    }
+    /**
+     * <pre>
+     * Represents None.
+     * </pre>
+     *
+     * <code>.tensorflow.NoneValue none_value = 1;</code>
+     */
+    public org.tensorflow.proto.framework.NoneValue getNoneValue() {
+      if (noneValueBuilder_ == null) {
+        if (kindCase_ == 1) {
+          return (org.tensorflow.proto.framework.NoneValue) kind_;
+        }
+        return org.tensorflow.proto.framework.NoneValue.getDefaultInstance();
+      } else {
+        if (kindCase_ == 1) {
+          return noneValueBuilder_.getMessage();
+        }
+        return org.tensorflow.proto.framework.NoneValue.getDefaultInstance();
+      }
+    }
+    /**
+     * <pre>
+     * Represents None.
+     * </pre>
+     *
+     * <code>.tensorflow.NoneValue none_value = 1;</code>
+     */
+    public Builder setNoneValue(org.tensorflow.proto.framework.NoneValue value) {
+      if (noneValueBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        kind_ = value;
+        onChanged();
+      } else {
+        noneValueBuilder_.setMessage(value);
+      }
+      kindCase_ = 1;
+      return this;
+    }
+    /**
+     * <pre>
+     * Represents None.
+     * </pre>
+     *
+     * <code>.tensorflow.NoneValue none_value = 1;</code>
+     */
+    public Builder setNoneValue(
+        org.tensorflow.proto.framework.NoneValue.Builder builderForValue) {
+      if (noneValueBuilder_ == null) {
+        kind_ = builderForValue.build();
+        onChanged();
+      } else {
+        noneValueBuilder_.setMessage(builderForValue.build());
+      }
+      kindCase_ = 1;
+      return this;
+    }
+    /**
+     * <pre>
+     * Represents None.
+     * </pre>
+     *
+     * <code>.tensorflow.NoneValue none_value = 1;</code>
+     */
+    public Builder mergeNoneValue(org.tensorflow.proto.framework.NoneValue value) {
+      if (noneValueBuilder_ == null) {
+        if (kindCase_ == 1 &&
+            kind_ != org.tensorflow.proto.framework.NoneValue.getDefaultInstance()) {
+          kind_ = org.tensorflow.proto.framework.NoneValue.newBuilder((org.tensorflow.proto.framework.NoneValue) kind_)
+              .mergeFrom(value).buildPartial();
+        } else {
+          kind_ = value;
+        }
+        onChanged();
+      } else {
+        if (kindCase_ == 1) {
+          noneValueBuilder_.mergeFrom(value);
+        }
+        noneValueBuilder_.setMessage(value);
+      }
+      kindCase_ = 1;
+      return this;
+    }
+    /**
+     * <pre>
+     * Represents None.
+     * </pre>
+     *
+     * <code>.tensorflow.NoneValue none_value = 1;</code>
+     */
+    public Builder clearNoneValue() {
+      if (noneValueBuilder_ == null) {
+        if (kindCase_ == 1) {
+          kindCase_ = 0;
+          kind_ = null;
+          onChanged();
+        }
+      } else {
+        if (kindCase_ == 1) {
+          kindCase_ = 0;
+          kind_ = null;
+        }
+        noneValueBuilder_.clear();
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Represents None.
+     * </pre>
+     *
+     * <code>.tensorflow.NoneValue none_value = 1;</code>
+     */
+    public org.tensorflow.proto.framework.NoneValue.Builder getNoneValueBuilder() {
+      return getNoneValueFieldBuilder().getBuilder();
+    }
+    /**
+     * <pre>
+     * Represents None.
+     * </pre>
+     *
+     * <code>.tensorflow.NoneValue none_value = 1;</code>
+     */
+    public org.tensorflow.proto.framework.NoneValueOrBuilder getNoneValueOrBuilder() {
+      if ((kindCase_ == 1) && (noneValueBuilder_ != null)) {
+        return noneValueBuilder_.getMessageOrBuilder();
+      } else {
+        if (kindCase_ == 1) {
+          return (org.tensorflow.proto.framework.NoneValue) kind_;
+        }
+        return org.tensorflow.proto.framework.NoneValue.getDefaultInstance();
+      }
+    }
+    /**
+     * <pre>
+     * Represents None.
+     * </pre>
+     *
+     * <code>.tensorflow.NoneValue none_value = 1;</code>
+     */
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.NoneValue, org.tensorflow.proto.framework.NoneValue.Builder, org.tensorflow.proto.framework.NoneValueOrBuilder> 
+        getNoneValueFieldBuilder() {
+      if (noneValueBuilder_ == null) {
+        if (!(kindCase_ == 1)) {
+          kind_ = org.tensorflow.proto.framework.NoneValue.getDefaultInstance();
+        }
+        noneValueBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+            org.tensorflow.proto.framework.NoneValue, org.tensorflow.proto.framework.NoneValue.Builder, org.tensorflow.proto.framework.NoneValueOrBuilder>(
+                (org.tensorflow.proto.framework.NoneValue) kind_,
+                getParentForChildren(),
+                isClean());
+        kind_ = null;
+      }
+      kindCase_ = 1;
+      onChanged();;
+      return noneValueBuilder_;
+    }
+
+    /**
+     * <pre>
+     * Represents a double-precision floating-point value (a Python `float`).
+     * </pre>
+     *
+     * <code>double float64_value = 11;</code>
+     */
+    public double getFloat64Value() {
+      if (kindCase_ == 11) {
+        return (java.lang.Double) kind_;
+      }
+      return 0D;
+    }
+    /**
+     * <pre>
+     * Represents a double-precision floating-point value (a Python `float`).
+     * </pre>
+     *
+     * <code>double float64_value = 11;</code>
+     */
+    public Builder setFloat64Value(double value) {
+      kindCase_ = 11;
+      kind_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Represents a double-precision floating-point value (a Python `float`).
+     * </pre>
+     *
+     * <code>double float64_value = 11;</code>
+     */
+    public Builder clearFloat64Value() {
+      if (kindCase_ == 11) {
+        kindCase_ = 0;
+        kind_ = null;
+        onChanged();
+      }
+      return this;
+    }
+
+    /**
+     * <pre>
+     * Represents a signed integer value, limited to 64 bits.
+     * Larger values from Python's arbitrary-precision integers are unsupported.
+     * </pre>
+     *
+     * <code>sint64 int64_value = 12;</code>
+     */
+    public long getInt64Value() {
+      if (kindCase_ == 12) {
+        return (java.lang.Long) kind_;
+      }
+      return 0L;
+    }
+    /**
+     * <pre>
+     * Represents a signed integer value, limited to 64 bits.
+     * Larger values from Python's arbitrary-precision integers are unsupported.
+     * </pre>
+     *
+     * <code>sint64 int64_value = 12;</code>
+     */
+    public Builder setInt64Value(long value) {
+      kindCase_ = 12;
+      kind_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Represents a signed integer value, limited to 64 bits.
+     * Larger values from Python's arbitrary-precision integers are unsupported.
+     * </pre>
+     *
+     * <code>sint64 int64_value = 12;</code>
+     */
+    public Builder clearInt64Value() {
+      if (kindCase_ == 12) {
+        kindCase_ = 0;
+        kind_ = null;
+        onChanged();
+      }
+      return this;
+    }
+
+    /**
+     * <pre>
+     * Represents a string of Unicode characters stored in a Python `str`.
+     * In Python 3, this is exactly what type `str` is.
+     * In Python 2, this is the UTF-8 encoding of the characters.
+     * For strings with ASCII characters only (as often used in TensorFlow code)
+     * there is effectively no difference between the language versions.
+     * The obsolescent `unicode` type of Python 2 is not supported here.
+     * </pre>
+     *
+     * <code>string string_value = 13;</code>
+     */
+    public java.lang.String getStringValue() {
+      java.lang.Object ref = "";
+      if (kindCase_ == 13) {
+        ref = kind_;
+      }
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        if (kindCase_ == 13) {
+          kind_ = s;
+        }
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Represents a string of Unicode characters stored in a Python `str`.
+     * In Python 3, this is exactly what type `str` is.
+     * In Python 2, this is the UTF-8 encoding of the characters.
+     * For strings with ASCII characters only (as often used in TensorFlow code)
+     * there is effectively no difference between the language versions.
+     * The obsolescent `unicode` type of Python 2 is not supported here.
+     * </pre>
+     *
+     * <code>string string_value = 13;</code>
+     */
+    public com.google.protobuf.ByteString
+        getStringValueBytes() {
+      java.lang.Object ref = "";
+      if (kindCase_ == 13) {
+        ref = kind_;
+      }
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        if (kindCase_ == 13) {
+          kind_ = b;
+        }
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Represents a string of Unicode characters stored in a Python `str`.
+     * In Python 3, this is exactly what type `str` is.
+     * In Python 2, this is the UTF-8 encoding of the characters.
+     * For strings with ASCII characters only (as often used in TensorFlow code)
+     * there is effectively no difference between the language versions.
+     * The obsolescent `unicode` type of Python 2 is not supported here.
+     * </pre>
+     *
+     * <code>string string_value = 13;</code>
+     */
+    public Builder setStringValue(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  kindCase_ = 13;
+      kind_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Represents a string of Unicode characters stored in a Python `str`.
+     * In Python 3, this is exactly what type `str` is.
+     * In Python 2, this is the UTF-8 encoding of the characters.
+     * For strings with ASCII characters only (as often used in TensorFlow code)
+     * there is effectively no difference between the language versions.
+     * The obsolescent `unicode` type of Python 2 is not supported here.
+     * </pre>
+     *
+     * <code>string string_value = 13;</code>
+     */
+    public Builder clearStringValue() {
+      if (kindCase_ == 13) {
+        kindCase_ = 0;
+        kind_ = null;
+        onChanged();
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Represents a string of Unicode characters stored in a Python `str`.
+     * In Python 3, this is exactly what type `str` is.
+     * In Python 2, this is the UTF-8 encoding of the characters.
+     * For strings with ASCII characters only (as often used in TensorFlow code)
+     * there is effectively no difference between the language versions.
+     * The obsolescent `unicode` type of Python 2 is not supported here.
+     * </pre>
+     *
+     * <code>string string_value = 13;</code>
+     */
+    public Builder setStringValueBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      kindCase_ = 13;
+      kind_ = value;
+      onChanged();
+      return this;
+    }
+
+    /**
+     * <pre>
+     * Represents a boolean value.
+     * </pre>
+     *
+     * <code>bool bool_value = 14;</code>
+     */
+    public boolean getBoolValue() {
+      if (kindCase_ == 14) {
+        return (java.lang.Boolean) kind_;
+      }
+      return false;
+    }
+    /**
+     * <pre>
+     * Represents a boolean value.
+     * </pre>
+     *
+     * <code>bool bool_value = 14;</code>
+     */
+    public Builder setBoolValue(boolean value) {
+      kindCase_ = 14;
+      kind_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Represents a boolean value.
+     * </pre>
+     *
+     * <code>bool bool_value = 14;</code>
+     */
+    public Builder clearBoolValue() {
+      if (kindCase_ == 14) {
+        kindCase_ = 0;
+        kind_ = null;
+        onChanged();
+      }
+      return this;
+    }
+
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.TensorShapeProto, org.tensorflow.proto.framework.TensorShapeProto.Builder, org.tensorflow.proto.framework.TensorShapeProtoOrBuilder> tensorShapeValueBuilder_;
+    /**
+     * <pre>
+     * Represents a TensorShape.
+     * </pre>
+     *
+     * <code>.tensorflow.TensorShapeProto tensor_shape_value = 31;</code>
+     */
+    public boolean hasTensorShapeValue() {
+      return kindCase_ == 31;
+    }
+    /**
+     * <pre>
+     * Represents a TensorShape.
+     * </pre>
+     *
+     * <code>.tensorflow.TensorShapeProto tensor_shape_value = 31;</code>
+     */
+    public org.tensorflow.proto.framework.TensorShapeProto getTensorShapeValue() {
+      if (tensorShapeValueBuilder_ == null) {
+        if (kindCase_ == 31) {
+          return (org.tensorflow.proto.framework.TensorShapeProto) kind_;
+        }
+        return org.tensorflow.proto.framework.TensorShapeProto.getDefaultInstance();
+      } else {
+        if (kindCase_ == 31) {
+          return tensorShapeValueBuilder_.getMessage();
+        }
+        return org.tensorflow.proto.framework.TensorShapeProto.getDefaultInstance();
+      }
+    }
+    /**
+     * <pre>
+     * Represents a TensorShape.
+     * </pre>
+     *
+     * <code>.tensorflow.TensorShapeProto tensor_shape_value = 31;</code>
+     */
+    public Builder setTensorShapeValue(org.tensorflow.proto.framework.TensorShapeProto value) {
+      if (tensorShapeValueBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        kind_ = value;
+        onChanged();
+      } else {
+        tensorShapeValueBuilder_.setMessage(value);
+      }
+      kindCase_ = 31;
+      return this;
+    }
+    /**
+     * <pre>
+     * Represents a TensorShape.
+     * </pre>
+     *
+     * <code>.tensorflow.TensorShapeProto tensor_shape_value = 31;</code>
+     */
+    public Builder setTensorShapeValue(
+        org.tensorflow.proto.framework.TensorShapeProto.Builder builderForValue) {
+      if (tensorShapeValueBuilder_ == null) {
+        kind_ = builderForValue.build();
+        onChanged();
+      } else {
+        tensorShapeValueBuilder_.setMessage(builderForValue.build());
+      }
+      kindCase_ = 31;
+      return this;
+    }
+    /**
+     * <pre>
+     * Represents a TensorShape.
+     * </pre>
+     *
+     * <code>.tensorflow.TensorShapeProto tensor_shape_value = 31;</code>
+     */
+    public Builder mergeTensorShapeValue(org.tensorflow.proto.framework.TensorShapeProto value) {
+      if (tensorShapeValueBuilder_ == null) {
+        if (kindCase_ == 31 &&
+            kind_ != org.tensorflow.proto.framework.TensorShapeProto.getDefaultInstance()) {
+          kind_ = org.tensorflow.proto.framework.TensorShapeProto.newBuilder((org.tensorflow.proto.framework.TensorShapeProto) kind_)
+              .mergeFrom(value).buildPartial();
+        } else {
+          kind_ = value;
+        }
+        onChanged();
+      } else {
+        if (kindCase_ == 31) {
+          tensorShapeValueBuilder_.mergeFrom(value);
+        }
+        tensorShapeValueBuilder_.setMessage(value);
+      }
+      kindCase_ = 31;
+      return this;
+    }
+    /**
+     * <pre>
+     * Represents a TensorShape.
+     * </pre>
+     *
+     * <code>.tensorflow.TensorShapeProto tensor_shape_value = 31;</code>
+     */
+    public Builder clearTensorShapeValue() {
+      if (tensorShapeValueBuilder_ == null) {
+        if (kindCase_ == 31) {
+          kindCase_ = 0;
+          kind_ = null;
+          onChanged();
+        }
+      } else {
+        if (kindCase_ == 31) {
+          kindCase_ = 0;
+          kind_ = null;
+        }
+        tensorShapeValueBuilder_.clear();
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Represents a TensorShape.
+     * </pre>
+     *
+     * <code>.tensorflow.TensorShapeProto tensor_shape_value = 31;</code>
+     */
+    public org.tensorflow.proto.framework.TensorShapeProto.Builder getTensorShapeValueBuilder() {
+      return getTensorShapeValueFieldBuilder().getBuilder();
+    }
+    /**
+     * <pre>
+     * Represents a TensorShape.
+     * </pre>
+     *
+     * <code>.tensorflow.TensorShapeProto tensor_shape_value = 31;</code>
+     */
+    public org.tensorflow.proto.framework.TensorShapeProtoOrBuilder getTensorShapeValueOrBuilder() {
+      if ((kindCase_ == 31) && (tensorShapeValueBuilder_ != null)) {
+        return tensorShapeValueBuilder_.getMessageOrBuilder();
+      } else {
+        if (kindCase_ == 31) {
+          return (org.tensorflow.proto.framework.TensorShapeProto) kind_;
+        }
+        return org.tensorflow.proto.framework.TensorShapeProto.getDefaultInstance();
+      }
+    }
+    /**
+     * <pre>
+     * Represents a TensorShape.
+     * </pre>
+     *
+     * <code>.tensorflow.TensorShapeProto tensor_shape_value = 31;</code>
+     */
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.TensorShapeProto, org.tensorflow.proto.framework.TensorShapeProto.Builder, org.tensorflow.proto.framework.TensorShapeProtoOrBuilder> 
+        getTensorShapeValueFieldBuilder() {
+      if (tensorShapeValueBuilder_ == null) {
+        if (!(kindCase_ == 31)) {
+          kind_ = org.tensorflow.proto.framework.TensorShapeProto.getDefaultInstance();
+        }
+        tensorShapeValueBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+            org.tensorflow.proto.framework.TensorShapeProto, org.tensorflow.proto.framework.TensorShapeProto.Builder, org.tensorflow.proto.framework.TensorShapeProtoOrBuilder>(
+                (org.tensorflow.proto.framework.TensorShapeProto) kind_,
+                getParentForChildren(),
+                isClean());
+        kind_ = null;
+      }
+      kindCase_ = 31;
+      onChanged();;
+      return tensorShapeValueBuilder_;
+    }
+
+    /**
+     * <pre>
+     * Represents an enum value for dtype.
+     * </pre>
+     *
+     * <code>.tensorflow.DataType tensor_dtype_value = 32;</code>
+     */
+    public int getTensorDtypeValueValue() {
+      if (kindCase_ == 32) {
+        return ((java.lang.Integer) kind_).intValue();
+      }
+      return 0;
+    }
+    /**
+     * <pre>
+     * Represents an enum value for dtype.
+     * </pre>
+     *
+     * <code>.tensorflow.DataType tensor_dtype_value = 32;</code>
+     */
+    public Builder setTensorDtypeValueValue(int value) {
+      kindCase_ = 32;
+      kind_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Represents an enum value for dtype.
+     * </pre>
+     *
+     * <code>.tensorflow.DataType tensor_dtype_value = 32;</code>
+     */
+    public org.tensorflow.proto.framework.DataType getTensorDtypeValue() {
+      if (kindCase_ == 32) {
+        @SuppressWarnings("deprecation")
+        org.tensorflow.proto.framework.DataType result = org.tensorflow.proto.framework.DataType.valueOf(
+            (java.lang.Integer) kind_);
+        return result == null ? org.tensorflow.proto.framework.DataType.UNRECOGNIZED : result;
+      }
+      return org.tensorflow.proto.framework.DataType.DT_INVALID;
+    }
+    /**
+     * <pre>
+     * Represents an enum value for dtype.
+     * </pre>
+     *
+     * <code>.tensorflow.DataType tensor_dtype_value = 32;</code>
+     */
+    public Builder setTensorDtypeValue(org.tensorflow.proto.framework.DataType value) {
+      if (value == null) {
+        throw new NullPointerException();
+      }
+      kindCase_ = 32;
+      kind_ = value.getNumber();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Represents an enum value for dtype.
+     * </pre>
+     *
+     * <code>.tensorflow.DataType tensor_dtype_value = 32;</code>
+     */
+    public Builder clearTensorDtypeValue() {
+      if (kindCase_ == 32) {
+        kindCase_ = 0;
+        kind_ = null;
+        onChanged();
+      }
+      return this;
+    }
+
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.TensorSpecProto, org.tensorflow.proto.framework.TensorSpecProto.Builder, org.tensorflow.proto.framework.TensorSpecProtoOrBuilder> tensorSpecValueBuilder_;
+    /**
+     * <pre>
+     * Represents a value for tf.TensorSpec.
+     * </pre>
+     *
+     * <code>.tensorflow.TensorSpecProto tensor_spec_value = 33;</code>
+     */
+    public boolean hasTensorSpecValue() {
+      return kindCase_ == 33;
+    }
+    /**
+     * <pre>
+     * Represents a value for tf.TensorSpec.
+     * </pre>
+     *
+     * <code>.tensorflow.TensorSpecProto tensor_spec_value = 33;</code>
+     */
+    public org.tensorflow.proto.framework.TensorSpecProto getTensorSpecValue() {
+      if (tensorSpecValueBuilder_ == null) {
+        if (kindCase_ == 33) {
+          return (org.tensorflow.proto.framework.TensorSpecProto) kind_;
+        }
+        return org.tensorflow.proto.framework.TensorSpecProto.getDefaultInstance();
+      } else {
+        if (kindCase_ == 33) {
+          return tensorSpecValueBuilder_.getMessage();
+        }
+        return org.tensorflow.proto.framework.TensorSpecProto.getDefaultInstance();
+      }
+    }
+    /**
+     * <pre>
+     * Represents a value for tf.TensorSpec.
+     * </pre>
+     *
+     * <code>.tensorflow.TensorSpecProto tensor_spec_value = 33;</code>
+     */
+    public Builder setTensorSpecValue(org.tensorflow.proto.framework.TensorSpecProto value) {
+      if (tensorSpecValueBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        kind_ = value;
+        onChanged();
+      } else {
+        tensorSpecValueBuilder_.setMessage(value);
+      }
+      kindCase_ = 33;
+      return this;
+    }
+    /**
+     * <pre>
+     * Represents a value for tf.TensorSpec.
+     * </pre>
+     *
+     * <code>.tensorflow.TensorSpecProto tensor_spec_value = 33;</code>
+     */
+    public Builder setTensorSpecValue(
+        org.tensorflow.proto.framework.TensorSpecProto.Builder builderForValue) {
+      if (tensorSpecValueBuilder_ == null) {
+        kind_ = builderForValue.build();
+        onChanged();
+      } else {
+        tensorSpecValueBuilder_.setMessage(builderForValue.build());
+      }
+      kindCase_ = 33;
+      return this;
+    }
+    /**
+     * <pre>
+     * Represents a value for tf.TensorSpec.
+     * </pre>
+     *
+     * <code>.tensorflow.TensorSpecProto tensor_spec_value = 33;</code>
+     */
+    public Builder mergeTensorSpecValue(org.tensorflow.proto.framework.TensorSpecProto value) {
+      if (tensorSpecValueBuilder_ == null) {
+        if (kindCase_ == 33 &&
+            kind_ != org.tensorflow.proto.framework.TensorSpecProto.getDefaultInstance()) {
+          kind_ = org.tensorflow.proto.framework.TensorSpecProto.newBuilder((org.tensorflow.proto.framework.TensorSpecProto) kind_)
+              .mergeFrom(value).buildPartial();
+        } else {
+          kind_ = value;
+        }
+        onChanged();
+      } else {
+        if (kindCase_ == 33) {
+          tensorSpecValueBuilder_.mergeFrom(value);
+        }
+        tensorSpecValueBuilder_.setMessage(value);
+      }
+      kindCase_ = 33;
+      return this;
+    }
+    /**
+     * <pre>
+     * Represents a value for tf.TensorSpec.
+     * </pre>
+     *
+     * <code>.tensorflow.TensorSpecProto tensor_spec_value = 33;</code>
+     */
+    public Builder clearTensorSpecValue() {
+      if (tensorSpecValueBuilder_ == null) {
+        if (kindCase_ == 33) {
+          kindCase_ = 0;
+          kind_ = null;
+          onChanged();
+        }
+      } else {
+        if (kindCase_ == 33) {
+          kindCase_ = 0;
+          kind_ = null;
+        }
+        tensorSpecValueBuilder_.clear();
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Represents a value for tf.TensorSpec.
+     * </pre>
+     *
+     * <code>.tensorflow.TensorSpecProto tensor_spec_value = 33;</code>
+     */
+    public org.tensorflow.proto.framework.TensorSpecProto.Builder getTensorSpecValueBuilder() {
+      return getTensorSpecValueFieldBuilder().getBuilder();
+    }
+    /**
+     * <pre>
+     * Represents a value for tf.TensorSpec.
+     * </pre>
+     *
+     * <code>.tensorflow.TensorSpecProto tensor_spec_value = 33;</code>
+     */
+    public org.tensorflow.proto.framework.TensorSpecProtoOrBuilder getTensorSpecValueOrBuilder() {
+      if ((kindCase_ == 33) && (tensorSpecValueBuilder_ != null)) {
+        return tensorSpecValueBuilder_.getMessageOrBuilder();
+      } else {
+        if (kindCase_ == 33) {
+          return (org.tensorflow.proto.framework.TensorSpecProto) kind_;
+        }
+        return org.tensorflow.proto.framework.TensorSpecProto.getDefaultInstance();
+      }
+    }
+    /**
+     * <pre>
+     * Represents a value for tf.TensorSpec.
+     * </pre>
+     *
+     * <code>.tensorflow.TensorSpecProto tensor_spec_value = 33;</code>
+     */
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.TensorSpecProto, org.tensorflow.proto.framework.TensorSpecProto.Builder, org.tensorflow.proto.framework.TensorSpecProtoOrBuilder> 
+        getTensorSpecValueFieldBuilder() {
+      if (tensorSpecValueBuilder_ == null) {
+        if (!(kindCase_ == 33)) {
+          kind_ = org.tensorflow.proto.framework.TensorSpecProto.getDefaultInstance();
+        }
+        tensorSpecValueBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+            org.tensorflow.proto.framework.TensorSpecProto, org.tensorflow.proto.framework.TensorSpecProto.Builder, org.tensorflow.proto.framework.TensorSpecProtoOrBuilder>(
+                (org.tensorflow.proto.framework.TensorSpecProto) kind_,
+                getParentForChildren(),
+                isClean());
+        kind_ = null;
+      }
+      kindCase_ = 33;
+      onChanged();;
+      return tensorSpecValueBuilder_;
+    }
+
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.TypeSpecProto, org.tensorflow.proto.framework.TypeSpecProto.Builder, org.tensorflow.proto.framework.TypeSpecProtoOrBuilder> typeSpecValueBuilder_;
+    /**
+     * <pre>
+     * Represents a value for tf.TypeSpec.
+     * </pre>
+     *
+     * <code>.tensorflow.TypeSpecProto type_spec_value = 34;</code>
+     */
+    public boolean hasTypeSpecValue() {
+      return kindCase_ == 34;
+    }
+    /**
+     * <pre>
+     * Represents a value for tf.TypeSpec.
+     * </pre>
+     *
+     * <code>.tensorflow.TypeSpecProto type_spec_value = 34;</code>
+     */
+    public org.tensorflow.proto.framework.TypeSpecProto getTypeSpecValue() {
+      if (typeSpecValueBuilder_ == null) {
+        if (kindCase_ == 34) {
+          return (org.tensorflow.proto.framework.TypeSpecProto) kind_;
+        }
+        return org.tensorflow.proto.framework.TypeSpecProto.getDefaultInstance();
+      } else {
+        if (kindCase_ == 34) {
+          return typeSpecValueBuilder_.getMessage();
+        }
+        return org.tensorflow.proto.framework.TypeSpecProto.getDefaultInstance();
+      }
+    }
+    /**
+     * <pre>
+     * Represents a value for tf.TypeSpec.
+     * </pre>
+     *
+     * <code>.tensorflow.TypeSpecProto type_spec_value = 34;</code>
+     */
+    public Builder setTypeSpecValue(org.tensorflow.proto.framework.TypeSpecProto value) {
+      if (typeSpecValueBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        kind_ = value;
+        onChanged();
+      } else {
+        typeSpecValueBuilder_.setMessage(value);
+      }
+      kindCase_ = 34;
+      return this;
+    }
+    /**
+     * <pre>
+     * Represents a value for tf.TypeSpec.
+     * </pre>
+     *
+     * <code>.tensorflow.TypeSpecProto type_spec_value = 34;</code>
+     */
+    public Builder setTypeSpecValue(
+        org.tensorflow.proto.framework.TypeSpecProto.Builder builderForValue) {
+      if (typeSpecValueBuilder_ == null) {
+        kind_ = builderForValue.build();
+        onChanged();
+      } else {
+        typeSpecValueBuilder_.setMessage(builderForValue.build());
+      }
+      kindCase_ = 34;
+      return this;
+    }
+    /**
+     * <pre>
+     * Represents a value for tf.TypeSpec.
+     * </pre>
+     *
+     * <code>.tensorflow.TypeSpecProto type_spec_value = 34;</code>
+     */
+    public Builder mergeTypeSpecValue(org.tensorflow.proto.framework.TypeSpecProto value) {
+      if (typeSpecValueBuilder_ == null) {
+        if (kindCase_ == 34 &&
+            kind_ != org.tensorflow.proto.framework.TypeSpecProto.getDefaultInstance()) {
+          kind_ = org.tensorflow.proto.framework.TypeSpecProto.newBuilder((org.tensorflow.proto.framework.TypeSpecProto) kind_)
+              .mergeFrom(value).buildPartial();
+        } else {
+          kind_ = value;
+        }
+        onChanged();
+      } else {
+        if (kindCase_ == 34) {
+          typeSpecValueBuilder_.mergeFrom(value);
+        }
+        typeSpecValueBuilder_.setMessage(value);
+      }
+      kindCase_ = 34;
+      return this;
+    }
+    /**
+     * <pre>
+     * Represents a value for tf.TypeSpec.
+     * </pre>
+     *
+     * <code>.tensorflow.TypeSpecProto type_spec_value = 34;</code>
+     */
+    public Builder clearTypeSpecValue() {
+      if (typeSpecValueBuilder_ == null) {
+        if (kindCase_ == 34) {
+          kindCase_ = 0;
+          kind_ = null;
+          onChanged();
+        }
+      } else {
+        if (kindCase_ == 34) {
+          kindCase_ = 0;
+          kind_ = null;
+        }
+        typeSpecValueBuilder_.clear();
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Represents a value for tf.TypeSpec.
+     * </pre>
+     *
+     * <code>.tensorflow.TypeSpecProto type_spec_value = 34;</code>
+     */
+    public org.tensorflow.proto.framework.TypeSpecProto.Builder getTypeSpecValueBuilder() {
+      return getTypeSpecValueFieldBuilder().getBuilder();
+    }
+    /**
+     * <pre>
+     * Represents a value for tf.TypeSpec.
+     * </pre>
+     *
+     * <code>.tensorflow.TypeSpecProto type_spec_value = 34;</code>
+     */
+    public org.tensorflow.proto.framework.TypeSpecProtoOrBuilder getTypeSpecValueOrBuilder() {
+      if ((kindCase_ == 34) && (typeSpecValueBuilder_ != null)) {
+        return typeSpecValueBuilder_.getMessageOrBuilder();
+      } else {
+        if (kindCase_ == 34) {
+          return (org.tensorflow.proto.framework.TypeSpecProto) kind_;
+        }
+        return org.tensorflow.proto.framework.TypeSpecProto.getDefaultInstance();
+      }
+    }
+    /**
+     * <pre>
+     * Represents a value for tf.TypeSpec.
+     * </pre>
+     *
+     * <code>.tensorflow.TypeSpecProto type_spec_value = 34;</code>
+     */
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.TypeSpecProto, org.tensorflow.proto.framework.TypeSpecProto.Builder, org.tensorflow.proto.framework.TypeSpecProtoOrBuilder> 
+        getTypeSpecValueFieldBuilder() {
+      if (typeSpecValueBuilder_ == null) {
+        if (!(kindCase_ == 34)) {
+          kind_ = org.tensorflow.proto.framework.TypeSpecProto.getDefaultInstance();
+        }
+        typeSpecValueBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+            org.tensorflow.proto.framework.TypeSpecProto, org.tensorflow.proto.framework.TypeSpecProto.Builder, org.tensorflow.proto.framework.TypeSpecProtoOrBuilder>(
+                (org.tensorflow.proto.framework.TypeSpecProto) kind_,
+                getParentForChildren(),
+                isClean());
+        kind_ = null;
+      }
+      kindCase_ = 34;
+      onChanged();;
+      return typeSpecValueBuilder_;
+    }
+
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.BoundedTensorSpecProto, org.tensorflow.proto.framework.BoundedTensorSpecProto.Builder, org.tensorflow.proto.framework.BoundedTensorSpecProtoOrBuilder> boundedTensorSpecValueBuilder_;
+    /**
+     * <pre>
+     * Represents a value for tf.BoundedTensorSpec.
+     * </pre>
+     *
+     * <code>.tensorflow.BoundedTensorSpecProto bounded_tensor_spec_value = 35;</code>
+     */
+    public boolean hasBoundedTensorSpecValue() {
+      return kindCase_ == 35;
+    }
+    /**
+     * <pre>
+     * Represents a value for tf.BoundedTensorSpec.
+     * </pre>
+     *
+     * <code>.tensorflow.BoundedTensorSpecProto bounded_tensor_spec_value = 35;</code>
+     */
+    public org.tensorflow.proto.framework.BoundedTensorSpecProto getBoundedTensorSpecValue() {
+      if (boundedTensorSpecValueBuilder_ == null) {
+        if (kindCase_ == 35) {
+          return (org.tensorflow.proto.framework.BoundedTensorSpecProto) kind_;
+        }
+        return org.tensorflow.proto.framework.BoundedTensorSpecProto.getDefaultInstance();
+      } else {
+        if (kindCase_ == 35) {
+          return boundedTensorSpecValueBuilder_.getMessage();
+        }
+        return org.tensorflow.proto.framework.BoundedTensorSpecProto.getDefaultInstance();
+      }
+    }
+    /**
+     * <pre>
+     * Represents a value for tf.BoundedTensorSpec.
+     * </pre>
+     *
+     * <code>.tensorflow.BoundedTensorSpecProto bounded_tensor_spec_value = 35;</code>
+     */
+    public Builder setBoundedTensorSpecValue(org.tensorflow.proto.framework.BoundedTensorSpecProto value) {
+      if (boundedTensorSpecValueBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        kind_ = value;
+        onChanged();
+      } else {
+        boundedTensorSpecValueBuilder_.setMessage(value);
+      }
+      kindCase_ = 35;
+      return this;
+    }
+    /**
+     * <pre>
+     * Represents a value for tf.BoundedTensorSpec.
+     * </pre>
+     *
+     * <code>.tensorflow.BoundedTensorSpecProto bounded_tensor_spec_value = 35;</code>
+     */
+    public Builder setBoundedTensorSpecValue(
+        org.tensorflow.proto.framework.BoundedTensorSpecProto.Builder builderForValue) {
+      if (boundedTensorSpecValueBuilder_ == null) {
+        kind_ = builderForValue.build();
+        onChanged();
+      } else {
+        boundedTensorSpecValueBuilder_.setMessage(builderForValue.build());
+      }
+      kindCase_ = 35;
+      return this;
+    }
+    /**
+     * <pre>
+     * Represents a value for tf.BoundedTensorSpec.
+     * </pre>
+     *
+     * <code>.tensorflow.BoundedTensorSpecProto bounded_tensor_spec_value = 35;</code>
+     */
+    public Builder mergeBoundedTensorSpecValue(org.tensorflow.proto.framework.BoundedTensorSpecProto value) {
+      if (boundedTensorSpecValueBuilder_ == null) {
+        if (kindCase_ == 35 &&
+            kind_ != org.tensorflow.proto.framework.BoundedTensorSpecProto.getDefaultInstance()) {
+          kind_ = org.tensorflow.proto.framework.BoundedTensorSpecProto.newBuilder((org.tensorflow.proto.framework.BoundedTensorSpecProto) kind_)
+              .mergeFrom(value).buildPartial();
+        } else {
+          kind_ = value;
+        }
+        onChanged();
+      } else {
+        if (kindCase_ == 35) {
+          boundedTensorSpecValueBuilder_.mergeFrom(value);
+        }
+        boundedTensorSpecValueBuilder_.setMessage(value);
+      }
+      kindCase_ = 35;
+      return this;
+    }
+    /**
+     * <pre>
+     * Represents a value for tf.BoundedTensorSpec.
+     * </pre>
+     *
+     * <code>.tensorflow.BoundedTensorSpecProto bounded_tensor_spec_value = 35;</code>
+     */
+    public Builder clearBoundedTensorSpecValue() {
+      if (boundedTensorSpecValueBuilder_ == null) {
+        if (kindCase_ == 35) {
+          kindCase_ = 0;
+          kind_ = null;
+          onChanged();
+        }
+      } else {
+        if (kindCase_ == 35) {
+          kindCase_ = 0;
+          kind_ = null;
+        }
+        boundedTensorSpecValueBuilder_.clear();
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Represents a value for tf.BoundedTensorSpec.
+     * </pre>
+     *
+     * <code>.tensorflow.BoundedTensorSpecProto bounded_tensor_spec_value = 35;</code>
+     */
+    public org.tensorflow.proto.framework.BoundedTensorSpecProto.Builder getBoundedTensorSpecValueBuilder() {
+      return getBoundedTensorSpecValueFieldBuilder().getBuilder();
+    }
+    /**
+     * <pre>
+     * Represents a value for tf.BoundedTensorSpec.
+     * </pre>
+     *
+     * <code>.tensorflow.BoundedTensorSpecProto bounded_tensor_spec_value = 35;</code>
+     */
+    public org.tensorflow.proto.framework.BoundedTensorSpecProtoOrBuilder getBoundedTensorSpecValueOrBuilder() {
+      if ((kindCase_ == 35) && (boundedTensorSpecValueBuilder_ != null)) {
+        return boundedTensorSpecValueBuilder_.getMessageOrBuilder();
+      } else {
+        if (kindCase_ == 35) {
+          return (org.tensorflow.proto.framework.BoundedTensorSpecProto) kind_;
+        }
+        return org.tensorflow.proto.framework.BoundedTensorSpecProto.getDefaultInstance();
+      }
+    }
+    /**
+     * <pre>
+     * Represents a value for tf.BoundedTensorSpec.
+     * </pre>
+     *
+     * <code>.tensorflow.BoundedTensorSpecProto bounded_tensor_spec_value = 35;</code>
+     */
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.BoundedTensorSpecProto, org.tensorflow.proto.framework.BoundedTensorSpecProto.Builder, org.tensorflow.proto.framework.BoundedTensorSpecProtoOrBuilder> 
+        getBoundedTensorSpecValueFieldBuilder() {
+      if (boundedTensorSpecValueBuilder_ == null) {
+        if (!(kindCase_ == 35)) {
+          kind_ = org.tensorflow.proto.framework.BoundedTensorSpecProto.getDefaultInstance();
+        }
+        boundedTensorSpecValueBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+            org.tensorflow.proto.framework.BoundedTensorSpecProto, org.tensorflow.proto.framework.BoundedTensorSpecProto.Builder, org.tensorflow.proto.framework.BoundedTensorSpecProtoOrBuilder>(
+                (org.tensorflow.proto.framework.BoundedTensorSpecProto) kind_,
+                getParentForChildren(),
+                isClean());
+        kind_ = null;
+      }
+      kindCase_ = 35;
+      onChanged();;
+      return boundedTensorSpecValueBuilder_;
+    }
+
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.ListValue, org.tensorflow.proto.framework.ListValue.Builder, org.tensorflow.proto.framework.ListValueOrBuilder> listValueBuilder_;
+    /**
+     * <pre>
+     * Represents a list of `Value`.
+     * </pre>
+     *
+     * <code>.tensorflow.ListValue list_value = 51;</code>
+     */
+    public boolean hasListValue() {
+      return kindCase_ == 51;
+    }
+    /**
+     * <pre>
+     * Represents a list of `Value`.
+     * </pre>
+     *
+     * <code>.tensorflow.ListValue list_value = 51;</code>
+     */
+    public org.tensorflow.proto.framework.ListValue getListValue() {
+      if (listValueBuilder_ == null) {
+        if (kindCase_ == 51) {
+          return (org.tensorflow.proto.framework.ListValue) kind_;
+        }
+        return org.tensorflow.proto.framework.ListValue.getDefaultInstance();
+      } else {
+        if (kindCase_ == 51) {
+          return listValueBuilder_.getMessage();
+        }
+        return org.tensorflow.proto.framework.ListValue.getDefaultInstance();
+      }
+    }
+    /**
+     * <pre>
+     * Represents a list of `Value`.
+     * </pre>
+     *
+     * <code>.tensorflow.ListValue list_value = 51;</code>
+     */
+    public Builder setListValue(org.tensorflow.proto.framework.ListValue value) {
+      if (listValueBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        kind_ = value;
+        onChanged();
+      } else {
+        listValueBuilder_.setMessage(value);
+      }
+      kindCase_ = 51;
+      return this;
+    }
+    /**
+     * <pre>
+     * Represents a list of `Value`.
+     * </pre>
+     *
+     * <code>.tensorflow.ListValue list_value = 51;</code>
+     */
+    public Builder setListValue(
+        org.tensorflow.proto.framework.ListValue.Builder builderForValue) {
+      if (listValueBuilder_ == null) {
+        kind_ = builderForValue.build();
+        onChanged();
+      } else {
+        listValueBuilder_.setMessage(builderForValue.build());
+      }
+      kindCase_ = 51;
+      return this;
+    }
+    /**
+     * <pre>
+     * Represents a list of `Value`.
+     * </pre>
+     *
+     * <code>.tensorflow.ListValue list_value = 51;</code>
+     */
+    public Builder mergeListValue(org.tensorflow.proto.framework.ListValue value) {
+      if (listValueBuilder_ == null) {
+        if (kindCase_ == 51 &&
+            kind_ != org.tensorflow.proto.framework.ListValue.getDefaultInstance()) {
+          kind_ = org.tensorflow.proto.framework.ListValue.newBuilder((org.tensorflow.proto.framework.ListValue) kind_)
+              .mergeFrom(value).buildPartial();
+        } else {
+          kind_ = value;
+        }
+        onChanged();
+      } else {
+        if (kindCase_ == 51) {
+          listValueBuilder_.mergeFrom(value);
+        }
+        listValueBuilder_.setMessage(value);
+      }
+      kindCase_ = 51;
+      return this;
+    }
+    /**
+     * <pre>
+     * Represents a list of `Value`.
+     * </pre>
+     *
+     * <code>.tensorflow.ListValue list_value = 51;</code>
+     */
+    public Builder clearListValue() {
+      if (listValueBuilder_ == null) {
+        if (kindCase_ == 51) {
+          kindCase_ = 0;
+          kind_ = null;
+          onChanged();
+        }
+      } else {
+        if (kindCase_ == 51) {
+          kindCase_ = 0;
+          kind_ = null;
+        }
+        listValueBuilder_.clear();
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Represents a list of `Value`.
+     * </pre>
+     *
+     * <code>.tensorflow.ListValue list_value = 51;</code>
+     */
+    public org.tensorflow.proto.framework.ListValue.Builder getListValueBuilder() {
+      return getListValueFieldBuilder().getBuilder();
+    }
+    /**
+     * <pre>
+     * Represents a list of `Value`.
+     * </pre>
+     *
+     * <code>.tensorflow.ListValue list_value = 51;</code>
+     */
+    public org.tensorflow.proto.framework.ListValueOrBuilder getListValueOrBuilder() {
+      if ((kindCase_ == 51) && (listValueBuilder_ != null)) {
+        return listValueBuilder_.getMessageOrBuilder();
+      } else {
+        if (kindCase_ == 51) {
+          return (org.tensorflow.proto.framework.ListValue) kind_;
+        }
+        return org.tensorflow.proto.framework.ListValue.getDefaultInstance();
+      }
+    }
+    /**
+     * <pre>
+     * Represents a list of `Value`.
+     * </pre>
+     *
+     * <code>.tensorflow.ListValue list_value = 51;</code>
+     */
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.ListValue, org.tensorflow.proto.framework.ListValue.Builder, org.tensorflow.proto.framework.ListValueOrBuilder> 
+        getListValueFieldBuilder() {
+      if (listValueBuilder_ == null) {
+        if (!(kindCase_ == 51)) {
+          kind_ = org.tensorflow.proto.framework.ListValue.getDefaultInstance();
+        }
+        listValueBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+            org.tensorflow.proto.framework.ListValue, org.tensorflow.proto.framework.ListValue.Builder, org.tensorflow.proto.framework.ListValueOrBuilder>(
+                (org.tensorflow.proto.framework.ListValue) kind_,
+                getParentForChildren(),
+                isClean());
+        kind_ = null;
+      }
+      kindCase_ = 51;
+      onChanged();;
+      return listValueBuilder_;
+    }
+
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.TupleValue, org.tensorflow.proto.framework.TupleValue.Builder, org.tensorflow.proto.framework.TupleValueOrBuilder> tupleValueBuilder_;
+    /**
+     * <pre>
+     * Represents a tuple of `Value`.
+     * </pre>
+     *
+     * <code>.tensorflow.TupleValue tuple_value = 52;</code>
+     */
+    public boolean hasTupleValue() {
+      return kindCase_ == 52;
+    }
+    /**
+     * <pre>
+     * Represents a tuple of `Value`.
+     * </pre>
+     *
+     * <code>.tensorflow.TupleValue tuple_value = 52;</code>
+     */
+    public org.tensorflow.proto.framework.TupleValue getTupleValue() {
+      if (tupleValueBuilder_ == null) {
+        if (kindCase_ == 52) {
+          return (org.tensorflow.proto.framework.TupleValue) kind_;
+        }
+        return org.tensorflow.proto.framework.TupleValue.getDefaultInstance();
+      } else {
+        if (kindCase_ == 52) {
+          return tupleValueBuilder_.getMessage();
+        }
+        return org.tensorflow.proto.framework.TupleValue.getDefaultInstance();
+      }
+    }
+    /**
+     * <pre>
+     * Represents a tuple of `Value`.
+     * </pre>
+     *
+     * <code>.tensorflow.TupleValue tuple_value = 52;</code>
+     */
+    public Builder setTupleValue(org.tensorflow.proto.framework.TupleValue value) {
+      if (tupleValueBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        kind_ = value;
+        onChanged();
+      } else {
+        tupleValueBuilder_.setMessage(value);
+      }
+      kindCase_ = 52;
+      return this;
+    }
+    /**
+     * <pre>
+     * Represents a tuple of `Value`.
+     * </pre>
+     *
+     * <code>.tensorflow.TupleValue tuple_value = 52;</code>
+     */
+    public Builder setTupleValue(
+        org.tensorflow.proto.framework.TupleValue.Builder builderForValue) {
+      if (tupleValueBuilder_ == null) {
+        kind_ = builderForValue.build();
+        onChanged();
+      } else {
+        tupleValueBuilder_.setMessage(builderForValue.build());
+      }
+      kindCase_ = 52;
+      return this;
+    }
+    /**
+     * <pre>
+     * Represents a tuple of `Value`.
+     * </pre>
+     *
+     * <code>.tensorflow.TupleValue tuple_value = 52;</code>
+     */
+    public Builder mergeTupleValue(org.tensorflow.proto.framework.TupleValue value) {
+      if (tupleValueBuilder_ == null) {
+        if (kindCase_ == 52 &&
+            kind_ != org.tensorflow.proto.framework.TupleValue.getDefaultInstance()) {
+          kind_ = org.tensorflow.proto.framework.TupleValue.newBuilder((org.tensorflow.proto.framework.TupleValue) kind_)
+              .mergeFrom(value).buildPartial();
+        } else {
+          kind_ = value;
+        }
+        onChanged();
+      } else {
+        if (kindCase_ == 52) {
+          tupleValueBuilder_.mergeFrom(value);
+        }
+        tupleValueBuilder_.setMessage(value);
+      }
+      kindCase_ = 52;
+      return this;
+    }
+    /**
+     * <pre>
+     * Represents a tuple of `Value`.
+     * </pre>
+     *
+     * <code>.tensorflow.TupleValue tuple_value = 52;</code>
+     */
+    public Builder clearTupleValue() {
+      if (tupleValueBuilder_ == null) {
+        if (kindCase_ == 52) {
+          kindCase_ = 0;
+          kind_ = null;
+          onChanged();
+        }
+      } else {
+        if (kindCase_ == 52) {
+          kindCase_ = 0;
+          kind_ = null;
+        }
+        tupleValueBuilder_.clear();
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Represents a tuple of `Value`.
+     * </pre>
+     *
+     * <code>.tensorflow.TupleValue tuple_value = 52;</code>
+     */
+    public org.tensorflow.proto.framework.TupleValue.Builder getTupleValueBuilder() {
+      return getTupleValueFieldBuilder().getBuilder();
+    }
+    /**
+     * <pre>
+     * Represents a tuple of `Value`.
+     * </pre>
+     *
+     * <code>.tensorflow.TupleValue tuple_value = 52;</code>
+     */
+    public org.tensorflow.proto.framework.TupleValueOrBuilder getTupleValueOrBuilder() {
+      if ((kindCase_ == 52) && (tupleValueBuilder_ != null)) {
+        return tupleValueBuilder_.getMessageOrBuilder();
+      } else {
+        if (kindCase_ == 52) {
+          return (org.tensorflow.proto.framework.TupleValue) kind_;
+        }
+        return org.tensorflow.proto.framework.TupleValue.getDefaultInstance();
+      }
+    }
+    /**
+     * <pre>
+     * Represents a tuple of `Value`.
+     * </pre>
+     *
+     * <code>.tensorflow.TupleValue tuple_value = 52;</code>
+     */
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.TupleValue, org.tensorflow.proto.framework.TupleValue.Builder, org.tensorflow.proto.framework.TupleValueOrBuilder> 
+        getTupleValueFieldBuilder() {
+      if (tupleValueBuilder_ == null) {
+        if (!(kindCase_ == 52)) {
+          kind_ = org.tensorflow.proto.framework.TupleValue.getDefaultInstance();
+        }
+        tupleValueBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+            org.tensorflow.proto.framework.TupleValue, org.tensorflow.proto.framework.TupleValue.Builder, org.tensorflow.proto.framework.TupleValueOrBuilder>(
+                (org.tensorflow.proto.framework.TupleValue) kind_,
+                getParentForChildren(),
+                isClean());
+        kind_ = null;
+      }
+      kindCase_ = 52;
+      onChanged();;
+      return tupleValueBuilder_;
+    }
+
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.DictValue, org.tensorflow.proto.framework.DictValue.Builder, org.tensorflow.proto.framework.DictValueOrBuilder> dictValueBuilder_;
+    /**
+     * <pre>
+     * Represents a dict `Value`.
+     * </pre>
+     *
+     * <code>.tensorflow.DictValue dict_value = 53;</code>
+     */
+    public boolean hasDictValue() {
+      return kindCase_ == 53;
+    }
+    /**
+     * <pre>
+     * Represents a dict `Value`.
+     * </pre>
+     *
+     * <code>.tensorflow.DictValue dict_value = 53;</code>
+     */
+    public org.tensorflow.proto.framework.DictValue getDictValue() {
+      if (dictValueBuilder_ == null) {
+        if (kindCase_ == 53) {
+          return (org.tensorflow.proto.framework.DictValue) kind_;
+        }
+        return org.tensorflow.proto.framework.DictValue.getDefaultInstance();
+      } else {
+        if (kindCase_ == 53) {
+          return dictValueBuilder_.getMessage();
+        }
+        return org.tensorflow.proto.framework.DictValue.getDefaultInstance();
+      }
+    }
+    /**
+     * <pre>
+     * Represents a dict `Value`.
+     * </pre>
+     *
+     * <code>.tensorflow.DictValue dict_value = 53;</code>
+     */
+    public Builder setDictValue(org.tensorflow.proto.framework.DictValue value) {
+      if (dictValueBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        kind_ = value;
+        onChanged();
+      } else {
+        dictValueBuilder_.setMessage(value);
+      }
+      kindCase_ = 53;
+      return this;
+    }
+    /**
+     * <pre>
+     * Represents a dict `Value`.
+     * </pre>
+     *
+     * <code>.tensorflow.DictValue dict_value = 53;</code>
+     */
+    public Builder setDictValue(
+        org.tensorflow.proto.framework.DictValue.Builder builderForValue) {
+      if (dictValueBuilder_ == null) {
+        kind_ = builderForValue.build();
+        onChanged();
+      } else {
+        dictValueBuilder_.setMessage(builderForValue.build());
+      }
+      kindCase_ = 53;
+      return this;
+    }
+    /**
+     * <pre>
+     * Represents a dict `Value`.
+     * </pre>
+     *
+     * <code>.tensorflow.DictValue dict_value = 53;</code>
+     */
+    public Builder mergeDictValue(org.tensorflow.proto.framework.DictValue value) {
+      if (dictValueBuilder_ == null) {
+        if (kindCase_ == 53 &&
+            kind_ != org.tensorflow.proto.framework.DictValue.getDefaultInstance()) {
+          kind_ = org.tensorflow.proto.framework.DictValue.newBuilder((org.tensorflow.proto.framework.DictValue) kind_)
+              .mergeFrom(value).buildPartial();
+        } else {
+          kind_ = value;
+        }
+        onChanged();
+      } else {
+        if (kindCase_ == 53) {
+          dictValueBuilder_.mergeFrom(value);
+        }
+        dictValueBuilder_.setMessage(value);
+      }
+      kindCase_ = 53;
+      return this;
+    }
+    /**
+     * <pre>
+     * Represents a dict `Value`.
+     * </pre>
+     *
+     * <code>.tensorflow.DictValue dict_value = 53;</code>
+     */
+    public Builder clearDictValue() {
+      if (dictValueBuilder_ == null) {
+        if (kindCase_ == 53) {
+          kindCase_ = 0;
+          kind_ = null;
+          onChanged();
+        }
+      } else {
+        if (kindCase_ == 53) {
+          kindCase_ = 0;
+          kind_ = null;
+        }
+        dictValueBuilder_.clear();
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Represents a dict `Value`.
+     * </pre>
+     *
+     * <code>.tensorflow.DictValue dict_value = 53;</code>
+     */
+    public org.tensorflow.proto.framework.DictValue.Builder getDictValueBuilder() {
+      return getDictValueFieldBuilder().getBuilder();
+    }
+    /**
+     * <pre>
+     * Represents a dict `Value`.
+     * </pre>
+     *
+     * <code>.tensorflow.DictValue dict_value = 53;</code>
+     */
+    public org.tensorflow.proto.framework.DictValueOrBuilder getDictValueOrBuilder() {
+      if ((kindCase_ == 53) && (dictValueBuilder_ != null)) {
+        return dictValueBuilder_.getMessageOrBuilder();
+      } else {
+        if (kindCase_ == 53) {
+          return (org.tensorflow.proto.framework.DictValue) kind_;
+        }
+        return org.tensorflow.proto.framework.DictValue.getDefaultInstance();
+      }
+    }
+    /**
+     * <pre>
+     * Represents a dict `Value`.
+     * </pre>
+     *
+     * <code>.tensorflow.DictValue dict_value = 53;</code>
+     */
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.DictValue, org.tensorflow.proto.framework.DictValue.Builder, org.tensorflow.proto.framework.DictValueOrBuilder> 
+        getDictValueFieldBuilder() {
+      if (dictValueBuilder_ == null) {
+        if (!(kindCase_ == 53)) {
+          kind_ = org.tensorflow.proto.framework.DictValue.getDefaultInstance();
+        }
+        dictValueBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+            org.tensorflow.proto.framework.DictValue, org.tensorflow.proto.framework.DictValue.Builder, org.tensorflow.proto.framework.DictValueOrBuilder>(
+                (org.tensorflow.proto.framework.DictValue) kind_,
+                getParentForChildren(),
+                isClean());
+        kind_ = null;
+      }
+      kindCase_ = 53;
+      onChanged();;
+      return dictValueBuilder_;
+    }
+
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.NamedTupleValue, org.tensorflow.proto.framework.NamedTupleValue.Builder, org.tensorflow.proto.framework.NamedTupleValueOrBuilder> namedTupleValueBuilder_;
+    /**
+     * <pre>
+     * Represents Python's namedtuple.
+     * </pre>
+     *
+     * <code>.tensorflow.NamedTupleValue named_tuple_value = 54;</code>
+     */
+    public boolean hasNamedTupleValue() {
+      return kindCase_ == 54;
+    }
+    /**
+     * <pre>
+     * Represents Python's namedtuple.
+     * </pre>
+     *
+     * <code>.tensorflow.NamedTupleValue named_tuple_value = 54;</code>
+     */
+    public org.tensorflow.proto.framework.NamedTupleValue getNamedTupleValue() {
+      if (namedTupleValueBuilder_ == null) {
+        if (kindCase_ == 54) {
+          return (org.tensorflow.proto.framework.NamedTupleValue) kind_;
+        }
+        return org.tensorflow.proto.framework.NamedTupleValue.getDefaultInstance();
+      } else {
+        if (kindCase_ == 54) {
+          return namedTupleValueBuilder_.getMessage();
+        }
+        return org.tensorflow.proto.framework.NamedTupleValue.getDefaultInstance();
+      }
+    }
+    /**
+     * <pre>
+     * Represents Python's namedtuple.
+     * </pre>
+     *
+     * <code>.tensorflow.NamedTupleValue named_tuple_value = 54;</code>
+     */
+    public Builder setNamedTupleValue(org.tensorflow.proto.framework.NamedTupleValue value) {
+      if (namedTupleValueBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        kind_ = value;
+        onChanged();
+      } else {
+        namedTupleValueBuilder_.setMessage(value);
+      }
+      kindCase_ = 54;
+      return this;
+    }
+    /**
+     * <pre>
+     * Represents Python's namedtuple.
+     * </pre>
+     *
+     * <code>.tensorflow.NamedTupleValue named_tuple_value = 54;</code>
+     */
+    public Builder setNamedTupleValue(
+        org.tensorflow.proto.framework.NamedTupleValue.Builder builderForValue) {
+      if (namedTupleValueBuilder_ == null) {
+        kind_ = builderForValue.build();
+        onChanged();
+      } else {
+        namedTupleValueBuilder_.setMessage(builderForValue.build());
+      }
+      kindCase_ = 54;
+      return this;
+    }
+    /**
+     * <pre>
+     * Represents Python's namedtuple.
+     * </pre>
+     *
+     * <code>.tensorflow.NamedTupleValue named_tuple_value = 54;</code>
+     */
+    public Builder mergeNamedTupleValue(org.tensorflow.proto.framework.NamedTupleValue value) {
+      if (namedTupleValueBuilder_ == null) {
+        if (kindCase_ == 54 &&
+            kind_ != org.tensorflow.proto.framework.NamedTupleValue.getDefaultInstance()) {
+          kind_ = org.tensorflow.proto.framework.NamedTupleValue.newBuilder((org.tensorflow.proto.framework.NamedTupleValue) kind_)
+              .mergeFrom(value).buildPartial();
+        } else {
+          kind_ = value;
+        }
+        onChanged();
+      } else {
+        if (kindCase_ == 54) {
+          namedTupleValueBuilder_.mergeFrom(value);
+        }
+        namedTupleValueBuilder_.setMessage(value);
+      }
+      kindCase_ = 54;
+      return this;
+    }
+    /**
+     * <pre>
+     * Represents Python's namedtuple.
+     * </pre>
+     *
+     * <code>.tensorflow.NamedTupleValue named_tuple_value = 54;</code>
+     */
+    public Builder clearNamedTupleValue() {
+      if (namedTupleValueBuilder_ == null) {
+        if (kindCase_ == 54) {
+          kindCase_ = 0;
+          kind_ = null;
+          onChanged();
+        }
+      } else {
+        if (kindCase_ == 54) {
+          kindCase_ = 0;
+          kind_ = null;
+        }
+        namedTupleValueBuilder_.clear();
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Represents Python's namedtuple.
+     * </pre>
+     *
+     * <code>.tensorflow.NamedTupleValue named_tuple_value = 54;</code>
+     */
+    public org.tensorflow.proto.framework.NamedTupleValue.Builder getNamedTupleValueBuilder() {
+      return getNamedTupleValueFieldBuilder().getBuilder();
+    }
+    /**
+     * <pre>
+     * Represents Python's namedtuple.
+     * </pre>
+     *
+     * <code>.tensorflow.NamedTupleValue named_tuple_value = 54;</code>
+     */
+    public org.tensorflow.proto.framework.NamedTupleValueOrBuilder getNamedTupleValueOrBuilder() {
+      if ((kindCase_ == 54) && (namedTupleValueBuilder_ != null)) {
+        return namedTupleValueBuilder_.getMessageOrBuilder();
+      } else {
+        if (kindCase_ == 54) {
+          return (org.tensorflow.proto.framework.NamedTupleValue) kind_;
+        }
+        return org.tensorflow.proto.framework.NamedTupleValue.getDefaultInstance();
+      }
+    }
+    /**
+     * <pre>
+     * Represents Python's namedtuple.
+     * </pre>
+     *
+     * <code>.tensorflow.NamedTupleValue named_tuple_value = 54;</code>
+     */
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.NamedTupleValue, org.tensorflow.proto.framework.NamedTupleValue.Builder, org.tensorflow.proto.framework.NamedTupleValueOrBuilder> 
+        getNamedTupleValueFieldBuilder() {
+      if (namedTupleValueBuilder_ == null) {
+        if (!(kindCase_ == 54)) {
+          kind_ = org.tensorflow.proto.framework.NamedTupleValue.getDefaultInstance();
+        }
+        namedTupleValueBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+            org.tensorflow.proto.framework.NamedTupleValue, org.tensorflow.proto.framework.NamedTupleValue.Builder, org.tensorflow.proto.framework.NamedTupleValueOrBuilder>(
+                (org.tensorflow.proto.framework.NamedTupleValue) kind_,
+                getParentForChildren(),
+                isClean());
+        kind_ = null;
+      }
+      kindCase_ = 54;
+      onChanged();;
+      return namedTupleValueBuilder_;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.StructuredValue)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.StructuredValue)
+  private static final org.tensorflow.proto.framework.StructuredValue DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.StructuredValue();
+  }
+
+  public static org.tensorflow.proto.framework.StructuredValue getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<StructuredValue>
+      PARSER = new com.google.protobuf.AbstractParser<StructuredValue>() {
+    @java.lang.Override
+    public StructuredValue parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new StructuredValue(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<StructuredValue> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<StructuredValue> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.framework.StructuredValue getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/StructuredValueOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/StructuredValueOrBuilder.java
new file mode 100644
index 00000000000..3ffb498eb22
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/StructuredValueOrBuilder.java
@@ -0,0 +1,309 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/struct.proto
+
+package org.tensorflow.proto.framework;
+
+public interface StructuredValueOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.StructuredValue)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <pre>
+   * Represents None.
+   * </pre>
+   *
+   * <code>.tensorflow.NoneValue none_value = 1;</code>
+   */
+  boolean hasNoneValue();
+  /**
+   * <pre>
+   * Represents None.
+   * </pre>
+   *
+   * <code>.tensorflow.NoneValue none_value = 1;</code>
+   */
+  org.tensorflow.proto.framework.NoneValue getNoneValue();
+  /**
+   * <pre>
+   * Represents None.
+   * </pre>
+   *
+   * <code>.tensorflow.NoneValue none_value = 1;</code>
+   */
+  org.tensorflow.proto.framework.NoneValueOrBuilder getNoneValueOrBuilder();
+
+  /**
+   * <pre>
+   * Represents a double-precision floating-point value (a Python `float`).
+   * </pre>
+   *
+   * <code>double float64_value = 11;</code>
+   */
+  double getFloat64Value();
+
+  /**
+   * <pre>
+   * Represents a signed integer value, limited to 64 bits.
+   * Larger values from Python's arbitrary-precision integers are unsupported.
+   * </pre>
+   *
+   * <code>sint64 int64_value = 12;</code>
+   */
+  long getInt64Value();
+
+  /**
+   * <pre>
+   * Represents a string of Unicode characters stored in a Python `str`.
+   * In Python 3, this is exactly what type `str` is.
+   * In Python 2, this is the UTF-8 encoding of the characters.
+   * For strings with ASCII characters only (as often used in TensorFlow code)
+   * there is effectively no difference between the language versions.
+   * The obsolescent `unicode` type of Python 2 is not supported here.
+   * </pre>
+   *
+   * <code>string string_value = 13;</code>
+   */
+  java.lang.String getStringValue();
+  /**
+   * <pre>
+   * Represents a string of Unicode characters stored in a Python `str`.
+   * In Python 3, this is exactly what type `str` is.
+   * In Python 2, this is the UTF-8 encoding of the characters.
+   * For strings with ASCII characters only (as often used in TensorFlow code)
+   * there is effectively no difference between the language versions.
+   * The obsolescent `unicode` type of Python 2 is not supported here.
+   * </pre>
+   *
+   * <code>string string_value = 13;</code>
+   */
+  com.google.protobuf.ByteString
+      getStringValueBytes();
+
+  /**
+   * <pre>
+   * Represents a boolean value.
+   * </pre>
+   *
+   * <code>bool bool_value = 14;</code>
+   */
+  boolean getBoolValue();
+
+  /**
+   * <pre>
+   * Represents a TensorShape.
+   * </pre>
+   *
+   * <code>.tensorflow.TensorShapeProto tensor_shape_value = 31;</code>
+   */
+  boolean hasTensorShapeValue();
+  /**
+   * <pre>
+   * Represents a TensorShape.
+   * </pre>
+   *
+   * <code>.tensorflow.TensorShapeProto tensor_shape_value = 31;</code>
+   */
+  org.tensorflow.proto.framework.TensorShapeProto getTensorShapeValue();
+  /**
+   * <pre>
+   * Represents a TensorShape.
+   * </pre>
+   *
+   * <code>.tensorflow.TensorShapeProto tensor_shape_value = 31;</code>
+   */
+  org.tensorflow.proto.framework.TensorShapeProtoOrBuilder getTensorShapeValueOrBuilder();
+
+  /**
+   * <pre>
+   * Represents an enum value for dtype.
+   * </pre>
+   *
+   * <code>.tensorflow.DataType tensor_dtype_value = 32;</code>
+   */
+  int getTensorDtypeValueValue();
+  /**
+   * <pre>
+   * Represents an enum value for dtype.
+   * </pre>
+   *
+   * <code>.tensorflow.DataType tensor_dtype_value = 32;</code>
+   */
+  org.tensorflow.proto.framework.DataType getTensorDtypeValue();
+
+  /**
+   * <pre>
+   * Represents a value for tf.TensorSpec.
+   * </pre>
+   *
+   * <code>.tensorflow.TensorSpecProto tensor_spec_value = 33;</code>
+   */
+  boolean hasTensorSpecValue();
+  /**
+   * <pre>
+   * Represents a value for tf.TensorSpec.
+   * </pre>
+   *
+   * <code>.tensorflow.TensorSpecProto tensor_spec_value = 33;</code>
+   */
+  org.tensorflow.proto.framework.TensorSpecProto getTensorSpecValue();
+  /**
+   * <pre>
+   * Represents a value for tf.TensorSpec.
+   * </pre>
+   *
+   * <code>.tensorflow.TensorSpecProto tensor_spec_value = 33;</code>
+   */
+  org.tensorflow.proto.framework.TensorSpecProtoOrBuilder getTensorSpecValueOrBuilder();
+
+  /**
+   * <pre>
+   * Represents a value for tf.TypeSpec.
+   * </pre>
+   *
+   * <code>.tensorflow.TypeSpecProto type_spec_value = 34;</code>
+   */
+  boolean hasTypeSpecValue();
+  /**
+   * <pre>
+   * Represents a value for tf.TypeSpec.
+   * </pre>
+   *
+   * <code>.tensorflow.TypeSpecProto type_spec_value = 34;</code>
+   */
+  org.tensorflow.proto.framework.TypeSpecProto getTypeSpecValue();
+  /**
+   * <pre>
+   * Represents a value for tf.TypeSpec.
+   * </pre>
+   *
+   * <code>.tensorflow.TypeSpecProto type_spec_value = 34;</code>
+   */
+  org.tensorflow.proto.framework.TypeSpecProtoOrBuilder getTypeSpecValueOrBuilder();
+
+  /**
+   * <pre>
+   * Represents a value for tf.BoundedTensorSpec.
+   * </pre>
+   *
+   * <code>.tensorflow.BoundedTensorSpecProto bounded_tensor_spec_value = 35;</code>
+   */
+  boolean hasBoundedTensorSpecValue();
+  /**
+   * <pre>
+   * Represents a value for tf.BoundedTensorSpec.
+   * </pre>
+   *
+   * <code>.tensorflow.BoundedTensorSpecProto bounded_tensor_spec_value = 35;</code>
+   */
+  org.tensorflow.proto.framework.BoundedTensorSpecProto getBoundedTensorSpecValue();
+  /**
+   * <pre>
+   * Represents a value for tf.BoundedTensorSpec.
+   * </pre>
+   *
+   * <code>.tensorflow.BoundedTensorSpecProto bounded_tensor_spec_value = 35;</code>
+   */
+  org.tensorflow.proto.framework.BoundedTensorSpecProtoOrBuilder getBoundedTensorSpecValueOrBuilder();
+
+  /**
+   * <pre>
+   * Represents a list of `Value`.
+   * </pre>
+   *
+   * <code>.tensorflow.ListValue list_value = 51;</code>
+   */
+  boolean hasListValue();
+  /**
+   * <pre>
+   * Represents a list of `Value`.
+   * </pre>
+   *
+   * <code>.tensorflow.ListValue list_value = 51;</code>
+   */
+  org.tensorflow.proto.framework.ListValue getListValue();
+  /**
+   * <pre>
+   * Represents a list of `Value`.
+   * </pre>
+   *
+   * <code>.tensorflow.ListValue list_value = 51;</code>
+   */
+  org.tensorflow.proto.framework.ListValueOrBuilder getListValueOrBuilder();
+
+  /**
+   * <pre>
+   * Represents a tuple of `Value`.
+   * </pre>
+   *
+   * <code>.tensorflow.TupleValue tuple_value = 52;</code>
+   */
+  boolean hasTupleValue();
+  /**
+   * <pre>
+   * Represents a tuple of `Value`.
+   * </pre>
+   *
+   * <code>.tensorflow.TupleValue tuple_value = 52;</code>
+   */
+  org.tensorflow.proto.framework.TupleValue getTupleValue();
+  /**
+   * <pre>
+   * Represents a tuple of `Value`.
+   * </pre>
+   *
+   * <code>.tensorflow.TupleValue tuple_value = 52;</code>
+   */
+  org.tensorflow.proto.framework.TupleValueOrBuilder getTupleValueOrBuilder();
+
+  /**
+   * <pre>
+   * Represents a dict `Value`.
+   * </pre>
+   *
+   * <code>.tensorflow.DictValue dict_value = 53;</code>
+   */
+  boolean hasDictValue();
+  /**
+   * <pre>
+   * Represents a dict `Value`.
+   * </pre>
+   *
+   * <code>.tensorflow.DictValue dict_value = 53;</code>
+   */
+  org.tensorflow.proto.framework.DictValue getDictValue();
+  /**
+   * <pre>
+   * Represents a dict `Value`.
+   * </pre>
+   *
+   * <code>.tensorflow.DictValue dict_value = 53;</code>
+   */
+  org.tensorflow.proto.framework.DictValueOrBuilder getDictValueOrBuilder();
+
+  /**
+   * <pre>
+   * Represents Python's namedtuple.
+   * </pre>
+   *
+   * <code>.tensorflow.NamedTupleValue named_tuple_value = 54;</code>
+   */
+  boolean hasNamedTupleValue();
+  /**
+   * <pre>
+   * Represents Python's namedtuple.
+   * </pre>
+   *
+   * <code>.tensorflow.NamedTupleValue named_tuple_value = 54;</code>
+   */
+  org.tensorflow.proto.framework.NamedTupleValue getNamedTupleValue();
+  /**
+   * <pre>
+   * Represents Python's namedtuple.
+   * </pre>
+   *
+   * <code>.tensorflow.NamedTupleValue named_tuple_value = 54;</code>
+   */
+  org.tensorflow.proto.framework.NamedTupleValueOrBuilder getNamedTupleValueOrBuilder();
+
+  public org.tensorflow.proto.framework.StructuredValue.KindCase getKindCase();
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/Summary.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/Summary.java
new file mode 100644
index 00000000000..19c5c6664a3
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/Summary.java
@@ -0,0 +1,4725 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/framework/summary.proto
+
+package org.tensorflow.proto.framework;
+
+/**
+ * <pre>
+ * A Summary is a set of named values to be displayed by the
+ * visualizer.
+ * Summaries are produced regularly during training, as controlled by
+ * the "summary_interval_secs" attribute of the training operation.
+ * Summaries are also produced at the end of an evaluation.
+ * </pre>
+ *
+ * Protobuf type {@code tensorflow.Summary}
+ */
+public  final class Summary extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.Summary)
+    SummaryOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use Summary.newBuilder() to construct.
+  private Summary(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private Summary() {
+    value_ = java.util.Collections.emptyList();
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new Summary();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private Summary(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    int mutable_bitField0_ = 0;
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 10: {
+            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
+              value_ = new java.util.ArrayList<org.tensorflow.proto.framework.Summary.Value>();
+              mutable_bitField0_ |= 0x00000001;
+            }
+            value_.add(
+                input.readMessage(org.tensorflow.proto.framework.Summary.Value.parser(), extensionRegistry));
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      if (((mutable_bitField0_ & 0x00000001) != 0)) {
+        value_ = java.util.Collections.unmodifiableList(value_);
+      }
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.framework.SummaryProtos.internal_static_tensorflow_Summary_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.framework.SummaryProtos.internal_static_tensorflow_Summary_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.framework.Summary.class, org.tensorflow.proto.framework.Summary.Builder.class);
+  }
+
+  public interface ImageOrBuilder extends
+      // @@protoc_insertion_point(interface_extends:tensorflow.Summary.Image)
+      com.google.protobuf.MessageOrBuilder {
+
+    /**
+     * <pre>
+     * Dimensions of the image.
+     * </pre>
+     *
+     * <code>int32 height = 1;</code>
+     */
+    int getHeight();
+
+    /**
+     * <code>int32 width = 2;</code>
+     */
+    int getWidth();
+
+    /**
+     * <pre>
+     * Valid colorspace values are
+     *   1 - grayscale
+     *   2 - grayscale + alpha
+     *   3 - RGB
+     *   4 - RGBA
+     *   5 - DIGITAL_YUV
+     *   6 - BGRA
+     * </pre>
+     *
+     * <code>int32 colorspace = 3;</code>
+     */
+    int getColorspace();
+
+    /**
+     * <pre>
+     * Image data in encoded format.  All image formats supported by
+     * image_codec::CoderUtil can be stored here.
+     * </pre>
+     *
+     * <code>bytes encoded_image_string = 4;</code>
+     */
+    com.google.protobuf.ByteString getEncodedImageString();
+  }
+  /**
+   * Protobuf type {@code tensorflow.Summary.Image}
+   */
+  public  static final class Image extends
+      com.google.protobuf.GeneratedMessageV3 implements
+      // @@protoc_insertion_point(message_implements:tensorflow.Summary.Image)
+      ImageOrBuilder {
+  private static final long serialVersionUID = 0L;
+    // Use Image.newBuilder() to construct.
+    private Image(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+      super(builder);
+    }
+    private Image() {
+      encodedImageString_ = com.google.protobuf.ByteString.EMPTY;
+    }
+
+    @java.lang.Override
+    @SuppressWarnings({"unused"})
+    protected java.lang.Object newInstance(
+        UnusedPrivateParameter unused) {
+      return new Image();
+    }
+
+    @java.lang.Override
+    public final com.google.protobuf.UnknownFieldSet
+    getUnknownFields() {
+      return this.unknownFields;
+    }
+    private Image(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      this();
+      if (extensionRegistry == null) {
+        throw new java.lang.NullPointerException();
+      }
+      com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+          com.google.protobuf.UnknownFieldSet.newBuilder();
+      try {
+        boolean done = false;
+        while (!done) {
+          int tag = input.readTag();
+          switch (tag) {
+            case 0:
+              done = true;
+              break;
+            case 8: {
+
+              height_ = input.readInt32();
+              break;
+            }
+            case 16: {
+
+              width_ = input.readInt32();
+              break;
+            }
+            case 24: {
+
+              colorspace_ = input.readInt32();
+              break;
+            }
+            case 34: {
+
+              encodedImageString_ = input.readBytes();
+              break;
+            }
+            default: {
+              if (!parseUnknownField(
+                  input, unknownFields, extensionRegistry, tag)) {
+                done = true;
+              }
+              break;
+            }
+          }
+        }
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        throw e.setUnfinishedMessage(this);
+      } catch (java.io.IOException e) {
+        throw new com.google.protobuf.InvalidProtocolBufferException(
+            e).setUnfinishedMessage(this);
+      } finally {
+        this.unknownFields = unknownFields.build();
+        makeExtensionsImmutable();
+      }
+    }
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.SummaryProtos.internal_static_tensorflow_Summary_Image_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.SummaryProtos.internal_static_tensorflow_Summary_Image_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.Summary.Image.class, org.tensorflow.proto.framework.Summary.Image.Builder.class);
+    }
+
+    public static final int HEIGHT_FIELD_NUMBER = 1;
+    private int height_;
+    /**
+     * <pre>
+     * Dimensions of the image.
+     * </pre>
+     *
+     * <code>int32 height = 1;</code>
+     */
+    public int getHeight() {
+      return height_;
+    }
+
+    public static final int WIDTH_FIELD_NUMBER = 2;
+    private int width_;
+    /**
+     * <code>int32 width = 2;</code>
+     */
+    public int getWidth() {
+      return width_;
+    }
+
+    public static final int COLORSPACE_FIELD_NUMBER = 3;
+    private int colorspace_;
+    /**
+     * <pre>
+     * Valid colorspace values are
+     *   1 - grayscale
+     *   2 - grayscale + alpha
+     *   3 - RGB
+     *   4 - RGBA
+     *   5 - DIGITAL_YUV
+     *   6 - BGRA
+     * </pre>
+     *
+     * <code>int32 colorspace = 3;</code>
+     */
+    public int getColorspace() {
+      return colorspace_;
+    }
+
+    public static final int ENCODED_IMAGE_STRING_FIELD_NUMBER = 4;
+    private com.google.protobuf.ByteString encodedImageString_;
+    /**
+     * <pre>
+     * Image data in encoded format.  All image formats supported by
+     * image_codec::CoderUtil can be stored here.
+     * </pre>
+     *
+     * <code>bytes encoded_image_string = 4;</code>
+     */
+    public com.google.protobuf.ByteString getEncodedImageString() {
+      return encodedImageString_;
+    }
+
+    private byte memoizedIsInitialized = -1;
+    @java.lang.Override
+    public final boolean isInitialized() {
+      byte isInitialized = memoizedIsInitialized;
+      if (isInitialized == 1) return true;
+      if (isInitialized == 0) return false;
+
+      memoizedIsInitialized = 1;
+      return true;
+    }
+
+    @java.lang.Override
+    public void writeTo(com.google.protobuf.CodedOutputStream output)
+                        throws java.io.IOException {
+      if (height_ != 0) {
+        output.writeInt32(1, height_);
+      }
+      if (width_ != 0) {
+        output.writeInt32(2, width_);
+      }
+      if (colorspace_ != 0) {
+        output.writeInt32(3, colorspace_);
+      }
+      if (!encodedImageString_.isEmpty()) {
+        output.writeBytes(4, encodedImageString_);
+      }
+      unknownFields.writeTo(output);
+    }
+
+    @java.lang.Override
+    public int getSerializedSize() {
+      int size = memoizedSize;
+      if (size != -1) return size;
+
+      size = 0;
+      if (height_ != 0) {
+        size += com.google.protobuf.CodedOutputStream
+          .computeInt32Size(1, height_);
+      }
+      if (width_ != 0) {
+        size += com.google.protobuf.CodedOutputStream
+          .computeInt32Size(2, width_);
+      }
+      if (colorspace_ != 0) {
+        size += com.google.protobuf.CodedOutputStream
+          .computeInt32Size(3, colorspace_);
+      }
+      if (!encodedImageString_.isEmpty()) {
+        size += com.google.protobuf.CodedOutputStream
+          .computeBytesSize(4, encodedImageString_);
+      }
+      size += unknownFields.getSerializedSize();
+      memoizedSize = size;
+      return size;
+    }
+
+    @java.lang.Override
+    public boolean equals(final java.lang.Object obj) {
+      if (obj == this) {
+       return true;
+      }
+      if (!(obj instanceof org.tensorflow.proto.framework.Summary.Image)) {
+        return super.equals(obj);
+      }
+      org.tensorflow.proto.framework.Summary.Image other = (org.tensorflow.proto.framework.Summary.Image) obj;
+
+      if (getHeight()
+          != other.getHeight()) return false;
+      if (getWidth()
+          != other.getWidth()) return false;
+      if (getColorspace()
+          != other.getColorspace()) return false;
+      if (!getEncodedImageString()
+          .equals(other.getEncodedImageString())) return false;
+      if (!unknownFields.equals(other.unknownFields)) return false;
+      return true;
+    }
+
+    @java.lang.Override
+    public int hashCode() {
+      if (memoizedHashCode != 0) {
+        return memoizedHashCode;
+      }
+      int hash = 41;
+      hash = (19 * hash) + getDescriptor().hashCode();
+      hash = (37 * hash) + HEIGHT_FIELD_NUMBER;
+      hash = (53 * hash) + getHeight();
+      hash = (37 * hash) + WIDTH_FIELD_NUMBER;
+      hash = (53 * hash) + getWidth();
+      hash = (37 * hash) + COLORSPACE_FIELD_NUMBER;
+      hash = (53 * hash) + getColorspace();
+      hash = (37 * hash) + ENCODED_IMAGE_STRING_FIELD_NUMBER;
+      hash = (53 * hash) + getEncodedImageString().hashCode();
+      hash = (29 * hash) + unknownFields.hashCode();
+      memoizedHashCode = hash;
+      return hash;
+    }
+
+    public static org.tensorflow.proto.framework.Summary.Image parseFrom(
+        java.nio.ByteBuffer data)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data);
+    }
+    public static org.tensorflow.proto.framework.Summary.Image parseFrom(
+        java.nio.ByteBuffer data,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.Summary.Image parseFrom(
+        com.google.protobuf.ByteString data)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data);
+    }
+    public static org.tensorflow.proto.framework.Summary.Image parseFrom(
+        com.google.protobuf.ByteString data,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.Summary.Image parseFrom(byte[] data)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data);
+    }
+    public static org.tensorflow.proto.framework.Summary.Image parseFrom(
+        byte[] data,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.Summary.Image parseFrom(java.io.InputStream input)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input);
+    }
+    public static org.tensorflow.proto.framework.Summary.Image parseFrom(
+        java.io.InputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.Summary.Image parseDelimitedFrom(java.io.InputStream input)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseDelimitedWithIOException(PARSER, input);
+    }
+    public static org.tensorflow.proto.framework.Summary.Image parseDelimitedFrom(
+        java.io.InputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.Summary.Image parseFrom(
+        com.google.protobuf.CodedInputStream input)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input);
+    }
+    public static org.tensorflow.proto.framework.Summary.Image parseFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input, extensionRegistry);
+    }
+
+    @java.lang.Override
+    public Builder newBuilderForType() { return newBuilder(); }
+    public static Builder newBuilder() {
+      return DEFAULT_INSTANCE.toBuilder();
+    }
+    public static Builder newBuilder(org.tensorflow.proto.framework.Summary.Image prototype) {
+      return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+    }
+    @java.lang.Override
+    public Builder toBuilder() {
+      return this == DEFAULT_INSTANCE
+          ? new Builder() : new Builder().mergeFrom(this);
+    }
+
+    @java.lang.Override
+    protected Builder newBuilderForType(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      Builder builder = new Builder(parent);
+      return builder;
+    }
+    /**
+     * Protobuf type {@code tensorflow.Summary.Image}
+     */
+    public static final class Builder extends
+        com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+        // @@protoc_insertion_point(builder_implements:tensorflow.Summary.Image)
+        org.tensorflow.proto.framework.Summary.ImageOrBuilder {
+      public static final com.google.protobuf.Descriptors.Descriptor
+          getDescriptor() {
+        return org.tensorflow.proto.framework.SummaryProtos.internal_static_tensorflow_Summary_Image_descriptor;
+      }
+
+      @java.lang.Override
+      protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+          internalGetFieldAccessorTable() {
+        return org.tensorflow.proto.framework.SummaryProtos.internal_static_tensorflow_Summary_Image_fieldAccessorTable
+            .ensureFieldAccessorsInitialized(
+                org.tensorflow.proto.framework.Summary.Image.class, org.tensorflow.proto.framework.Summary.Image.Builder.class);
+      }
+
+      // Construct using org.tensorflow.proto.framework.Summary.Image.newBuilder()
+      private Builder() {
+        maybeForceBuilderInitialization();
+      }
+
+      private Builder(
+          com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+        super(parent);
+        maybeForceBuilderInitialization();
+      }
+      private void maybeForceBuilderInitialization() {
+        if (com.google.protobuf.GeneratedMessageV3
+                .alwaysUseFieldBuilders) {
+        }
+      }
+      @java.lang.Override
+      public Builder clear() {
+        super.clear();
+        height_ = 0;
+
+        width_ = 0;
+
+        colorspace_ = 0;
+
+        encodedImageString_ = com.google.protobuf.ByteString.EMPTY;
+
+        return this;
+      }
+
+      @java.lang.Override
+      public com.google.protobuf.Descriptors.Descriptor
+          getDescriptorForType() {
+        return org.tensorflow.proto.framework.SummaryProtos.internal_static_tensorflow_Summary_Image_descriptor;
+      }
+
+      @java.lang.Override
+      public org.tensorflow.proto.framework.Summary.Image getDefaultInstanceForType() {
+        return org.tensorflow.proto.framework.Summary.Image.getDefaultInstance();
+      }
+
+      @java.lang.Override
+      public org.tensorflow.proto.framework.Summary.Image build() {
+        org.tensorflow.proto.framework.Summary.Image result = buildPartial();
+        if (!result.isInitialized()) {
+          throw newUninitializedMessageException(result);
+        }
+        return result;
+      }
+
+      @java.lang.Override
+      public org.tensorflow.proto.framework.Summary.Image buildPartial() {
+        org.tensorflow.proto.framework.Summary.Image result = new org.tensorflow.proto.framework.Summary.Image(this);
+        result.height_ = height_;
+        result.width_ = width_;
+        result.colorspace_ = colorspace_;
+        result.encodedImageString_ = encodedImageString_;
+        onBuilt();
+        return result;
+      }
+
+      @java.lang.Override
+      public Builder clone() {
+        return super.clone();
+      }
+      @java.lang.Override
+      public Builder setField(
+          com.google.protobuf.Descriptors.FieldDescriptor field,
+          java.lang.Object value) {
+        return super.setField(field, value);
+      }
+      @java.lang.Override
+      public Builder clearField(
+          com.google.protobuf.Descriptors.FieldDescriptor field) {
+        return super.clearField(field);
+      }
+      @java.lang.Override
+      public Builder clearOneof(
+          com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+        return super.clearOneof(oneof);
+      }
+      @java.lang.Override
+      public Builder setRepeatedField(
+          com.google.protobuf.Descriptors.FieldDescriptor field,
+          int index, java.lang.Object value) {
+        return super.setRepeatedField(field, index, value);
+      }
+      @java.lang.Override
+      public Builder addRepeatedField(
+          com.google.protobuf.Descriptors.FieldDescriptor field,
+          java.lang.Object value) {
+        return super.addRepeatedField(field, value);
+      }
+      @java.lang.Override
+      public Builder mergeFrom(com.google.protobuf.Message other) {
+        if (other instanceof org.tensorflow.proto.framework.Summary.Image) {
+          return mergeFrom((org.tensorflow.proto.framework.Summary.Image)other);
+        } else {
+          super.mergeFrom(other);
+          return this;
+        }
+      }
+
+      public Builder mergeFrom(org.tensorflow.proto.framework.Summary.Image other) {
+        if (other == org.tensorflow.proto.framework.Summary.Image.getDefaultInstance()) return this;
+        if (other.getHeight() != 0) {
+          setHeight(other.getHeight());
+        }
+        if (other.getWidth() != 0) {
+          setWidth(other.getWidth());
+        }
+        if (other.getColorspace() != 0) {
+          setColorspace(other.getColorspace());
+        }
+        if (other.getEncodedImageString() != com.google.protobuf.ByteString.EMPTY) {
+          setEncodedImageString(other.getEncodedImageString());
+        }
+        this.mergeUnknownFields(other.unknownFields);
+        onChanged();
+        return this;
+      }
+
+      @java.lang.Override
+      public final boolean isInitialized() {
+        return true;
+      }
+
+      @java.lang.Override
+      public Builder mergeFrom(
+          com.google.protobuf.CodedInputStream input,
+          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+          throws java.io.IOException {
+        org.tensorflow.proto.framework.Summary.Image parsedMessage = null;
+        try {
+          parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+        } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+          parsedMessage = (org.tensorflow.proto.framework.Summary.Image) e.getUnfinishedMessage();
+          throw e.unwrapIOException();
+        } finally {
+          if (parsedMessage != null) {
+            mergeFrom(parsedMessage);
+          }
+        }
+        return this;
+      }
+
+      private int height_ ;
+      /**
+       * <pre>
+       * Dimensions of the image.
+       * </pre>
+       *
+       * <code>int32 height = 1;</code>
+       */
+      public int getHeight() {
+        return height_;
+      }
+      /**
+       * <pre>
+       * Dimensions of the image.
+       * </pre>
+       *
+       * <code>int32 height = 1;</code>
+       */
+      public Builder setHeight(int value) {
+        
+        height_ = value;
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * Dimensions of the image.
+       * </pre>
+       *
+       * <code>int32 height = 1;</code>
+       */
+      public Builder clearHeight() {
+        
+        height_ = 0;
+        onChanged();
+        return this;
+      }
+
+      private int width_ ;
+      /**
+       * <code>int32 width = 2;</code>
+       */
+      public int getWidth() {
+        return width_;
+      }
+      /**
+       * <code>int32 width = 2;</code>
+       */
+      public Builder setWidth(int value) {
+        
+        width_ = value;
+        onChanged();
+        return this;
+      }
+      /**
+       * <code>int32 width = 2;</code>
+       */
+      public Builder clearWidth() {
+        
+        width_ = 0;
+        onChanged();
+        return this;
+      }
+
+      private int colorspace_ ;
+      /**
+       * <pre>
+       * Valid colorspace values are
+       *   1 - grayscale
+       *   2 - grayscale + alpha
+       *   3 - RGB
+       *   4 - RGBA
+       *   5 - DIGITAL_YUV
+       *   6 - BGRA
+       * </pre>
+       *
+       * <code>int32 colorspace = 3;</code>
+       */
+      public int getColorspace() {
+        return colorspace_;
+      }
+      /**
+       * <pre>
+       * Valid colorspace values are
+       *   1 - grayscale
+       *   2 - grayscale + alpha
+       *   3 - RGB
+       *   4 - RGBA
+       *   5 - DIGITAL_YUV
+       *   6 - BGRA
+       * </pre>
+       *
+       * <code>int32 colorspace = 3;</code>
+       */
+      public Builder setColorspace(int value) {
+        
+        colorspace_ = value;
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * Valid colorspace values are
+       *   1 - grayscale
+       *   2 - grayscale + alpha
+       *   3 - RGB
+       *   4 - RGBA
+       *   5 - DIGITAL_YUV
+       *   6 - BGRA
+       * </pre>
+       *
+       * <code>int32 colorspace = 3;</code>
+       */
+      public Builder clearColorspace() {
+        
+        colorspace_ = 0;
+        onChanged();
+        return this;
+      }
+
+      private com.google.protobuf.ByteString encodedImageString_ = com.google.protobuf.ByteString.EMPTY;
+      /**
+       * <pre>
+       * Image data in encoded format.  All image formats supported by
+       * image_codec::CoderUtil can be stored here.
+       * </pre>
+       *
+       * <code>bytes encoded_image_string = 4;</code>
+       */
+      public com.google.protobuf.ByteString getEncodedImageString() {
+        return encodedImageString_;
+      }
+      /**
+       * <pre>
+       * Image data in encoded format.  All image formats supported by
+       * image_codec::CoderUtil can be stored here.
+       * </pre>
+       *
+       * <code>bytes encoded_image_string = 4;</code>
+       */
+      public Builder setEncodedImageString(com.google.protobuf.ByteString value) {
+        if (value == null) {
+    throw new NullPointerException();
+  }
+  
+        encodedImageString_ = value;
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * Image data in encoded format.  All image formats supported by
+       * image_codec::CoderUtil can be stored here.
+       * </pre>
+       *
+       * <code>bytes encoded_image_string = 4;</code>
+       */
+      public Builder clearEncodedImageString() {
+        
+        encodedImageString_ = getDefaultInstance().getEncodedImageString();
+        onChanged();
+        return this;
+      }
+      @java.lang.Override
+      public final Builder setUnknownFields(
+          final com.google.protobuf.UnknownFieldSet unknownFields) {
+        return super.setUnknownFields(unknownFields);
+      }
+
+      @java.lang.Override
+      public final Builder mergeUnknownFields(
+          final com.google.protobuf.UnknownFieldSet unknownFields) {
+        return super.mergeUnknownFields(unknownFields);
+      }
+
+
+      // @@protoc_insertion_point(builder_scope:tensorflow.Summary.Image)
+    }
+
+    // @@protoc_insertion_point(class_scope:tensorflow.Summary.Image)
+    private static final org.tensorflow.proto.framework.Summary.Image DEFAULT_INSTANCE;
+    static {
+      DEFAULT_INSTANCE = new org.tensorflow.proto.framework.Summary.Image();
+    }
+
+    public static org.tensorflow.proto.framework.Summary.Image getDefaultInstance() {
+      return DEFAULT_INSTANCE;
+    }
+
+    private static final com.google.protobuf.Parser<Image>
+        PARSER = new com.google.protobuf.AbstractParser<Image>() {
+      @java.lang.Override
+      public Image parsePartialFrom(
+          com.google.protobuf.CodedInputStream input,
+          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+          throws com.google.protobuf.InvalidProtocolBufferException {
+        return new Image(input, extensionRegistry);
+      }
+    };
+
+    public static com.google.protobuf.Parser<Image> parser() {
+      return PARSER;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Parser<Image> getParserForType() {
+      return PARSER;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.Summary.Image getDefaultInstanceForType() {
+      return DEFAULT_INSTANCE;
+    }
+
+  }
+
+  public interface AudioOrBuilder extends
+      // @@protoc_insertion_point(interface_extends:tensorflow.Summary.Audio)
+      com.google.protobuf.MessageOrBuilder {
+
+    /**
+     * <pre>
+     * Sample rate of the audio in Hz.
+     * </pre>
+     *
+     * <code>float sample_rate = 1;</code>
+     */
+    float getSampleRate();
+
+    /**
+     * <pre>
+     * Number of channels of audio.
+     * </pre>
+     *
+     * <code>int64 num_channels = 2;</code>
+     */
+    long getNumChannels();
+
+    /**
+     * <pre>
+     * Length of the audio in frames (samples per channel).
+     * </pre>
+     *
+     * <code>int64 length_frames = 3;</code>
+     */
+    long getLengthFrames();
+
+    /**
+     * <pre>
+     * Encoded audio data and its associated RFC 2045 content type (e.g.
+     * "audio/wav").
+     * </pre>
+     *
+     * <code>bytes encoded_audio_string = 4;</code>
+     */
+    com.google.protobuf.ByteString getEncodedAudioString();
+
+    /**
+     * <code>string content_type = 5;</code>
+     */
+    java.lang.String getContentType();
+    /**
+     * <code>string content_type = 5;</code>
+     */
+    com.google.protobuf.ByteString
+        getContentTypeBytes();
+  }
+  /**
+   * Protobuf type {@code tensorflow.Summary.Audio}
+   */
+  public  static final class Audio extends
+      com.google.protobuf.GeneratedMessageV3 implements
+      // @@protoc_insertion_point(message_implements:tensorflow.Summary.Audio)
+      AudioOrBuilder {
+  private static final long serialVersionUID = 0L;
+    // Use Audio.newBuilder() to construct.
+    private Audio(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+      super(builder);
+    }
+    private Audio() {
+      encodedAudioString_ = com.google.protobuf.ByteString.EMPTY;
+      contentType_ = "";
+    }
+
+    @java.lang.Override
+    @SuppressWarnings({"unused"})
+    protected java.lang.Object newInstance(
+        UnusedPrivateParameter unused) {
+      return new Audio();
+    }
+
+    @java.lang.Override
+    public final com.google.protobuf.UnknownFieldSet
+    getUnknownFields() {
+      return this.unknownFields;
+    }
+    private Audio(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      this();
+      if (extensionRegistry == null) {
+        throw new java.lang.NullPointerException();
+      }
+      com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+          com.google.protobuf.UnknownFieldSet.newBuilder();
+      try {
+        boolean done = false;
+        while (!done) {
+          int tag = input.readTag();
+          switch (tag) {
+            case 0:
+              done = true;
+              break;
+            case 13: {
+
+              sampleRate_ = input.readFloat();
+              break;
+            }
+            case 16: {
+
+              numChannels_ = input.readInt64();
+              break;
+            }
+            case 24: {
+
+              lengthFrames_ = input.readInt64();
+              break;
+            }
+            case 34: {
+
+              encodedAudioString_ = input.readBytes();
+              break;
+            }
+            case 42: {
+              java.lang.String s = input.readStringRequireUtf8();
+
+              contentType_ = s;
+              break;
+            }
+            default: {
+              if (!parseUnknownField(
+                  input, unknownFields, extensionRegistry, tag)) {
+                done = true;
+              }
+              break;
+            }
+          }
+        }
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        throw e.setUnfinishedMessage(this);
+      } catch (java.io.IOException e) {
+        throw new com.google.protobuf.InvalidProtocolBufferException(
+            e).setUnfinishedMessage(this);
+      } finally {
+        this.unknownFields = unknownFields.build();
+        makeExtensionsImmutable();
+      }
+    }
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.SummaryProtos.internal_static_tensorflow_Summary_Audio_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.SummaryProtos.internal_static_tensorflow_Summary_Audio_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.Summary.Audio.class, org.tensorflow.proto.framework.Summary.Audio.Builder.class);
+    }
+
+    public static final int SAMPLE_RATE_FIELD_NUMBER = 1;
+    private float sampleRate_;
+    /**
+     * <pre>
+     * Sample rate of the audio in Hz.
+     * </pre>
+     *
+     * <code>float sample_rate = 1;</code>
+     */
+    public float getSampleRate() {
+      return sampleRate_;
+    }
+
+    public static final int NUM_CHANNELS_FIELD_NUMBER = 2;
+    private long numChannels_;
+    /**
+     * <pre>
+     * Number of channels of audio.
+     * </pre>
+     *
+     * <code>int64 num_channels = 2;</code>
+     */
+    public long getNumChannels() {
+      return numChannels_;
+    }
+
+    public static final int LENGTH_FRAMES_FIELD_NUMBER = 3;
+    private long lengthFrames_;
+    /**
+     * <pre>
+     * Length of the audio in frames (samples per channel).
+     * </pre>
+     *
+     * <code>int64 length_frames = 3;</code>
+     */
+    public long getLengthFrames() {
+      return lengthFrames_;
+    }
+
+    public static final int ENCODED_AUDIO_STRING_FIELD_NUMBER = 4;
+    private com.google.protobuf.ByteString encodedAudioString_;
+    /**
+     * <pre>
+     * Encoded audio data and its associated RFC 2045 content type (e.g.
+     * "audio/wav").
+     * </pre>
+     *
+     * <code>bytes encoded_audio_string = 4;</code>
+     */
+    public com.google.protobuf.ByteString getEncodedAudioString() {
+      return encodedAudioString_;
+    }
+
+    public static final int CONTENT_TYPE_FIELD_NUMBER = 5;
+    private volatile java.lang.Object contentType_;
+    /**
+     * <code>string content_type = 5;</code>
+     */
+    public java.lang.String getContentType() {
+      java.lang.Object ref = contentType_;
+      if (ref instanceof java.lang.String) {
+        return (java.lang.String) ref;
+      } else {
+        com.google.protobuf.ByteString bs = 
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        contentType_ = s;
+        return s;
+      }
+    }
+    /**
+     * <code>string content_type = 5;</code>
+     */
+    public com.google.protobuf.ByteString
+        getContentTypeBytes() {
+      java.lang.Object ref = contentType_;
+      if (ref instanceof java.lang.String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        contentType_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+
+    private byte memoizedIsInitialized = -1;
+    @java.lang.Override
+    public final boolean isInitialized() {
+      byte isInitialized = memoizedIsInitialized;
+      if (isInitialized == 1) return true;
+      if (isInitialized == 0) return false;
+
+      memoizedIsInitialized = 1;
+      return true;
+    }
+
+    @java.lang.Override
+    public void writeTo(com.google.protobuf.CodedOutputStream output)
+                        throws java.io.IOException {
+      if (sampleRate_ != 0F) {
+        output.writeFloat(1, sampleRate_);
+      }
+      if (numChannels_ != 0L) {
+        output.writeInt64(2, numChannels_);
+      }
+      if (lengthFrames_ != 0L) {
+        output.writeInt64(3, lengthFrames_);
+      }
+      if (!encodedAudioString_.isEmpty()) {
+        output.writeBytes(4, encodedAudioString_);
+      }
+      if (!getContentTypeBytes().isEmpty()) {
+        com.google.protobuf.GeneratedMessageV3.writeString(output, 5, contentType_);
+      }
+      unknownFields.writeTo(output);
+    }
+
+    @java.lang.Override
+    public int getSerializedSize() {
+      int size = memoizedSize;
+      if (size != -1) return size;
+
+      size = 0;
+      if (sampleRate_ != 0F) {
+        size += com.google.protobuf.CodedOutputStream
+          .computeFloatSize(1, sampleRate_);
+      }
+      if (numChannels_ != 0L) {
+        size += com.google.protobuf.CodedOutputStream
+          .computeInt64Size(2, numChannels_);
+      }
+      if (lengthFrames_ != 0L) {
+        size += com.google.protobuf.CodedOutputStream
+          .computeInt64Size(3, lengthFrames_);
+      }
+      if (!encodedAudioString_.isEmpty()) {
+        size += com.google.protobuf.CodedOutputStream
+          .computeBytesSize(4, encodedAudioString_);
+      }
+      if (!getContentTypeBytes().isEmpty()) {
+        size += com.google.protobuf.GeneratedMessageV3.computeStringSize(5, contentType_);
+      }
+      size += unknownFields.getSerializedSize();
+      memoizedSize = size;
+      return size;
+    }
+
+    @java.lang.Override
+    public boolean equals(final java.lang.Object obj) {
+      if (obj == this) {
+       return true;
+      }
+      if (!(obj instanceof org.tensorflow.proto.framework.Summary.Audio)) {
+        return super.equals(obj);
+      }
+      org.tensorflow.proto.framework.Summary.Audio other = (org.tensorflow.proto.framework.Summary.Audio) obj;
+
+      if (java.lang.Float.floatToIntBits(getSampleRate())
+          != java.lang.Float.floatToIntBits(
+              other.getSampleRate())) return false;
+      if (getNumChannels()
+          != other.getNumChannels()) return false;
+      if (getLengthFrames()
+          != other.getLengthFrames()) return false;
+      if (!getEncodedAudioString()
+          .equals(other.getEncodedAudioString())) return false;
+      if (!getContentType()
+          .equals(other.getContentType())) return false;
+      if (!unknownFields.equals(other.unknownFields)) return false;
+      return true;
+    }
+
+    @java.lang.Override
+    public int hashCode() {
+      if (memoizedHashCode != 0) {
+        return memoizedHashCode;
+      }
+      int hash = 41;
+      hash = (19 * hash) + getDescriptor().hashCode();
+      hash = (37 * hash) + SAMPLE_RATE_FIELD_NUMBER;
+      hash = (53 * hash) + java.lang.Float.floatToIntBits(
+          getSampleRate());
+      hash = (37 * hash) + NUM_CHANNELS_FIELD_NUMBER;
+      hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+          getNumChannels());
+      hash = (37 * hash) + LENGTH_FRAMES_FIELD_NUMBER;
+      hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+          getLengthFrames());
+      hash = (37 * hash) + ENCODED_AUDIO_STRING_FIELD_NUMBER;
+      hash = (53 * hash) + getEncodedAudioString().hashCode();
+      hash = (37 * hash) + CONTENT_TYPE_FIELD_NUMBER;
+      hash = (53 * hash) + getContentType().hashCode();
+      hash = (29 * hash) + unknownFields.hashCode();
+      memoizedHashCode = hash;
+      return hash;
+    }
+
+    public static org.tensorflow.proto.framework.Summary.Audio parseFrom(
+        java.nio.ByteBuffer data)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data);
+    }
+    public static org.tensorflow.proto.framework.Summary.Audio parseFrom(
+        java.nio.ByteBuffer data,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.Summary.Audio parseFrom(
+        com.google.protobuf.ByteString data)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data);
+    }
+    public static org.tensorflow.proto.framework.Summary.Audio parseFrom(
+        com.google.protobuf.ByteString data,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.Summary.Audio parseFrom(byte[] data)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data);
+    }
+    public static org.tensorflow.proto.framework.Summary.Audio parseFrom(
+        byte[] data,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.Summary.Audio parseFrom(java.io.InputStream input)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input);
+    }
+    public static org.tensorflow.proto.framework.Summary.Audio parseFrom(
+        java.io.InputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.Summary.Audio parseDelimitedFrom(java.io.InputStream input)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseDelimitedWithIOException(PARSER, input);
+    }
+    public static org.tensorflow.proto.framework.Summary.Audio parseDelimitedFrom(
+        java.io.InputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.Summary.Audio parseFrom(
+        com.google.protobuf.CodedInputStream input)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input);
+    }
+    public static org.tensorflow.proto.framework.Summary.Audio parseFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input, extensionRegistry);
+    }
+
+    @java.lang.Override
+    public Builder newBuilderForType() { return newBuilder(); }
+    public static Builder newBuilder() {
+      return DEFAULT_INSTANCE.toBuilder();
+    }
+    public static Builder newBuilder(org.tensorflow.proto.framework.Summary.Audio prototype) {
+      return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+    }
+    @java.lang.Override
+    public Builder toBuilder() {
+      return this == DEFAULT_INSTANCE
+          ? new Builder() : new Builder().mergeFrom(this);
+    }
+
+    @java.lang.Override
+    protected Builder newBuilderForType(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      Builder builder = new Builder(parent);
+      return builder;
+    }
+    /**
+     * Protobuf type {@code tensorflow.Summary.Audio}
+     */
+    public static final class Builder extends
+        com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+        // @@protoc_insertion_point(builder_implements:tensorflow.Summary.Audio)
+        org.tensorflow.proto.framework.Summary.AudioOrBuilder {
+      public static final com.google.protobuf.Descriptors.Descriptor
+          getDescriptor() {
+        return org.tensorflow.proto.framework.SummaryProtos.internal_static_tensorflow_Summary_Audio_descriptor;
+      }
+
+      @java.lang.Override
+      protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+          internalGetFieldAccessorTable() {
+        return org.tensorflow.proto.framework.SummaryProtos.internal_static_tensorflow_Summary_Audio_fieldAccessorTable
+            .ensureFieldAccessorsInitialized(
+                org.tensorflow.proto.framework.Summary.Audio.class, org.tensorflow.proto.framework.Summary.Audio.Builder.class);
+      }
+
+      // Construct using org.tensorflow.proto.framework.Summary.Audio.newBuilder()
+      private Builder() {
+        maybeForceBuilderInitialization();
+      }
+
+      private Builder(
+          com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+        super(parent);
+        maybeForceBuilderInitialization();
+      }
+      private void maybeForceBuilderInitialization() {
+        if (com.google.protobuf.GeneratedMessageV3
+                .alwaysUseFieldBuilders) {
+        }
+      }
+      @java.lang.Override
+      public Builder clear() {
+        super.clear();
+        sampleRate_ = 0F;
+
+        numChannels_ = 0L;
+
+        lengthFrames_ = 0L;
+
+        encodedAudioString_ = com.google.protobuf.ByteString.EMPTY;
+
+        contentType_ = "";
+
+        return this;
+      }
+
+      @java.lang.Override
+      public com.google.protobuf.Descriptors.Descriptor
+          getDescriptorForType() {
+        return org.tensorflow.proto.framework.SummaryProtos.internal_static_tensorflow_Summary_Audio_descriptor;
+      }
+
+      @java.lang.Override
+      public org.tensorflow.proto.framework.Summary.Audio getDefaultInstanceForType() {
+        return org.tensorflow.proto.framework.Summary.Audio.getDefaultInstance();
+      }
+
+      @java.lang.Override
+      public org.tensorflow.proto.framework.Summary.Audio build() {
+        org.tensorflow.proto.framework.Summary.Audio result = buildPartial();
+        if (!result.isInitialized()) {
+          throw newUninitializedMessageException(result);
+        }
+        return result;
+      }
+
+      @java.lang.Override
+      public org.tensorflow.proto.framework.Summary.Audio buildPartial() {
+        org.tensorflow.proto.framework.Summary.Audio result = new org.tensorflow.proto.framework.Summary.Audio(this);
+        result.sampleRate_ = sampleRate_;
+        result.numChannels_ = numChannels_;
+        result.lengthFrames_ = lengthFrames_;
+        result.encodedAudioString_ = encodedAudioString_;
+        result.contentType_ = contentType_;
+        onBuilt();
+        return result;
+      }
+
+      @java.lang.Override
+      public Builder clone() {
+        return super.clone();
+      }
+      @java.lang.Override
+      public Builder setField(
+          com.google.protobuf.Descriptors.FieldDescriptor field,
+          java.lang.Object value) {
+        return super.setField(field, value);
+      }
+      @java.lang.Override
+      public Builder clearField(
+          com.google.protobuf.Descriptors.FieldDescriptor field) {
+        return super.clearField(field);
+      }
+      @java.lang.Override
+      public Builder clearOneof(
+          com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+        return super.clearOneof(oneof);
+      }
+      @java.lang.Override
+      public Builder setRepeatedField(
+          com.google.protobuf.Descriptors.FieldDescriptor field,
+          int index, java.lang.Object value) {
+        return super.setRepeatedField(field, index, value);
+      }
+      @java.lang.Override
+      public Builder addRepeatedField(
+          com.google.protobuf.Descriptors.FieldDescriptor field,
+          java.lang.Object value) {
+        return super.addRepeatedField(field, value);
+      }
+      @java.lang.Override
+      public Builder mergeFrom(com.google.protobuf.Message other) {
+        if (other instanceof org.tensorflow.proto.framework.Summary.Audio) {
+          return mergeFrom((org.tensorflow.proto.framework.Summary.Audio)other);
+        } else {
+          super.mergeFrom(other);
+          return this;
+        }
+      }
+
+      public Builder mergeFrom(org.tensorflow.proto.framework.Summary.Audio other) {
+        if (other == org.tensorflow.proto.framework.Summary.Audio.getDefaultInstance()) return this;
+        if (other.getSampleRate() != 0F) {
+          setSampleRate(other.getSampleRate());
+        }
+        if (other.getNumChannels() != 0L) {
+          setNumChannels(other.getNumChannels());
+        }
+        if (other.getLengthFrames() != 0L) {
+          setLengthFrames(other.getLengthFrames());
+        }
+        if (other.getEncodedAudioString() != com.google.protobuf.ByteString.EMPTY) {
+          setEncodedAudioString(other.getEncodedAudioString());
+        }
+        if (!other.getContentType().isEmpty()) {
+          contentType_ = other.contentType_;
+          onChanged();
+        }
+        this.mergeUnknownFields(other.unknownFields);
+        onChanged();
+        return this;
+      }
+
+      @java.lang.Override
+      public final boolean isInitialized() {
+        return true;
+      }
+
+      @java.lang.Override
+      public Builder mergeFrom(
+          com.google.protobuf.CodedInputStream input,
+          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+          throws java.io.IOException {
+        org.tensorflow.proto.framework.Summary.Audio parsedMessage = null;
+        try {
+          parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+        } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+          parsedMessage = (org.tensorflow.proto.framework.Summary.Audio) e.getUnfinishedMessage();
+          throw e.unwrapIOException();
+        } finally {
+          if (parsedMessage != null) {
+            mergeFrom(parsedMessage);
+          }
+        }
+        return this;
+      }
+
+      private float sampleRate_ ;
+      /**
+       * <pre>
+       * Sample rate of the audio in Hz.
+       * </pre>
+       *
+       * <code>float sample_rate = 1;</code>
+       */
+      public float getSampleRate() {
+        return sampleRate_;
+      }
+      /**
+       * <pre>
+       * Sample rate of the audio in Hz.
+       * </pre>
+       *
+       * <code>float sample_rate = 1;</code>
+       */
+      public Builder setSampleRate(float value) {
+        
+        sampleRate_ = value;
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * Sample rate of the audio in Hz.
+       * </pre>
+       *
+       * <code>float sample_rate = 1;</code>
+       */
+      public Builder clearSampleRate() {
+        
+        sampleRate_ = 0F;
+        onChanged();
+        return this;
+      }
+
+      private long numChannels_ ;
+      /**
+       * <pre>
+       * Number of channels of audio.
+       * </pre>
+       *
+       * <code>int64 num_channels = 2;</code>
+       */
+      public long getNumChannels() {
+        return numChannels_;
+      }
+      /**
+       * <pre>
+       * Number of channels of audio.
+       * </pre>
+       *
+       * <code>int64 num_channels = 2;</code>
+       */
+      public Builder setNumChannels(long value) {
+        
+        numChannels_ = value;
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * Number of channels of audio.
+       * </pre>
+       *
+       * <code>int64 num_channels = 2;</code>
+       */
+      public Builder clearNumChannels() {
+        
+        numChannels_ = 0L;
+        onChanged();
+        return this;
+      }
+
+      private long lengthFrames_ ;
+      /**
+       * <pre>
+       * Length of the audio in frames (samples per channel).
+       * </pre>
+       *
+       * <code>int64 length_frames = 3;</code>
+       */
+      public long getLengthFrames() {
+        return lengthFrames_;
+      }
+      /**
+       * <pre>
+       * Length of the audio in frames (samples per channel).
+       * </pre>
+       *
+       * <code>int64 length_frames = 3;</code>
+       */
+      public Builder setLengthFrames(long value) {
+        
+        lengthFrames_ = value;
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * Length of the audio in frames (samples per channel).
+       * </pre>
+       *
+       * <code>int64 length_frames = 3;</code>
+       */
+      public Builder clearLengthFrames() {
+        
+        lengthFrames_ = 0L;
+        onChanged();
+        return this;
+      }
+
+      private com.google.protobuf.ByteString encodedAudioString_ = com.google.protobuf.ByteString.EMPTY;
+      /**
+       * <pre>
+       * Encoded audio data and its associated RFC 2045 content type (e.g.
+       * "audio/wav").
+       * </pre>
+       *
+       * <code>bytes encoded_audio_string = 4;</code>
+       */
+      public com.google.protobuf.ByteString getEncodedAudioString() {
+        return encodedAudioString_;
+      }
+      /**
+       * <pre>
+       * Encoded audio data and its associated RFC 2045 content type (e.g.
+       * "audio/wav").
+       * </pre>
+       *
+       * <code>bytes encoded_audio_string = 4;</code>
+       */
+      public Builder setEncodedAudioString(com.google.protobuf.ByteString value) {
+        if (value == null) {
+    throw new NullPointerException();
+  }
+  
+        encodedAudioString_ = value;
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * Encoded audio data and its associated RFC 2045 content type (e.g.
+       * "audio/wav").
+       * </pre>
+       *
+       * <code>bytes encoded_audio_string = 4;</code>
+       */
+      public Builder clearEncodedAudioString() {
+        
+        encodedAudioString_ = getDefaultInstance().getEncodedAudioString();
+        onChanged();
+        return this;
+      }
+
+      private java.lang.Object contentType_ = "";
+      /**
+       * <code>string content_type = 5;</code>
+       */
+      public java.lang.String getContentType() {
+        java.lang.Object ref = contentType_;
+        if (!(ref instanceof java.lang.String)) {
+          com.google.protobuf.ByteString bs =
+              (com.google.protobuf.ByteString) ref;
+          java.lang.String s = bs.toStringUtf8();
+          contentType_ = s;
+          return s;
+        } else {
+          return (java.lang.String) ref;
+        }
+      }
+      /**
+       * <code>string content_type = 5;</code>
+       */
+      public com.google.protobuf.ByteString
+          getContentTypeBytes() {
+        java.lang.Object ref = contentType_;
+        if (ref instanceof String) {
+          com.google.protobuf.ByteString b = 
+              com.google.protobuf.ByteString.copyFromUtf8(
+                  (java.lang.String) ref);
+          contentType_ = b;
+          return b;
+        } else {
+          return (com.google.protobuf.ByteString) ref;
+        }
+      }
+      /**
+       * <code>string content_type = 5;</code>
+       */
+      public Builder setContentType(
+          java.lang.String value) {
+        if (value == null) {
+    throw new NullPointerException();
+  }
+  
+        contentType_ = value;
+        onChanged();
+        return this;
+      }
+      /**
+       * <code>string content_type = 5;</code>
+       */
+      public Builder clearContentType() {
+        
+        contentType_ = getDefaultInstance().getContentType();
+        onChanged();
+        return this;
+      }
+      /**
+       * <code>string content_type = 5;</code>
+       */
+      public Builder setContentTypeBytes(
+          com.google.protobuf.ByteString value) {
+        if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+        
+        contentType_ = value;
+        onChanged();
+        return this;
+      }
+      @java.lang.Override
+      public final Builder setUnknownFields(
+          final com.google.protobuf.UnknownFieldSet unknownFields) {
+        return super.setUnknownFields(unknownFields);
+      }
+
+      @java.lang.Override
+      public final Builder mergeUnknownFields(
+          final com.google.protobuf.UnknownFieldSet unknownFields) {
+        return super.mergeUnknownFields(unknownFields);
+      }
+
+
+      // @@protoc_insertion_point(builder_scope:tensorflow.Summary.Audio)
+    }
+
+    // @@protoc_insertion_point(class_scope:tensorflow.Summary.Audio)
+    private static final org.tensorflow.proto.framework.Summary.Audio DEFAULT_INSTANCE;
+    static {
+      DEFAULT_INSTANCE = new org.tensorflow.proto.framework.Summary.Audio();
+    }
+
+    public static org.tensorflow.proto.framework.Summary.Audio getDefaultInstance() {
+      return DEFAULT_INSTANCE;
+    }
+
+    private static final com.google.protobuf.Parser<Audio>
+        PARSER = new com.google.protobuf.AbstractParser<Audio>() {
+      @java.lang.Override
+      public Audio parsePartialFrom(
+          com.google.protobuf.CodedInputStream input,
+          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+          throws com.google.protobuf.InvalidProtocolBufferException {
+        return new Audio(input, extensionRegistry);
+      }
+    };
+
+    public static com.google.protobuf.Parser<Audio> parser() {
+      return PARSER;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Parser<Audio> getParserForType() {
+      return PARSER;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.Summary.Audio getDefaultInstanceForType() {
+      return DEFAULT_INSTANCE;
+    }
+
+  }
+
+  public interface ValueOrBuilder extends
+      // @@protoc_insertion_point(interface_extends:tensorflow.Summary.Value)
+      com.google.protobuf.MessageOrBuilder {
+
+    /**
+     * <pre>
+     * This field is deprecated and will not be set.
+     * </pre>
+     *
+     * <code>string node_name = 7;</code>
+     */
+    java.lang.String getNodeName();
+    /**
+     * <pre>
+     * This field is deprecated and will not be set.
+     * </pre>
+     *
+     * <code>string node_name = 7;</code>
+     */
+    com.google.protobuf.ByteString
+        getNodeNameBytes();
+
+    /**
+     * <pre>
+     * Tag name for the data. Used by TensorBoard plugins to organize data. Tags
+     * are often organized by scope (which contains slashes to convey
+     * hierarchy). For example: foo/bar/0
+     * </pre>
+     *
+     * <code>string tag = 1;</code>
+     */
+    java.lang.String getTag();
+    /**
+     * <pre>
+     * Tag name for the data. Used by TensorBoard plugins to organize data. Tags
+     * are often organized by scope (which contains slashes to convey
+     * hierarchy). For example: foo/bar/0
+     * </pre>
+     *
+     * <code>string tag = 1;</code>
+     */
+    com.google.protobuf.ByteString
+        getTagBytes();
+
+    /**
+     * <pre>
+     * Contains metadata on the summary value such as which plugins may use it.
+     * Take note that many summary values may lack a metadata field. This is
+     * because the FileWriter only keeps a metadata object on the first summary
+     * value with a certain tag for each tag. TensorBoard then remembers which
+     * tags are associated with which plugins. This saves space.
+     * </pre>
+     *
+     * <code>.tensorflow.SummaryMetadata metadata = 9;</code>
+     */
+    boolean hasMetadata();
+    /**
+     * <pre>
+     * Contains metadata on the summary value such as which plugins may use it.
+     * Take note that many summary values may lack a metadata field. This is
+     * because the FileWriter only keeps a metadata object on the first summary
+     * value with a certain tag for each tag. TensorBoard then remembers which
+     * tags are associated with which plugins. This saves space.
+     * </pre>
+     *
+     * <code>.tensorflow.SummaryMetadata metadata = 9;</code>
+     */
+    org.tensorflow.proto.framework.SummaryMetadata getMetadata();
+    /**
+     * <pre>
+     * Contains metadata on the summary value such as which plugins may use it.
+     * Take note that many summary values may lack a metadata field. This is
+     * because the FileWriter only keeps a metadata object on the first summary
+     * value with a certain tag for each tag. TensorBoard then remembers which
+     * tags are associated with which plugins. This saves space.
+     * </pre>
+     *
+     * <code>.tensorflow.SummaryMetadata metadata = 9;</code>
+     */
+    org.tensorflow.proto.framework.SummaryMetadataOrBuilder getMetadataOrBuilder();
+
+    /**
+     * <code>float simple_value = 2;</code>
+     */
+    float getSimpleValue();
+
+    /**
+     * <code>bytes obsolete_old_style_histogram = 3;</code>
+     */
+    com.google.protobuf.ByteString getObsoleteOldStyleHistogram();
+
+    /**
+     * <code>.tensorflow.Summary.Image image = 4;</code>
+     */
+    boolean hasImage();
+    /**
+     * <code>.tensorflow.Summary.Image image = 4;</code>
+     */
+    org.tensorflow.proto.framework.Summary.Image getImage();
+    /**
+     * <code>.tensorflow.Summary.Image image = 4;</code>
+     */
+    org.tensorflow.proto.framework.Summary.ImageOrBuilder getImageOrBuilder();
+
+    /**
+     * <code>.tensorflow.HistogramProto histo = 5;</code>
+     */
+    boolean hasHisto();
+    /**
+     * <code>.tensorflow.HistogramProto histo = 5;</code>
+     */
+    org.tensorflow.proto.framework.HistogramProto getHisto();
+    /**
+     * <code>.tensorflow.HistogramProto histo = 5;</code>
+     */
+    org.tensorflow.proto.framework.HistogramProtoOrBuilder getHistoOrBuilder();
+
+    /**
+     * <code>.tensorflow.Summary.Audio audio = 6;</code>
+     */
+    boolean hasAudio();
+    /**
+     * <code>.tensorflow.Summary.Audio audio = 6;</code>
+     */
+    org.tensorflow.proto.framework.Summary.Audio getAudio();
+    /**
+     * <code>.tensorflow.Summary.Audio audio = 6;</code>
+     */
+    org.tensorflow.proto.framework.Summary.AudioOrBuilder getAudioOrBuilder();
+
+    /**
+     * <code>.tensorflow.TensorProto tensor = 8;</code>
+     */
+    boolean hasTensor();
+    /**
+     * <code>.tensorflow.TensorProto tensor = 8;</code>
+     */
+    org.tensorflow.proto.framework.TensorProto getTensor();
+    /**
+     * <code>.tensorflow.TensorProto tensor = 8;</code>
+     */
+    org.tensorflow.proto.framework.TensorProtoOrBuilder getTensorOrBuilder();
+
+    public org.tensorflow.proto.framework.Summary.Value.ValueCase getValueCase();
+  }
+  /**
+   * Protobuf type {@code tensorflow.Summary.Value}
+   */
+  public  static final class Value extends
+      com.google.protobuf.GeneratedMessageV3 implements
+      // @@protoc_insertion_point(message_implements:tensorflow.Summary.Value)
+      ValueOrBuilder {
+  private static final long serialVersionUID = 0L;
+    // Use Value.newBuilder() to construct.
+    private Value(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+      super(builder);
+    }
+    private Value() {
+      nodeName_ = "";
+      tag_ = "";
+    }
+
+    @java.lang.Override
+    @SuppressWarnings({"unused"})
+    protected java.lang.Object newInstance(
+        UnusedPrivateParameter unused) {
+      return new Value();
+    }
+
+    @java.lang.Override
+    public final com.google.protobuf.UnknownFieldSet
+    getUnknownFields() {
+      return this.unknownFields;
+    }
+    private Value(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      this();
+      if (extensionRegistry == null) {
+        throw new java.lang.NullPointerException();
+      }
+      com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+          com.google.protobuf.UnknownFieldSet.newBuilder();
+      try {
+        boolean done = false;
+        while (!done) {
+          int tag = input.readTag();
+          switch (tag) {
+            case 0:
+              done = true;
+              break;
+            case 10: {
+              java.lang.String s = input.readStringRequireUtf8();
+
+              tag_ = s;
+              break;
+            }
+            case 21: {
+              valueCase_ = 2;
+              value_ = input.readFloat();
+              break;
+            }
+            case 26: {
+              valueCase_ = 3;
+              value_ = input.readBytes();
+              break;
+            }
+            case 34: {
+              org.tensorflow.proto.framework.Summary.Image.Builder subBuilder = null;
+              if (valueCase_ == 4) {
+                subBuilder = ((org.tensorflow.proto.framework.Summary.Image) value_).toBuilder();
+              }
+              value_ =
+                  input.readMessage(org.tensorflow.proto.framework.Summary.Image.parser(), extensionRegistry);
+              if (subBuilder != null) {
+                subBuilder.mergeFrom((org.tensorflow.proto.framework.Summary.Image) value_);
+                value_ = subBuilder.buildPartial();
+              }
+              valueCase_ = 4;
+              break;
+            }
+            case 42: {
+              org.tensorflow.proto.framework.HistogramProto.Builder subBuilder = null;
+              if (valueCase_ == 5) {
+                subBuilder = ((org.tensorflow.proto.framework.HistogramProto) value_).toBuilder();
+              }
+              value_ =
+                  input.readMessage(org.tensorflow.proto.framework.HistogramProto.parser(), extensionRegistry);
+              if (subBuilder != null) {
+                subBuilder.mergeFrom((org.tensorflow.proto.framework.HistogramProto) value_);
+                value_ = subBuilder.buildPartial();
+              }
+              valueCase_ = 5;
+              break;
+            }
+            case 50: {
+              org.tensorflow.proto.framework.Summary.Audio.Builder subBuilder = null;
+              if (valueCase_ == 6) {
+                subBuilder = ((org.tensorflow.proto.framework.Summary.Audio) value_).toBuilder();
+              }
+              value_ =
+                  input.readMessage(org.tensorflow.proto.framework.Summary.Audio.parser(), extensionRegistry);
+              if (subBuilder != null) {
+                subBuilder.mergeFrom((org.tensorflow.proto.framework.Summary.Audio) value_);
+                value_ = subBuilder.buildPartial();
+              }
+              valueCase_ = 6;
+              break;
+            }
+            case 58: {
+              java.lang.String s = input.readStringRequireUtf8();
+
+              nodeName_ = s;
+              break;
+            }
+            case 66: {
+              org.tensorflow.proto.framework.TensorProto.Builder subBuilder = null;
+              if (valueCase_ == 8) {
+                subBuilder = ((org.tensorflow.proto.framework.TensorProto) value_).toBuilder();
+              }
+              value_ =
+                  input.readMessage(org.tensorflow.proto.framework.TensorProto.parser(), extensionRegistry);
+              if (subBuilder != null) {
+                subBuilder.mergeFrom((org.tensorflow.proto.framework.TensorProto) value_);
+                value_ = subBuilder.buildPartial();
+              }
+              valueCase_ = 8;
+              break;
+            }
+            case 74: {
+              org.tensorflow.proto.framework.SummaryMetadata.Builder subBuilder = null;
+              if (metadata_ != null) {
+                subBuilder = metadata_.toBuilder();
+              }
+              metadata_ = input.readMessage(org.tensorflow.proto.framework.SummaryMetadata.parser(), extensionRegistry);
+              if (subBuilder != null) {
+                subBuilder.mergeFrom(metadata_);
+                metadata_ = subBuilder.buildPartial();
+              }
+
+              break;
+            }
+            default: {
+              if (!parseUnknownField(
+                  input, unknownFields, extensionRegistry, tag)) {
+                done = true;
+              }
+              break;
+            }
+          }
+        }
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        throw e.setUnfinishedMessage(this);
+      } catch (java.io.IOException e) {
+        throw new com.google.protobuf.InvalidProtocolBufferException(
+            e).setUnfinishedMessage(this);
+      } finally {
+        this.unknownFields = unknownFields.build();
+        makeExtensionsImmutable();
+      }
+    }
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.SummaryProtos.internal_static_tensorflow_Summary_Value_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.SummaryProtos.internal_static_tensorflow_Summary_Value_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.Summary.Value.class, org.tensorflow.proto.framework.Summary.Value.Builder.class);
+    }
+
+    private int valueCase_ = 0;
+    private java.lang.Object value_;
+    public enum ValueCase
+        implements com.google.protobuf.Internal.EnumLite {
+      SIMPLE_VALUE(2),
+      OBSOLETE_OLD_STYLE_HISTOGRAM(3),
+      IMAGE(4),
+      HISTO(5),
+      AUDIO(6),
+      TENSOR(8),
+      VALUE_NOT_SET(0);
+      private final int value;
+      private ValueCase(int value) {
+        this.value = value;
+      }
+      /**
+       * @deprecated Use {@link #forNumber(int)} instead.
+       */
+      @java.lang.Deprecated
+      public static ValueCase valueOf(int value) {
+        return forNumber(value);
+      }
+
+      public static ValueCase forNumber(int value) {
+        switch (value) {
+          case 2: return SIMPLE_VALUE;
+          case 3: return OBSOLETE_OLD_STYLE_HISTOGRAM;
+          case 4: return IMAGE;
+          case 5: return HISTO;
+          case 6: return AUDIO;
+          case 8: return TENSOR;
+          case 0: return VALUE_NOT_SET;
+          default: return null;
+        }
+      }
+      public int getNumber() {
+        return this.value;
+      }
+    };
+
+    public ValueCase
+    getValueCase() {
+      return ValueCase.forNumber(
+          valueCase_);
+    }
+
+    public static final int NODE_NAME_FIELD_NUMBER = 7;
+    private volatile java.lang.Object nodeName_;
+    /**
+     * <pre>
+     * This field is deprecated and will not be set.
+     * </pre>
+     *
+     * <code>string node_name = 7;</code>
+     */
+    public java.lang.String getNodeName() {
+      java.lang.Object ref = nodeName_;
+      if (ref instanceof java.lang.String) {
+        return (java.lang.String) ref;
+      } else {
+        com.google.protobuf.ByteString bs = 
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        nodeName_ = s;
+        return s;
+      }
+    }
+    /**
+     * <pre>
+     * This field is deprecated and will not be set.
+     * </pre>
+     *
+     * <code>string node_name = 7;</code>
+     */
+    public com.google.protobuf.ByteString
+        getNodeNameBytes() {
+      java.lang.Object ref = nodeName_;
+      if (ref instanceof java.lang.String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        nodeName_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+
+    public static final int TAG_FIELD_NUMBER = 1;
+    private volatile java.lang.Object tag_;
+    /**
+     * <pre>
+     * Tag name for the data. Used by TensorBoard plugins to organize data. Tags
+     * are often organized by scope (which contains slashes to convey
+     * hierarchy). For example: foo/bar/0
+     * </pre>
+     *
+     * <code>string tag = 1;</code>
+     */
+    public java.lang.String getTag() {
+      java.lang.Object ref = tag_;
+      if (ref instanceof java.lang.String) {
+        return (java.lang.String) ref;
+      } else {
+        com.google.protobuf.ByteString bs = 
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        tag_ = s;
+        return s;
+      }
+    }
+    /**
+     * <pre>
+     * Tag name for the data. Used by TensorBoard plugins to organize data. Tags
+     * are often organized by scope (which contains slashes to convey
+     * hierarchy). For example: foo/bar/0
+     * </pre>
+     *
+     * <code>string tag = 1;</code>
+     */
+    public com.google.protobuf.ByteString
+        getTagBytes() {
+      java.lang.Object ref = tag_;
+      if (ref instanceof java.lang.String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        tag_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+
+    public static final int METADATA_FIELD_NUMBER = 9;
+    private org.tensorflow.proto.framework.SummaryMetadata metadata_;
+    /**
+     * <pre>
+     * Contains metadata on the summary value such as which plugins may use it.
+     * Take note that many summary values may lack a metadata field. This is
+     * because the FileWriter only keeps a metadata object on the first summary
+     * value with a certain tag for each tag. TensorBoard then remembers which
+     * tags are associated with which plugins. This saves space.
+     * </pre>
+     *
+     * <code>.tensorflow.SummaryMetadata metadata = 9;</code>
+     */
+    public boolean hasMetadata() {
+      return metadata_ != null;
+    }
+    /**
+     * <pre>
+     * Contains metadata on the summary value such as which plugins may use it.
+     * Take note that many summary values may lack a metadata field. This is
+     * because the FileWriter only keeps a metadata object on the first summary
+     * value with a certain tag for each tag. TensorBoard then remembers which
+     * tags are associated with which plugins. This saves space.
+     * </pre>
+     *
+     * <code>.tensorflow.SummaryMetadata metadata = 9;</code>
+     */
+    public org.tensorflow.proto.framework.SummaryMetadata getMetadata() {
+      return metadata_ == null ? org.tensorflow.proto.framework.SummaryMetadata.getDefaultInstance() : metadata_;
+    }
+    /**
+     * <pre>
+     * Contains metadata on the summary value such as which plugins may use it.
+     * Take note that many summary values may lack a metadata field. This is
+     * because the FileWriter only keeps a metadata object on the first summary
+     * value with a certain tag for each tag. TensorBoard then remembers which
+     * tags are associated with which plugins. This saves space.
+     * </pre>
+     *
+     * <code>.tensorflow.SummaryMetadata metadata = 9;</code>
+     */
+    public org.tensorflow.proto.framework.SummaryMetadataOrBuilder getMetadataOrBuilder() {
+      return getMetadata();
+    }
+
+    public static final int SIMPLE_VALUE_FIELD_NUMBER = 2;
+    /**
+     * <code>float simple_value = 2;</code>
+     */
+    public float getSimpleValue() {
+      if (valueCase_ == 2) {
+        return (java.lang.Float) value_;
+      }
+      return 0F;
+    }
+
+    public static final int OBSOLETE_OLD_STYLE_HISTOGRAM_FIELD_NUMBER = 3;
+    /**
+     * <code>bytes obsolete_old_style_histogram = 3;</code>
+     */
+    public com.google.protobuf.ByteString getObsoleteOldStyleHistogram() {
+      if (valueCase_ == 3) {
+        return (com.google.protobuf.ByteString) value_;
+      }
+      return com.google.protobuf.ByteString.EMPTY;
+    }
+
+    public static final int IMAGE_FIELD_NUMBER = 4;
+    /**
+     * <code>.tensorflow.Summary.Image image = 4;</code>
+     */
+    public boolean hasImage() {
+      return valueCase_ == 4;
+    }
+    /**
+     * <code>.tensorflow.Summary.Image image = 4;</code>
+     */
+    public org.tensorflow.proto.framework.Summary.Image getImage() {
+      if (valueCase_ == 4) {
+         return (org.tensorflow.proto.framework.Summary.Image) value_;
+      }
+      return org.tensorflow.proto.framework.Summary.Image.getDefaultInstance();
+    }
+    /**
+     * <code>.tensorflow.Summary.Image image = 4;</code>
+     */
+    public org.tensorflow.proto.framework.Summary.ImageOrBuilder getImageOrBuilder() {
+      if (valueCase_ == 4) {
+         return (org.tensorflow.proto.framework.Summary.Image) value_;
+      }
+      return org.tensorflow.proto.framework.Summary.Image.getDefaultInstance();
+    }
+
+    public static final int HISTO_FIELD_NUMBER = 5;
+    /**
+     * <code>.tensorflow.HistogramProto histo = 5;</code>
+     */
+    public boolean hasHisto() {
+      return valueCase_ == 5;
+    }
+    /**
+     * <code>.tensorflow.HistogramProto histo = 5;</code>
+     */
+    public org.tensorflow.proto.framework.HistogramProto getHisto() {
+      if (valueCase_ == 5) {
+         return (org.tensorflow.proto.framework.HistogramProto) value_;
+      }
+      return org.tensorflow.proto.framework.HistogramProto.getDefaultInstance();
+    }
+    /**
+     * <code>.tensorflow.HistogramProto histo = 5;</code>
+     */
+    public org.tensorflow.proto.framework.HistogramProtoOrBuilder getHistoOrBuilder() {
+      if (valueCase_ == 5) {
+         return (org.tensorflow.proto.framework.HistogramProto) value_;
+      }
+      return org.tensorflow.proto.framework.HistogramProto.getDefaultInstance();
+    }
+
+    public static final int AUDIO_FIELD_NUMBER = 6;
+    /**
+     * <code>.tensorflow.Summary.Audio audio = 6;</code>
+     */
+    public boolean hasAudio() {
+      return valueCase_ == 6;
+    }
+    /**
+     * <code>.tensorflow.Summary.Audio audio = 6;</code>
+     */
+    public org.tensorflow.proto.framework.Summary.Audio getAudio() {
+      if (valueCase_ == 6) {
+         return (org.tensorflow.proto.framework.Summary.Audio) value_;
+      }
+      return org.tensorflow.proto.framework.Summary.Audio.getDefaultInstance();
+    }
+    /**
+     * <code>.tensorflow.Summary.Audio audio = 6;</code>
+     */
+    public org.tensorflow.proto.framework.Summary.AudioOrBuilder getAudioOrBuilder() {
+      if (valueCase_ == 6) {
+         return (org.tensorflow.proto.framework.Summary.Audio) value_;
+      }
+      return org.tensorflow.proto.framework.Summary.Audio.getDefaultInstance();
+    }
+
+    public static final int TENSOR_FIELD_NUMBER = 8;
+    /**
+     * <code>.tensorflow.TensorProto tensor = 8;</code>
+     */
+    public boolean hasTensor() {
+      return valueCase_ == 8;
+    }
+    /**
+     * <code>.tensorflow.TensorProto tensor = 8;</code>
+     */
+    public org.tensorflow.proto.framework.TensorProto getTensor() {
+      if (valueCase_ == 8) {
+         return (org.tensorflow.proto.framework.TensorProto) value_;
+      }
+      return org.tensorflow.proto.framework.TensorProto.getDefaultInstance();
+    }
+    /**
+     * <code>.tensorflow.TensorProto tensor = 8;</code>
+     */
+    public org.tensorflow.proto.framework.TensorProtoOrBuilder getTensorOrBuilder() {
+      if (valueCase_ == 8) {
+         return (org.tensorflow.proto.framework.TensorProto) value_;
+      }
+      return org.tensorflow.proto.framework.TensorProto.getDefaultInstance();
+    }
+
+    private byte memoizedIsInitialized = -1;
+    @java.lang.Override
+    public final boolean isInitialized() {
+      byte isInitialized = memoizedIsInitialized;
+      if (isInitialized == 1) return true;
+      if (isInitialized == 0) return false;
+
+      memoizedIsInitialized = 1;
+      return true;
+    }
+
+    @java.lang.Override
+    public void writeTo(com.google.protobuf.CodedOutputStream output)
+                        throws java.io.IOException {
+      if (!getTagBytes().isEmpty()) {
+        com.google.protobuf.GeneratedMessageV3.writeString(output, 1, tag_);
+      }
+      if (valueCase_ == 2) {
+        output.writeFloat(
+            2, (float)((java.lang.Float) value_));
+      }
+      if (valueCase_ == 3) {
+        output.writeBytes(
+            3, (com.google.protobuf.ByteString) value_);
+      }
+      if (valueCase_ == 4) {
+        output.writeMessage(4, (org.tensorflow.proto.framework.Summary.Image) value_);
+      }
+      if (valueCase_ == 5) {
+        output.writeMessage(5, (org.tensorflow.proto.framework.HistogramProto) value_);
+      }
+      if (valueCase_ == 6) {
+        output.writeMessage(6, (org.tensorflow.proto.framework.Summary.Audio) value_);
+      }
+      if (!getNodeNameBytes().isEmpty()) {
+        com.google.protobuf.GeneratedMessageV3.writeString(output, 7, nodeName_);
+      }
+      if (valueCase_ == 8) {
+        output.writeMessage(8, (org.tensorflow.proto.framework.TensorProto) value_);
+      }
+      if (metadata_ != null) {
+        output.writeMessage(9, getMetadata());
+      }
+      unknownFields.writeTo(output);
+    }
+
+    @java.lang.Override
+    public int getSerializedSize() {
+      int size = memoizedSize;
+      if (size != -1) return size;
+
+      size = 0;
+      if (!getTagBytes().isEmpty()) {
+        size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, tag_);
+      }
+      if (valueCase_ == 2) {
+        size += com.google.protobuf.CodedOutputStream
+          .computeFloatSize(
+              2, (float)((java.lang.Float) value_));
+      }
+      if (valueCase_ == 3) {
+        size += com.google.protobuf.CodedOutputStream
+          .computeBytesSize(
+              3, (com.google.protobuf.ByteString) value_);
+      }
+      if (valueCase_ == 4) {
+        size += com.google.protobuf.CodedOutputStream
+          .computeMessageSize(4, (org.tensorflow.proto.framework.Summary.Image) value_);
+      }
+      if (valueCase_ == 5) {
+        size += com.google.protobuf.CodedOutputStream
+          .computeMessageSize(5, (org.tensorflow.proto.framework.HistogramProto) value_);
+      }
+      if (valueCase_ == 6) {
+        size += com.google.protobuf.CodedOutputStream
+          .computeMessageSize(6, (org.tensorflow.proto.framework.Summary.Audio) value_);
+      }
+      if (!getNodeNameBytes().isEmpty()) {
+        size += com.google.protobuf.GeneratedMessageV3.computeStringSize(7, nodeName_);
+      }
+      if (valueCase_ == 8) {
+        size += com.google.protobuf.CodedOutputStream
+          .computeMessageSize(8, (org.tensorflow.proto.framework.TensorProto) value_);
+      }
+      if (metadata_ != null) {
+        size += com.google.protobuf.CodedOutputStream
+          .computeMessageSize(9, getMetadata());
+      }
+      size += unknownFields.getSerializedSize();
+      memoizedSize = size;
+      return size;
+    }
+
+    @java.lang.Override
+    public boolean equals(final java.lang.Object obj) {
+      if (obj == this) {
+       return true;
+      }
+      if (!(obj instanceof org.tensorflow.proto.framework.Summary.Value)) {
+        return super.equals(obj);
+      }
+      org.tensorflow.proto.framework.Summary.Value other = (org.tensorflow.proto.framework.Summary.Value) obj;
+
+      if (!getNodeName()
+          .equals(other.getNodeName())) return false;
+      if (!getTag()
+          .equals(other.getTag())) return false;
+      if (hasMetadata() != other.hasMetadata()) return false;
+      if (hasMetadata()) {
+        if (!getMetadata()
+            .equals(other.getMetadata())) return false;
+      }
+      if (!getValueCase().equals(other.getValueCase())) return false;
+      switch (valueCase_) {
+        case 2:
+          if (java.lang.Float.floatToIntBits(getSimpleValue())
+              != java.lang.Float.floatToIntBits(
+                  other.getSimpleValue())) return false;
+          break;
+        case 3:
+          if (!getObsoleteOldStyleHistogram()
+              .equals(other.getObsoleteOldStyleHistogram())) return false;
+          break;
+        case 4:
+          if (!getImage()
+              .equals(other.getImage())) return false;
+          break;
+        case 5:
+          if (!getHisto()
+              .equals(other.getHisto())) return false;
+          break;
+        case 6:
+          if (!getAudio()
+              .equals(other.getAudio())) return false;
+          break;
+        case 8:
+          if (!getTensor()
+              .equals(other.getTensor())) return false;
+          break;
+        case 0:
+        default:
+      }
+      if (!unknownFields.equals(other.unknownFields)) return false;
+      return true;
+    }
+
+    @java.lang.Override
+    public int hashCode() {
+      if (memoizedHashCode != 0) {
+        return memoizedHashCode;
+      }
+      int hash = 41;
+      hash = (19 * hash) + getDescriptor().hashCode();
+      hash = (37 * hash) + NODE_NAME_FIELD_NUMBER;
+      hash = (53 * hash) + getNodeName().hashCode();
+      hash = (37 * hash) + TAG_FIELD_NUMBER;
+      hash = (53 * hash) + getTag().hashCode();
+      if (hasMetadata()) {
+        hash = (37 * hash) + METADATA_FIELD_NUMBER;
+        hash = (53 * hash) + getMetadata().hashCode();
+      }
+      switch (valueCase_) {
+        case 2:
+          hash = (37 * hash) + SIMPLE_VALUE_FIELD_NUMBER;
+          hash = (53 * hash) + java.lang.Float.floatToIntBits(
+              getSimpleValue());
+          break;
+        case 3:
+          hash = (37 * hash) + OBSOLETE_OLD_STYLE_HISTOGRAM_FIELD_NUMBER;
+          hash = (53 * hash) + getObsoleteOldStyleHistogram().hashCode();
+          break;
+        case 4:
+          hash = (37 * hash) + IMAGE_FIELD_NUMBER;
+          hash = (53 * hash) + getImage().hashCode();
+          break;
+        case 5:
+          hash = (37 * hash) + HISTO_FIELD_NUMBER;
+          hash = (53 * hash) + getHisto().hashCode();
+          break;
+        case 6:
+          hash = (37 * hash) + AUDIO_FIELD_NUMBER;
+          hash = (53 * hash) + getAudio().hashCode();
+          break;
+        case 8:
+          hash = (37 * hash) + TENSOR_FIELD_NUMBER;
+          hash = (53 * hash) + getTensor().hashCode();
+          break;
+        case 0:
+        default:
+      }
+      hash = (29 * hash) + unknownFields.hashCode();
+      memoizedHashCode = hash;
+      return hash;
+    }
+
+    public static org.tensorflow.proto.framework.Summary.Value parseFrom(
+        java.nio.ByteBuffer data)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data);
+    }
+    public static org.tensorflow.proto.framework.Summary.Value parseFrom(
+        java.nio.ByteBuffer data,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.Summary.Value parseFrom(
+        com.google.protobuf.ByteString data)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data);
+    }
+    public static org.tensorflow.proto.framework.Summary.Value parseFrom(
+        com.google.protobuf.ByteString data,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.Summary.Value parseFrom(byte[] data)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data);
+    }
+    public static org.tensorflow.proto.framework.Summary.Value parseFrom(
+        byte[] data,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.Summary.Value parseFrom(java.io.InputStream input)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input);
+    }
+    public static org.tensorflow.proto.framework.Summary.Value parseFrom(
+        java.io.InputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.Summary.Value parseDelimitedFrom(java.io.InputStream input)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseDelimitedWithIOException(PARSER, input);
+    }
+    public static org.tensorflow.proto.framework.Summary.Value parseDelimitedFrom(
+        java.io.InputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.Summary.Value parseFrom(
+        com.google.protobuf.CodedInputStream input)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input);
+    }
+    public static org.tensorflow.proto.framework.Summary.Value parseFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input, extensionRegistry);
+    }
+
+    @java.lang.Override
+    public Builder newBuilderForType() { return newBuilder(); }
+    public static Builder newBuilder() {
+      return DEFAULT_INSTANCE.toBuilder();
+    }
+    public static Builder newBuilder(org.tensorflow.proto.framework.Summary.Value prototype) {
+      return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+    }
+    @java.lang.Override
+    public Builder toBuilder() {
+      return this == DEFAULT_INSTANCE
+          ? new Builder() : new Builder().mergeFrom(this);
+    }
+
+    @java.lang.Override
+    protected Builder newBuilderForType(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      Builder builder = new Builder(parent);
+      return builder;
+    }
+    /**
+     * Protobuf type {@code tensorflow.Summary.Value}
+     */
+    public static final class Builder extends
+        com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+        // @@protoc_insertion_point(builder_implements:tensorflow.Summary.Value)
+        org.tensorflow.proto.framework.Summary.ValueOrBuilder {
+      public static final com.google.protobuf.Descriptors.Descriptor
+          getDescriptor() {
+        return org.tensorflow.proto.framework.SummaryProtos.internal_static_tensorflow_Summary_Value_descriptor;
+      }
+
+      @java.lang.Override
+      protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+          internalGetFieldAccessorTable() {
+        return org.tensorflow.proto.framework.SummaryProtos.internal_static_tensorflow_Summary_Value_fieldAccessorTable
+            .ensureFieldAccessorsInitialized(
+                org.tensorflow.proto.framework.Summary.Value.class, org.tensorflow.proto.framework.Summary.Value.Builder.class);
+      }
+
+      // Construct using org.tensorflow.proto.framework.Summary.Value.newBuilder()
+      private Builder() {
+        maybeForceBuilderInitialization();
+      }
+
+      private Builder(
+          com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+        super(parent);
+        maybeForceBuilderInitialization();
+      }
+      private void maybeForceBuilderInitialization() {
+        if (com.google.protobuf.GeneratedMessageV3
+                .alwaysUseFieldBuilders) {
+        }
+      }
+      @java.lang.Override
+      public Builder clear() {
+        super.clear();
+        nodeName_ = "";
+
+        tag_ = "";
+
+        if (metadataBuilder_ == null) {
+          metadata_ = null;
+        } else {
+          metadata_ = null;
+          metadataBuilder_ = null;
+        }
+        valueCase_ = 0;
+        value_ = null;
+        return this;
+      }
+
+      @java.lang.Override
+      public com.google.protobuf.Descriptors.Descriptor
+          getDescriptorForType() {
+        return org.tensorflow.proto.framework.SummaryProtos.internal_static_tensorflow_Summary_Value_descriptor;
+      }
+
+      @java.lang.Override
+      public org.tensorflow.proto.framework.Summary.Value getDefaultInstanceForType() {
+        return org.tensorflow.proto.framework.Summary.Value.getDefaultInstance();
+      }
+
+      @java.lang.Override
+      public org.tensorflow.proto.framework.Summary.Value build() {
+        org.tensorflow.proto.framework.Summary.Value result = buildPartial();
+        if (!result.isInitialized()) {
+          throw newUninitializedMessageException(result);
+        }
+        return result;
+      }
+
+      @java.lang.Override
+      public org.tensorflow.proto.framework.Summary.Value buildPartial() {
+        org.tensorflow.proto.framework.Summary.Value result = new org.tensorflow.proto.framework.Summary.Value(this);
+        result.nodeName_ = nodeName_;
+        result.tag_ = tag_;
+        if (metadataBuilder_ == null) {
+          result.metadata_ = metadata_;
+        } else {
+          result.metadata_ = metadataBuilder_.build();
+        }
+        if (valueCase_ == 2) {
+          result.value_ = value_;
+        }
+        if (valueCase_ == 3) {
+          result.value_ = value_;
+        }
+        if (valueCase_ == 4) {
+          if (imageBuilder_ == null) {
+            result.value_ = value_;
+          } else {
+            result.value_ = imageBuilder_.build();
+          }
+        }
+        if (valueCase_ == 5) {
+          if (histoBuilder_ == null) {
+            result.value_ = value_;
+          } else {
+            result.value_ = histoBuilder_.build();
+          }
+        }
+        if (valueCase_ == 6) {
+          if (audioBuilder_ == null) {
+            result.value_ = value_;
+          } else {
+            result.value_ = audioBuilder_.build();
+          }
+        }
+        if (valueCase_ == 8) {
+          if (tensorBuilder_ == null) {
+            result.value_ = value_;
+          } else {
+            result.value_ = tensorBuilder_.build();
+          }
+        }
+        result.valueCase_ = valueCase_;
+        onBuilt();
+        return result;
+      }
+
+      @java.lang.Override
+      public Builder clone() {
+        return super.clone();
+      }
+      @java.lang.Override
+      public Builder setField(
+          com.google.protobuf.Descriptors.FieldDescriptor field,
+          java.lang.Object value) {
+        return super.setField(field, value);
+      }
+      @java.lang.Override
+      public Builder clearField(
+          com.google.protobuf.Descriptors.FieldDescriptor field) {
+        return super.clearField(field);
+      }
+      @java.lang.Override
+      public Builder clearOneof(
+          com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+        return super.clearOneof(oneof);
+      }
+      @java.lang.Override
+      public Builder setRepeatedField(
+          com.google.protobuf.Descriptors.FieldDescriptor field,
+          int index, java.lang.Object value) {
+        return super.setRepeatedField(field, index, value);
+      }
+      @java.lang.Override
+      public Builder addRepeatedField(
+          com.google.protobuf.Descriptors.FieldDescriptor field,
+          java.lang.Object value) {
+        return super.addRepeatedField(field, value);
+      }
+      @java.lang.Override
+      public Builder mergeFrom(com.google.protobuf.Message other) {
+        if (other instanceof org.tensorflow.proto.framework.Summary.Value) {
+          return mergeFrom((org.tensorflow.proto.framework.Summary.Value)other);
+        } else {
+          super.mergeFrom(other);
+          return this;
+        }
+      }
+
+      public Builder mergeFrom(org.tensorflow.proto.framework.Summary.Value other) {
+        if (other == org.tensorflow.proto.framework.Summary.Value.getDefaultInstance()) return this;
+        if (!other.getNodeName().isEmpty()) {
+          nodeName_ = other.nodeName_;
+          onChanged();
+        }
+        if (!other.getTag().isEmpty()) {
+          tag_ = other.tag_;
+          onChanged();
+        }
+        if (other.hasMetadata()) {
+          mergeMetadata(other.getMetadata());
+        }
+        switch (other.getValueCase()) {
+          case SIMPLE_VALUE: {
+            setSimpleValue(other.getSimpleValue());
+            break;
+          }
+          case OBSOLETE_OLD_STYLE_HISTOGRAM: {
+            setObsoleteOldStyleHistogram(other.getObsoleteOldStyleHistogram());
+            break;
+          }
+          case IMAGE: {
+            mergeImage(other.getImage());
+            break;
+          }
+          case HISTO: {
+            mergeHisto(other.getHisto());
+            break;
+          }
+          case AUDIO: {
+            mergeAudio(other.getAudio());
+            break;
+          }
+          case TENSOR: {
+            mergeTensor(other.getTensor());
+            break;
+          }
+          case VALUE_NOT_SET: {
+            break;
+          }
+        }
+        this.mergeUnknownFields(other.unknownFields);
+        onChanged();
+        return this;
+      }
+
+      @java.lang.Override
+      public final boolean isInitialized() {
+        return true;
+      }
+
+      @java.lang.Override
+      public Builder mergeFrom(
+          com.google.protobuf.CodedInputStream input,
+          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+          throws java.io.IOException {
+        org.tensorflow.proto.framework.Summary.Value parsedMessage = null;
+        try {
+          parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+        } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+          parsedMessage = (org.tensorflow.proto.framework.Summary.Value) e.getUnfinishedMessage();
+          throw e.unwrapIOException();
+        } finally {
+          if (parsedMessage != null) {
+            mergeFrom(parsedMessage);
+          }
+        }
+        return this;
+      }
+      private int valueCase_ = 0;
+      private java.lang.Object value_;
+      public ValueCase
+          getValueCase() {
+        return ValueCase.forNumber(
+            valueCase_);
+      }
+
+      public Builder clearValue() {
+        valueCase_ = 0;
+        value_ = null;
+        onChanged();
+        return this;
+      }
+
+
+      private java.lang.Object nodeName_ = "";
+      /**
+       * <pre>
+       * This field is deprecated and will not be set.
+       * </pre>
+       *
+       * <code>string node_name = 7;</code>
+       */
+      public java.lang.String getNodeName() {
+        java.lang.Object ref = nodeName_;
+        if (!(ref instanceof java.lang.String)) {
+          com.google.protobuf.ByteString bs =
+              (com.google.protobuf.ByteString) ref;
+          java.lang.String s = bs.toStringUtf8();
+          nodeName_ = s;
+          return s;
+        } else {
+          return (java.lang.String) ref;
+        }
+      }
+      /**
+       * <pre>
+       * This field is deprecated and will not be set.
+       * </pre>
+       *
+       * <code>string node_name = 7;</code>
+       */
+      public com.google.protobuf.ByteString
+          getNodeNameBytes() {
+        java.lang.Object ref = nodeName_;
+        if (ref instanceof String) {
+          com.google.protobuf.ByteString b = 
+              com.google.protobuf.ByteString.copyFromUtf8(
+                  (java.lang.String) ref);
+          nodeName_ = b;
+          return b;
+        } else {
+          return (com.google.protobuf.ByteString) ref;
+        }
+      }
+      /**
+       * <pre>
+       * This field is deprecated and will not be set.
+       * </pre>
+       *
+       * <code>string node_name = 7;</code>
+       */
+      public Builder setNodeName(
+          java.lang.String value) {
+        if (value == null) {
+    throw new NullPointerException();
+  }
+  
+        nodeName_ = value;
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * This field is deprecated and will not be set.
+       * </pre>
+       *
+       * <code>string node_name = 7;</code>
+       */
+      public Builder clearNodeName() {
+        
+        nodeName_ = getDefaultInstance().getNodeName();
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * This field is deprecated and will not be set.
+       * </pre>
+       *
+       * <code>string node_name = 7;</code>
+       */
+      public Builder setNodeNameBytes(
+          com.google.protobuf.ByteString value) {
+        if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+        
+        nodeName_ = value;
+        onChanged();
+        return this;
+      }
+
+      private java.lang.Object tag_ = "";
+      /**
+       * <pre>
+       * Tag name for the data. Used by TensorBoard plugins to organize data. Tags
+       * are often organized by scope (which contains slashes to convey
+       * hierarchy). For example: foo/bar/0
+       * </pre>
+       *
+       * <code>string tag = 1;</code>
+       */
+      public java.lang.String getTag() {
+        java.lang.Object ref = tag_;
+        if (!(ref instanceof java.lang.String)) {
+          com.google.protobuf.ByteString bs =
+              (com.google.protobuf.ByteString) ref;
+          java.lang.String s = bs.toStringUtf8();
+          tag_ = s;
+          return s;
+        } else {
+          return (java.lang.String) ref;
+        }
+      }
+      /**
+       * <pre>
+       * Tag name for the data. Used by TensorBoard plugins to organize data. Tags
+       * are often organized by scope (which contains slashes to convey
+       * hierarchy). For example: foo/bar/0
+       * </pre>
+       *
+       * <code>string tag = 1;</code>
+       */
+      public com.google.protobuf.ByteString
+          getTagBytes() {
+        java.lang.Object ref = tag_;
+        if (ref instanceof String) {
+          com.google.protobuf.ByteString b = 
+              com.google.protobuf.ByteString.copyFromUtf8(
+                  (java.lang.String) ref);
+          tag_ = b;
+          return b;
+        } else {
+          return (com.google.protobuf.ByteString) ref;
+        }
+      }
+      /**
+       * <pre>
+       * Tag name for the data. Used by TensorBoard plugins to organize data. Tags
+       * are often organized by scope (which contains slashes to convey
+       * hierarchy). For example: foo/bar/0
+       * </pre>
+       *
+       * <code>string tag = 1;</code>
+       */
+      public Builder setTag(
+          java.lang.String value) {
+        if (value == null) {
+    throw new NullPointerException();
+  }
+  
+        tag_ = value;
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * Tag name for the data. Used by TensorBoard plugins to organize data. Tags
+       * are often organized by scope (which contains slashes to convey
+       * hierarchy). For example: foo/bar/0
+       * </pre>
+       *
+       * <code>string tag = 1;</code>
+       */
+      public Builder clearTag() {
+        
+        tag_ = getDefaultInstance().getTag();
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * Tag name for the data. Used by TensorBoard plugins to organize data. Tags
+       * are often organized by scope (which contains slashes to convey
+       * hierarchy). For example: foo/bar/0
+       * </pre>
+       *
+       * <code>string tag = 1;</code>
+       */
+      public Builder setTagBytes(
+          com.google.protobuf.ByteString value) {
+        if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+        
+        tag_ = value;
+        onChanged();
+        return this;
+      }
+
+      private org.tensorflow.proto.framework.SummaryMetadata metadata_;
+      private com.google.protobuf.SingleFieldBuilderV3<
+          org.tensorflow.proto.framework.SummaryMetadata, org.tensorflow.proto.framework.SummaryMetadata.Builder, org.tensorflow.proto.framework.SummaryMetadataOrBuilder> metadataBuilder_;
+      /**
+       * <pre>
+       * Contains metadata on the summary value such as which plugins may use it.
+       * Take note that many summary values may lack a metadata field. This is
+       * because the FileWriter only keeps a metadata object on the first summary
+       * value with a certain tag for each tag. TensorBoard then remembers which
+       * tags are associated with which plugins. This saves space.
+       * </pre>
+       *
+       * <code>.tensorflow.SummaryMetadata metadata = 9;</code>
+       */
+      public boolean hasMetadata() {
+        return metadataBuilder_ != null || metadata_ != null;
+      }
+      /**
+       * <pre>
+       * Contains metadata on the summary value such as which plugins may use it.
+       * Take note that many summary values may lack a metadata field. This is
+       * because the FileWriter only keeps a metadata object on the first summary
+       * value with a certain tag for each tag. TensorBoard then remembers which
+       * tags are associated with which plugins. This saves space.
+       * </pre>
+       *
+       * <code>.tensorflow.SummaryMetadata metadata = 9;</code>
+       */
+      public org.tensorflow.proto.framework.SummaryMetadata getMetadata() {
+        if (metadataBuilder_ == null) {
+          return metadata_ == null ? org.tensorflow.proto.framework.SummaryMetadata.getDefaultInstance() : metadata_;
+        } else {
+          return metadataBuilder_.getMessage();
+        }
+      }
+      /**
+       * <pre>
+       * Contains metadata on the summary value such as which plugins may use it.
+       * Take note that many summary values may lack a metadata field. This is
+       * because the FileWriter only keeps a metadata object on the first summary
+       * value with a certain tag for each tag. TensorBoard then remembers which
+       * tags are associated with which plugins. This saves space.
+       * </pre>
+       *
+       * <code>.tensorflow.SummaryMetadata metadata = 9;</code>
+       */
+      public Builder setMetadata(org.tensorflow.proto.framework.SummaryMetadata value) {
+        if (metadataBuilder_ == null) {
+          if (value == null) {
+            throw new NullPointerException();
+          }
+          metadata_ = value;
+          onChanged();
+        } else {
+          metadataBuilder_.setMessage(value);
+        }
+
+        return this;
+      }
+      /**
+       * <pre>
+       * Contains metadata on the summary value such as which plugins may use it.
+       * Take note that many summary values may lack a metadata field. This is
+       * because the FileWriter only keeps a metadata object on the first summary
+       * value with a certain tag for each tag. TensorBoard then remembers which
+       * tags are associated with which plugins. This saves space.
+       * </pre>
+       *
+       * <code>.tensorflow.SummaryMetadata metadata = 9;</code>
+       */
+      public Builder setMetadata(
+          org.tensorflow.proto.framework.SummaryMetadata.Builder builderForValue) {
+        if (metadataBuilder_ == null) {
+          metadata_ = builderForValue.build();
+          onChanged();
+        } else {
+          metadataBuilder_.setMessage(builderForValue.build());
+        }
+
+        return this;
+      }
+      /**
+       * <pre>
+       * Contains metadata on the summary value such as which plugins may use it.
+       * Take note that many summary values may lack a metadata field. This is
+       * because the FileWriter only keeps a metadata object on the first summary
+       * value with a certain tag for each tag. TensorBoard then remembers which
+       * tags are associated with which plugins. This saves space.
+       * </pre>
+       *
+       * <code>.tensorflow.SummaryMetadata metadata = 9;</code>
+       */
+      public Builder mergeMetadata(org.tensorflow.proto.framework.SummaryMetadata value) {
+        if (metadataBuilder_ == null) {
+          if (metadata_ != null) {
+            metadata_ =
+              org.tensorflow.proto.framework.SummaryMetadata.newBuilder(metadata_).mergeFrom(value).buildPartial();
+          } else {
+            metadata_ = value;
+          }
+          onChanged();
+        } else {
+          metadataBuilder_.mergeFrom(value);
+        }
+
+        return this;
+      }
+      /**
+       * <pre>
+       * Contains metadata on the summary value such as which plugins may use it.
+       * Take note that many summary values may lack a metadata field. This is
+       * because the FileWriter only keeps a metadata object on the first summary
+       * value with a certain tag for each tag. TensorBoard then remembers which
+       * tags are associated with which plugins. This saves space.
+       * </pre>
+       *
+       * <code>.tensorflow.SummaryMetadata metadata = 9;</code>
+       */
+      public Builder clearMetadata() {
+        if (metadataBuilder_ == null) {
+          metadata_ = null;
+          onChanged();
+        } else {
+          metadata_ = null;
+          metadataBuilder_ = null;
+        }
+
+        return this;
+      }
+      /**
+       * <pre>
+       * Contains metadata on the summary value such as which plugins may use it.
+       * Take note that many summary values may lack a metadata field. This is
+       * because the FileWriter only keeps a metadata object on the first summary
+       * value with a certain tag for each tag. TensorBoard then remembers which
+       * tags are associated with which plugins. This saves space.
+       * </pre>
+       *
+       * <code>.tensorflow.SummaryMetadata metadata = 9;</code>
+       */
+      public org.tensorflow.proto.framework.SummaryMetadata.Builder getMetadataBuilder() {
+        
+        onChanged();
+        return getMetadataFieldBuilder().getBuilder();
+      }
+      /**
+       * <pre>
+       * Contains metadata on the summary value such as which plugins may use it.
+       * Take note that many summary values may lack a metadata field. This is
+       * because the FileWriter only keeps a metadata object on the first summary
+       * value with a certain tag for each tag. TensorBoard then remembers which
+       * tags are associated with which plugins. This saves space.
+       * </pre>
+       *
+       * <code>.tensorflow.SummaryMetadata metadata = 9;</code>
+       */
+      public org.tensorflow.proto.framework.SummaryMetadataOrBuilder getMetadataOrBuilder() {
+        if (metadataBuilder_ != null) {
+          return metadataBuilder_.getMessageOrBuilder();
+        } else {
+          return metadata_ == null ?
+              org.tensorflow.proto.framework.SummaryMetadata.getDefaultInstance() : metadata_;
+        }
+      }
+      /**
+       * <pre>
+       * Contains metadata on the summary value such as which plugins may use it.
+       * Take note that many summary values may lack a metadata field. This is
+       * because the FileWriter only keeps a metadata object on the first summary
+       * value with a certain tag for each tag. TensorBoard then remembers which
+       * tags are associated with which plugins. This saves space.
+       * </pre>
+       *
+       * <code>.tensorflow.SummaryMetadata metadata = 9;</code>
+       */
+      private com.google.protobuf.SingleFieldBuilderV3<
+          org.tensorflow.proto.framework.SummaryMetadata, org.tensorflow.proto.framework.SummaryMetadata.Builder, org.tensorflow.proto.framework.SummaryMetadataOrBuilder> 
+          getMetadataFieldBuilder() {
+        if (metadataBuilder_ == null) {
+          metadataBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+              org.tensorflow.proto.framework.SummaryMetadata, org.tensorflow.proto.framework.SummaryMetadata.Builder, org.tensorflow.proto.framework.SummaryMetadataOrBuilder>(
+                  getMetadata(),
+                  getParentForChildren(),
+                  isClean());
+          metadata_ = null;
+        }
+        return metadataBuilder_;
+      }
+
+      /**
+       * <code>float simple_value = 2;</code>
+       */
+      public float getSimpleValue() {
+        if (valueCase_ == 2) {
+          return (java.lang.Float) value_;
+        }
+        return 0F;
+      }
+      /**
+       * <code>float simple_value = 2;</code>
+       */
+      public Builder setSimpleValue(float value) {
+        valueCase_ = 2;
+        value_ = value;
+        onChanged();
+        return this;
+      }
+      /**
+       * <code>float simple_value = 2;</code>
+       */
+      public Builder clearSimpleValue() {
+        if (valueCase_ == 2) {
+          valueCase_ = 0;
+          value_ = null;
+          onChanged();
+        }
+        return this;
+      }
+
+      /**
+       * <code>bytes obsolete_old_style_histogram = 3;</code>
+       */
+      public com.google.protobuf.ByteString getObsoleteOldStyleHistogram() {
+        if (valueCase_ == 3) {
+          return (com.google.protobuf.ByteString) value_;
+        }
+        return com.google.protobuf.ByteString.EMPTY;
+      }
+      /**
+       * <code>bytes obsolete_old_style_histogram = 3;</code>
+       */
+      public Builder setObsoleteOldStyleHistogram(com.google.protobuf.ByteString value) {
+        if (value == null) {
+    throw new NullPointerException();
+  }
+  valueCase_ = 3;
+        value_ = value;
+        onChanged();
+        return this;
+      }
+      /**
+       * <code>bytes obsolete_old_style_histogram = 3;</code>
+       */
+      public Builder clearObsoleteOldStyleHistogram() {
+        if (valueCase_ == 3) {
+          valueCase_ = 0;
+          value_ = null;
+          onChanged();
+        }
+        return this;
+      }
+
+      private com.google.protobuf.SingleFieldBuilderV3<
+          org.tensorflow.proto.framework.Summary.Image, org.tensorflow.proto.framework.Summary.Image.Builder, org.tensorflow.proto.framework.Summary.ImageOrBuilder> imageBuilder_;
+      /**
+       * <code>.tensorflow.Summary.Image image = 4;</code>
+       */
+      public boolean hasImage() {
+        return valueCase_ == 4;
+      }
+      /**
+       * <code>.tensorflow.Summary.Image image = 4;</code>
+       */
+      public org.tensorflow.proto.framework.Summary.Image getImage() {
+        if (imageBuilder_ == null) {
+          if (valueCase_ == 4) {
+            return (org.tensorflow.proto.framework.Summary.Image) value_;
+          }
+          return org.tensorflow.proto.framework.Summary.Image.getDefaultInstance();
+        } else {
+          if (valueCase_ == 4) {
+            return imageBuilder_.getMessage();
+          }
+          return org.tensorflow.proto.framework.Summary.Image.getDefaultInstance();
+        }
+      }
+      /**
+       * <code>.tensorflow.Summary.Image image = 4;</code>
+       */
+      public Builder setImage(org.tensorflow.proto.framework.Summary.Image value) {
+        if (imageBuilder_ == null) {
+          if (value == null) {
+            throw new NullPointerException();
+          }
+          value_ = value;
+          onChanged();
+        } else {
+          imageBuilder_.setMessage(value);
+        }
+        valueCase_ = 4;
+        return this;
+      }
+      /**
+       * <code>.tensorflow.Summary.Image image = 4;</code>
+       */
+      public Builder setImage(
+          org.tensorflow.proto.framework.Summary.Image.Builder builderForValue) {
+        if (imageBuilder_ == null) {
+          value_ = builderForValue.build();
+          onChanged();
+        } else {
+          imageBuilder_.setMessage(builderForValue.build());
+        }
+        valueCase_ = 4;
+        return this;
+      }
+      /**
+       * <code>.tensorflow.Summary.Image image = 4;</code>
+       */
+      public Builder mergeImage(org.tensorflow.proto.framework.Summary.Image value) {
+        if (imageBuilder_ == null) {
+          if (valueCase_ == 4 &&
+              value_ != org.tensorflow.proto.framework.Summary.Image.getDefaultInstance()) {
+            value_ = org.tensorflow.proto.framework.Summary.Image.newBuilder((org.tensorflow.proto.framework.Summary.Image) value_)
+                .mergeFrom(value).buildPartial();
+          } else {
+            value_ = value;
+          }
+          onChanged();
+        } else {
+          if (valueCase_ == 4) {
+            imageBuilder_.mergeFrom(value);
+          }
+          imageBuilder_.setMessage(value);
+        }
+        valueCase_ = 4;
+        return this;
+      }
+      /**
+       * <code>.tensorflow.Summary.Image image = 4;</code>
+       */
+      public Builder clearImage() {
+        if (imageBuilder_ == null) {
+          if (valueCase_ == 4) {
+            valueCase_ = 0;
+            value_ = null;
+            onChanged();
+          }
+        } else {
+          if (valueCase_ == 4) {
+            valueCase_ = 0;
+            value_ = null;
+          }
+          imageBuilder_.clear();
+        }
+        return this;
+      }
+      /**
+       * <code>.tensorflow.Summary.Image image = 4;</code>
+       */
+      public org.tensorflow.proto.framework.Summary.Image.Builder getImageBuilder() {
+        return getImageFieldBuilder().getBuilder();
+      }
+      /**
+       * <code>.tensorflow.Summary.Image image = 4;</code>
+       */
+      public org.tensorflow.proto.framework.Summary.ImageOrBuilder getImageOrBuilder() {
+        if ((valueCase_ == 4) && (imageBuilder_ != null)) {
+          return imageBuilder_.getMessageOrBuilder();
+        } else {
+          if (valueCase_ == 4) {
+            return (org.tensorflow.proto.framework.Summary.Image) value_;
+          }
+          return org.tensorflow.proto.framework.Summary.Image.getDefaultInstance();
+        }
+      }
+      /**
+       * <code>.tensorflow.Summary.Image image = 4;</code>
+       */
+      private com.google.protobuf.SingleFieldBuilderV3<
+          org.tensorflow.proto.framework.Summary.Image, org.tensorflow.proto.framework.Summary.Image.Builder, org.tensorflow.proto.framework.Summary.ImageOrBuilder> 
+          getImageFieldBuilder() {
+        if (imageBuilder_ == null) {
+          if (!(valueCase_ == 4)) {
+            value_ = org.tensorflow.proto.framework.Summary.Image.getDefaultInstance();
+          }
+          imageBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+              org.tensorflow.proto.framework.Summary.Image, org.tensorflow.proto.framework.Summary.Image.Builder, org.tensorflow.proto.framework.Summary.ImageOrBuilder>(
+                  (org.tensorflow.proto.framework.Summary.Image) value_,
+                  getParentForChildren(),
+                  isClean());
+          value_ = null;
+        }
+        valueCase_ = 4;
+        onChanged();;
+        return imageBuilder_;
+      }
+
+      private com.google.protobuf.SingleFieldBuilderV3<
+          org.tensorflow.proto.framework.HistogramProto, org.tensorflow.proto.framework.HistogramProto.Builder, org.tensorflow.proto.framework.HistogramProtoOrBuilder> histoBuilder_;
+      /**
+       * <code>.tensorflow.HistogramProto histo = 5;</code>
+       */
+      public boolean hasHisto() {
+        return valueCase_ == 5;
+      }
+      /**
+       * <code>.tensorflow.HistogramProto histo = 5;</code>
+       */
+      public org.tensorflow.proto.framework.HistogramProto getHisto() {
+        if (histoBuilder_ == null) {
+          if (valueCase_ == 5) {
+            return (org.tensorflow.proto.framework.HistogramProto) value_;
+          }
+          return org.tensorflow.proto.framework.HistogramProto.getDefaultInstance();
+        } else {
+          if (valueCase_ == 5) {
+            return histoBuilder_.getMessage();
+          }
+          return org.tensorflow.proto.framework.HistogramProto.getDefaultInstance();
+        }
+      }
+      /**
+       * <code>.tensorflow.HistogramProto histo = 5;</code>
+       */
+      public Builder setHisto(org.tensorflow.proto.framework.HistogramProto value) {
+        if (histoBuilder_ == null) {
+          if (value == null) {
+            throw new NullPointerException();
+          }
+          value_ = value;
+          onChanged();
+        } else {
+          histoBuilder_.setMessage(value);
+        }
+        valueCase_ = 5;
+        return this;
+      }
+      /**
+       * <code>.tensorflow.HistogramProto histo = 5;</code>
+       */
+      public Builder setHisto(
+          org.tensorflow.proto.framework.HistogramProto.Builder builderForValue) {
+        if (histoBuilder_ == null) {
+          value_ = builderForValue.build();
+          onChanged();
+        } else {
+          histoBuilder_.setMessage(builderForValue.build());
+        }
+        valueCase_ = 5;
+        return this;
+      }
+      /**
+       * <code>.tensorflow.HistogramProto histo = 5;</code>
+       */
+      public Builder mergeHisto(org.tensorflow.proto.framework.HistogramProto value) {
+        if (histoBuilder_ == null) {
+          if (valueCase_ == 5 &&
+              value_ != org.tensorflow.proto.framework.HistogramProto.getDefaultInstance()) {
+            value_ = org.tensorflow.proto.framework.HistogramProto.newBuilder((org.tensorflow.proto.framework.HistogramProto) value_)
+                .mergeFrom(value).buildPartial();
+          } else {
+            value_ = value;
+          }
+          onChanged();
+        } else {
+          if (valueCase_ == 5) {
+            histoBuilder_.mergeFrom(value);
+          }
+          histoBuilder_.setMessage(value);
+        }
+        valueCase_ = 5;
+        return this;
+      }
+      /**
+       * <code>.tensorflow.HistogramProto histo = 5;</code>
+       */
+      public Builder clearHisto() {
+        if (histoBuilder_ == null) {
+          if (valueCase_ == 5) {
+            valueCase_ = 0;
+            value_ = null;
+            onChanged();
+          }
+        } else {
+          if (valueCase_ == 5) {
+            valueCase_ = 0;
+            value_ = null;
+          }
+          histoBuilder_.clear();
+        }
+        return this;
+      }
+      /**
+       * <code>.tensorflow.HistogramProto histo = 5;</code>
+       */
+      public org.tensorflow.proto.framework.HistogramProto.Builder getHistoBuilder() {
+        return getHistoFieldBuilder().getBuilder();
+      }
+      /**
+       * <code>.tensorflow.HistogramProto histo = 5;</code>
+       */
+      public org.tensorflow.proto.framework.HistogramProtoOrBuilder getHistoOrBuilder() {
+        if ((valueCase_ == 5) && (histoBuilder_ != null)) {
+          return histoBuilder_.getMessageOrBuilder();
+        } else {
+          if (valueCase_ == 5) {
+            return (org.tensorflow.proto.framework.HistogramProto) value_;
+          }
+          return org.tensorflow.proto.framework.HistogramProto.getDefaultInstance();
+        }
+      }
+      /**
+       * <code>.tensorflow.HistogramProto histo = 5;</code>
+       */
+      private com.google.protobuf.SingleFieldBuilderV3<
+          org.tensorflow.proto.framework.HistogramProto, org.tensorflow.proto.framework.HistogramProto.Builder, org.tensorflow.proto.framework.HistogramProtoOrBuilder> 
+          getHistoFieldBuilder() {
+        if (histoBuilder_ == null) {
+          if (!(valueCase_ == 5)) {
+            value_ = org.tensorflow.proto.framework.HistogramProto.getDefaultInstance();
+          }
+          histoBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+              org.tensorflow.proto.framework.HistogramProto, org.tensorflow.proto.framework.HistogramProto.Builder, org.tensorflow.proto.framework.HistogramProtoOrBuilder>(
+                  (org.tensorflow.proto.framework.HistogramProto) value_,
+                  getParentForChildren(),
+                  isClean());
+          value_ = null;
+        }
+        valueCase_ = 5;
+        onChanged();;
+        return histoBuilder_;
+      }
+
+      private com.google.protobuf.SingleFieldBuilderV3<
+          org.tensorflow.proto.framework.Summary.Audio, org.tensorflow.proto.framework.Summary.Audio.Builder, org.tensorflow.proto.framework.Summary.AudioOrBuilder> audioBuilder_;
+      /**
+       * <code>.tensorflow.Summary.Audio audio = 6;</code>
+       */
+      public boolean hasAudio() {
+        return valueCase_ == 6;
+      }
+      /**
+       * <code>.tensorflow.Summary.Audio audio = 6;</code>
+       */
+      public org.tensorflow.proto.framework.Summary.Audio getAudio() {
+        if (audioBuilder_ == null) {
+          if (valueCase_ == 6) {
+            return (org.tensorflow.proto.framework.Summary.Audio) value_;
+          }
+          return org.tensorflow.proto.framework.Summary.Audio.getDefaultInstance();
+        } else {
+          if (valueCase_ == 6) {
+            return audioBuilder_.getMessage();
+          }
+          return org.tensorflow.proto.framework.Summary.Audio.getDefaultInstance();
+        }
+      }
+      /**
+       * <code>.tensorflow.Summary.Audio audio = 6;</code>
+       */
+      public Builder setAudio(org.tensorflow.proto.framework.Summary.Audio value) {
+        if (audioBuilder_ == null) {
+          if (value == null) {
+            throw new NullPointerException();
+          }
+          value_ = value;
+          onChanged();
+        } else {
+          audioBuilder_.setMessage(value);
+        }
+        valueCase_ = 6;
+        return this;
+      }
+      /**
+       * <code>.tensorflow.Summary.Audio audio = 6;</code>
+       */
+      public Builder setAudio(
+          org.tensorflow.proto.framework.Summary.Audio.Builder builderForValue) {
+        if (audioBuilder_ == null) {
+          value_ = builderForValue.build();
+          onChanged();
+        } else {
+          audioBuilder_.setMessage(builderForValue.build());
+        }
+        valueCase_ = 6;
+        return this;
+      }
+      /**
+       * <code>.tensorflow.Summary.Audio audio = 6;</code>
+       */
+      public Builder mergeAudio(org.tensorflow.proto.framework.Summary.Audio value) {
+        if (audioBuilder_ == null) {
+          if (valueCase_ == 6 &&
+              value_ != org.tensorflow.proto.framework.Summary.Audio.getDefaultInstance()) {
+            value_ = org.tensorflow.proto.framework.Summary.Audio.newBuilder((org.tensorflow.proto.framework.Summary.Audio) value_)
+                .mergeFrom(value).buildPartial();
+          } else {
+            value_ = value;
+          }
+          onChanged();
+        } else {
+          if (valueCase_ == 6) {
+            audioBuilder_.mergeFrom(value);
+          }
+          audioBuilder_.setMessage(value);
+        }
+        valueCase_ = 6;
+        return this;
+      }
+      /**
+       * <code>.tensorflow.Summary.Audio audio = 6;</code>
+       */
+      public Builder clearAudio() {
+        if (audioBuilder_ == null) {
+          if (valueCase_ == 6) {
+            valueCase_ = 0;
+            value_ = null;
+            onChanged();
+          }
+        } else {
+          if (valueCase_ == 6) {
+            valueCase_ = 0;
+            value_ = null;
+          }
+          audioBuilder_.clear();
+        }
+        return this;
+      }
+      /**
+       * <code>.tensorflow.Summary.Audio audio = 6;</code>
+       */
+      public org.tensorflow.proto.framework.Summary.Audio.Builder getAudioBuilder() {
+        return getAudioFieldBuilder().getBuilder();
+      }
+      /**
+       * <code>.tensorflow.Summary.Audio audio = 6;</code>
+       */
+      public org.tensorflow.proto.framework.Summary.AudioOrBuilder getAudioOrBuilder() {
+        if ((valueCase_ == 6) && (audioBuilder_ != null)) {
+          return audioBuilder_.getMessageOrBuilder();
+        } else {
+          if (valueCase_ == 6) {
+            return (org.tensorflow.proto.framework.Summary.Audio) value_;
+          }
+          return org.tensorflow.proto.framework.Summary.Audio.getDefaultInstance();
+        }
+      }
+      /**
+       * <code>.tensorflow.Summary.Audio audio = 6;</code>
+       */
+      private com.google.protobuf.SingleFieldBuilderV3<
+          org.tensorflow.proto.framework.Summary.Audio, org.tensorflow.proto.framework.Summary.Audio.Builder, org.tensorflow.proto.framework.Summary.AudioOrBuilder> 
+          getAudioFieldBuilder() {
+        if (audioBuilder_ == null) {
+          if (!(valueCase_ == 6)) {
+            value_ = org.tensorflow.proto.framework.Summary.Audio.getDefaultInstance();
+          }
+          audioBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+              org.tensorflow.proto.framework.Summary.Audio, org.tensorflow.proto.framework.Summary.Audio.Builder, org.tensorflow.proto.framework.Summary.AudioOrBuilder>(
+                  (org.tensorflow.proto.framework.Summary.Audio) value_,
+                  getParentForChildren(),
+                  isClean());
+          value_ = null;
+        }
+        valueCase_ = 6;
+        onChanged();;
+        return audioBuilder_;
+      }
+
+      private com.google.protobuf.SingleFieldBuilderV3<
+          org.tensorflow.proto.framework.TensorProto, org.tensorflow.proto.framework.TensorProto.Builder, org.tensorflow.proto.framework.TensorProtoOrBuilder> tensorBuilder_;
+      /**
+       * <code>.tensorflow.TensorProto tensor = 8;</code>
+       */
+      public boolean hasTensor() {
+        return valueCase_ == 8;
+      }
+      /**
+       * <code>.tensorflow.TensorProto tensor = 8;</code>
+       */
+      public org.tensorflow.proto.framework.TensorProto getTensor() {
+        if (tensorBuilder_ == null) {
+          if (valueCase_ == 8) {
+            return (org.tensorflow.proto.framework.TensorProto) value_;
+          }
+          return org.tensorflow.proto.framework.TensorProto.getDefaultInstance();
+        } else {
+          if (valueCase_ == 8) {
+            return tensorBuilder_.getMessage();
+          }
+          return org.tensorflow.proto.framework.TensorProto.getDefaultInstance();
+        }
+      }
+      /**
+       * <code>.tensorflow.TensorProto tensor = 8;</code>
+       */
+      public Builder setTensor(org.tensorflow.proto.framework.TensorProto value) {
+        if (tensorBuilder_ == null) {
+          if (value == null) {
+            throw new NullPointerException();
+          }
+          value_ = value;
+          onChanged();
+        } else {
+          tensorBuilder_.setMessage(value);
+        }
+        valueCase_ = 8;
+        return this;
+      }
+      /**
+       * <code>.tensorflow.TensorProto tensor = 8;</code>
+       */
+      public Builder setTensor(
+          org.tensorflow.proto.framework.TensorProto.Builder builderForValue) {
+        if (tensorBuilder_ == null) {
+          value_ = builderForValue.build();
+          onChanged();
+        } else {
+          tensorBuilder_.setMessage(builderForValue.build());
+        }
+        valueCase_ = 8;
+        return this;
+      }
+      /**
+       * <code>.tensorflow.TensorProto tensor = 8;</code>
+       */
+      public Builder mergeTensor(org.tensorflow.proto.framework.TensorProto value) {
+        if (tensorBuilder_ == null) {
+          if (valueCase_ == 8 &&
+              value_ != org.tensorflow.proto.framework.TensorProto.getDefaultInstance()) {
+            value_ = org.tensorflow.proto.framework.TensorProto.newBuilder((org.tensorflow.proto.framework.TensorProto) value_)
+                .mergeFrom(value).buildPartial();
+          } else {
+            value_ = value;
+          }
+          onChanged();
+        } else {
+          if (valueCase_ == 8) {
+            tensorBuilder_.mergeFrom(value);
+          }
+          tensorBuilder_.setMessage(value);
+        }
+        valueCase_ = 8;
+        return this;
+      }
+      /**
+       * <code>.tensorflow.TensorProto tensor = 8;</code>
+       */
+      public Builder clearTensor() {
+        if (tensorBuilder_ == null) {
+          if (valueCase_ == 8) {
+            valueCase_ = 0;
+            value_ = null;
+            onChanged();
+          }
+        } else {
+          if (valueCase_ == 8) {
+            valueCase_ = 0;
+            value_ = null;
+          }
+          tensorBuilder_.clear();
+        }
+        return this;
+      }
+      /**
+       * <code>.tensorflow.TensorProto tensor = 8;</code>
+       */
+      public org.tensorflow.proto.framework.TensorProto.Builder getTensorBuilder() {
+        return getTensorFieldBuilder().getBuilder();
+      }
+      /**
+       * <code>.tensorflow.TensorProto tensor = 8;</code>
+       */
+      public org.tensorflow.proto.framework.TensorProtoOrBuilder getTensorOrBuilder() {
+        if ((valueCase_ == 8) && (tensorBuilder_ != null)) {
+          return tensorBuilder_.getMessageOrBuilder();
+        } else {
+          if (valueCase_ == 8) {
+            return (org.tensorflow.proto.framework.TensorProto) value_;
+          }
+          return org.tensorflow.proto.framework.TensorProto.getDefaultInstance();
+        }
+      }
+      /**
+       * <code>.tensorflow.TensorProto tensor = 8;</code>
+       */
+      private com.google.protobuf.SingleFieldBuilderV3<
+          org.tensorflow.proto.framework.TensorProto, org.tensorflow.proto.framework.TensorProto.Builder, org.tensorflow.proto.framework.TensorProtoOrBuilder> 
+          getTensorFieldBuilder() {
+        if (tensorBuilder_ == null) {
+          if (!(valueCase_ == 8)) {
+            value_ = org.tensorflow.proto.framework.TensorProto.getDefaultInstance();
+          }
+          tensorBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+              org.tensorflow.proto.framework.TensorProto, org.tensorflow.proto.framework.TensorProto.Builder, org.tensorflow.proto.framework.TensorProtoOrBuilder>(
+                  (org.tensorflow.proto.framework.TensorProto) value_,
+                  getParentForChildren(),
+                  isClean());
+          value_ = null;
+        }
+        valueCase_ = 8;
+        onChanged();;
+        return tensorBuilder_;
+      }
+      @java.lang.Override
+      public final Builder setUnknownFields(
+          final com.google.protobuf.UnknownFieldSet unknownFields) {
+        return super.setUnknownFields(unknownFields);
+      }
+
+      @java.lang.Override
+      public final Builder mergeUnknownFields(
+          final com.google.protobuf.UnknownFieldSet unknownFields) {
+        return super.mergeUnknownFields(unknownFields);
+      }
+
+
+      // @@protoc_insertion_point(builder_scope:tensorflow.Summary.Value)
+    }
+
+    // @@protoc_insertion_point(class_scope:tensorflow.Summary.Value)
+    private static final org.tensorflow.proto.framework.Summary.Value DEFAULT_INSTANCE;
+    static {
+      DEFAULT_INSTANCE = new org.tensorflow.proto.framework.Summary.Value();
+    }
+
+    public static org.tensorflow.proto.framework.Summary.Value getDefaultInstance() {
+      return DEFAULT_INSTANCE;
+    }
+
+    private static final com.google.protobuf.Parser<Value>
+        PARSER = new com.google.protobuf.AbstractParser<Value>() {
+      @java.lang.Override
+      public Value parsePartialFrom(
+          com.google.protobuf.CodedInputStream input,
+          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+          throws com.google.protobuf.InvalidProtocolBufferException {
+        return new Value(input, extensionRegistry);
+      }
+    };
+
+    public static com.google.protobuf.Parser<Value> parser() {
+      return PARSER;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Parser<Value> getParserForType() {
+      return PARSER;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.Summary.Value getDefaultInstanceForType() {
+      return DEFAULT_INSTANCE;
+    }
+
+  }
+
+  public static final int VALUE_FIELD_NUMBER = 1;
+  private java.util.List<org.tensorflow.proto.framework.Summary.Value> value_;
+  /**
+   * <pre>
+   * Set of values for the summary.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.Summary.Value value = 1;</code>
+   */
+  public java.util.List<org.tensorflow.proto.framework.Summary.Value> getValueList() {
+    return value_;
+  }
+  /**
+   * <pre>
+   * Set of values for the summary.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.Summary.Value value = 1;</code>
+   */
+  public java.util.List<? extends org.tensorflow.proto.framework.Summary.ValueOrBuilder> 
+      getValueOrBuilderList() {
+    return value_;
+  }
+  /**
+   * <pre>
+   * Set of values for the summary.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.Summary.Value value = 1;</code>
+   */
+  public int getValueCount() {
+    return value_.size();
+  }
+  /**
+   * <pre>
+   * Set of values for the summary.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.Summary.Value value = 1;</code>
+   */
+  public org.tensorflow.proto.framework.Summary.Value getValue(int index) {
+    return value_.get(index);
+  }
+  /**
+   * <pre>
+   * Set of values for the summary.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.Summary.Value value = 1;</code>
+   */
+  public org.tensorflow.proto.framework.Summary.ValueOrBuilder getValueOrBuilder(
+      int index) {
+    return value_.get(index);
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    for (int i = 0; i < value_.size(); i++) {
+      output.writeMessage(1, value_.get(i));
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    for (int i = 0; i < value_.size(); i++) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(1, value_.get(i));
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.framework.Summary)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.framework.Summary other = (org.tensorflow.proto.framework.Summary) obj;
+
+    if (!getValueList()
+        .equals(other.getValueList())) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    if (getValueCount() > 0) {
+      hash = (37 * hash) + VALUE_FIELD_NUMBER;
+      hash = (53 * hash) + getValueList().hashCode();
+    }
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.framework.Summary parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.Summary parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.Summary parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.Summary parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.Summary parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.Summary parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.Summary parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.Summary parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.Summary parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.Summary parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.Summary parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.Summary parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.framework.Summary prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * <pre>
+   * A Summary is a set of named values to be displayed by the
+   * visualizer.
+   * Summaries are produced regularly during training, as controlled by
+   * the "summary_interval_secs" attribute of the training operation.
+   * Summaries are also produced at the end of an evaluation.
+   * </pre>
+   *
+   * Protobuf type {@code tensorflow.Summary}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.Summary)
+      org.tensorflow.proto.framework.SummaryOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.SummaryProtos.internal_static_tensorflow_Summary_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.SummaryProtos.internal_static_tensorflow_Summary_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.Summary.class, org.tensorflow.proto.framework.Summary.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.framework.Summary.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+        getValueFieldBuilder();
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      if (valueBuilder_ == null) {
+        value_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000001);
+      } else {
+        valueBuilder_.clear();
+      }
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.framework.SummaryProtos.internal_static_tensorflow_Summary_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.Summary getDefaultInstanceForType() {
+      return org.tensorflow.proto.framework.Summary.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.Summary build() {
+      org.tensorflow.proto.framework.Summary result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.Summary buildPartial() {
+      org.tensorflow.proto.framework.Summary result = new org.tensorflow.proto.framework.Summary(this);
+      int from_bitField0_ = bitField0_;
+      if (valueBuilder_ == null) {
+        if (((bitField0_ & 0x00000001) != 0)) {
+          value_ = java.util.Collections.unmodifiableList(value_);
+          bitField0_ = (bitField0_ & ~0x00000001);
+        }
+        result.value_ = value_;
+      } else {
+        result.value_ = valueBuilder_.build();
+      }
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.framework.Summary) {
+        return mergeFrom((org.tensorflow.proto.framework.Summary)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.framework.Summary other) {
+      if (other == org.tensorflow.proto.framework.Summary.getDefaultInstance()) return this;
+      if (valueBuilder_ == null) {
+        if (!other.value_.isEmpty()) {
+          if (value_.isEmpty()) {
+            value_ = other.value_;
+            bitField0_ = (bitField0_ & ~0x00000001);
+          } else {
+            ensureValueIsMutable();
+            value_.addAll(other.value_);
+          }
+          onChanged();
+        }
+      } else {
+        if (!other.value_.isEmpty()) {
+          if (valueBuilder_.isEmpty()) {
+            valueBuilder_.dispose();
+            valueBuilder_ = null;
+            value_ = other.value_;
+            bitField0_ = (bitField0_ & ~0x00000001);
+            valueBuilder_ = 
+              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
+                 getValueFieldBuilder() : null;
+          } else {
+            valueBuilder_.addAllMessages(other.value_);
+          }
+        }
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.framework.Summary parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.framework.Summary) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+    private int bitField0_;
+
+    private java.util.List<org.tensorflow.proto.framework.Summary.Value> value_ =
+      java.util.Collections.emptyList();
+    private void ensureValueIsMutable() {
+      if (!((bitField0_ & 0x00000001) != 0)) {
+        value_ = new java.util.ArrayList<org.tensorflow.proto.framework.Summary.Value>(value_);
+        bitField0_ |= 0x00000001;
+       }
+    }
+
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.framework.Summary.Value, org.tensorflow.proto.framework.Summary.Value.Builder, org.tensorflow.proto.framework.Summary.ValueOrBuilder> valueBuilder_;
+
+    /**
+     * <pre>
+     * Set of values for the summary.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.Summary.Value value = 1;</code>
+     */
+    public java.util.List<org.tensorflow.proto.framework.Summary.Value> getValueList() {
+      if (valueBuilder_ == null) {
+        return java.util.Collections.unmodifiableList(value_);
+      } else {
+        return valueBuilder_.getMessageList();
+      }
+    }
+    /**
+     * <pre>
+     * Set of values for the summary.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.Summary.Value value = 1;</code>
+     */
+    public int getValueCount() {
+      if (valueBuilder_ == null) {
+        return value_.size();
+      } else {
+        return valueBuilder_.getCount();
+      }
+    }
+    /**
+     * <pre>
+     * Set of values for the summary.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.Summary.Value value = 1;</code>
+     */
+    public org.tensorflow.proto.framework.Summary.Value getValue(int index) {
+      if (valueBuilder_ == null) {
+        return value_.get(index);
+      } else {
+        return valueBuilder_.getMessage(index);
+      }
+    }
+    /**
+     * <pre>
+     * Set of values for the summary.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.Summary.Value value = 1;</code>
+     */
+    public Builder setValue(
+        int index, org.tensorflow.proto.framework.Summary.Value value) {
+      if (valueBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureValueIsMutable();
+        value_.set(index, value);
+        onChanged();
+      } else {
+        valueBuilder_.setMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Set of values for the summary.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.Summary.Value value = 1;</code>
+     */
+    public Builder setValue(
+        int index, org.tensorflow.proto.framework.Summary.Value.Builder builderForValue) {
+      if (valueBuilder_ == null) {
+        ensureValueIsMutable();
+        value_.set(index, builderForValue.build());
+        onChanged();
+      } else {
+        valueBuilder_.setMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Set of values for the summary.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.Summary.Value value = 1;</code>
+     */
+    public Builder addValue(org.tensorflow.proto.framework.Summary.Value value) {
+      if (valueBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureValueIsMutable();
+        value_.add(value);
+        onChanged();
+      } else {
+        valueBuilder_.addMessage(value);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Set of values for the summary.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.Summary.Value value = 1;</code>
+     */
+    public Builder addValue(
+        int index, org.tensorflow.proto.framework.Summary.Value value) {
+      if (valueBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureValueIsMutable();
+        value_.add(index, value);
+        onChanged();
+      } else {
+        valueBuilder_.addMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Set of values for the summary.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.Summary.Value value = 1;</code>
+     */
+    public Builder addValue(
+        org.tensorflow.proto.framework.Summary.Value.Builder builderForValue) {
+      if (valueBuilder_ == null) {
+        ensureValueIsMutable();
+        value_.add(builderForValue.build());
+        onChanged();
+      } else {
+        valueBuilder_.addMessage(builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Set of values for the summary.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.Summary.Value value = 1;</code>
+     */
+    public Builder addValue(
+        int index, org.tensorflow.proto.framework.Summary.Value.Builder builderForValue) {
+      if (valueBuilder_ == null) {
+        ensureValueIsMutable();
+        value_.add(index, builderForValue.build());
+        onChanged();
+      } else {
+        valueBuilder_.addMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Set of values for the summary.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.Summary.Value value = 1;</code>
+     */
+    public Builder addAllValue(
+        java.lang.Iterable<? extends org.tensorflow.proto.framework.Summary.Value> values) {
+      if (valueBuilder_ == null) {
+        ensureValueIsMutable();
+        com.google.protobuf.AbstractMessageLite.Builder.addAll(
+            values, value_);
+        onChanged();
+      } else {
+        valueBuilder_.addAllMessages(values);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Set of values for the summary.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.Summary.Value value = 1;</code>
+     */
+    public Builder clearValue() {
+      if (valueBuilder_ == null) {
+        value_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000001);
+        onChanged();
+      } else {
+        valueBuilder_.clear();
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Set of values for the summary.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.Summary.Value value = 1;</code>
+     */
+    public Builder removeValue(int index) {
+      if (valueBuilder_ == null) {
+        ensureValueIsMutable();
+        value_.remove(index);
+        onChanged();
+      } else {
+        valueBuilder_.remove(index);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Set of values for the summary.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.Summary.Value value = 1;</code>
+     */
+    public org.tensorflow.proto.framework.Summary.Value.Builder getValueBuilder(
+        int index) {
+      return getValueFieldBuilder().getBuilder(index);
+    }
+    /**
+     * <pre>
+     * Set of values for the summary.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.Summary.Value value = 1;</code>
+     */
+    public org.tensorflow.proto.framework.Summary.ValueOrBuilder getValueOrBuilder(
+        int index) {
+      if (valueBuilder_ == null) {
+        return value_.get(index);  } else {
+        return valueBuilder_.getMessageOrBuilder(index);
+      }
+    }
+    /**
+     * <pre>
+     * Set of values for the summary.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.Summary.Value value = 1;</code>
+     */
+    public java.util.List<? extends org.tensorflow.proto.framework.Summary.ValueOrBuilder> 
+         getValueOrBuilderList() {
+      if (valueBuilder_ != null) {
+        return valueBuilder_.getMessageOrBuilderList();
+      } else {
+        return java.util.Collections.unmodifiableList(value_);
+      }
+    }
+    /**
+     * <pre>
+     * Set of values for the summary.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.Summary.Value value = 1;</code>
+     */
+    public org.tensorflow.proto.framework.Summary.Value.Builder addValueBuilder() {
+      return getValueFieldBuilder().addBuilder(
+          org.tensorflow.proto.framework.Summary.Value.getDefaultInstance());
+    }
+    /**
+     * <pre>
+     * Set of values for the summary.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.Summary.Value value = 1;</code>
+     */
+    public org.tensorflow.proto.framework.Summary.Value.Builder addValueBuilder(
+        int index) {
+      return getValueFieldBuilder().addBuilder(
+          index, org.tensorflow.proto.framework.Summary.Value.getDefaultInstance());
+    }
+    /**
+     * <pre>
+     * Set of values for the summary.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.Summary.Value value = 1;</code>
+     */
+    public java.util.List<org.tensorflow.proto.framework.Summary.Value.Builder> 
+         getValueBuilderList() {
+      return getValueFieldBuilder().getBuilderList();
+    }
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.framework.Summary.Value, org.tensorflow.proto.framework.Summary.Value.Builder, org.tensorflow.proto.framework.Summary.ValueOrBuilder> 
+        getValueFieldBuilder() {
+      if (valueBuilder_ == null) {
+        valueBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
+            org.tensorflow.proto.framework.Summary.Value, org.tensorflow.proto.framework.Summary.Value.Builder, org.tensorflow.proto.framework.Summary.ValueOrBuilder>(
+                value_,
+                ((bitField0_ & 0x00000001) != 0),
+                getParentForChildren(),
+                isClean());
+        value_ = null;
+      }
+      return valueBuilder_;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.Summary)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.Summary)
+  private static final org.tensorflow.proto.framework.Summary DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.Summary();
+  }
+
+  public static org.tensorflow.proto.framework.Summary getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<Summary>
+      PARSER = new com.google.protobuf.AbstractParser<Summary>() {
+    @java.lang.Override
+    public Summary parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new Summary(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<Summary> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<Summary> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.framework.Summary getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SummaryDescription.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SummaryDescription.java
new file mode 100644
index 00000000000..1543fd841d1
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SummaryDescription.java
@@ -0,0 +1,589 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/framework/summary.proto
+
+package org.tensorflow.proto.framework;
+
+/**
+ * <pre>
+ * Metadata associated with a series of Summary data
+ * </pre>
+ *
+ * Protobuf type {@code tensorflow.SummaryDescription}
+ */
+public  final class SummaryDescription extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.SummaryDescription)
+    SummaryDescriptionOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use SummaryDescription.newBuilder() to construct.
+  private SummaryDescription(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private SummaryDescription() {
+    typeHint_ = "";
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new SummaryDescription();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private SummaryDescription(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 10: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            typeHint_ = s;
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.framework.SummaryProtos.internal_static_tensorflow_SummaryDescription_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.framework.SummaryProtos.internal_static_tensorflow_SummaryDescription_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.framework.SummaryDescription.class, org.tensorflow.proto.framework.SummaryDescription.Builder.class);
+  }
+
+  public static final int TYPE_HINT_FIELD_NUMBER = 1;
+  private volatile java.lang.Object typeHint_;
+  /**
+   * <pre>
+   * Hint on how plugins should process the data in this series.
+   * Supported values include "scalar", "histogram", "image", "audio"
+   * </pre>
+   *
+   * <code>string type_hint = 1;</code>
+   */
+  public java.lang.String getTypeHint() {
+    java.lang.Object ref = typeHint_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      typeHint_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * Hint on how plugins should process the data in this series.
+   * Supported values include "scalar", "histogram", "image", "audio"
+   * </pre>
+   *
+   * <code>string type_hint = 1;</code>
+   */
+  public com.google.protobuf.ByteString
+      getTypeHintBytes() {
+    java.lang.Object ref = typeHint_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      typeHint_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    if (!getTypeHintBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 1, typeHint_);
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (!getTypeHintBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, typeHint_);
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.framework.SummaryDescription)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.framework.SummaryDescription other = (org.tensorflow.proto.framework.SummaryDescription) obj;
+
+    if (!getTypeHint()
+        .equals(other.getTypeHint())) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    hash = (37 * hash) + TYPE_HINT_FIELD_NUMBER;
+    hash = (53 * hash) + getTypeHint().hashCode();
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.framework.SummaryDescription parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.SummaryDescription parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.SummaryDescription parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.SummaryDescription parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.SummaryDescription parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.SummaryDescription parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.SummaryDescription parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.SummaryDescription parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.SummaryDescription parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.SummaryDescription parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.SummaryDescription parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.SummaryDescription parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.framework.SummaryDescription prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * <pre>
+   * Metadata associated with a series of Summary data
+   * </pre>
+   *
+   * Protobuf type {@code tensorflow.SummaryDescription}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.SummaryDescription)
+      org.tensorflow.proto.framework.SummaryDescriptionOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.SummaryProtos.internal_static_tensorflow_SummaryDescription_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.SummaryProtos.internal_static_tensorflow_SummaryDescription_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.SummaryDescription.class, org.tensorflow.proto.framework.SummaryDescription.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.framework.SummaryDescription.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      typeHint_ = "";
+
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.framework.SummaryProtos.internal_static_tensorflow_SummaryDescription_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.SummaryDescription getDefaultInstanceForType() {
+      return org.tensorflow.proto.framework.SummaryDescription.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.SummaryDescription build() {
+      org.tensorflow.proto.framework.SummaryDescription result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.SummaryDescription buildPartial() {
+      org.tensorflow.proto.framework.SummaryDescription result = new org.tensorflow.proto.framework.SummaryDescription(this);
+      result.typeHint_ = typeHint_;
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.framework.SummaryDescription) {
+        return mergeFrom((org.tensorflow.proto.framework.SummaryDescription)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.framework.SummaryDescription other) {
+      if (other == org.tensorflow.proto.framework.SummaryDescription.getDefaultInstance()) return this;
+      if (!other.getTypeHint().isEmpty()) {
+        typeHint_ = other.typeHint_;
+        onChanged();
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.framework.SummaryDescription parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.framework.SummaryDescription) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+
+    private java.lang.Object typeHint_ = "";
+    /**
+     * <pre>
+     * Hint on how plugins should process the data in this series.
+     * Supported values include "scalar", "histogram", "image", "audio"
+     * </pre>
+     *
+     * <code>string type_hint = 1;</code>
+     */
+    public java.lang.String getTypeHint() {
+      java.lang.Object ref = typeHint_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        typeHint_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Hint on how plugins should process the data in this series.
+     * Supported values include "scalar", "histogram", "image", "audio"
+     * </pre>
+     *
+     * <code>string type_hint = 1;</code>
+     */
+    public com.google.protobuf.ByteString
+        getTypeHintBytes() {
+      java.lang.Object ref = typeHint_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        typeHint_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Hint on how plugins should process the data in this series.
+     * Supported values include "scalar", "histogram", "image", "audio"
+     * </pre>
+     *
+     * <code>string type_hint = 1;</code>
+     */
+    public Builder setTypeHint(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      typeHint_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Hint on how plugins should process the data in this series.
+     * Supported values include "scalar", "histogram", "image", "audio"
+     * </pre>
+     *
+     * <code>string type_hint = 1;</code>
+     */
+    public Builder clearTypeHint() {
+      
+      typeHint_ = getDefaultInstance().getTypeHint();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Hint on how plugins should process the data in this series.
+     * Supported values include "scalar", "histogram", "image", "audio"
+     * </pre>
+     *
+     * <code>string type_hint = 1;</code>
+     */
+    public Builder setTypeHintBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      typeHint_ = value;
+      onChanged();
+      return this;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.SummaryDescription)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.SummaryDescription)
+  private static final org.tensorflow.proto.framework.SummaryDescription DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.SummaryDescription();
+  }
+
+  public static org.tensorflow.proto.framework.SummaryDescription getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<SummaryDescription>
+      PARSER = new com.google.protobuf.AbstractParser<SummaryDescription>() {
+    @java.lang.Override
+    public SummaryDescription parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new SummaryDescription(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<SummaryDescription> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<SummaryDescription> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.framework.SummaryDescription getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SummaryDescriptionOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SummaryDescriptionOrBuilder.java
new file mode 100644
index 00000000000..a6b581e65e4
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SummaryDescriptionOrBuilder.java
@@ -0,0 +1,29 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/framework/summary.proto
+
+package org.tensorflow.proto.framework;
+
+public interface SummaryDescriptionOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.SummaryDescription)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <pre>
+   * Hint on how plugins should process the data in this series.
+   * Supported values include "scalar", "histogram", "image", "audio"
+   * </pre>
+   *
+   * <code>string type_hint = 1;</code>
+   */
+  java.lang.String getTypeHint();
+  /**
+   * <pre>
+   * Hint on how plugins should process the data in this series.
+   * Supported values include "scalar", "histogram", "image", "audio"
+   * </pre>
+   *
+   * <code>string type_hint = 1;</code>
+   */
+  com.google.protobuf.ByteString
+      getTypeHintBytes();
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SummaryMetadata.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SummaryMetadata.java
new file mode 100644
index 00000000000..969ed4d6d71
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SummaryMetadata.java
@@ -0,0 +1,1784 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/framework/summary.proto
+
+package org.tensorflow.proto.framework;
+
+/**
+ * <pre>
+ * A SummaryMetadata encapsulates information on which plugins are able to make
+ * use of a certain summary value.
+ * </pre>
+ *
+ * Protobuf type {@code tensorflow.SummaryMetadata}
+ */
+public  final class SummaryMetadata extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.SummaryMetadata)
+    SummaryMetadataOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use SummaryMetadata.newBuilder() to construct.
+  private SummaryMetadata(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private SummaryMetadata() {
+    displayName_ = "";
+    summaryDescription_ = "";
+    dataClass_ = 0;
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new SummaryMetadata();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private SummaryMetadata(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 10: {
+            org.tensorflow.proto.framework.SummaryMetadata.PluginData.Builder subBuilder = null;
+            if (pluginData_ != null) {
+              subBuilder = pluginData_.toBuilder();
+            }
+            pluginData_ = input.readMessage(org.tensorflow.proto.framework.SummaryMetadata.PluginData.parser(), extensionRegistry);
+            if (subBuilder != null) {
+              subBuilder.mergeFrom(pluginData_);
+              pluginData_ = subBuilder.buildPartial();
+            }
+
+            break;
+          }
+          case 18: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            displayName_ = s;
+            break;
+          }
+          case 26: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            summaryDescription_ = s;
+            break;
+          }
+          case 32: {
+            int rawValue = input.readEnum();
+
+            dataClass_ = rawValue;
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.framework.SummaryProtos.internal_static_tensorflow_SummaryMetadata_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.framework.SummaryProtos.internal_static_tensorflow_SummaryMetadata_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.framework.SummaryMetadata.class, org.tensorflow.proto.framework.SummaryMetadata.Builder.class);
+  }
+
+  public interface PluginDataOrBuilder extends
+      // @@protoc_insertion_point(interface_extends:tensorflow.SummaryMetadata.PluginData)
+      com.google.protobuf.MessageOrBuilder {
+
+    /**
+     * <pre>
+     * The name of the plugin this data pertains to.
+     * </pre>
+     *
+     * <code>string plugin_name = 1;</code>
+     */
+    java.lang.String getPluginName();
+    /**
+     * <pre>
+     * The name of the plugin this data pertains to.
+     * </pre>
+     *
+     * <code>string plugin_name = 1;</code>
+     */
+    com.google.protobuf.ByteString
+        getPluginNameBytes();
+
+    /**
+     * <pre>
+     * The content to store for the plugin. The best practice is for this to be
+     * a binary serialized protocol buffer.
+     * </pre>
+     *
+     * <code>bytes content = 2;</code>
+     */
+    com.google.protobuf.ByteString getContent();
+  }
+  /**
+   * Protobuf type {@code tensorflow.SummaryMetadata.PluginData}
+   */
+  public  static final class PluginData extends
+      com.google.protobuf.GeneratedMessageV3 implements
+      // @@protoc_insertion_point(message_implements:tensorflow.SummaryMetadata.PluginData)
+      PluginDataOrBuilder {
+  private static final long serialVersionUID = 0L;
+    // Use PluginData.newBuilder() to construct.
+    private PluginData(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+      super(builder);
+    }
+    private PluginData() {
+      pluginName_ = "";
+      content_ = com.google.protobuf.ByteString.EMPTY;
+    }
+
+    @java.lang.Override
+    @SuppressWarnings({"unused"})
+    protected java.lang.Object newInstance(
+        UnusedPrivateParameter unused) {
+      return new PluginData();
+    }
+
+    @java.lang.Override
+    public final com.google.protobuf.UnknownFieldSet
+    getUnknownFields() {
+      return this.unknownFields;
+    }
+    private PluginData(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      this();
+      if (extensionRegistry == null) {
+        throw new java.lang.NullPointerException();
+      }
+      com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+          com.google.protobuf.UnknownFieldSet.newBuilder();
+      try {
+        boolean done = false;
+        while (!done) {
+          int tag = input.readTag();
+          switch (tag) {
+            case 0:
+              done = true;
+              break;
+            case 10: {
+              java.lang.String s = input.readStringRequireUtf8();
+
+              pluginName_ = s;
+              break;
+            }
+            case 18: {
+
+              content_ = input.readBytes();
+              break;
+            }
+            default: {
+              if (!parseUnknownField(
+                  input, unknownFields, extensionRegistry, tag)) {
+                done = true;
+              }
+              break;
+            }
+          }
+        }
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        throw e.setUnfinishedMessage(this);
+      } catch (java.io.IOException e) {
+        throw new com.google.protobuf.InvalidProtocolBufferException(
+            e).setUnfinishedMessage(this);
+      } finally {
+        this.unknownFields = unknownFields.build();
+        makeExtensionsImmutable();
+      }
+    }
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.SummaryProtos.internal_static_tensorflow_SummaryMetadata_PluginData_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.SummaryProtos.internal_static_tensorflow_SummaryMetadata_PluginData_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.SummaryMetadata.PluginData.class, org.tensorflow.proto.framework.SummaryMetadata.PluginData.Builder.class);
+    }
+
+    public static final int PLUGIN_NAME_FIELD_NUMBER = 1;
+    private volatile java.lang.Object pluginName_;
+    /**
+     * <pre>
+     * The name of the plugin this data pertains to.
+     * </pre>
+     *
+     * <code>string plugin_name = 1;</code>
+     */
+    public java.lang.String getPluginName() {
+      java.lang.Object ref = pluginName_;
+      if (ref instanceof java.lang.String) {
+        return (java.lang.String) ref;
+      } else {
+        com.google.protobuf.ByteString bs = 
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        pluginName_ = s;
+        return s;
+      }
+    }
+    /**
+     * <pre>
+     * The name of the plugin this data pertains to.
+     * </pre>
+     *
+     * <code>string plugin_name = 1;</code>
+     */
+    public com.google.protobuf.ByteString
+        getPluginNameBytes() {
+      java.lang.Object ref = pluginName_;
+      if (ref instanceof java.lang.String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        pluginName_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+
+    public static final int CONTENT_FIELD_NUMBER = 2;
+    private com.google.protobuf.ByteString content_;
+    /**
+     * <pre>
+     * The content to store for the plugin. The best practice is for this to be
+     * a binary serialized protocol buffer.
+     * </pre>
+     *
+     * <code>bytes content = 2;</code>
+     */
+    public com.google.protobuf.ByteString getContent() {
+      return content_;
+    }
+
+    private byte memoizedIsInitialized = -1;
+    @java.lang.Override
+    public final boolean isInitialized() {
+      byte isInitialized = memoizedIsInitialized;
+      if (isInitialized == 1) return true;
+      if (isInitialized == 0) return false;
+
+      memoizedIsInitialized = 1;
+      return true;
+    }
+
+    @java.lang.Override
+    public void writeTo(com.google.protobuf.CodedOutputStream output)
+                        throws java.io.IOException {
+      if (!getPluginNameBytes().isEmpty()) {
+        com.google.protobuf.GeneratedMessageV3.writeString(output, 1, pluginName_);
+      }
+      if (!content_.isEmpty()) {
+        output.writeBytes(2, content_);
+      }
+      unknownFields.writeTo(output);
+    }
+
+    @java.lang.Override
+    public int getSerializedSize() {
+      int size = memoizedSize;
+      if (size != -1) return size;
+
+      size = 0;
+      if (!getPluginNameBytes().isEmpty()) {
+        size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, pluginName_);
+      }
+      if (!content_.isEmpty()) {
+        size += com.google.protobuf.CodedOutputStream
+          .computeBytesSize(2, content_);
+      }
+      size += unknownFields.getSerializedSize();
+      memoizedSize = size;
+      return size;
+    }
+
+    @java.lang.Override
+    public boolean equals(final java.lang.Object obj) {
+      if (obj == this) {
+       return true;
+      }
+      if (!(obj instanceof org.tensorflow.proto.framework.SummaryMetadata.PluginData)) {
+        return super.equals(obj);
+      }
+      org.tensorflow.proto.framework.SummaryMetadata.PluginData other = (org.tensorflow.proto.framework.SummaryMetadata.PluginData) obj;
+
+      if (!getPluginName()
+          .equals(other.getPluginName())) return false;
+      if (!getContent()
+          .equals(other.getContent())) return false;
+      if (!unknownFields.equals(other.unknownFields)) return false;
+      return true;
+    }
+
+    @java.lang.Override
+    public int hashCode() {
+      if (memoizedHashCode != 0) {
+        return memoizedHashCode;
+      }
+      int hash = 41;
+      hash = (19 * hash) + getDescriptor().hashCode();
+      hash = (37 * hash) + PLUGIN_NAME_FIELD_NUMBER;
+      hash = (53 * hash) + getPluginName().hashCode();
+      hash = (37 * hash) + CONTENT_FIELD_NUMBER;
+      hash = (53 * hash) + getContent().hashCode();
+      hash = (29 * hash) + unknownFields.hashCode();
+      memoizedHashCode = hash;
+      return hash;
+    }
+
+    public static org.tensorflow.proto.framework.SummaryMetadata.PluginData parseFrom(
+        java.nio.ByteBuffer data)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data);
+    }
+    public static org.tensorflow.proto.framework.SummaryMetadata.PluginData parseFrom(
+        java.nio.ByteBuffer data,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.SummaryMetadata.PluginData parseFrom(
+        com.google.protobuf.ByteString data)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data);
+    }
+    public static org.tensorflow.proto.framework.SummaryMetadata.PluginData parseFrom(
+        com.google.protobuf.ByteString data,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.SummaryMetadata.PluginData parseFrom(byte[] data)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data);
+    }
+    public static org.tensorflow.proto.framework.SummaryMetadata.PluginData parseFrom(
+        byte[] data,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.SummaryMetadata.PluginData parseFrom(java.io.InputStream input)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input);
+    }
+    public static org.tensorflow.proto.framework.SummaryMetadata.PluginData parseFrom(
+        java.io.InputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.SummaryMetadata.PluginData parseDelimitedFrom(java.io.InputStream input)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseDelimitedWithIOException(PARSER, input);
+    }
+    public static org.tensorflow.proto.framework.SummaryMetadata.PluginData parseDelimitedFrom(
+        java.io.InputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.SummaryMetadata.PluginData parseFrom(
+        com.google.protobuf.CodedInputStream input)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input);
+    }
+    public static org.tensorflow.proto.framework.SummaryMetadata.PluginData parseFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input, extensionRegistry);
+    }
+
+    @java.lang.Override
+    public Builder newBuilderForType() { return newBuilder(); }
+    public static Builder newBuilder() {
+      return DEFAULT_INSTANCE.toBuilder();
+    }
+    public static Builder newBuilder(org.tensorflow.proto.framework.SummaryMetadata.PluginData prototype) {
+      return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+    }
+    @java.lang.Override
+    public Builder toBuilder() {
+      return this == DEFAULT_INSTANCE
+          ? new Builder() : new Builder().mergeFrom(this);
+    }
+
+    @java.lang.Override
+    protected Builder newBuilderForType(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      Builder builder = new Builder(parent);
+      return builder;
+    }
+    /**
+     * Protobuf type {@code tensorflow.SummaryMetadata.PluginData}
+     */
+    public static final class Builder extends
+        com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+        // @@protoc_insertion_point(builder_implements:tensorflow.SummaryMetadata.PluginData)
+        org.tensorflow.proto.framework.SummaryMetadata.PluginDataOrBuilder {
+      public static final com.google.protobuf.Descriptors.Descriptor
+          getDescriptor() {
+        return org.tensorflow.proto.framework.SummaryProtos.internal_static_tensorflow_SummaryMetadata_PluginData_descriptor;
+      }
+
+      @java.lang.Override
+      protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+          internalGetFieldAccessorTable() {
+        return org.tensorflow.proto.framework.SummaryProtos.internal_static_tensorflow_SummaryMetadata_PluginData_fieldAccessorTable
+            .ensureFieldAccessorsInitialized(
+                org.tensorflow.proto.framework.SummaryMetadata.PluginData.class, org.tensorflow.proto.framework.SummaryMetadata.PluginData.Builder.class);
+      }
+
+      // Construct using org.tensorflow.proto.framework.SummaryMetadata.PluginData.newBuilder()
+      private Builder() {
+        maybeForceBuilderInitialization();
+      }
+
+      private Builder(
+          com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+        super(parent);
+        maybeForceBuilderInitialization();
+      }
+      private void maybeForceBuilderInitialization() {
+        if (com.google.protobuf.GeneratedMessageV3
+                .alwaysUseFieldBuilders) {
+        }
+      }
+      @java.lang.Override
+      public Builder clear() {
+        super.clear();
+        pluginName_ = "";
+
+        content_ = com.google.protobuf.ByteString.EMPTY;
+
+        return this;
+      }
+
+      @java.lang.Override
+      public com.google.protobuf.Descriptors.Descriptor
+          getDescriptorForType() {
+        return org.tensorflow.proto.framework.SummaryProtos.internal_static_tensorflow_SummaryMetadata_PluginData_descriptor;
+      }
+
+      @java.lang.Override
+      public org.tensorflow.proto.framework.SummaryMetadata.PluginData getDefaultInstanceForType() {
+        return org.tensorflow.proto.framework.SummaryMetadata.PluginData.getDefaultInstance();
+      }
+
+      @java.lang.Override
+      public org.tensorflow.proto.framework.SummaryMetadata.PluginData build() {
+        org.tensorflow.proto.framework.SummaryMetadata.PluginData result = buildPartial();
+        if (!result.isInitialized()) {
+          throw newUninitializedMessageException(result);
+        }
+        return result;
+      }
+
+      @java.lang.Override
+      public org.tensorflow.proto.framework.SummaryMetadata.PluginData buildPartial() {
+        org.tensorflow.proto.framework.SummaryMetadata.PluginData result = new org.tensorflow.proto.framework.SummaryMetadata.PluginData(this);
+        result.pluginName_ = pluginName_;
+        result.content_ = content_;
+        onBuilt();
+        return result;
+      }
+
+      @java.lang.Override
+      public Builder clone() {
+        return super.clone();
+      }
+      @java.lang.Override
+      public Builder setField(
+          com.google.protobuf.Descriptors.FieldDescriptor field,
+          java.lang.Object value) {
+        return super.setField(field, value);
+      }
+      @java.lang.Override
+      public Builder clearField(
+          com.google.protobuf.Descriptors.FieldDescriptor field) {
+        return super.clearField(field);
+      }
+      @java.lang.Override
+      public Builder clearOneof(
+          com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+        return super.clearOneof(oneof);
+      }
+      @java.lang.Override
+      public Builder setRepeatedField(
+          com.google.protobuf.Descriptors.FieldDescriptor field,
+          int index, java.lang.Object value) {
+        return super.setRepeatedField(field, index, value);
+      }
+      @java.lang.Override
+      public Builder addRepeatedField(
+          com.google.protobuf.Descriptors.FieldDescriptor field,
+          java.lang.Object value) {
+        return super.addRepeatedField(field, value);
+      }
+      @java.lang.Override
+      public Builder mergeFrom(com.google.protobuf.Message other) {
+        if (other instanceof org.tensorflow.proto.framework.SummaryMetadata.PluginData) {
+          return mergeFrom((org.tensorflow.proto.framework.SummaryMetadata.PluginData)other);
+        } else {
+          super.mergeFrom(other);
+          return this;
+        }
+      }
+
+      public Builder mergeFrom(org.tensorflow.proto.framework.SummaryMetadata.PluginData other) {
+        if (other == org.tensorflow.proto.framework.SummaryMetadata.PluginData.getDefaultInstance()) return this;
+        if (!other.getPluginName().isEmpty()) {
+          pluginName_ = other.pluginName_;
+          onChanged();
+        }
+        if (other.getContent() != com.google.protobuf.ByteString.EMPTY) {
+          setContent(other.getContent());
+        }
+        this.mergeUnknownFields(other.unknownFields);
+        onChanged();
+        return this;
+      }
+
+      @java.lang.Override
+      public final boolean isInitialized() {
+        return true;
+      }
+
+      @java.lang.Override
+      public Builder mergeFrom(
+          com.google.protobuf.CodedInputStream input,
+          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+          throws java.io.IOException {
+        org.tensorflow.proto.framework.SummaryMetadata.PluginData parsedMessage = null;
+        try {
+          parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+        } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+          parsedMessage = (org.tensorflow.proto.framework.SummaryMetadata.PluginData) e.getUnfinishedMessage();
+          throw e.unwrapIOException();
+        } finally {
+          if (parsedMessage != null) {
+            mergeFrom(parsedMessage);
+          }
+        }
+        return this;
+      }
+
+      private java.lang.Object pluginName_ = "";
+      /**
+       * <pre>
+       * The name of the plugin this data pertains to.
+       * </pre>
+       *
+       * <code>string plugin_name = 1;</code>
+       */
+      public java.lang.String getPluginName() {
+        java.lang.Object ref = pluginName_;
+        if (!(ref instanceof java.lang.String)) {
+          com.google.protobuf.ByteString bs =
+              (com.google.protobuf.ByteString) ref;
+          java.lang.String s = bs.toStringUtf8();
+          pluginName_ = s;
+          return s;
+        } else {
+          return (java.lang.String) ref;
+        }
+      }
+      /**
+       * <pre>
+       * The name of the plugin this data pertains to.
+       * </pre>
+       *
+       * <code>string plugin_name = 1;</code>
+       */
+      public com.google.protobuf.ByteString
+          getPluginNameBytes() {
+        java.lang.Object ref = pluginName_;
+        if (ref instanceof String) {
+          com.google.protobuf.ByteString b = 
+              com.google.protobuf.ByteString.copyFromUtf8(
+                  (java.lang.String) ref);
+          pluginName_ = b;
+          return b;
+        } else {
+          return (com.google.protobuf.ByteString) ref;
+        }
+      }
+      /**
+       * <pre>
+       * The name of the plugin this data pertains to.
+       * </pre>
+       *
+       * <code>string plugin_name = 1;</code>
+       */
+      public Builder setPluginName(
+          java.lang.String value) {
+        if (value == null) {
+    throw new NullPointerException();
+  }
+  
+        pluginName_ = value;
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * The name of the plugin this data pertains to.
+       * </pre>
+       *
+       * <code>string plugin_name = 1;</code>
+       */
+      public Builder clearPluginName() {
+        
+        pluginName_ = getDefaultInstance().getPluginName();
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * The name of the plugin this data pertains to.
+       * </pre>
+       *
+       * <code>string plugin_name = 1;</code>
+       */
+      public Builder setPluginNameBytes(
+          com.google.protobuf.ByteString value) {
+        if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+        
+        pluginName_ = value;
+        onChanged();
+        return this;
+      }
+
+      private com.google.protobuf.ByteString content_ = com.google.protobuf.ByteString.EMPTY;
+      /**
+       * <pre>
+       * The content to store for the plugin. The best practice is for this to be
+       * a binary serialized protocol buffer.
+       * </pre>
+       *
+       * <code>bytes content = 2;</code>
+       */
+      public com.google.protobuf.ByteString getContent() {
+        return content_;
+      }
+      /**
+       * <pre>
+       * The content to store for the plugin. The best practice is for this to be
+       * a binary serialized protocol buffer.
+       * </pre>
+       *
+       * <code>bytes content = 2;</code>
+       */
+      public Builder setContent(com.google.protobuf.ByteString value) {
+        if (value == null) {
+    throw new NullPointerException();
+  }
+  
+        content_ = value;
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * The content to store for the plugin. The best practice is for this to be
+       * a binary serialized protocol buffer.
+       * </pre>
+       *
+       * <code>bytes content = 2;</code>
+       */
+      public Builder clearContent() {
+        
+        content_ = getDefaultInstance().getContent();
+        onChanged();
+        return this;
+      }
+      @java.lang.Override
+      public final Builder setUnknownFields(
+          final com.google.protobuf.UnknownFieldSet unknownFields) {
+        return super.setUnknownFields(unknownFields);
+      }
+
+      @java.lang.Override
+      public final Builder mergeUnknownFields(
+          final com.google.protobuf.UnknownFieldSet unknownFields) {
+        return super.mergeUnknownFields(unknownFields);
+      }
+
+
+      // @@protoc_insertion_point(builder_scope:tensorflow.SummaryMetadata.PluginData)
+    }
+
+    // @@protoc_insertion_point(class_scope:tensorflow.SummaryMetadata.PluginData)
+    private static final org.tensorflow.proto.framework.SummaryMetadata.PluginData DEFAULT_INSTANCE;
+    static {
+      DEFAULT_INSTANCE = new org.tensorflow.proto.framework.SummaryMetadata.PluginData();
+    }
+
+    public static org.tensorflow.proto.framework.SummaryMetadata.PluginData getDefaultInstance() {
+      return DEFAULT_INSTANCE;
+    }
+
+    private static final com.google.protobuf.Parser<PluginData>
+        PARSER = new com.google.protobuf.AbstractParser<PluginData>() {
+      @java.lang.Override
+      public PluginData parsePartialFrom(
+          com.google.protobuf.CodedInputStream input,
+          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+          throws com.google.protobuf.InvalidProtocolBufferException {
+        return new PluginData(input, extensionRegistry);
+      }
+    };
+
+    public static com.google.protobuf.Parser<PluginData> parser() {
+      return PARSER;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Parser<PluginData> getParserForType() {
+      return PARSER;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.SummaryMetadata.PluginData getDefaultInstanceForType() {
+      return DEFAULT_INSTANCE;
+    }
+
+  }
+
+  public static final int PLUGIN_DATA_FIELD_NUMBER = 1;
+  private org.tensorflow.proto.framework.SummaryMetadata.PluginData pluginData_;
+  /**
+   * <pre>
+   * Data that associates a summary with a certain plugin.
+   * </pre>
+   *
+   * <code>.tensorflow.SummaryMetadata.PluginData plugin_data = 1;</code>
+   */
+  public boolean hasPluginData() {
+    return pluginData_ != null;
+  }
+  /**
+   * <pre>
+   * Data that associates a summary with a certain plugin.
+   * </pre>
+   *
+   * <code>.tensorflow.SummaryMetadata.PluginData plugin_data = 1;</code>
+   */
+  public org.tensorflow.proto.framework.SummaryMetadata.PluginData getPluginData() {
+    return pluginData_ == null ? org.tensorflow.proto.framework.SummaryMetadata.PluginData.getDefaultInstance() : pluginData_;
+  }
+  /**
+   * <pre>
+   * Data that associates a summary with a certain plugin.
+   * </pre>
+   *
+   * <code>.tensorflow.SummaryMetadata.PluginData plugin_data = 1;</code>
+   */
+  public org.tensorflow.proto.framework.SummaryMetadata.PluginDataOrBuilder getPluginDataOrBuilder() {
+    return getPluginData();
+  }
+
+  public static final int DISPLAY_NAME_FIELD_NUMBER = 2;
+  private volatile java.lang.Object displayName_;
+  /**
+   * <pre>
+   * Display name for viewing in TensorBoard.
+   * </pre>
+   *
+   * <code>string display_name = 2;</code>
+   */
+  public java.lang.String getDisplayName() {
+    java.lang.Object ref = displayName_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      displayName_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * Display name for viewing in TensorBoard.
+   * </pre>
+   *
+   * <code>string display_name = 2;</code>
+   */
+  public com.google.protobuf.ByteString
+      getDisplayNameBytes() {
+    java.lang.Object ref = displayName_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      displayName_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int SUMMARY_DESCRIPTION_FIELD_NUMBER = 3;
+  private volatile java.lang.Object summaryDescription_;
+  /**
+   * <pre>
+   * Longform readable description of the summary sequence. Markdown supported.
+   * </pre>
+   *
+   * <code>string summary_description = 3;</code>
+   */
+  public java.lang.String getSummaryDescription() {
+    java.lang.Object ref = summaryDescription_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      summaryDescription_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * Longform readable description of the summary sequence. Markdown supported.
+   * </pre>
+   *
+   * <code>string summary_description = 3;</code>
+   */
+  public com.google.protobuf.ByteString
+      getSummaryDescriptionBytes() {
+    java.lang.Object ref = summaryDescription_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      summaryDescription_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int DATA_CLASS_FIELD_NUMBER = 4;
+  private int dataClass_;
+  /**
+   * <pre>
+   * Class of data stored in this time series. Required for compatibility with
+   * TensorBoard's generic data facilities (`DataProvider`, et al.). This value
+   * imposes constraints on the dtype and shape of the corresponding tensor
+   * values. See `DataClass` docs for details.
+   * </pre>
+   *
+   * <code>.tensorflow.DataClass data_class = 4;</code>
+   */
+  public int getDataClassValue() {
+    return dataClass_;
+  }
+  /**
+   * <pre>
+   * Class of data stored in this time series. Required for compatibility with
+   * TensorBoard's generic data facilities (`DataProvider`, et al.). This value
+   * imposes constraints on the dtype and shape of the corresponding tensor
+   * values. See `DataClass` docs for details.
+   * </pre>
+   *
+   * <code>.tensorflow.DataClass data_class = 4;</code>
+   */
+  public org.tensorflow.proto.framework.DataClass getDataClass() {
+    @SuppressWarnings("deprecation")
+    org.tensorflow.proto.framework.DataClass result = org.tensorflow.proto.framework.DataClass.valueOf(dataClass_);
+    return result == null ? org.tensorflow.proto.framework.DataClass.UNRECOGNIZED : result;
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    if (pluginData_ != null) {
+      output.writeMessage(1, getPluginData());
+    }
+    if (!getDisplayNameBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 2, displayName_);
+    }
+    if (!getSummaryDescriptionBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 3, summaryDescription_);
+    }
+    if (dataClass_ != org.tensorflow.proto.framework.DataClass.DATA_CLASS_UNKNOWN.getNumber()) {
+      output.writeEnum(4, dataClass_);
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (pluginData_ != null) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(1, getPluginData());
+    }
+    if (!getDisplayNameBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(2, displayName_);
+    }
+    if (!getSummaryDescriptionBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(3, summaryDescription_);
+    }
+    if (dataClass_ != org.tensorflow.proto.framework.DataClass.DATA_CLASS_UNKNOWN.getNumber()) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeEnumSize(4, dataClass_);
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.framework.SummaryMetadata)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.framework.SummaryMetadata other = (org.tensorflow.proto.framework.SummaryMetadata) obj;
+
+    if (hasPluginData() != other.hasPluginData()) return false;
+    if (hasPluginData()) {
+      if (!getPluginData()
+          .equals(other.getPluginData())) return false;
+    }
+    if (!getDisplayName()
+        .equals(other.getDisplayName())) return false;
+    if (!getSummaryDescription()
+        .equals(other.getSummaryDescription())) return false;
+    if (dataClass_ != other.dataClass_) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    if (hasPluginData()) {
+      hash = (37 * hash) + PLUGIN_DATA_FIELD_NUMBER;
+      hash = (53 * hash) + getPluginData().hashCode();
+    }
+    hash = (37 * hash) + DISPLAY_NAME_FIELD_NUMBER;
+    hash = (53 * hash) + getDisplayName().hashCode();
+    hash = (37 * hash) + SUMMARY_DESCRIPTION_FIELD_NUMBER;
+    hash = (53 * hash) + getSummaryDescription().hashCode();
+    hash = (37 * hash) + DATA_CLASS_FIELD_NUMBER;
+    hash = (53 * hash) + dataClass_;
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.framework.SummaryMetadata parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.SummaryMetadata parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.SummaryMetadata parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.SummaryMetadata parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.SummaryMetadata parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.SummaryMetadata parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.SummaryMetadata parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.SummaryMetadata parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.SummaryMetadata parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.SummaryMetadata parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.SummaryMetadata parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.SummaryMetadata parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.framework.SummaryMetadata prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * <pre>
+   * A SummaryMetadata encapsulates information on which plugins are able to make
+   * use of a certain summary value.
+   * </pre>
+   *
+   * Protobuf type {@code tensorflow.SummaryMetadata}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.SummaryMetadata)
+      org.tensorflow.proto.framework.SummaryMetadataOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.SummaryProtos.internal_static_tensorflow_SummaryMetadata_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.SummaryProtos.internal_static_tensorflow_SummaryMetadata_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.SummaryMetadata.class, org.tensorflow.proto.framework.SummaryMetadata.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.framework.SummaryMetadata.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      if (pluginDataBuilder_ == null) {
+        pluginData_ = null;
+      } else {
+        pluginData_ = null;
+        pluginDataBuilder_ = null;
+      }
+      displayName_ = "";
+
+      summaryDescription_ = "";
+
+      dataClass_ = 0;
+
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.framework.SummaryProtos.internal_static_tensorflow_SummaryMetadata_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.SummaryMetadata getDefaultInstanceForType() {
+      return org.tensorflow.proto.framework.SummaryMetadata.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.SummaryMetadata build() {
+      org.tensorflow.proto.framework.SummaryMetadata result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.SummaryMetadata buildPartial() {
+      org.tensorflow.proto.framework.SummaryMetadata result = new org.tensorflow.proto.framework.SummaryMetadata(this);
+      if (pluginDataBuilder_ == null) {
+        result.pluginData_ = pluginData_;
+      } else {
+        result.pluginData_ = pluginDataBuilder_.build();
+      }
+      result.displayName_ = displayName_;
+      result.summaryDescription_ = summaryDescription_;
+      result.dataClass_ = dataClass_;
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.framework.SummaryMetadata) {
+        return mergeFrom((org.tensorflow.proto.framework.SummaryMetadata)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.framework.SummaryMetadata other) {
+      if (other == org.tensorflow.proto.framework.SummaryMetadata.getDefaultInstance()) return this;
+      if (other.hasPluginData()) {
+        mergePluginData(other.getPluginData());
+      }
+      if (!other.getDisplayName().isEmpty()) {
+        displayName_ = other.displayName_;
+        onChanged();
+      }
+      if (!other.getSummaryDescription().isEmpty()) {
+        summaryDescription_ = other.summaryDescription_;
+        onChanged();
+      }
+      if (other.dataClass_ != 0) {
+        setDataClassValue(other.getDataClassValue());
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.framework.SummaryMetadata parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.framework.SummaryMetadata) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+
+    private org.tensorflow.proto.framework.SummaryMetadata.PluginData pluginData_;
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.SummaryMetadata.PluginData, org.tensorflow.proto.framework.SummaryMetadata.PluginData.Builder, org.tensorflow.proto.framework.SummaryMetadata.PluginDataOrBuilder> pluginDataBuilder_;
+    /**
+     * <pre>
+     * Data that associates a summary with a certain plugin.
+     * </pre>
+     *
+     * <code>.tensorflow.SummaryMetadata.PluginData plugin_data = 1;</code>
+     */
+    public boolean hasPluginData() {
+      return pluginDataBuilder_ != null || pluginData_ != null;
+    }
+    /**
+     * <pre>
+     * Data that associates a summary with a certain plugin.
+     * </pre>
+     *
+     * <code>.tensorflow.SummaryMetadata.PluginData plugin_data = 1;</code>
+     */
+    public org.tensorflow.proto.framework.SummaryMetadata.PluginData getPluginData() {
+      if (pluginDataBuilder_ == null) {
+        return pluginData_ == null ? org.tensorflow.proto.framework.SummaryMetadata.PluginData.getDefaultInstance() : pluginData_;
+      } else {
+        return pluginDataBuilder_.getMessage();
+      }
+    }
+    /**
+     * <pre>
+     * Data that associates a summary with a certain plugin.
+     * </pre>
+     *
+     * <code>.tensorflow.SummaryMetadata.PluginData plugin_data = 1;</code>
+     */
+    public Builder setPluginData(org.tensorflow.proto.framework.SummaryMetadata.PluginData value) {
+      if (pluginDataBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        pluginData_ = value;
+        onChanged();
+      } else {
+        pluginDataBuilder_.setMessage(value);
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Data that associates a summary with a certain plugin.
+     * </pre>
+     *
+     * <code>.tensorflow.SummaryMetadata.PluginData plugin_data = 1;</code>
+     */
+    public Builder setPluginData(
+        org.tensorflow.proto.framework.SummaryMetadata.PluginData.Builder builderForValue) {
+      if (pluginDataBuilder_ == null) {
+        pluginData_ = builderForValue.build();
+        onChanged();
+      } else {
+        pluginDataBuilder_.setMessage(builderForValue.build());
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Data that associates a summary with a certain plugin.
+     * </pre>
+     *
+     * <code>.tensorflow.SummaryMetadata.PluginData plugin_data = 1;</code>
+     */
+    public Builder mergePluginData(org.tensorflow.proto.framework.SummaryMetadata.PluginData value) {
+      if (pluginDataBuilder_ == null) {
+        if (pluginData_ != null) {
+          pluginData_ =
+            org.tensorflow.proto.framework.SummaryMetadata.PluginData.newBuilder(pluginData_).mergeFrom(value).buildPartial();
+        } else {
+          pluginData_ = value;
+        }
+        onChanged();
+      } else {
+        pluginDataBuilder_.mergeFrom(value);
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Data that associates a summary with a certain plugin.
+     * </pre>
+     *
+     * <code>.tensorflow.SummaryMetadata.PluginData plugin_data = 1;</code>
+     */
+    public Builder clearPluginData() {
+      if (pluginDataBuilder_ == null) {
+        pluginData_ = null;
+        onChanged();
+      } else {
+        pluginData_ = null;
+        pluginDataBuilder_ = null;
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Data that associates a summary with a certain plugin.
+     * </pre>
+     *
+     * <code>.tensorflow.SummaryMetadata.PluginData plugin_data = 1;</code>
+     */
+    public org.tensorflow.proto.framework.SummaryMetadata.PluginData.Builder getPluginDataBuilder() {
+      
+      onChanged();
+      return getPluginDataFieldBuilder().getBuilder();
+    }
+    /**
+     * <pre>
+     * Data that associates a summary with a certain plugin.
+     * </pre>
+     *
+     * <code>.tensorflow.SummaryMetadata.PluginData plugin_data = 1;</code>
+     */
+    public org.tensorflow.proto.framework.SummaryMetadata.PluginDataOrBuilder getPluginDataOrBuilder() {
+      if (pluginDataBuilder_ != null) {
+        return pluginDataBuilder_.getMessageOrBuilder();
+      } else {
+        return pluginData_ == null ?
+            org.tensorflow.proto.framework.SummaryMetadata.PluginData.getDefaultInstance() : pluginData_;
+      }
+    }
+    /**
+     * <pre>
+     * Data that associates a summary with a certain plugin.
+     * </pre>
+     *
+     * <code>.tensorflow.SummaryMetadata.PluginData plugin_data = 1;</code>
+     */
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.SummaryMetadata.PluginData, org.tensorflow.proto.framework.SummaryMetadata.PluginData.Builder, org.tensorflow.proto.framework.SummaryMetadata.PluginDataOrBuilder> 
+        getPluginDataFieldBuilder() {
+      if (pluginDataBuilder_ == null) {
+        pluginDataBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+            org.tensorflow.proto.framework.SummaryMetadata.PluginData, org.tensorflow.proto.framework.SummaryMetadata.PluginData.Builder, org.tensorflow.proto.framework.SummaryMetadata.PluginDataOrBuilder>(
+                getPluginData(),
+                getParentForChildren(),
+                isClean());
+        pluginData_ = null;
+      }
+      return pluginDataBuilder_;
+    }
+
+    private java.lang.Object displayName_ = "";
+    /**
+     * <pre>
+     * Display name for viewing in TensorBoard.
+     * </pre>
+     *
+     * <code>string display_name = 2;</code>
+     */
+    public java.lang.String getDisplayName() {
+      java.lang.Object ref = displayName_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        displayName_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Display name for viewing in TensorBoard.
+     * </pre>
+     *
+     * <code>string display_name = 2;</code>
+     */
+    public com.google.protobuf.ByteString
+        getDisplayNameBytes() {
+      java.lang.Object ref = displayName_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        displayName_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Display name for viewing in TensorBoard.
+     * </pre>
+     *
+     * <code>string display_name = 2;</code>
+     */
+    public Builder setDisplayName(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      displayName_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Display name for viewing in TensorBoard.
+     * </pre>
+     *
+     * <code>string display_name = 2;</code>
+     */
+    public Builder clearDisplayName() {
+      
+      displayName_ = getDefaultInstance().getDisplayName();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Display name for viewing in TensorBoard.
+     * </pre>
+     *
+     * <code>string display_name = 2;</code>
+     */
+    public Builder setDisplayNameBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      displayName_ = value;
+      onChanged();
+      return this;
+    }
+
+    private java.lang.Object summaryDescription_ = "";
+    /**
+     * <pre>
+     * Longform readable description of the summary sequence. Markdown supported.
+     * </pre>
+     *
+     * <code>string summary_description = 3;</code>
+     */
+    public java.lang.String getSummaryDescription() {
+      java.lang.Object ref = summaryDescription_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        summaryDescription_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Longform readable description of the summary sequence. Markdown supported.
+     * </pre>
+     *
+     * <code>string summary_description = 3;</code>
+     */
+    public com.google.protobuf.ByteString
+        getSummaryDescriptionBytes() {
+      java.lang.Object ref = summaryDescription_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        summaryDescription_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Longform readable description of the summary sequence. Markdown supported.
+     * </pre>
+     *
+     * <code>string summary_description = 3;</code>
+     */
+    public Builder setSummaryDescription(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      summaryDescription_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Longform readable description of the summary sequence. Markdown supported.
+     * </pre>
+     *
+     * <code>string summary_description = 3;</code>
+     */
+    public Builder clearSummaryDescription() {
+      
+      summaryDescription_ = getDefaultInstance().getSummaryDescription();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Longform readable description of the summary sequence. Markdown supported.
+     * </pre>
+     *
+     * <code>string summary_description = 3;</code>
+     */
+    public Builder setSummaryDescriptionBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      summaryDescription_ = value;
+      onChanged();
+      return this;
+    }
+
+    private int dataClass_ = 0;
+    /**
+     * <pre>
+     * Class of data stored in this time series. Required for compatibility with
+     * TensorBoard's generic data facilities (`DataProvider`, et al.). This value
+     * imposes constraints on the dtype and shape of the corresponding tensor
+     * values. See `DataClass` docs for details.
+     * </pre>
+     *
+     * <code>.tensorflow.DataClass data_class = 4;</code>
+     */
+    public int getDataClassValue() {
+      return dataClass_;
+    }
+    /**
+     * <pre>
+     * Class of data stored in this time series. Required for compatibility with
+     * TensorBoard's generic data facilities (`DataProvider`, et al.). This value
+     * imposes constraints on the dtype and shape of the corresponding tensor
+     * values. See `DataClass` docs for details.
+     * </pre>
+     *
+     * <code>.tensorflow.DataClass data_class = 4;</code>
+     */
+    public Builder setDataClassValue(int value) {
+      dataClass_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Class of data stored in this time series. Required for compatibility with
+     * TensorBoard's generic data facilities (`DataProvider`, et al.). This value
+     * imposes constraints on the dtype and shape of the corresponding tensor
+     * values. See `DataClass` docs for details.
+     * </pre>
+     *
+     * <code>.tensorflow.DataClass data_class = 4;</code>
+     */
+    public org.tensorflow.proto.framework.DataClass getDataClass() {
+      @SuppressWarnings("deprecation")
+      org.tensorflow.proto.framework.DataClass result = org.tensorflow.proto.framework.DataClass.valueOf(dataClass_);
+      return result == null ? org.tensorflow.proto.framework.DataClass.UNRECOGNIZED : result;
+    }
+    /**
+     * <pre>
+     * Class of data stored in this time series. Required for compatibility with
+     * TensorBoard's generic data facilities (`DataProvider`, et al.). This value
+     * imposes constraints on the dtype and shape of the corresponding tensor
+     * values. See `DataClass` docs for details.
+     * </pre>
+     *
+     * <code>.tensorflow.DataClass data_class = 4;</code>
+     */
+    public Builder setDataClass(org.tensorflow.proto.framework.DataClass value) {
+      if (value == null) {
+        throw new NullPointerException();
+      }
+      
+      dataClass_ = value.getNumber();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Class of data stored in this time series. Required for compatibility with
+     * TensorBoard's generic data facilities (`DataProvider`, et al.). This value
+     * imposes constraints on the dtype and shape of the corresponding tensor
+     * values. See `DataClass` docs for details.
+     * </pre>
+     *
+     * <code>.tensorflow.DataClass data_class = 4;</code>
+     */
+    public Builder clearDataClass() {
+      
+      dataClass_ = 0;
+      onChanged();
+      return this;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.SummaryMetadata)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.SummaryMetadata)
+  private static final org.tensorflow.proto.framework.SummaryMetadata DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.SummaryMetadata();
+  }
+
+  public static org.tensorflow.proto.framework.SummaryMetadata getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<SummaryMetadata>
+      PARSER = new com.google.protobuf.AbstractParser<SummaryMetadata>() {
+    @java.lang.Override
+    public SummaryMetadata parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new SummaryMetadata(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<SummaryMetadata> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<SummaryMetadata> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.framework.SummaryMetadata getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SummaryMetadataOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SummaryMetadataOrBuilder.java
new file mode 100644
index 00000000000..9fd35f7d1de
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SummaryMetadataOrBuilder.java
@@ -0,0 +1,93 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/framework/summary.proto
+
+package org.tensorflow.proto.framework;
+
+public interface SummaryMetadataOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.SummaryMetadata)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <pre>
+   * Data that associates a summary with a certain plugin.
+   * </pre>
+   *
+   * <code>.tensorflow.SummaryMetadata.PluginData plugin_data = 1;</code>
+   */
+  boolean hasPluginData();
+  /**
+   * <pre>
+   * Data that associates a summary with a certain plugin.
+   * </pre>
+   *
+   * <code>.tensorflow.SummaryMetadata.PluginData plugin_data = 1;</code>
+   */
+  org.tensorflow.proto.framework.SummaryMetadata.PluginData getPluginData();
+  /**
+   * <pre>
+   * Data that associates a summary with a certain plugin.
+   * </pre>
+   *
+   * <code>.tensorflow.SummaryMetadata.PluginData plugin_data = 1;</code>
+   */
+  org.tensorflow.proto.framework.SummaryMetadata.PluginDataOrBuilder getPluginDataOrBuilder();
+
+  /**
+   * <pre>
+   * Display name for viewing in TensorBoard.
+   * </pre>
+   *
+   * <code>string display_name = 2;</code>
+   */
+  java.lang.String getDisplayName();
+  /**
+   * <pre>
+   * Display name for viewing in TensorBoard.
+   * </pre>
+   *
+   * <code>string display_name = 2;</code>
+   */
+  com.google.protobuf.ByteString
+      getDisplayNameBytes();
+
+  /**
+   * <pre>
+   * Longform readable description of the summary sequence. Markdown supported.
+   * </pre>
+   *
+   * <code>string summary_description = 3;</code>
+   */
+  java.lang.String getSummaryDescription();
+  /**
+   * <pre>
+   * Longform readable description of the summary sequence. Markdown supported.
+   * </pre>
+   *
+   * <code>string summary_description = 3;</code>
+   */
+  com.google.protobuf.ByteString
+      getSummaryDescriptionBytes();
+
+  /**
+   * <pre>
+   * Class of data stored in this time series. Required for compatibility with
+   * TensorBoard's generic data facilities (`DataProvider`, et al.). This value
+   * imposes constraints on the dtype and shape of the corresponding tensor
+   * values. See `DataClass` docs for details.
+   * </pre>
+   *
+   * <code>.tensorflow.DataClass data_class = 4;</code>
+   */
+  int getDataClassValue();
+  /**
+   * <pre>
+   * Class of data stored in this time series. Required for compatibility with
+   * TensorBoard's generic data facilities (`DataProvider`, et al.). This value
+   * imposes constraints on the dtype and shape of the corresponding tensor
+   * values. See `DataClass` docs for details.
+   * </pre>
+   *
+   * <code>.tensorflow.DataClass data_class = 4;</code>
+   */
+  org.tensorflow.proto.framework.DataClass getDataClass();
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SummaryOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SummaryOrBuilder.java
new file mode 100644
index 00000000000..09de7983ac5
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SummaryOrBuilder.java
@@ -0,0 +1,53 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/framework/summary.proto
+
+package org.tensorflow.proto.framework;
+
+public interface SummaryOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.Summary)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <pre>
+   * Set of values for the summary.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.Summary.Value value = 1;</code>
+   */
+  java.util.List<org.tensorflow.proto.framework.Summary.Value> 
+      getValueList();
+  /**
+   * <pre>
+   * Set of values for the summary.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.Summary.Value value = 1;</code>
+   */
+  org.tensorflow.proto.framework.Summary.Value getValue(int index);
+  /**
+   * <pre>
+   * Set of values for the summary.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.Summary.Value value = 1;</code>
+   */
+  int getValueCount();
+  /**
+   * <pre>
+   * Set of values for the summary.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.Summary.Value value = 1;</code>
+   */
+  java.util.List<? extends org.tensorflow.proto.framework.Summary.ValueOrBuilder> 
+      getValueOrBuilderList();
+  /**
+   * <pre>
+   * Set of values for the summary.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.Summary.Value value = 1;</code>
+   */
+  org.tensorflow.proto.framework.Summary.ValueOrBuilder getValueOrBuilder(
+      int index);
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SummaryProtos.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SummaryProtos.java
new file mode 100644
index 00000000000..8e0373a5f18
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/SummaryProtos.java
@@ -0,0 +1,159 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/framework/summary.proto
+
+package org.tensorflow.proto.framework;
+
+public final class SummaryProtos {
+  private SummaryProtos() {}
+  public static void registerAllExtensions(
+      com.google.protobuf.ExtensionRegistryLite registry) {
+  }
+
+  public static void registerAllExtensions(
+      com.google.protobuf.ExtensionRegistry registry) {
+    registerAllExtensions(
+        (com.google.protobuf.ExtensionRegistryLite) registry);
+  }
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_SummaryDescription_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_SummaryDescription_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_HistogramProto_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_HistogramProto_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_SummaryMetadata_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_SummaryMetadata_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_SummaryMetadata_PluginData_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_SummaryMetadata_PluginData_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_Summary_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_Summary_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_Summary_Image_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_Summary_Image_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_Summary_Audio_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_Summary_Audio_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_Summary_Value_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_Summary_Value_fieldAccessorTable;
+
+  public static com.google.protobuf.Descriptors.FileDescriptor
+      getDescriptor() {
+    return descriptor;
+  }
+  private static  com.google.protobuf.Descriptors.FileDescriptor
+      descriptor;
+  static {
+    java.lang.String[] descriptorData = {
+      "\n\'tensorflow/core/framework/summary.prot" +
+      "o\022\ntensorflow\032&tensorflow/core/framework" +
+      "/tensor.proto\"\'\n\022SummaryDescription\022\021\n\tt" +
+      "ype_hint\030\001 \001(\t\"\207\001\n\016HistogramProto\022\013\n\003min" +
+      "\030\001 \001(\001\022\013\n\003max\030\002 \001(\001\022\013\n\003num\030\003 \001(\001\022\013\n\003sum\030" +
+      "\004 \001(\001\022\023\n\013sum_squares\030\005 \001(\001\022\030\n\014bucket_lim" +
+      "it\030\006 \003(\001B\002\020\001\022\022\n\006bucket\030\007 \003(\001B\002\020\001\"\340\001\n\017Sum" +
+      "maryMetadata\022;\n\013plugin_data\030\001 \001(\0132&.tens" +
+      "orflow.SummaryMetadata.PluginData\022\024\n\014dis" +
+      "play_name\030\002 \001(\t\022\033\n\023summary_description\030\003" +
+      " \001(\t\022)\n\ndata_class\030\004 \001(\0162\025.tensorflow.Da" +
+      "taClass\0322\n\nPluginData\022\023\n\013plugin_name\030\001 \001" +
+      "(\t\022\017\n\007content\030\002 \001(\014\"\336\004\n\007Summary\022(\n\005value" +
+      "\030\001 \003(\0132\031.tensorflow.Summary.Value\032X\n\005Ima" +
+      "ge\022\016\n\006height\030\001 \001(\005\022\r\n\005width\030\002 \001(\005\022\022\n\ncol" +
+      "orspace\030\003 \001(\005\022\034\n\024encoded_image_string\030\004 " +
+      "\001(\014\032}\n\005Audio\022\023\n\013sample_rate\030\001 \001(\002\022\024\n\014num" +
+      "_channels\030\002 \001(\003\022\025\n\rlength_frames\030\003 \001(\003\022\034" +
+      "\n\024encoded_audio_string\030\004 \001(\014\022\024\n\014content_" +
+      "type\030\005 \001(\t\032\317\002\n\005Value\022\021\n\tnode_name\030\007 \001(\t\022" +
+      "\013\n\003tag\030\001 \001(\t\022-\n\010metadata\030\t \001(\0132\033.tensorf" +
+      "low.SummaryMetadata\022\026\n\014simple_value\030\002 \001(" +
+      "\002H\000\022&\n\034obsolete_old_style_histogram\030\003 \001(" +
+      "\014H\000\022*\n\005image\030\004 \001(\0132\031.tensorflow.Summary." +
+      "ImageH\000\022+\n\005histo\030\005 \001(\0132\032.tensorflow.Hist" +
+      "ogramProtoH\000\022*\n\005audio\030\006 \001(\0132\031.tensorflow" +
+      ".Summary.AudioH\000\022)\n\006tensor\030\010 \001(\0132\027.tenso" +
+      "rflow.TensorProtoH\000B\007\n\005value*o\n\tDataClas" +
+      "s\022\026\n\022DATA_CLASS_UNKNOWN\020\000\022\025\n\021DATA_CLASS_" +
+      "SCALAR\020\001\022\025\n\021DATA_CLASS_TENSOR\020\002\022\034\n\030DATA_" +
+      "CLASS_BLOB_SEQUENCE\020\003B\204\001\n\036org.tensorflow" +
+      ".proto.frameworkB\rSummaryProtosP\001ZNgithu" +
+      "b.com/tensorflow/tensorflow/tensorflow/g" +
+      "o/core/framework/summary_go_proto\370\001\001b\006pr" +
+      "oto3"
+    };
+    descriptor = com.google.protobuf.Descriptors.FileDescriptor
+      .internalBuildGeneratedFileFrom(descriptorData,
+        new com.google.protobuf.Descriptors.FileDescriptor[] {
+          org.tensorflow.proto.framework.TensorProtos.getDescriptor(),
+        });
+    internal_static_tensorflow_SummaryDescription_descriptor =
+      getDescriptor().getMessageTypes().get(0);
+    internal_static_tensorflow_SummaryDescription_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_SummaryDescription_descriptor,
+        new java.lang.String[] { "TypeHint", });
+    internal_static_tensorflow_HistogramProto_descriptor =
+      getDescriptor().getMessageTypes().get(1);
+    internal_static_tensorflow_HistogramProto_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_HistogramProto_descriptor,
+        new java.lang.String[] { "Min", "Max", "Num", "Sum", "SumSquares", "BucketLimit", "Bucket", });
+    internal_static_tensorflow_SummaryMetadata_descriptor =
+      getDescriptor().getMessageTypes().get(2);
+    internal_static_tensorflow_SummaryMetadata_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_SummaryMetadata_descriptor,
+        new java.lang.String[] { "PluginData", "DisplayName", "SummaryDescription", "DataClass", });
+    internal_static_tensorflow_SummaryMetadata_PluginData_descriptor =
+      internal_static_tensorflow_SummaryMetadata_descriptor.getNestedTypes().get(0);
+    internal_static_tensorflow_SummaryMetadata_PluginData_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_SummaryMetadata_PluginData_descriptor,
+        new java.lang.String[] { "PluginName", "Content", });
+    internal_static_tensorflow_Summary_descriptor =
+      getDescriptor().getMessageTypes().get(3);
+    internal_static_tensorflow_Summary_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_Summary_descriptor,
+        new java.lang.String[] { "Value", });
+    internal_static_tensorflow_Summary_Image_descriptor =
+      internal_static_tensorflow_Summary_descriptor.getNestedTypes().get(0);
+    internal_static_tensorflow_Summary_Image_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_Summary_Image_descriptor,
+        new java.lang.String[] { "Height", "Width", "Colorspace", "EncodedImageString", });
+    internal_static_tensorflow_Summary_Audio_descriptor =
+      internal_static_tensorflow_Summary_descriptor.getNestedTypes().get(1);
+    internal_static_tensorflow_Summary_Audio_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_Summary_Audio_descriptor,
+        new java.lang.String[] { "SampleRate", "NumChannels", "LengthFrames", "EncodedAudioString", "ContentType", });
+    internal_static_tensorflow_Summary_Value_descriptor =
+      internal_static_tensorflow_Summary_descriptor.getNestedTypes().get(2);
+    internal_static_tensorflow_Summary_Value_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_Summary_Value_descriptor,
+        new java.lang.String[] { "NodeName", "Tag", "Metadata", "SimpleValue", "ObsoleteOldStyleHistogram", "Image", "Histo", "Audio", "Tensor", "Value", });
+    org.tensorflow.proto.framework.TensorProtos.getDescriptor();
+  }
+
+  // @@protoc_insertion_point(outer_class_scope)
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorConnection.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorConnection.java
new file mode 100644
index 00000000000..5be946ae717
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorConnection.java
@@ -0,0 +1,751 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/config.proto
+
+package org.tensorflow.proto.framework;
+
+/**
+ * <pre>
+ * Defines a connection between two tensors in a `GraphDef`.
+ * </pre>
+ *
+ * Protobuf type {@code tensorflow.TensorConnection}
+ */
+public  final class TensorConnection extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.TensorConnection)
+    TensorConnectionOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use TensorConnection.newBuilder() to construct.
+  private TensorConnection(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private TensorConnection() {
+    fromTensor_ = "";
+    toTensor_ = "";
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new TensorConnection();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private TensorConnection(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 10: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            fromTensor_ = s;
+            break;
+          }
+          case 18: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            toTensor_ = s;
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_TensorConnection_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_TensorConnection_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.framework.TensorConnection.class, org.tensorflow.proto.framework.TensorConnection.Builder.class);
+  }
+
+  public static final int FROM_TENSOR_FIELD_NUMBER = 1;
+  private volatile java.lang.Object fromTensor_;
+  /**
+   * <pre>
+   * A tensor name. The value of this tensor will be substituted for
+   * the tensor named in `to_tensor`.
+   * </pre>
+   *
+   * <code>string from_tensor = 1;</code>
+   */
+  public java.lang.String getFromTensor() {
+    java.lang.Object ref = fromTensor_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      fromTensor_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * A tensor name. The value of this tensor will be substituted for
+   * the tensor named in `to_tensor`.
+   * </pre>
+   *
+   * <code>string from_tensor = 1;</code>
+   */
+  public com.google.protobuf.ByteString
+      getFromTensorBytes() {
+    java.lang.Object ref = fromTensor_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      fromTensor_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int TO_TENSOR_FIELD_NUMBER = 2;
+  private volatile java.lang.Object toTensor_;
+  /**
+   * <pre>
+   * A tensor name. The value of this tensor will be bound to the
+   * value of the tensor named in `from_tensor`.
+   * </pre>
+   *
+   * <code>string to_tensor = 2;</code>
+   */
+  public java.lang.String getToTensor() {
+    java.lang.Object ref = toTensor_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      toTensor_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * A tensor name. The value of this tensor will be bound to the
+   * value of the tensor named in `from_tensor`.
+   * </pre>
+   *
+   * <code>string to_tensor = 2;</code>
+   */
+  public com.google.protobuf.ByteString
+      getToTensorBytes() {
+    java.lang.Object ref = toTensor_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      toTensor_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    if (!getFromTensorBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 1, fromTensor_);
+    }
+    if (!getToTensorBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 2, toTensor_);
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (!getFromTensorBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, fromTensor_);
+    }
+    if (!getToTensorBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(2, toTensor_);
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.framework.TensorConnection)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.framework.TensorConnection other = (org.tensorflow.proto.framework.TensorConnection) obj;
+
+    if (!getFromTensor()
+        .equals(other.getFromTensor())) return false;
+    if (!getToTensor()
+        .equals(other.getToTensor())) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    hash = (37 * hash) + FROM_TENSOR_FIELD_NUMBER;
+    hash = (53 * hash) + getFromTensor().hashCode();
+    hash = (37 * hash) + TO_TENSOR_FIELD_NUMBER;
+    hash = (53 * hash) + getToTensor().hashCode();
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.framework.TensorConnection parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.TensorConnection parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.TensorConnection parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.TensorConnection parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.TensorConnection parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.TensorConnection parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.TensorConnection parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.TensorConnection parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.TensorConnection parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.TensorConnection parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.TensorConnection parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.TensorConnection parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.framework.TensorConnection prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * <pre>
+   * Defines a connection between two tensors in a `GraphDef`.
+   * </pre>
+   *
+   * Protobuf type {@code tensorflow.TensorConnection}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.TensorConnection)
+      org.tensorflow.proto.framework.TensorConnectionOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_TensorConnection_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_TensorConnection_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.TensorConnection.class, org.tensorflow.proto.framework.TensorConnection.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.framework.TensorConnection.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      fromTensor_ = "";
+
+      toTensor_ = "";
+
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_TensorConnection_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.TensorConnection getDefaultInstanceForType() {
+      return org.tensorflow.proto.framework.TensorConnection.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.TensorConnection build() {
+      org.tensorflow.proto.framework.TensorConnection result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.TensorConnection buildPartial() {
+      org.tensorflow.proto.framework.TensorConnection result = new org.tensorflow.proto.framework.TensorConnection(this);
+      result.fromTensor_ = fromTensor_;
+      result.toTensor_ = toTensor_;
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.framework.TensorConnection) {
+        return mergeFrom((org.tensorflow.proto.framework.TensorConnection)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.framework.TensorConnection other) {
+      if (other == org.tensorflow.proto.framework.TensorConnection.getDefaultInstance()) return this;
+      if (!other.getFromTensor().isEmpty()) {
+        fromTensor_ = other.fromTensor_;
+        onChanged();
+      }
+      if (!other.getToTensor().isEmpty()) {
+        toTensor_ = other.toTensor_;
+        onChanged();
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.framework.TensorConnection parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.framework.TensorConnection) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+
+    private java.lang.Object fromTensor_ = "";
+    /**
+     * <pre>
+     * A tensor name. The value of this tensor will be substituted for
+     * the tensor named in `to_tensor`.
+     * </pre>
+     *
+     * <code>string from_tensor = 1;</code>
+     */
+    public java.lang.String getFromTensor() {
+      java.lang.Object ref = fromTensor_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        fromTensor_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * A tensor name. The value of this tensor will be substituted for
+     * the tensor named in `to_tensor`.
+     * </pre>
+     *
+     * <code>string from_tensor = 1;</code>
+     */
+    public com.google.protobuf.ByteString
+        getFromTensorBytes() {
+      java.lang.Object ref = fromTensor_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        fromTensor_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * A tensor name. The value of this tensor will be substituted for
+     * the tensor named in `to_tensor`.
+     * </pre>
+     *
+     * <code>string from_tensor = 1;</code>
+     */
+    public Builder setFromTensor(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      fromTensor_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * A tensor name. The value of this tensor will be substituted for
+     * the tensor named in `to_tensor`.
+     * </pre>
+     *
+     * <code>string from_tensor = 1;</code>
+     */
+    public Builder clearFromTensor() {
+      
+      fromTensor_ = getDefaultInstance().getFromTensor();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * A tensor name. The value of this tensor will be substituted for
+     * the tensor named in `to_tensor`.
+     * </pre>
+     *
+     * <code>string from_tensor = 1;</code>
+     */
+    public Builder setFromTensorBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      fromTensor_ = value;
+      onChanged();
+      return this;
+    }
+
+    private java.lang.Object toTensor_ = "";
+    /**
+     * <pre>
+     * A tensor name. The value of this tensor will be bound to the
+     * value of the tensor named in `from_tensor`.
+     * </pre>
+     *
+     * <code>string to_tensor = 2;</code>
+     */
+    public java.lang.String getToTensor() {
+      java.lang.Object ref = toTensor_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        toTensor_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * A tensor name. The value of this tensor will be bound to the
+     * value of the tensor named in `from_tensor`.
+     * </pre>
+     *
+     * <code>string to_tensor = 2;</code>
+     */
+    public com.google.protobuf.ByteString
+        getToTensorBytes() {
+      java.lang.Object ref = toTensor_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        toTensor_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * A tensor name. The value of this tensor will be bound to the
+     * value of the tensor named in `from_tensor`.
+     * </pre>
+     *
+     * <code>string to_tensor = 2;</code>
+     */
+    public Builder setToTensor(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      toTensor_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * A tensor name. The value of this tensor will be bound to the
+     * value of the tensor named in `from_tensor`.
+     * </pre>
+     *
+     * <code>string to_tensor = 2;</code>
+     */
+    public Builder clearToTensor() {
+      
+      toTensor_ = getDefaultInstance().getToTensor();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * A tensor name. The value of this tensor will be bound to the
+     * value of the tensor named in `from_tensor`.
+     * </pre>
+     *
+     * <code>string to_tensor = 2;</code>
+     */
+    public Builder setToTensorBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      toTensor_ = value;
+      onChanged();
+      return this;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.TensorConnection)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.TensorConnection)
+  private static final org.tensorflow.proto.framework.TensorConnection DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.TensorConnection();
+  }
+
+  public static org.tensorflow.proto.framework.TensorConnection getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<TensorConnection>
+      PARSER = new com.google.protobuf.AbstractParser<TensorConnection>() {
+    @java.lang.Override
+    public TensorConnection parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new TensorConnection(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<TensorConnection> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<TensorConnection> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.framework.TensorConnection getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorConnectionOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorConnectionOrBuilder.java
new file mode 100644
index 00000000000..377a25aa7ad
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorConnectionOrBuilder.java
@@ -0,0 +1,49 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/config.proto
+
+package org.tensorflow.proto.framework;
+
+public interface TensorConnectionOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.TensorConnection)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <pre>
+   * A tensor name. The value of this tensor will be substituted for
+   * the tensor named in `to_tensor`.
+   * </pre>
+   *
+   * <code>string from_tensor = 1;</code>
+   */
+  java.lang.String getFromTensor();
+  /**
+   * <pre>
+   * A tensor name. The value of this tensor will be substituted for
+   * the tensor named in `to_tensor`.
+   * </pre>
+   *
+   * <code>string from_tensor = 1;</code>
+   */
+  com.google.protobuf.ByteString
+      getFromTensorBytes();
+
+  /**
+   * <pre>
+   * A tensor name. The value of this tensor will be bound to the
+   * value of the tensor named in `from_tensor`.
+   * </pre>
+   *
+   * <code>string to_tensor = 2;</code>
+   */
+  java.lang.String getToTensor();
+  /**
+   * <pre>
+   * A tensor name. The value of this tensor will be bound to the
+   * value of the tensor named in `from_tensor`.
+   * </pre>
+   *
+   * <code>string to_tensor = 2;</code>
+   */
+  com.google.protobuf.ByteString
+      getToTensorBytes();
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorDescription.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorDescription.java
new file mode 100644
index 00000000000..7a79838fb16
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorDescription.java
@@ -0,0 +1,990 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/framework/tensor_description.proto
+
+package org.tensorflow.proto.framework;
+
+/**
+ * Protobuf type {@code tensorflow.TensorDescription}
+ */
+public  final class TensorDescription extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.TensorDescription)
+    TensorDescriptionOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use TensorDescription.newBuilder() to construct.
+  private TensorDescription(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private TensorDescription() {
+    dtype_ = 0;
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new TensorDescription();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private TensorDescription(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 8: {
+            int rawValue = input.readEnum();
+
+            dtype_ = rawValue;
+            break;
+          }
+          case 18: {
+            org.tensorflow.proto.framework.TensorShapeProto.Builder subBuilder = null;
+            if (shape_ != null) {
+              subBuilder = shape_.toBuilder();
+            }
+            shape_ = input.readMessage(org.tensorflow.proto.framework.TensorShapeProto.parser(), extensionRegistry);
+            if (subBuilder != null) {
+              subBuilder.mergeFrom(shape_);
+              shape_ = subBuilder.buildPartial();
+            }
+
+            break;
+          }
+          case 34: {
+            org.tensorflow.proto.framework.AllocationDescription.Builder subBuilder = null;
+            if (allocationDescription_ != null) {
+              subBuilder = allocationDescription_.toBuilder();
+            }
+            allocationDescription_ = input.readMessage(org.tensorflow.proto.framework.AllocationDescription.parser(), extensionRegistry);
+            if (subBuilder != null) {
+              subBuilder.mergeFrom(allocationDescription_);
+              allocationDescription_ = subBuilder.buildPartial();
+            }
+
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.framework.TensorDescriptionProtos.internal_static_tensorflow_TensorDescription_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.framework.TensorDescriptionProtos.internal_static_tensorflow_TensorDescription_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.framework.TensorDescription.class, org.tensorflow.proto.framework.TensorDescription.Builder.class);
+  }
+
+  public static final int DTYPE_FIELD_NUMBER = 1;
+  private int dtype_;
+  /**
+   * <pre>
+   * Data type of tensor elements
+   * </pre>
+   *
+   * <code>.tensorflow.DataType dtype = 1;</code>
+   */
+  public int getDtypeValue() {
+    return dtype_;
+  }
+  /**
+   * <pre>
+   * Data type of tensor elements
+   * </pre>
+   *
+   * <code>.tensorflow.DataType dtype = 1;</code>
+   */
+  public org.tensorflow.proto.framework.DataType getDtype() {
+    @SuppressWarnings("deprecation")
+    org.tensorflow.proto.framework.DataType result = org.tensorflow.proto.framework.DataType.valueOf(dtype_);
+    return result == null ? org.tensorflow.proto.framework.DataType.UNRECOGNIZED : result;
+  }
+
+  public static final int SHAPE_FIELD_NUMBER = 2;
+  private org.tensorflow.proto.framework.TensorShapeProto shape_;
+  /**
+   * <pre>
+   * Shape of the tensor.
+   * </pre>
+   *
+   * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+   */
+  public boolean hasShape() {
+    return shape_ != null;
+  }
+  /**
+   * <pre>
+   * Shape of the tensor.
+   * </pre>
+   *
+   * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+   */
+  public org.tensorflow.proto.framework.TensorShapeProto getShape() {
+    return shape_ == null ? org.tensorflow.proto.framework.TensorShapeProto.getDefaultInstance() : shape_;
+  }
+  /**
+   * <pre>
+   * Shape of the tensor.
+   * </pre>
+   *
+   * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+   */
+  public org.tensorflow.proto.framework.TensorShapeProtoOrBuilder getShapeOrBuilder() {
+    return getShape();
+  }
+
+  public static final int ALLOCATION_DESCRIPTION_FIELD_NUMBER = 4;
+  private org.tensorflow.proto.framework.AllocationDescription allocationDescription_;
+  /**
+   * <pre>
+   * Information about the size and allocator used for the data
+   * </pre>
+   *
+   * <code>.tensorflow.AllocationDescription allocation_description = 4;</code>
+   */
+  public boolean hasAllocationDescription() {
+    return allocationDescription_ != null;
+  }
+  /**
+   * <pre>
+   * Information about the size and allocator used for the data
+   * </pre>
+   *
+   * <code>.tensorflow.AllocationDescription allocation_description = 4;</code>
+   */
+  public org.tensorflow.proto.framework.AllocationDescription getAllocationDescription() {
+    return allocationDescription_ == null ? org.tensorflow.proto.framework.AllocationDescription.getDefaultInstance() : allocationDescription_;
+  }
+  /**
+   * <pre>
+   * Information about the size and allocator used for the data
+   * </pre>
+   *
+   * <code>.tensorflow.AllocationDescription allocation_description = 4;</code>
+   */
+  public org.tensorflow.proto.framework.AllocationDescriptionOrBuilder getAllocationDescriptionOrBuilder() {
+    return getAllocationDescription();
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    if (dtype_ != org.tensorflow.proto.framework.DataType.DT_INVALID.getNumber()) {
+      output.writeEnum(1, dtype_);
+    }
+    if (shape_ != null) {
+      output.writeMessage(2, getShape());
+    }
+    if (allocationDescription_ != null) {
+      output.writeMessage(4, getAllocationDescription());
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (dtype_ != org.tensorflow.proto.framework.DataType.DT_INVALID.getNumber()) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeEnumSize(1, dtype_);
+    }
+    if (shape_ != null) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(2, getShape());
+    }
+    if (allocationDescription_ != null) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(4, getAllocationDescription());
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.framework.TensorDescription)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.framework.TensorDescription other = (org.tensorflow.proto.framework.TensorDescription) obj;
+
+    if (dtype_ != other.dtype_) return false;
+    if (hasShape() != other.hasShape()) return false;
+    if (hasShape()) {
+      if (!getShape()
+          .equals(other.getShape())) return false;
+    }
+    if (hasAllocationDescription() != other.hasAllocationDescription()) return false;
+    if (hasAllocationDescription()) {
+      if (!getAllocationDescription()
+          .equals(other.getAllocationDescription())) return false;
+    }
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    hash = (37 * hash) + DTYPE_FIELD_NUMBER;
+    hash = (53 * hash) + dtype_;
+    if (hasShape()) {
+      hash = (37 * hash) + SHAPE_FIELD_NUMBER;
+      hash = (53 * hash) + getShape().hashCode();
+    }
+    if (hasAllocationDescription()) {
+      hash = (37 * hash) + ALLOCATION_DESCRIPTION_FIELD_NUMBER;
+      hash = (53 * hash) + getAllocationDescription().hashCode();
+    }
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.framework.TensorDescription parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.TensorDescription parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.TensorDescription parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.TensorDescription parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.TensorDescription parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.TensorDescription parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.TensorDescription parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.TensorDescription parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.TensorDescription parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.TensorDescription parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.TensorDescription parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.TensorDescription parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.framework.TensorDescription prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * Protobuf type {@code tensorflow.TensorDescription}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.TensorDescription)
+      org.tensorflow.proto.framework.TensorDescriptionOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.TensorDescriptionProtos.internal_static_tensorflow_TensorDescription_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.TensorDescriptionProtos.internal_static_tensorflow_TensorDescription_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.TensorDescription.class, org.tensorflow.proto.framework.TensorDescription.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.framework.TensorDescription.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      dtype_ = 0;
+
+      if (shapeBuilder_ == null) {
+        shape_ = null;
+      } else {
+        shape_ = null;
+        shapeBuilder_ = null;
+      }
+      if (allocationDescriptionBuilder_ == null) {
+        allocationDescription_ = null;
+      } else {
+        allocationDescription_ = null;
+        allocationDescriptionBuilder_ = null;
+      }
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.framework.TensorDescriptionProtos.internal_static_tensorflow_TensorDescription_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.TensorDescription getDefaultInstanceForType() {
+      return org.tensorflow.proto.framework.TensorDescription.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.TensorDescription build() {
+      org.tensorflow.proto.framework.TensorDescription result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.TensorDescription buildPartial() {
+      org.tensorflow.proto.framework.TensorDescription result = new org.tensorflow.proto.framework.TensorDescription(this);
+      result.dtype_ = dtype_;
+      if (shapeBuilder_ == null) {
+        result.shape_ = shape_;
+      } else {
+        result.shape_ = shapeBuilder_.build();
+      }
+      if (allocationDescriptionBuilder_ == null) {
+        result.allocationDescription_ = allocationDescription_;
+      } else {
+        result.allocationDescription_ = allocationDescriptionBuilder_.build();
+      }
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.framework.TensorDescription) {
+        return mergeFrom((org.tensorflow.proto.framework.TensorDescription)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.framework.TensorDescription other) {
+      if (other == org.tensorflow.proto.framework.TensorDescription.getDefaultInstance()) return this;
+      if (other.dtype_ != 0) {
+        setDtypeValue(other.getDtypeValue());
+      }
+      if (other.hasShape()) {
+        mergeShape(other.getShape());
+      }
+      if (other.hasAllocationDescription()) {
+        mergeAllocationDescription(other.getAllocationDescription());
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.framework.TensorDescription parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.framework.TensorDescription) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+
+    private int dtype_ = 0;
+    /**
+     * <pre>
+     * Data type of tensor elements
+     * </pre>
+     *
+     * <code>.tensorflow.DataType dtype = 1;</code>
+     */
+    public int getDtypeValue() {
+      return dtype_;
+    }
+    /**
+     * <pre>
+     * Data type of tensor elements
+     * </pre>
+     *
+     * <code>.tensorflow.DataType dtype = 1;</code>
+     */
+    public Builder setDtypeValue(int value) {
+      dtype_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Data type of tensor elements
+     * </pre>
+     *
+     * <code>.tensorflow.DataType dtype = 1;</code>
+     */
+    public org.tensorflow.proto.framework.DataType getDtype() {
+      @SuppressWarnings("deprecation")
+      org.tensorflow.proto.framework.DataType result = org.tensorflow.proto.framework.DataType.valueOf(dtype_);
+      return result == null ? org.tensorflow.proto.framework.DataType.UNRECOGNIZED : result;
+    }
+    /**
+     * <pre>
+     * Data type of tensor elements
+     * </pre>
+     *
+     * <code>.tensorflow.DataType dtype = 1;</code>
+     */
+    public Builder setDtype(org.tensorflow.proto.framework.DataType value) {
+      if (value == null) {
+        throw new NullPointerException();
+      }
+      
+      dtype_ = value.getNumber();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Data type of tensor elements
+     * </pre>
+     *
+     * <code>.tensorflow.DataType dtype = 1;</code>
+     */
+    public Builder clearDtype() {
+      
+      dtype_ = 0;
+      onChanged();
+      return this;
+    }
+
+    private org.tensorflow.proto.framework.TensorShapeProto shape_;
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.TensorShapeProto, org.tensorflow.proto.framework.TensorShapeProto.Builder, org.tensorflow.proto.framework.TensorShapeProtoOrBuilder> shapeBuilder_;
+    /**
+     * <pre>
+     * Shape of the tensor.
+     * </pre>
+     *
+     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+     */
+    public boolean hasShape() {
+      return shapeBuilder_ != null || shape_ != null;
+    }
+    /**
+     * <pre>
+     * Shape of the tensor.
+     * </pre>
+     *
+     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+     */
+    public org.tensorflow.proto.framework.TensorShapeProto getShape() {
+      if (shapeBuilder_ == null) {
+        return shape_ == null ? org.tensorflow.proto.framework.TensorShapeProto.getDefaultInstance() : shape_;
+      } else {
+        return shapeBuilder_.getMessage();
+      }
+    }
+    /**
+     * <pre>
+     * Shape of the tensor.
+     * </pre>
+     *
+     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+     */
+    public Builder setShape(org.tensorflow.proto.framework.TensorShapeProto value) {
+      if (shapeBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        shape_ = value;
+        onChanged();
+      } else {
+        shapeBuilder_.setMessage(value);
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Shape of the tensor.
+     * </pre>
+     *
+     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+     */
+    public Builder setShape(
+        org.tensorflow.proto.framework.TensorShapeProto.Builder builderForValue) {
+      if (shapeBuilder_ == null) {
+        shape_ = builderForValue.build();
+        onChanged();
+      } else {
+        shapeBuilder_.setMessage(builderForValue.build());
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Shape of the tensor.
+     * </pre>
+     *
+     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+     */
+    public Builder mergeShape(org.tensorflow.proto.framework.TensorShapeProto value) {
+      if (shapeBuilder_ == null) {
+        if (shape_ != null) {
+          shape_ =
+            org.tensorflow.proto.framework.TensorShapeProto.newBuilder(shape_).mergeFrom(value).buildPartial();
+        } else {
+          shape_ = value;
+        }
+        onChanged();
+      } else {
+        shapeBuilder_.mergeFrom(value);
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Shape of the tensor.
+     * </pre>
+     *
+     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+     */
+    public Builder clearShape() {
+      if (shapeBuilder_ == null) {
+        shape_ = null;
+        onChanged();
+      } else {
+        shape_ = null;
+        shapeBuilder_ = null;
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Shape of the tensor.
+     * </pre>
+     *
+     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+     */
+    public org.tensorflow.proto.framework.TensorShapeProto.Builder getShapeBuilder() {
+      
+      onChanged();
+      return getShapeFieldBuilder().getBuilder();
+    }
+    /**
+     * <pre>
+     * Shape of the tensor.
+     * </pre>
+     *
+     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+     */
+    public org.tensorflow.proto.framework.TensorShapeProtoOrBuilder getShapeOrBuilder() {
+      if (shapeBuilder_ != null) {
+        return shapeBuilder_.getMessageOrBuilder();
+      } else {
+        return shape_ == null ?
+            org.tensorflow.proto.framework.TensorShapeProto.getDefaultInstance() : shape_;
+      }
+    }
+    /**
+     * <pre>
+     * Shape of the tensor.
+     * </pre>
+     *
+     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+     */
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.TensorShapeProto, org.tensorflow.proto.framework.TensorShapeProto.Builder, org.tensorflow.proto.framework.TensorShapeProtoOrBuilder> 
+        getShapeFieldBuilder() {
+      if (shapeBuilder_ == null) {
+        shapeBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+            org.tensorflow.proto.framework.TensorShapeProto, org.tensorflow.proto.framework.TensorShapeProto.Builder, org.tensorflow.proto.framework.TensorShapeProtoOrBuilder>(
+                getShape(),
+                getParentForChildren(),
+                isClean());
+        shape_ = null;
+      }
+      return shapeBuilder_;
+    }
+
+    private org.tensorflow.proto.framework.AllocationDescription allocationDescription_;
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.AllocationDescription, org.tensorflow.proto.framework.AllocationDescription.Builder, org.tensorflow.proto.framework.AllocationDescriptionOrBuilder> allocationDescriptionBuilder_;
+    /**
+     * <pre>
+     * Information about the size and allocator used for the data
+     * </pre>
+     *
+     * <code>.tensorflow.AllocationDescription allocation_description = 4;</code>
+     */
+    public boolean hasAllocationDescription() {
+      return allocationDescriptionBuilder_ != null || allocationDescription_ != null;
+    }
+    /**
+     * <pre>
+     * Information about the size and allocator used for the data
+     * </pre>
+     *
+     * <code>.tensorflow.AllocationDescription allocation_description = 4;</code>
+     */
+    public org.tensorflow.proto.framework.AllocationDescription getAllocationDescription() {
+      if (allocationDescriptionBuilder_ == null) {
+        return allocationDescription_ == null ? org.tensorflow.proto.framework.AllocationDescription.getDefaultInstance() : allocationDescription_;
+      } else {
+        return allocationDescriptionBuilder_.getMessage();
+      }
+    }
+    /**
+     * <pre>
+     * Information about the size and allocator used for the data
+     * </pre>
+     *
+     * <code>.tensorflow.AllocationDescription allocation_description = 4;</code>
+     */
+    public Builder setAllocationDescription(org.tensorflow.proto.framework.AllocationDescription value) {
+      if (allocationDescriptionBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        allocationDescription_ = value;
+        onChanged();
+      } else {
+        allocationDescriptionBuilder_.setMessage(value);
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Information about the size and allocator used for the data
+     * </pre>
+     *
+     * <code>.tensorflow.AllocationDescription allocation_description = 4;</code>
+     */
+    public Builder setAllocationDescription(
+        org.tensorflow.proto.framework.AllocationDescription.Builder builderForValue) {
+      if (allocationDescriptionBuilder_ == null) {
+        allocationDescription_ = builderForValue.build();
+        onChanged();
+      } else {
+        allocationDescriptionBuilder_.setMessage(builderForValue.build());
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Information about the size and allocator used for the data
+     * </pre>
+     *
+     * <code>.tensorflow.AllocationDescription allocation_description = 4;</code>
+     */
+    public Builder mergeAllocationDescription(org.tensorflow.proto.framework.AllocationDescription value) {
+      if (allocationDescriptionBuilder_ == null) {
+        if (allocationDescription_ != null) {
+          allocationDescription_ =
+            org.tensorflow.proto.framework.AllocationDescription.newBuilder(allocationDescription_).mergeFrom(value).buildPartial();
+        } else {
+          allocationDescription_ = value;
+        }
+        onChanged();
+      } else {
+        allocationDescriptionBuilder_.mergeFrom(value);
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Information about the size and allocator used for the data
+     * </pre>
+     *
+     * <code>.tensorflow.AllocationDescription allocation_description = 4;</code>
+     */
+    public Builder clearAllocationDescription() {
+      if (allocationDescriptionBuilder_ == null) {
+        allocationDescription_ = null;
+        onChanged();
+      } else {
+        allocationDescription_ = null;
+        allocationDescriptionBuilder_ = null;
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Information about the size and allocator used for the data
+     * </pre>
+     *
+     * <code>.tensorflow.AllocationDescription allocation_description = 4;</code>
+     */
+    public org.tensorflow.proto.framework.AllocationDescription.Builder getAllocationDescriptionBuilder() {
+      
+      onChanged();
+      return getAllocationDescriptionFieldBuilder().getBuilder();
+    }
+    /**
+     * <pre>
+     * Information about the size and allocator used for the data
+     * </pre>
+     *
+     * <code>.tensorflow.AllocationDescription allocation_description = 4;</code>
+     */
+    public org.tensorflow.proto.framework.AllocationDescriptionOrBuilder getAllocationDescriptionOrBuilder() {
+      if (allocationDescriptionBuilder_ != null) {
+        return allocationDescriptionBuilder_.getMessageOrBuilder();
+      } else {
+        return allocationDescription_ == null ?
+            org.tensorflow.proto.framework.AllocationDescription.getDefaultInstance() : allocationDescription_;
+      }
+    }
+    /**
+     * <pre>
+     * Information about the size and allocator used for the data
+     * </pre>
+     *
+     * <code>.tensorflow.AllocationDescription allocation_description = 4;</code>
+     */
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.AllocationDescription, org.tensorflow.proto.framework.AllocationDescription.Builder, org.tensorflow.proto.framework.AllocationDescriptionOrBuilder> 
+        getAllocationDescriptionFieldBuilder() {
+      if (allocationDescriptionBuilder_ == null) {
+        allocationDescriptionBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+            org.tensorflow.proto.framework.AllocationDescription, org.tensorflow.proto.framework.AllocationDescription.Builder, org.tensorflow.proto.framework.AllocationDescriptionOrBuilder>(
+                getAllocationDescription(),
+                getParentForChildren(),
+                isClean());
+        allocationDescription_ = null;
+      }
+      return allocationDescriptionBuilder_;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.TensorDescription)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.TensorDescription)
+  private static final org.tensorflow.proto.framework.TensorDescription DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.TensorDescription();
+  }
+
+  public static org.tensorflow.proto.framework.TensorDescription getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<TensorDescription>
+      PARSER = new com.google.protobuf.AbstractParser<TensorDescription>() {
+    @java.lang.Override
+    public TensorDescription parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new TensorDescription(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<TensorDescription> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<TensorDescription> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.framework.TensorDescription getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorDescriptionOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorDescriptionOrBuilder.java
new file mode 100644
index 00000000000..e10f0e90d6b
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorDescriptionOrBuilder.java
@@ -0,0 +1,76 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/framework/tensor_description.proto
+
+package org.tensorflow.proto.framework;
+
+public interface TensorDescriptionOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.TensorDescription)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <pre>
+   * Data type of tensor elements
+   * </pre>
+   *
+   * <code>.tensorflow.DataType dtype = 1;</code>
+   */
+  int getDtypeValue();
+  /**
+   * <pre>
+   * Data type of tensor elements
+   * </pre>
+   *
+   * <code>.tensorflow.DataType dtype = 1;</code>
+   */
+  org.tensorflow.proto.framework.DataType getDtype();
+
+  /**
+   * <pre>
+   * Shape of the tensor.
+   * </pre>
+   *
+   * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+   */
+  boolean hasShape();
+  /**
+   * <pre>
+   * Shape of the tensor.
+   * </pre>
+   *
+   * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+   */
+  org.tensorflow.proto.framework.TensorShapeProto getShape();
+  /**
+   * <pre>
+   * Shape of the tensor.
+   * </pre>
+   *
+   * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+   */
+  org.tensorflow.proto.framework.TensorShapeProtoOrBuilder getShapeOrBuilder();
+
+  /**
+   * <pre>
+   * Information about the size and allocator used for the data
+   * </pre>
+   *
+   * <code>.tensorflow.AllocationDescription allocation_description = 4;</code>
+   */
+  boolean hasAllocationDescription();
+  /**
+   * <pre>
+   * Information about the size and allocator used for the data
+   * </pre>
+   *
+   * <code>.tensorflow.AllocationDescription allocation_description = 4;</code>
+   */
+  org.tensorflow.proto.framework.AllocationDescription getAllocationDescription();
+  /**
+   * <pre>
+   * Information about the size and allocator used for the data
+   * </pre>
+   *
+   * <code>.tensorflow.AllocationDescription allocation_description = 4;</code>
+   */
+  org.tensorflow.proto.framework.AllocationDescriptionOrBuilder getAllocationDescriptionOrBuilder();
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorDescriptionProtos.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorDescriptionProtos.java
new file mode 100644
index 00000000000..840b3c4068d
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorDescriptionProtos.java
@@ -0,0 +1,65 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/framework/tensor_description.proto
+
+package org.tensorflow.proto.framework;
+
+public final class TensorDescriptionProtos {
+  private TensorDescriptionProtos() {}
+  public static void registerAllExtensions(
+      com.google.protobuf.ExtensionRegistryLite registry) {
+  }
+
+  public static void registerAllExtensions(
+      com.google.protobuf.ExtensionRegistry registry) {
+    registerAllExtensions(
+        (com.google.protobuf.ExtensionRegistryLite) registry);
+  }
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_TensorDescription_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_TensorDescription_fieldAccessorTable;
+
+  public static com.google.protobuf.Descriptors.FileDescriptor
+      getDescriptor() {
+    return descriptor;
+  }
+  private static  com.google.protobuf.Descriptors.FileDescriptor
+      descriptor;
+  static {
+    java.lang.String[] descriptorData = {
+      "\n2tensorflow/core/framework/tensor_descr" +
+      "iption.proto\022\ntensorflow\0326tensorflow/cor" +
+      "e/framework/allocation_description.proto" +
+      "\032,tensorflow/core/framework/tensor_shape" +
+      ".proto\032%tensorflow/core/framework/types." +
+      "proto\"\250\001\n\021TensorDescription\022#\n\005dtype\030\001 \001" +
+      "(\0162\024.tensorflow.DataType\022+\n\005shape\030\002 \001(\0132" +
+      "\034.tensorflow.TensorShapeProto\022A\n\026allocat" +
+      "ion_description\030\004 \001(\0132!.tensorflow.Alloc" +
+      "ationDescriptionB\231\001\n\036org.tensorflow.prot" +
+      "o.frameworkB\027TensorDescriptionProtosP\001ZY" +
+      "github.com/tensorflow/tensorflow/tensorf" +
+      "low/go/core/framework/tensor_description" +
+      "_go_proto\370\001\001b\006proto3"
+    };
+    descriptor = com.google.protobuf.Descriptors.FileDescriptor
+      .internalBuildGeneratedFileFrom(descriptorData,
+        new com.google.protobuf.Descriptors.FileDescriptor[] {
+          org.tensorflow.proto.framework.AllocationDescriptionProtos.getDescriptor(),
+          org.tensorflow.proto.framework.TensorShapeProtos.getDescriptor(),
+          org.tensorflow.proto.framework.TypesProtos.getDescriptor(),
+        });
+    internal_static_tensorflow_TensorDescription_descriptor =
+      getDescriptor().getMessageTypes().get(0);
+    internal_static_tensorflow_TensorDescription_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_TensorDescription_descriptor,
+        new java.lang.String[] { "Dtype", "Shape", "AllocationDescription", });
+    org.tensorflow.proto.framework.AllocationDescriptionProtos.getDescriptor();
+    org.tensorflow.proto.framework.TensorShapeProtos.getDescriptor();
+    org.tensorflow.proto.framework.TypesProtos.getDescriptor();
+  }
+
+  // @@protoc_insertion_point(outer_class_scope)
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorInfo.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorInfo.java
new file mode 100644
index 00000000000..4e873da94be
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorInfo.java
@@ -0,0 +1,3680 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/meta_graph.proto
+
+package org.tensorflow.proto.framework;
+
+/**
+ * <pre>
+ * Information about a Tensor necessary for feeding or retrieval.
+ * </pre>
+ *
+ * Protobuf type {@code tensorflow.TensorInfo}
+ */
+public  final class TensorInfo extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.TensorInfo)
+    TensorInfoOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use TensorInfo.newBuilder() to construct.
+  private TensorInfo(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private TensorInfo() {
+    dtype_ = 0;
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new TensorInfo();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private TensorInfo(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 10: {
+            java.lang.String s = input.readStringRequireUtf8();
+            encodingCase_ = 1;
+            encoding_ = s;
+            break;
+          }
+          case 16: {
+            int rawValue = input.readEnum();
+
+            dtype_ = rawValue;
+            break;
+          }
+          case 26: {
+            org.tensorflow.proto.framework.TensorShapeProto.Builder subBuilder = null;
+            if (tensorShape_ != null) {
+              subBuilder = tensorShape_.toBuilder();
+            }
+            tensorShape_ = input.readMessage(org.tensorflow.proto.framework.TensorShapeProto.parser(), extensionRegistry);
+            if (subBuilder != null) {
+              subBuilder.mergeFrom(tensorShape_);
+              tensorShape_ = subBuilder.buildPartial();
+            }
+
+            break;
+          }
+          case 34: {
+            org.tensorflow.proto.framework.TensorInfo.CooSparse.Builder subBuilder = null;
+            if (encodingCase_ == 4) {
+              subBuilder = ((org.tensorflow.proto.framework.TensorInfo.CooSparse) encoding_).toBuilder();
+            }
+            encoding_ =
+                input.readMessage(org.tensorflow.proto.framework.TensorInfo.CooSparse.parser(), extensionRegistry);
+            if (subBuilder != null) {
+              subBuilder.mergeFrom((org.tensorflow.proto.framework.TensorInfo.CooSparse) encoding_);
+              encoding_ = subBuilder.buildPartial();
+            }
+            encodingCase_ = 4;
+            break;
+          }
+          case 42: {
+            org.tensorflow.proto.framework.TensorInfo.CompositeTensor.Builder subBuilder = null;
+            if (encodingCase_ == 5) {
+              subBuilder = ((org.tensorflow.proto.framework.TensorInfo.CompositeTensor) encoding_).toBuilder();
+            }
+            encoding_ =
+                input.readMessage(org.tensorflow.proto.framework.TensorInfo.CompositeTensor.parser(), extensionRegistry);
+            if (subBuilder != null) {
+              subBuilder.mergeFrom((org.tensorflow.proto.framework.TensorInfo.CompositeTensor) encoding_);
+              encoding_ = subBuilder.buildPartial();
+            }
+            encodingCase_ = 5;
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.framework.MetaGraphProtos.internal_static_tensorflow_TensorInfo_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.framework.MetaGraphProtos.internal_static_tensorflow_TensorInfo_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.framework.TensorInfo.class, org.tensorflow.proto.framework.TensorInfo.Builder.class);
+  }
+
+  public interface CooSparseOrBuilder extends
+      // @@protoc_insertion_point(interface_extends:tensorflow.TensorInfo.CooSparse)
+      com.google.protobuf.MessageOrBuilder {
+
+    /**
+     * <pre>
+     * The shape of the values Tensor is [?].  Its dtype must be the dtype of
+     * the SparseTensor as a whole, given in the enclosing TensorInfo.
+     * </pre>
+     *
+     * <code>string values_tensor_name = 1;</code>
+     */
+    java.lang.String getValuesTensorName();
+    /**
+     * <pre>
+     * The shape of the values Tensor is [?].  Its dtype must be the dtype of
+     * the SparseTensor as a whole, given in the enclosing TensorInfo.
+     * </pre>
+     *
+     * <code>string values_tensor_name = 1;</code>
+     */
+    com.google.protobuf.ByteString
+        getValuesTensorNameBytes();
+
+    /**
+     * <pre>
+     * The indices Tensor must have dtype int64 and shape [?, ?].
+     * </pre>
+     *
+     * <code>string indices_tensor_name = 2;</code>
+     */
+    java.lang.String getIndicesTensorName();
+    /**
+     * <pre>
+     * The indices Tensor must have dtype int64 and shape [?, ?].
+     * </pre>
+     *
+     * <code>string indices_tensor_name = 2;</code>
+     */
+    com.google.protobuf.ByteString
+        getIndicesTensorNameBytes();
+
+    /**
+     * <pre>
+     * The dynamic logical shape represented by the SparseTensor is recorded in
+     * the Tensor referenced here.  It must have dtype int64 and shape [?].
+     * </pre>
+     *
+     * <code>string dense_shape_tensor_name = 3;</code>
+     */
+    java.lang.String getDenseShapeTensorName();
+    /**
+     * <pre>
+     * The dynamic logical shape represented by the SparseTensor is recorded in
+     * the Tensor referenced here.  It must have dtype int64 and shape [?].
+     * </pre>
+     *
+     * <code>string dense_shape_tensor_name = 3;</code>
+     */
+    com.google.protobuf.ByteString
+        getDenseShapeTensorNameBytes();
+  }
+  /**
+   * <pre>
+   * For sparse tensors, The COO encoding stores a triple of values, indices,
+   * and shape.
+   * </pre>
+   *
+   * Protobuf type {@code tensorflow.TensorInfo.CooSparse}
+   */
+  public  static final class CooSparse extends
+      com.google.protobuf.GeneratedMessageV3 implements
+      // @@protoc_insertion_point(message_implements:tensorflow.TensorInfo.CooSparse)
+      CooSparseOrBuilder {
+  private static final long serialVersionUID = 0L;
+    // Use CooSparse.newBuilder() to construct.
+    private CooSparse(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+      super(builder);
+    }
+    private CooSparse() {
+      valuesTensorName_ = "";
+      indicesTensorName_ = "";
+      denseShapeTensorName_ = "";
+    }
+
+    @java.lang.Override
+    @SuppressWarnings({"unused"})
+    protected java.lang.Object newInstance(
+        UnusedPrivateParameter unused) {
+      return new CooSparse();
+    }
+
+    @java.lang.Override
+    public final com.google.protobuf.UnknownFieldSet
+    getUnknownFields() {
+      return this.unknownFields;
+    }
+    private CooSparse(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      this();
+      if (extensionRegistry == null) {
+        throw new java.lang.NullPointerException();
+      }
+      com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+          com.google.protobuf.UnknownFieldSet.newBuilder();
+      try {
+        boolean done = false;
+        while (!done) {
+          int tag = input.readTag();
+          switch (tag) {
+            case 0:
+              done = true;
+              break;
+            case 10: {
+              java.lang.String s = input.readStringRequireUtf8();
+
+              valuesTensorName_ = s;
+              break;
+            }
+            case 18: {
+              java.lang.String s = input.readStringRequireUtf8();
+
+              indicesTensorName_ = s;
+              break;
+            }
+            case 26: {
+              java.lang.String s = input.readStringRequireUtf8();
+
+              denseShapeTensorName_ = s;
+              break;
+            }
+            default: {
+              if (!parseUnknownField(
+                  input, unknownFields, extensionRegistry, tag)) {
+                done = true;
+              }
+              break;
+            }
+          }
+        }
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        throw e.setUnfinishedMessage(this);
+      } catch (java.io.IOException e) {
+        throw new com.google.protobuf.InvalidProtocolBufferException(
+            e).setUnfinishedMessage(this);
+      } finally {
+        this.unknownFields = unknownFields.build();
+        makeExtensionsImmutable();
+      }
+    }
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.MetaGraphProtos.internal_static_tensorflow_TensorInfo_CooSparse_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.MetaGraphProtos.internal_static_tensorflow_TensorInfo_CooSparse_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.TensorInfo.CooSparse.class, org.tensorflow.proto.framework.TensorInfo.CooSparse.Builder.class);
+    }
+
+    public static final int VALUES_TENSOR_NAME_FIELD_NUMBER = 1;
+    private volatile java.lang.Object valuesTensorName_;
+    /**
+     * <pre>
+     * The shape of the values Tensor is [?].  Its dtype must be the dtype of
+     * the SparseTensor as a whole, given in the enclosing TensorInfo.
+     * </pre>
+     *
+     * <code>string values_tensor_name = 1;</code>
+     */
+    public java.lang.String getValuesTensorName() {
+      java.lang.Object ref = valuesTensorName_;
+      if (ref instanceof java.lang.String) {
+        return (java.lang.String) ref;
+      } else {
+        com.google.protobuf.ByteString bs = 
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        valuesTensorName_ = s;
+        return s;
+      }
+    }
+    /**
+     * <pre>
+     * The shape of the values Tensor is [?].  Its dtype must be the dtype of
+     * the SparseTensor as a whole, given in the enclosing TensorInfo.
+     * </pre>
+     *
+     * <code>string values_tensor_name = 1;</code>
+     */
+    public com.google.protobuf.ByteString
+        getValuesTensorNameBytes() {
+      java.lang.Object ref = valuesTensorName_;
+      if (ref instanceof java.lang.String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        valuesTensorName_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+
+    public static final int INDICES_TENSOR_NAME_FIELD_NUMBER = 2;
+    private volatile java.lang.Object indicesTensorName_;
+    /**
+     * <pre>
+     * The indices Tensor must have dtype int64 and shape [?, ?].
+     * </pre>
+     *
+     * <code>string indices_tensor_name = 2;</code>
+     */
+    public java.lang.String getIndicesTensorName() {
+      java.lang.Object ref = indicesTensorName_;
+      if (ref instanceof java.lang.String) {
+        return (java.lang.String) ref;
+      } else {
+        com.google.protobuf.ByteString bs = 
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        indicesTensorName_ = s;
+        return s;
+      }
+    }
+    /**
+     * <pre>
+     * The indices Tensor must have dtype int64 and shape [?, ?].
+     * </pre>
+     *
+     * <code>string indices_tensor_name = 2;</code>
+     */
+    public com.google.protobuf.ByteString
+        getIndicesTensorNameBytes() {
+      java.lang.Object ref = indicesTensorName_;
+      if (ref instanceof java.lang.String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        indicesTensorName_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+
+    public static final int DENSE_SHAPE_TENSOR_NAME_FIELD_NUMBER = 3;
+    private volatile java.lang.Object denseShapeTensorName_;
+    /**
+     * <pre>
+     * The dynamic logical shape represented by the SparseTensor is recorded in
+     * the Tensor referenced here.  It must have dtype int64 and shape [?].
+     * </pre>
+     *
+     * <code>string dense_shape_tensor_name = 3;</code>
+     */
+    public java.lang.String getDenseShapeTensorName() {
+      java.lang.Object ref = denseShapeTensorName_;
+      if (ref instanceof java.lang.String) {
+        return (java.lang.String) ref;
+      } else {
+        com.google.protobuf.ByteString bs = 
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        denseShapeTensorName_ = s;
+        return s;
+      }
+    }
+    /**
+     * <pre>
+     * The dynamic logical shape represented by the SparseTensor is recorded in
+     * the Tensor referenced here.  It must have dtype int64 and shape [?].
+     * </pre>
+     *
+     * <code>string dense_shape_tensor_name = 3;</code>
+     */
+    public com.google.protobuf.ByteString
+        getDenseShapeTensorNameBytes() {
+      java.lang.Object ref = denseShapeTensorName_;
+      if (ref instanceof java.lang.String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        denseShapeTensorName_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+
+    private byte memoizedIsInitialized = -1;
+    @java.lang.Override
+    public final boolean isInitialized() {
+      byte isInitialized = memoizedIsInitialized;
+      if (isInitialized == 1) return true;
+      if (isInitialized == 0) return false;
+
+      memoizedIsInitialized = 1;
+      return true;
+    }
+
+    @java.lang.Override
+    public void writeTo(com.google.protobuf.CodedOutputStream output)
+                        throws java.io.IOException {
+      if (!getValuesTensorNameBytes().isEmpty()) {
+        com.google.protobuf.GeneratedMessageV3.writeString(output, 1, valuesTensorName_);
+      }
+      if (!getIndicesTensorNameBytes().isEmpty()) {
+        com.google.protobuf.GeneratedMessageV3.writeString(output, 2, indicesTensorName_);
+      }
+      if (!getDenseShapeTensorNameBytes().isEmpty()) {
+        com.google.protobuf.GeneratedMessageV3.writeString(output, 3, denseShapeTensorName_);
+      }
+      unknownFields.writeTo(output);
+    }
+
+    @java.lang.Override
+    public int getSerializedSize() {
+      int size = memoizedSize;
+      if (size != -1) return size;
+
+      size = 0;
+      if (!getValuesTensorNameBytes().isEmpty()) {
+        size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, valuesTensorName_);
+      }
+      if (!getIndicesTensorNameBytes().isEmpty()) {
+        size += com.google.protobuf.GeneratedMessageV3.computeStringSize(2, indicesTensorName_);
+      }
+      if (!getDenseShapeTensorNameBytes().isEmpty()) {
+        size += com.google.protobuf.GeneratedMessageV3.computeStringSize(3, denseShapeTensorName_);
+      }
+      size += unknownFields.getSerializedSize();
+      memoizedSize = size;
+      return size;
+    }
+
+    @java.lang.Override
+    public boolean equals(final java.lang.Object obj) {
+      if (obj == this) {
+       return true;
+      }
+      if (!(obj instanceof org.tensorflow.proto.framework.TensorInfo.CooSparse)) {
+        return super.equals(obj);
+      }
+      org.tensorflow.proto.framework.TensorInfo.CooSparse other = (org.tensorflow.proto.framework.TensorInfo.CooSparse) obj;
+
+      if (!getValuesTensorName()
+          .equals(other.getValuesTensorName())) return false;
+      if (!getIndicesTensorName()
+          .equals(other.getIndicesTensorName())) return false;
+      if (!getDenseShapeTensorName()
+          .equals(other.getDenseShapeTensorName())) return false;
+      if (!unknownFields.equals(other.unknownFields)) return false;
+      return true;
+    }
+
+    @java.lang.Override
+    public int hashCode() {
+      if (memoizedHashCode != 0) {
+        return memoizedHashCode;
+      }
+      int hash = 41;
+      hash = (19 * hash) + getDescriptor().hashCode();
+      hash = (37 * hash) + VALUES_TENSOR_NAME_FIELD_NUMBER;
+      hash = (53 * hash) + getValuesTensorName().hashCode();
+      hash = (37 * hash) + INDICES_TENSOR_NAME_FIELD_NUMBER;
+      hash = (53 * hash) + getIndicesTensorName().hashCode();
+      hash = (37 * hash) + DENSE_SHAPE_TENSOR_NAME_FIELD_NUMBER;
+      hash = (53 * hash) + getDenseShapeTensorName().hashCode();
+      hash = (29 * hash) + unknownFields.hashCode();
+      memoizedHashCode = hash;
+      return hash;
+    }
+
+    public static org.tensorflow.proto.framework.TensorInfo.CooSparse parseFrom(
+        java.nio.ByteBuffer data)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data);
+    }
+    public static org.tensorflow.proto.framework.TensorInfo.CooSparse parseFrom(
+        java.nio.ByteBuffer data,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.TensorInfo.CooSparse parseFrom(
+        com.google.protobuf.ByteString data)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data);
+    }
+    public static org.tensorflow.proto.framework.TensorInfo.CooSparse parseFrom(
+        com.google.protobuf.ByteString data,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.TensorInfo.CooSparse parseFrom(byte[] data)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data);
+    }
+    public static org.tensorflow.proto.framework.TensorInfo.CooSparse parseFrom(
+        byte[] data,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.TensorInfo.CooSparse parseFrom(java.io.InputStream input)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input);
+    }
+    public static org.tensorflow.proto.framework.TensorInfo.CooSparse parseFrom(
+        java.io.InputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.TensorInfo.CooSparse parseDelimitedFrom(java.io.InputStream input)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseDelimitedWithIOException(PARSER, input);
+    }
+    public static org.tensorflow.proto.framework.TensorInfo.CooSparse parseDelimitedFrom(
+        java.io.InputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.TensorInfo.CooSparse parseFrom(
+        com.google.protobuf.CodedInputStream input)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input);
+    }
+    public static org.tensorflow.proto.framework.TensorInfo.CooSparse parseFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input, extensionRegistry);
+    }
+
+    @java.lang.Override
+    public Builder newBuilderForType() { return newBuilder(); }
+    public static Builder newBuilder() {
+      return DEFAULT_INSTANCE.toBuilder();
+    }
+    public static Builder newBuilder(org.tensorflow.proto.framework.TensorInfo.CooSparse prototype) {
+      return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+    }
+    @java.lang.Override
+    public Builder toBuilder() {
+      return this == DEFAULT_INSTANCE
+          ? new Builder() : new Builder().mergeFrom(this);
+    }
+
+    @java.lang.Override
+    protected Builder newBuilderForType(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      Builder builder = new Builder(parent);
+      return builder;
+    }
+    /**
+     * <pre>
+     * For sparse tensors, The COO encoding stores a triple of values, indices,
+     * and shape.
+     * </pre>
+     *
+     * Protobuf type {@code tensorflow.TensorInfo.CooSparse}
+     */
+    public static final class Builder extends
+        com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+        // @@protoc_insertion_point(builder_implements:tensorflow.TensorInfo.CooSparse)
+        org.tensorflow.proto.framework.TensorInfo.CooSparseOrBuilder {
+      public static final com.google.protobuf.Descriptors.Descriptor
+          getDescriptor() {
+        return org.tensorflow.proto.framework.MetaGraphProtos.internal_static_tensorflow_TensorInfo_CooSparse_descriptor;
+      }
+
+      @java.lang.Override
+      protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+          internalGetFieldAccessorTable() {
+        return org.tensorflow.proto.framework.MetaGraphProtos.internal_static_tensorflow_TensorInfo_CooSparse_fieldAccessorTable
+            .ensureFieldAccessorsInitialized(
+                org.tensorflow.proto.framework.TensorInfo.CooSparse.class, org.tensorflow.proto.framework.TensorInfo.CooSparse.Builder.class);
+      }
+
+      // Construct using org.tensorflow.proto.framework.TensorInfo.CooSparse.newBuilder()
+      private Builder() {
+        maybeForceBuilderInitialization();
+      }
+
+      private Builder(
+          com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+        super(parent);
+        maybeForceBuilderInitialization();
+      }
+      private void maybeForceBuilderInitialization() {
+        if (com.google.protobuf.GeneratedMessageV3
+                .alwaysUseFieldBuilders) {
+        }
+      }
+      @java.lang.Override
+      public Builder clear() {
+        super.clear();
+        valuesTensorName_ = "";
+
+        indicesTensorName_ = "";
+
+        denseShapeTensorName_ = "";
+
+        return this;
+      }
+
+      @java.lang.Override
+      public com.google.protobuf.Descriptors.Descriptor
+          getDescriptorForType() {
+        return org.tensorflow.proto.framework.MetaGraphProtos.internal_static_tensorflow_TensorInfo_CooSparse_descriptor;
+      }
+
+      @java.lang.Override
+      public org.tensorflow.proto.framework.TensorInfo.CooSparse getDefaultInstanceForType() {
+        return org.tensorflow.proto.framework.TensorInfo.CooSparse.getDefaultInstance();
+      }
+
+      @java.lang.Override
+      public org.tensorflow.proto.framework.TensorInfo.CooSparse build() {
+        org.tensorflow.proto.framework.TensorInfo.CooSparse result = buildPartial();
+        if (!result.isInitialized()) {
+          throw newUninitializedMessageException(result);
+        }
+        return result;
+      }
+
+      @java.lang.Override
+      public org.tensorflow.proto.framework.TensorInfo.CooSparse buildPartial() {
+        org.tensorflow.proto.framework.TensorInfo.CooSparse result = new org.tensorflow.proto.framework.TensorInfo.CooSparse(this);
+        result.valuesTensorName_ = valuesTensorName_;
+        result.indicesTensorName_ = indicesTensorName_;
+        result.denseShapeTensorName_ = denseShapeTensorName_;
+        onBuilt();
+        return result;
+      }
+
+      @java.lang.Override
+      public Builder clone() {
+        return super.clone();
+      }
+      @java.lang.Override
+      public Builder setField(
+          com.google.protobuf.Descriptors.FieldDescriptor field,
+          java.lang.Object value) {
+        return super.setField(field, value);
+      }
+      @java.lang.Override
+      public Builder clearField(
+          com.google.protobuf.Descriptors.FieldDescriptor field) {
+        return super.clearField(field);
+      }
+      @java.lang.Override
+      public Builder clearOneof(
+          com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+        return super.clearOneof(oneof);
+      }
+      @java.lang.Override
+      public Builder setRepeatedField(
+          com.google.protobuf.Descriptors.FieldDescriptor field,
+          int index, java.lang.Object value) {
+        return super.setRepeatedField(field, index, value);
+      }
+      @java.lang.Override
+      public Builder addRepeatedField(
+          com.google.protobuf.Descriptors.FieldDescriptor field,
+          java.lang.Object value) {
+        return super.addRepeatedField(field, value);
+      }
+      @java.lang.Override
+      public Builder mergeFrom(com.google.protobuf.Message other) {
+        if (other instanceof org.tensorflow.proto.framework.TensorInfo.CooSparse) {
+          return mergeFrom((org.tensorflow.proto.framework.TensorInfo.CooSparse)other);
+        } else {
+          super.mergeFrom(other);
+          return this;
+        }
+      }
+
+      public Builder mergeFrom(org.tensorflow.proto.framework.TensorInfo.CooSparse other) {
+        if (other == org.tensorflow.proto.framework.TensorInfo.CooSparse.getDefaultInstance()) return this;
+        if (!other.getValuesTensorName().isEmpty()) {
+          valuesTensorName_ = other.valuesTensorName_;
+          onChanged();
+        }
+        if (!other.getIndicesTensorName().isEmpty()) {
+          indicesTensorName_ = other.indicesTensorName_;
+          onChanged();
+        }
+        if (!other.getDenseShapeTensorName().isEmpty()) {
+          denseShapeTensorName_ = other.denseShapeTensorName_;
+          onChanged();
+        }
+        this.mergeUnknownFields(other.unknownFields);
+        onChanged();
+        return this;
+      }
+
+      @java.lang.Override
+      public final boolean isInitialized() {
+        return true;
+      }
+
+      @java.lang.Override
+      public Builder mergeFrom(
+          com.google.protobuf.CodedInputStream input,
+          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+          throws java.io.IOException {
+        org.tensorflow.proto.framework.TensorInfo.CooSparse parsedMessage = null;
+        try {
+          parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+        } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+          parsedMessage = (org.tensorflow.proto.framework.TensorInfo.CooSparse) e.getUnfinishedMessage();
+          throw e.unwrapIOException();
+        } finally {
+          if (parsedMessage != null) {
+            mergeFrom(parsedMessage);
+          }
+        }
+        return this;
+      }
+
+      private java.lang.Object valuesTensorName_ = "";
+      /**
+       * <pre>
+       * The shape of the values Tensor is [?].  Its dtype must be the dtype of
+       * the SparseTensor as a whole, given in the enclosing TensorInfo.
+       * </pre>
+       *
+       * <code>string values_tensor_name = 1;</code>
+       */
+      public java.lang.String getValuesTensorName() {
+        java.lang.Object ref = valuesTensorName_;
+        if (!(ref instanceof java.lang.String)) {
+          com.google.protobuf.ByteString bs =
+              (com.google.protobuf.ByteString) ref;
+          java.lang.String s = bs.toStringUtf8();
+          valuesTensorName_ = s;
+          return s;
+        } else {
+          return (java.lang.String) ref;
+        }
+      }
+      /**
+       * <pre>
+       * The shape of the values Tensor is [?].  Its dtype must be the dtype of
+       * the SparseTensor as a whole, given in the enclosing TensorInfo.
+       * </pre>
+       *
+       * <code>string values_tensor_name = 1;</code>
+       */
+      public com.google.protobuf.ByteString
+          getValuesTensorNameBytes() {
+        java.lang.Object ref = valuesTensorName_;
+        if (ref instanceof String) {
+          com.google.protobuf.ByteString b = 
+              com.google.protobuf.ByteString.copyFromUtf8(
+                  (java.lang.String) ref);
+          valuesTensorName_ = b;
+          return b;
+        } else {
+          return (com.google.protobuf.ByteString) ref;
+        }
+      }
+      /**
+       * <pre>
+       * The shape of the values Tensor is [?].  Its dtype must be the dtype of
+       * the SparseTensor as a whole, given in the enclosing TensorInfo.
+       * </pre>
+       *
+       * <code>string values_tensor_name = 1;</code>
+       */
+      public Builder setValuesTensorName(
+          java.lang.String value) {
+        if (value == null) {
+    throw new NullPointerException();
+  }
+  
+        valuesTensorName_ = value;
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * The shape of the values Tensor is [?].  Its dtype must be the dtype of
+       * the SparseTensor as a whole, given in the enclosing TensorInfo.
+       * </pre>
+       *
+       * <code>string values_tensor_name = 1;</code>
+       */
+      public Builder clearValuesTensorName() {
+        
+        valuesTensorName_ = getDefaultInstance().getValuesTensorName();
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * The shape of the values Tensor is [?].  Its dtype must be the dtype of
+       * the SparseTensor as a whole, given in the enclosing TensorInfo.
+       * </pre>
+       *
+       * <code>string values_tensor_name = 1;</code>
+       */
+      public Builder setValuesTensorNameBytes(
+          com.google.protobuf.ByteString value) {
+        if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+        
+        valuesTensorName_ = value;
+        onChanged();
+        return this;
+      }
+
+      private java.lang.Object indicesTensorName_ = "";
+      /**
+       * <pre>
+       * The indices Tensor must have dtype int64 and shape [?, ?].
+       * </pre>
+       *
+       * <code>string indices_tensor_name = 2;</code>
+       */
+      public java.lang.String getIndicesTensorName() {
+        java.lang.Object ref = indicesTensorName_;
+        if (!(ref instanceof java.lang.String)) {
+          com.google.protobuf.ByteString bs =
+              (com.google.protobuf.ByteString) ref;
+          java.lang.String s = bs.toStringUtf8();
+          indicesTensorName_ = s;
+          return s;
+        } else {
+          return (java.lang.String) ref;
+        }
+      }
+      /**
+       * <pre>
+       * The indices Tensor must have dtype int64 and shape [?, ?].
+       * </pre>
+       *
+       * <code>string indices_tensor_name = 2;</code>
+       */
+      public com.google.protobuf.ByteString
+          getIndicesTensorNameBytes() {
+        java.lang.Object ref = indicesTensorName_;
+        if (ref instanceof String) {
+          com.google.protobuf.ByteString b = 
+              com.google.protobuf.ByteString.copyFromUtf8(
+                  (java.lang.String) ref);
+          indicesTensorName_ = b;
+          return b;
+        } else {
+          return (com.google.protobuf.ByteString) ref;
+        }
+      }
+      /**
+       * <pre>
+       * The indices Tensor must have dtype int64 and shape [?, ?].
+       * </pre>
+       *
+       * <code>string indices_tensor_name = 2;</code>
+       */
+      public Builder setIndicesTensorName(
+          java.lang.String value) {
+        if (value == null) {
+    throw new NullPointerException();
+  }
+  
+        indicesTensorName_ = value;
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * The indices Tensor must have dtype int64 and shape [?, ?].
+       * </pre>
+       *
+       * <code>string indices_tensor_name = 2;</code>
+       */
+      public Builder clearIndicesTensorName() {
+        
+        indicesTensorName_ = getDefaultInstance().getIndicesTensorName();
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * The indices Tensor must have dtype int64 and shape [?, ?].
+       * </pre>
+       *
+       * <code>string indices_tensor_name = 2;</code>
+       */
+      public Builder setIndicesTensorNameBytes(
+          com.google.protobuf.ByteString value) {
+        if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+        
+        indicesTensorName_ = value;
+        onChanged();
+        return this;
+      }
+
+      private java.lang.Object denseShapeTensorName_ = "";
+      /**
+       * <pre>
+       * The dynamic logical shape represented by the SparseTensor is recorded in
+       * the Tensor referenced here.  It must have dtype int64 and shape [?].
+       * </pre>
+       *
+       * <code>string dense_shape_tensor_name = 3;</code>
+       */
+      public java.lang.String getDenseShapeTensorName() {
+        java.lang.Object ref = denseShapeTensorName_;
+        if (!(ref instanceof java.lang.String)) {
+          com.google.protobuf.ByteString bs =
+              (com.google.protobuf.ByteString) ref;
+          java.lang.String s = bs.toStringUtf8();
+          denseShapeTensorName_ = s;
+          return s;
+        } else {
+          return (java.lang.String) ref;
+        }
+      }
+      /**
+       * <pre>
+       * The dynamic logical shape represented by the SparseTensor is recorded in
+       * the Tensor referenced here.  It must have dtype int64 and shape [?].
+       * </pre>
+       *
+       * <code>string dense_shape_tensor_name = 3;</code>
+       */
+      public com.google.protobuf.ByteString
+          getDenseShapeTensorNameBytes() {
+        java.lang.Object ref = denseShapeTensorName_;
+        if (ref instanceof String) {
+          com.google.protobuf.ByteString b = 
+              com.google.protobuf.ByteString.copyFromUtf8(
+                  (java.lang.String) ref);
+          denseShapeTensorName_ = b;
+          return b;
+        } else {
+          return (com.google.protobuf.ByteString) ref;
+        }
+      }
+      /**
+       * <pre>
+       * The dynamic logical shape represented by the SparseTensor is recorded in
+       * the Tensor referenced here.  It must have dtype int64 and shape [?].
+       * </pre>
+       *
+       * <code>string dense_shape_tensor_name = 3;</code>
+       */
+      public Builder setDenseShapeTensorName(
+          java.lang.String value) {
+        if (value == null) {
+    throw new NullPointerException();
+  }
+  
+        denseShapeTensorName_ = value;
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * The dynamic logical shape represented by the SparseTensor is recorded in
+       * the Tensor referenced here.  It must have dtype int64 and shape [?].
+       * </pre>
+       *
+       * <code>string dense_shape_tensor_name = 3;</code>
+       */
+      public Builder clearDenseShapeTensorName() {
+        
+        denseShapeTensorName_ = getDefaultInstance().getDenseShapeTensorName();
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * The dynamic logical shape represented by the SparseTensor is recorded in
+       * the Tensor referenced here.  It must have dtype int64 and shape [?].
+       * </pre>
+       *
+       * <code>string dense_shape_tensor_name = 3;</code>
+       */
+      public Builder setDenseShapeTensorNameBytes(
+          com.google.protobuf.ByteString value) {
+        if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+        
+        denseShapeTensorName_ = value;
+        onChanged();
+        return this;
+      }
+      @java.lang.Override
+      public final Builder setUnknownFields(
+          final com.google.protobuf.UnknownFieldSet unknownFields) {
+        return super.setUnknownFields(unknownFields);
+      }
+
+      @java.lang.Override
+      public final Builder mergeUnknownFields(
+          final com.google.protobuf.UnknownFieldSet unknownFields) {
+        return super.mergeUnknownFields(unknownFields);
+      }
+
+
+      // @@protoc_insertion_point(builder_scope:tensorflow.TensorInfo.CooSparse)
+    }
+
+    // @@protoc_insertion_point(class_scope:tensorflow.TensorInfo.CooSparse)
+    private static final org.tensorflow.proto.framework.TensorInfo.CooSparse DEFAULT_INSTANCE;
+    static {
+      DEFAULT_INSTANCE = new org.tensorflow.proto.framework.TensorInfo.CooSparse();
+    }
+
+    public static org.tensorflow.proto.framework.TensorInfo.CooSparse getDefaultInstance() {
+      return DEFAULT_INSTANCE;
+    }
+
+    private static final com.google.protobuf.Parser<CooSparse>
+        PARSER = new com.google.protobuf.AbstractParser<CooSparse>() {
+      @java.lang.Override
+      public CooSparse parsePartialFrom(
+          com.google.protobuf.CodedInputStream input,
+          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+          throws com.google.protobuf.InvalidProtocolBufferException {
+        return new CooSparse(input, extensionRegistry);
+      }
+    };
+
+    public static com.google.protobuf.Parser<CooSparse> parser() {
+      return PARSER;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Parser<CooSparse> getParserForType() {
+      return PARSER;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.TensorInfo.CooSparse getDefaultInstanceForType() {
+      return DEFAULT_INSTANCE;
+    }
+
+  }
+
+  public interface CompositeTensorOrBuilder extends
+      // @@protoc_insertion_point(interface_extends:tensorflow.TensorInfo.CompositeTensor)
+      com.google.protobuf.MessageOrBuilder {
+
+    /**
+     * <pre>
+     * The serialized TypeSpec for the composite tensor.
+     * </pre>
+     *
+     * <code>.tensorflow.TypeSpecProto type_spec = 1;</code>
+     */
+    boolean hasTypeSpec();
+    /**
+     * <pre>
+     * The serialized TypeSpec for the composite tensor.
+     * </pre>
+     *
+     * <code>.tensorflow.TypeSpecProto type_spec = 1;</code>
+     */
+    org.tensorflow.proto.framework.TypeSpecProto getTypeSpec();
+    /**
+     * <pre>
+     * The serialized TypeSpec for the composite tensor.
+     * </pre>
+     *
+     * <code>.tensorflow.TypeSpecProto type_spec = 1;</code>
+     */
+    org.tensorflow.proto.framework.TypeSpecProtoOrBuilder getTypeSpecOrBuilder();
+
+    /**
+     * <pre>
+     * A TensorInfo for each flattened component tensor.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorInfo components = 2;</code>
+     */
+    java.util.List<org.tensorflow.proto.framework.TensorInfo> 
+        getComponentsList();
+    /**
+     * <pre>
+     * A TensorInfo for each flattened component tensor.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorInfo components = 2;</code>
+     */
+    org.tensorflow.proto.framework.TensorInfo getComponents(int index);
+    /**
+     * <pre>
+     * A TensorInfo for each flattened component tensor.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorInfo components = 2;</code>
+     */
+    int getComponentsCount();
+    /**
+     * <pre>
+     * A TensorInfo for each flattened component tensor.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorInfo components = 2;</code>
+     */
+    java.util.List<? extends org.tensorflow.proto.framework.TensorInfoOrBuilder> 
+        getComponentsOrBuilderList();
+    /**
+     * <pre>
+     * A TensorInfo for each flattened component tensor.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorInfo components = 2;</code>
+     */
+    org.tensorflow.proto.framework.TensorInfoOrBuilder getComponentsOrBuilder(
+        int index);
+  }
+  /**
+   * <pre>
+   * Generic encoding for composite tensors.
+   * </pre>
+   *
+   * Protobuf type {@code tensorflow.TensorInfo.CompositeTensor}
+   */
+  public  static final class CompositeTensor extends
+      com.google.protobuf.GeneratedMessageV3 implements
+      // @@protoc_insertion_point(message_implements:tensorflow.TensorInfo.CompositeTensor)
+      CompositeTensorOrBuilder {
+  private static final long serialVersionUID = 0L;
+    // Use CompositeTensor.newBuilder() to construct.
+    private CompositeTensor(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+      super(builder);
+    }
+    private CompositeTensor() {
+      components_ = java.util.Collections.emptyList();
+    }
+
+    @java.lang.Override
+    @SuppressWarnings({"unused"})
+    protected java.lang.Object newInstance(
+        UnusedPrivateParameter unused) {
+      return new CompositeTensor();
+    }
+
+    @java.lang.Override
+    public final com.google.protobuf.UnknownFieldSet
+    getUnknownFields() {
+      return this.unknownFields;
+    }
+    private CompositeTensor(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      this();
+      if (extensionRegistry == null) {
+        throw new java.lang.NullPointerException();
+      }
+      int mutable_bitField0_ = 0;
+      com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+          com.google.protobuf.UnknownFieldSet.newBuilder();
+      try {
+        boolean done = false;
+        while (!done) {
+          int tag = input.readTag();
+          switch (tag) {
+            case 0:
+              done = true;
+              break;
+            case 10: {
+              org.tensorflow.proto.framework.TypeSpecProto.Builder subBuilder = null;
+              if (typeSpec_ != null) {
+                subBuilder = typeSpec_.toBuilder();
+              }
+              typeSpec_ = input.readMessage(org.tensorflow.proto.framework.TypeSpecProto.parser(), extensionRegistry);
+              if (subBuilder != null) {
+                subBuilder.mergeFrom(typeSpec_);
+                typeSpec_ = subBuilder.buildPartial();
+              }
+
+              break;
+            }
+            case 18: {
+              if (!((mutable_bitField0_ & 0x00000001) != 0)) {
+                components_ = new java.util.ArrayList<org.tensorflow.proto.framework.TensorInfo>();
+                mutable_bitField0_ |= 0x00000001;
+              }
+              components_.add(
+                  input.readMessage(org.tensorflow.proto.framework.TensorInfo.parser(), extensionRegistry));
+              break;
+            }
+            default: {
+              if (!parseUnknownField(
+                  input, unknownFields, extensionRegistry, tag)) {
+                done = true;
+              }
+              break;
+            }
+          }
+        }
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        throw e.setUnfinishedMessage(this);
+      } catch (java.io.IOException e) {
+        throw new com.google.protobuf.InvalidProtocolBufferException(
+            e).setUnfinishedMessage(this);
+      } finally {
+        if (((mutable_bitField0_ & 0x00000001) != 0)) {
+          components_ = java.util.Collections.unmodifiableList(components_);
+        }
+        this.unknownFields = unknownFields.build();
+        makeExtensionsImmutable();
+      }
+    }
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.MetaGraphProtos.internal_static_tensorflow_TensorInfo_CompositeTensor_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.MetaGraphProtos.internal_static_tensorflow_TensorInfo_CompositeTensor_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.TensorInfo.CompositeTensor.class, org.tensorflow.proto.framework.TensorInfo.CompositeTensor.Builder.class);
+    }
+
+    public static final int TYPE_SPEC_FIELD_NUMBER = 1;
+    private org.tensorflow.proto.framework.TypeSpecProto typeSpec_;
+    /**
+     * <pre>
+     * The serialized TypeSpec for the composite tensor.
+     * </pre>
+     *
+     * <code>.tensorflow.TypeSpecProto type_spec = 1;</code>
+     */
+    public boolean hasTypeSpec() {
+      return typeSpec_ != null;
+    }
+    /**
+     * <pre>
+     * The serialized TypeSpec for the composite tensor.
+     * </pre>
+     *
+     * <code>.tensorflow.TypeSpecProto type_spec = 1;</code>
+     */
+    public org.tensorflow.proto.framework.TypeSpecProto getTypeSpec() {
+      return typeSpec_ == null ? org.tensorflow.proto.framework.TypeSpecProto.getDefaultInstance() : typeSpec_;
+    }
+    /**
+     * <pre>
+     * The serialized TypeSpec for the composite tensor.
+     * </pre>
+     *
+     * <code>.tensorflow.TypeSpecProto type_spec = 1;</code>
+     */
+    public org.tensorflow.proto.framework.TypeSpecProtoOrBuilder getTypeSpecOrBuilder() {
+      return getTypeSpec();
+    }
+
+    public static final int COMPONENTS_FIELD_NUMBER = 2;
+    private java.util.List<org.tensorflow.proto.framework.TensorInfo> components_;
+    /**
+     * <pre>
+     * A TensorInfo for each flattened component tensor.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorInfo components = 2;</code>
+     */
+    public java.util.List<org.tensorflow.proto.framework.TensorInfo> getComponentsList() {
+      return components_;
+    }
+    /**
+     * <pre>
+     * A TensorInfo for each flattened component tensor.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorInfo components = 2;</code>
+     */
+    public java.util.List<? extends org.tensorflow.proto.framework.TensorInfoOrBuilder> 
+        getComponentsOrBuilderList() {
+      return components_;
+    }
+    /**
+     * <pre>
+     * A TensorInfo for each flattened component tensor.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorInfo components = 2;</code>
+     */
+    public int getComponentsCount() {
+      return components_.size();
+    }
+    /**
+     * <pre>
+     * A TensorInfo for each flattened component tensor.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorInfo components = 2;</code>
+     */
+    public org.tensorflow.proto.framework.TensorInfo getComponents(int index) {
+      return components_.get(index);
+    }
+    /**
+     * <pre>
+     * A TensorInfo for each flattened component tensor.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorInfo components = 2;</code>
+     */
+    public org.tensorflow.proto.framework.TensorInfoOrBuilder getComponentsOrBuilder(
+        int index) {
+      return components_.get(index);
+    }
+
+    private byte memoizedIsInitialized = -1;
+    @java.lang.Override
+    public final boolean isInitialized() {
+      byte isInitialized = memoizedIsInitialized;
+      if (isInitialized == 1) return true;
+      if (isInitialized == 0) return false;
+
+      memoizedIsInitialized = 1;
+      return true;
+    }
+
+    @java.lang.Override
+    public void writeTo(com.google.protobuf.CodedOutputStream output)
+                        throws java.io.IOException {
+      if (typeSpec_ != null) {
+        output.writeMessage(1, getTypeSpec());
+      }
+      for (int i = 0; i < components_.size(); i++) {
+        output.writeMessage(2, components_.get(i));
+      }
+      unknownFields.writeTo(output);
+    }
+
+    @java.lang.Override
+    public int getSerializedSize() {
+      int size = memoizedSize;
+      if (size != -1) return size;
+
+      size = 0;
+      if (typeSpec_ != null) {
+        size += com.google.protobuf.CodedOutputStream
+          .computeMessageSize(1, getTypeSpec());
+      }
+      for (int i = 0; i < components_.size(); i++) {
+        size += com.google.protobuf.CodedOutputStream
+          .computeMessageSize(2, components_.get(i));
+      }
+      size += unknownFields.getSerializedSize();
+      memoizedSize = size;
+      return size;
+    }
+
+    @java.lang.Override
+    public boolean equals(final java.lang.Object obj) {
+      if (obj == this) {
+       return true;
+      }
+      if (!(obj instanceof org.tensorflow.proto.framework.TensorInfo.CompositeTensor)) {
+        return super.equals(obj);
+      }
+      org.tensorflow.proto.framework.TensorInfo.CompositeTensor other = (org.tensorflow.proto.framework.TensorInfo.CompositeTensor) obj;
+
+      if (hasTypeSpec() != other.hasTypeSpec()) return false;
+      if (hasTypeSpec()) {
+        if (!getTypeSpec()
+            .equals(other.getTypeSpec())) return false;
+      }
+      if (!getComponentsList()
+          .equals(other.getComponentsList())) return false;
+      if (!unknownFields.equals(other.unknownFields)) return false;
+      return true;
+    }
+
+    @java.lang.Override
+    public int hashCode() {
+      if (memoizedHashCode != 0) {
+        return memoizedHashCode;
+      }
+      int hash = 41;
+      hash = (19 * hash) + getDescriptor().hashCode();
+      if (hasTypeSpec()) {
+        hash = (37 * hash) + TYPE_SPEC_FIELD_NUMBER;
+        hash = (53 * hash) + getTypeSpec().hashCode();
+      }
+      if (getComponentsCount() > 0) {
+        hash = (37 * hash) + COMPONENTS_FIELD_NUMBER;
+        hash = (53 * hash) + getComponentsList().hashCode();
+      }
+      hash = (29 * hash) + unknownFields.hashCode();
+      memoizedHashCode = hash;
+      return hash;
+    }
+
+    public static org.tensorflow.proto.framework.TensorInfo.CompositeTensor parseFrom(
+        java.nio.ByteBuffer data)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data);
+    }
+    public static org.tensorflow.proto.framework.TensorInfo.CompositeTensor parseFrom(
+        java.nio.ByteBuffer data,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.TensorInfo.CompositeTensor parseFrom(
+        com.google.protobuf.ByteString data)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data);
+    }
+    public static org.tensorflow.proto.framework.TensorInfo.CompositeTensor parseFrom(
+        com.google.protobuf.ByteString data,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.TensorInfo.CompositeTensor parseFrom(byte[] data)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data);
+    }
+    public static org.tensorflow.proto.framework.TensorInfo.CompositeTensor parseFrom(
+        byte[] data,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.TensorInfo.CompositeTensor parseFrom(java.io.InputStream input)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input);
+    }
+    public static org.tensorflow.proto.framework.TensorInfo.CompositeTensor parseFrom(
+        java.io.InputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.TensorInfo.CompositeTensor parseDelimitedFrom(java.io.InputStream input)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseDelimitedWithIOException(PARSER, input);
+    }
+    public static org.tensorflow.proto.framework.TensorInfo.CompositeTensor parseDelimitedFrom(
+        java.io.InputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.TensorInfo.CompositeTensor parseFrom(
+        com.google.protobuf.CodedInputStream input)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input);
+    }
+    public static org.tensorflow.proto.framework.TensorInfo.CompositeTensor parseFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input, extensionRegistry);
+    }
+
+    @java.lang.Override
+    public Builder newBuilderForType() { return newBuilder(); }
+    public static Builder newBuilder() {
+      return DEFAULT_INSTANCE.toBuilder();
+    }
+    public static Builder newBuilder(org.tensorflow.proto.framework.TensorInfo.CompositeTensor prototype) {
+      return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+    }
+    @java.lang.Override
+    public Builder toBuilder() {
+      return this == DEFAULT_INSTANCE
+          ? new Builder() : new Builder().mergeFrom(this);
+    }
+
+    @java.lang.Override
+    protected Builder newBuilderForType(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      Builder builder = new Builder(parent);
+      return builder;
+    }
+    /**
+     * <pre>
+     * Generic encoding for composite tensors.
+     * </pre>
+     *
+     * Protobuf type {@code tensorflow.TensorInfo.CompositeTensor}
+     */
+    public static final class Builder extends
+        com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+        // @@protoc_insertion_point(builder_implements:tensorflow.TensorInfo.CompositeTensor)
+        org.tensorflow.proto.framework.TensorInfo.CompositeTensorOrBuilder {
+      public static final com.google.protobuf.Descriptors.Descriptor
+          getDescriptor() {
+        return org.tensorflow.proto.framework.MetaGraphProtos.internal_static_tensorflow_TensorInfo_CompositeTensor_descriptor;
+      }
+
+      @java.lang.Override
+      protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+          internalGetFieldAccessorTable() {
+        return org.tensorflow.proto.framework.MetaGraphProtos.internal_static_tensorflow_TensorInfo_CompositeTensor_fieldAccessorTable
+            .ensureFieldAccessorsInitialized(
+                org.tensorflow.proto.framework.TensorInfo.CompositeTensor.class, org.tensorflow.proto.framework.TensorInfo.CompositeTensor.Builder.class);
+      }
+
+      // Construct using org.tensorflow.proto.framework.TensorInfo.CompositeTensor.newBuilder()
+      private Builder() {
+        maybeForceBuilderInitialization();
+      }
+
+      private Builder(
+          com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+        super(parent);
+        maybeForceBuilderInitialization();
+      }
+      private void maybeForceBuilderInitialization() {
+        if (com.google.protobuf.GeneratedMessageV3
+                .alwaysUseFieldBuilders) {
+          getComponentsFieldBuilder();
+        }
+      }
+      @java.lang.Override
+      public Builder clear() {
+        super.clear();
+        if (typeSpecBuilder_ == null) {
+          typeSpec_ = null;
+        } else {
+          typeSpec_ = null;
+          typeSpecBuilder_ = null;
+        }
+        if (componentsBuilder_ == null) {
+          components_ = java.util.Collections.emptyList();
+          bitField0_ = (bitField0_ & ~0x00000001);
+        } else {
+          componentsBuilder_.clear();
+        }
+        return this;
+      }
+
+      @java.lang.Override
+      public com.google.protobuf.Descriptors.Descriptor
+          getDescriptorForType() {
+        return org.tensorflow.proto.framework.MetaGraphProtos.internal_static_tensorflow_TensorInfo_CompositeTensor_descriptor;
+      }
+
+      @java.lang.Override
+      public org.tensorflow.proto.framework.TensorInfo.CompositeTensor getDefaultInstanceForType() {
+        return org.tensorflow.proto.framework.TensorInfo.CompositeTensor.getDefaultInstance();
+      }
+
+      @java.lang.Override
+      public org.tensorflow.proto.framework.TensorInfo.CompositeTensor build() {
+        org.tensorflow.proto.framework.TensorInfo.CompositeTensor result = buildPartial();
+        if (!result.isInitialized()) {
+          throw newUninitializedMessageException(result);
+        }
+        return result;
+      }
+
+      @java.lang.Override
+      public org.tensorflow.proto.framework.TensorInfo.CompositeTensor buildPartial() {
+        org.tensorflow.proto.framework.TensorInfo.CompositeTensor result = new org.tensorflow.proto.framework.TensorInfo.CompositeTensor(this);
+        int from_bitField0_ = bitField0_;
+        if (typeSpecBuilder_ == null) {
+          result.typeSpec_ = typeSpec_;
+        } else {
+          result.typeSpec_ = typeSpecBuilder_.build();
+        }
+        if (componentsBuilder_ == null) {
+          if (((bitField0_ & 0x00000001) != 0)) {
+            components_ = java.util.Collections.unmodifiableList(components_);
+            bitField0_ = (bitField0_ & ~0x00000001);
+          }
+          result.components_ = components_;
+        } else {
+          result.components_ = componentsBuilder_.build();
+        }
+        onBuilt();
+        return result;
+      }
+
+      @java.lang.Override
+      public Builder clone() {
+        return super.clone();
+      }
+      @java.lang.Override
+      public Builder setField(
+          com.google.protobuf.Descriptors.FieldDescriptor field,
+          java.lang.Object value) {
+        return super.setField(field, value);
+      }
+      @java.lang.Override
+      public Builder clearField(
+          com.google.protobuf.Descriptors.FieldDescriptor field) {
+        return super.clearField(field);
+      }
+      @java.lang.Override
+      public Builder clearOneof(
+          com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+        return super.clearOneof(oneof);
+      }
+      @java.lang.Override
+      public Builder setRepeatedField(
+          com.google.protobuf.Descriptors.FieldDescriptor field,
+          int index, java.lang.Object value) {
+        return super.setRepeatedField(field, index, value);
+      }
+      @java.lang.Override
+      public Builder addRepeatedField(
+          com.google.protobuf.Descriptors.FieldDescriptor field,
+          java.lang.Object value) {
+        return super.addRepeatedField(field, value);
+      }
+      @java.lang.Override
+      public Builder mergeFrom(com.google.protobuf.Message other) {
+        if (other instanceof org.tensorflow.proto.framework.TensorInfo.CompositeTensor) {
+          return mergeFrom((org.tensorflow.proto.framework.TensorInfo.CompositeTensor)other);
+        } else {
+          super.mergeFrom(other);
+          return this;
+        }
+      }
+
+      public Builder mergeFrom(org.tensorflow.proto.framework.TensorInfo.CompositeTensor other) {
+        if (other == org.tensorflow.proto.framework.TensorInfo.CompositeTensor.getDefaultInstance()) return this;
+        if (other.hasTypeSpec()) {
+          mergeTypeSpec(other.getTypeSpec());
+        }
+        if (componentsBuilder_ == null) {
+          if (!other.components_.isEmpty()) {
+            if (components_.isEmpty()) {
+              components_ = other.components_;
+              bitField0_ = (bitField0_ & ~0x00000001);
+            } else {
+              ensureComponentsIsMutable();
+              components_.addAll(other.components_);
+            }
+            onChanged();
+          }
+        } else {
+          if (!other.components_.isEmpty()) {
+            if (componentsBuilder_.isEmpty()) {
+              componentsBuilder_.dispose();
+              componentsBuilder_ = null;
+              components_ = other.components_;
+              bitField0_ = (bitField0_ & ~0x00000001);
+              componentsBuilder_ = 
+                com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
+                   getComponentsFieldBuilder() : null;
+            } else {
+              componentsBuilder_.addAllMessages(other.components_);
+            }
+          }
+        }
+        this.mergeUnknownFields(other.unknownFields);
+        onChanged();
+        return this;
+      }
+
+      @java.lang.Override
+      public final boolean isInitialized() {
+        return true;
+      }
+
+      @java.lang.Override
+      public Builder mergeFrom(
+          com.google.protobuf.CodedInputStream input,
+          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+          throws java.io.IOException {
+        org.tensorflow.proto.framework.TensorInfo.CompositeTensor parsedMessage = null;
+        try {
+          parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+        } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+          parsedMessage = (org.tensorflow.proto.framework.TensorInfo.CompositeTensor) e.getUnfinishedMessage();
+          throw e.unwrapIOException();
+        } finally {
+          if (parsedMessage != null) {
+            mergeFrom(parsedMessage);
+          }
+        }
+        return this;
+      }
+      private int bitField0_;
+
+      private org.tensorflow.proto.framework.TypeSpecProto typeSpec_;
+      private com.google.protobuf.SingleFieldBuilderV3<
+          org.tensorflow.proto.framework.TypeSpecProto, org.tensorflow.proto.framework.TypeSpecProto.Builder, org.tensorflow.proto.framework.TypeSpecProtoOrBuilder> typeSpecBuilder_;
+      /**
+       * <pre>
+       * The serialized TypeSpec for the composite tensor.
+       * </pre>
+       *
+       * <code>.tensorflow.TypeSpecProto type_spec = 1;</code>
+       */
+      public boolean hasTypeSpec() {
+        return typeSpecBuilder_ != null || typeSpec_ != null;
+      }
+      /**
+       * <pre>
+       * The serialized TypeSpec for the composite tensor.
+       * </pre>
+       *
+       * <code>.tensorflow.TypeSpecProto type_spec = 1;</code>
+       */
+      public org.tensorflow.proto.framework.TypeSpecProto getTypeSpec() {
+        if (typeSpecBuilder_ == null) {
+          return typeSpec_ == null ? org.tensorflow.proto.framework.TypeSpecProto.getDefaultInstance() : typeSpec_;
+        } else {
+          return typeSpecBuilder_.getMessage();
+        }
+      }
+      /**
+       * <pre>
+       * The serialized TypeSpec for the composite tensor.
+       * </pre>
+       *
+       * <code>.tensorflow.TypeSpecProto type_spec = 1;</code>
+       */
+      public Builder setTypeSpec(org.tensorflow.proto.framework.TypeSpecProto value) {
+        if (typeSpecBuilder_ == null) {
+          if (value == null) {
+            throw new NullPointerException();
+          }
+          typeSpec_ = value;
+          onChanged();
+        } else {
+          typeSpecBuilder_.setMessage(value);
+        }
+
+        return this;
+      }
+      /**
+       * <pre>
+       * The serialized TypeSpec for the composite tensor.
+       * </pre>
+       *
+       * <code>.tensorflow.TypeSpecProto type_spec = 1;</code>
+       */
+      public Builder setTypeSpec(
+          org.tensorflow.proto.framework.TypeSpecProto.Builder builderForValue) {
+        if (typeSpecBuilder_ == null) {
+          typeSpec_ = builderForValue.build();
+          onChanged();
+        } else {
+          typeSpecBuilder_.setMessage(builderForValue.build());
+        }
+
+        return this;
+      }
+      /**
+       * <pre>
+       * The serialized TypeSpec for the composite tensor.
+       * </pre>
+       *
+       * <code>.tensorflow.TypeSpecProto type_spec = 1;</code>
+       */
+      public Builder mergeTypeSpec(org.tensorflow.proto.framework.TypeSpecProto value) {
+        if (typeSpecBuilder_ == null) {
+          if (typeSpec_ != null) {
+            typeSpec_ =
+              org.tensorflow.proto.framework.TypeSpecProto.newBuilder(typeSpec_).mergeFrom(value).buildPartial();
+          } else {
+            typeSpec_ = value;
+          }
+          onChanged();
+        } else {
+          typeSpecBuilder_.mergeFrom(value);
+        }
+
+        return this;
+      }
+      /**
+       * <pre>
+       * The serialized TypeSpec for the composite tensor.
+       * </pre>
+       *
+       * <code>.tensorflow.TypeSpecProto type_spec = 1;</code>
+       */
+      public Builder clearTypeSpec() {
+        if (typeSpecBuilder_ == null) {
+          typeSpec_ = null;
+          onChanged();
+        } else {
+          typeSpec_ = null;
+          typeSpecBuilder_ = null;
+        }
+
+        return this;
+      }
+      /**
+       * <pre>
+       * The serialized TypeSpec for the composite tensor.
+       * </pre>
+       *
+       * <code>.tensorflow.TypeSpecProto type_spec = 1;</code>
+       */
+      public org.tensorflow.proto.framework.TypeSpecProto.Builder getTypeSpecBuilder() {
+        
+        onChanged();
+        return getTypeSpecFieldBuilder().getBuilder();
+      }
+      /**
+       * <pre>
+       * The serialized TypeSpec for the composite tensor.
+       * </pre>
+       *
+       * <code>.tensorflow.TypeSpecProto type_spec = 1;</code>
+       */
+      public org.tensorflow.proto.framework.TypeSpecProtoOrBuilder getTypeSpecOrBuilder() {
+        if (typeSpecBuilder_ != null) {
+          return typeSpecBuilder_.getMessageOrBuilder();
+        } else {
+          return typeSpec_ == null ?
+              org.tensorflow.proto.framework.TypeSpecProto.getDefaultInstance() : typeSpec_;
+        }
+      }
+      /**
+       * <pre>
+       * The serialized TypeSpec for the composite tensor.
+       * </pre>
+       *
+       * <code>.tensorflow.TypeSpecProto type_spec = 1;</code>
+       */
+      private com.google.protobuf.SingleFieldBuilderV3<
+          org.tensorflow.proto.framework.TypeSpecProto, org.tensorflow.proto.framework.TypeSpecProto.Builder, org.tensorflow.proto.framework.TypeSpecProtoOrBuilder> 
+          getTypeSpecFieldBuilder() {
+        if (typeSpecBuilder_ == null) {
+          typeSpecBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+              org.tensorflow.proto.framework.TypeSpecProto, org.tensorflow.proto.framework.TypeSpecProto.Builder, org.tensorflow.proto.framework.TypeSpecProtoOrBuilder>(
+                  getTypeSpec(),
+                  getParentForChildren(),
+                  isClean());
+          typeSpec_ = null;
+        }
+        return typeSpecBuilder_;
+      }
+
+      private java.util.List<org.tensorflow.proto.framework.TensorInfo> components_ =
+        java.util.Collections.emptyList();
+      private void ensureComponentsIsMutable() {
+        if (!((bitField0_ & 0x00000001) != 0)) {
+          components_ = new java.util.ArrayList<org.tensorflow.proto.framework.TensorInfo>(components_);
+          bitField0_ |= 0x00000001;
+         }
+      }
+
+      private com.google.protobuf.RepeatedFieldBuilderV3<
+          org.tensorflow.proto.framework.TensorInfo, org.tensorflow.proto.framework.TensorInfo.Builder, org.tensorflow.proto.framework.TensorInfoOrBuilder> componentsBuilder_;
+
+      /**
+       * <pre>
+       * A TensorInfo for each flattened component tensor.
+       * </pre>
+       *
+       * <code>repeated .tensorflow.TensorInfo components = 2;</code>
+       */
+      public java.util.List<org.tensorflow.proto.framework.TensorInfo> getComponentsList() {
+        if (componentsBuilder_ == null) {
+          return java.util.Collections.unmodifiableList(components_);
+        } else {
+          return componentsBuilder_.getMessageList();
+        }
+      }
+      /**
+       * <pre>
+       * A TensorInfo for each flattened component tensor.
+       * </pre>
+       *
+       * <code>repeated .tensorflow.TensorInfo components = 2;</code>
+       */
+      public int getComponentsCount() {
+        if (componentsBuilder_ == null) {
+          return components_.size();
+        } else {
+          return componentsBuilder_.getCount();
+        }
+      }
+      /**
+       * <pre>
+       * A TensorInfo for each flattened component tensor.
+       * </pre>
+       *
+       * <code>repeated .tensorflow.TensorInfo components = 2;</code>
+       */
+      public org.tensorflow.proto.framework.TensorInfo getComponents(int index) {
+        if (componentsBuilder_ == null) {
+          return components_.get(index);
+        } else {
+          return componentsBuilder_.getMessage(index);
+        }
+      }
+      /**
+       * <pre>
+       * A TensorInfo for each flattened component tensor.
+       * </pre>
+       *
+       * <code>repeated .tensorflow.TensorInfo components = 2;</code>
+       */
+      public Builder setComponents(
+          int index, org.tensorflow.proto.framework.TensorInfo value) {
+        if (componentsBuilder_ == null) {
+          if (value == null) {
+            throw new NullPointerException();
+          }
+          ensureComponentsIsMutable();
+          components_.set(index, value);
+          onChanged();
+        } else {
+          componentsBuilder_.setMessage(index, value);
+        }
+        return this;
+      }
+      /**
+       * <pre>
+       * A TensorInfo for each flattened component tensor.
+       * </pre>
+       *
+       * <code>repeated .tensorflow.TensorInfo components = 2;</code>
+       */
+      public Builder setComponents(
+          int index, org.tensorflow.proto.framework.TensorInfo.Builder builderForValue) {
+        if (componentsBuilder_ == null) {
+          ensureComponentsIsMutable();
+          components_.set(index, builderForValue.build());
+          onChanged();
+        } else {
+          componentsBuilder_.setMessage(index, builderForValue.build());
+        }
+        return this;
+      }
+      /**
+       * <pre>
+       * A TensorInfo for each flattened component tensor.
+       * </pre>
+       *
+       * <code>repeated .tensorflow.TensorInfo components = 2;</code>
+       */
+      public Builder addComponents(org.tensorflow.proto.framework.TensorInfo value) {
+        if (componentsBuilder_ == null) {
+          if (value == null) {
+            throw new NullPointerException();
+          }
+          ensureComponentsIsMutable();
+          components_.add(value);
+          onChanged();
+        } else {
+          componentsBuilder_.addMessage(value);
+        }
+        return this;
+      }
+      /**
+       * <pre>
+       * A TensorInfo for each flattened component tensor.
+       * </pre>
+       *
+       * <code>repeated .tensorflow.TensorInfo components = 2;</code>
+       */
+      public Builder addComponents(
+          int index, org.tensorflow.proto.framework.TensorInfo value) {
+        if (componentsBuilder_ == null) {
+          if (value == null) {
+            throw new NullPointerException();
+          }
+          ensureComponentsIsMutable();
+          components_.add(index, value);
+          onChanged();
+        } else {
+          componentsBuilder_.addMessage(index, value);
+        }
+        return this;
+      }
+      /**
+       * <pre>
+       * A TensorInfo for each flattened component tensor.
+       * </pre>
+       *
+       * <code>repeated .tensorflow.TensorInfo components = 2;</code>
+       */
+      public Builder addComponents(
+          org.tensorflow.proto.framework.TensorInfo.Builder builderForValue) {
+        if (componentsBuilder_ == null) {
+          ensureComponentsIsMutable();
+          components_.add(builderForValue.build());
+          onChanged();
+        } else {
+          componentsBuilder_.addMessage(builderForValue.build());
+        }
+        return this;
+      }
+      /**
+       * <pre>
+       * A TensorInfo for each flattened component tensor.
+       * </pre>
+       *
+       * <code>repeated .tensorflow.TensorInfo components = 2;</code>
+       */
+      public Builder addComponents(
+          int index, org.tensorflow.proto.framework.TensorInfo.Builder builderForValue) {
+        if (componentsBuilder_ == null) {
+          ensureComponentsIsMutable();
+          components_.add(index, builderForValue.build());
+          onChanged();
+        } else {
+          componentsBuilder_.addMessage(index, builderForValue.build());
+        }
+        return this;
+      }
+      /**
+       * <pre>
+       * A TensorInfo for each flattened component tensor.
+       * </pre>
+       *
+       * <code>repeated .tensorflow.TensorInfo components = 2;</code>
+       */
+      public Builder addAllComponents(
+          java.lang.Iterable<? extends org.tensorflow.proto.framework.TensorInfo> values) {
+        if (componentsBuilder_ == null) {
+          ensureComponentsIsMutable();
+          com.google.protobuf.AbstractMessageLite.Builder.addAll(
+              values, components_);
+          onChanged();
+        } else {
+          componentsBuilder_.addAllMessages(values);
+        }
+        return this;
+      }
+      /**
+       * <pre>
+       * A TensorInfo for each flattened component tensor.
+       * </pre>
+       *
+       * <code>repeated .tensorflow.TensorInfo components = 2;</code>
+       */
+      public Builder clearComponents() {
+        if (componentsBuilder_ == null) {
+          components_ = java.util.Collections.emptyList();
+          bitField0_ = (bitField0_ & ~0x00000001);
+          onChanged();
+        } else {
+          componentsBuilder_.clear();
+        }
+        return this;
+      }
+      /**
+       * <pre>
+       * A TensorInfo for each flattened component tensor.
+       * </pre>
+       *
+       * <code>repeated .tensorflow.TensorInfo components = 2;</code>
+       */
+      public Builder removeComponents(int index) {
+        if (componentsBuilder_ == null) {
+          ensureComponentsIsMutable();
+          components_.remove(index);
+          onChanged();
+        } else {
+          componentsBuilder_.remove(index);
+        }
+        return this;
+      }
+      /**
+       * <pre>
+       * A TensorInfo for each flattened component tensor.
+       * </pre>
+       *
+       * <code>repeated .tensorflow.TensorInfo components = 2;</code>
+       */
+      public org.tensorflow.proto.framework.TensorInfo.Builder getComponentsBuilder(
+          int index) {
+        return getComponentsFieldBuilder().getBuilder(index);
+      }
+      /**
+       * <pre>
+       * A TensorInfo for each flattened component tensor.
+       * </pre>
+       *
+       * <code>repeated .tensorflow.TensorInfo components = 2;</code>
+       */
+      public org.tensorflow.proto.framework.TensorInfoOrBuilder getComponentsOrBuilder(
+          int index) {
+        if (componentsBuilder_ == null) {
+          return components_.get(index);  } else {
+          return componentsBuilder_.getMessageOrBuilder(index);
+        }
+      }
+      /**
+       * <pre>
+       * A TensorInfo for each flattened component tensor.
+       * </pre>
+       *
+       * <code>repeated .tensorflow.TensorInfo components = 2;</code>
+       */
+      public java.util.List<? extends org.tensorflow.proto.framework.TensorInfoOrBuilder> 
+           getComponentsOrBuilderList() {
+        if (componentsBuilder_ != null) {
+          return componentsBuilder_.getMessageOrBuilderList();
+        } else {
+          return java.util.Collections.unmodifiableList(components_);
+        }
+      }
+      /**
+       * <pre>
+       * A TensorInfo for each flattened component tensor.
+       * </pre>
+       *
+       * <code>repeated .tensorflow.TensorInfo components = 2;</code>
+       */
+      public org.tensorflow.proto.framework.TensorInfo.Builder addComponentsBuilder() {
+        return getComponentsFieldBuilder().addBuilder(
+            org.tensorflow.proto.framework.TensorInfo.getDefaultInstance());
+      }
+      /**
+       * <pre>
+       * A TensorInfo for each flattened component tensor.
+       * </pre>
+       *
+       * <code>repeated .tensorflow.TensorInfo components = 2;</code>
+       */
+      public org.tensorflow.proto.framework.TensorInfo.Builder addComponentsBuilder(
+          int index) {
+        return getComponentsFieldBuilder().addBuilder(
+            index, org.tensorflow.proto.framework.TensorInfo.getDefaultInstance());
+      }
+      /**
+       * <pre>
+       * A TensorInfo for each flattened component tensor.
+       * </pre>
+       *
+       * <code>repeated .tensorflow.TensorInfo components = 2;</code>
+       */
+      public java.util.List<org.tensorflow.proto.framework.TensorInfo.Builder> 
+           getComponentsBuilderList() {
+        return getComponentsFieldBuilder().getBuilderList();
+      }
+      private com.google.protobuf.RepeatedFieldBuilderV3<
+          org.tensorflow.proto.framework.TensorInfo, org.tensorflow.proto.framework.TensorInfo.Builder, org.tensorflow.proto.framework.TensorInfoOrBuilder> 
+          getComponentsFieldBuilder() {
+        if (componentsBuilder_ == null) {
+          componentsBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
+              org.tensorflow.proto.framework.TensorInfo, org.tensorflow.proto.framework.TensorInfo.Builder, org.tensorflow.proto.framework.TensorInfoOrBuilder>(
+                  components_,
+                  ((bitField0_ & 0x00000001) != 0),
+                  getParentForChildren(),
+                  isClean());
+          components_ = null;
+        }
+        return componentsBuilder_;
+      }
+      @java.lang.Override
+      public final Builder setUnknownFields(
+          final com.google.protobuf.UnknownFieldSet unknownFields) {
+        return super.setUnknownFields(unknownFields);
+      }
+
+      @java.lang.Override
+      public final Builder mergeUnknownFields(
+          final com.google.protobuf.UnknownFieldSet unknownFields) {
+        return super.mergeUnknownFields(unknownFields);
+      }
+
+
+      // @@protoc_insertion_point(builder_scope:tensorflow.TensorInfo.CompositeTensor)
+    }
+
+    // @@protoc_insertion_point(class_scope:tensorflow.TensorInfo.CompositeTensor)
+    private static final org.tensorflow.proto.framework.TensorInfo.CompositeTensor DEFAULT_INSTANCE;
+    static {
+      DEFAULT_INSTANCE = new org.tensorflow.proto.framework.TensorInfo.CompositeTensor();
+    }
+
+    public static org.tensorflow.proto.framework.TensorInfo.CompositeTensor getDefaultInstance() {
+      return DEFAULT_INSTANCE;
+    }
+
+    private static final com.google.protobuf.Parser<CompositeTensor>
+        PARSER = new com.google.protobuf.AbstractParser<CompositeTensor>() {
+      @java.lang.Override
+      public CompositeTensor parsePartialFrom(
+          com.google.protobuf.CodedInputStream input,
+          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+          throws com.google.protobuf.InvalidProtocolBufferException {
+        return new CompositeTensor(input, extensionRegistry);
+      }
+    };
+
+    public static com.google.protobuf.Parser<CompositeTensor> parser() {
+      return PARSER;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Parser<CompositeTensor> getParserForType() {
+      return PARSER;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.TensorInfo.CompositeTensor getDefaultInstanceForType() {
+      return DEFAULT_INSTANCE;
+    }
+
+  }
+
+  private int encodingCase_ = 0;
+  private java.lang.Object encoding_;
+  public enum EncodingCase
+      implements com.google.protobuf.Internal.EnumLite {
+    NAME(1),
+    COO_SPARSE(4),
+    COMPOSITE_TENSOR(5),
+    ENCODING_NOT_SET(0);
+    private final int value;
+    private EncodingCase(int value) {
+      this.value = value;
+    }
+    /**
+     * @deprecated Use {@link #forNumber(int)} instead.
+     */
+    @java.lang.Deprecated
+    public static EncodingCase valueOf(int value) {
+      return forNumber(value);
+    }
+
+    public static EncodingCase forNumber(int value) {
+      switch (value) {
+        case 1: return NAME;
+        case 4: return COO_SPARSE;
+        case 5: return COMPOSITE_TENSOR;
+        case 0: return ENCODING_NOT_SET;
+        default: return null;
+      }
+    }
+    public int getNumber() {
+      return this.value;
+    }
+  };
+
+  public EncodingCase
+  getEncodingCase() {
+    return EncodingCase.forNumber(
+        encodingCase_);
+  }
+
+  public static final int NAME_FIELD_NUMBER = 1;
+  /**
+   * <pre>
+   * For dense `Tensor`s, the name of the tensor in the graph.
+   * </pre>
+   *
+   * <code>string name = 1;</code>
+   */
+  public java.lang.String getName() {
+    java.lang.Object ref = "";
+    if (encodingCase_ == 1) {
+      ref = encoding_;
+    }
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      if (encodingCase_ == 1) {
+        encoding_ = s;
+      }
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * For dense `Tensor`s, the name of the tensor in the graph.
+   * </pre>
+   *
+   * <code>string name = 1;</code>
+   */
+  public com.google.protobuf.ByteString
+      getNameBytes() {
+    java.lang.Object ref = "";
+    if (encodingCase_ == 1) {
+      ref = encoding_;
+    }
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      if (encodingCase_ == 1) {
+        encoding_ = b;
+      }
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int COO_SPARSE_FIELD_NUMBER = 4;
+  /**
+   * <pre>
+   * There are many possible encodings of sparse matrices
+   * (https://en.wikipedia.org/wiki/Sparse_matrix).  Currently, TensorFlow
+   * uses only the COO encoding.  This is supported and documented in the
+   * SparseTensor Python class.
+   * </pre>
+   *
+   * <code>.tensorflow.TensorInfo.CooSparse coo_sparse = 4;</code>
+   */
+  public boolean hasCooSparse() {
+    return encodingCase_ == 4;
+  }
+  /**
+   * <pre>
+   * There are many possible encodings of sparse matrices
+   * (https://en.wikipedia.org/wiki/Sparse_matrix).  Currently, TensorFlow
+   * uses only the COO encoding.  This is supported and documented in the
+   * SparseTensor Python class.
+   * </pre>
+   *
+   * <code>.tensorflow.TensorInfo.CooSparse coo_sparse = 4;</code>
+   */
+  public org.tensorflow.proto.framework.TensorInfo.CooSparse getCooSparse() {
+    if (encodingCase_ == 4) {
+       return (org.tensorflow.proto.framework.TensorInfo.CooSparse) encoding_;
+    }
+    return org.tensorflow.proto.framework.TensorInfo.CooSparse.getDefaultInstance();
+  }
+  /**
+   * <pre>
+   * There are many possible encodings of sparse matrices
+   * (https://en.wikipedia.org/wiki/Sparse_matrix).  Currently, TensorFlow
+   * uses only the COO encoding.  This is supported and documented in the
+   * SparseTensor Python class.
+   * </pre>
+   *
+   * <code>.tensorflow.TensorInfo.CooSparse coo_sparse = 4;</code>
+   */
+  public org.tensorflow.proto.framework.TensorInfo.CooSparseOrBuilder getCooSparseOrBuilder() {
+    if (encodingCase_ == 4) {
+       return (org.tensorflow.proto.framework.TensorInfo.CooSparse) encoding_;
+    }
+    return org.tensorflow.proto.framework.TensorInfo.CooSparse.getDefaultInstance();
+  }
+
+  public static final int COMPOSITE_TENSOR_FIELD_NUMBER = 5;
+  /**
+   * <pre>
+   * Generic encoding for CompositeTensors.
+   * </pre>
+   *
+   * <code>.tensorflow.TensorInfo.CompositeTensor composite_tensor = 5;</code>
+   */
+  public boolean hasCompositeTensor() {
+    return encodingCase_ == 5;
+  }
+  /**
+   * <pre>
+   * Generic encoding for CompositeTensors.
+   * </pre>
+   *
+   * <code>.tensorflow.TensorInfo.CompositeTensor composite_tensor = 5;</code>
+   */
+  public org.tensorflow.proto.framework.TensorInfo.CompositeTensor getCompositeTensor() {
+    if (encodingCase_ == 5) {
+       return (org.tensorflow.proto.framework.TensorInfo.CompositeTensor) encoding_;
+    }
+    return org.tensorflow.proto.framework.TensorInfo.CompositeTensor.getDefaultInstance();
+  }
+  /**
+   * <pre>
+   * Generic encoding for CompositeTensors.
+   * </pre>
+   *
+   * <code>.tensorflow.TensorInfo.CompositeTensor composite_tensor = 5;</code>
+   */
+  public org.tensorflow.proto.framework.TensorInfo.CompositeTensorOrBuilder getCompositeTensorOrBuilder() {
+    if (encodingCase_ == 5) {
+       return (org.tensorflow.proto.framework.TensorInfo.CompositeTensor) encoding_;
+    }
+    return org.tensorflow.proto.framework.TensorInfo.CompositeTensor.getDefaultInstance();
+  }
+
+  public static final int DTYPE_FIELD_NUMBER = 2;
+  private int dtype_;
+  /**
+   * <code>.tensorflow.DataType dtype = 2;</code>
+   */
+  public int getDtypeValue() {
+    return dtype_;
+  }
+  /**
+   * <code>.tensorflow.DataType dtype = 2;</code>
+   */
+  public org.tensorflow.proto.framework.DataType getDtype() {
+    @SuppressWarnings("deprecation")
+    org.tensorflow.proto.framework.DataType result = org.tensorflow.proto.framework.DataType.valueOf(dtype_);
+    return result == null ? org.tensorflow.proto.framework.DataType.UNRECOGNIZED : result;
+  }
+
+  public static final int TENSOR_SHAPE_FIELD_NUMBER = 3;
+  private org.tensorflow.proto.framework.TensorShapeProto tensorShape_;
+  /**
+   * <pre>
+   * The static shape should be recorded here, to the extent that it can
+   * be known in advance.  In the case of a SparseTensor, this field describes
+   * the logical shape of the represented tensor (aka dense_shape).
+   * </pre>
+   *
+   * <code>.tensorflow.TensorShapeProto tensor_shape = 3;</code>
+   */
+  public boolean hasTensorShape() {
+    return tensorShape_ != null;
+  }
+  /**
+   * <pre>
+   * The static shape should be recorded here, to the extent that it can
+   * be known in advance.  In the case of a SparseTensor, this field describes
+   * the logical shape of the represented tensor (aka dense_shape).
+   * </pre>
+   *
+   * <code>.tensorflow.TensorShapeProto tensor_shape = 3;</code>
+   */
+  public org.tensorflow.proto.framework.TensorShapeProto getTensorShape() {
+    return tensorShape_ == null ? org.tensorflow.proto.framework.TensorShapeProto.getDefaultInstance() : tensorShape_;
+  }
+  /**
+   * <pre>
+   * The static shape should be recorded here, to the extent that it can
+   * be known in advance.  In the case of a SparseTensor, this field describes
+   * the logical shape of the represented tensor (aka dense_shape).
+   * </pre>
+   *
+   * <code>.tensorflow.TensorShapeProto tensor_shape = 3;</code>
+   */
+  public org.tensorflow.proto.framework.TensorShapeProtoOrBuilder getTensorShapeOrBuilder() {
+    return getTensorShape();
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    if (encodingCase_ == 1) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 1, encoding_);
+    }
+    if (dtype_ != org.tensorflow.proto.framework.DataType.DT_INVALID.getNumber()) {
+      output.writeEnum(2, dtype_);
+    }
+    if (tensorShape_ != null) {
+      output.writeMessage(3, getTensorShape());
+    }
+    if (encodingCase_ == 4) {
+      output.writeMessage(4, (org.tensorflow.proto.framework.TensorInfo.CooSparse) encoding_);
+    }
+    if (encodingCase_ == 5) {
+      output.writeMessage(5, (org.tensorflow.proto.framework.TensorInfo.CompositeTensor) encoding_);
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (encodingCase_ == 1) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, encoding_);
+    }
+    if (dtype_ != org.tensorflow.proto.framework.DataType.DT_INVALID.getNumber()) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeEnumSize(2, dtype_);
+    }
+    if (tensorShape_ != null) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(3, getTensorShape());
+    }
+    if (encodingCase_ == 4) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(4, (org.tensorflow.proto.framework.TensorInfo.CooSparse) encoding_);
+    }
+    if (encodingCase_ == 5) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(5, (org.tensorflow.proto.framework.TensorInfo.CompositeTensor) encoding_);
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.framework.TensorInfo)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.framework.TensorInfo other = (org.tensorflow.proto.framework.TensorInfo) obj;
+
+    if (dtype_ != other.dtype_) return false;
+    if (hasTensorShape() != other.hasTensorShape()) return false;
+    if (hasTensorShape()) {
+      if (!getTensorShape()
+          .equals(other.getTensorShape())) return false;
+    }
+    if (!getEncodingCase().equals(other.getEncodingCase())) return false;
+    switch (encodingCase_) {
+      case 1:
+        if (!getName()
+            .equals(other.getName())) return false;
+        break;
+      case 4:
+        if (!getCooSparse()
+            .equals(other.getCooSparse())) return false;
+        break;
+      case 5:
+        if (!getCompositeTensor()
+            .equals(other.getCompositeTensor())) return false;
+        break;
+      case 0:
+      default:
+    }
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    hash = (37 * hash) + DTYPE_FIELD_NUMBER;
+    hash = (53 * hash) + dtype_;
+    if (hasTensorShape()) {
+      hash = (37 * hash) + TENSOR_SHAPE_FIELD_NUMBER;
+      hash = (53 * hash) + getTensorShape().hashCode();
+    }
+    switch (encodingCase_) {
+      case 1:
+        hash = (37 * hash) + NAME_FIELD_NUMBER;
+        hash = (53 * hash) + getName().hashCode();
+        break;
+      case 4:
+        hash = (37 * hash) + COO_SPARSE_FIELD_NUMBER;
+        hash = (53 * hash) + getCooSparse().hashCode();
+        break;
+      case 5:
+        hash = (37 * hash) + COMPOSITE_TENSOR_FIELD_NUMBER;
+        hash = (53 * hash) + getCompositeTensor().hashCode();
+        break;
+      case 0:
+      default:
+    }
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.framework.TensorInfo parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.TensorInfo parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.TensorInfo parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.TensorInfo parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.TensorInfo parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.TensorInfo parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.TensorInfo parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.TensorInfo parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.TensorInfo parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.TensorInfo parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.TensorInfo parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.TensorInfo parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.framework.TensorInfo prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * <pre>
+   * Information about a Tensor necessary for feeding or retrieval.
+   * </pre>
+   *
+   * Protobuf type {@code tensorflow.TensorInfo}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.TensorInfo)
+      org.tensorflow.proto.framework.TensorInfoOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.MetaGraphProtos.internal_static_tensorflow_TensorInfo_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.MetaGraphProtos.internal_static_tensorflow_TensorInfo_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.TensorInfo.class, org.tensorflow.proto.framework.TensorInfo.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.framework.TensorInfo.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      dtype_ = 0;
+
+      if (tensorShapeBuilder_ == null) {
+        tensorShape_ = null;
+      } else {
+        tensorShape_ = null;
+        tensorShapeBuilder_ = null;
+      }
+      encodingCase_ = 0;
+      encoding_ = null;
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.framework.MetaGraphProtos.internal_static_tensorflow_TensorInfo_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.TensorInfo getDefaultInstanceForType() {
+      return org.tensorflow.proto.framework.TensorInfo.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.TensorInfo build() {
+      org.tensorflow.proto.framework.TensorInfo result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.TensorInfo buildPartial() {
+      org.tensorflow.proto.framework.TensorInfo result = new org.tensorflow.proto.framework.TensorInfo(this);
+      if (encodingCase_ == 1) {
+        result.encoding_ = encoding_;
+      }
+      if (encodingCase_ == 4) {
+        if (cooSparseBuilder_ == null) {
+          result.encoding_ = encoding_;
+        } else {
+          result.encoding_ = cooSparseBuilder_.build();
+        }
+      }
+      if (encodingCase_ == 5) {
+        if (compositeTensorBuilder_ == null) {
+          result.encoding_ = encoding_;
+        } else {
+          result.encoding_ = compositeTensorBuilder_.build();
+        }
+      }
+      result.dtype_ = dtype_;
+      if (tensorShapeBuilder_ == null) {
+        result.tensorShape_ = tensorShape_;
+      } else {
+        result.tensorShape_ = tensorShapeBuilder_.build();
+      }
+      result.encodingCase_ = encodingCase_;
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.framework.TensorInfo) {
+        return mergeFrom((org.tensorflow.proto.framework.TensorInfo)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.framework.TensorInfo other) {
+      if (other == org.tensorflow.proto.framework.TensorInfo.getDefaultInstance()) return this;
+      if (other.dtype_ != 0) {
+        setDtypeValue(other.getDtypeValue());
+      }
+      if (other.hasTensorShape()) {
+        mergeTensorShape(other.getTensorShape());
+      }
+      switch (other.getEncodingCase()) {
+        case NAME: {
+          encodingCase_ = 1;
+          encoding_ = other.encoding_;
+          onChanged();
+          break;
+        }
+        case COO_SPARSE: {
+          mergeCooSparse(other.getCooSparse());
+          break;
+        }
+        case COMPOSITE_TENSOR: {
+          mergeCompositeTensor(other.getCompositeTensor());
+          break;
+        }
+        case ENCODING_NOT_SET: {
+          break;
+        }
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.framework.TensorInfo parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.framework.TensorInfo) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+    private int encodingCase_ = 0;
+    private java.lang.Object encoding_;
+    public EncodingCase
+        getEncodingCase() {
+      return EncodingCase.forNumber(
+          encodingCase_);
+    }
+
+    public Builder clearEncoding() {
+      encodingCase_ = 0;
+      encoding_ = null;
+      onChanged();
+      return this;
+    }
+
+
+    /**
+     * <pre>
+     * For dense `Tensor`s, the name of the tensor in the graph.
+     * </pre>
+     *
+     * <code>string name = 1;</code>
+     */
+    public java.lang.String getName() {
+      java.lang.Object ref = "";
+      if (encodingCase_ == 1) {
+        ref = encoding_;
+      }
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        if (encodingCase_ == 1) {
+          encoding_ = s;
+        }
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * For dense `Tensor`s, the name of the tensor in the graph.
+     * </pre>
+     *
+     * <code>string name = 1;</code>
+     */
+    public com.google.protobuf.ByteString
+        getNameBytes() {
+      java.lang.Object ref = "";
+      if (encodingCase_ == 1) {
+        ref = encoding_;
+      }
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        if (encodingCase_ == 1) {
+          encoding_ = b;
+        }
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * For dense `Tensor`s, the name of the tensor in the graph.
+     * </pre>
+     *
+     * <code>string name = 1;</code>
+     */
+    public Builder setName(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  encodingCase_ = 1;
+      encoding_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * For dense `Tensor`s, the name of the tensor in the graph.
+     * </pre>
+     *
+     * <code>string name = 1;</code>
+     */
+    public Builder clearName() {
+      if (encodingCase_ == 1) {
+        encodingCase_ = 0;
+        encoding_ = null;
+        onChanged();
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * For dense `Tensor`s, the name of the tensor in the graph.
+     * </pre>
+     *
+     * <code>string name = 1;</code>
+     */
+    public Builder setNameBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      encodingCase_ = 1;
+      encoding_ = value;
+      onChanged();
+      return this;
+    }
+
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.TensorInfo.CooSparse, org.tensorflow.proto.framework.TensorInfo.CooSparse.Builder, org.tensorflow.proto.framework.TensorInfo.CooSparseOrBuilder> cooSparseBuilder_;
+    /**
+     * <pre>
+     * There are many possible encodings of sparse matrices
+     * (https://en.wikipedia.org/wiki/Sparse_matrix).  Currently, TensorFlow
+     * uses only the COO encoding.  This is supported and documented in the
+     * SparseTensor Python class.
+     * </pre>
+     *
+     * <code>.tensorflow.TensorInfo.CooSparse coo_sparse = 4;</code>
+     */
+    public boolean hasCooSparse() {
+      return encodingCase_ == 4;
+    }
+    /**
+     * <pre>
+     * There are many possible encodings of sparse matrices
+     * (https://en.wikipedia.org/wiki/Sparse_matrix).  Currently, TensorFlow
+     * uses only the COO encoding.  This is supported and documented in the
+     * SparseTensor Python class.
+     * </pre>
+     *
+     * <code>.tensorflow.TensorInfo.CooSparse coo_sparse = 4;</code>
+     */
+    public org.tensorflow.proto.framework.TensorInfo.CooSparse getCooSparse() {
+      if (cooSparseBuilder_ == null) {
+        if (encodingCase_ == 4) {
+          return (org.tensorflow.proto.framework.TensorInfo.CooSparse) encoding_;
+        }
+        return org.tensorflow.proto.framework.TensorInfo.CooSparse.getDefaultInstance();
+      } else {
+        if (encodingCase_ == 4) {
+          return cooSparseBuilder_.getMessage();
+        }
+        return org.tensorflow.proto.framework.TensorInfo.CooSparse.getDefaultInstance();
+      }
+    }
+    /**
+     * <pre>
+     * There are many possible encodings of sparse matrices
+     * (https://en.wikipedia.org/wiki/Sparse_matrix).  Currently, TensorFlow
+     * uses only the COO encoding.  This is supported and documented in the
+     * SparseTensor Python class.
+     * </pre>
+     *
+     * <code>.tensorflow.TensorInfo.CooSparse coo_sparse = 4;</code>
+     */
+    public Builder setCooSparse(org.tensorflow.proto.framework.TensorInfo.CooSparse value) {
+      if (cooSparseBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        encoding_ = value;
+        onChanged();
+      } else {
+        cooSparseBuilder_.setMessage(value);
+      }
+      encodingCase_ = 4;
+      return this;
+    }
+    /**
+     * <pre>
+     * There are many possible encodings of sparse matrices
+     * (https://en.wikipedia.org/wiki/Sparse_matrix).  Currently, TensorFlow
+     * uses only the COO encoding.  This is supported and documented in the
+     * SparseTensor Python class.
+     * </pre>
+     *
+     * <code>.tensorflow.TensorInfo.CooSparse coo_sparse = 4;</code>
+     */
+    public Builder setCooSparse(
+        org.tensorflow.proto.framework.TensorInfo.CooSparse.Builder builderForValue) {
+      if (cooSparseBuilder_ == null) {
+        encoding_ = builderForValue.build();
+        onChanged();
+      } else {
+        cooSparseBuilder_.setMessage(builderForValue.build());
+      }
+      encodingCase_ = 4;
+      return this;
+    }
+    /**
+     * <pre>
+     * There are many possible encodings of sparse matrices
+     * (https://en.wikipedia.org/wiki/Sparse_matrix).  Currently, TensorFlow
+     * uses only the COO encoding.  This is supported and documented in the
+     * SparseTensor Python class.
+     * </pre>
+     *
+     * <code>.tensorflow.TensorInfo.CooSparse coo_sparse = 4;</code>
+     */
+    public Builder mergeCooSparse(org.tensorflow.proto.framework.TensorInfo.CooSparse value) {
+      if (cooSparseBuilder_ == null) {
+        if (encodingCase_ == 4 &&
+            encoding_ != org.tensorflow.proto.framework.TensorInfo.CooSparse.getDefaultInstance()) {
+          encoding_ = org.tensorflow.proto.framework.TensorInfo.CooSparse.newBuilder((org.tensorflow.proto.framework.TensorInfo.CooSparse) encoding_)
+              .mergeFrom(value).buildPartial();
+        } else {
+          encoding_ = value;
+        }
+        onChanged();
+      } else {
+        if (encodingCase_ == 4) {
+          cooSparseBuilder_.mergeFrom(value);
+        }
+        cooSparseBuilder_.setMessage(value);
+      }
+      encodingCase_ = 4;
+      return this;
+    }
+    /**
+     * <pre>
+     * There are many possible encodings of sparse matrices
+     * (https://en.wikipedia.org/wiki/Sparse_matrix).  Currently, TensorFlow
+     * uses only the COO encoding.  This is supported and documented in the
+     * SparseTensor Python class.
+     * </pre>
+     *
+     * <code>.tensorflow.TensorInfo.CooSparse coo_sparse = 4;</code>
+     */
+    public Builder clearCooSparse() {
+      if (cooSparseBuilder_ == null) {
+        if (encodingCase_ == 4) {
+          encodingCase_ = 0;
+          encoding_ = null;
+          onChanged();
+        }
+      } else {
+        if (encodingCase_ == 4) {
+          encodingCase_ = 0;
+          encoding_ = null;
+        }
+        cooSparseBuilder_.clear();
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * There are many possible encodings of sparse matrices
+     * (https://en.wikipedia.org/wiki/Sparse_matrix).  Currently, TensorFlow
+     * uses only the COO encoding.  This is supported and documented in the
+     * SparseTensor Python class.
+     * </pre>
+     *
+     * <code>.tensorflow.TensorInfo.CooSparse coo_sparse = 4;</code>
+     */
+    public org.tensorflow.proto.framework.TensorInfo.CooSparse.Builder getCooSparseBuilder() {
+      return getCooSparseFieldBuilder().getBuilder();
+    }
+    /**
+     * <pre>
+     * There are many possible encodings of sparse matrices
+     * (https://en.wikipedia.org/wiki/Sparse_matrix).  Currently, TensorFlow
+     * uses only the COO encoding.  This is supported and documented in the
+     * SparseTensor Python class.
+     * </pre>
+     *
+     * <code>.tensorflow.TensorInfo.CooSparse coo_sparse = 4;</code>
+     */
+    public org.tensorflow.proto.framework.TensorInfo.CooSparseOrBuilder getCooSparseOrBuilder() {
+      if ((encodingCase_ == 4) && (cooSparseBuilder_ != null)) {
+        return cooSparseBuilder_.getMessageOrBuilder();
+      } else {
+        if (encodingCase_ == 4) {
+          return (org.tensorflow.proto.framework.TensorInfo.CooSparse) encoding_;
+        }
+        return org.tensorflow.proto.framework.TensorInfo.CooSparse.getDefaultInstance();
+      }
+    }
+    /**
+     * <pre>
+     * There are many possible encodings of sparse matrices
+     * (https://en.wikipedia.org/wiki/Sparse_matrix).  Currently, TensorFlow
+     * uses only the COO encoding.  This is supported and documented in the
+     * SparseTensor Python class.
+     * </pre>
+     *
+     * <code>.tensorflow.TensorInfo.CooSparse coo_sparse = 4;</code>
+     */
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.TensorInfo.CooSparse, org.tensorflow.proto.framework.TensorInfo.CooSparse.Builder, org.tensorflow.proto.framework.TensorInfo.CooSparseOrBuilder> 
+        getCooSparseFieldBuilder() {
+      if (cooSparseBuilder_ == null) {
+        if (!(encodingCase_ == 4)) {
+          encoding_ = org.tensorflow.proto.framework.TensorInfo.CooSparse.getDefaultInstance();
+        }
+        cooSparseBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+            org.tensorflow.proto.framework.TensorInfo.CooSparse, org.tensorflow.proto.framework.TensorInfo.CooSparse.Builder, org.tensorflow.proto.framework.TensorInfo.CooSparseOrBuilder>(
+                (org.tensorflow.proto.framework.TensorInfo.CooSparse) encoding_,
+                getParentForChildren(),
+                isClean());
+        encoding_ = null;
+      }
+      encodingCase_ = 4;
+      onChanged();;
+      return cooSparseBuilder_;
+    }
+
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.TensorInfo.CompositeTensor, org.tensorflow.proto.framework.TensorInfo.CompositeTensor.Builder, org.tensorflow.proto.framework.TensorInfo.CompositeTensorOrBuilder> compositeTensorBuilder_;
+    /**
+     * <pre>
+     * Generic encoding for CompositeTensors.
+     * </pre>
+     *
+     * <code>.tensorflow.TensorInfo.CompositeTensor composite_tensor = 5;</code>
+     */
+    public boolean hasCompositeTensor() {
+      return encodingCase_ == 5;
+    }
+    /**
+     * <pre>
+     * Generic encoding for CompositeTensors.
+     * </pre>
+     *
+     * <code>.tensorflow.TensorInfo.CompositeTensor composite_tensor = 5;</code>
+     */
+    public org.tensorflow.proto.framework.TensorInfo.CompositeTensor getCompositeTensor() {
+      if (compositeTensorBuilder_ == null) {
+        if (encodingCase_ == 5) {
+          return (org.tensorflow.proto.framework.TensorInfo.CompositeTensor) encoding_;
+        }
+        return org.tensorflow.proto.framework.TensorInfo.CompositeTensor.getDefaultInstance();
+      } else {
+        if (encodingCase_ == 5) {
+          return compositeTensorBuilder_.getMessage();
+        }
+        return org.tensorflow.proto.framework.TensorInfo.CompositeTensor.getDefaultInstance();
+      }
+    }
+    /**
+     * <pre>
+     * Generic encoding for CompositeTensors.
+     * </pre>
+     *
+     * <code>.tensorflow.TensorInfo.CompositeTensor composite_tensor = 5;</code>
+     */
+    public Builder setCompositeTensor(org.tensorflow.proto.framework.TensorInfo.CompositeTensor value) {
+      if (compositeTensorBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        encoding_ = value;
+        onChanged();
+      } else {
+        compositeTensorBuilder_.setMessage(value);
+      }
+      encodingCase_ = 5;
+      return this;
+    }
+    /**
+     * <pre>
+     * Generic encoding for CompositeTensors.
+     * </pre>
+     *
+     * <code>.tensorflow.TensorInfo.CompositeTensor composite_tensor = 5;</code>
+     */
+    public Builder setCompositeTensor(
+        org.tensorflow.proto.framework.TensorInfo.CompositeTensor.Builder builderForValue) {
+      if (compositeTensorBuilder_ == null) {
+        encoding_ = builderForValue.build();
+        onChanged();
+      } else {
+        compositeTensorBuilder_.setMessage(builderForValue.build());
+      }
+      encodingCase_ = 5;
+      return this;
+    }
+    /**
+     * <pre>
+     * Generic encoding for CompositeTensors.
+     * </pre>
+     *
+     * <code>.tensorflow.TensorInfo.CompositeTensor composite_tensor = 5;</code>
+     */
+    public Builder mergeCompositeTensor(org.tensorflow.proto.framework.TensorInfo.CompositeTensor value) {
+      if (compositeTensorBuilder_ == null) {
+        if (encodingCase_ == 5 &&
+            encoding_ != org.tensorflow.proto.framework.TensorInfo.CompositeTensor.getDefaultInstance()) {
+          encoding_ = org.tensorflow.proto.framework.TensorInfo.CompositeTensor.newBuilder((org.tensorflow.proto.framework.TensorInfo.CompositeTensor) encoding_)
+              .mergeFrom(value).buildPartial();
+        } else {
+          encoding_ = value;
+        }
+        onChanged();
+      } else {
+        if (encodingCase_ == 5) {
+          compositeTensorBuilder_.mergeFrom(value);
+        }
+        compositeTensorBuilder_.setMessage(value);
+      }
+      encodingCase_ = 5;
+      return this;
+    }
+    /**
+     * <pre>
+     * Generic encoding for CompositeTensors.
+     * </pre>
+     *
+     * <code>.tensorflow.TensorInfo.CompositeTensor composite_tensor = 5;</code>
+     */
+    public Builder clearCompositeTensor() {
+      if (compositeTensorBuilder_ == null) {
+        if (encodingCase_ == 5) {
+          encodingCase_ = 0;
+          encoding_ = null;
+          onChanged();
+        }
+      } else {
+        if (encodingCase_ == 5) {
+          encodingCase_ = 0;
+          encoding_ = null;
+        }
+        compositeTensorBuilder_.clear();
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Generic encoding for CompositeTensors.
+     * </pre>
+     *
+     * <code>.tensorflow.TensorInfo.CompositeTensor composite_tensor = 5;</code>
+     */
+    public org.tensorflow.proto.framework.TensorInfo.CompositeTensor.Builder getCompositeTensorBuilder() {
+      return getCompositeTensorFieldBuilder().getBuilder();
+    }
+    /**
+     * <pre>
+     * Generic encoding for CompositeTensors.
+     * </pre>
+     *
+     * <code>.tensorflow.TensorInfo.CompositeTensor composite_tensor = 5;</code>
+     */
+    public org.tensorflow.proto.framework.TensorInfo.CompositeTensorOrBuilder getCompositeTensorOrBuilder() {
+      if ((encodingCase_ == 5) && (compositeTensorBuilder_ != null)) {
+        return compositeTensorBuilder_.getMessageOrBuilder();
+      } else {
+        if (encodingCase_ == 5) {
+          return (org.tensorflow.proto.framework.TensorInfo.CompositeTensor) encoding_;
+        }
+        return org.tensorflow.proto.framework.TensorInfo.CompositeTensor.getDefaultInstance();
+      }
+    }
+    /**
+     * <pre>
+     * Generic encoding for CompositeTensors.
+     * </pre>
+     *
+     * <code>.tensorflow.TensorInfo.CompositeTensor composite_tensor = 5;</code>
+     */
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.TensorInfo.CompositeTensor, org.tensorflow.proto.framework.TensorInfo.CompositeTensor.Builder, org.tensorflow.proto.framework.TensorInfo.CompositeTensorOrBuilder> 
+        getCompositeTensorFieldBuilder() {
+      if (compositeTensorBuilder_ == null) {
+        if (!(encodingCase_ == 5)) {
+          encoding_ = org.tensorflow.proto.framework.TensorInfo.CompositeTensor.getDefaultInstance();
+        }
+        compositeTensorBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+            org.tensorflow.proto.framework.TensorInfo.CompositeTensor, org.tensorflow.proto.framework.TensorInfo.CompositeTensor.Builder, org.tensorflow.proto.framework.TensorInfo.CompositeTensorOrBuilder>(
+                (org.tensorflow.proto.framework.TensorInfo.CompositeTensor) encoding_,
+                getParentForChildren(),
+                isClean());
+        encoding_ = null;
+      }
+      encodingCase_ = 5;
+      onChanged();;
+      return compositeTensorBuilder_;
+    }
+
+    private int dtype_ = 0;
+    /**
+     * <code>.tensorflow.DataType dtype = 2;</code>
+     */
+    public int getDtypeValue() {
+      return dtype_;
+    }
+    /**
+     * <code>.tensorflow.DataType dtype = 2;</code>
+     */
+    public Builder setDtypeValue(int value) {
+      dtype_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>.tensorflow.DataType dtype = 2;</code>
+     */
+    public org.tensorflow.proto.framework.DataType getDtype() {
+      @SuppressWarnings("deprecation")
+      org.tensorflow.proto.framework.DataType result = org.tensorflow.proto.framework.DataType.valueOf(dtype_);
+      return result == null ? org.tensorflow.proto.framework.DataType.UNRECOGNIZED : result;
+    }
+    /**
+     * <code>.tensorflow.DataType dtype = 2;</code>
+     */
+    public Builder setDtype(org.tensorflow.proto.framework.DataType value) {
+      if (value == null) {
+        throw new NullPointerException();
+      }
+      
+      dtype_ = value.getNumber();
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>.tensorflow.DataType dtype = 2;</code>
+     */
+    public Builder clearDtype() {
+      
+      dtype_ = 0;
+      onChanged();
+      return this;
+    }
+
+    private org.tensorflow.proto.framework.TensorShapeProto tensorShape_;
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.TensorShapeProto, org.tensorflow.proto.framework.TensorShapeProto.Builder, org.tensorflow.proto.framework.TensorShapeProtoOrBuilder> tensorShapeBuilder_;
+    /**
+     * <pre>
+     * The static shape should be recorded here, to the extent that it can
+     * be known in advance.  In the case of a SparseTensor, this field describes
+     * the logical shape of the represented tensor (aka dense_shape).
+     * </pre>
+     *
+     * <code>.tensorflow.TensorShapeProto tensor_shape = 3;</code>
+     */
+    public boolean hasTensorShape() {
+      return tensorShapeBuilder_ != null || tensorShape_ != null;
+    }
+    /**
+     * <pre>
+     * The static shape should be recorded here, to the extent that it can
+     * be known in advance.  In the case of a SparseTensor, this field describes
+     * the logical shape of the represented tensor (aka dense_shape).
+     * </pre>
+     *
+     * <code>.tensorflow.TensorShapeProto tensor_shape = 3;</code>
+     */
+    public org.tensorflow.proto.framework.TensorShapeProto getTensorShape() {
+      if (tensorShapeBuilder_ == null) {
+        return tensorShape_ == null ? org.tensorflow.proto.framework.TensorShapeProto.getDefaultInstance() : tensorShape_;
+      } else {
+        return tensorShapeBuilder_.getMessage();
+      }
+    }
+    /**
+     * <pre>
+     * The static shape should be recorded here, to the extent that it can
+     * be known in advance.  In the case of a SparseTensor, this field describes
+     * the logical shape of the represented tensor (aka dense_shape).
+     * </pre>
+     *
+     * <code>.tensorflow.TensorShapeProto tensor_shape = 3;</code>
+     */
+    public Builder setTensorShape(org.tensorflow.proto.framework.TensorShapeProto value) {
+      if (tensorShapeBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        tensorShape_ = value;
+        onChanged();
+      } else {
+        tensorShapeBuilder_.setMessage(value);
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * The static shape should be recorded here, to the extent that it can
+     * be known in advance.  In the case of a SparseTensor, this field describes
+     * the logical shape of the represented tensor (aka dense_shape).
+     * </pre>
+     *
+     * <code>.tensorflow.TensorShapeProto tensor_shape = 3;</code>
+     */
+    public Builder setTensorShape(
+        org.tensorflow.proto.framework.TensorShapeProto.Builder builderForValue) {
+      if (tensorShapeBuilder_ == null) {
+        tensorShape_ = builderForValue.build();
+        onChanged();
+      } else {
+        tensorShapeBuilder_.setMessage(builderForValue.build());
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * The static shape should be recorded here, to the extent that it can
+     * be known in advance.  In the case of a SparseTensor, this field describes
+     * the logical shape of the represented tensor (aka dense_shape).
+     * </pre>
+     *
+     * <code>.tensorflow.TensorShapeProto tensor_shape = 3;</code>
+     */
+    public Builder mergeTensorShape(org.tensorflow.proto.framework.TensorShapeProto value) {
+      if (tensorShapeBuilder_ == null) {
+        if (tensorShape_ != null) {
+          tensorShape_ =
+            org.tensorflow.proto.framework.TensorShapeProto.newBuilder(tensorShape_).mergeFrom(value).buildPartial();
+        } else {
+          tensorShape_ = value;
+        }
+        onChanged();
+      } else {
+        tensorShapeBuilder_.mergeFrom(value);
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * The static shape should be recorded here, to the extent that it can
+     * be known in advance.  In the case of a SparseTensor, this field describes
+     * the logical shape of the represented tensor (aka dense_shape).
+     * </pre>
+     *
+     * <code>.tensorflow.TensorShapeProto tensor_shape = 3;</code>
+     */
+    public Builder clearTensorShape() {
+      if (tensorShapeBuilder_ == null) {
+        tensorShape_ = null;
+        onChanged();
+      } else {
+        tensorShape_ = null;
+        tensorShapeBuilder_ = null;
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * The static shape should be recorded here, to the extent that it can
+     * be known in advance.  In the case of a SparseTensor, this field describes
+     * the logical shape of the represented tensor (aka dense_shape).
+     * </pre>
+     *
+     * <code>.tensorflow.TensorShapeProto tensor_shape = 3;</code>
+     */
+    public org.tensorflow.proto.framework.TensorShapeProto.Builder getTensorShapeBuilder() {
+      
+      onChanged();
+      return getTensorShapeFieldBuilder().getBuilder();
+    }
+    /**
+     * <pre>
+     * The static shape should be recorded here, to the extent that it can
+     * be known in advance.  In the case of a SparseTensor, this field describes
+     * the logical shape of the represented tensor (aka dense_shape).
+     * </pre>
+     *
+     * <code>.tensorflow.TensorShapeProto tensor_shape = 3;</code>
+     */
+    public org.tensorflow.proto.framework.TensorShapeProtoOrBuilder getTensorShapeOrBuilder() {
+      if (tensorShapeBuilder_ != null) {
+        return tensorShapeBuilder_.getMessageOrBuilder();
+      } else {
+        return tensorShape_ == null ?
+            org.tensorflow.proto.framework.TensorShapeProto.getDefaultInstance() : tensorShape_;
+      }
+    }
+    /**
+     * <pre>
+     * The static shape should be recorded here, to the extent that it can
+     * be known in advance.  In the case of a SparseTensor, this field describes
+     * the logical shape of the represented tensor (aka dense_shape).
+     * </pre>
+     *
+     * <code>.tensorflow.TensorShapeProto tensor_shape = 3;</code>
+     */
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.TensorShapeProto, org.tensorflow.proto.framework.TensorShapeProto.Builder, org.tensorflow.proto.framework.TensorShapeProtoOrBuilder> 
+        getTensorShapeFieldBuilder() {
+      if (tensorShapeBuilder_ == null) {
+        tensorShapeBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+            org.tensorflow.proto.framework.TensorShapeProto, org.tensorflow.proto.framework.TensorShapeProto.Builder, org.tensorflow.proto.framework.TensorShapeProtoOrBuilder>(
+                getTensorShape(),
+                getParentForChildren(),
+                isClean());
+        tensorShape_ = null;
+      }
+      return tensorShapeBuilder_;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.TensorInfo)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.TensorInfo)
+  private static final org.tensorflow.proto.framework.TensorInfo DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.TensorInfo();
+  }
+
+  public static org.tensorflow.proto.framework.TensorInfo getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<TensorInfo>
+      PARSER = new com.google.protobuf.AbstractParser<TensorInfo>() {
+    @java.lang.Override
+    public TensorInfo parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new TensorInfo(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<TensorInfo> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<TensorInfo> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.framework.TensorInfo getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorInfoOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorInfoOrBuilder.java
new file mode 100644
index 00000000000..d30570a1969
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorInfoOrBuilder.java
@@ -0,0 +1,128 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/meta_graph.proto
+
+package org.tensorflow.proto.framework;
+
+public interface TensorInfoOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.TensorInfo)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <pre>
+   * For dense `Tensor`s, the name of the tensor in the graph.
+   * </pre>
+   *
+   * <code>string name = 1;</code>
+   */
+  java.lang.String getName();
+  /**
+   * <pre>
+   * For dense `Tensor`s, the name of the tensor in the graph.
+   * </pre>
+   *
+   * <code>string name = 1;</code>
+   */
+  com.google.protobuf.ByteString
+      getNameBytes();
+
+  /**
+   * <pre>
+   * There are many possible encodings of sparse matrices
+   * (https://en.wikipedia.org/wiki/Sparse_matrix).  Currently, TensorFlow
+   * uses only the COO encoding.  This is supported and documented in the
+   * SparseTensor Python class.
+   * </pre>
+   *
+   * <code>.tensorflow.TensorInfo.CooSparse coo_sparse = 4;</code>
+   */
+  boolean hasCooSparse();
+  /**
+   * <pre>
+   * There are many possible encodings of sparse matrices
+   * (https://en.wikipedia.org/wiki/Sparse_matrix).  Currently, TensorFlow
+   * uses only the COO encoding.  This is supported and documented in the
+   * SparseTensor Python class.
+   * </pre>
+   *
+   * <code>.tensorflow.TensorInfo.CooSparse coo_sparse = 4;</code>
+   */
+  org.tensorflow.proto.framework.TensorInfo.CooSparse getCooSparse();
+  /**
+   * <pre>
+   * There are many possible encodings of sparse matrices
+   * (https://en.wikipedia.org/wiki/Sparse_matrix).  Currently, TensorFlow
+   * uses only the COO encoding.  This is supported and documented in the
+   * SparseTensor Python class.
+   * </pre>
+   *
+   * <code>.tensorflow.TensorInfo.CooSparse coo_sparse = 4;</code>
+   */
+  org.tensorflow.proto.framework.TensorInfo.CooSparseOrBuilder getCooSparseOrBuilder();
+
+  /**
+   * <pre>
+   * Generic encoding for CompositeTensors.
+   * </pre>
+   *
+   * <code>.tensorflow.TensorInfo.CompositeTensor composite_tensor = 5;</code>
+   */
+  boolean hasCompositeTensor();
+  /**
+   * <pre>
+   * Generic encoding for CompositeTensors.
+   * </pre>
+   *
+   * <code>.tensorflow.TensorInfo.CompositeTensor composite_tensor = 5;</code>
+   */
+  org.tensorflow.proto.framework.TensorInfo.CompositeTensor getCompositeTensor();
+  /**
+   * <pre>
+   * Generic encoding for CompositeTensors.
+   * </pre>
+   *
+   * <code>.tensorflow.TensorInfo.CompositeTensor composite_tensor = 5;</code>
+   */
+  org.tensorflow.proto.framework.TensorInfo.CompositeTensorOrBuilder getCompositeTensorOrBuilder();
+
+  /**
+   * <code>.tensorflow.DataType dtype = 2;</code>
+   */
+  int getDtypeValue();
+  /**
+   * <code>.tensorflow.DataType dtype = 2;</code>
+   */
+  org.tensorflow.proto.framework.DataType getDtype();
+
+  /**
+   * <pre>
+   * The static shape should be recorded here, to the extent that it can
+   * be known in advance.  In the case of a SparseTensor, this field describes
+   * the logical shape of the represented tensor (aka dense_shape).
+   * </pre>
+   *
+   * <code>.tensorflow.TensorShapeProto tensor_shape = 3;</code>
+   */
+  boolean hasTensorShape();
+  /**
+   * <pre>
+   * The static shape should be recorded here, to the extent that it can
+   * be known in advance.  In the case of a SparseTensor, this field describes
+   * the logical shape of the represented tensor (aka dense_shape).
+   * </pre>
+   *
+   * <code>.tensorflow.TensorShapeProto tensor_shape = 3;</code>
+   */
+  org.tensorflow.proto.framework.TensorShapeProto getTensorShape();
+  /**
+   * <pre>
+   * The static shape should be recorded here, to the extent that it can
+   * be known in advance.  In the case of a SparseTensor, this field describes
+   * the logical shape of the represented tensor (aka dense_shape).
+   * </pre>
+   *
+   * <code>.tensorflow.TensorShapeProto tensor_shape = 3;</code>
+   */
+  org.tensorflow.proto.framework.TensorShapeProtoOrBuilder getTensorShapeOrBuilder();
+
+  public org.tensorflow.proto.framework.TensorInfo.EncodingCase getEncodingCase();
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorProto.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorProto.java
new file mode 100644
index 00000000000..756d17fef2a
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorProto.java
@@ -0,0 +1,3980 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/framework/tensor.proto
+
+package org.tensorflow.proto.framework;
+
+/**
+ * <pre>
+ * Protocol buffer representing a tensor.
+ * </pre>
+ *
+ * Protobuf type {@code tensorflow.TensorProto}
+ */
+public  final class TensorProto extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.TensorProto)
+    TensorProtoOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use TensorProto.newBuilder() to construct.
+  private TensorProto(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private TensorProto() {
+    dtype_ = 0;
+    tensorContent_ = com.google.protobuf.ByteString.EMPTY;
+    halfVal_ = emptyIntList();
+    floatVal_ = emptyFloatList();
+    doubleVal_ = emptyDoubleList();
+    intVal_ = emptyIntList();
+    stringVal_ = java.util.Collections.emptyList();
+    scomplexVal_ = emptyFloatList();
+    int64Val_ = emptyLongList();
+    boolVal_ = emptyBooleanList();
+    dcomplexVal_ = emptyDoubleList();
+    resourceHandleVal_ = java.util.Collections.emptyList();
+    variantVal_ = java.util.Collections.emptyList();
+    uint32Val_ = emptyIntList();
+    uint64Val_ = emptyLongList();
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new TensorProto();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private TensorProto(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    int mutable_bitField0_ = 0;
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 8: {
+            int rawValue = input.readEnum();
+
+            dtype_ = rawValue;
+            break;
+          }
+          case 18: {
+            org.tensorflow.proto.framework.TensorShapeProto.Builder subBuilder = null;
+            if (tensorShape_ != null) {
+              subBuilder = tensorShape_.toBuilder();
+            }
+            tensorShape_ = input.readMessage(org.tensorflow.proto.framework.TensorShapeProto.parser(), extensionRegistry);
+            if (subBuilder != null) {
+              subBuilder.mergeFrom(tensorShape_);
+              tensorShape_ = subBuilder.buildPartial();
+            }
+
+            break;
+          }
+          case 24: {
+
+            versionNumber_ = input.readInt32();
+            break;
+          }
+          case 34: {
+
+            tensorContent_ = input.readBytes();
+            break;
+          }
+          case 45: {
+            if (!((mutable_bitField0_ & 0x00000002) != 0)) {
+              floatVal_ = newFloatList();
+              mutable_bitField0_ |= 0x00000002;
+            }
+            floatVal_.addFloat(input.readFloat());
+            break;
+          }
+          case 42: {
+            int length = input.readRawVarint32();
+            int limit = input.pushLimit(length);
+            if (!((mutable_bitField0_ & 0x00000002) != 0) && input.getBytesUntilLimit() > 0) {
+              floatVal_ = newFloatList();
+              mutable_bitField0_ |= 0x00000002;
+            }
+            while (input.getBytesUntilLimit() > 0) {
+              floatVal_.addFloat(input.readFloat());
+            }
+            input.popLimit(limit);
+            break;
+          }
+          case 49: {
+            if (!((mutable_bitField0_ & 0x00000004) != 0)) {
+              doubleVal_ = newDoubleList();
+              mutable_bitField0_ |= 0x00000004;
+            }
+            doubleVal_.addDouble(input.readDouble());
+            break;
+          }
+          case 50: {
+            int length = input.readRawVarint32();
+            int limit = input.pushLimit(length);
+            if (!((mutable_bitField0_ & 0x00000004) != 0) && input.getBytesUntilLimit() > 0) {
+              doubleVal_ = newDoubleList();
+              mutable_bitField0_ |= 0x00000004;
+            }
+            while (input.getBytesUntilLimit() > 0) {
+              doubleVal_.addDouble(input.readDouble());
+            }
+            input.popLimit(limit);
+            break;
+          }
+          case 56: {
+            if (!((mutable_bitField0_ & 0x00000008) != 0)) {
+              intVal_ = newIntList();
+              mutable_bitField0_ |= 0x00000008;
+            }
+            intVal_.addInt(input.readInt32());
+            break;
+          }
+          case 58: {
+            int length = input.readRawVarint32();
+            int limit = input.pushLimit(length);
+            if (!((mutable_bitField0_ & 0x00000008) != 0) && input.getBytesUntilLimit() > 0) {
+              intVal_ = newIntList();
+              mutable_bitField0_ |= 0x00000008;
+            }
+            while (input.getBytesUntilLimit() > 0) {
+              intVal_.addInt(input.readInt32());
+            }
+            input.popLimit(limit);
+            break;
+          }
+          case 66: {
+            if (!((mutable_bitField0_ & 0x00000010) != 0)) {
+              stringVal_ = new java.util.ArrayList<com.google.protobuf.ByteString>();
+              mutable_bitField0_ |= 0x00000010;
+            }
+            stringVal_.add(input.readBytes());
+            break;
+          }
+          case 77: {
+            if (!((mutable_bitField0_ & 0x00000020) != 0)) {
+              scomplexVal_ = newFloatList();
+              mutable_bitField0_ |= 0x00000020;
+            }
+            scomplexVal_.addFloat(input.readFloat());
+            break;
+          }
+          case 74: {
+            int length = input.readRawVarint32();
+            int limit = input.pushLimit(length);
+            if (!((mutable_bitField0_ & 0x00000020) != 0) && input.getBytesUntilLimit() > 0) {
+              scomplexVal_ = newFloatList();
+              mutable_bitField0_ |= 0x00000020;
+            }
+            while (input.getBytesUntilLimit() > 0) {
+              scomplexVal_.addFloat(input.readFloat());
+            }
+            input.popLimit(limit);
+            break;
+          }
+          case 80: {
+            if (!((mutable_bitField0_ & 0x00000040) != 0)) {
+              int64Val_ = newLongList();
+              mutable_bitField0_ |= 0x00000040;
+            }
+            int64Val_.addLong(input.readInt64());
+            break;
+          }
+          case 82: {
+            int length = input.readRawVarint32();
+            int limit = input.pushLimit(length);
+            if (!((mutable_bitField0_ & 0x00000040) != 0) && input.getBytesUntilLimit() > 0) {
+              int64Val_ = newLongList();
+              mutable_bitField0_ |= 0x00000040;
+            }
+            while (input.getBytesUntilLimit() > 0) {
+              int64Val_.addLong(input.readInt64());
+            }
+            input.popLimit(limit);
+            break;
+          }
+          case 88: {
+            if (!((mutable_bitField0_ & 0x00000080) != 0)) {
+              boolVal_ = newBooleanList();
+              mutable_bitField0_ |= 0x00000080;
+            }
+            boolVal_.addBoolean(input.readBool());
+            break;
+          }
+          case 90: {
+            int length = input.readRawVarint32();
+            int limit = input.pushLimit(length);
+            if (!((mutable_bitField0_ & 0x00000080) != 0) && input.getBytesUntilLimit() > 0) {
+              boolVal_ = newBooleanList();
+              mutable_bitField0_ |= 0x00000080;
+            }
+            while (input.getBytesUntilLimit() > 0) {
+              boolVal_.addBoolean(input.readBool());
+            }
+            input.popLimit(limit);
+            break;
+          }
+          case 97: {
+            if (!((mutable_bitField0_ & 0x00000100) != 0)) {
+              dcomplexVal_ = newDoubleList();
+              mutable_bitField0_ |= 0x00000100;
+            }
+            dcomplexVal_.addDouble(input.readDouble());
+            break;
+          }
+          case 98: {
+            int length = input.readRawVarint32();
+            int limit = input.pushLimit(length);
+            if (!((mutable_bitField0_ & 0x00000100) != 0) && input.getBytesUntilLimit() > 0) {
+              dcomplexVal_ = newDoubleList();
+              mutable_bitField0_ |= 0x00000100;
+            }
+            while (input.getBytesUntilLimit() > 0) {
+              dcomplexVal_.addDouble(input.readDouble());
+            }
+            input.popLimit(limit);
+            break;
+          }
+          case 104: {
+            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
+              halfVal_ = newIntList();
+              mutable_bitField0_ |= 0x00000001;
+            }
+            halfVal_.addInt(input.readInt32());
+            break;
+          }
+          case 106: {
+            int length = input.readRawVarint32();
+            int limit = input.pushLimit(length);
+            if (!((mutable_bitField0_ & 0x00000001) != 0) && input.getBytesUntilLimit() > 0) {
+              halfVal_ = newIntList();
+              mutable_bitField0_ |= 0x00000001;
+            }
+            while (input.getBytesUntilLimit() > 0) {
+              halfVal_.addInt(input.readInt32());
+            }
+            input.popLimit(limit);
+            break;
+          }
+          case 114: {
+            if (!((mutable_bitField0_ & 0x00000200) != 0)) {
+              resourceHandleVal_ = new java.util.ArrayList<org.tensorflow.proto.framework.ResourceHandleProto>();
+              mutable_bitField0_ |= 0x00000200;
+            }
+            resourceHandleVal_.add(
+                input.readMessage(org.tensorflow.proto.framework.ResourceHandleProto.parser(), extensionRegistry));
+            break;
+          }
+          case 122: {
+            if (!((mutable_bitField0_ & 0x00000400) != 0)) {
+              variantVal_ = new java.util.ArrayList<org.tensorflow.proto.framework.VariantTensorDataProto>();
+              mutable_bitField0_ |= 0x00000400;
+            }
+            variantVal_.add(
+                input.readMessage(org.tensorflow.proto.framework.VariantTensorDataProto.parser(), extensionRegistry));
+            break;
+          }
+          case 128: {
+            if (!((mutable_bitField0_ & 0x00000800) != 0)) {
+              uint32Val_ = newIntList();
+              mutable_bitField0_ |= 0x00000800;
+            }
+            uint32Val_.addInt(input.readUInt32());
+            break;
+          }
+          case 130: {
+            int length = input.readRawVarint32();
+            int limit = input.pushLimit(length);
+            if (!((mutable_bitField0_ & 0x00000800) != 0) && input.getBytesUntilLimit() > 0) {
+              uint32Val_ = newIntList();
+              mutable_bitField0_ |= 0x00000800;
+            }
+            while (input.getBytesUntilLimit() > 0) {
+              uint32Val_.addInt(input.readUInt32());
+            }
+            input.popLimit(limit);
+            break;
+          }
+          case 136: {
+            if (!((mutable_bitField0_ & 0x00001000) != 0)) {
+              uint64Val_ = newLongList();
+              mutable_bitField0_ |= 0x00001000;
+            }
+            uint64Val_.addLong(input.readUInt64());
+            break;
+          }
+          case 138: {
+            int length = input.readRawVarint32();
+            int limit = input.pushLimit(length);
+            if (!((mutable_bitField0_ & 0x00001000) != 0) && input.getBytesUntilLimit() > 0) {
+              uint64Val_ = newLongList();
+              mutable_bitField0_ |= 0x00001000;
+            }
+            while (input.getBytesUntilLimit() > 0) {
+              uint64Val_.addLong(input.readUInt64());
+            }
+            input.popLimit(limit);
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      if (((mutable_bitField0_ & 0x00000002) != 0)) {
+        floatVal_.makeImmutable(); // C
+      }
+      if (((mutable_bitField0_ & 0x00000004) != 0)) {
+        doubleVal_.makeImmutable(); // C
+      }
+      if (((mutable_bitField0_ & 0x00000008) != 0)) {
+        intVal_.makeImmutable(); // C
+      }
+      if (((mutable_bitField0_ & 0x00000010) != 0)) {
+        stringVal_ = java.util.Collections.unmodifiableList(stringVal_); // C
+      }
+      if (((mutable_bitField0_ & 0x00000020) != 0)) {
+        scomplexVal_.makeImmutable(); // C
+      }
+      if (((mutable_bitField0_ & 0x00000040) != 0)) {
+        int64Val_.makeImmutable(); // C
+      }
+      if (((mutable_bitField0_ & 0x00000080) != 0)) {
+        boolVal_.makeImmutable(); // C
+      }
+      if (((mutable_bitField0_ & 0x00000100) != 0)) {
+        dcomplexVal_.makeImmutable(); // C
+      }
+      if (((mutable_bitField0_ & 0x00000001) != 0)) {
+        halfVal_.makeImmutable(); // C
+      }
+      if (((mutable_bitField0_ & 0x00000200) != 0)) {
+        resourceHandleVal_ = java.util.Collections.unmodifiableList(resourceHandleVal_);
+      }
+      if (((mutable_bitField0_ & 0x00000400) != 0)) {
+        variantVal_ = java.util.Collections.unmodifiableList(variantVal_);
+      }
+      if (((mutable_bitField0_ & 0x00000800) != 0)) {
+        uint32Val_.makeImmutable(); // C
+      }
+      if (((mutable_bitField0_ & 0x00001000) != 0)) {
+        uint64Val_.makeImmutable(); // C
+      }
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.framework.TensorProtos.internal_static_tensorflow_TensorProto_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.framework.TensorProtos.internal_static_tensorflow_TensorProto_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.framework.TensorProto.class, org.tensorflow.proto.framework.TensorProto.Builder.class);
+  }
+
+  public static final int DTYPE_FIELD_NUMBER = 1;
+  private int dtype_;
+  /**
+   * <code>.tensorflow.DataType dtype = 1;</code>
+   */
+  public int getDtypeValue() {
+    return dtype_;
+  }
+  /**
+   * <code>.tensorflow.DataType dtype = 1;</code>
+   */
+  public org.tensorflow.proto.framework.DataType getDtype() {
+    @SuppressWarnings("deprecation")
+    org.tensorflow.proto.framework.DataType result = org.tensorflow.proto.framework.DataType.valueOf(dtype_);
+    return result == null ? org.tensorflow.proto.framework.DataType.UNRECOGNIZED : result;
+  }
+
+  public static final int TENSOR_SHAPE_FIELD_NUMBER = 2;
+  private org.tensorflow.proto.framework.TensorShapeProto tensorShape_;
+  /**
+   * <pre>
+   * Shape of the tensor.  TODO(touts): sort out the 0-rank issues.
+   * </pre>
+   *
+   * <code>.tensorflow.TensorShapeProto tensor_shape = 2;</code>
+   */
+  public boolean hasTensorShape() {
+    return tensorShape_ != null;
+  }
+  /**
+   * <pre>
+   * Shape of the tensor.  TODO(touts): sort out the 0-rank issues.
+   * </pre>
+   *
+   * <code>.tensorflow.TensorShapeProto tensor_shape = 2;</code>
+   */
+  public org.tensorflow.proto.framework.TensorShapeProto getTensorShape() {
+    return tensorShape_ == null ? org.tensorflow.proto.framework.TensorShapeProto.getDefaultInstance() : tensorShape_;
+  }
+  /**
+   * <pre>
+   * Shape of the tensor.  TODO(touts): sort out the 0-rank issues.
+   * </pre>
+   *
+   * <code>.tensorflow.TensorShapeProto tensor_shape = 2;</code>
+   */
+  public org.tensorflow.proto.framework.TensorShapeProtoOrBuilder getTensorShapeOrBuilder() {
+    return getTensorShape();
+  }
+
+  public static final int VERSION_NUMBER_FIELD_NUMBER = 3;
+  private int versionNumber_;
+  /**
+   * <pre>
+   * Version number.
+   * In version 0, if the "repeated xxx" representations contain only one
+   * element, that element is repeated to fill the shape.  This makes it easy
+   * to represent a constant Tensor with a single value.
+   * </pre>
+   *
+   * <code>int32 version_number = 3;</code>
+   */
+  public int getVersionNumber() {
+    return versionNumber_;
+  }
+
+  public static final int TENSOR_CONTENT_FIELD_NUMBER = 4;
+  private com.google.protobuf.ByteString tensorContent_;
+  /**
+   * <pre>
+   * Serialized raw tensor content from either Tensor::AsProtoTensorContent or
+   * memcpy in tensorflow::grpc::EncodeTensorToByteBuffer. This representation
+   * can be used for all tensor types. The purpose of this representation is to
+   * reduce serialization overhead during RPC call by avoiding serialization of
+   * many repeated small items.
+   * </pre>
+   *
+   * <code>bytes tensor_content = 4;</code>
+   */
+  public com.google.protobuf.ByteString getTensorContent() {
+    return tensorContent_;
+  }
+
+  public static final int HALF_VAL_FIELD_NUMBER = 13;
+  private com.google.protobuf.Internal.IntList halfVal_;
+  /**
+   * <pre>
+   * DT_HALF, DT_BFLOAT16. Note that since protobuf has no int16 type, we'll
+   * have some pointless zero padding for each value here.
+   * </pre>
+   *
+   * <code>repeated int32 half_val = 13 [packed = true];</code>
+   */
+  public java.util.List<java.lang.Integer>
+      getHalfValList() {
+    return halfVal_;
+  }
+  /**
+   * <pre>
+   * DT_HALF, DT_BFLOAT16. Note that since protobuf has no int16 type, we'll
+   * have some pointless zero padding for each value here.
+   * </pre>
+   *
+   * <code>repeated int32 half_val = 13 [packed = true];</code>
+   */
+  public int getHalfValCount() {
+    return halfVal_.size();
+  }
+  /**
+   * <pre>
+   * DT_HALF, DT_BFLOAT16. Note that since protobuf has no int16 type, we'll
+   * have some pointless zero padding for each value here.
+   * </pre>
+   *
+   * <code>repeated int32 half_val = 13 [packed = true];</code>
+   */
+  public int getHalfVal(int index) {
+    return halfVal_.getInt(index);
+  }
+  private int halfValMemoizedSerializedSize = -1;
+
+  public static final int FLOAT_VAL_FIELD_NUMBER = 5;
+  private com.google.protobuf.Internal.FloatList floatVal_;
+  /**
+   * <pre>
+   * DT_FLOAT.
+   * </pre>
+   *
+   * <code>repeated float float_val = 5 [packed = true];</code>
+   */
+  public java.util.List<java.lang.Float>
+      getFloatValList() {
+    return floatVal_;
+  }
+  /**
+   * <pre>
+   * DT_FLOAT.
+   * </pre>
+   *
+   * <code>repeated float float_val = 5 [packed = true];</code>
+   */
+  public int getFloatValCount() {
+    return floatVal_.size();
+  }
+  /**
+   * <pre>
+   * DT_FLOAT.
+   * </pre>
+   *
+   * <code>repeated float float_val = 5 [packed = true];</code>
+   */
+  public float getFloatVal(int index) {
+    return floatVal_.getFloat(index);
+  }
+  private int floatValMemoizedSerializedSize = -1;
+
+  public static final int DOUBLE_VAL_FIELD_NUMBER = 6;
+  private com.google.protobuf.Internal.DoubleList doubleVal_;
+  /**
+   * <pre>
+   * DT_DOUBLE.
+   * </pre>
+   *
+   * <code>repeated double double_val = 6 [packed = true];</code>
+   */
+  public java.util.List<java.lang.Double>
+      getDoubleValList() {
+    return doubleVal_;
+  }
+  /**
+   * <pre>
+   * DT_DOUBLE.
+   * </pre>
+   *
+   * <code>repeated double double_val = 6 [packed = true];</code>
+   */
+  public int getDoubleValCount() {
+    return doubleVal_.size();
+  }
+  /**
+   * <pre>
+   * DT_DOUBLE.
+   * </pre>
+   *
+   * <code>repeated double double_val = 6 [packed = true];</code>
+   */
+  public double getDoubleVal(int index) {
+    return doubleVal_.getDouble(index);
+  }
+  private int doubleValMemoizedSerializedSize = -1;
+
+  public static final int INT_VAL_FIELD_NUMBER = 7;
+  private com.google.protobuf.Internal.IntList intVal_;
+  /**
+   * <pre>
+   * DT_INT32, DT_INT16, DT_INT8, DT_UINT8.
+   * </pre>
+   *
+   * <code>repeated int32 int_val = 7 [packed = true];</code>
+   */
+  public java.util.List<java.lang.Integer>
+      getIntValList() {
+    return intVal_;
+  }
+  /**
+   * <pre>
+   * DT_INT32, DT_INT16, DT_INT8, DT_UINT8.
+   * </pre>
+   *
+   * <code>repeated int32 int_val = 7 [packed = true];</code>
+   */
+  public int getIntValCount() {
+    return intVal_.size();
+  }
+  /**
+   * <pre>
+   * DT_INT32, DT_INT16, DT_INT8, DT_UINT8.
+   * </pre>
+   *
+   * <code>repeated int32 int_val = 7 [packed = true];</code>
+   */
+  public int getIntVal(int index) {
+    return intVal_.getInt(index);
+  }
+  private int intValMemoizedSerializedSize = -1;
+
+  public static final int STRING_VAL_FIELD_NUMBER = 8;
+  private java.util.List<com.google.protobuf.ByteString> stringVal_;
+  /**
+   * <pre>
+   * DT_STRING
+   * </pre>
+   *
+   * <code>repeated bytes string_val = 8;</code>
+   */
+  public java.util.List<com.google.protobuf.ByteString>
+      getStringValList() {
+    return stringVal_;
+  }
+  /**
+   * <pre>
+   * DT_STRING
+   * </pre>
+   *
+   * <code>repeated bytes string_val = 8;</code>
+   */
+  public int getStringValCount() {
+    return stringVal_.size();
+  }
+  /**
+   * <pre>
+   * DT_STRING
+   * </pre>
+   *
+   * <code>repeated bytes string_val = 8;</code>
+   */
+  public com.google.protobuf.ByteString getStringVal(int index) {
+    return stringVal_.get(index);
+  }
+
+  public static final int SCOMPLEX_VAL_FIELD_NUMBER = 9;
+  private com.google.protobuf.Internal.FloatList scomplexVal_;
+  /**
+   * <pre>
+   * DT_COMPLEX64. scomplex_val(2*i) and scomplex_val(2*i+1) are real
+   * and imaginary parts of i-th single precision complex.
+   * </pre>
+   *
+   * <code>repeated float scomplex_val = 9 [packed = true];</code>
+   */
+  public java.util.List<java.lang.Float>
+      getScomplexValList() {
+    return scomplexVal_;
+  }
+  /**
+   * <pre>
+   * DT_COMPLEX64. scomplex_val(2*i) and scomplex_val(2*i+1) are real
+   * and imaginary parts of i-th single precision complex.
+   * </pre>
+   *
+   * <code>repeated float scomplex_val = 9 [packed = true];</code>
+   */
+  public int getScomplexValCount() {
+    return scomplexVal_.size();
+  }
+  /**
+   * <pre>
+   * DT_COMPLEX64. scomplex_val(2*i) and scomplex_val(2*i+1) are real
+   * and imaginary parts of i-th single precision complex.
+   * </pre>
+   *
+   * <code>repeated float scomplex_val = 9 [packed = true];</code>
+   */
+  public float getScomplexVal(int index) {
+    return scomplexVal_.getFloat(index);
+  }
+  private int scomplexValMemoizedSerializedSize = -1;
+
+  public static final int INT64_VAL_FIELD_NUMBER = 10;
+  private com.google.protobuf.Internal.LongList int64Val_;
+  /**
+   * <pre>
+   * DT_INT64
+   * </pre>
+   *
+   * <code>repeated int64 int64_val = 10 [packed = true];</code>
+   */
+  public java.util.List<java.lang.Long>
+      getInt64ValList() {
+    return int64Val_;
+  }
+  /**
+   * <pre>
+   * DT_INT64
+   * </pre>
+   *
+   * <code>repeated int64 int64_val = 10 [packed = true];</code>
+   */
+  public int getInt64ValCount() {
+    return int64Val_.size();
+  }
+  /**
+   * <pre>
+   * DT_INT64
+   * </pre>
+   *
+   * <code>repeated int64 int64_val = 10 [packed = true];</code>
+   */
+  public long getInt64Val(int index) {
+    return int64Val_.getLong(index);
+  }
+  private int int64ValMemoizedSerializedSize = -1;
+
+  public static final int BOOL_VAL_FIELD_NUMBER = 11;
+  private com.google.protobuf.Internal.BooleanList boolVal_;
+  /**
+   * <pre>
+   * DT_BOOL
+   * </pre>
+   *
+   * <code>repeated bool bool_val = 11 [packed = true];</code>
+   */
+  public java.util.List<java.lang.Boolean>
+      getBoolValList() {
+    return boolVal_;
+  }
+  /**
+   * <pre>
+   * DT_BOOL
+   * </pre>
+   *
+   * <code>repeated bool bool_val = 11 [packed = true];</code>
+   */
+  public int getBoolValCount() {
+    return boolVal_.size();
+  }
+  /**
+   * <pre>
+   * DT_BOOL
+   * </pre>
+   *
+   * <code>repeated bool bool_val = 11 [packed = true];</code>
+   */
+  public boolean getBoolVal(int index) {
+    return boolVal_.getBoolean(index);
+  }
+  private int boolValMemoizedSerializedSize = -1;
+
+  public static final int DCOMPLEX_VAL_FIELD_NUMBER = 12;
+  private com.google.protobuf.Internal.DoubleList dcomplexVal_;
+  /**
+   * <pre>
+   * DT_COMPLEX128. dcomplex_val(2*i) and dcomplex_val(2*i+1) are real
+   * and imaginary parts of i-th double precision complex.
+   * </pre>
+   *
+   * <code>repeated double dcomplex_val = 12 [packed = true];</code>
+   */
+  public java.util.List<java.lang.Double>
+      getDcomplexValList() {
+    return dcomplexVal_;
+  }
+  /**
+   * <pre>
+   * DT_COMPLEX128. dcomplex_val(2*i) and dcomplex_val(2*i+1) are real
+   * and imaginary parts of i-th double precision complex.
+   * </pre>
+   *
+   * <code>repeated double dcomplex_val = 12 [packed = true];</code>
+   */
+  public int getDcomplexValCount() {
+    return dcomplexVal_.size();
+  }
+  /**
+   * <pre>
+   * DT_COMPLEX128. dcomplex_val(2*i) and dcomplex_val(2*i+1) are real
+   * and imaginary parts of i-th double precision complex.
+   * </pre>
+   *
+   * <code>repeated double dcomplex_val = 12 [packed = true];</code>
+   */
+  public double getDcomplexVal(int index) {
+    return dcomplexVal_.getDouble(index);
+  }
+  private int dcomplexValMemoizedSerializedSize = -1;
+
+  public static final int RESOURCE_HANDLE_VAL_FIELD_NUMBER = 14;
+  private java.util.List<org.tensorflow.proto.framework.ResourceHandleProto> resourceHandleVal_;
+  /**
+   * <pre>
+   * DT_RESOURCE
+   * </pre>
+   *
+   * <code>repeated .tensorflow.ResourceHandleProto resource_handle_val = 14;</code>
+   */
+  public java.util.List<org.tensorflow.proto.framework.ResourceHandleProto> getResourceHandleValList() {
+    return resourceHandleVal_;
+  }
+  /**
+   * <pre>
+   * DT_RESOURCE
+   * </pre>
+   *
+   * <code>repeated .tensorflow.ResourceHandleProto resource_handle_val = 14;</code>
+   */
+  public java.util.List<? extends org.tensorflow.proto.framework.ResourceHandleProtoOrBuilder> 
+      getResourceHandleValOrBuilderList() {
+    return resourceHandleVal_;
+  }
+  /**
+   * <pre>
+   * DT_RESOURCE
+   * </pre>
+   *
+   * <code>repeated .tensorflow.ResourceHandleProto resource_handle_val = 14;</code>
+   */
+  public int getResourceHandleValCount() {
+    return resourceHandleVal_.size();
+  }
+  /**
+   * <pre>
+   * DT_RESOURCE
+   * </pre>
+   *
+   * <code>repeated .tensorflow.ResourceHandleProto resource_handle_val = 14;</code>
+   */
+  public org.tensorflow.proto.framework.ResourceHandleProto getResourceHandleVal(int index) {
+    return resourceHandleVal_.get(index);
+  }
+  /**
+   * <pre>
+   * DT_RESOURCE
+   * </pre>
+   *
+   * <code>repeated .tensorflow.ResourceHandleProto resource_handle_val = 14;</code>
+   */
+  public org.tensorflow.proto.framework.ResourceHandleProtoOrBuilder getResourceHandleValOrBuilder(
+      int index) {
+    return resourceHandleVal_.get(index);
+  }
+
+  public static final int VARIANT_VAL_FIELD_NUMBER = 15;
+  private java.util.List<org.tensorflow.proto.framework.VariantTensorDataProto> variantVal_;
+  /**
+   * <pre>
+   * DT_VARIANT
+   * </pre>
+   *
+   * <code>repeated .tensorflow.VariantTensorDataProto variant_val = 15;</code>
+   */
+  public java.util.List<org.tensorflow.proto.framework.VariantTensorDataProto> getVariantValList() {
+    return variantVal_;
+  }
+  /**
+   * <pre>
+   * DT_VARIANT
+   * </pre>
+   *
+   * <code>repeated .tensorflow.VariantTensorDataProto variant_val = 15;</code>
+   */
+  public java.util.List<? extends org.tensorflow.proto.framework.VariantTensorDataProtoOrBuilder> 
+      getVariantValOrBuilderList() {
+    return variantVal_;
+  }
+  /**
+   * <pre>
+   * DT_VARIANT
+   * </pre>
+   *
+   * <code>repeated .tensorflow.VariantTensorDataProto variant_val = 15;</code>
+   */
+  public int getVariantValCount() {
+    return variantVal_.size();
+  }
+  /**
+   * <pre>
+   * DT_VARIANT
+   * </pre>
+   *
+   * <code>repeated .tensorflow.VariantTensorDataProto variant_val = 15;</code>
+   */
+  public org.tensorflow.proto.framework.VariantTensorDataProto getVariantVal(int index) {
+    return variantVal_.get(index);
+  }
+  /**
+   * <pre>
+   * DT_VARIANT
+   * </pre>
+   *
+   * <code>repeated .tensorflow.VariantTensorDataProto variant_val = 15;</code>
+   */
+  public org.tensorflow.proto.framework.VariantTensorDataProtoOrBuilder getVariantValOrBuilder(
+      int index) {
+    return variantVal_.get(index);
+  }
+
+  public static final int UINT32_VAL_FIELD_NUMBER = 16;
+  private com.google.protobuf.Internal.IntList uint32Val_;
+  /**
+   * <pre>
+   * DT_UINT32
+   * </pre>
+   *
+   * <code>repeated uint32 uint32_val = 16 [packed = true];</code>
+   */
+  public java.util.List<java.lang.Integer>
+      getUint32ValList() {
+    return uint32Val_;
+  }
+  /**
+   * <pre>
+   * DT_UINT32
+   * </pre>
+   *
+   * <code>repeated uint32 uint32_val = 16 [packed = true];</code>
+   */
+  public int getUint32ValCount() {
+    return uint32Val_.size();
+  }
+  /**
+   * <pre>
+   * DT_UINT32
+   * </pre>
+   *
+   * <code>repeated uint32 uint32_val = 16 [packed = true];</code>
+   */
+  public int getUint32Val(int index) {
+    return uint32Val_.getInt(index);
+  }
+  private int uint32ValMemoizedSerializedSize = -1;
+
+  public static final int UINT64_VAL_FIELD_NUMBER = 17;
+  private com.google.protobuf.Internal.LongList uint64Val_;
+  /**
+   * <pre>
+   * DT_UINT64
+   * </pre>
+   *
+   * <code>repeated uint64 uint64_val = 17 [packed = true];</code>
+   */
+  public java.util.List<java.lang.Long>
+      getUint64ValList() {
+    return uint64Val_;
+  }
+  /**
+   * <pre>
+   * DT_UINT64
+   * </pre>
+   *
+   * <code>repeated uint64 uint64_val = 17 [packed = true];</code>
+   */
+  public int getUint64ValCount() {
+    return uint64Val_.size();
+  }
+  /**
+   * <pre>
+   * DT_UINT64
+   * </pre>
+   *
+   * <code>repeated uint64 uint64_val = 17 [packed = true];</code>
+   */
+  public long getUint64Val(int index) {
+    return uint64Val_.getLong(index);
+  }
+  private int uint64ValMemoizedSerializedSize = -1;
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    getSerializedSize();
+    if (dtype_ != org.tensorflow.proto.framework.DataType.DT_INVALID.getNumber()) {
+      output.writeEnum(1, dtype_);
+    }
+    if (tensorShape_ != null) {
+      output.writeMessage(2, getTensorShape());
+    }
+    if (versionNumber_ != 0) {
+      output.writeInt32(3, versionNumber_);
+    }
+    if (!tensorContent_.isEmpty()) {
+      output.writeBytes(4, tensorContent_);
+    }
+    if (getFloatValList().size() > 0) {
+      output.writeUInt32NoTag(42);
+      output.writeUInt32NoTag(floatValMemoizedSerializedSize);
+    }
+    for (int i = 0; i < floatVal_.size(); i++) {
+      output.writeFloatNoTag(floatVal_.getFloat(i));
+    }
+    if (getDoubleValList().size() > 0) {
+      output.writeUInt32NoTag(50);
+      output.writeUInt32NoTag(doubleValMemoizedSerializedSize);
+    }
+    for (int i = 0; i < doubleVal_.size(); i++) {
+      output.writeDoubleNoTag(doubleVal_.getDouble(i));
+    }
+    if (getIntValList().size() > 0) {
+      output.writeUInt32NoTag(58);
+      output.writeUInt32NoTag(intValMemoizedSerializedSize);
+    }
+    for (int i = 0; i < intVal_.size(); i++) {
+      output.writeInt32NoTag(intVal_.getInt(i));
+    }
+    for (int i = 0; i < stringVal_.size(); i++) {
+      output.writeBytes(8, stringVal_.get(i));
+    }
+    if (getScomplexValList().size() > 0) {
+      output.writeUInt32NoTag(74);
+      output.writeUInt32NoTag(scomplexValMemoizedSerializedSize);
+    }
+    for (int i = 0; i < scomplexVal_.size(); i++) {
+      output.writeFloatNoTag(scomplexVal_.getFloat(i));
+    }
+    if (getInt64ValList().size() > 0) {
+      output.writeUInt32NoTag(82);
+      output.writeUInt32NoTag(int64ValMemoizedSerializedSize);
+    }
+    for (int i = 0; i < int64Val_.size(); i++) {
+      output.writeInt64NoTag(int64Val_.getLong(i));
+    }
+    if (getBoolValList().size() > 0) {
+      output.writeUInt32NoTag(90);
+      output.writeUInt32NoTag(boolValMemoizedSerializedSize);
+    }
+    for (int i = 0; i < boolVal_.size(); i++) {
+      output.writeBoolNoTag(boolVal_.getBoolean(i));
+    }
+    if (getDcomplexValList().size() > 0) {
+      output.writeUInt32NoTag(98);
+      output.writeUInt32NoTag(dcomplexValMemoizedSerializedSize);
+    }
+    for (int i = 0; i < dcomplexVal_.size(); i++) {
+      output.writeDoubleNoTag(dcomplexVal_.getDouble(i));
+    }
+    if (getHalfValList().size() > 0) {
+      output.writeUInt32NoTag(106);
+      output.writeUInt32NoTag(halfValMemoizedSerializedSize);
+    }
+    for (int i = 0; i < halfVal_.size(); i++) {
+      output.writeInt32NoTag(halfVal_.getInt(i));
+    }
+    for (int i = 0; i < resourceHandleVal_.size(); i++) {
+      output.writeMessage(14, resourceHandleVal_.get(i));
+    }
+    for (int i = 0; i < variantVal_.size(); i++) {
+      output.writeMessage(15, variantVal_.get(i));
+    }
+    if (getUint32ValList().size() > 0) {
+      output.writeUInt32NoTag(130);
+      output.writeUInt32NoTag(uint32ValMemoizedSerializedSize);
+    }
+    for (int i = 0; i < uint32Val_.size(); i++) {
+      output.writeUInt32NoTag(uint32Val_.getInt(i));
+    }
+    if (getUint64ValList().size() > 0) {
+      output.writeUInt32NoTag(138);
+      output.writeUInt32NoTag(uint64ValMemoizedSerializedSize);
+    }
+    for (int i = 0; i < uint64Val_.size(); i++) {
+      output.writeUInt64NoTag(uint64Val_.getLong(i));
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (dtype_ != org.tensorflow.proto.framework.DataType.DT_INVALID.getNumber()) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeEnumSize(1, dtype_);
+    }
+    if (tensorShape_ != null) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(2, getTensorShape());
+    }
+    if (versionNumber_ != 0) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt32Size(3, versionNumber_);
+    }
+    if (!tensorContent_.isEmpty()) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeBytesSize(4, tensorContent_);
+    }
+    {
+      int dataSize = 0;
+      dataSize = 4 * getFloatValList().size();
+      size += dataSize;
+      if (!getFloatValList().isEmpty()) {
+        size += 1;
+        size += com.google.protobuf.CodedOutputStream
+            .computeInt32SizeNoTag(dataSize);
+      }
+      floatValMemoizedSerializedSize = dataSize;
+    }
+    {
+      int dataSize = 0;
+      dataSize = 8 * getDoubleValList().size();
+      size += dataSize;
+      if (!getDoubleValList().isEmpty()) {
+        size += 1;
+        size += com.google.protobuf.CodedOutputStream
+            .computeInt32SizeNoTag(dataSize);
+      }
+      doubleValMemoizedSerializedSize = dataSize;
+    }
+    {
+      int dataSize = 0;
+      for (int i = 0; i < intVal_.size(); i++) {
+        dataSize += com.google.protobuf.CodedOutputStream
+          .computeInt32SizeNoTag(intVal_.getInt(i));
+      }
+      size += dataSize;
+      if (!getIntValList().isEmpty()) {
+        size += 1;
+        size += com.google.protobuf.CodedOutputStream
+            .computeInt32SizeNoTag(dataSize);
+      }
+      intValMemoizedSerializedSize = dataSize;
+    }
+    {
+      int dataSize = 0;
+      for (int i = 0; i < stringVal_.size(); i++) {
+        dataSize += com.google.protobuf.CodedOutputStream
+          .computeBytesSizeNoTag(stringVal_.get(i));
+      }
+      size += dataSize;
+      size += 1 * getStringValList().size();
+    }
+    {
+      int dataSize = 0;
+      dataSize = 4 * getScomplexValList().size();
+      size += dataSize;
+      if (!getScomplexValList().isEmpty()) {
+        size += 1;
+        size += com.google.protobuf.CodedOutputStream
+            .computeInt32SizeNoTag(dataSize);
+      }
+      scomplexValMemoizedSerializedSize = dataSize;
+    }
+    {
+      int dataSize = 0;
+      for (int i = 0; i < int64Val_.size(); i++) {
+        dataSize += com.google.protobuf.CodedOutputStream
+          .computeInt64SizeNoTag(int64Val_.getLong(i));
+      }
+      size += dataSize;
+      if (!getInt64ValList().isEmpty()) {
+        size += 1;
+        size += com.google.protobuf.CodedOutputStream
+            .computeInt32SizeNoTag(dataSize);
+      }
+      int64ValMemoizedSerializedSize = dataSize;
+    }
+    {
+      int dataSize = 0;
+      dataSize = 1 * getBoolValList().size();
+      size += dataSize;
+      if (!getBoolValList().isEmpty()) {
+        size += 1;
+        size += com.google.protobuf.CodedOutputStream
+            .computeInt32SizeNoTag(dataSize);
+      }
+      boolValMemoizedSerializedSize = dataSize;
+    }
+    {
+      int dataSize = 0;
+      dataSize = 8 * getDcomplexValList().size();
+      size += dataSize;
+      if (!getDcomplexValList().isEmpty()) {
+        size += 1;
+        size += com.google.protobuf.CodedOutputStream
+            .computeInt32SizeNoTag(dataSize);
+      }
+      dcomplexValMemoizedSerializedSize = dataSize;
+    }
+    {
+      int dataSize = 0;
+      for (int i = 0; i < halfVal_.size(); i++) {
+        dataSize += com.google.protobuf.CodedOutputStream
+          .computeInt32SizeNoTag(halfVal_.getInt(i));
+      }
+      size += dataSize;
+      if (!getHalfValList().isEmpty()) {
+        size += 1;
+        size += com.google.protobuf.CodedOutputStream
+            .computeInt32SizeNoTag(dataSize);
+      }
+      halfValMemoizedSerializedSize = dataSize;
+    }
+    for (int i = 0; i < resourceHandleVal_.size(); i++) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(14, resourceHandleVal_.get(i));
+    }
+    for (int i = 0; i < variantVal_.size(); i++) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(15, variantVal_.get(i));
+    }
+    {
+      int dataSize = 0;
+      for (int i = 0; i < uint32Val_.size(); i++) {
+        dataSize += com.google.protobuf.CodedOutputStream
+          .computeUInt32SizeNoTag(uint32Val_.getInt(i));
+      }
+      size += dataSize;
+      if (!getUint32ValList().isEmpty()) {
+        size += 2;
+        size += com.google.protobuf.CodedOutputStream
+            .computeInt32SizeNoTag(dataSize);
+      }
+      uint32ValMemoizedSerializedSize = dataSize;
+    }
+    {
+      int dataSize = 0;
+      for (int i = 0; i < uint64Val_.size(); i++) {
+        dataSize += com.google.protobuf.CodedOutputStream
+          .computeUInt64SizeNoTag(uint64Val_.getLong(i));
+      }
+      size += dataSize;
+      if (!getUint64ValList().isEmpty()) {
+        size += 2;
+        size += com.google.protobuf.CodedOutputStream
+            .computeInt32SizeNoTag(dataSize);
+      }
+      uint64ValMemoizedSerializedSize = dataSize;
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.framework.TensorProto)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.framework.TensorProto other = (org.tensorflow.proto.framework.TensorProto) obj;
+
+    if (dtype_ != other.dtype_) return false;
+    if (hasTensorShape() != other.hasTensorShape()) return false;
+    if (hasTensorShape()) {
+      if (!getTensorShape()
+          .equals(other.getTensorShape())) return false;
+    }
+    if (getVersionNumber()
+        != other.getVersionNumber()) return false;
+    if (!getTensorContent()
+        .equals(other.getTensorContent())) return false;
+    if (!getHalfValList()
+        .equals(other.getHalfValList())) return false;
+    if (!getFloatValList()
+        .equals(other.getFloatValList())) return false;
+    if (!getDoubleValList()
+        .equals(other.getDoubleValList())) return false;
+    if (!getIntValList()
+        .equals(other.getIntValList())) return false;
+    if (!getStringValList()
+        .equals(other.getStringValList())) return false;
+    if (!getScomplexValList()
+        .equals(other.getScomplexValList())) return false;
+    if (!getInt64ValList()
+        .equals(other.getInt64ValList())) return false;
+    if (!getBoolValList()
+        .equals(other.getBoolValList())) return false;
+    if (!getDcomplexValList()
+        .equals(other.getDcomplexValList())) return false;
+    if (!getResourceHandleValList()
+        .equals(other.getResourceHandleValList())) return false;
+    if (!getVariantValList()
+        .equals(other.getVariantValList())) return false;
+    if (!getUint32ValList()
+        .equals(other.getUint32ValList())) return false;
+    if (!getUint64ValList()
+        .equals(other.getUint64ValList())) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    hash = (37 * hash) + DTYPE_FIELD_NUMBER;
+    hash = (53 * hash) + dtype_;
+    if (hasTensorShape()) {
+      hash = (37 * hash) + TENSOR_SHAPE_FIELD_NUMBER;
+      hash = (53 * hash) + getTensorShape().hashCode();
+    }
+    hash = (37 * hash) + VERSION_NUMBER_FIELD_NUMBER;
+    hash = (53 * hash) + getVersionNumber();
+    hash = (37 * hash) + TENSOR_CONTENT_FIELD_NUMBER;
+    hash = (53 * hash) + getTensorContent().hashCode();
+    if (getHalfValCount() > 0) {
+      hash = (37 * hash) + HALF_VAL_FIELD_NUMBER;
+      hash = (53 * hash) + getHalfValList().hashCode();
+    }
+    if (getFloatValCount() > 0) {
+      hash = (37 * hash) + FLOAT_VAL_FIELD_NUMBER;
+      hash = (53 * hash) + getFloatValList().hashCode();
+    }
+    if (getDoubleValCount() > 0) {
+      hash = (37 * hash) + DOUBLE_VAL_FIELD_NUMBER;
+      hash = (53 * hash) + getDoubleValList().hashCode();
+    }
+    if (getIntValCount() > 0) {
+      hash = (37 * hash) + INT_VAL_FIELD_NUMBER;
+      hash = (53 * hash) + getIntValList().hashCode();
+    }
+    if (getStringValCount() > 0) {
+      hash = (37 * hash) + STRING_VAL_FIELD_NUMBER;
+      hash = (53 * hash) + getStringValList().hashCode();
+    }
+    if (getScomplexValCount() > 0) {
+      hash = (37 * hash) + SCOMPLEX_VAL_FIELD_NUMBER;
+      hash = (53 * hash) + getScomplexValList().hashCode();
+    }
+    if (getInt64ValCount() > 0) {
+      hash = (37 * hash) + INT64_VAL_FIELD_NUMBER;
+      hash = (53 * hash) + getInt64ValList().hashCode();
+    }
+    if (getBoolValCount() > 0) {
+      hash = (37 * hash) + BOOL_VAL_FIELD_NUMBER;
+      hash = (53 * hash) + getBoolValList().hashCode();
+    }
+    if (getDcomplexValCount() > 0) {
+      hash = (37 * hash) + DCOMPLEX_VAL_FIELD_NUMBER;
+      hash = (53 * hash) + getDcomplexValList().hashCode();
+    }
+    if (getResourceHandleValCount() > 0) {
+      hash = (37 * hash) + RESOURCE_HANDLE_VAL_FIELD_NUMBER;
+      hash = (53 * hash) + getResourceHandleValList().hashCode();
+    }
+    if (getVariantValCount() > 0) {
+      hash = (37 * hash) + VARIANT_VAL_FIELD_NUMBER;
+      hash = (53 * hash) + getVariantValList().hashCode();
+    }
+    if (getUint32ValCount() > 0) {
+      hash = (37 * hash) + UINT32_VAL_FIELD_NUMBER;
+      hash = (53 * hash) + getUint32ValList().hashCode();
+    }
+    if (getUint64ValCount() > 0) {
+      hash = (37 * hash) + UINT64_VAL_FIELD_NUMBER;
+      hash = (53 * hash) + getUint64ValList().hashCode();
+    }
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.framework.TensorProto parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.TensorProto parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.TensorProto parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.TensorProto parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.TensorProto parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.TensorProto parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.TensorProto parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.TensorProto parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.TensorProto parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.TensorProto parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.TensorProto parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.TensorProto parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.framework.TensorProto prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * <pre>
+   * Protocol buffer representing a tensor.
+   * </pre>
+   *
+   * Protobuf type {@code tensorflow.TensorProto}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.TensorProto)
+      org.tensorflow.proto.framework.TensorProtoOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.TensorProtos.internal_static_tensorflow_TensorProto_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.TensorProtos.internal_static_tensorflow_TensorProto_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.TensorProto.class, org.tensorflow.proto.framework.TensorProto.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.framework.TensorProto.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+        getResourceHandleValFieldBuilder();
+        getVariantValFieldBuilder();
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      dtype_ = 0;
+
+      if (tensorShapeBuilder_ == null) {
+        tensorShape_ = null;
+      } else {
+        tensorShape_ = null;
+        tensorShapeBuilder_ = null;
+      }
+      versionNumber_ = 0;
+
+      tensorContent_ = com.google.protobuf.ByteString.EMPTY;
+
+      halfVal_ = emptyIntList();
+      bitField0_ = (bitField0_ & ~0x00000001);
+      floatVal_ = emptyFloatList();
+      bitField0_ = (bitField0_ & ~0x00000002);
+      doubleVal_ = emptyDoubleList();
+      bitField0_ = (bitField0_ & ~0x00000004);
+      intVal_ = emptyIntList();
+      bitField0_ = (bitField0_ & ~0x00000008);
+      stringVal_ = java.util.Collections.emptyList();
+      bitField0_ = (bitField0_ & ~0x00000010);
+      scomplexVal_ = emptyFloatList();
+      bitField0_ = (bitField0_ & ~0x00000020);
+      int64Val_ = emptyLongList();
+      bitField0_ = (bitField0_ & ~0x00000040);
+      boolVal_ = emptyBooleanList();
+      bitField0_ = (bitField0_ & ~0x00000080);
+      dcomplexVal_ = emptyDoubleList();
+      bitField0_ = (bitField0_ & ~0x00000100);
+      if (resourceHandleValBuilder_ == null) {
+        resourceHandleVal_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000200);
+      } else {
+        resourceHandleValBuilder_.clear();
+      }
+      if (variantValBuilder_ == null) {
+        variantVal_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000400);
+      } else {
+        variantValBuilder_.clear();
+      }
+      uint32Val_ = emptyIntList();
+      bitField0_ = (bitField0_ & ~0x00000800);
+      uint64Val_ = emptyLongList();
+      bitField0_ = (bitField0_ & ~0x00001000);
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.framework.TensorProtos.internal_static_tensorflow_TensorProto_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.TensorProto getDefaultInstanceForType() {
+      return org.tensorflow.proto.framework.TensorProto.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.TensorProto build() {
+      org.tensorflow.proto.framework.TensorProto result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.TensorProto buildPartial() {
+      org.tensorflow.proto.framework.TensorProto result = new org.tensorflow.proto.framework.TensorProto(this);
+      int from_bitField0_ = bitField0_;
+      result.dtype_ = dtype_;
+      if (tensorShapeBuilder_ == null) {
+        result.tensorShape_ = tensorShape_;
+      } else {
+        result.tensorShape_ = tensorShapeBuilder_.build();
+      }
+      result.versionNumber_ = versionNumber_;
+      result.tensorContent_ = tensorContent_;
+      if (((bitField0_ & 0x00000001) != 0)) {
+        halfVal_.makeImmutable();
+        bitField0_ = (bitField0_ & ~0x00000001);
+      }
+      result.halfVal_ = halfVal_;
+      if (((bitField0_ & 0x00000002) != 0)) {
+        floatVal_.makeImmutable();
+        bitField0_ = (bitField0_ & ~0x00000002);
+      }
+      result.floatVal_ = floatVal_;
+      if (((bitField0_ & 0x00000004) != 0)) {
+        doubleVal_.makeImmutable();
+        bitField0_ = (bitField0_ & ~0x00000004);
+      }
+      result.doubleVal_ = doubleVal_;
+      if (((bitField0_ & 0x00000008) != 0)) {
+        intVal_.makeImmutable();
+        bitField0_ = (bitField0_ & ~0x00000008);
+      }
+      result.intVal_ = intVal_;
+      if (((bitField0_ & 0x00000010) != 0)) {
+        stringVal_ = java.util.Collections.unmodifiableList(stringVal_);
+        bitField0_ = (bitField0_ & ~0x00000010);
+      }
+      result.stringVal_ = stringVal_;
+      if (((bitField0_ & 0x00000020) != 0)) {
+        scomplexVal_.makeImmutable();
+        bitField0_ = (bitField0_ & ~0x00000020);
+      }
+      result.scomplexVal_ = scomplexVal_;
+      if (((bitField0_ & 0x00000040) != 0)) {
+        int64Val_.makeImmutable();
+        bitField0_ = (bitField0_ & ~0x00000040);
+      }
+      result.int64Val_ = int64Val_;
+      if (((bitField0_ & 0x00000080) != 0)) {
+        boolVal_.makeImmutable();
+        bitField0_ = (bitField0_ & ~0x00000080);
+      }
+      result.boolVal_ = boolVal_;
+      if (((bitField0_ & 0x00000100) != 0)) {
+        dcomplexVal_.makeImmutable();
+        bitField0_ = (bitField0_ & ~0x00000100);
+      }
+      result.dcomplexVal_ = dcomplexVal_;
+      if (resourceHandleValBuilder_ == null) {
+        if (((bitField0_ & 0x00000200) != 0)) {
+          resourceHandleVal_ = java.util.Collections.unmodifiableList(resourceHandleVal_);
+          bitField0_ = (bitField0_ & ~0x00000200);
+        }
+        result.resourceHandleVal_ = resourceHandleVal_;
+      } else {
+        result.resourceHandleVal_ = resourceHandleValBuilder_.build();
+      }
+      if (variantValBuilder_ == null) {
+        if (((bitField0_ & 0x00000400) != 0)) {
+          variantVal_ = java.util.Collections.unmodifiableList(variantVal_);
+          bitField0_ = (bitField0_ & ~0x00000400);
+        }
+        result.variantVal_ = variantVal_;
+      } else {
+        result.variantVal_ = variantValBuilder_.build();
+      }
+      if (((bitField0_ & 0x00000800) != 0)) {
+        uint32Val_.makeImmutable();
+        bitField0_ = (bitField0_ & ~0x00000800);
+      }
+      result.uint32Val_ = uint32Val_;
+      if (((bitField0_ & 0x00001000) != 0)) {
+        uint64Val_.makeImmutable();
+        bitField0_ = (bitField0_ & ~0x00001000);
+      }
+      result.uint64Val_ = uint64Val_;
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.framework.TensorProto) {
+        return mergeFrom((org.tensorflow.proto.framework.TensorProto)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.framework.TensorProto other) {
+      if (other == org.tensorflow.proto.framework.TensorProto.getDefaultInstance()) return this;
+      if (other.dtype_ != 0) {
+        setDtypeValue(other.getDtypeValue());
+      }
+      if (other.hasTensorShape()) {
+        mergeTensorShape(other.getTensorShape());
+      }
+      if (other.getVersionNumber() != 0) {
+        setVersionNumber(other.getVersionNumber());
+      }
+      if (other.getTensorContent() != com.google.protobuf.ByteString.EMPTY) {
+        setTensorContent(other.getTensorContent());
+      }
+      if (!other.halfVal_.isEmpty()) {
+        if (halfVal_.isEmpty()) {
+          halfVal_ = other.halfVal_;
+          bitField0_ = (bitField0_ & ~0x00000001);
+        } else {
+          ensureHalfValIsMutable();
+          halfVal_.addAll(other.halfVal_);
+        }
+        onChanged();
+      }
+      if (!other.floatVal_.isEmpty()) {
+        if (floatVal_.isEmpty()) {
+          floatVal_ = other.floatVal_;
+          bitField0_ = (bitField0_ & ~0x00000002);
+        } else {
+          ensureFloatValIsMutable();
+          floatVal_.addAll(other.floatVal_);
+        }
+        onChanged();
+      }
+      if (!other.doubleVal_.isEmpty()) {
+        if (doubleVal_.isEmpty()) {
+          doubleVal_ = other.doubleVal_;
+          bitField0_ = (bitField0_ & ~0x00000004);
+        } else {
+          ensureDoubleValIsMutable();
+          doubleVal_.addAll(other.doubleVal_);
+        }
+        onChanged();
+      }
+      if (!other.intVal_.isEmpty()) {
+        if (intVal_.isEmpty()) {
+          intVal_ = other.intVal_;
+          bitField0_ = (bitField0_ & ~0x00000008);
+        } else {
+          ensureIntValIsMutable();
+          intVal_.addAll(other.intVal_);
+        }
+        onChanged();
+      }
+      if (!other.stringVal_.isEmpty()) {
+        if (stringVal_.isEmpty()) {
+          stringVal_ = other.stringVal_;
+          bitField0_ = (bitField0_ & ~0x00000010);
+        } else {
+          ensureStringValIsMutable();
+          stringVal_.addAll(other.stringVal_);
+        }
+        onChanged();
+      }
+      if (!other.scomplexVal_.isEmpty()) {
+        if (scomplexVal_.isEmpty()) {
+          scomplexVal_ = other.scomplexVal_;
+          bitField0_ = (bitField0_ & ~0x00000020);
+        } else {
+          ensureScomplexValIsMutable();
+          scomplexVal_.addAll(other.scomplexVal_);
+        }
+        onChanged();
+      }
+      if (!other.int64Val_.isEmpty()) {
+        if (int64Val_.isEmpty()) {
+          int64Val_ = other.int64Val_;
+          bitField0_ = (bitField0_ & ~0x00000040);
+        } else {
+          ensureInt64ValIsMutable();
+          int64Val_.addAll(other.int64Val_);
+        }
+        onChanged();
+      }
+      if (!other.boolVal_.isEmpty()) {
+        if (boolVal_.isEmpty()) {
+          boolVal_ = other.boolVal_;
+          bitField0_ = (bitField0_ & ~0x00000080);
+        } else {
+          ensureBoolValIsMutable();
+          boolVal_.addAll(other.boolVal_);
+        }
+        onChanged();
+      }
+      if (!other.dcomplexVal_.isEmpty()) {
+        if (dcomplexVal_.isEmpty()) {
+          dcomplexVal_ = other.dcomplexVal_;
+          bitField0_ = (bitField0_ & ~0x00000100);
+        } else {
+          ensureDcomplexValIsMutable();
+          dcomplexVal_.addAll(other.dcomplexVal_);
+        }
+        onChanged();
+      }
+      if (resourceHandleValBuilder_ == null) {
+        if (!other.resourceHandleVal_.isEmpty()) {
+          if (resourceHandleVal_.isEmpty()) {
+            resourceHandleVal_ = other.resourceHandleVal_;
+            bitField0_ = (bitField0_ & ~0x00000200);
+          } else {
+            ensureResourceHandleValIsMutable();
+            resourceHandleVal_.addAll(other.resourceHandleVal_);
+          }
+          onChanged();
+        }
+      } else {
+        if (!other.resourceHandleVal_.isEmpty()) {
+          if (resourceHandleValBuilder_.isEmpty()) {
+            resourceHandleValBuilder_.dispose();
+            resourceHandleValBuilder_ = null;
+            resourceHandleVal_ = other.resourceHandleVal_;
+            bitField0_ = (bitField0_ & ~0x00000200);
+            resourceHandleValBuilder_ = 
+              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
+                 getResourceHandleValFieldBuilder() : null;
+          } else {
+            resourceHandleValBuilder_.addAllMessages(other.resourceHandleVal_);
+          }
+        }
+      }
+      if (variantValBuilder_ == null) {
+        if (!other.variantVal_.isEmpty()) {
+          if (variantVal_.isEmpty()) {
+            variantVal_ = other.variantVal_;
+            bitField0_ = (bitField0_ & ~0x00000400);
+          } else {
+            ensureVariantValIsMutable();
+            variantVal_.addAll(other.variantVal_);
+          }
+          onChanged();
+        }
+      } else {
+        if (!other.variantVal_.isEmpty()) {
+          if (variantValBuilder_.isEmpty()) {
+            variantValBuilder_.dispose();
+            variantValBuilder_ = null;
+            variantVal_ = other.variantVal_;
+            bitField0_ = (bitField0_ & ~0x00000400);
+            variantValBuilder_ = 
+              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
+                 getVariantValFieldBuilder() : null;
+          } else {
+            variantValBuilder_.addAllMessages(other.variantVal_);
+          }
+        }
+      }
+      if (!other.uint32Val_.isEmpty()) {
+        if (uint32Val_.isEmpty()) {
+          uint32Val_ = other.uint32Val_;
+          bitField0_ = (bitField0_ & ~0x00000800);
+        } else {
+          ensureUint32ValIsMutable();
+          uint32Val_.addAll(other.uint32Val_);
+        }
+        onChanged();
+      }
+      if (!other.uint64Val_.isEmpty()) {
+        if (uint64Val_.isEmpty()) {
+          uint64Val_ = other.uint64Val_;
+          bitField0_ = (bitField0_ & ~0x00001000);
+        } else {
+          ensureUint64ValIsMutable();
+          uint64Val_.addAll(other.uint64Val_);
+        }
+        onChanged();
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.framework.TensorProto parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.framework.TensorProto) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+    private int bitField0_;
+
+    private int dtype_ = 0;
+    /**
+     * <code>.tensorflow.DataType dtype = 1;</code>
+     */
+    public int getDtypeValue() {
+      return dtype_;
+    }
+    /**
+     * <code>.tensorflow.DataType dtype = 1;</code>
+     */
+    public Builder setDtypeValue(int value) {
+      dtype_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>.tensorflow.DataType dtype = 1;</code>
+     */
+    public org.tensorflow.proto.framework.DataType getDtype() {
+      @SuppressWarnings("deprecation")
+      org.tensorflow.proto.framework.DataType result = org.tensorflow.proto.framework.DataType.valueOf(dtype_);
+      return result == null ? org.tensorflow.proto.framework.DataType.UNRECOGNIZED : result;
+    }
+    /**
+     * <code>.tensorflow.DataType dtype = 1;</code>
+     */
+    public Builder setDtype(org.tensorflow.proto.framework.DataType value) {
+      if (value == null) {
+        throw new NullPointerException();
+      }
+      
+      dtype_ = value.getNumber();
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>.tensorflow.DataType dtype = 1;</code>
+     */
+    public Builder clearDtype() {
+      
+      dtype_ = 0;
+      onChanged();
+      return this;
+    }
+
+    private org.tensorflow.proto.framework.TensorShapeProto tensorShape_;
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.TensorShapeProto, org.tensorflow.proto.framework.TensorShapeProto.Builder, org.tensorflow.proto.framework.TensorShapeProtoOrBuilder> tensorShapeBuilder_;
+    /**
+     * <pre>
+     * Shape of the tensor.  TODO(touts): sort out the 0-rank issues.
+     * </pre>
+     *
+     * <code>.tensorflow.TensorShapeProto tensor_shape = 2;</code>
+     */
+    public boolean hasTensorShape() {
+      return tensorShapeBuilder_ != null || tensorShape_ != null;
+    }
+    /**
+     * <pre>
+     * Shape of the tensor.  TODO(touts): sort out the 0-rank issues.
+     * </pre>
+     *
+     * <code>.tensorflow.TensorShapeProto tensor_shape = 2;</code>
+     */
+    public org.tensorflow.proto.framework.TensorShapeProto getTensorShape() {
+      if (tensorShapeBuilder_ == null) {
+        return tensorShape_ == null ? org.tensorflow.proto.framework.TensorShapeProto.getDefaultInstance() : tensorShape_;
+      } else {
+        return tensorShapeBuilder_.getMessage();
+      }
+    }
+    /**
+     * <pre>
+     * Shape of the tensor.  TODO(touts): sort out the 0-rank issues.
+     * </pre>
+     *
+     * <code>.tensorflow.TensorShapeProto tensor_shape = 2;</code>
+     */
+    public Builder setTensorShape(org.tensorflow.proto.framework.TensorShapeProto value) {
+      if (tensorShapeBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        tensorShape_ = value;
+        onChanged();
+      } else {
+        tensorShapeBuilder_.setMessage(value);
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Shape of the tensor.  TODO(touts): sort out the 0-rank issues.
+     * </pre>
+     *
+     * <code>.tensorflow.TensorShapeProto tensor_shape = 2;</code>
+     */
+    public Builder setTensorShape(
+        org.tensorflow.proto.framework.TensorShapeProto.Builder builderForValue) {
+      if (tensorShapeBuilder_ == null) {
+        tensorShape_ = builderForValue.build();
+        onChanged();
+      } else {
+        tensorShapeBuilder_.setMessage(builderForValue.build());
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Shape of the tensor.  TODO(touts): sort out the 0-rank issues.
+     * </pre>
+     *
+     * <code>.tensorflow.TensorShapeProto tensor_shape = 2;</code>
+     */
+    public Builder mergeTensorShape(org.tensorflow.proto.framework.TensorShapeProto value) {
+      if (tensorShapeBuilder_ == null) {
+        if (tensorShape_ != null) {
+          tensorShape_ =
+            org.tensorflow.proto.framework.TensorShapeProto.newBuilder(tensorShape_).mergeFrom(value).buildPartial();
+        } else {
+          tensorShape_ = value;
+        }
+        onChanged();
+      } else {
+        tensorShapeBuilder_.mergeFrom(value);
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Shape of the tensor.  TODO(touts): sort out the 0-rank issues.
+     * </pre>
+     *
+     * <code>.tensorflow.TensorShapeProto tensor_shape = 2;</code>
+     */
+    public Builder clearTensorShape() {
+      if (tensorShapeBuilder_ == null) {
+        tensorShape_ = null;
+        onChanged();
+      } else {
+        tensorShape_ = null;
+        tensorShapeBuilder_ = null;
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Shape of the tensor.  TODO(touts): sort out the 0-rank issues.
+     * </pre>
+     *
+     * <code>.tensorflow.TensorShapeProto tensor_shape = 2;</code>
+     */
+    public org.tensorflow.proto.framework.TensorShapeProto.Builder getTensorShapeBuilder() {
+      
+      onChanged();
+      return getTensorShapeFieldBuilder().getBuilder();
+    }
+    /**
+     * <pre>
+     * Shape of the tensor.  TODO(touts): sort out the 0-rank issues.
+     * </pre>
+     *
+     * <code>.tensorflow.TensorShapeProto tensor_shape = 2;</code>
+     */
+    public org.tensorflow.proto.framework.TensorShapeProtoOrBuilder getTensorShapeOrBuilder() {
+      if (tensorShapeBuilder_ != null) {
+        return tensorShapeBuilder_.getMessageOrBuilder();
+      } else {
+        return tensorShape_ == null ?
+            org.tensorflow.proto.framework.TensorShapeProto.getDefaultInstance() : tensorShape_;
+      }
+    }
+    /**
+     * <pre>
+     * Shape of the tensor.  TODO(touts): sort out the 0-rank issues.
+     * </pre>
+     *
+     * <code>.tensorflow.TensorShapeProto tensor_shape = 2;</code>
+     */
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.TensorShapeProto, org.tensorflow.proto.framework.TensorShapeProto.Builder, org.tensorflow.proto.framework.TensorShapeProtoOrBuilder> 
+        getTensorShapeFieldBuilder() {
+      if (tensorShapeBuilder_ == null) {
+        tensorShapeBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+            org.tensorflow.proto.framework.TensorShapeProto, org.tensorflow.proto.framework.TensorShapeProto.Builder, org.tensorflow.proto.framework.TensorShapeProtoOrBuilder>(
+                getTensorShape(),
+                getParentForChildren(),
+                isClean());
+        tensorShape_ = null;
+      }
+      return tensorShapeBuilder_;
+    }
+
+    private int versionNumber_ ;
+    /**
+     * <pre>
+     * Version number.
+     * In version 0, if the "repeated xxx" representations contain only one
+     * element, that element is repeated to fill the shape.  This makes it easy
+     * to represent a constant Tensor with a single value.
+     * </pre>
+     *
+     * <code>int32 version_number = 3;</code>
+     */
+    public int getVersionNumber() {
+      return versionNumber_;
+    }
+    /**
+     * <pre>
+     * Version number.
+     * In version 0, if the "repeated xxx" representations contain only one
+     * element, that element is repeated to fill the shape.  This makes it easy
+     * to represent a constant Tensor with a single value.
+     * </pre>
+     *
+     * <code>int32 version_number = 3;</code>
+     */
+    public Builder setVersionNumber(int value) {
+      
+      versionNumber_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Version number.
+     * In version 0, if the "repeated xxx" representations contain only one
+     * element, that element is repeated to fill the shape.  This makes it easy
+     * to represent a constant Tensor with a single value.
+     * </pre>
+     *
+     * <code>int32 version_number = 3;</code>
+     */
+    public Builder clearVersionNumber() {
+      
+      versionNumber_ = 0;
+      onChanged();
+      return this;
+    }
+
+    private com.google.protobuf.ByteString tensorContent_ = com.google.protobuf.ByteString.EMPTY;
+    /**
+     * <pre>
+     * Serialized raw tensor content from either Tensor::AsProtoTensorContent or
+     * memcpy in tensorflow::grpc::EncodeTensorToByteBuffer. This representation
+     * can be used for all tensor types. The purpose of this representation is to
+     * reduce serialization overhead during RPC call by avoiding serialization of
+     * many repeated small items.
+     * </pre>
+     *
+     * <code>bytes tensor_content = 4;</code>
+     */
+    public com.google.protobuf.ByteString getTensorContent() {
+      return tensorContent_;
+    }
+    /**
+     * <pre>
+     * Serialized raw tensor content from either Tensor::AsProtoTensorContent or
+     * memcpy in tensorflow::grpc::EncodeTensorToByteBuffer. This representation
+     * can be used for all tensor types. The purpose of this representation is to
+     * reduce serialization overhead during RPC call by avoiding serialization of
+     * many repeated small items.
+     * </pre>
+     *
+     * <code>bytes tensor_content = 4;</code>
+     */
+    public Builder setTensorContent(com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      tensorContent_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Serialized raw tensor content from either Tensor::AsProtoTensorContent or
+     * memcpy in tensorflow::grpc::EncodeTensorToByteBuffer. This representation
+     * can be used for all tensor types. The purpose of this representation is to
+     * reduce serialization overhead during RPC call by avoiding serialization of
+     * many repeated small items.
+     * </pre>
+     *
+     * <code>bytes tensor_content = 4;</code>
+     */
+    public Builder clearTensorContent() {
+      
+      tensorContent_ = getDefaultInstance().getTensorContent();
+      onChanged();
+      return this;
+    }
+
+    private com.google.protobuf.Internal.IntList halfVal_ = emptyIntList();
+    private void ensureHalfValIsMutable() {
+      if (!((bitField0_ & 0x00000001) != 0)) {
+        halfVal_ = mutableCopy(halfVal_);
+        bitField0_ |= 0x00000001;
+       }
+    }
+    /**
+     * <pre>
+     * DT_HALF, DT_BFLOAT16. Note that since protobuf has no int16 type, we'll
+     * have some pointless zero padding for each value here.
+     * </pre>
+     *
+     * <code>repeated int32 half_val = 13 [packed = true];</code>
+     */
+    public java.util.List<java.lang.Integer>
+        getHalfValList() {
+      return ((bitField0_ & 0x00000001) != 0) ?
+               java.util.Collections.unmodifiableList(halfVal_) : halfVal_;
+    }
+    /**
+     * <pre>
+     * DT_HALF, DT_BFLOAT16. Note that since protobuf has no int16 type, we'll
+     * have some pointless zero padding for each value here.
+     * </pre>
+     *
+     * <code>repeated int32 half_val = 13 [packed = true];</code>
+     */
+    public int getHalfValCount() {
+      return halfVal_.size();
+    }
+    /**
+     * <pre>
+     * DT_HALF, DT_BFLOAT16. Note that since protobuf has no int16 type, we'll
+     * have some pointless zero padding for each value here.
+     * </pre>
+     *
+     * <code>repeated int32 half_val = 13 [packed = true];</code>
+     */
+    public int getHalfVal(int index) {
+      return halfVal_.getInt(index);
+    }
+    /**
+     * <pre>
+     * DT_HALF, DT_BFLOAT16. Note that since protobuf has no int16 type, we'll
+     * have some pointless zero padding for each value here.
+     * </pre>
+     *
+     * <code>repeated int32 half_val = 13 [packed = true];</code>
+     */
+    public Builder setHalfVal(
+        int index, int value) {
+      ensureHalfValIsMutable();
+      halfVal_.setInt(index, value);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * DT_HALF, DT_BFLOAT16. Note that since protobuf has no int16 type, we'll
+     * have some pointless zero padding for each value here.
+     * </pre>
+     *
+     * <code>repeated int32 half_val = 13 [packed = true];</code>
+     */
+    public Builder addHalfVal(int value) {
+      ensureHalfValIsMutable();
+      halfVal_.addInt(value);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * DT_HALF, DT_BFLOAT16. Note that since protobuf has no int16 type, we'll
+     * have some pointless zero padding for each value here.
+     * </pre>
+     *
+     * <code>repeated int32 half_val = 13 [packed = true];</code>
+     */
+    public Builder addAllHalfVal(
+        java.lang.Iterable<? extends java.lang.Integer> values) {
+      ensureHalfValIsMutable();
+      com.google.protobuf.AbstractMessageLite.Builder.addAll(
+          values, halfVal_);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * DT_HALF, DT_BFLOAT16. Note that since protobuf has no int16 type, we'll
+     * have some pointless zero padding for each value here.
+     * </pre>
+     *
+     * <code>repeated int32 half_val = 13 [packed = true];</code>
+     */
+    public Builder clearHalfVal() {
+      halfVal_ = emptyIntList();
+      bitField0_ = (bitField0_ & ~0x00000001);
+      onChanged();
+      return this;
+    }
+
+    private com.google.protobuf.Internal.FloatList floatVal_ = emptyFloatList();
+    private void ensureFloatValIsMutable() {
+      if (!((bitField0_ & 0x00000002) != 0)) {
+        floatVal_ = mutableCopy(floatVal_);
+        bitField0_ |= 0x00000002;
+       }
+    }
+    /**
+     * <pre>
+     * DT_FLOAT.
+     * </pre>
+     *
+     * <code>repeated float float_val = 5 [packed = true];</code>
+     */
+    public java.util.List<java.lang.Float>
+        getFloatValList() {
+      return ((bitField0_ & 0x00000002) != 0) ?
+               java.util.Collections.unmodifiableList(floatVal_) : floatVal_;
+    }
+    /**
+     * <pre>
+     * DT_FLOAT.
+     * </pre>
+     *
+     * <code>repeated float float_val = 5 [packed = true];</code>
+     */
+    public int getFloatValCount() {
+      return floatVal_.size();
+    }
+    /**
+     * <pre>
+     * DT_FLOAT.
+     * </pre>
+     *
+     * <code>repeated float float_val = 5 [packed = true];</code>
+     */
+    public float getFloatVal(int index) {
+      return floatVal_.getFloat(index);
+    }
+    /**
+     * <pre>
+     * DT_FLOAT.
+     * </pre>
+     *
+     * <code>repeated float float_val = 5 [packed = true];</code>
+     */
+    public Builder setFloatVal(
+        int index, float value) {
+      ensureFloatValIsMutable();
+      floatVal_.setFloat(index, value);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * DT_FLOAT.
+     * </pre>
+     *
+     * <code>repeated float float_val = 5 [packed = true];</code>
+     */
+    public Builder addFloatVal(float value) {
+      ensureFloatValIsMutable();
+      floatVal_.addFloat(value);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * DT_FLOAT.
+     * </pre>
+     *
+     * <code>repeated float float_val = 5 [packed = true];</code>
+     */
+    public Builder addAllFloatVal(
+        java.lang.Iterable<? extends java.lang.Float> values) {
+      ensureFloatValIsMutable();
+      com.google.protobuf.AbstractMessageLite.Builder.addAll(
+          values, floatVal_);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * DT_FLOAT.
+     * </pre>
+     *
+     * <code>repeated float float_val = 5 [packed = true];</code>
+     */
+    public Builder clearFloatVal() {
+      floatVal_ = emptyFloatList();
+      bitField0_ = (bitField0_ & ~0x00000002);
+      onChanged();
+      return this;
+    }
+
+    private com.google.protobuf.Internal.DoubleList doubleVal_ = emptyDoubleList();
+    private void ensureDoubleValIsMutable() {
+      if (!((bitField0_ & 0x00000004) != 0)) {
+        doubleVal_ = mutableCopy(doubleVal_);
+        bitField0_ |= 0x00000004;
+       }
+    }
+    /**
+     * <pre>
+     * DT_DOUBLE.
+     * </pre>
+     *
+     * <code>repeated double double_val = 6 [packed = true];</code>
+     */
+    public java.util.List<java.lang.Double>
+        getDoubleValList() {
+      return ((bitField0_ & 0x00000004) != 0) ?
+               java.util.Collections.unmodifiableList(doubleVal_) : doubleVal_;
+    }
+    /**
+     * <pre>
+     * DT_DOUBLE.
+     * </pre>
+     *
+     * <code>repeated double double_val = 6 [packed = true];</code>
+     */
+    public int getDoubleValCount() {
+      return doubleVal_.size();
+    }
+    /**
+     * <pre>
+     * DT_DOUBLE.
+     * </pre>
+     *
+     * <code>repeated double double_val = 6 [packed = true];</code>
+     */
+    public double getDoubleVal(int index) {
+      return doubleVal_.getDouble(index);
+    }
+    /**
+     * <pre>
+     * DT_DOUBLE.
+     * </pre>
+     *
+     * <code>repeated double double_val = 6 [packed = true];</code>
+     */
+    public Builder setDoubleVal(
+        int index, double value) {
+      ensureDoubleValIsMutable();
+      doubleVal_.setDouble(index, value);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * DT_DOUBLE.
+     * </pre>
+     *
+     * <code>repeated double double_val = 6 [packed = true];</code>
+     */
+    public Builder addDoubleVal(double value) {
+      ensureDoubleValIsMutable();
+      doubleVal_.addDouble(value);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * DT_DOUBLE.
+     * </pre>
+     *
+     * <code>repeated double double_val = 6 [packed = true];</code>
+     */
+    public Builder addAllDoubleVal(
+        java.lang.Iterable<? extends java.lang.Double> values) {
+      ensureDoubleValIsMutable();
+      com.google.protobuf.AbstractMessageLite.Builder.addAll(
+          values, doubleVal_);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * DT_DOUBLE.
+     * </pre>
+     *
+     * <code>repeated double double_val = 6 [packed = true];</code>
+     */
+    public Builder clearDoubleVal() {
+      doubleVal_ = emptyDoubleList();
+      bitField0_ = (bitField0_ & ~0x00000004);
+      onChanged();
+      return this;
+    }
+
+    private com.google.protobuf.Internal.IntList intVal_ = emptyIntList();
+    private void ensureIntValIsMutable() {
+      if (!((bitField0_ & 0x00000008) != 0)) {
+        intVal_ = mutableCopy(intVal_);
+        bitField0_ |= 0x00000008;
+       }
+    }
+    /**
+     * <pre>
+     * DT_INT32, DT_INT16, DT_INT8, DT_UINT8.
+     * </pre>
+     *
+     * <code>repeated int32 int_val = 7 [packed = true];</code>
+     */
+    public java.util.List<java.lang.Integer>
+        getIntValList() {
+      return ((bitField0_ & 0x00000008) != 0) ?
+               java.util.Collections.unmodifiableList(intVal_) : intVal_;
+    }
+    /**
+     * <pre>
+     * DT_INT32, DT_INT16, DT_INT8, DT_UINT8.
+     * </pre>
+     *
+     * <code>repeated int32 int_val = 7 [packed = true];</code>
+     */
+    public int getIntValCount() {
+      return intVal_.size();
+    }
+    /**
+     * <pre>
+     * DT_INT32, DT_INT16, DT_INT8, DT_UINT8.
+     * </pre>
+     *
+     * <code>repeated int32 int_val = 7 [packed = true];</code>
+     */
+    public int getIntVal(int index) {
+      return intVal_.getInt(index);
+    }
+    /**
+     * <pre>
+     * DT_INT32, DT_INT16, DT_INT8, DT_UINT8.
+     * </pre>
+     *
+     * <code>repeated int32 int_val = 7 [packed = true];</code>
+     */
+    public Builder setIntVal(
+        int index, int value) {
+      ensureIntValIsMutable();
+      intVal_.setInt(index, value);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * DT_INT32, DT_INT16, DT_INT8, DT_UINT8.
+     * </pre>
+     *
+     * <code>repeated int32 int_val = 7 [packed = true];</code>
+     */
+    public Builder addIntVal(int value) {
+      ensureIntValIsMutable();
+      intVal_.addInt(value);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * DT_INT32, DT_INT16, DT_INT8, DT_UINT8.
+     * </pre>
+     *
+     * <code>repeated int32 int_val = 7 [packed = true];</code>
+     */
+    public Builder addAllIntVal(
+        java.lang.Iterable<? extends java.lang.Integer> values) {
+      ensureIntValIsMutable();
+      com.google.protobuf.AbstractMessageLite.Builder.addAll(
+          values, intVal_);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * DT_INT32, DT_INT16, DT_INT8, DT_UINT8.
+     * </pre>
+     *
+     * <code>repeated int32 int_val = 7 [packed = true];</code>
+     */
+    public Builder clearIntVal() {
+      intVal_ = emptyIntList();
+      bitField0_ = (bitField0_ & ~0x00000008);
+      onChanged();
+      return this;
+    }
+
+    private java.util.List<com.google.protobuf.ByteString> stringVal_ = java.util.Collections.emptyList();
+    private void ensureStringValIsMutable() {
+      if (!((bitField0_ & 0x00000010) != 0)) {
+        stringVal_ = new java.util.ArrayList<com.google.protobuf.ByteString>(stringVal_);
+        bitField0_ |= 0x00000010;
+       }
+    }
+    /**
+     * <pre>
+     * DT_STRING
+     * </pre>
+     *
+     * <code>repeated bytes string_val = 8;</code>
+     */
+    public java.util.List<com.google.protobuf.ByteString>
+        getStringValList() {
+      return ((bitField0_ & 0x00000010) != 0) ?
+               java.util.Collections.unmodifiableList(stringVal_) : stringVal_;
+    }
+    /**
+     * <pre>
+     * DT_STRING
+     * </pre>
+     *
+     * <code>repeated bytes string_val = 8;</code>
+     */
+    public int getStringValCount() {
+      return stringVal_.size();
+    }
+    /**
+     * <pre>
+     * DT_STRING
+     * </pre>
+     *
+     * <code>repeated bytes string_val = 8;</code>
+     */
+    public com.google.protobuf.ByteString getStringVal(int index) {
+      return stringVal_.get(index);
+    }
+    /**
+     * <pre>
+     * DT_STRING
+     * </pre>
+     *
+     * <code>repeated bytes string_val = 8;</code>
+     */
+    public Builder setStringVal(
+        int index, com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  ensureStringValIsMutable();
+      stringVal_.set(index, value);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * DT_STRING
+     * </pre>
+     *
+     * <code>repeated bytes string_val = 8;</code>
+     */
+    public Builder addStringVal(com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  ensureStringValIsMutable();
+      stringVal_.add(value);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * DT_STRING
+     * </pre>
+     *
+     * <code>repeated bytes string_val = 8;</code>
+     */
+    public Builder addAllStringVal(
+        java.lang.Iterable<? extends com.google.protobuf.ByteString> values) {
+      ensureStringValIsMutable();
+      com.google.protobuf.AbstractMessageLite.Builder.addAll(
+          values, stringVal_);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * DT_STRING
+     * </pre>
+     *
+     * <code>repeated bytes string_val = 8;</code>
+     */
+    public Builder clearStringVal() {
+      stringVal_ = java.util.Collections.emptyList();
+      bitField0_ = (bitField0_ & ~0x00000010);
+      onChanged();
+      return this;
+    }
+
+    private com.google.protobuf.Internal.FloatList scomplexVal_ = emptyFloatList();
+    private void ensureScomplexValIsMutable() {
+      if (!((bitField0_ & 0x00000020) != 0)) {
+        scomplexVal_ = mutableCopy(scomplexVal_);
+        bitField0_ |= 0x00000020;
+       }
+    }
+    /**
+     * <pre>
+     * DT_COMPLEX64. scomplex_val(2*i) and scomplex_val(2*i+1) are real
+     * and imaginary parts of i-th single precision complex.
+     * </pre>
+     *
+     * <code>repeated float scomplex_val = 9 [packed = true];</code>
+     */
+    public java.util.List<java.lang.Float>
+        getScomplexValList() {
+      return ((bitField0_ & 0x00000020) != 0) ?
+               java.util.Collections.unmodifiableList(scomplexVal_) : scomplexVal_;
+    }
+    /**
+     * <pre>
+     * DT_COMPLEX64. scomplex_val(2*i) and scomplex_val(2*i+1) are real
+     * and imaginary parts of i-th single precision complex.
+     * </pre>
+     *
+     * <code>repeated float scomplex_val = 9 [packed = true];</code>
+     */
+    public int getScomplexValCount() {
+      return scomplexVal_.size();
+    }
+    /**
+     * <pre>
+     * DT_COMPLEX64. scomplex_val(2*i) and scomplex_val(2*i+1) are real
+     * and imaginary parts of i-th single precision complex.
+     * </pre>
+     *
+     * <code>repeated float scomplex_val = 9 [packed = true];</code>
+     */
+    public float getScomplexVal(int index) {
+      return scomplexVal_.getFloat(index);
+    }
+    /**
+     * <pre>
+     * DT_COMPLEX64. scomplex_val(2*i) and scomplex_val(2*i+1) are real
+     * and imaginary parts of i-th single precision complex.
+     * </pre>
+     *
+     * <code>repeated float scomplex_val = 9 [packed = true];</code>
+     */
+    public Builder setScomplexVal(
+        int index, float value) {
+      ensureScomplexValIsMutable();
+      scomplexVal_.setFloat(index, value);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * DT_COMPLEX64. scomplex_val(2*i) and scomplex_val(2*i+1) are real
+     * and imaginary parts of i-th single precision complex.
+     * </pre>
+     *
+     * <code>repeated float scomplex_val = 9 [packed = true];</code>
+     */
+    public Builder addScomplexVal(float value) {
+      ensureScomplexValIsMutable();
+      scomplexVal_.addFloat(value);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * DT_COMPLEX64. scomplex_val(2*i) and scomplex_val(2*i+1) are real
+     * and imaginary parts of i-th single precision complex.
+     * </pre>
+     *
+     * <code>repeated float scomplex_val = 9 [packed = true];</code>
+     */
+    public Builder addAllScomplexVal(
+        java.lang.Iterable<? extends java.lang.Float> values) {
+      ensureScomplexValIsMutable();
+      com.google.protobuf.AbstractMessageLite.Builder.addAll(
+          values, scomplexVal_);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * DT_COMPLEX64. scomplex_val(2*i) and scomplex_val(2*i+1) are real
+     * and imaginary parts of i-th single precision complex.
+     * </pre>
+     *
+     * <code>repeated float scomplex_val = 9 [packed = true];</code>
+     */
+    public Builder clearScomplexVal() {
+      scomplexVal_ = emptyFloatList();
+      bitField0_ = (bitField0_ & ~0x00000020);
+      onChanged();
+      return this;
+    }
+
+    private com.google.protobuf.Internal.LongList int64Val_ = emptyLongList();
+    private void ensureInt64ValIsMutable() {
+      if (!((bitField0_ & 0x00000040) != 0)) {
+        int64Val_ = mutableCopy(int64Val_);
+        bitField0_ |= 0x00000040;
+       }
+    }
+    /**
+     * <pre>
+     * DT_INT64
+     * </pre>
+     *
+     * <code>repeated int64 int64_val = 10 [packed = true];</code>
+     */
+    public java.util.List<java.lang.Long>
+        getInt64ValList() {
+      return ((bitField0_ & 0x00000040) != 0) ?
+               java.util.Collections.unmodifiableList(int64Val_) : int64Val_;
+    }
+    /**
+     * <pre>
+     * DT_INT64
+     * </pre>
+     *
+     * <code>repeated int64 int64_val = 10 [packed = true];</code>
+     */
+    public int getInt64ValCount() {
+      return int64Val_.size();
+    }
+    /**
+     * <pre>
+     * DT_INT64
+     * </pre>
+     *
+     * <code>repeated int64 int64_val = 10 [packed = true];</code>
+     */
+    public long getInt64Val(int index) {
+      return int64Val_.getLong(index);
+    }
+    /**
+     * <pre>
+     * DT_INT64
+     * </pre>
+     *
+     * <code>repeated int64 int64_val = 10 [packed = true];</code>
+     */
+    public Builder setInt64Val(
+        int index, long value) {
+      ensureInt64ValIsMutable();
+      int64Val_.setLong(index, value);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * DT_INT64
+     * </pre>
+     *
+     * <code>repeated int64 int64_val = 10 [packed = true];</code>
+     */
+    public Builder addInt64Val(long value) {
+      ensureInt64ValIsMutable();
+      int64Val_.addLong(value);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * DT_INT64
+     * </pre>
+     *
+     * <code>repeated int64 int64_val = 10 [packed = true];</code>
+     */
+    public Builder addAllInt64Val(
+        java.lang.Iterable<? extends java.lang.Long> values) {
+      ensureInt64ValIsMutable();
+      com.google.protobuf.AbstractMessageLite.Builder.addAll(
+          values, int64Val_);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * DT_INT64
+     * </pre>
+     *
+     * <code>repeated int64 int64_val = 10 [packed = true];</code>
+     */
+    public Builder clearInt64Val() {
+      int64Val_ = emptyLongList();
+      bitField0_ = (bitField0_ & ~0x00000040);
+      onChanged();
+      return this;
+    }
+
+    private com.google.protobuf.Internal.BooleanList boolVal_ = emptyBooleanList();
+    private void ensureBoolValIsMutable() {
+      if (!((bitField0_ & 0x00000080) != 0)) {
+        boolVal_ = mutableCopy(boolVal_);
+        bitField0_ |= 0x00000080;
+       }
+    }
+    /**
+     * <pre>
+     * DT_BOOL
+     * </pre>
+     *
+     * <code>repeated bool bool_val = 11 [packed = true];</code>
+     */
+    public java.util.List<java.lang.Boolean>
+        getBoolValList() {
+      return ((bitField0_ & 0x00000080) != 0) ?
+               java.util.Collections.unmodifiableList(boolVal_) : boolVal_;
+    }
+    /**
+     * <pre>
+     * DT_BOOL
+     * </pre>
+     *
+     * <code>repeated bool bool_val = 11 [packed = true];</code>
+     */
+    public int getBoolValCount() {
+      return boolVal_.size();
+    }
+    /**
+     * <pre>
+     * DT_BOOL
+     * </pre>
+     *
+     * <code>repeated bool bool_val = 11 [packed = true];</code>
+     */
+    public boolean getBoolVal(int index) {
+      return boolVal_.getBoolean(index);
+    }
+    /**
+     * <pre>
+     * DT_BOOL
+     * </pre>
+     *
+     * <code>repeated bool bool_val = 11 [packed = true];</code>
+     */
+    public Builder setBoolVal(
+        int index, boolean value) {
+      ensureBoolValIsMutable();
+      boolVal_.setBoolean(index, value);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * DT_BOOL
+     * </pre>
+     *
+     * <code>repeated bool bool_val = 11 [packed = true];</code>
+     */
+    public Builder addBoolVal(boolean value) {
+      ensureBoolValIsMutable();
+      boolVal_.addBoolean(value);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * DT_BOOL
+     * </pre>
+     *
+     * <code>repeated bool bool_val = 11 [packed = true];</code>
+     */
+    public Builder addAllBoolVal(
+        java.lang.Iterable<? extends java.lang.Boolean> values) {
+      ensureBoolValIsMutable();
+      com.google.protobuf.AbstractMessageLite.Builder.addAll(
+          values, boolVal_);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * DT_BOOL
+     * </pre>
+     *
+     * <code>repeated bool bool_val = 11 [packed = true];</code>
+     */
+    public Builder clearBoolVal() {
+      boolVal_ = emptyBooleanList();
+      bitField0_ = (bitField0_ & ~0x00000080);
+      onChanged();
+      return this;
+    }
+
+    private com.google.protobuf.Internal.DoubleList dcomplexVal_ = emptyDoubleList();
+    private void ensureDcomplexValIsMutable() {
+      if (!((bitField0_ & 0x00000100) != 0)) {
+        dcomplexVal_ = mutableCopy(dcomplexVal_);
+        bitField0_ |= 0x00000100;
+       }
+    }
+    /**
+     * <pre>
+     * DT_COMPLEX128. dcomplex_val(2*i) and dcomplex_val(2*i+1) are real
+     * and imaginary parts of i-th double precision complex.
+     * </pre>
+     *
+     * <code>repeated double dcomplex_val = 12 [packed = true];</code>
+     */
+    public java.util.List<java.lang.Double>
+        getDcomplexValList() {
+      return ((bitField0_ & 0x00000100) != 0) ?
+               java.util.Collections.unmodifiableList(dcomplexVal_) : dcomplexVal_;
+    }
+    /**
+     * <pre>
+     * DT_COMPLEX128. dcomplex_val(2*i) and dcomplex_val(2*i+1) are real
+     * and imaginary parts of i-th double precision complex.
+     * </pre>
+     *
+     * <code>repeated double dcomplex_val = 12 [packed = true];</code>
+     */
+    public int getDcomplexValCount() {
+      return dcomplexVal_.size();
+    }
+    /**
+     * <pre>
+     * DT_COMPLEX128. dcomplex_val(2*i) and dcomplex_val(2*i+1) are real
+     * and imaginary parts of i-th double precision complex.
+     * </pre>
+     *
+     * <code>repeated double dcomplex_val = 12 [packed = true];</code>
+     */
+    public double getDcomplexVal(int index) {
+      return dcomplexVal_.getDouble(index);
+    }
+    /**
+     * <pre>
+     * DT_COMPLEX128. dcomplex_val(2*i) and dcomplex_val(2*i+1) are real
+     * and imaginary parts of i-th double precision complex.
+     * </pre>
+     *
+     * <code>repeated double dcomplex_val = 12 [packed = true];</code>
+     */
+    public Builder setDcomplexVal(
+        int index, double value) {
+      ensureDcomplexValIsMutable();
+      dcomplexVal_.setDouble(index, value);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * DT_COMPLEX128. dcomplex_val(2*i) and dcomplex_val(2*i+1) are real
+     * and imaginary parts of i-th double precision complex.
+     * </pre>
+     *
+     * <code>repeated double dcomplex_val = 12 [packed = true];</code>
+     */
+    public Builder addDcomplexVal(double value) {
+      ensureDcomplexValIsMutable();
+      dcomplexVal_.addDouble(value);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * DT_COMPLEX128. dcomplex_val(2*i) and dcomplex_val(2*i+1) are real
+     * and imaginary parts of i-th double precision complex.
+     * </pre>
+     *
+     * <code>repeated double dcomplex_val = 12 [packed = true];</code>
+     */
+    public Builder addAllDcomplexVal(
+        java.lang.Iterable<? extends java.lang.Double> values) {
+      ensureDcomplexValIsMutable();
+      com.google.protobuf.AbstractMessageLite.Builder.addAll(
+          values, dcomplexVal_);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * DT_COMPLEX128. dcomplex_val(2*i) and dcomplex_val(2*i+1) are real
+     * and imaginary parts of i-th double precision complex.
+     * </pre>
+     *
+     * <code>repeated double dcomplex_val = 12 [packed = true];</code>
+     */
+    public Builder clearDcomplexVal() {
+      dcomplexVal_ = emptyDoubleList();
+      bitField0_ = (bitField0_ & ~0x00000100);
+      onChanged();
+      return this;
+    }
+
+    private java.util.List<org.tensorflow.proto.framework.ResourceHandleProto> resourceHandleVal_ =
+      java.util.Collections.emptyList();
+    private void ensureResourceHandleValIsMutable() {
+      if (!((bitField0_ & 0x00000200) != 0)) {
+        resourceHandleVal_ = new java.util.ArrayList<org.tensorflow.proto.framework.ResourceHandleProto>(resourceHandleVal_);
+        bitField0_ |= 0x00000200;
+       }
+    }
+
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.framework.ResourceHandleProto, org.tensorflow.proto.framework.ResourceHandleProto.Builder, org.tensorflow.proto.framework.ResourceHandleProtoOrBuilder> resourceHandleValBuilder_;
+
+    /**
+     * <pre>
+     * DT_RESOURCE
+     * </pre>
+     *
+     * <code>repeated .tensorflow.ResourceHandleProto resource_handle_val = 14;</code>
+     */
+    public java.util.List<org.tensorflow.proto.framework.ResourceHandleProto> getResourceHandleValList() {
+      if (resourceHandleValBuilder_ == null) {
+        return java.util.Collections.unmodifiableList(resourceHandleVal_);
+      } else {
+        return resourceHandleValBuilder_.getMessageList();
+      }
+    }
+    /**
+     * <pre>
+     * DT_RESOURCE
+     * </pre>
+     *
+     * <code>repeated .tensorflow.ResourceHandleProto resource_handle_val = 14;</code>
+     */
+    public int getResourceHandleValCount() {
+      if (resourceHandleValBuilder_ == null) {
+        return resourceHandleVal_.size();
+      } else {
+        return resourceHandleValBuilder_.getCount();
+      }
+    }
+    /**
+     * <pre>
+     * DT_RESOURCE
+     * </pre>
+     *
+     * <code>repeated .tensorflow.ResourceHandleProto resource_handle_val = 14;</code>
+     */
+    public org.tensorflow.proto.framework.ResourceHandleProto getResourceHandleVal(int index) {
+      if (resourceHandleValBuilder_ == null) {
+        return resourceHandleVal_.get(index);
+      } else {
+        return resourceHandleValBuilder_.getMessage(index);
+      }
+    }
+    /**
+     * <pre>
+     * DT_RESOURCE
+     * </pre>
+     *
+     * <code>repeated .tensorflow.ResourceHandleProto resource_handle_val = 14;</code>
+     */
+    public Builder setResourceHandleVal(
+        int index, org.tensorflow.proto.framework.ResourceHandleProto value) {
+      if (resourceHandleValBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureResourceHandleValIsMutable();
+        resourceHandleVal_.set(index, value);
+        onChanged();
+      } else {
+        resourceHandleValBuilder_.setMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * DT_RESOURCE
+     * </pre>
+     *
+     * <code>repeated .tensorflow.ResourceHandleProto resource_handle_val = 14;</code>
+     */
+    public Builder setResourceHandleVal(
+        int index, org.tensorflow.proto.framework.ResourceHandleProto.Builder builderForValue) {
+      if (resourceHandleValBuilder_ == null) {
+        ensureResourceHandleValIsMutable();
+        resourceHandleVal_.set(index, builderForValue.build());
+        onChanged();
+      } else {
+        resourceHandleValBuilder_.setMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * DT_RESOURCE
+     * </pre>
+     *
+     * <code>repeated .tensorflow.ResourceHandleProto resource_handle_val = 14;</code>
+     */
+    public Builder addResourceHandleVal(org.tensorflow.proto.framework.ResourceHandleProto value) {
+      if (resourceHandleValBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureResourceHandleValIsMutable();
+        resourceHandleVal_.add(value);
+        onChanged();
+      } else {
+        resourceHandleValBuilder_.addMessage(value);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * DT_RESOURCE
+     * </pre>
+     *
+     * <code>repeated .tensorflow.ResourceHandleProto resource_handle_val = 14;</code>
+     */
+    public Builder addResourceHandleVal(
+        int index, org.tensorflow.proto.framework.ResourceHandleProto value) {
+      if (resourceHandleValBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureResourceHandleValIsMutable();
+        resourceHandleVal_.add(index, value);
+        onChanged();
+      } else {
+        resourceHandleValBuilder_.addMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * DT_RESOURCE
+     * </pre>
+     *
+     * <code>repeated .tensorflow.ResourceHandleProto resource_handle_val = 14;</code>
+     */
+    public Builder addResourceHandleVal(
+        org.tensorflow.proto.framework.ResourceHandleProto.Builder builderForValue) {
+      if (resourceHandleValBuilder_ == null) {
+        ensureResourceHandleValIsMutable();
+        resourceHandleVal_.add(builderForValue.build());
+        onChanged();
+      } else {
+        resourceHandleValBuilder_.addMessage(builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * DT_RESOURCE
+     * </pre>
+     *
+     * <code>repeated .tensorflow.ResourceHandleProto resource_handle_val = 14;</code>
+     */
+    public Builder addResourceHandleVal(
+        int index, org.tensorflow.proto.framework.ResourceHandleProto.Builder builderForValue) {
+      if (resourceHandleValBuilder_ == null) {
+        ensureResourceHandleValIsMutable();
+        resourceHandleVal_.add(index, builderForValue.build());
+        onChanged();
+      } else {
+        resourceHandleValBuilder_.addMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * DT_RESOURCE
+     * </pre>
+     *
+     * <code>repeated .tensorflow.ResourceHandleProto resource_handle_val = 14;</code>
+     */
+    public Builder addAllResourceHandleVal(
+        java.lang.Iterable<? extends org.tensorflow.proto.framework.ResourceHandleProto> values) {
+      if (resourceHandleValBuilder_ == null) {
+        ensureResourceHandleValIsMutable();
+        com.google.protobuf.AbstractMessageLite.Builder.addAll(
+            values, resourceHandleVal_);
+        onChanged();
+      } else {
+        resourceHandleValBuilder_.addAllMessages(values);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * DT_RESOURCE
+     * </pre>
+     *
+     * <code>repeated .tensorflow.ResourceHandleProto resource_handle_val = 14;</code>
+     */
+    public Builder clearResourceHandleVal() {
+      if (resourceHandleValBuilder_ == null) {
+        resourceHandleVal_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000200);
+        onChanged();
+      } else {
+        resourceHandleValBuilder_.clear();
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * DT_RESOURCE
+     * </pre>
+     *
+     * <code>repeated .tensorflow.ResourceHandleProto resource_handle_val = 14;</code>
+     */
+    public Builder removeResourceHandleVal(int index) {
+      if (resourceHandleValBuilder_ == null) {
+        ensureResourceHandleValIsMutable();
+        resourceHandleVal_.remove(index);
+        onChanged();
+      } else {
+        resourceHandleValBuilder_.remove(index);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * DT_RESOURCE
+     * </pre>
+     *
+     * <code>repeated .tensorflow.ResourceHandleProto resource_handle_val = 14;</code>
+     */
+    public org.tensorflow.proto.framework.ResourceHandleProto.Builder getResourceHandleValBuilder(
+        int index) {
+      return getResourceHandleValFieldBuilder().getBuilder(index);
+    }
+    /**
+     * <pre>
+     * DT_RESOURCE
+     * </pre>
+     *
+     * <code>repeated .tensorflow.ResourceHandleProto resource_handle_val = 14;</code>
+     */
+    public org.tensorflow.proto.framework.ResourceHandleProtoOrBuilder getResourceHandleValOrBuilder(
+        int index) {
+      if (resourceHandleValBuilder_ == null) {
+        return resourceHandleVal_.get(index);  } else {
+        return resourceHandleValBuilder_.getMessageOrBuilder(index);
+      }
+    }
+    /**
+     * <pre>
+     * DT_RESOURCE
+     * </pre>
+     *
+     * <code>repeated .tensorflow.ResourceHandleProto resource_handle_val = 14;</code>
+     */
+    public java.util.List<? extends org.tensorflow.proto.framework.ResourceHandleProtoOrBuilder> 
+         getResourceHandleValOrBuilderList() {
+      if (resourceHandleValBuilder_ != null) {
+        return resourceHandleValBuilder_.getMessageOrBuilderList();
+      } else {
+        return java.util.Collections.unmodifiableList(resourceHandleVal_);
+      }
+    }
+    /**
+     * <pre>
+     * DT_RESOURCE
+     * </pre>
+     *
+     * <code>repeated .tensorflow.ResourceHandleProto resource_handle_val = 14;</code>
+     */
+    public org.tensorflow.proto.framework.ResourceHandleProto.Builder addResourceHandleValBuilder() {
+      return getResourceHandleValFieldBuilder().addBuilder(
+          org.tensorflow.proto.framework.ResourceHandleProto.getDefaultInstance());
+    }
+    /**
+     * <pre>
+     * DT_RESOURCE
+     * </pre>
+     *
+     * <code>repeated .tensorflow.ResourceHandleProto resource_handle_val = 14;</code>
+     */
+    public org.tensorflow.proto.framework.ResourceHandleProto.Builder addResourceHandleValBuilder(
+        int index) {
+      return getResourceHandleValFieldBuilder().addBuilder(
+          index, org.tensorflow.proto.framework.ResourceHandleProto.getDefaultInstance());
+    }
+    /**
+     * <pre>
+     * DT_RESOURCE
+     * </pre>
+     *
+     * <code>repeated .tensorflow.ResourceHandleProto resource_handle_val = 14;</code>
+     */
+    public java.util.List<org.tensorflow.proto.framework.ResourceHandleProto.Builder> 
+         getResourceHandleValBuilderList() {
+      return getResourceHandleValFieldBuilder().getBuilderList();
+    }
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.framework.ResourceHandleProto, org.tensorflow.proto.framework.ResourceHandleProto.Builder, org.tensorflow.proto.framework.ResourceHandleProtoOrBuilder> 
+        getResourceHandleValFieldBuilder() {
+      if (resourceHandleValBuilder_ == null) {
+        resourceHandleValBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
+            org.tensorflow.proto.framework.ResourceHandleProto, org.tensorflow.proto.framework.ResourceHandleProto.Builder, org.tensorflow.proto.framework.ResourceHandleProtoOrBuilder>(
+                resourceHandleVal_,
+                ((bitField0_ & 0x00000200) != 0),
+                getParentForChildren(),
+                isClean());
+        resourceHandleVal_ = null;
+      }
+      return resourceHandleValBuilder_;
+    }
+
+    private java.util.List<org.tensorflow.proto.framework.VariantTensorDataProto> variantVal_ =
+      java.util.Collections.emptyList();
+    private void ensureVariantValIsMutable() {
+      if (!((bitField0_ & 0x00000400) != 0)) {
+        variantVal_ = new java.util.ArrayList<org.tensorflow.proto.framework.VariantTensorDataProto>(variantVal_);
+        bitField0_ |= 0x00000400;
+       }
+    }
+
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.framework.VariantTensorDataProto, org.tensorflow.proto.framework.VariantTensorDataProto.Builder, org.tensorflow.proto.framework.VariantTensorDataProtoOrBuilder> variantValBuilder_;
+
+    /**
+     * <pre>
+     * DT_VARIANT
+     * </pre>
+     *
+     * <code>repeated .tensorflow.VariantTensorDataProto variant_val = 15;</code>
+     */
+    public java.util.List<org.tensorflow.proto.framework.VariantTensorDataProto> getVariantValList() {
+      if (variantValBuilder_ == null) {
+        return java.util.Collections.unmodifiableList(variantVal_);
+      } else {
+        return variantValBuilder_.getMessageList();
+      }
+    }
+    /**
+     * <pre>
+     * DT_VARIANT
+     * </pre>
+     *
+     * <code>repeated .tensorflow.VariantTensorDataProto variant_val = 15;</code>
+     */
+    public int getVariantValCount() {
+      if (variantValBuilder_ == null) {
+        return variantVal_.size();
+      } else {
+        return variantValBuilder_.getCount();
+      }
+    }
+    /**
+     * <pre>
+     * DT_VARIANT
+     * </pre>
+     *
+     * <code>repeated .tensorflow.VariantTensorDataProto variant_val = 15;</code>
+     */
+    public org.tensorflow.proto.framework.VariantTensorDataProto getVariantVal(int index) {
+      if (variantValBuilder_ == null) {
+        return variantVal_.get(index);
+      } else {
+        return variantValBuilder_.getMessage(index);
+      }
+    }
+    /**
+     * <pre>
+     * DT_VARIANT
+     * </pre>
+     *
+     * <code>repeated .tensorflow.VariantTensorDataProto variant_val = 15;</code>
+     */
+    public Builder setVariantVal(
+        int index, org.tensorflow.proto.framework.VariantTensorDataProto value) {
+      if (variantValBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureVariantValIsMutable();
+        variantVal_.set(index, value);
+        onChanged();
+      } else {
+        variantValBuilder_.setMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * DT_VARIANT
+     * </pre>
+     *
+     * <code>repeated .tensorflow.VariantTensorDataProto variant_val = 15;</code>
+     */
+    public Builder setVariantVal(
+        int index, org.tensorflow.proto.framework.VariantTensorDataProto.Builder builderForValue) {
+      if (variantValBuilder_ == null) {
+        ensureVariantValIsMutable();
+        variantVal_.set(index, builderForValue.build());
+        onChanged();
+      } else {
+        variantValBuilder_.setMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * DT_VARIANT
+     * </pre>
+     *
+     * <code>repeated .tensorflow.VariantTensorDataProto variant_val = 15;</code>
+     */
+    public Builder addVariantVal(org.tensorflow.proto.framework.VariantTensorDataProto value) {
+      if (variantValBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureVariantValIsMutable();
+        variantVal_.add(value);
+        onChanged();
+      } else {
+        variantValBuilder_.addMessage(value);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * DT_VARIANT
+     * </pre>
+     *
+     * <code>repeated .tensorflow.VariantTensorDataProto variant_val = 15;</code>
+     */
+    public Builder addVariantVal(
+        int index, org.tensorflow.proto.framework.VariantTensorDataProto value) {
+      if (variantValBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureVariantValIsMutable();
+        variantVal_.add(index, value);
+        onChanged();
+      } else {
+        variantValBuilder_.addMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * DT_VARIANT
+     * </pre>
+     *
+     * <code>repeated .tensorflow.VariantTensorDataProto variant_val = 15;</code>
+     */
+    public Builder addVariantVal(
+        org.tensorflow.proto.framework.VariantTensorDataProto.Builder builderForValue) {
+      if (variantValBuilder_ == null) {
+        ensureVariantValIsMutable();
+        variantVal_.add(builderForValue.build());
+        onChanged();
+      } else {
+        variantValBuilder_.addMessage(builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * DT_VARIANT
+     * </pre>
+     *
+     * <code>repeated .tensorflow.VariantTensorDataProto variant_val = 15;</code>
+     */
+    public Builder addVariantVal(
+        int index, org.tensorflow.proto.framework.VariantTensorDataProto.Builder builderForValue) {
+      if (variantValBuilder_ == null) {
+        ensureVariantValIsMutable();
+        variantVal_.add(index, builderForValue.build());
+        onChanged();
+      } else {
+        variantValBuilder_.addMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * DT_VARIANT
+     * </pre>
+     *
+     * <code>repeated .tensorflow.VariantTensorDataProto variant_val = 15;</code>
+     */
+    public Builder addAllVariantVal(
+        java.lang.Iterable<? extends org.tensorflow.proto.framework.VariantTensorDataProto> values) {
+      if (variantValBuilder_ == null) {
+        ensureVariantValIsMutable();
+        com.google.protobuf.AbstractMessageLite.Builder.addAll(
+            values, variantVal_);
+        onChanged();
+      } else {
+        variantValBuilder_.addAllMessages(values);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * DT_VARIANT
+     * </pre>
+     *
+     * <code>repeated .tensorflow.VariantTensorDataProto variant_val = 15;</code>
+     */
+    public Builder clearVariantVal() {
+      if (variantValBuilder_ == null) {
+        variantVal_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000400);
+        onChanged();
+      } else {
+        variantValBuilder_.clear();
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * DT_VARIANT
+     * </pre>
+     *
+     * <code>repeated .tensorflow.VariantTensorDataProto variant_val = 15;</code>
+     */
+    public Builder removeVariantVal(int index) {
+      if (variantValBuilder_ == null) {
+        ensureVariantValIsMutable();
+        variantVal_.remove(index);
+        onChanged();
+      } else {
+        variantValBuilder_.remove(index);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * DT_VARIANT
+     * </pre>
+     *
+     * <code>repeated .tensorflow.VariantTensorDataProto variant_val = 15;</code>
+     */
+    public org.tensorflow.proto.framework.VariantTensorDataProto.Builder getVariantValBuilder(
+        int index) {
+      return getVariantValFieldBuilder().getBuilder(index);
+    }
+    /**
+     * <pre>
+     * DT_VARIANT
+     * </pre>
+     *
+     * <code>repeated .tensorflow.VariantTensorDataProto variant_val = 15;</code>
+     */
+    public org.tensorflow.proto.framework.VariantTensorDataProtoOrBuilder getVariantValOrBuilder(
+        int index) {
+      if (variantValBuilder_ == null) {
+        return variantVal_.get(index);  } else {
+        return variantValBuilder_.getMessageOrBuilder(index);
+      }
+    }
+    /**
+     * <pre>
+     * DT_VARIANT
+     * </pre>
+     *
+     * <code>repeated .tensorflow.VariantTensorDataProto variant_val = 15;</code>
+     */
+    public java.util.List<? extends org.tensorflow.proto.framework.VariantTensorDataProtoOrBuilder> 
+         getVariantValOrBuilderList() {
+      if (variantValBuilder_ != null) {
+        return variantValBuilder_.getMessageOrBuilderList();
+      } else {
+        return java.util.Collections.unmodifiableList(variantVal_);
+      }
+    }
+    /**
+     * <pre>
+     * DT_VARIANT
+     * </pre>
+     *
+     * <code>repeated .tensorflow.VariantTensorDataProto variant_val = 15;</code>
+     */
+    public org.tensorflow.proto.framework.VariantTensorDataProto.Builder addVariantValBuilder() {
+      return getVariantValFieldBuilder().addBuilder(
+          org.tensorflow.proto.framework.VariantTensorDataProto.getDefaultInstance());
+    }
+    /**
+     * <pre>
+     * DT_VARIANT
+     * </pre>
+     *
+     * <code>repeated .tensorflow.VariantTensorDataProto variant_val = 15;</code>
+     */
+    public org.tensorflow.proto.framework.VariantTensorDataProto.Builder addVariantValBuilder(
+        int index) {
+      return getVariantValFieldBuilder().addBuilder(
+          index, org.tensorflow.proto.framework.VariantTensorDataProto.getDefaultInstance());
+    }
+    /**
+     * <pre>
+     * DT_VARIANT
+     * </pre>
+     *
+     * <code>repeated .tensorflow.VariantTensorDataProto variant_val = 15;</code>
+     */
+    public java.util.List<org.tensorflow.proto.framework.VariantTensorDataProto.Builder> 
+         getVariantValBuilderList() {
+      return getVariantValFieldBuilder().getBuilderList();
+    }
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.framework.VariantTensorDataProto, org.tensorflow.proto.framework.VariantTensorDataProto.Builder, org.tensorflow.proto.framework.VariantTensorDataProtoOrBuilder> 
+        getVariantValFieldBuilder() {
+      if (variantValBuilder_ == null) {
+        variantValBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
+            org.tensorflow.proto.framework.VariantTensorDataProto, org.tensorflow.proto.framework.VariantTensorDataProto.Builder, org.tensorflow.proto.framework.VariantTensorDataProtoOrBuilder>(
+                variantVal_,
+                ((bitField0_ & 0x00000400) != 0),
+                getParentForChildren(),
+                isClean());
+        variantVal_ = null;
+      }
+      return variantValBuilder_;
+    }
+
+    private com.google.protobuf.Internal.IntList uint32Val_ = emptyIntList();
+    private void ensureUint32ValIsMutable() {
+      if (!((bitField0_ & 0x00000800) != 0)) {
+        uint32Val_ = mutableCopy(uint32Val_);
+        bitField0_ |= 0x00000800;
+       }
+    }
+    /**
+     * <pre>
+     * DT_UINT32
+     * </pre>
+     *
+     * <code>repeated uint32 uint32_val = 16 [packed = true];</code>
+     */
+    public java.util.List<java.lang.Integer>
+        getUint32ValList() {
+      return ((bitField0_ & 0x00000800) != 0) ?
+               java.util.Collections.unmodifiableList(uint32Val_) : uint32Val_;
+    }
+    /**
+     * <pre>
+     * DT_UINT32
+     * </pre>
+     *
+     * <code>repeated uint32 uint32_val = 16 [packed = true];</code>
+     */
+    public int getUint32ValCount() {
+      return uint32Val_.size();
+    }
+    /**
+     * <pre>
+     * DT_UINT32
+     * </pre>
+     *
+     * <code>repeated uint32 uint32_val = 16 [packed = true];</code>
+     */
+    public int getUint32Val(int index) {
+      return uint32Val_.getInt(index);
+    }
+    /**
+     * <pre>
+     * DT_UINT32
+     * </pre>
+     *
+     * <code>repeated uint32 uint32_val = 16 [packed = true];</code>
+     */
+    public Builder setUint32Val(
+        int index, int value) {
+      ensureUint32ValIsMutable();
+      uint32Val_.setInt(index, value);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * DT_UINT32
+     * </pre>
+     *
+     * <code>repeated uint32 uint32_val = 16 [packed = true];</code>
+     */
+    public Builder addUint32Val(int value) {
+      ensureUint32ValIsMutable();
+      uint32Val_.addInt(value);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * DT_UINT32
+     * </pre>
+     *
+     * <code>repeated uint32 uint32_val = 16 [packed = true];</code>
+     */
+    public Builder addAllUint32Val(
+        java.lang.Iterable<? extends java.lang.Integer> values) {
+      ensureUint32ValIsMutable();
+      com.google.protobuf.AbstractMessageLite.Builder.addAll(
+          values, uint32Val_);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * DT_UINT32
+     * </pre>
+     *
+     * <code>repeated uint32 uint32_val = 16 [packed = true];</code>
+     */
+    public Builder clearUint32Val() {
+      uint32Val_ = emptyIntList();
+      bitField0_ = (bitField0_ & ~0x00000800);
+      onChanged();
+      return this;
+    }
+
+    private com.google.protobuf.Internal.LongList uint64Val_ = emptyLongList();
+    private void ensureUint64ValIsMutable() {
+      if (!((bitField0_ & 0x00001000) != 0)) {
+        uint64Val_ = mutableCopy(uint64Val_);
+        bitField0_ |= 0x00001000;
+       }
+    }
+    /**
+     * <pre>
+     * DT_UINT64
+     * </pre>
+     *
+     * <code>repeated uint64 uint64_val = 17 [packed = true];</code>
+     */
+    public java.util.List<java.lang.Long>
+        getUint64ValList() {
+      return ((bitField0_ & 0x00001000) != 0) ?
+               java.util.Collections.unmodifiableList(uint64Val_) : uint64Val_;
+    }
+    /**
+     * <pre>
+     * DT_UINT64
+     * </pre>
+     *
+     * <code>repeated uint64 uint64_val = 17 [packed = true];</code>
+     */
+    public int getUint64ValCount() {
+      return uint64Val_.size();
+    }
+    /**
+     * <pre>
+     * DT_UINT64
+     * </pre>
+     *
+     * <code>repeated uint64 uint64_val = 17 [packed = true];</code>
+     */
+    public long getUint64Val(int index) {
+      return uint64Val_.getLong(index);
+    }
+    /**
+     * <pre>
+     * DT_UINT64
+     * </pre>
+     *
+     * <code>repeated uint64 uint64_val = 17 [packed = true];</code>
+     */
+    public Builder setUint64Val(
+        int index, long value) {
+      ensureUint64ValIsMutable();
+      uint64Val_.setLong(index, value);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * DT_UINT64
+     * </pre>
+     *
+     * <code>repeated uint64 uint64_val = 17 [packed = true];</code>
+     */
+    public Builder addUint64Val(long value) {
+      ensureUint64ValIsMutable();
+      uint64Val_.addLong(value);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * DT_UINT64
+     * </pre>
+     *
+     * <code>repeated uint64 uint64_val = 17 [packed = true];</code>
+     */
+    public Builder addAllUint64Val(
+        java.lang.Iterable<? extends java.lang.Long> values) {
+      ensureUint64ValIsMutable();
+      com.google.protobuf.AbstractMessageLite.Builder.addAll(
+          values, uint64Val_);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * DT_UINT64
+     * </pre>
+     *
+     * <code>repeated uint64 uint64_val = 17 [packed = true];</code>
+     */
+    public Builder clearUint64Val() {
+      uint64Val_ = emptyLongList();
+      bitField0_ = (bitField0_ & ~0x00001000);
+      onChanged();
+      return this;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.TensorProto)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.TensorProto)
+  private static final org.tensorflow.proto.framework.TensorProto DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.TensorProto();
+  }
+
+  public static org.tensorflow.proto.framework.TensorProto getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<TensorProto>
+      PARSER = new com.google.protobuf.AbstractParser<TensorProto>() {
+    @java.lang.Override
+    public TensorProto parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new TensorProto(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<TensorProto> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<TensorProto> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.framework.TensorProto getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorProtoOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorProtoOrBuilder.java
new file mode 100644
index 00000000000..b5369066023
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorProtoOrBuilder.java
@@ -0,0 +1,440 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/framework/tensor.proto
+
+package org.tensorflow.proto.framework;
+
+public interface TensorProtoOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.TensorProto)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <code>.tensorflow.DataType dtype = 1;</code>
+   */
+  int getDtypeValue();
+  /**
+   * <code>.tensorflow.DataType dtype = 1;</code>
+   */
+  org.tensorflow.proto.framework.DataType getDtype();
+
+  /**
+   * <pre>
+   * Shape of the tensor.  TODO(touts): sort out the 0-rank issues.
+   * </pre>
+   *
+   * <code>.tensorflow.TensorShapeProto tensor_shape = 2;</code>
+   */
+  boolean hasTensorShape();
+  /**
+   * <pre>
+   * Shape of the tensor.  TODO(touts): sort out the 0-rank issues.
+   * </pre>
+   *
+   * <code>.tensorflow.TensorShapeProto tensor_shape = 2;</code>
+   */
+  org.tensorflow.proto.framework.TensorShapeProto getTensorShape();
+  /**
+   * <pre>
+   * Shape of the tensor.  TODO(touts): sort out the 0-rank issues.
+   * </pre>
+   *
+   * <code>.tensorflow.TensorShapeProto tensor_shape = 2;</code>
+   */
+  org.tensorflow.proto.framework.TensorShapeProtoOrBuilder getTensorShapeOrBuilder();
+
+  /**
+   * <pre>
+   * Version number.
+   * In version 0, if the "repeated xxx" representations contain only one
+   * element, that element is repeated to fill the shape.  This makes it easy
+   * to represent a constant Tensor with a single value.
+   * </pre>
+   *
+   * <code>int32 version_number = 3;</code>
+   */
+  int getVersionNumber();
+
+  /**
+   * <pre>
+   * Serialized raw tensor content from either Tensor::AsProtoTensorContent or
+   * memcpy in tensorflow::grpc::EncodeTensorToByteBuffer. This representation
+   * can be used for all tensor types. The purpose of this representation is to
+   * reduce serialization overhead during RPC call by avoiding serialization of
+   * many repeated small items.
+   * </pre>
+   *
+   * <code>bytes tensor_content = 4;</code>
+   */
+  com.google.protobuf.ByteString getTensorContent();
+
+  /**
+   * <pre>
+   * DT_HALF, DT_BFLOAT16. Note that since protobuf has no int16 type, we'll
+   * have some pointless zero padding for each value here.
+   * </pre>
+   *
+   * <code>repeated int32 half_val = 13 [packed = true];</code>
+   */
+  java.util.List<java.lang.Integer> getHalfValList();
+  /**
+   * <pre>
+   * DT_HALF, DT_BFLOAT16. Note that since protobuf has no int16 type, we'll
+   * have some pointless zero padding for each value here.
+   * </pre>
+   *
+   * <code>repeated int32 half_val = 13 [packed = true];</code>
+   */
+  int getHalfValCount();
+  /**
+   * <pre>
+   * DT_HALF, DT_BFLOAT16. Note that since protobuf has no int16 type, we'll
+   * have some pointless zero padding for each value here.
+   * </pre>
+   *
+   * <code>repeated int32 half_val = 13 [packed = true];</code>
+   */
+  int getHalfVal(int index);
+
+  /**
+   * <pre>
+   * DT_FLOAT.
+   * </pre>
+   *
+   * <code>repeated float float_val = 5 [packed = true];</code>
+   */
+  java.util.List<java.lang.Float> getFloatValList();
+  /**
+   * <pre>
+   * DT_FLOAT.
+   * </pre>
+   *
+   * <code>repeated float float_val = 5 [packed = true];</code>
+   */
+  int getFloatValCount();
+  /**
+   * <pre>
+   * DT_FLOAT.
+   * </pre>
+   *
+   * <code>repeated float float_val = 5 [packed = true];</code>
+   */
+  float getFloatVal(int index);
+
+  /**
+   * <pre>
+   * DT_DOUBLE.
+   * </pre>
+   *
+   * <code>repeated double double_val = 6 [packed = true];</code>
+   */
+  java.util.List<java.lang.Double> getDoubleValList();
+  /**
+   * <pre>
+   * DT_DOUBLE.
+   * </pre>
+   *
+   * <code>repeated double double_val = 6 [packed = true];</code>
+   */
+  int getDoubleValCount();
+  /**
+   * <pre>
+   * DT_DOUBLE.
+   * </pre>
+   *
+   * <code>repeated double double_val = 6 [packed = true];</code>
+   */
+  double getDoubleVal(int index);
+
+  /**
+   * <pre>
+   * DT_INT32, DT_INT16, DT_INT8, DT_UINT8.
+   * </pre>
+   *
+   * <code>repeated int32 int_val = 7 [packed = true];</code>
+   */
+  java.util.List<java.lang.Integer> getIntValList();
+  /**
+   * <pre>
+   * DT_INT32, DT_INT16, DT_INT8, DT_UINT8.
+   * </pre>
+   *
+   * <code>repeated int32 int_val = 7 [packed = true];</code>
+   */
+  int getIntValCount();
+  /**
+   * <pre>
+   * DT_INT32, DT_INT16, DT_INT8, DT_UINT8.
+   * </pre>
+   *
+   * <code>repeated int32 int_val = 7 [packed = true];</code>
+   */
+  int getIntVal(int index);
+
+  /**
+   * <pre>
+   * DT_STRING
+   * </pre>
+   *
+   * <code>repeated bytes string_val = 8;</code>
+   */
+  java.util.List<com.google.protobuf.ByteString> getStringValList();
+  /**
+   * <pre>
+   * DT_STRING
+   * </pre>
+   *
+   * <code>repeated bytes string_val = 8;</code>
+   */
+  int getStringValCount();
+  /**
+   * <pre>
+   * DT_STRING
+   * </pre>
+   *
+   * <code>repeated bytes string_val = 8;</code>
+   */
+  com.google.protobuf.ByteString getStringVal(int index);
+
+  /**
+   * <pre>
+   * DT_COMPLEX64. scomplex_val(2*i) and scomplex_val(2*i+1) are real
+   * and imaginary parts of i-th single precision complex.
+   * </pre>
+   *
+   * <code>repeated float scomplex_val = 9 [packed = true];</code>
+   */
+  java.util.List<java.lang.Float> getScomplexValList();
+  /**
+   * <pre>
+   * DT_COMPLEX64. scomplex_val(2*i) and scomplex_val(2*i+1) are real
+   * and imaginary parts of i-th single precision complex.
+   * </pre>
+   *
+   * <code>repeated float scomplex_val = 9 [packed = true];</code>
+   */
+  int getScomplexValCount();
+  /**
+   * <pre>
+   * DT_COMPLEX64. scomplex_val(2*i) and scomplex_val(2*i+1) are real
+   * and imaginary parts of i-th single precision complex.
+   * </pre>
+   *
+   * <code>repeated float scomplex_val = 9 [packed = true];</code>
+   */
+  float getScomplexVal(int index);
+
+  /**
+   * <pre>
+   * DT_INT64
+   * </pre>
+   *
+   * <code>repeated int64 int64_val = 10 [packed = true];</code>
+   */
+  java.util.List<java.lang.Long> getInt64ValList();
+  /**
+   * <pre>
+   * DT_INT64
+   * </pre>
+   *
+   * <code>repeated int64 int64_val = 10 [packed = true];</code>
+   */
+  int getInt64ValCount();
+  /**
+   * <pre>
+   * DT_INT64
+   * </pre>
+   *
+   * <code>repeated int64 int64_val = 10 [packed = true];</code>
+   */
+  long getInt64Val(int index);
+
+  /**
+   * <pre>
+   * DT_BOOL
+   * </pre>
+   *
+   * <code>repeated bool bool_val = 11 [packed = true];</code>
+   */
+  java.util.List<java.lang.Boolean> getBoolValList();
+  /**
+   * <pre>
+   * DT_BOOL
+   * </pre>
+   *
+   * <code>repeated bool bool_val = 11 [packed = true];</code>
+   */
+  int getBoolValCount();
+  /**
+   * <pre>
+   * DT_BOOL
+   * </pre>
+   *
+   * <code>repeated bool bool_val = 11 [packed = true];</code>
+   */
+  boolean getBoolVal(int index);
+
+  /**
+   * <pre>
+   * DT_COMPLEX128. dcomplex_val(2*i) and dcomplex_val(2*i+1) are real
+   * and imaginary parts of i-th double precision complex.
+   * </pre>
+   *
+   * <code>repeated double dcomplex_val = 12 [packed = true];</code>
+   */
+  java.util.List<java.lang.Double> getDcomplexValList();
+  /**
+   * <pre>
+   * DT_COMPLEX128. dcomplex_val(2*i) and dcomplex_val(2*i+1) are real
+   * and imaginary parts of i-th double precision complex.
+   * </pre>
+   *
+   * <code>repeated double dcomplex_val = 12 [packed = true];</code>
+   */
+  int getDcomplexValCount();
+  /**
+   * <pre>
+   * DT_COMPLEX128. dcomplex_val(2*i) and dcomplex_val(2*i+1) are real
+   * and imaginary parts of i-th double precision complex.
+   * </pre>
+   *
+   * <code>repeated double dcomplex_val = 12 [packed = true];</code>
+   */
+  double getDcomplexVal(int index);
+
+  /**
+   * <pre>
+   * DT_RESOURCE
+   * </pre>
+   *
+   * <code>repeated .tensorflow.ResourceHandleProto resource_handle_val = 14;</code>
+   */
+  java.util.List<org.tensorflow.proto.framework.ResourceHandleProto> 
+      getResourceHandleValList();
+  /**
+   * <pre>
+   * DT_RESOURCE
+   * </pre>
+   *
+   * <code>repeated .tensorflow.ResourceHandleProto resource_handle_val = 14;</code>
+   */
+  org.tensorflow.proto.framework.ResourceHandleProto getResourceHandleVal(int index);
+  /**
+   * <pre>
+   * DT_RESOURCE
+   * </pre>
+   *
+   * <code>repeated .tensorflow.ResourceHandleProto resource_handle_val = 14;</code>
+   */
+  int getResourceHandleValCount();
+  /**
+   * <pre>
+   * DT_RESOURCE
+   * </pre>
+   *
+   * <code>repeated .tensorflow.ResourceHandleProto resource_handle_val = 14;</code>
+   */
+  java.util.List<? extends org.tensorflow.proto.framework.ResourceHandleProtoOrBuilder> 
+      getResourceHandleValOrBuilderList();
+  /**
+   * <pre>
+   * DT_RESOURCE
+   * </pre>
+   *
+   * <code>repeated .tensorflow.ResourceHandleProto resource_handle_val = 14;</code>
+   */
+  org.tensorflow.proto.framework.ResourceHandleProtoOrBuilder getResourceHandleValOrBuilder(
+      int index);
+
+  /**
+   * <pre>
+   * DT_VARIANT
+   * </pre>
+   *
+   * <code>repeated .tensorflow.VariantTensorDataProto variant_val = 15;</code>
+   */
+  java.util.List<org.tensorflow.proto.framework.VariantTensorDataProto> 
+      getVariantValList();
+  /**
+   * <pre>
+   * DT_VARIANT
+   * </pre>
+   *
+   * <code>repeated .tensorflow.VariantTensorDataProto variant_val = 15;</code>
+   */
+  org.tensorflow.proto.framework.VariantTensorDataProto getVariantVal(int index);
+  /**
+   * <pre>
+   * DT_VARIANT
+   * </pre>
+   *
+   * <code>repeated .tensorflow.VariantTensorDataProto variant_val = 15;</code>
+   */
+  int getVariantValCount();
+  /**
+   * <pre>
+   * DT_VARIANT
+   * </pre>
+   *
+   * <code>repeated .tensorflow.VariantTensorDataProto variant_val = 15;</code>
+   */
+  java.util.List<? extends org.tensorflow.proto.framework.VariantTensorDataProtoOrBuilder> 
+      getVariantValOrBuilderList();
+  /**
+   * <pre>
+   * DT_VARIANT
+   * </pre>
+   *
+   * <code>repeated .tensorflow.VariantTensorDataProto variant_val = 15;</code>
+   */
+  org.tensorflow.proto.framework.VariantTensorDataProtoOrBuilder getVariantValOrBuilder(
+      int index);
+
+  /**
+   * <pre>
+   * DT_UINT32
+   * </pre>
+   *
+   * <code>repeated uint32 uint32_val = 16 [packed = true];</code>
+   */
+  java.util.List<java.lang.Integer> getUint32ValList();
+  /**
+   * <pre>
+   * DT_UINT32
+   * </pre>
+   *
+   * <code>repeated uint32 uint32_val = 16 [packed = true];</code>
+   */
+  int getUint32ValCount();
+  /**
+   * <pre>
+   * DT_UINT32
+   * </pre>
+   *
+   * <code>repeated uint32 uint32_val = 16 [packed = true];</code>
+   */
+  int getUint32Val(int index);
+
+  /**
+   * <pre>
+   * DT_UINT64
+   * </pre>
+   *
+   * <code>repeated uint64 uint64_val = 17 [packed = true];</code>
+   */
+  java.util.List<java.lang.Long> getUint64ValList();
+  /**
+   * <pre>
+   * DT_UINT64
+   * </pre>
+   *
+   * <code>repeated uint64 uint64_val = 17 [packed = true];</code>
+   */
+  int getUint64ValCount();
+  /**
+   * <pre>
+   * DT_UINT64
+   * </pre>
+   *
+   * <code>repeated uint64 uint64_val = 17 [packed = true];</code>
+   */
+  long getUint64Val(int index);
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorProtos.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorProtos.java
new file mode 100644
index 00000000000..02a9fdadd6e
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorProtos.java
@@ -0,0 +1,86 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/framework/tensor.proto
+
+package org.tensorflow.proto.framework;
+
+public final class TensorProtos {
+  private TensorProtos() {}
+  public static void registerAllExtensions(
+      com.google.protobuf.ExtensionRegistryLite registry) {
+  }
+
+  public static void registerAllExtensions(
+      com.google.protobuf.ExtensionRegistry registry) {
+    registerAllExtensions(
+        (com.google.protobuf.ExtensionRegistryLite) registry);
+  }
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_TensorProto_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_TensorProto_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_VariantTensorDataProto_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_VariantTensorDataProto_fieldAccessorTable;
+
+  public static com.google.protobuf.Descriptors.FileDescriptor
+      getDescriptor() {
+    return descriptor;
+  }
+  private static  com.google.protobuf.Descriptors.FileDescriptor
+      descriptor;
+  static {
+    java.lang.String[] descriptorData = {
+      "\n&tensorflow/core/framework/tensor.proto" +
+      "\022\ntensorflow\032/tensorflow/core/framework/" +
+      "resource_handle.proto\032,tensorflow/core/f" +
+      "ramework/tensor_shape.proto\032%tensorflow/" +
+      "core/framework/types.proto\"\214\004\n\013TensorPro" +
+      "to\022#\n\005dtype\030\001 \001(\0162\024.tensorflow.DataType\022" +
+      "2\n\014tensor_shape\030\002 \001(\0132\034.tensorflow.Tenso" +
+      "rShapeProto\022\026\n\016version_number\030\003 \001(\005\022\026\n\016t" +
+      "ensor_content\030\004 \001(\014\022\024\n\010half_val\030\r \003(\005B\002\020" +
+      "\001\022\025\n\tfloat_val\030\005 \003(\002B\002\020\001\022\026\n\ndouble_val\030\006" +
+      " \003(\001B\002\020\001\022\023\n\007int_val\030\007 \003(\005B\002\020\001\022\022\n\nstring_" +
+      "val\030\010 \003(\014\022\030\n\014scomplex_val\030\t \003(\002B\002\020\001\022\025\n\ti" +
+      "nt64_val\030\n \003(\003B\002\020\001\022\024\n\010bool_val\030\013 \003(\010B\002\020\001" +
+      "\022\030\n\014dcomplex_val\030\014 \003(\001B\002\020\001\022<\n\023resource_h" +
+      "andle_val\030\016 \003(\0132\037.tensorflow.ResourceHan" +
+      "dleProto\0227\n\013variant_val\030\017 \003(\0132\".tensorfl" +
+      "ow.VariantTensorDataProto\022\026\n\nuint32_val\030" +
+      "\020 \003(\rB\002\020\001\022\026\n\nuint64_val\030\021 \003(\004B\002\020\001\"g\n\026Var" +
+      "iantTensorDataProto\022\021\n\ttype_name\030\001 \001(\t\022\020" +
+      "\n\010metadata\030\002 \001(\014\022(\n\007tensors\030\003 \003(\0132\027.tens" +
+      "orflow.TensorProtoB\202\001\n\036org.tensorflow.pr" +
+      "oto.frameworkB\014TensorProtosP\001ZMgithub.co" +
+      "m/tensorflow/tensorflow/tensorflow/go/co" +
+      "re/framework/tensor_go_proto\370\001\001b\006proto3"
+    };
+    descriptor = com.google.protobuf.Descriptors.FileDescriptor
+      .internalBuildGeneratedFileFrom(descriptorData,
+        new com.google.protobuf.Descriptors.FileDescriptor[] {
+          org.tensorflow.proto.framework.ResourceHandle.getDescriptor(),
+          org.tensorflow.proto.framework.TensorShapeProtos.getDescriptor(),
+          org.tensorflow.proto.framework.TypesProtos.getDescriptor(),
+        });
+    internal_static_tensorflow_TensorProto_descriptor =
+      getDescriptor().getMessageTypes().get(0);
+    internal_static_tensorflow_TensorProto_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_TensorProto_descriptor,
+        new java.lang.String[] { "Dtype", "TensorShape", "VersionNumber", "TensorContent", "HalfVal", "FloatVal", "DoubleVal", "IntVal", "StringVal", "ScomplexVal", "Int64Val", "BoolVal", "DcomplexVal", "ResourceHandleVal", "VariantVal", "Uint32Val", "Uint64Val", });
+    internal_static_tensorflow_VariantTensorDataProto_descriptor =
+      getDescriptor().getMessageTypes().get(1);
+    internal_static_tensorflow_VariantTensorDataProto_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_VariantTensorDataProto_descriptor,
+        new java.lang.String[] { "TypeName", "Metadata", "Tensors", });
+    org.tensorflow.proto.framework.ResourceHandle.getDescriptor();
+    org.tensorflow.proto.framework.TensorShapeProtos.getDescriptor();
+    org.tensorflow.proto.framework.TypesProtos.getDescriptor();
+  }
+
+  // @@protoc_insertion_point(outer_class_scope)
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorShapeProto.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorShapeProto.java
new file mode 100644
index 00000000000..1d0d543c656
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorShapeProto.java
@@ -0,0 +1,1852 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/framework/tensor_shape.proto
+
+package org.tensorflow.proto.framework;
+
+/**
+ * <pre>
+ * Dimensions of a tensor.
+ * </pre>
+ *
+ * Protobuf type {@code tensorflow.TensorShapeProto}
+ */
+public  final class TensorShapeProto extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.TensorShapeProto)
+    TensorShapeProtoOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use TensorShapeProto.newBuilder() to construct.
+  private TensorShapeProto(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private TensorShapeProto() {
+    dim_ = java.util.Collections.emptyList();
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new TensorShapeProto();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private TensorShapeProto(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    int mutable_bitField0_ = 0;
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 18: {
+            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
+              dim_ = new java.util.ArrayList<org.tensorflow.proto.framework.TensorShapeProto.Dim>();
+              mutable_bitField0_ |= 0x00000001;
+            }
+            dim_.add(
+                input.readMessage(org.tensorflow.proto.framework.TensorShapeProto.Dim.parser(), extensionRegistry));
+            break;
+          }
+          case 24: {
+
+            unknownRank_ = input.readBool();
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      if (((mutable_bitField0_ & 0x00000001) != 0)) {
+        dim_ = java.util.Collections.unmodifiableList(dim_);
+      }
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.framework.TensorShapeProtos.internal_static_tensorflow_TensorShapeProto_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.framework.TensorShapeProtos.internal_static_tensorflow_TensorShapeProto_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.framework.TensorShapeProto.class, org.tensorflow.proto.framework.TensorShapeProto.Builder.class);
+  }
+
+  public interface DimOrBuilder extends
+      // @@protoc_insertion_point(interface_extends:tensorflow.TensorShapeProto.Dim)
+      com.google.protobuf.MessageOrBuilder {
+
+    /**
+     * <pre>
+     * Size of the tensor in that dimension.
+     * This value must be &gt;= -1, but values of -1 are reserved for "unknown"
+     * shapes (values of -1 mean "unknown" dimension).  Certain wrappers
+     * that work with TensorShapeProto may fail at runtime when deserializing
+     * a TensorShapeProto containing a dim value of -1.
+     * </pre>
+     *
+     * <code>int64 size = 1;</code>
+     */
+    long getSize();
+
+    /**
+     * <pre>
+     * Optional name of the tensor dimension.
+     * </pre>
+     *
+     * <code>string name = 2;</code>
+     */
+    java.lang.String getName();
+    /**
+     * <pre>
+     * Optional name of the tensor dimension.
+     * </pre>
+     *
+     * <code>string name = 2;</code>
+     */
+    com.google.protobuf.ByteString
+        getNameBytes();
+  }
+  /**
+   * <pre>
+   * One dimension of the tensor.
+   * </pre>
+   *
+   * Protobuf type {@code tensorflow.TensorShapeProto.Dim}
+   */
+  public  static final class Dim extends
+      com.google.protobuf.GeneratedMessageV3 implements
+      // @@protoc_insertion_point(message_implements:tensorflow.TensorShapeProto.Dim)
+      DimOrBuilder {
+  private static final long serialVersionUID = 0L;
+    // Use Dim.newBuilder() to construct.
+    private Dim(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+      super(builder);
+    }
+    private Dim() {
+      name_ = "";
+    }
+
+    @java.lang.Override
+    @SuppressWarnings({"unused"})
+    protected java.lang.Object newInstance(
+        UnusedPrivateParameter unused) {
+      return new Dim();
+    }
+
+    @java.lang.Override
+    public final com.google.protobuf.UnknownFieldSet
+    getUnknownFields() {
+      return this.unknownFields;
+    }
+    private Dim(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      this();
+      if (extensionRegistry == null) {
+        throw new java.lang.NullPointerException();
+      }
+      com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+          com.google.protobuf.UnknownFieldSet.newBuilder();
+      try {
+        boolean done = false;
+        while (!done) {
+          int tag = input.readTag();
+          switch (tag) {
+            case 0:
+              done = true;
+              break;
+            case 8: {
+
+              size_ = input.readInt64();
+              break;
+            }
+            case 18: {
+              java.lang.String s = input.readStringRequireUtf8();
+
+              name_ = s;
+              break;
+            }
+            default: {
+              if (!parseUnknownField(
+                  input, unknownFields, extensionRegistry, tag)) {
+                done = true;
+              }
+              break;
+            }
+          }
+        }
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        throw e.setUnfinishedMessage(this);
+      } catch (java.io.IOException e) {
+        throw new com.google.protobuf.InvalidProtocolBufferException(
+            e).setUnfinishedMessage(this);
+      } finally {
+        this.unknownFields = unknownFields.build();
+        makeExtensionsImmutable();
+      }
+    }
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.TensorShapeProtos.internal_static_tensorflow_TensorShapeProto_Dim_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.TensorShapeProtos.internal_static_tensorflow_TensorShapeProto_Dim_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.TensorShapeProto.Dim.class, org.tensorflow.proto.framework.TensorShapeProto.Dim.Builder.class);
+    }
+
+    public static final int SIZE_FIELD_NUMBER = 1;
+    private long size_;
+    /**
+     * <pre>
+     * Size of the tensor in that dimension.
+     * This value must be &gt;= -1, but values of -1 are reserved for "unknown"
+     * shapes (values of -1 mean "unknown" dimension).  Certain wrappers
+     * that work with TensorShapeProto may fail at runtime when deserializing
+     * a TensorShapeProto containing a dim value of -1.
+     * </pre>
+     *
+     * <code>int64 size = 1;</code>
+     */
+    public long getSize() {
+      return size_;
+    }
+
+    public static final int NAME_FIELD_NUMBER = 2;
+    private volatile java.lang.Object name_;
+    /**
+     * <pre>
+     * Optional name of the tensor dimension.
+     * </pre>
+     *
+     * <code>string name = 2;</code>
+     */
+    public java.lang.String getName() {
+      java.lang.Object ref = name_;
+      if (ref instanceof java.lang.String) {
+        return (java.lang.String) ref;
+      } else {
+        com.google.protobuf.ByteString bs = 
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        name_ = s;
+        return s;
+      }
+    }
+    /**
+     * <pre>
+     * Optional name of the tensor dimension.
+     * </pre>
+     *
+     * <code>string name = 2;</code>
+     */
+    public com.google.protobuf.ByteString
+        getNameBytes() {
+      java.lang.Object ref = name_;
+      if (ref instanceof java.lang.String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        name_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+
+    private byte memoizedIsInitialized = -1;
+    @java.lang.Override
+    public final boolean isInitialized() {
+      byte isInitialized = memoizedIsInitialized;
+      if (isInitialized == 1) return true;
+      if (isInitialized == 0) return false;
+
+      memoizedIsInitialized = 1;
+      return true;
+    }
+
+    @java.lang.Override
+    public void writeTo(com.google.protobuf.CodedOutputStream output)
+                        throws java.io.IOException {
+      if (size_ != 0L) {
+        output.writeInt64(1, size_);
+      }
+      if (!getNameBytes().isEmpty()) {
+        com.google.protobuf.GeneratedMessageV3.writeString(output, 2, name_);
+      }
+      unknownFields.writeTo(output);
+    }
+
+    @java.lang.Override
+    public int getSerializedSize() {
+      int size = memoizedSize;
+      if (size != -1) return size;
+
+      size = 0;
+      if (size_ != 0L) {
+        size += com.google.protobuf.CodedOutputStream
+          .computeInt64Size(1, size_);
+      }
+      if (!getNameBytes().isEmpty()) {
+        size += com.google.protobuf.GeneratedMessageV3.computeStringSize(2, name_);
+      }
+      size += unknownFields.getSerializedSize();
+      memoizedSize = size;
+      return size;
+    }
+
+    @java.lang.Override
+    public boolean equals(final java.lang.Object obj) {
+      if (obj == this) {
+       return true;
+      }
+      if (!(obj instanceof org.tensorflow.proto.framework.TensorShapeProto.Dim)) {
+        return super.equals(obj);
+      }
+      org.tensorflow.proto.framework.TensorShapeProto.Dim other = (org.tensorflow.proto.framework.TensorShapeProto.Dim) obj;
+
+      if (getSize()
+          != other.getSize()) return false;
+      if (!getName()
+          .equals(other.getName())) return false;
+      if (!unknownFields.equals(other.unknownFields)) return false;
+      return true;
+    }
+
+    @java.lang.Override
+    public int hashCode() {
+      if (memoizedHashCode != 0) {
+        return memoizedHashCode;
+      }
+      int hash = 41;
+      hash = (19 * hash) + getDescriptor().hashCode();
+      hash = (37 * hash) + SIZE_FIELD_NUMBER;
+      hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+          getSize());
+      hash = (37 * hash) + NAME_FIELD_NUMBER;
+      hash = (53 * hash) + getName().hashCode();
+      hash = (29 * hash) + unknownFields.hashCode();
+      memoizedHashCode = hash;
+      return hash;
+    }
+
+    public static org.tensorflow.proto.framework.TensorShapeProto.Dim parseFrom(
+        java.nio.ByteBuffer data)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data);
+    }
+    public static org.tensorflow.proto.framework.TensorShapeProto.Dim parseFrom(
+        java.nio.ByteBuffer data,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.TensorShapeProto.Dim parseFrom(
+        com.google.protobuf.ByteString data)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data);
+    }
+    public static org.tensorflow.proto.framework.TensorShapeProto.Dim parseFrom(
+        com.google.protobuf.ByteString data,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.TensorShapeProto.Dim parseFrom(byte[] data)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data);
+    }
+    public static org.tensorflow.proto.framework.TensorShapeProto.Dim parseFrom(
+        byte[] data,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.TensorShapeProto.Dim parseFrom(java.io.InputStream input)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input);
+    }
+    public static org.tensorflow.proto.framework.TensorShapeProto.Dim parseFrom(
+        java.io.InputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.TensorShapeProto.Dim parseDelimitedFrom(java.io.InputStream input)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseDelimitedWithIOException(PARSER, input);
+    }
+    public static org.tensorflow.proto.framework.TensorShapeProto.Dim parseDelimitedFrom(
+        java.io.InputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.TensorShapeProto.Dim parseFrom(
+        com.google.protobuf.CodedInputStream input)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input);
+    }
+    public static org.tensorflow.proto.framework.TensorShapeProto.Dim parseFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input, extensionRegistry);
+    }
+
+    @java.lang.Override
+    public Builder newBuilderForType() { return newBuilder(); }
+    public static Builder newBuilder() {
+      return DEFAULT_INSTANCE.toBuilder();
+    }
+    public static Builder newBuilder(org.tensorflow.proto.framework.TensorShapeProto.Dim prototype) {
+      return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+    }
+    @java.lang.Override
+    public Builder toBuilder() {
+      return this == DEFAULT_INSTANCE
+          ? new Builder() : new Builder().mergeFrom(this);
+    }
+
+    @java.lang.Override
+    protected Builder newBuilderForType(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      Builder builder = new Builder(parent);
+      return builder;
+    }
+    /**
+     * <pre>
+     * One dimension of the tensor.
+     * </pre>
+     *
+     * Protobuf type {@code tensorflow.TensorShapeProto.Dim}
+     */
+    public static final class Builder extends
+        com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+        // @@protoc_insertion_point(builder_implements:tensorflow.TensorShapeProto.Dim)
+        org.tensorflow.proto.framework.TensorShapeProto.DimOrBuilder {
+      public static final com.google.protobuf.Descriptors.Descriptor
+          getDescriptor() {
+        return org.tensorflow.proto.framework.TensorShapeProtos.internal_static_tensorflow_TensorShapeProto_Dim_descriptor;
+      }
+
+      @java.lang.Override
+      protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+          internalGetFieldAccessorTable() {
+        return org.tensorflow.proto.framework.TensorShapeProtos.internal_static_tensorflow_TensorShapeProto_Dim_fieldAccessorTable
+            .ensureFieldAccessorsInitialized(
+                org.tensorflow.proto.framework.TensorShapeProto.Dim.class, org.tensorflow.proto.framework.TensorShapeProto.Dim.Builder.class);
+      }
+
+      // Construct using org.tensorflow.proto.framework.TensorShapeProto.Dim.newBuilder()
+      private Builder() {
+        maybeForceBuilderInitialization();
+      }
+
+      private Builder(
+          com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+        super(parent);
+        maybeForceBuilderInitialization();
+      }
+      private void maybeForceBuilderInitialization() {
+        if (com.google.protobuf.GeneratedMessageV3
+                .alwaysUseFieldBuilders) {
+        }
+      }
+      @java.lang.Override
+      public Builder clear() {
+        super.clear();
+        size_ = 0L;
+
+        name_ = "";
+
+        return this;
+      }
+
+      @java.lang.Override
+      public com.google.protobuf.Descriptors.Descriptor
+          getDescriptorForType() {
+        return org.tensorflow.proto.framework.TensorShapeProtos.internal_static_tensorflow_TensorShapeProto_Dim_descriptor;
+      }
+
+      @java.lang.Override
+      public org.tensorflow.proto.framework.TensorShapeProto.Dim getDefaultInstanceForType() {
+        return org.tensorflow.proto.framework.TensorShapeProto.Dim.getDefaultInstance();
+      }
+
+      @java.lang.Override
+      public org.tensorflow.proto.framework.TensorShapeProto.Dim build() {
+        org.tensorflow.proto.framework.TensorShapeProto.Dim result = buildPartial();
+        if (!result.isInitialized()) {
+          throw newUninitializedMessageException(result);
+        }
+        return result;
+      }
+
+      @java.lang.Override
+      public org.tensorflow.proto.framework.TensorShapeProto.Dim buildPartial() {
+        org.tensorflow.proto.framework.TensorShapeProto.Dim result = new org.tensorflow.proto.framework.TensorShapeProto.Dim(this);
+        result.size_ = size_;
+        result.name_ = name_;
+        onBuilt();
+        return result;
+      }
+
+      @java.lang.Override
+      public Builder clone() {
+        return super.clone();
+      }
+      @java.lang.Override
+      public Builder setField(
+          com.google.protobuf.Descriptors.FieldDescriptor field,
+          java.lang.Object value) {
+        return super.setField(field, value);
+      }
+      @java.lang.Override
+      public Builder clearField(
+          com.google.protobuf.Descriptors.FieldDescriptor field) {
+        return super.clearField(field);
+      }
+      @java.lang.Override
+      public Builder clearOneof(
+          com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+        return super.clearOneof(oneof);
+      }
+      @java.lang.Override
+      public Builder setRepeatedField(
+          com.google.protobuf.Descriptors.FieldDescriptor field,
+          int index, java.lang.Object value) {
+        return super.setRepeatedField(field, index, value);
+      }
+      @java.lang.Override
+      public Builder addRepeatedField(
+          com.google.protobuf.Descriptors.FieldDescriptor field,
+          java.lang.Object value) {
+        return super.addRepeatedField(field, value);
+      }
+      @java.lang.Override
+      public Builder mergeFrom(com.google.protobuf.Message other) {
+        if (other instanceof org.tensorflow.proto.framework.TensorShapeProto.Dim) {
+          return mergeFrom((org.tensorflow.proto.framework.TensorShapeProto.Dim)other);
+        } else {
+          super.mergeFrom(other);
+          return this;
+        }
+      }
+
+      public Builder mergeFrom(org.tensorflow.proto.framework.TensorShapeProto.Dim other) {
+        if (other == org.tensorflow.proto.framework.TensorShapeProto.Dim.getDefaultInstance()) return this;
+        if (other.getSize() != 0L) {
+          setSize(other.getSize());
+        }
+        if (!other.getName().isEmpty()) {
+          name_ = other.name_;
+          onChanged();
+        }
+        this.mergeUnknownFields(other.unknownFields);
+        onChanged();
+        return this;
+      }
+
+      @java.lang.Override
+      public final boolean isInitialized() {
+        return true;
+      }
+
+      @java.lang.Override
+      public Builder mergeFrom(
+          com.google.protobuf.CodedInputStream input,
+          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+          throws java.io.IOException {
+        org.tensorflow.proto.framework.TensorShapeProto.Dim parsedMessage = null;
+        try {
+          parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+        } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+          parsedMessage = (org.tensorflow.proto.framework.TensorShapeProto.Dim) e.getUnfinishedMessage();
+          throw e.unwrapIOException();
+        } finally {
+          if (parsedMessage != null) {
+            mergeFrom(parsedMessage);
+          }
+        }
+        return this;
+      }
+
+      private long size_ ;
+      /**
+       * <pre>
+       * Size of the tensor in that dimension.
+       * This value must be &gt;= -1, but values of -1 are reserved for "unknown"
+       * shapes (values of -1 mean "unknown" dimension).  Certain wrappers
+       * that work with TensorShapeProto may fail at runtime when deserializing
+       * a TensorShapeProto containing a dim value of -1.
+       * </pre>
+       *
+       * <code>int64 size = 1;</code>
+       */
+      public long getSize() {
+        return size_;
+      }
+      /**
+       * <pre>
+       * Size of the tensor in that dimension.
+       * This value must be &gt;= -1, but values of -1 are reserved for "unknown"
+       * shapes (values of -1 mean "unknown" dimension).  Certain wrappers
+       * that work with TensorShapeProto may fail at runtime when deserializing
+       * a TensorShapeProto containing a dim value of -1.
+       * </pre>
+       *
+       * <code>int64 size = 1;</code>
+       */
+      public Builder setSize(long value) {
+        
+        size_ = value;
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * Size of the tensor in that dimension.
+       * This value must be &gt;= -1, but values of -1 are reserved for "unknown"
+       * shapes (values of -1 mean "unknown" dimension).  Certain wrappers
+       * that work with TensorShapeProto may fail at runtime when deserializing
+       * a TensorShapeProto containing a dim value of -1.
+       * </pre>
+       *
+       * <code>int64 size = 1;</code>
+       */
+      public Builder clearSize() {
+        
+        size_ = 0L;
+        onChanged();
+        return this;
+      }
+
+      private java.lang.Object name_ = "";
+      /**
+       * <pre>
+       * Optional name of the tensor dimension.
+       * </pre>
+       *
+       * <code>string name = 2;</code>
+       */
+      public java.lang.String getName() {
+        java.lang.Object ref = name_;
+        if (!(ref instanceof java.lang.String)) {
+          com.google.protobuf.ByteString bs =
+              (com.google.protobuf.ByteString) ref;
+          java.lang.String s = bs.toStringUtf8();
+          name_ = s;
+          return s;
+        } else {
+          return (java.lang.String) ref;
+        }
+      }
+      /**
+       * <pre>
+       * Optional name of the tensor dimension.
+       * </pre>
+       *
+       * <code>string name = 2;</code>
+       */
+      public com.google.protobuf.ByteString
+          getNameBytes() {
+        java.lang.Object ref = name_;
+        if (ref instanceof String) {
+          com.google.protobuf.ByteString b = 
+              com.google.protobuf.ByteString.copyFromUtf8(
+                  (java.lang.String) ref);
+          name_ = b;
+          return b;
+        } else {
+          return (com.google.protobuf.ByteString) ref;
+        }
+      }
+      /**
+       * <pre>
+       * Optional name of the tensor dimension.
+       * </pre>
+       *
+       * <code>string name = 2;</code>
+       */
+      public Builder setName(
+          java.lang.String value) {
+        if (value == null) {
+    throw new NullPointerException();
+  }
+  
+        name_ = value;
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * Optional name of the tensor dimension.
+       * </pre>
+       *
+       * <code>string name = 2;</code>
+       */
+      public Builder clearName() {
+        
+        name_ = getDefaultInstance().getName();
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * Optional name of the tensor dimension.
+       * </pre>
+       *
+       * <code>string name = 2;</code>
+       */
+      public Builder setNameBytes(
+          com.google.protobuf.ByteString value) {
+        if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+        
+        name_ = value;
+        onChanged();
+        return this;
+      }
+      @java.lang.Override
+      public final Builder setUnknownFields(
+          final com.google.protobuf.UnknownFieldSet unknownFields) {
+        return super.setUnknownFields(unknownFields);
+      }
+
+      @java.lang.Override
+      public final Builder mergeUnknownFields(
+          final com.google.protobuf.UnknownFieldSet unknownFields) {
+        return super.mergeUnknownFields(unknownFields);
+      }
+
+
+      // @@protoc_insertion_point(builder_scope:tensorflow.TensorShapeProto.Dim)
+    }
+
+    // @@protoc_insertion_point(class_scope:tensorflow.TensorShapeProto.Dim)
+    private static final org.tensorflow.proto.framework.TensorShapeProto.Dim DEFAULT_INSTANCE;
+    static {
+      DEFAULT_INSTANCE = new org.tensorflow.proto.framework.TensorShapeProto.Dim();
+    }
+
+    public static org.tensorflow.proto.framework.TensorShapeProto.Dim getDefaultInstance() {
+      return DEFAULT_INSTANCE;
+    }
+
+    private static final com.google.protobuf.Parser<Dim>
+        PARSER = new com.google.protobuf.AbstractParser<Dim>() {
+      @java.lang.Override
+      public Dim parsePartialFrom(
+          com.google.protobuf.CodedInputStream input,
+          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+          throws com.google.protobuf.InvalidProtocolBufferException {
+        return new Dim(input, extensionRegistry);
+      }
+    };
+
+    public static com.google.protobuf.Parser<Dim> parser() {
+      return PARSER;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Parser<Dim> getParserForType() {
+      return PARSER;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.TensorShapeProto.Dim getDefaultInstanceForType() {
+      return DEFAULT_INSTANCE;
+    }
+
+  }
+
+  public static final int DIM_FIELD_NUMBER = 2;
+  private java.util.List<org.tensorflow.proto.framework.TensorShapeProto.Dim> dim_;
+  /**
+   * <pre>
+   * Dimensions of the tensor, such as {"input", 30}, {"output", 40}
+   * for a 30 x 40 2D tensor.  If an entry has size -1, this
+   * corresponds to a dimension of unknown size. The names are
+   * optional.
+   * The order of entries in "dim" matters: It indicates the layout of the
+   * values in the tensor in-memory representation.
+   * The first entry in "dim" is the outermost dimension used to layout the
+   * values, the last entry is the innermost dimension.  This matches the
+   * in-memory layout of RowMajor Eigen tensors.
+   * If "dim.size()" &gt; 0, "unknown_rank" must be false.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.TensorShapeProto.Dim dim = 2;</code>
+   */
+  public java.util.List<org.tensorflow.proto.framework.TensorShapeProto.Dim> getDimList() {
+    return dim_;
+  }
+  /**
+   * <pre>
+   * Dimensions of the tensor, such as {"input", 30}, {"output", 40}
+   * for a 30 x 40 2D tensor.  If an entry has size -1, this
+   * corresponds to a dimension of unknown size. The names are
+   * optional.
+   * The order of entries in "dim" matters: It indicates the layout of the
+   * values in the tensor in-memory representation.
+   * The first entry in "dim" is the outermost dimension used to layout the
+   * values, the last entry is the innermost dimension.  This matches the
+   * in-memory layout of RowMajor Eigen tensors.
+   * If "dim.size()" &gt; 0, "unknown_rank" must be false.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.TensorShapeProto.Dim dim = 2;</code>
+   */
+  public java.util.List<? extends org.tensorflow.proto.framework.TensorShapeProto.DimOrBuilder> 
+      getDimOrBuilderList() {
+    return dim_;
+  }
+  /**
+   * <pre>
+   * Dimensions of the tensor, such as {"input", 30}, {"output", 40}
+   * for a 30 x 40 2D tensor.  If an entry has size -1, this
+   * corresponds to a dimension of unknown size. The names are
+   * optional.
+   * The order of entries in "dim" matters: It indicates the layout of the
+   * values in the tensor in-memory representation.
+   * The first entry in "dim" is the outermost dimension used to layout the
+   * values, the last entry is the innermost dimension.  This matches the
+   * in-memory layout of RowMajor Eigen tensors.
+   * If "dim.size()" &gt; 0, "unknown_rank" must be false.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.TensorShapeProto.Dim dim = 2;</code>
+   */
+  public int getDimCount() {
+    return dim_.size();
+  }
+  /**
+   * <pre>
+   * Dimensions of the tensor, such as {"input", 30}, {"output", 40}
+   * for a 30 x 40 2D tensor.  If an entry has size -1, this
+   * corresponds to a dimension of unknown size. The names are
+   * optional.
+   * The order of entries in "dim" matters: It indicates the layout of the
+   * values in the tensor in-memory representation.
+   * The first entry in "dim" is the outermost dimension used to layout the
+   * values, the last entry is the innermost dimension.  This matches the
+   * in-memory layout of RowMajor Eigen tensors.
+   * If "dim.size()" &gt; 0, "unknown_rank" must be false.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.TensorShapeProto.Dim dim = 2;</code>
+   */
+  public org.tensorflow.proto.framework.TensorShapeProto.Dim getDim(int index) {
+    return dim_.get(index);
+  }
+  /**
+   * <pre>
+   * Dimensions of the tensor, such as {"input", 30}, {"output", 40}
+   * for a 30 x 40 2D tensor.  If an entry has size -1, this
+   * corresponds to a dimension of unknown size. The names are
+   * optional.
+   * The order of entries in "dim" matters: It indicates the layout of the
+   * values in the tensor in-memory representation.
+   * The first entry in "dim" is the outermost dimension used to layout the
+   * values, the last entry is the innermost dimension.  This matches the
+   * in-memory layout of RowMajor Eigen tensors.
+   * If "dim.size()" &gt; 0, "unknown_rank" must be false.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.TensorShapeProto.Dim dim = 2;</code>
+   */
+  public org.tensorflow.proto.framework.TensorShapeProto.DimOrBuilder getDimOrBuilder(
+      int index) {
+    return dim_.get(index);
+  }
+
+  public static final int UNKNOWN_RANK_FIELD_NUMBER = 3;
+  private boolean unknownRank_;
+  /**
+   * <pre>
+   * If true, the number of dimensions in the shape is unknown.
+   * If true, "dim.size()" must be 0.
+   * </pre>
+   *
+   * <code>bool unknown_rank = 3;</code>
+   */
+  public boolean getUnknownRank() {
+    return unknownRank_;
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    for (int i = 0; i < dim_.size(); i++) {
+      output.writeMessage(2, dim_.get(i));
+    }
+    if (unknownRank_ != false) {
+      output.writeBool(3, unknownRank_);
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    for (int i = 0; i < dim_.size(); i++) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(2, dim_.get(i));
+    }
+    if (unknownRank_ != false) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeBoolSize(3, unknownRank_);
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.framework.TensorShapeProto)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.framework.TensorShapeProto other = (org.tensorflow.proto.framework.TensorShapeProto) obj;
+
+    if (!getDimList()
+        .equals(other.getDimList())) return false;
+    if (getUnknownRank()
+        != other.getUnknownRank()) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    if (getDimCount() > 0) {
+      hash = (37 * hash) + DIM_FIELD_NUMBER;
+      hash = (53 * hash) + getDimList().hashCode();
+    }
+    hash = (37 * hash) + UNKNOWN_RANK_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashBoolean(
+        getUnknownRank());
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.framework.TensorShapeProto parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.TensorShapeProto parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.TensorShapeProto parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.TensorShapeProto parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.TensorShapeProto parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.TensorShapeProto parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.TensorShapeProto parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.TensorShapeProto parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.TensorShapeProto parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.TensorShapeProto parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.TensorShapeProto parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.TensorShapeProto parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.framework.TensorShapeProto prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * <pre>
+   * Dimensions of a tensor.
+   * </pre>
+   *
+   * Protobuf type {@code tensorflow.TensorShapeProto}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.TensorShapeProto)
+      org.tensorflow.proto.framework.TensorShapeProtoOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.TensorShapeProtos.internal_static_tensorflow_TensorShapeProto_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.TensorShapeProtos.internal_static_tensorflow_TensorShapeProto_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.TensorShapeProto.class, org.tensorflow.proto.framework.TensorShapeProto.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.framework.TensorShapeProto.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+        getDimFieldBuilder();
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      if (dimBuilder_ == null) {
+        dim_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000001);
+      } else {
+        dimBuilder_.clear();
+      }
+      unknownRank_ = false;
+
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.framework.TensorShapeProtos.internal_static_tensorflow_TensorShapeProto_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.TensorShapeProto getDefaultInstanceForType() {
+      return org.tensorflow.proto.framework.TensorShapeProto.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.TensorShapeProto build() {
+      org.tensorflow.proto.framework.TensorShapeProto result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.TensorShapeProto buildPartial() {
+      org.tensorflow.proto.framework.TensorShapeProto result = new org.tensorflow.proto.framework.TensorShapeProto(this);
+      int from_bitField0_ = bitField0_;
+      if (dimBuilder_ == null) {
+        if (((bitField0_ & 0x00000001) != 0)) {
+          dim_ = java.util.Collections.unmodifiableList(dim_);
+          bitField0_ = (bitField0_ & ~0x00000001);
+        }
+        result.dim_ = dim_;
+      } else {
+        result.dim_ = dimBuilder_.build();
+      }
+      result.unknownRank_ = unknownRank_;
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.framework.TensorShapeProto) {
+        return mergeFrom((org.tensorflow.proto.framework.TensorShapeProto)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.framework.TensorShapeProto other) {
+      if (other == org.tensorflow.proto.framework.TensorShapeProto.getDefaultInstance()) return this;
+      if (dimBuilder_ == null) {
+        if (!other.dim_.isEmpty()) {
+          if (dim_.isEmpty()) {
+            dim_ = other.dim_;
+            bitField0_ = (bitField0_ & ~0x00000001);
+          } else {
+            ensureDimIsMutable();
+            dim_.addAll(other.dim_);
+          }
+          onChanged();
+        }
+      } else {
+        if (!other.dim_.isEmpty()) {
+          if (dimBuilder_.isEmpty()) {
+            dimBuilder_.dispose();
+            dimBuilder_ = null;
+            dim_ = other.dim_;
+            bitField0_ = (bitField0_ & ~0x00000001);
+            dimBuilder_ = 
+              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
+                 getDimFieldBuilder() : null;
+          } else {
+            dimBuilder_.addAllMessages(other.dim_);
+          }
+        }
+      }
+      if (other.getUnknownRank() != false) {
+        setUnknownRank(other.getUnknownRank());
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.framework.TensorShapeProto parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.framework.TensorShapeProto) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+    private int bitField0_;
+
+    private java.util.List<org.tensorflow.proto.framework.TensorShapeProto.Dim> dim_ =
+      java.util.Collections.emptyList();
+    private void ensureDimIsMutable() {
+      if (!((bitField0_ & 0x00000001) != 0)) {
+        dim_ = new java.util.ArrayList<org.tensorflow.proto.framework.TensorShapeProto.Dim>(dim_);
+        bitField0_ |= 0x00000001;
+       }
+    }
+
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.framework.TensorShapeProto.Dim, org.tensorflow.proto.framework.TensorShapeProto.Dim.Builder, org.tensorflow.proto.framework.TensorShapeProto.DimOrBuilder> dimBuilder_;
+
+    /**
+     * <pre>
+     * Dimensions of the tensor, such as {"input", 30}, {"output", 40}
+     * for a 30 x 40 2D tensor.  If an entry has size -1, this
+     * corresponds to a dimension of unknown size. The names are
+     * optional.
+     * The order of entries in "dim" matters: It indicates the layout of the
+     * values in the tensor in-memory representation.
+     * The first entry in "dim" is the outermost dimension used to layout the
+     * values, the last entry is the innermost dimension.  This matches the
+     * in-memory layout of RowMajor Eigen tensors.
+     * If "dim.size()" &gt; 0, "unknown_rank" must be false.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorShapeProto.Dim dim = 2;</code>
+     */
+    public java.util.List<org.tensorflow.proto.framework.TensorShapeProto.Dim> getDimList() {
+      if (dimBuilder_ == null) {
+        return java.util.Collections.unmodifiableList(dim_);
+      } else {
+        return dimBuilder_.getMessageList();
+      }
+    }
+    /**
+     * <pre>
+     * Dimensions of the tensor, such as {"input", 30}, {"output", 40}
+     * for a 30 x 40 2D tensor.  If an entry has size -1, this
+     * corresponds to a dimension of unknown size. The names are
+     * optional.
+     * The order of entries in "dim" matters: It indicates the layout of the
+     * values in the tensor in-memory representation.
+     * The first entry in "dim" is the outermost dimension used to layout the
+     * values, the last entry is the innermost dimension.  This matches the
+     * in-memory layout of RowMajor Eigen tensors.
+     * If "dim.size()" &gt; 0, "unknown_rank" must be false.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorShapeProto.Dim dim = 2;</code>
+     */
+    public int getDimCount() {
+      if (dimBuilder_ == null) {
+        return dim_.size();
+      } else {
+        return dimBuilder_.getCount();
+      }
+    }
+    /**
+     * <pre>
+     * Dimensions of the tensor, such as {"input", 30}, {"output", 40}
+     * for a 30 x 40 2D tensor.  If an entry has size -1, this
+     * corresponds to a dimension of unknown size. The names are
+     * optional.
+     * The order of entries in "dim" matters: It indicates the layout of the
+     * values in the tensor in-memory representation.
+     * The first entry in "dim" is the outermost dimension used to layout the
+     * values, the last entry is the innermost dimension.  This matches the
+     * in-memory layout of RowMajor Eigen tensors.
+     * If "dim.size()" &gt; 0, "unknown_rank" must be false.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorShapeProto.Dim dim = 2;</code>
+     */
+    public org.tensorflow.proto.framework.TensorShapeProto.Dim getDim(int index) {
+      if (dimBuilder_ == null) {
+        return dim_.get(index);
+      } else {
+        return dimBuilder_.getMessage(index);
+      }
+    }
+    /**
+     * <pre>
+     * Dimensions of the tensor, such as {"input", 30}, {"output", 40}
+     * for a 30 x 40 2D tensor.  If an entry has size -1, this
+     * corresponds to a dimension of unknown size. The names are
+     * optional.
+     * The order of entries in "dim" matters: It indicates the layout of the
+     * values in the tensor in-memory representation.
+     * The first entry in "dim" is the outermost dimension used to layout the
+     * values, the last entry is the innermost dimension.  This matches the
+     * in-memory layout of RowMajor Eigen tensors.
+     * If "dim.size()" &gt; 0, "unknown_rank" must be false.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorShapeProto.Dim dim = 2;</code>
+     */
+    public Builder setDim(
+        int index, org.tensorflow.proto.framework.TensorShapeProto.Dim value) {
+      if (dimBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureDimIsMutable();
+        dim_.set(index, value);
+        onChanged();
+      } else {
+        dimBuilder_.setMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Dimensions of the tensor, such as {"input", 30}, {"output", 40}
+     * for a 30 x 40 2D tensor.  If an entry has size -1, this
+     * corresponds to a dimension of unknown size. The names are
+     * optional.
+     * The order of entries in "dim" matters: It indicates the layout of the
+     * values in the tensor in-memory representation.
+     * The first entry in "dim" is the outermost dimension used to layout the
+     * values, the last entry is the innermost dimension.  This matches the
+     * in-memory layout of RowMajor Eigen tensors.
+     * If "dim.size()" &gt; 0, "unknown_rank" must be false.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorShapeProto.Dim dim = 2;</code>
+     */
+    public Builder setDim(
+        int index, org.tensorflow.proto.framework.TensorShapeProto.Dim.Builder builderForValue) {
+      if (dimBuilder_ == null) {
+        ensureDimIsMutable();
+        dim_.set(index, builderForValue.build());
+        onChanged();
+      } else {
+        dimBuilder_.setMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Dimensions of the tensor, such as {"input", 30}, {"output", 40}
+     * for a 30 x 40 2D tensor.  If an entry has size -1, this
+     * corresponds to a dimension of unknown size. The names are
+     * optional.
+     * The order of entries in "dim" matters: It indicates the layout of the
+     * values in the tensor in-memory representation.
+     * The first entry in "dim" is the outermost dimension used to layout the
+     * values, the last entry is the innermost dimension.  This matches the
+     * in-memory layout of RowMajor Eigen tensors.
+     * If "dim.size()" &gt; 0, "unknown_rank" must be false.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorShapeProto.Dim dim = 2;</code>
+     */
+    public Builder addDim(org.tensorflow.proto.framework.TensorShapeProto.Dim value) {
+      if (dimBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureDimIsMutable();
+        dim_.add(value);
+        onChanged();
+      } else {
+        dimBuilder_.addMessage(value);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Dimensions of the tensor, such as {"input", 30}, {"output", 40}
+     * for a 30 x 40 2D tensor.  If an entry has size -1, this
+     * corresponds to a dimension of unknown size. The names are
+     * optional.
+     * The order of entries in "dim" matters: It indicates the layout of the
+     * values in the tensor in-memory representation.
+     * The first entry in "dim" is the outermost dimension used to layout the
+     * values, the last entry is the innermost dimension.  This matches the
+     * in-memory layout of RowMajor Eigen tensors.
+     * If "dim.size()" &gt; 0, "unknown_rank" must be false.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorShapeProto.Dim dim = 2;</code>
+     */
+    public Builder addDim(
+        int index, org.tensorflow.proto.framework.TensorShapeProto.Dim value) {
+      if (dimBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureDimIsMutable();
+        dim_.add(index, value);
+        onChanged();
+      } else {
+        dimBuilder_.addMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Dimensions of the tensor, such as {"input", 30}, {"output", 40}
+     * for a 30 x 40 2D tensor.  If an entry has size -1, this
+     * corresponds to a dimension of unknown size. The names are
+     * optional.
+     * The order of entries in "dim" matters: It indicates the layout of the
+     * values in the tensor in-memory representation.
+     * The first entry in "dim" is the outermost dimension used to layout the
+     * values, the last entry is the innermost dimension.  This matches the
+     * in-memory layout of RowMajor Eigen tensors.
+     * If "dim.size()" &gt; 0, "unknown_rank" must be false.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorShapeProto.Dim dim = 2;</code>
+     */
+    public Builder addDim(
+        org.tensorflow.proto.framework.TensorShapeProto.Dim.Builder builderForValue) {
+      if (dimBuilder_ == null) {
+        ensureDimIsMutable();
+        dim_.add(builderForValue.build());
+        onChanged();
+      } else {
+        dimBuilder_.addMessage(builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Dimensions of the tensor, such as {"input", 30}, {"output", 40}
+     * for a 30 x 40 2D tensor.  If an entry has size -1, this
+     * corresponds to a dimension of unknown size. The names are
+     * optional.
+     * The order of entries in "dim" matters: It indicates the layout of the
+     * values in the tensor in-memory representation.
+     * The first entry in "dim" is the outermost dimension used to layout the
+     * values, the last entry is the innermost dimension.  This matches the
+     * in-memory layout of RowMajor Eigen tensors.
+     * If "dim.size()" &gt; 0, "unknown_rank" must be false.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorShapeProto.Dim dim = 2;</code>
+     */
+    public Builder addDim(
+        int index, org.tensorflow.proto.framework.TensorShapeProto.Dim.Builder builderForValue) {
+      if (dimBuilder_ == null) {
+        ensureDimIsMutable();
+        dim_.add(index, builderForValue.build());
+        onChanged();
+      } else {
+        dimBuilder_.addMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Dimensions of the tensor, such as {"input", 30}, {"output", 40}
+     * for a 30 x 40 2D tensor.  If an entry has size -1, this
+     * corresponds to a dimension of unknown size. The names are
+     * optional.
+     * The order of entries in "dim" matters: It indicates the layout of the
+     * values in the tensor in-memory representation.
+     * The first entry in "dim" is the outermost dimension used to layout the
+     * values, the last entry is the innermost dimension.  This matches the
+     * in-memory layout of RowMajor Eigen tensors.
+     * If "dim.size()" &gt; 0, "unknown_rank" must be false.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorShapeProto.Dim dim = 2;</code>
+     */
+    public Builder addAllDim(
+        java.lang.Iterable<? extends org.tensorflow.proto.framework.TensorShapeProto.Dim> values) {
+      if (dimBuilder_ == null) {
+        ensureDimIsMutable();
+        com.google.protobuf.AbstractMessageLite.Builder.addAll(
+            values, dim_);
+        onChanged();
+      } else {
+        dimBuilder_.addAllMessages(values);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Dimensions of the tensor, such as {"input", 30}, {"output", 40}
+     * for a 30 x 40 2D tensor.  If an entry has size -1, this
+     * corresponds to a dimension of unknown size. The names are
+     * optional.
+     * The order of entries in "dim" matters: It indicates the layout of the
+     * values in the tensor in-memory representation.
+     * The first entry in "dim" is the outermost dimension used to layout the
+     * values, the last entry is the innermost dimension.  This matches the
+     * in-memory layout of RowMajor Eigen tensors.
+     * If "dim.size()" &gt; 0, "unknown_rank" must be false.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorShapeProto.Dim dim = 2;</code>
+     */
+    public Builder clearDim() {
+      if (dimBuilder_ == null) {
+        dim_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000001);
+        onChanged();
+      } else {
+        dimBuilder_.clear();
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Dimensions of the tensor, such as {"input", 30}, {"output", 40}
+     * for a 30 x 40 2D tensor.  If an entry has size -1, this
+     * corresponds to a dimension of unknown size. The names are
+     * optional.
+     * The order of entries in "dim" matters: It indicates the layout of the
+     * values in the tensor in-memory representation.
+     * The first entry in "dim" is the outermost dimension used to layout the
+     * values, the last entry is the innermost dimension.  This matches the
+     * in-memory layout of RowMajor Eigen tensors.
+     * If "dim.size()" &gt; 0, "unknown_rank" must be false.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorShapeProto.Dim dim = 2;</code>
+     */
+    public Builder removeDim(int index) {
+      if (dimBuilder_ == null) {
+        ensureDimIsMutable();
+        dim_.remove(index);
+        onChanged();
+      } else {
+        dimBuilder_.remove(index);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Dimensions of the tensor, such as {"input", 30}, {"output", 40}
+     * for a 30 x 40 2D tensor.  If an entry has size -1, this
+     * corresponds to a dimension of unknown size. The names are
+     * optional.
+     * The order of entries in "dim" matters: It indicates the layout of the
+     * values in the tensor in-memory representation.
+     * The first entry in "dim" is the outermost dimension used to layout the
+     * values, the last entry is the innermost dimension.  This matches the
+     * in-memory layout of RowMajor Eigen tensors.
+     * If "dim.size()" &gt; 0, "unknown_rank" must be false.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorShapeProto.Dim dim = 2;</code>
+     */
+    public org.tensorflow.proto.framework.TensorShapeProto.Dim.Builder getDimBuilder(
+        int index) {
+      return getDimFieldBuilder().getBuilder(index);
+    }
+    /**
+     * <pre>
+     * Dimensions of the tensor, such as {"input", 30}, {"output", 40}
+     * for a 30 x 40 2D tensor.  If an entry has size -1, this
+     * corresponds to a dimension of unknown size. The names are
+     * optional.
+     * The order of entries in "dim" matters: It indicates the layout of the
+     * values in the tensor in-memory representation.
+     * The first entry in "dim" is the outermost dimension used to layout the
+     * values, the last entry is the innermost dimension.  This matches the
+     * in-memory layout of RowMajor Eigen tensors.
+     * If "dim.size()" &gt; 0, "unknown_rank" must be false.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorShapeProto.Dim dim = 2;</code>
+     */
+    public org.tensorflow.proto.framework.TensorShapeProto.DimOrBuilder getDimOrBuilder(
+        int index) {
+      if (dimBuilder_ == null) {
+        return dim_.get(index);  } else {
+        return dimBuilder_.getMessageOrBuilder(index);
+      }
+    }
+    /**
+     * <pre>
+     * Dimensions of the tensor, such as {"input", 30}, {"output", 40}
+     * for a 30 x 40 2D tensor.  If an entry has size -1, this
+     * corresponds to a dimension of unknown size. The names are
+     * optional.
+     * The order of entries in "dim" matters: It indicates the layout of the
+     * values in the tensor in-memory representation.
+     * The first entry in "dim" is the outermost dimension used to layout the
+     * values, the last entry is the innermost dimension.  This matches the
+     * in-memory layout of RowMajor Eigen tensors.
+     * If "dim.size()" &gt; 0, "unknown_rank" must be false.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorShapeProto.Dim dim = 2;</code>
+     */
+    public java.util.List<? extends org.tensorflow.proto.framework.TensorShapeProto.DimOrBuilder> 
+         getDimOrBuilderList() {
+      if (dimBuilder_ != null) {
+        return dimBuilder_.getMessageOrBuilderList();
+      } else {
+        return java.util.Collections.unmodifiableList(dim_);
+      }
+    }
+    /**
+     * <pre>
+     * Dimensions of the tensor, such as {"input", 30}, {"output", 40}
+     * for a 30 x 40 2D tensor.  If an entry has size -1, this
+     * corresponds to a dimension of unknown size. The names are
+     * optional.
+     * The order of entries in "dim" matters: It indicates the layout of the
+     * values in the tensor in-memory representation.
+     * The first entry in "dim" is the outermost dimension used to layout the
+     * values, the last entry is the innermost dimension.  This matches the
+     * in-memory layout of RowMajor Eigen tensors.
+     * If "dim.size()" &gt; 0, "unknown_rank" must be false.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorShapeProto.Dim dim = 2;</code>
+     */
+    public org.tensorflow.proto.framework.TensorShapeProto.Dim.Builder addDimBuilder() {
+      return getDimFieldBuilder().addBuilder(
+          org.tensorflow.proto.framework.TensorShapeProto.Dim.getDefaultInstance());
+    }
+    /**
+     * <pre>
+     * Dimensions of the tensor, such as {"input", 30}, {"output", 40}
+     * for a 30 x 40 2D tensor.  If an entry has size -1, this
+     * corresponds to a dimension of unknown size. The names are
+     * optional.
+     * The order of entries in "dim" matters: It indicates the layout of the
+     * values in the tensor in-memory representation.
+     * The first entry in "dim" is the outermost dimension used to layout the
+     * values, the last entry is the innermost dimension.  This matches the
+     * in-memory layout of RowMajor Eigen tensors.
+     * If "dim.size()" &gt; 0, "unknown_rank" must be false.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorShapeProto.Dim dim = 2;</code>
+     */
+    public org.tensorflow.proto.framework.TensorShapeProto.Dim.Builder addDimBuilder(
+        int index) {
+      return getDimFieldBuilder().addBuilder(
+          index, org.tensorflow.proto.framework.TensorShapeProto.Dim.getDefaultInstance());
+    }
+    /**
+     * <pre>
+     * Dimensions of the tensor, such as {"input", 30}, {"output", 40}
+     * for a 30 x 40 2D tensor.  If an entry has size -1, this
+     * corresponds to a dimension of unknown size. The names are
+     * optional.
+     * The order of entries in "dim" matters: It indicates the layout of the
+     * values in the tensor in-memory representation.
+     * The first entry in "dim" is the outermost dimension used to layout the
+     * values, the last entry is the innermost dimension.  This matches the
+     * in-memory layout of RowMajor Eigen tensors.
+     * If "dim.size()" &gt; 0, "unknown_rank" must be false.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorShapeProto.Dim dim = 2;</code>
+     */
+    public java.util.List<org.tensorflow.proto.framework.TensorShapeProto.Dim.Builder> 
+         getDimBuilderList() {
+      return getDimFieldBuilder().getBuilderList();
+    }
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.framework.TensorShapeProto.Dim, org.tensorflow.proto.framework.TensorShapeProto.Dim.Builder, org.tensorflow.proto.framework.TensorShapeProto.DimOrBuilder> 
+        getDimFieldBuilder() {
+      if (dimBuilder_ == null) {
+        dimBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
+            org.tensorflow.proto.framework.TensorShapeProto.Dim, org.tensorflow.proto.framework.TensorShapeProto.Dim.Builder, org.tensorflow.proto.framework.TensorShapeProto.DimOrBuilder>(
+                dim_,
+                ((bitField0_ & 0x00000001) != 0),
+                getParentForChildren(),
+                isClean());
+        dim_ = null;
+      }
+      return dimBuilder_;
+    }
+
+    private boolean unknownRank_ ;
+    /**
+     * <pre>
+     * If true, the number of dimensions in the shape is unknown.
+     * If true, "dim.size()" must be 0.
+     * </pre>
+     *
+     * <code>bool unknown_rank = 3;</code>
+     */
+    public boolean getUnknownRank() {
+      return unknownRank_;
+    }
+    /**
+     * <pre>
+     * If true, the number of dimensions in the shape is unknown.
+     * If true, "dim.size()" must be 0.
+     * </pre>
+     *
+     * <code>bool unknown_rank = 3;</code>
+     */
+    public Builder setUnknownRank(boolean value) {
+      
+      unknownRank_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * If true, the number of dimensions in the shape is unknown.
+     * If true, "dim.size()" must be 0.
+     * </pre>
+     *
+     * <code>bool unknown_rank = 3;</code>
+     */
+    public Builder clearUnknownRank() {
+      
+      unknownRank_ = false;
+      onChanged();
+      return this;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.TensorShapeProto)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.TensorShapeProto)
+  private static final org.tensorflow.proto.framework.TensorShapeProto DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.TensorShapeProto();
+  }
+
+  public static org.tensorflow.proto.framework.TensorShapeProto getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<TensorShapeProto>
+      PARSER = new com.google.protobuf.AbstractParser<TensorShapeProto>() {
+    @java.lang.Override
+    public TensorShapeProto parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new TensorShapeProto(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<TensorShapeProto> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<TensorShapeProto> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.framework.TensorShapeProto getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorShapeProtoOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorShapeProtoOrBuilder.java
new file mode 100644
index 00000000000..7835368954b
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorShapeProtoOrBuilder.java
@@ -0,0 +1,108 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/framework/tensor_shape.proto
+
+package org.tensorflow.proto.framework;
+
+public interface TensorShapeProtoOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.TensorShapeProto)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <pre>
+   * Dimensions of the tensor, such as {"input", 30}, {"output", 40}
+   * for a 30 x 40 2D tensor.  If an entry has size -1, this
+   * corresponds to a dimension of unknown size. The names are
+   * optional.
+   * The order of entries in "dim" matters: It indicates the layout of the
+   * values in the tensor in-memory representation.
+   * The first entry in "dim" is the outermost dimension used to layout the
+   * values, the last entry is the innermost dimension.  This matches the
+   * in-memory layout of RowMajor Eigen tensors.
+   * If "dim.size()" &gt; 0, "unknown_rank" must be false.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.TensorShapeProto.Dim dim = 2;</code>
+   */
+  java.util.List<org.tensorflow.proto.framework.TensorShapeProto.Dim> 
+      getDimList();
+  /**
+   * <pre>
+   * Dimensions of the tensor, such as {"input", 30}, {"output", 40}
+   * for a 30 x 40 2D tensor.  If an entry has size -1, this
+   * corresponds to a dimension of unknown size. The names are
+   * optional.
+   * The order of entries in "dim" matters: It indicates the layout of the
+   * values in the tensor in-memory representation.
+   * The first entry in "dim" is the outermost dimension used to layout the
+   * values, the last entry is the innermost dimension.  This matches the
+   * in-memory layout of RowMajor Eigen tensors.
+   * If "dim.size()" &gt; 0, "unknown_rank" must be false.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.TensorShapeProto.Dim dim = 2;</code>
+   */
+  org.tensorflow.proto.framework.TensorShapeProto.Dim getDim(int index);
+  /**
+   * <pre>
+   * Dimensions of the tensor, such as {"input", 30}, {"output", 40}
+   * for a 30 x 40 2D tensor.  If an entry has size -1, this
+   * corresponds to a dimension of unknown size. The names are
+   * optional.
+   * The order of entries in "dim" matters: It indicates the layout of the
+   * values in the tensor in-memory representation.
+   * The first entry in "dim" is the outermost dimension used to layout the
+   * values, the last entry is the innermost dimension.  This matches the
+   * in-memory layout of RowMajor Eigen tensors.
+   * If "dim.size()" &gt; 0, "unknown_rank" must be false.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.TensorShapeProto.Dim dim = 2;</code>
+   */
+  int getDimCount();
+  /**
+   * <pre>
+   * Dimensions of the tensor, such as {"input", 30}, {"output", 40}
+   * for a 30 x 40 2D tensor.  If an entry has size -1, this
+   * corresponds to a dimension of unknown size. The names are
+   * optional.
+   * The order of entries in "dim" matters: It indicates the layout of the
+   * values in the tensor in-memory representation.
+   * The first entry in "dim" is the outermost dimension used to layout the
+   * values, the last entry is the innermost dimension.  This matches the
+   * in-memory layout of RowMajor Eigen tensors.
+   * If "dim.size()" &gt; 0, "unknown_rank" must be false.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.TensorShapeProto.Dim dim = 2;</code>
+   */
+  java.util.List<? extends org.tensorflow.proto.framework.TensorShapeProto.DimOrBuilder> 
+      getDimOrBuilderList();
+  /**
+   * <pre>
+   * Dimensions of the tensor, such as {"input", 30}, {"output", 40}
+   * for a 30 x 40 2D tensor.  If an entry has size -1, this
+   * corresponds to a dimension of unknown size. The names are
+   * optional.
+   * The order of entries in "dim" matters: It indicates the layout of the
+   * values in the tensor in-memory representation.
+   * The first entry in "dim" is the outermost dimension used to layout the
+   * values, the last entry is the innermost dimension.  This matches the
+   * in-memory layout of RowMajor Eigen tensors.
+   * If "dim.size()" &gt; 0, "unknown_rank" must be false.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.TensorShapeProto.Dim dim = 2;</code>
+   */
+  org.tensorflow.proto.framework.TensorShapeProto.DimOrBuilder getDimOrBuilder(
+      int index);
+
+  /**
+   * <pre>
+   * If true, the number of dimensions in the shape is unknown.
+   * If true, "dim.size()" must be 0.
+   * </pre>
+   *
+   * <code>bool unknown_rank = 3;</code>
+   */
+  boolean getUnknownRank();
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorShapeProtos.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorShapeProtos.java
new file mode 100644
index 00000000000..e8214c97ec1
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorShapeProtos.java
@@ -0,0 +1,65 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/framework/tensor_shape.proto
+
+package org.tensorflow.proto.framework;
+
+public final class TensorShapeProtos {
+  private TensorShapeProtos() {}
+  public static void registerAllExtensions(
+      com.google.protobuf.ExtensionRegistryLite registry) {
+  }
+
+  public static void registerAllExtensions(
+      com.google.protobuf.ExtensionRegistry registry) {
+    registerAllExtensions(
+        (com.google.protobuf.ExtensionRegistryLite) registry);
+  }
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_TensorShapeProto_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_TensorShapeProto_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_TensorShapeProto_Dim_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_TensorShapeProto_Dim_fieldAccessorTable;
+
+  public static com.google.protobuf.Descriptors.FileDescriptor
+      getDescriptor() {
+    return descriptor;
+  }
+  private static  com.google.protobuf.Descriptors.FileDescriptor
+      descriptor;
+  static {
+    java.lang.String[] descriptorData = {
+      "\n,tensorflow/core/framework/tensor_shape" +
+      ".proto\022\ntensorflow\"z\n\020TensorShapeProto\022-" +
+      "\n\003dim\030\002 \003(\0132 .tensorflow.TensorShapeProt" +
+      "o.Dim\022\024\n\014unknown_rank\030\003 \001(\010\032!\n\003Dim\022\014\n\004si" +
+      "ze\030\001 \001(\003\022\014\n\004name\030\002 \001(\tB\215\001\n\036org.tensorflo" +
+      "w.proto.frameworkB\021TensorShapeProtosP\001ZS" +
+      "github.com/tensorflow/tensorflow/tensorf" +
+      "low/go/core/framework/tensor_shape_go_pr" +
+      "oto\370\001\001b\006proto3"
+    };
+    descriptor = com.google.protobuf.Descriptors.FileDescriptor
+      .internalBuildGeneratedFileFrom(descriptorData,
+        new com.google.protobuf.Descriptors.FileDescriptor[] {
+        });
+    internal_static_tensorflow_TensorShapeProto_descriptor =
+      getDescriptor().getMessageTypes().get(0);
+    internal_static_tensorflow_TensorShapeProto_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_TensorShapeProto_descriptor,
+        new java.lang.String[] { "Dim", "UnknownRank", });
+    internal_static_tensorflow_TensorShapeProto_Dim_descriptor =
+      internal_static_tensorflow_TensorShapeProto_descriptor.getNestedTypes().get(0);
+    internal_static_tensorflow_TensorShapeProto_Dim_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_TensorShapeProto_Dim_descriptor,
+        new java.lang.String[] { "Size", "Name", });
+  }
+
+  // @@protoc_insertion_point(outer_class_scope)
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorSliceProto.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorSliceProto.java
new file mode 100644
index 00000000000..f1f7b1cf89d
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorSliceProto.java
@@ -0,0 +1,1589 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/framework/tensor_slice.proto
+
+package org.tensorflow.proto.framework;
+
+/**
+ * <pre>
+ * Can only be interpreted if you know the corresponding TensorShape.
+ * </pre>
+ *
+ * Protobuf type {@code tensorflow.TensorSliceProto}
+ */
+public  final class TensorSliceProto extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.TensorSliceProto)
+    TensorSliceProtoOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use TensorSliceProto.newBuilder() to construct.
+  private TensorSliceProto(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private TensorSliceProto() {
+    extent_ = java.util.Collections.emptyList();
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new TensorSliceProto();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private TensorSliceProto(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    int mutable_bitField0_ = 0;
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 10: {
+            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
+              extent_ = new java.util.ArrayList<org.tensorflow.proto.framework.TensorSliceProto.Extent>();
+              mutable_bitField0_ |= 0x00000001;
+            }
+            extent_.add(
+                input.readMessage(org.tensorflow.proto.framework.TensorSliceProto.Extent.parser(), extensionRegistry));
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      if (((mutable_bitField0_ & 0x00000001) != 0)) {
+        extent_ = java.util.Collections.unmodifiableList(extent_);
+      }
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.framework.TensorSliceProtos.internal_static_tensorflow_TensorSliceProto_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.framework.TensorSliceProtos.internal_static_tensorflow_TensorSliceProto_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.framework.TensorSliceProto.class, org.tensorflow.proto.framework.TensorSliceProto.Builder.class);
+  }
+
+  public interface ExtentOrBuilder extends
+      // @@protoc_insertion_point(interface_extends:tensorflow.TensorSliceProto.Extent)
+      com.google.protobuf.MessageOrBuilder {
+
+    /**
+     * <pre>
+     * Start index of the slice, starting at 0.
+     * </pre>
+     *
+     * <code>int64 start = 1;</code>
+     */
+    long getStart();
+
+    /**
+     * <code>int64 length = 2;</code>
+     */
+    long getLength();
+
+    public org.tensorflow.proto.framework.TensorSliceProto.Extent.HasLengthCase getHasLengthCase();
+  }
+  /**
+   * <pre>
+   * Extent of the slice in one dimension.
+   * </pre>
+   *
+   * Protobuf type {@code tensorflow.TensorSliceProto.Extent}
+   */
+  public  static final class Extent extends
+      com.google.protobuf.GeneratedMessageV3 implements
+      // @@protoc_insertion_point(message_implements:tensorflow.TensorSliceProto.Extent)
+      ExtentOrBuilder {
+  private static final long serialVersionUID = 0L;
+    // Use Extent.newBuilder() to construct.
+    private Extent(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+      super(builder);
+    }
+    private Extent() {
+    }
+
+    @java.lang.Override
+    @SuppressWarnings({"unused"})
+    protected java.lang.Object newInstance(
+        UnusedPrivateParameter unused) {
+      return new Extent();
+    }
+
+    @java.lang.Override
+    public final com.google.protobuf.UnknownFieldSet
+    getUnknownFields() {
+      return this.unknownFields;
+    }
+    private Extent(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      this();
+      if (extensionRegistry == null) {
+        throw new java.lang.NullPointerException();
+      }
+      com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+          com.google.protobuf.UnknownFieldSet.newBuilder();
+      try {
+        boolean done = false;
+        while (!done) {
+          int tag = input.readTag();
+          switch (tag) {
+            case 0:
+              done = true;
+              break;
+            case 8: {
+
+              start_ = input.readInt64();
+              break;
+            }
+            case 16: {
+              hasLengthCase_ = 2;
+              hasLength_ = input.readInt64();
+              break;
+            }
+            default: {
+              if (!parseUnknownField(
+                  input, unknownFields, extensionRegistry, tag)) {
+                done = true;
+              }
+              break;
+            }
+          }
+        }
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        throw e.setUnfinishedMessage(this);
+      } catch (java.io.IOException e) {
+        throw new com.google.protobuf.InvalidProtocolBufferException(
+            e).setUnfinishedMessage(this);
+      } finally {
+        this.unknownFields = unknownFields.build();
+        makeExtensionsImmutable();
+      }
+    }
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.TensorSliceProtos.internal_static_tensorflow_TensorSliceProto_Extent_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.TensorSliceProtos.internal_static_tensorflow_TensorSliceProto_Extent_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.TensorSliceProto.Extent.class, org.tensorflow.proto.framework.TensorSliceProto.Extent.Builder.class);
+    }
+
+    private int hasLengthCase_ = 0;
+    private java.lang.Object hasLength_;
+    public enum HasLengthCase
+        implements com.google.protobuf.Internal.EnumLite {
+      LENGTH(2),
+      HASLENGTH_NOT_SET(0);
+      private final int value;
+      private HasLengthCase(int value) {
+        this.value = value;
+      }
+      /**
+       * @deprecated Use {@link #forNumber(int)} instead.
+       */
+      @java.lang.Deprecated
+      public static HasLengthCase valueOf(int value) {
+        return forNumber(value);
+      }
+
+      public static HasLengthCase forNumber(int value) {
+        switch (value) {
+          case 2: return LENGTH;
+          case 0: return HASLENGTH_NOT_SET;
+          default: return null;
+        }
+      }
+      public int getNumber() {
+        return this.value;
+      }
+    };
+
+    public HasLengthCase
+    getHasLengthCase() {
+      return HasLengthCase.forNumber(
+          hasLengthCase_);
+    }
+
+    public static final int START_FIELD_NUMBER = 1;
+    private long start_;
+    /**
+     * <pre>
+     * Start index of the slice, starting at 0.
+     * </pre>
+     *
+     * <code>int64 start = 1;</code>
+     */
+    public long getStart() {
+      return start_;
+    }
+
+    public static final int LENGTH_FIELD_NUMBER = 2;
+    /**
+     * <code>int64 length = 2;</code>
+     */
+    public long getLength() {
+      if (hasLengthCase_ == 2) {
+        return (java.lang.Long) hasLength_;
+      }
+      return 0L;
+    }
+
+    private byte memoizedIsInitialized = -1;
+    @java.lang.Override
+    public final boolean isInitialized() {
+      byte isInitialized = memoizedIsInitialized;
+      if (isInitialized == 1) return true;
+      if (isInitialized == 0) return false;
+
+      memoizedIsInitialized = 1;
+      return true;
+    }
+
+    @java.lang.Override
+    public void writeTo(com.google.protobuf.CodedOutputStream output)
+                        throws java.io.IOException {
+      if (start_ != 0L) {
+        output.writeInt64(1, start_);
+      }
+      if (hasLengthCase_ == 2) {
+        output.writeInt64(
+            2, (long)((java.lang.Long) hasLength_));
+      }
+      unknownFields.writeTo(output);
+    }
+
+    @java.lang.Override
+    public int getSerializedSize() {
+      int size = memoizedSize;
+      if (size != -1) return size;
+
+      size = 0;
+      if (start_ != 0L) {
+        size += com.google.protobuf.CodedOutputStream
+          .computeInt64Size(1, start_);
+      }
+      if (hasLengthCase_ == 2) {
+        size += com.google.protobuf.CodedOutputStream
+          .computeInt64Size(
+              2, (long)((java.lang.Long) hasLength_));
+      }
+      size += unknownFields.getSerializedSize();
+      memoizedSize = size;
+      return size;
+    }
+
+    @java.lang.Override
+    public boolean equals(final java.lang.Object obj) {
+      if (obj == this) {
+       return true;
+      }
+      if (!(obj instanceof org.tensorflow.proto.framework.TensorSliceProto.Extent)) {
+        return super.equals(obj);
+      }
+      org.tensorflow.proto.framework.TensorSliceProto.Extent other = (org.tensorflow.proto.framework.TensorSliceProto.Extent) obj;
+
+      if (getStart()
+          != other.getStart()) return false;
+      if (!getHasLengthCase().equals(other.getHasLengthCase())) return false;
+      switch (hasLengthCase_) {
+        case 2:
+          if (getLength()
+              != other.getLength()) return false;
+          break;
+        case 0:
+        default:
+      }
+      if (!unknownFields.equals(other.unknownFields)) return false;
+      return true;
+    }
+
+    @java.lang.Override
+    public int hashCode() {
+      if (memoizedHashCode != 0) {
+        return memoizedHashCode;
+      }
+      int hash = 41;
+      hash = (19 * hash) + getDescriptor().hashCode();
+      hash = (37 * hash) + START_FIELD_NUMBER;
+      hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+          getStart());
+      switch (hasLengthCase_) {
+        case 2:
+          hash = (37 * hash) + LENGTH_FIELD_NUMBER;
+          hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+              getLength());
+          break;
+        case 0:
+        default:
+      }
+      hash = (29 * hash) + unknownFields.hashCode();
+      memoizedHashCode = hash;
+      return hash;
+    }
+
+    public static org.tensorflow.proto.framework.TensorSliceProto.Extent parseFrom(
+        java.nio.ByteBuffer data)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data);
+    }
+    public static org.tensorflow.proto.framework.TensorSliceProto.Extent parseFrom(
+        java.nio.ByteBuffer data,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.TensorSliceProto.Extent parseFrom(
+        com.google.protobuf.ByteString data)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data);
+    }
+    public static org.tensorflow.proto.framework.TensorSliceProto.Extent parseFrom(
+        com.google.protobuf.ByteString data,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.TensorSliceProto.Extent parseFrom(byte[] data)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data);
+    }
+    public static org.tensorflow.proto.framework.TensorSliceProto.Extent parseFrom(
+        byte[] data,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.TensorSliceProto.Extent parseFrom(java.io.InputStream input)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input);
+    }
+    public static org.tensorflow.proto.framework.TensorSliceProto.Extent parseFrom(
+        java.io.InputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.TensorSliceProto.Extent parseDelimitedFrom(java.io.InputStream input)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseDelimitedWithIOException(PARSER, input);
+    }
+    public static org.tensorflow.proto.framework.TensorSliceProto.Extent parseDelimitedFrom(
+        java.io.InputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.TensorSliceProto.Extent parseFrom(
+        com.google.protobuf.CodedInputStream input)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input);
+    }
+    public static org.tensorflow.proto.framework.TensorSliceProto.Extent parseFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input, extensionRegistry);
+    }
+
+    @java.lang.Override
+    public Builder newBuilderForType() { return newBuilder(); }
+    public static Builder newBuilder() {
+      return DEFAULT_INSTANCE.toBuilder();
+    }
+    public static Builder newBuilder(org.tensorflow.proto.framework.TensorSliceProto.Extent prototype) {
+      return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+    }
+    @java.lang.Override
+    public Builder toBuilder() {
+      return this == DEFAULT_INSTANCE
+          ? new Builder() : new Builder().mergeFrom(this);
+    }
+
+    @java.lang.Override
+    protected Builder newBuilderForType(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      Builder builder = new Builder(parent);
+      return builder;
+    }
+    /**
+     * <pre>
+     * Extent of the slice in one dimension.
+     * </pre>
+     *
+     * Protobuf type {@code tensorflow.TensorSliceProto.Extent}
+     */
+    public static final class Builder extends
+        com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+        // @@protoc_insertion_point(builder_implements:tensorflow.TensorSliceProto.Extent)
+        org.tensorflow.proto.framework.TensorSliceProto.ExtentOrBuilder {
+      public static final com.google.protobuf.Descriptors.Descriptor
+          getDescriptor() {
+        return org.tensorflow.proto.framework.TensorSliceProtos.internal_static_tensorflow_TensorSliceProto_Extent_descriptor;
+      }
+
+      @java.lang.Override
+      protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+          internalGetFieldAccessorTable() {
+        return org.tensorflow.proto.framework.TensorSliceProtos.internal_static_tensorflow_TensorSliceProto_Extent_fieldAccessorTable
+            .ensureFieldAccessorsInitialized(
+                org.tensorflow.proto.framework.TensorSliceProto.Extent.class, org.tensorflow.proto.framework.TensorSliceProto.Extent.Builder.class);
+      }
+
+      // Construct using org.tensorflow.proto.framework.TensorSliceProto.Extent.newBuilder()
+      private Builder() {
+        maybeForceBuilderInitialization();
+      }
+
+      private Builder(
+          com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+        super(parent);
+        maybeForceBuilderInitialization();
+      }
+      private void maybeForceBuilderInitialization() {
+        if (com.google.protobuf.GeneratedMessageV3
+                .alwaysUseFieldBuilders) {
+        }
+      }
+      @java.lang.Override
+      public Builder clear() {
+        super.clear();
+        start_ = 0L;
+
+        hasLengthCase_ = 0;
+        hasLength_ = null;
+        return this;
+      }
+
+      @java.lang.Override
+      public com.google.protobuf.Descriptors.Descriptor
+          getDescriptorForType() {
+        return org.tensorflow.proto.framework.TensorSliceProtos.internal_static_tensorflow_TensorSliceProto_Extent_descriptor;
+      }
+
+      @java.lang.Override
+      public org.tensorflow.proto.framework.TensorSliceProto.Extent getDefaultInstanceForType() {
+        return org.tensorflow.proto.framework.TensorSliceProto.Extent.getDefaultInstance();
+      }
+
+      @java.lang.Override
+      public org.tensorflow.proto.framework.TensorSliceProto.Extent build() {
+        org.tensorflow.proto.framework.TensorSliceProto.Extent result = buildPartial();
+        if (!result.isInitialized()) {
+          throw newUninitializedMessageException(result);
+        }
+        return result;
+      }
+
+      @java.lang.Override
+      public org.tensorflow.proto.framework.TensorSliceProto.Extent buildPartial() {
+        org.tensorflow.proto.framework.TensorSliceProto.Extent result = new org.tensorflow.proto.framework.TensorSliceProto.Extent(this);
+        result.start_ = start_;
+        if (hasLengthCase_ == 2) {
+          result.hasLength_ = hasLength_;
+        }
+        result.hasLengthCase_ = hasLengthCase_;
+        onBuilt();
+        return result;
+      }
+
+      @java.lang.Override
+      public Builder clone() {
+        return super.clone();
+      }
+      @java.lang.Override
+      public Builder setField(
+          com.google.protobuf.Descriptors.FieldDescriptor field,
+          java.lang.Object value) {
+        return super.setField(field, value);
+      }
+      @java.lang.Override
+      public Builder clearField(
+          com.google.protobuf.Descriptors.FieldDescriptor field) {
+        return super.clearField(field);
+      }
+      @java.lang.Override
+      public Builder clearOneof(
+          com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+        return super.clearOneof(oneof);
+      }
+      @java.lang.Override
+      public Builder setRepeatedField(
+          com.google.protobuf.Descriptors.FieldDescriptor field,
+          int index, java.lang.Object value) {
+        return super.setRepeatedField(field, index, value);
+      }
+      @java.lang.Override
+      public Builder addRepeatedField(
+          com.google.protobuf.Descriptors.FieldDescriptor field,
+          java.lang.Object value) {
+        return super.addRepeatedField(field, value);
+      }
+      @java.lang.Override
+      public Builder mergeFrom(com.google.protobuf.Message other) {
+        if (other instanceof org.tensorflow.proto.framework.TensorSliceProto.Extent) {
+          return mergeFrom((org.tensorflow.proto.framework.TensorSliceProto.Extent)other);
+        } else {
+          super.mergeFrom(other);
+          return this;
+        }
+      }
+
+      public Builder mergeFrom(org.tensorflow.proto.framework.TensorSliceProto.Extent other) {
+        if (other == org.tensorflow.proto.framework.TensorSliceProto.Extent.getDefaultInstance()) return this;
+        if (other.getStart() != 0L) {
+          setStart(other.getStart());
+        }
+        switch (other.getHasLengthCase()) {
+          case LENGTH: {
+            setLength(other.getLength());
+            break;
+          }
+          case HASLENGTH_NOT_SET: {
+            break;
+          }
+        }
+        this.mergeUnknownFields(other.unknownFields);
+        onChanged();
+        return this;
+      }
+
+      @java.lang.Override
+      public final boolean isInitialized() {
+        return true;
+      }
+
+      @java.lang.Override
+      public Builder mergeFrom(
+          com.google.protobuf.CodedInputStream input,
+          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+          throws java.io.IOException {
+        org.tensorflow.proto.framework.TensorSliceProto.Extent parsedMessage = null;
+        try {
+          parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+        } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+          parsedMessage = (org.tensorflow.proto.framework.TensorSliceProto.Extent) e.getUnfinishedMessage();
+          throw e.unwrapIOException();
+        } finally {
+          if (parsedMessage != null) {
+            mergeFrom(parsedMessage);
+          }
+        }
+        return this;
+      }
+      private int hasLengthCase_ = 0;
+      private java.lang.Object hasLength_;
+      public HasLengthCase
+          getHasLengthCase() {
+        return HasLengthCase.forNumber(
+            hasLengthCase_);
+      }
+
+      public Builder clearHasLength() {
+        hasLengthCase_ = 0;
+        hasLength_ = null;
+        onChanged();
+        return this;
+      }
+
+
+      private long start_ ;
+      /**
+       * <pre>
+       * Start index of the slice, starting at 0.
+       * </pre>
+       *
+       * <code>int64 start = 1;</code>
+       */
+      public long getStart() {
+        return start_;
+      }
+      /**
+       * <pre>
+       * Start index of the slice, starting at 0.
+       * </pre>
+       *
+       * <code>int64 start = 1;</code>
+       */
+      public Builder setStart(long value) {
+        
+        start_ = value;
+        onChanged();
+        return this;
+      }
+      /**
+       * <pre>
+       * Start index of the slice, starting at 0.
+       * </pre>
+       *
+       * <code>int64 start = 1;</code>
+       */
+      public Builder clearStart() {
+        
+        start_ = 0L;
+        onChanged();
+        return this;
+      }
+
+      /**
+       * <code>int64 length = 2;</code>
+       */
+      public long getLength() {
+        if (hasLengthCase_ == 2) {
+          return (java.lang.Long) hasLength_;
+        }
+        return 0L;
+      }
+      /**
+       * <code>int64 length = 2;</code>
+       */
+      public Builder setLength(long value) {
+        hasLengthCase_ = 2;
+        hasLength_ = value;
+        onChanged();
+        return this;
+      }
+      /**
+       * <code>int64 length = 2;</code>
+       */
+      public Builder clearLength() {
+        if (hasLengthCase_ == 2) {
+          hasLengthCase_ = 0;
+          hasLength_ = null;
+          onChanged();
+        }
+        return this;
+      }
+      @java.lang.Override
+      public final Builder setUnknownFields(
+          final com.google.protobuf.UnknownFieldSet unknownFields) {
+        return super.setUnknownFields(unknownFields);
+      }
+
+      @java.lang.Override
+      public final Builder mergeUnknownFields(
+          final com.google.protobuf.UnknownFieldSet unknownFields) {
+        return super.mergeUnknownFields(unknownFields);
+      }
+
+
+      // @@protoc_insertion_point(builder_scope:tensorflow.TensorSliceProto.Extent)
+    }
+
+    // @@protoc_insertion_point(class_scope:tensorflow.TensorSliceProto.Extent)
+    private static final org.tensorflow.proto.framework.TensorSliceProto.Extent DEFAULT_INSTANCE;
+    static {
+      DEFAULT_INSTANCE = new org.tensorflow.proto.framework.TensorSliceProto.Extent();
+    }
+
+    public static org.tensorflow.proto.framework.TensorSliceProto.Extent getDefaultInstance() {
+      return DEFAULT_INSTANCE;
+    }
+
+    private static final com.google.protobuf.Parser<Extent>
+        PARSER = new com.google.protobuf.AbstractParser<Extent>() {
+      @java.lang.Override
+      public Extent parsePartialFrom(
+          com.google.protobuf.CodedInputStream input,
+          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+          throws com.google.protobuf.InvalidProtocolBufferException {
+        return new Extent(input, extensionRegistry);
+      }
+    };
+
+    public static com.google.protobuf.Parser<Extent> parser() {
+      return PARSER;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Parser<Extent> getParserForType() {
+      return PARSER;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.TensorSliceProto.Extent getDefaultInstanceForType() {
+      return DEFAULT_INSTANCE;
+    }
+
+  }
+
+  public static final int EXTENT_FIELD_NUMBER = 1;
+  private java.util.List<org.tensorflow.proto.framework.TensorSliceProto.Extent> extent_;
+  /**
+   * <pre>
+   * Extent of the slice in all tensor dimensions.
+   * Must have one entry for each of the dimension of the tensor that this
+   * slice belongs to.  The order of sizes is the same as the order of
+   * dimensions in the TensorShape.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.TensorSliceProto.Extent extent = 1;</code>
+   */
+  public java.util.List<org.tensorflow.proto.framework.TensorSliceProto.Extent> getExtentList() {
+    return extent_;
+  }
+  /**
+   * <pre>
+   * Extent of the slice in all tensor dimensions.
+   * Must have one entry for each of the dimension of the tensor that this
+   * slice belongs to.  The order of sizes is the same as the order of
+   * dimensions in the TensorShape.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.TensorSliceProto.Extent extent = 1;</code>
+   */
+  public java.util.List<? extends org.tensorflow.proto.framework.TensorSliceProto.ExtentOrBuilder> 
+      getExtentOrBuilderList() {
+    return extent_;
+  }
+  /**
+   * <pre>
+   * Extent of the slice in all tensor dimensions.
+   * Must have one entry for each of the dimension of the tensor that this
+   * slice belongs to.  The order of sizes is the same as the order of
+   * dimensions in the TensorShape.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.TensorSliceProto.Extent extent = 1;</code>
+   */
+  public int getExtentCount() {
+    return extent_.size();
+  }
+  /**
+   * <pre>
+   * Extent of the slice in all tensor dimensions.
+   * Must have one entry for each of the dimension of the tensor that this
+   * slice belongs to.  The order of sizes is the same as the order of
+   * dimensions in the TensorShape.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.TensorSliceProto.Extent extent = 1;</code>
+   */
+  public org.tensorflow.proto.framework.TensorSliceProto.Extent getExtent(int index) {
+    return extent_.get(index);
+  }
+  /**
+   * <pre>
+   * Extent of the slice in all tensor dimensions.
+   * Must have one entry for each of the dimension of the tensor that this
+   * slice belongs to.  The order of sizes is the same as the order of
+   * dimensions in the TensorShape.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.TensorSliceProto.Extent extent = 1;</code>
+   */
+  public org.tensorflow.proto.framework.TensorSliceProto.ExtentOrBuilder getExtentOrBuilder(
+      int index) {
+    return extent_.get(index);
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    for (int i = 0; i < extent_.size(); i++) {
+      output.writeMessage(1, extent_.get(i));
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    for (int i = 0; i < extent_.size(); i++) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(1, extent_.get(i));
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.framework.TensorSliceProto)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.framework.TensorSliceProto other = (org.tensorflow.proto.framework.TensorSliceProto) obj;
+
+    if (!getExtentList()
+        .equals(other.getExtentList())) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    if (getExtentCount() > 0) {
+      hash = (37 * hash) + EXTENT_FIELD_NUMBER;
+      hash = (53 * hash) + getExtentList().hashCode();
+    }
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.framework.TensorSliceProto parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.TensorSliceProto parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.TensorSliceProto parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.TensorSliceProto parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.TensorSliceProto parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.TensorSliceProto parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.TensorSliceProto parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.TensorSliceProto parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.TensorSliceProto parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.TensorSliceProto parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.TensorSliceProto parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.TensorSliceProto parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.framework.TensorSliceProto prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * <pre>
+   * Can only be interpreted if you know the corresponding TensorShape.
+   * </pre>
+   *
+   * Protobuf type {@code tensorflow.TensorSliceProto}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.TensorSliceProto)
+      org.tensorflow.proto.framework.TensorSliceProtoOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.TensorSliceProtos.internal_static_tensorflow_TensorSliceProto_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.TensorSliceProtos.internal_static_tensorflow_TensorSliceProto_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.TensorSliceProto.class, org.tensorflow.proto.framework.TensorSliceProto.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.framework.TensorSliceProto.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+        getExtentFieldBuilder();
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      if (extentBuilder_ == null) {
+        extent_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000001);
+      } else {
+        extentBuilder_.clear();
+      }
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.framework.TensorSliceProtos.internal_static_tensorflow_TensorSliceProto_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.TensorSliceProto getDefaultInstanceForType() {
+      return org.tensorflow.proto.framework.TensorSliceProto.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.TensorSliceProto build() {
+      org.tensorflow.proto.framework.TensorSliceProto result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.TensorSliceProto buildPartial() {
+      org.tensorflow.proto.framework.TensorSliceProto result = new org.tensorflow.proto.framework.TensorSliceProto(this);
+      int from_bitField0_ = bitField0_;
+      if (extentBuilder_ == null) {
+        if (((bitField0_ & 0x00000001) != 0)) {
+          extent_ = java.util.Collections.unmodifiableList(extent_);
+          bitField0_ = (bitField0_ & ~0x00000001);
+        }
+        result.extent_ = extent_;
+      } else {
+        result.extent_ = extentBuilder_.build();
+      }
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.framework.TensorSliceProto) {
+        return mergeFrom((org.tensorflow.proto.framework.TensorSliceProto)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.framework.TensorSliceProto other) {
+      if (other == org.tensorflow.proto.framework.TensorSliceProto.getDefaultInstance()) return this;
+      if (extentBuilder_ == null) {
+        if (!other.extent_.isEmpty()) {
+          if (extent_.isEmpty()) {
+            extent_ = other.extent_;
+            bitField0_ = (bitField0_ & ~0x00000001);
+          } else {
+            ensureExtentIsMutable();
+            extent_.addAll(other.extent_);
+          }
+          onChanged();
+        }
+      } else {
+        if (!other.extent_.isEmpty()) {
+          if (extentBuilder_.isEmpty()) {
+            extentBuilder_.dispose();
+            extentBuilder_ = null;
+            extent_ = other.extent_;
+            bitField0_ = (bitField0_ & ~0x00000001);
+            extentBuilder_ = 
+              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
+                 getExtentFieldBuilder() : null;
+          } else {
+            extentBuilder_.addAllMessages(other.extent_);
+          }
+        }
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.framework.TensorSliceProto parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.framework.TensorSliceProto) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+    private int bitField0_;
+
+    private java.util.List<org.tensorflow.proto.framework.TensorSliceProto.Extent> extent_ =
+      java.util.Collections.emptyList();
+    private void ensureExtentIsMutable() {
+      if (!((bitField0_ & 0x00000001) != 0)) {
+        extent_ = new java.util.ArrayList<org.tensorflow.proto.framework.TensorSliceProto.Extent>(extent_);
+        bitField0_ |= 0x00000001;
+       }
+    }
+
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.framework.TensorSliceProto.Extent, org.tensorflow.proto.framework.TensorSliceProto.Extent.Builder, org.tensorflow.proto.framework.TensorSliceProto.ExtentOrBuilder> extentBuilder_;
+
+    /**
+     * <pre>
+     * Extent of the slice in all tensor dimensions.
+     * Must have one entry for each of the dimension of the tensor that this
+     * slice belongs to.  The order of sizes is the same as the order of
+     * dimensions in the TensorShape.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorSliceProto.Extent extent = 1;</code>
+     */
+    public java.util.List<org.tensorflow.proto.framework.TensorSliceProto.Extent> getExtentList() {
+      if (extentBuilder_ == null) {
+        return java.util.Collections.unmodifiableList(extent_);
+      } else {
+        return extentBuilder_.getMessageList();
+      }
+    }
+    /**
+     * <pre>
+     * Extent of the slice in all tensor dimensions.
+     * Must have one entry for each of the dimension of the tensor that this
+     * slice belongs to.  The order of sizes is the same as the order of
+     * dimensions in the TensorShape.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorSliceProto.Extent extent = 1;</code>
+     */
+    public int getExtentCount() {
+      if (extentBuilder_ == null) {
+        return extent_.size();
+      } else {
+        return extentBuilder_.getCount();
+      }
+    }
+    /**
+     * <pre>
+     * Extent of the slice in all tensor dimensions.
+     * Must have one entry for each of the dimension of the tensor that this
+     * slice belongs to.  The order of sizes is the same as the order of
+     * dimensions in the TensorShape.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorSliceProto.Extent extent = 1;</code>
+     */
+    public org.tensorflow.proto.framework.TensorSliceProto.Extent getExtent(int index) {
+      if (extentBuilder_ == null) {
+        return extent_.get(index);
+      } else {
+        return extentBuilder_.getMessage(index);
+      }
+    }
+    /**
+     * <pre>
+     * Extent of the slice in all tensor dimensions.
+     * Must have one entry for each of the dimension of the tensor that this
+     * slice belongs to.  The order of sizes is the same as the order of
+     * dimensions in the TensorShape.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorSliceProto.Extent extent = 1;</code>
+     */
+    public Builder setExtent(
+        int index, org.tensorflow.proto.framework.TensorSliceProto.Extent value) {
+      if (extentBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureExtentIsMutable();
+        extent_.set(index, value);
+        onChanged();
+      } else {
+        extentBuilder_.setMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Extent of the slice in all tensor dimensions.
+     * Must have one entry for each of the dimension of the tensor that this
+     * slice belongs to.  The order of sizes is the same as the order of
+     * dimensions in the TensorShape.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorSliceProto.Extent extent = 1;</code>
+     */
+    public Builder setExtent(
+        int index, org.tensorflow.proto.framework.TensorSliceProto.Extent.Builder builderForValue) {
+      if (extentBuilder_ == null) {
+        ensureExtentIsMutable();
+        extent_.set(index, builderForValue.build());
+        onChanged();
+      } else {
+        extentBuilder_.setMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Extent of the slice in all tensor dimensions.
+     * Must have one entry for each of the dimension of the tensor that this
+     * slice belongs to.  The order of sizes is the same as the order of
+     * dimensions in the TensorShape.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorSliceProto.Extent extent = 1;</code>
+     */
+    public Builder addExtent(org.tensorflow.proto.framework.TensorSliceProto.Extent value) {
+      if (extentBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureExtentIsMutable();
+        extent_.add(value);
+        onChanged();
+      } else {
+        extentBuilder_.addMessage(value);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Extent of the slice in all tensor dimensions.
+     * Must have one entry for each of the dimension of the tensor that this
+     * slice belongs to.  The order of sizes is the same as the order of
+     * dimensions in the TensorShape.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorSliceProto.Extent extent = 1;</code>
+     */
+    public Builder addExtent(
+        int index, org.tensorflow.proto.framework.TensorSliceProto.Extent value) {
+      if (extentBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureExtentIsMutable();
+        extent_.add(index, value);
+        onChanged();
+      } else {
+        extentBuilder_.addMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Extent of the slice in all tensor dimensions.
+     * Must have one entry for each of the dimension of the tensor that this
+     * slice belongs to.  The order of sizes is the same as the order of
+     * dimensions in the TensorShape.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorSliceProto.Extent extent = 1;</code>
+     */
+    public Builder addExtent(
+        org.tensorflow.proto.framework.TensorSliceProto.Extent.Builder builderForValue) {
+      if (extentBuilder_ == null) {
+        ensureExtentIsMutable();
+        extent_.add(builderForValue.build());
+        onChanged();
+      } else {
+        extentBuilder_.addMessage(builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Extent of the slice in all tensor dimensions.
+     * Must have one entry for each of the dimension of the tensor that this
+     * slice belongs to.  The order of sizes is the same as the order of
+     * dimensions in the TensorShape.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorSliceProto.Extent extent = 1;</code>
+     */
+    public Builder addExtent(
+        int index, org.tensorflow.proto.framework.TensorSliceProto.Extent.Builder builderForValue) {
+      if (extentBuilder_ == null) {
+        ensureExtentIsMutable();
+        extent_.add(index, builderForValue.build());
+        onChanged();
+      } else {
+        extentBuilder_.addMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Extent of the slice in all tensor dimensions.
+     * Must have one entry for each of the dimension of the tensor that this
+     * slice belongs to.  The order of sizes is the same as the order of
+     * dimensions in the TensorShape.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorSliceProto.Extent extent = 1;</code>
+     */
+    public Builder addAllExtent(
+        java.lang.Iterable<? extends org.tensorflow.proto.framework.TensorSliceProto.Extent> values) {
+      if (extentBuilder_ == null) {
+        ensureExtentIsMutable();
+        com.google.protobuf.AbstractMessageLite.Builder.addAll(
+            values, extent_);
+        onChanged();
+      } else {
+        extentBuilder_.addAllMessages(values);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Extent of the slice in all tensor dimensions.
+     * Must have one entry for each of the dimension of the tensor that this
+     * slice belongs to.  The order of sizes is the same as the order of
+     * dimensions in the TensorShape.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorSliceProto.Extent extent = 1;</code>
+     */
+    public Builder clearExtent() {
+      if (extentBuilder_ == null) {
+        extent_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000001);
+        onChanged();
+      } else {
+        extentBuilder_.clear();
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Extent of the slice in all tensor dimensions.
+     * Must have one entry for each of the dimension of the tensor that this
+     * slice belongs to.  The order of sizes is the same as the order of
+     * dimensions in the TensorShape.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorSliceProto.Extent extent = 1;</code>
+     */
+    public Builder removeExtent(int index) {
+      if (extentBuilder_ == null) {
+        ensureExtentIsMutable();
+        extent_.remove(index);
+        onChanged();
+      } else {
+        extentBuilder_.remove(index);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Extent of the slice in all tensor dimensions.
+     * Must have one entry for each of the dimension of the tensor that this
+     * slice belongs to.  The order of sizes is the same as the order of
+     * dimensions in the TensorShape.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorSliceProto.Extent extent = 1;</code>
+     */
+    public org.tensorflow.proto.framework.TensorSliceProto.Extent.Builder getExtentBuilder(
+        int index) {
+      return getExtentFieldBuilder().getBuilder(index);
+    }
+    /**
+     * <pre>
+     * Extent of the slice in all tensor dimensions.
+     * Must have one entry for each of the dimension of the tensor that this
+     * slice belongs to.  The order of sizes is the same as the order of
+     * dimensions in the TensorShape.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorSliceProto.Extent extent = 1;</code>
+     */
+    public org.tensorflow.proto.framework.TensorSliceProto.ExtentOrBuilder getExtentOrBuilder(
+        int index) {
+      if (extentBuilder_ == null) {
+        return extent_.get(index);  } else {
+        return extentBuilder_.getMessageOrBuilder(index);
+      }
+    }
+    /**
+     * <pre>
+     * Extent of the slice in all tensor dimensions.
+     * Must have one entry for each of the dimension of the tensor that this
+     * slice belongs to.  The order of sizes is the same as the order of
+     * dimensions in the TensorShape.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorSliceProto.Extent extent = 1;</code>
+     */
+    public java.util.List<? extends org.tensorflow.proto.framework.TensorSliceProto.ExtentOrBuilder> 
+         getExtentOrBuilderList() {
+      if (extentBuilder_ != null) {
+        return extentBuilder_.getMessageOrBuilderList();
+      } else {
+        return java.util.Collections.unmodifiableList(extent_);
+      }
+    }
+    /**
+     * <pre>
+     * Extent of the slice in all tensor dimensions.
+     * Must have one entry for each of the dimension of the tensor that this
+     * slice belongs to.  The order of sizes is the same as the order of
+     * dimensions in the TensorShape.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorSliceProto.Extent extent = 1;</code>
+     */
+    public org.tensorflow.proto.framework.TensorSliceProto.Extent.Builder addExtentBuilder() {
+      return getExtentFieldBuilder().addBuilder(
+          org.tensorflow.proto.framework.TensorSliceProto.Extent.getDefaultInstance());
+    }
+    /**
+     * <pre>
+     * Extent of the slice in all tensor dimensions.
+     * Must have one entry for each of the dimension of the tensor that this
+     * slice belongs to.  The order of sizes is the same as the order of
+     * dimensions in the TensorShape.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorSliceProto.Extent extent = 1;</code>
+     */
+    public org.tensorflow.proto.framework.TensorSliceProto.Extent.Builder addExtentBuilder(
+        int index) {
+      return getExtentFieldBuilder().addBuilder(
+          index, org.tensorflow.proto.framework.TensorSliceProto.Extent.getDefaultInstance());
+    }
+    /**
+     * <pre>
+     * Extent of the slice in all tensor dimensions.
+     * Must have one entry for each of the dimension of the tensor that this
+     * slice belongs to.  The order of sizes is the same as the order of
+     * dimensions in the TensorShape.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorSliceProto.Extent extent = 1;</code>
+     */
+    public java.util.List<org.tensorflow.proto.framework.TensorSliceProto.Extent.Builder> 
+         getExtentBuilderList() {
+      return getExtentFieldBuilder().getBuilderList();
+    }
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.framework.TensorSliceProto.Extent, org.tensorflow.proto.framework.TensorSliceProto.Extent.Builder, org.tensorflow.proto.framework.TensorSliceProto.ExtentOrBuilder> 
+        getExtentFieldBuilder() {
+      if (extentBuilder_ == null) {
+        extentBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
+            org.tensorflow.proto.framework.TensorSliceProto.Extent, org.tensorflow.proto.framework.TensorSliceProto.Extent.Builder, org.tensorflow.proto.framework.TensorSliceProto.ExtentOrBuilder>(
+                extent_,
+                ((bitField0_ & 0x00000001) != 0),
+                getParentForChildren(),
+                isClean());
+        extent_ = null;
+      }
+      return extentBuilder_;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.TensorSliceProto)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.TensorSliceProto)
+  private static final org.tensorflow.proto.framework.TensorSliceProto DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.TensorSliceProto();
+  }
+
+  public static org.tensorflow.proto.framework.TensorSliceProto getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<TensorSliceProto>
+      PARSER = new com.google.protobuf.AbstractParser<TensorSliceProto>() {
+    @java.lang.Override
+    public TensorSliceProto parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new TensorSliceProto(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<TensorSliceProto> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<TensorSliceProto> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.framework.TensorSliceProto getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorSliceProtoOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorSliceProtoOrBuilder.java
new file mode 100644
index 00000000000..187cbc4962f
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorSliceProtoOrBuilder.java
@@ -0,0 +1,68 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/framework/tensor_slice.proto
+
+package org.tensorflow.proto.framework;
+
+public interface TensorSliceProtoOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.TensorSliceProto)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <pre>
+   * Extent of the slice in all tensor dimensions.
+   * Must have one entry for each of the dimension of the tensor that this
+   * slice belongs to.  The order of sizes is the same as the order of
+   * dimensions in the TensorShape.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.TensorSliceProto.Extent extent = 1;</code>
+   */
+  java.util.List<org.tensorflow.proto.framework.TensorSliceProto.Extent> 
+      getExtentList();
+  /**
+   * <pre>
+   * Extent of the slice in all tensor dimensions.
+   * Must have one entry for each of the dimension of the tensor that this
+   * slice belongs to.  The order of sizes is the same as the order of
+   * dimensions in the TensorShape.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.TensorSliceProto.Extent extent = 1;</code>
+   */
+  org.tensorflow.proto.framework.TensorSliceProto.Extent getExtent(int index);
+  /**
+   * <pre>
+   * Extent of the slice in all tensor dimensions.
+   * Must have one entry for each of the dimension of the tensor that this
+   * slice belongs to.  The order of sizes is the same as the order of
+   * dimensions in the TensorShape.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.TensorSliceProto.Extent extent = 1;</code>
+   */
+  int getExtentCount();
+  /**
+   * <pre>
+   * Extent of the slice in all tensor dimensions.
+   * Must have one entry for each of the dimension of the tensor that this
+   * slice belongs to.  The order of sizes is the same as the order of
+   * dimensions in the TensorShape.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.TensorSliceProto.Extent extent = 1;</code>
+   */
+  java.util.List<? extends org.tensorflow.proto.framework.TensorSliceProto.ExtentOrBuilder> 
+      getExtentOrBuilderList();
+  /**
+   * <pre>
+   * Extent of the slice in all tensor dimensions.
+   * Must have one entry for each of the dimension of the tensor that this
+   * slice belongs to.  The order of sizes is the same as the order of
+   * dimensions in the TensorShape.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.TensorSliceProto.Extent extent = 1;</code>
+   */
+  org.tensorflow.proto.framework.TensorSliceProto.ExtentOrBuilder getExtentOrBuilder(
+      int index);
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorSliceProtos.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorSliceProtos.java
new file mode 100644
index 00000000000..8a3b4497177
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorSliceProtos.java
@@ -0,0 +1,65 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/framework/tensor_slice.proto
+
+package org.tensorflow.proto.framework;
+
+public final class TensorSliceProtos {
+  private TensorSliceProtos() {}
+  public static void registerAllExtensions(
+      com.google.protobuf.ExtensionRegistryLite registry) {
+  }
+
+  public static void registerAllExtensions(
+      com.google.protobuf.ExtensionRegistry registry) {
+    registerAllExtensions(
+        (com.google.protobuf.ExtensionRegistryLite) registry);
+  }
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_TensorSliceProto_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_TensorSliceProto_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_TensorSliceProto_Extent_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_TensorSliceProto_Extent_fieldAccessorTable;
+
+  public static com.google.protobuf.Descriptors.FileDescriptor
+      getDescriptor() {
+    return descriptor;
+  }
+  private static  com.google.protobuf.Descriptors.FileDescriptor
+      descriptor;
+  static {
+    java.lang.String[] descriptorData = {
+      "\n,tensorflow/core/framework/tensor_slice" +
+      ".proto\022\ntensorflow\"\200\001\n\020TensorSliceProto\022" +
+      "3\n\006extent\030\001 \003(\0132#.tensorflow.TensorSlice" +
+      "Proto.Extent\0327\n\006Extent\022\r\n\005start\030\001 \001(\003\022\020\n" +
+      "\006length\030\002 \001(\003H\000B\014\n\nhas_lengthB\215\001\n\036org.te" +
+      "nsorflow.proto.frameworkB\021TensorSlicePro" +
+      "tosP\001ZSgithub.com/tensorflow/tensorflow/" +
+      "tensorflow/go/core/framework/tensor_slic" +
+      "e_go_proto\370\001\001b\006proto3"
+    };
+    descriptor = com.google.protobuf.Descriptors.FileDescriptor
+      .internalBuildGeneratedFileFrom(descriptorData,
+        new com.google.protobuf.Descriptors.FileDescriptor[] {
+        });
+    internal_static_tensorflow_TensorSliceProto_descriptor =
+      getDescriptor().getMessageTypes().get(0);
+    internal_static_tensorflow_TensorSliceProto_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_TensorSliceProto_descriptor,
+        new java.lang.String[] { "Extent", });
+    internal_static_tensorflow_TensorSliceProto_Extent_descriptor =
+      internal_static_tensorflow_TensorSliceProto_descriptor.getNestedTypes().get(0);
+    internal_static_tensorflow_TensorSliceProto_Extent_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_TensorSliceProto_Extent_descriptor,
+        new java.lang.String[] { "Start", "Length", "HasLength", });
+  }
+
+  // @@protoc_insertion_point(outer_class_scope)
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorSpecProto.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorSpecProto.java
new file mode 100644
index 00000000000..1dadc728091
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorSpecProto.java
@@ -0,0 +1,820 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/struct.proto
+
+package org.tensorflow.proto.framework;
+
+/**
+ * <pre>
+ * A protobuf to represent tf.TensorSpec.
+ * </pre>
+ *
+ * Protobuf type {@code tensorflow.TensorSpecProto}
+ */
+public  final class TensorSpecProto extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.TensorSpecProto)
+    TensorSpecProtoOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use TensorSpecProto.newBuilder() to construct.
+  private TensorSpecProto(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private TensorSpecProto() {
+    name_ = "";
+    dtype_ = 0;
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new TensorSpecProto();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private TensorSpecProto(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 10: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            name_ = s;
+            break;
+          }
+          case 18: {
+            org.tensorflow.proto.framework.TensorShapeProto.Builder subBuilder = null;
+            if (shape_ != null) {
+              subBuilder = shape_.toBuilder();
+            }
+            shape_ = input.readMessage(org.tensorflow.proto.framework.TensorShapeProto.parser(), extensionRegistry);
+            if (subBuilder != null) {
+              subBuilder.mergeFrom(shape_);
+              shape_ = subBuilder.buildPartial();
+            }
+
+            break;
+          }
+          case 24: {
+            int rawValue = input.readEnum();
+
+            dtype_ = rawValue;
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.framework.StructProtos.internal_static_tensorflow_TensorSpecProto_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.framework.StructProtos.internal_static_tensorflow_TensorSpecProto_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.framework.TensorSpecProto.class, org.tensorflow.proto.framework.TensorSpecProto.Builder.class);
+  }
+
+  public static final int NAME_FIELD_NUMBER = 1;
+  private volatile java.lang.Object name_;
+  /**
+   * <code>string name = 1;</code>
+   */
+  public java.lang.String getName() {
+    java.lang.Object ref = name_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      name_ = s;
+      return s;
+    }
+  }
+  /**
+   * <code>string name = 1;</code>
+   */
+  public com.google.protobuf.ByteString
+      getNameBytes() {
+    java.lang.Object ref = name_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      name_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int SHAPE_FIELD_NUMBER = 2;
+  private org.tensorflow.proto.framework.TensorShapeProto shape_;
+  /**
+   * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+   */
+  public boolean hasShape() {
+    return shape_ != null;
+  }
+  /**
+   * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+   */
+  public org.tensorflow.proto.framework.TensorShapeProto getShape() {
+    return shape_ == null ? org.tensorflow.proto.framework.TensorShapeProto.getDefaultInstance() : shape_;
+  }
+  /**
+   * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+   */
+  public org.tensorflow.proto.framework.TensorShapeProtoOrBuilder getShapeOrBuilder() {
+    return getShape();
+  }
+
+  public static final int DTYPE_FIELD_NUMBER = 3;
+  private int dtype_;
+  /**
+   * <code>.tensorflow.DataType dtype = 3;</code>
+   */
+  public int getDtypeValue() {
+    return dtype_;
+  }
+  /**
+   * <code>.tensorflow.DataType dtype = 3;</code>
+   */
+  public org.tensorflow.proto.framework.DataType getDtype() {
+    @SuppressWarnings("deprecation")
+    org.tensorflow.proto.framework.DataType result = org.tensorflow.proto.framework.DataType.valueOf(dtype_);
+    return result == null ? org.tensorflow.proto.framework.DataType.UNRECOGNIZED : result;
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    if (!getNameBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 1, name_);
+    }
+    if (shape_ != null) {
+      output.writeMessage(2, getShape());
+    }
+    if (dtype_ != org.tensorflow.proto.framework.DataType.DT_INVALID.getNumber()) {
+      output.writeEnum(3, dtype_);
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (!getNameBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, name_);
+    }
+    if (shape_ != null) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(2, getShape());
+    }
+    if (dtype_ != org.tensorflow.proto.framework.DataType.DT_INVALID.getNumber()) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeEnumSize(3, dtype_);
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.framework.TensorSpecProto)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.framework.TensorSpecProto other = (org.tensorflow.proto.framework.TensorSpecProto) obj;
+
+    if (!getName()
+        .equals(other.getName())) return false;
+    if (hasShape() != other.hasShape()) return false;
+    if (hasShape()) {
+      if (!getShape()
+          .equals(other.getShape())) return false;
+    }
+    if (dtype_ != other.dtype_) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    hash = (37 * hash) + NAME_FIELD_NUMBER;
+    hash = (53 * hash) + getName().hashCode();
+    if (hasShape()) {
+      hash = (37 * hash) + SHAPE_FIELD_NUMBER;
+      hash = (53 * hash) + getShape().hashCode();
+    }
+    hash = (37 * hash) + DTYPE_FIELD_NUMBER;
+    hash = (53 * hash) + dtype_;
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.framework.TensorSpecProto parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.TensorSpecProto parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.TensorSpecProto parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.TensorSpecProto parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.TensorSpecProto parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.TensorSpecProto parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.TensorSpecProto parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.TensorSpecProto parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.TensorSpecProto parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.TensorSpecProto parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.TensorSpecProto parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.TensorSpecProto parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.framework.TensorSpecProto prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * <pre>
+   * A protobuf to represent tf.TensorSpec.
+   * </pre>
+   *
+   * Protobuf type {@code tensorflow.TensorSpecProto}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.TensorSpecProto)
+      org.tensorflow.proto.framework.TensorSpecProtoOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.StructProtos.internal_static_tensorflow_TensorSpecProto_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.StructProtos.internal_static_tensorflow_TensorSpecProto_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.TensorSpecProto.class, org.tensorflow.proto.framework.TensorSpecProto.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.framework.TensorSpecProto.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      name_ = "";
+
+      if (shapeBuilder_ == null) {
+        shape_ = null;
+      } else {
+        shape_ = null;
+        shapeBuilder_ = null;
+      }
+      dtype_ = 0;
+
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.framework.StructProtos.internal_static_tensorflow_TensorSpecProto_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.TensorSpecProto getDefaultInstanceForType() {
+      return org.tensorflow.proto.framework.TensorSpecProto.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.TensorSpecProto build() {
+      org.tensorflow.proto.framework.TensorSpecProto result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.TensorSpecProto buildPartial() {
+      org.tensorflow.proto.framework.TensorSpecProto result = new org.tensorflow.proto.framework.TensorSpecProto(this);
+      result.name_ = name_;
+      if (shapeBuilder_ == null) {
+        result.shape_ = shape_;
+      } else {
+        result.shape_ = shapeBuilder_.build();
+      }
+      result.dtype_ = dtype_;
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.framework.TensorSpecProto) {
+        return mergeFrom((org.tensorflow.proto.framework.TensorSpecProto)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.framework.TensorSpecProto other) {
+      if (other == org.tensorflow.proto.framework.TensorSpecProto.getDefaultInstance()) return this;
+      if (!other.getName().isEmpty()) {
+        name_ = other.name_;
+        onChanged();
+      }
+      if (other.hasShape()) {
+        mergeShape(other.getShape());
+      }
+      if (other.dtype_ != 0) {
+        setDtypeValue(other.getDtypeValue());
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.framework.TensorSpecProto parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.framework.TensorSpecProto) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+
+    private java.lang.Object name_ = "";
+    /**
+     * <code>string name = 1;</code>
+     */
+    public java.lang.String getName() {
+      java.lang.Object ref = name_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        name_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <code>string name = 1;</code>
+     */
+    public com.google.protobuf.ByteString
+        getNameBytes() {
+      java.lang.Object ref = name_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        name_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <code>string name = 1;</code>
+     */
+    public Builder setName(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      name_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>string name = 1;</code>
+     */
+    public Builder clearName() {
+      
+      name_ = getDefaultInstance().getName();
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>string name = 1;</code>
+     */
+    public Builder setNameBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      name_ = value;
+      onChanged();
+      return this;
+    }
+
+    private org.tensorflow.proto.framework.TensorShapeProto shape_;
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.TensorShapeProto, org.tensorflow.proto.framework.TensorShapeProto.Builder, org.tensorflow.proto.framework.TensorShapeProtoOrBuilder> shapeBuilder_;
+    /**
+     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+     */
+    public boolean hasShape() {
+      return shapeBuilder_ != null || shape_ != null;
+    }
+    /**
+     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+     */
+    public org.tensorflow.proto.framework.TensorShapeProto getShape() {
+      if (shapeBuilder_ == null) {
+        return shape_ == null ? org.tensorflow.proto.framework.TensorShapeProto.getDefaultInstance() : shape_;
+      } else {
+        return shapeBuilder_.getMessage();
+      }
+    }
+    /**
+     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+     */
+    public Builder setShape(org.tensorflow.proto.framework.TensorShapeProto value) {
+      if (shapeBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        shape_ = value;
+        onChanged();
+      } else {
+        shapeBuilder_.setMessage(value);
+      }
+
+      return this;
+    }
+    /**
+     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+     */
+    public Builder setShape(
+        org.tensorflow.proto.framework.TensorShapeProto.Builder builderForValue) {
+      if (shapeBuilder_ == null) {
+        shape_ = builderForValue.build();
+        onChanged();
+      } else {
+        shapeBuilder_.setMessage(builderForValue.build());
+      }
+
+      return this;
+    }
+    /**
+     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+     */
+    public Builder mergeShape(org.tensorflow.proto.framework.TensorShapeProto value) {
+      if (shapeBuilder_ == null) {
+        if (shape_ != null) {
+          shape_ =
+            org.tensorflow.proto.framework.TensorShapeProto.newBuilder(shape_).mergeFrom(value).buildPartial();
+        } else {
+          shape_ = value;
+        }
+        onChanged();
+      } else {
+        shapeBuilder_.mergeFrom(value);
+      }
+
+      return this;
+    }
+    /**
+     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+     */
+    public Builder clearShape() {
+      if (shapeBuilder_ == null) {
+        shape_ = null;
+        onChanged();
+      } else {
+        shape_ = null;
+        shapeBuilder_ = null;
+      }
+
+      return this;
+    }
+    /**
+     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+     */
+    public org.tensorflow.proto.framework.TensorShapeProto.Builder getShapeBuilder() {
+      
+      onChanged();
+      return getShapeFieldBuilder().getBuilder();
+    }
+    /**
+     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+     */
+    public org.tensorflow.proto.framework.TensorShapeProtoOrBuilder getShapeOrBuilder() {
+      if (shapeBuilder_ != null) {
+        return shapeBuilder_.getMessageOrBuilder();
+      } else {
+        return shape_ == null ?
+            org.tensorflow.proto.framework.TensorShapeProto.getDefaultInstance() : shape_;
+      }
+    }
+    /**
+     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+     */
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.TensorShapeProto, org.tensorflow.proto.framework.TensorShapeProto.Builder, org.tensorflow.proto.framework.TensorShapeProtoOrBuilder> 
+        getShapeFieldBuilder() {
+      if (shapeBuilder_ == null) {
+        shapeBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+            org.tensorflow.proto.framework.TensorShapeProto, org.tensorflow.proto.framework.TensorShapeProto.Builder, org.tensorflow.proto.framework.TensorShapeProtoOrBuilder>(
+                getShape(),
+                getParentForChildren(),
+                isClean());
+        shape_ = null;
+      }
+      return shapeBuilder_;
+    }
+
+    private int dtype_ = 0;
+    /**
+     * <code>.tensorflow.DataType dtype = 3;</code>
+     */
+    public int getDtypeValue() {
+      return dtype_;
+    }
+    /**
+     * <code>.tensorflow.DataType dtype = 3;</code>
+     */
+    public Builder setDtypeValue(int value) {
+      dtype_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>.tensorflow.DataType dtype = 3;</code>
+     */
+    public org.tensorflow.proto.framework.DataType getDtype() {
+      @SuppressWarnings("deprecation")
+      org.tensorflow.proto.framework.DataType result = org.tensorflow.proto.framework.DataType.valueOf(dtype_);
+      return result == null ? org.tensorflow.proto.framework.DataType.UNRECOGNIZED : result;
+    }
+    /**
+     * <code>.tensorflow.DataType dtype = 3;</code>
+     */
+    public Builder setDtype(org.tensorflow.proto.framework.DataType value) {
+      if (value == null) {
+        throw new NullPointerException();
+      }
+      
+      dtype_ = value.getNumber();
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>.tensorflow.DataType dtype = 3;</code>
+     */
+    public Builder clearDtype() {
+      
+      dtype_ = 0;
+      onChanged();
+      return this;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.TensorSpecProto)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.TensorSpecProto)
+  private static final org.tensorflow.proto.framework.TensorSpecProto DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.TensorSpecProto();
+  }
+
+  public static org.tensorflow.proto.framework.TensorSpecProto getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<TensorSpecProto>
+      PARSER = new com.google.protobuf.AbstractParser<TensorSpecProto>() {
+    @java.lang.Override
+    public TensorSpecProto parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new TensorSpecProto(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<TensorSpecProto> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<TensorSpecProto> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.framework.TensorSpecProto getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorSpecProtoOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorSpecProtoOrBuilder.java
new file mode 100644
index 00000000000..f9f8ec12d9d
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TensorSpecProtoOrBuilder.java
@@ -0,0 +1,41 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/struct.proto
+
+package org.tensorflow.proto.framework;
+
+public interface TensorSpecProtoOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.TensorSpecProto)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <code>string name = 1;</code>
+   */
+  java.lang.String getName();
+  /**
+   * <code>string name = 1;</code>
+   */
+  com.google.protobuf.ByteString
+      getNameBytes();
+
+  /**
+   * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+   */
+  boolean hasShape();
+  /**
+   * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+   */
+  org.tensorflow.proto.framework.TensorShapeProto getShape();
+  /**
+   * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+   */
+  org.tensorflow.proto.framework.TensorShapeProtoOrBuilder getShapeOrBuilder();
+
+  /**
+   * <code>.tensorflow.DataType dtype = 3;</code>
+   */
+  int getDtypeValue();
+  /**
+   * <code>.tensorflow.DataType dtype = 3;</code>
+   */
+  org.tensorflow.proto.framework.DataType getDtype();
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ThreadPoolOptionProto.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ThreadPoolOptionProto.java
new file mode 100644
index 00000000000..9894b24788e
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ThreadPoolOptionProto.java
@@ -0,0 +1,739 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/config.proto
+
+package org.tensorflow.proto.framework;
+
+/**
+ * Protobuf type {@code tensorflow.ThreadPoolOptionProto}
+ */
+public  final class ThreadPoolOptionProto extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.ThreadPoolOptionProto)
+    ThreadPoolOptionProtoOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use ThreadPoolOptionProto.newBuilder() to construct.
+  private ThreadPoolOptionProto(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private ThreadPoolOptionProto() {
+    globalName_ = "";
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new ThreadPoolOptionProto();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private ThreadPoolOptionProto(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 8: {
+
+            numThreads_ = input.readInt32();
+            break;
+          }
+          case 18: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            globalName_ = s;
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_ThreadPoolOptionProto_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_ThreadPoolOptionProto_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.framework.ThreadPoolOptionProto.class, org.tensorflow.proto.framework.ThreadPoolOptionProto.Builder.class);
+  }
+
+  public static final int NUM_THREADS_FIELD_NUMBER = 1;
+  private int numThreads_;
+  /**
+   * <pre>
+   * The number of threads in the pool.
+   * 0 means the system picks a value based on where this option proto is used
+   * (see the declaration of the specific field for more info).
+   * </pre>
+   *
+   * <code>int32 num_threads = 1;</code>
+   */
+  public int getNumThreads() {
+    return numThreads_;
+  }
+
+  public static final int GLOBAL_NAME_FIELD_NUMBER = 2;
+  private volatile java.lang.Object globalName_;
+  /**
+   * <pre>
+   * The global name of the threadpool.
+   * If empty, then the threadpool is made and used according to the scope it's
+   * in - e.g., for a session threadpool, it is used by that session only.
+   * If non-empty, then:
+   * - a global threadpool associated with this name is looked
+   *   up or created. This allows, for example, sharing one threadpool across
+   *   many sessions (e.g., like the default behavior, if
+   *   inter_op_parallelism_threads is not configured), but still partitioning
+   *   into a large and small pool.
+   * - if the threadpool for this global_name already exists, then it is an
+   *   error if the existing pool was created using a different num_threads
+   *   value as is specified on this call.
+   * - threadpools created this way are never garbage collected.
+   * </pre>
+   *
+   * <code>string global_name = 2;</code>
+   */
+  public java.lang.String getGlobalName() {
+    java.lang.Object ref = globalName_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      globalName_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * The global name of the threadpool.
+   * If empty, then the threadpool is made and used according to the scope it's
+   * in - e.g., for a session threadpool, it is used by that session only.
+   * If non-empty, then:
+   * - a global threadpool associated with this name is looked
+   *   up or created. This allows, for example, sharing one threadpool across
+   *   many sessions (e.g., like the default behavior, if
+   *   inter_op_parallelism_threads is not configured), but still partitioning
+   *   into a large and small pool.
+   * - if the threadpool for this global_name already exists, then it is an
+   *   error if the existing pool was created using a different num_threads
+   *   value as is specified on this call.
+   * - threadpools created this way are never garbage collected.
+   * </pre>
+   *
+   * <code>string global_name = 2;</code>
+   */
+  public com.google.protobuf.ByteString
+      getGlobalNameBytes() {
+    java.lang.Object ref = globalName_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      globalName_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    if (numThreads_ != 0) {
+      output.writeInt32(1, numThreads_);
+    }
+    if (!getGlobalNameBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 2, globalName_);
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (numThreads_ != 0) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt32Size(1, numThreads_);
+    }
+    if (!getGlobalNameBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(2, globalName_);
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.framework.ThreadPoolOptionProto)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.framework.ThreadPoolOptionProto other = (org.tensorflow.proto.framework.ThreadPoolOptionProto) obj;
+
+    if (getNumThreads()
+        != other.getNumThreads()) return false;
+    if (!getGlobalName()
+        .equals(other.getGlobalName())) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    hash = (37 * hash) + NUM_THREADS_FIELD_NUMBER;
+    hash = (53 * hash) + getNumThreads();
+    hash = (37 * hash) + GLOBAL_NAME_FIELD_NUMBER;
+    hash = (53 * hash) + getGlobalName().hashCode();
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.framework.ThreadPoolOptionProto parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.ThreadPoolOptionProto parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.ThreadPoolOptionProto parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.ThreadPoolOptionProto parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.ThreadPoolOptionProto parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.ThreadPoolOptionProto parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.ThreadPoolOptionProto parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.ThreadPoolOptionProto parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.ThreadPoolOptionProto parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.ThreadPoolOptionProto parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.ThreadPoolOptionProto parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.ThreadPoolOptionProto parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.framework.ThreadPoolOptionProto prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * Protobuf type {@code tensorflow.ThreadPoolOptionProto}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.ThreadPoolOptionProto)
+      org.tensorflow.proto.framework.ThreadPoolOptionProtoOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_ThreadPoolOptionProto_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_ThreadPoolOptionProto_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.ThreadPoolOptionProto.class, org.tensorflow.proto.framework.ThreadPoolOptionProto.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.framework.ThreadPoolOptionProto.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      numThreads_ = 0;
+
+      globalName_ = "";
+
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.framework.ConfigProtos.internal_static_tensorflow_ThreadPoolOptionProto_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.ThreadPoolOptionProto getDefaultInstanceForType() {
+      return org.tensorflow.proto.framework.ThreadPoolOptionProto.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.ThreadPoolOptionProto build() {
+      org.tensorflow.proto.framework.ThreadPoolOptionProto result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.ThreadPoolOptionProto buildPartial() {
+      org.tensorflow.proto.framework.ThreadPoolOptionProto result = new org.tensorflow.proto.framework.ThreadPoolOptionProto(this);
+      result.numThreads_ = numThreads_;
+      result.globalName_ = globalName_;
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.framework.ThreadPoolOptionProto) {
+        return mergeFrom((org.tensorflow.proto.framework.ThreadPoolOptionProto)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.framework.ThreadPoolOptionProto other) {
+      if (other == org.tensorflow.proto.framework.ThreadPoolOptionProto.getDefaultInstance()) return this;
+      if (other.getNumThreads() != 0) {
+        setNumThreads(other.getNumThreads());
+      }
+      if (!other.getGlobalName().isEmpty()) {
+        globalName_ = other.globalName_;
+        onChanged();
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.framework.ThreadPoolOptionProto parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.framework.ThreadPoolOptionProto) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+
+    private int numThreads_ ;
+    /**
+     * <pre>
+     * The number of threads in the pool.
+     * 0 means the system picks a value based on where this option proto is used
+     * (see the declaration of the specific field for more info).
+     * </pre>
+     *
+     * <code>int32 num_threads = 1;</code>
+     */
+    public int getNumThreads() {
+      return numThreads_;
+    }
+    /**
+     * <pre>
+     * The number of threads in the pool.
+     * 0 means the system picks a value based on where this option proto is used
+     * (see the declaration of the specific field for more info).
+     * </pre>
+     *
+     * <code>int32 num_threads = 1;</code>
+     */
+    public Builder setNumThreads(int value) {
+      
+      numThreads_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * The number of threads in the pool.
+     * 0 means the system picks a value based on where this option proto is used
+     * (see the declaration of the specific field for more info).
+     * </pre>
+     *
+     * <code>int32 num_threads = 1;</code>
+     */
+    public Builder clearNumThreads() {
+      
+      numThreads_ = 0;
+      onChanged();
+      return this;
+    }
+
+    private java.lang.Object globalName_ = "";
+    /**
+     * <pre>
+     * The global name of the threadpool.
+     * If empty, then the threadpool is made and used according to the scope it's
+     * in - e.g., for a session threadpool, it is used by that session only.
+     * If non-empty, then:
+     * - a global threadpool associated with this name is looked
+     *   up or created. This allows, for example, sharing one threadpool across
+     *   many sessions (e.g., like the default behavior, if
+     *   inter_op_parallelism_threads is not configured), but still partitioning
+     *   into a large and small pool.
+     * - if the threadpool for this global_name already exists, then it is an
+     *   error if the existing pool was created using a different num_threads
+     *   value as is specified on this call.
+     * - threadpools created this way are never garbage collected.
+     * </pre>
+     *
+     * <code>string global_name = 2;</code>
+     */
+    public java.lang.String getGlobalName() {
+      java.lang.Object ref = globalName_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        globalName_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * The global name of the threadpool.
+     * If empty, then the threadpool is made and used according to the scope it's
+     * in - e.g., for a session threadpool, it is used by that session only.
+     * If non-empty, then:
+     * - a global threadpool associated with this name is looked
+     *   up or created. This allows, for example, sharing one threadpool across
+     *   many sessions (e.g., like the default behavior, if
+     *   inter_op_parallelism_threads is not configured), but still partitioning
+     *   into a large and small pool.
+     * - if the threadpool for this global_name already exists, then it is an
+     *   error if the existing pool was created using a different num_threads
+     *   value as is specified on this call.
+     * - threadpools created this way are never garbage collected.
+     * </pre>
+     *
+     * <code>string global_name = 2;</code>
+     */
+    public com.google.protobuf.ByteString
+        getGlobalNameBytes() {
+      java.lang.Object ref = globalName_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        globalName_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * The global name of the threadpool.
+     * If empty, then the threadpool is made and used according to the scope it's
+     * in - e.g., for a session threadpool, it is used by that session only.
+     * If non-empty, then:
+     * - a global threadpool associated with this name is looked
+     *   up or created. This allows, for example, sharing one threadpool across
+     *   many sessions (e.g., like the default behavior, if
+     *   inter_op_parallelism_threads is not configured), but still partitioning
+     *   into a large and small pool.
+     * - if the threadpool for this global_name already exists, then it is an
+     *   error if the existing pool was created using a different num_threads
+     *   value as is specified on this call.
+     * - threadpools created this way are never garbage collected.
+     * </pre>
+     *
+     * <code>string global_name = 2;</code>
+     */
+    public Builder setGlobalName(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      globalName_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * The global name of the threadpool.
+     * If empty, then the threadpool is made and used according to the scope it's
+     * in - e.g., for a session threadpool, it is used by that session only.
+     * If non-empty, then:
+     * - a global threadpool associated with this name is looked
+     *   up or created. This allows, for example, sharing one threadpool across
+     *   many sessions (e.g., like the default behavior, if
+     *   inter_op_parallelism_threads is not configured), but still partitioning
+     *   into a large and small pool.
+     * - if the threadpool for this global_name already exists, then it is an
+     *   error if the existing pool was created using a different num_threads
+     *   value as is specified on this call.
+     * - threadpools created this way are never garbage collected.
+     * </pre>
+     *
+     * <code>string global_name = 2;</code>
+     */
+    public Builder clearGlobalName() {
+      
+      globalName_ = getDefaultInstance().getGlobalName();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * The global name of the threadpool.
+     * If empty, then the threadpool is made and used according to the scope it's
+     * in - e.g., for a session threadpool, it is used by that session only.
+     * If non-empty, then:
+     * - a global threadpool associated with this name is looked
+     *   up or created. This allows, for example, sharing one threadpool across
+     *   many sessions (e.g., like the default behavior, if
+     *   inter_op_parallelism_threads is not configured), but still partitioning
+     *   into a large and small pool.
+     * - if the threadpool for this global_name already exists, then it is an
+     *   error if the existing pool was created using a different num_threads
+     *   value as is specified on this call.
+     * - threadpools created this way are never garbage collected.
+     * </pre>
+     *
+     * <code>string global_name = 2;</code>
+     */
+    public Builder setGlobalNameBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      globalName_ = value;
+      onChanged();
+      return this;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.ThreadPoolOptionProto)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.ThreadPoolOptionProto)
+  private static final org.tensorflow.proto.framework.ThreadPoolOptionProto DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.ThreadPoolOptionProto();
+  }
+
+  public static org.tensorflow.proto.framework.ThreadPoolOptionProto getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<ThreadPoolOptionProto>
+      PARSER = new com.google.protobuf.AbstractParser<ThreadPoolOptionProto>() {
+    @java.lang.Override
+    public ThreadPoolOptionProto parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new ThreadPoolOptionProto(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<ThreadPoolOptionProto> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<ThreadPoolOptionProto> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.framework.ThreadPoolOptionProto getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ThreadPoolOptionProtoOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ThreadPoolOptionProtoOrBuilder.java
new file mode 100644
index 00000000000..d1d3d837519
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ThreadPoolOptionProtoOrBuilder.java
@@ -0,0 +1,62 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/config.proto
+
+package org.tensorflow.proto.framework;
+
+public interface ThreadPoolOptionProtoOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.ThreadPoolOptionProto)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <pre>
+   * The number of threads in the pool.
+   * 0 means the system picks a value based on where this option proto is used
+   * (see the declaration of the specific field for more info).
+   * </pre>
+   *
+   * <code>int32 num_threads = 1;</code>
+   */
+  int getNumThreads();
+
+  /**
+   * <pre>
+   * The global name of the threadpool.
+   * If empty, then the threadpool is made and used according to the scope it's
+   * in - e.g., for a session threadpool, it is used by that session only.
+   * If non-empty, then:
+   * - a global threadpool associated with this name is looked
+   *   up or created. This allows, for example, sharing one threadpool across
+   *   many sessions (e.g., like the default behavior, if
+   *   inter_op_parallelism_threads is not configured), but still partitioning
+   *   into a large and small pool.
+   * - if the threadpool for this global_name already exists, then it is an
+   *   error if the existing pool was created using a different num_threads
+   *   value as is specified on this call.
+   * - threadpools created this way are never garbage collected.
+   * </pre>
+   *
+   * <code>string global_name = 2;</code>
+   */
+  java.lang.String getGlobalName();
+  /**
+   * <pre>
+   * The global name of the threadpool.
+   * If empty, then the threadpool is made and used according to the scope it's
+   * in - e.g., for a session threadpool, it is used by that session only.
+   * If non-empty, then:
+   * - a global threadpool associated with this name is looked
+   *   up or created. This allows, for example, sharing one threadpool across
+   *   many sessions (e.g., like the default behavior, if
+   *   inter_op_parallelism_threads is not configured), but still partitioning
+   *   into a large and small pool.
+   * - if the threadpool for this global_name already exists, then it is an
+   *   error if the existing pool was created using a different num_threads
+   *   value as is specified on this call.
+   * - threadpools created this way are never garbage collected.
+   * </pre>
+   *
+   * <code>string global_name = 2;</code>
+   */
+  com.google.protobuf.ByteString
+      getGlobalNameBytes();
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TrackableObjectGraph.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TrackableObjectGraph.java
new file mode 100644
index 00000000000..5d516f926dc
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TrackableObjectGraph.java
@@ -0,0 +1,5141 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/trackable_object_graph.proto
+
+package org.tensorflow.proto.framework;
+
+/**
+ * Protobuf type {@code tensorflow.TrackableObjectGraph}
+ */
+public  final class TrackableObjectGraph extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.TrackableObjectGraph)
+    TrackableObjectGraphOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use TrackableObjectGraph.newBuilder() to construct.
+  private TrackableObjectGraph(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private TrackableObjectGraph() {
+    nodes_ = java.util.Collections.emptyList();
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new TrackableObjectGraph();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private TrackableObjectGraph(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    int mutable_bitField0_ = 0;
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 10: {
+            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
+              nodes_ = new java.util.ArrayList<org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject>();
+              mutable_bitField0_ |= 0x00000001;
+            }
+            nodes_.add(
+                input.readMessage(org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.parser(), extensionRegistry));
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      if (((mutable_bitField0_ & 0x00000001) != 0)) {
+        nodes_ = java.util.Collections.unmodifiableList(nodes_);
+      }
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.framework.TrackableObjectGraphProtos.internal_static_tensorflow_TrackableObjectGraph_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.framework.TrackableObjectGraphProtos.internal_static_tensorflow_TrackableObjectGraph_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.framework.TrackableObjectGraph.class, org.tensorflow.proto.framework.TrackableObjectGraph.Builder.class);
+  }
+
+  public interface TrackableObjectOrBuilder extends
+      // @@protoc_insertion_point(interface_extends:tensorflow.TrackableObjectGraph.TrackableObject)
+      com.google.protobuf.MessageOrBuilder {
+
+    /**
+     * <pre>
+     * Objects which this object depends on.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.ObjectReference children = 1;</code>
+     */
+    java.util.List<org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference> 
+        getChildrenList();
+    /**
+     * <pre>
+     * Objects which this object depends on.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.ObjectReference children = 1;</code>
+     */
+    org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference getChildren(int index);
+    /**
+     * <pre>
+     * Objects which this object depends on.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.ObjectReference children = 1;</code>
+     */
+    int getChildrenCount();
+    /**
+     * <pre>
+     * Objects which this object depends on.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.ObjectReference children = 1;</code>
+     */
+    java.util.List<? extends org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReferenceOrBuilder> 
+        getChildrenOrBuilderList();
+    /**
+     * <pre>
+     * Objects which this object depends on.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.ObjectReference children = 1;</code>
+     */
+    org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReferenceOrBuilder getChildrenOrBuilder(
+        int index);
+
+    /**
+     * <pre>
+     * Serialized data specific to this object.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SerializedTensor attributes = 2;</code>
+     */
+    java.util.List<org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensor> 
+        getAttributesList();
+    /**
+     * <pre>
+     * Serialized data specific to this object.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SerializedTensor attributes = 2;</code>
+     */
+    org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensor getAttributes(int index);
+    /**
+     * <pre>
+     * Serialized data specific to this object.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SerializedTensor attributes = 2;</code>
+     */
+    int getAttributesCount();
+    /**
+     * <pre>
+     * Serialized data specific to this object.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SerializedTensor attributes = 2;</code>
+     */
+    java.util.List<? extends org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensorOrBuilder> 
+        getAttributesOrBuilderList();
+    /**
+     * <pre>
+     * Serialized data specific to this object.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SerializedTensor attributes = 2;</code>
+     */
+    org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensorOrBuilder getAttributesOrBuilder(
+        int index);
+
+    /**
+     * <pre>
+     * Slot variables owned by this object.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SlotVariableReference slot_variables = 3;</code>
+     */
+    java.util.List<org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference> 
+        getSlotVariablesList();
+    /**
+     * <pre>
+     * Slot variables owned by this object.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SlotVariableReference slot_variables = 3;</code>
+     */
+    org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference getSlotVariables(int index);
+    /**
+     * <pre>
+     * Slot variables owned by this object.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SlotVariableReference slot_variables = 3;</code>
+     */
+    int getSlotVariablesCount();
+    /**
+     * <pre>
+     * Slot variables owned by this object.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SlotVariableReference slot_variables = 3;</code>
+     */
+    java.util.List<? extends org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReferenceOrBuilder> 
+        getSlotVariablesOrBuilderList();
+    /**
+     * <pre>
+     * Slot variables owned by this object.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SlotVariableReference slot_variables = 3;</code>
+     */
+    org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReferenceOrBuilder getSlotVariablesOrBuilder(
+        int index);
+  }
+  /**
+   * Protobuf type {@code tensorflow.TrackableObjectGraph.TrackableObject}
+   */
+  public  static final class TrackableObject extends
+      com.google.protobuf.GeneratedMessageV3 implements
+      // @@protoc_insertion_point(message_implements:tensorflow.TrackableObjectGraph.TrackableObject)
+      TrackableObjectOrBuilder {
+  private static final long serialVersionUID = 0L;
+    // Use TrackableObject.newBuilder() to construct.
+    private TrackableObject(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+      super(builder);
+    }
+    private TrackableObject() {
+      children_ = java.util.Collections.emptyList();
+      attributes_ = java.util.Collections.emptyList();
+      slotVariables_ = java.util.Collections.emptyList();
+    }
+
+    @java.lang.Override
+    @SuppressWarnings({"unused"})
+    protected java.lang.Object newInstance(
+        UnusedPrivateParameter unused) {
+      return new TrackableObject();
+    }
+
+    @java.lang.Override
+    public final com.google.protobuf.UnknownFieldSet
+    getUnknownFields() {
+      return this.unknownFields;
+    }
+    private TrackableObject(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      this();
+      if (extensionRegistry == null) {
+        throw new java.lang.NullPointerException();
+      }
+      int mutable_bitField0_ = 0;
+      com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+          com.google.protobuf.UnknownFieldSet.newBuilder();
+      try {
+        boolean done = false;
+        while (!done) {
+          int tag = input.readTag();
+          switch (tag) {
+            case 0:
+              done = true;
+              break;
+            case 10: {
+              if (!((mutable_bitField0_ & 0x00000001) != 0)) {
+                children_ = new java.util.ArrayList<org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference>();
+                mutable_bitField0_ |= 0x00000001;
+              }
+              children_.add(
+                  input.readMessage(org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference.parser(), extensionRegistry));
+              break;
+            }
+            case 18: {
+              if (!((mutable_bitField0_ & 0x00000002) != 0)) {
+                attributes_ = new java.util.ArrayList<org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensor>();
+                mutable_bitField0_ |= 0x00000002;
+              }
+              attributes_.add(
+                  input.readMessage(org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensor.parser(), extensionRegistry));
+              break;
+            }
+            case 26: {
+              if (!((mutable_bitField0_ & 0x00000004) != 0)) {
+                slotVariables_ = new java.util.ArrayList<org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference>();
+                mutable_bitField0_ |= 0x00000004;
+              }
+              slotVariables_.add(
+                  input.readMessage(org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference.parser(), extensionRegistry));
+              break;
+            }
+            default: {
+              if (!parseUnknownField(
+                  input, unknownFields, extensionRegistry, tag)) {
+                done = true;
+              }
+              break;
+            }
+          }
+        }
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        throw e.setUnfinishedMessage(this);
+      } catch (java.io.IOException e) {
+        throw new com.google.protobuf.InvalidProtocolBufferException(
+            e).setUnfinishedMessage(this);
+      } finally {
+        if (((mutable_bitField0_ & 0x00000001) != 0)) {
+          children_ = java.util.Collections.unmodifiableList(children_);
+        }
+        if (((mutable_bitField0_ & 0x00000002) != 0)) {
+          attributes_ = java.util.Collections.unmodifiableList(attributes_);
+        }
+        if (((mutable_bitField0_ & 0x00000004) != 0)) {
+          slotVariables_ = java.util.Collections.unmodifiableList(slotVariables_);
+        }
+        this.unknownFields = unknownFields.build();
+        makeExtensionsImmutable();
+      }
+    }
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.TrackableObjectGraphProtos.internal_static_tensorflow_TrackableObjectGraph_TrackableObject_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.TrackableObjectGraphProtos.internal_static_tensorflow_TrackableObjectGraph_TrackableObject_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.class, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.Builder.class);
+    }
+
+    public interface ObjectReferenceOrBuilder extends
+        // @@protoc_insertion_point(interface_extends:tensorflow.TrackableObjectGraph.TrackableObject.ObjectReference)
+        com.google.protobuf.MessageOrBuilder {
+
+      /**
+       * <pre>
+       * An index into `TrackableObjectGraph.nodes`, indicating the object
+       * being referenced.
+       * </pre>
+       *
+       * <code>int32 node_id = 1;</code>
+       */
+      int getNodeId();
+
+      /**
+       * <pre>
+       * A user-provided name for the edge.
+       * </pre>
+       *
+       * <code>string local_name = 2;</code>
+       */
+      java.lang.String getLocalName();
+      /**
+       * <pre>
+       * A user-provided name for the edge.
+       * </pre>
+       *
+       * <code>string local_name = 2;</code>
+       */
+      com.google.protobuf.ByteString
+          getLocalNameBytes();
+    }
+    /**
+     * Protobuf type {@code tensorflow.TrackableObjectGraph.TrackableObject.ObjectReference}
+     */
+    public  static final class ObjectReference extends
+        com.google.protobuf.GeneratedMessageV3 implements
+        // @@protoc_insertion_point(message_implements:tensorflow.TrackableObjectGraph.TrackableObject.ObjectReference)
+        ObjectReferenceOrBuilder {
+    private static final long serialVersionUID = 0L;
+      // Use ObjectReference.newBuilder() to construct.
+      private ObjectReference(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+        super(builder);
+      }
+      private ObjectReference() {
+        localName_ = "";
+      }
+
+      @java.lang.Override
+      @SuppressWarnings({"unused"})
+      protected java.lang.Object newInstance(
+          UnusedPrivateParameter unused) {
+        return new ObjectReference();
+      }
+
+      @java.lang.Override
+      public final com.google.protobuf.UnknownFieldSet
+      getUnknownFields() {
+        return this.unknownFields;
+      }
+      private ObjectReference(
+          com.google.protobuf.CodedInputStream input,
+          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+          throws com.google.protobuf.InvalidProtocolBufferException {
+        this();
+        if (extensionRegistry == null) {
+          throw new java.lang.NullPointerException();
+        }
+        com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+            com.google.protobuf.UnknownFieldSet.newBuilder();
+        try {
+          boolean done = false;
+          while (!done) {
+            int tag = input.readTag();
+            switch (tag) {
+              case 0:
+                done = true;
+                break;
+              case 8: {
+
+                nodeId_ = input.readInt32();
+                break;
+              }
+              case 18: {
+                java.lang.String s = input.readStringRequireUtf8();
+
+                localName_ = s;
+                break;
+              }
+              default: {
+                if (!parseUnknownField(
+                    input, unknownFields, extensionRegistry, tag)) {
+                  done = true;
+                }
+                break;
+              }
+            }
+          }
+        } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+          throw e.setUnfinishedMessage(this);
+        } catch (java.io.IOException e) {
+          throw new com.google.protobuf.InvalidProtocolBufferException(
+              e).setUnfinishedMessage(this);
+        } finally {
+          this.unknownFields = unknownFields.build();
+          makeExtensionsImmutable();
+        }
+      }
+      public static final com.google.protobuf.Descriptors.Descriptor
+          getDescriptor() {
+        return org.tensorflow.proto.framework.TrackableObjectGraphProtos.internal_static_tensorflow_TrackableObjectGraph_TrackableObject_ObjectReference_descriptor;
+      }
+
+      @java.lang.Override
+      protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+          internalGetFieldAccessorTable() {
+        return org.tensorflow.proto.framework.TrackableObjectGraphProtos.internal_static_tensorflow_TrackableObjectGraph_TrackableObject_ObjectReference_fieldAccessorTable
+            .ensureFieldAccessorsInitialized(
+                org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference.class, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference.Builder.class);
+      }
+
+      public static final int NODE_ID_FIELD_NUMBER = 1;
+      private int nodeId_;
+      /**
+       * <pre>
+       * An index into `TrackableObjectGraph.nodes`, indicating the object
+       * being referenced.
+       * </pre>
+       *
+       * <code>int32 node_id = 1;</code>
+       */
+      public int getNodeId() {
+        return nodeId_;
+      }
+
+      public static final int LOCAL_NAME_FIELD_NUMBER = 2;
+      private volatile java.lang.Object localName_;
+      /**
+       * <pre>
+       * A user-provided name for the edge.
+       * </pre>
+       *
+       * <code>string local_name = 2;</code>
+       */
+      public java.lang.String getLocalName() {
+        java.lang.Object ref = localName_;
+        if (ref instanceof java.lang.String) {
+          return (java.lang.String) ref;
+        } else {
+          com.google.protobuf.ByteString bs = 
+              (com.google.protobuf.ByteString) ref;
+          java.lang.String s = bs.toStringUtf8();
+          localName_ = s;
+          return s;
+        }
+      }
+      /**
+       * <pre>
+       * A user-provided name for the edge.
+       * </pre>
+       *
+       * <code>string local_name = 2;</code>
+       */
+      public com.google.protobuf.ByteString
+          getLocalNameBytes() {
+        java.lang.Object ref = localName_;
+        if (ref instanceof java.lang.String) {
+          com.google.protobuf.ByteString b = 
+              com.google.protobuf.ByteString.copyFromUtf8(
+                  (java.lang.String) ref);
+          localName_ = b;
+          return b;
+        } else {
+          return (com.google.protobuf.ByteString) ref;
+        }
+      }
+
+      private byte memoizedIsInitialized = -1;
+      @java.lang.Override
+      public final boolean isInitialized() {
+        byte isInitialized = memoizedIsInitialized;
+        if (isInitialized == 1) return true;
+        if (isInitialized == 0) return false;
+
+        memoizedIsInitialized = 1;
+        return true;
+      }
+
+      @java.lang.Override
+      public void writeTo(com.google.protobuf.CodedOutputStream output)
+                          throws java.io.IOException {
+        if (nodeId_ != 0) {
+          output.writeInt32(1, nodeId_);
+        }
+        if (!getLocalNameBytes().isEmpty()) {
+          com.google.protobuf.GeneratedMessageV3.writeString(output, 2, localName_);
+        }
+        unknownFields.writeTo(output);
+      }
+
+      @java.lang.Override
+      public int getSerializedSize() {
+        int size = memoizedSize;
+        if (size != -1) return size;
+
+        size = 0;
+        if (nodeId_ != 0) {
+          size += com.google.protobuf.CodedOutputStream
+            .computeInt32Size(1, nodeId_);
+        }
+        if (!getLocalNameBytes().isEmpty()) {
+          size += com.google.protobuf.GeneratedMessageV3.computeStringSize(2, localName_);
+        }
+        size += unknownFields.getSerializedSize();
+        memoizedSize = size;
+        return size;
+      }
+
+      @java.lang.Override
+      public boolean equals(final java.lang.Object obj) {
+        if (obj == this) {
+         return true;
+        }
+        if (!(obj instanceof org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference)) {
+          return super.equals(obj);
+        }
+        org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference other = (org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference) obj;
+
+        if (getNodeId()
+            != other.getNodeId()) return false;
+        if (!getLocalName()
+            .equals(other.getLocalName())) return false;
+        if (!unknownFields.equals(other.unknownFields)) return false;
+        return true;
+      }
+
+      @java.lang.Override
+      public int hashCode() {
+        if (memoizedHashCode != 0) {
+          return memoizedHashCode;
+        }
+        int hash = 41;
+        hash = (19 * hash) + getDescriptor().hashCode();
+        hash = (37 * hash) + NODE_ID_FIELD_NUMBER;
+        hash = (53 * hash) + getNodeId();
+        hash = (37 * hash) + LOCAL_NAME_FIELD_NUMBER;
+        hash = (53 * hash) + getLocalName().hashCode();
+        hash = (29 * hash) + unknownFields.hashCode();
+        memoizedHashCode = hash;
+        return hash;
+      }
+
+      public static org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference parseFrom(
+          java.nio.ByteBuffer data)
+          throws com.google.protobuf.InvalidProtocolBufferException {
+        return PARSER.parseFrom(data);
+      }
+      public static org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference parseFrom(
+          java.nio.ByteBuffer data,
+          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+          throws com.google.protobuf.InvalidProtocolBufferException {
+        return PARSER.parseFrom(data, extensionRegistry);
+      }
+      public static org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference parseFrom(
+          com.google.protobuf.ByteString data)
+          throws com.google.protobuf.InvalidProtocolBufferException {
+        return PARSER.parseFrom(data);
+      }
+      public static org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference parseFrom(
+          com.google.protobuf.ByteString data,
+          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+          throws com.google.protobuf.InvalidProtocolBufferException {
+        return PARSER.parseFrom(data, extensionRegistry);
+      }
+      public static org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference parseFrom(byte[] data)
+          throws com.google.protobuf.InvalidProtocolBufferException {
+        return PARSER.parseFrom(data);
+      }
+      public static org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference parseFrom(
+          byte[] data,
+          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+          throws com.google.protobuf.InvalidProtocolBufferException {
+        return PARSER.parseFrom(data, extensionRegistry);
+      }
+      public static org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference parseFrom(java.io.InputStream input)
+          throws java.io.IOException {
+        return com.google.protobuf.GeneratedMessageV3
+            .parseWithIOException(PARSER, input);
+      }
+      public static org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference parseFrom(
+          java.io.InputStream input,
+          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+          throws java.io.IOException {
+        return com.google.protobuf.GeneratedMessageV3
+            .parseWithIOException(PARSER, input, extensionRegistry);
+      }
+      public static org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference parseDelimitedFrom(java.io.InputStream input)
+          throws java.io.IOException {
+        return com.google.protobuf.GeneratedMessageV3
+            .parseDelimitedWithIOException(PARSER, input);
+      }
+      public static org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference parseDelimitedFrom(
+          java.io.InputStream input,
+          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+          throws java.io.IOException {
+        return com.google.protobuf.GeneratedMessageV3
+            .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+      }
+      public static org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference parseFrom(
+          com.google.protobuf.CodedInputStream input)
+          throws java.io.IOException {
+        return com.google.protobuf.GeneratedMessageV3
+            .parseWithIOException(PARSER, input);
+      }
+      public static org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference parseFrom(
+          com.google.protobuf.CodedInputStream input,
+          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+          throws java.io.IOException {
+        return com.google.protobuf.GeneratedMessageV3
+            .parseWithIOException(PARSER, input, extensionRegistry);
+      }
+
+      @java.lang.Override
+      public Builder newBuilderForType() { return newBuilder(); }
+      public static Builder newBuilder() {
+        return DEFAULT_INSTANCE.toBuilder();
+      }
+      public static Builder newBuilder(org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference prototype) {
+        return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+      }
+      @java.lang.Override
+      public Builder toBuilder() {
+        return this == DEFAULT_INSTANCE
+            ? new Builder() : new Builder().mergeFrom(this);
+      }
+
+      @java.lang.Override
+      protected Builder newBuilderForType(
+          com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+        Builder builder = new Builder(parent);
+        return builder;
+      }
+      /**
+       * Protobuf type {@code tensorflow.TrackableObjectGraph.TrackableObject.ObjectReference}
+       */
+      public static final class Builder extends
+          com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+          // @@protoc_insertion_point(builder_implements:tensorflow.TrackableObjectGraph.TrackableObject.ObjectReference)
+          org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReferenceOrBuilder {
+        public static final com.google.protobuf.Descriptors.Descriptor
+            getDescriptor() {
+          return org.tensorflow.proto.framework.TrackableObjectGraphProtos.internal_static_tensorflow_TrackableObjectGraph_TrackableObject_ObjectReference_descriptor;
+        }
+
+        @java.lang.Override
+        protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+            internalGetFieldAccessorTable() {
+          return org.tensorflow.proto.framework.TrackableObjectGraphProtos.internal_static_tensorflow_TrackableObjectGraph_TrackableObject_ObjectReference_fieldAccessorTable
+              .ensureFieldAccessorsInitialized(
+                  org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference.class, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference.Builder.class);
+        }
+
+        // Construct using org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference.newBuilder()
+        private Builder() {
+          maybeForceBuilderInitialization();
+        }
+
+        private Builder(
+            com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+          super(parent);
+          maybeForceBuilderInitialization();
+        }
+        private void maybeForceBuilderInitialization() {
+          if (com.google.protobuf.GeneratedMessageV3
+                  .alwaysUseFieldBuilders) {
+          }
+        }
+        @java.lang.Override
+        public Builder clear() {
+          super.clear();
+          nodeId_ = 0;
+
+          localName_ = "";
+
+          return this;
+        }
+
+        @java.lang.Override
+        public com.google.protobuf.Descriptors.Descriptor
+            getDescriptorForType() {
+          return org.tensorflow.proto.framework.TrackableObjectGraphProtos.internal_static_tensorflow_TrackableObjectGraph_TrackableObject_ObjectReference_descriptor;
+        }
+
+        @java.lang.Override
+        public org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference getDefaultInstanceForType() {
+          return org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference.getDefaultInstance();
+        }
+
+        @java.lang.Override
+        public org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference build() {
+          org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference result = buildPartial();
+          if (!result.isInitialized()) {
+            throw newUninitializedMessageException(result);
+          }
+          return result;
+        }
+
+        @java.lang.Override
+        public org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference buildPartial() {
+          org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference result = new org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference(this);
+          result.nodeId_ = nodeId_;
+          result.localName_ = localName_;
+          onBuilt();
+          return result;
+        }
+
+        @java.lang.Override
+        public Builder clone() {
+          return super.clone();
+        }
+        @java.lang.Override
+        public Builder setField(
+            com.google.protobuf.Descriptors.FieldDescriptor field,
+            java.lang.Object value) {
+          return super.setField(field, value);
+        }
+        @java.lang.Override
+        public Builder clearField(
+            com.google.protobuf.Descriptors.FieldDescriptor field) {
+          return super.clearField(field);
+        }
+        @java.lang.Override
+        public Builder clearOneof(
+            com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+          return super.clearOneof(oneof);
+        }
+        @java.lang.Override
+        public Builder setRepeatedField(
+            com.google.protobuf.Descriptors.FieldDescriptor field,
+            int index, java.lang.Object value) {
+          return super.setRepeatedField(field, index, value);
+        }
+        @java.lang.Override
+        public Builder addRepeatedField(
+            com.google.protobuf.Descriptors.FieldDescriptor field,
+            java.lang.Object value) {
+          return super.addRepeatedField(field, value);
+        }
+        @java.lang.Override
+        public Builder mergeFrom(com.google.protobuf.Message other) {
+          if (other instanceof org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference) {
+            return mergeFrom((org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference)other);
+          } else {
+            super.mergeFrom(other);
+            return this;
+          }
+        }
+
+        public Builder mergeFrom(org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference other) {
+          if (other == org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference.getDefaultInstance()) return this;
+          if (other.getNodeId() != 0) {
+            setNodeId(other.getNodeId());
+          }
+          if (!other.getLocalName().isEmpty()) {
+            localName_ = other.localName_;
+            onChanged();
+          }
+          this.mergeUnknownFields(other.unknownFields);
+          onChanged();
+          return this;
+        }
+
+        @java.lang.Override
+        public final boolean isInitialized() {
+          return true;
+        }
+
+        @java.lang.Override
+        public Builder mergeFrom(
+            com.google.protobuf.CodedInputStream input,
+            com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+            throws java.io.IOException {
+          org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference parsedMessage = null;
+          try {
+            parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+          } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+            parsedMessage = (org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference) e.getUnfinishedMessage();
+            throw e.unwrapIOException();
+          } finally {
+            if (parsedMessage != null) {
+              mergeFrom(parsedMessage);
+            }
+          }
+          return this;
+        }
+
+        private int nodeId_ ;
+        /**
+         * <pre>
+         * An index into `TrackableObjectGraph.nodes`, indicating the object
+         * being referenced.
+         * </pre>
+         *
+         * <code>int32 node_id = 1;</code>
+         */
+        public int getNodeId() {
+          return nodeId_;
+        }
+        /**
+         * <pre>
+         * An index into `TrackableObjectGraph.nodes`, indicating the object
+         * being referenced.
+         * </pre>
+         *
+         * <code>int32 node_id = 1;</code>
+         */
+        public Builder setNodeId(int value) {
+          
+          nodeId_ = value;
+          onChanged();
+          return this;
+        }
+        /**
+         * <pre>
+         * An index into `TrackableObjectGraph.nodes`, indicating the object
+         * being referenced.
+         * </pre>
+         *
+         * <code>int32 node_id = 1;</code>
+         */
+        public Builder clearNodeId() {
+          
+          nodeId_ = 0;
+          onChanged();
+          return this;
+        }
+
+        private java.lang.Object localName_ = "";
+        /**
+         * <pre>
+         * A user-provided name for the edge.
+         * </pre>
+         *
+         * <code>string local_name = 2;</code>
+         */
+        public java.lang.String getLocalName() {
+          java.lang.Object ref = localName_;
+          if (!(ref instanceof java.lang.String)) {
+            com.google.protobuf.ByteString bs =
+                (com.google.protobuf.ByteString) ref;
+            java.lang.String s = bs.toStringUtf8();
+            localName_ = s;
+            return s;
+          } else {
+            return (java.lang.String) ref;
+          }
+        }
+        /**
+         * <pre>
+         * A user-provided name for the edge.
+         * </pre>
+         *
+         * <code>string local_name = 2;</code>
+         */
+        public com.google.protobuf.ByteString
+            getLocalNameBytes() {
+          java.lang.Object ref = localName_;
+          if (ref instanceof String) {
+            com.google.protobuf.ByteString b = 
+                com.google.protobuf.ByteString.copyFromUtf8(
+                    (java.lang.String) ref);
+            localName_ = b;
+            return b;
+          } else {
+            return (com.google.protobuf.ByteString) ref;
+          }
+        }
+        /**
+         * <pre>
+         * A user-provided name for the edge.
+         * </pre>
+         *
+         * <code>string local_name = 2;</code>
+         */
+        public Builder setLocalName(
+            java.lang.String value) {
+          if (value == null) {
+    throw new NullPointerException();
+  }
+  
+          localName_ = value;
+          onChanged();
+          return this;
+        }
+        /**
+         * <pre>
+         * A user-provided name for the edge.
+         * </pre>
+         *
+         * <code>string local_name = 2;</code>
+         */
+        public Builder clearLocalName() {
+          
+          localName_ = getDefaultInstance().getLocalName();
+          onChanged();
+          return this;
+        }
+        /**
+         * <pre>
+         * A user-provided name for the edge.
+         * </pre>
+         *
+         * <code>string local_name = 2;</code>
+         */
+        public Builder setLocalNameBytes(
+            com.google.protobuf.ByteString value) {
+          if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+          
+          localName_ = value;
+          onChanged();
+          return this;
+        }
+        @java.lang.Override
+        public final Builder setUnknownFields(
+            final com.google.protobuf.UnknownFieldSet unknownFields) {
+          return super.setUnknownFields(unknownFields);
+        }
+
+        @java.lang.Override
+        public final Builder mergeUnknownFields(
+            final com.google.protobuf.UnknownFieldSet unknownFields) {
+          return super.mergeUnknownFields(unknownFields);
+        }
+
+
+        // @@protoc_insertion_point(builder_scope:tensorflow.TrackableObjectGraph.TrackableObject.ObjectReference)
+      }
+
+      // @@protoc_insertion_point(class_scope:tensorflow.TrackableObjectGraph.TrackableObject.ObjectReference)
+      private static final org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference DEFAULT_INSTANCE;
+      static {
+        DEFAULT_INSTANCE = new org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference();
+      }
+
+      public static org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference getDefaultInstance() {
+        return DEFAULT_INSTANCE;
+      }
+
+      private static final com.google.protobuf.Parser<ObjectReference>
+          PARSER = new com.google.protobuf.AbstractParser<ObjectReference>() {
+        @java.lang.Override
+        public ObjectReference parsePartialFrom(
+            com.google.protobuf.CodedInputStream input,
+            com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+            throws com.google.protobuf.InvalidProtocolBufferException {
+          return new ObjectReference(input, extensionRegistry);
+        }
+      };
+
+      public static com.google.protobuf.Parser<ObjectReference> parser() {
+        return PARSER;
+      }
+
+      @java.lang.Override
+      public com.google.protobuf.Parser<ObjectReference> getParserForType() {
+        return PARSER;
+      }
+
+      @java.lang.Override
+      public org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference getDefaultInstanceForType() {
+        return DEFAULT_INSTANCE;
+      }
+
+    }
+
+    public interface SerializedTensorOrBuilder extends
+        // @@protoc_insertion_point(interface_extends:tensorflow.TrackableObjectGraph.TrackableObject.SerializedTensor)
+        com.google.protobuf.MessageOrBuilder {
+
+      /**
+       * <pre>
+       * A name for the Tensor. Simple variables have only one
+       * `SerializedTensor` named "VARIABLE_VALUE" by convention. This value may
+       * be restored on object creation as an optimization.
+       * </pre>
+       *
+       * <code>string name = 1;</code>
+       */
+      java.lang.String getName();
+      /**
+       * <pre>
+       * A name for the Tensor. Simple variables have only one
+       * `SerializedTensor` named "VARIABLE_VALUE" by convention. This value may
+       * be restored on object creation as an optimization.
+       * </pre>
+       *
+       * <code>string name = 1;</code>
+       */
+      com.google.protobuf.ByteString
+          getNameBytes();
+
+      /**
+       * <pre>
+       * The full name of the variable/tensor, if applicable. Used to allow
+       * name-based loading of checkpoints which were saved using an
+       * object-based API. Should match the checkpoint key which would have been
+       * assigned by tf.train.Saver.
+       * </pre>
+       *
+       * <code>string full_name = 2;</code>
+       */
+      java.lang.String getFullName();
+      /**
+       * <pre>
+       * The full name of the variable/tensor, if applicable. Used to allow
+       * name-based loading of checkpoints which were saved using an
+       * object-based API. Should match the checkpoint key which would have been
+       * assigned by tf.train.Saver.
+       * </pre>
+       *
+       * <code>string full_name = 2;</code>
+       */
+      com.google.protobuf.ByteString
+          getFullNameBytes();
+
+      /**
+       * <pre>
+       * The generated name of the Tensor in the checkpoint.
+       * </pre>
+       *
+       * <code>string checkpoint_key = 3;</code>
+       */
+      java.lang.String getCheckpointKey();
+      /**
+       * <pre>
+       * The generated name of the Tensor in the checkpoint.
+       * </pre>
+       *
+       * <code>string checkpoint_key = 3;</code>
+       */
+      com.google.protobuf.ByteString
+          getCheckpointKeyBytes();
+
+      /**
+       * <pre>
+       * Whether checkpoints should be considered as matching even without this
+       * value restored. Used for non-critical values which don't affect the
+       * TensorFlow graph, such as layer configurations.
+       * </pre>
+       *
+       * <code>bool optional_restore = 4;</code>
+       */
+      boolean getOptionalRestore();
+    }
+    /**
+     * Protobuf type {@code tensorflow.TrackableObjectGraph.TrackableObject.SerializedTensor}
+     */
+    public  static final class SerializedTensor extends
+        com.google.protobuf.GeneratedMessageV3 implements
+        // @@protoc_insertion_point(message_implements:tensorflow.TrackableObjectGraph.TrackableObject.SerializedTensor)
+        SerializedTensorOrBuilder {
+    private static final long serialVersionUID = 0L;
+      // Use SerializedTensor.newBuilder() to construct.
+      private SerializedTensor(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+        super(builder);
+      }
+      private SerializedTensor() {
+        name_ = "";
+        fullName_ = "";
+        checkpointKey_ = "";
+      }
+
+      @java.lang.Override
+      @SuppressWarnings({"unused"})
+      protected java.lang.Object newInstance(
+          UnusedPrivateParameter unused) {
+        return new SerializedTensor();
+      }
+
+      @java.lang.Override
+      public final com.google.protobuf.UnknownFieldSet
+      getUnknownFields() {
+        return this.unknownFields;
+      }
+      private SerializedTensor(
+          com.google.protobuf.CodedInputStream input,
+          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+          throws com.google.protobuf.InvalidProtocolBufferException {
+        this();
+        if (extensionRegistry == null) {
+          throw new java.lang.NullPointerException();
+        }
+        com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+            com.google.protobuf.UnknownFieldSet.newBuilder();
+        try {
+          boolean done = false;
+          while (!done) {
+            int tag = input.readTag();
+            switch (tag) {
+              case 0:
+                done = true;
+                break;
+              case 10: {
+                java.lang.String s = input.readStringRequireUtf8();
+
+                name_ = s;
+                break;
+              }
+              case 18: {
+                java.lang.String s = input.readStringRequireUtf8();
+
+                fullName_ = s;
+                break;
+              }
+              case 26: {
+                java.lang.String s = input.readStringRequireUtf8();
+
+                checkpointKey_ = s;
+                break;
+              }
+              case 32: {
+
+                optionalRestore_ = input.readBool();
+                break;
+              }
+              default: {
+                if (!parseUnknownField(
+                    input, unknownFields, extensionRegistry, tag)) {
+                  done = true;
+                }
+                break;
+              }
+            }
+          }
+        } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+          throw e.setUnfinishedMessage(this);
+        } catch (java.io.IOException e) {
+          throw new com.google.protobuf.InvalidProtocolBufferException(
+              e).setUnfinishedMessage(this);
+        } finally {
+          this.unknownFields = unknownFields.build();
+          makeExtensionsImmutable();
+        }
+      }
+      public static final com.google.protobuf.Descriptors.Descriptor
+          getDescriptor() {
+        return org.tensorflow.proto.framework.TrackableObjectGraphProtos.internal_static_tensorflow_TrackableObjectGraph_TrackableObject_SerializedTensor_descriptor;
+      }
+
+      @java.lang.Override
+      protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+          internalGetFieldAccessorTable() {
+        return org.tensorflow.proto.framework.TrackableObjectGraphProtos.internal_static_tensorflow_TrackableObjectGraph_TrackableObject_SerializedTensor_fieldAccessorTable
+            .ensureFieldAccessorsInitialized(
+                org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensor.class, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensor.Builder.class);
+      }
+
+      public static final int NAME_FIELD_NUMBER = 1;
+      private volatile java.lang.Object name_;
+      /**
+       * <pre>
+       * A name for the Tensor. Simple variables have only one
+       * `SerializedTensor` named "VARIABLE_VALUE" by convention. This value may
+       * be restored on object creation as an optimization.
+       * </pre>
+       *
+       * <code>string name = 1;</code>
+       */
+      public java.lang.String getName() {
+        java.lang.Object ref = name_;
+        if (ref instanceof java.lang.String) {
+          return (java.lang.String) ref;
+        } else {
+          com.google.protobuf.ByteString bs = 
+              (com.google.protobuf.ByteString) ref;
+          java.lang.String s = bs.toStringUtf8();
+          name_ = s;
+          return s;
+        }
+      }
+      /**
+       * <pre>
+       * A name for the Tensor. Simple variables have only one
+       * `SerializedTensor` named "VARIABLE_VALUE" by convention. This value may
+       * be restored on object creation as an optimization.
+       * </pre>
+       *
+       * <code>string name = 1;</code>
+       */
+      public com.google.protobuf.ByteString
+          getNameBytes() {
+        java.lang.Object ref = name_;
+        if (ref instanceof java.lang.String) {
+          com.google.protobuf.ByteString b = 
+              com.google.protobuf.ByteString.copyFromUtf8(
+                  (java.lang.String) ref);
+          name_ = b;
+          return b;
+        } else {
+          return (com.google.protobuf.ByteString) ref;
+        }
+      }
+
+      public static final int FULL_NAME_FIELD_NUMBER = 2;
+      private volatile java.lang.Object fullName_;
+      /**
+       * <pre>
+       * The full name of the variable/tensor, if applicable. Used to allow
+       * name-based loading of checkpoints which were saved using an
+       * object-based API. Should match the checkpoint key which would have been
+       * assigned by tf.train.Saver.
+       * </pre>
+       *
+       * <code>string full_name = 2;</code>
+       */
+      public java.lang.String getFullName() {
+        java.lang.Object ref = fullName_;
+        if (ref instanceof java.lang.String) {
+          return (java.lang.String) ref;
+        } else {
+          com.google.protobuf.ByteString bs = 
+              (com.google.protobuf.ByteString) ref;
+          java.lang.String s = bs.toStringUtf8();
+          fullName_ = s;
+          return s;
+        }
+      }
+      /**
+       * <pre>
+       * The full name of the variable/tensor, if applicable. Used to allow
+       * name-based loading of checkpoints which were saved using an
+       * object-based API. Should match the checkpoint key which would have been
+       * assigned by tf.train.Saver.
+       * </pre>
+       *
+       * <code>string full_name = 2;</code>
+       */
+      public com.google.protobuf.ByteString
+          getFullNameBytes() {
+        java.lang.Object ref = fullName_;
+        if (ref instanceof java.lang.String) {
+          com.google.protobuf.ByteString b = 
+              com.google.protobuf.ByteString.copyFromUtf8(
+                  (java.lang.String) ref);
+          fullName_ = b;
+          return b;
+        } else {
+          return (com.google.protobuf.ByteString) ref;
+        }
+      }
+
+      public static final int CHECKPOINT_KEY_FIELD_NUMBER = 3;
+      private volatile java.lang.Object checkpointKey_;
+      /**
+       * <pre>
+       * The generated name of the Tensor in the checkpoint.
+       * </pre>
+       *
+       * <code>string checkpoint_key = 3;</code>
+       */
+      public java.lang.String getCheckpointKey() {
+        java.lang.Object ref = checkpointKey_;
+        if (ref instanceof java.lang.String) {
+          return (java.lang.String) ref;
+        } else {
+          com.google.protobuf.ByteString bs = 
+              (com.google.protobuf.ByteString) ref;
+          java.lang.String s = bs.toStringUtf8();
+          checkpointKey_ = s;
+          return s;
+        }
+      }
+      /**
+       * <pre>
+       * The generated name of the Tensor in the checkpoint.
+       * </pre>
+       *
+       * <code>string checkpoint_key = 3;</code>
+       */
+      public com.google.protobuf.ByteString
+          getCheckpointKeyBytes() {
+        java.lang.Object ref = checkpointKey_;
+        if (ref instanceof java.lang.String) {
+          com.google.protobuf.ByteString b = 
+              com.google.protobuf.ByteString.copyFromUtf8(
+                  (java.lang.String) ref);
+          checkpointKey_ = b;
+          return b;
+        } else {
+          return (com.google.protobuf.ByteString) ref;
+        }
+      }
+
+      public static final int OPTIONAL_RESTORE_FIELD_NUMBER = 4;
+      private boolean optionalRestore_;
+      /**
+       * <pre>
+       * Whether checkpoints should be considered as matching even without this
+       * value restored. Used for non-critical values which don't affect the
+       * TensorFlow graph, such as layer configurations.
+       * </pre>
+       *
+       * <code>bool optional_restore = 4;</code>
+       */
+      public boolean getOptionalRestore() {
+        return optionalRestore_;
+      }
+
+      private byte memoizedIsInitialized = -1;
+      @java.lang.Override
+      public final boolean isInitialized() {
+        byte isInitialized = memoizedIsInitialized;
+        if (isInitialized == 1) return true;
+        if (isInitialized == 0) return false;
+
+        memoizedIsInitialized = 1;
+        return true;
+      }
+
+      @java.lang.Override
+      public void writeTo(com.google.protobuf.CodedOutputStream output)
+                          throws java.io.IOException {
+        if (!getNameBytes().isEmpty()) {
+          com.google.protobuf.GeneratedMessageV3.writeString(output, 1, name_);
+        }
+        if (!getFullNameBytes().isEmpty()) {
+          com.google.protobuf.GeneratedMessageV3.writeString(output, 2, fullName_);
+        }
+        if (!getCheckpointKeyBytes().isEmpty()) {
+          com.google.protobuf.GeneratedMessageV3.writeString(output, 3, checkpointKey_);
+        }
+        if (optionalRestore_ != false) {
+          output.writeBool(4, optionalRestore_);
+        }
+        unknownFields.writeTo(output);
+      }
+
+      @java.lang.Override
+      public int getSerializedSize() {
+        int size = memoizedSize;
+        if (size != -1) return size;
+
+        size = 0;
+        if (!getNameBytes().isEmpty()) {
+          size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, name_);
+        }
+        if (!getFullNameBytes().isEmpty()) {
+          size += com.google.protobuf.GeneratedMessageV3.computeStringSize(2, fullName_);
+        }
+        if (!getCheckpointKeyBytes().isEmpty()) {
+          size += com.google.protobuf.GeneratedMessageV3.computeStringSize(3, checkpointKey_);
+        }
+        if (optionalRestore_ != false) {
+          size += com.google.protobuf.CodedOutputStream
+            .computeBoolSize(4, optionalRestore_);
+        }
+        size += unknownFields.getSerializedSize();
+        memoizedSize = size;
+        return size;
+      }
+
+      @java.lang.Override
+      public boolean equals(final java.lang.Object obj) {
+        if (obj == this) {
+         return true;
+        }
+        if (!(obj instanceof org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensor)) {
+          return super.equals(obj);
+        }
+        org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensor other = (org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensor) obj;
+
+        if (!getName()
+            .equals(other.getName())) return false;
+        if (!getFullName()
+            .equals(other.getFullName())) return false;
+        if (!getCheckpointKey()
+            .equals(other.getCheckpointKey())) return false;
+        if (getOptionalRestore()
+            != other.getOptionalRestore()) return false;
+        if (!unknownFields.equals(other.unknownFields)) return false;
+        return true;
+      }
+
+      @java.lang.Override
+      public int hashCode() {
+        if (memoizedHashCode != 0) {
+          return memoizedHashCode;
+        }
+        int hash = 41;
+        hash = (19 * hash) + getDescriptor().hashCode();
+        hash = (37 * hash) + NAME_FIELD_NUMBER;
+        hash = (53 * hash) + getName().hashCode();
+        hash = (37 * hash) + FULL_NAME_FIELD_NUMBER;
+        hash = (53 * hash) + getFullName().hashCode();
+        hash = (37 * hash) + CHECKPOINT_KEY_FIELD_NUMBER;
+        hash = (53 * hash) + getCheckpointKey().hashCode();
+        hash = (37 * hash) + OPTIONAL_RESTORE_FIELD_NUMBER;
+        hash = (53 * hash) + com.google.protobuf.Internal.hashBoolean(
+            getOptionalRestore());
+        hash = (29 * hash) + unknownFields.hashCode();
+        memoizedHashCode = hash;
+        return hash;
+      }
+
+      public static org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensor parseFrom(
+          java.nio.ByteBuffer data)
+          throws com.google.protobuf.InvalidProtocolBufferException {
+        return PARSER.parseFrom(data);
+      }
+      public static org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensor parseFrom(
+          java.nio.ByteBuffer data,
+          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+          throws com.google.protobuf.InvalidProtocolBufferException {
+        return PARSER.parseFrom(data, extensionRegistry);
+      }
+      public static org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensor parseFrom(
+          com.google.protobuf.ByteString data)
+          throws com.google.protobuf.InvalidProtocolBufferException {
+        return PARSER.parseFrom(data);
+      }
+      public static org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensor parseFrom(
+          com.google.protobuf.ByteString data,
+          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+          throws com.google.protobuf.InvalidProtocolBufferException {
+        return PARSER.parseFrom(data, extensionRegistry);
+      }
+      public static org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensor parseFrom(byte[] data)
+          throws com.google.protobuf.InvalidProtocolBufferException {
+        return PARSER.parseFrom(data);
+      }
+      public static org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensor parseFrom(
+          byte[] data,
+          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+          throws com.google.protobuf.InvalidProtocolBufferException {
+        return PARSER.parseFrom(data, extensionRegistry);
+      }
+      public static org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensor parseFrom(java.io.InputStream input)
+          throws java.io.IOException {
+        return com.google.protobuf.GeneratedMessageV3
+            .parseWithIOException(PARSER, input);
+      }
+      public static org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensor parseFrom(
+          java.io.InputStream input,
+          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+          throws java.io.IOException {
+        return com.google.protobuf.GeneratedMessageV3
+            .parseWithIOException(PARSER, input, extensionRegistry);
+      }
+      public static org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensor parseDelimitedFrom(java.io.InputStream input)
+          throws java.io.IOException {
+        return com.google.protobuf.GeneratedMessageV3
+            .parseDelimitedWithIOException(PARSER, input);
+      }
+      public static org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensor parseDelimitedFrom(
+          java.io.InputStream input,
+          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+          throws java.io.IOException {
+        return com.google.protobuf.GeneratedMessageV3
+            .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+      }
+      public static org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensor parseFrom(
+          com.google.protobuf.CodedInputStream input)
+          throws java.io.IOException {
+        return com.google.protobuf.GeneratedMessageV3
+            .parseWithIOException(PARSER, input);
+      }
+      public static org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensor parseFrom(
+          com.google.protobuf.CodedInputStream input,
+          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+          throws java.io.IOException {
+        return com.google.protobuf.GeneratedMessageV3
+            .parseWithIOException(PARSER, input, extensionRegistry);
+      }
+
+      @java.lang.Override
+      public Builder newBuilderForType() { return newBuilder(); }
+      public static Builder newBuilder() {
+        return DEFAULT_INSTANCE.toBuilder();
+      }
+      public static Builder newBuilder(org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensor prototype) {
+        return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+      }
+      @java.lang.Override
+      public Builder toBuilder() {
+        return this == DEFAULT_INSTANCE
+            ? new Builder() : new Builder().mergeFrom(this);
+      }
+
+      @java.lang.Override
+      protected Builder newBuilderForType(
+          com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+        Builder builder = new Builder(parent);
+        return builder;
+      }
+      /**
+       * Protobuf type {@code tensorflow.TrackableObjectGraph.TrackableObject.SerializedTensor}
+       */
+      public static final class Builder extends
+          com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+          // @@protoc_insertion_point(builder_implements:tensorflow.TrackableObjectGraph.TrackableObject.SerializedTensor)
+          org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensorOrBuilder {
+        public static final com.google.protobuf.Descriptors.Descriptor
+            getDescriptor() {
+          return org.tensorflow.proto.framework.TrackableObjectGraphProtos.internal_static_tensorflow_TrackableObjectGraph_TrackableObject_SerializedTensor_descriptor;
+        }
+
+        @java.lang.Override
+        protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+            internalGetFieldAccessorTable() {
+          return org.tensorflow.proto.framework.TrackableObjectGraphProtos.internal_static_tensorflow_TrackableObjectGraph_TrackableObject_SerializedTensor_fieldAccessorTable
+              .ensureFieldAccessorsInitialized(
+                  org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensor.class, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensor.Builder.class);
+        }
+
+        // Construct using org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensor.newBuilder()
+        private Builder() {
+          maybeForceBuilderInitialization();
+        }
+
+        private Builder(
+            com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+          super(parent);
+          maybeForceBuilderInitialization();
+        }
+        private void maybeForceBuilderInitialization() {
+          if (com.google.protobuf.GeneratedMessageV3
+                  .alwaysUseFieldBuilders) {
+          }
+        }
+        @java.lang.Override
+        public Builder clear() {
+          super.clear();
+          name_ = "";
+
+          fullName_ = "";
+
+          checkpointKey_ = "";
+
+          optionalRestore_ = false;
+
+          return this;
+        }
+
+        @java.lang.Override
+        public com.google.protobuf.Descriptors.Descriptor
+            getDescriptorForType() {
+          return org.tensorflow.proto.framework.TrackableObjectGraphProtos.internal_static_tensorflow_TrackableObjectGraph_TrackableObject_SerializedTensor_descriptor;
+        }
+
+        @java.lang.Override
+        public org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensor getDefaultInstanceForType() {
+          return org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensor.getDefaultInstance();
+        }
+
+        @java.lang.Override
+        public org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensor build() {
+          org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensor result = buildPartial();
+          if (!result.isInitialized()) {
+            throw newUninitializedMessageException(result);
+          }
+          return result;
+        }
+
+        @java.lang.Override
+        public org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensor buildPartial() {
+          org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensor result = new org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensor(this);
+          result.name_ = name_;
+          result.fullName_ = fullName_;
+          result.checkpointKey_ = checkpointKey_;
+          result.optionalRestore_ = optionalRestore_;
+          onBuilt();
+          return result;
+        }
+
+        @java.lang.Override
+        public Builder clone() {
+          return super.clone();
+        }
+        @java.lang.Override
+        public Builder setField(
+            com.google.protobuf.Descriptors.FieldDescriptor field,
+            java.lang.Object value) {
+          return super.setField(field, value);
+        }
+        @java.lang.Override
+        public Builder clearField(
+            com.google.protobuf.Descriptors.FieldDescriptor field) {
+          return super.clearField(field);
+        }
+        @java.lang.Override
+        public Builder clearOneof(
+            com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+          return super.clearOneof(oneof);
+        }
+        @java.lang.Override
+        public Builder setRepeatedField(
+            com.google.protobuf.Descriptors.FieldDescriptor field,
+            int index, java.lang.Object value) {
+          return super.setRepeatedField(field, index, value);
+        }
+        @java.lang.Override
+        public Builder addRepeatedField(
+            com.google.protobuf.Descriptors.FieldDescriptor field,
+            java.lang.Object value) {
+          return super.addRepeatedField(field, value);
+        }
+        @java.lang.Override
+        public Builder mergeFrom(com.google.protobuf.Message other) {
+          if (other instanceof org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensor) {
+            return mergeFrom((org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensor)other);
+          } else {
+            super.mergeFrom(other);
+            return this;
+          }
+        }
+
+        public Builder mergeFrom(org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensor other) {
+          if (other == org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensor.getDefaultInstance()) return this;
+          if (!other.getName().isEmpty()) {
+            name_ = other.name_;
+            onChanged();
+          }
+          if (!other.getFullName().isEmpty()) {
+            fullName_ = other.fullName_;
+            onChanged();
+          }
+          if (!other.getCheckpointKey().isEmpty()) {
+            checkpointKey_ = other.checkpointKey_;
+            onChanged();
+          }
+          if (other.getOptionalRestore() != false) {
+            setOptionalRestore(other.getOptionalRestore());
+          }
+          this.mergeUnknownFields(other.unknownFields);
+          onChanged();
+          return this;
+        }
+
+        @java.lang.Override
+        public final boolean isInitialized() {
+          return true;
+        }
+
+        @java.lang.Override
+        public Builder mergeFrom(
+            com.google.protobuf.CodedInputStream input,
+            com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+            throws java.io.IOException {
+          org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensor parsedMessage = null;
+          try {
+            parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+          } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+            parsedMessage = (org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensor) e.getUnfinishedMessage();
+            throw e.unwrapIOException();
+          } finally {
+            if (parsedMessage != null) {
+              mergeFrom(parsedMessage);
+            }
+          }
+          return this;
+        }
+
+        private java.lang.Object name_ = "";
+        /**
+         * <pre>
+         * A name for the Tensor. Simple variables have only one
+         * `SerializedTensor` named "VARIABLE_VALUE" by convention. This value may
+         * be restored on object creation as an optimization.
+         * </pre>
+         *
+         * <code>string name = 1;</code>
+         */
+        public java.lang.String getName() {
+          java.lang.Object ref = name_;
+          if (!(ref instanceof java.lang.String)) {
+            com.google.protobuf.ByteString bs =
+                (com.google.protobuf.ByteString) ref;
+            java.lang.String s = bs.toStringUtf8();
+            name_ = s;
+            return s;
+          } else {
+            return (java.lang.String) ref;
+          }
+        }
+        /**
+         * <pre>
+         * A name for the Tensor. Simple variables have only one
+         * `SerializedTensor` named "VARIABLE_VALUE" by convention. This value may
+         * be restored on object creation as an optimization.
+         * </pre>
+         *
+         * <code>string name = 1;</code>
+         */
+        public com.google.protobuf.ByteString
+            getNameBytes() {
+          java.lang.Object ref = name_;
+          if (ref instanceof String) {
+            com.google.protobuf.ByteString b = 
+                com.google.protobuf.ByteString.copyFromUtf8(
+                    (java.lang.String) ref);
+            name_ = b;
+            return b;
+          } else {
+            return (com.google.protobuf.ByteString) ref;
+          }
+        }
+        /**
+         * <pre>
+         * A name for the Tensor. Simple variables have only one
+         * `SerializedTensor` named "VARIABLE_VALUE" by convention. This value may
+         * be restored on object creation as an optimization.
+         * </pre>
+         *
+         * <code>string name = 1;</code>
+         */
+        public Builder setName(
+            java.lang.String value) {
+          if (value == null) {
+    throw new NullPointerException();
+  }
+  
+          name_ = value;
+          onChanged();
+          return this;
+        }
+        /**
+         * <pre>
+         * A name for the Tensor. Simple variables have only one
+         * `SerializedTensor` named "VARIABLE_VALUE" by convention. This value may
+         * be restored on object creation as an optimization.
+         * </pre>
+         *
+         * <code>string name = 1;</code>
+         */
+        public Builder clearName() {
+          
+          name_ = getDefaultInstance().getName();
+          onChanged();
+          return this;
+        }
+        /**
+         * <pre>
+         * A name for the Tensor. Simple variables have only one
+         * `SerializedTensor` named "VARIABLE_VALUE" by convention. This value may
+         * be restored on object creation as an optimization.
+         * </pre>
+         *
+         * <code>string name = 1;</code>
+         */
+        public Builder setNameBytes(
+            com.google.protobuf.ByteString value) {
+          if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+          
+          name_ = value;
+          onChanged();
+          return this;
+        }
+
+        private java.lang.Object fullName_ = "";
+        /**
+         * <pre>
+         * The full name of the variable/tensor, if applicable. Used to allow
+         * name-based loading of checkpoints which were saved using an
+         * object-based API. Should match the checkpoint key which would have been
+         * assigned by tf.train.Saver.
+         * </pre>
+         *
+         * <code>string full_name = 2;</code>
+         */
+        public java.lang.String getFullName() {
+          java.lang.Object ref = fullName_;
+          if (!(ref instanceof java.lang.String)) {
+            com.google.protobuf.ByteString bs =
+                (com.google.protobuf.ByteString) ref;
+            java.lang.String s = bs.toStringUtf8();
+            fullName_ = s;
+            return s;
+          } else {
+            return (java.lang.String) ref;
+          }
+        }
+        /**
+         * <pre>
+         * The full name of the variable/tensor, if applicable. Used to allow
+         * name-based loading of checkpoints which were saved using an
+         * object-based API. Should match the checkpoint key which would have been
+         * assigned by tf.train.Saver.
+         * </pre>
+         *
+         * <code>string full_name = 2;</code>
+         */
+        public com.google.protobuf.ByteString
+            getFullNameBytes() {
+          java.lang.Object ref = fullName_;
+          if (ref instanceof String) {
+            com.google.protobuf.ByteString b = 
+                com.google.protobuf.ByteString.copyFromUtf8(
+                    (java.lang.String) ref);
+            fullName_ = b;
+            return b;
+          } else {
+            return (com.google.protobuf.ByteString) ref;
+          }
+        }
+        /**
+         * <pre>
+         * The full name of the variable/tensor, if applicable. Used to allow
+         * name-based loading of checkpoints which were saved using an
+         * object-based API. Should match the checkpoint key which would have been
+         * assigned by tf.train.Saver.
+         * </pre>
+         *
+         * <code>string full_name = 2;</code>
+         */
+        public Builder setFullName(
+            java.lang.String value) {
+          if (value == null) {
+    throw new NullPointerException();
+  }
+  
+          fullName_ = value;
+          onChanged();
+          return this;
+        }
+        /**
+         * <pre>
+         * The full name of the variable/tensor, if applicable. Used to allow
+         * name-based loading of checkpoints which were saved using an
+         * object-based API. Should match the checkpoint key which would have been
+         * assigned by tf.train.Saver.
+         * </pre>
+         *
+         * <code>string full_name = 2;</code>
+         */
+        public Builder clearFullName() {
+          
+          fullName_ = getDefaultInstance().getFullName();
+          onChanged();
+          return this;
+        }
+        /**
+         * <pre>
+         * The full name of the variable/tensor, if applicable. Used to allow
+         * name-based loading of checkpoints which were saved using an
+         * object-based API. Should match the checkpoint key which would have been
+         * assigned by tf.train.Saver.
+         * </pre>
+         *
+         * <code>string full_name = 2;</code>
+         */
+        public Builder setFullNameBytes(
+            com.google.protobuf.ByteString value) {
+          if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+          
+          fullName_ = value;
+          onChanged();
+          return this;
+        }
+
+        private java.lang.Object checkpointKey_ = "";
+        /**
+         * <pre>
+         * The generated name of the Tensor in the checkpoint.
+         * </pre>
+         *
+         * <code>string checkpoint_key = 3;</code>
+         */
+        public java.lang.String getCheckpointKey() {
+          java.lang.Object ref = checkpointKey_;
+          if (!(ref instanceof java.lang.String)) {
+            com.google.protobuf.ByteString bs =
+                (com.google.protobuf.ByteString) ref;
+            java.lang.String s = bs.toStringUtf8();
+            checkpointKey_ = s;
+            return s;
+          } else {
+            return (java.lang.String) ref;
+          }
+        }
+        /**
+         * <pre>
+         * The generated name of the Tensor in the checkpoint.
+         * </pre>
+         *
+         * <code>string checkpoint_key = 3;</code>
+         */
+        public com.google.protobuf.ByteString
+            getCheckpointKeyBytes() {
+          java.lang.Object ref = checkpointKey_;
+          if (ref instanceof String) {
+            com.google.protobuf.ByteString b = 
+                com.google.protobuf.ByteString.copyFromUtf8(
+                    (java.lang.String) ref);
+            checkpointKey_ = b;
+            return b;
+          } else {
+            return (com.google.protobuf.ByteString) ref;
+          }
+        }
+        /**
+         * <pre>
+         * The generated name of the Tensor in the checkpoint.
+         * </pre>
+         *
+         * <code>string checkpoint_key = 3;</code>
+         */
+        public Builder setCheckpointKey(
+            java.lang.String value) {
+          if (value == null) {
+    throw new NullPointerException();
+  }
+  
+          checkpointKey_ = value;
+          onChanged();
+          return this;
+        }
+        /**
+         * <pre>
+         * The generated name of the Tensor in the checkpoint.
+         * </pre>
+         *
+         * <code>string checkpoint_key = 3;</code>
+         */
+        public Builder clearCheckpointKey() {
+          
+          checkpointKey_ = getDefaultInstance().getCheckpointKey();
+          onChanged();
+          return this;
+        }
+        /**
+         * <pre>
+         * The generated name of the Tensor in the checkpoint.
+         * </pre>
+         *
+         * <code>string checkpoint_key = 3;</code>
+         */
+        public Builder setCheckpointKeyBytes(
+            com.google.protobuf.ByteString value) {
+          if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+          
+          checkpointKey_ = value;
+          onChanged();
+          return this;
+        }
+
+        private boolean optionalRestore_ ;
+        /**
+         * <pre>
+         * Whether checkpoints should be considered as matching even without this
+         * value restored. Used for non-critical values which don't affect the
+         * TensorFlow graph, such as layer configurations.
+         * </pre>
+         *
+         * <code>bool optional_restore = 4;</code>
+         */
+        public boolean getOptionalRestore() {
+          return optionalRestore_;
+        }
+        /**
+         * <pre>
+         * Whether checkpoints should be considered as matching even without this
+         * value restored. Used for non-critical values which don't affect the
+         * TensorFlow graph, such as layer configurations.
+         * </pre>
+         *
+         * <code>bool optional_restore = 4;</code>
+         */
+        public Builder setOptionalRestore(boolean value) {
+          
+          optionalRestore_ = value;
+          onChanged();
+          return this;
+        }
+        /**
+         * <pre>
+         * Whether checkpoints should be considered as matching even without this
+         * value restored. Used for non-critical values which don't affect the
+         * TensorFlow graph, such as layer configurations.
+         * </pre>
+         *
+         * <code>bool optional_restore = 4;</code>
+         */
+        public Builder clearOptionalRestore() {
+          
+          optionalRestore_ = false;
+          onChanged();
+          return this;
+        }
+        @java.lang.Override
+        public final Builder setUnknownFields(
+            final com.google.protobuf.UnknownFieldSet unknownFields) {
+          return super.setUnknownFields(unknownFields);
+        }
+
+        @java.lang.Override
+        public final Builder mergeUnknownFields(
+            final com.google.protobuf.UnknownFieldSet unknownFields) {
+          return super.mergeUnknownFields(unknownFields);
+        }
+
+
+        // @@protoc_insertion_point(builder_scope:tensorflow.TrackableObjectGraph.TrackableObject.SerializedTensor)
+      }
+
+      // @@protoc_insertion_point(class_scope:tensorflow.TrackableObjectGraph.TrackableObject.SerializedTensor)
+      private static final org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensor DEFAULT_INSTANCE;
+      static {
+        DEFAULT_INSTANCE = new org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensor();
+      }
+
+      public static org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensor getDefaultInstance() {
+        return DEFAULT_INSTANCE;
+      }
+
+      private static final com.google.protobuf.Parser<SerializedTensor>
+          PARSER = new com.google.protobuf.AbstractParser<SerializedTensor>() {
+        @java.lang.Override
+        public SerializedTensor parsePartialFrom(
+            com.google.protobuf.CodedInputStream input,
+            com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+            throws com.google.protobuf.InvalidProtocolBufferException {
+          return new SerializedTensor(input, extensionRegistry);
+        }
+      };
+
+      public static com.google.protobuf.Parser<SerializedTensor> parser() {
+        return PARSER;
+      }
+
+      @java.lang.Override
+      public com.google.protobuf.Parser<SerializedTensor> getParserForType() {
+        return PARSER;
+      }
+
+      @java.lang.Override
+      public org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensor getDefaultInstanceForType() {
+        return DEFAULT_INSTANCE;
+      }
+
+    }
+
+    public interface SlotVariableReferenceOrBuilder extends
+        // @@protoc_insertion_point(interface_extends:tensorflow.TrackableObjectGraph.TrackableObject.SlotVariableReference)
+        com.google.protobuf.MessageOrBuilder {
+
+      /**
+       * <pre>
+       * An index into `TrackableObjectGraph.nodes`, indicating the
+       * variable object this slot was created for.
+       * </pre>
+       *
+       * <code>int32 original_variable_node_id = 1;</code>
+       */
+      int getOriginalVariableNodeId();
+
+      /**
+       * <pre>
+       * The name of the slot (e.g. "m"/"v").
+       * </pre>
+       *
+       * <code>string slot_name = 2;</code>
+       */
+      java.lang.String getSlotName();
+      /**
+       * <pre>
+       * The name of the slot (e.g. "m"/"v").
+       * </pre>
+       *
+       * <code>string slot_name = 2;</code>
+       */
+      com.google.protobuf.ByteString
+          getSlotNameBytes();
+
+      /**
+       * <pre>
+       * An index into `TrackableObjectGraph.nodes`, indicating the
+       * `Object` with the value of the slot variable.
+       * </pre>
+       *
+       * <code>int32 slot_variable_node_id = 3;</code>
+       */
+      int getSlotVariableNodeId();
+    }
+    /**
+     * Protobuf type {@code tensorflow.TrackableObjectGraph.TrackableObject.SlotVariableReference}
+     */
+    public  static final class SlotVariableReference extends
+        com.google.protobuf.GeneratedMessageV3 implements
+        // @@protoc_insertion_point(message_implements:tensorflow.TrackableObjectGraph.TrackableObject.SlotVariableReference)
+        SlotVariableReferenceOrBuilder {
+    private static final long serialVersionUID = 0L;
+      // Use SlotVariableReference.newBuilder() to construct.
+      private SlotVariableReference(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+        super(builder);
+      }
+      private SlotVariableReference() {
+        slotName_ = "";
+      }
+
+      @java.lang.Override
+      @SuppressWarnings({"unused"})
+      protected java.lang.Object newInstance(
+          UnusedPrivateParameter unused) {
+        return new SlotVariableReference();
+      }
+
+      @java.lang.Override
+      public final com.google.protobuf.UnknownFieldSet
+      getUnknownFields() {
+        return this.unknownFields;
+      }
+      private SlotVariableReference(
+          com.google.protobuf.CodedInputStream input,
+          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+          throws com.google.protobuf.InvalidProtocolBufferException {
+        this();
+        if (extensionRegistry == null) {
+          throw new java.lang.NullPointerException();
+        }
+        com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+            com.google.protobuf.UnknownFieldSet.newBuilder();
+        try {
+          boolean done = false;
+          while (!done) {
+            int tag = input.readTag();
+            switch (tag) {
+              case 0:
+                done = true;
+                break;
+              case 8: {
+
+                originalVariableNodeId_ = input.readInt32();
+                break;
+              }
+              case 18: {
+                java.lang.String s = input.readStringRequireUtf8();
+
+                slotName_ = s;
+                break;
+              }
+              case 24: {
+
+                slotVariableNodeId_ = input.readInt32();
+                break;
+              }
+              default: {
+                if (!parseUnknownField(
+                    input, unknownFields, extensionRegistry, tag)) {
+                  done = true;
+                }
+                break;
+              }
+            }
+          }
+        } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+          throw e.setUnfinishedMessage(this);
+        } catch (java.io.IOException e) {
+          throw new com.google.protobuf.InvalidProtocolBufferException(
+              e).setUnfinishedMessage(this);
+        } finally {
+          this.unknownFields = unknownFields.build();
+          makeExtensionsImmutable();
+        }
+      }
+      public static final com.google.protobuf.Descriptors.Descriptor
+          getDescriptor() {
+        return org.tensorflow.proto.framework.TrackableObjectGraphProtos.internal_static_tensorflow_TrackableObjectGraph_TrackableObject_SlotVariableReference_descriptor;
+      }
+
+      @java.lang.Override
+      protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+          internalGetFieldAccessorTable() {
+        return org.tensorflow.proto.framework.TrackableObjectGraphProtos.internal_static_tensorflow_TrackableObjectGraph_TrackableObject_SlotVariableReference_fieldAccessorTable
+            .ensureFieldAccessorsInitialized(
+                org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference.class, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference.Builder.class);
+      }
+
+      public static final int ORIGINAL_VARIABLE_NODE_ID_FIELD_NUMBER = 1;
+      private int originalVariableNodeId_;
+      /**
+       * <pre>
+       * An index into `TrackableObjectGraph.nodes`, indicating the
+       * variable object this slot was created for.
+       * </pre>
+       *
+       * <code>int32 original_variable_node_id = 1;</code>
+       */
+      public int getOriginalVariableNodeId() {
+        return originalVariableNodeId_;
+      }
+
+      public static final int SLOT_NAME_FIELD_NUMBER = 2;
+      private volatile java.lang.Object slotName_;
+      /**
+       * <pre>
+       * The name of the slot (e.g. "m"/"v").
+       * </pre>
+       *
+       * <code>string slot_name = 2;</code>
+       */
+      public java.lang.String getSlotName() {
+        java.lang.Object ref = slotName_;
+        if (ref instanceof java.lang.String) {
+          return (java.lang.String) ref;
+        } else {
+          com.google.protobuf.ByteString bs = 
+              (com.google.protobuf.ByteString) ref;
+          java.lang.String s = bs.toStringUtf8();
+          slotName_ = s;
+          return s;
+        }
+      }
+      /**
+       * <pre>
+       * The name of the slot (e.g. "m"/"v").
+       * </pre>
+       *
+       * <code>string slot_name = 2;</code>
+       */
+      public com.google.protobuf.ByteString
+          getSlotNameBytes() {
+        java.lang.Object ref = slotName_;
+        if (ref instanceof java.lang.String) {
+          com.google.protobuf.ByteString b = 
+              com.google.protobuf.ByteString.copyFromUtf8(
+                  (java.lang.String) ref);
+          slotName_ = b;
+          return b;
+        } else {
+          return (com.google.protobuf.ByteString) ref;
+        }
+      }
+
+      public static final int SLOT_VARIABLE_NODE_ID_FIELD_NUMBER = 3;
+      private int slotVariableNodeId_;
+      /**
+       * <pre>
+       * An index into `TrackableObjectGraph.nodes`, indicating the
+       * `Object` with the value of the slot variable.
+       * </pre>
+       *
+       * <code>int32 slot_variable_node_id = 3;</code>
+       */
+      public int getSlotVariableNodeId() {
+        return slotVariableNodeId_;
+      }
+
+      private byte memoizedIsInitialized = -1;
+      @java.lang.Override
+      public final boolean isInitialized() {
+        byte isInitialized = memoizedIsInitialized;
+        if (isInitialized == 1) return true;
+        if (isInitialized == 0) return false;
+
+        memoizedIsInitialized = 1;
+        return true;
+      }
+
+      @java.lang.Override
+      public void writeTo(com.google.protobuf.CodedOutputStream output)
+                          throws java.io.IOException {
+        if (originalVariableNodeId_ != 0) {
+          output.writeInt32(1, originalVariableNodeId_);
+        }
+        if (!getSlotNameBytes().isEmpty()) {
+          com.google.protobuf.GeneratedMessageV3.writeString(output, 2, slotName_);
+        }
+        if (slotVariableNodeId_ != 0) {
+          output.writeInt32(3, slotVariableNodeId_);
+        }
+        unknownFields.writeTo(output);
+      }
+
+      @java.lang.Override
+      public int getSerializedSize() {
+        int size = memoizedSize;
+        if (size != -1) return size;
+
+        size = 0;
+        if (originalVariableNodeId_ != 0) {
+          size += com.google.protobuf.CodedOutputStream
+            .computeInt32Size(1, originalVariableNodeId_);
+        }
+        if (!getSlotNameBytes().isEmpty()) {
+          size += com.google.protobuf.GeneratedMessageV3.computeStringSize(2, slotName_);
+        }
+        if (slotVariableNodeId_ != 0) {
+          size += com.google.protobuf.CodedOutputStream
+            .computeInt32Size(3, slotVariableNodeId_);
+        }
+        size += unknownFields.getSerializedSize();
+        memoizedSize = size;
+        return size;
+      }
+
+      @java.lang.Override
+      public boolean equals(final java.lang.Object obj) {
+        if (obj == this) {
+         return true;
+        }
+        if (!(obj instanceof org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference)) {
+          return super.equals(obj);
+        }
+        org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference other = (org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference) obj;
+
+        if (getOriginalVariableNodeId()
+            != other.getOriginalVariableNodeId()) return false;
+        if (!getSlotName()
+            .equals(other.getSlotName())) return false;
+        if (getSlotVariableNodeId()
+            != other.getSlotVariableNodeId()) return false;
+        if (!unknownFields.equals(other.unknownFields)) return false;
+        return true;
+      }
+
+      @java.lang.Override
+      public int hashCode() {
+        if (memoizedHashCode != 0) {
+          return memoizedHashCode;
+        }
+        int hash = 41;
+        hash = (19 * hash) + getDescriptor().hashCode();
+        hash = (37 * hash) + ORIGINAL_VARIABLE_NODE_ID_FIELD_NUMBER;
+        hash = (53 * hash) + getOriginalVariableNodeId();
+        hash = (37 * hash) + SLOT_NAME_FIELD_NUMBER;
+        hash = (53 * hash) + getSlotName().hashCode();
+        hash = (37 * hash) + SLOT_VARIABLE_NODE_ID_FIELD_NUMBER;
+        hash = (53 * hash) + getSlotVariableNodeId();
+        hash = (29 * hash) + unknownFields.hashCode();
+        memoizedHashCode = hash;
+        return hash;
+      }
+
+      public static org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference parseFrom(
+          java.nio.ByteBuffer data)
+          throws com.google.protobuf.InvalidProtocolBufferException {
+        return PARSER.parseFrom(data);
+      }
+      public static org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference parseFrom(
+          java.nio.ByteBuffer data,
+          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+          throws com.google.protobuf.InvalidProtocolBufferException {
+        return PARSER.parseFrom(data, extensionRegistry);
+      }
+      public static org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference parseFrom(
+          com.google.protobuf.ByteString data)
+          throws com.google.protobuf.InvalidProtocolBufferException {
+        return PARSER.parseFrom(data);
+      }
+      public static org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference parseFrom(
+          com.google.protobuf.ByteString data,
+          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+          throws com.google.protobuf.InvalidProtocolBufferException {
+        return PARSER.parseFrom(data, extensionRegistry);
+      }
+      public static org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference parseFrom(byte[] data)
+          throws com.google.protobuf.InvalidProtocolBufferException {
+        return PARSER.parseFrom(data);
+      }
+      public static org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference parseFrom(
+          byte[] data,
+          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+          throws com.google.protobuf.InvalidProtocolBufferException {
+        return PARSER.parseFrom(data, extensionRegistry);
+      }
+      public static org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference parseFrom(java.io.InputStream input)
+          throws java.io.IOException {
+        return com.google.protobuf.GeneratedMessageV3
+            .parseWithIOException(PARSER, input);
+      }
+      public static org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference parseFrom(
+          java.io.InputStream input,
+          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+          throws java.io.IOException {
+        return com.google.protobuf.GeneratedMessageV3
+            .parseWithIOException(PARSER, input, extensionRegistry);
+      }
+      public static org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference parseDelimitedFrom(java.io.InputStream input)
+          throws java.io.IOException {
+        return com.google.protobuf.GeneratedMessageV3
+            .parseDelimitedWithIOException(PARSER, input);
+      }
+      public static org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference parseDelimitedFrom(
+          java.io.InputStream input,
+          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+          throws java.io.IOException {
+        return com.google.protobuf.GeneratedMessageV3
+            .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+      }
+      public static org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference parseFrom(
+          com.google.protobuf.CodedInputStream input)
+          throws java.io.IOException {
+        return com.google.protobuf.GeneratedMessageV3
+            .parseWithIOException(PARSER, input);
+      }
+      public static org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference parseFrom(
+          com.google.protobuf.CodedInputStream input,
+          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+          throws java.io.IOException {
+        return com.google.protobuf.GeneratedMessageV3
+            .parseWithIOException(PARSER, input, extensionRegistry);
+      }
+
+      @java.lang.Override
+      public Builder newBuilderForType() { return newBuilder(); }
+      public static Builder newBuilder() {
+        return DEFAULT_INSTANCE.toBuilder();
+      }
+      public static Builder newBuilder(org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference prototype) {
+        return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+      }
+      @java.lang.Override
+      public Builder toBuilder() {
+        return this == DEFAULT_INSTANCE
+            ? new Builder() : new Builder().mergeFrom(this);
+      }
+
+      @java.lang.Override
+      protected Builder newBuilderForType(
+          com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+        Builder builder = new Builder(parent);
+        return builder;
+      }
+      /**
+       * Protobuf type {@code tensorflow.TrackableObjectGraph.TrackableObject.SlotVariableReference}
+       */
+      public static final class Builder extends
+          com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+          // @@protoc_insertion_point(builder_implements:tensorflow.TrackableObjectGraph.TrackableObject.SlotVariableReference)
+          org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReferenceOrBuilder {
+        public static final com.google.protobuf.Descriptors.Descriptor
+            getDescriptor() {
+          return org.tensorflow.proto.framework.TrackableObjectGraphProtos.internal_static_tensorflow_TrackableObjectGraph_TrackableObject_SlotVariableReference_descriptor;
+        }
+
+        @java.lang.Override
+        protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+            internalGetFieldAccessorTable() {
+          return org.tensorflow.proto.framework.TrackableObjectGraphProtos.internal_static_tensorflow_TrackableObjectGraph_TrackableObject_SlotVariableReference_fieldAccessorTable
+              .ensureFieldAccessorsInitialized(
+                  org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference.class, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference.Builder.class);
+        }
+
+        // Construct using org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference.newBuilder()
+        private Builder() {
+          maybeForceBuilderInitialization();
+        }
+
+        private Builder(
+            com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+          super(parent);
+          maybeForceBuilderInitialization();
+        }
+        private void maybeForceBuilderInitialization() {
+          if (com.google.protobuf.GeneratedMessageV3
+                  .alwaysUseFieldBuilders) {
+          }
+        }
+        @java.lang.Override
+        public Builder clear() {
+          super.clear();
+          originalVariableNodeId_ = 0;
+
+          slotName_ = "";
+
+          slotVariableNodeId_ = 0;
+
+          return this;
+        }
+
+        @java.lang.Override
+        public com.google.protobuf.Descriptors.Descriptor
+            getDescriptorForType() {
+          return org.tensorflow.proto.framework.TrackableObjectGraphProtos.internal_static_tensorflow_TrackableObjectGraph_TrackableObject_SlotVariableReference_descriptor;
+        }
+
+        @java.lang.Override
+        public org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference getDefaultInstanceForType() {
+          return org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference.getDefaultInstance();
+        }
+
+        @java.lang.Override
+        public org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference build() {
+          org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference result = buildPartial();
+          if (!result.isInitialized()) {
+            throw newUninitializedMessageException(result);
+          }
+          return result;
+        }
+
+        @java.lang.Override
+        public org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference buildPartial() {
+          org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference result = new org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference(this);
+          result.originalVariableNodeId_ = originalVariableNodeId_;
+          result.slotName_ = slotName_;
+          result.slotVariableNodeId_ = slotVariableNodeId_;
+          onBuilt();
+          return result;
+        }
+
+        @java.lang.Override
+        public Builder clone() {
+          return super.clone();
+        }
+        @java.lang.Override
+        public Builder setField(
+            com.google.protobuf.Descriptors.FieldDescriptor field,
+            java.lang.Object value) {
+          return super.setField(field, value);
+        }
+        @java.lang.Override
+        public Builder clearField(
+            com.google.protobuf.Descriptors.FieldDescriptor field) {
+          return super.clearField(field);
+        }
+        @java.lang.Override
+        public Builder clearOneof(
+            com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+          return super.clearOneof(oneof);
+        }
+        @java.lang.Override
+        public Builder setRepeatedField(
+            com.google.protobuf.Descriptors.FieldDescriptor field,
+            int index, java.lang.Object value) {
+          return super.setRepeatedField(field, index, value);
+        }
+        @java.lang.Override
+        public Builder addRepeatedField(
+            com.google.protobuf.Descriptors.FieldDescriptor field,
+            java.lang.Object value) {
+          return super.addRepeatedField(field, value);
+        }
+        @java.lang.Override
+        public Builder mergeFrom(com.google.protobuf.Message other) {
+          if (other instanceof org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference) {
+            return mergeFrom((org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference)other);
+          } else {
+            super.mergeFrom(other);
+            return this;
+          }
+        }
+
+        public Builder mergeFrom(org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference other) {
+          if (other == org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference.getDefaultInstance()) return this;
+          if (other.getOriginalVariableNodeId() != 0) {
+            setOriginalVariableNodeId(other.getOriginalVariableNodeId());
+          }
+          if (!other.getSlotName().isEmpty()) {
+            slotName_ = other.slotName_;
+            onChanged();
+          }
+          if (other.getSlotVariableNodeId() != 0) {
+            setSlotVariableNodeId(other.getSlotVariableNodeId());
+          }
+          this.mergeUnknownFields(other.unknownFields);
+          onChanged();
+          return this;
+        }
+
+        @java.lang.Override
+        public final boolean isInitialized() {
+          return true;
+        }
+
+        @java.lang.Override
+        public Builder mergeFrom(
+            com.google.protobuf.CodedInputStream input,
+            com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+            throws java.io.IOException {
+          org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference parsedMessage = null;
+          try {
+            parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+          } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+            parsedMessage = (org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference) e.getUnfinishedMessage();
+            throw e.unwrapIOException();
+          } finally {
+            if (parsedMessage != null) {
+              mergeFrom(parsedMessage);
+            }
+          }
+          return this;
+        }
+
+        private int originalVariableNodeId_ ;
+        /**
+         * <pre>
+         * An index into `TrackableObjectGraph.nodes`, indicating the
+         * variable object this slot was created for.
+         * </pre>
+         *
+         * <code>int32 original_variable_node_id = 1;</code>
+         */
+        public int getOriginalVariableNodeId() {
+          return originalVariableNodeId_;
+        }
+        /**
+         * <pre>
+         * An index into `TrackableObjectGraph.nodes`, indicating the
+         * variable object this slot was created for.
+         * </pre>
+         *
+         * <code>int32 original_variable_node_id = 1;</code>
+         */
+        public Builder setOriginalVariableNodeId(int value) {
+          
+          originalVariableNodeId_ = value;
+          onChanged();
+          return this;
+        }
+        /**
+         * <pre>
+         * An index into `TrackableObjectGraph.nodes`, indicating the
+         * variable object this slot was created for.
+         * </pre>
+         *
+         * <code>int32 original_variable_node_id = 1;</code>
+         */
+        public Builder clearOriginalVariableNodeId() {
+          
+          originalVariableNodeId_ = 0;
+          onChanged();
+          return this;
+        }
+
+        private java.lang.Object slotName_ = "";
+        /**
+         * <pre>
+         * The name of the slot (e.g. "m"/"v").
+         * </pre>
+         *
+         * <code>string slot_name = 2;</code>
+         */
+        public java.lang.String getSlotName() {
+          java.lang.Object ref = slotName_;
+          if (!(ref instanceof java.lang.String)) {
+            com.google.protobuf.ByteString bs =
+                (com.google.protobuf.ByteString) ref;
+            java.lang.String s = bs.toStringUtf8();
+            slotName_ = s;
+            return s;
+          } else {
+            return (java.lang.String) ref;
+          }
+        }
+        /**
+         * <pre>
+         * The name of the slot (e.g. "m"/"v").
+         * </pre>
+         *
+         * <code>string slot_name = 2;</code>
+         */
+        public com.google.protobuf.ByteString
+            getSlotNameBytes() {
+          java.lang.Object ref = slotName_;
+          if (ref instanceof String) {
+            com.google.protobuf.ByteString b = 
+                com.google.protobuf.ByteString.copyFromUtf8(
+                    (java.lang.String) ref);
+            slotName_ = b;
+            return b;
+          } else {
+            return (com.google.protobuf.ByteString) ref;
+          }
+        }
+        /**
+         * <pre>
+         * The name of the slot (e.g. "m"/"v").
+         * </pre>
+         *
+         * <code>string slot_name = 2;</code>
+         */
+        public Builder setSlotName(
+            java.lang.String value) {
+          if (value == null) {
+    throw new NullPointerException();
+  }
+  
+          slotName_ = value;
+          onChanged();
+          return this;
+        }
+        /**
+         * <pre>
+         * The name of the slot (e.g. "m"/"v").
+         * </pre>
+         *
+         * <code>string slot_name = 2;</code>
+         */
+        public Builder clearSlotName() {
+          
+          slotName_ = getDefaultInstance().getSlotName();
+          onChanged();
+          return this;
+        }
+        /**
+         * <pre>
+         * The name of the slot (e.g. "m"/"v").
+         * </pre>
+         *
+         * <code>string slot_name = 2;</code>
+         */
+        public Builder setSlotNameBytes(
+            com.google.protobuf.ByteString value) {
+          if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+          
+          slotName_ = value;
+          onChanged();
+          return this;
+        }
+
+        private int slotVariableNodeId_ ;
+        /**
+         * <pre>
+         * An index into `TrackableObjectGraph.nodes`, indicating the
+         * `Object` with the value of the slot variable.
+         * </pre>
+         *
+         * <code>int32 slot_variable_node_id = 3;</code>
+         */
+        public int getSlotVariableNodeId() {
+          return slotVariableNodeId_;
+        }
+        /**
+         * <pre>
+         * An index into `TrackableObjectGraph.nodes`, indicating the
+         * `Object` with the value of the slot variable.
+         * </pre>
+         *
+         * <code>int32 slot_variable_node_id = 3;</code>
+         */
+        public Builder setSlotVariableNodeId(int value) {
+          
+          slotVariableNodeId_ = value;
+          onChanged();
+          return this;
+        }
+        /**
+         * <pre>
+         * An index into `TrackableObjectGraph.nodes`, indicating the
+         * `Object` with the value of the slot variable.
+         * </pre>
+         *
+         * <code>int32 slot_variable_node_id = 3;</code>
+         */
+        public Builder clearSlotVariableNodeId() {
+          
+          slotVariableNodeId_ = 0;
+          onChanged();
+          return this;
+        }
+        @java.lang.Override
+        public final Builder setUnknownFields(
+            final com.google.protobuf.UnknownFieldSet unknownFields) {
+          return super.setUnknownFields(unknownFields);
+        }
+
+        @java.lang.Override
+        public final Builder mergeUnknownFields(
+            final com.google.protobuf.UnknownFieldSet unknownFields) {
+          return super.mergeUnknownFields(unknownFields);
+        }
+
+
+        // @@protoc_insertion_point(builder_scope:tensorflow.TrackableObjectGraph.TrackableObject.SlotVariableReference)
+      }
+
+      // @@protoc_insertion_point(class_scope:tensorflow.TrackableObjectGraph.TrackableObject.SlotVariableReference)
+      private static final org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference DEFAULT_INSTANCE;
+      static {
+        DEFAULT_INSTANCE = new org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference();
+      }
+
+      public static org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference getDefaultInstance() {
+        return DEFAULT_INSTANCE;
+      }
+
+      private static final com.google.protobuf.Parser<SlotVariableReference>
+          PARSER = new com.google.protobuf.AbstractParser<SlotVariableReference>() {
+        @java.lang.Override
+        public SlotVariableReference parsePartialFrom(
+            com.google.protobuf.CodedInputStream input,
+            com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+            throws com.google.protobuf.InvalidProtocolBufferException {
+          return new SlotVariableReference(input, extensionRegistry);
+        }
+      };
+
+      public static com.google.protobuf.Parser<SlotVariableReference> parser() {
+        return PARSER;
+      }
+
+      @java.lang.Override
+      public com.google.protobuf.Parser<SlotVariableReference> getParserForType() {
+        return PARSER;
+      }
+
+      @java.lang.Override
+      public org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference getDefaultInstanceForType() {
+        return DEFAULT_INSTANCE;
+      }
+
+    }
+
+    public static final int CHILDREN_FIELD_NUMBER = 1;
+    private java.util.List<org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference> children_;
+    /**
+     * <pre>
+     * Objects which this object depends on.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.ObjectReference children = 1;</code>
+     */
+    public java.util.List<org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference> getChildrenList() {
+      return children_;
+    }
+    /**
+     * <pre>
+     * Objects which this object depends on.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.ObjectReference children = 1;</code>
+     */
+    public java.util.List<? extends org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReferenceOrBuilder> 
+        getChildrenOrBuilderList() {
+      return children_;
+    }
+    /**
+     * <pre>
+     * Objects which this object depends on.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.ObjectReference children = 1;</code>
+     */
+    public int getChildrenCount() {
+      return children_.size();
+    }
+    /**
+     * <pre>
+     * Objects which this object depends on.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.ObjectReference children = 1;</code>
+     */
+    public org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference getChildren(int index) {
+      return children_.get(index);
+    }
+    /**
+     * <pre>
+     * Objects which this object depends on.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.ObjectReference children = 1;</code>
+     */
+    public org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReferenceOrBuilder getChildrenOrBuilder(
+        int index) {
+      return children_.get(index);
+    }
+
+    public static final int ATTRIBUTES_FIELD_NUMBER = 2;
+    private java.util.List<org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensor> attributes_;
+    /**
+     * <pre>
+     * Serialized data specific to this object.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SerializedTensor attributes = 2;</code>
+     */
+    public java.util.List<org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensor> getAttributesList() {
+      return attributes_;
+    }
+    /**
+     * <pre>
+     * Serialized data specific to this object.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SerializedTensor attributes = 2;</code>
+     */
+    public java.util.List<? extends org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensorOrBuilder> 
+        getAttributesOrBuilderList() {
+      return attributes_;
+    }
+    /**
+     * <pre>
+     * Serialized data specific to this object.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SerializedTensor attributes = 2;</code>
+     */
+    public int getAttributesCount() {
+      return attributes_.size();
+    }
+    /**
+     * <pre>
+     * Serialized data specific to this object.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SerializedTensor attributes = 2;</code>
+     */
+    public org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensor getAttributes(int index) {
+      return attributes_.get(index);
+    }
+    /**
+     * <pre>
+     * Serialized data specific to this object.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SerializedTensor attributes = 2;</code>
+     */
+    public org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensorOrBuilder getAttributesOrBuilder(
+        int index) {
+      return attributes_.get(index);
+    }
+
+    public static final int SLOT_VARIABLES_FIELD_NUMBER = 3;
+    private java.util.List<org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference> slotVariables_;
+    /**
+     * <pre>
+     * Slot variables owned by this object.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SlotVariableReference slot_variables = 3;</code>
+     */
+    public java.util.List<org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference> getSlotVariablesList() {
+      return slotVariables_;
+    }
+    /**
+     * <pre>
+     * Slot variables owned by this object.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SlotVariableReference slot_variables = 3;</code>
+     */
+    public java.util.List<? extends org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReferenceOrBuilder> 
+        getSlotVariablesOrBuilderList() {
+      return slotVariables_;
+    }
+    /**
+     * <pre>
+     * Slot variables owned by this object.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SlotVariableReference slot_variables = 3;</code>
+     */
+    public int getSlotVariablesCount() {
+      return slotVariables_.size();
+    }
+    /**
+     * <pre>
+     * Slot variables owned by this object.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SlotVariableReference slot_variables = 3;</code>
+     */
+    public org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference getSlotVariables(int index) {
+      return slotVariables_.get(index);
+    }
+    /**
+     * <pre>
+     * Slot variables owned by this object.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SlotVariableReference slot_variables = 3;</code>
+     */
+    public org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReferenceOrBuilder getSlotVariablesOrBuilder(
+        int index) {
+      return slotVariables_.get(index);
+    }
+
+    private byte memoizedIsInitialized = -1;
+    @java.lang.Override
+    public final boolean isInitialized() {
+      byte isInitialized = memoizedIsInitialized;
+      if (isInitialized == 1) return true;
+      if (isInitialized == 0) return false;
+
+      memoizedIsInitialized = 1;
+      return true;
+    }
+
+    @java.lang.Override
+    public void writeTo(com.google.protobuf.CodedOutputStream output)
+                        throws java.io.IOException {
+      for (int i = 0; i < children_.size(); i++) {
+        output.writeMessage(1, children_.get(i));
+      }
+      for (int i = 0; i < attributes_.size(); i++) {
+        output.writeMessage(2, attributes_.get(i));
+      }
+      for (int i = 0; i < slotVariables_.size(); i++) {
+        output.writeMessage(3, slotVariables_.get(i));
+      }
+      unknownFields.writeTo(output);
+    }
+
+    @java.lang.Override
+    public int getSerializedSize() {
+      int size = memoizedSize;
+      if (size != -1) return size;
+
+      size = 0;
+      for (int i = 0; i < children_.size(); i++) {
+        size += com.google.protobuf.CodedOutputStream
+          .computeMessageSize(1, children_.get(i));
+      }
+      for (int i = 0; i < attributes_.size(); i++) {
+        size += com.google.protobuf.CodedOutputStream
+          .computeMessageSize(2, attributes_.get(i));
+      }
+      for (int i = 0; i < slotVariables_.size(); i++) {
+        size += com.google.protobuf.CodedOutputStream
+          .computeMessageSize(3, slotVariables_.get(i));
+      }
+      size += unknownFields.getSerializedSize();
+      memoizedSize = size;
+      return size;
+    }
+
+    @java.lang.Override
+    public boolean equals(final java.lang.Object obj) {
+      if (obj == this) {
+       return true;
+      }
+      if (!(obj instanceof org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject)) {
+        return super.equals(obj);
+      }
+      org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject other = (org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject) obj;
+
+      if (!getChildrenList()
+          .equals(other.getChildrenList())) return false;
+      if (!getAttributesList()
+          .equals(other.getAttributesList())) return false;
+      if (!getSlotVariablesList()
+          .equals(other.getSlotVariablesList())) return false;
+      if (!unknownFields.equals(other.unknownFields)) return false;
+      return true;
+    }
+
+    @java.lang.Override
+    public int hashCode() {
+      if (memoizedHashCode != 0) {
+        return memoizedHashCode;
+      }
+      int hash = 41;
+      hash = (19 * hash) + getDescriptor().hashCode();
+      if (getChildrenCount() > 0) {
+        hash = (37 * hash) + CHILDREN_FIELD_NUMBER;
+        hash = (53 * hash) + getChildrenList().hashCode();
+      }
+      if (getAttributesCount() > 0) {
+        hash = (37 * hash) + ATTRIBUTES_FIELD_NUMBER;
+        hash = (53 * hash) + getAttributesList().hashCode();
+      }
+      if (getSlotVariablesCount() > 0) {
+        hash = (37 * hash) + SLOT_VARIABLES_FIELD_NUMBER;
+        hash = (53 * hash) + getSlotVariablesList().hashCode();
+      }
+      hash = (29 * hash) + unknownFields.hashCode();
+      memoizedHashCode = hash;
+      return hash;
+    }
+
+    public static org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject parseFrom(
+        java.nio.ByteBuffer data)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data);
+    }
+    public static org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject parseFrom(
+        java.nio.ByteBuffer data,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject parseFrom(
+        com.google.protobuf.ByteString data)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data);
+    }
+    public static org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject parseFrom(
+        com.google.protobuf.ByteString data,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject parseFrom(byte[] data)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data);
+    }
+    public static org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject parseFrom(
+        byte[] data,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return PARSER.parseFrom(data, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject parseFrom(java.io.InputStream input)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input);
+    }
+    public static org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject parseFrom(
+        java.io.InputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject parseDelimitedFrom(java.io.InputStream input)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseDelimitedWithIOException(PARSER, input);
+    }
+    public static org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject parseDelimitedFrom(
+        java.io.InputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+    }
+    public static org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject parseFrom(
+        com.google.protobuf.CodedInputStream input)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input);
+    }
+    public static org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject parseFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      return com.google.protobuf.GeneratedMessageV3
+          .parseWithIOException(PARSER, input, extensionRegistry);
+    }
+
+    @java.lang.Override
+    public Builder newBuilderForType() { return newBuilder(); }
+    public static Builder newBuilder() {
+      return DEFAULT_INSTANCE.toBuilder();
+    }
+    public static Builder newBuilder(org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject prototype) {
+      return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+    }
+    @java.lang.Override
+    public Builder toBuilder() {
+      return this == DEFAULT_INSTANCE
+          ? new Builder() : new Builder().mergeFrom(this);
+    }
+
+    @java.lang.Override
+    protected Builder newBuilderForType(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      Builder builder = new Builder(parent);
+      return builder;
+    }
+    /**
+     * Protobuf type {@code tensorflow.TrackableObjectGraph.TrackableObject}
+     */
+    public static final class Builder extends
+        com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+        // @@protoc_insertion_point(builder_implements:tensorflow.TrackableObjectGraph.TrackableObject)
+        org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObjectOrBuilder {
+      public static final com.google.protobuf.Descriptors.Descriptor
+          getDescriptor() {
+        return org.tensorflow.proto.framework.TrackableObjectGraphProtos.internal_static_tensorflow_TrackableObjectGraph_TrackableObject_descriptor;
+      }
+
+      @java.lang.Override
+      protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+          internalGetFieldAccessorTable() {
+        return org.tensorflow.proto.framework.TrackableObjectGraphProtos.internal_static_tensorflow_TrackableObjectGraph_TrackableObject_fieldAccessorTable
+            .ensureFieldAccessorsInitialized(
+                org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.class, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.Builder.class);
+      }
+
+      // Construct using org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.newBuilder()
+      private Builder() {
+        maybeForceBuilderInitialization();
+      }
+
+      private Builder(
+          com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+        super(parent);
+        maybeForceBuilderInitialization();
+      }
+      private void maybeForceBuilderInitialization() {
+        if (com.google.protobuf.GeneratedMessageV3
+                .alwaysUseFieldBuilders) {
+          getChildrenFieldBuilder();
+          getAttributesFieldBuilder();
+          getSlotVariablesFieldBuilder();
+        }
+      }
+      @java.lang.Override
+      public Builder clear() {
+        super.clear();
+        if (childrenBuilder_ == null) {
+          children_ = java.util.Collections.emptyList();
+          bitField0_ = (bitField0_ & ~0x00000001);
+        } else {
+          childrenBuilder_.clear();
+        }
+        if (attributesBuilder_ == null) {
+          attributes_ = java.util.Collections.emptyList();
+          bitField0_ = (bitField0_ & ~0x00000002);
+        } else {
+          attributesBuilder_.clear();
+        }
+        if (slotVariablesBuilder_ == null) {
+          slotVariables_ = java.util.Collections.emptyList();
+          bitField0_ = (bitField0_ & ~0x00000004);
+        } else {
+          slotVariablesBuilder_.clear();
+        }
+        return this;
+      }
+
+      @java.lang.Override
+      public com.google.protobuf.Descriptors.Descriptor
+          getDescriptorForType() {
+        return org.tensorflow.proto.framework.TrackableObjectGraphProtos.internal_static_tensorflow_TrackableObjectGraph_TrackableObject_descriptor;
+      }
+
+      @java.lang.Override
+      public org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject getDefaultInstanceForType() {
+        return org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.getDefaultInstance();
+      }
+
+      @java.lang.Override
+      public org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject build() {
+        org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject result = buildPartial();
+        if (!result.isInitialized()) {
+          throw newUninitializedMessageException(result);
+        }
+        return result;
+      }
+
+      @java.lang.Override
+      public org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject buildPartial() {
+        org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject result = new org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject(this);
+        int from_bitField0_ = bitField0_;
+        if (childrenBuilder_ == null) {
+          if (((bitField0_ & 0x00000001) != 0)) {
+            children_ = java.util.Collections.unmodifiableList(children_);
+            bitField0_ = (bitField0_ & ~0x00000001);
+          }
+          result.children_ = children_;
+        } else {
+          result.children_ = childrenBuilder_.build();
+        }
+        if (attributesBuilder_ == null) {
+          if (((bitField0_ & 0x00000002) != 0)) {
+            attributes_ = java.util.Collections.unmodifiableList(attributes_);
+            bitField0_ = (bitField0_ & ~0x00000002);
+          }
+          result.attributes_ = attributes_;
+        } else {
+          result.attributes_ = attributesBuilder_.build();
+        }
+        if (slotVariablesBuilder_ == null) {
+          if (((bitField0_ & 0x00000004) != 0)) {
+            slotVariables_ = java.util.Collections.unmodifiableList(slotVariables_);
+            bitField0_ = (bitField0_ & ~0x00000004);
+          }
+          result.slotVariables_ = slotVariables_;
+        } else {
+          result.slotVariables_ = slotVariablesBuilder_.build();
+        }
+        onBuilt();
+        return result;
+      }
+
+      @java.lang.Override
+      public Builder clone() {
+        return super.clone();
+      }
+      @java.lang.Override
+      public Builder setField(
+          com.google.protobuf.Descriptors.FieldDescriptor field,
+          java.lang.Object value) {
+        return super.setField(field, value);
+      }
+      @java.lang.Override
+      public Builder clearField(
+          com.google.protobuf.Descriptors.FieldDescriptor field) {
+        return super.clearField(field);
+      }
+      @java.lang.Override
+      public Builder clearOneof(
+          com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+        return super.clearOneof(oneof);
+      }
+      @java.lang.Override
+      public Builder setRepeatedField(
+          com.google.protobuf.Descriptors.FieldDescriptor field,
+          int index, java.lang.Object value) {
+        return super.setRepeatedField(field, index, value);
+      }
+      @java.lang.Override
+      public Builder addRepeatedField(
+          com.google.protobuf.Descriptors.FieldDescriptor field,
+          java.lang.Object value) {
+        return super.addRepeatedField(field, value);
+      }
+      @java.lang.Override
+      public Builder mergeFrom(com.google.protobuf.Message other) {
+        if (other instanceof org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject) {
+          return mergeFrom((org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject)other);
+        } else {
+          super.mergeFrom(other);
+          return this;
+        }
+      }
+
+      public Builder mergeFrom(org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject other) {
+        if (other == org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.getDefaultInstance()) return this;
+        if (childrenBuilder_ == null) {
+          if (!other.children_.isEmpty()) {
+            if (children_.isEmpty()) {
+              children_ = other.children_;
+              bitField0_ = (bitField0_ & ~0x00000001);
+            } else {
+              ensureChildrenIsMutable();
+              children_.addAll(other.children_);
+            }
+            onChanged();
+          }
+        } else {
+          if (!other.children_.isEmpty()) {
+            if (childrenBuilder_.isEmpty()) {
+              childrenBuilder_.dispose();
+              childrenBuilder_ = null;
+              children_ = other.children_;
+              bitField0_ = (bitField0_ & ~0x00000001);
+              childrenBuilder_ = 
+                com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
+                   getChildrenFieldBuilder() : null;
+            } else {
+              childrenBuilder_.addAllMessages(other.children_);
+            }
+          }
+        }
+        if (attributesBuilder_ == null) {
+          if (!other.attributes_.isEmpty()) {
+            if (attributes_.isEmpty()) {
+              attributes_ = other.attributes_;
+              bitField0_ = (bitField0_ & ~0x00000002);
+            } else {
+              ensureAttributesIsMutable();
+              attributes_.addAll(other.attributes_);
+            }
+            onChanged();
+          }
+        } else {
+          if (!other.attributes_.isEmpty()) {
+            if (attributesBuilder_.isEmpty()) {
+              attributesBuilder_.dispose();
+              attributesBuilder_ = null;
+              attributes_ = other.attributes_;
+              bitField0_ = (bitField0_ & ~0x00000002);
+              attributesBuilder_ = 
+                com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
+                   getAttributesFieldBuilder() : null;
+            } else {
+              attributesBuilder_.addAllMessages(other.attributes_);
+            }
+          }
+        }
+        if (slotVariablesBuilder_ == null) {
+          if (!other.slotVariables_.isEmpty()) {
+            if (slotVariables_.isEmpty()) {
+              slotVariables_ = other.slotVariables_;
+              bitField0_ = (bitField0_ & ~0x00000004);
+            } else {
+              ensureSlotVariablesIsMutable();
+              slotVariables_.addAll(other.slotVariables_);
+            }
+            onChanged();
+          }
+        } else {
+          if (!other.slotVariables_.isEmpty()) {
+            if (slotVariablesBuilder_.isEmpty()) {
+              slotVariablesBuilder_.dispose();
+              slotVariablesBuilder_ = null;
+              slotVariables_ = other.slotVariables_;
+              bitField0_ = (bitField0_ & ~0x00000004);
+              slotVariablesBuilder_ = 
+                com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
+                   getSlotVariablesFieldBuilder() : null;
+            } else {
+              slotVariablesBuilder_.addAllMessages(other.slotVariables_);
+            }
+          }
+        }
+        this.mergeUnknownFields(other.unknownFields);
+        onChanged();
+        return this;
+      }
+
+      @java.lang.Override
+      public final boolean isInitialized() {
+        return true;
+      }
+
+      @java.lang.Override
+      public Builder mergeFrom(
+          com.google.protobuf.CodedInputStream input,
+          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+          throws java.io.IOException {
+        org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject parsedMessage = null;
+        try {
+          parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+        } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+          parsedMessage = (org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject) e.getUnfinishedMessage();
+          throw e.unwrapIOException();
+        } finally {
+          if (parsedMessage != null) {
+            mergeFrom(parsedMessage);
+          }
+        }
+        return this;
+      }
+      private int bitField0_;
+
+      private java.util.List<org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference> children_ =
+        java.util.Collections.emptyList();
+      private void ensureChildrenIsMutable() {
+        if (!((bitField0_ & 0x00000001) != 0)) {
+          children_ = new java.util.ArrayList<org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference>(children_);
+          bitField0_ |= 0x00000001;
+         }
+      }
+
+      private com.google.protobuf.RepeatedFieldBuilderV3<
+          org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference.Builder, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReferenceOrBuilder> childrenBuilder_;
+
+      /**
+       * <pre>
+       * Objects which this object depends on.
+       * </pre>
+       *
+       * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.ObjectReference children = 1;</code>
+       */
+      public java.util.List<org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference> getChildrenList() {
+        if (childrenBuilder_ == null) {
+          return java.util.Collections.unmodifiableList(children_);
+        } else {
+          return childrenBuilder_.getMessageList();
+        }
+      }
+      /**
+       * <pre>
+       * Objects which this object depends on.
+       * </pre>
+       *
+       * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.ObjectReference children = 1;</code>
+       */
+      public int getChildrenCount() {
+        if (childrenBuilder_ == null) {
+          return children_.size();
+        } else {
+          return childrenBuilder_.getCount();
+        }
+      }
+      /**
+       * <pre>
+       * Objects which this object depends on.
+       * </pre>
+       *
+       * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.ObjectReference children = 1;</code>
+       */
+      public org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference getChildren(int index) {
+        if (childrenBuilder_ == null) {
+          return children_.get(index);
+        } else {
+          return childrenBuilder_.getMessage(index);
+        }
+      }
+      /**
+       * <pre>
+       * Objects which this object depends on.
+       * </pre>
+       *
+       * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.ObjectReference children = 1;</code>
+       */
+      public Builder setChildren(
+          int index, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference value) {
+        if (childrenBuilder_ == null) {
+          if (value == null) {
+            throw new NullPointerException();
+          }
+          ensureChildrenIsMutable();
+          children_.set(index, value);
+          onChanged();
+        } else {
+          childrenBuilder_.setMessage(index, value);
+        }
+        return this;
+      }
+      /**
+       * <pre>
+       * Objects which this object depends on.
+       * </pre>
+       *
+       * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.ObjectReference children = 1;</code>
+       */
+      public Builder setChildren(
+          int index, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference.Builder builderForValue) {
+        if (childrenBuilder_ == null) {
+          ensureChildrenIsMutable();
+          children_.set(index, builderForValue.build());
+          onChanged();
+        } else {
+          childrenBuilder_.setMessage(index, builderForValue.build());
+        }
+        return this;
+      }
+      /**
+       * <pre>
+       * Objects which this object depends on.
+       * </pre>
+       *
+       * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.ObjectReference children = 1;</code>
+       */
+      public Builder addChildren(org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference value) {
+        if (childrenBuilder_ == null) {
+          if (value == null) {
+            throw new NullPointerException();
+          }
+          ensureChildrenIsMutable();
+          children_.add(value);
+          onChanged();
+        } else {
+          childrenBuilder_.addMessage(value);
+        }
+        return this;
+      }
+      /**
+       * <pre>
+       * Objects which this object depends on.
+       * </pre>
+       *
+       * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.ObjectReference children = 1;</code>
+       */
+      public Builder addChildren(
+          int index, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference value) {
+        if (childrenBuilder_ == null) {
+          if (value == null) {
+            throw new NullPointerException();
+          }
+          ensureChildrenIsMutable();
+          children_.add(index, value);
+          onChanged();
+        } else {
+          childrenBuilder_.addMessage(index, value);
+        }
+        return this;
+      }
+      /**
+       * <pre>
+       * Objects which this object depends on.
+       * </pre>
+       *
+       * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.ObjectReference children = 1;</code>
+       */
+      public Builder addChildren(
+          org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference.Builder builderForValue) {
+        if (childrenBuilder_ == null) {
+          ensureChildrenIsMutable();
+          children_.add(builderForValue.build());
+          onChanged();
+        } else {
+          childrenBuilder_.addMessage(builderForValue.build());
+        }
+        return this;
+      }
+      /**
+       * <pre>
+       * Objects which this object depends on.
+       * </pre>
+       *
+       * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.ObjectReference children = 1;</code>
+       */
+      public Builder addChildren(
+          int index, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference.Builder builderForValue) {
+        if (childrenBuilder_ == null) {
+          ensureChildrenIsMutable();
+          children_.add(index, builderForValue.build());
+          onChanged();
+        } else {
+          childrenBuilder_.addMessage(index, builderForValue.build());
+        }
+        return this;
+      }
+      /**
+       * <pre>
+       * Objects which this object depends on.
+       * </pre>
+       *
+       * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.ObjectReference children = 1;</code>
+       */
+      public Builder addAllChildren(
+          java.lang.Iterable<? extends org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference> values) {
+        if (childrenBuilder_ == null) {
+          ensureChildrenIsMutable();
+          com.google.protobuf.AbstractMessageLite.Builder.addAll(
+              values, children_);
+          onChanged();
+        } else {
+          childrenBuilder_.addAllMessages(values);
+        }
+        return this;
+      }
+      /**
+       * <pre>
+       * Objects which this object depends on.
+       * </pre>
+       *
+       * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.ObjectReference children = 1;</code>
+       */
+      public Builder clearChildren() {
+        if (childrenBuilder_ == null) {
+          children_ = java.util.Collections.emptyList();
+          bitField0_ = (bitField0_ & ~0x00000001);
+          onChanged();
+        } else {
+          childrenBuilder_.clear();
+        }
+        return this;
+      }
+      /**
+       * <pre>
+       * Objects which this object depends on.
+       * </pre>
+       *
+       * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.ObjectReference children = 1;</code>
+       */
+      public Builder removeChildren(int index) {
+        if (childrenBuilder_ == null) {
+          ensureChildrenIsMutable();
+          children_.remove(index);
+          onChanged();
+        } else {
+          childrenBuilder_.remove(index);
+        }
+        return this;
+      }
+      /**
+       * <pre>
+       * Objects which this object depends on.
+       * </pre>
+       *
+       * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.ObjectReference children = 1;</code>
+       */
+      public org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference.Builder getChildrenBuilder(
+          int index) {
+        return getChildrenFieldBuilder().getBuilder(index);
+      }
+      /**
+       * <pre>
+       * Objects which this object depends on.
+       * </pre>
+       *
+       * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.ObjectReference children = 1;</code>
+       */
+      public org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReferenceOrBuilder getChildrenOrBuilder(
+          int index) {
+        if (childrenBuilder_ == null) {
+          return children_.get(index);  } else {
+          return childrenBuilder_.getMessageOrBuilder(index);
+        }
+      }
+      /**
+       * <pre>
+       * Objects which this object depends on.
+       * </pre>
+       *
+       * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.ObjectReference children = 1;</code>
+       */
+      public java.util.List<? extends org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReferenceOrBuilder> 
+           getChildrenOrBuilderList() {
+        if (childrenBuilder_ != null) {
+          return childrenBuilder_.getMessageOrBuilderList();
+        } else {
+          return java.util.Collections.unmodifiableList(children_);
+        }
+      }
+      /**
+       * <pre>
+       * Objects which this object depends on.
+       * </pre>
+       *
+       * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.ObjectReference children = 1;</code>
+       */
+      public org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference.Builder addChildrenBuilder() {
+        return getChildrenFieldBuilder().addBuilder(
+            org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference.getDefaultInstance());
+      }
+      /**
+       * <pre>
+       * Objects which this object depends on.
+       * </pre>
+       *
+       * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.ObjectReference children = 1;</code>
+       */
+      public org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference.Builder addChildrenBuilder(
+          int index) {
+        return getChildrenFieldBuilder().addBuilder(
+            index, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference.getDefaultInstance());
+      }
+      /**
+       * <pre>
+       * Objects which this object depends on.
+       * </pre>
+       *
+       * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.ObjectReference children = 1;</code>
+       */
+      public java.util.List<org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference.Builder> 
+           getChildrenBuilderList() {
+        return getChildrenFieldBuilder().getBuilderList();
+      }
+      private com.google.protobuf.RepeatedFieldBuilderV3<
+          org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference.Builder, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReferenceOrBuilder> 
+          getChildrenFieldBuilder() {
+        if (childrenBuilder_ == null) {
+          childrenBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
+              org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReference.Builder, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.ObjectReferenceOrBuilder>(
+                  children_,
+                  ((bitField0_ & 0x00000001) != 0),
+                  getParentForChildren(),
+                  isClean());
+          children_ = null;
+        }
+        return childrenBuilder_;
+      }
+
+      private java.util.List<org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensor> attributes_ =
+        java.util.Collections.emptyList();
+      private void ensureAttributesIsMutable() {
+        if (!((bitField0_ & 0x00000002) != 0)) {
+          attributes_ = new java.util.ArrayList<org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensor>(attributes_);
+          bitField0_ |= 0x00000002;
+         }
+      }
+
+      private com.google.protobuf.RepeatedFieldBuilderV3<
+          org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensor, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensor.Builder, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensorOrBuilder> attributesBuilder_;
+
+      /**
+       * <pre>
+       * Serialized data specific to this object.
+       * </pre>
+       *
+       * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SerializedTensor attributes = 2;</code>
+       */
+      public java.util.List<org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensor> getAttributesList() {
+        if (attributesBuilder_ == null) {
+          return java.util.Collections.unmodifiableList(attributes_);
+        } else {
+          return attributesBuilder_.getMessageList();
+        }
+      }
+      /**
+       * <pre>
+       * Serialized data specific to this object.
+       * </pre>
+       *
+       * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SerializedTensor attributes = 2;</code>
+       */
+      public int getAttributesCount() {
+        if (attributesBuilder_ == null) {
+          return attributes_.size();
+        } else {
+          return attributesBuilder_.getCount();
+        }
+      }
+      /**
+       * <pre>
+       * Serialized data specific to this object.
+       * </pre>
+       *
+       * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SerializedTensor attributes = 2;</code>
+       */
+      public org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensor getAttributes(int index) {
+        if (attributesBuilder_ == null) {
+          return attributes_.get(index);
+        } else {
+          return attributesBuilder_.getMessage(index);
+        }
+      }
+      /**
+       * <pre>
+       * Serialized data specific to this object.
+       * </pre>
+       *
+       * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SerializedTensor attributes = 2;</code>
+       */
+      public Builder setAttributes(
+          int index, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensor value) {
+        if (attributesBuilder_ == null) {
+          if (value == null) {
+            throw new NullPointerException();
+          }
+          ensureAttributesIsMutable();
+          attributes_.set(index, value);
+          onChanged();
+        } else {
+          attributesBuilder_.setMessage(index, value);
+        }
+        return this;
+      }
+      /**
+       * <pre>
+       * Serialized data specific to this object.
+       * </pre>
+       *
+       * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SerializedTensor attributes = 2;</code>
+       */
+      public Builder setAttributes(
+          int index, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensor.Builder builderForValue) {
+        if (attributesBuilder_ == null) {
+          ensureAttributesIsMutable();
+          attributes_.set(index, builderForValue.build());
+          onChanged();
+        } else {
+          attributesBuilder_.setMessage(index, builderForValue.build());
+        }
+        return this;
+      }
+      /**
+       * <pre>
+       * Serialized data specific to this object.
+       * </pre>
+       *
+       * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SerializedTensor attributes = 2;</code>
+       */
+      public Builder addAttributes(org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensor value) {
+        if (attributesBuilder_ == null) {
+          if (value == null) {
+            throw new NullPointerException();
+          }
+          ensureAttributesIsMutable();
+          attributes_.add(value);
+          onChanged();
+        } else {
+          attributesBuilder_.addMessage(value);
+        }
+        return this;
+      }
+      /**
+       * <pre>
+       * Serialized data specific to this object.
+       * </pre>
+       *
+       * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SerializedTensor attributes = 2;</code>
+       */
+      public Builder addAttributes(
+          int index, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensor value) {
+        if (attributesBuilder_ == null) {
+          if (value == null) {
+            throw new NullPointerException();
+          }
+          ensureAttributesIsMutable();
+          attributes_.add(index, value);
+          onChanged();
+        } else {
+          attributesBuilder_.addMessage(index, value);
+        }
+        return this;
+      }
+      /**
+       * <pre>
+       * Serialized data specific to this object.
+       * </pre>
+       *
+       * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SerializedTensor attributes = 2;</code>
+       */
+      public Builder addAttributes(
+          org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensor.Builder builderForValue) {
+        if (attributesBuilder_ == null) {
+          ensureAttributesIsMutable();
+          attributes_.add(builderForValue.build());
+          onChanged();
+        } else {
+          attributesBuilder_.addMessage(builderForValue.build());
+        }
+        return this;
+      }
+      /**
+       * <pre>
+       * Serialized data specific to this object.
+       * </pre>
+       *
+       * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SerializedTensor attributes = 2;</code>
+       */
+      public Builder addAttributes(
+          int index, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensor.Builder builderForValue) {
+        if (attributesBuilder_ == null) {
+          ensureAttributesIsMutable();
+          attributes_.add(index, builderForValue.build());
+          onChanged();
+        } else {
+          attributesBuilder_.addMessage(index, builderForValue.build());
+        }
+        return this;
+      }
+      /**
+       * <pre>
+       * Serialized data specific to this object.
+       * </pre>
+       *
+       * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SerializedTensor attributes = 2;</code>
+       */
+      public Builder addAllAttributes(
+          java.lang.Iterable<? extends org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensor> values) {
+        if (attributesBuilder_ == null) {
+          ensureAttributesIsMutable();
+          com.google.protobuf.AbstractMessageLite.Builder.addAll(
+              values, attributes_);
+          onChanged();
+        } else {
+          attributesBuilder_.addAllMessages(values);
+        }
+        return this;
+      }
+      /**
+       * <pre>
+       * Serialized data specific to this object.
+       * </pre>
+       *
+       * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SerializedTensor attributes = 2;</code>
+       */
+      public Builder clearAttributes() {
+        if (attributesBuilder_ == null) {
+          attributes_ = java.util.Collections.emptyList();
+          bitField0_ = (bitField0_ & ~0x00000002);
+          onChanged();
+        } else {
+          attributesBuilder_.clear();
+        }
+        return this;
+      }
+      /**
+       * <pre>
+       * Serialized data specific to this object.
+       * </pre>
+       *
+       * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SerializedTensor attributes = 2;</code>
+       */
+      public Builder removeAttributes(int index) {
+        if (attributesBuilder_ == null) {
+          ensureAttributesIsMutable();
+          attributes_.remove(index);
+          onChanged();
+        } else {
+          attributesBuilder_.remove(index);
+        }
+        return this;
+      }
+      /**
+       * <pre>
+       * Serialized data specific to this object.
+       * </pre>
+       *
+       * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SerializedTensor attributes = 2;</code>
+       */
+      public org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensor.Builder getAttributesBuilder(
+          int index) {
+        return getAttributesFieldBuilder().getBuilder(index);
+      }
+      /**
+       * <pre>
+       * Serialized data specific to this object.
+       * </pre>
+       *
+       * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SerializedTensor attributes = 2;</code>
+       */
+      public org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensorOrBuilder getAttributesOrBuilder(
+          int index) {
+        if (attributesBuilder_ == null) {
+          return attributes_.get(index);  } else {
+          return attributesBuilder_.getMessageOrBuilder(index);
+        }
+      }
+      /**
+       * <pre>
+       * Serialized data specific to this object.
+       * </pre>
+       *
+       * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SerializedTensor attributes = 2;</code>
+       */
+      public java.util.List<? extends org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensorOrBuilder> 
+           getAttributesOrBuilderList() {
+        if (attributesBuilder_ != null) {
+          return attributesBuilder_.getMessageOrBuilderList();
+        } else {
+          return java.util.Collections.unmodifiableList(attributes_);
+        }
+      }
+      /**
+       * <pre>
+       * Serialized data specific to this object.
+       * </pre>
+       *
+       * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SerializedTensor attributes = 2;</code>
+       */
+      public org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensor.Builder addAttributesBuilder() {
+        return getAttributesFieldBuilder().addBuilder(
+            org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensor.getDefaultInstance());
+      }
+      /**
+       * <pre>
+       * Serialized data specific to this object.
+       * </pre>
+       *
+       * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SerializedTensor attributes = 2;</code>
+       */
+      public org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensor.Builder addAttributesBuilder(
+          int index) {
+        return getAttributesFieldBuilder().addBuilder(
+            index, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensor.getDefaultInstance());
+      }
+      /**
+       * <pre>
+       * Serialized data specific to this object.
+       * </pre>
+       *
+       * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SerializedTensor attributes = 2;</code>
+       */
+      public java.util.List<org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensor.Builder> 
+           getAttributesBuilderList() {
+        return getAttributesFieldBuilder().getBuilderList();
+      }
+      private com.google.protobuf.RepeatedFieldBuilderV3<
+          org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensor, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensor.Builder, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensorOrBuilder> 
+          getAttributesFieldBuilder() {
+        if (attributesBuilder_ == null) {
+          attributesBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
+              org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensor, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensor.Builder, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SerializedTensorOrBuilder>(
+                  attributes_,
+                  ((bitField0_ & 0x00000002) != 0),
+                  getParentForChildren(),
+                  isClean());
+          attributes_ = null;
+        }
+        return attributesBuilder_;
+      }
+
+      private java.util.List<org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference> slotVariables_ =
+        java.util.Collections.emptyList();
+      private void ensureSlotVariablesIsMutable() {
+        if (!((bitField0_ & 0x00000004) != 0)) {
+          slotVariables_ = new java.util.ArrayList<org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference>(slotVariables_);
+          bitField0_ |= 0x00000004;
+         }
+      }
+
+      private com.google.protobuf.RepeatedFieldBuilderV3<
+          org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference.Builder, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReferenceOrBuilder> slotVariablesBuilder_;
+
+      /**
+       * <pre>
+       * Slot variables owned by this object.
+       * </pre>
+       *
+       * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SlotVariableReference slot_variables = 3;</code>
+       */
+      public java.util.List<org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference> getSlotVariablesList() {
+        if (slotVariablesBuilder_ == null) {
+          return java.util.Collections.unmodifiableList(slotVariables_);
+        } else {
+          return slotVariablesBuilder_.getMessageList();
+        }
+      }
+      /**
+       * <pre>
+       * Slot variables owned by this object.
+       * </pre>
+       *
+       * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SlotVariableReference slot_variables = 3;</code>
+       */
+      public int getSlotVariablesCount() {
+        if (slotVariablesBuilder_ == null) {
+          return slotVariables_.size();
+        } else {
+          return slotVariablesBuilder_.getCount();
+        }
+      }
+      /**
+       * <pre>
+       * Slot variables owned by this object.
+       * </pre>
+       *
+       * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SlotVariableReference slot_variables = 3;</code>
+       */
+      public org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference getSlotVariables(int index) {
+        if (slotVariablesBuilder_ == null) {
+          return slotVariables_.get(index);
+        } else {
+          return slotVariablesBuilder_.getMessage(index);
+        }
+      }
+      /**
+       * <pre>
+       * Slot variables owned by this object.
+       * </pre>
+       *
+       * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SlotVariableReference slot_variables = 3;</code>
+       */
+      public Builder setSlotVariables(
+          int index, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference value) {
+        if (slotVariablesBuilder_ == null) {
+          if (value == null) {
+            throw new NullPointerException();
+          }
+          ensureSlotVariablesIsMutable();
+          slotVariables_.set(index, value);
+          onChanged();
+        } else {
+          slotVariablesBuilder_.setMessage(index, value);
+        }
+        return this;
+      }
+      /**
+       * <pre>
+       * Slot variables owned by this object.
+       * </pre>
+       *
+       * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SlotVariableReference slot_variables = 3;</code>
+       */
+      public Builder setSlotVariables(
+          int index, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference.Builder builderForValue) {
+        if (slotVariablesBuilder_ == null) {
+          ensureSlotVariablesIsMutable();
+          slotVariables_.set(index, builderForValue.build());
+          onChanged();
+        } else {
+          slotVariablesBuilder_.setMessage(index, builderForValue.build());
+        }
+        return this;
+      }
+      /**
+       * <pre>
+       * Slot variables owned by this object.
+       * </pre>
+       *
+       * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SlotVariableReference slot_variables = 3;</code>
+       */
+      public Builder addSlotVariables(org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference value) {
+        if (slotVariablesBuilder_ == null) {
+          if (value == null) {
+            throw new NullPointerException();
+          }
+          ensureSlotVariablesIsMutable();
+          slotVariables_.add(value);
+          onChanged();
+        } else {
+          slotVariablesBuilder_.addMessage(value);
+        }
+        return this;
+      }
+      /**
+       * <pre>
+       * Slot variables owned by this object.
+       * </pre>
+       *
+       * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SlotVariableReference slot_variables = 3;</code>
+       */
+      public Builder addSlotVariables(
+          int index, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference value) {
+        if (slotVariablesBuilder_ == null) {
+          if (value == null) {
+            throw new NullPointerException();
+          }
+          ensureSlotVariablesIsMutable();
+          slotVariables_.add(index, value);
+          onChanged();
+        } else {
+          slotVariablesBuilder_.addMessage(index, value);
+        }
+        return this;
+      }
+      /**
+       * <pre>
+       * Slot variables owned by this object.
+       * </pre>
+       *
+       * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SlotVariableReference slot_variables = 3;</code>
+       */
+      public Builder addSlotVariables(
+          org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference.Builder builderForValue) {
+        if (slotVariablesBuilder_ == null) {
+          ensureSlotVariablesIsMutable();
+          slotVariables_.add(builderForValue.build());
+          onChanged();
+        } else {
+          slotVariablesBuilder_.addMessage(builderForValue.build());
+        }
+        return this;
+      }
+      /**
+       * <pre>
+       * Slot variables owned by this object.
+       * </pre>
+       *
+       * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SlotVariableReference slot_variables = 3;</code>
+       */
+      public Builder addSlotVariables(
+          int index, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference.Builder builderForValue) {
+        if (slotVariablesBuilder_ == null) {
+          ensureSlotVariablesIsMutable();
+          slotVariables_.add(index, builderForValue.build());
+          onChanged();
+        } else {
+          slotVariablesBuilder_.addMessage(index, builderForValue.build());
+        }
+        return this;
+      }
+      /**
+       * <pre>
+       * Slot variables owned by this object.
+       * </pre>
+       *
+       * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SlotVariableReference slot_variables = 3;</code>
+       */
+      public Builder addAllSlotVariables(
+          java.lang.Iterable<? extends org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference> values) {
+        if (slotVariablesBuilder_ == null) {
+          ensureSlotVariablesIsMutable();
+          com.google.protobuf.AbstractMessageLite.Builder.addAll(
+              values, slotVariables_);
+          onChanged();
+        } else {
+          slotVariablesBuilder_.addAllMessages(values);
+        }
+        return this;
+      }
+      /**
+       * <pre>
+       * Slot variables owned by this object.
+       * </pre>
+       *
+       * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SlotVariableReference slot_variables = 3;</code>
+       */
+      public Builder clearSlotVariables() {
+        if (slotVariablesBuilder_ == null) {
+          slotVariables_ = java.util.Collections.emptyList();
+          bitField0_ = (bitField0_ & ~0x00000004);
+          onChanged();
+        } else {
+          slotVariablesBuilder_.clear();
+        }
+        return this;
+      }
+      /**
+       * <pre>
+       * Slot variables owned by this object.
+       * </pre>
+       *
+       * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SlotVariableReference slot_variables = 3;</code>
+       */
+      public Builder removeSlotVariables(int index) {
+        if (slotVariablesBuilder_ == null) {
+          ensureSlotVariablesIsMutable();
+          slotVariables_.remove(index);
+          onChanged();
+        } else {
+          slotVariablesBuilder_.remove(index);
+        }
+        return this;
+      }
+      /**
+       * <pre>
+       * Slot variables owned by this object.
+       * </pre>
+       *
+       * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SlotVariableReference slot_variables = 3;</code>
+       */
+      public org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference.Builder getSlotVariablesBuilder(
+          int index) {
+        return getSlotVariablesFieldBuilder().getBuilder(index);
+      }
+      /**
+       * <pre>
+       * Slot variables owned by this object.
+       * </pre>
+       *
+       * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SlotVariableReference slot_variables = 3;</code>
+       */
+      public org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReferenceOrBuilder getSlotVariablesOrBuilder(
+          int index) {
+        if (slotVariablesBuilder_ == null) {
+          return slotVariables_.get(index);  } else {
+          return slotVariablesBuilder_.getMessageOrBuilder(index);
+        }
+      }
+      /**
+       * <pre>
+       * Slot variables owned by this object.
+       * </pre>
+       *
+       * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SlotVariableReference slot_variables = 3;</code>
+       */
+      public java.util.List<? extends org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReferenceOrBuilder> 
+           getSlotVariablesOrBuilderList() {
+        if (slotVariablesBuilder_ != null) {
+          return slotVariablesBuilder_.getMessageOrBuilderList();
+        } else {
+          return java.util.Collections.unmodifiableList(slotVariables_);
+        }
+      }
+      /**
+       * <pre>
+       * Slot variables owned by this object.
+       * </pre>
+       *
+       * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SlotVariableReference slot_variables = 3;</code>
+       */
+      public org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference.Builder addSlotVariablesBuilder() {
+        return getSlotVariablesFieldBuilder().addBuilder(
+            org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference.getDefaultInstance());
+      }
+      /**
+       * <pre>
+       * Slot variables owned by this object.
+       * </pre>
+       *
+       * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SlotVariableReference slot_variables = 3;</code>
+       */
+      public org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference.Builder addSlotVariablesBuilder(
+          int index) {
+        return getSlotVariablesFieldBuilder().addBuilder(
+            index, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference.getDefaultInstance());
+      }
+      /**
+       * <pre>
+       * Slot variables owned by this object.
+       * </pre>
+       *
+       * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject.SlotVariableReference slot_variables = 3;</code>
+       */
+      public java.util.List<org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference.Builder> 
+           getSlotVariablesBuilderList() {
+        return getSlotVariablesFieldBuilder().getBuilderList();
+      }
+      private com.google.protobuf.RepeatedFieldBuilderV3<
+          org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference.Builder, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReferenceOrBuilder> 
+          getSlotVariablesFieldBuilder() {
+        if (slotVariablesBuilder_ == null) {
+          slotVariablesBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
+              org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReference.Builder, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.SlotVariableReferenceOrBuilder>(
+                  slotVariables_,
+                  ((bitField0_ & 0x00000004) != 0),
+                  getParentForChildren(),
+                  isClean());
+          slotVariables_ = null;
+        }
+        return slotVariablesBuilder_;
+      }
+      @java.lang.Override
+      public final Builder setUnknownFields(
+          final com.google.protobuf.UnknownFieldSet unknownFields) {
+        return super.setUnknownFields(unknownFields);
+      }
+
+      @java.lang.Override
+      public final Builder mergeUnknownFields(
+          final com.google.protobuf.UnknownFieldSet unknownFields) {
+        return super.mergeUnknownFields(unknownFields);
+      }
+
+
+      // @@protoc_insertion_point(builder_scope:tensorflow.TrackableObjectGraph.TrackableObject)
+    }
+
+    // @@protoc_insertion_point(class_scope:tensorflow.TrackableObjectGraph.TrackableObject)
+    private static final org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject DEFAULT_INSTANCE;
+    static {
+      DEFAULT_INSTANCE = new org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject();
+    }
+
+    public static org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject getDefaultInstance() {
+      return DEFAULT_INSTANCE;
+    }
+
+    private static final com.google.protobuf.Parser<TrackableObject>
+        PARSER = new com.google.protobuf.AbstractParser<TrackableObject>() {
+      @java.lang.Override
+      public TrackableObject parsePartialFrom(
+          com.google.protobuf.CodedInputStream input,
+          com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+          throws com.google.protobuf.InvalidProtocolBufferException {
+        return new TrackableObject(input, extensionRegistry);
+      }
+    };
+
+    public static com.google.protobuf.Parser<TrackableObject> parser() {
+      return PARSER;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Parser<TrackableObject> getParserForType() {
+      return PARSER;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject getDefaultInstanceForType() {
+      return DEFAULT_INSTANCE;
+    }
+
+  }
+
+  public static final int NODES_FIELD_NUMBER = 1;
+  private java.util.List<org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject> nodes_;
+  /**
+   * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject nodes = 1;</code>
+   */
+  public java.util.List<org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject> getNodesList() {
+    return nodes_;
+  }
+  /**
+   * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject nodes = 1;</code>
+   */
+  public java.util.List<? extends org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObjectOrBuilder> 
+      getNodesOrBuilderList() {
+    return nodes_;
+  }
+  /**
+   * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject nodes = 1;</code>
+   */
+  public int getNodesCount() {
+    return nodes_.size();
+  }
+  /**
+   * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject nodes = 1;</code>
+   */
+  public org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject getNodes(int index) {
+    return nodes_.get(index);
+  }
+  /**
+   * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject nodes = 1;</code>
+   */
+  public org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObjectOrBuilder getNodesOrBuilder(
+      int index) {
+    return nodes_.get(index);
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    for (int i = 0; i < nodes_.size(); i++) {
+      output.writeMessage(1, nodes_.get(i));
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    for (int i = 0; i < nodes_.size(); i++) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(1, nodes_.get(i));
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.framework.TrackableObjectGraph)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.framework.TrackableObjectGraph other = (org.tensorflow.proto.framework.TrackableObjectGraph) obj;
+
+    if (!getNodesList()
+        .equals(other.getNodesList())) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    if (getNodesCount() > 0) {
+      hash = (37 * hash) + NODES_FIELD_NUMBER;
+      hash = (53 * hash) + getNodesList().hashCode();
+    }
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.framework.TrackableObjectGraph parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.TrackableObjectGraph parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.TrackableObjectGraph parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.TrackableObjectGraph parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.TrackableObjectGraph parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.TrackableObjectGraph parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.TrackableObjectGraph parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.TrackableObjectGraph parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.TrackableObjectGraph parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.TrackableObjectGraph parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.TrackableObjectGraph parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.TrackableObjectGraph parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.framework.TrackableObjectGraph prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * Protobuf type {@code tensorflow.TrackableObjectGraph}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.TrackableObjectGraph)
+      org.tensorflow.proto.framework.TrackableObjectGraphOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.TrackableObjectGraphProtos.internal_static_tensorflow_TrackableObjectGraph_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.TrackableObjectGraphProtos.internal_static_tensorflow_TrackableObjectGraph_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.TrackableObjectGraph.class, org.tensorflow.proto.framework.TrackableObjectGraph.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.framework.TrackableObjectGraph.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+        getNodesFieldBuilder();
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      if (nodesBuilder_ == null) {
+        nodes_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000001);
+      } else {
+        nodesBuilder_.clear();
+      }
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.framework.TrackableObjectGraphProtos.internal_static_tensorflow_TrackableObjectGraph_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.TrackableObjectGraph getDefaultInstanceForType() {
+      return org.tensorflow.proto.framework.TrackableObjectGraph.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.TrackableObjectGraph build() {
+      org.tensorflow.proto.framework.TrackableObjectGraph result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.TrackableObjectGraph buildPartial() {
+      org.tensorflow.proto.framework.TrackableObjectGraph result = new org.tensorflow.proto.framework.TrackableObjectGraph(this);
+      int from_bitField0_ = bitField0_;
+      if (nodesBuilder_ == null) {
+        if (((bitField0_ & 0x00000001) != 0)) {
+          nodes_ = java.util.Collections.unmodifiableList(nodes_);
+          bitField0_ = (bitField0_ & ~0x00000001);
+        }
+        result.nodes_ = nodes_;
+      } else {
+        result.nodes_ = nodesBuilder_.build();
+      }
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.framework.TrackableObjectGraph) {
+        return mergeFrom((org.tensorflow.proto.framework.TrackableObjectGraph)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.framework.TrackableObjectGraph other) {
+      if (other == org.tensorflow.proto.framework.TrackableObjectGraph.getDefaultInstance()) return this;
+      if (nodesBuilder_ == null) {
+        if (!other.nodes_.isEmpty()) {
+          if (nodes_.isEmpty()) {
+            nodes_ = other.nodes_;
+            bitField0_ = (bitField0_ & ~0x00000001);
+          } else {
+            ensureNodesIsMutable();
+            nodes_.addAll(other.nodes_);
+          }
+          onChanged();
+        }
+      } else {
+        if (!other.nodes_.isEmpty()) {
+          if (nodesBuilder_.isEmpty()) {
+            nodesBuilder_.dispose();
+            nodesBuilder_ = null;
+            nodes_ = other.nodes_;
+            bitField0_ = (bitField0_ & ~0x00000001);
+            nodesBuilder_ = 
+              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
+                 getNodesFieldBuilder() : null;
+          } else {
+            nodesBuilder_.addAllMessages(other.nodes_);
+          }
+        }
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.framework.TrackableObjectGraph parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.framework.TrackableObjectGraph) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+    private int bitField0_;
+
+    private java.util.List<org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject> nodes_ =
+      java.util.Collections.emptyList();
+    private void ensureNodesIsMutable() {
+      if (!((bitField0_ & 0x00000001) != 0)) {
+        nodes_ = new java.util.ArrayList<org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject>(nodes_);
+        bitField0_ |= 0x00000001;
+       }
+    }
+
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.Builder, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObjectOrBuilder> nodesBuilder_;
+
+    /**
+     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject nodes = 1;</code>
+     */
+    public java.util.List<org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject> getNodesList() {
+      if (nodesBuilder_ == null) {
+        return java.util.Collections.unmodifiableList(nodes_);
+      } else {
+        return nodesBuilder_.getMessageList();
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject nodes = 1;</code>
+     */
+    public int getNodesCount() {
+      if (nodesBuilder_ == null) {
+        return nodes_.size();
+      } else {
+        return nodesBuilder_.getCount();
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject nodes = 1;</code>
+     */
+    public org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject getNodes(int index) {
+      if (nodesBuilder_ == null) {
+        return nodes_.get(index);
+      } else {
+        return nodesBuilder_.getMessage(index);
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject nodes = 1;</code>
+     */
+    public Builder setNodes(
+        int index, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject value) {
+      if (nodesBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureNodesIsMutable();
+        nodes_.set(index, value);
+        onChanged();
+      } else {
+        nodesBuilder_.setMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject nodes = 1;</code>
+     */
+    public Builder setNodes(
+        int index, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.Builder builderForValue) {
+      if (nodesBuilder_ == null) {
+        ensureNodesIsMutable();
+        nodes_.set(index, builderForValue.build());
+        onChanged();
+      } else {
+        nodesBuilder_.setMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject nodes = 1;</code>
+     */
+    public Builder addNodes(org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject value) {
+      if (nodesBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureNodesIsMutable();
+        nodes_.add(value);
+        onChanged();
+      } else {
+        nodesBuilder_.addMessage(value);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject nodes = 1;</code>
+     */
+    public Builder addNodes(
+        int index, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject value) {
+      if (nodesBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureNodesIsMutable();
+        nodes_.add(index, value);
+        onChanged();
+      } else {
+        nodesBuilder_.addMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject nodes = 1;</code>
+     */
+    public Builder addNodes(
+        org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.Builder builderForValue) {
+      if (nodesBuilder_ == null) {
+        ensureNodesIsMutable();
+        nodes_.add(builderForValue.build());
+        onChanged();
+      } else {
+        nodesBuilder_.addMessage(builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject nodes = 1;</code>
+     */
+    public Builder addNodes(
+        int index, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.Builder builderForValue) {
+      if (nodesBuilder_ == null) {
+        ensureNodesIsMutable();
+        nodes_.add(index, builderForValue.build());
+        onChanged();
+      } else {
+        nodesBuilder_.addMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject nodes = 1;</code>
+     */
+    public Builder addAllNodes(
+        java.lang.Iterable<? extends org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject> values) {
+      if (nodesBuilder_ == null) {
+        ensureNodesIsMutable();
+        com.google.protobuf.AbstractMessageLite.Builder.addAll(
+            values, nodes_);
+        onChanged();
+      } else {
+        nodesBuilder_.addAllMessages(values);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject nodes = 1;</code>
+     */
+    public Builder clearNodes() {
+      if (nodesBuilder_ == null) {
+        nodes_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000001);
+        onChanged();
+      } else {
+        nodesBuilder_.clear();
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject nodes = 1;</code>
+     */
+    public Builder removeNodes(int index) {
+      if (nodesBuilder_ == null) {
+        ensureNodesIsMutable();
+        nodes_.remove(index);
+        onChanged();
+      } else {
+        nodesBuilder_.remove(index);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject nodes = 1;</code>
+     */
+    public org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.Builder getNodesBuilder(
+        int index) {
+      return getNodesFieldBuilder().getBuilder(index);
+    }
+    /**
+     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject nodes = 1;</code>
+     */
+    public org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObjectOrBuilder getNodesOrBuilder(
+        int index) {
+      if (nodesBuilder_ == null) {
+        return nodes_.get(index);  } else {
+        return nodesBuilder_.getMessageOrBuilder(index);
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject nodes = 1;</code>
+     */
+    public java.util.List<? extends org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObjectOrBuilder> 
+         getNodesOrBuilderList() {
+      if (nodesBuilder_ != null) {
+        return nodesBuilder_.getMessageOrBuilderList();
+      } else {
+        return java.util.Collections.unmodifiableList(nodes_);
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject nodes = 1;</code>
+     */
+    public org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.Builder addNodesBuilder() {
+      return getNodesFieldBuilder().addBuilder(
+          org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.getDefaultInstance());
+    }
+    /**
+     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject nodes = 1;</code>
+     */
+    public org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.Builder addNodesBuilder(
+        int index) {
+      return getNodesFieldBuilder().addBuilder(
+          index, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.getDefaultInstance());
+    }
+    /**
+     * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject nodes = 1;</code>
+     */
+    public java.util.List<org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.Builder> 
+         getNodesBuilderList() {
+      return getNodesFieldBuilder().getBuilderList();
+    }
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.Builder, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObjectOrBuilder> 
+        getNodesFieldBuilder() {
+      if (nodesBuilder_ == null) {
+        nodesBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
+            org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject.Builder, org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObjectOrBuilder>(
+                nodes_,
+                ((bitField0_ & 0x00000001) != 0),
+                getParentForChildren(),
+                isClean());
+        nodes_ = null;
+      }
+      return nodesBuilder_;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.TrackableObjectGraph)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.TrackableObjectGraph)
+  private static final org.tensorflow.proto.framework.TrackableObjectGraph DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.TrackableObjectGraph();
+  }
+
+  public static org.tensorflow.proto.framework.TrackableObjectGraph getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<TrackableObjectGraph>
+      PARSER = new com.google.protobuf.AbstractParser<TrackableObjectGraph>() {
+    @java.lang.Override
+    public TrackableObjectGraph parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new TrackableObjectGraph(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<TrackableObjectGraph> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<TrackableObjectGraph> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.framework.TrackableObjectGraph getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TrackableObjectGraphOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TrackableObjectGraphOrBuilder.java
new file mode 100644
index 00000000000..b099d054d9a
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TrackableObjectGraphOrBuilder.java
@@ -0,0 +1,33 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/trackable_object_graph.proto
+
+package org.tensorflow.proto.framework;
+
+public interface TrackableObjectGraphOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.TrackableObjectGraph)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject nodes = 1;</code>
+   */
+  java.util.List<org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject> 
+      getNodesList();
+  /**
+   * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject nodes = 1;</code>
+   */
+  org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObject getNodes(int index);
+  /**
+   * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject nodes = 1;</code>
+   */
+  int getNodesCount();
+  /**
+   * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject nodes = 1;</code>
+   */
+  java.util.List<? extends org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObjectOrBuilder> 
+      getNodesOrBuilderList();
+  /**
+   * <code>repeated .tensorflow.TrackableObjectGraph.TrackableObject nodes = 1;</code>
+   */
+  org.tensorflow.proto.framework.TrackableObjectGraph.TrackableObjectOrBuilder getNodesOrBuilder(
+      int index);
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TrackableObjectGraphProtos.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TrackableObjectGraphProtos.java
new file mode 100644
index 00000000000..22a3eef371a
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TrackableObjectGraphProtos.java
@@ -0,0 +1,111 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/trackable_object_graph.proto
+
+package org.tensorflow.proto.framework;
+
+public final class TrackableObjectGraphProtos {
+  private TrackableObjectGraphProtos() {}
+  public static void registerAllExtensions(
+      com.google.protobuf.ExtensionRegistryLite registry) {
+  }
+
+  public static void registerAllExtensions(
+      com.google.protobuf.ExtensionRegistry registry) {
+    registerAllExtensions(
+        (com.google.protobuf.ExtensionRegistryLite) registry);
+  }
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_TrackableObjectGraph_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_TrackableObjectGraph_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_TrackableObjectGraph_TrackableObject_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_TrackableObjectGraph_TrackableObject_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_TrackableObjectGraph_TrackableObject_ObjectReference_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_TrackableObjectGraph_TrackableObject_ObjectReference_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_TrackableObjectGraph_TrackableObject_SerializedTensor_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_TrackableObjectGraph_TrackableObject_SerializedTensor_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_TrackableObjectGraph_TrackableObject_SlotVariableReference_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_TrackableObjectGraph_TrackableObject_SlotVariableReference_fieldAccessorTable;
+
+  public static com.google.protobuf.Descriptors.FileDescriptor
+      getDescriptor() {
+    return descriptor;
+  }
+  private static  com.google.protobuf.Descriptors.FileDescriptor
+      descriptor;
+  static {
+    java.lang.String[] descriptorData = {
+      "\n5tensorflow/core/protobuf/trackable_obj" +
+      "ect_graph.proto\022\ntensorflow\"\203\005\n\024Trackabl" +
+      "eObjectGraph\022?\n\005nodes\030\001 \003(\01320.tensorflow" +
+      ".TrackableObjectGraph.TrackableObject\032\251\004" +
+      "\n\017TrackableObject\022R\n\010children\030\001 \003(\0132@.te" +
+      "nsorflow.TrackableObjectGraph.TrackableO" +
+      "bject.ObjectReference\022U\n\nattributes\030\002 \003(" +
+      "\0132A.tensorflow.TrackableObjectGraph.Trac" +
+      "kableObject.SerializedTensor\022^\n\016slot_var" +
+      "iables\030\003 \003(\0132F.tensorflow.TrackableObjec" +
+      "tGraph.TrackableObject.SlotVariableRefer" +
+      "ence\0326\n\017ObjectReference\022\017\n\007node_id\030\001 \001(\005" +
+      "\022\022\n\nlocal_name\030\002 \001(\t\032e\n\020SerializedTensor" +
+      "\022\014\n\004name\030\001 \001(\t\022\021\n\tfull_name\030\002 \001(\t\022\026\n\016che" +
+      "ckpoint_key\030\003 \001(\t\022\030\n\020optional_restore\030\004 " +
+      "\001(\010\032l\n\025SlotVariableReference\022!\n\031original" +
+      "_variable_node_id\030\001 \001(\005\022\021\n\tslot_name\030\002 \001" +
+      "(\t\022\035\n\025slot_variable_node_id\030\003 \001(\005B\213\001\n\036or" +
+      "g.tensorflow.proto.frameworkB\032TrackableO" +
+      "bjectGraphProtosP\001ZHgithub.com/tensorflo" +
+      "w/tensorflow/tensorflow/go/core/core_pro" +
+      "tos_go_proto\370\001\001b\006proto3"
+    };
+    descriptor = com.google.protobuf.Descriptors.FileDescriptor
+      .internalBuildGeneratedFileFrom(descriptorData,
+        new com.google.protobuf.Descriptors.FileDescriptor[] {
+        });
+    internal_static_tensorflow_TrackableObjectGraph_descriptor =
+      getDescriptor().getMessageTypes().get(0);
+    internal_static_tensorflow_TrackableObjectGraph_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_TrackableObjectGraph_descriptor,
+        new java.lang.String[] { "Nodes", });
+    internal_static_tensorflow_TrackableObjectGraph_TrackableObject_descriptor =
+      internal_static_tensorflow_TrackableObjectGraph_descriptor.getNestedTypes().get(0);
+    internal_static_tensorflow_TrackableObjectGraph_TrackableObject_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_TrackableObjectGraph_TrackableObject_descriptor,
+        new java.lang.String[] { "Children", "Attributes", "SlotVariables", });
+    internal_static_tensorflow_TrackableObjectGraph_TrackableObject_ObjectReference_descriptor =
+      internal_static_tensorflow_TrackableObjectGraph_TrackableObject_descriptor.getNestedTypes().get(0);
+    internal_static_tensorflow_TrackableObjectGraph_TrackableObject_ObjectReference_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_TrackableObjectGraph_TrackableObject_ObjectReference_descriptor,
+        new java.lang.String[] { "NodeId", "LocalName", });
+    internal_static_tensorflow_TrackableObjectGraph_TrackableObject_SerializedTensor_descriptor =
+      internal_static_tensorflow_TrackableObjectGraph_TrackableObject_descriptor.getNestedTypes().get(1);
+    internal_static_tensorflow_TrackableObjectGraph_TrackableObject_SerializedTensor_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_TrackableObjectGraph_TrackableObject_SerializedTensor_descriptor,
+        new java.lang.String[] { "Name", "FullName", "CheckpointKey", "OptionalRestore", });
+    internal_static_tensorflow_TrackableObjectGraph_TrackableObject_SlotVariableReference_descriptor =
+      internal_static_tensorflow_TrackableObjectGraph_TrackableObject_descriptor.getNestedTypes().get(2);
+    internal_static_tensorflow_TrackableObjectGraph_TrackableObject_SlotVariableReference_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_TrackableObjectGraph_TrackableObject_SlotVariableReference_descriptor,
+        new java.lang.String[] { "OriginalVariableNodeId", "SlotName", "SlotVariableNodeId", });
+  }
+
+  // @@protoc_insertion_point(outer_class_scope)
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TupleValue.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TupleValue.java
new file mode 100644
index 00000000000..68f4c9d8d0b
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TupleValue.java
@@ -0,0 +1,773 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/struct.proto
+
+package org.tensorflow.proto.framework;
+
+/**
+ * <pre>
+ * Represents a Python tuple.
+ * </pre>
+ *
+ * Protobuf type {@code tensorflow.TupleValue}
+ */
+public  final class TupleValue extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.TupleValue)
+    TupleValueOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use TupleValue.newBuilder() to construct.
+  private TupleValue(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private TupleValue() {
+    values_ = java.util.Collections.emptyList();
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new TupleValue();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private TupleValue(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    int mutable_bitField0_ = 0;
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 10: {
+            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
+              values_ = new java.util.ArrayList<org.tensorflow.proto.framework.StructuredValue>();
+              mutable_bitField0_ |= 0x00000001;
+            }
+            values_.add(
+                input.readMessage(org.tensorflow.proto.framework.StructuredValue.parser(), extensionRegistry));
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      if (((mutable_bitField0_ & 0x00000001) != 0)) {
+        values_ = java.util.Collections.unmodifiableList(values_);
+      }
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.framework.StructProtos.internal_static_tensorflow_TupleValue_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.framework.StructProtos.internal_static_tensorflow_TupleValue_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.framework.TupleValue.class, org.tensorflow.proto.framework.TupleValue.Builder.class);
+  }
+
+  public static final int VALUES_FIELD_NUMBER = 1;
+  private java.util.List<org.tensorflow.proto.framework.StructuredValue> values_;
+  /**
+   * <code>repeated .tensorflow.StructuredValue values = 1;</code>
+   */
+  public java.util.List<org.tensorflow.proto.framework.StructuredValue> getValuesList() {
+    return values_;
+  }
+  /**
+   * <code>repeated .tensorflow.StructuredValue values = 1;</code>
+   */
+  public java.util.List<? extends org.tensorflow.proto.framework.StructuredValueOrBuilder> 
+      getValuesOrBuilderList() {
+    return values_;
+  }
+  /**
+   * <code>repeated .tensorflow.StructuredValue values = 1;</code>
+   */
+  public int getValuesCount() {
+    return values_.size();
+  }
+  /**
+   * <code>repeated .tensorflow.StructuredValue values = 1;</code>
+   */
+  public org.tensorflow.proto.framework.StructuredValue getValues(int index) {
+    return values_.get(index);
+  }
+  /**
+   * <code>repeated .tensorflow.StructuredValue values = 1;</code>
+   */
+  public org.tensorflow.proto.framework.StructuredValueOrBuilder getValuesOrBuilder(
+      int index) {
+    return values_.get(index);
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    for (int i = 0; i < values_.size(); i++) {
+      output.writeMessage(1, values_.get(i));
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    for (int i = 0; i < values_.size(); i++) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(1, values_.get(i));
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.framework.TupleValue)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.framework.TupleValue other = (org.tensorflow.proto.framework.TupleValue) obj;
+
+    if (!getValuesList()
+        .equals(other.getValuesList())) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    if (getValuesCount() > 0) {
+      hash = (37 * hash) + VALUES_FIELD_NUMBER;
+      hash = (53 * hash) + getValuesList().hashCode();
+    }
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.framework.TupleValue parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.TupleValue parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.TupleValue parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.TupleValue parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.TupleValue parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.TupleValue parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.TupleValue parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.TupleValue parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.TupleValue parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.TupleValue parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.TupleValue parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.TupleValue parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.framework.TupleValue prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * <pre>
+   * Represents a Python tuple.
+   * </pre>
+   *
+   * Protobuf type {@code tensorflow.TupleValue}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.TupleValue)
+      org.tensorflow.proto.framework.TupleValueOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.StructProtos.internal_static_tensorflow_TupleValue_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.StructProtos.internal_static_tensorflow_TupleValue_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.TupleValue.class, org.tensorflow.proto.framework.TupleValue.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.framework.TupleValue.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+        getValuesFieldBuilder();
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      if (valuesBuilder_ == null) {
+        values_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000001);
+      } else {
+        valuesBuilder_.clear();
+      }
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.framework.StructProtos.internal_static_tensorflow_TupleValue_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.TupleValue getDefaultInstanceForType() {
+      return org.tensorflow.proto.framework.TupleValue.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.TupleValue build() {
+      org.tensorflow.proto.framework.TupleValue result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.TupleValue buildPartial() {
+      org.tensorflow.proto.framework.TupleValue result = new org.tensorflow.proto.framework.TupleValue(this);
+      int from_bitField0_ = bitField0_;
+      if (valuesBuilder_ == null) {
+        if (((bitField0_ & 0x00000001) != 0)) {
+          values_ = java.util.Collections.unmodifiableList(values_);
+          bitField0_ = (bitField0_ & ~0x00000001);
+        }
+        result.values_ = values_;
+      } else {
+        result.values_ = valuesBuilder_.build();
+      }
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.framework.TupleValue) {
+        return mergeFrom((org.tensorflow.proto.framework.TupleValue)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.framework.TupleValue other) {
+      if (other == org.tensorflow.proto.framework.TupleValue.getDefaultInstance()) return this;
+      if (valuesBuilder_ == null) {
+        if (!other.values_.isEmpty()) {
+          if (values_.isEmpty()) {
+            values_ = other.values_;
+            bitField0_ = (bitField0_ & ~0x00000001);
+          } else {
+            ensureValuesIsMutable();
+            values_.addAll(other.values_);
+          }
+          onChanged();
+        }
+      } else {
+        if (!other.values_.isEmpty()) {
+          if (valuesBuilder_.isEmpty()) {
+            valuesBuilder_.dispose();
+            valuesBuilder_ = null;
+            values_ = other.values_;
+            bitField0_ = (bitField0_ & ~0x00000001);
+            valuesBuilder_ = 
+              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
+                 getValuesFieldBuilder() : null;
+          } else {
+            valuesBuilder_.addAllMessages(other.values_);
+          }
+        }
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.framework.TupleValue parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.framework.TupleValue) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+    private int bitField0_;
+
+    private java.util.List<org.tensorflow.proto.framework.StructuredValue> values_ =
+      java.util.Collections.emptyList();
+    private void ensureValuesIsMutable() {
+      if (!((bitField0_ & 0x00000001) != 0)) {
+        values_ = new java.util.ArrayList<org.tensorflow.proto.framework.StructuredValue>(values_);
+        bitField0_ |= 0x00000001;
+       }
+    }
+
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.framework.StructuredValue, org.tensorflow.proto.framework.StructuredValue.Builder, org.tensorflow.proto.framework.StructuredValueOrBuilder> valuesBuilder_;
+
+    /**
+     * <code>repeated .tensorflow.StructuredValue values = 1;</code>
+     */
+    public java.util.List<org.tensorflow.proto.framework.StructuredValue> getValuesList() {
+      if (valuesBuilder_ == null) {
+        return java.util.Collections.unmodifiableList(values_);
+      } else {
+        return valuesBuilder_.getMessageList();
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.StructuredValue values = 1;</code>
+     */
+    public int getValuesCount() {
+      if (valuesBuilder_ == null) {
+        return values_.size();
+      } else {
+        return valuesBuilder_.getCount();
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.StructuredValue values = 1;</code>
+     */
+    public org.tensorflow.proto.framework.StructuredValue getValues(int index) {
+      if (valuesBuilder_ == null) {
+        return values_.get(index);
+      } else {
+        return valuesBuilder_.getMessage(index);
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.StructuredValue values = 1;</code>
+     */
+    public Builder setValues(
+        int index, org.tensorflow.proto.framework.StructuredValue value) {
+      if (valuesBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureValuesIsMutable();
+        values_.set(index, value);
+        onChanged();
+      } else {
+        valuesBuilder_.setMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.StructuredValue values = 1;</code>
+     */
+    public Builder setValues(
+        int index, org.tensorflow.proto.framework.StructuredValue.Builder builderForValue) {
+      if (valuesBuilder_ == null) {
+        ensureValuesIsMutable();
+        values_.set(index, builderForValue.build());
+        onChanged();
+      } else {
+        valuesBuilder_.setMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.StructuredValue values = 1;</code>
+     */
+    public Builder addValues(org.tensorflow.proto.framework.StructuredValue value) {
+      if (valuesBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureValuesIsMutable();
+        values_.add(value);
+        onChanged();
+      } else {
+        valuesBuilder_.addMessage(value);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.StructuredValue values = 1;</code>
+     */
+    public Builder addValues(
+        int index, org.tensorflow.proto.framework.StructuredValue value) {
+      if (valuesBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureValuesIsMutable();
+        values_.add(index, value);
+        onChanged();
+      } else {
+        valuesBuilder_.addMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.StructuredValue values = 1;</code>
+     */
+    public Builder addValues(
+        org.tensorflow.proto.framework.StructuredValue.Builder builderForValue) {
+      if (valuesBuilder_ == null) {
+        ensureValuesIsMutable();
+        values_.add(builderForValue.build());
+        onChanged();
+      } else {
+        valuesBuilder_.addMessage(builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.StructuredValue values = 1;</code>
+     */
+    public Builder addValues(
+        int index, org.tensorflow.proto.framework.StructuredValue.Builder builderForValue) {
+      if (valuesBuilder_ == null) {
+        ensureValuesIsMutable();
+        values_.add(index, builderForValue.build());
+        onChanged();
+      } else {
+        valuesBuilder_.addMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.StructuredValue values = 1;</code>
+     */
+    public Builder addAllValues(
+        java.lang.Iterable<? extends org.tensorflow.proto.framework.StructuredValue> values) {
+      if (valuesBuilder_ == null) {
+        ensureValuesIsMutable();
+        com.google.protobuf.AbstractMessageLite.Builder.addAll(
+            values, values_);
+        onChanged();
+      } else {
+        valuesBuilder_.addAllMessages(values);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.StructuredValue values = 1;</code>
+     */
+    public Builder clearValues() {
+      if (valuesBuilder_ == null) {
+        values_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000001);
+        onChanged();
+      } else {
+        valuesBuilder_.clear();
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.StructuredValue values = 1;</code>
+     */
+    public Builder removeValues(int index) {
+      if (valuesBuilder_ == null) {
+        ensureValuesIsMutable();
+        values_.remove(index);
+        onChanged();
+      } else {
+        valuesBuilder_.remove(index);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.StructuredValue values = 1;</code>
+     */
+    public org.tensorflow.proto.framework.StructuredValue.Builder getValuesBuilder(
+        int index) {
+      return getValuesFieldBuilder().getBuilder(index);
+    }
+    /**
+     * <code>repeated .tensorflow.StructuredValue values = 1;</code>
+     */
+    public org.tensorflow.proto.framework.StructuredValueOrBuilder getValuesOrBuilder(
+        int index) {
+      if (valuesBuilder_ == null) {
+        return values_.get(index);  } else {
+        return valuesBuilder_.getMessageOrBuilder(index);
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.StructuredValue values = 1;</code>
+     */
+    public java.util.List<? extends org.tensorflow.proto.framework.StructuredValueOrBuilder> 
+         getValuesOrBuilderList() {
+      if (valuesBuilder_ != null) {
+        return valuesBuilder_.getMessageOrBuilderList();
+      } else {
+        return java.util.Collections.unmodifiableList(values_);
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.StructuredValue values = 1;</code>
+     */
+    public org.tensorflow.proto.framework.StructuredValue.Builder addValuesBuilder() {
+      return getValuesFieldBuilder().addBuilder(
+          org.tensorflow.proto.framework.StructuredValue.getDefaultInstance());
+    }
+    /**
+     * <code>repeated .tensorflow.StructuredValue values = 1;</code>
+     */
+    public org.tensorflow.proto.framework.StructuredValue.Builder addValuesBuilder(
+        int index) {
+      return getValuesFieldBuilder().addBuilder(
+          index, org.tensorflow.proto.framework.StructuredValue.getDefaultInstance());
+    }
+    /**
+     * <code>repeated .tensorflow.StructuredValue values = 1;</code>
+     */
+    public java.util.List<org.tensorflow.proto.framework.StructuredValue.Builder> 
+         getValuesBuilderList() {
+      return getValuesFieldBuilder().getBuilderList();
+    }
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.framework.StructuredValue, org.tensorflow.proto.framework.StructuredValue.Builder, org.tensorflow.proto.framework.StructuredValueOrBuilder> 
+        getValuesFieldBuilder() {
+      if (valuesBuilder_ == null) {
+        valuesBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
+            org.tensorflow.proto.framework.StructuredValue, org.tensorflow.proto.framework.StructuredValue.Builder, org.tensorflow.proto.framework.StructuredValueOrBuilder>(
+                values_,
+                ((bitField0_ & 0x00000001) != 0),
+                getParentForChildren(),
+                isClean());
+        values_ = null;
+      }
+      return valuesBuilder_;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.TupleValue)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.TupleValue)
+  private static final org.tensorflow.proto.framework.TupleValue DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.TupleValue();
+  }
+
+  public static org.tensorflow.proto.framework.TupleValue getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<TupleValue>
+      PARSER = new com.google.protobuf.AbstractParser<TupleValue>() {
+    @java.lang.Override
+    public TupleValue parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new TupleValue(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<TupleValue> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<TupleValue> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.framework.TupleValue getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TupleValueOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TupleValueOrBuilder.java
new file mode 100644
index 00000000000..aea74c1d73b
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TupleValueOrBuilder.java
@@ -0,0 +1,33 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/struct.proto
+
+package org.tensorflow.proto.framework;
+
+public interface TupleValueOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.TupleValue)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <code>repeated .tensorflow.StructuredValue values = 1;</code>
+   */
+  java.util.List<org.tensorflow.proto.framework.StructuredValue> 
+      getValuesList();
+  /**
+   * <code>repeated .tensorflow.StructuredValue values = 1;</code>
+   */
+  org.tensorflow.proto.framework.StructuredValue getValues(int index);
+  /**
+   * <code>repeated .tensorflow.StructuredValue values = 1;</code>
+   */
+  int getValuesCount();
+  /**
+   * <code>repeated .tensorflow.StructuredValue values = 1;</code>
+   */
+  java.util.List<? extends org.tensorflow.proto.framework.StructuredValueOrBuilder> 
+      getValuesOrBuilderList();
+  /**
+   * <code>repeated .tensorflow.StructuredValue values = 1;</code>
+   */
+  org.tensorflow.proto.framework.StructuredValueOrBuilder getValuesOrBuilder(
+      int index);
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TypeSpecProto.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TypeSpecProto.java
new file mode 100644
index 00000000000..10c64eeb41c
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TypeSpecProto.java
@@ -0,0 +1,1176 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/struct.proto
+
+package org.tensorflow.proto.framework;
+
+/**
+ * <pre>
+ * Represents a tf.TypeSpec
+ * </pre>
+ *
+ * Protobuf type {@code tensorflow.TypeSpecProto}
+ */
+public  final class TypeSpecProto extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.TypeSpecProto)
+    TypeSpecProtoOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use TypeSpecProto.newBuilder() to construct.
+  private TypeSpecProto(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private TypeSpecProto() {
+    typeSpecClass_ = 0;
+    typeSpecClassName_ = "";
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new TypeSpecProto();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private TypeSpecProto(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 8: {
+            int rawValue = input.readEnum();
+
+            typeSpecClass_ = rawValue;
+            break;
+          }
+          case 18: {
+            org.tensorflow.proto.framework.StructuredValue.Builder subBuilder = null;
+            if (typeState_ != null) {
+              subBuilder = typeState_.toBuilder();
+            }
+            typeState_ = input.readMessage(org.tensorflow.proto.framework.StructuredValue.parser(), extensionRegistry);
+            if (subBuilder != null) {
+              subBuilder.mergeFrom(typeState_);
+              typeState_ = subBuilder.buildPartial();
+            }
+
+            break;
+          }
+          case 26: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            typeSpecClassName_ = s;
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.framework.StructProtos.internal_static_tensorflow_TypeSpecProto_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.framework.StructProtos.internal_static_tensorflow_TypeSpecProto_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.framework.TypeSpecProto.class, org.tensorflow.proto.framework.TypeSpecProto.Builder.class);
+  }
+
+  /**
+   * Protobuf enum {@code tensorflow.TypeSpecProto.TypeSpecClass}
+   */
+  public enum TypeSpecClass
+      implements com.google.protobuf.ProtocolMessageEnum {
+    /**
+     * <code>UNKNOWN = 0;</code>
+     */
+    UNKNOWN(0),
+    /**
+     * <pre>
+     * tf.SparseTensorSpec
+     * </pre>
+     *
+     * <code>SPARSE_TENSOR_SPEC = 1;</code>
+     */
+    SPARSE_TENSOR_SPEC(1),
+    /**
+     * <pre>
+     * tf.IndexedSlicesSpec
+     * </pre>
+     *
+     * <code>INDEXED_SLICES_SPEC = 2;</code>
+     */
+    INDEXED_SLICES_SPEC(2),
+    /**
+     * <pre>
+     * tf.RaggedTensorSpec
+     * </pre>
+     *
+     * <code>RAGGED_TENSOR_SPEC = 3;</code>
+     */
+    RAGGED_TENSOR_SPEC(3),
+    /**
+     * <pre>
+     * tf.TensorArraySpec
+     * </pre>
+     *
+     * <code>TENSOR_ARRAY_SPEC = 4;</code>
+     */
+    TENSOR_ARRAY_SPEC(4),
+    /**
+     * <pre>
+     * tf.data.DatasetSpec
+     * </pre>
+     *
+     * <code>DATA_DATASET_SPEC = 5;</code>
+     */
+    DATA_DATASET_SPEC(5),
+    /**
+     * <pre>
+     * IteratorSpec from data/ops/iterator_ops.py
+     * </pre>
+     *
+     * <code>DATA_ITERATOR_SPEC = 6;</code>
+     */
+    DATA_ITERATOR_SPEC(6),
+    /**
+     * <pre>
+     * tf.OptionalSpec
+     * </pre>
+     *
+     * <code>OPTIONAL_SPEC = 7;</code>
+     */
+    OPTIONAL_SPEC(7),
+    /**
+     * <pre>
+     * PerReplicaSpec from distribute/values.py
+     * </pre>
+     *
+     * <code>PER_REPLICA_SPEC = 8;</code>
+     */
+    PER_REPLICA_SPEC(8),
+    /**
+     * <pre>
+     * tf.VariableSpec
+     * </pre>
+     *
+     * <code>VARIABLE_SPEC = 9;</code>
+     */
+    VARIABLE_SPEC(9),
+    /**
+     * <pre>
+     * RowPartitionSpec from ragged/row_partition.py
+     * </pre>
+     *
+     * <code>ROW_PARTITION_SPEC = 10;</code>
+     */
+    ROW_PARTITION_SPEC(10),
+    UNRECOGNIZED(-1),
+    ;
+
+    /**
+     * <code>UNKNOWN = 0;</code>
+     */
+    public static final int UNKNOWN_VALUE = 0;
+    /**
+     * <pre>
+     * tf.SparseTensorSpec
+     * </pre>
+     *
+     * <code>SPARSE_TENSOR_SPEC = 1;</code>
+     */
+    public static final int SPARSE_TENSOR_SPEC_VALUE = 1;
+    /**
+     * <pre>
+     * tf.IndexedSlicesSpec
+     * </pre>
+     *
+     * <code>INDEXED_SLICES_SPEC = 2;</code>
+     */
+    public static final int INDEXED_SLICES_SPEC_VALUE = 2;
+    /**
+     * <pre>
+     * tf.RaggedTensorSpec
+     * </pre>
+     *
+     * <code>RAGGED_TENSOR_SPEC = 3;</code>
+     */
+    public static final int RAGGED_TENSOR_SPEC_VALUE = 3;
+    /**
+     * <pre>
+     * tf.TensorArraySpec
+     * </pre>
+     *
+     * <code>TENSOR_ARRAY_SPEC = 4;</code>
+     */
+    public static final int TENSOR_ARRAY_SPEC_VALUE = 4;
+    /**
+     * <pre>
+     * tf.data.DatasetSpec
+     * </pre>
+     *
+     * <code>DATA_DATASET_SPEC = 5;</code>
+     */
+    public static final int DATA_DATASET_SPEC_VALUE = 5;
+    /**
+     * <pre>
+     * IteratorSpec from data/ops/iterator_ops.py
+     * </pre>
+     *
+     * <code>DATA_ITERATOR_SPEC = 6;</code>
+     */
+    public static final int DATA_ITERATOR_SPEC_VALUE = 6;
+    /**
+     * <pre>
+     * tf.OptionalSpec
+     * </pre>
+     *
+     * <code>OPTIONAL_SPEC = 7;</code>
+     */
+    public static final int OPTIONAL_SPEC_VALUE = 7;
+    /**
+     * <pre>
+     * PerReplicaSpec from distribute/values.py
+     * </pre>
+     *
+     * <code>PER_REPLICA_SPEC = 8;</code>
+     */
+    public static final int PER_REPLICA_SPEC_VALUE = 8;
+    /**
+     * <pre>
+     * tf.VariableSpec
+     * </pre>
+     *
+     * <code>VARIABLE_SPEC = 9;</code>
+     */
+    public static final int VARIABLE_SPEC_VALUE = 9;
+    /**
+     * <pre>
+     * RowPartitionSpec from ragged/row_partition.py
+     * </pre>
+     *
+     * <code>ROW_PARTITION_SPEC = 10;</code>
+     */
+    public static final int ROW_PARTITION_SPEC_VALUE = 10;
+
+
+    public final int getNumber() {
+      if (this == UNRECOGNIZED) {
+        throw new java.lang.IllegalArgumentException(
+            "Can't get the number of an unknown enum value.");
+      }
+      return value;
+    }
+
+    /**
+     * @deprecated Use {@link #forNumber(int)} instead.
+     */
+    @java.lang.Deprecated
+    public static TypeSpecClass valueOf(int value) {
+      return forNumber(value);
+    }
+
+    public static TypeSpecClass forNumber(int value) {
+      switch (value) {
+        case 0: return UNKNOWN;
+        case 1: return SPARSE_TENSOR_SPEC;
+        case 2: return INDEXED_SLICES_SPEC;
+        case 3: return RAGGED_TENSOR_SPEC;
+        case 4: return TENSOR_ARRAY_SPEC;
+        case 5: return DATA_DATASET_SPEC;
+        case 6: return DATA_ITERATOR_SPEC;
+        case 7: return OPTIONAL_SPEC;
+        case 8: return PER_REPLICA_SPEC;
+        case 9: return VARIABLE_SPEC;
+        case 10: return ROW_PARTITION_SPEC;
+        default: return null;
+      }
+    }
+
+    public static com.google.protobuf.Internal.EnumLiteMap<TypeSpecClass>
+        internalGetValueMap() {
+      return internalValueMap;
+    }
+    private static final com.google.protobuf.Internal.EnumLiteMap<
+        TypeSpecClass> internalValueMap =
+          new com.google.protobuf.Internal.EnumLiteMap<TypeSpecClass>() {
+            public TypeSpecClass findValueByNumber(int number) {
+              return TypeSpecClass.forNumber(number);
+            }
+          };
+
+    public final com.google.protobuf.Descriptors.EnumValueDescriptor
+        getValueDescriptor() {
+      return getDescriptor().getValues().get(ordinal());
+    }
+    public final com.google.protobuf.Descriptors.EnumDescriptor
+        getDescriptorForType() {
+      return getDescriptor();
+    }
+    public static final com.google.protobuf.Descriptors.EnumDescriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.TypeSpecProto.getDescriptor().getEnumTypes().get(0);
+    }
+
+    private static final TypeSpecClass[] VALUES = values();
+
+    public static TypeSpecClass valueOf(
+        com.google.protobuf.Descriptors.EnumValueDescriptor desc) {
+      if (desc.getType() != getDescriptor()) {
+        throw new java.lang.IllegalArgumentException(
+          "EnumValueDescriptor is not for this type.");
+      }
+      if (desc.getIndex() == -1) {
+        return UNRECOGNIZED;
+      }
+      return VALUES[desc.getIndex()];
+    }
+
+    private final int value;
+
+    private TypeSpecClass(int value) {
+      this.value = value;
+    }
+
+    // @@protoc_insertion_point(enum_scope:tensorflow.TypeSpecProto.TypeSpecClass)
+  }
+
+  public static final int TYPE_SPEC_CLASS_FIELD_NUMBER = 1;
+  private int typeSpecClass_;
+  /**
+   * <code>.tensorflow.TypeSpecProto.TypeSpecClass type_spec_class = 1;</code>
+   */
+  public int getTypeSpecClassValue() {
+    return typeSpecClass_;
+  }
+  /**
+   * <code>.tensorflow.TypeSpecProto.TypeSpecClass type_spec_class = 1;</code>
+   */
+  public org.tensorflow.proto.framework.TypeSpecProto.TypeSpecClass getTypeSpecClass() {
+    @SuppressWarnings("deprecation")
+    org.tensorflow.proto.framework.TypeSpecProto.TypeSpecClass result = org.tensorflow.proto.framework.TypeSpecProto.TypeSpecClass.valueOf(typeSpecClass_);
+    return result == null ? org.tensorflow.proto.framework.TypeSpecProto.TypeSpecClass.UNRECOGNIZED : result;
+  }
+
+  public static final int TYPE_STATE_FIELD_NUMBER = 2;
+  private org.tensorflow.proto.framework.StructuredValue typeState_;
+  /**
+   * <pre>
+   * The value returned by TypeSpec._serialize().
+   * </pre>
+   *
+   * <code>.tensorflow.StructuredValue type_state = 2;</code>
+   */
+  public boolean hasTypeState() {
+    return typeState_ != null;
+  }
+  /**
+   * <pre>
+   * The value returned by TypeSpec._serialize().
+   * </pre>
+   *
+   * <code>.tensorflow.StructuredValue type_state = 2;</code>
+   */
+  public org.tensorflow.proto.framework.StructuredValue getTypeState() {
+    return typeState_ == null ? org.tensorflow.proto.framework.StructuredValue.getDefaultInstance() : typeState_;
+  }
+  /**
+   * <pre>
+   * The value returned by TypeSpec._serialize().
+   * </pre>
+   *
+   * <code>.tensorflow.StructuredValue type_state = 2;</code>
+   */
+  public org.tensorflow.proto.framework.StructuredValueOrBuilder getTypeStateOrBuilder() {
+    return getTypeState();
+  }
+
+  public static final int TYPE_SPEC_CLASS_NAME_FIELD_NUMBER = 3;
+  private volatile java.lang.Object typeSpecClassName_;
+  /**
+   * <pre>
+   * This is currently redundant with the type_spec_class enum, and is only
+   * used for error reporting.  In particular, if you use an older binary to
+   * load a newer model, and the model uses a TypeSpecClass that the older
+   * binary doesn't support, then this lets us display a useful error message.
+   * </pre>
+   *
+   * <code>string type_spec_class_name = 3;</code>
+   */
+  public java.lang.String getTypeSpecClassName() {
+    java.lang.Object ref = typeSpecClassName_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      typeSpecClassName_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * This is currently redundant with the type_spec_class enum, and is only
+   * used for error reporting.  In particular, if you use an older binary to
+   * load a newer model, and the model uses a TypeSpecClass that the older
+   * binary doesn't support, then this lets us display a useful error message.
+   * </pre>
+   *
+   * <code>string type_spec_class_name = 3;</code>
+   */
+  public com.google.protobuf.ByteString
+      getTypeSpecClassNameBytes() {
+    java.lang.Object ref = typeSpecClassName_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      typeSpecClassName_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    if (typeSpecClass_ != org.tensorflow.proto.framework.TypeSpecProto.TypeSpecClass.UNKNOWN.getNumber()) {
+      output.writeEnum(1, typeSpecClass_);
+    }
+    if (typeState_ != null) {
+      output.writeMessage(2, getTypeState());
+    }
+    if (!getTypeSpecClassNameBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 3, typeSpecClassName_);
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (typeSpecClass_ != org.tensorflow.proto.framework.TypeSpecProto.TypeSpecClass.UNKNOWN.getNumber()) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeEnumSize(1, typeSpecClass_);
+    }
+    if (typeState_ != null) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(2, getTypeState());
+    }
+    if (!getTypeSpecClassNameBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(3, typeSpecClassName_);
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.framework.TypeSpecProto)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.framework.TypeSpecProto other = (org.tensorflow.proto.framework.TypeSpecProto) obj;
+
+    if (typeSpecClass_ != other.typeSpecClass_) return false;
+    if (hasTypeState() != other.hasTypeState()) return false;
+    if (hasTypeState()) {
+      if (!getTypeState()
+          .equals(other.getTypeState())) return false;
+    }
+    if (!getTypeSpecClassName()
+        .equals(other.getTypeSpecClassName())) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    hash = (37 * hash) + TYPE_SPEC_CLASS_FIELD_NUMBER;
+    hash = (53 * hash) + typeSpecClass_;
+    if (hasTypeState()) {
+      hash = (37 * hash) + TYPE_STATE_FIELD_NUMBER;
+      hash = (53 * hash) + getTypeState().hashCode();
+    }
+    hash = (37 * hash) + TYPE_SPEC_CLASS_NAME_FIELD_NUMBER;
+    hash = (53 * hash) + getTypeSpecClassName().hashCode();
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.framework.TypeSpecProto parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.TypeSpecProto parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.TypeSpecProto parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.TypeSpecProto parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.TypeSpecProto parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.TypeSpecProto parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.TypeSpecProto parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.TypeSpecProto parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.TypeSpecProto parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.TypeSpecProto parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.TypeSpecProto parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.TypeSpecProto parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.framework.TypeSpecProto prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * <pre>
+   * Represents a tf.TypeSpec
+   * </pre>
+   *
+   * Protobuf type {@code tensorflow.TypeSpecProto}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.TypeSpecProto)
+      org.tensorflow.proto.framework.TypeSpecProtoOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.StructProtos.internal_static_tensorflow_TypeSpecProto_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.StructProtos.internal_static_tensorflow_TypeSpecProto_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.TypeSpecProto.class, org.tensorflow.proto.framework.TypeSpecProto.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.framework.TypeSpecProto.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      typeSpecClass_ = 0;
+
+      if (typeStateBuilder_ == null) {
+        typeState_ = null;
+      } else {
+        typeState_ = null;
+        typeStateBuilder_ = null;
+      }
+      typeSpecClassName_ = "";
+
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.framework.StructProtos.internal_static_tensorflow_TypeSpecProto_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.TypeSpecProto getDefaultInstanceForType() {
+      return org.tensorflow.proto.framework.TypeSpecProto.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.TypeSpecProto build() {
+      org.tensorflow.proto.framework.TypeSpecProto result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.TypeSpecProto buildPartial() {
+      org.tensorflow.proto.framework.TypeSpecProto result = new org.tensorflow.proto.framework.TypeSpecProto(this);
+      result.typeSpecClass_ = typeSpecClass_;
+      if (typeStateBuilder_ == null) {
+        result.typeState_ = typeState_;
+      } else {
+        result.typeState_ = typeStateBuilder_.build();
+      }
+      result.typeSpecClassName_ = typeSpecClassName_;
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.framework.TypeSpecProto) {
+        return mergeFrom((org.tensorflow.proto.framework.TypeSpecProto)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.framework.TypeSpecProto other) {
+      if (other == org.tensorflow.proto.framework.TypeSpecProto.getDefaultInstance()) return this;
+      if (other.typeSpecClass_ != 0) {
+        setTypeSpecClassValue(other.getTypeSpecClassValue());
+      }
+      if (other.hasTypeState()) {
+        mergeTypeState(other.getTypeState());
+      }
+      if (!other.getTypeSpecClassName().isEmpty()) {
+        typeSpecClassName_ = other.typeSpecClassName_;
+        onChanged();
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.framework.TypeSpecProto parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.framework.TypeSpecProto) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+
+    private int typeSpecClass_ = 0;
+    /**
+     * <code>.tensorflow.TypeSpecProto.TypeSpecClass type_spec_class = 1;</code>
+     */
+    public int getTypeSpecClassValue() {
+      return typeSpecClass_;
+    }
+    /**
+     * <code>.tensorflow.TypeSpecProto.TypeSpecClass type_spec_class = 1;</code>
+     */
+    public Builder setTypeSpecClassValue(int value) {
+      typeSpecClass_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>.tensorflow.TypeSpecProto.TypeSpecClass type_spec_class = 1;</code>
+     */
+    public org.tensorflow.proto.framework.TypeSpecProto.TypeSpecClass getTypeSpecClass() {
+      @SuppressWarnings("deprecation")
+      org.tensorflow.proto.framework.TypeSpecProto.TypeSpecClass result = org.tensorflow.proto.framework.TypeSpecProto.TypeSpecClass.valueOf(typeSpecClass_);
+      return result == null ? org.tensorflow.proto.framework.TypeSpecProto.TypeSpecClass.UNRECOGNIZED : result;
+    }
+    /**
+     * <code>.tensorflow.TypeSpecProto.TypeSpecClass type_spec_class = 1;</code>
+     */
+    public Builder setTypeSpecClass(org.tensorflow.proto.framework.TypeSpecProto.TypeSpecClass value) {
+      if (value == null) {
+        throw new NullPointerException();
+      }
+      
+      typeSpecClass_ = value.getNumber();
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>.tensorflow.TypeSpecProto.TypeSpecClass type_spec_class = 1;</code>
+     */
+    public Builder clearTypeSpecClass() {
+      
+      typeSpecClass_ = 0;
+      onChanged();
+      return this;
+    }
+
+    private org.tensorflow.proto.framework.StructuredValue typeState_;
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.StructuredValue, org.tensorflow.proto.framework.StructuredValue.Builder, org.tensorflow.proto.framework.StructuredValueOrBuilder> typeStateBuilder_;
+    /**
+     * <pre>
+     * The value returned by TypeSpec._serialize().
+     * </pre>
+     *
+     * <code>.tensorflow.StructuredValue type_state = 2;</code>
+     */
+    public boolean hasTypeState() {
+      return typeStateBuilder_ != null || typeState_ != null;
+    }
+    /**
+     * <pre>
+     * The value returned by TypeSpec._serialize().
+     * </pre>
+     *
+     * <code>.tensorflow.StructuredValue type_state = 2;</code>
+     */
+    public org.tensorflow.proto.framework.StructuredValue getTypeState() {
+      if (typeStateBuilder_ == null) {
+        return typeState_ == null ? org.tensorflow.proto.framework.StructuredValue.getDefaultInstance() : typeState_;
+      } else {
+        return typeStateBuilder_.getMessage();
+      }
+    }
+    /**
+     * <pre>
+     * The value returned by TypeSpec._serialize().
+     * </pre>
+     *
+     * <code>.tensorflow.StructuredValue type_state = 2;</code>
+     */
+    public Builder setTypeState(org.tensorflow.proto.framework.StructuredValue value) {
+      if (typeStateBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        typeState_ = value;
+        onChanged();
+      } else {
+        typeStateBuilder_.setMessage(value);
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * The value returned by TypeSpec._serialize().
+     * </pre>
+     *
+     * <code>.tensorflow.StructuredValue type_state = 2;</code>
+     */
+    public Builder setTypeState(
+        org.tensorflow.proto.framework.StructuredValue.Builder builderForValue) {
+      if (typeStateBuilder_ == null) {
+        typeState_ = builderForValue.build();
+        onChanged();
+      } else {
+        typeStateBuilder_.setMessage(builderForValue.build());
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * The value returned by TypeSpec._serialize().
+     * </pre>
+     *
+     * <code>.tensorflow.StructuredValue type_state = 2;</code>
+     */
+    public Builder mergeTypeState(org.tensorflow.proto.framework.StructuredValue value) {
+      if (typeStateBuilder_ == null) {
+        if (typeState_ != null) {
+          typeState_ =
+            org.tensorflow.proto.framework.StructuredValue.newBuilder(typeState_).mergeFrom(value).buildPartial();
+        } else {
+          typeState_ = value;
+        }
+        onChanged();
+      } else {
+        typeStateBuilder_.mergeFrom(value);
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * The value returned by TypeSpec._serialize().
+     * </pre>
+     *
+     * <code>.tensorflow.StructuredValue type_state = 2;</code>
+     */
+    public Builder clearTypeState() {
+      if (typeStateBuilder_ == null) {
+        typeState_ = null;
+        onChanged();
+      } else {
+        typeState_ = null;
+        typeStateBuilder_ = null;
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * The value returned by TypeSpec._serialize().
+     * </pre>
+     *
+     * <code>.tensorflow.StructuredValue type_state = 2;</code>
+     */
+    public org.tensorflow.proto.framework.StructuredValue.Builder getTypeStateBuilder() {
+      
+      onChanged();
+      return getTypeStateFieldBuilder().getBuilder();
+    }
+    /**
+     * <pre>
+     * The value returned by TypeSpec._serialize().
+     * </pre>
+     *
+     * <code>.tensorflow.StructuredValue type_state = 2;</code>
+     */
+    public org.tensorflow.proto.framework.StructuredValueOrBuilder getTypeStateOrBuilder() {
+      if (typeStateBuilder_ != null) {
+        return typeStateBuilder_.getMessageOrBuilder();
+      } else {
+        return typeState_ == null ?
+            org.tensorflow.proto.framework.StructuredValue.getDefaultInstance() : typeState_;
+      }
+    }
+    /**
+     * <pre>
+     * The value returned by TypeSpec._serialize().
+     * </pre>
+     *
+     * <code>.tensorflow.StructuredValue type_state = 2;</code>
+     */
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.StructuredValue, org.tensorflow.proto.framework.StructuredValue.Builder, org.tensorflow.proto.framework.StructuredValueOrBuilder> 
+        getTypeStateFieldBuilder() {
+      if (typeStateBuilder_ == null) {
+        typeStateBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+            org.tensorflow.proto.framework.StructuredValue, org.tensorflow.proto.framework.StructuredValue.Builder, org.tensorflow.proto.framework.StructuredValueOrBuilder>(
+                getTypeState(),
+                getParentForChildren(),
+                isClean());
+        typeState_ = null;
+      }
+      return typeStateBuilder_;
+    }
+
+    private java.lang.Object typeSpecClassName_ = "";
+    /**
+     * <pre>
+     * This is currently redundant with the type_spec_class enum, and is only
+     * used for error reporting.  In particular, if you use an older binary to
+     * load a newer model, and the model uses a TypeSpecClass that the older
+     * binary doesn't support, then this lets us display a useful error message.
+     * </pre>
+     *
+     * <code>string type_spec_class_name = 3;</code>
+     */
+    public java.lang.String getTypeSpecClassName() {
+      java.lang.Object ref = typeSpecClassName_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        typeSpecClassName_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * This is currently redundant with the type_spec_class enum, and is only
+     * used for error reporting.  In particular, if you use an older binary to
+     * load a newer model, and the model uses a TypeSpecClass that the older
+     * binary doesn't support, then this lets us display a useful error message.
+     * </pre>
+     *
+     * <code>string type_spec_class_name = 3;</code>
+     */
+    public com.google.protobuf.ByteString
+        getTypeSpecClassNameBytes() {
+      java.lang.Object ref = typeSpecClassName_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        typeSpecClassName_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * This is currently redundant with the type_spec_class enum, and is only
+     * used for error reporting.  In particular, if you use an older binary to
+     * load a newer model, and the model uses a TypeSpecClass that the older
+     * binary doesn't support, then this lets us display a useful error message.
+     * </pre>
+     *
+     * <code>string type_spec_class_name = 3;</code>
+     */
+    public Builder setTypeSpecClassName(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      typeSpecClassName_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * This is currently redundant with the type_spec_class enum, and is only
+     * used for error reporting.  In particular, if you use an older binary to
+     * load a newer model, and the model uses a TypeSpecClass that the older
+     * binary doesn't support, then this lets us display a useful error message.
+     * </pre>
+     *
+     * <code>string type_spec_class_name = 3;</code>
+     */
+    public Builder clearTypeSpecClassName() {
+      
+      typeSpecClassName_ = getDefaultInstance().getTypeSpecClassName();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * This is currently redundant with the type_spec_class enum, and is only
+     * used for error reporting.  In particular, if you use an older binary to
+     * load a newer model, and the model uses a TypeSpecClass that the older
+     * binary doesn't support, then this lets us display a useful error message.
+     * </pre>
+     *
+     * <code>string type_spec_class_name = 3;</code>
+     */
+    public Builder setTypeSpecClassNameBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      typeSpecClassName_ = value;
+      onChanged();
+      return this;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.TypeSpecProto)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.TypeSpecProto)
+  private static final org.tensorflow.proto.framework.TypeSpecProto DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.TypeSpecProto();
+  }
+
+  public static org.tensorflow.proto.framework.TypeSpecProto getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<TypeSpecProto>
+      PARSER = new com.google.protobuf.AbstractParser<TypeSpecProto>() {
+    @java.lang.Override
+    public TypeSpecProto parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new TypeSpecProto(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<TypeSpecProto> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<TypeSpecProto> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.framework.TypeSpecProto getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TypeSpecProtoOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TypeSpecProtoOrBuilder.java
new file mode 100644
index 00000000000..db9ce3b7ab8
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TypeSpecProtoOrBuilder.java
@@ -0,0 +1,67 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/struct.proto
+
+package org.tensorflow.proto.framework;
+
+public interface TypeSpecProtoOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.TypeSpecProto)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <code>.tensorflow.TypeSpecProto.TypeSpecClass type_spec_class = 1;</code>
+   */
+  int getTypeSpecClassValue();
+  /**
+   * <code>.tensorflow.TypeSpecProto.TypeSpecClass type_spec_class = 1;</code>
+   */
+  org.tensorflow.proto.framework.TypeSpecProto.TypeSpecClass getTypeSpecClass();
+
+  /**
+   * <pre>
+   * The value returned by TypeSpec._serialize().
+   * </pre>
+   *
+   * <code>.tensorflow.StructuredValue type_state = 2;</code>
+   */
+  boolean hasTypeState();
+  /**
+   * <pre>
+   * The value returned by TypeSpec._serialize().
+   * </pre>
+   *
+   * <code>.tensorflow.StructuredValue type_state = 2;</code>
+   */
+  org.tensorflow.proto.framework.StructuredValue getTypeState();
+  /**
+   * <pre>
+   * The value returned by TypeSpec._serialize().
+   * </pre>
+   *
+   * <code>.tensorflow.StructuredValue type_state = 2;</code>
+   */
+  org.tensorflow.proto.framework.StructuredValueOrBuilder getTypeStateOrBuilder();
+
+  /**
+   * <pre>
+   * This is currently redundant with the type_spec_class enum, and is only
+   * used for error reporting.  In particular, if you use an older binary to
+   * load a newer model, and the model uses a TypeSpecClass that the older
+   * binary doesn't support, then this lets us display a useful error message.
+   * </pre>
+   *
+   * <code>string type_spec_class_name = 3;</code>
+   */
+  java.lang.String getTypeSpecClassName();
+  /**
+   * <pre>
+   * This is currently redundant with the type_spec_class enum, and is only
+   * used for error reporting.  In particular, if you use an older binary to
+   * load a newer model, and the model uses a TypeSpecClass that the older
+   * binary doesn't support, then this lets us display a useful error message.
+   * </pre>
+   *
+   * <code>string type_spec_class_name = 3;</code>
+   */
+  com.google.protobuf.ByteString
+      getTypeSpecClassNameBytes();
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TypesProtos.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TypesProtos.java
new file mode 100644
index 00000000000..9c343efcb7c
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/TypesProtos.java
@@ -0,0 +1,60 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/framework/types.proto
+
+package org.tensorflow.proto.framework;
+
+public final class TypesProtos {
+  private TypesProtos() {}
+  public static void registerAllExtensions(
+      com.google.protobuf.ExtensionRegistryLite registry) {
+  }
+
+  public static void registerAllExtensions(
+      com.google.protobuf.ExtensionRegistry registry) {
+    registerAllExtensions(
+        (com.google.protobuf.ExtensionRegistryLite) registry);
+  }
+
+  public static com.google.protobuf.Descriptors.FileDescriptor
+      getDescriptor() {
+    return descriptor;
+  }
+  private static  com.google.protobuf.Descriptors.FileDescriptor
+      descriptor;
+  static {
+    java.lang.String[] descriptorData = {
+      "\n%tensorflow/core/framework/types.proto\022" +
+      "\ntensorflow*\252\006\n\010DataType\022\016\n\nDT_INVALID\020\000" +
+      "\022\014\n\010DT_FLOAT\020\001\022\r\n\tDT_DOUBLE\020\002\022\014\n\010DT_INT3" +
+      "2\020\003\022\014\n\010DT_UINT8\020\004\022\014\n\010DT_INT16\020\005\022\013\n\007DT_IN" +
+      "T8\020\006\022\r\n\tDT_STRING\020\007\022\020\n\014DT_COMPLEX64\020\010\022\014\n" +
+      "\010DT_INT64\020\t\022\013\n\007DT_BOOL\020\n\022\014\n\010DT_QINT8\020\013\022\r" +
+      "\n\tDT_QUINT8\020\014\022\r\n\tDT_QINT32\020\r\022\017\n\013DT_BFLOA" +
+      "T16\020\016\022\r\n\tDT_QINT16\020\017\022\016\n\nDT_QUINT16\020\020\022\r\n\t" +
+      "DT_UINT16\020\021\022\021\n\rDT_COMPLEX128\020\022\022\013\n\007DT_HAL" +
+      "F\020\023\022\017\n\013DT_RESOURCE\020\024\022\016\n\nDT_VARIANT\020\025\022\r\n\t" +
+      "DT_UINT32\020\026\022\r\n\tDT_UINT64\020\027\022\020\n\014DT_FLOAT_R" +
+      "EF\020e\022\021\n\rDT_DOUBLE_REF\020f\022\020\n\014DT_INT32_REF\020" +
+      "g\022\020\n\014DT_UINT8_REF\020h\022\020\n\014DT_INT16_REF\020i\022\017\n" +
+      "\013DT_INT8_REF\020j\022\021\n\rDT_STRING_REF\020k\022\024\n\020DT_" +
+      "COMPLEX64_REF\020l\022\020\n\014DT_INT64_REF\020m\022\017\n\013DT_" +
+      "BOOL_REF\020n\022\020\n\014DT_QINT8_REF\020o\022\021\n\rDT_QUINT" +
+      "8_REF\020p\022\021\n\rDT_QINT32_REF\020q\022\023\n\017DT_BFLOAT1" +
+      "6_REF\020r\022\021\n\rDT_QINT16_REF\020s\022\022\n\016DT_QUINT16" +
+      "_REF\020t\022\021\n\rDT_UINT16_REF\020u\022\025\n\021DT_COMPLEX1" +
+      "28_REF\020v\022\017\n\013DT_HALF_REF\020w\022\023\n\017DT_RESOURCE" +
+      "_REF\020x\022\022\n\016DT_VARIANT_REF\020y\022\021\n\rDT_UINT32_" +
+      "REF\020z\022\021\n\rDT_UINT64_REF\020{B\200\001\n\036org.tensorf" +
+      "low.proto.frameworkB\013TypesProtosP\001ZLgith" +
+      "ub.com/tensorflow/tensorflow/tensorflow/" +
+      "go/core/framework/types_go_proto\370\001\001b\006pro" +
+      "to3"
+    };
+    descriptor = com.google.protobuf.Descriptors.FileDescriptor
+      .internalBuildGeneratedFileFrom(descriptorData,
+        new com.google.protobuf.Descriptors.FileDescriptor[] {
+        });
+  }
+
+  // @@protoc_insertion_point(outer_class_scope)
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ValuesDef.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ValuesDef.java
new file mode 100644
index 00000000000..d6819b5bbfa
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ValuesDef.java
@@ -0,0 +1,969 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/control_flow.proto
+
+package org.tensorflow.proto.framework;
+
+/**
+ * <pre>
+ * Protocol buffer representing the values in ControlFlowContext.
+ * </pre>
+ *
+ * Protobuf type {@code tensorflow.ValuesDef}
+ */
+public  final class ValuesDef extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.ValuesDef)
+    ValuesDefOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use ValuesDef.newBuilder() to construct.
+  private ValuesDef(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private ValuesDef() {
+    values_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new ValuesDef();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private ValuesDef(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    int mutable_bitField0_ = 0;
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 10: {
+            java.lang.String s = input.readStringRequireUtf8();
+            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
+              values_ = new com.google.protobuf.LazyStringArrayList();
+              mutable_bitField0_ |= 0x00000001;
+            }
+            values_.add(s);
+            break;
+          }
+          case 18: {
+            if (!((mutable_bitField0_ & 0x00000002) != 0)) {
+              externalValues_ = com.google.protobuf.MapField.newMapField(
+                  ExternalValuesDefaultEntryHolder.defaultEntry);
+              mutable_bitField0_ |= 0x00000002;
+            }
+            com.google.protobuf.MapEntry<java.lang.String, java.lang.String>
+            externalValues__ = input.readMessage(
+                ExternalValuesDefaultEntryHolder.defaultEntry.getParserForType(), extensionRegistry);
+            externalValues_.getMutableMap().put(
+                externalValues__.getKey(), externalValues__.getValue());
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      if (((mutable_bitField0_ & 0x00000001) != 0)) {
+        values_ = values_.getUnmodifiableView();
+      }
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.framework.ControlFlowProtos.internal_static_tensorflow_ValuesDef_descriptor;
+  }
+
+  @SuppressWarnings({"rawtypes"})
+  @java.lang.Override
+  protected com.google.protobuf.MapField internalGetMapField(
+      int number) {
+    switch (number) {
+      case 2:
+        return internalGetExternalValues();
+      default:
+        throw new RuntimeException(
+            "Invalid map field number: " + number);
+    }
+  }
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.framework.ControlFlowProtos.internal_static_tensorflow_ValuesDef_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.framework.ValuesDef.class, org.tensorflow.proto.framework.ValuesDef.Builder.class);
+  }
+
+  public static final int VALUES_FIELD_NUMBER = 1;
+  private com.google.protobuf.LazyStringList values_;
+  /**
+   * <pre>
+   * Value names that have been seen in this context.
+   * </pre>
+   *
+   * <code>repeated string values = 1;</code>
+   */
+  public com.google.protobuf.ProtocolStringList
+      getValuesList() {
+    return values_;
+  }
+  /**
+   * <pre>
+   * Value names that have been seen in this context.
+   * </pre>
+   *
+   * <code>repeated string values = 1;</code>
+   */
+  public int getValuesCount() {
+    return values_.size();
+  }
+  /**
+   * <pre>
+   * Value names that have been seen in this context.
+   * </pre>
+   *
+   * <code>repeated string values = 1;</code>
+   */
+  public java.lang.String getValues(int index) {
+    return values_.get(index);
+  }
+  /**
+   * <pre>
+   * Value names that have been seen in this context.
+   * </pre>
+   *
+   * <code>repeated string values = 1;</code>
+   */
+  public com.google.protobuf.ByteString
+      getValuesBytes(int index) {
+    return values_.getByteString(index);
+  }
+
+  public static final int EXTERNAL_VALUES_FIELD_NUMBER = 2;
+  private static final class ExternalValuesDefaultEntryHolder {
+    static final com.google.protobuf.MapEntry<
+        java.lang.String, java.lang.String> defaultEntry =
+            com.google.protobuf.MapEntry
+            .<java.lang.String, java.lang.String>newDefaultInstance(
+                org.tensorflow.proto.framework.ControlFlowProtos.internal_static_tensorflow_ValuesDef_ExternalValuesEntry_descriptor, 
+                com.google.protobuf.WireFormat.FieldType.STRING,
+                "",
+                com.google.protobuf.WireFormat.FieldType.STRING,
+                "");
+  }
+  private com.google.protobuf.MapField<
+      java.lang.String, java.lang.String> externalValues_;
+  private com.google.protobuf.MapField<java.lang.String, java.lang.String>
+  internalGetExternalValues() {
+    if (externalValues_ == null) {
+      return com.google.protobuf.MapField.emptyMapField(
+          ExternalValuesDefaultEntryHolder.defaultEntry);
+    }
+    return externalValues_;
+  }
+
+  public int getExternalValuesCount() {
+    return internalGetExternalValues().getMap().size();
+  }
+  /**
+   * <pre>
+   * Value names referenced by but external to this context.
+   * </pre>
+   *
+   * <code>map&lt;string, string&gt; external_values = 2;</code>
+   */
+
+  public boolean containsExternalValues(
+      java.lang.String key) {
+    if (key == null) { throw new java.lang.NullPointerException(); }
+    return internalGetExternalValues().getMap().containsKey(key);
+  }
+  /**
+   * Use {@link #getExternalValuesMap()} instead.
+   */
+  @java.lang.Deprecated
+  public java.util.Map<java.lang.String, java.lang.String> getExternalValues() {
+    return getExternalValuesMap();
+  }
+  /**
+   * <pre>
+   * Value names referenced by but external to this context.
+   * </pre>
+   *
+   * <code>map&lt;string, string&gt; external_values = 2;</code>
+   */
+
+  public java.util.Map<java.lang.String, java.lang.String> getExternalValuesMap() {
+    return internalGetExternalValues().getMap();
+  }
+  /**
+   * <pre>
+   * Value names referenced by but external to this context.
+   * </pre>
+   *
+   * <code>map&lt;string, string&gt; external_values = 2;</code>
+   */
+
+  public java.lang.String getExternalValuesOrDefault(
+      java.lang.String key,
+      java.lang.String defaultValue) {
+    if (key == null) { throw new java.lang.NullPointerException(); }
+    java.util.Map<java.lang.String, java.lang.String> map =
+        internalGetExternalValues().getMap();
+    return map.containsKey(key) ? map.get(key) : defaultValue;
+  }
+  /**
+   * <pre>
+   * Value names referenced by but external to this context.
+   * </pre>
+   *
+   * <code>map&lt;string, string&gt; external_values = 2;</code>
+   */
+
+  public java.lang.String getExternalValuesOrThrow(
+      java.lang.String key) {
+    if (key == null) { throw new java.lang.NullPointerException(); }
+    java.util.Map<java.lang.String, java.lang.String> map =
+        internalGetExternalValues().getMap();
+    if (!map.containsKey(key)) {
+      throw new java.lang.IllegalArgumentException();
+    }
+    return map.get(key);
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    for (int i = 0; i < values_.size(); i++) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 1, values_.getRaw(i));
+    }
+    com.google.protobuf.GeneratedMessageV3
+      .serializeStringMapTo(
+        output,
+        internalGetExternalValues(),
+        ExternalValuesDefaultEntryHolder.defaultEntry,
+        2);
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    {
+      int dataSize = 0;
+      for (int i = 0; i < values_.size(); i++) {
+        dataSize += computeStringSizeNoTag(values_.getRaw(i));
+      }
+      size += dataSize;
+      size += 1 * getValuesList().size();
+    }
+    for (java.util.Map.Entry<java.lang.String, java.lang.String> entry
+         : internalGetExternalValues().getMap().entrySet()) {
+      com.google.protobuf.MapEntry<java.lang.String, java.lang.String>
+      externalValues__ = ExternalValuesDefaultEntryHolder.defaultEntry.newBuilderForType()
+          .setKey(entry.getKey())
+          .setValue(entry.getValue())
+          .build();
+      size += com.google.protobuf.CodedOutputStream
+          .computeMessageSize(2, externalValues__);
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.framework.ValuesDef)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.framework.ValuesDef other = (org.tensorflow.proto.framework.ValuesDef) obj;
+
+    if (!getValuesList()
+        .equals(other.getValuesList())) return false;
+    if (!internalGetExternalValues().equals(
+        other.internalGetExternalValues())) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    if (getValuesCount() > 0) {
+      hash = (37 * hash) + VALUES_FIELD_NUMBER;
+      hash = (53 * hash) + getValuesList().hashCode();
+    }
+    if (!internalGetExternalValues().getMap().isEmpty()) {
+      hash = (37 * hash) + EXTERNAL_VALUES_FIELD_NUMBER;
+      hash = (53 * hash) + internalGetExternalValues().hashCode();
+    }
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.framework.ValuesDef parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.ValuesDef parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.ValuesDef parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.ValuesDef parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.ValuesDef parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.ValuesDef parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.ValuesDef parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.ValuesDef parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.ValuesDef parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.ValuesDef parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.ValuesDef parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.ValuesDef parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.framework.ValuesDef prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * <pre>
+   * Protocol buffer representing the values in ControlFlowContext.
+   * </pre>
+   *
+   * Protobuf type {@code tensorflow.ValuesDef}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.ValuesDef)
+      org.tensorflow.proto.framework.ValuesDefOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.ControlFlowProtos.internal_static_tensorflow_ValuesDef_descriptor;
+    }
+
+    @SuppressWarnings({"rawtypes"})
+    protected com.google.protobuf.MapField internalGetMapField(
+        int number) {
+      switch (number) {
+        case 2:
+          return internalGetExternalValues();
+        default:
+          throw new RuntimeException(
+              "Invalid map field number: " + number);
+      }
+    }
+    @SuppressWarnings({"rawtypes"})
+    protected com.google.protobuf.MapField internalGetMutableMapField(
+        int number) {
+      switch (number) {
+        case 2:
+          return internalGetMutableExternalValues();
+        default:
+          throw new RuntimeException(
+              "Invalid map field number: " + number);
+      }
+    }
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.ControlFlowProtos.internal_static_tensorflow_ValuesDef_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.ValuesDef.class, org.tensorflow.proto.framework.ValuesDef.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.framework.ValuesDef.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      values_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+      bitField0_ = (bitField0_ & ~0x00000001);
+      internalGetMutableExternalValues().clear();
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.framework.ControlFlowProtos.internal_static_tensorflow_ValuesDef_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.ValuesDef getDefaultInstanceForType() {
+      return org.tensorflow.proto.framework.ValuesDef.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.ValuesDef build() {
+      org.tensorflow.proto.framework.ValuesDef result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.ValuesDef buildPartial() {
+      org.tensorflow.proto.framework.ValuesDef result = new org.tensorflow.proto.framework.ValuesDef(this);
+      int from_bitField0_ = bitField0_;
+      if (((bitField0_ & 0x00000001) != 0)) {
+        values_ = values_.getUnmodifiableView();
+        bitField0_ = (bitField0_ & ~0x00000001);
+      }
+      result.values_ = values_;
+      result.externalValues_ = internalGetExternalValues();
+      result.externalValues_.makeImmutable();
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.framework.ValuesDef) {
+        return mergeFrom((org.tensorflow.proto.framework.ValuesDef)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.framework.ValuesDef other) {
+      if (other == org.tensorflow.proto.framework.ValuesDef.getDefaultInstance()) return this;
+      if (!other.values_.isEmpty()) {
+        if (values_.isEmpty()) {
+          values_ = other.values_;
+          bitField0_ = (bitField0_ & ~0x00000001);
+        } else {
+          ensureValuesIsMutable();
+          values_.addAll(other.values_);
+        }
+        onChanged();
+      }
+      internalGetMutableExternalValues().mergeFrom(
+          other.internalGetExternalValues());
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.framework.ValuesDef parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.framework.ValuesDef) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+    private int bitField0_;
+
+    private com.google.protobuf.LazyStringList values_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+    private void ensureValuesIsMutable() {
+      if (!((bitField0_ & 0x00000001) != 0)) {
+        values_ = new com.google.protobuf.LazyStringArrayList(values_);
+        bitField0_ |= 0x00000001;
+       }
+    }
+    /**
+     * <pre>
+     * Value names that have been seen in this context.
+     * </pre>
+     *
+     * <code>repeated string values = 1;</code>
+     */
+    public com.google.protobuf.ProtocolStringList
+        getValuesList() {
+      return values_.getUnmodifiableView();
+    }
+    /**
+     * <pre>
+     * Value names that have been seen in this context.
+     * </pre>
+     *
+     * <code>repeated string values = 1;</code>
+     */
+    public int getValuesCount() {
+      return values_.size();
+    }
+    /**
+     * <pre>
+     * Value names that have been seen in this context.
+     * </pre>
+     *
+     * <code>repeated string values = 1;</code>
+     */
+    public java.lang.String getValues(int index) {
+      return values_.get(index);
+    }
+    /**
+     * <pre>
+     * Value names that have been seen in this context.
+     * </pre>
+     *
+     * <code>repeated string values = 1;</code>
+     */
+    public com.google.protobuf.ByteString
+        getValuesBytes(int index) {
+      return values_.getByteString(index);
+    }
+    /**
+     * <pre>
+     * Value names that have been seen in this context.
+     * </pre>
+     *
+     * <code>repeated string values = 1;</code>
+     */
+    public Builder setValues(
+        int index, java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  ensureValuesIsMutable();
+      values_.set(index, value);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Value names that have been seen in this context.
+     * </pre>
+     *
+     * <code>repeated string values = 1;</code>
+     */
+    public Builder addValues(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  ensureValuesIsMutable();
+      values_.add(value);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Value names that have been seen in this context.
+     * </pre>
+     *
+     * <code>repeated string values = 1;</code>
+     */
+    public Builder addAllValues(
+        java.lang.Iterable<java.lang.String> values) {
+      ensureValuesIsMutable();
+      com.google.protobuf.AbstractMessageLite.Builder.addAll(
+          values, values_);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Value names that have been seen in this context.
+     * </pre>
+     *
+     * <code>repeated string values = 1;</code>
+     */
+    public Builder clearValues() {
+      values_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+      bitField0_ = (bitField0_ & ~0x00000001);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Value names that have been seen in this context.
+     * </pre>
+     *
+     * <code>repeated string values = 1;</code>
+     */
+    public Builder addValuesBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      ensureValuesIsMutable();
+      values_.add(value);
+      onChanged();
+      return this;
+    }
+
+    private com.google.protobuf.MapField<
+        java.lang.String, java.lang.String> externalValues_;
+    private com.google.protobuf.MapField<java.lang.String, java.lang.String>
+    internalGetExternalValues() {
+      if (externalValues_ == null) {
+        return com.google.protobuf.MapField.emptyMapField(
+            ExternalValuesDefaultEntryHolder.defaultEntry);
+      }
+      return externalValues_;
+    }
+    private com.google.protobuf.MapField<java.lang.String, java.lang.String>
+    internalGetMutableExternalValues() {
+      onChanged();;
+      if (externalValues_ == null) {
+        externalValues_ = com.google.protobuf.MapField.newMapField(
+            ExternalValuesDefaultEntryHolder.defaultEntry);
+      }
+      if (!externalValues_.isMutable()) {
+        externalValues_ = externalValues_.copy();
+      }
+      return externalValues_;
+    }
+
+    public int getExternalValuesCount() {
+      return internalGetExternalValues().getMap().size();
+    }
+    /**
+     * <pre>
+     * Value names referenced by but external to this context.
+     * </pre>
+     *
+     * <code>map&lt;string, string&gt; external_values = 2;</code>
+     */
+
+    public boolean containsExternalValues(
+        java.lang.String key) {
+      if (key == null) { throw new java.lang.NullPointerException(); }
+      return internalGetExternalValues().getMap().containsKey(key);
+    }
+    /**
+     * Use {@link #getExternalValuesMap()} instead.
+     */
+    @java.lang.Deprecated
+    public java.util.Map<java.lang.String, java.lang.String> getExternalValues() {
+      return getExternalValuesMap();
+    }
+    /**
+     * <pre>
+     * Value names referenced by but external to this context.
+     * </pre>
+     *
+     * <code>map&lt;string, string&gt; external_values = 2;</code>
+     */
+
+    public java.util.Map<java.lang.String, java.lang.String> getExternalValuesMap() {
+      return internalGetExternalValues().getMap();
+    }
+    /**
+     * <pre>
+     * Value names referenced by but external to this context.
+     * </pre>
+     *
+     * <code>map&lt;string, string&gt; external_values = 2;</code>
+     */
+
+    public java.lang.String getExternalValuesOrDefault(
+        java.lang.String key,
+        java.lang.String defaultValue) {
+      if (key == null) { throw new java.lang.NullPointerException(); }
+      java.util.Map<java.lang.String, java.lang.String> map =
+          internalGetExternalValues().getMap();
+      return map.containsKey(key) ? map.get(key) : defaultValue;
+    }
+    /**
+     * <pre>
+     * Value names referenced by but external to this context.
+     * </pre>
+     *
+     * <code>map&lt;string, string&gt; external_values = 2;</code>
+     */
+
+    public java.lang.String getExternalValuesOrThrow(
+        java.lang.String key) {
+      if (key == null) { throw new java.lang.NullPointerException(); }
+      java.util.Map<java.lang.String, java.lang.String> map =
+          internalGetExternalValues().getMap();
+      if (!map.containsKey(key)) {
+        throw new java.lang.IllegalArgumentException();
+      }
+      return map.get(key);
+    }
+
+    public Builder clearExternalValues() {
+      internalGetMutableExternalValues().getMutableMap()
+          .clear();
+      return this;
+    }
+    /**
+     * <pre>
+     * Value names referenced by but external to this context.
+     * </pre>
+     *
+     * <code>map&lt;string, string&gt; external_values = 2;</code>
+     */
+
+    public Builder removeExternalValues(
+        java.lang.String key) {
+      if (key == null) { throw new java.lang.NullPointerException(); }
+      internalGetMutableExternalValues().getMutableMap()
+          .remove(key);
+      return this;
+    }
+    /**
+     * Use alternate mutation accessors instead.
+     */
+    @java.lang.Deprecated
+    public java.util.Map<java.lang.String, java.lang.String>
+    getMutableExternalValues() {
+      return internalGetMutableExternalValues().getMutableMap();
+    }
+    /**
+     * <pre>
+     * Value names referenced by but external to this context.
+     * </pre>
+     *
+     * <code>map&lt;string, string&gt; external_values = 2;</code>
+     */
+    public Builder putExternalValues(
+        java.lang.String key,
+        java.lang.String value) {
+      if (key == null) { throw new java.lang.NullPointerException(); }
+      if (value == null) { throw new java.lang.NullPointerException(); }
+      internalGetMutableExternalValues().getMutableMap()
+          .put(key, value);
+      return this;
+    }
+    /**
+     * <pre>
+     * Value names referenced by but external to this context.
+     * </pre>
+     *
+     * <code>map&lt;string, string&gt; external_values = 2;</code>
+     */
+
+    public Builder putAllExternalValues(
+        java.util.Map<java.lang.String, java.lang.String> values) {
+      internalGetMutableExternalValues().getMutableMap()
+          .putAll(values);
+      return this;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.ValuesDef)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.ValuesDef)
+  private static final org.tensorflow.proto.framework.ValuesDef DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.ValuesDef();
+  }
+
+  public static org.tensorflow.proto.framework.ValuesDef getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<ValuesDef>
+      PARSER = new com.google.protobuf.AbstractParser<ValuesDef>() {
+    @java.lang.Override
+    public ValuesDef parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new ValuesDef(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<ValuesDef> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<ValuesDef> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.framework.ValuesDef getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ValuesDefOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ValuesDefOrBuilder.java
new file mode 100644
index 00000000000..d9135689d22
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/ValuesDefOrBuilder.java
@@ -0,0 +1,98 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/control_flow.proto
+
+package org.tensorflow.proto.framework;
+
+public interface ValuesDefOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.ValuesDef)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <pre>
+   * Value names that have been seen in this context.
+   * </pre>
+   *
+   * <code>repeated string values = 1;</code>
+   */
+  java.util.List<java.lang.String>
+      getValuesList();
+  /**
+   * <pre>
+   * Value names that have been seen in this context.
+   * </pre>
+   *
+   * <code>repeated string values = 1;</code>
+   */
+  int getValuesCount();
+  /**
+   * <pre>
+   * Value names that have been seen in this context.
+   * </pre>
+   *
+   * <code>repeated string values = 1;</code>
+   */
+  java.lang.String getValues(int index);
+  /**
+   * <pre>
+   * Value names that have been seen in this context.
+   * </pre>
+   *
+   * <code>repeated string values = 1;</code>
+   */
+  com.google.protobuf.ByteString
+      getValuesBytes(int index);
+
+  /**
+   * <pre>
+   * Value names referenced by but external to this context.
+   * </pre>
+   *
+   * <code>map&lt;string, string&gt; external_values = 2;</code>
+   */
+  int getExternalValuesCount();
+  /**
+   * <pre>
+   * Value names referenced by but external to this context.
+   * </pre>
+   *
+   * <code>map&lt;string, string&gt; external_values = 2;</code>
+   */
+  boolean containsExternalValues(
+      java.lang.String key);
+  /**
+   * Use {@link #getExternalValuesMap()} instead.
+   */
+  @java.lang.Deprecated
+  java.util.Map<java.lang.String, java.lang.String>
+  getExternalValues();
+  /**
+   * <pre>
+   * Value names referenced by but external to this context.
+   * </pre>
+   *
+   * <code>map&lt;string, string&gt; external_values = 2;</code>
+   */
+  java.util.Map<java.lang.String, java.lang.String>
+  getExternalValuesMap();
+  /**
+   * <pre>
+   * Value names referenced by but external to this context.
+   * </pre>
+   *
+   * <code>map&lt;string, string&gt; external_values = 2;</code>
+   */
+
+  java.lang.String getExternalValuesOrDefault(
+      java.lang.String key,
+      java.lang.String defaultValue);
+  /**
+   * <pre>
+   * Value names referenced by but external to this context.
+   * </pre>
+   *
+   * <code>map&lt;string, string&gt; external_values = 2;</code>
+   */
+
+  java.lang.String getExternalValuesOrThrow(
+      java.lang.String key);
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/VariableAggregation.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/VariableAggregation.java
new file mode 100644
index 00000000000..584cd9b19bd
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/VariableAggregation.java
@@ -0,0 +1,165 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/framework/variable.proto
+
+package org.tensorflow.proto.framework;
+
+/**
+ * <pre>
+ * Indicates how a distributed variable will be aggregated.
+ * </pre>
+ *
+ * Protobuf enum {@code tensorflow.VariableAggregation}
+ */
+public enum VariableAggregation
+    implements com.google.protobuf.ProtocolMessageEnum {
+  /**
+   * <pre>
+   * `NONE`: This is the default, giving an error if you use a
+   * variable-update operation with multiple replicas.
+   * </pre>
+   *
+   * <code>VARIABLE_AGGREGATION_NONE = 0;</code>
+   */
+  VARIABLE_AGGREGATION_NONE(0),
+  /**
+   * <pre>
+   * `SUM`: Add the updates across replicas.
+   * </pre>
+   *
+   * <code>VARIABLE_AGGREGATION_SUM = 1;</code>
+   */
+  VARIABLE_AGGREGATION_SUM(1),
+  /**
+   * <pre>
+   * `MEAN`: Take the arithmetic mean ("average") of the updates across
+   * replicas.
+   * </pre>
+   *
+   * <code>VARIABLE_AGGREGATION_MEAN = 2;</code>
+   */
+  VARIABLE_AGGREGATION_MEAN(2),
+  /**
+   * <pre>
+   * `ONLY_FIRST_REPLICA`: This is for when every replica is performing the same
+   * update, but we only want to perform the update once. Used, e.g., for the
+   * global step counter.
+   * </pre>
+   *
+   * <code>VARIABLE_AGGREGATION_ONLY_FIRST_REPLICA = 3;</code>
+   */
+  VARIABLE_AGGREGATION_ONLY_FIRST_REPLICA(3),
+  UNRECOGNIZED(-1),
+  ;
+
+  /**
+   * <pre>
+   * `NONE`: This is the default, giving an error if you use a
+   * variable-update operation with multiple replicas.
+   * </pre>
+   *
+   * <code>VARIABLE_AGGREGATION_NONE = 0;</code>
+   */
+  public static final int VARIABLE_AGGREGATION_NONE_VALUE = 0;
+  /**
+   * <pre>
+   * `SUM`: Add the updates across replicas.
+   * </pre>
+   *
+   * <code>VARIABLE_AGGREGATION_SUM = 1;</code>
+   */
+  public static final int VARIABLE_AGGREGATION_SUM_VALUE = 1;
+  /**
+   * <pre>
+   * `MEAN`: Take the arithmetic mean ("average") of the updates across
+   * replicas.
+   * </pre>
+   *
+   * <code>VARIABLE_AGGREGATION_MEAN = 2;</code>
+   */
+  public static final int VARIABLE_AGGREGATION_MEAN_VALUE = 2;
+  /**
+   * <pre>
+   * `ONLY_FIRST_REPLICA`: This is for when every replica is performing the same
+   * update, but we only want to perform the update once. Used, e.g., for the
+   * global step counter.
+   * </pre>
+   *
+   * <code>VARIABLE_AGGREGATION_ONLY_FIRST_REPLICA = 3;</code>
+   */
+  public static final int VARIABLE_AGGREGATION_ONLY_FIRST_REPLICA_VALUE = 3;
+
+
+  public final int getNumber() {
+    if (this == UNRECOGNIZED) {
+      throw new java.lang.IllegalArgumentException(
+          "Can't get the number of an unknown enum value.");
+    }
+    return value;
+  }
+
+  /**
+   * @deprecated Use {@link #forNumber(int)} instead.
+   */
+  @java.lang.Deprecated
+  public static VariableAggregation valueOf(int value) {
+    return forNumber(value);
+  }
+
+  public static VariableAggregation forNumber(int value) {
+    switch (value) {
+      case 0: return VARIABLE_AGGREGATION_NONE;
+      case 1: return VARIABLE_AGGREGATION_SUM;
+      case 2: return VARIABLE_AGGREGATION_MEAN;
+      case 3: return VARIABLE_AGGREGATION_ONLY_FIRST_REPLICA;
+      default: return null;
+    }
+  }
+
+  public static com.google.protobuf.Internal.EnumLiteMap<VariableAggregation>
+      internalGetValueMap() {
+    return internalValueMap;
+  }
+  private static final com.google.protobuf.Internal.EnumLiteMap<
+      VariableAggregation> internalValueMap =
+        new com.google.protobuf.Internal.EnumLiteMap<VariableAggregation>() {
+          public VariableAggregation findValueByNumber(int number) {
+            return VariableAggregation.forNumber(number);
+          }
+        };
+
+  public final com.google.protobuf.Descriptors.EnumValueDescriptor
+      getValueDescriptor() {
+    return getDescriptor().getValues().get(ordinal());
+  }
+  public final com.google.protobuf.Descriptors.EnumDescriptor
+      getDescriptorForType() {
+    return getDescriptor();
+  }
+  public static final com.google.protobuf.Descriptors.EnumDescriptor
+      getDescriptor() {
+    return org.tensorflow.proto.framework.VariableProtos.getDescriptor().getEnumTypes().get(1);
+  }
+
+  private static final VariableAggregation[] VALUES = values();
+
+  public static VariableAggregation valueOf(
+      com.google.protobuf.Descriptors.EnumValueDescriptor desc) {
+    if (desc.getType() != getDescriptor()) {
+      throw new java.lang.IllegalArgumentException(
+        "EnumValueDescriptor is not for this type.");
+    }
+    if (desc.getIndex() == -1) {
+      return UNRECOGNIZED;
+    }
+    return VALUES[desc.getIndex()];
+  }
+
+  private final int value;
+
+  private VariableAggregation(int value) {
+    this.value = value;
+  }
+
+  // @@protoc_insertion_point(enum_scope:tensorflow.VariableAggregation)
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/VariableDef.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/VariableDef.java
new file mode 100644
index 00000000000..e0ecee5dc06
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/VariableDef.java
@@ -0,0 +1,1650 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/framework/variable.proto
+
+package org.tensorflow.proto.framework;
+
+/**
+ * <pre>
+ * Protocol buffer representing a Variable.
+ * </pre>
+ *
+ * Protobuf type {@code tensorflow.VariableDef}
+ */
+public  final class VariableDef extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.VariableDef)
+    VariableDefOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use VariableDef.newBuilder() to construct.
+  private VariableDef(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private VariableDef() {
+    variableName_ = "";
+    initialValueName_ = "";
+    initializerName_ = "";
+    snapshotName_ = "";
+    synchronization_ = 0;
+    aggregation_ = 0;
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new VariableDef();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private VariableDef(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 10: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            variableName_ = s;
+            break;
+          }
+          case 18: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            initializerName_ = s;
+            break;
+          }
+          case 26: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            snapshotName_ = s;
+            break;
+          }
+          case 34: {
+            org.tensorflow.proto.framework.SaveSliceInfoDef.Builder subBuilder = null;
+            if (saveSliceInfoDef_ != null) {
+              subBuilder = saveSliceInfoDef_.toBuilder();
+            }
+            saveSliceInfoDef_ = input.readMessage(org.tensorflow.proto.framework.SaveSliceInfoDef.parser(), extensionRegistry);
+            if (subBuilder != null) {
+              subBuilder.mergeFrom(saveSliceInfoDef_);
+              saveSliceInfoDef_ = subBuilder.buildPartial();
+            }
+
+            break;
+          }
+          case 40: {
+
+            isResource_ = input.readBool();
+            break;
+          }
+          case 50: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            initialValueName_ = s;
+            break;
+          }
+          case 56: {
+
+            trainable_ = input.readBool();
+            break;
+          }
+          case 64: {
+            int rawValue = input.readEnum();
+
+            synchronization_ = rawValue;
+            break;
+          }
+          case 72: {
+            int rawValue = input.readEnum();
+
+            aggregation_ = rawValue;
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.framework.VariableProtos.internal_static_tensorflow_VariableDef_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.framework.VariableProtos.internal_static_tensorflow_VariableDef_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.framework.VariableDef.class, org.tensorflow.proto.framework.VariableDef.Builder.class);
+  }
+
+  public static final int VARIABLE_NAME_FIELD_NUMBER = 1;
+  private volatile java.lang.Object variableName_;
+  /**
+   * <pre>
+   * Name of the variable tensor.
+   * </pre>
+   *
+   * <code>string variable_name = 1;</code>
+   */
+  public java.lang.String getVariableName() {
+    java.lang.Object ref = variableName_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      variableName_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * Name of the variable tensor.
+   * </pre>
+   *
+   * <code>string variable_name = 1;</code>
+   */
+  public com.google.protobuf.ByteString
+      getVariableNameBytes() {
+    java.lang.Object ref = variableName_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      variableName_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int INITIAL_VALUE_NAME_FIELD_NUMBER = 6;
+  private volatile java.lang.Object initialValueName_;
+  /**
+   * <pre>
+   * Name of the tensor holding the variable's initial value.
+   * </pre>
+   *
+   * <code>string initial_value_name = 6;</code>
+   */
+  public java.lang.String getInitialValueName() {
+    java.lang.Object ref = initialValueName_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      initialValueName_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * Name of the tensor holding the variable's initial value.
+   * </pre>
+   *
+   * <code>string initial_value_name = 6;</code>
+   */
+  public com.google.protobuf.ByteString
+      getInitialValueNameBytes() {
+    java.lang.Object ref = initialValueName_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      initialValueName_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int INITIALIZER_NAME_FIELD_NUMBER = 2;
+  private volatile java.lang.Object initializerName_;
+  /**
+   * <pre>
+   * Name of the initializer op.
+   * </pre>
+   *
+   * <code>string initializer_name = 2;</code>
+   */
+  public java.lang.String getInitializerName() {
+    java.lang.Object ref = initializerName_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      initializerName_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * Name of the initializer op.
+   * </pre>
+   *
+   * <code>string initializer_name = 2;</code>
+   */
+  public com.google.protobuf.ByteString
+      getInitializerNameBytes() {
+    java.lang.Object ref = initializerName_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      initializerName_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int SNAPSHOT_NAME_FIELD_NUMBER = 3;
+  private volatile java.lang.Object snapshotName_;
+  /**
+   * <pre>
+   * Name of the snapshot tensor.
+   * </pre>
+   *
+   * <code>string snapshot_name = 3;</code>
+   */
+  public java.lang.String getSnapshotName() {
+    java.lang.Object ref = snapshotName_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      snapshotName_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * Name of the snapshot tensor.
+   * </pre>
+   *
+   * <code>string snapshot_name = 3;</code>
+   */
+  public com.google.protobuf.ByteString
+      getSnapshotNameBytes() {
+    java.lang.Object ref = snapshotName_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      snapshotName_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int SAVE_SLICE_INFO_DEF_FIELD_NUMBER = 4;
+  private org.tensorflow.proto.framework.SaveSliceInfoDef saveSliceInfoDef_;
+  /**
+   * <pre>
+   * Support for saving variables as slices of a larger variable.
+   * </pre>
+   *
+   * <code>.tensorflow.SaveSliceInfoDef save_slice_info_def = 4;</code>
+   */
+  public boolean hasSaveSliceInfoDef() {
+    return saveSliceInfoDef_ != null;
+  }
+  /**
+   * <pre>
+   * Support for saving variables as slices of a larger variable.
+   * </pre>
+   *
+   * <code>.tensorflow.SaveSliceInfoDef save_slice_info_def = 4;</code>
+   */
+  public org.tensorflow.proto.framework.SaveSliceInfoDef getSaveSliceInfoDef() {
+    return saveSliceInfoDef_ == null ? org.tensorflow.proto.framework.SaveSliceInfoDef.getDefaultInstance() : saveSliceInfoDef_;
+  }
+  /**
+   * <pre>
+   * Support for saving variables as slices of a larger variable.
+   * </pre>
+   *
+   * <code>.tensorflow.SaveSliceInfoDef save_slice_info_def = 4;</code>
+   */
+  public org.tensorflow.proto.framework.SaveSliceInfoDefOrBuilder getSaveSliceInfoDefOrBuilder() {
+    return getSaveSliceInfoDef();
+  }
+
+  public static final int IS_RESOURCE_FIELD_NUMBER = 5;
+  private boolean isResource_;
+  /**
+   * <pre>
+   * Whether to represent this as a ResourceVariable.
+   * </pre>
+   *
+   * <code>bool is_resource = 5;</code>
+   */
+  public boolean getIsResource() {
+    return isResource_;
+  }
+
+  public static final int TRAINABLE_FIELD_NUMBER = 7;
+  private boolean trainable_;
+  /**
+   * <pre>
+   * Whether this variable should be trained.
+   * </pre>
+   *
+   * <code>bool trainable = 7;</code>
+   */
+  public boolean getTrainable() {
+    return trainable_;
+  }
+
+  public static final int SYNCHRONIZATION_FIELD_NUMBER = 8;
+  private int synchronization_;
+  /**
+   * <pre>
+   * Indicates when a distributed variable will be synced.
+   * </pre>
+   *
+   * <code>.tensorflow.VariableSynchronization synchronization = 8;</code>
+   */
+  public int getSynchronizationValue() {
+    return synchronization_;
+  }
+  /**
+   * <pre>
+   * Indicates when a distributed variable will be synced.
+   * </pre>
+   *
+   * <code>.tensorflow.VariableSynchronization synchronization = 8;</code>
+   */
+  public org.tensorflow.proto.framework.VariableSynchronization getSynchronization() {
+    @SuppressWarnings("deprecation")
+    org.tensorflow.proto.framework.VariableSynchronization result = org.tensorflow.proto.framework.VariableSynchronization.valueOf(synchronization_);
+    return result == null ? org.tensorflow.proto.framework.VariableSynchronization.UNRECOGNIZED : result;
+  }
+
+  public static final int AGGREGATION_FIELD_NUMBER = 9;
+  private int aggregation_;
+  /**
+   * <pre>
+   * Indicates how a distributed variable will be aggregated.
+   * </pre>
+   *
+   * <code>.tensorflow.VariableAggregation aggregation = 9;</code>
+   */
+  public int getAggregationValue() {
+    return aggregation_;
+  }
+  /**
+   * <pre>
+   * Indicates how a distributed variable will be aggregated.
+   * </pre>
+   *
+   * <code>.tensorflow.VariableAggregation aggregation = 9;</code>
+   */
+  public org.tensorflow.proto.framework.VariableAggregation getAggregation() {
+    @SuppressWarnings("deprecation")
+    org.tensorflow.proto.framework.VariableAggregation result = org.tensorflow.proto.framework.VariableAggregation.valueOf(aggregation_);
+    return result == null ? org.tensorflow.proto.framework.VariableAggregation.UNRECOGNIZED : result;
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    if (!getVariableNameBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 1, variableName_);
+    }
+    if (!getInitializerNameBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 2, initializerName_);
+    }
+    if (!getSnapshotNameBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 3, snapshotName_);
+    }
+    if (saveSliceInfoDef_ != null) {
+      output.writeMessage(4, getSaveSliceInfoDef());
+    }
+    if (isResource_ != false) {
+      output.writeBool(5, isResource_);
+    }
+    if (!getInitialValueNameBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 6, initialValueName_);
+    }
+    if (trainable_ != false) {
+      output.writeBool(7, trainable_);
+    }
+    if (synchronization_ != org.tensorflow.proto.framework.VariableSynchronization.VARIABLE_SYNCHRONIZATION_AUTO.getNumber()) {
+      output.writeEnum(8, synchronization_);
+    }
+    if (aggregation_ != org.tensorflow.proto.framework.VariableAggregation.VARIABLE_AGGREGATION_NONE.getNumber()) {
+      output.writeEnum(9, aggregation_);
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (!getVariableNameBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, variableName_);
+    }
+    if (!getInitializerNameBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(2, initializerName_);
+    }
+    if (!getSnapshotNameBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(3, snapshotName_);
+    }
+    if (saveSliceInfoDef_ != null) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(4, getSaveSliceInfoDef());
+    }
+    if (isResource_ != false) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeBoolSize(5, isResource_);
+    }
+    if (!getInitialValueNameBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(6, initialValueName_);
+    }
+    if (trainable_ != false) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeBoolSize(7, trainable_);
+    }
+    if (synchronization_ != org.tensorflow.proto.framework.VariableSynchronization.VARIABLE_SYNCHRONIZATION_AUTO.getNumber()) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeEnumSize(8, synchronization_);
+    }
+    if (aggregation_ != org.tensorflow.proto.framework.VariableAggregation.VARIABLE_AGGREGATION_NONE.getNumber()) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeEnumSize(9, aggregation_);
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.framework.VariableDef)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.framework.VariableDef other = (org.tensorflow.proto.framework.VariableDef) obj;
+
+    if (!getVariableName()
+        .equals(other.getVariableName())) return false;
+    if (!getInitialValueName()
+        .equals(other.getInitialValueName())) return false;
+    if (!getInitializerName()
+        .equals(other.getInitializerName())) return false;
+    if (!getSnapshotName()
+        .equals(other.getSnapshotName())) return false;
+    if (hasSaveSliceInfoDef() != other.hasSaveSliceInfoDef()) return false;
+    if (hasSaveSliceInfoDef()) {
+      if (!getSaveSliceInfoDef()
+          .equals(other.getSaveSliceInfoDef())) return false;
+    }
+    if (getIsResource()
+        != other.getIsResource()) return false;
+    if (getTrainable()
+        != other.getTrainable()) return false;
+    if (synchronization_ != other.synchronization_) return false;
+    if (aggregation_ != other.aggregation_) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    hash = (37 * hash) + VARIABLE_NAME_FIELD_NUMBER;
+    hash = (53 * hash) + getVariableName().hashCode();
+    hash = (37 * hash) + INITIAL_VALUE_NAME_FIELD_NUMBER;
+    hash = (53 * hash) + getInitialValueName().hashCode();
+    hash = (37 * hash) + INITIALIZER_NAME_FIELD_NUMBER;
+    hash = (53 * hash) + getInitializerName().hashCode();
+    hash = (37 * hash) + SNAPSHOT_NAME_FIELD_NUMBER;
+    hash = (53 * hash) + getSnapshotName().hashCode();
+    if (hasSaveSliceInfoDef()) {
+      hash = (37 * hash) + SAVE_SLICE_INFO_DEF_FIELD_NUMBER;
+      hash = (53 * hash) + getSaveSliceInfoDef().hashCode();
+    }
+    hash = (37 * hash) + IS_RESOURCE_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashBoolean(
+        getIsResource());
+    hash = (37 * hash) + TRAINABLE_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashBoolean(
+        getTrainable());
+    hash = (37 * hash) + SYNCHRONIZATION_FIELD_NUMBER;
+    hash = (53 * hash) + synchronization_;
+    hash = (37 * hash) + AGGREGATION_FIELD_NUMBER;
+    hash = (53 * hash) + aggregation_;
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.framework.VariableDef parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.VariableDef parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.VariableDef parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.VariableDef parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.VariableDef parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.VariableDef parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.VariableDef parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.VariableDef parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.VariableDef parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.VariableDef parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.VariableDef parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.VariableDef parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.framework.VariableDef prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * <pre>
+   * Protocol buffer representing a Variable.
+   * </pre>
+   *
+   * Protobuf type {@code tensorflow.VariableDef}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.VariableDef)
+      org.tensorflow.proto.framework.VariableDefOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.VariableProtos.internal_static_tensorflow_VariableDef_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.VariableProtos.internal_static_tensorflow_VariableDef_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.VariableDef.class, org.tensorflow.proto.framework.VariableDef.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.framework.VariableDef.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      variableName_ = "";
+
+      initialValueName_ = "";
+
+      initializerName_ = "";
+
+      snapshotName_ = "";
+
+      if (saveSliceInfoDefBuilder_ == null) {
+        saveSliceInfoDef_ = null;
+      } else {
+        saveSliceInfoDef_ = null;
+        saveSliceInfoDefBuilder_ = null;
+      }
+      isResource_ = false;
+
+      trainable_ = false;
+
+      synchronization_ = 0;
+
+      aggregation_ = 0;
+
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.framework.VariableProtos.internal_static_tensorflow_VariableDef_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.VariableDef getDefaultInstanceForType() {
+      return org.tensorflow.proto.framework.VariableDef.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.VariableDef build() {
+      org.tensorflow.proto.framework.VariableDef result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.VariableDef buildPartial() {
+      org.tensorflow.proto.framework.VariableDef result = new org.tensorflow.proto.framework.VariableDef(this);
+      result.variableName_ = variableName_;
+      result.initialValueName_ = initialValueName_;
+      result.initializerName_ = initializerName_;
+      result.snapshotName_ = snapshotName_;
+      if (saveSliceInfoDefBuilder_ == null) {
+        result.saveSliceInfoDef_ = saveSliceInfoDef_;
+      } else {
+        result.saveSliceInfoDef_ = saveSliceInfoDefBuilder_.build();
+      }
+      result.isResource_ = isResource_;
+      result.trainable_ = trainable_;
+      result.synchronization_ = synchronization_;
+      result.aggregation_ = aggregation_;
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.framework.VariableDef) {
+        return mergeFrom((org.tensorflow.proto.framework.VariableDef)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.framework.VariableDef other) {
+      if (other == org.tensorflow.proto.framework.VariableDef.getDefaultInstance()) return this;
+      if (!other.getVariableName().isEmpty()) {
+        variableName_ = other.variableName_;
+        onChanged();
+      }
+      if (!other.getInitialValueName().isEmpty()) {
+        initialValueName_ = other.initialValueName_;
+        onChanged();
+      }
+      if (!other.getInitializerName().isEmpty()) {
+        initializerName_ = other.initializerName_;
+        onChanged();
+      }
+      if (!other.getSnapshotName().isEmpty()) {
+        snapshotName_ = other.snapshotName_;
+        onChanged();
+      }
+      if (other.hasSaveSliceInfoDef()) {
+        mergeSaveSliceInfoDef(other.getSaveSliceInfoDef());
+      }
+      if (other.getIsResource() != false) {
+        setIsResource(other.getIsResource());
+      }
+      if (other.getTrainable() != false) {
+        setTrainable(other.getTrainable());
+      }
+      if (other.synchronization_ != 0) {
+        setSynchronizationValue(other.getSynchronizationValue());
+      }
+      if (other.aggregation_ != 0) {
+        setAggregationValue(other.getAggregationValue());
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.framework.VariableDef parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.framework.VariableDef) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+
+    private java.lang.Object variableName_ = "";
+    /**
+     * <pre>
+     * Name of the variable tensor.
+     * </pre>
+     *
+     * <code>string variable_name = 1;</code>
+     */
+    public java.lang.String getVariableName() {
+      java.lang.Object ref = variableName_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        variableName_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Name of the variable tensor.
+     * </pre>
+     *
+     * <code>string variable_name = 1;</code>
+     */
+    public com.google.protobuf.ByteString
+        getVariableNameBytes() {
+      java.lang.Object ref = variableName_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        variableName_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Name of the variable tensor.
+     * </pre>
+     *
+     * <code>string variable_name = 1;</code>
+     */
+    public Builder setVariableName(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      variableName_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Name of the variable tensor.
+     * </pre>
+     *
+     * <code>string variable_name = 1;</code>
+     */
+    public Builder clearVariableName() {
+      
+      variableName_ = getDefaultInstance().getVariableName();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Name of the variable tensor.
+     * </pre>
+     *
+     * <code>string variable_name = 1;</code>
+     */
+    public Builder setVariableNameBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      variableName_ = value;
+      onChanged();
+      return this;
+    }
+
+    private java.lang.Object initialValueName_ = "";
+    /**
+     * <pre>
+     * Name of the tensor holding the variable's initial value.
+     * </pre>
+     *
+     * <code>string initial_value_name = 6;</code>
+     */
+    public java.lang.String getInitialValueName() {
+      java.lang.Object ref = initialValueName_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        initialValueName_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Name of the tensor holding the variable's initial value.
+     * </pre>
+     *
+     * <code>string initial_value_name = 6;</code>
+     */
+    public com.google.protobuf.ByteString
+        getInitialValueNameBytes() {
+      java.lang.Object ref = initialValueName_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        initialValueName_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Name of the tensor holding the variable's initial value.
+     * </pre>
+     *
+     * <code>string initial_value_name = 6;</code>
+     */
+    public Builder setInitialValueName(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      initialValueName_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Name of the tensor holding the variable's initial value.
+     * </pre>
+     *
+     * <code>string initial_value_name = 6;</code>
+     */
+    public Builder clearInitialValueName() {
+      
+      initialValueName_ = getDefaultInstance().getInitialValueName();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Name of the tensor holding the variable's initial value.
+     * </pre>
+     *
+     * <code>string initial_value_name = 6;</code>
+     */
+    public Builder setInitialValueNameBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      initialValueName_ = value;
+      onChanged();
+      return this;
+    }
+
+    private java.lang.Object initializerName_ = "";
+    /**
+     * <pre>
+     * Name of the initializer op.
+     * </pre>
+     *
+     * <code>string initializer_name = 2;</code>
+     */
+    public java.lang.String getInitializerName() {
+      java.lang.Object ref = initializerName_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        initializerName_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Name of the initializer op.
+     * </pre>
+     *
+     * <code>string initializer_name = 2;</code>
+     */
+    public com.google.protobuf.ByteString
+        getInitializerNameBytes() {
+      java.lang.Object ref = initializerName_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        initializerName_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Name of the initializer op.
+     * </pre>
+     *
+     * <code>string initializer_name = 2;</code>
+     */
+    public Builder setInitializerName(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      initializerName_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Name of the initializer op.
+     * </pre>
+     *
+     * <code>string initializer_name = 2;</code>
+     */
+    public Builder clearInitializerName() {
+      
+      initializerName_ = getDefaultInstance().getInitializerName();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Name of the initializer op.
+     * </pre>
+     *
+     * <code>string initializer_name = 2;</code>
+     */
+    public Builder setInitializerNameBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      initializerName_ = value;
+      onChanged();
+      return this;
+    }
+
+    private java.lang.Object snapshotName_ = "";
+    /**
+     * <pre>
+     * Name of the snapshot tensor.
+     * </pre>
+     *
+     * <code>string snapshot_name = 3;</code>
+     */
+    public java.lang.String getSnapshotName() {
+      java.lang.Object ref = snapshotName_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        snapshotName_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Name of the snapshot tensor.
+     * </pre>
+     *
+     * <code>string snapshot_name = 3;</code>
+     */
+    public com.google.protobuf.ByteString
+        getSnapshotNameBytes() {
+      java.lang.Object ref = snapshotName_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        snapshotName_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Name of the snapshot tensor.
+     * </pre>
+     *
+     * <code>string snapshot_name = 3;</code>
+     */
+    public Builder setSnapshotName(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      snapshotName_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Name of the snapshot tensor.
+     * </pre>
+     *
+     * <code>string snapshot_name = 3;</code>
+     */
+    public Builder clearSnapshotName() {
+      
+      snapshotName_ = getDefaultInstance().getSnapshotName();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Name of the snapshot tensor.
+     * </pre>
+     *
+     * <code>string snapshot_name = 3;</code>
+     */
+    public Builder setSnapshotNameBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      snapshotName_ = value;
+      onChanged();
+      return this;
+    }
+
+    private org.tensorflow.proto.framework.SaveSliceInfoDef saveSliceInfoDef_;
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.SaveSliceInfoDef, org.tensorflow.proto.framework.SaveSliceInfoDef.Builder, org.tensorflow.proto.framework.SaveSliceInfoDefOrBuilder> saveSliceInfoDefBuilder_;
+    /**
+     * <pre>
+     * Support for saving variables as slices of a larger variable.
+     * </pre>
+     *
+     * <code>.tensorflow.SaveSliceInfoDef save_slice_info_def = 4;</code>
+     */
+    public boolean hasSaveSliceInfoDef() {
+      return saveSliceInfoDefBuilder_ != null || saveSliceInfoDef_ != null;
+    }
+    /**
+     * <pre>
+     * Support for saving variables as slices of a larger variable.
+     * </pre>
+     *
+     * <code>.tensorflow.SaveSliceInfoDef save_slice_info_def = 4;</code>
+     */
+    public org.tensorflow.proto.framework.SaveSliceInfoDef getSaveSliceInfoDef() {
+      if (saveSliceInfoDefBuilder_ == null) {
+        return saveSliceInfoDef_ == null ? org.tensorflow.proto.framework.SaveSliceInfoDef.getDefaultInstance() : saveSliceInfoDef_;
+      } else {
+        return saveSliceInfoDefBuilder_.getMessage();
+      }
+    }
+    /**
+     * <pre>
+     * Support for saving variables as slices of a larger variable.
+     * </pre>
+     *
+     * <code>.tensorflow.SaveSliceInfoDef save_slice_info_def = 4;</code>
+     */
+    public Builder setSaveSliceInfoDef(org.tensorflow.proto.framework.SaveSliceInfoDef value) {
+      if (saveSliceInfoDefBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        saveSliceInfoDef_ = value;
+        onChanged();
+      } else {
+        saveSliceInfoDefBuilder_.setMessage(value);
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Support for saving variables as slices of a larger variable.
+     * </pre>
+     *
+     * <code>.tensorflow.SaveSliceInfoDef save_slice_info_def = 4;</code>
+     */
+    public Builder setSaveSliceInfoDef(
+        org.tensorflow.proto.framework.SaveSliceInfoDef.Builder builderForValue) {
+      if (saveSliceInfoDefBuilder_ == null) {
+        saveSliceInfoDef_ = builderForValue.build();
+        onChanged();
+      } else {
+        saveSliceInfoDefBuilder_.setMessage(builderForValue.build());
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Support for saving variables as slices of a larger variable.
+     * </pre>
+     *
+     * <code>.tensorflow.SaveSliceInfoDef save_slice_info_def = 4;</code>
+     */
+    public Builder mergeSaveSliceInfoDef(org.tensorflow.proto.framework.SaveSliceInfoDef value) {
+      if (saveSliceInfoDefBuilder_ == null) {
+        if (saveSliceInfoDef_ != null) {
+          saveSliceInfoDef_ =
+            org.tensorflow.proto.framework.SaveSliceInfoDef.newBuilder(saveSliceInfoDef_).mergeFrom(value).buildPartial();
+        } else {
+          saveSliceInfoDef_ = value;
+        }
+        onChanged();
+      } else {
+        saveSliceInfoDefBuilder_.mergeFrom(value);
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Support for saving variables as slices of a larger variable.
+     * </pre>
+     *
+     * <code>.tensorflow.SaveSliceInfoDef save_slice_info_def = 4;</code>
+     */
+    public Builder clearSaveSliceInfoDef() {
+      if (saveSliceInfoDefBuilder_ == null) {
+        saveSliceInfoDef_ = null;
+        onChanged();
+      } else {
+        saveSliceInfoDef_ = null;
+        saveSliceInfoDefBuilder_ = null;
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Support for saving variables as slices of a larger variable.
+     * </pre>
+     *
+     * <code>.tensorflow.SaveSliceInfoDef save_slice_info_def = 4;</code>
+     */
+    public org.tensorflow.proto.framework.SaveSliceInfoDef.Builder getSaveSliceInfoDefBuilder() {
+      
+      onChanged();
+      return getSaveSliceInfoDefFieldBuilder().getBuilder();
+    }
+    /**
+     * <pre>
+     * Support for saving variables as slices of a larger variable.
+     * </pre>
+     *
+     * <code>.tensorflow.SaveSliceInfoDef save_slice_info_def = 4;</code>
+     */
+    public org.tensorflow.proto.framework.SaveSliceInfoDefOrBuilder getSaveSliceInfoDefOrBuilder() {
+      if (saveSliceInfoDefBuilder_ != null) {
+        return saveSliceInfoDefBuilder_.getMessageOrBuilder();
+      } else {
+        return saveSliceInfoDef_ == null ?
+            org.tensorflow.proto.framework.SaveSliceInfoDef.getDefaultInstance() : saveSliceInfoDef_;
+      }
+    }
+    /**
+     * <pre>
+     * Support for saving variables as slices of a larger variable.
+     * </pre>
+     *
+     * <code>.tensorflow.SaveSliceInfoDef save_slice_info_def = 4;</code>
+     */
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.SaveSliceInfoDef, org.tensorflow.proto.framework.SaveSliceInfoDef.Builder, org.tensorflow.proto.framework.SaveSliceInfoDefOrBuilder> 
+        getSaveSliceInfoDefFieldBuilder() {
+      if (saveSliceInfoDefBuilder_ == null) {
+        saveSliceInfoDefBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+            org.tensorflow.proto.framework.SaveSliceInfoDef, org.tensorflow.proto.framework.SaveSliceInfoDef.Builder, org.tensorflow.proto.framework.SaveSliceInfoDefOrBuilder>(
+                getSaveSliceInfoDef(),
+                getParentForChildren(),
+                isClean());
+        saveSliceInfoDef_ = null;
+      }
+      return saveSliceInfoDefBuilder_;
+    }
+
+    private boolean isResource_ ;
+    /**
+     * <pre>
+     * Whether to represent this as a ResourceVariable.
+     * </pre>
+     *
+     * <code>bool is_resource = 5;</code>
+     */
+    public boolean getIsResource() {
+      return isResource_;
+    }
+    /**
+     * <pre>
+     * Whether to represent this as a ResourceVariable.
+     * </pre>
+     *
+     * <code>bool is_resource = 5;</code>
+     */
+    public Builder setIsResource(boolean value) {
+      
+      isResource_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Whether to represent this as a ResourceVariable.
+     * </pre>
+     *
+     * <code>bool is_resource = 5;</code>
+     */
+    public Builder clearIsResource() {
+      
+      isResource_ = false;
+      onChanged();
+      return this;
+    }
+
+    private boolean trainable_ ;
+    /**
+     * <pre>
+     * Whether this variable should be trained.
+     * </pre>
+     *
+     * <code>bool trainable = 7;</code>
+     */
+    public boolean getTrainable() {
+      return trainable_;
+    }
+    /**
+     * <pre>
+     * Whether this variable should be trained.
+     * </pre>
+     *
+     * <code>bool trainable = 7;</code>
+     */
+    public Builder setTrainable(boolean value) {
+      
+      trainable_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Whether this variable should be trained.
+     * </pre>
+     *
+     * <code>bool trainable = 7;</code>
+     */
+    public Builder clearTrainable() {
+      
+      trainable_ = false;
+      onChanged();
+      return this;
+    }
+
+    private int synchronization_ = 0;
+    /**
+     * <pre>
+     * Indicates when a distributed variable will be synced.
+     * </pre>
+     *
+     * <code>.tensorflow.VariableSynchronization synchronization = 8;</code>
+     */
+    public int getSynchronizationValue() {
+      return synchronization_;
+    }
+    /**
+     * <pre>
+     * Indicates when a distributed variable will be synced.
+     * </pre>
+     *
+     * <code>.tensorflow.VariableSynchronization synchronization = 8;</code>
+     */
+    public Builder setSynchronizationValue(int value) {
+      synchronization_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Indicates when a distributed variable will be synced.
+     * </pre>
+     *
+     * <code>.tensorflow.VariableSynchronization synchronization = 8;</code>
+     */
+    public org.tensorflow.proto.framework.VariableSynchronization getSynchronization() {
+      @SuppressWarnings("deprecation")
+      org.tensorflow.proto.framework.VariableSynchronization result = org.tensorflow.proto.framework.VariableSynchronization.valueOf(synchronization_);
+      return result == null ? org.tensorflow.proto.framework.VariableSynchronization.UNRECOGNIZED : result;
+    }
+    /**
+     * <pre>
+     * Indicates when a distributed variable will be synced.
+     * </pre>
+     *
+     * <code>.tensorflow.VariableSynchronization synchronization = 8;</code>
+     */
+    public Builder setSynchronization(org.tensorflow.proto.framework.VariableSynchronization value) {
+      if (value == null) {
+        throw new NullPointerException();
+      }
+      
+      synchronization_ = value.getNumber();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Indicates when a distributed variable will be synced.
+     * </pre>
+     *
+     * <code>.tensorflow.VariableSynchronization synchronization = 8;</code>
+     */
+    public Builder clearSynchronization() {
+      
+      synchronization_ = 0;
+      onChanged();
+      return this;
+    }
+
+    private int aggregation_ = 0;
+    /**
+     * <pre>
+     * Indicates how a distributed variable will be aggregated.
+     * </pre>
+     *
+     * <code>.tensorflow.VariableAggregation aggregation = 9;</code>
+     */
+    public int getAggregationValue() {
+      return aggregation_;
+    }
+    /**
+     * <pre>
+     * Indicates how a distributed variable will be aggregated.
+     * </pre>
+     *
+     * <code>.tensorflow.VariableAggregation aggregation = 9;</code>
+     */
+    public Builder setAggregationValue(int value) {
+      aggregation_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Indicates how a distributed variable will be aggregated.
+     * </pre>
+     *
+     * <code>.tensorflow.VariableAggregation aggregation = 9;</code>
+     */
+    public org.tensorflow.proto.framework.VariableAggregation getAggregation() {
+      @SuppressWarnings("deprecation")
+      org.tensorflow.proto.framework.VariableAggregation result = org.tensorflow.proto.framework.VariableAggregation.valueOf(aggregation_);
+      return result == null ? org.tensorflow.proto.framework.VariableAggregation.UNRECOGNIZED : result;
+    }
+    /**
+     * <pre>
+     * Indicates how a distributed variable will be aggregated.
+     * </pre>
+     *
+     * <code>.tensorflow.VariableAggregation aggregation = 9;</code>
+     */
+    public Builder setAggregation(org.tensorflow.proto.framework.VariableAggregation value) {
+      if (value == null) {
+        throw new NullPointerException();
+      }
+      
+      aggregation_ = value.getNumber();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Indicates how a distributed variable will be aggregated.
+     * </pre>
+     *
+     * <code>.tensorflow.VariableAggregation aggregation = 9;</code>
+     */
+    public Builder clearAggregation() {
+      
+      aggregation_ = 0;
+      onChanged();
+      return this;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.VariableDef)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.VariableDef)
+  private static final org.tensorflow.proto.framework.VariableDef DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.VariableDef();
+  }
+
+  public static org.tensorflow.proto.framework.VariableDef getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<VariableDef>
+      PARSER = new com.google.protobuf.AbstractParser<VariableDef>() {
+    @java.lang.Override
+    public VariableDef parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new VariableDef(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<VariableDef> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<VariableDef> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.framework.VariableDef getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/VariableDefOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/VariableDefOrBuilder.java
new file mode 100644
index 00000000000..31cf5f0207a
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/VariableDefOrBuilder.java
@@ -0,0 +1,158 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/framework/variable.proto
+
+package org.tensorflow.proto.framework;
+
+public interface VariableDefOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.VariableDef)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <pre>
+   * Name of the variable tensor.
+   * </pre>
+   *
+   * <code>string variable_name = 1;</code>
+   */
+  java.lang.String getVariableName();
+  /**
+   * <pre>
+   * Name of the variable tensor.
+   * </pre>
+   *
+   * <code>string variable_name = 1;</code>
+   */
+  com.google.protobuf.ByteString
+      getVariableNameBytes();
+
+  /**
+   * <pre>
+   * Name of the tensor holding the variable's initial value.
+   * </pre>
+   *
+   * <code>string initial_value_name = 6;</code>
+   */
+  java.lang.String getInitialValueName();
+  /**
+   * <pre>
+   * Name of the tensor holding the variable's initial value.
+   * </pre>
+   *
+   * <code>string initial_value_name = 6;</code>
+   */
+  com.google.protobuf.ByteString
+      getInitialValueNameBytes();
+
+  /**
+   * <pre>
+   * Name of the initializer op.
+   * </pre>
+   *
+   * <code>string initializer_name = 2;</code>
+   */
+  java.lang.String getInitializerName();
+  /**
+   * <pre>
+   * Name of the initializer op.
+   * </pre>
+   *
+   * <code>string initializer_name = 2;</code>
+   */
+  com.google.protobuf.ByteString
+      getInitializerNameBytes();
+
+  /**
+   * <pre>
+   * Name of the snapshot tensor.
+   * </pre>
+   *
+   * <code>string snapshot_name = 3;</code>
+   */
+  java.lang.String getSnapshotName();
+  /**
+   * <pre>
+   * Name of the snapshot tensor.
+   * </pre>
+   *
+   * <code>string snapshot_name = 3;</code>
+   */
+  com.google.protobuf.ByteString
+      getSnapshotNameBytes();
+
+  /**
+   * <pre>
+   * Support for saving variables as slices of a larger variable.
+   * </pre>
+   *
+   * <code>.tensorflow.SaveSliceInfoDef save_slice_info_def = 4;</code>
+   */
+  boolean hasSaveSliceInfoDef();
+  /**
+   * <pre>
+   * Support for saving variables as slices of a larger variable.
+   * </pre>
+   *
+   * <code>.tensorflow.SaveSliceInfoDef save_slice_info_def = 4;</code>
+   */
+  org.tensorflow.proto.framework.SaveSliceInfoDef getSaveSliceInfoDef();
+  /**
+   * <pre>
+   * Support for saving variables as slices of a larger variable.
+   * </pre>
+   *
+   * <code>.tensorflow.SaveSliceInfoDef save_slice_info_def = 4;</code>
+   */
+  org.tensorflow.proto.framework.SaveSliceInfoDefOrBuilder getSaveSliceInfoDefOrBuilder();
+
+  /**
+   * <pre>
+   * Whether to represent this as a ResourceVariable.
+   * </pre>
+   *
+   * <code>bool is_resource = 5;</code>
+   */
+  boolean getIsResource();
+
+  /**
+   * <pre>
+   * Whether this variable should be trained.
+   * </pre>
+   *
+   * <code>bool trainable = 7;</code>
+   */
+  boolean getTrainable();
+
+  /**
+   * <pre>
+   * Indicates when a distributed variable will be synced.
+   * </pre>
+   *
+   * <code>.tensorflow.VariableSynchronization synchronization = 8;</code>
+   */
+  int getSynchronizationValue();
+  /**
+   * <pre>
+   * Indicates when a distributed variable will be synced.
+   * </pre>
+   *
+   * <code>.tensorflow.VariableSynchronization synchronization = 8;</code>
+   */
+  org.tensorflow.proto.framework.VariableSynchronization getSynchronization();
+
+  /**
+   * <pre>
+   * Indicates how a distributed variable will be aggregated.
+   * </pre>
+   *
+   * <code>.tensorflow.VariableAggregation aggregation = 9;</code>
+   */
+  int getAggregationValue();
+  /**
+   * <pre>
+   * Indicates how a distributed variable will be aggregated.
+   * </pre>
+   *
+   * <code>.tensorflow.VariableAggregation aggregation = 9;</code>
+   */
+  org.tensorflow.proto.framework.VariableAggregation getAggregation();
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/VariableProtos.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/VariableProtos.java
new file mode 100644
index 00000000000..1d8c7d85e0a
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/VariableProtos.java
@@ -0,0 +1,81 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/framework/variable.proto
+
+package org.tensorflow.proto.framework;
+
+public final class VariableProtos {
+  private VariableProtos() {}
+  public static void registerAllExtensions(
+      com.google.protobuf.ExtensionRegistryLite registry) {
+  }
+
+  public static void registerAllExtensions(
+      com.google.protobuf.ExtensionRegistry registry) {
+    registerAllExtensions(
+        (com.google.protobuf.ExtensionRegistryLite) registry);
+  }
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_VariableDef_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_VariableDef_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_SaveSliceInfoDef_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_SaveSliceInfoDef_fieldAccessorTable;
+
+  public static com.google.protobuf.Descriptors.FileDescriptor
+      getDescriptor() {
+    return descriptor;
+  }
+  private static  com.google.protobuf.Descriptors.FileDescriptor
+      descriptor;
+  static {
+    java.lang.String[] descriptorData = {
+      "\n(tensorflow/core/framework/variable.pro" +
+      "to\022\ntensorflow\"\310\002\n\013VariableDef\022\025\n\rvariab" +
+      "le_name\030\001 \001(\t\022\032\n\022initial_value_name\030\006 \001(" +
+      "\t\022\030\n\020initializer_name\030\002 \001(\t\022\025\n\rsnapshot_" +
+      "name\030\003 \001(\t\0229\n\023save_slice_info_def\030\004 \001(\0132" +
+      "\034.tensorflow.SaveSliceInfoDef\022\023\n\013is_reso" +
+      "urce\030\005 \001(\010\022\021\n\ttrainable\030\007 \001(\010\022<\n\017synchro" +
+      "nization\030\010 \001(\0162#.tensorflow.VariableSync" +
+      "hronization\0224\n\013aggregation\030\t \001(\0162\037.tenso" +
+      "rflow.VariableAggregation\"`\n\020SaveSliceIn" +
+      "foDef\022\021\n\tfull_name\030\001 \001(\t\022\022\n\nfull_shape\030\002" +
+      " \003(\003\022\022\n\nvar_offset\030\003 \003(\003\022\021\n\tvar_shape\030\004 " +
+      "\003(\003*\254\001\n\027VariableSynchronization\022!\n\035VARIA" +
+      "BLE_SYNCHRONIZATION_AUTO\020\000\022!\n\035VARIABLE_S" +
+      "YNCHRONIZATION_NONE\020\001\022%\n!VARIABLE_SYNCHR" +
+      "ONIZATION_ON_WRITE\020\002\022$\n VARIABLE_SYNCHRO" +
+      "NIZATION_ON_READ\020\003*\236\001\n\023VariableAggregati" +
+      "on\022\035\n\031VARIABLE_AGGREGATION_NONE\020\000\022\034\n\030VAR" +
+      "IABLE_AGGREGATION_SUM\020\001\022\035\n\031VARIABLE_AGGR" +
+      "EGATION_MEAN\020\002\022+\n\'VARIABLE_AGGREGATION_O" +
+      "NLY_FIRST_REPLICA\020\003B\206\001\n\036org.tensorflow.p" +
+      "roto.frameworkB\016VariableProtosP\001ZOgithub" +
+      ".com/tensorflow/tensorflow/tensorflow/go" +
+      "/core/framework/variable_go_proto\370\001\001b\006pr" +
+      "oto3"
+    };
+    descriptor = com.google.protobuf.Descriptors.FileDescriptor
+      .internalBuildGeneratedFileFrom(descriptorData,
+        new com.google.protobuf.Descriptors.FileDescriptor[] {
+        });
+    internal_static_tensorflow_VariableDef_descriptor =
+      getDescriptor().getMessageTypes().get(0);
+    internal_static_tensorflow_VariableDef_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_VariableDef_descriptor,
+        new java.lang.String[] { "VariableName", "InitialValueName", "InitializerName", "SnapshotName", "SaveSliceInfoDef", "IsResource", "Trainable", "Synchronization", "Aggregation", });
+    internal_static_tensorflow_SaveSliceInfoDef_descriptor =
+      getDescriptor().getMessageTypes().get(1);
+    internal_static_tensorflow_SaveSliceInfoDef_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_SaveSliceInfoDef_descriptor,
+        new java.lang.String[] { "FullName", "FullShape", "VarOffset", "VarShape", });
+  }
+
+  // @@protoc_insertion_point(outer_class_scope)
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/VariableSynchronization.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/VariableSynchronization.java
new file mode 100644
index 00000000000..b71b26cdcc1
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/VariableSynchronization.java
@@ -0,0 +1,169 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/framework/variable.proto
+
+package org.tensorflow.proto.framework;
+
+/**
+ * <pre>
+ * Indicates when a distributed variable will be synced.
+ * </pre>
+ *
+ * Protobuf enum {@code tensorflow.VariableSynchronization}
+ */
+public enum VariableSynchronization
+    implements com.google.protobuf.ProtocolMessageEnum {
+  /**
+   * <pre>
+   * `AUTO`: Indicates that the synchronization will be determined by the
+   * current `DistributionStrategy` (eg. With `MirroredStrategy` this would be
+   * `ON_WRITE`).
+   * </pre>
+   *
+   * <code>VARIABLE_SYNCHRONIZATION_AUTO = 0;</code>
+   */
+  VARIABLE_SYNCHRONIZATION_AUTO(0),
+  /**
+   * <pre>
+   * `NONE`: Indicates that there will only be one copy of the variable, so
+   * there is no need to sync.
+   * </pre>
+   *
+   * <code>VARIABLE_SYNCHRONIZATION_NONE = 1;</code>
+   */
+  VARIABLE_SYNCHRONIZATION_NONE(1),
+  /**
+   * <pre>
+   * `ON_WRITE`: Indicates that the variable will be updated across devices
+   * every time it is written.
+   * </pre>
+   *
+   * <code>VARIABLE_SYNCHRONIZATION_ON_WRITE = 2;</code>
+   */
+  VARIABLE_SYNCHRONIZATION_ON_WRITE(2),
+  /**
+   * <pre>
+   * `ON_READ`: Indicates that the variable will be aggregated across devices
+   * when it is read (eg. when checkpointing or when evaluating an op that uses
+   * the variable).
+   * </pre>
+   *
+   * <code>VARIABLE_SYNCHRONIZATION_ON_READ = 3;</code>
+   */
+  VARIABLE_SYNCHRONIZATION_ON_READ(3),
+  UNRECOGNIZED(-1),
+  ;
+
+  /**
+   * <pre>
+   * `AUTO`: Indicates that the synchronization will be determined by the
+   * current `DistributionStrategy` (eg. With `MirroredStrategy` this would be
+   * `ON_WRITE`).
+   * </pre>
+   *
+   * <code>VARIABLE_SYNCHRONIZATION_AUTO = 0;</code>
+   */
+  public static final int VARIABLE_SYNCHRONIZATION_AUTO_VALUE = 0;
+  /**
+   * <pre>
+   * `NONE`: Indicates that there will only be one copy of the variable, so
+   * there is no need to sync.
+   * </pre>
+   *
+   * <code>VARIABLE_SYNCHRONIZATION_NONE = 1;</code>
+   */
+  public static final int VARIABLE_SYNCHRONIZATION_NONE_VALUE = 1;
+  /**
+   * <pre>
+   * `ON_WRITE`: Indicates that the variable will be updated across devices
+   * every time it is written.
+   * </pre>
+   *
+   * <code>VARIABLE_SYNCHRONIZATION_ON_WRITE = 2;</code>
+   */
+  public static final int VARIABLE_SYNCHRONIZATION_ON_WRITE_VALUE = 2;
+  /**
+   * <pre>
+   * `ON_READ`: Indicates that the variable will be aggregated across devices
+   * when it is read (eg. when checkpointing or when evaluating an op that uses
+   * the variable).
+   * </pre>
+   *
+   * <code>VARIABLE_SYNCHRONIZATION_ON_READ = 3;</code>
+   */
+  public static final int VARIABLE_SYNCHRONIZATION_ON_READ_VALUE = 3;
+
+
+  public final int getNumber() {
+    if (this == UNRECOGNIZED) {
+      throw new java.lang.IllegalArgumentException(
+          "Can't get the number of an unknown enum value.");
+    }
+    return value;
+  }
+
+  /**
+   * @deprecated Use {@link #forNumber(int)} instead.
+   */
+  @java.lang.Deprecated
+  public static VariableSynchronization valueOf(int value) {
+    return forNumber(value);
+  }
+
+  public static VariableSynchronization forNumber(int value) {
+    switch (value) {
+      case 0: return VARIABLE_SYNCHRONIZATION_AUTO;
+      case 1: return VARIABLE_SYNCHRONIZATION_NONE;
+      case 2: return VARIABLE_SYNCHRONIZATION_ON_WRITE;
+      case 3: return VARIABLE_SYNCHRONIZATION_ON_READ;
+      default: return null;
+    }
+  }
+
+  public static com.google.protobuf.Internal.EnumLiteMap<VariableSynchronization>
+      internalGetValueMap() {
+    return internalValueMap;
+  }
+  private static final com.google.protobuf.Internal.EnumLiteMap<
+      VariableSynchronization> internalValueMap =
+        new com.google.protobuf.Internal.EnumLiteMap<VariableSynchronization>() {
+          public VariableSynchronization findValueByNumber(int number) {
+            return VariableSynchronization.forNumber(number);
+          }
+        };
+
+  public final com.google.protobuf.Descriptors.EnumValueDescriptor
+      getValueDescriptor() {
+    return getDescriptor().getValues().get(ordinal());
+  }
+  public final com.google.protobuf.Descriptors.EnumDescriptor
+      getDescriptorForType() {
+    return getDescriptor();
+  }
+  public static final com.google.protobuf.Descriptors.EnumDescriptor
+      getDescriptor() {
+    return org.tensorflow.proto.framework.VariableProtos.getDescriptor().getEnumTypes().get(0);
+  }
+
+  private static final VariableSynchronization[] VALUES = values();
+
+  public static VariableSynchronization valueOf(
+      com.google.protobuf.Descriptors.EnumValueDescriptor desc) {
+    if (desc.getType() != getDescriptor()) {
+      throw new java.lang.IllegalArgumentException(
+        "EnumValueDescriptor is not for this type.");
+    }
+    if (desc.getIndex() == -1) {
+      return UNRECOGNIZED;
+    }
+    return VALUES[desc.getIndex()];
+  }
+
+  private final int value;
+
+  private VariableSynchronization(int value) {
+    this.value = value;
+  }
+
+  // @@protoc_insertion_point(enum_scope:tensorflow.VariableSynchronization)
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/VariantTensorDataProto.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/VariantTensorDataProto.java
new file mode 100644
index 00000000000..47555bc3d7a
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/VariantTensorDataProto.java
@@ -0,0 +1,1097 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/framework/tensor.proto
+
+package org.tensorflow.proto.framework;
+
+/**
+ * <pre>
+ * Protocol buffer representing the serialization format of DT_VARIANT tensors.
+ * </pre>
+ *
+ * Protobuf type {@code tensorflow.VariantTensorDataProto}
+ */
+public  final class VariantTensorDataProto extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.VariantTensorDataProto)
+    VariantTensorDataProtoOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use VariantTensorDataProto.newBuilder() to construct.
+  private VariantTensorDataProto(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private VariantTensorDataProto() {
+    typeName_ = "";
+    metadata_ = com.google.protobuf.ByteString.EMPTY;
+    tensors_ = java.util.Collections.emptyList();
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new VariantTensorDataProto();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private VariantTensorDataProto(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    int mutable_bitField0_ = 0;
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 10: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            typeName_ = s;
+            break;
+          }
+          case 18: {
+
+            metadata_ = input.readBytes();
+            break;
+          }
+          case 26: {
+            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
+              tensors_ = new java.util.ArrayList<org.tensorflow.proto.framework.TensorProto>();
+              mutable_bitField0_ |= 0x00000001;
+            }
+            tensors_.add(
+                input.readMessage(org.tensorflow.proto.framework.TensorProto.parser(), extensionRegistry));
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      if (((mutable_bitField0_ & 0x00000001) != 0)) {
+        tensors_ = java.util.Collections.unmodifiableList(tensors_);
+      }
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.framework.TensorProtos.internal_static_tensorflow_VariantTensorDataProto_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.framework.TensorProtos.internal_static_tensorflow_VariantTensorDataProto_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.framework.VariantTensorDataProto.class, org.tensorflow.proto.framework.VariantTensorDataProto.Builder.class);
+  }
+
+  public static final int TYPE_NAME_FIELD_NUMBER = 1;
+  private volatile java.lang.Object typeName_;
+  /**
+   * <pre>
+   * Name of the type of objects being serialized.
+   * </pre>
+   *
+   * <code>string type_name = 1;</code>
+   */
+  public java.lang.String getTypeName() {
+    java.lang.Object ref = typeName_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      typeName_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * Name of the type of objects being serialized.
+   * </pre>
+   *
+   * <code>string type_name = 1;</code>
+   */
+  public com.google.protobuf.ByteString
+      getTypeNameBytes() {
+    java.lang.Object ref = typeName_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      typeName_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int METADATA_FIELD_NUMBER = 2;
+  private com.google.protobuf.ByteString metadata_;
+  /**
+   * <pre>
+   * Portions of the object that are not Tensors.
+   * </pre>
+   *
+   * <code>bytes metadata = 2;</code>
+   */
+  public com.google.protobuf.ByteString getMetadata() {
+    return metadata_;
+  }
+
+  public static final int TENSORS_FIELD_NUMBER = 3;
+  private java.util.List<org.tensorflow.proto.framework.TensorProto> tensors_;
+  /**
+   * <pre>
+   * Tensors contained within objects being serialized.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.TensorProto tensors = 3;</code>
+   */
+  public java.util.List<org.tensorflow.proto.framework.TensorProto> getTensorsList() {
+    return tensors_;
+  }
+  /**
+   * <pre>
+   * Tensors contained within objects being serialized.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.TensorProto tensors = 3;</code>
+   */
+  public java.util.List<? extends org.tensorflow.proto.framework.TensorProtoOrBuilder> 
+      getTensorsOrBuilderList() {
+    return tensors_;
+  }
+  /**
+   * <pre>
+   * Tensors contained within objects being serialized.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.TensorProto tensors = 3;</code>
+   */
+  public int getTensorsCount() {
+    return tensors_.size();
+  }
+  /**
+   * <pre>
+   * Tensors contained within objects being serialized.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.TensorProto tensors = 3;</code>
+   */
+  public org.tensorflow.proto.framework.TensorProto getTensors(int index) {
+    return tensors_.get(index);
+  }
+  /**
+   * <pre>
+   * Tensors contained within objects being serialized.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.TensorProto tensors = 3;</code>
+   */
+  public org.tensorflow.proto.framework.TensorProtoOrBuilder getTensorsOrBuilder(
+      int index) {
+    return tensors_.get(index);
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    if (!getTypeNameBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 1, typeName_);
+    }
+    if (!metadata_.isEmpty()) {
+      output.writeBytes(2, metadata_);
+    }
+    for (int i = 0; i < tensors_.size(); i++) {
+      output.writeMessage(3, tensors_.get(i));
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (!getTypeNameBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, typeName_);
+    }
+    if (!metadata_.isEmpty()) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeBytesSize(2, metadata_);
+    }
+    for (int i = 0; i < tensors_.size(); i++) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(3, tensors_.get(i));
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.framework.VariantTensorDataProto)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.framework.VariantTensorDataProto other = (org.tensorflow.proto.framework.VariantTensorDataProto) obj;
+
+    if (!getTypeName()
+        .equals(other.getTypeName())) return false;
+    if (!getMetadata()
+        .equals(other.getMetadata())) return false;
+    if (!getTensorsList()
+        .equals(other.getTensorsList())) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    hash = (37 * hash) + TYPE_NAME_FIELD_NUMBER;
+    hash = (53 * hash) + getTypeName().hashCode();
+    hash = (37 * hash) + METADATA_FIELD_NUMBER;
+    hash = (53 * hash) + getMetadata().hashCode();
+    if (getTensorsCount() > 0) {
+      hash = (37 * hash) + TENSORS_FIELD_NUMBER;
+      hash = (53 * hash) + getTensorsList().hashCode();
+    }
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.framework.VariantTensorDataProto parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.VariantTensorDataProto parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.VariantTensorDataProto parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.VariantTensorDataProto parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.VariantTensorDataProto parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.VariantTensorDataProto parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.VariantTensorDataProto parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.VariantTensorDataProto parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.VariantTensorDataProto parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.VariantTensorDataProto parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.VariantTensorDataProto parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.VariantTensorDataProto parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.framework.VariantTensorDataProto prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * <pre>
+   * Protocol buffer representing the serialization format of DT_VARIANT tensors.
+   * </pre>
+   *
+   * Protobuf type {@code tensorflow.VariantTensorDataProto}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.VariantTensorDataProto)
+      org.tensorflow.proto.framework.VariantTensorDataProtoOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.TensorProtos.internal_static_tensorflow_VariantTensorDataProto_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.TensorProtos.internal_static_tensorflow_VariantTensorDataProto_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.VariantTensorDataProto.class, org.tensorflow.proto.framework.VariantTensorDataProto.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.framework.VariantTensorDataProto.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+        getTensorsFieldBuilder();
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      typeName_ = "";
+
+      metadata_ = com.google.protobuf.ByteString.EMPTY;
+
+      if (tensorsBuilder_ == null) {
+        tensors_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000001);
+      } else {
+        tensorsBuilder_.clear();
+      }
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.framework.TensorProtos.internal_static_tensorflow_VariantTensorDataProto_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.VariantTensorDataProto getDefaultInstanceForType() {
+      return org.tensorflow.proto.framework.VariantTensorDataProto.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.VariantTensorDataProto build() {
+      org.tensorflow.proto.framework.VariantTensorDataProto result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.VariantTensorDataProto buildPartial() {
+      org.tensorflow.proto.framework.VariantTensorDataProto result = new org.tensorflow.proto.framework.VariantTensorDataProto(this);
+      int from_bitField0_ = bitField0_;
+      result.typeName_ = typeName_;
+      result.metadata_ = metadata_;
+      if (tensorsBuilder_ == null) {
+        if (((bitField0_ & 0x00000001) != 0)) {
+          tensors_ = java.util.Collections.unmodifiableList(tensors_);
+          bitField0_ = (bitField0_ & ~0x00000001);
+        }
+        result.tensors_ = tensors_;
+      } else {
+        result.tensors_ = tensorsBuilder_.build();
+      }
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.framework.VariantTensorDataProto) {
+        return mergeFrom((org.tensorflow.proto.framework.VariantTensorDataProto)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.framework.VariantTensorDataProto other) {
+      if (other == org.tensorflow.proto.framework.VariantTensorDataProto.getDefaultInstance()) return this;
+      if (!other.getTypeName().isEmpty()) {
+        typeName_ = other.typeName_;
+        onChanged();
+      }
+      if (other.getMetadata() != com.google.protobuf.ByteString.EMPTY) {
+        setMetadata(other.getMetadata());
+      }
+      if (tensorsBuilder_ == null) {
+        if (!other.tensors_.isEmpty()) {
+          if (tensors_.isEmpty()) {
+            tensors_ = other.tensors_;
+            bitField0_ = (bitField0_ & ~0x00000001);
+          } else {
+            ensureTensorsIsMutable();
+            tensors_.addAll(other.tensors_);
+          }
+          onChanged();
+        }
+      } else {
+        if (!other.tensors_.isEmpty()) {
+          if (tensorsBuilder_.isEmpty()) {
+            tensorsBuilder_.dispose();
+            tensorsBuilder_ = null;
+            tensors_ = other.tensors_;
+            bitField0_ = (bitField0_ & ~0x00000001);
+            tensorsBuilder_ = 
+              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
+                 getTensorsFieldBuilder() : null;
+          } else {
+            tensorsBuilder_.addAllMessages(other.tensors_);
+          }
+        }
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.framework.VariantTensorDataProto parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.framework.VariantTensorDataProto) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+    private int bitField0_;
+
+    private java.lang.Object typeName_ = "";
+    /**
+     * <pre>
+     * Name of the type of objects being serialized.
+     * </pre>
+     *
+     * <code>string type_name = 1;</code>
+     */
+    public java.lang.String getTypeName() {
+      java.lang.Object ref = typeName_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        typeName_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Name of the type of objects being serialized.
+     * </pre>
+     *
+     * <code>string type_name = 1;</code>
+     */
+    public com.google.protobuf.ByteString
+        getTypeNameBytes() {
+      java.lang.Object ref = typeName_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        typeName_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Name of the type of objects being serialized.
+     * </pre>
+     *
+     * <code>string type_name = 1;</code>
+     */
+    public Builder setTypeName(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      typeName_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Name of the type of objects being serialized.
+     * </pre>
+     *
+     * <code>string type_name = 1;</code>
+     */
+    public Builder clearTypeName() {
+      
+      typeName_ = getDefaultInstance().getTypeName();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Name of the type of objects being serialized.
+     * </pre>
+     *
+     * <code>string type_name = 1;</code>
+     */
+    public Builder setTypeNameBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      typeName_ = value;
+      onChanged();
+      return this;
+    }
+
+    private com.google.protobuf.ByteString metadata_ = com.google.protobuf.ByteString.EMPTY;
+    /**
+     * <pre>
+     * Portions of the object that are not Tensors.
+     * </pre>
+     *
+     * <code>bytes metadata = 2;</code>
+     */
+    public com.google.protobuf.ByteString getMetadata() {
+      return metadata_;
+    }
+    /**
+     * <pre>
+     * Portions of the object that are not Tensors.
+     * </pre>
+     *
+     * <code>bytes metadata = 2;</code>
+     */
+    public Builder setMetadata(com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      metadata_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Portions of the object that are not Tensors.
+     * </pre>
+     *
+     * <code>bytes metadata = 2;</code>
+     */
+    public Builder clearMetadata() {
+      
+      metadata_ = getDefaultInstance().getMetadata();
+      onChanged();
+      return this;
+    }
+
+    private java.util.List<org.tensorflow.proto.framework.TensorProto> tensors_ =
+      java.util.Collections.emptyList();
+    private void ensureTensorsIsMutable() {
+      if (!((bitField0_ & 0x00000001) != 0)) {
+        tensors_ = new java.util.ArrayList<org.tensorflow.proto.framework.TensorProto>(tensors_);
+        bitField0_ |= 0x00000001;
+       }
+    }
+
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.framework.TensorProto, org.tensorflow.proto.framework.TensorProto.Builder, org.tensorflow.proto.framework.TensorProtoOrBuilder> tensorsBuilder_;
+
+    /**
+     * <pre>
+     * Tensors contained within objects being serialized.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorProto tensors = 3;</code>
+     */
+    public java.util.List<org.tensorflow.proto.framework.TensorProto> getTensorsList() {
+      if (tensorsBuilder_ == null) {
+        return java.util.Collections.unmodifiableList(tensors_);
+      } else {
+        return tensorsBuilder_.getMessageList();
+      }
+    }
+    /**
+     * <pre>
+     * Tensors contained within objects being serialized.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorProto tensors = 3;</code>
+     */
+    public int getTensorsCount() {
+      if (tensorsBuilder_ == null) {
+        return tensors_.size();
+      } else {
+        return tensorsBuilder_.getCount();
+      }
+    }
+    /**
+     * <pre>
+     * Tensors contained within objects being serialized.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorProto tensors = 3;</code>
+     */
+    public org.tensorflow.proto.framework.TensorProto getTensors(int index) {
+      if (tensorsBuilder_ == null) {
+        return tensors_.get(index);
+      } else {
+        return tensorsBuilder_.getMessage(index);
+      }
+    }
+    /**
+     * <pre>
+     * Tensors contained within objects being serialized.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorProto tensors = 3;</code>
+     */
+    public Builder setTensors(
+        int index, org.tensorflow.proto.framework.TensorProto value) {
+      if (tensorsBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureTensorsIsMutable();
+        tensors_.set(index, value);
+        onChanged();
+      } else {
+        tensorsBuilder_.setMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Tensors contained within objects being serialized.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorProto tensors = 3;</code>
+     */
+    public Builder setTensors(
+        int index, org.tensorflow.proto.framework.TensorProto.Builder builderForValue) {
+      if (tensorsBuilder_ == null) {
+        ensureTensorsIsMutable();
+        tensors_.set(index, builderForValue.build());
+        onChanged();
+      } else {
+        tensorsBuilder_.setMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Tensors contained within objects being serialized.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorProto tensors = 3;</code>
+     */
+    public Builder addTensors(org.tensorflow.proto.framework.TensorProto value) {
+      if (tensorsBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureTensorsIsMutable();
+        tensors_.add(value);
+        onChanged();
+      } else {
+        tensorsBuilder_.addMessage(value);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Tensors contained within objects being serialized.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorProto tensors = 3;</code>
+     */
+    public Builder addTensors(
+        int index, org.tensorflow.proto.framework.TensorProto value) {
+      if (tensorsBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureTensorsIsMutable();
+        tensors_.add(index, value);
+        onChanged();
+      } else {
+        tensorsBuilder_.addMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Tensors contained within objects being serialized.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorProto tensors = 3;</code>
+     */
+    public Builder addTensors(
+        org.tensorflow.proto.framework.TensorProto.Builder builderForValue) {
+      if (tensorsBuilder_ == null) {
+        ensureTensorsIsMutable();
+        tensors_.add(builderForValue.build());
+        onChanged();
+      } else {
+        tensorsBuilder_.addMessage(builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Tensors contained within objects being serialized.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorProto tensors = 3;</code>
+     */
+    public Builder addTensors(
+        int index, org.tensorflow.proto.framework.TensorProto.Builder builderForValue) {
+      if (tensorsBuilder_ == null) {
+        ensureTensorsIsMutable();
+        tensors_.add(index, builderForValue.build());
+        onChanged();
+      } else {
+        tensorsBuilder_.addMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Tensors contained within objects being serialized.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorProto tensors = 3;</code>
+     */
+    public Builder addAllTensors(
+        java.lang.Iterable<? extends org.tensorflow.proto.framework.TensorProto> values) {
+      if (tensorsBuilder_ == null) {
+        ensureTensorsIsMutable();
+        com.google.protobuf.AbstractMessageLite.Builder.addAll(
+            values, tensors_);
+        onChanged();
+      } else {
+        tensorsBuilder_.addAllMessages(values);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Tensors contained within objects being serialized.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorProto tensors = 3;</code>
+     */
+    public Builder clearTensors() {
+      if (tensorsBuilder_ == null) {
+        tensors_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000001);
+        onChanged();
+      } else {
+        tensorsBuilder_.clear();
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Tensors contained within objects being serialized.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorProto tensors = 3;</code>
+     */
+    public Builder removeTensors(int index) {
+      if (tensorsBuilder_ == null) {
+        ensureTensorsIsMutable();
+        tensors_.remove(index);
+        onChanged();
+      } else {
+        tensorsBuilder_.remove(index);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Tensors contained within objects being serialized.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorProto tensors = 3;</code>
+     */
+    public org.tensorflow.proto.framework.TensorProto.Builder getTensorsBuilder(
+        int index) {
+      return getTensorsFieldBuilder().getBuilder(index);
+    }
+    /**
+     * <pre>
+     * Tensors contained within objects being serialized.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorProto tensors = 3;</code>
+     */
+    public org.tensorflow.proto.framework.TensorProtoOrBuilder getTensorsOrBuilder(
+        int index) {
+      if (tensorsBuilder_ == null) {
+        return tensors_.get(index);  } else {
+        return tensorsBuilder_.getMessageOrBuilder(index);
+      }
+    }
+    /**
+     * <pre>
+     * Tensors contained within objects being serialized.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorProto tensors = 3;</code>
+     */
+    public java.util.List<? extends org.tensorflow.proto.framework.TensorProtoOrBuilder> 
+         getTensorsOrBuilderList() {
+      if (tensorsBuilder_ != null) {
+        return tensorsBuilder_.getMessageOrBuilderList();
+      } else {
+        return java.util.Collections.unmodifiableList(tensors_);
+      }
+    }
+    /**
+     * <pre>
+     * Tensors contained within objects being serialized.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorProto tensors = 3;</code>
+     */
+    public org.tensorflow.proto.framework.TensorProto.Builder addTensorsBuilder() {
+      return getTensorsFieldBuilder().addBuilder(
+          org.tensorflow.proto.framework.TensorProto.getDefaultInstance());
+    }
+    /**
+     * <pre>
+     * Tensors contained within objects being serialized.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorProto tensors = 3;</code>
+     */
+    public org.tensorflow.proto.framework.TensorProto.Builder addTensorsBuilder(
+        int index) {
+      return getTensorsFieldBuilder().addBuilder(
+          index, org.tensorflow.proto.framework.TensorProto.getDefaultInstance());
+    }
+    /**
+     * <pre>
+     * Tensors contained within objects being serialized.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorProto tensors = 3;</code>
+     */
+    public java.util.List<org.tensorflow.proto.framework.TensorProto.Builder> 
+         getTensorsBuilderList() {
+      return getTensorsFieldBuilder().getBuilderList();
+    }
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.framework.TensorProto, org.tensorflow.proto.framework.TensorProto.Builder, org.tensorflow.proto.framework.TensorProtoOrBuilder> 
+        getTensorsFieldBuilder() {
+      if (tensorsBuilder_ == null) {
+        tensorsBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
+            org.tensorflow.proto.framework.TensorProto, org.tensorflow.proto.framework.TensorProto.Builder, org.tensorflow.proto.framework.TensorProtoOrBuilder>(
+                tensors_,
+                ((bitField0_ & 0x00000001) != 0),
+                getParentForChildren(),
+                isClean());
+        tensors_ = null;
+      }
+      return tensorsBuilder_;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.VariantTensorDataProto)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.VariantTensorDataProto)
+  private static final org.tensorflow.proto.framework.VariantTensorDataProto DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.VariantTensorDataProto();
+  }
+
+  public static org.tensorflow.proto.framework.VariantTensorDataProto getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<VariantTensorDataProto>
+      PARSER = new com.google.protobuf.AbstractParser<VariantTensorDataProto>() {
+    @java.lang.Override
+    public VariantTensorDataProto parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new VariantTensorDataProto(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<VariantTensorDataProto> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<VariantTensorDataProto> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.framework.VariantTensorDataProto getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/VariantTensorDataProtoOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/VariantTensorDataProtoOrBuilder.java
new file mode 100644
index 00000000000..916d583c328
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/VariantTensorDataProtoOrBuilder.java
@@ -0,0 +1,80 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/framework/tensor.proto
+
+package org.tensorflow.proto.framework;
+
+public interface VariantTensorDataProtoOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.VariantTensorDataProto)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <pre>
+   * Name of the type of objects being serialized.
+   * </pre>
+   *
+   * <code>string type_name = 1;</code>
+   */
+  java.lang.String getTypeName();
+  /**
+   * <pre>
+   * Name of the type of objects being serialized.
+   * </pre>
+   *
+   * <code>string type_name = 1;</code>
+   */
+  com.google.protobuf.ByteString
+      getTypeNameBytes();
+
+  /**
+   * <pre>
+   * Portions of the object that are not Tensors.
+   * </pre>
+   *
+   * <code>bytes metadata = 2;</code>
+   */
+  com.google.protobuf.ByteString getMetadata();
+
+  /**
+   * <pre>
+   * Tensors contained within objects being serialized.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.TensorProto tensors = 3;</code>
+   */
+  java.util.List<org.tensorflow.proto.framework.TensorProto> 
+      getTensorsList();
+  /**
+   * <pre>
+   * Tensors contained within objects being serialized.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.TensorProto tensors = 3;</code>
+   */
+  org.tensorflow.proto.framework.TensorProto getTensors(int index);
+  /**
+   * <pre>
+   * Tensors contained within objects being serialized.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.TensorProto tensors = 3;</code>
+   */
+  int getTensorsCount();
+  /**
+   * <pre>
+   * Tensors contained within objects being serialized.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.TensorProto tensors = 3;</code>
+   */
+  java.util.List<? extends org.tensorflow.proto.framework.TensorProtoOrBuilder> 
+      getTensorsOrBuilderList();
+  /**
+   * <pre>
+   * Tensors contained within objects being serialized.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.TensorProto tensors = 3;</code>
+   */
+  org.tensorflow.proto.framework.TensorProtoOrBuilder getTensorsOrBuilder(
+      int index);
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/VerifierConfig.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/VerifierConfig.java
new file mode 100644
index 00000000000..c6099d9d2df
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/VerifierConfig.java
@@ -0,0 +1,725 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/verifier_config.proto
+
+package org.tensorflow.proto.framework;
+
+/**
+ * <pre>
+ * The config for graph verifiers.
+ * </pre>
+ *
+ * Protobuf type {@code tensorflow.VerifierConfig}
+ */
+public  final class VerifierConfig extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.VerifierConfig)
+    VerifierConfigOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use VerifierConfig.newBuilder() to construct.
+  private VerifierConfig(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private VerifierConfig() {
+    structureVerifier_ = 0;
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new VerifierConfig();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private VerifierConfig(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 8: {
+
+            verificationTimeoutInMs_ = input.readInt64();
+            break;
+          }
+          case 16: {
+            int rawValue = input.readEnum();
+
+            structureVerifier_ = rawValue;
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.framework.VerifierConfigProtos.internal_static_tensorflow_VerifierConfig_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.framework.VerifierConfigProtos.internal_static_tensorflow_VerifierConfig_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.framework.VerifierConfig.class, org.tensorflow.proto.framework.VerifierConfig.Builder.class);
+  }
+
+  /**
+   * Protobuf enum {@code tensorflow.VerifierConfig.Toggle}
+   */
+  public enum Toggle
+      implements com.google.protobuf.ProtocolMessageEnum {
+    /**
+     * <code>DEFAULT = 0;</code>
+     */
+    DEFAULT(0),
+    /**
+     * <code>ON = 1;</code>
+     */
+    ON(1),
+    /**
+     * <code>OFF = 2;</code>
+     */
+    OFF(2),
+    UNRECOGNIZED(-1),
+    ;
+
+    /**
+     * <code>DEFAULT = 0;</code>
+     */
+    public static final int DEFAULT_VALUE = 0;
+    /**
+     * <code>ON = 1;</code>
+     */
+    public static final int ON_VALUE = 1;
+    /**
+     * <code>OFF = 2;</code>
+     */
+    public static final int OFF_VALUE = 2;
+
+
+    public final int getNumber() {
+      if (this == UNRECOGNIZED) {
+        throw new java.lang.IllegalArgumentException(
+            "Can't get the number of an unknown enum value.");
+      }
+      return value;
+    }
+
+    /**
+     * @deprecated Use {@link #forNumber(int)} instead.
+     */
+    @java.lang.Deprecated
+    public static Toggle valueOf(int value) {
+      return forNumber(value);
+    }
+
+    public static Toggle forNumber(int value) {
+      switch (value) {
+        case 0: return DEFAULT;
+        case 1: return ON;
+        case 2: return OFF;
+        default: return null;
+      }
+    }
+
+    public static com.google.protobuf.Internal.EnumLiteMap<Toggle>
+        internalGetValueMap() {
+      return internalValueMap;
+    }
+    private static final com.google.protobuf.Internal.EnumLiteMap<
+        Toggle> internalValueMap =
+          new com.google.protobuf.Internal.EnumLiteMap<Toggle>() {
+            public Toggle findValueByNumber(int number) {
+              return Toggle.forNumber(number);
+            }
+          };
+
+    public final com.google.protobuf.Descriptors.EnumValueDescriptor
+        getValueDescriptor() {
+      return getDescriptor().getValues().get(ordinal());
+    }
+    public final com.google.protobuf.Descriptors.EnumDescriptor
+        getDescriptorForType() {
+      return getDescriptor();
+    }
+    public static final com.google.protobuf.Descriptors.EnumDescriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.VerifierConfig.getDescriptor().getEnumTypes().get(0);
+    }
+
+    private static final Toggle[] VALUES = values();
+
+    public static Toggle valueOf(
+        com.google.protobuf.Descriptors.EnumValueDescriptor desc) {
+      if (desc.getType() != getDescriptor()) {
+        throw new java.lang.IllegalArgumentException(
+          "EnumValueDescriptor is not for this type.");
+      }
+      if (desc.getIndex() == -1) {
+        return UNRECOGNIZED;
+      }
+      return VALUES[desc.getIndex()];
+    }
+
+    private final int value;
+
+    private Toggle(int value) {
+      this.value = value;
+    }
+
+    // @@protoc_insertion_point(enum_scope:tensorflow.VerifierConfig.Toggle)
+  }
+
+  public static final int VERIFICATION_TIMEOUT_IN_MS_FIELD_NUMBER = 1;
+  private long verificationTimeoutInMs_;
+  /**
+   * <pre>
+   * Deadline for completion of all verification i.e. all the Toggle ON
+   * verifiers must complete execution within this time.
+   * </pre>
+   *
+   * <code>int64 verification_timeout_in_ms = 1;</code>
+   */
+  public long getVerificationTimeoutInMs() {
+    return verificationTimeoutInMs_;
+  }
+
+  public static final int STRUCTURE_VERIFIER_FIELD_NUMBER = 2;
+  private int structureVerifier_;
+  /**
+   * <pre>
+   * Perform structural validation on a tensorflow graph. Default is OFF.
+   * </pre>
+   *
+   * <code>.tensorflow.VerifierConfig.Toggle structure_verifier = 2;</code>
+   */
+  public int getStructureVerifierValue() {
+    return structureVerifier_;
+  }
+  /**
+   * <pre>
+   * Perform structural validation on a tensorflow graph. Default is OFF.
+   * </pre>
+   *
+   * <code>.tensorflow.VerifierConfig.Toggle structure_verifier = 2;</code>
+   */
+  public org.tensorflow.proto.framework.VerifierConfig.Toggle getStructureVerifier() {
+    @SuppressWarnings("deprecation")
+    org.tensorflow.proto.framework.VerifierConfig.Toggle result = org.tensorflow.proto.framework.VerifierConfig.Toggle.valueOf(structureVerifier_);
+    return result == null ? org.tensorflow.proto.framework.VerifierConfig.Toggle.UNRECOGNIZED : result;
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    if (verificationTimeoutInMs_ != 0L) {
+      output.writeInt64(1, verificationTimeoutInMs_);
+    }
+    if (structureVerifier_ != org.tensorflow.proto.framework.VerifierConfig.Toggle.DEFAULT.getNumber()) {
+      output.writeEnum(2, structureVerifier_);
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (verificationTimeoutInMs_ != 0L) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt64Size(1, verificationTimeoutInMs_);
+    }
+    if (structureVerifier_ != org.tensorflow.proto.framework.VerifierConfig.Toggle.DEFAULT.getNumber()) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeEnumSize(2, structureVerifier_);
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.framework.VerifierConfig)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.framework.VerifierConfig other = (org.tensorflow.proto.framework.VerifierConfig) obj;
+
+    if (getVerificationTimeoutInMs()
+        != other.getVerificationTimeoutInMs()) return false;
+    if (structureVerifier_ != other.structureVerifier_) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    hash = (37 * hash) + VERIFICATION_TIMEOUT_IN_MS_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+        getVerificationTimeoutInMs());
+    hash = (37 * hash) + STRUCTURE_VERIFIER_FIELD_NUMBER;
+    hash = (53 * hash) + structureVerifier_;
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.framework.VerifierConfig parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.VerifierConfig parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.VerifierConfig parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.VerifierConfig parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.VerifierConfig parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.VerifierConfig parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.VerifierConfig parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.VerifierConfig parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.VerifierConfig parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.VerifierConfig parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.VerifierConfig parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.VerifierConfig parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.framework.VerifierConfig prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * <pre>
+   * The config for graph verifiers.
+   * </pre>
+   *
+   * Protobuf type {@code tensorflow.VerifierConfig}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.VerifierConfig)
+      org.tensorflow.proto.framework.VerifierConfigOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.VerifierConfigProtos.internal_static_tensorflow_VerifierConfig_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.VerifierConfigProtos.internal_static_tensorflow_VerifierConfig_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.VerifierConfig.class, org.tensorflow.proto.framework.VerifierConfig.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.framework.VerifierConfig.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      verificationTimeoutInMs_ = 0L;
+
+      structureVerifier_ = 0;
+
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.framework.VerifierConfigProtos.internal_static_tensorflow_VerifierConfig_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.VerifierConfig getDefaultInstanceForType() {
+      return org.tensorflow.proto.framework.VerifierConfig.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.VerifierConfig build() {
+      org.tensorflow.proto.framework.VerifierConfig result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.VerifierConfig buildPartial() {
+      org.tensorflow.proto.framework.VerifierConfig result = new org.tensorflow.proto.framework.VerifierConfig(this);
+      result.verificationTimeoutInMs_ = verificationTimeoutInMs_;
+      result.structureVerifier_ = structureVerifier_;
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.framework.VerifierConfig) {
+        return mergeFrom((org.tensorflow.proto.framework.VerifierConfig)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.framework.VerifierConfig other) {
+      if (other == org.tensorflow.proto.framework.VerifierConfig.getDefaultInstance()) return this;
+      if (other.getVerificationTimeoutInMs() != 0L) {
+        setVerificationTimeoutInMs(other.getVerificationTimeoutInMs());
+      }
+      if (other.structureVerifier_ != 0) {
+        setStructureVerifierValue(other.getStructureVerifierValue());
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.framework.VerifierConfig parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.framework.VerifierConfig) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+
+    private long verificationTimeoutInMs_ ;
+    /**
+     * <pre>
+     * Deadline for completion of all verification i.e. all the Toggle ON
+     * verifiers must complete execution within this time.
+     * </pre>
+     *
+     * <code>int64 verification_timeout_in_ms = 1;</code>
+     */
+    public long getVerificationTimeoutInMs() {
+      return verificationTimeoutInMs_;
+    }
+    /**
+     * <pre>
+     * Deadline for completion of all verification i.e. all the Toggle ON
+     * verifiers must complete execution within this time.
+     * </pre>
+     *
+     * <code>int64 verification_timeout_in_ms = 1;</code>
+     */
+    public Builder setVerificationTimeoutInMs(long value) {
+      
+      verificationTimeoutInMs_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Deadline for completion of all verification i.e. all the Toggle ON
+     * verifiers must complete execution within this time.
+     * </pre>
+     *
+     * <code>int64 verification_timeout_in_ms = 1;</code>
+     */
+    public Builder clearVerificationTimeoutInMs() {
+      
+      verificationTimeoutInMs_ = 0L;
+      onChanged();
+      return this;
+    }
+
+    private int structureVerifier_ = 0;
+    /**
+     * <pre>
+     * Perform structural validation on a tensorflow graph. Default is OFF.
+     * </pre>
+     *
+     * <code>.tensorflow.VerifierConfig.Toggle structure_verifier = 2;</code>
+     */
+    public int getStructureVerifierValue() {
+      return structureVerifier_;
+    }
+    /**
+     * <pre>
+     * Perform structural validation on a tensorflow graph. Default is OFF.
+     * </pre>
+     *
+     * <code>.tensorflow.VerifierConfig.Toggle structure_verifier = 2;</code>
+     */
+    public Builder setStructureVerifierValue(int value) {
+      structureVerifier_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Perform structural validation on a tensorflow graph. Default is OFF.
+     * </pre>
+     *
+     * <code>.tensorflow.VerifierConfig.Toggle structure_verifier = 2;</code>
+     */
+    public org.tensorflow.proto.framework.VerifierConfig.Toggle getStructureVerifier() {
+      @SuppressWarnings("deprecation")
+      org.tensorflow.proto.framework.VerifierConfig.Toggle result = org.tensorflow.proto.framework.VerifierConfig.Toggle.valueOf(structureVerifier_);
+      return result == null ? org.tensorflow.proto.framework.VerifierConfig.Toggle.UNRECOGNIZED : result;
+    }
+    /**
+     * <pre>
+     * Perform structural validation on a tensorflow graph. Default is OFF.
+     * </pre>
+     *
+     * <code>.tensorflow.VerifierConfig.Toggle structure_verifier = 2;</code>
+     */
+    public Builder setStructureVerifier(org.tensorflow.proto.framework.VerifierConfig.Toggle value) {
+      if (value == null) {
+        throw new NullPointerException();
+      }
+      
+      structureVerifier_ = value.getNumber();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Perform structural validation on a tensorflow graph. Default is OFF.
+     * </pre>
+     *
+     * <code>.tensorflow.VerifierConfig.Toggle structure_verifier = 2;</code>
+     */
+    public Builder clearStructureVerifier() {
+      
+      structureVerifier_ = 0;
+      onChanged();
+      return this;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.VerifierConfig)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.VerifierConfig)
+  private static final org.tensorflow.proto.framework.VerifierConfig DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.VerifierConfig();
+  }
+
+  public static org.tensorflow.proto.framework.VerifierConfig getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<VerifierConfig>
+      PARSER = new com.google.protobuf.AbstractParser<VerifierConfig>() {
+    @java.lang.Override
+    public VerifierConfig parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new VerifierConfig(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<VerifierConfig> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<VerifierConfig> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.framework.VerifierConfig getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/VerifierConfigOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/VerifierConfigOrBuilder.java
new file mode 100644
index 00000000000..230a728e05f
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/VerifierConfigOrBuilder.java
@@ -0,0 +1,36 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/verifier_config.proto
+
+package org.tensorflow.proto.framework;
+
+public interface VerifierConfigOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.VerifierConfig)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <pre>
+   * Deadline for completion of all verification i.e. all the Toggle ON
+   * verifiers must complete execution within this time.
+   * </pre>
+   *
+   * <code>int64 verification_timeout_in_ms = 1;</code>
+   */
+  long getVerificationTimeoutInMs();
+
+  /**
+   * <pre>
+   * Perform structural validation on a tensorflow graph. Default is OFF.
+   * </pre>
+   *
+   * <code>.tensorflow.VerifierConfig.Toggle structure_verifier = 2;</code>
+   */
+  int getStructureVerifierValue();
+  /**
+   * <pre>
+   * Perform structural validation on a tensorflow graph. Default is OFF.
+   * </pre>
+   *
+   * <code>.tensorflow.VerifierConfig.Toggle structure_verifier = 2;</code>
+   */
+  org.tensorflow.proto.framework.VerifierConfig.Toggle getStructureVerifier();
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/VerifierConfigProtos.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/VerifierConfigProtos.java
new file mode 100644
index 00000000000..45b39000486
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/VerifierConfigProtos.java
@@ -0,0 +1,55 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/verifier_config.proto
+
+package org.tensorflow.proto.framework;
+
+public final class VerifierConfigProtos {
+  private VerifierConfigProtos() {}
+  public static void registerAllExtensions(
+      com.google.protobuf.ExtensionRegistryLite registry) {
+  }
+
+  public static void registerAllExtensions(
+      com.google.protobuf.ExtensionRegistry registry) {
+    registerAllExtensions(
+        (com.google.protobuf.ExtensionRegistryLite) registry);
+  }
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_VerifierConfig_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_VerifierConfig_fieldAccessorTable;
+
+  public static com.google.protobuf.Descriptors.FileDescriptor
+      getDescriptor() {
+    return descriptor;
+  }
+  private static  com.google.protobuf.Descriptors.FileDescriptor
+      descriptor;
+  static {
+    java.lang.String[] descriptorData = {
+      "\n.tensorflow/core/protobuf/verifier_conf" +
+      "ig.proto\022\ntensorflow\"\233\001\n\016VerifierConfig\022" +
+      "\"\n\032verification_timeout_in_ms\030\001 \001(\003\022=\n\022s" +
+      "tructure_verifier\030\002 \001(\0162!.tensorflow.Ver" +
+      "ifierConfig.Toggle\"&\n\006Toggle\022\013\n\007DEFAULT\020" +
+      "\000\022\006\n\002ON\020\001\022\007\n\003OFF\020\002B\205\001\n\036org.tensorflow.pr" +
+      "oto.frameworkB\024VerifierConfigProtosP\001ZHg" +
+      "ithub.com/tensorflow/tensorflow/tensorfl" +
+      "ow/go/core/core_protos_go_proto\370\001\001b\006prot" +
+      "o3"
+    };
+    descriptor = com.google.protobuf.Descriptors.FileDescriptor
+      .internalBuildGeneratedFileFrom(descriptorData,
+        new com.google.protobuf.Descriptors.FileDescriptor[] {
+        });
+    internal_static_tensorflow_VerifierConfig_descriptor =
+      getDescriptor().getMessageTypes().get(0);
+    internal_static_tensorflow_VerifierConfig_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_VerifierConfig_descriptor,
+        new java.lang.String[] { "VerificationTimeoutInMs", "StructureVerifier", });
+  }
+
+  // @@protoc_insertion_point(outer_class_scope)
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/VersionDef.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/VersionDef.java
new file mode 100644
index 00000000000..6e83b2aa792
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/VersionDef.java
@@ -0,0 +1,792 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/framework/versions.proto
+
+package org.tensorflow.proto.framework;
+
+/**
+ * <pre>
+ * Version information for a piece of serialized data
+ * There are different types of versions for each type of data
+ * (GraphDef, etc.), but they all have the same common shape
+ * described here.
+ * Each consumer has "consumer" and "min_producer" versions (specified
+ * elsewhere).  A consumer is allowed to consume this data if
+ *   producer &gt;= min_producer
+ *   consumer &gt;= min_consumer
+ *   consumer not in bad_consumers
+ * </pre>
+ *
+ * Protobuf type {@code tensorflow.VersionDef}
+ */
+public  final class VersionDef extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.VersionDef)
+    VersionDefOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use VersionDef.newBuilder() to construct.
+  private VersionDef(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private VersionDef() {
+    badConsumers_ = emptyIntList();
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new VersionDef();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private VersionDef(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    int mutable_bitField0_ = 0;
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 8: {
+
+            producer_ = input.readInt32();
+            break;
+          }
+          case 16: {
+
+            minConsumer_ = input.readInt32();
+            break;
+          }
+          case 24: {
+            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
+              badConsumers_ = newIntList();
+              mutable_bitField0_ |= 0x00000001;
+            }
+            badConsumers_.addInt(input.readInt32());
+            break;
+          }
+          case 26: {
+            int length = input.readRawVarint32();
+            int limit = input.pushLimit(length);
+            if (!((mutable_bitField0_ & 0x00000001) != 0) && input.getBytesUntilLimit() > 0) {
+              badConsumers_ = newIntList();
+              mutable_bitField0_ |= 0x00000001;
+            }
+            while (input.getBytesUntilLimit() > 0) {
+              badConsumers_.addInt(input.readInt32());
+            }
+            input.popLimit(limit);
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      if (((mutable_bitField0_ & 0x00000001) != 0)) {
+        badConsumers_.makeImmutable(); // C
+      }
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.framework.VersionsProtos.internal_static_tensorflow_VersionDef_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.framework.VersionsProtos.internal_static_tensorflow_VersionDef_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.framework.VersionDef.class, org.tensorflow.proto.framework.VersionDef.Builder.class);
+  }
+
+  public static final int PRODUCER_FIELD_NUMBER = 1;
+  private int producer_;
+  /**
+   * <pre>
+   * The version of the code that produced this data.
+   * </pre>
+   *
+   * <code>int32 producer = 1;</code>
+   */
+  public int getProducer() {
+    return producer_;
+  }
+
+  public static final int MIN_CONSUMER_FIELD_NUMBER = 2;
+  private int minConsumer_;
+  /**
+   * <pre>
+   * Any consumer below this version is not allowed to consume this data.
+   * </pre>
+   *
+   * <code>int32 min_consumer = 2;</code>
+   */
+  public int getMinConsumer() {
+    return minConsumer_;
+  }
+
+  public static final int BAD_CONSUMERS_FIELD_NUMBER = 3;
+  private com.google.protobuf.Internal.IntList badConsumers_;
+  /**
+   * <pre>
+   * Specific consumer versions which are disallowed (e.g. due to bugs).
+   * </pre>
+   *
+   * <code>repeated int32 bad_consumers = 3;</code>
+   */
+  public java.util.List<java.lang.Integer>
+      getBadConsumersList() {
+    return badConsumers_;
+  }
+  /**
+   * <pre>
+   * Specific consumer versions which are disallowed (e.g. due to bugs).
+   * </pre>
+   *
+   * <code>repeated int32 bad_consumers = 3;</code>
+   */
+  public int getBadConsumersCount() {
+    return badConsumers_.size();
+  }
+  /**
+   * <pre>
+   * Specific consumer versions which are disallowed (e.g. due to bugs).
+   * </pre>
+   *
+   * <code>repeated int32 bad_consumers = 3;</code>
+   */
+  public int getBadConsumers(int index) {
+    return badConsumers_.getInt(index);
+  }
+  private int badConsumersMemoizedSerializedSize = -1;
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    getSerializedSize();
+    if (producer_ != 0) {
+      output.writeInt32(1, producer_);
+    }
+    if (minConsumer_ != 0) {
+      output.writeInt32(2, minConsumer_);
+    }
+    if (getBadConsumersList().size() > 0) {
+      output.writeUInt32NoTag(26);
+      output.writeUInt32NoTag(badConsumersMemoizedSerializedSize);
+    }
+    for (int i = 0; i < badConsumers_.size(); i++) {
+      output.writeInt32NoTag(badConsumers_.getInt(i));
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (producer_ != 0) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt32Size(1, producer_);
+    }
+    if (minConsumer_ != 0) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt32Size(2, minConsumer_);
+    }
+    {
+      int dataSize = 0;
+      for (int i = 0; i < badConsumers_.size(); i++) {
+        dataSize += com.google.protobuf.CodedOutputStream
+          .computeInt32SizeNoTag(badConsumers_.getInt(i));
+      }
+      size += dataSize;
+      if (!getBadConsumersList().isEmpty()) {
+        size += 1;
+        size += com.google.protobuf.CodedOutputStream
+            .computeInt32SizeNoTag(dataSize);
+      }
+      badConsumersMemoizedSerializedSize = dataSize;
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.framework.VersionDef)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.framework.VersionDef other = (org.tensorflow.proto.framework.VersionDef) obj;
+
+    if (getProducer()
+        != other.getProducer()) return false;
+    if (getMinConsumer()
+        != other.getMinConsumer()) return false;
+    if (!getBadConsumersList()
+        .equals(other.getBadConsumersList())) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    hash = (37 * hash) + PRODUCER_FIELD_NUMBER;
+    hash = (53 * hash) + getProducer();
+    hash = (37 * hash) + MIN_CONSUMER_FIELD_NUMBER;
+    hash = (53 * hash) + getMinConsumer();
+    if (getBadConsumersCount() > 0) {
+      hash = (37 * hash) + BAD_CONSUMERS_FIELD_NUMBER;
+      hash = (53 * hash) + getBadConsumersList().hashCode();
+    }
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.framework.VersionDef parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.VersionDef parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.VersionDef parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.VersionDef parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.VersionDef parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.VersionDef parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.VersionDef parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.VersionDef parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.VersionDef parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.VersionDef parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.VersionDef parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.VersionDef parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.framework.VersionDef prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * <pre>
+   * Version information for a piece of serialized data
+   * There are different types of versions for each type of data
+   * (GraphDef, etc.), but they all have the same common shape
+   * described here.
+   * Each consumer has "consumer" and "min_producer" versions (specified
+   * elsewhere).  A consumer is allowed to consume this data if
+   *   producer &gt;= min_producer
+   *   consumer &gt;= min_consumer
+   *   consumer not in bad_consumers
+   * </pre>
+   *
+   * Protobuf type {@code tensorflow.VersionDef}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.VersionDef)
+      org.tensorflow.proto.framework.VersionDefOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.VersionsProtos.internal_static_tensorflow_VersionDef_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.VersionsProtos.internal_static_tensorflow_VersionDef_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.VersionDef.class, org.tensorflow.proto.framework.VersionDef.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.framework.VersionDef.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      producer_ = 0;
+
+      minConsumer_ = 0;
+
+      badConsumers_ = emptyIntList();
+      bitField0_ = (bitField0_ & ~0x00000001);
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.framework.VersionsProtos.internal_static_tensorflow_VersionDef_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.VersionDef getDefaultInstanceForType() {
+      return org.tensorflow.proto.framework.VersionDef.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.VersionDef build() {
+      org.tensorflow.proto.framework.VersionDef result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.VersionDef buildPartial() {
+      org.tensorflow.proto.framework.VersionDef result = new org.tensorflow.proto.framework.VersionDef(this);
+      int from_bitField0_ = bitField0_;
+      result.producer_ = producer_;
+      result.minConsumer_ = minConsumer_;
+      if (((bitField0_ & 0x00000001) != 0)) {
+        badConsumers_.makeImmutable();
+        bitField0_ = (bitField0_ & ~0x00000001);
+      }
+      result.badConsumers_ = badConsumers_;
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.framework.VersionDef) {
+        return mergeFrom((org.tensorflow.proto.framework.VersionDef)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.framework.VersionDef other) {
+      if (other == org.tensorflow.proto.framework.VersionDef.getDefaultInstance()) return this;
+      if (other.getProducer() != 0) {
+        setProducer(other.getProducer());
+      }
+      if (other.getMinConsumer() != 0) {
+        setMinConsumer(other.getMinConsumer());
+      }
+      if (!other.badConsumers_.isEmpty()) {
+        if (badConsumers_.isEmpty()) {
+          badConsumers_ = other.badConsumers_;
+          bitField0_ = (bitField0_ & ~0x00000001);
+        } else {
+          ensureBadConsumersIsMutable();
+          badConsumers_.addAll(other.badConsumers_);
+        }
+        onChanged();
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.framework.VersionDef parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.framework.VersionDef) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+    private int bitField0_;
+
+    private int producer_ ;
+    /**
+     * <pre>
+     * The version of the code that produced this data.
+     * </pre>
+     *
+     * <code>int32 producer = 1;</code>
+     */
+    public int getProducer() {
+      return producer_;
+    }
+    /**
+     * <pre>
+     * The version of the code that produced this data.
+     * </pre>
+     *
+     * <code>int32 producer = 1;</code>
+     */
+    public Builder setProducer(int value) {
+      
+      producer_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * The version of the code that produced this data.
+     * </pre>
+     *
+     * <code>int32 producer = 1;</code>
+     */
+    public Builder clearProducer() {
+      
+      producer_ = 0;
+      onChanged();
+      return this;
+    }
+
+    private int minConsumer_ ;
+    /**
+     * <pre>
+     * Any consumer below this version is not allowed to consume this data.
+     * </pre>
+     *
+     * <code>int32 min_consumer = 2;</code>
+     */
+    public int getMinConsumer() {
+      return minConsumer_;
+    }
+    /**
+     * <pre>
+     * Any consumer below this version is not allowed to consume this data.
+     * </pre>
+     *
+     * <code>int32 min_consumer = 2;</code>
+     */
+    public Builder setMinConsumer(int value) {
+      
+      minConsumer_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Any consumer below this version is not allowed to consume this data.
+     * </pre>
+     *
+     * <code>int32 min_consumer = 2;</code>
+     */
+    public Builder clearMinConsumer() {
+      
+      minConsumer_ = 0;
+      onChanged();
+      return this;
+    }
+
+    private com.google.protobuf.Internal.IntList badConsumers_ = emptyIntList();
+    private void ensureBadConsumersIsMutable() {
+      if (!((bitField0_ & 0x00000001) != 0)) {
+        badConsumers_ = mutableCopy(badConsumers_);
+        bitField0_ |= 0x00000001;
+       }
+    }
+    /**
+     * <pre>
+     * Specific consumer versions which are disallowed (e.g. due to bugs).
+     * </pre>
+     *
+     * <code>repeated int32 bad_consumers = 3;</code>
+     */
+    public java.util.List<java.lang.Integer>
+        getBadConsumersList() {
+      return ((bitField0_ & 0x00000001) != 0) ?
+               java.util.Collections.unmodifiableList(badConsumers_) : badConsumers_;
+    }
+    /**
+     * <pre>
+     * Specific consumer versions which are disallowed (e.g. due to bugs).
+     * </pre>
+     *
+     * <code>repeated int32 bad_consumers = 3;</code>
+     */
+    public int getBadConsumersCount() {
+      return badConsumers_.size();
+    }
+    /**
+     * <pre>
+     * Specific consumer versions which are disallowed (e.g. due to bugs).
+     * </pre>
+     *
+     * <code>repeated int32 bad_consumers = 3;</code>
+     */
+    public int getBadConsumers(int index) {
+      return badConsumers_.getInt(index);
+    }
+    /**
+     * <pre>
+     * Specific consumer versions which are disallowed (e.g. due to bugs).
+     * </pre>
+     *
+     * <code>repeated int32 bad_consumers = 3;</code>
+     */
+    public Builder setBadConsumers(
+        int index, int value) {
+      ensureBadConsumersIsMutable();
+      badConsumers_.setInt(index, value);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Specific consumer versions which are disallowed (e.g. due to bugs).
+     * </pre>
+     *
+     * <code>repeated int32 bad_consumers = 3;</code>
+     */
+    public Builder addBadConsumers(int value) {
+      ensureBadConsumersIsMutable();
+      badConsumers_.addInt(value);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Specific consumer versions which are disallowed (e.g. due to bugs).
+     * </pre>
+     *
+     * <code>repeated int32 bad_consumers = 3;</code>
+     */
+    public Builder addAllBadConsumers(
+        java.lang.Iterable<? extends java.lang.Integer> values) {
+      ensureBadConsumersIsMutable();
+      com.google.protobuf.AbstractMessageLite.Builder.addAll(
+          values, badConsumers_);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Specific consumer versions which are disallowed (e.g. due to bugs).
+     * </pre>
+     *
+     * <code>repeated int32 bad_consumers = 3;</code>
+     */
+    public Builder clearBadConsumers() {
+      badConsumers_ = emptyIntList();
+      bitField0_ = (bitField0_ & ~0x00000001);
+      onChanged();
+      return this;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.VersionDef)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.VersionDef)
+  private static final org.tensorflow.proto.framework.VersionDef DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.VersionDef();
+  }
+
+  public static org.tensorflow.proto.framework.VersionDef getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<VersionDef>
+      PARSER = new com.google.protobuf.AbstractParser<VersionDef>() {
+    @java.lang.Override
+    public VersionDef parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new VersionDef(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<VersionDef> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<VersionDef> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.framework.VersionDef getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/VersionDefOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/VersionDefOrBuilder.java
new file mode 100644
index 00000000000..519769551aa
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/VersionDefOrBuilder.java
@@ -0,0 +1,52 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/framework/versions.proto
+
+package org.tensorflow.proto.framework;
+
+public interface VersionDefOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.VersionDef)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <pre>
+   * The version of the code that produced this data.
+   * </pre>
+   *
+   * <code>int32 producer = 1;</code>
+   */
+  int getProducer();
+
+  /**
+   * <pre>
+   * Any consumer below this version is not allowed to consume this data.
+   * </pre>
+   *
+   * <code>int32 min_consumer = 2;</code>
+   */
+  int getMinConsumer();
+
+  /**
+   * <pre>
+   * Specific consumer versions which are disallowed (e.g. due to bugs).
+   * </pre>
+   *
+   * <code>repeated int32 bad_consumers = 3;</code>
+   */
+  java.util.List<java.lang.Integer> getBadConsumersList();
+  /**
+   * <pre>
+   * Specific consumer versions which are disallowed (e.g. due to bugs).
+   * </pre>
+   *
+   * <code>repeated int32 bad_consumers = 3;</code>
+   */
+  int getBadConsumersCount();
+  /**
+   * <pre>
+   * Specific consumer versions which are disallowed (e.g. due to bugs).
+   * </pre>
+   *
+   * <code>repeated int32 bad_consumers = 3;</code>
+   */
+  int getBadConsumers(int index);
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/VersionsProtos.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/VersionsProtos.java
new file mode 100644
index 00000000000..0597ff6e213
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/VersionsProtos.java
@@ -0,0 +1,52 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/framework/versions.proto
+
+package org.tensorflow.proto.framework;
+
+public final class VersionsProtos {
+  private VersionsProtos() {}
+  public static void registerAllExtensions(
+      com.google.protobuf.ExtensionRegistryLite registry) {
+  }
+
+  public static void registerAllExtensions(
+      com.google.protobuf.ExtensionRegistry registry) {
+    registerAllExtensions(
+        (com.google.protobuf.ExtensionRegistryLite) registry);
+  }
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_VersionDef_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_VersionDef_fieldAccessorTable;
+
+  public static com.google.protobuf.Descriptors.FileDescriptor
+      getDescriptor() {
+    return descriptor;
+  }
+  private static  com.google.protobuf.Descriptors.FileDescriptor
+      descriptor;
+  static {
+    java.lang.String[] descriptorData = {
+      "\n(tensorflow/core/framework/versions.pro" +
+      "to\022\ntensorflow\"K\n\nVersionDef\022\020\n\010producer" +
+      "\030\001 \001(\005\022\024\n\014min_consumer\030\002 \001(\005\022\025\n\rbad_cons" +
+      "umers\030\003 \003(\005B\206\001\n\036org.tensorflow.proto.fra" +
+      "meworkB\016VersionsProtosP\001ZOgithub.com/ten" +
+      "sorflow/tensorflow/tensorflow/go/core/fr" +
+      "amework/versions_go_proto\370\001\001b\006proto3"
+    };
+    descriptor = com.google.protobuf.Descriptors.FileDescriptor
+      .internalBuildGeneratedFileFrom(descriptorData,
+        new com.google.protobuf.Descriptors.FileDescriptor[] {
+        });
+    internal_static_tensorflow_VersionDef_descriptor =
+      getDescriptor().getMessageTypes().get(0);
+    internal_static_tensorflow_VersionDef_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_VersionDef_descriptor,
+        new java.lang.String[] { "Producer", "MinConsumer", "BadConsumers", });
+  }
+
+  // @@protoc_insertion_point(outer_class_scope)
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/WhileContextDef.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/WhileContextDef.java
new file mode 100644
index 00000000000..9bd67ab7bcd
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/WhileContextDef.java
@@ -0,0 +1,2534 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/control_flow.proto
+
+package org.tensorflow.proto.framework;
+
+/**
+ * <pre>
+ * Protocol buffer representing a WhileContext object.
+ * </pre>
+ *
+ * Protobuf type {@code tensorflow.WhileContextDef}
+ */
+public  final class WhileContextDef extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.WhileContextDef)
+    WhileContextDefOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use WhileContextDef.newBuilder() to construct.
+  private WhileContextDef(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private WhileContextDef() {
+    contextName_ = "";
+    pivotName_ = "";
+    pivotForPredName_ = "";
+    pivotForBodyName_ = "";
+    loopExitNames_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+    loopEnterNames_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+    maximumIterationsName_ = "";
+    nestedContexts_ = java.util.Collections.emptyList();
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new WhileContextDef();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private WhileContextDef(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    int mutable_bitField0_ = 0;
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 10: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            contextName_ = s;
+            break;
+          }
+          case 16: {
+
+            parallelIterations_ = input.readInt32();
+            break;
+          }
+          case 24: {
+
+            backProp_ = input.readBool();
+            break;
+          }
+          case 32: {
+
+            swapMemory_ = input.readBool();
+            break;
+          }
+          case 42: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            pivotName_ = s;
+            break;
+          }
+          case 50: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            pivotForPredName_ = s;
+            break;
+          }
+          case 58: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            pivotForBodyName_ = s;
+            break;
+          }
+          case 66: {
+            java.lang.String s = input.readStringRequireUtf8();
+            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
+              loopExitNames_ = new com.google.protobuf.LazyStringArrayList();
+              mutable_bitField0_ |= 0x00000001;
+            }
+            loopExitNames_.add(s);
+            break;
+          }
+          case 74: {
+            org.tensorflow.proto.framework.ValuesDef.Builder subBuilder = null;
+            if (valuesDef_ != null) {
+              subBuilder = valuesDef_.toBuilder();
+            }
+            valuesDef_ = input.readMessage(org.tensorflow.proto.framework.ValuesDef.parser(), extensionRegistry);
+            if (subBuilder != null) {
+              subBuilder.mergeFrom(valuesDef_);
+              valuesDef_ = subBuilder.buildPartial();
+            }
+
+            break;
+          }
+          case 82: {
+            java.lang.String s = input.readStringRequireUtf8();
+            if (!((mutable_bitField0_ & 0x00000002) != 0)) {
+              loopEnterNames_ = new com.google.protobuf.LazyStringArrayList();
+              mutable_bitField0_ |= 0x00000002;
+            }
+            loopEnterNames_.add(s);
+            break;
+          }
+          case 90: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            maximumIterationsName_ = s;
+            break;
+          }
+          case 98: {
+            if (!((mutable_bitField0_ & 0x00000004) != 0)) {
+              nestedContexts_ = new java.util.ArrayList<org.tensorflow.proto.framework.ControlFlowContextDef>();
+              mutable_bitField0_ |= 0x00000004;
+            }
+            nestedContexts_.add(
+                input.readMessage(org.tensorflow.proto.framework.ControlFlowContextDef.parser(), extensionRegistry));
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      if (((mutable_bitField0_ & 0x00000001) != 0)) {
+        loopExitNames_ = loopExitNames_.getUnmodifiableView();
+      }
+      if (((mutable_bitField0_ & 0x00000002) != 0)) {
+        loopEnterNames_ = loopEnterNames_.getUnmodifiableView();
+      }
+      if (((mutable_bitField0_ & 0x00000004) != 0)) {
+        nestedContexts_ = java.util.Collections.unmodifiableList(nestedContexts_);
+      }
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.framework.ControlFlowProtos.internal_static_tensorflow_WhileContextDef_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.framework.ControlFlowProtos.internal_static_tensorflow_WhileContextDef_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.framework.WhileContextDef.class, org.tensorflow.proto.framework.WhileContextDef.Builder.class);
+  }
+
+  public static final int CONTEXT_NAME_FIELD_NUMBER = 1;
+  private volatile java.lang.Object contextName_;
+  /**
+   * <pre>
+   * Name of the context.
+   * </pre>
+   *
+   * <code>string context_name = 1;</code>
+   */
+  public java.lang.String getContextName() {
+    java.lang.Object ref = contextName_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      contextName_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * Name of the context.
+   * </pre>
+   *
+   * <code>string context_name = 1;</code>
+   */
+  public com.google.protobuf.ByteString
+      getContextNameBytes() {
+    java.lang.Object ref = contextName_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      contextName_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int PARALLEL_ITERATIONS_FIELD_NUMBER = 2;
+  private int parallelIterations_;
+  /**
+   * <pre>
+   * The number of iterations allowed to run in parallel.
+   * </pre>
+   *
+   * <code>int32 parallel_iterations = 2;</code>
+   */
+  public int getParallelIterations() {
+    return parallelIterations_;
+  }
+
+  public static final int BACK_PROP_FIELD_NUMBER = 3;
+  private boolean backProp_;
+  /**
+   * <pre>
+   * Whether backprop is enabled for this while loop.
+   * </pre>
+   *
+   * <code>bool back_prop = 3;</code>
+   */
+  public boolean getBackProp() {
+    return backProp_;
+  }
+
+  public static final int SWAP_MEMORY_FIELD_NUMBER = 4;
+  private boolean swapMemory_;
+  /**
+   * <pre>
+   * Whether GPU-CPU memory swap is enabled for this loop.
+   * </pre>
+   *
+   * <code>bool swap_memory = 4;</code>
+   */
+  public boolean getSwapMemory() {
+    return swapMemory_;
+  }
+
+  public static final int PIVOT_NAME_FIELD_NUMBER = 5;
+  private volatile java.lang.Object pivotName_;
+  /**
+   * <pre>
+   * Name of the pivot tensor.
+   * </pre>
+   *
+   * <code>string pivot_name = 5;</code>
+   */
+  public java.lang.String getPivotName() {
+    java.lang.Object ref = pivotName_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      pivotName_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * Name of the pivot tensor.
+   * </pre>
+   *
+   * <code>string pivot_name = 5;</code>
+   */
+  public com.google.protobuf.ByteString
+      getPivotNameBytes() {
+    java.lang.Object ref = pivotName_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      pivotName_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int PIVOT_FOR_PRED_NAME_FIELD_NUMBER = 6;
+  private volatile java.lang.Object pivotForPredName_;
+  /**
+   * <pre>
+   * Name of the pivot_for_pred tensor.
+   * </pre>
+   *
+   * <code>string pivot_for_pred_name = 6;</code>
+   */
+  public java.lang.String getPivotForPredName() {
+    java.lang.Object ref = pivotForPredName_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      pivotForPredName_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * Name of the pivot_for_pred tensor.
+   * </pre>
+   *
+   * <code>string pivot_for_pred_name = 6;</code>
+   */
+  public com.google.protobuf.ByteString
+      getPivotForPredNameBytes() {
+    java.lang.Object ref = pivotForPredName_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      pivotForPredName_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int PIVOT_FOR_BODY_NAME_FIELD_NUMBER = 7;
+  private volatile java.lang.Object pivotForBodyName_;
+  /**
+   * <pre>
+   * Name of the pivot_for_body tensor.
+   * </pre>
+   *
+   * <code>string pivot_for_body_name = 7;</code>
+   */
+  public java.lang.String getPivotForBodyName() {
+    java.lang.Object ref = pivotForBodyName_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      pivotForBodyName_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * Name of the pivot_for_body tensor.
+   * </pre>
+   *
+   * <code>string pivot_for_body_name = 7;</code>
+   */
+  public com.google.protobuf.ByteString
+      getPivotForBodyNameBytes() {
+    java.lang.Object ref = pivotForBodyName_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      pivotForBodyName_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int LOOP_EXIT_NAMES_FIELD_NUMBER = 8;
+  private com.google.protobuf.LazyStringList loopExitNames_;
+  /**
+   * <pre>
+   * List of names for exit tensors.
+   * </pre>
+   *
+   * <code>repeated string loop_exit_names = 8;</code>
+   */
+  public com.google.protobuf.ProtocolStringList
+      getLoopExitNamesList() {
+    return loopExitNames_;
+  }
+  /**
+   * <pre>
+   * List of names for exit tensors.
+   * </pre>
+   *
+   * <code>repeated string loop_exit_names = 8;</code>
+   */
+  public int getLoopExitNamesCount() {
+    return loopExitNames_.size();
+  }
+  /**
+   * <pre>
+   * List of names for exit tensors.
+   * </pre>
+   *
+   * <code>repeated string loop_exit_names = 8;</code>
+   */
+  public java.lang.String getLoopExitNames(int index) {
+    return loopExitNames_.get(index);
+  }
+  /**
+   * <pre>
+   * List of names for exit tensors.
+   * </pre>
+   *
+   * <code>repeated string loop_exit_names = 8;</code>
+   */
+  public com.google.protobuf.ByteString
+      getLoopExitNamesBytes(int index) {
+    return loopExitNames_.getByteString(index);
+  }
+
+  public static final int LOOP_ENTER_NAMES_FIELD_NUMBER = 10;
+  private com.google.protobuf.LazyStringList loopEnterNames_;
+  /**
+   * <pre>
+   * List of names for enter tensors.
+   * </pre>
+   *
+   * <code>repeated string loop_enter_names = 10;</code>
+   */
+  public com.google.protobuf.ProtocolStringList
+      getLoopEnterNamesList() {
+    return loopEnterNames_;
+  }
+  /**
+   * <pre>
+   * List of names for enter tensors.
+   * </pre>
+   *
+   * <code>repeated string loop_enter_names = 10;</code>
+   */
+  public int getLoopEnterNamesCount() {
+    return loopEnterNames_.size();
+  }
+  /**
+   * <pre>
+   * List of names for enter tensors.
+   * </pre>
+   *
+   * <code>repeated string loop_enter_names = 10;</code>
+   */
+  public java.lang.String getLoopEnterNames(int index) {
+    return loopEnterNames_.get(index);
+  }
+  /**
+   * <pre>
+   * List of names for enter tensors.
+   * </pre>
+   *
+   * <code>repeated string loop_enter_names = 10;</code>
+   */
+  public com.google.protobuf.ByteString
+      getLoopEnterNamesBytes(int index) {
+    return loopEnterNames_.getByteString(index);
+  }
+
+  public static final int VALUES_DEF_FIELD_NUMBER = 9;
+  private org.tensorflow.proto.framework.ValuesDef valuesDef_;
+  /**
+   * <pre>
+   * Values and external values in control flow context.
+   * </pre>
+   *
+   * <code>.tensorflow.ValuesDef values_def = 9;</code>
+   */
+  public boolean hasValuesDef() {
+    return valuesDef_ != null;
+  }
+  /**
+   * <pre>
+   * Values and external values in control flow context.
+   * </pre>
+   *
+   * <code>.tensorflow.ValuesDef values_def = 9;</code>
+   */
+  public org.tensorflow.proto.framework.ValuesDef getValuesDef() {
+    return valuesDef_ == null ? org.tensorflow.proto.framework.ValuesDef.getDefaultInstance() : valuesDef_;
+  }
+  /**
+   * <pre>
+   * Values and external values in control flow context.
+   * </pre>
+   *
+   * <code>.tensorflow.ValuesDef values_def = 9;</code>
+   */
+  public org.tensorflow.proto.framework.ValuesDefOrBuilder getValuesDefOrBuilder() {
+    return getValuesDef();
+  }
+
+  public static final int MAXIMUM_ITERATIONS_NAME_FIELD_NUMBER = 11;
+  private volatile java.lang.Object maximumIterationsName_;
+  /**
+   * <pre>
+   * Optional name of the maximum_iterations tensor.
+   * </pre>
+   *
+   * <code>string maximum_iterations_name = 11;</code>
+   */
+  public java.lang.String getMaximumIterationsName() {
+    java.lang.Object ref = maximumIterationsName_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      maximumIterationsName_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * Optional name of the maximum_iterations tensor.
+   * </pre>
+   *
+   * <code>string maximum_iterations_name = 11;</code>
+   */
+  public com.google.protobuf.ByteString
+      getMaximumIterationsNameBytes() {
+    java.lang.Object ref = maximumIterationsName_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      maximumIterationsName_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int NESTED_CONTEXTS_FIELD_NUMBER = 12;
+  private java.util.List<org.tensorflow.proto.framework.ControlFlowContextDef> nestedContexts_;
+  /**
+   * <pre>
+   * Contexts contained inside this context (e.g. nested whiles).
+   * </pre>
+   *
+   * <code>repeated .tensorflow.ControlFlowContextDef nested_contexts = 12;</code>
+   */
+  public java.util.List<org.tensorflow.proto.framework.ControlFlowContextDef> getNestedContextsList() {
+    return nestedContexts_;
+  }
+  /**
+   * <pre>
+   * Contexts contained inside this context (e.g. nested whiles).
+   * </pre>
+   *
+   * <code>repeated .tensorflow.ControlFlowContextDef nested_contexts = 12;</code>
+   */
+  public java.util.List<? extends org.tensorflow.proto.framework.ControlFlowContextDefOrBuilder> 
+      getNestedContextsOrBuilderList() {
+    return nestedContexts_;
+  }
+  /**
+   * <pre>
+   * Contexts contained inside this context (e.g. nested whiles).
+   * </pre>
+   *
+   * <code>repeated .tensorflow.ControlFlowContextDef nested_contexts = 12;</code>
+   */
+  public int getNestedContextsCount() {
+    return nestedContexts_.size();
+  }
+  /**
+   * <pre>
+   * Contexts contained inside this context (e.g. nested whiles).
+   * </pre>
+   *
+   * <code>repeated .tensorflow.ControlFlowContextDef nested_contexts = 12;</code>
+   */
+  public org.tensorflow.proto.framework.ControlFlowContextDef getNestedContexts(int index) {
+    return nestedContexts_.get(index);
+  }
+  /**
+   * <pre>
+   * Contexts contained inside this context (e.g. nested whiles).
+   * </pre>
+   *
+   * <code>repeated .tensorflow.ControlFlowContextDef nested_contexts = 12;</code>
+   */
+  public org.tensorflow.proto.framework.ControlFlowContextDefOrBuilder getNestedContextsOrBuilder(
+      int index) {
+    return nestedContexts_.get(index);
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    if (!getContextNameBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 1, contextName_);
+    }
+    if (parallelIterations_ != 0) {
+      output.writeInt32(2, parallelIterations_);
+    }
+    if (backProp_ != false) {
+      output.writeBool(3, backProp_);
+    }
+    if (swapMemory_ != false) {
+      output.writeBool(4, swapMemory_);
+    }
+    if (!getPivotNameBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 5, pivotName_);
+    }
+    if (!getPivotForPredNameBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 6, pivotForPredName_);
+    }
+    if (!getPivotForBodyNameBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 7, pivotForBodyName_);
+    }
+    for (int i = 0; i < loopExitNames_.size(); i++) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 8, loopExitNames_.getRaw(i));
+    }
+    if (valuesDef_ != null) {
+      output.writeMessage(9, getValuesDef());
+    }
+    for (int i = 0; i < loopEnterNames_.size(); i++) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 10, loopEnterNames_.getRaw(i));
+    }
+    if (!getMaximumIterationsNameBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 11, maximumIterationsName_);
+    }
+    for (int i = 0; i < nestedContexts_.size(); i++) {
+      output.writeMessage(12, nestedContexts_.get(i));
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (!getContextNameBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, contextName_);
+    }
+    if (parallelIterations_ != 0) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt32Size(2, parallelIterations_);
+    }
+    if (backProp_ != false) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeBoolSize(3, backProp_);
+    }
+    if (swapMemory_ != false) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeBoolSize(4, swapMemory_);
+    }
+    if (!getPivotNameBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(5, pivotName_);
+    }
+    if (!getPivotForPredNameBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(6, pivotForPredName_);
+    }
+    if (!getPivotForBodyNameBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(7, pivotForBodyName_);
+    }
+    {
+      int dataSize = 0;
+      for (int i = 0; i < loopExitNames_.size(); i++) {
+        dataSize += computeStringSizeNoTag(loopExitNames_.getRaw(i));
+      }
+      size += dataSize;
+      size += 1 * getLoopExitNamesList().size();
+    }
+    if (valuesDef_ != null) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(9, getValuesDef());
+    }
+    {
+      int dataSize = 0;
+      for (int i = 0; i < loopEnterNames_.size(); i++) {
+        dataSize += computeStringSizeNoTag(loopEnterNames_.getRaw(i));
+      }
+      size += dataSize;
+      size += 1 * getLoopEnterNamesList().size();
+    }
+    if (!getMaximumIterationsNameBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(11, maximumIterationsName_);
+    }
+    for (int i = 0; i < nestedContexts_.size(); i++) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(12, nestedContexts_.get(i));
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.framework.WhileContextDef)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.framework.WhileContextDef other = (org.tensorflow.proto.framework.WhileContextDef) obj;
+
+    if (!getContextName()
+        .equals(other.getContextName())) return false;
+    if (getParallelIterations()
+        != other.getParallelIterations()) return false;
+    if (getBackProp()
+        != other.getBackProp()) return false;
+    if (getSwapMemory()
+        != other.getSwapMemory()) return false;
+    if (!getPivotName()
+        .equals(other.getPivotName())) return false;
+    if (!getPivotForPredName()
+        .equals(other.getPivotForPredName())) return false;
+    if (!getPivotForBodyName()
+        .equals(other.getPivotForBodyName())) return false;
+    if (!getLoopExitNamesList()
+        .equals(other.getLoopExitNamesList())) return false;
+    if (!getLoopEnterNamesList()
+        .equals(other.getLoopEnterNamesList())) return false;
+    if (hasValuesDef() != other.hasValuesDef()) return false;
+    if (hasValuesDef()) {
+      if (!getValuesDef()
+          .equals(other.getValuesDef())) return false;
+    }
+    if (!getMaximumIterationsName()
+        .equals(other.getMaximumIterationsName())) return false;
+    if (!getNestedContextsList()
+        .equals(other.getNestedContextsList())) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    hash = (37 * hash) + CONTEXT_NAME_FIELD_NUMBER;
+    hash = (53 * hash) + getContextName().hashCode();
+    hash = (37 * hash) + PARALLEL_ITERATIONS_FIELD_NUMBER;
+    hash = (53 * hash) + getParallelIterations();
+    hash = (37 * hash) + BACK_PROP_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashBoolean(
+        getBackProp());
+    hash = (37 * hash) + SWAP_MEMORY_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashBoolean(
+        getSwapMemory());
+    hash = (37 * hash) + PIVOT_NAME_FIELD_NUMBER;
+    hash = (53 * hash) + getPivotName().hashCode();
+    hash = (37 * hash) + PIVOT_FOR_PRED_NAME_FIELD_NUMBER;
+    hash = (53 * hash) + getPivotForPredName().hashCode();
+    hash = (37 * hash) + PIVOT_FOR_BODY_NAME_FIELD_NUMBER;
+    hash = (53 * hash) + getPivotForBodyName().hashCode();
+    if (getLoopExitNamesCount() > 0) {
+      hash = (37 * hash) + LOOP_EXIT_NAMES_FIELD_NUMBER;
+      hash = (53 * hash) + getLoopExitNamesList().hashCode();
+    }
+    if (getLoopEnterNamesCount() > 0) {
+      hash = (37 * hash) + LOOP_ENTER_NAMES_FIELD_NUMBER;
+      hash = (53 * hash) + getLoopEnterNamesList().hashCode();
+    }
+    if (hasValuesDef()) {
+      hash = (37 * hash) + VALUES_DEF_FIELD_NUMBER;
+      hash = (53 * hash) + getValuesDef().hashCode();
+    }
+    hash = (37 * hash) + MAXIMUM_ITERATIONS_NAME_FIELD_NUMBER;
+    hash = (53 * hash) + getMaximumIterationsName().hashCode();
+    if (getNestedContextsCount() > 0) {
+      hash = (37 * hash) + NESTED_CONTEXTS_FIELD_NUMBER;
+      hash = (53 * hash) + getNestedContextsList().hashCode();
+    }
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.framework.WhileContextDef parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.WhileContextDef parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.WhileContextDef parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.WhileContextDef parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.WhileContextDef parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.framework.WhileContextDef parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.WhileContextDef parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.WhileContextDef parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.WhileContextDef parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.WhileContextDef parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.framework.WhileContextDef parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.framework.WhileContextDef parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.framework.WhileContextDef prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * <pre>
+   * Protocol buffer representing a WhileContext object.
+   * </pre>
+   *
+   * Protobuf type {@code tensorflow.WhileContextDef}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.WhileContextDef)
+      org.tensorflow.proto.framework.WhileContextDefOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.framework.ControlFlowProtos.internal_static_tensorflow_WhileContextDef_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.framework.ControlFlowProtos.internal_static_tensorflow_WhileContextDef_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.framework.WhileContextDef.class, org.tensorflow.proto.framework.WhileContextDef.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.framework.WhileContextDef.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+        getNestedContextsFieldBuilder();
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      contextName_ = "";
+
+      parallelIterations_ = 0;
+
+      backProp_ = false;
+
+      swapMemory_ = false;
+
+      pivotName_ = "";
+
+      pivotForPredName_ = "";
+
+      pivotForBodyName_ = "";
+
+      loopExitNames_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+      bitField0_ = (bitField0_ & ~0x00000001);
+      loopEnterNames_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+      bitField0_ = (bitField0_ & ~0x00000002);
+      if (valuesDefBuilder_ == null) {
+        valuesDef_ = null;
+      } else {
+        valuesDef_ = null;
+        valuesDefBuilder_ = null;
+      }
+      maximumIterationsName_ = "";
+
+      if (nestedContextsBuilder_ == null) {
+        nestedContexts_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000004);
+      } else {
+        nestedContextsBuilder_.clear();
+      }
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.framework.ControlFlowProtos.internal_static_tensorflow_WhileContextDef_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.WhileContextDef getDefaultInstanceForType() {
+      return org.tensorflow.proto.framework.WhileContextDef.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.WhileContextDef build() {
+      org.tensorflow.proto.framework.WhileContextDef result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.framework.WhileContextDef buildPartial() {
+      org.tensorflow.proto.framework.WhileContextDef result = new org.tensorflow.proto.framework.WhileContextDef(this);
+      int from_bitField0_ = bitField0_;
+      result.contextName_ = contextName_;
+      result.parallelIterations_ = parallelIterations_;
+      result.backProp_ = backProp_;
+      result.swapMemory_ = swapMemory_;
+      result.pivotName_ = pivotName_;
+      result.pivotForPredName_ = pivotForPredName_;
+      result.pivotForBodyName_ = pivotForBodyName_;
+      if (((bitField0_ & 0x00000001) != 0)) {
+        loopExitNames_ = loopExitNames_.getUnmodifiableView();
+        bitField0_ = (bitField0_ & ~0x00000001);
+      }
+      result.loopExitNames_ = loopExitNames_;
+      if (((bitField0_ & 0x00000002) != 0)) {
+        loopEnterNames_ = loopEnterNames_.getUnmodifiableView();
+        bitField0_ = (bitField0_ & ~0x00000002);
+      }
+      result.loopEnterNames_ = loopEnterNames_;
+      if (valuesDefBuilder_ == null) {
+        result.valuesDef_ = valuesDef_;
+      } else {
+        result.valuesDef_ = valuesDefBuilder_.build();
+      }
+      result.maximumIterationsName_ = maximumIterationsName_;
+      if (nestedContextsBuilder_ == null) {
+        if (((bitField0_ & 0x00000004) != 0)) {
+          nestedContexts_ = java.util.Collections.unmodifiableList(nestedContexts_);
+          bitField0_ = (bitField0_ & ~0x00000004);
+        }
+        result.nestedContexts_ = nestedContexts_;
+      } else {
+        result.nestedContexts_ = nestedContextsBuilder_.build();
+      }
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.framework.WhileContextDef) {
+        return mergeFrom((org.tensorflow.proto.framework.WhileContextDef)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.framework.WhileContextDef other) {
+      if (other == org.tensorflow.proto.framework.WhileContextDef.getDefaultInstance()) return this;
+      if (!other.getContextName().isEmpty()) {
+        contextName_ = other.contextName_;
+        onChanged();
+      }
+      if (other.getParallelIterations() != 0) {
+        setParallelIterations(other.getParallelIterations());
+      }
+      if (other.getBackProp() != false) {
+        setBackProp(other.getBackProp());
+      }
+      if (other.getSwapMemory() != false) {
+        setSwapMemory(other.getSwapMemory());
+      }
+      if (!other.getPivotName().isEmpty()) {
+        pivotName_ = other.pivotName_;
+        onChanged();
+      }
+      if (!other.getPivotForPredName().isEmpty()) {
+        pivotForPredName_ = other.pivotForPredName_;
+        onChanged();
+      }
+      if (!other.getPivotForBodyName().isEmpty()) {
+        pivotForBodyName_ = other.pivotForBodyName_;
+        onChanged();
+      }
+      if (!other.loopExitNames_.isEmpty()) {
+        if (loopExitNames_.isEmpty()) {
+          loopExitNames_ = other.loopExitNames_;
+          bitField0_ = (bitField0_ & ~0x00000001);
+        } else {
+          ensureLoopExitNamesIsMutable();
+          loopExitNames_.addAll(other.loopExitNames_);
+        }
+        onChanged();
+      }
+      if (!other.loopEnterNames_.isEmpty()) {
+        if (loopEnterNames_.isEmpty()) {
+          loopEnterNames_ = other.loopEnterNames_;
+          bitField0_ = (bitField0_ & ~0x00000002);
+        } else {
+          ensureLoopEnterNamesIsMutable();
+          loopEnterNames_.addAll(other.loopEnterNames_);
+        }
+        onChanged();
+      }
+      if (other.hasValuesDef()) {
+        mergeValuesDef(other.getValuesDef());
+      }
+      if (!other.getMaximumIterationsName().isEmpty()) {
+        maximumIterationsName_ = other.maximumIterationsName_;
+        onChanged();
+      }
+      if (nestedContextsBuilder_ == null) {
+        if (!other.nestedContexts_.isEmpty()) {
+          if (nestedContexts_.isEmpty()) {
+            nestedContexts_ = other.nestedContexts_;
+            bitField0_ = (bitField0_ & ~0x00000004);
+          } else {
+            ensureNestedContextsIsMutable();
+            nestedContexts_.addAll(other.nestedContexts_);
+          }
+          onChanged();
+        }
+      } else {
+        if (!other.nestedContexts_.isEmpty()) {
+          if (nestedContextsBuilder_.isEmpty()) {
+            nestedContextsBuilder_.dispose();
+            nestedContextsBuilder_ = null;
+            nestedContexts_ = other.nestedContexts_;
+            bitField0_ = (bitField0_ & ~0x00000004);
+            nestedContextsBuilder_ = 
+              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
+                 getNestedContextsFieldBuilder() : null;
+          } else {
+            nestedContextsBuilder_.addAllMessages(other.nestedContexts_);
+          }
+        }
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.framework.WhileContextDef parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.framework.WhileContextDef) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+    private int bitField0_;
+
+    private java.lang.Object contextName_ = "";
+    /**
+     * <pre>
+     * Name of the context.
+     * </pre>
+     *
+     * <code>string context_name = 1;</code>
+     */
+    public java.lang.String getContextName() {
+      java.lang.Object ref = contextName_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        contextName_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Name of the context.
+     * </pre>
+     *
+     * <code>string context_name = 1;</code>
+     */
+    public com.google.protobuf.ByteString
+        getContextNameBytes() {
+      java.lang.Object ref = contextName_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        contextName_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Name of the context.
+     * </pre>
+     *
+     * <code>string context_name = 1;</code>
+     */
+    public Builder setContextName(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      contextName_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Name of the context.
+     * </pre>
+     *
+     * <code>string context_name = 1;</code>
+     */
+    public Builder clearContextName() {
+      
+      contextName_ = getDefaultInstance().getContextName();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Name of the context.
+     * </pre>
+     *
+     * <code>string context_name = 1;</code>
+     */
+    public Builder setContextNameBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      contextName_ = value;
+      onChanged();
+      return this;
+    }
+
+    private int parallelIterations_ ;
+    /**
+     * <pre>
+     * The number of iterations allowed to run in parallel.
+     * </pre>
+     *
+     * <code>int32 parallel_iterations = 2;</code>
+     */
+    public int getParallelIterations() {
+      return parallelIterations_;
+    }
+    /**
+     * <pre>
+     * The number of iterations allowed to run in parallel.
+     * </pre>
+     *
+     * <code>int32 parallel_iterations = 2;</code>
+     */
+    public Builder setParallelIterations(int value) {
+      
+      parallelIterations_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * The number of iterations allowed to run in parallel.
+     * </pre>
+     *
+     * <code>int32 parallel_iterations = 2;</code>
+     */
+    public Builder clearParallelIterations() {
+      
+      parallelIterations_ = 0;
+      onChanged();
+      return this;
+    }
+
+    private boolean backProp_ ;
+    /**
+     * <pre>
+     * Whether backprop is enabled for this while loop.
+     * </pre>
+     *
+     * <code>bool back_prop = 3;</code>
+     */
+    public boolean getBackProp() {
+      return backProp_;
+    }
+    /**
+     * <pre>
+     * Whether backprop is enabled for this while loop.
+     * </pre>
+     *
+     * <code>bool back_prop = 3;</code>
+     */
+    public Builder setBackProp(boolean value) {
+      
+      backProp_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Whether backprop is enabled for this while loop.
+     * </pre>
+     *
+     * <code>bool back_prop = 3;</code>
+     */
+    public Builder clearBackProp() {
+      
+      backProp_ = false;
+      onChanged();
+      return this;
+    }
+
+    private boolean swapMemory_ ;
+    /**
+     * <pre>
+     * Whether GPU-CPU memory swap is enabled for this loop.
+     * </pre>
+     *
+     * <code>bool swap_memory = 4;</code>
+     */
+    public boolean getSwapMemory() {
+      return swapMemory_;
+    }
+    /**
+     * <pre>
+     * Whether GPU-CPU memory swap is enabled for this loop.
+     * </pre>
+     *
+     * <code>bool swap_memory = 4;</code>
+     */
+    public Builder setSwapMemory(boolean value) {
+      
+      swapMemory_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Whether GPU-CPU memory swap is enabled for this loop.
+     * </pre>
+     *
+     * <code>bool swap_memory = 4;</code>
+     */
+    public Builder clearSwapMemory() {
+      
+      swapMemory_ = false;
+      onChanged();
+      return this;
+    }
+
+    private java.lang.Object pivotName_ = "";
+    /**
+     * <pre>
+     * Name of the pivot tensor.
+     * </pre>
+     *
+     * <code>string pivot_name = 5;</code>
+     */
+    public java.lang.String getPivotName() {
+      java.lang.Object ref = pivotName_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        pivotName_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Name of the pivot tensor.
+     * </pre>
+     *
+     * <code>string pivot_name = 5;</code>
+     */
+    public com.google.protobuf.ByteString
+        getPivotNameBytes() {
+      java.lang.Object ref = pivotName_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        pivotName_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Name of the pivot tensor.
+     * </pre>
+     *
+     * <code>string pivot_name = 5;</code>
+     */
+    public Builder setPivotName(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      pivotName_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Name of the pivot tensor.
+     * </pre>
+     *
+     * <code>string pivot_name = 5;</code>
+     */
+    public Builder clearPivotName() {
+      
+      pivotName_ = getDefaultInstance().getPivotName();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Name of the pivot tensor.
+     * </pre>
+     *
+     * <code>string pivot_name = 5;</code>
+     */
+    public Builder setPivotNameBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      pivotName_ = value;
+      onChanged();
+      return this;
+    }
+
+    private java.lang.Object pivotForPredName_ = "";
+    /**
+     * <pre>
+     * Name of the pivot_for_pred tensor.
+     * </pre>
+     *
+     * <code>string pivot_for_pred_name = 6;</code>
+     */
+    public java.lang.String getPivotForPredName() {
+      java.lang.Object ref = pivotForPredName_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        pivotForPredName_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Name of the pivot_for_pred tensor.
+     * </pre>
+     *
+     * <code>string pivot_for_pred_name = 6;</code>
+     */
+    public com.google.protobuf.ByteString
+        getPivotForPredNameBytes() {
+      java.lang.Object ref = pivotForPredName_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        pivotForPredName_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Name of the pivot_for_pred tensor.
+     * </pre>
+     *
+     * <code>string pivot_for_pred_name = 6;</code>
+     */
+    public Builder setPivotForPredName(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      pivotForPredName_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Name of the pivot_for_pred tensor.
+     * </pre>
+     *
+     * <code>string pivot_for_pred_name = 6;</code>
+     */
+    public Builder clearPivotForPredName() {
+      
+      pivotForPredName_ = getDefaultInstance().getPivotForPredName();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Name of the pivot_for_pred tensor.
+     * </pre>
+     *
+     * <code>string pivot_for_pred_name = 6;</code>
+     */
+    public Builder setPivotForPredNameBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      pivotForPredName_ = value;
+      onChanged();
+      return this;
+    }
+
+    private java.lang.Object pivotForBodyName_ = "";
+    /**
+     * <pre>
+     * Name of the pivot_for_body tensor.
+     * </pre>
+     *
+     * <code>string pivot_for_body_name = 7;</code>
+     */
+    public java.lang.String getPivotForBodyName() {
+      java.lang.Object ref = pivotForBodyName_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        pivotForBodyName_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Name of the pivot_for_body tensor.
+     * </pre>
+     *
+     * <code>string pivot_for_body_name = 7;</code>
+     */
+    public com.google.protobuf.ByteString
+        getPivotForBodyNameBytes() {
+      java.lang.Object ref = pivotForBodyName_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        pivotForBodyName_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Name of the pivot_for_body tensor.
+     * </pre>
+     *
+     * <code>string pivot_for_body_name = 7;</code>
+     */
+    public Builder setPivotForBodyName(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      pivotForBodyName_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Name of the pivot_for_body tensor.
+     * </pre>
+     *
+     * <code>string pivot_for_body_name = 7;</code>
+     */
+    public Builder clearPivotForBodyName() {
+      
+      pivotForBodyName_ = getDefaultInstance().getPivotForBodyName();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Name of the pivot_for_body tensor.
+     * </pre>
+     *
+     * <code>string pivot_for_body_name = 7;</code>
+     */
+    public Builder setPivotForBodyNameBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      pivotForBodyName_ = value;
+      onChanged();
+      return this;
+    }
+
+    private com.google.protobuf.LazyStringList loopExitNames_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+    private void ensureLoopExitNamesIsMutable() {
+      if (!((bitField0_ & 0x00000001) != 0)) {
+        loopExitNames_ = new com.google.protobuf.LazyStringArrayList(loopExitNames_);
+        bitField0_ |= 0x00000001;
+       }
+    }
+    /**
+     * <pre>
+     * List of names for exit tensors.
+     * </pre>
+     *
+     * <code>repeated string loop_exit_names = 8;</code>
+     */
+    public com.google.protobuf.ProtocolStringList
+        getLoopExitNamesList() {
+      return loopExitNames_.getUnmodifiableView();
+    }
+    /**
+     * <pre>
+     * List of names for exit tensors.
+     * </pre>
+     *
+     * <code>repeated string loop_exit_names = 8;</code>
+     */
+    public int getLoopExitNamesCount() {
+      return loopExitNames_.size();
+    }
+    /**
+     * <pre>
+     * List of names for exit tensors.
+     * </pre>
+     *
+     * <code>repeated string loop_exit_names = 8;</code>
+     */
+    public java.lang.String getLoopExitNames(int index) {
+      return loopExitNames_.get(index);
+    }
+    /**
+     * <pre>
+     * List of names for exit tensors.
+     * </pre>
+     *
+     * <code>repeated string loop_exit_names = 8;</code>
+     */
+    public com.google.protobuf.ByteString
+        getLoopExitNamesBytes(int index) {
+      return loopExitNames_.getByteString(index);
+    }
+    /**
+     * <pre>
+     * List of names for exit tensors.
+     * </pre>
+     *
+     * <code>repeated string loop_exit_names = 8;</code>
+     */
+    public Builder setLoopExitNames(
+        int index, java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  ensureLoopExitNamesIsMutable();
+      loopExitNames_.set(index, value);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * List of names for exit tensors.
+     * </pre>
+     *
+     * <code>repeated string loop_exit_names = 8;</code>
+     */
+    public Builder addLoopExitNames(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  ensureLoopExitNamesIsMutable();
+      loopExitNames_.add(value);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * List of names for exit tensors.
+     * </pre>
+     *
+     * <code>repeated string loop_exit_names = 8;</code>
+     */
+    public Builder addAllLoopExitNames(
+        java.lang.Iterable<java.lang.String> values) {
+      ensureLoopExitNamesIsMutable();
+      com.google.protobuf.AbstractMessageLite.Builder.addAll(
+          values, loopExitNames_);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * List of names for exit tensors.
+     * </pre>
+     *
+     * <code>repeated string loop_exit_names = 8;</code>
+     */
+    public Builder clearLoopExitNames() {
+      loopExitNames_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+      bitField0_ = (bitField0_ & ~0x00000001);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * List of names for exit tensors.
+     * </pre>
+     *
+     * <code>repeated string loop_exit_names = 8;</code>
+     */
+    public Builder addLoopExitNamesBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      ensureLoopExitNamesIsMutable();
+      loopExitNames_.add(value);
+      onChanged();
+      return this;
+    }
+
+    private com.google.protobuf.LazyStringList loopEnterNames_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+    private void ensureLoopEnterNamesIsMutable() {
+      if (!((bitField0_ & 0x00000002) != 0)) {
+        loopEnterNames_ = new com.google.protobuf.LazyStringArrayList(loopEnterNames_);
+        bitField0_ |= 0x00000002;
+       }
+    }
+    /**
+     * <pre>
+     * List of names for enter tensors.
+     * </pre>
+     *
+     * <code>repeated string loop_enter_names = 10;</code>
+     */
+    public com.google.protobuf.ProtocolStringList
+        getLoopEnterNamesList() {
+      return loopEnterNames_.getUnmodifiableView();
+    }
+    /**
+     * <pre>
+     * List of names for enter tensors.
+     * </pre>
+     *
+     * <code>repeated string loop_enter_names = 10;</code>
+     */
+    public int getLoopEnterNamesCount() {
+      return loopEnterNames_.size();
+    }
+    /**
+     * <pre>
+     * List of names for enter tensors.
+     * </pre>
+     *
+     * <code>repeated string loop_enter_names = 10;</code>
+     */
+    public java.lang.String getLoopEnterNames(int index) {
+      return loopEnterNames_.get(index);
+    }
+    /**
+     * <pre>
+     * List of names for enter tensors.
+     * </pre>
+     *
+     * <code>repeated string loop_enter_names = 10;</code>
+     */
+    public com.google.protobuf.ByteString
+        getLoopEnterNamesBytes(int index) {
+      return loopEnterNames_.getByteString(index);
+    }
+    /**
+     * <pre>
+     * List of names for enter tensors.
+     * </pre>
+     *
+     * <code>repeated string loop_enter_names = 10;</code>
+     */
+    public Builder setLoopEnterNames(
+        int index, java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  ensureLoopEnterNamesIsMutable();
+      loopEnterNames_.set(index, value);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * List of names for enter tensors.
+     * </pre>
+     *
+     * <code>repeated string loop_enter_names = 10;</code>
+     */
+    public Builder addLoopEnterNames(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  ensureLoopEnterNamesIsMutable();
+      loopEnterNames_.add(value);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * List of names for enter tensors.
+     * </pre>
+     *
+     * <code>repeated string loop_enter_names = 10;</code>
+     */
+    public Builder addAllLoopEnterNames(
+        java.lang.Iterable<java.lang.String> values) {
+      ensureLoopEnterNamesIsMutable();
+      com.google.protobuf.AbstractMessageLite.Builder.addAll(
+          values, loopEnterNames_);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * List of names for enter tensors.
+     * </pre>
+     *
+     * <code>repeated string loop_enter_names = 10;</code>
+     */
+    public Builder clearLoopEnterNames() {
+      loopEnterNames_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+      bitField0_ = (bitField0_ & ~0x00000002);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * List of names for enter tensors.
+     * </pre>
+     *
+     * <code>repeated string loop_enter_names = 10;</code>
+     */
+    public Builder addLoopEnterNamesBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      ensureLoopEnterNamesIsMutable();
+      loopEnterNames_.add(value);
+      onChanged();
+      return this;
+    }
+
+    private org.tensorflow.proto.framework.ValuesDef valuesDef_;
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.ValuesDef, org.tensorflow.proto.framework.ValuesDef.Builder, org.tensorflow.proto.framework.ValuesDefOrBuilder> valuesDefBuilder_;
+    /**
+     * <pre>
+     * Values and external values in control flow context.
+     * </pre>
+     *
+     * <code>.tensorflow.ValuesDef values_def = 9;</code>
+     */
+    public boolean hasValuesDef() {
+      return valuesDefBuilder_ != null || valuesDef_ != null;
+    }
+    /**
+     * <pre>
+     * Values and external values in control flow context.
+     * </pre>
+     *
+     * <code>.tensorflow.ValuesDef values_def = 9;</code>
+     */
+    public org.tensorflow.proto.framework.ValuesDef getValuesDef() {
+      if (valuesDefBuilder_ == null) {
+        return valuesDef_ == null ? org.tensorflow.proto.framework.ValuesDef.getDefaultInstance() : valuesDef_;
+      } else {
+        return valuesDefBuilder_.getMessage();
+      }
+    }
+    /**
+     * <pre>
+     * Values and external values in control flow context.
+     * </pre>
+     *
+     * <code>.tensorflow.ValuesDef values_def = 9;</code>
+     */
+    public Builder setValuesDef(org.tensorflow.proto.framework.ValuesDef value) {
+      if (valuesDefBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        valuesDef_ = value;
+        onChanged();
+      } else {
+        valuesDefBuilder_.setMessage(value);
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Values and external values in control flow context.
+     * </pre>
+     *
+     * <code>.tensorflow.ValuesDef values_def = 9;</code>
+     */
+    public Builder setValuesDef(
+        org.tensorflow.proto.framework.ValuesDef.Builder builderForValue) {
+      if (valuesDefBuilder_ == null) {
+        valuesDef_ = builderForValue.build();
+        onChanged();
+      } else {
+        valuesDefBuilder_.setMessage(builderForValue.build());
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Values and external values in control flow context.
+     * </pre>
+     *
+     * <code>.tensorflow.ValuesDef values_def = 9;</code>
+     */
+    public Builder mergeValuesDef(org.tensorflow.proto.framework.ValuesDef value) {
+      if (valuesDefBuilder_ == null) {
+        if (valuesDef_ != null) {
+          valuesDef_ =
+            org.tensorflow.proto.framework.ValuesDef.newBuilder(valuesDef_).mergeFrom(value).buildPartial();
+        } else {
+          valuesDef_ = value;
+        }
+        onChanged();
+      } else {
+        valuesDefBuilder_.mergeFrom(value);
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Values and external values in control flow context.
+     * </pre>
+     *
+     * <code>.tensorflow.ValuesDef values_def = 9;</code>
+     */
+    public Builder clearValuesDef() {
+      if (valuesDefBuilder_ == null) {
+        valuesDef_ = null;
+        onChanged();
+      } else {
+        valuesDef_ = null;
+        valuesDefBuilder_ = null;
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Values and external values in control flow context.
+     * </pre>
+     *
+     * <code>.tensorflow.ValuesDef values_def = 9;</code>
+     */
+    public org.tensorflow.proto.framework.ValuesDef.Builder getValuesDefBuilder() {
+      
+      onChanged();
+      return getValuesDefFieldBuilder().getBuilder();
+    }
+    /**
+     * <pre>
+     * Values and external values in control flow context.
+     * </pre>
+     *
+     * <code>.tensorflow.ValuesDef values_def = 9;</code>
+     */
+    public org.tensorflow.proto.framework.ValuesDefOrBuilder getValuesDefOrBuilder() {
+      if (valuesDefBuilder_ != null) {
+        return valuesDefBuilder_.getMessageOrBuilder();
+      } else {
+        return valuesDef_ == null ?
+            org.tensorflow.proto.framework.ValuesDef.getDefaultInstance() : valuesDef_;
+      }
+    }
+    /**
+     * <pre>
+     * Values and external values in control flow context.
+     * </pre>
+     *
+     * <code>.tensorflow.ValuesDef values_def = 9;</code>
+     */
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.ValuesDef, org.tensorflow.proto.framework.ValuesDef.Builder, org.tensorflow.proto.framework.ValuesDefOrBuilder> 
+        getValuesDefFieldBuilder() {
+      if (valuesDefBuilder_ == null) {
+        valuesDefBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+            org.tensorflow.proto.framework.ValuesDef, org.tensorflow.proto.framework.ValuesDef.Builder, org.tensorflow.proto.framework.ValuesDefOrBuilder>(
+                getValuesDef(),
+                getParentForChildren(),
+                isClean());
+        valuesDef_ = null;
+      }
+      return valuesDefBuilder_;
+    }
+
+    private java.lang.Object maximumIterationsName_ = "";
+    /**
+     * <pre>
+     * Optional name of the maximum_iterations tensor.
+     * </pre>
+     *
+     * <code>string maximum_iterations_name = 11;</code>
+     */
+    public java.lang.String getMaximumIterationsName() {
+      java.lang.Object ref = maximumIterationsName_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        maximumIterationsName_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Optional name of the maximum_iterations tensor.
+     * </pre>
+     *
+     * <code>string maximum_iterations_name = 11;</code>
+     */
+    public com.google.protobuf.ByteString
+        getMaximumIterationsNameBytes() {
+      java.lang.Object ref = maximumIterationsName_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        maximumIterationsName_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Optional name of the maximum_iterations tensor.
+     * </pre>
+     *
+     * <code>string maximum_iterations_name = 11;</code>
+     */
+    public Builder setMaximumIterationsName(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      maximumIterationsName_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Optional name of the maximum_iterations tensor.
+     * </pre>
+     *
+     * <code>string maximum_iterations_name = 11;</code>
+     */
+    public Builder clearMaximumIterationsName() {
+      
+      maximumIterationsName_ = getDefaultInstance().getMaximumIterationsName();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Optional name of the maximum_iterations tensor.
+     * </pre>
+     *
+     * <code>string maximum_iterations_name = 11;</code>
+     */
+    public Builder setMaximumIterationsNameBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      maximumIterationsName_ = value;
+      onChanged();
+      return this;
+    }
+
+    private java.util.List<org.tensorflow.proto.framework.ControlFlowContextDef> nestedContexts_ =
+      java.util.Collections.emptyList();
+    private void ensureNestedContextsIsMutable() {
+      if (!((bitField0_ & 0x00000004) != 0)) {
+        nestedContexts_ = new java.util.ArrayList<org.tensorflow.proto.framework.ControlFlowContextDef>(nestedContexts_);
+        bitField0_ |= 0x00000004;
+       }
+    }
+
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.framework.ControlFlowContextDef, org.tensorflow.proto.framework.ControlFlowContextDef.Builder, org.tensorflow.proto.framework.ControlFlowContextDefOrBuilder> nestedContextsBuilder_;
+
+    /**
+     * <pre>
+     * Contexts contained inside this context (e.g. nested whiles).
+     * </pre>
+     *
+     * <code>repeated .tensorflow.ControlFlowContextDef nested_contexts = 12;</code>
+     */
+    public java.util.List<org.tensorflow.proto.framework.ControlFlowContextDef> getNestedContextsList() {
+      if (nestedContextsBuilder_ == null) {
+        return java.util.Collections.unmodifiableList(nestedContexts_);
+      } else {
+        return nestedContextsBuilder_.getMessageList();
+      }
+    }
+    /**
+     * <pre>
+     * Contexts contained inside this context (e.g. nested whiles).
+     * </pre>
+     *
+     * <code>repeated .tensorflow.ControlFlowContextDef nested_contexts = 12;</code>
+     */
+    public int getNestedContextsCount() {
+      if (nestedContextsBuilder_ == null) {
+        return nestedContexts_.size();
+      } else {
+        return nestedContextsBuilder_.getCount();
+      }
+    }
+    /**
+     * <pre>
+     * Contexts contained inside this context (e.g. nested whiles).
+     * </pre>
+     *
+     * <code>repeated .tensorflow.ControlFlowContextDef nested_contexts = 12;</code>
+     */
+    public org.tensorflow.proto.framework.ControlFlowContextDef getNestedContexts(int index) {
+      if (nestedContextsBuilder_ == null) {
+        return nestedContexts_.get(index);
+      } else {
+        return nestedContextsBuilder_.getMessage(index);
+      }
+    }
+    /**
+     * <pre>
+     * Contexts contained inside this context (e.g. nested whiles).
+     * </pre>
+     *
+     * <code>repeated .tensorflow.ControlFlowContextDef nested_contexts = 12;</code>
+     */
+    public Builder setNestedContexts(
+        int index, org.tensorflow.proto.framework.ControlFlowContextDef value) {
+      if (nestedContextsBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureNestedContextsIsMutable();
+        nestedContexts_.set(index, value);
+        onChanged();
+      } else {
+        nestedContextsBuilder_.setMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Contexts contained inside this context (e.g. nested whiles).
+     * </pre>
+     *
+     * <code>repeated .tensorflow.ControlFlowContextDef nested_contexts = 12;</code>
+     */
+    public Builder setNestedContexts(
+        int index, org.tensorflow.proto.framework.ControlFlowContextDef.Builder builderForValue) {
+      if (nestedContextsBuilder_ == null) {
+        ensureNestedContextsIsMutable();
+        nestedContexts_.set(index, builderForValue.build());
+        onChanged();
+      } else {
+        nestedContextsBuilder_.setMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Contexts contained inside this context (e.g. nested whiles).
+     * </pre>
+     *
+     * <code>repeated .tensorflow.ControlFlowContextDef nested_contexts = 12;</code>
+     */
+    public Builder addNestedContexts(org.tensorflow.proto.framework.ControlFlowContextDef value) {
+      if (nestedContextsBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureNestedContextsIsMutable();
+        nestedContexts_.add(value);
+        onChanged();
+      } else {
+        nestedContextsBuilder_.addMessage(value);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Contexts contained inside this context (e.g. nested whiles).
+     * </pre>
+     *
+     * <code>repeated .tensorflow.ControlFlowContextDef nested_contexts = 12;</code>
+     */
+    public Builder addNestedContexts(
+        int index, org.tensorflow.proto.framework.ControlFlowContextDef value) {
+      if (nestedContextsBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureNestedContextsIsMutable();
+        nestedContexts_.add(index, value);
+        onChanged();
+      } else {
+        nestedContextsBuilder_.addMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Contexts contained inside this context (e.g. nested whiles).
+     * </pre>
+     *
+     * <code>repeated .tensorflow.ControlFlowContextDef nested_contexts = 12;</code>
+     */
+    public Builder addNestedContexts(
+        org.tensorflow.proto.framework.ControlFlowContextDef.Builder builderForValue) {
+      if (nestedContextsBuilder_ == null) {
+        ensureNestedContextsIsMutable();
+        nestedContexts_.add(builderForValue.build());
+        onChanged();
+      } else {
+        nestedContextsBuilder_.addMessage(builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Contexts contained inside this context (e.g. nested whiles).
+     * </pre>
+     *
+     * <code>repeated .tensorflow.ControlFlowContextDef nested_contexts = 12;</code>
+     */
+    public Builder addNestedContexts(
+        int index, org.tensorflow.proto.framework.ControlFlowContextDef.Builder builderForValue) {
+      if (nestedContextsBuilder_ == null) {
+        ensureNestedContextsIsMutable();
+        nestedContexts_.add(index, builderForValue.build());
+        onChanged();
+      } else {
+        nestedContextsBuilder_.addMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Contexts contained inside this context (e.g. nested whiles).
+     * </pre>
+     *
+     * <code>repeated .tensorflow.ControlFlowContextDef nested_contexts = 12;</code>
+     */
+    public Builder addAllNestedContexts(
+        java.lang.Iterable<? extends org.tensorflow.proto.framework.ControlFlowContextDef> values) {
+      if (nestedContextsBuilder_ == null) {
+        ensureNestedContextsIsMutable();
+        com.google.protobuf.AbstractMessageLite.Builder.addAll(
+            values, nestedContexts_);
+        onChanged();
+      } else {
+        nestedContextsBuilder_.addAllMessages(values);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Contexts contained inside this context (e.g. nested whiles).
+     * </pre>
+     *
+     * <code>repeated .tensorflow.ControlFlowContextDef nested_contexts = 12;</code>
+     */
+    public Builder clearNestedContexts() {
+      if (nestedContextsBuilder_ == null) {
+        nestedContexts_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000004);
+        onChanged();
+      } else {
+        nestedContextsBuilder_.clear();
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Contexts contained inside this context (e.g. nested whiles).
+     * </pre>
+     *
+     * <code>repeated .tensorflow.ControlFlowContextDef nested_contexts = 12;</code>
+     */
+    public Builder removeNestedContexts(int index) {
+      if (nestedContextsBuilder_ == null) {
+        ensureNestedContextsIsMutable();
+        nestedContexts_.remove(index);
+        onChanged();
+      } else {
+        nestedContextsBuilder_.remove(index);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Contexts contained inside this context (e.g. nested whiles).
+     * </pre>
+     *
+     * <code>repeated .tensorflow.ControlFlowContextDef nested_contexts = 12;</code>
+     */
+    public org.tensorflow.proto.framework.ControlFlowContextDef.Builder getNestedContextsBuilder(
+        int index) {
+      return getNestedContextsFieldBuilder().getBuilder(index);
+    }
+    /**
+     * <pre>
+     * Contexts contained inside this context (e.g. nested whiles).
+     * </pre>
+     *
+     * <code>repeated .tensorflow.ControlFlowContextDef nested_contexts = 12;</code>
+     */
+    public org.tensorflow.proto.framework.ControlFlowContextDefOrBuilder getNestedContextsOrBuilder(
+        int index) {
+      if (nestedContextsBuilder_ == null) {
+        return nestedContexts_.get(index);  } else {
+        return nestedContextsBuilder_.getMessageOrBuilder(index);
+      }
+    }
+    /**
+     * <pre>
+     * Contexts contained inside this context (e.g. nested whiles).
+     * </pre>
+     *
+     * <code>repeated .tensorflow.ControlFlowContextDef nested_contexts = 12;</code>
+     */
+    public java.util.List<? extends org.tensorflow.proto.framework.ControlFlowContextDefOrBuilder> 
+         getNestedContextsOrBuilderList() {
+      if (nestedContextsBuilder_ != null) {
+        return nestedContextsBuilder_.getMessageOrBuilderList();
+      } else {
+        return java.util.Collections.unmodifiableList(nestedContexts_);
+      }
+    }
+    /**
+     * <pre>
+     * Contexts contained inside this context (e.g. nested whiles).
+     * </pre>
+     *
+     * <code>repeated .tensorflow.ControlFlowContextDef nested_contexts = 12;</code>
+     */
+    public org.tensorflow.proto.framework.ControlFlowContextDef.Builder addNestedContextsBuilder() {
+      return getNestedContextsFieldBuilder().addBuilder(
+          org.tensorflow.proto.framework.ControlFlowContextDef.getDefaultInstance());
+    }
+    /**
+     * <pre>
+     * Contexts contained inside this context (e.g. nested whiles).
+     * </pre>
+     *
+     * <code>repeated .tensorflow.ControlFlowContextDef nested_contexts = 12;</code>
+     */
+    public org.tensorflow.proto.framework.ControlFlowContextDef.Builder addNestedContextsBuilder(
+        int index) {
+      return getNestedContextsFieldBuilder().addBuilder(
+          index, org.tensorflow.proto.framework.ControlFlowContextDef.getDefaultInstance());
+    }
+    /**
+     * <pre>
+     * Contexts contained inside this context (e.g. nested whiles).
+     * </pre>
+     *
+     * <code>repeated .tensorflow.ControlFlowContextDef nested_contexts = 12;</code>
+     */
+    public java.util.List<org.tensorflow.proto.framework.ControlFlowContextDef.Builder> 
+         getNestedContextsBuilderList() {
+      return getNestedContextsFieldBuilder().getBuilderList();
+    }
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.framework.ControlFlowContextDef, org.tensorflow.proto.framework.ControlFlowContextDef.Builder, org.tensorflow.proto.framework.ControlFlowContextDefOrBuilder> 
+        getNestedContextsFieldBuilder() {
+      if (nestedContextsBuilder_ == null) {
+        nestedContextsBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
+            org.tensorflow.proto.framework.ControlFlowContextDef, org.tensorflow.proto.framework.ControlFlowContextDef.Builder, org.tensorflow.proto.framework.ControlFlowContextDefOrBuilder>(
+                nestedContexts_,
+                ((bitField0_ & 0x00000004) != 0),
+                getParentForChildren(),
+                isClean());
+        nestedContexts_ = null;
+      }
+      return nestedContextsBuilder_;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.WhileContextDef)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.WhileContextDef)
+  private static final org.tensorflow.proto.framework.WhileContextDef DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.framework.WhileContextDef();
+  }
+
+  public static org.tensorflow.proto.framework.WhileContextDef getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<WhileContextDef>
+      PARSER = new com.google.protobuf.AbstractParser<WhileContextDef>() {
+    @java.lang.Override
+    public WhileContextDef parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new WhileContextDef(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<WhileContextDef> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<WhileContextDef> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.framework.WhileContextDef getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/WhileContextDefOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/WhileContextDefOrBuilder.java
new file mode 100644
index 00000000000..6fd3e2639e1
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/framework/WhileContextDefOrBuilder.java
@@ -0,0 +1,265 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/control_flow.proto
+
+package org.tensorflow.proto.framework;
+
+public interface WhileContextDefOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.WhileContextDef)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <pre>
+   * Name of the context.
+   * </pre>
+   *
+   * <code>string context_name = 1;</code>
+   */
+  java.lang.String getContextName();
+  /**
+   * <pre>
+   * Name of the context.
+   * </pre>
+   *
+   * <code>string context_name = 1;</code>
+   */
+  com.google.protobuf.ByteString
+      getContextNameBytes();
+
+  /**
+   * <pre>
+   * The number of iterations allowed to run in parallel.
+   * </pre>
+   *
+   * <code>int32 parallel_iterations = 2;</code>
+   */
+  int getParallelIterations();
+
+  /**
+   * <pre>
+   * Whether backprop is enabled for this while loop.
+   * </pre>
+   *
+   * <code>bool back_prop = 3;</code>
+   */
+  boolean getBackProp();
+
+  /**
+   * <pre>
+   * Whether GPU-CPU memory swap is enabled for this loop.
+   * </pre>
+   *
+   * <code>bool swap_memory = 4;</code>
+   */
+  boolean getSwapMemory();
+
+  /**
+   * <pre>
+   * Name of the pivot tensor.
+   * </pre>
+   *
+   * <code>string pivot_name = 5;</code>
+   */
+  java.lang.String getPivotName();
+  /**
+   * <pre>
+   * Name of the pivot tensor.
+   * </pre>
+   *
+   * <code>string pivot_name = 5;</code>
+   */
+  com.google.protobuf.ByteString
+      getPivotNameBytes();
+
+  /**
+   * <pre>
+   * Name of the pivot_for_pred tensor.
+   * </pre>
+   *
+   * <code>string pivot_for_pred_name = 6;</code>
+   */
+  java.lang.String getPivotForPredName();
+  /**
+   * <pre>
+   * Name of the pivot_for_pred tensor.
+   * </pre>
+   *
+   * <code>string pivot_for_pred_name = 6;</code>
+   */
+  com.google.protobuf.ByteString
+      getPivotForPredNameBytes();
+
+  /**
+   * <pre>
+   * Name of the pivot_for_body tensor.
+   * </pre>
+   *
+   * <code>string pivot_for_body_name = 7;</code>
+   */
+  java.lang.String getPivotForBodyName();
+  /**
+   * <pre>
+   * Name of the pivot_for_body tensor.
+   * </pre>
+   *
+   * <code>string pivot_for_body_name = 7;</code>
+   */
+  com.google.protobuf.ByteString
+      getPivotForBodyNameBytes();
+
+  /**
+   * <pre>
+   * List of names for exit tensors.
+   * </pre>
+   *
+   * <code>repeated string loop_exit_names = 8;</code>
+   */
+  java.util.List<java.lang.String>
+      getLoopExitNamesList();
+  /**
+   * <pre>
+   * List of names for exit tensors.
+   * </pre>
+   *
+   * <code>repeated string loop_exit_names = 8;</code>
+   */
+  int getLoopExitNamesCount();
+  /**
+   * <pre>
+   * List of names for exit tensors.
+   * </pre>
+   *
+   * <code>repeated string loop_exit_names = 8;</code>
+   */
+  java.lang.String getLoopExitNames(int index);
+  /**
+   * <pre>
+   * List of names for exit tensors.
+   * </pre>
+   *
+   * <code>repeated string loop_exit_names = 8;</code>
+   */
+  com.google.protobuf.ByteString
+      getLoopExitNamesBytes(int index);
+
+  /**
+   * <pre>
+   * List of names for enter tensors.
+   * </pre>
+   *
+   * <code>repeated string loop_enter_names = 10;</code>
+   */
+  java.util.List<java.lang.String>
+      getLoopEnterNamesList();
+  /**
+   * <pre>
+   * List of names for enter tensors.
+   * </pre>
+   *
+   * <code>repeated string loop_enter_names = 10;</code>
+   */
+  int getLoopEnterNamesCount();
+  /**
+   * <pre>
+   * List of names for enter tensors.
+   * </pre>
+   *
+   * <code>repeated string loop_enter_names = 10;</code>
+   */
+  java.lang.String getLoopEnterNames(int index);
+  /**
+   * <pre>
+   * List of names for enter tensors.
+   * </pre>
+   *
+   * <code>repeated string loop_enter_names = 10;</code>
+   */
+  com.google.protobuf.ByteString
+      getLoopEnterNamesBytes(int index);
+
+  /**
+   * <pre>
+   * Values and external values in control flow context.
+   * </pre>
+   *
+   * <code>.tensorflow.ValuesDef values_def = 9;</code>
+   */
+  boolean hasValuesDef();
+  /**
+   * <pre>
+   * Values and external values in control flow context.
+   * </pre>
+   *
+   * <code>.tensorflow.ValuesDef values_def = 9;</code>
+   */
+  org.tensorflow.proto.framework.ValuesDef getValuesDef();
+  /**
+   * <pre>
+   * Values and external values in control flow context.
+   * </pre>
+   *
+   * <code>.tensorflow.ValuesDef values_def = 9;</code>
+   */
+  org.tensorflow.proto.framework.ValuesDefOrBuilder getValuesDefOrBuilder();
+
+  /**
+   * <pre>
+   * Optional name of the maximum_iterations tensor.
+   * </pre>
+   *
+   * <code>string maximum_iterations_name = 11;</code>
+   */
+  java.lang.String getMaximumIterationsName();
+  /**
+   * <pre>
+   * Optional name of the maximum_iterations tensor.
+   * </pre>
+   *
+   * <code>string maximum_iterations_name = 11;</code>
+   */
+  com.google.protobuf.ByteString
+      getMaximumIterationsNameBytes();
+
+  /**
+   * <pre>
+   * Contexts contained inside this context (e.g. nested whiles).
+   * </pre>
+   *
+   * <code>repeated .tensorflow.ControlFlowContextDef nested_contexts = 12;</code>
+   */
+  java.util.List<org.tensorflow.proto.framework.ControlFlowContextDef> 
+      getNestedContextsList();
+  /**
+   * <pre>
+   * Contexts contained inside this context (e.g. nested whiles).
+   * </pre>
+   *
+   * <code>repeated .tensorflow.ControlFlowContextDef nested_contexts = 12;</code>
+   */
+  org.tensorflow.proto.framework.ControlFlowContextDef getNestedContexts(int index);
+  /**
+   * <pre>
+   * Contexts contained inside this context (e.g. nested whiles).
+   * </pre>
+   *
+   * <code>repeated .tensorflow.ControlFlowContextDef nested_contexts = 12;</code>
+   */
+  int getNestedContextsCount();
+  /**
+   * <pre>
+   * Contexts contained inside this context (e.g. nested whiles).
+   * </pre>
+   *
+   * <code>repeated .tensorflow.ControlFlowContextDef nested_contexts = 12;</code>
+   */
+  java.util.List<? extends org.tensorflow.proto.framework.ControlFlowContextDefOrBuilder> 
+      getNestedContextsOrBuilderList();
+  /**
+   * <pre>
+   * Contexts contained inside this context (e.g. nested whiles).
+   * </pre>
+   *
+   * <code>repeated .tensorflow.ControlFlowContextDef nested_contexts = 12;</code>
+   */
+  org.tensorflow.proto.framework.ControlFlowContextDefOrBuilder getNestedContextsOrBuilder(
+      int index);
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/ProfileOptions.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/ProfileOptions.java
new file mode 100644
index 00000000000..1a071369549
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/ProfileOptions.java
@@ -0,0 +1,1176 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/profiler/profiler_options.proto
+
+package org.tensorflow.proto.profiler;
+
+/**
+ * Protobuf type {@code tensorflow.ProfileOptions}
+ */
+public  final class ProfileOptions extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.ProfileOptions)
+    ProfileOptionsOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use ProfileOptions.newBuilder() to construct.
+  private ProfileOptions(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private ProfileOptions() {
+    deviceType_ = 0;
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new ProfileOptions();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private ProfileOptions(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 8: {
+
+            includeDatasetOps_ = input.readBool();
+            break;
+          }
+          case 16: {
+
+            hostTracerLevel_ = input.readUInt32();
+            break;
+          }
+          case 24: {
+
+            deviceTracerLevel_ = input.readUInt32();
+            break;
+          }
+          case 32: {
+
+            pythonTracerLevel_ = input.readUInt32();
+            break;
+          }
+          case 40: {
+
+            version_ = input.readUInt32();
+            break;
+          }
+          case 48: {
+            int rawValue = input.readEnum();
+
+            deviceType_ = rawValue;
+            break;
+          }
+          case 56: {
+
+            enableHloProto_ = input.readBool();
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.profiler.ProfilerOptionsProtos.internal_static_tensorflow_ProfileOptions_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.profiler.ProfilerOptionsProtos.internal_static_tensorflow_ProfileOptions_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.profiler.ProfileOptions.class, org.tensorflow.proto.profiler.ProfileOptions.Builder.class);
+  }
+
+  /**
+   * Protobuf enum {@code tensorflow.ProfileOptions.DeviceType}
+   */
+  public enum DeviceType
+      implements com.google.protobuf.ProtocolMessageEnum {
+    /**
+     * <code>UNSPECIFIED = 0;</code>
+     */
+    UNSPECIFIED(0),
+    /**
+     * <code>CPU = 1;</code>
+     */
+    CPU(1),
+    /**
+     * <code>GPU = 2;</code>
+     */
+    GPU(2),
+    /**
+     * <code>TPU = 3;</code>
+     */
+    TPU(3),
+    UNRECOGNIZED(-1),
+    ;
+
+    /**
+     * <code>UNSPECIFIED = 0;</code>
+     */
+    public static final int UNSPECIFIED_VALUE = 0;
+    /**
+     * <code>CPU = 1;</code>
+     */
+    public static final int CPU_VALUE = 1;
+    /**
+     * <code>GPU = 2;</code>
+     */
+    public static final int GPU_VALUE = 2;
+    /**
+     * <code>TPU = 3;</code>
+     */
+    public static final int TPU_VALUE = 3;
+
+
+    public final int getNumber() {
+      if (this == UNRECOGNIZED) {
+        throw new java.lang.IllegalArgumentException(
+            "Can't get the number of an unknown enum value.");
+      }
+      return value;
+    }
+
+    /**
+     * @deprecated Use {@link #forNumber(int)} instead.
+     */
+    @java.lang.Deprecated
+    public static DeviceType valueOf(int value) {
+      return forNumber(value);
+    }
+
+    public static DeviceType forNumber(int value) {
+      switch (value) {
+        case 0: return UNSPECIFIED;
+        case 1: return CPU;
+        case 2: return GPU;
+        case 3: return TPU;
+        default: return null;
+      }
+    }
+
+    public static com.google.protobuf.Internal.EnumLiteMap<DeviceType>
+        internalGetValueMap() {
+      return internalValueMap;
+    }
+    private static final com.google.protobuf.Internal.EnumLiteMap<
+        DeviceType> internalValueMap =
+          new com.google.protobuf.Internal.EnumLiteMap<DeviceType>() {
+            public DeviceType findValueByNumber(int number) {
+              return DeviceType.forNumber(number);
+            }
+          };
+
+    public final com.google.protobuf.Descriptors.EnumValueDescriptor
+        getValueDescriptor() {
+      return getDescriptor().getValues().get(ordinal());
+    }
+    public final com.google.protobuf.Descriptors.EnumDescriptor
+        getDescriptorForType() {
+      return getDescriptor();
+    }
+    public static final com.google.protobuf.Descriptors.EnumDescriptor
+        getDescriptor() {
+      return org.tensorflow.proto.profiler.ProfileOptions.getDescriptor().getEnumTypes().get(0);
+    }
+
+    private static final DeviceType[] VALUES = values();
+
+    public static DeviceType valueOf(
+        com.google.protobuf.Descriptors.EnumValueDescriptor desc) {
+      if (desc.getType() != getDescriptor()) {
+        throw new java.lang.IllegalArgumentException(
+          "EnumValueDescriptor is not for this type.");
+      }
+      if (desc.getIndex() == -1) {
+        return UNRECOGNIZED;
+      }
+      return VALUES[desc.getIndex()];
+    }
+
+    private final int value;
+
+    private DeviceType(int value) {
+      this.value = value;
+    }
+
+    // @@protoc_insertion_point(enum_scope:tensorflow.ProfileOptions.DeviceType)
+  }
+
+  public static final int VERSION_FIELD_NUMBER = 5;
+  private int version_;
+  /**
+   * <pre>
+   * Some default value of option are not proto3 default value. Use this version
+   * to determine if we should use default option value instead of proto3
+   * default value.
+   * </pre>
+   *
+   * <code>uint32 version = 5;</code>
+   */
+  public int getVersion() {
+    return version_;
+  }
+
+  public static final int DEVICE_TYPE_FIELD_NUMBER = 6;
+  private int deviceType_;
+  /**
+   * <pre>
+   * Device type to profile/trace: (version &gt;= 1)
+   * DeviceType::UNSPECIFIED: All registered device profiler will be enabled.
+   * DeviceType::CPU: only CPU will be profiled.
+   * DeviceType::GPU: only CPU/GPU will be profiled.
+   * DeviceType::TPU: only CPU/TPU will be profiled.
+   * </pre>
+   *
+   * <code>.tensorflow.ProfileOptions.DeviceType device_type = 6;</code>
+   */
+  public int getDeviceTypeValue() {
+    return deviceType_;
+  }
+  /**
+   * <pre>
+   * Device type to profile/trace: (version &gt;= 1)
+   * DeviceType::UNSPECIFIED: All registered device profiler will be enabled.
+   * DeviceType::CPU: only CPU will be profiled.
+   * DeviceType::GPU: only CPU/GPU will be profiled.
+   * DeviceType::TPU: only CPU/TPU will be profiled.
+   * </pre>
+   *
+   * <code>.tensorflow.ProfileOptions.DeviceType device_type = 6;</code>
+   */
+  public org.tensorflow.proto.profiler.ProfileOptions.DeviceType getDeviceType() {
+    @SuppressWarnings("deprecation")
+    org.tensorflow.proto.profiler.ProfileOptions.DeviceType result = org.tensorflow.proto.profiler.ProfileOptions.DeviceType.valueOf(deviceType_);
+    return result == null ? org.tensorflow.proto.profiler.ProfileOptions.DeviceType.UNRECOGNIZED : result;
+  }
+
+  public static final int INCLUDE_DATASET_OPS_FIELD_NUMBER = 1;
+  private boolean includeDatasetOps_;
+  /**
+   * <pre>
+   * We don't collect the dataset ops by default for better trace-viewer
+   * scalability. The caller can mannually set this field to include the ops.
+   * </pre>
+   *
+   * <code>bool include_dataset_ops = 1;</code>
+   */
+  public boolean getIncludeDatasetOps() {
+    return includeDatasetOps_;
+  }
+
+  public static final int HOST_TRACER_LEVEL_FIELD_NUMBER = 2;
+  private int hostTracerLevel_;
+  /**
+   * <pre>
+   * Levels of host tracing: (version &gt;= 1)
+   * - Level 0 is used to disable host traces.
+   * - Level 1 enables tracing of only user instrumented (or default) TraceMe.
+   * - Level 2 enables tracing of all level 1 TraceMe(s) and instrumented high
+   *           level program execution details (expensive TF ops, XLA ops, etc).
+   *           This is the default.
+   * - Level 3 enables tracing of all level 2 TraceMe(s) and more verbose
+   *           (low-level) program execution details (cheap TF ops, etc).
+   * </pre>
+   *
+   * <code>uint32 host_tracer_level = 2;</code>
+   */
+  public int getHostTracerLevel() {
+    return hostTracerLevel_;
+  }
+
+  public static final int DEVICE_TRACER_LEVEL_FIELD_NUMBER = 3;
+  private int deviceTracerLevel_;
+  /**
+   * <pre>
+   * Levels of device tracing: (version &gt;= 1)
+   * - Level 0 is used to disable device traces.
+   * - Level 1 is used to enable device traces.
+   * - More levels might be defined for specific device for controlling the
+   *   verbosity of the trace.
+   * </pre>
+   *
+   * <code>uint32 device_tracer_level = 3;</code>
+   */
+  public int getDeviceTracerLevel() {
+    return deviceTracerLevel_;
+  }
+
+  public static final int PYTHON_TRACER_LEVEL_FIELD_NUMBER = 4;
+  private int pythonTracerLevel_;
+  /**
+   * <pre>
+   * Whether enable python function calls tracing. Runtime overhead ensues if
+   * enabled. Default off. (version &gt;= 1)
+   * </pre>
+   *
+   * <code>uint32 python_tracer_level = 4;</code>
+   */
+  public int getPythonTracerLevel() {
+    return pythonTracerLevel_;
+  }
+
+  public static final int ENABLE_HLO_PROTO_FIELD_NUMBER = 7;
+  private boolean enableHloProto_;
+  /**
+   * <pre>
+   * Whether serialize hlo_proto when XLA is used. (version &gt;= 1)
+   * </pre>
+   *
+   * <code>bool enable_hlo_proto = 7;</code>
+   */
+  public boolean getEnableHloProto() {
+    return enableHloProto_;
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    if (includeDatasetOps_ != false) {
+      output.writeBool(1, includeDatasetOps_);
+    }
+    if (hostTracerLevel_ != 0) {
+      output.writeUInt32(2, hostTracerLevel_);
+    }
+    if (deviceTracerLevel_ != 0) {
+      output.writeUInt32(3, deviceTracerLevel_);
+    }
+    if (pythonTracerLevel_ != 0) {
+      output.writeUInt32(4, pythonTracerLevel_);
+    }
+    if (version_ != 0) {
+      output.writeUInt32(5, version_);
+    }
+    if (deviceType_ != org.tensorflow.proto.profiler.ProfileOptions.DeviceType.UNSPECIFIED.getNumber()) {
+      output.writeEnum(6, deviceType_);
+    }
+    if (enableHloProto_ != false) {
+      output.writeBool(7, enableHloProto_);
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (includeDatasetOps_ != false) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeBoolSize(1, includeDatasetOps_);
+    }
+    if (hostTracerLevel_ != 0) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeUInt32Size(2, hostTracerLevel_);
+    }
+    if (deviceTracerLevel_ != 0) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeUInt32Size(3, deviceTracerLevel_);
+    }
+    if (pythonTracerLevel_ != 0) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeUInt32Size(4, pythonTracerLevel_);
+    }
+    if (version_ != 0) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeUInt32Size(5, version_);
+    }
+    if (deviceType_ != org.tensorflow.proto.profiler.ProfileOptions.DeviceType.UNSPECIFIED.getNumber()) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeEnumSize(6, deviceType_);
+    }
+    if (enableHloProto_ != false) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeBoolSize(7, enableHloProto_);
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.profiler.ProfileOptions)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.profiler.ProfileOptions other = (org.tensorflow.proto.profiler.ProfileOptions) obj;
+
+    if (getVersion()
+        != other.getVersion()) return false;
+    if (deviceType_ != other.deviceType_) return false;
+    if (getIncludeDatasetOps()
+        != other.getIncludeDatasetOps()) return false;
+    if (getHostTracerLevel()
+        != other.getHostTracerLevel()) return false;
+    if (getDeviceTracerLevel()
+        != other.getDeviceTracerLevel()) return false;
+    if (getPythonTracerLevel()
+        != other.getPythonTracerLevel()) return false;
+    if (getEnableHloProto()
+        != other.getEnableHloProto()) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    hash = (37 * hash) + VERSION_FIELD_NUMBER;
+    hash = (53 * hash) + getVersion();
+    hash = (37 * hash) + DEVICE_TYPE_FIELD_NUMBER;
+    hash = (53 * hash) + deviceType_;
+    hash = (37 * hash) + INCLUDE_DATASET_OPS_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashBoolean(
+        getIncludeDatasetOps());
+    hash = (37 * hash) + HOST_TRACER_LEVEL_FIELD_NUMBER;
+    hash = (53 * hash) + getHostTracerLevel();
+    hash = (37 * hash) + DEVICE_TRACER_LEVEL_FIELD_NUMBER;
+    hash = (53 * hash) + getDeviceTracerLevel();
+    hash = (37 * hash) + PYTHON_TRACER_LEVEL_FIELD_NUMBER;
+    hash = (53 * hash) + getPythonTracerLevel();
+    hash = (37 * hash) + ENABLE_HLO_PROTO_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashBoolean(
+        getEnableHloProto());
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.profiler.ProfileOptions parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.profiler.ProfileOptions parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.profiler.ProfileOptions parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.profiler.ProfileOptions parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.profiler.ProfileOptions parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.profiler.ProfileOptions parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.profiler.ProfileOptions parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.profiler.ProfileOptions parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.profiler.ProfileOptions parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.profiler.ProfileOptions parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.profiler.ProfileOptions parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.profiler.ProfileOptions parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.profiler.ProfileOptions prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * Protobuf type {@code tensorflow.ProfileOptions}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.ProfileOptions)
+      org.tensorflow.proto.profiler.ProfileOptionsOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.profiler.ProfilerOptionsProtos.internal_static_tensorflow_ProfileOptions_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.profiler.ProfilerOptionsProtos.internal_static_tensorflow_ProfileOptions_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.profiler.ProfileOptions.class, org.tensorflow.proto.profiler.ProfileOptions.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.profiler.ProfileOptions.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      version_ = 0;
+
+      deviceType_ = 0;
+
+      includeDatasetOps_ = false;
+
+      hostTracerLevel_ = 0;
+
+      deviceTracerLevel_ = 0;
+
+      pythonTracerLevel_ = 0;
+
+      enableHloProto_ = false;
+
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.profiler.ProfilerOptionsProtos.internal_static_tensorflow_ProfileOptions_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.profiler.ProfileOptions getDefaultInstanceForType() {
+      return org.tensorflow.proto.profiler.ProfileOptions.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.profiler.ProfileOptions build() {
+      org.tensorflow.proto.profiler.ProfileOptions result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.profiler.ProfileOptions buildPartial() {
+      org.tensorflow.proto.profiler.ProfileOptions result = new org.tensorflow.proto.profiler.ProfileOptions(this);
+      result.version_ = version_;
+      result.deviceType_ = deviceType_;
+      result.includeDatasetOps_ = includeDatasetOps_;
+      result.hostTracerLevel_ = hostTracerLevel_;
+      result.deviceTracerLevel_ = deviceTracerLevel_;
+      result.pythonTracerLevel_ = pythonTracerLevel_;
+      result.enableHloProto_ = enableHloProto_;
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.profiler.ProfileOptions) {
+        return mergeFrom((org.tensorflow.proto.profiler.ProfileOptions)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.profiler.ProfileOptions other) {
+      if (other == org.tensorflow.proto.profiler.ProfileOptions.getDefaultInstance()) return this;
+      if (other.getVersion() != 0) {
+        setVersion(other.getVersion());
+      }
+      if (other.deviceType_ != 0) {
+        setDeviceTypeValue(other.getDeviceTypeValue());
+      }
+      if (other.getIncludeDatasetOps() != false) {
+        setIncludeDatasetOps(other.getIncludeDatasetOps());
+      }
+      if (other.getHostTracerLevel() != 0) {
+        setHostTracerLevel(other.getHostTracerLevel());
+      }
+      if (other.getDeviceTracerLevel() != 0) {
+        setDeviceTracerLevel(other.getDeviceTracerLevel());
+      }
+      if (other.getPythonTracerLevel() != 0) {
+        setPythonTracerLevel(other.getPythonTracerLevel());
+      }
+      if (other.getEnableHloProto() != false) {
+        setEnableHloProto(other.getEnableHloProto());
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.profiler.ProfileOptions parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.profiler.ProfileOptions) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+
+    private int version_ ;
+    /**
+     * <pre>
+     * Some default value of option are not proto3 default value. Use this version
+     * to determine if we should use default option value instead of proto3
+     * default value.
+     * </pre>
+     *
+     * <code>uint32 version = 5;</code>
+     */
+    public int getVersion() {
+      return version_;
+    }
+    /**
+     * <pre>
+     * Some default value of option are not proto3 default value. Use this version
+     * to determine if we should use default option value instead of proto3
+     * default value.
+     * </pre>
+     *
+     * <code>uint32 version = 5;</code>
+     */
+    public Builder setVersion(int value) {
+      
+      version_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Some default value of option are not proto3 default value. Use this version
+     * to determine if we should use default option value instead of proto3
+     * default value.
+     * </pre>
+     *
+     * <code>uint32 version = 5;</code>
+     */
+    public Builder clearVersion() {
+      
+      version_ = 0;
+      onChanged();
+      return this;
+    }
+
+    private int deviceType_ = 0;
+    /**
+     * <pre>
+     * Device type to profile/trace: (version &gt;= 1)
+     * DeviceType::UNSPECIFIED: All registered device profiler will be enabled.
+     * DeviceType::CPU: only CPU will be profiled.
+     * DeviceType::GPU: only CPU/GPU will be profiled.
+     * DeviceType::TPU: only CPU/TPU will be profiled.
+     * </pre>
+     *
+     * <code>.tensorflow.ProfileOptions.DeviceType device_type = 6;</code>
+     */
+    public int getDeviceTypeValue() {
+      return deviceType_;
+    }
+    /**
+     * <pre>
+     * Device type to profile/trace: (version &gt;= 1)
+     * DeviceType::UNSPECIFIED: All registered device profiler will be enabled.
+     * DeviceType::CPU: only CPU will be profiled.
+     * DeviceType::GPU: only CPU/GPU will be profiled.
+     * DeviceType::TPU: only CPU/TPU will be profiled.
+     * </pre>
+     *
+     * <code>.tensorflow.ProfileOptions.DeviceType device_type = 6;</code>
+     */
+    public Builder setDeviceTypeValue(int value) {
+      deviceType_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Device type to profile/trace: (version &gt;= 1)
+     * DeviceType::UNSPECIFIED: All registered device profiler will be enabled.
+     * DeviceType::CPU: only CPU will be profiled.
+     * DeviceType::GPU: only CPU/GPU will be profiled.
+     * DeviceType::TPU: only CPU/TPU will be profiled.
+     * </pre>
+     *
+     * <code>.tensorflow.ProfileOptions.DeviceType device_type = 6;</code>
+     */
+    public org.tensorflow.proto.profiler.ProfileOptions.DeviceType getDeviceType() {
+      @SuppressWarnings("deprecation")
+      org.tensorflow.proto.profiler.ProfileOptions.DeviceType result = org.tensorflow.proto.profiler.ProfileOptions.DeviceType.valueOf(deviceType_);
+      return result == null ? org.tensorflow.proto.profiler.ProfileOptions.DeviceType.UNRECOGNIZED : result;
+    }
+    /**
+     * <pre>
+     * Device type to profile/trace: (version &gt;= 1)
+     * DeviceType::UNSPECIFIED: All registered device profiler will be enabled.
+     * DeviceType::CPU: only CPU will be profiled.
+     * DeviceType::GPU: only CPU/GPU will be profiled.
+     * DeviceType::TPU: only CPU/TPU will be profiled.
+     * </pre>
+     *
+     * <code>.tensorflow.ProfileOptions.DeviceType device_type = 6;</code>
+     */
+    public Builder setDeviceType(org.tensorflow.proto.profiler.ProfileOptions.DeviceType value) {
+      if (value == null) {
+        throw new NullPointerException();
+      }
+      
+      deviceType_ = value.getNumber();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Device type to profile/trace: (version &gt;= 1)
+     * DeviceType::UNSPECIFIED: All registered device profiler will be enabled.
+     * DeviceType::CPU: only CPU will be profiled.
+     * DeviceType::GPU: only CPU/GPU will be profiled.
+     * DeviceType::TPU: only CPU/TPU will be profiled.
+     * </pre>
+     *
+     * <code>.tensorflow.ProfileOptions.DeviceType device_type = 6;</code>
+     */
+    public Builder clearDeviceType() {
+      
+      deviceType_ = 0;
+      onChanged();
+      return this;
+    }
+
+    private boolean includeDatasetOps_ ;
+    /**
+     * <pre>
+     * We don't collect the dataset ops by default for better trace-viewer
+     * scalability. The caller can mannually set this field to include the ops.
+     * </pre>
+     *
+     * <code>bool include_dataset_ops = 1;</code>
+     */
+    public boolean getIncludeDatasetOps() {
+      return includeDatasetOps_;
+    }
+    /**
+     * <pre>
+     * We don't collect the dataset ops by default for better trace-viewer
+     * scalability. The caller can mannually set this field to include the ops.
+     * </pre>
+     *
+     * <code>bool include_dataset_ops = 1;</code>
+     */
+    public Builder setIncludeDatasetOps(boolean value) {
+      
+      includeDatasetOps_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * We don't collect the dataset ops by default for better trace-viewer
+     * scalability. The caller can mannually set this field to include the ops.
+     * </pre>
+     *
+     * <code>bool include_dataset_ops = 1;</code>
+     */
+    public Builder clearIncludeDatasetOps() {
+      
+      includeDatasetOps_ = false;
+      onChanged();
+      return this;
+    }
+
+    private int hostTracerLevel_ ;
+    /**
+     * <pre>
+     * Levels of host tracing: (version &gt;= 1)
+     * - Level 0 is used to disable host traces.
+     * - Level 1 enables tracing of only user instrumented (or default) TraceMe.
+     * - Level 2 enables tracing of all level 1 TraceMe(s) and instrumented high
+     *           level program execution details (expensive TF ops, XLA ops, etc).
+     *           This is the default.
+     * - Level 3 enables tracing of all level 2 TraceMe(s) and more verbose
+     *           (low-level) program execution details (cheap TF ops, etc).
+     * </pre>
+     *
+     * <code>uint32 host_tracer_level = 2;</code>
+     */
+    public int getHostTracerLevel() {
+      return hostTracerLevel_;
+    }
+    /**
+     * <pre>
+     * Levels of host tracing: (version &gt;= 1)
+     * - Level 0 is used to disable host traces.
+     * - Level 1 enables tracing of only user instrumented (or default) TraceMe.
+     * - Level 2 enables tracing of all level 1 TraceMe(s) and instrumented high
+     *           level program execution details (expensive TF ops, XLA ops, etc).
+     *           This is the default.
+     * - Level 3 enables tracing of all level 2 TraceMe(s) and more verbose
+     *           (low-level) program execution details (cheap TF ops, etc).
+     * </pre>
+     *
+     * <code>uint32 host_tracer_level = 2;</code>
+     */
+    public Builder setHostTracerLevel(int value) {
+      
+      hostTracerLevel_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Levels of host tracing: (version &gt;= 1)
+     * - Level 0 is used to disable host traces.
+     * - Level 1 enables tracing of only user instrumented (or default) TraceMe.
+     * - Level 2 enables tracing of all level 1 TraceMe(s) and instrumented high
+     *           level program execution details (expensive TF ops, XLA ops, etc).
+     *           This is the default.
+     * - Level 3 enables tracing of all level 2 TraceMe(s) and more verbose
+     *           (low-level) program execution details (cheap TF ops, etc).
+     * </pre>
+     *
+     * <code>uint32 host_tracer_level = 2;</code>
+     */
+    public Builder clearHostTracerLevel() {
+      
+      hostTracerLevel_ = 0;
+      onChanged();
+      return this;
+    }
+
+    private int deviceTracerLevel_ ;
+    /**
+     * <pre>
+     * Levels of device tracing: (version &gt;= 1)
+     * - Level 0 is used to disable device traces.
+     * - Level 1 is used to enable device traces.
+     * - More levels might be defined for specific device for controlling the
+     *   verbosity of the trace.
+     * </pre>
+     *
+     * <code>uint32 device_tracer_level = 3;</code>
+     */
+    public int getDeviceTracerLevel() {
+      return deviceTracerLevel_;
+    }
+    /**
+     * <pre>
+     * Levels of device tracing: (version &gt;= 1)
+     * - Level 0 is used to disable device traces.
+     * - Level 1 is used to enable device traces.
+     * - More levels might be defined for specific device for controlling the
+     *   verbosity of the trace.
+     * </pre>
+     *
+     * <code>uint32 device_tracer_level = 3;</code>
+     */
+    public Builder setDeviceTracerLevel(int value) {
+      
+      deviceTracerLevel_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Levels of device tracing: (version &gt;= 1)
+     * - Level 0 is used to disable device traces.
+     * - Level 1 is used to enable device traces.
+     * - More levels might be defined for specific device for controlling the
+     *   verbosity of the trace.
+     * </pre>
+     *
+     * <code>uint32 device_tracer_level = 3;</code>
+     */
+    public Builder clearDeviceTracerLevel() {
+      
+      deviceTracerLevel_ = 0;
+      onChanged();
+      return this;
+    }
+
+    private int pythonTracerLevel_ ;
+    /**
+     * <pre>
+     * Whether enable python function calls tracing. Runtime overhead ensues if
+     * enabled. Default off. (version &gt;= 1)
+     * </pre>
+     *
+     * <code>uint32 python_tracer_level = 4;</code>
+     */
+    public int getPythonTracerLevel() {
+      return pythonTracerLevel_;
+    }
+    /**
+     * <pre>
+     * Whether enable python function calls tracing. Runtime overhead ensues if
+     * enabled. Default off. (version &gt;= 1)
+     * </pre>
+     *
+     * <code>uint32 python_tracer_level = 4;</code>
+     */
+    public Builder setPythonTracerLevel(int value) {
+      
+      pythonTracerLevel_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Whether enable python function calls tracing. Runtime overhead ensues if
+     * enabled. Default off. (version &gt;= 1)
+     * </pre>
+     *
+     * <code>uint32 python_tracer_level = 4;</code>
+     */
+    public Builder clearPythonTracerLevel() {
+      
+      pythonTracerLevel_ = 0;
+      onChanged();
+      return this;
+    }
+
+    private boolean enableHloProto_ ;
+    /**
+     * <pre>
+     * Whether serialize hlo_proto when XLA is used. (version &gt;= 1)
+     * </pre>
+     *
+     * <code>bool enable_hlo_proto = 7;</code>
+     */
+    public boolean getEnableHloProto() {
+      return enableHloProto_;
+    }
+    /**
+     * <pre>
+     * Whether serialize hlo_proto when XLA is used. (version &gt;= 1)
+     * </pre>
+     *
+     * <code>bool enable_hlo_proto = 7;</code>
+     */
+    public Builder setEnableHloProto(boolean value) {
+      
+      enableHloProto_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Whether serialize hlo_proto when XLA is used. (version &gt;= 1)
+     * </pre>
+     *
+     * <code>bool enable_hlo_proto = 7;</code>
+     */
+    public Builder clearEnableHloProto() {
+      
+      enableHloProto_ = false;
+      onChanged();
+      return this;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.ProfileOptions)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.ProfileOptions)
+  private static final org.tensorflow.proto.profiler.ProfileOptions DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.profiler.ProfileOptions();
+  }
+
+  public static org.tensorflow.proto.profiler.ProfileOptions getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<ProfileOptions>
+      PARSER = new com.google.protobuf.AbstractParser<ProfileOptions>() {
+    @java.lang.Override
+    public ProfileOptions parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new ProfileOptions(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<ProfileOptions> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<ProfileOptions> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.profiler.ProfileOptions getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/ProfileOptionsOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/ProfileOptionsOrBuilder.java
new file mode 100644
index 00000000000..e508bb52c79
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/ProfileOptionsOrBuilder.java
@@ -0,0 +1,103 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/profiler/profiler_options.proto
+
+package org.tensorflow.proto.profiler;
+
+public interface ProfileOptionsOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.ProfileOptions)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <pre>
+   * Some default value of option are not proto3 default value. Use this version
+   * to determine if we should use default option value instead of proto3
+   * default value.
+   * </pre>
+   *
+   * <code>uint32 version = 5;</code>
+   */
+  int getVersion();
+
+  /**
+   * <pre>
+   * Device type to profile/trace: (version &gt;= 1)
+   * DeviceType::UNSPECIFIED: All registered device profiler will be enabled.
+   * DeviceType::CPU: only CPU will be profiled.
+   * DeviceType::GPU: only CPU/GPU will be profiled.
+   * DeviceType::TPU: only CPU/TPU will be profiled.
+   * </pre>
+   *
+   * <code>.tensorflow.ProfileOptions.DeviceType device_type = 6;</code>
+   */
+  int getDeviceTypeValue();
+  /**
+   * <pre>
+   * Device type to profile/trace: (version &gt;= 1)
+   * DeviceType::UNSPECIFIED: All registered device profiler will be enabled.
+   * DeviceType::CPU: only CPU will be profiled.
+   * DeviceType::GPU: only CPU/GPU will be profiled.
+   * DeviceType::TPU: only CPU/TPU will be profiled.
+   * </pre>
+   *
+   * <code>.tensorflow.ProfileOptions.DeviceType device_type = 6;</code>
+   */
+  org.tensorflow.proto.profiler.ProfileOptions.DeviceType getDeviceType();
+
+  /**
+   * <pre>
+   * We don't collect the dataset ops by default for better trace-viewer
+   * scalability. The caller can mannually set this field to include the ops.
+   * </pre>
+   *
+   * <code>bool include_dataset_ops = 1;</code>
+   */
+  boolean getIncludeDatasetOps();
+
+  /**
+   * <pre>
+   * Levels of host tracing: (version &gt;= 1)
+   * - Level 0 is used to disable host traces.
+   * - Level 1 enables tracing of only user instrumented (or default) TraceMe.
+   * - Level 2 enables tracing of all level 1 TraceMe(s) and instrumented high
+   *           level program execution details (expensive TF ops, XLA ops, etc).
+   *           This is the default.
+   * - Level 3 enables tracing of all level 2 TraceMe(s) and more verbose
+   *           (low-level) program execution details (cheap TF ops, etc).
+   * </pre>
+   *
+   * <code>uint32 host_tracer_level = 2;</code>
+   */
+  int getHostTracerLevel();
+
+  /**
+   * <pre>
+   * Levels of device tracing: (version &gt;= 1)
+   * - Level 0 is used to disable device traces.
+   * - Level 1 is used to enable device traces.
+   * - More levels might be defined for specific device for controlling the
+   *   verbosity of the trace.
+   * </pre>
+   *
+   * <code>uint32 device_tracer_level = 3;</code>
+   */
+  int getDeviceTracerLevel();
+
+  /**
+   * <pre>
+   * Whether enable python function calls tracing. Runtime overhead ensues if
+   * enabled. Default off. (version &gt;= 1)
+   * </pre>
+   *
+   * <code>uint32 python_tracer_level = 4;</code>
+   */
+  int getPythonTracerLevel();
+
+  /**
+   * <pre>
+   * Whether serialize hlo_proto when XLA is used. (version &gt;= 1)
+   * </pre>
+   *
+   * <code>bool enable_hlo_proto = 7;</code>
+   */
+  boolean getEnableHloProto();
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/ProfilerOptionsProtos.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/ProfilerOptionsProtos.java
new file mode 100644
index 00000000000..669d1ee04b5
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/ProfilerOptionsProtos.java
@@ -0,0 +1,56 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/profiler/profiler_options.proto
+
+package org.tensorflow.proto.profiler;
+
+public final class ProfilerOptionsProtos {
+  private ProfilerOptionsProtos() {}
+  public static void registerAllExtensions(
+      com.google.protobuf.ExtensionRegistryLite registry) {
+  }
+
+  public static void registerAllExtensions(
+      com.google.protobuf.ExtensionRegistry registry) {
+    registerAllExtensions(
+        (com.google.protobuf.ExtensionRegistryLite) registry);
+  }
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_ProfileOptions_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_ProfileOptions_fieldAccessorTable;
+
+  public static com.google.protobuf.Descriptors.FileDescriptor
+      getDescriptor() {
+    return descriptor;
+  }
+  private static  com.google.protobuf.Descriptors.FileDescriptor
+      descriptor;
+  static {
+    java.lang.String[] descriptorData = {
+      "\n/tensorflow/core/profiler/profiler_opti" +
+      "ons.proto\022\ntensorflow\"\243\002\n\016ProfileOptions" +
+      "\022\017\n\007version\030\005 \001(\r\022:\n\013device_type\030\006 \001(\0162%" +
+      ".tensorflow.ProfileOptions.DeviceType\022\033\n" +
+      "\023include_dataset_ops\030\001 \001(\010\022\031\n\021host_trace" +
+      "r_level\030\002 \001(\r\022\033\n\023device_tracer_level\030\003 \001" +
+      "(\r\022\033\n\023python_tracer_level\030\004 \001(\r\022\030\n\020enabl" +
+      "e_hlo_proto\030\007 \001(\010\"8\n\nDeviceType\022\017\n\013UNSPE" +
+      "CIFIED\020\000\022\007\n\003CPU\020\001\022\007\n\003GPU\020\002\022\007\n\003TPU\020\003B8\n\035o" +
+      "rg.tensorflow.proto.profilerB\025ProfilerOp" +
+      "tionsProtosP\001b\006proto3"
+    };
+    descriptor = com.google.protobuf.Descriptors.FileDescriptor
+      .internalBuildGeneratedFileFrom(descriptorData,
+        new com.google.protobuf.Descriptors.FileDescriptor[] {
+        });
+    internal_static_tensorflow_ProfileOptions_descriptor =
+      getDescriptor().getMessageTypes().get(0);
+    internal_static_tensorflow_ProfileOptions_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_ProfileOptions_descriptor,
+        new java.lang.String[] { "Version", "DeviceType", "IncludeDatasetOps", "HostTracerLevel", "DeviceTracerLevel", "PythonTracerLevel", "EnableHloProto", });
+  }
+
+  // @@protoc_insertion_point(outer_class_scope)
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/XEvent.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/XEvent.java
new file mode 100644
index 00000000000..2921881fee2
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/XEvent.java
@@ -0,0 +1,1263 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/profiler/protobuf/xplane.proto
+
+package org.tensorflow.proto.profiler;
+
+/**
+ * <pre>
+ * An XEvent is a trace event, optionally annotated with XStats.
+ * Next ID: 6
+ * </pre>
+ *
+ * Protobuf type {@code tensorflow.profiler.XEvent}
+ */
+public  final class XEvent extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.profiler.XEvent)
+    XEventOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use XEvent.newBuilder() to construct.
+  private XEvent(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private XEvent() {
+    stats_ = java.util.Collections.emptyList();
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new XEvent();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private XEvent(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    int mutable_bitField0_ = 0;
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 8: {
+
+            metadataId_ = input.readInt64();
+            break;
+          }
+          case 16: {
+            dataCase_ = 2;
+            data_ = input.readInt64();
+            break;
+          }
+          case 24: {
+
+            durationPs_ = input.readInt64();
+            break;
+          }
+          case 34: {
+            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
+              stats_ = new java.util.ArrayList<org.tensorflow.proto.profiler.XStat>();
+              mutable_bitField0_ |= 0x00000001;
+            }
+            stats_.add(
+                input.readMessage(org.tensorflow.proto.profiler.XStat.parser(), extensionRegistry));
+            break;
+          }
+          case 40: {
+            dataCase_ = 5;
+            data_ = input.readInt64();
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      if (((mutable_bitField0_ & 0x00000001) != 0)) {
+        stats_ = java.util.Collections.unmodifiableList(stats_);
+      }
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.profiler.XPlaneProtos.internal_static_tensorflow_profiler_XEvent_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.profiler.XPlaneProtos.internal_static_tensorflow_profiler_XEvent_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.profiler.XEvent.class, org.tensorflow.proto.profiler.XEvent.Builder.class);
+  }
+
+  private int dataCase_ = 0;
+  private java.lang.Object data_;
+  public enum DataCase
+      implements com.google.protobuf.Internal.EnumLite {
+    OFFSET_PS(2),
+    NUM_OCCURRENCES(5),
+    DATA_NOT_SET(0);
+    private final int value;
+    private DataCase(int value) {
+      this.value = value;
+    }
+    /**
+     * @deprecated Use {@link #forNumber(int)} instead.
+     */
+    @java.lang.Deprecated
+    public static DataCase valueOf(int value) {
+      return forNumber(value);
+    }
+
+    public static DataCase forNumber(int value) {
+      switch (value) {
+        case 2: return OFFSET_PS;
+        case 5: return NUM_OCCURRENCES;
+        case 0: return DATA_NOT_SET;
+        default: return null;
+      }
+    }
+    public int getNumber() {
+      return this.value;
+    }
+  };
+
+  public DataCase
+  getDataCase() {
+    return DataCase.forNumber(
+        dataCase_);
+  }
+
+  public static final int METADATA_ID_FIELD_NUMBER = 1;
+  private long metadataId_;
+  /**
+   * <pre>
+   * XEventMetadata.id of corresponding metadata.
+   * </pre>
+   *
+   * <code>int64 metadata_id = 1;</code>
+   */
+  public long getMetadataId() {
+    return metadataId_;
+  }
+
+  public static final int OFFSET_PS_FIELD_NUMBER = 2;
+  /**
+   * <pre>
+   * Start time of the event in picoseconds, as offset from
+   * XLine.timestamp_ns().
+   * </pre>
+   *
+   * <code>int64 offset_ps = 2;</code>
+   */
+  public long getOffsetPs() {
+    if (dataCase_ == 2) {
+      return (java.lang.Long) data_;
+    }
+    return 0L;
+  }
+
+  public static final int NUM_OCCURRENCES_FIELD_NUMBER = 5;
+  /**
+   * <pre>
+   * Number of occurrences of the event, if aggregated.
+   * </pre>
+   *
+   * <code>int64 num_occurrences = 5;</code>
+   */
+  public long getNumOccurrences() {
+    if (dataCase_ == 5) {
+      return (java.lang.Long) data_;
+    }
+    return 0L;
+  }
+
+  public static final int DURATION_PS_FIELD_NUMBER = 3;
+  private long durationPs_;
+  /**
+   * <pre>
+   * Duration of the event in picoseconds. Can be zero for an instant event.
+   * </pre>
+   *
+   * <code>int64 duration_ps = 3;</code>
+   */
+  public long getDurationPs() {
+    return durationPs_;
+  }
+
+  public static final int STATS_FIELD_NUMBER = 4;
+  private java.util.List<org.tensorflow.proto.profiler.XStat> stats_;
+  /**
+   * <pre>
+   * XStats associated with the event.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.profiler.XStat stats = 4;</code>
+   */
+  public java.util.List<org.tensorflow.proto.profiler.XStat> getStatsList() {
+    return stats_;
+  }
+  /**
+   * <pre>
+   * XStats associated with the event.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.profiler.XStat stats = 4;</code>
+   */
+  public java.util.List<? extends org.tensorflow.proto.profiler.XStatOrBuilder> 
+      getStatsOrBuilderList() {
+    return stats_;
+  }
+  /**
+   * <pre>
+   * XStats associated with the event.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.profiler.XStat stats = 4;</code>
+   */
+  public int getStatsCount() {
+    return stats_.size();
+  }
+  /**
+   * <pre>
+   * XStats associated with the event.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.profiler.XStat stats = 4;</code>
+   */
+  public org.tensorflow.proto.profiler.XStat getStats(int index) {
+    return stats_.get(index);
+  }
+  /**
+   * <pre>
+   * XStats associated with the event.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.profiler.XStat stats = 4;</code>
+   */
+  public org.tensorflow.proto.profiler.XStatOrBuilder getStatsOrBuilder(
+      int index) {
+    return stats_.get(index);
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    if (metadataId_ != 0L) {
+      output.writeInt64(1, metadataId_);
+    }
+    if (dataCase_ == 2) {
+      output.writeInt64(
+          2, (long)((java.lang.Long) data_));
+    }
+    if (durationPs_ != 0L) {
+      output.writeInt64(3, durationPs_);
+    }
+    for (int i = 0; i < stats_.size(); i++) {
+      output.writeMessage(4, stats_.get(i));
+    }
+    if (dataCase_ == 5) {
+      output.writeInt64(
+          5, (long)((java.lang.Long) data_));
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (metadataId_ != 0L) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt64Size(1, metadataId_);
+    }
+    if (dataCase_ == 2) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt64Size(
+            2, (long)((java.lang.Long) data_));
+    }
+    if (durationPs_ != 0L) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt64Size(3, durationPs_);
+    }
+    for (int i = 0; i < stats_.size(); i++) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(4, stats_.get(i));
+    }
+    if (dataCase_ == 5) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt64Size(
+            5, (long)((java.lang.Long) data_));
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.profiler.XEvent)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.profiler.XEvent other = (org.tensorflow.proto.profiler.XEvent) obj;
+
+    if (getMetadataId()
+        != other.getMetadataId()) return false;
+    if (getDurationPs()
+        != other.getDurationPs()) return false;
+    if (!getStatsList()
+        .equals(other.getStatsList())) return false;
+    if (!getDataCase().equals(other.getDataCase())) return false;
+    switch (dataCase_) {
+      case 2:
+        if (getOffsetPs()
+            != other.getOffsetPs()) return false;
+        break;
+      case 5:
+        if (getNumOccurrences()
+            != other.getNumOccurrences()) return false;
+        break;
+      case 0:
+      default:
+    }
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    hash = (37 * hash) + METADATA_ID_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+        getMetadataId());
+    hash = (37 * hash) + DURATION_PS_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+        getDurationPs());
+    if (getStatsCount() > 0) {
+      hash = (37 * hash) + STATS_FIELD_NUMBER;
+      hash = (53 * hash) + getStatsList().hashCode();
+    }
+    switch (dataCase_) {
+      case 2:
+        hash = (37 * hash) + OFFSET_PS_FIELD_NUMBER;
+        hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+            getOffsetPs());
+        break;
+      case 5:
+        hash = (37 * hash) + NUM_OCCURRENCES_FIELD_NUMBER;
+        hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+            getNumOccurrences());
+        break;
+      case 0:
+      default:
+    }
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.profiler.XEvent parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.profiler.XEvent parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.profiler.XEvent parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.profiler.XEvent parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.profiler.XEvent parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.profiler.XEvent parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.profiler.XEvent parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.profiler.XEvent parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.profiler.XEvent parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.profiler.XEvent parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.profiler.XEvent parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.profiler.XEvent parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.profiler.XEvent prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * <pre>
+   * An XEvent is a trace event, optionally annotated with XStats.
+   * Next ID: 6
+   * </pre>
+   *
+   * Protobuf type {@code tensorflow.profiler.XEvent}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.profiler.XEvent)
+      org.tensorflow.proto.profiler.XEventOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.profiler.XPlaneProtos.internal_static_tensorflow_profiler_XEvent_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.profiler.XPlaneProtos.internal_static_tensorflow_profiler_XEvent_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.profiler.XEvent.class, org.tensorflow.proto.profiler.XEvent.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.profiler.XEvent.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+        getStatsFieldBuilder();
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      metadataId_ = 0L;
+
+      durationPs_ = 0L;
+
+      if (statsBuilder_ == null) {
+        stats_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000001);
+      } else {
+        statsBuilder_.clear();
+      }
+      dataCase_ = 0;
+      data_ = null;
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.profiler.XPlaneProtos.internal_static_tensorflow_profiler_XEvent_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.profiler.XEvent getDefaultInstanceForType() {
+      return org.tensorflow.proto.profiler.XEvent.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.profiler.XEvent build() {
+      org.tensorflow.proto.profiler.XEvent result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.profiler.XEvent buildPartial() {
+      org.tensorflow.proto.profiler.XEvent result = new org.tensorflow.proto.profiler.XEvent(this);
+      int from_bitField0_ = bitField0_;
+      result.metadataId_ = metadataId_;
+      if (dataCase_ == 2) {
+        result.data_ = data_;
+      }
+      if (dataCase_ == 5) {
+        result.data_ = data_;
+      }
+      result.durationPs_ = durationPs_;
+      if (statsBuilder_ == null) {
+        if (((bitField0_ & 0x00000001) != 0)) {
+          stats_ = java.util.Collections.unmodifiableList(stats_);
+          bitField0_ = (bitField0_ & ~0x00000001);
+        }
+        result.stats_ = stats_;
+      } else {
+        result.stats_ = statsBuilder_.build();
+      }
+      result.dataCase_ = dataCase_;
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.profiler.XEvent) {
+        return mergeFrom((org.tensorflow.proto.profiler.XEvent)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.profiler.XEvent other) {
+      if (other == org.tensorflow.proto.profiler.XEvent.getDefaultInstance()) return this;
+      if (other.getMetadataId() != 0L) {
+        setMetadataId(other.getMetadataId());
+      }
+      if (other.getDurationPs() != 0L) {
+        setDurationPs(other.getDurationPs());
+      }
+      if (statsBuilder_ == null) {
+        if (!other.stats_.isEmpty()) {
+          if (stats_.isEmpty()) {
+            stats_ = other.stats_;
+            bitField0_ = (bitField0_ & ~0x00000001);
+          } else {
+            ensureStatsIsMutable();
+            stats_.addAll(other.stats_);
+          }
+          onChanged();
+        }
+      } else {
+        if (!other.stats_.isEmpty()) {
+          if (statsBuilder_.isEmpty()) {
+            statsBuilder_.dispose();
+            statsBuilder_ = null;
+            stats_ = other.stats_;
+            bitField0_ = (bitField0_ & ~0x00000001);
+            statsBuilder_ = 
+              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
+                 getStatsFieldBuilder() : null;
+          } else {
+            statsBuilder_.addAllMessages(other.stats_);
+          }
+        }
+      }
+      switch (other.getDataCase()) {
+        case OFFSET_PS: {
+          setOffsetPs(other.getOffsetPs());
+          break;
+        }
+        case NUM_OCCURRENCES: {
+          setNumOccurrences(other.getNumOccurrences());
+          break;
+        }
+        case DATA_NOT_SET: {
+          break;
+        }
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.profiler.XEvent parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.profiler.XEvent) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+    private int dataCase_ = 0;
+    private java.lang.Object data_;
+    public DataCase
+        getDataCase() {
+      return DataCase.forNumber(
+          dataCase_);
+    }
+
+    public Builder clearData() {
+      dataCase_ = 0;
+      data_ = null;
+      onChanged();
+      return this;
+    }
+
+    private int bitField0_;
+
+    private long metadataId_ ;
+    /**
+     * <pre>
+     * XEventMetadata.id of corresponding metadata.
+     * </pre>
+     *
+     * <code>int64 metadata_id = 1;</code>
+     */
+    public long getMetadataId() {
+      return metadataId_;
+    }
+    /**
+     * <pre>
+     * XEventMetadata.id of corresponding metadata.
+     * </pre>
+     *
+     * <code>int64 metadata_id = 1;</code>
+     */
+    public Builder setMetadataId(long value) {
+      
+      metadataId_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * XEventMetadata.id of corresponding metadata.
+     * </pre>
+     *
+     * <code>int64 metadata_id = 1;</code>
+     */
+    public Builder clearMetadataId() {
+      
+      metadataId_ = 0L;
+      onChanged();
+      return this;
+    }
+
+    /**
+     * <pre>
+     * Start time of the event in picoseconds, as offset from
+     * XLine.timestamp_ns().
+     * </pre>
+     *
+     * <code>int64 offset_ps = 2;</code>
+     */
+    public long getOffsetPs() {
+      if (dataCase_ == 2) {
+        return (java.lang.Long) data_;
+      }
+      return 0L;
+    }
+    /**
+     * <pre>
+     * Start time of the event in picoseconds, as offset from
+     * XLine.timestamp_ns().
+     * </pre>
+     *
+     * <code>int64 offset_ps = 2;</code>
+     */
+    public Builder setOffsetPs(long value) {
+      dataCase_ = 2;
+      data_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Start time of the event in picoseconds, as offset from
+     * XLine.timestamp_ns().
+     * </pre>
+     *
+     * <code>int64 offset_ps = 2;</code>
+     */
+    public Builder clearOffsetPs() {
+      if (dataCase_ == 2) {
+        dataCase_ = 0;
+        data_ = null;
+        onChanged();
+      }
+      return this;
+    }
+
+    /**
+     * <pre>
+     * Number of occurrences of the event, if aggregated.
+     * </pre>
+     *
+     * <code>int64 num_occurrences = 5;</code>
+     */
+    public long getNumOccurrences() {
+      if (dataCase_ == 5) {
+        return (java.lang.Long) data_;
+      }
+      return 0L;
+    }
+    /**
+     * <pre>
+     * Number of occurrences of the event, if aggregated.
+     * </pre>
+     *
+     * <code>int64 num_occurrences = 5;</code>
+     */
+    public Builder setNumOccurrences(long value) {
+      dataCase_ = 5;
+      data_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Number of occurrences of the event, if aggregated.
+     * </pre>
+     *
+     * <code>int64 num_occurrences = 5;</code>
+     */
+    public Builder clearNumOccurrences() {
+      if (dataCase_ == 5) {
+        dataCase_ = 0;
+        data_ = null;
+        onChanged();
+      }
+      return this;
+    }
+
+    private long durationPs_ ;
+    /**
+     * <pre>
+     * Duration of the event in picoseconds. Can be zero for an instant event.
+     * </pre>
+     *
+     * <code>int64 duration_ps = 3;</code>
+     */
+    public long getDurationPs() {
+      return durationPs_;
+    }
+    /**
+     * <pre>
+     * Duration of the event in picoseconds. Can be zero for an instant event.
+     * </pre>
+     *
+     * <code>int64 duration_ps = 3;</code>
+     */
+    public Builder setDurationPs(long value) {
+      
+      durationPs_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Duration of the event in picoseconds. Can be zero for an instant event.
+     * </pre>
+     *
+     * <code>int64 duration_ps = 3;</code>
+     */
+    public Builder clearDurationPs() {
+      
+      durationPs_ = 0L;
+      onChanged();
+      return this;
+    }
+
+    private java.util.List<org.tensorflow.proto.profiler.XStat> stats_ =
+      java.util.Collections.emptyList();
+    private void ensureStatsIsMutable() {
+      if (!((bitField0_ & 0x00000001) != 0)) {
+        stats_ = new java.util.ArrayList<org.tensorflow.proto.profiler.XStat>(stats_);
+        bitField0_ |= 0x00000001;
+       }
+    }
+
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.profiler.XStat, org.tensorflow.proto.profiler.XStat.Builder, org.tensorflow.proto.profiler.XStatOrBuilder> statsBuilder_;
+
+    /**
+     * <pre>
+     * XStats associated with the event.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.profiler.XStat stats = 4;</code>
+     */
+    public java.util.List<org.tensorflow.proto.profiler.XStat> getStatsList() {
+      if (statsBuilder_ == null) {
+        return java.util.Collections.unmodifiableList(stats_);
+      } else {
+        return statsBuilder_.getMessageList();
+      }
+    }
+    /**
+     * <pre>
+     * XStats associated with the event.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.profiler.XStat stats = 4;</code>
+     */
+    public int getStatsCount() {
+      if (statsBuilder_ == null) {
+        return stats_.size();
+      } else {
+        return statsBuilder_.getCount();
+      }
+    }
+    /**
+     * <pre>
+     * XStats associated with the event.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.profiler.XStat stats = 4;</code>
+     */
+    public org.tensorflow.proto.profiler.XStat getStats(int index) {
+      if (statsBuilder_ == null) {
+        return stats_.get(index);
+      } else {
+        return statsBuilder_.getMessage(index);
+      }
+    }
+    /**
+     * <pre>
+     * XStats associated with the event.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.profiler.XStat stats = 4;</code>
+     */
+    public Builder setStats(
+        int index, org.tensorflow.proto.profiler.XStat value) {
+      if (statsBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureStatsIsMutable();
+        stats_.set(index, value);
+        onChanged();
+      } else {
+        statsBuilder_.setMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * XStats associated with the event.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.profiler.XStat stats = 4;</code>
+     */
+    public Builder setStats(
+        int index, org.tensorflow.proto.profiler.XStat.Builder builderForValue) {
+      if (statsBuilder_ == null) {
+        ensureStatsIsMutable();
+        stats_.set(index, builderForValue.build());
+        onChanged();
+      } else {
+        statsBuilder_.setMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * XStats associated with the event.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.profiler.XStat stats = 4;</code>
+     */
+    public Builder addStats(org.tensorflow.proto.profiler.XStat value) {
+      if (statsBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureStatsIsMutable();
+        stats_.add(value);
+        onChanged();
+      } else {
+        statsBuilder_.addMessage(value);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * XStats associated with the event.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.profiler.XStat stats = 4;</code>
+     */
+    public Builder addStats(
+        int index, org.tensorflow.proto.profiler.XStat value) {
+      if (statsBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureStatsIsMutable();
+        stats_.add(index, value);
+        onChanged();
+      } else {
+        statsBuilder_.addMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * XStats associated with the event.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.profiler.XStat stats = 4;</code>
+     */
+    public Builder addStats(
+        org.tensorflow.proto.profiler.XStat.Builder builderForValue) {
+      if (statsBuilder_ == null) {
+        ensureStatsIsMutable();
+        stats_.add(builderForValue.build());
+        onChanged();
+      } else {
+        statsBuilder_.addMessage(builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * XStats associated with the event.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.profiler.XStat stats = 4;</code>
+     */
+    public Builder addStats(
+        int index, org.tensorflow.proto.profiler.XStat.Builder builderForValue) {
+      if (statsBuilder_ == null) {
+        ensureStatsIsMutable();
+        stats_.add(index, builderForValue.build());
+        onChanged();
+      } else {
+        statsBuilder_.addMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * XStats associated with the event.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.profiler.XStat stats = 4;</code>
+     */
+    public Builder addAllStats(
+        java.lang.Iterable<? extends org.tensorflow.proto.profiler.XStat> values) {
+      if (statsBuilder_ == null) {
+        ensureStatsIsMutable();
+        com.google.protobuf.AbstractMessageLite.Builder.addAll(
+            values, stats_);
+        onChanged();
+      } else {
+        statsBuilder_.addAllMessages(values);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * XStats associated with the event.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.profiler.XStat stats = 4;</code>
+     */
+    public Builder clearStats() {
+      if (statsBuilder_ == null) {
+        stats_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000001);
+        onChanged();
+      } else {
+        statsBuilder_.clear();
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * XStats associated with the event.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.profiler.XStat stats = 4;</code>
+     */
+    public Builder removeStats(int index) {
+      if (statsBuilder_ == null) {
+        ensureStatsIsMutable();
+        stats_.remove(index);
+        onChanged();
+      } else {
+        statsBuilder_.remove(index);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * XStats associated with the event.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.profiler.XStat stats = 4;</code>
+     */
+    public org.tensorflow.proto.profiler.XStat.Builder getStatsBuilder(
+        int index) {
+      return getStatsFieldBuilder().getBuilder(index);
+    }
+    /**
+     * <pre>
+     * XStats associated with the event.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.profiler.XStat stats = 4;</code>
+     */
+    public org.tensorflow.proto.profiler.XStatOrBuilder getStatsOrBuilder(
+        int index) {
+      if (statsBuilder_ == null) {
+        return stats_.get(index);  } else {
+        return statsBuilder_.getMessageOrBuilder(index);
+      }
+    }
+    /**
+     * <pre>
+     * XStats associated with the event.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.profiler.XStat stats = 4;</code>
+     */
+    public java.util.List<? extends org.tensorflow.proto.profiler.XStatOrBuilder> 
+         getStatsOrBuilderList() {
+      if (statsBuilder_ != null) {
+        return statsBuilder_.getMessageOrBuilderList();
+      } else {
+        return java.util.Collections.unmodifiableList(stats_);
+      }
+    }
+    /**
+     * <pre>
+     * XStats associated with the event.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.profiler.XStat stats = 4;</code>
+     */
+    public org.tensorflow.proto.profiler.XStat.Builder addStatsBuilder() {
+      return getStatsFieldBuilder().addBuilder(
+          org.tensorflow.proto.profiler.XStat.getDefaultInstance());
+    }
+    /**
+     * <pre>
+     * XStats associated with the event.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.profiler.XStat stats = 4;</code>
+     */
+    public org.tensorflow.proto.profiler.XStat.Builder addStatsBuilder(
+        int index) {
+      return getStatsFieldBuilder().addBuilder(
+          index, org.tensorflow.proto.profiler.XStat.getDefaultInstance());
+    }
+    /**
+     * <pre>
+     * XStats associated with the event.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.profiler.XStat stats = 4;</code>
+     */
+    public java.util.List<org.tensorflow.proto.profiler.XStat.Builder> 
+         getStatsBuilderList() {
+      return getStatsFieldBuilder().getBuilderList();
+    }
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.profiler.XStat, org.tensorflow.proto.profiler.XStat.Builder, org.tensorflow.proto.profiler.XStatOrBuilder> 
+        getStatsFieldBuilder() {
+      if (statsBuilder_ == null) {
+        statsBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
+            org.tensorflow.proto.profiler.XStat, org.tensorflow.proto.profiler.XStat.Builder, org.tensorflow.proto.profiler.XStatOrBuilder>(
+                stats_,
+                ((bitField0_ & 0x00000001) != 0),
+                getParentForChildren(),
+                isClean());
+        stats_ = null;
+      }
+      return statsBuilder_;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.profiler.XEvent)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.profiler.XEvent)
+  private static final org.tensorflow.proto.profiler.XEvent DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.profiler.XEvent();
+  }
+
+  public static org.tensorflow.proto.profiler.XEvent getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<XEvent>
+      PARSER = new com.google.protobuf.AbstractParser<XEvent>() {
+    @java.lang.Override
+    public XEvent parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new XEvent(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<XEvent> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<XEvent> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.profiler.XEvent getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/XEventMetadata.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/XEventMetadata.java
new file mode 100644
index 00000000000..c7551d7f4f5
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/XEventMetadata.java
@@ -0,0 +1,890 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/profiler/protobuf/xplane.proto
+
+package org.tensorflow.proto.profiler;
+
+/**
+ * <pre>
+ * Metadata for an XEvent, shared by all instances of the same event.
+ * Next ID: 5
+ * </pre>
+ *
+ * Protobuf type {@code tensorflow.profiler.XEventMetadata}
+ */
+public  final class XEventMetadata extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.profiler.XEventMetadata)
+    XEventMetadataOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use XEventMetadata.newBuilder() to construct.
+  private XEventMetadata(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private XEventMetadata() {
+    name_ = "";
+    displayName_ = "";
+    metadata_ = com.google.protobuf.ByteString.EMPTY;
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new XEventMetadata();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private XEventMetadata(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 8: {
+
+            id_ = input.readInt64();
+            break;
+          }
+          case 18: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            name_ = s;
+            break;
+          }
+          case 26: {
+
+            metadata_ = input.readBytes();
+            break;
+          }
+          case 34: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            displayName_ = s;
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.profiler.XPlaneProtos.internal_static_tensorflow_profiler_XEventMetadata_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.profiler.XPlaneProtos.internal_static_tensorflow_profiler_XEventMetadata_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.profiler.XEventMetadata.class, org.tensorflow.proto.profiler.XEventMetadata.Builder.class);
+  }
+
+  public static final int ID_FIELD_NUMBER = 1;
+  private long id_;
+  /**
+   * <pre>
+   * XPlane.event_metadata map key.
+   * </pre>
+   *
+   * <code>int64 id = 1;</code>
+   */
+  public long getId() {
+    return id_;
+  }
+
+  public static final int NAME_FIELD_NUMBER = 2;
+  private volatile java.lang.Object name_;
+  /**
+   * <pre>
+   * Name of the event.
+   * </pre>
+   *
+   * <code>string name = 2;</code>
+   */
+  public java.lang.String getName() {
+    java.lang.Object ref = name_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      name_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * Name of the event.
+   * </pre>
+   *
+   * <code>string name = 2;</code>
+   */
+  public com.google.protobuf.ByteString
+      getNameBytes() {
+    java.lang.Object ref = name_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      name_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int DISPLAY_NAME_FIELD_NUMBER = 4;
+  private volatile java.lang.Object displayName_;
+  /**
+   * <pre>
+   * Name of the event shown in trace viewer.
+   * </pre>
+   *
+   * <code>string display_name = 4;</code>
+   */
+  public java.lang.String getDisplayName() {
+    java.lang.Object ref = displayName_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      displayName_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * Name of the event shown in trace viewer.
+   * </pre>
+   *
+   * <code>string display_name = 4;</code>
+   */
+  public com.google.protobuf.ByteString
+      getDisplayNameBytes() {
+    java.lang.Object ref = displayName_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      displayName_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int METADATA_FIELD_NUMBER = 3;
+  private com.google.protobuf.ByteString metadata_;
+  /**
+   * <pre>
+   * Additional metadata in serialized format.
+   * </pre>
+   *
+   * <code>bytes metadata = 3;</code>
+   */
+  public com.google.protobuf.ByteString getMetadata() {
+    return metadata_;
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    if (id_ != 0L) {
+      output.writeInt64(1, id_);
+    }
+    if (!getNameBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 2, name_);
+    }
+    if (!metadata_.isEmpty()) {
+      output.writeBytes(3, metadata_);
+    }
+    if (!getDisplayNameBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 4, displayName_);
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (id_ != 0L) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt64Size(1, id_);
+    }
+    if (!getNameBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(2, name_);
+    }
+    if (!metadata_.isEmpty()) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeBytesSize(3, metadata_);
+    }
+    if (!getDisplayNameBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(4, displayName_);
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.profiler.XEventMetadata)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.profiler.XEventMetadata other = (org.tensorflow.proto.profiler.XEventMetadata) obj;
+
+    if (getId()
+        != other.getId()) return false;
+    if (!getName()
+        .equals(other.getName())) return false;
+    if (!getDisplayName()
+        .equals(other.getDisplayName())) return false;
+    if (!getMetadata()
+        .equals(other.getMetadata())) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    hash = (37 * hash) + ID_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+        getId());
+    hash = (37 * hash) + NAME_FIELD_NUMBER;
+    hash = (53 * hash) + getName().hashCode();
+    hash = (37 * hash) + DISPLAY_NAME_FIELD_NUMBER;
+    hash = (53 * hash) + getDisplayName().hashCode();
+    hash = (37 * hash) + METADATA_FIELD_NUMBER;
+    hash = (53 * hash) + getMetadata().hashCode();
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.profiler.XEventMetadata parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.profiler.XEventMetadata parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.profiler.XEventMetadata parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.profiler.XEventMetadata parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.profiler.XEventMetadata parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.profiler.XEventMetadata parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.profiler.XEventMetadata parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.profiler.XEventMetadata parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.profiler.XEventMetadata parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.profiler.XEventMetadata parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.profiler.XEventMetadata parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.profiler.XEventMetadata parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.profiler.XEventMetadata prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * <pre>
+   * Metadata for an XEvent, shared by all instances of the same event.
+   * Next ID: 5
+   * </pre>
+   *
+   * Protobuf type {@code tensorflow.profiler.XEventMetadata}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.profiler.XEventMetadata)
+      org.tensorflow.proto.profiler.XEventMetadataOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.profiler.XPlaneProtos.internal_static_tensorflow_profiler_XEventMetadata_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.profiler.XPlaneProtos.internal_static_tensorflow_profiler_XEventMetadata_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.profiler.XEventMetadata.class, org.tensorflow.proto.profiler.XEventMetadata.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.profiler.XEventMetadata.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      id_ = 0L;
+
+      name_ = "";
+
+      displayName_ = "";
+
+      metadata_ = com.google.protobuf.ByteString.EMPTY;
+
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.profiler.XPlaneProtos.internal_static_tensorflow_profiler_XEventMetadata_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.profiler.XEventMetadata getDefaultInstanceForType() {
+      return org.tensorflow.proto.profiler.XEventMetadata.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.profiler.XEventMetadata build() {
+      org.tensorflow.proto.profiler.XEventMetadata result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.profiler.XEventMetadata buildPartial() {
+      org.tensorflow.proto.profiler.XEventMetadata result = new org.tensorflow.proto.profiler.XEventMetadata(this);
+      result.id_ = id_;
+      result.name_ = name_;
+      result.displayName_ = displayName_;
+      result.metadata_ = metadata_;
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.profiler.XEventMetadata) {
+        return mergeFrom((org.tensorflow.proto.profiler.XEventMetadata)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.profiler.XEventMetadata other) {
+      if (other == org.tensorflow.proto.profiler.XEventMetadata.getDefaultInstance()) return this;
+      if (other.getId() != 0L) {
+        setId(other.getId());
+      }
+      if (!other.getName().isEmpty()) {
+        name_ = other.name_;
+        onChanged();
+      }
+      if (!other.getDisplayName().isEmpty()) {
+        displayName_ = other.displayName_;
+        onChanged();
+      }
+      if (other.getMetadata() != com.google.protobuf.ByteString.EMPTY) {
+        setMetadata(other.getMetadata());
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.profiler.XEventMetadata parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.profiler.XEventMetadata) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+
+    private long id_ ;
+    /**
+     * <pre>
+     * XPlane.event_metadata map key.
+     * </pre>
+     *
+     * <code>int64 id = 1;</code>
+     */
+    public long getId() {
+      return id_;
+    }
+    /**
+     * <pre>
+     * XPlane.event_metadata map key.
+     * </pre>
+     *
+     * <code>int64 id = 1;</code>
+     */
+    public Builder setId(long value) {
+      
+      id_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * XPlane.event_metadata map key.
+     * </pre>
+     *
+     * <code>int64 id = 1;</code>
+     */
+    public Builder clearId() {
+      
+      id_ = 0L;
+      onChanged();
+      return this;
+    }
+
+    private java.lang.Object name_ = "";
+    /**
+     * <pre>
+     * Name of the event.
+     * </pre>
+     *
+     * <code>string name = 2;</code>
+     */
+    public java.lang.String getName() {
+      java.lang.Object ref = name_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        name_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Name of the event.
+     * </pre>
+     *
+     * <code>string name = 2;</code>
+     */
+    public com.google.protobuf.ByteString
+        getNameBytes() {
+      java.lang.Object ref = name_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        name_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Name of the event.
+     * </pre>
+     *
+     * <code>string name = 2;</code>
+     */
+    public Builder setName(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      name_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Name of the event.
+     * </pre>
+     *
+     * <code>string name = 2;</code>
+     */
+    public Builder clearName() {
+      
+      name_ = getDefaultInstance().getName();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Name of the event.
+     * </pre>
+     *
+     * <code>string name = 2;</code>
+     */
+    public Builder setNameBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      name_ = value;
+      onChanged();
+      return this;
+    }
+
+    private java.lang.Object displayName_ = "";
+    /**
+     * <pre>
+     * Name of the event shown in trace viewer.
+     * </pre>
+     *
+     * <code>string display_name = 4;</code>
+     */
+    public java.lang.String getDisplayName() {
+      java.lang.Object ref = displayName_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        displayName_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Name of the event shown in trace viewer.
+     * </pre>
+     *
+     * <code>string display_name = 4;</code>
+     */
+    public com.google.protobuf.ByteString
+        getDisplayNameBytes() {
+      java.lang.Object ref = displayName_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        displayName_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Name of the event shown in trace viewer.
+     * </pre>
+     *
+     * <code>string display_name = 4;</code>
+     */
+    public Builder setDisplayName(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      displayName_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Name of the event shown in trace viewer.
+     * </pre>
+     *
+     * <code>string display_name = 4;</code>
+     */
+    public Builder clearDisplayName() {
+      
+      displayName_ = getDefaultInstance().getDisplayName();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Name of the event shown in trace viewer.
+     * </pre>
+     *
+     * <code>string display_name = 4;</code>
+     */
+    public Builder setDisplayNameBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      displayName_ = value;
+      onChanged();
+      return this;
+    }
+
+    private com.google.protobuf.ByteString metadata_ = com.google.protobuf.ByteString.EMPTY;
+    /**
+     * <pre>
+     * Additional metadata in serialized format.
+     * </pre>
+     *
+     * <code>bytes metadata = 3;</code>
+     */
+    public com.google.protobuf.ByteString getMetadata() {
+      return metadata_;
+    }
+    /**
+     * <pre>
+     * Additional metadata in serialized format.
+     * </pre>
+     *
+     * <code>bytes metadata = 3;</code>
+     */
+    public Builder setMetadata(com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      metadata_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Additional metadata in serialized format.
+     * </pre>
+     *
+     * <code>bytes metadata = 3;</code>
+     */
+    public Builder clearMetadata() {
+      
+      metadata_ = getDefaultInstance().getMetadata();
+      onChanged();
+      return this;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.profiler.XEventMetadata)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.profiler.XEventMetadata)
+  private static final org.tensorflow.proto.profiler.XEventMetadata DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.profiler.XEventMetadata();
+  }
+
+  public static org.tensorflow.proto.profiler.XEventMetadata getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<XEventMetadata>
+      PARSER = new com.google.protobuf.AbstractParser<XEventMetadata>() {
+    @java.lang.Override
+    public XEventMetadata parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new XEventMetadata(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<XEventMetadata> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<XEventMetadata> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.profiler.XEventMetadata getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/XEventMetadataOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/XEventMetadataOrBuilder.java
new file mode 100644
index 00000000000..b150f6e1850
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/XEventMetadataOrBuilder.java
@@ -0,0 +1,63 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/profiler/protobuf/xplane.proto
+
+package org.tensorflow.proto.profiler;
+
+public interface XEventMetadataOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.profiler.XEventMetadata)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <pre>
+   * XPlane.event_metadata map key.
+   * </pre>
+   *
+   * <code>int64 id = 1;</code>
+   */
+  long getId();
+
+  /**
+   * <pre>
+   * Name of the event.
+   * </pre>
+   *
+   * <code>string name = 2;</code>
+   */
+  java.lang.String getName();
+  /**
+   * <pre>
+   * Name of the event.
+   * </pre>
+   *
+   * <code>string name = 2;</code>
+   */
+  com.google.protobuf.ByteString
+      getNameBytes();
+
+  /**
+   * <pre>
+   * Name of the event shown in trace viewer.
+   * </pre>
+   *
+   * <code>string display_name = 4;</code>
+   */
+  java.lang.String getDisplayName();
+  /**
+   * <pre>
+   * Name of the event shown in trace viewer.
+   * </pre>
+   *
+   * <code>string display_name = 4;</code>
+   */
+  com.google.protobuf.ByteString
+      getDisplayNameBytes();
+
+  /**
+   * <pre>
+   * Additional metadata in serialized format.
+   * </pre>
+   *
+   * <code>bytes metadata = 3;</code>
+   */
+  com.google.protobuf.ByteString getMetadata();
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/XEventOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/XEventOrBuilder.java
new file mode 100644
index 00000000000..3d9e4c2eec5
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/XEventOrBuilder.java
@@ -0,0 +1,92 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/profiler/protobuf/xplane.proto
+
+package org.tensorflow.proto.profiler;
+
+public interface XEventOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.profiler.XEvent)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <pre>
+   * XEventMetadata.id of corresponding metadata.
+   * </pre>
+   *
+   * <code>int64 metadata_id = 1;</code>
+   */
+  long getMetadataId();
+
+  /**
+   * <pre>
+   * Start time of the event in picoseconds, as offset from
+   * XLine.timestamp_ns().
+   * </pre>
+   *
+   * <code>int64 offset_ps = 2;</code>
+   */
+  long getOffsetPs();
+
+  /**
+   * <pre>
+   * Number of occurrences of the event, if aggregated.
+   * </pre>
+   *
+   * <code>int64 num_occurrences = 5;</code>
+   */
+  long getNumOccurrences();
+
+  /**
+   * <pre>
+   * Duration of the event in picoseconds. Can be zero for an instant event.
+   * </pre>
+   *
+   * <code>int64 duration_ps = 3;</code>
+   */
+  long getDurationPs();
+
+  /**
+   * <pre>
+   * XStats associated with the event.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.profiler.XStat stats = 4;</code>
+   */
+  java.util.List<org.tensorflow.proto.profiler.XStat> 
+      getStatsList();
+  /**
+   * <pre>
+   * XStats associated with the event.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.profiler.XStat stats = 4;</code>
+   */
+  org.tensorflow.proto.profiler.XStat getStats(int index);
+  /**
+   * <pre>
+   * XStats associated with the event.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.profiler.XStat stats = 4;</code>
+   */
+  int getStatsCount();
+  /**
+   * <pre>
+   * XStats associated with the event.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.profiler.XStat stats = 4;</code>
+   */
+  java.util.List<? extends org.tensorflow.proto.profiler.XStatOrBuilder> 
+      getStatsOrBuilderList();
+  /**
+   * <pre>
+   * XStats associated with the event.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.profiler.XStat stats = 4;</code>
+   */
+  org.tensorflow.proto.profiler.XStatOrBuilder getStatsOrBuilder(
+      int index);
+
+  public org.tensorflow.proto.profiler.XEvent.DataCase getDataCase();
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/XLine.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/XLine.java
new file mode 100644
index 00000000000..e9342bbb66f
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/XLine.java
@@ -0,0 +1,1508 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/profiler/protobuf/xplane.proto
+
+package org.tensorflow.proto.profiler;
+
+/**
+ * <pre>
+ * An XLine is a timeline of trace events (XEvents).
+ * Next ID: 12
+ * </pre>
+ *
+ * Protobuf type {@code tensorflow.profiler.XLine}
+ */
+public  final class XLine extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.profiler.XLine)
+    XLineOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use XLine.newBuilder() to construct.
+  private XLine(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private XLine() {
+    name_ = "";
+    displayName_ = "";
+    events_ = java.util.Collections.emptyList();
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new XLine();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private XLine(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    int mutable_bitField0_ = 0;
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 8: {
+
+            id_ = input.readInt64();
+            break;
+          }
+          case 18: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            name_ = s;
+            break;
+          }
+          case 24: {
+
+            timestampNs_ = input.readInt64();
+            break;
+          }
+          case 34: {
+            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
+              events_ = new java.util.ArrayList<org.tensorflow.proto.profiler.XEvent>();
+              mutable_bitField0_ |= 0x00000001;
+            }
+            events_.add(
+                input.readMessage(org.tensorflow.proto.profiler.XEvent.parser(), extensionRegistry));
+            break;
+          }
+          case 72: {
+
+            durationPs_ = input.readInt64();
+            break;
+          }
+          case 80: {
+
+            displayId_ = input.readInt64();
+            break;
+          }
+          case 90: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            displayName_ = s;
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      if (((mutable_bitField0_ & 0x00000001) != 0)) {
+        events_ = java.util.Collections.unmodifiableList(events_);
+      }
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.profiler.XPlaneProtos.internal_static_tensorflow_profiler_XLine_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.profiler.XPlaneProtos.internal_static_tensorflow_profiler_XLine_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.profiler.XLine.class, org.tensorflow.proto.profiler.XLine.Builder.class);
+  }
+
+  public static final int ID_FIELD_NUMBER = 1;
+  private long id_;
+  /**
+   * <pre>
+   * Id of this line, can be repeated within an XPlane. All XLines with the
+   * same id are effectively the same timeline.
+   * </pre>
+   *
+   * <code>int64 id = 1;</code>
+   */
+  public long getId() {
+    return id_;
+  }
+
+  public static final int DISPLAY_ID_FIELD_NUMBER = 10;
+  private long displayId_;
+  /**
+   * <pre>
+   * Display id of this line. Multiple lines with the same display_id are
+   * grouped together in the same trace viewer row.
+   * </pre>
+   *
+   * <code>int64 display_id = 10;</code>
+   */
+  public long getDisplayId() {
+    return displayId_;
+  }
+
+  public static final int NAME_FIELD_NUMBER = 2;
+  private volatile java.lang.Object name_;
+  /**
+   * <pre>
+   * Name of this XLine.
+   * </pre>
+   *
+   * <code>string name = 2;</code>
+   */
+  public java.lang.String getName() {
+    java.lang.Object ref = name_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      name_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * Name of this XLine.
+   * </pre>
+   *
+   * <code>string name = 2;</code>
+   */
+  public com.google.protobuf.ByteString
+      getNameBytes() {
+    java.lang.Object ref = name_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      name_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int DISPLAY_NAME_FIELD_NUMBER = 11;
+  private volatile java.lang.Object displayName_;
+  /**
+   * <pre>
+   * Name of this XLine to display in trace viewer.
+   * </pre>
+   *
+   * <code>string display_name = 11;</code>
+   */
+  public java.lang.String getDisplayName() {
+    java.lang.Object ref = displayName_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      displayName_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * Name of this XLine to display in trace viewer.
+   * </pre>
+   *
+   * <code>string display_name = 11;</code>
+   */
+  public com.google.protobuf.ByteString
+      getDisplayNameBytes() {
+    java.lang.Object ref = displayName_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      displayName_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int TIMESTAMP_NS_FIELD_NUMBER = 3;
+  private long timestampNs_;
+  /**
+   * <pre>
+   * Start time of this line in nanoseconds since the UNIX epoch.
+   * XEvent.offset_ps is relative to this timestamp.
+   * </pre>
+   *
+   * <code>int64 timestamp_ns = 3;</code>
+   */
+  public long getTimestampNs() {
+    return timestampNs_;
+  }
+
+  public static final int DURATION_PS_FIELD_NUMBER = 9;
+  private long durationPs_;
+  /**
+   * <pre>
+   * Profiling duration for this line in picoseconds.
+   * </pre>
+   *
+   * <code>int64 duration_ps = 9;</code>
+   */
+  public long getDurationPs() {
+    return durationPs_;
+  }
+
+  public static final int EVENTS_FIELD_NUMBER = 4;
+  private java.util.List<org.tensorflow.proto.profiler.XEvent> events_;
+  /**
+   * <pre>
+   * XEvents within the same XLine should not overlap in time, but they can be
+   * nested.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.profiler.XEvent events = 4;</code>
+   */
+  public java.util.List<org.tensorflow.proto.profiler.XEvent> getEventsList() {
+    return events_;
+  }
+  /**
+   * <pre>
+   * XEvents within the same XLine should not overlap in time, but they can be
+   * nested.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.profiler.XEvent events = 4;</code>
+   */
+  public java.util.List<? extends org.tensorflow.proto.profiler.XEventOrBuilder> 
+      getEventsOrBuilderList() {
+    return events_;
+  }
+  /**
+   * <pre>
+   * XEvents within the same XLine should not overlap in time, but they can be
+   * nested.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.profiler.XEvent events = 4;</code>
+   */
+  public int getEventsCount() {
+    return events_.size();
+  }
+  /**
+   * <pre>
+   * XEvents within the same XLine should not overlap in time, but they can be
+   * nested.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.profiler.XEvent events = 4;</code>
+   */
+  public org.tensorflow.proto.profiler.XEvent getEvents(int index) {
+    return events_.get(index);
+  }
+  /**
+   * <pre>
+   * XEvents within the same XLine should not overlap in time, but they can be
+   * nested.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.profiler.XEvent events = 4;</code>
+   */
+  public org.tensorflow.proto.profiler.XEventOrBuilder getEventsOrBuilder(
+      int index) {
+    return events_.get(index);
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    if (id_ != 0L) {
+      output.writeInt64(1, id_);
+    }
+    if (!getNameBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 2, name_);
+    }
+    if (timestampNs_ != 0L) {
+      output.writeInt64(3, timestampNs_);
+    }
+    for (int i = 0; i < events_.size(); i++) {
+      output.writeMessage(4, events_.get(i));
+    }
+    if (durationPs_ != 0L) {
+      output.writeInt64(9, durationPs_);
+    }
+    if (displayId_ != 0L) {
+      output.writeInt64(10, displayId_);
+    }
+    if (!getDisplayNameBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 11, displayName_);
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (id_ != 0L) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt64Size(1, id_);
+    }
+    if (!getNameBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(2, name_);
+    }
+    if (timestampNs_ != 0L) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt64Size(3, timestampNs_);
+    }
+    for (int i = 0; i < events_.size(); i++) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(4, events_.get(i));
+    }
+    if (durationPs_ != 0L) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt64Size(9, durationPs_);
+    }
+    if (displayId_ != 0L) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt64Size(10, displayId_);
+    }
+    if (!getDisplayNameBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(11, displayName_);
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.profiler.XLine)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.profiler.XLine other = (org.tensorflow.proto.profiler.XLine) obj;
+
+    if (getId()
+        != other.getId()) return false;
+    if (getDisplayId()
+        != other.getDisplayId()) return false;
+    if (!getName()
+        .equals(other.getName())) return false;
+    if (!getDisplayName()
+        .equals(other.getDisplayName())) return false;
+    if (getTimestampNs()
+        != other.getTimestampNs()) return false;
+    if (getDurationPs()
+        != other.getDurationPs()) return false;
+    if (!getEventsList()
+        .equals(other.getEventsList())) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    hash = (37 * hash) + ID_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+        getId());
+    hash = (37 * hash) + DISPLAY_ID_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+        getDisplayId());
+    hash = (37 * hash) + NAME_FIELD_NUMBER;
+    hash = (53 * hash) + getName().hashCode();
+    hash = (37 * hash) + DISPLAY_NAME_FIELD_NUMBER;
+    hash = (53 * hash) + getDisplayName().hashCode();
+    hash = (37 * hash) + TIMESTAMP_NS_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+        getTimestampNs());
+    hash = (37 * hash) + DURATION_PS_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+        getDurationPs());
+    if (getEventsCount() > 0) {
+      hash = (37 * hash) + EVENTS_FIELD_NUMBER;
+      hash = (53 * hash) + getEventsList().hashCode();
+    }
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.profiler.XLine parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.profiler.XLine parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.profiler.XLine parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.profiler.XLine parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.profiler.XLine parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.profiler.XLine parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.profiler.XLine parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.profiler.XLine parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.profiler.XLine parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.profiler.XLine parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.profiler.XLine parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.profiler.XLine parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.profiler.XLine prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * <pre>
+   * An XLine is a timeline of trace events (XEvents).
+   * Next ID: 12
+   * </pre>
+   *
+   * Protobuf type {@code tensorflow.profiler.XLine}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.profiler.XLine)
+      org.tensorflow.proto.profiler.XLineOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.profiler.XPlaneProtos.internal_static_tensorflow_profiler_XLine_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.profiler.XPlaneProtos.internal_static_tensorflow_profiler_XLine_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.profiler.XLine.class, org.tensorflow.proto.profiler.XLine.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.profiler.XLine.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+        getEventsFieldBuilder();
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      id_ = 0L;
+
+      displayId_ = 0L;
+
+      name_ = "";
+
+      displayName_ = "";
+
+      timestampNs_ = 0L;
+
+      durationPs_ = 0L;
+
+      if (eventsBuilder_ == null) {
+        events_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000001);
+      } else {
+        eventsBuilder_.clear();
+      }
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.profiler.XPlaneProtos.internal_static_tensorflow_profiler_XLine_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.profiler.XLine getDefaultInstanceForType() {
+      return org.tensorflow.proto.profiler.XLine.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.profiler.XLine build() {
+      org.tensorflow.proto.profiler.XLine result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.profiler.XLine buildPartial() {
+      org.tensorflow.proto.profiler.XLine result = new org.tensorflow.proto.profiler.XLine(this);
+      int from_bitField0_ = bitField0_;
+      result.id_ = id_;
+      result.displayId_ = displayId_;
+      result.name_ = name_;
+      result.displayName_ = displayName_;
+      result.timestampNs_ = timestampNs_;
+      result.durationPs_ = durationPs_;
+      if (eventsBuilder_ == null) {
+        if (((bitField0_ & 0x00000001) != 0)) {
+          events_ = java.util.Collections.unmodifiableList(events_);
+          bitField0_ = (bitField0_ & ~0x00000001);
+        }
+        result.events_ = events_;
+      } else {
+        result.events_ = eventsBuilder_.build();
+      }
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.profiler.XLine) {
+        return mergeFrom((org.tensorflow.proto.profiler.XLine)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.profiler.XLine other) {
+      if (other == org.tensorflow.proto.profiler.XLine.getDefaultInstance()) return this;
+      if (other.getId() != 0L) {
+        setId(other.getId());
+      }
+      if (other.getDisplayId() != 0L) {
+        setDisplayId(other.getDisplayId());
+      }
+      if (!other.getName().isEmpty()) {
+        name_ = other.name_;
+        onChanged();
+      }
+      if (!other.getDisplayName().isEmpty()) {
+        displayName_ = other.displayName_;
+        onChanged();
+      }
+      if (other.getTimestampNs() != 0L) {
+        setTimestampNs(other.getTimestampNs());
+      }
+      if (other.getDurationPs() != 0L) {
+        setDurationPs(other.getDurationPs());
+      }
+      if (eventsBuilder_ == null) {
+        if (!other.events_.isEmpty()) {
+          if (events_.isEmpty()) {
+            events_ = other.events_;
+            bitField0_ = (bitField0_ & ~0x00000001);
+          } else {
+            ensureEventsIsMutable();
+            events_.addAll(other.events_);
+          }
+          onChanged();
+        }
+      } else {
+        if (!other.events_.isEmpty()) {
+          if (eventsBuilder_.isEmpty()) {
+            eventsBuilder_.dispose();
+            eventsBuilder_ = null;
+            events_ = other.events_;
+            bitField0_ = (bitField0_ & ~0x00000001);
+            eventsBuilder_ = 
+              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
+                 getEventsFieldBuilder() : null;
+          } else {
+            eventsBuilder_.addAllMessages(other.events_);
+          }
+        }
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.profiler.XLine parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.profiler.XLine) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+    private int bitField0_;
+
+    private long id_ ;
+    /**
+     * <pre>
+     * Id of this line, can be repeated within an XPlane. All XLines with the
+     * same id are effectively the same timeline.
+     * </pre>
+     *
+     * <code>int64 id = 1;</code>
+     */
+    public long getId() {
+      return id_;
+    }
+    /**
+     * <pre>
+     * Id of this line, can be repeated within an XPlane. All XLines with the
+     * same id are effectively the same timeline.
+     * </pre>
+     *
+     * <code>int64 id = 1;</code>
+     */
+    public Builder setId(long value) {
+      
+      id_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Id of this line, can be repeated within an XPlane. All XLines with the
+     * same id are effectively the same timeline.
+     * </pre>
+     *
+     * <code>int64 id = 1;</code>
+     */
+    public Builder clearId() {
+      
+      id_ = 0L;
+      onChanged();
+      return this;
+    }
+
+    private long displayId_ ;
+    /**
+     * <pre>
+     * Display id of this line. Multiple lines with the same display_id are
+     * grouped together in the same trace viewer row.
+     * </pre>
+     *
+     * <code>int64 display_id = 10;</code>
+     */
+    public long getDisplayId() {
+      return displayId_;
+    }
+    /**
+     * <pre>
+     * Display id of this line. Multiple lines with the same display_id are
+     * grouped together in the same trace viewer row.
+     * </pre>
+     *
+     * <code>int64 display_id = 10;</code>
+     */
+    public Builder setDisplayId(long value) {
+      
+      displayId_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Display id of this line. Multiple lines with the same display_id are
+     * grouped together in the same trace viewer row.
+     * </pre>
+     *
+     * <code>int64 display_id = 10;</code>
+     */
+    public Builder clearDisplayId() {
+      
+      displayId_ = 0L;
+      onChanged();
+      return this;
+    }
+
+    private java.lang.Object name_ = "";
+    /**
+     * <pre>
+     * Name of this XLine.
+     * </pre>
+     *
+     * <code>string name = 2;</code>
+     */
+    public java.lang.String getName() {
+      java.lang.Object ref = name_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        name_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Name of this XLine.
+     * </pre>
+     *
+     * <code>string name = 2;</code>
+     */
+    public com.google.protobuf.ByteString
+        getNameBytes() {
+      java.lang.Object ref = name_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        name_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Name of this XLine.
+     * </pre>
+     *
+     * <code>string name = 2;</code>
+     */
+    public Builder setName(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      name_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Name of this XLine.
+     * </pre>
+     *
+     * <code>string name = 2;</code>
+     */
+    public Builder clearName() {
+      
+      name_ = getDefaultInstance().getName();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Name of this XLine.
+     * </pre>
+     *
+     * <code>string name = 2;</code>
+     */
+    public Builder setNameBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      name_ = value;
+      onChanged();
+      return this;
+    }
+
+    private java.lang.Object displayName_ = "";
+    /**
+     * <pre>
+     * Name of this XLine to display in trace viewer.
+     * </pre>
+     *
+     * <code>string display_name = 11;</code>
+     */
+    public java.lang.String getDisplayName() {
+      java.lang.Object ref = displayName_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        displayName_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Name of this XLine to display in trace viewer.
+     * </pre>
+     *
+     * <code>string display_name = 11;</code>
+     */
+    public com.google.protobuf.ByteString
+        getDisplayNameBytes() {
+      java.lang.Object ref = displayName_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        displayName_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Name of this XLine to display in trace viewer.
+     * </pre>
+     *
+     * <code>string display_name = 11;</code>
+     */
+    public Builder setDisplayName(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      displayName_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Name of this XLine to display in trace viewer.
+     * </pre>
+     *
+     * <code>string display_name = 11;</code>
+     */
+    public Builder clearDisplayName() {
+      
+      displayName_ = getDefaultInstance().getDisplayName();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Name of this XLine to display in trace viewer.
+     * </pre>
+     *
+     * <code>string display_name = 11;</code>
+     */
+    public Builder setDisplayNameBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      displayName_ = value;
+      onChanged();
+      return this;
+    }
+
+    private long timestampNs_ ;
+    /**
+     * <pre>
+     * Start time of this line in nanoseconds since the UNIX epoch.
+     * XEvent.offset_ps is relative to this timestamp.
+     * </pre>
+     *
+     * <code>int64 timestamp_ns = 3;</code>
+     */
+    public long getTimestampNs() {
+      return timestampNs_;
+    }
+    /**
+     * <pre>
+     * Start time of this line in nanoseconds since the UNIX epoch.
+     * XEvent.offset_ps is relative to this timestamp.
+     * </pre>
+     *
+     * <code>int64 timestamp_ns = 3;</code>
+     */
+    public Builder setTimestampNs(long value) {
+      
+      timestampNs_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Start time of this line in nanoseconds since the UNIX epoch.
+     * XEvent.offset_ps is relative to this timestamp.
+     * </pre>
+     *
+     * <code>int64 timestamp_ns = 3;</code>
+     */
+    public Builder clearTimestampNs() {
+      
+      timestampNs_ = 0L;
+      onChanged();
+      return this;
+    }
+
+    private long durationPs_ ;
+    /**
+     * <pre>
+     * Profiling duration for this line in picoseconds.
+     * </pre>
+     *
+     * <code>int64 duration_ps = 9;</code>
+     */
+    public long getDurationPs() {
+      return durationPs_;
+    }
+    /**
+     * <pre>
+     * Profiling duration for this line in picoseconds.
+     * </pre>
+     *
+     * <code>int64 duration_ps = 9;</code>
+     */
+    public Builder setDurationPs(long value) {
+      
+      durationPs_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Profiling duration for this line in picoseconds.
+     * </pre>
+     *
+     * <code>int64 duration_ps = 9;</code>
+     */
+    public Builder clearDurationPs() {
+      
+      durationPs_ = 0L;
+      onChanged();
+      return this;
+    }
+
+    private java.util.List<org.tensorflow.proto.profiler.XEvent> events_ =
+      java.util.Collections.emptyList();
+    private void ensureEventsIsMutable() {
+      if (!((bitField0_ & 0x00000001) != 0)) {
+        events_ = new java.util.ArrayList<org.tensorflow.proto.profiler.XEvent>(events_);
+        bitField0_ |= 0x00000001;
+       }
+    }
+
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.profiler.XEvent, org.tensorflow.proto.profiler.XEvent.Builder, org.tensorflow.proto.profiler.XEventOrBuilder> eventsBuilder_;
+
+    /**
+     * <pre>
+     * XEvents within the same XLine should not overlap in time, but they can be
+     * nested.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.profiler.XEvent events = 4;</code>
+     */
+    public java.util.List<org.tensorflow.proto.profiler.XEvent> getEventsList() {
+      if (eventsBuilder_ == null) {
+        return java.util.Collections.unmodifiableList(events_);
+      } else {
+        return eventsBuilder_.getMessageList();
+      }
+    }
+    /**
+     * <pre>
+     * XEvents within the same XLine should not overlap in time, but they can be
+     * nested.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.profiler.XEvent events = 4;</code>
+     */
+    public int getEventsCount() {
+      if (eventsBuilder_ == null) {
+        return events_.size();
+      } else {
+        return eventsBuilder_.getCount();
+      }
+    }
+    /**
+     * <pre>
+     * XEvents within the same XLine should not overlap in time, but they can be
+     * nested.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.profiler.XEvent events = 4;</code>
+     */
+    public org.tensorflow.proto.profiler.XEvent getEvents(int index) {
+      if (eventsBuilder_ == null) {
+        return events_.get(index);
+      } else {
+        return eventsBuilder_.getMessage(index);
+      }
+    }
+    /**
+     * <pre>
+     * XEvents within the same XLine should not overlap in time, but they can be
+     * nested.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.profiler.XEvent events = 4;</code>
+     */
+    public Builder setEvents(
+        int index, org.tensorflow.proto.profiler.XEvent value) {
+      if (eventsBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureEventsIsMutable();
+        events_.set(index, value);
+        onChanged();
+      } else {
+        eventsBuilder_.setMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * XEvents within the same XLine should not overlap in time, but they can be
+     * nested.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.profiler.XEvent events = 4;</code>
+     */
+    public Builder setEvents(
+        int index, org.tensorflow.proto.profiler.XEvent.Builder builderForValue) {
+      if (eventsBuilder_ == null) {
+        ensureEventsIsMutable();
+        events_.set(index, builderForValue.build());
+        onChanged();
+      } else {
+        eventsBuilder_.setMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * XEvents within the same XLine should not overlap in time, but they can be
+     * nested.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.profiler.XEvent events = 4;</code>
+     */
+    public Builder addEvents(org.tensorflow.proto.profiler.XEvent value) {
+      if (eventsBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureEventsIsMutable();
+        events_.add(value);
+        onChanged();
+      } else {
+        eventsBuilder_.addMessage(value);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * XEvents within the same XLine should not overlap in time, but they can be
+     * nested.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.profiler.XEvent events = 4;</code>
+     */
+    public Builder addEvents(
+        int index, org.tensorflow.proto.profiler.XEvent value) {
+      if (eventsBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureEventsIsMutable();
+        events_.add(index, value);
+        onChanged();
+      } else {
+        eventsBuilder_.addMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * XEvents within the same XLine should not overlap in time, but they can be
+     * nested.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.profiler.XEvent events = 4;</code>
+     */
+    public Builder addEvents(
+        org.tensorflow.proto.profiler.XEvent.Builder builderForValue) {
+      if (eventsBuilder_ == null) {
+        ensureEventsIsMutable();
+        events_.add(builderForValue.build());
+        onChanged();
+      } else {
+        eventsBuilder_.addMessage(builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * XEvents within the same XLine should not overlap in time, but they can be
+     * nested.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.profiler.XEvent events = 4;</code>
+     */
+    public Builder addEvents(
+        int index, org.tensorflow.proto.profiler.XEvent.Builder builderForValue) {
+      if (eventsBuilder_ == null) {
+        ensureEventsIsMutable();
+        events_.add(index, builderForValue.build());
+        onChanged();
+      } else {
+        eventsBuilder_.addMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * XEvents within the same XLine should not overlap in time, but they can be
+     * nested.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.profiler.XEvent events = 4;</code>
+     */
+    public Builder addAllEvents(
+        java.lang.Iterable<? extends org.tensorflow.proto.profiler.XEvent> values) {
+      if (eventsBuilder_ == null) {
+        ensureEventsIsMutable();
+        com.google.protobuf.AbstractMessageLite.Builder.addAll(
+            values, events_);
+        onChanged();
+      } else {
+        eventsBuilder_.addAllMessages(values);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * XEvents within the same XLine should not overlap in time, but they can be
+     * nested.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.profiler.XEvent events = 4;</code>
+     */
+    public Builder clearEvents() {
+      if (eventsBuilder_ == null) {
+        events_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000001);
+        onChanged();
+      } else {
+        eventsBuilder_.clear();
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * XEvents within the same XLine should not overlap in time, but they can be
+     * nested.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.profiler.XEvent events = 4;</code>
+     */
+    public Builder removeEvents(int index) {
+      if (eventsBuilder_ == null) {
+        ensureEventsIsMutable();
+        events_.remove(index);
+        onChanged();
+      } else {
+        eventsBuilder_.remove(index);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * XEvents within the same XLine should not overlap in time, but they can be
+     * nested.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.profiler.XEvent events = 4;</code>
+     */
+    public org.tensorflow.proto.profiler.XEvent.Builder getEventsBuilder(
+        int index) {
+      return getEventsFieldBuilder().getBuilder(index);
+    }
+    /**
+     * <pre>
+     * XEvents within the same XLine should not overlap in time, but they can be
+     * nested.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.profiler.XEvent events = 4;</code>
+     */
+    public org.tensorflow.proto.profiler.XEventOrBuilder getEventsOrBuilder(
+        int index) {
+      if (eventsBuilder_ == null) {
+        return events_.get(index);  } else {
+        return eventsBuilder_.getMessageOrBuilder(index);
+      }
+    }
+    /**
+     * <pre>
+     * XEvents within the same XLine should not overlap in time, but they can be
+     * nested.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.profiler.XEvent events = 4;</code>
+     */
+    public java.util.List<? extends org.tensorflow.proto.profiler.XEventOrBuilder> 
+         getEventsOrBuilderList() {
+      if (eventsBuilder_ != null) {
+        return eventsBuilder_.getMessageOrBuilderList();
+      } else {
+        return java.util.Collections.unmodifiableList(events_);
+      }
+    }
+    /**
+     * <pre>
+     * XEvents within the same XLine should not overlap in time, but they can be
+     * nested.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.profiler.XEvent events = 4;</code>
+     */
+    public org.tensorflow.proto.profiler.XEvent.Builder addEventsBuilder() {
+      return getEventsFieldBuilder().addBuilder(
+          org.tensorflow.proto.profiler.XEvent.getDefaultInstance());
+    }
+    /**
+     * <pre>
+     * XEvents within the same XLine should not overlap in time, but they can be
+     * nested.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.profiler.XEvent events = 4;</code>
+     */
+    public org.tensorflow.proto.profiler.XEvent.Builder addEventsBuilder(
+        int index) {
+      return getEventsFieldBuilder().addBuilder(
+          index, org.tensorflow.proto.profiler.XEvent.getDefaultInstance());
+    }
+    /**
+     * <pre>
+     * XEvents within the same XLine should not overlap in time, but they can be
+     * nested.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.profiler.XEvent events = 4;</code>
+     */
+    public java.util.List<org.tensorflow.proto.profiler.XEvent.Builder> 
+         getEventsBuilderList() {
+      return getEventsFieldBuilder().getBuilderList();
+    }
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.profiler.XEvent, org.tensorflow.proto.profiler.XEvent.Builder, org.tensorflow.proto.profiler.XEventOrBuilder> 
+        getEventsFieldBuilder() {
+      if (eventsBuilder_ == null) {
+        eventsBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
+            org.tensorflow.proto.profiler.XEvent, org.tensorflow.proto.profiler.XEvent.Builder, org.tensorflow.proto.profiler.XEventOrBuilder>(
+                events_,
+                ((bitField0_ & 0x00000001) != 0),
+                getParentForChildren(),
+                isClean());
+        events_ = null;
+      }
+      return eventsBuilder_;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.profiler.XLine)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.profiler.XLine)
+  private static final org.tensorflow.proto.profiler.XLine DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.profiler.XLine();
+  }
+
+  public static org.tensorflow.proto.profiler.XLine getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<XLine>
+      PARSER = new com.google.protobuf.AbstractParser<XLine>() {
+    @java.lang.Override
+    public XLine parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new XLine(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<XLine> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<XLine> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.profiler.XLine getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/XLineOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/XLineOrBuilder.java
new file mode 100644
index 00000000000..1cf5477dca7
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/XLineOrBuilder.java
@@ -0,0 +1,133 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/profiler/protobuf/xplane.proto
+
+package org.tensorflow.proto.profiler;
+
+public interface XLineOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.profiler.XLine)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <pre>
+   * Id of this line, can be repeated within an XPlane. All XLines with the
+   * same id are effectively the same timeline.
+   * </pre>
+   *
+   * <code>int64 id = 1;</code>
+   */
+  long getId();
+
+  /**
+   * <pre>
+   * Display id of this line. Multiple lines with the same display_id are
+   * grouped together in the same trace viewer row.
+   * </pre>
+   *
+   * <code>int64 display_id = 10;</code>
+   */
+  long getDisplayId();
+
+  /**
+   * <pre>
+   * Name of this XLine.
+   * </pre>
+   *
+   * <code>string name = 2;</code>
+   */
+  java.lang.String getName();
+  /**
+   * <pre>
+   * Name of this XLine.
+   * </pre>
+   *
+   * <code>string name = 2;</code>
+   */
+  com.google.protobuf.ByteString
+      getNameBytes();
+
+  /**
+   * <pre>
+   * Name of this XLine to display in trace viewer.
+   * </pre>
+   *
+   * <code>string display_name = 11;</code>
+   */
+  java.lang.String getDisplayName();
+  /**
+   * <pre>
+   * Name of this XLine to display in trace viewer.
+   * </pre>
+   *
+   * <code>string display_name = 11;</code>
+   */
+  com.google.protobuf.ByteString
+      getDisplayNameBytes();
+
+  /**
+   * <pre>
+   * Start time of this line in nanoseconds since the UNIX epoch.
+   * XEvent.offset_ps is relative to this timestamp.
+   * </pre>
+   *
+   * <code>int64 timestamp_ns = 3;</code>
+   */
+  long getTimestampNs();
+
+  /**
+   * <pre>
+   * Profiling duration for this line in picoseconds.
+   * </pre>
+   *
+   * <code>int64 duration_ps = 9;</code>
+   */
+  long getDurationPs();
+
+  /**
+   * <pre>
+   * XEvents within the same XLine should not overlap in time, but they can be
+   * nested.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.profiler.XEvent events = 4;</code>
+   */
+  java.util.List<org.tensorflow.proto.profiler.XEvent> 
+      getEventsList();
+  /**
+   * <pre>
+   * XEvents within the same XLine should not overlap in time, but they can be
+   * nested.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.profiler.XEvent events = 4;</code>
+   */
+  org.tensorflow.proto.profiler.XEvent getEvents(int index);
+  /**
+   * <pre>
+   * XEvents within the same XLine should not overlap in time, but they can be
+   * nested.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.profiler.XEvent events = 4;</code>
+   */
+  int getEventsCount();
+  /**
+   * <pre>
+   * XEvents within the same XLine should not overlap in time, but they can be
+   * nested.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.profiler.XEvent events = 4;</code>
+   */
+  java.util.List<? extends org.tensorflow.proto.profiler.XEventOrBuilder> 
+      getEventsOrBuilderList();
+  /**
+   * <pre>
+   * XEvents within the same XLine should not overlap in time, but they can be
+   * nested.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.profiler.XEvent events = 4;</code>
+   */
+  org.tensorflow.proto.profiler.XEventOrBuilder getEventsOrBuilder(
+      int index);
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/XPlane.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/XPlane.java
new file mode 100644
index 00000000000..7668a7f1d74
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/XPlane.java
@@ -0,0 +1,2168 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/profiler/protobuf/xplane.proto
+
+package org.tensorflow.proto.profiler;
+
+/**
+ * <pre>
+ * An XPlane is a container of parallel timelines (XLines), generated by a
+ * profiling source or by post-processing one or more XPlanes.
+ * Next ID: 7
+ * </pre>
+ *
+ * Protobuf type {@code tensorflow.profiler.XPlane}
+ */
+public  final class XPlane extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.profiler.XPlane)
+    XPlaneOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use XPlane.newBuilder() to construct.
+  private XPlane(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private XPlane() {
+    name_ = "";
+    lines_ = java.util.Collections.emptyList();
+    stats_ = java.util.Collections.emptyList();
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new XPlane();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private XPlane(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    int mutable_bitField0_ = 0;
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 8: {
+
+            id_ = input.readInt64();
+            break;
+          }
+          case 18: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            name_ = s;
+            break;
+          }
+          case 26: {
+            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
+              lines_ = new java.util.ArrayList<org.tensorflow.proto.profiler.XLine>();
+              mutable_bitField0_ |= 0x00000001;
+            }
+            lines_.add(
+                input.readMessage(org.tensorflow.proto.profiler.XLine.parser(), extensionRegistry));
+            break;
+          }
+          case 34: {
+            if (!((mutable_bitField0_ & 0x00000002) != 0)) {
+              eventMetadata_ = com.google.protobuf.MapField.newMapField(
+                  EventMetadataDefaultEntryHolder.defaultEntry);
+              mutable_bitField0_ |= 0x00000002;
+            }
+            com.google.protobuf.MapEntry<java.lang.Long, org.tensorflow.proto.profiler.XEventMetadata>
+            eventMetadata__ = input.readMessage(
+                EventMetadataDefaultEntryHolder.defaultEntry.getParserForType(), extensionRegistry);
+            eventMetadata_.getMutableMap().put(
+                eventMetadata__.getKey(), eventMetadata__.getValue());
+            break;
+          }
+          case 42: {
+            if (!((mutable_bitField0_ & 0x00000004) != 0)) {
+              statMetadata_ = com.google.protobuf.MapField.newMapField(
+                  StatMetadataDefaultEntryHolder.defaultEntry);
+              mutable_bitField0_ |= 0x00000004;
+            }
+            com.google.protobuf.MapEntry<java.lang.Long, org.tensorflow.proto.profiler.XStatMetadata>
+            statMetadata__ = input.readMessage(
+                StatMetadataDefaultEntryHolder.defaultEntry.getParserForType(), extensionRegistry);
+            statMetadata_.getMutableMap().put(
+                statMetadata__.getKey(), statMetadata__.getValue());
+            break;
+          }
+          case 50: {
+            if (!((mutable_bitField0_ & 0x00000008) != 0)) {
+              stats_ = new java.util.ArrayList<org.tensorflow.proto.profiler.XStat>();
+              mutable_bitField0_ |= 0x00000008;
+            }
+            stats_.add(
+                input.readMessage(org.tensorflow.proto.profiler.XStat.parser(), extensionRegistry));
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      if (((mutable_bitField0_ & 0x00000001) != 0)) {
+        lines_ = java.util.Collections.unmodifiableList(lines_);
+      }
+      if (((mutable_bitField0_ & 0x00000008) != 0)) {
+        stats_ = java.util.Collections.unmodifiableList(stats_);
+      }
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.profiler.XPlaneProtos.internal_static_tensorflow_profiler_XPlane_descriptor;
+  }
+
+  @SuppressWarnings({"rawtypes"})
+  @java.lang.Override
+  protected com.google.protobuf.MapField internalGetMapField(
+      int number) {
+    switch (number) {
+      case 4:
+        return internalGetEventMetadata();
+      case 5:
+        return internalGetStatMetadata();
+      default:
+        throw new RuntimeException(
+            "Invalid map field number: " + number);
+    }
+  }
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.profiler.XPlaneProtos.internal_static_tensorflow_profiler_XPlane_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.profiler.XPlane.class, org.tensorflow.proto.profiler.XPlane.Builder.class);
+  }
+
+  public static final int ID_FIELD_NUMBER = 1;
+  private long id_;
+  /**
+   * <code>int64 id = 1;</code>
+   */
+  public long getId() {
+    return id_;
+  }
+
+  public static final int NAME_FIELD_NUMBER = 2;
+  private volatile java.lang.Object name_;
+  /**
+   * <pre>
+   * Name of this line.
+   * </pre>
+   *
+   * <code>string name = 2;</code>
+   */
+  public java.lang.String getName() {
+    java.lang.Object ref = name_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      name_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * Name of this line.
+   * </pre>
+   *
+   * <code>string name = 2;</code>
+   */
+  public com.google.protobuf.ByteString
+      getNameBytes() {
+    java.lang.Object ref = name_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      name_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int LINES_FIELD_NUMBER = 3;
+  private java.util.List<org.tensorflow.proto.profiler.XLine> lines_;
+  /**
+   * <pre>
+   * Parallel timelines grouped in this plane. XLines with the same id
+   * are effectively the same timeline.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.profiler.XLine lines = 3;</code>
+   */
+  public java.util.List<org.tensorflow.proto.profiler.XLine> getLinesList() {
+    return lines_;
+  }
+  /**
+   * <pre>
+   * Parallel timelines grouped in this plane. XLines with the same id
+   * are effectively the same timeline.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.profiler.XLine lines = 3;</code>
+   */
+  public java.util.List<? extends org.tensorflow.proto.profiler.XLineOrBuilder> 
+      getLinesOrBuilderList() {
+    return lines_;
+  }
+  /**
+   * <pre>
+   * Parallel timelines grouped in this plane. XLines with the same id
+   * are effectively the same timeline.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.profiler.XLine lines = 3;</code>
+   */
+  public int getLinesCount() {
+    return lines_.size();
+  }
+  /**
+   * <pre>
+   * Parallel timelines grouped in this plane. XLines with the same id
+   * are effectively the same timeline.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.profiler.XLine lines = 3;</code>
+   */
+  public org.tensorflow.proto.profiler.XLine getLines(int index) {
+    return lines_.get(index);
+  }
+  /**
+   * <pre>
+   * Parallel timelines grouped in this plane. XLines with the same id
+   * are effectively the same timeline.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.profiler.XLine lines = 3;</code>
+   */
+  public org.tensorflow.proto.profiler.XLineOrBuilder getLinesOrBuilder(
+      int index) {
+    return lines_.get(index);
+  }
+
+  public static final int EVENT_METADATA_FIELD_NUMBER = 4;
+  private static final class EventMetadataDefaultEntryHolder {
+    static final com.google.protobuf.MapEntry<
+        java.lang.Long, org.tensorflow.proto.profiler.XEventMetadata> defaultEntry =
+            com.google.protobuf.MapEntry
+            .<java.lang.Long, org.tensorflow.proto.profiler.XEventMetadata>newDefaultInstance(
+                org.tensorflow.proto.profiler.XPlaneProtos.internal_static_tensorflow_profiler_XPlane_EventMetadataEntry_descriptor, 
+                com.google.protobuf.WireFormat.FieldType.INT64,
+                0L,
+                com.google.protobuf.WireFormat.FieldType.MESSAGE,
+                org.tensorflow.proto.profiler.XEventMetadata.getDefaultInstance());
+  }
+  private com.google.protobuf.MapField<
+      java.lang.Long, org.tensorflow.proto.profiler.XEventMetadata> eventMetadata_;
+  private com.google.protobuf.MapField<java.lang.Long, org.tensorflow.proto.profiler.XEventMetadata>
+  internalGetEventMetadata() {
+    if (eventMetadata_ == null) {
+      return com.google.protobuf.MapField.emptyMapField(
+          EventMetadataDefaultEntryHolder.defaultEntry);
+    }
+    return eventMetadata_;
+  }
+
+  public int getEventMetadataCount() {
+    return internalGetEventMetadata().getMap().size();
+  }
+  /**
+   * <pre>
+   * XEventMetadata map, each entry uses the XEventMetadata.id as key. This map
+   * should be used for events that share the same ID over the whole XPlane.
+   * </pre>
+   *
+   * <code>map&lt;int64, .tensorflow.profiler.XEventMetadata&gt; event_metadata = 4;</code>
+   */
+
+  public boolean containsEventMetadata(
+      long key) {
+    
+    return internalGetEventMetadata().getMap().containsKey(key);
+  }
+  /**
+   * Use {@link #getEventMetadataMap()} instead.
+   */
+  @java.lang.Deprecated
+  public java.util.Map<java.lang.Long, org.tensorflow.proto.profiler.XEventMetadata> getEventMetadata() {
+    return getEventMetadataMap();
+  }
+  /**
+   * <pre>
+   * XEventMetadata map, each entry uses the XEventMetadata.id as key. This map
+   * should be used for events that share the same ID over the whole XPlane.
+   * </pre>
+   *
+   * <code>map&lt;int64, .tensorflow.profiler.XEventMetadata&gt; event_metadata = 4;</code>
+   */
+
+  public java.util.Map<java.lang.Long, org.tensorflow.proto.profiler.XEventMetadata> getEventMetadataMap() {
+    return internalGetEventMetadata().getMap();
+  }
+  /**
+   * <pre>
+   * XEventMetadata map, each entry uses the XEventMetadata.id as key. This map
+   * should be used for events that share the same ID over the whole XPlane.
+   * </pre>
+   *
+   * <code>map&lt;int64, .tensorflow.profiler.XEventMetadata&gt; event_metadata = 4;</code>
+   */
+
+  public org.tensorflow.proto.profiler.XEventMetadata getEventMetadataOrDefault(
+      long key,
+      org.tensorflow.proto.profiler.XEventMetadata defaultValue) {
+    
+    java.util.Map<java.lang.Long, org.tensorflow.proto.profiler.XEventMetadata> map =
+        internalGetEventMetadata().getMap();
+    return map.containsKey(key) ? map.get(key) : defaultValue;
+  }
+  /**
+   * <pre>
+   * XEventMetadata map, each entry uses the XEventMetadata.id as key. This map
+   * should be used for events that share the same ID over the whole XPlane.
+   * </pre>
+   *
+   * <code>map&lt;int64, .tensorflow.profiler.XEventMetadata&gt; event_metadata = 4;</code>
+   */
+
+  public org.tensorflow.proto.profiler.XEventMetadata getEventMetadataOrThrow(
+      long key) {
+    
+    java.util.Map<java.lang.Long, org.tensorflow.proto.profiler.XEventMetadata> map =
+        internalGetEventMetadata().getMap();
+    if (!map.containsKey(key)) {
+      throw new java.lang.IllegalArgumentException();
+    }
+    return map.get(key);
+  }
+
+  public static final int STAT_METADATA_FIELD_NUMBER = 5;
+  private static final class StatMetadataDefaultEntryHolder {
+    static final com.google.protobuf.MapEntry<
+        java.lang.Long, org.tensorflow.proto.profiler.XStatMetadata> defaultEntry =
+            com.google.protobuf.MapEntry
+            .<java.lang.Long, org.tensorflow.proto.profiler.XStatMetadata>newDefaultInstance(
+                org.tensorflow.proto.profiler.XPlaneProtos.internal_static_tensorflow_profiler_XPlane_StatMetadataEntry_descriptor, 
+                com.google.protobuf.WireFormat.FieldType.INT64,
+                0L,
+                com.google.protobuf.WireFormat.FieldType.MESSAGE,
+                org.tensorflow.proto.profiler.XStatMetadata.getDefaultInstance());
+  }
+  private com.google.protobuf.MapField<
+      java.lang.Long, org.tensorflow.proto.profiler.XStatMetadata> statMetadata_;
+  private com.google.protobuf.MapField<java.lang.Long, org.tensorflow.proto.profiler.XStatMetadata>
+  internalGetStatMetadata() {
+    if (statMetadata_ == null) {
+      return com.google.protobuf.MapField.emptyMapField(
+          StatMetadataDefaultEntryHolder.defaultEntry);
+    }
+    return statMetadata_;
+  }
+
+  public int getStatMetadataCount() {
+    return internalGetStatMetadata().getMap().size();
+  }
+  /**
+   * <pre>
+   * XStatMetadata map, each entry uses the XStatMetadata.id as key. This map
+   * should be used for stats that share the same ID over the whole XPlane.
+   * </pre>
+   *
+   * <code>map&lt;int64, .tensorflow.profiler.XStatMetadata&gt; stat_metadata = 5;</code>
+   */
+
+  public boolean containsStatMetadata(
+      long key) {
+    
+    return internalGetStatMetadata().getMap().containsKey(key);
+  }
+  /**
+   * Use {@link #getStatMetadataMap()} instead.
+   */
+  @java.lang.Deprecated
+  public java.util.Map<java.lang.Long, org.tensorflow.proto.profiler.XStatMetadata> getStatMetadata() {
+    return getStatMetadataMap();
+  }
+  /**
+   * <pre>
+   * XStatMetadata map, each entry uses the XStatMetadata.id as key. This map
+   * should be used for stats that share the same ID over the whole XPlane.
+   * </pre>
+   *
+   * <code>map&lt;int64, .tensorflow.profiler.XStatMetadata&gt; stat_metadata = 5;</code>
+   */
+
+  public java.util.Map<java.lang.Long, org.tensorflow.proto.profiler.XStatMetadata> getStatMetadataMap() {
+    return internalGetStatMetadata().getMap();
+  }
+  /**
+   * <pre>
+   * XStatMetadata map, each entry uses the XStatMetadata.id as key. This map
+   * should be used for stats that share the same ID over the whole XPlane.
+   * </pre>
+   *
+   * <code>map&lt;int64, .tensorflow.profiler.XStatMetadata&gt; stat_metadata = 5;</code>
+   */
+
+  public org.tensorflow.proto.profiler.XStatMetadata getStatMetadataOrDefault(
+      long key,
+      org.tensorflow.proto.profiler.XStatMetadata defaultValue) {
+    
+    java.util.Map<java.lang.Long, org.tensorflow.proto.profiler.XStatMetadata> map =
+        internalGetStatMetadata().getMap();
+    return map.containsKey(key) ? map.get(key) : defaultValue;
+  }
+  /**
+   * <pre>
+   * XStatMetadata map, each entry uses the XStatMetadata.id as key. This map
+   * should be used for stats that share the same ID over the whole XPlane.
+   * </pre>
+   *
+   * <code>map&lt;int64, .tensorflow.profiler.XStatMetadata&gt; stat_metadata = 5;</code>
+   */
+
+  public org.tensorflow.proto.profiler.XStatMetadata getStatMetadataOrThrow(
+      long key) {
+    
+    java.util.Map<java.lang.Long, org.tensorflow.proto.profiler.XStatMetadata> map =
+        internalGetStatMetadata().getMap();
+    if (!map.containsKey(key)) {
+      throw new java.lang.IllegalArgumentException();
+    }
+    return map.get(key);
+  }
+
+  public static final int STATS_FIELD_NUMBER = 6;
+  private java.util.List<org.tensorflow.proto.profiler.XStat> stats_;
+  /**
+   * <pre>
+   * XStats associated with this plane, e.g. device capabilities.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.profiler.XStat stats = 6;</code>
+   */
+  public java.util.List<org.tensorflow.proto.profiler.XStat> getStatsList() {
+    return stats_;
+  }
+  /**
+   * <pre>
+   * XStats associated with this plane, e.g. device capabilities.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.profiler.XStat stats = 6;</code>
+   */
+  public java.util.List<? extends org.tensorflow.proto.profiler.XStatOrBuilder> 
+      getStatsOrBuilderList() {
+    return stats_;
+  }
+  /**
+   * <pre>
+   * XStats associated with this plane, e.g. device capabilities.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.profiler.XStat stats = 6;</code>
+   */
+  public int getStatsCount() {
+    return stats_.size();
+  }
+  /**
+   * <pre>
+   * XStats associated with this plane, e.g. device capabilities.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.profiler.XStat stats = 6;</code>
+   */
+  public org.tensorflow.proto.profiler.XStat getStats(int index) {
+    return stats_.get(index);
+  }
+  /**
+   * <pre>
+   * XStats associated with this plane, e.g. device capabilities.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.profiler.XStat stats = 6;</code>
+   */
+  public org.tensorflow.proto.profiler.XStatOrBuilder getStatsOrBuilder(
+      int index) {
+    return stats_.get(index);
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    if (id_ != 0L) {
+      output.writeInt64(1, id_);
+    }
+    if (!getNameBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 2, name_);
+    }
+    for (int i = 0; i < lines_.size(); i++) {
+      output.writeMessage(3, lines_.get(i));
+    }
+    com.google.protobuf.GeneratedMessageV3
+      .serializeLongMapTo(
+        output,
+        internalGetEventMetadata(),
+        EventMetadataDefaultEntryHolder.defaultEntry,
+        4);
+    com.google.protobuf.GeneratedMessageV3
+      .serializeLongMapTo(
+        output,
+        internalGetStatMetadata(),
+        StatMetadataDefaultEntryHolder.defaultEntry,
+        5);
+    for (int i = 0; i < stats_.size(); i++) {
+      output.writeMessage(6, stats_.get(i));
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (id_ != 0L) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt64Size(1, id_);
+    }
+    if (!getNameBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(2, name_);
+    }
+    for (int i = 0; i < lines_.size(); i++) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(3, lines_.get(i));
+    }
+    for (java.util.Map.Entry<java.lang.Long, org.tensorflow.proto.profiler.XEventMetadata> entry
+         : internalGetEventMetadata().getMap().entrySet()) {
+      com.google.protobuf.MapEntry<java.lang.Long, org.tensorflow.proto.profiler.XEventMetadata>
+      eventMetadata__ = EventMetadataDefaultEntryHolder.defaultEntry.newBuilderForType()
+          .setKey(entry.getKey())
+          .setValue(entry.getValue())
+          .build();
+      size += com.google.protobuf.CodedOutputStream
+          .computeMessageSize(4, eventMetadata__);
+    }
+    for (java.util.Map.Entry<java.lang.Long, org.tensorflow.proto.profiler.XStatMetadata> entry
+         : internalGetStatMetadata().getMap().entrySet()) {
+      com.google.protobuf.MapEntry<java.lang.Long, org.tensorflow.proto.profiler.XStatMetadata>
+      statMetadata__ = StatMetadataDefaultEntryHolder.defaultEntry.newBuilderForType()
+          .setKey(entry.getKey())
+          .setValue(entry.getValue())
+          .build();
+      size += com.google.protobuf.CodedOutputStream
+          .computeMessageSize(5, statMetadata__);
+    }
+    for (int i = 0; i < stats_.size(); i++) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(6, stats_.get(i));
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.profiler.XPlane)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.profiler.XPlane other = (org.tensorflow.proto.profiler.XPlane) obj;
+
+    if (getId()
+        != other.getId()) return false;
+    if (!getName()
+        .equals(other.getName())) return false;
+    if (!getLinesList()
+        .equals(other.getLinesList())) return false;
+    if (!internalGetEventMetadata().equals(
+        other.internalGetEventMetadata())) return false;
+    if (!internalGetStatMetadata().equals(
+        other.internalGetStatMetadata())) return false;
+    if (!getStatsList()
+        .equals(other.getStatsList())) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    hash = (37 * hash) + ID_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+        getId());
+    hash = (37 * hash) + NAME_FIELD_NUMBER;
+    hash = (53 * hash) + getName().hashCode();
+    if (getLinesCount() > 0) {
+      hash = (37 * hash) + LINES_FIELD_NUMBER;
+      hash = (53 * hash) + getLinesList().hashCode();
+    }
+    if (!internalGetEventMetadata().getMap().isEmpty()) {
+      hash = (37 * hash) + EVENT_METADATA_FIELD_NUMBER;
+      hash = (53 * hash) + internalGetEventMetadata().hashCode();
+    }
+    if (!internalGetStatMetadata().getMap().isEmpty()) {
+      hash = (37 * hash) + STAT_METADATA_FIELD_NUMBER;
+      hash = (53 * hash) + internalGetStatMetadata().hashCode();
+    }
+    if (getStatsCount() > 0) {
+      hash = (37 * hash) + STATS_FIELD_NUMBER;
+      hash = (53 * hash) + getStatsList().hashCode();
+    }
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.profiler.XPlane parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.profiler.XPlane parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.profiler.XPlane parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.profiler.XPlane parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.profiler.XPlane parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.profiler.XPlane parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.profiler.XPlane parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.profiler.XPlane parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.profiler.XPlane parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.profiler.XPlane parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.profiler.XPlane parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.profiler.XPlane parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.profiler.XPlane prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * <pre>
+   * An XPlane is a container of parallel timelines (XLines), generated by a
+   * profiling source or by post-processing one or more XPlanes.
+   * Next ID: 7
+   * </pre>
+   *
+   * Protobuf type {@code tensorflow.profiler.XPlane}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.profiler.XPlane)
+      org.tensorflow.proto.profiler.XPlaneOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.profiler.XPlaneProtos.internal_static_tensorflow_profiler_XPlane_descriptor;
+    }
+
+    @SuppressWarnings({"rawtypes"})
+    protected com.google.protobuf.MapField internalGetMapField(
+        int number) {
+      switch (number) {
+        case 4:
+          return internalGetEventMetadata();
+        case 5:
+          return internalGetStatMetadata();
+        default:
+          throw new RuntimeException(
+              "Invalid map field number: " + number);
+      }
+    }
+    @SuppressWarnings({"rawtypes"})
+    protected com.google.protobuf.MapField internalGetMutableMapField(
+        int number) {
+      switch (number) {
+        case 4:
+          return internalGetMutableEventMetadata();
+        case 5:
+          return internalGetMutableStatMetadata();
+        default:
+          throw new RuntimeException(
+              "Invalid map field number: " + number);
+      }
+    }
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.profiler.XPlaneProtos.internal_static_tensorflow_profiler_XPlane_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.profiler.XPlane.class, org.tensorflow.proto.profiler.XPlane.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.profiler.XPlane.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+        getLinesFieldBuilder();
+        getStatsFieldBuilder();
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      id_ = 0L;
+
+      name_ = "";
+
+      if (linesBuilder_ == null) {
+        lines_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000001);
+      } else {
+        linesBuilder_.clear();
+      }
+      internalGetMutableEventMetadata().clear();
+      internalGetMutableStatMetadata().clear();
+      if (statsBuilder_ == null) {
+        stats_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000008);
+      } else {
+        statsBuilder_.clear();
+      }
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.profiler.XPlaneProtos.internal_static_tensorflow_profiler_XPlane_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.profiler.XPlane getDefaultInstanceForType() {
+      return org.tensorflow.proto.profiler.XPlane.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.profiler.XPlane build() {
+      org.tensorflow.proto.profiler.XPlane result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.profiler.XPlane buildPartial() {
+      org.tensorflow.proto.profiler.XPlane result = new org.tensorflow.proto.profiler.XPlane(this);
+      int from_bitField0_ = bitField0_;
+      result.id_ = id_;
+      result.name_ = name_;
+      if (linesBuilder_ == null) {
+        if (((bitField0_ & 0x00000001) != 0)) {
+          lines_ = java.util.Collections.unmodifiableList(lines_);
+          bitField0_ = (bitField0_ & ~0x00000001);
+        }
+        result.lines_ = lines_;
+      } else {
+        result.lines_ = linesBuilder_.build();
+      }
+      result.eventMetadata_ = internalGetEventMetadata();
+      result.eventMetadata_.makeImmutable();
+      result.statMetadata_ = internalGetStatMetadata();
+      result.statMetadata_.makeImmutable();
+      if (statsBuilder_ == null) {
+        if (((bitField0_ & 0x00000008) != 0)) {
+          stats_ = java.util.Collections.unmodifiableList(stats_);
+          bitField0_ = (bitField0_ & ~0x00000008);
+        }
+        result.stats_ = stats_;
+      } else {
+        result.stats_ = statsBuilder_.build();
+      }
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.profiler.XPlane) {
+        return mergeFrom((org.tensorflow.proto.profiler.XPlane)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.profiler.XPlane other) {
+      if (other == org.tensorflow.proto.profiler.XPlane.getDefaultInstance()) return this;
+      if (other.getId() != 0L) {
+        setId(other.getId());
+      }
+      if (!other.getName().isEmpty()) {
+        name_ = other.name_;
+        onChanged();
+      }
+      if (linesBuilder_ == null) {
+        if (!other.lines_.isEmpty()) {
+          if (lines_.isEmpty()) {
+            lines_ = other.lines_;
+            bitField0_ = (bitField0_ & ~0x00000001);
+          } else {
+            ensureLinesIsMutable();
+            lines_.addAll(other.lines_);
+          }
+          onChanged();
+        }
+      } else {
+        if (!other.lines_.isEmpty()) {
+          if (linesBuilder_.isEmpty()) {
+            linesBuilder_.dispose();
+            linesBuilder_ = null;
+            lines_ = other.lines_;
+            bitField0_ = (bitField0_ & ~0x00000001);
+            linesBuilder_ = 
+              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
+                 getLinesFieldBuilder() : null;
+          } else {
+            linesBuilder_.addAllMessages(other.lines_);
+          }
+        }
+      }
+      internalGetMutableEventMetadata().mergeFrom(
+          other.internalGetEventMetadata());
+      internalGetMutableStatMetadata().mergeFrom(
+          other.internalGetStatMetadata());
+      if (statsBuilder_ == null) {
+        if (!other.stats_.isEmpty()) {
+          if (stats_.isEmpty()) {
+            stats_ = other.stats_;
+            bitField0_ = (bitField0_ & ~0x00000008);
+          } else {
+            ensureStatsIsMutable();
+            stats_.addAll(other.stats_);
+          }
+          onChanged();
+        }
+      } else {
+        if (!other.stats_.isEmpty()) {
+          if (statsBuilder_.isEmpty()) {
+            statsBuilder_.dispose();
+            statsBuilder_ = null;
+            stats_ = other.stats_;
+            bitField0_ = (bitField0_ & ~0x00000008);
+            statsBuilder_ = 
+              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
+                 getStatsFieldBuilder() : null;
+          } else {
+            statsBuilder_.addAllMessages(other.stats_);
+          }
+        }
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.profiler.XPlane parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.profiler.XPlane) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+    private int bitField0_;
+
+    private long id_ ;
+    /**
+     * <code>int64 id = 1;</code>
+     */
+    public long getId() {
+      return id_;
+    }
+    /**
+     * <code>int64 id = 1;</code>
+     */
+    public Builder setId(long value) {
+      
+      id_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>int64 id = 1;</code>
+     */
+    public Builder clearId() {
+      
+      id_ = 0L;
+      onChanged();
+      return this;
+    }
+
+    private java.lang.Object name_ = "";
+    /**
+     * <pre>
+     * Name of this line.
+     * </pre>
+     *
+     * <code>string name = 2;</code>
+     */
+    public java.lang.String getName() {
+      java.lang.Object ref = name_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        name_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Name of this line.
+     * </pre>
+     *
+     * <code>string name = 2;</code>
+     */
+    public com.google.protobuf.ByteString
+        getNameBytes() {
+      java.lang.Object ref = name_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        name_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Name of this line.
+     * </pre>
+     *
+     * <code>string name = 2;</code>
+     */
+    public Builder setName(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      name_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Name of this line.
+     * </pre>
+     *
+     * <code>string name = 2;</code>
+     */
+    public Builder clearName() {
+      
+      name_ = getDefaultInstance().getName();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Name of this line.
+     * </pre>
+     *
+     * <code>string name = 2;</code>
+     */
+    public Builder setNameBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      name_ = value;
+      onChanged();
+      return this;
+    }
+
+    private java.util.List<org.tensorflow.proto.profiler.XLine> lines_ =
+      java.util.Collections.emptyList();
+    private void ensureLinesIsMutable() {
+      if (!((bitField0_ & 0x00000001) != 0)) {
+        lines_ = new java.util.ArrayList<org.tensorflow.proto.profiler.XLine>(lines_);
+        bitField0_ |= 0x00000001;
+       }
+    }
+
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.profiler.XLine, org.tensorflow.proto.profiler.XLine.Builder, org.tensorflow.proto.profiler.XLineOrBuilder> linesBuilder_;
+
+    /**
+     * <pre>
+     * Parallel timelines grouped in this plane. XLines with the same id
+     * are effectively the same timeline.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.profiler.XLine lines = 3;</code>
+     */
+    public java.util.List<org.tensorflow.proto.profiler.XLine> getLinesList() {
+      if (linesBuilder_ == null) {
+        return java.util.Collections.unmodifiableList(lines_);
+      } else {
+        return linesBuilder_.getMessageList();
+      }
+    }
+    /**
+     * <pre>
+     * Parallel timelines grouped in this plane. XLines with the same id
+     * are effectively the same timeline.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.profiler.XLine lines = 3;</code>
+     */
+    public int getLinesCount() {
+      if (linesBuilder_ == null) {
+        return lines_.size();
+      } else {
+        return linesBuilder_.getCount();
+      }
+    }
+    /**
+     * <pre>
+     * Parallel timelines grouped in this plane. XLines with the same id
+     * are effectively the same timeline.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.profiler.XLine lines = 3;</code>
+     */
+    public org.tensorflow.proto.profiler.XLine getLines(int index) {
+      if (linesBuilder_ == null) {
+        return lines_.get(index);
+      } else {
+        return linesBuilder_.getMessage(index);
+      }
+    }
+    /**
+     * <pre>
+     * Parallel timelines grouped in this plane. XLines with the same id
+     * are effectively the same timeline.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.profiler.XLine lines = 3;</code>
+     */
+    public Builder setLines(
+        int index, org.tensorflow.proto.profiler.XLine value) {
+      if (linesBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureLinesIsMutable();
+        lines_.set(index, value);
+        onChanged();
+      } else {
+        linesBuilder_.setMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Parallel timelines grouped in this plane. XLines with the same id
+     * are effectively the same timeline.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.profiler.XLine lines = 3;</code>
+     */
+    public Builder setLines(
+        int index, org.tensorflow.proto.profiler.XLine.Builder builderForValue) {
+      if (linesBuilder_ == null) {
+        ensureLinesIsMutable();
+        lines_.set(index, builderForValue.build());
+        onChanged();
+      } else {
+        linesBuilder_.setMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Parallel timelines grouped in this plane. XLines with the same id
+     * are effectively the same timeline.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.profiler.XLine lines = 3;</code>
+     */
+    public Builder addLines(org.tensorflow.proto.profiler.XLine value) {
+      if (linesBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureLinesIsMutable();
+        lines_.add(value);
+        onChanged();
+      } else {
+        linesBuilder_.addMessage(value);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Parallel timelines grouped in this plane. XLines with the same id
+     * are effectively the same timeline.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.profiler.XLine lines = 3;</code>
+     */
+    public Builder addLines(
+        int index, org.tensorflow.proto.profiler.XLine value) {
+      if (linesBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureLinesIsMutable();
+        lines_.add(index, value);
+        onChanged();
+      } else {
+        linesBuilder_.addMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Parallel timelines grouped in this plane. XLines with the same id
+     * are effectively the same timeline.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.profiler.XLine lines = 3;</code>
+     */
+    public Builder addLines(
+        org.tensorflow.proto.profiler.XLine.Builder builderForValue) {
+      if (linesBuilder_ == null) {
+        ensureLinesIsMutable();
+        lines_.add(builderForValue.build());
+        onChanged();
+      } else {
+        linesBuilder_.addMessage(builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Parallel timelines grouped in this plane. XLines with the same id
+     * are effectively the same timeline.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.profiler.XLine lines = 3;</code>
+     */
+    public Builder addLines(
+        int index, org.tensorflow.proto.profiler.XLine.Builder builderForValue) {
+      if (linesBuilder_ == null) {
+        ensureLinesIsMutable();
+        lines_.add(index, builderForValue.build());
+        onChanged();
+      } else {
+        linesBuilder_.addMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Parallel timelines grouped in this plane. XLines with the same id
+     * are effectively the same timeline.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.profiler.XLine lines = 3;</code>
+     */
+    public Builder addAllLines(
+        java.lang.Iterable<? extends org.tensorflow.proto.profiler.XLine> values) {
+      if (linesBuilder_ == null) {
+        ensureLinesIsMutable();
+        com.google.protobuf.AbstractMessageLite.Builder.addAll(
+            values, lines_);
+        onChanged();
+      } else {
+        linesBuilder_.addAllMessages(values);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Parallel timelines grouped in this plane. XLines with the same id
+     * are effectively the same timeline.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.profiler.XLine lines = 3;</code>
+     */
+    public Builder clearLines() {
+      if (linesBuilder_ == null) {
+        lines_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000001);
+        onChanged();
+      } else {
+        linesBuilder_.clear();
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Parallel timelines grouped in this plane. XLines with the same id
+     * are effectively the same timeline.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.profiler.XLine lines = 3;</code>
+     */
+    public Builder removeLines(int index) {
+      if (linesBuilder_ == null) {
+        ensureLinesIsMutable();
+        lines_.remove(index);
+        onChanged();
+      } else {
+        linesBuilder_.remove(index);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Parallel timelines grouped in this plane. XLines with the same id
+     * are effectively the same timeline.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.profiler.XLine lines = 3;</code>
+     */
+    public org.tensorflow.proto.profiler.XLine.Builder getLinesBuilder(
+        int index) {
+      return getLinesFieldBuilder().getBuilder(index);
+    }
+    /**
+     * <pre>
+     * Parallel timelines grouped in this plane. XLines with the same id
+     * are effectively the same timeline.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.profiler.XLine lines = 3;</code>
+     */
+    public org.tensorflow.proto.profiler.XLineOrBuilder getLinesOrBuilder(
+        int index) {
+      if (linesBuilder_ == null) {
+        return lines_.get(index);  } else {
+        return linesBuilder_.getMessageOrBuilder(index);
+      }
+    }
+    /**
+     * <pre>
+     * Parallel timelines grouped in this plane. XLines with the same id
+     * are effectively the same timeline.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.profiler.XLine lines = 3;</code>
+     */
+    public java.util.List<? extends org.tensorflow.proto.profiler.XLineOrBuilder> 
+         getLinesOrBuilderList() {
+      if (linesBuilder_ != null) {
+        return linesBuilder_.getMessageOrBuilderList();
+      } else {
+        return java.util.Collections.unmodifiableList(lines_);
+      }
+    }
+    /**
+     * <pre>
+     * Parallel timelines grouped in this plane. XLines with the same id
+     * are effectively the same timeline.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.profiler.XLine lines = 3;</code>
+     */
+    public org.tensorflow.proto.profiler.XLine.Builder addLinesBuilder() {
+      return getLinesFieldBuilder().addBuilder(
+          org.tensorflow.proto.profiler.XLine.getDefaultInstance());
+    }
+    /**
+     * <pre>
+     * Parallel timelines grouped in this plane. XLines with the same id
+     * are effectively the same timeline.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.profiler.XLine lines = 3;</code>
+     */
+    public org.tensorflow.proto.profiler.XLine.Builder addLinesBuilder(
+        int index) {
+      return getLinesFieldBuilder().addBuilder(
+          index, org.tensorflow.proto.profiler.XLine.getDefaultInstance());
+    }
+    /**
+     * <pre>
+     * Parallel timelines grouped in this plane. XLines with the same id
+     * are effectively the same timeline.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.profiler.XLine lines = 3;</code>
+     */
+    public java.util.List<org.tensorflow.proto.profiler.XLine.Builder> 
+         getLinesBuilderList() {
+      return getLinesFieldBuilder().getBuilderList();
+    }
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.profiler.XLine, org.tensorflow.proto.profiler.XLine.Builder, org.tensorflow.proto.profiler.XLineOrBuilder> 
+        getLinesFieldBuilder() {
+      if (linesBuilder_ == null) {
+        linesBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
+            org.tensorflow.proto.profiler.XLine, org.tensorflow.proto.profiler.XLine.Builder, org.tensorflow.proto.profiler.XLineOrBuilder>(
+                lines_,
+                ((bitField0_ & 0x00000001) != 0),
+                getParentForChildren(),
+                isClean());
+        lines_ = null;
+      }
+      return linesBuilder_;
+    }
+
+    private com.google.protobuf.MapField<
+        java.lang.Long, org.tensorflow.proto.profiler.XEventMetadata> eventMetadata_;
+    private com.google.protobuf.MapField<java.lang.Long, org.tensorflow.proto.profiler.XEventMetadata>
+    internalGetEventMetadata() {
+      if (eventMetadata_ == null) {
+        return com.google.protobuf.MapField.emptyMapField(
+            EventMetadataDefaultEntryHolder.defaultEntry);
+      }
+      return eventMetadata_;
+    }
+    private com.google.protobuf.MapField<java.lang.Long, org.tensorflow.proto.profiler.XEventMetadata>
+    internalGetMutableEventMetadata() {
+      onChanged();;
+      if (eventMetadata_ == null) {
+        eventMetadata_ = com.google.protobuf.MapField.newMapField(
+            EventMetadataDefaultEntryHolder.defaultEntry);
+      }
+      if (!eventMetadata_.isMutable()) {
+        eventMetadata_ = eventMetadata_.copy();
+      }
+      return eventMetadata_;
+    }
+
+    public int getEventMetadataCount() {
+      return internalGetEventMetadata().getMap().size();
+    }
+    /**
+     * <pre>
+     * XEventMetadata map, each entry uses the XEventMetadata.id as key. This map
+     * should be used for events that share the same ID over the whole XPlane.
+     * </pre>
+     *
+     * <code>map&lt;int64, .tensorflow.profiler.XEventMetadata&gt; event_metadata = 4;</code>
+     */
+
+    public boolean containsEventMetadata(
+        long key) {
+      
+      return internalGetEventMetadata().getMap().containsKey(key);
+    }
+    /**
+     * Use {@link #getEventMetadataMap()} instead.
+     */
+    @java.lang.Deprecated
+    public java.util.Map<java.lang.Long, org.tensorflow.proto.profiler.XEventMetadata> getEventMetadata() {
+      return getEventMetadataMap();
+    }
+    /**
+     * <pre>
+     * XEventMetadata map, each entry uses the XEventMetadata.id as key. This map
+     * should be used for events that share the same ID over the whole XPlane.
+     * </pre>
+     *
+     * <code>map&lt;int64, .tensorflow.profiler.XEventMetadata&gt; event_metadata = 4;</code>
+     */
+
+    public java.util.Map<java.lang.Long, org.tensorflow.proto.profiler.XEventMetadata> getEventMetadataMap() {
+      return internalGetEventMetadata().getMap();
+    }
+    /**
+     * <pre>
+     * XEventMetadata map, each entry uses the XEventMetadata.id as key. This map
+     * should be used for events that share the same ID over the whole XPlane.
+     * </pre>
+     *
+     * <code>map&lt;int64, .tensorflow.profiler.XEventMetadata&gt; event_metadata = 4;</code>
+     */
+
+    public org.tensorflow.proto.profiler.XEventMetadata getEventMetadataOrDefault(
+        long key,
+        org.tensorflow.proto.profiler.XEventMetadata defaultValue) {
+      
+      java.util.Map<java.lang.Long, org.tensorflow.proto.profiler.XEventMetadata> map =
+          internalGetEventMetadata().getMap();
+      return map.containsKey(key) ? map.get(key) : defaultValue;
+    }
+    /**
+     * <pre>
+     * XEventMetadata map, each entry uses the XEventMetadata.id as key. This map
+     * should be used for events that share the same ID over the whole XPlane.
+     * </pre>
+     *
+     * <code>map&lt;int64, .tensorflow.profiler.XEventMetadata&gt; event_metadata = 4;</code>
+     */
+
+    public org.tensorflow.proto.profiler.XEventMetadata getEventMetadataOrThrow(
+        long key) {
+      
+      java.util.Map<java.lang.Long, org.tensorflow.proto.profiler.XEventMetadata> map =
+          internalGetEventMetadata().getMap();
+      if (!map.containsKey(key)) {
+        throw new java.lang.IllegalArgumentException();
+      }
+      return map.get(key);
+    }
+
+    public Builder clearEventMetadata() {
+      internalGetMutableEventMetadata().getMutableMap()
+          .clear();
+      return this;
+    }
+    /**
+     * <pre>
+     * XEventMetadata map, each entry uses the XEventMetadata.id as key. This map
+     * should be used for events that share the same ID over the whole XPlane.
+     * </pre>
+     *
+     * <code>map&lt;int64, .tensorflow.profiler.XEventMetadata&gt; event_metadata = 4;</code>
+     */
+
+    public Builder removeEventMetadata(
+        long key) {
+      
+      internalGetMutableEventMetadata().getMutableMap()
+          .remove(key);
+      return this;
+    }
+    /**
+     * Use alternate mutation accessors instead.
+     */
+    @java.lang.Deprecated
+    public java.util.Map<java.lang.Long, org.tensorflow.proto.profiler.XEventMetadata>
+    getMutableEventMetadata() {
+      return internalGetMutableEventMetadata().getMutableMap();
+    }
+    /**
+     * <pre>
+     * XEventMetadata map, each entry uses the XEventMetadata.id as key. This map
+     * should be used for events that share the same ID over the whole XPlane.
+     * </pre>
+     *
+     * <code>map&lt;int64, .tensorflow.profiler.XEventMetadata&gt; event_metadata = 4;</code>
+     */
+    public Builder putEventMetadata(
+        long key,
+        org.tensorflow.proto.profiler.XEventMetadata value) {
+      
+      if (value == null) { throw new java.lang.NullPointerException(); }
+      internalGetMutableEventMetadata().getMutableMap()
+          .put(key, value);
+      return this;
+    }
+    /**
+     * <pre>
+     * XEventMetadata map, each entry uses the XEventMetadata.id as key. This map
+     * should be used for events that share the same ID over the whole XPlane.
+     * </pre>
+     *
+     * <code>map&lt;int64, .tensorflow.profiler.XEventMetadata&gt; event_metadata = 4;</code>
+     */
+
+    public Builder putAllEventMetadata(
+        java.util.Map<java.lang.Long, org.tensorflow.proto.profiler.XEventMetadata> values) {
+      internalGetMutableEventMetadata().getMutableMap()
+          .putAll(values);
+      return this;
+    }
+
+    private com.google.protobuf.MapField<
+        java.lang.Long, org.tensorflow.proto.profiler.XStatMetadata> statMetadata_;
+    private com.google.protobuf.MapField<java.lang.Long, org.tensorflow.proto.profiler.XStatMetadata>
+    internalGetStatMetadata() {
+      if (statMetadata_ == null) {
+        return com.google.protobuf.MapField.emptyMapField(
+            StatMetadataDefaultEntryHolder.defaultEntry);
+      }
+      return statMetadata_;
+    }
+    private com.google.protobuf.MapField<java.lang.Long, org.tensorflow.proto.profiler.XStatMetadata>
+    internalGetMutableStatMetadata() {
+      onChanged();;
+      if (statMetadata_ == null) {
+        statMetadata_ = com.google.protobuf.MapField.newMapField(
+            StatMetadataDefaultEntryHolder.defaultEntry);
+      }
+      if (!statMetadata_.isMutable()) {
+        statMetadata_ = statMetadata_.copy();
+      }
+      return statMetadata_;
+    }
+
+    public int getStatMetadataCount() {
+      return internalGetStatMetadata().getMap().size();
+    }
+    /**
+     * <pre>
+     * XStatMetadata map, each entry uses the XStatMetadata.id as key. This map
+     * should be used for stats that share the same ID over the whole XPlane.
+     * </pre>
+     *
+     * <code>map&lt;int64, .tensorflow.profiler.XStatMetadata&gt; stat_metadata = 5;</code>
+     */
+
+    public boolean containsStatMetadata(
+        long key) {
+      
+      return internalGetStatMetadata().getMap().containsKey(key);
+    }
+    /**
+     * Use {@link #getStatMetadataMap()} instead.
+     */
+    @java.lang.Deprecated
+    public java.util.Map<java.lang.Long, org.tensorflow.proto.profiler.XStatMetadata> getStatMetadata() {
+      return getStatMetadataMap();
+    }
+    /**
+     * <pre>
+     * XStatMetadata map, each entry uses the XStatMetadata.id as key. This map
+     * should be used for stats that share the same ID over the whole XPlane.
+     * </pre>
+     *
+     * <code>map&lt;int64, .tensorflow.profiler.XStatMetadata&gt; stat_metadata = 5;</code>
+     */
+
+    public java.util.Map<java.lang.Long, org.tensorflow.proto.profiler.XStatMetadata> getStatMetadataMap() {
+      return internalGetStatMetadata().getMap();
+    }
+    /**
+     * <pre>
+     * XStatMetadata map, each entry uses the XStatMetadata.id as key. This map
+     * should be used for stats that share the same ID over the whole XPlane.
+     * </pre>
+     *
+     * <code>map&lt;int64, .tensorflow.profiler.XStatMetadata&gt; stat_metadata = 5;</code>
+     */
+
+    public org.tensorflow.proto.profiler.XStatMetadata getStatMetadataOrDefault(
+        long key,
+        org.tensorflow.proto.profiler.XStatMetadata defaultValue) {
+      
+      java.util.Map<java.lang.Long, org.tensorflow.proto.profiler.XStatMetadata> map =
+          internalGetStatMetadata().getMap();
+      return map.containsKey(key) ? map.get(key) : defaultValue;
+    }
+    /**
+     * <pre>
+     * XStatMetadata map, each entry uses the XStatMetadata.id as key. This map
+     * should be used for stats that share the same ID over the whole XPlane.
+     * </pre>
+     *
+     * <code>map&lt;int64, .tensorflow.profiler.XStatMetadata&gt; stat_metadata = 5;</code>
+     */
+
+    public org.tensorflow.proto.profiler.XStatMetadata getStatMetadataOrThrow(
+        long key) {
+      
+      java.util.Map<java.lang.Long, org.tensorflow.proto.profiler.XStatMetadata> map =
+          internalGetStatMetadata().getMap();
+      if (!map.containsKey(key)) {
+        throw new java.lang.IllegalArgumentException();
+      }
+      return map.get(key);
+    }
+
+    public Builder clearStatMetadata() {
+      internalGetMutableStatMetadata().getMutableMap()
+          .clear();
+      return this;
+    }
+    /**
+     * <pre>
+     * XStatMetadata map, each entry uses the XStatMetadata.id as key. This map
+     * should be used for stats that share the same ID over the whole XPlane.
+     * </pre>
+     *
+     * <code>map&lt;int64, .tensorflow.profiler.XStatMetadata&gt; stat_metadata = 5;</code>
+     */
+
+    public Builder removeStatMetadata(
+        long key) {
+      
+      internalGetMutableStatMetadata().getMutableMap()
+          .remove(key);
+      return this;
+    }
+    /**
+     * Use alternate mutation accessors instead.
+     */
+    @java.lang.Deprecated
+    public java.util.Map<java.lang.Long, org.tensorflow.proto.profiler.XStatMetadata>
+    getMutableStatMetadata() {
+      return internalGetMutableStatMetadata().getMutableMap();
+    }
+    /**
+     * <pre>
+     * XStatMetadata map, each entry uses the XStatMetadata.id as key. This map
+     * should be used for stats that share the same ID over the whole XPlane.
+     * </pre>
+     *
+     * <code>map&lt;int64, .tensorflow.profiler.XStatMetadata&gt; stat_metadata = 5;</code>
+     */
+    public Builder putStatMetadata(
+        long key,
+        org.tensorflow.proto.profiler.XStatMetadata value) {
+      
+      if (value == null) { throw new java.lang.NullPointerException(); }
+      internalGetMutableStatMetadata().getMutableMap()
+          .put(key, value);
+      return this;
+    }
+    /**
+     * <pre>
+     * XStatMetadata map, each entry uses the XStatMetadata.id as key. This map
+     * should be used for stats that share the same ID over the whole XPlane.
+     * </pre>
+     *
+     * <code>map&lt;int64, .tensorflow.profiler.XStatMetadata&gt; stat_metadata = 5;</code>
+     */
+
+    public Builder putAllStatMetadata(
+        java.util.Map<java.lang.Long, org.tensorflow.proto.profiler.XStatMetadata> values) {
+      internalGetMutableStatMetadata().getMutableMap()
+          .putAll(values);
+      return this;
+    }
+
+    private java.util.List<org.tensorflow.proto.profiler.XStat> stats_ =
+      java.util.Collections.emptyList();
+    private void ensureStatsIsMutable() {
+      if (!((bitField0_ & 0x00000008) != 0)) {
+        stats_ = new java.util.ArrayList<org.tensorflow.proto.profiler.XStat>(stats_);
+        bitField0_ |= 0x00000008;
+       }
+    }
+
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.profiler.XStat, org.tensorflow.proto.profiler.XStat.Builder, org.tensorflow.proto.profiler.XStatOrBuilder> statsBuilder_;
+
+    /**
+     * <pre>
+     * XStats associated with this plane, e.g. device capabilities.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.profiler.XStat stats = 6;</code>
+     */
+    public java.util.List<org.tensorflow.proto.profiler.XStat> getStatsList() {
+      if (statsBuilder_ == null) {
+        return java.util.Collections.unmodifiableList(stats_);
+      } else {
+        return statsBuilder_.getMessageList();
+      }
+    }
+    /**
+     * <pre>
+     * XStats associated with this plane, e.g. device capabilities.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.profiler.XStat stats = 6;</code>
+     */
+    public int getStatsCount() {
+      if (statsBuilder_ == null) {
+        return stats_.size();
+      } else {
+        return statsBuilder_.getCount();
+      }
+    }
+    /**
+     * <pre>
+     * XStats associated with this plane, e.g. device capabilities.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.profiler.XStat stats = 6;</code>
+     */
+    public org.tensorflow.proto.profiler.XStat getStats(int index) {
+      if (statsBuilder_ == null) {
+        return stats_.get(index);
+      } else {
+        return statsBuilder_.getMessage(index);
+      }
+    }
+    /**
+     * <pre>
+     * XStats associated with this plane, e.g. device capabilities.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.profiler.XStat stats = 6;</code>
+     */
+    public Builder setStats(
+        int index, org.tensorflow.proto.profiler.XStat value) {
+      if (statsBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureStatsIsMutable();
+        stats_.set(index, value);
+        onChanged();
+      } else {
+        statsBuilder_.setMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * XStats associated with this plane, e.g. device capabilities.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.profiler.XStat stats = 6;</code>
+     */
+    public Builder setStats(
+        int index, org.tensorflow.proto.profiler.XStat.Builder builderForValue) {
+      if (statsBuilder_ == null) {
+        ensureStatsIsMutable();
+        stats_.set(index, builderForValue.build());
+        onChanged();
+      } else {
+        statsBuilder_.setMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * XStats associated with this plane, e.g. device capabilities.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.profiler.XStat stats = 6;</code>
+     */
+    public Builder addStats(org.tensorflow.proto.profiler.XStat value) {
+      if (statsBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureStatsIsMutable();
+        stats_.add(value);
+        onChanged();
+      } else {
+        statsBuilder_.addMessage(value);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * XStats associated with this plane, e.g. device capabilities.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.profiler.XStat stats = 6;</code>
+     */
+    public Builder addStats(
+        int index, org.tensorflow.proto.profiler.XStat value) {
+      if (statsBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureStatsIsMutable();
+        stats_.add(index, value);
+        onChanged();
+      } else {
+        statsBuilder_.addMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * XStats associated with this plane, e.g. device capabilities.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.profiler.XStat stats = 6;</code>
+     */
+    public Builder addStats(
+        org.tensorflow.proto.profiler.XStat.Builder builderForValue) {
+      if (statsBuilder_ == null) {
+        ensureStatsIsMutable();
+        stats_.add(builderForValue.build());
+        onChanged();
+      } else {
+        statsBuilder_.addMessage(builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * XStats associated with this plane, e.g. device capabilities.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.profiler.XStat stats = 6;</code>
+     */
+    public Builder addStats(
+        int index, org.tensorflow.proto.profiler.XStat.Builder builderForValue) {
+      if (statsBuilder_ == null) {
+        ensureStatsIsMutable();
+        stats_.add(index, builderForValue.build());
+        onChanged();
+      } else {
+        statsBuilder_.addMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * XStats associated with this plane, e.g. device capabilities.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.profiler.XStat stats = 6;</code>
+     */
+    public Builder addAllStats(
+        java.lang.Iterable<? extends org.tensorflow.proto.profiler.XStat> values) {
+      if (statsBuilder_ == null) {
+        ensureStatsIsMutable();
+        com.google.protobuf.AbstractMessageLite.Builder.addAll(
+            values, stats_);
+        onChanged();
+      } else {
+        statsBuilder_.addAllMessages(values);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * XStats associated with this plane, e.g. device capabilities.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.profiler.XStat stats = 6;</code>
+     */
+    public Builder clearStats() {
+      if (statsBuilder_ == null) {
+        stats_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000008);
+        onChanged();
+      } else {
+        statsBuilder_.clear();
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * XStats associated with this plane, e.g. device capabilities.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.profiler.XStat stats = 6;</code>
+     */
+    public Builder removeStats(int index) {
+      if (statsBuilder_ == null) {
+        ensureStatsIsMutable();
+        stats_.remove(index);
+        onChanged();
+      } else {
+        statsBuilder_.remove(index);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * XStats associated with this plane, e.g. device capabilities.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.profiler.XStat stats = 6;</code>
+     */
+    public org.tensorflow.proto.profiler.XStat.Builder getStatsBuilder(
+        int index) {
+      return getStatsFieldBuilder().getBuilder(index);
+    }
+    /**
+     * <pre>
+     * XStats associated with this plane, e.g. device capabilities.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.profiler.XStat stats = 6;</code>
+     */
+    public org.tensorflow.proto.profiler.XStatOrBuilder getStatsOrBuilder(
+        int index) {
+      if (statsBuilder_ == null) {
+        return stats_.get(index);  } else {
+        return statsBuilder_.getMessageOrBuilder(index);
+      }
+    }
+    /**
+     * <pre>
+     * XStats associated with this plane, e.g. device capabilities.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.profiler.XStat stats = 6;</code>
+     */
+    public java.util.List<? extends org.tensorflow.proto.profiler.XStatOrBuilder> 
+         getStatsOrBuilderList() {
+      if (statsBuilder_ != null) {
+        return statsBuilder_.getMessageOrBuilderList();
+      } else {
+        return java.util.Collections.unmodifiableList(stats_);
+      }
+    }
+    /**
+     * <pre>
+     * XStats associated with this plane, e.g. device capabilities.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.profiler.XStat stats = 6;</code>
+     */
+    public org.tensorflow.proto.profiler.XStat.Builder addStatsBuilder() {
+      return getStatsFieldBuilder().addBuilder(
+          org.tensorflow.proto.profiler.XStat.getDefaultInstance());
+    }
+    /**
+     * <pre>
+     * XStats associated with this plane, e.g. device capabilities.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.profiler.XStat stats = 6;</code>
+     */
+    public org.tensorflow.proto.profiler.XStat.Builder addStatsBuilder(
+        int index) {
+      return getStatsFieldBuilder().addBuilder(
+          index, org.tensorflow.proto.profiler.XStat.getDefaultInstance());
+    }
+    /**
+     * <pre>
+     * XStats associated with this plane, e.g. device capabilities.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.profiler.XStat stats = 6;</code>
+     */
+    public java.util.List<org.tensorflow.proto.profiler.XStat.Builder> 
+         getStatsBuilderList() {
+      return getStatsFieldBuilder().getBuilderList();
+    }
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.profiler.XStat, org.tensorflow.proto.profiler.XStat.Builder, org.tensorflow.proto.profiler.XStatOrBuilder> 
+        getStatsFieldBuilder() {
+      if (statsBuilder_ == null) {
+        statsBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
+            org.tensorflow.proto.profiler.XStat, org.tensorflow.proto.profiler.XStat.Builder, org.tensorflow.proto.profiler.XStatOrBuilder>(
+                stats_,
+                ((bitField0_ & 0x00000008) != 0),
+                getParentForChildren(),
+                isClean());
+        stats_ = null;
+      }
+      return statsBuilder_;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.profiler.XPlane)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.profiler.XPlane)
+  private static final org.tensorflow.proto.profiler.XPlane DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.profiler.XPlane();
+  }
+
+  public static org.tensorflow.proto.profiler.XPlane getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<XPlane>
+      PARSER = new com.google.protobuf.AbstractParser<XPlane>() {
+    @java.lang.Override
+    public XPlane parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new XPlane(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<XPlane> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<XPlane> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.profiler.XPlane getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/XPlaneOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/XPlaneOrBuilder.java
new file mode 100644
index 00000000000..a6b97a4c6dd
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/XPlaneOrBuilder.java
@@ -0,0 +1,243 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/profiler/protobuf/xplane.proto
+
+package org.tensorflow.proto.profiler;
+
+public interface XPlaneOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.profiler.XPlane)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <code>int64 id = 1;</code>
+   */
+  long getId();
+
+  /**
+   * <pre>
+   * Name of this line.
+   * </pre>
+   *
+   * <code>string name = 2;</code>
+   */
+  java.lang.String getName();
+  /**
+   * <pre>
+   * Name of this line.
+   * </pre>
+   *
+   * <code>string name = 2;</code>
+   */
+  com.google.protobuf.ByteString
+      getNameBytes();
+
+  /**
+   * <pre>
+   * Parallel timelines grouped in this plane. XLines with the same id
+   * are effectively the same timeline.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.profiler.XLine lines = 3;</code>
+   */
+  java.util.List<org.tensorflow.proto.profiler.XLine> 
+      getLinesList();
+  /**
+   * <pre>
+   * Parallel timelines grouped in this plane. XLines with the same id
+   * are effectively the same timeline.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.profiler.XLine lines = 3;</code>
+   */
+  org.tensorflow.proto.profiler.XLine getLines(int index);
+  /**
+   * <pre>
+   * Parallel timelines grouped in this plane. XLines with the same id
+   * are effectively the same timeline.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.profiler.XLine lines = 3;</code>
+   */
+  int getLinesCount();
+  /**
+   * <pre>
+   * Parallel timelines grouped in this plane. XLines with the same id
+   * are effectively the same timeline.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.profiler.XLine lines = 3;</code>
+   */
+  java.util.List<? extends org.tensorflow.proto.profiler.XLineOrBuilder> 
+      getLinesOrBuilderList();
+  /**
+   * <pre>
+   * Parallel timelines grouped in this plane. XLines with the same id
+   * are effectively the same timeline.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.profiler.XLine lines = 3;</code>
+   */
+  org.tensorflow.proto.profiler.XLineOrBuilder getLinesOrBuilder(
+      int index);
+
+  /**
+   * <pre>
+   * XEventMetadata map, each entry uses the XEventMetadata.id as key. This map
+   * should be used for events that share the same ID over the whole XPlane.
+   * </pre>
+   *
+   * <code>map&lt;int64, .tensorflow.profiler.XEventMetadata&gt; event_metadata = 4;</code>
+   */
+  int getEventMetadataCount();
+  /**
+   * <pre>
+   * XEventMetadata map, each entry uses the XEventMetadata.id as key. This map
+   * should be used for events that share the same ID over the whole XPlane.
+   * </pre>
+   *
+   * <code>map&lt;int64, .tensorflow.profiler.XEventMetadata&gt; event_metadata = 4;</code>
+   */
+  boolean containsEventMetadata(
+      long key);
+  /**
+   * Use {@link #getEventMetadataMap()} instead.
+   */
+  @java.lang.Deprecated
+  java.util.Map<java.lang.Long, org.tensorflow.proto.profiler.XEventMetadata>
+  getEventMetadata();
+  /**
+   * <pre>
+   * XEventMetadata map, each entry uses the XEventMetadata.id as key. This map
+   * should be used for events that share the same ID over the whole XPlane.
+   * </pre>
+   *
+   * <code>map&lt;int64, .tensorflow.profiler.XEventMetadata&gt; event_metadata = 4;</code>
+   */
+  java.util.Map<java.lang.Long, org.tensorflow.proto.profiler.XEventMetadata>
+  getEventMetadataMap();
+  /**
+   * <pre>
+   * XEventMetadata map, each entry uses the XEventMetadata.id as key. This map
+   * should be used for events that share the same ID over the whole XPlane.
+   * </pre>
+   *
+   * <code>map&lt;int64, .tensorflow.profiler.XEventMetadata&gt; event_metadata = 4;</code>
+   */
+
+  org.tensorflow.proto.profiler.XEventMetadata getEventMetadataOrDefault(
+      long key,
+      org.tensorflow.proto.profiler.XEventMetadata defaultValue);
+  /**
+   * <pre>
+   * XEventMetadata map, each entry uses the XEventMetadata.id as key. This map
+   * should be used for events that share the same ID over the whole XPlane.
+   * </pre>
+   *
+   * <code>map&lt;int64, .tensorflow.profiler.XEventMetadata&gt; event_metadata = 4;</code>
+   */
+
+  org.tensorflow.proto.profiler.XEventMetadata getEventMetadataOrThrow(
+      long key);
+
+  /**
+   * <pre>
+   * XStatMetadata map, each entry uses the XStatMetadata.id as key. This map
+   * should be used for stats that share the same ID over the whole XPlane.
+   * </pre>
+   *
+   * <code>map&lt;int64, .tensorflow.profiler.XStatMetadata&gt; stat_metadata = 5;</code>
+   */
+  int getStatMetadataCount();
+  /**
+   * <pre>
+   * XStatMetadata map, each entry uses the XStatMetadata.id as key. This map
+   * should be used for stats that share the same ID over the whole XPlane.
+   * </pre>
+   *
+   * <code>map&lt;int64, .tensorflow.profiler.XStatMetadata&gt; stat_metadata = 5;</code>
+   */
+  boolean containsStatMetadata(
+      long key);
+  /**
+   * Use {@link #getStatMetadataMap()} instead.
+   */
+  @java.lang.Deprecated
+  java.util.Map<java.lang.Long, org.tensorflow.proto.profiler.XStatMetadata>
+  getStatMetadata();
+  /**
+   * <pre>
+   * XStatMetadata map, each entry uses the XStatMetadata.id as key. This map
+   * should be used for stats that share the same ID over the whole XPlane.
+   * </pre>
+   *
+   * <code>map&lt;int64, .tensorflow.profiler.XStatMetadata&gt; stat_metadata = 5;</code>
+   */
+  java.util.Map<java.lang.Long, org.tensorflow.proto.profiler.XStatMetadata>
+  getStatMetadataMap();
+  /**
+   * <pre>
+   * XStatMetadata map, each entry uses the XStatMetadata.id as key. This map
+   * should be used for stats that share the same ID over the whole XPlane.
+   * </pre>
+   *
+   * <code>map&lt;int64, .tensorflow.profiler.XStatMetadata&gt; stat_metadata = 5;</code>
+   */
+
+  org.tensorflow.proto.profiler.XStatMetadata getStatMetadataOrDefault(
+      long key,
+      org.tensorflow.proto.profiler.XStatMetadata defaultValue);
+  /**
+   * <pre>
+   * XStatMetadata map, each entry uses the XStatMetadata.id as key. This map
+   * should be used for stats that share the same ID over the whole XPlane.
+   * </pre>
+   *
+   * <code>map&lt;int64, .tensorflow.profiler.XStatMetadata&gt; stat_metadata = 5;</code>
+   */
+
+  org.tensorflow.proto.profiler.XStatMetadata getStatMetadataOrThrow(
+      long key);
+
+  /**
+   * <pre>
+   * XStats associated with this plane, e.g. device capabilities.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.profiler.XStat stats = 6;</code>
+   */
+  java.util.List<org.tensorflow.proto.profiler.XStat> 
+      getStatsList();
+  /**
+   * <pre>
+   * XStats associated with this plane, e.g. device capabilities.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.profiler.XStat stats = 6;</code>
+   */
+  org.tensorflow.proto.profiler.XStat getStats(int index);
+  /**
+   * <pre>
+   * XStats associated with this plane, e.g. device capabilities.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.profiler.XStat stats = 6;</code>
+   */
+  int getStatsCount();
+  /**
+   * <pre>
+   * XStats associated with this plane, e.g. device capabilities.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.profiler.XStat stats = 6;</code>
+   */
+  java.util.List<? extends org.tensorflow.proto.profiler.XStatOrBuilder> 
+      getStatsOrBuilderList();
+  /**
+   * <pre>
+   * XStats associated with this plane, e.g. device capabilities.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.profiler.XStat stats = 6;</code>
+   */
+  org.tensorflow.proto.profiler.XStatOrBuilder getStatsOrBuilder(
+      int index);
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/XPlaneProtos.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/XPlaneProtos.java
new file mode 100644
index 00000000000..4fdb2cce99d
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/XPlaneProtos.java
@@ -0,0 +1,167 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/profiler/protobuf/xplane.proto
+
+package org.tensorflow.proto.profiler;
+
+public final class XPlaneProtos {
+  private XPlaneProtos() {}
+  public static void registerAllExtensions(
+      com.google.protobuf.ExtensionRegistryLite registry) {
+  }
+
+  public static void registerAllExtensions(
+      com.google.protobuf.ExtensionRegistry registry) {
+    registerAllExtensions(
+        (com.google.protobuf.ExtensionRegistryLite) registry);
+  }
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_profiler_XSpace_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_profiler_XSpace_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_profiler_XPlane_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_profiler_XPlane_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_profiler_XPlane_EventMetadataEntry_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_profiler_XPlane_EventMetadataEntry_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_profiler_XPlane_StatMetadataEntry_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_profiler_XPlane_StatMetadataEntry_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_profiler_XLine_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_profiler_XLine_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_profiler_XEvent_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_profiler_XEvent_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_profiler_XStat_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_profiler_XStat_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_profiler_XEventMetadata_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_profiler_XEventMetadata_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_profiler_XStatMetadata_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_profiler_XStatMetadata_fieldAccessorTable;
+
+  public static com.google.protobuf.Descriptors.FileDescriptor
+      getDescriptor() {
+    return descriptor;
+  }
+  private static  com.google.protobuf.Descriptors.FileDescriptor
+      descriptor;
+  static {
+    java.lang.String[] descriptorData = {
+      "\n.tensorflow/core/profiler/protobuf/xpla" +
+      "ne.proto\022\023tensorflow.profiler\"W\n\006XSpace\022" +
+      "+\n\006planes\030\001 \003(\0132\033.tensorflow.profiler.XP" +
+      "lane\022\016\n\006errors\030\002 \003(\t\022\020\n\010warnings\030\003 \003(\t\"\272" +
+      "\003\n\006XPlane\022\n\n\002id\030\001 \001(\003\022\014\n\004name\030\002 \001(\t\022)\n\005l" +
+      "ines\030\003 \003(\0132\032.tensorflow.profiler.XLine\022F" +
+      "\n\016event_metadata\030\004 \003(\0132..tensorflow.prof" +
+      "iler.XPlane.EventMetadataEntry\022D\n\rstat_m" +
+      "etadata\030\005 \003(\0132-.tensorflow.profiler.XPla" +
+      "ne.StatMetadataEntry\022)\n\005stats\030\006 \003(\0132\032.te" +
+      "nsorflow.profiler.XStat\032Y\n\022EventMetadata" +
+      "Entry\022\013\n\003key\030\001 \001(\003\0222\n\005value\030\002 \001(\0132#.tens" +
+      "orflow.profiler.XEventMetadata:\0028\001\032W\n\021St" +
+      "atMetadataEntry\022\013\n\003key\030\001 \001(\003\0221\n\005value\030\002 " +
+      "\001(\0132\".tensorflow.profiler.XStatMetadata:" +
+      "\0028\001\"\273\001\n\005XLine\022\n\n\002id\030\001 \001(\003\022\022\n\ndisplay_id\030" +
+      "\n \001(\003\022\014\n\004name\030\002 \001(\t\022\024\n\014display_name\030\013 \001(" +
+      "\t\022\024\n\014timestamp_ns\030\003 \001(\003\022\023\n\013duration_ps\030\t" +
+      " \001(\003\022+\n\006events\030\004 \003(\0132\033.tensorflow.profil" +
+      "er.XEventJ\004\010\005\020\006J\004\010\006\020\007J\004\010\007\020\010J\004\010\010\020\t\"\225\001\n\006XE" +
+      "vent\022\023\n\013metadata_id\030\001 \001(\003\022\023\n\toffset_ps\030\002" +
+      " \001(\003H\000\022\031\n\017num_occurrences\030\005 \001(\003H\000\022\023\n\013dur" +
+      "ation_ps\030\003 \001(\003\022)\n\005stats\030\004 \003(\0132\032.tensorfl" +
+      "ow.profiler.XStatB\006\n\004data\"\255\001\n\005XStat\022\023\n\013m" +
+      "etadata_id\030\001 \001(\003\022\026\n\014double_value\030\002 \001(\001H\000" +
+      "\022\026\n\014uint64_value\030\003 \001(\004H\000\022\025\n\013int64_value\030" +
+      "\004 \001(\003H\000\022\023\n\tstr_value\030\005 \001(\tH\000\022\025\n\013bytes_va" +
+      "lue\030\006 \001(\014H\000\022\023\n\tref_value\030\007 \001(\004H\000B\007\n\005valu" +
+      "e\"R\n\016XEventMetadata\022\n\n\002id\030\001 \001(\003\022\014\n\004name\030" +
+      "\002 \001(\t\022\024\n\014display_name\030\004 \001(\t\022\020\n\010metadata\030" +
+      "\003 \001(\014\">\n\rXStatMetadata\022\n\n\002id\030\001 \001(\003\022\014\n\004na" +
+      "me\030\002 \001(\t\022\023\n\013description\030\003 \001(\tB2\n\035org.ten" +
+      "sorflow.proto.profilerB\014XPlaneProtosP\001\370\001" +
+      "\001b\006proto3"
+    };
+    descriptor = com.google.protobuf.Descriptors.FileDescriptor
+      .internalBuildGeneratedFileFrom(descriptorData,
+        new com.google.protobuf.Descriptors.FileDescriptor[] {
+        });
+    internal_static_tensorflow_profiler_XSpace_descriptor =
+      getDescriptor().getMessageTypes().get(0);
+    internal_static_tensorflow_profiler_XSpace_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_profiler_XSpace_descriptor,
+        new java.lang.String[] { "Planes", "Errors", "Warnings", });
+    internal_static_tensorflow_profiler_XPlane_descriptor =
+      getDescriptor().getMessageTypes().get(1);
+    internal_static_tensorflow_profiler_XPlane_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_profiler_XPlane_descriptor,
+        new java.lang.String[] { "Id", "Name", "Lines", "EventMetadata", "StatMetadata", "Stats", });
+    internal_static_tensorflow_profiler_XPlane_EventMetadataEntry_descriptor =
+      internal_static_tensorflow_profiler_XPlane_descriptor.getNestedTypes().get(0);
+    internal_static_tensorflow_profiler_XPlane_EventMetadataEntry_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_profiler_XPlane_EventMetadataEntry_descriptor,
+        new java.lang.String[] { "Key", "Value", });
+    internal_static_tensorflow_profiler_XPlane_StatMetadataEntry_descriptor =
+      internal_static_tensorflow_profiler_XPlane_descriptor.getNestedTypes().get(1);
+    internal_static_tensorflow_profiler_XPlane_StatMetadataEntry_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_profiler_XPlane_StatMetadataEntry_descriptor,
+        new java.lang.String[] { "Key", "Value", });
+    internal_static_tensorflow_profiler_XLine_descriptor =
+      getDescriptor().getMessageTypes().get(2);
+    internal_static_tensorflow_profiler_XLine_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_profiler_XLine_descriptor,
+        new java.lang.String[] { "Id", "DisplayId", "Name", "DisplayName", "TimestampNs", "DurationPs", "Events", });
+    internal_static_tensorflow_profiler_XEvent_descriptor =
+      getDescriptor().getMessageTypes().get(3);
+    internal_static_tensorflow_profiler_XEvent_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_profiler_XEvent_descriptor,
+        new java.lang.String[] { "MetadataId", "OffsetPs", "NumOccurrences", "DurationPs", "Stats", "Data", });
+    internal_static_tensorflow_profiler_XStat_descriptor =
+      getDescriptor().getMessageTypes().get(4);
+    internal_static_tensorflow_profiler_XStat_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_profiler_XStat_descriptor,
+        new java.lang.String[] { "MetadataId", "DoubleValue", "Uint64Value", "Int64Value", "StrValue", "BytesValue", "RefValue", "Value", });
+    internal_static_tensorflow_profiler_XEventMetadata_descriptor =
+      getDescriptor().getMessageTypes().get(5);
+    internal_static_tensorflow_profiler_XEventMetadata_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_profiler_XEventMetadata_descriptor,
+        new java.lang.String[] { "Id", "Name", "DisplayName", "Metadata", });
+    internal_static_tensorflow_profiler_XStatMetadata_descriptor =
+      getDescriptor().getMessageTypes().get(6);
+    internal_static_tensorflow_profiler_XStatMetadata_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_profiler_XStatMetadata_descriptor,
+        new java.lang.String[] { "Id", "Name", "Description", });
+  }
+
+  // @@protoc_insertion_point(outer_class_scope)
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/XSpace.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/XSpace.java
new file mode 100644
index 00000000000..956b5ab793b
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/XSpace.java
@@ -0,0 +1,1219 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/profiler/protobuf/xplane.proto
+
+package org.tensorflow.proto.profiler;
+
+/**
+ * <pre>
+ * A container of parallel XPlanes, generated by one or more profiling sources.
+ * Next ID: 4
+ * </pre>
+ *
+ * Protobuf type {@code tensorflow.profiler.XSpace}
+ */
+public  final class XSpace extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.profiler.XSpace)
+    XSpaceOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use XSpace.newBuilder() to construct.
+  private XSpace(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private XSpace() {
+    planes_ = java.util.Collections.emptyList();
+    errors_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+    warnings_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new XSpace();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private XSpace(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    int mutable_bitField0_ = 0;
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 10: {
+            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
+              planes_ = new java.util.ArrayList<org.tensorflow.proto.profiler.XPlane>();
+              mutable_bitField0_ |= 0x00000001;
+            }
+            planes_.add(
+                input.readMessage(org.tensorflow.proto.profiler.XPlane.parser(), extensionRegistry));
+            break;
+          }
+          case 18: {
+            java.lang.String s = input.readStringRequireUtf8();
+            if (!((mutable_bitField0_ & 0x00000002) != 0)) {
+              errors_ = new com.google.protobuf.LazyStringArrayList();
+              mutable_bitField0_ |= 0x00000002;
+            }
+            errors_.add(s);
+            break;
+          }
+          case 26: {
+            java.lang.String s = input.readStringRequireUtf8();
+            if (!((mutable_bitField0_ & 0x00000004) != 0)) {
+              warnings_ = new com.google.protobuf.LazyStringArrayList();
+              mutable_bitField0_ |= 0x00000004;
+            }
+            warnings_.add(s);
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      if (((mutable_bitField0_ & 0x00000001) != 0)) {
+        planes_ = java.util.Collections.unmodifiableList(planes_);
+      }
+      if (((mutable_bitField0_ & 0x00000002) != 0)) {
+        errors_ = errors_.getUnmodifiableView();
+      }
+      if (((mutable_bitField0_ & 0x00000004) != 0)) {
+        warnings_ = warnings_.getUnmodifiableView();
+      }
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.profiler.XPlaneProtos.internal_static_tensorflow_profiler_XSpace_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.profiler.XPlaneProtos.internal_static_tensorflow_profiler_XSpace_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.profiler.XSpace.class, org.tensorflow.proto.profiler.XSpace.Builder.class);
+  }
+
+  public static final int PLANES_FIELD_NUMBER = 1;
+  private java.util.List<org.tensorflow.proto.profiler.XPlane> planes_;
+  /**
+   * <code>repeated .tensorflow.profiler.XPlane planes = 1;</code>
+   */
+  public java.util.List<org.tensorflow.proto.profiler.XPlane> getPlanesList() {
+    return planes_;
+  }
+  /**
+   * <code>repeated .tensorflow.profiler.XPlane planes = 1;</code>
+   */
+  public java.util.List<? extends org.tensorflow.proto.profiler.XPlaneOrBuilder> 
+      getPlanesOrBuilderList() {
+    return planes_;
+  }
+  /**
+   * <code>repeated .tensorflow.profiler.XPlane planes = 1;</code>
+   */
+  public int getPlanesCount() {
+    return planes_.size();
+  }
+  /**
+   * <code>repeated .tensorflow.profiler.XPlane planes = 1;</code>
+   */
+  public org.tensorflow.proto.profiler.XPlane getPlanes(int index) {
+    return planes_.get(index);
+  }
+  /**
+   * <code>repeated .tensorflow.profiler.XPlane planes = 1;</code>
+   */
+  public org.tensorflow.proto.profiler.XPlaneOrBuilder getPlanesOrBuilder(
+      int index) {
+    return planes_.get(index);
+  }
+
+  public static final int ERRORS_FIELD_NUMBER = 2;
+  private com.google.protobuf.LazyStringList errors_;
+  /**
+   * <pre>
+   * Errors (if any) in the generation of planes.
+   * </pre>
+   *
+   * <code>repeated string errors = 2;</code>
+   */
+  public com.google.protobuf.ProtocolStringList
+      getErrorsList() {
+    return errors_;
+  }
+  /**
+   * <pre>
+   * Errors (if any) in the generation of planes.
+   * </pre>
+   *
+   * <code>repeated string errors = 2;</code>
+   */
+  public int getErrorsCount() {
+    return errors_.size();
+  }
+  /**
+   * <pre>
+   * Errors (if any) in the generation of planes.
+   * </pre>
+   *
+   * <code>repeated string errors = 2;</code>
+   */
+  public java.lang.String getErrors(int index) {
+    return errors_.get(index);
+  }
+  /**
+   * <pre>
+   * Errors (if any) in the generation of planes.
+   * </pre>
+   *
+   * <code>repeated string errors = 2;</code>
+   */
+  public com.google.protobuf.ByteString
+      getErrorsBytes(int index) {
+    return errors_.getByteString(index);
+  }
+
+  public static final int WARNINGS_FIELD_NUMBER = 3;
+  private com.google.protobuf.LazyStringList warnings_;
+  /**
+   * <pre>
+   * Warnings (if any) in the generation of planes;
+   * </pre>
+   *
+   * <code>repeated string warnings = 3;</code>
+   */
+  public com.google.protobuf.ProtocolStringList
+      getWarningsList() {
+    return warnings_;
+  }
+  /**
+   * <pre>
+   * Warnings (if any) in the generation of planes;
+   * </pre>
+   *
+   * <code>repeated string warnings = 3;</code>
+   */
+  public int getWarningsCount() {
+    return warnings_.size();
+  }
+  /**
+   * <pre>
+   * Warnings (if any) in the generation of planes;
+   * </pre>
+   *
+   * <code>repeated string warnings = 3;</code>
+   */
+  public java.lang.String getWarnings(int index) {
+    return warnings_.get(index);
+  }
+  /**
+   * <pre>
+   * Warnings (if any) in the generation of planes;
+   * </pre>
+   *
+   * <code>repeated string warnings = 3;</code>
+   */
+  public com.google.protobuf.ByteString
+      getWarningsBytes(int index) {
+    return warnings_.getByteString(index);
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    for (int i = 0; i < planes_.size(); i++) {
+      output.writeMessage(1, planes_.get(i));
+    }
+    for (int i = 0; i < errors_.size(); i++) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 2, errors_.getRaw(i));
+    }
+    for (int i = 0; i < warnings_.size(); i++) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 3, warnings_.getRaw(i));
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    for (int i = 0; i < planes_.size(); i++) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(1, planes_.get(i));
+    }
+    {
+      int dataSize = 0;
+      for (int i = 0; i < errors_.size(); i++) {
+        dataSize += computeStringSizeNoTag(errors_.getRaw(i));
+      }
+      size += dataSize;
+      size += 1 * getErrorsList().size();
+    }
+    {
+      int dataSize = 0;
+      for (int i = 0; i < warnings_.size(); i++) {
+        dataSize += computeStringSizeNoTag(warnings_.getRaw(i));
+      }
+      size += dataSize;
+      size += 1 * getWarningsList().size();
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.profiler.XSpace)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.profiler.XSpace other = (org.tensorflow.proto.profiler.XSpace) obj;
+
+    if (!getPlanesList()
+        .equals(other.getPlanesList())) return false;
+    if (!getErrorsList()
+        .equals(other.getErrorsList())) return false;
+    if (!getWarningsList()
+        .equals(other.getWarningsList())) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    if (getPlanesCount() > 0) {
+      hash = (37 * hash) + PLANES_FIELD_NUMBER;
+      hash = (53 * hash) + getPlanesList().hashCode();
+    }
+    if (getErrorsCount() > 0) {
+      hash = (37 * hash) + ERRORS_FIELD_NUMBER;
+      hash = (53 * hash) + getErrorsList().hashCode();
+    }
+    if (getWarningsCount() > 0) {
+      hash = (37 * hash) + WARNINGS_FIELD_NUMBER;
+      hash = (53 * hash) + getWarningsList().hashCode();
+    }
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.profiler.XSpace parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.profiler.XSpace parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.profiler.XSpace parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.profiler.XSpace parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.profiler.XSpace parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.profiler.XSpace parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.profiler.XSpace parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.profiler.XSpace parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.profiler.XSpace parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.profiler.XSpace parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.profiler.XSpace parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.profiler.XSpace parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.profiler.XSpace prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * <pre>
+   * A container of parallel XPlanes, generated by one or more profiling sources.
+   * Next ID: 4
+   * </pre>
+   *
+   * Protobuf type {@code tensorflow.profiler.XSpace}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.profiler.XSpace)
+      org.tensorflow.proto.profiler.XSpaceOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.profiler.XPlaneProtos.internal_static_tensorflow_profiler_XSpace_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.profiler.XPlaneProtos.internal_static_tensorflow_profiler_XSpace_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.profiler.XSpace.class, org.tensorflow.proto.profiler.XSpace.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.profiler.XSpace.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+        getPlanesFieldBuilder();
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      if (planesBuilder_ == null) {
+        planes_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000001);
+      } else {
+        planesBuilder_.clear();
+      }
+      errors_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+      bitField0_ = (bitField0_ & ~0x00000002);
+      warnings_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+      bitField0_ = (bitField0_ & ~0x00000004);
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.profiler.XPlaneProtos.internal_static_tensorflow_profiler_XSpace_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.profiler.XSpace getDefaultInstanceForType() {
+      return org.tensorflow.proto.profiler.XSpace.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.profiler.XSpace build() {
+      org.tensorflow.proto.profiler.XSpace result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.profiler.XSpace buildPartial() {
+      org.tensorflow.proto.profiler.XSpace result = new org.tensorflow.proto.profiler.XSpace(this);
+      int from_bitField0_ = bitField0_;
+      if (planesBuilder_ == null) {
+        if (((bitField0_ & 0x00000001) != 0)) {
+          planes_ = java.util.Collections.unmodifiableList(planes_);
+          bitField0_ = (bitField0_ & ~0x00000001);
+        }
+        result.planes_ = planes_;
+      } else {
+        result.planes_ = planesBuilder_.build();
+      }
+      if (((bitField0_ & 0x00000002) != 0)) {
+        errors_ = errors_.getUnmodifiableView();
+        bitField0_ = (bitField0_ & ~0x00000002);
+      }
+      result.errors_ = errors_;
+      if (((bitField0_ & 0x00000004) != 0)) {
+        warnings_ = warnings_.getUnmodifiableView();
+        bitField0_ = (bitField0_ & ~0x00000004);
+      }
+      result.warnings_ = warnings_;
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.profiler.XSpace) {
+        return mergeFrom((org.tensorflow.proto.profiler.XSpace)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.profiler.XSpace other) {
+      if (other == org.tensorflow.proto.profiler.XSpace.getDefaultInstance()) return this;
+      if (planesBuilder_ == null) {
+        if (!other.planes_.isEmpty()) {
+          if (planes_.isEmpty()) {
+            planes_ = other.planes_;
+            bitField0_ = (bitField0_ & ~0x00000001);
+          } else {
+            ensurePlanesIsMutable();
+            planes_.addAll(other.planes_);
+          }
+          onChanged();
+        }
+      } else {
+        if (!other.planes_.isEmpty()) {
+          if (planesBuilder_.isEmpty()) {
+            planesBuilder_.dispose();
+            planesBuilder_ = null;
+            planes_ = other.planes_;
+            bitField0_ = (bitField0_ & ~0x00000001);
+            planesBuilder_ = 
+              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
+                 getPlanesFieldBuilder() : null;
+          } else {
+            planesBuilder_.addAllMessages(other.planes_);
+          }
+        }
+      }
+      if (!other.errors_.isEmpty()) {
+        if (errors_.isEmpty()) {
+          errors_ = other.errors_;
+          bitField0_ = (bitField0_ & ~0x00000002);
+        } else {
+          ensureErrorsIsMutable();
+          errors_.addAll(other.errors_);
+        }
+        onChanged();
+      }
+      if (!other.warnings_.isEmpty()) {
+        if (warnings_.isEmpty()) {
+          warnings_ = other.warnings_;
+          bitField0_ = (bitField0_ & ~0x00000004);
+        } else {
+          ensureWarningsIsMutable();
+          warnings_.addAll(other.warnings_);
+        }
+        onChanged();
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.profiler.XSpace parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.profiler.XSpace) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+    private int bitField0_;
+
+    private java.util.List<org.tensorflow.proto.profiler.XPlane> planes_ =
+      java.util.Collections.emptyList();
+    private void ensurePlanesIsMutable() {
+      if (!((bitField0_ & 0x00000001) != 0)) {
+        planes_ = new java.util.ArrayList<org.tensorflow.proto.profiler.XPlane>(planes_);
+        bitField0_ |= 0x00000001;
+       }
+    }
+
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.profiler.XPlane, org.tensorflow.proto.profiler.XPlane.Builder, org.tensorflow.proto.profiler.XPlaneOrBuilder> planesBuilder_;
+
+    /**
+     * <code>repeated .tensorflow.profiler.XPlane planes = 1;</code>
+     */
+    public java.util.List<org.tensorflow.proto.profiler.XPlane> getPlanesList() {
+      if (planesBuilder_ == null) {
+        return java.util.Collections.unmodifiableList(planes_);
+      } else {
+        return planesBuilder_.getMessageList();
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.profiler.XPlane planes = 1;</code>
+     */
+    public int getPlanesCount() {
+      if (planesBuilder_ == null) {
+        return planes_.size();
+      } else {
+        return planesBuilder_.getCount();
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.profiler.XPlane planes = 1;</code>
+     */
+    public org.tensorflow.proto.profiler.XPlane getPlanes(int index) {
+      if (planesBuilder_ == null) {
+        return planes_.get(index);
+      } else {
+        return planesBuilder_.getMessage(index);
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.profiler.XPlane planes = 1;</code>
+     */
+    public Builder setPlanes(
+        int index, org.tensorflow.proto.profiler.XPlane value) {
+      if (planesBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensurePlanesIsMutable();
+        planes_.set(index, value);
+        onChanged();
+      } else {
+        planesBuilder_.setMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.profiler.XPlane planes = 1;</code>
+     */
+    public Builder setPlanes(
+        int index, org.tensorflow.proto.profiler.XPlane.Builder builderForValue) {
+      if (planesBuilder_ == null) {
+        ensurePlanesIsMutable();
+        planes_.set(index, builderForValue.build());
+        onChanged();
+      } else {
+        planesBuilder_.setMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.profiler.XPlane planes = 1;</code>
+     */
+    public Builder addPlanes(org.tensorflow.proto.profiler.XPlane value) {
+      if (planesBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensurePlanesIsMutable();
+        planes_.add(value);
+        onChanged();
+      } else {
+        planesBuilder_.addMessage(value);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.profiler.XPlane planes = 1;</code>
+     */
+    public Builder addPlanes(
+        int index, org.tensorflow.proto.profiler.XPlane value) {
+      if (planesBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensurePlanesIsMutable();
+        planes_.add(index, value);
+        onChanged();
+      } else {
+        planesBuilder_.addMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.profiler.XPlane planes = 1;</code>
+     */
+    public Builder addPlanes(
+        org.tensorflow.proto.profiler.XPlane.Builder builderForValue) {
+      if (planesBuilder_ == null) {
+        ensurePlanesIsMutable();
+        planes_.add(builderForValue.build());
+        onChanged();
+      } else {
+        planesBuilder_.addMessage(builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.profiler.XPlane planes = 1;</code>
+     */
+    public Builder addPlanes(
+        int index, org.tensorflow.proto.profiler.XPlane.Builder builderForValue) {
+      if (planesBuilder_ == null) {
+        ensurePlanesIsMutable();
+        planes_.add(index, builderForValue.build());
+        onChanged();
+      } else {
+        planesBuilder_.addMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.profiler.XPlane planes = 1;</code>
+     */
+    public Builder addAllPlanes(
+        java.lang.Iterable<? extends org.tensorflow.proto.profiler.XPlane> values) {
+      if (planesBuilder_ == null) {
+        ensurePlanesIsMutable();
+        com.google.protobuf.AbstractMessageLite.Builder.addAll(
+            values, planes_);
+        onChanged();
+      } else {
+        planesBuilder_.addAllMessages(values);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.profiler.XPlane planes = 1;</code>
+     */
+    public Builder clearPlanes() {
+      if (planesBuilder_ == null) {
+        planes_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000001);
+        onChanged();
+      } else {
+        planesBuilder_.clear();
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.profiler.XPlane planes = 1;</code>
+     */
+    public Builder removePlanes(int index) {
+      if (planesBuilder_ == null) {
+        ensurePlanesIsMutable();
+        planes_.remove(index);
+        onChanged();
+      } else {
+        planesBuilder_.remove(index);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.profiler.XPlane planes = 1;</code>
+     */
+    public org.tensorflow.proto.profiler.XPlane.Builder getPlanesBuilder(
+        int index) {
+      return getPlanesFieldBuilder().getBuilder(index);
+    }
+    /**
+     * <code>repeated .tensorflow.profiler.XPlane planes = 1;</code>
+     */
+    public org.tensorflow.proto.profiler.XPlaneOrBuilder getPlanesOrBuilder(
+        int index) {
+      if (planesBuilder_ == null) {
+        return planes_.get(index);  } else {
+        return planesBuilder_.getMessageOrBuilder(index);
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.profiler.XPlane planes = 1;</code>
+     */
+    public java.util.List<? extends org.tensorflow.proto.profiler.XPlaneOrBuilder> 
+         getPlanesOrBuilderList() {
+      if (planesBuilder_ != null) {
+        return planesBuilder_.getMessageOrBuilderList();
+      } else {
+        return java.util.Collections.unmodifiableList(planes_);
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.profiler.XPlane planes = 1;</code>
+     */
+    public org.tensorflow.proto.profiler.XPlane.Builder addPlanesBuilder() {
+      return getPlanesFieldBuilder().addBuilder(
+          org.tensorflow.proto.profiler.XPlane.getDefaultInstance());
+    }
+    /**
+     * <code>repeated .tensorflow.profiler.XPlane planes = 1;</code>
+     */
+    public org.tensorflow.proto.profiler.XPlane.Builder addPlanesBuilder(
+        int index) {
+      return getPlanesFieldBuilder().addBuilder(
+          index, org.tensorflow.proto.profiler.XPlane.getDefaultInstance());
+    }
+    /**
+     * <code>repeated .tensorflow.profiler.XPlane planes = 1;</code>
+     */
+    public java.util.List<org.tensorflow.proto.profiler.XPlane.Builder> 
+         getPlanesBuilderList() {
+      return getPlanesFieldBuilder().getBuilderList();
+    }
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.profiler.XPlane, org.tensorflow.proto.profiler.XPlane.Builder, org.tensorflow.proto.profiler.XPlaneOrBuilder> 
+        getPlanesFieldBuilder() {
+      if (planesBuilder_ == null) {
+        planesBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
+            org.tensorflow.proto.profiler.XPlane, org.tensorflow.proto.profiler.XPlane.Builder, org.tensorflow.proto.profiler.XPlaneOrBuilder>(
+                planes_,
+                ((bitField0_ & 0x00000001) != 0),
+                getParentForChildren(),
+                isClean());
+        planes_ = null;
+      }
+      return planesBuilder_;
+    }
+
+    private com.google.protobuf.LazyStringList errors_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+    private void ensureErrorsIsMutable() {
+      if (!((bitField0_ & 0x00000002) != 0)) {
+        errors_ = new com.google.protobuf.LazyStringArrayList(errors_);
+        bitField0_ |= 0x00000002;
+       }
+    }
+    /**
+     * <pre>
+     * Errors (if any) in the generation of planes.
+     * </pre>
+     *
+     * <code>repeated string errors = 2;</code>
+     */
+    public com.google.protobuf.ProtocolStringList
+        getErrorsList() {
+      return errors_.getUnmodifiableView();
+    }
+    /**
+     * <pre>
+     * Errors (if any) in the generation of planes.
+     * </pre>
+     *
+     * <code>repeated string errors = 2;</code>
+     */
+    public int getErrorsCount() {
+      return errors_.size();
+    }
+    /**
+     * <pre>
+     * Errors (if any) in the generation of planes.
+     * </pre>
+     *
+     * <code>repeated string errors = 2;</code>
+     */
+    public java.lang.String getErrors(int index) {
+      return errors_.get(index);
+    }
+    /**
+     * <pre>
+     * Errors (if any) in the generation of planes.
+     * </pre>
+     *
+     * <code>repeated string errors = 2;</code>
+     */
+    public com.google.protobuf.ByteString
+        getErrorsBytes(int index) {
+      return errors_.getByteString(index);
+    }
+    /**
+     * <pre>
+     * Errors (if any) in the generation of planes.
+     * </pre>
+     *
+     * <code>repeated string errors = 2;</code>
+     */
+    public Builder setErrors(
+        int index, java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  ensureErrorsIsMutable();
+      errors_.set(index, value);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Errors (if any) in the generation of planes.
+     * </pre>
+     *
+     * <code>repeated string errors = 2;</code>
+     */
+    public Builder addErrors(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  ensureErrorsIsMutable();
+      errors_.add(value);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Errors (if any) in the generation of planes.
+     * </pre>
+     *
+     * <code>repeated string errors = 2;</code>
+     */
+    public Builder addAllErrors(
+        java.lang.Iterable<java.lang.String> values) {
+      ensureErrorsIsMutable();
+      com.google.protobuf.AbstractMessageLite.Builder.addAll(
+          values, errors_);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Errors (if any) in the generation of planes.
+     * </pre>
+     *
+     * <code>repeated string errors = 2;</code>
+     */
+    public Builder clearErrors() {
+      errors_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+      bitField0_ = (bitField0_ & ~0x00000002);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Errors (if any) in the generation of planes.
+     * </pre>
+     *
+     * <code>repeated string errors = 2;</code>
+     */
+    public Builder addErrorsBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      ensureErrorsIsMutable();
+      errors_.add(value);
+      onChanged();
+      return this;
+    }
+
+    private com.google.protobuf.LazyStringList warnings_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+    private void ensureWarningsIsMutable() {
+      if (!((bitField0_ & 0x00000004) != 0)) {
+        warnings_ = new com.google.protobuf.LazyStringArrayList(warnings_);
+        bitField0_ |= 0x00000004;
+       }
+    }
+    /**
+     * <pre>
+     * Warnings (if any) in the generation of planes;
+     * </pre>
+     *
+     * <code>repeated string warnings = 3;</code>
+     */
+    public com.google.protobuf.ProtocolStringList
+        getWarningsList() {
+      return warnings_.getUnmodifiableView();
+    }
+    /**
+     * <pre>
+     * Warnings (if any) in the generation of planes;
+     * </pre>
+     *
+     * <code>repeated string warnings = 3;</code>
+     */
+    public int getWarningsCount() {
+      return warnings_.size();
+    }
+    /**
+     * <pre>
+     * Warnings (if any) in the generation of planes;
+     * </pre>
+     *
+     * <code>repeated string warnings = 3;</code>
+     */
+    public java.lang.String getWarnings(int index) {
+      return warnings_.get(index);
+    }
+    /**
+     * <pre>
+     * Warnings (if any) in the generation of planes;
+     * </pre>
+     *
+     * <code>repeated string warnings = 3;</code>
+     */
+    public com.google.protobuf.ByteString
+        getWarningsBytes(int index) {
+      return warnings_.getByteString(index);
+    }
+    /**
+     * <pre>
+     * Warnings (if any) in the generation of planes;
+     * </pre>
+     *
+     * <code>repeated string warnings = 3;</code>
+     */
+    public Builder setWarnings(
+        int index, java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  ensureWarningsIsMutable();
+      warnings_.set(index, value);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Warnings (if any) in the generation of planes;
+     * </pre>
+     *
+     * <code>repeated string warnings = 3;</code>
+     */
+    public Builder addWarnings(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  ensureWarningsIsMutable();
+      warnings_.add(value);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Warnings (if any) in the generation of planes;
+     * </pre>
+     *
+     * <code>repeated string warnings = 3;</code>
+     */
+    public Builder addAllWarnings(
+        java.lang.Iterable<java.lang.String> values) {
+      ensureWarningsIsMutable();
+      com.google.protobuf.AbstractMessageLite.Builder.addAll(
+          values, warnings_);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Warnings (if any) in the generation of planes;
+     * </pre>
+     *
+     * <code>repeated string warnings = 3;</code>
+     */
+    public Builder clearWarnings() {
+      warnings_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+      bitField0_ = (bitField0_ & ~0x00000004);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Warnings (if any) in the generation of planes;
+     * </pre>
+     *
+     * <code>repeated string warnings = 3;</code>
+     */
+    public Builder addWarningsBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      ensureWarningsIsMutable();
+      warnings_.add(value);
+      onChanged();
+      return this;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.profiler.XSpace)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.profiler.XSpace)
+  private static final org.tensorflow.proto.profiler.XSpace DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.profiler.XSpace();
+  }
+
+  public static org.tensorflow.proto.profiler.XSpace getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<XSpace>
+      PARSER = new com.google.protobuf.AbstractParser<XSpace>() {
+    @java.lang.Override
+    public XSpace parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new XSpace(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<XSpace> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<XSpace> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.profiler.XSpace getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/XSpaceOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/XSpaceOrBuilder.java
new file mode 100644
index 00000000000..dd081ad954f
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/XSpaceOrBuilder.java
@@ -0,0 +1,103 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/profiler/protobuf/xplane.proto
+
+package org.tensorflow.proto.profiler;
+
+public interface XSpaceOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.profiler.XSpace)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <code>repeated .tensorflow.profiler.XPlane planes = 1;</code>
+   */
+  java.util.List<org.tensorflow.proto.profiler.XPlane> 
+      getPlanesList();
+  /**
+   * <code>repeated .tensorflow.profiler.XPlane planes = 1;</code>
+   */
+  org.tensorflow.proto.profiler.XPlane getPlanes(int index);
+  /**
+   * <code>repeated .tensorflow.profiler.XPlane planes = 1;</code>
+   */
+  int getPlanesCount();
+  /**
+   * <code>repeated .tensorflow.profiler.XPlane planes = 1;</code>
+   */
+  java.util.List<? extends org.tensorflow.proto.profiler.XPlaneOrBuilder> 
+      getPlanesOrBuilderList();
+  /**
+   * <code>repeated .tensorflow.profiler.XPlane planes = 1;</code>
+   */
+  org.tensorflow.proto.profiler.XPlaneOrBuilder getPlanesOrBuilder(
+      int index);
+
+  /**
+   * <pre>
+   * Errors (if any) in the generation of planes.
+   * </pre>
+   *
+   * <code>repeated string errors = 2;</code>
+   */
+  java.util.List<java.lang.String>
+      getErrorsList();
+  /**
+   * <pre>
+   * Errors (if any) in the generation of planes.
+   * </pre>
+   *
+   * <code>repeated string errors = 2;</code>
+   */
+  int getErrorsCount();
+  /**
+   * <pre>
+   * Errors (if any) in the generation of planes.
+   * </pre>
+   *
+   * <code>repeated string errors = 2;</code>
+   */
+  java.lang.String getErrors(int index);
+  /**
+   * <pre>
+   * Errors (if any) in the generation of planes.
+   * </pre>
+   *
+   * <code>repeated string errors = 2;</code>
+   */
+  com.google.protobuf.ByteString
+      getErrorsBytes(int index);
+
+  /**
+   * <pre>
+   * Warnings (if any) in the generation of planes;
+   * </pre>
+   *
+   * <code>repeated string warnings = 3;</code>
+   */
+  java.util.List<java.lang.String>
+      getWarningsList();
+  /**
+   * <pre>
+   * Warnings (if any) in the generation of planes;
+   * </pre>
+   *
+   * <code>repeated string warnings = 3;</code>
+   */
+  int getWarningsCount();
+  /**
+   * <pre>
+   * Warnings (if any) in the generation of planes;
+   * </pre>
+   *
+   * <code>repeated string warnings = 3;</code>
+   */
+  java.lang.String getWarnings(int index);
+  /**
+   * <pre>
+   * Warnings (if any) in the generation of planes;
+   * </pre>
+   *
+   * <code>repeated string warnings = 3;</code>
+   */
+  com.google.protobuf.ByteString
+      getWarningsBytes(int index);
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/XStat.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/XStat.java
new file mode 100644
index 00000000000..3ab8c18a43a
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/XStat.java
@@ -0,0 +1,1111 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/profiler/protobuf/xplane.proto
+
+package org.tensorflow.proto.profiler;
+
+/**
+ * <pre>
+ * An XStat is a named value associated with an XEvent, e.g., a performance
+ * counter value, a metric computed by a formula applied over nested XEvents
+ * and XStats.
+ * Next ID: 8
+ * </pre>
+ *
+ * Protobuf type {@code tensorflow.profiler.XStat}
+ */
+public  final class XStat extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.profiler.XStat)
+    XStatOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use XStat.newBuilder() to construct.
+  private XStat(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private XStat() {
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new XStat();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private XStat(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 8: {
+
+            metadataId_ = input.readInt64();
+            break;
+          }
+          case 17: {
+            valueCase_ = 2;
+            value_ = input.readDouble();
+            break;
+          }
+          case 24: {
+            valueCase_ = 3;
+            value_ = input.readUInt64();
+            break;
+          }
+          case 32: {
+            valueCase_ = 4;
+            value_ = input.readInt64();
+            break;
+          }
+          case 42: {
+            java.lang.String s = input.readStringRequireUtf8();
+            valueCase_ = 5;
+            value_ = s;
+            break;
+          }
+          case 50: {
+            valueCase_ = 6;
+            value_ = input.readBytes();
+            break;
+          }
+          case 56: {
+            valueCase_ = 7;
+            value_ = input.readUInt64();
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.profiler.XPlaneProtos.internal_static_tensorflow_profiler_XStat_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.profiler.XPlaneProtos.internal_static_tensorflow_profiler_XStat_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.profiler.XStat.class, org.tensorflow.proto.profiler.XStat.Builder.class);
+  }
+
+  private int valueCase_ = 0;
+  private java.lang.Object value_;
+  public enum ValueCase
+      implements com.google.protobuf.Internal.EnumLite {
+    DOUBLE_VALUE(2),
+    UINT64_VALUE(3),
+    INT64_VALUE(4),
+    STR_VALUE(5),
+    BYTES_VALUE(6),
+    REF_VALUE(7),
+    VALUE_NOT_SET(0);
+    private final int value;
+    private ValueCase(int value) {
+      this.value = value;
+    }
+    /**
+     * @deprecated Use {@link #forNumber(int)} instead.
+     */
+    @java.lang.Deprecated
+    public static ValueCase valueOf(int value) {
+      return forNumber(value);
+    }
+
+    public static ValueCase forNumber(int value) {
+      switch (value) {
+        case 2: return DOUBLE_VALUE;
+        case 3: return UINT64_VALUE;
+        case 4: return INT64_VALUE;
+        case 5: return STR_VALUE;
+        case 6: return BYTES_VALUE;
+        case 7: return REF_VALUE;
+        case 0: return VALUE_NOT_SET;
+        default: return null;
+      }
+    }
+    public int getNumber() {
+      return this.value;
+    }
+  };
+
+  public ValueCase
+  getValueCase() {
+    return ValueCase.forNumber(
+        valueCase_);
+  }
+
+  public static final int METADATA_ID_FIELD_NUMBER = 1;
+  private long metadataId_;
+  /**
+   * <pre>
+   * XStatMetadata.id of corresponding metadata.
+   * </pre>
+   *
+   * <code>int64 metadata_id = 1;</code>
+   */
+  public long getMetadataId() {
+    return metadataId_;
+  }
+
+  public static final int DOUBLE_VALUE_FIELD_NUMBER = 2;
+  /**
+   * <code>double double_value = 2;</code>
+   */
+  public double getDoubleValue() {
+    if (valueCase_ == 2) {
+      return (java.lang.Double) value_;
+    }
+    return 0D;
+  }
+
+  public static final int UINT64_VALUE_FIELD_NUMBER = 3;
+  /**
+   * <code>uint64 uint64_value = 3;</code>
+   */
+  public long getUint64Value() {
+    if (valueCase_ == 3) {
+      return (java.lang.Long) value_;
+    }
+    return 0L;
+  }
+
+  public static final int INT64_VALUE_FIELD_NUMBER = 4;
+  /**
+   * <code>int64 int64_value = 4;</code>
+   */
+  public long getInt64Value() {
+    if (valueCase_ == 4) {
+      return (java.lang.Long) value_;
+    }
+    return 0L;
+  }
+
+  public static final int STR_VALUE_FIELD_NUMBER = 5;
+  /**
+   * <code>string str_value = 5;</code>
+   */
+  public java.lang.String getStrValue() {
+    java.lang.Object ref = "";
+    if (valueCase_ == 5) {
+      ref = value_;
+    }
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      if (valueCase_ == 5) {
+        value_ = s;
+      }
+      return s;
+    }
+  }
+  /**
+   * <code>string str_value = 5;</code>
+   */
+  public com.google.protobuf.ByteString
+      getStrValueBytes() {
+    java.lang.Object ref = "";
+    if (valueCase_ == 5) {
+      ref = value_;
+    }
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      if (valueCase_ == 5) {
+        value_ = b;
+      }
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int BYTES_VALUE_FIELD_NUMBER = 6;
+  /**
+   * <code>bytes bytes_value = 6;</code>
+   */
+  public com.google.protobuf.ByteString getBytesValue() {
+    if (valueCase_ == 6) {
+      return (com.google.protobuf.ByteString) value_;
+    }
+    return com.google.protobuf.ByteString.EMPTY;
+  }
+
+  public static final int REF_VALUE_FIELD_NUMBER = 7;
+  /**
+   * <pre>
+   * A string value that stored in XStatMetadata::name.
+   * </pre>
+   *
+   * <code>uint64 ref_value = 7;</code>
+   */
+  public long getRefValue() {
+    if (valueCase_ == 7) {
+      return (java.lang.Long) value_;
+    }
+    return 0L;
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    if (metadataId_ != 0L) {
+      output.writeInt64(1, metadataId_);
+    }
+    if (valueCase_ == 2) {
+      output.writeDouble(
+          2, (double)((java.lang.Double) value_));
+    }
+    if (valueCase_ == 3) {
+      output.writeUInt64(
+          3, (long)((java.lang.Long) value_));
+    }
+    if (valueCase_ == 4) {
+      output.writeInt64(
+          4, (long)((java.lang.Long) value_));
+    }
+    if (valueCase_ == 5) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 5, value_);
+    }
+    if (valueCase_ == 6) {
+      output.writeBytes(
+          6, (com.google.protobuf.ByteString) value_);
+    }
+    if (valueCase_ == 7) {
+      output.writeUInt64(
+          7, (long)((java.lang.Long) value_));
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (metadataId_ != 0L) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt64Size(1, metadataId_);
+    }
+    if (valueCase_ == 2) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeDoubleSize(
+            2, (double)((java.lang.Double) value_));
+    }
+    if (valueCase_ == 3) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeUInt64Size(
+            3, (long)((java.lang.Long) value_));
+    }
+    if (valueCase_ == 4) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt64Size(
+            4, (long)((java.lang.Long) value_));
+    }
+    if (valueCase_ == 5) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(5, value_);
+    }
+    if (valueCase_ == 6) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeBytesSize(
+            6, (com.google.protobuf.ByteString) value_);
+    }
+    if (valueCase_ == 7) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeUInt64Size(
+            7, (long)((java.lang.Long) value_));
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.profiler.XStat)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.profiler.XStat other = (org.tensorflow.proto.profiler.XStat) obj;
+
+    if (getMetadataId()
+        != other.getMetadataId()) return false;
+    if (!getValueCase().equals(other.getValueCase())) return false;
+    switch (valueCase_) {
+      case 2:
+        if (java.lang.Double.doubleToLongBits(getDoubleValue())
+            != java.lang.Double.doubleToLongBits(
+                other.getDoubleValue())) return false;
+        break;
+      case 3:
+        if (getUint64Value()
+            != other.getUint64Value()) return false;
+        break;
+      case 4:
+        if (getInt64Value()
+            != other.getInt64Value()) return false;
+        break;
+      case 5:
+        if (!getStrValue()
+            .equals(other.getStrValue())) return false;
+        break;
+      case 6:
+        if (!getBytesValue()
+            .equals(other.getBytesValue())) return false;
+        break;
+      case 7:
+        if (getRefValue()
+            != other.getRefValue()) return false;
+        break;
+      case 0:
+      default:
+    }
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    hash = (37 * hash) + METADATA_ID_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+        getMetadataId());
+    switch (valueCase_) {
+      case 2:
+        hash = (37 * hash) + DOUBLE_VALUE_FIELD_NUMBER;
+        hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+            java.lang.Double.doubleToLongBits(getDoubleValue()));
+        break;
+      case 3:
+        hash = (37 * hash) + UINT64_VALUE_FIELD_NUMBER;
+        hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+            getUint64Value());
+        break;
+      case 4:
+        hash = (37 * hash) + INT64_VALUE_FIELD_NUMBER;
+        hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+            getInt64Value());
+        break;
+      case 5:
+        hash = (37 * hash) + STR_VALUE_FIELD_NUMBER;
+        hash = (53 * hash) + getStrValue().hashCode();
+        break;
+      case 6:
+        hash = (37 * hash) + BYTES_VALUE_FIELD_NUMBER;
+        hash = (53 * hash) + getBytesValue().hashCode();
+        break;
+      case 7:
+        hash = (37 * hash) + REF_VALUE_FIELD_NUMBER;
+        hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+            getRefValue());
+        break;
+      case 0:
+      default:
+    }
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.profiler.XStat parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.profiler.XStat parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.profiler.XStat parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.profiler.XStat parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.profiler.XStat parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.profiler.XStat parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.profiler.XStat parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.profiler.XStat parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.profiler.XStat parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.profiler.XStat parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.profiler.XStat parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.profiler.XStat parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.profiler.XStat prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * <pre>
+   * An XStat is a named value associated with an XEvent, e.g., a performance
+   * counter value, a metric computed by a formula applied over nested XEvents
+   * and XStats.
+   * Next ID: 8
+   * </pre>
+   *
+   * Protobuf type {@code tensorflow.profiler.XStat}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.profiler.XStat)
+      org.tensorflow.proto.profiler.XStatOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.profiler.XPlaneProtos.internal_static_tensorflow_profiler_XStat_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.profiler.XPlaneProtos.internal_static_tensorflow_profiler_XStat_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.profiler.XStat.class, org.tensorflow.proto.profiler.XStat.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.profiler.XStat.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      metadataId_ = 0L;
+
+      valueCase_ = 0;
+      value_ = null;
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.profiler.XPlaneProtos.internal_static_tensorflow_profiler_XStat_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.profiler.XStat getDefaultInstanceForType() {
+      return org.tensorflow.proto.profiler.XStat.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.profiler.XStat build() {
+      org.tensorflow.proto.profiler.XStat result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.profiler.XStat buildPartial() {
+      org.tensorflow.proto.profiler.XStat result = new org.tensorflow.proto.profiler.XStat(this);
+      result.metadataId_ = metadataId_;
+      if (valueCase_ == 2) {
+        result.value_ = value_;
+      }
+      if (valueCase_ == 3) {
+        result.value_ = value_;
+      }
+      if (valueCase_ == 4) {
+        result.value_ = value_;
+      }
+      if (valueCase_ == 5) {
+        result.value_ = value_;
+      }
+      if (valueCase_ == 6) {
+        result.value_ = value_;
+      }
+      if (valueCase_ == 7) {
+        result.value_ = value_;
+      }
+      result.valueCase_ = valueCase_;
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.profiler.XStat) {
+        return mergeFrom((org.tensorflow.proto.profiler.XStat)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.profiler.XStat other) {
+      if (other == org.tensorflow.proto.profiler.XStat.getDefaultInstance()) return this;
+      if (other.getMetadataId() != 0L) {
+        setMetadataId(other.getMetadataId());
+      }
+      switch (other.getValueCase()) {
+        case DOUBLE_VALUE: {
+          setDoubleValue(other.getDoubleValue());
+          break;
+        }
+        case UINT64_VALUE: {
+          setUint64Value(other.getUint64Value());
+          break;
+        }
+        case INT64_VALUE: {
+          setInt64Value(other.getInt64Value());
+          break;
+        }
+        case STR_VALUE: {
+          valueCase_ = 5;
+          value_ = other.value_;
+          onChanged();
+          break;
+        }
+        case BYTES_VALUE: {
+          setBytesValue(other.getBytesValue());
+          break;
+        }
+        case REF_VALUE: {
+          setRefValue(other.getRefValue());
+          break;
+        }
+        case VALUE_NOT_SET: {
+          break;
+        }
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.profiler.XStat parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.profiler.XStat) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+    private int valueCase_ = 0;
+    private java.lang.Object value_;
+    public ValueCase
+        getValueCase() {
+      return ValueCase.forNumber(
+          valueCase_);
+    }
+
+    public Builder clearValue() {
+      valueCase_ = 0;
+      value_ = null;
+      onChanged();
+      return this;
+    }
+
+
+    private long metadataId_ ;
+    /**
+     * <pre>
+     * XStatMetadata.id of corresponding metadata.
+     * </pre>
+     *
+     * <code>int64 metadata_id = 1;</code>
+     */
+    public long getMetadataId() {
+      return metadataId_;
+    }
+    /**
+     * <pre>
+     * XStatMetadata.id of corresponding metadata.
+     * </pre>
+     *
+     * <code>int64 metadata_id = 1;</code>
+     */
+    public Builder setMetadataId(long value) {
+      
+      metadataId_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * XStatMetadata.id of corresponding metadata.
+     * </pre>
+     *
+     * <code>int64 metadata_id = 1;</code>
+     */
+    public Builder clearMetadataId() {
+      
+      metadataId_ = 0L;
+      onChanged();
+      return this;
+    }
+
+    /**
+     * <code>double double_value = 2;</code>
+     */
+    public double getDoubleValue() {
+      if (valueCase_ == 2) {
+        return (java.lang.Double) value_;
+      }
+      return 0D;
+    }
+    /**
+     * <code>double double_value = 2;</code>
+     */
+    public Builder setDoubleValue(double value) {
+      valueCase_ = 2;
+      value_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>double double_value = 2;</code>
+     */
+    public Builder clearDoubleValue() {
+      if (valueCase_ == 2) {
+        valueCase_ = 0;
+        value_ = null;
+        onChanged();
+      }
+      return this;
+    }
+
+    /**
+     * <code>uint64 uint64_value = 3;</code>
+     */
+    public long getUint64Value() {
+      if (valueCase_ == 3) {
+        return (java.lang.Long) value_;
+      }
+      return 0L;
+    }
+    /**
+     * <code>uint64 uint64_value = 3;</code>
+     */
+    public Builder setUint64Value(long value) {
+      valueCase_ = 3;
+      value_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>uint64 uint64_value = 3;</code>
+     */
+    public Builder clearUint64Value() {
+      if (valueCase_ == 3) {
+        valueCase_ = 0;
+        value_ = null;
+        onChanged();
+      }
+      return this;
+    }
+
+    /**
+     * <code>int64 int64_value = 4;</code>
+     */
+    public long getInt64Value() {
+      if (valueCase_ == 4) {
+        return (java.lang.Long) value_;
+      }
+      return 0L;
+    }
+    /**
+     * <code>int64 int64_value = 4;</code>
+     */
+    public Builder setInt64Value(long value) {
+      valueCase_ = 4;
+      value_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>int64 int64_value = 4;</code>
+     */
+    public Builder clearInt64Value() {
+      if (valueCase_ == 4) {
+        valueCase_ = 0;
+        value_ = null;
+        onChanged();
+      }
+      return this;
+    }
+
+    /**
+     * <code>string str_value = 5;</code>
+     */
+    public java.lang.String getStrValue() {
+      java.lang.Object ref = "";
+      if (valueCase_ == 5) {
+        ref = value_;
+      }
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        if (valueCase_ == 5) {
+          value_ = s;
+        }
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <code>string str_value = 5;</code>
+     */
+    public com.google.protobuf.ByteString
+        getStrValueBytes() {
+      java.lang.Object ref = "";
+      if (valueCase_ == 5) {
+        ref = value_;
+      }
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        if (valueCase_ == 5) {
+          value_ = b;
+        }
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <code>string str_value = 5;</code>
+     */
+    public Builder setStrValue(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  valueCase_ = 5;
+      value_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>string str_value = 5;</code>
+     */
+    public Builder clearStrValue() {
+      if (valueCase_ == 5) {
+        valueCase_ = 0;
+        value_ = null;
+        onChanged();
+      }
+      return this;
+    }
+    /**
+     * <code>string str_value = 5;</code>
+     */
+    public Builder setStrValueBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      valueCase_ = 5;
+      value_ = value;
+      onChanged();
+      return this;
+    }
+
+    /**
+     * <code>bytes bytes_value = 6;</code>
+     */
+    public com.google.protobuf.ByteString getBytesValue() {
+      if (valueCase_ == 6) {
+        return (com.google.protobuf.ByteString) value_;
+      }
+      return com.google.protobuf.ByteString.EMPTY;
+    }
+    /**
+     * <code>bytes bytes_value = 6;</code>
+     */
+    public Builder setBytesValue(com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  valueCase_ = 6;
+      value_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>bytes bytes_value = 6;</code>
+     */
+    public Builder clearBytesValue() {
+      if (valueCase_ == 6) {
+        valueCase_ = 0;
+        value_ = null;
+        onChanged();
+      }
+      return this;
+    }
+
+    /**
+     * <pre>
+     * A string value that stored in XStatMetadata::name.
+     * </pre>
+     *
+     * <code>uint64 ref_value = 7;</code>
+     */
+    public long getRefValue() {
+      if (valueCase_ == 7) {
+        return (java.lang.Long) value_;
+      }
+      return 0L;
+    }
+    /**
+     * <pre>
+     * A string value that stored in XStatMetadata::name.
+     * </pre>
+     *
+     * <code>uint64 ref_value = 7;</code>
+     */
+    public Builder setRefValue(long value) {
+      valueCase_ = 7;
+      value_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * A string value that stored in XStatMetadata::name.
+     * </pre>
+     *
+     * <code>uint64 ref_value = 7;</code>
+     */
+    public Builder clearRefValue() {
+      if (valueCase_ == 7) {
+        valueCase_ = 0;
+        value_ = null;
+        onChanged();
+      }
+      return this;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.profiler.XStat)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.profiler.XStat)
+  private static final org.tensorflow.proto.profiler.XStat DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.profiler.XStat();
+  }
+
+  public static org.tensorflow.proto.profiler.XStat getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<XStat>
+      PARSER = new com.google.protobuf.AbstractParser<XStat>() {
+    @java.lang.Override
+    public XStat parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new XStat(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<XStat> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<XStat> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.profiler.XStat getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/XStatMetadata.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/XStatMetadata.java
new file mode 100644
index 00000000000..8967f6e6c1e
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/XStatMetadata.java
@@ -0,0 +1,813 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/profiler/protobuf/xplane.proto
+
+package org.tensorflow.proto.profiler;
+
+/**
+ * <pre>
+ * Metadata for an XStat, shared by all instances of the same stat.
+ * Next ID: 4
+ * </pre>
+ *
+ * Protobuf type {@code tensorflow.profiler.XStatMetadata}
+ */
+public  final class XStatMetadata extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.profiler.XStatMetadata)
+    XStatMetadataOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use XStatMetadata.newBuilder() to construct.
+  private XStatMetadata(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private XStatMetadata() {
+    name_ = "";
+    description_ = "";
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new XStatMetadata();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private XStatMetadata(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 8: {
+
+            id_ = input.readInt64();
+            break;
+          }
+          case 18: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            name_ = s;
+            break;
+          }
+          case 26: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            description_ = s;
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.profiler.XPlaneProtos.internal_static_tensorflow_profiler_XStatMetadata_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.profiler.XPlaneProtos.internal_static_tensorflow_profiler_XStatMetadata_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.profiler.XStatMetadata.class, org.tensorflow.proto.profiler.XStatMetadata.Builder.class);
+  }
+
+  public static final int ID_FIELD_NUMBER = 1;
+  private long id_;
+  /**
+   * <pre>
+   * XPlane.stat_metadata map key.
+   * </pre>
+   *
+   * <code>int64 id = 1;</code>
+   */
+  public long getId() {
+    return id_;
+  }
+
+  public static final int NAME_FIELD_NUMBER = 2;
+  private volatile java.lang.Object name_;
+  /**
+   * <pre>
+   * Name of the stat (should be short).
+   * </pre>
+   *
+   * <code>string name = 2;</code>
+   */
+  public java.lang.String getName() {
+    java.lang.Object ref = name_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      name_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * Name of the stat (should be short).
+   * </pre>
+   *
+   * <code>string name = 2;</code>
+   */
+  public com.google.protobuf.ByteString
+      getNameBytes() {
+    java.lang.Object ref = name_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      name_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int DESCRIPTION_FIELD_NUMBER = 3;
+  private volatile java.lang.Object description_;
+  /**
+   * <pre>
+   * Description of the stat (might be long).
+   * </pre>
+   *
+   * <code>string description = 3;</code>
+   */
+  public java.lang.String getDescription() {
+    java.lang.Object ref = description_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      description_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * Description of the stat (might be long).
+   * </pre>
+   *
+   * <code>string description = 3;</code>
+   */
+  public com.google.protobuf.ByteString
+      getDescriptionBytes() {
+    java.lang.Object ref = description_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      description_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    if (id_ != 0L) {
+      output.writeInt64(1, id_);
+    }
+    if (!getNameBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 2, name_);
+    }
+    if (!getDescriptionBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 3, description_);
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (id_ != 0L) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt64Size(1, id_);
+    }
+    if (!getNameBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(2, name_);
+    }
+    if (!getDescriptionBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(3, description_);
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.profiler.XStatMetadata)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.profiler.XStatMetadata other = (org.tensorflow.proto.profiler.XStatMetadata) obj;
+
+    if (getId()
+        != other.getId()) return false;
+    if (!getName()
+        .equals(other.getName())) return false;
+    if (!getDescription()
+        .equals(other.getDescription())) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    hash = (37 * hash) + ID_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+        getId());
+    hash = (37 * hash) + NAME_FIELD_NUMBER;
+    hash = (53 * hash) + getName().hashCode();
+    hash = (37 * hash) + DESCRIPTION_FIELD_NUMBER;
+    hash = (53 * hash) + getDescription().hashCode();
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.profiler.XStatMetadata parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.profiler.XStatMetadata parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.profiler.XStatMetadata parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.profiler.XStatMetadata parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.profiler.XStatMetadata parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.profiler.XStatMetadata parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.profiler.XStatMetadata parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.profiler.XStatMetadata parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.profiler.XStatMetadata parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.profiler.XStatMetadata parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.profiler.XStatMetadata parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.profiler.XStatMetadata parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.profiler.XStatMetadata prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * <pre>
+   * Metadata for an XStat, shared by all instances of the same stat.
+   * Next ID: 4
+   * </pre>
+   *
+   * Protobuf type {@code tensorflow.profiler.XStatMetadata}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.profiler.XStatMetadata)
+      org.tensorflow.proto.profiler.XStatMetadataOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.profiler.XPlaneProtos.internal_static_tensorflow_profiler_XStatMetadata_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.profiler.XPlaneProtos.internal_static_tensorflow_profiler_XStatMetadata_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.profiler.XStatMetadata.class, org.tensorflow.proto.profiler.XStatMetadata.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.profiler.XStatMetadata.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      id_ = 0L;
+
+      name_ = "";
+
+      description_ = "";
+
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.profiler.XPlaneProtos.internal_static_tensorflow_profiler_XStatMetadata_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.profiler.XStatMetadata getDefaultInstanceForType() {
+      return org.tensorflow.proto.profiler.XStatMetadata.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.profiler.XStatMetadata build() {
+      org.tensorflow.proto.profiler.XStatMetadata result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.profiler.XStatMetadata buildPartial() {
+      org.tensorflow.proto.profiler.XStatMetadata result = new org.tensorflow.proto.profiler.XStatMetadata(this);
+      result.id_ = id_;
+      result.name_ = name_;
+      result.description_ = description_;
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.profiler.XStatMetadata) {
+        return mergeFrom((org.tensorflow.proto.profiler.XStatMetadata)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.profiler.XStatMetadata other) {
+      if (other == org.tensorflow.proto.profiler.XStatMetadata.getDefaultInstance()) return this;
+      if (other.getId() != 0L) {
+        setId(other.getId());
+      }
+      if (!other.getName().isEmpty()) {
+        name_ = other.name_;
+        onChanged();
+      }
+      if (!other.getDescription().isEmpty()) {
+        description_ = other.description_;
+        onChanged();
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.profiler.XStatMetadata parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.profiler.XStatMetadata) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+
+    private long id_ ;
+    /**
+     * <pre>
+     * XPlane.stat_metadata map key.
+     * </pre>
+     *
+     * <code>int64 id = 1;</code>
+     */
+    public long getId() {
+      return id_;
+    }
+    /**
+     * <pre>
+     * XPlane.stat_metadata map key.
+     * </pre>
+     *
+     * <code>int64 id = 1;</code>
+     */
+    public Builder setId(long value) {
+      
+      id_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * XPlane.stat_metadata map key.
+     * </pre>
+     *
+     * <code>int64 id = 1;</code>
+     */
+    public Builder clearId() {
+      
+      id_ = 0L;
+      onChanged();
+      return this;
+    }
+
+    private java.lang.Object name_ = "";
+    /**
+     * <pre>
+     * Name of the stat (should be short).
+     * </pre>
+     *
+     * <code>string name = 2;</code>
+     */
+    public java.lang.String getName() {
+      java.lang.Object ref = name_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        name_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Name of the stat (should be short).
+     * </pre>
+     *
+     * <code>string name = 2;</code>
+     */
+    public com.google.protobuf.ByteString
+        getNameBytes() {
+      java.lang.Object ref = name_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        name_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Name of the stat (should be short).
+     * </pre>
+     *
+     * <code>string name = 2;</code>
+     */
+    public Builder setName(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      name_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Name of the stat (should be short).
+     * </pre>
+     *
+     * <code>string name = 2;</code>
+     */
+    public Builder clearName() {
+      
+      name_ = getDefaultInstance().getName();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Name of the stat (should be short).
+     * </pre>
+     *
+     * <code>string name = 2;</code>
+     */
+    public Builder setNameBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      name_ = value;
+      onChanged();
+      return this;
+    }
+
+    private java.lang.Object description_ = "";
+    /**
+     * <pre>
+     * Description of the stat (might be long).
+     * </pre>
+     *
+     * <code>string description = 3;</code>
+     */
+    public java.lang.String getDescription() {
+      java.lang.Object ref = description_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        description_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Description of the stat (might be long).
+     * </pre>
+     *
+     * <code>string description = 3;</code>
+     */
+    public com.google.protobuf.ByteString
+        getDescriptionBytes() {
+      java.lang.Object ref = description_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        description_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Description of the stat (might be long).
+     * </pre>
+     *
+     * <code>string description = 3;</code>
+     */
+    public Builder setDescription(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      description_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Description of the stat (might be long).
+     * </pre>
+     *
+     * <code>string description = 3;</code>
+     */
+    public Builder clearDescription() {
+      
+      description_ = getDefaultInstance().getDescription();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Description of the stat (might be long).
+     * </pre>
+     *
+     * <code>string description = 3;</code>
+     */
+    public Builder setDescriptionBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      description_ = value;
+      onChanged();
+      return this;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.profiler.XStatMetadata)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.profiler.XStatMetadata)
+  private static final org.tensorflow.proto.profiler.XStatMetadata DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.profiler.XStatMetadata();
+  }
+
+  public static org.tensorflow.proto.profiler.XStatMetadata getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<XStatMetadata>
+      PARSER = new com.google.protobuf.AbstractParser<XStatMetadata>() {
+    @java.lang.Override
+    public XStatMetadata parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new XStatMetadata(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<XStatMetadata> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<XStatMetadata> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.profiler.XStatMetadata getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/XStatMetadataOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/XStatMetadataOrBuilder.java
new file mode 100644
index 00000000000..ebd901428ad
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/XStatMetadataOrBuilder.java
@@ -0,0 +1,54 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/profiler/protobuf/xplane.proto
+
+package org.tensorflow.proto.profiler;
+
+public interface XStatMetadataOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.profiler.XStatMetadata)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <pre>
+   * XPlane.stat_metadata map key.
+   * </pre>
+   *
+   * <code>int64 id = 1;</code>
+   */
+  long getId();
+
+  /**
+   * <pre>
+   * Name of the stat (should be short).
+   * </pre>
+   *
+   * <code>string name = 2;</code>
+   */
+  java.lang.String getName();
+  /**
+   * <pre>
+   * Name of the stat (should be short).
+   * </pre>
+   *
+   * <code>string name = 2;</code>
+   */
+  com.google.protobuf.ByteString
+      getNameBytes();
+
+  /**
+   * <pre>
+   * Description of the stat (might be long).
+   * </pre>
+   *
+   * <code>string description = 3;</code>
+   */
+  java.lang.String getDescription();
+  /**
+   * <pre>
+   * Description of the stat (might be long).
+   * </pre>
+   *
+   * <code>string description = 3;</code>
+   */
+  com.google.protobuf.ByteString
+      getDescriptionBytes();
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/XStatOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/XStatOrBuilder.java
new file mode 100644
index 00000000000..ddbd0ef1584
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/profiler/XStatOrBuilder.java
@@ -0,0 +1,59 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/profiler/protobuf/xplane.proto
+
+package org.tensorflow.proto.profiler;
+
+public interface XStatOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.profiler.XStat)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <pre>
+   * XStatMetadata.id of corresponding metadata.
+   * </pre>
+   *
+   * <code>int64 metadata_id = 1;</code>
+   */
+  long getMetadataId();
+
+  /**
+   * <code>double double_value = 2;</code>
+   */
+  double getDoubleValue();
+
+  /**
+   * <code>uint64 uint64_value = 3;</code>
+   */
+  long getUint64Value();
+
+  /**
+   * <code>int64 int64_value = 4;</code>
+   */
+  long getInt64Value();
+
+  /**
+   * <code>string str_value = 5;</code>
+   */
+  java.lang.String getStrValue();
+  /**
+   * <code>string str_value = 5;</code>
+   */
+  com.google.protobuf.ByteString
+      getStrValueBytes();
+
+  /**
+   * <code>bytes bytes_value = 6;</code>
+   */
+  com.google.protobuf.ByteString getBytesValue();
+
+  /**
+   * <pre>
+   * A string value that stored in XStatMetadata::name.
+   * </pre>
+   *
+   * <code>uint64 ref_value = 7;</code>
+   */
+  long getRefValue();
+
+  public org.tensorflow.proto.profiler.XStat.ValueCase getValueCase();
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/BfcMemoryMapProtos.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/BfcMemoryMapProtos.java
new file mode 100644
index 00000000000..b5870cc34d5
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/BfcMemoryMapProtos.java
@@ -0,0 +1,112 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/bfc_memory_map.proto
+
+package org.tensorflow.proto.util;
+
+public final class BfcMemoryMapProtos {
+  private BfcMemoryMapProtos() {}
+  public static void registerAllExtensions(
+      com.google.protobuf.ExtensionRegistryLite registry) {
+  }
+
+  public static void registerAllExtensions(
+      com.google.protobuf.ExtensionRegistry registry) {
+    registerAllExtensions(
+        (com.google.protobuf.ExtensionRegistryLite) registry);
+  }
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_MemAllocatorStats_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_MemAllocatorStats_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_MemChunk_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_MemChunk_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_BinSummary_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_BinSummary_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_SnapShot_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_SnapShot_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_MemoryDump_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_MemoryDump_fieldAccessorTable;
+
+  public static com.google.protobuf.Descriptors.FileDescriptor
+      getDescriptor() {
+    return descriptor;
+  }
+  private static  com.google.protobuf.Descriptors.FileDescriptor
+      descriptor;
+  static {
+    java.lang.String[] descriptorData = {
+      "\n-tensorflow/core/protobuf/bfc_memory_ma" +
+      "p.proto\022\ntensorflow\"\222\001\n\021MemAllocatorStat" +
+      "s\022\022\n\nnum_allocs\030\001 \001(\003\022\024\n\014bytes_in_use\030\002 " +
+      "\001(\003\022\031\n\021peak_bytes_in_use\030\003 \001(\003\022\032\n\022larges" +
+      "t_alloc_size\030\004 \001(\003\022\034\n\024fragmentation_metr" +
+      "ic\030\005 \001(\002\"\256\001\n\010MemChunk\022\017\n\007address\030\001 \001(\004\022\014" +
+      "\n\004size\030\002 \001(\003\022\026\n\016requested_size\030\003 \001(\003\022\013\n\003" +
+      "bin\030\004 \001(\005\022\017\n\007op_name\030\005 \001(\t\022\026\n\016freed_at_c" +
+      "ount\030\006 \001(\004\022\024\n\014action_count\030\007 \001(\004\022\016\n\006in_u" +
+      "se\030\010 \001(\010\022\017\n\007step_id\030\t \001(\004\"\213\001\n\nBinSummary" +
+      "\022\013\n\003bin\030\001 \001(\005\022\032\n\022total_bytes_in_use\030\002 \001(" +
+      "\003\022\032\n\022total_bytes_in_bin\030\003 \001(\003\022\033\n\023total_c" +
+      "hunks_in_use\030\004 \001(\003\022\033\n\023total_chunks_in_bi" +
+      "n\030\005 \001(\003\".\n\010SnapShot\022\024\n\014action_count\030\001 \001(" +
+      "\004\022\014\n\004size\030\002 \001(\003\"\315\001\n\nMemoryDump\022\026\n\016alloca" +
+      "tor_name\030\001 \001(\t\022+\n\013bin_summary\030\002 \003(\0132\026.te" +
+      "nsorflow.BinSummary\022#\n\005chunk\030\003 \003(\0132\024.ten" +
+      "sorflow.MemChunk\022\'\n\tsnap_shot\030\004 \003(\0132\024.te" +
+      "nsorflow.SnapShot\022,\n\005stats\030\005 \001(\0132\035.tenso" +
+      "rflow.MemAllocatorStatsB{\n\031org.tensorflo" +
+      "w.proto.utilB\022BfcMemoryMapProtosP\001ZHgith" +
+      "ub.com/tensorflow/tensorflow/tensorflow/" +
+      "go/core/core_protos_go_protob\006proto3"
+    };
+    descriptor = com.google.protobuf.Descriptors.FileDescriptor
+      .internalBuildGeneratedFileFrom(descriptorData,
+        new com.google.protobuf.Descriptors.FileDescriptor[] {
+        });
+    internal_static_tensorflow_MemAllocatorStats_descriptor =
+      getDescriptor().getMessageTypes().get(0);
+    internal_static_tensorflow_MemAllocatorStats_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_MemAllocatorStats_descriptor,
+        new java.lang.String[] { "NumAllocs", "BytesInUse", "PeakBytesInUse", "LargestAllocSize", "FragmentationMetric", });
+    internal_static_tensorflow_MemChunk_descriptor =
+      getDescriptor().getMessageTypes().get(1);
+    internal_static_tensorflow_MemChunk_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_MemChunk_descriptor,
+        new java.lang.String[] { "Address", "Size", "RequestedSize", "Bin", "OpName", "FreedAtCount", "ActionCount", "InUse", "StepId", });
+    internal_static_tensorflow_BinSummary_descriptor =
+      getDescriptor().getMessageTypes().get(2);
+    internal_static_tensorflow_BinSummary_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_BinSummary_descriptor,
+        new java.lang.String[] { "Bin", "TotalBytesInUse", "TotalBytesInBin", "TotalChunksInUse", "TotalChunksInBin", });
+    internal_static_tensorflow_SnapShot_descriptor =
+      getDescriptor().getMessageTypes().get(3);
+    internal_static_tensorflow_SnapShot_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_SnapShot_descriptor,
+        new java.lang.String[] { "ActionCount", "Size", });
+    internal_static_tensorflow_MemoryDump_descriptor =
+      getDescriptor().getMessageTypes().get(4);
+    internal_static_tensorflow_MemoryDump_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_MemoryDump_descriptor,
+        new java.lang.String[] { "AllocatorName", "BinSummary", "Chunk", "SnapShot", "Stats", });
+  }
+
+  // @@protoc_insertion_point(outer_class_scope)
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/BinSummary.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/BinSummary.java
new file mode 100644
index 00000000000..66f221b62f0
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/BinSummary.java
@@ -0,0 +1,708 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/bfc_memory_map.proto
+
+package org.tensorflow.proto.util;
+
+/**
+ * Protobuf type {@code tensorflow.BinSummary}
+ */
+public  final class BinSummary extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.BinSummary)
+    BinSummaryOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use BinSummary.newBuilder() to construct.
+  private BinSummary(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private BinSummary() {
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new BinSummary();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private BinSummary(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 8: {
+
+            bin_ = input.readInt32();
+            break;
+          }
+          case 16: {
+
+            totalBytesInUse_ = input.readInt64();
+            break;
+          }
+          case 24: {
+
+            totalBytesInBin_ = input.readInt64();
+            break;
+          }
+          case 32: {
+
+            totalChunksInUse_ = input.readInt64();
+            break;
+          }
+          case 40: {
+
+            totalChunksInBin_ = input.readInt64();
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.util.BfcMemoryMapProtos.internal_static_tensorflow_BinSummary_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.util.BfcMemoryMapProtos.internal_static_tensorflow_BinSummary_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.util.BinSummary.class, org.tensorflow.proto.util.BinSummary.Builder.class);
+  }
+
+  public static final int BIN_FIELD_NUMBER = 1;
+  private int bin_;
+  /**
+   * <code>int32 bin = 1;</code>
+   */
+  public int getBin() {
+    return bin_;
+  }
+
+  public static final int TOTAL_BYTES_IN_USE_FIELD_NUMBER = 2;
+  private long totalBytesInUse_;
+  /**
+   * <code>int64 total_bytes_in_use = 2;</code>
+   */
+  public long getTotalBytesInUse() {
+    return totalBytesInUse_;
+  }
+
+  public static final int TOTAL_BYTES_IN_BIN_FIELD_NUMBER = 3;
+  private long totalBytesInBin_;
+  /**
+   * <code>int64 total_bytes_in_bin = 3;</code>
+   */
+  public long getTotalBytesInBin() {
+    return totalBytesInBin_;
+  }
+
+  public static final int TOTAL_CHUNKS_IN_USE_FIELD_NUMBER = 4;
+  private long totalChunksInUse_;
+  /**
+   * <code>int64 total_chunks_in_use = 4;</code>
+   */
+  public long getTotalChunksInUse() {
+    return totalChunksInUse_;
+  }
+
+  public static final int TOTAL_CHUNKS_IN_BIN_FIELD_NUMBER = 5;
+  private long totalChunksInBin_;
+  /**
+   * <code>int64 total_chunks_in_bin = 5;</code>
+   */
+  public long getTotalChunksInBin() {
+    return totalChunksInBin_;
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    if (bin_ != 0) {
+      output.writeInt32(1, bin_);
+    }
+    if (totalBytesInUse_ != 0L) {
+      output.writeInt64(2, totalBytesInUse_);
+    }
+    if (totalBytesInBin_ != 0L) {
+      output.writeInt64(3, totalBytesInBin_);
+    }
+    if (totalChunksInUse_ != 0L) {
+      output.writeInt64(4, totalChunksInUse_);
+    }
+    if (totalChunksInBin_ != 0L) {
+      output.writeInt64(5, totalChunksInBin_);
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (bin_ != 0) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt32Size(1, bin_);
+    }
+    if (totalBytesInUse_ != 0L) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt64Size(2, totalBytesInUse_);
+    }
+    if (totalBytesInBin_ != 0L) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt64Size(3, totalBytesInBin_);
+    }
+    if (totalChunksInUse_ != 0L) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt64Size(4, totalChunksInUse_);
+    }
+    if (totalChunksInBin_ != 0L) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt64Size(5, totalChunksInBin_);
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.util.BinSummary)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.util.BinSummary other = (org.tensorflow.proto.util.BinSummary) obj;
+
+    if (getBin()
+        != other.getBin()) return false;
+    if (getTotalBytesInUse()
+        != other.getTotalBytesInUse()) return false;
+    if (getTotalBytesInBin()
+        != other.getTotalBytesInBin()) return false;
+    if (getTotalChunksInUse()
+        != other.getTotalChunksInUse()) return false;
+    if (getTotalChunksInBin()
+        != other.getTotalChunksInBin()) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    hash = (37 * hash) + BIN_FIELD_NUMBER;
+    hash = (53 * hash) + getBin();
+    hash = (37 * hash) + TOTAL_BYTES_IN_USE_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+        getTotalBytesInUse());
+    hash = (37 * hash) + TOTAL_BYTES_IN_BIN_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+        getTotalBytesInBin());
+    hash = (37 * hash) + TOTAL_CHUNKS_IN_USE_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+        getTotalChunksInUse());
+    hash = (37 * hash) + TOTAL_CHUNKS_IN_BIN_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+        getTotalChunksInBin());
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.util.BinSummary parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.BinSummary parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.BinSummary parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.BinSummary parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.BinSummary parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.BinSummary parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.BinSummary parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.BinSummary parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.BinSummary parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.BinSummary parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.BinSummary parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.BinSummary parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.util.BinSummary prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * Protobuf type {@code tensorflow.BinSummary}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.BinSummary)
+      org.tensorflow.proto.util.BinSummaryOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.util.BfcMemoryMapProtos.internal_static_tensorflow_BinSummary_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.util.BfcMemoryMapProtos.internal_static_tensorflow_BinSummary_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.util.BinSummary.class, org.tensorflow.proto.util.BinSummary.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.util.BinSummary.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      bin_ = 0;
+
+      totalBytesInUse_ = 0L;
+
+      totalBytesInBin_ = 0L;
+
+      totalChunksInUse_ = 0L;
+
+      totalChunksInBin_ = 0L;
+
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.util.BfcMemoryMapProtos.internal_static_tensorflow_BinSummary_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.BinSummary getDefaultInstanceForType() {
+      return org.tensorflow.proto.util.BinSummary.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.BinSummary build() {
+      org.tensorflow.proto.util.BinSummary result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.BinSummary buildPartial() {
+      org.tensorflow.proto.util.BinSummary result = new org.tensorflow.proto.util.BinSummary(this);
+      result.bin_ = bin_;
+      result.totalBytesInUse_ = totalBytesInUse_;
+      result.totalBytesInBin_ = totalBytesInBin_;
+      result.totalChunksInUse_ = totalChunksInUse_;
+      result.totalChunksInBin_ = totalChunksInBin_;
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.util.BinSummary) {
+        return mergeFrom((org.tensorflow.proto.util.BinSummary)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.util.BinSummary other) {
+      if (other == org.tensorflow.proto.util.BinSummary.getDefaultInstance()) return this;
+      if (other.getBin() != 0) {
+        setBin(other.getBin());
+      }
+      if (other.getTotalBytesInUse() != 0L) {
+        setTotalBytesInUse(other.getTotalBytesInUse());
+      }
+      if (other.getTotalBytesInBin() != 0L) {
+        setTotalBytesInBin(other.getTotalBytesInBin());
+      }
+      if (other.getTotalChunksInUse() != 0L) {
+        setTotalChunksInUse(other.getTotalChunksInUse());
+      }
+      if (other.getTotalChunksInBin() != 0L) {
+        setTotalChunksInBin(other.getTotalChunksInBin());
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.util.BinSummary parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.util.BinSummary) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+
+    private int bin_ ;
+    /**
+     * <code>int32 bin = 1;</code>
+     */
+    public int getBin() {
+      return bin_;
+    }
+    /**
+     * <code>int32 bin = 1;</code>
+     */
+    public Builder setBin(int value) {
+      
+      bin_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>int32 bin = 1;</code>
+     */
+    public Builder clearBin() {
+      
+      bin_ = 0;
+      onChanged();
+      return this;
+    }
+
+    private long totalBytesInUse_ ;
+    /**
+     * <code>int64 total_bytes_in_use = 2;</code>
+     */
+    public long getTotalBytesInUse() {
+      return totalBytesInUse_;
+    }
+    /**
+     * <code>int64 total_bytes_in_use = 2;</code>
+     */
+    public Builder setTotalBytesInUse(long value) {
+      
+      totalBytesInUse_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>int64 total_bytes_in_use = 2;</code>
+     */
+    public Builder clearTotalBytesInUse() {
+      
+      totalBytesInUse_ = 0L;
+      onChanged();
+      return this;
+    }
+
+    private long totalBytesInBin_ ;
+    /**
+     * <code>int64 total_bytes_in_bin = 3;</code>
+     */
+    public long getTotalBytesInBin() {
+      return totalBytesInBin_;
+    }
+    /**
+     * <code>int64 total_bytes_in_bin = 3;</code>
+     */
+    public Builder setTotalBytesInBin(long value) {
+      
+      totalBytesInBin_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>int64 total_bytes_in_bin = 3;</code>
+     */
+    public Builder clearTotalBytesInBin() {
+      
+      totalBytesInBin_ = 0L;
+      onChanged();
+      return this;
+    }
+
+    private long totalChunksInUse_ ;
+    /**
+     * <code>int64 total_chunks_in_use = 4;</code>
+     */
+    public long getTotalChunksInUse() {
+      return totalChunksInUse_;
+    }
+    /**
+     * <code>int64 total_chunks_in_use = 4;</code>
+     */
+    public Builder setTotalChunksInUse(long value) {
+      
+      totalChunksInUse_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>int64 total_chunks_in_use = 4;</code>
+     */
+    public Builder clearTotalChunksInUse() {
+      
+      totalChunksInUse_ = 0L;
+      onChanged();
+      return this;
+    }
+
+    private long totalChunksInBin_ ;
+    /**
+     * <code>int64 total_chunks_in_bin = 5;</code>
+     */
+    public long getTotalChunksInBin() {
+      return totalChunksInBin_;
+    }
+    /**
+     * <code>int64 total_chunks_in_bin = 5;</code>
+     */
+    public Builder setTotalChunksInBin(long value) {
+      
+      totalChunksInBin_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>int64 total_chunks_in_bin = 5;</code>
+     */
+    public Builder clearTotalChunksInBin() {
+      
+      totalChunksInBin_ = 0L;
+      onChanged();
+      return this;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.BinSummary)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.BinSummary)
+  private static final org.tensorflow.proto.util.BinSummary DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.util.BinSummary();
+  }
+
+  public static org.tensorflow.proto.util.BinSummary getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<BinSummary>
+      PARSER = new com.google.protobuf.AbstractParser<BinSummary>() {
+    @java.lang.Override
+    public BinSummary parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new BinSummary(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<BinSummary> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<BinSummary> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.util.BinSummary getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/BinSummaryOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/BinSummaryOrBuilder.java
new file mode 100644
index 00000000000..077107aa85f
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/BinSummaryOrBuilder.java
@@ -0,0 +1,34 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/bfc_memory_map.proto
+
+package org.tensorflow.proto.util;
+
+public interface BinSummaryOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.BinSummary)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <code>int32 bin = 1;</code>
+   */
+  int getBin();
+
+  /**
+   * <code>int64 total_bytes_in_use = 2;</code>
+   */
+  long getTotalBytesInUse();
+
+  /**
+   * <code>int64 total_bytes_in_bin = 3;</code>
+   */
+  long getTotalBytesInBin();
+
+  /**
+   * <code>int64 total_chunks_in_use = 4;</code>
+   */
+  long getTotalChunksInUse();
+
+  /**
+   * <code>int64 total_chunks_in_bin = 5;</code>
+   */
+  long getTotalChunksInBin();
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/BundleEntryProto.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/BundleEntryProto.java
new file mode 100644
index 00000000000..a74911cdde6
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/BundleEntryProto.java
@@ -0,0 +1,1540 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/tensor_bundle.proto
+
+package org.tensorflow.proto.util;
+
+/**
+ * <pre>
+ * Describes the metadata related to a checkpointed tensor.
+ * </pre>
+ *
+ * Protobuf type {@code tensorflow.BundleEntryProto}
+ */
+public  final class BundleEntryProto extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.BundleEntryProto)
+    BundleEntryProtoOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use BundleEntryProto.newBuilder() to construct.
+  private BundleEntryProto(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private BundleEntryProto() {
+    dtype_ = 0;
+    slices_ = java.util.Collections.emptyList();
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new BundleEntryProto();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private BundleEntryProto(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    int mutable_bitField0_ = 0;
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 8: {
+            int rawValue = input.readEnum();
+
+            dtype_ = rawValue;
+            break;
+          }
+          case 18: {
+            org.tensorflow.proto.framework.TensorShapeProto.Builder subBuilder = null;
+            if (shape_ != null) {
+              subBuilder = shape_.toBuilder();
+            }
+            shape_ = input.readMessage(org.tensorflow.proto.framework.TensorShapeProto.parser(), extensionRegistry);
+            if (subBuilder != null) {
+              subBuilder.mergeFrom(shape_);
+              shape_ = subBuilder.buildPartial();
+            }
+
+            break;
+          }
+          case 24: {
+
+            shardId_ = input.readInt32();
+            break;
+          }
+          case 32: {
+
+            offset_ = input.readInt64();
+            break;
+          }
+          case 40: {
+
+            size_ = input.readInt64();
+            break;
+          }
+          case 53: {
+
+            crc32C_ = input.readFixed32();
+            break;
+          }
+          case 58: {
+            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
+              slices_ = new java.util.ArrayList<org.tensorflow.proto.framework.TensorSliceProto>();
+              mutable_bitField0_ |= 0x00000001;
+            }
+            slices_.add(
+                input.readMessage(org.tensorflow.proto.framework.TensorSliceProto.parser(), extensionRegistry));
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      if (((mutable_bitField0_ & 0x00000001) != 0)) {
+        slices_ = java.util.Collections.unmodifiableList(slices_);
+      }
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.util.TensorBundleProtos.internal_static_tensorflow_BundleEntryProto_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.util.TensorBundleProtos.internal_static_tensorflow_BundleEntryProto_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.util.BundleEntryProto.class, org.tensorflow.proto.util.BundleEntryProto.Builder.class);
+  }
+
+  public static final int DTYPE_FIELD_NUMBER = 1;
+  private int dtype_;
+  /**
+   * <pre>
+   * The tensor dtype and shape.
+   * </pre>
+   *
+   * <code>.tensorflow.DataType dtype = 1;</code>
+   */
+  public int getDtypeValue() {
+    return dtype_;
+  }
+  /**
+   * <pre>
+   * The tensor dtype and shape.
+   * </pre>
+   *
+   * <code>.tensorflow.DataType dtype = 1;</code>
+   */
+  public org.tensorflow.proto.framework.DataType getDtype() {
+    @SuppressWarnings("deprecation")
+    org.tensorflow.proto.framework.DataType result = org.tensorflow.proto.framework.DataType.valueOf(dtype_);
+    return result == null ? org.tensorflow.proto.framework.DataType.UNRECOGNIZED : result;
+  }
+
+  public static final int SHAPE_FIELD_NUMBER = 2;
+  private org.tensorflow.proto.framework.TensorShapeProto shape_;
+  /**
+   * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+   */
+  public boolean hasShape() {
+    return shape_ != null;
+  }
+  /**
+   * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+   */
+  public org.tensorflow.proto.framework.TensorShapeProto getShape() {
+    return shape_ == null ? org.tensorflow.proto.framework.TensorShapeProto.getDefaultInstance() : shape_;
+  }
+  /**
+   * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+   */
+  public org.tensorflow.proto.framework.TensorShapeProtoOrBuilder getShapeOrBuilder() {
+    return getShape();
+  }
+
+  public static final int SHARD_ID_FIELD_NUMBER = 3;
+  private int shardId_;
+  /**
+   * <pre>
+   * The binary content of the tensor lies in:
+   *   File "shard_id": bytes [offset, offset + size).
+   * </pre>
+   *
+   * <code>int32 shard_id = 3;</code>
+   */
+  public int getShardId() {
+    return shardId_;
+  }
+
+  public static final int OFFSET_FIELD_NUMBER = 4;
+  private long offset_;
+  /**
+   * <code>int64 offset = 4;</code>
+   */
+  public long getOffset() {
+    return offset_;
+  }
+
+  public static final int SIZE_FIELD_NUMBER = 5;
+  private long size_;
+  /**
+   * <code>int64 size = 5;</code>
+   */
+  public long getSize() {
+    return size_;
+  }
+
+  public static final int CRC32C_FIELD_NUMBER = 6;
+  private int crc32C_;
+  /**
+   * <pre>
+   * The CRC32C checksum of the tensor bytes.
+   * </pre>
+   *
+   * <code>fixed32 crc32c = 6;</code>
+   */
+  public int getCrc32C() {
+    return crc32C_;
+  }
+
+  public static final int SLICES_FIELD_NUMBER = 7;
+  private java.util.List<org.tensorflow.proto.framework.TensorSliceProto> slices_;
+  /**
+   * <pre>
+   * Iff present, this entry represents a partitioned tensor.  The previous
+   * fields are interpreted as follows:
+   *   "dtype", "shape": describe the full tensor.
+   *   "shard_id", "offset", "size", "crc32c": all IGNORED.
+   *      These information for each slice can be looked up in their own
+   *      BundleEntryProto, keyed by each "slice_name".
+   * </pre>
+   *
+   * <code>repeated .tensorflow.TensorSliceProto slices = 7;</code>
+   */
+  public java.util.List<org.tensorflow.proto.framework.TensorSliceProto> getSlicesList() {
+    return slices_;
+  }
+  /**
+   * <pre>
+   * Iff present, this entry represents a partitioned tensor.  The previous
+   * fields are interpreted as follows:
+   *   "dtype", "shape": describe the full tensor.
+   *   "shard_id", "offset", "size", "crc32c": all IGNORED.
+   *      These information for each slice can be looked up in their own
+   *      BundleEntryProto, keyed by each "slice_name".
+   * </pre>
+   *
+   * <code>repeated .tensorflow.TensorSliceProto slices = 7;</code>
+   */
+  public java.util.List<? extends org.tensorflow.proto.framework.TensorSliceProtoOrBuilder> 
+      getSlicesOrBuilderList() {
+    return slices_;
+  }
+  /**
+   * <pre>
+   * Iff present, this entry represents a partitioned tensor.  The previous
+   * fields are interpreted as follows:
+   *   "dtype", "shape": describe the full tensor.
+   *   "shard_id", "offset", "size", "crc32c": all IGNORED.
+   *      These information for each slice can be looked up in their own
+   *      BundleEntryProto, keyed by each "slice_name".
+   * </pre>
+   *
+   * <code>repeated .tensorflow.TensorSliceProto slices = 7;</code>
+   */
+  public int getSlicesCount() {
+    return slices_.size();
+  }
+  /**
+   * <pre>
+   * Iff present, this entry represents a partitioned tensor.  The previous
+   * fields are interpreted as follows:
+   *   "dtype", "shape": describe the full tensor.
+   *   "shard_id", "offset", "size", "crc32c": all IGNORED.
+   *      These information for each slice can be looked up in their own
+   *      BundleEntryProto, keyed by each "slice_name".
+   * </pre>
+   *
+   * <code>repeated .tensorflow.TensorSliceProto slices = 7;</code>
+   */
+  public org.tensorflow.proto.framework.TensorSliceProto getSlices(int index) {
+    return slices_.get(index);
+  }
+  /**
+   * <pre>
+   * Iff present, this entry represents a partitioned tensor.  The previous
+   * fields are interpreted as follows:
+   *   "dtype", "shape": describe the full tensor.
+   *   "shard_id", "offset", "size", "crc32c": all IGNORED.
+   *      These information for each slice can be looked up in their own
+   *      BundleEntryProto, keyed by each "slice_name".
+   * </pre>
+   *
+   * <code>repeated .tensorflow.TensorSliceProto slices = 7;</code>
+   */
+  public org.tensorflow.proto.framework.TensorSliceProtoOrBuilder getSlicesOrBuilder(
+      int index) {
+    return slices_.get(index);
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    if (dtype_ != org.tensorflow.proto.framework.DataType.DT_INVALID.getNumber()) {
+      output.writeEnum(1, dtype_);
+    }
+    if (shape_ != null) {
+      output.writeMessage(2, getShape());
+    }
+    if (shardId_ != 0) {
+      output.writeInt32(3, shardId_);
+    }
+    if (offset_ != 0L) {
+      output.writeInt64(4, offset_);
+    }
+    if (size_ != 0L) {
+      output.writeInt64(5, size_);
+    }
+    if (crc32C_ != 0) {
+      output.writeFixed32(6, crc32C_);
+    }
+    for (int i = 0; i < slices_.size(); i++) {
+      output.writeMessage(7, slices_.get(i));
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (dtype_ != org.tensorflow.proto.framework.DataType.DT_INVALID.getNumber()) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeEnumSize(1, dtype_);
+    }
+    if (shape_ != null) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(2, getShape());
+    }
+    if (shardId_ != 0) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt32Size(3, shardId_);
+    }
+    if (offset_ != 0L) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt64Size(4, offset_);
+    }
+    if (size_ != 0L) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt64Size(5, size_);
+    }
+    if (crc32C_ != 0) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeFixed32Size(6, crc32C_);
+    }
+    for (int i = 0; i < slices_.size(); i++) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(7, slices_.get(i));
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.util.BundleEntryProto)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.util.BundleEntryProto other = (org.tensorflow.proto.util.BundleEntryProto) obj;
+
+    if (dtype_ != other.dtype_) return false;
+    if (hasShape() != other.hasShape()) return false;
+    if (hasShape()) {
+      if (!getShape()
+          .equals(other.getShape())) return false;
+    }
+    if (getShardId()
+        != other.getShardId()) return false;
+    if (getOffset()
+        != other.getOffset()) return false;
+    if (getSize()
+        != other.getSize()) return false;
+    if (getCrc32C()
+        != other.getCrc32C()) return false;
+    if (!getSlicesList()
+        .equals(other.getSlicesList())) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    hash = (37 * hash) + DTYPE_FIELD_NUMBER;
+    hash = (53 * hash) + dtype_;
+    if (hasShape()) {
+      hash = (37 * hash) + SHAPE_FIELD_NUMBER;
+      hash = (53 * hash) + getShape().hashCode();
+    }
+    hash = (37 * hash) + SHARD_ID_FIELD_NUMBER;
+    hash = (53 * hash) + getShardId();
+    hash = (37 * hash) + OFFSET_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+        getOffset());
+    hash = (37 * hash) + SIZE_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+        getSize());
+    hash = (37 * hash) + CRC32C_FIELD_NUMBER;
+    hash = (53 * hash) + getCrc32C();
+    if (getSlicesCount() > 0) {
+      hash = (37 * hash) + SLICES_FIELD_NUMBER;
+      hash = (53 * hash) + getSlicesList().hashCode();
+    }
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.util.BundleEntryProto parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.BundleEntryProto parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.BundleEntryProto parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.BundleEntryProto parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.BundleEntryProto parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.BundleEntryProto parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.BundleEntryProto parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.BundleEntryProto parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.BundleEntryProto parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.BundleEntryProto parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.BundleEntryProto parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.BundleEntryProto parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.util.BundleEntryProto prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * <pre>
+   * Describes the metadata related to a checkpointed tensor.
+   * </pre>
+   *
+   * Protobuf type {@code tensorflow.BundleEntryProto}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.BundleEntryProto)
+      org.tensorflow.proto.util.BundleEntryProtoOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.util.TensorBundleProtos.internal_static_tensorflow_BundleEntryProto_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.util.TensorBundleProtos.internal_static_tensorflow_BundleEntryProto_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.util.BundleEntryProto.class, org.tensorflow.proto.util.BundleEntryProto.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.util.BundleEntryProto.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+        getSlicesFieldBuilder();
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      dtype_ = 0;
+
+      if (shapeBuilder_ == null) {
+        shape_ = null;
+      } else {
+        shape_ = null;
+        shapeBuilder_ = null;
+      }
+      shardId_ = 0;
+
+      offset_ = 0L;
+
+      size_ = 0L;
+
+      crc32C_ = 0;
+
+      if (slicesBuilder_ == null) {
+        slices_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000001);
+      } else {
+        slicesBuilder_.clear();
+      }
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.util.TensorBundleProtos.internal_static_tensorflow_BundleEntryProto_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.BundleEntryProto getDefaultInstanceForType() {
+      return org.tensorflow.proto.util.BundleEntryProto.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.BundleEntryProto build() {
+      org.tensorflow.proto.util.BundleEntryProto result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.BundleEntryProto buildPartial() {
+      org.tensorflow.proto.util.BundleEntryProto result = new org.tensorflow.proto.util.BundleEntryProto(this);
+      int from_bitField0_ = bitField0_;
+      result.dtype_ = dtype_;
+      if (shapeBuilder_ == null) {
+        result.shape_ = shape_;
+      } else {
+        result.shape_ = shapeBuilder_.build();
+      }
+      result.shardId_ = shardId_;
+      result.offset_ = offset_;
+      result.size_ = size_;
+      result.crc32C_ = crc32C_;
+      if (slicesBuilder_ == null) {
+        if (((bitField0_ & 0x00000001) != 0)) {
+          slices_ = java.util.Collections.unmodifiableList(slices_);
+          bitField0_ = (bitField0_ & ~0x00000001);
+        }
+        result.slices_ = slices_;
+      } else {
+        result.slices_ = slicesBuilder_.build();
+      }
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.util.BundleEntryProto) {
+        return mergeFrom((org.tensorflow.proto.util.BundleEntryProto)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.util.BundleEntryProto other) {
+      if (other == org.tensorflow.proto.util.BundleEntryProto.getDefaultInstance()) return this;
+      if (other.dtype_ != 0) {
+        setDtypeValue(other.getDtypeValue());
+      }
+      if (other.hasShape()) {
+        mergeShape(other.getShape());
+      }
+      if (other.getShardId() != 0) {
+        setShardId(other.getShardId());
+      }
+      if (other.getOffset() != 0L) {
+        setOffset(other.getOffset());
+      }
+      if (other.getSize() != 0L) {
+        setSize(other.getSize());
+      }
+      if (other.getCrc32C() != 0) {
+        setCrc32C(other.getCrc32C());
+      }
+      if (slicesBuilder_ == null) {
+        if (!other.slices_.isEmpty()) {
+          if (slices_.isEmpty()) {
+            slices_ = other.slices_;
+            bitField0_ = (bitField0_ & ~0x00000001);
+          } else {
+            ensureSlicesIsMutable();
+            slices_.addAll(other.slices_);
+          }
+          onChanged();
+        }
+      } else {
+        if (!other.slices_.isEmpty()) {
+          if (slicesBuilder_.isEmpty()) {
+            slicesBuilder_.dispose();
+            slicesBuilder_ = null;
+            slices_ = other.slices_;
+            bitField0_ = (bitField0_ & ~0x00000001);
+            slicesBuilder_ = 
+              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
+                 getSlicesFieldBuilder() : null;
+          } else {
+            slicesBuilder_.addAllMessages(other.slices_);
+          }
+        }
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.util.BundleEntryProto parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.util.BundleEntryProto) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+    private int bitField0_;
+
+    private int dtype_ = 0;
+    /**
+     * <pre>
+     * The tensor dtype and shape.
+     * </pre>
+     *
+     * <code>.tensorflow.DataType dtype = 1;</code>
+     */
+    public int getDtypeValue() {
+      return dtype_;
+    }
+    /**
+     * <pre>
+     * The tensor dtype and shape.
+     * </pre>
+     *
+     * <code>.tensorflow.DataType dtype = 1;</code>
+     */
+    public Builder setDtypeValue(int value) {
+      dtype_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * The tensor dtype and shape.
+     * </pre>
+     *
+     * <code>.tensorflow.DataType dtype = 1;</code>
+     */
+    public org.tensorflow.proto.framework.DataType getDtype() {
+      @SuppressWarnings("deprecation")
+      org.tensorflow.proto.framework.DataType result = org.tensorflow.proto.framework.DataType.valueOf(dtype_);
+      return result == null ? org.tensorflow.proto.framework.DataType.UNRECOGNIZED : result;
+    }
+    /**
+     * <pre>
+     * The tensor dtype and shape.
+     * </pre>
+     *
+     * <code>.tensorflow.DataType dtype = 1;</code>
+     */
+    public Builder setDtype(org.tensorflow.proto.framework.DataType value) {
+      if (value == null) {
+        throw new NullPointerException();
+      }
+      
+      dtype_ = value.getNumber();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * The tensor dtype and shape.
+     * </pre>
+     *
+     * <code>.tensorflow.DataType dtype = 1;</code>
+     */
+    public Builder clearDtype() {
+      
+      dtype_ = 0;
+      onChanged();
+      return this;
+    }
+
+    private org.tensorflow.proto.framework.TensorShapeProto shape_;
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.TensorShapeProto, org.tensorflow.proto.framework.TensorShapeProto.Builder, org.tensorflow.proto.framework.TensorShapeProtoOrBuilder> shapeBuilder_;
+    /**
+     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+     */
+    public boolean hasShape() {
+      return shapeBuilder_ != null || shape_ != null;
+    }
+    /**
+     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+     */
+    public org.tensorflow.proto.framework.TensorShapeProto getShape() {
+      if (shapeBuilder_ == null) {
+        return shape_ == null ? org.tensorflow.proto.framework.TensorShapeProto.getDefaultInstance() : shape_;
+      } else {
+        return shapeBuilder_.getMessage();
+      }
+    }
+    /**
+     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+     */
+    public Builder setShape(org.tensorflow.proto.framework.TensorShapeProto value) {
+      if (shapeBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        shape_ = value;
+        onChanged();
+      } else {
+        shapeBuilder_.setMessage(value);
+      }
+
+      return this;
+    }
+    /**
+     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+     */
+    public Builder setShape(
+        org.tensorflow.proto.framework.TensorShapeProto.Builder builderForValue) {
+      if (shapeBuilder_ == null) {
+        shape_ = builderForValue.build();
+        onChanged();
+      } else {
+        shapeBuilder_.setMessage(builderForValue.build());
+      }
+
+      return this;
+    }
+    /**
+     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+     */
+    public Builder mergeShape(org.tensorflow.proto.framework.TensorShapeProto value) {
+      if (shapeBuilder_ == null) {
+        if (shape_ != null) {
+          shape_ =
+            org.tensorflow.proto.framework.TensorShapeProto.newBuilder(shape_).mergeFrom(value).buildPartial();
+        } else {
+          shape_ = value;
+        }
+        onChanged();
+      } else {
+        shapeBuilder_.mergeFrom(value);
+      }
+
+      return this;
+    }
+    /**
+     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+     */
+    public Builder clearShape() {
+      if (shapeBuilder_ == null) {
+        shape_ = null;
+        onChanged();
+      } else {
+        shape_ = null;
+        shapeBuilder_ = null;
+      }
+
+      return this;
+    }
+    /**
+     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+     */
+    public org.tensorflow.proto.framework.TensorShapeProto.Builder getShapeBuilder() {
+      
+      onChanged();
+      return getShapeFieldBuilder().getBuilder();
+    }
+    /**
+     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+     */
+    public org.tensorflow.proto.framework.TensorShapeProtoOrBuilder getShapeOrBuilder() {
+      if (shapeBuilder_ != null) {
+        return shapeBuilder_.getMessageOrBuilder();
+      } else {
+        return shape_ == null ?
+            org.tensorflow.proto.framework.TensorShapeProto.getDefaultInstance() : shape_;
+      }
+    }
+    /**
+     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+     */
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.TensorShapeProto, org.tensorflow.proto.framework.TensorShapeProto.Builder, org.tensorflow.proto.framework.TensorShapeProtoOrBuilder> 
+        getShapeFieldBuilder() {
+      if (shapeBuilder_ == null) {
+        shapeBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+            org.tensorflow.proto.framework.TensorShapeProto, org.tensorflow.proto.framework.TensorShapeProto.Builder, org.tensorflow.proto.framework.TensorShapeProtoOrBuilder>(
+                getShape(),
+                getParentForChildren(),
+                isClean());
+        shape_ = null;
+      }
+      return shapeBuilder_;
+    }
+
+    private int shardId_ ;
+    /**
+     * <pre>
+     * The binary content of the tensor lies in:
+     *   File "shard_id": bytes [offset, offset + size).
+     * </pre>
+     *
+     * <code>int32 shard_id = 3;</code>
+     */
+    public int getShardId() {
+      return shardId_;
+    }
+    /**
+     * <pre>
+     * The binary content of the tensor lies in:
+     *   File "shard_id": bytes [offset, offset + size).
+     * </pre>
+     *
+     * <code>int32 shard_id = 3;</code>
+     */
+    public Builder setShardId(int value) {
+      
+      shardId_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * The binary content of the tensor lies in:
+     *   File "shard_id": bytes [offset, offset + size).
+     * </pre>
+     *
+     * <code>int32 shard_id = 3;</code>
+     */
+    public Builder clearShardId() {
+      
+      shardId_ = 0;
+      onChanged();
+      return this;
+    }
+
+    private long offset_ ;
+    /**
+     * <code>int64 offset = 4;</code>
+     */
+    public long getOffset() {
+      return offset_;
+    }
+    /**
+     * <code>int64 offset = 4;</code>
+     */
+    public Builder setOffset(long value) {
+      
+      offset_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>int64 offset = 4;</code>
+     */
+    public Builder clearOffset() {
+      
+      offset_ = 0L;
+      onChanged();
+      return this;
+    }
+
+    private long size_ ;
+    /**
+     * <code>int64 size = 5;</code>
+     */
+    public long getSize() {
+      return size_;
+    }
+    /**
+     * <code>int64 size = 5;</code>
+     */
+    public Builder setSize(long value) {
+      
+      size_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>int64 size = 5;</code>
+     */
+    public Builder clearSize() {
+      
+      size_ = 0L;
+      onChanged();
+      return this;
+    }
+
+    private int crc32C_ ;
+    /**
+     * <pre>
+     * The CRC32C checksum of the tensor bytes.
+     * </pre>
+     *
+     * <code>fixed32 crc32c = 6;</code>
+     */
+    public int getCrc32C() {
+      return crc32C_;
+    }
+    /**
+     * <pre>
+     * The CRC32C checksum of the tensor bytes.
+     * </pre>
+     *
+     * <code>fixed32 crc32c = 6;</code>
+     */
+    public Builder setCrc32C(int value) {
+      
+      crc32C_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * The CRC32C checksum of the tensor bytes.
+     * </pre>
+     *
+     * <code>fixed32 crc32c = 6;</code>
+     */
+    public Builder clearCrc32C() {
+      
+      crc32C_ = 0;
+      onChanged();
+      return this;
+    }
+
+    private java.util.List<org.tensorflow.proto.framework.TensorSliceProto> slices_ =
+      java.util.Collections.emptyList();
+    private void ensureSlicesIsMutable() {
+      if (!((bitField0_ & 0x00000001) != 0)) {
+        slices_ = new java.util.ArrayList<org.tensorflow.proto.framework.TensorSliceProto>(slices_);
+        bitField0_ |= 0x00000001;
+       }
+    }
+
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.framework.TensorSliceProto, org.tensorflow.proto.framework.TensorSliceProto.Builder, org.tensorflow.proto.framework.TensorSliceProtoOrBuilder> slicesBuilder_;
+
+    /**
+     * <pre>
+     * Iff present, this entry represents a partitioned tensor.  The previous
+     * fields are interpreted as follows:
+     *   "dtype", "shape": describe the full tensor.
+     *   "shard_id", "offset", "size", "crc32c": all IGNORED.
+     *      These information for each slice can be looked up in their own
+     *      BundleEntryProto, keyed by each "slice_name".
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorSliceProto slices = 7;</code>
+     */
+    public java.util.List<org.tensorflow.proto.framework.TensorSliceProto> getSlicesList() {
+      if (slicesBuilder_ == null) {
+        return java.util.Collections.unmodifiableList(slices_);
+      } else {
+        return slicesBuilder_.getMessageList();
+      }
+    }
+    /**
+     * <pre>
+     * Iff present, this entry represents a partitioned tensor.  The previous
+     * fields are interpreted as follows:
+     *   "dtype", "shape": describe the full tensor.
+     *   "shard_id", "offset", "size", "crc32c": all IGNORED.
+     *      These information for each slice can be looked up in their own
+     *      BundleEntryProto, keyed by each "slice_name".
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorSliceProto slices = 7;</code>
+     */
+    public int getSlicesCount() {
+      if (slicesBuilder_ == null) {
+        return slices_.size();
+      } else {
+        return slicesBuilder_.getCount();
+      }
+    }
+    /**
+     * <pre>
+     * Iff present, this entry represents a partitioned tensor.  The previous
+     * fields are interpreted as follows:
+     *   "dtype", "shape": describe the full tensor.
+     *   "shard_id", "offset", "size", "crc32c": all IGNORED.
+     *      These information for each slice can be looked up in their own
+     *      BundleEntryProto, keyed by each "slice_name".
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorSliceProto slices = 7;</code>
+     */
+    public org.tensorflow.proto.framework.TensorSliceProto getSlices(int index) {
+      if (slicesBuilder_ == null) {
+        return slices_.get(index);
+      } else {
+        return slicesBuilder_.getMessage(index);
+      }
+    }
+    /**
+     * <pre>
+     * Iff present, this entry represents a partitioned tensor.  The previous
+     * fields are interpreted as follows:
+     *   "dtype", "shape": describe the full tensor.
+     *   "shard_id", "offset", "size", "crc32c": all IGNORED.
+     *      These information for each slice can be looked up in their own
+     *      BundleEntryProto, keyed by each "slice_name".
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorSliceProto slices = 7;</code>
+     */
+    public Builder setSlices(
+        int index, org.tensorflow.proto.framework.TensorSliceProto value) {
+      if (slicesBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureSlicesIsMutable();
+        slices_.set(index, value);
+        onChanged();
+      } else {
+        slicesBuilder_.setMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Iff present, this entry represents a partitioned tensor.  The previous
+     * fields are interpreted as follows:
+     *   "dtype", "shape": describe the full tensor.
+     *   "shard_id", "offset", "size", "crc32c": all IGNORED.
+     *      These information for each slice can be looked up in their own
+     *      BundleEntryProto, keyed by each "slice_name".
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorSliceProto slices = 7;</code>
+     */
+    public Builder setSlices(
+        int index, org.tensorflow.proto.framework.TensorSliceProto.Builder builderForValue) {
+      if (slicesBuilder_ == null) {
+        ensureSlicesIsMutable();
+        slices_.set(index, builderForValue.build());
+        onChanged();
+      } else {
+        slicesBuilder_.setMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Iff present, this entry represents a partitioned tensor.  The previous
+     * fields are interpreted as follows:
+     *   "dtype", "shape": describe the full tensor.
+     *   "shard_id", "offset", "size", "crc32c": all IGNORED.
+     *      These information for each slice can be looked up in their own
+     *      BundleEntryProto, keyed by each "slice_name".
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorSliceProto slices = 7;</code>
+     */
+    public Builder addSlices(org.tensorflow.proto.framework.TensorSliceProto value) {
+      if (slicesBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureSlicesIsMutable();
+        slices_.add(value);
+        onChanged();
+      } else {
+        slicesBuilder_.addMessage(value);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Iff present, this entry represents a partitioned tensor.  The previous
+     * fields are interpreted as follows:
+     *   "dtype", "shape": describe the full tensor.
+     *   "shard_id", "offset", "size", "crc32c": all IGNORED.
+     *      These information for each slice can be looked up in their own
+     *      BundleEntryProto, keyed by each "slice_name".
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorSliceProto slices = 7;</code>
+     */
+    public Builder addSlices(
+        int index, org.tensorflow.proto.framework.TensorSliceProto value) {
+      if (slicesBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureSlicesIsMutable();
+        slices_.add(index, value);
+        onChanged();
+      } else {
+        slicesBuilder_.addMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Iff present, this entry represents a partitioned tensor.  The previous
+     * fields are interpreted as follows:
+     *   "dtype", "shape": describe the full tensor.
+     *   "shard_id", "offset", "size", "crc32c": all IGNORED.
+     *      These information for each slice can be looked up in their own
+     *      BundleEntryProto, keyed by each "slice_name".
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorSliceProto slices = 7;</code>
+     */
+    public Builder addSlices(
+        org.tensorflow.proto.framework.TensorSliceProto.Builder builderForValue) {
+      if (slicesBuilder_ == null) {
+        ensureSlicesIsMutable();
+        slices_.add(builderForValue.build());
+        onChanged();
+      } else {
+        slicesBuilder_.addMessage(builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Iff present, this entry represents a partitioned tensor.  The previous
+     * fields are interpreted as follows:
+     *   "dtype", "shape": describe the full tensor.
+     *   "shard_id", "offset", "size", "crc32c": all IGNORED.
+     *      These information for each slice can be looked up in their own
+     *      BundleEntryProto, keyed by each "slice_name".
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorSliceProto slices = 7;</code>
+     */
+    public Builder addSlices(
+        int index, org.tensorflow.proto.framework.TensorSliceProto.Builder builderForValue) {
+      if (slicesBuilder_ == null) {
+        ensureSlicesIsMutable();
+        slices_.add(index, builderForValue.build());
+        onChanged();
+      } else {
+        slicesBuilder_.addMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Iff present, this entry represents a partitioned tensor.  The previous
+     * fields are interpreted as follows:
+     *   "dtype", "shape": describe the full tensor.
+     *   "shard_id", "offset", "size", "crc32c": all IGNORED.
+     *      These information for each slice can be looked up in their own
+     *      BundleEntryProto, keyed by each "slice_name".
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorSliceProto slices = 7;</code>
+     */
+    public Builder addAllSlices(
+        java.lang.Iterable<? extends org.tensorflow.proto.framework.TensorSliceProto> values) {
+      if (slicesBuilder_ == null) {
+        ensureSlicesIsMutable();
+        com.google.protobuf.AbstractMessageLite.Builder.addAll(
+            values, slices_);
+        onChanged();
+      } else {
+        slicesBuilder_.addAllMessages(values);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Iff present, this entry represents a partitioned tensor.  The previous
+     * fields are interpreted as follows:
+     *   "dtype", "shape": describe the full tensor.
+     *   "shard_id", "offset", "size", "crc32c": all IGNORED.
+     *      These information for each slice can be looked up in their own
+     *      BundleEntryProto, keyed by each "slice_name".
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorSliceProto slices = 7;</code>
+     */
+    public Builder clearSlices() {
+      if (slicesBuilder_ == null) {
+        slices_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000001);
+        onChanged();
+      } else {
+        slicesBuilder_.clear();
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Iff present, this entry represents a partitioned tensor.  The previous
+     * fields are interpreted as follows:
+     *   "dtype", "shape": describe the full tensor.
+     *   "shard_id", "offset", "size", "crc32c": all IGNORED.
+     *      These information for each slice can be looked up in their own
+     *      BundleEntryProto, keyed by each "slice_name".
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorSliceProto slices = 7;</code>
+     */
+    public Builder removeSlices(int index) {
+      if (slicesBuilder_ == null) {
+        ensureSlicesIsMutable();
+        slices_.remove(index);
+        onChanged();
+      } else {
+        slicesBuilder_.remove(index);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Iff present, this entry represents a partitioned tensor.  The previous
+     * fields are interpreted as follows:
+     *   "dtype", "shape": describe the full tensor.
+     *   "shard_id", "offset", "size", "crc32c": all IGNORED.
+     *      These information for each slice can be looked up in their own
+     *      BundleEntryProto, keyed by each "slice_name".
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorSliceProto slices = 7;</code>
+     */
+    public org.tensorflow.proto.framework.TensorSliceProto.Builder getSlicesBuilder(
+        int index) {
+      return getSlicesFieldBuilder().getBuilder(index);
+    }
+    /**
+     * <pre>
+     * Iff present, this entry represents a partitioned tensor.  The previous
+     * fields are interpreted as follows:
+     *   "dtype", "shape": describe the full tensor.
+     *   "shard_id", "offset", "size", "crc32c": all IGNORED.
+     *      These information for each slice can be looked up in their own
+     *      BundleEntryProto, keyed by each "slice_name".
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorSliceProto slices = 7;</code>
+     */
+    public org.tensorflow.proto.framework.TensorSliceProtoOrBuilder getSlicesOrBuilder(
+        int index) {
+      if (slicesBuilder_ == null) {
+        return slices_.get(index);  } else {
+        return slicesBuilder_.getMessageOrBuilder(index);
+      }
+    }
+    /**
+     * <pre>
+     * Iff present, this entry represents a partitioned tensor.  The previous
+     * fields are interpreted as follows:
+     *   "dtype", "shape": describe the full tensor.
+     *   "shard_id", "offset", "size", "crc32c": all IGNORED.
+     *      These information for each slice can be looked up in their own
+     *      BundleEntryProto, keyed by each "slice_name".
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorSliceProto slices = 7;</code>
+     */
+    public java.util.List<? extends org.tensorflow.proto.framework.TensorSliceProtoOrBuilder> 
+         getSlicesOrBuilderList() {
+      if (slicesBuilder_ != null) {
+        return slicesBuilder_.getMessageOrBuilderList();
+      } else {
+        return java.util.Collections.unmodifiableList(slices_);
+      }
+    }
+    /**
+     * <pre>
+     * Iff present, this entry represents a partitioned tensor.  The previous
+     * fields are interpreted as follows:
+     *   "dtype", "shape": describe the full tensor.
+     *   "shard_id", "offset", "size", "crc32c": all IGNORED.
+     *      These information for each slice can be looked up in their own
+     *      BundleEntryProto, keyed by each "slice_name".
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorSliceProto slices = 7;</code>
+     */
+    public org.tensorflow.proto.framework.TensorSliceProto.Builder addSlicesBuilder() {
+      return getSlicesFieldBuilder().addBuilder(
+          org.tensorflow.proto.framework.TensorSliceProto.getDefaultInstance());
+    }
+    /**
+     * <pre>
+     * Iff present, this entry represents a partitioned tensor.  The previous
+     * fields are interpreted as follows:
+     *   "dtype", "shape": describe the full tensor.
+     *   "shard_id", "offset", "size", "crc32c": all IGNORED.
+     *      These information for each slice can be looked up in their own
+     *      BundleEntryProto, keyed by each "slice_name".
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorSliceProto slices = 7;</code>
+     */
+    public org.tensorflow.proto.framework.TensorSliceProto.Builder addSlicesBuilder(
+        int index) {
+      return getSlicesFieldBuilder().addBuilder(
+          index, org.tensorflow.proto.framework.TensorSliceProto.getDefaultInstance());
+    }
+    /**
+     * <pre>
+     * Iff present, this entry represents a partitioned tensor.  The previous
+     * fields are interpreted as follows:
+     *   "dtype", "shape": describe the full tensor.
+     *   "shard_id", "offset", "size", "crc32c": all IGNORED.
+     *      These information for each slice can be looked up in their own
+     *      BundleEntryProto, keyed by each "slice_name".
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorSliceProto slices = 7;</code>
+     */
+    public java.util.List<org.tensorflow.proto.framework.TensorSliceProto.Builder> 
+         getSlicesBuilderList() {
+      return getSlicesFieldBuilder().getBuilderList();
+    }
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.framework.TensorSliceProto, org.tensorflow.proto.framework.TensorSliceProto.Builder, org.tensorflow.proto.framework.TensorSliceProtoOrBuilder> 
+        getSlicesFieldBuilder() {
+      if (slicesBuilder_ == null) {
+        slicesBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
+            org.tensorflow.proto.framework.TensorSliceProto, org.tensorflow.proto.framework.TensorSliceProto.Builder, org.tensorflow.proto.framework.TensorSliceProtoOrBuilder>(
+                slices_,
+                ((bitField0_ & 0x00000001) != 0),
+                getParentForChildren(),
+                isClean());
+        slices_ = null;
+      }
+      return slicesBuilder_;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.BundleEntryProto)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.BundleEntryProto)
+  private static final org.tensorflow.proto.util.BundleEntryProto DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.util.BundleEntryProto();
+  }
+
+  public static org.tensorflow.proto.util.BundleEntryProto getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<BundleEntryProto>
+      PARSER = new com.google.protobuf.AbstractParser<BundleEntryProto>() {
+    @java.lang.Override
+    public BundleEntryProto parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new BundleEntryProto(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<BundleEntryProto> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<BundleEntryProto> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.util.BundleEntryProto getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/BundleEntryProtoOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/BundleEntryProtoOrBuilder.java
new file mode 100644
index 00000000000..a95ca5d54e6
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/BundleEntryProtoOrBuilder.java
@@ -0,0 +1,137 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/tensor_bundle.proto
+
+package org.tensorflow.proto.util;
+
+public interface BundleEntryProtoOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.BundleEntryProto)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <pre>
+   * The tensor dtype and shape.
+   * </pre>
+   *
+   * <code>.tensorflow.DataType dtype = 1;</code>
+   */
+  int getDtypeValue();
+  /**
+   * <pre>
+   * The tensor dtype and shape.
+   * </pre>
+   *
+   * <code>.tensorflow.DataType dtype = 1;</code>
+   */
+  org.tensorflow.proto.framework.DataType getDtype();
+
+  /**
+   * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+   */
+  boolean hasShape();
+  /**
+   * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+   */
+  org.tensorflow.proto.framework.TensorShapeProto getShape();
+  /**
+   * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+   */
+  org.tensorflow.proto.framework.TensorShapeProtoOrBuilder getShapeOrBuilder();
+
+  /**
+   * <pre>
+   * The binary content of the tensor lies in:
+   *   File "shard_id": bytes [offset, offset + size).
+   * </pre>
+   *
+   * <code>int32 shard_id = 3;</code>
+   */
+  int getShardId();
+
+  /**
+   * <code>int64 offset = 4;</code>
+   */
+  long getOffset();
+
+  /**
+   * <code>int64 size = 5;</code>
+   */
+  long getSize();
+
+  /**
+   * <pre>
+   * The CRC32C checksum of the tensor bytes.
+   * </pre>
+   *
+   * <code>fixed32 crc32c = 6;</code>
+   */
+  int getCrc32C();
+
+  /**
+   * <pre>
+   * Iff present, this entry represents a partitioned tensor.  The previous
+   * fields are interpreted as follows:
+   *   "dtype", "shape": describe the full tensor.
+   *   "shard_id", "offset", "size", "crc32c": all IGNORED.
+   *      These information for each slice can be looked up in their own
+   *      BundleEntryProto, keyed by each "slice_name".
+   * </pre>
+   *
+   * <code>repeated .tensorflow.TensorSliceProto slices = 7;</code>
+   */
+  java.util.List<org.tensorflow.proto.framework.TensorSliceProto> 
+      getSlicesList();
+  /**
+   * <pre>
+   * Iff present, this entry represents a partitioned tensor.  The previous
+   * fields are interpreted as follows:
+   *   "dtype", "shape": describe the full tensor.
+   *   "shard_id", "offset", "size", "crc32c": all IGNORED.
+   *      These information for each slice can be looked up in their own
+   *      BundleEntryProto, keyed by each "slice_name".
+   * </pre>
+   *
+   * <code>repeated .tensorflow.TensorSliceProto slices = 7;</code>
+   */
+  org.tensorflow.proto.framework.TensorSliceProto getSlices(int index);
+  /**
+   * <pre>
+   * Iff present, this entry represents a partitioned tensor.  The previous
+   * fields are interpreted as follows:
+   *   "dtype", "shape": describe the full tensor.
+   *   "shard_id", "offset", "size", "crc32c": all IGNORED.
+   *      These information for each slice can be looked up in their own
+   *      BundleEntryProto, keyed by each "slice_name".
+   * </pre>
+   *
+   * <code>repeated .tensorflow.TensorSliceProto slices = 7;</code>
+   */
+  int getSlicesCount();
+  /**
+   * <pre>
+   * Iff present, this entry represents a partitioned tensor.  The previous
+   * fields are interpreted as follows:
+   *   "dtype", "shape": describe the full tensor.
+   *   "shard_id", "offset", "size", "crc32c": all IGNORED.
+   *      These information for each slice can be looked up in their own
+   *      BundleEntryProto, keyed by each "slice_name".
+   * </pre>
+   *
+   * <code>repeated .tensorflow.TensorSliceProto slices = 7;</code>
+   */
+  java.util.List<? extends org.tensorflow.proto.framework.TensorSliceProtoOrBuilder> 
+      getSlicesOrBuilderList();
+  /**
+   * <pre>
+   * Iff present, this entry represents a partitioned tensor.  The previous
+   * fields are interpreted as follows:
+   *   "dtype", "shape": describe the full tensor.
+   *   "shard_id", "offset", "size", "crc32c": all IGNORED.
+   *      These information for each slice can be looked up in their own
+   *      BundleEntryProto, keyed by each "slice_name".
+   * </pre>
+   *
+   * <code>repeated .tensorflow.TensorSliceProto slices = 7;</code>
+   */
+  org.tensorflow.proto.framework.TensorSliceProtoOrBuilder getSlicesOrBuilder(
+      int index);
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/BundleHeaderProto.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/BundleHeaderProto.java
new file mode 100644
index 00000000000..1a30188c8a5
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/BundleHeaderProto.java
@@ -0,0 +1,929 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/tensor_bundle.proto
+
+package org.tensorflow.proto.util;
+
+/**
+ * <pre>
+ * Special header that is associated with a bundle.
+ * TODO(zongheng,zhifengc): maybe in the future, we can add information about
+ * which binary produced this checkpoint, timestamp, etc. Sometime, these can be
+ * valuable debugging information. And if needed, these can be used as defensive
+ * information ensuring reader (binary version) of the checkpoint and the writer
+ * (binary version) must match within certain range, etc.
+ * </pre>
+ *
+ * Protobuf type {@code tensorflow.BundleHeaderProto}
+ */
+public  final class BundleHeaderProto extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.BundleHeaderProto)
+    BundleHeaderProtoOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use BundleHeaderProto.newBuilder() to construct.
+  private BundleHeaderProto(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private BundleHeaderProto() {
+    endianness_ = 0;
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new BundleHeaderProto();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private BundleHeaderProto(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 8: {
+
+            numShards_ = input.readInt32();
+            break;
+          }
+          case 16: {
+            int rawValue = input.readEnum();
+
+            endianness_ = rawValue;
+            break;
+          }
+          case 26: {
+            org.tensorflow.proto.framework.VersionDef.Builder subBuilder = null;
+            if (version_ != null) {
+              subBuilder = version_.toBuilder();
+            }
+            version_ = input.readMessage(org.tensorflow.proto.framework.VersionDef.parser(), extensionRegistry);
+            if (subBuilder != null) {
+              subBuilder.mergeFrom(version_);
+              version_ = subBuilder.buildPartial();
+            }
+
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.util.TensorBundleProtos.internal_static_tensorflow_BundleHeaderProto_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.util.TensorBundleProtos.internal_static_tensorflow_BundleHeaderProto_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.util.BundleHeaderProto.class, org.tensorflow.proto.util.BundleHeaderProto.Builder.class);
+  }
+
+  /**
+   * <pre>
+   * An enum indicating the endianness of the platform that produced this
+   * bundle.  A bundle can only be read by a platform with matching endianness.
+   * Defaults to LITTLE, as most modern platforms are little-endian.
+   * Affects the binary tensor data bytes only, not the metadata in protobufs.
+   * </pre>
+   *
+   * Protobuf enum {@code tensorflow.BundleHeaderProto.Endianness}
+   */
+  public enum Endianness
+      implements com.google.protobuf.ProtocolMessageEnum {
+    /**
+     * <code>LITTLE = 0;</code>
+     */
+    LITTLE(0),
+    /**
+     * <code>BIG = 1;</code>
+     */
+    BIG(1),
+    UNRECOGNIZED(-1),
+    ;
+
+    /**
+     * <code>LITTLE = 0;</code>
+     */
+    public static final int LITTLE_VALUE = 0;
+    /**
+     * <code>BIG = 1;</code>
+     */
+    public static final int BIG_VALUE = 1;
+
+
+    public final int getNumber() {
+      if (this == UNRECOGNIZED) {
+        throw new java.lang.IllegalArgumentException(
+            "Can't get the number of an unknown enum value.");
+      }
+      return value;
+    }
+
+    /**
+     * @deprecated Use {@link #forNumber(int)} instead.
+     */
+    @java.lang.Deprecated
+    public static Endianness valueOf(int value) {
+      return forNumber(value);
+    }
+
+    public static Endianness forNumber(int value) {
+      switch (value) {
+        case 0: return LITTLE;
+        case 1: return BIG;
+        default: return null;
+      }
+    }
+
+    public static com.google.protobuf.Internal.EnumLiteMap<Endianness>
+        internalGetValueMap() {
+      return internalValueMap;
+    }
+    private static final com.google.protobuf.Internal.EnumLiteMap<
+        Endianness> internalValueMap =
+          new com.google.protobuf.Internal.EnumLiteMap<Endianness>() {
+            public Endianness findValueByNumber(int number) {
+              return Endianness.forNumber(number);
+            }
+          };
+
+    public final com.google.protobuf.Descriptors.EnumValueDescriptor
+        getValueDescriptor() {
+      return getDescriptor().getValues().get(ordinal());
+    }
+    public final com.google.protobuf.Descriptors.EnumDescriptor
+        getDescriptorForType() {
+      return getDescriptor();
+    }
+    public static final com.google.protobuf.Descriptors.EnumDescriptor
+        getDescriptor() {
+      return org.tensorflow.proto.util.BundleHeaderProto.getDescriptor().getEnumTypes().get(0);
+    }
+
+    private static final Endianness[] VALUES = values();
+
+    public static Endianness valueOf(
+        com.google.protobuf.Descriptors.EnumValueDescriptor desc) {
+      if (desc.getType() != getDescriptor()) {
+        throw new java.lang.IllegalArgumentException(
+          "EnumValueDescriptor is not for this type.");
+      }
+      if (desc.getIndex() == -1) {
+        return UNRECOGNIZED;
+      }
+      return VALUES[desc.getIndex()];
+    }
+
+    private final int value;
+
+    private Endianness(int value) {
+      this.value = value;
+    }
+
+    // @@protoc_insertion_point(enum_scope:tensorflow.BundleHeaderProto.Endianness)
+  }
+
+  public static final int NUM_SHARDS_FIELD_NUMBER = 1;
+  private int numShards_;
+  /**
+   * <pre>
+   * Number of data files in the bundle.
+   * </pre>
+   *
+   * <code>int32 num_shards = 1;</code>
+   */
+  public int getNumShards() {
+    return numShards_;
+  }
+
+  public static final int ENDIANNESS_FIELD_NUMBER = 2;
+  private int endianness_;
+  /**
+   * <code>.tensorflow.BundleHeaderProto.Endianness endianness = 2;</code>
+   */
+  public int getEndiannessValue() {
+    return endianness_;
+  }
+  /**
+   * <code>.tensorflow.BundleHeaderProto.Endianness endianness = 2;</code>
+   */
+  public org.tensorflow.proto.util.BundleHeaderProto.Endianness getEndianness() {
+    @SuppressWarnings("deprecation")
+    org.tensorflow.proto.util.BundleHeaderProto.Endianness result = org.tensorflow.proto.util.BundleHeaderProto.Endianness.valueOf(endianness_);
+    return result == null ? org.tensorflow.proto.util.BundleHeaderProto.Endianness.UNRECOGNIZED : result;
+  }
+
+  public static final int VERSION_FIELD_NUMBER = 3;
+  private org.tensorflow.proto.framework.VersionDef version_;
+  /**
+   * <pre>
+   * Versioning of the tensor bundle format.
+   * </pre>
+   *
+   * <code>.tensorflow.VersionDef version = 3;</code>
+   */
+  public boolean hasVersion() {
+    return version_ != null;
+  }
+  /**
+   * <pre>
+   * Versioning of the tensor bundle format.
+   * </pre>
+   *
+   * <code>.tensorflow.VersionDef version = 3;</code>
+   */
+  public org.tensorflow.proto.framework.VersionDef getVersion() {
+    return version_ == null ? org.tensorflow.proto.framework.VersionDef.getDefaultInstance() : version_;
+  }
+  /**
+   * <pre>
+   * Versioning of the tensor bundle format.
+   * </pre>
+   *
+   * <code>.tensorflow.VersionDef version = 3;</code>
+   */
+  public org.tensorflow.proto.framework.VersionDefOrBuilder getVersionOrBuilder() {
+    return getVersion();
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    if (numShards_ != 0) {
+      output.writeInt32(1, numShards_);
+    }
+    if (endianness_ != org.tensorflow.proto.util.BundleHeaderProto.Endianness.LITTLE.getNumber()) {
+      output.writeEnum(2, endianness_);
+    }
+    if (version_ != null) {
+      output.writeMessage(3, getVersion());
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (numShards_ != 0) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt32Size(1, numShards_);
+    }
+    if (endianness_ != org.tensorflow.proto.util.BundleHeaderProto.Endianness.LITTLE.getNumber()) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeEnumSize(2, endianness_);
+    }
+    if (version_ != null) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(3, getVersion());
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.util.BundleHeaderProto)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.util.BundleHeaderProto other = (org.tensorflow.proto.util.BundleHeaderProto) obj;
+
+    if (getNumShards()
+        != other.getNumShards()) return false;
+    if (endianness_ != other.endianness_) return false;
+    if (hasVersion() != other.hasVersion()) return false;
+    if (hasVersion()) {
+      if (!getVersion()
+          .equals(other.getVersion())) return false;
+    }
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    hash = (37 * hash) + NUM_SHARDS_FIELD_NUMBER;
+    hash = (53 * hash) + getNumShards();
+    hash = (37 * hash) + ENDIANNESS_FIELD_NUMBER;
+    hash = (53 * hash) + endianness_;
+    if (hasVersion()) {
+      hash = (37 * hash) + VERSION_FIELD_NUMBER;
+      hash = (53 * hash) + getVersion().hashCode();
+    }
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.util.BundleHeaderProto parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.BundleHeaderProto parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.BundleHeaderProto parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.BundleHeaderProto parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.BundleHeaderProto parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.BundleHeaderProto parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.BundleHeaderProto parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.BundleHeaderProto parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.BundleHeaderProto parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.BundleHeaderProto parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.BundleHeaderProto parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.BundleHeaderProto parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.util.BundleHeaderProto prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * <pre>
+   * Special header that is associated with a bundle.
+   * TODO(zongheng,zhifengc): maybe in the future, we can add information about
+   * which binary produced this checkpoint, timestamp, etc. Sometime, these can be
+   * valuable debugging information. And if needed, these can be used as defensive
+   * information ensuring reader (binary version) of the checkpoint and the writer
+   * (binary version) must match within certain range, etc.
+   * </pre>
+   *
+   * Protobuf type {@code tensorflow.BundleHeaderProto}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.BundleHeaderProto)
+      org.tensorflow.proto.util.BundleHeaderProtoOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.util.TensorBundleProtos.internal_static_tensorflow_BundleHeaderProto_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.util.TensorBundleProtos.internal_static_tensorflow_BundleHeaderProto_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.util.BundleHeaderProto.class, org.tensorflow.proto.util.BundleHeaderProto.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.util.BundleHeaderProto.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      numShards_ = 0;
+
+      endianness_ = 0;
+
+      if (versionBuilder_ == null) {
+        version_ = null;
+      } else {
+        version_ = null;
+        versionBuilder_ = null;
+      }
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.util.TensorBundleProtos.internal_static_tensorflow_BundleHeaderProto_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.BundleHeaderProto getDefaultInstanceForType() {
+      return org.tensorflow.proto.util.BundleHeaderProto.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.BundleHeaderProto build() {
+      org.tensorflow.proto.util.BundleHeaderProto result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.BundleHeaderProto buildPartial() {
+      org.tensorflow.proto.util.BundleHeaderProto result = new org.tensorflow.proto.util.BundleHeaderProto(this);
+      result.numShards_ = numShards_;
+      result.endianness_ = endianness_;
+      if (versionBuilder_ == null) {
+        result.version_ = version_;
+      } else {
+        result.version_ = versionBuilder_.build();
+      }
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.util.BundleHeaderProto) {
+        return mergeFrom((org.tensorflow.proto.util.BundleHeaderProto)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.util.BundleHeaderProto other) {
+      if (other == org.tensorflow.proto.util.BundleHeaderProto.getDefaultInstance()) return this;
+      if (other.getNumShards() != 0) {
+        setNumShards(other.getNumShards());
+      }
+      if (other.endianness_ != 0) {
+        setEndiannessValue(other.getEndiannessValue());
+      }
+      if (other.hasVersion()) {
+        mergeVersion(other.getVersion());
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.util.BundleHeaderProto parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.util.BundleHeaderProto) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+
+    private int numShards_ ;
+    /**
+     * <pre>
+     * Number of data files in the bundle.
+     * </pre>
+     *
+     * <code>int32 num_shards = 1;</code>
+     */
+    public int getNumShards() {
+      return numShards_;
+    }
+    /**
+     * <pre>
+     * Number of data files in the bundle.
+     * </pre>
+     *
+     * <code>int32 num_shards = 1;</code>
+     */
+    public Builder setNumShards(int value) {
+      
+      numShards_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Number of data files in the bundle.
+     * </pre>
+     *
+     * <code>int32 num_shards = 1;</code>
+     */
+    public Builder clearNumShards() {
+      
+      numShards_ = 0;
+      onChanged();
+      return this;
+    }
+
+    private int endianness_ = 0;
+    /**
+     * <code>.tensorflow.BundleHeaderProto.Endianness endianness = 2;</code>
+     */
+    public int getEndiannessValue() {
+      return endianness_;
+    }
+    /**
+     * <code>.tensorflow.BundleHeaderProto.Endianness endianness = 2;</code>
+     */
+    public Builder setEndiannessValue(int value) {
+      endianness_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>.tensorflow.BundleHeaderProto.Endianness endianness = 2;</code>
+     */
+    public org.tensorflow.proto.util.BundleHeaderProto.Endianness getEndianness() {
+      @SuppressWarnings("deprecation")
+      org.tensorflow.proto.util.BundleHeaderProto.Endianness result = org.tensorflow.proto.util.BundleHeaderProto.Endianness.valueOf(endianness_);
+      return result == null ? org.tensorflow.proto.util.BundleHeaderProto.Endianness.UNRECOGNIZED : result;
+    }
+    /**
+     * <code>.tensorflow.BundleHeaderProto.Endianness endianness = 2;</code>
+     */
+    public Builder setEndianness(org.tensorflow.proto.util.BundleHeaderProto.Endianness value) {
+      if (value == null) {
+        throw new NullPointerException();
+      }
+      
+      endianness_ = value.getNumber();
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>.tensorflow.BundleHeaderProto.Endianness endianness = 2;</code>
+     */
+    public Builder clearEndianness() {
+      
+      endianness_ = 0;
+      onChanged();
+      return this;
+    }
+
+    private org.tensorflow.proto.framework.VersionDef version_;
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.VersionDef, org.tensorflow.proto.framework.VersionDef.Builder, org.tensorflow.proto.framework.VersionDefOrBuilder> versionBuilder_;
+    /**
+     * <pre>
+     * Versioning of the tensor bundle format.
+     * </pre>
+     *
+     * <code>.tensorflow.VersionDef version = 3;</code>
+     */
+    public boolean hasVersion() {
+      return versionBuilder_ != null || version_ != null;
+    }
+    /**
+     * <pre>
+     * Versioning of the tensor bundle format.
+     * </pre>
+     *
+     * <code>.tensorflow.VersionDef version = 3;</code>
+     */
+    public org.tensorflow.proto.framework.VersionDef getVersion() {
+      if (versionBuilder_ == null) {
+        return version_ == null ? org.tensorflow.proto.framework.VersionDef.getDefaultInstance() : version_;
+      } else {
+        return versionBuilder_.getMessage();
+      }
+    }
+    /**
+     * <pre>
+     * Versioning of the tensor bundle format.
+     * </pre>
+     *
+     * <code>.tensorflow.VersionDef version = 3;</code>
+     */
+    public Builder setVersion(org.tensorflow.proto.framework.VersionDef value) {
+      if (versionBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        version_ = value;
+        onChanged();
+      } else {
+        versionBuilder_.setMessage(value);
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Versioning of the tensor bundle format.
+     * </pre>
+     *
+     * <code>.tensorflow.VersionDef version = 3;</code>
+     */
+    public Builder setVersion(
+        org.tensorflow.proto.framework.VersionDef.Builder builderForValue) {
+      if (versionBuilder_ == null) {
+        version_ = builderForValue.build();
+        onChanged();
+      } else {
+        versionBuilder_.setMessage(builderForValue.build());
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Versioning of the tensor bundle format.
+     * </pre>
+     *
+     * <code>.tensorflow.VersionDef version = 3;</code>
+     */
+    public Builder mergeVersion(org.tensorflow.proto.framework.VersionDef value) {
+      if (versionBuilder_ == null) {
+        if (version_ != null) {
+          version_ =
+            org.tensorflow.proto.framework.VersionDef.newBuilder(version_).mergeFrom(value).buildPartial();
+        } else {
+          version_ = value;
+        }
+        onChanged();
+      } else {
+        versionBuilder_.mergeFrom(value);
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Versioning of the tensor bundle format.
+     * </pre>
+     *
+     * <code>.tensorflow.VersionDef version = 3;</code>
+     */
+    public Builder clearVersion() {
+      if (versionBuilder_ == null) {
+        version_ = null;
+        onChanged();
+      } else {
+        version_ = null;
+        versionBuilder_ = null;
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Versioning of the tensor bundle format.
+     * </pre>
+     *
+     * <code>.tensorflow.VersionDef version = 3;</code>
+     */
+    public org.tensorflow.proto.framework.VersionDef.Builder getVersionBuilder() {
+      
+      onChanged();
+      return getVersionFieldBuilder().getBuilder();
+    }
+    /**
+     * <pre>
+     * Versioning of the tensor bundle format.
+     * </pre>
+     *
+     * <code>.tensorflow.VersionDef version = 3;</code>
+     */
+    public org.tensorflow.proto.framework.VersionDefOrBuilder getVersionOrBuilder() {
+      if (versionBuilder_ != null) {
+        return versionBuilder_.getMessageOrBuilder();
+      } else {
+        return version_ == null ?
+            org.tensorflow.proto.framework.VersionDef.getDefaultInstance() : version_;
+      }
+    }
+    /**
+     * <pre>
+     * Versioning of the tensor bundle format.
+     * </pre>
+     *
+     * <code>.tensorflow.VersionDef version = 3;</code>
+     */
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.VersionDef, org.tensorflow.proto.framework.VersionDef.Builder, org.tensorflow.proto.framework.VersionDefOrBuilder> 
+        getVersionFieldBuilder() {
+      if (versionBuilder_ == null) {
+        versionBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+            org.tensorflow.proto.framework.VersionDef, org.tensorflow.proto.framework.VersionDef.Builder, org.tensorflow.proto.framework.VersionDefOrBuilder>(
+                getVersion(),
+                getParentForChildren(),
+                isClean());
+        version_ = null;
+      }
+      return versionBuilder_;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.BundleHeaderProto)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.BundleHeaderProto)
+  private static final org.tensorflow.proto.util.BundleHeaderProto DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.util.BundleHeaderProto();
+  }
+
+  public static org.tensorflow.proto.util.BundleHeaderProto getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<BundleHeaderProto>
+      PARSER = new com.google.protobuf.AbstractParser<BundleHeaderProto>() {
+    @java.lang.Override
+    public BundleHeaderProto parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new BundleHeaderProto(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<BundleHeaderProto> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<BundleHeaderProto> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.util.BundleHeaderProto getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/BundleHeaderProtoOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/BundleHeaderProtoOrBuilder.java
new file mode 100644
index 00000000000..f80c5fcef1c
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/BundleHeaderProtoOrBuilder.java
@@ -0,0 +1,52 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/tensor_bundle.proto
+
+package org.tensorflow.proto.util;
+
+public interface BundleHeaderProtoOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.BundleHeaderProto)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <pre>
+   * Number of data files in the bundle.
+   * </pre>
+   *
+   * <code>int32 num_shards = 1;</code>
+   */
+  int getNumShards();
+
+  /**
+   * <code>.tensorflow.BundleHeaderProto.Endianness endianness = 2;</code>
+   */
+  int getEndiannessValue();
+  /**
+   * <code>.tensorflow.BundleHeaderProto.Endianness endianness = 2;</code>
+   */
+  org.tensorflow.proto.util.BundleHeaderProto.Endianness getEndianness();
+
+  /**
+   * <pre>
+   * Versioning of the tensor bundle format.
+   * </pre>
+   *
+   * <code>.tensorflow.VersionDef version = 3;</code>
+   */
+  boolean hasVersion();
+  /**
+   * <pre>
+   * Versioning of the tensor bundle format.
+   * </pre>
+   *
+   * <code>.tensorflow.VersionDef version = 3;</code>
+   */
+  org.tensorflow.proto.framework.VersionDef getVersion();
+  /**
+   * <pre>
+   * Versioning of the tensor bundle format.
+   * </pre>
+   *
+   * <code>.tensorflow.VersionDef version = 3;</code>
+   */
+  org.tensorflow.proto.framework.VersionDefOrBuilder getVersionOrBuilder();
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/CodeLocation.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/CodeLocation.java
new file mode 100644
index 00000000000..28d856a52f3
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/CodeLocation.java
@@ -0,0 +1,837 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/debug_event.proto
+
+package org.tensorflow.proto.util;
+
+/**
+ * <pre>
+ * Code location information: A stack trace with host-name information.
+ * Instead of encoding the detailed stack trace, this proto refers to IDs of
+ * stack frames stored as `StackFrameWithId` protos.
+ * </pre>
+ *
+ * Protobuf type {@code tensorflow.CodeLocation}
+ */
+public  final class CodeLocation extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.CodeLocation)
+    CodeLocationOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use CodeLocation.newBuilder() to construct.
+  private CodeLocation(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private CodeLocation() {
+    hostName_ = "";
+    stackFrameIds_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new CodeLocation();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private CodeLocation(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    int mutable_bitField0_ = 0;
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 10: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            hostName_ = s;
+            break;
+          }
+          case 18: {
+            java.lang.String s = input.readStringRequireUtf8();
+            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
+              stackFrameIds_ = new com.google.protobuf.LazyStringArrayList();
+              mutable_bitField0_ |= 0x00000001;
+            }
+            stackFrameIds_.add(s);
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      if (((mutable_bitField0_ & 0x00000001) != 0)) {
+        stackFrameIds_ = stackFrameIds_.getUnmodifiableView();
+      }
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.util.DebugEventProtos.internal_static_tensorflow_CodeLocation_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.util.DebugEventProtos.internal_static_tensorflow_CodeLocation_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.util.CodeLocation.class, org.tensorflow.proto.util.CodeLocation.Builder.class);
+  }
+
+  public static final int HOST_NAME_FIELD_NUMBER = 1;
+  private volatile java.lang.Object hostName_;
+  /**
+   * <pre>
+   * Host name on which the source files are located.
+   * </pre>
+   *
+   * <code>string host_name = 1;</code>
+   */
+  public java.lang.String getHostName() {
+    java.lang.Object ref = hostName_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      hostName_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * Host name on which the source files are located.
+   * </pre>
+   *
+   * <code>string host_name = 1;</code>
+   */
+  public com.google.protobuf.ByteString
+      getHostNameBytes() {
+    java.lang.Object ref = hostName_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      hostName_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int STACK_FRAME_IDS_FIELD_NUMBER = 2;
+  private com.google.protobuf.LazyStringList stackFrameIds_;
+  /**
+   * <pre>
+   * ID to a stack frame, each of which is pointed to
+   * by a unique ID. The ordering of the frames is consistent with Python's
+   * `traceback.extract_tb()`.
+   * </pre>
+   *
+   * <code>repeated string stack_frame_ids = 2;</code>
+   */
+  public com.google.protobuf.ProtocolStringList
+      getStackFrameIdsList() {
+    return stackFrameIds_;
+  }
+  /**
+   * <pre>
+   * ID to a stack frame, each of which is pointed to
+   * by a unique ID. The ordering of the frames is consistent with Python's
+   * `traceback.extract_tb()`.
+   * </pre>
+   *
+   * <code>repeated string stack_frame_ids = 2;</code>
+   */
+  public int getStackFrameIdsCount() {
+    return stackFrameIds_.size();
+  }
+  /**
+   * <pre>
+   * ID to a stack frame, each of which is pointed to
+   * by a unique ID. The ordering of the frames is consistent with Python's
+   * `traceback.extract_tb()`.
+   * </pre>
+   *
+   * <code>repeated string stack_frame_ids = 2;</code>
+   */
+  public java.lang.String getStackFrameIds(int index) {
+    return stackFrameIds_.get(index);
+  }
+  /**
+   * <pre>
+   * ID to a stack frame, each of which is pointed to
+   * by a unique ID. The ordering of the frames is consistent with Python's
+   * `traceback.extract_tb()`.
+   * </pre>
+   *
+   * <code>repeated string stack_frame_ids = 2;</code>
+   */
+  public com.google.protobuf.ByteString
+      getStackFrameIdsBytes(int index) {
+    return stackFrameIds_.getByteString(index);
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    if (!getHostNameBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 1, hostName_);
+    }
+    for (int i = 0; i < stackFrameIds_.size(); i++) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 2, stackFrameIds_.getRaw(i));
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (!getHostNameBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, hostName_);
+    }
+    {
+      int dataSize = 0;
+      for (int i = 0; i < stackFrameIds_.size(); i++) {
+        dataSize += computeStringSizeNoTag(stackFrameIds_.getRaw(i));
+      }
+      size += dataSize;
+      size += 1 * getStackFrameIdsList().size();
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.util.CodeLocation)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.util.CodeLocation other = (org.tensorflow.proto.util.CodeLocation) obj;
+
+    if (!getHostName()
+        .equals(other.getHostName())) return false;
+    if (!getStackFrameIdsList()
+        .equals(other.getStackFrameIdsList())) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    hash = (37 * hash) + HOST_NAME_FIELD_NUMBER;
+    hash = (53 * hash) + getHostName().hashCode();
+    if (getStackFrameIdsCount() > 0) {
+      hash = (37 * hash) + STACK_FRAME_IDS_FIELD_NUMBER;
+      hash = (53 * hash) + getStackFrameIdsList().hashCode();
+    }
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.util.CodeLocation parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.CodeLocation parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.CodeLocation parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.CodeLocation parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.CodeLocation parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.CodeLocation parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.CodeLocation parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.CodeLocation parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.CodeLocation parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.CodeLocation parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.CodeLocation parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.CodeLocation parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.util.CodeLocation prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * <pre>
+   * Code location information: A stack trace with host-name information.
+   * Instead of encoding the detailed stack trace, this proto refers to IDs of
+   * stack frames stored as `StackFrameWithId` protos.
+   * </pre>
+   *
+   * Protobuf type {@code tensorflow.CodeLocation}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.CodeLocation)
+      org.tensorflow.proto.util.CodeLocationOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.util.DebugEventProtos.internal_static_tensorflow_CodeLocation_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.util.DebugEventProtos.internal_static_tensorflow_CodeLocation_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.util.CodeLocation.class, org.tensorflow.proto.util.CodeLocation.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.util.CodeLocation.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      hostName_ = "";
+
+      stackFrameIds_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+      bitField0_ = (bitField0_ & ~0x00000001);
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.util.DebugEventProtos.internal_static_tensorflow_CodeLocation_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.CodeLocation getDefaultInstanceForType() {
+      return org.tensorflow.proto.util.CodeLocation.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.CodeLocation build() {
+      org.tensorflow.proto.util.CodeLocation result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.CodeLocation buildPartial() {
+      org.tensorflow.proto.util.CodeLocation result = new org.tensorflow.proto.util.CodeLocation(this);
+      int from_bitField0_ = bitField0_;
+      result.hostName_ = hostName_;
+      if (((bitField0_ & 0x00000001) != 0)) {
+        stackFrameIds_ = stackFrameIds_.getUnmodifiableView();
+        bitField0_ = (bitField0_ & ~0x00000001);
+      }
+      result.stackFrameIds_ = stackFrameIds_;
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.util.CodeLocation) {
+        return mergeFrom((org.tensorflow.proto.util.CodeLocation)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.util.CodeLocation other) {
+      if (other == org.tensorflow.proto.util.CodeLocation.getDefaultInstance()) return this;
+      if (!other.getHostName().isEmpty()) {
+        hostName_ = other.hostName_;
+        onChanged();
+      }
+      if (!other.stackFrameIds_.isEmpty()) {
+        if (stackFrameIds_.isEmpty()) {
+          stackFrameIds_ = other.stackFrameIds_;
+          bitField0_ = (bitField0_ & ~0x00000001);
+        } else {
+          ensureStackFrameIdsIsMutable();
+          stackFrameIds_.addAll(other.stackFrameIds_);
+        }
+        onChanged();
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.util.CodeLocation parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.util.CodeLocation) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+    private int bitField0_;
+
+    private java.lang.Object hostName_ = "";
+    /**
+     * <pre>
+     * Host name on which the source files are located.
+     * </pre>
+     *
+     * <code>string host_name = 1;</code>
+     */
+    public java.lang.String getHostName() {
+      java.lang.Object ref = hostName_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        hostName_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Host name on which the source files are located.
+     * </pre>
+     *
+     * <code>string host_name = 1;</code>
+     */
+    public com.google.protobuf.ByteString
+        getHostNameBytes() {
+      java.lang.Object ref = hostName_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        hostName_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Host name on which the source files are located.
+     * </pre>
+     *
+     * <code>string host_name = 1;</code>
+     */
+    public Builder setHostName(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      hostName_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Host name on which the source files are located.
+     * </pre>
+     *
+     * <code>string host_name = 1;</code>
+     */
+    public Builder clearHostName() {
+      
+      hostName_ = getDefaultInstance().getHostName();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Host name on which the source files are located.
+     * </pre>
+     *
+     * <code>string host_name = 1;</code>
+     */
+    public Builder setHostNameBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      hostName_ = value;
+      onChanged();
+      return this;
+    }
+
+    private com.google.protobuf.LazyStringList stackFrameIds_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+    private void ensureStackFrameIdsIsMutable() {
+      if (!((bitField0_ & 0x00000001) != 0)) {
+        stackFrameIds_ = new com.google.protobuf.LazyStringArrayList(stackFrameIds_);
+        bitField0_ |= 0x00000001;
+       }
+    }
+    /**
+     * <pre>
+     * ID to a stack frame, each of which is pointed to
+     * by a unique ID. The ordering of the frames is consistent with Python's
+     * `traceback.extract_tb()`.
+     * </pre>
+     *
+     * <code>repeated string stack_frame_ids = 2;</code>
+     */
+    public com.google.protobuf.ProtocolStringList
+        getStackFrameIdsList() {
+      return stackFrameIds_.getUnmodifiableView();
+    }
+    /**
+     * <pre>
+     * ID to a stack frame, each of which is pointed to
+     * by a unique ID. The ordering of the frames is consistent with Python's
+     * `traceback.extract_tb()`.
+     * </pre>
+     *
+     * <code>repeated string stack_frame_ids = 2;</code>
+     */
+    public int getStackFrameIdsCount() {
+      return stackFrameIds_.size();
+    }
+    /**
+     * <pre>
+     * ID to a stack frame, each of which is pointed to
+     * by a unique ID. The ordering of the frames is consistent with Python's
+     * `traceback.extract_tb()`.
+     * </pre>
+     *
+     * <code>repeated string stack_frame_ids = 2;</code>
+     */
+    public java.lang.String getStackFrameIds(int index) {
+      return stackFrameIds_.get(index);
+    }
+    /**
+     * <pre>
+     * ID to a stack frame, each of which is pointed to
+     * by a unique ID. The ordering of the frames is consistent with Python's
+     * `traceback.extract_tb()`.
+     * </pre>
+     *
+     * <code>repeated string stack_frame_ids = 2;</code>
+     */
+    public com.google.protobuf.ByteString
+        getStackFrameIdsBytes(int index) {
+      return stackFrameIds_.getByteString(index);
+    }
+    /**
+     * <pre>
+     * ID to a stack frame, each of which is pointed to
+     * by a unique ID. The ordering of the frames is consistent with Python's
+     * `traceback.extract_tb()`.
+     * </pre>
+     *
+     * <code>repeated string stack_frame_ids = 2;</code>
+     */
+    public Builder setStackFrameIds(
+        int index, java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  ensureStackFrameIdsIsMutable();
+      stackFrameIds_.set(index, value);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * ID to a stack frame, each of which is pointed to
+     * by a unique ID. The ordering of the frames is consistent with Python's
+     * `traceback.extract_tb()`.
+     * </pre>
+     *
+     * <code>repeated string stack_frame_ids = 2;</code>
+     */
+    public Builder addStackFrameIds(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  ensureStackFrameIdsIsMutable();
+      stackFrameIds_.add(value);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * ID to a stack frame, each of which is pointed to
+     * by a unique ID. The ordering of the frames is consistent with Python's
+     * `traceback.extract_tb()`.
+     * </pre>
+     *
+     * <code>repeated string stack_frame_ids = 2;</code>
+     */
+    public Builder addAllStackFrameIds(
+        java.lang.Iterable<java.lang.String> values) {
+      ensureStackFrameIdsIsMutable();
+      com.google.protobuf.AbstractMessageLite.Builder.addAll(
+          values, stackFrameIds_);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * ID to a stack frame, each of which is pointed to
+     * by a unique ID. The ordering of the frames is consistent with Python's
+     * `traceback.extract_tb()`.
+     * </pre>
+     *
+     * <code>repeated string stack_frame_ids = 2;</code>
+     */
+    public Builder clearStackFrameIds() {
+      stackFrameIds_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+      bitField0_ = (bitField0_ & ~0x00000001);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * ID to a stack frame, each of which is pointed to
+     * by a unique ID. The ordering of the frames is consistent with Python's
+     * `traceback.extract_tb()`.
+     * </pre>
+     *
+     * <code>repeated string stack_frame_ids = 2;</code>
+     */
+    public Builder addStackFrameIdsBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      ensureStackFrameIdsIsMutable();
+      stackFrameIds_.add(value);
+      onChanged();
+      return this;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.CodeLocation)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.CodeLocation)
+  private static final org.tensorflow.proto.util.CodeLocation DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.util.CodeLocation();
+  }
+
+  public static org.tensorflow.proto.util.CodeLocation getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<CodeLocation>
+      PARSER = new com.google.protobuf.AbstractParser<CodeLocation>() {
+    @java.lang.Override
+    public CodeLocation parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new CodeLocation(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<CodeLocation> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<CodeLocation> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.util.CodeLocation getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/CodeLocationOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/CodeLocationOrBuilder.java
new file mode 100644
index 00000000000..b75306258be
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/CodeLocationOrBuilder.java
@@ -0,0 +1,70 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/debug_event.proto
+
+package org.tensorflow.proto.util;
+
+public interface CodeLocationOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.CodeLocation)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <pre>
+   * Host name on which the source files are located.
+   * </pre>
+   *
+   * <code>string host_name = 1;</code>
+   */
+  java.lang.String getHostName();
+  /**
+   * <pre>
+   * Host name on which the source files are located.
+   * </pre>
+   *
+   * <code>string host_name = 1;</code>
+   */
+  com.google.protobuf.ByteString
+      getHostNameBytes();
+
+  /**
+   * <pre>
+   * ID to a stack frame, each of which is pointed to
+   * by a unique ID. The ordering of the frames is consistent with Python's
+   * `traceback.extract_tb()`.
+   * </pre>
+   *
+   * <code>repeated string stack_frame_ids = 2;</code>
+   */
+  java.util.List<java.lang.String>
+      getStackFrameIdsList();
+  /**
+   * <pre>
+   * ID to a stack frame, each of which is pointed to
+   * by a unique ID. The ordering of the frames is consistent with Python's
+   * `traceback.extract_tb()`.
+   * </pre>
+   *
+   * <code>repeated string stack_frame_ids = 2;</code>
+   */
+  int getStackFrameIdsCount();
+  /**
+   * <pre>
+   * ID to a stack frame, each of which is pointed to
+   * by a unique ID. The ordering of the frames is consistent with Python's
+   * `traceback.extract_tb()`.
+   * </pre>
+   *
+   * <code>repeated string stack_frame_ids = 2;</code>
+   */
+  java.lang.String getStackFrameIds(int index);
+  /**
+   * <pre>
+   * ID to a stack frame, each of which is pointed to
+   * by a unique ID. The ordering of the frames is consistent with Python's
+   * `traceback.extract_tb()`.
+   * </pre>
+   *
+   * <code>repeated string stack_frame_ids = 2;</code>
+   */
+  com.google.protobuf.ByteString
+      getStackFrameIdsBytes(int index);
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/DebugEvent.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/DebugEvent.java
new file mode 100644
index 00000000000..a1b0b986a3d
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/DebugEvent.java
@@ -0,0 +1,2883 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/debug_event.proto
+
+package org.tensorflow.proto.util;
+
+/**
+ * <pre>
+ * An Event related to the debugging of a TensorFlow program.
+ * </pre>
+ *
+ * Protobuf type {@code tensorflow.DebugEvent}
+ */
+public  final class DebugEvent extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.DebugEvent)
+    DebugEventOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use DebugEvent.newBuilder() to construct.
+  private DebugEvent(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private DebugEvent() {
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new DebugEvent();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private DebugEvent(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 9: {
+
+            wallTime_ = input.readDouble();
+            break;
+          }
+          case 16: {
+
+            step_ = input.readInt64();
+            break;
+          }
+          case 26: {
+            org.tensorflow.proto.util.DebugMetadata.Builder subBuilder = null;
+            if (whatCase_ == 3) {
+              subBuilder = ((org.tensorflow.proto.util.DebugMetadata) what_).toBuilder();
+            }
+            what_ =
+                input.readMessage(org.tensorflow.proto.util.DebugMetadata.parser(), extensionRegistry);
+            if (subBuilder != null) {
+              subBuilder.mergeFrom((org.tensorflow.proto.util.DebugMetadata) what_);
+              what_ = subBuilder.buildPartial();
+            }
+            whatCase_ = 3;
+            break;
+          }
+          case 34: {
+            org.tensorflow.proto.util.SourceFile.Builder subBuilder = null;
+            if (whatCase_ == 4) {
+              subBuilder = ((org.tensorflow.proto.util.SourceFile) what_).toBuilder();
+            }
+            what_ =
+                input.readMessage(org.tensorflow.proto.util.SourceFile.parser(), extensionRegistry);
+            if (subBuilder != null) {
+              subBuilder.mergeFrom((org.tensorflow.proto.util.SourceFile) what_);
+              what_ = subBuilder.buildPartial();
+            }
+            whatCase_ = 4;
+            break;
+          }
+          case 50: {
+            org.tensorflow.proto.util.StackFrameWithId.Builder subBuilder = null;
+            if (whatCase_ == 6) {
+              subBuilder = ((org.tensorflow.proto.util.StackFrameWithId) what_).toBuilder();
+            }
+            what_ =
+                input.readMessage(org.tensorflow.proto.util.StackFrameWithId.parser(), extensionRegistry);
+            if (subBuilder != null) {
+              subBuilder.mergeFrom((org.tensorflow.proto.util.StackFrameWithId) what_);
+              what_ = subBuilder.buildPartial();
+            }
+            whatCase_ = 6;
+            break;
+          }
+          case 58: {
+            org.tensorflow.proto.util.GraphOpCreation.Builder subBuilder = null;
+            if (whatCase_ == 7) {
+              subBuilder = ((org.tensorflow.proto.util.GraphOpCreation) what_).toBuilder();
+            }
+            what_ =
+                input.readMessage(org.tensorflow.proto.util.GraphOpCreation.parser(), extensionRegistry);
+            if (subBuilder != null) {
+              subBuilder.mergeFrom((org.tensorflow.proto.util.GraphOpCreation) what_);
+              what_ = subBuilder.buildPartial();
+            }
+            whatCase_ = 7;
+            break;
+          }
+          case 66: {
+            org.tensorflow.proto.util.DebuggedGraph.Builder subBuilder = null;
+            if (whatCase_ == 8) {
+              subBuilder = ((org.tensorflow.proto.util.DebuggedGraph) what_).toBuilder();
+            }
+            what_ =
+                input.readMessage(org.tensorflow.proto.util.DebuggedGraph.parser(), extensionRegistry);
+            if (subBuilder != null) {
+              subBuilder.mergeFrom((org.tensorflow.proto.util.DebuggedGraph) what_);
+              what_ = subBuilder.buildPartial();
+            }
+            whatCase_ = 8;
+            break;
+          }
+          case 74: {
+            org.tensorflow.proto.util.Execution.Builder subBuilder = null;
+            if (whatCase_ == 9) {
+              subBuilder = ((org.tensorflow.proto.util.Execution) what_).toBuilder();
+            }
+            what_ =
+                input.readMessage(org.tensorflow.proto.util.Execution.parser(), extensionRegistry);
+            if (subBuilder != null) {
+              subBuilder.mergeFrom((org.tensorflow.proto.util.Execution) what_);
+              what_ = subBuilder.buildPartial();
+            }
+            whatCase_ = 9;
+            break;
+          }
+          case 82: {
+            org.tensorflow.proto.util.GraphExecutionTrace.Builder subBuilder = null;
+            if (whatCase_ == 10) {
+              subBuilder = ((org.tensorflow.proto.util.GraphExecutionTrace) what_).toBuilder();
+            }
+            what_ =
+                input.readMessage(org.tensorflow.proto.util.GraphExecutionTrace.parser(), extensionRegistry);
+            if (subBuilder != null) {
+              subBuilder.mergeFrom((org.tensorflow.proto.util.GraphExecutionTrace) what_);
+              what_ = subBuilder.buildPartial();
+            }
+            whatCase_ = 10;
+            break;
+          }
+          case 90: {
+            java.lang.String s = input.readStringRequireUtf8();
+            whatCase_ = 11;
+            what_ = s;
+            break;
+          }
+          case 98: {
+            org.tensorflow.proto.util.DebuggedDevice.Builder subBuilder = null;
+            if (whatCase_ == 12) {
+              subBuilder = ((org.tensorflow.proto.util.DebuggedDevice) what_).toBuilder();
+            }
+            what_ =
+                input.readMessage(org.tensorflow.proto.util.DebuggedDevice.parser(), extensionRegistry);
+            if (subBuilder != null) {
+              subBuilder.mergeFrom((org.tensorflow.proto.util.DebuggedDevice) what_);
+              what_ = subBuilder.buildPartial();
+            }
+            whatCase_ = 12;
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.util.DebugEventProtos.internal_static_tensorflow_DebugEvent_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.util.DebugEventProtos.internal_static_tensorflow_DebugEvent_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.util.DebugEvent.class, org.tensorflow.proto.util.DebugEvent.Builder.class);
+  }
+
+  private int whatCase_ = 0;
+  private java.lang.Object what_;
+  public enum WhatCase
+      implements com.google.protobuf.Internal.EnumLite {
+    DEBUG_METADATA(3),
+    SOURCE_FILE(4),
+    STACK_FRAME_WITH_ID(6),
+    GRAPH_OP_CREATION(7),
+    DEBUGGED_GRAPH(8),
+    EXECUTION(9),
+    GRAPH_EXECUTION_TRACE(10),
+    GRAPH_ID(11),
+    DEBUGGED_DEVICE(12),
+    WHAT_NOT_SET(0);
+    private final int value;
+    private WhatCase(int value) {
+      this.value = value;
+    }
+    /**
+     * @deprecated Use {@link #forNumber(int)} instead.
+     */
+    @java.lang.Deprecated
+    public static WhatCase valueOf(int value) {
+      return forNumber(value);
+    }
+
+    public static WhatCase forNumber(int value) {
+      switch (value) {
+        case 3: return DEBUG_METADATA;
+        case 4: return SOURCE_FILE;
+        case 6: return STACK_FRAME_WITH_ID;
+        case 7: return GRAPH_OP_CREATION;
+        case 8: return DEBUGGED_GRAPH;
+        case 9: return EXECUTION;
+        case 10: return GRAPH_EXECUTION_TRACE;
+        case 11: return GRAPH_ID;
+        case 12: return DEBUGGED_DEVICE;
+        case 0: return WHAT_NOT_SET;
+        default: return null;
+      }
+    }
+    public int getNumber() {
+      return this.value;
+    }
+  };
+
+  public WhatCase
+  getWhatCase() {
+    return WhatCase.forNumber(
+        whatCase_);
+  }
+
+  public static final int WALL_TIME_FIELD_NUMBER = 1;
+  private double wallTime_;
+  /**
+   * <pre>
+   * Timestamp in seconds (with microsecond precision).
+   * </pre>
+   *
+   * <code>double wall_time = 1;</code>
+   */
+  public double getWallTime() {
+    return wallTime_;
+  }
+
+  public static final int STEP_FIELD_NUMBER = 2;
+  private long step_;
+  /**
+   * <pre>
+   * Step of training (if available).
+   * </pre>
+   *
+   * <code>int64 step = 2;</code>
+   */
+  public long getStep() {
+    return step_;
+  }
+
+  public static final int DEBUG_METADATA_FIELD_NUMBER = 3;
+  /**
+   * <pre>
+   * Metadata related to this debugging data.
+   * </pre>
+   *
+   * <code>.tensorflow.DebugMetadata debug_metadata = 3;</code>
+   */
+  public boolean hasDebugMetadata() {
+    return whatCase_ == 3;
+  }
+  /**
+   * <pre>
+   * Metadata related to this debugging data.
+   * </pre>
+   *
+   * <code>.tensorflow.DebugMetadata debug_metadata = 3;</code>
+   */
+  public org.tensorflow.proto.util.DebugMetadata getDebugMetadata() {
+    if (whatCase_ == 3) {
+       return (org.tensorflow.proto.util.DebugMetadata) what_;
+    }
+    return org.tensorflow.proto.util.DebugMetadata.getDefaultInstance();
+  }
+  /**
+   * <pre>
+   * Metadata related to this debugging data.
+   * </pre>
+   *
+   * <code>.tensorflow.DebugMetadata debug_metadata = 3;</code>
+   */
+  public org.tensorflow.proto.util.DebugMetadataOrBuilder getDebugMetadataOrBuilder() {
+    if (whatCase_ == 3) {
+       return (org.tensorflow.proto.util.DebugMetadata) what_;
+    }
+    return org.tensorflow.proto.util.DebugMetadata.getDefaultInstance();
+  }
+
+  public static final int SOURCE_FILE_FIELD_NUMBER = 4;
+  /**
+   * <pre>
+   * The content of a source file.
+   * </pre>
+   *
+   * <code>.tensorflow.SourceFile source_file = 4;</code>
+   */
+  public boolean hasSourceFile() {
+    return whatCase_ == 4;
+  }
+  /**
+   * <pre>
+   * The content of a source file.
+   * </pre>
+   *
+   * <code>.tensorflow.SourceFile source_file = 4;</code>
+   */
+  public org.tensorflow.proto.util.SourceFile getSourceFile() {
+    if (whatCase_ == 4) {
+       return (org.tensorflow.proto.util.SourceFile) what_;
+    }
+    return org.tensorflow.proto.util.SourceFile.getDefaultInstance();
+  }
+  /**
+   * <pre>
+   * The content of a source file.
+   * </pre>
+   *
+   * <code>.tensorflow.SourceFile source_file = 4;</code>
+   */
+  public org.tensorflow.proto.util.SourceFileOrBuilder getSourceFileOrBuilder() {
+    if (whatCase_ == 4) {
+       return (org.tensorflow.proto.util.SourceFile) what_;
+    }
+    return org.tensorflow.proto.util.SourceFile.getDefaultInstance();
+  }
+
+  public static final int STACK_FRAME_WITH_ID_FIELD_NUMBER = 6;
+  /**
+   * <pre>
+   * A stack frame (filename, line number and column number, function name and
+   * code string) with ID.
+   * </pre>
+   *
+   * <code>.tensorflow.StackFrameWithId stack_frame_with_id = 6;</code>
+   */
+  public boolean hasStackFrameWithId() {
+    return whatCase_ == 6;
+  }
+  /**
+   * <pre>
+   * A stack frame (filename, line number and column number, function name and
+   * code string) with ID.
+   * </pre>
+   *
+   * <code>.tensorflow.StackFrameWithId stack_frame_with_id = 6;</code>
+   */
+  public org.tensorflow.proto.util.StackFrameWithId getStackFrameWithId() {
+    if (whatCase_ == 6) {
+       return (org.tensorflow.proto.util.StackFrameWithId) what_;
+    }
+    return org.tensorflow.proto.util.StackFrameWithId.getDefaultInstance();
+  }
+  /**
+   * <pre>
+   * A stack frame (filename, line number and column number, function name and
+   * code string) with ID.
+   * </pre>
+   *
+   * <code>.tensorflow.StackFrameWithId stack_frame_with_id = 6;</code>
+   */
+  public org.tensorflow.proto.util.StackFrameWithIdOrBuilder getStackFrameWithIdOrBuilder() {
+    if (whatCase_ == 6) {
+       return (org.tensorflow.proto.util.StackFrameWithId) what_;
+    }
+    return org.tensorflow.proto.util.StackFrameWithId.getDefaultInstance();
+  }
+
+  public static final int GRAPH_OP_CREATION_FIELD_NUMBER = 7;
+  /**
+   * <pre>
+   * The creation of an op within a graph (e.g., a FuncGraph compiled from
+   * a Python function).
+   * </pre>
+   *
+   * <code>.tensorflow.GraphOpCreation graph_op_creation = 7;</code>
+   */
+  public boolean hasGraphOpCreation() {
+    return whatCase_ == 7;
+  }
+  /**
+   * <pre>
+   * The creation of an op within a graph (e.g., a FuncGraph compiled from
+   * a Python function).
+   * </pre>
+   *
+   * <code>.tensorflow.GraphOpCreation graph_op_creation = 7;</code>
+   */
+  public org.tensorflow.proto.util.GraphOpCreation getGraphOpCreation() {
+    if (whatCase_ == 7) {
+       return (org.tensorflow.proto.util.GraphOpCreation) what_;
+    }
+    return org.tensorflow.proto.util.GraphOpCreation.getDefaultInstance();
+  }
+  /**
+   * <pre>
+   * The creation of an op within a graph (e.g., a FuncGraph compiled from
+   * a Python function).
+   * </pre>
+   *
+   * <code>.tensorflow.GraphOpCreation graph_op_creation = 7;</code>
+   */
+  public org.tensorflow.proto.util.GraphOpCreationOrBuilder getGraphOpCreationOrBuilder() {
+    if (whatCase_ == 7) {
+       return (org.tensorflow.proto.util.GraphOpCreation) what_;
+    }
+    return org.tensorflow.proto.util.GraphOpCreation.getDefaultInstance();
+  }
+
+  public static final int DEBUGGED_GRAPH_FIELD_NUMBER = 8;
+  /**
+   * <pre>
+   * Information about a debugged graph.
+   * </pre>
+   *
+   * <code>.tensorflow.DebuggedGraph debugged_graph = 8;</code>
+   */
+  public boolean hasDebuggedGraph() {
+    return whatCase_ == 8;
+  }
+  /**
+   * <pre>
+   * Information about a debugged graph.
+   * </pre>
+   *
+   * <code>.tensorflow.DebuggedGraph debugged_graph = 8;</code>
+   */
+  public org.tensorflow.proto.util.DebuggedGraph getDebuggedGraph() {
+    if (whatCase_ == 8) {
+       return (org.tensorflow.proto.util.DebuggedGraph) what_;
+    }
+    return org.tensorflow.proto.util.DebuggedGraph.getDefaultInstance();
+  }
+  /**
+   * <pre>
+   * Information about a debugged graph.
+   * </pre>
+   *
+   * <code>.tensorflow.DebuggedGraph debugged_graph = 8;</code>
+   */
+  public org.tensorflow.proto.util.DebuggedGraphOrBuilder getDebuggedGraphOrBuilder() {
+    if (whatCase_ == 8) {
+       return (org.tensorflow.proto.util.DebuggedGraph) what_;
+    }
+    return org.tensorflow.proto.util.DebuggedGraph.getDefaultInstance();
+  }
+
+  public static final int EXECUTION_FIELD_NUMBER = 9;
+  /**
+   * <pre>
+   * Execution of an op or a Graph (e.g., a tf.function).
+   * </pre>
+   *
+   * <code>.tensorflow.Execution execution = 9;</code>
+   */
+  public boolean hasExecution() {
+    return whatCase_ == 9;
+  }
+  /**
+   * <pre>
+   * Execution of an op or a Graph (e.g., a tf.function).
+   * </pre>
+   *
+   * <code>.tensorflow.Execution execution = 9;</code>
+   */
+  public org.tensorflow.proto.util.Execution getExecution() {
+    if (whatCase_ == 9) {
+       return (org.tensorflow.proto.util.Execution) what_;
+    }
+    return org.tensorflow.proto.util.Execution.getDefaultInstance();
+  }
+  /**
+   * <pre>
+   * Execution of an op or a Graph (e.g., a tf.function).
+   * </pre>
+   *
+   * <code>.tensorflow.Execution execution = 9;</code>
+   */
+  public org.tensorflow.proto.util.ExecutionOrBuilder getExecutionOrBuilder() {
+    if (whatCase_ == 9) {
+       return (org.tensorflow.proto.util.Execution) what_;
+    }
+    return org.tensorflow.proto.util.Execution.getDefaultInstance();
+  }
+
+  public static final int GRAPH_EXECUTION_TRACE_FIELD_NUMBER = 10;
+  /**
+   * <pre>
+   * A graph execution trace: Contains information about the intermediate
+   * tensors computed during the graph execution.
+   * </pre>
+   *
+   * <code>.tensorflow.GraphExecutionTrace graph_execution_trace = 10;</code>
+   */
+  public boolean hasGraphExecutionTrace() {
+    return whatCase_ == 10;
+  }
+  /**
+   * <pre>
+   * A graph execution trace: Contains information about the intermediate
+   * tensors computed during the graph execution.
+   * </pre>
+   *
+   * <code>.tensorflow.GraphExecutionTrace graph_execution_trace = 10;</code>
+   */
+  public org.tensorflow.proto.util.GraphExecutionTrace getGraphExecutionTrace() {
+    if (whatCase_ == 10) {
+       return (org.tensorflow.proto.util.GraphExecutionTrace) what_;
+    }
+    return org.tensorflow.proto.util.GraphExecutionTrace.getDefaultInstance();
+  }
+  /**
+   * <pre>
+   * A graph execution trace: Contains information about the intermediate
+   * tensors computed during the graph execution.
+   * </pre>
+   *
+   * <code>.tensorflow.GraphExecutionTrace graph_execution_trace = 10;</code>
+   */
+  public org.tensorflow.proto.util.GraphExecutionTraceOrBuilder getGraphExecutionTraceOrBuilder() {
+    if (whatCase_ == 10) {
+       return (org.tensorflow.proto.util.GraphExecutionTrace) what_;
+    }
+    return org.tensorflow.proto.util.GraphExecutionTrace.getDefaultInstance();
+  }
+
+  public static final int GRAPH_ID_FIELD_NUMBER = 11;
+  /**
+   * <pre>
+   * The ID of the graph (i.e., FuncGraph) executed here: applicable only
+   * to the execution of a FuncGraph.
+   * </pre>
+   *
+   * <code>string graph_id = 11;</code>
+   */
+  public java.lang.String getGraphId() {
+    java.lang.Object ref = "";
+    if (whatCase_ == 11) {
+      ref = what_;
+    }
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      if (whatCase_ == 11) {
+        what_ = s;
+      }
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * The ID of the graph (i.e., FuncGraph) executed here: applicable only
+   * to the execution of a FuncGraph.
+   * </pre>
+   *
+   * <code>string graph_id = 11;</code>
+   */
+  public com.google.protobuf.ByteString
+      getGraphIdBytes() {
+    java.lang.Object ref = "";
+    if (whatCase_ == 11) {
+      ref = what_;
+    }
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      if (whatCase_ == 11) {
+        what_ = b;
+      }
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int DEBUGGED_DEVICE_FIELD_NUMBER = 12;
+  /**
+   * <pre>
+   * A device on which debugger-instrumented ops and/or tensors reside.
+   * </pre>
+   *
+   * <code>.tensorflow.DebuggedDevice debugged_device = 12;</code>
+   */
+  public boolean hasDebuggedDevice() {
+    return whatCase_ == 12;
+  }
+  /**
+   * <pre>
+   * A device on which debugger-instrumented ops and/or tensors reside.
+   * </pre>
+   *
+   * <code>.tensorflow.DebuggedDevice debugged_device = 12;</code>
+   */
+  public org.tensorflow.proto.util.DebuggedDevice getDebuggedDevice() {
+    if (whatCase_ == 12) {
+       return (org.tensorflow.proto.util.DebuggedDevice) what_;
+    }
+    return org.tensorflow.proto.util.DebuggedDevice.getDefaultInstance();
+  }
+  /**
+   * <pre>
+   * A device on which debugger-instrumented ops and/or tensors reside.
+   * </pre>
+   *
+   * <code>.tensorflow.DebuggedDevice debugged_device = 12;</code>
+   */
+  public org.tensorflow.proto.util.DebuggedDeviceOrBuilder getDebuggedDeviceOrBuilder() {
+    if (whatCase_ == 12) {
+       return (org.tensorflow.proto.util.DebuggedDevice) what_;
+    }
+    return org.tensorflow.proto.util.DebuggedDevice.getDefaultInstance();
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    if (wallTime_ != 0D) {
+      output.writeDouble(1, wallTime_);
+    }
+    if (step_ != 0L) {
+      output.writeInt64(2, step_);
+    }
+    if (whatCase_ == 3) {
+      output.writeMessage(3, (org.tensorflow.proto.util.DebugMetadata) what_);
+    }
+    if (whatCase_ == 4) {
+      output.writeMessage(4, (org.tensorflow.proto.util.SourceFile) what_);
+    }
+    if (whatCase_ == 6) {
+      output.writeMessage(6, (org.tensorflow.proto.util.StackFrameWithId) what_);
+    }
+    if (whatCase_ == 7) {
+      output.writeMessage(7, (org.tensorflow.proto.util.GraphOpCreation) what_);
+    }
+    if (whatCase_ == 8) {
+      output.writeMessage(8, (org.tensorflow.proto.util.DebuggedGraph) what_);
+    }
+    if (whatCase_ == 9) {
+      output.writeMessage(9, (org.tensorflow.proto.util.Execution) what_);
+    }
+    if (whatCase_ == 10) {
+      output.writeMessage(10, (org.tensorflow.proto.util.GraphExecutionTrace) what_);
+    }
+    if (whatCase_ == 11) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 11, what_);
+    }
+    if (whatCase_ == 12) {
+      output.writeMessage(12, (org.tensorflow.proto.util.DebuggedDevice) what_);
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (wallTime_ != 0D) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeDoubleSize(1, wallTime_);
+    }
+    if (step_ != 0L) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt64Size(2, step_);
+    }
+    if (whatCase_ == 3) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(3, (org.tensorflow.proto.util.DebugMetadata) what_);
+    }
+    if (whatCase_ == 4) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(4, (org.tensorflow.proto.util.SourceFile) what_);
+    }
+    if (whatCase_ == 6) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(6, (org.tensorflow.proto.util.StackFrameWithId) what_);
+    }
+    if (whatCase_ == 7) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(7, (org.tensorflow.proto.util.GraphOpCreation) what_);
+    }
+    if (whatCase_ == 8) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(8, (org.tensorflow.proto.util.DebuggedGraph) what_);
+    }
+    if (whatCase_ == 9) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(9, (org.tensorflow.proto.util.Execution) what_);
+    }
+    if (whatCase_ == 10) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(10, (org.tensorflow.proto.util.GraphExecutionTrace) what_);
+    }
+    if (whatCase_ == 11) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(11, what_);
+    }
+    if (whatCase_ == 12) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(12, (org.tensorflow.proto.util.DebuggedDevice) what_);
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.util.DebugEvent)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.util.DebugEvent other = (org.tensorflow.proto.util.DebugEvent) obj;
+
+    if (java.lang.Double.doubleToLongBits(getWallTime())
+        != java.lang.Double.doubleToLongBits(
+            other.getWallTime())) return false;
+    if (getStep()
+        != other.getStep()) return false;
+    if (!getWhatCase().equals(other.getWhatCase())) return false;
+    switch (whatCase_) {
+      case 3:
+        if (!getDebugMetadata()
+            .equals(other.getDebugMetadata())) return false;
+        break;
+      case 4:
+        if (!getSourceFile()
+            .equals(other.getSourceFile())) return false;
+        break;
+      case 6:
+        if (!getStackFrameWithId()
+            .equals(other.getStackFrameWithId())) return false;
+        break;
+      case 7:
+        if (!getGraphOpCreation()
+            .equals(other.getGraphOpCreation())) return false;
+        break;
+      case 8:
+        if (!getDebuggedGraph()
+            .equals(other.getDebuggedGraph())) return false;
+        break;
+      case 9:
+        if (!getExecution()
+            .equals(other.getExecution())) return false;
+        break;
+      case 10:
+        if (!getGraphExecutionTrace()
+            .equals(other.getGraphExecutionTrace())) return false;
+        break;
+      case 11:
+        if (!getGraphId()
+            .equals(other.getGraphId())) return false;
+        break;
+      case 12:
+        if (!getDebuggedDevice()
+            .equals(other.getDebuggedDevice())) return false;
+        break;
+      case 0:
+      default:
+    }
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    hash = (37 * hash) + WALL_TIME_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+        java.lang.Double.doubleToLongBits(getWallTime()));
+    hash = (37 * hash) + STEP_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+        getStep());
+    switch (whatCase_) {
+      case 3:
+        hash = (37 * hash) + DEBUG_METADATA_FIELD_NUMBER;
+        hash = (53 * hash) + getDebugMetadata().hashCode();
+        break;
+      case 4:
+        hash = (37 * hash) + SOURCE_FILE_FIELD_NUMBER;
+        hash = (53 * hash) + getSourceFile().hashCode();
+        break;
+      case 6:
+        hash = (37 * hash) + STACK_FRAME_WITH_ID_FIELD_NUMBER;
+        hash = (53 * hash) + getStackFrameWithId().hashCode();
+        break;
+      case 7:
+        hash = (37 * hash) + GRAPH_OP_CREATION_FIELD_NUMBER;
+        hash = (53 * hash) + getGraphOpCreation().hashCode();
+        break;
+      case 8:
+        hash = (37 * hash) + DEBUGGED_GRAPH_FIELD_NUMBER;
+        hash = (53 * hash) + getDebuggedGraph().hashCode();
+        break;
+      case 9:
+        hash = (37 * hash) + EXECUTION_FIELD_NUMBER;
+        hash = (53 * hash) + getExecution().hashCode();
+        break;
+      case 10:
+        hash = (37 * hash) + GRAPH_EXECUTION_TRACE_FIELD_NUMBER;
+        hash = (53 * hash) + getGraphExecutionTrace().hashCode();
+        break;
+      case 11:
+        hash = (37 * hash) + GRAPH_ID_FIELD_NUMBER;
+        hash = (53 * hash) + getGraphId().hashCode();
+        break;
+      case 12:
+        hash = (37 * hash) + DEBUGGED_DEVICE_FIELD_NUMBER;
+        hash = (53 * hash) + getDebuggedDevice().hashCode();
+        break;
+      case 0:
+      default:
+    }
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.util.DebugEvent parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.DebugEvent parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.DebugEvent parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.DebugEvent parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.DebugEvent parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.DebugEvent parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.DebugEvent parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.DebugEvent parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.DebugEvent parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.DebugEvent parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.DebugEvent parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.DebugEvent parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.util.DebugEvent prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * <pre>
+   * An Event related to the debugging of a TensorFlow program.
+   * </pre>
+   *
+   * Protobuf type {@code tensorflow.DebugEvent}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.DebugEvent)
+      org.tensorflow.proto.util.DebugEventOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.util.DebugEventProtos.internal_static_tensorflow_DebugEvent_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.util.DebugEventProtos.internal_static_tensorflow_DebugEvent_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.util.DebugEvent.class, org.tensorflow.proto.util.DebugEvent.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.util.DebugEvent.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      wallTime_ = 0D;
+
+      step_ = 0L;
+
+      whatCase_ = 0;
+      what_ = null;
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.util.DebugEventProtos.internal_static_tensorflow_DebugEvent_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.DebugEvent getDefaultInstanceForType() {
+      return org.tensorflow.proto.util.DebugEvent.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.DebugEvent build() {
+      org.tensorflow.proto.util.DebugEvent result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.DebugEvent buildPartial() {
+      org.tensorflow.proto.util.DebugEvent result = new org.tensorflow.proto.util.DebugEvent(this);
+      result.wallTime_ = wallTime_;
+      result.step_ = step_;
+      if (whatCase_ == 3) {
+        if (debugMetadataBuilder_ == null) {
+          result.what_ = what_;
+        } else {
+          result.what_ = debugMetadataBuilder_.build();
+        }
+      }
+      if (whatCase_ == 4) {
+        if (sourceFileBuilder_ == null) {
+          result.what_ = what_;
+        } else {
+          result.what_ = sourceFileBuilder_.build();
+        }
+      }
+      if (whatCase_ == 6) {
+        if (stackFrameWithIdBuilder_ == null) {
+          result.what_ = what_;
+        } else {
+          result.what_ = stackFrameWithIdBuilder_.build();
+        }
+      }
+      if (whatCase_ == 7) {
+        if (graphOpCreationBuilder_ == null) {
+          result.what_ = what_;
+        } else {
+          result.what_ = graphOpCreationBuilder_.build();
+        }
+      }
+      if (whatCase_ == 8) {
+        if (debuggedGraphBuilder_ == null) {
+          result.what_ = what_;
+        } else {
+          result.what_ = debuggedGraphBuilder_.build();
+        }
+      }
+      if (whatCase_ == 9) {
+        if (executionBuilder_ == null) {
+          result.what_ = what_;
+        } else {
+          result.what_ = executionBuilder_.build();
+        }
+      }
+      if (whatCase_ == 10) {
+        if (graphExecutionTraceBuilder_ == null) {
+          result.what_ = what_;
+        } else {
+          result.what_ = graphExecutionTraceBuilder_.build();
+        }
+      }
+      if (whatCase_ == 11) {
+        result.what_ = what_;
+      }
+      if (whatCase_ == 12) {
+        if (debuggedDeviceBuilder_ == null) {
+          result.what_ = what_;
+        } else {
+          result.what_ = debuggedDeviceBuilder_.build();
+        }
+      }
+      result.whatCase_ = whatCase_;
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.util.DebugEvent) {
+        return mergeFrom((org.tensorflow.proto.util.DebugEvent)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.util.DebugEvent other) {
+      if (other == org.tensorflow.proto.util.DebugEvent.getDefaultInstance()) return this;
+      if (other.getWallTime() != 0D) {
+        setWallTime(other.getWallTime());
+      }
+      if (other.getStep() != 0L) {
+        setStep(other.getStep());
+      }
+      switch (other.getWhatCase()) {
+        case DEBUG_METADATA: {
+          mergeDebugMetadata(other.getDebugMetadata());
+          break;
+        }
+        case SOURCE_FILE: {
+          mergeSourceFile(other.getSourceFile());
+          break;
+        }
+        case STACK_FRAME_WITH_ID: {
+          mergeStackFrameWithId(other.getStackFrameWithId());
+          break;
+        }
+        case GRAPH_OP_CREATION: {
+          mergeGraphOpCreation(other.getGraphOpCreation());
+          break;
+        }
+        case DEBUGGED_GRAPH: {
+          mergeDebuggedGraph(other.getDebuggedGraph());
+          break;
+        }
+        case EXECUTION: {
+          mergeExecution(other.getExecution());
+          break;
+        }
+        case GRAPH_EXECUTION_TRACE: {
+          mergeGraphExecutionTrace(other.getGraphExecutionTrace());
+          break;
+        }
+        case GRAPH_ID: {
+          whatCase_ = 11;
+          what_ = other.what_;
+          onChanged();
+          break;
+        }
+        case DEBUGGED_DEVICE: {
+          mergeDebuggedDevice(other.getDebuggedDevice());
+          break;
+        }
+        case WHAT_NOT_SET: {
+          break;
+        }
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.util.DebugEvent parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.util.DebugEvent) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+    private int whatCase_ = 0;
+    private java.lang.Object what_;
+    public WhatCase
+        getWhatCase() {
+      return WhatCase.forNumber(
+          whatCase_);
+    }
+
+    public Builder clearWhat() {
+      whatCase_ = 0;
+      what_ = null;
+      onChanged();
+      return this;
+    }
+
+
+    private double wallTime_ ;
+    /**
+     * <pre>
+     * Timestamp in seconds (with microsecond precision).
+     * </pre>
+     *
+     * <code>double wall_time = 1;</code>
+     */
+    public double getWallTime() {
+      return wallTime_;
+    }
+    /**
+     * <pre>
+     * Timestamp in seconds (with microsecond precision).
+     * </pre>
+     *
+     * <code>double wall_time = 1;</code>
+     */
+    public Builder setWallTime(double value) {
+      
+      wallTime_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Timestamp in seconds (with microsecond precision).
+     * </pre>
+     *
+     * <code>double wall_time = 1;</code>
+     */
+    public Builder clearWallTime() {
+      
+      wallTime_ = 0D;
+      onChanged();
+      return this;
+    }
+
+    private long step_ ;
+    /**
+     * <pre>
+     * Step of training (if available).
+     * </pre>
+     *
+     * <code>int64 step = 2;</code>
+     */
+    public long getStep() {
+      return step_;
+    }
+    /**
+     * <pre>
+     * Step of training (if available).
+     * </pre>
+     *
+     * <code>int64 step = 2;</code>
+     */
+    public Builder setStep(long value) {
+      
+      step_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Step of training (if available).
+     * </pre>
+     *
+     * <code>int64 step = 2;</code>
+     */
+    public Builder clearStep() {
+      
+      step_ = 0L;
+      onChanged();
+      return this;
+    }
+
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.util.DebugMetadata, org.tensorflow.proto.util.DebugMetadata.Builder, org.tensorflow.proto.util.DebugMetadataOrBuilder> debugMetadataBuilder_;
+    /**
+     * <pre>
+     * Metadata related to this debugging data.
+     * </pre>
+     *
+     * <code>.tensorflow.DebugMetadata debug_metadata = 3;</code>
+     */
+    public boolean hasDebugMetadata() {
+      return whatCase_ == 3;
+    }
+    /**
+     * <pre>
+     * Metadata related to this debugging data.
+     * </pre>
+     *
+     * <code>.tensorflow.DebugMetadata debug_metadata = 3;</code>
+     */
+    public org.tensorflow.proto.util.DebugMetadata getDebugMetadata() {
+      if (debugMetadataBuilder_ == null) {
+        if (whatCase_ == 3) {
+          return (org.tensorflow.proto.util.DebugMetadata) what_;
+        }
+        return org.tensorflow.proto.util.DebugMetadata.getDefaultInstance();
+      } else {
+        if (whatCase_ == 3) {
+          return debugMetadataBuilder_.getMessage();
+        }
+        return org.tensorflow.proto.util.DebugMetadata.getDefaultInstance();
+      }
+    }
+    /**
+     * <pre>
+     * Metadata related to this debugging data.
+     * </pre>
+     *
+     * <code>.tensorflow.DebugMetadata debug_metadata = 3;</code>
+     */
+    public Builder setDebugMetadata(org.tensorflow.proto.util.DebugMetadata value) {
+      if (debugMetadataBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        what_ = value;
+        onChanged();
+      } else {
+        debugMetadataBuilder_.setMessage(value);
+      }
+      whatCase_ = 3;
+      return this;
+    }
+    /**
+     * <pre>
+     * Metadata related to this debugging data.
+     * </pre>
+     *
+     * <code>.tensorflow.DebugMetadata debug_metadata = 3;</code>
+     */
+    public Builder setDebugMetadata(
+        org.tensorflow.proto.util.DebugMetadata.Builder builderForValue) {
+      if (debugMetadataBuilder_ == null) {
+        what_ = builderForValue.build();
+        onChanged();
+      } else {
+        debugMetadataBuilder_.setMessage(builderForValue.build());
+      }
+      whatCase_ = 3;
+      return this;
+    }
+    /**
+     * <pre>
+     * Metadata related to this debugging data.
+     * </pre>
+     *
+     * <code>.tensorflow.DebugMetadata debug_metadata = 3;</code>
+     */
+    public Builder mergeDebugMetadata(org.tensorflow.proto.util.DebugMetadata value) {
+      if (debugMetadataBuilder_ == null) {
+        if (whatCase_ == 3 &&
+            what_ != org.tensorflow.proto.util.DebugMetadata.getDefaultInstance()) {
+          what_ = org.tensorflow.proto.util.DebugMetadata.newBuilder((org.tensorflow.proto.util.DebugMetadata) what_)
+              .mergeFrom(value).buildPartial();
+        } else {
+          what_ = value;
+        }
+        onChanged();
+      } else {
+        if (whatCase_ == 3) {
+          debugMetadataBuilder_.mergeFrom(value);
+        }
+        debugMetadataBuilder_.setMessage(value);
+      }
+      whatCase_ = 3;
+      return this;
+    }
+    /**
+     * <pre>
+     * Metadata related to this debugging data.
+     * </pre>
+     *
+     * <code>.tensorflow.DebugMetadata debug_metadata = 3;</code>
+     */
+    public Builder clearDebugMetadata() {
+      if (debugMetadataBuilder_ == null) {
+        if (whatCase_ == 3) {
+          whatCase_ = 0;
+          what_ = null;
+          onChanged();
+        }
+      } else {
+        if (whatCase_ == 3) {
+          whatCase_ = 0;
+          what_ = null;
+        }
+        debugMetadataBuilder_.clear();
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Metadata related to this debugging data.
+     * </pre>
+     *
+     * <code>.tensorflow.DebugMetadata debug_metadata = 3;</code>
+     */
+    public org.tensorflow.proto.util.DebugMetadata.Builder getDebugMetadataBuilder() {
+      return getDebugMetadataFieldBuilder().getBuilder();
+    }
+    /**
+     * <pre>
+     * Metadata related to this debugging data.
+     * </pre>
+     *
+     * <code>.tensorflow.DebugMetadata debug_metadata = 3;</code>
+     */
+    public org.tensorflow.proto.util.DebugMetadataOrBuilder getDebugMetadataOrBuilder() {
+      if ((whatCase_ == 3) && (debugMetadataBuilder_ != null)) {
+        return debugMetadataBuilder_.getMessageOrBuilder();
+      } else {
+        if (whatCase_ == 3) {
+          return (org.tensorflow.proto.util.DebugMetadata) what_;
+        }
+        return org.tensorflow.proto.util.DebugMetadata.getDefaultInstance();
+      }
+    }
+    /**
+     * <pre>
+     * Metadata related to this debugging data.
+     * </pre>
+     *
+     * <code>.tensorflow.DebugMetadata debug_metadata = 3;</code>
+     */
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.util.DebugMetadata, org.tensorflow.proto.util.DebugMetadata.Builder, org.tensorflow.proto.util.DebugMetadataOrBuilder> 
+        getDebugMetadataFieldBuilder() {
+      if (debugMetadataBuilder_ == null) {
+        if (!(whatCase_ == 3)) {
+          what_ = org.tensorflow.proto.util.DebugMetadata.getDefaultInstance();
+        }
+        debugMetadataBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+            org.tensorflow.proto.util.DebugMetadata, org.tensorflow.proto.util.DebugMetadata.Builder, org.tensorflow.proto.util.DebugMetadataOrBuilder>(
+                (org.tensorflow.proto.util.DebugMetadata) what_,
+                getParentForChildren(),
+                isClean());
+        what_ = null;
+      }
+      whatCase_ = 3;
+      onChanged();;
+      return debugMetadataBuilder_;
+    }
+
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.util.SourceFile, org.tensorflow.proto.util.SourceFile.Builder, org.tensorflow.proto.util.SourceFileOrBuilder> sourceFileBuilder_;
+    /**
+     * <pre>
+     * The content of a source file.
+     * </pre>
+     *
+     * <code>.tensorflow.SourceFile source_file = 4;</code>
+     */
+    public boolean hasSourceFile() {
+      return whatCase_ == 4;
+    }
+    /**
+     * <pre>
+     * The content of a source file.
+     * </pre>
+     *
+     * <code>.tensorflow.SourceFile source_file = 4;</code>
+     */
+    public org.tensorflow.proto.util.SourceFile getSourceFile() {
+      if (sourceFileBuilder_ == null) {
+        if (whatCase_ == 4) {
+          return (org.tensorflow.proto.util.SourceFile) what_;
+        }
+        return org.tensorflow.proto.util.SourceFile.getDefaultInstance();
+      } else {
+        if (whatCase_ == 4) {
+          return sourceFileBuilder_.getMessage();
+        }
+        return org.tensorflow.proto.util.SourceFile.getDefaultInstance();
+      }
+    }
+    /**
+     * <pre>
+     * The content of a source file.
+     * </pre>
+     *
+     * <code>.tensorflow.SourceFile source_file = 4;</code>
+     */
+    public Builder setSourceFile(org.tensorflow.proto.util.SourceFile value) {
+      if (sourceFileBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        what_ = value;
+        onChanged();
+      } else {
+        sourceFileBuilder_.setMessage(value);
+      }
+      whatCase_ = 4;
+      return this;
+    }
+    /**
+     * <pre>
+     * The content of a source file.
+     * </pre>
+     *
+     * <code>.tensorflow.SourceFile source_file = 4;</code>
+     */
+    public Builder setSourceFile(
+        org.tensorflow.proto.util.SourceFile.Builder builderForValue) {
+      if (sourceFileBuilder_ == null) {
+        what_ = builderForValue.build();
+        onChanged();
+      } else {
+        sourceFileBuilder_.setMessage(builderForValue.build());
+      }
+      whatCase_ = 4;
+      return this;
+    }
+    /**
+     * <pre>
+     * The content of a source file.
+     * </pre>
+     *
+     * <code>.tensorflow.SourceFile source_file = 4;</code>
+     */
+    public Builder mergeSourceFile(org.tensorflow.proto.util.SourceFile value) {
+      if (sourceFileBuilder_ == null) {
+        if (whatCase_ == 4 &&
+            what_ != org.tensorflow.proto.util.SourceFile.getDefaultInstance()) {
+          what_ = org.tensorflow.proto.util.SourceFile.newBuilder((org.tensorflow.proto.util.SourceFile) what_)
+              .mergeFrom(value).buildPartial();
+        } else {
+          what_ = value;
+        }
+        onChanged();
+      } else {
+        if (whatCase_ == 4) {
+          sourceFileBuilder_.mergeFrom(value);
+        }
+        sourceFileBuilder_.setMessage(value);
+      }
+      whatCase_ = 4;
+      return this;
+    }
+    /**
+     * <pre>
+     * The content of a source file.
+     * </pre>
+     *
+     * <code>.tensorflow.SourceFile source_file = 4;</code>
+     */
+    public Builder clearSourceFile() {
+      if (sourceFileBuilder_ == null) {
+        if (whatCase_ == 4) {
+          whatCase_ = 0;
+          what_ = null;
+          onChanged();
+        }
+      } else {
+        if (whatCase_ == 4) {
+          whatCase_ = 0;
+          what_ = null;
+        }
+        sourceFileBuilder_.clear();
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * The content of a source file.
+     * </pre>
+     *
+     * <code>.tensorflow.SourceFile source_file = 4;</code>
+     */
+    public org.tensorflow.proto.util.SourceFile.Builder getSourceFileBuilder() {
+      return getSourceFileFieldBuilder().getBuilder();
+    }
+    /**
+     * <pre>
+     * The content of a source file.
+     * </pre>
+     *
+     * <code>.tensorflow.SourceFile source_file = 4;</code>
+     */
+    public org.tensorflow.proto.util.SourceFileOrBuilder getSourceFileOrBuilder() {
+      if ((whatCase_ == 4) && (sourceFileBuilder_ != null)) {
+        return sourceFileBuilder_.getMessageOrBuilder();
+      } else {
+        if (whatCase_ == 4) {
+          return (org.tensorflow.proto.util.SourceFile) what_;
+        }
+        return org.tensorflow.proto.util.SourceFile.getDefaultInstance();
+      }
+    }
+    /**
+     * <pre>
+     * The content of a source file.
+     * </pre>
+     *
+     * <code>.tensorflow.SourceFile source_file = 4;</code>
+     */
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.util.SourceFile, org.tensorflow.proto.util.SourceFile.Builder, org.tensorflow.proto.util.SourceFileOrBuilder> 
+        getSourceFileFieldBuilder() {
+      if (sourceFileBuilder_ == null) {
+        if (!(whatCase_ == 4)) {
+          what_ = org.tensorflow.proto.util.SourceFile.getDefaultInstance();
+        }
+        sourceFileBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+            org.tensorflow.proto.util.SourceFile, org.tensorflow.proto.util.SourceFile.Builder, org.tensorflow.proto.util.SourceFileOrBuilder>(
+                (org.tensorflow.proto.util.SourceFile) what_,
+                getParentForChildren(),
+                isClean());
+        what_ = null;
+      }
+      whatCase_ = 4;
+      onChanged();;
+      return sourceFileBuilder_;
+    }
+
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.util.StackFrameWithId, org.tensorflow.proto.util.StackFrameWithId.Builder, org.tensorflow.proto.util.StackFrameWithIdOrBuilder> stackFrameWithIdBuilder_;
+    /**
+     * <pre>
+     * A stack frame (filename, line number and column number, function name and
+     * code string) with ID.
+     * </pre>
+     *
+     * <code>.tensorflow.StackFrameWithId stack_frame_with_id = 6;</code>
+     */
+    public boolean hasStackFrameWithId() {
+      return whatCase_ == 6;
+    }
+    /**
+     * <pre>
+     * A stack frame (filename, line number and column number, function name and
+     * code string) with ID.
+     * </pre>
+     *
+     * <code>.tensorflow.StackFrameWithId stack_frame_with_id = 6;</code>
+     */
+    public org.tensorflow.proto.util.StackFrameWithId getStackFrameWithId() {
+      if (stackFrameWithIdBuilder_ == null) {
+        if (whatCase_ == 6) {
+          return (org.tensorflow.proto.util.StackFrameWithId) what_;
+        }
+        return org.tensorflow.proto.util.StackFrameWithId.getDefaultInstance();
+      } else {
+        if (whatCase_ == 6) {
+          return stackFrameWithIdBuilder_.getMessage();
+        }
+        return org.tensorflow.proto.util.StackFrameWithId.getDefaultInstance();
+      }
+    }
+    /**
+     * <pre>
+     * A stack frame (filename, line number and column number, function name and
+     * code string) with ID.
+     * </pre>
+     *
+     * <code>.tensorflow.StackFrameWithId stack_frame_with_id = 6;</code>
+     */
+    public Builder setStackFrameWithId(org.tensorflow.proto.util.StackFrameWithId value) {
+      if (stackFrameWithIdBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        what_ = value;
+        onChanged();
+      } else {
+        stackFrameWithIdBuilder_.setMessage(value);
+      }
+      whatCase_ = 6;
+      return this;
+    }
+    /**
+     * <pre>
+     * A stack frame (filename, line number and column number, function name and
+     * code string) with ID.
+     * </pre>
+     *
+     * <code>.tensorflow.StackFrameWithId stack_frame_with_id = 6;</code>
+     */
+    public Builder setStackFrameWithId(
+        org.tensorflow.proto.util.StackFrameWithId.Builder builderForValue) {
+      if (stackFrameWithIdBuilder_ == null) {
+        what_ = builderForValue.build();
+        onChanged();
+      } else {
+        stackFrameWithIdBuilder_.setMessage(builderForValue.build());
+      }
+      whatCase_ = 6;
+      return this;
+    }
+    /**
+     * <pre>
+     * A stack frame (filename, line number and column number, function name and
+     * code string) with ID.
+     * </pre>
+     *
+     * <code>.tensorflow.StackFrameWithId stack_frame_with_id = 6;</code>
+     */
+    public Builder mergeStackFrameWithId(org.tensorflow.proto.util.StackFrameWithId value) {
+      if (stackFrameWithIdBuilder_ == null) {
+        if (whatCase_ == 6 &&
+            what_ != org.tensorflow.proto.util.StackFrameWithId.getDefaultInstance()) {
+          what_ = org.tensorflow.proto.util.StackFrameWithId.newBuilder((org.tensorflow.proto.util.StackFrameWithId) what_)
+              .mergeFrom(value).buildPartial();
+        } else {
+          what_ = value;
+        }
+        onChanged();
+      } else {
+        if (whatCase_ == 6) {
+          stackFrameWithIdBuilder_.mergeFrom(value);
+        }
+        stackFrameWithIdBuilder_.setMessage(value);
+      }
+      whatCase_ = 6;
+      return this;
+    }
+    /**
+     * <pre>
+     * A stack frame (filename, line number and column number, function name and
+     * code string) with ID.
+     * </pre>
+     *
+     * <code>.tensorflow.StackFrameWithId stack_frame_with_id = 6;</code>
+     */
+    public Builder clearStackFrameWithId() {
+      if (stackFrameWithIdBuilder_ == null) {
+        if (whatCase_ == 6) {
+          whatCase_ = 0;
+          what_ = null;
+          onChanged();
+        }
+      } else {
+        if (whatCase_ == 6) {
+          whatCase_ = 0;
+          what_ = null;
+        }
+        stackFrameWithIdBuilder_.clear();
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * A stack frame (filename, line number and column number, function name and
+     * code string) with ID.
+     * </pre>
+     *
+     * <code>.tensorflow.StackFrameWithId stack_frame_with_id = 6;</code>
+     */
+    public org.tensorflow.proto.util.StackFrameWithId.Builder getStackFrameWithIdBuilder() {
+      return getStackFrameWithIdFieldBuilder().getBuilder();
+    }
+    /**
+     * <pre>
+     * A stack frame (filename, line number and column number, function name and
+     * code string) with ID.
+     * </pre>
+     *
+     * <code>.tensorflow.StackFrameWithId stack_frame_with_id = 6;</code>
+     */
+    public org.tensorflow.proto.util.StackFrameWithIdOrBuilder getStackFrameWithIdOrBuilder() {
+      if ((whatCase_ == 6) && (stackFrameWithIdBuilder_ != null)) {
+        return stackFrameWithIdBuilder_.getMessageOrBuilder();
+      } else {
+        if (whatCase_ == 6) {
+          return (org.tensorflow.proto.util.StackFrameWithId) what_;
+        }
+        return org.tensorflow.proto.util.StackFrameWithId.getDefaultInstance();
+      }
+    }
+    /**
+     * <pre>
+     * A stack frame (filename, line number and column number, function name and
+     * code string) with ID.
+     * </pre>
+     *
+     * <code>.tensorflow.StackFrameWithId stack_frame_with_id = 6;</code>
+     */
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.util.StackFrameWithId, org.tensorflow.proto.util.StackFrameWithId.Builder, org.tensorflow.proto.util.StackFrameWithIdOrBuilder> 
+        getStackFrameWithIdFieldBuilder() {
+      if (stackFrameWithIdBuilder_ == null) {
+        if (!(whatCase_ == 6)) {
+          what_ = org.tensorflow.proto.util.StackFrameWithId.getDefaultInstance();
+        }
+        stackFrameWithIdBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+            org.tensorflow.proto.util.StackFrameWithId, org.tensorflow.proto.util.StackFrameWithId.Builder, org.tensorflow.proto.util.StackFrameWithIdOrBuilder>(
+                (org.tensorflow.proto.util.StackFrameWithId) what_,
+                getParentForChildren(),
+                isClean());
+        what_ = null;
+      }
+      whatCase_ = 6;
+      onChanged();;
+      return stackFrameWithIdBuilder_;
+    }
+
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.util.GraphOpCreation, org.tensorflow.proto.util.GraphOpCreation.Builder, org.tensorflow.proto.util.GraphOpCreationOrBuilder> graphOpCreationBuilder_;
+    /**
+     * <pre>
+     * The creation of an op within a graph (e.g., a FuncGraph compiled from
+     * a Python function).
+     * </pre>
+     *
+     * <code>.tensorflow.GraphOpCreation graph_op_creation = 7;</code>
+     */
+    public boolean hasGraphOpCreation() {
+      return whatCase_ == 7;
+    }
+    /**
+     * <pre>
+     * The creation of an op within a graph (e.g., a FuncGraph compiled from
+     * a Python function).
+     * </pre>
+     *
+     * <code>.tensorflow.GraphOpCreation graph_op_creation = 7;</code>
+     */
+    public org.tensorflow.proto.util.GraphOpCreation getGraphOpCreation() {
+      if (graphOpCreationBuilder_ == null) {
+        if (whatCase_ == 7) {
+          return (org.tensorflow.proto.util.GraphOpCreation) what_;
+        }
+        return org.tensorflow.proto.util.GraphOpCreation.getDefaultInstance();
+      } else {
+        if (whatCase_ == 7) {
+          return graphOpCreationBuilder_.getMessage();
+        }
+        return org.tensorflow.proto.util.GraphOpCreation.getDefaultInstance();
+      }
+    }
+    /**
+     * <pre>
+     * The creation of an op within a graph (e.g., a FuncGraph compiled from
+     * a Python function).
+     * </pre>
+     *
+     * <code>.tensorflow.GraphOpCreation graph_op_creation = 7;</code>
+     */
+    public Builder setGraphOpCreation(org.tensorflow.proto.util.GraphOpCreation value) {
+      if (graphOpCreationBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        what_ = value;
+        onChanged();
+      } else {
+        graphOpCreationBuilder_.setMessage(value);
+      }
+      whatCase_ = 7;
+      return this;
+    }
+    /**
+     * <pre>
+     * The creation of an op within a graph (e.g., a FuncGraph compiled from
+     * a Python function).
+     * </pre>
+     *
+     * <code>.tensorflow.GraphOpCreation graph_op_creation = 7;</code>
+     */
+    public Builder setGraphOpCreation(
+        org.tensorflow.proto.util.GraphOpCreation.Builder builderForValue) {
+      if (graphOpCreationBuilder_ == null) {
+        what_ = builderForValue.build();
+        onChanged();
+      } else {
+        graphOpCreationBuilder_.setMessage(builderForValue.build());
+      }
+      whatCase_ = 7;
+      return this;
+    }
+    /**
+     * <pre>
+     * The creation of an op within a graph (e.g., a FuncGraph compiled from
+     * a Python function).
+     * </pre>
+     *
+     * <code>.tensorflow.GraphOpCreation graph_op_creation = 7;</code>
+     */
+    public Builder mergeGraphOpCreation(org.tensorflow.proto.util.GraphOpCreation value) {
+      if (graphOpCreationBuilder_ == null) {
+        if (whatCase_ == 7 &&
+            what_ != org.tensorflow.proto.util.GraphOpCreation.getDefaultInstance()) {
+          what_ = org.tensorflow.proto.util.GraphOpCreation.newBuilder((org.tensorflow.proto.util.GraphOpCreation) what_)
+              .mergeFrom(value).buildPartial();
+        } else {
+          what_ = value;
+        }
+        onChanged();
+      } else {
+        if (whatCase_ == 7) {
+          graphOpCreationBuilder_.mergeFrom(value);
+        }
+        graphOpCreationBuilder_.setMessage(value);
+      }
+      whatCase_ = 7;
+      return this;
+    }
+    /**
+     * <pre>
+     * The creation of an op within a graph (e.g., a FuncGraph compiled from
+     * a Python function).
+     * </pre>
+     *
+     * <code>.tensorflow.GraphOpCreation graph_op_creation = 7;</code>
+     */
+    public Builder clearGraphOpCreation() {
+      if (graphOpCreationBuilder_ == null) {
+        if (whatCase_ == 7) {
+          whatCase_ = 0;
+          what_ = null;
+          onChanged();
+        }
+      } else {
+        if (whatCase_ == 7) {
+          whatCase_ = 0;
+          what_ = null;
+        }
+        graphOpCreationBuilder_.clear();
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * The creation of an op within a graph (e.g., a FuncGraph compiled from
+     * a Python function).
+     * </pre>
+     *
+     * <code>.tensorflow.GraphOpCreation graph_op_creation = 7;</code>
+     */
+    public org.tensorflow.proto.util.GraphOpCreation.Builder getGraphOpCreationBuilder() {
+      return getGraphOpCreationFieldBuilder().getBuilder();
+    }
+    /**
+     * <pre>
+     * The creation of an op within a graph (e.g., a FuncGraph compiled from
+     * a Python function).
+     * </pre>
+     *
+     * <code>.tensorflow.GraphOpCreation graph_op_creation = 7;</code>
+     */
+    public org.tensorflow.proto.util.GraphOpCreationOrBuilder getGraphOpCreationOrBuilder() {
+      if ((whatCase_ == 7) && (graphOpCreationBuilder_ != null)) {
+        return graphOpCreationBuilder_.getMessageOrBuilder();
+      } else {
+        if (whatCase_ == 7) {
+          return (org.tensorflow.proto.util.GraphOpCreation) what_;
+        }
+        return org.tensorflow.proto.util.GraphOpCreation.getDefaultInstance();
+      }
+    }
+    /**
+     * <pre>
+     * The creation of an op within a graph (e.g., a FuncGraph compiled from
+     * a Python function).
+     * </pre>
+     *
+     * <code>.tensorflow.GraphOpCreation graph_op_creation = 7;</code>
+     */
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.util.GraphOpCreation, org.tensorflow.proto.util.GraphOpCreation.Builder, org.tensorflow.proto.util.GraphOpCreationOrBuilder> 
+        getGraphOpCreationFieldBuilder() {
+      if (graphOpCreationBuilder_ == null) {
+        if (!(whatCase_ == 7)) {
+          what_ = org.tensorflow.proto.util.GraphOpCreation.getDefaultInstance();
+        }
+        graphOpCreationBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+            org.tensorflow.proto.util.GraphOpCreation, org.tensorflow.proto.util.GraphOpCreation.Builder, org.tensorflow.proto.util.GraphOpCreationOrBuilder>(
+                (org.tensorflow.proto.util.GraphOpCreation) what_,
+                getParentForChildren(),
+                isClean());
+        what_ = null;
+      }
+      whatCase_ = 7;
+      onChanged();;
+      return graphOpCreationBuilder_;
+    }
+
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.util.DebuggedGraph, org.tensorflow.proto.util.DebuggedGraph.Builder, org.tensorflow.proto.util.DebuggedGraphOrBuilder> debuggedGraphBuilder_;
+    /**
+     * <pre>
+     * Information about a debugged graph.
+     * </pre>
+     *
+     * <code>.tensorflow.DebuggedGraph debugged_graph = 8;</code>
+     */
+    public boolean hasDebuggedGraph() {
+      return whatCase_ == 8;
+    }
+    /**
+     * <pre>
+     * Information about a debugged graph.
+     * </pre>
+     *
+     * <code>.tensorflow.DebuggedGraph debugged_graph = 8;</code>
+     */
+    public org.tensorflow.proto.util.DebuggedGraph getDebuggedGraph() {
+      if (debuggedGraphBuilder_ == null) {
+        if (whatCase_ == 8) {
+          return (org.tensorflow.proto.util.DebuggedGraph) what_;
+        }
+        return org.tensorflow.proto.util.DebuggedGraph.getDefaultInstance();
+      } else {
+        if (whatCase_ == 8) {
+          return debuggedGraphBuilder_.getMessage();
+        }
+        return org.tensorflow.proto.util.DebuggedGraph.getDefaultInstance();
+      }
+    }
+    /**
+     * <pre>
+     * Information about a debugged graph.
+     * </pre>
+     *
+     * <code>.tensorflow.DebuggedGraph debugged_graph = 8;</code>
+     */
+    public Builder setDebuggedGraph(org.tensorflow.proto.util.DebuggedGraph value) {
+      if (debuggedGraphBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        what_ = value;
+        onChanged();
+      } else {
+        debuggedGraphBuilder_.setMessage(value);
+      }
+      whatCase_ = 8;
+      return this;
+    }
+    /**
+     * <pre>
+     * Information about a debugged graph.
+     * </pre>
+     *
+     * <code>.tensorflow.DebuggedGraph debugged_graph = 8;</code>
+     */
+    public Builder setDebuggedGraph(
+        org.tensorflow.proto.util.DebuggedGraph.Builder builderForValue) {
+      if (debuggedGraphBuilder_ == null) {
+        what_ = builderForValue.build();
+        onChanged();
+      } else {
+        debuggedGraphBuilder_.setMessage(builderForValue.build());
+      }
+      whatCase_ = 8;
+      return this;
+    }
+    /**
+     * <pre>
+     * Information about a debugged graph.
+     * </pre>
+     *
+     * <code>.tensorflow.DebuggedGraph debugged_graph = 8;</code>
+     */
+    public Builder mergeDebuggedGraph(org.tensorflow.proto.util.DebuggedGraph value) {
+      if (debuggedGraphBuilder_ == null) {
+        if (whatCase_ == 8 &&
+            what_ != org.tensorflow.proto.util.DebuggedGraph.getDefaultInstance()) {
+          what_ = org.tensorflow.proto.util.DebuggedGraph.newBuilder((org.tensorflow.proto.util.DebuggedGraph) what_)
+              .mergeFrom(value).buildPartial();
+        } else {
+          what_ = value;
+        }
+        onChanged();
+      } else {
+        if (whatCase_ == 8) {
+          debuggedGraphBuilder_.mergeFrom(value);
+        }
+        debuggedGraphBuilder_.setMessage(value);
+      }
+      whatCase_ = 8;
+      return this;
+    }
+    /**
+     * <pre>
+     * Information about a debugged graph.
+     * </pre>
+     *
+     * <code>.tensorflow.DebuggedGraph debugged_graph = 8;</code>
+     */
+    public Builder clearDebuggedGraph() {
+      if (debuggedGraphBuilder_ == null) {
+        if (whatCase_ == 8) {
+          whatCase_ = 0;
+          what_ = null;
+          onChanged();
+        }
+      } else {
+        if (whatCase_ == 8) {
+          whatCase_ = 0;
+          what_ = null;
+        }
+        debuggedGraphBuilder_.clear();
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Information about a debugged graph.
+     * </pre>
+     *
+     * <code>.tensorflow.DebuggedGraph debugged_graph = 8;</code>
+     */
+    public org.tensorflow.proto.util.DebuggedGraph.Builder getDebuggedGraphBuilder() {
+      return getDebuggedGraphFieldBuilder().getBuilder();
+    }
+    /**
+     * <pre>
+     * Information about a debugged graph.
+     * </pre>
+     *
+     * <code>.tensorflow.DebuggedGraph debugged_graph = 8;</code>
+     */
+    public org.tensorflow.proto.util.DebuggedGraphOrBuilder getDebuggedGraphOrBuilder() {
+      if ((whatCase_ == 8) && (debuggedGraphBuilder_ != null)) {
+        return debuggedGraphBuilder_.getMessageOrBuilder();
+      } else {
+        if (whatCase_ == 8) {
+          return (org.tensorflow.proto.util.DebuggedGraph) what_;
+        }
+        return org.tensorflow.proto.util.DebuggedGraph.getDefaultInstance();
+      }
+    }
+    /**
+     * <pre>
+     * Information about a debugged graph.
+     * </pre>
+     *
+     * <code>.tensorflow.DebuggedGraph debugged_graph = 8;</code>
+     */
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.util.DebuggedGraph, org.tensorflow.proto.util.DebuggedGraph.Builder, org.tensorflow.proto.util.DebuggedGraphOrBuilder> 
+        getDebuggedGraphFieldBuilder() {
+      if (debuggedGraphBuilder_ == null) {
+        if (!(whatCase_ == 8)) {
+          what_ = org.tensorflow.proto.util.DebuggedGraph.getDefaultInstance();
+        }
+        debuggedGraphBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+            org.tensorflow.proto.util.DebuggedGraph, org.tensorflow.proto.util.DebuggedGraph.Builder, org.tensorflow.proto.util.DebuggedGraphOrBuilder>(
+                (org.tensorflow.proto.util.DebuggedGraph) what_,
+                getParentForChildren(),
+                isClean());
+        what_ = null;
+      }
+      whatCase_ = 8;
+      onChanged();;
+      return debuggedGraphBuilder_;
+    }
+
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.util.Execution, org.tensorflow.proto.util.Execution.Builder, org.tensorflow.proto.util.ExecutionOrBuilder> executionBuilder_;
+    /**
+     * <pre>
+     * Execution of an op or a Graph (e.g., a tf.function).
+     * </pre>
+     *
+     * <code>.tensorflow.Execution execution = 9;</code>
+     */
+    public boolean hasExecution() {
+      return whatCase_ == 9;
+    }
+    /**
+     * <pre>
+     * Execution of an op or a Graph (e.g., a tf.function).
+     * </pre>
+     *
+     * <code>.tensorflow.Execution execution = 9;</code>
+     */
+    public org.tensorflow.proto.util.Execution getExecution() {
+      if (executionBuilder_ == null) {
+        if (whatCase_ == 9) {
+          return (org.tensorflow.proto.util.Execution) what_;
+        }
+        return org.tensorflow.proto.util.Execution.getDefaultInstance();
+      } else {
+        if (whatCase_ == 9) {
+          return executionBuilder_.getMessage();
+        }
+        return org.tensorflow.proto.util.Execution.getDefaultInstance();
+      }
+    }
+    /**
+     * <pre>
+     * Execution of an op or a Graph (e.g., a tf.function).
+     * </pre>
+     *
+     * <code>.tensorflow.Execution execution = 9;</code>
+     */
+    public Builder setExecution(org.tensorflow.proto.util.Execution value) {
+      if (executionBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        what_ = value;
+        onChanged();
+      } else {
+        executionBuilder_.setMessage(value);
+      }
+      whatCase_ = 9;
+      return this;
+    }
+    /**
+     * <pre>
+     * Execution of an op or a Graph (e.g., a tf.function).
+     * </pre>
+     *
+     * <code>.tensorflow.Execution execution = 9;</code>
+     */
+    public Builder setExecution(
+        org.tensorflow.proto.util.Execution.Builder builderForValue) {
+      if (executionBuilder_ == null) {
+        what_ = builderForValue.build();
+        onChanged();
+      } else {
+        executionBuilder_.setMessage(builderForValue.build());
+      }
+      whatCase_ = 9;
+      return this;
+    }
+    /**
+     * <pre>
+     * Execution of an op or a Graph (e.g., a tf.function).
+     * </pre>
+     *
+     * <code>.tensorflow.Execution execution = 9;</code>
+     */
+    public Builder mergeExecution(org.tensorflow.proto.util.Execution value) {
+      if (executionBuilder_ == null) {
+        if (whatCase_ == 9 &&
+            what_ != org.tensorflow.proto.util.Execution.getDefaultInstance()) {
+          what_ = org.tensorflow.proto.util.Execution.newBuilder((org.tensorflow.proto.util.Execution) what_)
+              .mergeFrom(value).buildPartial();
+        } else {
+          what_ = value;
+        }
+        onChanged();
+      } else {
+        if (whatCase_ == 9) {
+          executionBuilder_.mergeFrom(value);
+        }
+        executionBuilder_.setMessage(value);
+      }
+      whatCase_ = 9;
+      return this;
+    }
+    /**
+     * <pre>
+     * Execution of an op or a Graph (e.g., a tf.function).
+     * </pre>
+     *
+     * <code>.tensorflow.Execution execution = 9;</code>
+     */
+    public Builder clearExecution() {
+      if (executionBuilder_ == null) {
+        if (whatCase_ == 9) {
+          whatCase_ = 0;
+          what_ = null;
+          onChanged();
+        }
+      } else {
+        if (whatCase_ == 9) {
+          whatCase_ = 0;
+          what_ = null;
+        }
+        executionBuilder_.clear();
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Execution of an op or a Graph (e.g., a tf.function).
+     * </pre>
+     *
+     * <code>.tensorflow.Execution execution = 9;</code>
+     */
+    public org.tensorflow.proto.util.Execution.Builder getExecutionBuilder() {
+      return getExecutionFieldBuilder().getBuilder();
+    }
+    /**
+     * <pre>
+     * Execution of an op or a Graph (e.g., a tf.function).
+     * </pre>
+     *
+     * <code>.tensorflow.Execution execution = 9;</code>
+     */
+    public org.tensorflow.proto.util.ExecutionOrBuilder getExecutionOrBuilder() {
+      if ((whatCase_ == 9) && (executionBuilder_ != null)) {
+        return executionBuilder_.getMessageOrBuilder();
+      } else {
+        if (whatCase_ == 9) {
+          return (org.tensorflow.proto.util.Execution) what_;
+        }
+        return org.tensorflow.proto.util.Execution.getDefaultInstance();
+      }
+    }
+    /**
+     * <pre>
+     * Execution of an op or a Graph (e.g., a tf.function).
+     * </pre>
+     *
+     * <code>.tensorflow.Execution execution = 9;</code>
+     */
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.util.Execution, org.tensorflow.proto.util.Execution.Builder, org.tensorflow.proto.util.ExecutionOrBuilder> 
+        getExecutionFieldBuilder() {
+      if (executionBuilder_ == null) {
+        if (!(whatCase_ == 9)) {
+          what_ = org.tensorflow.proto.util.Execution.getDefaultInstance();
+        }
+        executionBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+            org.tensorflow.proto.util.Execution, org.tensorflow.proto.util.Execution.Builder, org.tensorflow.proto.util.ExecutionOrBuilder>(
+                (org.tensorflow.proto.util.Execution) what_,
+                getParentForChildren(),
+                isClean());
+        what_ = null;
+      }
+      whatCase_ = 9;
+      onChanged();;
+      return executionBuilder_;
+    }
+
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.util.GraphExecutionTrace, org.tensorflow.proto.util.GraphExecutionTrace.Builder, org.tensorflow.proto.util.GraphExecutionTraceOrBuilder> graphExecutionTraceBuilder_;
+    /**
+     * <pre>
+     * A graph execution trace: Contains information about the intermediate
+     * tensors computed during the graph execution.
+     * </pre>
+     *
+     * <code>.tensorflow.GraphExecutionTrace graph_execution_trace = 10;</code>
+     */
+    public boolean hasGraphExecutionTrace() {
+      return whatCase_ == 10;
+    }
+    /**
+     * <pre>
+     * A graph execution trace: Contains information about the intermediate
+     * tensors computed during the graph execution.
+     * </pre>
+     *
+     * <code>.tensorflow.GraphExecutionTrace graph_execution_trace = 10;</code>
+     */
+    public org.tensorflow.proto.util.GraphExecutionTrace getGraphExecutionTrace() {
+      if (graphExecutionTraceBuilder_ == null) {
+        if (whatCase_ == 10) {
+          return (org.tensorflow.proto.util.GraphExecutionTrace) what_;
+        }
+        return org.tensorflow.proto.util.GraphExecutionTrace.getDefaultInstance();
+      } else {
+        if (whatCase_ == 10) {
+          return graphExecutionTraceBuilder_.getMessage();
+        }
+        return org.tensorflow.proto.util.GraphExecutionTrace.getDefaultInstance();
+      }
+    }
+    /**
+     * <pre>
+     * A graph execution trace: Contains information about the intermediate
+     * tensors computed during the graph execution.
+     * </pre>
+     *
+     * <code>.tensorflow.GraphExecutionTrace graph_execution_trace = 10;</code>
+     */
+    public Builder setGraphExecutionTrace(org.tensorflow.proto.util.GraphExecutionTrace value) {
+      if (graphExecutionTraceBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        what_ = value;
+        onChanged();
+      } else {
+        graphExecutionTraceBuilder_.setMessage(value);
+      }
+      whatCase_ = 10;
+      return this;
+    }
+    /**
+     * <pre>
+     * A graph execution trace: Contains information about the intermediate
+     * tensors computed during the graph execution.
+     * </pre>
+     *
+     * <code>.tensorflow.GraphExecutionTrace graph_execution_trace = 10;</code>
+     */
+    public Builder setGraphExecutionTrace(
+        org.tensorflow.proto.util.GraphExecutionTrace.Builder builderForValue) {
+      if (graphExecutionTraceBuilder_ == null) {
+        what_ = builderForValue.build();
+        onChanged();
+      } else {
+        graphExecutionTraceBuilder_.setMessage(builderForValue.build());
+      }
+      whatCase_ = 10;
+      return this;
+    }
+    /**
+     * <pre>
+     * A graph execution trace: Contains information about the intermediate
+     * tensors computed during the graph execution.
+     * </pre>
+     *
+     * <code>.tensorflow.GraphExecutionTrace graph_execution_trace = 10;</code>
+     */
+    public Builder mergeGraphExecutionTrace(org.tensorflow.proto.util.GraphExecutionTrace value) {
+      if (graphExecutionTraceBuilder_ == null) {
+        if (whatCase_ == 10 &&
+            what_ != org.tensorflow.proto.util.GraphExecutionTrace.getDefaultInstance()) {
+          what_ = org.tensorflow.proto.util.GraphExecutionTrace.newBuilder((org.tensorflow.proto.util.GraphExecutionTrace) what_)
+              .mergeFrom(value).buildPartial();
+        } else {
+          what_ = value;
+        }
+        onChanged();
+      } else {
+        if (whatCase_ == 10) {
+          graphExecutionTraceBuilder_.mergeFrom(value);
+        }
+        graphExecutionTraceBuilder_.setMessage(value);
+      }
+      whatCase_ = 10;
+      return this;
+    }
+    /**
+     * <pre>
+     * A graph execution trace: Contains information about the intermediate
+     * tensors computed during the graph execution.
+     * </pre>
+     *
+     * <code>.tensorflow.GraphExecutionTrace graph_execution_trace = 10;</code>
+     */
+    public Builder clearGraphExecutionTrace() {
+      if (graphExecutionTraceBuilder_ == null) {
+        if (whatCase_ == 10) {
+          whatCase_ = 0;
+          what_ = null;
+          onChanged();
+        }
+      } else {
+        if (whatCase_ == 10) {
+          whatCase_ = 0;
+          what_ = null;
+        }
+        graphExecutionTraceBuilder_.clear();
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * A graph execution trace: Contains information about the intermediate
+     * tensors computed during the graph execution.
+     * </pre>
+     *
+     * <code>.tensorflow.GraphExecutionTrace graph_execution_trace = 10;</code>
+     */
+    public org.tensorflow.proto.util.GraphExecutionTrace.Builder getGraphExecutionTraceBuilder() {
+      return getGraphExecutionTraceFieldBuilder().getBuilder();
+    }
+    /**
+     * <pre>
+     * A graph execution trace: Contains information about the intermediate
+     * tensors computed during the graph execution.
+     * </pre>
+     *
+     * <code>.tensorflow.GraphExecutionTrace graph_execution_trace = 10;</code>
+     */
+    public org.tensorflow.proto.util.GraphExecutionTraceOrBuilder getGraphExecutionTraceOrBuilder() {
+      if ((whatCase_ == 10) && (graphExecutionTraceBuilder_ != null)) {
+        return graphExecutionTraceBuilder_.getMessageOrBuilder();
+      } else {
+        if (whatCase_ == 10) {
+          return (org.tensorflow.proto.util.GraphExecutionTrace) what_;
+        }
+        return org.tensorflow.proto.util.GraphExecutionTrace.getDefaultInstance();
+      }
+    }
+    /**
+     * <pre>
+     * A graph execution trace: Contains information about the intermediate
+     * tensors computed during the graph execution.
+     * </pre>
+     *
+     * <code>.tensorflow.GraphExecutionTrace graph_execution_trace = 10;</code>
+     */
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.util.GraphExecutionTrace, org.tensorflow.proto.util.GraphExecutionTrace.Builder, org.tensorflow.proto.util.GraphExecutionTraceOrBuilder> 
+        getGraphExecutionTraceFieldBuilder() {
+      if (graphExecutionTraceBuilder_ == null) {
+        if (!(whatCase_ == 10)) {
+          what_ = org.tensorflow.proto.util.GraphExecutionTrace.getDefaultInstance();
+        }
+        graphExecutionTraceBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+            org.tensorflow.proto.util.GraphExecutionTrace, org.tensorflow.proto.util.GraphExecutionTrace.Builder, org.tensorflow.proto.util.GraphExecutionTraceOrBuilder>(
+                (org.tensorflow.proto.util.GraphExecutionTrace) what_,
+                getParentForChildren(),
+                isClean());
+        what_ = null;
+      }
+      whatCase_ = 10;
+      onChanged();;
+      return graphExecutionTraceBuilder_;
+    }
+
+    /**
+     * <pre>
+     * The ID of the graph (i.e., FuncGraph) executed here: applicable only
+     * to the execution of a FuncGraph.
+     * </pre>
+     *
+     * <code>string graph_id = 11;</code>
+     */
+    public java.lang.String getGraphId() {
+      java.lang.Object ref = "";
+      if (whatCase_ == 11) {
+        ref = what_;
+      }
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        if (whatCase_ == 11) {
+          what_ = s;
+        }
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * The ID of the graph (i.e., FuncGraph) executed here: applicable only
+     * to the execution of a FuncGraph.
+     * </pre>
+     *
+     * <code>string graph_id = 11;</code>
+     */
+    public com.google.protobuf.ByteString
+        getGraphIdBytes() {
+      java.lang.Object ref = "";
+      if (whatCase_ == 11) {
+        ref = what_;
+      }
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        if (whatCase_ == 11) {
+          what_ = b;
+        }
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * The ID of the graph (i.e., FuncGraph) executed here: applicable only
+     * to the execution of a FuncGraph.
+     * </pre>
+     *
+     * <code>string graph_id = 11;</code>
+     */
+    public Builder setGraphId(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  whatCase_ = 11;
+      what_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * The ID of the graph (i.e., FuncGraph) executed here: applicable only
+     * to the execution of a FuncGraph.
+     * </pre>
+     *
+     * <code>string graph_id = 11;</code>
+     */
+    public Builder clearGraphId() {
+      if (whatCase_ == 11) {
+        whatCase_ = 0;
+        what_ = null;
+        onChanged();
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * The ID of the graph (i.e., FuncGraph) executed here: applicable only
+     * to the execution of a FuncGraph.
+     * </pre>
+     *
+     * <code>string graph_id = 11;</code>
+     */
+    public Builder setGraphIdBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      whatCase_ = 11;
+      what_ = value;
+      onChanged();
+      return this;
+    }
+
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.util.DebuggedDevice, org.tensorflow.proto.util.DebuggedDevice.Builder, org.tensorflow.proto.util.DebuggedDeviceOrBuilder> debuggedDeviceBuilder_;
+    /**
+     * <pre>
+     * A device on which debugger-instrumented ops and/or tensors reside.
+     * </pre>
+     *
+     * <code>.tensorflow.DebuggedDevice debugged_device = 12;</code>
+     */
+    public boolean hasDebuggedDevice() {
+      return whatCase_ == 12;
+    }
+    /**
+     * <pre>
+     * A device on which debugger-instrumented ops and/or tensors reside.
+     * </pre>
+     *
+     * <code>.tensorflow.DebuggedDevice debugged_device = 12;</code>
+     */
+    public org.tensorflow.proto.util.DebuggedDevice getDebuggedDevice() {
+      if (debuggedDeviceBuilder_ == null) {
+        if (whatCase_ == 12) {
+          return (org.tensorflow.proto.util.DebuggedDevice) what_;
+        }
+        return org.tensorflow.proto.util.DebuggedDevice.getDefaultInstance();
+      } else {
+        if (whatCase_ == 12) {
+          return debuggedDeviceBuilder_.getMessage();
+        }
+        return org.tensorflow.proto.util.DebuggedDevice.getDefaultInstance();
+      }
+    }
+    /**
+     * <pre>
+     * A device on which debugger-instrumented ops and/or tensors reside.
+     * </pre>
+     *
+     * <code>.tensorflow.DebuggedDevice debugged_device = 12;</code>
+     */
+    public Builder setDebuggedDevice(org.tensorflow.proto.util.DebuggedDevice value) {
+      if (debuggedDeviceBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        what_ = value;
+        onChanged();
+      } else {
+        debuggedDeviceBuilder_.setMessage(value);
+      }
+      whatCase_ = 12;
+      return this;
+    }
+    /**
+     * <pre>
+     * A device on which debugger-instrumented ops and/or tensors reside.
+     * </pre>
+     *
+     * <code>.tensorflow.DebuggedDevice debugged_device = 12;</code>
+     */
+    public Builder setDebuggedDevice(
+        org.tensorflow.proto.util.DebuggedDevice.Builder builderForValue) {
+      if (debuggedDeviceBuilder_ == null) {
+        what_ = builderForValue.build();
+        onChanged();
+      } else {
+        debuggedDeviceBuilder_.setMessage(builderForValue.build());
+      }
+      whatCase_ = 12;
+      return this;
+    }
+    /**
+     * <pre>
+     * A device on which debugger-instrumented ops and/or tensors reside.
+     * </pre>
+     *
+     * <code>.tensorflow.DebuggedDevice debugged_device = 12;</code>
+     */
+    public Builder mergeDebuggedDevice(org.tensorflow.proto.util.DebuggedDevice value) {
+      if (debuggedDeviceBuilder_ == null) {
+        if (whatCase_ == 12 &&
+            what_ != org.tensorflow.proto.util.DebuggedDevice.getDefaultInstance()) {
+          what_ = org.tensorflow.proto.util.DebuggedDevice.newBuilder((org.tensorflow.proto.util.DebuggedDevice) what_)
+              .mergeFrom(value).buildPartial();
+        } else {
+          what_ = value;
+        }
+        onChanged();
+      } else {
+        if (whatCase_ == 12) {
+          debuggedDeviceBuilder_.mergeFrom(value);
+        }
+        debuggedDeviceBuilder_.setMessage(value);
+      }
+      whatCase_ = 12;
+      return this;
+    }
+    /**
+     * <pre>
+     * A device on which debugger-instrumented ops and/or tensors reside.
+     * </pre>
+     *
+     * <code>.tensorflow.DebuggedDevice debugged_device = 12;</code>
+     */
+    public Builder clearDebuggedDevice() {
+      if (debuggedDeviceBuilder_ == null) {
+        if (whatCase_ == 12) {
+          whatCase_ = 0;
+          what_ = null;
+          onChanged();
+        }
+      } else {
+        if (whatCase_ == 12) {
+          whatCase_ = 0;
+          what_ = null;
+        }
+        debuggedDeviceBuilder_.clear();
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * A device on which debugger-instrumented ops and/or tensors reside.
+     * </pre>
+     *
+     * <code>.tensorflow.DebuggedDevice debugged_device = 12;</code>
+     */
+    public org.tensorflow.proto.util.DebuggedDevice.Builder getDebuggedDeviceBuilder() {
+      return getDebuggedDeviceFieldBuilder().getBuilder();
+    }
+    /**
+     * <pre>
+     * A device on which debugger-instrumented ops and/or tensors reside.
+     * </pre>
+     *
+     * <code>.tensorflow.DebuggedDevice debugged_device = 12;</code>
+     */
+    public org.tensorflow.proto.util.DebuggedDeviceOrBuilder getDebuggedDeviceOrBuilder() {
+      if ((whatCase_ == 12) && (debuggedDeviceBuilder_ != null)) {
+        return debuggedDeviceBuilder_.getMessageOrBuilder();
+      } else {
+        if (whatCase_ == 12) {
+          return (org.tensorflow.proto.util.DebuggedDevice) what_;
+        }
+        return org.tensorflow.proto.util.DebuggedDevice.getDefaultInstance();
+      }
+    }
+    /**
+     * <pre>
+     * A device on which debugger-instrumented ops and/or tensors reside.
+     * </pre>
+     *
+     * <code>.tensorflow.DebuggedDevice debugged_device = 12;</code>
+     */
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.util.DebuggedDevice, org.tensorflow.proto.util.DebuggedDevice.Builder, org.tensorflow.proto.util.DebuggedDeviceOrBuilder> 
+        getDebuggedDeviceFieldBuilder() {
+      if (debuggedDeviceBuilder_ == null) {
+        if (!(whatCase_ == 12)) {
+          what_ = org.tensorflow.proto.util.DebuggedDevice.getDefaultInstance();
+        }
+        debuggedDeviceBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+            org.tensorflow.proto.util.DebuggedDevice, org.tensorflow.proto.util.DebuggedDevice.Builder, org.tensorflow.proto.util.DebuggedDeviceOrBuilder>(
+                (org.tensorflow.proto.util.DebuggedDevice) what_,
+                getParentForChildren(),
+                isClean());
+        what_ = null;
+      }
+      whatCase_ = 12;
+      onChanged();;
+      return debuggedDeviceBuilder_;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.DebugEvent)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.DebugEvent)
+  private static final org.tensorflow.proto.util.DebugEvent DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.util.DebugEvent();
+  }
+
+  public static org.tensorflow.proto.util.DebugEvent getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<DebugEvent>
+      PARSER = new com.google.protobuf.AbstractParser<DebugEvent>() {
+    @java.lang.Override
+    public DebugEvent parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new DebugEvent(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<DebugEvent> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<DebugEvent> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.util.DebugEvent getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/DebugEventOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/DebugEventOrBuilder.java
new file mode 100644
index 00000000000..dd322c5d995
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/DebugEventOrBuilder.java
@@ -0,0 +1,258 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/debug_event.proto
+
+package org.tensorflow.proto.util;
+
+public interface DebugEventOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.DebugEvent)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <pre>
+   * Timestamp in seconds (with microsecond precision).
+   * </pre>
+   *
+   * <code>double wall_time = 1;</code>
+   */
+  double getWallTime();
+
+  /**
+   * <pre>
+   * Step of training (if available).
+   * </pre>
+   *
+   * <code>int64 step = 2;</code>
+   */
+  long getStep();
+
+  /**
+   * <pre>
+   * Metadata related to this debugging data.
+   * </pre>
+   *
+   * <code>.tensorflow.DebugMetadata debug_metadata = 3;</code>
+   */
+  boolean hasDebugMetadata();
+  /**
+   * <pre>
+   * Metadata related to this debugging data.
+   * </pre>
+   *
+   * <code>.tensorflow.DebugMetadata debug_metadata = 3;</code>
+   */
+  org.tensorflow.proto.util.DebugMetadata getDebugMetadata();
+  /**
+   * <pre>
+   * Metadata related to this debugging data.
+   * </pre>
+   *
+   * <code>.tensorflow.DebugMetadata debug_metadata = 3;</code>
+   */
+  org.tensorflow.proto.util.DebugMetadataOrBuilder getDebugMetadataOrBuilder();
+
+  /**
+   * <pre>
+   * The content of a source file.
+   * </pre>
+   *
+   * <code>.tensorflow.SourceFile source_file = 4;</code>
+   */
+  boolean hasSourceFile();
+  /**
+   * <pre>
+   * The content of a source file.
+   * </pre>
+   *
+   * <code>.tensorflow.SourceFile source_file = 4;</code>
+   */
+  org.tensorflow.proto.util.SourceFile getSourceFile();
+  /**
+   * <pre>
+   * The content of a source file.
+   * </pre>
+   *
+   * <code>.tensorflow.SourceFile source_file = 4;</code>
+   */
+  org.tensorflow.proto.util.SourceFileOrBuilder getSourceFileOrBuilder();
+
+  /**
+   * <pre>
+   * A stack frame (filename, line number and column number, function name and
+   * code string) with ID.
+   * </pre>
+   *
+   * <code>.tensorflow.StackFrameWithId stack_frame_with_id = 6;</code>
+   */
+  boolean hasStackFrameWithId();
+  /**
+   * <pre>
+   * A stack frame (filename, line number and column number, function name and
+   * code string) with ID.
+   * </pre>
+   *
+   * <code>.tensorflow.StackFrameWithId stack_frame_with_id = 6;</code>
+   */
+  org.tensorflow.proto.util.StackFrameWithId getStackFrameWithId();
+  /**
+   * <pre>
+   * A stack frame (filename, line number and column number, function name and
+   * code string) with ID.
+   * </pre>
+   *
+   * <code>.tensorflow.StackFrameWithId stack_frame_with_id = 6;</code>
+   */
+  org.tensorflow.proto.util.StackFrameWithIdOrBuilder getStackFrameWithIdOrBuilder();
+
+  /**
+   * <pre>
+   * The creation of an op within a graph (e.g., a FuncGraph compiled from
+   * a Python function).
+   * </pre>
+   *
+   * <code>.tensorflow.GraphOpCreation graph_op_creation = 7;</code>
+   */
+  boolean hasGraphOpCreation();
+  /**
+   * <pre>
+   * The creation of an op within a graph (e.g., a FuncGraph compiled from
+   * a Python function).
+   * </pre>
+   *
+   * <code>.tensorflow.GraphOpCreation graph_op_creation = 7;</code>
+   */
+  org.tensorflow.proto.util.GraphOpCreation getGraphOpCreation();
+  /**
+   * <pre>
+   * The creation of an op within a graph (e.g., a FuncGraph compiled from
+   * a Python function).
+   * </pre>
+   *
+   * <code>.tensorflow.GraphOpCreation graph_op_creation = 7;</code>
+   */
+  org.tensorflow.proto.util.GraphOpCreationOrBuilder getGraphOpCreationOrBuilder();
+
+  /**
+   * <pre>
+   * Information about a debugged graph.
+   * </pre>
+   *
+   * <code>.tensorflow.DebuggedGraph debugged_graph = 8;</code>
+   */
+  boolean hasDebuggedGraph();
+  /**
+   * <pre>
+   * Information about a debugged graph.
+   * </pre>
+   *
+   * <code>.tensorflow.DebuggedGraph debugged_graph = 8;</code>
+   */
+  org.tensorflow.proto.util.DebuggedGraph getDebuggedGraph();
+  /**
+   * <pre>
+   * Information about a debugged graph.
+   * </pre>
+   *
+   * <code>.tensorflow.DebuggedGraph debugged_graph = 8;</code>
+   */
+  org.tensorflow.proto.util.DebuggedGraphOrBuilder getDebuggedGraphOrBuilder();
+
+  /**
+   * <pre>
+   * Execution of an op or a Graph (e.g., a tf.function).
+   * </pre>
+   *
+   * <code>.tensorflow.Execution execution = 9;</code>
+   */
+  boolean hasExecution();
+  /**
+   * <pre>
+   * Execution of an op or a Graph (e.g., a tf.function).
+   * </pre>
+   *
+   * <code>.tensorflow.Execution execution = 9;</code>
+   */
+  org.tensorflow.proto.util.Execution getExecution();
+  /**
+   * <pre>
+   * Execution of an op or a Graph (e.g., a tf.function).
+   * </pre>
+   *
+   * <code>.tensorflow.Execution execution = 9;</code>
+   */
+  org.tensorflow.proto.util.ExecutionOrBuilder getExecutionOrBuilder();
+
+  /**
+   * <pre>
+   * A graph execution trace: Contains information about the intermediate
+   * tensors computed during the graph execution.
+   * </pre>
+   *
+   * <code>.tensorflow.GraphExecutionTrace graph_execution_trace = 10;</code>
+   */
+  boolean hasGraphExecutionTrace();
+  /**
+   * <pre>
+   * A graph execution trace: Contains information about the intermediate
+   * tensors computed during the graph execution.
+   * </pre>
+   *
+   * <code>.tensorflow.GraphExecutionTrace graph_execution_trace = 10;</code>
+   */
+  org.tensorflow.proto.util.GraphExecutionTrace getGraphExecutionTrace();
+  /**
+   * <pre>
+   * A graph execution trace: Contains information about the intermediate
+   * tensors computed during the graph execution.
+   * </pre>
+   *
+   * <code>.tensorflow.GraphExecutionTrace graph_execution_trace = 10;</code>
+   */
+  org.tensorflow.proto.util.GraphExecutionTraceOrBuilder getGraphExecutionTraceOrBuilder();
+
+  /**
+   * <pre>
+   * The ID of the graph (i.e., FuncGraph) executed here: applicable only
+   * to the execution of a FuncGraph.
+   * </pre>
+   *
+   * <code>string graph_id = 11;</code>
+   */
+  java.lang.String getGraphId();
+  /**
+   * <pre>
+   * The ID of the graph (i.e., FuncGraph) executed here: applicable only
+   * to the execution of a FuncGraph.
+   * </pre>
+   *
+   * <code>string graph_id = 11;</code>
+   */
+  com.google.protobuf.ByteString
+      getGraphIdBytes();
+
+  /**
+   * <pre>
+   * A device on which debugger-instrumented ops and/or tensors reside.
+   * </pre>
+   *
+   * <code>.tensorflow.DebuggedDevice debugged_device = 12;</code>
+   */
+  boolean hasDebuggedDevice();
+  /**
+   * <pre>
+   * A device on which debugger-instrumented ops and/or tensors reside.
+   * </pre>
+   *
+   * <code>.tensorflow.DebuggedDevice debugged_device = 12;</code>
+   */
+  org.tensorflow.proto.util.DebuggedDevice getDebuggedDevice();
+  /**
+   * <pre>
+   * A device on which debugger-instrumented ops and/or tensors reside.
+   * </pre>
+   *
+   * <code>.tensorflow.DebuggedDevice debugged_device = 12;</code>
+   */
+  org.tensorflow.proto.util.DebuggedDeviceOrBuilder getDebuggedDeviceOrBuilder();
+
+  public org.tensorflow.proto.util.DebugEvent.WhatCase getWhatCase();
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/DebugEventProtos.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/DebugEventProtos.java
new file mode 100644
index 00000000000..411b002619a
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/DebugEventProtos.java
@@ -0,0 +1,205 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/debug_event.proto
+
+package org.tensorflow.proto.util;
+
+public final class DebugEventProtos {
+  private DebugEventProtos() {}
+  public static void registerAllExtensions(
+      com.google.protobuf.ExtensionRegistryLite registry) {
+  }
+
+  public static void registerAllExtensions(
+      com.google.protobuf.ExtensionRegistry registry) {
+    registerAllExtensions(
+        (com.google.protobuf.ExtensionRegistryLite) registry);
+  }
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_DebugEvent_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_DebugEvent_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_DebugMetadata_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_DebugMetadata_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_SourceFile_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_SourceFile_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_StackFrameWithId_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_StackFrameWithId_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_CodeLocation_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_CodeLocation_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_GraphOpCreation_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_GraphOpCreation_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_DebuggedGraph_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_DebuggedGraph_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_DebuggedDevice_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_DebuggedDevice_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_Execution_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_Execution_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_GraphExecutionTrace_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_GraphExecutionTrace_fieldAccessorTable;
+
+  public static com.google.protobuf.Descriptors.FileDescriptor
+      getDescriptor() {
+    return descriptor;
+  }
+  private static  com.google.protobuf.Descriptors.FileDescriptor
+      descriptor;
+  static {
+    java.lang.String[] descriptorData = {
+      "\n*tensorflow/core/protobuf/debug_event.p" +
+      "roto\022\ntensorflow\032&tensorflow/core/framew" +
+      "ork/tensor.proto\032/tensorflow/core/protob" +
+      "uf/graph_debug_info.proto\"\376\003\n\nDebugEvent" +
+      "\022\021\n\twall_time\030\001 \001(\001\022\014\n\004step\030\002 \001(\003\0223\n\016deb" +
+      "ug_metadata\030\003 \001(\0132\031.tensorflow.DebugMeta" +
+      "dataH\000\022-\n\013source_file\030\004 \001(\0132\026.tensorflow" +
+      ".SourceFileH\000\022;\n\023stack_frame_with_id\030\006 \001" +
+      "(\0132\034.tensorflow.StackFrameWithIdH\000\0228\n\021gr" +
+      "aph_op_creation\030\007 \001(\0132\033.tensorflow.Graph" +
+      "OpCreationH\000\0223\n\016debugged_graph\030\010 \001(\0132\031.t" +
+      "ensorflow.DebuggedGraphH\000\022*\n\texecution\030\t" +
+      " \001(\0132\025.tensorflow.ExecutionH\000\022@\n\025graph_e" +
+      "xecution_trace\030\n \001(\0132\037.tensorflow.GraphE" +
+      "xecutionTraceH\000\022\022\n\010graph_id\030\013 \001(\tH\000\0225\n\017d" +
+      "ebugged_device\030\014 \001(\0132\032.tensorflow.Debugg" +
+      "edDeviceH\000B\006\n\004what\"W\n\rDebugMetadata\022\032\n\022t" +
+      "ensorflow_version\030\001 \001(\t\022\024\n\014file_version\030" +
+      "\002 \001(\t\022\024\n\014tfdbg_run_id\030\003 \001(\t\"A\n\nSourceFil" +
+      "e\022\021\n\tfile_path\030\001 \001(\t\022\021\n\thost_name\030\002 \001(\t\022" +
+      "\r\n\005lines\030\003 \003(\t\"]\n\020StackFrameWithId\022\n\n\002id" +
+      "\030\001 \001(\t\022=\n\rfile_line_col\030\002 \001(\0132&.tensorfl" +
+      "ow.GraphDebugInfo.FileLineCol\":\n\014CodeLoc" +
+      "ation\022\021\n\thost_name\030\001 \001(\t\022\027\n\017stack_frame_" +
+      "ids\030\002 \003(\t\"\344\001\n\017GraphOpCreation\022\017\n\007op_type" +
+      "\030\001 \001(\t\022\017\n\007op_name\030\002 \001(\t\022\022\n\ngraph_name\030\003 " +
+      "\001(\t\022\020\n\010graph_id\030\004 \001(\t\022\023\n\013device_name\030\005 \001" +
+      "(\t\022\023\n\013input_names\030\006 \003(\t\022\023\n\013num_outputs\030\007" +
+      " \001(\005\022/\n\rcode_location\030\010 \001(\0132\030.tensorflow" +
+      ".CodeLocation\022\031\n\021output_tensor_ids\030\t \003(\005" +
+      "\"\245\001\n\rDebuggedGraph\022\020\n\010graph_id\030\001 \001(\t\022\022\n\n" +
+      "graph_name\030\002 \001(\t\022\030\n\020instrumented_ops\030\003 \003" +
+      "(\t\022\032\n\022original_graph_def\030\004 \001(\014\022\036\n\026instru" +
+      "mented_graph_def\030\005 \001(\014\022\030\n\020outer_context_" +
+      "id\030\006 \001(\t\"8\n\016DebuggedDevice\022\023\n\013device_nam" +
+      "e\030\001 \001(\t\022\021\n\tdevice_id\030\002 \001(\005\"\263\002\n\tExecution" +
+      "\022\017\n\007op_type\030\001 \001(\t\022\023\n\013num_outputs\030\002 \001(\005\022\020" +
+      "\n\010graph_id\030\003 \001(\t\022\030\n\020input_tensor_ids\030\004 \003" +
+      "(\003\022\031\n\021output_tensor_ids\030\005 \003(\003\0226\n\021tensor_" +
+      "debug_mode\030\006 \001(\0162\033.tensorflow.TensorDebu" +
+      "gMode\022.\n\rtensor_protos\030\007 \003(\0132\027.tensorflo" +
+      "w.TensorProto\022/\n\rcode_location\030\010 \001(\0132\030.t" +
+      "ensorflow.CodeLocation\022 \n\030output_tensor_" +
+      "device_ids\030\t \003(\005\"\321\001\n\023GraphExecutionTrace" +
+      "\022\030\n\020tfdbg_context_id\030\001 \001(\t\022\017\n\007op_name\030\002 " +
+      "\001(\t\022\023\n\013output_slot\030\003 \001(\005\0226\n\021tensor_debug" +
+      "_mode\030\004 \001(\0162\033.tensorflow.TensorDebugMode" +
+      "\022-\n\014tensor_proto\030\005 \001(\0132\027.tensorflow.Tens" +
+      "orProto\022\023\n\013device_name\030\006 \001(\t*\266\001\n\017TensorD" +
+      "ebugMode\022\017\n\013UNSPECIFIED\020\000\022\r\n\tNO_TENSOR\020\001" +
+      "\022\017\n\013CURT_HEALTH\020\002\022\022\n\016CONCISE_HEALTH\020\003\022\017\n" +
+      "\013FULL_HEALTH\020\004\022\t\n\005SHAPE\020\005\022\021\n\rFULL_NUMERI" +
+      "CS\020\006\022\017\n\013FULL_TENSOR\020\007\022\036\n\032REDUCE_INF_NAN_" +
+      "THREE_SLOTS\020\010B|\n\031org.tensorflow.proto.ut" +
+      "ilB\020DebugEventProtosP\001ZHgithub.com/tenso" +
+      "rflow/tensorflow/tensorflow/go/core/core" +
+      "_protos_go_proto\370\001\001b\006proto3"
+    };
+    descriptor = com.google.protobuf.Descriptors.FileDescriptor
+      .internalBuildGeneratedFileFrom(descriptorData,
+        new com.google.protobuf.Descriptors.FileDescriptor[] {
+          org.tensorflow.proto.framework.TensorProtos.getDescriptor(),
+          org.tensorflow.proto.framework.GraphDebugInfoProtos.getDescriptor(),
+        });
+    internal_static_tensorflow_DebugEvent_descriptor =
+      getDescriptor().getMessageTypes().get(0);
+    internal_static_tensorflow_DebugEvent_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_DebugEvent_descriptor,
+        new java.lang.String[] { "WallTime", "Step", "DebugMetadata", "SourceFile", "StackFrameWithId", "GraphOpCreation", "DebuggedGraph", "Execution", "GraphExecutionTrace", "GraphId", "DebuggedDevice", "What", });
+    internal_static_tensorflow_DebugMetadata_descriptor =
+      getDescriptor().getMessageTypes().get(1);
+    internal_static_tensorflow_DebugMetadata_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_DebugMetadata_descriptor,
+        new java.lang.String[] { "TensorflowVersion", "FileVersion", "TfdbgRunId", });
+    internal_static_tensorflow_SourceFile_descriptor =
+      getDescriptor().getMessageTypes().get(2);
+    internal_static_tensorflow_SourceFile_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_SourceFile_descriptor,
+        new java.lang.String[] { "FilePath", "HostName", "Lines", });
+    internal_static_tensorflow_StackFrameWithId_descriptor =
+      getDescriptor().getMessageTypes().get(3);
+    internal_static_tensorflow_StackFrameWithId_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_StackFrameWithId_descriptor,
+        new java.lang.String[] { "Id", "FileLineCol", });
+    internal_static_tensorflow_CodeLocation_descriptor =
+      getDescriptor().getMessageTypes().get(4);
+    internal_static_tensorflow_CodeLocation_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_CodeLocation_descriptor,
+        new java.lang.String[] { "HostName", "StackFrameIds", });
+    internal_static_tensorflow_GraphOpCreation_descriptor =
+      getDescriptor().getMessageTypes().get(5);
+    internal_static_tensorflow_GraphOpCreation_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_GraphOpCreation_descriptor,
+        new java.lang.String[] { "OpType", "OpName", "GraphName", "GraphId", "DeviceName", "InputNames", "NumOutputs", "CodeLocation", "OutputTensorIds", });
+    internal_static_tensorflow_DebuggedGraph_descriptor =
+      getDescriptor().getMessageTypes().get(6);
+    internal_static_tensorflow_DebuggedGraph_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_DebuggedGraph_descriptor,
+        new java.lang.String[] { "GraphId", "GraphName", "InstrumentedOps", "OriginalGraphDef", "InstrumentedGraphDef", "OuterContextId", });
+    internal_static_tensorflow_DebuggedDevice_descriptor =
+      getDescriptor().getMessageTypes().get(7);
+    internal_static_tensorflow_DebuggedDevice_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_DebuggedDevice_descriptor,
+        new java.lang.String[] { "DeviceName", "DeviceId", });
+    internal_static_tensorflow_Execution_descriptor =
+      getDescriptor().getMessageTypes().get(8);
+    internal_static_tensorflow_Execution_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_Execution_descriptor,
+        new java.lang.String[] { "OpType", "NumOutputs", "GraphId", "InputTensorIds", "OutputTensorIds", "TensorDebugMode", "TensorProtos", "CodeLocation", "OutputTensorDeviceIds", });
+    internal_static_tensorflow_GraphExecutionTrace_descriptor =
+      getDescriptor().getMessageTypes().get(9);
+    internal_static_tensorflow_GraphExecutionTrace_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_GraphExecutionTrace_descriptor,
+        new java.lang.String[] { "TfdbgContextId", "OpName", "OutputSlot", "TensorDebugMode", "TensorProto", "DeviceName", });
+    org.tensorflow.proto.framework.TensorProtos.getDescriptor();
+    org.tensorflow.proto.framework.GraphDebugInfoProtos.getDescriptor();
+  }
+
+  // @@protoc_insertion_point(outer_class_scope)
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/DebugMetadata.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/DebugMetadata.java
new file mode 100644
index 00000000000..e0679182c6e
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/DebugMetadata.java
@@ -0,0 +1,920 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/debug_event.proto
+
+package org.tensorflow.proto.util;
+
+/**
+ * <pre>
+ * Metadata about the debugger and the debugged TensorFlow program.
+ * </pre>
+ *
+ * Protobuf type {@code tensorflow.DebugMetadata}
+ */
+public  final class DebugMetadata extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.DebugMetadata)
+    DebugMetadataOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use DebugMetadata.newBuilder() to construct.
+  private DebugMetadata(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private DebugMetadata() {
+    tensorflowVersion_ = "";
+    fileVersion_ = "";
+    tfdbgRunId_ = "";
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new DebugMetadata();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private DebugMetadata(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 10: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            tensorflowVersion_ = s;
+            break;
+          }
+          case 18: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            fileVersion_ = s;
+            break;
+          }
+          case 26: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            tfdbgRunId_ = s;
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.util.DebugEventProtos.internal_static_tensorflow_DebugMetadata_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.util.DebugEventProtos.internal_static_tensorflow_DebugMetadata_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.util.DebugMetadata.class, org.tensorflow.proto.util.DebugMetadata.Builder.class);
+  }
+
+  public static final int TENSORFLOW_VERSION_FIELD_NUMBER = 1;
+  private volatile java.lang.Object tensorflowVersion_;
+  /**
+   * <pre>
+   * Version of TensorFlow.
+   * </pre>
+   *
+   * <code>string tensorflow_version = 1;</code>
+   */
+  public java.lang.String getTensorflowVersion() {
+    java.lang.Object ref = tensorflowVersion_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      tensorflowVersion_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * Version of TensorFlow.
+   * </pre>
+   *
+   * <code>string tensorflow_version = 1;</code>
+   */
+  public com.google.protobuf.ByteString
+      getTensorflowVersionBytes() {
+    java.lang.Object ref = tensorflowVersion_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      tensorflowVersion_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int FILE_VERSION_FIELD_NUMBER = 2;
+  private volatile java.lang.Object fileVersion_;
+  /**
+   * <pre>
+   * Version of the DebugEvent file format.
+   * Has a format of "debug.Event:&lt;number&gt;", e.g., "debug.Event:1".
+   * </pre>
+   *
+   * <code>string file_version = 2;</code>
+   */
+  public java.lang.String getFileVersion() {
+    java.lang.Object ref = fileVersion_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      fileVersion_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * Version of the DebugEvent file format.
+   * Has a format of "debug.Event:&lt;number&gt;", e.g., "debug.Event:1".
+   * </pre>
+   *
+   * <code>string file_version = 2;</code>
+   */
+  public com.google.protobuf.ByteString
+      getFileVersionBytes() {
+    java.lang.Object ref = fileVersion_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      fileVersion_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int TFDBG_RUN_ID_FIELD_NUMBER = 3;
+  private volatile java.lang.Object tfdbgRunId_;
+  /**
+   * <pre>
+   * A unique ID for the current run of tfdbg.
+   * A run of tfdbg is defined as a TensorFlow job instrumented by tfdbg.
+   * Multiple hosts in a distributed TensorFlow job instrumented by tfdbg
+   * have the same ID.
+   * </pre>
+   *
+   * <code>string tfdbg_run_id = 3;</code>
+   */
+  public java.lang.String getTfdbgRunId() {
+    java.lang.Object ref = tfdbgRunId_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      tfdbgRunId_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * A unique ID for the current run of tfdbg.
+   * A run of tfdbg is defined as a TensorFlow job instrumented by tfdbg.
+   * Multiple hosts in a distributed TensorFlow job instrumented by tfdbg
+   * have the same ID.
+   * </pre>
+   *
+   * <code>string tfdbg_run_id = 3;</code>
+   */
+  public com.google.protobuf.ByteString
+      getTfdbgRunIdBytes() {
+    java.lang.Object ref = tfdbgRunId_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      tfdbgRunId_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    if (!getTensorflowVersionBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 1, tensorflowVersion_);
+    }
+    if (!getFileVersionBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 2, fileVersion_);
+    }
+    if (!getTfdbgRunIdBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 3, tfdbgRunId_);
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (!getTensorflowVersionBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, tensorflowVersion_);
+    }
+    if (!getFileVersionBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(2, fileVersion_);
+    }
+    if (!getTfdbgRunIdBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(3, tfdbgRunId_);
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.util.DebugMetadata)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.util.DebugMetadata other = (org.tensorflow.proto.util.DebugMetadata) obj;
+
+    if (!getTensorflowVersion()
+        .equals(other.getTensorflowVersion())) return false;
+    if (!getFileVersion()
+        .equals(other.getFileVersion())) return false;
+    if (!getTfdbgRunId()
+        .equals(other.getTfdbgRunId())) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    hash = (37 * hash) + TENSORFLOW_VERSION_FIELD_NUMBER;
+    hash = (53 * hash) + getTensorflowVersion().hashCode();
+    hash = (37 * hash) + FILE_VERSION_FIELD_NUMBER;
+    hash = (53 * hash) + getFileVersion().hashCode();
+    hash = (37 * hash) + TFDBG_RUN_ID_FIELD_NUMBER;
+    hash = (53 * hash) + getTfdbgRunId().hashCode();
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.util.DebugMetadata parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.DebugMetadata parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.DebugMetadata parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.DebugMetadata parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.DebugMetadata parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.DebugMetadata parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.DebugMetadata parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.DebugMetadata parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.DebugMetadata parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.DebugMetadata parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.DebugMetadata parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.DebugMetadata parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.util.DebugMetadata prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * <pre>
+   * Metadata about the debugger and the debugged TensorFlow program.
+   * </pre>
+   *
+   * Protobuf type {@code tensorflow.DebugMetadata}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.DebugMetadata)
+      org.tensorflow.proto.util.DebugMetadataOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.util.DebugEventProtos.internal_static_tensorflow_DebugMetadata_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.util.DebugEventProtos.internal_static_tensorflow_DebugMetadata_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.util.DebugMetadata.class, org.tensorflow.proto.util.DebugMetadata.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.util.DebugMetadata.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      tensorflowVersion_ = "";
+
+      fileVersion_ = "";
+
+      tfdbgRunId_ = "";
+
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.util.DebugEventProtos.internal_static_tensorflow_DebugMetadata_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.DebugMetadata getDefaultInstanceForType() {
+      return org.tensorflow.proto.util.DebugMetadata.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.DebugMetadata build() {
+      org.tensorflow.proto.util.DebugMetadata result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.DebugMetadata buildPartial() {
+      org.tensorflow.proto.util.DebugMetadata result = new org.tensorflow.proto.util.DebugMetadata(this);
+      result.tensorflowVersion_ = tensorflowVersion_;
+      result.fileVersion_ = fileVersion_;
+      result.tfdbgRunId_ = tfdbgRunId_;
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.util.DebugMetadata) {
+        return mergeFrom((org.tensorflow.proto.util.DebugMetadata)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.util.DebugMetadata other) {
+      if (other == org.tensorflow.proto.util.DebugMetadata.getDefaultInstance()) return this;
+      if (!other.getTensorflowVersion().isEmpty()) {
+        tensorflowVersion_ = other.tensorflowVersion_;
+        onChanged();
+      }
+      if (!other.getFileVersion().isEmpty()) {
+        fileVersion_ = other.fileVersion_;
+        onChanged();
+      }
+      if (!other.getTfdbgRunId().isEmpty()) {
+        tfdbgRunId_ = other.tfdbgRunId_;
+        onChanged();
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.util.DebugMetadata parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.util.DebugMetadata) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+
+    private java.lang.Object tensorflowVersion_ = "";
+    /**
+     * <pre>
+     * Version of TensorFlow.
+     * </pre>
+     *
+     * <code>string tensorflow_version = 1;</code>
+     */
+    public java.lang.String getTensorflowVersion() {
+      java.lang.Object ref = tensorflowVersion_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        tensorflowVersion_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Version of TensorFlow.
+     * </pre>
+     *
+     * <code>string tensorflow_version = 1;</code>
+     */
+    public com.google.protobuf.ByteString
+        getTensorflowVersionBytes() {
+      java.lang.Object ref = tensorflowVersion_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        tensorflowVersion_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Version of TensorFlow.
+     * </pre>
+     *
+     * <code>string tensorflow_version = 1;</code>
+     */
+    public Builder setTensorflowVersion(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      tensorflowVersion_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Version of TensorFlow.
+     * </pre>
+     *
+     * <code>string tensorflow_version = 1;</code>
+     */
+    public Builder clearTensorflowVersion() {
+      
+      tensorflowVersion_ = getDefaultInstance().getTensorflowVersion();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Version of TensorFlow.
+     * </pre>
+     *
+     * <code>string tensorflow_version = 1;</code>
+     */
+    public Builder setTensorflowVersionBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      tensorflowVersion_ = value;
+      onChanged();
+      return this;
+    }
+
+    private java.lang.Object fileVersion_ = "";
+    /**
+     * <pre>
+     * Version of the DebugEvent file format.
+     * Has a format of "debug.Event:&lt;number&gt;", e.g., "debug.Event:1".
+     * </pre>
+     *
+     * <code>string file_version = 2;</code>
+     */
+    public java.lang.String getFileVersion() {
+      java.lang.Object ref = fileVersion_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        fileVersion_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Version of the DebugEvent file format.
+     * Has a format of "debug.Event:&lt;number&gt;", e.g., "debug.Event:1".
+     * </pre>
+     *
+     * <code>string file_version = 2;</code>
+     */
+    public com.google.protobuf.ByteString
+        getFileVersionBytes() {
+      java.lang.Object ref = fileVersion_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        fileVersion_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Version of the DebugEvent file format.
+     * Has a format of "debug.Event:&lt;number&gt;", e.g., "debug.Event:1".
+     * </pre>
+     *
+     * <code>string file_version = 2;</code>
+     */
+    public Builder setFileVersion(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      fileVersion_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Version of the DebugEvent file format.
+     * Has a format of "debug.Event:&lt;number&gt;", e.g., "debug.Event:1".
+     * </pre>
+     *
+     * <code>string file_version = 2;</code>
+     */
+    public Builder clearFileVersion() {
+      
+      fileVersion_ = getDefaultInstance().getFileVersion();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Version of the DebugEvent file format.
+     * Has a format of "debug.Event:&lt;number&gt;", e.g., "debug.Event:1".
+     * </pre>
+     *
+     * <code>string file_version = 2;</code>
+     */
+    public Builder setFileVersionBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      fileVersion_ = value;
+      onChanged();
+      return this;
+    }
+
+    private java.lang.Object tfdbgRunId_ = "";
+    /**
+     * <pre>
+     * A unique ID for the current run of tfdbg.
+     * A run of tfdbg is defined as a TensorFlow job instrumented by tfdbg.
+     * Multiple hosts in a distributed TensorFlow job instrumented by tfdbg
+     * have the same ID.
+     * </pre>
+     *
+     * <code>string tfdbg_run_id = 3;</code>
+     */
+    public java.lang.String getTfdbgRunId() {
+      java.lang.Object ref = tfdbgRunId_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        tfdbgRunId_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * A unique ID for the current run of tfdbg.
+     * A run of tfdbg is defined as a TensorFlow job instrumented by tfdbg.
+     * Multiple hosts in a distributed TensorFlow job instrumented by tfdbg
+     * have the same ID.
+     * </pre>
+     *
+     * <code>string tfdbg_run_id = 3;</code>
+     */
+    public com.google.protobuf.ByteString
+        getTfdbgRunIdBytes() {
+      java.lang.Object ref = tfdbgRunId_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        tfdbgRunId_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * A unique ID for the current run of tfdbg.
+     * A run of tfdbg is defined as a TensorFlow job instrumented by tfdbg.
+     * Multiple hosts in a distributed TensorFlow job instrumented by tfdbg
+     * have the same ID.
+     * </pre>
+     *
+     * <code>string tfdbg_run_id = 3;</code>
+     */
+    public Builder setTfdbgRunId(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      tfdbgRunId_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * A unique ID for the current run of tfdbg.
+     * A run of tfdbg is defined as a TensorFlow job instrumented by tfdbg.
+     * Multiple hosts in a distributed TensorFlow job instrumented by tfdbg
+     * have the same ID.
+     * </pre>
+     *
+     * <code>string tfdbg_run_id = 3;</code>
+     */
+    public Builder clearTfdbgRunId() {
+      
+      tfdbgRunId_ = getDefaultInstance().getTfdbgRunId();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * A unique ID for the current run of tfdbg.
+     * A run of tfdbg is defined as a TensorFlow job instrumented by tfdbg.
+     * Multiple hosts in a distributed TensorFlow job instrumented by tfdbg
+     * have the same ID.
+     * </pre>
+     *
+     * <code>string tfdbg_run_id = 3;</code>
+     */
+    public Builder setTfdbgRunIdBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      tfdbgRunId_ = value;
+      onChanged();
+      return this;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.DebugMetadata)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.DebugMetadata)
+  private static final org.tensorflow.proto.util.DebugMetadata DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.util.DebugMetadata();
+  }
+
+  public static org.tensorflow.proto.util.DebugMetadata getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<DebugMetadata>
+      PARSER = new com.google.protobuf.AbstractParser<DebugMetadata>() {
+    @java.lang.Override
+    public DebugMetadata parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new DebugMetadata(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<DebugMetadata> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<DebugMetadata> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.util.DebugMetadata getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/DebugMetadataOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/DebugMetadataOrBuilder.java
new file mode 100644
index 00000000000..91015599f7d
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/DebugMetadataOrBuilder.java
@@ -0,0 +1,71 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/debug_event.proto
+
+package org.tensorflow.proto.util;
+
+public interface DebugMetadataOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.DebugMetadata)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <pre>
+   * Version of TensorFlow.
+   * </pre>
+   *
+   * <code>string tensorflow_version = 1;</code>
+   */
+  java.lang.String getTensorflowVersion();
+  /**
+   * <pre>
+   * Version of TensorFlow.
+   * </pre>
+   *
+   * <code>string tensorflow_version = 1;</code>
+   */
+  com.google.protobuf.ByteString
+      getTensorflowVersionBytes();
+
+  /**
+   * <pre>
+   * Version of the DebugEvent file format.
+   * Has a format of "debug.Event:&lt;number&gt;", e.g., "debug.Event:1".
+   * </pre>
+   *
+   * <code>string file_version = 2;</code>
+   */
+  java.lang.String getFileVersion();
+  /**
+   * <pre>
+   * Version of the DebugEvent file format.
+   * Has a format of "debug.Event:&lt;number&gt;", e.g., "debug.Event:1".
+   * </pre>
+   *
+   * <code>string file_version = 2;</code>
+   */
+  com.google.protobuf.ByteString
+      getFileVersionBytes();
+
+  /**
+   * <pre>
+   * A unique ID for the current run of tfdbg.
+   * A run of tfdbg is defined as a TensorFlow job instrumented by tfdbg.
+   * Multiple hosts in a distributed TensorFlow job instrumented by tfdbg
+   * have the same ID.
+   * </pre>
+   *
+   * <code>string tfdbg_run_id = 3;</code>
+   */
+  java.lang.String getTfdbgRunId();
+  /**
+   * <pre>
+   * A unique ID for the current run of tfdbg.
+   * A run of tfdbg is defined as a TensorFlow job instrumented by tfdbg.
+   * Multiple hosts in a distributed TensorFlow job instrumented by tfdbg
+   * have the same ID.
+   * </pre>
+   *
+   * <code>string tfdbg_run_id = 3;</code>
+   */
+  com.google.protobuf.ByteString
+      getTfdbgRunIdBytes();
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/DebuggedDevice.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/DebuggedDevice.java
new file mode 100644
index 00000000000..18f528481cd
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/DebuggedDevice.java
@@ -0,0 +1,667 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/debug_event.proto
+
+package org.tensorflow.proto.util;
+
+/**
+ * <pre>
+ * A device on which ops and/or tensors are instrumented by the debugger.
+ * </pre>
+ *
+ * Protobuf type {@code tensorflow.DebuggedDevice}
+ */
+public  final class DebuggedDevice extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.DebuggedDevice)
+    DebuggedDeviceOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use DebuggedDevice.newBuilder() to construct.
+  private DebuggedDevice(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private DebuggedDevice() {
+    deviceName_ = "";
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new DebuggedDevice();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private DebuggedDevice(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 10: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            deviceName_ = s;
+            break;
+          }
+          case 16: {
+
+            deviceId_ = input.readInt32();
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.util.DebugEventProtos.internal_static_tensorflow_DebuggedDevice_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.util.DebugEventProtos.internal_static_tensorflow_DebuggedDevice_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.util.DebuggedDevice.class, org.tensorflow.proto.util.DebuggedDevice.Builder.class);
+  }
+
+  public static final int DEVICE_NAME_FIELD_NUMBER = 1;
+  private volatile java.lang.Object deviceName_;
+  /**
+   * <pre>
+   * Name of the device.
+   * </pre>
+   *
+   * <code>string device_name = 1;</code>
+   */
+  public java.lang.String getDeviceName() {
+    java.lang.Object ref = deviceName_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      deviceName_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * Name of the device.
+   * </pre>
+   *
+   * <code>string device_name = 1;</code>
+   */
+  public com.google.protobuf.ByteString
+      getDeviceNameBytes() {
+    java.lang.Object ref = deviceName_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      deviceName_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int DEVICE_ID_FIELD_NUMBER = 2;
+  private int deviceId_;
+  /**
+   * <pre>
+   * A debugger-generated ID for the device. Guaranteed to be unique within
+   * the scope of the debugged TensorFlow program, including single-host and
+   * multi-host settings.
+   * TODO(cais): Test the uniqueness guarantee in multi-host settings.
+   * </pre>
+   *
+   * <code>int32 device_id = 2;</code>
+   */
+  public int getDeviceId() {
+    return deviceId_;
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    if (!getDeviceNameBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 1, deviceName_);
+    }
+    if (deviceId_ != 0) {
+      output.writeInt32(2, deviceId_);
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (!getDeviceNameBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, deviceName_);
+    }
+    if (deviceId_ != 0) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt32Size(2, deviceId_);
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.util.DebuggedDevice)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.util.DebuggedDevice other = (org.tensorflow.proto.util.DebuggedDevice) obj;
+
+    if (!getDeviceName()
+        .equals(other.getDeviceName())) return false;
+    if (getDeviceId()
+        != other.getDeviceId()) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    hash = (37 * hash) + DEVICE_NAME_FIELD_NUMBER;
+    hash = (53 * hash) + getDeviceName().hashCode();
+    hash = (37 * hash) + DEVICE_ID_FIELD_NUMBER;
+    hash = (53 * hash) + getDeviceId();
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.util.DebuggedDevice parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.DebuggedDevice parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.DebuggedDevice parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.DebuggedDevice parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.DebuggedDevice parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.DebuggedDevice parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.DebuggedDevice parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.DebuggedDevice parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.DebuggedDevice parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.DebuggedDevice parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.DebuggedDevice parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.DebuggedDevice parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.util.DebuggedDevice prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * <pre>
+   * A device on which ops and/or tensors are instrumented by the debugger.
+   * </pre>
+   *
+   * Protobuf type {@code tensorflow.DebuggedDevice}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.DebuggedDevice)
+      org.tensorflow.proto.util.DebuggedDeviceOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.util.DebugEventProtos.internal_static_tensorflow_DebuggedDevice_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.util.DebugEventProtos.internal_static_tensorflow_DebuggedDevice_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.util.DebuggedDevice.class, org.tensorflow.proto.util.DebuggedDevice.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.util.DebuggedDevice.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      deviceName_ = "";
+
+      deviceId_ = 0;
+
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.util.DebugEventProtos.internal_static_tensorflow_DebuggedDevice_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.DebuggedDevice getDefaultInstanceForType() {
+      return org.tensorflow.proto.util.DebuggedDevice.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.DebuggedDevice build() {
+      org.tensorflow.proto.util.DebuggedDevice result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.DebuggedDevice buildPartial() {
+      org.tensorflow.proto.util.DebuggedDevice result = new org.tensorflow.proto.util.DebuggedDevice(this);
+      result.deviceName_ = deviceName_;
+      result.deviceId_ = deviceId_;
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.util.DebuggedDevice) {
+        return mergeFrom((org.tensorflow.proto.util.DebuggedDevice)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.util.DebuggedDevice other) {
+      if (other == org.tensorflow.proto.util.DebuggedDevice.getDefaultInstance()) return this;
+      if (!other.getDeviceName().isEmpty()) {
+        deviceName_ = other.deviceName_;
+        onChanged();
+      }
+      if (other.getDeviceId() != 0) {
+        setDeviceId(other.getDeviceId());
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.util.DebuggedDevice parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.util.DebuggedDevice) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+
+    private java.lang.Object deviceName_ = "";
+    /**
+     * <pre>
+     * Name of the device.
+     * </pre>
+     *
+     * <code>string device_name = 1;</code>
+     */
+    public java.lang.String getDeviceName() {
+      java.lang.Object ref = deviceName_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        deviceName_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Name of the device.
+     * </pre>
+     *
+     * <code>string device_name = 1;</code>
+     */
+    public com.google.protobuf.ByteString
+        getDeviceNameBytes() {
+      java.lang.Object ref = deviceName_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        deviceName_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Name of the device.
+     * </pre>
+     *
+     * <code>string device_name = 1;</code>
+     */
+    public Builder setDeviceName(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      deviceName_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Name of the device.
+     * </pre>
+     *
+     * <code>string device_name = 1;</code>
+     */
+    public Builder clearDeviceName() {
+      
+      deviceName_ = getDefaultInstance().getDeviceName();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Name of the device.
+     * </pre>
+     *
+     * <code>string device_name = 1;</code>
+     */
+    public Builder setDeviceNameBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      deviceName_ = value;
+      onChanged();
+      return this;
+    }
+
+    private int deviceId_ ;
+    /**
+     * <pre>
+     * A debugger-generated ID for the device. Guaranteed to be unique within
+     * the scope of the debugged TensorFlow program, including single-host and
+     * multi-host settings.
+     * TODO(cais): Test the uniqueness guarantee in multi-host settings.
+     * </pre>
+     *
+     * <code>int32 device_id = 2;</code>
+     */
+    public int getDeviceId() {
+      return deviceId_;
+    }
+    /**
+     * <pre>
+     * A debugger-generated ID for the device. Guaranteed to be unique within
+     * the scope of the debugged TensorFlow program, including single-host and
+     * multi-host settings.
+     * TODO(cais): Test the uniqueness guarantee in multi-host settings.
+     * </pre>
+     *
+     * <code>int32 device_id = 2;</code>
+     */
+    public Builder setDeviceId(int value) {
+      
+      deviceId_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * A debugger-generated ID for the device. Guaranteed to be unique within
+     * the scope of the debugged TensorFlow program, including single-host and
+     * multi-host settings.
+     * TODO(cais): Test the uniqueness guarantee in multi-host settings.
+     * </pre>
+     *
+     * <code>int32 device_id = 2;</code>
+     */
+    public Builder clearDeviceId() {
+      
+      deviceId_ = 0;
+      onChanged();
+      return this;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.DebuggedDevice)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.DebuggedDevice)
+  private static final org.tensorflow.proto.util.DebuggedDevice DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.util.DebuggedDevice();
+  }
+
+  public static org.tensorflow.proto.util.DebuggedDevice getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<DebuggedDevice>
+      PARSER = new com.google.protobuf.AbstractParser<DebuggedDevice>() {
+    @java.lang.Override
+    public DebuggedDevice parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new DebuggedDevice(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<DebuggedDevice> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<DebuggedDevice> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.util.DebuggedDevice getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/DebuggedDeviceOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/DebuggedDeviceOrBuilder.java
new file mode 100644
index 00000000000..3915849b3ae
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/DebuggedDeviceOrBuilder.java
@@ -0,0 +1,39 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/debug_event.proto
+
+package org.tensorflow.proto.util;
+
+public interface DebuggedDeviceOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.DebuggedDevice)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <pre>
+   * Name of the device.
+   * </pre>
+   *
+   * <code>string device_name = 1;</code>
+   */
+  java.lang.String getDeviceName();
+  /**
+   * <pre>
+   * Name of the device.
+   * </pre>
+   *
+   * <code>string device_name = 1;</code>
+   */
+  com.google.protobuf.ByteString
+      getDeviceNameBytes();
+
+  /**
+   * <pre>
+   * A debugger-generated ID for the device. Guaranteed to be unique within
+   * the scope of the debugged TensorFlow program, including single-host and
+   * multi-host settings.
+   * TODO(cais): Test the uniqueness guarantee in multi-host settings.
+   * </pre>
+   *
+   * <code>int32 device_id = 2;</code>
+   */
+  int getDeviceId();
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/DebuggedGraph.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/DebuggedGraph.java
new file mode 100644
index 00000000000..6fa3e5adcff
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/DebuggedGraph.java
@@ -0,0 +1,1295 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/debug_event.proto
+
+package org.tensorflow.proto.util;
+
+/**
+ * <pre>
+ * A debugger-instrumented graph.
+ * </pre>
+ *
+ * Protobuf type {@code tensorflow.DebuggedGraph}
+ */
+public  final class DebuggedGraph extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.DebuggedGraph)
+    DebuggedGraphOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use DebuggedGraph.newBuilder() to construct.
+  private DebuggedGraph(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private DebuggedGraph() {
+    graphId_ = "";
+    graphName_ = "";
+    instrumentedOps_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+    originalGraphDef_ = com.google.protobuf.ByteString.EMPTY;
+    instrumentedGraphDef_ = com.google.protobuf.ByteString.EMPTY;
+    outerContextId_ = "";
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new DebuggedGraph();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private DebuggedGraph(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    int mutable_bitField0_ = 0;
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 10: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            graphId_ = s;
+            break;
+          }
+          case 18: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            graphName_ = s;
+            break;
+          }
+          case 26: {
+            java.lang.String s = input.readStringRequireUtf8();
+            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
+              instrumentedOps_ = new com.google.protobuf.LazyStringArrayList();
+              mutable_bitField0_ |= 0x00000001;
+            }
+            instrumentedOps_.add(s);
+            break;
+          }
+          case 34: {
+
+            originalGraphDef_ = input.readBytes();
+            break;
+          }
+          case 42: {
+
+            instrumentedGraphDef_ = input.readBytes();
+            break;
+          }
+          case 50: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            outerContextId_ = s;
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      if (((mutable_bitField0_ & 0x00000001) != 0)) {
+        instrumentedOps_ = instrumentedOps_.getUnmodifiableView();
+      }
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.util.DebugEventProtos.internal_static_tensorflow_DebuggedGraph_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.util.DebugEventProtos.internal_static_tensorflow_DebuggedGraph_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.util.DebuggedGraph.class, org.tensorflow.proto.util.DebuggedGraph.Builder.class);
+  }
+
+  public static final int GRAPH_ID_FIELD_NUMBER = 1;
+  private volatile java.lang.Object graphId_;
+  /**
+   * <pre>
+   * An ID for the graph.
+   * This can be used up to look up graph names. Generated by the debugger.
+   * </pre>
+   *
+   * <code>string graph_id = 1;</code>
+   */
+  public java.lang.String getGraphId() {
+    java.lang.Object ref = graphId_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      graphId_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * An ID for the graph.
+   * This can be used up to look up graph names. Generated by the debugger.
+   * </pre>
+   *
+   * <code>string graph_id = 1;</code>
+   */
+  public com.google.protobuf.ByteString
+      getGraphIdBytes() {
+    java.lang.Object ref = graphId_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      graphId_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int GRAPH_NAME_FIELD_NUMBER = 2;
+  private volatile java.lang.Object graphName_;
+  /**
+   * <pre>
+   * Name of the graph (if available).
+   * </pre>
+   *
+   * <code>string graph_name = 2;</code>
+   */
+  public java.lang.String getGraphName() {
+    java.lang.Object ref = graphName_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      graphName_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * Name of the graph (if available).
+   * </pre>
+   *
+   * <code>string graph_name = 2;</code>
+   */
+  public com.google.protobuf.ByteString
+      getGraphNameBytes() {
+    java.lang.Object ref = graphName_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      graphName_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int INSTRUMENTED_OPS_FIELD_NUMBER = 3;
+  private com.google.protobuf.LazyStringList instrumentedOps_;
+  /**
+   * <pre>
+   * Names of the instrumented ops. This can be used to look up op name
+   * based on the numeric-summary tensors (2nd column).
+   * </pre>
+   *
+   * <code>repeated string instrumented_ops = 3;</code>
+   */
+  public com.google.protobuf.ProtocolStringList
+      getInstrumentedOpsList() {
+    return instrumentedOps_;
+  }
+  /**
+   * <pre>
+   * Names of the instrumented ops. This can be used to look up op name
+   * based on the numeric-summary tensors (2nd column).
+   * </pre>
+   *
+   * <code>repeated string instrumented_ops = 3;</code>
+   */
+  public int getInstrumentedOpsCount() {
+    return instrumentedOps_.size();
+  }
+  /**
+   * <pre>
+   * Names of the instrumented ops. This can be used to look up op name
+   * based on the numeric-summary tensors (2nd column).
+   * </pre>
+   *
+   * <code>repeated string instrumented_ops = 3;</code>
+   */
+  public java.lang.String getInstrumentedOps(int index) {
+    return instrumentedOps_.get(index);
+  }
+  /**
+   * <pre>
+   * Names of the instrumented ops. This can be used to look up op name
+   * based on the numeric-summary tensors (2nd column).
+   * </pre>
+   *
+   * <code>repeated string instrumented_ops = 3;</code>
+   */
+  public com.google.protobuf.ByteString
+      getInstrumentedOpsBytes(int index) {
+    return instrumentedOps_.getByteString(index);
+  }
+
+  public static final int ORIGINAL_GRAPH_DEF_FIELD_NUMBER = 4;
+  private com.google.protobuf.ByteString originalGraphDef_;
+  /**
+   * <pre>
+   * Original (uninstrumented) GraphDef (if available).
+   * </pre>
+   *
+   * <code>bytes original_graph_def = 4;</code>
+   */
+  public com.google.protobuf.ByteString getOriginalGraphDef() {
+    return originalGraphDef_;
+  }
+
+  public static final int INSTRUMENTED_GRAPH_DEF_FIELD_NUMBER = 5;
+  private com.google.protobuf.ByteString instrumentedGraphDef_;
+  /**
+   * <pre>
+   * An encoded version of a GraphDef.
+   * This graph may include the debugger-inserted ops.
+   * </pre>
+   *
+   * <code>bytes instrumented_graph_def = 5;</code>
+   */
+  public com.google.protobuf.ByteString getInstrumentedGraphDef() {
+    return instrumentedGraphDef_;
+  }
+
+  public static final int OUTER_CONTEXT_ID_FIELD_NUMBER = 6;
+  private volatile java.lang.Object outerContextId_;
+  /**
+   * <pre>
+   * IDs of the immediate enclosing context (graph), if any.
+   * </pre>
+   *
+   * <code>string outer_context_id = 6;</code>
+   */
+  public java.lang.String getOuterContextId() {
+    java.lang.Object ref = outerContextId_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      outerContextId_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * IDs of the immediate enclosing context (graph), if any.
+   * </pre>
+   *
+   * <code>string outer_context_id = 6;</code>
+   */
+  public com.google.protobuf.ByteString
+      getOuterContextIdBytes() {
+    java.lang.Object ref = outerContextId_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      outerContextId_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    if (!getGraphIdBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 1, graphId_);
+    }
+    if (!getGraphNameBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 2, graphName_);
+    }
+    for (int i = 0; i < instrumentedOps_.size(); i++) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 3, instrumentedOps_.getRaw(i));
+    }
+    if (!originalGraphDef_.isEmpty()) {
+      output.writeBytes(4, originalGraphDef_);
+    }
+    if (!instrumentedGraphDef_.isEmpty()) {
+      output.writeBytes(5, instrumentedGraphDef_);
+    }
+    if (!getOuterContextIdBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 6, outerContextId_);
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (!getGraphIdBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, graphId_);
+    }
+    if (!getGraphNameBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(2, graphName_);
+    }
+    {
+      int dataSize = 0;
+      for (int i = 0; i < instrumentedOps_.size(); i++) {
+        dataSize += computeStringSizeNoTag(instrumentedOps_.getRaw(i));
+      }
+      size += dataSize;
+      size += 1 * getInstrumentedOpsList().size();
+    }
+    if (!originalGraphDef_.isEmpty()) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeBytesSize(4, originalGraphDef_);
+    }
+    if (!instrumentedGraphDef_.isEmpty()) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeBytesSize(5, instrumentedGraphDef_);
+    }
+    if (!getOuterContextIdBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(6, outerContextId_);
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.util.DebuggedGraph)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.util.DebuggedGraph other = (org.tensorflow.proto.util.DebuggedGraph) obj;
+
+    if (!getGraphId()
+        .equals(other.getGraphId())) return false;
+    if (!getGraphName()
+        .equals(other.getGraphName())) return false;
+    if (!getInstrumentedOpsList()
+        .equals(other.getInstrumentedOpsList())) return false;
+    if (!getOriginalGraphDef()
+        .equals(other.getOriginalGraphDef())) return false;
+    if (!getInstrumentedGraphDef()
+        .equals(other.getInstrumentedGraphDef())) return false;
+    if (!getOuterContextId()
+        .equals(other.getOuterContextId())) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    hash = (37 * hash) + GRAPH_ID_FIELD_NUMBER;
+    hash = (53 * hash) + getGraphId().hashCode();
+    hash = (37 * hash) + GRAPH_NAME_FIELD_NUMBER;
+    hash = (53 * hash) + getGraphName().hashCode();
+    if (getInstrumentedOpsCount() > 0) {
+      hash = (37 * hash) + INSTRUMENTED_OPS_FIELD_NUMBER;
+      hash = (53 * hash) + getInstrumentedOpsList().hashCode();
+    }
+    hash = (37 * hash) + ORIGINAL_GRAPH_DEF_FIELD_NUMBER;
+    hash = (53 * hash) + getOriginalGraphDef().hashCode();
+    hash = (37 * hash) + INSTRUMENTED_GRAPH_DEF_FIELD_NUMBER;
+    hash = (53 * hash) + getInstrumentedGraphDef().hashCode();
+    hash = (37 * hash) + OUTER_CONTEXT_ID_FIELD_NUMBER;
+    hash = (53 * hash) + getOuterContextId().hashCode();
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.util.DebuggedGraph parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.DebuggedGraph parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.DebuggedGraph parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.DebuggedGraph parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.DebuggedGraph parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.DebuggedGraph parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.DebuggedGraph parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.DebuggedGraph parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.DebuggedGraph parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.DebuggedGraph parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.DebuggedGraph parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.DebuggedGraph parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.util.DebuggedGraph prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * <pre>
+   * A debugger-instrumented graph.
+   * </pre>
+   *
+   * Protobuf type {@code tensorflow.DebuggedGraph}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.DebuggedGraph)
+      org.tensorflow.proto.util.DebuggedGraphOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.util.DebugEventProtos.internal_static_tensorflow_DebuggedGraph_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.util.DebugEventProtos.internal_static_tensorflow_DebuggedGraph_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.util.DebuggedGraph.class, org.tensorflow.proto.util.DebuggedGraph.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.util.DebuggedGraph.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      graphId_ = "";
+
+      graphName_ = "";
+
+      instrumentedOps_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+      bitField0_ = (bitField0_ & ~0x00000001);
+      originalGraphDef_ = com.google.protobuf.ByteString.EMPTY;
+
+      instrumentedGraphDef_ = com.google.protobuf.ByteString.EMPTY;
+
+      outerContextId_ = "";
+
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.util.DebugEventProtos.internal_static_tensorflow_DebuggedGraph_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.DebuggedGraph getDefaultInstanceForType() {
+      return org.tensorflow.proto.util.DebuggedGraph.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.DebuggedGraph build() {
+      org.tensorflow.proto.util.DebuggedGraph result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.DebuggedGraph buildPartial() {
+      org.tensorflow.proto.util.DebuggedGraph result = new org.tensorflow.proto.util.DebuggedGraph(this);
+      int from_bitField0_ = bitField0_;
+      result.graphId_ = graphId_;
+      result.graphName_ = graphName_;
+      if (((bitField0_ & 0x00000001) != 0)) {
+        instrumentedOps_ = instrumentedOps_.getUnmodifiableView();
+        bitField0_ = (bitField0_ & ~0x00000001);
+      }
+      result.instrumentedOps_ = instrumentedOps_;
+      result.originalGraphDef_ = originalGraphDef_;
+      result.instrumentedGraphDef_ = instrumentedGraphDef_;
+      result.outerContextId_ = outerContextId_;
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.util.DebuggedGraph) {
+        return mergeFrom((org.tensorflow.proto.util.DebuggedGraph)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.util.DebuggedGraph other) {
+      if (other == org.tensorflow.proto.util.DebuggedGraph.getDefaultInstance()) return this;
+      if (!other.getGraphId().isEmpty()) {
+        graphId_ = other.graphId_;
+        onChanged();
+      }
+      if (!other.getGraphName().isEmpty()) {
+        graphName_ = other.graphName_;
+        onChanged();
+      }
+      if (!other.instrumentedOps_.isEmpty()) {
+        if (instrumentedOps_.isEmpty()) {
+          instrumentedOps_ = other.instrumentedOps_;
+          bitField0_ = (bitField0_ & ~0x00000001);
+        } else {
+          ensureInstrumentedOpsIsMutable();
+          instrumentedOps_.addAll(other.instrumentedOps_);
+        }
+        onChanged();
+      }
+      if (other.getOriginalGraphDef() != com.google.protobuf.ByteString.EMPTY) {
+        setOriginalGraphDef(other.getOriginalGraphDef());
+      }
+      if (other.getInstrumentedGraphDef() != com.google.protobuf.ByteString.EMPTY) {
+        setInstrumentedGraphDef(other.getInstrumentedGraphDef());
+      }
+      if (!other.getOuterContextId().isEmpty()) {
+        outerContextId_ = other.outerContextId_;
+        onChanged();
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.util.DebuggedGraph parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.util.DebuggedGraph) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+    private int bitField0_;
+
+    private java.lang.Object graphId_ = "";
+    /**
+     * <pre>
+     * An ID for the graph.
+     * This can be used up to look up graph names. Generated by the debugger.
+     * </pre>
+     *
+     * <code>string graph_id = 1;</code>
+     */
+    public java.lang.String getGraphId() {
+      java.lang.Object ref = graphId_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        graphId_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * An ID for the graph.
+     * This can be used up to look up graph names. Generated by the debugger.
+     * </pre>
+     *
+     * <code>string graph_id = 1;</code>
+     */
+    public com.google.protobuf.ByteString
+        getGraphIdBytes() {
+      java.lang.Object ref = graphId_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        graphId_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * An ID for the graph.
+     * This can be used up to look up graph names. Generated by the debugger.
+     * </pre>
+     *
+     * <code>string graph_id = 1;</code>
+     */
+    public Builder setGraphId(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      graphId_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * An ID for the graph.
+     * This can be used up to look up graph names. Generated by the debugger.
+     * </pre>
+     *
+     * <code>string graph_id = 1;</code>
+     */
+    public Builder clearGraphId() {
+      
+      graphId_ = getDefaultInstance().getGraphId();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * An ID for the graph.
+     * This can be used up to look up graph names. Generated by the debugger.
+     * </pre>
+     *
+     * <code>string graph_id = 1;</code>
+     */
+    public Builder setGraphIdBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      graphId_ = value;
+      onChanged();
+      return this;
+    }
+
+    private java.lang.Object graphName_ = "";
+    /**
+     * <pre>
+     * Name of the graph (if available).
+     * </pre>
+     *
+     * <code>string graph_name = 2;</code>
+     */
+    public java.lang.String getGraphName() {
+      java.lang.Object ref = graphName_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        graphName_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Name of the graph (if available).
+     * </pre>
+     *
+     * <code>string graph_name = 2;</code>
+     */
+    public com.google.protobuf.ByteString
+        getGraphNameBytes() {
+      java.lang.Object ref = graphName_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        graphName_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Name of the graph (if available).
+     * </pre>
+     *
+     * <code>string graph_name = 2;</code>
+     */
+    public Builder setGraphName(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      graphName_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Name of the graph (if available).
+     * </pre>
+     *
+     * <code>string graph_name = 2;</code>
+     */
+    public Builder clearGraphName() {
+      
+      graphName_ = getDefaultInstance().getGraphName();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Name of the graph (if available).
+     * </pre>
+     *
+     * <code>string graph_name = 2;</code>
+     */
+    public Builder setGraphNameBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      graphName_ = value;
+      onChanged();
+      return this;
+    }
+
+    private com.google.protobuf.LazyStringList instrumentedOps_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+    private void ensureInstrumentedOpsIsMutable() {
+      if (!((bitField0_ & 0x00000001) != 0)) {
+        instrumentedOps_ = new com.google.protobuf.LazyStringArrayList(instrumentedOps_);
+        bitField0_ |= 0x00000001;
+       }
+    }
+    /**
+     * <pre>
+     * Names of the instrumented ops. This can be used to look up op name
+     * based on the numeric-summary tensors (2nd column).
+     * </pre>
+     *
+     * <code>repeated string instrumented_ops = 3;</code>
+     */
+    public com.google.protobuf.ProtocolStringList
+        getInstrumentedOpsList() {
+      return instrumentedOps_.getUnmodifiableView();
+    }
+    /**
+     * <pre>
+     * Names of the instrumented ops. This can be used to look up op name
+     * based on the numeric-summary tensors (2nd column).
+     * </pre>
+     *
+     * <code>repeated string instrumented_ops = 3;</code>
+     */
+    public int getInstrumentedOpsCount() {
+      return instrumentedOps_.size();
+    }
+    /**
+     * <pre>
+     * Names of the instrumented ops. This can be used to look up op name
+     * based on the numeric-summary tensors (2nd column).
+     * </pre>
+     *
+     * <code>repeated string instrumented_ops = 3;</code>
+     */
+    public java.lang.String getInstrumentedOps(int index) {
+      return instrumentedOps_.get(index);
+    }
+    /**
+     * <pre>
+     * Names of the instrumented ops. This can be used to look up op name
+     * based on the numeric-summary tensors (2nd column).
+     * </pre>
+     *
+     * <code>repeated string instrumented_ops = 3;</code>
+     */
+    public com.google.protobuf.ByteString
+        getInstrumentedOpsBytes(int index) {
+      return instrumentedOps_.getByteString(index);
+    }
+    /**
+     * <pre>
+     * Names of the instrumented ops. This can be used to look up op name
+     * based on the numeric-summary tensors (2nd column).
+     * </pre>
+     *
+     * <code>repeated string instrumented_ops = 3;</code>
+     */
+    public Builder setInstrumentedOps(
+        int index, java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  ensureInstrumentedOpsIsMutable();
+      instrumentedOps_.set(index, value);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Names of the instrumented ops. This can be used to look up op name
+     * based on the numeric-summary tensors (2nd column).
+     * </pre>
+     *
+     * <code>repeated string instrumented_ops = 3;</code>
+     */
+    public Builder addInstrumentedOps(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  ensureInstrumentedOpsIsMutable();
+      instrumentedOps_.add(value);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Names of the instrumented ops. This can be used to look up op name
+     * based on the numeric-summary tensors (2nd column).
+     * </pre>
+     *
+     * <code>repeated string instrumented_ops = 3;</code>
+     */
+    public Builder addAllInstrumentedOps(
+        java.lang.Iterable<java.lang.String> values) {
+      ensureInstrumentedOpsIsMutable();
+      com.google.protobuf.AbstractMessageLite.Builder.addAll(
+          values, instrumentedOps_);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Names of the instrumented ops. This can be used to look up op name
+     * based on the numeric-summary tensors (2nd column).
+     * </pre>
+     *
+     * <code>repeated string instrumented_ops = 3;</code>
+     */
+    public Builder clearInstrumentedOps() {
+      instrumentedOps_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+      bitField0_ = (bitField0_ & ~0x00000001);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Names of the instrumented ops. This can be used to look up op name
+     * based on the numeric-summary tensors (2nd column).
+     * </pre>
+     *
+     * <code>repeated string instrumented_ops = 3;</code>
+     */
+    public Builder addInstrumentedOpsBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      ensureInstrumentedOpsIsMutable();
+      instrumentedOps_.add(value);
+      onChanged();
+      return this;
+    }
+
+    private com.google.protobuf.ByteString originalGraphDef_ = com.google.protobuf.ByteString.EMPTY;
+    /**
+     * <pre>
+     * Original (uninstrumented) GraphDef (if available).
+     * </pre>
+     *
+     * <code>bytes original_graph_def = 4;</code>
+     */
+    public com.google.protobuf.ByteString getOriginalGraphDef() {
+      return originalGraphDef_;
+    }
+    /**
+     * <pre>
+     * Original (uninstrumented) GraphDef (if available).
+     * </pre>
+     *
+     * <code>bytes original_graph_def = 4;</code>
+     */
+    public Builder setOriginalGraphDef(com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      originalGraphDef_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Original (uninstrumented) GraphDef (if available).
+     * </pre>
+     *
+     * <code>bytes original_graph_def = 4;</code>
+     */
+    public Builder clearOriginalGraphDef() {
+      
+      originalGraphDef_ = getDefaultInstance().getOriginalGraphDef();
+      onChanged();
+      return this;
+    }
+
+    private com.google.protobuf.ByteString instrumentedGraphDef_ = com.google.protobuf.ByteString.EMPTY;
+    /**
+     * <pre>
+     * An encoded version of a GraphDef.
+     * This graph may include the debugger-inserted ops.
+     * </pre>
+     *
+     * <code>bytes instrumented_graph_def = 5;</code>
+     */
+    public com.google.protobuf.ByteString getInstrumentedGraphDef() {
+      return instrumentedGraphDef_;
+    }
+    /**
+     * <pre>
+     * An encoded version of a GraphDef.
+     * This graph may include the debugger-inserted ops.
+     * </pre>
+     *
+     * <code>bytes instrumented_graph_def = 5;</code>
+     */
+    public Builder setInstrumentedGraphDef(com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      instrumentedGraphDef_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * An encoded version of a GraphDef.
+     * This graph may include the debugger-inserted ops.
+     * </pre>
+     *
+     * <code>bytes instrumented_graph_def = 5;</code>
+     */
+    public Builder clearInstrumentedGraphDef() {
+      
+      instrumentedGraphDef_ = getDefaultInstance().getInstrumentedGraphDef();
+      onChanged();
+      return this;
+    }
+
+    private java.lang.Object outerContextId_ = "";
+    /**
+     * <pre>
+     * IDs of the immediate enclosing context (graph), if any.
+     * </pre>
+     *
+     * <code>string outer_context_id = 6;</code>
+     */
+    public java.lang.String getOuterContextId() {
+      java.lang.Object ref = outerContextId_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        outerContextId_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * IDs of the immediate enclosing context (graph), if any.
+     * </pre>
+     *
+     * <code>string outer_context_id = 6;</code>
+     */
+    public com.google.protobuf.ByteString
+        getOuterContextIdBytes() {
+      java.lang.Object ref = outerContextId_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        outerContextId_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * IDs of the immediate enclosing context (graph), if any.
+     * </pre>
+     *
+     * <code>string outer_context_id = 6;</code>
+     */
+    public Builder setOuterContextId(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      outerContextId_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * IDs of the immediate enclosing context (graph), if any.
+     * </pre>
+     *
+     * <code>string outer_context_id = 6;</code>
+     */
+    public Builder clearOuterContextId() {
+      
+      outerContextId_ = getDefaultInstance().getOuterContextId();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * IDs of the immediate enclosing context (graph), if any.
+     * </pre>
+     *
+     * <code>string outer_context_id = 6;</code>
+     */
+    public Builder setOuterContextIdBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      outerContextId_ = value;
+      onChanged();
+      return this;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.DebuggedGraph)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.DebuggedGraph)
+  private static final org.tensorflow.proto.util.DebuggedGraph DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.util.DebuggedGraph();
+  }
+
+  public static org.tensorflow.proto.util.DebuggedGraph getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<DebuggedGraph>
+      PARSER = new com.google.protobuf.AbstractParser<DebuggedGraph>() {
+    @java.lang.Override
+    public DebuggedGraph parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new DebuggedGraph(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<DebuggedGraph> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<DebuggedGraph> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.util.DebuggedGraph getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/DebuggedGraphOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/DebuggedGraphOrBuilder.java
new file mode 100644
index 00000000000..c5f78a334e6
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/DebuggedGraphOrBuilder.java
@@ -0,0 +1,123 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/debug_event.proto
+
+package org.tensorflow.proto.util;
+
+public interface DebuggedGraphOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.DebuggedGraph)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <pre>
+   * An ID for the graph.
+   * This can be used up to look up graph names. Generated by the debugger.
+   * </pre>
+   *
+   * <code>string graph_id = 1;</code>
+   */
+  java.lang.String getGraphId();
+  /**
+   * <pre>
+   * An ID for the graph.
+   * This can be used up to look up graph names. Generated by the debugger.
+   * </pre>
+   *
+   * <code>string graph_id = 1;</code>
+   */
+  com.google.protobuf.ByteString
+      getGraphIdBytes();
+
+  /**
+   * <pre>
+   * Name of the graph (if available).
+   * </pre>
+   *
+   * <code>string graph_name = 2;</code>
+   */
+  java.lang.String getGraphName();
+  /**
+   * <pre>
+   * Name of the graph (if available).
+   * </pre>
+   *
+   * <code>string graph_name = 2;</code>
+   */
+  com.google.protobuf.ByteString
+      getGraphNameBytes();
+
+  /**
+   * <pre>
+   * Names of the instrumented ops. This can be used to look up op name
+   * based on the numeric-summary tensors (2nd column).
+   * </pre>
+   *
+   * <code>repeated string instrumented_ops = 3;</code>
+   */
+  java.util.List<java.lang.String>
+      getInstrumentedOpsList();
+  /**
+   * <pre>
+   * Names of the instrumented ops. This can be used to look up op name
+   * based on the numeric-summary tensors (2nd column).
+   * </pre>
+   *
+   * <code>repeated string instrumented_ops = 3;</code>
+   */
+  int getInstrumentedOpsCount();
+  /**
+   * <pre>
+   * Names of the instrumented ops. This can be used to look up op name
+   * based on the numeric-summary tensors (2nd column).
+   * </pre>
+   *
+   * <code>repeated string instrumented_ops = 3;</code>
+   */
+  java.lang.String getInstrumentedOps(int index);
+  /**
+   * <pre>
+   * Names of the instrumented ops. This can be used to look up op name
+   * based on the numeric-summary tensors (2nd column).
+   * </pre>
+   *
+   * <code>repeated string instrumented_ops = 3;</code>
+   */
+  com.google.protobuf.ByteString
+      getInstrumentedOpsBytes(int index);
+
+  /**
+   * <pre>
+   * Original (uninstrumented) GraphDef (if available).
+   * </pre>
+   *
+   * <code>bytes original_graph_def = 4;</code>
+   */
+  com.google.protobuf.ByteString getOriginalGraphDef();
+
+  /**
+   * <pre>
+   * An encoded version of a GraphDef.
+   * This graph may include the debugger-inserted ops.
+   * </pre>
+   *
+   * <code>bytes instrumented_graph_def = 5;</code>
+   */
+  com.google.protobuf.ByteString getInstrumentedGraphDef();
+
+  /**
+   * <pre>
+   * IDs of the immediate enclosing context (graph), if any.
+   * </pre>
+   *
+   * <code>string outer_context_id = 6;</code>
+   */
+  java.lang.String getOuterContextId();
+  /**
+   * <pre>
+   * IDs of the immediate enclosing context (graph), if any.
+   * </pre>
+   *
+   * <code>string outer_context_id = 6;</code>
+   */
+  com.google.protobuf.ByteString
+      getOuterContextIdBytes();
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/Event.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/Event.java
new file mode 100644
index 00000000000..9b30419c091
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/Event.java
@@ -0,0 +1,2049 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/util/event.proto
+
+package org.tensorflow.proto.util;
+
+/**
+ * <pre>
+ * Protocol buffer representing an event that happened during
+ * the execution of a Brain model.
+ * </pre>
+ *
+ * Protobuf type {@code tensorflow.Event}
+ */
+public  final class Event extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.Event)
+    EventOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use Event.newBuilder() to construct.
+  private Event(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private Event() {
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new Event();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private Event(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 9: {
+
+            wallTime_ = input.readDouble();
+            break;
+          }
+          case 16: {
+
+            step_ = input.readInt64();
+            break;
+          }
+          case 26: {
+            java.lang.String s = input.readStringRequireUtf8();
+            whatCase_ = 3;
+            what_ = s;
+            break;
+          }
+          case 34: {
+            whatCase_ = 4;
+            what_ = input.readBytes();
+            break;
+          }
+          case 42: {
+            org.tensorflow.proto.framework.Summary.Builder subBuilder = null;
+            if (whatCase_ == 5) {
+              subBuilder = ((org.tensorflow.proto.framework.Summary) what_).toBuilder();
+            }
+            what_ =
+                input.readMessage(org.tensorflow.proto.framework.Summary.parser(), extensionRegistry);
+            if (subBuilder != null) {
+              subBuilder.mergeFrom((org.tensorflow.proto.framework.Summary) what_);
+              what_ = subBuilder.buildPartial();
+            }
+            whatCase_ = 5;
+            break;
+          }
+          case 50: {
+            org.tensorflow.proto.util.LogMessage.Builder subBuilder = null;
+            if (whatCase_ == 6) {
+              subBuilder = ((org.tensorflow.proto.util.LogMessage) what_).toBuilder();
+            }
+            what_ =
+                input.readMessage(org.tensorflow.proto.util.LogMessage.parser(), extensionRegistry);
+            if (subBuilder != null) {
+              subBuilder.mergeFrom((org.tensorflow.proto.util.LogMessage) what_);
+              what_ = subBuilder.buildPartial();
+            }
+            whatCase_ = 6;
+            break;
+          }
+          case 58: {
+            org.tensorflow.proto.util.SessionLog.Builder subBuilder = null;
+            if (whatCase_ == 7) {
+              subBuilder = ((org.tensorflow.proto.util.SessionLog) what_).toBuilder();
+            }
+            what_ =
+                input.readMessage(org.tensorflow.proto.util.SessionLog.parser(), extensionRegistry);
+            if (subBuilder != null) {
+              subBuilder.mergeFrom((org.tensorflow.proto.util.SessionLog) what_);
+              what_ = subBuilder.buildPartial();
+            }
+            whatCase_ = 7;
+            break;
+          }
+          case 66: {
+            org.tensorflow.proto.util.TaggedRunMetadata.Builder subBuilder = null;
+            if (whatCase_ == 8) {
+              subBuilder = ((org.tensorflow.proto.util.TaggedRunMetadata) what_).toBuilder();
+            }
+            what_ =
+                input.readMessage(org.tensorflow.proto.util.TaggedRunMetadata.parser(), extensionRegistry);
+            if (subBuilder != null) {
+              subBuilder.mergeFrom((org.tensorflow.proto.util.TaggedRunMetadata) what_);
+              what_ = subBuilder.buildPartial();
+            }
+            whatCase_ = 8;
+            break;
+          }
+          case 74: {
+            whatCase_ = 9;
+            what_ = input.readBytes();
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.util.EventProtos.internal_static_tensorflow_Event_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.util.EventProtos.internal_static_tensorflow_Event_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.util.Event.class, org.tensorflow.proto.util.Event.Builder.class);
+  }
+
+  private int whatCase_ = 0;
+  private java.lang.Object what_;
+  public enum WhatCase
+      implements com.google.protobuf.Internal.EnumLite {
+    FILE_VERSION(3),
+    GRAPH_DEF(4),
+    SUMMARY(5),
+    LOG_MESSAGE(6),
+    SESSION_LOG(7),
+    TAGGED_RUN_METADATA(8),
+    META_GRAPH_DEF(9),
+    WHAT_NOT_SET(0);
+    private final int value;
+    private WhatCase(int value) {
+      this.value = value;
+    }
+    /**
+     * @deprecated Use {@link #forNumber(int)} instead.
+     */
+    @java.lang.Deprecated
+    public static WhatCase valueOf(int value) {
+      return forNumber(value);
+    }
+
+    public static WhatCase forNumber(int value) {
+      switch (value) {
+        case 3: return FILE_VERSION;
+        case 4: return GRAPH_DEF;
+        case 5: return SUMMARY;
+        case 6: return LOG_MESSAGE;
+        case 7: return SESSION_LOG;
+        case 8: return TAGGED_RUN_METADATA;
+        case 9: return META_GRAPH_DEF;
+        case 0: return WHAT_NOT_SET;
+        default: return null;
+      }
+    }
+    public int getNumber() {
+      return this.value;
+    }
+  };
+
+  public WhatCase
+  getWhatCase() {
+    return WhatCase.forNumber(
+        whatCase_);
+  }
+
+  public static final int WALL_TIME_FIELD_NUMBER = 1;
+  private double wallTime_;
+  /**
+   * <pre>
+   * Timestamp of the event.
+   * </pre>
+   *
+   * <code>double wall_time = 1;</code>
+   */
+  public double getWallTime() {
+    return wallTime_;
+  }
+
+  public static final int STEP_FIELD_NUMBER = 2;
+  private long step_;
+  /**
+   * <pre>
+   * Global step of the event.
+   * </pre>
+   *
+   * <code>int64 step = 2;</code>
+   */
+  public long getStep() {
+    return step_;
+  }
+
+  public static final int FILE_VERSION_FIELD_NUMBER = 3;
+  /**
+   * <pre>
+   * An event file was started, with the specified version.
+   * This is use to identify the contents of the record IO files
+   * easily.  Current version is "brain.Event:2".  All versions
+   * start with "brain.Event:".
+   * </pre>
+   *
+   * <code>string file_version = 3;</code>
+   */
+  public java.lang.String getFileVersion() {
+    java.lang.Object ref = "";
+    if (whatCase_ == 3) {
+      ref = what_;
+    }
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      if (whatCase_ == 3) {
+        what_ = s;
+      }
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * An event file was started, with the specified version.
+   * This is use to identify the contents of the record IO files
+   * easily.  Current version is "brain.Event:2".  All versions
+   * start with "brain.Event:".
+   * </pre>
+   *
+   * <code>string file_version = 3;</code>
+   */
+  public com.google.protobuf.ByteString
+      getFileVersionBytes() {
+    java.lang.Object ref = "";
+    if (whatCase_ == 3) {
+      ref = what_;
+    }
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      if (whatCase_ == 3) {
+        what_ = b;
+      }
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int GRAPH_DEF_FIELD_NUMBER = 4;
+  /**
+   * <pre>
+   * An encoded version of a GraphDef.
+   * </pre>
+   *
+   * <code>bytes graph_def = 4;</code>
+   */
+  public com.google.protobuf.ByteString getGraphDef() {
+    if (whatCase_ == 4) {
+      return (com.google.protobuf.ByteString) what_;
+    }
+    return com.google.protobuf.ByteString.EMPTY;
+  }
+
+  public static final int SUMMARY_FIELD_NUMBER = 5;
+  /**
+   * <pre>
+   * A summary was generated.
+   * </pre>
+   *
+   * <code>.tensorflow.Summary summary = 5;</code>
+   */
+  public boolean hasSummary() {
+    return whatCase_ == 5;
+  }
+  /**
+   * <pre>
+   * A summary was generated.
+   * </pre>
+   *
+   * <code>.tensorflow.Summary summary = 5;</code>
+   */
+  public org.tensorflow.proto.framework.Summary getSummary() {
+    if (whatCase_ == 5) {
+       return (org.tensorflow.proto.framework.Summary) what_;
+    }
+    return org.tensorflow.proto.framework.Summary.getDefaultInstance();
+  }
+  /**
+   * <pre>
+   * A summary was generated.
+   * </pre>
+   *
+   * <code>.tensorflow.Summary summary = 5;</code>
+   */
+  public org.tensorflow.proto.framework.SummaryOrBuilder getSummaryOrBuilder() {
+    if (whatCase_ == 5) {
+       return (org.tensorflow.proto.framework.Summary) what_;
+    }
+    return org.tensorflow.proto.framework.Summary.getDefaultInstance();
+  }
+
+  public static final int LOG_MESSAGE_FIELD_NUMBER = 6;
+  /**
+   * <pre>
+   * The user output a log message. Not all messages are logged, only ones
+   * generated via the Python tensorboard_logging module.
+   * </pre>
+   *
+   * <code>.tensorflow.LogMessage log_message = 6;</code>
+   */
+  public boolean hasLogMessage() {
+    return whatCase_ == 6;
+  }
+  /**
+   * <pre>
+   * The user output a log message. Not all messages are logged, only ones
+   * generated via the Python tensorboard_logging module.
+   * </pre>
+   *
+   * <code>.tensorflow.LogMessage log_message = 6;</code>
+   */
+  public org.tensorflow.proto.util.LogMessage getLogMessage() {
+    if (whatCase_ == 6) {
+       return (org.tensorflow.proto.util.LogMessage) what_;
+    }
+    return org.tensorflow.proto.util.LogMessage.getDefaultInstance();
+  }
+  /**
+   * <pre>
+   * The user output a log message. Not all messages are logged, only ones
+   * generated via the Python tensorboard_logging module.
+   * </pre>
+   *
+   * <code>.tensorflow.LogMessage log_message = 6;</code>
+   */
+  public org.tensorflow.proto.util.LogMessageOrBuilder getLogMessageOrBuilder() {
+    if (whatCase_ == 6) {
+       return (org.tensorflow.proto.util.LogMessage) what_;
+    }
+    return org.tensorflow.proto.util.LogMessage.getDefaultInstance();
+  }
+
+  public static final int SESSION_LOG_FIELD_NUMBER = 7;
+  /**
+   * <pre>
+   * The state of the session which can be used for restarting after crashes.
+   * </pre>
+   *
+   * <code>.tensorflow.SessionLog session_log = 7;</code>
+   */
+  public boolean hasSessionLog() {
+    return whatCase_ == 7;
+  }
+  /**
+   * <pre>
+   * The state of the session which can be used for restarting after crashes.
+   * </pre>
+   *
+   * <code>.tensorflow.SessionLog session_log = 7;</code>
+   */
+  public org.tensorflow.proto.util.SessionLog getSessionLog() {
+    if (whatCase_ == 7) {
+       return (org.tensorflow.proto.util.SessionLog) what_;
+    }
+    return org.tensorflow.proto.util.SessionLog.getDefaultInstance();
+  }
+  /**
+   * <pre>
+   * The state of the session which can be used for restarting after crashes.
+   * </pre>
+   *
+   * <code>.tensorflow.SessionLog session_log = 7;</code>
+   */
+  public org.tensorflow.proto.util.SessionLogOrBuilder getSessionLogOrBuilder() {
+    if (whatCase_ == 7) {
+       return (org.tensorflow.proto.util.SessionLog) what_;
+    }
+    return org.tensorflow.proto.util.SessionLog.getDefaultInstance();
+  }
+
+  public static final int TAGGED_RUN_METADATA_FIELD_NUMBER = 8;
+  /**
+   * <pre>
+   * The metadata returned by running a session.run() call.
+   * </pre>
+   *
+   * <code>.tensorflow.TaggedRunMetadata tagged_run_metadata = 8;</code>
+   */
+  public boolean hasTaggedRunMetadata() {
+    return whatCase_ == 8;
+  }
+  /**
+   * <pre>
+   * The metadata returned by running a session.run() call.
+   * </pre>
+   *
+   * <code>.tensorflow.TaggedRunMetadata tagged_run_metadata = 8;</code>
+   */
+  public org.tensorflow.proto.util.TaggedRunMetadata getTaggedRunMetadata() {
+    if (whatCase_ == 8) {
+       return (org.tensorflow.proto.util.TaggedRunMetadata) what_;
+    }
+    return org.tensorflow.proto.util.TaggedRunMetadata.getDefaultInstance();
+  }
+  /**
+   * <pre>
+   * The metadata returned by running a session.run() call.
+   * </pre>
+   *
+   * <code>.tensorflow.TaggedRunMetadata tagged_run_metadata = 8;</code>
+   */
+  public org.tensorflow.proto.util.TaggedRunMetadataOrBuilder getTaggedRunMetadataOrBuilder() {
+    if (whatCase_ == 8) {
+       return (org.tensorflow.proto.util.TaggedRunMetadata) what_;
+    }
+    return org.tensorflow.proto.util.TaggedRunMetadata.getDefaultInstance();
+  }
+
+  public static final int META_GRAPH_DEF_FIELD_NUMBER = 9;
+  /**
+   * <pre>
+   * An encoded version of a MetaGraphDef.
+   * </pre>
+   *
+   * <code>bytes meta_graph_def = 9;</code>
+   */
+  public com.google.protobuf.ByteString getMetaGraphDef() {
+    if (whatCase_ == 9) {
+      return (com.google.protobuf.ByteString) what_;
+    }
+    return com.google.protobuf.ByteString.EMPTY;
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    if (wallTime_ != 0D) {
+      output.writeDouble(1, wallTime_);
+    }
+    if (step_ != 0L) {
+      output.writeInt64(2, step_);
+    }
+    if (whatCase_ == 3) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 3, what_);
+    }
+    if (whatCase_ == 4) {
+      output.writeBytes(
+          4, (com.google.protobuf.ByteString) what_);
+    }
+    if (whatCase_ == 5) {
+      output.writeMessage(5, (org.tensorflow.proto.framework.Summary) what_);
+    }
+    if (whatCase_ == 6) {
+      output.writeMessage(6, (org.tensorflow.proto.util.LogMessage) what_);
+    }
+    if (whatCase_ == 7) {
+      output.writeMessage(7, (org.tensorflow.proto.util.SessionLog) what_);
+    }
+    if (whatCase_ == 8) {
+      output.writeMessage(8, (org.tensorflow.proto.util.TaggedRunMetadata) what_);
+    }
+    if (whatCase_ == 9) {
+      output.writeBytes(
+          9, (com.google.protobuf.ByteString) what_);
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (wallTime_ != 0D) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeDoubleSize(1, wallTime_);
+    }
+    if (step_ != 0L) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt64Size(2, step_);
+    }
+    if (whatCase_ == 3) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(3, what_);
+    }
+    if (whatCase_ == 4) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeBytesSize(
+            4, (com.google.protobuf.ByteString) what_);
+    }
+    if (whatCase_ == 5) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(5, (org.tensorflow.proto.framework.Summary) what_);
+    }
+    if (whatCase_ == 6) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(6, (org.tensorflow.proto.util.LogMessage) what_);
+    }
+    if (whatCase_ == 7) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(7, (org.tensorflow.proto.util.SessionLog) what_);
+    }
+    if (whatCase_ == 8) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(8, (org.tensorflow.proto.util.TaggedRunMetadata) what_);
+    }
+    if (whatCase_ == 9) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeBytesSize(
+            9, (com.google.protobuf.ByteString) what_);
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.util.Event)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.util.Event other = (org.tensorflow.proto.util.Event) obj;
+
+    if (java.lang.Double.doubleToLongBits(getWallTime())
+        != java.lang.Double.doubleToLongBits(
+            other.getWallTime())) return false;
+    if (getStep()
+        != other.getStep()) return false;
+    if (!getWhatCase().equals(other.getWhatCase())) return false;
+    switch (whatCase_) {
+      case 3:
+        if (!getFileVersion()
+            .equals(other.getFileVersion())) return false;
+        break;
+      case 4:
+        if (!getGraphDef()
+            .equals(other.getGraphDef())) return false;
+        break;
+      case 5:
+        if (!getSummary()
+            .equals(other.getSummary())) return false;
+        break;
+      case 6:
+        if (!getLogMessage()
+            .equals(other.getLogMessage())) return false;
+        break;
+      case 7:
+        if (!getSessionLog()
+            .equals(other.getSessionLog())) return false;
+        break;
+      case 8:
+        if (!getTaggedRunMetadata()
+            .equals(other.getTaggedRunMetadata())) return false;
+        break;
+      case 9:
+        if (!getMetaGraphDef()
+            .equals(other.getMetaGraphDef())) return false;
+        break;
+      case 0:
+      default:
+    }
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    hash = (37 * hash) + WALL_TIME_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+        java.lang.Double.doubleToLongBits(getWallTime()));
+    hash = (37 * hash) + STEP_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+        getStep());
+    switch (whatCase_) {
+      case 3:
+        hash = (37 * hash) + FILE_VERSION_FIELD_NUMBER;
+        hash = (53 * hash) + getFileVersion().hashCode();
+        break;
+      case 4:
+        hash = (37 * hash) + GRAPH_DEF_FIELD_NUMBER;
+        hash = (53 * hash) + getGraphDef().hashCode();
+        break;
+      case 5:
+        hash = (37 * hash) + SUMMARY_FIELD_NUMBER;
+        hash = (53 * hash) + getSummary().hashCode();
+        break;
+      case 6:
+        hash = (37 * hash) + LOG_MESSAGE_FIELD_NUMBER;
+        hash = (53 * hash) + getLogMessage().hashCode();
+        break;
+      case 7:
+        hash = (37 * hash) + SESSION_LOG_FIELD_NUMBER;
+        hash = (53 * hash) + getSessionLog().hashCode();
+        break;
+      case 8:
+        hash = (37 * hash) + TAGGED_RUN_METADATA_FIELD_NUMBER;
+        hash = (53 * hash) + getTaggedRunMetadata().hashCode();
+        break;
+      case 9:
+        hash = (37 * hash) + META_GRAPH_DEF_FIELD_NUMBER;
+        hash = (53 * hash) + getMetaGraphDef().hashCode();
+        break;
+      case 0:
+      default:
+    }
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.util.Event parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.Event parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.Event parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.Event parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.Event parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.Event parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.Event parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.Event parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.Event parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.Event parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.Event parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.Event parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.util.Event prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * <pre>
+   * Protocol buffer representing an event that happened during
+   * the execution of a Brain model.
+   * </pre>
+   *
+   * Protobuf type {@code tensorflow.Event}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.Event)
+      org.tensorflow.proto.util.EventOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.util.EventProtos.internal_static_tensorflow_Event_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.util.EventProtos.internal_static_tensorflow_Event_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.util.Event.class, org.tensorflow.proto.util.Event.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.util.Event.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      wallTime_ = 0D;
+
+      step_ = 0L;
+
+      whatCase_ = 0;
+      what_ = null;
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.util.EventProtos.internal_static_tensorflow_Event_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.Event getDefaultInstanceForType() {
+      return org.tensorflow.proto.util.Event.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.Event build() {
+      org.tensorflow.proto.util.Event result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.Event buildPartial() {
+      org.tensorflow.proto.util.Event result = new org.tensorflow.proto.util.Event(this);
+      result.wallTime_ = wallTime_;
+      result.step_ = step_;
+      if (whatCase_ == 3) {
+        result.what_ = what_;
+      }
+      if (whatCase_ == 4) {
+        result.what_ = what_;
+      }
+      if (whatCase_ == 5) {
+        if (summaryBuilder_ == null) {
+          result.what_ = what_;
+        } else {
+          result.what_ = summaryBuilder_.build();
+        }
+      }
+      if (whatCase_ == 6) {
+        if (logMessageBuilder_ == null) {
+          result.what_ = what_;
+        } else {
+          result.what_ = logMessageBuilder_.build();
+        }
+      }
+      if (whatCase_ == 7) {
+        if (sessionLogBuilder_ == null) {
+          result.what_ = what_;
+        } else {
+          result.what_ = sessionLogBuilder_.build();
+        }
+      }
+      if (whatCase_ == 8) {
+        if (taggedRunMetadataBuilder_ == null) {
+          result.what_ = what_;
+        } else {
+          result.what_ = taggedRunMetadataBuilder_.build();
+        }
+      }
+      if (whatCase_ == 9) {
+        result.what_ = what_;
+      }
+      result.whatCase_ = whatCase_;
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.util.Event) {
+        return mergeFrom((org.tensorflow.proto.util.Event)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.util.Event other) {
+      if (other == org.tensorflow.proto.util.Event.getDefaultInstance()) return this;
+      if (other.getWallTime() != 0D) {
+        setWallTime(other.getWallTime());
+      }
+      if (other.getStep() != 0L) {
+        setStep(other.getStep());
+      }
+      switch (other.getWhatCase()) {
+        case FILE_VERSION: {
+          whatCase_ = 3;
+          what_ = other.what_;
+          onChanged();
+          break;
+        }
+        case GRAPH_DEF: {
+          setGraphDef(other.getGraphDef());
+          break;
+        }
+        case SUMMARY: {
+          mergeSummary(other.getSummary());
+          break;
+        }
+        case LOG_MESSAGE: {
+          mergeLogMessage(other.getLogMessage());
+          break;
+        }
+        case SESSION_LOG: {
+          mergeSessionLog(other.getSessionLog());
+          break;
+        }
+        case TAGGED_RUN_METADATA: {
+          mergeTaggedRunMetadata(other.getTaggedRunMetadata());
+          break;
+        }
+        case META_GRAPH_DEF: {
+          setMetaGraphDef(other.getMetaGraphDef());
+          break;
+        }
+        case WHAT_NOT_SET: {
+          break;
+        }
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.util.Event parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.util.Event) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+    private int whatCase_ = 0;
+    private java.lang.Object what_;
+    public WhatCase
+        getWhatCase() {
+      return WhatCase.forNumber(
+          whatCase_);
+    }
+
+    public Builder clearWhat() {
+      whatCase_ = 0;
+      what_ = null;
+      onChanged();
+      return this;
+    }
+
+
+    private double wallTime_ ;
+    /**
+     * <pre>
+     * Timestamp of the event.
+     * </pre>
+     *
+     * <code>double wall_time = 1;</code>
+     */
+    public double getWallTime() {
+      return wallTime_;
+    }
+    /**
+     * <pre>
+     * Timestamp of the event.
+     * </pre>
+     *
+     * <code>double wall_time = 1;</code>
+     */
+    public Builder setWallTime(double value) {
+      
+      wallTime_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Timestamp of the event.
+     * </pre>
+     *
+     * <code>double wall_time = 1;</code>
+     */
+    public Builder clearWallTime() {
+      
+      wallTime_ = 0D;
+      onChanged();
+      return this;
+    }
+
+    private long step_ ;
+    /**
+     * <pre>
+     * Global step of the event.
+     * </pre>
+     *
+     * <code>int64 step = 2;</code>
+     */
+    public long getStep() {
+      return step_;
+    }
+    /**
+     * <pre>
+     * Global step of the event.
+     * </pre>
+     *
+     * <code>int64 step = 2;</code>
+     */
+    public Builder setStep(long value) {
+      
+      step_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Global step of the event.
+     * </pre>
+     *
+     * <code>int64 step = 2;</code>
+     */
+    public Builder clearStep() {
+      
+      step_ = 0L;
+      onChanged();
+      return this;
+    }
+
+    /**
+     * <pre>
+     * An event file was started, with the specified version.
+     * This is use to identify the contents of the record IO files
+     * easily.  Current version is "brain.Event:2".  All versions
+     * start with "brain.Event:".
+     * </pre>
+     *
+     * <code>string file_version = 3;</code>
+     */
+    public java.lang.String getFileVersion() {
+      java.lang.Object ref = "";
+      if (whatCase_ == 3) {
+        ref = what_;
+      }
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        if (whatCase_ == 3) {
+          what_ = s;
+        }
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * An event file was started, with the specified version.
+     * This is use to identify the contents of the record IO files
+     * easily.  Current version is "brain.Event:2".  All versions
+     * start with "brain.Event:".
+     * </pre>
+     *
+     * <code>string file_version = 3;</code>
+     */
+    public com.google.protobuf.ByteString
+        getFileVersionBytes() {
+      java.lang.Object ref = "";
+      if (whatCase_ == 3) {
+        ref = what_;
+      }
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        if (whatCase_ == 3) {
+          what_ = b;
+        }
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * An event file was started, with the specified version.
+     * This is use to identify the contents of the record IO files
+     * easily.  Current version is "brain.Event:2".  All versions
+     * start with "brain.Event:".
+     * </pre>
+     *
+     * <code>string file_version = 3;</code>
+     */
+    public Builder setFileVersion(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  whatCase_ = 3;
+      what_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * An event file was started, with the specified version.
+     * This is use to identify the contents of the record IO files
+     * easily.  Current version is "brain.Event:2".  All versions
+     * start with "brain.Event:".
+     * </pre>
+     *
+     * <code>string file_version = 3;</code>
+     */
+    public Builder clearFileVersion() {
+      if (whatCase_ == 3) {
+        whatCase_ = 0;
+        what_ = null;
+        onChanged();
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * An event file was started, with the specified version.
+     * This is use to identify the contents of the record IO files
+     * easily.  Current version is "brain.Event:2".  All versions
+     * start with "brain.Event:".
+     * </pre>
+     *
+     * <code>string file_version = 3;</code>
+     */
+    public Builder setFileVersionBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      whatCase_ = 3;
+      what_ = value;
+      onChanged();
+      return this;
+    }
+
+    /**
+     * <pre>
+     * An encoded version of a GraphDef.
+     * </pre>
+     *
+     * <code>bytes graph_def = 4;</code>
+     */
+    public com.google.protobuf.ByteString getGraphDef() {
+      if (whatCase_ == 4) {
+        return (com.google.protobuf.ByteString) what_;
+      }
+      return com.google.protobuf.ByteString.EMPTY;
+    }
+    /**
+     * <pre>
+     * An encoded version of a GraphDef.
+     * </pre>
+     *
+     * <code>bytes graph_def = 4;</code>
+     */
+    public Builder setGraphDef(com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  whatCase_ = 4;
+      what_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * An encoded version of a GraphDef.
+     * </pre>
+     *
+     * <code>bytes graph_def = 4;</code>
+     */
+    public Builder clearGraphDef() {
+      if (whatCase_ == 4) {
+        whatCase_ = 0;
+        what_ = null;
+        onChanged();
+      }
+      return this;
+    }
+
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.Summary, org.tensorflow.proto.framework.Summary.Builder, org.tensorflow.proto.framework.SummaryOrBuilder> summaryBuilder_;
+    /**
+     * <pre>
+     * A summary was generated.
+     * </pre>
+     *
+     * <code>.tensorflow.Summary summary = 5;</code>
+     */
+    public boolean hasSummary() {
+      return whatCase_ == 5;
+    }
+    /**
+     * <pre>
+     * A summary was generated.
+     * </pre>
+     *
+     * <code>.tensorflow.Summary summary = 5;</code>
+     */
+    public org.tensorflow.proto.framework.Summary getSummary() {
+      if (summaryBuilder_ == null) {
+        if (whatCase_ == 5) {
+          return (org.tensorflow.proto.framework.Summary) what_;
+        }
+        return org.tensorflow.proto.framework.Summary.getDefaultInstance();
+      } else {
+        if (whatCase_ == 5) {
+          return summaryBuilder_.getMessage();
+        }
+        return org.tensorflow.proto.framework.Summary.getDefaultInstance();
+      }
+    }
+    /**
+     * <pre>
+     * A summary was generated.
+     * </pre>
+     *
+     * <code>.tensorflow.Summary summary = 5;</code>
+     */
+    public Builder setSummary(org.tensorflow.proto.framework.Summary value) {
+      if (summaryBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        what_ = value;
+        onChanged();
+      } else {
+        summaryBuilder_.setMessage(value);
+      }
+      whatCase_ = 5;
+      return this;
+    }
+    /**
+     * <pre>
+     * A summary was generated.
+     * </pre>
+     *
+     * <code>.tensorflow.Summary summary = 5;</code>
+     */
+    public Builder setSummary(
+        org.tensorflow.proto.framework.Summary.Builder builderForValue) {
+      if (summaryBuilder_ == null) {
+        what_ = builderForValue.build();
+        onChanged();
+      } else {
+        summaryBuilder_.setMessage(builderForValue.build());
+      }
+      whatCase_ = 5;
+      return this;
+    }
+    /**
+     * <pre>
+     * A summary was generated.
+     * </pre>
+     *
+     * <code>.tensorflow.Summary summary = 5;</code>
+     */
+    public Builder mergeSummary(org.tensorflow.proto.framework.Summary value) {
+      if (summaryBuilder_ == null) {
+        if (whatCase_ == 5 &&
+            what_ != org.tensorflow.proto.framework.Summary.getDefaultInstance()) {
+          what_ = org.tensorflow.proto.framework.Summary.newBuilder((org.tensorflow.proto.framework.Summary) what_)
+              .mergeFrom(value).buildPartial();
+        } else {
+          what_ = value;
+        }
+        onChanged();
+      } else {
+        if (whatCase_ == 5) {
+          summaryBuilder_.mergeFrom(value);
+        }
+        summaryBuilder_.setMessage(value);
+      }
+      whatCase_ = 5;
+      return this;
+    }
+    /**
+     * <pre>
+     * A summary was generated.
+     * </pre>
+     *
+     * <code>.tensorflow.Summary summary = 5;</code>
+     */
+    public Builder clearSummary() {
+      if (summaryBuilder_ == null) {
+        if (whatCase_ == 5) {
+          whatCase_ = 0;
+          what_ = null;
+          onChanged();
+        }
+      } else {
+        if (whatCase_ == 5) {
+          whatCase_ = 0;
+          what_ = null;
+        }
+        summaryBuilder_.clear();
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * A summary was generated.
+     * </pre>
+     *
+     * <code>.tensorflow.Summary summary = 5;</code>
+     */
+    public org.tensorflow.proto.framework.Summary.Builder getSummaryBuilder() {
+      return getSummaryFieldBuilder().getBuilder();
+    }
+    /**
+     * <pre>
+     * A summary was generated.
+     * </pre>
+     *
+     * <code>.tensorflow.Summary summary = 5;</code>
+     */
+    public org.tensorflow.proto.framework.SummaryOrBuilder getSummaryOrBuilder() {
+      if ((whatCase_ == 5) && (summaryBuilder_ != null)) {
+        return summaryBuilder_.getMessageOrBuilder();
+      } else {
+        if (whatCase_ == 5) {
+          return (org.tensorflow.proto.framework.Summary) what_;
+        }
+        return org.tensorflow.proto.framework.Summary.getDefaultInstance();
+      }
+    }
+    /**
+     * <pre>
+     * A summary was generated.
+     * </pre>
+     *
+     * <code>.tensorflow.Summary summary = 5;</code>
+     */
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.Summary, org.tensorflow.proto.framework.Summary.Builder, org.tensorflow.proto.framework.SummaryOrBuilder> 
+        getSummaryFieldBuilder() {
+      if (summaryBuilder_ == null) {
+        if (!(whatCase_ == 5)) {
+          what_ = org.tensorflow.proto.framework.Summary.getDefaultInstance();
+        }
+        summaryBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+            org.tensorflow.proto.framework.Summary, org.tensorflow.proto.framework.Summary.Builder, org.tensorflow.proto.framework.SummaryOrBuilder>(
+                (org.tensorflow.proto.framework.Summary) what_,
+                getParentForChildren(),
+                isClean());
+        what_ = null;
+      }
+      whatCase_ = 5;
+      onChanged();;
+      return summaryBuilder_;
+    }
+
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.util.LogMessage, org.tensorflow.proto.util.LogMessage.Builder, org.tensorflow.proto.util.LogMessageOrBuilder> logMessageBuilder_;
+    /**
+     * <pre>
+     * The user output a log message. Not all messages are logged, only ones
+     * generated via the Python tensorboard_logging module.
+     * </pre>
+     *
+     * <code>.tensorflow.LogMessage log_message = 6;</code>
+     */
+    public boolean hasLogMessage() {
+      return whatCase_ == 6;
+    }
+    /**
+     * <pre>
+     * The user output a log message. Not all messages are logged, only ones
+     * generated via the Python tensorboard_logging module.
+     * </pre>
+     *
+     * <code>.tensorflow.LogMessage log_message = 6;</code>
+     */
+    public org.tensorflow.proto.util.LogMessage getLogMessage() {
+      if (logMessageBuilder_ == null) {
+        if (whatCase_ == 6) {
+          return (org.tensorflow.proto.util.LogMessage) what_;
+        }
+        return org.tensorflow.proto.util.LogMessage.getDefaultInstance();
+      } else {
+        if (whatCase_ == 6) {
+          return logMessageBuilder_.getMessage();
+        }
+        return org.tensorflow.proto.util.LogMessage.getDefaultInstance();
+      }
+    }
+    /**
+     * <pre>
+     * The user output a log message. Not all messages are logged, only ones
+     * generated via the Python tensorboard_logging module.
+     * </pre>
+     *
+     * <code>.tensorflow.LogMessage log_message = 6;</code>
+     */
+    public Builder setLogMessage(org.tensorflow.proto.util.LogMessage value) {
+      if (logMessageBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        what_ = value;
+        onChanged();
+      } else {
+        logMessageBuilder_.setMessage(value);
+      }
+      whatCase_ = 6;
+      return this;
+    }
+    /**
+     * <pre>
+     * The user output a log message. Not all messages are logged, only ones
+     * generated via the Python tensorboard_logging module.
+     * </pre>
+     *
+     * <code>.tensorflow.LogMessage log_message = 6;</code>
+     */
+    public Builder setLogMessage(
+        org.tensorflow.proto.util.LogMessage.Builder builderForValue) {
+      if (logMessageBuilder_ == null) {
+        what_ = builderForValue.build();
+        onChanged();
+      } else {
+        logMessageBuilder_.setMessage(builderForValue.build());
+      }
+      whatCase_ = 6;
+      return this;
+    }
+    /**
+     * <pre>
+     * The user output a log message. Not all messages are logged, only ones
+     * generated via the Python tensorboard_logging module.
+     * </pre>
+     *
+     * <code>.tensorflow.LogMessage log_message = 6;</code>
+     */
+    public Builder mergeLogMessage(org.tensorflow.proto.util.LogMessage value) {
+      if (logMessageBuilder_ == null) {
+        if (whatCase_ == 6 &&
+            what_ != org.tensorflow.proto.util.LogMessage.getDefaultInstance()) {
+          what_ = org.tensorflow.proto.util.LogMessage.newBuilder((org.tensorflow.proto.util.LogMessage) what_)
+              .mergeFrom(value).buildPartial();
+        } else {
+          what_ = value;
+        }
+        onChanged();
+      } else {
+        if (whatCase_ == 6) {
+          logMessageBuilder_.mergeFrom(value);
+        }
+        logMessageBuilder_.setMessage(value);
+      }
+      whatCase_ = 6;
+      return this;
+    }
+    /**
+     * <pre>
+     * The user output a log message. Not all messages are logged, only ones
+     * generated via the Python tensorboard_logging module.
+     * </pre>
+     *
+     * <code>.tensorflow.LogMessage log_message = 6;</code>
+     */
+    public Builder clearLogMessage() {
+      if (logMessageBuilder_ == null) {
+        if (whatCase_ == 6) {
+          whatCase_ = 0;
+          what_ = null;
+          onChanged();
+        }
+      } else {
+        if (whatCase_ == 6) {
+          whatCase_ = 0;
+          what_ = null;
+        }
+        logMessageBuilder_.clear();
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * The user output a log message. Not all messages are logged, only ones
+     * generated via the Python tensorboard_logging module.
+     * </pre>
+     *
+     * <code>.tensorflow.LogMessage log_message = 6;</code>
+     */
+    public org.tensorflow.proto.util.LogMessage.Builder getLogMessageBuilder() {
+      return getLogMessageFieldBuilder().getBuilder();
+    }
+    /**
+     * <pre>
+     * The user output a log message. Not all messages are logged, only ones
+     * generated via the Python tensorboard_logging module.
+     * </pre>
+     *
+     * <code>.tensorflow.LogMessage log_message = 6;</code>
+     */
+    public org.tensorflow.proto.util.LogMessageOrBuilder getLogMessageOrBuilder() {
+      if ((whatCase_ == 6) && (logMessageBuilder_ != null)) {
+        return logMessageBuilder_.getMessageOrBuilder();
+      } else {
+        if (whatCase_ == 6) {
+          return (org.tensorflow.proto.util.LogMessage) what_;
+        }
+        return org.tensorflow.proto.util.LogMessage.getDefaultInstance();
+      }
+    }
+    /**
+     * <pre>
+     * The user output a log message. Not all messages are logged, only ones
+     * generated via the Python tensorboard_logging module.
+     * </pre>
+     *
+     * <code>.tensorflow.LogMessage log_message = 6;</code>
+     */
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.util.LogMessage, org.tensorflow.proto.util.LogMessage.Builder, org.tensorflow.proto.util.LogMessageOrBuilder> 
+        getLogMessageFieldBuilder() {
+      if (logMessageBuilder_ == null) {
+        if (!(whatCase_ == 6)) {
+          what_ = org.tensorflow.proto.util.LogMessage.getDefaultInstance();
+        }
+        logMessageBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+            org.tensorflow.proto.util.LogMessage, org.tensorflow.proto.util.LogMessage.Builder, org.tensorflow.proto.util.LogMessageOrBuilder>(
+                (org.tensorflow.proto.util.LogMessage) what_,
+                getParentForChildren(),
+                isClean());
+        what_ = null;
+      }
+      whatCase_ = 6;
+      onChanged();;
+      return logMessageBuilder_;
+    }
+
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.util.SessionLog, org.tensorflow.proto.util.SessionLog.Builder, org.tensorflow.proto.util.SessionLogOrBuilder> sessionLogBuilder_;
+    /**
+     * <pre>
+     * The state of the session which can be used for restarting after crashes.
+     * </pre>
+     *
+     * <code>.tensorflow.SessionLog session_log = 7;</code>
+     */
+    public boolean hasSessionLog() {
+      return whatCase_ == 7;
+    }
+    /**
+     * <pre>
+     * The state of the session which can be used for restarting after crashes.
+     * </pre>
+     *
+     * <code>.tensorflow.SessionLog session_log = 7;</code>
+     */
+    public org.tensorflow.proto.util.SessionLog getSessionLog() {
+      if (sessionLogBuilder_ == null) {
+        if (whatCase_ == 7) {
+          return (org.tensorflow.proto.util.SessionLog) what_;
+        }
+        return org.tensorflow.proto.util.SessionLog.getDefaultInstance();
+      } else {
+        if (whatCase_ == 7) {
+          return sessionLogBuilder_.getMessage();
+        }
+        return org.tensorflow.proto.util.SessionLog.getDefaultInstance();
+      }
+    }
+    /**
+     * <pre>
+     * The state of the session which can be used for restarting after crashes.
+     * </pre>
+     *
+     * <code>.tensorflow.SessionLog session_log = 7;</code>
+     */
+    public Builder setSessionLog(org.tensorflow.proto.util.SessionLog value) {
+      if (sessionLogBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        what_ = value;
+        onChanged();
+      } else {
+        sessionLogBuilder_.setMessage(value);
+      }
+      whatCase_ = 7;
+      return this;
+    }
+    /**
+     * <pre>
+     * The state of the session which can be used for restarting after crashes.
+     * </pre>
+     *
+     * <code>.tensorflow.SessionLog session_log = 7;</code>
+     */
+    public Builder setSessionLog(
+        org.tensorflow.proto.util.SessionLog.Builder builderForValue) {
+      if (sessionLogBuilder_ == null) {
+        what_ = builderForValue.build();
+        onChanged();
+      } else {
+        sessionLogBuilder_.setMessage(builderForValue.build());
+      }
+      whatCase_ = 7;
+      return this;
+    }
+    /**
+     * <pre>
+     * The state of the session which can be used for restarting after crashes.
+     * </pre>
+     *
+     * <code>.tensorflow.SessionLog session_log = 7;</code>
+     */
+    public Builder mergeSessionLog(org.tensorflow.proto.util.SessionLog value) {
+      if (sessionLogBuilder_ == null) {
+        if (whatCase_ == 7 &&
+            what_ != org.tensorflow.proto.util.SessionLog.getDefaultInstance()) {
+          what_ = org.tensorflow.proto.util.SessionLog.newBuilder((org.tensorflow.proto.util.SessionLog) what_)
+              .mergeFrom(value).buildPartial();
+        } else {
+          what_ = value;
+        }
+        onChanged();
+      } else {
+        if (whatCase_ == 7) {
+          sessionLogBuilder_.mergeFrom(value);
+        }
+        sessionLogBuilder_.setMessage(value);
+      }
+      whatCase_ = 7;
+      return this;
+    }
+    /**
+     * <pre>
+     * The state of the session which can be used for restarting after crashes.
+     * </pre>
+     *
+     * <code>.tensorflow.SessionLog session_log = 7;</code>
+     */
+    public Builder clearSessionLog() {
+      if (sessionLogBuilder_ == null) {
+        if (whatCase_ == 7) {
+          whatCase_ = 0;
+          what_ = null;
+          onChanged();
+        }
+      } else {
+        if (whatCase_ == 7) {
+          whatCase_ = 0;
+          what_ = null;
+        }
+        sessionLogBuilder_.clear();
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * The state of the session which can be used for restarting after crashes.
+     * </pre>
+     *
+     * <code>.tensorflow.SessionLog session_log = 7;</code>
+     */
+    public org.tensorflow.proto.util.SessionLog.Builder getSessionLogBuilder() {
+      return getSessionLogFieldBuilder().getBuilder();
+    }
+    /**
+     * <pre>
+     * The state of the session which can be used for restarting after crashes.
+     * </pre>
+     *
+     * <code>.tensorflow.SessionLog session_log = 7;</code>
+     */
+    public org.tensorflow.proto.util.SessionLogOrBuilder getSessionLogOrBuilder() {
+      if ((whatCase_ == 7) && (sessionLogBuilder_ != null)) {
+        return sessionLogBuilder_.getMessageOrBuilder();
+      } else {
+        if (whatCase_ == 7) {
+          return (org.tensorflow.proto.util.SessionLog) what_;
+        }
+        return org.tensorflow.proto.util.SessionLog.getDefaultInstance();
+      }
+    }
+    /**
+     * <pre>
+     * The state of the session which can be used for restarting after crashes.
+     * </pre>
+     *
+     * <code>.tensorflow.SessionLog session_log = 7;</code>
+     */
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.util.SessionLog, org.tensorflow.proto.util.SessionLog.Builder, org.tensorflow.proto.util.SessionLogOrBuilder> 
+        getSessionLogFieldBuilder() {
+      if (sessionLogBuilder_ == null) {
+        if (!(whatCase_ == 7)) {
+          what_ = org.tensorflow.proto.util.SessionLog.getDefaultInstance();
+        }
+        sessionLogBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+            org.tensorflow.proto.util.SessionLog, org.tensorflow.proto.util.SessionLog.Builder, org.tensorflow.proto.util.SessionLogOrBuilder>(
+                (org.tensorflow.proto.util.SessionLog) what_,
+                getParentForChildren(),
+                isClean());
+        what_ = null;
+      }
+      whatCase_ = 7;
+      onChanged();;
+      return sessionLogBuilder_;
+    }
+
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.util.TaggedRunMetadata, org.tensorflow.proto.util.TaggedRunMetadata.Builder, org.tensorflow.proto.util.TaggedRunMetadataOrBuilder> taggedRunMetadataBuilder_;
+    /**
+     * <pre>
+     * The metadata returned by running a session.run() call.
+     * </pre>
+     *
+     * <code>.tensorflow.TaggedRunMetadata tagged_run_metadata = 8;</code>
+     */
+    public boolean hasTaggedRunMetadata() {
+      return whatCase_ == 8;
+    }
+    /**
+     * <pre>
+     * The metadata returned by running a session.run() call.
+     * </pre>
+     *
+     * <code>.tensorflow.TaggedRunMetadata tagged_run_metadata = 8;</code>
+     */
+    public org.tensorflow.proto.util.TaggedRunMetadata getTaggedRunMetadata() {
+      if (taggedRunMetadataBuilder_ == null) {
+        if (whatCase_ == 8) {
+          return (org.tensorflow.proto.util.TaggedRunMetadata) what_;
+        }
+        return org.tensorflow.proto.util.TaggedRunMetadata.getDefaultInstance();
+      } else {
+        if (whatCase_ == 8) {
+          return taggedRunMetadataBuilder_.getMessage();
+        }
+        return org.tensorflow.proto.util.TaggedRunMetadata.getDefaultInstance();
+      }
+    }
+    /**
+     * <pre>
+     * The metadata returned by running a session.run() call.
+     * </pre>
+     *
+     * <code>.tensorflow.TaggedRunMetadata tagged_run_metadata = 8;</code>
+     */
+    public Builder setTaggedRunMetadata(org.tensorflow.proto.util.TaggedRunMetadata value) {
+      if (taggedRunMetadataBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        what_ = value;
+        onChanged();
+      } else {
+        taggedRunMetadataBuilder_.setMessage(value);
+      }
+      whatCase_ = 8;
+      return this;
+    }
+    /**
+     * <pre>
+     * The metadata returned by running a session.run() call.
+     * </pre>
+     *
+     * <code>.tensorflow.TaggedRunMetadata tagged_run_metadata = 8;</code>
+     */
+    public Builder setTaggedRunMetadata(
+        org.tensorflow.proto.util.TaggedRunMetadata.Builder builderForValue) {
+      if (taggedRunMetadataBuilder_ == null) {
+        what_ = builderForValue.build();
+        onChanged();
+      } else {
+        taggedRunMetadataBuilder_.setMessage(builderForValue.build());
+      }
+      whatCase_ = 8;
+      return this;
+    }
+    /**
+     * <pre>
+     * The metadata returned by running a session.run() call.
+     * </pre>
+     *
+     * <code>.tensorflow.TaggedRunMetadata tagged_run_metadata = 8;</code>
+     */
+    public Builder mergeTaggedRunMetadata(org.tensorflow.proto.util.TaggedRunMetadata value) {
+      if (taggedRunMetadataBuilder_ == null) {
+        if (whatCase_ == 8 &&
+            what_ != org.tensorflow.proto.util.TaggedRunMetadata.getDefaultInstance()) {
+          what_ = org.tensorflow.proto.util.TaggedRunMetadata.newBuilder((org.tensorflow.proto.util.TaggedRunMetadata) what_)
+              .mergeFrom(value).buildPartial();
+        } else {
+          what_ = value;
+        }
+        onChanged();
+      } else {
+        if (whatCase_ == 8) {
+          taggedRunMetadataBuilder_.mergeFrom(value);
+        }
+        taggedRunMetadataBuilder_.setMessage(value);
+      }
+      whatCase_ = 8;
+      return this;
+    }
+    /**
+     * <pre>
+     * The metadata returned by running a session.run() call.
+     * </pre>
+     *
+     * <code>.tensorflow.TaggedRunMetadata tagged_run_metadata = 8;</code>
+     */
+    public Builder clearTaggedRunMetadata() {
+      if (taggedRunMetadataBuilder_ == null) {
+        if (whatCase_ == 8) {
+          whatCase_ = 0;
+          what_ = null;
+          onChanged();
+        }
+      } else {
+        if (whatCase_ == 8) {
+          whatCase_ = 0;
+          what_ = null;
+        }
+        taggedRunMetadataBuilder_.clear();
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * The metadata returned by running a session.run() call.
+     * </pre>
+     *
+     * <code>.tensorflow.TaggedRunMetadata tagged_run_metadata = 8;</code>
+     */
+    public org.tensorflow.proto.util.TaggedRunMetadata.Builder getTaggedRunMetadataBuilder() {
+      return getTaggedRunMetadataFieldBuilder().getBuilder();
+    }
+    /**
+     * <pre>
+     * The metadata returned by running a session.run() call.
+     * </pre>
+     *
+     * <code>.tensorflow.TaggedRunMetadata tagged_run_metadata = 8;</code>
+     */
+    public org.tensorflow.proto.util.TaggedRunMetadataOrBuilder getTaggedRunMetadataOrBuilder() {
+      if ((whatCase_ == 8) && (taggedRunMetadataBuilder_ != null)) {
+        return taggedRunMetadataBuilder_.getMessageOrBuilder();
+      } else {
+        if (whatCase_ == 8) {
+          return (org.tensorflow.proto.util.TaggedRunMetadata) what_;
+        }
+        return org.tensorflow.proto.util.TaggedRunMetadata.getDefaultInstance();
+      }
+    }
+    /**
+     * <pre>
+     * The metadata returned by running a session.run() call.
+     * </pre>
+     *
+     * <code>.tensorflow.TaggedRunMetadata tagged_run_metadata = 8;</code>
+     */
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.util.TaggedRunMetadata, org.tensorflow.proto.util.TaggedRunMetadata.Builder, org.tensorflow.proto.util.TaggedRunMetadataOrBuilder> 
+        getTaggedRunMetadataFieldBuilder() {
+      if (taggedRunMetadataBuilder_ == null) {
+        if (!(whatCase_ == 8)) {
+          what_ = org.tensorflow.proto.util.TaggedRunMetadata.getDefaultInstance();
+        }
+        taggedRunMetadataBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+            org.tensorflow.proto.util.TaggedRunMetadata, org.tensorflow.proto.util.TaggedRunMetadata.Builder, org.tensorflow.proto.util.TaggedRunMetadataOrBuilder>(
+                (org.tensorflow.proto.util.TaggedRunMetadata) what_,
+                getParentForChildren(),
+                isClean());
+        what_ = null;
+      }
+      whatCase_ = 8;
+      onChanged();;
+      return taggedRunMetadataBuilder_;
+    }
+
+    /**
+     * <pre>
+     * An encoded version of a MetaGraphDef.
+     * </pre>
+     *
+     * <code>bytes meta_graph_def = 9;</code>
+     */
+    public com.google.protobuf.ByteString getMetaGraphDef() {
+      if (whatCase_ == 9) {
+        return (com.google.protobuf.ByteString) what_;
+      }
+      return com.google.protobuf.ByteString.EMPTY;
+    }
+    /**
+     * <pre>
+     * An encoded version of a MetaGraphDef.
+     * </pre>
+     *
+     * <code>bytes meta_graph_def = 9;</code>
+     */
+    public Builder setMetaGraphDef(com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  whatCase_ = 9;
+      what_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * An encoded version of a MetaGraphDef.
+     * </pre>
+     *
+     * <code>bytes meta_graph_def = 9;</code>
+     */
+    public Builder clearMetaGraphDef() {
+      if (whatCase_ == 9) {
+        whatCase_ = 0;
+        what_ = null;
+        onChanged();
+      }
+      return this;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.Event)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.Event)
+  private static final org.tensorflow.proto.util.Event DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.util.Event();
+  }
+
+  public static org.tensorflow.proto.util.Event getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<Event>
+      PARSER = new com.google.protobuf.AbstractParser<Event>() {
+    @java.lang.Override
+    public Event parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new Event(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<Event> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<Event> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.util.Event getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/EventOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/EventOrBuilder.java
new file mode 100644
index 00000000000..dbe86dad114
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/EventOrBuilder.java
@@ -0,0 +1,174 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/util/event.proto
+
+package org.tensorflow.proto.util;
+
+public interface EventOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.Event)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <pre>
+   * Timestamp of the event.
+   * </pre>
+   *
+   * <code>double wall_time = 1;</code>
+   */
+  double getWallTime();
+
+  /**
+   * <pre>
+   * Global step of the event.
+   * </pre>
+   *
+   * <code>int64 step = 2;</code>
+   */
+  long getStep();
+
+  /**
+   * <pre>
+   * An event file was started, with the specified version.
+   * This is use to identify the contents of the record IO files
+   * easily.  Current version is "brain.Event:2".  All versions
+   * start with "brain.Event:".
+   * </pre>
+   *
+   * <code>string file_version = 3;</code>
+   */
+  java.lang.String getFileVersion();
+  /**
+   * <pre>
+   * An event file was started, with the specified version.
+   * This is use to identify the contents of the record IO files
+   * easily.  Current version is "brain.Event:2".  All versions
+   * start with "brain.Event:".
+   * </pre>
+   *
+   * <code>string file_version = 3;</code>
+   */
+  com.google.protobuf.ByteString
+      getFileVersionBytes();
+
+  /**
+   * <pre>
+   * An encoded version of a GraphDef.
+   * </pre>
+   *
+   * <code>bytes graph_def = 4;</code>
+   */
+  com.google.protobuf.ByteString getGraphDef();
+
+  /**
+   * <pre>
+   * A summary was generated.
+   * </pre>
+   *
+   * <code>.tensorflow.Summary summary = 5;</code>
+   */
+  boolean hasSummary();
+  /**
+   * <pre>
+   * A summary was generated.
+   * </pre>
+   *
+   * <code>.tensorflow.Summary summary = 5;</code>
+   */
+  org.tensorflow.proto.framework.Summary getSummary();
+  /**
+   * <pre>
+   * A summary was generated.
+   * </pre>
+   *
+   * <code>.tensorflow.Summary summary = 5;</code>
+   */
+  org.tensorflow.proto.framework.SummaryOrBuilder getSummaryOrBuilder();
+
+  /**
+   * <pre>
+   * The user output a log message. Not all messages are logged, only ones
+   * generated via the Python tensorboard_logging module.
+   * </pre>
+   *
+   * <code>.tensorflow.LogMessage log_message = 6;</code>
+   */
+  boolean hasLogMessage();
+  /**
+   * <pre>
+   * The user output a log message. Not all messages are logged, only ones
+   * generated via the Python tensorboard_logging module.
+   * </pre>
+   *
+   * <code>.tensorflow.LogMessage log_message = 6;</code>
+   */
+  org.tensorflow.proto.util.LogMessage getLogMessage();
+  /**
+   * <pre>
+   * The user output a log message. Not all messages are logged, only ones
+   * generated via the Python tensorboard_logging module.
+   * </pre>
+   *
+   * <code>.tensorflow.LogMessage log_message = 6;</code>
+   */
+  org.tensorflow.proto.util.LogMessageOrBuilder getLogMessageOrBuilder();
+
+  /**
+   * <pre>
+   * The state of the session which can be used for restarting after crashes.
+   * </pre>
+   *
+   * <code>.tensorflow.SessionLog session_log = 7;</code>
+   */
+  boolean hasSessionLog();
+  /**
+   * <pre>
+   * The state of the session which can be used for restarting after crashes.
+   * </pre>
+   *
+   * <code>.tensorflow.SessionLog session_log = 7;</code>
+   */
+  org.tensorflow.proto.util.SessionLog getSessionLog();
+  /**
+   * <pre>
+   * The state of the session which can be used for restarting after crashes.
+   * </pre>
+   *
+   * <code>.tensorflow.SessionLog session_log = 7;</code>
+   */
+  org.tensorflow.proto.util.SessionLogOrBuilder getSessionLogOrBuilder();
+
+  /**
+   * <pre>
+   * The metadata returned by running a session.run() call.
+   * </pre>
+   *
+   * <code>.tensorflow.TaggedRunMetadata tagged_run_metadata = 8;</code>
+   */
+  boolean hasTaggedRunMetadata();
+  /**
+   * <pre>
+   * The metadata returned by running a session.run() call.
+   * </pre>
+   *
+   * <code>.tensorflow.TaggedRunMetadata tagged_run_metadata = 8;</code>
+   */
+  org.tensorflow.proto.util.TaggedRunMetadata getTaggedRunMetadata();
+  /**
+   * <pre>
+   * The metadata returned by running a session.run() call.
+   * </pre>
+   *
+   * <code>.tensorflow.TaggedRunMetadata tagged_run_metadata = 8;</code>
+   */
+  org.tensorflow.proto.util.TaggedRunMetadataOrBuilder getTaggedRunMetadataOrBuilder();
+
+  /**
+   * <pre>
+   * An encoded version of a MetaGraphDef.
+   * </pre>
+   *
+   * <code>bytes meta_graph_def = 9;</code>
+   */
+  com.google.protobuf.ByteString getMetaGraphDef();
+
+  public org.tensorflow.proto.util.Event.WhatCase getWhatCase();
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/EventProtos.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/EventProtos.java
new file mode 100644
index 00000000000..121a7223e1d
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/EventProtos.java
@@ -0,0 +1,161 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/util/event.proto
+
+package org.tensorflow.proto.util;
+
+public final class EventProtos {
+  private EventProtos() {}
+  public static void registerAllExtensions(
+      com.google.protobuf.ExtensionRegistryLite registry) {
+  }
+
+  public static void registerAllExtensions(
+      com.google.protobuf.ExtensionRegistry registry) {
+    registerAllExtensions(
+        (com.google.protobuf.ExtensionRegistryLite) registry);
+  }
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_Event_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_Event_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_LogMessage_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_LogMessage_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_SessionLog_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_SessionLog_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_TaggedRunMetadata_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_TaggedRunMetadata_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_WatchdogConfig_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_WatchdogConfig_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_RequestedExitCode_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_RequestedExitCode_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_WorkerHeartbeatRequest_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_WorkerHeartbeatRequest_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_WorkerHeartbeatResponse_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_WorkerHeartbeatResponse_fieldAccessorTable;
+
+  public static com.google.protobuf.Descriptors.FileDescriptor
+      getDescriptor() {
+    return descriptor;
+  }
+  private static  com.google.protobuf.Descriptors.FileDescriptor
+      descriptor;
+  static {
+    java.lang.String[] descriptorData = {
+      "\n tensorflow/core/util/event.proto\022\ntens" +
+      "orflow\032\'tensorflow/core/framework/summar" +
+      "y.proto\"\273\002\n\005Event\022\021\n\twall_time\030\001 \001(\001\022\014\n\004" +
+      "step\030\002 \001(\003\022\026\n\014file_version\030\003 \001(\tH\000\022\023\n\tgr" +
+      "aph_def\030\004 \001(\014H\000\022&\n\007summary\030\005 \001(\0132\023.tenso" +
+      "rflow.SummaryH\000\022-\n\013log_message\030\006 \001(\0132\026.t" +
+      "ensorflow.LogMessageH\000\022-\n\013session_log\030\007 " +
+      "\001(\0132\026.tensorflow.SessionLogH\000\022<\n\023tagged_" +
+      "run_metadata\030\010 \001(\0132\035.tensorflow.TaggedRu" +
+      "nMetadataH\000\022\030\n\016meta_graph_def\030\t \001(\014H\000B\006\n" +
+      "\004what\"\231\001\n\nLogMessage\022+\n\005level\030\001 \001(\0162\034.te" +
+      "nsorflow.LogMessage.Level\022\017\n\007message\030\002 \001" +
+      "(\t\"M\n\005Level\022\013\n\007UNKNOWN\020\000\022\r\n\tDEBUGGING\020\n\022" +
+      "\010\n\004INFO\020\024\022\010\n\004WARN\020\036\022\t\n\005ERROR\020(\022\t\n\005FATAL\020" +
+      "2\"\266\001\n\nSessionLog\0224\n\006status\030\001 \001(\0162$.tenso" +
+      "rflow.SessionLog.SessionStatus\022\027\n\017checkp" +
+      "oint_path\030\002 \001(\t\022\013\n\003msg\030\003 \001(\t\"L\n\rSessionS" +
+      "tatus\022\026\n\022STATUS_UNSPECIFIED\020\000\022\t\n\005START\020\001" +
+      "\022\010\n\004STOP\020\002\022\016\n\nCHECKPOINT\020\003\"6\n\021TaggedRunM" +
+      "etadata\022\013\n\003tag\030\001 \001(\t\022\024\n\014run_metadata\030\002 \001" +
+      "(\014\"$\n\016WatchdogConfig\022\022\n\ntimeout_ms\030\001 \001(\003" +
+      "\"&\n\021RequestedExitCode\022\021\n\texit_code\030\001 \001(\005" +
+      "\"\266\001\n\026WorkerHeartbeatRequest\0225\n\rshutdown_" +
+      "mode\030\001 \001(\0162\036.tensorflow.WorkerShutdownMo" +
+      "de\0223\n\017watchdog_config\030\002 \001(\0132\032.tensorflow" +
+      ".WatchdogConfig\0220\n\texit_code\030\003 \001(\0132\035.ten" +
+      "sorflow.RequestedExitCode\"\203\001\n\027WorkerHear" +
+      "tbeatResponse\022/\n\rhealth_status\030\001 \001(\0162\030.t" +
+      "ensorflow.WorkerHealth\022%\n\nworker_log\030\002 \003" +
+      "(\0132\021.tensorflow.Event\022\020\n\010hostname\030\003 \001(\t*" +
+      "[\n\014WorkerHealth\022\006\n\002OK\020\000\022\034\n\030RECEIVED_SHUT" +
+      "DOWN_SIGNAL\020\001\022\022\n\016INTERNAL_ERROR\020\002\022\021\n\rSHU" +
+      "TTING_DOWN\020\003*k\n\022WorkerShutdownMode\022\013\n\007DE" +
+      "FAULT\020\000\022\022\n\016NOT_CONFIGURED\020\001\022\030\n\024WAIT_FOR_" +
+      "COORDINATOR\020\002\022\032\n\026SHUTDOWN_AFTER_TIMEOUT\020" +
+      "\003B-\n\031org.tensorflow.proto.utilB\013EventPro" +
+      "tosP\001\370\001\001b\006proto3"
+    };
+    descriptor = com.google.protobuf.Descriptors.FileDescriptor
+      .internalBuildGeneratedFileFrom(descriptorData,
+        new com.google.protobuf.Descriptors.FileDescriptor[] {
+          org.tensorflow.proto.framework.SummaryProtos.getDescriptor(),
+        });
+    internal_static_tensorflow_Event_descriptor =
+      getDescriptor().getMessageTypes().get(0);
+    internal_static_tensorflow_Event_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_Event_descriptor,
+        new java.lang.String[] { "WallTime", "Step", "FileVersion", "GraphDef", "Summary", "LogMessage", "SessionLog", "TaggedRunMetadata", "MetaGraphDef", "What", });
+    internal_static_tensorflow_LogMessage_descriptor =
+      getDescriptor().getMessageTypes().get(1);
+    internal_static_tensorflow_LogMessage_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_LogMessage_descriptor,
+        new java.lang.String[] { "Level", "Message", });
+    internal_static_tensorflow_SessionLog_descriptor =
+      getDescriptor().getMessageTypes().get(2);
+    internal_static_tensorflow_SessionLog_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_SessionLog_descriptor,
+        new java.lang.String[] { "Status", "CheckpointPath", "Msg", });
+    internal_static_tensorflow_TaggedRunMetadata_descriptor =
+      getDescriptor().getMessageTypes().get(3);
+    internal_static_tensorflow_TaggedRunMetadata_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_TaggedRunMetadata_descriptor,
+        new java.lang.String[] { "Tag", "RunMetadata", });
+    internal_static_tensorflow_WatchdogConfig_descriptor =
+      getDescriptor().getMessageTypes().get(4);
+    internal_static_tensorflow_WatchdogConfig_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_WatchdogConfig_descriptor,
+        new java.lang.String[] { "TimeoutMs", });
+    internal_static_tensorflow_RequestedExitCode_descriptor =
+      getDescriptor().getMessageTypes().get(5);
+    internal_static_tensorflow_RequestedExitCode_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_RequestedExitCode_descriptor,
+        new java.lang.String[] { "ExitCode", });
+    internal_static_tensorflow_WorkerHeartbeatRequest_descriptor =
+      getDescriptor().getMessageTypes().get(6);
+    internal_static_tensorflow_WorkerHeartbeatRequest_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_WorkerHeartbeatRequest_descriptor,
+        new java.lang.String[] { "ShutdownMode", "WatchdogConfig", "ExitCode", });
+    internal_static_tensorflow_WorkerHeartbeatResponse_descriptor =
+      getDescriptor().getMessageTypes().get(7);
+    internal_static_tensorflow_WorkerHeartbeatResponse_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_WorkerHeartbeatResponse_descriptor,
+        new java.lang.String[] { "HealthStatus", "WorkerLog", "Hostname", });
+    org.tensorflow.proto.framework.SummaryProtos.getDescriptor();
+  }
+
+  // @@protoc_insertion_point(outer_class_scope)
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/Execution.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/Execution.java
new file mode 100644
index 00000000000..42d3a1d835a
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/Execution.java
@@ -0,0 +1,2259 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/debug_event.proto
+
+package org.tensorflow.proto.util;
+
+/**
+ * <pre>
+ * Data relating to the eager execution of an op or a Graph.
+ * For a op that generates N output tensors (N &gt;= 0), only one
+ * Execution proto will be used to describe the execution event.
+ * </pre>
+ *
+ * Protobuf type {@code tensorflow.Execution}
+ */
+public  final class Execution extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.Execution)
+    ExecutionOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use Execution.newBuilder() to construct.
+  private Execution(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private Execution() {
+    opType_ = "";
+    graphId_ = "";
+    inputTensorIds_ = emptyLongList();
+    outputTensorIds_ = emptyLongList();
+    tensorDebugMode_ = 0;
+    tensorProtos_ = java.util.Collections.emptyList();
+    outputTensorDeviceIds_ = emptyIntList();
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new Execution();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private Execution(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    int mutable_bitField0_ = 0;
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 10: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            opType_ = s;
+            break;
+          }
+          case 16: {
+
+            numOutputs_ = input.readInt32();
+            break;
+          }
+          case 26: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            graphId_ = s;
+            break;
+          }
+          case 32: {
+            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
+              inputTensorIds_ = newLongList();
+              mutable_bitField0_ |= 0x00000001;
+            }
+            inputTensorIds_.addLong(input.readInt64());
+            break;
+          }
+          case 34: {
+            int length = input.readRawVarint32();
+            int limit = input.pushLimit(length);
+            if (!((mutable_bitField0_ & 0x00000001) != 0) && input.getBytesUntilLimit() > 0) {
+              inputTensorIds_ = newLongList();
+              mutable_bitField0_ |= 0x00000001;
+            }
+            while (input.getBytesUntilLimit() > 0) {
+              inputTensorIds_.addLong(input.readInt64());
+            }
+            input.popLimit(limit);
+            break;
+          }
+          case 40: {
+            if (!((mutable_bitField0_ & 0x00000002) != 0)) {
+              outputTensorIds_ = newLongList();
+              mutable_bitField0_ |= 0x00000002;
+            }
+            outputTensorIds_.addLong(input.readInt64());
+            break;
+          }
+          case 42: {
+            int length = input.readRawVarint32();
+            int limit = input.pushLimit(length);
+            if (!((mutable_bitField0_ & 0x00000002) != 0) && input.getBytesUntilLimit() > 0) {
+              outputTensorIds_ = newLongList();
+              mutable_bitField0_ |= 0x00000002;
+            }
+            while (input.getBytesUntilLimit() > 0) {
+              outputTensorIds_.addLong(input.readInt64());
+            }
+            input.popLimit(limit);
+            break;
+          }
+          case 48: {
+            int rawValue = input.readEnum();
+
+            tensorDebugMode_ = rawValue;
+            break;
+          }
+          case 58: {
+            if (!((mutable_bitField0_ & 0x00000004) != 0)) {
+              tensorProtos_ = new java.util.ArrayList<org.tensorflow.proto.framework.TensorProto>();
+              mutable_bitField0_ |= 0x00000004;
+            }
+            tensorProtos_.add(
+                input.readMessage(org.tensorflow.proto.framework.TensorProto.parser(), extensionRegistry));
+            break;
+          }
+          case 66: {
+            org.tensorflow.proto.util.CodeLocation.Builder subBuilder = null;
+            if (codeLocation_ != null) {
+              subBuilder = codeLocation_.toBuilder();
+            }
+            codeLocation_ = input.readMessage(org.tensorflow.proto.util.CodeLocation.parser(), extensionRegistry);
+            if (subBuilder != null) {
+              subBuilder.mergeFrom(codeLocation_);
+              codeLocation_ = subBuilder.buildPartial();
+            }
+
+            break;
+          }
+          case 72: {
+            if (!((mutable_bitField0_ & 0x00000008) != 0)) {
+              outputTensorDeviceIds_ = newIntList();
+              mutable_bitField0_ |= 0x00000008;
+            }
+            outputTensorDeviceIds_.addInt(input.readInt32());
+            break;
+          }
+          case 74: {
+            int length = input.readRawVarint32();
+            int limit = input.pushLimit(length);
+            if (!((mutable_bitField0_ & 0x00000008) != 0) && input.getBytesUntilLimit() > 0) {
+              outputTensorDeviceIds_ = newIntList();
+              mutable_bitField0_ |= 0x00000008;
+            }
+            while (input.getBytesUntilLimit() > 0) {
+              outputTensorDeviceIds_.addInt(input.readInt32());
+            }
+            input.popLimit(limit);
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      if (((mutable_bitField0_ & 0x00000001) != 0)) {
+        inputTensorIds_.makeImmutable(); // C
+      }
+      if (((mutable_bitField0_ & 0x00000002) != 0)) {
+        outputTensorIds_.makeImmutable(); // C
+      }
+      if (((mutable_bitField0_ & 0x00000004) != 0)) {
+        tensorProtos_ = java.util.Collections.unmodifiableList(tensorProtos_);
+      }
+      if (((mutable_bitField0_ & 0x00000008) != 0)) {
+        outputTensorDeviceIds_.makeImmutable(); // C
+      }
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.util.DebugEventProtos.internal_static_tensorflow_Execution_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.util.DebugEventProtos.internal_static_tensorflow_Execution_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.util.Execution.class, org.tensorflow.proto.util.Execution.Builder.class);
+  }
+
+  public static final int OP_TYPE_FIELD_NUMBER = 1;
+  private volatile java.lang.Object opType_;
+  /**
+   * <pre>
+   * Op type (e.g., "MatMul").
+   * In the case of a Graph, this is the name of the Graph.
+   * </pre>
+   *
+   * <code>string op_type = 1;</code>
+   */
+  public java.lang.String getOpType() {
+    java.lang.Object ref = opType_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      opType_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * Op type (e.g., "MatMul").
+   * In the case of a Graph, this is the name of the Graph.
+   * </pre>
+   *
+   * <code>string op_type = 1;</code>
+   */
+  public com.google.protobuf.ByteString
+      getOpTypeBytes() {
+    java.lang.Object ref = opType_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      opType_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int NUM_OUTPUTS_FIELD_NUMBER = 2;
+  private int numOutputs_;
+  /**
+   * <pre>
+   * Number of output tensors.
+   * </pre>
+   *
+   * <code>int32 num_outputs = 2;</code>
+   */
+  public int getNumOutputs() {
+    return numOutputs_;
+  }
+
+  public static final int GRAPH_ID_FIELD_NUMBER = 3;
+  private volatile java.lang.Object graphId_;
+  /**
+   * <pre>
+   * The graph that's executed: applicable only to the eager
+   * execution of a FuncGraph.
+   * </pre>
+   *
+   * <code>string graph_id = 3;</code>
+   */
+  public java.lang.String getGraphId() {
+    java.lang.Object ref = graphId_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      graphId_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * The graph that's executed: applicable only to the eager
+   * execution of a FuncGraph.
+   * </pre>
+   *
+   * <code>string graph_id = 3;</code>
+   */
+  public com.google.protobuf.ByteString
+      getGraphIdBytes() {
+    java.lang.Object ref = graphId_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      graphId_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int INPUT_TENSOR_IDS_FIELD_NUMBER = 4;
+  private com.google.protobuf.Internal.LongList inputTensorIds_;
+  /**
+   * <pre>
+   * IDs of the input tensors (if available).
+   * </pre>
+   *
+   * <code>repeated int64 input_tensor_ids = 4;</code>
+   */
+  public java.util.List<java.lang.Long>
+      getInputTensorIdsList() {
+    return inputTensorIds_;
+  }
+  /**
+   * <pre>
+   * IDs of the input tensors (if available).
+   * </pre>
+   *
+   * <code>repeated int64 input_tensor_ids = 4;</code>
+   */
+  public int getInputTensorIdsCount() {
+    return inputTensorIds_.size();
+  }
+  /**
+   * <pre>
+   * IDs of the input tensors (if available).
+   * </pre>
+   *
+   * <code>repeated int64 input_tensor_ids = 4;</code>
+   */
+  public long getInputTensorIds(int index) {
+    return inputTensorIds_.getLong(index);
+  }
+  private int inputTensorIdsMemoizedSerializedSize = -1;
+
+  public static final int OUTPUT_TENSOR_IDS_FIELD_NUMBER = 5;
+  private com.google.protobuf.Internal.LongList outputTensorIds_;
+  /**
+   * <pre>
+   * IDs of the output tensors (if availbable).
+   * If specified, must have the same length as tensor_protos.
+   * </pre>
+   *
+   * <code>repeated int64 output_tensor_ids = 5;</code>
+   */
+  public java.util.List<java.lang.Long>
+      getOutputTensorIdsList() {
+    return outputTensorIds_;
+  }
+  /**
+   * <pre>
+   * IDs of the output tensors (if availbable).
+   * If specified, must have the same length as tensor_protos.
+   * </pre>
+   *
+   * <code>repeated int64 output_tensor_ids = 5;</code>
+   */
+  public int getOutputTensorIdsCount() {
+    return outputTensorIds_.size();
+  }
+  /**
+   * <pre>
+   * IDs of the output tensors (if availbable).
+   * If specified, must have the same length as tensor_protos.
+   * </pre>
+   *
+   * <code>repeated int64 output_tensor_ids = 5;</code>
+   */
+  public long getOutputTensorIds(int index) {
+    return outputTensorIds_.getLong(index);
+  }
+  private int outputTensorIdsMemoizedSerializedSize = -1;
+
+  public static final int TENSOR_DEBUG_MODE_FIELD_NUMBER = 6;
+  private int tensorDebugMode_;
+  /**
+   * <pre>
+   * Type of the tensor value encapsulated in this proto.
+   * </pre>
+   *
+   * <code>.tensorflow.TensorDebugMode tensor_debug_mode = 6;</code>
+   */
+  public int getTensorDebugModeValue() {
+    return tensorDebugMode_;
+  }
+  /**
+   * <pre>
+   * Type of the tensor value encapsulated in this proto.
+   * </pre>
+   *
+   * <code>.tensorflow.TensorDebugMode tensor_debug_mode = 6;</code>
+   */
+  public org.tensorflow.proto.util.TensorDebugMode getTensorDebugMode() {
+    @SuppressWarnings("deprecation")
+    org.tensorflow.proto.util.TensorDebugMode result = org.tensorflow.proto.util.TensorDebugMode.valueOf(tensorDebugMode_);
+    return result == null ? org.tensorflow.proto.util.TensorDebugMode.UNRECOGNIZED : result;
+  }
+
+  public static final int TENSOR_PROTOS_FIELD_NUMBER = 7;
+  private java.util.List<org.tensorflow.proto.framework.TensorProto> tensorProtos_;
+  /**
+   * <pre>
+   * Output Tensor values in the type described by `tensor_value_type`.
+   * The length of this should match `num_outputs`.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.TensorProto tensor_protos = 7;</code>
+   */
+  public java.util.List<org.tensorflow.proto.framework.TensorProto> getTensorProtosList() {
+    return tensorProtos_;
+  }
+  /**
+   * <pre>
+   * Output Tensor values in the type described by `tensor_value_type`.
+   * The length of this should match `num_outputs`.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.TensorProto tensor_protos = 7;</code>
+   */
+  public java.util.List<? extends org.tensorflow.proto.framework.TensorProtoOrBuilder> 
+      getTensorProtosOrBuilderList() {
+    return tensorProtos_;
+  }
+  /**
+   * <pre>
+   * Output Tensor values in the type described by `tensor_value_type`.
+   * The length of this should match `num_outputs`.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.TensorProto tensor_protos = 7;</code>
+   */
+  public int getTensorProtosCount() {
+    return tensorProtos_.size();
+  }
+  /**
+   * <pre>
+   * Output Tensor values in the type described by `tensor_value_type`.
+   * The length of this should match `num_outputs`.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.TensorProto tensor_protos = 7;</code>
+   */
+  public org.tensorflow.proto.framework.TensorProto getTensorProtos(int index) {
+    return tensorProtos_.get(index);
+  }
+  /**
+   * <pre>
+   * Output Tensor values in the type described by `tensor_value_type`.
+   * The length of this should match `num_outputs`.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.TensorProto tensor_protos = 7;</code>
+   */
+  public org.tensorflow.proto.framework.TensorProtoOrBuilder getTensorProtosOrBuilder(
+      int index) {
+    return tensorProtos_.get(index);
+  }
+
+  public static final int CODE_LOCATION_FIELD_NUMBER = 8;
+  private org.tensorflow.proto.util.CodeLocation codeLocation_;
+  /**
+   * <pre>
+   * Stack trace of the eager execution.
+   * </pre>
+   *
+   * <code>.tensorflow.CodeLocation code_location = 8;</code>
+   */
+  public boolean hasCodeLocation() {
+    return codeLocation_ != null;
+  }
+  /**
+   * <pre>
+   * Stack trace of the eager execution.
+   * </pre>
+   *
+   * <code>.tensorflow.CodeLocation code_location = 8;</code>
+   */
+  public org.tensorflow.proto.util.CodeLocation getCodeLocation() {
+    return codeLocation_ == null ? org.tensorflow.proto.util.CodeLocation.getDefaultInstance() : codeLocation_;
+  }
+  /**
+   * <pre>
+   * Stack trace of the eager execution.
+   * </pre>
+   *
+   * <code>.tensorflow.CodeLocation code_location = 8;</code>
+   */
+  public org.tensorflow.proto.util.CodeLocationOrBuilder getCodeLocationOrBuilder() {
+    return getCodeLocation();
+  }
+
+  public static final int OUTPUT_TENSOR_DEVICE_IDS_FIELD_NUMBER = 9;
+  private com.google.protobuf.Internal.IntList outputTensorDeviceIds_;
+  /**
+   * <pre>
+   * Debugged-generated IDs of the devices on which the output tensors reside.
+   * To look up details about the device (e.g., name), cross-reference this
+   * field with the DebuggedDevice messages.
+   * </pre>
+   *
+   * <code>repeated int32 output_tensor_device_ids = 9;</code>
+   */
+  public java.util.List<java.lang.Integer>
+      getOutputTensorDeviceIdsList() {
+    return outputTensorDeviceIds_;
+  }
+  /**
+   * <pre>
+   * Debugged-generated IDs of the devices on which the output tensors reside.
+   * To look up details about the device (e.g., name), cross-reference this
+   * field with the DebuggedDevice messages.
+   * </pre>
+   *
+   * <code>repeated int32 output_tensor_device_ids = 9;</code>
+   */
+  public int getOutputTensorDeviceIdsCount() {
+    return outputTensorDeviceIds_.size();
+  }
+  /**
+   * <pre>
+   * Debugged-generated IDs of the devices on which the output tensors reside.
+   * To look up details about the device (e.g., name), cross-reference this
+   * field with the DebuggedDevice messages.
+   * </pre>
+   *
+   * <code>repeated int32 output_tensor_device_ids = 9;</code>
+   */
+  public int getOutputTensorDeviceIds(int index) {
+    return outputTensorDeviceIds_.getInt(index);
+  }
+  private int outputTensorDeviceIdsMemoizedSerializedSize = -1;
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    getSerializedSize();
+    if (!getOpTypeBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 1, opType_);
+    }
+    if (numOutputs_ != 0) {
+      output.writeInt32(2, numOutputs_);
+    }
+    if (!getGraphIdBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 3, graphId_);
+    }
+    if (getInputTensorIdsList().size() > 0) {
+      output.writeUInt32NoTag(34);
+      output.writeUInt32NoTag(inputTensorIdsMemoizedSerializedSize);
+    }
+    for (int i = 0; i < inputTensorIds_.size(); i++) {
+      output.writeInt64NoTag(inputTensorIds_.getLong(i));
+    }
+    if (getOutputTensorIdsList().size() > 0) {
+      output.writeUInt32NoTag(42);
+      output.writeUInt32NoTag(outputTensorIdsMemoizedSerializedSize);
+    }
+    for (int i = 0; i < outputTensorIds_.size(); i++) {
+      output.writeInt64NoTag(outputTensorIds_.getLong(i));
+    }
+    if (tensorDebugMode_ != org.tensorflow.proto.util.TensorDebugMode.UNSPECIFIED.getNumber()) {
+      output.writeEnum(6, tensorDebugMode_);
+    }
+    for (int i = 0; i < tensorProtos_.size(); i++) {
+      output.writeMessage(7, tensorProtos_.get(i));
+    }
+    if (codeLocation_ != null) {
+      output.writeMessage(8, getCodeLocation());
+    }
+    if (getOutputTensorDeviceIdsList().size() > 0) {
+      output.writeUInt32NoTag(74);
+      output.writeUInt32NoTag(outputTensorDeviceIdsMemoizedSerializedSize);
+    }
+    for (int i = 0; i < outputTensorDeviceIds_.size(); i++) {
+      output.writeInt32NoTag(outputTensorDeviceIds_.getInt(i));
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (!getOpTypeBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, opType_);
+    }
+    if (numOutputs_ != 0) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt32Size(2, numOutputs_);
+    }
+    if (!getGraphIdBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(3, graphId_);
+    }
+    {
+      int dataSize = 0;
+      for (int i = 0; i < inputTensorIds_.size(); i++) {
+        dataSize += com.google.protobuf.CodedOutputStream
+          .computeInt64SizeNoTag(inputTensorIds_.getLong(i));
+      }
+      size += dataSize;
+      if (!getInputTensorIdsList().isEmpty()) {
+        size += 1;
+        size += com.google.protobuf.CodedOutputStream
+            .computeInt32SizeNoTag(dataSize);
+      }
+      inputTensorIdsMemoizedSerializedSize = dataSize;
+    }
+    {
+      int dataSize = 0;
+      for (int i = 0; i < outputTensorIds_.size(); i++) {
+        dataSize += com.google.protobuf.CodedOutputStream
+          .computeInt64SizeNoTag(outputTensorIds_.getLong(i));
+      }
+      size += dataSize;
+      if (!getOutputTensorIdsList().isEmpty()) {
+        size += 1;
+        size += com.google.protobuf.CodedOutputStream
+            .computeInt32SizeNoTag(dataSize);
+      }
+      outputTensorIdsMemoizedSerializedSize = dataSize;
+    }
+    if (tensorDebugMode_ != org.tensorflow.proto.util.TensorDebugMode.UNSPECIFIED.getNumber()) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeEnumSize(6, tensorDebugMode_);
+    }
+    for (int i = 0; i < tensorProtos_.size(); i++) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(7, tensorProtos_.get(i));
+    }
+    if (codeLocation_ != null) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(8, getCodeLocation());
+    }
+    {
+      int dataSize = 0;
+      for (int i = 0; i < outputTensorDeviceIds_.size(); i++) {
+        dataSize += com.google.protobuf.CodedOutputStream
+          .computeInt32SizeNoTag(outputTensorDeviceIds_.getInt(i));
+      }
+      size += dataSize;
+      if (!getOutputTensorDeviceIdsList().isEmpty()) {
+        size += 1;
+        size += com.google.protobuf.CodedOutputStream
+            .computeInt32SizeNoTag(dataSize);
+      }
+      outputTensorDeviceIdsMemoizedSerializedSize = dataSize;
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.util.Execution)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.util.Execution other = (org.tensorflow.proto.util.Execution) obj;
+
+    if (!getOpType()
+        .equals(other.getOpType())) return false;
+    if (getNumOutputs()
+        != other.getNumOutputs()) return false;
+    if (!getGraphId()
+        .equals(other.getGraphId())) return false;
+    if (!getInputTensorIdsList()
+        .equals(other.getInputTensorIdsList())) return false;
+    if (!getOutputTensorIdsList()
+        .equals(other.getOutputTensorIdsList())) return false;
+    if (tensorDebugMode_ != other.tensorDebugMode_) return false;
+    if (!getTensorProtosList()
+        .equals(other.getTensorProtosList())) return false;
+    if (hasCodeLocation() != other.hasCodeLocation()) return false;
+    if (hasCodeLocation()) {
+      if (!getCodeLocation()
+          .equals(other.getCodeLocation())) return false;
+    }
+    if (!getOutputTensorDeviceIdsList()
+        .equals(other.getOutputTensorDeviceIdsList())) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    hash = (37 * hash) + OP_TYPE_FIELD_NUMBER;
+    hash = (53 * hash) + getOpType().hashCode();
+    hash = (37 * hash) + NUM_OUTPUTS_FIELD_NUMBER;
+    hash = (53 * hash) + getNumOutputs();
+    hash = (37 * hash) + GRAPH_ID_FIELD_NUMBER;
+    hash = (53 * hash) + getGraphId().hashCode();
+    if (getInputTensorIdsCount() > 0) {
+      hash = (37 * hash) + INPUT_TENSOR_IDS_FIELD_NUMBER;
+      hash = (53 * hash) + getInputTensorIdsList().hashCode();
+    }
+    if (getOutputTensorIdsCount() > 0) {
+      hash = (37 * hash) + OUTPUT_TENSOR_IDS_FIELD_NUMBER;
+      hash = (53 * hash) + getOutputTensorIdsList().hashCode();
+    }
+    hash = (37 * hash) + TENSOR_DEBUG_MODE_FIELD_NUMBER;
+    hash = (53 * hash) + tensorDebugMode_;
+    if (getTensorProtosCount() > 0) {
+      hash = (37 * hash) + TENSOR_PROTOS_FIELD_NUMBER;
+      hash = (53 * hash) + getTensorProtosList().hashCode();
+    }
+    if (hasCodeLocation()) {
+      hash = (37 * hash) + CODE_LOCATION_FIELD_NUMBER;
+      hash = (53 * hash) + getCodeLocation().hashCode();
+    }
+    if (getOutputTensorDeviceIdsCount() > 0) {
+      hash = (37 * hash) + OUTPUT_TENSOR_DEVICE_IDS_FIELD_NUMBER;
+      hash = (53 * hash) + getOutputTensorDeviceIdsList().hashCode();
+    }
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.util.Execution parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.Execution parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.Execution parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.Execution parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.Execution parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.Execution parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.Execution parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.Execution parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.Execution parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.Execution parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.Execution parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.Execution parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.util.Execution prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * <pre>
+   * Data relating to the eager execution of an op or a Graph.
+   * For a op that generates N output tensors (N &gt;= 0), only one
+   * Execution proto will be used to describe the execution event.
+   * </pre>
+   *
+   * Protobuf type {@code tensorflow.Execution}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.Execution)
+      org.tensorflow.proto.util.ExecutionOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.util.DebugEventProtos.internal_static_tensorflow_Execution_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.util.DebugEventProtos.internal_static_tensorflow_Execution_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.util.Execution.class, org.tensorflow.proto.util.Execution.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.util.Execution.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+        getTensorProtosFieldBuilder();
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      opType_ = "";
+
+      numOutputs_ = 0;
+
+      graphId_ = "";
+
+      inputTensorIds_ = emptyLongList();
+      bitField0_ = (bitField0_ & ~0x00000001);
+      outputTensorIds_ = emptyLongList();
+      bitField0_ = (bitField0_ & ~0x00000002);
+      tensorDebugMode_ = 0;
+
+      if (tensorProtosBuilder_ == null) {
+        tensorProtos_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000004);
+      } else {
+        tensorProtosBuilder_.clear();
+      }
+      if (codeLocationBuilder_ == null) {
+        codeLocation_ = null;
+      } else {
+        codeLocation_ = null;
+        codeLocationBuilder_ = null;
+      }
+      outputTensorDeviceIds_ = emptyIntList();
+      bitField0_ = (bitField0_ & ~0x00000008);
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.util.DebugEventProtos.internal_static_tensorflow_Execution_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.Execution getDefaultInstanceForType() {
+      return org.tensorflow.proto.util.Execution.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.Execution build() {
+      org.tensorflow.proto.util.Execution result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.Execution buildPartial() {
+      org.tensorflow.proto.util.Execution result = new org.tensorflow.proto.util.Execution(this);
+      int from_bitField0_ = bitField0_;
+      result.opType_ = opType_;
+      result.numOutputs_ = numOutputs_;
+      result.graphId_ = graphId_;
+      if (((bitField0_ & 0x00000001) != 0)) {
+        inputTensorIds_.makeImmutable();
+        bitField0_ = (bitField0_ & ~0x00000001);
+      }
+      result.inputTensorIds_ = inputTensorIds_;
+      if (((bitField0_ & 0x00000002) != 0)) {
+        outputTensorIds_.makeImmutable();
+        bitField0_ = (bitField0_ & ~0x00000002);
+      }
+      result.outputTensorIds_ = outputTensorIds_;
+      result.tensorDebugMode_ = tensorDebugMode_;
+      if (tensorProtosBuilder_ == null) {
+        if (((bitField0_ & 0x00000004) != 0)) {
+          tensorProtos_ = java.util.Collections.unmodifiableList(tensorProtos_);
+          bitField0_ = (bitField0_ & ~0x00000004);
+        }
+        result.tensorProtos_ = tensorProtos_;
+      } else {
+        result.tensorProtos_ = tensorProtosBuilder_.build();
+      }
+      if (codeLocationBuilder_ == null) {
+        result.codeLocation_ = codeLocation_;
+      } else {
+        result.codeLocation_ = codeLocationBuilder_.build();
+      }
+      if (((bitField0_ & 0x00000008) != 0)) {
+        outputTensorDeviceIds_.makeImmutable();
+        bitField0_ = (bitField0_ & ~0x00000008);
+      }
+      result.outputTensorDeviceIds_ = outputTensorDeviceIds_;
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.util.Execution) {
+        return mergeFrom((org.tensorflow.proto.util.Execution)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.util.Execution other) {
+      if (other == org.tensorflow.proto.util.Execution.getDefaultInstance()) return this;
+      if (!other.getOpType().isEmpty()) {
+        opType_ = other.opType_;
+        onChanged();
+      }
+      if (other.getNumOutputs() != 0) {
+        setNumOutputs(other.getNumOutputs());
+      }
+      if (!other.getGraphId().isEmpty()) {
+        graphId_ = other.graphId_;
+        onChanged();
+      }
+      if (!other.inputTensorIds_.isEmpty()) {
+        if (inputTensorIds_.isEmpty()) {
+          inputTensorIds_ = other.inputTensorIds_;
+          bitField0_ = (bitField0_ & ~0x00000001);
+        } else {
+          ensureInputTensorIdsIsMutable();
+          inputTensorIds_.addAll(other.inputTensorIds_);
+        }
+        onChanged();
+      }
+      if (!other.outputTensorIds_.isEmpty()) {
+        if (outputTensorIds_.isEmpty()) {
+          outputTensorIds_ = other.outputTensorIds_;
+          bitField0_ = (bitField0_ & ~0x00000002);
+        } else {
+          ensureOutputTensorIdsIsMutable();
+          outputTensorIds_.addAll(other.outputTensorIds_);
+        }
+        onChanged();
+      }
+      if (other.tensorDebugMode_ != 0) {
+        setTensorDebugModeValue(other.getTensorDebugModeValue());
+      }
+      if (tensorProtosBuilder_ == null) {
+        if (!other.tensorProtos_.isEmpty()) {
+          if (tensorProtos_.isEmpty()) {
+            tensorProtos_ = other.tensorProtos_;
+            bitField0_ = (bitField0_ & ~0x00000004);
+          } else {
+            ensureTensorProtosIsMutable();
+            tensorProtos_.addAll(other.tensorProtos_);
+          }
+          onChanged();
+        }
+      } else {
+        if (!other.tensorProtos_.isEmpty()) {
+          if (tensorProtosBuilder_.isEmpty()) {
+            tensorProtosBuilder_.dispose();
+            tensorProtosBuilder_ = null;
+            tensorProtos_ = other.tensorProtos_;
+            bitField0_ = (bitField0_ & ~0x00000004);
+            tensorProtosBuilder_ = 
+              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
+                 getTensorProtosFieldBuilder() : null;
+          } else {
+            tensorProtosBuilder_.addAllMessages(other.tensorProtos_);
+          }
+        }
+      }
+      if (other.hasCodeLocation()) {
+        mergeCodeLocation(other.getCodeLocation());
+      }
+      if (!other.outputTensorDeviceIds_.isEmpty()) {
+        if (outputTensorDeviceIds_.isEmpty()) {
+          outputTensorDeviceIds_ = other.outputTensorDeviceIds_;
+          bitField0_ = (bitField0_ & ~0x00000008);
+        } else {
+          ensureOutputTensorDeviceIdsIsMutable();
+          outputTensorDeviceIds_.addAll(other.outputTensorDeviceIds_);
+        }
+        onChanged();
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.util.Execution parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.util.Execution) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+    private int bitField0_;
+
+    private java.lang.Object opType_ = "";
+    /**
+     * <pre>
+     * Op type (e.g., "MatMul").
+     * In the case of a Graph, this is the name of the Graph.
+     * </pre>
+     *
+     * <code>string op_type = 1;</code>
+     */
+    public java.lang.String getOpType() {
+      java.lang.Object ref = opType_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        opType_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Op type (e.g., "MatMul").
+     * In the case of a Graph, this is the name of the Graph.
+     * </pre>
+     *
+     * <code>string op_type = 1;</code>
+     */
+    public com.google.protobuf.ByteString
+        getOpTypeBytes() {
+      java.lang.Object ref = opType_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        opType_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Op type (e.g., "MatMul").
+     * In the case of a Graph, this is the name of the Graph.
+     * </pre>
+     *
+     * <code>string op_type = 1;</code>
+     */
+    public Builder setOpType(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      opType_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Op type (e.g., "MatMul").
+     * In the case of a Graph, this is the name of the Graph.
+     * </pre>
+     *
+     * <code>string op_type = 1;</code>
+     */
+    public Builder clearOpType() {
+      
+      opType_ = getDefaultInstance().getOpType();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Op type (e.g., "MatMul").
+     * In the case of a Graph, this is the name of the Graph.
+     * </pre>
+     *
+     * <code>string op_type = 1;</code>
+     */
+    public Builder setOpTypeBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      opType_ = value;
+      onChanged();
+      return this;
+    }
+
+    private int numOutputs_ ;
+    /**
+     * <pre>
+     * Number of output tensors.
+     * </pre>
+     *
+     * <code>int32 num_outputs = 2;</code>
+     */
+    public int getNumOutputs() {
+      return numOutputs_;
+    }
+    /**
+     * <pre>
+     * Number of output tensors.
+     * </pre>
+     *
+     * <code>int32 num_outputs = 2;</code>
+     */
+    public Builder setNumOutputs(int value) {
+      
+      numOutputs_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Number of output tensors.
+     * </pre>
+     *
+     * <code>int32 num_outputs = 2;</code>
+     */
+    public Builder clearNumOutputs() {
+      
+      numOutputs_ = 0;
+      onChanged();
+      return this;
+    }
+
+    private java.lang.Object graphId_ = "";
+    /**
+     * <pre>
+     * The graph that's executed: applicable only to the eager
+     * execution of a FuncGraph.
+     * </pre>
+     *
+     * <code>string graph_id = 3;</code>
+     */
+    public java.lang.String getGraphId() {
+      java.lang.Object ref = graphId_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        graphId_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * The graph that's executed: applicable only to the eager
+     * execution of a FuncGraph.
+     * </pre>
+     *
+     * <code>string graph_id = 3;</code>
+     */
+    public com.google.protobuf.ByteString
+        getGraphIdBytes() {
+      java.lang.Object ref = graphId_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        graphId_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * The graph that's executed: applicable only to the eager
+     * execution of a FuncGraph.
+     * </pre>
+     *
+     * <code>string graph_id = 3;</code>
+     */
+    public Builder setGraphId(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      graphId_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * The graph that's executed: applicable only to the eager
+     * execution of a FuncGraph.
+     * </pre>
+     *
+     * <code>string graph_id = 3;</code>
+     */
+    public Builder clearGraphId() {
+      
+      graphId_ = getDefaultInstance().getGraphId();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * The graph that's executed: applicable only to the eager
+     * execution of a FuncGraph.
+     * </pre>
+     *
+     * <code>string graph_id = 3;</code>
+     */
+    public Builder setGraphIdBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      graphId_ = value;
+      onChanged();
+      return this;
+    }
+
+    private com.google.protobuf.Internal.LongList inputTensorIds_ = emptyLongList();
+    private void ensureInputTensorIdsIsMutable() {
+      if (!((bitField0_ & 0x00000001) != 0)) {
+        inputTensorIds_ = mutableCopy(inputTensorIds_);
+        bitField0_ |= 0x00000001;
+       }
+    }
+    /**
+     * <pre>
+     * IDs of the input tensors (if available).
+     * </pre>
+     *
+     * <code>repeated int64 input_tensor_ids = 4;</code>
+     */
+    public java.util.List<java.lang.Long>
+        getInputTensorIdsList() {
+      return ((bitField0_ & 0x00000001) != 0) ?
+               java.util.Collections.unmodifiableList(inputTensorIds_) : inputTensorIds_;
+    }
+    /**
+     * <pre>
+     * IDs of the input tensors (if available).
+     * </pre>
+     *
+     * <code>repeated int64 input_tensor_ids = 4;</code>
+     */
+    public int getInputTensorIdsCount() {
+      return inputTensorIds_.size();
+    }
+    /**
+     * <pre>
+     * IDs of the input tensors (if available).
+     * </pre>
+     *
+     * <code>repeated int64 input_tensor_ids = 4;</code>
+     */
+    public long getInputTensorIds(int index) {
+      return inputTensorIds_.getLong(index);
+    }
+    /**
+     * <pre>
+     * IDs of the input tensors (if available).
+     * </pre>
+     *
+     * <code>repeated int64 input_tensor_ids = 4;</code>
+     */
+    public Builder setInputTensorIds(
+        int index, long value) {
+      ensureInputTensorIdsIsMutable();
+      inputTensorIds_.setLong(index, value);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * IDs of the input tensors (if available).
+     * </pre>
+     *
+     * <code>repeated int64 input_tensor_ids = 4;</code>
+     */
+    public Builder addInputTensorIds(long value) {
+      ensureInputTensorIdsIsMutable();
+      inputTensorIds_.addLong(value);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * IDs of the input tensors (if available).
+     * </pre>
+     *
+     * <code>repeated int64 input_tensor_ids = 4;</code>
+     */
+    public Builder addAllInputTensorIds(
+        java.lang.Iterable<? extends java.lang.Long> values) {
+      ensureInputTensorIdsIsMutable();
+      com.google.protobuf.AbstractMessageLite.Builder.addAll(
+          values, inputTensorIds_);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * IDs of the input tensors (if available).
+     * </pre>
+     *
+     * <code>repeated int64 input_tensor_ids = 4;</code>
+     */
+    public Builder clearInputTensorIds() {
+      inputTensorIds_ = emptyLongList();
+      bitField0_ = (bitField0_ & ~0x00000001);
+      onChanged();
+      return this;
+    }
+
+    private com.google.protobuf.Internal.LongList outputTensorIds_ = emptyLongList();
+    private void ensureOutputTensorIdsIsMutable() {
+      if (!((bitField0_ & 0x00000002) != 0)) {
+        outputTensorIds_ = mutableCopy(outputTensorIds_);
+        bitField0_ |= 0x00000002;
+       }
+    }
+    /**
+     * <pre>
+     * IDs of the output tensors (if availbable).
+     * If specified, must have the same length as tensor_protos.
+     * </pre>
+     *
+     * <code>repeated int64 output_tensor_ids = 5;</code>
+     */
+    public java.util.List<java.lang.Long>
+        getOutputTensorIdsList() {
+      return ((bitField0_ & 0x00000002) != 0) ?
+               java.util.Collections.unmodifiableList(outputTensorIds_) : outputTensorIds_;
+    }
+    /**
+     * <pre>
+     * IDs of the output tensors (if availbable).
+     * If specified, must have the same length as tensor_protos.
+     * </pre>
+     *
+     * <code>repeated int64 output_tensor_ids = 5;</code>
+     */
+    public int getOutputTensorIdsCount() {
+      return outputTensorIds_.size();
+    }
+    /**
+     * <pre>
+     * IDs of the output tensors (if availbable).
+     * If specified, must have the same length as tensor_protos.
+     * </pre>
+     *
+     * <code>repeated int64 output_tensor_ids = 5;</code>
+     */
+    public long getOutputTensorIds(int index) {
+      return outputTensorIds_.getLong(index);
+    }
+    /**
+     * <pre>
+     * IDs of the output tensors (if availbable).
+     * If specified, must have the same length as tensor_protos.
+     * </pre>
+     *
+     * <code>repeated int64 output_tensor_ids = 5;</code>
+     */
+    public Builder setOutputTensorIds(
+        int index, long value) {
+      ensureOutputTensorIdsIsMutable();
+      outputTensorIds_.setLong(index, value);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * IDs of the output tensors (if availbable).
+     * If specified, must have the same length as tensor_protos.
+     * </pre>
+     *
+     * <code>repeated int64 output_tensor_ids = 5;</code>
+     */
+    public Builder addOutputTensorIds(long value) {
+      ensureOutputTensorIdsIsMutable();
+      outputTensorIds_.addLong(value);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * IDs of the output tensors (if availbable).
+     * If specified, must have the same length as tensor_protos.
+     * </pre>
+     *
+     * <code>repeated int64 output_tensor_ids = 5;</code>
+     */
+    public Builder addAllOutputTensorIds(
+        java.lang.Iterable<? extends java.lang.Long> values) {
+      ensureOutputTensorIdsIsMutable();
+      com.google.protobuf.AbstractMessageLite.Builder.addAll(
+          values, outputTensorIds_);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * IDs of the output tensors (if availbable).
+     * If specified, must have the same length as tensor_protos.
+     * </pre>
+     *
+     * <code>repeated int64 output_tensor_ids = 5;</code>
+     */
+    public Builder clearOutputTensorIds() {
+      outputTensorIds_ = emptyLongList();
+      bitField0_ = (bitField0_ & ~0x00000002);
+      onChanged();
+      return this;
+    }
+
+    private int tensorDebugMode_ = 0;
+    /**
+     * <pre>
+     * Type of the tensor value encapsulated in this proto.
+     * </pre>
+     *
+     * <code>.tensorflow.TensorDebugMode tensor_debug_mode = 6;</code>
+     */
+    public int getTensorDebugModeValue() {
+      return tensorDebugMode_;
+    }
+    /**
+     * <pre>
+     * Type of the tensor value encapsulated in this proto.
+     * </pre>
+     *
+     * <code>.tensorflow.TensorDebugMode tensor_debug_mode = 6;</code>
+     */
+    public Builder setTensorDebugModeValue(int value) {
+      tensorDebugMode_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Type of the tensor value encapsulated in this proto.
+     * </pre>
+     *
+     * <code>.tensorflow.TensorDebugMode tensor_debug_mode = 6;</code>
+     */
+    public org.tensorflow.proto.util.TensorDebugMode getTensorDebugMode() {
+      @SuppressWarnings("deprecation")
+      org.tensorflow.proto.util.TensorDebugMode result = org.tensorflow.proto.util.TensorDebugMode.valueOf(tensorDebugMode_);
+      return result == null ? org.tensorflow.proto.util.TensorDebugMode.UNRECOGNIZED : result;
+    }
+    /**
+     * <pre>
+     * Type of the tensor value encapsulated in this proto.
+     * </pre>
+     *
+     * <code>.tensorflow.TensorDebugMode tensor_debug_mode = 6;</code>
+     */
+    public Builder setTensorDebugMode(org.tensorflow.proto.util.TensorDebugMode value) {
+      if (value == null) {
+        throw new NullPointerException();
+      }
+      
+      tensorDebugMode_ = value.getNumber();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Type of the tensor value encapsulated in this proto.
+     * </pre>
+     *
+     * <code>.tensorflow.TensorDebugMode tensor_debug_mode = 6;</code>
+     */
+    public Builder clearTensorDebugMode() {
+      
+      tensorDebugMode_ = 0;
+      onChanged();
+      return this;
+    }
+
+    private java.util.List<org.tensorflow.proto.framework.TensorProto> tensorProtos_ =
+      java.util.Collections.emptyList();
+    private void ensureTensorProtosIsMutable() {
+      if (!((bitField0_ & 0x00000004) != 0)) {
+        tensorProtos_ = new java.util.ArrayList<org.tensorflow.proto.framework.TensorProto>(tensorProtos_);
+        bitField0_ |= 0x00000004;
+       }
+    }
+
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.framework.TensorProto, org.tensorflow.proto.framework.TensorProto.Builder, org.tensorflow.proto.framework.TensorProtoOrBuilder> tensorProtosBuilder_;
+
+    /**
+     * <pre>
+     * Output Tensor values in the type described by `tensor_value_type`.
+     * The length of this should match `num_outputs`.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorProto tensor_protos = 7;</code>
+     */
+    public java.util.List<org.tensorflow.proto.framework.TensorProto> getTensorProtosList() {
+      if (tensorProtosBuilder_ == null) {
+        return java.util.Collections.unmodifiableList(tensorProtos_);
+      } else {
+        return tensorProtosBuilder_.getMessageList();
+      }
+    }
+    /**
+     * <pre>
+     * Output Tensor values in the type described by `tensor_value_type`.
+     * The length of this should match `num_outputs`.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorProto tensor_protos = 7;</code>
+     */
+    public int getTensorProtosCount() {
+      if (tensorProtosBuilder_ == null) {
+        return tensorProtos_.size();
+      } else {
+        return tensorProtosBuilder_.getCount();
+      }
+    }
+    /**
+     * <pre>
+     * Output Tensor values in the type described by `tensor_value_type`.
+     * The length of this should match `num_outputs`.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorProto tensor_protos = 7;</code>
+     */
+    public org.tensorflow.proto.framework.TensorProto getTensorProtos(int index) {
+      if (tensorProtosBuilder_ == null) {
+        return tensorProtos_.get(index);
+      } else {
+        return tensorProtosBuilder_.getMessage(index);
+      }
+    }
+    /**
+     * <pre>
+     * Output Tensor values in the type described by `tensor_value_type`.
+     * The length of this should match `num_outputs`.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorProto tensor_protos = 7;</code>
+     */
+    public Builder setTensorProtos(
+        int index, org.tensorflow.proto.framework.TensorProto value) {
+      if (tensorProtosBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureTensorProtosIsMutable();
+        tensorProtos_.set(index, value);
+        onChanged();
+      } else {
+        tensorProtosBuilder_.setMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Output Tensor values in the type described by `tensor_value_type`.
+     * The length of this should match `num_outputs`.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorProto tensor_protos = 7;</code>
+     */
+    public Builder setTensorProtos(
+        int index, org.tensorflow.proto.framework.TensorProto.Builder builderForValue) {
+      if (tensorProtosBuilder_ == null) {
+        ensureTensorProtosIsMutable();
+        tensorProtos_.set(index, builderForValue.build());
+        onChanged();
+      } else {
+        tensorProtosBuilder_.setMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Output Tensor values in the type described by `tensor_value_type`.
+     * The length of this should match `num_outputs`.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorProto tensor_protos = 7;</code>
+     */
+    public Builder addTensorProtos(org.tensorflow.proto.framework.TensorProto value) {
+      if (tensorProtosBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureTensorProtosIsMutable();
+        tensorProtos_.add(value);
+        onChanged();
+      } else {
+        tensorProtosBuilder_.addMessage(value);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Output Tensor values in the type described by `tensor_value_type`.
+     * The length of this should match `num_outputs`.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorProto tensor_protos = 7;</code>
+     */
+    public Builder addTensorProtos(
+        int index, org.tensorflow.proto.framework.TensorProto value) {
+      if (tensorProtosBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureTensorProtosIsMutable();
+        tensorProtos_.add(index, value);
+        onChanged();
+      } else {
+        tensorProtosBuilder_.addMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Output Tensor values in the type described by `tensor_value_type`.
+     * The length of this should match `num_outputs`.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorProto tensor_protos = 7;</code>
+     */
+    public Builder addTensorProtos(
+        org.tensorflow.proto.framework.TensorProto.Builder builderForValue) {
+      if (tensorProtosBuilder_ == null) {
+        ensureTensorProtosIsMutable();
+        tensorProtos_.add(builderForValue.build());
+        onChanged();
+      } else {
+        tensorProtosBuilder_.addMessage(builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Output Tensor values in the type described by `tensor_value_type`.
+     * The length of this should match `num_outputs`.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorProto tensor_protos = 7;</code>
+     */
+    public Builder addTensorProtos(
+        int index, org.tensorflow.proto.framework.TensorProto.Builder builderForValue) {
+      if (tensorProtosBuilder_ == null) {
+        ensureTensorProtosIsMutable();
+        tensorProtos_.add(index, builderForValue.build());
+        onChanged();
+      } else {
+        tensorProtosBuilder_.addMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Output Tensor values in the type described by `tensor_value_type`.
+     * The length of this should match `num_outputs`.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorProto tensor_protos = 7;</code>
+     */
+    public Builder addAllTensorProtos(
+        java.lang.Iterable<? extends org.tensorflow.proto.framework.TensorProto> values) {
+      if (tensorProtosBuilder_ == null) {
+        ensureTensorProtosIsMutable();
+        com.google.protobuf.AbstractMessageLite.Builder.addAll(
+            values, tensorProtos_);
+        onChanged();
+      } else {
+        tensorProtosBuilder_.addAllMessages(values);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Output Tensor values in the type described by `tensor_value_type`.
+     * The length of this should match `num_outputs`.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorProto tensor_protos = 7;</code>
+     */
+    public Builder clearTensorProtos() {
+      if (tensorProtosBuilder_ == null) {
+        tensorProtos_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000004);
+        onChanged();
+      } else {
+        tensorProtosBuilder_.clear();
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Output Tensor values in the type described by `tensor_value_type`.
+     * The length of this should match `num_outputs`.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorProto tensor_protos = 7;</code>
+     */
+    public Builder removeTensorProtos(int index) {
+      if (tensorProtosBuilder_ == null) {
+        ensureTensorProtosIsMutable();
+        tensorProtos_.remove(index);
+        onChanged();
+      } else {
+        tensorProtosBuilder_.remove(index);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Output Tensor values in the type described by `tensor_value_type`.
+     * The length of this should match `num_outputs`.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorProto tensor_protos = 7;</code>
+     */
+    public org.tensorflow.proto.framework.TensorProto.Builder getTensorProtosBuilder(
+        int index) {
+      return getTensorProtosFieldBuilder().getBuilder(index);
+    }
+    /**
+     * <pre>
+     * Output Tensor values in the type described by `tensor_value_type`.
+     * The length of this should match `num_outputs`.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorProto tensor_protos = 7;</code>
+     */
+    public org.tensorflow.proto.framework.TensorProtoOrBuilder getTensorProtosOrBuilder(
+        int index) {
+      if (tensorProtosBuilder_ == null) {
+        return tensorProtos_.get(index);  } else {
+        return tensorProtosBuilder_.getMessageOrBuilder(index);
+      }
+    }
+    /**
+     * <pre>
+     * Output Tensor values in the type described by `tensor_value_type`.
+     * The length of this should match `num_outputs`.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorProto tensor_protos = 7;</code>
+     */
+    public java.util.List<? extends org.tensorflow.proto.framework.TensorProtoOrBuilder> 
+         getTensorProtosOrBuilderList() {
+      if (tensorProtosBuilder_ != null) {
+        return tensorProtosBuilder_.getMessageOrBuilderList();
+      } else {
+        return java.util.Collections.unmodifiableList(tensorProtos_);
+      }
+    }
+    /**
+     * <pre>
+     * Output Tensor values in the type described by `tensor_value_type`.
+     * The length of this should match `num_outputs`.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorProto tensor_protos = 7;</code>
+     */
+    public org.tensorflow.proto.framework.TensorProto.Builder addTensorProtosBuilder() {
+      return getTensorProtosFieldBuilder().addBuilder(
+          org.tensorflow.proto.framework.TensorProto.getDefaultInstance());
+    }
+    /**
+     * <pre>
+     * Output Tensor values in the type described by `tensor_value_type`.
+     * The length of this should match `num_outputs`.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorProto tensor_protos = 7;</code>
+     */
+    public org.tensorflow.proto.framework.TensorProto.Builder addTensorProtosBuilder(
+        int index) {
+      return getTensorProtosFieldBuilder().addBuilder(
+          index, org.tensorflow.proto.framework.TensorProto.getDefaultInstance());
+    }
+    /**
+     * <pre>
+     * Output Tensor values in the type described by `tensor_value_type`.
+     * The length of this should match `num_outputs`.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorProto tensor_protos = 7;</code>
+     */
+    public java.util.List<org.tensorflow.proto.framework.TensorProto.Builder> 
+         getTensorProtosBuilderList() {
+      return getTensorProtosFieldBuilder().getBuilderList();
+    }
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.framework.TensorProto, org.tensorflow.proto.framework.TensorProto.Builder, org.tensorflow.proto.framework.TensorProtoOrBuilder> 
+        getTensorProtosFieldBuilder() {
+      if (tensorProtosBuilder_ == null) {
+        tensorProtosBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
+            org.tensorflow.proto.framework.TensorProto, org.tensorflow.proto.framework.TensorProto.Builder, org.tensorflow.proto.framework.TensorProtoOrBuilder>(
+                tensorProtos_,
+                ((bitField0_ & 0x00000004) != 0),
+                getParentForChildren(),
+                isClean());
+        tensorProtos_ = null;
+      }
+      return tensorProtosBuilder_;
+    }
+
+    private org.tensorflow.proto.util.CodeLocation codeLocation_;
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.util.CodeLocation, org.tensorflow.proto.util.CodeLocation.Builder, org.tensorflow.proto.util.CodeLocationOrBuilder> codeLocationBuilder_;
+    /**
+     * <pre>
+     * Stack trace of the eager execution.
+     * </pre>
+     *
+     * <code>.tensorflow.CodeLocation code_location = 8;</code>
+     */
+    public boolean hasCodeLocation() {
+      return codeLocationBuilder_ != null || codeLocation_ != null;
+    }
+    /**
+     * <pre>
+     * Stack trace of the eager execution.
+     * </pre>
+     *
+     * <code>.tensorflow.CodeLocation code_location = 8;</code>
+     */
+    public org.tensorflow.proto.util.CodeLocation getCodeLocation() {
+      if (codeLocationBuilder_ == null) {
+        return codeLocation_ == null ? org.tensorflow.proto.util.CodeLocation.getDefaultInstance() : codeLocation_;
+      } else {
+        return codeLocationBuilder_.getMessage();
+      }
+    }
+    /**
+     * <pre>
+     * Stack trace of the eager execution.
+     * </pre>
+     *
+     * <code>.tensorflow.CodeLocation code_location = 8;</code>
+     */
+    public Builder setCodeLocation(org.tensorflow.proto.util.CodeLocation value) {
+      if (codeLocationBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        codeLocation_ = value;
+        onChanged();
+      } else {
+        codeLocationBuilder_.setMessage(value);
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Stack trace of the eager execution.
+     * </pre>
+     *
+     * <code>.tensorflow.CodeLocation code_location = 8;</code>
+     */
+    public Builder setCodeLocation(
+        org.tensorflow.proto.util.CodeLocation.Builder builderForValue) {
+      if (codeLocationBuilder_ == null) {
+        codeLocation_ = builderForValue.build();
+        onChanged();
+      } else {
+        codeLocationBuilder_.setMessage(builderForValue.build());
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Stack trace of the eager execution.
+     * </pre>
+     *
+     * <code>.tensorflow.CodeLocation code_location = 8;</code>
+     */
+    public Builder mergeCodeLocation(org.tensorflow.proto.util.CodeLocation value) {
+      if (codeLocationBuilder_ == null) {
+        if (codeLocation_ != null) {
+          codeLocation_ =
+            org.tensorflow.proto.util.CodeLocation.newBuilder(codeLocation_).mergeFrom(value).buildPartial();
+        } else {
+          codeLocation_ = value;
+        }
+        onChanged();
+      } else {
+        codeLocationBuilder_.mergeFrom(value);
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Stack trace of the eager execution.
+     * </pre>
+     *
+     * <code>.tensorflow.CodeLocation code_location = 8;</code>
+     */
+    public Builder clearCodeLocation() {
+      if (codeLocationBuilder_ == null) {
+        codeLocation_ = null;
+        onChanged();
+      } else {
+        codeLocation_ = null;
+        codeLocationBuilder_ = null;
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Stack trace of the eager execution.
+     * </pre>
+     *
+     * <code>.tensorflow.CodeLocation code_location = 8;</code>
+     */
+    public org.tensorflow.proto.util.CodeLocation.Builder getCodeLocationBuilder() {
+      
+      onChanged();
+      return getCodeLocationFieldBuilder().getBuilder();
+    }
+    /**
+     * <pre>
+     * Stack trace of the eager execution.
+     * </pre>
+     *
+     * <code>.tensorflow.CodeLocation code_location = 8;</code>
+     */
+    public org.tensorflow.proto.util.CodeLocationOrBuilder getCodeLocationOrBuilder() {
+      if (codeLocationBuilder_ != null) {
+        return codeLocationBuilder_.getMessageOrBuilder();
+      } else {
+        return codeLocation_ == null ?
+            org.tensorflow.proto.util.CodeLocation.getDefaultInstance() : codeLocation_;
+      }
+    }
+    /**
+     * <pre>
+     * Stack trace of the eager execution.
+     * </pre>
+     *
+     * <code>.tensorflow.CodeLocation code_location = 8;</code>
+     */
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.util.CodeLocation, org.tensorflow.proto.util.CodeLocation.Builder, org.tensorflow.proto.util.CodeLocationOrBuilder> 
+        getCodeLocationFieldBuilder() {
+      if (codeLocationBuilder_ == null) {
+        codeLocationBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+            org.tensorflow.proto.util.CodeLocation, org.tensorflow.proto.util.CodeLocation.Builder, org.tensorflow.proto.util.CodeLocationOrBuilder>(
+                getCodeLocation(),
+                getParentForChildren(),
+                isClean());
+        codeLocation_ = null;
+      }
+      return codeLocationBuilder_;
+    }
+
+    private com.google.protobuf.Internal.IntList outputTensorDeviceIds_ = emptyIntList();
+    private void ensureOutputTensorDeviceIdsIsMutable() {
+      if (!((bitField0_ & 0x00000008) != 0)) {
+        outputTensorDeviceIds_ = mutableCopy(outputTensorDeviceIds_);
+        bitField0_ |= 0x00000008;
+       }
+    }
+    /**
+     * <pre>
+     * Debugged-generated IDs of the devices on which the output tensors reside.
+     * To look up details about the device (e.g., name), cross-reference this
+     * field with the DebuggedDevice messages.
+     * </pre>
+     *
+     * <code>repeated int32 output_tensor_device_ids = 9;</code>
+     */
+    public java.util.List<java.lang.Integer>
+        getOutputTensorDeviceIdsList() {
+      return ((bitField0_ & 0x00000008) != 0) ?
+               java.util.Collections.unmodifiableList(outputTensorDeviceIds_) : outputTensorDeviceIds_;
+    }
+    /**
+     * <pre>
+     * Debugged-generated IDs of the devices on which the output tensors reside.
+     * To look up details about the device (e.g., name), cross-reference this
+     * field with the DebuggedDevice messages.
+     * </pre>
+     *
+     * <code>repeated int32 output_tensor_device_ids = 9;</code>
+     */
+    public int getOutputTensorDeviceIdsCount() {
+      return outputTensorDeviceIds_.size();
+    }
+    /**
+     * <pre>
+     * Debugged-generated IDs of the devices on which the output tensors reside.
+     * To look up details about the device (e.g., name), cross-reference this
+     * field with the DebuggedDevice messages.
+     * </pre>
+     *
+     * <code>repeated int32 output_tensor_device_ids = 9;</code>
+     */
+    public int getOutputTensorDeviceIds(int index) {
+      return outputTensorDeviceIds_.getInt(index);
+    }
+    /**
+     * <pre>
+     * Debugged-generated IDs of the devices on which the output tensors reside.
+     * To look up details about the device (e.g., name), cross-reference this
+     * field with the DebuggedDevice messages.
+     * </pre>
+     *
+     * <code>repeated int32 output_tensor_device_ids = 9;</code>
+     */
+    public Builder setOutputTensorDeviceIds(
+        int index, int value) {
+      ensureOutputTensorDeviceIdsIsMutable();
+      outputTensorDeviceIds_.setInt(index, value);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Debugged-generated IDs of the devices on which the output tensors reside.
+     * To look up details about the device (e.g., name), cross-reference this
+     * field with the DebuggedDevice messages.
+     * </pre>
+     *
+     * <code>repeated int32 output_tensor_device_ids = 9;</code>
+     */
+    public Builder addOutputTensorDeviceIds(int value) {
+      ensureOutputTensorDeviceIdsIsMutable();
+      outputTensorDeviceIds_.addInt(value);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Debugged-generated IDs of the devices on which the output tensors reside.
+     * To look up details about the device (e.g., name), cross-reference this
+     * field with the DebuggedDevice messages.
+     * </pre>
+     *
+     * <code>repeated int32 output_tensor_device_ids = 9;</code>
+     */
+    public Builder addAllOutputTensorDeviceIds(
+        java.lang.Iterable<? extends java.lang.Integer> values) {
+      ensureOutputTensorDeviceIdsIsMutable();
+      com.google.protobuf.AbstractMessageLite.Builder.addAll(
+          values, outputTensorDeviceIds_);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Debugged-generated IDs of the devices on which the output tensors reside.
+     * To look up details about the device (e.g., name), cross-reference this
+     * field with the DebuggedDevice messages.
+     * </pre>
+     *
+     * <code>repeated int32 output_tensor_device_ids = 9;</code>
+     */
+    public Builder clearOutputTensorDeviceIds() {
+      outputTensorDeviceIds_ = emptyIntList();
+      bitField0_ = (bitField0_ & ~0x00000008);
+      onChanged();
+      return this;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.Execution)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.Execution)
+  private static final org.tensorflow.proto.util.Execution DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.util.Execution();
+  }
+
+  public static org.tensorflow.proto.util.Execution getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<Execution>
+      PARSER = new com.google.protobuf.AbstractParser<Execution>() {
+    @java.lang.Override
+    public Execution parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new Execution(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<Execution> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<Execution> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.util.Execution getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/ExecutionOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/ExecutionOrBuilder.java
new file mode 100644
index 00000000000..c08db8762e5
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/ExecutionOrBuilder.java
@@ -0,0 +1,233 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/debug_event.proto
+
+package org.tensorflow.proto.util;
+
+public interface ExecutionOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.Execution)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <pre>
+   * Op type (e.g., "MatMul").
+   * In the case of a Graph, this is the name of the Graph.
+   * </pre>
+   *
+   * <code>string op_type = 1;</code>
+   */
+  java.lang.String getOpType();
+  /**
+   * <pre>
+   * Op type (e.g., "MatMul").
+   * In the case of a Graph, this is the name of the Graph.
+   * </pre>
+   *
+   * <code>string op_type = 1;</code>
+   */
+  com.google.protobuf.ByteString
+      getOpTypeBytes();
+
+  /**
+   * <pre>
+   * Number of output tensors.
+   * </pre>
+   *
+   * <code>int32 num_outputs = 2;</code>
+   */
+  int getNumOutputs();
+
+  /**
+   * <pre>
+   * The graph that's executed: applicable only to the eager
+   * execution of a FuncGraph.
+   * </pre>
+   *
+   * <code>string graph_id = 3;</code>
+   */
+  java.lang.String getGraphId();
+  /**
+   * <pre>
+   * The graph that's executed: applicable only to the eager
+   * execution of a FuncGraph.
+   * </pre>
+   *
+   * <code>string graph_id = 3;</code>
+   */
+  com.google.protobuf.ByteString
+      getGraphIdBytes();
+
+  /**
+   * <pre>
+   * IDs of the input tensors (if available).
+   * </pre>
+   *
+   * <code>repeated int64 input_tensor_ids = 4;</code>
+   */
+  java.util.List<java.lang.Long> getInputTensorIdsList();
+  /**
+   * <pre>
+   * IDs of the input tensors (if available).
+   * </pre>
+   *
+   * <code>repeated int64 input_tensor_ids = 4;</code>
+   */
+  int getInputTensorIdsCount();
+  /**
+   * <pre>
+   * IDs of the input tensors (if available).
+   * </pre>
+   *
+   * <code>repeated int64 input_tensor_ids = 4;</code>
+   */
+  long getInputTensorIds(int index);
+
+  /**
+   * <pre>
+   * IDs of the output tensors (if availbable).
+   * If specified, must have the same length as tensor_protos.
+   * </pre>
+   *
+   * <code>repeated int64 output_tensor_ids = 5;</code>
+   */
+  java.util.List<java.lang.Long> getOutputTensorIdsList();
+  /**
+   * <pre>
+   * IDs of the output tensors (if availbable).
+   * If specified, must have the same length as tensor_protos.
+   * </pre>
+   *
+   * <code>repeated int64 output_tensor_ids = 5;</code>
+   */
+  int getOutputTensorIdsCount();
+  /**
+   * <pre>
+   * IDs of the output tensors (if availbable).
+   * If specified, must have the same length as tensor_protos.
+   * </pre>
+   *
+   * <code>repeated int64 output_tensor_ids = 5;</code>
+   */
+  long getOutputTensorIds(int index);
+
+  /**
+   * <pre>
+   * Type of the tensor value encapsulated in this proto.
+   * </pre>
+   *
+   * <code>.tensorflow.TensorDebugMode tensor_debug_mode = 6;</code>
+   */
+  int getTensorDebugModeValue();
+  /**
+   * <pre>
+   * Type of the tensor value encapsulated in this proto.
+   * </pre>
+   *
+   * <code>.tensorflow.TensorDebugMode tensor_debug_mode = 6;</code>
+   */
+  org.tensorflow.proto.util.TensorDebugMode getTensorDebugMode();
+
+  /**
+   * <pre>
+   * Output Tensor values in the type described by `tensor_value_type`.
+   * The length of this should match `num_outputs`.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.TensorProto tensor_protos = 7;</code>
+   */
+  java.util.List<org.tensorflow.proto.framework.TensorProto> 
+      getTensorProtosList();
+  /**
+   * <pre>
+   * Output Tensor values in the type described by `tensor_value_type`.
+   * The length of this should match `num_outputs`.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.TensorProto tensor_protos = 7;</code>
+   */
+  org.tensorflow.proto.framework.TensorProto getTensorProtos(int index);
+  /**
+   * <pre>
+   * Output Tensor values in the type described by `tensor_value_type`.
+   * The length of this should match `num_outputs`.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.TensorProto tensor_protos = 7;</code>
+   */
+  int getTensorProtosCount();
+  /**
+   * <pre>
+   * Output Tensor values in the type described by `tensor_value_type`.
+   * The length of this should match `num_outputs`.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.TensorProto tensor_protos = 7;</code>
+   */
+  java.util.List<? extends org.tensorflow.proto.framework.TensorProtoOrBuilder> 
+      getTensorProtosOrBuilderList();
+  /**
+   * <pre>
+   * Output Tensor values in the type described by `tensor_value_type`.
+   * The length of this should match `num_outputs`.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.TensorProto tensor_protos = 7;</code>
+   */
+  org.tensorflow.proto.framework.TensorProtoOrBuilder getTensorProtosOrBuilder(
+      int index);
+
+  /**
+   * <pre>
+   * Stack trace of the eager execution.
+   * </pre>
+   *
+   * <code>.tensorflow.CodeLocation code_location = 8;</code>
+   */
+  boolean hasCodeLocation();
+  /**
+   * <pre>
+   * Stack trace of the eager execution.
+   * </pre>
+   *
+   * <code>.tensorflow.CodeLocation code_location = 8;</code>
+   */
+  org.tensorflow.proto.util.CodeLocation getCodeLocation();
+  /**
+   * <pre>
+   * Stack trace of the eager execution.
+   * </pre>
+   *
+   * <code>.tensorflow.CodeLocation code_location = 8;</code>
+   */
+  org.tensorflow.proto.util.CodeLocationOrBuilder getCodeLocationOrBuilder();
+
+  /**
+   * <pre>
+   * Debugged-generated IDs of the devices on which the output tensors reside.
+   * To look up details about the device (e.g., name), cross-reference this
+   * field with the DebuggedDevice messages.
+   * </pre>
+   *
+   * <code>repeated int32 output_tensor_device_ids = 9;</code>
+   */
+  java.util.List<java.lang.Integer> getOutputTensorDeviceIdsList();
+  /**
+   * <pre>
+   * Debugged-generated IDs of the devices on which the output tensors reside.
+   * To look up details about the device (e.g., name), cross-reference this
+   * field with the DebuggedDevice messages.
+   * </pre>
+   *
+   * <code>repeated int32 output_tensor_device_ids = 9;</code>
+   */
+  int getOutputTensorDeviceIdsCount();
+  /**
+   * <pre>
+   * Debugged-generated IDs of the devices on which the output tensors reside.
+   * To look up details about the device (e.g., name), cross-reference this
+   * field with the DebuggedDevice messages.
+   * </pre>
+   *
+   * <code>repeated int32 output_tensor_device_ids = 9;</code>
+   */
+  int getOutputTensorDeviceIds(int index);
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/GraphExecutionTrace.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/GraphExecutionTrace.java
new file mode 100644
index 00000000000..3ea9e244ff2
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/GraphExecutionTrace.java
@@ -0,0 +1,1359 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/debug_event.proto
+
+package org.tensorflow.proto.util;
+
+/**
+ * <pre>
+ * Data relating to an execution of a Graph (e.g., an eager execution of a
+ * FuncGraph).
+ * The values of the intermediate tensors computed in the graph are recorded
+ * in this proto. A graph execution may correspond to one or more pieces of
+ * `GraphExecutionTrace`, depending on whether the instrumented tensor values
+ * are summarized in an aggregated or separate fashion.
+ * </pre>
+ *
+ * Protobuf type {@code tensorflow.GraphExecutionTrace}
+ */
+public  final class GraphExecutionTrace extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.GraphExecutionTrace)
+    GraphExecutionTraceOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use GraphExecutionTrace.newBuilder() to construct.
+  private GraphExecutionTrace(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private GraphExecutionTrace() {
+    tfdbgContextId_ = "";
+    opName_ = "";
+    tensorDebugMode_ = 0;
+    deviceName_ = "";
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new GraphExecutionTrace();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private GraphExecutionTrace(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 10: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            tfdbgContextId_ = s;
+            break;
+          }
+          case 18: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            opName_ = s;
+            break;
+          }
+          case 24: {
+
+            outputSlot_ = input.readInt32();
+            break;
+          }
+          case 32: {
+            int rawValue = input.readEnum();
+
+            tensorDebugMode_ = rawValue;
+            break;
+          }
+          case 42: {
+            org.tensorflow.proto.framework.TensorProto.Builder subBuilder = null;
+            if (tensorProto_ != null) {
+              subBuilder = tensorProto_.toBuilder();
+            }
+            tensorProto_ = input.readMessage(org.tensorflow.proto.framework.TensorProto.parser(), extensionRegistry);
+            if (subBuilder != null) {
+              subBuilder.mergeFrom(tensorProto_);
+              tensorProto_ = subBuilder.buildPartial();
+            }
+
+            break;
+          }
+          case 50: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            deviceName_ = s;
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.util.DebugEventProtos.internal_static_tensorflow_GraphExecutionTrace_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.util.DebugEventProtos.internal_static_tensorflow_GraphExecutionTrace_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.util.GraphExecutionTrace.class, org.tensorflow.proto.util.GraphExecutionTrace.Builder.class);
+  }
+
+  public static final int TFDBG_CONTEXT_ID_FIELD_NUMBER = 1;
+  private volatile java.lang.Object tfdbgContextId_;
+  /**
+   * <pre>
+   * Unique ID of the context that the executed op(s) belong to (e.g., a
+   * compiled concrete tf.function).
+   * </pre>
+   *
+   * <code>string tfdbg_context_id = 1;</code>
+   */
+  public java.lang.String getTfdbgContextId() {
+    java.lang.Object ref = tfdbgContextId_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      tfdbgContextId_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * Unique ID of the context that the executed op(s) belong to (e.g., a
+   * compiled concrete tf.function).
+   * </pre>
+   *
+   * <code>string tfdbg_context_id = 1;</code>
+   */
+  public com.google.protobuf.ByteString
+      getTfdbgContextIdBytes() {
+    java.lang.Object ref = tfdbgContextId_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      tfdbgContextId_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int OP_NAME_FIELD_NUMBER = 2;
+  private volatile java.lang.Object opName_;
+  /**
+   * <pre>
+   * Name of the op (applicable only in the case of the `FULL_TENSOR` trace
+   * level).
+   * </pre>
+   *
+   * <code>string op_name = 2;</code>
+   */
+  public java.lang.String getOpName() {
+    java.lang.Object ref = opName_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      opName_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * Name of the op (applicable only in the case of the `FULL_TENSOR` trace
+   * level).
+   * </pre>
+   *
+   * <code>string op_name = 2;</code>
+   */
+  public com.google.protobuf.ByteString
+      getOpNameBytes() {
+    java.lang.Object ref = opName_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      opName_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int OUTPUT_SLOT_FIELD_NUMBER = 3;
+  private int outputSlot_;
+  /**
+   * <pre>
+   * Output slot of the tensor (applicable only in the case of the `FULL_TENSOR`
+   * trace level).
+   * </pre>
+   *
+   * <code>int32 output_slot = 3;</code>
+   */
+  public int getOutputSlot() {
+    return outputSlot_;
+  }
+
+  public static final int TENSOR_DEBUG_MODE_FIELD_NUMBER = 4;
+  private int tensorDebugMode_;
+  /**
+   * <pre>
+   * Type of the tensor value encapsulated in this proto.
+   * </pre>
+   *
+   * <code>.tensorflow.TensorDebugMode tensor_debug_mode = 4;</code>
+   */
+  public int getTensorDebugModeValue() {
+    return tensorDebugMode_;
+  }
+  /**
+   * <pre>
+   * Type of the tensor value encapsulated in this proto.
+   * </pre>
+   *
+   * <code>.tensorflow.TensorDebugMode tensor_debug_mode = 4;</code>
+   */
+  public org.tensorflow.proto.util.TensorDebugMode getTensorDebugMode() {
+    @SuppressWarnings("deprecation")
+    org.tensorflow.proto.util.TensorDebugMode result = org.tensorflow.proto.util.TensorDebugMode.valueOf(tensorDebugMode_);
+    return result == null ? org.tensorflow.proto.util.TensorDebugMode.UNRECOGNIZED : result;
+  }
+
+  public static final int TENSOR_PROTO_FIELD_NUMBER = 5;
+  private org.tensorflow.proto.framework.TensorProto tensorProto_;
+  /**
+   * <pre>
+   * Tensor value in the type described by `tensor_value_type`.
+   * This tensor may summarize the value of a single intermediate op of the
+   * graph, or those of multiple intermediate tensors.
+   * </pre>
+   *
+   * <code>.tensorflow.TensorProto tensor_proto = 5;</code>
+   */
+  public boolean hasTensorProto() {
+    return tensorProto_ != null;
+  }
+  /**
+   * <pre>
+   * Tensor value in the type described by `tensor_value_type`.
+   * This tensor may summarize the value of a single intermediate op of the
+   * graph, or those of multiple intermediate tensors.
+   * </pre>
+   *
+   * <code>.tensorflow.TensorProto tensor_proto = 5;</code>
+   */
+  public org.tensorflow.proto.framework.TensorProto getTensorProto() {
+    return tensorProto_ == null ? org.tensorflow.proto.framework.TensorProto.getDefaultInstance() : tensorProto_;
+  }
+  /**
+   * <pre>
+   * Tensor value in the type described by `tensor_value_type`.
+   * This tensor may summarize the value of a single intermediate op of the
+   * graph, or those of multiple intermediate tensors.
+   * </pre>
+   *
+   * <code>.tensorflow.TensorProto tensor_proto = 5;</code>
+   */
+  public org.tensorflow.proto.framework.TensorProtoOrBuilder getTensorProtoOrBuilder() {
+    return getTensorProto();
+  }
+
+  public static final int DEVICE_NAME_FIELD_NUMBER = 6;
+  private volatile java.lang.Object deviceName_;
+  /**
+   * <pre>
+   * Name of the device that the op belongs to.
+   * </pre>
+   *
+   * <code>string device_name = 6;</code>
+   */
+  public java.lang.String getDeviceName() {
+    java.lang.Object ref = deviceName_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      deviceName_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * Name of the device that the op belongs to.
+   * </pre>
+   *
+   * <code>string device_name = 6;</code>
+   */
+  public com.google.protobuf.ByteString
+      getDeviceNameBytes() {
+    java.lang.Object ref = deviceName_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      deviceName_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    if (!getTfdbgContextIdBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 1, tfdbgContextId_);
+    }
+    if (!getOpNameBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 2, opName_);
+    }
+    if (outputSlot_ != 0) {
+      output.writeInt32(3, outputSlot_);
+    }
+    if (tensorDebugMode_ != org.tensorflow.proto.util.TensorDebugMode.UNSPECIFIED.getNumber()) {
+      output.writeEnum(4, tensorDebugMode_);
+    }
+    if (tensorProto_ != null) {
+      output.writeMessage(5, getTensorProto());
+    }
+    if (!getDeviceNameBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 6, deviceName_);
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (!getTfdbgContextIdBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, tfdbgContextId_);
+    }
+    if (!getOpNameBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(2, opName_);
+    }
+    if (outputSlot_ != 0) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt32Size(3, outputSlot_);
+    }
+    if (tensorDebugMode_ != org.tensorflow.proto.util.TensorDebugMode.UNSPECIFIED.getNumber()) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeEnumSize(4, tensorDebugMode_);
+    }
+    if (tensorProto_ != null) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(5, getTensorProto());
+    }
+    if (!getDeviceNameBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(6, deviceName_);
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.util.GraphExecutionTrace)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.util.GraphExecutionTrace other = (org.tensorflow.proto.util.GraphExecutionTrace) obj;
+
+    if (!getTfdbgContextId()
+        .equals(other.getTfdbgContextId())) return false;
+    if (!getOpName()
+        .equals(other.getOpName())) return false;
+    if (getOutputSlot()
+        != other.getOutputSlot()) return false;
+    if (tensorDebugMode_ != other.tensorDebugMode_) return false;
+    if (hasTensorProto() != other.hasTensorProto()) return false;
+    if (hasTensorProto()) {
+      if (!getTensorProto()
+          .equals(other.getTensorProto())) return false;
+    }
+    if (!getDeviceName()
+        .equals(other.getDeviceName())) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    hash = (37 * hash) + TFDBG_CONTEXT_ID_FIELD_NUMBER;
+    hash = (53 * hash) + getTfdbgContextId().hashCode();
+    hash = (37 * hash) + OP_NAME_FIELD_NUMBER;
+    hash = (53 * hash) + getOpName().hashCode();
+    hash = (37 * hash) + OUTPUT_SLOT_FIELD_NUMBER;
+    hash = (53 * hash) + getOutputSlot();
+    hash = (37 * hash) + TENSOR_DEBUG_MODE_FIELD_NUMBER;
+    hash = (53 * hash) + tensorDebugMode_;
+    if (hasTensorProto()) {
+      hash = (37 * hash) + TENSOR_PROTO_FIELD_NUMBER;
+      hash = (53 * hash) + getTensorProto().hashCode();
+    }
+    hash = (37 * hash) + DEVICE_NAME_FIELD_NUMBER;
+    hash = (53 * hash) + getDeviceName().hashCode();
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.util.GraphExecutionTrace parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.GraphExecutionTrace parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.GraphExecutionTrace parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.GraphExecutionTrace parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.GraphExecutionTrace parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.GraphExecutionTrace parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.GraphExecutionTrace parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.GraphExecutionTrace parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.GraphExecutionTrace parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.GraphExecutionTrace parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.GraphExecutionTrace parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.GraphExecutionTrace parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.util.GraphExecutionTrace prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * <pre>
+   * Data relating to an execution of a Graph (e.g., an eager execution of a
+   * FuncGraph).
+   * The values of the intermediate tensors computed in the graph are recorded
+   * in this proto. A graph execution may correspond to one or more pieces of
+   * `GraphExecutionTrace`, depending on whether the instrumented tensor values
+   * are summarized in an aggregated or separate fashion.
+   * </pre>
+   *
+   * Protobuf type {@code tensorflow.GraphExecutionTrace}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.GraphExecutionTrace)
+      org.tensorflow.proto.util.GraphExecutionTraceOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.util.DebugEventProtos.internal_static_tensorflow_GraphExecutionTrace_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.util.DebugEventProtos.internal_static_tensorflow_GraphExecutionTrace_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.util.GraphExecutionTrace.class, org.tensorflow.proto.util.GraphExecutionTrace.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.util.GraphExecutionTrace.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      tfdbgContextId_ = "";
+
+      opName_ = "";
+
+      outputSlot_ = 0;
+
+      tensorDebugMode_ = 0;
+
+      if (tensorProtoBuilder_ == null) {
+        tensorProto_ = null;
+      } else {
+        tensorProto_ = null;
+        tensorProtoBuilder_ = null;
+      }
+      deviceName_ = "";
+
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.util.DebugEventProtos.internal_static_tensorflow_GraphExecutionTrace_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.GraphExecutionTrace getDefaultInstanceForType() {
+      return org.tensorflow.proto.util.GraphExecutionTrace.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.GraphExecutionTrace build() {
+      org.tensorflow.proto.util.GraphExecutionTrace result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.GraphExecutionTrace buildPartial() {
+      org.tensorflow.proto.util.GraphExecutionTrace result = new org.tensorflow.proto.util.GraphExecutionTrace(this);
+      result.tfdbgContextId_ = tfdbgContextId_;
+      result.opName_ = opName_;
+      result.outputSlot_ = outputSlot_;
+      result.tensorDebugMode_ = tensorDebugMode_;
+      if (tensorProtoBuilder_ == null) {
+        result.tensorProto_ = tensorProto_;
+      } else {
+        result.tensorProto_ = tensorProtoBuilder_.build();
+      }
+      result.deviceName_ = deviceName_;
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.util.GraphExecutionTrace) {
+        return mergeFrom((org.tensorflow.proto.util.GraphExecutionTrace)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.util.GraphExecutionTrace other) {
+      if (other == org.tensorflow.proto.util.GraphExecutionTrace.getDefaultInstance()) return this;
+      if (!other.getTfdbgContextId().isEmpty()) {
+        tfdbgContextId_ = other.tfdbgContextId_;
+        onChanged();
+      }
+      if (!other.getOpName().isEmpty()) {
+        opName_ = other.opName_;
+        onChanged();
+      }
+      if (other.getOutputSlot() != 0) {
+        setOutputSlot(other.getOutputSlot());
+      }
+      if (other.tensorDebugMode_ != 0) {
+        setTensorDebugModeValue(other.getTensorDebugModeValue());
+      }
+      if (other.hasTensorProto()) {
+        mergeTensorProto(other.getTensorProto());
+      }
+      if (!other.getDeviceName().isEmpty()) {
+        deviceName_ = other.deviceName_;
+        onChanged();
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.util.GraphExecutionTrace parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.util.GraphExecutionTrace) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+
+    private java.lang.Object tfdbgContextId_ = "";
+    /**
+     * <pre>
+     * Unique ID of the context that the executed op(s) belong to (e.g., a
+     * compiled concrete tf.function).
+     * </pre>
+     *
+     * <code>string tfdbg_context_id = 1;</code>
+     */
+    public java.lang.String getTfdbgContextId() {
+      java.lang.Object ref = tfdbgContextId_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        tfdbgContextId_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Unique ID of the context that the executed op(s) belong to (e.g., a
+     * compiled concrete tf.function).
+     * </pre>
+     *
+     * <code>string tfdbg_context_id = 1;</code>
+     */
+    public com.google.protobuf.ByteString
+        getTfdbgContextIdBytes() {
+      java.lang.Object ref = tfdbgContextId_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        tfdbgContextId_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Unique ID of the context that the executed op(s) belong to (e.g., a
+     * compiled concrete tf.function).
+     * </pre>
+     *
+     * <code>string tfdbg_context_id = 1;</code>
+     */
+    public Builder setTfdbgContextId(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      tfdbgContextId_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Unique ID of the context that the executed op(s) belong to (e.g., a
+     * compiled concrete tf.function).
+     * </pre>
+     *
+     * <code>string tfdbg_context_id = 1;</code>
+     */
+    public Builder clearTfdbgContextId() {
+      
+      tfdbgContextId_ = getDefaultInstance().getTfdbgContextId();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Unique ID of the context that the executed op(s) belong to (e.g., a
+     * compiled concrete tf.function).
+     * </pre>
+     *
+     * <code>string tfdbg_context_id = 1;</code>
+     */
+    public Builder setTfdbgContextIdBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      tfdbgContextId_ = value;
+      onChanged();
+      return this;
+    }
+
+    private java.lang.Object opName_ = "";
+    /**
+     * <pre>
+     * Name of the op (applicable only in the case of the `FULL_TENSOR` trace
+     * level).
+     * </pre>
+     *
+     * <code>string op_name = 2;</code>
+     */
+    public java.lang.String getOpName() {
+      java.lang.Object ref = opName_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        opName_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Name of the op (applicable only in the case of the `FULL_TENSOR` trace
+     * level).
+     * </pre>
+     *
+     * <code>string op_name = 2;</code>
+     */
+    public com.google.protobuf.ByteString
+        getOpNameBytes() {
+      java.lang.Object ref = opName_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        opName_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Name of the op (applicable only in the case of the `FULL_TENSOR` trace
+     * level).
+     * </pre>
+     *
+     * <code>string op_name = 2;</code>
+     */
+    public Builder setOpName(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      opName_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Name of the op (applicable only in the case of the `FULL_TENSOR` trace
+     * level).
+     * </pre>
+     *
+     * <code>string op_name = 2;</code>
+     */
+    public Builder clearOpName() {
+      
+      opName_ = getDefaultInstance().getOpName();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Name of the op (applicable only in the case of the `FULL_TENSOR` trace
+     * level).
+     * </pre>
+     *
+     * <code>string op_name = 2;</code>
+     */
+    public Builder setOpNameBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      opName_ = value;
+      onChanged();
+      return this;
+    }
+
+    private int outputSlot_ ;
+    /**
+     * <pre>
+     * Output slot of the tensor (applicable only in the case of the `FULL_TENSOR`
+     * trace level).
+     * </pre>
+     *
+     * <code>int32 output_slot = 3;</code>
+     */
+    public int getOutputSlot() {
+      return outputSlot_;
+    }
+    /**
+     * <pre>
+     * Output slot of the tensor (applicable only in the case of the `FULL_TENSOR`
+     * trace level).
+     * </pre>
+     *
+     * <code>int32 output_slot = 3;</code>
+     */
+    public Builder setOutputSlot(int value) {
+      
+      outputSlot_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Output slot of the tensor (applicable only in the case of the `FULL_TENSOR`
+     * trace level).
+     * </pre>
+     *
+     * <code>int32 output_slot = 3;</code>
+     */
+    public Builder clearOutputSlot() {
+      
+      outputSlot_ = 0;
+      onChanged();
+      return this;
+    }
+
+    private int tensorDebugMode_ = 0;
+    /**
+     * <pre>
+     * Type of the tensor value encapsulated in this proto.
+     * </pre>
+     *
+     * <code>.tensorflow.TensorDebugMode tensor_debug_mode = 4;</code>
+     */
+    public int getTensorDebugModeValue() {
+      return tensorDebugMode_;
+    }
+    /**
+     * <pre>
+     * Type of the tensor value encapsulated in this proto.
+     * </pre>
+     *
+     * <code>.tensorflow.TensorDebugMode tensor_debug_mode = 4;</code>
+     */
+    public Builder setTensorDebugModeValue(int value) {
+      tensorDebugMode_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Type of the tensor value encapsulated in this proto.
+     * </pre>
+     *
+     * <code>.tensorflow.TensorDebugMode tensor_debug_mode = 4;</code>
+     */
+    public org.tensorflow.proto.util.TensorDebugMode getTensorDebugMode() {
+      @SuppressWarnings("deprecation")
+      org.tensorflow.proto.util.TensorDebugMode result = org.tensorflow.proto.util.TensorDebugMode.valueOf(tensorDebugMode_);
+      return result == null ? org.tensorflow.proto.util.TensorDebugMode.UNRECOGNIZED : result;
+    }
+    /**
+     * <pre>
+     * Type of the tensor value encapsulated in this proto.
+     * </pre>
+     *
+     * <code>.tensorflow.TensorDebugMode tensor_debug_mode = 4;</code>
+     */
+    public Builder setTensorDebugMode(org.tensorflow.proto.util.TensorDebugMode value) {
+      if (value == null) {
+        throw new NullPointerException();
+      }
+      
+      tensorDebugMode_ = value.getNumber();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Type of the tensor value encapsulated in this proto.
+     * </pre>
+     *
+     * <code>.tensorflow.TensorDebugMode tensor_debug_mode = 4;</code>
+     */
+    public Builder clearTensorDebugMode() {
+      
+      tensorDebugMode_ = 0;
+      onChanged();
+      return this;
+    }
+
+    private org.tensorflow.proto.framework.TensorProto tensorProto_;
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.TensorProto, org.tensorflow.proto.framework.TensorProto.Builder, org.tensorflow.proto.framework.TensorProtoOrBuilder> tensorProtoBuilder_;
+    /**
+     * <pre>
+     * Tensor value in the type described by `tensor_value_type`.
+     * This tensor may summarize the value of a single intermediate op of the
+     * graph, or those of multiple intermediate tensors.
+     * </pre>
+     *
+     * <code>.tensorflow.TensorProto tensor_proto = 5;</code>
+     */
+    public boolean hasTensorProto() {
+      return tensorProtoBuilder_ != null || tensorProto_ != null;
+    }
+    /**
+     * <pre>
+     * Tensor value in the type described by `tensor_value_type`.
+     * This tensor may summarize the value of a single intermediate op of the
+     * graph, or those of multiple intermediate tensors.
+     * </pre>
+     *
+     * <code>.tensorflow.TensorProto tensor_proto = 5;</code>
+     */
+    public org.tensorflow.proto.framework.TensorProto getTensorProto() {
+      if (tensorProtoBuilder_ == null) {
+        return tensorProto_ == null ? org.tensorflow.proto.framework.TensorProto.getDefaultInstance() : tensorProto_;
+      } else {
+        return tensorProtoBuilder_.getMessage();
+      }
+    }
+    /**
+     * <pre>
+     * Tensor value in the type described by `tensor_value_type`.
+     * This tensor may summarize the value of a single intermediate op of the
+     * graph, or those of multiple intermediate tensors.
+     * </pre>
+     *
+     * <code>.tensorflow.TensorProto tensor_proto = 5;</code>
+     */
+    public Builder setTensorProto(org.tensorflow.proto.framework.TensorProto value) {
+      if (tensorProtoBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        tensorProto_ = value;
+        onChanged();
+      } else {
+        tensorProtoBuilder_.setMessage(value);
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Tensor value in the type described by `tensor_value_type`.
+     * This tensor may summarize the value of a single intermediate op of the
+     * graph, or those of multiple intermediate tensors.
+     * </pre>
+     *
+     * <code>.tensorflow.TensorProto tensor_proto = 5;</code>
+     */
+    public Builder setTensorProto(
+        org.tensorflow.proto.framework.TensorProto.Builder builderForValue) {
+      if (tensorProtoBuilder_ == null) {
+        tensorProto_ = builderForValue.build();
+        onChanged();
+      } else {
+        tensorProtoBuilder_.setMessage(builderForValue.build());
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Tensor value in the type described by `tensor_value_type`.
+     * This tensor may summarize the value of a single intermediate op of the
+     * graph, or those of multiple intermediate tensors.
+     * </pre>
+     *
+     * <code>.tensorflow.TensorProto tensor_proto = 5;</code>
+     */
+    public Builder mergeTensorProto(org.tensorflow.proto.framework.TensorProto value) {
+      if (tensorProtoBuilder_ == null) {
+        if (tensorProto_ != null) {
+          tensorProto_ =
+            org.tensorflow.proto.framework.TensorProto.newBuilder(tensorProto_).mergeFrom(value).buildPartial();
+        } else {
+          tensorProto_ = value;
+        }
+        onChanged();
+      } else {
+        tensorProtoBuilder_.mergeFrom(value);
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Tensor value in the type described by `tensor_value_type`.
+     * This tensor may summarize the value of a single intermediate op of the
+     * graph, or those of multiple intermediate tensors.
+     * </pre>
+     *
+     * <code>.tensorflow.TensorProto tensor_proto = 5;</code>
+     */
+    public Builder clearTensorProto() {
+      if (tensorProtoBuilder_ == null) {
+        tensorProto_ = null;
+        onChanged();
+      } else {
+        tensorProto_ = null;
+        tensorProtoBuilder_ = null;
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Tensor value in the type described by `tensor_value_type`.
+     * This tensor may summarize the value of a single intermediate op of the
+     * graph, or those of multiple intermediate tensors.
+     * </pre>
+     *
+     * <code>.tensorflow.TensorProto tensor_proto = 5;</code>
+     */
+    public org.tensorflow.proto.framework.TensorProto.Builder getTensorProtoBuilder() {
+      
+      onChanged();
+      return getTensorProtoFieldBuilder().getBuilder();
+    }
+    /**
+     * <pre>
+     * Tensor value in the type described by `tensor_value_type`.
+     * This tensor may summarize the value of a single intermediate op of the
+     * graph, or those of multiple intermediate tensors.
+     * </pre>
+     *
+     * <code>.tensorflow.TensorProto tensor_proto = 5;</code>
+     */
+    public org.tensorflow.proto.framework.TensorProtoOrBuilder getTensorProtoOrBuilder() {
+      if (tensorProtoBuilder_ != null) {
+        return tensorProtoBuilder_.getMessageOrBuilder();
+      } else {
+        return tensorProto_ == null ?
+            org.tensorflow.proto.framework.TensorProto.getDefaultInstance() : tensorProto_;
+      }
+    }
+    /**
+     * <pre>
+     * Tensor value in the type described by `tensor_value_type`.
+     * This tensor may summarize the value of a single intermediate op of the
+     * graph, or those of multiple intermediate tensors.
+     * </pre>
+     *
+     * <code>.tensorflow.TensorProto tensor_proto = 5;</code>
+     */
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.TensorProto, org.tensorflow.proto.framework.TensorProto.Builder, org.tensorflow.proto.framework.TensorProtoOrBuilder> 
+        getTensorProtoFieldBuilder() {
+      if (tensorProtoBuilder_ == null) {
+        tensorProtoBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+            org.tensorflow.proto.framework.TensorProto, org.tensorflow.proto.framework.TensorProto.Builder, org.tensorflow.proto.framework.TensorProtoOrBuilder>(
+                getTensorProto(),
+                getParentForChildren(),
+                isClean());
+        tensorProto_ = null;
+      }
+      return tensorProtoBuilder_;
+    }
+
+    private java.lang.Object deviceName_ = "";
+    /**
+     * <pre>
+     * Name of the device that the op belongs to.
+     * </pre>
+     *
+     * <code>string device_name = 6;</code>
+     */
+    public java.lang.String getDeviceName() {
+      java.lang.Object ref = deviceName_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        deviceName_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Name of the device that the op belongs to.
+     * </pre>
+     *
+     * <code>string device_name = 6;</code>
+     */
+    public com.google.protobuf.ByteString
+        getDeviceNameBytes() {
+      java.lang.Object ref = deviceName_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        deviceName_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Name of the device that the op belongs to.
+     * </pre>
+     *
+     * <code>string device_name = 6;</code>
+     */
+    public Builder setDeviceName(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      deviceName_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Name of the device that the op belongs to.
+     * </pre>
+     *
+     * <code>string device_name = 6;</code>
+     */
+    public Builder clearDeviceName() {
+      
+      deviceName_ = getDefaultInstance().getDeviceName();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Name of the device that the op belongs to.
+     * </pre>
+     *
+     * <code>string device_name = 6;</code>
+     */
+    public Builder setDeviceNameBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      deviceName_ = value;
+      onChanged();
+      return this;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.GraphExecutionTrace)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.GraphExecutionTrace)
+  private static final org.tensorflow.proto.util.GraphExecutionTrace DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.util.GraphExecutionTrace();
+  }
+
+  public static org.tensorflow.proto.util.GraphExecutionTrace getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<GraphExecutionTrace>
+      PARSER = new com.google.protobuf.AbstractParser<GraphExecutionTrace>() {
+    @java.lang.Override
+    public GraphExecutionTrace parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new GraphExecutionTrace(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<GraphExecutionTrace> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<GraphExecutionTrace> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.util.GraphExecutionTrace getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/GraphExecutionTraceOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/GraphExecutionTraceOrBuilder.java
new file mode 100644
index 00000000000..d83c52096e3
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/GraphExecutionTraceOrBuilder.java
@@ -0,0 +1,125 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/debug_event.proto
+
+package org.tensorflow.proto.util;
+
+public interface GraphExecutionTraceOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.GraphExecutionTrace)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <pre>
+   * Unique ID of the context that the executed op(s) belong to (e.g., a
+   * compiled concrete tf.function).
+   * </pre>
+   *
+   * <code>string tfdbg_context_id = 1;</code>
+   */
+  java.lang.String getTfdbgContextId();
+  /**
+   * <pre>
+   * Unique ID of the context that the executed op(s) belong to (e.g., a
+   * compiled concrete tf.function).
+   * </pre>
+   *
+   * <code>string tfdbg_context_id = 1;</code>
+   */
+  com.google.protobuf.ByteString
+      getTfdbgContextIdBytes();
+
+  /**
+   * <pre>
+   * Name of the op (applicable only in the case of the `FULL_TENSOR` trace
+   * level).
+   * </pre>
+   *
+   * <code>string op_name = 2;</code>
+   */
+  java.lang.String getOpName();
+  /**
+   * <pre>
+   * Name of the op (applicable only in the case of the `FULL_TENSOR` trace
+   * level).
+   * </pre>
+   *
+   * <code>string op_name = 2;</code>
+   */
+  com.google.protobuf.ByteString
+      getOpNameBytes();
+
+  /**
+   * <pre>
+   * Output slot of the tensor (applicable only in the case of the `FULL_TENSOR`
+   * trace level).
+   * </pre>
+   *
+   * <code>int32 output_slot = 3;</code>
+   */
+  int getOutputSlot();
+
+  /**
+   * <pre>
+   * Type of the tensor value encapsulated in this proto.
+   * </pre>
+   *
+   * <code>.tensorflow.TensorDebugMode tensor_debug_mode = 4;</code>
+   */
+  int getTensorDebugModeValue();
+  /**
+   * <pre>
+   * Type of the tensor value encapsulated in this proto.
+   * </pre>
+   *
+   * <code>.tensorflow.TensorDebugMode tensor_debug_mode = 4;</code>
+   */
+  org.tensorflow.proto.util.TensorDebugMode getTensorDebugMode();
+
+  /**
+   * <pre>
+   * Tensor value in the type described by `tensor_value_type`.
+   * This tensor may summarize the value of a single intermediate op of the
+   * graph, or those of multiple intermediate tensors.
+   * </pre>
+   *
+   * <code>.tensorflow.TensorProto tensor_proto = 5;</code>
+   */
+  boolean hasTensorProto();
+  /**
+   * <pre>
+   * Tensor value in the type described by `tensor_value_type`.
+   * This tensor may summarize the value of a single intermediate op of the
+   * graph, or those of multiple intermediate tensors.
+   * </pre>
+   *
+   * <code>.tensorflow.TensorProto tensor_proto = 5;</code>
+   */
+  org.tensorflow.proto.framework.TensorProto getTensorProto();
+  /**
+   * <pre>
+   * Tensor value in the type described by `tensor_value_type`.
+   * This tensor may summarize the value of a single intermediate op of the
+   * graph, or those of multiple intermediate tensors.
+   * </pre>
+   *
+   * <code>.tensorflow.TensorProto tensor_proto = 5;</code>
+   */
+  org.tensorflow.proto.framework.TensorProtoOrBuilder getTensorProtoOrBuilder();
+
+  /**
+   * <pre>
+   * Name of the device that the op belongs to.
+   * </pre>
+   *
+   * <code>string device_name = 6;</code>
+   */
+  java.lang.String getDeviceName();
+  /**
+   * <pre>
+   * Name of the device that the op belongs to.
+   * </pre>
+   *
+   * <code>string device_name = 6;</code>
+   */
+  com.google.protobuf.ByteString
+      getDeviceNameBytes();
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/GraphOpCreation.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/GraphOpCreation.java
new file mode 100644
index 00000000000..da8a4523ec0
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/GraphOpCreation.java
@@ -0,0 +1,1936 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/debug_event.proto
+
+package org.tensorflow.proto.util;
+
+/**
+ * <pre>
+ * The creation of an op in a TensorFlow Graph (e.g., FuncGraph in TF2).
+ * </pre>
+ *
+ * Protobuf type {@code tensorflow.GraphOpCreation}
+ */
+public  final class GraphOpCreation extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.GraphOpCreation)
+    GraphOpCreationOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use GraphOpCreation.newBuilder() to construct.
+  private GraphOpCreation(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private GraphOpCreation() {
+    opType_ = "";
+    opName_ = "";
+    graphName_ = "";
+    graphId_ = "";
+    deviceName_ = "";
+    inputNames_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+    outputTensorIds_ = emptyIntList();
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new GraphOpCreation();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private GraphOpCreation(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    int mutable_bitField0_ = 0;
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 10: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            opType_ = s;
+            break;
+          }
+          case 18: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            opName_ = s;
+            break;
+          }
+          case 26: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            graphName_ = s;
+            break;
+          }
+          case 34: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            graphId_ = s;
+            break;
+          }
+          case 42: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            deviceName_ = s;
+            break;
+          }
+          case 50: {
+            java.lang.String s = input.readStringRequireUtf8();
+            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
+              inputNames_ = new com.google.protobuf.LazyStringArrayList();
+              mutable_bitField0_ |= 0x00000001;
+            }
+            inputNames_.add(s);
+            break;
+          }
+          case 56: {
+
+            numOutputs_ = input.readInt32();
+            break;
+          }
+          case 66: {
+            org.tensorflow.proto.util.CodeLocation.Builder subBuilder = null;
+            if (codeLocation_ != null) {
+              subBuilder = codeLocation_.toBuilder();
+            }
+            codeLocation_ = input.readMessage(org.tensorflow.proto.util.CodeLocation.parser(), extensionRegistry);
+            if (subBuilder != null) {
+              subBuilder.mergeFrom(codeLocation_);
+              codeLocation_ = subBuilder.buildPartial();
+            }
+
+            break;
+          }
+          case 72: {
+            if (!((mutable_bitField0_ & 0x00000002) != 0)) {
+              outputTensorIds_ = newIntList();
+              mutable_bitField0_ |= 0x00000002;
+            }
+            outputTensorIds_.addInt(input.readInt32());
+            break;
+          }
+          case 74: {
+            int length = input.readRawVarint32();
+            int limit = input.pushLimit(length);
+            if (!((mutable_bitField0_ & 0x00000002) != 0) && input.getBytesUntilLimit() > 0) {
+              outputTensorIds_ = newIntList();
+              mutable_bitField0_ |= 0x00000002;
+            }
+            while (input.getBytesUntilLimit() > 0) {
+              outputTensorIds_.addInt(input.readInt32());
+            }
+            input.popLimit(limit);
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      if (((mutable_bitField0_ & 0x00000001) != 0)) {
+        inputNames_ = inputNames_.getUnmodifiableView();
+      }
+      if (((mutable_bitField0_ & 0x00000002) != 0)) {
+        outputTensorIds_.makeImmutable(); // C
+      }
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.util.DebugEventProtos.internal_static_tensorflow_GraphOpCreation_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.util.DebugEventProtos.internal_static_tensorflow_GraphOpCreation_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.util.GraphOpCreation.class, org.tensorflow.proto.util.GraphOpCreation.Builder.class);
+  }
+
+  public static final int OP_TYPE_FIELD_NUMBER = 1;
+  private volatile java.lang.Object opType_;
+  /**
+   * <pre>
+   * Type of the op (e.g., "MatMul").
+   * </pre>
+   *
+   * <code>string op_type = 1;</code>
+   */
+  public java.lang.String getOpType() {
+    java.lang.Object ref = opType_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      opType_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * Type of the op (e.g., "MatMul").
+   * </pre>
+   *
+   * <code>string op_type = 1;</code>
+   */
+  public com.google.protobuf.ByteString
+      getOpTypeBytes() {
+    java.lang.Object ref = opType_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      opType_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int OP_NAME_FIELD_NUMBER = 2;
+  private volatile java.lang.Object opName_;
+  /**
+   * <pre>
+   * Name of the op (e.g., "Dense/MatMul_1").
+   * </pre>
+   *
+   * <code>string op_name = 2;</code>
+   */
+  public java.lang.String getOpName() {
+    java.lang.Object ref = opName_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      opName_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * Name of the op (e.g., "Dense/MatMul_1").
+   * </pre>
+   *
+   * <code>string op_name = 2;</code>
+   */
+  public com.google.protobuf.ByteString
+      getOpNameBytes() {
+    java.lang.Object ref = opName_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      opName_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int GRAPH_NAME_FIELD_NUMBER = 3;
+  private volatile java.lang.Object graphName_;
+  /**
+   * <pre>
+   * Name of the graph that the op is a part of (if available).
+   * </pre>
+   *
+   * <code>string graph_name = 3;</code>
+   */
+  public java.lang.String getGraphName() {
+    java.lang.Object ref = graphName_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      graphName_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * Name of the graph that the op is a part of (if available).
+   * </pre>
+   *
+   * <code>string graph_name = 3;</code>
+   */
+  public com.google.protobuf.ByteString
+      getGraphNameBytes() {
+    java.lang.Object ref = graphName_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      graphName_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int GRAPH_ID_FIELD_NUMBER = 4;
+  private volatile java.lang.Object graphId_;
+  /**
+   * <pre>
+   * Unique ID of the graph (generated by debugger).
+   * This is the ID of the immediately-enclosing graph.
+   * </pre>
+   *
+   * <code>string graph_id = 4;</code>
+   */
+  public java.lang.String getGraphId() {
+    java.lang.Object ref = graphId_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      graphId_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * Unique ID of the graph (generated by debugger).
+   * This is the ID of the immediately-enclosing graph.
+   * </pre>
+   *
+   * <code>string graph_id = 4;</code>
+   */
+  public com.google.protobuf.ByteString
+      getGraphIdBytes() {
+    java.lang.Object ref = graphId_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      graphId_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int DEVICE_NAME_FIELD_NUMBER = 5;
+  private volatile java.lang.Object deviceName_;
+  /**
+   * <pre>
+   * Name of the device that the op is assigned to (if available).
+   * </pre>
+   *
+   * <code>string device_name = 5;</code>
+   */
+  public java.lang.String getDeviceName() {
+    java.lang.Object ref = deviceName_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      deviceName_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * Name of the device that the op is assigned to (if available).
+   * </pre>
+   *
+   * <code>string device_name = 5;</code>
+   */
+  public com.google.protobuf.ByteString
+      getDeviceNameBytes() {
+    java.lang.Object ref = deviceName_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      deviceName_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int INPUT_NAMES_FIELD_NUMBER = 6;
+  private com.google.protobuf.LazyStringList inputNames_;
+  /**
+   * <pre>
+   * Names of the input tensors to the op.
+   * </pre>
+   *
+   * <code>repeated string input_names = 6;</code>
+   */
+  public com.google.protobuf.ProtocolStringList
+      getInputNamesList() {
+    return inputNames_;
+  }
+  /**
+   * <pre>
+   * Names of the input tensors to the op.
+   * </pre>
+   *
+   * <code>repeated string input_names = 6;</code>
+   */
+  public int getInputNamesCount() {
+    return inputNames_.size();
+  }
+  /**
+   * <pre>
+   * Names of the input tensors to the op.
+   * </pre>
+   *
+   * <code>repeated string input_names = 6;</code>
+   */
+  public java.lang.String getInputNames(int index) {
+    return inputNames_.get(index);
+  }
+  /**
+   * <pre>
+   * Names of the input tensors to the op.
+   * </pre>
+   *
+   * <code>repeated string input_names = 6;</code>
+   */
+  public com.google.protobuf.ByteString
+      getInputNamesBytes(int index) {
+    return inputNames_.getByteString(index);
+  }
+
+  public static final int NUM_OUTPUTS_FIELD_NUMBER = 7;
+  private int numOutputs_;
+  /**
+   * <pre>
+   * Number of output tensors emitted by the op.
+   * </pre>
+   *
+   * <code>int32 num_outputs = 7;</code>
+   */
+  public int getNumOutputs() {
+    return numOutputs_;
+  }
+
+  public static final int CODE_LOCATION_FIELD_NUMBER = 8;
+  private org.tensorflow.proto.util.CodeLocation codeLocation_;
+  /**
+   * <pre>
+   * The unique ID for code location (stack trace) of the op's creation.
+   * </pre>
+   *
+   * <code>.tensorflow.CodeLocation code_location = 8;</code>
+   */
+  public boolean hasCodeLocation() {
+    return codeLocation_ != null;
+  }
+  /**
+   * <pre>
+   * The unique ID for code location (stack trace) of the op's creation.
+   * </pre>
+   *
+   * <code>.tensorflow.CodeLocation code_location = 8;</code>
+   */
+  public org.tensorflow.proto.util.CodeLocation getCodeLocation() {
+    return codeLocation_ == null ? org.tensorflow.proto.util.CodeLocation.getDefaultInstance() : codeLocation_;
+  }
+  /**
+   * <pre>
+   * The unique ID for code location (stack trace) of the op's creation.
+   * </pre>
+   *
+   * <code>.tensorflow.CodeLocation code_location = 8;</code>
+   */
+  public org.tensorflow.proto.util.CodeLocationOrBuilder getCodeLocationOrBuilder() {
+    return getCodeLocation();
+  }
+
+  public static final int OUTPUT_TENSOR_IDS_FIELD_NUMBER = 9;
+  private com.google.protobuf.Internal.IntList outputTensorIds_;
+  /**
+   * <pre>
+   * Unique IDs for the output tensors of this op.
+   * </pre>
+   *
+   * <code>repeated int32 output_tensor_ids = 9;</code>
+   */
+  public java.util.List<java.lang.Integer>
+      getOutputTensorIdsList() {
+    return outputTensorIds_;
+  }
+  /**
+   * <pre>
+   * Unique IDs for the output tensors of this op.
+   * </pre>
+   *
+   * <code>repeated int32 output_tensor_ids = 9;</code>
+   */
+  public int getOutputTensorIdsCount() {
+    return outputTensorIds_.size();
+  }
+  /**
+   * <pre>
+   * Unique IDs for the output tensors of this op.
+   * </pre>
+   *
+   * <code>repeated int32 output_tensor_ids = 9;</code>
+   */
+  public int getOutputTensorIds(int index) {
+    return outputTensorIds_.getInt(index);
+  }
+  private int outputTensorIdsMemoizedSerializedSize = -1;
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    getSerializedSize();
+    if (!getOpTypeBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 1, opType_);
+    }
+    if (!getOpNameBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 2, opName_);
+    }
+    if (!getGraphNameBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 3, graphName_);
+    }
+    if (!getGraphIdBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 4, graphId_);
+    }
+    if (!getDeviceNameBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 5, deviceName_);
+    }
+    for (int i = 0; i < inputNames_.size(); i++) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 6, inputNames_.getRaw(i));
+    }
+    if (numOutputs_ != 0) {
+      output.writeInt32(7, numOutputs_);
+    }
+    if (codeLocation_ != null) {
+      output.writeMessage(8, getCodeLocation());
+    }
+    if (getOutputTensorIdsList().size() > 0) {
+      output.writeUInt32NoTag(74);
+      output.writeUInt32NoTag(outputTensorIdsMemoizedSerializedSize);
+    }
+    for (int i = 0; i < outputTensorIds_.size(); i++) {
+      output.writeInt32NoTag(outputTensorIds_.getInt(i));
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (!getOpTypeBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, opType_);
+    }
+    if (!getOpNameBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(2, opName_);
+    }
+    if (!getGraphNameBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(3, graphName_);
+    }
+    if (!getGraphIdBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(4, graphId_);
+    }
+    if (!getDeviceNameBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(5, deviceName_);
+    }
+    {
+      int dataSize = 0;
+      for (int i = 0; i < inputNames_.size(); i++) {
+        dataSize += computeStringSizeNoTag(inputNames_.getRaw(i));
+      }
+      size += dataSize;
+      size += 1 * getInputNamesList().size();
+    }
+    if (numOutputs_ != 0) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt32Size(7, numOutputs_);
+    }
+    if (codeLocation_ != null) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(8, getCodeLocation());
+    }
+    {
+      int dataSize = 0;
+      for (int i = 0; i < outputTensorIds_.size(); i++) {
+        dataSize += com.google.protobuf.CodedOutputStream
+          .computeInt32SizeNoTag(outputTensorIds_.getInt(i));
+      }
+      size += dataSize;
+      if (!getOutputTensorIdsList().isEmpty()) {
+        size += 1;
+        size += com.google.protobuf.CodedOutputStream
+            .computeInt32SizeNoTag(dataSize);
+      }
+      outputTensorIdsMemoizedSerializedSize = dataSize;
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.util.GraphOpCreation)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.util.GraphOpCreation other = (org.tensorflow.proto.util.GraphOpCreation) obj;
+
+    if (!getOpType()
+        .equals(other.getOpType())) return false;
+    if (!getOpName()
+        .equals(other.getOpName())) return false;
+    if (!getGraphName()
+        .equals(other.getGraphName())) return false;
+    if (!getGraphId()
+        .equals(other.getGraphId())) return false;
+    if (!getDeviceName()
+        .equals(other.getDeviceName())) return false;
+    if (!getInputNamesList()
+        .equals(other.getInputNamesList())) return false;
+    if (getNumOutputs()
+        != other.getNumOutputs()) return false;
+    if (hasCodeLocation() != other.hasCodeLocation()) return false;
+    if (hasCodeLocation()) {
+      if (!getCodeLocation()
+          .equals(other.getCodeLocation())) return false;
+    }
+    if (!getOutputTensorIdsList()
+        .equals(other.getOutputTensorIdsList())) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    hash = (37 * hash) + OP_TYPE_FIELD_NUMBER;
+    hash = (53 * hash) + getOpType().hashCode();
+    hash = (37 * hash) + OP_NAME_FIELD_NUMBER;
+    hash = (53 * hash) + getOpName().hashCode();
+    hash = (37 * hash) + GRAPH_NAME_FIELD_NUMBER;
+    hash = (53 * hash) + getGraphName().hashCode();
+    hash = (37 * hash) + GRAPH_ID_FIELD_NUMBER;
+    hash = (53 * hash) + getGraphId().hashCode();
+    hash = (37 * hash) + DEVICE_NAME_FIELD_NUMBER;
+    hash = (53 * hash) + getDeviceName().hashCode();
+    if (getInputNamesCount() > 0) {
+      hash = (37 * hash) + INPUT_NAMES_FIELD_NUMBER;
+      hash = (53 * hash) + getInputNamesList().hashCode();
+    }
+    hash = (37 * hash) + NUM_OUTPUTS_FIELD_NUMBER;
+    hash = (53 * hash) + getNumOutputs();
+    if (hasCodeLocation()) {
+      hash = (37 * hash) + CODE_LOCATION_FIELD_NUMBER;
+      hash = (53 * hash) + getCodeLocation().hashCode();
+    }
+    if (getOutputTensorIdsCount() > 0) {
+      hash = (37 * hash) + OUTPUT_TENSOR_IDS_FIELD_NUMBER;
+      hash = (53 * hash) + getOutputTensorIdsList().hashCode();
+    }
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.util.GraphOpCreation parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.GraphOpCreation parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.GraphOpCreation parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.GraphOpCreation parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.GraphOpCreation parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.GraphOpCreation parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.GraphOpCreation parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.GraphOpCreation parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.GraphOpCreation parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.GraphOpCreation parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.GraphOpCreation parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.GraphOpCreation parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.util.GraphOpCreation prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * <pre>
+   * The creation of an op in a TensorFlow Graph (e.g., FuncGraph in TF2).
+   * </pre>
+   *
+   * Protobuf type {@code tensorflow.GraphOpCreation}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.GraphOpCreation)
+      org.tensorflow.proto.util.GraphOpCreationOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.util.DebugEventProtos.internal_static_tensorflow_GraphOpCreation_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.util.DebugEventProtos.internal_static_tensorflow_GraphOpCreation_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.util.GraphOpCreation.class, org.tensorflow.proto.util.GraphOpCreation.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.util.GraphOpCreation.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      opType_ = "";
+
+      opName_ = "";
+
+      graphName_ = "";
+
+      graphId_ = "";
+
+      deviceName_ = "";
+
+      inputNames_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+      bitField0_ = (bitField0_ & ~0x00000001);
+      numOutputs_ = 0;
+
+      if (codeLocationBuilder_ == null) {
+        codeLocation_ = null;
+      } else {
+        codeLocation_ = null;
+        codeLocationBuilder_ = null;
+      }
+      outputTensorIds_ = emptyIntList();
+      bitField0_ = (bitField0_ & ~0x00000002);
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.util.DebugEventProtos.internal_static_tensorflow_GraphOpCreation_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.GraphOpCreation getDefaultInstanceForType() {
+      return org.tensorflow.proto.util.GraphOpCreation.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.GraphOpCreation build() {
+      org.tensorflow.proto.util.GraphOpCreation result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.GraphOpCreation buildPartial() {
+      org.tensorflow.proto.util.GraphOpCreation result = new org.tensorflow.proto.util.GraphOpCreation(this);
+      int from_bitField0_ = bitField0_;
+      result.opType_ = opType_;
+      result.opName_ = opName_;
+      result.graphName_ = graphName_;
+      result.graphId_ = graphId_;
+      result.deviceName_ = deviceName_;
+      if (((bitField0_ & 0x00000001) != 0)) {
+        inputNames_ = inputNames_.getUnmodifiableView();
+        bitField0_ = (bitField0_ & ~0x00000001);
+      }
+      result.inputNames_ = inputNames_;
+      result.numOutputs_ = numOutputs_;
+      if (codeLocationBuilder_ == null) {
+        result.codeLocation_ = codeLocation_;
+      } else {
+        result.codeLocation_ = codeLocationBuilder_.build();
+      }
+      if (((bitField0_ & 0x00000002) != 0)) {
+        outputTensorIds_.makeImmutable();
+        bitField0_ = (bitField0_ & ~0x00000002);
+      }
+      result.outputTensorIds_ = outputTensorIds_;
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.util.GraphOpCreation) {
+        return mergeFrom((org.tensorflow.proto.util.GraphOpCreation)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.util.GraphOpCreation other) {
+      if (other == org.tensorflow.proto.util.GraphOpCreation.getDefaultInstance()) return this;
+      if (!other.getOpType().isEmpty()) {
+        opType_ = other.opType_;
+        onChanged();
+      }
+      if (!other.getOpName().isEmpty()) {
+        opName_ = other.opName_;
+        onChanged();
+      }
+      if (!other.getGraphName().isEmpty()) {
+        graphName_ = other.graphName_;
+        onChanged();
+      }
+      if (!other.getGraphId().isEmpty()) {
+        graphId_ = other.graphId_;
+        onChanged();
+      }
+      if (!other.getDeviceName().isEmpty()) {
+        deviceName_ = other.deviceName_;
+        onChanged();
+      }
+      if (!other.inputNames_.isEmpty()) {
+        if (inputNames_.isEmpty()) {
+          inputNames_ = other.inputNames_;
+          bitField0_ = (bitField0_ & ~0x00000001);
+        } else {
+          ensureInputNamesIsMutable();
+          inputNames_.addAll(other.inputNames_);
+        }
+        onChanged();
+      }
+      if (other.getNumOutputs() != 0) {
+        setNumOutputs(other.getNumOutputs());
+      }
+      if (other.hasCodeLocation()) {
+        mergeCodeLocation(other.getCodeLocation());
+      }
+      if (!other.outputTensorIds_.isEmpty()) {
+        if (outputTensorIds_.isEmpty()) {
+          outputTensorIds_ = other.outputTensorIds_;
+          bitField0_ = (bitField0_ & ~0x00000002);
+        } else {
+          ensureOutputTensorIdsIsMutable();
+          outputTensorIds_.addAll(other.outputTensorIds_);
+        }
+        onChanged();
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.util.GraphOpCreation parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.util.GraphOpCreation) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+    private int bitField0_;
+
+    private java.lang.Object opType_ = "";
+    /**
+     * <pre>
+     * Type of the op (e.g., "MatMul").
+     * </pre>
+     *
+     * <code>string op_type = 1;</code>
+     */
+    public java.lang.String getOpType() {
+      java.lang.Object ref = opType_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        opType_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Type of the op (e.g., "MatMul").
+     * </pre>
+     *
+     * <code>string op_type = 1;</code>
+     */
+    public com.google.protobuf.ByteString
+        getOpTypeBytes() {
+      java.lang.Object ref = opType_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        opType_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Type of the op (e.g., "MatMul").
+     * </pre>
+     *
+     * <code>string op_type = 1;</code>
+     */
+    public Builder setOpType(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      opType_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Type of the op (e.g., "MatMul").
+     * </pre>
+     *
+     * <code>string op_type = 1;</code>
+     */
+    public Builder clearOpType() {
+      
+      opType_ = getDefaultInstance().getOpType();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Type of the op (e.g., "MatMul").
+     * </pre>
+     *
+     * <code>string op_type = 1;</code>
+     */
+    public Builder setOpTypeBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      opType_ = value;
+      onChanged();
+      return this;
+    }
+
+    private java.lang.Object opName_ = "";
+    /**
+     * <pre>
+     * Name of the op (e.g., "Dense/MatMul_1").
+     * </pre>
+     *
+     * <code>string op_name = 2;</code>
+     */
+    public java.lang.String getOpName() {
+      java.lang.Object ref = opName_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        opName_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Name of the op (e.g., "Dense/MatMul_1").
+     * </pre>
+     *
+     * <code>string op_name = 2;</code>
+     */
+    public com.google.protobuf.ByteString
+        getOpNameBytes() {
+      java.lang.Object ref = opName_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        opName_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Name of the op (e.g., "Dense/MatMul_1").
+     * </pre>
+     *
+     * <code>string op_name = 2;</code>
+     */
+    public Builder setOpName(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      opName_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Name of the op (e.g., "Dense/MatMul_1").
+     * </pre>
+     *
+     * <code>string op_name = 2;</code>
+     */
+    public Builder clearOpName() {
+      
+      opName_ = getDefaultInstance().getOpName();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Name of the op (e.g., "Dense/MatMul_1").
+     * </pre>
+     *
+     * <code>string op_name = 2;</code>
+     */
+    public Builder setOpNameBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      opName_ = value;
+      onChanged();
+      return this;
+    }
+
+    private java.lang.Object graphName_ = "";
+    /**
+     * <pre>
+     * Name of the graph that the op is a part of (if available).
+     * </pre>
+     *
+     * <code>string graph_name = 3;</code>
+     */
+    public java.lang.String getGraphName() {
+      java.lang.Object ref = graphName_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        graphName_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Name of the graph that the op is a part of (if available).
+     * </pre>
+     *
+     * <code>string graph_name = 3;</code>
+     */
+    public com.google.protobuf.ByteString
+        getGraphNameBytes() {
+      java.lang.Object ref = graphName_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        graphName_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Name of the graph that the op is a part of (if available).
+     * </pre>
+     *
+     * <code>string graph_name = 3;</code>
+     */
+    public Builder setGraphName(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      graphName_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Name of the graph that the op is a part of (if available).
+     * </pre>
+     *
+     * <code>string graph_name = 3;</code>
+     */
+    public Builder clearGraphName() {
+      
+      graphName_ = getDefaultInstance().getGraphName();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Name of the graph that the op is a part of (if available).
+     * </pre>
+     *
+     * <code>string graph_name = 3;</code>
+     */
+    public Builder setGraphNameBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      graphName_ = value;
+      onChanged();
+      return this;
+    }
+
+    private java.lang.Object graphId_ = "";
+    /**
+     * <pre>
+     * Unique ID of the graph (generated by debugger).
+     * This is the ID of the immediately-enclosing graph.
+     * </pre>
+     *
+     * <code>string graph_id = 4;</code>
+     */
+    public java.lang.String getGraphId() {
+      java.lang.Object ref = graphId_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        graphId_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Unique ID of the graph (generated by debugger).
+     * This is the ID of the immediately-enclosing graph.
+     * </pre>
+     *
+     * <code>string graph_id = 4;</code>
+     */
+    public com.google.protobuf.ByteString
+        getGraphIdBytes() {
+      java.lang.Object ref = graphId_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        graphId_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Unique ID of the graph (generated by debugger).
+     * This is the ID of the immediately-enclosing graph.
+     * </pre>
+     *
+     * <code>string graph_id = 4;</code>
+     */
+    public Builder setGraphId(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      graphId_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Unique ID of the graph (generated by debugger).
+     * This is the ID of the immediately-enclosing graph.
+     * </pre>
+     *
+     * <code>string graph_id = 4;</code>
+     */
+    public Builder clearGraphId() {
+      
+      graphId_ = getDefaultInstance().getGraphId();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Unique ID of the graph (generated by debugger).
+     * This is the ID of the immediately-enclosing graph.
+     * </pre>
+     *
+     * <code>string graph_id = 4;</code>
+     */
+    public Builder setGraphIdBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      graphId_ = value;
+      onChanged();
+      return this;
+    }
+
+    private java.lang.Object deviceName_ = "";
+    /**
+     * <pre>
+     * Name of the device that the op is assigned to (if available).
+     * </pre>
+     *
+     * <code>string device_name = 5;</code>
+     */
+    public java.lang.String getDeviceName() {
+      java.lang.Object ref = deviceName_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        deviceName_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Name of the device that the op is assigned to (if available).
+     * </pre>
+     *
+     * <code>string device_name = 5;</code>
+     */
+    public com.google.protobuf.ByteString
+        getDeviceNameBytes() {
+      java.lang.Object ref = deviceName_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        deviceName_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Name of the device that the op is assigned to (if available).
+     * </pre>
+     *
+     * <code>string device_name = 5;</code>
+     */
+    public Builder setDeviceName(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      deviceName_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Name of the device that the op is assigned to (if available).
+     * </pre>
+     *
+     * <code>string device_name = 5;</code>
+     */
+    public Builder clearDeviceName() {
+      
+      deviceName_ = getDefaultInstance().getDeviceName();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Name of the device that the op is assigned to (if available).
+     * </pre>
+     *
+     * <code>string device_name = 5;</code>
+     */
+    public Builder setDeviceNameBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      deviceName_ = value;
+      onChanged();
+      return this;
+    }
+
+    private com.google.protobuf.LazyStringList inputNames_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+    private void ensureInputNamesIsMutable() {
+      if (!((bitField0_ & 0x00000001) != 0)) {
+        inputNames_ = new com.google.protobuf.LazyStringArrayList(inputNames_);
+        bitField0_ |= 0x00000001;
+       }
+    }
+    /**
+     * <pre>
+     * Names of the input tensors to the op.
+     * </pre>
+     *
+     * <code>repeated string input_names = 6;</code>
+     */
+    public com.google.protobuf.ProtocolStringList
+        getInputNamesList() {
+      return inputNames_.getUnmodifiableView();
+    }
+    /**
+     * <pre>
+     * Names of the input tensors to the op.
+     * </pre>
+     *
+     * <code>repeated string input_names = 6;</code>
+     */
+    public int getInputNamesCount() {
+      return inputNames_.size();
+    }
+    /**
+     * <pre>
+     * Names of the input tensors to the op.
+     * </pre>
+     *
+     * <code>repeated string input_names = 6;</code>
+     */
+    public java.lang.String getInputNames(int index) {
+      return inputNames_.get(index);
+    }
+    /**
+     * <pre>
+     * Names of the input tensors to the op.
+     * </pre>
+     *
+     * <code>repeated string input_names = 6;</code>
+     */
+    public com.google.protobuf.ByteString
+        getInputNamesBytes(int index) {
+      return inputNames_.getByteString(index);
+    }
+    /**
+     * <pre>
+     * Names of the input tensors to the op.
+     * </pre>
+     *
+     * <code>repeated string input_names = 6;</code>
+     */
+    public Builder setInputNames(
+        int index, java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  ensureInputNamesIsMutable();
+      inputNames_.set(index, value);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Names of the input tensors to the op.
+     * </pre>
+     *
+     * <code>repeated string input_names = 6;</code>
+     */
+    public Builder addInputNames(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  ensureInputNamesIsMutable();
+      inputNames_.add(value);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Names of the input tensors to the op.
+     * </pre>
+     *
+     * <code>repeated string input_names = 6;</code>
+     */
+    public Builder addAllInputNames(
+        java.lang.Iterable<java.lang.String> values) {
+      ensureInputNamesIsMutable();
+      com.google.protobuf.AbstractMessageLite.Builder.addAll(
+          values, inputNames_);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Names of the input tensors to the op.
+     * </pre>
+     *
+     * <code>repeated string input_names = 6;</code>
+     */
+    public Builder clearInputNames() {
+      inputNames_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+      bitField0_ = (bitField0_ & ~0x00000001);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Names of the input tensors to the op.
+     * </pre>
+     *
+     * <code>repeated string input_names = 6;</code>
+     */
+    public Builder addInputNamesBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      ensureInputNamesIsMutable();
+      inputNames_.add(value);
+      onChanged();
+      return this;
+    }
+
+    private int numOutputs_ ;
+    /**
+     * <pre>
+     * Number of output tensors emitted by the op.
+     * </pre>
+     *
+     * <code>int32 num_outputs = 7;</code>
+     */
+    public int getNumOutputs() {
+      return numOutputs_;
+    }
+    /**
+     * <pre>
+     * Number of output tensors emitted by the op.
+     * </pre>
+     *
+     * <code>int32 num_outputs = 7;</code>
+     */
+    public Builder setNumOutputs(int value) {
+      
+      numOutputs_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Number of output tensors emitted by the op.
+     * </pre>
+     *
+     * <code>int32 num_outputs = 7;</code>
+     */
+    public Builder clearNumOutputs() {
+      
+      numOutputs_ = 0;
+      onChanged();
+      return this;
+    }
+
+    private org.tensorflow.proto.util.CodeLocation codeLocation_;
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.util.CodeLocation, org.tensorflow.proto.util.CodeLocation.Builder, org.tensorflow.proto.util.CodeLocationOrBuilder> codeLocationBuilder_;
+    /**
+     * <pre>
+     * The unique ID for code location (stack trace) of the op's creation.
+     * </pre>
+     *
+     * <code>.tensorflow.CodeLocation code_location = 8;</code>
+     */
+    public boolean hasCodeLocation() {
+      return codeLocationBuilder_ != null || codeLocation_ != null;
+    }
+    /**
+     * <pre>
+     * The unique ID for code location (stack trace) of the op's creation.
+     * </pre>
+     *
+     * <code>.tensorflow.CodeLocation code_location = 8;</code>
+     */
+    public org.tensorflow.proto.util.CodeLocation getCodeLocation() {
+      if (codeLocationBuilder_ == null) {
+        return codeLocation_ == null ? org.tensorflow.proto.util.CodeLocation.getDefaultInstance() : codeLocation_;
+      } else {
+        return codeLocationBuilder_.getMessage();
+      }
+    }
+    /**
+     * <pre>
+     * The unique ID for code location (stack trace) of the op's creation.
+     * </pre>
+     *
+     * <code>.tensorflow.CodeLocation code_location = 8;</code>
+     */
+    public Builder setCodeLocation(org.tensorflow.proto.util.CodeLocation value) {
+      if (codeLocationBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        codeLocation_ = value;
+        onChanged();
+      } else {
+        codeLocationBuilder_.setMessage(value);
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * The unique ID for code location (stack trace) of the op's creation.
+     * </pre>
+     *
+     * <code>.tensorflow.CodeLocation code_location = 8;</code>
+     */
+    public Builder setCodeLocation(
+        org.tensorflow.proto.util.CodeLocation.Builder builderForValue) {
+      if (codeLocationBuilder_ == null) {
+        codeLocation_ = builderForValue.build();
+        onChanged();
+      } else {
+        codeLocationBuilder_.setMessage(builderForValue.build());
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * The unique ID for code location (stack trace) of the op's creation.
+     * </pre>
+     *
+     * <code>.tensorflow.CodeLocation code_location = 8;</code>
+     */
+    public Builder mergeCodeLocation(org.tensorflow.proto.util.CodeLocation value) {
+      if (codeLocationBuilder_ == null) {
+        if (codeLocation_ != null) {
+          codeLocation_ =
+            org.tensorflow.proto.util.CodeLocation.newBuilder(codeLocation_).mergeFrom(value).buildPartial();
+        } else {
+          codeLocation_ = value;
+        }
+        onChanged();
+      } else {
+        codeLocationBuilder_.mergeFrom(value);
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * The unique ID for code location (stack trace) of the op's creation.
+     * </pre>
+     *
+     * <code>.tensorflow.CodeLocation code_location = 8;</code>
+     */
+    public Builder clearCodeLocation() {
+      if (codeLocationBuilder_ == null) {
+        codeLocation_ = null;
+        onChanged();
+      } else {
+        codeLocation_ = null;
+        codeLocationBuilder_ = null;
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * The unique ID for code location (stack trace) of the op's creation.
+     * </pre>
+     *
+     * <code>.tensorflow.CodeLocation code_location = 8;</code>
+     */
+    public org.tensorflow.proto.util.CodeLocation.Builder getCodeLocationBuilder() {
+      
+      onChanged();
+      return getCodeLocationFieldBuilder().getBuilder();
+    }
+    /**
+     * <pre>
+     * The unique ID for code location (stack trace) of the op's creation.
+     * </pre>
+     *
+     * <code>.tensorflow.CodeLocation code_location = 8;</code>
+     */
+    public org.tensorflow.proto.util.CodeLocationOrBuilder getCodeLocationOrBuilder() {
+      if (codeLocationBuilder_ != null) {
+        return codeLocationBuilder_.getMessageOrBuilder();
+      } else {
+        return codeLocation_ == null ?
+            org.tensorflow.proto.util.CodeLocation.getDefaultInstance() : codeLocation_;
+      }
+    }
+    /**
+     * <pre>
+     * The unique ID for code location (stack trace) of the op's creation.
+     * </pre>
+     *
+     * <code>.tensorflow.CodeLocation code_location = 8;</code>
+     */
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.util.CodeLocation, org.tensorflow.proto.util.CodeLocation.Builder, org.tensorflow.proto.util.CodeLocationOrBuilder> 
+        getCodeLocationFieldBuilder() {
+      if (codeLocationBuilder_ == null) {
+        codeLocationBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+            org.tensorflow.proto.util.CodeLocation, org.tensorflow.proto.util.CodeLocation.Builder, org.tensorflow.proto.util.CodeLocationOrBuilder>(
+                getCodeLocation(),
+                getParentForChildren(),
+                isClean());
+        codeLocation_ = null;
+      }
+      return codeLocationBuilder_;
+    }
+
+    private com.google.protobuf.Internal.IntList outputTensorIds_ = emptyIntList();
+    private void ensureOutputTensorIdsIsMutable() {
+      if (!((bitField0_ & 0x00000002) != 0)) {
+        outputTensorIds_ = mutableCopy(outputTensorIds_);
+        bitField0_ |= 0x00000002;
+       }
+    }
+    /**
+     * <pre>
+     * Unique IDs for the output tensors of this op.
+     * </pre>
+     *
+     * <code>repeated int32 output_tensor_ids = 9;</code>
+     */
+    public java.util.List<java.lang.Integer>
+        getOutputTensorIdsList() {
+      return ((bitField0_ & 0x00000002) != 0) ?
+               java.util.Collections.unmodifiableList(outputTensorIds_) : outputTensorIds_;
+    }
+    /**
+     * <pre>
+     * Unique IDs for the output tensors of this op.
+     * </pre>
+     *
+     * <code>repeated int32 output_tensor_ids = 9;</code>
+     */
+    public int getOutputTensorIdsCount() {
+      return outputTensorIds_.size();
+    }
+    /**
+     * <pre>
+     * Unique IDs for the output tensors of this op.
+     * </pre>
+     *
+     * <code>repeated int32 output_tensor_ids = 9;</code>
+     */
+    public int getOutputTensorIds(int index) {
+      return outputTensorIds_.getInt(index);
+    }
+    /**
+     * <pre>
+     * Unique IDs for the output tensors of this op.
+     * </pre>
+     *
+     * <code>repeated int32 output_tensor_ids = 9;</code>
+     */
+    public Builder setOutputTensorIds(
+        int index, int value) {
+      ensureOutputTensorIdsIsMutable();
+      outputTensorIds_.setInt(index, value);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Unique IDs for the output tensors of this op.
+     * </pre>
+     *
+     * <code>repeated int32 output_tensor_ids = 9;</code>
+     */
+    public Builder addOutputTensorIds(int value) {
+      ensureOutputTensorIdsIsMutable();
+      outputTensorIds_.addInt(value);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Unique IDs for the output tensors of this op.
+     * </pre>
+     *
+     * <code>repeated int32 output_tensor_ids = 9;</code>
+     */
+    public Builder addAllOutputTensorIds(
+        java.lang.Iterable<? extends java.lang.Integer> values) {
+      ensureOutputTensorIdsIsMutable();
+      com.google.protobuf.AbstractMessageLite.Builder.addAll(
+          values, outputTensorIds_);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Unique IDs for the output tensors of this op.
+     * </pre>
+     *
+     * <code>repeated int32 output_tensor_ids = 9;</code>
+     */
+    public Builder clearOutputTensorIds() {
+      outputTensorIds_ = emptyIntList();
+      bitField0_ = (bitField0_ & ~0x00000002);
+      onChanged();
+      return this;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.GraphOpCreation)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.GraphOpCreation)
+  private static final org.tensorflow.proto.util.GraphOpCreation DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.util.GraphOpCreation();
+  }
+
+  public static org.tensorflow.proto.util.GraphOpCreation getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<GraphOpCreation>
+      PARSER = new com.google.protobuf.AbstractParser<GraphOpCreation>() {
+    @java.lang.Override
+    public GraphOpCreation parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new GraphOpCreation(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<GraphOpCreation> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<GraphOpCreation> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.util.GraphOpCreation getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/GraphOpCreationOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/GraphOpCreationOrBuilder.java
new file mode 100644
index 00000000000..60abaf9cdfe
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/GraphOpCreationOrBuilder.java
@@ -0,0 +1,195 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/debug_event.proto
+
+package org.tensorflow.proto.util;
+
+public interface GraphOpCreationOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.GraphOpCreation)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <pre>
+   * Type of the op (e.g., "MatMul").
+   * </pre>
+   *
+   * <code>string op_type = 1;</code>
+   */
+  java.lang.String getOpType();
+  /**
+   * <pre>
+   * Type of the op (e.g., "MatMul").
+   * </pre>
+   *
+   * <code>string op_type = 1;</code>
+   */
+  com.google.protobuf.ByteString
+      getOpTypeBytes();
+
+  /**
+   * <pre>
+   * Name of the op (e.g., "Dense/MatMul_1").
+   * </pre>
+   *
+   * <code>string op_name = 2;</code>
+   */
+  java.lang.String getOpName();
+  /**
+   * <pre>
+   * Name of the op (e.g., "Dense/MatMul_1").
+   * </pre>
+   *
+   * <code>string op_name = 2;</code>
+   */
+  com.google.protobuf.ByteString
+      getOpNameBytes();
+
+  /**
+   * <pre>
+   * Name of the graph that the op is a part of (if available).
+   * </pre>
+   *
+   * <code>string graph_name = 3;</code>
+   */
+  java.lang.String getGraphName();
+  /**
+   * <pre>
+   * Name of the graph that the op is a part of (if available).
+   * </pre>
+   *
+   * <code>string graph_name = 3;</code>
+   */
+  com.google.protobuf.ByteString
+      getGraphNameBytes();
+
+  /**
+   * <pre>
+   * Unique ID of the graph (generated by debugger).
+   * This is the ID of the immediately-enclosing graph.
+   * </pre>
+   *
+   * <code>string graph_id = 4;</code>
+   */
+  java.lang.String getGraphId();
+  /**
+   * <pre>
+   * Unique ID of the graph (generated by debugger).
+   * This is the ID of the immediately-enclosing graph.
+   * </pre>
+   *
+   * <code>string graph_id = 4;</code>
+   */
+  com.google.protobuf.ByteString
+      getGraphIdBytes();
+
+  /**
+   * <pre>
+   * Name of the device that the op is assigned to (if available).
+   * </pre>
+   *
+   * <code>string device_name = 5;</code>
+   */
+  java.lang.String getDeviceName();
+  /**
+   * <pre>
+   * Name of the device that the op is assigned to (if available).
+   * </pre>
+   *
+   * <code>string device_name = 5;</code>
+   */
+  com.google.protobuf.ByteString
+      getDeviceNameBytes();
+
+  /**
+   * <pre>
+   * Names of the input tensors to the op.
+   * </pre>
+   *
+   * <code>repeated string input_names = 6;</code>
+   */
+  java.util.List<java.lang.String>
+      getInputNamesList();
+  /**
+   * <pre>
+   * Names of the input tensors to the op.
+   * </pre>
+   *
+   * <code>repeated string input_names = 6;</code>
+   */
+  int getInputNamesCount();
+  /**
+   * <pre>
+   * Names of the input tensors to the op.
+   * </pre>
+   *
+   * <code>repeated string input_names = 6;</code>
+   */
+  java.lang.String getInputNames(int index);
+  /**
+   * <pre>
+   * Names of the input tensors to the op.
+   * </pre>
+   *
+   * <code>repeated string input_names = 6;</code>
+   */
+  com.google.protobuf.ByteString
+      getInputNamesBytes(int index);
+
+  /**
+   * <pre>
+   * Number of output tensors emitted by the op.
+   * </pre>
+   *
+   * <code>int32 num_outputs = 7;</code>
+   */
+  int getNumOutputs();
+
+  /**
+   * <pre>
+   * The unique ID for code location (stack trace) of the op's creation.
+   * </pre>
+   *
+   * <code>.tensorflow.CodeLocation code_location = 8;</code>
+   */
+  boolean hasCodeLocation();
+  /**
+   * <pre>
+   * The unique ID for code location (stack trace) of the op's creation.
+   * </pre>
+   *
+   * <code>.tensorflow.CodeLocation code_location = 8;</code>
+   */
+  org.tensorflow.proto.util.CodeLocation getCodeLocation();
+  /**
+   * <pre>
+   * The unique ID for code location (stack trace) of the op's creation.
+   * </pre>
+   *
+   * <code>.tensorflow.CodeLocation code_location = 8;</code>
+   */
+  org.tensorflow.proto.util.CodeLocationOrBuilder getCodeLocationOrBuilder();
+
+  /**
+   * <pre>
+   * Unique IDs for the output tensors of this op.
+   * </pre>
+   *
+   * <code>repeated int32 output_tensor_ids = 9;</code>
+   */
+  java.util.List<java.lang.Integer> getOutputTensorIdsList();
+  /**
+   * <pre>
+   * Unique IDs for the output tensors of this op.
+   * </pre>
+   *
+   * <code>repeated int32 output_tensor_ids = 9;</code>
+   */
+  int getOutputTensorIdsCount();
+  /**
+   * <pre>
+   * Unique IDs for the output tensors of this op.
+   * </pre>
+   *
+   * <code>repeated int32 output_tensor_ids = 9;</code>
+   */
+  int getOutputTensorIds(int index);
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/LogMessage.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/LogMessage.java
new file mode 100644
index 00000000000..3a7dd55b4c8
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/LogMessage.java
@@ -0,0 +1,787 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/util/event.proto
+
+package org.tensorflow.proto.util;
+
+/**
+ * <pre>
+ * Protocol buffer used for logging messages to the events file.
+ * </pre>
+ *
+ * Protobuf type {@code tensorflow.LogMessage}
+ */
+public  final class LogMessage extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.LogMessage)
+    LogMessageOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use LogMessage.newBuilder() to construct.
+  private LogMessage(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private LogMessage() {
+    level_ = 0;
+    message_ = "";
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new LogMessage();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private LogMessage(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 8: {
+            int rawValue = input.readEnum();
+
+            level_ = rawValue;
+            break;
+          }
+          case 18: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            message_ = s;
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.util.EventProtos.internal_static_tensorflow_LogMessage_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.util.EventProtos.internal_static_tensorflow_LogMessage_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.util.LogMessage.class, org.tensorflow.proto.util.LogMessage.Builder.class);
+  }
+
+  /**
+   * Protobuf enum {@code tensorflow.LogMessage.Level}
+   */
+  public enum Level
+      implements com.google.protobuf.ProtocolMessageEnum {
+    /**
+     * <code>UNKNOWN = 0;</code>
+     */
+    UNKNOWN(0),
+    /**
+     * <pre>
+     * Note: The logging level 10 cannot be named DEBUG. Some software
+     * projects compile their C/C++ code with -DDEBUG in debug builds. So the
+     * C++ code generated from this file should not have an identifier named
+     * DEBUG.
+     * </pre>
+     *
+     * <code>DEBUGGING = 10;</code>
+     */
+    DEBUGGING(10),
+    /**
+     * <code>INFO = 20;</code>
+     */
+    INFO(20),
+    /**
+     * <code>WARN = 30;</code>
+     */
+    WARN(30),
+    /**
+     * <code>ERROR = 40;</code>
+     */
+    ERROR(40),
+    /**
+     * <code>FATAL = 50;</code>
+     */
+    FATAL(50),
+    UNRECOGNIZED(-1),
+    ;
+
+    /**
+     * <code>UNKNOWN = 0;</code>
+     */
+    public static final int UNKNOWN_VALUE = 0;
+    /**
+     * <pre>
+     * Note: The logging level 10 cannot be named DEBUG. Some software
+     * projects compile their C/C++ code with -DDEBUG in debug builds. So the
+     * C++ code generated from this file should not have an identifier named
+     * DEBUG.
+     * </pre>
+     *
+     * <code>DEBUGGING = 10;</code>
+     */
+    public static final int DEBUGGING_VALUE = 10;
+    /**
+     * <code>INFO = 20;</code>
+     */
+    public static final int INFO_VALUE = 20;
+    /**
+     * <code>WARN = 30;</code>
+     */
+    public static final int WARN_VALUE = 30;
+    /**
+     * <code>ERROR = 40;</code>
+     */
+    public static final int ERROR_VALUE = 40;
+    /**
+     * <code>FATAL = 50;</code>
+     */
+    public static final int FATAL_VALUE = 50;
+
+
+    public final int getNumber() {
+      if (this == UNRECOGNIZED) {
+        throw new java.lang.IllegalArgumentException(
+            "Can't get the number of an unknown enum value.");
+      }
+      return value;
+    }
+
+    /**
+     * @deprecated Use {@link #forNumber(int)} instead.
+     */
+    @java.lang.Deprecated
+    public static Level valueOf(int value) {
+      return forNumber(value);
+    }
+
+    public static Level forNumber(int value) {
+      switch (value) {
+        case 0: return UNKNOWN;
+        case 10: return DEBUGGING;
+        case 20: return INFO;
+        case 30: return WARN;
+        case 40: return ERROR;
+        case 50: return FATAL;
+        default: return null;
+      }
+    }
+
+    public static com.google.protobuf.Internal.EnumLiteMap<Level>
+        internalGetValueMap() {
+      return internalValueMap;
+    }
+    private static final com.google.protobuf.Internal.EnumLiteMap<
+        Level> internalValueMap =
+          new com.google.protobuf.Internal.EnumLiteMap<Level>() {
+            public Level findValueByNumber(int number) {
+              return Level.forNumber(number);
+            }
+          };
+
+    public final com.google.protobuf.Descriptors.EnumValueDescriptor
+        getValueDescriptor() {
+      return getDescriptor().getValues().get(ordinal());
+    }
+    public final com.google.protobuf.Descriptors.EnumDescriptor
+        getDescriptorForType() {
+      return getDescriptor();
+    }
+    public static final com.google.protobuf.Descriptors.EnumDescriptor
+        getDescriptor() {
+      return org.tensorflow.proto.util.LogMessage.getDescriptor().getEnumTypes().get(0);
+    }
+
+    private static final Level[] VALUES = values();
+
+    public static Level valueOf(
+        com.google.protobuf.Descriptors.EnumValueDescriptor desc) {
+      if (desc.getType() != getDescriptor()) {
+        throw new java.lang.IllegalArgumentException(
+          "EnumValueDescriptor is not for this type.");
+      }
+      if (desc.getIndex() == -1) {
+        return UNRECOGNIZED;
+      }
+      return VALUES[desc.getIndex()];
+    }
+
+    private final int value;
+
+    private Level(int value) {
+      this.value = value;
+    }
+
+    // @@protoc_insertion_point(enum_scope:tensorflow.LogMessage.Level)
+  }
+
+  public static final int LEVEL_FIELD_NUMBER = 1;
+  private int level_;
+  /**
+   * <code>.tensorflow.LogMessage.Level level = 1;</code>
+   */
+  public int getLevelValue() {
+    return level_;
+  }
+  /**
+   * <code>.tensorflow.LogMessage.Level level = 1;</code>
+   */
+  public org.tensorflow.proto.util.LogMessage.Level getLevel() {
+    @SuppressWarnings("deprecation")
+    org.tensorflow.proto.util.LogMessage.Level result = org.tensorflow.proto.util.LogMessage.Level.valueOf(level_);
+    return result == null ? org.tensorflow.proto.util.LogMessage.Level.UNRECOGNIZED : result;
+  }
+
+  public static final int MESSAGE_FIELD_NUMBER = 2;
+  private volatile java.lang.Object message_;
+  /**
+   * <code>string message = 2;</code>
+   */
+  public java.lang.String getMessage() {
+    java.lang.Object ref = message_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      message_ = s;
+      return s;
+    }
+  }
+  /**
+   * <code>string message = 2;</code>
+   */
+  public com.google.protobuf.ByteString
+      getMessageBytes() {
+    java.lang.Object ref = message_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      message_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    if (level_ != org.tensorflow.proto.util.LogMessage.Level.UNKNOWN.getNumber()) {
+      output.writeEnum(1, level_);
+    }
+    if (!getMessageBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 2, message_);
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (level_ != org.tensorflow.proto.util.LogMessage.Level.UNKNOWN.getNumber()) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeEnumSize(1, level_);
+    }
+    if (!getMessageBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(2, message_);
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.util.LogMessage)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.util.LogMessage other = (org.tensorflow.proto.util.LogMessage) obj;
+
+    if (level_ != other.level_) return false;
+    if (!getMessage()
+        .equals(other.getMessage())) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    hash = (37 * hash) + LEVEL_FIELD_NUMBER;
+    hash = (53 * hash) + level_;
+    hash = (37 * hash) + MESSAGE_FIELD_NUMBER;
+    hash = (53 * hash) + getMessage().hashCode();
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.util.LogMessage parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.LogMessage parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.LogMessage parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.LogMessage parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.LogMessage parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.LogMessage parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.LogMessage parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.LogMessage parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.LogMessage parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.LogMessage parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.LogMessage parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.LogMessage parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.util.LogMessage prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * <pre>
+   * Protocol buffer used for logging messages to the events file.
+   * </pre>
+   *
+   * Protobuf type {@code tensorflow.LogMessage}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.LogMessage)
+      org.tensorflow.proto.util.LogMessageOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.util.EventProtos.internal_static_tensorflow_LogMessage_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.util.EventProtos.internal_static_tensorflow_LogMessage_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.util.LogMessage.class, org.tensorflow.proto.util.LogMessage.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.util.LogMessage.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      level_ = 0;
+
+      message_ = "";
+
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.util.EventProtos.internal_static_tensorflow_LogMessage_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.LogMessage getDefaultInstanceForType() {
+      return org.tensorflow.proto.util.LogMessage.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.LogMessage build() {
+      org.tensorflow.proto.util.LogMessage result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.LogMessage buildPartial() {
+      org.tensorflow.proto.util.LogMessage result = new org.tensorflow.proto.util.LogMessage(this);
+      result.level_ = level_;
+      result.message_ = message_;
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.util.LogMessage) {
+        return mergeFrom((org.tensorflow.proto.util.LogMessage)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.util.LogMessage other) {
+      if (other == org.tensorflow.proto.util.LogMessage.getDefaultInstance()) return this;
+      if (other.level_ != 0) {
+        setLevelValue(other.getLevelValue());
+      }
+      if (!other.getMessage().isEmpty()) {
+        message_ = other.message_;
+        onChanged();
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.util.LogMessage parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.util.LogMessage) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+
+    private int level_ = 0;
+    /**
+     * <code>.tensorflow.LogMessage.Level level = 1;</code>
+     */
+    public int getLevelValue() {
+      return level_;
+    }
+    /**
+     * <code>.tensorflow.LogMessage.Level level = 1;</code>
+     */
+    public Builder setLevelValue(int value) {
+      level_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>.tensorflow.LogMessage.Level level = 1;</code>
+     */
+    public org.tensorflow.proto.util.LogMessage.Level getLevel() {
+      @SuppressWarnings("deprecation")
+      org.tensorflow.proto.util.LogMessage.Level result = org.tensorflow.proto.util.LogMessage.Level.valueOf(level_);
+      return result == null ? org.tensorflow.proto.util.LogMessage.Level.UNRECOGNIZED : result;
+    }
+    /**
+     * <code>.tensorflow.LogMessage.Level level = 1;</code>
+     */
+    public Builder setLevel(org.tensorflow.proto.util.LogMessage.Level value) {
+      if (value == null) {
+        throw new NullPointerException();
+      }
+      
+      level_ = value.getNumber();
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>.tensorflow.LogMessage.Level level = 1;</code>
+     */
+    public Builder clearLevel() {
+      
+      level_ = 0;
+      onChanged();
+      return this;
+    }
+
+    private java.lang.Object message_ = "";
+    /**
+     * <code>string message = 2;</code>
+     */
+    public java.lang.String getMessage() {
+      java.lang.Object ref = message_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        message_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <code>string message = 2;</code>
+     */
+    public com.google.protobuf.ByteString
+        getMessageBytes() {
+      java.lang.Object ref = message_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        message_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <code>string message = 2;</code>
+     */
+    public Builder setMessage(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      message_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>string message = 2;</code>
+     */
+    public Builder clearMessage() {
+      
+      message_ = getDefaultInstance().getMessage();
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>string message = 2;</code>
+     */
+    public Builder setMessageBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      message_ = value;
+      onChanged();
+      return this;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.LogMessage)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.LogMessage)
+  private static final org.tensorflow.proto.util.LogMessage DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.util.LogMessage();
+  }
+
+  public static org.tensorflow.proto.util.LogMessage getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<LogMessage>
+      PARSER = new com.google.protobuf.AbstractParser<LogMessage>() {
+    @java.lang.Override
+    public LogMessage parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new LogMessage(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<LogMessage> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<LogMessage> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.util.LogMessage getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/LogMessageOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/LogMessageOrBuilder.java
new file mode 100644
index 00000000000..a19c21eebdf
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/LogMessageOrBuilder.java
@@ -0,0 +1,28 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/util/event.proto
+
+package org.tensorflow.proto.util;
+
+public interface LogMessageOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.LogMessage)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <code>.tensorflow.LogMessage.Level level = 1;</code>
+   */
+  int getLevelValue();
+  /**
+   * <code>.tensorflow.LogMessage.Level level = 1;</code>
+   */
+  org.tensorflow.proto.util.LogMessage.Level getLevel();
+
+  /**
+   * <code>string message = 2;</code>
+   */
+  java.lang.String getMessage();
+  /**
+   * <code>string message = 2;</code>
+   */
+  com.google.protobuf.ByteString
+      getMessageBytes();
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/MemAllocatorStats.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/MemAllocatorStats.java
new file mode 100644
index 00000000000..78a84b41b9c
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/MemAllocatorStats.java
@@ -0,0 +1,718 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/bfc_memory_map.proto
+
+package org.tensorflow.proto.util;
+
+/**
+ * <pre>
+ * Some of the data from AllocatorStats
+ * </pre>
+ *
+ * Protobuf type {@code tensorflow.MemAllocatorStats}
+ */
+public  final class MemAllocatorStats extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.MemAllocatorStats)
+    MemAllocatorStatsOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use MemAllocatorStats.newBuilder() to construct.
+  private MemAllocatorStats(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private MemAllocatorStats() {
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new MemAllocatorStats();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private MemAllocatorStats(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 8: {
+
+            numAllocs_ = input.readInt64();
+            break;
+          }
+          case 16: {
+
+            bytesInUse_ = input.readInt64();
+            break;
+          }
+          case 24: {
+
+            peakBytesInUse_ = input.readInt64();
+            break;
+          }
+          case 32: {
+
+            largestAllocSize_ = input.readInt64();
+            break;
+          }
+          case 45: {
+
+            fragmentationMetric_ = input.readFloat();
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.util.BfcMemoryMapProtos.internal_static_tensorflow_MemAllocatorStats_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.util.BfcMemoryMapProtos.internal_static_tensorflow_MemAllocatorStats_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.util.MemAllocatorStats.class, org.tensorflow.proto.util.MemAllocatorStats.Builder.class);
+  }
+
+  public static final int NUM_ALLOCS_FIELD_NUMBER = 1;
+  private long numAllocs_;
+  /**
+   * <code>int64 num_allocs = 1;</code>
+   */
+  public long getNumAllocs() {
+    return numAllocs_;
+  }
+
+  public static final int BYTES_IN_USE_FIELD_NUMBER = 2;
+  private long bytesInUse_;
+  /**
+   * <code>int64 bytes_in_use = 2;</code>
+   */
+  public long getBytesInUse() {
+    return bytesInUse_;
+  }
+
+  public static final int PEAK_BYTES_IN_USE_FIELD_NUMBER = 3;
+  private long peakBytesInUse_;
+  /**
+   * <code>int64 peak_bytes_in_use = 3;</code>
+   */
+  public long getPeakBytesInUse() {
+    return peakBytesInUse_;
+  }
+
+  public static final int LARGEST_ALLOC_SIZE_FIELD_NUMBER = 4;
+  private long largestAllocSize_;
+  /**
+   * <code>int64 largest_alloc_size = 4;</code>
+   */
+  public long getLargestAllocSize() {
+    return largestAllocSize_;
+  }
+
+  public static final int FRAGMENTATION_METRIC_FIELD_NUMBER = 5;
+  private float fragmentationMetric_;
+  /**
+   * <code>float fragmentation_metric = 5;</code>
+   */
+  public float getFragmentationMetric() {
+    return fragmentationMetric_;
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    if (numAllocs_ != 0L) {
+      output.writeInt64(1, numAllocs_);
+    }
+    if (bytesInUse_ != 0L) {
+      output.writeInt64(2, bytesInUse_);
+    }
+    if (peakBytesInUse_ != 0L) {
+      output.writeInt64(3, peakBytesInUse_);
+    }
+    if (largestAllocSize_ != 0L) {
+      output.writeInt64(4, largestAllocSize_);
+    }
+    if (fragmentationMetric_ != 0F) {
+      output.writeFloat(5, fragmentationMetric_);
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (numAllocs_ != 0L) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt64Size(1, numAllocs_);
+    }
+    if (bytesInUse_ != 0L) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt64Size(2, bytesInUse_);
+    }
+    if (peakBytesInUse_ != 0L) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt64Size(3, peakBytesInUse_);
+    }
+    if (largestAllocSize_ != 0L) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt64Size(4, largestAllocSize_);
+    }
+    if (fragmentationMetric_ != 0F) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeFloatSize(5, fragmentationMetric_);
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.util.MemAllocatorStats)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.util.MemAllocatorStats other = (org.tensorflow.proto.util.MemAllocatorStats) obj;
+
+    if (getNumAllocs()
+        != other.getNumAllocs()) return false;
+    if (getBytesInUse()
+        != other.getBytesInUse()) return false;
+    if (getPeakBytesInUse()
+        != other.getPeakBytesInUse()) return false;
+    if (getLargestAllocSize()
+        != other.getLargestAllocSize()) return false;
+    if (java.lang.Float.floatToIntBits(getFragmentationMetric())
+        != java.lang.Float.floatToIntBits(
+            other.getFragmentationMetric())) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    hash = (37 * hash) + NUM_ALLOCS_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+        getNumAllocs());
+    hash = (37 * hash) + BYTES_IN_USE_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+        getBytesInUse());
+    hash = (37 * hash) + PEAK_BYTES_IN_USE_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+        getPeakBytesInUse());
+    hash = (37 * hash) + LARGEST_ALLOC_SIZE_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+        getLargestAllocSize());
+    hash = (37 * hash) + FRAGMENTATION_METRIC_FIELD_NUMBER;
+    hash = (53 * hash) + java.lang.Float.floatToIntBits(
+        getFragmentationMetric());
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.util.MemAllocatorStats parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.MemAllocatorStats parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.MemAllocatorStats parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.MemAllocatorStats parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.MemAllocatorStats parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.MemAllocatorStats parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.MemAllocatorStats parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.MemAllocatorStats parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.MemAllocatorStats parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.MemAllocatorStats parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.MemAllocatorStats parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.MemAllocatorStats parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.util.MemAllocatorStats prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * <pre>
+   * Some of the data from AllocatorStats
+   * </pre>
+   *
+   * Protobuf type {@code tensorflow.MemAllocatorStats}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.MemAllocatorStats)
+      org.tensorflow.proto.util.MemAllocatorStatsOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.util.BfcMemoryMapProtos.internal_static_tensorflow_MemAllocatorStats_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.util.BfcMemoryMapProtos.internal_static_tensorflow_MemAllocatorStats_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.util.MemAllocatorStats.class, org.tensorflow.proto.util.MemAllocatorStats.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.util.MemAllocatorStats.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      numAllocs_ = 0L;
+
+      bytesInUse_ = 0L;
+
+      peakBytesInUse_ = 0L;
+
+      largestAllocSize_ = 0L;
+
+      fragmentationMetric_ = 0F;
+
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.util.BfcMemoryMapProtos.internal_static_tensorflow_MemAllocatorStats_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.MemAllocatorStats getDefaultInstanceForType() {
+      return org.tensorflow.proto.util.MemAllocatorStats.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.MemAllocatorStats build() {
+      org.tensorflow.proto.util.MemAllocatorStats result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.MemAllocatorStats buildPartial() {
+      org.tensorflow.proto.util.MemAllocatorStats result = new org.tensorflow.proto.util.MemAllocatorStats(this);
+      result.numAllocs_ = numAllocs_;
+      result.bytesInUse_ = bytesInUse_;
+      result.peakBytesInUse_ = peakBytesInUse_;
+      result.largestAllocSize_ = largestAllocSize_;
+      result.fragmentationMetric_ = fragmentationMetric_;
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.util.MemAllocatorStats) {
+        return mergeFrom((org.tensorflow.proto.util.MemAllocatorStats)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.util.MemAllocatorStats other) {
+      if (other == org.tensorflow.proto.util.MemAllocatorStats.getDefaultInstance()) return this;
+      if (other.getNumAllocs() != 0L) {
+        setNumAllocs(other.getNumAllocs());
+      }
+      if (other.getBytesInUse() != 0L) {
+        setBytesInUse(other.getBytesInUse());
+      }
+      if (other.getPeakBytesInUse() != 0L) {
+        setPeakBytesInUse(other.getPeakBytesInUse());
+      }
+      if (other.getLargestAllocSize() != 0L) {
+        setLargestAllocSize(other.getLargestAllocSize());
+      }
+      if (other.getFragmentationMetric() != 0F) {
+        setFragmentationMetric(other.getFragmentationMetric());
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.util.MemAllocatorStats parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.util.MemAllocatorStats) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+
+    private long numAllocs_ ;
+    /**
+     * <code>int64 num_allocs = 1;</code>
+     */
+    public long getNumAllocs() {
+      return numAllocs_;
+    }
+    /**
+     * <code>int64 num_allocs = 1;</code>
+     */
+    public Builder setNumAllocs(long value) {
+      
+      numAllocs_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>int64 num_allocs = 1;</code>
+     */
+    public Builder clearNumAllocs() {
+      
+      numAllocs_ = 0L;
+      onChanged();
+      return this;
+    }
+
+    private long bytesInUse_ ;
+    /**
+     * <code>int64 bytes_in_use = 2;</code>
+     */
+    public long getBytesInUse() {
+      return bytesInUse_;
+    }
+    /**
+     * <code>int64 bytes_in_use = 2;</code>
+     */
+    public Builder setBytesInUse(long value) {
+      
+      bytesInUse_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>int64 bytes_in_use = 2;</code>
+     */
+    public Builder clearBytesInUse() {
+      
+      bytesInUse_ = 0L;
+      onChanged();
+      return this;
+    }
+
+    private long peakBytesInUse_ ;
+    /**
+     * <code>int64 peak_bytes_in_use = 3;</code>
+     */
+    public long getPeakBytesInUse() {
+      return peakBytesInUse_;
+    }
+    /**
+     * <code>int64 peak_bytes_in_use = 3;</code>
+     */
+    public Builder setPeakBytesInUse(long value) {
+      
+      peakBytesInUse_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>int64 peak_bytes_in_use = 3;</code>
+     */
+    public Builder clearPeakBytesInUse() {
+      
+      peakBytesInUse_ = 0L;
+      onChanged();
+      return this;
+    }
+
+    private long largestAllocSize_ ;
+    /**
+     * <code>int64 largest_alloc_size = 4;</code>
+     */
+    public long getLargestAllocSize() {
+      return largestAllocSize_;
+    }
+    /**
+     * <code>int64 largest_alloc_size = 4;</code>
+     */
+    public Builder setLargestAllocSize(long value) {
+      
+      largestAllocSize_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>int64 largest_alloc_size = 4;</code>
+     */
+    public Builder clearLargestAllocSize() {
+      
+      largestAllocSize_ = 0L;
+      onChanged();
+      return this;
+    }
+
+    private float fragmentationMetric_ ;
+    /**
+     * <code>float fragmentation_metric = 5;</code>
+     */
+    public float getFragmentationMetric() {
+      return fragmentationMetric_;
+    }
+    /**
+     * <code>float fragmentation_metric = 5;</code>
+     */
+    public Builder setFragmentationMetric(float value) {
+      
+      fragmentationMetric_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>float fragmentation_metric = 5;</code>
+     */
+    public Builder clearFragmentationMetric() {
+      
+      fragmentationMetric_ = 0F;
+      onChanged();
+      return this;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.MemAllocatorStats)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.MemAllocatorStats)
+  private static final org.tensorflow.proto.util.MemAllocatorStats DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.util.MemAllocatorStats();
+  }
+
+  public static org.tensorflow.proto.util.MemAllocatorStats getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<MemAllocatorStats>
+      PARSER = new com.google.protobuf.AbstractParser<MemAllocatorStats>() {
+    @java.lang.Override
+    public MemAllocatorStats parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new MemAllocatorStats(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<MemAllocatorStats> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<MemAllocatorStats> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.util.MemAllocatorStats getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/MemAllocatorStatsOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/MemAllocatorStatsOrBuilder.java
new file mode 100644
index 00000000000..119204effcb
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/MemAllocatorStatsOrBuilder.java
@@ -0,0 +1,34 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/bfc_memory_map.proto
+
+package org.tensorflow.proto.util;
+
+public interface MemAllocatorStatsOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.MemAllocatorStats)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <code>int64 num_allocs = 1;</code>
+   */
+  long getNumAllocs();
+
+  /**
+   * <code>int64 bytes_in_use = 2;</code>
+   */
+  long getBytesInUse();
+
+  /**
+   * <code>int64 peak_bytes_in_use = 3;</code>
+   */
+  long getPeakBytesInUse();
+
+  /**
+   * <code>int64 largest_alloc_size = 4;</code>
+   */
+  long getLargestAllocSize();
+
+  /**
+   * <code>float fragmentation_metric = 5;</code>
+   */
+  float getFragmentationMetric();
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/MemChunk.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/MemChunk.java
new file mode 100644
index 00000000000..3212e0b9c85
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/MemChunk.java
@@ -0,0 +1,1009 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/bfc_memory_map.proto
+
+package org.tensorflow.proto.util;
+
+/**
+ * Protobuf type {@code tensorflow.MemChunk}
+ */
+public  final class MemChunk extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.MemChunk)
+    MemChunkOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use MemChunk.newBuilder() to construct.
+  private MemChunk(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private MemChunk() {
+    opName_ = "";
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new MemChunk();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private MemChunk(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 8: {
+
+            address_ = input.readUInt64();
+            break;
+          }
+          case 16: {
+
+            size_ = input.readInt64();
+            break;
+          }
+          case 24: {
+
+            requestedSize_ = input.readInt64();
+            break;
+          }
+          case 32: {
+
+            bin_ = input.readInt32();
+            break;
+          }
+          case 42: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            opName_ = s;
+            break;
+          }
+          case 48: {
+
+            freedAtCount_ = input.readUInt64();
+            break;
+          }
+          case 56: {
+
+            actionCount_ = input.readUInt64();
+            break;
+          }
+          case 64: {
+
+            inUse_ = input.readBool();
+            break;
+          }
+          case 72: {
+
+            stepId_ = input.readUInt64();
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.util.BfcMemoryMapProtos.internal_static_tensorflow_MemChunk_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.util.BfcMemoryMapProtos.internal_static_tensorflow_MemChunk_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.util.MemChunk.class, org.tensorflow.proto.util.MemChunk.Builder.class);
+  }
+
+  public static final int ADDRESS_FIELD_NUMBER = 1;
+  private long address_;
+  /**
+   * <code>uint64 address = 1;</code>
+   */
+  public long getAddress() {
+    return address_;
+  }
+
+  public static final int SIZE_FIELD_NUMBER = 2;
+  private long size_;
+  /**
+   * <code>int64 size = 2;</code>
+   */
+  public long getSize() {
+    return size_;
+  }
+
+  public static final int REQUESTED_SIZE_FIELD_NUMBER = 3;
+  private long requestedSize_;
+  /**
+   * <code>int64 requested_size = 3;</code>
+   */
+  public long getRequestedSize() {
+    return requestedSize_;
+  }
+
+  public static final int BIN_FIELD_NUMBER = 4;
+  private int bin_;
+  /**
+   * <code>int32 bin = 4;</code>
+   */
+  public int getBin() {
+    return bin_;
+  }
+
+  public static final int OP_NAME_FIELD_NUMBER = 5;
+  private volatile java.lang.Object opName_;
+  /**
+   * <code>string op_name = 5;</code>
+   */
+  public java.lang.String getOpName() {
+    java.lang.Object ref = opName_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      opName_ = s;
+      return s;
+    }
+  }
+  /**
+   * <code>string op_name = 5;</code>
+   */
+  public com.google.protobuf.ByteString
+      getOpNameBytes() {
+    java.lang.Object ref = opName_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      opName_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int FREED_AT_COUNT_FIELD_NUMBER = 6;
+  private long freedAtCount_;
+  /**
+   * <code>uint64 freed_at_count = 6;</code>
+   */
+  public long getFreedAtCount() {
+    return freedAtCount_;
+  }
+
+  public static final int ACTION_COUNT_FIELD_NUMBER = 7;
+  private long actionCount_;
+  /**
+   * <code>uint64 action_count = 7;</code>
+   */
+  public long getActionCount() {
+    return actionCount_;
+  }
+
+  public static final int IN_USE_FIELD_NUMBER = 8;
+  private boolean inUse_;
+  /**
+   * <code>bool in_use = 8;</code>
+   */
+  public boolean getInUse() {
+    return inUse_;
+  }
+
+  public static final int STEP_ID_FIELD_NUMBER = 9;
+  private long stepId_;
+  /**
+   * <code>uint64 step_id = 9;</code>
+   */
+  public long getStepId() {
+    return stepId_;
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    if (address_ != 0L) {
+      output.writeUInt64(1, address_);
+    }
+    if (size_ != 0L) {
+      output.writeInt64(2, size_);
+    }
+    if (requestedSize_ != 0L) {
+      output.writeInt64(3, requestedSize_);
+    }
+    if (bin_ != 0) {
+      output.writeInt32(4, bin_);
+    }
+    if (!getOpNameBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 5, opName_);
+    }
+    if (freedAtCount_ != 0L) {
+      output.writeUInt64(6, freedAtCount_);
+    }
+    if (actionCount_ != 0L) {
+      output.writeUInt64(7, actionCount_);
+    }
+    if (inUse_ != false) {
+      output.writeBool(8, inUse_);
+    }
+    if (stepId_ != 0L) {
+      output.writeUInt64(9, stepId_);
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (address_ != 0L) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeUInt64Size(1, address_);
+    }
+    if (size_ != 0L) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt64Size(2, size_);
+    }
+    if (requestedSize_ != 0L) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt64Size(3, requestedSize_);
+    }
+    if (bin_ != 0) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt32Size(4, bin_);
+    }
+    if (!getOpNameBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(5, opName_);
+    }
+    if (freedAtCount_ != 0L) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeUInt64Size(6, freedAtCount_);
+    }
+    if (actionCount_ != 0L) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeUInt64Size(7, actionCount_);
+    }
+    if (inUse_ != false) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeBoolSize(8, inUse_);
+    }
+    if (stepId_ != 0L) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeUInt64Size(9, stepId_);
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.util.MemChunk)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.util.MemChunk other = (org.tensorflow.proto.util.MemChunk) obj;
+
+    if (getAddress()
+        != other.getAddress()) return false;
+    if (getSize()
+        != other.getSize()) return false;
+    if (getRequestedSize()
+        != other.getRequestedSize()) return false;
+    if (getBin()
+        != other.getBin()) return false;
+    if (!getOpName()
+        .equals(other.getOpName())) return false;
+    if (getFreedAtCount()
+        != other.getFreedAtCount()) return false;
+    if (getActionCount()
+        != other.getActionCount()) return false;
+    if (getInUse()
+        != other.getInUse()) return false;
+    if (getStepId()
+        != other.getStepId()) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    hash = (37 * hash) + ADDRESS_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+        getAddress());
+    hash = (37 * hash) + SIZE_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+        getSize());
+    hash = (37 * hash) + REQUESTED_SIZE_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+        getRequestedSize());
+    hash = (37 * hash) + BIN_FIELD_NUMBER;
+    hash = (53 * hash) + getBin();
+    hash = (37 * hash) + OP_NAME_FIELD_NUMBER;
+    hash = (53 * hash) + getOpName().hashCode();
+    hash = (37 * hash) + FREED_AT_COUNT_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+        getFreedAtCount());
+    hash = (37 * hash) + ACTION_COUNT_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+        getActionCount());
+    hash = (37 * hash) + IN_USE_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashBoolean(
+        getInUse());
+    hash = (37 * hash) + STEP_ID_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+        getStepId());
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.util.MemChunk parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.MemChunk parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.MemChunk parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.MemChunk parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.MemChunk parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.MemChunk parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.MemChunk parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.MemChunk parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.MemChunk parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.MemChunk parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.MemChunk parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.MemChunk parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.util.MemChunk prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * Protobuf type {@code tensorflow.MemChunk}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.MemChunk)
+      org.tensorflow.proto.util.MemChunkOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.util.BfcMemoryMapProtos.internal_static_tensorflow_MemChunk_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.util.BfcMemoryMapProtos.internal_static_tensorflow_MemChunk_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.util.MemChunk.class, org.tensorflow.proto.util.MemChunk.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.util.MemChunk.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      address_ = 0L;
+
+      size_ = 0L;
+
+      requestedSize_ = 0L;
+
+      bin_ = 0;
+
+      opName_ = "";
+
+      freedAtCount_ = 0L;
+
+      actionCount_ = 0L;
+
+      inUse_ = false;
+
+      stepId_ = 0L;
+
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.util.BfcMemoryMapProtos.internal_static_tensorflow_MemChunk_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.MemChunk getDefaultInstanceForType() {
+      return org.tensorflow.proto.util.MemChunk.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.MemChunk build() {
+      org.tensorflow.proto.util.MemChunk result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.MemChunk buildPartial() {
+      org.tensorflow.proto.util.MemChunk result = new org.tensorflow.proto.util.MemChunk(this);
+      result.address_ = address_;
+      result.size_ = size_;
+      result.requestedSize_ = requestedSize_;
+      result.bin_ = bin_;
+      result.opName_ = opName_;
+      result.freedAtCount_ = freedAtCount_;
+      result.actionCount_ = actionCount_;
+      result.inUse_ = inUse_;
+      result.stepId_ = stepId_;
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.util.MemChunk) {
+        return mergeFrom((org.tensorflow.proto.util.MemChunk)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.util.MemChunk other) {
+      if (other == org.tensorflow.proto.util.MemChunk.getDefaultInstance()) return this;
+      if (other.getAddress() != 0L) {
+        setAddress(other.getAddress());
+      }
+      if (other.getSize() != 0L) {
+        setSize(other.getSize());
+      }
+      if (other.getRequestedSize() != 0L) {
+        setRequestedSize(other.getRequestedSize());
+      }
+      if (other.getBin() != 0) {
+        setBin(other.getBin());
+      }
+      if (!other.getOpName().isEmpty()) {
+        opName_ = other.opName_;
+        onChanged();
+      }
+      if (other.getFreedAtCount() != 0L) {
+        setFreedAtCount(other.getFreedAtCount());
+      }
+      if (other.getActionCount() != 0L) {
+        setActionCount(other.getActionCount());
+      }
+      if (other.getInUse() != false) {
+        setInUse(other.getInUse());
+      }
+      if (other.getStepId() != 0L) {
+        setStepId(other.getStepId());
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.util.MemChunk parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.util.MemChunk) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+
+    private long address_ ;
+    /**
+     * <code>uint64 address = 1;</code>
+     */
+    public long getAddress() {
+      return address_;
+    }
+    /**
+     * <code>uint64 address = 1;</code>
+     */
+    public Builder setAddress(long value) {
+      
+      address_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>uint64 address = 1;</code>
+     */
+    public Builder clearAddress() {
+      
+      address_ = 0L;
+      onChanged();
+      return this;
+    }
+
+    private long size_ ;
+    /**
+     * <code>int64 size = 2;</code>
+     */
+    public long getSize() {
+      return size_;
+    }
+    /**
+     * <code>int64 size = 2;</code>
+     */
+    public Builder setSize(long value) {
+      
+      size_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>int64 size = 2;</code>
+     */
+    public Builder clearSize() {
+      
+      size_ = 0L;
+      onChanged();
+      return this;
+    }
+
+    private long requestedSize_ ;
+    /**
+     * <code>int64 requested_size = 3;</code>
+     */
+    public long getRequestedSize() {
+      return requestedSize_;
+    }
+    /**
+     * <code>int64 requested_size = 3;</code>
+     */
+    public Builder setRequestedSize(long value) {
+      
+      requestedSize_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>int64 requested_size = 3;</code>
+     */
+    public Builder clearRequestedSize() {
+      
+      requestedSize_ = 0L;
+      onChanged();
+      return this;
+    }
+
+    private int bin_ ;
+    /**
+     * <code>int32 bin = 4;</code>
+     */
+    public int getBin() {
+      return bin_;
+    }
+    /**
+     * <code>int32 bin = 4;</code>
+     */
+    public Builder setBin(int value) {
+      
+      bin_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>int32 bin = 4;</code>
+     */
+    public Builder clearBin() {
+      
+      bin_ = 0;
+      onChanged();
+      return this;
+    }
+
+    private java.lang.Object opName_ = "";
+    /**
+     * <code>string op_name = 5;</code>
+     */
+    public java.lang.String getOpName() {
+      java.lang.Object ref = opName_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        opName_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <code>string op_name = 5;</code>
+     */
+    public com.google.protobuf.ByteString
+        getOpNameBytes() {
+      java.lang.Object ref = opName_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        opName_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <code>string op_name = 5;</code>
+     */
+    public Builder setOpName(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      opName_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>string op_name = 5;</code>
+     */
+    public Builder clearOpName() {
+      
+      opName_ = getDefaultInstance().getOpName();
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>string op_name = 5;</code>
+     */
+    public Builder setOpNameBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      opName_ = value;
+      onChanged();
+      return this;
+    }
+
+    private long freedAtCount_ ;
+    /**
+     * <code>uint64 freed_at_count = 6;</code>
+     */
+    public long getFreedAtCount() {
+      return freedAtCount_;
+    }
+    /**
+     * <code>uint64 freed_at_count = 6;</code>
+     */
+    public Builder setFreedAtCount(long value) {
+      
+      freedAtCount_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>uint64 freed_at_count = 6;</code>
+     */
+    public Builder clearFreedAtCount() {
+      
+      freedAtCount_ = 0L;
+      onChanged();
+      return this;
+    }
+
+    private long actionCount_ ;
+    /**
+     * <code>uint64 action_count = 7;</code>
+     */
+    public long getActionCount() {
+      return actionCount_;
+    }
+    /**
+     * <code>uint64 action_count = 7;</code>
+     */
+    public Builder setActionCount(long value) {
+      
+      actionCount_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>uint64 action_count = 7;</code>
+     */
+    public Builder clearActionCount() {
+      
+      actionCount_ = 0L;
+      onChanged();
+      return this;
+    }
+
+    private boolean inUse_ ;
+    /**
+     * <code>bool in_use = 8;</code>
+     */
+    public boolean getInUse() {
+      return inUse_;
+    }
+    /**
+     * <code>bool in_use = 8;</code>
+     */
+    public Builder setInUse(boolean value) {
+      
+      inUse_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>bool in_use = 8;</code>
+     */
+    public Builder clearInUse() {
+      
+      inUse_ = false;
+      onChanged();
+      return this;
+    }
+
+    private long stepId_ ;
+    /**
+     * <code>uint64 step_id = 9;</code>
+     */
+    public long getStepId() {
+      return stepId_;
+    }
+    /**
+     * <code>uint64 step_id = 9;</code>
+     */
+    public Builder setStepId(long value) {
+      
+      stepId_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>uint64 step_id = 9;</code>
+     */
+    public Builder clearStepId() {
+      
+      stepId_ = 0L;
+      onChanged();
+      return this;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.MemChunk)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.MemChunk)
+  private static final org.tensorflow.proto.util.MemChunk DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.util.MemChunk();
+  }
+
+  public static org.tensorflow.proto.util.MemChunk getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<MemChunk>
+      PARSER = new com.google.protobuf.AbstractParser<MemChunk>() {
+    @java.lang.Override
+    public MemChunk parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new MemChunk(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<MemChunk> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<MemChunk> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.util.MemChunk getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/MemChunkOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/MemChunkOrBuilder.java
new file mode 100644
index 00000000000..6dbce3e91ab
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/MemChunkOrBuilder.java
@@ -0,0 +1,59 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/bfc_memory_map.proto
+
+package org.tensorflow.proto.util;
+
+public interface MemChunkOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.MemChunk)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <code>uint64 address = 1;</code>
+   */
+  long getAddress();
+
+  /**
+   * <code>int64 size = 2;</code>
+   */
+  long getSize();
+
+  /**
+   * <code>int64 requested_size = 3;</code>
+   */
+  long getRequestedSize();
+
+  /**
+   * <code>int32 bin = 4;</code>
+   */
+  int getBin();
+
+  /**
+   * <code>string op_name = 5;</code>
+   */
+  java.lang.String getOpName();
+  /**
+   * <code>string op_name = 5;</code>
+   */
+  com.google.protobuf.ByteString
+      getOpNameBytes();
+
+  /**
+   * <code>uint64 freed_at_count = 6;</code>
+   */
+  long getFreedAtCount();
+
+  /**
+   * <code>uint64 action_count = 7;</code>
+   */
+  long getActionCount();
+
+  /**
+   * <code>bool in_use = 8;</code>
+   */
+  boolean getInUse();
+
+  /**
+   * <code>uint64 step_id = 9;</code>
+   */
+  long getStepId();
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/MemmappedFileSystemDirectory.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/MemmappedFileSystemDirectory.java
new file mode 100644
index 00000000000..0f7f977819f
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/MemmappedFileSystemDirectory.java
@@ -0,0 +1,773 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/util/memmapped_file_system.proto
+
+package org.tensorflow.proto.util;
+
+/**
+ * <pre>
+ * A directory of regions in a memmapped file.
+ * </pre>
+ *
+ * Protobuf type {@code tensorflow.MemmappedFileSystemDirectory}
+ */
+public  final class MemmappedFileSystemDirectory extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.MemmappedFileSystemDirectory)
+    MemmappedFileSystemDirectoryOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use MemmappedFileSystemDirectory.newBuilder() to construct.
+  private MemmappedFileSystemDirectory(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private MemmappedFileSystemDirectory() {
+    element_ = java.util.Collections.emptyList();
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new MemmappedFileSystemDirectory();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private MemmappedFileSystemDirectory(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    int mutable_bitField0_ = 0;
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 10: {
+            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
+              element_ = new java.util.ArrayList<org.tensorflow.proto.util.MemmappedFileSystemDirectoryElement>();
+              mutable_bitField0_ |= 0x00000001;
+            }
+            element_.add(
+                input.readMessage(org.tensorflow.proto.util.MemmappedFileSystemDirectoryElement.parser(), extensionRegistry));
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      if (((mutable_bitField0_ & 0x00000001) != 0)) {
+        element_ = java.util.Collections.unmodifiableList(element_);
+      }
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.util.MemmappedFileSystemProtos.internal_static_tensorflow_MemmappedFileSystemDirectory_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.util.MemmappedFileSystemProtos.internal_static_tensorflow_MemmappedFileSystemDirectory_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.util.MemmappedFileSystemDirectory.class, org.tensorflow.proto.util.MemmappedFileSystemDirectory.Builder.class);
+  }
+
+  public static final int ELEMENT_FIELD_NUMBER = 1;
+  private java.util.List<org.tensorflow.proto.util.MemmappedFileSystemDirectoryElement> element_;
+  /**
+   * <code>repeated .tensorflow.MemmappedFileSystemDirectoryElement element = 1;</code>
+   */
+  public java.util.List<org.tensorflow.proto.util.MemmappedFileSystemDirectoryElement> getElementList() {
+    return element_;
+  }
+  /**
+   * <code>repeated .tensorflow.MemmappedFileSystemDirectoryElement element = 1;</code>
+   */
+  public java.util.List<? extends org.tensorflow.proto.util.MemmappedFileSystemDirectoryElementOrBuilder> 
+      getElementOrBuilderList() {
+    return element_;
+  }
+  /**
+   * <code>repeated .tensorflow.MemmappedFileSystemDirectoryElement element = 1;</code>
+   */
+  public int getElementCount() {
+    return element_.size();
+  }
+  /**
+   * <code>repeated .tensorflow.MemmappedFileSystemDirectoryElement element = 1;</code>
+   */
+  public org.tensorflow.proto.util.MemmappedFileSystemDirectoryElement getElement(int index) {
+    return element_.get(index);
+  }
+  /**
+   * <code>repeated .tensorflow.MemmappedFileSystemDirectoryElement element = 1;</code>
+   */
+  public org.tensorflow.proto.util.MemmappedFileSystemDirectoryElementOrBuilder getElementOrBuilder(
+      int index) {
+    return element_.get(index);
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    for (int i = 0; i < element_.size(); i++) {
+      output.writeMessage(1, element_.get(i));
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    for (int i = 0; i < element_.size(); i++) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(1, element_.get(i));
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.util.MemmappedFileSystemDirectory)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.util.MemmappedFileSystemDirectory other = (org.tensorflow.proto.util.MemmappedFileSystemDirectory) obj;
+
+    if (!getElementList()
+        .equals(other.getElementList())) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    if (getElementCount() > 0) {
+      hash = (37 * hash) + ELEMENT_FIELD_NUMBER;
+      hash = (53 * hash) + getElementList().hashCode();
+    }
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.util.MemmappedFileSystemDirectory parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.MemmappedFileSystemDirectory parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.MemmappedFileSystemDirectory parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.MemmappedFileSystemDirectory parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.MemmappedFileSystemDirectory parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.MemmappedFileSystemDirectory parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.MemmappedFileSystemDirectory parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.MemmappedFileSystemDirectory parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.MemmappedFileSystemDirectory parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.MemmappedFileSystemDirectory parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.MemmappedFileSystemDirectory parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.MemmappedFileSystemDirectory parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.util.MemmappedFileSystemDirectory prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * <pre>
+   * A directory of regions in a memmapped file.
+   * </pre>
+   *
+   * Protobuf type {@code tensorflow.MemmappedFileSystemDirectory}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.MemmappedFileSystemDirectory)
+      org.tensorflow.proto.util.MemmappedFileSystemDirectoryOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.util.MemmappedFileSystemProtos.internal_static_tensorflow_MemmappedFileSystemDirectory_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.util.MemmappedFileSystemProtos.internal_static_tensorflow_MemmappedFileSystemDirectory_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.util.MemmappedFileSystemDirectory.class, org.tensorflow.proto.util.MemmappedFileSystemDirectory.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.util.MemmappedFileSystemDirectory.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+        getElementFieldBuilder();
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      if (elementBuilder_ == null) {
+        element_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000001);
+      } else {
+        elementBuilder_.clear();
+      }
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.util.MemmappedFileSystemProtos.internal_static_tensorflow_MemmappedFileSystemDirectory_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.MemmappedFileSystemDirectory getDefaultInstanceForType() {
+      return org.tensorflow.proto.util.MemmappedFileSystemDirectory.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.MemmappedFileSystemDirectory build() {
+      org.tensorflow.proto.util.MemmappedFileSystemDirectory result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.MemmappedFileSystemDirectory buildPartial() {
+      org.tensorflow.proto.util.MemmappedFileSystemDirectory result = new org.tensorflow.proto.util.MemmappedFileSystemDirectory(this);
+      int from_bitField0_ = bitField0_;
+      if (elementBuilder_ == null) {
+        if (((bitField0_ & 0x00000001) != 0)) {
+          element_ = java.util.Collections.unmodifiableList(element_);
+          bitField0_ = (bitField0_ & ~0x00000001);
+        }
+        result.element_ = element_;
+      } else {
+        result.element_ = elementBuilder_.build();
+      }
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.util.MemmappedFileSystemDirectory) {
+        return mergeFrom((org.tensorflow.proto.util.MemmappedFileSystemDirectory)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.util.MemmappedFileSystemDirectory other) {
+      if (other == org.tensorflow.proto.util.MemmappedFileSystemDirectory.getDefaultInstance()) return this;
+      if (elementBuilder_ == null) {
+        if (!other.element_.isEmpty()) {
+          if (element_.isEmpty()) {
+            element_ = other.element_;
+            bitField0_ = (bitField0_ & ~0x00000001);
+          } else {
+            ensureElementIsMutable();
+            element_.addAll(other.element_);
+          }
+          onChanged();
+        }
+      } else {
+        if (!other.element_.isEmpty()) {
+          if (elementBuilder_.isEmpty()) {
+            elementBuilder_.dispose();
+            elementBuilder_ = null;
+            element_ = other.element_;
+            bitField0_ = (bitField0_ & ~0x00000001);
+            elementBuilder_ = 
+              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
+                 getElementFieldBuilder() : null;
+          } else {
+            elementBuilder_.addAllMessages(other.element_);
+          }
+        }
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.util.MemmappedFileSystemDirectory parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.util.MemmappedFileSystemDirectory) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+    private int bitField0_;
+
+    private java.util.List<org.tensorflow.proto.util.MemmappedFileSystemDirectoryElement> element_ =
+      java.util.Collections.emptyList();
+    private void ensureElementIsMutable() {
+      if (!((bitField0_ & 0x00000001) != 0)) {
+        element_ = new java.util.ArrayList<org.tensorflow.proto.util.MemmappedFileSystemDirectoryElement>(element_);
+        bitField0_ |= 0x00000001;
+       }
+    }
+
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.util.MemmappedFileSystemDirectoryElement, org.tensorflow.proto.util.MemmappedFileSystemDirectoryElement.Builder, org.tensorflow.proto.util.MemmappedFileSystemDirectoryElementOrBuilder> elementBuilder_;
+
+    /**
+     * <code>repeated .tensorflow.MemmappedFileSystemDirectoryElement element = 1;</code>
+     */
+    public java.util.List<org.tensorflow.proto.util.MemmappedFileSystemDirectoryElement> getElementList() {
+      if (elementBuilder_ == null) {
+        return java.util.Collections.unmodifiableList(element_);
+      } else {
+        return elementBuilder_.getMessageList();
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.MemmappedFileSystemDirectoryElement element = 1;</code>
+     */
+    public int getElementCount() {
+      if (elementBuilder_ == null) {
+        return element_.size();
+      } else {
+        return elementBuilder_.getCount();
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.MemmappedFileSystemDirectoryElement element = 1;</code>
+     */
+    public org.tensorflow.proto.util.MemmappedFileSystemDirectoryElement getElement(int index) {
+      if (elementBuilder_ == null) {
+        return element_.get(index);
+      } else {
+        return elementBuilder_.getMessage(index);
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.MemmappedFileSystemDirectoryElement element = 1;</code>
+     */
+    public Builder setElement(
+        int index, org.tensorflow.proto.util.MemmappedFileSystemDirectoryElement value) {
+      if (elementBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureElementIsMutable();
+        element_.set(index, value);
+        onChanged();
+      } else {
+        elementBuilder_.setMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.MemmappedFileSystemDirectoryElement element = 1;</code>
+     */
+    public Builder setElement(
+        int index, org.tensorflow.proto.util.MemmappedFileSystemDirectoryElement.Builder builderForValue) {
+      if (elementBuilder_ == null) {
+        ensureElementIsMutable();
+        element_.set(index, builderForValue.build());
+        onChanged();
+      } else {
+        elementBuilder_.setMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.MemmappedFileSystemDirectoryElement element = 1;</code>
+     */
+    public Builder addElement(org.tensorflow.proto.util.MemmappedFileSystemDirectoryElement value) {
+      if (elementBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureElementIsMutable();
+        element_.add(value);
+        onChanged();
+      } else {
+        elementBuilder_.addMessage(value);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.MemmappedFileSystemDirectoryElement element = 1;</code>
+     */
+    public Builder addElement(
+        int index, org.tensorflow.proto.util.MemmappedFileSystemDirectoryElement value) {
+      if (elementBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureElementIsMutable();
+        element_.add(index, value);
+        onChanged();
+      } else {
+        elementBuilder_.addMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.MemmappedFileSystemDirectoryElement element = 1;</code>
+     */
+    public Builder addElement(
+        org.tensorflow.proto.util.MemmappedFileSystemDirectoryElement.Builder builderForValue) {
+      if (elementBuilder_ == null) {
+        ensureElementIsMutable();
+        element_.add(builderForValue.build());
+        onChanged();
+      } else {
+        elementBuilder_.addMessage(builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.MemmappedFileSystemDirectoryElement element = 1;</code>
+     */
+    public Builder addElement(
+        int index, org.tensorflow.proto.util.MemmappedFileSystemDirectoryElement.Builder builderForValue) {
+      if (elementBuilder_ == null) {
+        ensureElementIsMutable();
+        element_.add(index, builderForValue.build());
+        onChanged();
+      } else {
+        elementBuilder_.addMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.MemmappedFileSystemDirectoryElement element = 1;</code>
+     */
+    public Builder addAllElement(
+        java.lang.Iterable<? extends org.tensorflow.proto.util.MemmappedFileSystemDirectoryElement> values) {
+      if (elementBuilder_ == null) {
+        ensureElementIsMutable();
+        com.google.protobuf.AbstractMessageLite.Builder.addAll(
+            values, element_);
+        onChanged();
+      } else {
+        elementBuilder_.addAllMessages(values);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.MemmappedFileSystemDirectoryElement element = 1;</code>
+     */
+    public Builder clearElement() {
+      if (elementBuilder_ == null) {
+        element_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000001);
+        onChanged();
+      } else {
+        elementBuilder_.clear();
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.MemmappedFileSystemDirectoryElement element = 1;</code>
+     */
+    public Builder removeElement(int index) {
+      if (elementBuilder_ == null) {
+        ensureElementIsMutable();
+        element_.remove(index);
+        onChanged();
+      } else {
+        elementBuilder_.remove(index);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.MemmappedFileSystemDirectoryElement element = 1;</code>
+     */
+    public org.tensorflow.proto.util.MemmappedFileSystemDirectoryElement.Builder getElementBuilder(
+        int index) {
+      return getElementFieldBuilder().getBuilder(index);
+    }
+    /**
+     * <code>repeated .tensorflow.MemmappedFileSystemDirectoryElement element = 1;</code>
+     */
+    public org.tensorflow.proto.util.MemmappedFileSystemDirectoryElementOrBuilder getElementOrBuilder(
+        int index) {
+      if (elementBuilder_ == null) {
+        return element_.get(index);  } else {
+        return elementBuilder_.getMessageOrBuilder(index);
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.MemmappedFileSystemDirectoryElement element = 1;</code>
+     */
+    public java.util.List<? extends org.tensorflow.proto.util.MemmappedFileSystemDirectoryElementOrBuilder> 
+         getElementOrBuilderList() {
+      if (elementBuilder_ != null) {
+        return elementBuilder_.getMessageOrBuilderList();
+      } else {
+        return java.util.Collections.unmodifiableList(element_);
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.MemmappedFileSystemDirectoryElement element = 1;</code>
+     */
+    public org.tensorflow.proto.util.MemmappedFileSystemDirectoryElement.Builder addElementBuilder() {
+      return getElementFieldBuilder().addBuilder(
+          org.tensorflow.proto.util.MemmappedFileSystemDirectoryElement.getDefaultInstance());
+    }
+    /**
+     * <code>repeated .tensorflow.MemmappedFileSystemDirectoryElement element = 1;</code>
+     */
+    public org.tensorflow.proto.util.MemmappedFileSystemDirectoryElement.Builder addElementBuilder(
+        int index) {
+      return getElementFieldBuilder().addBuilder(
+          index, org.tensorflow.proto.util.MemmappedFileSystemDirectoryElement.getDefaultInstance());
+    }
+    /**
+     * <code>repeated .tensorflow.MemmappedFileSystemDirectoryElement element = 1;</code>
+     */
+    public java.util.List<org.tensorflow.proto.util.MemmappedFileSystemDirectoryElement.Builder> 
+         getElementBuilderList() {
+      return getElementFieldBuilder().getBuilderList();
+    }
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.util.MemmappedFileSystemDirectoryElement, org.tensorflow.proto.util.MemmappedFileSystemDirectoryElement.Builder, org.tensorflow.proto.util.MemmappedFileSystemDirectoryElementOrBuilder> 
+        getElementFieldBuilder() {
+      if (elementBuilder_ == null) {
+        elementBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
+            org.tensorflow.proto.util.MemmappedFileSystemDirectoryElement, org.tensorflow.proto.util.MemmappedFileSystemDirectoryElement.Builder, org.tensorflow.proto.util.MemmappedFileSystemDirectoryElementOrBuilder>(
+                element_,
+                ((bitField0_ & 0x00000001) != 0),
+                getParentForChildren(),
+                isClean());
+        element_ = null;
+      }
+      return elementBuilder_;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.MemmappedFileSystemDirectory)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.MemmappedFileSystemDirectory)
+  private static final org.tensorflow.proto.util.MemmappedFileSystemDirectory DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.util.MemmappedFileSystemDirectory();
+  }
+
+  public static org.tensorflow.proto.util.MemmappedFileSystemDirectory getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<MemmappedFileSystemDirectory>
+      PARSER = new com.google.protobuf.AbstractParser<MemmappedFileSystemDirectory>() {
+    @java.lang.Override
+    public MemmappedFileSystemDirectory parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new MemmappedFileSystemDirectory(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<MemmappedFileSystemDirectory> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<MemmappedFileSystemDirectory> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.util.MemmappedFileSystemDirectory getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/MemmappedFileSystemDirectoryElement.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/MemmappedFileSystemDirectoryElement.java
new file mode 100644
index 00000000000..d2dafba3ef2
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/MemmappedFileSystemDirectoryElement.java
@@ -0,0 +1,670 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/util/memmapped_file_system.proto
+
+package org.tensorflow.proto.util;
+
+/**
+ * <pre>
+ * A message that describes one region of memmapped file.
+ * </pre>
+ *
+ * Protobuf type {@code tensorflow.MemmappedFileSystemDirectoryElement}
+ */
+public  final class MemmappedFileSystemDirectoryElement extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.MemmappedFileSystemDirectoryElement)
+    MemmappedFileSystemDirectoryElementOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use MemmappedFileSystemDirectoryElement.newBuilder() to construct.
+  private MemmappedFileSystemDirectoryElement(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private MemmappedFileSystemDirectoryElement() {
+    name_ = "";
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new MemmappedFileSystemDirectoryElement();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private MemmappedFileSystemDirectoryElement(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 8: {
+
+            offset_ = input.readUInt64();
+            break;
+          }
+          case 18: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            name_ = s;
+            break;
+          }
+          case 24: {
+
+            length_ = input.readUInt64();
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.util.MemmappedFileSystemProtos.internal_static_tensorflow_MemmappedFileSystemDirectoryElement_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.util.MemmappedFileSystemProtos.internal_static_tensorflow_MemmappedFileSystemDirectoryElement_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.util.MemmappedFileSystemDirectoryElement.class, org.tensorflow.proto.util.MemmappedFileSystemDirectoryElement.Builder.class);
+  }
+
+  public static final int OFFSET_FIELD_NUMBER = 1;
+  private long offset_;
+  /**
+   * <code>uint64 offset = 1;</code>
+   */
+  public long getOffset() {
+    return offset_;
+  }
+
+  public static final int NAME_FIELD_NUMBER = 2;
+  private volatile java.lang.Object name_;
+  /**
+   * <code>string name = 2;</code>
+   */
+  public java.lang.String getName() {
+    java.lang.Object ref = name_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      name_ = s;
+      return s;
+    }
+  }
+  /**
+   * <code>string name = 2;</code>
+   */
+  public com.google.protobuf.ByteString
+      getNameBytes() {
+    java.lang.Object ref = name_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      name_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int LENGTH_FIELD_NUMBER = 3;
+  private long length_;
+  /**
+   * <code>uint64 length = 3;</code>
+   */
+  public long getLength() {
+    return length_;
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    if (offset_ != 0L) {
+      output.writeUInt64(1, offset_);
+    }
+    if (!getNameBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 2, name_);
+    }
+    if (length_ != 0L) {
+      output.writeUInt64(3, length_);
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (offset_ != 0L) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeUInt64Size(1, offset_);
+    }
+    if (!getNameBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(2, name_);
+    }
+    if (length_ != 0L) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeUInt64Size(3, length_);
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.util.MemmappedFileSystemDirectoryElement)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.util.MemmappedFileSystemDirectoryElement other = (org.tensorflow.proto.util.MemmappedFileSystemDirectoryElement) obj;
+
+    if (getOffset()
+        != other.getOffset()) return false;
+    if (!getName()
+        .equals(other.getName())) return false;
+    if (getLength()
+        != other.getLength()) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    hash = (37 * hash) + OFFSET_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+        getOffset());
+    hash = (37 * hash) + NAME_FIELD_NUMBER;
+    hash = (53 * hash) + getName().hashCode();
+    hash = (37 * hash) + LENGTH_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+        getLength());
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.util.MemmappedFileSystemDirectoryElement parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.MemmappedFileSystemDirectoryElement parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.MemmappedFileSystemDirectoryElement parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.MemmappedFileSystemDirectoryElement parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.MemmappedFileSystemDirectoryElement parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.MemmappedFileSystemDirectoryElement parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.MemmappedFileSystemDirectoryElement parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.MemmappedFileSystemDirectoryElement parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.MemmappedFileSystemDirectoryElement parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.MemmappedFileSystemDirectoryElement parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.MemmappedFileSystemDirectoryElement parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.MemmappedFileSystemDirectoryElement parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.util.MemmappedFileSystemDirectoryElement prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * <pre>
+   * A message that describes one region of memmapped file.
+   * </pre>
+   *
+   * Protobuf type {@code tensorflow.MemmappedFileSystemDirectoryElement}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.MemmappedFileSystemDirectoryElement)
+      org.tensorflow.proto.util.MemmappedFileSystemDirectoryElementOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.util.MemmappedFileSystemProtos.internal_static_tensorflow_MemmappedFileSystemDirectoryElement_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.util.MemmappedFileSystemProtos.internal_static_tensorflow_MemmappedFileSystemDirectoryElement_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.util.MemmappedFileSystemDirectoryElement.class, org.tensorflow.proto.util.MemmappedFileSystemDirectoryElement.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.util.MemmappedFileSystemDirectoryElement.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      offset_ = 0L;
+
+      name_ = "";
+
+      length_ = 0L;
+
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.util.MemmappedFileSystemProtos.internal_static_tensorflow_MemmappedFileSystemDirectoryElement_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.MemmappedFileSystemDirectoryElement getDefaultInstanceForType() {
+      return org.tensorflow.proto.util.MemmappedFileSystemDirectoryElement.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.MemmappedFileSystemDirectoryElement build() {
+      org.tensorflow.proto.util.MemmappedFileSystemDirectoryElement result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.MemmappedFileSystemDirectoryElement buildPartial() {
+      org.tensorflow.proto.util.MemmappedFileSystemDirectoryElement result = new org.tensorflow.proto.util.MemmappedFileSystemDirectoryElement(this);
+      result.offset_ = offset_;
+      result.name_ = name_;
+      result.length_ = length_;
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.util.MemmappedFileSystemDirectoryElement) {
+        return mergeFrom((org.tensorflow.proto.util.MemmappedFileSystemDirectoryElement)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.util.MemmappedFileSystemDirectoryElement other) {
+      if (other == org.tensorflow.proto.util.MemmappedFileSystemDirectoryElement.getDefaultInstance()) return this;
+      if (other.getOffset() != 0L) {
+        setOffset(other.getOffset());
+      }
+      if (!other.getName().isEmpty()) {
+        name_ = other.name_;
+        onChanged();
+      }
+      if (other.getLength() != 0L) {
+        setLength(other.getLength());
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.util.MemmappedFileSystemDirectoryElement parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.util.MemmappedFileSystemDirectoryElement) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+
+    private long offset_ ;
+    /**
+     * <code>uint64 offset = 1;</code>
+     */
+    public long getOffset() {
+      return offset_;
+    }
+    /**
+     * <code>uint64 offset = 1;</code>
+     */
+    public Builder setOffset(long value) {
+      
+      offset_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>uint64 offset = 1;</code>
+     */
+    public Builder clearOffset() {
+      
+      offset_ = 0L;
+      onChanged();
+      return this;
+    }
+
+    private java.lang.Object name_ = "";
+    /**
+     * <code>string name = 2;</code>
+     */
+    public java.lang.String getName() {
+      java.lang.Object ref = name_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        name_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <code>string name = 2;</code>
+     */
+    public com.google.protobuf.ByteString
+        getNameBytes() {
+      java.lang.Object ref = name_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        name_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <code>string name = 2;</code>
+     */
+    public Builder setName(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      name_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>string name = 2;</code>
+     */
+    public Builder clearName() {
+      
+      name_ = getDefaultInstance().getName();
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>string name = 2;</code>
+     */
+    public Builder setNameBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      name_ = value;
+      onChanged();
+      return this;
+    }
+
+    private long length_ ;
+    /**
+     * <code>uint64 length = 3;</code>
+     */
+    public long getLength() {
+      return length_;
+    }
+    /**
+     * <code>uint64 length = 3;</code>
+     */
+    public Builder setLength(long value) {
+      
+      length_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>uint64 length = 3;</code>
+     */
+    public Builder clearLength() {
+      
+      length_ = 0L;
+      onChanged();
+      return this;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.MemmappedFileSystemDirectoryElement)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.MemmappedFileSystemDirectoryElement)
+  private static final org.tensorflow.proto.util.MemmappedFileSystemDirectoryElement DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.util.MemmappedFileSystemDirectoryElement();
+  }
+
+  public static org.tensorflow.proto.util.MemmappedFileSystemDirectoryElement getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<MemmappedFileSystemDirectoryElement>
+      PARSER = new com.google.protobuf.AbstractParser<MemmappedFileSystemDirectoryElement>() {
+    @java.lang.Override
+    public MemmappedFileSystemDirectoryElement parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new MemmappedFileSystemDirectoryElement(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<MemmappedFileSystemDirectoryElement> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<MemmappedFileSystemDirectoryElement> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.util.MemmappedFileSystemDirectoryElement getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/MemmappedFileSystemDirectoryElementOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/MemmappedFileSystemDirectoryElementOrBuilder.java
new file mode 100644
index 00000000000..93fd60f2faa
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/MemmappedFileSystemDirectoryElementOrBuilder.java
@@ -0,0 +1,29 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/util/memmapped_file_system.proto
+
+package org.tensorflow.proto.util;
+
+public interface MemmappedFileSystemDirectoryElementOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.MemmappedFileSystemDirectoryElement)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <code>uint64 offset = 1;</code>
+   */
+  long getOffset();
+
+  /**
+   * <code>string name = 2;</code>
+   */
+  java.lang.String getName();
+  /**
+   * <code>string name = 2;</code>
+   */
+  com.google.protobuf.ByteString
+      getNameBytes();
+
+  /**
+   * <code>uint64 length = 3;</code>
+   */
+  long getLength();
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/MemmappedFileSystemDirectoryOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/MemmappedFileSystemDirectoryOrBuilder.java
new file mode 100644
index 00000000000..c19c30645f5
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/MemmappedFileSystemDirectoryOrBuilder.java
@@ -0,0 +1,33 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/util/memmapped_file_system.proto
+
+package org.tensorflow.proto.util;
+
+public interface MemmappedFileSystemDirectoryOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.MemmappedFileSystemDirectory)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <code>repeated .tensorflow.MemmappedFileSystemDirectoryElement element = 1;</code>
+   */
+  java.util.List<org.tensorflow.proto.util.MemmappedFileSystemDirectoryElement> 
+      getElementList();
+  /**
+   * <code>repeated .tensorflow.MemmappedFileSystemDirectoryElement element = 1;</code>
+   */
+  org.tensorflow.proto.util.MemmappedFileSystemDirectoryElement getElement(int index);
+  /**
+   * <code>repeated .tensorflow.MemmappedFileSystemDirectoryElement element = 1;</code>
+   */
+  int getElementCount();
+  /**
+   * <code>repeated .tensorflow.MemmappedFileSystemDirectoryElement element = 1;</code>
+   */
+  java.util.List<? extends org.tensorflow.proto.util.MemmappedFileSystemDirectoryElementOrBuilder> 
+      getElementOrBuilderList();
+  /**
+   * <code>repeated .tensorflow.MemmappedFileSystemDirectoryElement element = 1;</code>
+   */
+  org.tensorflow.proto.util.MemmappedFileSystemDirectoryElementOrBuilder getElementOrBuilder(
+      int index);
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/MemmappedFileSystemProtos.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/MemmappedFileSystemProtos.java
new file mode 100644
index 00000000000..2e2809cce45
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/MemmappedFileSystemProtos.java
@@ -0,0 +1,64 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/util/memmapped_file_system.proto
+
+package org.tensorflow.proto.util;
+
+public final class MemmappedFileSystemProtos {
+  private MemmappedFileSystemProtos() {}
+  public static void registerAllExtensions(
+      com.google.protobuf.ExtensionRegistryLite registry) {
+  }
+
+  public static void registerAllExtensions(
+      com.google.protobuf.ExtensionRegistry registry) {
+    registerAllExtensions(
+        (com.google.protobuf.ExtensionRegistryLite) registry);
+  }
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_MemmappedFileSystemDirectoryElement_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_MemmappedFileSystemDirectoryElement_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_MemmappedFileSystemDirectory_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_MemmappedFileSystemDirectory_fieldAccessorTable;
+
+  public static com.google.protobuf.Descriptors.FileDescriptor
+      getDescriptor() {
+    return descriptor;
+  }
+  private static  com.google.protobuf.Descriptors.FileDescriptor
+      descriptor;
+  static {
+    java.lang.String[] descriptorData = {
+      "\n0tensorflow/core/util/memmapped_file_sy" +
+      "stem.proto\022\ntensorflow\"S\n#MemmappedFileS" +
+      "ystemDirectoryElement\022\016\n\006offset\030\001 \001(\004\022\014\n" +
+      "\004name\030\002 \001(\t\022\016\n\006length\030\003 \001(\004\"`\n\034Memmapped" +
+      "FileSystemDirectory\022@\n\007element\030\001 \003(\0132/.t" +
+      "ensorflow.MemmappedFileSystemDirectoryEl" +
+      "ementB;\n\031org.tensorflow.proto.utilB\031Memm" +
+      "appedFileSystemProtosP\001\370\001\001b\006proto3"
+    };
+    descriptor = com.google.protobuf.Descriptors.FileDescriptor
+      .internalBuildGeneratedFileFrom(descriptorData,
+        new com.google.protobuf.Descriptors.FileDescriptor[] {
+        });
+    internal_static_tensorflow_MemmappedFileSystemDirectoryElement_descriptor =
+      getDescriptor().getMessageTypes().get(0);
+    internal_static_tensorflow_MemmappedFileSystemDirectoryElement_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_MemmappedFileSystemDirectoryElement_descriptor,
+        new java.lang.String[] { "Offset", "Name", "Length", });
+    internal_static_tensorflow_MemmappedFileSystemDirectory_descriptor =
+      getDescriptor().getMessageTypes().get(1);
+    internal_static_tensorflow_MemmappedFileSystemDirectory_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_MemmappedFileSystemDirectory_descriptor,
+        new java.lang.String[] { "Element", });
+  }
+
+  // @@protoc_insertion_point(outer_class_scope)
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/MemoryDump.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/MemoryDump.java
new file mode 100644
index 00000000000..e7408bfc997
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/MemoryDump.java
@@ -0,0 +1,1759 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/bfc_memory_map.proto
+
+package org.tensorflow.proto.util;
+
+/**
+ * Protobuf type {@code tensorflow.MemoryDump}
+ */
+public  final class MemoryDump extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.MemoryDump)
+    MemoryDumpOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use MemoryDump.newBuilder() to construct.
+  private MemoryDump(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private MemoryDump() {
+    allocatorName_ = "";
+    binSummary_ = java.util.Collections.emptyList();
+    chunk_ = java.util.Collections.emptyList();
+    snapShot_ = java.util.Collections.emptyList();
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new MemoryDump();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private MemoryDump(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    int mutable_bitField0_ = 0;
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 10: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            allocatorName_ = s;
+            break;
+          }
+          case 18: {
+            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
+              binSummary_ = new java.util.ArrayList<org.tensorflow.proto.util.BinSummary>();
+              mutable_bitField0_ |= 0x00000001;
+            }
+            binSummary_.add(
+                input.readMessage(org.tensorflow.proto.util.BinSummary.parser(), extensionRegistry));
+            break;
+          }
+          case 26: {
+            if (!((mutable_bitField0_ & 0x00000002) != 0)) {
+              chunk_ = new java.util.ArrayList<org.tensorflow.proto.util.MemChunk>();
+              mutable_bitField0_ |= 0x00000002;
+            }
+            chunk_.add(
+                input.readMessage(org.tensorflow.proto.util.MemChunk.parser(), extensionRegistry));
+            break;
+          }
+          case 34: {
+            if (!((mutable_bitField0_ & 0x00000004) != 0)) {
+              snapShot_ = new java.util.ArrayList<org.tensorflow.proto.util.SnapShot>();
+              mutable_bitField0_ |= 0x00000004;
+            }
+            snapShot_.add(
+                input.readMessage(org.tensorflow.proto.util.SnapShot.parser(), extensionRegistry));
+            break;
+          }
+          case 42: {
+            org.tensorflow.proto.util.MemAllocatorStats.Builder subBuilder = null;
+            if (stats_ != null) {
+              subBuilder = stats_.toBuilder();
+            }
+            stats_ = input.readMessage(org.tensorflow.proto.util.MemAllocatorStats.parser(), extensionRegistry);
+            if (subBuilder != null) {
+              subBuilder.mergeFrom(stats_);
+              stats_ = subBuilder.buildPartial();
+            }
+
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      if (((mutable_bitField0_ & 0x00000001) != 0)) {
+        binSummary_ = java.util.Collections.unmodifiableList(binSummary_);
+      }
+      if (((mutable_bitField0_ & 0x00000002) != 0)) {
+        chunk_ = java.util.Collections.unmodifiableList(chunk_);
+      }
+      if (((mutable_bitField0_ & 0x00000004) != 0)) {
+        snapShot_ = java.util.Collections.unmodifiableList(snapShot_);
+      }
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.util.BfcMemoryMapProtos.internal_static_tensorflow_MemoryDump_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.util.BfcMemoryMapProtos.internal_static_tensorflow_MemoryDump_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.util.MemoryDump.class, org.tensorflow.proto.util.MemoryDump.Builder.class);
+  }
+
+  public static final int ALLOCATOR_NAME_FIELD_NUMBER = 1;
+  private volatile java.lang.Object allocatorName_;
+  /**
+   * <code>string allocator_name = 1;</code>
+   */
+  public java.lang.String getAllocatorName() {
+    java.lang.Object ref = allocatorName_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      allocatorName_ = s;
+      return s;
+    }
+  }
+  /**
+   * <code>string allocator_name = 1;</code>
+   */
+  public com.google.protobuf.ByteString
+      getAllocatorNameBytes() {
+    java.lang.Object ref = allocatorName_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      allocatorName_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int BIN_SUMMARY_FIELD_NUMBER = 2;
+  private java.util.List<org.tensorflow.proto.util.BinSummary> binSummary_;
+  /**
+   * <code>repeated .tensorflow.BinSummary bin_summary = 2;</code>
+   */
+  public java.util.List<org.tensorflow.proto.util.BinSummary> getBinSummaryList() {
+    return binSummary_;
+  }
+  /**
+   * <code>repeated .tensorflow.BinSummary bin_summary = 2;</code>
+   */
+  public java.util.List<? extends org.tensorflow.proto.util.BinSummaryOrBuilder> 
+      getBinSummaryOrBuilderList() {
+    return binSummary_;
+  }
+  /**
+   * <code>repeated .tensorflow.BinSummary bin_summary = 2;</code>
+   */
+  public int getBinSummaryCount() {
+    return binSummary_.size();
+  }
+  /**
+   * <code>repeated .tensorflow.BinSummary bin_summary = 2;</code>
+   */
+  public org.tensorflow.proto.util.BinSummary getBinSummary(int index) {
+    return binSummary_.get(index);
+  }
+  /**
+   * <code>repeated .tensorflow.BinSummary bin_summary = 2;</code>
+   */
+  public org.tensorflow.proto.util.BinSummaryOrBuilder getBinSummaryOrBuilder(
+      int index) {
+    return binSummary_.get(index);
+  }
+
+  public static final int CHUNK_FIELD_NUMBER = 3;
+  private java.util.List<org.tensorflow.proto.util.MemChunk> chunk_;
+  /**
+   * <code>repeated .tensorflow.MemChunk chunk = 3;</code>
+   */
+  public java.util.List<org.tensorflow.proto.util.MemChunk> getChunkList() {
+    return chunk_;
+  }
+  /**
+   * <code>repeated .tensorflow.MemChunk chunk = 3;</code>
+   */
+  public java.util.List<? extends org.tensorflow.proto.util.MemChunkOrBuilder> 
+      getChunkOrBuilderList() {
+    return chunk_;
+  }
+  /**
+   * <code>repeated .tensorflow.MemChunk chunk = 3;</code>
+   */
+  public int getChunkCount() {
+    return chunk_.size();
+  }
+  /**
+   * <code>repeated .tensorflow.MemChunk chunk = 3;</code>
+   */
+  public org.tensorflow.proto.util.MemChunk getChunk(int index) {
+    return chunk_.get(index);
+  }
+  /**
+   * <code>repeated .tensorflow.MemChunk chunk = 3;</code>
+   */
+  public org.tensorflow.proto.util.MemChunkOrBuilder getChunkOrBuilder(
+      int index) {
+    return chunk_.get(index);
+  }
+
+  public static final int SNAP_SHOT_FIELD_NUMBER = 4;
+  private java.util.List<org.tensorflow.proto.util.SnapShot> snapShot_;
+  /**
+   * <code>repeated .tensorflow.SnapShot snap_shot = 4;</code>
+   */
+  public java.util.List<org.tensorflow.proto.util.SnapShot> getSnapShotList() {
+    return snapShot_;
+  }
+  /**
+   * <code>repeated .tensorflow.SnapShot snap_shot = 4;</code>
+   */
+  public java.util.List<? extends org.tensorflow.proto.util.SnapShotOrBuilder> 
+      getSnapShotOrBuilderList() {
+    return snapShot_;
+  }
+  /**
+   * <code>repeated .tensorflow.SnapShot snap_shot = 4;</code>
+   */
+  public int getSnapShotCount() {
+    return snapShot_.size();
+  }
+  /**
+   * <code>repeated .tensorflow.SnapShot snap_shot = 4;</code>
+   */
+  public org.tensorflow.proto.util.SnapShot getSnapShot(int index) {
+    return snapShot_.get(index);
+  }
+  /**
+   * <code>repeated .tensorflow.SnapShot snap_shot = 4;</code>
+   */
+  public org.tensorflow.proto.util.SnapShotOrBuilder getSnapShotOrBuilder(
+      int index) {
+    return snapShot_.get(index);
+  }
+
+  public static final int STATS_FIELD_NUMBER = 5;
+  private org.tensorflow.proto.util.MemAllocatorStats stats_;
+  /**
+   * <code>.tensorflow.MemAllocatorStats stats = 5;</code>
+   */
+  public boolean hasStats() {
+    return stats_ != null;
+  }
+  /**
+   * <code>.tensorflow.MemAllocatorStats stats = 5;</code>
+   */
+  public org.tensorflow.proto.util.MemAllocatorStats getStats() {
+    return stats_ == null ? org.tensorflow.proto.util.MemAllocatorStats.getDefaultInstance() : stats_;
+  }
+  /**
+   * <code>.tensorflow.MemAllocatorStats stats = 5;</code>
+   */
+  public org.tensorflow.proto.util.MemAllocatorStatsOrBuilder getStatsOrBuilder() {
+    return getStats();
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    if (!getAllocatorNameBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 1, allocatorName_);
+    }
+    for (int i = 0; i < binSummary_.size(); i++) {
+      output.writeMessage(2, binSummary_.get(i));
+    }
+    for (int i = 0; i < chunk_.size(); i++) {
+      output.writeMessage(3, chunk_.get(i));
+    }
+    for (int i = 0; i < snapShot_.size(); i++) {
+      output.writeMessage(4, snapShot_.get(i));
+    }
+    if (stats_ != null) {
+      output.writeMessage(5, getStats());
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (!getAllocatorNameBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, allocatorName_);
+    }
+    for (int i = 0; i < binSummary_.size(); i++) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(2, binSummary_.get(i));
+    }
+    for (int i = 0; i < chunk_.size(); i++) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(3, chunk_.get(i));
+    }
+    for (int i = 0; i < snapShot_.size(); i++) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(4, snapShot_.get(i));
+    }
+    if (stats_ != null) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(5, getStats());
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.util.MemoryDump)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.util.MemoryDump other = (org.tensorflow.proto.util.MemoryDump) obj;
+
+    if (!getAllocatorName()
+        .equals(other.getAllocatorName())) return false;
+    if (!getBinSummaryList()
+        .equals(other.getBinSummaryList())) return false;
+    if (!getChunkList()
+        .equals(other.getChunkList())) return false;
+    if (!getSnapShotList()
+        .equals(other.getSnapShotList())) return false;
+    if (hasStats() != other.hasStats()) return false;
+    if (hasStats()) {
+      if (!getStats()
+          .equals(other.getStats())) return false;
+    }
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    hash = (37 * hash) + ALLOCATOR_NAME_FIELD_NUMBER;
+    hash = (53 * hash) + getAllocatorName().hashCode();
+    if (getBinSummaryCount() > 0) {
+      hash = (37 * hash) + BIN_SUMMARY_FIELD_NUMBER;
+      hash = (53 * hash) + getBinSummaryList().hashCode();
+    }
+    if (getChunkCount() > 0) {
+      hash = (37 * hash) + CHUNK_FIELD_NUMBER;
+      hash = (53 * hash) + getChunkList().hashCode();
+    }
+    if (getSnapShotCount() > 0) {
+      hash = (37 * hash) + SNAP_SHOT_FIELD_NUMBER;
+      hash = (53 * hash) + getSnapShotList().hashCode();
+    }
+    if (hasStats()) {
+      hash = (37 * hash) + STATS_FIELD_NUMBER;
+      hash = (53 * hash) + getStats().hashCode();
+    }
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.util.MemoryDump parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.MemoryDump parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.MemoryDump parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.MemoryDump parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.MemoryDump parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.MemoryDump parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.MemoryDump parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.MemoryDump parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.MemoryDump parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.MemoryDump parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.MemoryDump parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.MemoryDump parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.util.MemoryDump prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * Protobuf type {@code tensorflow.MemoryDump}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.MemoryDump)
+      org.tensorflow.proto.util.MemoryDumpOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.util.BfcMemoryMapProtos.internal_static_tensorflow_MemoryDump_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.util.BfcMemoryMapProtos.internal_static_tensorflow_MemoryDump_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.util.MemoryDump.class, org.tensorflow.proto.util.MemoryDump.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.util.MemoryDump.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+        getBinSummaryFieldBuilder();
+        getChunkFieldBuilder();
+        getSnapShotFieldBuilder();
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      allocatorName_ = "";
+
+      if (binSummaryBuilder_ == null) {
+        binSummary_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000001);
+      } else {
+        binSummaryBuilder_.clear();
+      }
+      if (chunkBuilder_ == null) {
+        chunk_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000002);
+      } else {
+        chunkBuilder_.clear();
+      }
+      if (snapShotBuilder_ == null) {
+        snapShot_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000004);
+      } else {
+        snapShotBuilder_.clear();
+      }
+      if (statsBuilder_ == null) {
+        stats_ = null;
+      } else {
+        stats_ = null;
+        statsBuilder_ = null;
+      }
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.util.BfcMemoryMapProtos.internal_static_tensorflow_MemoryDump_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.MemoryDump getDefaultInstanceForType() {
+      return org.tensorflow.proto.util.MemoryDump.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.MemoryDump build() {
+      org.tensorflow.proto.util.MemoryDump result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.MemoryDump buildPartial() {
+      org.tensorflow.proto.util.MemoryDump result = new org.tensorflow.proto.util.MemoryDump(this);
+      int from_bitField0_ = bitField0_;
+      result.allocatorName_ = allocatorName_;
+      if (binSummaryBuilder_ == null) {
+        if (((bitField0_ & 0x00000001) != 0)) {
+          binSummary_ = java.util.Collections.unmodifiableList(binSummary_);
+          bitField0_ = (bitField0_ & ~0x00000001);
+        }
+        result.binSummary_ = binSummary_;
+      } else {
+        result.binSummary_ = binSummaryBuilder_.build();
+      }
+      if (chunkBuilder_ == null) {
+        if (((bitField0_ & 0x00000002) != 0)) {
+          chunk_ = java.util.Collections.unmodifiableList(chunk_);
+          bitField0_ = (bitField0_ & ~0x00000002);
+        }
+        result.chunk_ = chunk_;
+      } else {
+        result.chunk_ = chunkBuilder_.build();
+      }
+      if (snapShotBuilder_ == null) {
+        if (((bitField0_ & 0x00000004) != 0)) {
+          snapShot_ = java.util.Collections.unmodifiableList(snapShot_);
+          bitField0_ = (bitField0_ & ~0x00000004);
+        }
+        result.snapShot_ = snapShot_;
+      } else {
+        result.snapShot_ = snapShotBuilder_.build();
+      }
+      if (statsBuilder_ == null) {
+        result.stats_ = stats_;
+      } else {
+        result.stats_ = statsBuilder_.build();
+      }
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.util.MemoryDump) {
+        return mergeFrom((org.tensorflow.proto.util.MemoryDump)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.util.MemoryDump other) {
+      if (other == org.tensorflow.proto.util.MemoryDump.getDefaultInstance()) return this;
+      if (!other.getAllocatorName().isEmpty()) {
+        allocatorName_ = other.allocatorName_;
+        onChanged();
+      }
+      if (binSummaryBuilder_ == null) {
+        if (!other.binSummary_.isEmpty()) {
+          if (binSummary_.isEmpty()) {
+            binSummary_ = other.binSummary_;
+            bitField0_ = (bitField0_ & ~0x00000001);
+          } else {
+            ensureBinSummaryIsMutable();
+            binSummary_.addAll(other.binSummary_);
+          }
+          onChanged();
+        }
+      } else {
+        if (!other.binSummary_.isEmpty()) {
+          if (binSummaryBuilder_.isEmpty()) {
+            binSummaryBuilder_.dispose();
+            binSummaryBuilder_ = null;
+            binSummary_ = other.binSummary_;
+            bitField0_ = (bitField0_ & ~0x00000001);
+            binSummaryBuilder_ = 
+              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
+                 getBinSummaryFieldBuilder() : null;
+          } else {
+            binSummaryBuilder_.addAllMessages(other.binSummary_);
+          }
+        }
+      }
+      if (chunkBuilder_ == null) {
+        if (!other.chunk_.isEmpty()) {
+          if (chunk_.isEmpty()) {
+            chunk_ = other.chunk_;
+            bitField0_ = (bitField0_ & ~0x00000002);
+          } else {
+            ensureChunkIsMutable();
+            chunk_.addAll(other.chunk_);
+          }
+          onChanged();
+        }
+      } else {
+        if (!other.chunk_.isEmpty()) {
+          if (chunkBuilder_.isEmpty()) {
+            chunkBuilder_.dispose();
+            chunkBuilder_ = null;
+            chunk_ = other.chunk_;
+            bitField0_ = (bitField0_ & ~0x00000002);
+            chunkBuilder_ = 
+              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
+                 getChunkFieldBuilder() : null;
+          } else {
+            chunkBuilder_.addAllMessages(other.chunk_);
+          }
+        }
+      }
+      if (snapShotBuilder_ == null) {
+        if (!other.snapShot_.isEmpty()) {
+          if (snapShot_.isEmpty()) {
+            snapShot_ = other.snapShot_;
+            bitField0_ = (bitField0_ & ~0x00000004);
+          } else {
+            ensureSnapShotIsMutable();
+            snapShot_.addAll(other.snapShot_);
+          }
+          onChanged();
+        }
+      } else {
+        if (!other.snapShot_.isEmpty()) {
+          if (snapShotBuilder_.isEmpty()) {
+            snapShotBuilder_.dispose();
+            snapShotBuilder_ = null;
+            snapShot_ = other.snapShot_;
+            bitField0_ = (bitField0_ & ~0x00000004);
+            snapShotBuilder_ = 
+              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
+                 getSnapShotFieldBuilder() : null;
+          } else {
+            snapShotBuilder_.addAllMessages(other.snapShot_);
+          }
+        }
+      }
+      if (other.hasStats()) {
+        mergeStats(other.getStats());
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.util.MemoryDump parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.util.MemoryDump) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+    private int bitField0_;
+
+    private java.lang.Object allocatorName_ = "";
+    /**
+     * <code>string allocator_name = 1;</code>
+     */
+    public java.lang.String getAllocatorName() {
+      java.lang.Object ref = allocatorName_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        allocatorName_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <code>string allocator_name = 1;</code>
+     */
+    public com.google.protobuf.ByteString
+        getAllocatorNameBytes() {
+      java.lang.Object ref = allocatorName_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        allocatorName_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <code>string allocator_name = 1;</code>
+     */
+    public Builder setAllocatorName(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      allocatorName_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>string allocator_name = 1;</code>
+     */
+    public Builder clearAllocatorName() {
+      
+      allocatorName_ = getDefaultInstance().getAllocatorName();
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>string allocator_name = 1;</code>
+     */
+    public Builder setAllocatorNameBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      allocatorName_ = value;
+      onChanged();
+      return this;
+    }
+
+    private java.util.List<org.tensorflow.proto.util.BinSummary> binSummary_ =
+      java.util.Collections.emptyList();
+    private void ensureBinSummaryIsMutable() {
+      if (!((bitField0_ & 0x00000001) != 0)) {
+        binSummary_ = new java.util.ArrayList<org.tensorflow.proto.util.BinSummary>(binSummary_);
+        bitField0_ |= 0x00000001;
+       }
+    }
+
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.util.BinSummary, org.tensorflow.proto.util.BinSummary.Builder, org.tensorflow.proto.util.BinSummaryOrBuilder> binSummaryBuilder_;
+
+    /**
+     * <code>repeated .tensorflow.BinSummary bin_summary = 2;</code>
+     */
+    public java.util.List<org.tensorflow.proto.util.BinSummary> getBinSummaryList() {
+      if (binSummaryBuilder_ == null) {
+        return java.util.Collections.unmodifiableList(binSummary_);
+      } else {
+        return binSummaryBuilder_.getMessageList();
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.BinSummary bin_summary = 2;</code>
+     */
+    public int getBinSummaryCount() {
+      if (binSummaryBuilder_ == null) {
+        return binSummary_.size();
+      } else {
+        return binSummaryBuilder_.getCount();
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.BinSummary bin_summary = 2;</code>
+     */
+    public org.tensorflow.proto.util.BinSummary getBinSummary(int index) {
+      if (binSummaryBuilder_ == null) {
+        return binSummary_.get(index);
+      } else {
+        return binSummaryBuilder_.getMessage(index);
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.BinSummary bin_summary = 2;</code>
+     */
+    public Builder setBinSummary(
+        int index, org.tensorflow.proto.util.BinSummary value) {
+      if (binSummaryBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureBinSummaryIsMutable();
+        binSummary_.set(index, value);
+        onChanged();
+      } else {
+        binSummaryBuilder_.setMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.BinSummary bin_summary = 2;</code>
+     */
+    public Builder setBinSummary(
+        int index, org.tensorflow.proto.util.BinSummary.Builder builderForValue) {
+      if (binSummaryBuilder_ == null) {
+        ensureBinSummaryIsMutable();
+        binSummary_.set(index, builderForValue.build());
+        onChanged();
+      } else {
+        binSummaryBuilder_.setMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.BinSummary bin_summary = 2;</code>
+     */
+    public Builder addBinSummary(org.tensorflow.proto.util.BinSummary value) {
+      if (binSummaryBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureBinSummaryIsMutable();
+        binSummary_.add(value);
+        onChanged();
+      } else {
+        binSummaryBuilder_.addMessage(value);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.BinSummary bin_summary = 2;</code>
+     */
+    public Builder addBinSummary(
+        int index, org.tensorflow.proto.util.BinSummary value) {
+      if (binSummaryBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureBinSummaryIsMutable();
+        binSummary_.add(index, value);
+        onChanged();
+      } else {
+        binSummaryBuilder_.addMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.BinSummary bin_summary = 2;</code>
+     */
+    public Builder addBinSummary(
+        org.tensorflow.proto.util.BinSummary.Builder builderForValue) {
+      if (binSummaryBuilder_ == null) {
+        ensureBinSummaryIsMutable();
+        binSummary_.add(builderForValue.build());
+        onChanged();
+      } else {
+        binSummaryBuilder_.addMessage(builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.BinSummary bin_summary = 2;</code>
+     */
+    public Builder addBinSummary(
+        int index, org.tensorflow.proto.util.BinSummary.Builder builderForValue) {
+      if (binSummaryBuilder_ == null) {
+        ensureBinSummaryIsMutable();
+        binSummary_.add(index, builderForValue.build());
+        onChanged();
+      } else {
+        binSummaryBuilder_.addMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.BinSummary bin_summary = 2;</code>
+     */
+    public Builder addAllBinSummary(
+        java.lang.Iterable<? extends org.tensorflow.proto.util.BinSummary> values) {
+      if (binSummaryBuilder_ == null) {
+        ensureBinSummaryIsMutable();
+        com.google.protobuf.AbstractMessageLite.Builder.addAll(
+            values, binSummary_);
+        onChanged();
+      } else {
+        binSummaryBuilder_.addAllMessages(values);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.BinSummary bin_summary = 2;</code>
+     */
+    public Builder clearBinSummary() {
+      if (binSummaryBuilder_ == null) {
+        binSummary_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000001);
+        onChanged();
+      } else {
+        binSummaryBuilder_.clear();
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.BinSummary bin_summary = 2;</code>
+     */
+    public Builder removeBinSummary(int index) {
+      if (binSummaryBuilder_ == null) {
+        ensureBinSummaryIsMutable();
+        binSummary_.remove(index);
+        onChanged();
+      } else {
+        binSummaryBuilder_.remove(index);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.BinSummary bin_summary = 2;</code>
+     */
+    public org.tensorflow.proto.util.BinSummary.Builder getBinSummaryBuilder(
+        int index) {
+      return getBinSummaryFieldBuilder().getBuilder(index);
+    }
+    /**
+     * <code>repeated .tensorflow.BinSummary bin_summary = 2;</code>
+     */
+    public org.tensorflow.proto.util.BinSummaryOrBuilder getBinSummaryOrBuilder(
+        int index) {
+      if (binSummaryBuilder_ == null) {
+        return binSummary_.get(index);  } else {
+        return binSummaryBuilder_.getMessageOrBuilder(index);
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.BinSummary bin_summary = 2;</code>
+     */
+    public java.util.List<? extends org.tensorflow.proto.util.BinSummaryOrBuilder> 
+         getBinSummaryOrBuilderList() {
+      if (binSummaryBuilder_ != null) {
+        return binSummaryBuilder_.getMessageOrBuilderList();
+      } else {
+        return java.util.Collections.unmodifiableList(binSummary_);
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.BinSummary bin_summary = 2;</code>
+     */
+    public org.tensorflow.proto.util.BinSummary.Builder addBinSummaryBuilder() {
+      return getBinSummaryFieldBuilder().addBuilder(
+          org.tensorflow.proto.util.BinSummary.getDefaultInstance());
+    }
+    /**
+     * <code>repeated .tensorflow.BinSummary bin_summary = 2;</code>
+     */
+    public org.tensorflow.proto.util.BinSummary.Builder addBinSummaryBuilder(
+        int index) {
+      return getBinSummaryFieldBuilder().addBuilder(
+          index, org.tensorflow.proto.util.BinSummary.getDefaultInstance());
+    }
+    /**
+     * <code>repeated .tensorflow.BinSummary bin_summary = 2;</code>
+     */
+    public java.util.List<org.tensorflow.proto.util.BinSummary.Builder> 
+         getBinSummaryBuilderList() {
+      return getBinSummaryFieldBuilder().getBuilderList();
+    }
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.util.BinSummary, org.tensorflow.proto.util.BinSummary.Builder, org.tensorflow.proto.util.BinSummaryOrBuilder> 
+        getBinSummaryFieldBuilder() {
+      if (binSummaryBuilder_ == null) {
+        binSummaryBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
+            org.tensorflow.proto.util.BinSummary, org.tensorflow.proto.util.BinSummary.Builder, org.tensorflow.proto.util.BinSummaryOrBuilder>(
+                binSummary_,
+                ((bitField0_ & 0x00000001) != 0),
+                getParentForChildren(),
+                isClean());
+        binSummary_ = null;
+      }
+      return binSummaryBuilder_;
+    }
+
+    private java.util.List<org.tensorflow.proto.util.MemChunk> chunk_ =
+      java.util.Collections.emptyList();
+    private void ensureChunkIsMutable() {
+      if (!((bitField0_ & 0x00000002) != 0)) {
+        chunk_ = new java.util.ArrayList<org.tensorflow.proto.util.MemChunk>(chunk_);
+        bitField0_ |= 0x00000002;
+       }
+    }
+
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.util.MemChunk, org.tensorflow.proto.util.MemChunk.Builder, org.tensorflow.proto.util.MemChunkOrBuilder> chunkBuilder_;
+
+    /**
+     * <code>repeated .tensorflow.MemChunk chunk = 3;</code>
+     */
+    public java.util.List<org.tensorflow.proto.util.MemChunk> getChunkList() {
+      if (chunkBuilder_ == null) {
+        return java.util.Collections.unmodifiableList(chunk_);
+      } else {
+        return chunkBuilder_.getMessageList();
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.MemChunk chunk = 3;</code>
+     */
+    public int getChunkCount() {
+      if (chunkBuilder_ == null) {
+        return chunk_.size();
+      } else {
+        return chunkBuilder_.getCount();
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.MemChunk chunk = 3;</code>
+     */
+    public org.tensorflow.proto.util.MemChunk getChunk(int index) {
+      if (chunkBuilder_ == null) {
+        return chunk_.get(index);
+      } else {
+        return chunkBuilder_.getMessage(index);
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.MemChunk chunk = 3;</code>
+     */
+    public Builder setChunk(
+        int index, org.tensorflow.proto.util.MemChunk value) {
+      if (chunkBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureChunkIsMutable();
+        chunk_.set(index, value);
+        onChanged();
+      } else {
+        chunkBuilder_.setMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.MemChunk chunk = 3;</code>
+     */
+    public Builder setChunk(
+        int index, org.tensorflow.proto.util.MemChunk.Builder builderForValue) {
+      if (chunkBuilder_ == null) {
+        ensureChunkIsMutable();
+        chunk_.set(index, builderForValue.build());
+        onChanged();
+      } else {
+        chunkBuilder_.setMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.MemChunk chunk = 3;</code>
+     */
+    public Builder addChunk(org.tensorflow.proto.util.MemChunk value) {
+      if (chunkBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureChunkIsMutable();
+        chunk_.add(value);
+        onChanged();
+      } else {
+        chunkBuilder_.addMessage(value);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.MemChunk chunk = 3;</code>
+     */
+    public Builder addChunk(
+        int index, org.tensorflow.proto.util.MemChunk value) {
+      if (chunkBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureChunkIsMutable();
+        chunk_.add(index, value);
+        onChanged();
+      } else {
+        chunkBuilder_.addMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.MemChunk chunk = 3;</code>
+     */
+    public Builder addChunk(
+        org.tensorflow.proto.util.MemChunk.Builder builderForValue) {
+      if (chunkBuilder_ == null) {
+        ensureChunkIsMutable();
+        chunk_.add(builderForValue.build());
+        onChanged();
+      } else {
+        chunkBuilder_.addMessage(builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.MemChunk chunk = 3;</code>
+     */
+    public Builder addChunk(
+        int index, org.tensorflow.proto.util.MemChunk.Builder builderForValue) {
+      if (chunkBuilder_ == null) {
+        ensureChunkIsMutable();
+        chunk_.add(index, builderForValue.build());
+        onChanged();
+      } else {
+        chunkBuilder_.addMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.MemChunk chunk = 3;</code>
+     */
+    public Builder addAllChunk(
+        java.lang.Iterable<? extends org.tensorflow.proto.util.MemChunk> values) {
+      if (chunkBuilder_ == null) {
+        ensureChunkIsMutable();
+        com.google.protobuf.AbstractMessageLite.Builder.addAll(
+            values, chunk_);
+        onChanged();
+      } else {
+        chunkBuilder_.addAllMessages(values);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.MemChunk chunk = 3;</code>
+     */
+    public Builder clearChunk() {
+      if (chunkBuilder_ == null) {
+        chunk_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000002);
+        onChanged();
+      } else {
+        chunkBuilder_.clear();
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.MemChunk chunk = 3;</code>
+     */
+    public Builder removeChunk(int index) {
+      if (chunkBuilder_ == null) {
+        ensureChunkIsMutable();
+        chunk_.remove(index);
+        onChanged();
+      } else {
+        chunkBuilder_.remove(index);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.MemChunk chunk = 3;</code>
+     */
+    public org.tensorflow.proto.util.MemChunk.Builder getChunkBuilder(
+        int index) {
+      return getChunkFieldBuilder().getBuilder(index);
+    }
+    /**
+     * <code>repeated .tensorflow.MemChunk chunk = 3;</code>
+     */
+    public org.tensorflow.proto.util.MemChunkOrBuilder getChunkOrBuilder(
+        int index) {
+      if (chunkBuilder_ == null) {
+        return chunk_.get(index);  } else {
+        return chunkBuilder_.getMessageOrBuilder(index);
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.MemChunk chunk = 3;</code>
+     */
+    public java.util.List<? extends org.tensorflow.proto.util.MemChunkOrBuilder> 
+         getChunkOrBuilderList() {
+      if (chunkBuilder_ != null) {
+        return chunkBuilder_.getMessageOrBuilderList();
+      } else {
+        return java.util.Collections.unmodifiableList(chunk_);
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.MemChunk chunk = 3;</code>
+     */
+    public org.tensorflow.proto.util.MemChunk.Builder addChunkBuilder() {
+      return getChunkFieldBuilder().addBuilder(
+          org.tensorflow.proto.util.MemChunk.getDefaultInstance());
+    }
+    /**
+     * <code>repeated .tensorflow.MemChunk chunk = 3;</code>
+     */
+    public org.tensorflow.proto.util.MemChunk.Builder addChunkBuilder(
+        int index) {
+      return getChunkFieldBuilder().addBuilder(
+          index, org.tensorflow.proto.util.MemChunk.getDefaultInstance());
+    }
+    /**
+     * <code>repeated .tensorflow.MemChunk chunk = 3;</code>
+     */
+    public java.util.List<org.tensorflow.proto.util.MemChunk.Builder> 
+         getChunkBuilderList() {
+      return getChunkFieldBuilder().getBuilderList();
+    }
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.util.MemChunk, org.tensorflow.proto.util.MemChunk.Builder, org.tensorflow.proto.util.MemChunkOrBuilder> 
+        getChunkFieldBuilder() {
+      if (chunkBuilder_ == null) {
+        chunkBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
+            org.tensorflow.proto.util.MemChunk, org.tensorflow.proto.util.MemChunk.Builder, org.tensorflow.proto.util.MemChunkOrBuilder>(
+                chunk_,
+                ((bitField0_ & 0x00000002) != 0),
+                getParentForChildren(),
+                isClean());
+        chunk_ = null;
+      }
+      return chunkBuilder_;
+    }
+
+    private java.util.List<org.tensorflow.proto.util.SnapShot> snapShot_ =
+      java.util.Collections.emptyList();
+    private void ensureSnapShotIsMutable() {
+      if (!((bitField0_ & 0x00000004) != 0)) {
+        snapShot_ = new java.util.ArrayList<org.tensorflow.proto.util.SnapShot>(snapShot_);
+        bitField0_ |= 0x00000004;
+       }
+    }
+
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.util.SnapShot, org.tensorflow.proto.util.SnapShot.Builder, org.tensorflow.proto.util.SnapShotOrBuilder> snapShotBuilder_;
+
+    /**
+     * <code>repeated .tensorflow.SnapShot snap_shot = 4;</code>
+     */
+    public java.util.List<org.tensorflow.proto.util.SnapShot> getSnapShotList() {
+      if (snapShotBuilder_ == null) {
+        return java.util.Collections.unmodifiableList(snapShot_);
+      } else {
+        return snapShotBuilder_.getMessageList();
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.SnapShot snap_shot = 4;</code>
+     */
+    public int getSnapShotCount() {
+      if (snapShotBuilder_ == null) {
+        return snapShot_.size();
+      } else {
+        return snapShotBuilder_.getCount();
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.SnapShot snap_shot = 4;</code>
+     */
+    public org.tensorflow.proto.util.SnapShot getSnapShot(int index) {
+      if (snapShotBuilder_ == null) {
+        return snapShot_.get(index);
+      } else {
+        return snapShotBuilder_.getMessage(index);
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.SnapShot snap_shot = 4;</code>
+     */
+    public Builder setSnapShot(
+        int index, org.tensorflow.proto.util.SnapShot value) {
+      if (snapShotBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureSnapShotIsMutable();
+        snapShot_.set(index, value);
+        onChanged();
+      } else {
+        snapShotBuilder_.setMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.SnapShot snap_shot = 4;</code>
+     */
+    public Builder setSnapShot(
+        int index, org.tensorflow.proto.util.SnapShot.Builder builderForValue) {
+      if (snapShotBuilder_ == null) {
+        ensureSnapShotIsMutable();
+        snapShot_.set(index, builderForValue.build());
+        onChanged();
+      } else {
+        snapShotBuilder_.setMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.SnapShot snap_shot = 4;</code>
+     */
+    public Builder addSnapShot(org.tensorflow.proto.util.SnapShot value) {
+      if (snapShotBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureSnapShotIsMutable();
+        snapShot_.add(value);
+        onChanged();
+      } else {
+        snapShotBuilder_.addMessage(value);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.SnapShot snap_shot = 4;</code>
+     */
+    public Builder addSnapShot(
+        int index, org.tensorflow.proto.util.SnapShot value) {
+      if (snapShotBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureSnapShotIsMutable();
+        snapShot_.add(index, value);
+        onChanged();
+      } else {
+        snapShotBuilder_.addMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.SnapShot snap_shot = 4;</code>
+     */
+    public Builder addSnapShot(
+        org.tensorflow.proto.util.SnapShot.Builder builderForValue) {
+      if (snapShotBuilder_ == null) {
+        ensureSnapShotIsMutable();
+        snapShot_.add(builderForValue.build());
+        onChanged();
+      } else {
+        snapShotBuilder_.addMessage(builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.SnapShot snap_shot = 4;</code>
+     */
+    public Builder addSnapShot(
+        int index, org.tensorflow.proto.util.SnapShot.Builder builderForValue) {
+      if (snapShotBuilder_ == null) {
+        ensureSnapShotIsMutable();
+        snapShot_.add(index, builderForValue.build());
+        onChanged();
+      } else {
+        snapShotBuilder_.addMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.SnapShot snap_shot = 4;</code>
+     */
+    public Builder addAllSnapShot(
+        java.lang.Iterable<? extends org.tensorflow.proto.util.SnapShot> values) {
+      if (snapShotBuilder_ == null) {
+        ensureSnapShotIsMutable();
+        com.google.protobuf.AbstractMessageLite.Builder.addAll(
+            values, snapShot_);
+        onChanged();
+      } else {
+        snapShotBuilder_.addAllMessages(values);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.SnapShot snap_shot = 4;</code>
+     */
+    public Builder clearSnapShot() {
+      if (snapShotBuilder_ == null) {
+        snapShot_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000004);
+        onChanged();
+      } else {
+        snapShotBuilder_.clear();
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.SnapShot snap_shot = 4;</code>
+     */
+    public Builder removeSnapShot(int index) {
+      if (snapShotBuilder_ == null) {
+        ensureSnapShotIsMutable();
+        snapShot_.remove(index);
+        onChanged();
+      } else {
+        snapShotBuilder_.remove(index);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.SnapShot snap_shot = 4;</code>
+     */
+    public org.tensorflow.proto.util.SnapShot.Builder getSnapShotBuilder(
+        int index) {
+      return getSnapShotFieldBuilder().getBuilder(index);
+    }
+    /**
+     * <code>repeated .tensorflow.SnapShot snap_shot = 4;</code>
+     */
+    public org.tensorflow.proto.util.SnapShotOrBuilder getSnapShotOrBuilder(
+        int index) {
+      if (snapShotBuilder_ == null) {
+        return snapShot_.get(index);  } else {
+        return snapShotBuilder_.getMessageOrBuilder(index);
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.SnapShot snap_shot = 4;</code>
+     */
+    public java.util.List<? extends org.tensorflow.proto.util.SnapShotOrBuilder> 
+         getSnapShotOrBuilderList() {
+      if (snapShotBuilder_ != null) {
+        return snapShotBuilder_.getMessageOrBuilderList();
+      } else {
+        return java.util.Collections.unmodifiableList(snapShot_);
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.SnapShot snap_shot = 4;</code>
+     */
+    public org.tensorflow.proto.util.SnapShot.Builder addSnapShotBuilder() {
+      return getSnapShotFieldBuilder().addBuilder(
+          org.tensorflow.proto.util.SnapShot.getDefaultInstance());
+    }
+    /**
+     * <code>repeated .tensorflow.SnapShot snap_shot = 4;</code>
+     */
+    public org.tensorflow.proto.util.SnapShot.Builder addSnapShotBuilder(
+        int index) {
+      return getSnapShotFieldBuilder().addBuilder(
+          index, org.tensorflow.proto.util.SnapShot.getDefaultInstance());
+    }
+    /**
+     * <code>repeated .tensorflow.SnapShot snap_shot = 4;</code>
+     */
+    public java.util.List<org.tensorflow.proto.util.SnapShot.Builder> 
+         getSnapShotBuilderList() {
+      return getSnapShotFieldBuilder().getBuilderList();
+    }
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.util.SnapShot, org.tensorflow.proto.util.SnapShot.Builder, org.tensorflow.proto.util.SnapShotOrBuilder> 
+        getSnapShotFieldBuilder() {
+      if (snapShotBuilder_ == null) {
+        snapShotBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
+            org.tensorflow.proto.util.SnapShot, org.tensorflow.proto.util.SnapShot.Builder, org.tensorflow.proto.util.SnapShotOrBuilder>(
+                snapShot_,
+                ((bitField0_ & 0x00000004) != 0),
+                getParentForChildren(),
+                isClean());
+        snapShot_ = null;
+      }
+      return snapShotBuilder_;
+    }
+
+    private org.tensorflow.proto.util.MemAllocatorStats stats_;
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.util.MemAllocatorStats, org.tensorflow.proto.util.MemAllocatorStats.Builder, org.tensorflow.proto.util.MemAllocatorStatsOrBuilder> statsBuilder_;
+    /**
+     * <code>.tensorflow.MemAllocatorStats stats = 5;</code>
+     */
+    public boolean hasStats() {
+      return statsBuilder_ != null || stats_ != null;
+    }
+    /**
+     * <code>.tensorflow.MemAllocatorStats stats = 5;</code>
+     */
+    public org.tensorflow.proto.util.MemAllocatorStats getStats() {
+      if (statsBuilder_ == null) {
+        return stats_ == null ? org.tensorflow.proto.util.MemAllocatorStats.getDefaultInstance() : stats_;
+      } else {
+        return statsBuilder_.getMessage();
+      }
+    }
+    /**
+     * <code>.tensorflow.MemAllocatorStats stats = 5;</code>
+     */
+    public Builder setStats(org.tensorflow.proto.util.MemAllocatorStats value) {
+      if (statsBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        stats_ = value;
+        onChanged();
+      } else {
+        statsBuilder_.setMessage(value);
+      }
+
+      return this;
+    }
+    /**
+     * <code>.tensorflow.MemAllocatorStats stats = 5;</code>
+     */
+    public Builder setStats(
+        org.tensorflow.proto.util.MemAllocatorStats.Builder builderForValue) {
+      if (statsBuilder_ == null) {
+        stats_ = builderForValue.build();
+        onChanged();
+      } else {
+        statsBuilder_.setMessage(builderForValue.build());
+      }
+
+      return this;
+    }
+    /**
+     * <code>.tensorflow.MemAllocatorStats stats = 5;</code>
+     */
+    public Builder mergeStats(org.tensorflow.proto.util.MemAllocatorStats value) {
+      if (statsBuilder_ == null) {
+        if (stats_ != null) {
+          stats_ =
+            org.tensorflow.proto.util.MemAllocatorStats.newBuilder(stats_).mergeFrom(value).buildPartial();
+        } else {
+          stats_ = value;
+        }
+        onChanged();
+      } else {
+        statsBuilder_.mergeFrom(value);
+      }
+
+      return this;
+    }
+    /**
+     * <code>.tensorflow.MemAllocatorStats stats = 5;</code>
+     */
+    public Builder clearStats() {
+      if (statsBuilder_ == null) {
+        stats_ = null;
+        onChanged();
+      } else {
+        stats_ = null;
+        statsBuilder_ = null;
+      }
+
+      return this;
+    }
+    /**
+     * <code>.tensorflow.MemAllocatorStats stats = 5;</code>
+     */
+    public org.tensorflow.proto.util.MemAllocatorStats.Builder getStatsBuilder() {
+      
+      onChanged();
+      return getStatsFieldBuilder().getBuilder();
+    }
+    /**
+     * <code>.tensorflow.MemAllocatorStats stats = 5;</code>
+     */
+    public org.tensorflow.proto.util.MemAllocatorStatsOrBuilder getStatsOrBuilder() {
+      if (statsBuilder_ != null) {
+        return statsBuilder_.getMessageOrBuilder();
+      } else {
+        return stats_ == null ?
+            org.tensorflow.proto.util.MemAllocatorStats.getDefaultInstance() : stats_;
+      }
+    }
+    /**
+     * <code>.tensorflow.MemAllocatorStats stats = 5;</code>
+     */
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.util.MemAllocatorStats, org.tensorflow.proto.util.MemAllocatorStats.Builder, org.tensorflow.proto.util.MemAllocatorStatsOrBuilder> 
+        getStatsFieldBuilder() {
+      if (statsBuilder_ == null) {
+        statsBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+            org.tensorflow.proto.util.MemAllocatorStats, org.tensorflow.proto.util.MemAllocatorStats.Builder, org.tensorflow.proto.util.MemAllocatorStatsOrBuilder>(
+                getStats(),
+                getParentForChildren(),
+                isClean());
+        stats_ = null;
+      }
+      return statsBuilder_;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.MemoryDump)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.MemoryDump)
+  private static final org.tensorflow.proto.util.MemoryDump DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.util.MemoryDump();
+  }
+
+  public static org.tensorflow.proto.util.MemoryDump getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<MemoryDump>
+      PARSER = new com.google.protobuf.AbstractParser<MemoryDump>() {
+    @java.lang.Override
+    public MemoryDump parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new MemoryDump(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<MemoryDump> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<MemoryDump> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.util.MemoryDump getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/MemoryDumpOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/MemoryDumpOrBuilder.java
new file mode 100644
index 00000000000..5d15667734f
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/MemoryDumpOrBuilder.java
@@ -0,0 +1,104 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/bfc_memory_map.proto
+
+package org.tensorflow.proto.util;
+
+public interface MemoryDumpOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.MemoryDump)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <code>string allocator_name = 1;</code>
+   */
+  java.lang.String getAllocatorName();
+  /**
+   * <code>string allocator_name = 1;</code>
+   */
+  com.google.protobuf.ByteString
+      getAllocatorNameBytes();
+
+  /**
+   * <code>repeated .tensorflow.BinSummary bin_summary = 2;</code>
+   */
+  java.util.List<org.tensorflow.proto.util.BinSummary> 
+      getBinSummaryList();
+  /**
+   * <code>repeated .tensorflow.BinSummary bin_summary = 2;</code>
+   */
+  org.tensorflow.proto.util.BinSummary getBinSummary(int index);
+  /**
+   * <code>repeated .tensorflow.BinSummary bin_summary = 2;</code>
+   */
+  int getBinSummaryCount();
+  /**
+   * <code>repeated .tensorflow.BinSummary bin_summary = 2;</code>
+   */
+  java.util.List<? extends org.tensorflow.proto.util.BinSummaryOrBuilder> 
+      getBinSummaryOrBuilderList();
+  /**
+   * <code>repeated .tensorflow.BinSummary bin_summary = 2;</code>
+   */
+  org.tensorflow.proto.util.BinSummaryOrBuilder getBinSummaryOrBuilder(
+      int index);
+
+  /**
+   * <code>repeated .tensorflow.MemChunk chunk = 3;</code>
+   */
+  java.util.List<org.tensorflow.proto.util.MemChunk> 
+      getChunkList();
+  /**
+   * <code>repeated .tensorflow.MemChunk chunk = 3;</code>
+   */
+  org.tensorflow.proto.util.MemChunk getChunk(int index);
+  /**
+   * <code>repeated .tensorflow.MemChunk chunk = 3;</code>
+   */
+  int getChunkCount();
+  /**
+   * <code>repeated .tensorflow.MemChunk chunk = 3;</code>
+   */
+  java.util.List<? extends org.tensorflow.proto.util.MemChunkOrBuilder> 
+      getChunkOrBuilderList();
+  /**
+   * <code>repeated .tensorflow.MemChunk chunk = 3;</code>
+   */
+  org.tensorflow.proto.util.MemChunkOrBuilder getChunkOrBuilder(
+      int index);
+
+  /**
+   * <code>repeated .tensorflow.SnapShot snap_shot = 4;</code>
+   */
+  java.util.List<org.tensorflow.proto.util.SnapShot> 
+      getSnapShotList();
+  /**
+   * <code>repeated .tensorflow.SnapShot snap_shot = 4;</code>
+   */
+  org.tensorflow.proto.util.SnapShot getSnapShot(int index);
+  /**
+   * <code>repeated .tensorflow.SnapShot snap_shot = 4;</code>
+   */
+  int getSnapShotCount();
+  /**
+   * <code>repeated .tensorflow.SnapShot snap_shot = 4;</code>
+   */
+  java.util.List<? extends org.tensorflow.proto.util.SnapShotOrBuilder> 
+      getSnapShotOrBuilderList();
+  /**
+   * <code>repeated .tensorflow.SnapShot snap_shot = 4;</code>
+   */
+  org.tensorflow.proto.util.SnapShotOrBuilder getSnapShotOrBuilder(
+      int index);
+
+  /**
+   * <code>.tensorflow.MemAllocatorStats stats = 5;</code>
+   */
+  boolean hasStats();
+  /**
+   * <code>.tensorflow.MemAllocatorStats stats = 5;</code>
+   */
+  org.tensorflow.proto.util.MemAllocatorStats getStats();
+  /**
+   * <code>.tensorflow.MemAllocatorStats stats = 5;</code>
+   */
+  org.tensorflow.proto.util.MemAllocatorStatsOrBuilder getStatsOrBuilder();
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/RequestedExitCode.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/RequestedExitCode.java
new file mode 100644
index 00000000000..093cc8f7b5e
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/RequestedExitCode.java
@@ -0,0 +1,476 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/util/event.proto
+
+package org.tensorflow.proto.util;
+
+/**
+ * Protobuf type {@code tensorflow.RequestedExitCode}
+ */
+public  final class RequestedExitCode extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.RequestedExitCode)
+    RequestedExitCodeOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use RequestedExitCode.newBuilder() to construct.
+  private RequestedExitCode(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private RequestedExitCode() {
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new RequestedExitCode();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private RequestedExitCode(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 8: {
+
+            exitCode_ = input.readInt32();
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.util.EventProtos.internal_static_tensorflow_RequestedExitCode_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.util.EventProtos.internal_static_tensorflow_RequestedExitCode_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.util.RequestedExitCode.class, org.tensorflow.proto.util.RequestedExitCode.Builder.class);
+  }
+
+  public static final int EXIT_CODE_FIELD_NUMBER = 1;
+  private int exitCode_;
+  /**
+   * <code>int32 exit_code = 1;</code>
+   */
+  public int getExitCode() {
+    return exitCode_;
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    if (exitCode_ != 0) {
+      output.writeInt32(1, exitCode_);
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (exitCode_ != 0) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt32Size(1, exitCode_);
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.util.RequestedExitCode)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.util.RequestedExitCode other = (org.tensorflow.proto.util.RequestedExitCode) obj;
+
+    if (getExitCode()
+        != other.getExitCode()) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    hash = (37 * hash) + EXIT_CODE_FIELD_NUMBER;
+    hash = (53 * hash) + getExitCode();
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.util.RequestedExitCode parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.RequestedExitCode parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.RequestedExitCode parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.RequestedExitCode parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.RequestedExitCode parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.RequestedExitCode parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.RequestedExitCode parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.RequestedExitCode parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.RequestedExitCode parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.RequestedExitCode parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.RequestedExitCode parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.RequestedExitCode parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.util.RequestedExitCode prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * Protobuf type {@code tensorflow.RequestedExitCode}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.RequestedExitCode)
+      org.tensorflow.proto.util.RequestedExitCodeOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.util.EventProtos.internal_static_tensorflow_RequestedExitCode_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.util.EventProtos.internal_static_tensorflow_RequestedExitCode_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.util.RequestedExitCode.class, org.tensorflow.proto.util.RequestedExitCode.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.util.RequestedExitCode.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      exitCode_ = 0;
+
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.util.EventProtos.internal_static_tensorflow_RequestedExitCode_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.RequestedExitCode getDefaultInstanceForType() {
+      return org.tensorflow.proto.util.RequestedExitCode.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.RequestedExitCode build() {
+      org.tensorflow.proto.util.RequestedExitCode result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.RequestedExitCode buildPartial() {
+      org.tensorflow.proto.util.RequestedExitCode result = new org.tensorflow.proto.util.RequestedExitCode(this);
+      result.exitCode_ = exitCode_;
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.util.RequestedExitCode) {
+        return mergeFrom((org.tensorflow.proto.util.RequestedExitCode)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.util.RequestedExitCode other) {
+      if (other == org.tensorflow.proto.util.RequestedExitCode.getDefaultInstance()) return this;
+      if (other.getExitCode() != 0) {
+        setExitCode(other.getExitCode());
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.util.RequestedExitCode parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.util.RequestedExitCode) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+
+    private int exitCode_ ;
+    /**
+     * <code>int32 exit_code = 1;</code>
+     */
+    public int getExitCode() {
+      return exitCode_;
+    }
+    /**
+     * <code>int32 exit_code = 1;</code>
+     */
+    public Builder setExitCode(int value) {
+      
+      exitCode_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>int32 exit_code = 1;</code>
+     */
+    public Builder clearExitCode() {
+      
+      exitCode_ = 0;
+      onChanged();
+      return this;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.RequestedExitCode)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.RequestedExitCode)
+  private static final org.tensorflow.proto.util.RequestedExitCode DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.util.RequestedExitCode();
+  }
+
+  public static org.tensorflow.proto.util.RequestedExitCode getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<RequestedExitCode>
+      PARSER = new com.google.protobuf.AbstractParser<RequestedExitCode>() {
+    @java.lang.Override
+    public RequestedExitCode parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new RequestedExitCode(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<RequestedExitCode> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<RequestedExitCode> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.util.RequestedExitCode getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/RequestedExitCodeOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/RequestedExitCodeOrBuilder.java
new file mode 100644
index 00000000000..260e7609dc7
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/RequestedExitCodeOrBuilder.java
@@ -0,0 +1,14 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/util/event.proto
+
+package org.tensorflow.proto.util;
+
+public interface RequestedExitCodeOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.RequestedExitCode)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <code>int32 exit_code = 1;</code>
+   */
+  int getExitCode();
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SavedSlice.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SavedSlice.java
new file mode 100644
index 00000000000..b1473544ce7
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SavedSlice.java
@@ -0,0 +1,1073 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/util/saved_tensor_slice.proto
+
+package org.tensorflow.proto.util;
+
+/**
+ * <pre>
+ * Saved tensor slice: it stores the name of the tensors, the slice, and the
+ * raw data.
+ * </pre>
+ *
+ * Protobuf type {@code tensorflow.SavedSlice}
+ */
+public  final class SavedSlice extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.SavedSlice)
+    SavedSliceOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use SavedSlice.newBuilder() to construct.
+  private SavedSlice(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private SavedSlice() {
+    name_ = "";
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new SavedSlice();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private SavedSlice(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 10: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            name_ = s;
+            break;
+          }
+          case 18: {
+            org.tensorflow.proto.framework.TensorSliceProto.Builder subBuilder = null;
+            if (slice_ != null) {
+              subBuilder = slice_.toBuilder();
+            }
+            slice_ = input.readMessage(org.tensorflow.proto.framework.TensorSliceProto.parser(), extensionRegistry);
+            if (subBuilder != null) {
+              subBuilder.mergeFrom(slice_);
+              slice_ = subBuilder.buildPartial();
+            }
+
+            break;
+          }
+          case 26: {
+            org.tensorflow.proto.framework.TensorProto.Builder subBuilder = null;
+            if (data_ != null) {
+              subBuilder = data_.toBuilder();
+            }
+            data_ = input.readMessage(org.tensorflow.proto.framework.TensorProto.parser(), extensionRegistry);
+            if (subBuilder != null) {
+              subBuilder.mergeFrom(data_);
+              data_ = subBuilder.buildPartial();
+            }
+
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.util.SavedTensorSliceProtos.internal_static_tensorflow_SavedSlice_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.util.SavedTensorSliceProtos.internal_static_tensorflow_SavedSlice_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.util.SavedSlice.class, org.tensorflow.proto.util.SavedSlice.Builder.class);
+  }
+
+  public static final int NAME_FIELD_NUMBER = 1;
+  private volatile java.lang.Object name_;
+  /**
+   * <pre>
+   * Name of the tensor that this slice belongs to. This must be identical to
+   * the name used to encode the key for this record.
+   * </pre>
+   *
+   * <code>string name = 1;</code>
+   */
+  public java.lang.String getName() {
+    java.lang.Object ref = name_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      name_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * Name of the tensor that this slice belongs to. This must be identical to
+   * the name used to encode the key for this record.
+   * </pre>
+   *
+   * <code>string name = 1;</code>
+   */
+  public com.google.protobuf.ByteString
+      getNameBytes() {
+    java.lang.Object ref = name_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      name_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int SLICE_FIELD_NUMBER = 2;
+  private org.tensorflow.proto.framework.TensorSliceProto slice_;
+  /**
+   * <pre>
+   * Extent of the slice.  Must have one entry for each of the dimension of the
+   * tensor that this slice belongs to.
+   * </pre>
+   *
+   * <code>.tensorflow.TensorSliceProto slice = 2;</code>
+   */
+  public boolean hasSlice() {
+    return slice_ != null;
+  }
+  /**
+   * <pre>
+   * Extent of the slice.  Must have one entry for each of the dimension of the
+   * tensor that this slice belongs to.
+   * </pre>
+   *
+   * <code>.tensorflow.TensorSliceProto slice = 2;</code>
+   */
+  public org.tensorflow.proto.framework.TensorSliceProto getSlice() {
+    return slice_ == null ? org.tensorflow.proto.framework.TensorSliceProto.getDefaultInstance() : slice_;
+  }
+  /**
+   * <pre>
+   * Extent of the slice.  Must have one entry for each of the dimension of the
+   * tensor that this slice belongs to.
+   * </pre>
+   *
+   * <code>.tensorflow.TensorSliceProto slice = 2;</code>
+   */
+  public org.tensorflow.proto.framework.TensorSliceProtoOrBuilder getSliceOrBuilder() {
+    return getSlice();
+  }
+
+  public static final int DATA_FIELD_NUMBER = 3;
+  private org.tensorflow.proto.framework.TensorProto data_;
+  /**
+   * <pre>
+   * The raw data of the slice is stored as a TensorProto. Only raw data are
+   * stored (we don't fill in fields such as dtype or tensor_shape).
+   * </pre>
+   *
+   * <code>.tensorflow.TensorProto data = 3;</code>
+   */
+  public boolean hasData() {
+    return data_ != null;
+  }
+  /**
+   * <pre>
+   * The raw data of the slice is stored as a TensorProto. Only raw data are
+   * stored (we don't fill in fields such as dtype or tensor_shape).
+   * </pre>
+   *
+   * <code>.tensorflow.TensorProto data = 3;</code>
+   */
+  public org.tensorflow.proto.framework.TensorProto getData() {
+    return data_ == null ? org.tensorflow.proto.framework.TensorProto.getDefaultInstance() : data_;
+  }
+  /**
+   * <pre>
+   * The raw data of the slice is stored as a TensorProto. Only raw data are
+   * stored (we don't fill in fields such as dtype or tensor_shape).
+   * </pre>
+   *
+   * <code>.tensorflow.TensorProto data = 3;</code>
+   */
+  public org.tensorflow.proto.framework.TensorProtoOrBuilder getDataOrBuilder() {
+    return getData();
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    if (!getNameBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 1, name_);
+    }
+    if (slice_ != null) {
+      output.writeMessage(2, getSlice());
+    }
+    if (data_ != null) {
+      output.writeMessage(3, getData());
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (!getNameBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, name_);
+    }
+    if (slice_ != null) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(2, getSlice());
+    }
+    if (data_ != null) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(3, getData());
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.util.SavedSlice)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.util.SavedSlice other = (org.tensorflow.proto.util.SavedSlice) obj;
+
+    if (!getName()
+        .equals(other.getName())) return false;
+    if (hasSlice() != other.hasSlice()) return false;
+    if (hasSlice()) {
+      if (!getSlice()
+          .equals(other.getSlice())) return false;
+    }
+    if (hasData() != other.hasData()) return false;
+    if (hasData()) {
+      if (!getData()
+          .equals(other.getData())) return false;
+    }
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    hash = (37 * hash) + NAME_FIELD_NUMBER;
+    hash = (53 * hash) + getName().hashCode();
+    if (hasSlice()) {
+      hash = (37 * hash) + SLICE_FIELD_NUMBER;
+      hash = (53 * hash) + getSlice().hashCode();
+    }
+    if (hasData()) {
+      hash = (37 * hash) + DATA_FIELD_NUMBER;
+      hash = (53 * hash) + getData().hashCode();
+    }
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.util.SavedSlice parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.SavedSlice parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.SavedSlice parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.SavedSlice parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.SavedSlice parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.SavedSlice parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.SavedSlice parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.SavedSlice parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.SavedSlice parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.SavedSlice parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.SavedSlice parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.SavedSlice parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.util.SavedSlice prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * <pre>
+   * Saved tensor slice: it stores the name of the tensors, the slice, and the
+   * raw data.
+   * </pre>
+   *
+   * Protobuf type {@code tensorflow.SavedSlice}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.SavedSlice)
+      org.tensorflow.proto.util.SavedSliceOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.util.SavedTensorSliceProtos.internal_static_tensorflow_SavedSlice_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.util.SavedTensorSliceProtos.internal_static_tensorflow_SavedSlice_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.util.SavedSlice.class, org.tensorflow.proto.util.SavedSlice.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.util.SavedSlice.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      name_ = "";
+
+      if (sliceBuilder_ == null) {
+        slice_ = null;
+      } else {
+        slice_ = null;
+        sliceBuilder_ = null;
+      }
+      if (dataBuilder_ == null) {
+        data_ = null;
+      } else {
+        data_ = null;
+        dataBuilder_ = null;
+      }
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.util.SavedTensorSliceProtos.internal_static_tensorflow_SavedSlice_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.SavedSlice getDefaultInstanceForType() {
+      return org.tensorflow.proto.util.SavedSlice.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.SavedSlice build() {
+      org.tensorflow.proto.util.SavedSlice result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.SavedSlice buildPartial() {
+      org.tensorflow.proto.util.SavedSlice result = new org.tensorflow.proto.util.SavedSlice(this);
+      result.name_ = name_;
+      if (sliceBuilder_ == null) {
+        result.slice_ = slice_;
+      } else {
+        result.slice_ = sliceBuilder_.build();
+      }
+      if (dataBuilder_ == null) {
+        result.data_ = data_;
+      } else {
+        result.data_ = dataBuilder_.build();
+      }
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.util.SavedSlice) {
+        return mergeFrom((org.tensorflow.proto.util.SavedSlice)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.util.SavedSlice other) {
+      if (other == org.tensorflow.proto.util.SavedSlice.getDefaultInstance()) return this;
+      if (!other.getName().isEmpty()) {
+        name_ = other.name_;
+        onChanged();
+      }
+      if (other.hasSlice()) {
+        mergeSlice(other.getSlice());
+      }
+      if (other.hasData()) {
+        mergeData(other.getData());
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.util.SavedSlice parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.util.SavedSlice) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+
+    private java.lang.Object name_ = "";
+    /**
+     * <pre>
+     * Name of the tensor that this slice belongs to. This must be identical to
+     * the name used to encode the key for this record.
+     * </pre>
+     *
+     * <code>string name = 1;</code>
+     */
+    public java.lang.String getName() {
+      java.lang.Object ref = name_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        name_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Name of the tensor that this slice belongs to. This must be identical to
+     * the name used to encode the key for this record.
+     * </pre>
+     *
+     * <code>string name = 1;</code>
+     */
+    public com.google.protobuf.ByteString
+        getNameBytes() {
+      java.lang.Object ref = name_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        name_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Name of the tensor that this slice belongs to. This must be identical to
+     * the name used to encode the key for this record.
+     * </pre>
+     *
+     * <code>string name = 1;</code>
+     */
+    public Builder setName(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      name_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Name of the tensor that this slice belongs to. This must be identical to
+     * the name used to encode the key for this record.
+     * </pre>
+     *
+     * <code>string name = 1;</code>
+     */
+    public Builder clearName() {
+      
+      name_ = getDefaultInstance().getName();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Name of the tensor that this slice belongs to. This must be identical to
+     * the name used to encode the key for this record.
+     * </pre>
+     *
+     * <code>string name = 1;</code>
+     */
+    public Builder setNameBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      name_ = value;
+      onChanged();
+      return this;
+    }
+
+    private org.tensorflow.proto.framework.TensorSliceProto slice_;
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.TensorSliceProto, org.tensorflow.proto.framework.TensorSliceProto.Builder, org.tensorflow.proto.framework.TensorSliceProtoOrBuilder> sliceBuilder_;
+    /**
+     * <pre>
+     * Extent of the slice.  Must have one entry for each of the dimension of the
+     * tensor that this slice belongs to.
+     * </pre>
+     *
+     * <code>.tensorflow.TensorSliceProto slice = 2;</code>
+     */
+    public boolean hasSlice() {
+      return sliceBuilder_ != null || slice_ != null;
+    }
+    /**
+     * <pre>
+     * Extent of the slice.  Must have one entry for each of the dimension of the
+     * tensor that this slice belongs to.
+     * </pre>
+     *
+     * <code>.tensorflow.TensorSliceProto slice = 2;</code>
+     */
+    public org.tensorflow.proto.framework.TensorSliceProto getSlice() {
+      if (sliceBuilder_ == null) {
+        return slice_ == null ? org.tensorflow.proto.framework.TensorSliceProto.getDefaultInstance() : slice_;
+      } else {
+        return sliceBuilder_.getMessage();
+      }
+    }
+    /**
+     * <pre>
+     * Extent of the slice.  Must have one entry for each of the dimension of the
+     * tensor that this slice belongs to.
+     * </pre>
+     *
+     * <code>.tensorflow.TensorSliceProto slice = 2;</code>
+     */
+    public Builder setSlice(org.tensorflow.proto.framework.TensorSliceProto value) {
+      if (sliceBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        slice_ = value;
+        onChanged();
+      } else {
+        sliceBuilder_.setMessage(value);
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Extent of the slice.  Must have one entry for each of the dimension of the
+     * tensor that this slice belongs to.
+     * </pre>
+     *
+     * <code>.tensorflow.TensorSliceProto slice = 2;</code>
+     */
+    public Builder setSlice(
+        org.tensorflow.proto.framework.TensorSliceProto.Builder builderForValue) {
+      if (sliceBuilder_ == null) {
+        slice_ = builderForValue.build();
+        onChanged();
+      } else {
+        sliceBuilder_.setMessage(builderForValue.build());
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Extent of the slice.  Must have one entry for each of the dimension of the
+     * tensor that this slice belongs to.
+     * </pre>
+     *
+     * <code>.tensorflow.TensorSliceProto slice = 2;</code>
+     */
+    public Builder mergeSlice(org.tensorflow.proto.framework.TensorSliceProto value) {
+      if (sliceBuilder_ == null) {
+        if (slice_ != null) {
+          slice_ =
+            org.tensorflow.proto.framework.TensorSliceProto.newBuilder(slice_).mergeFrom(value).buildPartial();
+        } else {
+          slice_ = value;
+        }
+        onChanged();
+      } else {
+        sliceBuilder_.mergeFrom(value);
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Extent of the slice.  Must have one entry for each of the dimension of the
+     * tensor that this slice belongs to.
+     * </pre>
+     *
+     * <code>.tensorflow.TensorSliceProto slice = 2;</code>
+     */
+    public Builder clearSlice() {
+      if (sliceBuilder_ == null) {
+        slice_ = null;
+        onChanged();
+      } else {
+        slice_ = null;
+        sliceBuilder_ = null;
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Extent of the slice.  Must have one entry for each of the dimension of the
+     * tensor that this slice belongs to.
+     * </pre>
+     *
+     * <code>.tensorflow.TensorSliceProto slice = 2;</code>
+     */
+    public org.tensorflow.proto.framework.TensorSliceProto.Builder getSliceBuilder() {
+      
+      onChanged();
+      return getSliceFieldBuilder().getBuilder();
+    }
+    /**
+     * <pre>
+     * Extent of the slice.  Must have one entry for each of the dimension of the
+     * tensor that this slice belongs to.
+     * </pre>
+     *
+     * <code>.tensorflow.TensorSliceProto slice = 2;</code>
+     */
+    public org.tensorflow.proto.framework.TensorSliceProtoOrBuilder getSliceOrBuilder() {
+      if (sliceBuilder_ != null) {
+        return sliceBuilder_.getMessageOrBuilder();
+      } else {
+        return slice_ == null ?
+            org.tensorflow.proto.framework.TensorSliceProto.getDefaultInstance() : slice_;
+      }
+    }
+    /**
+     * <pre>
+     * Extent of the slice.  Must have one entry for each of the dimension of the
+     * tensor that this slice belongs to.
+     * </pre>
+     *
+     * <code>.tensorflow.TensorSliceProto slice = 2;</code>
+     */
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.TensorSliceProto, org.tensorflow.proto.framework.TensorSliceProto.Builder, org.tensorflow.proto.framework.TensorSliceProtoOrBuilder> 
+        getSliceFieldBuilder() {
+      if (sliceBuilder_ == null) {
+        sliceBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+            org.tensorflow.proto.framework.TensorSliceProto, org.tensorflow.proto.framework.TensorSliceProto.Builder, org.tensorflow.proto.framework.TensorSliceProtoOrBuilder>(
+                getSlice(),
+                getParentForChildren(),
+                isClean());
+        slice_ = null;
+      }
+      return sliceBuilder_;
+    }
+
+    private org.tensorflow.proto.framework.TensorProto data_;
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.TensorProto, org.tensorflow.proto.framework.TensorProto.Builder, org.tensorflow.proto.framework.TensorProtoOrBuilder> dataBuilder_;
+    /**
+     * <pre>
+     * The raw data of the slice is stored as a TensorProto. Only raw data are
+     * stored (we don't fill in fields such as dtype or tensor_shape).
+     * </pre>
+     *
+     * <code>.tensorflow.TensorProto data = 3;</code>
+     */
+    public boolean hasData() {
+      return dataBuilder_ != null || data_ != null;
+    }
+    /**
+     * <pre>
+     * The raw data of the slice is stored as a TensorProto. Only raw data are
+     * stored (we don't fill in fields such as dtype or tensor_shape).
+     * </pre>
+     *
+     * <code>.tensorflow.TensorProto data = 3;</code>
+     */
+    public org.tensorflow.proto.framework.TensorProto getData() {
+      if (dataBuilder_ == null) {
+        return data_ == null ? org.tensorflow.proto.framework.TensorProto.getDefaultInstance() : data_;
+      } else {
+        return dataBuilder_.getMessage();
+      }
+    }
+    /**
+     * <pre>
+     * The raw data of the slice is stored as a TensorProto. Only raw data are
+     * stored (we don't fill in fields such as dtype or tensor_shape).
+     * </pre>
+     *
+     * <code>.tensorflow.TensorProto data = 3;</code>
+     */
+    public Builder setData(org.tensorflow.proto.framework.TensorProto value) {
+      if (dataBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        data_ = value;
+        onChanged();
+      } else {
+        dataBuilder_.setMessage(value);
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * The raw data of the slice is stored as a TensorProto. Only raw data are
+     * stored (we don't fill in fields such as dtype or tensor_shape).
+     * </pre>
+     *
+     * <code>.tensorflow.TensorProto data = 3;</code>
+     */
+    public Builder setData(
+        org.tensorflow.proto.framework.TensorProto.Builder builderForValue) {
+      if (dataBuilder_ == null) {
+        data_ = builderForValue.build();
+        onChanged();
+      } else {
+        dataBuilder_.setMessage(builderForValue.build());
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * The raw data of the slice is stored as a TensorProto. Only raw data are
+     * stored (we don't fill in fields such as dtype or tensor_shape).
+     * </pre>
+     *
+     * <code>.tensorflow.TensorProto data = 3;</code>
+     */
+    public Builder mergeData(org.tensorflow.proto.framework.TensorProto value) {
+      if (dataBuilder_ == null) {
+        if (data_ != null) {
+          data_ =
+            org.tensorflow.proto.framework.TensorProto.newBuilder(data_).mergeFrom(value).buildPartial();
+        } else {
+          data_ = value;
+        }
+        onChanged();
+      } else {
+        dataBuilder_.mergeFrom(value);
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * The raw data of the slice is stored as a TensorProto. Only raw data are
+     * stored (we don't fill in fields such as dtype or tensor_shape).
+     * </pre>
+     *
+     * <code>.tensorflow.TensorProto data = 3;</code>
+     */
+    public Builder clearData() {
+      if (dataBuilder_ == null) {
+        data_ = null;
+        onChanged();
+      } else {
+        data_ = null;
+        dataBuilder_ = null;
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * The raw data of the slice is stored as a TensorProto. Only raw data are
+     * stored (we don't fill in fields such as dtype or tensor_shape).
+     * </pre>
+     *
+     * <code>.tensorflow.TensorProto data = 3;</code>
+     */
+    public org.tensorflow.proto.framework.TensorProto.Builder getDataBuilder() {
+      
+      onChanged();
+      return getDataFieldBuilder().getBuilder();
+    }
+    /**
+     * <pre>
+     * The raw data of the slice is stored as a TensorProto. Only raw data are
+     * stored (we don't fill in fields such as dtype or tensor_shape).
+     * </pre>
+     *
+     * <code>.tensorflow.TensorProto data = 3;</code>
+     */
+    public org.tensorflow.proto.framework.TensorProtoOrBuilder getDataOrBuilder() {
+      if (dataBuilder_ != null) {
+        return dataBuilder_.getMessageOrBuilder();
+      } else {
+        return data_ == null ?
+            org.tensorflow.proto.framework.TensorProto.getDefaultInstance() : data_;
+      }
+    }
+    /**
+     * <pre>
+     * The raw data of the slice is stored as a TensorProto. Only raw data are
+     * stored (we don't fill in fields such as dtype or tensor_shape).
+     * </pre>
+     *
+     * <code>.tensorflow.TensorProto data = 3;</code>
+     */
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.TensorProto, org.tensorflow.proto.framework.TensorProto.Builder, org.tensorflow.proto.framework.TensorProtoOrBuilder> 
+        getDataFieldBuilder() {
+      if (dataBuilder_ == null) {
+        dataBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+            org.tensorflow.proto.framework.TensorProto, org.tensorflow.proto.framework.TensorProto.Builder, org.tensorflow.proto.framework.TensorProtoOrBuilder>(
+                getData(),
+                getParentForChildren(),
+                isClean());
+        data_ = null;
+      }
+      return dataBuilder_;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.SavedSlice)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.SavedSlice)
+  private static final org.tensorflow.proto.util.SavedSlice DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.util.SavedSlice();
+  }
+
+  public static org.tensorflow.proto.util.SavedSlice getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<SavedSlice>
+      PARSER = new com.google.protobuf.AbstractParser<SavedSlice>() {
+    @java.lang.Override
+    public SavedSlice parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new SavedSlice(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<SavedSlice> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<SavedSlice> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.util.SavedSlice getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SavedSliceMeta.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SavedSliceMeta.java
new file mode 100644
index 00000000000..884792faec0
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SavedSliceMeta.java
@@ -0,0 +1,1364 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/util/saved_tensor_slice.proto
+
+package org.tensorflow.proto.util;
+
+/**
+ * <pre>
+ * Metadata describing the set of slices of the same tensor saved in a
+ * checkpoint file.
+ * </pre>
+ *
+ * Protobuf type {@code tensorflow.SavedSliceMeta}
+ */
+public  final class SavedSliceMeta extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.SavedSliceMeta)
+    SavedSliceMetaOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use SavedSliceMeta.newBuilder() to construct.
+  private SavedSliceMeta(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private SavedSliceMeta() {
+    name_ = "";
+    type_ = 0;
+    slice_ = java.util.Collections.emptyList();
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new SavedSliceMeta();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private SavedSliceMeta(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    int mutable_bitField0_ = 0;
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 10: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            name_ = s;
+            break;
+          }
+          case 18: {
+            org.tensorflow.proto.framework.TensorShapeProto.Builder subBuilder = null;
+            if (shape_ != null) {
+              subBuilder = shape_.toBuilder();
+            }
+            shape_ = input.readMessage(org.tensorflow.proto.framework.TensorShapeProto.parser(), extensionRegistry);
+            if (subBuilder != null) {
+              subBuilder.mergeFrom(shape_);
+              shape_ = subBuilder.buildPartial();
+            }
+
+            break;
+          }
+          case 24: {
+            int rawValue = input.readEnum();
+
+            type_ = rawValue;
+            break;
+          }
+          case 34: {
+            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
+              slice_ = new java.util.ArrayList<org.tensorflow.proto.framework.TensorSliceProto>();
+              mutable_bitField0_ |= 0x00000001;
+            }
+            slice_.add(
+                input.readMessage(org.tensorflow.proto.framework.TensorSliceProto.parser(), extensionRegistry));
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      if (((mutable_bitField0_ & 0x00000001) != 0)) {
+        slice_ = java.util.Collections.unmodifiableList(slice_);
+      }
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.util.SavedTensorSliceProtos.internal_static_tensorflow_SavedSliceMeta_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.util.SavedTensorSliceProtos.internal_static_tensorflow_SavedSliceMeta_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.util.SavedSliceMeta.class, org.tensorflow.proto.util.SavedSliceMeta.Builder.class);
+  }
+
+  public static final int NAME_FIELD_NUMBER = 1;
+  private volatile java.lang.Object name_;
+  /**
+   * <pre>
+   * Name of the tensor.
+   * </pre>
+   *
+   * <code>string name = 1;</code>
+   */
+  public java.lang.String getName() {
+    java.lang.Object ref = name_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      name_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * Name of the tensor.
+   * </pre>
+   *
+   * <code>string name = 1;</code>
+   */
+  public com.google.protobuf.ByteString
+      getNameBytes() {
+    java.lang.Object ref = name_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      name_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int SHAPE_FIELD_NUMBER = 2;
+  private org.tensorflow.proto.framework.TensorShapeProto shape_;
+  /**
+   * <pre>
+   * Shape of the tensor
+   * </pre>
+   *
+   * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+   */
+  public boolean hasShape() {
+    return shape_ != null;
+  }
+  /**
+   * <pre>
+   * Shape of the tensor
+   * </pre>
+   *
+   * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+   */
+  public org.tensorflow.proto.framework.TensorShapeProto getShape() {
+    return shape_ == null ? org.tensorflow.proto.framework.TensorShapeProto.getDefaultInstance() : shape_;
+  }
+  /**
+   * <pre>
+   * Shape of the tensor
+   * </pre>
+   *
+   * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+   */
+  public org.tensorflow.proto.framework.TensorShapeProtoOrBuilder getShapeOrBuilder() {
+    return getShape();
+  }
+
+  public static final int TYPE_FIELD_NUMBER = 3;
+  private int type_;
+  /**
+   * <pre>
+   * Type of the tensor
+   * </pre>
+   *
+   * <code>.tensorflow.DataType type = 3;</code>
+   */
+  public int getTypeValue() {
+    return type_;
+  }
+  /**
+   * <pre>
+   * Type of the tensor
+   * </pre>
+   *
+   * <code>.tensorflow.DataType type = 3;</code>
+   */
+  public org.tensorflow.proto.framework.DataType getType() {
+    @SuppressWarnings("deprecation")
+    org.tensorflow.proto.framework.DataType result = org.tensorflow.proto.framework.DataType.valueOf(type_);
+    return result == null ? org.tensorflow.proto.framework.DataType.UNRECOGNIZED : result;
+  }
+
+  public static final int SLICE_FIELD_NUMBER = 4;
+  private java.util.List<org.tensorflow.proto.framework.TensorSliceProto> slice_;
+  /**
+   * <pre>
+   * Explicit list of slices saved in the checkpoint file.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.TensorSliceProto slice = 4;</code>
+   */
+  public java.util.List<org.tensorflow.proto.framework.TensorSliceProto> getSliceList() {
+    return slice_;
+  }
+  /**
+   * <pre>
+   * Explicit list of slices saved in the checkpoint file.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.TensorSliceProto slice = 4;</code>
+   */
+  public java.util.List<? extends org.tensorflow.proto.framework.TensorSliceProtoOrBuilder> 
+      getSliceOrBuilderList() {
+    return slice_;
+  }
+  /**
+   * <pre>
+   * Explicit list of slices saved in the checkpoint file.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.TensorSliceProto slice = 4;</code>
+   */
+  public int getSliceCount() {
+    return slice_.size();
+  }
+  /**
+   * <pre>
+   * Explicit list of slices saved in the checkpoint file.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.TensorSliceProto slice = 4;</code>
+   */
+  public org.tensorflow.proto.framework.TensorSliceProto getSlice(int index) {
+    return slice_.get(index);
+  }
+  /**
+   * <pre>
+   * Explicit list of slices saved in the checkpoint file.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.TensorSliceProto slice = 4;</code>
+   */
+  public org.tensorflow.proto.framework.TensorSliceProtoOrBuilder getSliceOrBuilder(
+      int index) {
+    return slice_.get(index);
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    if (!getNameBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 1, name_);
+    }
+    if (shape_ != null) {
+      output.writeMessage(2, getShape());
+    }
+    if (type_ != org.tensorflow.proto.framework.DataType.DT_INVALID.getNumber()) {
+      output.writeEnum(3, type_);
+    }
+    for (int i = 0; i < slice_.size(); i++) {
+      output.writeMessage(4, slice_.get(i));
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (!getNameBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, name_);
+    }
+    if (shape_ != null) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(2, getShape());
+    }
+    if (type_ != org.tensorflow.proto.framework.DataType.DT_INVALID.getNumber()) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeEnumSize(3, type_);
+    }
+    for (int i = 0; i < slice_.size(); i++) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(4, slice_.get(i));
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.util.SavedSliceMeta)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.util.SavedSliceMeta other = (org.tensorflow.proto.util.SavedSliceMeta) obj;
+
+    if (!getName()
+        .equals(other.getName())) return false;
+    if (hasShape() != other.hasShape()) return false;
+    if (hasShape()) {
+      if (!getShape()
+          .equals(other.getShape())) return false;
+    }
+    if (type_ != other.type_) return false;
+    if (!getSliceList()
+        .equals(other.getSliceList())) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    hash = (37 * hash) + NAME_FIELD_NUMBER;
+    hash = (53 * hash) + getName().hashCode();
+    if (hasShape()) {
+      hash = (37 * hash) + SHAPE_FIELD_NUMBER;
+      hash = (53 * hash) + getShape().hashCode();
+    }
+    hash = (37 * hash) + TYPE_FIELD_NUMBER;
+    hash = (53 * hash) + type_;
+    if (getSliceCount() > 0) {
+      hash = (37 * hash) + SLICE_FIELD_NUMBER;
+      hash = (53 * hash) + getSliceList().hashCode();
+    }
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.util.SavedSliceMeta parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.SavedSliceMeta parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.SavedSliceMeta parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.SavedSliceMeta parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.SavedSliceMeta parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.SavedSliceMeta parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.SavedSliceMeta parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.SavedSliceMeta parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.SavedSliceMeta parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.SavedSliceMeta parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.SavedSliceMeta parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.SavedSliceMeta parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.util.SavedSliceMeta prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * <pre>
+   * Metadata describing the set of slices of the same tensor saved in a
+   * checkpoint file.
+   * </pre>
+   *
+   * Protobuf type {@code tensorflow.SavedSliceMeta}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.SavedSliceMeta)
+      org.tensorflow.proto.util.SavedSliceMetaOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.util.SavedTensorSliceProtos.internal_static_tensorflow_SavedSliceMeta_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.util.SavedTensorSliceProtos.internal_static_tensorflow_SavedSliceMeta_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.util.SavedSliceMeta.class, org.tensorflow.proto.util.SavedSliceMeta.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.util.SavedSliceMeta.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+        getSliceFieldBuilder();
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      name_ = "";
+
+      if (shapeBuilder_ == null) {
+        shape_ = null;
+      } else {
+        shape_ = null;
+        shapeBuilder_ = null;
+      }
+      type_ = 0;
+
+      if (sliceBuilder_ == null) {
+        slice_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000001);
+      } else {
+        sliceBuilder_.clear();
+      }
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.util.SavedTensorSliceProtos.internal_static_tensorflow_SavedSliceMeta_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.SavedSliceMeta getDefaultInstanceForType() {
+      return org.tensorflow.proto.util.SavedSliceMeta.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.SavedSliceMeta build() {
+      org.tensorflow.proto.util.SavedSliceMeta result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.SavedSliceMeta buildPartial() {
+      org.tensorflow.proto.util.SavedSliceMeta result = new org.tensorflow.proto.util.SavedSliceMeta(this);
+      int from_bitField0_ = bitField0_;
+      result.name_ = name_;
+      if (shapeBuilder_ == null) {
+        result.shape_ = shape_;
+      } else {
+        result.shape_ = shapeBuilder_.build();
+      }
+      result.type_ = type_;
+      if (sliceBuilder_ == null) {
+        if (((bitField0_ & 0x00000001) != 0)) {
+          slice_ = java.util.Collections.unmodifiableList(slice_);
+          bitField0_ = (bitField0_ & ~0x00000001);
+        }
+        result.slice_ = slice_;
+      } else {
+        result.slice_ = sliceBuilder_.build();
+      }
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.util.SavedSliceMeta) {
+        return mergeFrom((org.tensorflow.proto.util.SavedSliceMeta)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.util.SavedSliceMeta other) {
+      if (other == org.tensorflow.proto.util.SavedSliceMeta.getDefaultInstance()) return this;
+      if (!other.getName().isEmpty()) {
+        name_ = other.name_;
+        onChanged();
+      }
+      if (other.hasShape()) {
+        mergeShape(other.getShape());
+      }
+      if (other.type_ != 0) {
+        setTypeValue(other.getTypeValue());
+      }
+      if (sliceBuilder_ == null) {
+        if (!other.slice_.isEmpty()) {
+          if (slice_.isEmpty()) {
+            slice_ = other.slice_;
+            bitField0_ = (bitField0_ & ~0x00000001);
+          } else {
+            ensureSliceIsMutable();
+            slice_.addAll(other.slice_);
+          }
+          onChanged();
+        }
+      } else {
+        if (!other.slice_.isEmpty()) {
+          if (sliceBuilder_.isEmpty()) {
+            sliceBuilder_.dispose();
+            sliceBuilder_ = null;
+            slice_ = other.slice_;
+            bitField0_ = (bitField0_ & ~0x00000001);
+            sliceBuilder_ = 
+              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
+                 getSliceFieldBuilder() : null;
+          } else {
+            sliceBuilder_.addAllMessages(other.slice_);
+          }
+        }
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.util.SavedSliceMeta parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.util.SavedSliceMeta) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+    private int bitField0_;
+
+    private java.lang.Object name_ = "";
+    /**
+     * <pre>
+     * Name of the tensor.
+     * </pre>
+     *
+     * <code>string name = 1;</code>
+     */
+    public java.lang.String getName() {
+      java.lang.Object ref = name_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        name_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Name of the tensor.
+     * </pre>
+     *
+     * <code>string name = 1;</code>
+     */
+    public com.google.protobuf.ByteString
+        getNameBytes() {
+      java.lang.Object ref = name_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        name_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Name of the tensor.
+     * </pre>
+     *
+     * <code>string name = 1;</code>
+     */
+    public Builder setName(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      name_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Name of the tensor.
+     * </pre>
+     *
+     * <code>string name = 1;</code>
+     */
+    public Builder clearName() {
+      
+      name_ = getDefaultInstance().getName();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Name of the tensor.
+     * </pre>
+     *
+     * <code>string name = 1;</code>
+     */
+    public Builder setNameBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      name_ = value;
+      onChanged();
+      return this;
+    }
+
+    private org.tensorflow.proto.framework.TensorShapeProto shape_;
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.TensorShapeProto, org.tensorflow.proto.framework.TensorShapeProto.Builder, org.tensorflow.proto.framework.TensorShapeProtoOrBuilder> shapeBuilder_;
+    /**
+     * <pre>
+     * Shape of the tensor
+     * </pre>
+     *
+     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+     */
+    public boolean hasShape() {
+      return shapeBuilder_ != null || shape_ != null;
+    }
+    /**
+     * <pre>
+     * Shape of the tensor
+     * </pre>
+     *
+     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+     */
+    public org.tensorflow.proto.framework.TensorShapeProto getShape() {
+      if (shapeBuilder_ == null) {
+        return shape_ == null ? org.tensorflow.proto.framework.TensorShapeProto.getDefaultInstance() : shape_;
+      } else {
+        return shapeBuilder_.getMessage();
+      }
+    }
+    /**
+     * <pre>
+     * Shape of the tensor
+     * </pre>
+     *
+     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+     */
+    public Builder setShape(org.tensorflow.proto.framework.TensorShapeProto value) {
+      if (shapeBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        shape_ = value;
+        onChanged();
+      } else {
+        shapeBuilder_.setMessage(value);
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Shape of the tensor
+     * </pre>
+     *
+     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+     */
+    public Builder setShape(
+        org.tensorflow.proto.framework.TensorShapeProto.Builder builderForValue) {
+      if (shapeBuilder_ == null) {
+        shape_ = builderForValue.build();
+        onChanged();
+      } else {
+        shapeBuilder_.setMessage(builderForValue.build());
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Shape of the tensor
+     * </pre>
+     *
+     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+     */
+    public Builder mergeShape(org.tensorflow.proto.framework.TensorShapeProto value) {
+      if (shapeBuilder_ == null) {
+        if (shape_ != null) {
+          shape_ =
+            org.tensorflow.proto.framework.TensorShapeProto.newBuilder(shape_).mergeFrom(value).buildPartial();
+        } else {
+          shape_ = value;
+        }
+        onChanged();
+      } else {
+        shapeBuilder_.mergeFrom(value);
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Shape of the tensor
+     * </pre>
+     *
+     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+     */
+    public Builder clearShape() {
+      if (shapeBuilder_ == null) {
+        shape_ = null;
+        onChanged();
+      } else {
+        shape_ = null;
+        shapeBuilder_ = null;
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Shape of the tensor
+     * </pre>
+     *
+     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+     */
+    public org.tensorflow.proto.framework.TensorShapeProto.Builder getShapeBuilder() {
+      
+      onChanged();
+      return getShapeFieldBuilder().getBuilder();
+    }
+    /**
+     * <pre>
+     * Shape of the tensor
+     * </pre>
+     *
+     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+     */
+    public org.tensorflow.proto.framework.TensorShapeProtoOrBuilder getShapeOrBuilder() {
+      if (shapeBuilder_ != null) {
+        return shapeBuilder_.getMessageOrBuilder();
+      } else {
+        return shape_ == null ?
+            org.tensorflow.proto.framework.TensorShapeProto.getDefaultInstance() : shape_;
+      }
+    }
+    /**
+     * <pre>
+     * Shape of the tensor
+     * </pre>
+     *
+     * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+     */
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.TensorShapeProto, org.tensorflow.proto.framework.TensorShapeProto.Builder, org.tensorflow.proto.framework.TensorShapeProtoOrBuilder> 
+        getShapeFieldBuilder() {
+      if (shapeBuilder_ == null) {
+        shapeBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+            org.tensorflow.proto.framework.TensorShapeProto, org.tensorflow.proto.framework.TensorShapeProto.Builder, org.tensorflow.proto.framework.TensorShapeProtoOrBuilder>(
+                getShape(),
+                getParentForChildren(),
+                isClean());
+        shape_ = null;
+      }
+      return shapeBuilder_;
+    }
+
+    private int type_ = 0;
+    /**
+     * <pre>
+     * Type of the tensor
+     * </pre>
+     *
+     * <code>.tensorflow.DataType type = 3;</code>
+     */
+    public int getTypeValue() {
+      return type_;
+    }
+    /**
+     * <pre>
+     * Type of the tensor
+     * </pre>
+     *
+     * <code>.tensorflow.DataType type = 3;</code>
+     */
+    public Builder setTypeValue(int value) {
+      type_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Type of the tensor
+     * </pre>
+     *
+     * <code>.tensorflow.DataType type = 3;</code>
+     */
+    public org.tensorflow.proto.framework.DataType getType() {
+      @SuppressWarnings("deprecation")
+      org.tensorflow.proto.framework.DataType result = org.tensorflow.proto.framework.DataType.valueOf(type_);
+      return result == null ? org.tensorflow.proto.framework.DataType.UNRECOGNIZED : result;
+    }
+    /**
+     * <pre>
+     * Type of the tensor
+     * </pre>
+     *
+     * <code>.tensorflow.DataType type = 3;</code>
+     */
+    public Builder setType(org.tensorflow.proto.framework.DataType value) {
+      if (value == null) {
+        throw new NullPointerException();
+      }
+      
+      type_ = value.getNumber();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Type of the tensor
+     * </pre>
+     *
+     * <code>.tensorflow.DataType type = 3;</code>
+     */
+    public Builder clearType() {
+      
+      type_ = 0;
+      onChanged();
+      return this;
+    }
+
+    private java.util.List<org.tensorflow.proto.framework.TensorSliceProto> slice_ =
+      java.util.Collections.emptyList();
+    private void ensureSliceIsMutable() {
+      if (!((bitField0_ & 0x00000001) != 0)) {
+        slice_ = new java.util.ArrayList<org.tensorflow.proto.framework.TensorSliceProto>(slice_);
+        bitField0_ |= 0x00000001;
+       }
+    }
+
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.framework.TensorSliceProto, org.tensorflow.proto.framework.TensorSliceProto.Builder, org.tensorflow.proto.framework.TensorSliceProtoOrBuilder> sliceBuilder_;
+
+    /**
+     * <pre>
+     * Explicit list of slices saved in the checkpoint file.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorSliceProto slice = 4;</code>
+     */
+    public java.util.List<org.tensorflow.proto.framework.TensorSliceProto> getSliceList() {
+      if (sliceBuilder_ == null) {
+        return java.util.Collections.unmodifiableList(slice_);
+      } else {
+        return sliceBuilder_.getMessageList();
+      }
+    }
+    /**
+     * <pre>
+     * Explicit list of slices saved in the checkpoint file.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorSliceProto slice = 4;</code>
+     */
+    public int getSliceCount() {
+      if (sliceBuilder_ == null) {
+        return slice_.size();
+      } else {
+        return sliceBuilder_.getCount();
+      }
+    }
+    /**
+     * <pre>
+     * Explicit list of slices saved in the checkpoint file.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorSliceProto slice = 4;</code>
+     */
+    public org.tensorflow.proto.framework.TensorSliceProto getSlice(int index) {
+      if (sliceBuilder_ == null) {
+        return slice_.get(index);
+      } else {
+        return sliceBuilder_.getMessage(index);
+      }
+    }
+    /**
+     * <pre>
+     * Explicit list of slices saved in the checkpoint file.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorSliceProto slice = 4;</code>
+     */
+    public Builder setSlice(
+        int index, org.tensorflow.proto.framework.TensorSliceProto value) {
+      if (sliceBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureSliceIsMutable();
+        slice_.set(index, value);
+        onChanged();
+      } else {
+        sliceBuilder_.setMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Explicit list of slices saved in the checkpoint file.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorSliceProto slice = 4;</code>
+     */
+    public Builder setSlice(
+        int index, org.tensorflow.proto.framework.TensorSliceProto.Builder builderForValue) {
+      if (sliceBuilder_ == null) {
+        ensureSliceIsMutable();
+        slice_.set(index, builderForValue.build());
+        onChanged();
+      } else {
+        sliceBuilder_.setMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Explicit list of slices saved in the checkpoint file.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorSliceProto slice = 4;</code>
+     */
+    public Builder addSlice(org.tensorflow.proto.framework.TensorSliceProto value) {
+      if (sliceBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureSliceIsMutable();
+        slice_.add(value);
+        onChanged();
+      } else {
+        sliceBuilder_.addMessage(value);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Explicit list of slices saved in the checkpoint file.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorSliceProto slice = 4;</code>
+     */
+    public Builder addSlice(
+        int index, org.tensorflow.proto.framework.TensorSliceProto value) {
+      if (sliceBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureSliceIsMutable();
+        slice_.add(index, value);
+        onChanged();
+      } else {
+        sliceBuilder_.addMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Explicit list of slices saved in the checkpoint file.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorSliceProto slice = 4;</code>
+     */
+    public Builder addSlice(
+        org.tensorflow.proto.framework.TensorSliceProto.Builder builderForValue) {
+      if (sliceBuilder_ == null) {
+        ensureSliceIsMutable();
+        slice_.add(builderForValue.build());
+        onChanged();
+      } else {
+        sliceBuilder_.addMessage(builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Explicit list of slices saved in the checkpoint file.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorSliceProto slice = 4;</code>
+     */
+    public Builder addSlice(
+        int index, org.tensorflow.proto.framework.TensorSliceProto.Builder builderForValue) {
+      if (sliceBuilder_ == null) {
+        ensureSliceIsMutable();
+        slice_.add(index, builderForValue.build());
+        onChanged();
+      } else {
+        sliceBuilder_.addMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Explicit list of slices saved in the checkpoint file.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorSliceProto slice = 4;</code>
+     */
+    public Builder addAllSlice(
+        java.lang.Iterable<? extends org.tensorflow.proto.framework.TensorSliceProto> values) {
+      if (sliceBuilder_ == null) {
+        ensureSliceIsMutable();
+        com.google.protobuf.AbstractMessageLite.Builder.addAll(
+            values, slice_);
+        onChanged();
+      } else {
+        sliceBuilder_.addAllMessages(values);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Explicit list of slices saved in the checkpoint file.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorSliceProto slice = 4;</code>
+     */
+    public Builder clearSlice() {
+      if (sliceBuilder_ == null) {
+        slice_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000001);
+        onChanged();
+      } else {
+        sliceBuilder_.clear();
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Explicit list of slices saved in the checkpoint file.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorSliceProto slice = 4;</code>
+     */
+    public Builder removeSlice(int index) {
+      if (sliceBuilder_ == null) {
+        ensureSliceIsMutable();
+        slice_.remove(index);
+        onChanged();
+      } else {
+        sliceBuilder_.remove(index);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Explicit list of slices saved in the checkpoint file.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorSliceProto slice = 4;</code>
+     */
+    public org.tensorflow.proto.framework.TensorSliceProto.Builder getSliceBuilder(
+        int index) {
+      return getSliceFieldBuilder().getBuilder(index);
+    }
+    /**
+     * <pre>
+     * Explicit list of slices saved in the checkpoint file.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorSliceProto slice = 4;</code>
+     */
+    public org.tensorflow.proto.framework.TensorSliceProtoOrBuilder getSliceOrBuilder(
+        int index) {
+      if (sliceBuilder_ == null) {
+        return slice_.get(index);  } else {
+        return sliceBuilder_.getMessageOrBuilder(index);
+      }
+    }
+    /**
+     * <pre>
+     * Explicit list of slices saved in the checkpoint file.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorSliceProto slice = 4;</code>
+     */
+    public java.util.List<? extends org.tensorflow.proto.framework.TensorSliceProtoOrBuilder> 
+         getSliceOrBuilderList() {
+      if (sliceBuilder_ != null) {
+        return sliceBuilder_.getMessageOrBuilderList();
+      } else {
+        return java.util.Collections.unmodifiableList(slice_);
+      }
+    }
+    /**
+     * <pre>
+     * Explicit list of slices saved in the checkpoint file.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorSliceProto slice = 4;</code>
+     */
+    public org.tensorflow.proto.framework.TensorSliceProto.Builder addSliceBuilder() {
+      return getSliceFieldBuilder().addBuilder(
+          org.tensorflow.proto.framework.TensorSliceProto.getDefaultInstance());
+    }
+    /**
+     * <pre>
+     * Explicit list of slices saved in the checkpoint file.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorSliceProto slice = 4;</code>
+     */
+    public org.tensorflow.proto.framework.TensorSliceProto.Builder addSliceBuilder(
+        int index) {
+      return getSliceFieldBuilder().addBuilder(
+          index, org.tensorflow.proto.framework.TensorSliceProto.getDefaultInstance());
+    }
+    /**
+     * <pre>
+     * Explicit list of slices saved in the checkpoint file.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.TensorSliceProto slice = 4;</code>
+     */
+    public java.util.List<org.tensorflow.proto.framework.TensorSliceProto.Builder> 
+         getSliceBuilderList() {
+      return getSliceFieldBuilder().getBuilderList();
+    }
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.framework.TensorSliceProto, org.tensorflow.proto.framework.TensorSliceProto.Builder, org.tensorflow.proto.framework.TensorSliceProtoOrBuilder> 
+        getSliceFieldBuilder() {
+      if (sliceBuilder_ == null) {
+        sliceBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
+            org.tensorflow.proto.framework.TensorSliceProto, org.tensorflow.proto.framework.TensorSliceProto.Builder, org.tensorflow.proto.framework.TensorSliceProtoOrBuilder>(
+                slice_,
+                ((bitField0_ & 0x00000001) != 0),
+                getParentForChildren(),
+                isClean());
+        slice_ = null;
+      }
+      return sliceBuilder_;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.SavedSliceMeta)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.SavedSliceMeta)
+  private static final org.tensorflow.proto.util.SavedSliceMeta DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.util.SavedSliceMeta();
+  }
+
+  public static org.tensorflow.proto.util.SavedSliceMeta getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<SavedSliceMeta>
+      PARSER = new com.google.protobuf.AbstractParser<SavedSliceMeta>() {
+    @java.lang.Override
+    public SavedSliceMeta parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new SavedSliceMeta(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<SavedSliceMeta> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<SavedSliceMeta> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.util.SavedSliceMeta getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SavedSliceMetaOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SavedSliceMetaOrBuilder.java
new file mode 100644
index 00000000000..76aff4e7eb5
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SavedSliceMetaOrBuilder.java
@@ -0,0 +1,113 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/util/saved_tensor_slice.proto
+
+package org.tensorflow.proto.util;
+
+public interface SavedSliceMetaOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.SavedSliceMeta)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <pre>
+   * Name of the tensor.
+   * </pre>
+   *
+   * <code>string name = 1;</code>
+   */
+  java.lang.String getName();
+  /**
+   * <pre>
+   * Name of the tensor.
+   * </pre>
+   *
+   * <code>string name = 1;</code>
+   */
+  com.google.protobuf.ByteString
+      getNameBytes();
+
+  /**
+   * <pre>
+   * Shape of the tensor
+   * </pre>
+   *
+   * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+   */
+  boolean hasShape();
+  /**
+   * <pre>
+   * Shape of the tensor
+   * </pre>
+   *
+   * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+   */
+  org.tensorflow.proto.framework.TensorShapeProto getShape();
+  /**
+   * <pre>
+   * Shape of the tensor
+   * </pre>
+   *
+   * <code>.tensorflow.TensorShapeProto shape = 2;</code>
+   */
+  org.tensorflow.proto.framework.TensorShapeProtoOrBuilder getShapeOrBuilder();
+
+  /**
+   * <pre>
+   * Type of the tensor
+   * </pre>
+   *
+   * <code>.tensorflow.DataType type = 3;</code>
+   */
+  int getTypeValue();
+  /**
+   * <pre>
+   * Type of the tensor
+   * </pre>
+   *
+   * <code>.tensorflow.DataType type = 3;</code>
+   */
+  org.tensorflow.proto.framework.DataType getType();
+
+  /**
+   * <pre>
+   * Explicit list of slices saved in the checkpoint file.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.TensorSliceProto slice = 4;</code>
+   */
+  java.util.List<org.tensorflow.proto.framework.TensorSliceProto> 
+      getSliceList();
+  /**
+   * <pre>
+   * Explicit list of slices saved in the checkpoint file.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.TensorSliceProto slice = 4;</code>
+   */
+  org.tensorflow.proto.framework.TensorSliceProto getSlice(int index);
+  /**
+   * <pre>
+   * Explicit list of slices saved in the checkpoint file.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.TensorSliceProto slice = 4;</code>
+   */
+  int getSliceCount();
+  /**
+   * <pre>
+   * Explicit list of slices saved in the checkpoint file.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.TensorSliceProto slice = 4;</code>
+   */
+  java.util.List<? extends org.tensorflow.proto.framework.TensorSliceProtoOrBuilder> 
+      getSliceOrBuilderList();
+  /**
+   * <pre>
+   * Explicit list of slices saved in the checkpoint file.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.TensorSliceProto slice = 4;</code>
+   */
+  org.tensorflow.proto.framework.TensorSliceProtoOrBuilder getSliceOrBuilder(
+      int index);
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SavedSliceOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SavedSliceOrBuilder.java
new file mode 100644
index 00000000000..bf76430d534
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SavedSliceOrBuilder.java
@@ -0,0 +1,85 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/util/saved_tensor_slice.proto
+
+package org.tensorflow.proto.util;
+
+public interface SavedSliceOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.SavedSlice)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <pre>
+   * Name of the tensor that this slice belongs to. This must be identical to
+   * the name used to encode the key for this record.
+   * </pre>
+   *
+   * <code>string name = 1;</code>
+   */
+  java.lang.String getName();
+  /**
+   * <pre>
+   * Name of the tensor that this slice belongs to. This must be identical to
+   * the name used to encode the key for this record.
+   * </pre>
+   *
+   * <code>string name = 1;</code>
+   */
+  com.google.protobuf.ByteString
+      getNameBytes();
+
+  /**
+   * <pre>
+   * Extent of the slice.  Must have one entry for each of the dimension of the
+   * tensor that this slice belongs to.
+   * </pre>
+   *
+   * <code>.tensorflow.TensorSliceProto slice = 2;</code>
+   */
+  boolean hasSlice();
+  /**
+   * <pre>
+   * Extent of the slice.  Must have one entry for each of the dimension of the
+   * tensor that this slice belongs to.
+   * </pre>
+   *
+   * <code>.tensorflow.TensorSliceProto slice = 2;</code>
+   */
+  org.tensorflow.proto.framework.TensorSliceProto getSlice();
+  /**
+   * <pre>
+   * Extent of the slice.  Must have one entry for each of the dimension of the
+   * tensor that this slice belongs to.
+   * </pre>
+   *
+   * <code>.tensorflow.TensorSliceProto slice = 2;</code>
+   */
+  org.tensorflow.proto.framework.TensorSliceProtoOrBuilder getSliceOrBuilder();
+
+  /**
+   * <pre>
+   * The raw data of the slice is stored as a TensorProto. Only raw data are
+   * stored (we don't fill in fields such as dtype or tensor_shape).
+   * </pre>
+   *
+   * <code>.tensorflow.TensorProto data = 3;</code>
+   */
+  boolean hasData();
+  /**
+   * <pre>
+   * The raw data of the slice is stored as a TensorProto. Only raw data are
+   * stored (we don't fill in fields such as dtype or tensor_shape).
+   * </pre>
+   *
+   * <code>.tensorflow.TensorProto data = 3;</code>
+   */
+  org.tensorflow.proto.framework.TensorProto getData();
+  /**
+   * <pre>
+   * The raw data of the slice is stored as a TensorProto. Only raw data are
+   * stored (we don't fill in fields such as dtype or tensor_shape).
+   * </pre>
+   *
+   * <code>.tensorflow.TensorProto data = 3;</code>
+   */
+  org.tensorflow.proto.framework.TensorProtoOrBuilder getDataOrBuilder();
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SavedTensorSliceMeta.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SavedTensorSliceMeta.java
new file mode 100644
index 00000000000..6217e7ed403
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SavedTensorSliceMeta.java
@@ -0,0 +1,1108 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/util/saved_tensor_slice.proto
+
+package org.tensorflow.proto.util;
+
+/**
+ * <pre>
+ * Metadata describing the set of tensor slices saved in a checkpoint file.
+ * It is always stored at the beginning of each checkpoint file.
+ * </pre>
+ *
+ * Protobuf type {@code tensorflow.SavedTensorSliceMeta}
+ */
+public  final class SavedTensorSliceMeta extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.SavedTensorSliceMeta)
+    SavedTensorSliceMetaOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use SavedTensorSliceMeta.newBuilder() to construct.
+  private SavedTensorSliceMeta(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private SavedTensorSliceMeta() {
+    tensor_ = java.util.Collections.emptyList();
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new SavedTensorSliceMeta();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private SavedTensorSliceMeta(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    int mutable_bitField0_ = 0;
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 10: {
+            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
+              tensor_ = new java.util.ArrayList<org.tensorflow.proto.util.SavedSliceMeta>();
+              mutable_bitField0_ |= 0x00000001;
+            }
+            tensor_.add(
+                input.readMessage(org.tensorflow.proto.util.SavedSliceMeta.parser(), extensionRegistry));
+            break;
+          }
+          case 18: {
+            org.tensorflow.proto.framework.VersionDef.Builder subBuilder = null;
+            if (versions_ != null) {
+              subBuilder = versions_.toBuilder();
+            }
+            versions_ = input.readMessage(org.tensorflow.proto.framework.VersionDef.parser(), extensionRegistry);
+            if (subBuilder != null) {
+              subBuilder.mergeFrom(versions_);
+              versions_ = subBuilder.buildPartial();
+            }
+
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      if (((mutable_bitField0_ & 0x00000001) != 0)) {
+        tensor_ = java.util.Collections.unmodifiableList(tensor_);
+      }
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.util.SavedTensorSliceProtos.internal_static_tensorflow_SavedTensorSliceMeta_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.util.SavedTensorSliceProtos.internal_static_tensorflow_SavedTensorSliceMeta_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.util.SavedTensorSliceMeta.class, org.tensorflow.proto.util.SavedTensorSliceMeta.Builder.class);
+  }
+
+  public static final int TENSOR_FIELD_NUMBER = 1;
+  private java.util.List<org.tensorflow.proto.util.SavedSliceMeta> tensor_;
+  /**
+   * <pre>
+   * Each SavedSliceMeta describes the slices for one tensor.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.SavedSliceMeta tensor = 1;</code>
+   */
+  public java.util.List<org.tensorflow.proto.util.SavedSliceMeta> getTensorList() {
+    return tensor_;
+  }
+  /**
+   * <pre>
+   * Each SavedSliceMeta describes the slices for one tensor.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.SavedSliceMeta tensor = 1;</code>
+   */
+  public java.util.List<? extends org.tensorflow.proto.util.SavedSliceMetaOrBuilder> 
+      getTensorOrBuilderList() {
+    return tensor_;
+  }
+  /**
+   * <pre>
+   * Each SavedSliceMeta describes the slices for one tensor.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.SavedSliceMeta tensor = 1;</code>
+   */
+  public int getTensorCount() {
+    return tensor_.size();
+  }
+  /**
+   * <pre>
+   * Each SavedSliceMeta describes the slices for one tensor.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.SavedSliceMeta tensor = 1;</code>
+   */
+  public org.tensorflow.proto.util.SavedSliceMeta getTensor(int index) {
+    return tensor_.get(index);
+  }
+  /**
+   * <pre>
+   * Each SavedSliceMeta describes the slices for one tensor.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.SavedSliceMeta tensor = 1;</code>
+   */
+  public org.tensorflow.proto.util.SavedSliceMetaOrBuilder getTensorOrBuilder(
+      int index) {
+    return tensor_.get(index);
+  }
+
+  public static final int VERSIONS_FIELD_NUMBER = 2;
+  private org.tensorflow.proto.framework.VersionDef versions_;
+  /**
+   * <pre>
+   * Compatibility version of this checkpoint.  See core/public/version.h
+   * for version history.
+   * </pre>
+   *
+   * <code>.tensorflow.VersionDef versions = 2;</code>
+   */
+  public boolean hasVersions() {
+    return versions_ != null;
+  }
+  /**
+   * <pre>
+   * Compatibility version of this checkpoint.  See core/public/version.h
+   * for version history.
+   * </pre>
+   *
+   * <code>.tensorflow.VersionDef versions = 2;</code>
+   */
+  public org.tensorflow.proto.framework.VersionDef getVersions() {
+    return versions_ == null ? org.tensorflow.proto.framework.VersionDef.getDefaultInstance() : versions_;
+  }
+  /**
+   * <pre>
+   * Compatibility version of this checkpoint.  See core/public/version.h
+   * for version history.
+   * </pre>
+   *
+   * <code>.tensorflow.VersionDef versions = 2;</code>
+   */
+  public org.tensorflow.proto.framework.VersionDefOrBuilder getVersionsOrBuilder() {
+    return getVersions();
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    for (int i = 0; i < tensor_.size(); i++) {
+      output.writeMessage(1, tensor_.get(i));
+    }
+    if (versions_ != null) {
+      output.writeMessage(2, getVersions());
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    for (int i = 0; i < tensor_.size(); i++) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(1, tensor_.get(i));
+    }
+    if (versions_ != null) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(2, getVersions());
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.util.SavedTensorSliceMeta)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.util.SavedTensorSliceMeta other = (org.tensorflow.proto.util.SavedTensorSliceMeta) obj;
+
+    if (!getTensorList()
+        .equals(other.getTensorList())) return false;
+    if (hasVersions() != other.hasVersions()) return false;
+    if (hasVersions()) {
+      if (!getVersions()
+          .equals(other.getVersions())) return false;
+    }
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    if (getTensorCount() > 0) {
+      hash = (37 * hash) + TENSOR_FIELD_NUMBER;
+      hash = (53 * hash) + getTensorList().hashCode();
+    }
+    if (hasVersions()) {
+      hash = (37 * hash) + VERSIONS_FIELD_NUMBER;
+      hash = (53 * hash) + getVersions().hashCode();
+    }
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.util.SavedTensorSliceMeta parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.SavedTensorSliceMeta parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.SavedTensorSliceMeta parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.SavedTensorSliceMeta parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.SavedTensorSliceMeta parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.SavedTensorSliceMeta parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.SavedTensorSliceMeta parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.SavedTensorSliceMeta parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.SavedTensorSliceMeta parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.SavedTensorSliceMeta parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.SavedTensorSliceMeta parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.SavedTensorSliceMeta parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.util.SavedTensorSliceMeta prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * <pre>
+   * Metadata describing the set of tensor slices saved in a checkpoint file.
+   * It is always stored at the beginning of each checkpoint file.
+   * </pre>
+   *
+   * Protobuf type {@code tensorflow.SavedTensorSliceMeta}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.SavedTensorSliceMeta)
+      org.tensorflow.proto.util.SavedTensorSliceMetaOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.util.SavedTensorSliceProtos.internal_static_tensorflow_SavedTensorSliceMeta_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.util.SavedTensorSliceProtos.internal_static_tensorflow_SavedTensorSliceMeta_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.util.SavedTensorSliceMeta.class, org.tensorflow.proto.util.SavedTensorSliceMeta.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.util.SavedTensorSliceMeta.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+        getTensorFieldBuilder();
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      if (tensorBuilder_ == null) {
+        tensor_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000001);
+      } else {
+        tensorBuilder_.clear();
+      }
+      if (versionsBuilder_ == null) {
+        versions_ = null;
+      } else {
+        versions_ = null;
+        versionsBuilder_ = null;
+      }
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.util.SavedTensorSliceProtos.internal_static_tensorflow_SavedTensorSliceMeta_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.SavedTensorSliceMeta getDefaultInstanceForType() {
+      return org.tensorflow.proto.util.SavedTensorSliceMeta.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.SavedTensorSliceMeta build() {
+      org.tensorflow.proto.util.SavedTensorSliceMeta result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.SavedTensorSliceMeta buildPartial() {
+      org.tensorflow.proto.util.SavedTensorSliceMeta result = new org.tensorflow.proto.util.SavedTensorSliceMeta(this);
+      int from_bitField0_ = bitField0_;
+      if (tensorBuilder_ == null) {
+        if (((bitField0_ & 0x00000001) != 0)) {
+          tensor_ = java.util.Collections.unmodifiableList(tensor_);
+          bitField0_ = (bitField0_ & ~0x00000001);
+        }
+        result.tensor_ = tensor_;
+      } else {
+        result.tensor_ = tensorBuilder_.build();
+      }
+      if (versionsBuilder_ == null) {
+        result.versions_ = versions_;
+      } else {
+        result.versions_ = versionsBuilder_.build();
+      }
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.util.SavedTensorSliceMeta) {
+        return mergeFrom((org.tensorflow.proto.util.SavedTensorSliceMeta)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.util.SavedTensorSliceMeta other) {
+      if (other == org.tensorflow.proto.util.SavedTensorSliceMeta.getDefaultInstance()) return this;
+      if (tensorBuilder_ == null) {
+        if (!other.tensor_.isEmpty()) {
+          if (tensor_.isEmpty()) {
+            tensor_ = other.tensor_;
+            bitField0_ = (bitField0_ & ~0x00000001);
+          } else {
+            ensureTensorIsMutable();
+            tensor_.addAll(other.tensor_);
+          }
+          onChanged();
+        }
+      } else {
+        if (!other.tensor_.isEmpty()) {
+          if (tensorBuilder_.isEmpty()) {
+            tensorBuilder_.dispose();
+            tensorBuilder_ = null;
+            tensor_ = other.tensor_;
+            bitField0_ = (bitField0_ & ~0x00000001);
+            tensorBuilder_ = 
+              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
+                 getTensorFieldBuilder() : null;
+          } else {
+            tensorBuilder_.addAllMessages(other.tensor_);
+          }
+        }
+      }
+      if (other.hasVersions()) {
+        mergeVersions(other.getVersions());
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.util.SavedTensorSliceMeta parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.util.SavedTensorSliceMeta) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+    private int bitField0_;
+
+    private java.util.List<org.tensorflow.proto.util.SavedSliceMeta> tensor_ =
+      java.util.Collections.emptyList();
+    private void ensureTensorIsMutable() {
+      if (!((bitField0_ & 0x00000001) != 0)) {
+        tensor_ = new java.util.ArrayList<org.tensorflow.proto.util.SavedSliceMeta>(tensor_);
+        bitField0_ |= 0x00000001;
+       }
+    }
+
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.util.SavedSliceMeta, org.tensorflow.proto.util.SavedSliceMeta.Builder, org.tensorflow.proto.util.SavedSliceMetaOrBuilder> tensorBuilder_;
+
+    /**
+     * <pre>
+     * Each SavedSliceMeta describes the slices for one tensor.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.SavedSliceMeta tensor = 1;</code>
+     */
+    public java.util.List<org.tensorflow.proto.util.SavedSliceMeta> getTensorList() {
+      if (tensorBuilder_ == null) {
+        return java.util.Collections.unmodifiableList(tensor_);
+      } else {
+        return tensorBuilder_.getMessageList();
+      }
+    }
+    /**
+     * <pre>
+     * Each SavedSliceMeta describes the slices for one tensor.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.SavedSliceMeta tensor = 1;</code>
+     */
+    public int getTensorCount() {
+      if (tensorBuilder_ == null) {
+        return tensor_.size();
+      } else {
+        return tensorBuilder_.getCount();
+      }
+    }
+    /**
+     * <pre>
+     * Each SavedSliceMeta describes the slices for one tensor.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.SavedSliceMeta tensor = 1;</code>
+     */
+    public org.tensorflow.proto.util.SavedSliceMeta getTensor(int index) {
+      if (tensorBuilder_ == null) {
+        return tensor_.get(index);
+      } else {
+        return tensorBuilder_.getMessage(index);
+      }
+    }
+    /**
+     * <pre>
+     * Each SavedSliceMeta describes the slices for one tensor.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.SavedSliceMeta tensor = 1;</code>
+     */
+    public Builder setTensor(
+        int index, org.tensorflow.proto.util.SavedSliceMeta value) {
+      if (tensorBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureTensorIsMutable();
+        tensor_.set(index, value);
+        onChanged();
+      } else {
+        tensorBuilder_.setMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Each SavedSliceMeta describes the slices for one tensor.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.SavedSliceMeta tensor = 1;</code>
+     */
+    public Builder setTensor(
+        int index, org.tensorflow.proto.util.SavedSliceMeta.Builder builderForValue) {
+      if (tensorBuilder_ == null) {
+        ensureTensorIsMutable();
+        tensor_.set(index, builderForValue.build());
+        onChanged();
+      } else {
+        tensorBuilder_.setMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Each SavedSliceMeta describes the slices for one tensor.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.SavedSliceMeta tensor = 1;</code>
+     */
+    public Builder addTensor(org.tensorflow.proto.util.SavedSliceMeta value) {
+      if (tensorBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureTensorIsMutable();
+        tensor_.add(value);
+        onChanged();
+      } else {
+        tensorBuilder_.addMessage(value);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Each SavedSliceMeta describes the slices for one tensor.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.SavedSliceMeta tensor = 1;</code>
+     */
+    public Builder addTensor(
+        int index, org.tensorflow.proto.util.SavedSliceMeta value) {
+      if (tensorBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureTensorIsMutable();
+        tensor_.add(index, value);
+        onChanged();
+      } else {
+        tensorBuilder_.addMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Each SavedSliceMeta describes the slices for one tensor.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.SavedSliceMeta tensor = 1;</code>
+     */
+    public Builder addTensor(
+        org.tensorflow.proto.util.SavedSliceMeta.Builder builderForValue) {
+      if (tensorBuilder_ == null) {
+        ensureTensorIsMutable();
+        tensor_.add(builderForValue.build());
+        onChanged();
+      } else {
+        tensorBuilder_.addMessage(builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Each SavedSliceMeta describes the slices for one tensor.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.SavedSliceMeta tensor = 1;</code>
+     */
+    public Builder addTensor(
+        int index, org.tensorflow.proto.util.SavedSliceMeta.Builder builderForValue) {
+      if (tensorBuilder_ == null) {
+        ensureTensorIsMutable();
+        tensor_.add(index, builderForValue.build());
+        onChanged();
+      } else {
+        tensorBuilder_.addMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Each SavedSliceMeta describes the slices for one tensor.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.SavedSliceMeta tensor = 1;</code>
+     */
+    public Builder addAllTensor(
+        java.lang.Iterable<? extends org.tensorflow.proto.util.SavedSliceMeta> values) {
+      if (tensorBuilder_ == null) {
+        ensureTensorIsMutable();
+        com.google.protobuf.AbstractMessageLite.Builder.addAll(
+            values, tensor_);
+        onChanged();
+      } else {
+        tensorBuilder_.addAllMessages(values);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Each SavedSliceMeta describes the slices for one tensor.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.SavedSliceMeta tensor = 1;</code>
+     */
+    public Builder clearTensor() {
+      if (tensorBuilder_ == null) {
+        tensor_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000001);
+        onChanged();
+      } else {
+        tensorBuilder_.clear();
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Each SavedSliceMeta describes the slices for one tensor.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.SavedSliceMeta tensor = 1;</code>
+     */
+    public Builder removeTensor(int index) {
+      if (tensorBuilder_ == null) {
+        ensureTensorIsMutable();
+        tensor_.remove(index);
+        onChanged();
+      } else {
+        tensorBuilder_.remove(index);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Each SavedSliceMeta describes the slices for one tensor.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.SavedSliceMeta tensor = 1;</code>
+     */
+    public org.tensorflow.proto.util.SavedSliceMeta.Builder getTensorBuilder(
+        int index) {
+      return getTensorFieldBuilder().getBuilder(index);
+    }
+    /**
+     * <pre>
+     * Each SavedSliceMeta describes the slices for one tensor.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.SavedSliceMeta tensor = 1;</code>
+     */
+    public org.tensorflow.proto.util.SavedSliceMetaOrBuilder getTensorOrBuilder(
+        int index) {
+      if (tensorBuilder_ == null) {
+        return tensor_.get(index);  } else {
+        return tensorBuilder_.getMessageOrBuilder(index);
+      }
+    }
+    /**
+     * <pre>
+     * Each SavedSliceMeta describes the slices for one tensor.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.SavedSliceMeta tensor = 1;</code>
+     */
+    public java.util.List<? extends org.tensorflow.proto.util.SavedSliceMetaOrBuilder> 
+         getTensorOrBuilderList() {
+      if (tensorBuilder_ != null) {
+        return tensorBuilder_.getMessageOrBuilderList();
+      } else {
+        return java.util.Collections.unmodifiableList(tensor_);
+      }
+    }
+    /**
+     * <pre>
+     * Each SavedSliceMeta describes the slices for one tensor.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.SavedSliceMeta tensor = 1;</code>
+     */
+    public org.tensorflow.proto.util.SavedSliceMeta.Builder addTensorBuilder() {
+      return getTensorFieldBuilder().addBuilder(
+          org.tensorflow.proto.util.SavedSliceMeta.getDefaultInstance());
+    }
+    /**
+     * <pre>
+     * Each SavedSliceMeta describes the slices for one tensor.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.SavedSliceMeta tensor = 1;</code>
+     */
+    public org.tensorflow.proto.util.SavedSliceMeta.Builder addTensorBuilder(
+        int index) {
+      return getTensorFieldBuilder().addBuilder(
+          index, org.tensorflow.proto.util.SavedSliceMeta.getDefaultInstance());
+    }
+    /**
+     * <pre>
+     * Each SavedSliceMeta describes the slices for one tensor.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.SavedSliceMeta tensor = 1;</code>
+     */
+    public java.util.List<org.tensorflow.proto.util.SavedSliceMeta.Builder> 
+         getTensorBuilderList() {
+      return getTensorFieldBuilder().getBuilderList();
+    }
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.util.SavedSliceMeta, org.tensorflow.proto.util.SavedSliceMeta.Builder, org.tensorflow.proto.util.SavedSliceMetaOrBuilder> 
+        getTensorFieldBuilder() {
+      if (tensorBuilder_ == null) {
+        tensorBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
+            org.tensorflow.proto.util.SavedSliceMeta, org.tensorflow.proto.util.SavedSliceMeta.Builder, org.tensorflow.proto.util.SavedSliceMetaOrBuilder>(
+                tensor_,
+                ((bitField0_ & 0x00000001) != 0),
+                getParentForChildren(),
+                isClean());
+        tensor_ = null;
+      }
+      return tensorBuilder_;
+    }
+
+    private org.tensorflow.proto.framework.VersionDef versions_;
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.VersionDef, org.tensorflow.proto.framework.VersionDef.Builder, org.tensorflow.proto.framework.VersionDefOrBuilder> versionsBuilder_;
+    /**
+     * <pre>
+     * Compatibility version of this checkpoint.  See core/public/version.h
+     * for version history.
+     * </pre>
+     *
+     * <code>.tensorflow.VersionDef versions = 2;</code>
+     */
+    public boolean hasVersions() {
+      return versionsBuilder_ != null || versions_ != null;
+    }
+    /**
+     * <pre>
+     * Compatibility version of this checkpoint.  See core/public/version.h
+     * for version history.
+     * </pre>
+     *
+     * <code>.tensorflow.VersionDef versions = 2;</code>
+     */
+    public org.tensorflow.proto.framework.VersionDef getVersions() {
+      if (versionsBuilder_ == null) {
+        return versions_ == null ? org.tensorflow.proto.framework.VersionDef.getDefaultInstance() : versions_;
+      } else {
+        return versionsBuilder_.getMessage();
+      }
+    }
+    /**
+     * <pre>
+     * Compatibility version of this checkpoint.  See core/public/version.h
+     * for version history.
+     * </pre>
+     *
+     * <code>.tensorflow.VersionDef versions = 2;</code>
+     */
+    public Builder setVersions(org.tensorflow.proto.framework.VersionDef value) {
+      if (versionsBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        versions_ = value;
+        onChanged();
+      } else {
+        versionsBuilder_.setMessage(value);
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Compatibility version of this checkpoint.  See core/public/version.h
+     * for version history.
+     * </pre>
+     *
+     * <code>.tensorflow.VersionDef versions = 2;</code>
+     */
+    public Builder setVersions(
+        org.tensorflow.proto.framework.VersionDef.Builder builderForValue) {
+      if (versionsBuilder_ == null) {
+        versions_ = builderForValue.build();
+        onChanged();
+      } else {
+        versionsBuilder_.setMessage(builderForValue.build());
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Compatibility version of this checkpoint.  See core/public/version.h
+     * for version history.
+     * </pre>
+     *
+     * <code>.tensorflow.VersionDef versions = 2;</code>
+     */
+    public Builder mergeVersions(org.tensorflow.proto.framework.VersionDef value) {
+      if (versionsBuilder_ == null) {
+        if (versions_ != null) {
+          versions_ =
+            org.tensorflow.proto.framework.VersionDef.newBuilder(versions_).mergeFrom(value).buildPartial();
+        } else {
+          versions_ = value;
+        }
+        onChanged();
+      } else {
+        versionsBuilder_.mergeFrom(value);
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Compatibility version of this checkpoint.  See core/public/version.h
+     * for version history.
+     * </pre>
+     *
+     * <code>.tensorflow.VersionDef versions = 2;</code>
+     */
+    public Builder clearVersions() {
+      if (versionsBuilder_ == null) {
+        versions_ = null;
+        onChanged();
+      } else {
+        versions_ = null;
+        versionsBuilder_ = null;
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Compatibility version of this checkpoint.  See core/public/version.h
+     * for version history.
+     * </pre>
+     *
+     * <code>.tensorflow.VersionDef versions = 2;</code>
+     */
+    public org.tensorflow.proto.framework.VersionDef.Builder getVersionsBuilder() {
+      
+      onChanged();
+      return getVersionsFieldBuilder().getBuilder();
+    }
+    /**
+     * <pre>
+     * Compatibility version of this checkpoint.  See core/public/version.h
+     * for version history.
+     * </pre>
+     *
+     * <code>.tensorflow.VersionDef versions = 2;</code>
+     */
+    public org.tensorflow.proto.framework.VersionDefOrBuilder getVersionsOrBuilder() {
+      if (versionsBuilder_ != null) {
+        return versionsBuilder_.getMessageOrBuilder();
+      } else {
+        return versions_ == null ?
+            org.tensorflow.proto.framework.VersionDef.getDefaultInstance() : versions_;
+      }
+    }
+    /**
+     * <pre>
+     * Compatibility version of this checkpoint.  See core/public/version.h
+     * for version history.
+     * </pre>
+     *
+     * <code>.tensorflow.VersionDef versions = 2;</code>
+     */
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.VersionDef, org.tensorflow.proto.framework.VersionDef.Builder, org.tensorflow.proto.framework.VersionDefOrBuilder> 
+        getVersionsFieldBuilder() {
+      if (versionsBuilder_ == null) {
+        versionsBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+            org.tensorflow.proto.framework.VersionDef, org.tensorflow.proto.framework.VersionDef.Builder, org.tensorflow.proto.framework.VersionDefOrBuilder>(
+                getVersions(),
+                getParentForChildren(),
+                isClean());
+        versions_ = null;
+      }
+      return versionsBuilder_;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.SavedTensorSliceMeta)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.SavedTensorSliceMeta)
+  private static final org.tensorflow.proto.util.SavedTensorSliceMeta DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.util.SavedTensorSliceMeta();
+  }
+
+  public static org.tensorflow.proto.util.SavedTensorSliceMeta getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<SavedTensorSliceMeta>
+      PARSER = new com.google.protobuf.AbstractParser<SavedTensorSliceMeta>() {
+    @java.lang.Override
+    public SavedTensorSliceMeta parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new SavedTensorSliceMeta(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<SavedTensorSliceMeta> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<SavedTensorSliceMeta> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.util.SavedTensorSliceMeta getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SavedTensorSliceMetaOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SavedTensorSliceMetaOrBuilder.java
new file mode 100644
index 00000000000..464fc6423f6
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SavedTensorSliceMetaOrBuilder.java
@@ -0,0 +1,81 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/util/saved_tensor_slice.proto
+
+package org.tensorflow.proto.util;
+
+public interface SavedTensorSliceMetaOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.SavedTensorSliceMeta)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <pre>
+   * Each SavedSliceMeta describes the slices for one tensor.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.SavedSliceMeta tensor = 1;</code>
+   */
+  java.util.List<org.tensorflow.proto.util.SavedSliceMeta> 
+      getTensorList();
+  /**
+   * <pre>
+   * Each SavedSliceMeta describes the slices for one tensor.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.SavedSliceMeta tensor = 1;</code>
+   */
+  org.tensorflow.proto.util.SavedSliceMeta getTensor(int index);
+  /**
+   * <pre>
+   * Each SavedSliceMeta describes the slices for one tensor.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.SavedSliceMeta tensor = 1;</code>
+   */
+  int getTensorCount();
+  /**
+   * <pre>
+   * Each SavedSliceMeta describes the slices for one tensor.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.SavedSliceMeta tensor = 1;</code>
+   */
+  java.util.List<? extends org.tensorflow.proto.util.SavedSliceMetaOrBuilder> 
+      getTensorOrBuilderList();
+  /**
+   * <pre>
+   * Each SavedSliceMeta describes the slices for one tensor.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.SavedSliceMeta tensor = 1;</code>
+   */
+  org.tensorflow.proto.util.SavedSliceMetaOrBuilder getTensorOrBuilder(
+      int index);
+
+  /**
+   * <pre>
+   * Compatibility version of this checkpoint.  See core/public/version.h
+   * for version history.
+   * </pre>
+   *
+   * <code>.tensorflow.VersionDef versions = 2;</code>
+   */
+  boolean hasVersions();
+  /**
+   * <pre>
+   * Compatibility version of this checkpoint.  See core/public/version.h
+   * for version history.
+   * </pre>
+   *
+   * <code>.tensorflow.VersionDef versions = 2;</code>
+   */
+  org.tensorflow.proto.framework.VersionDef getVersions();
+  /**
+   * <pre>
+   * Compatibility version of this checkpoint.  See core/public/version.h
+   * for version history.
+   * </pre>
+   *
+   * <code>.tensorflow.VersionDef versions = 2;</code>
+   */
+  org.tensorflow.proto.framework.VersionDefOrBuilder getVersionsOrBuilder();
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SavedTensorSliceProtos.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SavedTensorSliceProtos.java
new file mode 100644
index 00000000000..94de1b07571
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SavedTensorSliceProtos.java
@@ -0,0 +1,109 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/util/saved_tensor_slice.proto
+
+package org.tensorflow.proto.util;
+
+public final class SavedTensorSliceProtos {
+  private SavedTensorSliceProtos() {}
+  public static void registerAllExtensions(
+      com.google.protobuf.ExtensionRegistryLite registry) {
+  }
+
+  public static void registerAllExtensions(
+      com.google.protobuf.ExtensionRegistry registry) {
+    registerAllExtensions(
+        (com.google.protobuf.ExtensionRegistryLite) registry);
+  }
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_SavedSliceMeta_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_SavedSliceMeta_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_SavedTensorSliceMeta_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_SavedTensorSliceMeta_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_SavedSlice_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_SavedSlice_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_SavedTensorSlices_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_SavedTensorSlices_fieldAccessorTable;
+
+  public static com.google.protobuf.Descriptors.FileDescriptor
+      getDescriptor() {
+    return descriptor;
+  }
+  private static  com.google.protobuf.Descriptors.FileDescriptor
+      descriptor;
+  static {
+    java.lang.String[] descriptorData = {
+      "\n-tensorflow/core/util/saved_tensor_slic" +
+      "e.proto\022\ntensorflow\032,tensorflow/core/fra" +
+      "mework/tensor_shape.proto\032,tensorflow/co" +
+      "re/framework/tensor_slice.proto\032&tensorf" +
+      "low/core/framework/tensor.proto\032%tensorf" +
+      "low/core/framework/types.proto\032(tensorfl" +
+      "ow/core/framework/versions.proto\"\234\001\n\016Sav" +
+      "edSliceMeta\022\014\n\004name\030\001 \001(\t\022+\n\005shape\030\002 \001(\013" +
+      "2\034.tensorflow.TensorShapeProto\022\"\n\004type\030\003" +
+      " \001(\0162\024.tensorflow.DataType\022+\n\005slice\030\004 \003(" +
+      "\0132\034.tensorflow.TensorSliceProto\"l\n\024Saved" +
+      "TensorSliceMeta\022*\n\006tensor\030\001 \003(\0132\032.tensor" +
+      "flow.SavedSliceMeta\022(\n\010versions\030\002 \001(\0132\026." +
+      "tensorflow.VersionDef\"n\n\nSavedSlice\022\014\n\004n" +
+      "ame\030\001 \001(\t\022+\n\005slice\030\002 \001(\0132\034.tensorflow.Te" +
+      "nsorSliceProto\022%\n\004data\030\003 \001(\0132\027.tensorflo" +
+      "w.TensorProto\"i\n\021SavedTensorSlices\022.\n\004me" +
+      "ta\030\001 \001(\0132 .tensorflow.SavedTensorSliceMe" +
+      "ta\022$\n\004data\030\002 \001(\0132\026.tensorflow.SavedSlice" +
+      "B8\n\031org.tensorflow.proto.utilB\026SavedTens" +
+      "orSliceProtosP\001\370\001\001b\006proto3"
+    };
+    descriptor = com.google.protobuf.Descriptors.FileDescriptor
+      .internalBuildGeneratedFileFrom(descriptorData,
+        new com.google.protobuf.Descriptors.FileDescriptor[] {
+          org.tensorflow.proto.framework.TensorShapeProtos.getDescriptor(),
+          org.tensorflow.proto.framework.TensorSliceProtos.getDescriptor(),
+          org.tensorflow.proto.framework.TensorProtos.getDescriptor(),
+          org.tensorflow.proto.framework.TypesProtos.getDescriptor(),
+          org.tensorflow.proto.framework.VersionsProtos.getDescriptor(),
+        });
+    internal_static_tensorflow_SavedSliceMeta_descriptor =
+      getDescriptor().getMessageTypes().get(0);
+    internal_static_tensorflow_SavedSliceMeta_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_SavedSliceMeta_descriptor,
+        new java.lang.String[] { "Name", "Shape", "Type", "Slice", });
+    internal_static_tensorflow_SavedTensorSliceMeta_descriptor =
+      getDescriptor().getMessageTypes().get(1);
+    internal_static_tensorflow_SavedTensorSliceMeta_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_SavedTensorSliceMeta_descriptor,
+        new java.lang.String[] { "Tensor", "Versions", });
+    internal_static_tensorflow_SavedSlice_descriptor =
+      getDescriptor().getMessageTypes().get(2);
+    internal_static_tensorflow_SavedSlice_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_SavedSlice_descriptor,
+        new java.lang.String[] { "Name", "Slice", "Data", });
+    internal_static_tensorflow_SavedTensorSlices_descriptor =
+      getDescriptor().getMessageTypes().get(3);
+    internal_static_tensorflow_SavedTensorSlices_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_SavedTensorSlices_descriptor,
+        new java.lang.String[] { "Meta", "Data", });
+    org.tensorflow.proto.framework.TensorShapeProtos.getDescriptor();
+    org.tensorflow.proto.framework.TensorSliceProtos.getDescriptor();
+    org.tensorflow.proto.framework.TensorProtos.getDescriptor();
+    org.tensorflow.proto.framework.TypesProtos.getDescriptor();
+    org.tensorflow.proto.framework.VersionsProtos.getDescriptor();
+  }
+
+  // @@protoc_insertion_point(outer_class_scope)
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SavedTensorSlices.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SavedTensorSlices.java
new file mode 100644
index 00000000000..e058ca978ee
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SavedTensorSlices.java
@@ -0,0 +1,899 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/util/saved_tensor_slice.proto
+
+package org.tensorflow.proto.util;
+
+/**
+ * <pre>
+ * Each record in a v3 checkpoint file is a serialized SavedTensorSlices
+ * message.
+ * </pre>
+ *
+ * Protobuf type {@code tensorflow.SavedTensorSlices}
+ */
+public  final class SavedTensorSlices extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.SavedTensorSlices)
+    SavedTensorSlicesOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use SavedTensorSlices.newBuilder() to construct.
+  private SavedTensorSlices(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private SavedTensorSlices() {
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new SavedTensorSlices();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private SavedTensorSlices(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 10: {
+            org.tensorflow.proto.util.SavedTensorSliceMeta.Builder subBuilder = null;
+            if (meta_ != null) {
+              subBuilder = meta_.toBuilder();
+            }
+            meta_ = input.readMessage(org.tensorflow.proto.util.SavedTensorSliceMeta.parser(), extensionRegistry);
+            if (subBuilder != null) {
+              subBuilder.mergeFrom(meta_);
+              meta_ = subBuilder.buildPartial();
+            }
+
+            break;
+          }
+          case 18: {
+            org.tensorflow.proto.util.SavedSlice.Builder subBuilder = null;
+            if (data_ != null) {
+              subBuilder = data_.toBuilder();
+            }
+            data_ = input.readMessage(org.tensorflow.proto.util.SavedSlice.parser(), extensionRegistry);
+            if (subBuilder != null) {
+              subBuilder.mergeFrom(data_);
+              data_ = subBuilder.buildPartial();
+            }
+
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.util.SavedTensorSliceProtos.internal_static_tensorflow_SavedTensorSlices_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.util.SavedTensorSliceProtos.internal_static_tensorflow_SavedTensorSlices_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.util.SavedTensorSlices.class, org.tensorflow.proto.util.SavedTensorSlices.Builder.class);
+  }
+
+  public static final int META_FIELD_NUMBER = 1;
+  private org.tensorflow.proto.util.SavedTensorSliceMeta meta_;
+  /**
+   * <pre>
+   * This is only present at the first item of each checkpoint file and serves
+   * as a table of contents, listing all the tensor slices saved in this file.
+   * </pre>
+   *
+   * <code>.tensorflow.SavedTensorSliceMeta meta = 1;</code>
+   */
+  public boolean hasMeta() {
+    return meta_ != null;
+  }
+  /**
+   * <pre>
+   * This is only present at the first item of each checkpoint file and serves
+   * as a table of contents, listing all the tensor slices saved in this file.
+   * </pre>
+   *
+   * <code>.tensorflow.SavedTensorSliceMeta meta = 1;</code>
+   */
+  public org.tensorflow.proto.util.SavedTensorSliceMeta getMeta() {
+    return meta_ == null ? org.tensorflow.proto.util.SavedTensorSliceMeta.getDefaultInstance() : meta_;
+  }
+  /**
+   * <pre>
+   * This is only present at the first item of each checkpoint file and serves
+   * as a table of contents, listing all the tensor slices saved in this file.
+   * </pre>
+   *
+   * <code>.tensorflow.SavedTensorSliceMeta meta = 1;</code>
+   */
+  public org.tensorflow.proto.util.SavedTensorSliceMetaOrBuilder getMetaOrBuilder() {
+    return getMeta();
+  }
+
+  public static final int DATA_FIELD_NUMBER = 2;
+  private org.tensorflow.proto.util.SavedSlice data_;
+  /**
+   * <pre>
+   * This exists in all but the first item of each checkpoint file.
+   * </pre>
+   *
+   * <code>.tensorflow.SavedSlice data = 2;</code>
+   */
+  public boolean hasData() {
+    return data_ != null;
+  }
+  /**
+   * <pre>
+   * This exists in all but the first item of each checkpoint file.
+   * </pre>
+   *
+   * <code>.tensorflow.SavedSlice data = 2;</code>
+   */
+  public org.tensorflow.proto.util.SavedSlice getData() {
+    return data_ == null ? org.tensorflow.proto.util.SavedSlice.getDefaultInstance() : data_;
+  }
+  /**
+   * <pre>
+   * This exists in all but the first item of each checkpoint file.
+   * </pre>
+   *
+   * <code>.tensorflow.SavedSlice data = 2;</code>
+   */
+  public org.tensorflow.proto.util.SavedSliceOrBuilder getDataOrBuilder() {
+    return getData();
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    if (meta_ != null) {
+      output.writeMessage(1, getMeta());
+    }
+    if (data_ != null) {
+      output.writeMessage(2, getData());
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (meta_ != null) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(1, getMeta());
+    }
+    if (data_ != null) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(2, getData());
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.util.SavedTensorSlices)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.util.SavedTensorSlices other = (org.tensorflow.proto.util.SavedTensorSlices) obj;
+
+    if (hasMeta() != other.hasMeta()) return false;
+    if (hasMeta()) {
+      if (!getMeta()
+          .equals(other.getMeta())) return false;
+    }
+    if (hasData() != other.hasData()) return false;
+    if (hasData()) {
+      if (!getData()
+          .equals(other.getData())) return false;
+    }
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    if (hasMeta()) {
+      hash = (37 * hash) + META_FIELD_NUMBER;
+      hash = (53 * hash) + getMeta().hashCode();
+    }
+    if (hasData()) {
+      hash = (37 * hash) + DATA_FIELD_NUMBER;
+      hash = (53 * hash) + getData().hashCode();
+    }
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.util.SavedTensorSlices parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.SavedTensorSlices parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.SavedTensorSlices parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.SavedTensorSlices parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.SavedTensorSlices parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.SavedTensorSlices parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.SavedTensorSlices parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.SavedTensorSlices parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.SavedTensorSlices parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.SavedTensorSlices parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.SavedTensorSlices parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.SavedTensorSlices parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.util.SavedTensorSlices prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * <pre>
+   * Each record in a v3 checkpoint file is a serialized SavedTensorSlices
+   * message.
+   * </pre>
+   *
+   * Protobuf type {@code tensorflow.SavedTensorSlices}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.SavedTensorSlices)
+      org.tensorflow.proto.util.SavedTensorSlicesOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.util.SavedTensorSliceProtos.internal_static_tensorflow_SavedTensorSlices_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.util.SavedTensorSliceProtos.internal_static_tensorflow_SavedTensorSlices_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.util.SavedTensorSlices.class, org.tensorflow.proto.util.SavedTensorSlices.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.util.SavedTensorSlices.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      if (metaBuilder_ == null) {
+        meta_ = null;
+      } else {
+        meta_ = null;
+        metaBuilder_ = null;
+      }
+      if (dataBuilder_ == null) {
+        data_ = null;
+      } else {
+        data_ = null;
+        dataBuilder_ = null;
+      }
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.util.SavedTensorSliceProtos.internal_static_tensorflow_SavedTensorSlices_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.SavedTensorSlices getDefaultInstanceForType() {
+      return org.tensorflow.proto.util.SavedTensorSlices.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.SavedTensorSlices build() {
+      org.tensorflow.proto.util.SavedTensorSlices result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.SavedTensorSlices buildPartial() {
+      org.tensorflow.proto.util.SavedTensorSlices result = new org.tensorflow.proto.util.SavedTensorSlices(this);
+      if (metaBuilder_ == null) {
+        result.meta_ = meta_;
+      } else {
+        result.meta_ = metaBuilder_.build();
+      }
+      if (dataBuilder_ == null) {
+        result.data_ = data_;
+      } else {
+        result.data_ = dataBuilder_.build();
+      }
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.util.SavedTensorSlices) {
+        return mergeFrom((org.tensorflow.proto.util.SavedTensorSlices)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.util.SavedTensorSlices other) {
+      if (other == org.tensorflow.proto.util.SavedTensorSlices.getDefaultInstance()) return this;
+      if (other.hasMeta()) {
+        mergeMeta(other.getMeta());
+      }
+      if (other.hasData()) {
+        mergeData(other.getData());
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.util.SavedTensorSlices parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.util.SavedTensorSlices) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+
+    private org.tensorflow.proto.util.SavedTensorSliceMeta meta_;
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.util.SavedTensorSliceMeta, org.tensorflow.proto.util.SavedTensorSliceMeta.Builder, org.tensorflow.proto.util.SavedTensorSliceMetaOrBuilder> metaBuilder_;
+    /**
+     * <pre>
+     * This is only present at the first item of each checkpoint file and serves
+     * as a table of contents, listing all the tensor slices saved in this file.
+     * </pre>
+     *
+     * <code>.tensorflow.SavedTensorSliceMeta meta = 1;</code>
+     */
+    public boolean hasMeta() {
+      return metaBuilder_ != null || meta_ != null;
+    }
+    /**
+     * <pre>
+     * This is only present at the first item of each checkpoint file and serves
+     * as a table of contents, listing all the tensor slices saved in this file.
+     * </pre>
+     *
+     * <code>.tensorflow.SavedTensorSliceMeta meta = 1;</code>
+     */
+    public org.tensorflow.proto.util.SavedTensorSliceMeta getMeta() {
+      if (metaBuilder_ == null) {
+        return meta_ == null ? org.tensorflow.proto.util.SavedTensorSliceMeta.getDefaultInstance() : meta_;
+      } else {
+        return metaBuilder_.getMessage();
+      }
+    }
+    /**
+     * <pre>
+     * This is only present at the first item of each checkpoint file and serves
+     * as a table of contents, listing all the tensor slices saved in this file.
+     * </pre>
+     *
+     * <code>.tensorflow.SavedTensorSliceMeta meta = 1;</code>
+     */
+    public Builder setMeta(org.tensorflow.proto.util.SavedTensorSliceMeta value) {
+      if (metaBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        meta_ = value;
+        onChanged();
+      } else {
+        metaBuilder_.setMessage(value);
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * This is only present at the first item of each checkpoint file and serves
+     * as a table of contents, listing all the tensor slices saved in this file.
+     * </pre>
+     *
+     * <code>.tensorflow.SavedTensorSliceMeta meta = 1;</code>
+     */
+    public Builder setMeta(
+        org.tensorflow.proto.util.SavedTensorSliceMeta.Builder builderForValue) {
+      if (metaBuilder_ == null) {
+        meta_ = builderForValue.build();
+        onChanged();
+      } else {
+        metaBuilder_.setMessage(builderForValue.build());
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * This is only present at the first item of each checkpoint file and serves
+     * as a table of contents, listing all the tensor slices saved in this file.
+     * </pre>
+     *
+     * <code>.tensorflow.SavedTensorSliceMeta meta = 1;</code>
+     */
+    public Builder mergeMeta(org.tensorflow.proto.util.SavedTensorSliceMeta value) {
+      if (metaBuilder_ == null) {
+        if (meta_ != null) {
+          meta_ =
+            org.tensorflow.proto.util.SavedTensorSliceMeta.newBuilder(meta_).mergeFrom(value).buildPartial();
+        } else {
+          meta_ = value;
+        }
+        onChanged();
+      } else {
+        metaBuilder_.mergeFrom(value);
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * This is only present at the first item of each checkpoint file and serves
+     * as a table of contents, listing all the tensor slices saved in this file.
+     * </pre>
+     *
+     * <code>.tensorflow.SavedTensorSliceMeta meta = 1;</code>
+     */
+    public Builder clearMeta() {
+      if (metaBuilder_ == null) {
+        meta_ = null;
+        onChanged();
+      } else {
+        meta_ = null;
+        metaBuilder_ = null;
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * This is only present at the first item of each checkpoint file and serves
+     * as a table of contents, listing all the tensor slices saved in this file.
+     * </pre>
+     *
+     * <code>.tensorflow.SavedTensorSliceMeta meta = 1;</code>
+     */
+    public org.tensorflow.proto.util.SavedTensorSliceMeta.Builder getMetaBuilder() {
+      
+      onChanged();
+      return getMetaFieldBuilder().getBuilder();
+    }
+    /**
+     * <pre>
+     * This is only present at the first item of each checkpoint file and serves
+     * as a table of contents, listing all the tensor slices saved in this file.
+     * </pre>
+     *
+     * <code>.tensorflow.SavedTensorSliceMeta meta = 1;</code>
+     */
+    public org.tensorflow.proto.util.SavedTensorSliceMetaOrBuilder getMetaOrBuilder() {
+      if (metaBuilder_ != null) {
+        return metaBuilder_.getMessageOrBuilder();
+      } else {
+        return meta_ == null ?
+            org.tensorflow.proto.util.SavedTensorSliceMeta.getDefaultInstance() : meta_;
+      }
+    }
+    /**
+     * <pre>
+     * This is only present at the first item of each checkpoint file and serves
+     * as a table of contents, listing all the tensor slices saved in this file.
+     * </pre>
+     *
+     * <code>.tensorflow.SavedTensorSliceMeta meta = 1;</code>
+     */
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.util.SavedTensorSliceMeta, org.tensorflow.proto.util.SavedTensorSliceMeta.Builder, org.tensorflow.proto.util.SavedTensorSliceMetaOrBuilder> 
+        getMetaFieldBuilder() {
+      if (metaBuilder_ == null) {
+        metaBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+            org.tensorflow.proto.util.SavedTensorSliceMeta, org.tensorflow.proto.util.SavedTensorSliceMeta.Builder, org.tensorflow.proto.util.SavedTensorSliceMetaOrBuilder>(
+                getMeta(),
+                getParentForChildren(),
+                isClean());
+        meta_ = null;
+      }
+      return metaBuilder_;
+    }
+
+    private org.tensorflow.proto.util.SavedSlice data_;
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.util.SavedSlice, org.tensorflow.proto.util.SavedSlice.Builder, org.tensorflow.proto.util.SavedSliceOrBuilder> dataBuilder_;
+    /**
+     * <pre>
+     * This exists in all but the first item of each checkpoint file.
+     * </pre>
+     *
+     * <code>.tensorflow.SavedSlice data = 2;</code>
+     */
+    public boolean hasData() {
+      return dataBuilder_ != null || data_ != null;
+    }
+    /**
+     * <pre>
+     * This exists in all but the first item of each checkpoint file.
+     * </pre>
+     *
+     * <code>.tensorflow.SavedSlice data = 2;</code>
+     */
+    public org.tensorflow.proto.util.SavedSlice getData() {
+      if (dataBuilder_ == null) {
+        return data_ == null ? org.tensorflow.proto.util.SavedSlice.getDefaultInstance() : data_;
+      } else {
+        return dataBuilder_.getMessage();
+      }
+    }
+    /**
+     * <pre>
+     * This exists in all but the first item of each checkpoint file.
+     * </pre>
+     *
+     * <code>.tensorflow.SavedSlice data = 2;</code>
+     */
+    public Builder setData(org.tensorflow.proto.util.SavedSlice value) {
+      if (dataBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        data_ = value;
+        onChanged();
+      } else {
+        dataBuilder_.setMessage(value);
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * This exists in all but the first item of each checkpoint file.
+     * </pre>
+     *
+     * <code>.tensorflow.SavedSlice data = 2;</code>
+     */
+    public Builder setData(
+        org.tensorflow.proto.util.SavedSlice.Builder builderForValue) {
+      if (dataBuilder_ == null) {
+        data_ = builderForValue.build();
+        onChanged();
+      } else {
+        dataBuilder_.setMessage(builderForValue.build());
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * This exists in all but the first item of each checkpoint file.
+     * </pre>
+     *
+     * <code>.tensorflow.SavedSlice data = 2;</code>
+     */
+    public Builder mergeData(org.tensorflow.proto.util.SavedSlice value) {
+      if (dataBuilder_ == null) {
+        if (data_ != null) {
+          data_ =
+            org.tensorflow.proto.util.SavedSlice.newBuilder(data_).mergeFrom(value).buildPartial();
+        } else {
+          data_ = value;
+        }
+        onChanged();
+      } else {
+        dataBuilder_.mergeFrom(value);
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * This exists in all but the first item of each checkpoint file.
+     * </pre>
+     *
+     * <code>.tensorflow.SavedSlice data = 2;</code>
+     */
+    public Builder clearData() {
+      if (dataBuilder_ == null) {
+        data_ = null;
+        onChanged();
+      } else {
+        data_ = null;
+        dataBuilder_ = null;
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * This exists in all but the first item of each checkpoint file.
+     * </pre>
+     *
+     * <code>.tensorflow.SavedSlice data = 2;</code>
+     */
+    public org.tensorflow.proto.util.SavedSlice.Builder getDataBuilder() {
+      
+      onChanged();
+      return getDataFieldBuilder().getBuilder();
+    }
+    /**
+     * <pre>
+     * This exists in all but the first item of each checkpoint file.
+     * </pre>
+     *
+     * <code>.tensorflow.SavedSlice data = 2;</code>
+     */
+    public org.tensorflow.proto.util.SavedSliceOrBuilder getDataOrBuilder() {
+      if (dataBuilder_ != null) {
+        return dataBuilder_.getMessageOrBuilder();
+      } else {
+        return data_ == null ?
+            org.tensorflow.proto.util.SavedSlice.getDefaultInstance() : data_;
+      }
+    }
+    /**
+     * <pre>
+     * This exists in all but the first item of each checkpoint file.
+     * </pre>
+     *
+     * <code>.tensorflow.SavedSlice data = 2;</code>
+     */
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.util.SavedSlice, org.tensorflow.proto.util.SavedSlice.Builder, org.tensorflow.proto.util.SavedSliceOrBuilder> 
+        getDataFieldBuilder() {
+      if (dataBuilder_ == null) {
+        dataBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+            org.tensorflow.proto.util.SavedSlice, org.tensorflow.proto.util.SavedSlice.Builder, org.tensorflow.proto.util.SavedSliceOrBuilder>(
+                getData(),
+                getParentForChildren(),
+                isClean());
+        data_ = null;
+      }
+      return dataBuilder_;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.SavedTensorSlices)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.SavedTensorSlices)
+  private static final org.tensorflow.proto.util.SavedTensorSlices DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.util.SavedTensorSlices();
+  }
+
+  public static org.tensorflow.proto.util.SavedTensorSlices getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<SavedTensorSlices>
+      PARSER = new com.google.protobuf.AbstractParser<SavedTensorSlices>() {
+    @java.lang.Override
+    public SavedTensorSlices parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new SavedTensorSlices(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<SavedTensorSlices> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<SavedTensorSlices> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.util.SavedTensorSlices getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SavedTensorSlicesOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SavedTensorSlicesOrBuilder.java
new file mode 100644
index 00000000000..f8b12b1c7dd
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SavedTensorSlicesOrBuilder.java
@@ -0,0 +1,62 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/util/saved_tensor_slice.proto
+
+package org.tensorflow.proto.util;
+
+public interface SavedTensorSlicesOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.SavedTensorSlices)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <pre>
+   * This is only present at the first item of each checkpoint file and serves
+   * as a table of contents, listing all the tensor slices saved in this file.
+   * </pre>
+   *
+   * <code>.tensorflow.SavedTensorSliceMeta meta = 1;</code>
+   */
+  boolean hasMeta();
+  /**
+   * <pre>
+   * This is only present at the first item of each checkpoint file and serves
+   * as a table of contents, listing all the tensor slices saved in this file.
+   * </pre>
+   *
+   * <code>.tensorflow.SavedTensorSliceMeta meta = 1;</code>
+   */
+  org.tensorflow.proto.util.SavedTensorSliceMeta getMeta();
+  /**
+   * <pre>
+   * This is only present at the first item of each checkpoint file and serves
+   * as a table of contents, listing all the tensor slices saved in this file.
+   * </pre>
+   *
+   * <code>.tensorflow.SavedTensorSliceMeta meta = 1;</code>
+   */
+  org.tensorflow.proto.util.SavedTensorSliceMetaOrBuilder getMetaOrBuilder();
+
+  /**
+   * <pre>
+   * This exists in all but the first item of each checkpoint file.
+   * </pre>
+   *
+   * <code>.tensorflow.SavedSlice data = 2;</code>
+   */
+  boolean hasData();
+  /**
+   * <pre>
+   * This exists in all but the first item of each checkpoint file.
+   * </pre>
+   *
+   * <code>.tensorflow.SavedSlice data = 2;</code>
+   */
+  org.tensorflow.proto.util.SavedSlice getData();
+  /**
+   * <pre>
+   * This exists in all but the first item of each checkpoint file.
+   * </pre>
+   *
+   * <code>.tensorflow.SavedSlice data = 2;</code>
+   */
+  org.tensorflow.proto.util.SavedSliceOrBuilder getDataOrBuilder();
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SaverDef.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SaverDef.java
new file mode 100644
index 00000000000..daf4a6c91e4
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SaverDef.java
@@ -0,0 +1,1356 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/saver.proto
+
+package org.tensorflow.proto.util;
+
+/**
+ * <pre>
+ * Protocol buffer representing the configuration of a Saver.
+ * </pre>
+ *
+ * Protobuf type {@code tensorflow.SaverDef}
+ */
+public  final class SaverDef extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.SaverDef)
+    SaverDefOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use SaverDef.newBuilder() to construct.
+  private SaverDef(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private SaverDef() {
+    filenameTensorName_ = "";
+    saveTensorName_ = "";
+    restoreOpName_ = "";
+    version_ = 0;
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new SaverDef();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private SaverDef(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 10: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            filenameTensorName_ = s;
+            break;
+          }
+          case 18: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            saveTensorName_ = s;
+            break;
+          }
+          case 26: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            restoreOpName_ = s;
+            break;
+          }
+          case 32: {
+
+            maxToKeep_ = input.readInt32();
+            break;
+          }
+          case 40: {
+
+            sharded_ = input.readBool();
+            break;
+          }
+          case 53: {
+
+            keepCheckpointEveryNHours_ = input.readFloat();
+            break;
+          }
+          case 56: {
+            int rawValue = input.readEnum();
+
+            version_ = rawValue;
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.util.SaverProtos.internal_static_tensorflow_SaverDef_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.util.SaverProtos.internal_static_tensorflow_SaverDef_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.util.SaverDef.class, org.tensorflow.proto.util.SaverDef.Builder.class);
+  }
+
+  /**
+   * <pre>
+   * A version number that identifies a different on-disk checkpoint format.
+   * Usually, each subclass of BaseSaverBuilder works with a particular
+   * version/format.  However, it is possible that the same builder may be
+   * upgraded to support a newer checkpoint format in the future.
+   * </pre>
+   *
+   * Protobuf enum {@code tensorflow.SaverDef.CheckpointFormatVersion}
+   */
+  public enum CheckpointFormatVersion
+      implements com.google.protobuf.ProtocolMessageEnum {
+    /**
+     * <pre>
+     * Internal legacy format.
+     * </pre>
+     *
+     * <code>LEGACY = 0;</code>
+     */
+    LEGACY(0),
+    /**
+     * <pre>
+     * Deprecated format: tf.Saver() which works with tensorflow::table::Table.
+     * </pre>
+     *
+     * <code>V1 = 1;</code>
+     */
+    V1(1),
+    /**
+     * <pre>
+     * Current format: more efficient.
+     * </pre>
+     *
+     * <code>V2 = 2;</code>
+     */
+    V2(2),
+    UNRECOGNIZED(-1),
+    ;
+
+    /**
+     * <pre>
+     * Internal legacy format.
+     * </pre>
+     *
+     * <code>LEGACY = 0;</code>
+     */
+    public static final int LEGACY_VALUE = 0;
+    /**
+     * <pre>
+     * Deprecated format: tf.Saver() which works with tensorflow::table::Table.
+     * </pre>
+     *
+     * <code>V1 = 1;</code>
+     */
+    public static final int V1_VALUE = 1;
+    /**
+     * <pre>
+     * Current format: more efficient.
+     * </pre>
+     *
+     * <code>V2 = 2;</code>
+     */
+    public static final int V2_VALUE = 2;
+
+
+    public final int getNumber() {
+      if (this == UNRECOGNIZED) {
+        throw new java.lang.IllegalArgumentException(
+            "Can't get the number of an unknown enum value.");
+      }
+      return value;
+    }
+
+    /**
+     * @deprecated Use {@link #forNumber(int)} instead.
+     */
+    @java.lang.Deprecated
+    public static CheckpointFormatVersion valueOf(int value) {
+      return forNumber(value);
+    }
+
+    public static CheckpointFormatVersion forNumber(int value) {
+      switch (value) {
+        case 0: return LEGACY;
+        case 1: return V1;
+        case 2: return V2;
+        default: return null;
+      }
+    }
+
+    public static com.google.protobuf.Internal.EnumLiteMap<CheckpointFormatVersion>
+        internalGetValueMap() {
+      return internalValueMap;
+    }
+    private static final com.google.protobuf.Internal.EnumLiteMap<
+        CheckpointFormatVersion> internalValueMap =
+          new com.google.protobuf.Internal.EnumLiteMap<CheckpointFormatVersion>() {
+            public CheckpointFormatVersion findValueByNumber(int number) {
+              return CheckpointFormatVersion.forNumber(number);
+            }
+          };
+
+    public final com.google.protobuf.Descriptors.EnumValueDescriptor
+        getValueDescriptor() {
+      return getDescriptor().getValues().get(ordinal());
+    }
+    public final com.google.protobuf.Descriptors.EnumDescriptor
+        getDescriptorForType() {
+      return getDescriptor();
+    }
+    public static final com.google.protobuf.Descriptors.EnumDescriptor
+        getDescriptor() {
+      return org.tensorflow.proto.util.SaverDef.getDescriptor().getEnumTypes().get(0);
+    }
+
+    private static final CheckpointFormatVersion[] VALUES = values();
+
+    public static CheckpointFormatVersion valueOf(
+        com.google.protobuf.Descriptors.EnumValueDescriptor desc) {
+      if (desc.getType() != getDescriptor()) {
+        throw new java.lang.IllegalArgumentException(
+          "EnumValueDescriptor is not for this type.");
+      }
+      if (desc.getIndex() == -1) {
+        return UNRECOGNIZED;
+      }
+      return VALUES[desc.getIndex()];
+    }
+
+    private final int value;
+
+    private CheckpointFormatVersion(int value) {
+      this.value = value;
+    }
+
+    // @@protoc_insertion_point(enum_scope:tensorflow.SaverDef.CheckpointFormatVersion)
+  }
+
+  public static final int FILENAME_TENSOR_NAME_FIELD_NUMBER = 1;
+  private volatile java.lang.Object filenameTensorName_;
+  /**
+   * <pre>
+   * The name of the tensor in which to specify the filename when saving or
+   * restoring a model checkpoint.
+   * </pre>
+   *
+   * <code>string filename_tensor_name = 1;</code>
+   */
+  public java.lang.String getFilenameTensorName() {
+    java.lang.Object ref = filenameTensorName_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      filenameTensorName_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * The name of the tensor in which to specify the filename when saving or
+   * restoring a model checkpoint.
+   * </pre>
+   *
+   * <code>string filename_tensor_name = 1;</code>
+   */
+  public com.google.protobuf.ByteString
+      getFilenameTensorNameBytes() {
+    java.lang.Object ref = filenameTensorName_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      filenameTensorName_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int SAVE_TENSOR_NAME_FIELD_NUMBER = 2;
+  private volatile java.lang.Object saveTensorName_;
+  /**
+   * <pre>
+   * The operation to run when saving a model checkpoint.
+   * </pre>
+   *
+   * <code>string save_tensor_name = 2;</code>
+   */
+  public java.lang.String getSaveTensorName() {
+    java.lang.Object ref = saveTensorName_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      saveTensorName_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * The operation to run when saving a model checkpoint.
+   * </pre>
+   *
+   * <code>string save_tensor_name = 2;</code>
+   */
+  public com.google.protobuf.ByteString
+      getSaveTensorNameBytes() {
+    java.lang.Object ref = saveTensorName_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      saveTensorName_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int RESTORE_OP_NAME_FIELD_NUMBER = 3;
+  private volatile java.lang.Object restoreOpName_;
+  /**
+   * <pre>
+   * The operation to run when restoring a model checkpoint.
+   * </pre>
+   *
+   * <code>string restore_op_name = 3;</code>
+   */
+  public java.lang.String getRestoreOpName() {
+    java.lang.Object ref = restoreOpName_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      restoreOpName_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * The operation to run when restoring a model checkpoint.
+   * </pre>
+   *
+   * <code>string restore_op_name = 3;</code>
+   */
+  public com.google.protobuf.ByteString
+      getRestoreOpNameBytes() {
+    java.lang.Object ref = restoreOpName_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      restoreOpName_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int MAX_TO_KEEP_FIELD_NUMBER = 4;
+  private int maxToKeep_;
+  /**
+   * <pre>
+   * Maximum number of checkpoints to keep.  If 0, no checkpoints are deleted.
+   * </pre>
+   *
+   * <code>int32 max_to_keep = 4;</code>
+   */
+  public int getMaxToKeep() {
+    return maxToKeep_;
+  }
+
+  public static final int SHARDED_FIELD_NUMBER = 5;
+  private boolean sharded_;
+  /**
+   * <pre>
+   * Shard the save files, one per device that has Variable nodes.
+   * </pre>
+   *
+   * <code>bool sharded = 5;</code>
+   */
+  public boolean getSharded() {
+    return sharded_;
+  }
+
+  public static final int KEEP_CHECKPOINT_EVERY_N_HOURS_FIELD_NUMBER = 6;
+  private float keepCheckpointEveryNHours_;
+  /**
+   * <pre>
+   * How often to keep an additional checkpoint. If not specified, only the last
+   * "max_to_keep" checkpoints are kept; if specified, in addition to keeping
+   * the last "max_to_keep" checkpoints, an additional checkpoint will be kept
+   * for every n hours of training.
+   * </pre>
+   *
+   * <code>float keep_checkpoint_every_n_hours = 6;</code>
+   */
+  public float getKeepCheckpointEveryNHours() {
+    return keepCheckpointEveryNHours_;
+  }
+
+  public static final int VERSION_FIELD_NUMBER = 7;
+  private int version_;
+  /**
+   * <code>.tensorflow.SaverDef.CheckpointFormatVersion version = 7;</code>
+   */
+  public int getVersionValue() {
+    return version_;
+  }
+  /**
+   * <code>.tensorflow.SaverDef.CheckpointFormatVersion version = 7;</code>
+   */
+  public org.tensorflow.proto.util.SaverDef.CheckpointFormatVersion getVersion() {
+    @SuppressWarnings("deprecation")
+    org.tensorflow.proto.util.SaverDef.CheckpointFormatVersion result = org.tensorflow.proto.util.SaverDef.CheckpointFormatVersion.valueOf(version_);
+    return result == null ? org.tensorflow.proto.util.SaverDef.CheckpointFormatVersion.UNRECOGNIZED : result;
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    if (!getFilenameTensorNameBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 1, filenameTensorName_);
+    }
+    if (!getSaveTensorNameBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 2, saveTensorName_);
+    }
+    if (!getRestoreOpNameBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 3, restoreOpName_);
+    }
+    if (maxToKeep_ != 0) {
+      output.writeInt32(4, maxToKeep_);
+    }
+    if (sharded_ != false) {
+      output.writeBool(5, sharded_);
+    }
+    if (keepCheckpointEveryNHours_ != 0F) {
+      output.writeFloat(6, keepCheckpointEveryNHours_);
+    }
+    if (version_ != org.tensorflow.proto.util.SaverDef.CheckpointFormatVersion.LEGACY.getNumber()) {
+      output.writeEnum(7, version_);
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (!getFilenameTensorNameBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, filenameTensorName_);
+    }
+    if (!getSaveTensorNameBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(2, saveTensorName_);
+    }
+    if (!getRestoreOpNameBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(3, restoreOpName_);
+    }
+    if (maxToKeep_ != 0) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt32Size(4, maxToKeep_);
+    }
+    if (sharded_ != false) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeBoolSize(5, sharded_);
+    }
+    if (keepCheckpointEveryNHours_ != 0F) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeFloatSize(6, keepCheckpointEveryNHours_);
+    }
+    if (version_ != org.tensorflow.proto.util.SaverDef.CheckpointFormatVersion.LEGACY.getNumber()) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeEnumSize(7, version_);
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.util.SaverDef)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.util.SaverDef other = (org.tensorflow.proto.util.SaverDef) obj;
+
+    if (!getFilenameTensorName()
+        .equals(other.getFilenameTensorName())) return false;
+    if (!getSaveTensorName()
+        .equals(other.getSaveTensorName())) return false;
+    if (!getRestoreOpName()
+        .equals(other.getRestoreOpName())) return false;
+    if (getMaxToKeep()
+        != other.getMaxToKeep()) return false;
+    if (getSharded()
+        != other.getSharded()) return false;
+    if (java.lang.Float.floatToIntBits(getKeepCheckpointEveryNHours())
+        != java.lang.Float.floatToIntBits(
+            other.getKeepCheckpointEveryNHours())) return false;
+    if (version_ != other.version_) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    hash = (37 * hash) + FILENAME_TENSOR_NAME_FIELD_NUMBER;
+    hash = (53 * hash) + getFilenameTensorName().hashCode();
+    hash = (37 * hash) + SAVE_TENSOR_NAME_FIELD_NUMBER;
+    hash = (53 * hash) + getSaveTensorName().hashCode();
+    hash = (37 * hash) + RESTORE_OP_NAME_FIELD_NUMBER;
+    hash = (53 * hash) + getRestoreOpName().hashCode();
+    hash = (37 * hash) + MAX_TO_KEEP_FIELD_NUMBER;
+    hash = (53 * hash) + getMaxToKeep();
+    hash = (37 * hash) + SHARDED_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashBoolean(
+        getSharded());
+    hash = (37 * hash) + KEEP_CHECKPOINT_EVERY_N_HOURS_FIELD_NUMBER;
+    hash = (53 * hash) + java.lang.Float.floatToIntBits(
+        getKeepCheckpointEveryNHours());
+    hash = (37 * hash) + VERSION_FIELD_NUMBER;
+    hash = (53 * hash) + version_;
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.util.SaverDef parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.SaverDef parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.SaverDef parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.SaverDef parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.SaverDef parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.SaverDef parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.SaverDef parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.SaverDef parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.SaverDef parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.SaverDef parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.SaverDef parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.SaverDef parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.util.SaverDef prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * <pre>
+   * Protocol buffer representing the configuration of a Saver.
+   * </pre>
+   *
+   * Protobuf type {@code tensorflow.SaverDef}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.SaverDef)
+      org.tensorflow.proto.util.SaverDefOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.util.SaverProtos.internal_static_tensorflow_SaverDef_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.util.SaverProtos.internal_static_tensorflow_SaverDef_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.util.SaverDef.class, org.tensorflow.proto.util.SaverDef.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.util.SaverDef.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      filenameTensorName_ = "";
+
+      saveTensorName_ = "";
+
+      restoreOpName_ = "";
+
+      maxToKeep_ = 0;
+
+      sharded_ = false;
+
+      keepCheckpointEveryNHours_ = 0F;
+
+      version_ = 0;
+
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.util.SaverProtos.internal_static_tensorflow_SaverDef_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.SaverDef getDefaultInstanceForType() {
+      return org.tensorflow.proto.util.SaverDef.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.SaverDef build() {
+      org.tensorflow.proto.util.SaverDef result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.SaverDef buildPartial() {
+      org.tensorflow.proto.util.SaverDef result = new org.tensorflow.proto.util.SaverDef(this);
+      result.filenameTensorName_ = filenameTensorName_;
+      result.saveTensorName_ = saveTensorName_;
+      result.restoreOpName_ = restoreOpName_;
+      result.maxToKeep_ = maxToKeep_;
+      result.sharded_ = sharded_;
+      result.keepCheckpointEveryNHours_ = keepCheckpointEveryNHours_;
+      result.version_ = version_;
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.util.SaverDef) {
+        return mergeFrom((org.tensorflow.proto.util.SaverDef)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.util.SaverDef other) {
+      if (other == org.tensorflow.proto.util.SaverDef.getDefaultInstance()) return this;
+      if (!other.getFilenameTensorName().isEmpty()) {
+        filenameTensorName_ = other.filenameTensorName_;
+        onChanged();
+      }
+      if (!other.getSaveTensorName().isEmpty()) {
+        saveTensorName_ = other.saveTensorName_;
+        onChanged();
+      }
+      if (!other.getRestoreOpName().isEmpty()) {
+        restoreOpName_ = other.restoreOpName_;
+        onChanged();
+      }
+      if (other.getMaxToKeep() != 0) {
+        setMaxToKeep(other.getMaxToKeep());
+      }
+      if (other.getSharded() != false) {
+        setSharded(other.getSharded());
+      }
+      if (other.getKeepCheckpointEveryNHours() != 0F) {
+        setKeepCheckpointEveryNHours(other.getKeepCheckpointEveryNHours());
+      }
+      if (other.version_ != 0) {
+        setVersionValue(other.getVersionValue());
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.util.SaverDef parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.util.SaverDef) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+
+    private java.lang.Object filenameTensorName_ = "";
+    /**
+     * <pre>
+     * The name of the tensor in which to specify the filename when saving or
+     * restoring a model checkpoint.
+     * </pre>
+     *
+     * <code>string filename_tensor_name = 1;</code>
+     */
+    public java.lang.String getFilenameTensorName() {
+      java.lang.Object ref = filenameTensorName_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        filenameTensorName_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * The name of the tensor in which to specify the filename when saving or
+     * restoring a model checkpoint.
+     * </pre>
+     *
+     * <code>string filename_tensor_name = 1;</code>
+     */
+    public com.google.protobuf.ByteString
+        getFilenameTensorNameBytes() {
+      java.lang.Object ref = filenameTensorName_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        filenameTensorName_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * The name of the tensor in which to specify the filename when saving or
+     * restoring a model checkpoint.
+     * </pre>
+     *
+     * <code>string filename_tensor_name = 1;</code>
+     */
+    public Builder setFilenameTensorName(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      filenameTensorName_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * The name of the tensor in which to specify the filename when saving or
+     * restoring a model checkpoint.
+     * </pre>
+     *
+     * <code>string filename_tensor_name = 1;</code>
+     */
+    public Builder clearFilenameTensorName() {
+      
+      filenameTensorName_ = getDefaultInstance().getFilenameTensorName();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * The name of the tensor in which to specify the filename when saving or
+     * restoring a model checkpoint.
+     * </pre>
+     *
+     * <code>string filename_tensor_name = 1;</code>
+     */
+    public Builder setFilenameTensorNameBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      filenameTensorName_ = value;
+      onChanged();
+      return this;
+    }
+
+    private java.lang.Object saveTensorName_ = "";
+    /**
+     * <pre>
+     * The operation to run when saving a model checkpoint.
+     * </pre>
+     *
+     * <code>string save_tensor_name = 2;</code>
+     */
+    public java.lang.String getSaveTensorName() {
+      java.lang.Object ref = saveTensorName_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        saveTensorName_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * The operation to run when saving a model checkpoint.
+     * </pre>
+     *
+     * <code>string save_tensor_name = 2;</code>
+     */
+    public com.google.protobuf.ByteString
+        getSaveTensorNameBytes() {
+      java.lang.Object ref = saveTensorName_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        saveTensorName_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * The operation to run when saving a model checkpoint.
+     * </pre>
+     *
+     * <code>string save_tensor_name = 2;</code>
+     */
+    public Builder setSaveTensorName(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      saveTensorName_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * The operation to run when saving a model checkpoint.
+     * </pre>
+     *
+     * <code>string save_tensor_name = 2;</code>
+     */
+    public Builder clearSaveTensorName() {
+      
+      saveTensorName_ = getDefaultInstance().getSaveTensorName();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * The operation to run when saving a model checkpoint.
+     * </pre>
+     *
+     * <code>string save_tensor_name = 2;</code>
+     */
+    public Builder setSaveTensorNameBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      saveTensorName_ = value;
+      onChanged();
+      return this;
+    }
+
+    private java.lang.Object restoreOpName_ = "";
+    /**
+     * <pre>
+     * The operation to run when restoring a model checkpoint.
+     * </pre>
+     *
+     * <code>string restore_op_name = 3;</code>
+     */
+    public java.lang.String getRestoreOpName() {
+      java.lang.Object ref = restoreOpName_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        restoreOpName_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * The operation to run when restoring a model checkpoint.
+     * </pre>
+     *
+     * <code>string restore_op_name = 3;</code>
+     */
+    public com.google.protobuf.ByteString
+        getRestoreOpNameBytes() {
+      java.lang.Object ref = restoreOpName_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        restoreOpName_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * The operation to run when restoring a model checkpoint.
+     * </pre>
+     *
+     * <code>string restore_op_name = 3;</code>
+     */
+    public Builder setRestoreOpName(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      restoreOpName_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * The operation to run when restoring a model checkpoint.
+     * </pre>
+     *
+     * <code>string restore_op_name = 3;</code>
+     */
+    public Builder clearRestoreOpName() {
+      
+      restoreOpName_ = getDefaultInstance().getRestoreOpName();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * The operation to run when restoring a model checkpoint.
+     * </pre>
+     *
+     * <code>string restore_op_name = 3;</code>
+     */
+    public Builder setRestoreOpNameBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      restoreOpName_ = value;
+      onChanged();
+      return this;
+    }
+
+    private int maxToKeep_ ;
+    /**
+     * <pre>
+     * Maximum number of checkpoints to keep.  If 0, no checkpoints are deleted.
+     * </pre>
+     *
+     * <code>int32 max_to_keep = 4;</code>
+     */
+    public int getMaxToKeep() {
+      return maxToKeep_;
+    }
+    /**
+     * <pre>
+     * Maximum number of checkpoints to keep.  If 0, no checkpoints are deleted.
+     * </pre>
+     *
+     * <code>int32 max_to_keep = 4;</code>
+     */
+    public Builder setMaxToKeep(int value) {
+      
+      maxToKeep_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Maximum number of checkpoints to keep.  If 0, no checkpoints are deleted.
+     * </pre>
+     *
+     * <code>int32 max_to_keep = 4;</code>
+     */
+    public Builder clearMaxToKeep() {
+      
+      maxToKeep_ = 0;
+      onChanged();
+      return this;
+    }
+
+    private boolean sharded_ ;
+    /**
+     * <pre>
+     * Shard the save files, one per device that has Variable nodes.
+     * </pre>
+     *
+     * <code>bool sharded = 5;</code>
+     */
+    public boolean getSharded() {
+      return sharded_;
+    }
+    /**
+     * <pre>
+     * Shard the save files, one per device that has Variable nodes.
+     * </pre>
+     *
+     * <code>bool sharded = 5;</code>
+     */
+    public Builder setSharded(boolean value) {
+      
+      sharded_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Shard the save files, one per device that has Variable nodes.
+     * </pre>
+     *
+     * <code>bool sharded = 5;</code>
+     */
+    public Builder clearSharded() {
+      
+      sharded_ = false;
+      onChanged();
+      return this;
+    }
+
+    private float keepCheckpointEveryNHours_ ;
+    /**
+     * <pre>
+     * How often to keep an additional checkpoint. If not specified, only the last
+     * "max_to_keep" checkpoints are kept; if specified, in addition to keeping
+     * the last "max_to_keep" checkpoints, an additional checkpoint will be kept
+     * for every n hours of training.
+     * </pre>
+     *
+     * <code>float keep_checkpoint_every_n_hours = 6;</code>
+     */
+    public float getKeepCheckpointEveryNHours() {
+      return keepCheckpointEveryNHours_;
+    }
+    /**
+     * <pre>
+     * How often to keep an additional checkpoint. If not specified, only the last
+     * "max_to_keep" checkpoints are kept; if specified, in addition to keeping
+     * the last "max_to_keep" checkpoints, an additional checkpoint will be kept
+     * for every n hours of training.
+     * </pre>
+     *
+     * <code>float keep_checkpoint_every_n_hours = 6;</code>
+     */
+    public Builder setKeepCheckpointEveryNHours(float value) {
+      
+      keepCheckpointEveryNHours_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * How often to keep an additional checkpoint. If not specified, only the last
+     * "max_to_keep" checkpoints are kept; if specified, in addition to keeping
+     * the last "max_to_keep" checkpoints, an additional checkpoint will be kept
+     * for every n hours of training.
+     * </pre>
+     *
+     * <code>float keep_checkpoint_every_n_hours = 6;</code>
+     */
+    public Builder clearKeepCheckpointEveryNHours() {
+      
+      keepCheckpointEveryNHours_ = 0F;
+      onChanged();
+      return this;
+    }
+
+    private int version_ = 0;
+    /**
+     * <code>.tensorflow.SaverDef.CheckpointFormatVersion version = 7;</code>
+     */
+    public int getVersionValue() {
+      return version_;
+    }
+    /**
+     * <code>.tensorflow.SaverDef.CheckpointFormatVersion version = 7;</code>
+     */
+    public Builder setVersionValue(int value) {
+      version_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>.tensorflow.SaverDef.CheckpointFormatVersion version = 7;</code>
+     */
+    public org.tensorflow.proto.util.SaverDef.CheckpointFormatVersion getVersion() {
+      @SuppressWarnings("deprecation")
+      org.tensorflow.proto.util.SaverDef.CheckpointFormatVersion result = org.tensorflow.proto.util.SaverDef.CheckpointFormatVersion.valueOf(version_);
+      return result == null ? org.tensorflow.proto.util.SaverDef.CheckpointFormatVersion.UNRECOGNIZED : result;
+    }
+    /**
+     * <code>.tensorflow.SaverDef.CheckpointFormatVersion version = 7;</code>
+     */
+    public Builder setVersion(org.tensorflow.proto.util.SaverDef.CheckpointFormatVersion value) {
+      if (value == null) {
+        throw new NullPointerException();
+      }
+      
+      version_ = value.getNumber();
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>.tensorflow.SaverDef.CheckpointFormatVersion version = 7;</code>
+     */
+    public Builder clearVersion() {
+      
+      version_ = 0;
+      onChanged();
+      return this;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.SaverDef)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.SaverDef)
+  private static final org.tensorflow.proto.util.SaverDef DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.util.SaverDef();
+  }
+
+  public static org.tensorflow.proto.util.SaverDef getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<SaverDef>
+      PARSER = new com.google.protobuf.AbstractParser<SaverDef>() {
+    @java.lang.Override
+    public SaverDef parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new SaverDef(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<SaverDef> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<SaverDef> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.util.SaverDef getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SaverDefOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SaverDefOrBuilder.java
new file mode 100644
index 00000000000..6b2d7fa9e48
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SaverDefOrBuilder.java
@@ -0,0 +1,104 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/saver.proto
+
+package org.tensorflow.proto.util;
+
+public interface SaverDefOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.SaverDef)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <pre>
+   * The name of the tensor in which to specify the filename when saving or
+   * restoring a model checkpoint.
+   * </pre>
+   *
+   * <code>string filename_tensor_name = 1;</code>
+   */
+  java.lang.String getFilenameTensorName();
+  /**
+   * <pre>
+   * The name of the tensor in which to specify the filename when saving or
+   * restoring a model checkpoint.
+   * </pre>
+   *
+   * <code>string filename_tensor_name = 1;</code>
+   */
+  com.google.protobuf.ByteString
+      getFilenameTensorNameBytes();
+
+  /**
+   * <pre>
+   * The operation to run when saving a model checkpoint.
+   * </pre>
+   *
+   * <code>string save_tensor_name = 2;</code>
+   */
+  java.lang.String getSaveTensorName();
+  /**
+   * <pre>
+   * The operation to run when saving a model checkpoint.
+   * </pre>
+   *
+   * <code>string save_tensor_name = 2;</code>
+   */
+  com.google.protobuf.ByteString
+      getSaveTensorNameBytes();
+
+  /**
+   * <pre>
+   * The operation to run when restoring a model checkpoint.
+   * </pre>
+   *
+   * <code>string restore_op_name = 3;</code>
+   */
+  java.lang.String getRestoreOpName();
+  /**
+   * <pre>
+   * The operation to run when restoring a model checkpoint.
+   * </pre>
+   *
+   * <code>string restore_op_name = 3;</code>
+   */
+  com.google.protobuf.ByteString
+      getRestoreOpNameBytes();
+
+  /**
+   * <pre>
+   * Maximum number of checkpoints to keep.  If 0, no checkpoints are deleted.
+   * </pre>
+   *
+   * <code>int32 max_to_keep = 4;</code>
+   */
+  int getMaxToKeep();
+
+  /**
+   * <pre>
+   * Shard the save files, one per device that has Variable nodes.
+   * </pre>
+   *
+   * <code>bool sharded = 5;</code>
+   */
+  boolean getSharded();
+
+  /**
+   * <pre>
+   * How often to keep an additional checkpoint. If not specified, only the last
+   * "max_to_keep" checkpoints are kept; if specified, in addition to keeping
+   * the last "max_to_keep" checkpoints, an additional checkpoint will be kept
+   * for every n hours of training.
+   * </pre>
+   *
+   * <code>float keep_checkpoint_every_n_hours = 6;</code>
+   */
+  float getKeepCheckpointEveryNHours();
+
+  /**
+   * <code>.tensorflow.SaverDef.CheckpointFormatVersion version = 7;</code>
+   */
+  int getVersionValue();
+  /**
+   * <code>.tensorflow.SaverDef.CheckpointFormatVersion version = 7;</code>
+   */
+  org.tensorflow.proto.util.SaverDef.CheckpointFormatVersion getVersion();
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SaverProtos.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SaverProtos.java
new file mode 100644
index 00000000000..8739b58bf4f
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SaverProtos.java
@@ -0,0 +1,57 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/saver.proto
+
+package org.tensorflow.proto.util;
+
+public final class SaverProtos {
+  private SaverProtos() {}
+  public static void registerAllExtensions(
+      com.google.protobuf.ExtensionRegistryLite registry) {
+  }
+
+  public static void registerAllExtensions(
+      com.google.protobuf.ExtensionRegistry registry) {
+    registerAllExtensions(
+        (com.google.protobuf.ExtensionRegistryLite) registry);
+  }
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_SaverDef_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_SaverDef_fieldAccessorTable;
+
+  public static com.google.protobuf.Descriptors.FileDescriptor
+      getDescriptor() {
+    return descriptor;
+  }
+  private static  com.google.protobuf.Descriptors.FileDescriptor
+      descriptor;
+  static {
+    java.lang.String[] descriptorData = {
+      "\n$tensorflow/core/protobuf/saver.proto\022\n" +
+      "tensorflow\"\236\002\n\010SaverDef\022\034\n\024filename_tens" +
+      "or_name\030\001 \001(\t\022\030\n\020save_tensor_name\030\002 \001(\t\022" +
+      "\027\n\017restore_op_name\030\003 \001(\t\022\023\n\013max_to_keep\030" +
+      "\004 \001(\005\022\017\n\007sharded\030\005 \001(\010\022%\n\035keep_checkpoin" +
+      "t_every_n_hours\030\006 \001(\002\022=\n\007version\030\007 \001(\0162," +
+      ".tensorflow.SaverDef.CheckpointFormatVer" +
+      "sion\"5\n\027CheckpointFormatVersion\022\n\n\006LEGAC" +
+      "Y\020\000\022\006\n\002V1\020\001\022\006\n\002V2\020\002Bw\n\031org.tensorflow.pr" +
+      "oto.utilB\013SaverProtosP\001ZHgithub.com/tens" +
+      "orflow/tensorflow/tensorflow/go/core/cor" +
+      "e_protos_go_proto\370\001\001b\006proto3"
+    };
+    descriptor = com.google.protobuf.Descriptors.FileDescriptor
+      .internalBuildGeneratedFileFrom(descriptorData,
+        new com.google.protobuf.Descriptors.FileDescriptor[] {
+        });
+    internal_static_tensorflow_SaverDef_descriptor =
+      getDescriptor().getMessageTypes().get(0);
+    internal_static_tensorflow_SaverDef_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_SaverDef_descriptor,
+        new java.lang.String[] { "FilenameTensorName", "SaveTensorName", "RestoreOpName", "MaxToKeep", "Sharded", "KeepCheckpointEveryNHours", "Version", });
+  }
+
+  // @@protoc_insertion_point(outer_class_scope)
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SessionLog.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SessionLog.java
new file mode 100644
index 00000000000..824dba910f9
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SessionLog.java
@@ -0,0 +1,910 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/util/event.proto
+
+package org.tensorflow.proto.util;
+
+/**
+ * <pre>
+ * Protocol buffer used for logging session state.
+ * </pre>
+ *
+ * Protobuf type {@code tensorflow.SessionLog}
+ */
+public  final class SessionLog extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.SessionLog)
+    SessionLogOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use SessionLog.newBuilder() to construct.
+  private SessionLog(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private SessionLog() {
+    status_ = 0;
+    checkpointPath_ = "";
+    msg_ = "";
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new SessionLog();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private SessionLog(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 8: {
+            int rawValue = input.readEnum();
+
+            status_ = rawValue;
+            break;
+          }
+          case 18: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            checkpointPath_ = s;
+            break;
+          }
+          case 26: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            msg_ = s;
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.util.EventProtos.internal_static_tensorflow_SessionLog_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.util.EventProtos.internal_static_tensorflow_SessionLog_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.util.SessionLog.class, org.tensorflow.proto.util.SessionLog.Builder.class);
+  }
+
+  /**
+   * Protobuf enum {@code tensorflow.SessionLog.SessionStatus}
+   */
+  public enum SessionStatus
+      implements com.google.protobuf.ProtocolMessageEnum {
+    /**
+     * <code>STATUS_UNSPECIFIED = 0;</code>
+     */
+    STATUS_UNSPECIFIED(0),
+    /**
+     * <code>START = 1;</code>
+     */
+    START(1),
+    /**
+     * <code>STOP = 2;</code>
+     */
+    STOP(2),
+    /**
+     * <code>CHECKPOINT = 3;</code>
+     */
+    CHECKPOINT(3),
+    UNRECOGNIZED(-1),
+    ;
+
+    /**
+     * <code>STATUS_UNSPECIFIED = 0;</code>
+     */
+    public static final int STATUS_UNSPECIFIED_VALUE = 0;
+    /**
+     * <code>START = 1;</code>
+     */
+    public static final int START_VALUE = 1;
+    /**
+     * <code>STOP = 2;</code>
+     */
+    public static final int STOP_VALUE = 2;
+    /**
+     * <code>CHECKPOINT = 3;</code>
+     */
+    public static final int CHECKPOINT_VALUE = 3;
+
+
+    public final int getNumber() {
+      if (this == UNRECOGNIZED) {
+        throw new java.lang.IllegalArgumentException(
+            "Can't get the number of an unknown enum value.");
+      }
+      return value;
+    }
+
+    /**
+     * @deprecated Use {@link #forNumber(int)} instead.
+     */
+    @java.lang.Deprecated
+    public static SessionStatus valueOf(int value) {
+      return forNumber(value);
+    }
+
+    public static SessionStatus forNumber(int value) {
+      switch (value) {
+        case 0: return STATUS_UNSPECIFIED;
+        case 1: return START;
+        case 2: return STOP;
+        case 3: return CHECKPOINT;
+        default: return null;
+      }
+    }
+
+    public static com.google.protobuf.Internal.EnumLiteMap<SessionStatus>
+        internalGetValueMap() {
+      return internalValueMap;
+    }
+    private static final com.google.protobuf.Internal.EnumLiteMap<
+        SessionStatus> internalValueMap =
+          new com.google.protobuf.Internal.EnumLiteMap<SessionStatus>() {
+            public SessionStatus findValueByNumber(int number) {
+              return SessionStatus.forNumber(number);
+            }
+          };
+
+    public final com.google.protobuf.Descriptors.EnumValueDescriptor
+        getValueDescriptor() {
+      return getDescriptor().getValues().get(ordinal());
+    }
+    public final com.google.protobuf.Descriptors.EnumDescriptor
+        getDescriptorForType() {
+      return getDescriptor();
+    }
+    public static final com.google.protobuf.Descriptors.EnumDescriptor
+        getDescriptor() {
+      return org.tensorflow.proto.util.SessionLog.getDescriptor().getEnumTypes().get(0);
+    }
+
+    private static final SessionStatus[] VALUES = values();
+
+    public static SessionStatus valueOf(
+        com.google.protobuf.Descriptors.EnumValueDescriptor desc) {
+      if (desc.getType() != getDescriptor()) {
+        throw new java.lang.IllegalArgumentException(
+          "EnumValueDescriptor is not for this type.");
+      }
+      if (desc.getIndex() == -1) {
+        return UNRECOGNIZED;
+      }
+      return VALUES[desc.getIndex()];
+    }
+
+    private final int value;
+
+    private SessionStatus(int value) {
+      this.value = value;
+    }
+
+    // @@protoc_insertion_point(enum_scope:tensorflow.SessionLog.SessionStatus)
+  }
+
+  public static final int STATUS_FIELD_NUMBER = 1;
+  private int status_;
+  /**
+   * <code>.tensorflow.SessionLog.SessionStatus status = 1;</code>
+   */
+  public int getStatusValue() {
+    return status_;
+  }
+  /**
+   * <code>.tensorflow.SessionLog.SessionStatus status = 1;</code>
+   */
+  public org.tensorflow.proto.util.SessionLog.SessionStatus getStatus() {
+    @SuppressWarnings("deprecation")
+    org.tensorflow.proto.util.SessionLog.SessionStatus result = org.tensorflow.proto.util.SessionLog.SessionStatus.valueOf(status_);
+    return result == null ? org.tensorflow.proto.util.SessionLog.SessionStatus.UNRECOGNIZED : result;
+  }
+
+  public static final int CHECKPOINT_PATH_FIELD_NUMBER = 2;
+  private volatile java.lang.Object checkpointPath_;
+  /**
+   * <pre>
+   * This checkpoint_path contains both the path and filename.
+   * </pre>
+   *
+   * <code>string checkpoint_path = 2;</code>
+   */
+  public java.lang.String getCheckpointPath() {
+    java.lang.Object ref = checkpointPath_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      checkpointPath_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * This checkpoint_path contains both the path and filename.
+   * </pre>
+   *
+   * <code>string checkpoint_path = 2;</code>
+   */
+  public com.google.protobuf.ByteString
+      getCheckpointPathBytes() {
+    java.lang.Object ref = checkpointPath_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      checkpointPath_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int MSG_FIELD_NUMBER = 3;
+  private volatile java.lang.Object msg_;
+  /**
+   * <code>string msg = 3;</code>
+   */
+  public java.lang.String getMsg() {
+    java.lang.Object ref = msg_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      msg_ = s;
+      return s;
+    }
+  }
+  /**
+   * <code>string msg = 3;</code>
+   */
+  public com.google.protobuf.ByteString
+      getMsgBytes() {
+    java.lang.Object ref = msg_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      msg_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    if (status_ != org.tensorflow.proto.util.SessionLog.SessionStatus.STATUS_UNSPECIFIED.getNumber()) {
+      output.writeEnum(1, status_);
+    }
+    if (!getCheckpointPathBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 2, checkpointPath_);
+    }
+    if (!getMsgBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 3, msg_);
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (status_ != org.tensorflow.proto.util.SessionLog.SessionStatus.STATUS_UNSPECIFIED.getNumber()) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeEnumSize(1, status_);
+    }
+    if (!getCheckpointPathBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(2, checkpointPath_);
+    }
+    if (!getMsgBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(3, msg_);
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.util.SessionLog)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.util.SessionLog other = (org.tensorflow.proto.util.SessionLog) obj;
+
+    if (status_ != other.status_) return false;
+    if (!getCheckpointPath()
+        .equals(other.getCheckpointPath())) return false;
+    if (!getMsg()
+        .equals(other.getMsg())) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    hash = (37 * hash) + STATUS_FIELD_NUMBER;
+    hash = (53 * hash) + status_;
+    hash = (37 * hash) + CHECKPOINT_PATH_FIELD_NUMBER;
+    hash = (53 * hash) + getCheckpointPath().hashCode();
+    hash = (37 * hash) + MSG_FIELD_NUMBER;
+    hash = (53 * hash) + getMsg().hashCode();
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.util.SessionLog parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.SessionLog parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.SessionLog parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.SessionLog parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.SessionLog parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.SessionLog parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.SessionLog parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.SessionLog parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.SessionLog parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.SessionLog parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.SessionLog parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.SessionLog parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.util.SessionLog prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * <pre>
+   * Protocol buffer used for logging session state.
+   * </pre>
+   *
+   * Protobuf type {@code tensorflow.SessionLog}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.SessionLog)
+      org.tensorflow.proto.util.SessionLogOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.util.EventProtos.internal_static_tensorflow_SessionLog_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.util.EventProtos.internal_static_tensorflow_SessionLog_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.util.SessionLog.class, org.tensorflow.proto.util.SessionLog.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.util.SessionLog.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      status_ = 0;
+
+      checkpointPath_ = "";
+
+      msg_ = "";
+
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.util.EventProtos.internal_static_tensorflow_SessionLog_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.SessionLog getDefaultInstanceForType() {
+      return org.tensorflow.proto.util.SessionLog.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.SessionLog build() {
+      org.tensorflow.proto.util.SessionLog result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.SessionLog buildPartial() {
+      org.tensorflow.proto.util.SessionLog result = new org.tensorflow.proto.util.SessionLog(this);
+      result.status_ = status_;
+      result.checkpointPath_ = checkpointPath_;
+      result.msg_ = msg_;
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.util.SessionLog) {
+        return mergeFrom((org.tensorflow.proto.util.SessionLog)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.util.SessionLog other) {
+      if (other == org.tensorflow.proto.util.SessionLog.getDefaultInstance()) return this;
+      if (other.status_ != 0) {
+        setStatusValue(other.getStatusValue());
+      }
+      if (!other.getCheckpointPath().isEmpty()) {
+        checkpointPath_ = other.checkpointPath_;
+        onChanged();
+      }
+      if (!other.getMsg().isEmpty()) {
+        msg_ = other.msg_;
+        onChanged();
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.util.SessionLog parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.util.SessionLog) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+
+    private int status_ = 0;
+    /**
+     * <code>.tensorflow.SessionLog.SessionStatus status = 1;</code>
+     */
+    public int getStatusValue() {
+      return status_;
+    }
+    /**
+     * <code>.tensorflow.SessionLog.SessionStatus status = 1;</code>
+     */
+    public Builder setStatusValue(int value) {
+      status_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>.tensorflow.SessionLog.SessionStatus status = 1;</code>
+     */
+    public org.tensorflow.proto.util.SessionLog.SessionStatus getStatus() {
+      @SuppressWarnings("deprecation")
+      org.tensorflow.proto.util.SessionLog.SessionStatus result = org.tensorflow.proto.util.SessionLog.SessionStatus.valueOf(status_);
+      return result == null ? org.tensorflow.proto.util.SessionLog.SessionStatus.UNRECOGNIZED : result;
+    }
+    /**
+     * <code>.tensorflow.SessionLog.SessionStatus status = 1;</code>
+     */
+    public Builder setStatus(org.tensorflow.proto.util.SessionLog.SessionStatus value) {
+      if (value == null) {
+        throw new NullPointerException();
+      }
+      
+      status_ = value.getNumber();
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>.tensorflow.SessionLog.SessionStatus status = 1;</code>
+     */
+    public Builder clearStatus() {
+      
+      status_ = 0;
+      onChanged();
+      return this;
+    }
+
+    private java.lang.Object checkpointPath_ = "";
+    /**
+     * <pre>
+     * This checkpoint_path contains both the path and filename.
+     * </pre>
+     *
+     * <code>string checkpoint_path = 2;</code>
+     */
+    public java.lang.String getCheckpointPath() {
+      java.lang.Object ref = checkpointPath_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        checkpointPath_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * This checkpoint_path contains both the path and filename.
+     * </pre>
+     *
+     * <code>string checkpoint_path = 2;</code>
+     */
+    public com.google.protobuf.ByteString
+        getCheckpointPathBytes() {
+      java.lang.Object ref = checkpointPath_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        checkpointPath_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * This checkpoint_path contains both the path and filename.
+     * </pre>
+     *
+     * <code>string checkpoint_path = 2;</code>
+     */
+    public Builder setCheckpointPath(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      checkpointPath_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * This checkpoint_path contains both the path and filename.
+     * </pre>
+     *
+     * <code>string checkpoint_path = 2;</code>
+     */
+    public Builder clearCheckpointPath() {
+      
+      checkpointPath_ = getDefaultInstance().getCheckpointPath();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * This checkpoint_path contains both the path and filename.
+     * </pre>
+     *
+     * <code>string checkpoint_path = 2;</code>
+     */
+    public Builder setCheckpointPathBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      checkpointPath_ = value;
+      onChanged();
+      return this;
+    }
+
+    private java.lang.Object msg_ = "";
+    /**
+     * <code>string msg = 3;</code>
+     */
+    public java.lang.String getMsg() {
+      java.lang.Object ref = msg_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        msg_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <code>string msg = 3;</code>
+     */
+    public com.google.protobuf.ByteString
+        getMsgBytes() {
+      java.lang.Object ref = msg_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        msg_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <code>string msg = 3;</code>
+     */
+    public Builder setMsg(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      msg_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>string msg = 3;</code>
+     */
+    public Builder clearMsg() {
+      
+      msg_ = getDefaultInstance().getMsg();
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>string msg = 3;</code>
+     */
+    public Builder setMsgBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      msg_ = value;
+      onChanged();
+      return this;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.SessionLog)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.SessionLog)
+  private static final org.tensorflow.proto.util.SessionLog DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.util.SessionLog();
+  }
+
+  public static org.tensorflow.proto.util.SessionLog getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<SessionLog>
+      PARSER = new com.google.protobuf.AbstractParser<SessionLog>() {
+    @java.lang.Override
+    public SessionLog parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new SessionLog(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<SessionLog> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<SessionLog> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.util.SessionLog getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SessionLogOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SessionLogOrBuilder.java
new file mode 100644
index 00000000000..84ba691a95d
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SessionLogOrBuilder.java
@@ -0,0 +1,46 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/util/event.proto
+
+package org.tensorflow.proto.util;
+
+public interface SessionLogOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.SessionLog)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <code>.tensorflow.SessionLog.SessionStatus status = 1;</code>
+   */
+  int getStatusValue();
+  /**
+   * <code>.tensorflow.SessionLog.SessionStatus status = 1;</code>
+   */
+  org.tensorflow.proto.util.SessionLog.SessionStatus getStatus();
+
+  /**
+   * <pre>
+   * This checkpoint_path contains both the path and filename.
+   * </pre>
+   *
+   * <code>string checkpoint_path = 2;</code>
+   */
+  java.lang.String getCheckpointPath();
+  /**
+   * <pre>
+   * This checkpoint_path contains both the path and filename.
+   * </pre>
+   *
+   * <code>string checkpoint_path = 2;</code>
+   */
+  com.google.protobuf.ByteString
+      getCheckpointPathBytes();
+
+  /**
+   * <code>string msg = 3;</code>
+   */
+  java.lang.String getMsg();
+  /**
+   * <code>string msg = 3;</code>
+   */
+  com.google.protobuf.ByteString
+      getMsgBytes();
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SnapShot.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SnapShot.java
new file mode 100644
index 00000000000..bde929e4c6d
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SnapShot.java
@@ -0,0 +1,535 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/bfc_memory_map.proto
+
+package org.tensorflow.proto.util;
+
+/**
+ * Protobuf type {@code tensorflow.SnapShot}
+ */
+public  final class SnapShot extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.SnapShot)
+    SnapShotOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use SnapShot.newBuilder() to construct.
+  private SnapShot(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private SnapShot() {
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new SnapShot();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private SnapShot(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 8: {
+
+            actionCount_ = input.readUInt64();
+            break;
+          }
+          case 16: {
+
+            size_ = input.readInt64();
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.util.BfcMemoryMapProtos.internal_static_tensorflow_SnapShot_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.util.BfcMemoryMapProtos.internal_static_tensorflow_SnapShot_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.util.SnapShot.class, org.tensorflow.proto.util.SnapShot.Builder.class);
+  }
+
+  public static final int ACTION_COUNT_FIELD_NUMBER = 1;
+  private long actionCount_;
+  /**
+   * <code>uint64 action_count = 1;</code>
+   */
+  public long getActionCount() {
+    return actionCount_;
+  }
+
+  public static final int SIZE_FIELD_NUMBER = 2;
+  private long size_;
+  /**
+   * <code>int64 size = 2;</code>
+   */
+  public long getSize() {
+    return size_;
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    if (actionCount_ != 0L) {
+      output.writeUInt64(1, actionCount_);
+    }
+    if (size_ != 0L) {
+      output.writeInt64(2, size_);
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (actionCount_ != 0L) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeUInt64Size(1, actionCount_);
+    }
+    if (size_ != 0L) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt64Size(2, size_);
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.util.SnapShot)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.util.SnapShot other = (org.tensorflow.proto.util.SnapShot) obj;
+
+    if (getActionCount()
+        != other.getActionCount()) return false;
+    if (getSize()
+        != other.getSize()) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    hash = (37 * hash) + ACTION_COUNT_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+        getActionCount());
+    hash = (37 * hash) + SIZE_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+        getSize());
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.util.SnapShot parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.SnapShot parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.SnapShot parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.SnapShot parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.SnapShot parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.SnapShot parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.SnapShot parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.SnapShot parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.SnapShot parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.SnapShot parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.SnapShot parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.SnapShot parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.util.SnapShot prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * Protobuf type {@code tensorflow.SnapShot}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.SnapShot)
+      org.tensorflow.proto.util.SnapShotOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.util.BfcMemoryMapProtos.internal_static_tensorflow_SnapShot_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.util.BfcMemoryMapProtos.internal_static_tensorflow_SnapShot_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.util.SnapShot.class, org.tensorflow.proto.util.SnapShot.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.util.SnapShot.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      actionCount_ = 0L;
+
+      size_ = 0L;
+
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.util.BfcMemoryMapProtos.internal_static_tensorflow_SnapShot_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.SnapShot getDefaultInstanceForType() {
+      return org.tensorflow.proto.util.SnapShot.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.SnapShot build() {
+      org.tensorflow.proto.util.SnapShot result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.SnapShot buildPartial() {
+      org.tensorflow.proto.util.SnapShot result = new org.tensorflow.proto.util.SnapShot(this);
+      result.actionCount_ = actionCount_;
+      result.size_ = size_;
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.util.SnapShot) {
+        return mergeFrom((org.tensorflow.proto.util.SnapShot)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.util.SnapShot other) {
+      if (other == org.tensorflow.proto.util.SnapShot.getDefaultInstance()) return this;
+      if (other.getActionCount() != 0L) {
+        setActionCount(other.getActionCount());
+      }
+      if (other.getSize() != 0L) {
+        setSize(other.getSize());
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.util.SnapShot parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.util.SnapShot) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+
+    private long actionCount_ ;
+    /**
+     * <code>uint64 action_count = 1;</code>
+     */
+    public long getActionCount() {
+      return actionCount_;
+    }
+    /**
+     * <code>uint64 action_count = 1;</code>
+     */
+    public Builder setActionCount(long value) {
+      
+      actionCount_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>uint64 action_count = 1;</code>
+     */
+    public Builder clearActionCount() {
+      
+      actionCount_ = 0L;
+      onChanged();
+      return this;
+    }
+
+    private long size_ ;
+    /**
+     * <code>int64 size = 2;</code>
+     */
+    public long getSize() {
+      return size_;
+    }
+    /**
+     * <code>int64 size = 2;</code>
+     */
+    public Builder setSize(long value) {
+      
+      size_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>int64 size = 2;</code>
+     */
+    public Builder clearSize() {
+      
+      size_ = 0L;
+      onChanged();
+      return this;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.SnapShot)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.SnapShot)
+  private static final org.tensorflow.proto.util.SnapShot DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.util.SnapShot();
+  }
+
+  public static org.tensorflow.proto.util.SnapShot getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<SnapShot>
+      PARSER = new com.google.protobuf.AbstractParser<SnapShot>() {
+    @java.lang.Override
+    public SnapShot parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new SnapShot(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<SnapShot> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<SnapShot> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.util.SnapShot getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SnapShotOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SnapShotOrBuilder.java
new file mode 100644
index 00000000000..787a23ded48
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SnapShotOrBuilder.java
@@ -0,0 +1,19 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/bfc_memory_map.proto
+
+package org.tensorflow.proto.util;
+
+public interface SnapShotOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.SnapShot)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <code>uint64 action_count = 1;</code>
+   */
+  long getActionCount();
+
+  /**
+   * <code>int64 size = 2;</code>
+   */
+  long getSize();
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SnapshotMetadataRecord.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SnapshotMetadataRecord.java
new file mode 100644
index 00000000000..36e10e21bb1
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SnapshotMetadataRecord.java
@@ -0,0 +1,1328 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/data/experimental/snapshot.proto
+
+package org.tensorflow.proto.util;
+
+/**
+ * <pre>
+ * This stores the metadata information present in each snapshot record.
+ * </pre>
+ *
+ * Protobuf type {@code tensorflow.data.experimental.SnapshotMetadataRecord}
+ */
+public  final class SnapshotMetadataRecord extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.data.experimental.SnapshotMetadataRecord)
+    SnapshotMetadataRecordOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use SnapshotMetadataRecord.newBuilder() to construct.
+  private SnapshotMetadataRecord(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private SnapshotMetadataRecord() {
+    graphHash_ = "";
+    runId_ = "";
+    dtype_ = java.util.Collections.emptyList();
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new SnapshotMetadataRecord();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private SnapshotMetadataRecord(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    int mutable_bitField0_ = 0;
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 10: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            graphHash_ = s;
+            break;
+          }
+          case 18: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            runId_ = s;
+            break;
+          }
+          case 24: {
+
+            creationTimestamp_ = input.readInt64();
+            break;
+          }
+          case 32: {
+
+            version_ = input.readInt64();
+            break;
+          }
+          case 40: {
+            int rawValue = input.readEnum();
+            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
+              dtype_ = new java.util.ArrayList<java.lang.Integer>();
+              mutable_bitField0_ |= 0x00000001;
+            }
+            dtype_.add(rawValue);
+            break;
+          }
+          case 42: {
+            int length = input.readRawVarint32();
+            int oldLimit = input.pushLimit(length);
+            while(input.getBytesUntilLimit() > 0) {
+              int rawValue = input.readEnum();
+              if (!((mutable_bitField0_ & 0x00000001) != 0)) {
+                dtype_ = new java.util.ArrayList<java.lang.Integer>();
+                mutable_bitField0_ |= 0x00000001;
+              }
+              dtype_.add(rawValue);
+            }
+            input.popLimit(oldLimit);
+            break;
+          }
+          case 48: {
+
+            numElements_ = input.readInt64();
+            break;
+          }
+          case 8000: {
+
+            finalized_ = input.readBool();
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      if (((mutable_bitField0_ & 0x00000001) != 0)) {
+        dtype_ = java.util.Collections.unmodifiableList(dtype_);
+      }
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.util.SnapshotProtos.internal_static_tensorflow_data_experimental_SnapshotMetadataRecord_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.util.SnapshotProtos.internal_static_tensorflow_data_experimental_SnapshotMetadataRecord_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.util.SnapshotMetadataRecord.class, org.tensorflow.proto.util.SnapshotMetadataRecord.Builder.class);
+  }
+
+  public static final int GRAPH_HASH_FIELD_NUMBER = 1;
+  private volatile java.lang.Object graphHash_;
+  /**
+   * <pre>
+   * Stores the fingerprint of the graph that describes the dataset that is
+   * snapshotted.
+   * </pre>
+   *
+   * <code>string graph_hash = 1;</code>
+   */
+  public java.lang.String getGraphHash() {
+    java.lang.Object ref = graphHash_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      graphHash_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * Stores the fingerprint of the graph that describes the dataset that is
+   * snapshotted.
+   * </pre>
+   *
+   * <code>string graph_hash = 1;</code>
+   */
+  public com.google.protobuf.ByteString
+      getGraphHashBytes() {
+    java.lang.Object ref = graphHash_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      graphHash_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int RUN_ID_FIELD_NUMBER = 2;
+  private volatile java.lang.Object runId_;
+  /**
+   * <pre>
+   * Run ID that this snapshot corresponds to.
+   * </pre>
+   *
+   * <code>string run_id = 2;</code>
+   */
+  public java.lang.String getRunId() {
+    java.lang.Object ref = runId_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      runId_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * Run ID that this snapshot corresponds to.
+   * </pre>
+   *
+   * <code>string run_id = 2;</code>
+   */
+  public com.google.protobuf.ByteString
+      getRunIdBytes() {
+    java.lang.Object ref = runId_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      runId_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int CREATION_TIMESTAMP_FIELD_NUMBER = 3;
+  private long creationTimestamp_;
+  /**
+   * <pre>
+   * Time when we started creating this snapshot.
+   * </pre>
+   *
+   * <code>int64 creation_timestamp = 3;</code>
+   */
+  public long getCreationTimestamp() {
+    return creationTimestamp_;
+  }
+
+  public static final int VERSION_FIELD_NUMBER = 4;
+  private long version_;
+  /**
+   * <pre>
+   * Version of the snapshot data file format.
+   * </pre>
+   *
+   * <code>int64 version = 4;</code>
+   */
+  public long getVersion() {
+    return version_;
+  }
+
+  public static final int DTYPE_FIELD_NUMBER = 5;
+  private java.util.List<java.lang.Integer> dtype_;
+  private static final com.google.protobuf.Internal.ListAdapter.Converter<
+      java.lang.Integer, org.tensorflow.proto.framework.DataType> dtype_converter_ =
+          new com.google.protobuf.Internal.ListAdapter.Converter<
+              java.lang.Integer, org.tensorflow.proto.framework.DataType>() {
+            public org.tensorflow.proto.framework.DataType convert(java.lang.Integer from) {
+              @SuppressWarnings("deprecation")
+              org.tensorflow.proto.framework.DataType result = org.tensorflow.proto.framework.DataType.valueOf(from);
+              return result == null ? org.tensorflow.proto.framework.DataType.UNRECOGNIZED : result;
+            }
+          };
+  /**
+   * <pre>
+   * A list of tensor dtype corresponding to each element of the snapshot.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.DataType dtype = 5;</code>
+   */
+  public java.util.List<org.tensorflow.proto.framework.DataType> getDtypeList() {
+    return new com.google.protobuf.Internal.ListAdapter<
+        java.lang.Integer, org.tensorflow.proto.framework.DataType>(dtype_, dtype_converter_);
+  }
+  /**
+   * <pre>
+   * A list of tensor dtype corresponding to each element of the snapshot.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.DataType dtype = 5;</code>
+   */
+  public int getDtypeCount() {
+    return dtype_.size();
+  }
+  /**
+   * <pre>
+   * A list of tensor dtype corresponding to each element of the snapshot.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.DataType dtype = 5;</code>
+   */
+  public org.tensorflow.proto.framework.DataType getDtype(int index) {
+    return dtype_converter_.convert(dtype_.get(index));
+  }
+  /**
+   * <pre>
+   * A list of tensor dtype corresponding to each element of the snapshot.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.DataType dtype = 5;</code>
+   */
+  public java.util.List<java.lang.Integer>
+  getDtypeValueList() {
+    return dtype_;
+  }
+  /**
+   * <pre>
+   * A list of tensor dtype corresponding to each element of the snapshot.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.DataType dtype = 5;</code>
+   */
+  public int getDtypeValue(int index) {
+    return dtype_.get(index);
+  }
+  private int dtypeMemoizedSerializedSize;
+
+  public static final int NUM_ELEMENTS_FIELD_NUMBER = 6;
+  private long numElements_;
+  /**
+   * <pre>
+   * The number of elements in the snapshot.
+   * </pre>
+   *
+   * <code>int64 num_elements = 6;</code>
+   */
+  public long getNumElements() {
+    return numElements_;
+  }
+
+  public static final int FINALIZED_FIELD_NUMBER = 1000;
+  private boolean finalized_;
+  /**
+   * <code>bool finalized = 1000;</code>
+   */
+  public boolean getFinalized() {
+    return finalized_;
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    getSerializedSize();
+    if (!getGraphHashBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 1, graphHash_);
+    }
+    if (!getRunIdBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 2, runId_);
+    }
+    if (creationTimestamp_ != 0L) {
+      output.writeInt64(3, creationTimestamp_);
+    }
+    if (version_ != 0L) {
+      output.writeInt64(4, version_);
+    }
+    if (getDtypeList().size() > 0) {
+      output.writeUInt32NoTag(42);
+      output.writeUInt32NoTag(dtypeMemoizedSerializedSize);
+    }
+    for (int i = 0; i < dtype_.size(); i++) {
+      output.writeEnumNoTag(dtype_.get(i));
+    }
+    if (numElements_ != 0L) {
+      output.writeInt64(6, numElements_);
+    }
+    if (finalized_ != false) {
+      output.writeBool(1000, finalized_);
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (!getGraphHashBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, graphHash_);
+    }
+    if (!getRunIdBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(2, runId_);
+    }
+    if (creationTimestamp_ != 0L) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt64Size(3, creationTimestamp_);
+    }
+    if (version_ != 0L) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt64Size(4, version_);
+    }
+    {
+      int dataSize = 0;
+      for (int i = 0; i < dtype_.size(); i++) {
+        dataSize += com.google.protobuf.CodedOutputStream
+          .computeEnumSizeNoTag(dtype_.get(i));
+      }
+      size += dataSize;
+      if (!getDtypeList().isEmpty()) {  size += 1;
+        size += com.google.protobuf.CodedOutputStream
+          .computeUInt32SizeNoTag(dataSize);
+      }dtypeMemoizedSerializedSize = dataSize;
+    }
+    if (numElements_ != 0L) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt64Size(6, numElements_);
+    }
+    if (finalized_ != false) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeBoolSize(1000, finalized_);
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.util.SnapshotMetadataRecord)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.util.SnapshotMetadataRecord other = (org.tensorflow.proto.util.SnapshotMetadataRecord) obj;
+
+    if (!getGraphHash()
+        .equals(other.getGraphHash())) return false;
+    if (!getRunId()
+        .equals(other.getRunId())) return false;
+    if (getCreationTimestamp()
+        != other.getCreationTimestamp()) return false;
+    if (getVersion()
+        != other.getVersion()) return false;
+    if (!dtype_.equals(other.dtype_)) return false;
+    if (getNumElements()
+        != other.getNumElements()) return false;
+    if (getFinalized()
+        != other.getFinalized()) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    hash = (37 * hash) + GRAPH_HASH_FIELD_NUMBER;
+    hash = (53 * hash) + getGraphHash().hashCode();
+    hash = (37 * hash) + RUN_ID_FIELD_NUMBER;
+    hash = (53 * hash) + getRunId().hashCode();
+    hash = (37 * hash) + CREATION_TIMESTAMP_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+        getCreationTimestamp());
+    hash = (37 * hash) + VERSION_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+        getVersion());
+    if (getDtypeCount() > 0) {
+      hash = (37 * hash) + DTYPE_FIELD_NUMBER;
+      hash = (53 * hash) + dtype_.hashCode();
+    }
+    hash = (37 * hash) + NUM_ELEMENTS_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+        getNumElements());
+    hash = (37 * hash) + FINALIZED_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashBoolean(
+        getFinalized());
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.util.SnapshotMetadataRecord parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.SnapshotMetadataRecord parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.SnapshotMetadataRecord parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.SnapshotMetadataRecord parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.SnapshotMetadataRecord parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.SnapshotMetadataRecord parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.SnapshotMetadataRecord parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.SnapshotMetadataRecord parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.SnapshotMetadataRecord parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.SnapshotMetadataRecord parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.SnapshotMetadataRecord parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.SnapshotMetadataRecord parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.util.SnapshotMetadataRecord prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * <pre>
+   * This stores the metadata information present in each snapshot record.
+   * </pre>
+   *
+   * Protobuf type {@code tensorflow.data.experimental.SnapshotMetadataRecord}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.data.experimental.SnapshotMetadataRecord)
+      org.tensorflow.proto.util.SnapshotMetadataRecordOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.util.SnapshotProtos.internal_static_tensorflow_data_experimental_SnapshotMetadataRecord_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.util.SnapshotProtos.internal_static_tensorflow_data_experimental_SnapshotMetadataRecord_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.util.SnapshotMetadataRecord.class, org.tensorflow.proto.util.SnapshotMetadataRecord.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.util.SnapshotMetadataRecord.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      graphHash_ = "";
+
+      runId_ = "";
+
+      creationTimestamp_ = 0L;
+
+      version_ = 0L;
+
+      dtype_ = java.util.Collections.emptyList();
+      bitField0_ = (bitField0_ & ~0x00000001);
+      numElements_ = 0L;
+
+      finalized_ = false;
+
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.util.SnapshotProtos.internal_static_tensorflow_data_experimental_SnapshotMetadataRecord_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.SnapshotMetadataRecord getDefaultInstanceForType() {
+      return org.tensorflow.proto.util.SnapshotMetadataRecord.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.SnapshotMetadataRecord build() {
+      org.tensorflow.proto.util.SnapshotMetadataRecord result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.SnapshotMetadataRecord buildPartial() {
+      org.tensorflow.proto.util.SnapshotMetadataRecord result = new org.tensorflow.proto.util.SnapshotMetadataRecord(this);
+      int from_bitField0_ = bitField0_;
+      result.graphHash_ = graphHash_;
+      result.runId_ = runId_;
+      result.creationTimestamp_ = creationTimestamp_;
+      result.version_ = version_;
+      if (((bitField0_ & 0x00000001) != 0)) {
+        dtype_ = java.util.Collections.unmodifiableList(dtype_);
+        bitField0_ = (bitField0_ & ~0x00000001);
+      }
+      result.dtype_ = dtype_;
+      result.numElements_ = numElements_;
+      result.finalized_ = finalized_;
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.util.SnapshotMetadataRecord) {
+        return mergeFrom((org.tensorflow.proto.util.SnapshotMetadataRecord)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.util.SnapshotMetadataRecord other) {
+      if (other == org.tensorflow.proto.util.SnapshotMetadataRecord.getDefaultInstance()) return this;
+      if (!other.getGraphHash().isEmpty()) {
+        graphHash_ = other.graphHash_;
+        onChanged();
+      }
+      if (!other.getRunId().isEmpty()) {
+        runId_ = other.runId_;
+        onChanged();
+      }
+      if (other.getCreationTimestamp() != 0L) {
+        setCreationTimestamp(other.getCreationTimestamp());
+      }
+      if (other.getVersion() != 0L) {
+        setVersion(other.getVersion());
+      }
+      if (!other.dtype_.isEmpty()) {
+        if (dtype_.isEmpty()) {
+          dtype_ = other.dtype_;
+          bitField0_ = (bitField0_ & ~0x00000001);
+        } else {
+          ensureDtypeIsMutable();
+          dtype_.addAll(other.dtype_);
+        }
+        onChanged();
+      }
+      if (other.getNumElements() != 0L) {
+        setNumElements(other.getNumElements());
+      }
+      if (other.getFinalized() != false) {
+        setFinalized(other.getFinalized());
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.util.SnapshotMetadataRecord parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.util.SnapshotMetadataRecord) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+    private int bitField0_;
+
+    private java.lang.Object graphHash_ = "";
+    /**
+     * <pre>
+     * Stores the fingerprint of the graph that describes the dataset that is
+     * snapshotted.
+     * </pre>
+     *
+     * <code>string graph_hash = 1;</code>
+     */
+    public java.lang.String getGraphHash() {
+      java.lang.Object ref = graphHash_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        graphHash_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Stores the fingerprint of the graph that describes the dataset that is
+     * snapshotted.
+     * </pre>
+     *
+     * <code>string graph_hash = 1;</code>
+     */
+    public com.google.protobuf.ByteString
+        getGraphHashBytes() {
+      java.lang.Object ref = graphHash_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        graphHash_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Stores the fingerprint of the graph that describes the dataset that is
+     * snapshotted.
+     * </pre>
+     *
+     * <code>string graph_hash = 1;</code>
+     */
+    public Builder setGraphHash(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      graphHash_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Stores the fingerprint of the graph that describes the dataset that is
+     * snapshotted.
+     * </pre>
+     *
+     * <code>string graph_hash = 1;</code>
+     */
+    public Builder clearGraphHash() {
+      
+      graphHash_ = getDefaultInstance().getGraphHash();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Stores the fingerprint of the graph that describes the dataset that is
+     * snapshotted.
+     * </pre>
+     *
+     * <code>string graph_hash = 1;</code>
+     */
+    public Builder setGraphHashBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      graphHash_ = value;
+      onChanged();
+      return this;
+    }
+
+    private java.lang.Object runId_ = "";
+    /**
+     * <pre>
+     * Run ID that this snapshot corresponds to.
+     * </pre>
+     *
+     * <code>string run_id = 2;</code>
+     */
+    public java.lang.String getRunId() {
+      java.lang.Object ref = runId_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        runId_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Run ID that this snapshot corresponds to.
+     * </pre>
+     *
+     * <code>string run_id = 2;</code>
+     */
+    public com.google.protobuf.ByteString
+        getRunIdBytes() {
+      java.lang.Object ref = runId_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        runId_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Run ID that this snapshot corresponds to.
+     * </pre>
+     *
+     * <code>string run_id = 2;</code>
+     */
+    public Builder setRunId(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      runId_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Run ID that this snapshot corresponds to.
+     * </pre>
+     *
+     * <code>string run_id = 2;</code>
+     */
+    public Builder clearRunId() {
+      
+      runId_ = getDefaultInstance().getRunId();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Run ID that this snapshot corresponds to.
+     * </pre>
+     *
+     * <code>string run_id = 2;</code>
+     */
+    public Builder setRunIdBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      runId_ = value;
+      onChanged();
+      return this;
+    }
+
+    private long creationTimestamp_ ;
+    /**
+     * <pre>
+     * Time when we started creating this snapshot.
+     * </pre>
+     *
+     * <code>int64 creation_timestamp = 3;</code>
+     */
+    public long getCreationTimestamp() {
+      return creationTimestamp_;
+    }
+    /**
+     * <pre>
+     * Time when we started creating this snapshot.
+     * </pre>
+     *
+     * <code>int64 creation_timestamp = 3;</code>
+     */
+    public Builder setCreationTimestamp(long value) {
+      
+      creationTimestamp_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Time when we started creating this snapshot.
+     * </pre>
+     *
+     * <code>int64 creation_timestamp = 3;</code>
+     */
+    public Builder clearCreationTimestamp() {
+      
+      creationTimestamp_ = 0L;
+      onChanged();
+      return this;
+    }
+
+    private long version_ ;
+    /**
+     * <pre>
+     * Version of the snapshot data file format.
+     * </pre>
+     *
+     * <code>int64 version = 4;</code>
+     */
+    public long getVersion() {
+      return version_;
+    }
+    /**
+     * <pre>
+     * Version of the snapshot data file format.
+     * </pre>
+     *
+     * <code>int64 version = 4;</code>
+     */
+    public Builder setVersion(long value) {
+      
+      version_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Version of the snapshot data file format.
+     * </pre>
+     *
+     * <code>int64 version = 4;</code>
+     */
+    public Builder clearVersion() {
+      
+      version_ = 0L;
+      onChanged();
+      return this;
+    }
+
+    private java.util.List<java.lang.Integer> dtype_ =
+      java.util.Collections.emptyList();
+    private void ensureDtypeIsMutable() {
+      if (!((bitField0_ & 0x00000001) != 0)) {
+        dtype_ = new java.util.ArrayList<java.lang.Integer>(dtype_);
+        bitField0_ |= 0x00000001;
+      }
+    }
+    /**
+     * <pre>
+     * A list of tensor dtype corresponding to each element of the snapshot.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.DataType dtype = 5;</code>
+     */
+    public java.util.List<org.tensorflow.proto.framework.DataType> getDtypeList() {
+      return new com.google.protobuf.Internal.ListAdapter<
+          java.lang.Integer, org.tensorflow.proto.framework.DataType>(dtype_, dtype_converter_);
+    }
+    /**
+     * <pre>
+     * A list of tensor dtype corresponding to each element of the snapshot.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.DataType dtype = 5;</code>
+     */
+    public int getDtypeCount() {
+      return dtype_.size();
+    }
+    /**
+     * <pre>
+     * A list of tensor dtype corresponding to each element of the snapshot.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.DataType dtype = 5;</code>
+     */
+    public org.tensorflow.proto.framework.DataType getDtype(int index) {
+      return dtype_converter_.convert(dtype_.get(index));
+    }
+    /**
+     * <pre>
+     * A list of tensor dtype corresponding to each element of the snapshot.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.DataType dtype = 5;</code>
+     */
+    public Builder setDtype(
+        int index, org.tensorflow.proto.framework.DataType value) {
+      if (value == null) {
+        throw new NullPointerException();
+      }
+      ensureDtypeIsMutable();
+      dtype_.set(index, value.getNumber());
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * A list of tensor dtype corresponding to each element of the snapshot.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.DataType dtype = 5;</code>
+     */
+    public Builder addDtype(org.tensorflow.proto.framework.DataType value) {
+      if (value == null) {
+        throw new NullPointerException();
+      }
+      ensureDtypeIsMutable();
+      dtype_.add(value.getNumber());
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * A list of tensor dtype corresponding to each element of the snapshot.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.DataType dtype = 5;</code>
+     */
+    public Builder addAllDtype(
+        java.lang.Iterable<? extends org.tensorflow.proto.framework.DataType> values) {
+      ensureDtypeIsMutable();
+      for (org.tensorflow.proto.framework.DataType value : values) {
+        dtype_.add(value.getNumber());
+      }
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * A list of tensor dtype corresponding to each element of the snapshot.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.DataType dtype = 5;</code>
+     */
+    public Builder clearDtype() {
+      dtype_ = java.util.Collections.emptyList();
+      bitField0_ = (bitField0_ & ~0x00000001);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * A list of tensor dtype corresponding to each element of the snapshot.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.DataType dtype = 5;</code>
+     */
+    public java.util.List<java.lang.Integer>
+    getDtypeValueList() {
+      return java.util.Collections.unmodifiableList(dtype_);
+    }
+    /**
+     * <pre>
+     * A list of tensor dtype corresponding to each element of the snapshot.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.DataType dtype = 5;</code>
+     */
+    public int getDtypeValue(int index) {
+      return dtype_.get(index);
+    }
+    /**
+     * <pre>
+     * A list of tensor dtype corresponding to each element of the snapshot.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.DataType dtype = 5;</code>
+     */
+    public Builder setDtypeValue(
+        int index, int value) {
+      ensureDtypeIsMutable();
+      dtype_.set(index, value);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * A list of tensor dtype corresponding to each element of the snapshot.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.DataType dtype = 5;</code>
+     */
+    public Builder addDtypeValue(int value) {
+      ensureDtypeIsMutable();
+      dtype_.add(value);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * A list of tensor dtype corresponding to each element of the snapshot.
+     * </pre>
+     *
+     * <code>repeated .tensorflow.DataType dtype = 5;</code>
+     */
+    public Builder addAllDtypeValue(
+        java.lang.Iterable<java.lang.Integer> values) {
+      ensureDtypeIsMutable();
+      for (int value : values) {
+        dtype_.add(value);
+      }
+      onChanged();
+      return this;
+    }
+
+    private long numElements_ ;
+    /**
+     * <pre>
+     * The number of elements in the snapshot.
+     * </pre>
+     *
+     * <code>int64 num_elements = 6;</code>
+     */
+    public long getNumElements() {
+      return numElements_;
+    }
+    /**
+     * <pre>
+     * The number of elements in the snapshot.
+     * </pre>
+     *
+     * <code>int64 num_elements = 6;</code>
+     */
+    public Builder setNumElements(long value) {
+      
+      numElements_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * The number of elements in the snapshot.
+     * </pre>
+     *
+     * <code>int64 num_elements = 6;</code>
+     */
+    public Builder clearNumElements() {
+      
+      numElements_ = 0L;
+      onChanged();
+      return this;
+    }
+
+    private boolean finalized_ ;
+    /**
+     * <code>bool finalized = 1000;</code>
+     */
+    public boolean getFinalized() {
+      return finalized_;
+    }
+    /**
+     * <code>bool finalized = 1000;</code>
+     */
+    public Builder setFinalized(boolean value) {
+      
+      finalized_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>bool finalized = 1000;</code>
+     */
+    public Builder clearFinalized() {
+      
+      finalized_ = false;
+      onChanged();
+      return this;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.data.experimental.SnapshotMetadataRecord)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.data.experimental.SnapshotMetadataRecord)
+  private static final org.tensorflow.proto.util.SnapshotMetadataRecord DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.util.SnapshotMetadataRecord();
+  }
+
+  public static org.tensorflow.proto.util.SnapshotMetadataRecord getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<SnapshotMetadataRecord>
+      PARSER = new com.google.protobuf.AbstractParser<SnapshotMetadataRecord>() {
+    @java.lang.Override
+    public SnapshotMetadataRecord parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new SnapshotMetadataRecord(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<SnapshotMetadataRecord> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<SnapshotMetadataRecord> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.util.SnapshotMetadataRecord getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SnapshotMetadataRecordOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SnapshotMetadataRecordOrBuilder.java
new file mode 100644
index 00000000000..13686e12e57
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SnapshotMetadataRecordOrBuilder.java
@@ -0,0 +1,121 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/data/experimental/snapshot.proto
+
+package org.tensorflow.proto.util;
+
+public interface SnapshotMetadataRecordOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.data.experimental.SnapshotMetadataRecord)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <pre>
+   * Stores the fingerprint of the graph that describes the dataset that is
+   * snapshotted.
+   * </pre>
+   *
+   * <code>string graph_hash = 1;</code>
+   */
+  java.lang.String getGraphHash();
+  /**
+   * <pre>
+   * Stores the fingerprint of the graph that describes the dataset that is
+   * snapshotted.
+   * </pre>
+   *
+   * <code>string graph_hash = 1;</code>
+   */
+  com.google.protobuf.ByteString
+      getGraphHashBytes();
+
+  /**
+   * <pre>
+   * Run ID that this snapshot corresponds to.
+   * </pre>
+   *
+   * <code>string run_id = 2;</code>
+   */
+  java.lang.String getRunId();
+  /**
+   * <pre>
+   * Run ID that this snapshot corresponds to.
+   * </pre>
+   *
+   * <code>string run_id = 2;</code>
+   */
+  com.google.protobuf.ByteString
+      getRunIdBytes();
+
+  /**
+   * <pre>
+   * Time when we started creating this snapshot.
+   * </pre>
+   *
+   * <code>int64 creation_timestamp = 3;</code>
+   */
+  long getCreationTimestamp();
+
+  /**
+   * <pre>
+   * Version of the snapshot data file format.
+   * </pre>
+   *
+   * <code>int64 version = 4;</code>
+   */
+  long getVersion();
+
+  /**
+   * <pre>
+   * A list of tensor dtype corresponding to each element of the snapshot.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.DataType dtype = 5;</code>
+   */
+  java.util.List<org.tensorflow.proto.framework.DataType> getDtypeList();
+  /**
+   * <pre>
+   * A list of tensor dtype corresponding to each element of the snapshot.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.DataType dtype = 5;</code>
+   */
+  int getDtypeCount();
+  /**
+   * <pre>
+   * A list of tensor dtype corresponding to each element of the snapshot.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.DataType dtype = 5;</code>
+   */
+  org.tensorflow.proto.framework.DataType getDtype(int index);
+  /**
+   * <pre>
+   * A list of tensor dtype corresponding to each element of the snapshot.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.DataType dtype = 5;</code>
+   */
+  java.util.List<java.lang.Integer>
+  getDtypeValueList();
+  /**
+   * <pre>
+   * A list of tensor dtype corresponding to each element of the snapshot.
+   * </pre>
+   *
+   * <code>repeated .tensorflow.DataType dtype = 5;</code>
+   */
+  int getDtypeValue(int index);
+
+  /**
+   * <pre>
+   * The number of elements in the snapshot.
+   * </pre>
+   *
+   * <code>int64 num_elements = 6;</code>
+   */
+  long getNumElements();
+
+  /**
+   * <code>bool finalized = 1000;</code>
+   */
+  boolean getFinalized();
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SnapshotProtos.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SnapshotProtos.java
new file mode 100644
index 00000000000..2d8f207c550
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SnapshotProtos.java
@@ -0,0 +1,102 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/data/experimental/snapshot.proto
+
+package org.tensorflow.proto.util;
+
+public final class SnapshotProtos {
+  private SnapshotProtos() {}
+  public static void registerAllExtensions(
+      com.google.protobuf.ExtensionRegistryLite registry) {
+  }
+
+  public static void registerAllExtensions(
+      com.google.protobuf.ExtensionRegistry registry) {
+    registerAllExtensions(
+        (com.google.protobuf.ExtensionRegistryLite) registry);
+  }
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_data_experimental_SnapshotRecord_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_data_experimental_SnapshotRecord_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_data_experimental_SnapshotMetadataRecord_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_data_experimental_SnapshotMetadataRecord_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_data_experimental_TensorMetadata_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_data_experimental_TensorMetadata_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_data_experimental_SnapshotTensorMetadata_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_data_experimental_SnapshotTensorMetadata_fieldAccessorTable;
+
+  public static com.google.protobuf.Descriptors.FileDescriptor
+      getDescriptor() {
+    return descriptor;
+  }
+  private static  com.google.protobuf.Descriptors.FileDescriptor
+      descriptor;
+  static {
+    java.lang.String[] descriptorData = {
+      "\n9tensorflow/core/protobuf/data/experime" +
+      "ntal/snapshot.proto\022\034tensorflow.data.exp" +
+      "erimental\032&tensorflow/core/framework/ten" +
+      "sor.proto\032,tensorflow/core/framework/ten" +
+      "sor_shape.proto\032%tensorflow/core/framewo" +
+      "rk/types.proto\"9\n\016SnapshotRecord\022\'\n\006tens" +
+      "or\030\001 \003(\0132\027.tensorflow.TensorProto\"\270\001\n\026Sn" +
+      "apshotMetadataRecord\022\022\n\ngraph_hash\030\001 \001(\t" +
+      "\022\016\n\006run_id\030\002 \001(\t\022\032\n\022creation_timestamp\030\003" +
+      " \001(\003\022\017\n\007version\030\004 \001(\003\022#\n\005dtype\030\005 \003(\0162\024.t" +
+      "ensorflow.DataType\022\024\n\014num_elements\030\006 \001(\003" +
+      "\022\022\n\tfinalized\030\350\007 \001(\010\"_\n\016TensorMetadata\0222" +
+      "\n\014tensor_shape\030\002 \001(\0132\034.tensorflow.Tensor" +
+      "ShapeProto\022\031\n\021tensor_size_bytes\030\003 \001(\003\"_\n" +
+      "\026SnapshotTensorMetadata\022E\n\017tensor_metada" +
+      "ta\030\001 \003(\0132,.tensorflow.data.experimental." +
+      "TensorMetadataB-\n\031org.tensorflow.proto.u" +
+      "tilB\016SnapshotProtosP\001b\006proto3"
+    };
+    descriptor = com.google.protobuf.Descriptors.FileDescriptor
+      .internalBuildGeneratedFileFrom(descriptorData,
+        new com.google.protobuf.Descriptors.FileDescriptor[] {
+          org.tensorflow.proto.framework.TensorProtos.getDescriptor(),
+          org.tensorflow.proto.framework.TensorShapeProtos.getDescriptor(),
+          org.tensorflow.proto.framework.TypesProtos.getDescriptor(),
+        });
+    internal_static_tensorflow_data_experimental_SnapshotRecord_descriptor =
+      getDescriptor().getMessageTypes().get(0);
+    internal_static_tensorflow_data_experimental_SnapshotRecord_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_data_experimental_SnapshotRecord_descriptor,
+        new java.lang.String[] { "Tensor", });
+    internal_static_tensorflow_data_experimental_SnapshotMetadataRecord_descriptor =
+      getDescriptor().getMessageTypes().get(1);
+    internal_static_tensorflow_data_experimental_SnapshotMetadataRecord_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_data_experimental_SnapshotMetadataRecord_descriptor,
+        new java.lang.String[] { "GraphHash", "RunId", "CreationTimestamp", "Version", "Dtype", "NumElements", "Finalized", });
+    internal_static_tensorflow_data_experimental_TensorMetadata_descriptor =
+      getDescriptor().getMessageTypes().get(2);
+    internal_static_tensorflow_data_experimental_TensorMetadata_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_data_experimental_TensorMetadata_descriptor,
+        new java.lang.String[] { "TensorShape", "TensorSizeBytes", });
+    internal_static_tensorflow_data_experimental_SnapshotTensorMetadata_descriptor =
+      getDescriptor().getMessageTypes().get(3);
+    internal_static_tensorflow_data_experimental_SnapshotTensorMetadata_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_data_experimental_SnapshotTensorMetadata_descriptor,
+        new java.lang.String[] { "TensorMetadata", });
+    org.tensorflow.proto.framework.TensorProtos.getDescriptor();
+    org.tensorflow.proto.framework.TensorShapeProtos.getDescriptor();
+    org.tensorflow.proto.framework.TypesProtos.getDescriptor();
+  }
+
+  // @@protoc_insertion_point(outer_class_scope)
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SnapshotRecord.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SnapshotRecord.java
new file mode 100644
index 00000000000..223b22587d7
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SnapshotRecord.java
@@ -0,0 +1,777 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/data/experimental/snapshot.proto
+
+package org.tensorflow.proto.util;
+
+/**
+ * <pre>
+ * Each SnapshotRecord represents one batch of pre-processed input data. A batch
+ * consists of a list of tensors that we encode as TensorProtos. This message
+ * doesn't store the structure of the batch.
+ * </pre>
+ *
+ * Protobuf type {@code tensorflow.data.experimental.SnapshotRecord}
+ */
+public  final class SnapshotRecord extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.data.experimental.SnapshotRecord)
+    SnapshotRecordOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use SnapshotRecord.newBuilder() to construct.
+  private SnapshotRecord(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private SnapshotRecord() {
+    tensor_ = java.util.Collections.emptyList();
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new SnapshotRecord();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private SnapshotRecord(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    int mutable_bitField0_ = 0;
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 10: {
+            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
+              tensor_ = new java.util.ArrayList<org.tensorflow.proto.framework.TensorProto>();
+              mutable_bitField0_ |= 0x00000001;
+            }
+            tensor_.add(
+                input.readMessage(org.tensorflow.proto.framework.TensorProto.parser(), extensionRegistry));
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      if (((mutable_bitField0_ & 0x00000001) != 0)) {
+        tensor_ = java.util.Collections.unmodifiableList(tensor_);
+      }
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.util.SnapshotProtos.internal_static_tensorflow_data_experimental_SnapshotRecord_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.util.SnapshotProtos.internal_static_tensorflow_data_experimental_SnapshotRecord_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.util.SnapshotRecord.class, org.tensorflow.proto.util.SnapshotRecord.Builder.class);
+  }
+
+  public static final int TENSOR_FIELD_NUMBER = 1;
+  private java.util.List<org.tensorflow.proto.framework.TensorProto> tensor_;
+  /**
+   * <code>repeated .tensorflow.TensorProto tensor = 1;</code>
+   */
+  public java.util.List<org.tensorflow.proto.framework.TensorProto> getTensorList() {
+    return tensor_;
+  }
+  /**
+   * <code>repeated .tensorflow.TensorProto tensor = 1;</code>
+   */
+  public java.util.List<? extends org.tensorflow.proto.framework.TensorProtoOrBuilder> 
+      getTensorOrBuilderList() {
+    return tensor_;
+  }
+  /**
+   * <code>repeated .tensorflow.TensorProto tensor = 1;</code>
+   */
+  public int getTensorCount() {
+    return tensor_.size();
+  }
+  /**
+   * <code>repeated .tensorflow.TensorProto tensor = 1;</code>
+   */
+  public org.tensorflow.proto.framework.TensorProto getTensor(int index) {
+    return tensor_.get(index);
+  }
+  /**
+   * <code>repeated .tensorflow.TensorProto tensor = 1;</code>
+   */
+  public org.tensorflow.proto.framework.TensorProtoOrBuilder getTensorOrBuilder(
+      int index) {
+    return tensor_.get(index);
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    for (int i = 0; i < tensor_.size(); i++) {
+      output.writeMessage(1, tensor_.get(i));
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    for (int i = 0; i < tensor_.size(); i++) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(1, tensor_.get(i));
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.util.SnapshotRecord)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.util.SnapshotRecord other = (org.tensorflow.proto.util.SnapshotRecord) obj;
+
+    if (!getTensorList()
+        .equals(other.getTensorList())) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    if (getTensorCount() > 0) {
+      hash = (37 * hash) + TENSOR_FIELD_NUMBER;
+      hash = (53 * hash) + getTensorList().hashCode();
+    }
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.util.SnapshotRecord parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.SnapshotRecord parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.SnapshotRecord parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.SnapshotRecord parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.SnapshotRecord parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.SnapshotRecord parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.SnapshotRecord parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.SnapshotRecord parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.SnapshotRecord parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.SnapshotRecord parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.SnapshotRecord parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.SnapshotRecord parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.util.SnapshotRecord prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * <pre>
+   * Each SnapshotRecord represents one batch of pre-processed input data. A batch
+   * consists of a list of tensors that we encode as TensorProtos. This message
+   * doesn't store the structure of the batch.
+   * </pre>
+   *
+   * Protobuf type {@code tensorflow.data.experimental.SnapshotRecord}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.data.experimental.SnapshotRecord)
+      org.tensorflow.proto.util.SnapshotRecordOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.util.SnapshotProtos.internal_static_tensorflow_data_experimental_SnapshotRecord_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.util.SnapshotProtos.internal_static_tensorflow_data_experimental_SnapshotRecord_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.util.SnapshotRecord.class, org.tensorflow.proto.util.SnapshotRecord.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.util.SnapshotRecord.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+        getTensorFieldBuilder();
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      if (tensorBuilder_ == null) {
+        tensor_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000001);
+      } else {
+        tensorBuilder_.clear();
+      }
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.util.SnapshotProtos.internal_static_tensorflow_data_experimental_SnapshotRecord_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.SnapshotRecord getDefaultInstanceForType() {
+      return org.tensorflow.proto.util.SnapshotRecord.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.SnapshotRecord build() {
+      org.tensorflow.proto.util.SnapshotRecord result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.SnapshotRecord buildPartial() {
+      org.tensorflow.proto.util.SnapshotRecord result = new org.tensorflow.proto.util.SnapshotRecord(this);
+      int from_bitField0_ = bitField0_;
+      if (tensorBuilder_ == null) {
+        if (((bitField0_ & 0x00000001) != 0)) {
+          tensor_ = java.util.Collections.unmodifiableList(tensor_);
+          bitField0_ = (bitField0_ & ~0x00000001);
+        }
+        result.tensor_ = tensor_;
+      } else {
+        result.tensor_ = tensorBuilder_.build();
+      }
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.util.SnapshotRecord) {
+        return mergeFrom((org.tensorflow.proto.util.SnapshotRecord)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.util.SnapshotRecord other) {
+      if (other == org.tensorflow.proto.util.SnapshotRecord.getDefaultInstance()) return this;
+      if (tensorBuilder_ == null) {
+        if (!other.tensor_.isEmpty()) {
+          if (tensor_.isEmpty()) {
+            tensor_ = other.tensor_;
+            bitField0_ = (bitField0_ & ~0x00000001);
+          } else {
+            ensureTensorIsMutable();
+            tensor_.addAll(other.tensor_);
+          }
+          onChanged();
+        }
+      } else {
+        if (!other.tensor_.isEmpty()) {
+          if (tensorBuilder_.isEmpty()) {
+            tensorBuilder_.dispose();
+            tensorBuilder_ = null;
+            tensor_ = other.tensor_;
+            bitField0_ = (bitField0_ & ~0x00000001);
+            tensorBuilder_ = 
+              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
+                 getTensorFieldBuilder() : null;
+          } else {
+            tensorBuilder_.addAllMessages(other.tensor_);
+          }
+        }
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.util.SnapshotRecord parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.util.SnapshotRecord) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+    private int bitField0_;
+
+    private java.util.List<org.tensorflow.proto.framework.TensorProto> tensor_ =
+      java.util.Collections.emptyList();
+    private void ensureTensorIsMutable() {
+      if (!((bitField0_ & 0x00000001) != 0)) {
+        tensor_ = new java.util.ArrayList<org.tensorflow.proto.framework.TensorProto>(tensor_);
+        bitField0_ |= 0x00000001;
+       }
+    }
+
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.framework.TensorProto, org.tensorflow.proto.framework.TensorProto.Builder, org.tensorflow.proto.framework.TensorProtoOrBuilder> tensorBuilder_;
+
+    /**
+     * <code>repeated .tensorflow.TensorProto tensor = 1;</code>
+     */
+    public java.util.List<org.tensorflow.proto.framework.TensorProto> getTensorList() {
+      if (tensorBuilder_ == null) {
+        return java.util.Collections.unmodifiableList(tensor_);
+      } else {
+        return tensorBuilder_.getMessageList();
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.TensorProto tensor = 1;</code>
+     */
+    public int getTensorCount() {
+      if (tensorBuilder_ == null) {
+        return tensor_.size();
+      } else {
+        return tensorBuilder_.getCount();
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.TensorProto tensor = 1;</code>
+     */
+    public org.tensorflow.proto.framework.TensorProto getTensor(int index) {
+      if (tensorBuilder_ == null) {
+        return tensor_.get(index);
+      } else {
+        return tensorBuilder_.getMessage(index);
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.TensorProto tensor = 1;</code>
+     */
+    public Builder setTensor(
+        int index, org.tensorflow.proto.framework.TensorProto value) {
+      if (tensorBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureTensorIsMutable();
+        tensor_.set(index, value);
+        onChanged();
+      } else {
+        tensorBuilder_.setMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.TensorProto tensor = 1;</code>
+     */
+    public Builder setTensor(
+        int index, org.tensorflow.proto.framework.TensorProto.Builder builderForValue) {
+      if (tensorBuilder_ == null) {
+        ensureTensorIsMutable();
+        tensor_.set(index, builderForValue.build());
+        onChanged();
+      } else {
+        tensorBuilder_.setMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.TensorProto tensor = 1;</code>
+     */
+    public Builder addTensor(org.tensorflow.proto.framework.TensorProto value) {
+      if (tensorBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureTensorIsMutable();
+        tensor_.add(value);
+        onChanged();
+      } else {
+        tensorBuilder_.addMessage(value);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.TensorProto tensor = 1;</code>
+     */
+    public Builder addTensor(
+        int index, org.tensorflow.proto.framework.TensorProto value) {
+      if (tensorBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureTensorIsMutable();
+        tensor_.add(index, value);
+        onChanged();
+      } else {
+        tensorBuilder_.addMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.TensorProto tensor = 1;</code>
+     */
+    public Builder addTensor(
+        org.tensorflow.proto.framework.TensorProto.Builder builderForValue) {
+      if (tensorBuilder_ == null) {
+        ensureTensorIsMutable();
+        tensor_.add(builderForValue.build());
+        onChanged();
+      } else {
+        tensorBuilder_.addMessage(builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.TensorProto tensor = 1;</code>
+     */
+    public Builder addTensor(
+        int index, org.tensorflow.proto.framework.TensorProto.Builder builderForValue) {
+      if (tensorBuilder_ == null) {
+        ensureTensorIsMutable();
+        tensor_.add(index, builderForValue.build());
+        onChanged();
+      } else {
+        tensorBuilder_.addMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.TensorProto tensor = 1;</code>
+     */
+    public Builder addAllTensor(
+        java.lang.Iterable<? extends org.tensorflow.proto.framework.TensorProto> values) {
+      if (tensorBuilder_ == null) {
+        ensureTensorIsMutable();
+        com.google.protobuf.AbstractMessageLite.Builder.addAll(
+            values, tensor_);
+        onChanged();
+      } else {
+        tensorBuilder_.addAllMessages(values);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.TensorProto tensor = 1;</code>
+     */
+    public Builder clearTensor() {
+      if (tensorBuilder_ == null) {
+        tensor_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000001);
+        onChanged();
+      } else {
+        tensorBuilder_.clear();
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.TensorProto tensor = 1;</code>
+     */
+    public Builder removeTensor(int index) {
+      if (tensorBuilder_ == null) {
+        ensureTensorIsMutable();
+        tensor_.remove(index);
+        onChanged();
+      } else {
+        tensorBuilder_.remove(index);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.TensorProto tensor = 1;</code>
+     */
+    public org.tensorflow.proto.framework.TensorProto.Builder getTensorBuilder(
+        int index) {
+      return getTensorFieldBuilder().getBuilder(index);
+    }
+    /**
+     * <code>repeated .tensorflow.TensorProto tensor = 1;</code>
+     */
+    public org.tensorflow.proto.framework.TensorProtoOrBuilder getTensorOrBuilder(
+        int index) {
+      if (tensorBuilder_ == null) {
+        return tensor_.get(index);  } else {
+        return tensorBuilder_.getMessageOrBuilder(index);
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.TensorProto tensor = 1;</code>
+     */
+    public java.util.List<? extends org.tensorflow.proto.framework.TensorProtoOrBuilder> 
+         getTensorOrBuilderList() {
+      if (tensorBuilder_ != null) {
+        return tensorBuilder_.getMessageOrBuilderList();
+      } else {
+        return java.util.Collections.unmodifiableList(tensor_);
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.TensorProto tensor = 1;</code>
+     */
+    public org.tensorflow.proto.framework.TensorProto.Builder addTensorBuilder() {
+      return getTensorFieldBuilder().addBuilder(
+          org.tensorflow.proto.framework.TensorProto.getDefaultInstance());
+    }
+    /**
+     * <code>repeated .tensorflow.TensorProto tensor = 1;</code>
+     */
+    public org.tensorflow.proto.framework.TensorProto.Builder addTensorBuilder(
+        int index) {
+      return getTensorFieldBuilder().addBuilder(
+          index, org.tensorflow.proto.framework.TensorProto.getDefaultInstance());
+    }
+    /**
+     * <code>repeated .tensorflow.TensorProto tensor = 1;</code>
+     */
+    public java.util.List<org.tensorflow.proto.framework.TensorProto.Builder> 
+         getTensorBuilderList() {
+      return getTensorFieldBuilder().getBuilderList();
+    }
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.framework.TensorProto, org.tensorflow.proto.framework.TensorProto.Builder, org.tensorflow.proto.framework.TensorProtoOrBuilder> 
+        getTensorFieldBuilder() {
+      if (tensorBuilder_ == null) {
+        tensorBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
+            org.tensorflow.proto.framework.TensorProto, org.tensorflow.proto.framework.TensorProto.Builder, org.tensorflow.proto.framework.TensorProtoOrBuilder>(
+                tensor_,
+                ((bitField0_ & 0x00000001) != 0),
+                getParentForChildren(),
+                isClean());
+        tensor_ = null;
+      }
+      return tensorBuilder_;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.data.experimental.SnapshotRecord)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.data.experimental.SnapshotRecord)
+  private static final org.tensorflow.proto.util.SnapshotRecord DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.util.SnapshotRecord();
+  }
+
+  public static org.tensorflow.proto.util.SnapshotRecord getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<SnapshotRecord>
+      PARSER = new com.google.protobuf.AbstractParser<SnapshotRecord>() {
+    @java.lang.Override
+    public SnapshotRecord parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new SnapshotRecord(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<SnapshotRecord> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<SnapshotRecord> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.util.SnapshotRecord getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SnapshotRecordOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SnapshotRecordOrBuilder.java
new file mode 100644
index 00000000000..e420dac0f6b
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SnapshotRecordOrBuilder.java
@@ -0,0 +1,33 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/data/experimental/snapshot.proto
+
+package org.tensorflow.proto.util;
+
+public interface SnapshotRecordOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.data.experimental.SnapshotRecord)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <code>repeated .tensorflow.TensorProto tensor = 1;</code>
+   */
+  java.util.List<org.tensorflow.proto.framework.TensorProto> 
+      getTensorList();
+  /**
+   * <code>repeated .tensorflow.TensorProto tensor = 1;</code>
+   */
+  org.tensorflow.proto.framework.TensorProto getTensor(int index);
+  /**
+   * <code>repeated .tensorflow.TensorProto tensor = 1;</code>
+   */
+  int getTensorCount();
+  /**
+   * <code>repeated .tensorflow.TensorProto tensor = 1;</code>
+   */
+  java.util.List<? extends org.tensorflow.proto.framework.TensorProtoOrBuilder> 
+      getTensorOrBuilderList();
+  /**
+   * <code>repeated .tensorflow.TensorProto tensor = 1;</code>
+   */
+  org.tensorflow.proto.framework.TensorProtoOrBuilder getTensorOrBuilder(
+      int index);
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SnapshotTensorMetadata.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SnapshotTensorMetadata.java
new file mode 100644
index 00000000000..2d8d9dc96de
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SnapshotTensorMetadata.java
@@ -0,0 +1,773 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/data/experimental/snapshot.proto
+
+package org.tensorflow.proto.util;
+
+/**
+ * <pre>
+ * Metadata for all the tensors in a Snapshot Record.
+ * </pre>
+ *
+ * Protobuf type {@code tensorflow.data.experimental.SnapshotTensorMetadata}
+ */
+public  final class SnapshotTensorMetadata extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.data.experimental.SnapshotTensorMetadata)
+    SnapshotTensorMetadataOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use SnapshotTensorMetadata.newBuilder() to construct.
+  private SnapshotTensorMetadata(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private SnapshotTensorMetadata() {
+    tensorMetadata_ = java.util.Collections.emptyList();
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new SnapshotTensorMetadata();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private SnapshotTensorMetadata(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    int mutable_bitField0_ = 0;
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 10: {
+            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
+              tensorMetadata_ = new java.util.ArrayList<org.tensorflow.proto.util.TensorMetadata>();
+              mutable_bitField0_ |= 0x00000001;
+            }
+            tensorMetadata_.add(
+                input.readMessage(org.tensorflow.proto.util.TensorMetadata.parser(), extensionRegistry));
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      if (((mutable_bitField0_ & 0x00000001) != 0)) {
+        tensorMetadata_ = java.util.Collections.unmodifiableList(tensorMetadata_);
+      }
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.util.SnapshotProtos.internal_static_tensorflow_data_experimental_SnapshotTensorMetadata_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.util.SnapshotProtos.internal_static_tensorflow_data_experimental_SnapshotTensorMetadata_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.util.SnapshotTensorMetadata.class, org.tensorflow.proto.util.SnapshotTensorMetadata.Builder.class);
+  }
+
+  public static final int TENSOR_METADATA_FIELD_NUMBER = 1;
+  private java.util.List<org.tensorflow.proto.util.TensorMetadata> tensorMetadata_;
+  /**
+   * <code>repeated .tensorflow.data.experimental.TensorMetadata tensor_metadata = 1;</code>
+   */
+  public java.util.List<org.tensorflow.proto.util.TensorMetadata> getTensorMetadataList() {
+    return tensorMetadata_;
+  }
+  /**
+   * <code>repeated .tensorflow.data.experimental.TensorMetadata tensor_metadata = 1;</code>
+   */
+  public java.util.List<? extends org.tensorflow.proto.util.TensorMetadataOrBuilder> 
+      getTensorMetadataOrBuilderList() {
+    return tensorMetadata_;
+  }
+  /**
+   * <code>repeated .tensorflow.data.experimental.TensorMetadata tensor_metadata = 1;</code>
+   */
+  public int getTensorMetadataCount() {
+    return tensorMetadata_.size();
+  }
+  /**
+   * <code>repeated .tensorflow.data.experimental.TensorMetadata tensor_metadata = 1;</code>
+   */
+  public org.tensorflow.proto.util.TensorMetadata getTensorMetadata(int index) {
+    return tensorMetadata_.get(index);
+  }
+  /**
+   * <code>repeated .tensorflow.data.experimental.TensorMetadata tensor_metadata = 1;</code>
+   */
+  public org.tensorflow.proto.util.TensorMetadataOrBuilder getTensorMetadataOrBuilder(
+      int index) {
+    return tensorMetadata_.get(index);
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    for (int i = 0; i < tensorMetadata_.size(); i++) {
+      output.writeMessage(1, tensorMetadata_.get(i));
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    for (int i = 0; i < tensorMetadata_.size(); i++) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(1, tensorMetadata_.get(i));
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.util.SnapshotTensorMetadata)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.util.SnapshotTensorMetadata other = (org.tensorflow.proto.util.SnapshotTensorMetadata) obj;
+
+    if (!getTensorMetadataList()
+        .equals(other.getTensorMetadataList())) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    if (getTensorMetadataCount() > 0) {
+      hash = (37 * hash) + TENSOR_METADATA_FIELD_NUMBER;
+      hash = (53 * hash) + getTensorMetadataList().hashCode();
+    }
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.util.SnapshotTensorMetadata parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.SnapshotTensorMetadata parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.SnapshotTensorMetadata parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.SnapshotTensorMetadata parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.SnapshotTensorMetadata parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.SnapshotTensorMetadata parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.SnapshotTensorMetadata parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.SnapshotTensorMetadata parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.SnapshotTensorMetadata parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.SnapshotTensorMetadata parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.SnapshotTensorMetadata parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.SnapshotTensorMetadata parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.util.SnapshotTensorMetadata prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * <pre>
+   * Metadata for all the tensors in a Snapshot Record.
+   * </pre>
+   *
+   * Protobuf type {@code tensorflow.data.experimental.SnapshotTensorMetadata}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.data.experimental.SnapshotTensorMetadata)
+      org.tensorflow.proto.util.SnapshotTensorMetadataOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.util.SnapshotProtos.internal_static_tensorflow_data_experimental_SnapshotTensorMetadata_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.util.SnapshotProtos.internal_static_tensorflow_data_experimental_SnapshotTensorMetadata_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.util.SnapshotTensorMetadata.class, org.tensorflow.proto.util.SnapshotTensorMetadata.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.util.SnapshotTensorMetadata.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+        getTensorMetadataFieldBuilder();
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      if (tensorMetadataBuilder_ == null) {
+        tensorMetadata_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000001);
+      } else {
+        tensorMetadataBuilder_.clear();
+      }
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.util.SnapshotProtos.internal_static_tensorflow_data_experimental_SnapshotTensorMetadata_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.SnapshotTensorMetadata getDefaultInstanceForType() {
+      return org.tensorflow.proto.util.SnapshotTensorMetadata.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.SnapshotTensorMetadata build() {
+      org.tensorflow.proto.util.SnapshotTensorMetadata result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.SnapshotTensorMetadata buildPartial() {
+      org.tensorflow.proto.util.SnapshotTensorMetadata result = new org.tensorflow.proto.util.SnapshotTensorMetadata(this);
+      int from_bitField0_ = bitField0_;
+      if (tensorMetadataBuilder_ == null) {
+        if (((bitField0_ & 0x00000001) != 0)) {
+          tensorMetadata_ = java.util.Collections.unmodifiableList(tensorMetadata_);
+          bitField0_ = (bitField0_ & ~0x00000001);
+        }
+        result.tensorMetadata_ = tensorMetadata_;
+      } else {
+        result.tensorMetadata_ = tensorMetadataBuilder_.build();
+      }
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.util.SnapshotTensorMetadata) {
+        return mergeFrom((org.tensorflow.proto.util.SnapshotTensorMetadata)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.util.SnapshotTensorMetadata other) {
+      if (other == org.tensorflow.proto.util.SnapshotTensorMetadata.getDefaultInstance()) return this;
+      if (tensorMetadataBuilder_ == null) {
+        if (!other.tensorMetadata_.isEmpty()) {
+          if (tensorMetadata_.isEmpty()) {
+            tensorMetadata_ = other.tensorMetadata_;
+            bitField0_ = (bitField0_ & ~0x00000001);
+          } else {
+            ensureTensorMetadataIsMutable();
+            tensorMetadata_.addAll(other.tensorMetadata_);
+          }
+          onChanged();
+        }
+      } else {
+        if (!other.tensorMetadata_.isEmpty()) {
+          if (tensorMetadataBuilder_.isEmpty()) {
+            tensorMetadataBuilder_.dispose();
+            tensorMetadataBuilder_ = null;
+            tensorMetadata_ = other.tensorMetadata_;
+            bitField0_ = (bitField0_ & ~0x00000001);
+            tensorMetadataBuilder_ = 
+              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
+                 getTensorMetadataFieldBuilder() : null;
+          } else {
+            tensorMetadataBuilder_.addAllMessages(other.tensorMetadata_);
+          }
+        }
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.util.SnapshotTensorMetadata parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.util.SnapshotTensorMetadata) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+    private int bitField0_;
+
+    private java.util.List<org.tensorflow.proto.util.TensorMetadata> tensorMetadata_ =
+      java.util.Collections.emptyList();
+    private void ensureTensorMetadataIsMutable() {
+      if (!((bitField0_ & 0x00000001) != 0)) {
+        tensorMetadata_ = new java.util.ArrayList<org.tensorflow.proto.util.TensorMetadata>(tensorMetadata_);
+        bitField0_ |= 0x00000001;
+       }
+    }
+
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.util.TensorMetadata, org.tensorflow.proto.util.TensorMetadata.Builder, org.tensorflow.proto.util.TensorMetadataOrBuilder> tensorMetadataBuilder_;
+
+    /**
+     * <code>repeated .tensorflow.data.experimental.TensorMetadata tensor_metadata = 1;</code>
+     */
+    public java.util.List<org.tensorflow.proto.util.TensorMetadata> getTensorMetadataList() {
+      if (tensorMetadataBuilder_ == null) {
+        return java.util.Collections.unmodifiableList(tensorMetadata_);
+      } else {
+        return tensorMetadataBuilder_.getMessageList();
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.data.experimental.TensorMetadata tensor_metadata = 1;</code>
+     */
+    public int getTensorMetadataCount() {
+      if (tensorMetadataBuilder_ == null) {
+        return tensorMetadata_.size();
+      } else {
+        return tensorMetadataBuilder_.getCount();
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.data.experimental.TensorMetadata tensor_metadata = 1;</code>
+     */
+    public org.tensorflow.proto.util.TensorMetadata getTensorMetadata(int index) {
+      if (tensorMetadataBuilder_ == null) {
+        return tensorMetadata_.get(index);
+      } else {
+        return tensorMetadataBuilder_.getMessage(index);
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.data.experimental.TensorMetadata tensor_metadata = 1;</code>
+     */
+    public Builder setTensorMetadata(
+        int index, org.tensorflow.proto.util.TensorMetadata value) {
+      if (tensorMetadataBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureTensorMetadataIsMutable();
+        tensorMetadata_.set(index, value);
+        onChanged();
+      } else {
+        tensorMetadataBuilder_.setMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.data.experimental.TensorMetadata tensor_metadata = 1;</code>
+     */
+    public Builder setTensorMetadata(
+        int index, org.tensorflow.proto.util.TensorMetadata.Builder builderForValue) {
+      if (tensorMetadataBuilder_ == null) {
+        ensureTensorMetadataIsMutable();
+        tensorMetadata_.set(index, builderForValue.build());
+        onChanged();
+      } else {
+        tensorMetadataBuilder_.setMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.data.experimental.TensorMetadata tensor_metadata = 1;</code>
+     */
+    public Builder addTensorMetadata(org.tensorflow.proto.util.TensorMetadata value) {
+      if (tensorMetadataBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureTensorMetadataIsMutable();
+        tensorMetadata_.add(value);
+        onChanged();
+      } else {
+        tensorMetadataBuilder_.addMessage(value);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.data.experimental.TensorMetadata tensor_metadata = 1;</code>
+     */
+    public Builder addTensorMetadata(
+        int index, org.tensorflow.proto.util.TensorMetadata value) {
+      if (tensorMetadataBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureTensorMetadataIsMutable();
+        tensorMetadata_.add(index, value);
+        onChanged();
+      } else {
+        tensorMetadataBuilder_.addMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.data.experimental.TensorMetadata tensor_metadata = 1;</code>
+     */
+    public Builder addTensorMetadata(
+        org.tensorflow.proto.util.TensorMetadata.Builder builderForValue) {
+      if (tensorMetadataBuilder_ == null) {
+        ensureTensorMetadataIsMutable();
+        tensorMetadata_.add(builderForValue.build());
+        onChanged();
+      } else {
+        tensorMetadataBuilder_.addMessage(builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.data.experimental.TensorMetadata tensor_metadata = 1;</code>
+     */
+    public Builder addTensorMetadata(
+        int index, org.tensorflow.proto.util.TensorMetadata.Builder builderForValue) {
+      if (tensorMetadataBuilder_ == null) {
+        ensureTensorMetadataIsMutable();
+        tensorMetadata_.add(index, builderForValue.build());
+        onChanged();
+      } else {
+        tensorMetadataBuilder_.addMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.data.experimental.TensorMetadata tensor_metadata = 1;</code>
+     */
+    public Builder addAllTensorMetadata(
+        java.lang.Iterable<? extends org.tensorflow.proto.util.TensorMetadata> values) {
+      if (tensorMetadataBuilder_ == null) {
+        ensureTensorMetadataIsMutable();
+        com.google.protobuf.AbstractMessageLite.Builder.addAll(
+            values, tensorMetadata_);
+        onChanged();
+      } else {
+        tensorMetadataBuilder_.addAllMessages(values);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.data.experimental.TensorMetadata tensor_metadata = 1;</code>
+     */
+    public Builder clearTensorMetadata() {
+      if (tensorMetadataBuilder_ == null) {
+        tensorMetadata_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000001);
+        onChanged();
+      } else {
+        tensorMetadataBuilder_.clear();
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.data.experimental.TensorMetadata tensor_metadata = 1;</code>
+     */
+    public Builder removeTensorMetadata(int index) {
+      if (tensorMetadataBuilder_ == null) {
+        ensureTensorMetadataIsMutable();
+        tensorMetadata_.remove(index);
+        onChanged();
+      } else {
+        tensorMetadataBuilder_.remove(index);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.data.experimental.TensorMetadata tensor_metadata = 1;</code>
+     */
+    public org.tensorflow.proto.util.TensorMetadata.Builder getTensorMetadataBuilder(
+        int index) {
+      return getTensorMetadataFieldBuilder().getBuilder(index);
+    }
+    /**
+     * <code>repeated .tensorflow.data.experimental.TensorMetadata tensor_metadata = 1;</code>
+     */
+    public org.tensorflow.proto.util.TensorMetadataOrBuilder getTensorMetadataOrBuilder(
+        int index) {
+      if (tensorMetadataBuilder_ == null) {
+        return tensorMetadata_.get(index);  } else {
+        return tensorMetadataBuilder_.getMessageOrBuilder(index);
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.data.experimental.TensorMetadata tensor_metadata = 1;</code>
+     */
+    public java.util.List<? extends org.tensorflow.proto.util.TensorMetadataOrBuilder> 
+         getTensorMetadataOrBuilderList() {
+      if (tensorMetadataBuilder_ != null) {
+        return tensorMetadataBuilder_.getMessageOrBuilderList();
+      } else {
+        return java.util.Collections.unmodifiableList(tensorMetadata_);
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.data.experimental.TensorMetadata tensor_metadata = 1;</code>
+     */
+    public org.tensorflow.proto.util.TensorMetadata.Builder addTensorMetadataBuilder() {
+      return getTensorMetadataFieldBuilder().addBuilder(
+          org.tensorflow.proto.util.TensorMetadata.getDefaultInstance());
+    }
+    /**
+     * <code>repeated .tensorflow.data.experimental.TensorMetadata tensor_metadata = 1;</code>
+     */
+    public org.tensorflow.proto.util.TensorMetadata.Builder addTensorMetadataBuilder(
+        int index) {
+      return getTensorMetadataFieldBuilder().addBuilder(
+          index, org.tensorflow.proto.util.TensorMetadata.getDefaultInstance());
+    }
+    /**
+     * <code>repeated .tensorflow.data.experimental.TensorMetadata tensor_metadata = 1;</code>
+     */
+    public java.util.List<org.tensorflow.proto.util.TensorMetadata.Builder> 
+         getTensorMetadataBuilderList() {
+      return getTensorMetadataFieldBuilder().getBuilderList();
+    }
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.util.TensorMetadata, org.tensorflow.proto.util.TensorMetadata.Builder, org.tensorflow.proto.util.TensorMetadataOrBuilder> 
+        getTensorMetadataFieldBuilder() {
+      if (tensorMetadataBuilder_ == null) {
+        tensorMetadataBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
+            org.tensorflow.proto.util.TensorMetadata, org.tensorflow.proto.util.TensorMetadata.Builder, org.tensorflow.proto.util.TensorMetadataOrBuilder>(
+                tensorMetadata_,
+                ((bitField0_ & 0x00000001) != 0),
+                getParentForChildren(),
+                isClean());
+        tensorMetadata_ = null;
+      }
+      return tensorMetadataBuilder_;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.data.experimental.SnapshotTensorMetadata)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.data.experimental.SnapshotTensorMetadata)
+  private static final org.tensorflow.proto.util.SnapshotTensorMetadata DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.util.SnapshotTensorMetadata();
+  }
+
+  public static org.tensorflow.proto.util.SnapshotTensorMetadata getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<SnapshotTensorMetadata>
+      PARSER = new com.google.protobuf.AbstractParser<SnapshotTensorMetadata>() {
+    @java.lang.Override
+    public SnapshotTensorMetadata parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new SnapshotTensorMetadata(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<SnapshotTensorMetadata> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<SnapshotTensorMetadata> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.util.SnapshotTensorMetadata getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SnapshotTensorMetadataOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SnapshotTensorMetadataOrBuilder.java
new file mode 100644
index 00000000000..8c906ffb714
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SnapshotTensorMetadataOrBuilder.java
@@ -0,0 +1,33 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/data/experimental/snapshot.proto
+
+package org.tensorflow.proto.util;
+
+public interface SnapshotTensorMetadataOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.data.experimental.SnapshotTensorMetadata)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <code>repeated .tensorflow.data.experimental.TensorMetadata tensor_metadata = 1;</code>
+   */
+  java.util.List<org.tensorflow.proto.util.TensorMetadata> 
+      getTensorMetadataList();
+  /**
+   * <code>repeated .tensorflow.data.experimental.TensorMetadata tensor_metadata = 1;</code>
+   */
+  org.tensorflow.proto.util.TensorMetadata getTensorMetadata(int index);
+  /**
+   * <code>repeated .tensorflow.data.experimental.TensorMetadata tensor_metadata = 1;</code>
+   */
+  int getTensorMetadataCount();
+  /**
+   * <code>repeated .tensorflow.data.experimental.TensorMetadata tensor_metadata = 1;</code>
+   */
+  java.util.List<? extends org.tensorflow.proto.util.TensorMetadataOrBuilder> 
+      getTensorMetadataOrBuilderList();
+  /**
+   * <code>repeated .tensorflow.data.experimental.TensorMetadata tensor_metadata = 1;</code>
+   */
+  org.tensorflow.proto.util.TensorMetadataOrBuilder getTensorMetadataOrBuilder(
+      int index);
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SourceFile.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SourceFile.java
new file mode 100644
index 00000000000..b8f6c068649
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SourceFile.java
@@ -0,0 +1,964 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/debug_event.proto
+
+package org.tensorflow.proto.util;
+
+/**
+ * <pre>
+ * Content of a source file involved in the execution of the debugged TensorFlow
+ * program.
+ * </pre>
+ *
+ * Protobuf type {@code tensorflow.SourceFile}
+ */
+public  final class SourceFile extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.SourceFile)
+    SourceFileOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use SourceFile.newBuilder() to construct.
+  private SourceFile(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private SourceFile() {
+    filePath_ = "";
+    hostName_ = "";
+    lines_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new SourceFile();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private SourceFile(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    int mutable_bitField0_ = 0;
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 10: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            filePath_ = s;
+            break;
+          }
+          case 18: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            hostName_ = s;
+            break;
+          }
+          case 26: {
+            java.lang.String s = input.readStringRequireUtf8();
+            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
+              lines_ = new com.google.protobuf.LazyStringArrayList();
+              mutable_bitField0_ |= 0x00000001;
+            }
+            lines_.add(s);
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      if (((mutable_bitField0_ & 0x00000001) != 0)) {
+        lines_ = lines_.getUnmodifiableView();
+      }
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.util.DebugEventProtos.internal_static_tensorflow_SourceFile_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.util.DebugEventProtos.internal_static_tensorflow_SourceFile_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.util.SourceFile.class, org.tensorflow.proto.util.SourceFile.Builder.class);
+  }
+
+  public static final int FILE_PATH_FIELD_NUMBER = 1;
+  private volatile java.lang.Object filePath_;
+  /**
+   * <pre>
+   * Path to the file.
+   * </pre>
+   *
+   * <code>string file_path = 1;</code>
+   */
+  public java.lang.String getFilePath() {
+    java.lang.Object ref = filePath_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      filePath_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * Path to the file.
+   * </pre>
+   *
+   * <code>string file_path = 1;</code>
+   */
+  public com.google.protobuf.ByteString
+      getFilePathBytes() {
+    java.lang.Object ref = filePath_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      filePath_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int HOST_NAME_FIELD_NUMBER = 2;
+  private volatile java.lang.Object hostName_;
+  /**
+   * <pre>
+   * Name of the host on which the file is located.
+   * </pre>
+   *
+   * <code>string host_name = 2;</code>
+   */
+  public java.lang.String getHostName() {
+    java.lang.Object ref = hostName_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      hostName_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * Name of the host on which the file is located.
+   * </pre>
+   *
+   * <code>string host_name = 2;</code>
+   */
+  public com.google.protobuf.ByteString
+      getHostNameBytes() {
+    java.lang.Object ref = hostName_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      hostName_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int LINES_FIELD_NUMBER = 3;
+  private com.google.protobuf.LazyStringList lines_;
+  /**
+   * <pre>
+   * Line-by-line content of the file.
+   * </pre>
+   *
+   * <code>repeated string lines = 3;</code>
+   */
+  public com.google.protobuf.ProtocolStringList
+      getLinesList() {
+    return lines_;
+  }
+  /**
+   * <pre>
+   * Line-by-line content of the file.
+   * </pre>
+   *
+   * <code>repeated string lines = 3;</code>
+   */
+  public int getLinesCount() {
+    return lines_.size();
+  }
+  /**
+   * <pre>
+   * Line-by-line content of the file.
+   * </pre>
+   *
+   * <code>repeated string lines = 3;</code>
+   */
+  public java.lang.String getLines(int index) {
+    return lines_.get(index);
+  }
+  /**
+   * <pre>
+   * Line-by-line content of the file.
+   * </pre>
+   *
+   * <code>repeated string lines = 3;</code>
+   */
+  public com.google.protobuf.ByteString
+      getLinesBytes(int index) {
+    return lines_.getByteString(index);
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    if (!getFilePathBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 1, filePath_);
+    }
+    if (!getHostNameBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 2, hostName_);
+    }
+    for (int i = 0; i < lines_.size(); i++) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 3, lines_.getRaw(i));
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (!getFilePathBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, filePath_);
+    }
+    if (!getHostNameBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(2, hostName_);
+    }
+    {
+      int dataSize = 0;
+      for (int i = 0; i < lines_.size(); i++) {
+        dataSize += computeStringSizeNoTag(lines_.getRaw(i));
+      }
+      size += dataSize;
+      size += 1 * getLinesList().size();
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.util.SourceFile)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.util.SourceFile other = (org.tensorflow.proto.util.SourceFile) obj;
+
+    if (!getFilePath()
+        .equals(other.getFilePath())) return false;
+    if (!getHostName()
+        .equals(other.getHostName())) return false;
+    if (!getLinesList()
+        .equals(other.getLinesList())) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    hash = (37 * hash) + FILE_PATH_FIELD_NUMBER;
+    hash = (53 * hash) + getFilePath().hashCode();
+    hash = (37 * hash) + HOST_NAME_FIELD_NUMBER;
+    hash = (53 * hash) + getHostName().hashCode();
+    if (getLinesCount() > 0) {
+      hash = (37 * hash) + LINES_FIELD_NUMBER;
+      hash = (53 * hash) + getLinesList().hashCode();
+    }
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.util.SourceFile parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.SourceFile parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.SourceFile parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.SourceFile parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.SourceFile parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.SourceFile parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.SourceFile parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.SourceFile parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.SourceFile parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.SourceFile parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.SourceFile parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.SourceFile parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.util.SourceFile prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * <pre>
+   * Content of a source file involved in the execution of the debugged TensorFlow
+   * program.
+   * </pre>
+   *
+   * Protobuf type {@code tensorflow.SourceFile}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.SourceFile)
+      org.tensorflow.proto.util.SourceFileOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.util.DebugEventProtos.internal_static_tensorflow_SourceFile_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.util.DebugEventProtos.internal_static_tensorflow_SourceFile_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.util.SourceFile.class, org.tensorflow.proto.util.SourceFile.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.util.SourceFile.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      filePath_ = "";
+
+      hostName_ = "";
+
+      lines_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+      bitField0_ = (bitField0_ & ~0x00000001);
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.util.DebugEventProtos.internal_static_tensorflow_SourceFile_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.SourceFile getDefaultInstanceForType() {
+      return org.tensorflow.proto.util.SourceFile.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.SourceFile build() {
+      org.tensorflow.proto.util.SourceFile result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.SourceFile buildPartial() {
+      org.tensorflow.proto.util.SourceFile result = new org.tensorflow.proto.util.SourceFile(this);
+      int from_bitField0_ = bitField0_;
+      result.filePath_ = filePath_;
+      result.hostName_ = hostName_;
+      if (((bitField0_ & 0x00000001) != 0)) {
+        lines_ = lines_.getUnmodifiableView();
+        bitField0_ = (bitField0_ & ~0x00000001);
+      }
+      result.lines_ = lines_;
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.util.SourceFile) {
+        return mergeFrom((org.tensorflow.proto.util.SourceFile)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.util.SourceFile other) {
+      if (other == org.tensorflow.proto.util.SourceFile.getDefaultInstance()) return this;
+      if (!other.getFilePath().isEmpty()) {
+        filePath_ = other.filePath_;
+        onChanged();
+      }
+      if (!other.getHostName().isEmpty()) {
+        hostName_ = other.hostName_;
+        onChanged();
+      }
+      if (!other.lines_.isEmpty()) {
+        if (lines_.isEmpty()) {
+          lines_ = other.lines_;
+          bitField0_ = (bitField0_ & ~0x00000001);
+        } else {
+          ensureLinesIsMutable();
+          lines_.addAll(other.lines_);
+        }
+        onChanged();
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.util.SourceFile parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.util.SourceFile) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+    private int bitField0_;
+
+    private java.lang.Object filePath_ = "";
+    /**
+     * <pre>
+     * Path to the file.
+     * </pre>
+     *
+     * <code>string file_path = 1;</code>
+     */
+    public java.lang.String getFilePath() {
+      java.lang.Object ref = filePath_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        filePath_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Path to the file.
+     * </pre>
+     *
+     * <code>string file_path = 1;</code>
+     */
+    public com.google.protobuf.ByteString
+        getFilePathBytes() {
+      java.lang.Object ref = filePath_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        filePath_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Path to the file.
+     * </pre>
+     *
+     * <code>string file_path = 1;</code>
+     */
+    public Builder setFilePath(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      filePath_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Path to the file.
+     * </pre>
+     *
+     * <code>string file_path = 1;</code>
+     */
+    public Builder clearFilePath() {
+      
+      filePath_ = getDefaultInstance().getFilePath();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Path to the file.
+     * </pre>
+     *
+     * <code>string file_path = 1;</code>
+     */
+    public Builder setFilePathBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      filePath_ = value;
+      onChanged();
+      return this;
+    }
+
+    private java.lang.Object hostName_ = "";
+    /**
+     * <pre>
+     * Name of the host on which the file is located.
+     * </pre>
+     *
+     * <code>string host_name = 2;</code>
+     */
+    public java.lang.String getHostName() {
+      java.lang.Object ref = hostName_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        hostName_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Name of the host on which the file is located.
+     * </pre>
+     *
+     * <code>string host_name = 2;</code>
+     */
+    public com.google.protobuf.ByteString
+        getHostNameBytes() {
+      java.lang.Object ref = hostName_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        hostName_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Name of the host on which the file is located.
+     * </pre>
+     *
+     * <code>string host_name = 2;</code>
+     */
+    public Builder setHostName(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      hostName_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Name of the host on which the file is located.
+     * </pre>
+     *
+     * <code>string host_name = 2;</code>
+     */
+    public Builder clearHostName() {
+      
+      hostName_ = getDefaultInstance().getHostName();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Name of the host on which the file is located.
+     * </pre>
+     *
+     * <code>string host_name = 2;</code>
+     */
+    public Builder setHostNameBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      hostName_ = value;
+      onChanged();
+      return this;
+    }
+
+    private com.google.protobuf.LazyStringList lines_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+    private void ensureLinesIsMutable() {
+      if (!((bitField0_ & 0x00000001) != 0)) {
+        lines_ = new com.google.protobuf.LazyStringArrayList(lines_);
+        bitField0_ |= 0x00000001;
+       }
+    }
+    /**
+     * <pre>
+     * Line-by-line content of the file.
+     * </pre>
+     *
+     * <code>repeated string lines = 3;</code>
+     */
+    public com.google.protobuf.ProtocolStringList
+        getLinesList() {
+      return lines_.getUnmodifiableView();
+    }
+    /**
+     * <pre>
+     * Line-by-line content of the file.
+     * </pre>
+     *
+     * <code>repeated string lines = 3;</code>
+     */
+    public int getLinesCount() {
+      return lines_.size();
+    }
+    /**
+     * <pre>
+     * Line-by-line content of the file.
+     * </pre>
+     *
+     * <code>repeated string lines = 3;</code>
+     */
+    public java.lang.String getLines(int index) {
+      return lines_.get(index);
+    }
+    /**
+     * <pre>
+     * Line-by-line content of the file.
+     * </pre>
+     *
+     * <code>repeated string lines = 3;</code>
+     */
+    public com.google.protobuf.ByteString
+        getLinesBytes(int index) {
+      return lines_.getByteString(index);
+    }
+    /**
+     * <pre>
+     * Line-by-line content of the file.
+     * </pre>
+     *
+     * <code>repeated string lines = 3;</code>
+     */
+    public Builder setLines(
+        int index, java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  ensureLinesIsMutable();
+      lines_.set(index, value);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Line-by-line content of the file.
+     * </pre>
+     *
+     * <code>repeated string lines = 3;</code>
+     */
+    public Builder addLines(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  ensureLinesIsMutable();
+      lines_.add(value);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Line-by-line content of the file.
+     * </pre>
+     *
+     * <code>repeated string lines = 3;</code>
+     */
+    public Builder addAllLines(
+        java.lang.Iterable<java.lang.String> values) {
+      ensureLinesIsMutable();
+      com.google.protobuf.AbstractMessageLite.Builder.addAll(
+          values, lines_);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Line-by-line content of the file.
+     * </pre>
+     *
+     * <code>repeated string lines = 3;</code>
+     */
+    public Builder clearLines() {
+      lines_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+      bitField0_ = (bitField0_ & ~0x00000001);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Line-by-line content of the file.
+     * </pre>
+     *
+     * <code>repeated string lines = 3;</code>
+     */
+    public Builder addLinesBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      ensureLinesIsMutable();
+      lines_.add(value);
+      onChanged();
+      return this;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.SourceFile)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.SourceFile)
+  private static final org.tensorflow.proto.util.SourceFile DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.util.SourceFile();
+  }
+
+  public static org.tensorflow.proto.util.SourceFile getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<SourceFile>
+      PARSER = new com.google.protobuf.AbstractParser<SourceFile>() {
+    @java.lang.Override
+    public SourceFile parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new SourceFile(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<SourceFile> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<SourceFile> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.util.SourceFile getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SourceFileOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SourceFileOrBuilder.java
new file mode 100644
index 00000000000..b05d03bfb81
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/SourceFileOrBuilder.java
@@ -0,0 +1,80 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/debug_event.proto
+
+package org.tensorflow.proto.util;
+
+public interface SourceFileOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.SourceFile)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <pre>
+   * Path to the file.
+   * </pre>
+   *
+   * <code>string file_path = 1;</code>
+   */
+  java.lang.String getFilePath();
+  /**
+   * <pre>
+   * Path to the file.
+   * </pre>
+   *
+   * <code>string file_path = 1;</code>
+   */
+  com.google.protobuf.ByteString
+      getFilePathBytes();
+
+  /**
+   * <pre>
+   * Name of the host on which the file is located.
+   * </pre>
+   *
+   * <code>string host_name = 2;</code>
+   */
+  java.lang.String getHostName();
+  /**
+   * <pre>
+   * Name of the host on which the file is located.
+   * </pre>
+   *
+   * <code>string host_name = 2;</code>
+   */
+  com.google.protobuf.ByteString
+      getHostNameBytes();
+
+  /**
+   * <pre>
+   * Line-by-line content of the file.
+   * </pre>
+   *
+   * <code>repeated string lines = 3;</code>
+   */
+  java.util.List<java.lang.String>
+      getLinesList();
+  /**
+   * <pre>
+   * Line-by-line content of the file.
+   * </pre>
+   *
+   * <code>repeated string lines = 3;</code>
+   */
+  int getLinesCount();
+  /**
+   * <pre>
+   * Line-by-line content of the file.
+   * </pre>
+   *
+   * <code>repeated string lines = 3;</code>
+   */
+  java.lang.String getLines(int index);
+  /**
+   * <pre>
+   * Line-by-line content of the file.
+   * </pre>
+   *
+   * <code>repeated string lines = 3;</code>
+   */
+  com.google.protobuf.ByteString
+      getLinesBytes(int index);
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/StackFrameWithId.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/StackFrameWithId.java
new file mode 100644
index 00000000000..e41ffbd6cc3
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/StackFrameWithId.java
@@ -0,0 +1,835 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/debug_event.proto
+
+package org.tensorflow.proto.util;
+
+/**
+ * <pre>
+ * A stack frame with ID.
+ * </pre>
+ *
+ * Protobuf type {@code tensorflow.StackFrameWithId}
+ */
+public  final class StackFrameWithId extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.StackFrameWithId)
+    StackFrameWithIdOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use StackFrameWithId.newBuilder() to construct.
+  private StackFrameWithId(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private StackFrameWithId() {
+    id_ = "";
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new StackFrameWithId();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private StackFrameWithId(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 10: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            id_ = s;
+            break;
+          }
+          case 18: {
+            org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol.Builder subBuilder = null;
+            if (fileLineCol_ != null) {
+              subBuilder = fileLineCol_.toBuilder();
+            }
+            fileLineCol_ = input.readMessage(org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol.parser(), extensionRegistry);
+            if (subBuilder != null) {
+              subBuilder.mergeFrom(fileLineCol_);
+              fileLineCol_ = subBuilder.buildPartial();
+            }
+
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.util.DebugEventProtos.internal_static_tensorflow_StackFrameWithId_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.util.DebugEventProtos.internal_static_tensorflow_StackFrameWithId_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.util.StackFrameWithId.class, org.tensorflow.proto.util.StackFrameWithId.Builder.class);
+  }
+
+  public static final int ID_FIELD_NUMBER = 1;
+  private volatile java.lang.Object id_;
+  /**
+   * <pre>
+   * A unique ID for the stack frame: A UUID-like string.
+   * </pre>
+   *
+   * <code>string id = 1;</code>
+   */
+  public java.lang.String getId() {
+    java.lang.Object ref = id_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      id_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * A unique ID for the stack frame: A UUID-like string.
+   * </pre>
+   *
+   * <code>string id = 1;</code>
+   */
+  public com.google.protobuf.ByteString
+      getIdBytes() {
+    java.lang.Object ref = id_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      id_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int FILE_LINE_COL_FIELD_NUMBER = 2;
+  private org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol fileLineCol_;
+  /**
+   * <pre>
+   * Stack frame, i.e., a frame of a stack trace, containing information
+   * regarding the file name, line number, function name, code content
+   * of the line, and column number (if available).
+   * </pre>
+   *
+   * <code>.tensorflow.GraphDebugInfo.FileLineCol file_line_col = 2;</code>
+   */
+  public boolean hasFileLineCol() {
+    return fileLineCol_ != null;
+  }
+  /**
+   * <pre>
+   * Stack frame, i.e., a frame of a stack trace, containing information
+   * regarding the file name, line number, function name, code content
+   * of the line, and column number (if available).
+   * </pre>
+   *
+   * <code>.tensorflow.GraphDebugInfo.FileLineCol file_line_col = 2;</code>
+   */
+  public org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol getFileLineCol() {
+    return fileLineCol_ == null ? org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol.getDefaultInstance() : fileLineCol_;
+  }
+  /**
+   * <pre>
+   * Stack frame, i.e., a frame of a stack trace, containing information
+   * regarding the file name, line number, function name, code content
+   * of the line, and column number (if available).
+   * </pre>
+   *
+   * <code>.tensorflow.GraphDebugInfo.FileLineCol file_line_col = 2;</code>
+   */
+  public org.tensorflow.proto.framework.GraphDebugInfo.FileLineColOrBuilder getFileLineColOrBuilder() {
+    return getFileLineCol();
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    if (!getIdBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 1, id_);
+    }
+    if (fileLineCol_ != null) {
+      output.writeMessage(2, getFileLineCol());
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (!getIdBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, id_);
+    }
+    if (fileLineCol_ != null) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(2, getFileLineCol());
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.util.StackFrameWithId)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.util.StackFrameWithId other = (org.tensorflow.proto.util.StackFrameWithId) obj;
+
+    if (!getId()
+        .equals(other.getId())) return false;
+    if (hasFileLineCol() != other.hasFileLineCol()) return false;
+    if (hasFileLineCol()) {
+      if (!getFileLineCol()
+          .equals(other.getFileLineCol())) return false;
+    }
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    hash = (37 * hash) + ID_FIELD_NUMBER;
+    hash = (53 * hash) + getId().hashCode();
+    if (hasFileLineCol()) {
+      hash = (37 * hash) + FILE_LINE_COL_FIELD_NUMBER;
+      hash = (53 * hash) + getFileLineCol().hashCode();
+    }
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.util.StackFrameWithId parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.StackFrameWithId parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.StackFrameWithId parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.StackFrameWithId parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.StackFrameWithId parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.StackFrameWithId parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.StackFrameWithId parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.StackFrameWithId parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.StackFrameWithId parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.StackFrameWithId parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.StackFrameWithId parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.StackFrameWithId parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.util.StackFrameWithId prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * <pre>
+   * A stack frame with ID.
+   * </pre>
+   *
+   * Protobuf type {@code tensorflow.StackFrameWithId}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.StackFrameWithId)
+      org.tensorflow.proto.util.StackFrameWithIdOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.util.DebugEventProtos.internal_static_tensorflow_StackFrameWithId_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.util.DebugEventProtos.internal_static_tensorflow_StackFrameWithId_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.util.StackFrameWithId.class, org.tensorflow.proto.util.StackFrameWithId.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.util.StackFrameWithId.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      id_ = "";
+
+      if (fileLineColBuilder_ == null) {
+        fileLineCol_ = null;
+      } else {
+        fileLineCol_ = null;
+        fileLineColBuilder_ = null;
+      }
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.util.DebugEventProtos.internal_static_tensorflow_StackFrameWithId_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.StackFrameWithId getDefaultInstanceForType() {
+      return org.tensorflow.proto.util.StackFrameWithId.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.StackFrameWithId build() {
+      org.tensorflow.proto.util.StackFrameWithId result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.StackFrameWithId buildPartial() {
+      org.tensorflow.proto.util.StackFrameWithId result = new org.tensorflow.proto.util.StackFrameWithId(this);
+      result.id_ = id_;
+      if (fileLineColBuilder_ == null) {
+        result.fileLineCol_ = fileLineCol_;
+      } else {
+        result.fileLineCol_ = fileLineColBuilder_.build();
+      }
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.util.StackFrameWithId) {
+        return mergeFrom((org.tensorflow.proto.util.StackFrameWithId)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.util.StackFrameWithId other) {
+      if (other == org.tensorflow.proto.util.StackFrameWithId.getDefaultInstance()) return this;
+      if (!other.getId().isEmpty()) {
+        id_ = other.id_;
+        onChanged();
+      }
+      if (other.hasFileLineCol()) {
+        mergeFileLineCol(other.getFileLineCol());
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.util.StackFrameWithId parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.util.StackFrameWithId) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+
+    private java.lang.Object id_ = "";
+    /**
+     * <pre>
+     * A unique ID for the stack frame: A UUID-like string.
+     * </pre>
+     *
+     * <code>string id = 1;</code>
+     */
+    public java.lang.String getId() {
+      java.lang.Object ref = id_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        id_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * A unique ID for the stack frame: A UUID-like string.
+     * </pre>
+     *
+     * <code>string id = 1;</code>
+     */
+    public com.google.protobuf.ByteString
+        getIdBytes() {
+      java.lang.Object ref = id_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        id_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * A unique ID for the stack frame: A UUID-like string.
+     * </pre>
+     *
+     * <code>string id = 1;</code>
+     */
+    public Builder setId(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      id_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * A unique ID for the stack frame: A UUID-like string.
+     * </pre>
+     *
+     * <code>string id = 1;</code>
+     */
+    public Builder clearId() {
+      
+      id_ = getDefaultInstance().getId();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * A unique ID for the stack frame: A UUID-like string.
+     * </pre>
+     *
+     * <code>string id = 1;</code>
+     */
+    public Builder setIdBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      id_ = value;
+      onChanged();
+      return this;
+    }
+
+    private org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol fileLineCol_;
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol, org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol.Builder, org.tensorflow.proto.framework.GraphDebugInfo.FileLineColOrBuilder> fileLineColBuilder_;
+    /**
+     * <pre>
+     * Stack frame, i.e., a frame of a stack trace, containing information
+     * regarding the file name, line number, function name, code content
+     * of the line, and column number (if available).
+     * </pre>
+     *
+     * <code>.tensorflow.GraphDebugInfo.FileLineCol file_line_col = 2;</code>
+     */
+    public boolean hasFileLineCol() {
+      return fileLineColBuilder_ != null || fileLineCol_ != null;
+    }
+    /**
+     * <pre>
+     * Stack frame, i.e., a frame of a stack trace, containing information
+     * regarding the file name, line number, function name, code content
+     * of the line, and column number (if available).
+     * </pre>
+     *
+     * <code>.tensorflow.GraphDebugInfo.FileLineCol file_line_col = 2;</code>
+     */
+    public org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol getFileLineCol() {
+      if (fileLineColBuilder_ == null) {
+        return fileLineCol_ == null ? org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol.getDefaultInstance() : fileLineCol_;
+      } else {
+        return fileLineColBuilder_.getMessage();
+      }
+    }
+    /**
+     * <pre>
+     * Stack frame, i.e., a frame of a stack trace, containing information
+     * regarding the file name, line number, function name, code content
+     * of the line, and column number (if available).
+     * </pre>
+     *
+     * <code>.tensorflow.GraphDebugInfo.FileLineCol file_line_col = 2;</code>
+     */
+    public Builder setFileLineCol(org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol value) {
+      if (fileLineColBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        fileLineCol_ = value;
+        onChanged();
+      } else {
+        fileLineColBuilder_.setMessage(value);
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Stack frame, i.e., a frame of a stack trace, containing information
+     * regarding the file name, line number, function name, code content
+     * of the line, and column number (if available).
+     * </pre>
+     *
+     * <code>.tensorflow.GraphDebugInfo.FileLineCol file_line_col = 2;</code>
+     */
+    public Builder setFileLineCol(
+        org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol.Builder builderForValue) {
+      if (fileLineColBuilder_ == null) {
+        fileLineCol_ = builderForValue.build();
+        onChanged();
+      } else {
+        fileLineColBuilder_.setMessage(builderForValue.build());
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Stack frame, i.e., a frame of a stack trace, containing information
+     * regarding the file name, line number, function name, code content
+     * of the line, and column number (if available).
+     * </pre>
+     *
+     * <code>.tensorflow.GraphDebugInfo.FileLineCol file_line_col = 2;</code>
+     */
+    public Builder mergeFileLineCol(org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol value) {
+      if (fileLineColBuilder_ == null) {
+        if (fileLineCol_ != null) {
+          fileLineCol_ =
+            org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol.newBuilder(fileLineCol_).mergeFrom(value).buildPartial();
+        } else {
+          fileLineCol_ = value;
+        }
+        onChanged();
+      } else {
+        fileLineColBuilder_.mergeFrom(value);
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Stack frame, i.e., a frame of a stack trace, containing information
+     * regarding the file name, line number, function name, code content
+     * of the line, and column number (if available).
+     * </pre>
+     *
+     * <code>.tensorflow.GraphDebugInfo.FileLineCol file_line_col = 2;</code>
+     */
+    public Builder clearFileLineCol() {
+      if (fileLineColBuilder_ == null) {
+        fileLineCol_ = null;
+        onChanged();
+      } else {
+        fileLineCol_ = null;
+        fileLineColBuilder_ = null;
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Stack frame, i.e., a frame of a stack trace, containing information
+     * regarding the file name, line number, function name, code content
+     * of the line, and column number (if available).
+     * </pre>
+     *
+     * <code>.tensorflow.GraphDebugInfo.FileLineCol file_line_col = 2;</code>
+     */
+    public org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol.Builder getFileLineColBuilder() {
+      
+      onChanged();
+      return getFileLineColFieldBuilder().getBuilder();
+    }
+    /**
+     * <pre>
+     * Stack frame, i.e., a frame of a stack trace, containing information
+     * regarding the file name, line number, function name, code content
+     * of the line, and column number (if available).
+     * </pre>
+     *
+     * <code>.tensorflow.GraphDebugInfo.FileLineCol file_line_col = 2;</code>
+     */
+    public org.tensorflow.proto.framework.GraphDebugInfo.FileLineColOrBuilder getFileLineColOrBuilder() {
+      if (fileLineColBuilder_ != null) {
+        return fileLineColBuilder_.getMessageOrBuilder();
+      } else {
+        return fileLineCol_ == null ?
+            org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol.getDefaultInstance() : fileLineCol_;
+      }
+    }
+    /**
+     * <pre>
+     * Stack frame, i.e., a frame of a stack trace, containing information
+     * regarding the file name, line number, function name, code content
+     * of the line, and column number (if available).
+     * </pre>
+     *
+     * <code>.tensorflow.GraphDebugInfo.FileLineCol file_line_col = 2;</code>
+     */
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol, org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol.Builder, org.tensorflow.proto.framework.GraphDebugInfo.FileLineColOrBuilder> 
+        getFileLineColFieldBuilder() {
+      if (fileLineColBuilder_ == null) {
+        fileLineColBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+            org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol, org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol.Builder, org.tensorflow.proto.framework.GraphDebugInfo.FileLineColOrBuilder>(
+                getFileLineCol(),
+                getParentForChildren(),
+                isClean());
+        fileLineCol_ = null;
+      }
+      return fileLineColBuilder_;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.StackFrameWithId)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.StackFrameWithId)
+  private static final org.tensorflow.proto.util.StackFrameWithId DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.util.StackFrameWithId();
+  }
+
+  public static org.tensorflow.proto.util.StackFrameWithId getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<StackFrameWithId>
+      PARSER = new com.google.protobuf.AbstractParser<StackFrameWithId>() {
+    @java.lang.Override
+    public StackFrameWithId parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new StackFrameWithId(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<StackFrameWithId> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<StackFrameWithId> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.util.StackFrameWithId getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/StackFrameWithIdOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/StackFrameWithIdOrBuilder.java
new file mode 100644
index 00000000000..b1132ff9e72
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/StackFrameWithIdOrBuilder.java
@@ -0,0 +1,58 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/debug_event.proto
+
+package org.tensorflow.proto.util;
+
+public interface StackFrameWithIdOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.StackFrameWithId)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <pre>
+   * A unique ID for the stack frame: A UUID-like string.
+   * </pre>
+   *
+   * <code>string id = 1;</code>
+   */
+  java.lang.String getId();
+  /**
+   * <pre>
+   * A unique ID for the stack frame: A UUID-like string.
+   * </pre>
+   *
+   * <code>string id = 1;</code>
+   */
+  com.google.protobuf.ByteString
+      getIdBytes();
+
+  /**
+   * <pre>
+   * Stack frame, i.e., a frame of a stack trace, containing information
+   * regarding the file name, line number, function name, code content
+   * of the line, and column number (if available).
+   * </pre>
+   *
+   * <code>.tensorflow.GraphDebugInfo.FileLineCol file_line_col = 2;</code>
+   */
+  boolean hasFileLineCol();
+  /**
+   * <pre>
+   * Stack frame, i.e., a frame of a stack trace, containing information
+   * regarding the file name, line number, function name, code content
+   * of the line, and column number (if available).
+   * </pre>
+   *
+   * <code>.tensorflow.GraphDebugInfo.FileLineCol file_line_col = 2;</code>
+   */
+  org.tensorflow.proto.framework.GraphDebugInfo.FileLineCol getFileLineCol();
+  /**
+   * <pre>
+   * Stack frame, i.e., a frame of a stack trace, containing information
+   * regarding the file name, line number, function name, code content
+   * of the line, and column number (if available).
+   * </pre>
+   *
+   * <code>.tensorflow.GraphDebugInfo.FileLineCol file_line_col = 2;</code>
+   */
+  org.tensorflow.proto.framework.GraphDebugInfo.FileLineColOrBuilder getFileLineColOrBuilder();
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/TaggedRunMetadata.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/TaggedRunMetadata.java
new file mode 100644
index 00000000000..4880946d5df
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/TaggedRunMetadata.java
@@ -0,0 +1,663 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/util/event.proto
+
+package org.tensorflow.proto.util;
+
+/**
+ * <pre>
+ * For logging the metadata output for a single session.run() call.
+ * </pre>
+ *
+ * Protobuf type {@code tensorflow.TaggedRunMetadata}
+ */
+public  final class TaggedRunMetadata extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.TaggedRunMetadata)
+    TaggedRunMetadataOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use TaggedRunMetadata.newBuilder() to construct.
+  private TaggedRunMetadata(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private TaggedRunMetadata() {
+    tag_ = "";
+    runMetadata_ = com.google.protobuf.ByteString.EMPTY;
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new TaggedRunMetadata();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private TaggedRunMetadata(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 10: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            tag_ = s;
+            break;
+          }
+          case 18: {
+
+            runMetadata_ = input.readBytes();
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.util.EventProtos.internal_static_tensorflow_TaggedRunMetadata_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.util.EventProtos.internal_static_tensorflow_TaggedRunMetadata_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.util.TaggedRunMetadata.class, org.tensorflow.proto.util.TaggedRunMetadata.Builder.class);
+  }
+
+  public static final int TAG_FIELD_NUMBER = 1;
+  private volatile java.lang.Object tag_;
+  /**
+   * <pre>
+   * Tag name associated with this metadata.
+   * </pre>
+   *
+   * <code>string tag = 1;</code>
+   */
+  public java.lang.String getTag() {
+    java.lang.Object ref = tag_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      tag_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * Tag name associated with this metadata.
+   * </pre>
+   *
+   * <code>string tag = 1;</code>
+   */
+  public com.google.protobuf.ByteString
+      getTagBytes() {
+    java.lang.Object ref = tag_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      tag_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int RUN_METADATA_FIELD_NUMBER = 2;
+  private com.google.protobuf.ByteString runMetadata_;
+  /**
+   * <pre>
+   * Byte-encoded version of the `RunMetadata` proto in order to allow lazy
+   * deserialization.
+   * </pre>
+   *
+   * <code>bytes run_metadata = 2;</code>
+   */
+  public com.google.protobuf.ByteString getRunMetadata() {
+    return runMetadata_;
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    if (!getTagBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 1, tag_);
+    }
+    if (!runMetadata_.isEmpty()) {
+      output.writeBytes(2, runMetadata_);
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (!getTagBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, tag_);
+    }
+    if (!runMetadata_.isEmpty()) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeBytesSize(2, runMetadata_);
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.util.TaggedRunMetadata)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.util.TaggedRunMetadata other = (org.tensorflow.proto.util.TaggedRunMetadata) obj;
+
+    if (!getTag()
+        .equals(other.getTag())) return false;
+    if (!getRunMetadata()
+        .equals(other.getRunMetadata())) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    hash = (37 * hash) + TAG_FIELD_NUMBER;
+    hash = (53 * hash) + getTag().hashCode();
+    hash = (37 * hash) + RUN_METADATA_FIELD_NUMBER;
+    hash = (53 * hash) + getRunMetadata().hashCode();
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.util.TaggedRunMetadata parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.TaggedRunMetadata parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.TaggedRunMetadata parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.TaggedRunMetadata parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.TaggedRunMetadata parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.TaggedRunMetadata parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.TaggedRunMetadata parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.TaggedRunMetadata parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.TaggedRunMetadata parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.TaggedRunMetadata parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.TaggedRunMetadata parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.TaggedRunMetadata parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.util.TaggedRunMetadata prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * <pre>
+   * For logging the metadata output for a single session.run() call.
+   * </pre>
+   *
+   * Protobuf type {@code tensorflow.TaggedRunMetadata}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.TaggedRunMetadata)
+      org.tensorflow.proto.util.TaggedRunMetadataOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.util.EventProtos.internal_static_tensorflow_TaggedRunMetadata_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.util.EventProtos.internal_static_tensorflow_TaggedRunMetadata_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.util.TaggedRunMetadata.class, org.tensorflow.proto.util.TaggedRunMetadata.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.util.TaggedRunMetadata.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      tag_ = "";
+
+      runMetadata_ = com.google.protobuf.ByteString.EMPTY;
+
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.util.EventProtos.internal_static_tensorflow_TaggedRunMetadata_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.TaggedRunMetadata getDefaultInstanceForType() {
+      return org.tensorflow.proto.util.TaggedRunMetadata.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.TaggedRunMetadata build() {
+      org.tensorflow.proto.util.TaggedRunMetadata result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.TaggedRunMetadata buildPartial() {
+      org.tensorflow.proto.util.TaggedRunMetadata result = new org.tensorflow.proto.util.TaggedRunMetadata(this);
+      result.tag_ = tag_;
+      result.runMetadata_ = runMetadata_;
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.util.TaggedRunMetadata) {
+        return mergeFrom((org.tensorflow.proto.util.TaggedRunMetadata)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.util.TaggedRunMetadata other) {
+      if (other == org.tensorflow.proto.util.TaggedRunMetadata.getDefaultInstance()) return this;
+      if (!other.getTag().isEmpty()) {
+        tag_ = other.tag_;
+        onChanged();
+      }
+      if (other.getRunMetadata() != com.google.protobuf.ByteString.EMPTY) {
+        setRunMetadata(other.getRunMetadata());
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.util.TaggedRunMetadata parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.util.TaggedRunMetadata) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+
+    private java.lang.Object tag_ = "";
+    /**
+     * <pre>
+     * Tag name associated with this metadata.
+     * </pre>
+     *
+     * <code>string tag = 1;</code>
+     */
+    public java.lang.String getTag() {
+      java.lang.Object ref = tag_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        tag_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Tag name associated with this metadata.
+     * </pre>
+     *
+     * <code>string tag = 1;</code>
+     */
+    public com.google.protobuf.ByteString
+        getTagBytes() {
+      java.lang.Object ref = tag_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        tag_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Tag name associated with this metadata.
+     * </pre>
+     *
+     * <code>string tag = 1;</code>
+     */
+    public Builder setTag(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      tag_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Tag name associated with this metadata.
+     * </pre>
+     *
+     * <code>string tag = 1;</code>
+     */
+    public Builder clearTag() {
+      
+      tag_ = getDefaultInstance().getTag();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Tag name associated with this metadata.
+     * </pre>
+     *
+     * <code>string tag = 1;</code>
+     */
+    public Builder setTagBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      tag_ = value;
+      onChanged();
+      return this;
+    }
+
+    private com.google.protobuf.ByteString runMetadata_ = com.google.protobuf.ByteString.EMPTY;
+    /**
+     * <pre>
+     * Byte-encoded version of the `RunMetadata` proto in order to allow lazy
+     * deserialization.
+     * </pre>
+     *
+     * <code>bytes run_metadata = 2;</code>
+     */
+    public com.google.protobuf.ByteString getRunMetadata() {
+      return runMetadata_;
+    }
+    /**
+     * <pre>
+     * Byte-encoded version of the `RunMetadata` proto in order to allow lazy
+     * deserialization.
+     * </pre>
+     *
+     * <code>bytes run_metadata = 2;</code>
+     */
+    public Builder setRunMetadata(com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      runMetadata_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Byte-encoded version of the `RunMetadata` proto in order to allow lazy
+     * deserialization.
+     * </pre>
+     *
+     * <code>bytes run_metadata = 2;</code>
+     */
+    public Builder clearRunMetadata() {
+      
+      runMetadata_ = getDefaultInstance().getRunMetadata();
+      onChanged();
+      return this;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.TaggedRunMetadata)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.TaggedRunMetadata)
+  private static final org.tensorflow.proto.util.TaggedRunMetadata DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.util.TaggedRunMetadata();
+  }
+
+  public static org.tensorflow.proto.util.TaggedRunMetadata getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<TaggedRunMetadata>
+      PARSER = new com.google.protobuf.AbstractParser<TaggedRunMetadata>() {
+    @java.lang.Override
+    public TaggedRunMetadata parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new TaggedRunMetadata(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<TaggedRunMetadata> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<TaggedRunMetadata> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.util.TaggedRunMetadata getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/TaggedRunMetadataOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/TaggedRunMetadataOrBuilder.java
new file mode 100644
index 00000000000..d364543215c
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/TaggedRunMetadataOrBuilder.java
@@ -0,0 +1,37 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/util/event.proto
+
+package org.tensorflow.proto.util;
+
+public interface TaggedRunMetadataOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.TaggedRunMetadata)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <pre>
+   * Tag name associated with this metadata.
+   * </pre>
+   *
+   * <code>string tag = 1;</code>
+   */
+  java.lang.String getTag();
+  /**
+   * <pre>
+   * Tag name associated with this metadata.
+   * </pre>
+   *
+   * <code>string tag = 1;</code>
+   */
+  com.google.protobuf.ByteString
+      getTagBytes();
+
+  /**
+   * <pre>
+   * Byte-encoded version of the `RunMetadata` proto in order to allow lazy
+   * deserialization.
+   * </pre>
+   *
+   * <code>bytes run_metadata = 2;</code>
+   */
+  com.google.protobuf.ByteString getRunMetadata();
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/TensorBundleProtos.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/TensorBundleProtos.java
new file mode 100644
index 00000000000..3de209fb5a9
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/TensorBundleProtos.java
@@ -0,0 +1,83 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/tensor_bundle.proto
+
+package org.tensorflow.proto.util;
+
+public final class TensorBundleProtos {
+  private TensorBundleProtos() {}
+  public static void registerAllExtensions(
+      com.google.protobuf.ExtensionRegistryLite registry) {
+  }
+
+  public static void registerAllExtensions(
+      com.google.protobuf.ExtensionRegistry registry) {
+    registerAllExtensions(
+        (com.google.protobuf.ExtensionRegistryLite) registry);
+  }
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_BundleHeaderProto_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_BundleHeaderProto_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_BundleEntryProto_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_BundleEntryProto_fieldAccessorTable;
+
+  public static com.google.protobuf.Descriptors.FileDescriptor
+      getDescriptor() {
+    return descriptor;
+  }
+  private static  com.google.protobuf.Descriptors.FileDescriptor
+      descriptor;
+  static {
+    java.lang.String[] descriptorData = {
+      "\n,tensorflow/core/protobuf/tensor_bundle" +
+      ".proto\022\ntensorflow\032,tensorflow/core/fram" +
+      "ework/tensor_shape.proto\032,tensorflow/cor" +
+      "e/framework/tensor_slice.proto\032%tensorfl" +
+      "ow/core/framework/types.proto\032(tensorflo" +
+      "w/core/framework/versions.proto\"\261\001\n\021Bund" +
+      "leHeaderProto\022\022\n\nnum_shards\030\001 \001(\005\022<\n\nend" +
+      "ianness\030\002 \001(\0162(.tensorflow.BundleHeaderP" +
+      "roto.Endianness\022\'\n\007version\030\003 \001(\0132\026.tenso" +
+      "rflow.VersionDef\"!\n\nEndianness\022\n\n\006LITTLE" +
+      "\020\000\022\007\n\003BIG\020\001\"\322\001\n\020BundleEntryProto\022#\n\005dtyp" +
+      "e\030\001 \001(\0162\024.tensorflow.DataType\022+\n\005shape\030\002" +
+      " \001(\0132\034.tensorflow.TensorShapeProto\022\020\n\010sh" +
+      "ard_id\030\003 \001(\005\022\016\n\006offset\030\004 \001(\003\022\014\n\004size\030\005 \001" +
+      "(\003\022\016\n\006crc32c\030\006 \001(\007\022,\n\006slices\030\007 \003(\0132\034.ten" +
+      "sorflow.TensorSliceProtoB~\n\031org.tensorfl" +
+      "ow.proto.utilB\022TensorBundleProtosP\001ZHgit" +
+      "hub.com/tensorflow/tensorflow/tensorflow" +
+      "/go/core/core_protos_go_proto\370\001\001b\006proto3"
+    };
+    descriptor = com.google.protobuf.Descriptors.FileDescriptor
+      .internalBuildGeneratedFileFrom(descriptorData,
+        new com.google.protobuf.Descriptors.FileDescriptor[] {
+          org.tensorflow.proto.framework.TensorShapeProtos.getDescriptor(),
+          org.tensorflow.proto.framework.TensorSliceProtos.getDescriptor(),
+          org.tensorflow.proto.framework.TypesProtos.getDescriptor(),
+          org.tensorflow.proto.framework.VersionsProtos.getDescriptor(),
+        });
+    internal_static_tensorflow_BundleHeaderProto_descriptor =
+      getDescriptor().getMessageTypes().get(0);
+    internal_static_tensorflow_BundleHeaderProto_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_BundleHeaderProto_descriptor,
+        new java.lang.String[] { "NumShards", "Endianness", "Version", });
+    internal_static_tensorflow_BundleEntryProto_descriptor =
+      getDescriptor().getMessageTypes().get(1);
+    internal_static_tensorflow_BundleEntryProto_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_BundleEntryProto_descriptor,
+        new java.lang.String[] { "Dtype", "Shape", "ShardId", "Offset", "Size", "Crc32C", "Slices", });
+    org.tensorflow.proto.framework.TensorShapeProtos.getDescriptor();
+    org.tensorflow.proto.framework.TensorSliceProtos.getDescriptor();
+    org.tensorflow.proto.framework.TypesProtos.getDescriptor();
+    org.tensorflow.proto.framework.VersionsProtos.getDescriptor();
+  }
+
+  // @@protoc_insertion_point(outer_class_scope)
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/TensorDebugMode.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/TensorDebugMode.java
new file mode 100644
index 00000000000..074b1ca74c5
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/TensorDebugMode.java
@@ -0,0 +1,282 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/debug_event.proto
+
+package org.tensorflow.proto.util;
+
+/**
+ * <pre>
+ * Available modes for extracting debugging information from a Tensor.
+ * TODO(cais): Document the detailed column names and semantics in a separate
+ * markdown file once the implementation settles.
+ * </pre>
+ *
+ * Protobuf enum {@code tensorflow.TensorDebugMode}
+ */
+public enum TensorDebugMode
+    implements com.google.protobuf.ProtocolMessageEnum {
+  /**
+   * <code>UNSPECIFIED = 0;</code>
+   */
+  UNSPECIFIED(0),
+  /**
+   * <pre>
+   * Only records what tensors are computed, eagerly or in graphs.
+   * No information regarding the value of the tensor is available.
+   * </pre>
+   *
+   * <code>NO_TENSOR = 1;</code>
+   */
+  NO_TENSOR(1),
+  /**
+   * <pre>
+   * A minimalist health summary for float-type tensors.
+   * Contains information only about the presence/absence of pathological
+   * values including Infinity and NaN.
+   * Applicable only to float dtypes.
+   * </pre>
+   *
+   * <code>CURT_HEALTH = 2;</code>
+   */
+  CURT_HEALTH(2),
+  /**
+   * <pre>
+   * A concise health summary for float-type tensors.
+   * Contains more information that CURT_HEALTH.
+   * Infinity and NaN are treated differently.
+   * Applicable only to float and integer dtypes.
+   * </pre>
+   *
+   * <code>CONCISE_HEALTH = 3;</code>
+   */
+  CONCISE_HEALTH(3),
+  /**
+   * <pre>
+   * A detailed health summary.
+   * Contains further detailed information than `CONCISE_HEALTH`.
+   * Information about device, dtype and shape are included.
+   * Counts for various types of values (Infinity, NaN, negative, zero,
+   * positive) are included.
+   * Applicable to float, integer and boolean dtypes.
+   * </pre>
+   *
+   * <code>FULL_HEALTH = 4;</code>
+   */
+  FULL_HEALTH(4),
+  /**
+   * <pre>
+   * Provides full runtime shape information, up to a maximum rank, beyond
+   * which the dimension sizes are truncated.
+   * </pre>
+   *
+   * <code>SHAPE = 5;</code>
+   */
+  SHAPE(5),
+  /**
+   * <pre>
+   * Full numeric summary.
+   * Including device, dtype, shape, counts of various types of values
+   * (Infinity, NaN, negative, zero, positive), and summary statistics
+   * (minimum, maximum, mean and variance).
+   * Applicable to float, integer and boolean dtypes.
+   * </pre>
+   *
+   * <code>FULL_NUMERICS = 6;</code>
+   */
+  FULL_NUMERICS(6),
+  /**
+   * <pre>
+   * Full tensor value.
+   * </pre>
+   *
+   * <code>FULL_TENSOR = 7;</code>
+   */
+  FULL_TENSOR(7),
+  /**
+   * <pre>
+   * Reduce the elements of a tensor to a rank-1 tensor of shape [3], in which
+   * - the 1st element is -inf if any element of the tensor is -inf,
+   *   or zero otherwise.
+   * - the 2nd element is +inf if any element of the tensor is +inf,
+   *   or zero otherwise.
+   * - the 3rd element is nan if any element of the tensor is nan, or zero
+   *   otherwise.
+   * </pre>
+   *
+   * <code>REDUCE_INF_NAN_THREE_SLOTS = 8;</code>
+   */
+  REDUCE_INF_NAN_THREE_SLOTS(8),
+  UNRECOGNIZED(-1),
+  ;
+
+  /**
+   * <code>UNSPECIFIED = 0;</code>
+   */
+  public static final int UNSPECIFIED_VALUE = 0;
+  /**
+   * <pre>
+   * Only records what tensors are computed, eagerly or in graphs.
+   * No information regarding the value of the tensor is available.
+   * </pre>
+   *
+   * <code>NO_TENSOR = 1;</code>
+   */
+  public static final int NO_TENSOR_VALUE = 1;
+  /**
+   * <pre>
+   * A minimalist health summary for float-type tensors.
+   * Contains information only about the presence/absence of pathological
+   * values including Infinity and NaN.
+   * Applicable only to float dtypes.
+   * </pre>
+   *
+   * <code>CURT_HEALTH = 2;</code>
+   */
+  public static final int CURT_HEALTH_VALUE = 2;
+  /**
+   * <pre>
+   * A concise health summary for float-type tensors.
+   * Contains more information that CURT_HEALTH.
+   * Infinity and NaN are treated differently.
+   * Applicable only to float and integer dtypes.
+   * </pre>
+   *
+   * <code>CONCISE_HEALTH = 3;</code>
+   */
+  public static final int CONCISE_HEALTH_VALUE = 3;
+  /**
+   * <pre>
+   * A detailed health summary.
+   * Contains further detailed information than `CONCISE_HEALTH`.
+   * Information about device, dtype and shape are included.
+   * Counts for various types of values (Infinity, NaN, negative, zero,
+   * positive) are included.
+   * Applicable to float, integer and boolean dtypes.
+   * </pre>
+   *
+   * <code>FULL_HEALTH = 4;</code>
+   */
+  public static final int FULL_HEALTH_VALUE = 4;
+  /**
+   * <pre>
+   * Provides full runtime shape information, up to a maximum rank, beyond
+   * which the dimension sizes are truncated.
+   * </pre>
+   *
+   * <code>SHAPE = 5;</code>
+   */
+  public static final int SHAPE_VALUE = 5;
+  /**
+   * <pre>
+   * Full numeric summary.
+   * Including device, dtype, shape, counts of various types of values
+   * (Infinity, NaN, negative, zero, positive), and summary statistics
+   * (minimum, maximum, mean and variance).
+   * Applicable to float, integer and boolean dtypes.
+   * </pre>
+   *
+   * <code>FULL_NUMERICS = 6;</code>
+   */
+  public static final int FULL_NUMERICS_VALUE = 6;
+  /**
+   * <pre>
+   * Full tensor value.
+   * </pre>
+   *
+   * <code>FULL_TENSOR = 7;</code>
+   */
+  public static final int FULL_TENSOR_VALUE = 7;
+  /**
+   * <pre>
+   * Reduce the elements of a tensor to a rank-1 tensor of shape [3], in which
+   * - the 1st element is -inf if any element of the tensor is -inf,
+   *   or zero otherwise.
+   * - the 2nd element is +inf if any element of the tensor is +inf,
+   *   or zero otherwise.
+   * - the 3rd element is nan if any element of the tensor is nan, or zero
+   *   otherwise.
+   * </pre>
+   *
+   * <code>REDUCE_INF_NAN_THREE_SLOTS = 8;</code>
+   */
+  public static final int REDUCE_INF_NAN_THREE_SLOTS_VALUE = 8;
+
+
+  public final int getNumber() {
+    if (this == UNRECOGNIZED) {
+      throw new java.lang.IllegalArgumentException(
+          "Can't get the number of an unknown enum value.");
+    }
+    return value;
+  }
+
+  /**
+   * @deprecated Use {@link #forNumber(int)} instead.
+   */
+  @java.lang.Deprecated
+  public static TensorDebugMode valueOf(int value) {
+    return forNumber(value);
+  }
+
+  public static TensorDebugMode forNumber(int value) {
+    switch (value) {
+      case 0: return UNSPECIFIED;
+      case 1: return NO_TENSOR;
+      case 2: return CURT_HEALTH;
+      case 3: return CONCISE_HEALTH;
+      case 4: return FULL_HEALTH;
+      case 5: return SHAPE;
+      case 6: return FULL_NUMERICS;
+      case 7: return FULL_TENSOR;
+      case 8: return REDUCE_INF_NAN_THREE_SLOTS;
+      default: return null;
+    }
+  }
+
+  public static com.google.protobuf.Internal.EnumLiteMap<TensorDebugMode>
+      internalGetValueMap() {
+    return internalValueMap;
+  }
+  private static final com.google.protobuf.Internal.EnumLiteMap<
+      TensorDebugMode> internalValueMap =
+        new com.google.protobuf.Internal.EnumLiteMap<TensorDebugMode>() {
+          public TensorDebugMode findValueByNumber(int number) {
+            return TensorDebugMode.forNumber(number);
+          }
+        };
+
+  public final com.google.protobuf.Descriptors.EnumValueDescriptor
+      getValueDescriptor() {
+    return getDescriptor().getValues().get(ordinal());
+  }
+  public final com.google.protobuf.Descriptors.EnumDescriptor
+      getDescriptorForType() {
+    return getDescriptor();
+  }
+  public static final com.google.protobuf.Descriptors.EnumDescriptor
+      getDescriptor() {
+    return org.tensorflow.proto.util.DebugEventProtos.getDescriptor().getEnumTypes().get(0);
+  }
+
+  private static final TensorDebugMode[] VALUES = values();
+
+  public static TensorDebugMode valueOf(
+      com.google.protobuf.Descriptors.EnumValueDescriptor desc) {
+    if (desc.getType() != getDescriptor()) {
+      throw new java.lang.IllegalArgumentException(
+        "EnumValueDescriptor is not for this type.");
+    }
+    if (desc.getIndex() == -1) {
+      return UNRECOGNIZED;
+    }
+    return VALUES[desc.getIndex()];
+  }
+
+  private final int value;
+
+  private TensorDebugMode(int value) {
+    this.value = value;
+  }
+
+  // @@protoc_insertion_point(enum_scope:tensorflow.TensorDebugMode)
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/TensorMetadata.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/TensorMetadata.java
new file mode 100644
index 00000000000..c1ac5388ed4
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/TensorMetadata.java
@@ -0,0 +1,682 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/data/experimental/snapshot.proto
+
+package org.tensorflow.proto.util;
+
+/**
+ * <pre>
+ * Metadata for a single tensor in the Snapshot Record.
+ * </pre>
+ *
+ * Protobuf type {@code tensorflow.data.experimental.TensorMetadata}
+ */
+public  final class TensorMetadata extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.data.experimental.TensorMetadata)
+    TensorMetadataOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use TensorMetadata.newBuilder() to construct.
+  private TensorMetadata(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private TensorMetadata() {
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new TensorMetadata();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private TensorMetadata(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 18: {
+            org.tensorflow.proto.framework.TensorShapeProto.Builder subBuilder = null;
+            if (tensorShape_ != null) {
+              subBuilder = tensorShape_.toBuilder();
+            }
+            tensorShape_ = input.readMessage(org.tensorflow.proto.framework.TensorShapeProto.parser(), extensionRegistry);
+            if (subBuilder != null) {
+              subBuilder.mergeFrom(tensorShape_);
+              tensorShape_ = subBuilder.buildPartial();
+            }
+
+            break;
+          }
+          case 24: {
+
+            tensorSizeBytes_ = input.readInt64();
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.util.SnapshotProtos.internal_static_tensorflow_data_experimental_TensorMetadata_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.util.SnapshotProtos.internal_static_tensorflow_data_experimental_TensorMetadata_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.util.TensorMetadata.class, org.tensorflow.proto.util.TensorMetadata.Builder.class);
+  }
+
+  public static final int TENSOR_SHAPE_FIELD_NUMBER = 2;
+  private org.tensorflow.proto.framework.TensorShapeProto tensorShape_;
+  /**
+   * <code>.tensorflow.TensorShapeProto tensor_shape = 2;</code>
+   */
+  public boolean hasTensorShape() {
+    return tensorShape_ != null;
+  }
+  /**
+   * <code>.tensorflow.TensorShapeProto tensor_shape = 2;</code>
+   */
+  public org.tensorflow.proto.framework.TensorShapeProto getTensorShape() {
+    return tensorShape_ == null ? org.tensorflow.proto.framework.TensorShapeProto.getDefaultInstance() : tensorShape_;
+  }
+  /**
+   * <code>.tensorflow.TensorShapeProto tensor_shape = 2;</code>
+   */
+  public org.tensorflow.proto.framework.TensorShapeProtoOrBuilder getTensorShapeOrBuilder() {
+    return getTensorShape();
+  }
+
+  public static final int TENSOR_SIZE_BYTES_FIELD_NUMBER = 3;
+  private long tensorSizeBytes_;
+  /**
+   * <pre>
+   * Number of uncompressed bytes used to store the tensor representation.
+   * </pre>
+   *
+   * <code>int64 tensor_size_bytes = 3;</code>
+   */
+  public long getTensorSizeBytes() {
+    return tensorSizeBytes_;
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    if (tensorShape_ != null) {
+      output.writeMessage(2, getTensorShape());
+    }
+    if (tensorSizeBytes_ != 0L) {
+      output.writeInt64(3, tensorSizeBytes_);
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (tensorShape_ != null) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(2, getTensorShape());
+    }
+    if (tensorSizeBytes_ != 0L) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt64Size(3, tensorSizeBytes_);
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.util.TensorMetadata)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.util.TensorMetadata other = (org.tensorflow.proto.util.TensorMetadata) obj;
+
+    if (hasTensorShape() != other.hasTensorShape()) return false;
+    if (hasTensorShape()) {
+      if (!getTensorShape()
+          .equals(other.getTensorShape())) return false;
+    }
+    if (getTensorSizeBytes()
+        != other.getTensorSizeBytes()) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    if (hasTensorShape()) {
+      hash = (37 * hash) + TENSOR_SHAPE_FIELD_NUMBER;
+      hash = (53 * hash) + getTensorShape().hashCode();
+    }
+    hash = (37 * hash) + TENSOR_SIZE_BYTES_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+        getTensorSizeBytes());
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.util.TensorMetadata parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.TensorMetadata parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.TensorMetadata parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.TensorMetadata parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.TensorMetadata parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.TensorMetadata parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.TensorMetadata parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.TensorMetadata parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.TensorMetadata parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.TensorMetadata parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.TensorMetadata parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.TensorMetadata parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.util.TensorMetadata prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * <pre>
+   * Metadata for a single tensor in the Snapshot Record.
+   * </pre>
+   *
+   * Protobuf type {@code tensorflow.data.experimental.TensorMetadata}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.data.experimental.TensorMetadata)
+      org.tensorflow.proto.util.TensorMetadataOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.util.SnapshotProtos.internal_static_tensorflow_data_experimental_TensorMetadata_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.util.SnapshotProtos.internal_static_tensorflow_data_experimental_TensorMetadata_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.util.TensorMetadata.class, org.tensorflow.proto.util.TensorMetadata.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.util.TensorMetadata.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      if (tensorShapeBuilder_ == null) {
+        tensorShape_ = null;
+      } else {
+        tensorShape_ = null;
+        tensorShapeBuilder_ = null;
+      }
+      tensorSizeBytes_ = 0L;
+
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.util.SnapshotProtos.internal_static_tensorflow_data_experimental_TensorMetadata_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.TensorMetadata getDefaultInstanceForType() {
+      return org.tensorflow.proto.util.TensorMetadata.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.TensorMetadata build() {
+      org.tensorflow.proto.util.TensorMetadata result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.TensorMetadata buildPartial() {
+      org.tensorflow.proto.util.TensorMetadata result = new org.tensorflow.proto.util.TensorMetadata(this);
+      if (tensorShapeBuilder_ == null) {
+        result.tensorShape_ = tensorShape_;
+      } else {
+        result.tensorShape_ = tensorShapeBuilder_.build();
+      }
+      result.tensorSizeBytes_ = tensorSizeBytes_;
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.util.TensorMetadata) {
+        return mergeFrom((org.tensorflow.proto.util.TensorMetadata)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.util.TensorMetadata other) {
+      if (other == org.tensorflow.proto.util.TensorMetadata.getDefaultInstance()) return this;
+      if (other.hasTensorShape()) {
+        mergeTensorShape(other.getTensorShape());
+      }
+      if (other.getTensorSizeBytes() != 0L) {
+        setTensorSizeBytes(other.getTensorSizeBytes());
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.util.TensorMetadata parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.util.TensorMetadata) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+
+    private org.tensorflow.proto.framework.TensorShapeProto tensorShape_;
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.TensorShapeProto, org.tensorflow.proto.framework.TensorShapeProto.Builder, org.tensorflow.proto.framework.TensorShapeProtoOrBuilder> tensorShapeBuilder_;
+    /**
+     * <code>.tensorflow.TensorShapeProto tensor_shape = 2;</code>
+     */
+    public boolean hasTensorShape() {
+      return tensorShapeBuilder_ != null || tensorShape_ != null;
+    }
+    /**
+     * <code>.tensorflow.TensorShapeProto tensor_shape = 2;</code>
+     */
+    public org.tensorflow.proto.framework.TensorShapeProto getTensorShape() {
+      if (tensorShapeBuilder_ == null) {
+        return tensorShape_ == null ? org.tensorflow.proto.framework.TensorShapeProto.getDefaultInstance() : tensorShape_;
+      } else {
+        return tensorShapeBuilder_.getMessage();
+      }
+    }
+    /**
+     * <code>.tensorflow.TensorShapeProto tensor_shape = 2;</code>
+     */
+    public Builder setTensorShape(org.tensorflow.proto.framework.TensorShapeProto value) {
+      if (tensorShapeBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        tensorShape_ = value;
+        onChanged();
+      } else {
+        tensorShapeBuilder_.setMessage(value);
+      }
+
+      return this;
+    }
+    /**
+     * <code>.tensorflow.TensorShapeProto tensor_shape = 2;</code>
+     */
+    public Builder setTensorShape(
+        org.tensorflow.proto.framework.TensorShapeProto.Builder builderForValue) {
+      if (tensorShapeBuilder_ == null) {
+        tensorShape_ = builderForValue.build();
+        onChanged();
+      } else {
+        tensorShapeBuilder_.setMessage(builderForValue.build());
+      }
+
+      return this;
+    }
+    /**
+     * <code>.tensorflow.TensorShapeProto tensor_shape = 2;</code>
+     */
+    public Builder mergeTensorShape(org.tensorflow.proto.framework.TensorShapeProto value) {
+      if (tensorShapeBuilder_ == null) {
+        if (tensorShape_ != null) {
+          tensorShape_ =
+            org.tensorflow.proto.framework.TensorShapeProto.newBuilder(tensorShape_).mergeFrom(value).buildPartial();
+        } else {
+          tensorShape_ = value;
+        }
+        onChanged();
+      } else {
+        tensorShapeBuilder_.mergeFrom(value);
+      }
+
+      return this;
+    }
+    /**
+     * <code>.tensorflow.TensorShapeProto tensor_shape = 2;</code>
+     */
+    public Builder clearTensorShape() {
+      if (tensorShapeBuilder_ == null) {
+        tensorShape_ = null;
+        onChanged();
+      } else {
+        tensorShape_ = null;
+        tensorShapeBuilder_ = null;
+      }
+
+      return this;
+    }
+    /**
+     * <code>.tensorflow.TensorShapeProto tensor_shape = 2;</code>
+     */
+    public org.tensorflow.proto.framework.TensorShapeProto.Builder getTensorShapeBuilder() {
+      
+      onChanged();
+      return getTensorShapeFieldBuilder().getBuilder();
+    }
+    /**
+     * <code>.tensorflow.TensorShapeProto tensor_shape = 2;</code>
+     */
+    public org.tensorflow.proto.framework.TensorShapeProtoOrBuilder getTensorShapeOrBuilder() {
+      if (tensorShapeBuilder_ != null) {
+        return tensorShapeBuilder_.getMessageOrBuilder();
+      } else {
+        return tensorShape_ == null ?
+            org.tensorflow.proto.framework.TensorShapeProto.getDefaultInstance() : tensorShape_;
+      }
+    }
+    /**
+     * <code>.tensorflow.TensorShapeProto tensor_shape = 2;</code>
+     */
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.framework.TensorShapeProto, org.tensorflow.proto.framework.TensorShapeProto.Builder, org.tensorflow.proto.framework.TensorShapeProtoOrBuilder> 
+        getTensorShapeFieldBuilder() {
+      if (tensorShapeBuilder_ == null) {
+        tensorShapeBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+            org.tensorflow.proto.framework.TensorShapeProto, org.tensorflow.proto.framework.TensorShapeProto.Builder, org.tensorflow.proto.framework.TensorShapeProtoOrBuilder>(
+                getTensorShape(),
+                getParentForChildren(),
+                isClean());
+        tensorShape_ = null;
+      }
+      return tensorShapeBuilder_;
+    }
+
+    private long tensorSizeBytes_ ;
+    /**
+     * <pre>
+     * Number of uncompressed bytes used to store the tensor representation.
+     * </pre>
+     *
+     * <code>int64 tensor_size_bytes = 3;</code>
+     */
+    public long getTensorSizeBytes() {
+      return tensorSizeBytes_;
+    }
+    /**
+     * <pre>
+     * Number of uncompressed bytes used to store the tensor representation.
+     * </pre>
+     *
+     * <code>int64 tensor_size_bytes = 3;</code>
+     */
+    public Builder setTensorSizeBytes(long value) {
+      
+      tensorSizeBytes_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Number of uncompressed bytes used to store the tensor representation.
+     * </pre>
+     *
+     * <code>int64 tensor_size_bytes = 3;</code>
+     */
+    public Builder clearTensorSizeBytes() {
+      
+      tensorSizeBytes_ = 0L;
+      onChanged();
+      return this;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.data.experimental.TensorMetadata)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.data.experimental.TensorMetadata)
+  private static final org.tensorflow.proto.util.TensorMetadata DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.util.TensorMetadata();
+  }
+
+  public static org.tensorflow.proto.util.TensorMetadata getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<TensorMetadata>
+      PARSER = new com.google.protobuf.AbstractParser<TensorMetadata>() {
+    @java.lang.Override
+    public TensorMetadata parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new TensorMetadata(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<TensorMetadata> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<TensorMetadata> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.util.TensorMetadata getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/TensorMetadataOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/TensorMetadataOrBuilder.java
new file mode 100644
index 00000000000..0cfa470c9e6
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/TensorMetadataOrBuilder.java
@@ -0,0 +1,31 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/protobuf/data/experimental/snapshot.proto
+
+package org.tensorflow.proto.util;
+
+public interface TensorMetadataOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.data.experimental.TensorMetadata)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <code>.tensorflow.TensorShapeProto tensor_shape = 2;</code>
+   */
+  boolean hasTensorShape();
+  /**
+   * <code>.tensorflow.TensorShapeProto tensor_shape = 2;</code>
+   */
+  org.tensorflow.proto.framework.TensorShapeProto getTensorShape();
+  /**
+   * <code>.tensorflow.TensorShapeProto tensor_shape = 2;</code>
+   */
+  org.tensorflow.proto.framework.TensorShapeProtoOrBuilder getTensorShapeOrBuilder();
+
+  /**
+   * <pre>
+   * Number of uncompressed bytes used to store the tensor representation.
+   * </pre>
+   *
+   * <code>int64 tensor_size_bytes = 3;</code>
+   */
+  long getTensorSizeBytes();
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/WatchdogConfig.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/WatchdogConfig.java
new file mode 100644
index 00000000000..f818bbca13f
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/WatchdogConfig.java
@@ -0,0 +1,477 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/util/event.proto
+
+package org.tensorflow.proto.util;
+
+/**
+ * Protobuf type {@code tensorflow.WatchdogConfig}
+ */
+public  final class WatchdogConfig extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.WatchdogConfig)
+    WatchdogConfigOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use WatchdogConfig.newBuilder() to construct.
+  private WatchdogConfig(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private WatchdogConfig() {
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new WatchdogConfig();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private WatchdogConfig(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 8: {
+
+            timeoutMs_ = input.readInt64();
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.util.EventProtos.internal_static_tensorflow_WatchdogConfig_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.util.EventProtos.internal_static_tensorflow_WatchdogConfig_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.util.WatchdogConfig.class, org.tensorflow.proto.util.WatchdogConfig.Builder.class);
+  }
+
+  public static final int TIMEOUT_MS_FIELD_NUMBER = 1;
+  private long timeoutMs_;
+  /**
+   * <code>int64 timeout_ms = 1;</code>
+   */
+  public long getTimeoutMs() {
+    return timeoutMs_;
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    if (timeoutMs_ != 0L) {
+      output.writeInt64(1, timeoutMs_);
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (timeoutMs_ != 0L) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt64Size(1, timeoutMs_);
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.util.WatchdogConfig)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.util.WatchdogConfig other = (org.tensorflow.proto.util.WatchdogConfig) obj;
+
+    if (getTimeoutMs()
+        != other.getTimeoutMs()) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    hash = (37 * hash) + TIMEOUT_MS_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+        getTimeoutMs());
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.util.WatchdogConfig parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.WatchdogConfig parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.WatchdogConfig parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.WatchdogConfig parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.WatchdogConfig parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.WatchdogConfig parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.WatchdogConfig parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.WatchdogConfig parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.WatchdogConfig parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.WatchdogConfig parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.WatchdogConfig parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.WatchdogConfig parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.util.WatchdogConfig prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * Protobuf type {@code tensorflow.WatchdogConfig}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.WatchdogConfig)
+      org.tensorflow.proto.util.WatchdogConfigOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.util.EventProtos.internal_static_tensorflow_WatchdogConfig_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.util.EventProtos.internal_static_tensorflow_WatchdogConfig_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.util.WatchdogConfig.class, org.tensorflow.proto.util.WatchdogConfig.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.util.WatchdogConfig.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      timeoutMs_ = 0L;
+
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.util.EventProtos.internal_static_tensorflow_WatchdogConfig_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.WatchdogConfig getDefaultInstanceForType() {
+      return org.tensorflow.proto.util.WatchdogConfig.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.WatchdogConfig build() {
+      org.tensorflow.proto.util.WatchdogConfig result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.WatchdogConfig buildPartial() {
+      org.tensorflow.proto.util.WatchdogConfig result = new org.tensorflow.proto.util.WatchdogConfig(this);
+      result.timeoutMs_ = timeoutMs_;
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.util.WatchdogConfig) {
+        return mergeFrom((org.tensorflow.proto.util.WatchdogConfig)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.util.WatchdogConfig other) {
+      if (other == org.tensorflow.proto.util.WatchdogConfig.getDefaultInstance()) return this;
+      if (other.getTimeoutMs() != 0L) {
+        setTimeoutMs(other.getTimeoutMs());
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.util.WatchdogConfig parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.util.WatchdogConfig) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+
+    private long timeoutMs_ ;
+    /**
+     * <code>int64 timeout_ms = 1;</code>
+     */
+    public long getTimeoutMs() {
+      return timeoutMs_;
+    }
+    /**
+     * <code>int64 timeout_ms = 1;</code>
+     */
+    public Builder setTimeoutMs(long value) {
+      
+      timeoutMs_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>int64 timeout_ms = 1;</code>
+     */
+    public Builder clearTimeoutMs() {
+      
+      timeoutMs_ = 0L;
+      onChanged();
+      return this;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.WatchdogConfig)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.WatchdogConfig)
+  private static final org.tensorflow.proto.util.WatchdogConfig DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.util.WatchdogConfig();
+  }
+
+  public static org.tensorflow.proto.util.WatchdogConfig getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<WatchdogConfig>
+      PARSER = new com.google.protobuf.AbstractParser<WatchdogConfig>() {
+    @java.lang.Override
+    public WatchdogConfig parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new WatchdogConfig(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<WatchdogConfig> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<WatchdogConfig> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.util.WatchdogConfig getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/WatchdogConfigOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/WatchdogConfigOrBuilder.java
new file mode 100644
index 00000000000..74e6511d5ad
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/WatchdogConfigOrBuilder.java
@@ -0,0 +1,14 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/util/event.proto
+
+package org.tensorflow.proto.util;
+
+public interface WatchdogConfigOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.WatchdogConfig)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <code>int64 timeout_ms = 1;</code>
+   */
+  long getTimeoutMs();
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/WorkerHealth.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/WorkerHealth.java
new file mode 100644
index 00000000000..268ecd3a6a0
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/WorkerHealth.java
@@ -0,0 +1,141 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/util/event.proto
+
+package org.tensorflow.proto.util;
+
+/**
+ * <pre>
+ * Current health status of a worker.
+ * </pre>
+ *
+ * Protobuf enum {@code tensorflow.WorkerHealth}
+ */
+public enum WorkerHealth
+    implements com.google.protobuf.ProtocolMessageEnum {
+  /**
+   * <pre>
+   * By default a worker is healthy.
+   * </pre>
+   *
+   * <code>OK = 0;</code>
+   */
+  OK(0),
+  /**
+   * <code>RECEIVED_SHUTDOWN_SIGNAL = 1;</code>
+   */
+  RECEIVED_SHUTDOWN_SIGNAL(1),
+  /**
+   * <code>INTERNAL_ERROR = 2;</code>
+   */
+  INTERNAL_ERROR(2),
+  /**
+   * <pre>
+   * Worker has been instructed to shutdown after a timeout.
+   * </pre>
+   *
+   * <code>SHUTTING_DOWN = 3;</code>
+   */
+  SHUTTING_DOWN(3),
+  UNRECOGNIZED(-1),
+  ;
+
+  /**
+   * <pre>
+   * By default a worker is healthy.
+   * </pre>
+   *
+   * <code>OK = 0;</code>
+   */
+  public static final int OK_VALUE = 0;
+  /**
+   * <code>RECEIVED_SHUTDOWN_SIGNAL = 1;</code>
+   */
+  public static final int RECEIVED_SHUTDOWN_SIGNAL_VALUE = 1;
+  /**
+   * <code>INTERNAL_ERROR = 2;</code>
+   */
+  public static final int INTERNAL_ERROR_VALUE = 2;
+  /**
+   * <pre>
+   * Worker has been instructed to shutdown after a timeout.
+   * </pre>
+   *
+   * <code>SHUTTING_DOWN = 3;</code>
+   */
+  public static final int SHUTTING_DOWN_VALUE = 3;
+
+
+  public final int getNumber() {
+    if (this == UNRECOGNIZED) {
+      throw new java.lang.IllegalArgumentException(
+          "Can't get the number of an unknown enum value.");
+    }
+    return value;
+  }
+
+  /**
+   * @deprecated Use {@link #forNumber(int)} instead.
+   */
+  @java.lang.Deprecated
+  public static WorkerHealth valueOf(int value) {
+    return forNumber(value);
+  }
+
+  public static WorkerHealth forNumber(int value) {
+    switch (value) {
+      case 0: return OK;
+      case 1: return RECEIVED_SHUTDOWN_SIGNAL;
+      case 2: return INTERNAL_ERROR;
+      case 3: return SHUTTING_DOWN;
+      default: return null;
+    }
+  }
+
+  public static com.google.protobuf.Internal.EnumLiteMap<WorkerHealth>
+      internalGetValueMap() {
+    return internalValueMap;
+  }
+  private static final com.google.protobuf.Internal.EnumLiteMap<
+      WorkerHealth> internalValueMap =
+        new com.google.protobuf.Internal.EnumLiteMap<WorkerHealth>() {
+          public WorkerHealth findValueByNumber(int number) {
+            return WorkerHealth.forNumber(number);
+          }
+        };
+
+  public final com.google.protobuf.Descriptors.EnumValueDescriptor
+      getValueDescriptor() {
+    return getDescriptor().getValues().get(ordinal());
+  }
+  public final com.google.protobuf.Descriptors.EnumDescriptor
+      getDescriptorForType() {
+    return getDescriptor();
+  }
+  public static final com.google.protobuf.Descriptors.EnumDescriptor
+      getDescriptor() {
+    return org.tensorflow.proto.util.EventProtos.getDescriptor().getEnumTypes().get(0);
+  }
+
+  private static final WorkerHealth[] VALUES = values();
+
+  public static WorkerHealth valueOf(
+      com.google.protobuf.Descriptors.EnumValueDescriptor desc) {
+    if (desc.getType() != getDescriptor()) {
+      throw new java.lang.IllegalArgumentException(
+        "EnumValueDescriptor is not for this type.");
+    }
+    if (desc.getIndex() == -1) {
+      return UNRECOGNIZED;
+    }
+    return VALUES[desc.getIndex()];
+  }
+
+  private final int value;
+
+  private WorkerHealth(int value) {
+    this.value = value;
+  }
+
+  // @@protoc_insertion_point(enum_scope:tensorflow.WorkerHealth)
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/WorkerHeartbeatRequest.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/WorkerHeartbeatRequest.java
new file mode 100644
index 00000000000..9be331b3eca
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/WorkerHeartbeatRequest.java
@@ -0,0 +1,866 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/util/event.proto
+
+package org.tensorflow.proto.util;
+
+/**
+ * Protobuf type {@code tensorflow.WorkerHeartbeatRequest}
+ */
+public  final class WorkerHeartbeatRequest extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.WorkerHeartbeatRequest)
+    WorkerHeartbeatRequestOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use WorkerHeartbeatRequest.newBuilder() to construct.
+  private WorkerHeartbeatRequest(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private WorkerHeartbeatRequest() {
+    shutdownMode_ = 0;
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new WorkerHeartbeatRequest();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private WorkerHeartbeatRequest(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 8: {
+            int rawValue = input.readEnum();
+
+            shutdownMode_ = rawValue;
+            break;
+          }
+          case 18: {
+            org.tensorflow.proto.util.WatchdogConfig.Builder subBuilder = null;
+            if (watchdogConfig_ != null) {
+              subBuilder = watchdogConfig_.toBuilder();
+            }
+            watchdogConfig_ = input.readMessage(org.tensorflow.proto.util.WatchdogConfig.parser(), extensionRegistry);
+            if (subBuilder != null) {
+              subBuilder.mergeFrom(watchdogConfig_);
+              watchdogConfig_ = subBuilder.buildPartial();
+            }
+
+            break;
+          }
+          case 26: {
+            org.tensorflow.proto.util.RequestedExitCode.Builder subBuilder = null;
+            if (exitCode_ != null) {
+              subBuilder = exitCode_.toBuilder();
+            }
+            exitCode_ = input.readMessage(org.tensorflow.proto.util.RequestedExitCode.parser(), extensionRegistry);
+            if (subBuilder != null) {
+              subBuilder.mergeFrom(exitCode_);
+              exitCode_ = subBuilder.buildPartial();
+            }
+
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.util.EventProtos.internal_static_tensorflow_WorkerHeartbeatRequest_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.util.EventProtos.internal_static_tensorflow_WorkerHeartbeatRequest_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.util.WorkerHeartbeatRequest.class, org.tensorflow.proto.util.WorkerHeartbeatRequest.Builder.class);
+  }
+
+  public static final int SHUTDOWN_MODE_FIELD_NUMBER = 1;
+  private int shutdownMode_;
+  /**
+   * <code>.tensorflow.WorkerShutdownMode shutdown_mode = 1;</code>
+   */
+  public int getShutdownModeValue() {
+    return shutdownMode_;
+  }
+  /**
+   * <code>.tensorflow.WorkerShutdownMode shutdown_mode = 1;</code>
+   */
+  public org.tensorflow.proto.util.WorkerShutdownMode getShutdownMode() {
+    @SuppressWarnings("deprecation")
+    org.tensorflow.proto.util.WorkerShutdownMode result = org.tensorflow.proto.util.WorkerShutdownMode.valueOf(shutdownMode_);
+    return result == null ? org.tensorflow.proto.util.WorkerShutdownMode.UNRECOGNIZED : result;
+  }
+
+  public static final int WATCHDOG_CONFIG_FIELD_NUMBER = 2;
+  private org.tensorflow.proto.util.WatchdogConfig watchdogConfig_;
+  /**
+   * <code>.tensorflow.WatchdogConfig watchdog_config = 2;</code>
+   */
+  public boolean hasWatchdogConfig() {
+    return watchdogConfig_ != null;
+  }
+  /**
+   * <code>.tensorflow.WatchdogConfig watchdog_config = 2;</code>
+   */
+  public org.tensorflow.proto.util.WatchdogConfig getWatchdogConfig() {
+    return watchdogConfig_ == null ? org.tensorflow.proto.util.WatchdogConfig.getDefaultInstance() : watchdogConfig_;
+  }
+  /**
+   * <code>.tensorflow.WatchdogConfig watchdog_config = 2;</code>
+   */
+  public org.tensorflow.proto.util.WatchdogConfigOrBuilder getWatchdogConfigOrBuilder() {
+    return getWatchdogConfig();
+  }
+
+  public static final int EXIT_CODE_FIELD_NUMBER = 3;
+  private org.tensorflow.proto.util.RequestedExitCode exitCode_;
+  /**
+   * <code>.tensorflow.RequestedExitCode exit_code = 3;</code>
+   */
+  public boolean hasExitCode() {
+    return exitCode_ != null;
+  }
+  /**
+   * <code>.tensorflow.RequestedExitCode exit_code = 3;</code>
+   */
+  public org.tensorflow.proto.util.RequestedExitCode getExitCode() {
+    return exitCode_ == null ? org.tensorflow.proto.util.RequestedExitCode.getDefaultInstance() : exitCode_;
+  }
+  /**
+   * <code>.tensorflow.RequestedExitCode exit_code = 3;</code>
+   */
+  public org.tensorflow.proto.util.RequestedExitCodeOrBuilder getExitCodeOrBuilder() {
+    return getExitCode();
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    if (shutdownMode_ != org.tensorflow.proto.util.WorkerShutdownMode.DEFAULT.getNumber()) {
+      output.writeEnum(1, shutdownMode_);
+    }
+    if (watchdogConfig_ != null) {
+      output.writeMessage(2, getWatchdogConfig());
+    }
+    if (exitCode_ != null) {
+      output.writeMessage(3, getExitCode());
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (shutdownMode_ != org.tensorflow.proto.util.WorkerShutdownMode.DEFAULT.getNumber()) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeEnumSize(1, shutdownMode_);
+    }
+    if (watchdogConfig_ != null) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(2, getWatchdogConfig());
+    }
+    if (exitCode_ != null) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(3, getExitCode());
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.util.WorkerHeartbeatRequest)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.util.WorkerHeartbeatRequest other = (org.tensorflow.proto.util.WorkerHeartbeatRequest) obj;
+
+    if (shutdownMode_ != other.shutdownMode_) return false;
+    if (hasWatchdogConfig() != other.hasWatchdogConfig()) return false;
+    if (hasWatchdogConfig()) {
+      if (!getWatchdogConfig()
+          .equals(other.getWatchdogConfig())) return false;
+    }
+    if (hasExitCode() != other.hasExitCode()) return false;
+    if (hasExitCode()) {
+      if (!getExitCode()
+          .equals(other.getExitCode())) return false;
+    }
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    hash = (37 * hash) + SHUTDOWN_MODE_FIELD_NUMBER;
+    hash = (53 * hash) + shutdownMode_;
+    if (hasWatchdogConfig()) {
+      hash = (37 * hash) + WATCHDOG_CONFIG_FIELD_NUMBER;
+      hash = (53 * hash) + getWatchdogConfig().hashCode();
+    }
+    if (hasExitCode()) {
+      hash = (37 * hash) + EXIT_CODE_FIELD_NUMBER;
+      hash = (53 * hash) + getExitCode().hashCode();
+    }
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.util.WorkerHeartbeatRequest parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.WorkerHeartbeatRequest parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.WorkerHeartbeatRequest parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.WorkerHeartbeatRequest parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.WorkerHeartbeatRequest parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.WorkerHeartbeatRequest parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.WorkerHeartbeatRequest parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.WorkerHeartbeatRequest parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.WorkerHeartbeatRequest parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.WorkerHeartbeatRequest parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.WorkerHeartbeatRequest parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.WorkerHeartbeatRequest parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.util.WorkerHeartbeatRequest prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * Protobuf type {@code tensorflow.WorkerHeartbeatRequest}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.WorkerHeartbeatRequest)
+      org.tensorflow.proto.util.WorkerHeartbeatRequestOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.util.EventProtos.internal_static_tensorflow_WorkerHeartbeatRequest_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.util.EventProtos.internal_static_tensorflow_WorkerHeartbeatRequest_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.util.WorkerHeartbeatRequest.class, org.tensorflow.proto.util.WorkerHeartbeatRequest.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.util.WorkerHeartbeatRequest.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      shutdownMode_ = 0;
+
+      if (watchdogConfigBuilder_ == null) {
+        watchdogConfig_ = null;
+      } else {
+        watchdogConfig_ = null;
+        watchdogConfigBuilder_ = null;
+      }
+      if (exitCodeBuilder_ == null) {
+        exitCode_ = null;
+      } else {
+        exitCode_ = null;
+        exitCodeBuilder_ = null;
+      }
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.util.EventProtos.internal_static_tensorflow_WorkerHeartbeatRequest_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.WorkerHeartbeatRequest getDefaultInstanceForType() {
+      return org.tensorflow.proto.util.WorkerHeartbeatRequest.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.WorkerHeartbeatRequest build() {
+      org.tensorflow.proto.util.WorkerHeartbeatRequest result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.WorkerHeartbeatRequest buildPartial() {
+      org.tensorflow.proto.util.WorkerHeartbeatRequest result = new org.tensorflow.proto.util.WorkerHeartbeatRequest(this);
+      result.shutdownMode_ = shutdownMode_;
+      if (watchdogConfigBuilder_ == null) {
+        result.watchdogConfig_ = watchdogConfig_;
+      } else {
+        result.watchdogConfig_ = watchdogConfigBuilder_.build();
+      }
+      if (exitCodeBuilder_ == null) {
+        result.exitCode_ = exitCode_;
+      } else {
+        result.exitCode_ = exitCodeBuilder_.build();
+      }
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.util.WorkerHeartbeatRequest) {
+        return mergeFrom((org.tensorflow.proto.util.WorkerHeartbeatRequest)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.util.WorkerHeartbeatRequest other) {
+      if (other == org.tensorflow.proto.util.WorkerHeartbeatRequest.getDefaultInstance()) return this;
+      if (other.shutdownMode_ != 0) {
+        setShutdownModeValue(other.getShutdownModeValue());
+      }
+      if (other.hasWatchdogConfig()) {
+        mergeWatchdogConfig(other.getWatchdogConfig());
+      }
+      if (other.hasExitCode()) {
+        mergeExitCode(other.getExitCode());
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.util.WorkerHeartbeatRequest parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.util.WorkerHeartbeatRequest) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+
+    private int shutdownMode_ = 0;
+    /**
+     * <code>.tensorflow.WorkerShutdownMode shutdown_mode = 1;</code>
+     */
+    public int getShutdownModeValue() {
+      return shutdownMode_;
+    }
+    /**
+     * <code>.tensorflow.WorkerShutdownMode shutdown_mode = 1;</code>
+     */
+    public Builder setShutdownModeValue(int value) {
+      shutdownMode_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>.tensorflow.WorkerShutdownMode shutdown_mode = 1;</code>
+     */
+    public org.tensorflow.proto.util.WorkerShutdownMode getShutdownMode() {
+      @SuppressWarnings("deprecation")
+      org.tensorflow.proto.util.WorkerShutdownMode result = org.tensorflow.proto.util.WorkerShutdownMode.valueOf(shutdownMode_);
+      return result == null ? org.tensorflow.proto.util.WorkerShutdownMode.UNRECOGNIZED : result;
+    }
+    /**
+     * <code>.tensorflow.WorkerShutdownMode shutdown_mode = 1;</code>
+     */
+    public Builder setShutdownMode(org.tensorflow.proto.util.WorkerShutdownMode value) {
+      if (value == null) {
+        throw new NullPointerException();
+      }
+      
+      shutdownMode_ = value.getNumber();
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>.tensorflow.WorkerShutdownMode shutdown_mode = 1;</code>
+     */
+    public Builder clearShutdownMode() {
+      
+      shutdownMode_ = 0;
+      onChanged();
+      return this;
+    }
+
+    private org.tensorflow.proto.util.WatchdogConfig watchdogConfig_;
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.util.WatchdogConfig, org.tensorflow.proto.util.WatchdogConfig.Builder, org.tensorflow.proto.util.WatchdogConfigOrBuilder> watchdogConfigBuilder_;
+    /**
+     * <code>.tensorflow.WatchdogConfig watchdog_config = 2;</code>
+     */
+    public boolean hasWatchdogConfig() {
+      return watchdogConfigBuilder_ != null || watchdogConfig_ != null;
+    }
+    /**
+     * <code>.tensorflow.WatchdogConfig watchdog_config = 2;</code>
+     */
+    public org.tensorflow.proto.util.WatchdogConfig getWatchdogConfig() {
+      if (watchdogConfigBuilder_ == null) {
+        return watchdogConfig_ == null ? org.tensorflow.proto.util.WatchdogConfig.getDefaultInstance() : watchdogConfig_;
+      } else {
+        return watchdogConfigBuilder_.getMessage();
+      }
+    }
+    /**
+     * <code>.tensorflow.WatchdogConfig watchdog_config = 2;</code>
+     */
+    public Builder setWatchdogConfig(org.tensorflow.proto.util.WatchdogConfig value) {
+      if (watchdogConfigBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        watchdogConfig_ = value;
+        onChanged();
+      } else {
+        watchdogConfigBuilder_.setMessage(value);
+      }
+
+      return this;
+    }
+    /**
+     * <code>.tensorflow.WatchdogConfig watchdog_config = 2;</code>
+     */
+    public Builder setWatchdogConfig(
+        org.tensorflow.proto.util.WatchdogConfig.Builder builderForValue) {
+      if (watchdogConfigBuilder_ == null) {
+        watchdogConfig_ = builderForValue.build();
+        onChanged();
+      } else {
+        watchdogConfigBuilder_.setMessage(builderForValue.build());
+      }
+
+      return this;
+    }
+    /**
+     * <code>.tensorflow.WatchdogConfig watchdog_config = 2;</code>
+     */
+    public Builder mergeWatchdogConfig(org.tensorflow.proto.util.WatchdogConfig value) {
+      if (watchdogConfigBuilder_ == null) {
+        if (watchdogConfig_ != null) {
+          watchdogConfig_ =
+            org.tensorflow.proto.util.WatchdogConfig.newBuilder(watchdogConfig_).mergeFrom(value).buildPartial();
+        } else {
+          watchdogConfig_ = value;
+        }
+        onChanged();
+      } else {
+        watchdogConfigBuilder_.mergeFrom(value);
+      }
+
+      return this;
+    }
+    /**
+     * <code>.tensorflow.WatchdogConfig watchdog_config = 2;</code>
+     */
+    public Builder clearWatchdogConfig() {
+      if (watchdogConfigBuilder_ == null) {
+        watchdogConfig_ = null;
+        onChanged();
+      } else {
+        watchdogConfig_ = null;
+        watchdogConfigBuilder_ = null;
+      }
+
+      return this;
+    }
+    /**
+     * <code>.tensorflow.WatchdogConfig watchdog_config = 2;</code>
+     */
+    public org.tensorflow.proto.util.WatchdogConfig.Builder getWatchdogConfigBuilder() {
+      
+      onChanged();
+      return getWatchdogConfigFieldBuilder().getBuilder();
+    }
+    /**
+     * <code>.tensorflow.WatchdogConfig watchdog_config = 2;</code>
+     */
+    public org.tensorflow.proto.util.WatchdogConfigOrBuilder getWatchdogConfigOrBuilder() {
+      if (watchdogConfigBuilder_ != null) {
+        return watchdogConfigBuilder_.getMessageOrBuilder();
+      } else {
+        return watchdogConfig_ == null ?
+            org.tensorflow.proto.util.WatchdogConfig.getDefaultInstance() : watchdogConfig_;
+      }
+    }
+    /**
+     * <code>.tensorflow.WatchdogConfig watchdog_config = 2;</code>
+     */
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.util.WatchdogConfig, org.tensorflow.proto.util.WatchdogConfig.Builder, org.tensorflow.proto.util.WatchdogConfigOrBuilder> 
+        getWatchdogConfigFieldBuilder() {
+      if (watchdogConfigBuilder_ == null) {
+        watchdogConfigBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+            org.tensorflow.proto.util.WatchdogConfig, org.tensorflow.proto.util.WatchdogConfig.Builder, org.tensorflow.proto.util.WatchdogConfigOrBuilder>(
+                getWatchdogConfig(),
+                getParentForChildren(),
+                isClean());
+        watchdogConfig_ = null;
+      }
+      return watchdogConfigBuilder_;
+    }
+
+    private org.tensorflow.proto.util.RequestedExitCode exitCode_;
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.util.RequestedExitCode, org.tensorflow.proto.util.RequestedExitCode.Builder, org.tensorflow.proto.util.RequestedExitCodeOrBuilder> exitCodeBuilder_;
+    /**
+     * <code>.tensorflow.RequestedExitCode exit_code = 3;</code>
+     */
+    public boolean hasExitCode() {
+      return exitCodeBuilder_ != null || exitCode_ != null;
+    }
+    /**
+     * <code>.tensorflow.RequestedExitCode exit_code = 3;</code>
+     */
+    public org.tensorflow.proto.util.RequestedExitCode getExitCode() {
+      if (exitCodeBuilder_ == null) {
+        return exitCode_ == null ? org.tensorflow.proto.util.RequestedExitCode.getDefaultInstance() : exitCode_;
+      } else {
+        return exitCodeBuilder_.getMessage();
+      }
+    }
+    /**
+     * <code>.tensorflow.RequestedExitCode exit_code = 3;</code>
+     */
+    public Builder setExitCode(org.tensorflow.proto.util.RequestedExitCode value) {
+      if (exitCodeBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        exitCode_ = value;
+        onChanged();
+      } else {
+        exitCodeBuilder_.setMessage(value);
+      }
+
+      return this;
+    }
+    /**
+     * <code>.tensorflow.RequestedExitCode exit_code = 3;</code>
+     */
+    public Builder setExitCode(
+        org.tensorflow.proto.util.RequestedExitCode.Builder builderForValue) {
+      if (exitCodeBuilder_ == null) {
+        exitCode_ = builderForValue.build();
+        onChanged();
+      } else {
+        exitCodeBuilder_.setMessage(builderForValue.build());
+      }
+
+      return this;
+    }
+    /**
+     * <code>.tensorflow.RequestedExitCode exit_code = 3;</code>
+     */
+    public Builder mergeExitCode(org.tensorflow.proto.util.RequestedExitCode value) {
+      if (exitCodeBuilder_ == null) {
+        if (exitCode_ != null) {
+          exitCode_ =
+            org.tensorflow.proto.util.RequestedExitCode.newBuilder(exitCode_).mergeFrom(value).buildPartial();
+        } else {
+          exitCode_ = value;
+        }
+        onChanged();
+      } else {
+        exitCodeBuilder_.mergeFrom(value);
+      }
+
+      return this;
+    }
+    /**
+     * <code>.tensorflow.RequestedExitCode exit_code = 3;</code>
+     */
+    public Builder clearExitCode() {
+      if (exitCodeBuilder_ == null) {
+        exitCode_ = null;
+        onChanged();
+      } else {
+        exitCode_ = null;
+        exitCodeBuilder_ = null;
+      }
+
+      return this;
+    }
+    /**
+     * <code>.tensorflow.RequestedExitCode exit_code = 3;</code>
+     */
+    public org.tensorflow.proto.util.RequestedExitCode.Builder getExitCodeBuilder() {
+      
+      onChanged();
+      return getExitCodeFieldBuilder().getBuilder();
+    }
+    /**
+     * <code>.tensorflow.RequestedExitCode exit_code = 3;</code>
+     */
+    public org.tensorflow.proto.util.RequestedExitCodeOrBuilder getExitCodeOrBuilder() {
+      if (exitCodeBuilder_ != null) {
+        return exitCodeBuilder_.getMessageOrBuilder();
+      } else {
+        return exitCode_ == null ?
+            org.tensorflow.proto.util.RequestedExitCode.getDefaultInstance() : exitCode_;
+      }
+    }
+    /**
+     * <code>.tensorflow.RequestedExitCode exit_code = 3;</code>
+     */
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.util.RequestedExitCode, org.tensorflow.proto.util.RequestedExitCode.Builder, org.tensorflow.proto.util.RequestedExitCodeOrBuilder> 
+        getExitCodeFieldBuilder() {
+      if (exitCodeBuilder_ == null) {
+        exitCodeBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+            org.tensorflow.proto.util.RequestedExitCode, org.tensorflow.proto.util.RequestedExitCode.Builder, org.tensorflow.proto.util.RequestedExitCodeOrBuilder>(
+                getExitCode(),
+                getParentForChildren(),
+                isClean());
+        exitCode_ = null;
+      }
+      return exitCodeBuilder_;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.WorkerHeartbeatRequest)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.WorkerHeartbeatRequest)
+  private static final org.tensorflow.proto.util.WorkerHeartbeatRequest DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.util.WorkerHeartbeatRequest();
+  }
+
+  public static org.tensorflow.proto.util.WorkerHeartbeatRequest getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<WorkerHeartbeatRequest>
+      PARSER = new com.google.protobuf.AbstractParser<WorkerHeartbeatRequest>() {
+    @java.lang.Override
+    public WorkerHeartbeatRequest parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new WorkerHeartbeatRequest(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<WorkerHeartbeatRequest> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<WorkerHeartbeatRequest> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.util.WorkerHeartbeatRequest getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/WorkerHeartbeatRequestOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/WorkerHeartbeatRequestOrBuilder.java
new file mode 100644
index 00000000000..28e78122ba0
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/WorkerHeartbeatRequestOrBuilder.java
@@ -0,0 +1,44 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/util/event.proto
+
+package org.tensorflow.proto.util;
+
+public interface WorkerHeartbeatRequestOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.WorkerHeartbeatRequest)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <code>.tensorflow.WorkerShutdownMode shutdown_mode = 1;</code>
+   */
+  int getShutdownModeValue();
+  /**
+   * <code>.tensorflow.WorkerShutdownMode shutdown_mode = 1;</code>
+   */
+  org.tensorflow.proto.util.WorkerShutdownMode getShutdownMode();
+
+  /**
+   * <code>.tensorflow.WatchdogConfig watchdog_config = 2;</code>
+   */
+  boolean hasWatchdogConfig();
+  /**
+   * <code>.tensorflow.WatchdogConfig watchdog_config = 2;</code>
+   */
+  org.tensorflow.proto.util.WatchdogConfig getWatchdogConfig();
+  /**
+   * <code>.tensorflow.WatchdogConfig watchdog_config = 2;</code>
+   */
+  org.tensorflow.proto.util.WatchdogConfigOrBuilder getWatchdogConfigOrBuilder();
+
+  /**
+   * <code>.tensorflow.RequestedExitCode exit_code = 3;</code>
+   */
+  boolean hasExitCode();
+  /**
+   * <code>.tensorflow.RequestedExitCode exit_code = 3;</code>
+   */
+  org.tensorflow.proto.util.RequestedExitCode getExitCode();
+  /**
+   * <code>.tensorflow.RequestedExitCode exit_code = 3;</code>
+   */
+  org.tensorflow.proto.util.RequestedExitCodeOrBuilder getExitCodeOrBuilder();
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/WorkerHeartbeatResponse.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/WorkerHeartbeatResponse.java
new file mode 100644
index 00000000000..87262dd0405
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/WorkerHeartbeatResponse.java
@@ -0,0 +1,977 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/util/event.proto
+
+package org.tensorflow.proto.util;
+
+/**
+ * Protobuf type {@code tensorflow.WorkerHeartbeatResponse}
+ */
+public  final class WorkerHeartbeatResponse extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.WorkerHeartbeatResponse)
+    WorkerHeartbeatResponseOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use WorkerHeartbeatResponse.newBuilder() to construct.
+  private WorkerHeartbeatResponse(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private WorkerHeartbeatResponse() {
+    healthStatus_ = 0;
+    workerLog_ = java.util.Collections.emptyList();
+    hostname_ = "";
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new WorkerHeartbeatResponse();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private WorkerHeartbeatResponse(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    int mutable_bitField0_ = 0;
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 8: {
+            int rawValue = input.readEnum();
+
+            healthStatus_ = rawValue;
+            break;
+          }
+          case 18: {
+            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
+              workerLog_ = new java.util.ArrayList<org.tensorflow.proto.util.Event>();
+              mutable_bitField0_ |= 0x00000001;
+            }
+            workerLog_.add(
+                input.readMessage(org.tensorflow.proto.util.Event.parser(), extensionRegistry));
+            break;
+          }
+          case 26: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            hostname_ = s;
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      if (((mutable_bitField0_ & 0x00000001) != 0)) {
+        workerLog_ = java.util.Collections.unmodifiableList(workerLog_);
+      }
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.util.EventProtos.internal_static_tensorflow_WorkerHeartbeatResponse_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.util.EventProtos.internal_static_tensorflow_WorkerHeartbeatResponse_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.util.WorkerHeartbeatResponse.class, org.tensorflow.proto.util.WorkerHeartbeatResponse.Builder.class);
+  }
+
+  public static final int HEALTH_STATUS_FIELD_NUMBER = 1;
+  private int healthStatus_;
+  /**
+   * <code>.tensorflow.WorkerHealth health_status = 1;</code>
+   */
+  public int getHealthStatusValue() {
+    return healthStatus_;
+  }
+  /**
+   * <code>.tensorflow.WorkerHealth health_status = 1;</code>
+   */
+  public org.tensorflow.proto.util.WorkerHealth getHealthStatus() {
+    @SuppressWarnings("deprecation")
+    org.tensorflow.proto.util.WorkerHealth result = org.tensorflow.proto.util.WorkerHealth.valueOf(healthStatus_);
+    return result == null ? org.tensorflow.proto.util.WorkerHealth.UNRECOGNIZED : result;
+  }
+
+  public static final int WORKER_LOG_FIELD_NUMBER = 2;
+  private java.util.List<org.tensorflow.proto.util.Event> workerLog_;
+  /**
+   * <code>repeated .tensorflow.Event worker_log = 2;</code>
+   */
+  public java.util.List<org.tensorflow.proto.util.Event> getWorkerLogList() {
+    return workerLog_;
+  }
+  /**
+   * <code>repeated .tensorflow.Event worker_log = 2;</code>
+   */
+  public java.util.List<? extends org.tensorflow.proto.util.EventOrBuilder> 
+      getWorkerLogOrBuilderList() {
+    return workerLog_;
+  }
+  /**
+   * <code>repeated .tensorflow.Event worker_log = 2;</code>
+   */
+  public int getWorkerLogCount() {
+    return workerLog_.size();
+  }
+  /**
+   * <code>repeated .tensorflow.Event worker_log = 2;</code>
+   */
+  public org.tensorflow.proto.util.Event getWorkerLog(int index) {
+    return workerLog_.get(index);
+  }
+  /**
+   * <code>repeated .tensorflow.Event worker_log = 2;</code>
+   */
+  public org.tensorflow.proto.util.EventOrBuilder getWorkerLogOrBuilder(
+      int index) {
+    return workerLog_.get(index);
+  }
+
+  public static final int HOSTNAME_FIELD_NUMBER = 3;
+  private volatile java.lang.Object hostname_;
+  /**
+   * <code>string hostname = 3;</code>
+   */
+  public java.lang.String getHostname() {
+    java.lang.Object ref = hostname_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      hostname_ = s;
+      return s;
+    }
+  }
+  /**
+   * <code>string hostname = 3;</code>
+   */
+  public com.google.protobuf.ByteString
+      getHostnameBytes() {
+    java.lang.Object ref = hostname_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      hostname_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    if (healthStatus_ != org.tensorflow.proto.util.WorkerHealth.OK.getNumber()) {
+      output.writeEnum(1, healthStatus_);
+    }
+    for (int i = 0; i < workerLog_.size(); i++) {
+      output.writeMessage(2, workerLog_.get(i));
+    }
+    if (!getHostnameBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 3, hostname_);
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (healthStatus_ != org.tensorflow.proto.util.WorkerHealth.OK.getNumber()) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeEnumSize(1, healthStatus_);
+    }
+    for (int i = 0; i < workerLog_.size(); i++) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(2, workerLog_.get(i));
+    }
+    if (!getHostnameBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(3, hostname_);
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.util.WorkerHeartbeatResponse)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.util.WorkerHeartbeatResponse other = (org.tensorflow.proto.util.WorkerHeartbeatResponse) obj;
+
+    if (healthStatus_ != other.healthStatus_) return false;
+    if (!getWorkerLogList()
+        .equals(other.getWorkerLogList())) return false;
+    if (!getHostname()
+        .equals(other.getHostname())) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    hash = (37 * hash) + HEALTH_STATUS_FIELD_NUMBER;
+    hash = (53 * hash) + healthStatus_;
+    if (getWorkerLogCount() > 0) {
+      hash = (37 * hash) + WORKER_LOG_FIELD_NUMBER;
+      hash = (53 * hash) + getWorkerLogList().hashCode();
+    }
+    hash = (37 * hash) + HOSTNAME_FIELD_NUMBER;
+    hash = (53 * hash) + getHostname().hashCode();
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.util.WorkerHeartbeatResponse parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.WorkerHeartbeatResponse parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.WorkerHeartbeatResponse parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.WorkerHeartbeatResponse parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.WorkerHeartbeatResponse parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.WorkerHeartbeatResponse parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.WorkerHeartbeatResponse parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.WorkerHeartbeatResponse parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.WorkerHeartbeatResponse parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.WorkerHeartbeatResponse parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.WorkerHeartbeatResponse parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.WorkerHeartbeatResponse parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.util.WorkerHeartbeatResponse prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * Protobuf type {@code tensorflow.WorkerHeartbeatResponse}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.WorkerHeartbeatResponse)
+      org.tensorflow.proto.util.WorkerHeartbeatResponseOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.util.EventProtos.internal_static_tensorflow_WorkerHeartbeatResponse_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.util.EventProtos.internal_static_tensorflow_WorkerHeartbeatResponse_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.util.WorkerHeartbeatResponse.class, org.tensorflow.proto.util.WorkerHeartbeatResponse.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.util.WorkerHeartbeatResponse.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+        getWorkerLogFieldBuilder();
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      healthStatus_ = 0;
+
+      if (workerLogBuilder_ == null) {
+        workerLog_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000001);
+      } else {
+        workerLogBuilder_.clear();
+      }
+      hostname_ = "";
+
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.util.EventProtos.internal_static_tensorflow_WorkerHeartbeatResponse_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.WorkerHeartbeatResponse getDefaultInstanceForType() {
+      return org.tensorflow.proto.util.WorkerHeartbeatResponse.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.WorkerHeartbeatResponse build() {
+      org.tensorflow.proto.util.WorkerHeartbeatResponse result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.WorkerHeartbeatResponse buildPartial() {
+      org.tensorflow.proto.util.WorkerHeartbeatResponse result = new org.tensorflow.proto.util.WorkerHeartbeatResponse(this);
+      int from_bitField0_ = bitField0_;
+      result.healthStatus_ = healthStatus_;
+      if (workerLogBuilder_ == null) {
+        if (((bitField0_ & 0x00000001) != 0)) {
+          workerLog_ = java.util.Collections.unmodifiableList(workerLog_);
+          bitField0_ = (bitField0_ & ~0x00000001);
+        }
+        result.workerLog_ = workerLog_;
+      } else {
+        result.workerLog_ = workerLogBuilder_.build();
+      }
+      result.hostname_ = hostname_;
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.util.WorkerHeartbeatResponse) {
+        return mergeFrom((org.tensorflow.proto.util.WorkerHeartbeatResponse)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.util.WorkerHeartbeatResponse other) {
+      if (other == org.tensorflow.proto.util.WorkerHeartbeatResponse.getDefaultInstance()) return this;
+      if (other.healthStatus_ != 0) {
+        setHealthStatusValue(other.getHealthStatusValue());
+      }
+      if (workerLogBuilder_ == null) {
+        if (!other.workerLog_.isEmpty()) {
+          if (workerLog_.isEmpty()) {
+            workerLog_ = other.workerLog_;
+            bitField0_ = (bitField0_ & ~0x00000001);
+          } else {
+            ensureWorkerLogIsMutable();
+            workerLog_.addAll(other.workerLog_);
+          }
+          onChanged();
+        }
+      } else {
+        if (!other.workerLog_.isEmpty()) {
+          if (workerLogBuilder_.isEmpty()) {
+            workerLogBuilder_.dispose();
+            workerLogBuilder_ = null;
+            workerLog_ = other.workerLog_;
+            bitField0_ = (bitField0_ & ~0x00000001);
+            workerLogBuilder_ = 
+              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
+                 getWorkerLogFieldBuilder() : null;
+          } else {
+            workerLogBuilder_.addAllMessages(other.workerLog_);
+          }
+        }
+      }
+      if (!other.getHostname().isEmpty()) {
+        hostname_ = other.hostname_;
+        onChanged();
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.util.WorkerHeartbeatResponse parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.util.WorkerHeartbeatResponse) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+    private int bitField0_;
+
+    private int healthStatus_ = 0;
+    /**
+     * <code>.tensorflow.WorkerHealth health_status = 1;</code>
+     */
+    public int getHealthStatusValue() {
+      return healthStatus_;
+    }
+    /**
+     * <code>.tensorflow.WorkerHealth health_status = 1;</code>
+     */
+    public Builder setHealthStatusValue(int value) {
+      healthStatus_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>.tensorflow.WorkerHealth health_status = 1;</code>
+     */
+    public org.tensorflow.proto.util.WorkerHealth getHealthStatus() {
+      @SuppressWarnings("deprecation")
+      org.tensorflow.proto.util.WorkerHealth result = org.tensorflow.proto.util.WorkerHealth.valueOf(healthStatus_);
+      return result == null ? org.tensorflow.proto.util.WorkerHealth.UNRECOGNIZED : result;
+    }
+    /**
+     * <code>.tensorflow.WorkerHealth health_status = 1;</code>
+     */
+    public Builder setHealthStatus(org.tensorflow.proto.util.WorkerHealth value) {
+      if (value == null) {
+        throw new NullPointerException();
+      }
+      
+      healthStatus_ = value.getNumber();
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>.tensorflow.WorkerHealth health_status = 1;</code>
+     */
+    public Builder clearHealthStatus() {
+      
+      healthStatus_ = 0;
+      onChanged();
+      return this;
+    }
+
+    private java.util.List<org.tensorflow.proto.util.Event> workerLog_ =
+      java.util.Collections.emptyList();
+    private void ensureWorkerLogIsMutable() {
+      if (!((bitField0_ & 0x00000001) != 0)) {
+        workerLog_ = new java.util.ArrayList<org.tensorflow.proto.util.Event>(workerLog_);
+        bitField0_ |= 0x00000001;
+       }
+    }
+
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.util.Event, org.tensorflow.proto.util.Event.Builder, org.tensorflow.proto.util.EventOrBuilder> workerLogBuilder_;
+
+    /**
+     * <code>repeated .tensorflow.Event worker_log = 2;</code>
+     */
+    public java.util.List<org.tensorflow.proto.util.Event> getWorkerLogList() {
+      if (workerLogBuilder_ == null) {
+        return java.util.Collections.unmodifiableList(workerLog_);
+      } else {
+        return workerLogBuilder_.getMessageList();
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.Event worker_log = 2;</code>
+     */
+    public int getWorkerLogCount() {
+      if (workerLogBuilder_ == null) {
+        return workerLog_.size();
+      } else {
+        return workerLogBuilder_.getCount();
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.Event worker_log = 2;</code>
+     */
+    public org.tensorflow.proto.util.Event getWorkerLog(int index) {
+      if (workerLogBuilder_ == null) {
+        return workerLog_.get(index);
+      } else {
+        return workerLogBuilder_.getMessage(index);
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.Event worker_log = 2;</code>
+     */
+    public Builder setWorkerLog(
+        int index, org.tensorflow.proto.util.Event value) {
+      if (workerLogBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureWorkerLogIsMutable();
+        workerLog_.set(index, value);
+        onChanged();
+      } else {
+        workerLogBuilder_.setMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.Event worker_log = 2;</code>
+     */
+    public Builder setWorkerLog(
+        int index, org.tensorflow.proto.util.Event.Builder builderForValue) {
+      if (workerLogBuilder_ == null) {
+        ensureWorkerLogIsMutable();
+        workerLog_.set(index, builderForValue.build());
+        onChanged();
+      } else {
+        workerLogBuilder_.setMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.Event worker_log = 2;</code>
+     */
+    public Builder addWorkerLog(org.tensorflow.proto.util.Event value) {
+      if (workerLogBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureWorkerLogIsMutable();
+        workerLog_.add(value);
+        onChanged();
+      } else {
+        workerLogBuilder_.addMessage(value);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.Event worker_log = 2;</code>
+     */
+    public Builder addWorkerLog(
+        int index, org.tensorflow.proto.util.Event value) {
+      if (workerLogBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureWorkerLogIsMutable();
+        workerLog_.add(index, value);
+        onChanged();
+      } else {
+        workerLogBuilder_.addMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.Event worker_log = 2;</code>
+     */
+    public Builder addWorkerLog(
+        org.tensorflow.proto.util.Event.Builder builderForValue) {
+      if (workerLogBuilder_ == null) {
+        ensureWorkerLogIsMutable();
+        workerLog_.add(builderForValue.build());
+        onChanged();
+      } else {
+        workerLogBuilder_.addMessage(builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.Event worker_log = 2;</code>
+     */
+    public Builder addWorkerLog(
+        int index, org.tensorflow.proto.util.Event.Builder builderForValue) {
+      if (workerLogBuilder_ == null) {
+        ensureWorkerLogIsMutable();
+        workerLog_.add(index, builderForValue.build());
+        onChanged();
+      } else {
+        workerLogBuilder_.addMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.Event worker_log = 2;</code>
+     */
+    public Builder addAllWorkerLog(
+        java.lang.Iterable<? extends org.tensorflow.proto.util.Event> values) {
+      if (workerLogBuilder_ == null) {
+        ensureWorkerLogIsMutable();
+        com.google.protobuf.AbstractMessageLite.Builder.addAll(
+            values, workerLog_);
+        onChanged();
+      } else {
+        workerLogBuilder_.addAllMessages(values);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.Event worker_log = 2;</code>
+     */
+    public Builder clearWorkerLog() {
+      if (workerLogBuilder_ == null) {
+        workerLog_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000001);
+        onChanged();
+      } else {
+        workerLogBuilder_.clear();
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.Event worker_log = 2;</code>
+     */
+    public Builder removeWorkerLog(int index) {
+      if (workerLogBuilder_ == null) {
+        ensureWorkerLogIsMutable();
+        workerLog_.remove(index);
+        onChanged();
+      } else {
+        workerLogBuilder_.remove(index);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.Event worker_log = 2;</code>
+     */
+    public org.tensorflow.proto.util.Event.Builder getWorkerLogBuilder(
+        int index) {
+      return getWorkerLogFieldBuilder().getBuilder(index);
+    }
+    /**
+     * <code>repeated .tensorflow.Event worker_log = 2;</code>
+     */
+    public org.tensorflow.proto.util.EventOrBuilder getWorkerLogOrBuilder(
+        int index) {
+      if (workerLogBuilder_ == null) {
+        return workerLog_.get(index);  } else {
+        return workerLogBuilder_.getMessageOrBuilder(index);
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.Event worker_log = 2;</code>
+     */
+    public java.util.List<? extends org.tensorflow.proto.util.EventOrBuilder> 
+         getWorkerLogOrBuilderList() {
+      if (workerLogBuilder_ != null) {
+        return workerLogBuilder_.getMessageOrBuilderList();
+      } else {
+        return java.util.Collections.unmodifiableList(workerLog_);
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.Event worker_log = 2;</code>
+     */
+    public org.tensorflow.proto.util.Event.Builder addWorkerLogBuilder() {
+      return getWorkerLogFieldBuilder().addBuilder(
+          org.tensorflow.proto.util.Event.getDefaultInstance());
+    }
+    /**
+     * <code>repeated .tensorflow.Event worker_log = 2;</code>
+     */
+    public org.tensorflow.proto.util.Event.Builder addWorkerLogBuilder(
+        int index) {
+      return getWorkerLogFieldBuilder().addBuilder(
+          index, org.tensorflow.proto.util.Event.getDefaultInstance());
+    }
+    /**
+     * <code>repeated .tensorflow.Event worker_log = 2;</code>
+     */
+    public java.util.List<org.tensorflow.proto.util.Event.Builder> 
+         getWorkerLogBuilderList() {
+      return getWorkerLogFieldBuilder().getBuilderList();
+    }
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.util.Event, org.tensorflow.proto.util.Event.Builder, org.tensorflow.proto.util.EventOrBuilder> 
+        getWorkerLogFieldBuilder() {
+      if (workerLogBuilder_ == null) {
+        workerLogBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
+            org.tensorflow.proto.util.Event, org.tensorflow.proto.util.Event.Builder, org.tensorflow.proto.util.EventOrBuilder>(
+                workerLog_,
+                ((bitField0_ & 0x00000001) != 0),
+                getParentForChildren(),
+                isClean());
+        workerLog_ = null;
+      }
+      return workerLogBuilder_;
+    }
+
+    private java.lang.Object hostname_ = "";
+    /**
+     * <code>string hostname = 3;</code>
+     */
+    public java.lang.String getHostname() {
+      java.lang.Object ref = hostname_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        hostname_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <code>string hostname = 3;</code>
+     */
+    public com.google.protobuf.ByteString
+        getHostnameBytes() {
+      java.lang.Object ref = hostname_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        hostname_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <code>string hostname = 3;</code>
+     */
+    public Builder setHostname(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      hostname_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>string hostname = 3;</code>
+     */
+    public Builder clearHostname() {
+      
+      hostname_ = getDefaultInstance().getHostname();
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>string hostname = 3;</code>
+     */
+    public Builder setHostnameBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      hostname_ = value;
+      onChanged();
+      return this;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.WorkerHeartbeatResponse)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.WorkerHeartbeatResponse)
+  private static final org.tensorflow.proto.util.WorkerHeartbeatResponse DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.util.WorkerHeartbeatResponse();
+  }
+
+  public static org.tensorflow.proto.util.WorkerHeartbeatResponse getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<WorkerHeartbeatResponse>
+      PARSER = new com.google.protobuf.AbstractParser<WorkerHeartbeatResponse>() {
+    @java.lang.Override
+    public WorkerHeartbeatResponse parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new WorkerHeartbeatResponse(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<WorkerHeartbeatResponse> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<WorkerHeartbeatResponse> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.util.WorkerHeartbeatResponse getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/WorkerHeartbeatResponseOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/WorkerHeartbeatResponseOrBuilder.java
new file mode 100644
index 00000000000..5f42f819a51
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/WorkerHeartbeatResponseOrBuilder.java
@@ -0,0 +1,52 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/util/event.proto
+
+package org.tensorflow.proto.util;
+
+public interface WorkerHeartbeatResponseOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.WorkerHeartbeatResponse)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <code>.tensorflow.WorkerHealth health_status = 1;</code>
+   */
+  int getHealthStatusValue();
+  /**
+   * <code>.tensorflow.WorkerHealth health_status = 1;</code>
+   */
+  org.tensorflow.proto.util.WorkerHealth getHealthStatus();
+
+  /**
+   * <code>repeated .tensorflow.Event worker_log = 2;</code>
+   */
+  java.util.List<org.tensorflow.proto.util.Event> 
+      getWorkerLogList();
+  /**
+   * <code>repeated .tensorflow.Event worker_log = 2;</code>
+   */
+  org.tensorflow.proto.util.Event getWorkerLog(int index);
+  /**
+   * <code>repeated .tensorflow.Event worker_log = 2;</code>
+   */
+  int getWorkerLogCount();
+  /**
+   * <code>repeated .tensorflow.Event worker_log = 2;</code>
+   */
+  java.util.List<? extends org.tensorflow.proto.util.EventOrBuilder> 
+      getWorkerLogOrBuilderList();
+  /**
+   * <code>repeated .tensorflow.Event worker_log = 2;</code>
+   */
+  org.tensorflow.proto.util.EventOrBuilder getWorkerLogOrBuilder(
+      int index);
+
+  /**
+   * <code>string hostname = 3;</code>
+   */
+  java.lang.String getHostname();
+  /**
+   * <code>string hostname = 3;</code>
+   */
+  com.google.protobuf.ByteString
+      getHostnameBytes();
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/WorkerShutdownMode.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/WorkerShutdownMode.java
new file mode 100644
index 00000000000..41c4f45373c
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/WorkerShutdownMode.java
@@ -0,0 +1,126 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/util/event.proto
+
+package org.tensorflow.proto.util;
+
+/**
+ * <pre>
+ * Indicates the behavior of the worker when an internal error or shutdown
+ * signal is received.
+ * </pre>
+ *
+ * Protobuf enum {@code tensorflow.WorkerShutdownMode}
+ */
+public enum WorkerShutdownMode
+    implements com.google.protobuf.ProtocolMessageEnum {
+  /**
+   * <code>DEFAULT = 0;</code>
+   */
+  DEFAULT(0),
+  /**
+   * <code>NOT_CONFIGURED = 1;</code>
+   */
+  NOT_CONFIGURED(1),
+  /**
+   * <code>WAIT_FOR_COORDINATOR = 2;</code>
+   */
+  WAIT_FOR_COORDINATOR(2),
+  /**
+   * <code>SHUTDOWN_AFTER_TIMEOUT = 3;</code>
+   */
+  SHUTDOWN_AFTER_TIMEOUT(3),
+  UNRECOGNIZED(-1),
+  ;
+
+  /**
+   * <code>DEFAULT = 0;</code>
+   */
+  public static final int DEFAULT_VALUE = 0;
+  /**
+   * <code>NOT_CONFIGURED = 1;</code>
+   */
+  public static final int NOT_CONFIGURED_VALUE = 1;
+  /**
+   * <code>WAIT_FOR_COORDINATOR = 2;</code>
+   */
+  public static final int WAIT_FOR_COORDINATOR_VALUE = 2;
+  /**
+   * <code>SHUTDOWN_AFTER_TIMEOUT = 3;</code>
+   */
+  public static final int SHUTDOWN_AFTER_TIMEOUT_VALUE = 3;
+
+
+  public final int getNumber() {
+    if (this == UNRECOGNIZED) {
+      throw new java.lang.IllegalArgumentException(
+          "Can't get the number of an unknown enum value.");
+    }
+    return value;
+  }
+
+  /**
+   * @deprecated Use {@link #forNumber(int)} instead.
+   */
+  @java.lang.Deprecated
+  public static WorkerShutdownMode valueOf(int value) {
+    return forNumber(value);
+  }
+
+  public static WorkerShutdownMode forNumber(int value) {
+    switch (value) {
+      case 0: return DEFAULT;
+      case 1: return NOT_CONFIGURED;
+      case 2: return WAIT_FOR_COORDINATOR;
+      case 3: return SHUTDOWN_AFTER_TIMEOUT;
+      default: return null;
+    }
+  }
+
+  public static com.google.protobuf.Internal.EnumLiteMap<WorkerShutdownMode>
+      internalGetValueMap() {
+    return internalValueMap;
+  }
+  private static final com.google.protobuf.Internal.EnumLiteMap<
+      WorkerShutdownMode> internalValueMap =
+        new com.google.protobuf.Internal.EnumLiteMap<WorkerShutdownMode>() {
+          public WorkerShutdownMode findValueByNumber(int number) {
+            return WorkerShutdownMode.forNumber(number);
+          }
+        };
+
+  public final com.google.protobuf.Descriptors.EnumValueDescriptor
+      getValueDescriptor() {
+    return getDescriptor().getValues().get(ordinal());
+  }
+  public final com.google.protobuf.Descriptors.EnumDescriptor
+      getDescriptorForType() {
+    return getDescriptor();
+  }
+  public static final com.google.protobuf.Descriptors.EnumDescriptor
+      getDescriptor() {
+    return org.tensorflow.proto.util.EventProtos.getDescriptor().getEnumTypes().get(1);
+  }
+
+  private static final WorkerShutdownMode[] VALUES = values();
+
+  public static WorkerShutdownMode valueOf(
+      com.google.protobuf.Descriptors.EnumValueDescriptor desc) {
+    if (desc.getType() != getDescriptor()) {
+      throw new java.lang.IllegalArgumentException(
+        "EnumValueDescriptor is not for this type.");
+    }
+    if (desc.getIndex() == -1) {
+      return UNRECOGNIZED;
+    }
+    return VALUES[desc.getIndex()];
+  }
+
+  private final int value;
+
+  private WorkerShutdownMode(int value) {
+    this.value = value;
+  }
+
+  // @@protoc_insertion_point(enum_scope:tensorflow.WorkerShutdownMode)
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/AvailableDeviceInfo.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/AvailableDeviceInfo.java
new file mode 100644
index 00000000000..c80f9c4ff64
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/AvailableDeviceInfo.java
@@ -0,0 +1,966 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/util/test_log.proto
+
+package org.tensorflow.proto.util.testlog;
+
+/**
+ * <pre>
+ * Matches DeviceAttributes
+ * </pre>
+ *
+ * Protobuf type {@code tensorflow.AvailableDeviceInfo}
+ */
+public  final class AvailableDeviceInfo extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.AvailableDeviceInfo)
+    AvailableDeviceInfoOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use AvailableDeviceInfo.newBuilder() to construct.
+  private AvailableDeviceInfo(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private AvailableDeviceInfo() {
+    name_ = "";
+    type_ = "";
+    physicalDescription_ = "";
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new AvailableDeviceInfo();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private AvailableDeviceInfo(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 10: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            name_ = s;
+            break;
+          }
+          case 18: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            type_ = s;
+            break;
+          }
+          case 24: {
+
+            memoryLimit_ = input.readInt64();
+            break;
+          }
+          case 34: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            physicalDescription_ = s;
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_AvailableDeviceInfo_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_AvailableDeviceInfo_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.util.testlog.AvailableDeviceInfo.class, org.tensorflow.proto.util.testlog.AvailableDeviceInfo.Builder.class);
+  }
+
+  public static final int NAME_FIELD_NUMBER = 1;
+  private volatile java.lang.Object name_;
+  /**
+   * <pre>
+   * Device name.
+   * </pre>
+   *
+   * <code>string name = 1;</code>
+   */
+  public java.lang.String getName() {
+    java.lang.Object ref = name_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      name_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * Device name.
+   * </pre>
+   *
+   * <code>string name = 1;</code>
+   */
+  public com.google.protobuf.ByteString
+      getNameBytes() {
+    java.lang.Object ref = name_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      name_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int TYPE_FIELD_NUMBER = 2;
+  private volatile java.lang.Object type_;
+  /**
+   * <pre>
+   * Device type, e.g. 'CPU' or 'GPU'.
+   * </pre>
+   *
+   * <code>string type = 2;</code>
+   */
+  public java.lang.String getType() {
+    java.lang.Object ref = type_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      type_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * Device type, e.g. 'CPU' or 'GPU'.
+   * </pre>
+   *
+   * <code>string type = 2;</code>
+   */
+  public com.google.protobuf.ByteString
+      getTypeBytes() {
+    java.lang.Object ref = type_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      type_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int MEMORY_LIMIT_FIELD_NUMBER = 3;
+  private long memoryLimit_;
+  /**
+   * <pre>
+   * Memory capacity in bytes.
+   * </pre>
+   *
+   * <code>int64 memory_limit = 3;</code>
+   */
+  public long getMemoryLimit() {
+    return memoryLimit_;
+  }
+
+  public static final int PHYSICAL_DESCRIPTION_FIELD_NUMBER = 4;
+  private volatile java.lang.Object physicalDescription_;
+  /**
+   * <pre>
+   * The physical description of this device.
+   * </pre>
+   *
+   * <code>string physical_description = 4;</code>
+   */
+  public java.lang.String getPhysicalDescription() {
+    java.lang.Object ref = physicalDescription_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      physicalDescription_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * The physical description of this device.
+   * </pre>
+   *
+   * <code>string physical_description = 4;</code>
+   */
+  public com.google.protobuf.ByteString
+      getPhysicalDescriptionBytes() {
+    java.lang.Object ref = physicalDescription_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      physicalDescription_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    if (!getNameBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 1, name_);
+    }
+    if (!getTypeBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 2, type_);
+    }
+    if (memoryLimit_ != 0L) {
+      output.writeInt64(3, memoryLimit_);
+    }
+    if (!getPhysicalDescriptionBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 4, physicalDescription_);
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (!getNameBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, name_);
+    }
+    if (!getTypeBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(2, type_);
+    }
+    if (memoryLimit_ != 0L) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt64Size(3, memoryLimit_);
+    }
+    if (!getPhysicalDescriptionBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(4, physicalDescription_);
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.util.testlog.AvailableDeviceInfo)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.util.testlog.AvailableDeviceInfo other = (org.tensorflow.proto.util.testlog.AvailableDeviceInfo) obj;
+
+    if (!getName()
+        .equals(other.getName())) return false;
+    if (!getType()
+        .equals(other.getType())) return false;
+    if (getMemoryLimit()
+        != other.getMemoryLimit()) return false;
+    if (!getPhysicalDescription()
+        .equals(other.getPhysicalDescription())) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    hash = (37 * hash) + NAME_FIELD_NUMBER;
+    hash = (53 * hash) + getName().hashCode();
+    hash = (37 * hash) + TYPE_FIELD_NUMBER;
+    hash = (53 * hash) + getType().hashCode();
+    hash = (37 * hash) + MEMORY_LIMIT_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+        getMemoryLimit());
+    hash = (37 * hash) + PHYSICAL_DESCRIPTION_FIELD_NUMBER;
+    hash = (53 * hash) + getPhysicalDescription().hashCode();
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.util.testlog.AvailableDeviceInfo parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.testlog.AvailableDeviceInfo parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.testlog.AvailableDeviceInfo parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.testlog.AvailableDeviceInfo parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.testlog.AvailableDeviceInfo parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.testlog.AvailableDeviceInfo parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.testlog.AvailableDeviceInfo parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.testlog.AvailableDeviceInfo parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.testlog.AvailableDeviceInfo parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.testlog.AvailableDeviceInfo parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.testlog.AvailableDeviceInfo parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.testlog.AvailableDeviceInfo parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.util.testlog.AvailableDeviceInfo prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * <pre>
+   * Matches DeviceAttributes
+   * </pre>
+   *
+   * Protobuf type {@code tensorflow.AvailableDeviceInfo}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.AvailableDeviceInfo)
+      org.tensorflow.proto.util.testlog.AvailableDeviceInfoOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_AvailableDeviceInfo_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_AvailableDeviceInfo_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.util.testlog.AvailableDeviceInfo.class, org.tensorflow.proto.util.testlog.AvailableDeviceInfo.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.util.testlog.AvailableDeviceInfo.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      name_ = "";
+
+      type_ = "";
+
+      memoryLimit_ = 0L;
+
+      physicalDescription_ = "";
+
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_AvailableDeviceInfo_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.testlog.AvailableDeviceInfo getDefaultInstanceForType() {
+      return org.tensorflow.proto.util.testlog.AvailableDeviceInfo.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.testlog.AvailableDeviceInfo build() {
+      org.tensorflow.proto.util.testlog.AvailableDeviceInfo result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.testlog.AvailableDeviceInfo buildPartial() {
+      org.tensorflow.proto.util.testlog.AvailableDeviceInfo result = new org.tensorflow.proto.util.testlog.AvailableDeviceInfo(this);
+      result.name_ = name_;
+      result.type_ = type_;
+      result.memoryLimit_ = memoryLimit_;
+      result.physicalDescription_ = physicalDescription_;
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.util.testlog.AvailableDeviceInfo) {
+        return mergeFrom((org.tensorflow.proto.util.testlog.AvailableDeviceInfo)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.util.testlog.AvailableDeviceInfo other) {
+      if (other == org.tensorflow.proto.util.testlog.AvailableDeviceInfo.getDefaultInstance()) return this;
+      if (!other.getName().isEmpty()) {
+        name_ = other.name_;
+        onChanged();
+      }
+      if (!other.getType().isEmpty()) {
+        type_ = other.type_;
+        onChanged();
+      }
+      if (other.getMemoryLimit() != 0L) {
+        setMemoryLimit(other.getMemoryLimit());
+      }
+      if (!other.getPhysicalDescription().isEmpty()) {
+        physicalDescription_ = other.physicalDescription_;
+        onChanged();
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.util.testlog.AvailableDeviceInfo parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.util.testlog.AvailableDeviceInfo) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+
+    private java.lang.Object name_ = "";
+    /**
+     * <pre>
+     * Device name.
+     * </pre>
+     *
+     * <code>string name = 1;</code>
+     */
+    public java.lang.String getName() {
+      java.lang.Object ref = name_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        name_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Device name.
+     * </pre>
+     *
+     * <code>string name = 1;</code>
+     */
+    public com.google.protobuf.ByteString
+        getNameBytes() {
+      java.lang.Object ref = name_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        name_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Device name.
+     * </pre>
+     *
+     * <code>string name = 1;</code>
+     */
+    public Builder setName(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      name_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Device name.
+     * </pre>
+     *
+     * <code>string name = 1;</code>
+     */
+    public Builder clearName() {
+      
+      name_ = getDefaultInstance().getName();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Device name.
+     * </pre>
+     *
+     * <code>string name = 1;</code>
+     */
+    public Builder setNameBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      name_ = value;
+      onChanged();
+      return this;
+    }
+
+    private java.lang.Object type_ = "";
+    /**
+     * <pre>
+     * Device type, e.g. 'CPU' or 'GPU'.
+     * </pre>
+     *
+     * <code>string type = 2;</code>
+     */
+    public java.lang.String getType() {
+      java.lang.Object ref = type_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        type_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Device type, e.g. 'CPU' or 'GPU'.
+     * </pre>
+     *
+     * <code>string type = 2;</code>
+     */
+    public com.google.protobuf.ByteString
+        getTypeBytes() {
+      java.lang.Object ref = type_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        type_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Device type, e.g. 'CPU' or 'GPU'.
+     * </pre>
+     *
+     * <code>string type = 2;</code>
+     */
+    public Builder setType(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      type_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Device type, e.g. 'CPU' or 'GPU'.
+     * </pre>
+     *
+     * <code>string type = 2;</code>
+     */
+    public Builder clearType() {
+      
+      type_ = getDefaultInstance().getType();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Device type, e.g. 'CPU' or 'GPU'.
+     * </pre>
+     *
+     * <code>string type = 2;</code>
+     */
+    public Builder setTypeBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      type_ = value;
+      onChanged();
+      return this;
+    }
+
+    private long memoryLimit_ ;
+    /**
+     * <pre>
+     * Memory capacity in bytes.
+     * </pre>
+     *
+     * <code>int64 memory_limit = 3;</code>
+     */
+    public long getMemoryLimit() {
+      return memoryLimit_;
+    }
+    /**
+     * <pre>
+     * Memory capacity in bytes.
+     * </pre>
+     *
+     * <code>int64 memory_limit = 3;</code>
+     */
+    public Builder setMemoryLimit(long value) {
+      
+      memoryLimit_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Memory capacity in bytes.
+     * </pre>
+     *
+     * <code>int64 memory_limit = 3;</code>
+     */
+    public Builder clearMemoryLimit() {
+      
+      memoryLimit_ = 0L;
+      onChanged();
+      return this;
+    }
+
+    private java.lang.Object physicalDescription_ = "";
+    /**
+     * <pre>
+     * The physical description of this device.
+     * </pre>
+     *
+     * <code>string physical_description = 4;</code>
+     */
+    public java.lang.String getPhysicalDescription() {
+      java.lang.Object ref = physicalDescription_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        physicalDescription_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * The physical description of this device.
+     * </pre>
+     *
+     * <code>string physical_description = 4;</code>
+     */
+    public com.google.protobuf.ByteString
+        getPhysicalDescriptionBytes() {
+      java.lang.Object ref = physicalDescription_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        physicalDescription_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * The physical description of this device.
+     * </pre>
+     *
+     * <code>string physical_description = 4;</code>
+     */
+    public Builder setPhysicalDescription(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      physicalDescription_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * The physical description of this device.
+     * </pre>
+     *
+     * <code>string physical_description = 4;</code>
+     */
+    public Builder clearPhysicalDescription() {
+      
+      physicalDescription_ = getDefaultInstance().getPhysicalDescription();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * The physical description of this device.
+     * </pre>
+     *
+     * <code>string physical_description = 4;</code>
+     */
+    public Builder setPhysicalDescriptionBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      physicalDescription_ = value;
+      onChanged();
+      return this;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.AvailableDeviceInfo)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.AvailableDeviceInfo)
+  private static final org.tensorflow.proto.util.testlog.AvailableDeviceInfo DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.util.testlog.AvailableDeviceInfo();
+  }
+
+  public static org.tensorflow.proto.util.testlog.AvailableDeviceInfo getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<AvailableDeviceInfo>
+      PARSER = new com.google.protobuf.AbstractParser<AvailableDeviceInfo>() {
+    @java.lang.Override
+    public AvailableDeviceInfo parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new AvailableDeviceInfo(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<AvailableDeviceInfo> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<AvailableDeviceInfo> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.util.testlog.AvailableDeviceInfo getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/AvailableDeviceInfoOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/AvailableDeviceInfoOrBuilder.java
new file mode 100644
index 00000000000..138fcf086ef
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/AvailableDeviceInfoOrBuilder.java
@@ -0,0 +1,72 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/util/test_log.proto
+
+package org.tensorflow.proto.util.testlog;
+
+public interface AvailableDeviceInfoOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.AvailableDeviceInfo)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <pre>
+   * Device name.
+   * </pre>
+   *
+   * <code>string name = 1;</code>
+   */
+  java.lang.String getName();
+  /**
+   * <pre>
+   * Device name.
+   * </pre>
+   *
+   * <code>string name = 1;</code>
+   */
+  com.google.protobuf.ByteString
+      getNameBytes();
+
+  /**
+   * <pre>
+   * Device type, e.g. 'CPU' or 'GPU'.
+   * </pre>
+   *
+   * <code>string type = 2;</code>
+   */
+  java.lang.String getType();
+  /**
+   * <pre>
+   * Device type, e.g. 'CPU' or 'GPU'.
+   * </pre>
+   *
+   * <code>string type = 2;</code>
+   */
+  com.google.protobuf.ByteString
+      getTypeBytes();
+
+  /**
+   * <pre>
+   * Memory capacity in bytes.
+   * </pre>
+   *
+   * <code>int64 memory_limit = 3;</code>
+   */
+  long getMemoryLimit();
+
+  /**
+   * <pre>
+   * The physical description of this device.
+   * </pre>
+   *
+   * <code>string physical_description = 4;</code>
+   */
+  java.lang.String getPhysicalDescription();
+  /**
+   * <pre>
+   * The physical description of this device.
+   * </pre>
+   *
+   * <code>string physical_description = 4;</code>
+   */
+  com.google.protobuf.ByteString
+      getPhysicalDescriptionBytes();
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/BenchmarkEntries.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/BenchmarkEntries.java
new file mode 100644
index 00000000000..27fe4cdcc81
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/BenchmarkEntries.java
@@ -0,0 +1,765 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/util/test_log.proto
+
+package org.tensorflow.proto.util.testlog;
+
+/**
+ * Protobuf type {@code tensorflow.BenchmarkEntries}
+ */
+public  final class BenchmarkEntries extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.BenchmarkEntries)
+    BenchmarkEntriesOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use BenchmarkEntries.newBuilder() to construct.
+  private BenchmarkEntries(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private BenchmarkEntries() {
+    entry_ = java.util.Collections.emptyList();
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new BenchmarkEntries();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private BenchmarkEntries(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    int mutable_bitField0_ = 0;
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 10: {
+            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
+              entry_ = new java.util.ArrayList<org.tensorflow.proto.util.testlog.BenchmarkEntry>();
+              mutable_bitField0_ |= 0x00000001;
+            }
+            entry_.add(
+                input.readMessage(org.tensorflow.proto.util.testlog.BenchmarkEntry.parser(), extensionRegistry));
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      if (((mutable_bitField0_ & 0x00000001) != 0)) {
+        entry_ = java.util.Collections.unmodifiableList(entry_);
+      }
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_BenchmarkEntries_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_BenchmarkEntries_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.util.testlog.BenchmarkEntries.class, org.tensorflow.proto.util.testlog.BenchmarkEntries.Builder.class);
+  }
+
+  public static final int ENTRY_FIELD_NUMBER = 1;
+  private java.util.List<org.tensorflow.proto.util.testlog.BenchmarkEntry> entry_;
+  /**
+   * <code>repeated .tensorflow.BenchmarkEntry entry = 1;</code>
+   */
+  public java.util.List<org.tensorflow.proto.util.testlog.BenchmarkEntry> getEntryList() {
+    return entry_;
+  }
+  /**
+   * <code>repeated .tensorflow.BenchmarkEntry entry = 1;</code>
+   */
+  public java.util.List<? extends org.tensorflow.proto.util.testlog.BenchmarkEntryOrBuilder> 
+      getEntryOrBuilderList() {
+    return entry_;
+  }
+  /**
+   * <code>repeated .tensorflow.BenchmarkEntry entry = 1;</code>
+   */
+  public int getEntryCount() {
+    return entry_.size();
+  }
+  /**
+   * <code>repeated .tensorflow.BenchmarkEntry entry = 1;</code>
+   */
+  public org.tensorflow.proto.util.testlog.BenchmarkEntry getEntry(int index) {
+    return entry_.get(index);
+  }
+  /**
+   * <code>repeated .tensorflow.BenchmarkEntry entry = 1;</code>
+   */
+  public org.tensorflow.proto.util.testlog.BenchmarkEntryOrBuilder getEntryOrBuilder(
+      int index) {
+    return entry_.get(index);
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    for (int i = 0; i < entry_.size(); i++) {
+      output.writeMessage(1, entry_.get(i));
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    for (int i = 0; i < entry_.size(); i++) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(1, entry_.get(i));
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.util.testlog.BenchmarkEntries)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.util.testlog.BenchmarkEntries other = (org.tensorflow.proto.util.testlog.BenchmarkEntries) obj;
+
+    if (!getEntryList()
+        .equals(other.getEntryList())) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    if (getEntryCount() > 0) {
+      hash = (37 * hash) + ENTRY_FIELD_NUMBER;
+      hash = (53 * hash) + getEntryList().hashCode();
+    }
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.util.testlog.BenchmarkEntries parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.testlog.BenchmarkEntries parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.testlog.BenchmarkEntries parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.testlog.BenchmarkEntries parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.testlog.BenchmarkEntries parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.testlog.BenchmarkEntries parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.testlog.BenchmarkEntries parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.testlog.BenchmarkEntries parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.testlog.BenchmarkEntries parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.testlog.BenchmarkEntries parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.testlog.BenchmarkEntries parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.testlog.BenchmarkEntries parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.util.testlog.BenchmarkEntries prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * Protobuf type {@code tensorflow.BenchmarkEntries}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.BenchmarkEntries)
+      org.tensorflow.proto.util.testlog.BenchmarkEntriesOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_BenchmarkEntries_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_BenchmarkEntries_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.util.testlog.BenchmarkEntries.class, org.tensorflow.proto.util.testlog.BenchmarkEntries.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.util.testlog.BenchmarkEntries.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+        getEntryFieldBuilder();
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      if (entryBuilder_ == null) {
+        entry_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000001);
+      } else {
+        entryBuilder_.clear();
+      }
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_BenchmarkEntries_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.testlog.BenchmarkEntries getDefaultInstanceForType() {
+      return org.tensorflow.proto.util.testlog.BenchmarkEntries.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.testlog.BenchmarkEntries build() {
+      org.tensorflow.proto.util.testlog.BenchmarkEntries result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.testlog.BenchmarkEntries buildPartial() {
+      org.tensorflow.proto.util.testlog.BenchmarkEntries result = new org.tensorflow.proto.util.testlog.BenchmarkEntries(this);
+      int from_bitField0_ = bitField0_;
+      if (entryBuilder_ == null) {
+        if (((bitField0_ & 0x00000001) != 0)) {
+          entry_ = java.util.Collections.unmodifiableList(entry_);
+          bitField0_ = (bitField0_ & ~0x00000001);
+        }
+        result.entry_ = entry_;
+      } else {
+        result.entry_ = entryBuilder_.build();
+      }
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.util.testlog.BenchmarkEntries) {
+        return mergeFrom((org.tensorflow.proto.util.testlog.BenchmarkEntries)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.util.testlog.BenchmarkEntries other) {
+      if (other == org.tensorflow.proto.util.testlog.BenchmarkEntries.getDefaultInstance()) return this;
+      if (entryBuilder_ == null) {
+        if (!other.entry_.isEmpty()) {
+          if (entry_.isEmpty()) {
+            entry_ = other.entry_;
+            bitField0_ = (bitField0_ & ~0x00000001);
+          } else {
+            ensureEntryIsMutable();
+            entry_.addAll(other.entry_);
+          }
+          onChanged();
+        }
+      } else {
+        if (!other.entry_.isEmpty()) {
+          if (entryBuilder_.isEmpty()) {
+            entryBuilder_.dispose();
+            entryBuilder_ = null;
+            entry_ = other.entry_;
+            bitField0_ = (bitField0_ & ~0x00000001);
+            entryBuilder_ = 
+              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
+                 getEntryFieldBuilder() : null;
+          } else {
+            entryBuilder_.addAllMessages(other.entry_);
+          }
+        }
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.util.testlog.BenchmarkEntries parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.util.testlog.BenchmarkEntries) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+    private int bitField0_;
+
+    private java.util.List<org.tensorflow.proto.util.testlog.BenchmarkEntry> entry_ =
+      java.util.Collections.emptyList();
+    private void ensureEntryIsMutable() {
+      if (!((bitField0_ & 0x00000001) != 0)) {
+        entry_ = new java.util.ArrayList<org.tensorflow.proto.util.testlog.BenchmarkEntry>(entry_);
+        bitField0_ |= 0x00000001;
+       }
+    }
+
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.util.testlog.BenchmarkEntry, org.tensorflow.proto.util.testlog.BenchmarkEntry.Builder, org.tensorflow.proto.util.testlog.BenchmarkEntryOrBuilder> entryBuilder_;
+
+    /**
+     * <code>repeated .tensorflow.BenchmarkEntry entry = 1;</code>
+     */
+    public java.util.List<org.tensorflow.proto.util.testlog.BenchmarkEntry> getEntryList() {
+      if (entryBuilder_ == null) {
+        return java.util.Collections.unmodifiableList(entry_);
+      } else {
+        return entryBuilder_.getMessageList();
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.BenchmarkEntry entry = 1;</code>
+     */
+    public int getEntryCount() {
+      if (entryBuilder_ == null) {
+        return entry_.size();
+      } else {
+        return entryBuilder_.getCount();
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.BenchmarkEntry entry = 1;</code>
+     */
+    public org.tensorflow.proto.util.testlog.BenchmarkEntry getEntry(int index) {
+      if (entryBuilder_ == null) {
+        return entry_.get(index);
+      } else {
+        return entryBuilder_.getMessage(index);
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.BenchmarkEntry entry = 1;</code>
+     */
+    public Builder setEntry(
+        int index, org.tensorflow.proto.util.testlog.BenchmarkEntry value) {
+      if (entryBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureEntryIsMutable();
+        entry_.set(index, value);
+        onChanged();
+      } else {
+        entryBuilder_.setMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.BenchmarkEntry entry = 1;</code>
+     */
+    public Builder setEntry(
+        int index, org.tensorflow.proto.util.testlog.BenchmarkEntry.Builder builderForValue) {
+      if (entryBuilder_ == null) {
+        ensureEntryIsMutable();
+        entry_.set(index, builderForValue.build());
+        onChanged();
+      } else {
+        entryBuilder_.setMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.BenchmarkEntry entry = 1;</code>
+     */
+    public Builder addEntry(org.tensorflow.proto.util.testlog.BenchmarkEntry value) {
+      if (entryBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureEntryIsMutable();
+        entry_.add(value);
+        onChanged();
+      } else {
+        entryBuilder_.addMessage(value);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.BenchmarkEntry entry = 1;</code>
+     */
+    public Builder addEntry(
+        int index, org.tensorflow.proto.util.testlog.BenchmarkEntry value) {
+      if (entryBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureEntryIsMutable();
+        entry_.add(index, value);
+        onChanged();
+      } else {
+        entryBuilder_.addMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.BenchmarkEntry entry = 1;</code>
+     */
+    public Builder addEntry(
+        org.tensorflow.proto.util.testlog.BenchmarkEntry.Builder builderForValue) {
+      if (entryBuilder_ == null) {
+        ensureEntryIsMutable();
+        entry_.add(builderForValue.build());
+        onChanged();
+      } else {
+        entryBuilder_.addMessage(builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.BenchmarkEntry entry = 1;</code>
+     */
+    public Builder addEntry(
+        int index, org.tensorflow.proto.util.testlog.BenchmarkEntry.Builder builderForValue) {
+      if (entryBuilder_ == null) {
+        ensureEntryIsMutable();
+        entry_.add(index, builderForValue.build());
+        onChanged();
+      } else {
+        entryBuilder_.addMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.BenchmarkEntry entry = 1;</code>
+     */
+    public Builder addAllEntry(
+        java.lang.Iterable<? extends org.tensorflow.proto.util.testlog.BenchmarkEntry> values) {
+      if (entryBuilder_ == null) {
+        ensureEntryIsMutable();
+        com.google.protobuf.AbstractMessageLite.Builder.addAll(
+            values, entry_);
+        onChanged();
+      } else {
+        entryBuilder_.addAllMessages(values);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.BenchmarkEntry entry = 1;</code>
+     */
+    public Builder clearEntry() {
+      if (entryBuilder_ == null) {
+        entry_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000001);
+        onChanged();
+      } else {
+        entryBuilder_.clear();
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.BenchmarkEntry entry = 1;</code>
+     */
+    public Builder removeEntry(int index) {
+      if (entryBuilder_ == null) {
+        ensureEntryIsMutable();
+        entry_.remove(index);
+        onChanged();
+      } else {
+        entryBuilder_.remove(index);
+      }
+      return this;
+    }
+    /**
+     * <code>repeated .tensorflow.BenchmarkEntry entry = 1;</code>
+     */
+    public org.tensorflow.proto.util.testlog.BenchmarkEntry.Builder getEntryBuilder(
+        int index) {
+      return getEntryFieldBuilder().getBuilder(index);
+    }
+    /**
+     * <code>repeated .tensorflow.BenchmarkEntry entry = 1;</code>
+     */
+    public org.tensorflow.proto.util.testlog.BenchmarkEntryOrBuilder getEntryOrBuilder(
+        int index) {
+      if (entryBuilder_ == null) {
+        return entry_.get(index);  } else {
+        return entryBuilder_.getMessageOrBuilder(index);
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.BenchmarkEntry entry = 1;</code>
+     */
+    public java.util.List<? extends org.tensorflow.proto.util.testlog.BenchmarkEntryOrBuilder> 
+         getEntryOrBuilderList() {
+      if (entryBuilder_ != null) {
+        return entryBuilder_.getMessageOrBuilderList();
+      } else {
+        return java.util.Collections.unmodifiableList(entry_);
+      }
+    }
+    /**
+     * <code>repeated .tensorflow.BenchmarkEntry entry = 1;</code>
+     */
+    public org.tensorflow.proto.util.testlog.BenchmarkEntry.Builder addEntryBuilder() {
+      return getEntryFieldBuilder().addBuilder(
+          org.tensorflow.proto.util.testlog.BenchmarkEntry.getDefaultInstance());
+    }
+    /**
+     * <code>repeated .tensorflow.BenchmarkEntry entry = 1;</code>
+     */
+    public org.tensorflow.proto.util.testlog.BenchmarkEntry.Builder addEntryBuilder(
+        int index) {
+      return getEntryFieldBuilder().addBuilder(
+          index, org.tensorflow.proto.util.testlog.BenchmarkEntry.getDefaultInstance());
+    }
+    /**
+     * <code>repeated .tensorflow.BenchmarkEntry entry = 1;</code>
+     */
+    public java.util.List<org.tensorflow.proto.util.testlog.BenchmarkEntry.Builder> 
+         getEntryBuilderList() {
+      return getEntryFieldBuilder().getBuilderList();
+    }
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.util.testlog.BenchmarkEntry, org.tensorflow.proto.util.testlog.BenchmarkEntry.Builder, org.tensorflow.proto.util.testlog.BenchmarkEntryOrBuilder> 
+        getEntryFieldBuilder() {
+      if (entryBuilder_ == null) {
+        entryBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
+            org.tensorflow.proto.util.testlog.BenchmarkEntry, org.tensorflow.proto.util.testlog.BenchmarkEntry.Builder, org.tensorflow.proto.util.testlog.BenchmarkEntryOrBuilder>(
+                entry_,
+                ((bitField0_ & 0x00000001) != 0),
+                getParentForChildren(),
+                isClean());
+        entry_ = null;
+      }
+      return entryBuilder_;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.BenchmarkEntries)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.BenchmarkEntries)
+  private static final org.tensorflow.proto.util.testlog.BenchmarkEntries DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.util.testlog.BenchmarkEntries();
+  }
+
+  public static org.tensorflow.proto.util.testlog.BenchmarkEntries getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<BenchmarkEntries>
+      PARSER = new com.google.protobuf.AbstractParser<BenchmarkEntries>() {
+    @java.lang.Override
+    public BenchmarkEntries parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new BenchmarkEntries(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<BenchmarkEntries> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<BenchmarkEntries> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.util.testlog.BenchmarkEntries getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/BenchmarkEntriesOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/BenchmarkEntriesOrBuilder.java
new file mode 100644
index 00000000000..25a3905dfdf
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/BenchmarkEntriesOrBuilder.java
@@ -0,0 +1,33 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/util/test_log.proto
+
+package org.tensorflow.proto.util.testlog;
+
+public interface BenchmarkEntriesOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.BenchmarkEntries)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <code>repeated .tensorflow.BenchmarkEntry entry = 1;</code>
+   */
+  java.util.List<org.tensorflow.proto.util.testlog.BenchmarkEntry> 
+      getEntryList();
+  /**
+   * <code>repeated .tensorflow.BenchmarkEntry entry = 1;</code>
+   */
+  org.tensorflow.proto.util.testlog.BenchmarkEntry getEntry(int index);
+  /**
+   * <code>repeated .tensorflow.BenchmarkEntry entry = 1;</code>
+   */
+  int getEntryCount();
+  /**
+   * <code>repeated .tensorflow.BenchmarkEntry entry = 1;</code>
+   */
+  java.util.List<? extends org.tensorflow.proto.util.testlog.BenchmarkEntryOrBuilder> 
+      getEntryOrBuilderList();
+  /**
+   * <code>repeated .tensorflow.BenchmarkEntry entry = 1;</code>
+   */
+  org.tensorflow.proto.util.testlog.BenchmarkEntryOrBuilder getEntryOrBuilder(
+      int index);
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/BenchmarkEntry.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/BenchmarkEntry.java
new file mode 100644
index 00000000000..b83843aa744
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/BenchmarkEntry.java
@@ -0,0 +1,1676 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/util/test_log.proto
+
+package org.tensorflow.proto.util.testlog;
+
+/**
+ * <pre>
+ * Each unit test or benchmark in a test or benchmark run provides
+ * some set of information.  Here we provide some reasonable keys
+ * one would expect to see, with optional key/value pairs for things
+ * we haven't considered.
+ * This BenchmarkEntry should be emitted by each unit test or benchmark
+ * reporter.
+ * </pre>
+ *
+ * Protobuf type {@code tensorflow.BenchmarkEntry}
+ */
+public  final class BenchmarkEntry extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.BenchmarkEntry)
+    BenchmarkEntryOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use BenchmarkEntry.newBuilder() to construct.
+  private BenchmarkEntry(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private BenchmarkEntry() {
+    name_ = "";
+    metrics_ = java.util.Collections.emptyList();
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new BenchmarkEntry();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private BenchmarkEntry(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    int mutable_bitField0_ = 0;
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 10: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            name_ = s;
+            break;
+          }
+          case 16: {
+
+            iters_ = input.readInt64();
+            break;
+          }
+          case 25: {
+
+            cpuTime_ = input.readDouble();
+            break;
+          }
+          case 33: {
+
+            wallTime_ = input.readDouble();
+            break;
+          }
+          case 41: {
+
+            throughput_ = input.readDouble();
+            break;
+          }
+          case 50: {
+            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
+              extras_ = com.google.protobuf.MapField.newMapField(
+                  ExtrasDefaultEntryHolder.defaultEntry);
+              mutable_bitField0_ |= 0x00000001;
+            }
+            com.google.protobuf.MapEntry<java.lang.String, org.tensorflow.proto.util.testlog.EntryValue>
+            extras__ = input.readMessage(
+                ExtrasDefaultEntryHolder.defaultEntry.getParserForType(), extensionRegistry);
+            extras_.getMutableMap().put(
+                extras__.getKey(), extras__.getValue());
+            break;
+          }
+          case 58: {
+            if (!((mutable_bitField0_ & 0x00000002) != 0)) {
+              metrics_ = new java.util.ArrayList<org.tensorflow.proto.util.testlog.MetricEntry>();
+              mutable_bitField0_ |= 0x00000002;
+            }
+            metrics_.add(
+                input.readMessage(org.tensorflow.proto.util.testlog.MetricEntry.parser(), extensionRegistry));
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      if (((mutable_bitField0_ & 0x00000002) != 0)) {
+        metrics_ = java.util.Collections.unmodifiableList(metrics_);
+      }
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_BenchmarkEntry_descriptor;
+  }
+
+  @SuppressWarnings({"rawtypes"})
+  @java.lang.Override
+  protected com.google.protobuf.MapField internalGetMapField(
+      int number) {
+    switch (number) {
+      case 6:
+        return internalGetExtras();
+      default:
+        throw new RuntimeException(
+            "Invalid map field number: " + number);
+    }
+  }
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_BenchmarkEntry_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.util.testlog.BenchmarkEntry.class, org.tensorflow.proto.util.testlog.BenchmarkEntry.Builder.class);
+  }
+
+  public static final int NAME_FIELD_NUMBER = 1;
+  private volatile java.lang.Object name_;
+  /**
+   * <pre>
+   * The name of the specific benchmark or test
+   * (e.g. BM_AdjustContrast_gpu_B_W_H)
+   * </pre>
+   *
+   * <code>string name = 1;</code>
+   */
+  public java.lang.String getName() {
+    java.lang.Object ref = name_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      name_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * The name of the specific benchmark or test
+   * (e.g. BM_AdjustContrast_gpu_B_W_H)
+   * </pre>
+   *
+   * <code>string name = 1;</code>
+   */
+  public com.google.protobuf.ByteString
+      getNameBytes() {
+    java.lang.Object ref = name_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      name_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int ITERS_FIELD_NUMBER = 2;
+  private long iters_;
+  /**
+   * <pre>
+   * If a benchmark, how many iterations it was run for
+   * </pre>
+   *
+   * <code>int64 iters = 2;</code>
+   */
+  public long getIters() {
+    return iters_;
+  }
+
+  public static final int CPU_TIME_FIELD_NUMBER = 3;
+  private double cpuTime_;
+  /**
+   * <pre>
+   * Total cpu time used for all iterations (in seconds)
+   * </pre>
+   *
+   * <code>double cpu_time = 3;</code>
+   */
+  public double getCpuTime() {
+    return cpuTime_;
+  }
+
+  public static final int WALL_TIME_FIELD_NUMBER = 4;
+  private double wallTime_;
+  /**
+   * <pre>
+   * Total wall time used for all iterations (in seconds)
+   * </pre>
+   *
+   * <code>double wall_time = 4;</code>
+   */
+  public double getWallTime() {
+    return wallTime_;
+  }
+
+  public static final int THROUGHPUT_FIELD_NUMBER = 5;
+  private double throughput_;
+  /**
+   * <pre>
+   * Throughput (in MB/s)
+   * </pre>
+   *
+   * <code>double throughput = 5;</code>
+   */
+  public double getThroughput() {
+    return throughput_;
+  }
+
+  public static final int EXTRAS_FIELD_NUMBER = 6;
+  private static final class ExtrasDefaultEntryHolder {
+    static final com.google.protobuf.MapEntry<
+        java.lang.String, org.tensorflow.proto.util.testlog.EntryValue> defaultEntry =
+            com.google.protobuf.MapEntry
+            .<java.lang.String, org.tensorflow.proto.util.testlog.EntryValue>newDefaultInstance(
+                org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_BenchmarkEntry_ExtrasEntry_descriptor, 
+                com.google.protobuf.WireFormat.FieldType.STRING,
+                "",
+                com.google.protobuf.WireFormat.FieldType.MESSAGE,
+                org.tensorflow.proto.util.testlog.EntryValue.getDefaultInstance());
+  }
+  private com.google.protobuf.MapField<
+      java.lang.String, org.tensorflow.proto.util.testlog.EntryValue> extras_;
+  private com.google.protobuf.MapField<java.lang.String, org.tensorflow.proto.util.testlog.EntryValue>
+  internalGetExtras() {
+    if (extras_ == null) {
+      return com.google.protobuf.MapField.emptyMapField(
+          ExtrasDefaultEntryHolder.defaultEntry);
+    }
+    return extras_;
+  }
+
+  public int getExtrasCount() {
+    return internalGetExtras().getMap().size();
+  }
+  /**
+   * <pre>
+   * Generic map from result key to value.
+   * </pre>
+   *
+   * <code>map&lt;string, .tensorflow.EntryValue&gt; extras = 6;</code>
+   */
+
+  public boolean containsExtras(
+      java.lang.String key) {
+    if (key == null) { throw new java.lang.NullPointerException(); }
+    return internalGetExtras().getMap().containsKey(key);
+  }
+  /**
+   * Use {@link #getExtrasMap()} instead.
+   */
+  @java.lang.Deprecated
+  public java.util.Map<java.lang.String, org.tensorflow.proto.util.testlog.EntryValue> getExtras() {
+    return getExtrasMap();
+  }
+  /**
+   * <pre>
+   * Generic map from result key to value.
+   * </pre>
+   *
+   * <code>map&lt;string, .tensorflow.EntryValue&gt; extras = 6;</code>
+   */
+
+  public java.util.Map<java.lang.String, org.tensorflow.proto.util.testlog.EntryValue> getExtrasMap() {
+    return internalGetExtras().getMap();
+  }
+  /**
+   * <pre>
+   * Generic map from result key to value.
+   * </pre>
+   *
+   * <code>map&lt;string, .tensorflow.EntryValue&gt; extras = 6;</code>
+   */
+
+  public org.tensorflow.proto.util.testlog.EntryValue getExtrasOrDefault(
+      java.lang.String key,
+      org.tensorflow.proto.util.testlog.EntryValue defaultValue) {
+    if (key == null) { throw new java.lang.NullPointerException(); }
+    java.util.Map<java.lang.String, org.tensorflow.proto.util.testlog.EntryValue> map =
+        internalGetExtras().getMap();
+    return map.containsKey(key) ? map.get(key) : defaultValue;
+  }
+  /**
+   * <pre>
+   * Generic map from result key to value.
+   * </pre>
+   *
+   * <code>map&lt;string, .tensorflow.EntryValue&gt; extras = 6;</code>
+   */
+
+  public org.tensorflow.proto.util.testlog.EntryValue getExtrasOrThrow(
+      java.lang.String key) {
+    if (key == null) { throw new java.lang.NullPointerException(); }
+    java.util.Map<java.lang.String, org.tensorflow.proto.util.testlog.EntryValue> map =
+        internalGetExtras().getMap();
+    if (!map.containsKey(key)) {
+      throw new java.lang.IllegalArgumentException();
+    }
+    return map.get(key);
+  }
+
+  public static final int METRICS_FIELD_NUMBER = 7;
+  private java.util.List<org.tensorflow.proto.util.testlog.MetricEntry> metrics_;
+  /**
+   * <pre>
+   * Metric name, value and expected range. This can include accuracy metrics
+   * typically used to determine whether the accuracy test has passed
+   * </pre>
+   *
+   * <code>repeated .tensorflow.MetricEntry metrics = 7;</code>
+   */
+  public java.util.List<org.tensorflow.proto.util.testlog.MetricEntry> getMetricsList() {
+    return metrics_;
+  }
+  /**
+   * <pre>
+   * Metric name, value and expected range. This can include accuracy metrics
+   * typically used to determine whether the accuracy test has passed
+   * </pre>
+   *
+   * <code>repeated .tensorflow.MetricEntry metrics = 7;</code>
+   */
+  public java.util.List<? extends org.tensorflow.proto.util.testlog.MetricEntryOrBuilder> 
+      getMetricsOrBuilderList() {
+    return metrics_;
+  }
+  /**
+   * <pre>
+   * Metric name, value and expected range. This can include accuracy metrics
+   * typically used to determine whether the accuracy test has passed
+   * </pre>
+   *
+   * <code>repeated .tensorflow.MetricEntry metrics = 7;</code>
+   */
+  public int getMetricsCount() {
+    return metrics_.size();
+  }
+  /**
+   * <pre>
+   * Metric name, value and expected range. This can include accuracy metrics
+   * typically used to determine whether the accuracy test has passed
+   * </pre>
+   *
+   * <code>repeated .tensorflow.MetricEntry metrics = 7;</code>
+   */
+  public org.tensorflow.proto.util.testlog.MetricEntry getMetrics(int index) {
+    return metrics_.get(index);
+  }
+  /**
+   * <pre>
+   * Metric name, value and expected range. This can include accuracy metrics
+   * typically used to determine whether the accuracy test has passed
+   * </pre>
+   *
+   * <code>repeated .tensorflow.MetricEntry metrics = 7;</code>
+   */
+  public org.tensorflow.proto.util.testlog.MetricEntryOrBuilder getMetricsOrBuilder(
+      int index) {
+    return metrics_.get(index);
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    if (!getNameBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 1, name_);
+    }
+    if (iters_ != 0L) {
+      output.writeInt64(2, iters_);
+    }
+    if (cpuTime_ != 0D) {
+      output.writeDouble(3, cpuTime_);
+    }
+    if (wallTime_ != 0D) {
+      output.writeDouble(4, wallTime_);
+    }
+    if (throughput_ != 0D) {
+      output.writeDouble(5, throughput_);
+    }
+    com.google.protobuf.GeneratedMessageV3
+      .serializeStringMapTo(
+        output,
+        internalGetExtras(),
+        ExtrasDefaultEntryHolder.defaultEntry,
+        6);
+    for (int i = 0; i < metrics_.size(); i++) {
+      output.writeMessage(7, metrics_.get(i));
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (!getNameBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, name_);
+    }
+    if (iters_ != 0L) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt64Size(2, iters_);
+    }
+    if (cpuTime_ != 0D) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeDoubleSize(3, cpuTime_);
+    }
+    if (wallTime_ != 0D) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeDoubleSize(4, wallTime_);
+    }
+    if (throughput_ != 0D) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeDoubleSize(5, throughput_);
+    }
+    for (java.util.Map.Entry<java.lang.String, org.tensorflow.proto.util.testlog.EntryValue> entry
+         : internalGetExtras().getMap().entrySet()) {
+      com.google.protobuf.MapEntry<java.lang.String, org.tensorflow.proto.util.testlog.EntryValue>
+      extras__ = ExtrasDefaultEntryHolder.defaultEntry.newBuilderForType()
+          .setKey(entry.getKey())
+          .setValue(entry.getValue())
+          .build();
+      size += com.google.protobuf.CodedOutputStream
+          .computeMessageSize(6, extras__);
+    }
+    for (int i = 0; i < metrics_.size(); i++) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(7, metrics_.get(i));
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.util.testlog.BenchmarkEntry)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.util.testlog.BenchmarkEntry other = (org.tensorflow.proto.util.testlog.BenchmarkEntry) obj;
+
+    if (!getName()
+        .equals(other.getName())) return false;
+    if (getIters()
+        != other.getIters()) return false;
+    if (java.lang.Double.doubleToLongBits(getCpuTime())
+        != java.lang.Double.doubleToLongBits(
+            other.getCpuTime())) return false;
+    if (java.lang.Double.doubleToLongBits(getWallTime())
+        != java.lang.Double.doubleToLongBits(
+            other.getWallTime())) return false;
+    if (java.lang.Double.doubleToLongBits(getThroughput())
+        != java.lang.Double.doubleToLongBits(
+            other.getThroughput())) return false;
+    if (!internalGetExtras().equals(
+        other.internalGetExtras())) return false;
+    if (!getMetricsList()
+        .equals(other.getMetricsList())) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    hash = (37 * hash) + NAME_FIELD_NUMBER;
+    hash = (53 * hash) + getName().hashCode();
+    hash = (37 * hash) + ITERS_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+        getIters());
+    hash = (37 * hash) + CPU_TIME_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+        java.lang.Double.doubleToLongBits(getCpuTime()));
+    hash = (37 * hash) + WALL_TIME_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+        java.lang.Double.doubleToLongBits(getWallTime()));
+    hash = (37 * hash) + THROUGHPUT_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+        java.lang.Double.doubleToLongBits(getThroughput()));
+    if (!internalGetExtras().getMap().isEmpty()) {
+      hash = (37 * hash) + EXTRAS_FIELD_NUMBER;
+      hash = (53 * hash) + internalGetExtras().hashCode();
+    }
+    if (getMetricsCount() > 0) {
+      hash = (37 * hash) + METRICS_FIELD_NUMBER;
+      hash = (53 * hash) + getMetricsList().hashCode();
+    }
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.util.testlog.BenchmarkEntry parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.testlog.BenchmarkEntry parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.testlog.BenchmarkEntry parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.testlog.BenchmarkEntry parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.testlog.BenchmarkEntry parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.testlog.BenchmarkEntry parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.testlog.BenchmarkEntry parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.testlog.BenchmarkEntry parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.testlog.BenchmarkEntry parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.testlog.BenchmarkEntry parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.testlog.BenchmarkEntry parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.testlog.BenchmarkEntry parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.util.testlog.BenchmarkEntry prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * <pre>
+   * Each unit test or benchmark in a test or benchmark run provides
+   * some set of information.  Here we provide some reasonable keys
+   * one would expect to see, with optional key/value pairs for things
+   * we haven't considered.
+   * This BenchmarkEntry should be emitted by each unit test or benchmark
+   * reporter.
+   * </pre>
+   *
+   * Protobuf type {@code tensorflow.BenchmarkEntry}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.BenchmarkEntry)
+      org.tensorflow.proto.util.testlog.BenchmarkEntryOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_BenchmarkEntry_descriptor;
+    }
+
+    @SuppressWarnings({"rawtypes"})
+    protected com.google.protobuf.MapField internalGetMapField(
+        int number) {
+      switch (number) {
+        case 6:
+          return internalGetExtras();
+        default:
+          throw new RuntimeException(
+              "Invalid map field number: " + number);
+      }
+    }
+    @SuppressWarnings({"rawtypes"})
+    protected com.google.protobuf.MapField internalGetMutableMapField(
+        int number) {
+      switch (number) {
+        case 6:
+          return internalGetMutableExtras();
+        default:
+          throw new RuntimeException(
+              "Invalid map field number: " + number);
+      }
+    }
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_BenchmarkEntry_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.util.testlog.BenchmarkEntry.class, org.tensorflow.proto.util.testlog.BenchmarkEntry.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.util.testlog.BenchmarkEntry.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+        getMetricsFieldBuilder();
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      name_ = "";
+
+      iters_ = 0L;
+
+      cpuTime_ = 0D;
+
+      wallTime_ = 0D;
+
+      throughput_ = 0D;
+
+      internalGetMutableExtras().clear();
+      if (metricsBuilder_ == null) {
+        metrics_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000002);
+      } else {
+        metricsBuilder_.clear();
+      }
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_BenchmarkEntry_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.testlog.BenchmarkEntry getDefaultInstanceForType() {
+      return org.tensorflow.proto.util.testlog.BenchmarkEntry.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.testlog.BenchmarkEntry build() {
+      org.tensorflow.proto.util.testlog.BenchmarkEntry result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.testlog.BenchmarkEntry buildPartial() {
+      org.tensorflow.proto.util.testlog.BenchmarkEntry result = new org.tensorflow.proto.util.testlog.BenchmarkEntry(this);
+      int from_bitField0_ = bitField0_;
+      result.name_ = name_;
+      result.iters_ = iters_;
+      result.cpuTime_ = cpuTime_;
+      result.wallTime_ = wallTime_;
+      result.throughput_ = throughput_;
+      result.extras_ = internalGetExtras();
+      result.extras_.makeImmutable();
+      if (metricsBuilder_ == null) {
+        if (((bitField0_ & 0x00000002) != 0)) {
+          metrics_ = java.util.Collections.unmodifiableList(metrics_);
+          bitField0_ = (bitField0_ & ~0x00000002);
+        }
+        result.metrics_ = metrics_;
+      } else {
+        result.metrics_ = metricsBuilder_.build();
+      }
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.util.testlog.BenchmarkEntry) {
+        return mergeFrom((org.tensorflow.proto.util.testlog.BenchmarkEntry)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.util.testlog.BenchmarkEntry other) {
+      if (other == org.tensorflow.proto.util.testlog.BenchmarkEntry.getDefaultInstance()) return this;
+      if (!other.getName().isEmpty()) {
+        name_ = other.name_;
+        onChanged();
+      }
+      if (other.getIters() != 0L) {
+        setIters(other.getIters());
+      }
+      if (other.getCpuTime() != 0D) {
+        setCpuTime(other.getCpuTime());
+      }
+      if (other.getWallTime() != 0D) {
+        setWallTime(other.getWallTime());
+      }
+      if (other.getThroughput() != 0D) {
+        setThroughput(other.getThroughput());
+      }
+      internalGetMutableExtras().mergeFrom(
+          other.internalGetExtras());
+      if (metricsBuilder_ == null) {
+        if (!other.metrics_.isEmpty()) {
+          if (metrics_.isEmpty()) {
+            metrics_ = other.metrics_;
+            bitField0_ = (bitField0_ & ~0x00000002);
+          } else {
+            ensureMetricsIsMutable();
+            metrics_.addAll(other.metrics_);
+          }
+          onChanged();
+        }
+      } else {
+        if (!other.metrics_.isEmpty()) {
+          if (metricsBuilder_.isEmpty()) {
+            metricsBuilder_.dispose();
+            metricsBuilder_ = null;
+            metrics_ = other.metrics_;
+            bitField0_ = (bitField0_ & ~0x00000002);
+            metricsBuilder_ = 
+              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
+                 getMetricsFieldBuilder() : null;
+          } else {
+            metricsBuilder_.addAllMessages(other.metrics_);
+          }
+        }
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.util.testlog.BenchmarkEntry parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.util.testlog.BenchmarkEntry) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+    private int bitField0_;
+
+    private java.lang.Object name_ = "";
+    /**
+     * <pre>
+     * The name of the specific benchmark or test
+     * (e.g. BM_AdjustContrast_gpu_B_W_H)
+     * </pre>
+     *
+     * <code>string name = 1;</code>
+     */
+    public java.lang.String getName() {
+      java.lang.Object ref = name_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        name_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * The name of the specific benchmark or test
+     * (e.g. BM_AdjustContrast_gpu_B_W_H)
+     * </pre>
+     *
+     * <code>string name = 1;</code>
+     */
+    public com.google.protobuf.ByteString
+        getNameBytes() {
+      java.lang.Object ref = name_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        name_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * The name of the specific benchmark or test
+     * (e.g. BM_AdjustContrast_gpu_B_W_H)
+     * </pre>
+     *
+     * <code>string name = 1;</code>
+     */
+    public Builder setName(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      name_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * The name of the specific benchmark or test
+     * (e.g. BM_AdjustContrast_gpu_B_W_H)
+     * </pre>
+     *
+     * <code>string name = 1;</code>
+     */
+    public Builder clearName() {
+      
+      name_ = getDefaultInstance().getName();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * The name of the specific benchmark or test
+     * (e.g. BM_AdjustContrast_gpu_B_W_H)
+     * </pre>
+     *
+     * <code>string name = 1;</code>
+     */
+    public Builder setNameBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      name_ = value;
+      onChanged();
+      return this;
+    }
+
+    private long iters_ ;
+    /**
+     * <pre>
+     * If a benchmark, how many iterations it was run for
+     * </pre>
+     *
+     * <code>int64 iters = 2;</code>
+     */
+    public long getIters() {
+      return iters_;
+    }
+    /**
+     * <pre>
+     * If a benchmark, how many iterations it was run for
+     * </pre>
+     *
+     * <code>int64 iters = 2;</code>
+     */
+    public Builder setIters(long value) {
+      
+      iters_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * If a benchmark, how many iterations it was run for
+     * </pre>
+     *
+     * <code>int64 iters = 2;</code>
+     */
+    public Builder clearIters() {
+      
+      iters_ = 0L;
+      onChanged();
+      return this;
+    }
+
+    private double cpuTime_ ;
+    /**
+     * <pre>
+     * Total cpu time used for all iterations (in seconds)
+     * </pre>
+     *
+     * <code>double cpu_time = 3;</code>
+     */
+    public double getCpuTime() {
+      return cpuTime_;
+    }
+    /**
+     * <pre>
+     * Total cpu time used for all iterations (in seconds)
+     * </pre>
+     *
+     * <code>double cpu_time = 3;</code>
+     */
+    public Builder setCpuTime(double value) {
+      
+      cpuTime_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Total cpu time used for all iterations (in seconds)
+     * </pre>
+     *
+     * <code>double cpu_time = 3;</code>
+     */
+    public Builder clearCpuTime() {
+      
+      cpuTime_ = 0D;
+      onChanged();
+      return this;
+    }
+
+    private double wallTime_ ;
+    /**
+     * <pre>
+     * Total wall time used for all iterations (in seconds)
+     * </pre>
+     *
+     * <code>double wall_time = 4;</code>
+     */
+    public double getWallTime() {
+      return wallTime_;
+    }
+    /**
+     * <pre>
+     * Total wall time used for all iterations (in seconds)
+     * </pre>
+     *
+     * <code>double wall_time = 4;</code>
+     */
+    public Builder setWallTime(double value) {
+      
+      wallTime_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Total wall time used for all iterations (in seconds)
+     * </pre>
+     *
+     * <code>double wall_time = 4;</code>
+     */
+    public Builder clearWallTime() {
+      
+      wallTime_ = 0D;
+      onChanged();
+      return this;
+    }
+
+    private double throughput_ ;
+    /**
+     * <pre>
+     * Throughput (in MB/s)
+     * </pre>
+     *
+     * <code>double throughput = 5;</code>
+     */
+    public double getThroughput() {
+      return throughput_;
+    }
+    /**
+     * <pre>
+     * Throughput (in MB/s)
+     * </pre>
+     *
+     * <code>double throughput = 5;</code>
+     */
+    public Builder setThroughput(double value) {
+      
+      throughput_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Throughput (in MB/s)
+     * </pre>
+     *
+     * <code>double throughput = 5;</code>
+     */
+    public Builder clearThroughput() {
+      
+      throughput_ = 0D;
+      onChanged();
+      return this;
+    }
+
+    private com.google.protobuf.MapField<
+        java.lang.String, org.tensorflow.proto.util.testlog.EntryValue> extras_;
+    private com.google.protobuf.MapField<java.lang.String, org.tensorflow.proto.util.testlog.EntryValue>
+    internalGetExtras() {
+      if (extras_ == null) {
+        return com.google.protobuf.MapField.emptyMapField(
+            ExtrasDefaultEntryHolder.defaultEntry);
+      }
+      return extras_;
+    }
+    private com.google.protobuf.MapField<java.lang.String, org.tensorflow.proto.util.testlog.EntryValue>
+    internalGetMutableExtras() {
+      onChanged();;
+      if (extras_ == null) {
+        extras_ = com.google.protobuf.MapField.newMapField(
+            ExtrasDefaultEntryHolder.defaultEntry);
+      }
+      if (!extras_.isMutable()) {
+        extras_ = extras_.copy();
+      }
+      return extras_;
+    }
+
+    public int getExtrasCount() {
+      return internalGetExtras().getMap().size();
+    }
+    /**
+     * <pre>
+     * Generic map from result key to value.
+     * </pre>
+     *
+     * <code>map&lt;string, .tensorflow.EntryValue&gt; extras = 6;</code>
+     */
+
+    public boolean containsExtras(
+        java.lang.String key) {
+      if (key == null) { throw new java.lang.NullPointerException(); }
+      return internalGetExtras().getMap().containsKey(key);
+    }
+    /**
+     * Use {@link #getExtrasMap()} instead.
+     */
+    @java.lang.Deprecated
+    public java.util.Map<java.lang.String, org.tensorflow.proto.util.testlog.EntryValue> getExtras() {
+      return getExtrasMap();
+    }
+    /**
+     * <pre>
+     * Generic map from result key to value.
+     * </pre>
+     *
+     * <code>map&lt;string, .tensorflow.EntryValue&gt; extras = 6;</code>
+     */
+
+    public java.util.Map<java.lang.String, org.tensorflow.proto.util.testlog.EntryValue> getExtrasMap() {
+      return internalGetExtras().getMap();
+    }
+    /**
+     * <pre>
+     * Generic map from result key to value.
+     * </pre>
+     *
+     * <code>map&lt;string, .tensorflow.EntryValue&gt; extras = 6;</code>
+     */
+
+    public org.tensorflow.proto.util.testlog.EntryValue getExtrasOrDefault(
+        java.lang.String key,
+        org.tensorflow.proto.util.testlog.EntryValue defaultValue) {
+      if (key == null) { throw new java.lang.NullPointerException(); }
+      java.util.Map<java.lang.String, org.tensorflow.proto.util.testlog.EntryValue> map =
+          internalGetExtras().getMap();
+      return map.containsKey(key) ? map.get(key) : defaultValue;
+    }
+    /**
+     * <pre>
+     * Generic map from result key to value.
+     * </pre>
+     *
+     * <code>map&lt;string, .tensorflow.EntryValue&gt; extras = 6;</code>
+     */
+
+    public org.tensorflow.proto.util.testlog.EntryValue getExtrasOrThrow(
+        java.lang.String key) {
+      if (key == null) { throw new java.lang.NullPointerException(); }
+      java.util.Map<java.lang.String, org.tensorflow.proto.util.testlog.EntryValue> map =
+          internalGetExtras().getMap();
+      if (!map.containsKey(key)) {
+        throw new java.lang.IllegalArgumentException();
+      }
+      return map.get(key);
+    }
+
+    public Builder clearExtras() {
+      internalGetMutableExtras().getMutableMap()
+          .clear();
+      return this;
+    }
+    /**
+     * <pre>
+     * Generic map from result key to value.
+     * </pre>
+     *
+     * <code>map&lt;string, .tensorflow.EntryValue&gt; extras = 6;</code>
+     */
+
+    public Builder removeExtras(
+        java.lang.String key) {
+      if (key == null) { throw new java.lang.NullPointerException(); }
+      internalGetMutableExtras().getMutableMap()
+          .remove(key);
+      return this;
+    }
+    /**
+     * Use alternate mutation accessors instead.
+     */
+    @java.lang.Deprecated
+    public java.util.Map<java.lang.String, org.tensorflow.proto.util.testlog.EntryValue>
+    getMutableExtras() {
+      return internalGetMutableExtras().getMutableMap();
+    }
+    /**
+     * <pre>
+     * Generic map from result key to value.
+     * </pre>
+     *
+     * <code>map&lt;string, .tensorflow.EntryValue&gt; extras = 6;</code>
+     */
+    public Builder putExtras(
+        java.lang.String key,
+        org.tensorflow.proto.util.testlog.EntryValue value) {
+      if (key == null) { throw new java.lang.NullPointerException(); }
+      if (value == null) { throw new java.lang.NullPointerException(); }
+      internalGetMutableExtras().getMutableMap()
+          .put(key, value);
+      return this;
+    }
+    /**
+     * <pre>
+     * Generic map from result key to value.
+     * </pre>
+     *
+     * <code>map&lt;string, .tensorflow.EntryValue&gt; extras = 6;</code>
+     */
+
+    public Builder putAllExtras(
+        java.util.Map<java.lang.String, org.tensorflow.proto.util.testlog.EntryValue> values) {
+      internalGetMutableExtras().getMutableMap()
+          .putAll(values);
+      return this;
+    }
+
+    private java.util.List<org.tensorflow.proto.util.testlog.MetricEntry> metrics_ =
+      java.util.Collections.emptyList();
+    private void ensureMetricsIsMutable() {
+      if (!((bitField0_ & 0x00000002) != 0)) {
+        metrics_ = new java.util.ArrayList<org.tensorflow.proto.util.testlog.MetricEntry>(metrics_);
+        bitField0_ |= 0x00000002;
+       }
+    }
+
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.util.testlog.MetricEntry, org.tensorflow.proto.util.testlog.MetricEntry.Builder, org.tensorflow.proto.util.testlog.MetricEntryOrBuilder> metricsBuilder_;
+
+    /**
+     * <pre>
+     * Metric name, value and expected range. This can include accuracy metrics
+     * typically used to determine whether the accuracy test has passed
+     * </pre>
+     *
+     * <code>repeated .tensorflow.MetricEntry metrics = 7;</code>
+     */
+    public java.util.List<org.tensorflow.proto.util.testlog.MetricEntry> getMetricsList() {
+      if (metricsBuilder_ == null) {
+        return java.util.Collections.unmodifiableList(metrics_);
+      } else {
+        return metricsBuilder_.getMessageList();
+      }
+    }
+    /**
+     * <pre>
+     * Metric name, value and expected range. This can include accuracy metrics
+     * typically used to determine whether the accuracy test has passed
+     * </pre>
+     *
+     * <code>repeated .tensorflow.MetricEntry metrics = 7;</code>
+     */
+    public int getMetricsCount() {
+      if (metricsBuilder_ == null) {
+        return metrics_.size();
+      } else {
+        return metricsBuilder_.getCount();
+      }
+    }
+    /**
+     * <pre>
+     * Metric name, value and expected range. This can include accuracy metrics
+     * typically used to determine whether the accuracy test has passed
+     * </pre>
+     *
+     * <code>repeated .tensorflow.MetricEntry metrics = 7;</code>
+     */
+    public org.tensorflow.proto.util.testlog.MetricEntry getMetrics(int index) {
+      if (metricsBuilder_ == null) {
+        return metrics_.get(index);
+      } else {
+        return metricsBuilder_.getMessage(index);
+      }
+    }
+    /**
+     * <pre>
+     * Metric name, value and expected range. This can include accuracy metrics
+     * typically used to determine whether the accuracy test has passed
+     * </pre>
+     *
+     * <code>repeated .tensorflow.MetricEntry metrics = 7;</code>
+     */
+    public Builder setMetrics(
+        int index, org.tensorflow.proto.util.testlog.MetricEntry value) {
+      if (metricsBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureMetricsIsMutable();
+        metrics_.set(index, value);
+        onChanged();
+      } else {
+        metricsBuilder_.setMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Metric name, value and expected range. This can include accuracy metrics
+     * typically used to determine whether the accuracy test has passed
+     * </pre>
+     *
+     * <code>repeated .tensorflow.MetricEntry metrics = 7;</code>
+     */
+    public Builder setMetrics(
+        int index, org.tensorflow.proto.util.testlog.MetricEntry.Builder builderForValue) {
+      if (metricsBuilder_ == null) {
+        ensureMetricsIsMutable();
+        metrics_.set(index, builderForValue.build());
+        onChanged();
+      } else {
+        metricsBuilder_.setMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Metric name, value and expected range. This can include accuracy metrics
+     * typically used to determine whether the accuracy test has passed
+     * </pre>
+     *
+     * <code>repeated .tensorflow.MetricEntry metrics = 7;</code>
+     */
+    public Builder addMetrics(org.tensorflow.proto.util.testlog.MetricEntry value) {
+      if (metricsBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureMetricsIsMutable();
+        metrics_.add(value);
+        onChanged();
+      } else {
+        metricsBuilder_.addMessage(value);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Metric name, value and expected range. This can include accuracy metrics
+     * typically used to determine whether the accuracy test has passed
+     * </pre>
+     *
+     * <code>repeated .tensorflow.MetricEntry metrics = 7;</code>
+     */
+    public Builder addMetrics(
+        int index, org.tensorflow.proto.util.testlog.MetricEntry value) {
+      if (metricsBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureMetricsIsMutable();
+        metrics_.add(index, value);
+        onChanged();
+      } else {
+        metricsBuilder_.addMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Metric name, value and expected range. This can include accuracy metrics
+     * typically used to determine whether the accuracy test has passed
+     * </pre>
+     *
+     * <code>repeated .tensorflow.MetricEntry metrics = 7;</code>
+     */
+    public Builder addMetrics(
+        org.tensorflow.proto.util.testlog.MetricEntry.Builder builderForValue) {
+      if (metricsBuilder_ == null) {
+        ensureMetricsIsMutable();
+        metrics_.add(builderForValue.build());
+        onChanged();
+      } else {
+        metricsBuilder_.addMessage(builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Metric name, value and expected range. This can include accuracy metrics
+     * typically used to determine whether the accuracy test has passed
+     * </pre>
+     *
+     * <code>repeated .tensorflow.MetricEntry metrics = 7;</code>
+     */
+    public Builder addMetrics(
+        int index, org.tensorflow.proto.util.testlog.MetricEntry.Builder builderForValue) {
+      if (metricsBuilder_ == null) {
+        ensureMetricsIsMutable();
+        metrics_.add(index, builderForValue.build());
+        onChanged();
+      } else {
+        metricsBuilder_.addMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Metric name, value and expected range. This can include accuracy metrics
+     * typically used to determine whether the accuracy test has passed
+     * </pre>
+     *
+     * <code>repeated .tensorflow.MetricEntry metrics = 7;</code>
+     */
+    public Builder addAllMetrics(
+        java.lang.Iterable<? extends org.tensorflow.proto.util.testlog.MetricEntry> values) {
+      if (metricsBuilder_ == null) {
+        ensureMetricsIsMutable();
+        com.google.protobuf.AbstractMessageLite.Builder.addAll(
+            values, metrics_);
+        onChanged();
+      } else {
+        metricsBuilder_.addAllMessages(values);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Metric name, value and expected range. This can include accuracy metrics
+     * typically used to determine whether the accuracy test has passed
+     * </pre>
+     *
+     * <code>repeated .tensorflow.MetricEntry metrics = 7;</code>
+     */
+    public Builder clearMetrics() {
+      if (metricsBuilder_ == null) {
+        metrics_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000002);
+        onChanged();
+      } else {
+        metricsBuilder_.clear();
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Metric name, value and expected range. This can include accuracy metrics
+     * typically used to determine whether the accuracy test has passed
+     * </pre>
+     *
+     * <code>repeated .tensorflow.MetricEntry metrics = 7;</code>
+     */
+    public Builder removeMetrics(int index) {
+      if (metricsBuilder_ == null) {
+        ensureMetricsIsMutable();
+        metrics_.remove(index);
+        onChanged();
+      } else {
+        metricsBuilder_.remove(index);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Metric name, value and expected range. This can include accuracy metrics
+     * typically used to determine whether the accuracy test has passed
+     * </pre>
+     *
+     * <code>repeated .tensorflow.MetricEntry metrics = 7;</code>
+     */
+    public org.tensorflow.proto.util.testlog.MetricEntry.Builder getMetricsBuilder(
+        int index) {
+      return getMetricsFieldBuilder().getBuilder(index);
+    }
+    /**
+     * <pre>
+     * Metric name, value and expected range. This can include accuracy metrics
+     * typically used to determine whether the accuracy test has passed
+     * </pre>
+     *
+     * <code>repeated .tensorflow.MetricEntry metrics = 7;</code>
+     */
+    public org.tensorflow.proto.util.testlog.MetricEntryOrBuilder getMetricsOrBuilder(
+        int index) {
+      if (metricsBuilder_ == null) {
+        return metrics_.get(index);  } else {
+        return metricsBuilder_.getMessageOrBuilder(index);
+      }
+    }
+    /**
+     * <pre>
+     * Metric name, value and expected range. This can include accuracy metrics
+     * typically used to determine whether the accuracy test has passed
+     * </pre>
+     *
+     * <code>repeated .tensorflow.MetricEntry metrics = 7;</code>
+     */
+    public java.util.List<? extends org.tensorflow.proto.util.testlog.MetricEntryOrBuilder> 
+         getMetricsOrBuilderList() {
+      if (metricsBuilder_ != null) {
+        return metricsBuilder_.getMessageOrBuilderList();
+      } else {
+        return java.util.Collections.unmodifiableList(metrics_);
+      }
+    }
+    /**
+     * <pre>
+     * Metric name, value and expected range. This can include accuracy metrics
+     * typically used to determine whether the accuracy test has passed
+     * </pre>
+     *
+     * <code>repeated .tensorflow.MetricEntry metrics = 7;</code>
+     */
+    public org.tensorflow.proto.util.testlog.MetricEntry.Builder addMetricsBuilder() {
+      return getMetricsFieldBuilder().addBuilder(
+          org.tensorflow.proto.util.testlog.MetricEntry.getDefaultInstance());
+    }
+    /**
+     * <pre>
+     * Metric name, value and expected range. This can include accuracy metrics
+     * typically used to determine whether the accuracy test has passed
+     * </pre>
+     *
+     * <code>repeated .tensorflow.MetricEntry metrics = 7;</code>
+     */
+    public org.tensorflow.proto.util.testlog.MetricEntry.Builder addMetricsBuilder(
+        int index) {
+      return getMetricsFieldBuilder().addBuilder(
+          index, org.tensorflow.proto.util.testlog.MetricEntry.getDefaultInstance());
+    }
+    /**
+     * <pre>
+     * Metric name, value and expected range. This can include accuracy metrics
+     * typically used to determine whether the accuracy test has passed
+     * </pre>
+     *
+     * <code>repeated .tensorflow.MetricEntry metrics = 7;</code>
+     */
+    public java.util.List<org.tensorflow.proto.util.testlog.MetricEntry.Builder> 
+         getMetricsBuilderList() {
+      return getMetricsFieldBuilder().getBuilderList();
+    }
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.util.testlog.MetricEntry, org.tensorflow.proto.util.testlog.MetricEntry.Builder, org.tensorflow.proto.util.testlog.MetricEntryOrBuilder> 
+        getMetricsFieldBuilder() {
+      if (metricsBuilder_ == null) {
+        metricsBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
+            org.tensorflow.proto.util.testlog.MetricEntry, org.tensorflow.proto.util.testlog.MetricEntry.Builder, org.tensorflow.proto.util.testlog.MetricEntryOrBuilder>(
+                metrics_,
+                ((bitField0_ & 0x00000002) != 0),
+                getParentForChildren(),
+                isClean());
+        metrics_ = null;
+      }
+      return metricsBuilder_;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.BenchmarkEntry)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.BenchmarkEntry)
+  private static final org.tensorflow.proto.util.testlog.BenchmarkEntry DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.util.testlog.BenchmarkEntry();
+  }
+
+  public static org.tensorflow.proto.util.testlog.BenchmarkEntry getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<BenchmarkEntry>
+      PARSER = new com.google.protobuf.AbstractParser<BenchmarkEntry>() {
+    @java.lang.Override
+    public BenchmarkEntry parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new BenchmarkEntry(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<BenchmarkEntry> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<BenchmarkEntry> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.util.testlog.BenchmarkEntry getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/BenchmarkEntryOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/BenchmarkEntryOrBuilder.java
new file mode 100644
index 00000000000..475b3e31b7c
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/BenchmarkEntryOrBuilder.java
@@ -0,0 +1,168 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/util/test_log.proto
+
+package org.tensorflow.proto.util.testlog;
+
+public interface BenchmarkEntryOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.BenchmarkEntry)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <pre>
+   * The name of the specific benchmark or test
+   * (e.g. BM_AdjustContrast_gpu_B_W_H)
+   * </pre>
+   *
+   * <code>string name = 1;</code>
+   */
+  java.lang.String getName();
+  /**
+   * <pre>
+   * The name of the specific benchmark or test
+   * (e.g. BM_AdjustContrast_gpu_B_W_H)
+   * </pre>
+   *
+   * <code>string name = 1;</code>
+   */
+  com.google.protobuf.ByteString
+      getNameBytes();
+
+  /**
+   * <pre>
+   * If a benchmark, how many iterations it was run for
+   * </pre>
+   *
+   * <code>int64 iters = 2;</code>
+   */
+  long getIters();
+
+  /**
+   * <pre>
+   * Total cpu time used for all iterations (in seconds)
+   * </pre>
+   *
+   * <code>double cpu_time = 3;</code>
+   */
+  double getCpuTime();
+
+  /**
+   * <pre>
+   * Total wall time used for all iterations (in seconds)
+   * </pre>
+   *
+   * <code>double wall_time = 4;</code>
+   */
+  double getWallTime();
+
+  /**
+   * <pre>
+   * Throughput (in MB/s)
+   * </pre>
+   *
+   * <code>double throughput = 5;</code>
+   */
+  double getThroughput();
+
+  /**
+   * <pre>
+   * Generic map from result key to value.
+   * </pre>
+   *
+   * <code>map&lt;string, .tensorflow.EntryValue&gt; extras = 6;</code>
+   */
+  int getExtrasCount();
+  /**
+   * <pre>
+   * Generic map from result key to value.
+   * </pre>
+   *
+   * <code>map&lt;string, .tensorflow.EntryValue&gt; extras = 6;</code>
+   */
+  boolean containsExtras(
+      java.lang.String key);
+  /**
+   * Use {@link #getExtrasMap()} instead.
+   */
+  @java.lang.Deprecated
+  java.util.Map<java.lang.String, org.tensorflow.proto.util.testlog.EntryValue>
+  getExtras();
+  /**
+   * <pre>
+   * Generic map from result key to value.
+   * </pre>
+   *
+   * <code>map&lt;string, .tensorflow.EntryValue&gt; extras = 6;</code>
+   */
+  java.util.Map<java.lang.String, org.tensorflow.proto.util.testlog.EntryValue>
+  getExtrasMap();
+  /**
+   * <pre>
+   * Generic map from result key to value.
+   * </pre>
+   *
+   * <code>map&lt;string, .tensorflow.EntryValue&gt; extras = 6;</code>
+   */
+
+  org.tensorflow.proto.util.testlog.EntryValue getExtrasOrDefault(
+      java.lang.String key,
+      org.tensorflow.proto.util.testlog.EntryValue defaultValue);
+  /**
+   * <pre>
+   * Generic map from result key to value.
+   * </pre>
+   *
+   * <code>map&lt;string, .tensorflow.EntryValue&gt; extras = 6;</code>
+   */
+
+  org.tensorflow.proto.util.testlog.EntryValue getExtrasOrThrow(
+      java.lang.String key);
+
+  /**
+   * <pre>
+   * Metric name, value and expected range. This can include accuracy metrics
+   * typically used to determine whether the accuracy test has passed
+   * </pre>
+   *
+   * <code>repeated .tensorflow.MetricEntry metrics = 7;</code>
+   */
+  java.util.List<org.tensorflow.proto.util.testlog.MetricEntry> 
+      getMetricsList();
+  /**
+   * <pre>
+   * Metric name, value and expected range. This can include accuracy metrics
+   * typically used to determine whether the accuracy test has passed
+   * </pre>
+   *
+   * <code>repeated .tensorflow.MetricEntry metrics = 7;</code>
+   */
+  org.tensorflow.proto.util.testlog.MetricEntry getMetrics(int index);
+  /**
+   * <pre>
+   * Metric name, value and expected range. This can include accuracy metrics
+   * typically used to determine whether the accuracy test has passed
+   * </pre>
+   *
+   * <code>repeated .tensorflow.MetricEntry metrics = 7;</code>
+   */
+  int getMetricsCount();
+  /**
+   * <pre>
+   * Metric name, value and expected range. This can include accuracy metrics
+   * typically used to determine whether the accuracy test has passed
+   * </pre>
+   *
+   * <code>repeated .tensorflow.MetricEntry metrics = 7;</code>
+   */
+  java.util.List<? extends org.tensorflow.proto.util.testlog.MetricEntryOrBuilder> 
+      getMetricsOrBuilderList();
+  /**
+   * <pre>
+   * Metric name, value and expected range. This can include accuracy metrics
+   * typically used to determine whether the accuracy test has passed
+   * </pre>
+   *
+   * <code>repeated .tensorflow.MetricEntry metrics = 7;</code>
+   */
+  org.tensorflow.proto.util.testlog.MetricEntryOrBuilder getMetricsOrBuilder(
+      int index);
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/BuildConfiguration.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/BuildConfiguration.java
new file mode 100644
index 00000000000..8d571429f4f
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/BuildConfiguration.java
@@ -0,0 +1,1021 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/util/test_log.proto
+
+package org.tensorflow.proto.util.testlog;
+
+/**
+ * Protobuf type {@code tensorflow.BuildConfiguration}
+ */
+public  final class BuildConfiguration extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.BuildConfiguration)
+    BuildConfigurationOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use BuildConfiguration.newBuilder() to construct.
+  private BuildConfiguration(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private BuildConfiguration() {
+    mode_ = "";
+    ccFlags_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+    opts_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new BuildConfiguration();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private BuildConfiguration(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    int mutable_bitField0_ = 0;
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 10: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            mode_ = s;
+            break;
+          }
+          case 18: {
+            java.lang.String s = input.readStringRequireUtf8();
+            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
+              ccFlags_ = new com.google.protobuf.LazyStringArrayList();
+              mutable_bitField0_ |= 0x00000001;
+            }
+            ccFlags_.add(s);
+            break;
+          }
+          case 26: {
+            java.lang.String s = input.readStringRequireUtf8();
+            if (!((mutable_bitField0_ & 0x00000002) != 0)) {
+              opts_ = new com.google.protobuf.LazyStringArrayList();
+              mutable_bitField0_ |= 0x00000002;
+            }
+            opts_.add(s);
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      if (((mutable_bitField0_ & 0x00000001) != 0)) {
+        ccFlags_ = ccFlags_.getUnmodifiableView();
+      }
+      if (((mutable_bitField0_ & 0x00000002) != 0)) {
+        opts_ = opts_.getUnmodifiableView();
+      }
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_BuildConfiguration_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_BuildConfiguration_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.util.testlog.BuildConfiguration.class, org.tensorflow.proto.util.testlog.BuildConfiguration.Builder.class);
+  }
+
+  public static final int MODE_FIELD_NUMBER = 1;
+  private volatile java.lang.Object mode_;
+  /**
+   * <pre>
+   * opt, dbg, etc
+   * </pre>
+   *
+   * <code>string mode = 1;</code>
+   */
+  public java.lang.String getMode() {
+    java.lang.Object ref = mode_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      mode_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * opt, dbg, etc
+   * </pre>
+   *
+   * <code>string mode = 1;</code>
+   */
+  public com.google.protobuf.ByteString
+      getModeBytes() {
+    java.lang.Object ref = mode_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      mode_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int CC_FLAGS_FIELD_NUMBER = 2;
+  private com.google.protobuf.LazyStringList ccFlags_;
+  /**
+   * <pre>
+   * CC compiler flags, if known
+   * </pre>
+   *
+   * <code>repeated string cc_flags = 2;</code>
+   */
+  public com.google.protobuf.ProtocolStringList
+      getCcFlagsList() {
+    return ccFlags_;
+  }
+  /**
+   * <pre>
+   * CC compiler flags, if known
+   * </pre>
+   *
+   * <code>repeated string cc_flags = 2;</code>
+   */
+  public int getCcFlagsCount() {
+    return ccFlags_.size();
+  }
+  /**
+   * <pre>
+   * CC compiler flags, if known
+   * </pre>
+   *
+   * <code>repeated string cc_flags = 2;</code>
+   */
+  public java.lang.String getCcFlags(int index) {
+    return ccFlags_.get(index);
+  }
+  /**
+   * <pre>
+   * CC compiler flags, if known
+   * </pre>
+   *
+   * <code>repeated string cc_flags = 2;</code>
+   */
+  public com.google.protobuf.ByteString
+      getCcFlagsBytes(int index) {
+    return ccFlags_.getByteString(index);
+  }
+
+  public static final int OPTS_FIELD_NUMBER = 3;
+  private com.google.protobuf.LazyStringList opts_;
+  /**
+   * <pre>
+   * Bazel compilation options, if known
+   * </pre>
+   *
+   * <code>repeated string opts = 3;</code>
+   */
+  public com.google.protobuf.ProtocolStringList
+      getOptsList() {
+    return opts_;
+  }
+  /**
+   * <pre>
+   * Bazel compilation options, if known
+   * </pre>
+   *
+   * <code>repeated string opts = 3;</code>
+   */
+  public int getOptsCount() {
+    return opts_.size();
+  }
+  /**
+   * <pre>
+   * Bazel compilation options, if known
+   * </pre>
+   *
+   * <code>repeated string opts = 3;</code>
+   */
+  public java.lang.String getOpts(int index) {
+    return opts_.get(index);
+  }
+  /**
+   * <pre>
+   * Bazel compilation options, if known
+   * </pre>
+   *
+   * <code>repeated string opts = 3;</code>
+   */
+  public com.google.protobuf.ByteString
+      getOptsBytes(int index) {
+    return opts_.getByteString(index);
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    if (!getModeBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 1, mode_);
+    }
+    for (int i = 0; i < ccFlags_.size(); i++) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 2, ccFlags_.getRaw(i));
+    }
+    for (int i = 0; i < opts_.size(); i++) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 3, opts_.getRaw(i));
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (!getModeBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, mode_);
+    }
+    {
+      int dataSize = 0;
+      for (int i = 0; i < ccFlags_.size(); i++) {
+        dataSize += computeStringSizeNoTag(ccFlags_.getRaw(i));
+      }
+      size += dataSize;
+      size += 1 * getCcFlagsList().size();
+    }
+    {
+      int dataSize = 0;
+      for (int i = 0; i < opts_.size(); i++) {
+        dataSize += computeStringSizeNoTag(opts_.getRaw(i));
+      }
+      size += dataSize;
+      size += 1 * getOptsList().size();
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.util.testlog.BuildConfiguration)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.util.testlog.BuildConfiguration other = (org.tensorflow.proto.util.testlog.BuildConfiguration) obj;
+
+    if (!getMode()
+        .equals(other.getMode())) return false;
+    if (!getCcFlagsList()
+        .equals(other.getCcFlagsList())) return false;
+    if (!getOptsList()
+        .equals(other.getOptsList())) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    hash = (37 * hash) + MODE_FIELD_NUMBER;
+    hash = (53 * hash) + getMode().hashCode();
+    if (getCcFlagsCount() > 0) {
+      hash = (37 * hash) + CC_FLAGS_FIELD_NUMBER;
+      hash = (53 * hash) + getCcFlagsList().hashCode();
+    }
+    if (getOptsCount() > 0) {
+      hash = (37 * hash) + OPTS_FIELD_NUMBER;
+      hash = (53 * hash) + getOptsList().hashCode();
+    }
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.util.testlog.BuildConfiguration parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.testlog.BuildConfiguration parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.testlog.BuildConfiguration parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.testlog.BuildConfiguration parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.testlog.BuildConfiguration parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.testlog.BuildConfiguration parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.testlog.BuildConfiguration parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.testlog.BuildConfiguration parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.testlog.BuildConfiguration parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.testlog.BuildConfiguration parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.testlog.BuildConfiguration parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.testlog.BuildConfiguration parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.util.testlog.BuildConfiguration prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * Protobuf type {@code tensorflow.BuildConfiguration}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.BuildConfiguration)
+      org.tensorflow.proto.util.testlog.BuildConfigurationOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_BuildConfiguration_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_BuildConfiguration_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.util.testlog.BuildConfiguration.class, org.tensorflow.proto.util.testlog.BuildConfiguration.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.util.testlog.BuildConfiguration.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      mode_ = "";
+
+      ccFlags_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+      bitField0_ = (bitField0_ & ~0x00000001);
+      opts_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+      bitField0_ = (bitField0_ & ~0x00000002);
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_BuildConfiguration_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.testlog.BuildConfiguration getDefaultInstanceForType() {
+      return org.tensorflow.proto.util.testlog.BuildConfiguration.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.testlog.BuildConfiguration build() {
+      org.tensorflow.proto.util.testlog.BuildConfiguration result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.testlog.BuildConfiguration buildPartial() {
+      org.tensorflow.proto.util.testlog.BuildConfiguration result = new org.tensorflow.proto.util.testlog.BuildConfiguration(this);
+      int from_bitField0_ = bitField0_;
+      result.mode_ = mode_;
+      if (((bitField0_ & 0x00000001) != 0)) {
+        ccFlags_ = ccFlags_.getUnmodifiableView();
+        bitField0_ = (bitField0_ & ~0x00000001);
+      }
+      result.ccFlags_ = ccFlags_;
+      if (((bitField0_ & 0x00000002) != 0)) {
+        opts_ = opts_.getUnmodifiableView();
+        bitField0_ = (bitField0_ & ~0x00000002);
+      }
+      result.opts_ = opts_;
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.util.testlog.BuildConfiguration) {
+        return mergeFrom((org.tensorflow.proto.util.testlog.BuildConfiguration)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.util.testlog.BuildConfiguration other) {
+      if (other == org.tensorflow.proto.util.testlog.BuildConfiguration.getDefaultInstance()) return this;
+      if (!other.getMode().isEmpty()) {
+        mode_ = other.mode_;
+        onChanged();
+      }
+      if (!other.ccFlags_.isEmpty()) {
+        if (ccFlags_.isEmpty()) {
+          ccFlags_ = other.ccFlags_;
+          bitField0_ = (bitField0_ & ~0x00000001);
+        } else {
+          ensureCcFlagsIsMutable();
+          ccFlags_.addAll(other.ccFlags_);
+        }
+        onChanged();
+      }
+      if (!other.opts_.isEmpty()) {
+        if (opts_.isEmpty()) {
+          opts_ = other.opts_;
+          bitField0_ = (bitField0_ & ~0x00000002);
+        } else {
+          ensureOptsIsMutable();
+          opts_.addAll(other.opts_);
+        }
+        onChanged();
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.util.testlog.BuildConfiguration parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.util.testlog.BuildConfiguration) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+    private int bitField0_;
+
+    private java.lang.Object mode_ = "";
+    /**
+     * <pre>
+     * opt, dbg, etc
+     * </pre>
+     *
+     * <code>string mode = 1;</code>
+     */
+    public java.lang.String getMode() {
+      java.lang.Object ref = mode_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        mode_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * opt, dbg, etc
+     * </pre>
+     *
+     * <code>string mode = 1;</code>
+     */
+    public com.google.protobuf.ByteString
+        getModeBytes() {
+      java.lang.Object ref = mode_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        mode_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * opt, dbg, etc
+     * </pre>
+     *
+     * <code>string mode = 1;</code>
+     */
+    public Builder setMode(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      mode_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * opt, dbg, etc
+     * </pre>
+     *
+     * <code>string mode = 1;</code>
+     */
+    public Builder clearMode() {
+      
+      mode_ = getDefaultInstance().getMode();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * opt, dbg, etc
+     * </pre>
+     *
+     * <code>string mode = 1;</code>
+     */
+    public Builder setModeBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      mode_ = value;
+      onChanged();
+      return this;
+    }
+
+    private com.google.protobuf.LazyStringList ccFlags_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+    private void ensureCcFlagsIsMutable() {
+      if (!((bitField0_ & 0x00000001) != 0)) {
+        ccFlags_ = new com.google.protobuf.LazyStringArrayList(ccFlags_);
+        bitField0_ |= 0x00000001;
+       }
+    }
+    /**
+     * <pre>
+     * CC compiler flags, if known
+     * </pre>
+     *
+     * <code>repeated string cc_flags = 2;</code>
+     */
+    public com.google.protobuf.ProtocolStringList
+        getCcFlagsList() {
+      return ccFlags_.getUnmodifiableView();
+    }
+    /**
+     * <pre>
+     * CC compiler flags, if known
+     * </pre>
+     *
+     * <code>repeated string cc_flags = 2;</code>
+     */
+    public int getCcFlagsCount() {
+      return ccFlags_.size();
+    }
+    /**
+     * <pre>
+     * CC compiler flags, if known
+     * </pre>
+     *
+     * <code>repeated string cc_flags = 2;</code>
+     */
+    public java.lang.String getCcFlags(int index) {
+      return ccFlags_.get(index);
+    }
+    /**
+     * <pre>
+     * CC compiler flags, if known
+     * </pre>
+     *
+     * <code>repeated string cc_flags = 2;</code>
+     */
+    public com.google.protobuf.ByteString
+        getCcFlagsBytes(int index) {
+      return ccFlags_.getByteString(index);
+    }
+    /**
+     * <pre>
+     * CC compiler flags, if known
+     * </pre>
+     *
+     * <code>repeated string cc_flags = 2;</code>
+     */
+    public Builder setCcFlags(
+        int index, java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  ensureCcFlagsIsMutable();
+      ccFlags_.set(index, value);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * CC compiler flags, if known
+     * </pre>
+     *
+     * <code>repeated string cc_flags = 2;</code>
+     */
+    public Builder addCcFlags(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  ensureCcFlagsIsMutable();
+      ccFlags_.add(value);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * CC compiler flags, if known
+     * </pre>
+     *
+     * <code>repeated string cc_flags = 2;</code>
+     */
+    public Builder addAllCcFlags(
+        java.lang.Iterable<java.lang.String> values) {
+      ensureCcFlagsIsMutable();
+      com.google.protobuf.AbstractMessageLite.Builder.addAll(
+          values, ccFlags_);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * CC compiler flags, if known
+     * </pre>
+     *
+     * <code>repeated string cc_flags = 2;</code>
+     */
+    public Builder clearCcFlags() {
+      ccFlags_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+      bitField0_ = (bitField0_ & ~0x00000001);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * CC compiler flags, if known
+     * </pre>
+     *
+     * <code>repeated string cc_flags = 2;</code>
+     */
+    public Builder addCcFlagsBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      ensureCcFlagsIsMutable();
+      ccFlags_.add(value);
+      onChanged();
+      return this;
+    }
+
+    private com.google.protobuf.LazyStringList opts_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+    private void ensureOptsIsMutable() {
+      if (!((bitField0_ & 0x00000002) != 0)) {
+        opts_ = new com.google.protobuf.LazyStringArrayList(opts_);
+        bitField0_ |= 0x00000002;
+       }
+    }
+    /**
+     * <pre>
+     * Bazel compilation options, if known
+     * </pre>
+     *
+     * <code>repeated string opts = 3;</code>
+     */
+    public com.google.protobuf.ProtocolStringList
+        getOptsList() {
+      return opts_.getUnmodifiableView();
+    }
+    /**
+     * <pre>
+     * Bazel compilation options, if known
+     * </pre>
+     *
+     * <code>repeated string opts = 3;</code>
+     */
+    public int getOptsCount() {
+      return opts_.size();
+    }
+    /**
+     * <pre>
+     * Bazel compilation options, if known
+     * </pre>
+     *
+     * <code>repeated string opts = 3;</code>
+     */
+    public java.lang.String getOpts(int index) {
+      return opts_.get(index);
+    }
+    /**
+     * <pre>
+     * Bazel compilation options, if known
+     * </pre>
+     *
+     * <code>repeated string opts = 3;</code>
+     */
+    public com.google.protobuf.ByteString
+        getOptsBytes(int index) {
+      return opts_.getByteString(index);
+    }
+    /**
+     * <pre>
+     * Bazel compilation options, if known
+     * </pre>
+     *
+     * <code>repeated string opts = 3;</code>
+     */
+    public Builder setOpts(
+        int index, java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  ensureOptsIsMutable();
+      opts_.set(index, value);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Bazel compilation options, if known
+     * </pre>
+     *
+     * <code>repeated string opts = 3;</code>
+     */
+    public Builder addOpts(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  ensureOptsIsMutable();
+      opts_.add(value);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Bazel compilation options, if known
+     * </pre>
+     *
+     * <code>repeated string opts = 3;</code>
+     */
+    public Builder addAllOpts(
+        java.lang.Iterable<java.lang.String> values) {
+      ensureOptsIsMutable();
+      com.google.protobuf.AbstractMessageLite.Builder.addAll(
+          values, opts_);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Bazel compilation options, if known
+     * </pre>
+     *
+     * <code>repeated string opts = 3;</code>
+     */
+    public Builder clearOpts() {
+      opts_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+      bitField0_ = (bitField0_ & ~0x00000002);
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Bazel compilation options, if known
+     * </pre>
+     *
+     * <code>repeated string opts = 3;</code>
+     */
+    public Builder addOptsBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      ensureOptsIsMutable();
+      opts_.add(value);
+      onChanged();
+      return this;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.BuildConfiguration)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.BuildConfiguration)
+  private static final org.tensorflow.proto.util.testlog.BuildConfiguration DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.util.testlog.BuildConfiguration();
+  }
+
+  public static org.tensorflow.proto.util.testlog.BuildConfiguration getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<BuildConfiguration>
+      PARSER = new com.google.protobuf.AbstractParser<BuildConfiguration>() {
+    @java.lang.Override
+    public BuildConfiguration parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new BuildConfiguration(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<BuildConfiguration> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<BuildConfiguration> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.util.testlog.BuildConfiguration getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/BuildConfigurationOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/BuildConfigurationOrBuilder.java
new file mode 100644
index 00000000000..4fd04ebf65d
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/BuildConfigurationOrBuilder.java
@@ -0,0 +1,97 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/util/test_log.proto
+
+package org.tensorflow.proto.util.testlog;
+
+public interface BuildConfigurationOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.BuildConfiguration)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <pre>
+   * opt, dbg, etc
+   * </pre>
+   *
+   * <code>string mode = 1;</code>
+   */
+  java.lang.String getMode();
+  /**
+   * <pre>
+   * opt, dbg, etc
+   * </pre>
+   *
+   * <code>string mode = 1;</code>
+   */
+  com.google.protobuf.ByteString
+      getModeBytes();
+
+  /**
+   * <pre>
+   * CC compiler flags, if known
+   * </pre>
+   *
+   * <code>repeated string cc_flags = 2;</code>
+   */
+  java.util.List<java.lang.String>
+      getCcFlagsList();
+  /**
+   * <pre>
+   * CC compiler flags, if known
+   * </pre>
+   *
+   * <code>repeated string cc_flags = 2;</code>
+   */
+  int getCcFlagsCount();
+  /**
+   * <pre>
+   * CC compiler flags, if known
+   * </pre>
+   *
+   * <code>repeated string cc_flags = 2;</code>
+   */
+  java.lang.String getCcFlags(int index);
+  /**
+   * <pre>
+   * CC compiler flags, if known
+   * </pre>
+   *
+   * <code>repeated string cc_flags = 2;</code>
+   */
+  com.google.protobuf.ByteString
+      getCcFlagsBytes(int index);
+
+  /**
+   * <pre>
+   * Bazel compilation options, if known
+   * </pre>
+   *
+   * <code>repeated string opts = 3;</code>
+   */
+  java.util.List<java.lang.String>
+      getOptsList();
+  /**
+   * <pre>
+   * Bazel compilation options, if known
+   * </pre>
+   *
+   * <code>repeated string opts = 3;</code>
+   */
+  int getOptsCount();
+  /**
+   * <pre>
+   * Bazel compilation options, if known
+   * </pre>
+   *
+   * <code>repeated string opts = 3;</code>
+   */
+  java.lang.String getOpts(int index);
+  /**
+   * <pre>
+   * Bazel compilation options, if known
+   * </pre>
+   *
+   * <code>repeated string opts = 3;</code>
+   */
+  com.google.protobuf.ByteString
+      getOptsBytes(int index);
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/CPUInfo.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/CPUInfo.java
new file mode 100644
index 00000000000..b4ae942f481
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/CPUInfo.java
@@ -0,0 +1,1254 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/util/test_log.proto
+
+package org.tensorflow.proto.util.testlog;
+
+/**
+ * Protobuf type {@code tensorflow.CPUInfo}
+ */
+public  final class CPUInfo extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.CPUInfo)
+    CPUInfoOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use CPUInfo.newBuilder() to construct.
+  private CPUInfo(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private CPUInfo() {
+    cpuInfo_ = "";
+    cpuGovernor_ = "";
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new CPUInfo();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private CPUInfo(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    int mutable_bitField0_ = 0;
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 8: {
+
+            numCores_ = input.readInt64();
+            break;
+          }
+          case 16: {
+
+            numCoresAllowed_ = input.readInt64();
+            break;
+          }
+          case 25: {
+
+            mhzPerCpu_ = input.readDouble();
+            break;
+          }
+          case 34: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            cpuInfo_ = s;
+            break;
+          }
+          case 42: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            cpuGovernor_ = s;
+            break;
+          }
+          case 50: {
+            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
+              cacheSize_ = com.google.protobuf.MapField.newMapField(
+                  CacheSizeDefaultEntryHolder.defaultEntry);
+              mutable_bitField0_ |= 0x00000001;
+            }
+            com.google.protobuf.MapEntry<java.lang.String, java.lang.Long>
+            cacheSize__ = input.readMessage(
+                CacheSizeDefaultEntryHolder.defaultEntry.getParserForType(), extensionRegistry);
+            cacheSize_.getMutableMap().put(
+                cacheSize__.getKey(), cacheSize__.getValue());
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_CPUInfo_descriptor;
+  }
+
+  @SuppressWarnings({"rawtypes"})
+  @java.lang.Override
+  protected com.google.protobuf.MapField internalGetMapField(
+      int number) {
+    switch (number) {
+      case 6:
+        return internalGetCacheSize();
+      default:
+        throw new RuntimeException(
+            "Invalid map field number: " + number);
+    }
+  }
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_CPUInfo_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.util.testlog.CPUInfo.class, org.tensorflow.proto.util.testlog.CPUInfo.Builder.class);
+  }
+
+  public static final int NUM_CORES_FIELD_NUMBER = 1;
+  private long numCores_;
+  /**
+   * <code>int64 num_cores = 1;</code>
+   */
+  public long getNumCores() {
+    return numCores_;
+  }
+
+  public static final int NUM_CORES_ALLOWED_FIELD_NUMBER = 2;
+  private long numCoresAllowed_;
+  /**
+   * <code>int64 num_cores_allowed = 2;</code>
+   */
+  public long getNumCoresAllowed() {
+    return numCoresAllowed_;
+  }
+
+  public static final int MHZ_PER_CPU_FIELD_NUMBER = 3;
+  private double mhzPerCpu_;
+  /**
+   * <pre>
+   * How fast are these cpus?
+   * </pre>
+   *
+   * <code>double mhz_per_cpu = 3;</code>
+   */
+  public double getMhzPerCpu() {
+    return mhzPerCpu_;
+  }
+
+  public static final int CPU_INFO_FIELD_NUMBER = 4;
+  private volatile java.lang.Object cpuInfo_;
+  /**
+   * <pre>
+   * Additional cpu information. For example,
+   * Intel Ivybridge with HyperThreading (24 cores) dL1:32KB dL2:256KB dL3:30MB
+   * </pre>
+   *
+   * <code>string cpu_info = 4;</code>
+   */
+  public java.lang.String getCpuInfo() {
+    java.lang.Object ref = cpuInfo_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      cpuInfo_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * Additional cpu information. For example,
+   * Intel Ivybridge with HyperThreading (24 cores) dL1:32KB dL2:256KB dL3:30MB
+   * </pre>
+   *
+   * <code>string cpu_info = 4;</code>
+   */
+  public com.google.protobuf.ByteString
+      getCpuInfoBytes() {
+    java.lang.Object ref = cpuInfo_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      cpuInfo_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int CPU_GOVERNOR_FIELD_NUMBER = 5;
+  private volatile java.lang.Object cpuGovernor_;
+  /**
+   * <pre>
+   * What kind of cpu scaling is enabled on the host.
+   * Examples include "performance", "ondemand", "conservative", "mixed".
+   * </pre>
+   *
+   * <code>string cpu_governor = 5;</code>
+   */
+  public java.lang.String getCpuGovernor() {
+    java.lang.Object ref = cpuGovernor_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      cpuGovernor_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * What kind of cpu scaling is enabled on the host.
+   * Examples include "performance", "ondemand", "conservative", "mixed".
+   * </pre>
+   *
+   * <code>string cpu_governor = 5;</code>
+   */
+  public com.google.protobuf.ByteString
+      getCpuGovernorBytes() {
+    java.lang.Object ref = cpuGovernor_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      cpuGovernor_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int CACHE_SIZE_FIELD_NUMBER = 6;
+  private static final class CacheSizeDefaultEntryHolder {
+    static final com.google.protobuf.MapEntry<
+        java.lang.String, java.lang.Long> defaultEntry =
+            com.google.protobuf.MapEntry
+            .<java.lang.String, java.lang.Long>newDefaultInstance(
+                org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_CPUInfo_CacheSizeEntry_descriptor, 
+                com.google.protobuf.WireFormat.FieldType.STRING,
+                "",
+                com.google.protobuf.WireFormat.FieldType.INT64,
+                0L);
+  }
+  private com.google.protobuf.MapField<
+      java.lang.String, java.lang.Long> cacheSize_;
+  private com.google.protobuf.MapField<java.lang.String, java.lang.Long>
+  internalGetCacheSize() {
+    if (cacheSize_ == null) {
+      return com.google.protobuf.MapField.emptyMapField(
+          CacheSizeDefaultEntryHolder.defaultEntry);
+    }
+    return cacheSize_;
+  }
+
+  public int getCacheSizeCount() {
+    return internalGetCacheSize().getMap().size();
+  }
+  /**
+   * <pre>
+   * Cache sizes (in bytes), e.g. "L2": 262144 (for 256KB)
+   * </pre>
+   *
+   * <code>map&lt;string, int64&gt; cache_size = 6;</code>
+   */
+
+  public boolean containsCacheSize(
+      java.lang.String key) {
+    if (key == null) { throw new java.lang.NullPointerException(); }
+    return internalGetCacheSize().getMap().containsKey(key);
+  }
+  /**
+   * Use {@link #getCacheSizeMap()} instead.
+   */
+  @java.lang.Deprecated
+  public java.util.Map<java.lang.String, java.lang.Long> getCacheSize() {
+    return getCacheSizeMap();
+  }
+  /**
+   * <pre>
+   * Cache sizes (in bytes), e.g. "L2": 262144 (for 256KB)
+   * </pre>
+   *
+   * <code>map&lt;string, int64&gt; cache_size = 6;</code>
+   */
+
+  public java.util.Map<java.lang.String, java.lang.Long> getCacheSizeMap() {
+    return internalGetCacheSize().getMap();
+  }
+  /**
+   * <pre>
+   * Cache sizes (in bytes), e.g. "L2": 262144 (for 256KB)
+   * </pre>
+   *
+   * <code>map&lt;string, int64&gt; cache_size = 6;</code>
+   */
+
+  public long getCacheSizeOrDefault(
+      java.lang.String key,
+      long defaultValue) {
+    if (key == null) { throw new java.lang.NullPointerException(); }
+    java.util.Map<java.lang.String, java.lang.Long> map =
+        internalGetCacheSize().getMap();
+    return map.containsKey(key) ? map.get(key) : defaultValue;
+  }
+  /**
+   * <pre>
+   * Cache sizes (in bytes), e.g. "L2": 262144 (for 256KB)
+   * </pre>
+   *
+   * <code>map&lt;string, int64&gt; cache_size = 6;</code>
+   */
+
+  public long getCacheSizeOrThrow(
+      java.lang.String key) {
+    if (key == null) { throw new java.lang.NullPointerException(); }
+    java.util.Map<java.lang.String, java.lang.Long> map =
+        internalGetCacheSize().getMap();
+    if (!map.containsKey(key)) {
+      throw new java.lang.IllegalArgumentException();
+    }
+    return map.get(key);
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    if (numCores_ != 0L) {
+      output.writeInt64(1, numCores_);
+    }
+    if (numCoresAllowed_ != 0L) {
+      output.writeInt64(2, numCoresAllowed_);
+    }
+    if (mhzPerCpu_ != 0D) {
+      output.writeDouble(3, mhzPerCpu_);
+    }
+    if (!getCpuInfoBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 4, cpuInfo_);
+    }
+    if (!getCpuGovernorBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 5, cpuGovernor_);
+    }
+    com.google.protobuf.GeneratedMessageV3
+      .serializeStringMapTo(
+        output,
+        internalGetCacheSize(),
+        CacheSizeDefaultEntryHolder.defaultEntry,
+        6);
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (numCores_ != 0L) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt64Size(1, numCores_);
+    }
+    if (numCoresAllowed_ != 0L) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt64Size(2, numCoresAllowed_);
+    }
+    if (mhzPerCpu_ != 0D) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeDoubleSize(3, mhzPerCpu_);
+    }
+    if (!getCpuInfoBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(4, cpuInfo_);
+    }
+    if (!getCpuGovernorBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(5, cpuGovernor_);
+    }
+    for (java.util.Map.Entry<java.lang.String, java.lang.Long> entry
+         : internalGetCacheSize().getMap().entrySet()) {
+      com.google.protobuf.MapEntry<java.lang.String, java.lang.Long>
+      cacheSize__ = CacheSizeDefaultEntryHolder.defaultEntry.newBuilderForType()
+          .setKey(entry.getKey())
+          .setValue(entry.getValue())
+          .build();
+      size += com.google.protobuf.CodedOutputStream
+          .computeMessageSize(6, cacheSize__);
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.util.testlog.CPUInfo)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.util.testlog.CPUInfo other = (org.tensorflow.proto.util.testlog.CPUInfo) obj;
+
+    if (getNumCores()
+        != other.getNumCores()) return false;
+    if (getNumCoresAllowed()
+        != other.getNumCoresAllowed()) return false;
+    if (java.lang.Double.doubleToLongBits(getMhzPerCpu())
+        != java.lang.Double.doubleToLongBits(
+            other.getMhzPerCpu())) return false;
+    if (!getCpuInfo()
+        .equals(other.getCpuInfo())) return false;
+    if (!getCpuGovernor()
+        .equals(other.getCpuGovernor())) return false;
+    if (!internalGetCacheSize().equals(
+        other.internalGetCacheSize())) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    hash = (37 * hash) + NUM_CORES_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+        getNumCores());
+    hash = (37 * hash) + NUM_CORES_ALLOWED_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+        getNumCoresAllowed());
+    hash = (37 * hash) + MHZ_PER_CPU_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+        java.lang.Double.doubleToLongBits(getMhzPerCpu()));
+    hash = (37 * hash) + CPU_INFO_FIELD_NUMBER;
+    hash = (53 * hash) + getCpuInfo().hashCode();
+    hash = (37 * hash) + CPU_GOVERNOR_FIELD_NUMBER;
+    hash = (53 * hash) + getCpuGovernor().hashCode();
+    if (!internalGetCacheSize().getMap().isEmpty()) {
+      hash = (37 * hash) + CACHE_SIZE_FIELD_NUMBER;
+      hash = (53 * hash) + internalGetCacheSize().hashCode();
+    }
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.util.testlog.CPUInfo parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.testlog.CPUInfo parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.testlog.CPUInfo parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.testlog.CPUInfo parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.testlog.CPUInfo parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.testlog.CPUInfo parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.testlog.CPUInfo parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.testlog.CPUInfo parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.testlog.CPUInfo parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.testlog.CPUInfo parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.testlog.CPUInfo parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.testlog.CPUInfo parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.util.testlog.CPUInfo prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * Protobuf type {@code tensorflow.CPUInfo}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.CPUInfo)
+      org.tensorflow.proto.util.testlog.CPUInfoOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_CPUInfo_descriptor;
+    }
+
+    @SuppressWarnings({"rawtypes"})
+    protected com.google.protobuf.MapField internalGetMapField(
+        int number) {
+      switch (number) {
+        case 6:
+          return internalGetCacheSize();
+        default:
+          throw new RuntimeException(
+              "Invalid map field number: " + number);
+      }
+    }
+    @SuppressWarnings({"rawtypes"})
+    protected com.google.protobuf.MapField internalGetMutableMapField(
+        int number) {
+      switch (number) {
+        case 6:
+          return internalGetMutableCacheSize();
+        default:
+          throw new RuntimeException(
+              "Invalid map field number: " + number);
+      }
+    }
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_CPUInfo_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.util.testlog.CPUInfo.class, org.tensorflow.proto.util.testlog.CPUInfo.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.util.testlog.CPUInfo.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      numCores_ = 0L;
+
+      numCoresAllowed_ = 0L;
+
+      mhzPerCpu_ = 0D;
+
+      cpuInfo_ = "";
+
+      cpuGovernor_ = "";
+
+      internalGetMutableCacheSize().clear();
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_CPUInfo_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.testlog.CPUInfo getDefaultInstanceForType() {
+      return org.tensorflow.proto.util.testlog.CPUInfo.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.testlog.CPUInfo build() {
+      org.tensorflow.proto.util.testlog.CPUInfo result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.testlog.CPUInfo buildPartial() {
+      org.tensorflow.proto.util.testlog.CPUInfo result = new org.tensorflow.proto.util.testlog.CPUInfo(this);
+      int from_bitField0_ = bitField0_;
+      result.numCores_ = numCores_;
+      result.numCoresAllowed_ = numCoresAllowed_;
+      result.mhzPerCpu_ = mhzPerCpu_;
+      result.cpuInfo_ = cpuInfo_;
+      result.cpuGovernor_ = cpuGovernor_;
+      result.cacheSize_ = internalGetCacheSize();
+      result.cacheSize_.makeImmutable();
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.util.testlog.CPUInfo) {
+        return mergeFrom((org.tensorflow.proto.util.testlog.CPUInfo)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.util.testlog.CPUInfo other) {
+      if (other == org.tensorflow.proto.util.testlog.CPUInfo.getDefaultInstance()) return this;
+      if (other.getNumCores() != 0L) {
+        setNumCores(other.getNumCores());
+      }
+      if (other.getNumCoresAllowed() != 0L) {
+        setNumCoresAllowed(other.getNumCoresAllowed());
+      }
+      if (other.getMhzPerCpu() != 0D) {
+        setMhzPerCpu(other.getMhzPerCpu());
+      }
+      if (!other.getCpuInfo().isEmpty()) {
+        cpuInfo_ = other.cpuInfo_;
+        onChanged();
+      }
+      if (!other.getCpuGovernor().isEmpty()) {
+        cpuGovernor_ = other.cpuGovernor_;
+        onChanged();
+      }
+      internalGetMutableCacheSize().mergeFrom(
+          other.internalGetCacheSize());
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.util.testlog.CPUInfo parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.util.testlog.CPUInfo) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+    private int bitField0_;
+
+    private long numCores_ ;
+    /**
+     * <code>int64 num_cores = 1;</code>
+     */
+    public long getNumCores() {
+      return numCores_;
+    }
+    /**
+     * <code>int64 num_cores = 1;</code>
+     */
+    public Builder setNumCores(long value) {
+      
+      numCores_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>int64 num_cores = 1;</code>
+     */
+    public Builder clearNumCores() {
+      
+      numCores_ = 0L;
+      onChanged();
+      return this;
+    }
+
+    private long numCoresAllowed_ ;
+    /**
+     * <code>int64 num_cores_allowed = 2;</code>
+     */
+    public long getNumCoresAllowed() {
+      return numCoresAllowed_;
+    }
+    /**
+     * <code>int64 num_cores_allowed = 2;</code>
+     */
+    public Builder setNumCoresAllowed(long value) {
+      
+      numCoresAllowed_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>int64 num_cores_allowed = 2;</code>
+     */
+    public Builder clearNumCoresAllowed() {
+      
+      numCoresAllowed_ = 0L;
+      onChanged();
+      return this;
+    }
+
+    private double mhzPerCpu_ ;
+    /**
+     * <pre>
+     * How fast are these cpus?
+     * </pre>
+     *
+     * <code>double mhz_per_cpu = 3;</code>
+     */
+    public double getMhzPerCpu() {
+      return mhzPerCpu_;
+    }
+    /**
+     * <pre>
+     * How fast are these cpus?
+     * </pre>
+     *
+     * <code>double mhz_per_cpu = 3;</code>
+     */
+    public Builder setMhzPerCpu(double value) {
+      
+      mhzPerCpu_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * How fast are these cpus?
+     * </pre>
+     *
+     * <code>double mhz_per_cpu = 3;</code>
+     */
+    public Builder clearMhzPerCpu() {
+      
+      mhzPerCpu_ = 0D;
+      onChanged();
+      return this;
+    }
+
+    private java.lang.Object cpuInfo_ = "";
+    /**
+     * <pre>
+     * Additional cpu information. For example,
+     * Intel Ivybridge with HyperThreading (24 cores) dL1:32KB dL2:256KB dL3:30MB
+     * </pre>
+     *
+     * <code>string cpu_info = 4;</code>
+     */
+    public java.lang.String getCpuInfo() {
+      java.lang.Object ref = cpuInfo_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        cpuInfo_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Additional cpu information. For example,
+     * Intel Ivybridge with HyperThreading (24 cores) dL1:32KB dL2:256KB dL3:30MB
+     * </pre>
+     *
+     * <code>string cpu_info = 4;</code>
+     */
+    public com.google.protobuf.ByteString
+        getCpuInfoBytes() {
+      java.lang.Object ref = cpuInfo_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        cpuInfo_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Additional cpu information. For example,
+     * Intel Ivybridge with HyperThreading (24 cores) dL1:32KB dL2:256KB dL3:30MB
+     * </pre>
+     *
+     * <code>string cpu_info = 4;</code>
+     */
+    public Builder setCpuInfo(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      cpuInfo_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Additional cpu information. For example,
+     * Intel Ivybridge with HyperThreading (24 cores) dL1:32KB dL2:256KB dL3:30MB
+     * </pre>
+     *
+     * <code>string cpu_info = 4;</code>
+     */
+    public Builder clearCpuInfo() {
+      
+      cpuInfo_ = getDefaultInstance().getCpuInfo();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Additional cpu information. For example,
+     * Intel Ivybridge with HyperThreading (24 cores) dL1:32KB dL2:256KB dL3:30MB
+     * </pre>
+     *
+     * <code>string cpu_info = 4;</code>
+     */
+    public Builder setCpuInfoBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      cpuInfo_ = value;
+      onChanged();
+      return this;
+    }
+
+    private java.lang.Object cpuGovernor_ = "";
+    /**
+     * <pre>
+     * What kind of cpu scaling is enabled on the host.
+     * Examples include "performance", "ondemand", "conservative", "mixed".
+     * </pre>
+     *
+     * <code>string cpu_governor = 5;</code>
+     */
+    public java.lang.String getCpuGovernor() {
+      java.lang.Object ref = cpuGovernor_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        cpuGovernor_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * What kind of cpu scaling is enabled on the host.
+     * Examples include "performance", "ondemand", "conservative", "mixed".
+     * </pre>
+     *
+     * <code>string cpu_governor = 5;</code>
+     */
+    public com.google.protobuf.ByteString
+        getCpuGovernorBytes() {
+      java.lang.Object ref = cpuGovernor_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        cpuGovernor_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * What kind of cpu scaling is enabled on the host.
+     * Examples include "performance", "ondemand", "conservative", "mixed".
+     * </pre>
+     *
+     * <code>string cpu_governor = 5;</code>
+     */
+    public Builder setCpuGovernor(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      cpuGovernor_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * What kind of cpu scaling is enabled on the host.
+     * Examples include "performance", "ondemand", "conservative", "mixed".
+     * </pre>
+     *
+     * <code>string cpu_governor = 5;</code>
+     */
+    public Builder clearCpuGovernor() {
+      
+      cpuGovernor_ = getDefaultInstance().getCpuGovernor();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * What kind of cpu scaling is enabled on the host.
+     * Examples include "performance", "ondemand", "conservative", "mixed".
+     * </pre>
+     *
+     * <code>string cpu_governor = 5;</code>
+     */
+    public Builder setCpuGovernorBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      cpuGovernor_ = value;
+      onChanged();
+      return this;
+    }
+
+    private com.google.protobuf.MapField<
+        java.lang.String, java.lang.Long> cacheSize_;
+    private com.google.protobuf.MapField<java.lang.String, java.lang.Long>
+    internalGetCacheSize() {
+      if (cacheSize_ == null) {
+        return com.google.protobuf.MapField.emptyMapField(
+            CacheSizeDefaultEntryHolder.defaultEntry);
+      }
+      return cacheSize_;
+    }
+    private com.google.protobuf.MapField<java.lang.String, java.lang.Long>
+    internalGetMutableCacheSize() {
+      onChanged();;
+      if (cacheSize_ == null) {
+        cacheSize_ = com.google.protobuf.MapField.newMapField(
+            CacheSizeDefaultEntryHolder.defaultEntry);
+      }
+      if (!cacheSize_.isMutable()) {
+        cacheSize_ = cacheSize_.copy();
+      }
+      return cacheSize_;
+    }
+
+    public int getCacheSizeCount() {
+      return internalGetCacheSize().getMap().size();
+    }
+    /**
+     * <pre>
+     * Cache sizes (in bytes), e.g. "L2": 262144 (for 256KB)
+     * </pre>
+     *
+     * <code>map&lt;string, int64&gt; cache_size = 6;</code>
+     */
+
+    public boolean containsCacheSize(
+        java.lang.String key) {
+      if (key == null) { throw new java.lang.NullPointerException(); }
+      return internalGetCacheSize().getMap().containsKey(key);
+    }
+    /**
+     * Use {@link #getCacheSizeMap()} instead.
+     */
+    @java.lang.Deprecated
+    public java.util.Map<java.lang.String, java.lang.Long> getCacheSize() {
+      return getCacheSizeMap();
+    }
+    /**
+     * <pre>
+     * Cache sizes (in bytes), e.g. "L2": 262144 (for 256KB)
+     * </pre>
+     *
+     * <code>map&lt;string, int64&gt; cache_size = 6;</code>
+     */
+
+    public java.util.Map<java.lang.String, java.lang.Long> getCacheSizeMap() {
+      return internalGetCacheSize().getMap();
+    }
+    /**
+     * <pre>
+     * Cache sizes (in bytes), e.g. "L2": 262144 (for 256KB)
+     * </pre>
+     *
+     * <code>map&lt;string, int64&gt; cache_size = 6;</code>
+     */
+
+    public long getCacheSizeOrDefault(
+        java.lang.String key,
+        long defaultValue) {
+      if (key == null) { throw new java.lang.NullPointerException(); }
+      java.util.Map<java.lang.String, java.lang.Long> map =
+          internalGetCacheSize().getMap();
+      return map.containsKey(key) ? map.get(key) : defaultValue;
+    }
+    /**
+     * <pre>
+     * Cache sizes (in bytes), e.g. "L2": 262144 (for 256KB)
+     * </pre>
+     *
+     * <code>map&lt;string, int64&gt; cache_size = 6;</code>
+     */
+
+    public long getCacheSizeOrThrow(
+        java.lang.String key) {
+      if (key == null) { throw new java.lang.NullPointerException(); }
+      java.util.Map<java.lang.String, java.lang.Long> map =
+          internalGetCacheSize().getMap();
+      if (!map.containsKey(key)) {
+        throw new java.lang.IllegalArgumentException();
+      }
+      return map.get(key);
+    }
+
+    public Builder clearCacheSize() {
+      internalGetMutableCacheSize().getMutableMap()
+          .clear();
+      return this;
+    }
+    /**
+     * <pre>
+     * Cache sizes (in bytes), e.g. "L2": 262144 (for 256KB)
+     * </pre>
+     *
+     * <code>map&lt;string, int64&gt; cache_size = 6;</code>
+     */
+
+    public Builder removeCacheSize(
+        java.lang.String key) {
+      if (key == null) { throw new java.lang.NullPointerException(); }
+      internalGetMutableCacheSize().getMutableMap()
+          .remove(key);
+      return this;
+    }
+    /**
+     * Use alternate mutation accessors instead.
+     */
+    @java.lang.Deprecated
+    public java.util.Map<java.lang.String, java.lang.Long>
+    getMutableCacheSize() {
+      return internalGetMutableCacheSize().getMutableMap();
+    }
+    /**
+     * <pre>
+     * Cache sizes (in bytes), e.g. "L2": 262144 (for 256KB)
+     * </pre>
+     *
+     * <code>map&lt;string, int64&gt; cache_size = 6;</code>
+     */
+    public Builder putCacheSize(
+        java.lang.String key,
+        long value) {
+      if (key == null) { throw new java.lang.NullPointerException(); }
+      
+      internalGetMutableCacheSize().getMutableMap()
+          .put(key, value);
+      return this;
+    }
+    /**
+     * <pre>
+     * Cache sizes (in bytes), e.g. "L2": 262144 (for 256KB)
+     * </pre>
+     *
+     * <code>map&lt;string, int64&gt; cache_size = 6;</code>
+     */
+
+    public Builder putAllCacheSize(
+        java.util.Map<java.lang.String, java.lang.Long> values) {
+      internalGetMutableCacheSize().getMutableMap()
+          .putAll(values);
+      return this;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.CPUInfo)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.CPUInfo)
+  private static final org.tensorflow.proto.util.testlog.CPUInfo DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.util.testlog.CPUInfo();
+  }
+
+  public static org.tensorflow.proto.util.testlog.CPUInfo getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<CPUInfo>
+      PARSER = new com.google.protobuf.AbstractParser<CPUInfo>() {
+    @java.lang.Override
+    public CPUInfo parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new CPUInfo(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<CPUInfo> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<CPUInfo> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.util.testlog.CPUInfo getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/CPUInfoOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/CPUInfoOrBuilder.java
new file mode 100644
index 00000000000..be5a9556f41
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/CPUInfoOrBuilder.java
@@ -0,0 +1,122 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/util/test_log.proto
+
+package org.tensorflow.proto.util.testlog;
+
+public interface CPUInfoOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.CPUInfo)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <code>int64 num_cores = 1;</code>
+   */
+  long getNumCores();
+
+  /**
+   * <code>int64 num_cores_allowed = 2;</code>
+   */
+  long getNumCoresAllowed();
+
+  /**
+   * <pre>
+   * How fast are these cpus?
+   * </pre>
+   *
+   * <code>double mhz_per_cpu = 3;</code>
+   */
+  double getMhzPerCpu();
+
+  /**
+   * <pre>
+   * Additional cpu information. For example,
+   * Intel Ivybridge with HyperThreading (24 cores) dL1:32KB dL2:256KB dL3:30MB
+   * </pre>
+   *
+   * <code>string cpu_info = 4;</code>
+   */
+  java.lang.String getCpuInfo();
+  /**
+   * <pre>
+   * Additional cpu information. For example,
+   * Intel Ivybridge with HyperThreading (24 cores) dL1:32KB dL2:256KB dL3:30MB
+   * </pre>
+   *
+   * <code>string cpu_info = 4;</code>
+   */
+  com.google.protobuf.ByteString
+      getCpuInfoBytes();
+
+  /**
+   * <pre>
+   * What kind of cpu scaling is enabled on the host.
+   * Examples include "performance", "ondemand", "conservative", "mixed".
+   * </pre>
+   *
+   * <code>string cpu_governor = 5;</code>
+   */
+  java.lang.String getCpuGovernor();
+  /**
+   * <pre>
+   * What kind of cpu scaling is enabled on the host.
+   * Examples include "performance", "ondemand", "conservative", "mixed".
+   * </pre>
+   *
+   * <code>string cpu_governor = 5;</code>
+   */
+  com.google.protobuf.ByteString
+      getCpuGovernorBytes();
+
+  /**
+   * <pre>
+   * Cache sizes (in bytes), e.g. "L2": 262144 (for 256KB)
+   * </pre>
+   *
+   * <code>map&lt;string, int64&gt; cache_size = 6;</code>
+   */
+  int getCacheSizeCount();
+  /**
+   * <pre>
+   * Cache sizes (in bytes), e.g. "L2": 262144 (for 256KB)
+   * </pre>
+   *
+   * <code>map&lt;string, int64&gt; cache_size = 6;</code>
+   */
+  boolean containsCacheSize(
+      java.lang.String key);
+  /**
+   * Use {@link #getCacheSizeMap()} instead.
+   */
+  @java.lang.Deprecated
+  java.util.Map<java.lang.String, java.lang.Long>
+  getCacheSize();
+  /**
+   * <pre>
+   * Cache sizes (in bytes), e.g. "L2": 262144 (for 256KB)
+   * </pre>
+   *
+   * <code>map&lt;string, int64&gt; cache_size = 6;</code>
+   */
+  java.util.Map<java.lang.String, java.lang.Long>
+  getCacheSizeMap();
+  /**
+   * <pre>
+   * Cache sizes (in bytes), e.g. "L2": 262144 (for 256KB)
+   * </pre>
+   *
+   * <code>map&lt;string, int64&gt; cache_size = 6;</code>
+   */
+
+  long getCacheSizeOrDefault(
+      java.lang.String key,
+      long defaultValue);
+  /**
+   * <pre>
+   * Cache sizes (in bytes), e.g. "L2": 262144 (for 256KB)
+   * </pre>
+   *
+   * <code>map&lt;string, int64&gt; cache_size = 6;</code>
+   */
+
+  long getCacheSizeOrThrow(
+      java.lang.String key);
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/CommitId.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/CommitId.java
new file mode 100644
index 00000000000..beac981e730
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/CommitId.java
@@ -0,0 +1,964 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/util/test_log.proto
+
+package org.tensorflow.proto.util.testlog;
+
+/**
+ * Protobuf type {@code tensorflow.CommitId}
+ */
+public  final class CommitId extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.CommitId)
+    CommitIdOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use CommitId.newBuilder() to construct.
+  private CommitId(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private CommitId() {
+    snapshot_ = "";
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new CommitId();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private CommitId(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 8: {
+            kindCase_ = 1;
+            kind_ = input.readInt64();
+            break;
+          }
+          case 18: {
+            java.lang.String s = input.readStringRequireUtf8();
+            kindCase_ = 2;
+            kind_ = s;
+            break;
+          }
+          case 26: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            snapshot_ = s;
+            break;
+          }
+          case 32: {
+
+            pendingChangelist_ = input.readInt64();
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_CommitId_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_CommitId_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.util.testlog.CommitId.class, org.tensorflow.proto.util.testlog.CommitId.Builder.class);
+  }
+
+  private int kindCase_ = 0;
+  private java.lang.Object kind_;
+  public enum KindCase
+      implements com.google.protobuf.Internal.EnumLite {
+    CHANGELIST(1),
+    HASH(2),
+    KIND_NOT_SET(0);
+    private final int value;
+    private KindCase(int value) {
+      this.value = value;
+    }
+    /**
+     * @deprecated Use {@link #forNumber(int)} instead.
+     */
+    @java.lang.Deprecated
+    public static KindCase valueOf(int value) {
+      return forNumber(value);
+    }
+
+    public static KindCase forNumber(int value) {
+      switch (value) {
+        case 1: return CHANGELIST;
+        case 2: return HASH;
+        case 0: return KIND_NOT_SET;
+        default: return null;
+      }
+    }
+    public int getNumber() {
+      return this.value;
+    }
+  };
+
+  public KindCase
+  getKindCase() {
+    return KindCase.forNumber(
+        kindCase_);
+  }
+
+  public static final int CHANGELIST_FIELD_NUMBER = 1;
+  /**
+   * <pre>
+   * Submitted changelist.
+   * </pre>
+   *
+   * <code>int64 changelist = 1;</code>
+   */
+  public long getChangelist() {
+    if (kindCase_ == 1) {
+      return (java.lang.Long) kind_;
+    }
+    return 0L;
+  }
+
+  public static final int HASH_FIELD_NUMBER = 2;
+  /**
+   * <code>string hash = 2;</code>
+   */
+  public java.lang.String getHash() {
+    java.lang.Object ref = "";
+    if (kindCase_ == 2) {
+      ref = kind_;
+    }
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      if (kindCase_ == 2) {
+        kind_ = s;
+      }
+      return s;
+    }
+  }
+  /**
+   * <code>string hash = 2;</code>
+   */
+  public com.google.protobuf.ByteString
+      getHashBytes() {
+    java.lang.Object ref = "";
+    if (kindCase_ == 2) {
+      ref = kind_;
+    }
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      if (kindCase_ == 2) {
+        kind_ = b;
+      }
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int SNAPSHOT_FIELD_NUMBER = 3;
+  private volatile java.lang.Object snapshot_;
+  /**
+   * <pre>
+   * Hash of intermediate change between hash/changelist and what was tested.
+   * Not used if the build is from a commit without modifications.
+   * </pre>
+   *
+   * <code>string snapshot = 3;</code>
+   */
+  public java.lang.String getSnapshot() {
+    java.lang.Object ref = snapshot_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      snapshot_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * Hash of intermediate change between hash/changelist and what was tested.
+   * Not used if the build is from a commit without modifications.
+   * </pre>
+   *
+   * <code>string snapshot = 3;</code>
+   */
+  public com.google.protobuf.ByteString
+      getSnapshotBytes() {
+    java.lang.Object ref = snapshot_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      snapshot_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int PENDING_CHANGELIST_FIELD_NUMBER = 4;
+  private long pendingChangelist_;
+  /**
+   * <pre>
+   * Changelist tested if the change list is not already submitted.
+   * </pre>
+   *
+   * <code>int64 pending_changelist = 4;</code>
+   */
+  public long getPendingChangelist() {
+    return pendingChangelist_;
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    if (kindCase_ == 1) {
+      output.writeInt64(
+          1, (long)((java.lang.Long) kind_));
+    }
+    if (kindCase_ == 2) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 2, kind_);
+    }
+    if (!getSnapshotBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 3, snapshot_);
+    }
+    if (pendingChangelist_ != 0L) {
+      output.writeInt64(4, pendingChangelist_);
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (kindCase_ == 1) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt64Size(
+            1, (long)((java.lang.Long) kind_));
+    }
+    if (kindCase_ == 2) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(2, kind_);
+    }
+    if (!getSnapshotBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(3, snapshot_);
+    }
+    if (pendingChangelist_ != 0L) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt64Size(4, pendingChangelist_);
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.util.testlog.CommitId)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.util.testlog.CommitId other = (org.tensorflow.proto.util.testlog.CommitId) obj;
+
+    if (!getSnapshot()
+        .equals(other.getSnapshot())) return false;
+    if (getPendingChangelist()
+        != other.getPendingChangelist()) return false;
+    if (!getKindCase().equals(other.getKindCase())) return false;
+    switch (kindCase_) {
+      case 1:
+        if (getChangelist()
+            != other.getChangelist()) return false;
+        break;
+      case 2:
+        if (!getHash()
+            .equals(other.getHash())) return false;
+        break;
+      case 0:
+      default:
+    }
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    hash = (37 * hash) + SNAPSHOT_FIELD_NUMBER;
+    hash = (53 * hash) + getSnapshot().hashCode();
+    hash = (37 * hash) + PENDING_CHANGELIST_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+        getPendingChangelist());
+    switch (kindCase_) {
+      case 1:
+        hash = (37 * hash) + CHANGELIST_FIELD_NUMBER;
+        hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+            getChangelist());
+        break;
+      case 2:
+        hash = (37 * hash) + HASH_FIELD_NUMBER;
+        hash = (53 * hash) + getHash().hashCode();
+        break;
+      case 0:
+      default:
+    }
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.util.testlog.CommitId parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.testlog.CommitId parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.testlog.CommitId parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.testlog.CommitId parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.testlog.CommitId parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.testlog.CommitId parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.testlog.CommitId parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.testlog.CommitId parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.testlog.CommitId parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.testlog.CommitId parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.testlog.CommitId parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.testlog.CommitId parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.util.testlog.CommitId prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * Protobuf type {@code tensorflow.CommitId}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.CommitId)
+      org.tensorflow.proto.util.testlog.CommitIdOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_CommitId_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_CommitId_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.util.testlog.CommitId.class, org.tensorflow.proto.util.testlog.CommitId.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.util.testlog.CommitId.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      snapshot_ = "";
+
+      pendingChangelist_ = 0L;
+
+      kindCase_ = 0;
+      kind_ = null;
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_CommitId_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.testlog.CommitId getDefaultInstanceForType() {
+      return org.tensorflow.proto.util.testlog.CommitId.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.testlog.CommitId build() {
+      org.tensorflow.proto.util.testlog.CommitId result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.testlog.CommitId buildPartial() {
+      org.tensorflow.proto.util.testlog.CommitId result = new org.tensorflow.proto.util.testlog.CommitId(this);
+      if (kindCase_ == 1) {
+        result.kind_ = kind_;
+      }
+      if (kindCase_ == 2) {
+        result.kind_ = kind_;
+      }
+      result.snapshot_ = snapshot_;
+      result.pendingChangelist_ = pendingChangelist_;
+      result.kindCase_ = kindCase_;
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.util.testlog.CommitId) {
+        return mergeFrom((org.tensorflow.proto.util.testlog.CommitId)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.util.testlog.CommitId other) {
+      if (other == org.tensorflow.proto.util.testlog.CommitId.getDefaultInstance()) return this;
+      if (!other.getSnapshot().isEmpty()) {
+        snapshot_ = other.snapshot_;
+        onChanged();
+      }
+      if (other.getPendingChangelist() != 0L) {
+        setPendingChangelist(other.getPendingChangelist());
+      }
+      switch (other.getKindCase()) {
+        case CHANGELIST: {
+          setChangelist(other.getChangelist());
+          break;
+        }
+        case HASH: {
+          kindCase_ = 2;
+          kind_ = other.kind_;
+          onChanged();
+          break;
+        }
+        case KIND_NOT_SET: {
+          break;
+        }
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.util.testlog.CommitId parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.util.testlog.CommitId) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+    private int kindCase_ = 0;
+    private java.lang.Object kind_;
+    public KindCase
+        getKindCase() {
+      return KindCase.forNumber(
+          kindCase_);
+    }
+
+    public Builder clearKind() {
+      kindCase_ = 0;
+      kind_ = null;
+      onChanged();
+      return this;
+    }
+
+
+    /**
+     * <pre>
+     * Submitted changelist.
+     * </pre>
+     *
+     * <code>int64 changelist = 1;</code>
+     */
+    public long getChangelist() {
+      if (kindCase_ == 1) {
+        return (java.lang.Long) kind_;
+      }
+      return 0L;
+    }
+    /**
+     * <pre>
+     * Submitted changelist.
+     * </pre>
+     *
+     * <code>int64 changelist = 1;</code>
+     */
+    public Builder setChangelist(long value) {
+      kindCase_ = 1;
+      kind_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Submitted changelist.
+     * </pre>
+     *
+     * <code>int64 changelist = 1;</code>
+     */
+    public Builder clearChangelist() {
+      if (kindCase_ == 1) {
+        kindCase_ = 0;
+        kind_ = null;
+        onChanged();
+      }
+      return this;
+    }
+
+    /**
+     * <code>string hash = 2;</code>
+     */
+    public java.lang.String getHash() {
+      java.lang.Object ref = "";
+      if (kindCase_ == 2) {
+        ref = kind_;
+      }
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        if (kindCase_ == 2) {
+          kind_ = s;
+        }
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <code>string hash = 2;</code>
+     */
+    public com.google.protobuf.ByteString
+        getHashBytes() {
+      java.lang.Object ref = "";
+      if (kindCase_ == 2) {
+        ref = kind_;
+      }
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        if (kindCase_ == 2) {
+          kind_ = b;
+        }
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <code>string hash = 2;</code>
+     */
+    public Builder setHash(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  kindCase_ = 2;
+      kind_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>string hash = 2;</code>
+     */
+    public Builder clearHash() {
+      if (kindCase_ == 2) {
+        kindCase_ = 0;
+        kind_ = null;
+        onChanged();
+      }
+      return this;
+    }
+    /**
+     * <code>string hash = 2;</code>
+     */
+    public Builder setHashBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      kindCase_ = 2;
+      kind_ = value;
+      onChanged();
+      return this;
+    }
+
+    private java.lang.Object snapshot_ = "";
+    /**
+     * <pre>
+     * Hash of intermediate change between hash/changelist and what was tested.
+     * Not used if the build is from a commit without modifications.
+     * </pre>
+     *
+     * <code>string snapshot = 3;</code>
+     */
+    public java.lang.String getSnapshot() {
+      java.lang.Object ref = snapshot_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        snapshot_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Hash of intermediate change between hash/changelist and what was tested.
+     * Not used if the build is from a commit without modifications.
+     * </pre>
+     *
+     * <code>string snapshot = 3;</code>
+     */
+    public com.google.protobuf.ByteString
+        getSnapshotBytes() {
+      java.lang.Object ref = snapshot_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        snapshot_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Hash of intermediate change between hash/changelist and what was tested.
+     * Not used if the build is from a commit without modifications.
+     * </pre>
+     *
+     * <code>string snapshot = 3;</code>
+     */
+    public Builder setSnapshot(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      snapshot_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Hash of intermediate change between hash/changelist and what was tested.
+     * Not used if the build is from a commit without modifications.
+     * </pre>
+     *
+     * <code>string snapshot = 3;</code>
+     */
+    public Builder clearSnapshot() {
+      
+      snapshot_ = getDefaultInstance().getSnapshot();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Hash of intermediate change between hash/changelist and what was tested.
+     * Not used if the build is from a commit without modifications.
+     * </pre>
+     *
+     * <code>string snapshot = 3;</code>
+     */
+    public Builder setSnapshotBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      snapshot_ = value;
+      onChanged();
+      return this;
+    }
+
+    private long pendingChangelist_ ;
+    /**
+     * <pre>
+     * Changelist tested if the change list is not already submitted.
+     * </pre>
+     *
+     * <code>int64 pending_changelist = 4;</code>
+     */
+    public long getPendingChangelist() {
+      return pendingChangelist_;
+    }
+    /**
+     * <pre>
+     * Changelist tested if the change list is not already submitted.
+     * </pre>
+     *
+     * <code>int64 pending_changelist = 4;</code>
+     */
+    public Builder setPendingChangelist(long value) {
+      
+      pendingChangelist_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Changelist tested if the change list is not already submitted.
+     * </pre>
+     *
+     * <code>int64 pending_changelist = 4;</code>
+     */
+    public Builder clearPendingChangelist() {
+      
+      pendingChangelist_ = 0L;
+      onChanged();
+      return this;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.CommitId)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.CommitId)
+  private static final org.tensorflow.proto.util.testlog.CommitId DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.util.testlog.CommitId();
+  }
+
+  public static org.tensorflow.proto.util.testlog.CommitId getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<CommitId>
+      PARSER = new com.google.protobuf.AbstractParser<CommitId>() {
+    @java.lang.Override
+    public CommitId parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new CommitId(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<CommitId> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<CommitId> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.util.testlog.CommitId getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/CommitIdOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/CommitIdOrBuilder.java
new file mode 100644
index 00000000000..b63f4353379
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/CommitIdOrBuilder.java
@@ -0,0 +1,59 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/util/test_log.proto
+
+package org.tensorflow.proto.util.testlog;
+
+public interface CommitIdOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.CommitId)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <pre>
+   * Submitted changelist.
+   * </pre>
+   *
+   * <code>int64 changelist = 1;</code>
+   */
+  long getChangelist();
+
+  /**
+   * <code>string hash = 2;</code>
+   */
+  java.lang.String getHash();
+  /**
+   * <code>string hash = 2;</code>
+   */
+  com.google.protobuf.ByteString
+      getHashBytes();
+
+  /**
+   * <pre>
+   * Hash of intermediate change between hash/changelist and what was tested.
+   * Not used if the build is from a commit without modifications.
+   * </pre>
+   *
+   * <code>string snapshot = 3;</code>
+   */
+  java.lang.String getSnapshot();
+  /**
+   * <pre>
+   * Hash of intermediate change between hash/changelist and what was tested.
+   * Not used if the build is from a commit without modifications.
+   * </pre>
+   *
+   * <code>string snapshot = 3;</code>
+   */
+  com.google.protobuf.ByteString
+      getSnapshotBytes();
+
+  /**
+   * <pre>
+   * Changelist tested if the change list is not already submitted.
+   * </pre>
+   *
+   * <code>int64 pending_changelist = 4;</code>
+   */
+  long getPendingChangelist();
+
+  public org.tensorflow.proto.util.testlog.CommitId.KindCase getKindCase();
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/EntryValue.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/EntryValue.java
new file mode 100644
index 00000000000..d8ca9fffd45
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/EntryValue.java
@@ -0,0 +1,713 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/util/test_log.proto
+
+package org.tensorflow.proto.util.testlog;
+
+/**
+ * Protobuf type {@code tensorflow.EntryValue}
+ */
+public  final class EntryValue extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.EntryValue)
+    EntryValueOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use EntryValue.newBuilder() to construct.
+  private EntryValue(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private EntryValue() {
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new EntryValue();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private EntryValue(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 9: {
+            kindCase_ = 1;
+            kind_ = input.readDouble();
+            break;
+          }
+          case 18: {
+            java.lang.String s = input.readStringRequireUtf8();
+            kindCase_ = 2;
+            kind_ = s;
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_EntryValue_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_EntryValue_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.util.testlog.EntryValue.class, org.tensorflow.proto.util.testlog.EntryValue.Builder.class);
+  }
+
+  private int kindCase_ = 0;
+  private java.lang.Object kind_;
+  public enum KindCase
+      implements com.google.protobuf.Internal.EnumLite {
+    DOUBLE_VALUE(1),
+    STRING_VALUE(2),
+    KIND_NOT_SET(0);
+    private final int value;
+    private KindCase(int value) {
+      this.value = value;
+    }
+    /**
+     * @deprecated Use {@link #forNumber(int)} instead.
+     */
+    @java.lang.Deprecated
+    public static KindCase valueOf(int value) {
+      return forNumber(value);
+    }
+
+    public static KindCase forNumber(int value) {
+      switch (value) {
+        case 1: return DOUBLE_VALUE;
+        case 2: return STRING_VALUE;
+        case 0: return KIND_NOT_SET;
+        default: return null;
+      }
+    }
+    public int getNumber() {
+      return this.value;
+    }
+  };
+
+  public KindCase
+  getKindCase() {
+    return KindCase.forNumber(
+        kindCase_);
+  }
+
+  public static final int DOUBLE_VALUE_FIELD_NUMBER = 1;
+  /**
+   * <code>double double_value = 1;</code>
+   */
+  public double getDoubleValue() {
+    if (kindCase_ == 1) {
+      return (java.lang.Double) kind_;
+    }
+    return 0D;
+  }
+
+  public static final int STRING_VALUE_FIELD_NUMBER = 2;
+  /**
+   * <code>string string_value = 2;</code>
+   */
+  public java.lang.String getStringValue() {
+    java.lang.Object ref = "";
+    if (kindCase_ == 2) {
+      ref = kind_;
+    }
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      if (kindCase_ == 2) {
+        kind_ = s;
+      }
+      return s;
+    }
+  }
+  /**
+   * <code>string string_value = 2;</code>
+   */
+  public com.google.protobuf.ByteString
+      getStringValueBytes() {
+    java.lang.Object ref = "";
+    if (kindCase_ == 2) {
+      ref = kind_;
+    }
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      if (kindCase_ == 2) {
+        kind_ = b;
+      }
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    if (kindCase_ == 1) {
+      output.writeDouble(
+          1, (double)((java.lang.Double) kind_));
+    }
+    if (kindCase_ == 2) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 2, kind_);
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (kindCase_ == 1) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeDoubleSize(
+            1, (double)((java.lang.Double) kind_));
+    }
+    if (kindCase_ == 2) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(2, kind_);
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.util.testlog.EntryValue)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.util.testlog.EntryValue other = (org.tensorflow.proto.util.testlog.EntryValue) obj;
+
+    if (!getKindCase().equals(other.getKindCase())) return false;
+    switch (kindCase_) {
+      case 1:
+        if (java.lang.Double.doubleToLongBits(getDoubleValue())
+            != java.lang.Double.doubleToLongBits(
+                other.getDoubleValue())) return false;
+        break;
+      case 2:
+        if (!getStringValue()
+            .equals(other.getStringValue())) return false;
+        break;
+      case 0:
+      default:
+    }
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    switch (kindCase_) {
+      case 1:
+        hash = (37 * hash) + DOUBLE_VALUE_FIELD_NUMBER;
+        hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+            java.lang.Double.doubleToLongBits(getDoubleValue()));
+        break;
+      case 2:
+        hash = (37 * hash) + STRING_VALUE_FIELD_NUMBER;
+        hash = (53 * hash) + getStringValue().hashCode();
+        break;
+      case 0:
+      default:
+    }
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.util.testlog.EntryValue parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.testlog.EntryValue parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.testlog.EntryValue parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.testlog.EntryValue parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.testlog.EntryValue parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.testlog.EntryValue parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.testlog.EntryValue parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.testlog.EntryValue parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.testlog.EntryValue parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.testlog.EntryValue parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.testlog.EntryValue parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.testlog.EntryValue parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.util.testlog.EntryValue prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * Protobuf type {@code tensorflow.EntryValue}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.EntryValue)
+      org.tensorflow.proto.util.testlog.EntryValueOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_EntryValue_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_EntryValue_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.util.testlog.EntryValue.class, org.tensorflow.proto.util.testlog.EntryValue.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.util.testlog.EntryValue.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      kindCase_ = 0;
+      kind_ = null;
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_EntryValue_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.testlog.EntryValue getDefaultInstanceForType() {
+      return org.tensorflow.proto.util.testlog.EntryValue.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.testlog.EntryValue build() {
+      org.tensorflow.proto.util.testlog.EntryValue result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.testlog.EntryValue buildPartial() {
+      org.tensorflow.proto.util.testlog.EntryValue result = new org.tensorflow.proto.util.testlog.EntryValue(this);
+      if (kindCase_ == 1) {
+        result.kind_ = kind_;
+      }
+      if (kindCase_ == 2) {
+        result.kind_ = kind_;
+      }
+      result.kindCase_ = kindCase_;
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.util.testlog.EntryValue) {
+        return mergeFrom((org.tensorflow.proto.util.testlog.EntryValue)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.util.testlog.EntryValue other) {
+      if (other == org.tensorflow.proto.util.testlog.EntryValue.getDefaultInstance()) return this;
+      switch (other.getKindCase()) {
+        case DOUBLE_VALUE: {
+          setDoubleValue(other.getDoubleValue());
+          break;
+        }
+        case STRING_VALUE: {
+          kindCase_ = 2;
+          kind_ = other.kind_;
+          onChanged();
+          break;
+        }
+        case KIND_NOT_SET: {
+          break;
+        }
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.util.testlog.EntryValue parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.util.testlog.EntryValue) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+    private int kindCase_ = 0;
+    private java.lang.Object kind_;
+    public KindCase
+        getKindCase() {
+      return KindCase.forNumber(
+          kindCase_);
+    }
+
+    public Builder clearKind() {
+      kindCase_ = 0;
+      kind_ = null;
+      onChanged();
+      return this;
+    }
+
+
+    /**
+     * <code>double double_value = 1;</code>
+     */
+    public double getDoubleValue() {
+      if (kindCase_ == 1) {
+        return (java.lang.Double) kind_;
+      }
+      return 0D;
+    }
+    /**
+     * <code>double double_value = 1;</code>
+     */
+    public Builder setDoubleValue(double value) {
+      kindCase_ = 1;
+      kind_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>double double_value = 1;</code>
+     */
+    public Builder clearDoubleValue() {
+      if (kindCase_ == 1) {
+        kindCase_ = 0;
+        kind_ = null;
+        onChanged();
+      }
+      return this;
+    }
+
+    /**
+     * <code>string string_value = 2;</code>
+     */
+    public java.lang.String getStringValue() {
+      java.lang.Object ref = "";
+      if (kindCase_ == 2) {
+        ref = kind_;
+      }
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        if (kindCase_ == 2) {
+          kind_ = s;
+        }
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <code>string string_value = 2;</code>
+     */
+    public com.google.protobuf.ByteString
+        getStringValueBytes() {
+      java.lang.Object ref = "";
+      if (kindCase_ == 2) {
+        ref = kind_;
+      }
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        if (kindCase_ == 2) {
+          kind_ = b;
+        }
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <code>string string_value = 2;</code>
+     */
+    public Builder setStringValue(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  kindCase_ = 2;
+      kind_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>string string_value = 2;</code>
+     */
+    public Builder clearStringValue() {
+      if (kindCase_ == 2) {
+        kindCase_ = 0;
+        kind_ = null;
+        onChanged();
+      }
+      return this;
+    }
+    /**
+     * <code>string string_value = 2;</code>
+     */
+    public Builder setStringValueBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      kindCase_ = 2;
+      kind_ = value;
+      onChanged();
+      return this;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.EntryValue)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.EntryValue)
+  private static final org.tensorflow.proto.util.testlog.EntryValue DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.util.testlog.EntryValue();
+  }
+
+  public static org.tensorflow.proto.util.testlog.EntryValue getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<EntryValue>
+      PARSER = new com.google.protobuf.AbstractParser<EntryValue>() {
+    @java.lang.Override
+    public EntryValue parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new EntryValue(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<EntryValue> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<EntryValue> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.util.testlog.EntryValue getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/EntryValueOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/EntryValueOrBuilder.java
new file mode 100644
index 00000000000..4a2d71e54ba
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/EntryValueOrBuilder.java
@@ -0,0 +1,26 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/util/test_log.proto
+
+package org.tensorflow.proto.util.testlog;
+
+public interface EntryValueOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.EntryValue)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <code>double double_value = 1;</code>
+   */
+  double getDoubleValue();
+
+  /**
+   * <code>string string_value = 2;</code>
+   */
+  java.lang.String getStringValue();
+  /**
+   * <code>string string_value = 2;</code>
+   */
+  com.google.protobuf.ByteString
+      getStringValueBytes();
+
+  public org.tensorflow.proto.util.testlog.EntryValue.KindCase getKindCase();
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/GPUInfo.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/GPUInfo.java
new file mode 100644
index 00000000000..7f6878d36ad
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/GPUInfo.java
@@ -0,0 +1,884 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/util/test_log.proto
+
+package org.tensorflow.proto.util.testlog;
+
+/**
+ * Protobuf type {@code tensorflow.GPUInfo}
+ */
+public  final class GPUInfo extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.GPUInfo)
+    GPUInfoOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use GPUInfo.newBuilder() to construct.
+  private GPUInfo(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private GPUInfo() {
+    model_ = "";
+    uuid_ = "";
+    busId_ = "";
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new GPUInfo();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private GPUInfo(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 10: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            model_ = s;
+            break;
+          }
+          case 18: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            uuid_ = s;
+            break;
+          }
+          case 26: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            busId_ = s;
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_GPUInfo_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_GPUInfo_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.util.testlog.GPUInfo.class, org.tensorflow.proto.util.testlog.GPUInfo.Builder.class);
+  }
+
+  public static final int MODEL_FIELD_NUMBER = 1;
+  private volatile java.lang.Object model_;
+  /**
+   * <pre>
+   * e.g. "Tesla K40c"
+   * </pre>
+   *
+   * <code>string model = 1;</code>
+   */
+  public java.lang.String getModel() {
+    java.lang.Object ref = model_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      model_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * e.g. "Tesla K40c"
+   * </pre>
+   *
+   * <code>string model = 1;</code>
+   */
+  public com.google.protobuf.ByteString
+      getModelBytes() {
+    java.lang.Object ref = model_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      model_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int UUID_FIELD_NUMBER = 2;
+  private volatile java.lang.Object uuid_;
+  /**
+   * <pre>
+   * Final entry in output of "nvidia-smi -L"
+   * </pre>
+   *
+   * <code>string uuid = 2;</code>
+   */
+  public java.lang.String getUuid() {
+    java.lang.Object ref = uuid_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      uuid_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * Final entry in output of "nvidia-smi -L"
+   * </pre>
+   *
+   * <code>string uuid = 2;</code>
+   */
+  public com.google.protobuf.ByteString
+      getUuidBytes() {
+    java.lang.Object ref = uuid_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      uuid_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int BUS_ID_FIELD_NUMBER = 3;
+  private volatile java.lang.Object busId_;
+  /**
+   * <pre>
+   * e.g. "0000:04:00.0"
+   * </pre>
+   *
+   * <code>string bus_id = 3;</code>
+   */
+  public java.lang.String getBusId() {
+    java.lang.Object ref = busId_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      busId_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * e.g. "0000:04:00.0"
+   * </pre>
+   *
+   * <code>string bus_id = 3;</code>
+   */
+  public com.google.protobuf.ByteString
+      getBusIdBytes() {
+    java.lang.Object ref = busId_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      busId_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    if (!getModelBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 1, model_);
+    }
+    if (!getUuidBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 2, uuid_);
+    }
+    if (!getBusIdBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 3, busId_);
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (!getModelBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, model_);
+    }
+    if (!getUuidBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(2, uuid_);
+    }
+    if (!getBusIdBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(3, busId_);
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.util.testlog.GPUInfo)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.util.testlog.GPUInfo other = (org.tensorflow.proto.util.testlog.GPUInfo) obj;
+
+    if (!getModel()
+        .equals(other.getModel())) return false;
+    if (!getUuid()
+        .equals(other.getUuid())) return false;
+    if (!getBusId()
+        .equals(other.getBusId())) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    hash = (37 * hash) + MODEL_FIELD_NUMBER;
+    hash = (53 * hash) + getModel().hashCode();
+    hash = (37 * hash) + UUID_FIELD_NUMBER;
+    hash = (53 * hash) + getUuid().hashCode();
+    hash = (37 * hash) + BUS_ID_FIELD_NUMBER;
+    hash = (53 * hash) + getBusId().hashCode();
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.util.testlog.GPUInfo parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.testlog.GPUInfo parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.testlog.GPUInfo parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.testlog.GPUInfo parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.testlog.GPUInfo parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.testlog.GPUInfo parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.testlog.GPUInfo parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.testlog.GPUInfo parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.testlog.GPUInfo parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.testlog.GPUInfo parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.testlog.GPUInfo parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.testlog.GPUInfo parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.util.testlog.GPUInfo prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * Protobuf type {@code tensorflow.GPUInfo}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.GPUInfo)
+      org.tensorflow.proto.util.testlog.GPUInfoOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_GPUInfo_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_GPUInfo_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.util.testlog.GPUInfo.class, org.tensorflow.proto.util.testlog.GPUInfo.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.util.testlog.GPUInfo.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      model_ = "";
+
+      uuid_ = "";
+
+      busId_ = "";
+
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_GPUInfo_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.testlog.GPUInfo getDefaultInstanceForType() {
+      return org.tensorflow.proto.util.testlog.GPUInfo.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.testlog.GPUInfo build() {
+      org.tensorflow.proto.util.testlog.GPUInfo result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.testlog.GPUInfo buildPartial() {
+      org.tensorflow.proto.util.testlog.GPUInfo result = new org.tensorflow.proto.util.testlog.GPUInfo(this);
+      result.model_ = model_;
+      result.uuid_ = uuid_;
+      result.busId_ = busId_;
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.util.testlog.GPUInfo) {
+        return mergeFrom((org.tensorflow.proto.util.testlog.GPUInfo)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.util.testlog.GPUInfo other) {
+      if (other == org.tensorflow.proto.util.testlog.GPUInfo.getDefaultInstance()) return this;
+      if (!other.getModel().isEmpty()) {
+        model_ = other.model_;
+        onChanged();
+      }
+      if (!other.getUuid().isEmpty()) {
+        uuid_ = other.uuid_;
+        onChanged();
+      }
+      if (!other.getBusId().isEmpty()) {
+        busId_ = other.busId_;
+        onChanged();
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.util.testlog.GPUInfo parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.util.testlog.GPUInfo) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+
+    private java.lang.Object model_ = "";
+    /**
+     * <pre>
+     * e.g. "Tesla K40c"
+     * </pre>
+     *
+     * <code>string model = 1;</code>
+     */
+    public java.lang.String getModel() {
+      java.lang.Object ref = model_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        model_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * e.g. "Tesla K40c"
+     * </pre>
+     *
+     * <code>string model = 1;</code>
+     */
+    public com.google.protobuf.ByteString
+        getModelBytes() {
+      java.lang.Object ref = model_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        model_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * e.g. "Tesla K40c"
+     * </pre>
+     *
+     * <code>string model = 1;</code>
+     */
+    public Builder setModel(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      model_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * e.g. "Tesla K40c"
+     * </pre>
+     *
+     * <code>string model = 1;</code>
+     */
+    public Builder clearModel() {
+      
+      model_ = getDefaultInstance().getModel();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * e.g. "Tesla K40c"
+     * </pre>
+     *
+     * <code>string model = 1;</code>
+     */
+    public Builder setModelBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      model_ = value;
+      onChanged();
+      return this;
+    }
+
+    private java.lang.Object uuid_ = "";
+    /**
+     * <pre>
+     * Final entry in output of "nvidia-smi -L"
+     * </pre>
+     *
+     * <code>string uuid = 2;</code>
+     */
+    public java.lang.String getUuid() {
+      java.lang.Object ref = uuid_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        uuid_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Final entry in output of "nvidia-smi -L"
+     * </pre>
+     *
+     * <code>string uuid = 2;</code>
+     */
+    public com.google.protobuf.ByteString
+        getUuidBytes() {
+      java.lang.Object ref = uuid_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        uuid_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Final entry in output of "nvidia-smi -L"
+     * </pre>
+     *
+     * <code>string uuid = 2;</code>
+     */
+    public Builder setUuid(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      uuid_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Final entry in output of "nvidia-smi -L"
+     * </pre>
+     *
+     * <code>string uuid = 2;</code>
+     */
+    public Builder clearUuid() {
+      
+      uuid_ = getDefaultInstance().getUuid();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Final entry in output of "nvidia-smi -L"
+     * </pre>
+     *
+     * <code>string uuid = 2;</code>
+     */
+    public Builder setUuidBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      uuid_ = value;
+      onChanged();
+      return this;
+    }
+
+    private java.lang.Object busId_ = "";
+    /**
+     * <pre>
+     * e.g. "0000:04:00.0"
+     * </pre>
+     *
+     * <code>string bus_id = 3;</code>
+     */
+    public java.lang.String getBusId() {
+      java.lang.Object ref = busId_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        busId_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * e.g. "0000:04:00.0"
+     * </pre>
+     *
+     * <code>string bus_id = 3;</code>
+     */
+    public com.google.protobuf.ByteString
+        getBusIdBytes() {
+      java.lang.Object ref = busId_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        busId_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * e.g. "0000:04:00.0"
+     * </pre>
+     *
+     * <code>string bus_id = 3;</code>
+     */
+    public Builder setBusId(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      busId_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * e.g. "0000:04:00.0"
+     * </pre>
+     *
+     * <code>string bus_id = 3;</code>
+     */
+    public Builder clearBusId() {
+      
+      busId_ = getDefaultInstance().getBusId();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * e.g. "0000:04:00.0"
+     * </pre>
+     *
+     * <code>string bus_id = 3;</code>
+     */
+    public Builder setBusIdBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      busId_ = value;
+      onChanged();
+      return this;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.GPUInfo)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.GPUInfo)
+  private static final org.tensorflow.proto.util.testlog.GPUInfo DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.util.testlog.GPUInfo();
+  }
+
+  public static org.tensorflow.proto.util.testlog.GPUInfo getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<GPUInfo>
+      PARSER = new com.google.protobuf.AbstractParser<GPUInfo>() {
+    @java.lang.Override
+    public GPUInfo parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new GPUInfo(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<GPUInfo> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<GPUInfo> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.util.testlog.GPUInfo getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/GPUInfoOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/GPUInfoOrBuilder.java
new file mode 100644
index 00000000000..44985f7d574
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/GPUInfoOrBuilder.java
@@ -0,0 +1,63 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/util/test_log.proto
+
+package org.tensorflow.proto.util.testlog;
+
+public interface GPUInfoOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.GPUInfo)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <pre>
+   * e.g. "Tesla K40c"
+   * </pre>
+   *
+   * <code>string model = 1;</code>
+   */
+  java.lang.String getModel();
+  /**
+   * <pre>
+   * e.g. "Tesla K40c"
+   * </pre>
+   *
+   * <code>string model = 1;</code>
+   */
+  com.google.protobuf.ByteString
+      getModelBytes();
+
+  /**
+   * <pre>
+   * Final entry in output of "nvidia-smi -L"
+   * </pre>
+   *
+   * <code>string uuid = 2;</code>
+   */
+  java.lang.String getUuid();
+  /**
+   * <pre>
+   * Final entry in output of "nvidia-smi -L"
+   * </pre>
+   *
+   * <code>string uuid = 2;</code>
+   */
+  com.google.protobuf.ByteString
+      getUuidBytes();
+
+  /**
+   * <pre>
+   * e.g. "0000:04:00.0"
+   * </pre>
+   *
+   * <code>string bus_id = 3;</code>
+   */
+  java.lang.String getBusId();
+  /**
+   * <pre>
+   * e.g. "0000:04:00.0"
+   * </pre>
+   *
+   * <code>string bus_id = 3;</code>
+   */
+  com.google.protobuf.ByteString
+      getBusIdBytes();
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/MachineConfiguration.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/MachineConfiguration.java
new file mode 100644
index 00000000000..70b9d4a3d7c
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/MachineConfiguration.java
@@ -0,0 +1,2241 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/util/test_log.proto
+
+package org.tensorflow.proto.util.testlog;
+
+/**
+ * Protobuf type {@code tensorflow.MachineConfiguration}
+ */
+public  final class MachineConfiguration extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.MachineConfiguration)
+    MachineConfigurationOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use MachineConfiguration.newBuilder() to construct.
+  private MachineConfiguration(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private MachineConfiguration() {
+    hostname_ = "";
+    serialIdentifier_ = "";
+    deviceInfo_ = java.util.Collections.emptyList();
+    availableDeviceInfo_ = java.util.Collections.emptyList();
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new MachineConfiguration();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private MachineConfiguration(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    int mutable_bitField0_ = 0;
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 10: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            hostname_ = s;
+            break;
+          }
+          case 18: {
+            org.tensorflow.proto.util.testlog.PlatformInfo.Builder subBuilder = null;
+            if (platformInfo_ != null) {
+              subBuilder = platformInfo_.toBuilder();
+            }
+            platformInfo_ = input.readMessage(org.tensorflow.proto.util.testlog.PlatformInfo.parser(), extensionRegistry);
+            if (subBuilder != null) {
+              subBuilder.mergeFrom(platformInfo_);
+              platformInfo_ = subBuilder.buildPartial();
+            }
+
+            break;
+          }
+          case 26: {
+            org.tensorflow.proto.util.testlog.CPUInfo.Builder subBuilder = null;
+            if (cpuInfo_ != null) {
+              subBuilder = cpuInfo_.toBuilder();
+            }
+            cpuInfo_ = input.readMessage(org.tensorflow.proto.util.testlog.CPUInfo.parser(), extensionRegistry);
+            if (subBuilder != null) {
+              subBuilder.mergeFrom(cpuInfo_);
+              cpuInfo_ = subBuilder.buildPartial();
+            }
+
+            break;
+          }
+          case 34: {
+            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
+              deviceInfo_ = new java.util.ArrayList<com.google.protobuf.Any>();
+              mutable_bitField0_ |= 0x00000001;
+            }
+            deviceInfo_.add(
+                input.readMessage(com.google.protobuf.Any.parser(), extensionRegistry));
+            break;
+          }
+          case 42: {
+            if (!((mutable_bitField0_ & 0x00000002) != 0)) {
+              availableDeviceInfo_ = new java.util.ArrayList<org.tensorflow.proto.util.testlog.AvailableDeviceInfo>();
+              mutable_bitField0_ |= 0x00000002;
+            }
+            availableDeviceInfo_.add(
+                input.readMessage(org.tensorflow.proto.util.testlog.AvailableDeviceInfo.parser(), extensionRegistry));
+            break;
+          }
+          case 50: {
+            org.tensorflow.proto.util.testlog.MemoryInfo.Builder subBuilder = null;
+            if (memoryInfo_ != null) {
+              subBuilder = memoryInfo_.toBuilder();
+            }
+            memoryInfo_ = input.readMessage(org.tensorflow.proto.util.testlog.MemoryInfo.parser(), extensionRegistry);
+            if (subBuilder != null) {
+              subBuilder.mergeFrom(memoryInfo_);
+              memoryInfo_ = subBuilder.buildPartial();
+            }
+
+            break;
+          }
+          case 58: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            serialIdentifier_ = s;
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      if (((mutable_bitField0_ & 0x00000001) != 0)) {
+        deviceInfo_ = java.util.Collections.unmodifiableList(deviceInfo_);
+      }
+      if (((mutable_bitField0_ & 0x00000002) != 0)) {
+        availableDeviceInfo_ = java.util.Collections.unmodifiableList(availableDeviceInfo_);
+      }
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_MachineConfiguration_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_MachineConfiguration_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.util.testlog.MachineConfiguration.class, org.tensorflow.proto.util.testlog.MachineConfiguration.Builder.class);
+  }
+
+  public static final int HOSTNAME_FIELD_NUMBER = 1;
+  private volatile java.lang.Object hostname_;
+  /**
+   * <pre>
+   * Host name of machine that ran the benchmark.
+   * </pre>
+   *
+   * <code>string hostname = 1;</code>
+   */
+  public java.lang.String getHostname() {
+    java.lang.Object ref = hostname_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      hostname_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * Host name of machine that ran the benchmark.
+   * </pre>
+   *
+   * <code>string hostname = 1;</code>
+   */
+  public com.google.protobuf.ByteString
+      getHostnameBytes() {
+    java.lang.Object ref = hostname_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      hostname_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int SERIAL_IDENTIFIER_FIELD_NUMBER = 7;
+  private volatile java.lang.Object serialIdentifier_;
+  /**
+   * <pre>
+   * Unique serial number of the machine.
+   * </pre>
+   *
+   * <code>string serial_identifier = 7;</code>
+   */
+  public java.lang.String getSerialIdentifier() {
+    java.lang.Object ref = serialIdentifier_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      serialIdentifier_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * Unique serial number of the machine.
+   * </pre>
+   *
+   * <code>string serial_identifier = 7;</code>
+   */
+  public com.google.protobuf.ByteString
+      getSerialIdentifierBytes() {
+    java.lang.Object ref = serialIdentifier_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      serialIdentifier_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int PLATFORM_INFO_FIELD_NUMBER = 2;
+  private org.tensorflow.proto.util.testlog.PlatformInfo platformInfo_;
+  /**
+   * <pre>
+   * Additional platform information.
+   * </pre>
+   *
+   * <code>.tensorflow.PlatformInfo platform_info = 2;</code>
+   */
+  public boolean hasPlatformInfo() {
+    return platformInfo_ != null;
+  }
+  /**
+   * <pre>
+   * Additional platform information.
+   * </pre>
+   *
+   * <code>.tensorflow.PlatformInfo platform_info = 2;</code>
+   */
+  public org.tensorflow.proto.util.testlog.PlatformInfo getPlatformInfo() {
+    return platformInfo_ == null ? org.tensorflow.proto.util.testlog.PlatformInfo.getDefaultInstance() : platformInfo_;
+  }
+  /**
+   * <pre>
+   * Additional platform information.
+   * </pre>
+   *
+   * <code>.tensorflow.PlatformInfo platform_info = 2;</code>
+   */
+  public org.tensorflow.proto.util.testlog.PlatformInfoOrBuilder getPlatformInfoOrBuilder() {
+    return getPlatformInfo();
+  }
+
+  public static final int CPU_INFO_FIELD_NUMBER = 3;
+  private org.tensorflow.proto.util.testlog.CPUInfo cpuInfo_;
+  /**
+   * <pre>
+   * CPU Information.
+   * </pre>
+   *
+   * <code>.tensorflow.CPUInfo cpu_info = 3;</code>
+   */
+  public boolean hasCpuInfo() {
+    return cpuInfo_ != null;
+  }
+  /**
+   * <pre>
+   * CPU Information.
+   * </pre>
+   *
+   * <code>.tensorflow.CPUInfo cpu_info = 3;</code>
+   */
+  public org.tensorflow.proto.util.testlog.CPUInfo getCpuInfo() {
+    return cpuInfo_ == null ? org.tensorflow.proto.util.testlog.CPUInfo.getDefaultInstance() : cpuInfo_;
+  }
+  /**
+   * <pre>
+   * CPU Information.
+   * </pre>
+   *
+   * <code>.tensorflow.CPUInfo cpu_info = 3;</code>
+   */
+  public org.tensorflow.proto.util.testlog.CPUInfoOrBuilder getCpuInfoOrBuilder() {
+    return getCpuInfo();
+  }
+
+  public static final int DEVICE_INFO_FIELD_NUMBER = 4;
+  private java.util.List<com.google.protobuf.Any> deviceInfo_;
+  /**
+   * <pre>
+   * Other devices that are attached and relevant (e.g. GPUInfo).
+   * </pre>
+   *
+   * <code>repeated .google.protobuf.Any device_info = 4;</code>
+   */
+  public java.util.List<com.google.protobuf.Any> getDeviceInfoList() {
+    return deviceInfo_;
+  }
+  /**
+   * <pre>
+   * Other devices that are attached and relevant (e.g. GPUInfo).
+   * </pre>
+   *
+   * <code>repeated .google.protobuf.Any device_info = 4;</code>
+   */
+  public java.util.List<? extends com.google.protobuf.AnyOrBuilder> 
+      getDeviceInfoOrBuilderList() {
+    return deviceInfo_;
+  }
+  /**
+   * <pre>
+   * Other devices that are attached and relevant (e.g. GPUInfo).
+   * </pre>
+   *
+   * <code>repeated .google.protobuf.Any device_info = 4;</code>
+   */
+  public int getDeviceInfoCount() {
+    return deviceInfo_.size();
+  }
+  /**
+   * <pre>
+   * Other devices that are attached and relevant (e.g. GPUInfo).
+   * </pre>
+   *
+   * <code>repeated .google.protobuf.Any device_info = 4;</code>
+   */
+  public com.google.protobuf.Any getDeviceInfo(int index) {
+    return deviceInfo_.get(index);
+  }
+  /**
+   * <pre>
+   * Other devices that are attached and relevant (e.g. GPUInfo).
+   * </pre>
+   *
+   * <code>repeated .google.protobuf.Any device_info = 4;</code>
+   */
+  public com.google.protobuf.AnyOrBuilder getDeviceInfoOrBuilder(
+      int index) {
+    return deviceInfo_.get(index);
+  }
+
+  public static final int AVAILABLE_DEVICE_INFO_FIELD_NUMBER = 5;
+  private java.util.List<org.tensorflow.proto.util.testlog.AvailableDeviceInfo> availableDeviceInfo_;
+  /**
+   * <pre>
+   * Devices accessible to the test (e.g. as given by list_local_devices).
+   * </pre>
+   *
+   * <code>repeated .tensorflow.AvailableDeviceInfo available_device_info = 5;</code>
+   */
+  public java.util.List<org.tensorflow.proto.util.testlog.AvailableDeviceInfo> getAvailableDeviceInfoList() {
+    return availableDeviceInfo_;
+  }
+  /**
+   * <pre>
+   * Devices accessible to the test (e.g. as given by list_local_devices).
+   * </pre>
+   *
+   * <code>repeated .tensorflow.AvailableDeviceInfo available_device_info = 5;</code>
+   */
+  public java.util.List<? extends org.tensorflow.proto.util.testlog.AvailableDeviceInfoOrBuilder> 
+      getAvailableDeviceInfoOrBuilderList() {
+    return availableDeviceInfo_;
+  }
+  /**
+   * <pre>
+   * Devices accessible to the test (e.g. as given by list_local_devices).
+   * </pre>
+   *
+   * <code>repeated .tensorflow.AvailableDeviceInfo available_device_info = 5;</code>
+   */
+  public int getAvailableDeviceInfoCount() {
+    return availableDeviceInfo_.size();
+  }
+  /**
+   * <pre>
+   * Devices accessible to the test (e.g. as given by list_local_devices).
+   * </pre>
+   *
+   * <code>repeated .tensorflow.AvailableDeviceInfo available_device_info = 5;</code>
+   */
+  public org.tensorflow.proto.util.testlog.AvailableDeviceInfo getAvailableDeviceInfo(int index) {
+    return availableDeviceInfo_.get(index);
+  }
+  /**
+   * <pre>
+   * Devices accessible to the test (e.g. as given by list_local_devices).
+   * </pre>
+   *
+   * <code>repeated .tensorflow.AvailableDeviceInfo available_device_info = 5;</code>
+   */
+  public org.tensorflow.proto.util.testlog.AvailableDeviceInfoOrBuilder getAvailableDeviceInfoOrBuilder(
+      int index) {
+    return availableDeviceInfo_.get(index);
+  }
+
+  public static final int MEMORY_INFO_FIELD_NUMBER = 6;
+  private org.tensorflow.proto.util.testlog.MemoryInfo memoryInfo_;
+  /**
+   * <code>.tensorflow.MemoryInfo memory_info = 6;</code>
+   */
+  public boolean hasMemoryInfo() {
+    return memoryInfo_ != null;
+  }
+  /**
+   * <code>.tensorflow.MemoryInfo memory_info = 6;</code>
+   */
+  public org.tensorflow.proto.util.testlog.MemoryInfo getMemoryInfo() {
+    return memoryInfo_ == null ? org.tensorflow.proto.util.testlog.MemoryInfo.getDefaultInstance() : memoryInfo_;
+  }
+  /**
+   * <code>.tensorflow.MemoryInfo memory_info = 6;</code>
+   */
+  public org.tensorflow.proto.util.testlog.MemoryInfoOrBuilder getMemoryInfoOrBuilder() {
+    return getMemoryInfo();
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    if (!getHostnameBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 1, hostname_);
+    }
+    if (platformInfo_ != null) {
+      output.writeMessage(2, getPlatformInfo());
+    }
+    if (cpuInfo_ != null) {
+      output.writeMessage(3, getCpuInfo());
+    }
+    for (int i = 0; i < deviceInfo_.size(); i++) {
+      output.writeMessage(4, deviceInfo_.get(i));
+    }
+    for (int i = 0; i < availableDeviceInfo_.size(); i++) {
+      output.writeMessage(5, availableDeviceInfo_.get(i));
+    }
+    if (memoryInfo_ != null) {
+      output.writeMessage(6, getMemoryInfo());
+    }
+    if (!getSerialIdentifierBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 7, serialIdentifier_);
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (!getHostnameBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, hostname_);
+    }
+    if (platformInfo_ != null) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(2, getPlatformInfo());
+    }
+    if (cpuInfo_ != null) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(3, getCpuInfo());
+    }
+    for (int i = 0; i < deviceInfo_.size(); i++) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(4, deviceInfo_.get(i));
+    }
+    for (int i = 0; i < availableDeviceInfo_.size(); i++) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(5, availableDeviceInfo_.get(i));
+    }
+    if (memoryInfo_ != null) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(6, getMemoryInfo());
+    }
+    if (!getSerialIdentifierBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(7, serialIdentifier_);
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.util.testlog.MachineConfiguration)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.util.testlog.MachineConfiguration other = (org.tensorflow.proto.util.testlog.MachineConfiguration) obj;
+
+    if (!getHostname()
+        .equals(other.getHostname())) return false;
+    if (!getSerialIdentifier()
+        .equals(other.getSerialIdentifier())) return false;
+    if (hasPlatformInfo() != other.hasPlatformInfo()) return false;
+    if (hasPlatformInfo()) {
+      if (!getPlatformInfo()
+          .equals(other.getPlatformInfo())) return false;
+    }
+    if (hasCpuInfo() != other.hasCpuInfo()) return false;
+    if (hasCpuInfo()) {
+      if (!getCpuInfo()
+          .equals(other.getCpuInfo())) return false;
+    }
+    if (!getDeviceInfoList()
+        .equals(other.getDeviceInfoList())) return false;
+    if (!getAvailableDeviceInfoList()
+        .equals(other.getAvailableDeviceInfoList())) return false;
+    if (hasMemoryInfo() != other.hasMemoryInfo()) return false;
+    if (hasMemoryInfo()) {
+      if (!getMemoryInfo()
+          .equals(other.getMemoryInfo())) return false;
+    }
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    hash = (37 * hash) + HOSTNAME_FIELD_NUMBER;
+    hash = (53 * hash) + getHostname().hashCode();
+    hash = (37 * hash) + SERIAL_IDENTIFIER_FIELD_NUMBER;
+    hash = (53 * hash) + getSerialIdentifier().hashCode();
+    if (hasPlatformInfo()) {
+      hash = (37 * hash) + PLATFORM_INFO_FIELD_NUMBER;
+      hash = (53 * hash) + getPlatformInfo().hashCode();
+    }
+    if (hasCpuInfo()) {
+      hash = (37 * hash) + CPU_INFO_FIELD_NUMBER;
+      hash = (53 * hash) + getCpuInfo().hashCode();
+    }
+    if (getDeviceInfoCount() > 0) {
+      hash = (37 * hash) + DEVICE_INFO_FIELD_NUMBER;
+      hash = (53 * hash) + getDeviceInfoList().hashCode();
+    }
+    if (getAvailableDeviceInfoCount() > 0) {
+      hash = (37 * hash) + AVAILABLE_DEVICE_INFO_FIELD_NUMBER;
+      hash = (53 * hash) + getAvailableDeviceInfoList().hashCode();
+    }
+    if (hasMemoryInfo()) {
+      hash = (37 * hash) + MEMORY_INFO_FIELD_NUMBER;
+      hash = (53 * hash) + getMemoryInfo().hashCode();
+    }
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.util.testlog.MachineConfiguration parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.testlog.MachineConfiguration parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.testlog.MachineConfiguration parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.testlog.MachineConfiguration parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.testlog.MachineConfiguration parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.testlog.MachineConfiguration parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.testlog.MachineConfiguration parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.testlog.MachineConfiguration parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.testlog.MachineConfiguration parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.testlog.MachineConfiguration parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.testlog.MachineConfiguration parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.testlog.MachineConfiguration parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.util.testlog.MachineConfiguration prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * Protobuf type {@code tensorflow.MachineConfiguration}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.MachineConfiguration)
+      org.tensorflow.proto.util.testlog.MachineConfigurationOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_MachineConfiguration_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_MachineConfiguration_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.util.testlog.MachineConfiguration.class, org.tensorflow.proto.util.testlog.MachineConfiguration.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.util.testlog.MachineConfiguration.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+        getDeviceInfoFieldBuilder();
+        getAvailableDeviceInfoFieldBuilder();
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      hostname_ = "";
+
+      serialIdentifier_ = "";
+
+      if (platformInfoBuilder_ == null) {
+        platformInfo_ = null;
+      } else {
+        platformInfo_ = null;
+        platformInfoBuilder_ = null;
+      }
+      if (cpuInfoBuilder_ == null) {
+        cpuInfo_ = null;
+      } else {
+        cpuInfo_ = null;
+        cpuInfoBuilder_ = null;
+      }
+      if (deviceInfoBuilder_ == null) {
+        deviceInfo_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000001);
+      } else {
+        deviceInfoBuilder_.clear();
+      }
+      if (availableDeviceInfoBuilder_ == null) {
+        availableDeviceInfo_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000002);
+      } else {
+        availableDeviceInfoBuilder_.clear();
+      }
+      if (memoryInfoBuilder_ == null) {
+        memoryInfo_ = null;
+      } else {
+        memoryInfo_ = null;
+        memoryInfoBuilder_ = null;
+      }
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_MachineConfiguration_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.testlog.MachineConfiguration getDefaultInstanceForType() {
+      return org.tensorflow.proto.util.testlog.MachineConfiguration.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.testlog.MachineConfiguration build() {
+      org.tensorflow.proto.util.testlog.MachineConfiguration result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.testlog.MachineConfiguration buildPartial() {
+      org.tensorflow.proto.util.testlog.MachineConfiguration result = new org.tensorflow.proto.util.testlog.MachineConfiguration(this);
+      int from_bitField0_ = bitField0_;
+      result.hostname_ = hostname_;
+      result.serialIdentifier_ = serialIdentifier_;
+      if (platformInfoBuilder_ == null) {
+        result.platformInfo_ = platformInfo_;
+      } else {
+        result.platformInfo_ = platformInfoBuilder_.build();
+      }
+      if (cpuInfoBuilder_ == null) {
+        result.cpuInfo_ = cpuInfo_;
+      } else {
+        result.cpuInfo_ = cpuInfoBuilder_.build();
+      }
+      if (deviceInfoBuilder_ == null) {
+        if (((bitField0_ & 0x00000001) != 0)) {
+          deviceInfo_ = java.util.Collections.unmodifiableList(deviceInfo_);
+          bitField0_ = (bitField0_ & ~0x00000001);
+        }
+        result.deviceInfo_ = deviceInfo_;
+      } else {
+        result.deviceInfo_ = deviceInfoBuilder_.build();
+      }
+      if (availableDeviceInfoBuilder_ == null) {
+        if (((bitField0_ & 0x00000002) != 0)) {
+          availableDeviceInfo_ = java.util.Collections.unmodifiableList(availableDeviceInfo_);
+          bitField0_ = (bitField0_ & ~0x00000002);
+        }
+        result.availableDeviceInfo_ = availableDeviceInfo_;
+      } else {
+        result.availableDeviceInfo_ = availableDeviceInfoBuilder_.build();
+      }
+      if (memoryInfoBuilder_ == null) {
+        result.memoryInfo_ = memoryInfo_;
+      } else {
+        result.memoryInfo_ = memoryInfoBuilder_.build();
+      }
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.util.testlog.MachineConfiguration) {
+        return mergeFrom((org.tensorflow.proto.util.testlog.MachineConfiguration)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.util.testlog.MachineConfiguration other) {
+      if (other == org.tensorflow.proto.util.testlog.MachineConfiguration.getDefaultInstance()) return this;
+      if (!other.getHostname().isEmpty()) {
+        hostname_ = other.hostname_;
+        onChanged();
+      }
+      if (!other.getSerialIdentifier().isEmpty()) {
+        serialIdentifier_ = other.serialIdentifier_;
+        onChanged();
+      }
+      if (other.hasPlatformInfo()) {
+        mergePlatformInfo(other.getPlatformInfo());
+      }
+      if (other.hasCpuInfo()) {
+        mergeCpuInfo(other.getCpuInfo());
+      }
+      if (deviceInfoBuilder_ == null) {
+        if (!other.deviceInfo_.isEmpty()) {
+          if (deviceInfo_.isEmpty()) {
+            deviceInfo_ = other.deviceInfo_;
+            bitField0_ = (bitField0_ & ~0x00000001);
+          } else {
+            ensureDeviceInfoIsMutable();
+            deviceInfo_.addAll(other.deviceInfo_);
+          }
+          onChanged();
+        }
+      } else {
+        if (!other.deviceInfo_.isEmpty()) {
+          if (deviceInfoBuilder_.isEmpty()) {
+            deviceInfoBuilder_.dispose();
+            deviceInfoBuilder_ = null;
+            deviceInfo_ = other.deviceInfo_;
+            bitField0_ = (bitField0_ & ~0x00000001);
+            deviceInfoBuilder_ = 
+              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
+                 getDeviceInfoFieldBuilder() : null;
+          } else {
+            deviceInfoBuilder_.addAllMessages(other.deviceInfo_);
+          }
+        }
+      }
+      if (availableDeviceInfoBuilder_ == null) {
+        if (!other.availableDeviceInfo_.isEmpty()) {
+          if (availableDeviceInfo_.isEmpty()) {
+            availableDeviceInfo_ = other.availableDeviceInfo_;
+            bitField0_ = (bitField0_ & ~0x00000002);
+          } else {
+            ensureAvailableDeviceInfoIsMutable();
+            availableDeviceInfo_.addAll(other.availableDeviceInfo_);
+          }
+          onChanged();
+        }
+      } else {
+        if (!other.availableDeviceInfo_.isEmpty()) {
+          if (availableDeviceInfoBuilder_.isEmpty()) {
+            availableDeviceInfoBuilder_.dispose();
+            availableDeviceInfoBuilder_ = null;
+            availableDeviceInfo_ = other.availableDeviceInfo_;
+            bitField0_ = (bitField0_ & ~0x00000002);
+            availableDeviceInfoBuilder_ = 
+              com.google.protobuf.GeneratedMessageV3.alwaysUseFieldBuilders ?
+                 getAvailableDeviceInfoFieldBuilder() : null;
+          } else {
+            availableDeviceInfoBuilder_.addAllMessages(other.availableDeviceInfo_);
+          }
+        }
+      }
+      if (other.hasMemoryInfo()) {
+        mergeMemoryInfo(other.getMemoryInfo());
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.util.testlog.MachineConfiguration parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.util.testlog.MachineConfiguration) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+    private int bitField0_;
+
+    private java.lang.Object hostname_ = "";
+    /**
+     * <pre>
+     * Host name of machine that ran the benchmark.
+     * </pre>
+     *
+     * <code>string hostname = 1;</code>
+     */
+    public java.lang.String getHostname() {
+      java.lang.Object ref = hostname_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        hostname_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Host name of machine that ran the benchmark.
+     * </pre>
+     *
+     * <code>string hostname = 1;</code>
+     */
+    public com.google.protobuf.ByteString
+        getHostnameBytes() {
+      java.lang.Object ref = hostname_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        hostname_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Host name of machine that ran the benchmark.
+     * </pre>
+     *
+     * <code>string hostname = 1;</code>
+     */
+    public Builder setHostname(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      hostname_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Host name of machine that ran the benchmark.
+     * </pre>
+     *
+     * <code>string hostname = 1;</code>
+     */
+    public Builder clearHostname() {
+      
+      hostname_ = getDefaultInstance().getHostname();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Host name of machine that ran the benchmark.
+     * </pre>
+     *
+     * <code>string hostname = 1;</code>
+     */
+    public Builder setHostnameBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      hostname_ = value;
+      onChanged();
+      return this;
+    }
+
+    private java.lang.Object serialIdentifier_ = "";
+    /**
+     * <pre>
+     * Unique serial number of the machine.
+     * </pre>
+     *
+     * <code>string serial_identifier = 7;</code>
+     */
+    public java.lang.String getSerialIdentifier() {
+      java.lang.Object ref = serialIdentifier_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        serialIdentifier_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Unique serial number of the machine.
+     * </pre>
+     *
+     * <code>string serial_identifier = 7;</code>
+     */
+    public com.google.protobuf.ByteString
+        getSerialIdentifierBytes() {
+      java.lang.Object ref = serialIdentifier_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        serialIdentifier_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Unique serial number of the machine.
+     * </pre>
+     *
+     * <code>string serial_identifier = 7;</code>
+     */
+    public Builder setSerialIdentifier(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      serialIdentifier_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Unique serial number of the machine.
+     * </pre>
+     *
+     * <code>string serial_identifier = 7;</code>
+     */
+    public Builder clearSerialIdentifier() {
+      
+      serialIdentifier_ = getDefaultInstance().getSerialIdentifier();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Unique serial number of the machine.
+     * </pre>
+     *
+     * <code>string serial_identifier = 7;</code>
+     */
+    public Builder setSerialIdentifierBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      serialIdentifier_ = value;
+      onChanged();
+      return this;
+    }
+
+    private org.tensorflow.proto.util.testlog.PlatformInfo platformInfo_;
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.util.testlog.PlatformInfo, org.tensorflow.proto.util.testlog.PlatformInfo.Builder, org.tensorflow.proto.util.testlog.PlatformInfoOrBuilder> platformInfoBuilder_;
+    /**
+     * <pre>
+     * Additional platform information.
+     * </pre>
+     *
+     * <code>.tensorflow.PlatformInfo platform_info = 2;</code>
+     */
+    public boolean hasPlatformInfo() {
+      return platformInfoBuilder_ != null || platformInfo_ != null;
+    }
+    /**
+     * <pre>
+     * Additional platform information.
+     * </pre>
+     *
+     * <code>.tensorflow.PlatformInfo platform_info = 2;</code>
+     */
+    public org.tensorflow.proto.util.testlog.PlatformInfo getPlatformInfo() {
+      if (platformInfoBuilder_ == null) {
+        return platformInfo_ == null ? org.tensorflow.proto.util.testlog.PlatformInfo.getDefaultInstance() : platformInfo_;
+      } else {
+        return platformInfoBuilder_.getMessage();
+      }
+    }
+    /**
+     * <pre>
+     * Additional platform information.
+     * </pre>
+     *
+     * <code>.tensorflow.PlatformInfo platform_info = 2;</code>
+     */
+    public Builder setPlatformInfo(org.tensorflow.proto.util.testlog.PlatformInfo value) {
+      if (platformInfoBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        platformInfo_ = value;
+        onChanged();
+      } else {
+        platformInfoBuilder_.setMessage(value);
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Additional platform information.
+     * </pre>
+     *
+     * <code>.tensorflow.PlatformInfo platform_info = 2;</code>
+     */
+    public Builder setPlatformInfo(
+        org.tensorflow.proto.util.testlog.PlatformInfo.Builder builderForValue) {
+      if (platformInfoBuilder_ == null) {
+        platformInfo_ = builderForValue.build();
+        onChanged();
+      } else {
+        platformInfoBuilder_.setMessage(builderForValue.build());
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Additional platform information.
+     * </pre>
+     *
+     * <code>.tensorflow.PlatformInfo platform_info = 2;</code>
+     */
+    public Builder mergePlatformInfo(org.tensorflow.proto.util.testlog.PlatformInfo value) {
+      if (platformInfoBuilder_ == null) {
+        if (platformInfo_ != null) {
+          platformInfo_ =
+            org.tensorflow.proto.util.testlog.PlatformInfo.newBuilder(platformInfo_).mergeFrom(value).buildPartial();
+        } else {
+          platformInfo_ = value;
+        }
+        onChanged();
+      } else {
+        platformInfoBuilder_.mergeFrom(value);
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Additional platform information.
+     * </pre>
+     *
+     * <code>.tensorflow.PlatformInfo platform_info = 2;</code>
+     */
+    public Builder clearPlatformInfo() {
+      if (platformInfoBuilder_ == null) {
+        platformInfo_ = null;
+        onChanged();
+      } else {
+        platformInfo_ = null;
+        platformInfoBuilder_ = null;
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Additional platform information.
+     * </pre>
+     *
+     * <code>.tensorflow.PlatformInfo platform_info = 2;</code>
+     */
+    public org.tensorflow.proto.util.testlog.PlatformInfo.Builder getPlatformInfoBuilder() {
+      
+      onChanged();
+      return getPlatformInfoFieldBuilder().getBuilder();
+    }
+    /**
+     * <pre>
+     * Additional platform information.
+     * </pre>
+     *
+     * <code>.tensorflow.PlatformInfo platform_info = 2;</code>
+     */
+    public org.tensorflow.proto.util.testlog.PlatformInfoOrBuilder getPlatformInfoOrBuilder() {
+      if (platformInfoBuilder_ != null) {
+        return platformInfoBuilder_.getMessageOrBuilder();
+      } else {
+        return platformInfo_ == null ?
+            org.tensorflow.proto.util.testlog.PlatformInfo.getDefaultInstance() : platformInfo_;
+      }
+    }
+    /**
+     * <pre>
+     * Additional platform information.
+     * </pre>
+     *
+     * <code>.tensorflow.PlatformInfo platform_info = 2;</code>
+     */
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.util.testlog.PlatformInfo, org.tensorflow.proto.util.testlog.PlatformInfo.Builder, org.tensorflow.proto.util.testlog.PlatformInfoOrBuilder> 
+        getPlatformInfoFieldBuilder() {
+      if (platformInfoBuilder_ == null) {
+        platformInfoBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+            org.tensorflow.proto.util.testlog.PlatformInfo, org.tensorflow.proto.util.testlog.PlatformInfo.Builder, org.tensorflow.proto.util.testlog.PlatformInfoOrBuilder>(
+                getPlatformInfo(),
+                getParentForChildren(),
+                isClean());
+        platformInfo_ = null;
+      }
+      return platformInfoBuilder_;
+    }
+
+    private org.tensorflow.proto.util.testlog.CPUInfo cpuInfo_;
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.util.testlog.CPUInfo, org.tensorflow.proto.util.testlog.CPUInfo.Builder, org.tensorflow.proto.util.testlog.CPUInfoOrBuilder> cpuInfoBuilder_;
+    /**
+     * <pre>
+     * CPU Information.
+     * </pre>
+     *
+     * <code>.tensorflow.CPUInfo cpu_info = 3;</code>
+     */
+    public boolean hasCpuInfo() {
+      return cpuInfoBuilder_ != null || cpuInfo_ != null;
+    }
+    /**
+     * <pre>
+     * CPU Information.
+     * </pre>
+     *
+     * <code>.tensorflow.CPUInfo cpu_info = 3;</code>
+     */
+    public org.tensorflow.proto.util.testlog.CPUInfo getCpuInfo() {
+      if (cpuInfoBuilder_ == null) {
+        return cpuInfo_ == null ? org.tensorflow.proto.util.testlog.CPUInfo.getDefaultInstance() : cpuInfo_;
+      } else {
+        return cpuInfoBuilder_.getMessage();
+      }
+    }
+    /**
+     * <pre>
+     * CPU Information.
+     * </pre>
+     *
+     * <code>.tensorflow.CPUInfo cpu_info = 3;</code>
+     */
+    public Builder setCpuInfo(org.tensorflow.proto.util.testlog.CPUInfo value) {
+      if (cpuInfoBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        cpuInfo_ = value;
+        onChanged();
+      } else {
+        cpuInfoBuilder_.setMessage(value);
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * CPU Information.
+     * </pre>
+     *
+     * <code>.tensorflow.CPUInfo cpu_info = 3;</code>
+     */
+    public Builder setCpuInfo(
+        org.tensorflow.proto.util.testlog.CPUInfo.Builder builderForValue) {
+      if (cpuInfoBuilder_ == null) {
+        cpuInfo_ = builderForValue.build();
+        onChanged();
+      } else {
+        cpuInfoBuilder_.setMessage(builderForValue.build());
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * CPU Information.
+     * </pre>
+     *
+     * <code>.tensorflow.CPUInfo cpu_info = 3;</code>
+     */
+    public Builder mergeCpuInfo(org.tensorflow.proto.util.testlog.CPUInfo value) {
+      if (cpuInfoBuilder_ == null) {
+        if (cpuInfo_ != null) {
+          cpuInfo_ =
+            org.tensorflow.proto.util.testlog.CPUInfo.newBuilder(cpuInfo_).mergeFrom(value).buildPartial();
+        } else {
+          cpuInfo_ = value;
+        }
+        onChanged();
+      } else {
+        cpuInfoBuilder_.mergeFrom(value);
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * CPU Information.
+     * </pre>
+     *
+     * <code>.tensorflow.CPUInfo cpu_info = 3;</code>
+     */
+    public Builder clearCpuInfo() {
+      if (cpuInfoBuilder_ == null) {
+        cpuInfo_ = null;
+        onChanged();
+      } else {
+        cpuInfo_ = null;
+        cpuInfoBuilder_ = null;
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * CPU Information.
+     * </pre>
+     *
+     * <code>.tensorflow.CPUInfo cpu_info = 3;</code>
+     */
+    public org.tensorflow.proto.util.testlog.CPUInfo.Builder getCpuInfoBuilder() {
+      
+      onChanged();
+      return getCpuInfoFieldBuilder().getBuilder();
+    }
+    /**
+     * <pre>
+     * CPU Information.
+     * </pre>
+     *
+     * <code>.tensorflow.CPUInfo cpu_info = 3;</code>
+     */
+    public org.tensorflow.proto.util.testlog.CPUInfoOrBuilder getCpuInfoOrBuilder() {
+      if (cpuInfoBuilder_ != null) {
+        return cpuInfoBuilder_.getMessageOrBuilder();
+      } else {
+        return cpuInfo_ == null ?
+            org.tensorflow.proto.util.testlog.CPUInfo.getDefaultInstance() : cpuInfo_;
+      }
+    }
+    /**
+     * <pre>
+     * CPU Information.
+     * </pre>
+     *
+     * <code>.tensorflow.CPUInfo cpu_info = 3;</code>
+     */
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.util.testlog.CPUInfo, org.tensorflow.proto.util.testlog.CPUInfo.Builder, org.tensorflow.proto.util.testlog.CPUInfoOrBuilder> 
+        getCpuInfoFieldBuilder() {
+      if (cpuInfoBuilder_ == null) {
+        cpuInfoBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+            org.tensorflow.proto.util.testlog.CPUInfo, org.tensorflow.proto.util.testlog.CPUInfo.Builder, org.tensorflow.proto.util.testlog.CPUInfoOrBuilder>(
+                getCpuInfo(),
+                getParentForChildren(),
+                isClean());
+        cpuInfo_ = null;
+      }
+      return cpuInfoBuilder_;
+    }
+
+    private java.util.List<com.google.protobuf.Any> deviceInfo_ =
+      java.util.Collections.emptyList();
+    private void ensureDeviceInfoIsMutable() {
+      if (!((bitField0_ & 0x00000001) != 0)) {
+        deviceInfo_ = new java.util.ArrayList<com.google.protobuf.Any>(deviceInfo_);
+        bitField0_ |= 0x00000001;
+       }
+    }
+
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        com.google.protobuf.Any, com.google.protobuf.Any.Builder, com.google.protobuf.AnyOrBuilder> deviceInfoBuilder_;
+
+    /**
+     * <pre>
+     * Other devices that are attached and relevant (e.g. GPUInfo).
+     * </pre>
+     *
+     * <code>repeated .google.protobuf.Any device_info = 4;</code>
+     */
+    public java.util.List<com.google.protobuf.Any> getDeviceInfoList() {
+      if (deviceInfoBuilder_ == null) {
+        return java.util.Collections.unmodifiableList(deviceInfo_);
+      } else {
+        return deviceInfoBuilder_.getMessageList();
+      }
+    }
+    /**
+     * <pre>
+     * Other devices that are attached and relevant (e.g. GPUInfo).
+     * </pre>
+     *
+     * <code>repeated .google.protobuf.Any device_info = 4;</code>
+     */
+    public int getDeviceInfoCount() {
+      if (deviceInfoBuilder_ == null) {
+        return deviceInfo_.size();
+      } else {
+        return deviceInfoBuilder_.getCount();
+      }
+    }
+    /**
+     * <pre>
+     * Other devices that are attached and relevant (e.g. GPUInfo).
+     * </pre>
+     *
+     * <code>repeated .google.protobuf.Any device_info = 4;</code>
+     */
+    public com.google.protobuf.Any getDeviceInfo(int index) {
+      if (deviceInfoBuilder_ == null) {
+        return deviceInfo_.get(index);
+      } else {
+        return deviceInfoBuilder_.getMessage(index);
+      }
+    }
+    /**
+     * <pre>
+     * Other devices that are attached and relevant (e.g. GPUInfo).
+     * </pre>
+     *
+     * <code>repeated .google.protobuf.Any device_info = 4;</code>
+     */
+    public Builder setDeviceInfo(
+        int index, com.google.protobuf.Any value) {
+      if (deviceInfoBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureDeviceInfoIsMutable();
+        deviceInfo_.set(index, value);
+        onChanged();
+      } else {
+        deviceInfoBuilder_.setMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Other devices that are attached and relevant (e.g. GPUInfo).
+     * </pre>
+     *
+     * <code>repeated .google.protobuf.Any device_info = 4;</code>
+     */
+    public Builder setDeviceInfo(
+        int index, com.google.protobuf.Any.Builder builderForValue) {
+      if (deviceInfoBuilder_ == null) {
+        ensureDeviceInfoIsMutable();
+        deviceInfo_.set(index, builderForValue.build());
+        onChanged();
+      } else {
+        deviceInfoBuilder_.setMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Other devices that are attached and relevant (e.g. GPUInfo).
+     * </pre>
+     *
+     * <code>repeated .google.protobuf.Any device_info = 4;</code>
+     */
+    public Builder addDeviceInfo(com.google.protobuf.Any value) {
+      if (deviceInfoBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureDeviceInfoIsMutable();
+        deviceInfo_.add(value);
+        onChanged();
+      } else {
+        deviceInfoBuilder_.addMessage(value);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Other devices that are attached and relevant (e.g. GPUInfo).
+     * </pre>
+     *
+     * <code>repeated .google.protobuf.Any device_info = 4;</code>
+     */
+    public Builder addDeviceInfo(
+        int index, com.google.protobuf.Any value) {
+      if (deviceInfoBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureDeviceInfoIsMutable();
+        deviceInfo_.add(index, value);
+        onChanged();
+      } else {
+        deviceInfoBuilder_.addMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Other devices that are attached and relevant (e.g. GPUInfo).
+     * </pre>
+     *
+     * <code>repeated .google.protobuf.Any device_info = 4;</code>
+     */
+    public Builder addDeviceInfo(
+        com.google.protobuf.Any.Builder builderForValue) {
+      if (deviceInfoBuilder_ == null) {
+        ensureDeviceInfoIsMutable();
+        deviceInfo_.add(builderForValue.build());
+        onChanged();
+      } else {
+        deviceInfoBuilder_.addMessage(builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Other devices that are attached and relevant (e.g. GPUInfo).
+     * </pre>
+     *
+     * <code>repeated .google.protobuf.Any device_info = 4;</code>
+     */
+    public Builder addDeviceInfo(
+        int index, com.google.protobuf.Any.Builder builderForValue) {
+      if (deviceInfoBuilder_ == null) {
+        ensureDeviceInfoIsMutable();
+        deviceInfo_.add(index, builderForValue.build());
+        onChanged();
+      } else {
+        deviceInfoBuilder_.addMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Other devices that are attached and relevant (e.g. GPUInfo).
+     * </pre>
+     *
+     * <code>repeated .google.protobuf.Any device_info = 4;</code>
+     */
+    public Builder addAllDeviceInfo(
+        java.lang.Iterable<? extends com.google.protobuf.Any> values) {
+      if (deviceInfoBuilder_ == null) {
+        ensureDeviceInfoIsMutable();
+        com.google.protobuf.AbstractMessageLite.Builder.addAll(
+            values, deviceInfo_);
+        onChanged();
+      } else {
+        deviceInfoBuilder_.addAllMessages(values);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Other devices that are attached and relevant (e.g. GPUInfo).
+     * </pre>
+     *
+     * <code>repeated .google.protobuf.Any device_info = 4;</code>
+     */
+    public Builder clearDeviceInfo() {
+      if (deviceInfoBuilder_ == null) {
+        deviceInfo_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000001);
+        onChanged();
+      } else {
+        deviceInfoBuilder_.clear();
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Other devices that are attached and relevant (e.g. GPUInfo).
+     * </pre>
+     *
+     * <code>repeated .google.protobuf.Any device_info = 4;</code>
+     */
+    public Builder removeDeviceInfo(int index) {
+      if (deviceInfoBuilder_ == null) {
+        ensureDeviceInfoIsMutable();
+        deviceInfo_.remove(index);
+        onChanged();
+      } else {
+        deviceInfoBuilder_.remove(index);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Other devices that are attached and relevant (e.g. GPUInfo).
+     * </pre>
+     *
+     * <code>repeated .google.protobuf.Any device_info = 4;</code>
+     */
+    public com.google.protobuf.Any.Builder getDeviceInfoBuilder(
+        int index) {
+      return getDeviceInfoFieldBuilder().getBuilder(index);
+    }
+    /**
+     * <pre>
+     * Other devices that are attached and relevant (e.g. GPUInfo).
+     * </pre>
+     *
+     * <code>repeated .google.protobuf.Any device_info = 4;</code>
+     */
+    public com.google.protobuf.AnyOrBuilder getDeviceInfoOrBuilder(
+        int index) {
+      if (deviceInfoBuilder_ == null) {
+        return deviceInfo_.get(index);  } else {
+        return deviceInfoBuilder_.getMessageOrBuilder(index);
+      }
+    }
+    /**
+     * <pre>
+     * Other devices that are attached and relevant (e.g. GPUInfo).
+     * </pre>
+     *
+     * <code>repeated .google.protobuf.Any device_info = 4;</code>
+     */
+    public java.util.List<? extends com.google.protobuf.AnyOrBuilder> 
+         getDeviceInfoOrBuilderList() {
+      if (deviceInfoBuilder_ != null) {
+        return deviceInfoBuilder_.getMessageOrBuilderList();
+      } else {
+        return java.util.Collections.unmodifiableList(deviceInfo_);
+      }
+    }
+    /**
+     * <pre>
+     * Other devices that are attached and relevant (e.g. GPUInfo).
+     * </pre>
+     *
+     * <code>repeated .google.protobuf.Any device_info = 4;</code>
+     */
+    public com.google.protobuf.Any.Builder addDeviceInfoBuilder() {
+      return getDeviceInfoFieldBuilder().addBuilder(
+          com.google.protobuf.Any.getDefaultInstance());
+    }
+    /**
+     * <pre>
+     * Other devices that are attached and relevant (e.g. GPUInfo).
+     * </pre>
+     *
+     * <code>repeated .google.protobuf.Any device_info = 4;</code>
+     */
+    public com.google.protobuf.Any.Builder addDeviceInfoBuilder(
+        int index) {
+      return getDeviceInfoFieldBuilder().addBuilder(
+          index, com.google.protobuf.Any.getDefaultInstance());
+    }
+    /**
+     * <pre>
+     * Other devices that are attached and relevant (e.g. GPUInfo).
+     * </pre>
+     *
+     * <code>repeated .google.protobuf.Any device_info = 4;</code>
+     */
+    public java.util.List<com.google.protobuf.Any.Builder> 
+         getDeviceInfoBuilderList() {
+      return getDeviceInfoFieldBuilder().getBuilderList();
+    }
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        com.google.protobuf.Any, com.google.protobuf.Any.Builder, com.google.protobuf.AnyOrBuilder> 
+        getDeviceInfoFieldBuilder() {
+      if (deviceInfoBuilder_ == null) {
+        deviceInfoBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
+            com.google.protobuf.Any, com.google.protobuf.Any.Builder, com.google.protobuf.AnyOrBuilder>(
+                deviceInfo_,
+                ((bitField0_ & 0x00000001) != 0),
+                getParentForChildren(),
+                isClean());
+        deviceInfo_ = null;
+      }
+      return deviceInfoBuilder_;
+    }
+
+    private java.util.List<org.tensorflow.proto.util.testlog.AvailableDeviceInfo> availableDeviceInfo_ =
+      java.util.Collections.emptyList();
+    private void ensureAvailableDeviceInfoIsMutable() {
+      if (!((bitField0_ & 0x00000002) != 0)) {
+        availableDeviceInfo_ = new java.util.ArrayList<org.tensorflow.proto.util.testlog.AvailableDeviceInfo>(availableDeviceInfo_);
+        bitField0_ |= 0x00000002;
+       }
+    }
+
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.util.testlog.AvailableDeviceInfo, org.tensorflow.proto.util.testlog.AvailableDeviceInfo.Builder, org.tensorflow.proto.util.testlog.AvailableDeviceInfoOrBuilder> availableDeviceInfoBuilder_;
+
+    /**
+     * <pre>
+     * Devices accessible to the test (e.g. as given by list_local_devices).
+     * </pre>
+     *
+     * <code>repeated .tensorflow.AvailableDeviceInfo available_device_info = 5;</code>
+     */
+    public java.util.List<org.tensorflow.proto.util.testlog.AvailableDeviceInfo> getAvailableDeviceInfoList() {
+      if (availableDeviceInfoBuilder_ == null) {
+        return java.util.Collections.unmodifiableList(availableDeviceInfo_);
+      } else {
+        return availableDeviceInfoBuilder_.getMessageList();
+      }
+    }
+    /**
+     * <pre>
+     * Devices accessible to the test (e.g. as given by list_local_devices).
+     * </pre>
+     *
+     * <code>repeated .tensorflow.AvailableDeviceInfo available_device_info = 5;</code>
+     */
+    public int getAvailableDeviceInfoCount() {
+      if (availableDeviceInfoBuilder_ == null) {
+        return availableDeviceInfo_.size();
+      } else {
+        return availableDeviceInfoBuilder_.getCount();
+      }
+    }
+    /**
+     * <pre>
+     * Devices accessible to the test (e.g. as given by list_local_devices).
+     * </pre>
+     *
+     * <code>repeated .tensorflow.AvailableDeviceInfo available_device_info = 5;</code>
+     */
+    public org.tensorflow.proto.util.testlog.AvailableDeviceInfo getAvailableDeviceInfo(int index) {
+      if (availableDeviceInfoBuilder_ == null) {
+        return availableDeviceInfo_.get(index);
+      } else {
+        return availableDeviceInfoBuilder_.getMessage(index);
+      }
+    }
+    /**
+     * <pre>
+     * Devices accessible to the test (e.g. as given by list_local_devices).
+     * </pre>
+     *
+     * <code>repeated .tensorflow.AvailableDeviceInfo available_device_info = 5;</code>
+     */
+    public Builder setAvailableDeviceInfo(
+        int index, org.tensorflow.proto.util.testlog.AvailableDeviceInfo value) {
+      if (availableDeviceInfoBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureAvailableDeviceInfoIsMutable();
+        availableDeviceInfo_.set(index, value);
+        onChanged();
+      } else {
+        availableDeviceInfoBuilder_.setMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Devices accessible to the test (e.g. as given by list_local_devices).
+     * </pre>
+     *
+     * <code>repeated .tensorflow.AvailableDeviceInfo available_device_info = 5;</code>
+     */
+    public Builder setAvailableDeviceInfo(
+        int index, org.tensorflow.proto.util.testlog.AvailableDeviceInfo.Builder builderForValue) {
+      if (availableDeviceInfoBuilder_ == null) {
+        ensureAvailableDeviceInfoIsMutable();
+        availableDeviceInfo_.set(index, builderForValue.build());
+        onChanged();
+      } else {
+        availableDeviceInfoBuilder_.setMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Devices accessible to the test (e.g. as given by list_local_devices).
+     * </pre>
+     *
+     * <code>repeated .tensorflow.AvailableDeviceInfo available_device_info = 5;</code>
+     */
+    public Builder addAvailableDeviceInfo(org.tensorflow.proto.util.testlog.AvailableDeviceInfo value) {
+      if (availableDeviceInfoBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureAvailableDeviceInfoIsMutable();
+        availableDeviceInfo_.add(value);
+        onChanged();
+      } else {
+        availableDeviceInfoBuilder_.addMessage(value);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Devices accessible to the test (e.g. as given by list_local_devices).
+     * </pre>
+     *
+     * <code>repeated .tensorflow.AvailableDeviceInfo available_device_info = 5;</code>
+     */
+    public Builder addAvailableDeviceInfo(
+        int index, org.tensorflow.proto.util.testlog.AvailableDeviceInfo value) {
+      if (availableDeviceInfoBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        ensureAvailableDeviceInfoIsMutable();
+        availableDeviceInfo_.add(index, value);
+        onChanged();
+      } else {
+        availableDeviceInfoBuilder_.addMessage(index, value);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Devices accessible to the test (e.g. as given by list_local_devices).
+     * </pre>
+     *
+     * <code>repeated .tensorflow.AvailableDeviceInfo available_device_info = 5;</code>
+     */
+    public Builder addAvailableDeviceInfo(
+        org.tensorflow.proto.util.testlog.AvailableDeviceInfo.Builder builderForValue) {
+      if (availableDeviceInfoBuilder_ == null) {
+        ensureAvailableDeviceInfoIsMutable();
+        availableDeviceInfo_.add(builderForValue.build());
+        onChanged();
+      } else {
+        availableDeviceInfoBuilder_.addMessage(builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Devices accessible to the test (e.g. as given by list_local_devices).
+     * </pre>
+     *
+     * <code>repeated .tensorflow.AvailableDeviceInfo available_device_info = 5;</code>
+     */
+    public Builder addAvailableDeviceInfo(
+        int index, org.tensorflow.proto.util.testlog.AvailableDeviceInfo.Builder builderForValue) {
+      if (availableDeviceInfoBuilder_ == null) {
+        ensureAvailableDeviceInfoIsMutable();
+        availableDeviceInfo_.add(index, builderForValue.build());
+        onChanged();
+      } else {
+        availableDeviceInfoBuilder_.addMessage(index, builderForValue.build());
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Devices accessible to the test (e.g. as given by list_local_devices).
+     * </pre>
+     *
+     * <code>repeated .tensorflow.AvailableDeviceInfo available_device_info = 5;</code>
+     */
+    public Builder addAllAvailableDeviceInfo(
+        java.lang.Iterable<? extends org.tensorflow.proto.util.testlog.AvailableDeviceInfo> values) {
+      if (availableDeviceInfoBuilder_ == null) {
+        ensureAvailableDeviceInfoIsMutable();
+        com.google.protobuf.AbstractMessageLite.Builder.addAll(
+            values, availableDeviceInfo_);
+        onChanged();
+      } else {
+        availableDeviceInfoBuilder_.addAllMessages(values);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Devices accessible to the test (e.g. as given by list_local_devices).
+     * </pre>
+     *
+     * <code>repeated .tensorflow.AvailableDeviceInfo available_device_info = 5;</code>
+     */
+    public Builder clearAvailableDeviceInfo() {
+      if (availableDeviceInfoBuilder_ == null) {
+        availableDeviceInfo_ = java.util.Collections.emptyList();
+        bitField0_ = (bitField0_ & ~0x00000002);
+        onChanged();
+      } else {
+        availableDeviceInfoBuilder_.clear();
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Devices accessible to the test (e.g. as given by list_local_devices).
+     * </pre>
+     *
+     * <code>repeated .tensorflow.AvailableDeviceInfo available_device_info = 5;</code>
+     */
+    public Builder removeAvailableDeviceInfo(int index) {
+      if (availableDeviceInfoBuilder_ == null) {
+        ensureAvailableDeviceInfoIsMutable();
+        availableDeviceInfo_.remove(index);
+        onChanged();
+      } else {
+        availableDeviceInfoBuilder_.remove(index);
+      }
+      return this;
+    }
+    /**
+     * <pre>
+     * Devices accessible to the test (e.g. as given by list_local_devices).
+     * </pre>
+     *
+     * <code>repeated .tensorflow.AvailableDeviceInfo available_device_info = 5;</code>
+     */
+    public org.tensorflow.proto.util.testlog.AvailableDeviceInfo.Builder getAvailableDeviceInfoBuilder(
+        int index) {
+      return getAvailableDeviceInfoFieldBuilder().getBuilder(index);
+    }
+    /**
+     * <pre>
+     * Devices accessible to the test (e.g. as given by list_local_devices).
+     * </pre>
+     *
+     * <code>repeated .tensorflow.AvailableDeviceInfo available_device_info = 5;</code>
+     */
+    public org.tensorflow.proto.util.testlog.AvailableDeviceInfoOrBuilder getAvailableDeviceInfoOrBuilder(
+        int index) {
+      if (availableDeviceInfoBuilder_ == null) {
+        return availableDeviceInfo_.get(index);  } else {
+        return availableDeviceInfoBuilder_.getMessageOrBuilder(index);
+      }
+    }
+    /**
+     * <pre>
+     * Devices accessible to the test (e.g. as given by list_local_devices).
+     * </pre>
+     *
+     * <code>repeated .tensorflow.AvailableDeviceInfo available_device_info = 5;</code>
+     */
+    public java.util.List<? extends org.tensorflow.proto.util.testlog.AvailableDeviceInfoOrBuilder> 
+         getAvailableDeviceInfoOrBuilderList() {
+      if (availableDeviceInfoBuilder_ != null) {
+        return availableDeviceInfoBuilder_.getMessageOrBuilderList();
+      } else {
+        return java.util.Collections.unmodifiableList(availableDeviceInfo_);
+      }
+    }
+    /**
+     * <pre>
+     * Devices accessible to the test (e.g. as given by list_local_devices).
+     * </pre>
+     *
+     * <code>repeated .tensorflow.AvailableDeviceInfo available_device_info = 5;</code>
+     */
+    public org.tensorflow.proto.util.testlog.AvailableDeviceInfo.Builder addAvailableDeviceInfoBuilder() {
+      return getAvailableDeviceInfoFieldBuilder().addBuilder(
+          org.tensorflow.proto.util.testlog.AvailableDeviceInfo.getDefaultInstance());
+    }
+    /**
+     * <pre>
+     * Devices accessible to the test (e.g. as given by list_local_devices).
+     * </pre>
+     *
+     * <code>repeated .tensorflow.AvailableDeviceInfo available_device_info = 5;</code>
+     */
+    public org.tensorflow.proto.util.testlog.AvailableDeviceInfo.Builder addAvailableDeviceInfoBuilder(
+        int index) {
+      return getAvailableDeviceInfoFieldBuilder().addBuilder(
+          index, org.tensorflow.proto.util.testlog.AvailableDeviceInfo.getDefaultInstance());
+    }
+    /**
+     * <pre>
+     * Devices accessible to the test (e.g. as given by list_local_devices).
+     * </pre>
+     *
+     * <code>repeated .tensorflow.AvailableDeviceInfo available_device_info = 5;</code>
+     */
+    public java.util.List<org.tensorflow.proto.util.testlog.AvailableDeviceInfo.Builder> 
+         getAvailableDeviceInfoBuilderList() {
+      return getAvailableDeviceInfoFieldBuilder().getBuilderList();
+    }
+    private com.google.protobuf.RepeatedFieldBuilderV3<
+        org.tensorflow.proto.util.testlog.AvailableDeviceInfo, org.tensorflow.proto.util.testlog.AvailableDeviceInfo.Builder, org.tensorflow.proto.util.testlog.AvailableDeviceInfoOrBuilder> 
+        getAvailableDeviceInfoFieldBuilder() {
+      if (availableDeviceInfoBuilder_ == null) {
+        availableDeviceInfoBuilder_ = new com.google.protobuf.RepeatedFieldBuilderV3<
+            org.tensorflow.proto.util.testlog.AvailableDeviceInfo, org.tensorflow.proto.util.testlog.AvailableDeviceInfo.Builder, org.tensorflow.proto.util.testlog.AvailableDeviceInfoOrBuilder>(
+                availableDeviceInfo_,
+                ((bitField0_ & 0x00000002) != 0),
+                getParentForChildren(),
+                isClean());
+        availableDeviceInfo_ = null;
+      }
+      return availableDeviceInfoBuilder_;
+    }
+
+    private org.tensorflow.proto.util.testlog.MemoryInfo memoryInfo_;
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.util.testlog.MemoryInfo, org.tensorflow.proto.util.testlog.MemoryInfo.Builder, org.tensorflow.proto.util.testlog.MemoryInfoOrBuilder> memoryInfoBuilder_;
+    /**
+     * <code>.tensorflow.MemoryInfo memory_info = 6;</code>
+     */
+    public boolean hasMemoryInfo() {
+      return memoryInfoBuilder_ != null || memoryInfo_ != null;
+    }
+    /**
+     * <code>.tensorflow.MemoryInfo memory_info = 6;</code>
+     */
+    public org.tensorflow.proto.util.testlog.MemoryInfo getMemoryInfo() {
+      if (memoryInfoBuilder_ == null) {
+        return memoryInfo_ == null ? org.tensorflow.proto.util.testlog.MemoryInfo.getDefaultInstance() : memoryInfo_;
+      } else {
+        return memoryInfoBuilder_.getMessage();
+      }
+    }
+    /**
+     * <code>.tensorflow.MemoryInfo memory_info = 6;</code>
+     */
+    public Builder setMemoryInfo(org.tensorflow.proto.util.testlog.MemoryInfo value) {
+      if (memoryInfoBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        memoryInfo_ = value;
+        onChanged();
+      } else {
+        memoryInfoBuilder_.setMessage(value);
+      }
+
+      return this;
+    }
+    /**
+     * <code>.tensorflow.MemoryInfo memory_info = 6;</code>
+     */
+    public Builder setMemoryInfo(
+        org.tensorflow.proto.util.testlog.MemoryInfo.Builder builderForValue) {
+      if (memoryInfoBuilder_ == null) {
+        memoryInfo_ = builderForValue.build();
+        onChanged();
+      } else {
+        memoryInfoBuilder_.setMessage(builderForValue.build());
+      }
+
+      return this;
+    }
+    /**
+     * <code>.tensorflow.MemoryInfo memory_info = 6;</code>
+     */
+    public Builder mergeMemoryInfo(org.tensorflow.proto.util.testlog.MemoryInfo value) {
+      if (memoryInfoBuilder_ == null) {
+        if (memoryInfo_ != null) {
+          memoryInfo_ =
+            org.tensorflow.proto.util.testlog.MemoryInfo.newBuilder(memoryInfo_).mergeFrom(value).buildPartial();
+        } else {
+          memoryInfo_ = value;
+        }
+        onChanged();
+      } else {
+        memoryInfoBuilder_.mergeFrom(value);
+      }
+
+      return this;
+    }
+    /**
+     * <code>.tensorflow.MemoryInfo memory_info = 6;</code>
+     */
+    public Builder clearMemoryInfo() {
+      if (memoryInfoBuilder_ == null) {
+        memoryInfo_ = null;
+        onChanged();
+      } else {
+        memoryInfo_ = null;
+        memoryInfoBuilder_ = null;
+      }
+
+      return this;
+    }
+    /**
+     * <code>.tensorflow.MemoryInfo memory_info = 6;</code>
+     */
+    public org.tensorflow.proto.util.testlog.MemoryInfo.Builder getMemoryInfoBuilder() {
+      
+      onChanged();
+      return getMemoryInfoFieldBuilder().getBuilder();
+    }
+    /**
+     * <code>.tensorflow.MemoryInfo memory_info = 6;</code>
+     */
+    public org.tensorflow.proto.util.testlog.MemoryInfoOrBuilder getMemoryInfoOrBuilder() {
+      if (memoryInfoBuilder_ != null) {
+        return memoryInfoBuilder_.getMessageOrBuilder();
+      } else {
+        return memoryInfo_ == null ?
+            org.tensorflow.proto.util.testlog.MemoryInfo.getDefaultInstance() : memoryInfo_;
+      }
+    }
+    /**
+     * <code>.tensorflow.MemoryInfo memory_info = 6;</code>
+     */
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.util.testlog.MemoryInfo, org.tensorflow.proto.util.testlog.MemoryInfo.Builder, org.tensorflow.proto.util.testlog.MemoryInfoOrBuilder> 
+        getMemoryInfoFieldBuilder() {
+      if (memoryInfoBuilder_ == null) {
+        memoryInfoBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+            org.tensorflow.proto.util.testlog.MemoryInfo, org.tensorflow.proto.util.testlog.MemoryInfo.Builder, org.tensorflow.proto.util.testlog.MemoryInfoOrBuilder>(
+                getMemoryInfo(),
+                getParentForChildren(),
+                isClean());
+        memoryInfo_ = null;
+      }
+      return memoryInfoBuilder_;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.MachineConfiguration)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.MachineConfiguration)
+  private static final org.tensorflow.proto.util.testlog.MachineConfiguration DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.util.testlog.MachineConfiguration();
+  }
+
+  public static org.tensorflow.proto.util.testlog.MachineConfiguration getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<MachineConfiguration>
+      PARSER = new com.google.protobuf.AbstractParser<MachineConfiguration>() {
+    @java.lang.Override
+    public MachineConfiguration parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new MachineConfiguration(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<MachineConfiguration> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<MachineConfiguration> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.util.testlog.MachineConfiguration getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/MachineConfigurationOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/MachineConfigurationOrBuilder.java
new file mode 100644
index 00000000000..8b0f5f6f884
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/MachineConfigurationOrBuilder.java
@@ -0,0 +1,196 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/util/test_log.proto
+
+package org.tensorflow.proto.util.testlog;
+
+public interface MachineConfigurationOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.MachineConfiguration)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <pre>
+   * Host name of machine that ran the benchmark.
+   * </pre>
+   *
+   * <code>string hostname = 1;</code>
+   */
+  java.lang.String getHostname();
+  /**
+   * <pre>
+   * Host name of machine that ran the benchmark.
+   * </pre>
+   *
+   * <code>string hostname = 1;</code>
+   */
+  com.google.protobuf.ByteString
+      getHostnameBytes();
+
+  /**
+   * <pre>
+   * Unique serial number of the machine.
+   * </pre>
+   *
+   * <code>string serial_identifier = 7;</code>
+   */
+  java.lang.String getSerialIdentifier();
+  /**
+   * <pre>
+   * Unique serial number of the machine.
+   * </pre>
+   *
+   * <code>string serial_identifier = 7;</code>
+   */
+  com.google.protobuf.ByteString
+      getSerialIdentifierBytes();
+
+  /**
+   * <pre>
+   * Additional platform information.
+   * </pre>
+   *
+   * <code>.tensorflow.PlatformInfo platform_info = 2;</code>
+   */
+  boolean hasPlatformInfo();
+  /**
+   * <pre>
+   * Additional platform information.
+   * </pre>
+   *
+   * <code>.tensorflow.PlatformInfo platform_info = 2;</code>
+   */
+  org.tensorflow.proto.util.testlog.PlatformInfo getPlatformInfo();
+  /**
+   * <pre>
+   * Additional platform information.
+   * </pre>
+   *
+   * <code>.tensorflow.PlatformInfo platform_info = 2;</code>
+   */
+  org.tensorflow.proto.util.testlog.PlatformInfoOrBuilder getPlatformInfoOrBuilder();
+
+  /**
+   * <pre>
+   * CPU Information.
+   * </pre>
+   *
+   * <code>.tensorflow.CPUInfo cpu_info = 3;</code>
+   */
+  boolean hasCpuInfo();
+  /**
+   * <pre>
+   * CPU Information.
+   * </pre>
+   *
+   * <code>.tensorflow.CPUInfo cpu_info = 3;</code>
+   */
+  org.tensorflow.proto.util.testlog.CPUInfo getCpuInfo();
+  /**
+   * <pre>
+   * CPU Information.
+   * </pre>
+   *
+   * <code>.tensorflow.CPUInfo cpu_info = 3;</code>
+   */
+  org.tensorflow.proto.util.testlog.CPUInfoOrBuilder getCpuInfoOrBuilder();
+
+  /**
+   * <pre>
+   * Other devices that are attached and relevant (e.g. GPUInfo).
+   * </pre>
+   *
+   * <code>repeated .google.protobuf.Any device_info = 4;</code>
+   */
+  java.util.List<com.google.protobuf.Any> 
+      getDeviceInfoList();
+  /**
+   * <pre>
+   * Other devices that are attached and relevant (e.g. GPUInfo).
+   * </pre>
+   *
+   * <code>repeated .google.protobuf.Any device_info = 4;</code>
+   */
+  com.google.protobuf.Any getDeviceInfo(int index);
+  /**
+   * <pre>
+   * Other devices that are attached and relevant (e.g. GPUInfo).
+   * </pre>
+   *
+   * <code>repeated .google.protobuf.Any device_info = 4;</code>
+   */
+  int getDeviceInfoCount();
+  /**
+   * <pre>
+   * Other devices that are attached and relevant (e.g. GPUInfo).
+   * </pre>
+   *
+   * <code>repeated .google.protobuf.Any device_info = 4;</code>
+   */
+  java.util.List<? extends com.google.protobuf.AnyOrBuilder> 
+      getDeviceInfoOrBuilderList();
+  /**
+   * <pre>
+   * Other devices that are attached and relevant (e.g. GPUInfo).
+   * </pre>
+   *
+   * <code>repeated .google.protobuf.Any device_info = 4;</code>
+   */
+  com.google.protobuf.AnyOrBuilder getDeviceInfoOrBuilder(
+      int index);
+
+  /**
+   * <pre>
+   * Devices accessible to the test (e.g. as given by list_local_devices).
+   * </pre>
+   *
+   * <code>repeated .tensorflow.AvailableDeviceInfo available_device_info = 5;</code>
+   */
+  java.util.List<org.tensorflow.proto.util.testlog.AvailableDeviceInfo> 
+      getAvailableDeviceInfoList();
+  /**
+   * <pre>
+   * Devices accessible to the test (e.g. as given by list_local_devices).
+   * </pre>
+   *
+   * <code>repeated .tensorflow.AvailableDeviceInfo available_device_info = 5;</code>
+   */
+  org.tensorflow.proto.util.testlog.AvailableDeviceInfo getAvailableDeviceInfo(int index);
+  /**
+   * <pre>
+   * Devices accessible to the test (e.g. as given by list_local_devices).
+   * </pre>
+   *
+   * <code>repeated .tensorflow.AvailableDeviceInfo available_device_info = 5;</code>
+   */
+  int getAvailableDeviceInfoCount();
+  /**
+   * <pre>
+   * Devices accessible to the test (e.g. as given by list_local_devices).
+   * </pre>
+   *
+   * <code>repeated .tensorflow.AvailableDeviceInfo available_device_info = 5;</code>
+   */
+  java.util.List<? extends org.tensorflow.proto.util.testlog.AvailableDeviceInfoOrBuilder> 
+      getAvailableDeviceInfoOrBuilderList();
+  /**
+   * <pre>
+   * Devices accessible to the test (e.g. as given by list_local_devices).
+   * </pre>
+   *
+   * <code>repeated .tensorflow.AvailableDeviceInfo available_device_info = 5;</code>
+   */
+  org.tensorflow.proto.util.testlog.AvailableDeviceInfoOrBuilder getAvailableDeviceInfoOrBuilder(
+      int index);
+
+  /**
+   * <code>.tensorflow.MemoryInfo memory_info = 6;</code>
+   */
+  boolean hasMemoryInfo();
+  /**
+   * <code>.tensorflow.MemoryInfo memory_info = 6;</code>
+   */
+  org.tensorflow.proto.util.testlog.MemoryInfo getMemoryInfo();
+  /**
+   * <code>.tensorflow.MemoryInfo memory_info = 6;</code>
+   */
+  org.tensorflow.proto.util.testlog.MemoryInfoOrBuilder getMemoryInfoOrBuilder();
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/MemoryInfo.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/MemoryInfo.java
new file mode 100644
index 00000000000..e28e00134cc
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/MemoryInfo.java
@@ -0,0 +1,567 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/util/test_log.proto
+
+package org.tensorflow.proto.util.testlog;
+
+/**
+ * Protobuf type {@code tensorflow.MemoryInfo}
+ */
+public  final class MemoryInfo extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.MemoryInfo)
+    MemoryInfoOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use MemoryInfo.newBuilder() to construct.
+  private MemoryInfo(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private MemoryInfo() {
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new MemoryInfo();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private MemoryInfo(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 8: {
+
+            total_ = input.readInt64();
+            break;
+          }
+          case 16: {
+
+            available_ = input.readInt64();
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_MemoryInfo_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_MemoryInfo_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.util.testlog.MemoryInfo.class, org.tensorflow.proto.util.testlog.MemoryInfo.Builder.class);
+  }
+
+  public static final int TOTAL_FIELD_NUMBER = 1;
+  private long total_;
+  /**
+   * <pre>
+   * Total virtual memory in bytes
+   * </pre>
+   *
+   * <code>int64 total = 1;</code>
+   */
+  public long getTotal() {
+    return total_;
+  }
+
+  public static final int AVAILABLE_FIELD_NUMBER = 2;
+  private long available_;
+  /**
+   * <pre>
+   * Immediately available memory in bytes
+   * </pre>
+   *
+   * <code>int64 available = 2;</code>
+   */
+  public long getAvailable() {
+    return available_;
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    if (total_ != 0L) {
+      output.writeInt64(1, total_);
+    }
+    if (available_ != 0L) {
+      output.writeInt64(2, available_);
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (total_ != 0L) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt64Size(1, total_);
+    }
+    if (available_ != 0L) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt64Size(2, available_);
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.util.testlog.MemoryInfo)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.util.testlog.MemoryInfo other = (org.tensorflow.proto.util.testlog.MemoryInfo) obj;
+
+    if (getTotal()
+        != other.getTotal()) return false;
+    if (getAvailable()
+        != other.getAvailable()) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    hash = (37 * hash) + TOTAL_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+        getTotal());
+    hash = (37 * hash) + AVAILABLE_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+        getAvailable());
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.util.testlog.MemoryInfo parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.testlog.MemoryInfo parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.testlog.MemoryInfo parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.testlog.MemoryInfo parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.testlog.MemoryInfo parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.testlog.MemoryInfo parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.testlog.MemoryInfo parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.testlog.MemoryInfo parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.testlog.MemoryInfo parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.testlog.MemoryInfo parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.testlog.MemoryInfo parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.testlog.MemoryInfo parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.util.testlog.MemoryInfo prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * Protobuf type {@code tensorflow.MemoryInfo}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.MemoryInfo)
+      org.tensorflow.proto.util.testlog.MemoryInfoOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_MemoryInfo_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_MemoryInfo_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.util.testlog.MemoryInfo.class, org.tensorflow.proto.util.testlog.MemoryInfo.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.util.testlog.MemoryInfo.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      total_ = 0L;
+
+      available_ = 0L;
+
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_MemoryInfo_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.testlog.MemoryInfo getDefaultInstanceForType() {
+      return org.tensorflow.proto.util.testlog.MemoryInfo.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.testlog.MemoryInfo build() {
+      org.tensorflow.proto.util.testlog.MemoryInfo result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.testlog.MemoryInfo buildPartial() {
+      org.tensorflow.proto.util.testlog.MemoryInfo result = new org.tensorflow.proto.util.testlog.MemoryInfo(this);
+      result.total_ = total_;
+      result.available_ = available_;
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.util.testlog.MemoryInfo) {
+        return mergeFrom((org.tensorflow.proto.util.testlog.MemoryInfo)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.util.testlog.MemoryInfo other) {
+      if (other == org.tensorflow.proto.util.testlog.MemoryInfo.getDefaultInstance()) return this;
+      if (other.getTotal() != 0L) {
+        setTotal(other.getTotal());
+      }
+      if (other.getAvailable() != 0L) {
+        setAvailable(other.getAvailable());
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.util.testlog.MemoryInfo parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.util.testlog.MemoryInfo) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+
+    private long total_ ;
+    /**
+     * <pre>
+     * Total virtual memory in bytes
+     * </pre>
+     *
+     * <code>int64 total = 1;</code>
+     */
+    public long getTotal() {
+      return total_;
+    }
+    /**
+     * <pre>
+     * Total virtual memory in bytes
+     * </pre>
+     *
+     * <code>int64 total = 1;</code>
+     */
+    public Builder setTotal(long value) {
+      
+      total_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Total virtual memory in bytes
+     * </pre>
+     *
+     * <code>int64 total = 1;</code>
+     */
+    public Builder clearTotal() {
+      
+      total_ = 0L;
+      onChanged();
+      return this;
+    }
+
+    private long available_ ;
+    /**
+     * <pre>
+     * Immediately available memory in bytes
+     * </pre>
+     *
+     * <code>int64 available = 2;</code>
+     */
+    public long getAvailable() {
+      return available_;
+    }
+    /**
+     * <pre>
+     * Immediately available memory in bytes
+     * </pre>
+     *
+     * <code>int64 available = 2;</code>
+     */
+    public Builder setAvailable(long value) {
+      
+      available_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Immediately available memory in bytes
+     * </pre>
+     *
+     * <code>int64 available = 2;</code>
+     */
+    public Builder clearAvailable() {
+      
+      available_ = 0L;
+      onChanged();
+      return this;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.MemoryInfo)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.MemoryInfo)
+  private static final org.tensorflow.proto.util.testlog.MemoryInfo DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.util.testlog.MemoryInfo();
+  }
+
+  public static org.tensorflow.proto.util.testlog.MemoryInfo getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<MemoryInfo>
+      PARSER = new com.google.protobuf.AbstractParser<MemoryInfo>() {
+    @java.lang.Override
+    public MemoryInfo parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new MemoryInfo(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<MemoryInfo> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<MemoryInfo> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.util.testlog.MemoryInfo getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/MemoryInfoOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/MemoryInfoOrBuilder.java
new file mode 100644
index 00000000000..1700a289e4f
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/MemoryInfoOrBuilder.java
@@ -0,0 +1,27 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/util/test_log.proto
+
+package org.tensorflow.proto.util.testlog;
+
+public interface MemoryInfoOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.MemoryInfo)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <pre>
+   * Total virtual memory in bytes
+   * </pre>
+   *
+   * <code>int64 total = 1;</code>
+   */
+  long getTotal();
+
+  /**
+   * <pre>
+   * Immediately available memory in bytes
+   * </pre>
+   *
+   * <code>int64 available = 2;</code>
+   */
+  long getAvailable();
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/MetricEntry.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/MetricEntry.java
new file mode 100644
index 00000000000..0ebe3ce31ab
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/MetricEntry.java
@@ -0,0 +1,1107 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/util/test_log.proto
+
+package org.tensorflow.proto.util.testlog;
+
+/**
+ * Protobuf type {@code tensorflow.MetricEntry}
+ */
+public  final class MetricEntry extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.MetricEntry)
+    MetricEntryOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use MetricEntry.newBuilder() to construct.
+  private MetricEntry(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private MetricEntry() {
+    name_ = "";
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new MetricEntry();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private MetricEntry(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 10: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            name_ = s;
+            break;
+          }
+          case 17: {
+
+            value_ = input.readDouble();
+            break;
+          }
+          case 26: {
+            com.google.protobuf.DoubleValue.Builder subBuilder = null;
+            if (minValue_ != null) {
+              subBuilder = minValue_.toBuilder();
+            }
+            minValue_ = input.readMessage(com.google.protobuf.DoubleValue.parser(), extensionRegistry);
+            if (subBuilder != null) {
+              subBuilder.mergeFrom(minValue_);
+              minValue_ = subBuilder.buildPartial();
+            }
+
+            break;
+          }
+          case 34: {
+            com.google.protobuf.DoubleValue.Builder subBuilder = null;
+            if (maxValue_ != null) {
+              subBuilder = maxValue_.toBuilder();
+            }
+            maxValue_ = input.readMessage(com.google.protobuf.DoubleValue.parser(), extensionRegistry);
+            if (subBuilder != null) {
+              subBuilder.mergeFrom(maxValue_);
+              maxValue_ = subBuilder.buildPartial();
+            }
+
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_MetricEntry_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_MetricEntry_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.util.testlog.MetricEntry.class, org.tensorflow.proto.util.testlog.MetricEntry.Builder.class);
+  }
+
+  public static final int NAME_FIELD_NUMBER = 1;
+  private volatile java.lang.Object name_;
+  /**
+   * <pre>
+   * Metric name
+   * </pre>
+   *
+   * <code>string name = 1;</code>
+   */
+  public java.lang.String getName() {
+    java.lang.Object ref = name_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      name_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * Metric name
+   * </pre>
+   *
+   * <code>string name = 1;</code>
+   */
+  public com.google.protobuf.ByteString
+      getNameBytes() {
+    java.lang.Object ref = name_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      name_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int VALUE_FIELD_NUMBER = 2;
+  private double value_;
+  /**
+   * <pre>
+   * Metric value
+   * </pre>
+   *
+   * <code>double value = 2;</code>
+   */
+  public double getValue() {
+    return value_;
+  }
+
+  public static final int MIN_VALUE_FIELD_NUMBER = 3;
+  private com.google.protobuf.DoubleValue minValue_;
+  /**
+   * <pre>
+   * The minimum acceptable value for the metric if specified
+   * </pre>
+   *
+   * <code>.google.protobuf.DoubleValue min_value = 3;</code>
+   */
+  public boolean hasMinValue() {
+    return minValue_ != null;
+  }
+  /**
+   * <pre>
+   * The minimum acceptable value for the metric if specified
+   * </pre>
+   *
+   * <code>.google.protobuf.DoubleValue min_value = 3;</code>
+   */
+  public com.google.protobuf.DoubleValue getMinValue() {
+    return minValue_ == null ? com.google.protobuf.DoubleValue.getDefaultInstance() : minValue_;
+  }
+  /**
+   * <pre>
+   * The minimum acceptable value for the metric if specified
+   * </pre>
+   *
+   * <code>.google.protobuf.DoubleValue min_value = 3;</code>
+   */
+  public com.google.protobuf.DoubleValueOrBuilder getMinValueOrBuilder() {
+    return getMinValue();
+  }
+
+  public static final int MAX_VALUE_FIELD_NUMBER = 4;
+  private com.google.protobuf.DoubleValue maxValue_;
+  /**
+   * <pre>
+   * The maximum acceptable value for the metric if specified
+   * </pre>
+   *
+   * <code>.google.protobuf.DoubleValue max_value = 4;</code>
+   */
+  public boolean hasMaxValue() {
+    return maxValue_ != null;
+  }
+  /**
+   * <pre>
+   * The maximum acceptable value for the metric if specified
+   * </pre>
+   *
+   * <code>.google.protobuf.DoubleValue max_value = 4;</code>
+   */
+  public com.google.protobuf.DoubleValue getMaxValue() {
+    return maxValue_ == null ? com.google.protobuf.DoubleValue.getDefaultInstance() : maxValue_;
+  }
+  /**
+   * <pre>
+   * The maximum acceptable value for the metric if specified
+   * </pre>
+   *
+   * <code>.google.protobuf.DoubleValue max_value = 4;</code>
+   */
+  public com.google.protobuf.DoubleValueOrBuilder getMaxValueOrBuilder() {
+    return getMaxValue();
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    if (!getNameBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 1, name_);
+    }
+    if (value_ != 0D) {
+      output.writeDouble(2, value_);
+    }
+    if (minValue_ != null) {
+      output.writeMessage(3, getMinValue());
+    }
+    if (maxValue_ != null) {
+      output.writeMessage(4, getMaxValue());
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (!getNameBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, name_);
+    }
+    if (value_ != 0D) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeDoubleSize(2, value_);
+    }
+    if (minValue_ != null) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(3, getMinValue());
+    }
+    if (maxValue_ != null) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(4, getMaxValue());
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.util.testlog.MetricEntry)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.util.testlog.MetricEntry other = (org.tensorflow.proto.util.testlog.MetricEntry) obj;
+
+    if (!getName()
+        .equals(other.getName())) return false;
+    if (java.lang.Double.doubleToLongBits(getValue())
+        != java.lang.Double.doubleToLongBits(
+            other.getValue())) return false;
+    if (hasMinValue() != other.hasMinValue()) return false;
+    if (hasMinValue()) {
+      if (!getMinValue()
+          .equals(other.getMinValue())) return false;
+    }
+    if (hasMaxValue() != other.hasMaxValue()) return false;
+    if (hasMaxValue()) {
+      if (!getMaxValue()
+          .equals(other.getMaxValue())) return false;
+    }
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    hash = (37 * hash) + NAME_FIELD_NUMBER;
+    hash = (53 * hash) + getName().hashCode();
+    hash = (37 * hash) + VALUE_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+        java.lang.Double.doubleToLongBits(getValue()));
+    if (hasMinValue()) {
+      hash = (37 * hash) + MIN_VALUE_FIELD_NUMBER;
+      hash = (53 * hash) + getMinValue().hashCode();
+    }
+    if (hasMaxValue()) {
+      hash = (37 * hash) + MAX_VALUE_FIELD_NUMBER;
+      hash = (53 * hash) + getMaxValue().hashCode();
+    }
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.util.testlog.MetricEntry parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.testlog.MetricEntry parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.testlog.MetricEntry parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.testlog.MetricEntry parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.testlog.MetricEntry parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.testlog.MetricEntry parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.testlog.MetricEntry parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.testlog.MetricEntry parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.testlog.MetricEntry parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.testlog.MetricEntry parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.testlog.MetricEntry parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.testlog.MetricEntry parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.util.testlog.MetricEntry prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * Protobuf type {@code tensorflow.MetricEntry}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.MetricEntry)
+      org.tensorflow.proto.util.testlog.MetricEntryOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_MetricEntry_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_MetricEntry_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.util.testlog.MetricEntry.class, org.tensorflow.proto.util.testlog.MetricEntry.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.util.testlog.MetricEntry.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      name_ = "";
+
+      value_ = 0D;
+
+      if (minValueBuilder_ == null) {
+        minValue_ = null;
+      } else {
+        minValue_ = null;
+        minValueBuilder_ = null;
+      }
+      if (maxValueBuilder_ == null) {
+        maxValue_ = null;
+      } else {
+        maxValue_ = null;
+        maxValueBuilder_ = null;
+      }
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_MetricEntry_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.testlog.MetricEntry getDefaultInstanceForType() {
+      return org.tensorflow.proto.util.testlog.MetricEntry.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.testlog.MetricEntry build() {
+      org.tensorflow.proto.util.testlog.MetricEntry result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.testlog.MetricEntry buildPartial() {
+      org.tensorflow.proto.util.testlog.MetricEntry result = new org.tensorflow.proto.util.testlog.MetricEntry(this);
+      result.name_ = name_;
+      result.value_ = value_;
+      if (minValueBuilder_ == null) {
+        result.minValue_ = minValue_;
+      } else {
+        result.minValue_ = minValueBuilder_.build();
+      }
+      if (maxValueBuilder_ == null) {
+        result.maxValue_ = maxValue_;
+      } else {
+        result.maxValue_ = maxValueBuilder_.build();
+      }
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.util.testlog.MetricEntry) {
+        return mergeFrom((org.tensorflow.proto.util.testlog.MetricEntry)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.util.testlog.MetricEntry other) {
+      if (other == org.tensorflow.proto.util.testlog.MetricEntry.getDefaultInstance()) return this;
+      if (!other.getName().isEmpty()) {
+        name_ = other.name_;
+        onChanged();
+      }
+      if (other.getValue() != 0D) {
+        setValue(other.getValue());
+      }
+      if (other.hasMinValue()) {
+        mergeMinValue(other.getMinValue());
+      }
+      if (other.hasMaxValue()) {
+        mergeMaxValue(other.getMaxValue());
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.util.testlog.MetricEntry parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.util.testlog.MetricEntry) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+
+    private java.lang.Object name_ = "";
+    /**
+     * <pre>
+     * Metric name
+     * </pre>
+     *
+     * <code>string name = 1;</code>
+     */
+    public java.lang.String getName() {
+      java.lang.Object ref = name_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        name_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Metric name
+     * </pre>
+     *
+     * <code>string name = 1;</code>
+     */
+    public com.google.protobuf.ByteString
+        getNameBytes() {
+      java.lang.Object ref = name_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        name_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Metric name
+     * </pre>
+     *
+     * <code>string name = 1;</code>
+     */
+    public Builder setName(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      name_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Metric name
+     * </pre>
+     *
+     * <code>string name = 1;</code>
+     */
+    public Builder clearName() {
+      
+      name_ = getDefaultInstance().getName();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Metric name
+     * </pre>
+     *
+     * <code>string name = 1;</code>
+     */
+    public Builder setNameBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      name_ = value;
+      onChanged();
+      return this;
+    }
+
+    private double value_ ;
+    /**
+     * <pre>
+     * Metric value
+     * </pre>
+     *
+     * <code>double value = 2;</code>
+     */
+    public double getValue() {
+      return value_;
+    }
+    /**
+     * <pre>
+     * Metric value
+     * </pre>
+     *
+     * <code>double value = 2;</code>
+     */
+    public Builder setValue(double value) {
+      
+      value_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Metric value
+     * </pre>
+     *
+     * <code>double value = 2;</code>
+     */
+    public Builder clearValue() {
+      
+      value_ = 0D;
+      onChanged();
+      return this;
+    }
+
+    private com.google.protobuf.DoubleValue minValue_;
+    private com.google.protobuf.SingleFieldBuilderV3<
+        com.google.protobuf.DoubleValue, com.google.protobuf.DoubleValue.Builder, com.google.protobuf.DoubleValueOrBuilder> minValueBuilder_;
+    /**
+     * <pre>
+     * The minimum acceptable value for the metric if specified
+     * </pre>
+     *
+     * <code>.google.protobuf.DoubleValue min_value = 3;</code>
+     */
+    public boolean hasMinValue() {
+      return minValueBuilder_ != null || minValue_ != null;
+    }
+    /**
+     * <pre>
+     * The minimum acceptable value for the metric if specified
+     * </pre>
+     *
+     * <code>.google.protobuf.DoubleValue min_value = 3;</code>
+     */
+    public com.google.protobuf.DoubleValue getMinValue() {
+      if (minValueBuilder_ == null) {
+        return minValue_ == null ? com.google.protobuf.DoubleValue.getDefaultInstance() : minValue_;
+      } else {
+        return minValueBuilder_.getMessage();
+      }
+    }
+    /**
+     * <pre>
+     * The minimum acceptable value for the metric if specified
+     * </pre>
+     *
+     * <code>.google.protobuf.DoubleValue min_value = 3;</code>
+     */
+    public Builder setMinValue(com.google.protobuf.DoubleValue value) {
+      if (minValueBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        minValue_ = value;
+        onChanged();
+      } else {
+        minValueBuilder_.setMessage(value);
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * The minimum acceptable value for the metric if specified
+     * </pre>
+     *
+     * <code>.google.protobuf.DoubleValue min_value = 3;</code>
+     */
+    public Builder setMinValue(
+        com.google.protobuf.DoubleValue.Builder builderForValue) {
+      if (minValueBuilder_ == null) {
+        minValue_ = builderForValue.build();
+        onChanged();
+      } else {
+        minValueBuilder_.setMessage(builderForValue.build());
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * The minimum acceptable value for the metric if specified
+     * </pre>
+     *
+     * <code>.google.protobuf.DoubleValue min_value = 3;</code>
+     */
+    public Builder mergeMinValue(com.google.protobuf.DoubleValue value) {
+      if (minValueBuilder_ == null) {
+        if (minValue_ != null) {
+          minValue_ =
+            com.google.protobuf.DoubleValue.newBuilder(minValue_).mergeFrom(value).buildPartial();
+        } else {
+          minValue_ = value;
+        }
+        onChanged();
+      } else {
+        minValueBuilder_.mergeFrom(value);
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * The minimum acceptable value for the metric if specified
+     * </pre>
+     *
+     * <code>.google.protobuf.DoubleValue min_value = 3;</code>
+     */
+    public Builder clearMinValue() {
+      if (minValueBuilder_ == null) {
+        minValue_ = null;
+        onChanged();
+      } else {
+        minValue_ = null;
+        minValueBuilder_ = null;
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * The minimum acceptable value for the metric if specified
+     * </pre>
+     *
+     * <code>.google.protobuf.DoubleValue min_value = 3;</code>
+     */
+    public com.google.protobuf.DoubleValue.Builder getMinValueBuilder() {
+      
+      onChanged();
+      return getMinValueFieldBuilder().getBuilder();
+    }
+    /**
+     * <pre>
+     * The minimum acceptable value for the metric if specified
+     * </pre>
+     *
+     * <code>.google.protobuf.DoubleValue min_value = 3;</code>
+     */
+    public com.google.protobuf.DoubleValueOrBuilder getMinValueOrBuilder() {
+      if (minValueBuilder_ != null) {
+        return minValueBuilder_.getMessageOrBuilder();
+      } else {
+        return minValue_ == null ?
+            com.google.protobuf.DoubleValue.getDefaultInstance() : minValue_;
+      }
+    }
+    /**
+     * <pre>
+     * The minimum acceptable value for the metric if specified
+     * </pre>
+     *
+     * <code>.google.protobuf.DoubleValue min_value = 3;</code>
+     */
+    private com.google.protobuf.SingleFieldBuilderV3<
+        com.google.protobuf.DoubleValue, com.google.protobuf.DoubleValue.Builder, com.google.protobuf.DoubleValueOrBuilder> 
+        getMinValueFieldBuilder() {
+      if (minValueBuilder_ == null) {
+        minValueBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+            com.google.protobuf.DoubleValue, com.google.protobuf.DoubleValue.Builder, com.google.protobuf.DoubleValueOrBuilder>(
+                getMinValue(),
+                getParentForChildren(),
+                isClean());
+        minValue_ = null;
+      }
+      return minValueBuilder_;
+    }
+
+    private com.google.protobuf.DoubleValue maxValue_;
+    private com.google.protobuf.SingleFieldBuilderV3<
+        com.google.protobuf.DoubleValue, com.google.protobuf.DoubleValue.Builder, com.google.protobuf.DoubleValueOrBuilder> maxValueBuilder_;
+    /**
+     * <pre>
+     * The maximum acceptable value for the metric if specified
+     * </pre>
+     *
+     * <code>.google.protobuf.DoubleValue max_value = 4;</code>
+     */
+    public boolean hasMaxValue() {
+      return maxValueBuilder_ != null || maxValue_ != null;
+    }
+    /**
+     * <pre>
+     * The maximum acceptable value for the metric if specified
+     * </pre>
+     *
+     * <code>.google.protobuf.DoubleValue max_value = 4;</code>
+     */
+    public com.google.protobuf.DoubleValue getMaxValue() {
+      if (maxValueBuilder_ == null) {
+        return maxValue_ == null ? com.google.protobuf.DoubleValue.getDefaultInstance() : maxValue_;
+      } else {
+        return maxValueBuilder_.getMessage();
+      }
+    }
+    /**
+     * <pre>
+     * The maximum acceptable value for the metric if specified
+     * </pre>
+     *
+     * <code>.google.protobuf.DoubleValue max_value = 4;</code>
+     */
+    public Builder setMaxValue(com.google.protobuf.DoubleValue value) {
+      if (maxValueBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        maxValue_ = value;
+        onChanged();
+      } else {
+        maxValueBuilder_.setMessage(value);
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * The maximum acceptable value for the metric if specified
+     * </pre>
+     *
+     * <code>.google.protobuf.DoubleValue max_value = 4;</code>
+     */
+    public Builder setMaxValue(
+        com.google.protobuf.DoubleValue.Builder builderForValue) {
+      if (maxValueBuilder_ == null) {
+        maxValue_ = builderForValue.build();
+        onChanged();
+      } else {
+        maxValueBuilder_.setMessage(builderForValue.build());
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * The maximum acceptable value for the metric if specified
+     * </pre>
+     *
+     * <code>.google.protobuf.DoubleValue max_value = 4;</code>
+     */
+    public Builder mergeMaxValue(com.google.protobuf.DoubleValue value) {
+      if (maxValueBuilder_ == null) {
+        if (maxValue_ != null) {
+          maxValue_ =
+            com.google.protobuf.DoubleValue.newBuilder(maxValue_).mergeFrom(value).buildPartial();
+        } else {
+          maxValue_ = value;
+        }
+        onChanged();
+      } else {
+        maxValueBuilder_.mergeFrom(value);
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * The maximum acceptable value for the metric if specified
+     * </pre>
+     *
+     * <code>.google.protobuf.DoubleValue max_value = 4;</code>
+     */
+    public Builder clearMaxValue() {
+      if (maxValueBuilder_ == null) {
+        maxValue_ = null;
+        onChanged();
+      } else {
+        maxValue_ = null;
+        maxValueBuilder_ = null;
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * The maximum acceptable value for the metric if specified
+     * </pre>
+     *
+     * <code>.google.protobuf.DoubleValue max_value = 4;</code>
+     */
+    public com.google.protobuf.DoubleValue.Builder getMaxValueBuilder() {
+      
+      onChanged();
+      return getMaxValueFieldBuilder().getBuilder();
+    }
+    /**
+     * <pre>
+     * The maximum acceptable value for the metric if specified
+     * </pre>
+     *
+     * <code>.google.protobuf.DoubleValue max_value = 4;</code>
+     */
+    public com.google.protobuf.DoubleValueOrBuilder getMaxValueOrBuilder() {
+      if (maxValueBuilder_ != null) {
+        return maxValueBuilder_.getMessageOrBuilder();
+      } else {
+        return maxValue_ == null ?
+            com.google.protobuf.DoubleValue.getDefaultInstance() : maxValue_;
+      }
+    }
+    /**
+     * <pre>
+     * The maximum acceptable value for the metric if specified
+     * </pre>
+     *
+     * <code>.google.protobuf.DoubleValue max_value = 4;</code>
+     */
+    private com.google.protobuf.SingleFieldBuilderV3<
+        com.google.protobuf.DoubleValue, com.google.protobuf.DoubleValue.Builder, com.google.protobuf.DoubleValueOrBuilder> 
+        getMaxValueFieldBuilder() {
+      if (maxValueBuilder_ == null) {
+        maxValueBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+            com.google.protobuf.DoubleValue, com.google.protobuf.DoubleValue.Builder, com.google.protobuf.DoubleValueOrBuilder>(
+                getMaxValue(),
+                getParentForChildren(),
+                isClean());
+        maxValue_ = null;
+      }
+      return maxValueBuilder_;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.MetricEntry)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.MetricEntry)
+  private static final org.tensorflow.proto.util.testlog.MetricEntry DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.util.testlog.MetricEntry();
+  }
+
+  public static org.tensorflow.proto.util.testlog.MetricEntry getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<MetricEntry>
+      PARSER = new com.google.protobuf.AbstractParser<MetricEntry>() {
+    @java.lang.Override
+    public MetricEntry parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new MetricEntry(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<MetricEntry> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<MetricEntry> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.util.testlog.MetricEntry getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/MetricEntryOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/MetricEntryOrBuilder.java
new file mode 100644
index 00000000000..eeb12bca8a9
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/MetricEntryOrBuilder.java
@@ -0,0 +1,86 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/util/test_log.proto
+
+package org.tensorflow.proto.util.testlog;
+
+public interface MetricEntryOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.MetricEntry)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <pre>
+   * Metric name
+   * </pre>
+   *
+   * <code>string name = 1;</code>
+   */
+  java.lang.String getName();
+  /**
+   * <pre>
+   * Metric name
+   * </pre>
+   *
+   * <code>string name = 1;</code>
+   */
+  com.google.protobuf.ByteString
+      getNameBytes();
+
+  /**
+   * <pre>
+   * Metric value
+   * </pre>
+   *
+   * <code>double value = 2;</code>
+   */
+  double getValue();
+
+  /**
+   * <pre>
+   * The minimum acceptable value for the metric if specified
+   * </pre>
+   *
+   * <code>.google.protobuf.DoubleValue min_value = 3;</code>
+   */
+  boolean hasMinValue();
+  /**
+   * <pre>
+   * The minimum acceptable value for the metric if specified
+   * </pre>
+   *
+   * <code>.google.protobuf.DoubleValue min_value = 3;</code>
+   */
+  com.google.protobuf.DoubleValue getMinValue();
+  /**
+   * <pre>
+   * The minimum acceptable value for the metric if specified
+   * </pre>
+   *
+   * <code>.google.protobuf.DoubleValue min_value = 3;</code>
+   */
+  com.google.protobuf.DoubleValueOrBuilder getMinValueOrBuilder();
+
+  /**
+   * <pre>
+   * The maximum acceptable value for the metric if specified
+   * </pre>
+   *
+   * <code>.google.protobuf.DoubleValue max_value = 4;</code>
+   */
+  boolean hasMaxValue();
+  /**
+   * <pre>
+   * The maximum acceptable value for the metric if specified
+   * </pre>
+   *
+   * <code>.google.protobuf.DoubleValue max_value = 4;</code>
+   */
+  com.google.protobuf.DoubleValue getMaxValue();
+  /**
+   * <pre>
+   * The maximum acceptable value for the metric if specified
+   * </pre>
+   *
+   * <code>.google.protobuf.DoubleValue max_value = 4;</code>
+   */
+  com.google.protobuf.DoubleValueOrBuilder getMaxValueOrBuilder();
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/PlatformInfo.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/PlatformInfo.java
new file mode 100644
index 00000000000..efe3a52d92e
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/PlatformInfo.java
@@ -0,0 +1,1349 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/util/test_log.proto
+
+package org.tensorflow.proto.util.testlog;
+
+/**
+ * Protobuf type {@code tensorflow.PlatformInfo}
+ */
+public  final class PlatformInfo extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.PlatformInfo)
+    PlatformInfoOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use PlatformInfo.newBuilder() to construct.
+  private PlatformInfo(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private PlatformInfo() {
+    bits_ = "";
+    linkage_ = "";
+    machine_ = "";
+    release_ = "";
+    system_ = "";
+    version_ = "";
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new PlatformInfo();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private PlatformInfo(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 10: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            bits_ = s;
+            break;
+          }
+          case 18: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            linkage_ = s;
+            break;
+          }
+          case 26: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            machine_ = s;
+            break;
+          }
+          case 34: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            release_ = s;
+            break;
+          }
+          case 42: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            system_ = s;
+            break;
+          }
+          case 50: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            version_ = s;
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_PlatformInfo_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_PlatformInfo_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.util.testlog.PlatformInfo.class, org.tensorflow.proto.util.testlog.PlatformInfo.Builder.class);
+  }
+
+  public static final int BITS_FIELD_NUMBER = 1;
+  private volatile java.lang.Object bits_;
+  /**
+   * <pre>
+   * e.g. '64bit'
+   * </pre>
+   *
+   * <code>string bits = 1;</code>
+   */
+  public java.lang.String getBits() {
+    java.lang.Object ref = bits_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      bits_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * e.g. '64bit'
+   * </pre>
+   *
+   * <code>string bits = 1;</code>
+   */
+  public com.google.protobuf.ByteString
+      getBitsBytes() {
+    java.lang.Object ref = bits_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      bits_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int LINKAGE_FIELD_NUMBER = 2;
+  private volatile java.lang.Object linkage_;
+  /**
+   * <pre>
+   * e.g. 'ELF'
+   * </pre>
+   *
+   * <code>string linkage = 2;</code>
+   */
+  public java.lang.String getLinkage() {
+    java.lang.Object ref = linkage_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      linkage_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * e.g. 'ELF'
+   * </pre>
+   *
+   * <code>string linkage = 2;</code>
+   */
+  public com.google.protobuf.ByteString
+      getLinkageBytes() {
+    java.lang.Object ref = linkage_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      linkage_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int MACHINE_FIELD_NUMBER = 3;
+  private volatile java.lang.Object machine_;
+  /**
+   * <pre>
+   * e.g. 'i386'
+   * </pre>
+   *
+   * <code>string machine = 3;</code>
+   */
+  public java.lang.String getMachine() {
+    java.lang.Object ref = machine_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      machine_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * e.g. 'i386'
+   * </pre>
+   *
+   * <code>string machine = 3;</code>
+   */
+  public com.google.protobuf.ByteString
+      getMachineBytes() {
+    java.lang.Object ref = machine_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      machine_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int RELEASE_FIELD_NUMBER = 4;
+  private volatile java.lang.Object release_;
+  /**
+   * <pre>
+   * e.g. '3.13.0-76-generic'
+   * </pre>
+   *
+   * <code>string release = 4;</code>
+   */
+  public java.lang.String getRelease() {
+    java.lang.Object ref = release_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      release_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * e.g. '3.13.0-76-generic'
+   * </pre>
+   *
+   * <code>string release = 4;</code>
+   */
+  public com.google.protobuf.ByteString
+      getReleaseBytes() {
+    java.lang.Object ref = release_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      release_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int SYSTEM_FIELD_NUMBER = 5;
+  private volatile java.lang.Object system_;
+  /**
+   * <pre>
+   * e.g. 'Linux'
+   * </pre>
+   *
+   * <code>string system = 5;</code>
+   */
+  public java.lang.String getSystem() {
+    java.lang.Object ref = system_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      system_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * e.g. 'Linux'
+   * </pre>
+   *
+   * <code>string system = 5;</code>
+   */
+  public com.google.protobuf.ByteString
+      getSystemBytes() {
+    java.lang.Object ref = system_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      system_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int VERSION_FIELD_NUMBER = 6;
+  private volatile java.lang.Object version_;
+  /**
+   * <pre>
+   * e.g. '#120-Ubuntu SMP Mon Jan 18 15:59:10 UTC 2016'
+   * </pre>
+   *
+   * <code>string version = 6;</code>
+   */
+  public java.lang.String getVersion() {
+    java.lang.Object ref = version_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      version_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * e.g. '#120-Ubuntu SMP Mon Jan 18 15:59:10 UTC 2016'
+   * </pre>
+   *
+   * <code>string version = 6;</code>
+   */
+  public com.google.protobuf.ByteString
+      getVersionBytes() {
+    java.lang.Object ref = version_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      version_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    if (!getBitsBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 1, bits_);
+    }
+    if (!getLinkageBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 2, linkage_);
+    }
+    if (!getMachineBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 3, machine_);
+    }
+    if (!getReleaseBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 4, release_);
+    }
+    if (!getSystemBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 5, system_);
+    }
+    if (!getVersionBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 6, version_);
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (!getBitsBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, bits_);
+    }
+    if (!getLinkageBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(2, linkage_);
+    }
+    if (!getMachineBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(3, machine_);
+    }
+    if (!getReleaseBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(4, release_);
+    }
+    if (!getSystemBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(5, system_);
+    }
+    if (!getVersionBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(6, version_);
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.util.testlog.PlatformInfo)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.util.testlog.PlatformInfo other = (org.tensorflow.proto.util.testlog.PlatformInfo) obj;
+
+    if (!getBits()
+        .equals(other.getBits())) return false;
+    if (!getLinkage()
+        .equals(other.getLinkage())) return false;
+    if (!getMachine()
+        .equals(other.getMachine())) return false;
+    if (!getRelease()
+        .equals(other.getRelease())) return false;
+    if (!getSystem()
+        .equals(other.getSystem())) return false;
+    if (!getVersion()
+        .equals(other.getVersion())) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    hash = (37 * hash) + BITS_FIELD_NUMBER;
+    hash = (53 * hash) + getBits().hashCode();
+    hash = (37 * hash) + LINKAGE_FIELD_NUMBER;
+    hash = (53 * hash) + getLinkage().hashCode();
+    hash = (37 * hash) + MACHINE_FIELD_NUMBER;
+    hash = (53 * hash) + getMachine().hashCode();
+    hash = (37 * hash) + RELEASE_FIELD_NUMBER;
+    hash = (53 * hash) + getRelease().hashCode();
+    hash = (37 * hash) + SYSTEM_FIELD_NUMBER;
+    hash = (53 * hash) + getSystem().hashCode();
+    hash = (37 * hash) + VERSION_FIELD_NUMBER;
+    hash = (53 * hash) + getVersion().hashCode();
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.util.testlog.PlatformInfo parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.testlog.PlatformInfo parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.testlog.PlatformInfo parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.testlog.PlatformInfo parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.testlog.PlatformInfo parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.testlog.PlatformInfo parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.testlog.PlatformInfo parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.testlog.PlatformInfo parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.testlog.PlatformInfo parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.testlog.PlatformInfo parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.testlog.PlatformInfo parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.testlog.PlatformInfo parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.util.testlog.PlatformInfo prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * Protobuf type {@code tensorflow.PlatformInfo}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.PlatformInfo)
+      org.tensorflow.proto.util.testlog.PlatformInfoOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_PlatformInfo_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_PlatformInfo_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.util.testlog.PlatformInfo.class, org.tensorflow.proto.util.testlog.PlatformInfo.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.util.testlog.PlatformInfo.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      bits_ = "";
+
+      linkage_ = "";
+
+      machine_ = "";
+
+      release_ = "";
+
+      system_ = "";
+
+      version_ = "";
+
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_PlatformInfo_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.testlog.PlatformInfo getDefaultInstanceForType() {
+      return org.tensorflow.proto.util.testlog.PlatformInfo.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.testlog.PlatformInfo build() {
+      org.tensorflow.proto.util.testlog.PlatformInfo result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.testlog.PlatformInfo buildPartial() {
+      org.tensorflow.proto.util.testlog.PlatformInfo result = new org.tensorflow.proto.util.testlog.PlatformInfo(this);
+      result.bits_ = bits_;
+      result.linkage_ = linkage_;
+      result.machine_ = machine_;
+      result.release_ = release_;
+      result.system_ = system_;
+      result.version_ = version_;
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.util.testlog.PlatformInfo) {
+        return mergeFrom((org.tensorflow.proto.util.testlog.PlatformInfo)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.util.testlog.PlatformInfo other) {
+      if (other == org.tensorflow.proto.util.testlog.PlatformInfo.getDefaultInstance()) return this;
+      if (!other.getBits().isEmpty()) {
+        bits_ = other.bits_;
+        onChanged();
+      }
+      if (!other.getLinkage().isEmpty()) {
+        linkage_ = other.linkage_;
+        onChanged();
+      }
+      if (!other.getMachine().isEmpty()) {
+        machine_ = other.machine_;
+        onChanged();
+      }
+      if (!other.getRelease().isEmpty()) {
+        release_ = other.release_;
+        onChanged();
+      }
+      if (!other.getSystem().isEmpty()) {
+        system_ = other.system_;
+        onChanged();
+      }
+      if (!other.getVersion().isEmpty()) {
+        version_ = other.version_;
+        onChanged();
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.util.testlog.PlatformInfo parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.util.testlog.PlatformInfo) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+
+    private java.lang.Object bits_ = "";
+    /**
+     * <pre>
+     * e.g. '64bit'
+     * </pre>
+     *
+     * <code>string bits = 1;</code>
+     */
+    public java.lang.String getBits() {
+      java.lang.Object ref = bits_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        bits_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * e.g. '64bit'
+     * </pre>
+     *
+     * <code>string bits = 1;</code>
+     */
+    public com.google.protobuf.ByteString
+        getBitsBytes() {
+      java.lang.Object ref = bits_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        bits_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * e.g. '64bit'
+     * </pre>
+     *
+     * <code>string bits = 1;</code>
+     */
+    public Builder setBits(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      bits_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * e.g. '64bit'
+     * </pre>
+     *
+     * <code>string bits = 1;</code>
+     */
+    public Builder clearBits() {
+      
+      bits_ = getDefaultInstance().getBits();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * e.g. '64bit'
+     * </pre>
+     *
+     * <code>string bits = 1;</code>
+     */
+    public Builder setBitsBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      bits_ = value;
+      onChanged();
+      return this;
+    }
+
+    private java.lang.Object linkage_ = "";
+    /**
+     * <pre>
+     * e.g. 'ELF'
+     * </pre>
+     *
+     * <code>string linkage = 2;</code>
+     */
+    public java.lang.String getLinkage() {
+      java.lang.Object ref = linkage_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        linkage_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * e.g. 'ELF'
+     * </pre>
+     *
+     * <code>string linkage = 2;</code>
+     */
+    public com.google.protobuf.ByteString
+        getLinkageBytes() {
+      java.lang.Object ref = linkage_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        linkage_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * e.g. 'ELF'
+     * </pre>
+     *
+     * <code>string linkage = 2;</code>
+     */
+    public Builder setLinkage(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      linkage_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * e.g. 'ELF'
+     * </pre>
+     *
+     * <code>string linkage = 2;</code>
+     */
+    public Builder clearLinkage() {
+      
+      linkage_ = getDefaultInstance().getLinkage();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * e.g. 'ELF'
+     * </pre>
+     *
+     * <code>string linkage = 2;</code>
+     */
+    public Builder setLinkageBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      linkage_ = value;
+      onChanged();
+      return this;
+    }
+
+    private java.lang.Object machine_ = "";
+    /**
+     * <pre>
+     * e.g. 'i386'
+     * </pre>
+     *
+     * <code>string machine = 3;</code>
+     */
+    public java.lang.String getMachine() {
+      java.lang.Object ref = machine_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        machine_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * e.g. 'i386'
+     * </pre>
+     *
+     * <code>string machine = 3;</code>
+     */
+    public com.google.protobuf.ByteString
+        getMachineBytes() {
+      java.lang.Object ref = machine_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        machine_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * e.g. 'i386'
+     * </pre>
+     *
+     * <code>string machine = 3;</code>
+     */
+    public Builder setMachine(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      machine_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * e.g. 'i386'
+     * </pre>
+     *
+     * <code>string machine = 3;</code>
+     */
+    public Builder clearMachine() {
+      
+      machine_ = getDefaultInstance().getMachine();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * e.g. 'i386'
+     * </pre>
+     *
+     * <code>string machine = 3;</code>
+     */
+    public Builder setMachineBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      machine_ = value;
+      onChanged();
+      return this;
+    }
+
+    private java.lang.Object release_ = "";
+    /**
+     * <pre>
+     * e.g. '3.13.0-76-generic'
+     * </pre>
+     *
+     * <code>string release = 4;</code>
+     */
+    public java.lang.String getRelease() {
+      java.lang.Object ref = release_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        release_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * e.g. '3.13.0-76-generic'
+     * </pre>
+     *
+     * <code>string release = 4;</code>
+     */
+    public com.google.protobuf.ByteString
+        getReleaseBytes() {
+      java.lang.Object ref = release_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        release_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * e.g. '3.13.0-76-generic'
+     * </pre>
+     *
+     * <code>string release = 4;</code>
+     */
+    public Builder setRelease(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      release_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * e.g. '3.13.0-76-generic'
+     * </pre>
+     *
+     * <code>string release = 4;</code>
+     */
+    public Builder clearRelease() {
+      
+      release_ = getDefaultInstance().getRelease();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * e.g. '3.13.0-76-generic'
+     * </pre>
+     *
+     * <code>string release = 4;</code>
+     */
+    public Builder setReleaseBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      release_ = value;
+      onChanged();
+      return this;
+    }
+
+    private java.lang.Object system_ = "";
+    /**
+     * <pre>
+     * e.g. 'Linux'
+     * </pre>
+     *
+     * <code>string system = 5;</code>
+     */
+    public java.lang.String getSystem() {
+      java.lang.Object ref = system_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        system_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * e.g. 'Linux'
+     * </pre>
+     *
+     * <code>string system = 5;</code>
+     */
+    public com.google.protobuf.ByteString
+        getSystemBytes() {
+      java.lang.Object ref = system_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        system_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * e.g. 'Linux'
+     * </pre>
+     *
+     * <code>string system = 5;</code>
+     */
+    public Builder setSystem(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      system_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * e.g. 'Linux'
+     * </pre>
+     *
+     * <code>string system = 5;</code>
+     */
+    public Builder clearSystem() {
+      
+      system_ = getDefaultInstance().getSystem();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * e.g. 'Linux'
+     * </pre>
+     *
+     * <code>string system = 5;</code>
+     */
+    public Builder setSystemBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      system_ = value;
+      onChanged();
+      return this;
+    }
+
+    private java.lang.Object version_ = "";
+    /**
+     * <pre>
+     * e.g. '#120-Ubuntu SMP Mon Jan 18 15:59:10 UTC 2016'
+     * </pre>
+     *
+     * <code>string version = 6;</code>
+     */
+    public java.lang.String getVersion() {
+      java.lang.Object ref = version_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        version_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * e.g. '#120-Ubuntu SMP Mon Jan 18 15:59:10 UTC 2016'
+     * </pre>
+     *
+     * <code>string version = 6;</code>
+     */
+    public com.google.protobuf.ByteString
+        getVersionBytes() {
+      java.lang.Object ref = version_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        version_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * e.g. '#120-Ubuntu SMP Mon Jan 18 15:59:10 UTC 2016'
+     * </pre>
+     *
+     * <code>string version = 6;</code>
+     */
+    public Builder setVersion(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      version_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * e.g. '#120-Ubuntu SMP Mon Jan 18 15:59:10 UTC 2016'
+     * </pre>
+     *
+     * <code>string version = 6;</code>
+     */
+    public Builder clearVersion() {
+      
+      version_ = getDefaultInstance().getVersion();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * e.g. '#120-Ubuntu SMP Mon Jan 18 15:59:10 UTC 2016'
+     * </pre>
+     *
+     * <code>string version = 6;</code>
+     */
+    public Builder setVersionBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      version_ = value;
+      onChanged();
+      return this;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.PlatformInfo)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.PlatformInfo)
+  private static final org.tensorflow.proto.util.testlog.PlatformInfo DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.util.testlog.PlatformInfo();
+  }
+
+  public static org.tensorflow.proto.util.testlog.PlatformInfo getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<PlatformInfo>
+      PARSER = new com.google.protobuf.AbstractParser<PlatformInfo>() {
+    @java.lang.Override
+    public PlatformInfo parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new PlatformInfo(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<PlatformInfo> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<PlatformInfo> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.util.testlog.PlatformInfo getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/PlatformInfoOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/PlatformInfoOrBuilder.java
new file mode 100644
index 00000000000..6a5089fa309
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/PlatformInfoOrBuilder.java
@@ -0,0 +1,117 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/util/test_log.proto
+
+package org.tensorflow.proto.util.testlog;
+
+public interface PlatformInfoOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.PlatformInfo)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <pre>
+   * e.g. '64bit'
+   * </pre>
+   *
+   * <code>string bits = 1;</code>
+   */
+  java.lang.String getBits();
+  /**
+   * <pre>
+   * e.g. '64bit'
+   * </pre>
+   *
+   * <code>string bits = 1;</code>
+   */
+  com.google.protobuf.ByteString
+      getBitsBytes();
+
+  /**
+   * <pre>
+   * e.g. 'ELF'
+   * </pre>
+   *
+   * <code>string linkage = 2;</code>
+   */
+  java.lang.String getLinkage();
+  /**
+   * <pre>
+   * e.g. 'ELF'
+   * </pre>
+   *
+   * <code>string linkage = 2;</code>
+   */
+  com.google.protobuf.ByteString
+      getLinkageBytes();
+
+  /**
+   * <pre>
+   * e.g. 'i386'
+   * </pre>
+   *
+   * <code>string machine = 3;</code>
+   */
+  java.lang.String getMachine();
+  /**
+   * <pre>
+   * e.g. 'i386'
+   * </pre>
+   *
+   * <code>string machine = 3;</code>
+   */
+  com.google.protobuf.ByteString
+      getMachineBytes();
+
+  /**
+   * <pre>
+   * e.g. '3.13.0-76-generic'
+   * </pre>
+   *
+   * <code>string release = 4;</code>
+   */
+  java.lang.String getRelease();
+  /**
+   * <pre>
+   * e.g. '3.13.0-76-generic'
+   * </pre>
+   *
+   * <code>string release = 4;</code>
+   */
+  com.google.protobuf.ByteString
+      getReleaseBytes();
+
+  /**
+   * <pre>
+   * e.g. 'Linux'
+   * </pre>
+   *
+   * <code>string system = 5;</code>
+   */
+  java.lang.String getSystem();
+  /**
+   * <pre>
+   * e.g. 'Linux'
+   * </pre>
+   *
+   * <code>string system = 5;</code>
+   */
+  com.google.protobuf.ByteString
+      getSystemBytes();
+
+  /**
+   * <pre>
+   * e.g. '#120-Ubuntu SMP Mon Jan 18 15:59:10 UTC 2016'
+   * </pre>
+   *
+   * <code>string version = 6;</code>
+   */
+  java.lang.String getVersion();
+  /**
+   * <pre>
+   * e.g. '#120-Ubuntu SMP Mon Jan 18 15:59:10 UTC 2016'
+   * </pre>
+   *
+   * <code>string version = 6;</code>
+   */
+  com.google.protobuf.ByteString
+      getVersionBytes();
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/RunConfiguration.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/RunConfiguration.java
new file mode 100644
index 00000000000..1ec51529d32
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/RunConfiguration.java
@@ -0,0 +1,917 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/util/test_log.proto
+
+package org.tensorflow.proto.util.testlog;
+
+/**
+ * <pre>
+ * Run-specific items such as arguments to the test / benchmark.
+ * </pre>
+ *
+ * Protobuf type {@code tensorflow.RunConfiguration}
+ */
+public  final class RunConfiguration extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.RunConfiguration)
+    RunConfigurationOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use RunConfiguration.newBuilder() to construct.
+  private RunConfiguration(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private RunConfiguration() {
+    argument_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new RunConfiguration();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private RunConfiguration(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    int mutable_bitField0_ = 0;
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 10: {
+            java.lang.String s = input.readStringRequireUtf8();
+            if (!((mutable_bitField0_ & 0x00000001) != 0)) {
+              argument_ = new com.google.protobuf.LazyStringArrayList();
+              mutable_bitField0_ |= 0x00000001;
+            }
+            argument_.add(s);
+            break;
+          }
+          case 18: {
+            if (!((mutable_bitField0_ & 0x00000002) != 0)) {
+              envVars_ = com.google.protobuf.MapField.newMapField(
+                  EnvVarsDefaultEntryHolder.defaultEntry);
+              mutable_bitField0_ |= 0x00000002;
+            }
+            com.google.protobuf.MapEntry<java.lang.String, java.lang.String>
+            envVars__ = input.readMessage(
+                EnvVarsDefaultEntryHolder.defaultEntry.getParserForType(), extensionRegistry);
+            envVars_.getMutableMap().put(
+                envVars__.getKey(), envVars__.getValue());
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      if (((mutable_bitField0_ & 0x00000001) != 0)) {
+        argument_ = argument_.getUnmodifiableView();
+      }
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_RunConfiguration_descriptor;
+  }
+
+  @SuppressWarnings({"rawtypes"})
+  @java.lang.Override
+  protected com.google.protobuf.MapField internalGetMapField(
+      int number) {
+    switch (number) {
+      case 2:
+        return internalGetEnvVars();
+      default:
+        throw new RuntimeException(
+            "Invalid map field number: " + number);
+    }
+  }
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_RunConfiguration_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.util.testlog.RunConfiguration.class, org.tensorflow.proto.util.testlog.RunConfiguration.Builder.class);
+  }
+
+  public static final int ARGUMENT_FIELD_NUMBER = 1;
+  private com.google.protobuf.LazyStringList argument_;
+  /**
+   * <code>repeated string argument = 1;</code>
+   */
+  public com.google.protobuf.ProtocolStringList
+      getArgumentList() {
+    return argument_;
+  }
+  /**
+   * <code>repeated string argument = 1;</code>
+   */
+  public int getArgumentCount() {
+    return argument_.size();
+  }
+  /**
+   * <code>repeated string argument = 1;</code>
+   */
+  public java.lang.String getArgument(int index) {
+    return argument_.get(index);
+  }
+  /**
+   * <code>repeated string argument = 1;</code>
+   */
+  public com.google.protobuf.ByteString
+      getArgumentBytes(int index) {
+    return argument_.getByteString(index);
+  }
+
+  public static final int ENV_VARS_FIELD_NUMBER = 2;
+  private static final class EnvVarsDefaultEntryHolder {
+    static final com.google.protobuf.MapEntry<
+        java.lang.String, java.lang.String> defaultEntry =
+            com.google.protobuf.MapEntry
+            .<java.lang.String, java.lang.String>newDefaultInstance(
+                org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_RunConfiguration_EnvVarsEntry_descriptor, 
+                com.google.protobuf.WireFormat.FieldType.STRING,
+                "",
+                com.google.protobuf.WireFormat.FieldType.STRING,
+                "");
+  }
+  private com.google.protobuf.MapField<
+      java.lang.String, java.lang.String> envVars_;
+  private com.google.protobuf.MapField<java.lang.String, java.lang.String>
+  internalGetEnvVars() {
+    if (envVars_ == null) {
+      return com.google.protobuf.MapField.emptyMapField(
+          EnvVarsDefaultEntryHolder.defaultEntry);
+    }
+    return envVars_;
+  }
+
+  public int getEnvVarsCount() {
+    return internalGetEnvVars().getMap().size();
+  }
+  /**
+   * <pre>
+   * Environment variables used to run the test/benchmark.
+   * </pre>
+   *
+   * <code>map&lt;string, string&gt; env_vars = 2;</code>
+   */
+
+  public boolean containsEnvVars(
+      java.lang.String key) {
+    if (key == null) { throw new java.lang.NullPointerException(); }
+    return internalGetEnvVars().getMap().containsKey(key);
+  }
+  /**
+   * Use {@link #getEnvVarsMap()} instead.
+   */
+  @java.lang.Deprecated
+  public java.util.Map<java.lang.String, java.lang.String> getEnvVars() {
+    return getEnvVarsMap();
+  }
+  /**
+   * <pre>
+   * Environment variables used to run the test/benchmark.
+   * </pre>
+   *
+   * <code>map&lt;string, string&gt; env_vars = 2;</code>
+   */
+
+  public java.util.Map<java.lang.String, java.lang.String> getEnvVarsMap() {
+    return internalGetEnvVars().getMap();
+  }
+  /**
+   * <pre>
+   * Environment variables used to run the test/benchmark.
+   * </pre>
+   *
+   * <code>map&lt;string, string&gt; env_vars = 2;</code>
+   */
+
+  public java.lang.String getEnvVarsOrDefault(
+      java.lang.String key,
+      java.lang.String defaultValue) {
+    if (key == null) { throw new java.lang.NullPointerException(); }
+    java.util.Map<java.lang.String, java.lang.String> map =
+        internalGetEnvVars().getMap();
+    return map.containsKey(key) ? map.get(key) : defaultValue;
+  }
+  /**
+   * <pre>
+   * Environment variables used to run the test/benchmark.
+   * </pre>
+   *
+   * <code>map&lt;string, string&gt; env_vars = 2;</code>
+   */
+
+  public java.lang.String getEnvVarsOrThrow(
+      java.lang.String key) {
+    if (key == null) { throw new java.lang.NullPointerException(); }
+    java.util.Map<java.lang.String, java.lang.String> map =
+        internalGetEnvVars().getMap();
+    if (!map.containsKey(key)) {
+      throw new java.lang.IllegalArgumentException();
+    }
+    return map.get(key);
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    for (int i = 0; i < argument_.size(); i++) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 1, argument_.getRaw(i));
+    }
+    com.google.protobuf.GeneratedMessageV3
+      .serializeStringMapTo(
+        output,
+        internalGetEnvVars(),
+        EnvVarsDefaultEntryHolder.defaultEntry,
+        2);
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    {
+      int dataSize = 0;
+      for (int i = 0; i < argument_.size(); i++) {
+        dataSize += computeStringSizeNoTag(argument_.getRaw(i));
+      }
+      size += dataSize;
+      size += 1 * getArgumentList().size();
+    }
+    for (java.util.Map.Entry<java.lang.String, java.lang.String> entry
+         : internalGetEnvVars().getMap().entrySet()) {
+      com.google.protobuf.MapEntry<java.lang.String, java.lang.String>
+      envVars__ = EnvVarsDefaultEntryHolder.defaultEntry.newBuilderForType()
+          .setKey(entry.getKey())
+          .setValue(entry.getValue())
+          .build();
+      size += com.google.protobuf.CodedOutputStream
+          .computeMessageSize(2, envVars__);
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.util.testlog.RunConfiguration)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.util.testlog.RunConfiguration other = (org.tensorflow.proto.util.testlog.RunConfiguration) obj;
+
+    if (!getArgumentList()
+        .equals(other.getArgumentList())) return false;
+    if (!internalGetEnvVars().equals(
+        other.internalGetEnvVars())) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    if (getArgumentCount() > 0) {
+      hash = (37 * hash) + ARGUMENT_FIELD_NUMBER;
+      hash = (53 * hash) + getArgumentList().hashCode();
+    }
+    if (!internalGetEnvVars().getMap().isEmpty()) {
+      hash = (37 * hash) + ENV_VARS_FIELD_NUMBER;
+      hash = (53 * hash) + internalGetEnvVars().hashCode();
+    }
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.util.testlog.RunConfiguration parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.testlog.RunConfiguration parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.testlog.RunConfiguration parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.testlog.RunConfiguration parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.testlog.RunConfiguration parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.testlog.RunConfiguration parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.testlog.RunConfiguration parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.testlog.RunConfiguration parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.testlog.RunConfiguration parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.testlog.RunConfiguration parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.testlog.RunConfiguration parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.testlog.RunConfiguration parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.util.testlog.RunConfiguration prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * <pre>
+   * Run-specific items such as arguments to the test / benchmark.
+   * </pre>
+   *
+   * Protobuf type {@code tensorflow.RunConfiguration}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.RunConfiguration)
+      org.tensorflow.proto.util.testlog.RunConfigurationOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_RunConfiguration_descriptor;
+    }
+
+    @SuppressWarnings({"rawtypes"})
+    protected com.google.protobuf.MapField internalGetMapField(
+        int number) {
+      switch (number) {
+        case 2:
+          return internalGetEnvVars();
+        default:
+          throw new RuntimeException(
+              "Invalid map field number: " + number);
+      }
+    }
+    @SuppressWarnings({"rawtypes"})
+    protected com.google.protobuf.MapField internalGetMutableMapField(
+        int number) {
+      switch (number) {
+        case 2:
+          return internalGetMutableEnvVars();
+        default:
+          throw new RuntimeException(
+              "Invalid map field number: " + number);
+      }
+    }
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_RunConfiguration_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.util.testlog.RunConfiguration.class, org.tensorflow.proto.util.testlog.RunConfiguration.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.util.testlog.RunConfiguration.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      argument_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+      bitField0_ = (bitField0_ & ~0x00000001);
+      internalGetMutableEnvVars().clear();
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_RunConfiguration_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.testlog.RunConfiguration getDefaultInstanceForType() {
+      return org.tensorflow.proto.util.testlog.RunConfiguration.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.testlog.RunConfiguration build() {
+      org.tensorflow.proto.util.testlog.RunConfiguration result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.testlog.RunConfiguration buildPartial() {
+      org.tensorflow.proto.util.testlog.RunConfiguration result = new org.tensorflow.proto.util.testlog.RunConfiguration(this);
+      int from_bitField0_ = bitField0_;
+      if (((bitField0_ & 0x00000001) != 0)) {
+        argument_ = argument_.getUnmodifiableView();
+        bitField0_ = (bitField0_ & ~0x00000001);
+      }
+      result.argument_ = argument_;
+      result.envVars_ = internalGetEnvVars();
+      result.envVars_.makeImmutable();
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.util.testlog.RunConfiguration) {
+        return mergeFrom((org.tensorflow.proto.util.testlog.RunConfiguration)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.util.testlog.RunConfiguration other) {
+      if (other == org.tensorflow.proto.util.testlog.RunConfiguration.getDefaultInstance()) return this;
+      if (!other.argument_.isEmpty()) {
+        if (argument_.isEmpty()) {
+          argument_ = other.argument_;
+          bitField0_ = (bitField0_ & ~0x00000001);
+        } else {
+          ensureArgumentIsMutable();
+          argument_.addAll(other.argument_);
+        }
+        onChanged();
+      }
+      internalGetMutableEnvVars().mergeFrom(
+          other.internalGetEnvVars());
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.util.testlog.RunConfiguration parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.util.testlog.RunConfiguration) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+    private int bitField0_;
+
+    private com.google.protobuf.LazyStringList argument_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+    private void ensureArgumentIsMutable() {
+      if (!((bitField0_ & 0x00000001) != 0)) {
+        argument_ = new com.google.protobuf.LazyStringArrayList(argument_);
+        bitField0_ |= 0x00000001;
+       }
+    }
+    /**
+     * <code>repeated string argument = 1;</code>
+     */
+    public com.google.protobuf.ProtocolStringList
+        getArgumentList() {
+      return argument_.getUnmodifiableView();
+    }
+    /**
+     * <code>repeated string argument = 1;</code>
+     */
+    public int getArgumentCount() {
+      return argument_.size();
+    }
+    /**
+     * <code>repeated string argument = 1;</code>
+     */
+    public java.lang.String getArgument(int index) {
+      return argument_.get(index);
+    }
+    /**
+     * <code>repeated string argument = 1;</code>
+     */
+    public com.google.protobuf.ByteString
+        getArgumentBytes(int index) {
+      return argument_.getByteString(index);
+    }
+    /**
+     * <code>repeated string argument = 1;</code>
+     */
+    public Builder setArgument(
+        int index, java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  ensureArgumentIsMutable();
+      argument_.set(index, value);
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>repeated string argument = 1;</code>
+     */
+    public Builder addArgument(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  ensureArgumentIsMutable();
+      argument_.add(value);
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>repeated string argument = 1;</code>
+     */
+    public Builder addAllArgument(
+        java.lang.Iterable<java.lang.String> values) {
+      ensureArgumentIsMutable();
+      com.google.protobuf.AbstractMessageLite.Builder.addAll(
+          values, argument_);
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>repeated string argument = 1;</code>
+     */
+    public Builder clearArgument() {
+      argument_ = com.google.protobuf.LazyStringArrayList.EMPTY;
+      bitField0_ = (bitField0_ & ~0x00000001);
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>repeated string argument = 1;</code>
+     */
+    public Builder addArgumentBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      ensureArgumentIsMutable();
+      argument_.add(value);
+      onChanged();
+      return this;
+    }
+
+    private com.google.protobuf.MapField<
+        java.lang.String, java.lang.String> envVars_;
+    private com.google.protobuf.MapField<java.lang.String, java.lang.String>
+    internalGetEnvVars() {
+      if (envVars_ == null) {
+        return com.google.protobuf.MapField.emptyMapField(
+            EnvVarsDefaultEntryHolder.defaultEntry);
+      }
+      return envVars_;
+    }
+    private com.google.protobuf.MapField<java.lang.String, java.lang.String>
+    internalGetMutableEnvVars() {
+      onChanged();;
+      if (envVars_ == null) {
+        envVars_ = com.google.protobuf.MapField.newMapField(
+            EnvVarsDefaultEntryHolder.defaultEntry);
+      }
+      if (!envVars_.isMutable()) {
+        envVars_ = envVars_.copy();
+      }
+      return envVars_;
+    }
+
+    public int getEnvVarsCount() {
+      return internalGetEnvVars().getMap().size();
+    }
+    /**
+     * <pre>
+     * Environment variables used to run the test/benchmark.
+     * </pre>
+     *
+     * <code>map&lt;string, string&gt; env_vars = 2;</code>
+     */
+
+    public boolean containsEnvVars(
+        java.lang.String key) {
+      if (key == null) { throw new java.lang.NullPointerException(); }
+      return internalGetEnvVars().getMap().containsKey(key);
+    }
+    /**
+     * Use {@link #getEnvVarsMap()} instead.
+     */
+    @java.lang.Deprecated
+    public java.util.Map<java.lang.String, java.lang.String> getEnvVars() {
+      return getEnvVarsMap();
+    }
+    /**
+     * <pre>
+     * Environment variables used to run the test/benchmark.
+     * </pre>
+     *
+     * <code>map&lt;string, string&gt; env_vars = 2;</code>
+     */
+
+    public java.util.Map<java.lang.String, java.lang.String> getEnvVarsMap() {
+      return internalGetEnvVars().getMap();
+    }
+    /**
+     * <pre>
+     * Environment variables used to run the test/benchmark.
+     * </pre>
+     *
+     * <code>map&lt;string, string&gt; env_vars = 2;</code>
+     */
+
+    public java.lang.String getEnvVarsOrDefault(
+        java.lang.String key,
+        java.lang.String defaultValue) {
+      if (key == null) { throw new java.lang.NullPointerException(); }
+      java.util.Map<java.lang.String, java.lang.String> map =
+          internalGetEnvVars().getMap();
+      return map.containsKey(key) ? map.get(key) : defaultValue;
+    }
+    /**
+     * <pre>
+     * Environment variables used to run the test/benchmark.
+     * </pre>
+     *
+     * <code>map&lt;string, string&gt; env_vars = 2;</code>
+     */
+
+    public java.lang.String getEnvVarsOrThrow(
+        java.lang.String key) {
+      if (key == null) { throw new java.lang.NullPointerException(); }
+      java.util.Map<java.lang.String, java.lang.String> map =
+          internalGetEnvVars().getMap();
+      if (!map.containsKey(key)) {
+        throw new java.lang.IllegalArgumentException();
+      }
+      return map.get(key);
+    }
+
+    public Builder clearEnvVars() {
+      internalGetMutableEnvVars().getMutableMap()
+          .clear();
+      return this;
+    }
+    /**
+     * <pre>
+     * Environment variables used to run the test/benchmark.
+     * </pre>
+     *
+     * <code>map&lt;string, string&gt; env_vars = 2;</code>
+     */
+
+    public Builder removeEnvVars(
+        java.lang.String key) {
+      if (key == null) { throw new java.lang.NullPointerException(); }
+      internalGetMutableEnvVars().getMutableMap()
+          .remove(key);
+      return this;
+    }
+    /**
+     * Use alternate mutation accessors instead.
+     */
+    @java.lang.Deprecated
+    public java.util.Map<java.lang.String, java.lang.String>
+    getMutableEnvVars() {
+      return internalGetMutableEnvVars().getMutableMap();
+    }
+    /**
+     * <pre>
+     * Environment variables used to run the test/benchmark.
+     * </pre>
+     *
+     * <code>map&lt;string, string&gt; env_vars = 2;</code>
+     */
+    public Builder putEnvVars(
+        java.lang.String key,
+        java.lang.String value) {
+      if (key == null) { throw new java.lang.NullPointerException(); }
+      if (value == null) { throw new java.lang.NullPointerException(); }
+      internalGetMutableEnvVars().getMutableMap()
+          .put(key, value);
+      return this;
+    }
+    /**
+     * <pre>
+     * Environment variables used to run the test/benchmark.
+     * </pre>
+     *
+     * <code>map&lt;string, string&gt; env_vars = 2;</code>
+     */
+
+    public Builder putAllEnvVars(
+        java.util.Map<java.lang.String, java.lang.String> values) {
+      internalGetMutableEnvVars().getMutableMap()
+          .putAll(values);
+      return this;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.RunConfiguration)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.RunConfiguration)
+  private static final org.tensorflow.proto.util.testlog.RunConfiguration DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.util.testlog.RunConfiguration();
+  }
+
+  public static org.tensorflow.proto.util.testlog.RunConfiguration getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<RunConfiguration>
+      PARSER = new com.google.protobuf.AbstractParser<RunConfiguration>() {
+    @java.lang.Override
+    public RunConfiguration parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new RunConfiguration(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<RunConfiguration> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<RunConfiguration> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.util.testlog.RunConfiguration getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/RunConfigurationOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/RunConfigurationOrBuilder.java
new file mode 100644
index 00000000000..9150f64ee59
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/RunConfigurationOrBuilder.java
@@ -0,0 +1,82 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/util/test_log.proto
+
+package org.tensorflow.proto.util.testlog;
+
+public interface RunConfigurationOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.RunConfiguration)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <code>repeated string argument = 1;</code>
+   */
+  java.util.List<java.lang.String>
+      getArgumentList();
+  /**
+   * <code>repeated string argument = 1;</code>
+   */
+  int getArgumentCount();
+  /**
+   * <code>repeated string argument = 1;</code>
+   */
+  java.lang.String getArgument(int index);
+  /**
+   * <code>repeated string argument = 1;</code>
+   */
+  com.google.protobuf.ByteString
+      getArgumentBytes(int index);
+
+  /**
+   * <pre>
+   * Environment variables used to run the test/benchmark.
+   * </pre>
+   *
+   * <code>map&lt;string, string&gt; env_vars = 2;</code>
+   */
+  int getEnvVarsCount();
+  /**
+   * <pre>
+   * Environment variables used to run the test/benchmark.
+   * </pre>
+   *
+   * <code>map&lt;string, string&gt; env_vars = 2;</code>
+   */
+  boolean containsEnvVars(
+      java.lang.String key);
+  /**
+   * Use {@link #getEnvVarsMap()} instead.
+   */
+  @java.lang.Deprecated
+  java.util.Map<java.lang.String, java.lang.String>
+  getEnvVars();
+  /**
+   * <pre>
+   * Environment variables used to run the test/benchmark.
+   * </pre>
+   *
+   * <code>map&lt;string, string&gt; env_vars = 2;</code>
+   */
+  java.util.Map<java.lang.String, java.lang.String>
+  getEnvVarsMap();
+  /**
+   * <pre>
+   * Environment variables used to run the test/benchmark.
+   * </pre>
+   *
+   * <code>map&lt;string, string&gt; env_vars = 2;</code>
+   */
+
+  java.lang.String getEnvVarsOrDefault(
+      java.lang.String key,
+      java.lang.String defaultValue);
+  /**
+   * <pre>
+   * Environment variables used to run the test/benchmark.
+   * </pre>
+   *
+   * <code>map&lt;string, string&gt; env_vars = 2;</code>
+   */
+
+  java.lang.String getEnvVarsOrThrow(
+      java.lang.String key);
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/TestLogProtos.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/TestLogProtos.java
new file mode 100644
index 00000000000..bb4033abe44
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/TestLogProtos.java
@@ -0,0 +1,288 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/util/test_log.proto
+
+package org.tensorflow.proto.util.testlog;
+
+public final class TestLogProtos {
+  private TestLogProtos() {}
+  public static void registerAllExtensions(
+      com.google.protobuf.ExtensionRegistryLite registry) {
+  }
+
+  public static void registerAllExtensions(
+      com.google.protobuf.ExtensionRegistry registry) {
+    registerAllExtensions(
+        (com.google.protobuf.ExtensionRegistryLite) registry);
+  }
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_EntryValue_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_EntryValue_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_MetricEntry_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_MetricEntry_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_BenchmarkEntry_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_BenchmarkEntry_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_BenchmarkEntry_ExtrasEntry_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_BenchmarkEntry_ExtrasEntry_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_BenchmarkEntries_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_BenchmarkEntries_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_BuildConfiguration_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_BuildConfiguration_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_CommitId_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_CommitId_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_CPUInfo_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_CPUInfo_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_CPUInfo_CacheSizeEntry_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_CPUInfo_CacheSizeEntry_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_MemoryInfo_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_MemoryInfo_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_GPUInfo_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_GPUInfo_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_PlatformInfo_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_PlatformInfo_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_AvailableDeviceInfo_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_AvailableDeviceInfo_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_MachineConfiguration_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_MachineConfiguration_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_RunConfiguration_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_RunConfiguration_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_RunConfiguration_EnvVarsEntry_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_RunConfiguration_EnvVarsEntry_fieldAccessorTable;
+  static final com.google.protobuf.Descriptors.Descriptor
+    internal_static_tensorflow_TestResults_descriptor;
+  static final 
+    com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internal_static_tensorflow_TestResults_fieldAccessorTable;
+
+  public static com.google.protobuf.Descriptors.FileDescriptor
+      getDescriptor() {
+    return descriptor;
+  }
+  private static  com.google.protobuf.Descriptors.FileDescriptor
+      descriptor;
+  static {
+    java.lang.String[] descriptorData = {
+      "\n#tensorflow/core/util/test_log.proto\022\nt" +
+      "ensorflow\032\031google/protobuf/any.proto\032\036go" +
+      "ogle/protobuf/wrappers.proto\"D\n\nEntryVal" +
+      "ue\022\026\n\014double_value\030\001 \001(\001H\000\022\026\n\014string_val" +
+      "ue\030\002 \001(\tH\000B\006\n\004kind\"\214\001\n\013MetricEntry\022\014\n\004na" +
+      "me\030\001 \001(\t\022\r\n\005value\030\002 \001(\001\022/\n\tmin_value\030\003 \001" +
+      "(\0132\034.google.protobuf.DoubleValue\022/\n\tmax_" +
+      "value\030\004 \001(\0132\034.google.protobuf.DoubleValu" +
+      "e\"\217\002\n\016BenchmarkEntry\022\014\n\004name\030\001 \001(\t\022\r\n\005it" +
+      "ers\030\002 \001(\003\022\020\n\010cpu_time\030\003 \001(\001\022\021\n\twall_time" +
+      "\030\004 \001(\001\022\022\n\nthroughput\030\005 \001(\001\0226\n\006extras\030\006 \003" +
+      "(\0132&.tensorflow.BenchmarkEntry.ExtrasEnt" +
+      "ry\022(\n\007metrics\030\007 \003(\0132\027.tensorflow.MetricE" +
+      "ntry\032E\n\013ExtrasEntry\022\013\n\003key\030\001 \001(\t\022%\n\005valu" +
+      "e\030\002 \001(\0132\026.tensorflow.EntryValue:\0028\001\"=\n\020B" +
+      "enchmarkEntries\022)\n\005entry\030\001 \003(\0132\032.tensorf" +
+      "low.BenchmarkEntry\"B\n\022BuildConfiguration" +
+      "\022\014\n\004mode\030\001 \001(\t\022\020\n\010cc_flags\030\002 \003(\t\022\014\n\004opts" +
+      "\030\003 \003(\t\"f\n\010CommitId\022\024\n\nchangelist\030\001 \001(\003H\000" +
+      "\022\016\n\004hash\030\002 \001(\tH\000\022\020\n\010snapshot\030\003 \001(\t\022\032\n\022pe" +
+      "nding_changelist\030\004 \001(\003B\006\n\004kind\"\336\001\n\007CPUIn" +
+      "fo\022\021\n\tnum_cores\030\001 \001(\003\022\031\n\021num_cores_allow" +
+      "ed\030\002 \001(\003\022\023\n\013mhz_per_cpu\030\003 \001(\001\022\020\n\010cpu_inf" +
+      "o\030\004 \001(\t\022\024\n\014cpu_governor\030\005 \001(\t\0226\n\ncache_s" +
+      "ize\030\006 \003(\0132\".tensorflow.CPUInfo.CacheSize" +
+      "Entry\0320\n\016CacheSizeEntry\022\013\n\003key\030\001 \001(\t\022\r\n\005" +
+      "value\030\002 \001(\003:\0028\001\".\n\nMemoryInfo\022\r\n\005total\030\001" +
+      " \001(\003\022\021\n\tavailable\030\002 \001(\003\"6\n\007GPUInfo\022\r\n\005mo" +
+      "del\030\001 \001(\t\022\014\n\004uuid\030\002 \001(\t\022\016\n\006bus_id\030\003 \001(\t\"" +
+      "p\n\014PlatformInfo\022\014\n\004bits\030\001 \001(\t\022\017\n\007linkage" +
+      "\030\002 \001(\t\022\017\n\007machine\030\003 \001(\t\022\017\n\007release\030\004 \001(\t" +
+      "\022\016\n\006system\030\005 \001(\t\022\017\n\007version\030\006 \001(\t\"e\n\023Ava" +
+      "ilableDeviceInfo\022\014\n\004name\030\001 \001(\t\022\014\n\004type\030\002" +
+      " \001(\t\022\024\n\014memory_limit\030\003 \001(\003\022\034\n\024physical_d" +
+      "escription\030\004 \001(\t\"\263\002\n\024MachineConfiguratio" +
+      "n\022\020\n\010hostname\030\001 \001(\t\022\031\n\021serial_identifier" +
+      "\030\007 \001(\t\022/\n\rplatform_info\030\002 \001(\0132\030.tensorfl" +
+      "ow.PlatformInfo\022%\n\010cpu_info\030\003 \001(\0132\023.tens" +
+      "orflow.CPUInfo\022)\n\013device_info\030\004 \003(\0132\024.go" +
+      "ogle.protobuf.Any\022>\n\025available_device_in" +
+      "fo\030\005 \003(\0132\037.tensorflow.AvailableDeviceInf" +
+      "o\022+\n\013memory_info\030\006 \001(\0132\026.tensorflow.Memo" +
+      "ryInfo\"\221\001\n\020RunConfiguration\022\020\n\010argument\030" +
+      "\001 \003(\t\022;\n\010env_vars\030\002 \003(\0132).tensorflow.Run" +
+      "Configuration.EnvVarsEntry\032.\n\014EnvVarsEnt" +
+      "ry\022\013\n\003key\030\001 \001(\t\022\r\n\005value\030\002 \001(\t:\0028\001\"\320\004\n\013T" +
+      "estResults\022\016\n\006target\030\001 \001(\t\022-\n\007entries\030\002 " +
+      "\001(\0132\034.tensorflow.BenchmarkEntries\022;\n\023bui" +
+      "ld_configuration\030\003 \001(\0132\036.tensorflow.Buil" +
+      "dConfiguration\022\'\n\tcommit_id\030\004 \001(\0132\024.tens" +
+      "orflow.CommitId\022\022\n\nstart_time\030\005 \001(\003\022\020\n\010r" +
+      "un_time\030\006 \001(\001\022?\n\025machine_configuration\030\007" +
+      " \001(\0132 .tensorflow.MachineConfiguration\0227" +
+      "\n\021run_configuration\030\010 \001(\0132\034.tensorflow.R" +
+      "unConfiguration\022\014\n\004name\030\t \001(\t\022=\n\016benchma" +
+      "rk_type\030\n \001(\0162%.tensorflow.TestResults.B" +
+      "enchmarkType\022\020\n\010run_mode\030\013 \001(\t\022\022\n\ntf_ver" +
+      "sion\030\014 \001(\t\"\210\001\n\rBenchmarkType\022\013\n\007UNKNOWN\020" +
+      "\000\022\026\n\022CPP_MICROBENCHMARK\020\001\022\024\n\020PYTHON_BENC" +
+      "HMARK\020\002\022\025\n\021ANDROID_BENCHMARK\020\003\022\022\n\016EDGE_B" +
+      "ENCHMARK\020\004\022\021\n\rIOS_BENCHMARK\020\005B7\n!org.ten" +
+      "sorflow.proto.util.testlogB\rTestLogProto" +
+      "sP\001\370\001\001b\006proto3"
+    };
+    descriptor = com.google.protobuf.Descriptors.FileDescriptor
+      .internalBuildGeneratedFileFrom(descriptorData,
+        new com.google.protobuf.Descriptors.FileDescriptor[] {
+          com.google.protobuf.AnyProto.getDescriptor(),
+          com.google.protobuf.WrappersProto.getDescriptor(),
+        });
+    internal_static_tensorflow_EntryValue_descriptor =
+      getDescriptor().getMessageTypes().get(0);
+    internal_static_tensorflow_EntryValue_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_EntryValue_descriptor,
+        new java.lang.String[] { "DoubleValue", "StringValue", "Kind", });
+    internal_static_tensorflow_MetricEntry_descriptor =
+      getDescriptor().getMessageTypes().get(1);
+    internal_static_tensorflow_MetricEntry_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_MetricEntry_descriptor,
+        new java.lang.String[] { "Name", "Value", "MinValue", "MaxValue", });
+    internal_static_tensorflow_BenchmarkEntry_descriptor =
+      getDescriptor().getMessageTypes().get(2);
+    internal_static_tensorflow_BenchmarkEntry_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_BenchmarkEntry_descriptor,
+        new java.lang.String[] { "Name", "Iters", "CpuTime", "WallTime", "Throughput", "Extras", "Metrics", });
+    internal_static_tensorflow_BenchmarkEntry_ExtrasEntry_descriptor =
+      internal_static_tensorflow_BenchmarkEntry_descriptor.getNestedTypes().get(0);
+    internal_static_tensorflow_BenchmarkEntry_ExtrasEntry_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_BenchmarkEntry_ExtrasEntry_descriptor,
+        new java.lang.String[] { "Key", "Value", });
+    internal_static_tensorflow_BenchmarkEntries_descriptor =
+      getDescriptor().getMessageTypes().get(3);
+    internal_static_tensorflow_BenchmarkEntries_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_BenchmarkEntries_descriptor,
+        new java.lang.String[] { "Entry", });
+    internal_static_tensorflow_BuildConfiguration_descriptor =
+      getDescriptor().getMessageTypes().get(4);
+    internal_static_tensorflow_BuildConfiguration_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_BuildConfiguration_descriptor,
+        new java.lang.String[] { "Mode", "CcFlags", "Opts", });
+    internal_static_tensorflow_CommitId_descriptor =
+      getDescriptor().getMessageTypes().get(5);
+    internal_static_tensorflow_CommitId_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_CommitId_descriptor,
+        new java.lang.String[] { "Changelist", "Hash", "Snapshot", "PendingChangelist", "Kind", });
+    internal_static_tensorflow_CPUInfo_descriptor =
+      getDescriptor().getMessageTypes().get(6);
+    internal_static_tensorflow_CPUInfo_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_CPUInfo_descriptor,
+        new java.lang.String[] { "NumCores", "NumCoresAllowed", "MhzPerCpu", "CpuInfo", "CpuGovernor", "CacheSize", });
+    internal_static_tensorflow_CPUInfo_CacheSizeEntry_descriptor =
+      internal_static_tensorflow_CPUInfo_descriptor.getNestedTypes().get(0);
+    internal_static_tensorflow_CPUInfo_CacheSizeEntry_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_CPUInfo_CacheSizeEntry_descriptor,
+        new java.lang.String[] { "Key", "Value", });
+    internal_static_tensorflow_MemoryInfo_descriptor =
+      getDescriptor().getMessageTypes().get(7);
+    internal_static_tensorflow_MemoryInfo_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_MemoryInfo_descriptor,
+        new java.lang.String[] { "Total", "Available", });
+    internal_static_tensorflow_GPUInfo_descriptor =
+      getDescriptor().getMessageTypes().get(8);
+    internal_static_tensorflow_GPUInfo_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_GPUInfo_descriptor,
+        new java.lang.String[] { "Model", "Uuid", "BusId", });
+    internal_static_tensorflow_PlatformInfo_descriptor =
+      getDescriptor().getMessageTypes().get(9);
+    internal_static_tensorflow_PlatformInfo_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_PlatformInfo_descriptor,
+        new java.lang.String[] { "Bits", "Linkage", "Machine", "Release", "System", "Version", });
+    internal_static_tensorflow_AvailableDeviceInfo_descriptor =
+      getDescriptor().getMessageTypes().get(10);
+    internal_static_tensorflow_AvailableDeviceInfo_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_AvailableDeviceInfo_descriptor,
+        new java.lang.String[] { "Name", "Type", "MemoryLimit", "PhysicalDescription", });
+    internal_static_tensorflow_MachineConfiguration_descriptor =
+      getDescriptor().getMessageTypes().get(11);
+    internal_static_tensorflow_MachineConfiguration_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_MachineConfiguration_descriptor,
+        new java.lang.String[] { "Hostname", "SerialIdentifier", "PlatformInfo", "CpuInfo", "DeviceInfo", "AvailableDeviceInfo", "MemoryInfo", });
+    internal_static_tensorflow_RunConfiguration_descriptor =
+      getDescriptor().getMessageTypes().get(12);
+    internal_static_tensorflow_RunConfiguration_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_RunConfiguration_descriptor,
+        new java.lang.String[] { "Argument", "EnvVars", });
+    internal_static_tensorflow_RunConfiguration_EnvVarsEntry_descriptor =
+      internal_static_tensorflow_RunConfiguration_descriptor.getNestedTypes().get(0);
+    internal_static_tensorflow_RunConfiguration_EnvVarsEntry_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_RunConfiguration_EnvVarsEntry_descriptor,
+        new java.lang.String[] { "Key", "Value", });
+    internal_static_tensorflow_TestResults_descriptor =
+      getDescriptor().getMessageTypes().get(13);
+    internal_static_tensorflow_TestResults_fieldAccessorTable = new
+      com.google.protobuf.GeneratedMessageV3.FieldAccessorTable(
+        internal_static_tensorflow_TestResults_descriptor,
+        new java.lang.String[] { "Target", "Entries", "BuildConfiguration", "CommitId", "StartTime", "RunTime", "MachineConfiguration", "RunConfiguration", "Name", "BenchmarkType", "RunMode", "TfVersion", });
+    com.google.protobuf.AnyProto.getDescriptor();
+    com.google.protobuf.WrappersProto.getDescriptor();
+  }
+
+  // @@protoc_insertion_point(outer_class_scope)
+}
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/TestResults.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/TestResults.java
new file mode 100644
index 00000000000..15ae63f6d31
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/TestResults.java
@@ -0,0 +1,2624 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/util/test_log.proto
+
+package org.tensorflow.proto.util.testlog;
+
+/**
+ * <pre>
+ * The output of one benchmark / test run.  Each run contains a list of
+ * tests or benchmarks, stored as BenchmarkEntry messages.
+ * This message should be emitted by the reporter (which runs the
+ * test / BM in a subprocess and then reads the emitted BenchmarkEntry messages;
+ * usually from a serialized json file, finally collecting them along
+ * with additional information about the test run.
+ * </pre>
+ *
+ * Protobuf type {@code tensorflow.TestResults}
+ */
+public  final class TestResults extends
+    com.google.protobuf.GeneratedMessageV3 implements
+    // @@protoc_insertion_point(message_implements:tensorflow.TestResults)
+    TestResultsOrBuilder {
+private static final long serialVersionUID = 0L;
+  // Use TestResults.newBuilder() to construct.
+  private TestResults(com.google.protobuf.GeneratedMessageV3.Builder<?> builder) {
+    super(builder);
+  }
+  private TestResults() {
+    target_ = "";
+    name_ = "";
+    benchmarkType_ = 0;
+    runMode_ = "";
+    tfVersion_ = "";
+  }
+
+  @java.lang.Override
+  @SuppressWarnings({"unused"})
+  protected java.lang.Object newInstance(
+      UnusedPrivateParameter unused) {
+    return new TestResults();
+  }
+
+  @java.lang.Override
+  public final com.google.protobuf.UnknownFieldSet
+  getUnknownFields() {
+    return this.unknownFields;
+  }
+  private TestResults(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    this();
+    if (extensionRegistry == null) {
+      throw new java.lang.NullPointerException();
+    }
+    com.google.protobuf.UnknownFieldSet.Builder unknownFields =
+        com.google.protobuf.UnknownFieldSet.newBuilder();
+    try {
+      boolean done = false;
+      while (!done) {
+        int tag = input.readTag();
+        switch (tag) {
+          case 0:
+            done = true;
+            break;
+          case 10: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            target_ = s;
+            break;
+          }
+          case 18: {
+            org.tensorflow.proto.util.testlog.BenchmarkEntries.Builder subBuilder = null;
+            if (entries_ != null) {
+              subBuilder = entries_.toBuilder();
+            }
+            entries_ = input.readMessage(org.tensorflow.proto.util.testlog.BenchmarkEntries.parser(), extensionRegistry);
+            if (subBuilder != null) {
+              subBuilder.mergeFrom(entries_);
+              entries_ = subBuilder.buildPartial();
+            }
+
+            break;
+          }
+          case 26: {
+            org.tensorflow.proto.util.testlog.BuildConfiguration.Builder subBuilder = null;
+            if (buildConfiguration_ != null) {
+              subBuilder = buildConfiguration_.toBuilder();
+            }
+            buildConfiguration_ = input.readMessage(org.tensorflow.proto.util.testlog.BuildConfiguration.parser(), extensionRegistry);
+            if (subBuilder != null) {
+              subBuilder.mergeFrom(buildConfiguration_);
+              buildConfiguration_ = subBuilder.buildPartial();
+            }
+
+            break;
+          }
+          case 34: {
+            org.tensorflow.proto.util.testlog.CommitId.Builder subBuilder = null;
+            if (commitId_ != null) {
+              subBuilder = commitId_.toBuilder();
+            }
+            commitId_ = input.readMessage(org.tensorflow.proto.util.testlog.CommitId.parser(), extensionRegistry);
+            if (subBuilder != null) {
+              subBuilder.mergeFrom(commitId_);
+              commitId_ = subBuilder.buildPartial();
+            }
+
+            break;
+          }
+          case 40: {
+
+            startTime_ = input.readInt64();
+            break;
+          }
+          case 49: {
+
+            runTime_ = input.readDouble();
+            break;
+          }
+          case 58: {
+            org.tensorflow.proto.util.testlog.MachineConfiguration.Builder subBuilder = null;
+            if (machineConfiguration_ != null) {
+              subBuilder = machineConfiguration_.toBuilder();
+            }
+            machineConfiguration_ = input.readMessage(org.tensorflow.proto.util.testlog.MachineConfiguration.parser(), extensionRegistry);
+            if (subBuilder != null) {
+              subBuilder.mergeFrom(machineConfiguration_);
+              machineConfiguration_ = subBuilder.buildPartial();
+            }
+
+            break;
+          }
+          case 66: {
+            org.tensorflow.proto.util.testlog.RunConfiguration.Builder subBuilder = null;
+            if (runConfiguration_ != null) {
+              subBuilder = runConfiguration_.toBuilder();
+            }
+            runConfiguration_ = input.readMessage(org.tensorflow.proto.util.testlog.RunConfiguration.parser(), extensionRegistry);
+            if (subBuilder != null) {
+              subBuilder.mergeFrom(runConfiguration_);
+              runConfiguration_ = subBuilder.buildPartial();
+            }
+
+            break;
+          }
+          case 74: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            name_ = s;
+            break;
+          }
+          case 80: {
+            int rawValue = input.readEnum();
+
+            benchmarkType_ = rawValue;
+            break;
+          }
+          case 90: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            runMode_ = s;
+            break;
+          }
+          case 98: {
+            java.lang.String s = input.readStringRequireUtf8();
+
+            tfVersion_ = s;
+            break;
+          }
+          default: {
+            if (!parseUnknownField(
+                input, unknownFields, extensionRegistry, tag)) {
+              done = true;
+            }
+            break;
+          }
+        }
+      }
+    } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+      throw e.setUnfinishedMessage(this);
+    } catch (java.io.IOException e) {
+      throw new com.google.protobuf.InvalidProtocolBufferException(
+          e).setUnfinishedMessage(this);
+    } finally {
+      this.unknownFields = unknownFields.build();
+      makeExtensionsImmutable();
+    }
+  }
+  public static final com.google.protobuf.Descriptors.Descriptor
+      getDescriptor() {
+    return org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_TestResults_descriptor;
+  }
+
+  @java.lang.Override
+  protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+      internalGetFieldAccessorTable() {
+    return org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_TestResults_fieldAccessorTable
+        .ensureFieldAccessorsInitialized(
+            org.tensorflow.proto.util.testlog.TestResults.class, org.tensorflow.proto.util.testlog.TestResults.Builder.class);
+  }
+
+  /**
+   * <pre>
+   * The type of benchmark.
+   * </pre>
+   *
+   * Protobuf enum {@code tensorflow.TestResults.BenchmarkType}
+   */
+  public enum BenchmarkType
+      implements com.google.protobuf.ProtocolMessageEnum {
+    /**
+     * <pre>
+     * Fallback for protos written before Type was introduced.
+     * </pre>
+     *
+     * <code>UNKNOWN = 0;</code>
+     */
+    UNKNOWN(0),
+    /**
+     * <code>CPP_MICROBENCHMARK = 1;</code>
+     */
+    CPP_MICROBENCHMARK(1),
+    /**
+     * <code>PYTHON_BENCHMARK = 2;</code>
+     */
+    PYTHON_BENCHMARK(2),
+    /**
+     * <code>ANDROID_BENCHMARK = 3;</code>
+     */
+    ANDROID_BENCHMARK(3),
+    /**
+     * <code>EDGE_BENCHMARK = 4;</code>
+     */
+    EDGE_BENCHMARK(4),
+    /**
+     * <code>IOS_BENCHMARK = 5;</code>
+     */
+    IOS_BENCHMARK(5),
+    UNRECOGNIZED(-1),
+    ;
+
+    /**
+     * <pre>
+     * Fallback for protos written before Type was introduced.
+     * </pre>
+     *
+     * <code>UNKNOWN = 0;</code>
+     */
+    public static final int UNKNOWN_VALUE = 0;
+    /**
+     * <code>CPP_MICROBENCHMARK = 1;</code>
+     */
+    public static final int CPP_MICROBENCHMARK_VALUE = 1;
+    /**
+     * <code>PYTHON_BENCHMARK = 2;</code>
+     */
+    public static final int PYTHON_BENCHMARK_VALUE = 2;
+    /**
+     * <code>ANDROID_BENCHMARK = 3;</code>
+     */
+    public static final int ANDROID_BENCHMARK_VALUE = 3;
+    /**
+     * <code>EDGE_BENCHMARK = 4;</code>
+     */
+    public static final int EDGE_BENCHMARK_VALUE = 4;
+    /**
+     * <code>IOS_BENCHMARK = 5;</code>
+     */
+    public static final int IOS_BENCHMARK_VALUE = 5;
+
+
+    public final int getNumber() {
+      if (this == UNRECOGNIZED) {
+        throw new java.lang.IllegalArgumentException(
+            "Can't get the number of an unknown enum value.");
+      }
+      return value;
+    }
+
+    /**
+     * @deprecated Use {@link #forNumber(int)} instead.
+     */
+    @java.lang.Deprecated
+    public static BenchmarkType valueOf(int value) {
+      return forNumber(value);
+    }
+
+    public static BenchmarkType forNumber(int value) {
+      switch (value) {
+        case 0: return UNKNOWN;
+        case 1: return CPP_MICROBENCHMARK;
+        case 2: return PYTHON_BENCHMARK;
+        case 3: return ANDROID_BENCHMARK;
+        case 4: return EDGE_BENCHMARK;
+        case 5: return IOS_BENCHMARK;
+        default: return null;
+      }
+    }
+
+    public static com.google.protobuf.Internal.EnumLiteMap<BenchmarkType>
+        internalGetValueMap() {
+      return internalValueMap;
+    }
+    private static final com.google.protobuf.Internal.EnumLiteMap<
+        BenchmarkType> internalValueMap =
+          new com.google.protobuf.Internal.EnumLiteMap<BenchmarkType>() {
+            public BenchmarkType findValueByNumber(int number) {
+              return BenchmarkType.forNumber(number);
+            }
+          };
+
+    public final com.google.protobuf.Descriptors.EnumValueDescriptor
+        getValueDescriptor() {
+      return getDescriptor().getValues().get(ordinal());
+    }
+    public final com.google.protobuf.Descriptors.EnumDescriptor
+        getDescriptorForType() {
+      return getDescriptor();
+    }
+    public static final com.google.protobuf.Descriptors.EnumDescriptor
+        getDescriptor() {
+      return org.tensorflow.proto.util.testlog.TestResults.getDescriptor().getEnumTypes().get(0);
+    }
+
+    private static final BenchmarkType[] VALUES = values();
+
+    public static BenchmarkType valueOf(
+        com.google.protobuf.Descriptors.EnumValueDescriptor desc) {
+      if (desc.getType() != getDescriptor()) {
+        throw new java.lang.IllegalArgumentException(
+          "EnumValueDescriptor is not for this type.");
+      }
+      if (desc.getIndex() == -1) {
+        return UNRECOGNIZED;
+      }
+      return VALUES[desc.getIndex()];
+    }
+
+    private final int value;
+
+    private BenchmarkType(int value) {
+      this.value = value;
+    }
+
+    // @@protoc_insertion_point(enum_scope:tensorflow.TestResults.BenchmarkType)
+  }
+
+  public static final int TARGET_FIELD_NUMBER = 1;
+  private volatile java.lang.Object target_;
+  /**
+   * <pre>
+   * The target of the run, e.g.:
+   *  //tensorflow/core:kernels_adjust_contrast_op_benchmark_test
+   * </pre>
+   *
+   * <code>string target = 1;</code>
+   */
+  public java.lang.String getTarget() {
+    java.lang.Object ref = target_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      target_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * The target of the run, e.g.:
+   *  //tensorflow/core:kernels_adjust_contrast_op_benchmark_test
+   * </pre>
+   *
+   * <code>string target = 1;</code>
+   */
+  public com.google.protobuf.ByteString
+      getTargetBytes() {
+    java.lang.Object ref = target_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      target_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int ENTRIES_FIELD_NUMBER = 2;
+  private org.tensorflow.proto.util.testlog.BenchmarkEntries entries_;
+  /**
+   * <pre>
+   * The list of tests or benchmarks in this run.
+   * </pre>
+   *
+   * <code>.tensorflow.BenchmarkEntries entries = 2;</code>
+   */
+  public boolean hasEntries() {
+    return entries_ != null;
+  }
+  /**
+   * <pre>
+   * The list of tests or benchmarks in this run.
+   * </pre>
+   *
+   * <code>.tensorflow.BenchmarkEntries entries = 2;</code>
+   */
+  public org.tensorflow.proto.util.testlog.BenchmarkEntries getEntries() {
+    return entries_ == null ? org.tensorflow.proto.util.testlog.BenchmarkEntries.getDefaultInstance() : entries_;
+  }
+  /**
+   * <pre>
+   * The list of tests or benchmarks in this run.
+   * </pre>
+   *
+   * <code>.tensorflow.BenchmarkEntries entries = 2;</code>
+   */
+  public org.tensorflow.proto.util.testlog.BenchmarkEntriesOrBuilder getEntriesOrBuilder() {
+    return getEntries();
+  }
+
+  public static final int BUILD_CONFIGURATION_FIELD_NUMBER = 3;
+  private org.tensorflow.proto.util.testlog.BuildConfiguration buildConfiguration_;
+  /**
+   * <pre>
+   * The configuration of the build (compiled opt? with cuda? any copts?)
+   * </pre>
+   *
+   * <code>.tensorflow.BuildConfiguration build_configuration = 3;</code>
+   */
+  public boolean hasBuildConfiguration() {
+    return buildConfiguration_ != null;
+  }
+  /**
+   * <pre>
+   * The configuration of the build (compiled opt? with cuda? any copts?)
+   * </pre>
+   *
+   * <code>.tensorflow.BuildConfiguration build_configuration = 3;</code>
+   */
+  public org.tensorflow.proto.util.testlog.BuildConfiguration getBuildConfiguration() {
+    return buildConfiguration_ == null ? org.tensorflow.proto.util.testlog.BuildConfiguration.getDefaultInstance() : buildConfiguration_;
+  }
+  /**
+   * <pre>
+   * The configuration of the build (compiled opt? with cuda? any copts?)
+   * </pre>
+   *
+   * <code>.tensorflow.BuildConfiguration build_configuration = 3;</code>
+   */
+  public org.tensorflow.proto.util.testlog.BuildConfigurationOrBuilder getBuildConfigurationOrBuilder() {
+    return getBuildConfiguration();
+  }
+
+  public static final int COMMIT_ID_FIELD_NUMBER = 4;
+  private org.tensorflow.proto.util.testlog.CommitId commitId_;
+  /**
+   * <pre>
+   * The commit id (git hash or changelist)
+   * </pre>
+   *
+   * <code>.tensorflow.CommitId commit_id = 4;</code>
+   */
+  public boolean hasCommitId() {
+    return commitId_ != null;
+  }
+  /**
+   * <pre>
+   * The commit id (git hash or changelist)
+   * </pre>
+   *
+   * <code>.tensorflow.CommitId commit_id = 4;</code>
+   */
+  public org.tensorflow.proto.util.testlog.CommitId getCommitId() {
+    return commitId_ == null ? org.tensorflow.proto.util.testlog.CommitId.getDefaultInstance() : commitId_;
+  }
+  /**
+   * <pre>
+   * The commit id (git hash or changelist)
+   * </pre>
+   *
+   * <code>.tensorflow.CommitId commit_id = 4;</code>
+   */
+  public org.tensorflow.proto.util.testlog.CommitIdOrBuilder getCommitIdOrBuilder() {
+    return getCommitId();
+  }
+
+  public static final int START_TIME_FIELD_NUMBER = 5;
+  private long startTime_;
+  /**
+   * <pre>
+   * The time the run started (in seconds of UTC time since Unix epoch)
+   * </pre>
+   *
+   * <code>int64 start_time = 5;</code>
+   */
+  public long getStartTime() {
+    return startTime_;
+  }
+
+  public static final int RUN_TIME_FIELD_NUMBER = 6;
+  private double runTime_;
+  /**
+   * <pre>
+   * The amount of time the total run took (wall time in seconds)
+   * </pre>
+   *
+   * <code>double run_time = 6;</code>
+   */
+  public double getRunTime() {
+    return runTime_;
+  }
+
+  public static final int MACHINE_CONFIGURATION_FIELD_NUMBER = 7;
+  private org.tensorflow.proto.util.testlog.MachineConfiguration machineConfiguration_;
+  /**
+   * <pre>
+   * Machine-specific parameters (Platform and CPU info)
+   * </pre>
+   *
+   * <code>.tensorflow.MachineConfiguration machine_configuration = 7;</code>
+   */
+  public boolean hasMachineConfiguration() {
+    return machineConfiguration_ != null;
+  }
+  /**
+   * <pre>
+   * Machine-specific parameters (Platform and CPU info)
+   * </pre>
+   *
+   * <code>.tensorflow.MachineConfiguration machine_configuration = 7;</code>
+   */
+  public org.tensorflow.proto.util.testlog.MachineConfiguration getMachineConfiguration() {
+    return machineConfiguration_ == null ? org.tensorflow.proto.util.testlog.MachineConfiguration.getDefaultInstance() : machineConfiguration_;
+  }
+  /**
+   * <pre>
+   * Machine-specific parameters (Platform and CPU info)
+   * </pre>
+   *
+   * <code>.tensorflow.MachineConfiguration machine_configuration = 7;</code>
+   */
+  public org.tensorflow.proto.util.testlog.MachineConfigurationOrBuilder getMachineConfigurationOrBuilder() {
+    return getMachineConfiguration();
+  }
+
+  public static final int RUN_CONFIGURATION_FIELD_NUMBER = 8;
+  private org.tensorflow.proto.util.testlog.RunConfiguration runConfiguration_;
+  /**
+   * <pre>
+   * Run-specific parameters (arguments, etc)
+   * </pre>
+   *
+   * <code>.tensorflow.RunConfiguration run_configuration = 8;</code>
+   */
+  public boolean hasRunConfiguration() {
+    return runConfiguration_ != null;
+  }
+  /**
+   * <pre>
+   * Run-specific parameters (arguments, etc)
+   * </pre>
+   *
+   * <code>.tensorflow.RunConfiguration run_configuration = 8;</code>
+   */
+  public org.tensorflow.proto.util.testlog.RunConfiguration getRunConfiguration() {
+    return runConfiguration_ == null ? org.tensorflow.proto.util.testlog.RunConfiguration.getDefaultInstance() : runConfiguration_;
+  }
+  /**
+   * <pre>
+   * Run-specific parameters (arguments, etc)
+   * </pre>
+   *
+   * <code>.tensorflow.RunConfiguration run_configuration = 8;</code>
+   */
+  public org.tensorflow.proto.util.testlog.RunConfigurationOrBuilder getRunConfigurationOrBuilder() {
+    return getRunConfiguration();
+  }
+
+  public static final int NAME_FIELD_NUMBER = 9;
+  private volatile java.lang.Object name_;
+  /**
+   * <pre>
+   * Benchmark target identifier.
+   * </pre>
+   *
+   * <code>string name = 9;</code>
+   */
+  public java.lang.String getName() {
+    java.lang.Object ref = name_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      name_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * Benchmark target identifier.
+   * </pre>
+   *
+   * <code>string name = 9;</code>
+   */
+  public com.google.protobuf.ByteString
+      getNameBytes() {
+    java.lang.Object ref = name_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      name_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int BENCHMARK_TYPE_FIELD_NUMBER = 10;
+  private int benchmarkType_;
+  /**
+   * <code>.tensorflow.TestResults.BenchmarkType benchmark_type = 10;</code>
+   */
+  public int getBenchmarkTypeValue() {
+    return benchmarkType_;
+  }
+  /**
+   * <code>.tensorflow.TestResults.BenchmarkType benchmark_type = 10;</code>
+   */
+  public org.tensorflow.proto.util.testlog.TestResults.BenchmarkType getBenchmarkType() {
+    @SuppressWarnings("deprecation")
+    org.tensorflow.proto.util.testlog.TestResults.BenchmarkType result = org.tensorflow.proto.util.testlog.TestResults.BenchmarkType.valueOf(benchmarkType_);
+    return result == null ? org.tensorflow.proto.util.testlog.TestResults.BenchmarkType.UNRECOGNIZED : result;
+  }
+
+  public static final int RUN_MODE_FIELD_NUMBER = 11;
+  private volatile java.lang.Object runMode_;
+  /**
+   * <pre>
+   * Used for differentiating between continuous and debug builds.
+   * Must be one of:
+   * * cbuild: results from continuous build.
+   * * presubmit: results from oneshot requests.
+   * * culprit: results from culprit finder rerun.
+   * </pre>
+   *
+   * <code>string run_mode = 11;</code>
+   */
+  public java.lang.String getRunMode() {
+    java.lang.Object ref = runMode_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      runMode_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * Used for differentiating between continuous and debug builds.
+   * Must be one of:
+   * * cbuild: results from continuous build.
+   * * presubmit: results from oneshot requests.
+   * * culprit: results from culprit finder rerun.
+   * </pre>
+   *
+   * <code>string run_mode = 11;</code>
+   */
+  public com.google.protobuf.ByteString
+      getRunModeBytes() {
+    java.lang.Object ref = runMode_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      runMode_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  public static final int TF_VERSION_FIELD_NUMBER = 12;
+  private volatile java.lang.Object tfVersion_;
+  /**
+   * <pre>
+   * TensorFlow version this benchmark runs against.
+   * This can be either set to full version or just the major version.
+   * </pre>
+   *
+   * <code>string tf_version = 12;</code>
+   */
+  public java.lang.String getTfVersion() {
+    java.lang.Object ref = tfVersion_;
+    if (ref instanceof java.lang.String) {
+      return (java.lang.String) ref;
+    } else {
+      com.google.protobuf.ByteString bs = 
+          (com.google.protobuf.ByteString) ref;
+      java.lang.String s = bs.toStringUtf8();
+      tfVersion_ = s;
+      return s;
+    }
+  }
+  /**
+   * <pre>
+   * TensorFlow version this benchmark runs against.
+   * This can be either set to full version or just the major version.
+   * </pre>
+   *
+   * <code>string tf_version = 12;</code>
+   */
+  public com.google.protobuf.ByteString
+      getTfVersionBytes() {
+    java.lang.Object ref = tfVersion_;
+    if (ref instanceof java.lang.String) {
+      com.google.protobuf.ByteString b = 
+          com.google.protobuf.ByteString.copyFromUtf8(
+              (java.lang.String) ref);
+      tfVersion_ = b;
+      return b;
+    } else {
+      return (com.google.protobuf.ByteString) ref;
+    }
+  }
+
+  private byte memoizedIsInitialized = -1;
+  @java.lang.Override
+  public final boolean isInitialized() {
+    byte isInitialized = memoizedIsInitialized;
+    if (isInitialized == 1) return true;
+    if (isInitialized == 0) return false;
+
+    memoizedIsInitialized = 1;
+    return true;
+  }
+
+  @java.lang.Override
+  public void writeTo(com.google.protobuf.CodedOutputStream output)
+                      throws java.io.IOException {
+    if (!getTargetBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 1, target_);
+    }
+    if (entries_ != null) {
+      output.writeMessage(2, getEntries());
+    }
+    if (buildConfiguration_ != null) {
+      output.writeMessage(3, getBuildConfiguration());
+    }
+    if (commitId_ != null) {
+      output.writeMessage(4, getCommitId());
+    }
+    if (startTime_ != 0L) {
+      output.writeInt64(5, startTime_);
+    }
+    if (runTime_ != 0D) {
+      output.writeDouble(6, runTime_);
+    }
+    if (machineConfiguration_ != null) {
+      output.writeMessage(7, getMachineConfiguration());
+    }
+    if (runConfiguration_ != null) {
+      output.writeMessage(8, getRunConfiguration());
+    }
+    if (!getNameBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 9, name_);
+    }
+    if (benchmarkType_ != org.tensorflow.proto.util.testlog.TestResults.BenchmarkType.UNKNOWN.getNumber()) {
+      output.writeEnum(10, benchmarkType_);
+    }
+    if (!getRunModeBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 11, runMode_);
+    }
+    if (!getTfVersionBytes().isEmpty()) {
+      com.google.protobuf.GeneratedMessageV3.writeString(output, 12, tfVersion_);
+    }
+    unknownFields.writeTo(output);
+  }
+
+  @java.lang.Override
+  public int getSerializedSize() {
+    int size = memoizedSize;
+    if (size != -1) return size;
+
+    size = 0;
+    if (!getTargetBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(1, target_);
+    }
+    if (entries_ != null) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(2, getEntries());
+    }
+    if (buildConfiguration_ != null) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(3, getBuildConfiguration());
+    }
+    if (commitId_ != null) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(4, getCommitId());
+    }
+    if (startTime_ != 0L) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeInt64Size(5, startTime_);
+    }
+    if (runTime_ != 0D) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeDoubleSize(6, runTime_);
+    }
+    if (machineConfiguration_ != null) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(7, getMachineConfiguration());
+    }
+    if (runConfiguration_ != null) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeMessageSize(8, getRunConfiguration());
+    }
+    if (!getNameBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(9, name_);
+    }
+    if (benchmarkType_ != org.tensorflow.proto.util.testlog.TestResults.BenchmarkType.UNKNOWN.getNumber()) {
+      size += com.google.protobuf.CodedOutputStream
+        .computeEnumSize(10, benchmarkType_);
+    }
+    if (!getRunModeBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(11, runMode_);
+    }
+    if (!getTfVersionBytes().isEmpty()) {
+      size += com.google.protobuf.GeneratedMessageV3.computeStringSize(12, tfVersion_);
+    }
+    size += unknownFields.getSerializedSize();
+    memoizedSize = size;
+    return size;
+  }
+
+  @java.lang.Override
+  public boolean equals(final java.lang.Object obj) {
+    if (obj == this) {
+     return true;
+    }
+    if (!(obj instanceof org.tensorflow.proto.util.testlog.TestResults)) {
+      return super.equals(obj);
+    }
+    org.tensorflow.proto.util.testlog.TestResults other = (org.tensorflow.proto.util.testlog.TestResults) obj;
+
+    if (!getTarget()
+        .equals(other.getTarget())) return false;
+    if (hasEntries() != other.hasEntries()) return false;
+    if (hasEntries()) {
+      if (!getEntries()
+          .equals(other.getEntries())) return false;
+    }
+    if (hasBuildConfiguration() != other.hasBuildConfiguration()) return false;
+    if (hasBuildConfiguration()) {
+      if (!getBuildConfiguration()
+          .equals(other.getBuildConfiguration())) return false;
+    }
+    if (hasCommitId() != other.hasCommitId()) return false;
+    if (hasCommitId()) {
+      if (!getCommitId()
+          .equals(other.getCommitId())) return false;
+    }
+    if (getStartTime()
+        != other.getStartTime()) return false;
+    if (java.lang.Double.doubleToLongBits(getRunTime())
+        != java.lang.Double.doubleToLongBits(
+            other.getRunTime())) return false;
+    if (hasMachineConfiguration() != other.hasMachineConfiguration()) return false;
+    if (hasMachineConfiguration()) {
+      if (!getMachineConfiguration()
+          .equals(other.getMachineConfiguration())) return false;
+    }
+    if (hasRunConfiguration() != other.hasRunConfiguration()) return false;
+    if (hasRunConfiguration()) {
+      if (!getRunConfiguration()
+          .equals(other.getRunConfiguration())) return false;
+    }
+    if (!getName()
+        .equals(other.getName())) return false;
+    if (benchmarkType_ != other.benchmarkType_) return false;
+    if (!getRunMode()
+        .equals(other.getRunMode())) return false;
+    if (!getTfVersion()
+        .equals(other.getTfVersion())) return false;
+    if (!unknownFields.equals(other.unknownFields)) return false;
+    return true;
+  }
+
+  @java.lang.Override
+  public int hashCode() {
+    if (memoizedHashCode != 0) {
+      return memoizedHashCode;
+    }
+    int hash = 41;
+    hash = (19 * hash) + getDescriptor().hashCode();
+    hash = (37 * hash) + TARGET_FIELD_NUMBER;
+    hash = (53 * hash) + getTarget().hashCode();
+    if (hasEntries()) {
+      hash = (37 * hash) + ENTRIES_FIELD_NUMBER;
+      hash = (53 * hash) + getEntries().hashCode();
+    }
+    if (hasBuildConfiguration()) {
+      hash = (37 * hash) + BUILD_CONFIGURATION_FIELD_NUMBER;
+      hash = (53 * hash) + getBuildConfiguration().hashCode();
+    }
+    if (hasCommitId()) {
+      hash = (37 * hash) + COMMIT_ID_FIELD_NUMBER;
+      hash = (53 * hash) + getCommitId().hashCode();
+    }
+    hash = (37 * hash) + START_TIME_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+        getStartTime());
+    hash = (37 * hash) + RUN_TIME_FIELD_NUMBER;
+    hash = (53 * hash) + com.google.protobuf.Internal.hashLong(
+        java.lang.Double.doubleToLongBits(getRunTime()));
+    if (hasMachineConfiguration()) {
+      hash = (37 * hash) + MACHINE_CONFIGURATION_FIELD_NUMBER;
+      hash = (53 * hash) + getMachineConfiguration().hashCode();
+    }
+    if (hasRunConfiguration()) {
+      hash = (37 * hash) + RUN_CONFIGURATION_FIELD_NUMBER;
+      hash = (53 * hash) + getRunConfiguration().hashCode();
+    }
+    hash = (37 * hash) + NAME_FIELD_NUMBER;
+    hash = (53 * hash) + getName().hashCode();
+    hash = (37 * hash) + BENCHMARK_TYPE_FIELD_NUMBER;
+    hash = (53 * hash) + benchmarkType_;
+    hash = (37 * hash) + RUN_MODE_FIELD_NUMBER;
+    hash = (53 * hash) + getRunMode().hashCode();
+    hash = (37 * hash) + TF_VERSION_FIELD_NUMBER;
+    hash = (53 * hash) + getTfVersion().hashCode();
+    hash = (29 * hash) + unknownFields.hashCode();
+    memoizedHashCode = hash;
+    return hash;
+  }
+
+  public static org.tensorflow.proto.util.testlog.TestResults parseFrom(
+      java.nio.ByteBuffer data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.testlog.TestResults parseFrom(
+      java.nio.ByteBuffer data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.testlog.TestResults parseFrom(
+      com.google.protobuf.ByteString data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.testlog.TestResults parseFrom(
+      com.google.protobuf.ByteString data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.testlog.TestResults parseFrom(byte[] data)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data);
+  }
+  public static org.tensorflow.proto.util.testlog.TestResults parseFrom(
+      byte[] data,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws com.google.protobuf.InvalidProtocolBufferException {
+    return PARSER.parseFrom(data, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.testlog.TestResults parseFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.testlog.TestResults parseFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.testlog.TestResults parseDelimitedFrom(java.io.InputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.testlog.TestResults parseDelimitedFrom(
+      java.io.InputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseDelimitedWithIOException(PARSER, input, extensionRegistry);
+  }
+  public static org.tensorflow.proto.util.testlog.TestResults parseFrom(
+      com.google.protobuf.CodedInputStream input)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input);
+  }
+  public static org.tensorflow.proto.util.testlog.TestResults parseFrom(
+      com.google.protobuf.CodedInputStream input,
+      com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+      throws java.io.IOException {
+    return com.google.protobuf.GeneratedMessageV3
+        .parseWithIOException(PARSER, input, extensionRegistry);
+  }
+
+  @java.lang.Override
+  public Builder newBuilderForType() { return newBuilder(); }
+  public static Builder newBuilder() {
+    return DEFAULT_INSTANCE.toBuilder();
+  }
+  public static Builder newBuilder(org.tensorflow.proto.util.testlog.TestResults prototype) {
+    return DEFAULT_INSTANCE.toBuilder().mergeFrom(prototype);
+  }
+  @java.lang.Override
+  public Builder toBuilder() {
+    return this == DEFAULT_INSTANCE
+        ? new Builder() : new Builder().mergeFrom(this);
+  }
+
+  @java.lang.Override
+  protected Builder newBuilderForType(
+      com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+    Builder builder = new Builder(parent);
+    return builder;
+  }
+  /**
+   * <pre>
+   * The output of one benchmark / test run.  Each run contains a list of
+   * tests or benchmarks, stored as BenchmarkEntry messages.
+   * This message should be emitted by the reporter (which runs the
+   * test / BM in a subprocess and then reads the emitted BenchmarkEntry messages;
+   * usually from a serialized json file, finally collecting them along
+   * with additional information about the test run.
+   * </pre>
+   *
+   * Protobuf type {@code tensorflow.TestResults}
+   */
+  public static final class Builder extends
+      com.google.protobuf.GeneratedMessageV3.Builder<Builder> implements
+      // @@protoc_insertion_point(builder_implements:tensorflow.TestResults)
+      org.tensorflow.proto.util.testlog.TestResultsOrBuilder {
+    public static final com.google.protobuf.Descriptors.Descriptor
+        getDescriptor() {
+      return org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_TestResults_descriptor;
+    }
+
+    @java.lang.Override
+    protected com.google.protobuf.GeneratedMessageV3.FieldAccessorTable
+        internalGetFieldAccessorTable() {
+      return org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_TestResults_fieldAccessorTable
+          .ensureFieldAccessorsInitialized(
+              org.tensorflow.proto.util.testlog.TestResults.class, org.tensorflow.proto.util.testlog.TestResults.Builder.class);
+    }
+
+    // Construct using org.tensorflow.proto.util.testlog.TestResults.newBuilder()
+    private Builder() {
+      maybeForceBuilderInitialization();
+    }
+
+    private Builder(
+        com.google.protobuf.GeneratedMessageV3.BuilderParent parent) {
+      super(parent);
+      maybeForceBuilderInitialization();
+    }
+    private void maybeForceBuilderInitialization() {
+      if (com.google.protobuf.GeneratedMessageV3
+              .alwaysUseFieldBuilders) {
+      }
+    }
+    @java.lang.Override
+    public Builder clear() {
+      super.clear();
+      target_ = "";
+
+      if (entriesBuilder_ == null) {
+        entries_ = null;
+      } else {
+        entries_ = null;
+        entriesBuilder_ = null;
+      }
+      if (buildConfigurationBuilder_ == null) {
+        buildConfiguration_ = null;
+      } else {
+        buildConfiguration_ = null;
+        buildConfigurationBuilder_ = null;
+      }
+      if (commitIdBuilder_ == null) {
+        commitId_ = null;
+      } else {
+        commitId_ = null;
+        commitIdBuilder_ = null;
+      }
+      startTime_ = 0L;
+
+      runTime_ = 0D;
+
+      if (machineConfigurationBuilder_ == null) {
+        machineConfiguration_ = null;
+      } else {
+        machineConfiguration_ = null;
+        machineConfigurationBuilder_ = null;
+      }
+      if (runConfigurationBuilder_ == null) {
+        runConfiguration_ = null;
+      } else {
+        runConfiguration_ = null;
+        runConfigurationBuilder_ = null;
+      }
+      name_ = "";
+
+      benchmarkType_ = 0;
+
+      runMode_ = "";
+
+      tfVersion_ = "";
+
+      return this;
+    }
+
+    @java.lang.Override
+    public com.google.protobuf.Descriptors.Descriptor
+        getDescriptorForType() {
+      return org.tensorflow.proto.util.testlog.TestLogProtos.internal_static_tensorflow_TestResults_descriptor;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.testlog.TestResults getDefaultInstanceForType() {
+      return org.tensorflow.proto.util.testlog.TestResults.getDefaultInstance();
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.testlog.TestResults build() {
+      org.tensorflow.proto.util.testlog.TestResults result = buildPartial();
+      if (!result.isInitialized()) {
+        throw newUninitializedMessageException(result);
+      }
+      return result;
+    }
+
+    @java.lang.Override
+    public org.tensorflow.proto.util.testlog.TestResults buildPartial() {
+      org.tensorflow.proto.util.testlog.TestResults result = new org.tensorflow.proto.util.testlog.TestResults(this);
+      result.target_ = target_;
+      if (entriesBuilder_ == null) {
+        result.entries_ = entries_;
+      } else {
+        result.entries_ = entriesBuilder_.build();
+      }
+      if (buildConfigurationBuilder_ == null) {
+        result.buildConfiguration_ = buildConfiguration_;
+      } else {
+        result.buildConfiguration_ = buildConfigurationBuilder_.build();
+      }
+      if (commitIdBuilder_ == null) {
+        result.commitId_ = commitId_;
+      } else {
+        result.commitId_ = commitIdBuilder_.build();
+      }
+      result.startTime_ = startTime_;
+      result.runTime_ = runTime_;
+      if (machineConfigurationBuilder_ == null) {
+        result.machineConfiguration_ = machineConfiguration_;
+      } else {
+        result.machineConfiguration_ = machineConfigurationBuilder_.build();
+      }
+      if (runConfigurationBuilder_ == null) {
+        result.runConfiguration_ = runConfiguration_;
+      } else {
+        result.runConfiguration_ = runConfigurationBuilder_.build();
+      }
+      result.name_ = name_;
+      result.benchmarkType_ = benchmarkType_;
+      result.runMode_ = runMode_;
+      result.tfVersion_ = tfVersion_;
+      onBuilt();
+      return result;
+    }
+
+    @java.lang.Override
+    public Builder clone() {
+      return super.clone();
+    }
+    @java.lang.Override
+    public Builder setField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.setField(field, value);
+    }
+    @java.lang.Override
+    public Builder clearField(
+        com.google.protobuf.Descriptors.FieldDescriptor field) {
+      return super.clearField(field);
+    }
+    @java.lang.Override
+    public Builder clearOneof(
+        com.google.protobuf.Descriptors.OneofDescriptor oneof) {
+      return super.clearOneof(oneof);
+    }
+    @java.lang.Override
+    public Builder setRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        int index, java.lang.Object value) {
+      return super.setRepeatedField(field, index, value);
+    }
+    @java.lang.Override
+    public Builder addRepeatedField(
+        com.google.protobuf.Descriptors.FieldDescriptor field,
+        java.lang.Object value) {
+      return super.addRepeatedField(field, value);
+    }
+    @java.lang.Override
+    public Builder mergeFrom(com.google.protobuf.Message other) {
+      if (other instanceof org.tensorflow.proto.util.testlog.TestResults) {
+        return mergeFrom((org.tensorflow.proto.util.testlog.TestResults)other);
+      } else {
+        super.mergeFrom(other);
+        return this;
+      }
+    }
+
+    public Builder mergeFrom(org.tensorflow.proto.util.testlog.TestResults other) {
+      if (other == org.tensorflow.proto.util.testlog.TestResults.getDefaultInstance()) return this;
+      if (!other.getTarget().isEmpty()) {
+        target_ = other.target_;
+        onChanged();
+      }
+      if (other.hasEntries()) {
+        mergeEntries(other.getEntries());
+      }
+      if (other.hasBuildConfiguration()) {
+        mergeBuildConfiguration(other.getBuildConfiguration());
+      }
+      if (other.hasCommitId()) {
+        mergeCommitId(other.getCommitId());
+      }
+      if (other.getStartTime() != 0L) {
+        setStartTime(other.getStartTime());
+      }
+      if (other.getRunTime() != 0D) {
+        setRunTime(other.getRunTime());
+      }
+      if (other.hasMachineConfiguration()) {
+        mergeMachineConfiguration(other.getMachineConfiguration());
+      }
+      if (other.hasRunConfiguration()) {
+        mergeRunConfiguration(other.getRunConfiguration());
+      }
+      if (!other.getName().isEmpty()) {
+        name_ = other.name_;
+        onChanged();
+      }
+      if (other.benchmarkType_ != 0) {
+        setBenchmarkTypeValue(other.getBenchmarkTypeValue());
+      }
+      if (!other.getRunMode().isEmpty()) {
+        runMode_ = other.runMode_;
+        onChanged();
+      }
+      if (!other.getTfVersion().isEmpty()) {
+        tfVersion_ = other.tfVersion_;
+        onChanged();
+      }
+      this.mergeUnknownFields(other.unknownFields);
+      onChanged();
+      return this;
+    }
+
+    @java.lang.Override
+    public final boolean isInitialized() {
+      return true;
+    }
+
+    @java.lang.Override
+    public Builder mergeFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws java.io.IOException {
+      org.tensorflow.proto.util.testlog.TestResults parsedMessage = null;
+      try {
+        parsedMessage = PARSER.parsePartialFrom(input, extensionRegistry);
+      } catch (com.google.protobuf.InvalidProtocolBufferException e) {
+        parsedMessage = (org.tensorflow.proto.util.testlog.TestResults) e.getUnfinishedMessage();
+        throw e.unwrapIOException();
+      } finally {
+        if (parsedMessage != null) {
+          mergeFrom(parsedMessage);
+        }
+      }
+      return this;
+    }
+
+    private java.lang.Object target_ = "";
+    /**
+     * <pre>
+     * The target of the run, e.g.:
+     *  //tensorflow/core:kernels_adjust_contrast_op_benchmark_test
+     * </pre>
+     *
+     * <code>string target = 1;</code>
+     */
+    public java.lang.String getTarget() {
+      java.lang.Object ref = target_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        target_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * The target of the run, e.g.:
+     *  //tensorflow/core:kernels_adjust_contrast_op_benchmark_test
+     * </pre>
+     *
+     * <code>string target = 1;</code>
+     */
+    public com.google.protobuf.ByteString
+        getTargetBytes() {
+      java.lang.Object ref = target_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        target_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * The target of the run, e.g.:
+     *  //tensorflow/core:kernels_adjust_contrast_op_benchmark_test
+     * </pre>
+     *
+     * <code>string target = 1;</code>
+     */
+    public Builder setTarget(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      target_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * The target of the run, e.g.:
+     *  //tensorflow/core:kernels_adjust_contrast_op_benchmark_test
+     * </pre>
+     *
+     * <code>string target = 1;</code>
+     */
+    public Builder clearTarget() {
+      
+      target_ = getDefaultInstance().getTarget();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * The target of the run, e.g.:
+     *  //tensorflow/core:kernels_adjust_contrast_op_benchmark_test
+     * </pre>
+     *
+     * <code>string target = 1;</code>
+     */
+    public Builder setTargetBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      target_ = value;
+      onChanged();
+      return this;
+    }
+
+    private org.tensorflow.proto.util.testlog.BenchmarkEntries entries_;
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.util.testlog.BenchmarkEntries, org.tensorflow.proto.util.testlog.BenchmarkEntries.Builder, org.tensorflow.proto.util.testlog.BenchmarkEntriesOrBuilder> entriesBuilder_;
+    /**
+     * <pre>
+     * The list of tests or benchmarks in this run.
+     * </pre>
+     *
+     * <code>.tensorflow.BenchmarkEntries entries = 2;</code>
+     */
+    public boolean hasEntries() {
+      return entriesBuilder_ != null || entries_ != null;
+    }
+    /**
+     * <pre>
+     * The list of tests or benchmarks in this run.
+     * </pre>
+     *
+     * <code>.tensorflow.BenchmarkEntries entries = 2;</code>
+     */
+    public org.tensorflow.proto.util.testlog.BenchmarkEntries getEntries() {
+      if (entriesBuilder_ == null) {
+        return entries_ == null ? org.tensorflow.proto.util.testlog.BenchmarkEntries.getDefaultInstance() : entries_;
+      } else {
+        return entriesBuilder_.getMessage();
+      }
+    }
+    /**
+     * <pre>
+     * The list of tests or benchmarks in this run.
+     * </pre>
+     *
+     * <code>.tensorflow.BenchmarkEntries entries = 2;</code>
+     */
+    public Builder setEntries(org.tensorflow.proto.util.testlog.BenchmarkEntries value) {
+      if (entriesBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        entries_ = value;
+        onChanged();
+      } else {
+        entriesBuilder_.setMessage(value);
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * The list of tests or benchmarks in this run.
+     * </pre>
+     *
+     * <code>.tensorflow.BenchmarkEntries entries = 2;</code>
+     */
+    public Builder setEntries(
+        org.tensorflow.proto.util.testlog.BenchmarkEntries.Builder builderForValue) {
+      if (entriesBuilder_ == null) {
+        entries_ = builderForValue.build();
+        onChanged();
+      } else {
+        entriesBuilder_.setMessage(builderForValue.build());
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * The list of tests or benchmarks in this run.
+     * </pre>
+     *
+     * <code>.tensorflow.BenchmarkEntries entries = 2;</code>
+     */
+    public Builder mergeEntries(org.tensorflow.proto.util.testlog.BenchmarkEntries value) {
+      if (entriesBuilder_ == null) {
+        if (entries_ != null) {
+          entries_ =
+            org.tensorflow.proto.util.testlog.BenchmarkEntries.newBuilder(entries_).mergeFrom(value).buildPartial();
+        } else {
+          entries_ = value;
+        }
+        onChanged();
+      } else {
+        entriesBuilder_.mergeFrom(value);
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * The list of tests or benchmarks in this run.
+     * </pre>
+     *
+     * <code>.tensorflow.BenchmarkEntries entries = 2;</code>
+     */
+    public Builder clearEntries() {
+      if (entriesBuilder_ == null) {
+        entries_ = null;
+        onChanged();
+      } else {
+        entries_ = null;
+        entriesBuilder_ = null;
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * The list of tests or benchmarks in this run.
+     * </pre>
+     *
+     * <code>.tensorflow.BenchmarkEntries entries = 2;</code>
+     */
+    public org.tensorflow.proto.util.testlog.BenchmarkEntries.Builder getEntriesBuilder() {
+      
+      onChanged();
+      return getEntriesFieldBuilder().getBuilder();
+    }
+    /**
+     * <pre>
+     * The list of tests or benchmarks in this run.
+     * </pre>
+     *
+     * <code>.tensorflow.BenchmarkEntries entries = 2;</code>
+     */
+    public org.tensorflow.proto.util.testlog.BenchmarkEntriesOrBuilder getEntriesOrBuilder() {
+      if (entriesBuilder_ != null) {
+        return entriesBuilder_.getMessageOrBuilder();
+      } else {
+        return entries_ == null ?
+            org.tensorflow.proto.util.testlog.BenchmarkEntries.getDefaultInstance() : entries_;
+      }
+    }
+    /**
+     * <pre>
+     * The list of tests or benchmarks in this run.
+     * </pre>
+     *
+     * <code>.tensorflow.BenchmarkEntries entries = 2;</code>
+     */
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.util.testlog.BenchmarkEntries, org.tensorflow.proto.util.testlog.BenchmarkEntries.Builder, org.tensorflow.proto.util.testlog.BenchmarkEntriesOrBuilder> 
+        getEntriesFieldBuilder() {
+      if (entriesBuilder_ == null) {
+        entriesBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+            org.tensorflow.proto.util.testlog.BenchmarkEntries, org.tensorflow.proto.util.testlog.BenchmarkEntries.Builder, org.tensorflow.proto.util.testlog.BenchmarkEntriesOrBuilder>(
+                getEntries(),
+                getParentForChildren(),
+                isClean());
+        entries_ = null;
+      }
+      return entriesBuilder_;
+    }
+
+    private org.tensorflow.proto.util.testlog.BuildConfiguration buildConfiguration_;
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.util.testlog.BuildConfiguration, org.tensorflow.proto.util.testlog.BuildConfiguration.Builder, org.tensorflow.proto.util.testlog.BuildConfigurationOrBuilder> buildConfigurationBuilder_;
+    /**
+     * <pre>
+     * The configuration of the build (compiled opt? with cuda? any copts?)
+     * </pre>
+     *
+     * <code>.tensorflow.BuildConfiguration build_configuration = 3;</code>
+     */
+    public boolean hasBuildConfiguration() {
+      return buildConfigurationBuilder_ != null || buildConfiguration_ != null;
+    }
+    /**
+     * <pre>
+     * The configuration of the build (compiled opt? with cuda? any copts?)
+     * </pre>
+     *
+     * <code>.tensorflow.BuildConfiguration build_configuration = 3;</code>
+     */
+    public org.tensorflow.proto.util.testlog.BuildConfiguration getBuildConfiguration() {
+      if (buildConfigurationBuilder_ == null) {
+        return buildConfiguration_ == null ? org.tensorflow.proto.util.testlog.BuildConfiguration.getDefaultInstance() : buildConfiguration_;
+      } else {
+        return buildConfigurationBuilder_.getMessage();
+      }
+    }
+    /**
+     * <pre>
+     * The configuration of the build (compiled opt? with cuda? any copts?)
+     * </pre>
+     *
+     * <code>.tensorflow.BuildConfiguration build_configuration = 3;</code>
+     */
+    public Builder setBuildConfiguration(org.tensorflow.proto.util.testlog.BuildConfiguration value) {
+      if (buildConfigurationBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        buildConfiguration_ = value;
+        onChanged();
+      } else {
+        buildConfigurationBuilder_.setMessage(value);
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * The configuration of the build (compiled opt? with cuda? any copts?)
+     * </pre>
+     *
+     * <code>.tensorflow.BuildConfiguration build_configuration = 3;</code>
+     */
+    public Builder setBuildConfiguration(
+        org.tensorflow.proto.util.testlog.BuildConfiguration.Builder builderForValue) {
+      if (buildConfigurationBuilder_ == null) {
+        buildConfiguration_ = builderForValue.build();
+        onChanged();
+      } else {
+        buildConfigurationBuilder_.setMessage(builderForValue.build());
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * The configuration of the build (compiled opt? with cuda? any copts?)
+     * </pre>
+     *
+     * <code>.tensorflow.BuildConfiguration build_configuration = 3;</code>
+     */
+    public Builder mergeBuildConfiguration(org.tensorflow.proto.util.testlog.BuildConfiguration value) {
+      if (buildConfigurationBuilder_ == null) {
+        if (buildConfiguration_ != null) {
+          buildConfiguration_ =
+            org.tensorflow.proto.util.testlog.BuildConfiguration.newBuilder(buildConfiguration_).mergeFrom(value).buildPartial();
+        } else {
+          buildConfiguration_ = value;
+        }
+        onChanged();
+      } else {
+        buildConfigurationBuilder_.mergeFrom(value);
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * The configuration of the build (compiled opt? with cuda? any copts?)
+     * </pre>
+     *
+     * <code>.tensorflow.BuildConfiguration build_configuration = 3;</code>
+     */
+    public Builder clearBuildConfiguration() {
+      if (buildConfigurationBuilder_ == null) {
+        buildConfiguration_ = null;
+        onChanged();
+      } else {
+        buildConfiguration_ = null;
+        buildConfigurationBuilder_ = null;
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * The configuration of the build (compiled opt? with cuda? any copts?)
+     * </pre>
+     *
+     * <code>.tensorflow.BuildConfiguration build_configuration = 3;</code>
+     */
+    public org.tensorflow.proto.util.testlog.BuildConfiguration.Builder getBuildConfigurationBuilder() {
+      
+      onChanged();
+      return getBuildConfigurationFieldBuilder().getBuilder();
+    }
+    /**
+     * <pre>
+     * The configuration of the build (compiled opt? with cuda? any copts?)
+     * </pre>
+     *
+     * <code>.tensorflow.BuildConfiguration build_configuration = 3;</code>
+     */
+    public org.tensorflow.proto.util.testlog.BuildConfigurationOrBuilder getBuildConfigurationOrBuilder() {
+      if (buildConfigurationBuilder_ != null) {
+        return buildConfigurationBuilder_.getMessageOrBuilder();
+      } else {
+        return buildConfiguration_ == null ?
+            org.tensorflow.proto.util.testlog.BuildConfiguration.getDefaultInstance() : buildConfiguration_;
+      }
+    }
+    /**
+     * <pre>
+     * The configuration of the build (compiled opt? with cuda? any copts?)
+     * </pre>
+     *
+     * <code>.tensorflow.BuildConfiguration build_configuration = 3;</code>
+     */
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.util.testlog.BuildConfiguration, org.tensorflow.proto.util.testlog.BuildConfiguration.Builder, org.tensorflow.proto.util.testlog.BuildConfigurationOrBuilder> 
+        getBuildConfigurationFieldBuilder() {
+      if (buildConfigurationBuilder_ == null) {
+        buildConfigurationBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+            org.tensorflow.proto.util.testlog.BuildConfiguration, org.tensorflow.proto.util.testlog.BuildConfiguration.Builder, org.tensorflow.proto.util.testlog.BuildConfigurationOrBuilder>(
+                getBuildConfiguration(),
+                getParentForChildren(),
+                isClean());
+        buildConfiguration_ = null;
+      }
+      return buildConfigurationBuilder_;
+    }
+
+    private org.tensorflow.proto.util.testlog.CommitId commitId_;
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.util.testlog.CommitId, org.tensorflow.proto.util.testlog.CommitId.Builder, org.tensorflow.proto.util.testlog.CommitIdOrBuilder> commitIdBuilder_;
+    /**
+     * <pre>
+     * The commit id (git hash or changelist)
+     * </pre>
+     *
+     * <code>.tensorflow.CommitId commit_id = 4;</code>
+     */
+    public boolean hasCommitId() {
+      return commitIdBuilder_ != null || commitId_ != null;
+    }
+    /**
+     * <pre>
+     * The commit id (git hash or changelist)
+     * </pre>
+     *
+     * <code>.tensorflow.CommitId commit_id = 4;</code>
+     */
+    public org.tensorflow.proto.util.testlog.CommitId getCommitId() {
+      if (commitIdBuilder_ == null) {
+        return commitId_ == null ? org.tensorflow.proto.util.testlog.CommitId.getDefaultInstance() : commitId_;
+      } else {
+        return commitIdBuilder_.getMessage();
+      }
+    }
+    /**
+     * <pre>
+     * The commit id (git hash or changelist)
+     * </pre>
+     *
+     * <code>.tensorflow.CommitId commit_id = 4;</code>
+     */
+    public Builder setCommitId(org.tensorflow.proto.util.testlog.CommitId value) {
+      if (commitIdBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        commitId_ = value;
+        onChanged();
+      } else {
+        commitIdBuilder_.setMessage(value);
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * The commit id (git hash or changelist)
+     * </pre>
+     *
+     * <code>.tensorflow.CommitId commit_id = 4;</code>
+     */
+    public Builder setCommitId(
+        org.tensorflow.proto.util.testlog.CommitId.Builder builderForValue) {
+      if (commitIdBuilder_ == null) {
+        commitId_ = builderForValue.build();
+        onChanged();
+      } else {
+        commitIdBuilder_.setMessage(builderForValue.build());
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * The commit id (git hash or changelist)
+     * </pre>
+     *
+     * <code>.tensorflow.CommitId commit_id = 4;</code>
+     */
+    public Builder mergeCommitId(org.tensorflow.proto.util.testlog.CommitId value) {
+      if (commitIdBuilder_ == null) {
+        if (commitId_ != null) {
+          commitId_ =
+            org.tensorflow.proto.util.testlog.CommitId.newBuilder(commitId_).mergeFrom(value).buildPartial();
+        } else {
+          commitId_ = value;
+        }
+        onChanged();
+      } else {
+        commitIdBuilder_.mergeFrom(value);
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * The commit id (git hash or changelist)
+     * </pre>
+     *
+     * <code>.tensorflow.CommitId commit_id = 4;</code>
+     */
+    public Builder clearCommitId() {
+      if (commitIdBuilder_ == null) {
+        commitId_ = null;
+        onChanged();
+      } else {
+        commitId_ = null;
+        commitIdBuilder_ = null;
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * The commit id (git hash or changelist)
+     * </pre>
+     *
+     * <code>.tensorflow.CommitId commit_id = 4;</code>
+     */
+    public org.tensorflow.proto.util.testlog.CommitId.Builder getCommitIdBuilder() {
+      
+      onChanged();
+      return getCommitIdFieldBuilder().getBuilder();
+    }
+    /**
+     * <pre>
+     * The commit id (git hash or changelist)
+     * </pre>
+     *
+     * <code>.tensorflow.CommitId commit_id = 4;</code>
+     */
+    public org.tensorflow.proto.util.testlog.CommitIdOrBuilder getCommitIdOrBuilder() {
+      if (commitIdBuilder_ != null) {
+        return commitIdBuilder_.getMessageOrBuilder();
+      } else {
+        return commitId_ == null ?
+            org.tensorflow.proto.util.testlog.CommitId.getDefaultInstance() : commitId_;
+      }
+    }
+    /**
+     * <pre>
+     * The commit id (git hash or changelist)
+     * </pre>
+     *
+     * <code>.tensorflow.CommitId commit_id = 4;</code>
+     */
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.util.testlog.CommitId, org.tensorflow.proto.util.testlog.CommitId.Builder, org.tensorflow.proto.util.testlog.CommitIdOrBuilder> 
+        getCommitIdFieldBuilder() {
+      if (commitIdBuilder_ == null) {
+        commitIdBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+            org.tensorflow.proto.util.testlog.CommitId, org.tensorflow.proto.util.testlog.CommitId.Builder, org.tensorflow.proto.util.testlog.CommitIdOrBuilder>(
+                getCommitId(),
+                getParentForChildren(),
+                isClean());
+        commitId_ = null;
+      }
+      return commitIdBuilder_;
+    }
+
+    private long startTime_ ;
+    /**
+     * <pre>
+     * The time the run started (in seconds of UTC time since Unix epoch)
+     * </pre>
+     *
+     * <code>int64 start_time = 5;</code>
+     */
+    public long getStartTime() {
+      return startTime_;
+    }
+    /**
+     * <pre>
+     * The time the run started (in seconds of UTC time since Unix epoch)
+     * </pre>
+     *
+     * <code>int64 start_time = 5;</code>
+     */
+    public Builder setStartTime(long value) {
+      
+      startTime_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * The time the run started (in seconds of UTC time since Unix epoch)
+     * </pre>
+     *
+     * <code>int64 start_time = 5;</code>
+     */
+    public Builder clearStartTime() {
+      
+      startTime_ = 0L;
+      onChanged();
+      return this;
+    }
+
+    private double runTime_ ;
+    /**
+     * <pre>
+     * The amount of time the total run took (wall time in seconds)
+     * </pre>
+     *
+     * <code>double run_time = 6;</code>
+     */
+    public double getRunTime() {
+      return runTime_;
+    }
+    /**
+     * <pre>
+     * The amount of time the total run took (wall time in seconds)
+     * </pre>
+     *
+     * <code>double run_time = 6;</code>
+     */
+    public Builder setRunTime(double value) {
+      
+      runTime_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * The amount of time the total run took (wall time in seconds)
+     * </pre>
+     *
+     * <code>double run_time = 6;</code>
+     */
+    public Builder clearRunTime() {
+      
+      runTime_ = 0D;
+      onChanged();
+      return this;
+    }
+
+    private org.tensorflow.proto.util.testlog.MachineConfiguration machineConfiguration_;
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.util.testlog.MachineConfiguration, org.tensorflow.proto.util.testlog.MachineConfiguration.Builder, org.tensorflow.proto.util.testlog.MachineConfigurationOrBuilder> machineConfigurationBuilder_;
+    /**
+     * <pre>
+     * Machine-specific parameters (Platform and CPU info)
+     * </pre>
+     *
+     * <code>.tensorflow.MachineConfiguration machine_configuration = 7;</code>
+     */
+    public boolean hasMachineConfiguration() {
+      return machineConfigurationBuilder_ != null || machineConfiguration_ != null;
+    }
+    /**
+     * <pre>
+     * Machine-specific parameters (Platform and CPU info)
+     * </pre>
+     *
+     * <code>.tensorflow.MachineConfiguration machine_configuration = 7;</code>
+     */
+    public org.tensorflow.proto.util.testlog.MachineConfiguration getMachineConfiguration() {
+      if (machineConfigurationBuilder_ == null) {
+        return machineConfiguration_ == null ? org.tensorflow.proto.util.testlog.MachineConfiguration.getDefaultInstance() : machineConfiguration_;
+      } else {
+        return machineConfigurationBuilder_.getMessage();
+      }
+    }
+    /**
+     * <pre>
+     * Machine-specific parameters (Platform and CPU info)
+     * </pre>
+     *
+     * <code>.tensorflow.MachineConfiguration machine_configuration = 7;</code>
+     */
+    public Builder setMachineConfiguration(org.tensorflow.proto.util.testlog.MachineConfiguration value) {
+      if (machineConfigurationBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        machineConfiguration_ = value;
+        onChanged();
+      } else {
+        machineConfigurationBuilder_.setMessage(value);
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Machine-specific parameters (Platform and CPU info)
+     * </pre>
+     *
+     * <code>.tensorflow.MachineConfiguration machine_configuration = 7;</code>
+     */
+    public Builder setMachineConfiguration(
+        org.tensorflow.proto.util.testlog.MachineConfiguration.Builder builderForValue) {
+      if (machineConfigurationBuilder_ == null) {
+        machineConfiguration_ = builderForValue.build();
+        onChanged();
+      } else {
+        machineConfigurationBuilder_.setMessage(builderForValue.build());
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Machine-specific parameters (Platform and CPU info)
+     * </pre>
+     *
+     * <code>.tensorflow.MachineConfiguration machine_configuration = 7;</code>
+     */
+    public Builder mergeMachineConfiguration(org.tensorflow.proto.util.testlog.MachineConfiguration value) {
+      if (machineConfigurationBuilder_ == null) {
+        if (machineConfiguration_ != null) {
+          machineConfiguration_ =
+            org.tensorflow.proto.util.testlog.MachineConfiguration.newBuilder(machineConfiguration_).mergeFrom(value).buildPartial();
+        } else {
+          machineConfiguration_ = value;
+        }
+        onChanged();
+      } else {
+        machineConfigurationBuilder_.mergeFrom(value);
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Machine-specific parameters (Platform and CPU info)
+     * </pre>
+     *
+     * <code>.tensorflow.MachineConfiguration machine_configuration = 7;</code>
+     */
+    public Builder clearMachineConfiguration() {
+      if (machineConfigurationBuilder_ == null) {
+        machineConfiguration_ = null;
+        onChanged();
+      } else {
+        machineConfiguration_ = null;
+        machineConfigurationBuilder_ = null;
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Machine-specific parameters (Platform and CPU info)
+     * </pre>
+     *
+     * <code>.tensorflow.MachineConfiguration machine_configuration = 7;</code>
+     */
+    public org.tensorflow.proto.util.testlog.MachineConfiguration.Builder getMachineConfigurationBuilder() {
+      
+      onChanged();
+      return getMachineConfigurationFieldBuilder().getBuilder();
+    }
+    /**
+     * <pre>
+     * Machine-specific parameters (Platform and CPU info)
+     * </pre>
+     *
+     * <code>.tensorflow.MachineConfiguration machine_configuration = 7;</code>
+     */
+    public org.tensorflow.proto.util.testlog.MachineConfigurationOrBuilder getMachineConfigurationOrBuilder() {
+      if (machineConfigurationBuilder_ != null) {
+        return machineConfigurationBuilder_.getMessageOrBuilder();
+      } else {
+        return machineConfiguration_ == null ?
+            org.tensorflow.proto.util.testlog.MachineConfiguration.getDefaultInstance() : machineConfiguration_;
+      }
+    }
+    /**
+     * <pre>
+     * Machine-specific parameters (Platform and CPU info)
+     * </pre>
+     *
+     * <code>.tensorflow.MachineConfiguration machine_configuration = 7;</code>
+     */
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.util.testlog.MachineConfiguration, org.tensorflow.proto.util.testlog.MachineConfiguration.Builder, org.tensorflow.proto.util.testlog.MachineConfigurationOrBuilder> 
+        getMachineConfigurationFieldBuilder() {
+      if (machineConfigurationBuilder_ == null) {
+        machineConfigurationBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+            org.tensorflow.proto.util.testlog.MachineConfiguration, org.tensorflow.proto.util.testlog.MachineConfiguration.Builder, org.tensorflow.proto.util.testlog.MachineConfigurationOrBuilder>(
+                getMachineConfiguration(),
+                getParentForChildren(),
+                isClean());
+        machineConfiguration_ = null;
+      }
+      return machineConfigurationBuilder_;
+    }
+
+    private org.tensorflow.proto.util.testlog.RunConfiguration runConfiguration_;
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.util.testlog.RunConfiguration, org.tensorflow.proto.util.testlog.RunConfiguration.Builder, org.tensorflow.proto.util.testlog.RunConfigurationOrBuilder> runConfigurationBuilder_;
+    /**
+     * <pre>
+     * Run-specific parameters (arguments, etc)
+     * </pre>
+     *
+     * <code>.tensorflow.RunConfiguration run_configuration = 8;</code>
+     */
+    public boolean hasRunConfiguration() {
+      return runConfigurationBuilder_ != null || runConfiguration_ != null;
+    }
+    /**
+     * <pre>
+     * Run-specific parameters (arguments, etc)
+     * </pre>
+     *
+     * <code>.tensorflow.RunConfiguration run_configuration = 8;</code>
+     */
+    public org.tensorflow.proto.util.testlog.RunConfiguration getRunConfiguration() {
+      if (runConfigurationBuilder_ == null) {
+        return runConfiguration_ == null ? org.tensorflow.proto.util.testlog.RunConfiguration.getDefaultInstance() : runConfiguration_;
+      } else {
+        return runConfigurationBuilder_.getMessage();
+      }
+    }
+    /**
+     * <pre>
+     * Run-specific parameters (arguments, etc)
+     * </pre>
+     *
+     * <code>.tensorflow.RunConfiguration run_configuration = 8;</code>
+     */
+    public Builder setRunConfiguration(org.tensorflow.proto.util.testlog.RunConfiguration value) {
+      if (runConfigurationBuilder_ == null) {
+        if (value == null) {
+          throw new NullPointerException();
+        }
+        runConfiguration_ = value;
+        onChanged();
+      } else {
+        runConfigurationBuilder_.setMessage(value);
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Run-specific parameters (arguments, etc)
+     * </pre>
+     *
+     * <code>.tensorflow.RunConfiguration run_configuration = 8;</code>
+     */
+    public Builder setRunConfiguration(
+        org.tensorflow.proto.util.testlog.RunConfiguration.Builder builderForValue) {
+      if (runConfigurationBuilder_ == null) {
+        runConfiguration_ = builderForValue.build();
+        onChanged();
+      } else {
+        runConfigurationBuilder_.setMessage(builderForValue.build());
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Run-specific parameters (arguments, etc)
+     * </pre>
+     *
+     * <code>.tensorflow.RunConfiguration run_configuration = 8;</code>
+     */
+    public Builder mergeRunConfiguration(org.tensorflow.proto.util.testlog.RunConfiguration value) {
+      if (runConfigurationBuilder_ == null) {
+        if (runConfiguration_ != null) {
+          runConfiguration_ =
+            org.tensorflow.proto.util.testlog.RunConfiguration.newBuilder(runConfiguration_).mergeFrom(value).buildPartial();
+        } else {
+          runConfiguration_ = value;
+        }
+        onChanged();
+      } else {
+        runConfigurationBuilder_.mergeFrom(value);
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Run-specific parameters (arguments, etc)
+     * </pre>
+     *
+     * <code>.tensorflow.RunConfiguration run_configuration = 8;</code>
+     */
+    public Builder clearRunConfiguration() {
+      if (runConfigurationBuilder_ == null) {
+        runConfiguration_ = null;
+        onChanged();
+      } else {
+        runConfiguration_ = null;
+        runConfigurationBuilder_ = null;
+      }
+
+      return this;
+    }
+    /**
+     * <pre>
+     * Run-specific parameters (arguments, etc)
+     * </pre>
+     *
+     * <code>.tensorflow.RunConfiguration run_configuration = 8;</code>
+     */
+    public org.tensorflow.proto.util.testlog.RunConfiguration.Builder getRunConfigurationBuilder() {
+      
+      onChanged();
+      return getRunConfigurationFieldBuilder().getBuilder();
+    }
+    /**
+     * <pre>
+     * Run-specific parameters (arguments, etc)
+     * </pre>
+     *
+     * <code>.tensorflow.RunConfiguration run_configuration = 8;</code>
+     */
+    public org.tensorflow.proto.util.testlog.RunConfigurationOrBuilder getRunConfigurationOrBuilder() {
+      if (runConfigurationBuilder_ != null) {
+        return runConfigurationBuilder_.getMessageOrBuilder();
+      } else {
+        return runConfiguration_ == null ?
+            org.tensorflow.proto.util.testlog.RunConfiguration.getDefaultInstance() : runConfiguration_;
+      }
+    }
+    /**
+     * <pre>
+     * Run-specific parameters (arguments, etc)
+     * </pre>
+     *
+     * <code>.tensorflow.RunConfiguration run_configuration = 8;</code>
+     */
+    private com.google.protobuf.SingleFieldBuilderV3<
+        org.tensorflow.proto.util.testlog.RunConfiguration, org.tensorflow.proto.util.testlog.RunConfiguration.Builder, org.tensorflow.proto.util.testlog.RunConfigurationOrBuilder> 
+        getRunConfigurationFieldBuilder() {
+      if (runConfigurationBuilder_ == null) {
+        runConfigurationBuilder_ = new com.google.protobuf.SingleFieldBuilderV3<
+            org.tensorflow.proto.util.testlog.RunConfiguration, org.tensorflow.proto.util.testlog.RunConfiguration.Builder, org.tensorflow.proto.util.testlog.RunConfigurationOrBuilder>(
+                getRunConfiguration(),
+                getParentForChildren(),
+                isClean());
+        runConfiguration_ = null;
+      }
+      return runConfigurationBuilder_;
+    }
+
+    private java.lang.Object name_ = "";
+    /**
+     * <pre>
+     * Benchmark target identifier.
+     * </pre>
+     *
+     * <code>string name = 9;</code>
+     */
+    public java.lang.String getName() {
+      java.lang.Object ref = name_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        name_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Benchmark target identifier.
+     * </pre>
+     *
+     * <code>string name = 9;</code>
+     */
+    public com.google.protobuf.ByteString
+        getNameBytes() {
+      java.lang.Object ref = name_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        name_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Benchmark target identifier.
+     * </pre>
+     *
+     * <code>string name = 9;</code>
+     */
+    public Builder setName(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      name_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Benchmark target identifier.
+     * </pre>
+     *
+     * <code>string name = 9;</code>
+     */
+    public Builder clearName() {
+      
+      name_ = getDefaultInstance().getName();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Benchmark target identifier.
+     * </pre>
+     *
+     * <code>string name = 9;</code>
+     */
+    public Builder setNameBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      name_ = value;
+      onChanged();
+      return this;
+    }
+
+    private int benchmarkType_ = 0;
+    /**
+     * <code>.tensorflow.TestResults.BenchmarkType benchmark_type = 10;</code>
+     */
+    public int getBenchmarkTypeValue() {
+      return benchmarkType_;
+    }
+    /**
+     * <code>.tensorflow.TestResults.BenchmarkType benchmark_type = 10;</code>
+     */
+    public Builder setBenchmarkTypeValue(int value) {
+      benchmarkType_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>.tensorflow.TestResults.BenchmarkType benchmark_type = 10;</code>
+     */
+    public org.tensorflow.proto.util.testlog.TestResults.BenchmarkType getBenchmarkType() {
+      @SuppressWarnings("deprecation")
+      org.tensorflow.proto.util.testlog.TestResults.BenchmarkType result = org.tensorflow.proto.util.testlog.TestResults.BenchmarkType.valueOf(benchmarkType_);
+      return result == null ? org.tensorflow.proto.util.testlog.TestResults.BenchmarkType.UNRECOGNIZED : result;
+    }
+    /**
+     * <code>.tensorflow.TestResults.BenchmarkType benchmark_type = 10;</code>
+     */
+    public Builder setBenchmarkType(org.tensorflow.proto.util.testlog.TestResults.BenchmarkType value) {
+      if (value == null) {
+        throw new NullPointerException();
+      }
+      
+      benchmarkType_ = value.getNumber();
+      onChanged();
+      return this;
+    }
+    /**
+     * <code>.tensorflow.TestResults.BenchmarkType benchmark_type = 10;</code>
+     */
+    public Builder clearBenchmarkType() {
+      
+      benchmarkType_ = 0;
+      onChanged();
+      return this;
+    }
+
+    private java.lang.Object runMode_ = "";
+    /**
+     * <pre>
+     * Used for differentiating between continuous and debug builds.
+     * Must be one of:
+     * * cbuild: results from continuous build.
+     * * presubmit: results from oneshot requests.
+     * * culprit: results from culprit finder rerun.
+     * </pre>
+     *
+     * <code>string run_mode = 11;</code>
+     */
+    public java.lang.String getRunMode() {
+      java.lang.Object ref = runMode_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        runMode_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Used for differentiating between continuous and debug builds.
+     * Must be one of:
+     * * cbuild: results from continuous build.
+     * * presubmit: results from oneshot requests.
+     * * culprit: results from culprit finder rerun.
+     * </pre>
+     *
+     * <code>string run_mode = 11;</code>
+     */
+    public com.google.protobuf.ByteString
+        getRunModeBytes() {
+      java.lang.Object ref = runMode_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        runMode_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * Used for differentiating between continuous and debug builds.
+     * Must be one of:
+     * * cbuild: results from continuous build.
+     * * presubmit: results from oneshot requests.
+     * * culprit: results from culprit finder rerun.
+     * </pre>
+     *
+     * <code>string run_mode = 11;</code>
+     */
+    public Builder setRunMode(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      runMode_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Used for differentiating between continuous and debug builds.
+     * Must be one of:
+     * * cbuild: results from continuous build.
+     * * presubmit: results from oneshot requests.
+     * * culprit: results from culprit finder rerun.
+     * </pre>
+     *
+     * <code>string run_mode = 11;</code>
+     */
+    public Builder clearRunMode() {
+      
+      runMode_ = getDefaultInstance().getRunMode();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * Used for differentiating between continuous and debug builds.
+     * Must be one of:
+     * * cbuild: results from continuous build.
+     * * presubmit: results from oneshot requests.
+     * * culprit: results from culprit finder rerun.
+     * </pre>
+     *
+     * <code>string run_mode = 11;</code>
+     */
+    public Builder setRunModeBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      runMode_ = value;
+      onChanged();
+      return this;
+    }
+
+    private java.lang.Object tfVersion_ = "";
+    /**
+     * <pre>
+     * TensorFlow version this benchmark runs against.
+     * This can be either set to full version or just the major version.
+     * </pre>
+     *
+     * <code>string tf_version = 12;</code>
+     */
+    public java.lang.String getTfVersion() {
+      java.lang.Object ref = tfVersion_;
+      if (!(ref instanceof java.lang.String)) {
+        com.google.protobuf.ByteString bs =
+            (com.google.protobuf.ByteString) ref;
+        java.lang.String s = bs.toStringUtf8();
+        tfVersion_ = s;
+        return s;
+      } else {
+        return (java.lang.String) ref;
+      }
+    }
+    /**
+     * <pre>
+     * TensorFlow version this benchmark runs against.
+     * This can be either set to full version or just the major version.
+     * </pre>
+     *
+     * <code>string tf_version = 12;</code>
+     */
+    public com.google.protobuf.ByteString
+        getTfVersionBytes() {
+      java.lang.Object ref = tfVersion_;
+      if (ref instanceof String) {
+        com.google.protobuf.ByteString b = 
+            com.google.protobuf.ByteString.copyFromUtf8(
+                (java.lang.String) ref);
+        tfVersion_ = b;
+        return b;
+      } else {
+        return (com.google.protobuf.ByteString) ref;
+      }
+    }
+    /**
+     * <pre>
+     * TensorFlow version this benchmark runs against.
+     * This can be either set to full version or just the major version.
+     * </pre>
+     *
+     * <code>string tf_version = 12;</code>
+     */
+    public Builder setTfVersion(
+        java.lang.String value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  
+      tfVersion_ = value;
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * TensorFlow version this benchmark runs against.
+     * This can be either set to full version or just the major version.
+     * </pre>
+     *
+     * <code>string tf_version = 12;</code>
+     */
+    public Builder clearTfVersion() {
+      
+      tfVersion_ = getDefaultInstance().getTfVersion();
+      onChanged();
+      return this;
+    }
+    /**
+     * <pre>
+     * TensorFlow version this benchmark runs against.
+     * This can be either set to full version or just the major version.
+     * </pre>
+     *
+     * <code>string tf_version = 12;</code>
+     */
+    public Builder setTfVersionBytes(
+        com.google.protobuf.ByteString value) {
+      if (value == null) {
+    throw new NullPointerException();
+  }
+  checkByteStringIsUtf8(value);
+      
+      tfVersion_ = value;
+      onChanged();
+      return this;
+    }
+    @java.lang.Override
+    public final Builder setUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.setUnknownFields(unknownFields);
+    }
+
+    @java.lang.Override
+    public final Builder mergeUnknownFields(
+        final com.google.protobuf.UnknownFieldSet unknownFields) {
+      return super.mergeUnknownFields(unknownFields);
+    }
+
+
+    // @@protoc_insertion_point(builder_scope:tensorflow.TestResults)
+  }
+
+  // @@protoc_insertion_point(class_scope:tensorflow.TestResults)
+  private static final org.tensorflow.proto.util.testlog.TestResults DEFAULT_INSTANCE;
+  static {
+    DEFAULT_INSTANCE = new org.tensorflow.proto.util.testlog.TestResults();
+  }
+
+  public static org.tensorflow.proto.util.testlog.TestResults getDefaultInstance() {
+    return DEFAULT_INSTANCE;
+  }
+
+  private static final com.google.protobuf.Parser<TestResults>
+      PARSER = new com.google.protobuf.AbstractParser<TestResults>() {
+    @java.lang.Override
+    public TestResults parsePartialFrom(
+        com.google.protobuf.CodedInputStream input,
+        com.google.protobuf.ExtensionRegistryLite extensionRegistry)
+        throws com.google.protobuf.InvalidProtocolBufferException {
+      return new TestResults(input, extensionRegistry);
+    }
+  };
+
+  public static com.google.protobuf.Parser<TestResults> parser() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public com.google.protobuf.Parser<TestResults> getParserForType() {
+    return PARSER;
+  }
+
+  @java.lang.Override
+  public org.tensorflow.proto.util.testlog.TestResults getDefaultInstanceForType() {
+    return DEFAULT_INSTANCE;
+  }
+
+}
+
diff --git a/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/TestResultsOrBuilder.java b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/TestResultsOrBuilder.java
new file mode 100644
index 00000000000..28ef5b7bfd0
--- /dev/null
+++ b/tensorflow-core/tensorflow-core-api/src/gen/java/org/tensorflow/proto/util/testlog/TestResultsOrBuilder.java
@@ -0,0 +1,245 @@
+// Generated by the protocol buffer compiler.  DO NOT EDIT!
+// source: tensorflow/core/util/test_log.proto
+
+package org.tensorflow.proto.util.testlog;
+
+public interface TestResultsOrBuilder extends
+    // @@protoc_insertion_point(interface_extends:tensorflow.TestResults)
+    com.google.protobuf.MessageOrBuilder {
+
+  /**
+   * <pre>
+   * The target of the run, e.g.:
+   *  //tensorflow/core:kernels_adjust_contrast_op_benchmark_test
+   * </pre>
+   *
+   * <code>string target = 1;</code>
+   */
+  java.lang.String getTarget();
+  /**
+   * <pre>
+   * The target of the run, e.g.:
+   *  //tensorflow/core:kernels_adjust_contrast_op_benchmark_test
+   * </pre>
+   *
+   * <code>string target = 1;</code>
+   */
+  com.google.protobuf.ByteString
+      getTargetBytes();
+
+  /**
+   * <pre>
+   * The list of tests or benchmarks in this run.
+   * </pre>
+   *
+   * <code>.tensorflow.BenchmarkEntries entries = 2;</code>
+   */
+  boolean hasEntries();
+  /**
+   * <pre>
+   * The list of tests or benchmarks in this run.
+   * </pre>
+   *
+   * <code>.tensorflow.BenchmarkEntries entries = 2;</code>
+   */
+  org.tensorflow.proto.util.testlog.BenchmarkEntries getEntries();
+  /**
+   * <pre>
+   * The list of tests or benchmarks in this run.
+   * </pre>
+   *
+   * <code>.tensorflow.BenchmarkEntries entries = 2;</code>
+   */
+  org.tensorflow.proto.util.testlog.BenchmarkEntriesOrBuilder getEntriesOrBuilder();
+
+  /**
+   * <pre>
+   * The configuration of the build (compiled opt? with cuda? any copts?)
+   * </pre>
+   *
+   * <code>.tensorflow.BuildConfiguration build_configuration = 3;</code>
+   */
+  boolean hasBuildConfiguration();
+  /**
+   * <pre>
+   * The configuration of the build (compiled opt? with cuda? any copts?)
+   * </pre>
+   *
+   * <code>.tensorflow.BuildConfiguration build_configuration = 3;</code>
+   */
+  org.tensorflow.proto.util.testlog.BuildConfiguration getBuildConfiguration();
+  /**
+   * <pre>
+   * The configuration of the build (compiled opt? with cuda? any copts?)
+   * </pre>
+   *
+   * <code>.tensorflow.BuildConfiguration build_configuration = 3;</code>
+   */
+  org.tensorflow.proto.util.testlog.BuildConfigurationOrBuilder getBuildConfigurationOrBuilder();
+
+  /**
+   * <pre>
+   * The commit id (git hash or changelist)
+   * </pre>
+   *
+   * <code>.tensorflow.CommitId commit_id = 4;</code>
+   */
+  boolean hasCommitId();
+  /**
+   * <pre>
+   * The commit id (git hash or changelist)
+   * </pre>
+   *
+   * <code>.tensorflow.CommitId commit_id = 4;</code>
+   */
+  org.tensorflow.proto.util.testlog.CommitId getCommitId();
+  /**
+   * <pre>
+   * The commit id (git hash or changelist)
+   * </pre>
+   *
+   * <code>.tensorflow.CommitId commit_id = 4;</code>
+   */
+  org.tensorflow.proto.util.testlog.CommitIdOrBuilder getCommitIdOrBuilder();
+
+  /**
+   * <pre>
+   * The time the run started (in seconds of UTC time since Unix epoch)
+   * </pre>
+   *
+   * <code>int64 start_time = 5;</code>
+   */
+  long getStartTime();
+
+  /**
+   * <pre>
+   * The amount of time the total run took (wall time in seconds)
+   * </pre>
+   *
+   * <code>double run_time = 6;</code>
+   */
+  double getRunTime();
+
+  /**
+   * <pre>
+   * Machine-specific parameters (Platform and CPU info)
+   * </pre>
+   *
+   * <code>.tensorflow.MachineConfiguration machine_configuration = 7;</code>
+   */
+  boolean hasMachineConfiguration();
+  /**
+   * <pre>
+   * Machine-specific parameters (Platform and CPU info)
+   * </pre>
+   *
+   * <code>.tensorflow.MachineConfiguration machine_configuration = 7;</code>
+   */
+  org.tensorflow.proto.util.testlog.MachineConfiguration getMachineConfiguration();
+  /**
+   * <pre>
+   * Machine-specific parameters (Platform and CPU info)
+   * </pre>
+   *
+   * <code>.tensorflow.MachineConfiguration machine_configuration = 7;</code>
+   */
+  org.tensorflow.proto.util.testlog.MachineConfigurationOrBuilder getMachineConfigurationOrBuilder();
+
+  /**
+   * <pre>
+   * Run-specific parameters (arguments, etc)
+   * </pre>
+   *
+   * <code>.tensorflow.RunConfiguration run_configuration = 8;</code>
+   */
+  boolean hasRunConfiguration();
+  /**
+   * <pre>
+   * Run-specific parameters (arguments, etc)
+   * </pre>
+   *
+   * <code>.tensorflow.RunConfiguration run_configuration = 8;</code>
+   */
+  org.tensorflow.proto.util.testlog.RunConfiguration getRunConfiguration();
+  /**
+   * <pre>
+   * Run-specific parameters (arguments, etc)
+   * </pre>
+   *
+   * <code>.tensorflow.RunConfiguration run_configuration = 8;</code>
+   */
+  org.tensorflow.proto.util.testlog.RunConfigurationOrBuilder getRunConfigurationOrBuilder();
+
+  /**
+   * <pre>
+   * Benchmark target identifier.
+   * </pre>
+   *
+   * <code>string name = 9;</code>
+   */
+  java.lang.String getName();
+  /**
+   * <pre>
+   * Benchmark target identifier.
+   * </pre>
+   *
+   * <code>string name = 9;</code>
+   */
+  com.google.protobuf.ByteString
+      getNameBytes();
+
+  /**
+   * <code>.tensorflow.TestResults.BenchmarkType benchmark_type = 10;</code>
+   */
+  int getBenchmarkTypeValue();
+  /**
+   * <code>.tensorflow.TestResults.BenchmarkType benchmark_type = 10;</code>
+   */
+  org.tensorflow.proto.util.testlog.TestResults.BenchmarkType getBenchmarkType();
+
+  /**
+   * <pre>
+   * Used for differentiating between continuous and debug builds.
+   * Must be one of:
+   * * cbuild: results from continuous build.
+   * * presubmit: results from oneshot requests.
+   * * culprit: results from culprit finder rerun.
+   * </pre>
+   *
+   * <code>string run_mode = 11;</code>
+   */
+  java.lang.String getRunMode();
+  /**
+   * <pre>
+   * Used for differentiating between continuous and debug builds.
+   * Must be one of:
+   * * cbuild: results from continuous build.
+   * * presubmit: results from oneshot requests.
+   * * culprit: results from culprit finder rerun.
+   * </pre>
+   *
+   * <code>string run_mode = 11;</code>
+   */
+  com.google.protobuf.ByteString
+      getRunModeBytes();
+
+  /**
+   * <pre>
+   * TensorFlow version this benchmark runs against.
+   * This can be either set to full version or just the major version.
+   * </pre>
+   *
+   * <code>string tf_version = 12;</code>
+   */
+  java.lang.String getTfVersion();
+  /**
+   * <pre>
+   * TensorFlow version this benchmark runs against.
+   * This can be either set to full version or just the major version.
+   * </pre>
+   *
+   * <code>string tf_version = 12;</code>
+   */
+  com.google.protobuf.ByteString
+      getTfVersionBytes();
+}

From 4740bd44972a79b61163853df1325657880477e4 Mon Sep 17 00:00:00 2001
From: klessard <klessard@expedia.com>
Date: Wed, 18 Nov 2020 15:52:32 -0500
Subject: [PATCH 11/11] Rebase from master

---
 .../annotations/org/tensorflow/op/NnOps.java  | 178 ++++++++++++++----
 .../org/tensorflow/types/TStringTest.java     |   9 +-
 .../tensorflow/framework/activations/ELU.java |   3 +-
 .../framework/activations/HardSigmoid.java    |   3 +-
 .../framework/activations/ReLU.java           |   3 +-
 .../framework/losses/BinaryCrossentropy.java  |   4 +-
 .../losses/CategoricalCrossentropy.java       |   4 +-
 .../tensorflow/framework/losses/Hinge.java    |   6 +-
 .../tensorflow/framework/losses/Losses.java   |  43 +++--
 .../losses/SparseCategoricalCrossentropy.java |   4 +-
 .../framework/losses/SquaredHinge.java        |   6 +-
 .../framework/losses/impl/LossesHelper.java   |  11 +-
 .../framework/utils/CastHelper.java           |   5 +-
 .../framework/activations/ReLUTest.java       |   4 +-
 .../framework/utils/EagerTestSession.java     |   2 +-
 .../framework/utils/GraphTestSession.java     |  51 +++--
 16 files changed, 220 insertions(+), 116 deletions(-)

diff --git a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/NnOps.java b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/NnOps.java
index d6984bb1a95..0f523cf9dfb 100644
--- a/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/NnOps.java
+++ b/tensorflow-core/tensorflow-core-api/src/gen/annotations/org/tensorflow/op/NnOps.java
@@ -18,7 +18,6 @@
 package org.tensorflow.op;
 
 import java.util.List;
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.op.nn.AvgPool;
 import org.tensorflow.op.nn.AvgPool3d;
@@ -84,6 +83,7 @@
 import org.tensorflow.op.nn.Relu;
 import org.tensorflow.op.nn.Relu6;
 import org.tensorflow.op.nn.Selu;
+import org.tensorflow.op.nn.SigmoidCrossEntropyWithLogits;
 import org.tensorflow.op.nn.Softmax;
 import org.tensorflow.op.nn.SoftmaxCrossEntropyWithLogits;
 import org.tensorflow.op.nn.Softsign;
@@ -103,10 +103,13 @@
  * @see {@link Ops}
  */
 public final class NnOps {
+  public final NnRawOps raw;
+
   private final Scope scope;
 
   NnOps(Scope scope) {
     this.scope = scope;
+    raw = new NnRawOps(scope);
   }
 
   /**
@@ -1342,7 +1345,7 @@ public <T extends TNumber> MaxPool3d<T> maxPool3d(Operand<T> input, List<Long> k
   }
 
   /**
-   * Computes gradients of max pooling function.
+   * Computes gradients of 3D max pooling function.
    *
    * @param <U> data type for {@code output()} output
    * @param origInput The original input tensor.
@@ -1767,6 +1770,56 @@ public <T extends TNumber> Selu<T> selu(Operand<T> features) {
     return Selu.create(scope, features);
   }
 
+  /**
+   * Computes sigmoid cross entropy given <code>logits</code>.
+   *
+   *  <p>Measures the probability error in discrete classification tasks in which each class is
+   *  independent and not mutually exclusive. For instance, one could perform multilabel
+   *  classification where a picture can contain both an elephant and a dog at the same time.
+   *
+   *  <p>For brevity, let <code>x = logits</code>, <code>z = labels</code>. The logistic loss in
+   *  pseudo-code is
+   *
+   *  <pre>
+   *  z * -log(sigmoid(x)) + (1 - z) * -log(1 - sigmoid(x))
+   *   = z * -log(1 / (1 + exp(-x))) + (1 - z) * -log(exp(-x) / (1 + exp(-x)))
+   *   = z * log(1 + exp(-x)) + (1 - z) * (-log(exp(-x)) + log(1 + exp(-x)))
+   *   = z * log(1 + exp(-x)) + (1 - z) * (x + log(1 + exp(-x))
+   *   = (1 - z) * x + log(1 + exp(-x))
+   *   = x - x * z + log(1 + exp(-x))
+   *  </pre>
+   *
+   *  <p>For <code>x < 0</code>, to avoid overflow in <code>exp(-x)</code>, we reformulate the above
+   *
+   *  <pre>
+   *  x - x * z + log(1 + exp(-x))
+   *   = log(exp(x)) - x * z + log(1 + exp(-x))
+   *   = - x * z + log(1 + exp(x))
+   *  </pre>
+   *
+   *  <p>Hence, to ensure stability and avoid overflow, the implementation uses this equivalent
+   *  formulation
+   *
+   *  <pre>
+   *    max(x, 0) - x * z + log(1 + exp(-abs(x)))
+   *  </pre>
+   *
+   *  <p></ode>logits</code> and <code>labels</code> must have the same type and shape.
+   *
+   *  <p>
+   *
+   * @param scope The TensorFlow scope
+   * @param labels the labels
+   * @param logits the logits of type float32 or float64
+   * @param <T> the type of labels and logits
+   * @return the component-wise logistic losses.
+   * @throws IllegalArgumentException if logits' and labels' do not have the same shape
+   */
+  public <T extends TNumber> Operand<T> sigmoidCrossEntropyWithLogits(Operand<T> labels,
+      Operand<T> logits) {
+    return SigmoidCrossEntropyWithLogits.sigmoidCrossEntropyWithLogits(scope, labels, logits);
+  }
+
   /**
    * Computes softmax activations.
    *  <p>
@@ -1783,20 +1836,54 @@ public <T extends TNumber> Softmax<T> softmax(Operand<T> logits) {
   }
 
   /**
-   * Computes softmax cross entropy cost and gradients to backpropagate.
-   *  <p>
-   *  Inputs are the logits, not probabilities.
+   * Computes softmax cross entropy between <code>logits</code> and <code>labels</code>.
    *
-   * @param <T> data type for {@code loss()} output
-   * @param features batch_size x num_classes matrix
-   * @param labels batch_size x num_classes matrix
-   *  The caller must ensure that each batch of labels represents a valid
-   *  probability distribution.
-   * @return a new instance of SoftmaxCrossEntropyWithLogits
+   *  <p>Measures the probability error in discrete classification tasks in which the classes are
+   *  mutually exclusive (each entry is in exactly one class). For example, each CIFAR-10 image is
+   *  labeled with one and only one label: an image can be a dog or a truck, but not both.
+   *
+   *  <p><b>NOTE:</b>
+   *
+   *  <p>While the classes are mutually exclusive, their probabilities need not be. All that is
+   *  required is that each row of <code>labels</code> is a valid probability distribution. If they
+   *  are not, the computation of the gradient will be incorrect.
+   *
+   *  <p>If using exclusive <code>labels</code> (wherein one and only one class is true at a time),
+   *  see {@link org.tensorflow.op.NnOps#sparseSoftmaxCrossEntropyWithLogits}
+   *
+   *  <p>Usage:
+   *
+   *  <pre>
+   *    Operand&lt;TFloat32&gt; logits =
+   *        tf.constant(new float[][] {{4.0F, 2.0F, 1.0F}, {0.0F, 5.0F, 1.0F}} );
+   *    Operand&lt;TFloat32&gt; labels =
+   *        tf.constant(new float[][] {{1.0F, 0.0F, 0.0F}, {0.0F, 0.8F, 0.2F}} );
+   *    Operand&lt;TFloat32&gt; output =
+   *        tf.nn.softmaxCrossEntropyWithLogits(labels, logits, -1);
+   *    // output Shape = [2]
+   *    // dataType = FLOAT (1)
+   *    // values { 0.169846, 0.824745 }
+   *  </pre>
+   *
+   *  <p>Backpropagation will happen into both <code>logits</code> and <code>labels</code>. To
+   *  disallow backpropagation into <code>labels</code>, pass label tensors through <code>
+   *  tf.stopGradient</code> before feeding it to this function.
+   *
+   * @param scope current scope
+   * @param labels Each vector along the class dimension should hold a valid probability
+   *      distribution e.g. for the case in which labels are of shape <code>[batch_size, num_classes]
+   *      </code>, each row of <code>labels[i]</code> must be a valid probability distribution.
+   * @param logits Per-label activations, typically a linear output. These activation energies are
+   *      interpreted as unnormalized log probabilities.
+   * @param axis The class dimension. -1 is the last dimension.
+   * @param <T> the number type of the operands
+   * @return the softmax cross entropy loss. Its type is the same as <code>logits</code> and its
+   *      shape is the same as <code>labels</code> except that it does not have the last dimension of
+   *      <code>labels</code>.
    */
-  public <T extends TNumber> SoftmaxCrossEntropyWithLogits<T> softmaxCrossEntropyWithLogits(
-      Operand<T> features, Operand<T> labels) {
-    return SoftmaxCrossEntropyWithLogits.create(scope, features, labels);
+  public <T extends TNumber, U extends TNumber> Operand<T> softmaxCrossEntropyWithLogits(
+      Operand<U> labels, Operand<T> logits, int axis) {
+    return SoftmaxCrossEntropyWithLogits.softmaxCrossEntropyWithLogits(scope, labels, logits, axis);
   }
 
   /**
@@ -1988,24 +2075,51 @@ public <T extends TType> SpaceToDepth<T> spaceToDepth(Operand<T> input, Long blo
   }
 
   /**
-   * Computes softmax cross entropy cost and gradients to backpropagate.
-   *  <p>
-   *  Unlike `SoftmaxCrossEntropyWithLogits`, this operation does not accept
-   *  a matrix of label probabilities, but rather a single label per row
-   *  of features.  This label is considered to have probability 1.0 for the
-   *  given row.
-   *  <p>
-   *  Inputs are the logits, not probabilities.
-   *
-   * @param <T> data type for {@code loss()} output
-   * @param features batch_size x num_classes matrix
-   * @param labels batch_size vector with values in [0, num_classes).
-   *  This is the label for the given minibatch entry.
-   * @return a new instance of SparseSoftmaxCrossEntropyWithLogits
-   */
-  public <T extends TNumber, U extends TNumber> SparseSoftmaxCrossEntropyWithLogits<T> sparseSoftmaxCrossEntropyWithLogits(
-      Operand<T> features, Operand<U> labels) {
-    return SparseSoftmaxCrossEntropyWithLogits.create(scope, features, labels);
+   * Computes sparse softmax cross entropy between <code>logits</code> and <code>labels</code>.
+   *
+   *  <p>Measures the probability error in discrete classification tasks in which the classes are
+   *  mutually exclusive (each entry is in exactly one class). For example, each CIFAR-10 image is
+   *  labeled with one and only one label: an image can be a dog or a truck, but not both.
+   *
+   *  <p><b>NOTE:</b>
+   *
+   *  <p>For this operation, the probability of a given label is considered exclusive. That is, soft
+   *  classes are not allowed, and the <code>labels</code> vector must provide a single specific
+   *  index for the true class for each row of <code>logits</code> (each minibatch entry). For soft
+   *  softmax classification with a probability distribution for each entry, {@link
+   *  org.tensorflow.op.NnOps#softmaxCrossEntropyWithLogits}.
+   *
+   *  <p><b>WARNING:</b>
+   *
+   *  <p>This op expects unscaled logits, since it performs a <code>softmax</code> on <code>logits
+   *  </code> internally for efficiency. Do not call this op with the output of <code>softmax</code>,
+   *  as it will produce incorrect results.
+   *
+   *  <p>A common use case is to have logits of shape <code>[batchSize, numClasses]</code> and have
+   *  labels of shape <code>[batchSize]</code>, but higher dimensions are supported, in which case
+   *  the <code>dim</code>-th dimension is assumed to be of size <code>numClasses</code>. <code>
+   *  logits</code> must have the <cod>dataType</cod> of <code>TFloat16</code>, <code>TFloat32</code>
+   *  , or <code>TFloat64</code>, and <code>labels</code> must have the dtype of <code>TInt32</code>
+   *  or <code>TInt64</code>.
+   *
+   * @param scope current scope
+   * @param labels <code>Tensor</code> of shape <code>[d_0, d_1, ..., d_{r-1}]</code> (where <code>r
+   *      </code> is rank of <code>labels</code> and result) and the dataType is <code>TInt32</code>
+   *      or <code>TInt64</code>. Each entry in <code>labels</code> must be an index in <code>[0,
+   *      numClasses)</code>. Other values will raise an exception when this op is run on CPU, and
+   *      return <code>NaN</code> for corresponding loss and gradient rows on GPU.
+   * @param logits Per-label activations (typically a linear output) of shape <code>[d_0, d_1, ...,
+   *      d_{r-1}, numClasses]</code> and dataType of <code>TFloat16</code>, <code>TFloat32</code>,
+   *      or <code>TFloat64</code>. These activation energies are interpreted as unnormalized log
+   *      probabilities.
+   * @return A <code>Tensor</code> of the same shape as <code>labels</code> and of the same type as
+   *      <code>logits</code> with the softmax cross entropy loss.
+   * @throws IllegalArgumentException If logits are scalars (need to have rank >= 1) or if the rank
+   *      of the labels is not equal to the rank of the logits minus one.
+   */
+  public <T extends TNumber, U extends TNumber> Operand sparseSoftmaxCrossEntropyWithLogits(
+      Operand<T> labels, Operand<U> logits) {
+    return SparseSoftmaxCrossEntropyWithLogits.sparseSoftmaxCrossEntropyWithLogits(scope, labels, logits);
   }
 
   /**
diff --git a/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/types/TStringTest.java b/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/types/TStringTest.java
index cb4260b1b4c..08048947648 100644
--- a/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/types/TStringTest.java
+++ b/tensorflow-core/tensorflow-core-api/src/test/java/org/tensorflow/types/TStringTest.java
@@ -39,13 +39,10 @@ public void createScalar() {
 
     @Test
     public void createrScalarLongerThan127() {
-        Tensor<TString> tensor = TString.scalarOf("Long String 1234567890123456789012345678901234567890123456789012345678901234567890123456789012345678901234567890123456789012345678901234567890 !");
+        TString tensor = TString.scalarOf("Long String 1234567890123456789012345678901234567890123456789012345678901234567890123456789012345678901234567890123456789012345678901234567890 !");
         assertNotNull(tensor);
-
-        TString data = tensor.data();
-        assertNotNull(data);
-        assertEquals(Shape.scalar(), data.shape());
-        assertEquals("Long String 1234567890123456789012345678901234567890123456789012345678901234567890123456789012345678901234567890123456789012345678901234567890 !", data.getObject());
+        assertEquals(Shape.scalar(), tensor.shape());
+        assertEquals("Long String 1234567890123456789012345678901234567890123456789012345678901234567890123456789012345678901234567890123456789012345678901234567890 !", tensor.getObject());
     }
 
 
diff --git a/tensorflow-framework/src/main/java/org/tensorflow/framework/activations/ELU.java b/tensorflow-framework/src/main/java/org/tensorflow/framework/activations/ELU.java
index ae3d7e8c896..d421bc1155d 100644
--- a/tensorflow-framework/src/main/java/org/tensorflow/framework/activations/ELU.java
+++ b/tensorflow-framework/src/main/java/org/tensorflow/framework/activations/ELU.java
@@ -14,7 +14,6 @@
 =======================================================================*/
 package org.tensorflow.framework.activations;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.op.Ops;
 import org.tensorflow.types.TBool;
@@ -89,7 +88,7 @@ public Operand<T> call(Operand<T> input) {
     Operand<T> result = tf.nn.elu(input);
     if (alpha == 1.0) return result;
     else {
-      DataType<T> dataType = input.asOutput().dataType();
+      Class<T> dataType = input.asOutput().type();
       Operand<T> y = tf.math.mul(result, tf.dtypes.cast(tf.constant(alpha), dataType));
       Operand<TBool> cond = tf.math.greater(result, tf.dtypes.cast(tf.constant(0), dataType));
       return tf.select(cond, result, y);
diff --git a/tensorflow-framework/src/main/java/org/tensorflow/framework/activations/HardSigmoid.java b/tensorflow-framework/src/main/java/org/tensorflow/framework/activations/HardSigmoid.java
index a486cbdc601..9e254a8fa34 100644
--- a/tensorflow-framework/src/main/java/org/tensorflow/framework/activations/HardSigmoid.java
+++ b/tensorflow-framework/src/main/java/org/tensorflow/framework/activations/HardSigmoid.java
@@ -14,7 +14,6 @@
 =======================================================================*/
 package org.tensorflow.framework.activations;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.op.Ops;
 import org.tensorflow.types.family.TFloating;
@@ -63,7 +62,7 @@ public HardSigmoid(Ops tf) {
    */
   @Override
   public Operand<T> call(Operand<T> input) {
-    DataType<T> dataType = input.asOutput().dataType();
+    Class<T> dataType = input.asOutput().type();
     Operand<T> point2 = tf.dtypes.cast(tf.constant(0.2), dataType);
     Operand<T> point5 = tf.dtypes.cast(tf.constant(0.5), dataType);
 
diff --git a/tensorflow-framework/src/main/java/org/tensorflow/framework/activations/ReLU.java b/tensorflow-framework/src/main/java/org/tensorflow/framework/activations/ReLU.java
index c24cf71077d..d405aa1bc7d 100644
--- a/tensorflow-framework/src/main/java/org/tensorflow/framework/activations/ReLU.java
+++ b/tensorflow-framework/src/main/java/org/tensorflow/framework/activations/ReLU.java
@@ -14,7 +14,6 @@
 =======================================================================*/
 package org.tensorflow.framework.activations;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.op.Ops;
 import org.tensorflow.op.math.Greater;
@@ -99,7 +98,7 @@ public ReLU(Ops tf, float alpha, float maxValue, float threshold) {
   @Override
   public Operand<T> call(Operand<T> input) {
 
-    DataType<T> dataType = input.asOutput().dataType();
+    Class<T> dataType = input.asOutput().type();
 
     boolean clipMax = !Float.isNaN(maxValue);
     Operand<T> negativePart = null;
diff --git a/tensorflow-framework/src/main/java/org/tensorflow/framework/losses/BinaryCrossentropy.java b/tensorflow-framework/src/main/java/org/tensorflow/framework/losses/BinaryCrossentropy.java
index effdf990f71..88ab71eb4be 100644
--- a/tensorflow-framework/src/main/java/org/tensorflow/framework/losses/BinaryCrossentropy.java
+++ b/tensorflow-framework/src/main/java/org/tensorflow/framework/losses/BinaryCrossentropy.java
@@ -217,8 +217,8 @@ public <T extends TNumber, U extends TNumber> Operand<T> call(
               getTF(),
               "predictions range check [0-1]",
               predictions,
-              cast(getTF(), getTF().constant(0), predictions.asOutput().dataType()),
-              cast(getTF(), getTF().constant(1), predictions.asOutput().dataType()));
+              cast(getTF(), getTF().constant(0), predictions.asOutput().type()),
+              cast(getTF(), getTF().constant(1), predictions.asOutput().type()));
 
     } else {
       lPredictions = predictions;
diff --git a/tensorflow-framework/src/main/java/org/tensorflow/framework/losses/CategoricalCrossentropy.java b/tensorflow-framework/src/main/java/org/tensorflow/framework/losses/CategoricalCrossentropy.java
index 7701ebfb806..d7bb4ab5128 100644
--- a/tensorflow-framework/src/main/java/org/tensorflow/framework/losses/CategoricalCrossentropy.java
+++ b/tensorflow-framework/src/main/java/org/tensorflow/framework/losses/CategoricalCrossentropy.java
@@ -256,8 +256,8 @@ public <T extends TNumber, U extends TNumber> Operand<T> call(
               getTF(),
               "predictions range check [0-1]",
               predictions,
-              cast(getTF(), getTF().constant(0), predictions.asOutput().dataType()),
-              cast(getTF(), getTF().constant(1), predictions.asOutput().dataType()));
+              cast(getTF(), getTF().constant(0), predictions.asOutput().type()),
+              cast(getTF(), getTF().constant(1), predictions.asOutput().type()));
 
     } else {
       lPredictions = predictions;
diff --git a/tensorflow-framework/src/main/java/org/tensorflow/framework/losses/Hinge.java b/tensorflow-framework/src/main/java/org/tensorflow/framework/losses/Hinge.java
index 5fdfd4c9b96..b5bf2d32355 100644
--- a/tensorflow-framework/src/main/java/org/tensorflow/framework/losses/Hinge.java
+++ b/tensorflow-framework/src/main/java/org/tensorflow/framework/losses/Hinge.java
@@ -124,15 +124,15 @@ public Hinge(Ops tf, String name, Reduction reduction) {
   public <T extends TNumber, U extends TNumber> Operand<T> call(
       Operand<U> labels, Operand<T> predictions, Operand<T> sampleWeights) {
     @SuppressWarnings("unchecked")
-    Operand<T> tLabels = predictions.asOutput().dataType() == labels.asOutput().dataType() ?
+    Operand<T> tLabels = predictions.asOutput().type() == labels.asOutput().type() ?
             (Operand<T>)labels :
-            cast(tf,  labels, predictions.asOutput().dataType());
+            cast(tf,  labels, predictions.asOutput().type());
     tLabels = LossesHelper.valueCheck(
             getTF(),
             "labels value check [-1, 0, 1]",
             tLabels,
             cast(getTF(), getTF().constant(new int[] { -1, 0, 1}),
-                    predictions.asOutput().dataType()));
+                    predictions.asOutput().type()));
 
     Operand<T> losses = Losses.hinge(getTF(), tLabels, predictions);
     return LossesHelper.computeWeightedLoss(getTF(), losses, getReduction(), sampleWeights);
diff --git a/tensorflow-framework/src/main/java/org/tensorflow/framework/losses/Losses.java b/tensorflow-framework/src/main/java/org/tensorflow/framework/losses/Losses.java
index 7a633ede2bf..1d5bbbf41e5 100644
--- a/tensorflow-framework/src/main/java/org/tensorflow/framework/losses/Losses.java
+++ b/tensorflow-framework/src/main/java/org/tensorflow/framework/losses/Losses.java
@@ -14,7 +14,6 @@
 =======================================================================*/
 package org.tensorflow.framework.losses;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.framework.losses.impl.LossTuple;
 import org.tensorflow.framework.losses.impl.LossesHelper;
@@ -51,7 +50,7 @@ public class Losses {
    */
   public static <T extends TNumber, U extends TNumber> Operand<T> meanAbsoluteError(
       Ops tf, Operand<U> labels, Operand<T> predictions) {
-    Operand<T> tLabels = cast(tf, labels, predictions.asOutput().dataType());
+    Operand<T> tLabels = cast(tf, labels, predictions.asOutput().type());
     LossTuple<T> ops = LossesHelper.squeezeOrExpandDimensions(tf, tLabels, predictions, null);
     predictions = ops.getTarget();
     tLabels = ops.getLabels();
@@ -73,7 +72,7 @@ public static <T extends TNumber, U extends TNumber> Operand<T> meanAbsoluteErro
    */
   public static <T extends TNumber, U extends TNumber> Operand<T> meanSquaredError(
       Ops tf, Operand<U> labels, Operand<T> predictions) {
-    Operand<T> tLabels = cast(tf, labels, predictions.asOutput().dataType());
+    Operand<T> tLabels = cast(tf, labels, predictions.asOutput().type());
     LossTuple<T> ops = LossesHelper.squeezeOrExpandDimensions(tf, tLabels, predictions, null);
     predictions = ops.getTarget();
     tLabels = ops.getLabels();
@@ -94,7 +93,7 @@ public static <T extends TNumber, U extends TNumber> Operand<T> meanSquaredError
    */
   public static <T extends TNumber, U extends TNumber> Operand<T> meanAbsolutePercentageError(
       Ops tf, Operand<U> labels, Operand<T> predictions) {
-    DataType<T> dataType = predictions.asOutput().dataType();
+    Class<T> dataType = predictions.asOutput().type();
     Operand<T> tLabels = cast(tf, labels, dataType);
     LossTuple<T> ops = LossesHelper.squeezeOrExpandDimensions(tf, tLabels, predictions, null);
     predictions = ops.getTarget();
@@ -121,7 +120,7 @@ public static <T extends TNumber, U extends TNumber> Operand<T> meanAbsolutePerc
    */
   public static <T extends TNumber, U extends TNumber> Operand<T> meanSquaredLogarithmicError(
       Ops tf, Operand<U> labels, Operand<T> predictions) {
-    DataType<T> dataType = predictions.asOutput().dataType();
+    Class<T> dataType = predictions.asOutput().type();
     Operand<T> tLabels = cast(tf, labels, dataType);
     LossTuple<T> ops = LossesHelper.squeezeOrExpandDimensions(tf, tLabels, predictions, null);
     predictions = ops.getTarget();
@@ -152,7 +151,7 @@ public static <T extends TNumber, U extends TNumber> Operand<T> meanSquaredLogar
    */
   public static <T extends TNumber, U extends TNumber> Operand<T> binaryCrossentropy(
       Ops tf, Operand<U> labels, Operand<T> predictions, boolean fromLogits, float labelSmoothing) {
-    DataType<T> dataType = predictions.asOutput().dataType();
+    Class<T> dataType = predictions.asOutput().type();
     Operand<T> tLabels = cast(tf, labels, dataType);
     LossTuple<T> ops = LossesHelper.squeezeOrExpandDimensions(tf, tLabels, predictions, null);
     predictions = ops.getTarget();
@@ -193,7 +192,7 @@ private static <T extends TNumber> Operand<T> binaryCrossentropyHelper(
     }
     */
 
-    DataType<T> dataType = output.asOutput().dataType();
+    Class<T> dataType = output.asOutput().type();
     Operand<T> one = cast(tf, tf.constant(1), dataType);
     Operand<T> epsilonConst = cast(tf, tf.constant(EPSILON), dataType);
     Operand<T> oneMinusEpsilonConst = tf.math.sub(one, epsilonConst);
@@ -231,7 +230,7 @@ public static <T extends TNumber, U extends TNumber> Operand<T> categoricalCross
       boolean fromLogits,
       float labelSmoothing,
       int axis) {
-    DataType<T> dataType = predictions.asOutput().dataType();
+    Class<T> dataType = predictions.asOutput().type();
     Operand<T> tLabels = cast(tf, labels, dataType);
     LossTuple<T> ops = LossesHelper.squeezeOrExpandDimensions(tf, tLabels, predictions, null);
     predictions = ops.getTarget();
@@ -284,7 +283,7 @@ public static <T extends TNumber, U extends TNumber> Operand<T> categoricalCross
    */
   public static <T extends TNumber, U extends TNumber> Operand<T> categoricalHinge(
       Ops tf, Operand<U> labels, Operand<T> predictions) {
-    DataType<T> dataType = predictions.asOutput().dataType();
+    Class<T> dataType = predictions.asOutput().type();
     Operand<T> tLabels = cast(tf, labels, dataType);
     LossTuple<T> lossTuple = LossesHelper.squeezeOrExpandDimensions(tf, tLabels, predictions, null);
     predictions = lossTuple.getTarget();
@@ -330,7 +329,7 @@ public static <T extends TNumber, U extends TNumber> Operand<T> categoricalHinge
    */
   public static <T extends TNumber, U extends TNumber> Operand<T> cosineSimilarity(
       Ops tf, Operand<U> labels, Operand<T> predictions, int axis) {
-    DataType<T> dataType = predictions.asOutput().dataType();
+    Class<T> dataType = predictions.asOutput().type();
     Operand<T> tLabels = cast(tf, labels, dataType);
     LossTuple<T> lossTuple = LossesHelper.squeezeOrExpandDimensions(tf, tLabels, predictions, null);
     predictions = lossTuple.getTarget();
@@ -357,7 +356,7 @@ public static <T extends TNumber, U extends TNumber> Operand<T> cosineSimilarity
    */
   public static <T extends TNumber, U extends TNumber> Operand<T> hinge(
       Ops tf, Operand<U> labels, Operand<T> predictions) {
-    DataType<T> dataType = predictions.asOutput().dataType();
+    Class<T> dataType = predictions.asOutput().type();
     Operand<T> tLabels = cast(tf, labels, dataType);
     LossTuple<T> lossTuple = LossesHelper.squeezeOrExpandDimensions(tf, tLabels, predictions, null);
     predictions = lossTuple.getTarget();
@@ -393,7 +392,7 @@ public static <T extends TNumber, U extends TNumber> Operand<T> hinge(
    */
   public static <T extends TNumber, U extends TNumber> Operand<T> huber(
       Ops tf, Operand<U> labels, Operand<T> predictions, float delta) {
-    DataType<T> dataType = predictions.asOutput().dataType();
+    Class<T> dataType = predictions.asOutput().type();
     Operand<T> tLabels = cast(tf, labels, dataType);
     LossTuple<T> lossTuple = LossesHelper.squeezeOrExpandDimensions(tf, tLabels, predictions, null);
     predictions = lossTuple.getTarget();
@@ -424,7 +423,7 @@ public static <T extends TNumber, U extends TNumber> Operand<T> huber(
    */
   public static <T extends TNumber, U extends TNumber> Operand<T> kullbackLeiblerDivergence(
       Ops tf, Operand<U> labels, Operand<T> predictions) {
-    DataType<T> dataType = predictions.asOutput().dataType();
+    Class<T> dataType = predictions.asOutput().type();
     Operand<T> tLabels = cast(tf, labels, dataType);
     LossTuple<T> lossTuple = LossesHelper.squeezeOrExpandDimensions(tf, tLabels, predictions, null);
     predictions = lossTuple.getTarget();
@@ -454,7 +453,7 @@ public static <T extends TNumber, U extends TNumber> Operand<T> kullbackLeiblerD
    */
   public static <T extends TNumber, U extends TNumber> Operand<T> logCosh(
       Ops tf, Operand<U> labels, Operand<T> predictions) {
-    DataType<T> dataType = predictions.asOutput().dataType();
+    Class<T> dataType = predictions.asOutput().type();
     Operand<T> tLabels = cast(tf, labels, dataType);
     LossTuple<T> lossTuple = LossesHelper.squeezeOrExpandDimensions(tf, tLabels, predictions, null);
     predictions = lossTuple.getTarget();
@@ -482,7 +481,7 @@ public static <T extends TNumber, U extends TNumber> Operand<T> logCosh(
    */
   public static <T extends TNumber, U extends TNumber> Operand<T> poisson(
       Ops tf, Operand<U> labels, Operand<T> predictions) {
-    DataType<T> dataType = predictions.asOutput().dataType();
+    Class<T> dataType = predictions.asOutput().type();
     Operand<T> tLabels = cast(tf, labels, dataType);
     LossTuple<T> lossTuple = LossesHelper.squeezeOrExpandDimensions(tf, tLabels, predictions, null);
     predictions = lossTuple.getTarget();
@@ -509,7 +508,7 @@ public static <T extends TNumber, U extends TNumber> Operand<T> poisson(
    */
   public static <T extends TNumber, U extends TNumber> Operand<T> sparseCategoricalCrossentropy(
       Ops tf, Operand<U> labels, Operand<T> predictions, boolean fromLogits, int axis) {
-    DataType<T> dataType = predictions.asOutput().dataType();
+    Class<T> dataType = predictions.asOutput().type();
     Operand<T> epsilonConst = cast(tf, tf.constant(EPSILON), dataType);
     Operand<T> one = cast(tf, tf.constant(1), dataType);
     Operand<T> oneMinusEpsilonConst = tf.math.sub(one, epsilonConst);
@@ -546,7 +545,7 @@ public static <T extends TNumber, U extends TNumber> Operand<T> sparseCategorica
       predictions = tf.linalg.transpose(predictions, tf.constant(axisNew));
     }
 
-    Operand<TInt64> iLabels = cast(tf, labels, TInt64.DTYPE);
+    Operand<TInt64> iLabels = cast(tf, labels, TInt64.class);
 
     // Try to adjust the shape so that rank of labels = rank of logits - 1.
     Shape labelsShape = labels.asOutput().shape();
@@ -586,7 +585,7 @@ public static <T extends TNumber, U extends TNumber> Operand<T> sparseCategorica
    */
   public static <T extends TNumber, U extends TNumber> Operand<T> squaredHinge(
       Ops tf, Operand<U> labels, Operand<T> predictions) {
-    DataType<T> dataType = predictions.asOutput().dataType();
+    Class<T> dataType = predictions.asOutput().type();
     Operand<T> tLabels = cast(tf, labels, dataType);
     LossTuple<T> lossTuple = LossesHelper.squeezeOrExpandDimensions(tf, tLabels, predictions, null);
     predictions = lossTuple.getTarget();
@@ -614,7 +613,7 @@ public static <T extends TNumber, U extends TNumber> Operand<T> squaredHinge(
    */
   private static <T extends TNumber> Operand<T> smoothBinaryLabels(
       Ops tf, Operand<T> labels, float labelSmoothing) {
-    DataType<T> dataType = labels.asOutput().dataType();
+    Class<T> dataType = labels.asOutput().type();
     Operand<T> oneMinusSmoothing = cast(tf, tf.constant(1.f - labelSmoothing), dataType);
     Operand<T> halfSmoothing = cast(tf, tf.constant(0.5F * labelSmoothing), dataType);
     return tf.math.add(tf.math.mul(labels, oneMinusSmoothing), halfSmoothing);
@@ -633,7 +632,7 @@ private static <T extends TNumber> Operand<T> smoothBinaryLabels(
    */
   private static <T extends TNumber> Operand<T> smoothCategoricalLabels(
       Ops tf, Operand<T> labels, float labelSmoothing) {
-    DataType<T> dataType = labels.asOutput().dataType();
+    Class<T> dataType = labels.asOutput().type();
     Operand<T> smoothing = cast(tf, tf.constant(labelSmoothing), dataType);
     Shape labelsShape = labels.asOutput().shape();
     int numDims = labelsShape.numDimensions();
@@ -656,7 +655,7 @@ public static <T extends TNumber> Operand<T> l2Normalize(Ops tf, Operand<T> x, i
         tf.reduceSum(tf.math.square(x), tf.constant(axis), ReduceSum.keepDims(Boolean.TRUE));
     Operand<T> invNorm =
         tf.math.rsqrt(
-            tf.math.maximum(squareSum, cast(tf, tf.constant(1e-12F), x.asOutput().dataType())));
+            tf.math.maximum(squareSum, cast(tf, tf.constant(1e-12F), x.asOutput().type())));
     return tf.math.mul(x, invNorm);
   }
 
@@ -669,7 +668,7 @@ public static <T extends TNumber> Operand<T> l2Normalize(Ops tf, Operand<T> x, i
    * @return the labels, possibly converted into -1/1.
    */
   private static <T extends TNumber> Operand<T> maybeConvertLabels(Ops tf, Operand<T> labels) {
-    DataType<T> dataType = labels.asOutput().dataType();
+    Class<T> dataType = labels.asOutput().type();
 
     Operand<T> one = cast(tf, tf.constant(1), dataType);
     Operand<T> zero = cast(tf, tf.constant(0), dataType);
diff --git a/tensorflow-framework/src/main/java/org/tensorflow/framework/losses/SparseCategoricalCrossentropy.java b/tensorflow-framework/src/main/java/org/tensorflow/framework/losses/SparseCategoricalCrossentropy.java
index 5586a4da889..9372126d3ab 100644
--- a/tensorflow-framework/src/main/java/org/tensorflow/framework/losses/SparseCategoricalCrossentropy.java
+++ b/tensorflow-framework/src/main/java/org/tensorflow/framework/losses/SparseCategoricalCrossentropy.java
@@ -205,8 +205,8 @@ public <T extends TNumber, U extends TNumber> Operand<T> call(
                       getTF(),
                       "predictions range check [0-1]",
                       predictions,
-                      cast(getTF(), getTF().constant(0), predictions.asOutput().dataType()),
-                      cast(getTF(), getTF().constant(1), predictions.asOutput().dataType()));
+                      cast(getTF(), getTF().constant(0), predictions.asOutput().type()),
+                      cast(getTF(), getTF().constant(1), predictions.asOutput().type()));
 
     } else {
       lPredictions = predictions;
diff --git a/tensorflow-framework/src/main/java/org/tensorflow/framework/losses/SquaredHinge.java b/tensorflow-framework/src/main/java/org/tensorflow/framework/losses/SquaredHinge.java
index 182ce592e55..9b712dcc712 100644
--- a/tensorflow-framework/src/main/java/org/tensorflow/framework/losses/SquaredHinge.java
+++ b/tensorflow-framework/src/main/java/org/tensorflow/framework/losses/SquaredHinge.java
@@ -125,15 +125,15 @@ public SquaredHinge(Ops tf, String name, Reduction reduction) {
   public <T extends TNumber, U extends TNumber> Operand<T> call(
       Operand<U> labels, Operand<T> predictions, Operand<T> sampleWeights) {
     @SuppressWarnings("unchecked")
-    Operand<T> tLabels = predictions.asOutput().dataType() == labels.asOutput().dataType() ?
+    Operand<T> tLabels = predictions.asOutput().type() == labels.asOutput().type() ?
             (Operand<T>)labels :
-            cast(tf,  labels, predictions.asOutput().dataType());
+            cast(tf,  labels, predictions.asOutput().type());
     tLabels = LossesHelper.valueCheck(
             getTF(),
             "labels value check [-1, 0, 1]",
             tLabels,
             cast(getTF(), getTF().constant(new int[] { -1, 0, 1}),
-                    predictions.asOutput().dataType()));
+                    predictions.asOutput().type()));
     Operand<T> losses = Losses.squaredHinge(getTF(), tLabels, predictions);
     return LossesHelper.computeWeightedLoss(getTF(), losses, getReduction(), sampleWeights);
   }
diff --git a/tensorflow-framework/src/main/java/org/tensorflow/framework/losses/impl/LossesHelper.java b/tensorflow-framework/src/main/java/org/tensorflow/framework/losses/impl/LossesHelper.java
index 463296a1f50..1a9f1339e24 100644
--- a/tensorflow-framework/src/main/java/org/tensorflow/framework/losses/impl/LossesHelper.java
+++ b/tensorflow-framework/src/main/java/org/tensorflow/framework/losses/impl/LossesHelper.java
@@ -14,7 +14,6 @@
 =======================================================================*/
 package org.tensorflow.framework.losses.impl;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.framework.losses.Reduction;
 import org.tensorflow.ndarray.Shape;
@@ -249,7 +248,7 @@ public static <T extends TNumber> LossTuple<T> removeSqueezableDimensions(
    */
   public static <T extends TNumber> Operand<T> computeWeightedLoss(
       Ops tf, Operand<T> loss, Reduction reduction, Operand<T> sampleWeight) {
-    DataType<T> dataType = loss.asOutput().dataType();
+    Class<T> dataType = loss.asOutput().type();
     if (sampleWeight == null) {
       sampleWeight = cast(tf, tf.constant(1), dataType);
     }
@@ -300,7 +299,7 @@ public static <T extends TNumber> Operand<T> safeMean(
       Ops tf, Operand<T> losses, long numElements) {
     Operand<T> totalLoss = tf.reduceSum(losses, allAxes(tf, losses));
     return tf.math.divNoNan(
-        totalLoss, cast(tf, tf.constant(numElements), losses.asOutput().dataType()));
+        totalLoss, cast(tf, tf.constant(numElements), losses.asOutput().type()));
   }
 
   /**
@@ -361,7 +360,7 @@ public static <T extends TNumber> Operand<T> rangeCheck(
           tf.withSubScope("rangeCheck")
               .withControlDependencies(Collections.singletonList(assertThat));
       return ltf.identity(values);
-    } else if (!cond.asOutput().data().getBoolean())
+    } else if (!cond.asOutput().asTensor().getBoolean())
       throw new IllegalArgumentException(String.format("%s : values out of range", prefix));
     else return values;
   }
@@ -386,7 +385,7 @@ public static <T extends TNumber> Operand<T> valueCheck(
       Ops tf, String prefix, Operand<T> values, Operand<T> allowedValues) {
     Operand<T> flatValues =
         tf.reshape(values, tf.constant(Shape.of(values.asOutput().shape().size())));
-    SetDiff1d<T, TInt32> diff = tf.setDiff1d(flatValues, allowedValues, TInt32.DTYPE);
+    SetDiff1d<T, TInt32> diff = tf.setDiff1d(flatValues, allowedValues, TInt32.class);
     long diffSize = diff.out().asOutput().shape().size();
 
     if (diffSize != Shape.UNKNOWN_SIZE) {
@@ -409,7 +408,7 @@ public static <T extends TNumber> Operand<T> valueCheck(
             tf.withSubScope("valueCheck")
                 .withControlDependencies(Collections.singletonList(assertThat));
         return ltf.identity(values);
-      } else if (!cond.asOutput().data().getBoolean())
+      } else if (!cond.asOutput().asTensor().getBoolean())
         throw new IllegalArgumentException(String.format("%s : values not in value set", prefix));
       else return values;
     }
diff --git a/tensorflow-framework/src/main/java/org/tensorflow/framework/utils/CastHelper.java b/tensorflow-framework/src/main/java/org/tensorflow/framework/utils/CastHelper.java
index aec75e6078a..4e310ff56e3 100644
--- a/tensorflow-framework/src/main/java/org/tensorflow/framework/utils/CastHelper.java
+++ b/tensorflow-framework/src/main/java/org/tensorflow/framework/utils/CastHelper.java
@@ -15,7 +15,6 @@
  */
 package org.tensorflow.framework.utils;
 
-import org.tensorflow.DataType;
 import org.tensorflow.Operand;
 import org.tensorflow.op.Ops;
 import org.tensorflow.types.family.TType;
@@ -35,8 +34,8 @@ public class CastHelper {
    */
   @SuppressWarnings("unchecked")
   public static <T extends TType, U extends TType> Operand<T> cast(
-          Ops tf, Operand<U> value, DataType<T> requiredType) {
-    return (value.asOutput().dataType() == requiredType)
+          Ops tf, Operand<U> value, Class<T> requiredType) {
+    return (value.asOutput().type() == requiredType)
         ? (Operand<T>) value
         : tf.dtypes.cast(value, requiredType);
   }
diff --git a/tensorflow-framework/src/test/java/org/tensorflow/framework/activations/ReLUTest.java b/tensorflow-framework/src/test/java/org/tensorflow/framework/activations/ReLUTest.java
index f54401515ab..a0aa2c4b453 100644
--- a/tensorflow-framework/src/test/java/org/tensorflow/framework/activations/ReLUTest.java
+++ b/tensorflow-framework/src/test/java/org/tensorflow/framework/activations/ReLUTest.java
@@ -90,8 +90,8 @@ public void testCallFloat16() {
         Ops tf = session.getTF();
         ReLU<TFloat16> instance = new ReLU<>(tf);
         Operand<TFloat16> result =
-            instance.call(tf.dtypes.cast(tf.constant(input), TFloat16.DTYPE));
-        session.evaluate(tf.dtypes.cast(tf.constant(expected), TFloat16.DTYPE), result);
+            instance.call(tf.dtypes.cast(tf.constant(input), TFloat16.class));
+        session.evaluate(tf.dtypes.cast(tf.constant(expected), TFloat16.class), result);
       }
   }
 
diff --git a/tensorflow-framework/src/test/java/org/tensorflow/framework/utils/EagerTestSession.java b/tensorflow-framework/src/test/java/org/tensorflow/framework/utils/EagerTestSession.java
index 69fcee43ab3..f9033c49cc8 100644
--- a/tensorflow-framework/src/test/java/org/tensorflow/framework/utils/EagerTestSession.java
+++ b/tensorflow-framework/src/test/java/org/tensorflow/framework/utils/EagerTestSession.java
@@ -370,7 +370,7 @@ public <T extends TType> void evaluate(Output<T> input, Predicate<Number> predic
             .scalars()
             .forEachIndexed((idx, f) -> assertTrue(predicate.test(o.asTensor().getDouble())));
       }
-    } else if (type == TFloat16.DTYPE) {
+    } else if (type == TFloat16.class) {
       Output<TFloat16> o = (Output<TFloat16>) input;
       if (debug) {
         if (isScalar) {
diff --git a/tensorflow-framework/src/test/java/org/tensorflow/framework/utils/GraphTestSession.java b/tensorflow-framework/src/test/java/org/tensorflow/framework/utils/GraphTestSession.java
index 42d5f0fe696..d2e856e9f22 100644
--- a/tensorflow-framework/src/test/java/org/tensorflow/framework/utils/GraphTestSession.java
+++ b/tensorflow-framework/src/test/java/org/tensorflow/framework/utils/GraphTestSession.java
@@ -537,45 +537,44 @@ public <T extends TType> void evaluate(Output<T> expected, Output<T> input) {
                       assertEquals(expectedResult.getDouble(idx), f.getDouble(), epsilon));
         }
       }
-    } else if (dtype == TFloat16.DTYPE) {
-        final Output<TFloat16> finalExpected = (Output<TFloat16>) expected;
-        if (debug) {
-          try (TFloat16 result =
-                  this.getGraphSession().runner().fetch(input).run().get(0);
-              TFloat16 expectedResult =
-                  this.getGraphSession().runner().fetch(input).run().get(0)) {
-            if (isScalar) {
-              System.out.printf(
-                  "0). %f <==> %f\n", expectedResult.getFloat(), result.getFloat());
-            } else {
-              result
-                  .scalars()
-                  .forEachIndexed(
-                      (idx, f) ->
-                          System.out.printf(
-                              "%d). %f <==> %f\n",
-                              index.getAndIncrement(),
-                              finalExpected.asTensor().getFloat(idx),
-                              f.getFloat()));
-            }
-          }
-        }
-        index.set(0);
+    } else if (dtype == TFloat16.class) {
+      final Output<TFloat16> finalExpected = (Output<TFloat16>) expected;
+      if (debug) {
         try (TFloat16 result =
                 this.getGraphSession().runner().fetch(input).run().get(0);
             TFloat16 expectedResult =
                 this.getGraphSession().runner().fetch(input).run().get(0)) {
           if (isScalar) {
-            assertEquals(expectedResult.getFloat(), result.getFloat(), epsilon);
+            System.out.printf(
+                "0). %f <==> %f\n", expectedResult.getFloat(), result.getFloat());
           } else {
             result
                 .scalars()
                 .forEachIndexed(
                     (idx, f) ->
-                        assertEquals(expectedResult.getFloat(idx), f.getFloat(), epsilon));
+                        System.out.printf(
+                            "%d). %f <==> %f\n",
+                            index.getAndIncrement(),
+                            finalExpected.asTensor().getFloat(idx),
+                            f.getFloat()));
           }
         }
       }
+      index.set(0);
+      try (TFloat16 result =
+              this.getGraphSession().runner().fetch(input).run().get(0);
+          TFloat16 expectedResult =
+              this.getGraphSession().runner().fetch(input).run().get(0)) {
+        if (isScalar) {
+          assertEquals(expectedResult.getFloat(), result.getFloat(), epsilon);
+        } else {
+          result
+              .scalars()
+              .forEachIndexed(
+                  (idx, f) ->
+                      assertEquals(expectedResult.getFloat(idx), f.getFloat(), epsilon));
+        }
+      }
     } else if (dtype == TInt32.class) {
       final Output<TInt32> finalExpected = (Output<TInt32>) expected;
       if (debug) {